LLVMFuzzerTestOneInput: 60| 13.7k|extern "C" int LLVMFuzzerTestOneInput(const uint8_t* Data, size_t Size) { 61| 13.7k| assert(Data); ------------------ | Branch (61:3): [True: 13.7k, False: 0] ------------------ 62| | 63| 13.7k| const rawspeed::Buffer buffer( 64| 13.7k| Data, rawspeed::implicit_cast(Size)); 65| | 66| 13.7k| try { 67| 13.7k| rawspeed::PARSER parser(buffer); 68| | 69| 13.7k|#if GETDECODER 70| 13.7k|#if DECODE 71| 13.7k| auto decoder = 72| 13.7k|#endif 73| 13.7k| parser.getDecoder(); 74| 13.7k|#endif 75| | 76| 13.7k|#if DECODE 77| 13.7k| decoder->applyCrop = false; 78| 13.7k| decoder->interpolateBadPixels = false; 79| 13.7k| decoder->failOnUnknown = false; 80| | // decoder->checkSupport(&metadata); 81| | 82| 13.7k| decoder->decodeRaw(); 83| 13.7k| decoder->decodeMetaData(&metadata); 84| 13.7k|#endif 85| 13.7k| } catch (const rawspeed::RawParserException&) { 86| 1.22k| return 0; 87| 1.22k|#if GETDECODER 88| 9.27k| } catch (const rawspeed::RawDecoderException&) { 89| 9.27k| return 0; 90| 9.27k|#endif 91| 9.27k| } catch (const rawspeed::IOException&) { 92| 894| return 0; 93| 894|#if DECODE 94| 894| } catch (const rawspeed::RawspeedException&) { 95| 0| return 0; 96| 0|#endif 97| 0| } 98| | 99| 2.38k| return 0; 100| 13.7k|} _ZNK8rawspeed16AlignedAllocatorIhLi16EE10deallocateEPhm: 74| 7.93k| void deallocate(T* p, std::size_t n) const noexcept { 75| 7.93k| invariant(p); ------------------ | | 27| 7.93k|#define invariant(expr) assert(expr) ------------------ | Branch (75:5): [True: 7.93k, False: 0] ------------------ 76| 7.93k| invariant(n > 0); ------------------ | | 27| 7.93k|#define invariant(expr) assert(expr) ------------------ | Branch (76:5): [True: 7.93k, False: 0] ------------------ 77| 7.93k| invariant(isAligned(p, alignment)); ------------------ | | 27| 7.93k|#define invariant(expr) assert(expr) ------------------ | Branch (77:5): [True: 7.93k, False: 0] ------------------ 78| 7.93k| operator delete(p, static_cast(alignment)); 79| 7.93k| } _ZNK8rawspeed16AlignedAllocatorIhLi16EE8allocateEm: 47| 7.93k| [[nodiscard]] T* allocate(std::size_t numElts) const { 48| 7.93k| static_assert(size_t(alignment) >= alignof(T), "insufficient alignment"); 49| 7.93k| invariant(numElts > 0 && "Should not be trying to allocate no elements"); ------------------ | | 27| 7.93k|#define invariant(expr) assert(expr) ------------------ | Branch (49:5): [True: 7.93k, False: 0] | Branch (49:5): [True: 7.93k, Folded] | Branch (49:5): [True: 7.93k, False: 0] ------------------ 50| 7.93k| assert(numElts <= allocator_traits::max_size(*this) && ------------------ | Branch (50:5): [True: 7.93k, False: 0] | Branch (50:5): [True: 7.93k, Folded] | Branch (50:5): [True: 7.93k, False: 0] ------------------ 51| 7.93k| "Can allocate this many elements."); 52| 7.93k| invariant(numElts <= SIZE_MAX / sizeof(T) && ------------------ | | 27| 7.93k|#define invariant(expr) assert(expr) ------------------ | Branch (52:5): [True: 7.93k, False: 0] | Branch (52:5): [True: 7.93k, Folded] | Branch (52:5): [True: 7.93k, False: 0] ------------------ 53| 7.93k| "Byte count calculation will not overflow"); 54| | 55| 7.93k| std::size_t numBytes = sizeof(T) * numElts; 56| | 57| 7.93k|#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION 58| | // workaround ASAN's broken allocator_may_return_null option 59| | // plus, avoidance of libFuzzer's rss_limit_mb option 60| | // if trying to alloc more than 2GB, just return null. 61| | // else it would abort() the whole program... 62| 7.93k| if (numBytes > 2UL << 30UL) ------------------ | Branch (62:9): [True: 0, False: 7.93k] ------------------ 63| 7.93k| ThrowRSE("FUZZ alloc bailout (%zu bytes)", numBytes); ------------------ | | 95| 0| ThrowExceptionHelper(rawspeed::RawspeedException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 64| 7.93k|#endif 65| | 66| 7.93k| auto* r = static_cast(operator new( 67| 7.93k| numBytes, static_cast(alignment))); 68| 7.93k| invariant(isAligned(r, alignment)); ------------------ | | 27| 7.93k|#define invariant(expr) assert(expr) ------------------ | Branch (68:5): [True: 7.93k, False: 0] ------------------ 69| 7.93k| if (!r) ------------------ | Branch (69:9): [True: 0, False: 7.93k] ------------------ 70| 7.93k| ThrowRSE("Out of memory while trying to allocate %zu bytes", numBytes); ------------------ | | 95| 0| ThrowExceptionHelper(rawspeed::RawspeedException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 71| 7.93k| return r; 72| 7.93k| } _ZN8rawspeed10Array1DRefIKhEC2EPS1_i: 104| 83.0M| : data(data_), numElts(numElts_) { 105| 83.0M| establishClassInvariants(); 106| 83.0M|} _ZNK8rawspeed10Array1DRefIKhE24establishClassInvariantsEv: 97| 777M|Array1DRef::establishClassInvariants() const noexcept { 98| 777M| invariant(data); ------------------ | | 27| 777M|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 777M, False: 18.4E] ------------------ 99| 777M| invariant(numElts >= 0); ------------------ | | 27| 777M|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 777M, False: 18.4E] ------------------ 100| 777M|} _ZNK8rawspeed10Array1DRefIKhE5beginEv: 159| 52.1M|template inline T* Array1DRef::begin() const { 160| 52.1M| establishClassInvariants(); 161| 52.1M| return addressOf(/*eltIdx=*/0); 162| 52.1M|} _ZNK8rawspeed10Array1DRefIKhE9addressOfEi: 133| 85.6M|Array1DRef::addressOf(const int eltIdx) const { 134| 85.6M| establishClassInvariants(); 135| 85.6M| invariant(eltIdx >= 0); ------------------ | | 27| 85.6M|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 85.6M, False: 18.4E] ------------------ 136| 85.6M| invariant(eltIdx <= numElts); ------------------ | | 27| 85.6M|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 85.6M, False: 18.4E] ------------------ 137| 85.6M|#pragma GCC diagnostic push 138| 85.6M|#pragma GCC diagnostic ignored "-Wpragmas" 139| 85.6M|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 140| 85.6M|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 141| 85.6M| return data + eltIdx; 142| 85.6M|#pragma GCC diagnostic pop 143| 85.6M|} _ZNK8rawspeed10Array1DRefIKhE4sizeEv: 154| 392M|template inline int Array1DRef::size() const { 155| 392M| establishClassInvariants(); 156| 392M| return numElts; 157| 392M|} _ZNK8rawspeed10Array1DRefIKSt4byteE4sizeEv: 154| 202M|template inline int Array1DRef::size() const { 155| 202M| establishClassInvariants(); 156| 202M| return numElts; 157| 202M|} _ZNK8rawspeed10Array1DRefIKSt4byteE24establishClassInvariantsEv: 97| 337M|Array1DRef::establishClassInvariants() const noexcept { 98| 337M| invariant(data); ------------------ | | 27| 337M|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 348M, False: 18.4E] ------------------ 99| 348M| invariant(numElts >= 0); ------------------ | | 27| 348M|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 364M, False: 18.4E] ------------------ 100| 348M|} _ZNK8rawspeed10Array1DRefISt4byteE4sizeEv: 154| 217M|template inline int Array1DRef::size() const { 155| 217M| establishClassInvariants(); 156| 217M| return numElts; 157| 217M|} _ZNK8rawspeed10Array1DRefISt4byteE24establishClassInvariantsEv: 97| 393M|Array1DRef::establishClassInvariants() const noexcept { 98| 393M| invariant(data); ------------------ | | 27| 393M|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 395M, False: 18.4E] ------------------ 99| 395M| invariant(numElts >= 0); ------------------ | | 27| 395M|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 396M, False: 18.4E] ------------------ 100| 395M|} _ZNK8rawspeed10Array1DRefIKSt4byteE5beginEv: 159| 7.00M|template inline T* Array1DRef::begin() const { 160| 7.00M| establishClassInvariants(); 161| 7.00M| return addressOf(/*eltIdx=*/0); 162| 7.00M|} _ZNK8rawspeed10Array1DRefIKSt4byteE9addressOfEi: 133| 13.2M|Array1DRef::addressOf(const int eltIdx) const { 134| 13.2M| establishClassInvariants(); 135| 13.2M| invariant(eltIdx >= 0); ------------------ | | 27| 13.2M|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 13.0M, False: 174k] ------------------ 136| 13.0M| invariant(eltIdx <= numElts); ------------------ | | 27| 13.0M|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 13.0M, False: 44.5k] ------------------ 137| 13.0M|#pragma GCC diagnostic push 138| 13.0M|#pragma GCC diagnostic ignored "-Wpragmas" 139| 13.0M|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 140| 13.0M|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 141| 13.0M| return data + eltIdx; 142| 13.0M|#pragma GCC diagnostic pop 143| 13.0M|} _ZNK8rawspeed10Array1DRefIKSt4byteE3endEv: 163| 19.8k|template inline T* Array1DRef::end() const { 164| 19.8k| establishClassInvariants(); 165| 19.8k| return addressOf(/*eltIdx=*/numElts); 166| 19.8k|} _ZNK8rawspeed10Array1DRefISt4byteE5beginEv: 159| 7.93M|template inline T* Array1DRef::begin() const { 160| 7.93M| establishClassInvariants(); 161| 7.93M| return addressOf(/*eltIdx=*/0); 162| 7.93M|} _ZNK8rawspeed10Array1DRefISt4byteE9addressOfEi: 133| 30.7M|Array1DRef::addressOf(const int eltIdx) const { 134| 30.7M| establishClassInvariants(); 135| 30.7M| invariant(eltIdx >= 0); ------------------ | | 27| 30.7M|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 30.6M, False: 112k] ------------------ 136| 30.6M| invariant(eltIdx <= numElts); ------------------ | | 27| 30.6M|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 30.6M, False: 19.9k] ------------------ 137| 30.6M|#pragma GCC diagnostic push 138| 30.6M|#pragma GCC diagnostic ignored "-Wpragmas" 139| 30.6M|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 140| 30.6M|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 141| 30.6M| return data + eltIdx; 142| 30.6M|#pragma GCC diagnostic pop 143| 30.6M|} _ZNK8rawspeed10Array1DRefISt4byteE7getCropEii: 110| 15.5M|Array1DRef::getCrop(int offset, int size) const { 111| 15.5M| establishClassInvariants(); 112| 15.5M| invariant(offset >= 0); ------------------ | | 27| 15.5M|#define invariant(expr) assert(expr) ------------------ | Branch (112:3): [True: 15.6M, False: 18.4E] ------------------ 113| 15.6M| invariant(size >= 0); ------------------ | | 27| 15.6M|#define invariant(expr) assert(expr) ------------------ | Branch (113:3): [True: 15.6M, False: 493] ------------------ 114| 15.6M| invariant(offset <= numElts); ------------------ | | 27| 15.6M|#define invariant(expr) assert(expr) ------------------ | Branch (114:3): [True: 15.6M, False: 18.4E] ------------------ 115| 15.6M| invariant(size <= numElts); ------------------ | | 27| 15.6M|#define invariant(expr) assert(expr) ------------------ | Branch (115:3): [True: 15.6M, False: 18.4E] ------------------ 116| 15.6M| invariant(offset + size <= numElts); ------------------ | | 27| 15.6M|#define invariant(expr) assert(expr) ------------------ | Branch (116:3): [True: 15.6M, False: 18.4E] ------------------ 117| 15.6M| return {*this, offset, size}; 118| 15.6M|} _ZNK8rawspeed10Array1DRefIKSt4byteE7getCropEii: 110| 7.09M|Array1DRef::getCrop(int offset, int size) const { 111| 7.09M| establishClassInvariants(); 112| 7.09M| invariant(offset >= 0); ------------------ | | 27| 7.09M|#define invariant(expr) assert(expr) ------------------ | Branch (112:3): [True: 7.16M, False: 18.4E] ------------------ 113| 7.16M| invariant(size >= 0); ------------------ | | 27| 7.16M|#define invariant(expr) assert(expr) ------------------ | Branch (113:3): [True: 7.16M, False: 18.4E] ------------------ 114| 7.16M| invariant(offset <= numElts); ------------------ | | 27| 7.16M|#define invariant(expr) assert(expr) ------------------ | Branch (114:3): [True: 7.17M, False: 18.4E] ------------------ 115| 7.17M| invariant(size <= numElts); ------------------ | | 27| 7.17M|#define invariant(expr) assert(expr) ------------------ | Branch (115:3): [True: 7.17M, False: 18.4E] ------------------ 116| 7.17M| invariant(offset + size <= numElts); ------------------ | | 27| 7.17M|#define invariant(expr) assert(expr) ------------------ | Branch (116:3): [True: 7.17M, False: 18.4E] ------------------ 117| 7.17M| return {*this, offset, size}; 118| 7.17M|} _ZNK8rawspeed10Array1DRefIKhE7getCropEii: 110| 29.4M|Array1DRef::getCrop(int offset, int size) const { 111| 29.4M| establishClassInvariants(); 112| 29.4M| invariant(offset >= 0); ------------------ | | 27| 29.4M|#define invariant(expr) assert(expr) ------------------ | Branch (112:3): [True: 29.4M, False: 18.4E] ------------------ 113| 29.4M| invariant(size >= 0); ------------------ | | 27| 29.4M|#define invariant(expr) assert(expr) ------------------ | Branch (113:3): [True: 29.4M, False: 18.4E] ------------------ 114| 29.4M| invariant(offset <= numElts); ------------------ | | 27| 29.4M|#define invariant(expr) assert(expr) ------------------ | Branch (114:3): [True: 29.4M, False: 18.4E] ------------------ 115| 29.4M| invariant(size <= numElts); ------------------ | | 27| 29.4M|#define invariant(expr) assert(expr) ------------------ | Branch (115:3): [True: 29.4M, False: 18.4E] ------------------ 116| 29.4M| invariant(offset + size <= numElts); ------------------ | | 27| 29.4M|#define invariant(expr) assert(expr) ------------------ | Branch (116:3): [True: 29.4M, False: 18.4E] ------------------ 117| 29.4M| return {*this, offset, size}; 118| 29.4M|} _ZNK8rawspeed10Array1DRefIKhEclEi: 147| 2.73M|Array1DRef::operator()(const int eltIdx) const { 148| 2.73M| establishClassInvariants(); 149| 2.73M| invariant(eltIdx >= 0); ------------------ | | 27| 2.73M|#define invariant(expr) assert(expr) ------------------ | Branch (149:3): [True: 2.73M, False: 0] ------------------ 150| 2.73M| invariant(eltIdx < numElts); ------------------ | | 27| 2.73M|#define invariant(expr) assert(expr) ------------------ | Branch (150:3): [True: 2.73M, False: 0] ------------------ 151| 2.73M| return *addressOf(eltIdx); 152| 2.73M|} _ZNK8rawspeed10Array1DRefIKhE3endEv: 163| 1.27M|template inline T* Array1DRef::end() const { 164| 1.27M| establishClassInvariants(); 165| 1.27M| return addressOf(/*eltIdx=*/numElts); 166| 1.27M|} _ZNK8rawspeed10Array1DRefItE24establishClassInvariantsEv: 97| 1.21G|Array1DRef::establishClassInvariants() const noexcept { 98| 1.21G| invariant(data); ------------------ | | 27| 1.21G|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 1.23G, False: 18.4E] ------------------ 99| 1.23G| invariant(numElts >= 0); ------------------ | | 27| 1.23G|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 1.23G, False: 18.4E] ------------------ 100| 1.23G|} _ZNK8rawspeed10Array1DRefItE4sizeEv: 154| 611M|template inline int Array1DRef::size() const { 155| 611M| establishClassInvariants(); 156| 611M| return numElts; 157| 611M|} _ZNK8rawspeed10Array1DRefIfE24establishClassInvariantsEv: 97| 18.5M|Array1DRef::establishClassInvariants() const noexcept { 98| 18.5M| invariant(data); ------------------ | | 27| 18.5M|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 18.5M, False: 6] ------------------ 99| 18.5M| invariant(numElts >= 0); ------------------ | | 27| 18.5M|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 18.5M, False: 1] ------------------ 100| 18.5M|} _ZNK8rawspeed10Array1DRefIfE4sizeEv: 154| 10.2M|template inline int Array1DRef::size() const { 155| 10.2M| establishClassInvariants(); 156| 10.2M| return numElts; 157| 10.2M|} _ZN8rawspeed10Array1DRefIhEC2EPhi: 104| 19| : data(data_), numElts(numElts_) { 105| 19| establishClassInvariants(); 106| 19|} _ZNK8rawspeed10Array1DRefIhE24establishClassInvariantsEv: 97| 2.78M|Array1DRef::establishClassInvariants() const noexcept { 98| 2.78M| invariant(data); ------------------ | | 27| 2.78M|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 2.78M, False: 0] ------------------ 99| 2.78M| invariant(numElts >= 0); ------------------ | | 27| 2.78M|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 2.78M, False: 0] ------------------ 100| 2.78M|} _ZNK8rawspeed10Array1DRefIhE4sizeEv: 154| 1.66M|template inline int Array1DRef::size() const { 155| 1.66M| establishClassInvariants(); 156| 1.66M| return numElts; 157| 1.66M|} _ZNK8rawspeed10Array1DRefIKhE8getBlockEii: 122| 185k|Array1DRef::getBlock(int size, int index) const { 123| 185k| establishClassInvariants(); 124| 185k| invariant(index >= 0); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (124:3): [True: 185k, False: 0] ------------------ 125| 185k| invariant(size >= 0); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (125:3): [True: 185k, False: 0] ------------------ 126| 185k| invariant(index <= numElts); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (126:3): [True: 185k, False: 0] ------------------ 127| 185k| invariant(size <= numElts); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (127:3): [True: 185k, False: 0] ------------------ 128| 185k| return getCrop(size * index, size); 129| 185k|} _ZNK8rawspeed10Array1DRefIhE8getBlockEii: 122| 185k|Array1DRef::getBlock(int size, int index) const { 123| 185k| establishClassInvariants(); 124| 185k| invariant(index >= 0); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (124:3): [True: 185k, False: 0] ------------------ 125| 185k| invariant(size >= 0); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (125:3): [True: 185k, False: 0] ------------------ 126| 185k| invariant(index <= numElts); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (126:3): [True: 185k, False: 0] ------------------ 127| 185k| invariant(size <= numElts); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (127:3): [True: 185k, False: 0] ------------------ 128| 185k| return getCrop(size * index, size); 129| 185k|} _ZNK8rawspeed10Array1DRefIhE7getCropEii: 110| 185k|Array1DRef::getCrop(int offset, int size) const { 111| 185k| establishClassInvariants(); 112| 185k| invariant(offset >= 0); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (112:3): [True: 185k, False: 0] ------------------ 113| 185k| invariant(size >= 0); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (113:3): [True: 185k, False: 0] ------------------ 114| 185k| invariant(offset <= numElts); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (114:3): [True: 185k, False: 0] ------------------ 115| 185k| invariant(size <= numElts); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (115:3): [True: 185k, False: 0] ------------------ 116| 185k| invariant(offset + size <= numElts); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (116:3): [True: 185k, False: 0] ------------------ 117| 185k| return {*this, offset, size}; 118| 185k|} _ZNK8rawspeed10Array1DRefIhE9addressOfEi: 133| 185k|Array1DRef::addressOf(const int eltIdx) const { 134| 185k| establishClassInvariants(); 135| 185k| invariant(eltIdx >= 0); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 185k, False: 0] ------------------ 136| 185k| invariant(eltIdx <= numElts); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 185k, False: 0] ------------------ 137| 185k|#pragma GCC diagnostic push 138| 185k|#pragma GCC diagnostic ignored "-Wpragmas" 139| 185k|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 140| 185k|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 141| 185k| return data + eltIdx; 142| 185k|#pragma GCC diagnostic pop 143| 185k|} _ZN8rawspeed10Array1DRefIiEC2EPii: 104| 4.11k| : data(data_), numElts(numElts_) { 105| 4.11k| establishClassInvariants(); 106| 4.11k|} _ZNK8rawspeed10Array1DRefIiE24establishClassInvariantsEv: 97| 85.6k|Array1DRef::establishClassInvariants() const noexcept { 98| 85.6k| invariant(data); ------------------ | | 27| 85.6k|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 85.6k, False: 0] ------------------ 99| 85.6k| invariant(numElts >= 0); ------------------ | | 27| 85.6k|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 85.6k, False: 0] ------------------ 100| 85.6k|} _ZNK8rawspeed10Array1DRefIiE4sizeEv: 154| 41.9k|template inline int Array1DRef::size() const { 155| 41.9k| establishClassInvariants(); 156| 41.9k| return numElts; 157| 41.9k|} _ZNK8rawspeed10Array1DRefIiE7getCropEii: 110| 2.11k|Array1DRef::getCrop(int offset, int size) const { 111| 2.11k| establishClassInvariants(); 112| 2.11k| invariant(offset >= 0); ------------------ | | 27| 2.11k|#define invariant(expr) assert(expr) ------------------ | Branch (112:3): [True: 2.11k, False: 0] ------------------ 113| 2.11k| invariant(size >= 0); ------------------ | | 27| 2.11k|#define invariant(expr) assert(expr) ------------------ | Branch (113:3): [True: 2.11k, False: 0] ------------------ 114| 2.11k| invariant(offset <= numElts); ------------------ | | 27| 2.11k|#define invariant(expr) assert(expr) ------------------ | Branch (114:3): [True: 2.11k, False: 0] ------------------ 115| 2.11k| invariant(size <= numElts); ------------------ | | 27| 2.11k|#define invariant(expr) assert(expr) ------------------ | Branch (115:3): [True: 2.11k, False: 0] ------------------ 116| 2.11k| invariant(offset + size <= numElts); ------------------ | | 27| 2.11k|#define invariant(expr) assert(expr) ------------------ | Branch (116:3): [True: 2.11k, False: 0] ------------------ 117| 2.11k| return {*this, offset, size}; 118| 2.11k|} _ZNK8rawspeed10Array1DRefIiE9addressOfEi: 133| 7.26k|Array1DRef::addressOf(const int eltIdx) const { 134| 7.26k| establishClassInvariants(); 135| 7.26k| invariant(eltIdx >= 0); ------------------ | | 27| 7.26k|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 7.26k, False: 0] ------------------ 136| 7.26k| invariant(eltIdx <= numElts); ------------------ | | 27| 7.26k|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 7.26k, False: 0] ------------------ 137| 7.26k|#pragma GCC diagnostic push 138| 7.26k|#pragma GCC diagnostic ignored "-Wpragmas" 139| 7.26k|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 140| 7.26k|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 141| 7.26k| return data + eltIdx; 142| 7.26k|#pragma GCC diagnostic pop 143| 7.26k|} _ZNK8rawspeed10Array1DRefIiEclEi: 147| 1.53k|Array1DRef::operator()(const int eltIdx) const { 148| 1.53k| establishClassInvariants(); 149| 1.53k| invariant(eltIdx >= 0); ------------------ | | 27| 1.53k|#define invariant(expr) assert(expr) ------------------ | Branch (149:3): [True: 1.53k, False: 0] ------------------ 150| 1.53k| invariant(eltIdx < numElts); ------------------ | | 27| 1.53k|#define invariant(expr) assert(expr) ------------------ | Branch (150:3): [True: 1.53k, False: 0] ------------------ 151| 1.53k| return *addressOf(eltIdx); 152| 1.53k|} _ZNK8rawspeed10Array1DRefISt4byteE3endEv: 163| 8.01k|template inline T* Array1DRef::end() const { 164| 8.01k| establishClassInvariants(); 165| 8.01k| return addressOf(/*eltIdx=*/numElts); 166| 8.01k|} _ZNK8rawspeed10Array1DRefISt4byteEclEi: 147| 7.58M|Array1DRef::operator()(const int eltIdx) const { 148| 7.58M| establishClassInvariants(); 149| 7.58M| invariant(eltIdx >= 0); ------------------ | | 27| 7.58M|#define invariant(expr) assert(expr) ------------------ | Branch (149:3): [True: 7.58M, False: 18.4E] ------------------ 150| 7.58M| invariant(eltIdx < numElts); ------------------ | | 27| 7.58M|#define invariant(expr) assert(expr) ------------------ | Branch (150:3): [True: 7.58M, False: 1] ------------------ 151| 7.58M| return *addressOf(eltIdx); 152| 7.58M|} _ZN8rawspeed10Array1DRefItEC2EPti: 104| 46.5M| : data(data_), numElts(numElts_) { 105| 46.5M| establishClassInvariants(); 106| 46.5M|} _ZN8rawspeed10Array1DRefISt4byteEC2ItQaaaantaasr3stdE10is_const_vITL0__Entsr3stdE10is_const_vIT_Entsr3stdE9is_same_vIu14__remove_constIS5_Eu14__remove_constIS4_EEsr3stdE9is_same_vIS6_S1_EEENS0_IS5_EE: 78| 10.4k| : Array1DRef(reinterpret_cast(RHS.data), sizeof(T2) * RHS.numElts) {} _ZN8rawspeed10Array1DRefISt4byteEC2EPS1_i: 104| 21.3M| : data(data_), numElts(numElts_) { 105| 21.3M| establishClassInvariants(); 106| 21.3M|} _ZN8rawspeed10Array1DRefIfEC2EPfi: 104| 552k| : data(data_), numElts(numElts_) { 105| 552k| establishClassInvariants(); 106| 552k|} _ZN8rawspeed10Array1DRefISt4byteEC2IfQaaaantaasr3stdE10is_const_vITL0__Entsr3stdE10is_const_vIT_Entsr3stdE9is_same_vIu14__remove_constIS5_Eu14__remove_constIS4_EEsr3stdE9is_same_vIS6_S1_EEENS0_IS5_EE: 78| 545| : Array1DRef(reinterpret_cast(RHS.data), sizeof(T2) * RHS.numElts) {} _ZNK8rawspeed10Array1DRefItE9addressOfEi: 133| 169M|Array1DRef::addressOf(const int eltIdx) const { 134| 169M| establishClassInvariants(); 135| 169M| invariant(eltIdx >= 0); ------------------ | | 27| 169M|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 168M, False: 1.02M] ------------------ 136| 168M| invariant(eltIdx <= numElts); ------------------ | | 27| 168M|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 169M, False: 18.4E] ------------------ 137| 169M|#pragma GCC diagnostic push 138| 169M|#pragma GCC diagnostic ignored "-Wpragmas" 139| 169M|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 140| 169M|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 141| 169M| return data + eltIdx; 142| 168M|#pragma GCC diagnostic pop 143| 168M|} _ZNK8rawspeed10Array1DRefItE7getCropEii: 110| 43.2M|Array1DRef::getCrop(int offset, int size) const { 111| 43.2M| establishClassInvariants(); 112| 43.2M| invariant(offset >= 0); ------------------ | | 27| 43.2M|#define invariant(expr) assert(expr) ------------------ | Branch (112:3): [True: 47.3M, False: 18.4E] ------------------ 113| 47.3M| invariant(size >= 0); ------------------ | | 27| 47.3M|#define invariant(expr) assert(expr) ------------------ | Branch (113:3): [True: 47.4M, False: 18.4E] ------------------ 114| 47.4M| invariant(offset <= numElts); ------------------ | | 27| 47.4M|#define invariant(expr) assert(expr) ------------------ | Branch (114:3): [True: 47.4M, False: 18.4E] ------------------ 115| 47.4M| invariant(size <= numElts); ------------------ | | 27| 47.4M|#define invariant(expr) assert(expr) ------------------ | Branch (115:3): [True: 47.5M, False: 18.4E] ------------------ 116| 47.5M| invariant(offset + size <= numElts); ------------------ | | 27| 47.5M|#define invariant(expr) assert(expr) ------------------ | Branch (116:3): [True: 47.5M, False: 18.4E] ------------------ 117| 47.5M| return {*this, offset, size}; 118| 47.5M|} _ZNK8rawspeed10Array1DRefItEclEi: 147| 129M|Array1DRef::operator()(const int eltIdx) const { 148| 129M| establishClassInvariants(); 149| 129M| invariant(eltIdx >= 0); ------------------ | | 27| 129M|#define invariant(expr) assert(expr) ------------------ | Branch (149:3): [True: 130M, False: 18.4E] ------------------ 150| 130M| invariant(eltIdx < numElts); ------------------ | | 27| 130M|#define invariant(expr) assert(expr) ------------------ | Branch (150:3): [True: 129M, False: 1.15M] ------------------ 151| 129M| return *addressOf(eltIdx); 152| 130M|} _ZN8rawspeed10Array1DRefIKSt4byteEC2IKhQaaaantaasr3stdE10is_const_vITL0__Entsr3stdE10is_const_vIT_Entsr3stdE9is_same_vIu14__remove_constIS7_Eu14__remove_constIS6_EEsr3stdE9is_same_vIS8_S1_EEENS0_IS7_EE: 78| 415k| : Array1DRef(reinterpret_cast(RHS.data), sizeof(T2) * RHS.numElts) {} _ZN8rawspeed10Array1DRefIKSt4byteEC2EPS2_i: 104| 7.14M| : data(data_), numElts(numElts_) { 105| 7.14M| establishClassInvariants(); 106| 7.14M|} _ZNK8rawspeed10Array1DRefISt4byteE8getBlockEii: 122| 6.10M|Array1DRef::getBlock(int size, int index) const { 123| 6.10M| establishClassInvariants(); 124| 6.10M| invariant(index >= 0); ------------------ | | 27| 6.10M|#define invariant(expr) assert(expr) ------------------ | Branch (124:3): [True: 6.16M, False: 18.4E] ------------------ 125| 6.16M| invariant(size >= 0); ------------------ | | 27| 6.16M|#define invariant(expr) assert(expr) ------------------ | Branch (125:3): [True: 6.16M, False: 18.4E] ------------------ 126| 6.16M| invariant(index <= numElts); ------------------ | | 27| 6.16M|#define invariant(expr) assert(expr) ------------------ | Branch (126:3): [True: 6.16M, False: 18.4E] ------------------ 127| 6.16M| invariant(size <= numElts); ------------------ | | 27| 6.16M|#define invariant(expr) assert(expr) ------------------ | Branch (127:3): [True: 6.16M, False: 18.4E] ------------------ 128| 6.16M| return getCrop(size * index, size); 129| 6.16M|} _ZNK8rawspeed10Array1DRefIfE7getCropEii: 110| 590k|Array1DRef::getCrop(int offset, int size) const { 111| 590k| establishClassInvariants(); 112| 590k| invariant(offset >= 0); ------------------ | | 27| 590k|#define invariant(expr) assert(expr) ------------------ | Branch (112:3): [True: 590k, False: 18.4E] ------------------ 113| 590k| invariant(size >= 0); ------------------ | | 27| 590k|#define invariant(expr) assert(expr) ------------------ | Branch (113:3): [True: 590k, False: 0] ------------------ 114| 590k| invariant(offset <= numElts); ------------------ | | 27| 590k|#define invariant(expr) assert(expr) ------------------ | Branch (114:3): [True: 590k, False: 0] ------------------ 115| 590k| invariant(size <= numElts); ------------------ | | 27| 590k|#define invariant(expr) assert(expr) ------------------ | Branch (115:3): [True: 590k, False: 0] ------------------ 116| 590k| invariant(offset + size <= numElts); ------------------ | | 27| 590k|#define invariant(expr) assert(expr) ------------------ | Branch (116:3): [True: 590k, False: 0] ------------------ 117| 590k| return {*this, offset, size}; 118| 590k|} _ZNK8rawspeed10Array1DRefIfE9addressOfEi: 133| 1.10M|Array1DRef::addressOf(const int eltIdx) const { 134| 1.10M| establishClassInvariants(); 135| 1.10M| invariant(eltIdx >= 0); ------------------ | | 27| 1.10M|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 1.10M, False: 0] ------------------ 136| 1.10M| invariant(eltIdx <= numElts); ------------------ | | 27| 1.10M|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 1.10M, False: 0] ------------------ 137| 1.10M|#pragma GCC diagnostic push 138| 1.10M|#pragma GCC diagnostic ignored "-Wpragmas" 139| 1.10M|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 140| 1.10M|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 141| 1.10M| return data + eltIdx; 142| 1.10M|#pragma GCC diagnostic pop 143| 1.10M|} _ZNK8rawspeed10Array1DRefIfEclEi: 147| 511k|Array1DRef::operator()(const int eltIdx) const { 148| 511k| establishClassInvariants(); 149| 511k| invariant(eltIdx >= 0); ------------------ | | 27| 511k|#define invariant(expr) assert(expr) ------------------ | Branch (149:3): [True: 511k, False: 0] ------------------ 150| 511k| invariant(eltIdx < numElts); ------------------ | | 27| 511k|#define invariant(expr) assert(expr) ------------------ | Branch (150:3): [True: 511k, False: 0] ------------------ 151| 511k| return *addressOf(eltIdx); 152| 511k|} _ZNK8rawspeed10Array1DRefIKtE24establishClassInvariantsEv: 97| 103M|Array1DRef::establishClassInvariants() const noexcept { 98| 103M| invariant(data); ------------------ | | 27| 103M|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 103M, False: 18.4E] ------------------ 99| 103M| invariant(numElts >= 0); ------------------ | | 27| 103M|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 105M, False: 18.4E] ------------------ 100| 103M|} _ZNK8rawspeed10Array1DRefIKtE4sizeEv: 154| 60.4M|template inline int Array1DRef::size() const { 155| 60.4M| establishClassInvariants(); 156| 60.4M| return numElts; 157| 60.4M|} _ZNK8rawspeed10Array1DRefIKtE7getCropEii: 110| 3.71M|Array1DRef::getCrop(int offset, int size) const { 111| 3.71M| establishClassInvariants(); 112| 3.71M| invariant(offset >= 0); ------------------ | | 27| 3.71M|#define invariant(expr) assert(expr) ------------------ | Branch (112:3): [True: 3.72M, False: 18.4E] ------------------ 113| 3.72M| invariant(size >= 0); ------------------ | | 27| 3.72M|#define invariant(expr) assert(expr) ------------------ | Branch (113:3): [True: 3.72M, False: 18.4E] ------------------ 114| 3.72M| invariant(offset <= numElts); ------------------ | | 27| 3.72M|#define invariant(expr) assert(expr) ------------------ | Branch (114:3): [True: 3.72M, False: 18.4E] ------------------ 115| 3.72M| invariant(size <= numElts); ------------------ | | 27| 3.72M|#define invariant(expr) assert(expr) ------------------ | Branch (115:3): [True: 3.72M, False: 657] ------------------ 116| 3.72M| invariant(offset + size <= numElts); ------------------ | | 27| 3.72M|#define invariant(expr) assert(expr) ------------------ | Branch (116:3): [True: 3.72M, False: 18.4E] ------------------ 117| 3.72M| return {*this, offset, size}; 118| 3.72M|} _ZNK8rawspeed10Array1DRefIKtE9addressOfEi: 133| 7.41M|Array1DRef::addressOf(const int eltIdx) const { 134| 7.41M| establishClassInvariants(); 135| 7.41M| invariant(eltIdx >= 0); ------------------ | | 27| 7.41M|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 7.40M, False: 5.61k] ------------------ 136| 7.40M| invariant(eltIdx <= numElts); ------------------ | | 27| 7.40M|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 7.40M, False: 18.4E] ------------------ 137| 7.40M|#pragma GCC diagnostic push 138| 7.40M|#pragma GCC diagnostic ignored "-Wpragmas" 139| 7.40M|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 140| 7.40M|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 141| 7.40M| return data + eltIdx; 142| 7.40M|#pragma GCC diagnostic pop 143| 7.40M|} _ZN8rawspeed10Array1DRefIKtEC2EPS1_i: 104| 4.98M| : data(data_), numElts(numElts_) { 105| 4.98M| establishClassInvariants(); 106| 4.98M|} _ZNK8rawspeed10Array1DRefIKtEclEi: 147| 3.72M|Array1DRef::operator()(const int eltIdx) const { 148| 3.72M| establishClassInvariants(); 149| 3.72M| invariant(eltIdx >= 0); ------------------ | | 27| 3.72M|#define invariant(expr) assert(expr) ------------------ | Branch (149:3): [True: 3.72M, False: 18.4E] ------------------ 150| 3.72M| invariant(eltIdx < numElts); ------------------ | | 27| 3.72M|#define invariant(expr) assert(expr) ------------------ | Branch (150:3): [True: 3.72M, False: 3.24k] ------------------ 151| 3.72M| return *addressOf(eltIdx); 152| 3.72M|} _ZN8rawspeed10Array1DRefIKtEC2ItQaaaantsr3stdE10is_const_vITL0__Esr3stdE10is_const_vIT_Esr3stdE9is_same_vIu14__remove_constIS5_Eu14__remove_constIS4_EEEENS0_IS5_EE: 69| 1.27M| : Array1DRef(RHS.data, RHS.numElts) {} _ZNK8rawspeed10Array1DRefItE8getBlockEii: 122| 332k|Array1DRef::getBlock(int size, int index) const { 123| 332k| establishClassInvariants(); 124| 332k| invariant(index >= 0); ------------------ | | 27| 332k|#define invariant(expr) assert(expr) ------------------ | Branch (124:3): [True: 332k, False: 329] ------------------ 125| 332k| invariant(size >= 0); ------------------ | | 27| 332k|#define invariant(expr) assert(expr) ------------------ | Branch (125:3): [True: 332k, False: 18.4E] ------------------ 126| 332k| invariant(index <= numElts); ------------------ | | 27| 332k|#define invariant(expr) assert(expr) ------------------ | Branch (126:3): [True: 332k, False: 18.4E] ------------------ 127| 332k| invariant(size <= numElts); ------------------ | | 27| 332k|#define invariant(expr) assert(expr) ------------------ | Branch (127:3): [True: 332k, False: 18.4E] ------------------ 128| 332k| return getCrop(size * index, size); 129| 332k|} _ZNK8rawspeed10Array1DRefIiE5beginEv: 159| 1.80k|template inline T* Array1DRef::begin() const { 160| 1.80k| establishClassInvariants(); 161| 1.80k| return addressOf(/*eltIdx=*/0); 162| 1.80k|} _ZNK8rawspeed10Array1DRefIiE3endEv: 163| 1.80k|template inline T* Array1DRef::end() const { 164| 1.80k| establishClassInvariants(); 165| 1.80k| return addressOf(/*eltIdx=*/numElts); 166| 1.80k|} _ZN8rawspeed10Array1DRefIKNS_15DngSliceElementEEC2EPS2_i: 104| 13.7k| : data(data_), numElts(numElts_) { 105| 13.7k| establishClassInvariants(); 106| 13.7k|} _ZNK8rawspeed10Array1DRefIKNS_15DngSliceElementEE24establishClassInvariantsEv: 97| 67.7k|Array1DRef::establishClassInvariants() const noexcept { 98| 67.7k| invariant(data); ------------------ | | 27| 67.7k|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 67.7k, False: 18.4E] ------------------ 99| 67.7k| invariant(numElts >= 0); ------------------ | | 27| 67.7k|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 67.9k, False: 18.4E] ------------------ 100| 67.7k|} _ZNK8rawspeed10Array1DRefIKNS_15DngSliceElementEE3endEv: 163| 13.8k|template inline T* Array1DRef::end() const { 164| 13.8k| establishClassInvariants(); 165| 13.8k| return addressOf(/*eltIdx=*/numElts); 166| 13.8k|} _ZNK8rawspeed10Array1DRefIKNS_15DngSliceElementEE9addressOfEi: 133| 27.5k|Array1DRef::addressOf(const int eltIdx) const { 134| 27.5k| establishClassInvariants(); 135| 27.5k| invariant(eltIdx >= 0); ------------------ | | 27| 27.5k|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 27.3k, False: 162] ------------------ 136| 27.3k| invariant(eltIdx <= numElts); ------------------ | | 27| 27.3k|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 27.3k, False: 3] ------------------ 137| 27.3k|#pragma GCC diagnostic push 138| 27.3k|#pragma GCC diagnostic ignored "-Wpragmas" 139| 27.3k|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 140| 27.3k|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 141| 27.3k| return data + eltIdx; 142| 27.3k|#pragma GCC diagnostic pop 143| 27.3k|} _ZNK8rawspeed10Array1DRefIKNS_15DngSliceElementEE5beginEv: 159| 13.7k|template inline T* Array1DRef::begin() const { 160| 13.7k| establishClassInvariants(); 161| 13.7k| return addressOf(/*eltIdx=*/0); 162| 13.7k|} _ZNK8rawspeed10Array1DRefIKNS_8CFAColorEE24establishClassInvariantsEv: 97| 141k|Array1DRef::establishClassInvariants() const noexcept { 98| 141k| invariant(data); ------------------ | | 27| 141k|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 142k, False: 18.4E] ------------------ 99| 142k| invariant(numElts >= 0); ------------------ | | 27| 142k|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 142k, False: 18.4E] ------------------ 100| 142k|} _ZNK8rawspeed10Array1DRefIKNS_8CFAColorEE4sizeEv: 154| 78.2k|template inline int Array1DRef::size() const { 155| 78.2k| establishClassInvariants(); 156| 78.2k| return numElts; 157| 78.2k|} _ZNK8rawspeed10Array1DRefIKNS_8CFAColorEE7getCropEii: 110| 4.57k|Array1DRef::getCrop(int offset, int size) const { 111| 4.57k| establishClassInvariants(); 112| 4.57k| invariant(offset >= 0); ------------------ | | 27| 4.57k|#define invariant(expr) assert(expr) ------------------ | Branch (112:3): [True: 4.62k, False: 18.4E] ------------------ 113| 4.62k| invariant(size >= 0); ------------------ | | 27| 4.62k|#define invariant(expr) assert(expr) ------------------ | Branch (113:3): [True: 4.62k, False: 18.4E] ------------------ 114| 4.62k| invariant(offset <= numElts); ------------------ | | 27| 4.62k|#define invariant(expr) assert(expr) ------------------ | Branch (114:3): [True: 4.62k, False: 0] ------------------ 115| 4.62k| invariant(size <= numElts); ------------------ | | 27| 4.62k|#define invariant(expr) assert(expr) ------------------ | Branch (115:3): [True: 4.63k, False: 18.4E] ------------------ 116| 4.63k| invariant(offset + size <= numElts); ------------------ | | 27| 4.63k|#define invariant(expr) assert(expr) ------------------ | Branch (116:3): [True: 4.62k, False: 2] ------------------ 117| 4.62k| return {*this, offset, size}; 118| 4.63k|} _ZNK8rawspeed10Array1DRefIKNS_8CFAColorEE9addressOfEi: 133| 9.17k|Array1DRef::addressOf(const int eltIdx) const { 134| 9.17k| establishClassInvariants(); 135| 9.17k| invariant(eltIdx >= 0); ------------------ | | 27| 9.17k|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 9.13k, False: 31] ------------------ 136| 9.13k| invariant(eltIdx <= numElts); ------------------ | | 27| 9.13k|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 9.18k, False: 18.4E] ------------------ 137| 9.18k|#pragma GCC diagnostic push 138| 9.18k|#pragma GCC diagnostic ignored "-Wpragmas" 139| 9.18k|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 140| 9.18k|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 141| 9.18k| return data + eltIdx; 142| 9.13k|#pragma GCC diagnostic pop 143| 9.13k|} _ZNK8rawspeed10Array1DRefIKNS_8CFAColorEEclEi: 147| 4.62k|Array1DRef::operator()(const int eltIdx) const { 148| 4.62k| establishClassInvariants(); 149| 4.62k| invariant(eltIdx >= 0); ------------------ | | 27| 4.62k|#define invariant(expr) assert(expr) ------------------ | Branch (149:3): [True: 4.62k, False: 18.4E] ------------------ 150| 4.62k| invariant(eltIdx < numElts); ------------------ | | 27| 4.62k|#define invariant(expr) assert(expr) ------------------ | Branch (150:3): [True: 4.63k, False: 18.4E] ------------------ 151| 4.63k| return *addressOf(eltIdx); 152| 4.62k|} _ZNK8rawspeed10Array1DRefINS_8CFAColorEE24establishClassInvariantsEv: 97| 220k|Array1DRef::establishClassInvariants() const noexcept { 98| 220k| invariant(data); ------------------ | | 27| 220k|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 221k, False: 18.4E] ------------------ 99| 221k| invariant(numElts >= 0); ------------------ | | 27| 221k|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 222k, False: 18.4E] ------------------ 100| 221k|} _ZNK8rawspeed10Array1DRefINS_8CFAColorEE4sizeEv: 154| 121k|template inline int Array1DRef::size() const { 155| 121k| establishClassInvariants(); 156| 121k| return numElts; 157| 121k|} _ZNK8rawspeed10Array1DRefINS_8CFAColorEE7getCropEii: 110| 6.35k|Array1DRef::getCrop(int offset, int size) const { 111| 6.35k| establishClassInvariants(); 112| 6.35k| invariant(offset >= 0); ------------------ | | 27| 6.35k|#define invariant(expr) assert(expr) ------------------ | Branch (112:3): [True: 6.43k, False: 18.4E] ------------------ 113| 6.43k| invariant(size >= 0); ------------------ | | 27| 6.43k|#define invariant(expr) assert(expr) ------------------ | Branch (113:3): [True: 6.44k, False: 18.4E] ------------------ 114| 6.44k| invariant(offset <= numElts); ------------------ | | 27| 6.44k|#define invariant(expr) assert(expr) ------------------ | Branch (114:3): [True: 6.44k, False: 18.4E] ------------------ 115| 6.44k| invariant(size <= numElts); ------------------ | | 27| 6.44k|#define invariant(expr) assert(expr) ------------------ | Branch (115:3): [True: 6.44k, False: 18.4E] ------------------ 116| 6.44k| invariant(offset + size <= numElts); ------------------ | | 27| 6.44k|#define invariant(expr) assert(expr) ------------------ | Branch (116:3): [True: 6.44k, False: 0] ------------------ 117| 6.44k| return {*this, offset, size}; 118| 6.44k|} _ZNK8rawspeed10Array1DRefINS_8CFAColorEE9addressOfEi: 133| 12.7k|Array1DRef::addressOf(const int eltIdx) const { 134| 12.7k| establishClassInvariants(); 135| 12.7k| invariant(eltIdx >= 0); ------------------ | | 27| 12.7k|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 12.6k, False: 55] ------------------ 136| 12.6k| invariant(eltIdx <= numElts); ------------------ | | 27| 12.6k|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 12.7k, False: 18.4E] ------------------ 137| 12.7k|#pragma GCC diagnostic push 138| 12.7k|#pragma GCC diagnostic ignored "-Wpragmas" 139| 12.7k|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 140| 12.7k|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 141| 12.7k| return data + eltIdx; 142| 12.6k|#pragma GCC diagnostic pop 143| 12.6k|} _ZNK8rawspeed10Array1DRefINS_8CFAColorEEclEi: 147| 6.44k|Array1DRef::operator()(const int eltIdx) const { 148| 6.44k| establishClassInvariants(); 149| 6.44k| invariant(eltIdx >= 0); ------------------ | | 27| 6.44k|#define invariant(expr) assert(expr) ------------------ | Branch (149:3): [True: 6.46k, False: 18.4E] ------------------ 150| 6.46k| invariant(eltIdx < numElts); ------------------ | | 27| 6.46k|#define invariant(expr) assert(expr) ------------------ | Branch (150:3): [True: 6.44k, False: 27] ------------------ 151| 6.44k| return *addressOf(eltIdx); 152| 6.46k|} _ZN8rawspeed10Array1DRefINS_8CFAColorEEC2EPS1_i: 104| 8.03k| : data(data_), numElts(numElts_) { 105| 8.03k| establishClassInvariants(); 106| 8.03k|} _ZN8rawspeed10Array1DRefIKNS_8CFAColorEEC2EPS2_i: 104| 5.75k| : data(data_), numElts(numElts_) { 105| 5.75k| establishClassInvariants(); 106| 5.75k|} _ZN8rawspeed10Array1DRefIKNS_13PhaseOneStripEEC2EPS2_i: 104| 15.4k| : data(data_), numElts(numElts_) { 105| 15.4k| establishClassInvariants(); 106| 15.4k|} _ZNK8rawspeed10Array1DRefIKNS_13PhaseOneStripEE24establishClassInvariantsEv: 97| 74.9k|Array1DRef::establishClassInvariants() const noexcept { 98| 74.9k| invariant(data); ------------------ | | 27| 74.9k|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 75.0k, False: 18.4E] ------------------ 99| 75.0k| invariant(numElts >= 0); ------------------ | | 27| 75.0k|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 75.0k, False: 18.4E] ------------------ 100| 75.0k|} _ZNK8rawspeed10Array1DRefIKNS_13PhaseOneStripEE3endEv: 163| 15.3k|template inline T* Array1DRef::end() const { 164| 15.3k| establishClassInvariants(); 165| 15.3k| return addressOf(/*eltIdx=*/numElts); 166| 15.3k|} _ZNK8rawspeed10Array1DRefIKNS_13PhaseOneStripEE9addressOfEi: 133| 30.2k|Array1DRef::addressOf(const int eltIdx) const { 134| 30.2k| establishClassInvariants(); 135| 30.2k| invariant(eltIdx >= 0); ------------------ | | 27| 30.2k|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 30.0k, False: 240] ------------------ 136| 30.0k| invariant(eltIdx <= numElts); ------------------ | | 27| 30.0k|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 30.0k, False: 1] ------------------ 137| 30.0k|#pragma GCC diagnostic push 138| 30.0k|#pragma GCC diagnostic ignored "-Wpragmas" 139| 30.0k|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 140| 30.0k|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 141| 30.0k| return data + eltIdx; 142| 30.0k|#pragma GCC diagnostic pop 143| 30.0k|} _ZNK8rawspeed10Array1DRefIKNS_13PhaseOneStripEE5beginEv: 159| 15.1k|template inline T* Array1DRef::begin() const { 160| 15.1k| establishClassInvariants(); 161| 15.1k| return addressOf(/*eltIdx=*/0); 162| 15.1k|} _ZN8rawspeed10Array1DRefIKNS_23PanasonicV4Decompressor5BlockEEC2EPS3_i: 104| 4.77k| : data(data_), numElts(numElts_) { 105| 4.77k| establishClassInvariants(); 106| 4.77k|} _ZNK8rawspeed10Array1DRefIKNS_23PanasonicV4Decompressor5BlockEE24establishClassInvariantsEv: 97| 23.5k|Array1DRef::establishClassInvariants() const noexcept { 98| 23.5k| invariant(data); ------------------ | | 27| 23.5k|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 23.6k, False: 18.4E] ------------------ 99| 23.6k| invariant(numElts >= 0); ------------------ | | 27| 23.6k|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 23.6k, False: 15] ------------------ 100| 23.6k|} _ZNK8rawspeed10Array1DRefIKNS_23PanasonicV4Decompressor5BlockEE3endEv: 163| 4.76k|template inline T* Array1DRef::end() const { 164| 4.76k| establishClassInvariants(); 165| 4.76k| return addressOf(/*eltIdx=*/numElts); 166| 4.76k|} _ZNK8rawspeed10Array1DRefIKNS_23PanasonicV4Decompressor5BlockEE9addressOfEi: 133| 9.49k|Array1DRef::addressOf(const int eltIdx) const { 134| 9.49k| establishClassInvariants(); 135| 9.49k| invariant(eltIdx >= 0); ------------------ | | 27| 9.49k|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 9.49k, False: 3] ------------------ 136| 9.49k| invariant(eltIdx <= numElts); ------------------ | | 27| 9.49k|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 9.48k, False: 9] ------------------ 137| 9.48k|#pragma GCC diagnostic push 138| 9.48k|#pragma GCC diagnostic ignored "-Wpragmas" 139| 9.48k|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 140| 9.48k|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 141| 9.48k| return data + eltIdx; 142| 9.49k|#pragma GCC diagnostic pop 143| 9.49k|} _ZNK8rawspeed10Array1DRefIKNS_23PanasonicV4Decompressor5BlockEE5beginEv: 159| 4.74k|template inline T* Array1DRef::begin() const { 160| 4.74k| establishClassInvariants(); 161| 4.74k| return addressOf(/*eltIdx=*/0); 162| 4.74k|} _ZN8rawspeed10Array1DRefIKNS_23PanasonicV5Decompressor5BlockEEC2EPS3_i: 104| 2.44k| : data(data_), numElts(numElts_) { 105| 2.44k| establishClassInvariants(); 106| 2.44k|} _ZNK8rawspeed10Array1DRefIKNS_23PanasonicV5Decompressor5BlockEE24establishClassInvariantsEv: 97| 11.9k|Array1DRef::establishClassInvariants() const noexcept { 98| 11.9k| invariant(data); ------------------ | | 27| 11.9k|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 11.9k, False: 18.4E] ------------------ 99| 11.9k| invariant(numElts >= 0); ------------------ | | 27| 11.9k|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 11.9k, False: 18.4E] ------------------ 100| 11.9k|} _ZNK8rawspeed10Array1DRefIKNS_23PanasonicV5Decompressor5BlockEE3endEv: 163| 2.43k|template inline T* Array1DRef::end() const { 164| 2.43k| establishClassInvariants(); 165| 2.43k| return addressOf(/*eltIdx=*/numElts); 166| 2.43k|} _ZNK8rawspeed10Array1DRefIKNS_23PanasonicV5Decompressor5BlockEE9addressOfEi: 133| 4.82k|Array1DRef::addressOf(const int eltIdx) const { 134| 4.82k| establishClassInvariants(); 135| 4.82k| invariant(eltIdx >= 0); ------------------ | | 27| 4.82k|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 4.82k, False: 6] ------------------ 136| 4.82k| invariant(eltIdx <= numElts); ------------------ | | 27| 4.82k|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 4.81k, False: 4] ------------------ 137| 4.81k|#pragma GCC diagnostic push 138| 4.81k|#pragma GCC diagnostic ignored "-Wpragmas" 139| 4.81k|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 140| 4.81k|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 141| 4.81k| return data + eltIdx; 142| 4.82k|#pragma GCC diagnostic pop 143| 4.82k|} _ZNK8rawspeed10Array1DRefIKNS_23PanasonicV5Decompressor5BlockEE5beginEv: 159| 2.42k|template inline T* Array1DRef::begin() const { 160| 2.42k| establishClassInvariants(); 161| 2.42k| return addressOf(/*eltIdx=*/0); 162| 2.42k|} _ZNK8rawspeed10Array2DRefIKSt4byteE5widthEv: 165| 22.2k|__attribute__((always_inline)) inline int Array2DRef::width() const { 166| 22.2k| establishClassInvariants(); 167| 22.2k| return _width; 168| 22.2k|} _ZNK8rawspeed10Array2DRefIKSt4byteE24establishClassInvariantsEv: 127| 86.5k|Array2DRef::establishClassInvariants() const noexcept { 128| 86.5k| data.establishClassInvariants(); 129| 86.5k| invariant(_width >= 0); ------------------ | | 27| 86.5k|#define invariant(expr) assert(expr) ------------------ | Branch (129:3): [True: 86.5k, False: 3] ------------------ 130| 86.5k| invariant(_height >= 0); ------------------ | | 27| 86.5k|#define invariant(expr) assert(expr) ------------------ | Branch (130:3): [True: 86.5k, False: 0] ------------------ 131| 86.5k| invariant(_pitch != 0); ------------------ | | 27| 86.5k|#define invariant(expr) assert(expr) ------------------ | Branch (131:3): [True: 86.5k, False: 0] ------------------ 132| 86.5k| invariant(_pitch >= 0); ------------------ | | 27| 86.5k|#define invariant(expr) assert(expr) ------------------ | Branch (132:3): [True: 86.5k, False: 18.4E] ------------------ 133| 86.5k| invariant(_pitch >= _width); ------------------ | | 27| 86.5k|#define invariant(expr) assert(expr) ------------------ | Branch (133:3): [True: 86.5k, False: 0] ------------------ 134| 86.5k| invariant((_width == 0) == (_height == 0)); ------------------ | | 27| 86.5k|#define invariant(expr) assert(expr) ------------------ | Branch (134:3): [True: 86.5k, False: 0] ------------------ 135| 86.5k| invariant(data.size() == _pitch * _height); ------------------ | | 27| 86.5k|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 86.5k, False: 5] ------------------ 136| 86.5k|} _ZNK8rawspeed10Array2DRefIKSt4byteE6heightEv: 171| 42.5k|__attribute__((always_inline)) inline int Array2DRef::height() const { 172| 42.5k| establishClassInvariants(); 173| 42.5k| return _height; 174| 42.5k|} _ZNK8rawspeed10Array2DRefISt4byteE5widthEv: 165| 18.7M|__attribute__((always_inline)) inline int Array2DRef::width() const { 166| 18.7M| establishClassInvariants(); 167| 18.7M| return _width; 168| 18.7M|} _ZNK8rawspeed10Array2DRefISt4byteE24establishClassInvariantsEv: 127| 46.9M|Array2DRef::establishClassInvariants() const noexcept { 128| 46.9M| data.establishClassInvariants(); 129| 46.9M| invariant(_width >= 0); ------------------ | | 27| 46.9M|#define invariant(expr) assert(expr) ------------------ | Branch (129:3): [True: 46.9M, False: 18.4E] ------------------ 130| 46.9M| invariant(_height >= 0); ------------------ | | 27| 46.9M|#define invariant(expr) assert(expr) ------------------ | Branch (130:3): [True: 46.9M, False: 0] ------------------ 131| 46.9M| invariant(_pitch != 0); ------------------ | | 27| 46.9M|#define invariant(expr) assert(expr) ------------------ | Branch (131:3): [True: 46.9M, False: 0] ------------------ 132| 46.9M| invariant(_pitch >= 0); ------------------ | | 27| 46.9M|#define invariant(expr) assert(expr) ------------------ | Branch (132:3): [True: 46.9M, False: 1] ------------------ 133| 46.9M| invariant(_pitch >= _width); ------------------ | | 27| 46.9M|#define invariant(expr) assert(expr) ------------------ | Branch (133:3): [True: 46.9M, False: 0] ------------------ 134| 46.9M| invariant((_width == 0) == (_height == 0)); ------------------ | | 27| 46.9M|#define invariant(expr) assert(expr) ------------------ | Branch (134:3): [True: 46.9M, False: 18.4E] ------------------ 135| 46.9M| invariant(data.size() == _pitch * _height); ------------------ | | 27| 46.9M|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 46.9M, False: 18.4E] ------------------ 136| 46.9M|} _ZNK8rawspeed10Array2DRefISt4byteE6heightEv: 171| 13.0M|__attribute__((always_inline)) inline int Array2DRef::height() const { 172| 13.0M| establishClassInvariants(); 173| 13.0M| return _height; 174| 13.0M|} _ZNK8rawspeed10Array2DRefISt4byteE15getAsArray1DRefEv: 178| 706|Array2DRef::getAsArray1DRef() const { 179| 706| establishClassInvariants(); 180| 706| if (height() == 1 || _pitch == width()) ------------------ | Branch (180:7): [True: 164, False: 542] | Branch (180:24): [True: 32, False: 510] ------------------ 181| 196| return data.getCrop(/*offset=*/0, width() * height()).getAsArray1DRef(); 182| 510| return std::nullopt; 183| 706|} _ZNK8rawspeed10Array2DRefIKSt4byteE15getAsArray1DRefEv: 178| 706|Array2DRef::getAsArray1DRef() const { 179| 706| establishClassInvariants(); 180| 706| if (height() == 1 || _pitch == width()) ------------------ | Branch (180:7): [True: 166, False: 540] | Branch (180:24): [True: 534, False: 6] ------------------ 181| 700| return data.getCrop(/*offset=*/0, width() * height()).getAsArray1DRef(); 182| 6| return std::nullopt; 183| 706|} _ZNK8rawspeed10Array2DRefISt4byteEixEi: 186| 7.53M|inline Array1DRef Array2DRef::operator[](const int row) const { 187| 7.53M| establishClassInvariants(); 188| 7.53M| invariant(row >= 0); ------------------ | | 27| 7.53M|#define invariant(expr) assert(expr) ------------------ | Branch (188:3): [True: 7.53M, False: 18.4E] ------------------ 189| 7.53M| invariant(row < height()); ------------------ | | 27| 7.53M|#define invariant(expr) assert(expr) ------------------ | Branch (189:3): [True: 7.53M, False: 2] ------------------ 190| 7.53M| return data.getCrop(row * _pitch, width()).getAsArray1DRef(); 191| 7.53M|} _ZNK8rawspeed10Array2DRefIKSt4byteEixEi: 186| 19.6k|inline Array1DRef Array2DRef::operator[](const int row) const { 187| 19.6k| establishClassInvariants(); 188| 19.6k| invariant(row >= 0); ------------------ | | 27| 19.6k|#define invariant(expr) assert(expr) ------------------ | Branch (188:3): [True: 19.6k, False: 2] ------------------ 189| 19.6k| invariant(row < height()); ------------------ | | 27| 19.6k|#define invariant(expr) assert(expr) ------------------ | Branch (189:3): [True: 19.6k, False: 18.4E] ------------------ 190| 19.6k| return data.getCrop(row * _pitch, width()).getAsArray1DRef(); 191| 19.6k|} _ZNK8rawspeed10Array2DRefItE24establishClassInvariantsEv: 127| 200M|Array2DRef::establishClassInvariants() const noexcept { 128| 200M| data.establishClassInvariants(); 129| 200M| invariant(_width >= 0); ------------------ | | 27| 200M|#define invariant(expr) assert(expr) ------------------ | Branch (129:3): [True: 207M, False: 18.4E] ------------------ 130| 207M| invariant(_height >= 0); ------------------ | | 27| 207M|#define invariant(expr) assert(expr) ------------------ | Branch (130:3): [True: 209M, False: 18.4E] ------------------ 131| 209M| invariant(_pitch != 0); ------------------ | | 27| 209M|#define invariant(expr) assert(expr) ------------------ | Branch (131:3): [True: 209M, False: 18.4E] ------------------ 132| 209M| invariant(_pitch >= 0); ------------------ | | 27| 209M|#define invariant(expr) assert(expr) ------------------ | Branch (132:3): [True: 209M, False: 18.4E] ------------------ 133| 209M| invariant(_pitch >= _width); ------------------ | | 27| 209M|#define invariant(expr) assert(expr) ------------------ | Branch (133:3): [True: 209M, False: 255k] ------------------ 134| 209M| invariant((_width == 0) == (_height == 0)); ------------------ | | 27| 209M|#define invariant(expr) assert(expr) ------------------ | Branch (134:3): [True: 212M, False: 18.4E] ------------------ 135| 212M| invariant(data.size() == _pitch * _height); ------------------ | | 27| 212M|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 211M, False: 650k] ------------------ 136| 212M|} _ZNK8rawspeed10Array2DRefItE5widthEv: 165| 92.9M|__attribute__((always_inline)) inline int Array2DRef::width() const { 166| 92.9M| establishClassInvariants(); 167| 92.9M| return _width; 168| 92.9M|} _ZNK8rawspeed10Array2DRefItE6heightEv: 171| 57.5M|__attribute__((always_inline)) inline int Array2DRef::height() const { 172| 57.5M| establishClassInvariants(); 173| 57.5M| return _height; 174| 57.5M|} _ZNK8rawspeed10Array2DRefIfE24establishClassInvariantsEv: 127| 3.23M|Array2DRef::establishClassInvariants() const noexcept { 128| 3.23M| data.establishClassInvariants(); 129| 3.23M| invariant(_width >= 0); ------------------ | | 27| 3.23M|#define invariant(expr) assert(expr) ------------------ | Branch (129:3): [True: 3.23M, False: 0] ------------------ 130| 3.23M| invariant(_height >= 0); ------------------ | | 27| 3.23M|#define invariant(expr) assert(expr) ------------------ | Branch (130:3): [True: 3.23M, False: 0] ------------------ 131| 3.23M| invariant(_pitch != 0); ------------------ | | 27| 3.23M|#define invariant(expr) assert(expr) ------------------ | Branch (131:3): [True: 3.23M, False: 0] ------------------ 132| 3.23M| invariant(_pitch >= 0); ------------------ | | 27| 3.23M|#define invariant(expr) assert(expr) ------------------ | Branch (132:3): [True: 3.23M, False: 0] ------------------ 133| 3.23M| invariant(_pitch >= _width); ------------------ | | 27| 3.23M|#define invariant(expr) assert(expr) ------------------ | Branch (133:3): [True: 3.23M, False: 0] ------------------ 134| 3.23M| invariant((_width == 0) == (_height == 0)); ------------------ | | 27| 3.23M|#define invariant(expr) assert(expr) ------------------ | Branch (134:3): [True: 3.23M, False: 0] ------------------ 135| 3.23M| invariant(data.size() == _pitch * _height); ------------------ | | 27| 3.23M|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 3.23M, False: 18.4E] ------------------ 136| 3.23M|} _ZNK8rawspeed10Array2DRefIfE5widthEv: 165| 1.29M|__attribute__((always_inline)) inline int Array2DRef::width() const { 166| 1.29M| establishClassInvariants(); 167| 1.29M| return _width; 168| 1.29M|} _ZNK8rawspeed10Array2DRefIfE6heightEv: 171| 788k|__attribute__((always_inline)) inline int Array2DRef::height() const { 172| 788k| establishClassInvariants(); 173| 788k| return _height; 174| 788k|} _ZN8rawspeed10Array2DRefIiEC2EPiii: 154| 1.99k| : Array2DRef(data_, width_, height_, /*pitch=*/width_) { 155| 1.99k| establishClassInvariants(); 156| 1.99k|} _ZN8rawspeed10Array2DRefIiEC2EPiiii: 148| 1.99k| : Array2DRef({data_, pitch_ * height_}, width_, height_, pitch_) { 149| 1.99k| establishClassInvariants(); 150| 1.99k|} _ZN8rawspeed10Array2DRefIiEC2ENS_10Array1DRefIiEEiii: 141| 1.99k| : data(data_), _pitch(pitch_), _width(width_), _height(height_) { 142| 1.99k| establishClassInvariants(); 143| 1.99k|} _ZNK8rawspeed10Array2DRefIiE24establishClassInvariantsEv: 127| 16.5k|Array2DRef::establishClassInvariants() const noexcept { 128| 16.5k| data.establishClassInvariants(); 129| 16.5k| invariant(_width >= 0); ------------------ | | 27| 16.5k|#define invariant(expr) assert(expr) ------------------ | Branch (129:3): [True: 16.5k, False: 0] ------------------ 130| 16.5k| invariant(_height >= 0); ------------------ | | 27| 16.5k|#define invariant(expr) assert(expr) ------------------ | Branch (130:3): [True: 16.5k, False: 0] ------------------ 131| 16.5k| invariant(_pitch != 0); ------------------ | | 27| 16.5k|#define invariant(expr) assert(expr) ------------------ | Branch (131:3): [True: 16.5k, False: 0] ------------------ 132| 16.5k| invariant(_pitch >= 0); ------------------ | | 27| 16.5k|#define invariant(expr) assert(expr) ------------------ | Branch (132:3): [True: 16.5k, False: 0] ------------------ 133| 16.5k| invariant(_pitch >= _width); ------------------ | | 27| 16.5k|#define invariant(expr) assert(expr) ------------------ | Branch (133:3): [True: 16.5k, False: 0] ------------------ 134| 16.5k| invariant((_width == 0) == (_height == 0)); ------------------ | | 27| 16.5k|#define invariant(expr) assert(expr) ------------------ | Branch (134:3): [True: 16.5k, False: 0] ------------------ 135| 16.5k| invariant(data.size() == _pitch * _height); ------------------ | | 27| 16.5k|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 16.5k, False: 0] ------------------ 136| 16.5k|} _ZNK8rawspeed10Array2DRefIiE15getAsArray1DRefEv: 178| 2.11k|Array2DRef::getAsArray1DRef() const { 179| 2.11k| establishClassInvariants(); 180| 2.11k| if (height() == 1 || _pitch == width()) ------------------ | Branch (180:7): [True: 0, False: 2.11k] | Branch (180:24): [True: 2.11k, False: 0] ------------------ 181| 2.11k| return data.getCrop(/*offset=*/0, width() * height()).getAsArray1DRef(); 182| 0| return std::nullopt; 183| 2.11k|} _ZNK8rawspeed10Array2DRefIiE6heightEv: 171| 4.23k|__attribute__((always_inline)) inline int Array2DRef::height() const { 172| 4.23k| establishClassInvariants(); 173| 4.23k| return _height; 174| 4.23k|} _ZNK8rawspeed10Array2DRefIiE5widthEv: 165| 4.23k|__attribute__((always_inline)) inline int Array2DRef::width() const { 166| 4.23k| establishClassInvariants(); 167| 4.23k| return _width; 168| 4.23k|} _ZN8rawspeed10Array2DRefItEC2EPtiii: 148| 4.07M| : Array2DRef({data_, pitch_ * height_}, width_, height_, pitch_) { 149| 4.07M| establishClassInvariants(); 150| 4.07M|} _ZN8rawspeed10Array2DRefItEC2ENS_10Array1DRefItEEiii: 141| 4.08M| : data(data_), _pitch(pitch_), _width(width_), _height(height_) { 142| 4.08M| establishClassInvariants(); 143| 4.08M|} _ZN8rawspeed10Array2DRefISt4byteEC2ItQaaaantaasr3stdE10is_const_vITL0__Entsr3stdE10is_const_vIT_Entsr3stdE9is_same_vIu14__remove_constIS5_Eu14__remove_constIS4_EEsr3stdE9is_same_vIS6_S1_EEENS0_IS5_EE: 92| 10.4k| : Array2DRef(RHS.data, sizeof(T2) * RHS._width, RHS._height, 93| 10.4k| sizeof(T2) * RHS._pitch) {} _ZN8rawspeed10Array2DRefISt4byteEC2ENS_10Array1DRefIS1_EEiii: 141| 11.7k| : data(data_), _pitch(pitch_), _width(width_), _height(height_) { 142| 11.7k| establishClassInvariants(); 143| 11.7k|} _ZN8rawspeed10Array2DRefIfEC2EPfiii: 148| 1.70k| : Array2DRef({data_, pitch_ * height_}, width_, height_, pitch_) { 149| 1.70k| establishClassInvariants(); 150| 1.70k|} _ZN8rawspeed10Array2DRefIfEC2ENS_10Array1DRefIfEEiii: 141| 1.70k| : data(data_), _pitch(pitch_), _width(width_), _height(height_) { 142| 1.70k| establishClassInvariants(); 143| 1.70k|} _ZN8rawspeed10Array2DRefISt4byteEC2IfQaaaantaasr3stdE10is_const_vITL0__Entsr3stdE10is_const_vIT_Entsr3stdE9is_same_vIu14__remove_constIS5_Eu14__remove_constIS4_EEsr3stdE9is_same_vIS6_S1_EEENS0_IS5_EE: 92| 545| : Array2DRef(RHS.data, sizeof(T2) * RHS._width, RHS._height, 93| 545| sizeof(T2) * RHS._pitch) {} _ZNK8rawspeed10Array2DRefISt4byteEclEii: 195| 5.66M|Array2DRef::operator()(const int row, const int col) const { 196| 5.66M| establishClassInvariants(); 197| 5.66M| invariant(col >= 0); ------------------ | | 27| 5.66M|#define invariant(expr) assert(expr) ------------------ | Branch (197:3): [True: 5.66M, False: 0] ------------------ 198| 5.66M| invariant(col < width()); ------------------ | | 27| 5.66M|#define invariant(expr) assert(expr) ------------------ | Branch (198:3): [True: 5.66M, False: 0] ------------------ 199| 5.66M| return (operator[](row))(col); 200| 5.66M|} _ZNK8rawspeed10Array2DRefItEixEi: 186| 44.0M|inline Array1DRef Array2DRef::operator[](const int row) const { 187| 44.0M| establishClassInvariants(); 188| 44.0M| invariant(row >= 0); ------------------ | | 27| 44.0M|#define invariant(expr) assert(expr) ------------------ | Branch (188:3): [True: 45.3M, False: 18.4E] ------------------ 189| 45.3M| invariant(row < height()); ------------------ | | 27| 45.3M|#define invariant(expr) assert(expr) ------------------ | Branch (189:3): [True: 42.3M, False: 3.00M] ------------------ 190| 42.3M| return data.getCrop(row * _pitch, width()).getAsArray1DRef(); 191| 45.3M|} _ZN8rawspeed10Array2DRefItEC2EPtii: 154| 2.85k| : Array2DRef(data_, width_, height_, /*pitch=*/width_) { 155| 2.85k| establishClassInvariants(); 156| 2.85k|} _ZNK8rawspeed10Array2DRefItEclEii: 195| 41.8M|Array2DRef::operator()(const int row, const int col) const { 196| 41.8M| establishClassInvariants(); 197| 41.8M| invariant(col >= 0); ------------------ | | 27| 41.8M|#define invariant(expr) assert(expr) ------------------ | Branch (197:3): [True: 41.2M, False: 642k] ------------------ 198| 41.2M| invariant(col < width()); ------------------ | | 27| 41.2M|#define invariant(expr) assert(expr) ------------------ | Branch (198:3): [True: 41.9M, False: 18.4E] ------------------ 199| 41.9M| return (operator[](row))(col); 200| 41.2M|} _ZN8rawspeed10Array2DRefIKhEC2EPS1_ii: 154| 331| : Array2DRef(data_, width_, height_, /*pitch=*/width_) { 155| 331| establishClassInvariants(); 156| 331|} _ZN8rawspeed10Array2DRefIKhEC2EPS1_iii: 148| 331| : Array2DRef({data_, pitch_ * height_}, width_, height_, pitch_) { 149| 331| establishClassInvariants(); 150| 331|} _ZN8rawspeed10Array2DRefIKhEC2ENS_10Array1DRefIS1_EEiii: 141| 331| : data(data_), _pitch(pitch_), _width(width_), _height(height_) { 142| 331| establishClassInvariants(); 143| 331|} _ZNK8rawspeed10Array2DRefIKhE24establishClassInvariantsEv: 127| 13.6M|Array2DRef::establishClassInvariants() const noexcept { 128| 13.6M| data.establishClassInvariants(); 129| 13.6M| invariant(_width >= 0); ------------------ | | 27| 13.6M|#define invariant(expr) assert(expr) ------------------ | Branch (129:3): [True: 13.6M, False: 0] ------------------ 130| 13.6M| invariant(_height >= 0); ------------------ | | 27| 13.6M|#define invariant(expr) assert(expr) ------------------ | Branch (130:3): [True: 13.6M, False: 0] ------------------ 131| 13.6M| invariant(_pitch != 0); ------------------ | | 27| 13.6M|#define invariant(expr) assert(expr) ------------------ | Branch (131:3): [True: 13.6M, False: 0] ------------------ 132| 13.6M| invariant(_pitch >= 0); ------------------ | | 27| 13.6M|#define invariant(expr) assert(expr) ------------------ | Branch (132:3): [True: 13.6M, False: 0] ------------------ 133| 13.6M| invariant(_pitch >= _width); ------------------ | | 27| 13.6M|#define invariant(expr) assert(expr) ------------------ | Branch (133:3): [True: 13.6M, False: 0] ------------------ 134| 13.6M| invariant((_width == 0) == (_height == 0)); ------------------ | | 27| 13.6M|#define invariant(expr) assert(expr) ------------------ | Branch (134:3): [True: 13.6M, False: 0] ------------------ 135| 13.6M| invariant(data.size() == _pitch * _height); ------------------ | | 27| 13.6M|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 13.6M, False: 0] ------------------ 136| 13.6M|} _ZNK8rawspeed10Array2DRefIKhEclEii: 195| 2.72M|Array2DRef::operator()(const int row, const int col) const { 196| 2.72M| establishClassInvariants(); 197| 2.72M| invariant(col >= 0); ------------------ | | 27| 2.72M|#define invariant(expr) assert(expr) ------------------ | Branch (197:3): [True: 2.72M, False: 0] ------------------ 198| 2.72M| invariant(col < width()); ------------------ | | 27| 2.72M|#define invariant(expr) assert(expr) ------------------ | Branch (198:3): [True: 2.72M, False: 0] ------------------ 199| 2.72M| return (operator[](row))(col); 200| 2.72M|} _ZNK8rawspeed10Array2DRefIKhE5widthEv: 165| 5.44M|__attribute__((always_inline)) inline int Array2DRef::width() const { 166| 5.44M| establishClassInvariants(); 167| 5.44M| return _width; 168| 5.44M|} _ZNK8rawspeed10Array2DRefIKhEixEi: 186| 2.72M|inline Array1DRef Array2DRef::operator[](const int row) const { 187| 2.72M| establishClassInvariants(); 188| 2.72M| invariant(row >= 0); ------------------ | | 27| 2.72M|#define invariant(expr) assert(expr) ------------------ | Branch (188:3): [True: 2.72M, False: 0] ------------------ 189| 2.72M| invariant(row < height()); ------------------ | | 27| 2.72M|#define invariant(expr) assert(expr) ------------------ | Branch (189:3): [True: 2.72M, False: 0] ------------------ 190| 2.72M| return data.getCrop(row * _pitch, width()).getAsArray1DRef(); 191| 2.72M|} _ZNK8rawspeed10Array2DRefIKhE6heightEv: 171| 2.72M|__attribute__((always_inline)) inline int Array2DRef::height() const { 172| 2.72M| establishClassInvariants(); 173| 2.72M| return _height; 174| 2.72M|} _ZN8rawspeed10Array2DRefISt4byteEC2EPS1_iii: 148| 706| : Array2DRef({data_, pitch_ * height_}, width_, height_, pitch_) { 149| 706| establishClassInvariants(); 150| 706|} _ZN8rawspeed10Array2DRefIKSt4byteEC2EPS2_iii: 148| 707| : Array2DRef({data_, pitch_ * height_}, width_, height_, pitch_) { 149| 707| establishClassInvariants(); 150| 707|} _ZN8rawspeed10Array2DRefIKSt4byteEC2ENS_10Array1DRefIS2_EEiii: 141| 707| : data(data_), _pitch(pitch_), _width(width_), _height(height_) { 142| 707| establishClassInvariants(); 143| 707|} _ZNK8rawspeed10Array2DRefIfEclEii: 195| 511k|Array2DRef::operator()(const int row, const int col) const { 196| 511k| establishClassInvariants(); 197| 511k| invariant(col >= 0); ------------------ | | 27| 511k|#define invariant(expr) assert(expr) ------------------ | Branch (197:3): [True: 511k, False: 1] ------------------ 198| 511k| invariant(col < width()); ------------------ | | 27| 511k|#define invariant(expr) assert(expr) ------------------ | Branch (198:3): [True: 511k, False: 0] ------------------ 199| 511k| return (operator[](row))(col); 200| 511k|} _ZNK8rawspeed10Array2DRefIfEixEi: 186| 551k|inline Array1DRef Array2DRef::operator[](const int row) const { 187| 551k| establishClassInvariants(); 188| 551k| invariant(row >= 0); ------------------ | | 27| 551k|#define invariant(expr) assert(expr) ------------------ | Branch (188:3): [True: 551k, False: 0] ------------------ 189| 551k| invariant(row < height()); ------------------ | | 27| 551k|#define invariant(expr) assert(expr) ------------------ | Branch (189:3): [True: 551k, False: 0] ------------------ 190| 551k| return data.getCrop(row * _pitch, width()).getAsArray1DRef(); 191| 551k|} _ZNK8rawspeed10Array2DRefItE5pitchEv: 159| 1.29M|__attribute__((always_inline)) inline int Array2DRef::pitch() const { 160| 1.29M| establishClassInvariants(); 161| 1.29M| return _pitch; 162| 1.29M|} _ZNK8rawspeed10Array2DRefIKtE5widthEv: 165| 7.43M|__attribute__((always_inline)) inline int Array2DRef::width() const { 166| 7.43M| establishClassInvariants(); 167| 7.43M| return _width; 168| 7.43M|} _ZNK8rawspeed10Array2DRefIKtE24establishClassInvariantsEv: 127| 19.4M|Array2DRef::establishClassInvariants() const noexcept { 128| 19.4M| data.establishClassInvariants(); 129| 19.4M| invariant(_width >= 0); ------------------ | | 27| 19.4M|#define invariant(expr) assert(expr) ------------------ | Branch (129:3): [True: 19.5M, False: 18.4E] ------------------ 130| 19.5M| invariant(_height >= 0); ------------------ | | 27| 19.5M|#define invariant(expr) assert(expr) ------------------ | Branch (130:3): [True: 19.5M, False: 18.4E] ------------------ 131| 19.5M| invariant(_pitch != 0); ------------------ | | 27| 19.5M|#define invariant(expr) assert(expr) ------------------ | Branch (131:3): [True: 19.5M, False: 18.4E] ------------------ 132| 19.5M| invariant(_pitch >= 0); ------------------ | | 27| 19.5M|#define invariant(expr) assert(expr) ------------------ | Branch (132:3): [True: 19.5M, False: 18.4E] ------------------ 133| 19.5M| invariant(_pitch >= _width); ------------------ | | 27| 19.5M|#define invariant(expr) assert(expr) ------------------ | Branch (133:3): [True: 19.6M, False: 18.4E] ------------------ 134| 19.6M| invariant((_width == 0) == (_height == 0)); ------------------ | | 27| 19.6M|#define invariant(expr) assert(expr) ------------------ | Branch (134:3): [True: 19.5M, False: 62.1k] ------------------ 135| 19.5M| invariant(data.size() == _pitch * _height); ------------------ | | 27| 19.5M|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 19.4M, False: 67.3k] ------------------ 136| 19.5M|} _ZNK8rawspeed10Array2DRefIKtE6heightEv: 171| 3.74M|__attribute__((always_inline)) inline int Array2DRef::height() const { 172| 3.74M| establishClassInvariants(); 173| 3.74M| return _height; 174| 3.74M|} _ZNK8rawspeed10Array2DRefIKtEclEii: 195| 3.71M|Array2DRef::operator()(const int row, const int col) const { 196| 3.71M| establishClassInvariants(); 197| 3.71M| invariant(col >= 0); ------------------ | | 27| 3.71M|#define invariant(expr) assert(expr) ------------------ | Branch (197:3): [True: 3.71M, False: 3.71k] ------------------ 198| 3.71M| invariant(col < width()); ------------------ | | 27| 3.71M|#define invariant(expr) assert(expr) ------------------ | Branch (198:3): [True: 3.71M, False: 283] ------------------ 199| 3.71M| return (operator[](row))(col); 200| 3.71M|} _ZNK8rawspeed10Array2DRefIKtEixEi: 186| 3.71M|inline Array1DRef Array2DRef::operator[](const int row) const { 187| 3.71M| establishClassInvariants(); 188| 3.71M| invariant(row >= 0); ------------------ | | 27| 3.71M|#define invariant(expr) assert(expr) ------------------ | Branch (188:3): [True: 3.72M, False: 18.4E] ------------------ 189| 3.72M| invariant(row < height()); ------------------ | | 27| 3.72M|#define invariant(expr) assert(expr) ------------------ | Branch (189:3): [True: 3.71M, False: 7.46k] ------------------ 190| 3.71M| return data.getCrop(row * _pitch, width()).getAsArray1DRef(); 191| 3.72M|} _ZN8rawspeed10Array2DRefIKtEC2ItQaaaantsr3stdE10is_const_vITL0__Esr3stdE10is_const_vIT_Esr3stdE9is_same_vIu14__remove_constIS5_Eu14__remove_constIS4_EEEENS0_IS5_EE: 83| 1.27M| : Array2DRef(RHS.data, RHS._width, RHS._height, RHS._pitch) {} _ZN8rawspeed10Array2DRefIKtEC2ENS_10Array1DRefIS1_EEiii: 141| 1.27M| : data(data_), _pitch(pitch_), _width(width_), _height(height_) { 142| 1.27M| establishClassInvariants(); 143| 1.27M|} _ZNK8rawspeed10Array2DRefIKNS_8CFAColorEE24establishClassInvariantsEv: 127| 25.8k|Array2DRef::establishClassInvariants() const noexcept { 128| 25.8k| data.establishClassInvariants(); 129| 25.8k| invariant(_width >= 0); ------------------ | | 27| 25.8k|#define invariant(expr) assert(expr) ------------------ | Branch (129:3): [True: 25.9k, False: 18.4E] ------------------ 130| 25.9k| invariant(_height >= 0); ------------------ | | 27| 25.9k|#define invariant(expr) assert(expr) ------------------ | Branch (130:3): [True: 25.8k, False: 54] ------------------ 131| 25.8k| invariant(_pitch != 0); ------------------ | | 27| 25.8k|#define invariant(expr) assert(expr) ------------------ | Branch (131:3): [True: 25.8k, False: 18.4E] ------------------ 132| 25.8k| invariant(_pitch >= 0); ------------------ | | 27| 25.8k|#define invariant(expr) assert(expr) ------------------ | Branch (132:3): [True: 25.9k, False: 18.4E] ------------------ 133| 25.9k| invariant(_pitch >= _width); ------------------ | | 27| 25.9k|#define invariant(expr) assert(expr) ------------------ | Branch (133:3): [True: 25.9k, False: 18.4E] ------------------ 134| 25.9k| invariant((_width == 0) == (_height == 0)); ------------------ | | 27| 25.9k|#define invariant(expr) assert(expr) ------------------ | Branch (134:3): [True: 25.9k, False: 18.4E] ------------------ 135| 25.9k| invariant(data.size() == _pitch * _height); ------------------ | | 27| 25.9k|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 25.9k, False: 18.4E] ------------------ 136| 25.9k|} _ZNK8rawspeed10Array2DRefIKNS_8CFAColorEEclEii: 195| 4.63k|Array2DRef::operator()(const int row, const int col) const { 196| 4.63k| establishClassInvariants(); 197| 4.63k| invariant(col >= 0); ------------------ | | 27| 4.63k|#define invariant(expr) assert(expr) ------------------ | Branch (197:3): [True: 4.59k, False: 46] ------------------ 198| 4.59k| invariant(col < width()); ------------------ | | 27| 4.59k|#define invariant(expr) assert(expr) ------------------ | Branch (198:3): [True: 4.60k, False: 18.4E] ------------------ 199| 4.60k| return (operator[](row))(col); 200| 4.59k|} _ZNK8rawspeed10Array2DRefIKNS_8CFAColorEE5widthEv: 165| 9.12k|__attribute__((always_inline)) inline int Array2DRef::width() const { 166| 9.12k| establishClassInvariants(); 167| 9.12k| return _width; 168| 9.12k|} _ZNK8rawspeed10Array2DRefIKNS_8CFAColorEEixEi: 186| 4.59k|inline Array1DRef Array2DRef::operator[](const int row) const { 187| 4.59k| establishClassInvariants(); 188| 4.59k| invariant(row >= 0); ------------------ | | 27| 4.59k|#define invariant(expr) assert(expr) ------------------ | Branch (188:3): [True: 4.63k, False: 18.4E] ------------------ 189| 4.63k| invariant(row < height()); ------------------ | | 27| 4.63k|#define invariant(expr) assert(expr) ------------------ | Branch (189:3): [True: 4.59k, False: 44] ------------------ 190| 4.59k| return data.getCrop(row * _pitch, width()).getAsArray1DRef(); 191| 4.63k|} _ZNK8rawspeed10Array2DRefIKNS_8CFAColorEE6heightEv: 171| 4.63k|__attribute__((always_inline)) inline int Array2DRef::height() const { 172| 4.63k| establishClassInvariants(); 173| 4.63k| return _height; 174| 4.63k|} _ZNK8rawspeed10Array2DRefINS_8CFAColorEE24establishClassInvariantsEv: 127| 49.5k|Array2DRef::establishClassInvariants() const noexcept { 128| 49.5k| data.establishClassInvariants(); 129| 49.5k| invariant(_width >= 0); ------------------ | | 27| 49.5k|#define invariant(expr) assert(expr) ------------------ | Branch (129:3): [True: 49.8k, False: 18.4E] ------------------ 130| 49.8k| invariant(_height >= 0); ------------------ | | 27| 49.8k|#define invariant(expr) assert(expr) ------------------ | Branch (130:3): [True: 49.8k, False: 18.4E] ------------------ 131| 49.8k| invariant(_pitch != 0); ------------------ | | 27| 49.8k|#define invariant(expr) assert(expr) ------------------ | Branch (131:3): [True: 49.9k, False: 18.4E] ------------------ 132| 49.9k| invariant(_pitch >= 0); ------------------ | | 27| 49.9k|#define invariant(expr) assert(expr) ------------------ | Branch (132:3): [True: 49.9k, False: 18.4E] ------------------ 133| 49.9k| invariant(_pitch >= _width); ------------------ | | 27| 49.9k|#define invariant(expr) assert(expr) ------------------ | Branch (133:3): [True: 49.9k, False: 18.4E] ------------------ 134| 49.9k| invariant((_width == 0) == (_height == 0)); ------------------ | | 27| 49.9k|#define invariant(expr) assert(expr) ------------------ | Branch (134:3): [True: 49.9k, False: 18.4E] ------------------ 135| 49.9k| invariant(data.size() == _pitch * _height); ------------------ | | 27| 49.9k|#define invariant(expr) assert(expr) ------------------ | Branch (135:3): [True: 49.9k, False: 63] ------------------ 136| 49.9k|} _ZNK8rawspeed10Array2DRefINS_8CFAColorEE6heightEv: 171| 11.2k|__attribute__((always_inline)) inline int Array2DRef::height() const { 172| 11.2k| establishClassInvariants(); 173| 11.2k| return _height; 174| 11.2k|} _ZNK8rawspeed10Array2DRefINS_8CFAColorEE5widthEv: 165| 22.1k|__attribute__((always_inline)) inline int Array2DRef::width() const { 166| 22.1k| establishClassInvariants(); 167| 22.1k| return _width; 168| 22.1k|} _ZNK8rawspeed10Array2DRefINS_8CFAColorEEclEii: 195| 6.46k|Array2DRef::operator()(const int row, const int col) const { 196| 6.46k| establishClassInvariants(); 197| 6.46k| invariant(col >= 0); ------------------ | | 27| 6.46k|#define invariant(expr) assert(expr) ------------------ | Branch (197:3): [True: 6.43k, False: 39] ------------------ 198| 6.43k| invariant(col < width()); ------------------ | | 27| 6.43k|#define invariant(expr) assert(expr) ------------------ | Branch (198:3): [True: 6.38k, False: 41] ------------------ 199| 6.38k| return (operator[](row))(col); 200| 6.43k|} _ZNK8rawspeed10Array2DRefINS_8CFAColorEEixEi: 186| 6.39k|inline Array1DRef Array2DRef::operator[](const int row) const { 187| 6.39k| establishClassInvariants(); 188| 6.39k| invariant(row >= 0); ------------------ | | 27| 6.39k|#define invariant(expr) assert(expr) ------------------ | Branch (188:3): [True: 6.45k, False: 18.4E] ------------------ 189| 6.45k| invariant(row < height()); ------------------ | | 27| 6.45k|#define invariant(expr) assert(expr) ------------------ | Branch (189:3): [True: 6.37k, False: 75] ------------------ 190| 6.37k| return data.getCrop(row * _pitch, width()).getAsArray1DRef(); 191| 6.45k|} _ZN8rawspeed10Array2DRefINS_8CFAColorEEC2EPS1_ii: 154| 1.62k| : Array2DRef(data_, width_, height_, /*pitch=*/width_) { 155| 1.62k| establishClassInvariants(); 156| 1.62k|} _ZN8rawspeed10Array2DRefINS_8CFAColorEEC2EPS1_iii: 148| 1.62k| : Array2DRef({data_, pitch_ * height_}, width_, height_, pitch_) { 149| 1.62k| establishClassInvariants(); 150| 1.62k|} _ZN8rawspeed10Array2DRefINS_8CFAColorEEC2ENS_10Array1DRefIS1_EEiii: 141| 1.62k| : data(data_), _pitch(pitch_), _width(width_), _height(height_) { 142| 1.62k| establishClassInvariants(); 143| 1.62k|} _ZN8rawspeed10Array2DRefIKNS_8CFAColorEEC2EPS2_ii: 154| 1.16k| : Array2DRef(data_, width_, height_, /*pitch=*/width_) { 155| 1.16k| establishClassInvariants(); 156| 1.16k|} _ZN8rawspeed10Array2DRefIKNS_8CFAColorEEC2EPS2_iii: 148| 1.15k| : Array2DRef({data_, pitch_ * height_}, width_, height_, pitch_) { 149| 1.15k| establishClassInvariants(); 150| 1.15k|} _ZN8rawspeed10Array2DRefIKNS_8CFAColorEEC2ENS_10Array1DRefIS2_EEiii: 141| 1.16k| : data(data_), _pitch(pitch_), _width(width_), _height(height_) { 142| 1.16k| establishClassInvariants(); 143| 1.16k|} _ZN8rawspeed10Array2DRefIfE6createINSt3__19allocatorIfEEEES1_RNS3_6vectorIfT_EEii: 99| 266| int height) { 100| 266| using VectorTy = std::remove_reference_t; 101| 266| storage = VectorTy(width * height); 102| 266| return {storage.data(), width, height}; 103| 266| } _ZN8rawspeed10Array2DRefIfEC2EPfii: 154| 532| : Array2DRef(data_, width_, height_, /*pitch=*/width_) { 155| 532| establishClassInvariants(); 156| 532|} _ZN8rawspeed8bitwidthIjEEjT_: 43| 15.6M|constexpr unsigned RAWSPEED_READNONE bitwidth([[maybe_unused]] T unused = {}) { 44| | return CHAR_BIT * sizeof(T); 45| 15.6M|} _ZN8rawspeed8bitwidthImEEjT_: 43| 21.6M|constexpr unsigned RAWSPEED_READNONE bitwidth([[maybe_unused]] T unused = {}) { 44| | return CHAR_BIT * sizeof(T); 45| 21.6M|} _ZN8rawspeed12isPowerOfTwoIiEEbT_: 38| 36|template constexpr bool RAWSPEED_READNONE isPowerOfTwo(T val) { 39| 36| return (val & (~val + 1)) == val; 40| 36|} _ZN8rawspeed14extractLowBitsIjQsr3stdE17unsigned_integralIT_EEES1_S1_j: 96| 10.1M|constexpr RAWSPEED_READNONE T extractLowBits(T value, unsigned nBits) { 97| | // invariant(nBits >= 0); 98| 10.1M| invariant(nBits != 0); // Would result in out-of-bound shift. ------------------ | | 27| 10.1M|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 10.1M, False: 0] ------------------ 99| 10.1M| invariant(nBits <= bitwidth()); // No-op is fine. ------------------ | | 27| 10.1M|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 10.3M, False: 18.4E] ------------------ 100| 10.3M| unsigned numHighPaddingBits = bitwidth() - nBits; 101| | // invariant(numHighPaddingBits >= 0); 102| 10.3M| invariant(numHighPaddingBits < bitwidth()); // Shift is in-bounds. ------------------ | | 27| 10.3M|#define invariant(expr) assert(expr) ------------------ | Branch (102:3): [True: 10.3M, False: 0] ------------------ 103| 10.3M| value <<= numHighPaddingBits; 104| 10.3M| value >>= numHighPaddingBits; 105| 10.3M| return value; 106| 10.3M|} _ZN8rawspeed15extractHighBitsImQsr3stdE13is_unsigned_vIT_EEES1_S1_jj: 121| 21.1M| T value, unsigned nBits, unsigned effectiveBitwidth = bitwidth()) { 122| 21.1M| invariant(effectiveBitwidth <= bitwidth()); ------------------ | | 27| 21.1M|#define invariant(expr) assert(expr) ------------------ | Branch (122:3): [True: 21.1M, False: 0] ------------------ 123| 21.1M| invariant(nBits <= effectiveBitwidth); ------------------ | | 27| 21.1M|#define invariant(expr) assert(expr) ------------------ | Branch (123:3): [True: 21.1M, False: 18.4E] ------------------ 124| 21.1M| auto numLowBitsToSkip = effectiveBitwidth - nBits; 125| 21.1M| invariant(numLowBitsToSkip < bitwidth()); ------------------ | | 27| 21.1M|#define invariant(expr) assert(expr) ------------------ | Branch (125:3): [True: 21.1M, False: 0] ------------------ 126| 21.1M| return value >> numLowBitsToSkip; 127| 21.1M|} _ZN8rawspeed8bitwidthIiEEjT_: 43| 9.85M|constexpr unsigned RAWSPEED_READNONE bitwidth([[maybe_unused]] T unused = {}) { 44| | return CHAR_BIT * sizeof(T); 45| 9.85M|} _ZN8rawspeed8bitwidthItEEjT_: 43| 1.53M|constexpr unsigned RAWSPEED_READNONE bitwidth([[maybe_unused]] T unused = {}) { 44| | return CHAR_BIT * sizeof(T); 45| 1.53M|} _ZN8rawspeed15extractHighBitsItQsr3stdE13is_unsigned_vIT_EEES1_S1_jj: 121| 1.25M| T value, unsigned nBits, unsigned effectiveBitwidth = bitwidth()) { 122| 1.25M| invariant(effectiveBitwidth <= bitwidth()); ------------------ | | 27| 1.25M|#define invariant(expr) assert(expr) ------------------ | Branch (122:3): [True: 1.25M, False: 3] ------------------ 123| 1.25M| invariant(nBits <= effectiveBitwidth); ------------------ | | 27| 1.25M|#define invariant(expr) assert(expr) ------------------ | Branch (123:3): [True: 1.25M, False: 18.4E] ------------------ 124| 1.25M| auto numLowBitsToSkip = effectiveBitwidth - nBits; 125| 1.25M| invariant(numLowBitsToSkip < bitwidth()); ------------------ | | 27| 1.25M|#define invariant(expr) assert(expr) ------------------ | Branch (125:3): [True: 1.25M, False: 1] ------------------ 126| 1.25M| return value >> numLowBitsToSkip; 127| 1.25M|} _ZN8rawspeed9clampBitsIiQsr3stdE15is_arithmetic_vIT_EEEDaS1_j: 75| 9.19M|constexpr auto RAWSPEED_READNONE clampBits(T value, unsigned int nBits) { 76| | // We expect to produce uint16_t. 77| 9.19M| invariant(nBits <= 16); ------------------ | | 27| 9.19M|#define invariant(expr) assert(expr) ------------------ | Branch (77:3): [True: 9.19M, False: 0] ------------------ 78| | // Check that the clamp is not a no-op. Not of uint16_t to 16 bits e.g. 79| | // (Well, not really, if we are called from clampBits, it's ok..). 80| 9.19M| invariant(bitwidth() > nBits); // If nBits >= bitwidth, then shift is UB. ------------------ | | 27| 9.19M|#define invariant(expr) assert(expr) ------------------ | Branch (80:3): [True: 9.19M, False: 0] ------------------ 81| 9.19M| const auto maxVal = implicit_cast((T(1) << nBits) - T(1)); 82| 9.19M| return implicit_cast(std::clamp(value, T(0), maxVal)); 83| 9.19M|} _ZN8rawspeed6isIntNIiQsr3stdE15is_arithmetic_vIT_EEEbS1_j: 87| 655k|constexpr bool RAWSPEED_READNONE isIntN(T value, unsigned int nBits) { 88| 655k| invariant(nBits < bitwidth() && "Check must not be tautological."); ------------------ | | 27| 655k|#define invariant(expr) assert(expr) ------------------ | Branch (88:3): [True: 655k, False: 0] | Branch (88:3): [True: 655k, Folded] | Branch (88:3): [True: 655k, False: 0] ------------------ 89| 655k| using UnsignedT = std::make_unsigned_t; 90| 655k| const auto highBits = static_cast(value) >> nBits; 91| 655k| return highBits == 0; 92| 655k|} _ZN8rawspeed9clampBitsIjQsr3stdE15is_arithmetic_vIT_EEEDaS1_j: 75| 287k|constexpr auto RAWSPEED_READNONE clampBits(T value, unsigned int nBits) { 76| | // We expect to produce uint16_t. 77| 287k| invariant(nBits <= 16); ------------------ | | 27| 287k|#define invariant(expr) assert(expr) ------------------ | Branch (77:3): [True: 287k, False: 0] ------------------ 78| | // Check that the clamp is not a no-op. Not of uint16_t to 16 bits e.g. 79| | // (Well, not really, if we are called from clampBits, it's ok..). 80| 287k| invariant(bitwidth() > nBits); // If nBits >= bitwidth, then shift is UB. ------------------ | | 27| 287k|#define invariant(expr) assert(expr) ------------------ | Branch (80:3): [True: 299k, False: 18.4E] ------------------ 81| 299k| const auto maxVal = implicit_cast((T(1) << nBits) - T(1)); 82| 299k| return implicit_cast(std::clamp(value, T(0), maxVal)); 83| 287k|} _ZN8rawspeed15extractHighBitsIjQsr3stdE13is_unsigned_vIT_EEES1_S1_jj: 121| 3.35M| T value, unsigned nBits, unsigned effectiveBitwidth = bitwidth()) { 122| 3.35M| invariant(effectiveBitwidth <= bitwidth()); ------------------ | | 27| 3.35M|#define invariant(expr) assert(expr) ------------------ | Branch (122:3): [True: 3.35M, False: 0] ------------------ 123| 3.35M| invariant(nBits <= effectiveBitwidth); ------------------ | | 27| 3.35M|#define invariant(expr) assert(expr) ------------------ | Branch (123:3): [True: 3.36M, False: 18.4E] ------------------ 124| 3.36M| auto numLowBitsToSkip = effectiveBitwidth - nBits; 125| 3.36M| invariant(numLowBitsToSkip < bitwidth()); ------------------ | | 27| 3.36M|#define invariant(expr) assert(expr) ------------------ | Branch (125:3): [True: 3.36M, False: 0] ------------------ 126| 3.36M| return value >> numLowBitsToSkip; 127| 3.36M|} _ZN8rawspeed13numActiveBitsImQaasr3stdE17unsigned_integralIT_Egecl8bitwidthIS1_EEclL_ZNS_8bitwidthIjEEjS1_EilEEEEjS1_: 56| 544k|unsigned numActiveBits(const T v) { 57| 544k| return bitwidth(v) - std::countl_zero(v); 58| 544k|} _ZN8rawspeed13numActiveBitsItQaasr3stdE17unsigned_integralIT_Eltcl8bitwidthIS1_EEclL_ZNS_8bitwidthIjEEjS1_EilEEEEjS1_: 62| 1.45M|unsigned numActiveBits(const T v) { 63| 1.45M| return numActiveBits(static_cast(v)); 64| 1.45M|} _ZN8rawspeed13numActiveBitsIjQaasr3stdE17unsigned_integralIT_Egecl8bitwidthIS1_EEclL_ZNS_8bitwidthIjEEjS1_EilEEEEjS1_: 56| 1.45M|unsigned numActiveBits(const T v) { 57| 1.45M| return bitwidth(v) - std::countl_zero(v); 58| 1.45M|} _ZN8rawspeed10signExtendIjQsr3stdE13is_unsigned_vIT_EEEu13__make_signedIS1_ES1_j: 132| 182k| RAWSPEED_READNONE signExtend(T value, unsigned int nBits) { 133| 182k| invariant(nBits != 0 && "Only valid for non-zero bit count."); ------------------ | | 27| 182k|#define invariant(expr) assert(expr) ------------------ | Branch (133:3): [True: 182k, False: 0] | Branch (133:3): [True: 182k, Folded] | Branch (133:3): [True: 182k, False: 0] ------------------ 134| 182k| const T SpareSignBits = bitwidth() - nBits; 135| 182k| using SignedT = std::make_signed_t; 136| 182k| return static_cast(value << SpareSignBits) >> SpareSignBits; 137| 182k|} _ZN8rawspeed13implicit_castIjmQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 23.6M|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 23.6M|#pragma GCC diagnostic push 34| 23.6M|#pragma GCC diagnostic ignored "-Wconversion" 35| 23.6M|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 23.6M| return value; 37| 23.6M|#pragma GCC diagnostic pop 38| 23.6M|} _ZN8rawspeed13implicit_castIijQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 60.8M|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 60.8M|#pragma GCC diagnostic push 34| 60.8M|#pragma GCC diagnostic ignored "-Wconversion" 35| 60.8M|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 60.8M| return value; 37| 60.8M|#pragma GCC diagnostic pop 38| 60.8M|} _ZN8rawspeed13implicit_castItjQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 33.0M|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 33.0M|#pragma GCC diagnostic push 34| 33.0M|#pragma GCC diagnostic ignored "-Wconversion" 35| 33.0M|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 33.0M| return value; 37| 33.0M|#pragma GCC diagnostic pop 38| 33.0M|} _ZN8rawspeed13implicit_castIfdQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 244|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 244|#pragma GCC diagnostic push 34| 244|#pragma GCC diagnostic ignored "-Wconversion" 35| 244|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 244| return value; 37| 244|#pragma GCC diagnostic pop 38| 244|} _ZN8rawspeed13implicit_castItiQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 28.3M|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 28.3M|#pragma GCC diagnostic push 34| 28.3M|#pragma GCC diagnostic ignored "-Wconversion" 35| 28.3M|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 28.3M| return value; 37| 28.3M|#pragma GCC diagnostic pop 38| 28.3M|} _ZN8rawspeed13implicit_castIhiQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 82.4k|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 82.4k|#pragma GCC diagnostic push 34| 82.4k|#pragma GCC diagnostic ignored "-Wconversion" 35| 82.4k|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 82.4k| return value; 37| 82.4k|#pragma GCC diagnostic pop 38| 82.4k|} _ZN8rawspeed13implicit_castIimQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 13.7M|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 13.7M|#pragma GCC diagnostic push 34| 13.7M|#pragma GCC diagnostic ignored "-Wconversion" 35| 13.7M|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 13.7M| return value; 37| 13.7M|#pragma GCC diagnostic pop 38| 13.7M|} _ZN8rawspeed13implicit_castImmQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Esr3stdE9is_same_vIS1_S2_EEES2_S1_: 46| 953|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 47| 953| return value; 48| 953|} _ZN8rawspeed13implicit_castIiiQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Esr3stdE9is_same_vIS1_S2_EEES2_S1_: 46| 9.19M|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 47| 9.19M| return value; 48| 9.19M|} _ZN8rawspeed13implicit_castIjiQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 1.74k|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 1.74k|#pragma GCC diagnostic push 34| 1.74k|#pragma GCC diagnostic ignored "-Wconversion" 35| 1.74k|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 1.74k| return value; 37| 1.74k|#pragma GCC diagnostic pop 38| 1.74k|} _ZN8rawspeed13implicit_castIliQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 405|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 405|#pragma GCC diagnostic push 34| 405|#pragma GCC diagnostic ignored "-Wconversion" 35| 405|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 405| return value; 37| 405|#pragma GCC diagnostic pop 38| 405|} _ZN8rawspeed13implicit_castImiQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 360|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 360|#pragma GCC diagnostic push 34| 360|#pragma GCC diagnostic ignored "-Wconversion" 35| 360|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 360| return value; 37| 360|#pragma GCC diagnostic pop 38| 360|} _ZN8rawspeed13implicit_castIjjQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Esr3stdE9is_same_vIS1_S2_EEES2_S1_: 46| 302k|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 47| 302k| return value; 48| 302k|} _ZN8rawspeed13implicit_castIifQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 1.09k|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 1.09k|#pragma GCC diagnostic push 34| 1.09k|#pragma GCC diagnostic ignored "-Wconversion" 35| 1.09k|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 1.09k| return value; 37| 1.09k|#pragma GCC diagnostic pop 38| 1.09k|} _ZN8rawspeed13implicit_castIsiQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 275k|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 275k|#pragma GCC diagnostic push 34| 275k|#pragma GCC diagnostic ignored "-Wconversion" 35| 275k|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 275k| return value; 37| 275k|#pragma GCC diagnostic pop 38| 275k|} _ZN8rawspeed13implicit_castIjyQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 69|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 69|#pragma GCC diagnostic push 34| 69|#pragma GCC diagnostic ignored "-Wconversion" 35| 69|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 69| return value; 37| 69|#pragma GCC diagnostic pop 38| 69|} _ZN8rawspeed13implicit_castItmQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 1.14M|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 1.14M|#pragma GCC diagnostic push 34| 1.14M|#pragma GCC diagnostic ignored "-Wconversion" 35| 1.14M|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 1.14M| return value; 37| 1.14M|#pragma GCC diagnostic pop 38| 1.14M|} _ZN8rawspeed13implicit_castIdmQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 5.24M|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 5.24M|#pragma GCC diagnostic push 34| 5.24M|#pragma GCC diagnostic ignored "-Wconversion" 35| 5.24M|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 5.24M| return value; 37| 5.24M|#pragma GCC diagnostic pop 38| 5.24M|} _ZN8rawspeed13implicit_castItdQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 8.52M|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 8.52M|#pragma GCC diagnostic push 34| 8.52M|#pragma GCC diagnostic ignored "-Wconversion" 35| 8.52M|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 8.52M| return value; 37| 8.52M|#pragma GCC diagnostic pop 38| 8.52M|} _ZN8rawspeed13implicit_castIdfQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 599k|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 599k|#pragma GCC diagnostic push 34| 599k|#pragma GCC diagnostic ignored "-Wconversion" 35| 599k|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 599k| return value; 37| 599k|#pragma GCC diagnostic pop 38| 599k|} _ZN8rawspeed13implicit_castIjfQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 3.12k|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 3.12k|#pragma GCC diagnostic push 34| 3.12k|#pragma GCC diagnostic ignored "-Wconversion" 35| 3.12k|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 3.12k| return value; 37| 3.12k|#pragma GCC diagnostic pop 38| 3.12k|} _ZN8rawspeed13implicit_castIfjQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 167|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 167|#pragma GCC diagnostic push 34| 167|#pragma GCC diagnostic ignored "-Wconversion" 35| 167|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 167| return value; 37| 167|#pragma GCC diagnostic pop 38| 167|} _ZN8rawspeed13implicit_castIhjQaaaaoosr3stdE13is_integral_vIT0_Esr3stdE19is_floating_point_vIS1_Eoosr3stdE13is_integral_vIT_Esr3stdE19is_floating_point_vIS2_Entsr3stdE9is_same_vIS1_S2_EEES2_S1_: 32| 493|constexpr RAWSPEED_READNONE Ttgt implicit_cast(Tsrc value) { 33| 493|#pragma GCC diagnostic push 34| 493|#pragma GCC diagnostic ignored "-Wconversion" 35| 493|#pragma GCC diagnostic ignored "-Wdouble-promotion" 36| 493| return value; 37| 493|#pragma GCC diagnostic pop 38| 493|} _ZNK8rawspeed17CroppedArray1DRefISt4byteE24establishClassInvariantsEv: 94| 53.6M|CroppedArray1DRef::establishClassInvariants() const noexcept { 95| 53.6M| base.establishClassInvariants(); 96| 53.6M| invariant(offset >= 0); ------------------ | | 27| 53.6M|#define invariant(expr) assert(expr) ------------------ | Branch (96:3): [True: 53.9M, False: 18.4E] ------------------ 97| 53.9M| invariant(numElts >= 0); ------------------ | | 27| 53.9M|#define invariant(expr) assert(expr) ------------------ | Branch (97:3): [True: 53.9M, False: 18.4E] ------------------ 98| 53.9M| invariant(offset <= base.size()); ------------------ | | 27| 53.9M|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 53.7M, False: 159k] ------------------ 99| 53.7M| invariant(numElts <= base.size()); ------------------ | | 27| 53.7M|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 53.5M, False: 242k] ------------------ 100| 53.5M| invariant(offset + numElts <= base.size()); ------------------ | | 27| 53.5M|#define invariant(expr) assert(expr) ------------------ | Branch (100:3): [True: 53.4M, False: 120k] ------------------ 101| 53.5M|} _ZNK8rawspeed17CroppedArray1DRefISt4byteE15getAsArray1DRefEv: 71| 7.54M| [[nodiscard]] Array1DRef getAsArray1DRef() const { 72| 7.54M| return {begin(), size()}; 73| 7.54M| } _ZNK8rawspeed17CroppedArray1DRefISt4byteE5beginEv: 137| 15.4M|template inline T* CroppedArray1DRef::begin() const { 138| 15.4M| establishClassInvariants(); 139| 15.4M| return addressOf(/*eltIdx=*/0); 140| 15.4M|} _ZNK8rawspeed17CroppedArray1DRefISt4byteE9addressOfEi: 152| 15.4M|inline T* CroppedArray1DRef::addressOf(const int eltIdx) const { 153| 15.4M| establishClassInvariants(); 154| 15.4M| invariant(eltIdx >= 0); ------------------ | | 27| 15.4M|#define invariant(expr) assert(expr) ------------------ | Branch (154:3): [True: 15.4M, False: 18.4E] ------------------ 155| 15.4M| invariant(eltIdx <= numElts); ------------------ | | 27| 15.4M|#define invariant(expr) assert(expr) ------------------ | Branch (155:3): [True: 15.4M, False: 18.4E] ------------------ 156| 15.4M| return base.addressOf(offset + eltIdx); 157| 15.4M|} _ZNK8rawspeed17CroppedArray1DRefISt4byteE4sizeEv: 146| 7.54M|template inline int CroppedArray1DRef::size() const { 147| 7.54M| establishClassInvariants(); 148| 7.54M| return numElts; 149| 7.54M|} _ZNK8rawspeed17CroppedArray1DRefIKSt4byteE24establishClassInvariantsEv: 94| 25.3M|CroppedArray1DRef::establishClassInvariants() const noexcept { 95| 25.3M| base.establishClassInvariants(); 96| 25.3M| invariant(offset >= 0); ------------------ | | 27| 25.3M|#define invariant(expr) assert(expr) ------------------ | Branch (96:3): [True: 26.5M, False: 18.4E] ------------------ 97| 26.5M| invariant(numElts >= 0); ------------------ | | 27| 26.5M|#define invariant(expr) assert(expr) ------------------ | Branch (97:3): [True: 26.5M, False: 18.4E] ------------------ 98| 26.5M| invariant(offset <= base.size()); ------------------ | | 27| 26.5M|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 26.0M, False: 530k] ------------------ 99| 26.0M| invariant(numElts <= base.size()); ------------------ | | 27| 26.0M|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 25.2M, False: 786k] ------------------ 100| 25.2M| invariant(offset + numElts <= base.size()); ------------------ | | 27| 25.2M|#define invariant(expr) assert(expr) ------------------ | Branch (100:3): [True: 24.8M, False: 414k] ------------------ 101| 25.2M|} _ZNK8rawspeed17CroppedArray1DRefIKSt4byteE15getAsArray1DRefEv: 71| 6.81M| [[nodiscard]] Array1DRef getAsArray1DRef() const { 72| 6.81M| return {begin(), size()}; 73| 6.81M| } _ZNK8rawspeed17CroppedArray1DRefIKSt4byteE5beginEv: 137| 6.82M|template inline T* CroppedArray1DRef::begin() const { 138| 6.82M| establishClassInvariants(); 139| 6.82M| return addressOf(/*eltIdx=*/0); 140| 6.82M|} _ZNK8rawspeed17CroppedArray1DRefIKSt4byteE9addressOfEi: 152| 6.75M|inline T* CroppedArray1DRef::addressOf(const int eltIdx) const { 153| 6.75M| establishClassInvariants(); 154| 6.75M| invariant(eltIdx >= 0); ------------------ | | 27| 6.75M|#define invariant(expr) assert(expr) ------------------ | Branch (154:3): [True: 6.74M, False: 11.2k] ------------------ 155| 6.74M| invariant(eltIdx <= numElts); ------------------ | | 27| 6.74M|#define invariant(expr) assert(expr) ------------------ | Branch (155:3): [True: 6.75M, False: 18.4E] ------------------ 156| 6.75M| return base.addressOf(offset + eltIdx); 157| 6.74M|} _ZNK8rawspeed17CroppedArray1DRefIKSt4byteE4sizeEv: 146| 6.71M|template inline int CroppedArray1DRef::size() const { 147| 6.71M| establishClassInvariants(); 148| 6.71M| return numElts; 149| 6.71M|} _ZNK8rawspeed17CroppedArray1DRefIKhE24establishClassInvariantsEv: 94| 117M|CroppedArray1DRef::establishClassInvariants() const noexcept { 95| 117M| base.establishClassInvariants(); 96| 117M| invariant(offset >= 0); ------------------ | | 27| 117M|#define invariant(expr) assert(expr) ------------------ | Branch (96:3): [True: 117M, False: 3.80k] ------------------ 97| 117M| invariant(numElts >= 0); ------------------ | | 27| 117M|#define invariant(expr) assert(expr) ------------------ | Branch (97:3): [True: 117M, False: 18.4E] ------------------ 98| 117M| invariant(offset <= base.size()); ------------------ | | 27| 117M|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 117M, False: 19.4k] ------------------ 99| 117M| invariant(numElts <= base.size()); ------------------ | | 27| 117M|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 117M, False: 43.6k] ------------------ 100| 117M| invariant(offset + numElts <= base.size()); ------------------ | | 27| 117M|#define invariant(expr) assert(expr) ------------------ | Branch (100:3): [True: 117M, False: 19.9k] ------------------ 101| 117M|} _ZNK8rawspeed17CroppedArray1DRefIKhE15getAsArray1DRefEv: 71| 29.2M| [[nodiscard]] Array1DRef getAsArray1DRef() const { 72| 29.2M| return {begin(), size()}; 73| 29.2M| } _ZNK8rawspeed17CroppedArray1DRefIKhE5beginEv: 137| 29.4M|template inline T* CroppedArray1DRef::begin() const { 138| 29.4M| establishClassInvariants(); 139| 29.4M| return addressOf(/*eltIdx=*/0); 140| 29.4M|} _ZNK8rawspeed17CroppedArray1DRefIKhE9addressOfEi: 152| 29.4M|inline T* CroppedArray1DRef::addressOf(const int eltIdx) const { 153| 29.4M| establishClassInvariants(); 154| 29.4M| invariant(eltIdx >= 0); ------------------ | | 27| 29.4M|#define invariant(expr) assert(expr) ------------------ | Branch (154:3): [True: 29.4M, False: 18.4E] ------------------ 155| 29.4M| invariant(eltIdx <= numElts); ------------------ | | 27| 29.4M|#define invariant(expr) assert(expr) ------------------ | Branch (155:3): [True: 29.4M, False: 18.4E] ------------------ 156| 29.4M| return base.addressOf(offset + eltIdx); 157| 29.4M|} _ZNK8rawspeed17CroppedArray1DRefIKhE4sizeEv: 146| 29.3M|template inline int CroppedArray1DRef::size() const { 147| 29.3M| establishClassInvariants(); 148| 29.3M| return numElts; 149| 29.3M|} _ZN8rawspeed17CroppedArray1DRefIKhEC2ENS_10Array1DRefIS1_EEii: 107| 29.4M| : base(base_), offset(offset_), numElts(numElts_) { 108| 29.4M| establishClassInvariants(); 109| 29.4M|} _ZN8rawspeed17CroppedArray1DRefIhEC2ENS_10Array1DRefIhEEii: 107| 185k| : base(base_), offset(offset_), numElts(numElts_) { 108| 185k| establishClassInvariants(); 109| 185k|} _ZNK8rawspeed17CroppedArray1DRefIhE24establishClassInvariantsEv: 94| 556k|CroppedArray1DRef::establishClassInvariants() const noexcept { 95| 556k| base.establishClassInvariants(); 96| 556k| invariant(offset >= 0); ------------------ | | 27| 556k|#define invariant(expr) assert(expr) ------------------ | Branch (96:3): [True: 556k, False: 0] ------------------ 97| 556k| invariant(numElts >= 0); ------------------ | | 27| 556k|#define invariant(expr) assert(expr) ------------------ | Branch (97:3): [True: 556k, False: 0] ------------------ 98| 556k| invariant(offset <= base.size()); ------------------ | | 27| 556k|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 556k, False: 0] ------------------ 99| 556k| invariant(numElts <= base.size()); ------------------ | | 27| 556k|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 556k, False: 0] ------------------ 100| 556k| invariant(offset + numElts <= base.size()); ------------------ | | 27| 556k|#define invariant(expr) assert(expr) ------------------ | Branch (100:3): [True: 556k, False: 0] ------------------ 101| 556k|} _ZNK8rawspeed17CroppedArray1DRefIhE5beginEv: 137| 185k|template inline T* CroppedArray1DRef::begin() const { 138| 185k| establishClassInvariants(); 139| 185k| return addressOf(/*eltIdx=*/0); 140| 185k|} _ZNK8rawspeed17CroppedArray1DRefIhE9addressOfEi: 152| 185k|inline T* CroppedArray1DRef::addressOf(const int eltIdx) const { 153| 185k| establishClassInvariants(); 154| 185k| invariant(eltIdx >= 0); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (154:3): [True: 185k, False: 0] ------------------ 155| 185k| invariant(eltIdx <= numElts); ------------------ | | 27| 185k|#define invariant(expr) assert(expr) ------------------ | Branch (155:3): [True: 185k, False: 0] ------------------ 156| 185k| return base.addressOf(offset + eltIdx); 157| 185k|} _ZN8rawspeed17CroppedArray1DRefIiEC2ENS_10Array1DRefIiEEii: 107| 2.11k| : base(base_), offset(offset_), numElts(numElts_) { 108| 2.11k| establishClassInvariants(); 109| 2.11k|} _ZNK8rawspeed17CroppedArray1DRefIiE24establishClassInvariantsEv: 94| 8.46k|CroppedArray1DRef::establishClassInvariants() const noexcept { 95| 8.46k| base.establishClassInvariants(); 96| 8.46k| invariant(offset >= 0); ------------------ | | 27| 8.46k|#define invariant(expr) assert(expr) ------------------ | Branch (96:3): [True: 8.46k, False: 0] ------------------ 97| 8.46k| invariant(numElts >= 0); ------------------ | | 27| 8.46k|#define invariant(expr) assert(expr) ------------------ | Branch (97:3): [True: 8.46k, False: 0] ------------------ 98| 8.46k| invariant(offset <= base.size()); ------------------ | | 27| 8.46k|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 8.46k, False: 0] ------------------ 99| 8.46k| invariant(numElts <= base.size()); ------------------ | | 27| 8.46k|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 8.46k, False: 0] ------------------ 100| 8.46k| invariant(offset + numElts <= base.size()); ------------------ | | 27| 8.46k|#define invariant(expr) assert(expr) ------------------ | Branch (100:3): [True: 8.46k, False: 0] ------------------ 101| 8.46k|} _ZNK8rawspeed17CroppedArray1DRefIiE15getAsArray1DRefEv: 71| 2.11k| [[nodiscard]] Array1DRef getAsArray1DRef() const { 72| 2.11k| return {begin(), size()}; 73| 2.11k| } _ZNK8rawspeed17CroppedArray1DRefIiE5beginEv: 137| 2.11k|template inline T* CroppedArray1DRef::begin() const { 138| 2.11k| establishClassInvariants(); 139| 2.11k| return addressOf(/*eltIdx=*/0); 140| 2.11k|} _ZNK8rawspeed17CroppedArray1DRefIiE9addressOfEi: 152| 2.11k|inline T* CroppedArray1DRef::addressOf(const int eltIdx) const { 153| 2.11k| establishClassInvariants(); 154| 2.11k| invariant(eltIdx >= 0); ------------------ | | 27| 2.11k|#define invariant(expr) assert(expr) ------------------ | Branch (154:3): [True: 2.11k, False: 0] ------------------ 155| 2.11k| invariant(eltIdx <= numElts); ------------------ | | 27| 2.11k|#define invariant(expr) assert(expr) ------------------ | Branch (155:3): [True: 2.11k, False: 0] ------------------ 156| 2.11k| return base.addressOf(offset + eltIdx); 157| 2.11k|} _ZNK8rawspeed17CroppedArray1DRefIiE4sizeEv: 146| 2.11k|template inline int CroppedArray1DRef::size() const { 147| 2.11k| establishClassInvariants(); 148| 2.11k| return numElts; 149| 2.11k|} _ZN8rawspeed17CroppedArray1DRefISt4byteEC2ENS_10Array1DRefIS1_EEii: 107| 15.6M| : base(base_), offset(offset_), numElts(numElts_) { 108| 15.6M| establishClassInvariants(); 109| 15.6M|} _ZNK8rawspeed17CroppedArray1DRefItE15getAsArray1DRefEv: 71| 43.1M| [[nodiscard]] Array1DRef getAsArray1DRef() const { 72| 43.1M| return {begin(), size()}; 73| 43.1M| } _ZNK8rawspeed17CroppedArray1DRefItE5beginEv: 137| 44.5M|template inline T* CroppedArray1DRef::begin() const { 138| 44.5M| establishClassInvariants(); 139| 44.5M| return addressOf(/*eltIdx=*/0); 140| 44.5M|} _ZNK8rawspeed17CroppedArray1DRefItE24establishClassInvariantsEv: 94| 165M|CroppedArray1DRef::establishClassInvariants() const noexcept { 95| 165M| base.establishClassInvariants(); 96| 165M| invariant(offset >= 0); ------------------ | | 27| 165M|#define invariant(expr) assert(expr) ------------------ | Branch (96:3): [True: 171M, False: 18.4E] ------------------ 97| 171M| invariant(numElts >= 0); ------------------ | | 27| 171M|#define invariant(expr) assert(expr) ------------------ | Branch (97:3): [True: 171M, False: 18.4E] ------------------ 98| 171M| invariant(offset <= base.size()); ------------------ | | 27| 171M|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 171M, False: 111k] ------------------ 99| 171M| invariant(numElts <= base.size()); ------------------ | | 27| 171M|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 167M, False: 4.50M] ------------------ 100| 167M| invariant(offset + numElts <= base.size()); ------------------ | | 27| 167M|#define invariant(expr) assert(expr) ------------------ | Branch (100:3): [True: 163M, False: 3.82M] ------------------ 101| 167M|} _ZNK8rawspeed17CroppedArray1DRefItE9addressOfEi: 152| 46.1M|inline T* CroppedArray1DRef::addressOf(const int eltIdx) const { 153| 46.1M| establishClassInvariants(); 154| 46.1M| invariant(eltIdx >= 0); ------------------ | | 27| 46.1M|#define invariant(expr) assert(expr) ------------------ | Branch (154:3): [True: 46.0M, False: 40.1k] ------------------ 155| 46.0M| invariant(eltIdx <= numElts); ------------------ | | 27| 46.0M|#define invariant(expr) assert(expr) ------------------ | Branch (155:3): [True: 46.1M, False: 18.4E] ------------------ 156| 46.1M| return base.addressOf(offset + eltIdx); 157| 46.0M|} _ZNK8rawspeed17CroppedArray1DRefItE4sizeEv: 146| 42.2M|template inline int CroppedArray1DRef::size() const { 147| 42.2M| establishClassInvariants(); 148| 42.2M| return numElts; 149| 42.2M|} _ZN8rawspeed17CroppedArray1DRefItEC2ENS_10Array1DRefItEEii: 107| 47.5M| : base(base_), offset(offset_), numElts(numElts_) { 108| 47.5M| establishClassInvariants(); 109| 47.5M|} _ZNK8rawspeed17CroppedArray1DRefItEclEi: 160| 2.01M|inline T& CroppedArray1DRef::operator()(const int eltIdx) const { 161| 2.01M| establishClassInvariants(); 162| 2.01M| invariant(eltIdx >= 0); ------------------ | | 27| 2.01M|#define invariant(expr) assert(expr) ------------------ | Branch (162:3): [True: 1.94M, False: 75.0k] ------------------ 163| 1.94M| invariant(eltIdx < numElts); ------------------ | | 27| 1.94M|#define invariant(expr) assert(expr) ------------------ | Branch (163:3): [True: 1.95M, False: 18.4E] ------------------ 164| 1.95M| return *addressOf(eltIdx); 165| 1.94M|} _ZN8rawspeed17CroppedArray1DRefIKSt4byteEC2ENS_10Array1DRefIS2_EEii: 107| 7.17M| : base(base_), offset(offset_), numElts(numElts_) { 108| 7.17M| establishClassInvariants(); 109| 7.17M|} _ZN8rawspeed17CroppedArray1DRefIfEC2ENS_10Array1DRefIfEEii: 107| 590k| : base(base_), offset(offset_), numElts(numElts_) { 108| 590k| establishClassInvariants(); 109| 590k|} _ZNK8rawspeed17CroppedArray1DRefIfE24establishClassInvariantsEv: 94| 2.32M|CroppedArray1DRef::establishClassInvariants() const noexcept { 95| 2.32M| base.establishClassInvariants(); 96| 2.32M| invariant(offset >= 0); ------------------ | | 27| 2.32M|#define invariant(expr) assert(expr) ------------------ | Branch (96:3): [True: 2.32M, False: 0] ------------------ 97| 2.32M| invariant(numElts >= 0); ------------------ | | 27| 2.32M|#define invariant(expr) assert(expr) ------------------ | Branch (97:3): [True: 2.32M, False: 0] ------------------ 98| 2.32M| invariant(offset <= base.size()); ------------------ | | 27| 2.32M|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 2.32M, False: 1] ------------------ 99| 2.32M| invariant(numElts <= base.size()); ------------------ | | 27| 2.32M|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 2.32M, False: 18.4E] ------------------ 100| 2.32M| invariant(offset + numElts <= base.size()); ------------------ | | 27| 2.32M|#define invariant(expr) assert(expr) ------------------ | Branch (100:3): [True: 2.32M, False: 0] ------------------ 101| 2.32M|} _ZNK8rawspeed17CroppedArray1DRefIfE15getAsArray1DRefEv: 71| 551k| [[nodiscard]] Array1DRef getAsArray1DRef() const { 72| 551k| return {begin(), size()}; 73| 551k| } _ZNK8rawspeed17CroppedArray1DRefIfE5beginEv: 137| 551k|template inline T* CroppedArray1DRef::begin() const { 138| 551k| establishClassInvariants(); 139| 551k| return addressOf(/*eltIdx=*/0); 140| 551k|} _ZNK8rawspeed17CroppedArray1DRefIfE9addressOfEi: 152| 590k|inline T* CroppedArray1DRef::addressOf(const int eltIdx) const { 153| 590k| establishClassInvariants(); 154| 590k| invariant(eltIdx >= 0); ------------------ | | 27| 590k|#define invariant(expr) assert(expr) ------------------ | Branch (154:3): [True: 590k, False: 0] ------------------ 155| 590k| invariant(eltIdx <= numElts); ------------------ | | 27| 590k|#define invariant(expr) assert(expr) ------------------ | Branch (155:3): [True: 590k, False: 0] ------------------ 156| 590k| return base.addressOf(offset + eltIdx); 157| 590k|} _ZNK8rawspeed17CroppedArray1DRefIfE4sizeEv: 146| 551k|template inline int CroppedArray1DRef::size() const { 147| 551k| establishClassInvariants(); 148| 551k| return numElts; 149| 551k|} _ZN8rawspeed17CroppedArray1DRefIKtEC2ENS_10Array1DRefIS1_EEii: 107| 3.72M| : base(base_), offset(offset_), numElts(numElts_) { 108| 3.72M| establishClassInvariants(); 109| 3.72M|} _ZNK8rawspeed17CroppedArray1DRefIKtE24establishClassInvariantsEv: 94| 14.7M|CroppedArray1DRef::establishClassInvariants() const noexcept { 95| 14.7M| base.establishClassInvariants(); 96| 14.7M| invariant(offset >= 0); ------------------ | | 27| 14.7M|#define invariant(expr) assert(expr) ------------------ | Branch (96:3): [True: 14.8M, False: 18.4E] ------------------ 97| 14.8M| invariant(numElts >= 0); ------------------ | | 27| 14.8M|#define invariant(expr) assert(expr) ------------------ | Branch (97:3): [True: 14.8M, False: 18.4E] ------------------ 98| 14.8M| invariant(offset <= base.size()); ------------------ | | 27| 14.8M|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 14.8M, False: 6.18k] ------------------ 99| 14.8M| invariant(numElts <= base.size()); ------------------ | | 27| 14.8M|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 14.7M, False: 24.2k] ------------------ 100| 14.7M| invariant(offset + numElts <= base.size()); ------------------ | | 27| 14.7M|#define invariant(expr) assert(expr) ------------------ | Branch (100:3): [True: 14.7M, False: 20.7k] ------------------ 101| 14.7M|} _ZNK8rawspeed17CroppedArray1DRefIKtE15getAsArray1DRefEv: 71| 3.71M| [[nodiscard]] Array1DRef getAsArray1DRef() const { 72| 3.71M| return {begin(), size()}; 73| 3.71M| } _ZNK8rawspeed17CroppedArray1DRefIKtE5beginEv: 137| 3.71M|template inline T* CroppedArray1DRef::begin() const { 138| 3.71M| establishClassInvariants(); 139| 3.71M| return addressOf(/*eltIdx=*/0); 140| 3.71M|} _ZNK8rawspeed17CroppedArray1DRefIKtE9addressOfEi: 152| 3.71M|inline T* CroppedArray1DRef::addressOf(const int eltIdx) const { 153| 3.71M| establishClassInvariants(); 154| 3.71M| invariant(eltIdx >= 0); ------------------ | | 27| 3.71M|#define invariant(expr) assert(expr) ------------------ | Branch (154:3): [True: 3.71M, False: 18.4E] ------------------ 155| 3.71M| invariant(eltIdx <= numElts); ------------------ | | 27| 3.71M|#define invariant(expr) assert(expr) ------------------ | Branch (155:3): [True: 3.71M, False: 18.4E] ------------------ 156| 3.71M| return base.addressOf(offset + eltIdx); 157| 3.71M|} _ZNK8rawspeed17CroppedArray1DRefIKtE4sizeEv: 146| 3.71M|template inline int CroppedArray1DRef::size() const { 147| 3.71M| establishClassInvariants(); 148| 3.71M| return numElts; 149| 3.71M|} _ZNK8rawspeed17CroppedArray1DRefIfEclEi: 160| 39.4k|inline T& CroppedArray1DRef::operator()(const int eltIdx) const { 161| 39.4k| establishClassInvariants(); 162| 39.4k| invariant(eltIdx >= 0); ------------------ | | 27| 39.4k|#define invariant(expr) assert(expr) ------------------ | Branch (162:3): [True: 39.4k, False: 0] ------------------ 163| 39.4k| invariant(eltIdx < numElts); ------------------ | | 27| 39.4k|#define invariant(expr) assert(expr) ------------------ | Branch (163:3): [True: 39.4k, False: 0] ------------------ 164| 39.4k| return *addressOf(eltIdx); 165| 39.4k|} _ZNK8rawspeed17CroppedArray1DRefIKNS_8CFAColorEE24establishClassInvariantsEv: 94| 18.2k|CroppedArray1DRef::establishClassInvariants() const noexcept { 95| 18.2k| base.establishClassInvariants(); 96| 18.2k| invariant(offset >= 0); ------------------ | | 27| 18.2k|#define invariant(expr) assert(expr) ------------------ | Branch (96:3): [True: 18.2k, False: 18.4E] ------------------ 97| 18.2k| invariant(numElts >= 0); ------------------ | | 27| 18.2k|#define invariant(expr) assert(expr) ------------------ | Branch (97:3): [True: 18.3k, False: 18.4E] ------------------ 98| 18.3k| invariant(offset <= base.size()); ------------------ | | 27| 18.3k|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 18.2k, False: 39] ------------------ 99| 18.2k| invariant(numElts <= base.size()); ------------------ | | 27| 18.2k|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 18.2k, False: 71] ------------------ 100| 18.2k| invariant(offset + numElts <= base.size()); ------------------ | | 27| 18.2k|#define invariant(expr) assert(expr) ------------------ | Branch (100:3): [True: 18.1k, False: 52] ------------------ 101| 18.2k|} _ZNK8rawspeed17CroppedArray1DRefIKNS_8CFAColorEE15getAsArray1DRefEv: 71| 4.60k| [[nodiscard]] Array1DRef getAsArray1DRef() const { 72| 4.60k| return {begin(), size()}; 73| 4.60k| } _ZNK8rawspeed17CroppedArray1DRefIKNS_8CFAColorEE5beginEv: 137| 4.60k|template inline T* CroppedArray1DRef::begin() const { 138| 4.60k| establishClassInvariants(); 139| 4.60k| return addressOf(/*eltIdx=*/0); 140| 4.60k|} _ZNK8rawspeed17CroppedArray1DRefIKNS_8CFAColorEE9addressOfEi: 152| 4.59k|inline T* CroppedArray1DRef::addressOf(const int eltIdx) const { 153| 4.59k| establishClassInvariants(); 154| 4.59k| invariant(eltIdx >= 0); ------------------ | | 27| 4.59k|#define invariant(expr) assert(expr) ------------------ | Branch (154:3): [True: 4.61k, False: 18.4E] ------------------ 155| 4.61k| invariant(eltIdx <= numElts); ------------------ | | 27| 4.61k|#define invariant(expr) assert(expr) ------------------ | Branch (155:3): [True: 4.61k, False: 18.4E] ------------------ 156| 4.61k| return base.addressOf(offset + eltIdx); 157| 4.61k|} _ZNK8rawspeed17CroppedArray1DRefIKNS_8CFAColorEE4sizeEv: 146| 4.61k|template inline int CroppedArray1DRef::size() const { 147| 4.61k| establishClassInvariants(); 148| 4.61k| return numElts; 149| 4.61k|} _ZNK8rawspeed17CroppedArray1DRefINS_8CFAColorEE24establishClassInvariantsEv: 94| 25.2k|CroppedArray1DRef::establishClassInvariants() const noexcept { 95| 25.2k| base.establishClassInvariants(); 96| 25.2k| invariant(offset >= 0); ------------------ | | 27| 25.2k|#define invariant(expr) assert(expr) ------------------ | Branch (96:3): [True: 25.4k, False: 18.4E] ------------------ 97| 25.4k| invariant(numElts >= 0); ------------------ | | 27| 25.4k|#define invariant(expr) assert(expr) ------------------ | Branch (97:3): [True: 25.4k, False: 18.4E] ------------------ 98| 25.4k| invariant(offset <= base.size()); ------------------ | | 27| 25.4k|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 25.4k, False: 23] ------------------ 99| 25.4k| invariant(numElts <= base.size()); ------------------ | | 27| 25.4k|#define invariant(expr) assert(expr) ------------------ | Branch (99:3): [True: 25.2k, False: 175] ------------------ 100| 25.2k| invariant(offset + numElts <= base.size()); ------------------ | | 27| 25.2k|#define invariant(expr) assert(expr) ------------------ | Branch (100:3): [True: 25.1k, False: 71] ------------------ 101| 25.2k|} _ZNK8rawspeed17CroppedArray1DRefINS_8CFAColorEE15getAsArray1DRefEv: 71| 6.42k| [[nodiscard]] Array1DRef getAsArray1DRef() const { 72| 6.42k| return {begin(), size()}; 73| 6.42k| } _ZNK8rawspeed17CroppedArray1DRefINS_8CFAColorEE5beginEv: 137| 6.43k|template inline T* CroppedArray1DRef::begin() const { 138| 6.43k| establishClassInvariants(); 139| 6.43k| return addressOf(/*eltIdx=*/0); 140| 6.43k|} _ZNK8rawspeed17CroppedArray1DRefINS_8CFAColorEE9addressOfEi: 152| 6.42k|inline T* CroppedArray1DRef::addressOf(const int eltIdx) const { 153| 6.42k| establishClassInvariants(); 154| 6.42k| invariant(eltIdx >= 0); ------------------ | | 27| 6.42k|#define invariant(expr) assert(expr) ------------------ | Branch (154:3): [True: 6.44k, False: 18.4E] ------------------ 155| 6.44k| invariant(eltIdx <= numElts); ------------------ | | 27| 6.44k|#define invariant(expr) assert(expr) ------------------ | Branch (155:3): [True: 6.43k, False: 1] ------------------ 156| 6.43k| return base.addressOf(offset + eltIdx); 157| 6.44k|} _ZNK8rawspeed17CroppedArray1DRefINS_8CFAColorEE4sizeEv: 146| 6.41k|template inline int CroppedArray1DRef::size() const { 147| 6.41k| establishClassInvariants(); 148| 6.41k| return numElts; 149| 6.41k|} _ZN8rawspeed17CroppedArray1DRefINS_8CFAColorEEC2ENS_10Array1DRefIS1_EEii: 107| 6.44k| : base(base_), offset(offset_), numElts(numElts_) { 108| 6.44k| establishClassInvariants(); 109| 6.44k|} _ZN8rawspeed17CroppedArray1DRefIKNS_8CFAColorEEC2ENS_10Array1DRefIS2_EEii: 107| 4.62k| : base(base_), offset(offset_), numElts(numElts_) { 108| 4.62k| establishClassInvariants(); 109| 4.62k|} _ZNK8rawspeed17CroppedArray2DRefItE24establishClassInvariantsEv: 98| 4.27M|CroppedArray2DRef::establishClassInvariants() const noexcept { 99| 4.27M| base.establishClassInvariants(); 100| 4.27M| invariant(offsetCols >= 0); ------------------ | | 27| 4.27M|#define invariant(expr) assert(expr) ------------------ | Branch (100:3): [True: 4.28M, False: 18.4E] ------------------ 101| 4.28M| invariant(offsetRows >= 0); ------------------ | | 27| 4.28M|#define invariant(expr) assert(expr) ------------------ | Branch (101:3): [True: 4.28M, False: 18.4E] ------------------ 102| 4.28M| invariant(croppedWidth >= 0); ------------------ | | 27| 4.28M|#define invariant(expr) assert(expr) ------------------ | Branch (102:3): [True: 4.28M, False: 18.4E] ------------------ 103| 4.28M| invariant(croppedHeight >= 0); ------------------ | | 27| 4.28M|#define invariant(expr) assert(expr) ------------------ | Branch (103:3): [True: 4.28M, False: 18.4E] ------------------ 104| 4.28M| invariant(offsetCols <= base.width()); ------------------ | | 27| 4.28M|#define invariant(expr) assert(expr) ------------------ | Branch (104:3): [True: 4.27M, False: 12.4k] ------------------ 105| 4.27M| invariant(offsetRows <= base.height()); ------------------ | | 27| 4.27M|#define invariant(expr) assert(expr) ------------------ | Branch (105:3): [True: 4.26M, False: 4.91k] ------------------ 106| 4.26M| invariant(croppedWidth <= base.width()); ------------------ | | 27| 4.26M|#define invariant(expr) assert(expr) ------------------ | Branch (106:3): [True: 4.26M, False: 1.21k] ------------------ 107| 4.26M| invariant(croppedHeight <= base.height()); ------------------ | | 27| 4.26M|#define invariant(expr) assert(expr) ------------------ | Branch (107:3): [True: 4.26M, False: 690] ------------------ 108| 4.26M| invariant(offsetCols + croppedWidth <= base.width()); ------------------ | | 27| 4.26M|#define invariant(expr) assert(expr) ------------------ | Branch (108:3): [True: 4.26M, False: 18.4E] ------------------ 109| 4.26M| invariant(offsetRows + croppedHeight <= base.height()); ------------------ | | 27| 4.26M|#define invariant(expr) assert(expr) ------------------ | Branch (109:3): [True: 4.26M, False: 18.4E] ------------------ 110| 4.26M| invariant((croppedWidth == 0) == (croppedHeight == 0)); ------------------ | | 27| 4.26M|#define invariant(expr) assert(expr) ------------------ | Branch (110:3): [True: 4.26M, False: 18.4E] ------------------ 111| 4.26M|} _ZNK8rawspeed17CroppedArray2DRefIfE24establishClassInvariantsEv: 98| 79.1k|CroppedArray2DRef::establishClassInvariants() const noexcept { 99| 79.1k| base.establishClassInvariants(); 100| 79.1k| invariant(offsetCols >= 0); ------------------ | | 27| 79.1k|#define invariant(expr) assert(expr) ------------------ | Branch (100:3): [True: 79.1k, False: 0] ------------------ 101| 79.1k| invariant(offsetRows >= 0); ------------------ | | 27| 79.1k|#define invariant(expr) assert(expr) ------------------ | Branch (101:3): [True: 79.1k, False: 0] ------------------ 102| 79.1k| invariant(croppedWidth >= 0); ------------------ | | 27| 79.1k|#define invariant(expr) assert(expr) ------------------ | Branch (102:3): [True: 79.1k, False: 0] ------------------ 103| 79.1k| invariant(croppedHeight >= 0); ------------------ | | 27| 79.1k|#define invariant(expr) assert(expr) ------------------ | Branch (103:3): [True: 79.1k, False: 0] ------------------ 104| 79.1k| invariant(offsetCols <= base.width()); ------------------ | | 27| 79.1k|#define invariant(expr) assert(expr) ------------------ | Branch (104:3): [True: 79.1k, False: 0] ------------------ 105| 79.1k| invariant(offsetRows <= base.height()); ------------------ | | 27| 79.1k|#define invariant(expr) assert(expr) ------------------ | Branch (105:3): [True: 79.1k, False: 0] ------------------ 106| 79.1k| invariant(croppedWidth <= base.width()); ------------------ | | 27| 79.1k|#define invariant(expr) assert(expr) ------------------ | Branch (106:3): [True: 79.1k, False: 0] ------------------ 107| 79.1k| invariant(croppedHeight <= base.height()); ------------------ | | 27| 79.1k|#define invariant(expr) assert(expr) ------------------ | Branch (107:3): [True: 79.1k, False: 0] ------------------ 108| 79.1k| invariant(offsetCols + croppedWidth <= base.width()); ------------------ | | 27| 79.1k|#define invariant(expr) assert(expr) ------------------ | Branch (108:3): [True: 79.1k, False: 0] ------------------ 109| 79.1k| invariant(offsetRows + croppedHeight <= base.height()); ------------------ | | 27| 79.1k|#define invariant(expr) assert(expr) ------------------ | Branch (109:3): [True: 79.1k, False: 0] ------------------ 110| 79.1k| invariant((croppedWidth == 0) == (croppedHeight == 0)); ------------------ | | 27| 79.1k|#define invariant(expr) assert(expr) ------------------ | Branch (110:3): [True: 79.1k, False: 0] ------------------ 111| 79.1k|} _ZN8rawspeed17CroppedArray2DRefItEC2ENS_10Array2DRefItEEiiii: 118| 1.29M| : base(base_), offsetCols(offsetCols_), offsetRows(offsetRows_), 119| 1.29M| croppedWidth(croppedWidth_), croppedHeight(croppedHeight_) { 120| 1.29M| establishClassInvariants(); 121| 1.29M|} _ZNK8rawspeed17CroppedArray2DRefItEclEii: 134| 204k|inline T& CroppedArray2DRef::operator()(const int row, const int col) const { 135| 204k| establishClassInvariants(); 136| 204k| invariant(col >= 0); ------------------ | | 27| 204k|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 204k, False: 0] ------------------ 137| 204k| invariant(col < croppedWidth); ------------------ | | 27| 204k|#define invariant(expr) assert(expr) ------------------ | Branch (137:3): [True: 204k, False: 0] ------------------ 138| 204k| return (operator[](row))(col); 139| 204k|} _ZNK8rawspeed17CroppedArray2DRefItEixEi: 125| 1.49M|CroppedArray2DRef::operator[](const int row) const { 126| 1.49M| establishClassInvariants(); 127| 1.49M| invariant(row >= 0); ------------------ | | 27| 1.49M|#define invariant(expr) assert(expr) ------------------ | Branch (127:3): [True: 1.49M, False: 18.4E] ------------------ 128| 1.49M| invariant(row < croppedHeight); ------------------ | | 27| 1.49M|#define invariant(expr) assert(expr) ------------------ | Branch (128:3): [True: 1.49M, False: 18.4E] ------------------ 129| 1.49M| const Array1DRef fullLine = base.operator[](offsetRows + row); 130| 1.49M| return fullLine.getCrop(offsetCols, croppedWidth); 131| 1.49M|} _ZNK8rawspeed17CroppedArray2DRefItE15getAsArray2DRefEv: 79| 1.28M| [[nodiscard]] Array2DRef getAsArray2DRef() const { 80| 1.28M| establishClassInvariants(); 81| 1.28M| return {operator[](0).begin(), croppedWidth, croppedHeight, base.pitch()}; 82| 1.28M| } _ZNK8rawspeed17CroppedArray2DRefISt4byteEixEi: 125| 1.84M|CroppedArray2DRef::operator[](const int row) const { 126| 1.84M| establishClassInvariants(); 127| 1.84M| invariant(row >= 0); ------------------ | | 27| 1.84M|#define invariant(expr) assert(expr) ------------------ | Branch (127:3): [True: 1.84M, False: 0] ------------------ 128| 1.84M| invariant(row < croppedHeight); ------------------ | | 27| 1.84M|#define invariant(expr) assert(expr) ------------------ | Branch (128:3): [True: 1.84M, False: 0] ------------------ 129| 1.84M| const Array1DRef fullLine = base.operator[](offsetRows + row); 130| 1.84M| return fullLine.getCrop(offsetCols, croppedWidth); 131| 1.84M|} _ZNK8rawspeed17CroppedArray2DRefISt4byteE24establishClassInvariantsEv: 98| 1.85M|CroppedArray2DRef::establishClassInvariants() const noexcept { 99| 1.85M| base.establishClassInvariants(); 100| 1.85M| invariant(offsetCols >= 0); ------------------ | | 27| 1.85M|#define invariant(expr) assert(expr) ------------------ | Branch (100:3): [True: 1.85M, False: 0] ------------------ 101| 1.85M| invariant(offsetRows >= 0); ------------------ | | 27| 1.85M|#define invariant(expr) assert(expr) ------------------ | Branch (101:3): [True: 1.85M, False: 0] ------------------ 102| 1.85M| invariant(croppedWidth >= 0); ------------------ | | 27| 1.85M|#define invariant(expr) assert(expr) ------------------ | Branch (102:3): [True: 1.85M, False: 0] ------------------ 103| 1.85M| invariant(croppedHeight >= 0); ------------------ | | 27| 1.85M|#define invariant(expr) assert(expr) ------------------ | Branch (103:3): [True: 1.85M, False: 0] ------------------ 104| 1.85M| invariant(offsetCols <= base.width()); ------------------ | | 27| 1.85M|#define invariant(expr) assert(expr) ------------------ | Branch (104:3): [True: 1.85M, False: 0] ------------------ 105| 1.85M| invariant(offsetRows <= base.height()); ------------------ | | 27| 1.85M|#define invariant(expr) assert(expr) ------------------ | Branch (105:3): [True: 1.85M, False: 0] ------------------ 106| 1.85M| invariant(croppedWidth <= base.width()); ------------------ | | 27| 1.85M|#define invariant(expr) assert(expr) ------------------ | Branch (106:3): [True: 1.85M, False: 0] ------------------ 107| 1.85M| invariant(croppedHeight <= base.height()); ------------------ | | 27| 1.85M|#define invariant(expr) assert(expr) ------------------ | Branch (107:3): [True: 1.85M, False: 0] ------------------ 108| 1.85M| invariant(offsetCols + croppedWidth <= base.width()); ------------------ | | 27| 1.85M|#define invariant(expr) assert(expr) ------------------ | Branch (108:3): [True: 1.85M, False: 0] ------------------ 109| 1.85M| invariant(offsetRows + croppedHeight <= base.height()); ------------------ | | 27| 1.85M|#define invariant(expr) assert(expr) ------------------ | Branch (109:3): [True: 1.85M, False: 0] ------------------ 110| 1.85M| invariant((croppedWidth == 0) == (croppedHeight == 0)); ------------------ | | 27| 1.85M|#define invariant(expr) assert(expr) ------------------ | Branch (110:3): [True: 1.85M, False: 0] ------------------ 111| 1.85M|} _ZN8rawspeed17CroppedArray2DRefISt4byteEC2ENS_10Array2DRefIS1_EE: 64| 3.07k| : CroppedArray2DRef(RHS, /*offsetCols=*/0, /*offsetRows=*/0, RHS.width(), 65| 3.07k| RHS.height()) {} _ZN8rawspeed17CroppedArray2DRefISt4byteEC2ENS_10Array2DRefIS1_EEiiii: 118| 3.07k| : base(base_), offsetCols(offsetCols_), offsetRows(offsetRows_), 119| 3.07k| croppedWidth(croppedWidth_), croppedHeight(croppedHeight_) { 120| 3.07k| establishClassInvariants(); 121| 3.07k|} _ZN8rawspeed17CroppedArray2DRefIfEC2ENS_10Array2DRefIfEEiiii: 118| 285| : base(base_), offsetCols(offsetCols_), offsetRows(offsetRows_), 119| 285| croppedWidth(croppedWidth_), croppedHeight(croppedHeight_) { 120| 285| establishClassInvariants(); 121| 285|} _ZNK8rawspeed17CroppedArray2DRefIfEclEii: 134| 39.4k|inline T& CroppedArray2DRef::operator()(const int row, const int col) const { 135| 39.4k| establishClassInvariants(); 136| 39.4k| invariant(col >= 0); ------------------ | | 27| 39.4k|#define invariant(expr) assert(expr) ------------------ | Branch (136:3): [True: 39.4k, False: 0] ------------------ 137| 39.4k| invariant(col < croppedWidth); ------------------ | | 27| 39.4k|#define invariant(expr) assert(expr) ------------------ | Branch (137:3): [True: 39.4k, False: 0] ------------------ 138| 39.4k| return (operator[](row))(col); 139| 39.4k|} _ZNK8rawspeed17CroppedArray2DRefIfEixEi: 125| 39.4k|CroppedArray2DRef::operator[](const int row) const { 126| 39.4k| establishClassInvariants(); 127| 39.4k| invariant(row >= 0); ------------------ | | 27| 39.4k|#define invariant(expr) assert(expr) ------------------ | Branch (127:3): [True: 39.4k, False: 0] ------------------ 128| 39.4k| invariant(row < croppedHeight); ------------------ | | 27| 39.4k|#define invariant(expr) assert(expr) ------------------ | Branch (128:3): [True: 39.4k, False: 0] ------------------ 129| 39.4k| const Array1DRef fullLine = base.operator[](offsetRows + row); 130| 39.4k| return fullLine.getCrop(offsetCols, croppedWidth); 131| 39.4k|} _ZN8rawspeed27DefaultInitAllocatorAdaptorIhNS_16AlignedAllocatorIhLi16EEEE10deallocateEPhm: 69| 7.93k| void deallocate(T* p, std::size_t n) noexcept { 70| 7.93k| allocator_traits::deallocate(allocator, p, n); 71| 7.93k| } _ZNK8rawspeed27DefaultInitAllocatorAdaptorIhNS_16AlignedAllocatorIhLi16EEEE9constructIhEEvPT_: 75| 31.8G| noexcept(std::is_nothrow_default_constructible_v) { 76| 31.8G| ::new (static_cast(ptr)) U; // start the life-time, but do not init. 77| 31.8G| } _ZN8rawspeed27DefaultInitAllocatorAdaptorIhNS_16AlignedAllocatorIhLi16EEEE8allocateEm: 65| 7.93k| T* allocate(std::size_t n) { 66| 7.93k| return allocator_traits::allocate(allocator, n); 67| 7.93k| } _ZN8rawspeed5MutexC2Ev: 40| 30.8k| explicit Mutex() { omp_init_lock(&mutex); } _ZN8rawspeed5MutexD2Ev: 47| 30.8k| ~Mutex() { omp_destroy_lock(&mutex); } _ZN8rawspeed5Mutex4LockEv: 52| 25.1k| void Lock() ACQUIRE() { omp_set_lock(&mutex); } _ZN8rawspeed5Mutex6UnlockEv: 55| 25.2k| void Unlock() RELEASE() { omp_unset_lock(&mutex); } _ZN8rawspeed11MutexLockerC2EPNS_5MutexE: 107| 25.1k| explicit MutexLocker(Mutex* mu) ACQUIRE(mu) : mut(mu) { mu->Lock(); } _ZN8rawspeed11MutexLockerD2Ev: 114| 25.2k| ~MutexLocker() RELEASE() { mut->Unlock(); } _ZN8rawspeed11NORangesSetINS_6BufferEE6insertERKS1_: 61| 27.0k| bool insert(const T& newElt) { 62| 27.0k| if (rangeIsOverlappingExistingElementOfSortedSet(newElt)) ------------------ | Branch (62:9): [True: 2.34k, False: 24.7k] ------------------ 63| 2.34k| return false; 64| | 65| 24.7k| auto i = elts.insert(newElt); 66| 24.7k| assert(i.second && "Did not insert after all?"); ------------------ | Branch (66:5): [True: 24.7k, False: 0] | Branch (66:5): [True: 24.7k, Folded] | Branch (66:5): [True: 24.7k, False: 0] ------------------ 67| 24.7k| (void)i; 68| 24.7k| return true; 69| 24.7k| } _ZNK8rawspeed11NORangesSetINS_6BufferEE44rangeIsOverlappingExistingElementOfSortedSetERKS1_: 36| 27.0k| rangeIsOverlappingExistingElementOfSortedSet(const T& newElt) const { 37| | // If there are no elements in set, then the new element 38| | // does not overlap any existing elements. 39| 27.0k| if (elts.empty()) ------------------ | Branch (39:9): [True: 17.3k, False: 9.70k] ------------------ 40| 17.3k| return false; 41| | 42| | // Find the first element that is not less than the new element. 43| 9.70k| auto p = 44| 9.70k| std::partition_point(elts.begin(), elts.end(), 45| 9.70k| [newElt](const T& elt) { return elt < newElt; }); 46| | 47| | // If there is such an element, we must not overlap with it. 48| 9.70k| if (p != elts.end() && RangesOverlap(newElt, *p)) ------------------ | Branch (48:9): [True: 3.93k, False: 5.76k] | Branch (48:9): [True: 1.31k, False: 8.39k] | Branch (48:28): [True: 1.31k, False: 2.62k] ------------------ 49| 1.31k| return true; 50| | 51| | // Now, is there an element before the element we found? 52| 8.39k| if (p == elts.begin()) ------------------ | Branch (52:9): [True: 393, False: 7.99k] ------------------ 53| 393| return false; 54| | 55| | // There is. We *also* must not overlap with that element too. 56| 7.99k| auto prevBeforeP = std::prev(p); 57| 7.99k| return RangesOverlap(newElt, *prevBeforeP); 58| 8.39k| } _ZZNK8rawspeed11NORangesSetINS_6BufferEE44rangeIsOverlappingExistingElementOfSortedSetERKS1_ENKUlS4_E_clES4_: 45| 17.5k| [newElt](const T& elt) { return elt < newElt; }); _ZN8rawspeed12NotARationalIjEC2Ejj: 35| 3.54k| NotARational(T num_, T den_) : num(num_), den(den_) {} _ZN8rawspeed12NotARationalIiEC2Eii: 35| 74.0k| NotARational(T num_, T den_) : num(num_), den(den_) {} _ZNK8rawspeed12NotARationalIjEcvT_IfQsr3stdE19is_floating_point_vITL0__EEEv: 39| 2.15k| explicit operator T2() const { 40| 2.15k| return T2(num) / T2(den); 41| 2.15k| } _ZNK8rawspeed12NotARationalIiEcvT_IfQsr3stdE19is_floating_point_vITL0__EEEv: 39| 1.45k| explicit operator T2() const { 40| 1.45k| return T2(num) / T2(den); 41| 1.45k| } _ZNK8rawspeed12NotARationalIjEcvT_IdQsr3stdE19is_floating_point_vITL0__EEEv: 39| 96| explicit operator T2() const { 40| 96| return T2(num) / T2(den); 41| 96| } _ZNK8rawspeed8OptionalINS_10Array1DRefISt4byteEEEcvbEv: 89| 705| explicit operator bool() const { return has_value(); } _ZNK8rawspeed8OptionalINS_10Array1DRefISt4byteEEE9has_valueEv: 85| 899| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 899| return impl.has_value(); 87| 899| } _ZNK8rawspeed8OptionalINS_10Array1DRefIKSt4byteEEEcvbEv: 89| 196| explicit operator bool() const { return has_value(); } _ZNK8rawspeed8OptionalINS_10Array1DRefIKSt4byteEEE9has_valueEv: 85| 390| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 390| return impl.has_value(); 87| 390| } _ZNR8rawspeed8OptionalINS_10Array1DRefISt4byteEEEdeEv: 70| 194| T& operator*() & { 71| 194| invariant(has_value()); ------------------ | | 27| 194|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 194, False: 0] ------------------ 72| 194| return impl.value(); 73| 194| } _ZNR8rawspeed8OptionalINS_10Array1DRefIKSt4byteEEEdeEv: 70| 194| T& operator*() & { 71| 194| invariant(has_value()); ------------------ | | 27| 194|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 194, False: 0] ------------------ 72| 194| return impl.value(); 73| 194| } _ZN8rawspeed8OptionalINS_10ByteStreamEEC2Ev: 35| 175| Optional() = default; _ZN8rawspeed8OptionalINS_10ByteStreamEEaSIS1_Qsr3stdE7same_asITL0__T_EEERS2_OS5_: 46| 50| Optional& operator=(U&& value) { 47| 50| impl = std::forward(value); 48| 50| return *this; 49| 50| } _ZNR8rawspeed8OptionalINS_10ByteStreamEEdeEv: 70| 6.93k| T& operator*() & { 71| 6.93k| invariant(has_value()); ------------------ | | 27| 6.93k|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 6.93k, False: 0] ------------------ 72| 6.93k| return impl.value(); 73| 6.93k| } _ZNK8rawspeed8OptionalINS_10ByteStreamEE9has_valueEv: 85| 14.4k| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 14.4k| return impl.has_value(); 87| 14.4k| } _ZN8rawspeed8OptionalINSt3__15arrayIfLm4EEEEC2IRS3_Qaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS7_RSA_Entsr3stdE7same_asIS7_OSA_Entsr3stdE7same_asIS7_NS1_8optionalIS8_EEEEEOS8_: 42| 471| Optional(U&& value) : impl(std::forward(value)) {} _ZNK8rawspeed8OptionalIiEcvbEv: 89| 208| explicit operator bool() const { return has_value(); } _ZNK8rawspeed8OptionalIiE9has_valueEv: 85| 19.3k| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 19.3k| return impl.has_value(); 87| 19.3k| } _ZNR8rawspeed8OptionalIiEdeEv: 70| 12.5k| T& operator*() & { 71| 12.5k| invariant(has_value()); ------------------ | | 27| 12.5k|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 12.5k, False: 18.4E] ------------------ 72| 12.5k| return impl.value(); 73| 12.5k| } _ZN8rawspeed8OptionalIiEaSIiQsr3stdE7same_asITL0__T_EEERS1_OS4_: 46| 3.53k| Optional& operator=(U&& value) { 47| 3.53k| impl = std::forward(value); 48| 3.53k| return *this; 49| 3.53k| } _ZN8rawspeed8OptionalINS_10Array2DRefIiEEEaSIS2_Qsr3stdE7same_asITL0__T_EEERS3_OS6_: 46| 1.99k| Optional& operator=(U&& value) { 47| 1.99k| impl = std::forward(value); 48| 1.99k| return *this; 49| 1.99k| } _ZN8rawspeed8OptionalINS_10Array2DRefIiEEEptEv: 60| 2.11k| T* operator->() { 61| 2.11k| invariant(has_value()); ------------------ | | 27| 2.11k|#define invariant(expr) assert(expr) ------------------ | Branch (61:5): [True: 2.11k, False: 0] ------------------ 62| 2.11k| return &impl.value(); 63| 2.11k| } _ZNK8rawspeed8OptionalINS_10Array2DRefIiEEE9has_valueEv: 85| 2.11k| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 2.11k| return impl.has_value(); 87| 2.11k| } _ZN8rawspeed8OptionalINS_10Array1DRefIiEEEC2IS2_Qaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS5_RS8_Entsr3stdE7same_asIS5_OS8_Entsr3stdE7same_asIS5_NSt3__18optionalIS6_EEEEEOS6_: 42| 2.11k| Optional(U&& value) : impl(std::forward(value)) {} _ZNO8rawspeed8OptionalINS_10Array1DRefIiEEEdeEv: 75| 2.11k| T&& operator*() && { 76| 2.11k| invariant(has_value()); ------------------ | | 27| 2.11k|#define invariant(expr) assert(expr) ------------------ | Branch (76:5): [True: 2.11k, False: 0] ------------------ 77| 2.11k| return std::move(impl.value()); 78| 2.11k| } _ZNK8rawspeed8OptionalINS_10Array1DRefIiEEE9has_valueEv: 85| 2.11k| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 2.11k| return impl.has_value(); 87| 2.11k| } _ZNK8rawspeed8OptionalINS_10JpegMarkerEEcvbEv: 89| 4.20M| explicit operator bool() const { return has_value(); } _ZNK8rawspeed8OptionalINS_10JpegMarkerEE9has_valueEv: 85| 4.21M| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 4.21M| return impl.has_value(); 87| 4.21M| } _ZN8rawspeed8OptionalINS_10Array2DRefIiEEEC2Ev: 35| 15.4k| Optional() = default; _ZN8rawspeed8OptionalIiEC2Ev: 35| 19.3k| Optional() = default; _ZN8rawspeed8OptionalINSt3__15arrayIfLm4EEEEC2Ev: 35| 15.4k| Optional() = default; _ZN8rawspeed8OptionalINSt3__117basic_string_viewIcNS1_11char_traitsIcEEEEEC2IRA4_KcQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asISB_RSE_Entsr3stdE7same_asISB_OSE_Entsr3stdE7same_asISB_NS1_8optionalISC_EEEEEOSC_: 42| 26| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINSt3__117basic_string_viewIcNS1_11char_traitsIcEEEEEC2IRA6_KcQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asISB_RSE_Entsr3stdE7same_asISB_OSE_Entsr3stdE7same_asISB_NS1_8optionalISC_EEEEEOSC_: 42| 38| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINSt3__117basic_string_viewIcNS1_11char_traitsIcEEEEEC2IRA5_KcQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asISB_RSE_Entsr3stdE7same_asISB_OSE_Entsr3stdE7same_asISB_NS1_8optionalISC_EEEEEOSC_: 42| 30| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINSt3__117basic_string_viewIcNS1_11char_traitsIcEEEEEC2IRA8_KcQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asISB_RSE_Entsr3stdE7same_asISB_OSE_Entsr3stdE7same_asISB_NS1_8optionalISC_EEEEEOSC_: 42| 14| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINSt3__117basic_string_viewIcNS1_11char_traitsIcEEEEEC2IRA7_KcQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asISB_RSE_Entsr3stdE7same_asISB_OSE_Entsr3stdE7same_asISB_NS1_8optionalISC_EEEEEOSC_: 42| 18| Optional(U&& value) : impl(std::forward(value)) {} _ZNK8rawspeed8OptionalINSt3__117basic_string_viewIcNS1_11char_traitsIcEEEEEcvbEv: 89| 126| explicit operator bool() const { return has_value(); } _ZNK8rawspeed8OptionalINSt3__117basic_string_viewIcNS1_11char_traitsIcEEEEE9has_valueEv: 85| 252| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 252| return impl.has_value(); 87| 252| } _ZNR8rawspeed8OptionalINSt3__117basic_string_viewIcNS1_11char_traitsIcEEEEEdeEv: 70| 126| T& operator*() & { 71| 126| invariant(has_value()); ------------------ | | 27| 126|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 126, False: 0] ------------------ 72| 126| return impl.value(); 73| 126| } _ZN8rawspeed8OptionalINS_10Array1DRefISt4byteEEEC2IS3_Qaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS6_RS9_Entsr3stdE7same_asIS6_OS9_Entsr3stdE7same_asIS6_NSt3__18optionalIS7_EEEEEOS7_: 42| 196| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINS_10Array1DRefISt4byteEEEC2IRKNSt3__19nullopt_tEQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asISA_RSD_Entsr3stdE7same_asISA_OSD_Entsr3stdE7same_asISA_NS6_8optionalISB_EEEEEOSB_: 42| 510| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINS_10Array1DRefIKSt4byteEEEC2IS4_Qaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS7_RSA_Entsr3stdE7same_asIS7_OSA_Entsr3stdE7same_asIS7_NSt3__18optionalIS8_EEEEEOS8_: 42| 700| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINS_10Array1DRefIKSt4byteEEEC2IRKNSt3__19nullopt_tEQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asISB_RSE_Entsr3stdE7same_asISB_OSE_Entsr3stdE7same_asISB_NS7_8optionalISC_EEEEEOSC_: 42| 6| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINS_10ByteStreamEEC2IS1_Qaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS4_RS7_Entsr3stdE7same_asIS4_OS7_Entsr3stdE7same_asIS4_NSt3__18optionalIS5_EEEEEOS5_: 42| 6.88k| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINS_10ByteStreamEEC2IRKNSt3__19nullopt_tEQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS8_RSB_Entsr3stdE7same_asIS8_OSB_Entsr3stdE7same_asIS8_NS4_8optionalIS9_EEEEEOS9_: 42| 513| Optional(U&& value) : impl(std::forward(value)) {} _ZNK8rawspeed8OptionalINS_10ByteStreamEEcvbEv: 89| 7.57k| explicit operator bool() const { return has_value(); } _ZN8rawspeed8OptionalINS_10JpegMarkerEEC2IS1_Qaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS4_RS7_Entsr3stdE7same_asIS4_OS7_Entsr3stdE7same_asIS4_NSt3__18optionalIS5_EEEEEOS5_: 42| 13.7k| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINS_10JpegMarkerEEC2Ev: 35| 4.19M| Optional() = default; _ZNO8rawspeed8OptionalINS_10JpegMarkerEEdeEv: 75| 6.88k| T&& operator*() && { 76| 6.88k| invariant(has_value()); ------------------ | | 27| 6.88k|#define invariant(expr) assert(expr) ------------------ | Branch (76:5): [True: 6.88k, False: 0] ------------------ 77| 6.88k| return std::move(impl.value()); 78| 6.88k| } _ZN8rawspeed8OptionalIiEC2IiQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS3_RS6_Entsr3stdE7same_asIS3_OS6_Entsr3stdE7same_asIS3_NSt3__18optionalIS4_EEEEEOS4_: 42| 7| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalIiEC2IRKNSt3__19nullopt_tEQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS7_RSA_Entsr3stdE7same_asIS7_OSA_Entsr3stdE7same_asIS7_NS3_8optionalIS8_EEEEEOS8_: 42| 1| Optional(U&& value) : impl(std::forward(value)) {} _ZNR8rawspeed8OptionalINS_10JpegMarkerEEdeEv: 70| 8| T& operator*() & { 71| 8| invariant(has_value()); ------------------ | | 27| 8|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 8, False: 0] ------------------ 72| 8| return impl.value(); 73| 8| } Cr2Decoder.cpp:_ZN8rawspeed8OptionalINSt3__14pairINS_12_GLOBAL__N_115ColorDataFormatENS0_IiEEEEEC2IS6_Qaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS9_RSC_Entsr3stdE7same_asIS9_OSC_Entsr3stdE7same_asIS9_NS1_8optionalISA_EEEEEOSA_: 42| 16| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalIiEC2IRiQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS4_RS7_Entsr3stdE7same_asIS4_OS7_Entsr3stdE7same_asIS4_NSt3__18optionalIS5_EEEEEOS5_: 42| 16| Optional(U&& value) : impl(std::forward(value)) {} Cr2Decoder.cpp:_ZN8rawspeed8OptionalINSt3__14pairINS_12_GLOBAL__N_115ColorDataFormatENS0_IiEEEEEC2IRKNS1_9nullopt_tEQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asISC_RSF_Entsr3stdE7same_asISC_OSF_Entsr3stdE7same_asISC_NS1_8optionalISD_EEEEEOSD_: 42| 6| Optional(U&& value) : impl(std::forward(value)) {} Cr2Decoder.cpp:_ZNK8rawspeed8OptionalINSt3__14pairINS_12_GLOBAL__N_115ColorDataFormatENS0_IiEEEEEcvbEv: 89| 22| explicit operator bool() const { return has_value(); } Cr2Decoder.cpp:_ZNK8rawspeed8OptionalINSt3__14pairINS_12_GLOBAL__N_115ColorDataFormatENS0_IiEEEEE9has_valueEv: 85| 38| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 38| return impl.has_value(); 87| 38| } Cr2Decoder.cpp:_ZNR8rawspeed8OptionalINSt3__14pairINS_12_GLOBAL__N_115ColorDataFormatENS0_IiEEEEEdeEv: 70| 16| T& operator*() & { 71| 16| invariant(has_value()); ------------------ | | 27| 16|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 16, False: 0] ------------------ 72| 16| return impl.value(); 73| 16| } _ZN8rawspeed8OptionalINSt3__14pairIiiEEEC2IRKNS1_9nullopt_tEQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS9_RSC_Entsr3stdE7same_asIS9_OSC_Entsr3stdE7same_asIS9_NS1_8optionalISA_EEEEEOSA_: 42| 3| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINSt3__14pairIiiEEEC2IS3_Qaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS6_RS9_Entsr3stdE7same_asIS6_OS9_Entsr3stdE7same_asIS6_NS1_8optionalIS7_EEEEEOS7_: 42| 1| Optional(U&& value) : impl(std::forward(value)) {} _ZNK8rawspeed8OptionalINSt3__14pairIiiEEEcvbEv: 89| 4| explicit operator bool() const { return has_value(); } _ZNK8rawspeed8OptionalINSt3__14pairIiiEEE9has_valueEv: 85| 5| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 5| return impl.has_value(); 87| 5| } _ZN8rawspeed8OptionalINSt3__14pairIiiEEEptEv: 60| 1| T* operator->() { 61| 1| invariant(has_value()); ------------------ | | 27| 1|#define invariant(expr) assert(expr) ------------------ | Branch (61:5): [True: 1, False: 0] ------------------ 62| 1| return &impl.value(); 63| 1| } _ZNKR8rawspeed8OptionalIiEdeEv: 65| 6.59k| const T& operator*() const& { 66| 6.59k| invariant(has_value()); ------------------ | | 27| 6.59k|#define invariant(expr) assert(expr) ------------------ | Branch (66:5): [True: 6.59k, False: 0] ------------------ 67| 6.59k| return impl.value(); 68| 6.59k| } _ZN8rawspeed8OptionalIiEC2IjQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS3_RS6_Entsr3stdE7same_asIS3_OS6_Entsr3stdE7same_asIS3_NSt3__18optionalIS4_EEEEEOS4_: 42| 4.03k| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINS_12iRectangle2DEEC2Ev: 35| 3.12k| Optional() = default; _ZNK8rawspeed8OptionalINS_12iRectangle2DEEcvbEv: 89| 79.6k| explicit operator bool() const { return has_value(); } _ZNK8rawspeed8OptionalINS_12iRectangle2DEE9has_valueEv: 85| 156k| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 156k| return impl.has_value(); 87| 156k| } _ZNR8rawspeed8OptionalINS_12iRectangle2DEEdeEv: 70| 75.9k| T& operator*() & { 71| 75.9k| invariant(has_value()); ------------------ | | 27| 75.9k|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 75.9k, False: 0] ------------------ 72| 75.9k| return impl.value(); 73| 75.9k| } _ZN8rawspeed8OptionalINS_12iRectangle2DEEC2IRS1_Qaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS5_RS8_Entsr3stdE7same_asIS5_OS8_Entsr3stdE7same_asIS5_NSt3__18optionalIS6_EEEEEOS6_: 42| 75.9k| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINS_12iRectangle2DEEptEv: 60| 558| T* operator->() { 61| 558| invariant(has_value()); ------------------ | | 27| 558|#define invariant(expr) assert(expr) ------------------ | Branch (61:5): [True: 558, False: 0] ------------------ 62| 558| return &impl.value(); 63| 558| } _ZN8rawspeed8OptionalINS_12iRectangle2DEEC2IS1_Qaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS4_RS7_Entsr3stdE7same_asIS4_OS7_Entsr3stdE7same_asIS4_NSt3__18optionalIS5_EEEEEOS5_: 42| 16| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINS_8CFAColorEEC2IS1_Qaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS4_RS7_Entsr3stdE7same_asIS4_OS7_Entsr3stdE7same_asIS4_NSt3__18optionalIS5_EEEEEOS5_: 42| 1.18k| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINS_8CFAColorEEC2IRKNSt3__19nullopt_tEQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS8_RSB_Entsr3stdE7same_asIS8_OSB_Entsr3stdE7same_asIS8_NS4_8optionalIS9_EEEEEOS9_: 42| 14| Optional(U&& value) : impl(std::forward(value)) {} _ZNK8rawspeed8OptionalINS_8CFAColorEEcvbEv: 89| 1.19k| explicit operator bool() const { return has_value(); } _ZNK8rawspeed8OptionalINS_8CFAColorEE9has_valueEv: 85| 2.37k| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 2.37k| return impl.has_value(); 87| 2.37k| } _ZNR8rawspeed8OptionalINS_8CFAColorEEdeEv: 70| 1.18k| T& operator*() & { 71| 1.18k| invariant(has_value()); ------------------ | | 27| 1.18k|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 1.18k, False: 0] ------------------ 72| 1.18k| return impl.value(); 73| 1.18k| } _ZNK8rawspeed8OptionalINS_12iRectangle2DEEptEv: 55| 16| const T* operator->() const { 56| 16| invariant(has_value()); ------------------ | | 27| 16|#define invariant(expr) assert(expr) ------------------ | Branch (56:5): [True: 16, False: 0] ------------------ 57| 16| return &impl.value(); 58| 16| } _ZNKR8rawspeed8OptionalINS_12iRectangle2DEEdeEv: 65| 8| const T& operator*() const& { 66| 8| invariant(has_value()); ------------------ | | 27| 8|#define invariant(expr) assert(expr) ------------------ | Branch (66:5): [True: 8, False: 0] ------------------ 67| 8| return impl.value(); 68| 8| } _ZN8rawspeed8OptionalINS_20PrefixCodeLUTDecoderINS_10VC5CodeTagENS_23PrefixCodeVectorDecoderIS2_EEEEEC2Ev: 35| 572| Optional() = default; _ZN8rawspeed8OptionalItEC2Ev: 35| 1.13k| Optional() = default; _ZN8rawspeed8OptionalIsEC2Ev: 35| 568| Optional() = default; _ZN8rawspeed8OptionalINS_10BayerPhaseEEC2Ev: 35| 104| Optional() = default; _ZN8rawspeed8OptionalINS_10BayerPhaseEEC2IS1_Qaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS4_RS7_Entsr3stdE7same_asIS4_OS7_Entsr3stdE7same_asIS4_NSt3__18optionalIS5_EEEEEOS5_: 42| 387| Optional(U&& value) : impl(std::forward(value)) {} _ZNK8rawspeed8OptionalINS_10BayerPhaseEEcvbEv: 89| 490| explicit operator bool() const { return has_value(); } _ZNK8rawspeed8OptionalINS_10BayerPhaseEE9has_valueEv: 85| 871| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 871| return impl.has_value(); 87| 871| } _ZNR8rawspeed8OptionalINS_10BayerPhaseEEdeEv: 70| 379| T& operator*() & { 71| 379| invariant(has_value()); ------------------ | | 27| 379|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 379, False: 0] ------------------ 72| 379| return impl.value(); 73| 379| } _ZN8rawspeed8OptionalItEC2IRtQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS4_RS7_Entsr3stdE7same_asIS4_OS7_Entsr3stdE7same_asIS4_NSt3__18optionalIS5_EEEEEOS5_: 42| 60| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalIsEaSIsQsr3stdE7same_asITL0__T_EEERS1_OS4_: 46| 300| Optional& operator=(U&& value) { 47| 300| impl = std::forward(value); 48| 300| return *this; 49| 300| } _ZN8rawspeed8OptionalINS_15VC5Decompressor8BandDataEEC2Ev: 35| 24| Optional() = default; _ZNK8rawspeed8OptionalItE9has_valueEv: 85| 147| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 147| return impl.has_value(); 87| 147| } _ZNR8rawspeed8OptionalItEdeEv: 70| 58| T& operator*() & { 71| 58| invariant(has_value()); ------------------ | | 27| 58|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 58, False: 0] ------------------ 72| 58| return impl.value(); 73| 58| } _ZN8rawspeedeqItiEEbRKNS_8OptionalIT_EERKT0_: 99| 27|bool operator==(const Optional& lhs, const U& rhs) { 100| 27| return lhs && *lhs == rhs; ------------------ | Branch (100:10): [True: 27, False: 0] | Branch (100:17): [True: 4, False: 23] ------------------ 101| 27|} _ZNK8rawspeed8OptionalItEcvbEv: 89| 27| explicit operator bool() const { return has_value(); } _ZNKR8rawspeed8OptionalItEdeEv: 65| 27| const T& operator*() const& { 66| 27| invariant(has_value()); ------------------ | | 27| 27|#define invariant(expr) assert(expr) ------------------ | Branch (66:5): [True: 27, False: 0] ------------------ 67| 27| return impl.value(); 68| 27| } _ZN8rawspeed8OptionalItE5resetEv: 95| 24| void reset() { impl.reset(); } _ZNK8rawspeed8OptionalIsE9has_valueEv: 85| 45| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 45| return impl.has_value(); 87| 45| } _ZNR8rawspeed8OptionalIsEdeEv: 70| 22| T& operator*() & { 71| 22| invariant(has_value()); ------------------ | | 27| 22|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 22, False: 0] ------------------ 72| 22| return impl.value(); 73| 22| } _ZN8rawspeed8OptionalIsE5resetEv: 95| 22| void reset() { impl.reset(); } _ZNK8rawspeed8OptionalINSt3__14pairIPKcPFNS1_10unique_ptrINS_10DngOpcodes9DngOpcodeENS1_14default_deleteIS7_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DEEEEEcvbEv: 89| 1.34k| explicit operator bool() const { return has_value(); } _ZNK8rawspeed8OptionalINSt3__14pairIPKcPFNS1_10unique_ptrINS_10DngOpcodes9DngOpcodeENS1_14default_deleteIS7_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DEEEEE9has_valueEv: 85| 2.66k| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 2.66k| return impl.has_value(); 87| 2.66k| } _ZNR8rawspeed8OptionalINSt3__14pairIPKcPFNS1_10unique_ptrINS_10DngOpcodes9DngOpcodeENS1_14default_deleteIS7_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DEEEEEdeEv: 70| 1.32k| T& operator*() & { 71| 1.32k| invariant(has_value()); ------------------ | | 27| 1.32k|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 1.32k, False: 0] ------------------ 72| 1.32k| return impl.value(); 73| 1.32k| } _ZN8rawspeed8OptionalINSt3__14pairIPKcPFNS1_10unique_ptrINS_10DngOpcodes9DngOpcodeENS1_14default_deleteIS7_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DEEEEEC2INS2_IS4_DnEEQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asISO_RSR_Entsr3stdE7same_asISO_OSR_Entsr3stdE7same_asISO_NS1_8optionalISP_EEEEEOSP_: 42| 50| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINSt3__14pairIPKcPFNS1_10unique_ptrINS_10DngOpcodes9DngOpcodeENS1_14default_deleteIS7_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DEEEEEC2ISK_Qaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asISN_RSQ_Entsr3stdE7same_asISN_OSQ_Entsr3stdE7same_asISN_NS1_8optionalISO_EEEEEOSO_: 42| 1.27k| Optional(U&& value) : impl(std::forward(value)) {} _ZN8rawspeed8OptionalINSt3__14pairIPKcPFNS1_10unique_ptrINS_10DngOpcodes9DngOpcodeENS1_14default_deleteIS7_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DEEEEEC2IRKNS1_9nullopt_tEQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asISQ_RST_Entsr3stdE7same_asISQ_OST_Entsr3stdE7same_asISQ_NS1_8optionalISR_EEEEEOSR_: 42| 20| Optional(U&& value) : impl(std::forward(value)) {} IiqDecoder.cpp:_ZN8rawspeed8OptionalINS_12_GLOBAL__N_19IIQFormatEEC2Ev: 35| 1.12k| Optional() = default; _ZN8rawspeed8OptionalINS_6BufferEEC2Ev: 35| 1.12k| Optional() = default; IiqDecoder.cpp:_ZNK8rawspeed8OptionalINS_12_GLOBAL__N_19IIQFormatEEcvbEv: 89| 2.83k| explicit operator bool() const { return has_value(); } IiqDecoder.cpp:_ZNK8rawspeed8OptionalINS_12_GLOBAL__N_19IIQFormatEE9has_valueEv: 85| 3.76k| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 3.76k| return impl.has_value(); 87| 3.76k| } IiqDecoder.cpp:_ZN8rawspeed8OptionalINS_12_GLOBAL__N_19IIQFormatEEC2IS2_Qaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS5_RS8_Entsr3stdE7same_asIS5_OS8_Entsr3stdE7same_asIS5_NSt3__18optionalIS6_EEEEEOS6_: 42| 931| Optional(U&& value) : impl(std::forward(value)) {} IiqDecoder.cpp:_ZN8rawspeed8OptionalINS_12_GLOBAL__N_19IIQFormatEEC2IRKNSt3__19nullopt_tEQaaaaaantsr3stdE7same_asITL0__NS0_IT_EEEntsr3stdE7same_asIS9_RSC_Entsr3stdE7same_asIS9_OSC_Entsr3stdE7same_asIS9_NS5_8optionalISA_EEEEEOSA_: 42| 27| Optional(U&& value) : impl(std::forward(value)) {} IiqDecoder.cpp:_ZNR8rawspeed8OptionalINS_12_GLOBAL__N_19IIQFormatEEdeEv: 70| 931| T& operator*() & { 71| 931| invariant(has_value()); ------------------ | | 27| 931|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 931, False: 0] ------------------ 72| 931| return impl.value(); 73| 931| } _ZN8rawspeed8OptionalINS_6BufferEEaSIS1_Qsr3stdE7same_asITL0__T_EEERS2_OS5_: 46| 1.09k| Optional& operator=(U&& value) { 47| 1.09k| impl = std::forward(value); 48| 1.09k| return *this; 49| 1.09k| } _ZNK8rawspeed8OptionalINS_6BufferEEcvbEv: 89| 907| explicit operator bool() const { return has_value(); } _ZNK8rawspeed8OptionalINS_6BufferEE9has_valueEv: 85| 2.64k| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 2.64k| return impl.has_value(); 87| 2.64k| } _ZN8rawspeed8OptionalINS_6BufferEEptEv: 60| 870| T* operator->() { 61| 870| invariant(has_value()); ------------------ | | 27| 870|#define invariant(expr) assert(expr) ------------------ | Branch (61:5): [True: 870, False: 0] ------------------ 62| 870| return &impl.value(); 63| 870| } _ZNR8rawspeed8OptionalINS_6BufferEEdeEv: 70| 870| T& operator*() & { 71| 870| invariant(has_value()); ------------------ | | 27| 870|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 870, False: 0] ------------------ 72| 870| return impl.value(); 73| 870| } _ZN8rawspeed8OptionalINS_11HuffmanCodeINS_15BaselineCodeTagEEEEC2Ev: 35| 174| Optional() = default; _ZN8rawspeed8OptionalINS_11HuffmanCodeINS_15BaselineCodeTagEEEEaSIS3_Qsr3stdE7same_asITL0__T_EEERS4_OS7_: 46| 162| Optional& operator=(U&& value) { 47| 162| impl = std::forward(value); 48| 162| return *this; 49| 162| } _ZNR8rawspeed8OptionalINS_11HuffmanCodeINS_15BaselineCodeTagEEEEdeEv: 70| 162| T& operator*() & { 71| 162| invariant(has_value()); ------------------ | | 27| 162|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 162, False: 0] ------------------ 72| 162| return impl.value(); 73| 162| } _ZNK8rawspeed8OptionalINS_11HuffmanCodeINS_15BaselineCodeTagEEEE9has_valueEv: 85| 162| [[nodiscard]] bool has_value() const RAWSPEED_READNONE { 86| 162| return impl.has_value(); 87| 162| } _ZN8rawspeedeqINS_10BayerPhaseES1_EEbRKNS_8OptionalIT_EERKT0_: 99| 5|bool operator==(const Optional& lhs, const U& rhs) { 100| 5| return lhs && *lhs == rhs; ------------------ | Branch (100:10): [True: 5, False: 0] | Branch (100:17): [True: 2, False: 3] ------------------ 101| 5|} _ZNKR8rawspeed8OptionalINS_10BayerPhaseEEdeEv: 65| 5| const T& operator*() const& { 66| 5| invariant(has_value()); ------------------ | | 27| 5|#define invariant(expr) assert(expr) ------------------ | Branch (66:5): [True: 5, False: 0] ------------------ 67| 5| return impl.value(); 68| 5| } _ZN8rawspeed8iPoint2DC2Eii: 41| 1.80M| constexpr iPoint2D(value_type a, value_type b) : x(a), y(b) {} _ZNK8rawspeed8iPoint2DplERKS0_: 43| 517k| constexpr iPoint2D operator+(const iPoint2D& rhs) const { 44| 517k| return {x + rhs.x, y + rhs.y}; 45| 517k| } _ZNK8rawspeed8iPoint2DmiERKS0_: 46| 3.14k| constexpr iPoint2D operator-(const iPoint2D& rhs) const { 47| 3.14k| return {x - rhs.x, y - rhs.y}; 48| 3.14k| } _ZN8rawspeed8iPoint2DpLERKS0_: 50| 169| iPoint2D& operator+=(const iPoint2D& rhs) { 51| 169| *this = operator+(rhs); 52| 169| return *this; 53| 169| } _ZNK8rawspeed8iPoint2DeqERKS0_: 59| 182k| constexpr bool RAWSPEED_READONLY operator==(const iPoint2D& rhs) const { 60| 182k| return x == rhs.x && y == rhs.y; ------------------ | Branch (60:12): [True: 46.1k, False: 136k] | Branch (60:26): [True: 45.8k, False: 360] ------------------ 61| 182k| } _ZNK8rawspeed8iPoint2DgtERKS0_: 63| 19.9k| constexpr bool RAWSPEED_READONLY operator>(const iPoint2D& rhs) const { 64| 19.9k| return x > rhs.x && y > rhs.y; ------------------ | Branch (64:12): [True: 19.7k, False: 128] | Branch (64:25): [True: 19.6k, False: 136] ------------------ 65| 19.9k| } _ZNK8rawspeed8iPoint2DltERKS0_: 66| 2.33k| constexpr bool RAWSPEED_READONLY operator<(const iPoint2D& rhs) const { 67| 2.33k| return x < rhs.x && y < rhs.y; ------------------ | Branch (67:12): [True: 1.76k, False: 569] | Branch (67:25): [True: 1.73k, False: 34] ------------------ 68| 2.33k| } _ZNK8rawspeed8iPoint2DgeERKS0_: 70| 12.2k| constexpr bool RAWSPEED_READONLY operator>=(const iPoint2D& rhs) const { 71| 12.2k| return x >= rhs.x && y >= rhs.y; ------------------ | Branch (71:12): [True: 12.0k, False: 199] | Branch (71:26): [True: 11.8k, False: 167] ------------------ 72| 12.2k| } _ZNK8rawspeed8iPoint2DleERKS0_: 73| 84.8k| constexpr bool RAWSPEED_READONLY operator<=(const iPoint2D& rhs) const { 74| 84.8k| return x <= rhs.x && y <= rhs.y; ------------------ | Branch (74:12): [True: 83.9k, False: 896] | Branch (74:26): [True: 83.6k, False: 272] ------------------ 75| 84.8k| } _ZNK8rawspeed8iPoint2D15hasPositiveAreaEv: 77| 19.9k| [[nodiscard]] bool RAWSPEED_READONLY hasPositiveArea() const { 78| 19.9k| return *this > iPoint2D(0, 0); 79| 19.9k| } _ZNK8rawspeed8iPoint2D4areaEv: 81| 30.9k| [[nodiscard]] area_type RAWSPEED_READONLY area() const { 82| 30.9k| using signed_area = std::make_signed_t; 83| | 84| 30.9k| area_type x_abs = std::abs(static_cast(x)); 85| 30.9k| area_type y_abs = std::abs(static_cast(y)); 86| | 87| 30.9k| return x_abs * y_abs; 88| 30.9k| } _ZNK8rawspeed8iPoint2D12isThisInsideERKS0_: 91| 394| isThisInside(const iPoint2D& rhs) const { 92| 394| return *this <= rhs; 93| 394| } _ZN8rawspeed12iRectangle2DC2ERKNS_8iPoint2DES3_: 111| 401k| : pos(pos_), dim(dim_) {} _ZN8rawspeed12iRectangle2DC2Eiiii: 115| 713| : pos({x_pos, y_pos}), dim({w, h}) {} _ZNK8rawspeed12iRectangle2D6getTopEv: 117| 1.48M| [[nodiscard]] constexpr int getTop() const { return pos.y; } _ZNK8rawspeed12iRectangle2D9getBottomEv: 118| 1.66k| [[nodiscard]] constexpr int getBottom() const { return pos.y + dim.y; } _ZNK8rawspeed12iRectangle2D7getLeftEv: 119| 1.10M| [[nodiscard]] constexpr int getLeft() const { return pos.x; } _ZNK8rawspeed12iRectangle2D8getRightEv: 120| 68.9k| [[nodiscard]] constexpr int getRight() const { return pos.x + dim.x; } _ZNK8rawspeed12iRectangle2D8getWidthEv: 121| 1.17M| [[nodiscard]] constexpr int getWidth() const { return dim.x; } _ZNK8rawspeed12iRectangle2D9getHeightEv: 122| 641k| [[nodiscard]] constexpr int getHeight() const { return dim.y; } _ZNK8rawspeed12iRectangle2D10getTopLeftEv: 123| 92.7k| [[nodiscard]] constexpr iPoint2D getTopLeft() const { return pos; } _ZNK8rawspeed12iRectangle2D14getBottomRightEv: 124| 185k| [[nodiscard]] constexpr iPoint2D getBottomRight() const { return pos + dim; } _ZNK8rawspeed12iRectangle2D11getTopRightEv: 125| 19.6k| [[nodiscard]] constexpr iPoint2D getTopRight() const { 126| 19.6k| return pos + iPoint2D(dim.x, 0); 127| 19.6k| } _ZNK8rawspeed12iRectangle2D13getBottomLeftEv: 128| 312k| [[nodiscard]] constexpr iPoint2D getBottomLeft() const { 129| 312k| return pos + iPoint2D(0, dim.y); 130| 312k| } _ZNK8rawspeed12iRectangle2D15hasPositiveAreaEv: 131| 572| [[nodiscard]] constexpr bool RAWSPEED_READONLY hasPositiveArea() const { 132| 572| return (dim.x > 0) && (dim.y > 0); ------------------ | Branch (132:12): [True: 521, False: 51] | Branch (132:27): [True: 498, False: 23] ------------------ 133| 572| } _ZNK8rawspeed12iRectangle2D13isPointInsideERKNS_8iPoint2DE: 136| 1.21k| isPointInside(const iPoint2D& subPoint) const { 137| 1.21k| return subPoint >= getTopLeft() && subPoint < getBottomRight(); ------------------ | Branch (137:12): [True: 1.18k, False: 26] | Branch (137:40): [True: 1.14k, False: 45] ------------------ 138| 1.21k| } _ZNK8rawspeed12iRectangle2D22isPointInsideInclusiveERKNS_8iPoint2DE: 141| 5.55k| isPointInsideInclusive(const iPoint2D& subPoint) const { 142| 5.55k| return subPoint >= getTopLeft() && subPoint <= getBottomRight(); ------------------ | Branch (142:12): [True: 5.22k, False: 338] | Branch (142:40): [True: 4.73k, False: 485] ------------------ 143| 5.55k| } _ZNK8rawspeed12iRectangle2D12isThisInsideERKS0_: 146| 2.68k| isThisInside(const iRectangle2D& superRect) const { 147| 2.68k| return getTopLeft() >= superRect.getTopLeft() && ------------------ | Branch (147:12): [True: 2.68k, False: 0] ------------------ 148| 2.68k| getBottomRight() <= superRect.getBottomRight(); ------------------ | Branch (148:12): [True: 2.68k, False: 0] ------------------ 149| 2.68k| } _ZN8rawspeed12iRectangle2D10setTopLeftERKNS_8iPoint2DE: 158| 1.62k| void setTopLeft(const iPoint2D& top_left) { pos = top_left; } _ZN8rawspeed12iRectangle2D22setBottomRightAbsoluteERKNS_8iPoint2DE: 161| 1.62k| void setBottomRightAbsolute(const iPoint2D& bottom_right) { 162| 1.62k| dim = bottom_right - pos; 163| 1.62k| } _ZN8rawspeedeqERKNS_12iRectangle2DES0_: 225| 1.62k| const iRectangle2D b) { 226| 1.62k| return std::tie(a.pos, a.dim) == std::tie(b.pos, b.dim); 227| 1.62k|} _ZN8rawspeed8iPoint2DC2Ev: 40| 63.1k| constexpr iPoint2D() = default; _ZN8rawspeed12iRectangle2DC2Ev: 109| 4.48k| constexpr iRectangle2D() = default; _ZN8rawspeed13RangesOverlapINS_6BufferEEEbRKT_S4_: 62| 11.9k|constexpr bool RAWSPEED_READNONE RangesOverlap(const T& lhs, const T& rhs) { 63| 11.9k| if (&lhs == &rhs) ------------------ | Branch (63:7): [True: 0, False: 11.9k] ------------------ 64| 0| return true; 65| | 66| 11.9k| if (std::begin(lhs) == std::begin(rhs)) ------------------ | Branch (66:7): [True: 829, False: 11.1k] ------------------ 67| 829| return true; 68| | 69| 11.1k| const std::pair ordered = 70| 11.1k| std::minmax(lhs, rhs, [](const T& r0, const T& r1) { 71| 11.1k| assert(std::begin(r0) != std::begin(r1)); 72| 11.1k| return std::begin(r0) < std::begin(r1); 73| 11.1k| }); 74| | 75| 11.1k| assert(std::begin(ordered.first) < std::begin(ordered.second)); ------------------ | Branch (75:3): [True: 11.1k, False: 0] ------------------ 76| 11.1k| return RangeContains(ordered.first, std::begin(ordered.second)); 77| 11.1k|} _ZZN8rawspeed13RangesOverlapINS_6BufferEEEbRKT_S4_ENKUlRKS1_S6_E_clES6_S6_: 70| 11.1k| std::minmax(lhs, rhs, [](const T& r0, const T& r1) { 71| 11.1k| assert(std::begin(r0) != std::begin(r1)); ------------------ | Branch (71:9): [True: 11.1k, False: 0] ------------------ 72| 11.1k| return std::begin(r0) < std::begin(r1); 73| 11.1k| }); _ZN8rawspeed13RangeContainsINS_6BufferEPKhEEbRKT_T0_: 53| 11.1k|constexpr bool RAWSPEED_READNONE RangeContains(const Tr& r, Tv pos) { 54| 11.1k| if (pos < std::begin(r)) ------------------ | Branch (54:7): [True: 0, False: 11.1k] ------------------ 55| 0| return false; 56| | 57| 11.1k| assert(pos >= std::begin(r)); ------------------ | Branch (57:3): [True: 11.1k, False: 0] ------------------ 58| 11.1k| return std::end(r) > pos; 59| 11.1k|} _ZN8rawspeed32variableLengthLoadNaiveViaMemcpyENS_10Array1DRefISt4byteEENS0_IKS1_EEi: 150| 8.02k| int inPos) { 151| 8.02k| invariant(out.size() != 0); ------------------ | | 27| 8.02k|#define invariant(expr) assert(expr) ------------------ | Branch (151:3): [True: 8.02k, False: 0] ------------------ 152| 8.02k| invariant(in.size() != 0); ------------------ | | 27| 8.02k|#define invariant(expr) assert(expr) ------------------ | Branch (152:3): [True: 8.02k, False: 1] ------------------ 153| 8.02k| invariant(out.size() <= in.size()); ------------------ | | 27| 8.02k|#define invariant(expr) assert(expr) ------------------ | Branch (153:3): [True: 8.01k, False: 7] ------------------ 154| 8.01k| invariant(inPos >= 0); ------------------ | | 27| 8.01k|#define invariant(expr) assert(expr) ------------------ | Branch (154:3): [True: 8.01k, False: 0] ------------------ 155| | 156| 8.01k| std::fill(out.begin(), out.end(), std::byte{0x00}); 157| | 158| 8.01k| inPos = std::min(inPos, in.size()); 159| | 160| 8.01k| int inPosEnd = inPos + out.size(); 161| 8.01k| inPosEnd = std::min(inPosEnd, in.size()); 162| 8.01k| invariant(inPos <= inPosEnd); ------------------ | | 27| 8.01k|#define invariant(expr) assert(expr) ------------------ | Branch (162:3): [True: 8.01k, False: 2] ------------------ 163| | 164| 8.01k| const int copySize = inPosEnd - inPos; 165| 8.01k| invariant(copySize >= 0); ------------------ | | 27| 8.01k|#define invariant(expr) assert(expr) ------------------ | Branch (165:3): [True: 8.01k, False: 2] ------------------ 166| 8.01k| invariant(copySize <= out.size()); ------------------ | | 27| 8.01k|#define invariant(expr) assert(expr) ------------------ | Branch (166:3): [True: 8.01k, False: 0] ------------------ 167| | 168| 8.01k| out = out.getCrop(/*offset=*/0, copySize).getAsArray1DRef(); 169| 8.01k| in = in.getCrop(/*offset=*/inPos, copySize).getAsArray1DRef(); 170| 8.01k| invariant(in.size() == out.size()); ------------------ | | 27| 8.01k|#define invariant(expr) assert(expr) ------------------ | Branch (170:3): [True: 8.02k, False: 18.4E] ------------------ 171| | 172| 8.02k| memcpy(out.begin(), in.begin(), copySize); 173| 8.02k|} _ZN8rawspeed14iterator_rangeINS_16Cr2SliceIteratorEEC2ES1_S1_: 33| 1.08k| : begin_iterator(std::move(begin_iterator_)), 34| 1.08k| end_iterator(std::move(end_iterator_)) {} _ZNK8rawspeed14iterator_rangeINS_16Cr2SliceIteratorEE5beginEv: 36| 1.08k| [[nodiscard]] Iter begin() const { return begin_iterator; } _ZNK8rawspeed14iterator_rangeINS_16Cr2SliceIteratorEE3endEv: 37| 1.08k| [[nodiscard]] Iter end() const { return end_iterator; } _ZN8rawspeed14iterator_rangeINS_21Cr2OutputTileIteratorEEC2ES1_S1_: 33| 1.59k| : begin_iterator(std::move(begin_iterator_)), 34| 1.59k| end_iterator(std::move(end_iterator_)) {} _ZNK8rawspeed14iterator_rangeINS_21Cr2OutputTileIteratorEE5beginEv: 36| 1.59k| [[nodiscard]] Iter begin() const { return begin_iterator; } _ZNK8rawspeed14iterator_rangeINS_21Cr2OutputTileIteratorEE3endEv: 37| 2.61k| [[nodiscard]] Iter end() const { return end_iterator; } _ZN8rawspeed14iterator_rangeINS_30Cr2VerticalOutputStripIteratorEEC2ES1_S1_: 33| 512| : begin_iterator(std::move(begin_iterator_)), 34| 512| end_iterator(std::move(end_iterator_)) {} _ZNK8rawspeed14iterator_rangeINS_30Cr2VerticalOutputStripIteratorEE5beginEv: 36| 512| [[nodiscard]] Iter begin() const { return begin_iterator; } _ZNK8rawspeed14iterator_rangeINS_30Cr2VerticalOutputStripIteratorEE3endEv: 37| 512| [[nodiscard]] Iter end() const { return end_iterator; } _ZN8rawspeed28BitStreamCacheLeftInRightOut4pushEmi: 60| 4.13M| void push(uint64_t bits, int count) noexcept { 61| 4.13M| establishClassInvariants(); 62| 4.13M| invariant(count >= 0); ------------------ | | 27| 4.13M|#define invariant(expr) assert(expr) ------------------ | Branch (62:5): [True: 4.19M, False: 18.4E] ------------------ 63| | // NOTE: count may be zero! 64| 4.19M| invariant(count <= Size); ------------------ | | 27| 4.19M|#define invariant(expr) assert(expr) ------------------ | Branch (64:5): [True: 4.19M, False: 18.4E] ------------------ 65| 4.19M| invariant(count + fillLevel <= Size); ------------------ | | 27| 4.19M|#define invariant(expr) assert(expr) ------------------ | Branch (65:5): [True: 4.20M, False: 18.4E] ------------------ 66| 4.20M| cache |= bits << fillLevel; 67| 4.20M| fillLevel += count; 68| 4.20M| } _ZNK8rawspeed28BitStreamCacheLeftInRightOut4peekEi: 70| 10.1M| [[nodiscard]] uint32_t peek(int count) const noexcept { 71| 10.1M| establishClassInvariants(); 72| 10.1M| invariant(count >= 0); ------------------ | | 27| 10.1M|#define invariant(expr) assert(expr) ------------------ | Branch (72:5): [True: 10.2M, False: 18.4E] ------------------ 73| 10.2M| invariant(count <= MaxGetBits); ------------------ | | 27| 10.2M|#define invariant(expr) assert(expr) ------------------ | Branch (73:5): [True: 10.2M, False: 18.4E] ------------------ 74| 10.2M| invariant(count != 0); ------------------ | | 27| 10.2M|#define invariant(expr) assert(expr) ------------------ | Branch (74:5): [True: 10.3M, False: 18.4E] ------------------ 75| 10.3M| invariant(count <= Size); ------------------ | | 27| 10.3M|#define invariant(expr) assert(expr) ------------------ | Branch (75:5): [True: 10.3M, False: 0] ------------------ 76| 10.3M| invariant(count <= fillLevel); ------------------ | | 27| 10.3M|#define invariant(expr) assert(expr) ------------------ | Branch (76:5): [True: 10.3M, False: 18.4E] ------------------ 77| 10.3M| return extractLowBits(static_cast(cache), count); 78| 10.3M| } _ZN8rawspeed28BitStreamCacheLeftInRightOut4skipEi: 80| 10.8M| void skip(int count) noexcept { 81| 10.8M| establishClassInvariants(); 82| 10.8M| invariant(count >= 0); ------------------ | | 27| 10.8M|#define invariant(expr) assert(expr) ------------------ | Branch (82:5): [True: 10.7M, False: 77.1k] ------------------ 83| | // `count` *could* be larger than `MaxGetBits`. 84| | // `count` could be zero. 85| 10.7M| invariant(count <= Size); ------------------ | | 27| 10.7M|#define invariant(expr) assert(expr) ------------------ | Branch (85:5): [True: 10.7M, False: 18.4E] ------------------ 86| 10.7M| invariant(count <= fillLevel); ------------------ | | 27| 10.7M|#define invariant(expr) assert(expr) ------------------ | Branch (86:5): [True: 10.8M, False: 18.4E] ------------------ 87| 10.8M| cache >>= count; 88| 10.8M| fillLevel -= count; 89| 10.8M| } _ZN8rawspeed28BitStreamCacheRightInLeftOut4pushEmi: 93| 2.85M| void push(uint64_t bits, int count) noexcept { 94| 2.85M| establishClassInvariants(); 95| 2.85M| invariant(count >= 0); ------------------ | | 27| 2.85M|#define invariant(expr) assert(expr) ------------------ | Branch (95:5): [True: 2.85M, False: 18.4E] ------------------ 96| | // NOTE: count may be zero! 97| 2.85M| invariant(count <= Size); ------------------ | | 27| 2.85M|#define invariant(expr) assert(expr) ------------------ | Branch (97:5): [True: 2.85M, False: 18.4E] ------------------ 98| 2.85M| invariant(count + fillLevel <= Size); ------------------ | | 27| 2.85M|#define invariant(expr) assert(expr) ------------------ | Branch (98:5): [True: 2.85M, False: 18.4E] ------------------ 99| | // If the maximal size of the cache is BitStreamCacheBase::Size, and we 100| | // have fillLevel [high] bits set, how many empty [low] bits do we have? 101| 2.85M| const int vacantBits = BitStreamCacheBase::Size - fillLevel; 102| 2.85M| invariant(vacantBits >= 0); ------------------ | | 27| 2.85M|#define invariant(expr) assert(expr) ------------------ | Branch (102:5): [True: 2.85M, False: 2] ------------------ 103| 2.85M| invariant(vacantBits <= Size); ------------------ | | 27| 2.85M|#define invariant(expr) assert(expr) ------------------ | Branch (103:5): [True: 2.85M, False: 18.4E] ------------------ 104| 2.85M| invariant(vacantBits != 0); ------------------ | | 27| 2.85M|#define invariant(expr) assert(expr) ------------------ | Branch (104:5): [True: 2.85M, False: 0] ------------------ 105| 2.85M| invariant(vacantBits >= count); ------------------ | | 27| 2.85M|#define invariant(expr) assert(expr) ------------------ | Branch (105:5): [True: 2.85M, False: 1] ------------------ 106| | // If we just directly 'or' these low bits into the cache right now, 107| | // how many unfilled bits of a gap will there be in the middle of a cache? 108| 2.85M| const int emptyBitsGap = vacantBits - count; 109| 2.85M| invariant(emptyBitsGap >= 0); ------------------ | | 27| 2.85M|#define invariant(expr) assert(expr) ------------------ | Branch (109:5): [True: 2.85M, False: 0] ------------------ 110| 2.85M| invariant(emptyBitsGap <= Size); ------------------ | | 27| 2.85M|#define invariant(expr) assert(expr) ------------------ | Branch (110:5): [True: 2.85M, False: 18.4E] ------------------ 111| 2.85M| if (count != 0) { ------------------ | Branch (111:9): [True: 2.85M, False: 18.4E] ------------------ 112| 2.85M| invariant(emptyBitsGap < Size); ------------------ | | 27| 2.85M|#define invariant(expr) assert(expr) ------------------ | Branch (112:7): [True: 2.85M, False: 18.4E] ------------------ 113| | // So just shift the new bits so that there is no gap in the middle. 114| 2.85M| cache |= bits << emptyBitsGap; 115| 2.85M| } 116| 2.85M| fillLevel += count; 117| 2.85M| } _ZNK8rawspeed28BitStreamCacheRightInLeftOut4peekEi: 119| 21.1M| [[nodiscard]] auto peek(int count) const noexcept { 120| 21.1M| establishClassInvariants(); 121| 21.1M| invariant(count >= 0); ------------------ | | 27| 21.1M|#define invariant(expr) assert(expr) ------------------ | Branch (121:5): [True: 21.1M, False: 21.6k] ------------------ 122| 21.1M| invariant(count <= Size); ------------------ | | 27| 21.1M|#define invariant(expr) assert(expr) ------------------ | Branch (122:5): [True: 21.1M, False: 1] ------------------ 123| 21.1M| invariant(count <= MaxGetBits); ------------------ | | 27| 21.1M|#define invariant(expr) assert(expr) ------------------ | Branch (123:5): [True: 21.1M, False: 18.4E] ------------------ 124| 21.1M| invariant(count != 0); ------------------ | | 27| 21.1M|#define invariant(expr) assert(expr) ------------------ | Branch (124:5): [True: 21.1M, False: 0] ------------------ 125| 21.1M| invariant(count <= fillLevel); ------------------ | | 27| 21.1M|#define invariant(expr) assert(expr) ------------------ | Branch (125:5): [True: 21.1M, False: 18.4E] ------------------ 126| 21.1M| return implicit_cast( 127| 21.1M| extractHighBits(cache, count, 128| 21.1M| /*effectiveBitwidth=*/BitStreamCacheBase::Size)); 129| 21.1M| } _ZN8rawspeed28BitStreamCacheRightInLeftOut4skipEi: 131| 21.3M| void skip(int count) noexcept { 132| 21.3M| establishClassInvariants(); 133| 21.3M| invariant(count >= 0); ------------------ | | 27| 21.3M|#define invariant(expr) assert(expr) ------------------ | Branch (133:5): [True: 21.3M, False: 1.26k] ------------------ 134| | // `count` *could* be larger than `MaxGetBits`. 135| | // `count` could be zero. 136| 21.3M| invariant(count <= Size); ------------------ | | 27| 21.3M|#define invariant(expr) assert(expr) ------------------ | Branch (136:5): [True: 21.3M, False: 18.4E] ------------------ 137| 21.3M| invariant(count <= fillLevel); ------------------ | | 27| 21.3M|#define invariant(expr) assert(expr) ------------------ | Branch (137:5): [True: 21.3M, False: 18.4E] ------------------ 138| 21.3M| fillLevel -= count; 139| 21.3M| cache <<= count; 140| 21.3M| } _ZNK8rawspeed18BitStreamCacheBase24establishClassInvariantsEv: 54| 178M|BitStreamCacheBase::establishClassInvariants() const noexcept { 55| 178M| invariant(fillLevel >= 0); ------------------ | | 27| 178M|#define invariant(expr) assert(expr) ------------------ | Branch (55:3): [True: 178M, False: 18.4E] ------------------ 56| 178M| invariant(fillLevel <= Size); ------------------ | | 27| 178M|#define invariant(expr) assert(expr) ------------------ | Branch (56:3): [True: 179M, False: 18.4E] ------------------ 57| 178M|} _ZN8rawspeed11BitStreamerINS_14BitStreamerMSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEEC2ENS_10Array1DRefIKSt4byteEE: 192| 3.58k| explicit BitStreamer(Array1DRef input) : replenisher(input) { 193| 3.58k| establishClassInvariants(); 194| 3.58k| } _ZN8rawspeed26BitStreamerReplenisherBaseINS_14BitStreamerMSBEEC2ENS_10Array1DRefIKSt4byteEE: 57| 3.58k| : input(input_) { 58| 3.58k| if (input.size() < BitStreamerTraits::MaxProcessBytes) ------------------ | Branch (58:9): [True: 70, False: 3.51k] ------------------ 59| 3.58k| ThrowIOE("Bit stream size is smaller than MaxProcessBytes"); ------------------ | | 37| 70|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 70| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 70| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 70| __VA_ARGS__) | | ------------------ ------------------ 60| 3.58k| } _ZNK8rawspeed11BitStreamerINS_14BitStreamerMSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE24establishClassInvariantsEv: 185| 40.5M| void establishClassInvariants() const noexcept { 186| 40.5M| cache.establishClassInvariants(); 187| 40.5M| replenisher.establishClassInvariants(); 188| 40.5M| } _ZNK8rawspeed26BitStreamerReplenisherBaseINS_14BitStreamerMSBEE24establishClassInvariantsEv: 65| 47.5M|BitStreamerReplenisherBase::establishClassInvariants() const noexcept { 66| 47.5M| input.establishClassInvariants(); 67| 47.5M| invariant(input.size() >= BitStreamerTraits::MaxProcessBytes); ------------------ | | 27| 47.5M|#define invariant(expr) assert(expr) ------------------ | Branch (67:3): [True: 47.5M, False: 1] ------------------ 68| 47.5M| invariant(pos >= 0); ------------------ | | 27| 47.5M|#define invariant(expr) assert(expr) ------------------ | Branch (68:3): [True: 47.5M, False: 0] ------------------ 69| 47.5M| invariant(pos % StreamTraits::MinLoadStepByteMultiple == 0); ------------------ | | 27| 47.5M|#define invariant(expr) assert(expr) ------------------ | Branch (69:3): [True: 47.5M, False: 0] ------------------ 70| | // `pos` *could* be out-of-bounds of `input`. 71| 47.5M|} _ZN8rawspeed11BitStreamerINS_14BitStreamerMSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE4fillEi: 216| 7.61M| void fill(int nbits = Cache::MaxGetBits) { 217| 7.61M| establishClassInvariants(); 218| 7.61M| invariant(nbits >= 0); ------------------ | | 27| 7.61M|#define invariant(expr) assert(expr) ------------------ | Branch (218:5): [True: 7.61M, False: 0] ------------------ 219| 7.61M| invariant(nbits != 0); ------------------ | | 27| 7.61M|#define invariant(expr) assert(expr) ------------------ | Branch (219:5): [True: 7.61M, False: 0] ------------------ 220| 7.61M| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 7.61M|#define invariant(expr) assert(expr) ------------------ | Branch (220:5): [True: 7.61M, False: 0] ------------------ 221| | 222| 7.61M| if (cache.fillLevel >= nbits) ------------------ | Branch (222:9): [True: 5.85M, False: 1.76M] ------------------ 223| 5.85M| return; 224| | 225| 1.76M| const auto input = replenisher.getInput(); 226| 1.76M| const auto numBytes = static_cast(this)->fillCache(input); 227| 1.76M| replenisher.markNumBytesAsConsumed(numBytes); 228| | invariant(cache.fillLevel >= nbits); ------------------ | | 27| 1.76M|#define invariant(expr) assert(expr) ------------------ | Branch (228:5): [True: 1.75M, False: 257] ------------------ 229| 1.76M| } _ZN8rawspeed39BitStreamerForwardSequentialReplenisherINS_14BitStreamerMSBEE8getInputEv: 100| 1.76M| std::array::MaxProcessBytes> getInput() { 101| 1.76M| Base::establishClassInvariants(); 102| | 103| 1.76M| std::array::MaxProcessBytes> tmpStorage; 104| 1.76M| auto tmp = Array1DRef(tmpStorage.data(), 105| 1.76M| implicit_cast(tmpStorage.size())); 106| | 107| | // Do we have BitStreamerTraits::MaxProcessBytes or more bytes left in 108| | // the input buffer? If so, then we can just read from said buffer. 109| 1.76M| if (getPos() + BitStreamerTraits::MaxProcessBytes <= ------------------ | Branch (109:9): [True: 1.75M, False: 3.15k] ------------------ 110| 1.76M| Base::input.size()) [[likely]] { 111| 1.75M| auto currInput = 112| 1.75M| Base::input.getCrop(getPos(), BitStreamerTraits::MaxProcessBytes) 113| 1.75M| .getAsArray1DRef(); 114| 1.75M| invariant(currInput.size() == tmp.size()); ------------------ | | 27| 1.75M|#define invariant(expr) assert(expr) ------------------ | Branch (114:7): [True: 1.75M, False: 0] ------------------ 115| 1.75M| memcpy(tmp.begin(), currInput.begin(), 116| 1.75M| BitStreamerTraits::MaxProcessBytes); 117| 1.75M| return tmpStorage; 118| 1.75M| } 119| | 120| | // We have to use intermediate buffer, either because the input is running 121| | // out of bytes, or because we want to enforce bounds checking. 122| | 123| | // Note that in order to keep all fill-level invariants we must allow to 124| | // over-read past-the-end a bit. 125| 3.15k| if (getPos() > Base::input.size() + ------------------ | Branch (125:9): [True: 257, False: 2.90k] ------------------ 126| 3.15k| 2 * BitStreamerTraits::MaxProcessBytes) [[unlikely]] 127| 257| ThrowIOE("Buffer overflow read in BitStreamer"); ------------------ | | 37| 257|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 257| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 257| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 257| __VA_ARGS__) | | ------------------ ------------------ 128| | 129| 2.90k| variableLengthLoadNaiveViaMemcpy(tmp, Base::input, getPos()); 130| | 131| 2.90k| return tmpStorage; 132| 3.15k| } _ZNK8rawspeed39BitStreamerForwardSequentialReplenisherINS_14BitStreamerMSBEE6getPosEv: 84| 3.52M| [[nodiscard]] typename Base::size_type getPos() const { 85| 3.52M| Base::establishClassInvariants(); 86| 3.52M| return Base::pos; 87| 3.52M| } _ZN8rawspeed11BitStreamerINS_14BitStreamerMSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE9fillCacheENSt3__15arrayISt4byteLm4EEE: 157| 1.75M| inputStorage) { 158| 1.75M| static_assert(BitStreamCacheBase::MaxGetBits >= 32, "check implementation"); 159| 1.75M| establishClassInvariants(); 160| 1.75M| auto input = Array1DRef(inputStorage.data(), 161| 1.75M| implicit_cast(inputStorage.size())); 162| 1.75M| invariant(input.size() == Traits::MaxProcessBytes); ------------------ | | 27| 1.75M|#define invariant(expr) assert(expr) ------------------ | Branch (162:5): [True: 1.75M, False: 0] ------------------ 163| | 164| 1.75M| constexpr int StreamChunkBitwidth = 165| 1.75M| bitwidth(); 166| 1.75M| static_assert(CHAR_BIT * Traits::MaxProcessBytes >= StreamChunkBitwidth); 167| 1.75M| static_assert(CHAR_BIT * Traits::MaxProcessBytes % StreamChunkBitwidth == 168| 1.75M| 0); 169| 1.75M| constexpr int NumChunksNeeded = 170| 1.75M| (CHAR_BIT * Traits::MaxProcessBytes) / StreamChunkBitwidth; 171| 1.75M| static_assert(NumChunksNeeded >= 1); 172| | 173| 3.51M| for (int i = 0; i != NumChunksNeeded; ++i) { ------------------ | Branch (173:21): [True: 1.75M, False: 1.75M] ------------------ 174| 1.75M| auto chunkInput = 175| 1.75M| input.getBlock(sizeof(typename StreamTraits::ChunkType), i); 176| 1.75M| auto chunk = getByteSwapped( 177| 1.75M| chunkInput.begin(), 178| 1.75M| StreamTraits::ChunkEndianness != getHostEndianness()); 179| 1.75M| cache.push(chunk, StreamChunkBitwidth); 180| 1.75M| } 181| 1.75M| return Traits::MaxProcessBytes; 182| 1.75M| } _ZN8rawspeed39BitStreamerForwardSequentialReplenisherINS_14BitStreamerMSBEE22markNumBytesAsConsumedEi: 92| 1.75M| void markNumBytesAsConsumed(typename Base::size_type numBytes) { 93| 1.75M| Base::establishClassInvariants(); 94| 1.75M| invariant(numBytes >= 0); ------------------ | | 27| 1.75M|#define invariant(expr) assert(expr) ------------------ | Branch (94:5): [True: 1.75M, False: 0] ------------------ 95| 1.75M| invariant(numBytes != 0); ------------------ | | 27| 1.75M|#define invariant(expr) assert(expr) ------------------ | Branch (95:5): [True: 1.75M, False: 0] ------------------ 96| 1.75M| invariant(numBytes % StreamTraits::MinLoadStepByteMultiple == 0); ------------------ | | 27| 1.75M|#define invariant(expr) assert(expr) ------------------ | Branch (96:5): [True: 1.75M, False: 0] ------------------ 97| 1.75M| Base::pos += numBytes; 98| 1.75M| } _ZN8rawspeed11BitStreamerINS_14BitStreamerMSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE13getBitsNoFillEi: 269| 7.01M| uint32_t getBitsNoFill(int nbits) { 270| 7.01M| establishClassInvariants(); 271| 7.01M| invariant(nbits >= 0); ------------------ | | 27| 7.01M|#define invariant(expr) assert(expr) ------------------ | Branch (271:5): [True: 7.01M, False: 0] ------------------ 272| 7.01M| invariant(nbits != 0); ------------------ | | 27| 7.01M|#define invariant(expr) assert(expr) ------------------ | Branch (272:5): [True: 7.01M, False: 0] ------------------ 273| 7.01M| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 7.01M|#define invariant(expr) assert(expr) ------------------ | Branch (273:5): [True: 7.01M, False: 0] ------------------ 274| 7.01M| uint32_t ret = peekBitsNoFill(nbits); 275| 7.01M| skipBitsNoFill(nbits); 276| 7.01M| return ret; 277| 7.01M| } _ZN8rawspeed11BitStreamerINS_14BitStreamerMSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE14peekBitsNoFillEi: 253| 9.65M| uint32_t RAWSPEED_READONLY peekBitsNoFill(int nbits) { 254| 9.65M| establishClassInvariants(); 255| 9.65M| invariant(nbits >= 0); ------------------ | | 27| 9.65M|#define invariant(expr) assert(expr) ------------------ | Branch (255:5): [True: 9.65M, False: 0] ------------------ 256| 9.65M| invariant(nbits != 0); ------------------ | | 27| 9.65M|#define invariant(expr) assert(expr) ------------------ | Branch (256:5): [True: 9.65M, False: 0] ------------------ 257| 9.65M| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 9.65M|#define invariant(expr) assert(expr) ------------------ | Branch (257:5): [True: 9.65M, False: 0] ------------------ 258| 9.65M| return cache.peek(nbits); 259| 9.65M| } _ZN8rawspeed11BitStreamerINS_14BitStreamerMSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE14skipBitsNoFillEi: 261| 9.89M| void skipBitsNoFill(int nbits) { 262| 9.89M| establishClassInvariants(); 263| 9.89M| invariant(nbits >= 0); ------------------ | | 27| 9.89M|#define invariant(expr) assert(expr) ------------------ | Branch (263:5): [True: 9.89M, False: 0] ------------------ 264| | // `nbits` could be zero. 265| 9.89M| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 9.89M|#define invariant(expr) assert(expr) ------------------ | Branch (265:5): [True: 9.89M, False: 0] ------------------ 266| 9.89M| cache.skip(nbits); 267| 9.89M| } _ZN8rawspeed11BitStreamerINS_14BitStreamerMSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE7getBitsEi: 294| 4.40M| uint32_t getBits(int nbits) { 295| 4.40M| establishClassInvariants(); 296| 4.40M| invariant(nbits >= 0); ------------------ | | 27| 4.40M|#define invariant(expr) assert(expr) ------------------ | Branch (296:5): [True: 4.40M, False: 0] ------------------ 297| 4.40M| invariant(nbits != 0); ------------------ | | 27| 4.40M|#define invariant(expr) assert(expr) ------------------ | Branch (297:5): [True: 4.40M, False: 0] ------------------ 298| 4.40M| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 4.40M|#define invariant(expr) assert(expr) ------------------ | Branch (298:5): [True: 4.40M, False: 0] ------------------ 299| 4.40M| fill(nbits); 300| 4.40M| return getBitsNoFill(nbits); 301| 4.40M| } _ZN8rawspeed11BitStreamerINS_14BitStreamerMSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE9skipBytesEi: 321| 104k| void skipBytes(int nbytes) { 322| 104k| establishClassInvariants(); 323| 104k| int nbits = 8 * nbytes; 324| 104k| skipManyBits(nbits); 325| 104k| } _ZN8rawspeed11BitStreamerINS_14BitStreamerMSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE12skipManyBitsEi: 305| 104k| void skipManyBits(int nbits) { 306| 104k| establishClassInvariants(); 307| 104k| int remainingBitsToSkip = nbits; 308| 505k| for (; remainingBitsToSkip >= Cache::MaxGetBits; ------------------ | Branch (308:12): [True: 400k, False: 104k] ------------------ 309| 400k| remainingBitsToSkip -= Cache::MaxGetBits) { 310| 400k| fill(Cache::MaxGetBits); 311| 400k| skipBitsNoFill(Cache::MaxGetBits); 312| 400k| } 313| 104k| if (remainingBitsToSkip > 0) { ------------------ | Branch (313:9): [True: 426, False: 104k] ------------------ 314| 426| fill(remainingBitsToSkip); 315| 426| skipBitsNoFill(remainingBitsToSkip); 316| 426| } 317| 104k| } _ZN8rawspeed11BitStreamerINS_14BitStreamerLSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEEC2ENS_10Array1DRefIKSt4byteEE: 192| 397k| explicit BitStreamer(Array1DRef input) : replenisher(input) { 193| 397k| establishClassInvariants(); 194| 397k| } _ZN8rawspeed26BitStreamerReplenisherBaseINS_14BitStreamerLSBEEC2ENS_10Array1DRefIKSt4byteEE: 57| 397k| : input(input_) { 58| 397k| if (input.size() < BitStreamerTraits::MaxProcessBytes) ------------------ | Branch (58:9): [True: 29, False: 397k] ------------------ 59| 397k| ThrowIOE("Bit stream size is smaller than MaxProcessBytes"); ------------------ | | 37| 29|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 29| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 29| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 29| __VA_ARGS__) | | ------------------ ------------------ 60| 397k| } _ZNK8rawspeed11BitStreamerINS_14BitStreamerLSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE24establishClassInvariantsEv: 185| 41.5M| void establishClassInvariants() const noexcept { 186| 41.5M| cache.establishClassInvariants(); 187| 41.5M| replenisher.establishClassInvariants(); 188| 41.5M| } _ZNK8rawspeed26BitStreamerReplenisherBaseINS_14BitStreamerLSBEE24establishClassInvariantsEv: 65| 50.9M|BitStreamerReplenisherBase::establishClassInvariants() const noexcept { 66| 50.9M| input.establishClassInvariants(); 67| 50.9M| invariant(input.size() >= BitStreamerTraits::MaxProcessBytes); ------------------ | | 27| 50.9M|#define invariant(expr) assert(expr) ------------------ | Branch (67:3): [True: 48.7M, False: 2.15M] ------------------ 68| 48.7M| invariant(pos >= 0); ------------------ | | 27| 48.7M|#define invariant(expr) assert(expr) ------------------ | Branch (68:3): [True: 29.5M, False: 19.1M] ------------------ 69| 29.5M| invariant(pos % StreamTraits::MinLoadStepByteMultiple == 0); ------------------ | | 27| 29.5M|#define invariant(expr) assert(expr) ------------------ | Branch (69:3): [True: 28.8M, False: 750k] ------------------ 70| | // `pos` *could* be out-of-bounds of `input`. 71| 29.5M|} _ZN8rawspeed11BitStreamerINS_14BitStreamerLSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE7getBitsEi: 294| 4.93M| uint32_t getBits(int nbits) { 295| 4.93M| establishClassInvariants(); 296| 4.93M| invariant(nbits >= 0); ------------------ | | 27| 4.93M|#define invariant(expr) assert(expr) ------------------ | Branch (296:5): [True: 4.65M, False: 278k] ------------------ 297| 4.65M| invariant(nbits != 0); ------------------ | | 27| 4.65M|#define invariant(expr) assert(expr) ------------------ | Branch (297:5): [True: 4.65M, False: 18.4E] ------------------ 298| 4.65M| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 4.65M|#define invariant(expr) assert(expr) ------------------ | Branch (298:5): [True: 4.87M, False: 18.4E] ------------------ 299| 4.87M| fill(nbits); 300| 4.87M| return getBitsNoFill(nbits); 301| 4.65M| } _ZN8rawspeed11BitStreamerINS_14BitStreamerLSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE4fillEi: 216| 7.47M| void fill(int nbits = Cache::MaxGetBits) { 217| 7.47M| establishClassInvariants(); 218| 7.47M| invariant(nbits >= 0); ------------------ | | 27| 7.47M|#define invariant(expr) assert(expr) ------------------ | Branch (218:5): [True: 7.51M, False: 18.4E] ------------------ 219| 7.51M| invariant(nbits != 0); ------------------ | | 27| 7.51M|#define invariant(expr) assert(expr) ------------------ | Branch (219:5): [True: 7.53M, False: 18.4E] ------------------ 220| 7.53M| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 7.53M|#define invariant(expr) assert(expr) ------------------ | Branch (220:5): [True: 7.54M, False: 18.4E] ------------------ 221| | 222| 7.54M| if (cache.fillLevel >= nbits) ------------------ | Branch (222:9): [True: 3.51M, False: 4.03M] ------------------ 223| 3.51M| return; 224| | 225| 4.03M| const auto input = replenisher.getInput(); 226| 4.03M| const auto numBytes = static_cast(this)->fillCache(input); 227| 4.03M| replenisher.markNumBytesAsConsumed(numBytes); 228| | invariant(cache.fillLevel >= nbits); ------------------ | | 27| 4.03M|#define invariant(expr) assert(expr) ------------------ | Branch (228:5): [True: 4.18M, False: 18.4E] ------------------ 229| 4.03M| } _ZN8rawspeed39BitStreamerForwardSequentialReplenisherINS_14BitStreamerLSBEE8getInputEv: 100| 4.28M| std::array::MaxProcessBytes> getInput() { 101| 4.28M| Base::establishClassInvariants(); 102| | 103| 4.28M| std::array::MaxProcessBytes> tmpStorage; 104| 4.28M| auto tmp = Array1DRef(tmpStorage.data(), 105| 4.28M| implicit_cast(tmpStorage.size())); 106| | 107| | // Do we have BitStreamerTraits::MaxProcessBytes or more bytes left in 108| | // the input buffer? If so, then we can just read from said buffer. 109| 4.28M| if (getPos() + BitStreamerTraits::MaxProcessBytes <= ------------------ | Branch (109:9): [True: 4.29M, False: 18.4E] ------------------ 110| 4.29M| Base::input.size()) [[likely]] { 111| 4.29M| auto currInput = 112| 4.29M| Base::input.getCrop(getPos(), BitStreamerTraits::MaxProcessBytes) 113| 4.29M| .getAsArray1DRef(); 114| 4.29M| invariant(currInput.size() == tmp.size()); ------------------ | | 27| 4.29M|#define invariant(expr) assert(expr) ------------------ | Branch (114:7): [True: 4.10M, False: 189k] ------------------ 115| 4.10M| memcpy(tmp.begin(), currInput.begin(), 116| 4.10M| BitStreamerTraits::MaxProcessBytes); 117| 4.10M| return tmpStorage; 118| 4.29M| } 119| | 120| | // We have to use intermediate buffer, either because the input is running 121| | // out of bytes, or because we want to enforce bounds checking. 122| | 123| | // Note that in order to keep all fill-level invariants we must allow to 124| | // over-read past-the-end a bit. 125| 18.4E| if (getPos() > Base::input.size() + ------------------ | Branch (125:9): [True: 0, False: 18.4E] ------------------ 126| 18.4E| 2 * BitStreamerTraits::MaxProcessBytes) [[unlikely]] 127| 0| ThrowIOE("Buffer overflow read in BitStreamer"); ------------------ | | 37| 0|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 128| | 129| 18.4E| variableLengthLoadNaiveViaMemcpy(tmp, Base::input, getPos()); 130| | 131| 18.4E| return tmpStorage; 132| 18.4E| } _ZNK8rawspeed39BitStreamerForwardSequentialReplenisherINS_14BitStreamerLSBEE6getPosEv: 84| 8.13M| [[nodiscard]] typename Base::size_type getPos() const { 85| 8.13M| Base::establishClassInvariants(); 86| 8.13M| return Base::pos; 87| 8.13M| } _ZN8rawspeed11BitStreamerINS_14BitStreamerLSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE9fillCacheENSt3__15arrayISt4byteLm4EEE: 157| 4.13M| inputStorage) { 158| 4.13M| static_assert(BitStreamCacheBase::MaxGetBits >= 32, "check implementation"); 159| 4.13M| establishClassInvariants(); 160| 4.13M| auto input = Array1DRef(inputStorage.data(), 161| 4.13M| implicit_cast(inputStorage.size())); 162| 4.13M| invariant(input.size() == Traits::MaxProcessBytes); ------------------ | | 27| 4.13M|#define invariant(expr) assert(expr) ------------------ | Branch (162:5): [True: 4.20M, False: 18.4E] ------------------ 163| | 164| 4.20M| constexpr int StreamChunkBitwidth = 165| 4.20M| bitwidth(); 166| 4.20M| static_assert(CHAR_BIT * Traits::MaxProcessBytes >= StreamChunkBitwidth); 167| 4.20M| static_assert(CHAR_BIT * Traits::MaxProcessBytes % StreamChunkBitwidth == 168| 4.20M| 0); 169| 4.20M| constexpr int NumChunksNeeded = 170| 4.20M| (CHAR_BIT * Traits::MaxProcessBytes) / StreamChunkBitwidth; 171| 4.20M| static_assert(NumChunksNeeded >= 1); 172| | 173| 8.41M| for (int i = 0; i != NumChunksNeeded; ++i) { ------------------ | Branch (173:21): [True: 4.20M, False: 4.20M] ------------------ 174| 4.20M| auto chunkInput = 175| 4.20M| input.getBlock(sizeof(typename StreamTraits::ChunkType), i); 176| 4.20M| auto chunk = getByteSwapped( 177| 4.20M| chunkInput.begin(), 178| 4.20M| StreamTraits::ChunkEndianness != getHostEndianness()); 179| 4.20M| cache.push(chunk, StreamChunkBitwidth); 180| 4.20M| } 181| 4.20M| return Traits::MaxProcessBytes; 182| 4.13M| } _ZN8rawspeed39BitStreamerForwardSequentialReplenisherINS_14BitStreamerLSBEE22markNumBytesAsConsumedEi: 92| 4.21M| void markNumBytesAsConsumed(typename Base::size_type numBytes) { 93| 4.21M| Base::establishClassInvariants(); 94| 4.21M| invariant(numBytes >= 0); ------------------ | | 27| 4.21M|#define invariant(expr) assert(expr) ------------------ | Branch (94:5): [True: 4.15M, False: 55.7k] ------------------ 95| 4.15M| invariant(numBytes != 0); ------------------ | | 27| 4.15M|#define invariant(expr) assert(expr) ------------------ | Branch (95:5): [True: 4.17M, False: 18.4E] ------------------ 96| 4.17M| invariant(numBytes % StreamTraits::MinLoadStepByteMultiple == 0); ------------------ | | 27| 4.17M|#define invariant(expr) assert(expr) ------------------ | Branch (96:5): [True: 4.17M, False: 0] ------------------ 97| 4.17M| Base::pos += numBytes; 98| 4.17M| } _ZN8rawspeed11BitStreamerINS_14BitStreamerLSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE13getBitsNoFillEi: 269| 10.1M| uint32_t getBitsNoFill(int nbits) { 270| 10.1M| establishClassInvariants(); 271| 10.1M| invariant(nbits >= 0); ------------------ | | 27| 10.1M|#define invariant(expr) assert(expr) ------------------ | Branch (271:5): [True: 9.82M, False: 363k] ------------------ 272| 9.82M| invariant(nbits != 0); ------------------ | | 27| 9.82M|#define invariant(expr) assert(expr) ------------------ | Branch (272:5): [True: 9.84M, False: 18.4E] ------------------ 273| 9.84M| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 9.84M|#define invariant(expr) assert(expr) ------------------ | Branch (273:5): [True: 10.1M, False: 18.4E] ------------------ 274| 10.1M| uint32_t ret = peekBitsNoFill(nbits); 275| 10.1M| skipBitsNoFill(nbits); 276| 10.1M| return ret; 277| 9.84M| } _ZN8rawspeed11BitStreamerINS_14BitStreamerLSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE14peekBitsNoFillEi: 253| 10.1M| uint32_t RAWSPEED_READONLY peekBitsNoFill(int nbits) { 254| 10.1M| establishClassInvariants(); 255| 10.1M| invariant(nbits >= 0); ------------------ | | 27| 10.1M|#define invariant(expr) assert(expr) ------------------ | Branch (255:5): [True: 9.84M, False: 314k] ------------------ 256| 9.84M| invariant(nbits != 0); ------------------ | | 27| 9.84M|#define invariant(expr) assert(expr) ------------------ | Branch (256:5): [True: 9.85M, False: 18.4E] ------------------ 257| 9.85M| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 9.85M|#define invariant(expr) assert(expr) ------------------ | Branch (257:5): [True: 10.1M, False: 18.4E] ------------------ 258| 10.1M| return cache.peek(nbits); 259| 9.85M| } _ZN8rawspeed11BitStreamerINS_14BitStreamerLSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE14skipBitsNoFillEi: 261| 10.9M| void skipBitsNoFill(int nbits) { 262| 10.9M| establishClassInvariants(); 263| 10.9M| invariant(nbits >= 0); ------------------ | | 27| 10.9M|#define invariant(expr) assert(expr) ------------------ | Branch (263:5): [True: 10.2M, False: 757k] ------------------ 264| | // `nbits` could be zero. 265| 10.2M| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 10.2M|#define invariant(expr) assert(expr) ------------------ | Branch (265:5): [True: 10.2M, False: 18.4E] ------------------ 266| 10.2M| cache.skip(nbits); 267| 10.2M| } _ZN8rawspeed11BitStreamerINS_14BitStreamerLSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE9skipBytesEi: 321| 23.3k| void skipBytes(int nbytes) { 322| 23.3k| establishClassInvariants(); 323| 23.3k| int nbits = 8 * nbytes; 324| 23.3k| skipManyBits(nbits); 325| 23.3k| } _ZN8rawspeed11BitStreamerINS_14BitStreamerLSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE12skipManyBitsEi: 305| 23.3k| void skipManyBits(int nbits) { 306| 23.3k| establishClassInvariants(); 307| 23.3k| int remainingBitsToSkip = nbits; 308| 101k| for (; remainingBitsToSkip >= Cache::MaxGetBits; ------------------ | Branch (308:12): [True: 78.3k, False: 23.3k] ------------------ 309| 78.3k| remainingBitsToSkip -= Cache::MaxGetBits) { 310| 78.3k| fill(Cache::MaxGetBits); 311| 78.3k| skipBitsNoFill(Cache::MaxGetBits); 312| 78.3k| } 313| 23.3k| if (remainingBitsToSkip > 0) { ------------------ | Branch (313:9): [True: 878, False: 22.4k] ------------------ 314| 878| fill(remainingBitsToSkip); 315| 878| skipBitsNoFill(remainingBitsToSkip); 316| 878| } 317| 23.3k| } _ZN8rawspeed11BitStreamerINS_16BitStreamerMSB32ENS_39BitStreamerForwardSequentialReplenisherIS1_EEEC2ENS_10Array1DRefIKSt4byteEE: 192| 11.3k| explicit BitStreamer(Array1DRef input) : replenisher(input) { 193| 11.3k| establishClassInvariants(); 194| 11.3k| } _ZN8rawspeed26BitStreamerReplenisherBaseINS_16BitStreamerMSB32EEC2ENS_10Array1DRefIKSt4byteEE: 57| 11.3k| : input(input_) { 58| 11.3k| if (input.size() < BitStreamerTraits::MaxProcessBytes) ------------------ | Branch (58:9): [True: 230, False: 11.1k] ------------------ 59| 11.3k| ThrowIOE("Bit stream size is smaller than MaxProcessBytes"); ------------------ | | 37| 230|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 230| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 230| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 230| __VA_ARGS__) | | ------------------ ------------------ 60| 11.3k| } _ZNK8rawspeed11BitStreamerINS_16BitStreamerMSB32ENS_39BitStreamerForwardSequentialReplenisherIS1_EEE24establishClassInvariantsEv: 185| 3.08M| void establishClassInvariants() const noexcept { 186| 3.08M| cache.establishClassInvariants(); 187| 3.08M| replenisher.establishClassInvariants(); 188| 3.08M| } _ZNK8rawspeed26BitStreamerReplenisherBaseINS_16BitStreamerMSB32EE24establishClassInvariantsEv: 65| 3.60M|BitStreamerReplenisherBase::establishClassInvariants() const noexcept { 66| 3.60M| input.establishClassInvariants(); 67| 3.60M| invariant(input.size() >= BitStreamerTraits::MaxProcessBytes); ------------------ | | 27| 3.60M|#define invariant(expr) assert(expr) ------------------ | Branch (67:3): [True: 3.60M, False: 557] ------------------ 68| 3.60M| invariant(pos >= 0); ------------------ | | 27| 3.60M|#define invariant(expr) assert(expr) ------------------ | Branch (68:3): [True: 3.60M, False: 18.4E] ------------------ 69| 3.60M| invariant(pos % StreamTraits::MinLoadStepByteMultiple == 0); ------------------ | | 27| 3.60M|#define invariant(expr) assert(expr) ------------------ | Branch (69:3): [True: 3.60M, False: 18.4E] ------------------ 70| | // `pos` *could* be out-of-bounds of `input`. 71| 3.60M|} _ZN8rawspeed11BitStreamerINS_16BitStreamerMSB32ENS_39BitStreamerForwardSequentialReplenisherIS1_EEE7getBitsEi: 294| 310k| uint32_t getBits(int nbits) { 295| 310k| establishClassInvariants(); 296| 310k| invariant(nbits >= 0); ------------------ | | 27| 310k|#define invariant(expr) assert(expr) ------------------ | Branch (296:5): [True: 310k, False: 0] ------------------ 297| 310k| invariant(nbits != 0); ------------------ | | 27| 310k|#define invariant(expr) assert(expr) ------------------ | Branch (297:5): [True: 310k, False: 0] ------------------ 298| 310k| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 310k|#define invariant(expr) assert(expr) ------------------ | Branch (298:5): [True: 310k, False: 0] ------------------ 299| 310k| fill(nbits); 300| 310k| return getBitsNoFill(nbits); 301| 310k| } _ZN8rawspeed11BitStreamerINS_16BitStreamerMSB32ENS_39BitStreamerForwardSequentialReplenisherIS1_EEE4fillEi: 216| 524k| void fill(int nbits = Cache::MaxGetBits) { 217| 524k| establishClassInvariants(); 218| 524k| invariant(nbits >= 0); ------------------ | | 27| 524k|#define invariant(expr) assert(expr) ------------------ | Branch (218:5): [True: 524k, False: 18.4E] ------------------ 219| 524k| invariant(nbits != 0); ------------------ | | 27| 524k|#define invariant(expr) assert(expr) ------------------ | Branch (219:5): [True: 524k, False: 18.4E] ------------------ 220| 524k| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 524k|#define invariant(expr) assert(expr) ------------------ | Branch (220:5): [True: 524k, False: 0] ------------------ 221| | 222| 524k| if (cache.fillLevel >= nbits) ------------------ | Branch (222:9): [True: 397k, False: 127k] ------------------ 223| 397k| return; 224| | 225| 127k| const auto input = replenisher.getInput(); 226| 127k| const auto numBytes = static_cast(this)->fillCache(input); 227| 127k| replenisher.markNumBytesAsConsumed(numBytes); 228| | invariant(cache.fillLevel >= nbits); ------------------ | | 27| 127k|#define invariant(expr) assert(expr) ------------------ | Branch (228:5): [True: 126k, False: 475] ------------------ 229| 127k| } _ZN8rawspeed39BitStreamerForwardSequentialReplenisherINS_16BitStreamerMSB32EE8getInputEv: 100| 127k| std::array::MaxProcessBytes> getInput() { 101| 127k| Base::establishClassInvariants(); 102| | 103| 127k| std::array::MaxProcessBytes> tmpStorage; 104| 127k| auto tmp = Array1DRef(tmpStorage.data(), 105| 127k| implicit_cast(tmpStorage.size())); 106| | 107| | // Do we have BitStreamerTraits::MaxProcessBytes or more bytes left in 108| | // the input buffer? If so, then we can just read from said buffer. 109| 127k| if (getPos() + BitStreamerTraits::MaxProcessBytes <= ------------------ | Branch (109:9): [True: 124k, False: 2.67k] ------------------ 110| 127k| Base::input.size()) [[likely]] { 111| 124k| auto currInput = 112| 124k| Base::input.getCrop(getPos(), BitStreamerTraits::MaxProcessBytes) 113| 124k| .getAsArray1DRef(); 114| 124k| invariant(currInput.size() == tmp.size()); ------------------ | | 27| 124k|#define invariant(expr) assert(expr) ------------------ | Branch (114:7): [True: 124k, False: 87] ------------------ 115| 124k| memcpy(tmp.begin(), currInput.begin(), 116| 124k| BitStreamerTraits::MaxProcessBytes); 117| 124k| return tmpStorage; 118| 124k| } 119| | 120| | // We have to use intermediate buffer, either because the input is running 121| | // out of bytes, or because we want to enforce bounds checking. 122| | 123| | // Note that in order to keep all fill-level invariants we must allow to 124| | // over-read past-the-end a bit. 125| 2.67k| if (getPos() > Base::input.size() + ------------------ | Branch (125:9): [True: 400, False: 2.27k] ------------------ 126| 2.67k| 2 * BitStreamerTraits::MaxProcessBytes) [[unlikely]] 127| 400| ThrowIOE("Buffer overflow read in BitStreamer"); ------------------ | | 37| 400|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 400| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 400| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 400| __VA_ARGS__) | | ------------------ ------------------ 128| | 129| 2.27k| variableLengthLoadNaiveViaMemcpy(tmp, Base::input, getPos()); 130| | 131| 2.27k| return tmpStorage; 132| 2.67k| } _ZNK8rawspeed39BitStreamerForwardSequentialReplenisherINS_16BitStreamerMSB32EE6getPosEv: 84| 264k| [[nodiscard]] typename Base::size_type getPos() const { 85| 264k| Base::establishClassInvariants(); 86| 264k| return Base::pos; 87| 264k| } _ZN8rawspeed11BitStreamerINS_16BitStreamerMSB32ENS_39BitStreamerForwardSequentialReplenisherIS1_EEE9fillCacheENSt3__15arrayISt4byteLm4EEE: 157| 126k| inputStorage) { 158| 126k| static_assert(BitStreamCacheBase::MaxGetBits >= 32, "check implementation"); 159| 126k| establishClassInvariants(); 160| 126k| auto input = Array1DRef(inputStorage.data(), 161| 126k| implicit_cast(inputStorage.size())); 162| 126k| invariant(input.size() == Traits::MaxProcessBytes); ------------------ | | 27| 126k|#define invariant(expr) assert(expr) ------------------ | Branch (162:5): [True: 126k, False: 9] ------------------ 163| | 164| 126k| constexpr int StreamChunkBitwidth = 165| 126k| bitwidth(); 166| 126k| static_assert(CHAR_BIT * Traits::MaxProcessBytes >= StreamChunkBitwidth); 167| 126k| static_assert(CHAR_BIT * Traits::MaxProcessBytes % StreamChunkBitwidth == 168| 126k| 0); 169| 126k| constexpr int NumChunksNeeded = 170| 126k| (CHAR_BIT * Traits::MaxProcessBytes) / StreamChunkBitwidth; 171| 126k| static_assert(NumChunksNeeded >= 1); 172| | 173| 253k| for (int i = 0; i != NumChunksNeeded; ++i) { ------------------ | Branch (173:21): [True: 126k, False: 126k] ------------------ 174| 126k| auto chunkInput = 175| 126k| input.getBlock(sizeof(typename StreamTraits::ChunkType), i); 176| 126k| auto chunk = getByteSwapped( 177| 126k| chunkInput.begin(), 178| 126k| StreamTraits::ChunkEndianness != getHostEndianness()); 179| 126k| cache.push(chunk, StreamChunkBitwidth); 180| 126k| } 181| 126k| return Traits::MaxProcessBytes; 182| 126k| } _ZN8rawspeed39BitStreamerForwardSequentialReplenisherINS_16BitStreamerMSB32EE22markNumBytesAsConsumedEi: 92| 126k| void markNumBytesAsConsumed(typename Base::size_type numBytes) { 93| 126k| Base::establishClassInvariants(); 94| 126k| invariant(numBytes >= 0); ------------------ | | 27| 126k|#define invariant(expr) assert(expr) ------------------ | Branch (94:5): [True: 126k, False: 8] ------------------ 95| 126k| invariant(numBytes != 0); ------------------ | | 27| 126k|#define invariant(expr) assert(expr) ------------------ | Branch (95:5): [True: 126k, False: 18.4E] ------------------ 96| 126k| invariant(numBytes % StreamTraits::MinLoadStepByteMultiple == 0); ------------------ | | 27| 126k|#define invariant(expr) assert(expr) ------------------ | Branch (96:5): [True: 126k, False: 18.4E] ------------------ 97| 126k| Base::pos += numBytes; 98| 126k| } _ZN8rawspeed11BitStreamerINS_16BitStreamerMSB32ENS_39BitStreamerForwardSequentialReplenisherIS1_EEE13getBitsNoFillEi: 269| 686k| uint32_t getBitsNoFill(int nbits) { 270| 686k| establishClassInvariants(); 271| 686k| invariant(nbits >= 0); ------------------ | | 27| 686k|#define invariant(expr) assert(expr) ------------------ | Branch (271:5): [True: 686k, False: 18.4E] ------------------ 272| 686k| invariant(nbits != 0); ------------------ | | 27| 686k|#define invariant(expr) assert(expr) ------------------ | Branch (272:5): [True: 686k, False: 1] ------------------ 273| 686k| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 686k|#define invariant(expr) assert(expr) ------------------ | Branch (273:5): [True: 686k, False: 18.4E] ------------------ 274| 686k| uint32_t ret = peekBitsNoFill(nbits); 275| 686k| skipBitsNoFill(nbits); 276| 686k| return ret; 277| 686k| } _ZN8rawspeed11BitStreamerINS_16BitStreamerMSB32ENS_39BitStreamerForwardSequentialReplenisherIS1_EEE14peekBitsNoFillEi: 253| 707k| uint32_t RAWSPEED_READONLY peekBitsNoFill(int nbits) { 254| 707k| establishClassInvariants(); 255| 707k| invariant(nbits >= 0); ------------------ | | 27| 707k|#define invariant(expr) assert(expr) ------------------ | Branch (255:5): [True: 707k, False: 18.4E] ------------------ 256| 707k| invariant(nbits != 0); ------------------ | | 27| 707k|#define invariant(expr) assert(expr) ------------------ | Branch (256:5): [True: 707k, False: 18.4E] ------------------ 257| 707k| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 707k|#define invariant(expr) assert(expr) ------------------ | Branch (257:5): [True: 707k, False: 18.4E] ------------------ 258| 707k| return cache.peek(nbits); 259| 707k| } _ZN8rawspeed11BitStreamerINS_16BitStreamerMSB32ENS_39BitStreamerForwardSequentialReplenisherIS1_EEE14skipBitsNoFillEi: 261| 708k| void skipBitsNoFill(int nbits) { 262| 708k| establishClassInvariants(); 263| 708k| invariant(nbits >= 0); ------------------ | | 27| 708k|#define invariant(expr) assert(expr) ------------------ | Branch (263:5): [True: 708k, False: 35] ------------------ 264| | // `nbits` could be zero. 265| 708k| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 708k|#define invariant(expr) assert(expr) ------------------ | Branch (265:5): [True: 708k, False: 0] ------------------ 266| 708k| cache.skip(nbits); 267| 708k| } _ZN8rawspeed11BitStreamerINS_16BitStreamerMSB32ENS_39BitStreamerForwardSequentialReplenisherIS1_EEE9skipBytesEi: 321| 38| void skipBytes(int nbytes) { 322| 38| establishClassInvariants(); 323| 38| int nbits = 8 * nbytes; 324| 38| skipManyBits(nbits); 325| 38| } _ZN8rawspeed11BitStreamerINS_16BitStreamerMSB32ENS_39BitStreamerForwardSequentialReplenisherIS1_EEE12skipManyBitsEi: 305| 38| void skipManyBits(int nbits) { 306| 38| establishClassInvariants(); 307| 38| int remainingBitsToSkip = nbits; 308| 38| for (; remainingBitsToSkip >= Cache::MaxGetBits; ------------------ | Branch (308:12): [True: 0, False: 38] ------------------ 309| 38| remainingBitsToSkip -= Cache::MaxGetBits) { 310| 0| fill(Cache::MaxGetBits); 311| 0| skipBitsNoFill(Cache::MaxGetBits); 312| 0| } 313| 38| if (remainingBitsToSkip > 0) { ------------------ | Branch (313:9): [True: 0, False: 38] ------------------ 314| 0| fill(remainingBitsToSkip); 315| 0| skipBitsNoFill(remainingBitsToSkip); 316| 0| } 317| 38| } _ZN8rawspeed11BitStreamerINS_14BitStreamerLSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE8peekBitsEi: 279| 18.1k| uint32_t peekBits(int nbits) { 280| 18.1k| establishClassInvariants(); 281| 18.1k| invariant(nbits >= 0); ------------------ | | 27| 18.1k|#define invariant(expr) assert(expr) ------------------ | Branch (281:5): [True: 17.6k, False: 526] ------------------ 282| 17.6k| invariant(nbits != 0); ------------------ | | 27| 17.6k|#define invariant(expr) assert(expr) ------------------ | Branch (282:5): [True: 17.6k, False: 0] ------------------ 283| 17.6k| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 17.6k|#define invariant(expr) assert(expr) ------------------ | Branch (283:5): [True: 18.1k, False: 18.4E] ------------------ 284| 18.1k| fill(nbits); 285| 18.1k| return peekBitsNoFill(nbits); 286| 17.6k| } _ZNK8rawspeed11BitStreamerINS_15BitStreamerJPEGENS_39BitStreamerForwardSequentialReplenisherIS1_EEE24establishClassInvariantsEv: 185| 33.2M| void establishClassInvariants() const noexcept { 186| 33.2M| cache.establishClassInvariants(); 187| 33.2M| replenisher.establishClassInvariants(); 188| 33.2M| } _ZNK8rawspeed26BitStreamerReplenisherBaseINS_15BitStreamerJPEGEE24establishClassInvariantsEv: 65| 37.0M|BitStreamerReplenisherBase::establishClassInvariants() const noexcept { 66| 37.0M| input.establishClassInvariants(); 67| 37.0M| invariant(input.size() >= BitStreamerTraits::MaxProcessBytes); ------------------ | | 27| 37.0M|#define invariant(expr) assert(expr) ------------------ | Branch (67:3): [True: 37.0M, False: 6.30k] ------------------ 68| 37.0M| invariant(pos >= 0); ------------------ | | 27| 37.0M|#define invariant(expr) assert(expr) ------------------ | Branch (68:3): [True: 37.0M, False: 13.2k] ------------------ 69| 37.0M| invariant(pos % StreamTraits::MinLoadStepByteMultiple == 0); ------------------ | | 27| 37.0M|#define invariant(expr) assert(expr) ------------------ | Branch (69:3): [True: 37.0M, False: 0] ------------------ 70| | // `pos` *could* be out-of-bounds of `input`. 71| 37.0M|} _ZNK8rawspeed11BitStreamerINS_15BitStreamerJPEGENS_39BitStreamerForwardSequentialReplenisherIS1_EEE16getInputPositionEv: 232| 452| [[nodiscard]] size_type RAWSPEED_READONLY getInputPosition() const { 233| 452| establishClassInvariants(); 234| 452| return replenisher.getPos(); 235| 452| } _ZNK8rawspeed39BitStreamerForwardSequentialReplenisherINS_15BitStreamerJPEGEE6getPosEv: 84| 1.91M| [[nodiscard]] typename Base::size_type getPos() const { 85| 1.91M| Base::establishClassInvariants(); 86| 1.91M| return Base::pos; 87| 1.91M| } _ZNK8rawspeed11BitStreamerINS_15BitStreamerJPEGENS_39BitStreamerForwardSequentialReplenisherIS1_EEE16getRemainingSizeEv: 243| 301| [[nodiscard]] size_type getRemainingSize() const { 244| 301| establishClassInvariants(); 245| 301| return replenisher.getRemainingSize(); 246| 301| } _ZNK8rawspeed39BitStreamerForwardSequentialReplenisherINS_15BitStreamerJPEGEE16getRemainingSizeEv: 88| 301| [[nodiscard]] typename Base::size_type getRemainingSize() const { 89| 301| Base::establishClassInvariants(); 90| 301| return Base::input.size() - getPos(); 91| 301| } _ZN8rawspeed11BitStreamerINS_15BitStreamerJPEGENS_39BitStreamerForwardSequentialReplenisherIS1_EEEC2ENS_10Array1DRefIKSt4byteEE: 192| 918| explicit BitStreamer(Array1DRef input) : replenisher(input) { 193| 918| establishClassInvariants(); 194| 918| } _ZN8rawspeed26BitStreamerReplenisherBaseINS_15BitStreamerJPEGEEC2ENS_10Array1DRefIKSt4byteEE: 57| 918| : input(input_) { 58| 918| if (input.size() < BitStreamerTraits::MaxProcessBytes) ------------------ | Branch (58:9): [True: 49, False: 869] ------------------ 59| 918| ThrowIOE("Bit stream size is smaller than MaxProcessBytes"); ------------------ | | 37| 49|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 49| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 49| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 49| __VA_ARGS__) | | ------------------ ------------------ 60| 918| } _ZN8rawspeed11BitStreamerINS_15BitStreamerJPEGENS_39BitStreamerForwardSequentialReplenisherIS1_EEE4fillEi: 216| 10.7M| void fill(int nbits = Cache::MaxGetBits) { 217| 10.7M| establishClassInvariants(); 218| 10.7M| invariant(nbits >= 0); ------------------ | | 27| 10.7M|#define invariant(expr) assert(expr) ------------------ | Branch (218:5): [True: 10.7M, False: 18.4E] ------------------ 219| 10.7M| invariant(nbits != 0); ------------------ | | 27| 10.7M|#define invariant(expr) assert(expr) ------------------ | Branch (219:5): [True: 10.7M, False: 0] ------------------ 220| 10.7M| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 10.7M|#define invariant(expr) assert(expr) ------------------ | Branch (220:5): [True: 10.7M, False: 18.4E] ------------------ 221| | 222| 10.7M| if (cache.fillLevel >= nbits) ------------------ | Branch (222:9): [True: 9.79M, False: 955k] ------------------ 223| 9.79M| return; 224| | 225| 955k| const auto input = replenisher.getInput(); 226| 955k| const auto numBytes = static_cast(this)->fillCache(input); 227| 955k| replenisher.markNumBytesAsConsumed(numBytes); 228| | invariant(cache.fillLevel >= nbits); ------------------ | | 27| 955k|#define invariant(expr) assert(expr) ------------------ | Branch (228:5): [True: 956k, False: 18.4E] ------------------ 229| 955k| } _ZN8rawspeed39BitStreamerForwardSequentialReplenisherINS_15BitStreamerJPEGEE8getInputEv: 100| 956k| std::array::MaxProcessBytes> getInput() { 101| 956k| Base::establishClassInvariants(); 102| | 103| 956k| std::array::MaxProcessBytes> tmpStorage; 104| 956k| auto tmp = Array1DRef(tmpStorage.data(), 105| 956k| implicit_cast(tmpStorage.size())); 106| | 107| | // Do we have BitStreamerTraits::MaxProcessBytes or more bytes left in 108| | // the input buffer? If so, then we can just read from said buffer. 109| 956k| if (getPos() + BitStreamerTraits::MaxProcessBytes <= ------------------ | Branch (109:9): [True: 953k, False: 3.05k] ------------------ 110| 956k| Base::input.size()) [[likely]] { 111| 953k| auto currInput = 112| 953k| Base::input.getCrop(getPos(), BitStreamerTraits::MaxProcessBytes) 113| 953k| .getAsArray1DRef(); 114| 953k| invariant(currInput.size() == tmp.size()); ------------------ | | 27| 953k|#define invariant(expr) assert(expr) ------------------ | Branch (114:7): [True: 953k, False: 11] ------------------ 115| 953k| memcpy(tmp.begin(), currInput.begin(), 116| 953k| BitStreamerTraits::MaxProcessBytes); 117| 953k| return tmpStorage; 118| 953k| } 119| | 120| | // We have to use intermediate buffer, either because the input is running 121| | // out of bytes, or because we want to enforce bounds checking. 122| | 123| | // Note that in order to keep all fill-level invariants we must allow to 124| | // over-read past-the-end a bit. 125| 3.05k| if (getPos() > Base::input.size() + ------------------ | Branch (125:9): [True: 436, False: 2.61k] ------------------ 126| 3.05k| 2 * BitStreamerTraits::MaxProcessBytes) [[unlikely]] 127| 436| ThrowIOE("Buffer overflow read in BitStreamer"); ------------------ | | 37| 436|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 436| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 436| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 436| __VA_ARGS__) | | ------------------ ------------------ 128| | 129| 2.61k| variableLengthLoadNaiveViaMemcpy(tmp, Base::input, getPos()); 130| | 131| 2.61k| return tmpStorage; 132| 3.05k| } _ZN8rawspeed39BitStreamerForwardSequentialReplenisherINS_15BitStreamerJPEGEE22markNumBytesAsConsumedEi: 92| 956k| void markNumBytesAsConsumed(typename Base::size_type numBytes) { 93| 956k| Base::establishClassInvariants(); 94| 956k| invariant(numBytes >= 0); ------------------ | | 27| 956k|#define invariant(expr) assert(expr) ------------------ | Branch (94:5): [True: 956k, False: 18.4E] ------------------ 95| 956k| invariant(numBytes != 0); ------------------ | | 27| 956k|#define invariant(expr) assert(expr) ------------------ | Branch (95:5): [True: 956k, False: 18.4E] ------------------ 96| 956k| invariant(numBytes % StreamTraits::MinLoadStepByteMultiple == 0); ------------------ | | 27| 956k|#define invariant(expr) assert(expr) ------------------ | Branch (96:5): [True: 956k, False: 2] ------------------ 97| 956k| Base::pos += numBytes; 98| 956k| } _ZN8rawspeed11BitStreamerINS_15BitStreamerJPEGENS_39BitStreamerForwardSequentialReplenisherIS1_EEE14peekBitsNoFillEi: 253| 10.7M| uint32_t RAWSPEED_READONLY peekBitsNoFill(int nbits) { 254| 10.7M| establishClassInvariants(); 255| 10.7M| invariant(nbits >= 0); ------------------ | | 27| 10.7M|#define invariant(expr) assert(expr) ------------------ | Branch (255:5): [True: 10.7M, False: 18.4E] ------------------ 256| 10.7M| invariant(nbits != 0); ------------------ | | 27| 10.7M|#define invariant(expr) assert(expr) ------------------ | Branch (256:5): [True: 10.7M, False: 6] ------------------ 257| 10.7M| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 10.7M|#define invariant(expr) assert(expr) ------------------ | Branch (257:5): [True: 10.7M, False: 18.4E] ------------------ 258| 10.7M| return cache.peek(nbits); 259| 10.7M| } _ZN8rawspeed11BitStreamerINS_15BitStreamerJPEGENS_39BitStreamerForwardSequentialReplenisherIS1_EEE14skipBitsNoFillEi: 261| 10.7M| void skipBitsNoFill(int nbits) { 262| 10.7M| establishClassInvariants(); 263| 10.7M| invariant(nbits >= 0); ------------------ | | 27| 10.7M|#define invariant(expr) assert(expr) ------------------ | Branch (263:5): [True: 10.7M, False: 18.4E] ------------------ 264| | // `nbits` could be zero. 265| 10.7M| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 10.7M|#define invariant(expr) assert(expr) ------------------ | Branch (265:5): [True: 10.7M, False: 18.4E] ------------------ 266| 10.7M| cache.skip(nbits); 267| 10.7M| } _ZN8rawspeed11BitStreamerINS_15BitStreamerJPEGENS_39BitStreamerForwardSequentialReplenisherIS1_EEE13getBitsNoFillEi: 269| 20.3k| uint32_t getBitsNoFill(int nbits) { 270| 20.3k| establishClassInvariants(); 271| 20.3k| invariant(nbits >= 0); ------------------ | | 27| 20.3k|#define invariant(expr) assert(expr) ------------------ | Branch (271:5): [True: 20.3k, False: 0] ------------------ 272| 20.3k| invariant(nbits != 0); ------------------ | | 27| 20.3k|#define invariant(expr) assert(expr) ------------------ | Branch (272:5): [True: 20.3k, False: 0] ------------------ 273| 20.3k| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 20.3k|#define invariant(expr) assert(expr) ------------------ | Branch (273:5): [True: 20.3k, False: 0] ------------------ 274| 20.3k| uint32_t ret = peekBitsNoFill(nbits); 275| 20.3k| skipBitsNoFill(nbits); 276| 20.3k| return ret; 277| 20.3k| } _ZN8rawspeed11BitStreamerINS_14BitStreamerMSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE8peekBitsEi: 279| 128| uint32_t peekBits(int nbits) { 280| 128| establishClassInvariants(); 281| 128| invariant(nbits >= 0); ------------------ | | 27| 128|#define invariant(expr) assert(expr) ------------------ | Branch (281:5): [True: 128, False: 0] ------------------ 282| 128| invariant(nbits != 0); ------------------ | | 27| 128|#define invariant(expr) assert(expr) ------------------ | Branch (282:5): [True: 128, False: 0] ------------------ 283| 128| invariant(nbits <= Cache::MaxGetBits); ------------------ | | 27| 128|#define invariant(expr) assert(expr) ------------------ | Branch (283:5): [True: 128, False: 0] ------------------ 284| 128| fill(nbits); 285| 128| return peekBitsNoFill(nbits); 286| 128| } _ZNK8rawspeed11BitStreamerINS_14BitStreamerLSBENS_39BitStreamerForwardSequentialReplenisherIS1_EEE12getFillLevelEv: 248| 8.29M| [[nodiscard]] size_type RAWSPEED_READONLY getFillLevel() const { 249| 8.29M| establishClassInvariants(); 250| 8.29M| return cache.fillLevel; 251| 8.29M| } _ZNK8rawspeed11BitStreamerINS_16BitStreamerMSB32ENS_39BitStreamerForwardSequentialReplenisherIS1_EEE16getInputPositionEv: 232| 7.82k| [[nodiscard]] size_type RAWSPEED_READONLY getInputPosition() const { 233| 7.82k| establishClassInvariants(); 234| 7.82k| return replenisher.getPos(); 235| 7.82k| } _ZNK8rawspeed11BitStreamerINS_16BitStreamerMSB32ENS_39BitStreamerForwardSequentialReplenisherIS1_EEE16getRemainingSizeEv: 243| 189| [[nodiscard]] size_type getRemainingSize() const { 244| 189| establishClassInvariants(); 245| 189| return replenisher.getRemainingSize(); 246| 189| } _ZNK8rawspeed39BitStreamerForwardSequentialReplenisherINS_16BitStreamerMSB32EE16getRemainingSizeEv: 88| 189| [[nodiscard]] typename Base::size_type getRemainingSize() const { 89| 189| Base::establishClassInvariants(); 90| 189| return Base::input.size() - getPos(); 91| 189| } _ZNK8rawspeed11BitStreamerINS_16BitStreamerMSB32ENS_39BitStreamerForwardSequentialReplenisherIS1_EEE17getStreamPositionEv: 238| 7.42k| [[nodiscard]] size_type getStreamPosition() const { 239| 7.42k| establishClassInvariants(); 240| 7.42k| return getInputPosition() - (cache.fillLevel >> 3); 241| 7.42k| } _ZN8rawspeed15BitStreamerJPEG9fillCacheENSt3__15arrayISt4byteLm8EEE: 108| 956k| inputStorage) { 109| 956k| static_assert(BitStreamCacheBase::MaxGetBits >= 32, "check implementation"); 110| 956k| establishClassInvariants(); 111| 956k| auto input = Array1DRef(inputStorage.data(), 112| 956k| implicit_cast(inputStorage.size())); 113| 956k| invariant(input.size() == Traits::MaxProcessBytes); ------------------ | | 27| 956k|#define invariant(expr) assert(expr) ------------------ | Branch (113:3): [True: 956k, False: 2] ------------------ 114| | 115| 956k| constexpr int StreamChunkBitwidth = 116| 956k| bitwidth(); 117| | 118| 956k| auto speculativeOptimisticCache = cache; 119| 956k| auto speculativeOptimisticChunk = 120| 956k| getByteSwapped( 121| 956k| input.begin(), StreamTraits::ChunkEndianness != getHostEndianness()); 122| 956k| speculativeOptimisticCache.push(speculativeOptimisticChunk, 123| 956k| StreamChunkBitwidth); 124| | 125| | // short-cut path for the most common case (no FF marker in the next 4 bytes) 126| | // this is slightly faster than the else-case alone. 127| 956k| if (std::accumulate(&input(0), &input(4), true, [](bool b, std::byte byte) { ------------------ | Branch (127:7): [True: 954k, False: 2.25k] ------------------ 128| 956k| return b && (byte != std::byte{0xFF}); 129| 956k| })) { 130| 954k| cache = speculativeOptimisticCache; 131| 954k| return 4; 132| 954k| } 133| | 134| 2.25k| size_type p = 0; 135| 10.4k| for (size_type i = 0; i < 4; ++i) { ------------------ | Branch (135:25): [True: 8.51k, False: 1.95k] ------------------ 136| 8.51k| const int numBytesNeeded = 4 - i; 137| | 138| | // Pre-execute most common case, where next byte is 'normal'/non-FF 139| 8.51k| const std::byte c0 = input(p + 0); 140| 8.51k| cache.push(std::to_integer(c0), 8); 141| 8.51k| if (c0 != std::byte{0xFF}) { ------------------ | Branch (141:9): [True: 5.90k, False: 2.61k] ------------------ 142| 5.90k| p += 1; 143| 5.90k| continue; // Got normal byte. 144| 5.90k| } 145| | 146| | // Found FF -> pre-execute case of FF/00, which represents an FF data byte 147| 2.61k| const std::byte c1 = input(p + 1); 148| 2.61k| if (c1 == std::byte{0x00}) { ------------------ | Branch (148:9): [True: 2.30k, False: 301] ------------------ 149| | // Got FF/00, where 0x00 is a stuffing byte (that should be ignored), 150| | // so 0xFF is a normal byte. All good. 151| 2.30k| p += 2; 152| 2.30k| continue; 153| 2.30k| } 154| | 155| | // Found FF/xx with xx != 00. This is the end of stream marker. 156| 301| endOfStreamPos = getInputPosition() + p; 157| | 158| | // That means we shouldn't have pushed last 8 bits (0xFF, from c0). 159| | // We need to "unpush" them, and fill the vacant cache bits with zeros. 160| | 161| | // First, recover the cache fill level. 162| 301| cache.fillLevel -= 8; 163| | // Now, this code is incredibly underencapsulated, and the 164| | // implementation details are leaking into here. Thus, we know that 165| | // all the fillLevel bits in cache are all high bits. So to "unpush" 166| | // the last 8 bits, and fill the vacant cache bits with zeros, we only 167| | // need to keep the high fillLevel bits. So just create a mask with only 168| | // high fillLevel bits set, and 'and' the cache with it. 169| | // Caution, we know fillLevel won't be 64, but it may be 0, 170| | // so pick the mask-creation idiom accordingly. 171| 301| cache.cache &= ~((~0ULL) >> cache.fillLevel); 172| 301| cache.fillLevel = 64; 173| | 174| | // No further reading from this buffer shall happen. Do signal that by 175| | // claiming that we have consumed all the remaining bytes of the buffer. 176| | 177| 301| p = getRemainingSize() + numBytesNeeded; 178| 301| invariant(p >= 6); ------------------ | | 27| 301|#define invariant(expr) assert(expr) ------------------ | Branch (178:5): [True: 301, False: 0] ------------------ 179| 301| break; 180| 301| } 181| 2.25k| invariant(p >= 5); ------------------ | | 27| 2.25k|#define invariant(expr) assert(expr) ------------------ | Branch (181:3): [True: 2.24k, False: 14] ------------------ 182| 2.24k| return p; 183| 2.25k|} _ZNK8rawspeed15BitStreamerJPEG17getStreamPositionEv: 185| 151|inline BitStreamerJPEG::size_type BitStreamerJPEG::getStreamPosition() const { 186| | // The current number of bytes we consumed. 187| | // When at the end of the stream pos, it points to the JPEG marker FF 188| 151| return endOfStreamPos.value_or(getInputPosition()); 189| 151|} _ZN8rawspeed12PosOrUnknownIiEaSIiQsr3stdE7same_asITL0__T_EEERS1_S4_: 52| 301| PosOrUnknown& operator=(U newValue) { 53| 301| invariant(!has_value()); ------------------ | | 27| 301|#define invariant(expr) assert(expr) ------------------ | Branch (53:5): [True: 301, False: 0] ------------------ 54| 301| val = newValue; 55| 301| invariant(has_value()); ------------------ | | 27| 301|#define invariant(expr) assert(expr) ------------------ | Branch (55:5): [True: 301, False: 0] ------------------ 56| 301| return *this; 57| 301| } _ZNK8rawspeed12PosOrUnknownIiE9has_valueEv: 48| 753| [[nodiscard]] bool has_value() const RAWSPEED_READONLY { return val >= 0; } _ZNK8rawspeed12PosOrUnknownIiE8value_orIiQsr3stdE7same_asITL0__T_EEEiS4_: 61| 151| [[nodiscard]] T value_or(U fallback) const { 62| 151| if (has_value()) ------------------ | Branch (62:9): [True: 70, False: 81] ------------------ 63| 70| return val; 64| 81| return fallback; 65| 151| } _ZN8rawspeed12PosOrUnknownIiEC2Ev: 46| 869| PosOrUnknown() = default; _ZZN8rawspeed15BitStreamerJPEG9fillCacheENSt3__15arrayISt4byteLm8EEEENKUlbS3_E_clEbS3_: 127| 3.82M| if (std::accumulate(&input(0), &input(4), true, [](bool b, std::byte byte) { 128| 3.82M| return b && (byte != std::byte{0xFF}); ------------------ | Branch (128:16): [True: 3.82M, False: 3.48k] | Branch (128:21): [True: 3.81M, False: 2.24k] ------------------ 129| 3.82M| })) { _ZN8rawspeed18AbstractPrefixCodeINS_15BaselineCodeTagEEC2Ev: 178| 2.26k| AbstractPrefixCode() = default; _ZN8rawspeed18AbstractPrefixCodeINS_15BaselineCodeTagEE10CodeSymbolC2Eth: 151| 11.2k| : code(code_), code_len(code_len_) { 152| 11.2k| assert(code_len > 0); ------------------ | Branch (152:7): [True: 11.2k, False: 0] ------------------ 153| 11.2k| assert(code_len <= Traits::MaxCodeLenghtBits); ------------------ | Branch (153:7): [True: 11.2k, False: 0] ------------------ 154| 11.2k| assert(code <= ((1U << code_len) - 1U)); ------------------ | Branch (154:7): [True: 11.2k, False: 0] ------------------ 155| 11.2k| } _ZN8rawspeed18AbstractPrefixCodeINS_15BaselineCodeTagEEC2ENSt3__16vectorIhNS3_9allocatorIhEEEE: 181| 2.05k| : codeValues(std::move(codeValues_)) { 182| 2.05k| if (codeValues.empty()) ------------------ | Branch (182:9): [True: 0, False: 2.05k] ------------------ 183| 2.05k| ThrowRDE("Empty code alphabet?"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 184| 2.05k| assert( ------------------ | Branch (184:5): [True: 2.05k, False: 1] ------------------ 185| 2.05k| all_of(codeValues.begin(), codeValues.end(), 186| 2.05k| [](const CodeValueTy& v) { return v <= Traits::MaxCodeValue; })); 187| 2.05k| } _ZN8rawspeed18AbstractPrefixCodeINS_15BaselineCodeTagEE10CodeSymbol16HaveCommonPrefixERKS3_S5_: 163| 50.5k| const CodeSymbol& partial) { 164| 50.5k| assert(partial.code_len <= symbol.code_len); ------------------ | Branch (164:7): [True: 50.5k, False: 0] ------------------ 165| | 166| 50.5k| const auto s0 = extractHighBits(symbol.code, partial.code_len, 167| 50.5k| /*effectiveBitwidth=*/symbol.code_len); 168| 50.5k| const auto s1 = partial.code; 169| | 170| 50.5k| return s0 == s1; 171| 50.5k| } _ZN8rawspeed25AbstractPrefixCodeDecoderINS_15BaselineCodeTagEE6extendEjj: 70| 1.62M| static int RAWSPEED_READNONE extend(uint32_t diff, uint32_t len) { 71| 1.62M| invariant(len > 0); ------------------ | | 27| 1.62M|#define invariant(expr) assert(expr) ------------------ | Branch (71:5): [True: 1.62M, False: 0] ------------------ 72| 1.62M| auto ret = static_cast(diff); 73| 1.62M| if ((diff & (1 << (len - 1))) == 0) ------------------ | Branch (73:9): [True: 590k, False: 1.03M] ------------------ 74| 590k| ret -= (1 << len) - 1; 75| 1.62M| return ret; 76| 1.62M| } _ZNK8rawspeed25AbstractPrefixCodeDecoderINS_15BaselineCodeTagEE13processSymbolINS_15BitStreamerJPEGELb1EEEiRT_NS_18AbstractPrefixCodeIS1_E10CodeSymbolEh: 45| 16.3k| typename Traits::CodeValueTy codeValue) const { 46| 16.3k| invariant(symbol.code_len >= 0 && ------------------ | | 27| 16.3k|#define invariant(expr) assert(expr) ------------------ | Branch (46:5): [True: 16.3k, False: 0] | Branch (46:5): [True: 16.3k, False: 0] | Branch (46:5): [True: 16.3k, False: 0] ------------------ 47| 16.3k| symbol.code_len <= Traits::MaxCodeLenghtBits); 48| | 49| | // If we were only looking for symbol's code value, then just return it. 50| | if constexpr (!FULL_DECODE) 51| | return codeValue; 52| | 53| | // Else, treat it as the length of following difference 54| | // that we need to read and extend. 55| 16.3k| int diff_l = codeValue; 56| 16.3k| invariant(diff_l >= 0 && diff_l <= 16); ------------------ | | 27| 16.3k|#define invariant(expr) assert(expr) ------------------ | Branch (56:5): [True: 16.3k, False: 0] | Branch (56:5): [True: 16.3k, False: 0] | Branch (56:5): [True: 16.3k, False: 0] ------------------ 57| | 58| 16.3k| if (diff_l == 16) { ------------------ | Branch (58:9): [True: 3.32k, False: 13.0k] ------------------ 59| 3.32k| if (Base::handleDNGBug16()) ------------------ | Branch (59:11): [True: 1.85k, False: 1.47k] ------------------ 60| 1.85k| bs.skipBitsNoFill(16); 61| 3.32k| return -32768; 62| 3.32k| } 63| | 64| 13.0k| invariant(symbol.code_len + diff_l <= 32); ------------------ | | 27| 13.0k|#define invariant(expr) assert(expr) ------------------ | Branch (64:5): [True: 13.0k, False: 0] ------------------ 65| 13.0k| return diff_l ? extend(bs.getBitsNoFill(diff_l), diff_l) : 0; ------------------ | Branch (65:12): [True: 11.5k, False: 1.46k] ------------------ 66| 13.0k| } _ZNK8rawspeed25AbstractPrefixCodeDecoderINS_15BaselineCodeTagEE13processSymbolINS_14BitStreamerMSBELb1EEEiRT_NS_18AbstractPrefixCodeIS1_E10CodeSymbolEh: 45| 29.9k| typename Traits::CodeValueTy codeValue) const { 46| 29.9k| invariant(symbol.code_len >= 0 && ------------------ | | 27| 29.9k|#define invariant(expr) assert(expr) ------------------ | Branch (46:5): [True: 29.9k, False: 0] | Branch (46:5): [True: 29.9k, False: 0] | Branch (46:5): [True: 29.9k, False: 0] ------------------ 47| 29.9k| symbol.code_len <= Traits::MaxCodeLenghtBits); 48| | 49| | // If we were only looking for symbol's code value, then just return it. 50| | if constexpr (!FULL_DECODE) 51| | return codeValue; 52| | 53| | // Else, treat it as the length of following difference 54| | // that we need to read and extend. 55| 29.9k| int diff_l = codeValue; 56| 29.9k| invariant(diff_l >= 0 && diff_l <= 16); ------------------ | | 27| 29.9k|#define invariant(expr) assert(expr) ------------------ | Branch (56:5): [True: 29.9k, False: 0] | Branch (56:5): [True: 29.9k, False: 0] | Branch (56:5): [True: 29.9k, False: 0] ------------------ 57| | 58| 29.9k| if (diff_l == 16) { ------------------ | Branch (58:9): [True: 0, False: 29.9k] ------------------ 59| 0| if (Base::handleDNGBug16()) ------------------ | Branch (59:11): [True: 0, False: 0] ------------------ 60| 0| bs.skipBitsNoFill(16); 61| 0| return -32768; 62| 0| } 63| | 64| 29.9k| invariant(symbol.code_len + diff_l <= 32); ------------------ | | 27| 29.9k|#define invariant(expr) assert(expr) ------------------ | Branch (64:5): [True: 29.9k, False: 0] ------------------ 65| 29.9k| return diff_l ? extend(bs.getBitsNoFill(diff_l), diff_l) : 0; ------------------ | Branch (65:12): [True: 29.8k, False: 69] ------------------ 66| 29.9k| } _ZNK8rawspeed25AbstractPrefixCodeDecoderINS_15BaselineCodeTagEE13processSymbolINS_16BitStreamerMSB32ELb0EEEiRT_NS_18AbstractPrefixCodeIS1_E10CodeSymbolEh: 45| 580| typename Traits::CodeValueTy codeValue) const { 46| 580| invariant(symbol.code_len >= 0 && ------------------ | | 27| 580|#define invariant(expr) assert(expr) ------------------ | Branch (46:5): [True: 580, False: 0] | Branch (46:5): [True: 580, False: 0] | Branch (46:5): [True: 580, False: 0] ------------------ 47| 580| symbol.code_len <= Traits::MaxCodeLenghtBits); 48| | 49| | // If we were only looking for symbol's code value, then just return it. 50| | if constexpr (!FULL_DECODE) 51| 580| return codeValue; 52| | 53| | // Else, treat it as the length of following difference 54| | // that we need to read and extend. 55| 0| int diff_l = codeValue; 56| 580| invariant(diff_l >= 0 && diff_l <= 16); ------------------ | | 27| 580|#define invariant(expr) assert(expr) ------------------ | Branch (56:5): [True: 0, False: 580] | Branch (56:5): [True: 0, False: 0] | Branch (56:5): [True: 0, False: 580] ------------------ 57| | 58| 0| if (diff_l == 16) { ------------------ | Branch (58:9): [True: 0, False: 0] ------------------ 59| 0| if (Base::handleDNGBug16()) ------------------ | Branch (59:11): [True: 0, False: 0] ------------------ 60| 0| bs.skipBitsNoFill(16); 61| 0| return -32768; 62| 0| } 63| | 64| 0| invariant(symbol.code_len + diff_l <= 32); ------------------ | | 27| 0|#define invariant(expr) assert(expr) ------------------ | Branch (64:5): [True: 0, False: 0] ------------------ 65| 0| return diff_l ? extend(bs.getBitsNoFill(diff_l), diff_l) : 0; ------------------ | Branch (65:12): [True: 0, False: 0] ------------------ 66| 0| } _ZN8rawspeed28AbstractPrefixCodeTranscoderINS_15BaselineCodeTagEEC2ENS_10PrefixCodeIS1_EE: 48| 2.05k| : code(std::move(code_)) {} _ZN8rawspeed28AbstractPrefixCodeTranscoderINS_15BaselineCodeTagEE5setupEbb: 71| 2.05k| void setup(bool fullDecode_, bool fixDNGBug16_) { 72| 2.05k| invariant(!fullDecode_ || Traits::SupportsFullDecode); ------------------ | | 27| 2.05k|#define invariant(expr) assert(expr) ------------------ | Branch (72:5): [True: 207, False: 1.84k] | Branch (72:5): [True: 1.84k, Folded] | Branch (72:5): [True: 2.05k, False: 0] ------------------ 73| | 74| 2.05k| this->fullDecode = fullDecode_; 75| 2.05k| this->fixDNGBug16 = fixDNGBug16_; 76| | 77| 2.05k| if (fullDecode) { ------------------ | Branch (77:9): [True: 1.84k, False: 207] ------------------ 78| | // If we are in a full-decoding mode, we will be interpreting code values 79| | // as bit length of the following difference, which incurs hard limit 80| | // of 16 (since we want to need to read at most 32 bits max for a symbol 81| | // plus difference). Though we could enforce it per-code instead? 82| 1.84k| verifyCodeValuesAsDiffLengths(); 83| 1.84k| } 84| 2.05k| } _ZNK8rawspeed28AbstractPrefixCodeTranscoderINS_15BaselineCodeTagEE29verifyCodeValuesAsDiffLengthsEv: 50| 1.93k| void verifyCodeValuesAsDiffLengths() const { 51| 9.80k| for (const auto cValue : code.Base::codeValues) { ------------------ | Branch (51:28): [True: 9.80k, False: 1.90k] ------------------ 52| 9.80k| if (cValue <= Traits::MaxDiffLength) ------------------ | Branch (52:11): [True: 9.76k, False: 36] ------------------ 53| 9.76k| continue; 54| 9.80k| ThrowRDE("Corrupt Huffman code: difference length %u longer than %u", ------------------ | | 38| 36| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 36| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 36| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 36| __VA_ARGS__) | | ------------------ ------------------ 55| 9.80k| cValue, Traits::MaxDiffLength); 56| 9.80k| } 57| 1.93k| assert(maxCodePlusDiffLength() <= 32U); ------------------ | Branch (57:5): [True: 1.90k, False: 0] ------------------ 58| 1.90k| } _ZNK8rawspeed28AbstractPrefixCodeTranscoderINS_15BaselineCodeTagEE21maxCodePlusDiffLengthEv: 66| 1.90k| maxCodePlusDiffLength() const { 67| 1.90k| return maxCodeLength() + *(std::max_element(code.Base::codeValues.cbegin(), 68| 1.90k| code.Base::codeValues.cend())); 69| 1.90k| } _ZNK8rawspeed28AbstractPrefixCodeTranscoderINS_15BaselineCodeTagEE13maxCodeLengthEv: 61| 45.3k| [[nodiscard]] size_t RAWSPEED_READONLY maxCodeLength() const { 62| 45.3k| return code.nCodesPerLength.size() - 1; 63| 45.3k| } _ZNK8rawspeed28AbstractPrefixCodeTranscoderINS_15BaselineCodeTagEE12isFullDecodeEv: 87| 25.8M| [[nodiscard]] bool RAWSPEED_READONLY isFullDecode() const { 88| 25.8M| return fullDecode; 89| 25.8M| } _ZNK8rawspeed28AbstractPrefixCodeTranscoderINS_15BaselineCodeTagEE14handleDNGBug16Ev: 90| 150k| [[nodiscard]] bool RAWSPEED_READONLY handleDNGBug16() const { 91| 150k| return fixDNGBug16; 92| 150k| } _ZN8rawspeed11HuffmanCodeINS_15BaselineCodeTagEEC2Ev: 48| 2.26k| HuffmanCode() = default; _ZN8rawspeed11HuffmanCodeINS_15BaselineCodeTagEE18setNCodesPerLengthENS_6BufferE: 99| 2.24k| uint32_t setNCodesPerLength(Buffer data) { 100| 2.24k| invariant(data.getSize() == Traits::MaxCodeLenghtBits); ------------------ | | 27| 2.24k|#define invariant(expr) assert(expr) ------------------ | Branch (100:5): [True: 2.24k, False: 0] ------------------ 101| | 102| 2.24k| nCodesPerLength.resize((1 + Traits::MaxCodeLenghtBits), 0); 103| 2.24k| std::copy(data.begin(), data.end(), &nCodesPerLength[1]); 104| 2.24k| assert(nCodesPerLength[0] == 0); ------------------ | Branch (104:5): [True: 2.24k, False: 0] ------------------ 105| | 106| | // trim empty entries from the codes per length table on the right 107| 21.6k| while (!nCodesPerLength.empty() && nCodesPerLength.back() == 0) ------------------ | Branch (107:12): [True: 21.6k, False: 3] | Branch (107:40): [True: 19.3k, False: 2.23k] ------------------ 108| 19.3k| nCodesPerLength.pop_back(); 109| | 110| 2.24k| if (nCodesPerLength.empty()) ------------------ | Branch (110:9): [True: 3, False: 2.23k] ------------------ 111| 2.24k| ThrowRDE("Codes-per-length table is empty"); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 112| | 113| 2.24k| assert(nCodesPerLength.back() > 0); ------------------ | Branch (113:5): [True: 2.23k, False: 0] ------------------ 114| | 115| 2.23k| const auto count = maxCodesCount(); 116| 2.23k| invariant(count > 0); ------------------ | | 27| 2.23k|#define invariant(expr) assert(expr) ------------------ | Branch (116:5): [True: 2.23k, False: 0] ------------------ 117| | 118| 2.23k| if (count > Traits::MaxNumCodeValues) ------------------ | Branch (118:9): [True: 22, False: 2.21k] ------------------ 119| 2.23k| ThrowRDE("Too big code-values table"); ------------------ | | 38| 22| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 22| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 22| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 22| __VA_ARGS__) | | ------------------ ------------------ 120| | 121| | // We are at the Root node, len is 1, there are two possible child Nodes 122| 2.21k| unsigned maxCodes = 2; 123| | 124| 18.3k| for (auto codeLen = 1UL; codeLen < nCodesPerLength.size(); codeLen++) { ------------------ | Branch (124:30): [True: 16.1k, False: 2.19k] ------------------ 125| | // we have codeLen bits. make sure that that code count can actually fit 126| | // E.g. for len 1 we could have two codes: 0b0 and 0b1 127| | // (but in that case there can be no other codes (with higher lengths)) 128| 16.1k| const auto maxCodesInCurrLen = (1U << codeLen); 129| 16.1k| const auto nCodes = nCodesPerLength[codeLen]; 130| 16.1k| if (nCodes > maxCodesInCurrLen) { ------------------ | Branch (130:11): [True: 15, False: 16.1k] ------------------ 131| 15| ThrowRDE("Corrupt Huffman. Can never have %u codes in %lu-bit len", ------------------ | | 38| 15| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 15| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 15| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 15| __VA_ARGS__) | | ------------------ ------------------ 132| 15| nCodes, codeLen); 133| 15| } 134| | 135| | // Also, check that we actually can have this much leafs for this length 136| 16.1k| if (nCodes > maxCodes) { ------------------ | Branch (136:11): [True: 8, False: 16.1k] ------------------ 137| 8| ThrowRDE( ------------------ | | 38| 8| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 8| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 8| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 8| __VA_ARGS__) | | ------------------ ------------------ 138| 8| "Corrupt Huffman. Can only fit %u out of %u codes in %lu-bit len", 139| 8| maxCodes, nCodes, codeLen); 140| 8| } 141| | 142| | // There are nCodes leafs on this level, and those can not be branches 143| 16.1k| maxCodes -= nCodes; 144| | // On the next level, rest can be branches, and can have two child Nodes 145| 16.1k| maxCodes *= 2; 146| 16.1k| } 147| | 148| 2.19k| return count; 149| 2.21k| } _ZNK8rawspeed11HuffmanCodeINS_15BaselineCodeTagEE13maxCodesCountEv: 61| 16.9k| [[nodiscard]] unsigned int RAWSPEED_READONLY maxCodesCount() const { 62| 16.9k| return std::accumulate(nCodesPerLength.begin(), nCodesPerLength.end(), 0U); 63| 16.9k| } _ZN8rawspeed11HuffmanCodeINS_15BaselineCodeTagEE13setCodeValuesENS_10Array1DRefIKhEE: 151| 2.17k| void setCodeValues(Array1DRef data) { 152| 2.17k| invariant(data.size() <= Traits::MaxNumCodeValues); ------------------ | | 27| 2.17k|#define invariant(expr) assert(expr) ------------------ | Branch (152:5): [True: 2.17k, False: 0] ------------------ 153| 2.17k| invariant(static_cast(data.size()) == maxCodesCount()); ------------------ | | 27| 2.17k|#define invariant(expr) assert(expr) ------------------ | Branch (153:5): [True: 2.17k, False: 0] ------------------ 154| | 155| 2.17k| this->codeValues.clear(); 156| 2.17k| this->codeValues.reserve(maxCodesCount()); 157| 2.17k| std::copy(data.begin(), data.end(), std::back_inserter(this->codeValues)); 158| 2.17k| assert(this->codeValues.size() == maxCodesCount()); ------------------ | Branch (158:5): [True: 2.17k, False: 0] ------------------ 159| | 160| 11.5k| for (const auto& cValue : this->codeValues) { ------------------ | Branch (160:29): [True: 11.5k, False: 2.17k] ------------------ 161| 11.5k| if (cValue <= Traits::MaxCodeValue) ------------------ | Branch (161:11): [True: 11.5k, False: 0] ------------------ 162| 11.5k| continue; 163| 0| ThrowRDE("Corrupt Huffman code: code value %u is larger than maximum %u", ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 164| 11.5k| cValue, Traits::MaxCodeValue); 165| 11.5k| } 166| 2.17k| } _ZNK8rawspeed11HuffmanCodeINS_15BaselineCodeTagEEeqERKS2_: 94| 746| bool operator==(const HuffmanCode& other) const { 95| 746| return nCodesPerLength == other.nCodesPerLength && ------------------ | Branch (95:12): [True: 546, False: 200] ------------------ 96| 546| this->codeValues == other.codeValues; ------------------ | Branch (96:12): [True: 122, False: 424] ------------------ 97| 746| } _ZN8rawspeed11HuffmanCodeINS_15BaselineCodeTagEEcvNS_10PrefixCodeIS1_EEEv: 168| 2.05k| explicit operator PrefixCode() { 169| 2.05k| std::vector symbols = generateCodeSymbols(); 170| 2.05k| return {std::move(symbols), std::move(Parent::codeValues)}; 171| 2.05k| } _ZNK8rawspeed11HuffmanCodeINS_15BaselineCodeTagEE19generateCodeSymbolsEv: 66| 2.05k| [[nodiscard]] std::vector generateCodeSymbols() const { 67| 2.05k| std::vector symbols; 68| | 69| 2.05k| assert(!nCodesPerLength.empty()); ------------------ | Branch (69:5): [True: 2.05k, False: 0] ------------------ 70| 2.05k| assert(maxCodesCount() > 0); ------------------ | Branch (70:5): [True: 2.05k, False: 0] ------------------ 71| | 72| 2.05k| assert(this->codeValues.size() == maxCodesCount()); ------------------ | Branch (72:5): [True: 2.05k, False: 0] ------------------ 73| | 74| | // reserve all the memory. avoids lots of small allocs 75| 2.05k| symbols.reserve(maxCodesCount()); 76| | 77| | // Figure C.1: make table of Huffman code length for each symbol 78| | // Figure C.2: generate the codes themselves 79| 2.05k| uint32_t code = 0; 80| 17.5k| for (unsigned int l = 1; l <= maxCodeLength(); ++l) { ------------------ | Branch (80:30): [True: 15.5k, False: 2.05k] ------------------ 81| 26.7k| for (unsigned int i = 0; i < nCodesPerLength[l]; ++i) { ------------------ | Branch (81:32): [True: 11.2k, False: 15.5k] ------------------ 82| 11.2k| symbols.emplace_back(code, l); 83| 11.2k| code++; 84| 11.2k| } 85| | 86| 15.5k| code <<= 1; 87| 15.5k| } 88| | 89| 2.05k| assert(symbols.size() == maxCodesCount()); ------------------ | Branch (89:5): [True: 2.05k, False: 1] ------------------ 90| | 91| 2.05k| return symbols; 92| 2.05k| } _ZNK8rawspeed11HuffmanCodeINS_15BaselineCodeTagEE13maxCodeLengthEv: 51| 17.5k| [[nodiscard]] size_t RAWSPEED_READONLY maxCodeLength() const { 52| 17.5k| return nCodesPerLength.size() - 1; 53| 17.5k| } _ZN8rawspeed10PrefixCodeINS_15BaselineCodeTagEEC2ENSt3__16vectorINS_18AbstractPrefixCodeIS1_E10CodeSymbolENS3_9allocatorIS7_EEEENS4_IhNS8_IhEEEE: 52| 2.05k| : Base(std::move(codeValues_)), symbols(std::move(symbols_)) { 53| 2.05k| if (symbols.empty() || Base::codeValues.empty() || ------------------ | Branch (53:9): [True: 0, False: 2.05k] | Branch (53:28): [True: 0, False: 2.05k] ------------------ 54| 2.05k| symbols.size() != Base::codeValues.size()) ------------------ | Branch (54:9): [True: 0, False: 2.05k] ------------------ 55| 2.05k| ThrowRDE("Malformed code"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 56| | 57| 2.05k| nCodesPerLength.resize(1 + Traits::MaxCodeLenghtBits); 58| 11.2k| for (const CodeSymbol& s : symbols) { ------------------ | Branch (58:30): [True: 11.2k, False: 2.05k] ------------------ 59| 11.2k| assert(s.code_len > 0 && s.code_len <= Traits::MaxCodeLenghtBits); ------------------ | Branch (59:7): [True: 11.2k, False: 0] | Branch (59:7): [True: 11.2k, False: 0] | Branch (59:7): [True: 11.2k, False: 0] ------------------ 60| 11.2k| ++nCodesPerLength[s.code_len]; 61| 11.2k| } 62| 19.4k| while (nCodesPerLength.back() == 0) ------------------ | Branch (62:12): [True: 17.3k, False: 2.05k] ------------------ 63| 17.3k| nCodesPerLength.pop_back(); 64| 2.05k| assert(nCodesPerLength.size() > 1); ------------------ | Branch (64:5): [True: 2.05k, False: 18.4E] ------------------ 65| | 66| 2.05k| verifyCodeSymbols(); 67| 2.05k| } _ZN8rawspeed10PrefixCodeINS_15BaselineCodeTagEE17verifyCodeSymbolsEv: 70| 2.05k| void verifyCodeSymbols() { 71| | // We are at the Root node, len is 1, there are two possible child Nodes 72| 2.05k| unsigned maxCodes = 2; 73| 17.5k| for (auto codeLen = 1UL; codeLen < nCodesPerLength.size(); codeLen++) { ------------------ | Branch (73:30): [True: 15.5k, False: 2.05k] ------------------ 74| | // we have codeLen bits. make sure that that code count can actually fit 75| | // E.g. for len 1 we could have two codes: 0b0 and 0b1 76| | // (but in that case there can be no other codes (with higher lengths)) 77| 15.5k| const unsigned nCodes = nCodesPerLength[codeLen]; 78| 15.5k| if (nCodes > maxCodes) ------------------ | Branch (78:11): [True: 0, False: 15.5k] ------------------ 79| 15.5k| ThrowRDE("Too many codes of of length %lu.", codeLen); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 80| | // There are nCodes leafs on this level, and those can not be branches 81| 15.5k| maxCodes -= nCodes; 82| | // On the next level, rest can be branches, and can have two child Nodes 83| 15.5k| maxCodes *= 2; 84| 15.5k| } 85| | 86| | // The code symbols are ordered so that the code lengths are not decreasing. 87| | // NOTE: codes of same lenght are not nessesairly sorted! 88| 2.05k| if (std::adjacent_find( ------------------ | Branch (88:9): [True: 0, False: 2.05k] ------------------ 89| 2.05k| symbols.cbegin(), symbols.cend(), 90| 2.05k| [](const CodeSymbol& lhs, const CodeSymbol& rhs) -> bool { 91| 2.05k| return !std::less_equal<>()(lhs.code_len, rhs.code_len); 92| 2.05k| }) != symbols.cend()) 93| 2.05k| ThrowRDE("Code symbols are not globally ordered"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 94| | 95| | // No two symbols should have the same prefix (high bits) 96| | // Only analyze the lower triangular matrix, excluding diagonal 97| 13.3k| for (auto sId = 0UL; sId < symbols.size(); sId++) { ------------------ | Branch (97:26): [True: 11.2k, False: 2.05k] ------------------ 98| 61.7k| for (auto pId = 0UL; pId < sId; pId++) ------------------ | Branch (98:28): [True: 50.5k, False: 11.2k] ------------------ 99| 50.5k| if (CodeSymbol::HaveCommonPrefix(symbols[sId], symbols[pId])) ------------------ | Branch (99:13): [True: 0, False: 50.5k] ------------------ 100| 11.2k| ThrowRDE("Not prefix codes!"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 101| 11.2k| } 102| 2.05k| } _ZZN8rawspeed10PrefixCodeINS_15BaselineCodeTagEE17verifyCodeSymbolsEvENKUlRKNS_18AbstractPrefixCodeIS1_E10CodeSymbolES7_E_clES7_S7_: 90| 9.21k| [](const CodeSymbol& lhs, const CodeSymbol& rhs) -> bool { 91| 9.21k| return !std::less_equal<>()(lhs.code_len, rhs.code_len); 92| 9.21k| }) != symbols.cend()) _ZN8rawspeed20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS1_EEE5setupEbb: 95| 2.05k| void setup(bool fullDecode_, bool fixDNGBug16_) { 96| 2.05k| Base::setup(fullDecode_, fixDNGBug16_); 97| | 98| | // Generate lookup table for fast decoding lookup. 99| | // See definition of decodeLookup above 100| 2.05k| decodeLookup.resize(1 << LookupDepth); 101| 10.1k| for (size_t i = 0; i < Base::code.symbols.size(); i++) { ------------------ | Branch (101:24): [True: 8.67k, False: 1.42k] ------------------ 102| 8.67k| uint8_t code_l = Base::code.symbols[i].code_len; 103| 8.67k| if (code_l > static_cast(LookupDepth)) ------------------ | Branch (103:11): [True: 625, False: 8.05k] ------------------ 104| 625| break; 105| | 106| 8.05k| auto ll = implicit_cast(Base::code.symbols[i].code 107| 8.05k| << (LookupDepth - code_l)); 108| 8.05k| auto ul = 109| 8.05k| implicit_cast(ll | ((1 << (LookupDepth - code_l)) - 1)); 110| 8.05k| static_assert(Traits::MaxCodeValueLenghtBits <= 111| 8.05k| bitwidth() - PayloadShift); 112| 8.05k| LUTUnsignedEntryTy diff_l = Base::code.codeValues[i]; 113| 2.52M| for (uint16_t c = ll; c <= ul; c++) { ------------------ | Branch (113:29): [True: 2.51M, False: 8.05k] ------------------ 114| 2.51M| if (!(c < decodeLookup.size())) ------------------ | Branch (114:13): [True: 0, False: 2.51M] ------------------ 115| 2.51M| ThrowRDE("Corrupt Huffman"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 116| | 117| 2.51M| if (!FlagMask || !Base::isFullDecode() || code_l > LookupDepth || ------------------ | Branch (117:13): [Folded, False: 2.51M] | Branch (117:26): [True: 163k, False: 2.35M] | Branch (117:51): [True: 71, False: 2.35M] ------------------ 118| 2.35M| (code_l + diff_l > LookupDepth && diff_l != 16)) { ------------------ | Branch (118:14): [True: 521k, False: 1.83M] | Branch (118:47): [True: 374k, False: 147k] ------------------ 119| | // lookup bit depth is too small to fit both the encoded length 120| | // and the final difference value. 121| | // -> store only the length and do a normal sign extension later 122| 536k| invariant(!Base::isFullDecode() || diff_l > 0); ------------------ | | 27| 536k|#define invariant(expr) assert(expr) ------------------ | Branch (122:11): [True: 162k, False: 374k] | Branch (122:11): [True: 374k, False: 0] | Branch (122:11): [True: 536k, False: 0] ------------------ 123| 536k| decodeLookup[c] = diff_l << PayloadShift | code_l; 124| | 125| 536k| if (!Base::isFullDecode()) ------------------ | Branch (125:15): [True: 162k, False: 374k] ------------------ 126| 162k| decodeLookup[c] |= FlagMask; 127| 1.97M| } else { 128| | // Lookup bit depth is sufficient to encode the final value. 129| 1.97M| decodeLookup[c] = FlagMask | code_l; 130| 1.97M| if (diff_l != 16 || Base::handleDNGBug16()) ------------------ | Branch (130:15): [True: 1.83M, False: 147k] | Branch (130:31): [True: 362, False: 146k] ------------------ 131| 1.83M| decodeLookup[c] += diff_l; 132| | 133| 1.97M| if (diff_l) { ------------------ | Branch (133:15): [True: 1.06M, False: 913k] ------------------ 134| 1.06M| LUTUnsignedEntryTy diff; 135| 1.06M| if (diff_l != 16) { ------------------ | Branch (135:17): [True: 919k, False: 147k] ------------------ 136| 919k| diff = extractHighBits(c, code_l + diff_l, 137| 919k| /*effectiveBitwidth=*/LookupDepth); 138| 919k| diff &= ((1 << diff_l) - 1); 139| 919k| } else { 140| 147k| diff = LUTUnsignedEntryTy(-32768); 141| 147k| } 142| 1.06M| decodeLookup[c] |= static_cast( 143| 1.06M| static_cast(Base::extend(diff, diff_l)) 144| 1.06M| << PayloadShift); 145| 1.06M| } 146| 1.97M| } 147| 2.51M| } 148| 8.05k| } 149| 2.05k| } _ZNK8rawspeed20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS1_EEE16decodeDifferenceINS_15BitStreamerJPEGEEEiRT_: 161| 10.7M| __attribute__((always_inline)) int decodeDifference(BIT_STREAM& bs) const { 162| 10.7M| static_assert( 163| 10.7M| BitStreamerTraits::canUseWithPrefixCodeDecoder, 164| 10.7M| "This BitStreamer specialization is not marked as usable here"); 165| 10.7M| invariant(Base::isFullDecode()); ------------------ | | 27| 10.7M|#define invariant(expr) assert(expr) ------------------ | Branch (165:5): [True: 10.7M, False: 18.4E] ------------------ 166| 10.7M| return decode(bs); 167| 10.7M| } _ZNK8rawspeed20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS1_EEE6decodeINS_15BitStreamerJPEGELb1EEEiRT_: 174| 10.7M| __attribute__((always_inline)) int decode(BIT_STREAM& bs) const { 175| 10.7M| static_assert( 176| 10.7M| BitStreamerTraits::canUseWithPrefixCodeDecoder, 177| 10.7M| "This BitStreamer specialization is not marked as usable here"); 178| 10.7M| invariant(FULL_DECODE == Base::isFullDecode()); ------------------ | | 27| 10.7M|#define invariant(expr) assert(expr) ------------------ | Branch (178:5): [True: 10.7M, False: 3.07k] ------------------ 179| 10.7M| bs.fill(32); 180| | 181| 10.7M| typename Base::CodeSymbol partial; 182| 10.7M| partial.code_len = LookupDepth; 183| 10.7M| partial.code = implicit_cast( 184| 10.7M| bs.peekBitsNoFill(partial.code_len)); 185| | 186| 10.7M| assert(partial.code < decodeLookup.size()); ------------------ | Branch (186:5): [True: 10.7M, False: 18.4E] ------------------ 187| 10.7M| auto lutEntry = static_cast(decodeLookup[partial.code]); 188| 10.7M| int payload = static_cast(lutEntry) >> PayloadShift; 189| 10.7M| int len = lutEntry & LenMask; 190| | 191| | // How far did reading of those LookupDepth bits *actually* move us forward? 192| 10.7M| bs.skipBitsNoFill(len); 193| | 194| | // If the flag bit is set, then we have already skipped all the len bits 195| | // we needed to skip, and payload is the answer we were looking for. 196| 10.7M| if (lutEntry & FlagMask) ------------------ | Branch (196:9): [True: 10.7M, False: 15.2k] ------------------ 197| 10.7M| return payload; 198| | 199| 15.2k| typename Traits::CodeValueTy codeValue; 200| 15.2k| if (lutEntry) { ------------------ | Branch (200:9): [True: 11.3k, False: 3.91k] ------------------ 201| | // If the flag is not set, but the entry is not empty, 202| | // the payload is the code value for this symbol. 203| 11.3k| partial.code_len = implicit_cast(len); 204| 11.3k| codeValue = implicit_cast(payload); 205| 11.3k| invariant(!FULL_DECODE || codeValue /*aka diff_l*/ > 0); ------------------ | | 27| 11.3k|#define invariant(expr) assert(expr) ------------------ | Branch (205:7): [Folded, False: 11.3k] | Branch (205:7): [True: 11.3k, False: 0] | Branch (205:7): [True: 11.3k, False: 0] ------------------ 206| 11.3k| } else { 207| | // No match in the lookup table, because either the code is longer 208| | // than LookupDepth or the input is corrupt. Need to read more bits... 209| 3.91k| invariant(len == 0); ------------------ | | 27| 3.91k|#define invariant(expr) assert(expr) ------------------ | Branch (209:7): [True: 5.24k, False: 18.4E] ------------------ 210| 5.24k| bs.skipBitsNoFill(partial.code_len); 211| 5.24k| std::tie(partial, codeValue) = 212| 5.24k| Base::finishReadingPartialSymbol(bs, partial); 213| 5.24k| } 214| | 215| 16.6k| return Base::template processSymbol(bs, partial, 216| 16.6k| codeValue); 217| 15.2k| } _ZNK8rawspeed20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS1_EEE16decodeDifferenceINS_14BitStreamerMSBEEEiRT_: 161| 384k| __attribute__((always_inline)) int decodeDifference(BIT_STREAM& bs) const { 162| 384k| static_assert( 163| 384k| BitStreamerTraits::canUseWithPrefixCodeDecoder, 164| 384k| "This BitStreamer specialization is not marked as usable here"); 165| 384k| invariant(Base::isFullDecode()); ------------------ | | 27| 384k|#define invariant(expr) assert(expr) ------------------ | Branch (165:5): [True: 384k, False: 0] ------------------ 166| 384k| return decode(bs); 167| 384k| } _ZNK8rawspeed20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS1_EEE6decodeINS_14BitStreamerMSBELb1EEEiRT_: 174| 384k| __attribute__((always_inline)) int decode(BIT_STREAM& bs) const { 175| 384k| static_assert( 176| 384k| BitStreamerTraits::canUseWithPrefixCodeDecoder, 177| 384k| "This BitStreamer specialization is not marked as usable here"); 178| 384k| invariant(FULL_DECODE == Base::isFullDecode()); ------------------ | | 27| 384k|#define invariant(expr) assert(expr) ------------------ | Branch (178:5): [True: 384k, False: 0] ------------------ 179| 384k| bs.fill(32); 180| | 181| 384k| typename Base::CodeSymbol partial; 182| 384k| partial.code_len = LookupDepth; 183| 384k| partial.code = implicit_cast( 184| 384k| bs.peekBitsNoFill(partial.code_len)); 185| | 186| 384k| assert(partial.code < decodeLookup.size()); ------------------ | Branch (186:5): [True: 384k, False: 72] ------------------ 187| 384k| auto lutEntry = static_cast(decodeLookup[partial.code]); 188| 384k| int payload = static_cast(lutEntry) >> PayloadShift; 189| 384k| int len = lutEntry & LenMask; 190| | 191| | // How far did reading of those LookupDepth bits *actually* move us forward? 192| 384k| bs.skipBitsNoFill(len); 193| | 194| | // If the flag bit is set, then we have already skipped all the len bits 195| | // we needed to skip, and payload is the answer we were looking for. 196| 384k| if (lutEntry & FlagMask) ------------------ | Branch (196:9): [True: 354k, False: 29.9k] ------------------ 197| 354k| return payload; 198| | 199| 29.9k| typename Traits::CodeValueTy codeValue; 200| 29.9k| if (lutEntry) { ------------------ | Branch (200:9): [True: 29.8k, False: 108] ------------------ 201| | // If the flag is not set, but the entry is not empty, 202| | // the payload is the code value for this symbol. 203| 29.8k| partial.code_len = implicit_cast(len); 204| 29.8k| codeValue = implicit_cast(payload); 205| 29.8k| invariant(!FULL_DECODE || codeValue /*aka diff_l*/ > 0); ------------------ | | 27| 29.8k|#define invariant(expr) assert(expr) ------------------ | Branch (205:7): [Folded, False: 29.8k] | Branch (205:7): [True: 29.8k, False: 0] | Branch (205:7): [True: 29.8k, False: 0] ------------------ 206| 29.8k| } else { 207| | // No match in the lookup table, because either the code is longer 208| | // than LookupDepth or the input is corrupt. Need to read more bits... 209| 108| invariant(len == 0); ------------------ | | 27| 108|#define invariant(expr) assert(expr) ------------------ | Branch (209:7): [True: 108, False: 0] ------------------ 210| 108| bs.skipBitsNoFill(partial.code_len); 211| 108| std::tie(partial, codeValue) = 212| 108| Base::finishReadingPartialSymbol(bs, partial); 213| 108| } 214| | 215| 29.9k| return Base::template processSymbol(bs, partial, 216| 29.9k| codeValue); 217| 29.9k| } _ZNK8rawspeed20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS1_EEE15decodeCodeValueINS_16BitStreamerMSB32EEEiRT_: 152| 20.9k| __attribute__((always_inline)) int decodeCodeValue(BIT_STREAM& bs) const { 153| 20.9k| static_assert( 154| 20.9k| BitStreamerTraits::canUseWithPrefixCodeDecoder, 155| 20.9k| "This BitStreamer specialization is not marked as usable here"); 156| 20.9k| invariant(!Base::isFullDecode()); ------------------ | | 27| 20.9k|#define invariant(expr) assert(expr) ------------------ | Branch (156:5): [True: 20.9k, False: 0] ------------------ 157| 20.9k| return decode(bs); 158| 20.9k| } _ZNK8rawspeed20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS1_EEE6decodeINS_16BitStreamerMSB32ELb0EEEiRT_: 174| 20.9k| __attribute__((always_inline)) int decode(BIT_STREAM& bs) const { 175| 20.9k| static_assert( 176| 20.9k| BitStreamerTraits::canUseWithPrefixCodeDecoder, 177| 20.9k| "This BitStreamer specialization is not marked as usable here"); 178| 20.9k| invariant(FULL_DECODE == Base::isFullDecode()); ------------------ | | 27| 20.9k|#define invariant(expr) assert(expr) ------------------ | Branch (178:5): [True: 20.9k, False: 0] ------------------ 179| 20.9k| bs.fill(32); 180| | 181| 20.9k| typename Base::CodeSymbol partial; 182| 20.9k| partial.code_len = LookupDepth; 183| 20.9k| partial.code = implicit_cast( 184| 20.9k| bs.peekBitsNoFill(partial.code_len)); 185| | 186| 20.9k| assert(partial.code < decodeLookup.size()); ------------------ | Branch (186:5): [True: 20.9k, False: 0] ------------------ 187| 20.9k| auto lutEntry = static_cast(decodeLookup[partial.code]); 188| 20.9k| int payload = static_cast(lutEntry) >> PayloadShift; 189| 20.9k| int len = lutEntry & LenMask; 190| | 191| | // How far did reading of those LookupDepth bits *actually* move us forward? 192| 20.9k| bs.skipBitsNoFill(len); 193| | 194| | // If the flag bit is set, then we have already skipped all the len bits 195| | // we needed to skip, and payload is the answer we were looking for. 196| 20.9k| if (lutEntry & FlagMask) ------------------ | Branch (196:9): [True: 20.3k, False: 666] ------------------ 197| 20.3k| return payload; 198| | 199| 666| typename Traits::CodeValueTy codeValue; 200| 666| if (lutEntry) { ------------------ | Branch (200:9): [True: 0, False: 666] ------------------ 201| | // If the flag is not set, but the entry is not empty, 202| | // the payload is the code value for this symbol. 203| 0| partial.code_len = implicit_cast(len); 204| 0| codeValue = implicit_cast(payload); 205| 0| invariant(!FULL_DECODE || codeValue /*aka diff_l*/ > 0); ------------------ | | 27| 0|#define invariant(expr) assert(expr) ------------------ | Branch (205:7): [True: 0, Folded] | Branch (205:7): [True: 0, False: 0] | Branch (205:7): [True: 0, Folded] ------------------ 206| 666| } else { 207| | // No match in the lookup table, because either the code is longer 208| | // than LookupDepth or the input is corrupt. Need to read more bits... 209| 666| invariant(len == 0); ------------------ | | 27| 666|#define invariant(expr) assert(expr) ------------------ | Branch (209:7): [True: 666, False: 0] ------------------ 210| 666| bs.skipBitsNoFill(partial.code_len); 211| 666| std::tie(partial, codeValue) = 212| 666| Base::finishReadingPartialSymbol(bs, partial); 213| 666| } 214| | 215| 666| return Base::template processSymbol(bs, partial, 216| 666| codeValue); 217| 666| } _ZN8rawspeed23PrefixCodeLookupDecoderINS_15BaselineCodeTagEEC2EONS_11HuffmanCodeIS1_EE: 80| 2.05k| : Base(std::move(hc).operator rawspeed::PrefixCode()) {} _ZN8rawspeed23PrefixCodeLookupDecoderINS_15BaselineCodeTagEE5setupEbb: 97| 2.05k| void setup(bool fullDecode_, bool fixDNGBug16_) { 98| 2.05k| AbstractPrefixCodeDecoder::setup(fullDecode_, fixDNGBug16_); 99| | 100| | // Figure F.15: generate decoding tables 101| 2.05k| codeOffsetOL.resize(Base::maxCodeLength() + 1UL, MaxCodeValue); 102| 2.05k| maxCodeOL.resize(Base::maxCodeLength() + 1UL, MaxCodeValue); 103| 2.05k| for (unsigned int numCodesSoFar = 0, codeLen = 1; 104| 17.3k| codeLen <= Base::maxCodeLength(); codeLen++) { ------------------ | Branch (104:10): [True: 15.3k, False: 2.05k] ------------------ 105| 15.3k| if (!Base::code.nCodesPerLength[codeLen]) ------------------ | Branch (105:11): [True: 9.20k, False: 6.10k] ------------------ 106| 9.20k| continue; 107| 6.10k| codeOffsetOL[codeLen] = implicit_cast( 108| 6.10k| Base::code.symbols[numCodesSoFar].code - numCodesSoFar); 109| 6.10k| assert(codeOffsetOL[codeLen] != MaxCodeValue); ------------------ | Branch (109:7): [True: 6.10k, False: 0] ------------------ 110| 6.10k| numCodesSoFar += Base::code.nCodesPerLength[codeLen]; 111| 6.10k| maxCodeOL[codeLen] = Base::code.symbols[numCodesSoFar - 1].code; 112| 6.10k| } 113| 2.05k| } _ZNK8rawspeed23PrefixCodeLookupDecoderINS_15BaselineCodeTagEE26finishReadingPartialSymbolINS_15BitStreamerJPEGEEENSt3__14pairINS_18AbstractPrefixCodeIS1_E10CodeSymbolEiEERT_S9_: 136| 5.24k| typename Base::CodeSymbol partial) const { 137| 5.24k| static_assert( 138| 5.24k| BitStreamerTraits::canUseWithPrefixCodeDecoder, 139| 5.24k| "This BitStreamer specialization is not marked as usable here"); 140| 14.0k| while (partial.code_len < Base::maxCodeLength() && ------------------ | Branch (140:12): [True: 12.7k, False: 1.29k] ------------------ 141| 12.7k| (MaxCodeValue == maxCodeOL[partial.code_len] || ------------------ | Branch (141:13): [True: 8.58k, False: 4.17k] ------------------ 142| 8.80k| partial.code > maxCodeOL[partial.code_len])) { ------------------ | Branch (142:13): [True: 225, False: 3.95k] ------------------ 143| 8.80k| uint32_t temp = bs.getBitsNoFill(1); 144| 8.80k| partial.code = 145| 8.80k| implicit_cast((partial.code << 1) | temp); 146| 8.80k| partial.code_len++; 147| 8.80k| } 148| | 149| | // NOTE: when we are called from PrefixCodeLUTDecoder, the partial.code_len 150| | // *could* be larger than the largest code lenght for this huffman table, 151| | // which is a symptom of a corrupt code. 152| 5.24k| if (partial.code_len > Base::maxCodeLength() || ------------------ | Branch (152:9): [True: 145, False: 5.10k] ------------------ 153| 5.10k| partial.code > maxCodeOL[partial.code_len]) ------------------ | Branch (153:9): [True: 137, False: 4.96k] ------------------ 154| 5.24k| ThrowRDE("bad Huffman code: %u (len: %u)", partial.code, ------------------ | | 38| 282| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 282| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 282| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 282| __VA_ARGS__) | | ------------------ ------------------ 155| 4.96k| partial.code_len); 156| | 157| 5.24k| assert(MaxCodeValue != codeOffsetOL[partial.code_len]); ------------------ | Branch (157:5): [True: 4.96k, False: 0] ------------------ 158| 4.96k| assert(partial.code >= codeOffsetOL[partial.code_len]); ------------------ | Branch (158:5): [True: 4.96k, False: 0] ------------------ 159| 4.96k| unsigned codeIndex = partial.code - codeOffsetOL[partial.code_len]; 160| 4.96k| assert(codeIndex < Base::code.codeValues.size()); ------------------ | Branch (160:5): [True: 4.96k, False: 0] ------------------ 161| | 162| 4.96k| typename Traits::CodeValueTy codeValue = Base::code.codeValues[codeIndex]; 163| 4.96k| return {partial, codeValue}; 164| 4.96k| } _ZNK8rawspeed23PrefixCodeLookupDecoderINS_15BaselineCodeTagEE26finishReadingPartialSymbolINS_14BitStreamerMSBEEENSt3__14pairINS_18AbstractPrefixCodeIS1_E10CodeSymbolEiEERT_S9_: 136| 108| typename Base::CodeSymbol partial) const { 137| 108| static_assert( 138| 108| BitStreamerTraits::canUseWithPrefixCodeDecoder, 139| 108| "This BitStreamer specialization is not marked as usable here"); 140| 214| while (partial.code_len < Base::maxCodeLength() && ------------------ | Branch (140:12): [True: 106, False: 108] ------------------ 141| 106| (MaxCodeValue == maxCodeOL[partial.code_len] || ------------------ | Branch (141:13): [True: 92, False: 14] ------------------ 142| 106| partial.code > maxCodeOL[partial.code_len])) { ------------------ | Branch (142:13): [True: 14, False: 0] ------------------ 143| 106| uint32_t temp = bs.getBitsNoFill(1); 144| 106| partial.code = 145| 106| implicit_cast((partial.code << 1) | temp); 146| 106| partial.code_len++; 147| 106| } 148| | 149| | // NOTE: when we are called from PrefixCodeLUTDecoder, the partial.code_len 150| | // *could* be larger than the largest code lenght for this huffman table, 151| | // which is a symptom of a corrupt code. 152| 108| if (partial.code_len > Base::maxCodeLength() || ------------------ | Branch (152:9): [True: 2, False: 106] ------------------ 153| 106| partial.code > maxCodeOL[partial.code_len]) ------------------ | Branch (153:9): [True: 12, False: 94] ------------------ 154| 108| ThrowRDE("bad Huffman code: %u (len: %u)", partial.code, ------------------ | | 38| 14| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 14| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 14| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 14| __VA_ARGS__) | | ------------------ ------------------ 155| 94| partial.code_len); 156| | 157| 108| assert(MaxCodeValue != codeOffsetOL[partial.code_len]); ------------------ | Branch (157:5): [True: 94, False: 0] ------------------ 158| 94| assert(partial.code >= codeOffsetOL[partial.code_len]); ------------------ | Branch (158:5): [True: 94, False: 0] ------------------ 159| 94| unsigned codeIndex = partial.code - codeOffsetOL[partial.code_len]; 160| 94| assert(codeIndex < Base::code.codeValues.size()); ------------------ | Branch (160:5): [True: 94, False: 0] ------------------ 161| | 162| 94| typename Traits::CodeValueTy codeValue = Base::code.codeValues[codeIndex]; 163| 94| return {partial, codeValue}; 164| 94| } _ZNK8rawspeed23PrefixCodeLookupDecoderINS_15BaselineCodeTagEE26finishReadingPartialSymbolINS_16BitStreamerMSB32EEENSt3__14pairINS_18AbstractPrefixCodeIS1_E10CodeSymbolEiEERT_S9_: 136| 666| typename Base::CodeSymbol partial) const { 137| 666| static_assert( 138| 666| BitStreamerTraits::canUseWithPrefixCodeDecoder, 139| 666| "This BitStreamer specialization is not marked as usable here"); 140| 1.79k| while (partial.code_len < Base::maxCodeLength() && ------------------ | Branch (140:12): [True: 1.25k, False: 539] ------------------ 141| 1.25k| (MaxCodeValue == maxCodeOL[partial.code_len] || ------------------ | Branch (141:13): [True: 988, False: 264] ------------------ 142| 1.12k| partial.code > maxCodeOL[partial.code_len])) { ------------------ | Branch (142:13): [True: 137, False: 127] ------------------ 143| 1.12k| uint32_t temp = bs.getBitsNoFill(1); 144| 1.12k| partial.code = 145| 1.12k| implicit_cast((partial.code << 1) | temp); 146| 1.12k| partial.code_len++; 147| 1.12k| } 148| | 149| | // NOTE: when we are called from PrefixCodeLUTDecoder, the partial.code_len 150| | // *could* be larger than the largest code lenght for this huffman table, 151| | // which is a symptom of a corrupt code. 152| 666| if (partial.code_len > Base::maxCodeLength() || ------------------ | Branch (152:9): [True: 4, False: 662] ------------------ 153| 662| partial.code > maxCodeOL[partial.code_len]) ------------------ | Branch (153:9): [True: 82, False: 580] ------------------ 154| 666| ThrowRDE("bad Huffman code: %u (len: %u)", partial.code, ------------------ | | 38| 86| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 86| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 86| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 86| __VA_ARGS__) | | ------------------ ------------------ 155| 580| partial.code_len); 156| | 157| 666| assert(MaxCodeValue != codeOffsetOL[partial.code_len]); ------------------ | Branch (157:5): [True: 580, False: 0] ------------------ 158| 580| assert(partial.code >= codeOffsetOL[partial.code_len]); ------------------ | Branch (158:5): [True: 580, False: 0] ------------------ 159| 580| unsigned codeIndex = partial.code - codeOffsetOL[partial.code_len]; 160| 580| assert(codeIndex < Base::code.codeValues.size()); ------------------ | Branch (160:5): [True: 580, False: 0] ------------------ 161| | 162| 580| typename Traits::CodeValueTy codeValue = Base::code.codeValues[codeIndex]; 163| 580| return {partial, codeValue}; 164| 580| } _ZN8rawspeed22getTranslationalOffsetENS_10BayerPhaseES0_: 67| 1.16k|inline iPoint2D getTranslationalOffset(BayerPhase src, BayerPhase tgt) { 68| 1.16k| auto getCanonicalPosition = [](BayerPhase p) -> iPoint2D { 69| 1.16k| auto i = static_cast(p); 70| 1.16k| return {(i & 0b01) != 0, (i & 0b10) != 0}; 71| 1.16k| }; 72| | 73| 1.16k| iPoint2D off = getCanonicalPosition(tgt) - getCanonicalPosition(src); 74| 1.16k| return {std::abs(off.x), std::abs(off.y)}; 75| 1.16k|} _ZN8rawspeed14getAsCFAColorsENS_10BayerPhaseE: 97| 1.16k|inline std::array getAsCFAColors(BayerPhase p) { 98| 1.16k| const BayerPhase basePhase = BayerPhase::RGGB; 99| 1.16k| using enum CFAColor; 100| 1.16k| const std::array basePat = {RED, GREEN, GREEN, BLUE}; 101| 1.16k| return applyPhaseShift(basePat, basePhase, /*tgtPhase=*/p); 102| 1.16k|} _ZN8rawspeed15getAsBayerPhaseERKNS_16ColorFilterArrayE: 131| 483|inline Optional getAsBayerPhase(const ColorFilterArray& CFA) { 132| 483| if (CFA.getSize() != iPoint2D(2, 2)) ------------------ | Branch (132:7): [True: 23, False: 460] ------------------ 133| 23| return {}; 134| | 135| 460| std::array patData; 136| 460| const Array2DRef pat(patData.data(), 2, 2); 137| 1.38k| for (int row = 0; row < pat.height(); ++row) { ------------------ | Branch (137:21): [True: 927, False: 460] ------------------ 138| 2.77k| for (int col = 0; col < pat.width(); ++col) { ------------------ | Branch (138:23): [True: 1.84k, False: 927] ------------------ 139| 1.84k| pat(row, col) = CFA.getColorAt(col, row); 140| 1.84k| } 141| 927| } 142| | 143| 460| for (auto i = static_cast(BayerPhase::RGGB); 144| 1.24k| i <= static_cast(BayerPhase::BGGR); ++i) { ------------------ | Branch (144:8): [True: 1.16k, False: 73] ------------------ 145| 1.16k| if (auto p = static_cast(i); getAsCFAColors(p) == patData) ------------------ | Branch (145:46): [True: 387, False: 781] ------------------ 146| 387| return p; 147| 1.16k| } 148| | 149| 73| return {}; 150| 460|} _ZN8rawspeed15applyPhaseShiftINS_8CFAColorEEENSt3__15arrayIT_Lm4EEES5_NS_10BayerPhaseES6_: 81| 1.16k| BayerPhase tgtPhase) { 82| 1.16k| const iPoint2D coordOffset = getTranslationalOffset(srcPhase, tgtPhase); 83| 1.16k| assert(coordOffset >= iPoint2D(0, 0) && "Offset is non-negative."); ------------------ | Branch (83:3): [True: 1.15k, False: 2] | Branch (83:3): [True: 1.15k, Folded] | Branch (83:3): [True: 1.16k, False: 0] ------------------ 84| 1.16k| const Array2DRef src(srcData.data(), 2, 2); 85| | 86| 1.16k| std::array tgtData; 87| 1.16k| const Array2DRef tgt(tgtData.data(), 2, 2); 88| 3.48k| for (int row = 0; row < tgt.height(); ++row) { ------------------ | Branch (88:21): [True: 2.32k, False: 1.16k] ------------------ 89| 6.96k| for (int col = 0; col < tgt.width(); ++col) { ------------------ | Branch (89:23): [True: 4.64k, False: 2.32k] ------------------ 90| 4.64k| tgt(row, col) = src((coordOffset.y + row) % 2, (coordOffset.x + col) % 2); 91| 4.64k| } 92| 2.32k| } 93| | 94| 1.16k| return tgtData; 95| 1.16k|} _ZZN8rawspeed22getTranslationalOffsetENS_10BayerPhaseES0_ENKUlS0_E_clES0_: 68| 2.33k| auto getCanonicalPosition = [](BayerPhase p) -> iPoint2D { 69| 2.33k| auto i = static_cast(p); 70| 2.33k| return {(i & 0b01) != 0, (i & 0b10) != 0}; 71| 2.33k| }; _ZN8rawspeed8writeLogENS_10DEBUG_PRIOEPKcz: 31| 349k|void writeLog(DEBUG_PRIO priority, const char* format, ...) { 32| | // When fuzzing, any output is really undesirable. 33| 349k|} _ZN8rawspeed21getMisalignmentOffsetImQsr3stdE13is_integral_vIT_EEEmS1_m: 109| 8.05k| uint64_t multiple) { 110| 8.05k| if (multiple == 0) ------------------ | Branch (110:7): [True: 0, False: 8.05k] ------------------ 111| 0| return 0; 112| 8.05k| return value % multiple; 113| 8.05k|} _ZN8rawspeed15roundToMultipleImEET_S1_mb: 117| 8.05k| bool roundDown) { 118| 8.05k| uint64_t offset = getMisalignmentOffset(value, multiple); 119| 8.05k| if (offset == 0) ------------------ | Branch (119:7): [True: 3.15k, False: 4.90k] ------------------ 120| 3.15k| return value; 121| | // Drop remainder. 122| 4.90k| T roundedDown = value - offset; 123| 4.90k| if (roundDown) // If we were rounding down, then that's it. ------------------ | Branch (123:7): [True: 32, False: 4.87k] ------------------ 124| 32| return roundedDown; 125| | // Else, just add one multiple. 126| 4.87k| return roundedDown + multiple; 127| 4.90k|} _ZN8rawspeed14copyPixelsImplENS_10Array1DRefISt4byteEENS0_IKS1_EE: 56| 19.8k| Array1DRef src) { 57| 19.8k| invariant(src.size() == dest.size()); ------------------ | | 27| 19.8k|#define invariant(expr) assert(expr) ------------------ | Branch (57:3): [True: 19.8k, False: 18.4E] ------------------ 58| 19.8k| std::copy(src.begin(), src.end(), dest.begin()); 59| 19.8k|} _ZN8rawspeed14copyPixelsImplENS_10Array2DRefISt4byteEENS0_IKS1_EE: 62| 706| Array2DRef src) { 63| 706| invariant(src.width() > 0); ------------------ | | 27| 706|#define invariant(expr) assert(expr) ------------------ | Branch (63:3): [True: 707, False: 18.4E] ------------------ 64| 707| invariant(src.height() > 0); ------------------ | | 27| 707|#define invariant(expr) assert(expr) ------------------ | Branch (64:3): [True: 706, False: 1] ------------------ 65| 706| invariant(dest.width() > 0); ------------------ | | 27| 706|#define invariant(expr) assert(expr) ------------------ | Branch (65:3): [True: 706, False: 0] ------------------ 66| 706| invariant(dest.height() > 0); ------------------ | | 27| 706|#define invariant(expr) assert(expr) ------------------ | Branch (66:3): [True: 706, False: 0] ------------------ 67| 706| invariant(src.height() == dest.height()); ------------------ | | 27| 706|#define invariant(expr) assert(expr) ------------------ | Branch (67:3): [True: 706, False: 0] ------------------ 68| 706| invariant(src.width() == dest.width()); ------------------ | | 27| 706|#define invariant(expr) assert(expr) ------------------ | Branch (68:3): [True: 706, False: 0] ------------------ 69| 706| if (auto [destAsStrip, srcAsStrip] = 70| 706| std::make_tuple(dest.getAsArray1DRef(), src.getAsArray1DRef()); 71| 706| destAsStrip && srcAsStrip) { ------------------ | Branch (71:7): [True: 196, False: 510] | Branch (71:22): [True: 194, False: 2] ------------------ 72| 194| copyPixelsImpl(*destAsStrip, *srcAsStrip); 73| 194| return; 74| 194| } 75| 20.1k| for (int row = 0; row != src.height(); ++row) ------------------ | Branch (75:21): [True: 19.6k, False: 512] ------------------ 76| 19.6k| copyPixelsImpl(dest[row], src[row]); 77| 512|} _ZN8rawspeed10copyPixelsEPSt4byteiPKS0_iii: 81| 706| int height) { 82| 706| invariant(destPtr); ------------------ | | 27| 706|#define invariant(expr) assert(expr) ------------------ | Branch (82:3): [True: 706, False: 0] ------------------ 83| 706| invariant(dstPitch > 0); ------------------ | | 27| 706|#define invariant(expr) assert(expr) ------------------ | Branch (83:3): [True: 706, False: 0] ------------------ 84| 706| invariant(srcPtr); ------------------ | | 27| 706|#define invariant(expr) assert(expr) ------------------ | Branch (84:3): [True: 706, False: 0] ------------------ 85| 706| invariant(srcPitch > 0); ------------------ | | 27| 706|#define invariant(expr) assert(expr) ------------------ | Branch (85:3): [True: 706, False: 0] ------------------ 86| 706| invariant(rowSize > 0); ------------------ | | 27| 706|#define invariant(expr) assert(expr) ------------------ | Branch (86:3): [True: 706, False: 0] ------------------ 87| 706| invariant(height > 0); ------------------ | | 27| 706|#define invariant(expr) assert(expr) ------------------ | Branch (87:3): [True: 706, False: 0] ------------------ 88| 706| invariant(rowSize <= srcPitch); ------------------ | | 27| 706|#define invariant(expr) assert(expr) ------------------ | Branch (88:3): [True: 706, False: 0] ------------------ 89| 706| invariant(rowSize <= dstPitch); ------------------ | | 27| 706|#define invariant(expr) assert(expr) ------------------ | Branch (89:3): [True: 706, False: 0] ------------------ 90| 706| auto dest = Array2DRef(destPtr, rowSize, height, dstPitch); 91| 706| auto src = Array2DRef(srcPtr, rowSize, height, srcPitch); 92| 706| copyPixelsImpl(dest, src); 93| 706|} _ZN8rawspeed9roundDownEmm: 130| 48| uint64_t multiple) { 131| 48| return roundToMultiple(value, multiple, /*roundDown=*/true); 132| 48|} _ZN8rawspeed7roundUpEmm: 135| 8.01k| uint64_t multiple) { 136| 8.01k| return roundToMultiple(value, multiple, /*roundDown=*/false); 137| 8.01k|} _ZN8rawspeed19roundUpDivisionSafeEmm: 146| 24.8k| uint64_t div) { 147| 18.4E| return (value != 0) ? (1 + ((value - 1) / div)) : 0; ------------------ | Branch (147:10): [True: 24.8k, False: 18.4E] ------------------ 148| 24.8k|} _ZN8rawspeed10trimSpacesENSt3__117basic_string_viewIcNS0_11char_traitsIcEEEE: 163| 123k|inline std::string trimSpaces(std::string_view str) { 164| | // Find the first character position after excluding leading blank spaces 165| 123k| size_t startpos = str.find_first_not_of(" \t"); 166| | 167| | // Find the first character position from reverse af 168| 123k| size_t endpos = str.find_last_not_of(" \t"); 169| | 170| | // if all spaces or empty return an empty string 171| 123k| if ((startpos == std::string::npos) || (endpos == std::string::npos)) ------------------ | Branch (171:7): [True: 34.0k, False: 88.9k] | Branch (171:42): [True: 0, False: 88.9k] ------------------ 172| 34.0k| return ""; 173| | 174| 88.9k| str = str.substr(startpos, endpos - startpos + 1); 175| 88.9k| return {str.begin(), str.end()}; 176| 123k|} _ZN8rawspeed21getMisalignmentOffsetIPhQsr3stdE12is_pointer_vIT_EEEmS2_m: 98| 15.8k| uint64_t multiple) { 99| 15.8k| if (multiple == 0) ------------------ | Branch (99:7): [True: 0, False: 15.8k] ------------------ 100| 0| return 0; 101| 15.8k| static_assert(bitwidth() >= bitwidth(), 102| 15.8k| "uintptr_t can not represent all pointer values?"); 103| 15.8k| return reinterpret_cast(value) % multiple; 104| 15.8k|} _ZN8rawspeed9isAlignedIPhEEbT_m: 151| 15.8k|constexpr RAWSPEED_READNONE bool isAligned(T value, size_t multiple) { 152| 15.8k| return (multiple == 0) || (getMisalignmentOffset(value, multiple) == 0); ------------------ | Branch (152:10): [True: 0, False: 15.8k] | Branch (152:29): [True: 15.8k, False: 0] ------------------ 153| 15.8k|} _ZN8rawspeed4isInINS_7TiffTagES1_EEbT_RKSt16initializer_listIT0_E: 157| 178k| const std::initializer_list& list) { 158| 178k| return std::any_of(list.begin(), list.end(), 159| 178k| [value](const T2& t) { return t == value; }); 160| 178k|} _ZZN8rawspeed4isInINS_7TiffTagES1_EEbT_RKSt16initializer_listIT0_EENKUlRKS1_E_clES9_: 159| 976k| [value](const T2& t) { return t == value; }); _ZN8rawspeed9isAlignedIiEEbT_m: 151| 7.93k|constexpr RAWSPEED_READNONE bool isAligned(T value, size_t multiple) { 152| 7.93k| return (multiple == 0) || (getMisalignmentOffset(value, multiple) == 0); ------------------ | Branch (152:10): [True: 0, False: 7.93k] | Branch (152:29): [True: 7.93k, False: 0] ------------------ 153| 7.93k|} _ZN8rawspeed21getMisalignmentOffsetIiQsr3stdE13is_integral_vIT_EEEmS1_m: 109| 7.93k| uint64_t multiple) { 110| 7.93k| if (multiple == 0) ------------------ | Branch (110:7): [True: 0, False: 7.93k] ------------------ 111| 0| return 0; 112| 7.93k| return value % multiple; 113| 7.93k|} _ZN8rawspeed9isAlignedIPSt4byteEEbT_m: 151| 5.66M|constexpr RAWSPEED_READNONE bool isAligned(T value, size_t multiple) { 152| 5.66M| return (multiple == 0) || (getMisalignmentOffset(value, multiple) == 0); ------------------ | Branch (152:10): [True: 0, False: 5.66M] | Branch (152:29): [True: 5.66M, False: 0] ------------------ 153| 5.66M|} _ZN8rawspeed21getMisalignmentOffsetIPSt4byteQsr3stdE12is_pointer_vIT_EEEmS3_m: 98| 5.66M| uint64_t multiple) { 99| 5.66M| if (multiple == 0) ------------------ | Branch (99:7): [True: 0, False: 5.66M] ------------------ 100| 0| return 0; 101| 5.66M| static_assert(bitwidth() >= bitwidth(), 102| 5.66M| "uintptr_t can not represent all pointer values?"); 103| 5.66M| return reinterpret_cast(value) % multiple; 104| 5.66M|} _ZN8rawspeed10DngOpcodesC2ERKNS_8RawImageENS_10ByteStreamE: 672| 1.02k|DngOpcodes::DngOpcodes(const RawImage& ri, ByteStream bs) { 673| | // DNG opcodes are always stored in big-endian byte order. 674| 1.02k| bs.setByteOrder(Endianness::big); 675| | 676| 1.02k| const auto opcode_count = bs.getU32(); 677| 1.02k| auto origPos = bs.getPosition(); 678| | 679| | // validate opcode count. we either have to do this, or we can't preallocate 680| 68.8k| for (auto i = 0U; i < opcode_count; i++) { ------------------ | Branch (680:21): [True: 67.8k, False: 1.02k] ------------------ 681| 67.8k| bs.skipBytes(4); // code 682| 67.8k| bs.skipBytes(4); // version 683| 67.8k| bs.skipBytes(4); // flags 684| 67.8k| const auto opcode_size = bs.getU32(); 685| 67.8k| bs.skipBytes(opcode_size); 686| 67.8k| } 687| | 688| 1.02k| bs.setPosition(origPos); 689| | 690| | // okay, we may indeed have that many opcodes in here. now let's reserve 691| 1.02k| opcodes.reserve(opcode_count); 692| | 693| 1.02k| iRectangle2D integrated_subimg = getImageCropAsRectangle(ri); 694| | 695| 2.29k| for (auto i = 0U; i < opcode_count; i++) { ------------------ | Branch (695:21): [True: 1.34k, False: 950] ------------------ 696| 1.34k| auto code = bs.getU32(); 697| 1.34k| bs.skipBytes(4); // ignore version 698| |#ifdef DEBUG 699| | bs.skipBytes(4); // ignore flags 700| |#else 701| 1.34k| auto flags = bs.getU32(); 702| 1.34k|#endif 703| 1.34k| const auto opcode_size = bs.getU32(); 704| 1.34k| ByteStream opcode_bs = bs.getStream(opcode_size); 705| | 706| 1.34k| const char* opName = nullptr; 707| 1.34k| constructor_t opConstructor = nullptr; 708| 1.34k| if (auto Op = Map(code)) ------------------ | Branch (708:14): [True: 1.32k, False: 20] ------------------ 709| 1.32k| std::tie(opName, opConstructor) = *Op; 710| 20| else 711| 1.34k| ThrowRDE("Unknown unhandled Opcode: %u", code); ------------------ | | 38| 20| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 20| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 20| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 20| __VA_ARGS__) | | ------------------ ------------------ 712| | 713| 1.32k| if (opConstructor != nullptr) ------------------ | Branch (713:9): [True: 1.27k, False: 50] ------------------ 714| 1.27k| opcodes.emplace_back(opConstructor(ri, opcode_bs, integrated_subimg)); 715| 50| else { 716| 50|#ifndef DEBUG 717| | // Throw Error if not marked as optional 718| 50| if (!(flags & 1)) ------------------ | Branch (718:11): [True: 4, False: 46] ------------------ 719| 4|#endif 720| 50| ThrowRDE("Unsupported Opcode: %u (%s)", code, opName); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 721| 50| } 722| | 723| 1.31k| if (opcode_bs.getRemainSize() != 0) ------------------ | Branch (723:9): [True: 49, False: 1.27k] ------------------ 724| 1.31k| ThrowRDE("Inconsistent length of opcode"); ------------------ | | 38| 49| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 49| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 49| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 49| __VA_ARGS__) | | ------------------ ------------------ 725| 1.31k| } 726| | 727| |#ifdef DEBUG 728| | assert(opcodes.size() == opcode_count); 729| |#endif 730| 1.02k|} DngOpcodes.cpp:_ZN8rawspeed12_GLOBAL__N_123getImageCropAsRectangleERKNS_8RawImageE: 68| 2.61k|iRectangle2D getImageCropAsRectangle(const RawImage& ri) { 69| 2.61k| iRectangle2D rect; 70| 2.61k| switch (ri->getDataType()) { ------------------ | Branch (70:11): [True: 2.61k, False: 0] ------------------ 71| 2.39k| case RawImageType::UINT16: ------------------ | Branch (71:3): [True: 2.39k, False: 217] ------------------ 72| 2.39k| rect = getImageCropAsRectangle(ri->getU16DataAsCroppedArray2DRef()); 73| 2.39k| break; 74| 217| case RawImageType::F32: ------------------ | Branch (74:3): [True: 217, False: 2.39k] ------------------ 75| 217| rect = getImageCropAsRectangle(ri->getF32DataAsCroppedArray2DRef()); 76| 217| break; 77| 2.61k| } 78| 5.22k| for (int* col : {&rect.pos.x, &rect.dim.x}) { ------------------ | Branch (78:17): [True: 5.22k, False: 2.61k] ------------------ 79| 5.22k| assert(*col % ri->getCpp() == 0 && "Column is width * cpp"); ------------------ | Branch (79:5): [True: 5.22k, False: 0] | Branch (79:5): [True: 5.22k, Folded] | Branch (79:5): [True: 5.22k, False: 0] ------------------ 80| 5.22k| *col /= ri->getCpp(); 81| 5.22k| } 82| 2.61k| return rect; 83| 2.61k|} DngOpcodes.cpp:_ZN8rawspeed12_GLOBAL__N_123getImageCropAsRectangleItEENS_12iRectangle2DENS_17CroppedArray2DRefIT_EE: 63| 2.39k|iRectangle2D getImageCropAsRectangle(CroppedArray2DRef img) { 64| 2.39k| return {{img.offsetCols, img.offsetRows}, 65| 2.39k| {img.croppedWidth, img.croppedHeight}}; 66| 2.39k|} DngOpcodes.cpp:_ZN8rawspeed12_GLOBAL__N_123getImageCropAsRectangleIfEENS_12iRectangle2DENS_17CroppedArray2DRefIT_EE: 63| 217|iRectangle2D getImageCropAsRectangle(CroppedArray2DRef img) { 64| 217| return {{img.offsetCols, img.offsetRows}, 65| 217| {img.croppedWidth, img.croppedHeight}}; 66| 217|} _ZN8rawspeed10DngOpcodesD2Ev: 734| 387|DngOpcodes::~DngOpcodes() = default; _ZNK8rawspeed10DngOpcodes12applyOpCodesERKNS_8RawImageE: 736| 387|void DngOpcodes::applyOpCodes(const RawImage& ri) const { 737| 567| for (const auto& code : opcodes) { ------------------ | Branch (737:25): [True: 567, False: 387] ------------------ 738| 567| code->setup(ri); 739| 567| code->apply(ri); 740| 567| } 741| 387|} _ZN8rawspeed10DngOpcodes3MapEj: 753| 1.34k|DngOpcodes::Map(uint32_t code) { 754| 1.34k| switch (code) { 755| 16| case 1U: ------------------ | Branch (755:3): [True: 16, False: 1.32k] ------------------ 756| 16| return make_pair("WarpRectilinear", nullptr); 757| 10| case 2U: ------------------ | Branch (757:3): [True: 10, False: 1.33k] ------------------ 758| 10| return make_pair("WarpFisheye", nullptr); 759| 12| case 3U: ------------------ | Branch (759:3): [True: 12, False: 1.33k] ------------------ 760| 12| return make_pair("FixVignetteRadial", nullptr); 761| 123| case 4U: ------------------ | Branch (761:3): [True: 123, False: 1.22k] ------------------ 762| 123| return make_pair( 763| 123| "FixBadPixelsConstant", 764| 123| &DngOpcodes::constructor); 765| 413| case 5U: ------------------ | Branch (765:3): [True: 413, False: 930] ------------------ 766| 413| return make_pair("FixBadPixelsList", 767| 413| &DngOpcodes::constructor); 768| 67| case 6U: ------------------ | Branch (768:3): [True: 67, False: 1.27k] ------------------ 769| 67| return make_pair("TrimBounds", 770| 67| &DngOpcodes::constructor); 771| 100| case 7U: ------------------ | Branch (771:3): [True: 100, False: 1.24k] ------------------ 772| 100| return make_pair("MapTable", 773| 100| &DngOpcodes::constructor); 774| 82| case 8U: ------------------ | Branch (774:3): [True: 82, False: 1.26k] ------------------ 775| 82| return make_pair("MapPolynomial", 776| 82| &DngOpcodes::constructor); 777| 11| case 9U: ------------------ | Branch (777:3): [True: 11, False: 1.33k] ------------------ 778| 11| return make_pair("GainMap", nullptr); 779| 130| case 10U: ------------------ | Branch (779:3): [True: 130, False: 1.21k] ------------------ 780| 130| return make_pair( 781| 130| "DeltaPerRow", 782| 130| &DngOpcodes::constructor< 783| 130| DngOpcodes::OffsetPerRowOrCol>); 784| 143| case 11U: ------------------ | Branch (784:3): [True: 143, False: 1.20k] ------------------ 785| 143| return make_pair( 786| 143| "DeltaPerColumn", 787| 143| &DngOpcodes::constructor< 788| 143| DngOpcodes::OffsetPerRowOrCol>); 789| 102| case 12U: ------------------ | Branch (789:3): [True: 102, False: 1.24k] ------------------ 790| 102| return make_pair( 791| 102| "ScalePerRow", 792| 102| &DngOpcodes::constructor< 793| 102| DngOpcodes::ScalePerRowOrCol>); 794| 113| case 13U: ------------------ | Branch (794:3): [True: 113, False: 1.23k] ------------------ 795| 113| return make_pair( 796| 113| "ScalePerColumn", 797| 113| &DngOpcodes::constructor< 798| 113| DngOpcodes::ScalePerRowOrCol>); 799| 1| case 14U: ------------------ | Branch (799:3): [True: 1, False: 1.34k] ------------------ 800| 1| return make_pair("WarpRectilinear2", nullptr); 801| 20| default: ------------------ | Branch (801:3): [True: 20, False: 1.32k] ------------------ 802| 20| return std::nullopt; 803| 1.34k| } 804| 1.34k|} _ZN8rawspeed10DngOpcodes11constructorINS0_20FixBadPixelsConstantEEENSt3__110unique_ptrINS0_9DngOpcodeENS3_14default_deleteIS5_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DE: 746| 123| iRectangle2D& integrated_subimg) { 747| 123| return std::make_unique(ri, bs, integrated_subimg); 748| 123|} _ZN8rawspeed10DngOpcodes20FixBadPixelsConstantC2ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DE: 157| 123| : DngOpcodes::DngOpcode(integrated_subimg_), value(bs.getU32()) { 158| 123| bs.getU32(); // Bayer Phase not used 159| 123| } _ZN8rawspeed10DngOpcodes9DngOpcodeC2ERKNS_12iRectangle2DE: 108| 2.39k| : integrated_subimg(integrated_subimg_) 109| |#endif 110| 2.39k| { 111| | assert(std::uncaught_exceptions() == 0 && ------------------ | Branch (111:5): [True: 2.39k, False: 0] | Branch (111:5): [True: 2.39k, Folded] | Branch (111:5): [True: 2.39k, False: 0] ------------------ 112| 2.39k| "Creating DngOpcode during call stack unwinding?"); 113| 2.39k| } _ZN8rawspeed10DngOpcodes9DngOpcodeD2Ev: 121| 2.39k| virtual ~DngOpcode() { 122| | assert((std::uncaught_exceptions() > 0 || setup_was_called) && ------------------ | Branch (122:5): [True: 806, False: 1.58k] | Branch (122:5): [True: 1.58k, False: 0] | Branch (122:5): [True: 2.39k, Folded] | Branch (122:5): [True: 2.39k, False: 0] ------------------ 123| 2.39k| "Derived classes did not call our setup()!"); 124| 2.39k| } _ZN8rawspeed10DngOpcodes11constructorINS0_16FixBadPixelsListEEENSt3__110unique_ptrINS0_9DngOpcodeENS3_14default_deleteIS5_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DE: 746| 413| iRectangle2D& integrated_subimg) { 747| 413| return std::make_unique(ri, bs, integrated_subimg); 748| 413|} _ZN8rawspeed10DngOpcodes16FixBadPixelsListC2ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DE: 271| 413| : DngOpcodes::DngOpcode(integrated_subimg_) { 272| | // Although it is not really obvious from the spec, 273| | // the coordinates appear to be global/crop-independent, 274| | // and apply to the source uncropped image. 275| 413| const iRectangle2D fullImage({0, 0}, ri->getUncroppedDim()); 276| | 277| 413| bs.getU32(); // Skip phase - we don't care 278| 413| auto badPointCount = bs.getU32(); 279| 413| auto badRectCount = bs.getU32(); 280| | 281| | // first, check that we indeed have much enough data 282| 413| const auto origPos = bs.getPosition(); 283| 413| bs.skipBytes(badPointCount, 2 * 4); 284| 413| bs.skipBytes(badRectCount, 4 * 4); 285| 413| bs.setPosition(origPos); 286| | 287| | // Read points 288| 413| badPixels.reserve(badPixels.size() + badPointCount); 289| 1.55k| for (auto i = 0U; i < badPointCount; ++i) { ------------------ | Branch (289:23): [True: 1.21k, False: 342] ------------------ 290| 1.21k| auto y = bs.getU32(); 291| 1.21k| auto x = bs.getU32(); 292| | 293| 1.21k| if (const iPoint2D badPoint(x, y); !fullImage.isPointInside(badPoint)) ------------------ | Branch (293:42): [True: 71, False: 1.14k] ------------------ 294| 1.21k| ThrowRDE("Bad point not inside image."); ------------------ | | 38| 71| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 71| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 71| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 71| __VA_ARGS__) | | ------------------ ------------------ 295| | 296| 1.14k| badPixels.emplace_back(y << 16 | x); 297| 1.14k| } 298| | 299| | // Read rects 300| 1.40k| for (auto i = 0U; i < badRectCount; ++i) { ------------------ | Branch (300:23): [True: 1.11k, False: 285] ------------------ 301| 1.11k| iRectangle2D fullImage_ = fullImage; 302| 1.11k| const DummyROIOpcode dummy(ri, bs, fullImage_); 303| | 304| 1.11k| const iRectangle2D badRect = dummy.getRoi(); 305| 1.11k| assert(badRect.isThisInside(fullImage)); ------------------ | Branch (305:7): [True: 1.06k, False: 57] ------------------ 306| | 307| 1.06k| auto area = badRect.getHeight() * badRect.getWidth(); 308| 1.06k| badPixels.reserve(badPixels.size() + area); 309| 14.4k| for (auto y = 0; y < badRect.getHeight(); ++y) { ------------------ | Branch (309:24): [True: 13.4k, False: 1.06k] ------------------ 310| 998k| for (auto x = 0; x < badRect.getWidth(); ++x) { ------------------ | Branch (310:26): [True: 984k, False: 13.4k] ------------------ 311| 984k| badPixels.emplace_back((badRect.getTop() + y) << 16 | 312| 984k| (badRect.getLeft() + x)); 313| 984k| } 314| 13.4k| } 315| 1.06k| } 316| 342| } _ZN8rawspeed10DngOpcodes14DummyROIOpcodeC2ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DE: 244| 1.11k| : ROIOpcode(ri, bs, integrated_subimg_) { 245| 1.11k| DummyROIOpcode::setup(ri); 246| 1.11k| } _ZN8rawspeed10DngOpcodes9ROIOpcodeC2ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DE: 201| 1.85k| : DngOpcodes::DngOpcode(integrated_subimg_) { 202| 1.85k| const iRectangle2D subImage = {{0, 0}, integrated_subimg_.dim}; 203| | 204| 1.85k| uint32_t top = bs.getU32(); 205| 1.85k| uint32_t left = bs.getU32(); 206| 1.85k| uint32_t bottom = bs.getU32(); 207| 1.85k| uint32_t right = bs.getU32(); 208| | 209| 1.85k| const iPoint2D topLeft(left, top); 210| 1.85k| const iPoint2D bottomRight(right, bottom); 211| | 212| 1.85k| if (!(subImage.isPointInsideInclusive(topLeft) && ------------------ | Branch (212:11): [True: 1.70k, False: 150] ------------------ 213| 1.70k| subImage.isPointInsideInclusive(bottomRight) && ------------------ | Branch (213:11): [True: 1.61k, False: 92] ------------------ 214| 1.61k| bottomRight >= topLeft)) { ------------------ | Branch (214:11): [True: 1.61k, False: 1] ------------------ 215| 218| ThrowRDE("Rectangle (%d, %d, %d, %d) not inside image (%d, %d, %d, %d).", ------------------ | | 38| 218| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 218| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 218| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 218| __VA_ARGS__) | | ------------------ ------------------ 216| 218| topLeft.x, topLeft.y, bottomRight.x, bottomRight.y, 217| 218| subImage.getTopLeft().x, subImage.getTopLeft().y, 218| 218| subImage.getBottomRight().x, subImage.getBottomRight().y); 219| 218| } 220| | 221| 1.63k| roi.setTopLeft(topLeft); 222| 1.63k| roi.setBottomRightAbsolute(bottomRight); 223| | assert(roi.isThisInside(subImage)); ------------------ | Branch (223:5): [True: 1.61k, False: 25] ------------------ 224| 1.63k| } _ZNK8rawspeed10DngOpcodes9ROIOpcode6getRoiEv: 226| 2.64k| [[nodiscard]] const iRectangle2D& RAWSPEED_READONLY getRoi() const { 227| 2.64k| return roi; 228| 2.64k| } _ZN8rawspeed10DngOpcodes11constructorINS0_10TrimBoundsEEENSt3__110unique_ptrINS0_9DngOpcodeENS3_14default_deleteIS5_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DE: 746| 67| iRectangle2D& integrated_subimg) { 747| 67| return std::make_unique(ri, bs, integrated_subimg); 748| 67|} _ZN8rawspeed10DngOpcodes10TrimBoundsC2ERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DE: 338| 67| : ROIOpcode(ri, bs, integrated_subimg_) { 339| 67| integrated_subimg_.pos += getRoi().pos; 340| 67| integrated_subimg_.dim = getRoi().dim; 341| 67| } _ZN8rawspeed10DngOpcodes11constructorINS0_8TableMapEEENSt3__110unique_ptrINS0_9DngOpcodeENS3_14default_deleteIS5_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DE: 746| 100| iRectangle2D& integrated_subimg) { 747| 100| return std::make_unique(ri, bs, integrated_subimg); 748| 100|} _ZN8rawspeed10DngOpcodes8TableMapC2ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DE: 453| 100| : LookupOpcode(ri, bs, integrated_subimg_) { 454| 100| auto count = bs.getU32(); 455| | 456| 100| if (count == 0 || count > 65536) ------------------ | Branch (456:9): [True: 35, False: 65] | Branch (456:23): [True: 13, False: 52] ------------------ 457| 100| ThrowRDE("Invalid size of lookup table"); ------------------ | | 38| 14| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 14| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 14| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 14| __VA_ARGS__) | | ------------------ ------------------ 458| | 459| 210k| for (auto i = 0U; i < count; ++i) ------------------ | Branch (459:23): [True: 210k, False: 86] ------------------ 460| 210k| lookup[i] = bs.getU16(); 461| | 462| 86| if (count < lookup.size()) ------------------ | Branch (462:9): [True: 46, False: 40] ------------------ 463| 46| fill_n(&lookup[count], lookup.size() - count, lookup[count - 1]); 464| 86| } _ZN8rawspeed10DngOpcodes11PixelOpcodeC2ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DE: 364| 670| : ROIOpcode(ri, bs, integrated_subimg_), firstPlane(bs.getU32()), 365| 670| planes(bs.getU32()) { 366| | 367| 670| if (planes == 0 || firstPlane > ri->getCpp() || planes > ri->getCpp() || ------------------ | Branch (367:9): [True: 153, False: 517] | Branch (367:24): [True: 34, False: 483] | Branch (367:53): [True: 29, False: 454] ------------------ 368| 454| firstPlane + planes > ri->getCpp()) { ------------------ | Branch (368:9): [True: 1, False: 453] ------------------ 369| 65| ThrowRDE("Bad plane params (first %u, num %u), got planes = %u", ------------------ | | 38| 65| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 65| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 65| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 65| __VA_ARGS__) | | ------------------ ------------------ 370| 65| firstPlane, planes, ri->getCpp()); 371| 65| } 372| | 373| 605| rowPitch = bs.getU32(); 374| 605| colPitch = bs.getU32(); 375| | 376| 605| const iRectangle2D& ROI = getRoi(); 377| | 378| 605| if (rowPitch < 1 || rowPitch > static_cast(ROI.getHeight()) || ------------------ | Branch (378:9): [True: 154, False: 451] | Branch (378:25): [True: 10, False: 441] ------------------ 379| 441| colPitch < 1 || colPitch > static_cast(ROI.getWidth())) ------------------ | Branch (379:9): [True: 3, False: 438] | Branch (379:25): [True: 35, False: 403] ------------------ 380| 605| ThrowRDE("Invalid pitch"); ------------------ | | 38| 49| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 49| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 49| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 49| __VA_ARGS__) | | ------------------ ------------------ 381| 605| } _ZN8rawspeed10DngOpcodes11constructorINS0_13PolynomialMapEEENSt3__110unique_ptrINS0_9DngOpcodeENS3_14default_deleteIS5_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DE: 746| 82| iRectangle2D& integrated_subimg) { 747| 82| return std::make_unique(ri, bs, integrated_subimg); 748| 82|} _ZN8rawspeed10DngOpcodes13PolynomialMapC2ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DE: 480| 82| : LookupOpcode(ri, bs, integrated_subimg_) { 481| 82| vector polynomial; 482| | 483| 82| const auto polynomial_size = bs.getU32() + 1UL; 484| 82| (void)bs.check(implicit_cast(8UL * polynomial_size)); 485| 82| if (polynomial_size > 9) ------------------ | Branch (485:9): [True: 2, False: 80] ------------------ 486| 82| ThrowRDE("A polynomial with more than 8 degrees not allowed"); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 487| | 488| 80| polynomial.reserve(polynomial_size); 489| 80| std::generate_n(std::back_inserter(polynomial), polynomial_size, 490| 80| [&bs]() { return bs.get(); }); 491| | 492| | // Create lookup 493| 80| lookup.resize(65536); 494| 2.75M| for (auto i = 0UL; i < lookup.size(); ++i) { ------------------ | Branch (494:24): [True: 2.75M, False: 80] ------------------ 495| 2.75M| double val = polynomial[0]; 496| 5.37M| for (auto j = 1UL; j < polynomial.size(); ++j) ------------------ | Branch (496:26): [True: 2.62M, False: 2.75M] ------------------ 497| 2.62M| val += polynomial[j] * pow(implicit_cast(i) / 65536.0, 498| 2.62M| implicit_cast(j)); 499| 2.75M| lookup[i] = implicit_cast(std::clamp( 500| 2.75M| val * 65535.5, std::numeric_limits::min(), 501| 2.75M| std::numeric_limits::max())); 502| 2.75M| } 503| 80| } _ZZN8rawspeed10DngOpcodes13PolynomialMapC1ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DEENKUlvE_clEv: 490| 82| [&bs]() { return bs.get(); }); _ZN8rawspeed10DngOpcodes11constructorINS0_17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectYEEEEENSt3__110unique_ptrINS0_9DngOpcodeENS6_14default_deleteIS8_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DE: 746| 130| iRectangle2D& integrated_subimg) { 747| 130| return std::make_unique(ri, bs, integrated_subimg); 748| 130|} _ZN8rawspeed10DngOpcodes17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectYEEC2ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DE: 612| 130| : DeltaRowOrCol(ri, bs, integrated_subimg_, 65535.0F), 613| 130| absLimit(double(std::numeric_limits::max()) / 614| 130| implicit_cast(this->f2iScale)) {} _ZN8rawspeed10DngOpcodes13DeltaRowOrColINS0_17DeltaRowOrColBase7SelectYEEC2ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DEf: 565| 232| : DeltaRowOrColBase(ri, bs, integrated_subimg_), f2iScale(f2iScale_) { 566| 232| const auto deltaF_count = bs.getU32(); 567| 232| (void)bs.check(deltaF_count, 4); 568| | 569| | // See PixelOpcode::applyOP(). We will access deltaF/deltaI up to (excl.) 570| | // either ROI.getWidth() or ROI.getHeight() index. Thus, we need to have 571| | // either ROI.getRight() or ROI.getBottom() elements in there. 572| 232| if (const auto expectedSize = roundUpDivisionSafe( 573| 232| S::select(getRoi().getWidth(), getRoi().getHeight()), 574| 232| S::select(getPitch().x, getPitch().y)); 575| 232| expectedSize != deltaF_count) { ------------------ | Branch (575:9): [True: 4, False: 228] ------------------ 576| 4| ThrowRDE("Got unexpected number of elements (%" PRIu64 "), expected %u.", ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 577| 4| expectedSize, deltaF_count); 578| 4| } 579| | 580| 228| deltaF.reserve(deltaF_count); 581| 228| std::generate_n(std::back_inserter(deltaF), deltaF_count, [&bs]() { 582| 228| const auto F = bs.get(); 583| 228| switch (std::fpclassify(F)) { 584| 228| case FP_NORMAL: 585| 228| case FP_SUBNORMAL: 586| 228| case FP_ZERO: 587| 228| return F; 588| 228| default: 589| 228| ThrowRDE("Got bad float %f.", implicit_cast(F)); 590| 228| } 591| 228| }); 592| 228| } _ZN8rawspeed10DngOpcodes17DeltaRowOrColBaseC2ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DE: 528| 488| : PixelOpcode(ri, bs, integrated_subimg_) {} _ZN8rawspeed10DngOpcodes17DeltaRowOrColBase7SelectY6selectEjj: 522| 7.89k| static uint32_t select(uint32_t /*x*/, uint32_t y) { return y; } _ZNK8rawspeed10DngOpcodes11PixelOpcode8getPitchEv: 383| 275| [[nodiscard]] iPoint2D RAWSPEED_READONLY getPitch() const { 384| 275| return {static_cast(colPitch), static_cast(rowPitch)}; 385| 275| } _ZZN8rawspeed10DngOpcodes13DeltaRowOrColINS0_17DeltaRowOrColBase7SelectYEEC1ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DEfENKUlvE_clEv: 581| 195| std::generate_n(std::back_inserter(deltaF), deltaF_count, [&bs]() { 582| 195| const auto F = bs.get(); 583| 195| switch (std::fpclassify(F)) { 584| 150| case FP_NORMAL: ------------------ | Branch (584:7): [True: 150, False: 45] ------------------ 585| 150| case FP_SUBNORMAL: ------------------ | Branch (585:7): [True: 0, False: 195] ------------------ 586| 195| case FP_ZERO: ------------------ | Branch (586:7): [True: 45, False: 150] ------------------ 587| 195| return F; 588| 0| default: ------------------ | Branch (588:7): [True: 0, False: 195] ------------------ 589| 0| ThrowRDE("Got bad float %f.", implicit_cast(F)); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 590| 195| } 591| 195| }); _ZN8rawspeed10DngOpcodes13DeltaRowOrColINS0_17DeltaRowOrColBase7SelectYEE5setupERKNS_8RawImageE: 539| 107| void setup(const RawImage& ri) final { 540| 107| PixelOpcode::setup(ri); 541| | 542| | // If we are working on a float image, no need to convert to int 543| 107| if (ri->getDataType() != RawImageType::UINT16) ------------------ | Branch (543:9): [True: 35, False: 72] ------------------ 544| 35| return; 545| | 546| 72| deltaI.reserve(deltaF.size()); 547| 116| for (const auto f : deltaF) { ------------------ | Branch (547:23): [True: 116, False: 66] ------------------ 548| 116| if (!valueIsOk(f)) ------------------ | Branch (548:11): [True: 6, False: 110] ------------------ 549| 116| ThrowRDE("Got float %f which is unacceptable.", ------------------ | | 38| 6| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 6| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 6| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 6| __VA_ARGS__) | | ------------------ ------------------ 550| 110| implicit_cast(f)); 551| 110| deltaI.emplace_back(static_cast(f2iScale * f)); 552| 110| } 553| 72| } _ZN8rawspeed10DngOpcodes17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectYEE5applyERKNS_8RawImageE: 616| 47| void apply(const RawImage& ri) override { 617| 47| if (ri->getDataType() == RawImageType::UINT16) { ------------------ | Branch (617:9): [True: 33, False: 14] ------------------ 618| 33| this->template applyOP( 619| 33| ri, [this](uint32_t x, uint32_t y, uint16_t v) { 620| 33| return clampBits(this->deltaI[S::select(x, y)] + v, 16); 621| 33| }); 622| 33| } else { 623| 14| this->template applyOP(ri, 624| 14| [this](uint32_t x, uint32_t y, float v) { 625| 14| return this->deltaF[S::select(x, y)] + v; 626| 14| }); 627| 14| } 628| 47| } _ZNK8rawspeed10DngOpcodes11PixelOpcode7applyOPItZNS0_17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectYEE5applyERKNS_8RawImageEEUljjtE_EEvS9_T0_: 391| 33| void applyOP(const RawImage& ri, OP op) const { 392| 33| const CroppedArray2DRef img = getDataAsCroppedArray2DRef(ri); 393| 33| int cpp = ri->getCpp(); 394| 33| const iRectangle2D& ROI = getRoi(); 395| 33| const iPoint2D numAffected( 396| 33| implicit_cast(roundUpDivisionSafe(getRoi().dim.x, colPitch)), 397| 33| implicit_cast(roundUpDivisionSafe(getRoi().dim.y, rowPitch))); 398| 85| for (int y = 0; y < numAffected.y; ++y) { ------------------ | Branch (398:21): [True: 52, False: 33] ------------------ 399| 839| for (int x = 0; x < numAffected.x; ++x) { ------------------ | Branch (399:23): [True: 787, False: 52] ------------------ 400| 2.18k| for (auto p = 0U; p < planes; ++p) { ------------------ | Branch (400:27): [True: 1.39k, False: 787] ------------------ 401| 1.39k| T& pixel = 402| 1.39k| img(ROI.getTop() + (rowPitch * y), 403| 1.39k| firstPlane + ((ROI.getLeft() + colPitch * x) * cpp) + p); 404| 1.39k| pixel = op(x, y, pixel); 405| 1.39k| } 406| 787| } 407| 52| } 408| 33| } DngOpcodes.cpp:_ZN8rawspeed12_GLOBAL__N_126getDataAsCroppedArray2DRefItEENS_17CroppedArray2DRefIT_EERKNS_8RawImageE: 87| 179|CroppedArray2DRef getDataAsCroppedArray2DRef(const RawImage& ri) { 88| | if constexpr (std::is_same()) 89| 179| return ri->getU16DataAsCroppedArray2DRef(); 90| | if constexpr (std::is_same()) 91| | return ri->getF32DataAsCroppedArray2DRef(); 92| 179| __builtin_unreachable(); 93| 179|} _ZZN8rawspeed10DngOpcodes17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectYEE5applyERKNS_8RawImageEENKUljjtE_clEjjt: 619| 1.39k| ri, [this](uint32_t x, uint32_t y, uint16_t v) { 620| 1.39k| return clampBits(this->deltaI[S::select(x, y)] + v, 16); 621| 1.39k| }); _ZNK8rawspeed10DngOpcodes11PixelOpcode7applyOPIfZNS0_17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectYEE5applyERKNS_8RawImageEEUljjfE_EEvS9_T0_: 391| 14| void applyOP(const RawImage& ri, OP op) const { 392| 14| const CroppedArray2DRef img = getDataAsCroppedArray2DRef(ri); 393| 14| int cpp = ri->getCpp(); 394| 14| const iRectangle2D& ROI = getRoi(); 395| 14| const iPoint2D numAffected( 396| 14| implicit_cast(roundUpDivisionSafe(getRoi().dim.x, colPitch)), 397| 14| implicit_cast(roundUpDivisionSafe(getRoi().dim.y, rowPitch))); 398| 28| for (int y = 0; y < numAffected.y; ++y) { ------------------ | Branch (398:21): [True: 14, False: 14] ------------------ 399| 483| for (int x = 0; x < numAffected.x; ++x) { ------------------ | Branch (399:23): [True: 469, False: 14] ------------------ 400| 1.15k| for (auto p = 0U; p < planes; ++p) { ------------------ | Branch (400:27): [True: 687, False: 469] ------------------ 401| 687| T& pixel = 402| 687| img(ROI.getTop() + (rowPitch * y), 403| 687| firstPlane + ((ROI.getLeft() + colPitch * x) * cpp) + p); 404| 687| pixel = op(x, y, pixel); 405| 687| } 406| 469| } 407| 14| } 408| 14| } DngOpcodes.cpp:_ZN8rawspeed12_GLOBAL__N_126getDataAsCroppedArray2DRefIfEENS_17CroppedArray2DRefIT_EERKNS_8RawImageE: 87| 68|CroppedArray2DRef getDataAsCroppedArray2DRef(const RawImage& ri) { 88| | if constexpr (std::is_same()) 89| | return ri->getU16DataAsCroppedArray2DRef(); 90| | if constexpr (std::is_same()) 91| 68| return ri->getF32DataAsCroppedArray2DRef(); 92| 0| __builtin_unreachable(); 93| 68|} _ZZN8rawspeed10DngOpcodes17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectYEE5applyERKNS_8RawImageEENKUljjfE_clEjjf: 624| 687| [this](uint32_t x, uint32_t y, float v) { 625| 687| return this->deltaF[S::select(x, y)] + v; 626| 687| }); _ZN8rawspeed10DngOpcodes17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectYEE9valueIsOkEf: 605| 53| bool valueIsOk(float value) override { 606| 53| return implicit_cast(std::abs(value)) <= absLimit; 607| 53| } _ZN8rawspeed10DngOpcodes11constructorINS0_17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectXEEEEENSt3__110unique_ptrINS0_9DngOpcodeENS6_14default_deleteIS8_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DE: 746| 143| iRectangle2D& integrated_subimg) { 747| 143| return std::make_unique(ri, bs, integrated_subimg); 748| 143|} _ZN8rawspeed10DngOpcodes17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectXEEC2ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DE: 612| 143| : DeltaRowOrCol(ri, bs, integrated_subimg_, 65535.0F), 613| 143| absLimit(double(std::numeric_limits::max()) / 614| 143| implicit_cast(this->f2iScale)) {} _ZN8rawspeed10DngOpcodes13DeltaRowOrColINS0_17DeltaRowOrColBase7SelectXEEC2ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DEf: 565| 256| : DeltaRowOrColBase(ri, bs, integrated_subimg_), f2iScale(f2iScale_) { 566| 256| const auto deltaF_count = bs.getU32(); 567| 256| (void)bs.check(deltaF_count, 4); 568| | 569| | // See PixelOpcode::applyOP(). We will access deltaF/deltaI up to (excl.) 570| | // either ROI.getWidth() or ROI.getHeight() index. Thus, we need to have 571| | // either ROI.getRight() or ROI.getBottom() elements in there. 572| 256| if (const auto expectedSize = roundUpDivisionSafe( 573| 256| S::select(getRoi().getWidth(), getRoi().getHeight()), 574| 256| S::select(getPitch().x, getPitch().y)); 575| 256| expectedSize != deltaF_count) { ------------------ | Branch (575:9): [True: 4, False: 252] ------------------ 576| 4| ThrowRDE("Got unexpected number of elements (%" PRIu64 "), expected %u.", ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 577| 4| expectedSize, deltaF_count); 578| 4| } 579| | 580| 252| deltaF.reserve(deltaF_count); 581| 252| std::generate_n(std::back_inserter(deltaF), deltaF_count, [&bs]() { 582| 252| const auto F = bs.get(); 583| 252| switch (std::fpclassify(F)) { 584| 252| case FP_NORMAL: 585| 252| case FP_SUBNORMAL: 586| 252| case FP_ZERO: 587| 252| return F; 588| 252| default: 589| 252| ThrowRDE("Got bad float %f.", implicit_cast(F)); 590| 252| } 591| 252| }); 592| 252| } _ZN8rawspeed10DngOpcodes17DeltaRowOrColBase7SelectX6selectEjj: 518| 41.2k| static uint32_t select(uint32_t x, uint32_t /*y*/) { return x; } _ZZN8rawspeed10DngOpcodes13DeltaRowOrColINS0_17DeltaRowOrColBase7SelectXEEC1ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DEfENKUlvE_clEv: 581| 49.0k| std::generate_n(std::back_inserter(deltaF), deltaF_count, [&bs]() { 582| 49.0k| const auto F = bs.get(); 583| 49.0k| switch (std::fpclassify(F)) { 584| 21.6k| case FP_NORMAL: ------------------ | Branch (584:7): [True: 21.6k, False: 27.3k] ------------------ 585| 21.6k| case FP_SUBNORMAL: ------------------ | Branch (585:7): [True: 0, False: 49.0k] ------------------ 586| 49.0k| case FP_ZERO: ------------------ | Branch (586:7): [True: 27.3k, False: 21.6k] ------------------ 587| 49.0k| return F; 588| 0| default: ------------------ | Branch (588:7): [True: 0, False: 49.0k] ------------------ 589| 0| ThrowRDE("Got bad float %f.", implicit_cast(F)); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 590| 49.0k| } 591| 49.0k| }); _ZN8rawspeed10DngOpcodes13DeltaRowOrColINS0_17DeltaRowOrColBase7SelectXEE5setupERKNS_8RawImageE: 539| 116| void setup(const RawImage& ri) final { 540| 116| PixelOpcode::setup(ri); 541| | 542| | // If we are working on a float image, no need to convert to int 543| 116| if (ri->getDataType() != RawImageType::UINT16) ------------------ | Branch (543:9): [True: 33, False: 83] ------------------ 544| 33| return; 545| | 546| 83| deltaI.reserve(deltaF.size()); 547| 1.17k| for (const auto f : deltaF) { ------------------ | Branch (547:23): [True: 1.17k, False: 70] ------------------ 548| 1.17k| if (!valueIsOk(f)) ------------------ | Branch (548:11): [True: 13, False: 1.15k] ------------------ 549| 1.17k| ThrowRDE("Got float %f which is unacceptable.", ------------------ | | 38| 13| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 13| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 13| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 13| __VA_ARGS__) | | ------------------ ------------------ 550| 1.15k| implicit_cast(f)); 551| 1.15k| deltaI.emplace_back(static_cast(f2iScale * f)); 552| 1.15k| } 553| 83| } _ZN8rawspeed10DngOpcodes17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectXEE5applyERKNS_8RawImageE: 616| 56| void apply(const RawImage& ri) override { 617| 56| if (ri->getDataType() == RawImageType::UINT16) { ------------------ | Branch (617:9): [True: 37, False: 19] ------------------ 618| 37| this->template applyOP( 619| 37| ri, [this](uint32_t x, uint32_t y, uint16_t v) { 620| 37| return clampBits(this->deltaI[S::select(x, y)] + v, 16); 621| 37| }); 622| 37| } else { 623| 19| this->template applyOP(ri, 624| 19| [this](uint32_t x, uint32_t y, float v) { 625| 19| return this->deltaF[S::select(x, y)] + v; 626| 19| }); 627| 19| } 628| 56| } _ZNK8rawspeed10DngOpcodes11PixelOpcode7applyOPItZNS0_17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectXEE5applyERKNS_8RawImageEEUljjtE_EEvS9_T0_: 391| 37| void applyOP(const RawImage& ri, OP op) const { 392| 37| const CroppedArray2DRef img = getDataAsCroppedArray2DRef(ri); 393| 37| int cpp = ri->getCpp(); 394| 37| const iRectangle2D& ROI = getRoi(); 395| 37| const iPoint2D numAffected( 396| 37| implicit_cast(roundUpDivisionSafe(getRoi().dim.x, colPitch)), 397| 37| implicit_cast(roundUpDivisionSafe(getRoi().dim.y, rowPitch))); 398| 1.41k| for (int y = 0; y < numAffected.y; ++y) { ------------------ | Branch (398:21): [True: 1.38k, False: 37] ------------------ 399| 3.08k| for (int x = 0; x < numAffected.x; ++x) { ------------------ | Branch (399:23): [True: 1.70k, False: 1.38k] ------------------ 400| 3.52k| for (auto p = 0U; p < planes; ++p) { ------------------ | Branch (400:27): [True: 1.81k, False: 1.70k] ------------------ 401| 1.81k| T& pixel = 402| 1.81k| img(ROI.getTop() + (rowPitch * y), 403| 1.81k| firstPlane + ((ROI.getLeft() + colPitch * x) * cpp) + p); 404| 1.81k| pixel = op(x, y, pixel); 405| 1.81k| } 406| 1.70k| } 407| 1.38k| } 408| 37| } _ZZN8rawspeed10DngOpcodes17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectXEE5applyERKNS_8RawImageEENKUljjtE_clEjjt: 619| 1.81k| ri, [this](uint32_t x, uint32_t y, uint16_t v) { 620| 1.81k| return clampBits(this->deltaI[S::select(x, y)] + v, 16); 621| 1.81k| }); _ZNK8rawspeed10DngOpcodes11PixelOpcode7applyOPIfZNS0_17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectXEE5applyERKNS_8RawImageEEUljjfE_EEvS9_T0_: 391| 19| void applyOP(const RawImage& ri, OP op) const { 392| 19| const CroppedArray2DRef img = getDataAsCroppedArray2DRef(ri); 393| 19| int cpp = ri->getCpp(); 394| 19| const iRectangle2D& ROI = getRoi(); 395| 19| const iPoint2D numAffected( 396| 19| implicit_cast(roundUpDivisionSafe(getRoi().dim.x, colPitch)), 397| 19| implicit_cast(roundUpDivisionSafe(getRoi().dim.y, rowPitch))); 398| 484| for (int y = 0; y < numAffected.y; ++y) { ------------------ | Branch (398:21): [True: 465, False: 19] ------------------ 399| 13.1k| for (int x = 0; x < numAffected.x; ++x) { ------------------ | Branch (399:23): [True: 12.6k, False: 465] ------------------ 400| 49.8k| for (auto p = 0U; p < planes; ++p) { ------------------ | Branch (400:27): [True: 37.1k, False: 12.6k] ------------------ 401| 37.1k| T& pixel = 402| 37.1k| img(ROI.getTop() + (rowPitch * y), 403| 37.1k| firstPlane + ((ROI.getLeft() + colPitch * x) * cpp) + p); 404| 37.1k| pixel = op(x, y, pixel); 405| 37.1k| } 406| 12.6k| } 407| 465| } 408| 19| } _ZZN8rawspeed10DngOpcodes17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectXEE5applyERKNS_8RawImageEENKUljjfE_clEjjf: 624| 37.1k| [this](uint32_t x, uint32_t y, float v) { 625| 37.1k| return this->deltaF[S::select(x, y)] + v; 626| 37.1k| }); _ZN8rawspeed10DngOpcodes17OffsetPerRowOrColINS0_17DeltaRowOrColBase7SelectXEE9valueIsOkEf: 605| 740| bool valueIsOk(float value) override { 606| 740| return implicit_cast(std::abs(value)) <= absLimit; 607| 740| } _ZN8rawspeed10DngOpcodes11constructorINS0_16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectYEEEEENSt3__110unique_ptrINS0_9DngOpcodeENS6_14default_deleteIS8_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DE: 746| 102| iRectangle2D& integrated_subimg) { 747| 102| return std::make_unique(ri, bs, integrated_subimg); 748| 102|} _ZN8rawspeed10DngOpcodes16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectYEEC2ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DE: 650| 102| : DeltaRowOrCol(ri, bs, integrated_subimg_, 1024.0F), 651| 102| maxLimit((double(std::numeric_limits::max() - rounding) / 652| 102| double(std::numeric_limits::max())) / 653| 102| implicit_cast(this->f2iScale)) {} _ZN8rawspeed10DngOpcodes16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectYEE5applyERKNS_8RawImageE: 655| 54| void apply(const RawImage& ri) override { 656| 54| if (ri->getDataType() == RawImageType::UINT16) { ------------------ | Branch (656:9): [True: 33, False: 21] ------------------ 657| 33| this->template applyOP(ri, [this](uint32_t x, uint32_t y, 658| 33| uint16_t v) { 659| 33| return clampBits((this->deltaI[S::select(x, y)] * v + 512) >> 10, 16); 660| 33| }); 661| 33| } else { 662| 21| this->template applyOP(ri, 663| 21| [this](uint32_t x, uint32_t y, float v) { 664| 21| return this->deltaF[S::select(x, y)] * v; 665| 21| }); 666| 21| } 667| 54| } _ZNK8rawspeed10DngOpcodes11PixelOpcode7applyOPItZNS0_16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectYEE5applyERKNS_8RawImageEEUljjtE_EEvS9_T0_: 391| 33| void applyOP(const RawImage& ri, OP op) const { 392| 33| const CroppedArray2DRef img = getDataAsCroppedArray2DRef(ri); 393| 33| int cpp = ri->getCpp(); 394| 33| const iRectangle2D& ROI = getRoi(); 395| 33| const iPoint2D numAffected( 396| 33| implicit_cast(roundUpDivisionSafe(getRoi().dim.x, colPitch)), 397| 33| implicit_cast(roundUpDivisionSafe(getRoi().dim.y, rowPitch))); 398| 88| for (int y = 0; y < numAffected.y; ++y) { ------------------ | Branch (398:21): [True: 55, False: 33] ------------------ 399| 4.46k| for (int x = 0; x < numAffected.x; ++x) { ------------------ | Branch (399:23): [True: 4.40k, False: 55] ------------------ 400| 9.19k| for (auto p = 0U; p < planes; ++p) { ------------------ | Branch (400:27): [True: 4.79k, False: 4.40k] ------------------ 401| 4.79k| T& pixel = 402| 4.79k| img(ROI.getTop() + (rowPitch * y), 403| 4.79k| firstPlane + ((ROI.getLeft() + colPitch * x) * cpp) + p); 404| 4.79k| pixel = op(x, y, pixel); 405| 4.79k| } 406| 4.40k| } 407| 55| } 408| 33| } _ZZN8rawspeed10DngOpcodes16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectYEE5applyERKNS_8RawImageEENKUljjtE_clEjjt: 658| 4.79k| uint16_t v) { 659| 4.79k| return clampBits((this->deltaI[S::select(x, y)] * v + 512) >> 10, 16); 660| 4.79k| }); _ZNK8rawspeed10DngOpcodes11PixelOpcode7applyOPIfZNS0_16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectYEE5applyERKNS_8RawImageEEUljjfE_EEvS9_T0_: 391| 21| void applyOP(const RawImage& ri, OP op) const { 392| 21| const CroppedArray2DRef img = getDataAsCroppedArray2DRef(ri); 393| 21| int cpp = ri->getCpp(); 394| 21| const iRectangle2D& ROI = getRoi(); 395| 21| const iPoint2D numAffected( 396| 21| implicit_cast(roundUpDivisionSafe(getRoi().dim.x, colPitch)), 397| 21| implicit_cast(roundUpDivisionSafe(getRoi().dim.y, rowPitch))); 398| 42| for (int y = 0; y < numAffected.y; ++y) { ------------------ | Branch (398:21): [True: 21, False: 21] ------------------ 399| 525| for (int x = 0; x < numAffected.x; ++x) { ------------------ | Branch (399:23): [True: 504, False: 21] ------------------ 400| 1.25k| for (auto p = 0U; p < planes; ++p) { ------------------ | Branch (400:27): [True: 753, False: 504] ------------------ 401| 753| T& pixel = 402| 753| img(ROI.getTop() + (rowPitch * y), 403| 753| firstPlane + ((ROI.getLeft() + colPitch * x) * cpp) + p); 404| 753| pixel = op(x, y, pixel); 405| 753| } 406| 504| } 407| 21| } 408| 21| } _ZZN8rawspeed10DngOpcodes16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectYEE5applyERKNS_8RawImageEENKUljjfE_clEjjf: 663| 753| [this](uint32_t x, uint32_t y, float v) { 664| 753| return this->deltaF[S::select(x, y)] * v; 665| 753| }); _ZN8rawspeed10DngOpcodes16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectYEE9valueIsOkEf: 643| 63| bool valueIsOk(float value) override { 644| 63| return value >= minLimit && implicit_cast(value) <= maxLimit; ------------------ | Branch (644:12): [True: 59, False: 4] | Branch (644:33): [True: 58, False: 1] ------------------ 645| 63| } _ZN8rawspeed10DngOpcodes11constructorINS0_16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectXEEEEENSt3__110unique_ptrINS0_9DngOpcodeENS6_14default_deleteIS8_EEEERKNS_8RawImageERNS_10ByteStreamERNS_12iRectangle2DE: 746| 113| iRectangle2D& integrated_subimg) { 747| 113| return std::make_unique(ri, bs, integrated_subimg); 748| 113|} _ZN8rawspeed10DngOpcodes16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectXEEC2ERKNS_8RawImageERNS_10ByteStreamERKNS_12iRectangle2DE: 650| 113| : DeltaRowOrCol(ri, bs, integrated_subimg_, 1024.0F), 651| 113| maxLimit((double(std::numeric_limits::max() - rounding) / 652| 113| double(std::numeric_limits::max())) / 653| 113| implicit_cast(this->f2iScale)) {} _ZN8rawspeed10DngOpcodes16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectXEE5applyERKNS_8RawImageE: 655| 47| void apply(const RawImage& ri) override { 656| 47| if (ri->getDataType() == RawImageType::UINT16) { ------------------ | Branch (656:9): [True: 33, False: 14] ------------------ 657| 33| this->template applyOP(ri, [this](uint32_t x, uint32_t y, 658| 33| uint16_t v) { 659| 33| return clampBits((this->deltaI[S::select(x, y)] * v + 512) >> 10, 16); 660| 33| }); 661| 33| } else { 662| 14| this->template applyOP(ri, 663| 14| [this](uint32_t x, uint32_t y, float v) { 664| 14| return this->deltaF[S::select(x, y)] * v; 665| 14| }); 666| 14| } 667| 47| } _ZNK8rawspeed10DngOpcodes11PixelOpcode7applyOPItZNS0_16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectXEE5applyERKNS_8RawImageEEUljjtE_EEvS9_T0_: 391| 33| void applyOP(const RawImage& ri, OP op) const { 392| 33| const CroppedArray2DRef img = getDataAsCroppedArray2DRef(ri); 393| 33| int cpp = ri->getCpp(); 394| 33| const iRectangle2D& ROI = getRoi(); 395| 33| const iPoint2D numAffected( 396| 33| implicit_cast(roundUpDivisionSafe(getRoi().dim.x, colPitch)), 397| 33| implicit_cast(roundUpDivisionSafe(getRoi().dim.y, rowPitch))); 398| 707| for (int y = 0; y < numAffected.y; ++y) { ------------------ | Branch (398:21): [True: 674, False: 33] ------------------ 399| 1.61k| for (int x = 0; x < numAffected.x; ++x) { ------------------ | Branch (399:23): [True: 936, False: 674] ------------------ 400| 2.09k| for (auto p = 0U; p < planes; ++p) { ------------------ | Branch (400:27): [True: 1.16k, False: 936] ------------------ 401| 1.16k| T& pixel = 402| 1.16k| img(ROI.getTop() + (rowPitch * y), 403| 1.16k| firstPlane + ((ROI.getLeft() + colPitch * x) * cpp) + p); 404| 1.16k| pixel = op(x, y, pixel); 405| 1.16k| } 406| 936| } 407| 674| } 408| 33| } _ZZN8rawspeed10DngOpcodes16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectXEE5applyERKNS_8RawImageEENKUljjtE_clEjjt: 658| 1.16k| uint16_t v) { 659| 1.16k| return clampBits((this->deltaI[S::select(x, y)] * v + 512) >> 10, 16); 660| 1.16k| }); _ZNK8rawspeed10DngOpcodes11PixelOpcode7applyOPIfZNS0_16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectXEE5applyERKNS_8RawImageEEUljjfE_EEvS9_T0_: 391| 14| void applyOP(const RawImage& ri, OP op) const { 392| 14| const CroppedArray2DRef img = getDataAsCroppedArray2DRef(ri); 393| 14| int cpp = ri->getCpp(); 394| 14| const iRectangle2D& ROI = getRoi(); 395| 14| const iPoint2D numAffected( 396| 14| implicit_cast(roundUpDivisionSafe(getRoi().dim.x, colPitch)), 397| 14| implicit_cast(roundUpDivisionSafe(getRoi().dim.y, rowPitch))); 398| 453| for (int y = 0; y < numAffected.y; ++y) { ------------------ | Branch (398:21): [True: 439, False: 14] ------------------ 399| 886| for (int x = 0; x < numAffected.x; ++x) { ------------------ | Branch (399:23): [True: 447, False: 439] ------------------ 400| 1.23k| for (auto p = 0U; p < planes; ++p) { ------------------ | Branch (400:27): [True: 786, False: 447] ------------------ 401| 786| T& pixel = 402| 786| img(ROI.getTop() + (rowPitch * y), 403| 786| firstPlane + ((ROI.getLeft() + colPitch * x) * cpp) + p); 404| 786| pixel = op(x, y, pixel); 405| 786| } 406| 447| } 407| 439| } 408| 14| } _ZZN8rawspeed10DngOpcodes16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectXEE5applyERKNS_8RawImageEENKUljjfE_clEjjf: 663| 786| [this](uint32_t x, uint32_t y, float v) { 664| 786| return this->deltaF[S::select(x, y)] * v; 665| 786| }); _ZN8rawspeed10DngOpcodes16ScalePerRowOrColINS0_17DeltaRowOrColBase7SelectXEE9valueIsOkEf: 643| 430| bool valueIsOk(float value) override { 644| 430| return value >= minLimit && implicit_cast(value) <= maxLimit; ------------------ | Branch (644:12): [True: 426, False: 4] | Branch (644:33): [True: 423, False: 3] ------------------ 645| 430| } _ZN8rawspeed10DngOpcodes9DngOpcode5setupERKNS_8RawImageE: 128| 1.62k| virtual void setup(const RawImage& ri) { 129| 1.62k|#ifndef NDEBUG 130| 1.62k| setup_was_called = true; 131| 1.62k|#endif 132| 1.62k| assert(integrated_subimg == getImageCropAsRectangle(ri) && ------------------ | Branch (132:5): [True: 1.62k, False: 0] | Branch (132:5): [True: 1.62k, Folded] | Branch (132:5): [True: 1.62k, False: 0] ------------------ 133| 1.62k| "Current image sub-crop does not match the expected one!"); 134| | 135| | // NOP by default. child class shall final this if needed. 136| 1.62k| } _ZN8rawspeed10DngOpcodes20FixBadPixelsConstant5setupERKNS_8RawImageE: 161| 83| void setup(const RawImage& ri) override { 162| 83| DngOpcodes::DngOpcode::setup(ri); 163| | 164| | // These limitations are present within the DNG SDK as well. 165| 83| if (ri->getDataType() != RawImageType::UINT16) ------------------ | Branch (165:9): [True: 1, False: 82] ------------------ 166| 83| ThrowRDE("Only 16 bit images supported"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 167| | 168| 82| if (ri->getCpp() > 1) ------------------ | Branch (168:9): [True: 1, False: 81] ------------------ 169| 82| ThrowRDE("Only 1 component images supported"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 170| 82| } _ZN8rawspeed10DngOpcodes20FixBadPixelsConstant5applyERKNS_8RawImageE: 172| 81| void apply(const RawImage& ri) override { 173| 81| MutexLocker guard(&ri->mBadPixelMutex); 174| 81| const CroppedArray2DRef img(ri->getU16DataAsCroppedArray2DRef()); 175| 81| iPoint2D crop = ri->getCropOffset(); 176| 81| uint32_t offset = crop.x | (crop.y << 16); 177| 19.1k| for (auto row = 0; row < img.croppedHeight; ++row) { ------------------ | Branch (177:24): [True: 19.1k, False: 81] ------------------ 178| 212k| for (auto col = 0; col < img.croppedWidth; ++col) { ------------------ | Branch (178:26): [True: 193k, False: 19.1k] ------------------ 179| 193k| if (img(row, col) == value) ------------------ | Branch (179:13): [True: 114k, False: 79.1k] ------------------ 180| 114k| ri->mBadPixelPositions.push_back(offset + (row << 16 | col)); 181| 193k| } 182| 19.1k| } 183| 81| } _ZN8rawspeed10DngOpcodes16FixBadPixelsList5applyERKNS_8RawImageE: 318| 193| void apply(const RawImage& ri) override { 319| 193| MutexLocker guard(&ri->mBadPixelMutex); 320| 193| ri->mBadPixelPositions.insert(ri->mBadPixelPositions.begin(), 321| 193| badPixels.begin(), badPixels.end()); 322| 193| } _ZN8rawspeed10DngOpcodes10TrimBounds5applyERKNS_8RawImageE: 343| 24| void apply(const RawImage& ri) override { ri->subFrame(getRoi()); } _ZN8rawspeed10DngOpcodes12LookupOpcode5setupERKNS_8RawImageE: 426| 44| void setup(const RawImage& ri) final { 427| 44| PixelOpcode::setup(ri); 428| | 429| 44| if (ri->getDataType() != RawImageType::UINT16) ------------------ | Branch (429:9): [True: 1, False: 43] ------------------ 430| 44| ThrowRDE("Only 16 bit images supported"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 431| 44| } _ZN8rawspeed10DngOpcodes12LookupOpcode5applyERKNS_8RawImageE: 433| 43| void apply(const RawImage& ri) final { 434| 43| applyOP(ri, [this]([[maybe_unused]] uint32_t x, 435| 43| [[maybe_unused]] uint32_t y, 436| 43| uint16_t v) { return lookup[v]; }); 437| 43| } _ZNK8rawspeed10DngOpcodes11PixelOpcode7applyOPItZNS0_12LookupOpcode5applyERKNS_8RawImageEEUljjtE_EEvS6_T0_: 391| 43| void applyOP(const RawImage& ri, OP op) const { 392| 43| const CroppedArray2DRef img = getDataAsCroppedArray2DRef(ri); 393| 43| int cpp = ri->getCpp(); 394| 43| const iRectangle2D& ROI = getRoi(); 395| 43| const iPoint2D numAffected( 396| 43| implicit_cast(roundUpDivisionSafe(getRoi().dim.x, colPitch)), 397| 43| implicit_cast(roundUpDivisionSafe(getRoi().dim.y, rowPitch))); 398| 468| for (int y = 0; y < numAffected.y; ++y) { ------------------ | Branch (398:21): [True: 425, False: 43] ------------------ 399| 1.88k| for (int x = 0; x < numAffected.x; ++x) { ------------------ | Branch (399:23): [True: 1.46k, False: 425] ------------------ 400| 3.38k| for (auto p = 0U; p < planes; ++p) { ------------------ | Branch (400:27): [True: 1.91k, False: 1.46k] ------------------ 401| 1.91k| T& pixel = 402| 1.91k| img(ROI.getTop() + (rowPitch * y), 403| 1.91k| firstPlane + ((ROI.getLeft() + colPitch * x) * cpp) + p); 404| 1.91k| pixel = op(x, y, pixel); 405| 1.91k| } 406| 1.46k| } 407| 425| } 408| 43| } _ZZN8rawspeed10DngOpcodes12LookupOpcode5applyERKNS_8RawImageEENKUljjtE_clEjjt: 436| 1.91k| uint16_t v) { return lookup[v]; }); _ZN8rawspeed8ErrorLog8setErrorERKNSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEE: 29| 19.2k|void ErrorLog::setError(const std::string& err) { 30| 19.2k| MutexLocker guard(&mutex); 31| 19.2k| errors.push_back(err); 32| 19.2k|} _ZN8rawspeed8ErrorLog15isTooManyErrorsEjPNSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEE: 34| 3.93k|bool ErrorLog::isTooManyErrors(unsigned many, std::string* firstErr) { 35| 3.93k| MutexLocker guard(&mutex); 36| | 37| 3.93k| if (errors.size() < many) ------------------ | Branch (37:7): [True: 2.87k, False: 1.05k] ------------------ 38| 2.87k| return false; 39| | 40| 1.05k| if (!firstErr) ------------------ | Branch (40:7): [True: 0, False: 1.05k] ------------------ 41| 0| return true; 42| | 43| 1.05k| *firstErr = errors[0]; 44| 1.05k| return true; 45| 1.05k|} _ZN8rawspeed25extendBinaryFloatingPointINS_13ieee_754_20088Binary16ENS1_8Binary32EEEjj: 108| 114k|inline uint32_t extendBinaryFloatingPoint(uint32_t narrow) { 109| 114k| uint32_t sign = (narrow >> NarrowType::SignBitPos) & 1; 110| 114k| uint32_t narrow_exponent = (narrow >> NarrowType::ExponentPos) & 111| 114k| ((1 << NarrowType::ExponentWidth) - 1); 112| 114k| uint32_t narrow_fraction = narrow & ((1 << NarrowType::FractionWidth) - 1); 113| | 114| | // Normalized or zero 115| 114k| uint32_t wide_exponent = 116| 114k| static_cast(narrow_exponent) - NarrowType::Bias + WideType::Bias; 117| 114k| uint32_t wide_fraction = 118| 114k| narrow_fraction << (WideType::FractionWidth - NarrowType::FractionWidth); 119| | 120| 114k| if (narrow_exponent == ((1 << NarrowType::ExponentWidth) - 1)) { ------------------ | Branch (120:7): [True: 4.10k, False: 110k] ------------------ 121| | // Infinity or NaN 122| 4.10k| wide_exponent = ((1 << WideType::ExponentWidth) - 1); 123| | // Narrow fraction is kept/widened! 124| 110k| } else if (narrow_exponent == 0) { ------------------ | Branch (124:14): [True: 64.7k, False: 46.0k] ------------------ 125| 64.7k| if (narrow_fraction == 0) { ------------------ | Branch (125:9): [True: 46.7k, False: 17.9k] ------------------ 126| | // +-Zero 127| 46.7k| wide_exponent = 0; 128| 46.7k| wide_fraction = 0; 129| 46.7k| } else { 130| | // Subnormal numbers 131| | // We can represent it as a normalized value in wider type, 132| | // we have to shift fraction until we get 1.new_fraction 133| | // and decrement exponent for each shift. 134| | // FIXME; what is the implicit precondition here? 135| 17.9k| wide_exponent = 1 - NarrowType::Bias + WideType::Bias; 136| 100k| while (!(wide_fraction & (1 << WideType::FractionWidth))) { ------------------ | Branch (136:14): [True: 82.1k, False: 17.9k] ------------------ 137| 82.1k| wide_exponent -= 1; 138| 82.1k| wide_fraction <<= 1; 139| 82.1k| } 140| 17.9k| wide_fraction &= ((1 << WideType::FractionWidth) - 1); 141| 17.9k| } 142| 64.7k| } 143| 114k| return (sign << WideType::SignBitPos) | 144| 114k| (wide_exponent << WideType::ExponentPos) | wide_fraction; 145| 114k|} _ZN8rawspeed25extendBinaryFloatingPointINS_13ieee_754_20088Binary24ENS1_8Binary32EEEjj: 108| 87.0k|inline uint32_t extendBinaryFloatingPoint(uint32_t narrow) { 109| 87.0k| uint32_t sign = (narrow >> NarrowType::SignBitPos) & 1; 110| 87.0k| uint32_t narrow_exponent = (narrow >> NarrowType::ExponentPos) & 111| 87.0k| ((1 << NarrowType::ExponentWidth) - 1); 112| 87.0k| uint32_t narrow_fraction = narrow & ((1 << NarrowType::FractionWidth) - 1); 113| | 114| | // Normalized or zero 115| 87.0k| uint32_t wide_exponent = 116| 87.0k| static_cast(narrow_exponent) - NarrowType::Bias + WideType::Bias; 117| 87.0k| uint32_t wide_fraction = 118| 87.0k| narrow_fraction << (WideType::FractionWidth - NarrowType::FractionWidth); 119| | 120| 87.0k| if (narrow_exponent == ((1 << NarrowType::ExponentWidth) - 1)) { ------------------ | Branch (120:7): [True: 5.33k, False: 81.6k] ------------------ 121| | // Infinity or NaN 122| 5.33k| wide_exponent = ((1 << WideType::ExponentWidth) - 1); 123| | // Narrow fraction is kept/widened! 124| 81.6k| } else if (narrow_exponent == 0) { ------------------ | Branch (124:14): [True: 33.5k, False: 48.0k] ------------------ 125| 33.5k| if (narrow_fraction == 0) { ------------------ | Branch (125:9): [True: 23.7k, False: 9.85k] ------------------ 126| | // +-Zero 127| 23.7k| wide_exponent = 0; 128| 23.7k| wide_fraction = 0; 129| 23.7k| } else { 130| | // Subnormal numbers 131| | // We can represent it as a normalized value in wider type, 132| | // we have to shift fraction until we get 1.new_fraction 133| | // and decrement exponent for each shift. 134| | // FIXME; what is the implicit precondition here? 135| 9.85k| wide_exponent = 1 - NarrowType::Bias + WideType::Bias; 136| 79.8k| while (!(wide_fraction & (1 << WideType::FractionWidth))) { ------------------ | Branch (136:14): [True: 69.9k, False: 9.85k] ------------------ 137| 69.9k| wide_exponent -= 1; 138| 69.9k| wide_fraction <<= 1; 139| 69.9k| } 140| 9.85k| wide_fraction &= ((1 << WideType::FractionWidth) - 1); 141| 9.85k| } 142| 33.5k| } 143| 87.0k| return (sign << WideType::SignBitPos) | 144| 87.0k| (wide_exponent << WideType::ExponentPos) | wide_fraction; 145| 87.0k|} rawspeed_get_number_of_processor_cores: 31| 4.36k|rawspeed_get_number_of_processor_cores() { 32| 4.36k| return omp_get_max_threads(); 33| 4.36k|} _ZN8rawspeed12RawImageData10createDataEv: 68| 7.99k|void RawImageData::createData() { 69| 7.99k| static constexpr const auto alignment = 16; 70| | 71| 7.99k| if (dim.x > 65535 || dim.y > 65535) ------------------ | Branch (71:7): [True: 0, False: 7.99k] | Branch (71:24): [True: 0, False: 7.99k] ------------------ 72| 7.99k| ThrowRDE("Dimensions too large for allocation."); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 73| 7.99k| if (dim.x <= 0 || dim.y <= 0) ------------------ | Branch (73:7): [True: 32, False: 7.95k] | Branch (73:21): [True: 22, False: 7.93k] ------------------ 74| 7.99k| ThrowRDE("Dimension of one sides is less than 1 - cannot allocate image."); ------------------ | | 38| 54| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 54| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 54| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 54| __VA_ARGS__) | | ------------------ ------------------ 75| 7.93k| if (cpp <= 0 || bpp <= 0) ------------------ | Branch (75:7): [True: 0, False: 7.93k] | Branch (75:19): [True: 0, False: 7.93k] ------------------ 76| 7.93k| ThrowRDE("Unspecified component count - cannot allocate image."); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 77| 7.93k| if (isAllocated()) ------------------ | Branch (77:7): [True: 0, False: 7.93k] ------------------ 78| 7.93k| ThrowRDE("Duplicate data allocation in createData."); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 79| | 80| | // want each line to start at 16-byte aligned address 81| 7.93k| pitch = 82| 7.93k| implicit_cast(roundUp(static_cast(dim.x) * bpp, alignment)); 83| 7.93k| assert(isAligned(pitch, alignment)); ------------------ | Branch (83:3): [True: 7.93k, False: 0] ------------------ 84| | 85| |#if defined(DEBUG) || __has_feature(address_sanitizer) || \ 86| | defined(__SANITIZE_ADDRESS__) 87| | // want to ensure that we have some padding 88| | pitch += alignment * alignment; 89| | assert(isAligned(pitch, alignment)); 90| |#endif 91| | 92| 7.93k| padding = pitch - dim.x * bpp; 93| | 94| |#if defined(DEBUG) || __has_feature(address_sanitizer) || \ 95| | defined(__SANITIZE_ADDRESS__) 96| | assert(padding > 0); 97| |#endif 98| | 99| 7.93k| data.resize(static_cast(pitch) * dim.y); 100| | 101| 7.93k| uncropped_dim = dim; 102| | 103| 7.93k|#ifndef NDEBUG 104| 7.93k| const Array2DRef img = getByteDataAsUncroppedArray2DRef(); 105| | 106| 5.67M| for (int j = 0; j < dim.y; j++) { ------------------ | Branch (106:19): [True: 5.66M, False: 7.93k] ------------------ 107| | // each line is indeed 16-byte aligned 108| 5.66M| assert(isAligned(&img(j, 0), alignment)); ------------------ | Branch (108:5): [True: 5.66M, False: 0] ------------------ 109| 5.66M| } 110| 7.93k|#endif 111| | 112| 7.93k| poisonPadding(); 113| 7.93k|} _ZN8rawspeed12RawImageData13poisonPaddingEv: 127| 7.93k|void RawImageData::poisonPadding() { 128| | // if we are building without ASAN, then there is no need/way to poison. 129| | // however, i think it is better to have such an empty function rather 130| | // than making this whole function not exist in ASAN-less builds 131| 7.93k|} _ZN8rawspeed12RawImageData6setCppEj: 153| 3.78k|void RawImageData::setCpp(uint32_t val) { 154| 3.78k| if (isAllocated()) ------------------ | Branch (154:7): [True: 0, False: 3.78k] ------------------ 155| 3.78k| ThrowRDE("Attempted to set Components per pixel after data allocation"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 156| 3.78k| if (val > 4) { ------------------ | Branch (156:7): [True: 0, False: 3.78k] ------------------ 157| 0| ThrowRDE( ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 158| 0| "Only up to 4 components per pixel is support - attempted to set: %u", 159| 0| val); 160| 0| } 161| | 162| 3.78k| bpp /= cpp; 163| 3.78k| cpp = val; 164| 3.78k| bpp *= val; 165| 3.78k|} _ZNK8rawspeed12RawImageData15getUncroppedDimEv: 167| 646|iPoint2D RAWSPEED_READONLY rawspeed::RawImageData::getUncroppedDim() const { 168| 646| return uncropped_dim; 169| 646|} _ZNK8rawspeed12RawImageData13getCropOffsetEv: 171| 314|iPoint2D RAWSPEED_READONLY RawImageData::getCropOffset() const { 172| 314| return mOffset; 173| 314|} _ZN8rawspeed12RawImageData8subFrameENS_12iRectangle2DE: 175| 213|void RawImageData::subFrame(iRectangle2D crop) { 176| 213| if (!crop.hasPositiveArea()) ------------------ | Branch (176:7): [True: 74, False: 139] ------------------ 177| 213| ThrowRDE("No positive crop area"); ------------------ | | 38| 74| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 74| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 74| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 74| __VA_ARGS__) | | ------------------ ------------------ 178| | 179| 139| if (!crop.dim.isThisInside(dim - crop.pos)) { ------------------ | Branch (179:7): [True: 0, False: 139] ------------------ 180| 0| writeLog(DEBUG_PRIO::WARNING, "WARNING: RawImageData::subFrame - Attempted " 181| 0| "to create new subframe larger than original " 182| 0| "size. Crop skipped."); 183| 0| return; 184| 0| } 185| 139| if (crop.pos.x < 0 || crop.pos.y < 0 || crop.dim.x < 0 || crop.dim.y < 0) { ------------------ | Branch (185:7): [True: 0, False: 139] | Branch (185:25): [True: 0, False: 139] | Branch (185:43): [True: 0, False: 139] | Branch (185:61): [True: 0, False: 139] ------------------ 186| 0| writeLog(DEBUG_PRIO::WARNING, "WARNING: RawImageData::subFrame - Negative " 187| 0| "crop offset. Crop skipped."); 188| 0| return; 189| 0| } 190| | 191| | // if CFA, and not X-Trans, adjust. 192| 139| if (isCFA && cfa.getDcrawFilter() != 1 && cfa.getDcrawFilter() != 9) { ------------------ | Branch (192:7): [True: 19, False: 120] | Branch (192:16): [True: 18, False: 1] | Branch (192:45): [True: 18, False: 0] ------------------ 193| 18| cfa.shiftRight(crop.pos.x); 194| 18| cfa.shiftDown(crop.pos.y); 195| 18| } 196| | 197| 139| mOffset += crop.pos; 198| 139| dim = crop.dim; 199| 139|} _ZN8rawspeed12RawImageData11startWorkerENS_14RawImageWorker18RawImageWorkerTaskEb: 272| 66| bool cropped) { 273| 66| const int height = [&]() { 274| 66| int h = cropped ? dim.y : uncropped_dim.y; 275| 66| if (static_cast(task) & 276| 66| static_cast(RawImageWorker::RawImageWorkerTask::FULL_IMAGE)) { 277| 66| h = uncropped_dim.y; 278| 66| } 279| 66| return h; 280| 66| }(); 281| | 282| 66| const int threads = rawspeed_get_number_of_processor_cores(); 283| 66| const int y_per_thread = (height + threads - 1) / threads; 284| | 285| 66|#ifdef HAVE_OPENMP 286| 66|#pragma omp parallel for default(none) \ 287| 66| firstprivate(threads, y_per_thread, height, task) num_threads(threads) \ 288| 66| schedule(static) 289| 66|#endif 290| 66| for (int i = 0; i < threads; i++) { 291| 66| int y_offset = std::min(i * y_per_thread, height); 292| 66| int y_end = std::min((i + 1) * y_per_thread, height); 293| | 294| 66| RawImageWorker worker(this, task, y_offset, y_end); 295| 66| } 296| 66|} RawImage.cpp:_ZN8rawspeed12RawImageData11startWorkerENS_14RawImageWorker18RawImageWorkerTaskEb.omp_outlined_debug__: 290| 4.17k| for (int i = 0; i < threads; i++) { ------------------ | Branch (290:19): [True: 2.08k, False: 2.08k] ------------------ 291| 2.08k| int y_offset = std::min(i * y_per_thread, height); 292| 2.08k| int y_end = std::min((i + 1) * y_per_thread, height); 293| | 294| 2.08k| RawImageWorker worker(this, task, y_offset, y_end); 295| 2.08k| } RawImage.cpp:_ZZN8rawspeed12RawImageData11startWorkerENS_14RawImageWorker18RawImageWorkerTaskEbENK3$_0clEv: 273| 66| const int height = [&]() { 274| 66| int h = cropped ? dim.y : uncropped_dim.y; ------------------ | Branch (274:13): [True: 66, False: 0] ------------------ 275| 66| if (static_cast(task) & ------------------ | Branch (275:9): [True: 66, False: 0] ------------------ 276| 66| static_cast(RawImageWorker::RawImageWorkerTask::FULL_IMAGE)) { 277| 66| h = uncropped_dim.y; 278| 66| } 279| 66| return h; 280| 66| }(); _ZN8rawspeed14RawImageWorkerC2EPNS_12RawImageDataENS0_18RawImageWorkerTaskEii: 342| 2.08k| : data(_img), task(_task), start_y(_start_y), end_y(_end_y) { 343| 2.08k| performTask(); 344| 2.08k|} _ZN8rawspeed14RawImageWorker11performTaskEv: 346| 2.08k|void RawImageWorker::performTask() noexcept { 347| 2.08k| try { 348| 2.08k| switch (task) { 349| 0| using enum RawImageWorkerTask; 350| 0| case SCALE_VALUES: ------------------ | Branch (350:5): [True: 0, False: 2.08k] ------------------ 351| 0| data->scaleValues(start_y, end_y); 352| 0| break; 353| 0| case FIX_BAD_PIXELS: ------------------ | Branch (353:5): [True: 0, False: 2.08k] ------------------ 354| 0| data->fixBadPixelsThread(start_y, end_y); 355| 0| break; 356| 2.09k| case APPLY_LOOKUP: ------------------ | Branch (356:5): [True: 2.09k, False: 18.4E] ------------------ 357| 2.09k| data->doLookup(start_y, end_y); 358| 2.09k| break; 359| 0| default: ------------------ | Branch (359:5): [True: 0, False: 2.08k] ------------------ 360| 0| assert(false); ------------------ | Branch (360:7): [Folded, False: 0] ------------------ 361| 2.08k| } 362| 2.08k| } catch (const RawDecoderException& e) { 363| 256| data->setError(e.what()); 364| 256| } catch (const TiffParserException& e) { 365| 0| data->setError(e.what()); 366| 0| } catch (const IOException& e) { 367| 0| data->setError(e.what()); 368| 0| } catch (...) { 369| | // We should not get any other exception type here. 370| 0| __builtin_unreachable(); 371| 0| } 372| 2.08k|} _ZN8rawspeed12RawImageData16sixteenBitLookupEv: 374| 66|void RawImageData::sixteenBitLookup() { 375| 66| if (table == nullptr) { ------------------ | Branch (375:7): [True: 0, False: 66] ------------------ 376| 0| return; 377| 0| } 378| 66| startWorker(RawImageWorker::RawImageWorkerTask::APPLY_LOOKUP, true); 379| 66|} _ZN8rawspeed12RawImageData8setTableENSt3__110unique_ptrINS_11TableLookUpENS1_14default_deleteIS3_EEEE: 381| 1.31k|void RawImageData::setTable(std::unique_ptr t) { 382| 1.31k| table = std::move(t); 383| 1.31k|} _ZN8rawspeed12RawImageData8setTableERKNSt3__16vectorItNS1_9allocatorItEEEEb: 385| 659|void RawImageData::setTable(const std::vector& table_, bool dither) { 386| 659| assert(!table_.empty()); ------------------ | Branch (386:3): [True: 659, False: 0] ------------------ 387| | 388| 659| auto t = std::make_unique(1, dither); 389| 659| t->setTable(0, table_); 390| 659| this->setTable(std::move(t)); 391| 659|} _ZNK8rawspeed12RawImageData6getCppEv: 118| 63.8k| [[nodiscard]] uint32_t RAWSPEED_READONLY getCpp() const { return cpp; } _ZNK8rawspeed12RawImageData6getBppEv: 119| 7.29k| [[nodiscard]] uint32_t RAWSPEED_READONLY getBpp() const { return bpp; } _ZNK8rawspeed12RawImageData11getDataTypeEv: 125| 19.4k| [[nodiscard]] rawspeed::RawImageType getDataType() const { return dataType; } _ZNK8rawspeed12RawImageData11isAllocatedEv: 152| 11.7k| [[nodiscard]] bool isAllocated() const { return !data.empty(); } _ZNK8rawspeed8RawImageptEv: 253| 5.61M| RawImageData* RAWSPEED_READONLY operator->() const { return &*p_; } _ZNK8rawspeed8RawImagedeEv: 254| 18.1k| RawImageData& RAWSPEED_READONLY operator*() const { return *p_; } _ZN8rawspeed8RawImageC2ENSt3__110shared_ptrINS_12RawImageDataEEE: 257| 15.4k| explicit RawImage(std::shared_ptr p) : p_(std::move(p)) {} _ZN8rawspeed8RawImage3getEv: 259| 177| RawImageData* get() { return &*p_; } _ZN8rawspeed18RawImageCurveGuardC2EPKNS_8RawImageERKNSt3__16vectorItNS4_9allocatorItEEEEb: 379| 659| : mRaw(raw), curve(curve_), uncorrectedRawValues(uncorrectedRawValues_) { 380| 659| if (uncorrectedRawValues) ------------------ | Branch (380:9): [True: 0, False: 659] ------------------ 381| 0| return; 382| | 383| 659| (*mRaw)->setTable(curve, true); 384| 659| } _ZN8rawspeed18RawImageCurveGuardD2Ev: 386| 658| ~RawImageCurveGuard() { 387| | // Set the table, if it should be needed later. 388| 658| if (uncorrectedRawValues) ------------------ | Branch (388:9): [True: 0, False: 658] ------------------ 389| 0| (*mRaw)->setTable(curve, false); 390| 658| else 391| 658| (*mRaw)->setTable(nullptr); 392| 658| } _ZN8rawspeed8RawImage6createENS_12RawImageTypeE: 265| 15.4k|inline RawImage RawImage::create(RawImageType type) { 266| 15.4k| switch (type) { ------------------ | Branch (266:11): [True: 15.4k, False: 0] ------------------ 267| 15.1k| case RawImageType::UINT16: ------------------ | Branch (267:3): [True: 15.1k, False: 308] ------------------ 268| 15.1k| return RawImage(std::make_shared()); 269| 308| case RawImageType::F32: ------------------ | Branch (269:3): [True: 308, False: 15.1k] ------------------ 270| 308| return RawImage(std::make_shared()); 271| 15.4k| } 272| 0| writeLog(DEBUG_PRIO::ERROR, "RawImage::create: Unknown Image type!"); 273| 0| __builtin_unreachable(); 274| 15.4k|} _ZN8rawspeed12RawImageData31getU16DataAsUncroppedArray2DRefEv: 290| 2.76M|RawImageData::getU16DataAsUncroppedArray2DRef() noexcept { 291| 2.76M| assert(dataType == RawImageType::UINT16 && ------------------ | Branch (291:3): [True: 2.76M, False: 18.4E] | Branch (291:3): [True: 2.76M, Folded] | Branch (291:3): [True: 2.76M, False: 18.4E] ------------------ 292| 2.76M| "Attempting to access floating-point buffer as uint16_t."); 293| 2.76M| assert(!data.empty() && "Data not yet allocated."); ------------------ | Branch (293:3): [True: 2.76M, False: 18.4E] | Branch (293:3): [True: 2.76M, Folded] | Branch (293:3): [True: 2.76M, False: 18.4E] ------------------ 294| 2.76M|#pragma GCC diagnostic push 295| 2.76M|#pragma GCC diagnostic ignored "-Wpragmas" 296| 2.76M|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 297| 2.76M|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 298| 2.76M| return {reinterpret_cast(data.data()), cpp * uncropped_dim.x, 299| 2.76M| uncropped_dim.y, static_cast(pitch / sizeof(uint16_t))}; 300| 2.76M|#pragma GCC diagnostic pop 301| 2.76M|} _ZN8rawspeed12RawImageData29getU16DataAsCroppedArray2DRefEv: 304| 2.65k|RawImageData::getU16DataAsCroppedArray2DRef() noexcept { 305| 2.65k| return {getU16DataAsUncroppedArray2DRef(), cpp * mOffset.x, mOffset.y, 306| 2.65k| cpp * dim.x, dim.y}; 307| 2.65k|} _ZN8rawspeed12RawImageData31getF32DataAsUncroppedArray2DRefEv: 310| 1.17k|RawImageData::getF32DataAsUncroppedArray2DRef() noexcept { 311| 1.17k| assert(dataType == RawImageType::F32 && ------------------ | Branch (311:3): [True: 1.17k, False: 0] | Branch (311:3): [True: 1.17k, Folded] | Branch (311:3): [True: 1.17k, False: 0] ------------------ 312| 1.17k| "Attempting to access integer buffer as float."); 313| 1.17k| assert(!data.empty() && "Data not yet allocated."); ------------------ | Branch (313:3): [True: 1.17k, False: 0] | Branch (313:3): [True: 1.17k, Folded] | Branch (313:3): [True: 1.17k, False: 0] ------------------ 314| 1.17k|#pragma GCC diagnostic push 315| 1.17k|#pragma GCC diagnostic ignored "-Wpragmas" 316| 1.17k|#pragma GCC diagnostic ignored "-Wunknown-warning-option" 317| 1.17k|#pragma GCC diagnostic ignored "-Wunsafe-buffer-usage" 318| 1.17k| return {reinterpret_cast(data.data()), cpp * uncropped_dim.x, 319| 1.17k| uncropped_dim.y, static_cast(pitch / sizeof(float))}; 320| 1.17k|#pragma GCC diagnostic pop 321| 1.17k|} _ZN8rawspeed12RawImageData29getF32DataAsCroppedArray2DRefEv: 324| 285|RawImageData::getF32DataAsCroppedArray2DRef() noexcept { 325| 285| return {getF32DataAsUncroppedArray2DRef(), cpp * mOffset.x, mOffset.y, 326| 285| cpp * dim.x, dim.y}; 327| 285|} _ZN8rawspeed12RawImageData32getByteDataAsUncroppedArray2DRefEv: 330| 11.0k|RawImageData::getByteDataAsUncroppedArray2DRef() noexcept { 331| 11.0k| switch (dataType) { ------------------ | Branch (331:11): [True: 11.0k, False: 0] ------------------ 332| 10.4k| case RawImageType::UINT16: ------------------ | Branch (332:3): [True: 10.4k, False: 545] ------------------ 333| 10.4k| return getU16DataAsUncroppedArray2DRef(); 334| 545| case RawImageType::F32: ------------------ | Branch (334:3): [True: 545, False: 10.4k] ------------------ 335| 545| return getF32DataAsUncroppedArray2DRef(); 336| 11.0k| } 337| 0| __builtin_unreachable(); 338| 11.0k|} _ZN8rawspeed15RawImageDataU1613setWithLookUpEtPSt4bytePj: 346| 1.86M| uint32_t* random) { 347| 1.86M| auto* dest = reinterpret_cast(dst); 348| 1.86M| if (table == nullptr) { ------------------ | Branch (348:7): [True: 0, False: 1.86M] ------------------ 349| 0| *dest = value; 350| 0| return; 351| 0| } 352| 1.86M| if (table->dither) { ------------------ | Branch (352:7): [True: 1.86M, False: 5.21k] ------------------ 353| 1.86M| uint32_t base = table->tables[(2 * value) + 0]; 354| 1.86M| uint32_t delta = table->tables[(2 * value) + 1]; 355| 1.86M| uint32_t r = *random; 356| | 357| 1.86M| uint32_t pix = base + ((delta * (r & 2047) + 1024) >> 12); 358| 1.86M| *random = 15700 * (r & 65535) + (r >> 16); 359| 1.86M| *dest = implicit_cast(pix); 360| 1.86M| return; 361| 1.86M| } 362| 5.21k| *dest = table->tables[value]; 363| 5.21k|} _ZN8rawspeed12RawImageDataD2Ev: 117| 15.4k| virtual ~RawImageData() = default; _ZN8rawspeed12RawImageDataC2Ev: 191| 15.4k| RawImageData() = default; _ZN8rawspeed17RawImageDataFloatC2Ev: 43| 308|RawImageDataFloat::RawImageDataFloat() { 44| 308| bpp = sizeof(float); 45| 308| dataType = RawImageType::F32; 46| 308|} _ZN8rawspeed17RawImageDataFloat8doLookupEii: 262| 254|void RawImageDataFloat::doLookup(int start_y, int end_y) { 263| 254| ThrowRDE("Float point lookup tables not implemented"); ------------------ | | 38| 254| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 254| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 254| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 254| __VA_ARGS__) | | ------------------ ------------------ 264| 254|} _ZN8rawspeed15RawImageDataU16C2Ev: 52| 15.1k|RawImageDataU16::RawImageDataU16() { 53| 15.1k| dataType = RawImageType::UINT16; 54| 15.1k| bpp = sizeof(uint16_t); 55| 15.1k|} _ZN8rawspeed15RawImageDataU168doLookupEii: 488| 1.84k|void RawImageDataU16::doLookup(int start_y, int end_y) { 489| 1.84k| const Array2DRef img = getU16DataAsUncroppedArray2DRef(); 490| | 491| 1.84k| if (table->ntables == 1) { ------------------ | Branch (491:7): [True: 1.82k, False: 14] ------------------ 492| 1.83k| if (table->dither) { ------------------ | Branch (492:9): [True: 1.83k, False: 18.4E] ------------------ 493| 1.83k| int gw = uncropped_dim.x * cpp; 494| 1.83k| const auto t = table->getTable(0); 495| 7.97k| for (int y = start_y; y < end_y; y++) { ------------------ | Branch (495:29): [True: 6.14k, False: 1.83k] ------------------ 496| 6.14k| uint32_t v = (uncropped_dim.x + y * 13) ^ 0x45694584; 497| 363k| for (int x = 0; x < gw; x++) { ------------------ | Branch (497:25): [True: 357k, False: 6.14k] ------------------ 498| 357k| uint16_t p = img(y, x); 499| 357k| uint32_t base = t((2 * p) + 0); 500| 357k| uint32_t delta = t((2 * p) + 1); 501| 357k| v = 15700 * (v & 65535) + (v >> 16); 502| 357k| uint32_t pix = base + ((delta * (v & 2047) + 1024) >> 12); 503| 357k| img(y, x) = clampBits(pix, 16); 504| 357k| } 505| 6.14k| } 506| 1.83k| return; 507| 1.83k| } 508| | 509| 18.4E| int gw = uncropped_dim.x * cpp; 510| 18.4E| const auto t = table->getTable(0); 511| 18.4E| for (int y = start_y; y < end_y; y++) { ------------------ | Branch (511:27): [True: 0, False: 18.4E] ------------------ 512| 0| for (int x = 0; x < gw; x++) { ------------------ | Branch (512:23): [True: 0, False: 0] ------------------ 513| 0| img(y, x) = t(img(y, x)); 514| 0| } 515| 0| } 516| 18.4E| return; 517| 1.82k| } 518| 14| ThrowRDE("Table lookup with multiple components not implemented"); ------------------ | | 38| 14| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 14| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 14| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 14| __VA_ARGS__) | | ------------------ ------------------ 519| 1.84k|} _ZN8rawspeed17RawspeedException3logEPKc: 58| 172k| log(const char* msg) { 59| 172k| writeLog(DEBUG_PRIO::EXTRA, "EXCEPTION: %s", msg); 60| 172k| } _ZN8rawspeed17RawspeedExceptionC2EPKc: 67| 172k| : std::runtime_error(msg) { 68| 172k| log(msg); 69| 172k| } _ZN8rawspeed14ThrowExceptionINS_11IOExceptionEEEvPKcz: 36| 121k| ThrowException(const char* fmt, ...) { 37| 121k| static constexpr size_t bufSize = 8192; 38| 121k|#if defined(HAVE_CXX_THREAD_LOCAL) 39| 121k| static thread_local std::array buf; 40| |#elif defined(HAVE_GCC_THREAD_LOCAL) 41| | static __thread char buf[bufSize]; 42| |#else 43| |#pragma message \ 44| | "Don't have thread-local-storage! Exception text may be garbled if used multithreaded" 45| | static char buf[bufSize]; 46| |#endif 47| | 48| 121k| va_list val; 49| 121k| va_start(val, fmt); 50| 121k| vsnprintf(buf.data(), sizeof(buf), fmt, val); 51| | va_end(val); 52| 121k| writeLog(DEBUG_PRIO::EXTRA, "EXCEPTION: %s", buf.data()); 53| 121k| throw T(buf.data()); 54| 121k|} _ZN8rawspeed14ThrowExceptionINS_19TiffParserExceptionEEEvPKcz: 36| 25.1k| ThrowException(const char* fmt, ...) { 37| 25.1k| static constexpr size_t bufSize = 8192; 38| 25.1k|#if defined(HAVE_CXX_THREAD_LOCAL) 39| 25.1k| static thread_local std::array buf; 40| |#elif defined(HAVE_GCC_THREAD_LOCAL) 41| | static __thread char buf[bufSize]; 42| |#else 43| |#pragma message \ 44| | "Don't have thread-local-storage! Exception text may be garbled if used multithreaded" 45| | static char buf[bufSize]; 46| |#endif 47| | 48| 25.1k| va_list val; 49| 25.1k| va_start(val, fmt); 50| 25.1k| vsnprintf(buf.data(), sizeof(buf), fmt, val); 51| | va_end(val); 52| 25.1k| writeLog(DEBUG_PRIO::EXTRA, "EXCEPTION: %s", buf.data()); 53| 25.1k| throw T(buf.data()); 54| 25.1k|} _ZN8rawspeed14ThrowExceptionINS_19RawDecoderExceptionEEEvPKcz: 36| 25.7k| ThrowException(const char* fmt, ...) { 37| 25.7k| static constexpr size_t bufSize = 8192; 38| 25.7k|#if defined(HAVE_CXX_THREAD_LOCAL) 39| 25.7k| static thread_local std::array buf; 40| |#elif defined(HAVE_GCC_THREAD_LOCAL) 41| | static __thread char buf[bufSize]; 42| |#else 43| |#pragma message \ 44| | "Don't have thread-local-storage! Exception text may be garbled if used multithreaded" 45| | static char buf[bufSize]; 46| |#endif 47| | 48| 25.7k| va_list val; 49| 25.7k| va_start(val, fmt); 50| 25.7k| vsnprintf(buf.data(), sizeof(buf), fmt, val); 51| | va_end(val); 52| 25.7k| writeLog(DEBUG_PRIO::EXTRA, "EXCEPTION: %s", buf.data()); 53| 25.7k| throw T(buf.data()); 54| 25.7k|} _ZN8rawspeed9SimpleLUTIjLi12EEC2Ev: 38| 572| SimpleLUT() = default; OlympusDecompressor.cpp:_ZN8rawspeed9SimpleLUTIaLi12EEC2INS_12_GLOBAL__N_123OlympusDecompressorImpl5numLZMUlmjE_EQaantsr3stdE9is_same_vINS0_IT_XT0_EEEu11__remove_cvIu20__remove_reference_tITL0__EEEsr3stdE16is_convertible_vISA_NSt3__18functionIFS7_NDtL_ZNS_9SimpleLUT5tableEEE9size_typeESH_EEEEEES7_: 51| 133| explicit SimpleLUT(F f) { 52| 133| const auto fullTableSize = 1U << TableBitWidth; 53| 133| table.reserve(fullTableSize); 54| 133| std::generate_n(std::back_inserter(table), fullTableSize, 55| 133| [&f, table_ = &table]() { 56| | // which row [0..fullTableSize) are we filling? 57| 133| const auto i = table_->size(); 58| 133| return f(i, fullTableSize); 59| 133| }); 60| | assert(table.size() == fullTableSize); ------------------ | Branch (60:5): [True: 133, False: 0] ------------------ 61| 133| } OlympusDecompressor.cpp:_ZZN8rawspeed9SimpleLUTIaLi12EEC1INS_12_GLOBAL__N_123OlympusDecompressorImpl5numLZMUlmjE_EQaantsr3stdE9is_same_vINS0_IT_XT0_EEEu11__remove_cvIu20__remove_reference_tITL0__EEEsr3stdE16is_convertible_vISA_NSt3__18functionIFS7_NDtL_ZNS_9SimpleLUT5tableEEE9size_typeESH_EEEEEES7_ENKUlvE_clEv: 55| 544k| [&f, table_ = &table]() { 56| | // which row [0..fullTableSize) are we filling? 57| 544k| const auto i = table_->size(); 58| 544k| return f(i, fullTableSize); 59| 544k| }); _ZNK8rawspeed9SimpleLUTIaLi12EEixEi: 63| 1.45M| value_type operator[](int x) const { 64| 1.45M| unsigned clampedX = clampBits(x, TableBitWidth); 65| 1.45M| return table[clampedX]; 66| 1.45M| } _ZN8rawspeed6SplineItEC2ERKNSt3__16vectorINS_8iPoint2DENS2_9allocatorIS4_EEEE: 113| 88| explicit Spline(const std::vector& control_points) { 114| 88| assert(control_points.size() >= 2 && ------------------ | Branch (114:5): [True: 88, False: 0] | Branch (114:5): [True: 88, Folded] | Branch (114:5): [True: 88, False: 0] ------------------ 115| 88| "Need at least two points to interpolate between"); 116| | 117| | // Expect the X coords of the curve to start/end at the extreme values 118| 88| assert(control_points.front().x == 0); ------------------ | Branch (118:5): [True: 88, False: 0] ------------------ 119| 88| assert(control_points.back().x == 65535); ------------------ | Branch (119:5): [True: 88, False: 0] ------------------ 120| | 121| 88| assert(std::adjacent_find( ------------------ | Branch (121:5): [True: 88, False: 0] | Branch (121:5): [True: 88, Folded] | Branch (121:5): [True: 88, False: 0] ------------------ 122| 88| control_points.cbegin(), control_points.cend(), 123| 88| [](const iPoint2D& lhs, const iPoint2D& rhs) -> bool { 124| 88| return std::greater_equal<>()(lhs.x, rhs.x); 125| 88| }) == control_points.cend() && 126| 88| "The X coordinates must all be strictly increasing"); 127| | 128| 88|#ifndef NDEBUG 129| 88| if constexpr (!std::is_floating_point_v) { 130| | // The Y coords must be limited to the range of value_type 131| 88| std::for_each(control_points.cbegin(), control_points.cend(), 132| 88| [](const iPoint2D& p) { 133| 88| assert(p.y >= std::numeric_limits::min()); 134| 88| assert(p.y <= std::numeric_limits::max()); 135| 88| }); 136| 88| } 137| 88|#endif 138| | 139| 88| num_coords = implicit_cast(control_points.size()); 140| 88| num_segments = num_coords - 1; 141| | 142| 88| xCp.resize(num_coords); 143| 88| segments.resize(num_coords); 144| 880| for (int i = 0; i < num_coords; i++) { ------------------ | Branch (144:21): [True: 792, False: 88] ------------------ 145| 792| xCp[i] = control_points[i].x; 146| 792| segments[i].a = control_points[i].y; 147| 792| } 148| | 149| 88| prepare(); 150| 88| } _ZZN8rawspeed6SplineItEC1ERKNSt3__16vectorINS_8iPoint2DENS2_9allocatorIS4_EEEEENKUlRKS4_E_clESB_: 132| 792| [](const iPoint2D& p) { 133| 792| assert(p.y >= std::numeric_limits::min()); ------------------ | Branch (133:23): [True: 792, False: 0] ------------------ 134| 792| assert(p.y <= std::numeric_limits::max()); ------------------ | Branch (134:23): [True: 792, False: 0] ------------------ 135| 792| }); _ZN8rawspeed6SplineItE7prepareEv: 67| 88| void prepare() { 68| | // Extra values used during computation 69| 88| std::vector h(num_segments); 70| 88| std::vector alpha(num_segments); 71| 88| std::vector mu(num_coords); 72| 88| std::vector z(num_coords); 73| | 74| 792| for (int i = 0; i < num_segments; i++) ------------------ | Branch (74:21): [True: 704, False: 88] ------------------ 75| 704| h[i] = xCp[i + 1] - xCp[i]; 76| | 77| 704| for (int i = 1; i < num_segments; i++) { ------------------ | Branch (77:21): [True: 616, False: 88] ------------------ 78| 616| Segment& sp = segments[i - 1]; 79| 616| Segment& s = segments[i]; 80| 616| Segment& sn = segments[i + 1]; 81| | 82| 616| alpha[i] = (3. / h[i]) * (sn.a - s.a) - (3. / h[i - 1]) * (s.a - sp.a); 83| 616| } 84| | 85| 88| mu[0] = z[0] = 0; 86| | 87| 704| for (int i = 1; i < num_segments; i++) { ------------------ | Branch (87:21): [True: 616, False: 88] ------------------ 88| 616| const double l = (2 * (xCp[i + 1] - xCp[i - 1])) - (h[i - 1] * mu[i - 1]); 89| 616| mu[i] = h[i] / l; 90| 616| z[i] = (alpha[i] - h[i - 1] * z[i - 1]) / l; 91| 616| } 92| | 93| 88| z.back() = segments.back().c = 0; 94| | 95| 792| for (int i = num_segments - 1; i >= 0; i--) { ------------------ | Branch (95:36): [True: 704, False: 88] ------------------ 96| 704| Segment& s = segments[i]; 97| 704| Segment& sn = segments[i + 1]; 98| | 99| 704| s.c = z[i] - mu[i] * sn.c; 100| 704| s.b = (sn.a - s.a) / h[i] - h[i] * (sn.c + 2 * s.c) / 3.; 101| 704| s.d = (sn.c - s.c) / (3. * h[i]); 102| 704| } 103| | 104| | // The last segment is nonsensical, and was only used to temporarily store 105| | // the a and c to simplify calculations, so drop that 'segment' now 106| 88| segments.pop_back(); 107| | 108| | assert(static_cast(num_segments) == ------------------ | Branch (108:5): [True: 88, False: 0] ------------------ 109| 88| segments.size()); 110| 88| } _ZNK8rawspeed6SplineItE14calculateCurveEv: 154| 88| [[nodiscard]] std::vector calculateCurve() const { 155| 88| std::vector curve(65536); 156| | 157| 792| for (int i = 0; i < num_segments; i++) { ------------------ | Branch (157:21): [True: 704, False: 88] ------------------ 158| 704| const Segment& s = segments[i]; 159| | 160| 5.76M| for (int x = xCp[i]; x <= xCp[i + 1]; x++) { ------------------ | Branch (160:28): [True: 5.76M, False: 704] ------------------ 161| 5.76M| double diff = x - xCp[i]; 162| 5.76M| double diff_2 = diff * diff; 163| 5.76M| double diff_3 = diff * diff * diff; 164| | 165| 5.76M| double interpolated = 166| 5.76M| s.a + (s.b * diff) + (s.c * diff_2) + (s.d * diff_3); 167| | 168| 5.76M| if constexpr (!std::is_floating_point_v) { 169| 5.76M| interpolated = std::max( 170| 5.76M| interpolated, double(std::numeric_limits::min())); 171| 5.76M| interpolated = std::min( 172| 5.76M| interpolated, double(std::numeric_limits::max())); 173| 5.76M| } 174| | 175| 5.76M| curve[x] = implicit_cast(interpolated); 176| 5.76M| } 177| 704| } 178| | 179| 88| return curve; 180| 88| } _ZN8rawspeed11TableLookUpC2Eib: 43| 659| : ntables(_ntables), dither(_dither) { 44| 659| if (ntables < 1) { ------------------ | Branch (44:7): [True: 0, False: 659] ------------------ 45| 0| ThrowRDE("Cannot construct 0 tables"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 46| 0| } 47| 659| tables.resize(ntables * TABLE_SIZE, uint16_t(0)); 48| 659|} _ZN8rawspeed11TableLookUp8setTableEiRKNSt3__16vectorItNS1_9allocatorItEEEE: 50| 659|void TableLookUp::setTable(int ntable, const std::vector& table) { 51| 659| assert(!table.empty()); ------------------ | Branch (51:3): [True: 659, False: 0] ------------------ 52| | 53| 659| const auto nfilled = implicit_cast(table.size()); 54| 659| if (nfilled > TABLE_MAX_ELTS) ------------------ | Branch (54:7): [True: 1, False: 658] ------------------ 55| 659| ThrowRDE("Table lookup with %i entries is unsupported", nfilled); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 56| | 57| 658| if (ntable > ntables) { ------------------ | Branch (57:7): [True: 0, False: 658] ------------------ 58| 0| ThrowRDE("Table lookup with number greater than number of tables."); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 59| 0| } 60| | 61| 658| auto t = Array2DRef(tables.data(), TABLE_SIZE, ntables)[ntable]; 62| 658| if (!dither) { ------------------ | Branch (62:7): [True: 0, False: 658] ------------------ 63| 0| for (int i = 0; i < TABLE_MAX_ELTS; i++) { ------------------ | Branch (63:21): [True: 0, False: 0] ------------------ 64| 0| t(i) = (i < nfilled) ? table[i] : table[nfilled - 1]; ------------------ | Branch (64:14): [True: 0, False: 0] ------------------ 65| 0| } 66| 0| return; 67| 0| } 68| 4.79M| for (int i = 0; i < nfilled; i++) { ------------------ | Branch (68:19): [True: 4.79M, False: 658] ------------------ 69| 4.79M| int center = table[i]; 70| 4.79M| int lower = i > 0 ? table[i - 1] : center; ------------------ | Branch (70:17): [True: 4.79M, False: 658] ------------------ 71| 4.79M| int upper = i < (nfilled - 1) ? table[i + 1] : center; ------------------ | Branch (71:17): [True: 4.79M, False: 658] ------------------ 72| | // Non-monotonic LUT handling: don't interpolate across the cross-over. 73| 4.79M| lower = std::min(lower, center); 74| 4.79M| upper = std::max(upper, center); 75| 4.79M| int delta = upper - lower; 76| 4.79M| invariant(delta >= 0); ------------------ | | 27| 4.79M|#define invariant(expr) assert(expr) ------------------ | Branch (76:5): [True: 4.79M, False: 0] ------------------ 77| 4.79M| t(i * 2) = clampBits(center - ((upper - lower + 2) / 4), 16); 78| 4.79M| t((i * 2) + 1) = implicit_cast(delta); 79| 4.79M| } 80| | 81| 38.3M| for (int i = nfilled; i < TABLE_MAX_ELTS; i++) { ------------------ | Branch (81:25): [True: 38.3M, False: 658] ------------------ 82| 38.3M| t(i * 2) = table[nfilled - 1]; 83| 38.3M| t((i * 2) + 1) = 0; 84| 38.3M| } 85| 658|} _ZN8rawspeed11TableLookUp8getTableEi: 87| 1.83k|Array1DRef TableLookUp::getTable(int n) { 88| 1.83k| if (n > ntables) { ------------------ | Branch (88:7): [True: 0, False: 1.83k] ------------------ 89| 0| ThrowRDE("Table lookup with number greater than number of tables."); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 90| 0| } 91| 1.83k| return Array2DRef(tables.data(), TABLE_SIZE, ntables)[n]; 92| 1.83k|} _ZNK8rawspeed19AbstractTiffDecoder22getIFDWithLargestImageENS_7TiffTagE: 37| 1.27k|AbstractTiffDecoder::getIFDWithLargestImage(TiffTag filter) const { 38| 1.27k| std::vector ifds = mRootIFD->getIFDsWithTag(filter); 39| | 40| 1.27k| if (ifds.empty()) ------------------ | Branch (40:7): [True: 10, False: 1.26k] ------------------ 41| 1.27k| ThrowRDE("No suitable IFD with tag 0x%04x found.", ------------------ | | 38| 10| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 10| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 10| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 10| __VA_ARGS__) | | ------------------ ------------------ 42| 1.26k| static_cast(filter)); 43| | 44| 1.26k| const auto* res = ifds[0]; 45| 1.26k| uint32_t width = res->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 46| 1.53k| for (const auto* ifd : ifds) { ------------------ | Branch (46:24): [True: 1.53k, False: 1.26k] ------------------ 47| 1.53k| const TiffEntry* widthE = ifd->getEntry(TiffTag::IMAGEWIDTH); 48| | // guard against random maker note entries with the same tag 49| 1.53k| if (widthE->count == 1 && widthE->getU32() > width) { ------------------ | Branch (49:9): [True: 881, False: 655] | Branch (49:31): [True: 241, False: 640] ------------------ 50| 241| res = ifd; 51| 241| width = widthE->getU32(); 52| 241| } 53| 1.53k| } 54| | 55| 1.26k| return res; 56| 1.27k|} _ZN8rawspeed19AbstractTiffDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 44| 11.6k| : RawDecoder(file), mRootIFD(std::move(root)) {} _ZN8rawspeed19AbstractTiffDecoder10getRootIFDEv: 46| 226| TiffIFD* getRootIFD() final { return mRootIFD.get(); } _ZN8rawspeed19AbstractTiffDecoder20checkCameraSupportedEPKNS_14CameraMetaDataERKNS_6TiffIDERKNSt3__112basic_stringIcNS7_11char_traitsIcEENS7_9allocatorIcEEEE: 49| 39| const std::string& mode) { 50| 39| return RawDecoder::checkCameraSupported(meta, id.make, id.model, mode); 51| 39| } _ZN8rawspeed19AbstractTiffDecoder11setMetaDataEPKNS_14CameraMetaDataERKNS_6TiffIDERKNSt3__112basic_stringIcNS7_11char_traitsIcEENS7_9allocatorIcEEEEi: 56| 421| const std::string& mode, int iso_speed) { 57| 421| setMetaData(meta, id.make, id.model, mode, iso_speed); 58| 421| } _ZN8rawspeed19AbstractTiffDecoder11setMetaDataEPKNS_14CameraMetaDataERKNSt3__112basic_stringIcNS4_11char_traitsIcEENS4_9allocatorIcEEEEi: 61| 147| int iso_speed) { 62| 147| setMetaData(meta, mRootIFD->getID(), mode, iso_speed); 63| 147| } _ZN8rawspeed10ArwDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 61| 4.15k| [[maybe_unused]] Buffer file) { 62| 4.15k| const auto id = rootIFD->getID(); 63| 4.15k| const std::string& make = id.make; 64| | 65| | // FIXME: magic 66| | 67| 4.15k| return make == "SONY"; 68| 4.15k|} _ZN8rawspeed10ArwDecoder21decodeTransitionalArwEv: 120| 815|RawImage ArwDecoder::decodeTransitionalArw() { 121| 815| if (const TiffEntry* model = mRootIFD->getEntryRecursive(TiffTag::MODEL); 122| 815| model && model->getString() == "DSLR-A100") { ------------------ | Branch (122:7): [True: 815, False: 0] | Branch (122:7): [True: 794, False: 21] | Branch (122:16): [True: 794, False: 21] ------------------ 123| | // We've caught the elusive A100 in the wild, a transitional format 124| | // between the simple sanity of the MRW custom format and the wordly 125| | // wonderfullness of the Tiff-based ARW format, let's shoot from the hip 126| 794| const TiffIFD* raw = mRootIFD->getIFDWithTag(TiffTag::SUBIFDS); 127| 794| uint32_t off = raw->getEntry(TiffTag::SUBIFDS)->getU32(); 128| 794| uint32_t width = 3881; 129| 794| uint32_t height = 2608; 130| | 131| 794| mRaw->dim = iPoint2D(width, height); 132| | 133| 794| ByteStream input(DataBuffer(mFile.getSubView(off), Endianness::little)); 134| 794| SonyArw1Decompressor a(mRaw); 135| 794| mRaw->createData(); 136| 794| a.decompress(input); 137| | 138| 794| return mRaw; 139| 794| } 140| | 141| 21| if (hints.contains("srf_format")) ------------------ | Branch (141:7): [True: 0, False: 21] ------------------ 142| 0| return decodeSRF(); 143| | 144| 21| ThrowRDE("No image data found"); ------------------ | | 38| 21| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 21| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 21| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 21| __VA_ARGS__) | | ------------------ ------------------ 145| 21|} _ZN8rawspeed10ArwDecoder11decodeCurveEPKNS_7TiffIFDE: 147| 213|std::vector ArwDecoder::decodeCurve(const TiffIFD* raw) { 148| 213| std::vector curve(0x4001); 149| 213| const TiffEntry* c = raw->getEntry(TiffTag::SONYCURVE); 150| 213| std::array sony_curve = {{0, 0, 0, 0, 0, 4095}}; 151| | 152| 1.03k| for (uint32_t i = 0; i < 4; i++) ------------------ | Branch (152:24): [True: 817, False: 213] ------------------ 153| 817| sony_curve[i + 1] = (c->getU16(i) >> 2) & 0xfff; 154| | 155| 3.34M| for (uint32_t i = 0; i < 0x4001; i++) ------------------ | Branch (155:24): [True: 3.34M, False: 213] ------------------ 156| 3.34M| curve[i] = implicit_cast(i); 157| | 158| 1.23k| for (uint32_t i = 0; i < 5; i++) ------------------ | Branch (158:24): [True: 1.02k, False: 213] ------------------ 159| 1.03M| for (uint32_t j = sony_curve[i] + 1; j <= sony_curve[i + 1]; j++) ------------------ | Branch (159:42): [True: 1.03M, False: 1.02k] ------------------ 160| 1.03M| curve[j] = implicit_cast(curve[j - 1] + (1 << i)); 161| | 162| 213| return curve; 163| 213|} _ZN8rawspeed10ArwDecoder17decodeRawInternalEv: 165| 1.32k|RawImage ArwDecoder::decodeRawInternal() { 166| 1.32k| vector data = mRootIFD->getIFDsWithTag(TiffTag::STRIPOFFSETS); 167| | 168| 1.32k| if (data.empty()) ------------------ | Branch (168:7): [True: 815, False: 513] ------------------ 169| 815| return decodeTransitionalArw(); 170| | 171| 513| const TiffIFD* raw = data[0]; 172| 513| int compression = raw->getEntry(TiffTag::COMPRESSION)->getU32(); 173| 513| if (1 == compression) { ------------------ | Branch (173:7): [True: 152, False: 361] ------------------ 174| 152| DecodeUncompressed(raw); 175| 152| return mRaw; 176| 152| } 177| | 178| 361| if (7 == compression) { ------------------ | Branch (178:7): [True: 45, False: 316] ------------------ 179| 45| DecodeLJpeg(raw); 180| | // cropping of lossless compressed L files already done in Ljpeg decoder 181| 45| applyCrop = false; 182| 45| return mRaw; 183| 45| } 184| | 185| 316| if (32767 != compression) ------------------ | Branch (185:7): [True: 39, False: 277] ------------------ 186| 316| ThrowRDE("Unsupported compression %i", compression); ------------------ | | 38| 39| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 39| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 39| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 39| __VA_ARGS__) | | ------------------ ------------------ 187| | 188| 277| const TiffEntry* offsets = raw->getEntry(TiffTag::STRIPOFFSETS); 189| 277| const TiffEntry* counts = raw->getEntry(TiffTag::STRIPBYTECOUNTS); 190| | 191| 277| if (offsets->count != 1) { ------------------ | Branch (191:7): [True: 1, False: 276] ------------------ 192| 1| ThrowRDE("Multiple Strips found: %u", offsets->count); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 193| 1| } 194| 276| if (counts->count != offsets->count) { ------------------ | Branch (194:7): [True: 2, False: 274] ------------------ 195| 2| ThrowRDE( ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 196| 2| "Byte count number does not match strip size: count:%u, strips:%u ", 197| 2| counts->count, offsets->count); 198| 2| } 199| 274| uint32_t width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 200| 274| uint32_t height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 201| 274| uint32_t bitPerPixel = raw->getEntry(TiffTag::BITSPERSAMPLE)->getU32(); 202| | 203| 274| switch (bitPerPixel) { 204| 218| case 8: ------------------ | Branch (204:3): [True: 218, False: 56] ------------------ 205| 265| case 12: ------------------ | Branch (205:3): [True: 47, False: 227] ------------------ 206| 270| case 14: ------------------ | Branch (206:3): [True: 5, False: 269] ------------------ 207| 270| break; 208| 1| default: ------------------ | Branch (208:3): [True: 1, False: 273] ------------------ 209| 1| ThrowRDE("Unexpected bits per pixel: %u", bitPerPixel); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 210| 274| } 211| | 212| | // Sony E-550 marks compressed 8bpp ARW with 12 bit per pixel 213| | // this makes the compression detect it as a ARW v1. 214| | // This camera has however another MAKER entry, so we MAY be able 215| | // to detect it this way in the future. 216| 270| data = mRootIFD->getIFDsWithTag(TiffTag::MAKE); 217| 270| if (data.size() > 1) { ------------------ | Branch (217:7): [True: 13, False: 257] ------------------ 218| 36| for (auto& i : data) { ------------------ | Branch (218:18): [True: 36, False: 13] ------------------ 219| 36| std::string make = i->getEntry(TiffTag::MAKE)->getString(); 220| | /* Check for maker "SONY" without spaces */ 221| 36| if (make == "SONY") ------------------ | Branch (221:11): [True: 19, False: 17] ------------------ 222| 19| bitPerPixel = 8; 223| 36| } 224| 13| } 225| | 226| 270| if (width == 0 || height == 0 || height % 2 != 0 || width > 9600 || ------------------ | Branch (226:7): [True: 3, False: 267] | Branch (226:21): [True: 1, False: 266] | Branch (226:36): [True: 2, False: 264] | Branch (226:55): [True: 25, False: 239] ------------------ 227| 239| height > 6376) ------------------ | Branch (227:7): [True: 25, False: 214] ------------------ 228| 270| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 55| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 55| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 55| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 55| __VA_ARGS__) | | ------------------ ------------------ 229| | 230| 215| bool arw1 = uint64_t(counts->getU32()) * 8 != width * height * bitPerPixel; 231| 215| if (arw1) ------------------ | Branch (231:7): [True: 108, False: 107] ------------------ 232| 108| height += 8; 233| | 234| 215| mRaw->dim = iPoint2D(width, height); 235| | 236| 215| std::vector curve = decodeCurve(raw); 237| | 238| 215| RawImageCurveGuard curveHandler(&mRaw, curve, uncorrectedRawValues); 239| | 240| 215| uint32_t c2 = counts->getU32(); 241| 215| uint32_t off = offsets->getU32(); 242| | 243| 215| if (!mFile.isValid(off)) ------------------ | Branch (243:7): [True: 30, False: 185] ------------------ 244| 215| ThrowRDE("Data offset after EOF, file probably truncated"); ------------------ | | 38| 30| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 30| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 30| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 30| __VA_ARGS__) | | ------------------ ------------------ 245| | 246| 185| if (!mFile.isValid(off, c2)) ------------------ | Branch (246:7): [True: 84, False: 101] ------------------ 247| 84| c2 = mFile.getSize() - off; 248| | 249| 185| ByteStream input(DataBuffer(mFile.getSubView(off, c2), Endianness::little)); 250| | 251| 185| if (arw1) { ------------------ | Branch (251:7): [True: 81, False: 104] ------------------ 252| 81| SonyArw1Decompressor a(mRaw); 253| 81| mRaw->createData(); 254| 81| a.decompress(input); 255| 81| mShiftDownScaleForExif = 2; 256| 104| } else { 257| 104| DecodeARW2(input, width, height, bitPerPixel); 258| 104| } 259| | 260| 185| if (bitPerPixel == 12) ------------------ | Branch (260:7): [True: 2, False: 183] ------------------ 261| 2| mShiftDownScaleForExif = 2; 262| | 263| 185| return mRaw; 264| 215|} _ZNK8rawspeed10ArwDecoder18DecodeUncompressedEPKNS_7TiffIFDE: 266| 152|void ArwDecoder::DecodeUncompressed(const TiffIFD* raw) const { 267| 152| uint32_t width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 268| 152| uint32_t height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 269| 152| uint32_t off = raw->getEntry(TiffTag::STRIPOFFSETS)->getU32(); 270| 152| uint32_t c2 = raw->getEntry(TiffTag::STRIPBYTECOUNTS)->getU32(); 271| | 272| 152| mRaw->dim = iPoint2D(width, height); 273| | 274| 152| if (width == 0 || height == 0 || width > 9600 || height > 6376) ------------------ | Branch (274:7): [True: 2, False: 150] | Branch (274:21): [True: 1, False: 149] | Branch (274:36): [True: 28, False: 121] | Branch (274:52): [True: 17, False: 104] ------------------ 275| 152| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 47| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 47| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 47| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 47| __VA_ARGS__) | | ------------------ ------------------ 276| | 277| 105| if (c2 == 0) ------------------ | Branch (277:7): [True: 1, False: 104] ------------------ 278| 105| ThrowRDE("Strip is empty, nothing to decode!"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 279| | 280| 104| const Buffer buf(mFile.getSubView(off, c2)); 281| | 282| 104| if (hints.contains("sr2_format")) { ------------------ | Branch (282:7): [True: 0, False: 104] ------------------ 283| 0| UncompressedDecompressor u(ByteStream(DataBuffer(buf, Endianness::little)), 284| 0| mRaw, 285| 0| iRectangle2D({0, 0}, iPoint2D(width, height)), 286| 0| 2 * width, 16, BitOrder::MSB); 287| 0| mRaw->createData(); 288| 0| u.readUncompressedRaw(); 289| 104| } else { 290| 104| UncompressedDecompressor u(ByteStream(DataBuffer(buf, Endianness::little)), 291| 104| mRaw, 292| 104| iRectangle2D({0, 0}, iPoint2D(width, height)), 293| 104| 2 * width, 16, BitOrder::LSB); 294| 104| mRaw->createData(); 295| 104| u.readUncompressedRaw(); 296| 104| } 297| 104|} _ZN8rawspeed10ArwDecoder11DecodeLJpegEPKNS_7TiffIFDE: 299| 45|void ArwDecoder::DecodeLJpeg(const TiffIFD* raw) { 300| 45| uint32_t width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 301| 45| uint32_t height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 302| 45| uint32_t bitPerPixel = raw->getEntry(TiffTag::BITSPERSAMPLE)->getU32(); 303| 45| uint32_t photometric = 304| 45| raw->getEntry(TiffTag::PHOTOMETRICINTERPRETATION)->getU32(); 305| | 306| 45| if (photometric != 32803) ------------------ | Branch (306:7): [True: 5, False: 40] ------------------ 307| 45| ThrowRDE("Unsupported photometric interpretation: %u", photometric); ------------------ | | 38| 5| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 5| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 5| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 5| __VA_ARGS__) | | ------------------ ------------------ 308| | 309| 40| switch (bitPerPixel) { 310| 16| case 8: ------------------ | Branch (310:3): [True: 16, False: 24] ------------------ 311| 31| case 12: ------------------ | Branch (311:3): [True: 15, False: 25] ------------------ 312| 38| case 14: ------------------ | Branch (312:3): [True: 7, False: 33] ------------------ 313| 38| break; 314| 1| default: ------------------ | Branch (314:3): [True: 1, False: 39] ------------------ 315| 1| ThrowRDE("Unexpected bits per pixel: %u", bitPerPixel); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 316| 40| } 317| | 318| 38| if (width == 0 || height == 0 || width % 2 != 0 || height % 2 != 0 || ------------------ | Branch (318:7): [True: 1, False: 37] | Branch (318:21): [True: 1, False: 36] | Branch (318:36): [True: 1, False: 35] | Branch (318:54): [True: 1, False: 34] ------------------ 319| 34| width > 9728 || height > 6656) ------------------ | Branch (319:7): [True: 10, False: 24] | Branch (319:23): [True: 16, False: 8] ------------------ 320| 38| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 30| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 30| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 30| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 30| __VA_ARGS__) | | ------------------ ------------------ 321| | 322| 8| mRaw->dim = iPoint2D(width, height); 323| | 324| 8| auto tilew = uint64_t(raw->getEntry(TiffTag::TILEWIDTH)->getU32()); 325| 8| uint32_t tileh = raw->getEntry(TiffTag::TILELENGTH)->getU32(); 326| | 327| 8| if (tilew <= 0 || tileh <= 0 || tileh % 2 != 0) ------------------ | Branch (327:7): [True: 3, False: 5] | Branch (327:21): [True: 1, False: 4] | Branch (327:35): [True: 2, False: 2] ------------------ 328| 8| ThrowRDE("Invalid tile size: (%" PRIu64 ", %u)", tilew, tileh); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 329| | 330| 8| assert(tilew > 0); ------------------ | Branch (330:3): [True: 2, False: 2] ------------------ 331| 2| const auto tilesX = 332| 2| implicit_cast(roundUpDivisionSafe(mRaw->dim.x, tilew)); 333| 2| if (!tilesX) ------------------ | Branch (333:7): [True: 0, False: 2] ------------------ 334| 2| ThrowRDE("Zero tiles horizontally"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 335| | 336| 2| assert(tileh > 0); ------------------ | Branch (336:3): [True: 2, False: 0] ------------------ 337| 2| const auto tilesY = 338| 2| implicit_cast(roundUpDivisionSafe(mRaw->dim.y, tileh)); 339| 2| if (!tilesY) ------------------ | Branch (339:7): [True: 0, False: 2] ------------------ 340| 2| ThrowRDE("Zero tiles vertically"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 341| | 342| | // Math thoughs: if we know that the total size is 100, while tile size is 11, 343| | // we end up with 9 full tiles, and 1 partial tile (10 total). 344| | // 345| | // BUT! If we know that the total size is 100, and we have same 10 tiles, 346| | // we'd naively guess that each tile's size is 10, and not 11... 347| | 348| 2| const TiffEntry* offsets = raw->getEntry(TiffTag::TILEOFFSETS); 349| 2| const TiffEntry* counts = raw->getEntry(TiffTag::TILEBYTECOUNTS); 350| 2| if (offsets->count != counts->count) { ------------------ | Branch (350:7): [True: 0, False: 2] ------------------ 351| 0| ThrowRDE("Tile count mismatch: offsets:%u count:%u", offsets->count, ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 352| 0| counts->count); 353| 0| } 354| | 355| | // tilesX * tilesY may overflow, but division is fine, so let's do that. 356| 2| if ((offsets->count / tilesX != tilesY || (offsets->count % tilesX != 0)) || ------------------ | Branch (356:8): [True: 2, False: 0] | Branch (356:45): [True: 0, False: 0] ------------------ 357| 0| (offsets->count / tilesY != tilesX || (offsets->count % tilesY != 0))) { ------------------ | Branch (357:8): [True: 0, False: 0] | Branch (357:45): [True: 0, False: 0] ------------------ 358| 0| ThrowRDE("Tile X/Y count mismatch: total:%u X:%u, Y:%u", offsets->count, ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 359| 0| tilesX, tilesY); 360| 0| } 361| | 362| 2| NORangesSet tilesLegality; 363| 2| for (int tile = 0U; tile < implicit_cast(offsets->count); tile++) { ------------------ | Branch (363:23): [True: 0, False: 2] ------------------ 364| 0| const uint32_t offset = offsets->getU32(tile); 365| 0| const uint32_t length = counts->getU32(tile); 366| 0| if (!tilesLegality.insert(mFile.getSubView(offset, length))) ------------------ | Branch (366:9): [True: 0, False: 0] ------------------ 367| 0| ThrowRDE("Two tiles overlap. Raw corrupt!"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 368| 0| } 369| | 370| 2| mRaw->createData(); 371| 2|#ifdef HAVE_OPENMP 372| 2|#pragma omp parallel for schedule(static) default(none) \ 373| 2| shared(offsets, counts) firstprivate(tilesX, tilew, tileh) 374| 2|#endif 375| 2| for (int tile = 0U; tile < static_cast(offsets->count); tile++) { 376| 2| try { 377| 2| const uint32_t tileX = tile % tilesX; 378| 2| const uint32_t tileY = tile / tilesX; 379| 2| const uint32_t offset = offsets->getU32(tile); 380| 2| const uint32_t length = counts->getU32(tile); 381| | 382| 2| LJpegDecoder decoder( 383| 2| ByteStream( 384| 2| DataBuffer(mFile.getSubView(offset, length), Endianness::little)), 385| 2| mRaw); 386| 2| auto offsetX = implicit_cast(tileX * tilew); 387| 2| auto offsetY = tileY * tileh; 388| 2| auto tileWidth = implicit_cast(tilew); 389| 2| auto tileHeight = tileh; 390| 2| auto maxDim = iPoint2D{implicit_cast(tileWidth), 391| 2| implicit_cast(tileHeight)}; 392| 2| decoder.decode(offsetX, offsetY, tileWidth, tileHeight, maxDim, 393| 2| /*fixDng16Bug=*/false); 394| 2| } catch (const RawDecoderException& err) { 395| 2| mRaw->setError(err.what()); 396| 2| } catch (const IOException& err) { 397| 2| mRaw->setError(err.what()); 398| 2| } catch (...) { 399| | // We should not get any other exception type here. 400| 2| __builtin_unreachable(); 401| 2| } 402| 2| } 403| | 404| 2| std::string firstErr; 405| 2| if (mRaw->isTooManyErrors(1, &firstErr)) { ------------------ | Branch (405:7): [True: 0, False: 2] ------------------ 406| 0| ThrowRDE("Too many errors encountered. Giving up. First Error:\n%s", ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 407| 0| firstErr.c_str()); 408| 0| } 409| | 410| 2| const TiffEntry* size_entry = raw->getEntry(TiffTag::SONYRAWIMAGESIZE); 411| 2| iRectangle2D crop(0, 0, size_entry->getU32(0), size_entry->getU32(1)); 412| 2| mRaw->subFrame(crop); 413| 2|} _ZN8rawspeed10ArwDecoder10DecodeARW2ENS_10ByteStreamEjjj: 416| 93| uint32_t bpp) { 417| | 418| 93| if (bpp == 8) { ------------------ | Branch (418:7): [True: 92, False: 1] ------------------ 419| 92| SonyArw2Decompressor a2(mRaw, input); 420| 92| mRaw->createData(); 421| 92| a2.decompress(); 422| 92| return; 423| 92| } // End bpp = 8 424| | 425| 1| if (bpp == 12) { ------------------ | Branch (425:7): [True: 1, False: 0] ------------------ 426| 1| input.setByteOrder(Endianness::little); 427| 1| UncompressedDecompressor u(input, mRaw, 428| 1| iRectangle2D({0, 0}, iPoint2D(w, h)), 429| 1| bpp * w / 8, bpp, BitOrder::LSB); 430| 1| mRaw->createData(); 431| 1| u.readUncompressedRaw(); 432| | 433| | // Shift scales, since black and white are the same as compressed precision 434| 1| mShiftDownScale = 2; 435| 1| return; 436| 1| } 437| 0| ThrowRDE("Unsupported bit depth"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 438| 1|} _ZNK8rawspeed10ArwDecoder11ParseA100WBEv: 440| 24|void ArwDecoder::ParseA100WB() const { 441| 24| if (!mRootIFD->hasEntryRecursive(TiffTag::DNGPRIVATEDATA)) ------------------ | Branch (441:7): [True: 1, False: 23] ------------------ 442| 1| return; 443| | 444| | // only contains the offset, not the length! 445| 23| const TiffEntry* priv = mRootIFD->getEntryRecursive(TiffTag::DNGPRIVATEDATA); 446| 23| ByteStream bs = priv->getData(); 447| 23| bs.setByteOrder(Endianness::little); 448| 23| const uint32_t off = bs.getU32(); 449| | 450| 23| bs = ByteStream(DataBuffer(mFile.getSubView(off), Endianness::little)); 451| | 452| | // MRW style, see MrwDecoder 453| | 454| 23| bs.setByteOrder(Endianness::big); 455| 23| uint32_t tag = bs.getU32(); 456| 23| if (0x4D5249 != tag) // MRI ------------------ | Branch (456:7): [True: 7, False: 16] ------------------ 457| 23| ThrowRDE("Can not parse DNGPRIVATEDATA, invalid tag (0x%x).", tag); ------------------ | | 38| 7| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 7| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 7| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 7| __VA_ARGS__) | | ------------------ ------------------ 458| | 459| 16| bs.setByteOrder(Endianness::little); 460| 16| uint32_t len = bs.getU32(); 461| | 462| 16| bs = bs.getSubStream(bs.getPosition(), len); 463| | 464| 106| while (bs.getRemainSize() > 0) { ------------------ | Branch (464:10): [True: 104, False: 2] ------------------ 465| 104| bs.setByteOrder(Endianness::big); 466| 104| tag = bs.getU32(); 467| 104| bs.setByteOrder(Endianness::little); 468| 104| len = bs.getU32(); 469| 104| (void)bs.check(len); 470| 104| if (!len) ------------------ | Branch (470:9): [True: 2, False: 102] ------------------ 471| 104| ThrowRDE("Found entry of zero length, corrupt."); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 472| | 473| 102| if (0x574247 != tag) { // WBG ------------------ | Branch (473:9): [True: 90, False: 12] ------------------ 474| | // not the tag we are interested in, skip 475| 90| bs.skipBytes(len); 476| 90| continue; 477| 90| } 478| | 479| 12| bs.skipBytes(4); 480| | 481| 12| bs.setByteOrder(Endianness::little); 482| 12| std::array tmp; 483| 12| for (auto& coeff : tmp) ------------------ | Branch (483:22): [True: 4, False: 12] ------------------ 484| 4| coeff = bs.getU16(); 485| | 486| 12| std::array wbCoeffs = {}; 487| 12| wbCoeffs[0] = static_cast(tmp[0]); 488| 12| wbCoeffs[1] = static_cast(tmp[1]); 489| 12| wbCoeffs[2] = static_cast(tmp[3]); 490| 12| mRaw->metadata.wbCoeffs = wbCoeffs; 491| | 492| | // only need this one block, no need to process any further 493| 12| break; 494| 102| } 495| 16|} _ZN8rawspeed10ArwDecoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 497| 95|void ArwDecoder::decodeMetaDataInternal(const CameraMetaData* meta) { 498| | // Default 499| 95| int iso = 0; 500| | 501| 95| mRaw->cfa.setCFA(iPoint2D(2, 2), CFAColor::RED, CFAColor::GREEN, 502| 95| CFAColor::GREEN, CFAColor::BLUE); 503| | 504| 95| if (mRootIFD->hasEntryRecursive(TiffTag::ISOSPEEDRATINGS)) ------------------ | Branch (504:7): [True: 1, False: 94] ------------------ 505| 1| iso = mRootIFD->getEntryRecursive(TiffTag::ISOSPEEDRATINGS)->getU32(); 506| | 507| 95| auto id = mRootIFD->getID(); 508| | 509| 95| setMetaData(meta, id, "", iso); 510| 95| if (mRaw->whitePoint) ------------------ | Branch (510:7): [True: 0, False: 95] ------------------ 511| 0| mRaw->whitePoint = *mRaw->whitePoint >> mShiftDownScale; 512| 95| mRaw->blackLevel >>= mShiftDownScale; 513| | 514| | // Set the whitebalance 515| 95| try { 516| 95| if (id.model == "DSLR-A100") { // Handle the MRW style WB of the A100 ------------------ | Branch (516:9): [True: 24, False: 71] ------------------ 517| 24| ParseA100WB(); 518| 71| } else { // Everything else but the A100 519| 71| GetWB(); 520| 71| } 521| 95| } catch (const RawspeedException& e) { 522| 86| mRaw->setError(e.what()); 523| | // We caught an exception reading WB, just ignore it 524| 86| } 525| 95|} _ZN8rawspeed10ArwDecoder11SonyDecryptENS_10Array1DRefIKhEENS1_IhEEij: 528| 19| Array1DRef obuf, int len, uint32_t key) { 529| 19| invariant(ibuf.size() == obuf.size()); ------------------ | | 27| 19|#define invariant(expr) assert(expr) ------------------ | Branch (529:3): [True: 19, False: 0] ------------------ 530| 19| invariant(ibuf.size() == 4 * len); ------------------ | | 27| 19|#define invariant(expr) assert(expr) ------------------ | Branch (530:3): [True: 19, False: 0] ------------------ 531| 19| invariant(obuf.size() == 4 * len); ------------------ | | 27| 19|#define invariant(expr) assert(expr) ------------------ | Branch (531:3): [True: 19, False: 0] ------------------ 532| | 533| 19| if (0 == len) ------------------ | Branch (533:7): [True: 0, False: 19] ------------------ 534| 0| return; 535| | 536| 19| std::array pad; 537| | 538| | // Initialize the decryption pad from the key 539| 95| for (int p = 0; p < 4; p++) ------------------ | Branch (539:19): [True: 76, False: 19] ------------------ 540| 76| pad[p] = key = uint32_t((key * 48828125UL) + 1UL); 541| 19| pad[3] = pad[3] << 1 | (pad[0] ^ pad[2]) >> 31; 542| 2.35k| for (int p = 4; p < 127; p++) ------------------ | Branch (542:19): [True: 2.33k, False: 19] ------------------ 543| 2.33k| pad[p] = (pad[p - 4] ^ pad[p - 2]) << 1 | (pad[p - 3] ^ pad[p - 1]) >> 31; 544| 2.43k| for (int p = 0; p < 127; p++) ------------------ | Branch (544:19): [True: 2.41k, False: 19] ------------------ 545| 2.41k| pad[p] = getU32BE(&pad[p]); 546| | 547| 19| int p = 127; 548| | // Decrypt the buffer in place using the pad 549| 185k| for (int i = 0; i != len; ++i) { ------------------ | Branch (549:19): [True: 185k, False: 19] ------------------ 550| 185k| pad[p & 127] = pad[(p + 1) & 127] ^ pad[(p + 1 + 64) & 127]; 551| | 552| 185k| uint32_t pv = pad[p & 127]; 553| | 554| 185k| uint32_t bv; 555| 185k| memcpy(&bv, ibuf.getBlock(4, i).begin(), sizeof(uint32_t)); 556| | 557| 185k| bv ^= pv; 558| | 559| 185k| memcpy(obuf.getBlock(4, i).begin(), &bv, sizeof(uint32_t)); 560| | 561| 185k| p++; 562| 185k| } 563| 19|} _ZNK8rawspeed10ArwDecoder5GetWBEv: 565| 70|void ArwDecoder::GetWB() const { 566| | // Set the whitebalance for all the modern ARW formats (everything after A100) 567| 70| if (mRootIFD->hasEntryRecursive(TiffTag::DNGPRIVATEDATA)) { ------------------ | Branch (567:7): [True: 65, False: 5] ------------------ 568| 65| NORangesSet ifds_undecoded; 569| | 570| 65| const TiffEntry* priv = 571| 65| mRootIFD->getEntryRecursive(TiffTag::DNGPRIVATEDATA); 572| 65| TiffRootIFD makerNoteIFD(nullptr, &ifds_undecoded, priv->getRootIfdData(), 573| 65| priv->getU32()); 574| | 575| 65| const TiffEntry* sony_offset = 576| 65| makerNoteIFD.getEntryRecursive(TiffTag::SONYOFFSET); 577| 65| const TiffEntry* sony_length = 578| 65| makerNoteIFD.getEntryRecursive(TiffTag::SONYLENGTH); 579| 65| const TiffEntry* sony_key = 580| 65| makerNoteIFD.getEntryRecursive(TiffTag::SONYKEY); 581| 65| if (!sony_offset || !sony_length || !sony_key || sony_key->count != 4) ------------------ | Branch (581:9): [True: 11, False: 54] | Branch (581:25): [True: 4, False: 50] | Branch (581:41): [True: 1, False: 49] | Branch (581:54): [True: 0, False: 49] ------------------ 582| 65| ThrowRDE("couldn't find the correct metadata for WB decoding"); ------------------ | | 38| 10| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 10| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 10| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 10| __VA_ARGS__) | | ------------------ ------------------ 583| | 584| 65| assert(sony_offset != nullptr); ------------------ | Branch (584:5): [True: 49, False: 6] ------------------ 585| 49| uint32_t off = sony_offset->getU32(); 586| | 587| 49| assert(sony_length != nullptr); ------------------ | Branch (587:5): [True: 48, False: 1] ------------------ 588| | // The Decryption is done in blocks of 4 bytes. 589| 48| auto len = implicit_cast(roundDown(sony_length->getU32(), 4)); 590| 48| if (!len) ------------------ | Branch (590:9): [True: 1, False: 47] ------------------ 591| 48| ThrowRDE("No buffer to decrypt?"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 592| | 593| 48| assert(sony_key != nullptr); ------------------ | Branch (593:5): [True: 47, False: 0] ------------------ 594| 47| uint32_t key = getU32LE(sony_key->getData().getData(4)); 595| | 596| | // "Decrypt" IFD 597| 47| const auto& ifd_crypt = priv->getRootIfdData(); 598| 47| const auto EncryptedBuffer = 599| 47| ifd_crypt.getSubView(off, len).getAsArray1DRef(); 600| | // We do have to prepend 'off' padding, because TIFF uses absolute offsets. 601| 47| const auto DecryptedBufferSize = off + EncryptedBuffer.size(); 602| 47| std::vector DecryptedBuffer(DecryptedBufferSize); 603| | 604| 47| SonyDecrypt(EncryptedBuffer, 605| 47| {&DecryptedBuffer[off], implicit_cast(len)}, len / 4, key); 606| | 607| 47| NORangesSet ifds_decoded; 608| 47| Buffer decIFD(DecryptedBuffer.data(), DecryptedBufferSize); 609| 47| const Buffer Padding(decIFD.getSubView(0, off)); 610| | // The Decrypted Root Ifd can not point to preceding padding buffer. 611| 47| ifds_decoded.insert(Padding); 612| | 613| 47| DataBuffer dbIDD(decIFD, priv->getRootIfdData().getByteOrder()); 614| 47| TiffRootIFD encryptedIFD(nullptr, &ifds_decoded, dbIDD, off); 615| | 616| 47| if (encryptedIFD.hasEntry(TiffTag::SONYGRBGLEVELS)) { ------------------ | Branch (616:9): [True: 0, False: 47] ------------------ 617| 0| const TiffEntry* wb = encryptedIFD.getEntry(TiffTag::SONYGRBGLEVELS); 618| 0| if (wb->count != 4) ------------------ | Branch (618:11): [True: 0, False: 0] ------------------ 619| 0| ThrowRDE("WB has %u entries instead of 4", wb->count); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 620| 0| std::array wbCoeffs = {}; 621| 0| wbCoeffs[0] = wb->getFloat(1); 622| 0| wbCoeffs[1] = wb->getFloat(0); 623| 0| wbCoeffs[2] = wb->getFloat(2); 624| 0| mRaw->metadata.wbCoeffs = wbCoeffs; 625| 47| } else if (encryptedIFD.hasEntry(TiffTag::SONYRGGBLEVELS)) { ------------------ | Branch (625:16): [True: 0, False: 47] ------------------ 626| 0| const TiffEntry* wb = encryptedIFD.getEntry(TiffTag::SONYRGGBLEVELS); 627| 0| if (wb->count != 4) ------------------ | Branch (627:11): [True: 0, False: 0] ------------------ 628| 0| ThrowRDE("WB has %u entries instead of 4", wb->count); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 629| 0| std::array wbCoeffs = {}; 630| 0| wbCoeffs[0] = wb->getFloat(0); 631| 0| wbCoeffs[1] = wb->getFloat(1); 632| 0| wbCoeffs[2] = wb->getFloat(3); 633| 0| mRaw->metadata.wbCoeffs = wbCoeffs; 634| 0| } 635| | 636| 47| if (encryptedIFD.hasEntry(TiffTag::SONYBLACKLEVEL)) { ------------------ | Branch (636:9): [True: 0, False: 47] ------------------ 637| 0| const TiffEntry* bl = encryptedIFD.getEntry(TiffTag::SONYBLACKLEVEL); 638| 0| if (bl->count != 4) ------------------ | Branch (638:11): [True: 0, False: 0] ------------------ 639| 0| ThrowRDE("Black Level has %u entries instead of 4", bl->count); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 640| 0| mRaw->blackLevelSeparate = 641| 0| Array2DRef(mRaw->blackLevelSeparateStorage.data(), 2, 2); 642| 0| auto blackLevelSeparate1D = *mRaw->blackLevelSeparate->getAsArray1DRef(); 643| 0| for (int i = 0; i < 4; ++i) ------------------ | Branch (643:23): [True: 0, False: 0] ------------------ 644| 0| blackLevelSeparate1D(i) = bl->getU16(i) >> mShiftDownScaleForExif; 645| 0| } 646| | 647| 47| if (encryptedIFD.hasEntry(TiffTag::SONYWHITELEVEL)) { ------------------ | Branch (647:9): [True: 0, False: 47] ------------------ 648| 0| const TiffEntry* wl = encryptedIFD.getEntry(TiffTag::SONYWHITELEVEL); 649| 0| if (wl->count != 1 && wl->count != 3) ------------------ | Branch (649:11): [True: 0, False: 0] | Branch (649:29): [True: 0, False: 0] ------------------ 650| 0| ThrowRDE("White Level has %u entries instead of 1 or 3", wl->count); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 651| 0| mRaw->whitePoint = wl->getU16(0) >> mShiftDownScaleForExif; 652| 0| } 653| 47| } 654| 70|} _ZN8rawspeed10ArwDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 43| 1.32k| : AbstractTiffDecoder(std::move(root), file) {} _ZN8rawspeed10Cr2Decoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 58| 6.42k| [[maybe_unused]] Buffer file) { 59| 6.42k| const auto id = rootIFD->getID(); 60| 6.42k| const std::string& make = id.make; 61| 6.42k| const std::string& model = id.model; 62| | 63| | // FIXME: magic 64| | 65| 6.42k| return make == "Canon" || ------------------ | Branch (65:10): [True: 1.07k, False: 5.34k] ------------------ 66| 5.34k| (make == "Kodak" && (model == "DCS520C" || model == "DCS560C")); ------------------ | Branch (66:11): [True: 279, False: 5.06k] | Branch (66:31): [True: 1, False: 278] | Branch (66:53): [True: 0, False: 278] ------------------ 67| 6.42k|} _ZN8rawspeed10Cr2Decoder15decodeOldFormatEv: 69| 959|RawImage Cr2Decoder::decodeOldFormat() { 70| 959| uint32_t offset = 0; 71| 959| if (mRootIFD->getEntryRecursive(TiffTag::CANON_RAW_DATA_OFFSET)) { ------------------ | Branch (71:7): [True: 936, False: 23] ------------------ 72| 936| offset = 73| 936| mRootIFD->getEntryRecursive(TiffTag::CANON_RAW_DATA_OFFSET)->getU32(); 74| 936| } else { 75| | // D2000 is oh so special... 76| 23| const auto* ifd = mRootIFD->getIFDWithTag(TiffTag::CFAPATTERN); 77| 23| if (!ifd->hasEntry(TiffTag::STRIPOFFSETS)) ------------------ | Branch (77:9): [True: 6, False: 17] ------------------ 78| 23| ThrowRDE("Couldn't find offset"); ------------------ | | 38| 6| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 6| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 6| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 6| __VA_ARGS__) | | ------------------ ------------------ 79| | 80| 17| offset = ifd->getEntry(TiffTag::STRIPOFFSETS)->getU32(); 81| 17| } 82| | 83| 953| ByteStream b(DataBuffer(mFile.getSubView(offset), Endianness::big)); 84| 953| b.skipBytes(41); 85| 953| int height = b.getU16(); 86| 953| int width = b.getU16(); 87| | 88| | // some old models (1D/1DS/D2000C) encode two lines as one 89| | // see: FIX_CANON_HALF_HEIGHT_DOUBLE_WIDTH 90| 953| if (width > 2 * height) { ------------------ | Branch (90:7): [True: 751, False: 202] ------------------ 91| 751| height *= 2; 92| 751| width /= 2; 93| 751| } 94| 953| width *= 2; // components 95| | 96| 953| mRaw->dim = {width, height}; 97| | 98| 953| const ByteStream bs(DataBuffer(mFile.getSubView(offset), Endianness::little)); 99| | 100| 953| Cr2LJpegDecoder l(bs, mRaw); 101| 953| mRaw->createData(); 102| | 103| 953| Cr2SliceWidths slicing(/*numSlices=*/1, /*sliceWidth=don't care*/ 0, 104| 953| /*lastSliceWidth=*/implicit_cast(width)); 105| 953| l.decode(slicing); 106| 953| ljpegSamplePrecision = l.getSamplePrecision(); 107| | 108| | // deal with D2000 GrayResponseCurve 109| 953| if (const TiffEntry* curve = 110| 953| mRootIFD->getEntryRecursive(static_cast(0x123)); 111| 953| curve && curve->type == TiffDataType::SHORT && curve->count == 4096) { ------------------ | Branch (111:7): [True: 2, False: 951] | Branch (111:16): [True: 1, False: 1] | Branch (111:54): [True: 0, False: 1] ------------------ 112| 0| auto table = curve->getU16Array(curve->count); 113| 0| RawImageCurveGuard curveHandler(&mRaw, table, uncorrectedRawValues); 114| | 115| | // Apply table 116| 0| if (!uncorrectedRawValues) ------------------ | Branch (116:9): [True: 0, False: 0] ------------------ 117| 0| mRaw->sixteenBitLookup(); 118| 0| } 119| | 120| 953| return mRaw; 121| 959|} _ZN8rawspeed10Cr2Decoder15decodeNewFormatEv: 125| 114|RawImage Cr2Decoder::decodeNewFormat() { 126| 114| const TiffEntry* sensorInfoE = 127| 114| mRootIFD->getEntryRecursive(TiffTag::CANON_SENSOR_INFO); 128| 114| if (!sensorInfoE) ------------------ | Branch (128:7): [True: 2, False: 112] ------------------ 129| 114| ThrowTPE("failed to get SensorInfo from MakerNote"); ------------------ | | 39| 2| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 130| | 131| 114| assert(sensorInfoE != nullptr); ------------------ | Branch (131:3): [True: 112, False: 0] ------------------ 132| | 133| 112| if (isSubSampled() != (getSubSampling() != iPoint2D{1, 1})) ------------------ | Branch (133:7): [True: 2, False: 110] ------------------ 134| 112| ThrowTPE("Subsampling sanity check failed"); ------------------ | | 39| 2| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 135| | 136| 110| mRaw->dim = {sensorInfoE->getU16(1), sensorInfoE->getU16(2)}; 137| 110| mRaw->setCpp(1); 138| 110| mRaw->isCFA = !isSubSampled(); 139| | 140| 110| if (isSubSampled()) { ------------------ | Branch (140:7): [True: 3, False: 107] ------------------ 141| 3| iPoint2D& subSampling = mRaw->metadata.subsampling; 142| 3| subSampling = getSubSampling(); 143| 3| if (subSampling.x <= 1 && subSampling.y <= 1) ------------------ | Branch (143:9): [True: 0, False: 3] | Branch (143:31): [True: 0, False: 0] ------------------ 144| 3| ThrowRDE("RAW is expected to be subsampled, but it's not"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 145| | 146| 3| if (mRaw->dim.x % subSampling.x != 0) ------------------ | Branch (146:9): [True: 1, False: 2] ------------------ 147| 3| ThrowRDE("Raw width is not a multiple of horizontal subsampling factor"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 148| 2| mRaw->dim.x /= subSampling.x; 149| | 150| 2| if (mRaw->dim.y % subSampling.y != 0) ------------------ | Branch (150:9): [True: 1, False: 1] ------------------ 151| 2| ThrowRDE("Raw height is not a multiple of vertical subsampling factor"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 152| 1| mRaw->dim.y /= subSampling.y; 153| | 154| 1| mRaw->dim.x *= 2 + subSampling.x * subSampling.y; 155| 1| } 156| | 157| 108| const TiffIFD* raw = mRootIFD->getSubIFDs()[3].get(); 158| | 159| 108| Cr2SliceWidths slicing; 160| | // there are four cases: 161| | // * there is a tag with three components, 162| | // $ last two components are non-zero: all fine then. 163| | // $ first two components are zero, last component is non-zero 164| | // we let Cr2LJpegDecoder guess it (it'll throw if fails) 165| | // $ else the image is considered corrupt. 166| | // * there is a tag with not three components, the image is considered 167| | // corrupt. $ there is no tag, we let Cr2LJpegDecoder guess it (it'll throw if 168| | // fails) 169| 108| if (const TiffEntry* cr2SliceEntry = 170| 108| raw->getEntryRecursive(TiffTag::CANONCR2SLICE); 171| 108| cr2SliceEntry) { ------------------ | Branch (171:7): [True: 67, False: 41] ------------------ 172| 67| if (cr2SliceEntry->count != 3) { ------------------ | Branch (172:9): [True: 1, False: 66] ------------------ 173| 1| ThrowRDE("Found RawImageSegmentation tag with %u elements, should be 3.", ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 174| 1| cr2SliceEntry->count); 175| 1| } 176| | 177| 66| if (cr2SliceEntry->getU16(1) != 0 && cr2SliceEntry->getU16(2) != 0) { ------------------ | Branch (177:9): [True: 61, False: 5] | Branch (177:42): [True: 59, False: 2] ------------------ 178| | // first component can be either zero or non-zero, don't care 179| 59| slicing = Cr2SliceWidths(/*numSlices=*/1 + cr2SliceEntry->getU16(0), 180| 59| /*sliceWidth=*/cr2SliceEntry->getU16(1), 181| 59| /*lastSliceWidth=*/cr2SliceEntry->getU16(2)); 182| 59| } else if (cr2SliceEntry->getU16(0) == 0 && cr2SliceEntry->getU16(1) == 0 && ------------------ | Branch (182:16): [True: 6, False: 1] | Branch (182:49): [True: 5, False: 1] ------------------ 183| 5| cr2SliceEntry->getU16(2) != 0) { ------------------ | Branch (183:16): [True: 4, False: 1] ------------------ 184| | // PowerShot G16, PowerShot S120, let Cr2LJpegDecoder guess. 185| 4| } else { 186| 3| ThrowRDE("Strange RawImageSegmentation tag: (%d, %d, %d), image corrupt.", ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 187| 3| cr2SliceEntry->getU16(0), cr2SliceEntry->getU16(1), 188| 3| cr2SliceEntry->getU16(2)); 189| 3| } 190| 66| } // EOS 20D, EOS-1D Mark II, let Cr2LJpegDecoder guess. 191| | 192| 104| const uint32_t offset = raw->getEntry(TiffTag::STRIPOFFSETS)->getU32(); 193| 104| const uint32_t count = raw->getEntry(TiffTag::STRIPBYTECOUNTS)->getU32(); 194| | 195| 104| const ByteStream bs( 196| 104| DataBuffer(mFile.getSubView(offset, count), Endianness::little)); 197| | 198| 104| Cr2LJpegDecoder d(bs, mRaw); 199| 104| mRaw->createData(); 200| 104| d.decode(slicing); 201| 104| ljpegSamplePrecision = d.getSamplePrecision(); 202| | 203| 104| assert(getSubSampling() == mRaw->metadata.subsampling); ------------------ | Branch (203:3): [True: 1, False: 103] ------------------ 204| | 205| 1| if (mRaw->metadata.subsampling.x > 1 || mRaw->metadata.subsampling.y > 1) ------------------ | Branch (205:7): [True: 0, False: 1] | Branch (205:43): [True: 0, False: 1] ------------------ 206| 0| sRawInterpolate(); 207| | 208| 1| return mRaw; 209| 104|} _ZN8rawspeed10Cr2Decoder17decodeRawInternalEv: 211| 1.07k|RawImage Cr2Decoder::decodeRawInternal() { 212| 1.07k| if (mRootIFD->getSubIFDs().size() < 4) ------------------ | Branch (212:7): [True: 959, False: 114] ------------------ 213| 959| return decodeOldFormat(); 214| 114| else // NOLINT(readability-else-after-return): ok, here it make sense 215| 114| return decodeNewFormat(); 216| 1.07k|} _ZNK8rawspeed10Cr2Decoder20decodeCanonColorDataEv: 383| 40|bool Cr2Decoder::decodeCanonColorData() const { 384| 40| const TiffEntry* wb = mRootIFD->getEntryRecursive(TiffTag::CANONCOLORDATA); 385| 40| if (!wb) ------------------ | Branch (385:7): [True: 18, False: 22] ------------------ 386| 18| return false; 387| | 388| 22| auto dsc = deduceColorDataFormat(wb); 389| 22| if (!dsc) ------------------ | Branch (389:7): [True: 6, False: 16] ------------------ 390| 6| return false; 391| | 392| 16| auto [f, ver] = *dsc; 393| | 394| 16| int offset = getWhiteBalanceOffsetInColorData(f); 395| | 396| 16| offset /= 2; 397| | 398| 16| std::array wbCoeffs = {}; 399| 16| wbCoeffs[0] = static_cast(wb->getU16(offset + 0)); 400| 16| wbCoeffs[1] = static_cast(wb->getU16(offset + 1)); 401| 16| wbCoeffs[2] = static_cast(wb->getU16(offset + 3)); 402| 16| mRaw->metadata.wbCoeffs = wbCoeffs; 403| | 404| 16| auto levelOffsets = getBlackAndWhiteLevelOffsetsInColorData(f, ver); 405| 16| if (!levelOffsets) ------------------ | Branch (405:7): [True: 3, False: 13] ------------------ 406| 3| return false; 407| | 408| 13| mRaw->whitePoint = wb->getU16(levelOffsets->second); 409| | 410| 13| mRaw->blackLevelSeparate = 411| 13| Array2DRef(mRaw->blackLevelSeparateStorage.data(), 2, 2); 412| 13| auto blackLevelSeparate1D = *mRaw->blackLevelSeparate->getAsArray1DRef(); 413| 13| for (int c = 0; c != 4; ++c) ------------------ | Branch (413:19): [True: 0, False: 13] ------------------ 414| 0| blackLevelSeparate1D(c) = wb->getU16(c + levelOffsets->first); 415| | 416| | // In Canon MakerNotes, the levels are always unscaled, and are 14-bit, 417| | // and so if the LJpeg precision was lower, we need to adjust. 418| 13| constexpr int makernotesPrecision = 14; 419| 13| if (makernotesPrecision > ljpegSamplePrecision) { ------------------ | Branch (419:7): [True: 0, False: 13] ------------------ 420| 0| int bitDepthDiff = makernotesPrecision - ljpegSamplePrecision; 421| 0| assert(bitDepthDiff >= 1 && bitDepthDiff <= 12); ------------------ | Branch (421:5): [True: 0, False: 0] | Branch (421:5): [True: 0, False: 0] | Branch (421:5): [True: 0, False: 0] ------------------ 422| 0| if (shouldRescaleBlackLevels(f, ver)) { ------------------ | Branch (422:9): [True: 0, False: 0] ------------------ 423| 0| for (int c = 0; c != 4; ++c) ------------------ | Branch (423:23): [True: 0, False: 0] ------------------ 424| 0| blackLevelSeparate1D(c) >>= bitDepthDiff; 425| 0| } 426| 0| mRaw->whitePoint = *mRaw->whitePoint >> bitDepthDiff; 427| 0| } 428| | 429| 13| return true; 430| 13|} Cr2Decoder.cpp:_ZN8rawspeed12_GLOBAL__N_121deduceColorDataFormatEPKNS_9TiffEntryE: 243| 22|deduceColorDataFormat(const TiffEntry* ccd) { 244| | // The original ColorData, detect by it's fixed size. 245| 22| if (ccd->count == 582) ------------------ | Branch (245:7): [True: 0, False: 22] ------------------ 246| 0| return {{ColorDataFormat::ColorData1, {}}}; 247| | // Second incarnation of ColorData, still size-only detection. 248| 22| if (ccd->count == 653) ------------------ | Branch (248:7): [True: 0, False: 22] ------------------ 249| 0| return {{ColorDataFormat::ColorData2, {}}}; 250| | // From now onwards, Canon has finally added a `version` field, use it. 251| 22| switch (int colorDataVersion = static_cast(ccd->getU16(0)); 252| 22| colorDataVersion) { 253| 1| case 1: ------------------ | Branch (253:3): [True: 1, False: 21] ------------------ 254| 1| return {{ColorDataFormat::ColorData3, colorDataVersion}}; 255| 1| case 2: ------------------ | Branch (255:3): [True: 1, False: 21] ------------------ 256| 2| case 3: ------------------ | Branch (256:3): [True: 1, False: 21] ------------------ 257| 3| case 4: ------------------ | Branch (257:3): [True: 1, False: 21] ------------------ 258| 4| case 5: ------------------ | Branch (258:3): [True: 1, False: 21] ------------------ 259| 5| case 6: ------------------ | Branch (259:3): [True: 1, False: 21] ------------------ 260| 6| case 7: ------------------ | Branch (260:3): [True: 1, False: 21] ------------------ 261| 7| case 9: ------------------ | Branch (261:3): [True: 1, False: 21] ------------------ 262| 7| return {{ColorDataFormat::ColorData4, colorDataVersion}}; 263| 1| case -4: ------------------ | Branch (263:3): [True: 1, False: 21] ------------------ 264| 2| case -3: ------------------ | Branch (264:3): [True: 1, False: 21] ------------------ 265| 2| return {{ColorDataFormat::ColorData5, colorDataVersion}}; 266| 1| case 10: { ------------------ | Branch (266:3): [True: 1, False: 21] ------------------ 267| 1| ColorDataFormat f = [count = ccd->count]() { 268| 1| switch (count) { 269| 1| case 1273: 270| 1| case 1275: 271| 1| return ColorDataFormat::ColorData6; 272| 1| default: 273| 1| return ColorDataFormat::ColorData7; 274| 1| } 275| 1| }(); 276| 1| return {{f, colorDataVersion}}; 277| 1| } 278| 1| case 11: ------------------ | Branch (278:3): [True: 1, False: 21] ------------------ 279| 1| return {{ColorDataFormat::ColorData7, colorDataVersion}}; 280| 1| case 12: ------------------ | Branch (280:3): [True: 1, False: 21] ------------------ 281| 2| case 13: ------------------ | Branch (281:3): [True: 1, False: 21] ------------------ 282| 3| case 14: ------------------ | Branch (282:3): [True: 1, False: 21] ------------------ 283| 4| case 15: ------------------ | Branch (283:3): [True: 1, False: 21] ------------------ 284| 4| return {{ColorDataFormat::ColorData8, colorDataVersion}}; 285| 6| default: ------------------ | Branch (285:3): [True: 6, False: 16] ------------------ 286| 6| break; 287| 22| } 288| 6| return std::nullopt; 289| 22|} Cr2Decoder.cpp:_ZZN8rawspeed12_GLOBAL__N_121deduceColorDataFormatEPKNS_9TiffEntryEENK3$_0clEv: 267| 1| ColorDataFormat f = [count = ccd->count]() { 268| 1| switch (count) { 269| 0| case 1273: ------------------ | Branch (269:7): [True: 0, False: 1] ------------------ 270| 0| case 1275: ------------------ | Branch (270:7): [True: 0, False: 1] ------------------ 271| 0| return ColorDataFormat::ColorData6; 272| 1| default: ------------------ | Branch (272:7): [True: 1, False: 0] ------------------ 273| 1| return ColorDataFormat::ColorData7; 274| 1| } 275| 1| }(); Cr2Decoder.cpp:_ZN8rawspeed12_GLOBAL__N_132getWhiteBalanceOffsetInColorDataENS0_15ColorDataFormatE: 291| 16|[[nodiscard]] int getWhiteBalanceOffsetInColorData(ColorDataFormat f) { 292| 16| switch (f) { ------------------ | Branch (292:11): [True: 16, False: 0] ------------------ 293| 0| using enum ColorDataFormat; 294| 0| case ColorData1: ------------------ | Branch (294:3): [True: 0, False: 16] ------------------ 295| 0| return 50; 296| 0| case ColorData2: ------------------ | Branch (296:3): [True: 0, False: 16] ------------------ 297| 0| return 68; 298| 1| case ColorData3: ------------------ | Branch (298:3): [True: 1, False: 15] ------------------ 299| 8| case ColorData4: ------------------ | Branch (299:3): [True: 7, False: 9] ------------------ 300| 8| case ColorData6: ------------------ | Branch (300:3): [True: 0, False: 16] ------------------ 301| 10| case ColorData7: ------------------ | Branch (301:3): [True: 2, False: 14] ------------------ 302| 14| case ColorData8: ------------------ | Branch (302:3): [True: 4, False: 12] ------------------ 303| 14| return 126; 304| 2| case ColorData5: ------------------ | Branch (304:3): [True: 2, False: 14] ------------------ 305| 2| return 142; 306| 16| } 307| 0| __builtin_unreachable(); 308| 16|} Cr2Decoder.cpp:_ZN8rawspeed12_GLOBAL__N_139getBlackAndWhiteLevelOffsetsInColorDataENS0_15ColorDataFormatENS_8OptionalIiEE: 312| 4| Optional colorDataVersion) { 313| 4| switch (f) { ------------------ | Branch (313:11): [True: 4, False: 0] ------------------ 314| 0| using enum ColorDataFormat; 315| 0| case ColorData1: ------------------ | Branch (315:3): [True: 0, False: 4] ------------------ 316| 0| case ColorData2: ------------------ | Branch (316:3): [True: 0, False: 4] ------------------ 317| 1| case ColorData3: ------------------ | Branch (317:3): [True: 1, False: 3] ------------------ 318| | // These seemingly did not contain `SpecularWhiteLevel` yet. 319| 1| return std::nullopt; 320| 3| case ColorData4: ------------------ | Branch (320:3): [True: 3, False: 1] ------------------ 321| 3| switch (*colorDataVersion) { 322| 1| case 2: ------------------ | Branch (322:5): [True: 1, False: 2] ------------------ 323| 2| case 3: ------------------ | Branch (323:5): [True: 1, False: 2] ------------------ 324| 2| return std::nullopt; // Still no `SpecularWhiteLevel`. 325| 1| case 4: ------------------ | Branch (325:5): [True: 1, False: 2] ------------------ 326| 1| case 5: ------------------ | Branch (326:5): [True: 0, False: 3] ------------------ 327| 1| return {{692, 697}}; 328| 0| case 6: ------------------ | Branch (328:5): [True: 0, False: 3] ------------------ 329| 0| case 7: ------------------ | Branch (329:5): [True: 0, False: 3] ------------------ 330| 0| return {{715, 720}}; 331| 0| case 9: ------------------ | Branch (331:5): [True: 0, False: 3] ------------------ 332| 0| return {{719, 724}}; 333| 0| default: ------------------ | Branch (333:5): [True: 0, False: 3] ------------------ 334| 0| __builtin_unreachable(); 335| 3| } 336| 0| case ColorData5: ------------------ | Branch (336:3): [True: 0, False: 4] ------------------ 337| 0| switch (*colorDataVersion) { 338| 0| case -4: ------------------ | Branch (338:5): [True: 0, False: 0] ------------------ 339| 0| return {{333, 1386}}; 340| 0| case -3: ------------------ | Branch (340:5): [True: 0, False: 0] ------------------ 341| 0| return {{264, 662}}; 342| 0| default: ------------------ | Branch (342:5): [True: 0, False: 0] ------------------ 343| 0| __builtin_unreachable(); 344| 0| } 345| 0| case ColorData6: ------------------ | Branch (345:3): [True: 0, False: 4] ------------------ 346| 0| switch (*colorDataVersion) { 347| 0| case 10: ------------------ | Branch (347:5): [True: 0, False: 0] ------------------ 348| 0| return {{479, 484}}; 349| 0| default: ------------------ | Branch (349:5): [True: 0, False: 0] ------------------ 350| 0| __builtin_unreachable(); 351| 0| } 352| 0| case ColorData7: ------------------ | Branch (352:3): [True: 0, False: 4] ------------------ 353| 0| switch (*colorDataVersion) { 354| 0| case 10: ------------------ | Branch (354:5): [True: 0, False: 0] ------------------ 355| 0| return {{504, 509}}; 356| 0| case 11: ------------------ | Branch (356:5): [True: 0, False: 0] ------------------ 357| 0| return {{728, 733}}; 358| 0| default: ------------------ | Branch (358:5): [True: 0, False: 0] ------------------ 359| 0| __builtin_unreachable(); 360| 0| } 361| 0| case ColorData8: ------------------ | Branch (361:3): [True: 0, False: 4] ------------------ 362| 0| switch (*colorDataVersion) { 363| 0| case 12: ------------------ | Branch (363:5): [True: 0, False: 0] ------------------ 364| 0| case 13: ------------------ | Branch (364:5): [True: 0, False: 0] ------------------ 365| 0| case 15: ------------------ | Branch (365:5): [True: 0, False: 0] ------------------ 366| 0| return {{778, 783}}; 367| 0| case 14: ------------------ | Branch (367:5): [True: 0, False: 0] ------------------ 368| 0| return {{556, 561}}; 369| 0| default: ------------------ | Branch (369:5): [True: 0, False: 0] ------------------ 370| 0| __builtin_unreachable(); 371| 0| } 372| 4| } 373| 0| __builtin_unreachable(); 374| 4|} _ZNK8rawspeed10Cr2Decoder17parseWhiteBalanceEv: 432| 40|void Cr2Decoder::parseWhiteBalance() const { 433| | // Default white point is LJpeg sample precision. 434| 40| mRaw->whitePoint = (1U << ljpegSamplePrecision) - 1; 435| | 436| 40| if (decodeCanonColorData()) ------------------ | Branch (436:7): [True: 0, False: 40] ------------------ 437| 0| return; 438| | 439| 40| if (mRootIFD->hasEntryRecursive(TiffTag::CANONSHOTINFO) && ------------------ | Branch (439:7): [True: 7, False: 33] ------------------ 440| 7| mRootIFD->hasEntryRecursive(TiffTag::CANONPOWERSHOTG9WB)) { ------------------ | Branch (440:7): [True: 3, False: 4] ------------------ 441| 3| const TiffEntry* shot_info = 442| 3| mRootIFD->getEntryRecursive(TiffTag::CANONSHOTINFO); 443| 3| const TiffEntry* g9_wb = 444| 3| mRootIFD->getEntryRecursive(TiffTag::CANONPOWERSHOTG9WB); 445| | 446| 3| uint16_t wb_index = shot_info->getU16(7); 447| 3| int wb_offset = (wb_index < 18) ------------------ | Branch (447:21): [True: 1, False: 2] ------------------ 448| 3| ? std::string_view("012347800000005896")[wb_index] - '0' 449| 3| : 0; 450| 3| wb_offset = wb_offset * 8 + 2; 451| | 452| 3| std::array wbCoeffs = {}; 453| 3| wbCoeffs[0] = static_cast(g9_wb->getU32(wb_offset + 1)); 454| 3| wbCoeffs[1] = (static_cast(g9_wb->getU32(wb_offset + 0)) + 455| 3| static_cast(g9_wb->getU32(wb_offset + 3))) / 456| 3| 2.0F; 457| 3| wbCoeffs[2] = static_cast(g9_wb->getU32(wb_offset + 2)); 458| 3| mRaw->metadata.wbCoeffs = wbCoeffs; 459| 37| } else if (mRootIFD->hasEntryRecursive(static_cast(0xa4))) { ------------------ | Branch (459:14): [True: 9, False: 28] ------------------ 460| | // WB for the old 1D and 1DS 461| 9| const TiffEntry* wb = 462| 9| mRootIFD->getEntryRecursive(static_cast(0xa4)); 463| 9| if (wb->count >= 3) { ------------------ | Branch (463:9): [True: 7, False: 2] ------------------ 464| 7| std::array wbCoeffs = {}; 465| 7| wbCoeffs[0] = wb->getFloat(0); 466| 7| wbCoeffs[1] = wb->getFloat(1); 467| 7| wbCoeffs[2] = wb->getFloat(2); 468| 7| mRaw->metadata.wbCoeffs = wbCoeffs; 469| 7| } 470| 9| } 471| 40|} _ZN8rawspeed10Cr2Decoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 473| 41|void Cr2Decoder::decodeMetaDataInternal(const CameraMetaData* meta) { 474| 41| int iso = 0; 475| 41| mRaw->cfa.setCFA(iPoint2D(2, 2), CFAColor::RED, CFAColor::GREEN, 476| 41| CFAColor::GREEN, CFAColor::BLUE); 477| | 478| 41| std::string mode; 479| | 480| 41| if (mRaw->metadata.subsampling.y == 2 && mRaw->metadata.subsampling.x == 2) ------------------ | Branch (480:7): [True: 0, False: 41] | Branch (480:44): [True: 0, False: 0] ------------------ 481| 0| mode = "sRaw1"; 482| | 483| 41| if (mRaw->metadata.subsampling.y == 1 && mRaw->metadata.subsampling.x == 2) ------------------ | Branch (483:7): [True: 41, False: 0] | Branch (483:44): [True: 0, False: 41] ------------------ 484| 0| mode = "sRaw2"; 485| | 486| 41| if (mRootIFD->hasEntryRecursive(TiffTag::ISOSPEEDRATINGS)) ------------------ | Branch (486:7): [True: 9, False: 32] ------------------ 487| 9| iso = mRootIFD->getEntryRecursive(TiffTag::ISOSPEEDRATINGS)->getU32(); 488| 41| if (65535 == iso) { ------------------ | Branch (488:7): [True: 0, False: 41] ------------------ 489| | // ISOSPEEDRATINGS is a SHORT EXIF value. For larger values, we have to look 490| | // at RECOMMENDEDEXPOSUREINDEX (maybe Canon specific). 491| 0| if (mRootIFD->hasEntryRecursive(TiffTag::RECOMMENDEDEXPOSUREINDEX)) ------------------ | Branch (491:9): [True: 0, False: 0] ------------------ 492| 0| iso = mRootIFD->getEntryRecursive(TiffTag::RECOMMENDEDEXPOSUREINDEX) 493| 0| ->getU32(); 494| 0| } 495| | 496| | // Fetch the white balance 497| 41| try { 498| 41| parseWhiteBalance(); 499| 41| } catch (const RawspeedException& e) { 500| 23| mRaw->setError(e.what()); 501| | // We caught an exception reading WB, just ignore it 502| 23| } 503| 41| setMetaData(meta, mode, iso); 504| 40| assert(mShiftUpScaleForExif == 0 || mShiftUpScaleForExif == 2); ------------------ | Branch (504:3): [True: 40, False: 0] | Branch (504:3): [True: 0, False: 0] | Branch (504:3): [True: 40, False: 0] ------------------ 505| 40| if (mShiftUpScaleForExif) { ------------------ | Branch (505:7): [True: 0, False: 40] ------------------ 506| 0| mRaw->blackLevel = 0; 507| 0| mRaw->blackLevelSeparate = std::nullopt; 508| 0| } 509| 40| if (mShiftUpScaleForExif != 0 && isPowerOfTwo(1 + *mRaw->whitePoint)) ------------------ | Branch (509:7): [True: 0, False: 40] | Branch (509:36): [True: 0, False: 0] ------------------ 510| 0| mRaw->whitePoint = ((1 + *mRaw->whitePoint) << mShiftUpScaleForExif) - 1; 511| 40| else 512| 40| mRaw->whitePoint = *mRaw->whitePoint << mShiftUpScaleForExif; 513| 40|} _ZNK8rawspeed10Cr2Decoder12isSubSampledEv: 515| 318|bool Cr2Decoder::isSubSampled() const { 516| 318| if (mRootIFD->getSubIFDs().size() != 4) ------------------ | Branch (516:7): [True: 236, False: 82] ------------------ 517| 236| return false; 518| 82| const TiffEntry* typeE = 519| 82| mRootIFD->getSubIFDs()[3]->getEntryRecursive(TiffTag::CANON_SRAWTYPE); 520| 82| return typeE && typeE->getU32() == 4; ------------------ | Branch (520:10): [True: 75, False: 7] | Branch (520:19): [True: 10, False: 65] ------------------ 521| 318|} _ZNK8rawspeed10Cr2Decoder14getSubSamplingEv: 523| 116|iPoint2D Cr2Decoder::getSubSampling() const { 524| 116| const TiffEntry* CCS = 525| 116| mRootIFD->getEntryRecursive(TiffTag::CANON_CAMERA_SETTINGS); 526| 116| if (!CCS) ------------------ | Branch (526:7): [True: 1, False: 115] ------------------ 527| 116| ThrowRDE("CanonCameraSettings entry not found."); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 528| | 529| 115| if (CCS->type != TiffDataType::SHORT) ------------------ | Branch (529:7): [True: 2, False: 113] ------------------ 530| 115| ThrowRDE("Unexpected CanonCameraSettings entry type encountered "); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 531| | 532| 113| if (CCS->count < 47) ------------------ | Branch (532:7): [True: 99, False: 14] ------------------ 533| 99| return {1, 1}; 534| | 535| 14| switch (uint16_t qual = CCS->getU16(46)) { 536| 3| case 0: ------------------ | Branch (536:3): [True: 3, False: 11] ------------------ 537| 3| return {1, 1}; 538| 2| case 1: ------------------ | Branch (538:3): [True: 2, False: 12] ------------------ 539| 2| return {2, 2}; 540| 5| case 2: ------------------ | Branch (540:3): [True: 5, False: 9] ------------------ 541| 5| return {2, 1}; 542| 4| default: ------------------ | Branch (542:3): [True: 4, False: 10] ------------------ 543| 4| ThrowRDE("Unexpected SRAWQuality value found: %u", qual); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 544| 14| } 545| 14|} _ZN8rawspeed10Cr2DecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 39| 1.07k| : AbstractTiffDecoder(std::move(root), file) {} _ZN8rawspeed10DcrDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 44| 1.20k| [[maybe_unused]] Buffer file) { 45| 1.20k| const auto id = rootIFD->getID(); 46| 1.20k| const std::string& make = id.make; 47| | 48| | // FIXME: magic 49| | 50| 1.20k| return make == "Kodak"; 51| 1.20k|} _ZN8rawspeed10DcrDecoder20checkImageDimensionsEv: 53| 250|void DcrDecoder::checkImageDimensions() { 54| 250| if (width > 4516 || height > 3012) ------------------ | Branch (54:7): [True: 22, False: 228] | Branch (54:23): [True: 17, False: 211] ------------------ 55| 250| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 39| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 39| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 39| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 39| __VA_ARGS__) | | ------------------ ------------------ 56| 250|} _ZN8rawspeed10DcrDecoder17decodeRawInternalEv: 58| 278|RawImage DcrDecoder::decodeRawInternal() { 59| 278| SimpleTiffDecoder::prepareForRawDecoding(); 60| | 61| 278| ByteStream input(DataBuffer(mFile.getSubView(off), Endianness::little)); 62| | 63| 278| if (int compression = raw->getEntry(TiffTag::COMPRESSION)->getU32(); 64| 278| 65000 != compression) ------------------ | Branch (64:7): [True: 7, False: 271] ------------------ 65| 278| ThrowRDE("Unsupported compression %d", compression); ------------------ | | 38| 7| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 7| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 7| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 7| __VA_ARGS__) | | ------------------ ------------------ 66| | 67| 271| const TiffEntry* ifdoffset = mRootIFD->getEntryRecursive(TiffTag::KODAK_IFD); 68| 271| if (!ifdoffset) ------------------ | Branch (68:7): [True: 1, False: 270] ------------------ 69| 271| ThrowRDE("Couldn't find the Kodak IFD offset"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 70| | 71| 270| NORangesSet ifds; 72| | 73| 270| assert(ifdoffset != nullptr); ------------------ | Branch (73:3): [True: 202, False: 68] ------------------ 74| 202| TiffRootIFD kodakifd(nullptr, &ifds, ifdoffset->getRootIfdData(), 75| 202| ifdoffset->getU32()); 76| | 77| 202| const TiffEntry* linearization = 78| 202| kodakifd.getEntryRecursive(TiffTag::KODAK_LINEARIZATION); 79| 202| if (!linearization || ------------------ | Branch (79:7): [True: 126, False: 76] ------------------ 80| 76| (linearization->count != 1024 && linearization->count != 4096) || ------------------ | Branch (80:8): [True: 2, False: 74] | Branch (80:40): [True: 1, False: 1] ------------------ 81| 75| linearization->type != TiffDataType::SHORT) ------------------ | Branch (81:7): [True: 1, False: 74] ------------------ 82| 202| ThrowRDE("Couldn't find the linearization table"); ------------------ | | 38| 118| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 118| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 118| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 118| __VA_ARGS__) | | ------------------ ------------------ 83| | 84| 202| assert(linearization != nullptr); ------------------ | Branch (84:3): [True: 74, False: 10] ------------------ 85| 74| auto linTable = linearization->getU16Array(linearization->count); 86| | 87| 74| RawImageCurveGuard curveHandler(&mRaw, linTable, uncorrectedRawValues); 88| | 89| | // FIXME: dcraw does all sorts of crazy things besides this to fetch 90| | // WB from what appear to be presets and calculate it in weird ways 91| | // The only file I have only uses this method, if anybody careas look 92| | // in dcraw.c parse_kodak_ifd() for all that weirdness 93| 74| if (const TiffEntry* blob = 94| 74| kodakifd.getEntryRecursive(static_cast(0x03fd)); 95| 74| blob && blob->count == 72) { ------------------ | Branch (95:7): [True: 26, False: 48] | Branch (95:15): [True: 23, False: 3] ------------------ 96| 23| std::array wbCoeffs = {}; 97| 84| for (auto i = 0U; i < 3; i++) { ------------------ | Branch (97:23): [True: 64, False: 20] ------------------ 98| 64| const auto mul = blob->getU16(20 + i); 99| 64| if (0 == mul) ------------------ | Branch (99:11): [True: 3, False: 61] ------------------ 100| 64| ThrowRDE("WB coefficient is zero!"); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 101| 61| wbCoeffs[i] = 2048.0F / mul; 102| 61| } 103| 20| mRaw->metadata.wbCoeffs = wbCoeffs; 104| 20| } 105| | 106| 71| const int bps = [CurveSize = linearization->count]() { 107| 71| switch (CurveSize) { 108| 71| case 1024: 109| 71| return 10; 110| 71| case 4096: 111| 71| return 12; 112| 71| default: 113| 71| __builtin_unreachable(); 114| 71| } 115| 71| }(); 116| | 117| 71| KodakDecompressor k(mRaw, input, bps, uncorrectedRawValues); 118| 71| mRaw->createData(); 119| 71| k.decompress(); 120| | 121| 71| return mRaw; 122| 74|} DcrDecoder.cpp:_ZZN8rawspeed10DcrDecoder17decodeRawInternalEvENK3$_0clEv: 106| 70| const int bps = [CurveSize = linearization->count]() { 107| 70| switch (CurveSize) { 108| 69| case 1024: ------------------ | Branch (108:5): [True: 69, False: 1] ------------------ 109| 69| return 10; 110| 1| case 4096: ------------------ | Branch (110:5): [True: 1, False: 69] ------------------ 111| 1| return 12; 112| 0| default: ------------------ | Branch (112:5): [True: 0, False: 70] ------------------ 113| 0| __builtin_unreachable(); 114| 70| } 115| 70| }(); _ZN8rawspeed10DcrDecoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 124| 3|void DcrDecoder::decodeMetaDataInternal(const CameraMetaData* meta) { 125| 3| setMetaData(meta, "", 0); 126| 3|} _ZN8rawspeed10DcrDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 41| 278| : SimpleTiffDecoder(std::move(root), file) {} _ZN8rawspeed10DcsDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 44| 929| [[maybe_unused]] Buffer file) { 45| 929| const auto id = rootIFD->getID(); 46| 929| const std::string& make = id.make; 47| | 48| | // FIXME: magic 49| | 50| 929| return make == "KODAK"; 51| 929|} _ZN8rawspeed10DcsDecoder20checkImageDimensionsEv: 53| 75|void DcsDecoder::checkImageDimensions() { 54| 75| if (width > 3072 || height > 2048) ------------------ | Branch (54:7): [True: 30, False: 45] | Branch (54:23): [True: 19, False: 26] ------------------ 55| 75| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 49| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 49| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 49| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 49| __VA_ARGS__) | | ------------------ ------------------ 56| 75|} _ZN8rawspeed10DcsDecoder17decodeRawInternalEv: 58| 100|RawImage DcsDecoder::decodeRawInternal() { 59| 100| SimpleTiffDecoder::prepareForRawDecoding(); 60| | 61| 100| const TiffEntry* linearization = 62| 100| mRootIFD->getEntryRecursive(TiffTag::GRAYRESPONSECURVE); 63| 100| if (!linearization || linearization->count != 256 || ------------------ | Branch (63:7): [True: 78, False: 22] | Branch (63:25): [True: 3, False: 19] ------------------ 64| 19| linearization->type != TiffDataType::SHORT) ------------------ | Branch (64:7): [True: 1, False: 18] ------------------ 65| 100| ThrowRDE("Couldn't find the linearization table"); ------------------ | | 38| 8| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 8| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 8| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 8| __VA_ARGS__) | | ------------------ ------------------ 66| | 67| 100| assert(linearization != nullptr); ------------------ | Branch (67:3): [True: 18, False: 74] ------------------ 68| 18| auto table = linearization->getU16Array(256); 69| | 70| 18| RawImageCurveGuard curveHandler(&mRaw, table, uncorrectedRawValues); 71| | 72| 18| UncompressedDecompressor u( 73| 18| ByteStream(DataBuffer(mFile.getSubView(off, c2), Endianness::little)), 74| 18| mRaw, iRectangle2D({0, 0}, iPoint2D(width, height)), 8 * width / 8, 8, 75| 18| BitOrder::LSB); 76| 18| mRaw->createData(); 77| | 78| 18| if (uncorrectedRawValues) ------------------ | Branch (78:7): [True: 0, False: 18] ------------------ 79| 0| u.decode8BitRaw(); 80| 18| else 81| 18| u.decode8BitRaw(); 82| | 83| 18| return mRaw; 84| 92|} _ZN8rawspeed10DcsDecoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 86| 17|void DcsDecoder::decodeMetaDataInternal(const CameraMetaData* meta) { 87| 17| setMetaData(meta, "", 0); 88| 17|} _ZN8rawspeed10DcsDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 43| 100| : SimpleTiffDecoder(std::move(root), file) {} _ZN8rawspeed10DngDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 62| 12.8k| const TiffRootIFD* rootIFD, [[maybe_unused]] Buffer file) { 63| 12.8k| return rootIFD->hasEntryRecursive(TiffTag::DNGVERSION); 64| 12.8k|} _ZN8rawspeed10DngDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 67| 3.94k| : AbstractTiffDecoder(std::move(rootIFD), file) { 68| 3.94k| if (!mRootIFD->hasEntryRecursive(TiffTag::DNGVERSION)) ------------------ | Branch (68:7): [True: 0, False: 3.94k] ------------------ 69| 3.94k| ThrowRDE("DNG, but version tag is missing. Will not guess."); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 70| | 71| 3.94k| const auto v = 72| 3.94k| mRootIFD->getEntryRecursive(TiffTag::DNGVERSION)->getData().getBuffer(4); 73| | 74| 3.94k| if (v[0] != 1) { ------------------ | Branch (74:7): [True: 9, False: 3.93k] ------------------ 75| 9| ThrowRDE("Not a supported DNG image format: v%i.%i.%i.%i", ------------------ | | 38| 9| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 9| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 9| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 9| __VA_ARGS__) | | ------------------ ------------------ 76| 9| static_cast(v[0]), static_cast(v[1]), 77| 9| static_cast(v[2]), static_cast(v[3])); 78| 9| } 79| | // if (v[1] > 4) 80| | // ThrowRDE("Not a supported DNG image format: v%u.%u.%u.%u", (int)v[0], 81| | // (int)v[1], (int)v[2], (int)v[3]); 82| | 83| | // Prior to v1.1.xxx fix LJPEG encoding bug 84| 3.93k| mFixLjpeg = (v[0] <= 1) && (v[1] < 1); ------------------ | Branch (84:15): [True: 3.93k, False: 1] | Branch (84:30): [True: 358, False: 3.57k] ------------------ 85| 3.93k|} _ZN8rawspeed10DngDecoder20dropUnsuportedChunksEPNSt3__16vectorIPKNS_7TiffIFDENS1_9allocatorIS5_EEEE: 87| 3.91k|void DngDecoder::dropUnsuportedChunks(std::vector* data) { 88| 8.73k| for (auto i = data->begin(); i != data->end();) { ------------------ | Branch (88:32): [True: 4.82k, False: 3.90k] ------------------ 89| 4.82k| const auto& ifd = *i; 90| | 91| 4.82k| int comp = ifd->getEntry(TiffTag::COMPRESSION)->getU16(); 92| 4.82k| bool isSubsampled = false; 93| 4.82k| bool isAlpha = false; 94| | 95| 4.82k| if (ifd->hasEntry(TiffTag::NEWSUBFILETYPE) && ------------------ | Branch (95:9): [True: 430, False: 4.39k] ------------------ 96| 430| ifd->getEntry(TiffTag::NEWSUBFILETYPE)->isInt()) { ------------------ | Branch (96:9): [True: 407, False: 23] ------------------ 97| 407| const uint32_t NewSubFileType = 98| 407| (*i)->getEntry(TiffTag::NEWSUBFILETYPE)->getU32(); 99| | 100| | // bit 0 is on if image is subsampled. 101| | // the value itself can be either 1, or 0x10001. 102| | // or 5 for "Transparency information for subsampled raw images" 103| 407| isSubsampled = NewSubFileType & (1 << 0); 104| | 105| | // bit 2 is on if image contains transparency information. 106| | // the value itself can be either 4 or 5 107| 407| isAlpha = NewSubFileType & (1 << 2); 108| 407| } 109| | 110| | // normal raw? 111| 4.82k| bool supported = !isSubsampled && !isAlpha; ------------------ | Branch (111:22): [True: 4.53k, False: 285] | Branch (111:39): [True: 4.53k, False: 1] ------------------ 112| | 113| 4.82k| switch (comp) { 114| 3.16k| case 1: // uncompressed ------------------ | Branch (114:5): [True: 3.16k, False: 1.66k] ------------------ 115| 3.82k| case 7: // lossless JPEG ------------------ | Branch (115:5): [True: 663, False: 4.15k] ------------------ 116| |#ifdef HAVE_ZLIB 117| | case 8: // deflate 118| |#endif 119| 4.09k| case 9: // VC-5 as used by GoPro ------------------ | Branch (119:5): [True: 271, False: 4.55k] ------------------ 120| |#ifdef HAVE_JPEG 121| | case 0x884c: // lossy JPEG 122| |#endif 123| | // no change, if supported, then is still supported. 124| 4.09k| break; 125| | 126| 0|#ifndef HAVE_ZLIB 127| 14| case 8: // deflate ------------------ | Branch (127:5): [True: 14, False: 4.80k] ------------------ 128| 14|#pragma message \ 129| 14| "ZLIB is not present! Deflate compression will not be supported!" 130| 14| writeLog(DEBUG_PRIO::WARNING, "DNG Decoder: found Deflate-encoded chunk, " 131| 14| "but the deflate support was disabled at " 132| 14| "build!"); 133| 14| [[clang::fallthrough]]; 134| 14|#endif 135| 14|#ifndef HAVE_JPEG 136| 79| case 0x884c: // lossy JPEG ------------------ | Branch (136:5): [True: 65, False: 4.75k] ------------------ 137| 79|#pragma message \ 138| 79| "JPEG is not present! Lossy JPEG compression will not be supported!" 139| 79| writeLog(DEBUG_PRIO::WARNING, "DNG Decoder: found lossy JPEG-encoded " 140| 79| "chunk, but the jpeg support was " 141| 79| "disabled at build!"); 142| 79| [[clang::fallthrough]]; 143| 79|#endif 144| 720| default: ------------------ | Branch (144:5): [True: 641, False: 4.18k] ------------------ 145| 720| supported = false; 146| 720| break; 147| 4.82k| } 148| | 149| 4.81k| if (supported) ------------------ | Branch (149:9): [True: 3.94k, False: 870] ------------------ 150| 3.94k| ++i; 151| 870| else 152| 870| i = data->erase(i); 153| 4.81k| } 154| 3.91k|} _ZNK8rawspeed10DngDecoder15parseACTIVEAREAEPKNS_7TiffIFDE: 156| 2.76k|Optional DngDecoder::parseACTIVEAREA(const TiffIFD* raw) const { 157| 2.76k| if (!raw->hasEntry(TiffTag::ACTIVEAREA)) ------------------ | Branch (157:7): [True: 2.55k, False: 212] ------------------ 158| 2.55k| return {}; 159| | 160| 212| const TiffEntry* active_area = raw->getEntry(TiffTag::ACTIVEAREA); 161| 212| if (active_area->count != 4) ------------------ | Branch (161:7): [True: 2, False: 210] ------------------ 162| 212| ThrowRDE("active area has %u values instead of 4", active_area->count); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 163| | 164| 210| const iRectangle2D fullImage(0, 0, mRaw->dim.x, mRaw->dim.y); 165| | 166| 210| const auto corners = active_area->getU32Array(4); 167| 210| const iPoint2D topLeft(static_cast(corners[1]), 168| 210| static_cast(corners[0])); 169| 210| const iPoint2D bottomRight(static_cast(corners[3]), 170| 210| static_cast(corners[2])); 171| | 172| 210| if (!(fullImage.isPointInsideInclusive(topLeft) && ------------------ | Branch (172:9): [True: 107, False: 103] ------------------ 173| 107| fullImage.isPointInsideInclusive(bottomRight) && ------------------ | Branch (173:9): [True: 18, False: 89] ------------------ 174| 193| bottomRight >= topLeft)) { ------------------ | Branch (174:9): [True: 16, False: 2] ------------------ 175| 193| ThrowRDE("Rectangle (%i, %i, %i, %i) not inside image (%i, %i, %i, %i).", ------------------ | | 38| 193| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 193| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 193| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 193| __VA_ARGS__) | | ------------------ ------------------ 176| 193| topLeft.x, topLeft.y, bottomRight.x, bottomRight.y, 177| 193| fullImage.getTopLeft().x, fullImage.getTopLeft().y, 178| 193| fullImage.getBottomRight().x, fullImage.getBottomRight().y); 179| 193| } 180| | 181| 17| iRectangle2D crop; 182| 17| crop.setTopLeft(topLeft); 183| 17| crop.setBottomRightAbsolute(bottomRight); 184| 17| assert(fullImage.isThisInside(fullImage)); ------------------ | Branch (184:3): [True: 16, False: 1] ------------------ 185| | 186| 16| return crop; 187| 17|} _ZNK8rawspeed10DngDecoder8parseCFAEPKNS_7TiffIFDE: 213| 344|void DngDecoder::parseCFA(const TiffIFD* raw) const { 214| | 215| | // Check if layout is OK, if present 216| 344| if (raw->hasEntry(TiffTag::CFALAYOUT) && ------------------ | Branch (216:7): [True: 2, False: 342] ------------------ 217| 2| raw->getEntry(TiffTag::CFALAYOUT)->getU16() != 1) ------------------ | Branch (217:7): [True: 1, False: 1] ------------------ 218| 344| ThrowRDE("Unsupported CFA Layout."); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 219| | 220| 343| const TiffEntry* cfadim = raw->getEntry(TiffTag::CFAREPEATPATTERNDIM); 221| 343| if (cfadim->count != 2) ------------------ | Branch (221:7): [True: 1, False: 342] ------------------ 222| 343| ThrowRDE("Couldn't read CFA pattern dimension"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 223| | 224| | // Does NOT contain dimensions as some documents state 225| 342| const TiffEntry* cPat = raw->getEntry(TiffTag::CFAPATTERN); 226| 342| if (!cPat->count) ------------------ | Branch (226:7): [True: 1, False: 341] ------------------ 227| 342| ThrowRDE("CFA pattern is empty!"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 228| | 229| 341| iPoint2D cfaSize(cfadim->getU32(1), cfadim->getU32(0)); 230| 341| if (!cfaSize.hasPositiveArea() || cfaSize.area() != cPat->count) { ------------------ | Branch (230:7): [True: 26, False: 315] | Branch (230:37): [True: 73, False: 242] ------------------ 231| 92| ThrowRDE("CFA pattern dimension and pattern count does not " ------------------ | | 38| 92| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 92| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 92| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 92| __VA_ARGS__) | | ------------------ ------------------ 232| 92| "match: %u.", 233| 92| cPat->count); 234| 92| } 235| | 236| 249| mRaw->cfa.setSize(cfaSize); 237| | 238| 844| for (int y = 0; y < cfaSize.y; y++) { ------------------ | Branch (238:19): [True: 609, False: 235] ------------------ 239| 1.79k| for (int x = 0; x < cfaSize.x; x++) { ------------------ | Branch (239:21): [True: 1.19k, False: 595] ------------------ 240| 1.19k| uint32_t c1 = cPat->getByte(x + (y * cfaSize.x)); 241| | 242| 1.19k| auto c2 = getDNGCFAPatternAsCFAColor(c1); 243| 1.19k| if (!c2) ------------------ | Branch (243:11): [True: 14, False: 1.18k] ------------------ 244| 1.19k| ThrowRDE("Unsupported CFA Color: %u", c1); ------------------ | | 38| 14| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 14| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 14| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 14| __VA_ARGS__) | | ------------------ ------------------ 245| | 246| 1.18k| mRaw->cfa.setColorAt(iPoint2D(x, y), *c2); 247| 1.18k| } 248| 609| } 249| | 250| | // the cfa is specified relative to the ActiveArea. we want it relative (0,0) 251| | // Since in handleMetadata(), in subFrame() we unconditionally shift CFA by 252| | // activearea+DefaultCropOrigin; here we need to undo the 'ACTIVEAREA' part. 253| 235| const Optional aa = parseACTIVEAREA(raw); 254| 235| if (!aa) ------------------ | Branch (254:7): [True: 216, False: 19] ------------------ 255| 216| return; 256| | 257| | // To reverse the ActiveArea modifictions done earlier, we need to 258| | // use the negated ActiveArea x/y values. 259| 19| mRaw->cfa.shiftRight(-int(aa->pos.x)); 260| 19| mRaw->cfa.shiftDown(-int(aa->pos.y)); 261| 19|} DngDecoder.cpp:_ZN8rawspeed12_GLOBAL__N_126getDNGCFAPatternAsCFAColorEj: 190| 1.19k|Optional getDNGCFAPatternAsCFAColor(uint32_t c) { 191| 1.19k| switch (c) { 192| 0| using enum CFAColor; 193| 464| case 0: ------------------ | Branch (193:3): [True: 464, False: 730] ------------------ 194| 464| return RED; 195| 395| case 1: ------------------ | Branch (195:3): [True: 395, False: 799] ------------------ 196| 395| return GREEN; 197| 217| case 2: ------------------ | Branch (197:3): [True: 217, False: 977] ------------------ 198| 217| return BLUE; 199| 26| case 3: ------------------ | Branch (199:3): [True: 26, False: 1.16k] ------------------ 200| 26| return CYAN; 201| 24| case 4: ------------------ | Branch (201:3): [True: 24, False: 1.17k] ------------------ 202| 24| return MAGENTA; 203| 20| case 5: ------------------ | Branch (203:3): [True: 20, False: 1.17k] ------------------ 204| 20| return YELLOW; 205| 34| case 6: ------------------ | Branch (205:3): [True: 34, False: 1.16k] ------------------ 206| 34| return WHITE; 207| 14| default: ------------------ | Branch (207:3): [True: 14, False: 1.18k] ------------------ 208| 14| return std::nullopt; 209| 1.19k| } 210| 1.19k|} _ZNK8rawspeed10DngDecoder16parseColorMatrixEv: 263| 1.68k|void DngDecoder::parseColorMatrix() const { 264| | // Look for D65 calibrated color matrix 265| | 266| 1.68k| auto impl = [this](TiffTag I, TiffTag M) -> TiffEntry* { 267| 1.68k| if (!mRootIFD->hasEntryRecursive(I)) 268| 1.68k| return nullptr; 269| 1.68k| if (const TiffEntry* illuminant = mRootIFD->getEntryRecursive(I); 270| 1.68k| illuminant->getU16() != 21 || // D65 271| 1.68k| !mRootIFD->hasEntryRecursive(M)) 272| 1.68k| return nullptr; 273| 1.68k| return mRootIFD->getEntryRecursive(M); 274| 1.68k| }; 275| | 276| 1.68k| const TiffEntry* mat; 277| 1.68k| mat = impl(TiffTag::CALIBRATIONILLUMINANT1, TiffTag::COLORMATRIX1); 278| 1.68k| if (!mat) ------------------ | Branch (278:7): [True: 1.64k, False: 32] ------------------ 279| 1.64k| mat = impl(TiffTag::CALIBRATIONILLUMINANT2, TiffTag::COLORMATRIX2); 280| 1.68k| if (!mat) ------------------ | Branch (280:7): [True: 1.63k, False: 43] ------------------ 281| 1.63k| return; 282| | 283| | // Color matrix size *MUST* be a multiple of 3 (number of channels in XYZ). 284| 43| if (mat->count % 3 != 0) ------------------ | Branch (284:7): [True: 2, False: 41] ------------------ 285| 2| return; 286| | 287| 41| const auto srat_vals = mat->getSRationalArray(mat->count); 288| 41| bool Success = true; 289| 41| mRaw->metadata.colorMatrix.reserve(mat->count); 290| 1.75k| for (const auto& val : srat_vals) { ------------------ | Branch (290:24): [True: 1.75k, False: 36] ------------------ 291| 1.75k| Success &= val.den != 0; 292| 1.75k| if (!Success) ------------------ | Branch (292:9): [True: 5, False: 1.75k] ------------------ 293| 5| break; 294| 1.75k| mRaw->metadata.colorMatrix.emplace_back(val); 295| 1.75k| } 296| 41| if (!Success) ------------------ | Branch (296:7): [True: 5, False: 36] ------------------ 297| 5| mRaw->metadata.colorMatrix.clear(); 298| 41|} DngDecoder.cpp:_ZZNK8rawspeed10DngDecoder16parseColorMatrixEvENK3$_0clENS_7TiffTagES2_: 266| 3.33k| auto impl = [this](TiffTag I, TiffTag M) -> TiffEntry* { 267| 3.33k| if (!mRootIFD->hasEntryRecursive(I)) ------------------ | Branch (267:9): [True: 3.26k, False: 67] ------------------ 268| 3.26k| return nullptr; 269| 67| if (const TiffEntry* illuminant = mRootIFD->getEntryRecursive(I); 270| 67| illuminant->getU16() != 21 || // D65 ------------------ | Branch (270:9): [True: 37, False: 30] ------------------ 271| 30| !mRootIFD->hasEntryRecursive(M)) ------------------ | Branch (271:9): [True: 2, False: 28] ------------------ 272| 24| return nullptr; 273| 43| return mRootIFD->getEntryRecursive(M); 274| 67| }; _ZNK8rawspeed10DngDecoder20getTilingDescriptionEPKNS_7TiffIFDE: 301| 3.47k|DngDecoder::getTilingDescription(const TiffIFD* raw) const { 302| 3.47k| if (raw->hasEntry(TiffTag::TILEOFFSETS)) { ------------------ | Branch (302:7): [True: 322, False: 3.15k] ------------------ 303| 322| const uint32_t tilew = raw->getEntry(TiffTag::TILEWIDTH)->getU32(); 304| 322| const uint32_t tileh = raw->getEntry(TiffTag::TILELENGTH)->getU32(); 305| | 306| 322| if (tilew <= 0 || tileh <= 0) ------------------ | Branch (306:9): [True: 2, False: 320] | Branch (306:23): [True: 2, False: 318] ------------------ 307| 322| ThrowRDE("Invalid tile size: (%u, %u)", tilew, tileh); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 308| | 309| 322| assert(tilew > 0); ------------------ | Branch (309:5): [True: 318, False: 1] ------------------ 310| 318| const auto tilesX = 311| 318| implicit_cast(roundUpDivisionSafe(mRaw->dim.x, tilew)); 312| 318| if (!tilesX) ------------------ | Branch (312:9): [True: 0, False: 318] ------------------ 313| 318| ThrowRDE("Zero tiles horizontally"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 314| | 315| 318| assert(tileh > 0); ------------------ | Branch (315:5): [True: 318, False: 0] ------------------ 316| 318| const auto tilesY = 317| 318| implicit_cast(roundUpDivisionSafe(mRaw->dim.y, tileh)); 318| 318| if (!tilesY) ------------------ | Branch (318:9): [True: 0, False: 318] ------------------ 319| 318| ThrowRDE("Zero tiles vertically"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 320| | 321| 318| const TiffEntry* offsets = raw->getEntry(TiffTag::TILEOFFSETS); 322| 318| if (const TiffEntry* counts = raw->getEntry(TiffTag::TILEBYTECOUNTS); 323| 318| offsets->count != counts->count) { ------------------ | Branch (323:9): [True: 3, False: 315] ------------------ 324| 3| ThrowRDE("Tile count mismatch: offsets:%u count:%u", offsets->count, ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 325| 3| counts->count); 326| 3| } 327| | 328| | // tilesX * tilesY may overflow, but division is fine, so let's do that. 329| 315| if ((offsets->count / tilesX != tilesY || (offsets->count % tilesX != 0)) || ------------------ | Branch (329:10): [True: 11, False: 304] | Branch (329:47): [True: 1, False: 303] ------------------ 330| 303| (offsets->count / tilesY != tilesX || (offsets->count % tilesY != 0))) { ------------------ | Branch (330:10): [True: 0, False: 303] | Branch (330:47): [True: 0, False: 303] ------------------ 331| 12| ThrowRDE("Tile X/Y count mismatch: total:%u X:%u, Y:%u", offsets->count, ------------------ | | 38| 12| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 12| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 12| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 12| __VA_ARGS__) | | ------------------ ------------------ 332| 12| tilesX, tilesY); 333| 12| } 334| | 335| 303| return {mRaw->dim, tilew, tileh}; 336| 315| } 337| | 338| | // Strips 339| 3.15k| const TiffEntry* offsets = raw->getEntry(TiffTag::STRIPOFFSETS); 340| 3.15k| const TiffEntry* counts = raw->getEntry(TiffTag::STRIPBYTECOUNTS); 341| | 342| 3.15k| if (counts->count != offsets->count) { ------------------ | Branch (342:7): [True: 15, False: 3.14k] ------------------ 343| 15| ThrowRDE("Byte count number does not match strip size: " ------------------ | | 38| 15| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 15| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 15| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 15| __VA_ARGS__) | | ------------------ ------------------ 344| 15| "count:%u, stips:%u ", 345| 15| counts->count, offsets->count); 346| 15| } 347| | 348| 3.14k| uint32_t yPerSlice = raw->hasEntry(TiffTag::ROWSPERSTRIP) ------------------ | Branch (348:24): [True: 242, False: 2.89k] ------------------ 349| 3.14k| ? raw->getEntry(TiffTag::ROWSPERSTRIP)->getU32() 350| 3.14k| : mRaw->dim.y; 351| | 352| 3.14k| if (yPerSlice == 0 || ------------------ | Branch (352:7): [True: 21, False: 3.12k] ------------------ 353| 3.12k| roundUpDivisionSafe(mRaw->dim.y, yPerSlice) != counts->count) { ------------------ | Branch (353:7): [True: 7, False: 3.11k] ------------------ 354| 8| ThrowRDE("Invalid y per slice %u or strip count %u (height = %i)", ------------------ | | 38| 8| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 8| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 8| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 8| __VA_ARGS__) | | ------------------ ------------------ 355| 8| yPerSlice, counts->count, mRaw->dim.y); 356| 8| } 357| | 358| 3.13k| return {mRaw->dim, static_cast(mRaw->dim.x), yPerSlice}; 359| 3.14k|} _ZNK8rawspeed10DngDecoder10decodeDataEPKNS_7TiffIFDEj: 361| 3.47k|void DngDecoder::decodeData(const TiffIFD* raw, uint32_t sample_format) const { 362| 3.47k| if (compression == 8 && sample_format != 3) { ------------------ | Branch (362:7): [True: 0, False: 3.47k] | Branch (362:27): [True: 0, False: 0] ------------------ 363| 0| ThrowRDE("Only float format is supported for " ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 364| 0| "deflate-compressed data."); 365| 3.47k| } else if ((compression == 7 || compression == 0x884c) && ------------------ | Branch (365:15): [True: 531, False: 2.94k] | Branch (365:35): [True: 0, False: 2.94k] ------------------ 366| 531| sample_format != 1) { ------------------ | Branch (366:14): [True: 1, False: 530] ------------------ 367| 1| ThrowRDE("Only 16 bit unsigned data supported for " ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 368| 1| "JPEG-compressed data."); 369| 1| } 370| | 371| 3.47k| uint32_t predictor = ~0U; 372| 3.47k| if (raw->hasEntry(TiffTag::PREDICTOR)) ------------------ | Branch (372:7): [True: 1, False: 3.47k] ------------------ 373| 1| predictor = raw->getEntry(TiffTag::PREDICTOR)->getU32(); 374| | 375| 3.47k| if (mRaw->getDataType() == RawImageType::UINT16) { ------------------ | Branch (375:7): [True: 3.17k, False: 304] ------------------ 376| | // Default white level is (2 ** BitsPerSample) - 1 377| 3.17k| mRaw->whitePoint = implicit_cast((1UL << *bps) - 1UL); 378| 3.17k| } else if (mRaw->getDataType() == RawImageType::F32) { ------------------ | Branch (378:14): [True: 304, False: 0] ------------------ 379| | // 1. We divide by white level to normalize the image, 380| | // s.t. the 1.0 becomes the white level. 381| | // 2. In DNG spec, white level is always an integer, 382| | // there can not be a non-integral white level. 383| | // 3. Additionally, no white level for FP DNG's means that 384| | // it is pre-normalized (default is 1.0). 385| | // 386| | // Therefore, we can surmise that: 387| | // a) FP DNG's never have white level of <1.0 388| | // b) FP DNG's always have integral white level 389| | // c) if FP DNG has a white level of `w`, it is always correct 390| | // to divide by `float(w)` to normalize the image 391| | // ... therefore, we can, as an optimization, 392| | // store FP DNG white level as an integer. 393| 304| mRaw->whitePoint = 1; 394| 304| } 395| | 396| | // Some decompressors (such as VC5) may depend on the white point 397| 3.47k| if (raw->hasEntry(TiffTag::WHITELEVEL)) { ------------------ | Branch (397:7): [True: 148, False: 3.33k] ------------------ 398| 148| const TiffEntry* whitelevel = raw->getEntry(TiffTag::WHITELEVEL); 399| 148| if (whitelevel->isInt()) ------------------ | Branch (399:9): [True: 143, False: 5] ------------------ 400| 143| mRaw->whitePoint = whitelevel->getU32(); 401| 148| } 402| | 403| 3.47k| AbstractDngDecompressor slices(mRaw, getTilingDescription(raw), compression, 404| 3.47k| mFixLjpeg, *bps, predictor); 405| | 406| 3.47k| slices.slices.reserve(slices.dsc.numTiles); 407| | 408| 3.47k| const TiffEntry* offsets = nullptr; 409| 3.47k| const TiffEntry* counts = nullptr; 410| 3.47k| if (raw->hasEntry(TiffTag::TILEOFFSETS)) { ------------------ | Branch (410:7): [True: 303, False: 3.17k] ------------------ 411| 303| offsets = raw->getEntry(TiffTag::TILEOFFSETS); 412| 303| counts = raw->getEntry(TiffTag::TILEBYTECOUNTS); 413| 3.17k| } else { // Strips 414| 3.17k| offsets = raw->getEntry(TiffTag::STRIPOFFSETS); 415| 3.17k| counts = raw->getEntry(TiffTag::STRIPBYTECOUNTS); 416| 3.17k| } 417| 3.47k| assert(slices.dsc.numTiles == offsets->count); ------------------ | Branch (417:3): [True: 3.41k, False: 62] ------------------ 418| 3.47k| assert(slices.dsc.numTiles == counts->count); ------------------ | Branch (418:3): [True: 3.41k, False: 0] ------------------ 419| | 420| 3.41k| NORangesSet tilesLegality; 421| 7.99k| for (auto n = 0U; n < slices.dsc.numTiles; n++) { ------------------ | Branch (421:21): [True: 4.57k, False: 3.41k] ------------------ 422| 4.57k| const auto offset = offsets->getU32(n); 423| 4.57k| const auto count = counts->getU32(n); 424| | 425| 4.57k| if (count < 1) ------------------ | Branch (425:9): [True: 1, False: 4.57k] ------------------ 426| 4.57k| ThrowRDE("Tile %u is empty", n); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 427| | 428| 4.57k| ByteStream bs(DataBuffer(mFile.getSubView(offset, count), 429| 4.57k| mRootIFD->rootBuffer.getByteOrder())); 430| | 431| 4.57k| if (!tilesLegality.insert(bs)) ------------------ | Branch (431:9): [True: 2, False: 4.57k] ------------------ 432| 4.57k| ThrowTPE("Two tiles overlap. Raw corrupt!"); ------------------ | | 39| 2| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 433| | 434| 4.57k| slices.slices.emplace_back(slices.dsc, n, bs); 435| 4.57k| } 436| | 437| 3.41k| assert(slices.slices.size() == slices.dsc.numTiles); ------------------ | Branch (437:3): [True: 3.36k, False: 47] ------------------ 438| 3.36k| if (slices.slices.empty()) ------------------ | Branch (438:7): [True: 0, False: 3.36k] ------------------ 439| 3.36k| ThrowRDE("No valid slices found."); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 440| | 441| | // FIXME: should we sort the tiles, to linearize the input reading? 442| | 443| 3.36k| mRaw->createData(); 444| | 445| 3.36k| slices.decompress(); 446| 3.36k|} _ZN8rawspeed10DngDecoder17decodeRawInternalEv: 448| 3.93k|RawImage DngDecoder::decodeRawInternal() { 449| 3.93k| vector data = mRootIFD->getIFDsWithTag(TiffTag::COMPRESSION); 450| | 451| 3.93k| if (data.empty()) ------------------ | Branch (451:7): [True: 19, False: 3.91k] ------------------ 452| 3.93k| ThrowRDE("No image data found"); ------------------ | | 38| 19| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 19| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 19| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 19| __VA_ARGS__) | | ------------------ ------------------ 453| | 454| 3.91k| dropUnsuportedChunks(&data); 455| | 456| 3.91k| if (data.empty()) ------------------ | Branch (456:7): [True: 93, False: 3.82k] ------------------ 457| 3.91k| ThrowRDE("No RAW chunks found"); ------------------ | | 38| 93| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 93| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 93| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 93| __VA_ARGS__) | | ------------------ ------------------ 458| | 459| 3.82k| if (data.size() > 1) { ------------------ | Branch (459:7): [True: 29, False: 3.79k] ------------------ 460| 29| writeLog(DEBUG_PRIO::EXTRA, 461| 29| "Multiple RAW chunks found - using first only!"); 462| 29| } 463| | 464| 3.82k| const TiffIFD* raw = data[0]; 465| | 466| 3.82k| bps = raw->getEntry(TiffTag::BITSPERSAMPLE)->getU32(); 467| 3.82k| if (*bps < 1 || *bps > 32) ------------------ | Branch (467:7): [True: 60, False: 3.76k] | Branch (467:19): [True: 21, False: 3.74k] ------------------ 468| 3.82k| ThrowRDE("Unsupported bit per sample count: %i.", *bps); ------------------ | | 38| 42| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 42| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 42| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 42| __VA_ARGS__) | | ------------------ ------------------ 469| | 470| 3.78k| uint32_t sample_format = 1; 471| 3.78k| if (raw->hasEntry(TiffTag::SAMPLEFORMAT)) ------------------ | Branch (471:7): [True: 309, False: 3.47k] ------------------ 472| 309| sample_format = raw->getEntry(TiffTag::SAMPLEFORMAT)->getU32(); 473| | 474| 3.78k| compression = raw->getEntry(TiffTag::COMPRESSION)->getU16(); 475| | 476| 3.78k| switch (sample_format) { 477| 3.43k| case 1: ------------------ | Branch (477:3): [True: 3.43k, False: 348] ------------------ 478| 3.43k| mRaw = RawImage::create(RawImageType::UINT16); 479| 3.43k| break; 480| 308| case 3: ------------------ | Branch (480:3): [True: 308, False: 3.47k] ------------------ 481| 308| mRaw = RawImage::create(RawImageType::F32); 482| 308| break; 483| 1| default: ------------------ | Branch (483:3): [True: 1, False: 3.78k] ------------------ 484| 1| ThrowRDE("Only 16 bit unsigned or float point data supported. Sample " ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 485| 3.78k| "format %u is not supported.", 486| 3.78k| sample_format); 487| 3.78k| } 488| | 489| 3.74k| mRaw->isCFA = 490| 3.74k| (raw->getEntry(TiffTag::PHOTOMETRICINTERPRETATION)->getU16() == 32803); 491| | 492| 3.74k| if (mRaw->isCFA) ------------------ | Branch (492:7): [True: 351, False: 3.39k] ------------------ 493| 351| writeLog(DEBUG_PRIO::EXTRA, "This is a CFA image"); 494| 3.39k| else { 495| 3.39k| writeLog(DEBUG_PRIO::EXTRA, "This is NOT a CFA image"); 496| 3.39k| } 497| | 498| 3.74k| if (sample_format == 1 && *bps > 16) ------------------ | Branch (498:7): [True: 3.42k, False: 319] | Branch (498:29): [True: 6, False: 3.41k] ------------------ 499| 3.74k| ThrowRDE("Integer precision larger than 16 bits currently not supported."); ------------------ | | 38| 6| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 6| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 6| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 6| __VA_ARGS__) | | ------------------ ------------------ 500| | 501| 3.73k| if (sample_format == 3 && *bps != 16 && *bps != 24 && *bps != 32) ------------------ | Branch (501:7): [True: 308, False: 3.42k] | Branch (501:29): [True: 150, False: 158] | Branch (501:43): [True: 26, False: 124] | Branch (501:57): [True: 2, False: 24] ------------------ 502| 3.73k| ThrowRDE("Floating point must be 16/24/32 bits per sample."); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 503| | 504| 3.73k| mRaw->dim.x = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 505| 3.73k| mRaw->dim.y = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 506| | 507| 3.73k| if (!mRaw->dim.hasPositiveArea()) ------------------ | Branch (507:7): [True: 40, False: 3.69k] ------------------ 508| 3.73k| ThrowRDE("Image has zero size"); ------------------ | | 38| 40| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 40| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 40| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 40| __VA_ARGS__) | | ------------------ ------------------ 509| | 510| 3.69k|#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION 511| | // Yeah, sure, here it would be just dumb to leave this for production :) 512| 3.69k| if (mRaw->dim.x > 9536 || mRaw->dim.y > 7680) { ------------------ | Branch (512:7): [True: 35, False: 3.65k] | Branch (512:29): [True: 12, False: 3.64k] ------------------ 513| 29| ThrowRDE("Unexpected image dimensions found: (%i; %i)", mRaw->dim.x, ------------------ | | 38| 29| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 29| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 29| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 29| __VA_ARGS__) | | ------------------ ------------------ 514| 29| mRaw->dim.y); 515| 29| } 516| 3.66k|#endif 517| | 518| 3.66k| if (mRaw->isCFA) ------------------ | Branch (518:7): [True: 344, False: 3.32k] ------------------ 519| 344| parseCFA(raw); 520| | 521| 3.66k| uint32_t cpp = raw->getEntry(TiffTag::SAMPLESPERPIXEL)->getU32(); 522| | 523| 3.66k| if (cpp < 1 || cpp > 4) ------------------ | Branch (523:7): [True: 153, False: 3.51k] | Branch (523:18): [True: 32, False: 3.47k] ------------------ 524| 3.66k| ThrowRDE("Unsupported samples per pixel count: %u.", cpp); ------------------ | | 38| 33| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 33| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 33| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 33| __VA_ARGS__) | | ------------------ ------------------ 525| | 526| 3.63k| mRaw->setCpp(cpp); 527| | 528| | // Now load the image 529| 3.63k| decodeData(raw, sample_format); 530| | 531| 3.63k| handleMetadata(raw); 532| | 533| 3.63k| return mRaw; 534| 3.66k|} _ZN8rawspeed10DngDecoder14handleMetadataEPKNS_7TiffIFDE: 536| 2.54k|void DngDecoder::handleMetadata(const TiffIFD* raw) { 537| | // Crop 538| 2.54k| if (const Optional aa = parseACTIVEAREA(raw)) ------------------ | Branch (538:36): [True: 8, False: 2.53k] ------------------ 539| 8| mRaw->subFrame(*aa); 540| | 541| 2.54k| if (raw->hasEntry(TiffTag::DEFAULTCROPORIGIN) && ------------------ | Branch (541:7): [True: 292, False: 2.25k] ------------------ 542| 292| raw->hasEntry(TiffTag::DEFAULTCROPSIZE)) { ------------------ | Branch (542:7): [True: 270, False: 22] ------------------ 543| 270| iRectangle2D cropped(0, 0, mRaw->dim.x, mRaw->dim.y); 544| 270| const TiffEntry* origin_entry = raw->getEntry(TiffTag::DEFAULTCROPORIGIN); 545| 270| const TiffEntry* size_entry = raw->getEntry(TiffTag::DEFAULTCROPSIZE); 546| | 547| 270| const auto tl_r = origin_entry->getRationalArray(2); 548| 270| std::array tl; 549| 270| std::transform(tl_r.begin(), tl_r.end(), tl.begin(), 550| 270| [](const NotARational& r) { 551| 270| if (r.den == 0 || r.num % r.den != 0) 552| 270| ThrowRDE("Error decoding default crop origin"); 553| 270| return r.num / r.den; 554| 270| }); 555| | 556| 270| if (iPoint2D cropOrigin(tl[0], tl[1]); 557| 270| cropped.isPointInsideInclusive(cropOrigin)) ------------------ | Branch (557:9): [True: 15, False: 255] ------------------ 558| 15| cropped = iRectangle2D(cropOrigin, {0, 0}); 559| | 560| 270| cropped.dim = mRaw->dim - cropped.pos; 561| | 562| 270| const auto sz_r = size_entry->getRationalArray(2); 563| 270| std::array sz; 564| 270| std::transform(sz_r.begin(), sz_r.end(), sz.begin(), 565| 270| [](const NotARational& r) { 566| 270| if (r.den == 0 || r.num % r.den != 0) 567| 270| ThrowRDE("Error decoding default crop size"); 568| 270| return r.num / r.den; 569| 270| }); 570| | 571| 270| if (iPoint2D size(sz[0], sz[1]); 572| 270| size.isThisInside(mRaw->dim) && ------------------ | Branch (572:9): [True: 74, False: 196] | Branch (572:9): [True: 72, False: 198] ------------------ 573| 74| (size + cropped.pos).isThisInside(mRaw->dim)) ------------------ | Branch (573:9): [True: 72, False: 2] ------------------ 574| 72| cropped.dim = size; 575| | 576| 270| mRaw->subFrame(cropped); 577| 270| } 578| | 579| | // Adapt DNG DefaultScale to aspect-ratio 580| 2.54k| if (raw->hasEntry(TiffTag::DEFAULTSCALE)) { ------------------ | Branch (580:7): [True: 56, False: 2.48k] ------------------ 581| 56| const TiffEntry* default_scale = raw->getEntry(TiffTag::DEFAULTSCALE); 582| 56| const auto scales = default_scale->getRationalArray(2); 583| 110| for (const auto& scale : scales) { ------------------ | Branch (583:28): [True: 110, False: 49] ------------------ 584| 110| if (scale.num == 0 || scale.den == 0) ------------------ | Branch (584:11): [True: 2, False: 108] | Branch (584:29): [True: 5, False: 103] ------------------ 585| 110| ThrowRDE("Error decoding default pixel scale"); ------------------ | | 38| 7| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 7| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 7| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 7| __VA_ARGS__) | | ------------------ ------------------ 586| 110| } 587| 49| mRaw->metadata.pixelAspectRatio = 588| 49| static_cast(scales[0]) / static_cast(scales[1]); 589| 49| } 590| | 591| | // Apply stage 1 opcodes 592| 2.53k| if (applyStage1DngOpcodes && raw->hasEntry(TiffTag::OPCODELIST1)) { ------------------ | Branch (592:7): [True: 2.18k, False: 354] | Branch (592:32): [True: 1.02k, False: 1.15k] ------------------ 593| 1.02k| try { 594| 1.02k| const TiffEntry* opcodes = raw->getEntry(TiffTag::OPCODELIST1); 595| | // The entry might exist, but it might be empty, which means no opcodes 596| 1.02k| if (opcodes->count > 0) { ------------------ | Branch (596:11): [True: 1.02k, False: 1] ------------------ 597| 1.02k| DngOpcodes codes(mRaw, opcodes->getData()); 598| 1.02k| codes.applyOpCodes(mRaw); 599| 1.02k| } 600| 1.02k| } catch (const RawDecoderException& e) { 601| | // We push back errors from the opcode parser, since the image may still 602| | // be usable 603| 527| mRaw->setError(e.what()); 604| 527| } 605| 1.02k| } 606| | 607| | // Linearization 608| 2.53k| if (raw->hasEntry(TiffTag::LINEARIZATIONTABLE) && ------------------ | Branch (608:7): [True: 76, False: 2.32k] ------------------ 609| 76| raw->getEntry(TiffTag::LINEARIZATIONTABLE)->count > 0) { ------------------ | Branch (609:7): [True: 75, False: 1] ------------------ 610| 75| const TiffEntry* lintable = raw->getEntry(TiffTag::LINEARIZATIONTABLE); 611| 75| auto table = lintable->getU16Array(lintable->count); 612| 75| RawImageCurveGuard curveHandler(&mRaw, table, uncorrectedRawValues); 613| 75| if (!uncorrectedRawValues) ------------------ | Branch (613:9): [True: 66, False: 9] ------------------ 614| 66| mRaw->sixteenBitLookup(); 615| 75| } 616| | 617| | // Set black 618| 2.39k| setBlack(raw); 619| | 620| | // Apply opcodes to lossy DNG 621| 2.39k| if (compression == 0x884c && !uncorrectedRawValues && ------------------ | Branch (621:7): [True: 0, False: 2.39k] | Branch (621:32): [True: 0, False: 0] ------------------ 622| 0| raw->hasEntry(TiffTag::OPCODELIST2)) { ------------------ | Branch (622:7): [True: 0, False: 0] ------------------ 623| | // We must apply black/white scaling 624| 0| mRaw->scaleBlackWhite(); 625| | 626| | // Apply stage 2 codes 627| 0| try { 628| 0| DngOpcodes codes(mRaw, raw->getEntry(TiffTag::OPCODELIST2)->getData()); 629| 0| codes.applyOpCodes(mRaw); 630| 0| } catch (const RawDecoderException& e) { 631| | // We push back errors from the opcode parser, since the image may still 632| | // be usable 633| 0| mRaw->setError(e.what()); 634| 0| } 635| 0| mRaw->blackAreas.clear(); 636| 0| mRaw->blackLevel = 0; 637| 0| mRaw->blackLevelSeparate = 638| 0| Array2DRef(mRaw->blackLevelSeparateStorage.data(), 2, 2); 639| 0| auto blackLevelSeparate1D = *mRaw->blackLevelSeparate->getAsArray1DRef(); 640| 0| std::fill(blackLevelSeparate1D.begin(), blackLevelSeparate1D.end(), 0); 641| | // FIXME: why do we provide both the `blackLevel` and `blackLevelSeparate`? 642| 0| mRaw->whitePoint = 65535; 643| 0| } 644| 2.39k|} DngDecoder.cpp:_ZZN8rawspeed10DngDecoder14handleMetadataEPKNS_7TiffIFDEENK3$_0clERKNS_12NotARationalIjEE: 550| 523| [](const NotARational& r) { 551| 523| if (r.den == 0 || r.num % r.den != 0) ------------------ | Branch (551:26): [True: 2, False: 521] | Branch (551:40): [True: 49, False: 472] ------------------ 552| 523| ThrowRDE("Error decoding default crop origin"); ------------------ | | 38| 51| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 51| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 51| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 51| __VA_ARGS__) | | ------------------ ------------------ 553| 472| return r.num / r.den; 554| 523| }); DngDecoder.cpp:_ZZN8rawspeed10DngDecoder14handleMetadataEPKNS_7TiffIFDEENK3$_1clERKNS_12NotARationalIjEE: 565| 389| [](const NotARational& r) { 566| 389| if (r.den == 0 || r.num % r.den != 0) ------------------ | Branch (566:26): [True: 2, False: 387] | Branch (566:40): [True: 21, False: 366] ------------------ 567| 389| ThrowRDE("Error decoding default crop size"); ------------------ | | 38| 23| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 23| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 23| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 23| __VA_ARGS__) | | ------------------ ------------------ 568| 366| return r.num / r.den; 569| 389| }); _ZNK8rawspeed10DngDecoder17parseWhiteBalanceEv: 646| 1.66k|void DngDecoder::parseWhiteBalance() const { 647| | // Fetch the white balance 648| 1.66k| if (mRootIFD->hasEntryRecursive(TiffTag::ASSHOTNEUTRAL)) { ------------------ | Branch (648:7): [True: 55, False: 1.60k] ------------------ 649| 55| const TiffEntry* as_shot_neutral = 650| 55| mRootIFD->getEntryRecursive(TiffTag::ASSHOTNEUTRAL); 651| 55| if (as_shot_neutral->count == 3) { ------------------ | Branch (651:9): [True: 43, False: 12] ------------------ 652| 43| std::array wbCoeffs = {}; 653| 172| for (uint32_t i = 0; i < 3; i++) { ------------------ | Branch (653:28): [True: 129, False: 43] ------------------ 654| 129| float c = as_shot_neutral->getFloat(i); 655| 129| wbCoeffs[i] = (c > 0.0F) ? (1.0F / c) : 0.0F; ------------------ | Branch (655:23): [True: 71, False: 58] ------------------ 656| 129| } 657| 43| mRaw->metadata.wbCoeffs = wbCoeffs; 658| 43| } 659| 55| return; 660| 55| } 661| | 662| 1.60k| if (!mRaw->metadata.colorMatrix.empty() && ------------------ | Branch (662:7): [True: 19, False: 1.59k] ------------------ 663| 19| mRootIFD->hasEntryRecursive(TiffTag::ASSHOTWHITEXY)) { ------------------ | Branch (663:7): [True: 10, False: 9] ------------------ 664| 10| const TiffEntry* as_shot_white_xy = 665| 10| mRootIFD->getEntryRecursive(TiffTag::ASSHOTWHITEXY); 666| 10| assert(mRaw->metadata.colorMatrix.size() % 3 == 0); ------------------ | Branch (666:5): [True: 10, False: 0] ------------------ 667| 10| const auto numColorPlanes = mRaw->metadata.colorMatrix.size() / 3; 668| 10| if (numColorPlanes == 3 && as_shot_white_xy->count == 2) { ------------------ | Branch (668:9): [True: 5, False: 5] | Branch (668:32): [True: 4, False: 1] ------------------ 669| | // See http://www.brucelindbloom.com/index.html?Eqn_xyY_to_XYZ.html 670| 4| const float x = as_shot_white_xy->getFloat(0); 671| 4| const float y = as_shot_white_xy->getFloat(1); 672| 4| if (y > 0.0F) { ------------------ | Branch (672:11): [True: 3, False: 1] ------------------ 673| 3| constexpr float Y = 1.0F; 674| 3| const std::array as_shot_white = { 675| 3| {x * Y / y, Y, (1 - x - y) * Y / y}}; 676| | 677| | // Convert from XYZ to camera reference values first 678| 3| std::array wbCoeffs = {}; 679| | 680| 12| for (uint32_t i = 0; i < 3; i++) { ------------------ | Branch (680:30): [True: 9, False: 3] ------------------ 681| 9| float c = (float(mRaw->metadata.colorMatrix[(i * 3) + 0]) * 682| 9| as_shot_white[0]) + 683| 9| (float(mRaw->metadata.colorMatrix[(i * 3) + 1]) * 684| 9| as_shot_white[1]) + 685| 9| (float(mRaw->metadata.colorMatrix[(i * 3) + 2]) * 686| 9| as_shot_white[2]); 687| 9| wbCoeffs[i] = (c > 0.0F) ? (1.0F / c) : 0.0F; ------------------ | Branch (687:25): [True: 4, False: 5] ------------------ 688| 9| } 689| 3| mRaw->metadata.wbCoeffs = wbCoeffs; 690| 3| } 691| 4| } 692| 10| } 693| 1.60k|} _ZN8rawspeed10DngDecoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 695| 1.70k|void DngDecoder::decodeMetaDataInternal(const CameraMetaData* meta) { 696| 1.70k| if (mRootIFD->hasEntryRecursive(TiffTag::ISOSPEEDRATINGS)) ------------------ | Branch (696:7): [True: 10, False: 1.69k] ------------------ 697| 10| mRaw->metadata.isoSpeed = 698| 10| mRootIFD->getEntryRecursive(TiffTag::ISOSPEEDRATINGS)->getU32(); 699| | 700| 1.70k| TiffID id; 701| | 702| 1.70k| try { 703| 1.70k| id = mRootIFD->getID(); 704| 1.70k| } catch (const RawspeedException& e) { 705| 1.60k| mRaw->setError(e.what()); 706| | // not all dngs have MAKE/MODEL entries, 707| | // will be dealt with by using UNIQUECAMERAMODEL below 708| 1.60k| } 709| | 710| | // Set the make and model 711| 1.70k| mRaw->metadata.make = id.make; 712| 1.69k| mRaw->metadata.model = id.model; 713| | 714| 1.69k| const Camera* cam = meta->getCamera(id.make, id.model, "dng"); 715| 1.69k| if (!cam) // Also look for non-DNG cameras in case it's a converted file ------------------ | Branch (715:7): [True: 1.69k, False: 0] ------------------ 716| 1.69k| cam = meta->getCamera(id.make, id.model, ""); 717| 1.69k| if (!cam) // Worst case scenario, look for any such camera. ------------------ | Branch (717:7): [True: 1.69k, False: 0] ------------------ 718| 1.69k| cam = meta->getCamera(id.make, id.model); 719| 1.69k| if (cam) { ------------------ | Branch (719:7): [True: 0, False: 1.69k] ------------------ 720| 0| mRaw->metadata.canonical_make = cam->canonical_make; 721| 0| mRaw->metadata.canonical_model = cam->canonical_model; 722| 0| mRaw->metadata.canonical_alias = cam->canonical_alias; 723| 0| mRaw->metadata.canonical_id = cam->canonical_id; 724| 1.69k| } else { 725| 1.69k| mRaw->metadata.canonical_make = id.make; 726| 1.69k| mRaw->metadata.canonical_model = mRaw->metadata.canonical_alias = id.model; 727| 1.69k| if (mRootIFD->hasEntryRecursive(TiffTag::UNIQUECAMERAMODEL)) { ------------------ | Branch (727:9): [True: 16, False: 1.67k] ------------------ 728| 16| mRaw->metadata.canonical_id = 729| 16| mRootIFD->getEntryRecursive(TiffTag::UNIQUECAMERAMODEL)->getString(); 730| 1.67k| } else { 731| 1.67k| mRaw->metadata.canonical_id = id.make + " " + id.model; 732| 1.67k| } 733| 1.69k| } 734| | 735| 1.69k| parseColorMatrix(); 736| | 737| 1.69k| parseWhiteBalance(); 738| 1.69k|} _ZNK8rawspeed10DngDecoder17decodeMaskedAreasEPKNS_7TiffIFDE: 762| 238|bool DngDecoder::decodeMaskedAreas(const TiffIFD* raw) const { 763| 238| const TiffEntry* masked = raw->getEntry(TiffTag::MASKEDAREAS); 764| | 765| 238| if (masked->type != TiffDataType::SHORT && masked->type != TiffDataType::LONG) ------------------ | Branch (765:7): [True: 168, False: 70] | Branch (765:46): [True: 3, False: 165] ------------------ 766| 3| return false; 767| | 768| 235| uint32_t nrects = masked->count / 4; 769| 235| if (0 == nrects) ------------------ | Branch (769:7): [True: 2, False: 233] ------------------ 770| 2| return false; 771| | 772| | /* Since we may both have short or int, copy it to int array. */ 773| 233| auto rects = masked->getU32Array(nrects * 4); 774| | 775| 233| const iRectangle2D fullImage(0, 0, mRaw->getUncroppedDim().x, 776| 233| mRaw->getUncroppedDim().y); 777| 233| const iPoint2D top = mRaw->getCropOffset(); 778| | 779| 813| for (uint32_t i = 0; i < nrects; i++) { ------------------ | Branch (779:24): [True: 802, False: 11] ------------------ 780| 802| iPoint2D topleft(rects[(i * 4UL) + 1UL], rects[i * 4UL]); 781| 802| iPoint2D bottomright(rects[(i * 4UL) + 3UL], rects[(i * 4UL) + 2UL]); 782| | 783| 802| if (!(fullImage.isPointInsideInclusive(topleft) && ------------------ | Branch (783:11): [True: 688, False: 114] ------------------ 784| 688| fullImage.isPointInsideInclusive(bottomright) && ------------------ | Branch (784:11): [True: 589, False: 99] ------------------ 785| 589| (topleft < bottomright))) ------------------ | Branch (785:11): [True: 580, False: 9] ------------------ 786| 802| ThrowRDE("Bad masked area."); ------------------ | | 38| 222| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 222| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 222| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 222| __VA_ARGS__) | | ------------------ ------------------ 787| | 788| | // Is this a horizontal box, only add it if it covers the active width of 789| | // the image 790| 580| if (topleft.x <= top.x && bottomright.x >= (mRaw->dim.x + top.x)) { ------------------ | Branch (790:9): [True: 493, False: 87] | Branch (790:31): [True: 178, False: 315] ------------------ 791| 178| mRaw->blackAreas.emplace_back(topleft.y, bottomright.y - topleft.y, 792| 178| false); 793| 178| } 794| | // Is it a vertical box, only add it if it covers the active height of the 795| | // image 796| 402| else if (topleft.y <= top.y && bottomright.y >= (mRaw->dim.y + top.y)) { ------------------ | Branch (796:14): [True: 301, False: 101] | Branch (796:36): [True: 135, False: 166] ------------------ 797| 135| mRaw->blackAreas.emplace_back(topleft.x, bottomright.x - topleft.x, true); 798| 135| } 799| 580| } 800| 11| return !mRaw->blackAreas.empty(); 801| 233|} _ZNK8rawspeed10DngDecoder17decodeBlackLevelsEPKNS_7TiffIFDE: 803| 318|bool DngDecoder::decodeBlackLevels(const TiffIFD* raw) const { 804| 318| iPoint2D blackdim(1, 1); 805| 318| if (raw->hasEntry(TiffTag::BLACKLEVELREPEATDIM)) { ------------------ | Branch (805:7): [True: 149, False: 169] ------------------ 806| 149| const TiffEntry* bleveldim = raw->getEntry(TiffTag::BLACKLEVELREPEATDIM); 807| 149| if (bleveldim->count == 2) ------------------ | Branch (807:9): [True: 125, False: 24] ------------------ 808| 125| blackdim = iPoint2D(bleveldim->getU32(0), bleveldim->getU32(1)); 809| 24| else if (bleveldim->count == 1) { ------------------ | Branch (809:14): [True: 23, False: 1] ------------------ 810| | // Non-spec-compliant quirk. Assuming NxN repeat dimensions. 811| 23| blackdim = iPoint2D(bleveldim->getU32(0), bleveldim->getU32(0)); 812| | // Let's only allow somewhat unambiguous case of 1x1 repeat dimensions. 813| 23| if (blackdim != iPoint2D(1, 1)) ------------------ | Branch (813:11): [True: 21, False: 2] ------------------ 814| 21| return false; 815| 23| } else { 816| 1| return false; 817| 1| } 818| 149| } 819| | 820| 296| if (!blackdim.hasPositiveArea()) ------------------ | Branch (820:7): [True: 41, False: 255] ------------------ 821| 41| return false; 822| | 823| 255| if (!raw->hasEntry(TiffTag::BLACKLEVEL)) ------------------ | Branch (823:7): [True: 0, False: 255] ------------------ 824| 0| return true; 825| | 826| 255| if (mRaw->getCpp() != 1) ------------------ | Branch (826:7): [True: 2, False: 253] ------------------ 827| 2| return false; 828| | 829| 253| const TiffEntry* black_entry = raw->getEntry(TiffTag::BLACKLEVEL); 830| 253| if (black_entry->count < blackdim.area()) ------------------ | Branch (830:7): [True: 70, False: 183] ------------------ 831| 253| ThrowRDE("BLACKLEVEL entry is too small"); ------------------ | | 38| 70| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 70| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 70| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 70| __VA_ARGS__) | | ------------------ ------------------ 832| | 833| 183| using BlackType = decltype(mRaw->blackLevelSeparateStorage)::value_type; 834| | 835| 183| if (blackdim.x < 2 || blackdim.y < 2) { ------------------ | Branch (835:7): [True: 169, False: 14] | Branch (835:25): [True: 1, False: 13] ------------------ 836| | // We so not have enough to fill all individually, read a single and copy it 837| 170| float value = black_entry->getFloat(); 838| | 839| 170| if (static_cast(value) < std::numeric_limits::min() || ------------------ | Branch (839:9): [True: 8, False: 162] ------------------ 840| 162| static_cast(value) > std::numeric_limits::max()) ------------------ | Branch (840:9): [True: 1, False: 161] ------------------ 841| 170| ThrowRDE("Error decoding black level"); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 842| | 843| 167| mRaw->blackLevelSeparate = 844| 167| Array2DRef(mRaw->blackLevelSeparateStorage.data(), 2, 2); 845| 167| auto blackLevelSeparate1D = *mRaw->blackLevelSeparate->getAsArray1DRef(); 846| 489| for (int y = 0; y < 2; y++) { ------------------ | Branch (846:21): [True: 322, False: 167] ------------------ 847| 966| for (int x = 0; x < 2; x++) ------------------ | Branch (847:23): [True: 644, False: 322] ------------------ 848| 644| blackLevelSeparate1D((y * 2) + x) = implicit_cast(value); 849| 322| } 850| 167| } else { 851| 13| mRaw->blackLevelSeparate = 852| 13| Array2DRef(mRaw->blackLevelSeparateStorage.data(), 2, 2); 853| 13| auto blackLevelSeparate1D = *mRaw->blackLevelSeparate->getAsArray1DRef(); 854| 25| for (int y = 0; y < 2; y++) { ------------------ | Branch (854:21): [True: 21, False: 4] ------------------ 855| 50| for (int x = 0; x < 2; x++) { ------------------ | Branch (855:23): [True: 38, False: 12] ------------------ 856| 38| float value = black_entry->getFloat((y * blackdim.x) + x); 857| | 858| 38| if (static_cast(value) < ------------------ | Branch (858:13): [True: 5, False: 33] ------------------ 859| 38| std::numeric_limits::min() || 860| 33| static_cast(value) > std::numeric_limits::max()) ------------------ | Branch (860:13): [True: 4, False: 29] ------------------ 861| 38| ThrowRDE("Error decoding black level"); ------------------ | | 38| 9| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 9| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 9| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 9| __VA_ARGS__) | | ------------------ ------------------ 862| | 863| 29| blackLevelSeparate1D((y * 2) + x) = implicit_cast(value); 864| 29| } 865| 21| } 866| 13| } 867| | 868| | // DNG Spec says we must add black in deltav and deltah 869| 171| if (raw->hasEntry(TiffTag::BLACKLEVELDELTAV)) { ------------------ | Branch (869:7): [True: 29, False: 142] ------------------ 870| 29| const TiffEntry* blackleveldeltav = 871| 29| raw->getEntry(TiffTag::BLACKLEVELDELTAV); 872| 29| if (static_cast(blackleveldeltav->count) < mRaw->dim.y) ------------------ | Branch (872:9): [True: 1, False: 28] ------------------ 873| 29| ThrowRDE("BLACKLEVELDELTAV array is too small"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 874| 28| std::array black_sum = {{}}; 875| 1.34k| for (int i = 0; i < mRaw->dim.y; i++) ------------------ | Branch (875:21): [True: 1.31k, False: 28] ------------------ 876| 1.31k| black_sum[i & 1] += blackleveldeltav->getFloat(i); 877| | 878| 28| auto blackLevelSeparate1D = *mRaw->blackLevelSeparate->getAsArray1DRef(); 879| 110| for (int i = 0; i < 4; i++) { ------------------ | Branch (879:21): [True: 91, False: 19] ------------------ 880| 91| const float value = 881| 91| black_sum[i >> 1] / static_cast(mRaw->dim.y) * 2.0F; 882| 91| if (static_cast(value) < std::numeric_limits::min() || ------------------ | Branch (882:11): [True: 4, False: 87] ------------------ 883| 87| static_cast(value) > std::numeric_limits::max()) ------------------ | Branch (883:11): [True: 2, False: 85] ------------------ 884| 91| ThrowRDE("Error decoding black level"); ------------------ | | 38| 6| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 6| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 6| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 6| __VA_ARGS__) | | ------------------ ------------------ 885| | 886| 85| if (__builtin_sadd_overflow(blackLevelSeparate1D(i), ------------------ | Branch (886:11): [True: 3, False: 82] ------------------ 887| 85| implicit_cast(value), 888| 85| &blackLevelSeparate1D(i))) 889| 85| ThrowRDE("Integer overflow when calculating black level"); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 890| 85| } 891| 28| } 892| | 893| 161| if (raw->hasEntry(TiffTag::BLACKLEVELDELTAH)) { ------------------ | Branch (893:7): [True: 95, False: 66] ------------------ 894| 95| const TiffEntry* blackleveldeltah = 895| 95| raw->getEntry(TiffTag::BLACKLEVELDELTAH); 896| 95| if (static_cast(blackleveldeltah->count) < mRaw->dim.x) ------------------ | Branch (896:9): [True: 2, False: 93] ------------------ 897| 95| ThrowRDE("BLACKLEVELDELTAH array is too small"); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 898| 93| std::array black_sum = {{}}; 899| 3.64k| for (int i = 0; i < mRaw->dim.x; i++) ------------------ | Branch (899:21): [True: 3.54k, False: 93] ------------------ 900| 3.54k| black_sum[i & 1] += blackleveldeltah->getFloat(i); 901| | 902| 93| auto blackLevelSeparate1D = *mRaw->blackLevelSeparate->getAsArray1DRef(); 903| 422| for (int i = 0; i < 4; i++) { ------------------ | Branch (903:21): [True: 341, False: 81] ------------------ 904| 341| const float value = 905| 341| black_sum[i & 1] / static_cast(mRaw->dim.x) * 2.0F; 906| 341| if (static_cast(value) < std::numeric_limits::min() || ------------------ | Branch (906:11): [True: 3, False: 338] ------------------ 907| 338| static_cast(value) > std::numeric_limits::max()) ------------------ | Branch (907:11): [True: 4, False: 334] ------------------ 908| 341| ThrowRDE("Error decoding black level"); ------------------ | | 38| 7| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 7| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 7| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 7| __VA_ARGS__) | | ------------------ ------------------ 909| | 910| 334| if (__builtin_sadd_overflow(blackLevelSeparate1D(i), ------------------ | Branch (910:11): [True: 5, False: 329] ------------------ 911| 334| implicit_cast(value), 912| 334| &blackLevelSeparate1D(i))) 913| 334| ThrowRDE("Integer overflow when calculating black level"); ------------------ | | 38| 5| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 5| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 5| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 5| __VA_ARGS__) | | ------------------ ------------------ 914| 334| } 915| 93| } 916| 147| return true; 917| 161|} _ZNK8rawspeed10DngDecoder8setBlackEPKNS_7TiffIFDE: 919| 2.03k|void DngDecoder::setBlack(const TiffIFD* raw) const { 920| | 921| 2.03k| if (raw->hasEntry(TiffTag::MASKEDAREAS) && decodeMaskedAreas(raw)) ------------------ | Branch (921:7): [True: 238, False: 1.79k] | Branch (921:46): [True: 7, False: 231] ------------------ 922| 7| return; 923| | 924| | // Black defaults to 0 925| | // FIXME: is this the right thing to do? 926| 2.02k| mRaw->blackLevelSeparate = 927| 2.02k| Array2DRef(mRaw->blackLevelSeparateStorage.data(), 2, 2); 928| 2.02k| auto blackLevelSeparate1D = *mRaw->blackLevelSeparate->getAsArray1DRef(); 929| 2.02k| std::fill(blackLevelSeparate1D.begin(), blackLevelSeparate1D.end(), 0); 930| | 931| 2.02k| if (raw->hasEntry(TiffTag::BLACKLEVEL)) ------------------ | Branch (931:7): [True: 318, False: 1.71k] ------------------ 932| 318| decodeBlackLevels(raw); 933| 2.02k|} _ZN8rawspeed10ErfDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 44| 742| [[maybe_unused]] Buffer file) { 45| 742| const auto id = rootIFD->getID(); 46| 742| const std::string& make = id.make; 47| | 48| | // FIXME: magic 49| | 50| 742| return make == "SEIKO EPSON CORP."; 51| 742|} _ZN8rawspeed10ErfDecoder20checkImageDimensionsEv: 53| 50|void ErfDecoder::checkImageDimensions() { 54| 50| if (width > 3040 || height > 2024) ------------------ | Branch (54:7): [True: 19, False: 31] | Branch (54:23): [True: 6, False: 25] ------------------ 55| 50| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 25| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 25| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 25| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 25| __VA_ARGS__) | | ------------------ ------------------ 56| 50|} _ZN8rawspeed10ErfDecoder17decodeRawInternalEv: 58| 53|RawImage ErfDecoder::decodeRawInternal() { 59| 53| SimpleTiffDecoder::prepareForRawDecoding(); 60| | 61| 53| UncompressedDecompressor u( 62| 53| ByteStream(DataBuffer(mFile.getSubView(off, c2), Endianness::little)), 63| 53| mRaw, iRectangle2D({0, 0}, iPoint2D(width, height)), 64| 53| (12 * width / 8) + ((width + 2) / 10), 12, BitOrder::MSB); 65| 53| mRaw->createData(); 66| | 67| 53| u.decode12BitRawWithControl(); 68| | 69| 53| return mRaw; 70| 53|} _ZN8rawspeed10ErfDecoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 72| 22|void ErfDecoder::decodeMetaDataInternal(const CameraMetaData* meta) { 73| 22| setMetaData(meta, "", 0); 74| | 75| 22| if (mRootIFD->hasEntryRecursive(TiffTag::EPSONWB)) { ------------------ | Branch (75:7): [True: 3, False: 19] ------------------ 76| 3| const TiffEntry* wb = mRootIFD->getEntryRecursive(TiffTag::EPSONWB); 77| 3| if (wb->count == 256) { ------------------ | Branch (77:9): [True: 0, False: 3] ------------------ 78| | // Magic values taken directly from dcraw 79| 0| std::array wbCoeffs = {}; 80| 0| wbCoeffs[0] = static_cast(wb->getU16(24)) * 508.0F * 1.078F / 81| 0| static_cast(0x10000); 82| 0| wbCoeffs[1] = 1.0F; 83| 0| wbCoeffs[2] = static_cast(wb->getU16(25)) * 382.0F * 1.173F / 84| 0| static_cast(0x10000); 85| 0| mRaw->metadata.wbCoeffs = wbCoeffs; 86| 0| } 87| 3| } 88| 22|} _ZN8rawspeed10ErfDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 41| 53| : SimpleTiffDecoder(std::move(root), file) {} _ZN8rawspeed10IiqDecoder20isAppropriateDecoderENS_6BufferE: 59| 8.23k|bool IiqDecoder::isAppropriateDecoder(Buffer file) { 60| 8.23k| const DataBuffer db(file, Endianness::little); 61| | 62| | // The IIQ magic. Is present for all IIQ raws. 63| 8.23k| return db.get(8) == 0x49494949; 64| 8.23k|} _ZN8rawspeed10IiqDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 66| 8.26k|bool IiqDecoder::isAppropriateDecoder(const TiffRootIFD* rootIFD, Buffer file) { 67| 8.26k| const auto id = rootIFD->getID(); 68| 8.26k| const std::string& make = id.make; 69| | 70| 8.26k| return IiqDecoder::isAppropriateDecoder(file) && ------------------ | Branch (70:10): [True: 1.20k, False: 7.05k] ------------------ 71| 1.20k| (make == "Phase One A/S" || make == "Phase One" || make == "Leaf"); ------------------ | Branch (71:11): [True: 53, False: 1.15k] | Branch (71:38): [True: 946, False: 210] | Branch (71:61): [True: 159, False: 51] ------------------ 72| 8.26k|} _ZN8rawspeed10IiqDecoder13computeSripesENS_6BufferENSt3__16vectorINS0_9IiqOffsetENS2_9allocatorIS4_EEEEj: 77| 870| uint32_t height) { 78| 870| assert(height > 0); ------------------ | Branch (78:3): [True: 870, False: 0] ------------------ 79| 870| assert(offsets.size() == (1 + height)); ------------------ | Branch (79:3): [True: 870, False: 0] ------------------ 80| | 81| 870| ByteStream bs(DataBuffer(raw_data, Endianness::little)); 82| | 83| | // so... here's the thing. offsets are not guaranteed to be in 84| | // monotonically increasing order. so for each element of 'offsets', 85| | // we need to find element which specifies next larger offset. 86| | // and only then by subtracting those two offsets we get the slice size. 87| | 88| 870| std::sort(offsets.begin(), offsets.end(), 89| 870| [](const IiqOffset& a, const IiqOffset& b) { 90| 870| if (a.offset == b.offset && &a != &b) 91| 870| ThrowRDE("Two identical offsets found. Corrupt raw."); 92| 870| return a.offset < b.offset; 93| 870| }); 94| | 95| 870| std::vector slices; 96| 870| slices.reserve(height); 97| | 98| 870| auto offset_iterator = std::begin(offsets); 99| 870| bs.skipBytes(offset_iterator->offset); 100| | 101| 870| auto next_offset_iterator = std::next(offset_iterator); 102| 4.77k| while (next_offset_iterator < std::end(offsets)) { ------------------ | Branch (102:10): [True: 3.90k, False: 870] ------------------ 103| 3.90k| assert(next_offset_iterator->offset > offset_iterator->offset); ------------------ | Branch (103:5): [True: 3.90k, False: 0] ------------------ 104| 3.90k| const auto size = next_offset_iterator->offset - offset_iterator->offset; 105| 3.90k| assert(size > 0); ------------------ | Branch (105:5): [True: 3.90k, False: 0] ------------------ 106| | 107| 3.90k| slices.emplace_back(offset_iterator->n, bs.getStream(size)); 108| | 109| 3.90k| std::advance(offset_iterator, 1); 110| 3.90k| std::advance(next_offset_iterator, 1); 111| 3.90k| } 112| | 113| 870| assert(slices.size() == height); ------------------ | Branch (113:3): [True: 488, False: 382] ------------------ 114| | 115| 488| return slices; 116| 870|} IiqDecoder.cpp:_ZZN8rawspeed10IiqDecoder13computeSripesENS_6BufferENSt3__16vectorINS0_9IiqOffsetENS2_9allocatorIS4_EEEEjENK3$_0clERKS4_SA_: 89| 1.06M| [](const IiqOffset& a, const IiqOffset& b) { 90| 1.06M| if (a.offset == b.offset && &a != &b) ------------------ | Branch (90:19): [True: 151, False: 1.06M] | Branch (90:43): [True: 151, False: 0] ------------------ 91| 1.06M| ThrowRDE("Two identical offsets found. Corrupt raw."); ------------------ | | 38| 151| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 151| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 151| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 151| __VA_ARGS__) | | ------------------ ------------------ 92| 1.06M| return a.offset < b.offset; 93| 1.06M| }); _ZN8rawspeed10IiqDecoder17decodeRawInternalEv: 154| 1.15k|RawImage IiqDecoder::decodeRawInternal() { 155| 1.15k| const Buffer buf(mFile.getSubView(8)); 156| 1.15k| const DataBuffer db(buf, Endianness::little); 157| 1.15k| ByteStream bs(db); 158| | 159| 1.15k| bs.skipBytes(4); // Phase One magic 160| 1.15k| bs.skipBytes(4); // padding? 161| | 162| 1.15k| const auto origPos = bs.getPosition(); 163| | 164| 1.15k| const uint32_t entries_offset = bs.getU32(); 165| | 166| 1.15k| bs.setPosition(entries_offset); 167| | 168| 1.15k| const uint32_t entries_count = bs.getU32(); 169| 1.15k| bs.skipBytes(4); // ??? 170| | 171| | // this is how much is to be read for all the entries 172| 1.15k| ByteStream es(bs.getStream(entries_count, 16)); 173| | 174| 1.15k| bs.setPosition(origPos); 175| | 176| 1.15k| uint32_t width = 0; 177| 1.15k| uint32_t height = 0; 178| 1.15k| uint32_t split_row = 0; 179| 1.15k| uint32_t split_col = 0; 180| | 181| 1.15k| Optional format; 182| 1.15k| Optional raw_data; 183| 1.15k| ByteStream block_offsets; 184| 1.15k| ByteStream wb; 185| 1.15k| ByteStream correction_meta_data; 186| | 187| 142k| for (uint32_t entry = 0; entry < entries_count; entry++) { ------------------ | Branch (187:28): [True: 140k, False: 1.09k] ------------------ 188| 140k| const uint32_t tag = es.getU32(); 189| 140k| es.skipBytes(4); // type 190| 140k| const uint32_t len = es.getU32(); 191| 140k| const uint32_t data = es.getU32(); 192| | 193| 140k| switch (tag) { 194| 21| case 0x107: ------------------ | Branch (194:5): [True: 21, False: 140k] ------------------ 195| 21| wb = bs.getSubStream(data, len); 196| 21| break; 197| 1.58k| case 0x108: ------------------ | Branch (197:5): [True: 1.58k, False: 139k] ------------------ 198| 1.58k| width = data; 199| 1.58k| break; 200| 1.50k| case 0x109: ------------------ | Branch (200:5): [True: 1.50k, False: 139k] ------------------ 201| 1.50k| height = data; 202| 1.50k| break; 203| 960| case 0x10e: // RawFormat ------------------ | Branch (203:5): [True: 960, False: 140k] ------------------ 204| 960| if (format) ------------------ | Branch (204:11): [True: 2, False: 958] ------------------ 205| 960| ThrowRDE("Duplicate RawFormat tag."); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 206| 958| format = getAsIIQFormat(data); 207| 958| if (!format || *format != IIQFormat::IIQ_L) ------------------ | Branch (207:11): [True: 27, False: 931] | Branch (207:22): [True: 0, False: 931] ------------------ 208| 958| ThrowRDE("Unsupported RawFormat: %u", data); ------------------ | | 38| 27| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 27| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 27| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 27| __VA_ARGS__) | | ------------------ ------------------ 209| 931| break; 210| 1.10k| case 0x10f: ------------------ | Branch (210:5): [True: 1.10k, False: 139k] ------------------ 211| 1.10k| raw_data = bs.getSubView(data, len); 212| 1.10k| break; 213| 647| case 0x110: ------------------ | Branch (213:5): [True: 647, False: 140k] ------------------ 214| 647| correction_meta_data = bs.getSubStream(data); 215| 647| break; 216| 1.32k| case 0x21c: ------------------ | Branch (216:5): [True: 1.32k, False: 139k] ------------------ 217| | // they are not guaranteed to be sequential! 218| 1.32k| block_offsets = bs.getSubStream(data, len); 219| 1.32k| break; 220| 143| case 0x21d: ------------------ | Branch (220:5): [True: 143, False: 140k] ------------------ 221| | // 16-bit black level adapted to 14-bit raw data (IIQFormat::IIQ_L) 222| 143| black_level = data >> 2; 223| 143| break; 224| 189| case 0x222: ------------------ | Branch (224:5): [True: 189, False: 140k] ------------------ 225| 189| split_col = data; 226| 189| break; 227| 121| case 0x224: ------------------ | Branch (227:5): [True: 121, False: 140k] ------------------ 228| 121| split_row = data; 229| 121| break; 230| 133k| default: ------------------ | Branch (230:5): [True: 133k, False: 7.59k] ------------------ 231| | // FIXME: is there a "block_sizes" entry? 232| 133k| break; 233| 140k| } 234| 140k| } 235| | 236| | // FIXME: could be wrong. max "active pixels" in "Sensor+" mode - "101 MP" 237| 1.09k| if (width == 0 || height == 0 || width > 11976 || height > 8854) ------------------ | Branch (237:7): [True: 85, False: 1.01k] | Branch (237:21): [True: 44, False: 968] | Branch (237:36): [True: 8, False: 960] | Branch (237:53): [True: 44, False: 916] ------------------ 238| 1.09k| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 152| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 152| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 152| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 152| __VA_ARGS__) | | ------------------ ------------------ 239| | 240| 945| if (!format) ------------------ | Branch (240:7): [True: 9, False: 936] ------------------ 241| 945| ThrowRDE("Unspecified RawFormat"); ------------------ | | 38| 9| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 9| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 9| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 9| __VA_ARGS__) | | ------------------ ------------------ 242| | 243| 936| if (!raw_data) ------------------ | Branch (243:7): [True: 2, False: 934] ------------------ 244| 936| ThrowRDE("No raw data found"); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 245| | 246| 934| if (split_col > width || split_row > height) ------------------ | Branch (246:7): [True: 58, False: 876] | Branch (246:28): [True: 3, False: 873] ------------------ 247| 934| ThrowRDE("Invalid sensor quadrant split values (%u, %u)", split_row, ------------------ | | 38| 32| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 32| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 32| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 32| __VA_ARGS__) | | ------------------ ------------------ 248| 902| split_col); 249| | 250| 902| block_offsets = block_offsets.getStream(height, sizeof(uint32_t)); 251| | 252| 902| std::vector offsets; 253| 902| offsets.reserve(1 + height); 254| | 255| 105k| for (uint32_t row = 0; row < height; row++) ------------------ | Branch (255:26): [True: 104k, False: 902] ------------------ 256| 104k| offsets.emplace_back(row, block_offsets.getU32()); 257| | 258| | // to simplify slice size calculation, we insert a dummy offset, 259| | // which will be used much like end() 260| 902| offsets.emplace_back(height, raw_data->getSize()); 261| | 262| 902| std::vector strips( 263| 902| computeSripes(*raw_data, std::move(offsets), height)); 264| | 265| 902| mRaw->dim = iPoint2D(width, height); 266| | 267| 902| PhaseOneDecompressor p(mRaw, std::move(strips)); 268| 902| mRaw->createData(); 269| 902| p.decompress(); 270| | 271| 902| if (correction_meta_data.getSize() != 0 && iiq) ------------------ | Branch (271:7): [True: 326, False: 576] | Branch (271:46): [True: 326, False: 0] ------------------ 272| 326| CorrectPhaseOneC(correction_meta_data, split_row, split_col); 273| | 274| 902| std::array wbCoeffs = {}; 275| 1.02k| for (int i = 0; i < 3; i++) ------------------ | Branch (275:19): [True: 127, False: 902] ------------------ 276| 127| wbCoeffs[i] = wb.getFloat(); 277| 902| mRaw->metadata.wbCoeffs = wbCoeffs; 278| | 279| 902| return mRaw; 280| 934|} IiqDecoder.cpp:_ZN8rawspeed12_GLOBAL__N_114getAsIIQFormatEj: 130| 958|Optional getAsIIQFormat(uint32_t v) { 131| 958| switch (v) { 132| 0| using enum IIQFormat; 133| 0| case 1: ------------------ | Branch (133:3): [True: 0, False: 958] ------------------ 134| 0| return RAW_1; 135| 0| case 2: ------------------ | Branch (135:3): [True: 0, False: 958] ------------------ 136| 0| return RAW_2; 137| 931| case 3: ------------------ | Branch (137:3): [True: 931, False: 27] ------------------ 138| 931| return IIQ_L; 139| 0| case 5: ------------------ | Branch (139:3): [True: 0, False: 958] ------------------ 140| 0| return IIQ_S; 141| 0| case 6: ------------------ | Branch (141:3): [True: 0, False: 958] ------------------ 142| 0| return IIQ_Sv2; 143| 0| case 8: ------------------ | Branch (143:3): [True: 0, False: 958] ------------------ 144| 0| return IIQ_L16; 145| 27| default: ------------------ | Branch (145:3): [True: 27, False: 931] ------------------ 146| 27| return std::nullopt; 147| 958| } 148| 958|} _ZNK8rawspeed10IiqDecoder16CorrectPhaseOneCENS_10ByteStreamEjj: 283| 326| uint32_t split_col) const { 284| 326| meta_data.skipBytes(8); 285| 326| const uint32_t bytes_to_entries = meta_data.getU32(); 286| 326| meta_data.setPosition(bytes_to_entries); 287| 326| const uint32_t entries_count = meta_data.getU32(); 288| 326| meta_data.skipBytes(4); 289| | 290| | // this is how much is to be read for all the entries 291| 326| ByteStream entries(meta_data.getStream(entries_count, 12)); 292| 326| meta_data.setPosition(0); 293| | 294| 326| bool QuadrantMultipliersSeen = false; 295| 326| bool SensorDefectsSeen = false; 296| | 297| 24.6k| for (uint32_t entry = 0; entry < entries_count; entry++) { ------------------ | Branch (297:28): [True: 24.5k, False: 103] ------------------ 298| 24.5k| const uint32_t tag = entries.getU32(); 299| 24.5k| const uint32_t len = entries.getU32(); 300| 24.5k| const uint32_t offset = entries.getU32(); 301| | 302| 24.5k| switch (tag) { 303| 61| case 0x400: // Sensor Defects ------------------ | Branch (303:5): [True: 61, False: 24.4k] ------------------ 304| 61| if (SensorDefectsSeen) ------------------ | Branch (304:11): [True: 2, False: 59] ------------------ 305| 61| ThrowRDE("Second sensor defects entry seen. Unexpected."); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 306| 59| correctSensorDefects(meta_data.getSubStream(offset, len)); 307| 59| SensorDefectsSeen = true; 308| 59| break; 309| 216| case 0x40b: // Chroma calibration ------------------ | Branch (309:5): [True: 216, False: 24.3k] ------------------ 310| 216| PhaseOneFlatField(meta_data.getSubStream(offset, len), IiqCorr::CHROMA); 311| 216| break; 312| 311| case 0x410: // Luma calibration ------------------ | Branch (312:5): [True: 311, False: 24.2k] ------------------ 313| 311| PhaseOneFlatField(meta_data.getSubStream(offset, len), IiqCorr::LUMA); 314| 311| break; 315| 151| case 0x431: ------------------ | Branch (315:5): [True: 151, False: 24.3k] ------------------ 316| 151| if (QuadrantMultipliersSeen) ------------------ | Branch (316:11): [True: 5, False: 146] ------------------ 317| 151| ThrowRDE("Second quadrant multipliers entry seen. Unexpected."); ------------------ | | 38| 5| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 5| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 5| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 5| __VA_ARGS__) | | ------------------ ------------------ 318| 146| if (iiq.quadrantMultipliers) ------------------ | Branch (318:11): [True: 146, False: 0] ------------------ 319| 146| CorrectQuadrantMultipliersCombined(meta_data.getSubStream(offset, len), 320| 146| split_row, split_col); 321| 146| QuadrantMultipliersSeen = true; 322| 146| break; 323| 23.7k| default: ------------------ | Branch (323:5): [True: 23.7k, False: 739] ------------------ 324| 23.7k| break; 325| 24.5k| } 326| 24.5k| } 327| 326|} _ZNK8rawspeed10IiqDecoder34CorrectQuadrantMultipliersCombinedENS_10ByteStreamEjj: 339| 146| uint32_t split_col) const { 340| 146| std::array shared_x_coords; 341| | 342| | // Read the middle seven points from the file 343| 146| std::generate_n(std::next(shared_x_coords.begin()), 7, 344| 146| [&data] { return data.getU32(); }); 345| | 346| | // All the curves include (0, 0) and (65535, 65535), 347| | // so the first and last points are predefined 348| 146| shared_x_coords.front() = 0; 349| 146| shared_x_coords.back() = 65535; 350| | 351| | // Check that the middle coordinates make sense. 352| 146| if (std::adjacent_find(shared_x_coords.cbegin(), shared_x_coords.cend(), ------------------ | Branch (352:7): [True: 107, False: 39] ------------------ 353| 146| std::greater_equal<>()) != shared_x_coords.cend()) 354| 146| ThrowRDE("The X coordinates must all be strictly increasing"); ------------------ | | 38| 107| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 107| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 107| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 107| __VA_ARGS__) | | ------------------ ------------------ 355| | 356| 39| std::array, 2>, 2> control_points; 357| 64| for (auto& quadRow : control_points) { ------------------ | Branch (357:22): [True: 64, False: 22] ------------------ 358| 115| for (auto& quadrant : quadRow) { ------------------ | Branch (358:25): [True: 115, False: 47] ------------------ 359| 115| quadrant.reserve(9); 360| 115| quadrant.emplace_back(0, 0); 361| | 362| 838| for (int i = 1; i < 8; i++) { ------------------ | Branch (362:23): [True: 739, False: 99] ------------------ 363| | // These multipliers are expressed in ten-thousandths in the 364| | // file 365| 739| const uint64_t y_coord = 366| 739| (uint64_t(data.getU32()) * shared_x_coords[i]) / 10000ULL; 367| 739| if (y_coord > 65535) ------------------ | Branch (367:13): [True: 16, False: 723] ------------------ 368| 739| ThrowRDE("The Y coordinate %" PRIu64 " is too large", y_coord); ------------------ | | 38| 16| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 16| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 16| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 16| __VA_ARGS__) | | ------------------ ------------------ 369| 723| quadrant.emplace_back(shared_x_coords[i], y_coord); 370| 723| } 371| | 372| 99| quadrant.emplace_back(65535, 65535); 373| 99| assert(quadrant.size() == 9); ------------------ | Branch (373:7): [True: 98, False: 1] ------------------ 374| 99| } 375| 64| } 376| | 377| 66| for (int quadRow = 0; quadRow < 2; quadRow++) { ------------------ | Branch (377:25): [True: 44, False: 22] ------------------ 378| 132| for (int quadCol = 0; quadCol < 2; quadCol++) { ------------------ | Branch (378:27): [True: 88, False: 44] ------------------ 379| 88| const Array2DRef img(mRaw->getU16DataAsUncroppedArray2DRef()); 380| | 381| 88| const Spline<> s(control_points[quadRow][quadCol]); 382| 88| const std::vector curve = s.calculateCurve(); 383| | 384| 88| int row_start = quadRow == 0 ? 0 : split_row; ------------------ | Branch (384:23): [True: 44, False: 44] ------------------ 385| 88| int row_end = quadRow == 0 ? split_row : img.height(); ------------------ | Branch (385:21): [True: 44, False: 44] ------------------ 386| 88| int col_start = quadCol == 0 ? 0 : split_col; ------------------ | Branch (386:23): [True: 44, False: 44] ------------------ 387| 88| int col_end = quadCol == 0 ? split_col : img.width(); ------------------ | Branch (387:21): [True: 44, False: 44] ------------------ 388| | 389| 266| for (int row = row_start; row < row_end; row++) { ------------------ | Branch (389:33): [True: 178, False: 88] ------------------ 390| 1.75k| for (int col = col_start; col < col_end; col++) { ------------------ | Branch (390:35): [True: 1.57k, False: 178] ------------------ 391| 1.57k| uint16_t& pixel = img(row, col); 392| | // This adjustment is expected to be made with the 393| | // black-level already subtracted from the pixel values. 394| | // Because this is kept as metadata and not subtracted at 395| | // this point, to make the correction work we subtract the 396| | // appropriate amount before indexing into the curve and 397| | // then add it back so that subtracting the black level 398| | // later will work as expected 399| 1.57k| const uint16_t diff = pixel < black_level ------------------ | Branch (399:33): [True: 624, False: 954] ------------------ 400| 1.57k| ? pixel 401| 1.57k| : implicit_cast(black_level); 402| 1.57k| pixel = curve[pixel - diff] + diff; 403| 1.57k| } 404| 178| } 405| 88| } 406| 44| } 407| 22|} IiqDecoder.cpp:_ZZNK8rawspeed10IiqDecoder34CorrectQuadrantMultipliersCombinedENS_10ByteStreamEjjENK3$_0clEv: 344| 1.02k| [&data] { return data.getU32(); }); _ZNK8rawspeed10IiqDecoder17PhaseOneFlatFieldENS_10ByteStreamENS0_7IiqCorrE: 412| 516|void IiqDecoder::PhaseOneFlatField(ByteStream data, IiqCorr corr) const { 413| 516| const Array2DRef img(mRaw->getU16DataAsUncroppedArray2DRef()); 414| | 415| 516| int nc = [corr]() { 416| 516| switch (corr) { 417| 516| case IiqCorr::LUMA: 418| 516| return 2; 419| 516| case IiqCorr::CHROMA: 420| 516| return 4; 421| 516| } 422| 516| ThrowRDE("Unsupported IIQ correction"); 423| 516| }(); 424| | 425| 516| std::array head; 426| 4.64k| for (int i = 0; i < 8; i++) ------------------ | Branch (426:19): [True: 4.12k, False: 516] ------------------ 427| 4.12k| head[i] = data.getU16(); 428| | 429| 516| if (head[2] == 0 || head[3] == 0 || head[4] == 0 || head[5] == 0) ------------------ | Branch (429:7): [True: 63, False: 453] | Branch (429:23): [True: 53, False: 400] | Branch (429:39): [True: 68, False: 332] | Branch (429:55): [True: 66, False: 266] ------------------ 430| 250| return; 431| | 432| 266| auto wide = implicit_cast(roundUpDivisionSafe(head[2], head[4])); 433| 266| auto high = implicit_cast(roundUpDivisionSafe(head[3], head[5])); 434| | 435| 266| std::vector mrow_storage; 436| 266| Array2DRef mrow = Array2DRef::create( 437| 266| mrow_storage, /*width=*/wide * nc, /*height=*/1); 438| 266| mrow = Array2DRef(mrow_storage.data(), /*width=*/nc, /*height=*/wide); 439| | 440| 24.6k| for (int y = 0; y < high; y++) { ------------------ | Branch (440:19): [True: 24.4k, False: 266] ------------------ 441| 67.0k| for (int x = 0; x < wide; x++) { ------------------ | Branch (441:21): [True: 42.5k, False: 24.4k] ------------------ 442| 86.7k| for (int c = 0; c < nc; c += 2) { ------------------ | Branch (442:23): [True: 44.1k, False: 42.5k] ------------------ 443| 44.1k| float num = data.getU16() / 32768.0F; 444| 44.1k| if (y == 0) ------------------ | Branch (444:13): [True: 9.53k, False: 34.6k] ------------------ 445| 9.53k| mrow(x, c) = num; 446| 34.6k| else 447| 34.6k| mrow(x, c + 1) = (num - mrow(x, c)) / head[5]; 448| 44.1k| } 449| 42.5k| } 450| 24.4k| if (y == 0) ------------------ | Branch (450:9): [True: 257, False: 24.1k] ------------------ 451| 257| continue; 452| 24.1k| for (int rend = head[1] + (y * head[5]), row = rend - head[5]; 453| 25.6k| row < mRaw->dim.y && row < rend && row < (head[1] + head[3] - head[5]); ------------------ | Branch (453:10): [True: 1.59k, False: 24.0k] | Branch (453:31): [True: 1.46k, False: 126] | Branch (453:45): [True: 1.43k, False: 34] ------------------ 454| 24.1k| row++) { 455| 57.2k| for (int x = 1; x < wide; x++) { ------------------ | Branch (455:23): [True: 55.8k, False: 1.43k] ------------------ 456| 55.8k| std::array mult; 457| 112k| for (int c = 0; c < nc; c += 2) { ------------------ | Branch (457:25): [True: 56.7k, False: 55.8k] ------------------ 458| 56.7k| mult[c] = mrow(x - 1, c); 459| 56.7k| mult[c + 1] = (mrow(x, c) - mult[c]) / head[4]; 460| 56.7k| } 461| 55.8k| for (int cend = head[0] + (x * head[4]), col = cend - head[4]; 462| 58.9k| col < mRaw->dim.x && col < cend && ------------------ | Branch (462:14): [True: 3.40k, False: 55.5k] | Branch (462:35): [True: 3.31k, False: 85] ------------------ 463| 3.31k| col < head[0] + head[2] - head[4]; ------------------ | Branch (463:14): [True: 3.13k, False: 188] ------------------ 464| 55.8k| col++) { 465| 3.13k| if (int c = 466| 3.13k| nc > 2 ? static_cast(mRaw->cfa.getColorAt(row, col)) ------------------ | Branch (466:19): [True: 2, False: 3.12k] ------------------ 467| 3.13k| : 0; 468| 3.13k| !(c & 1)) { ------------------ | Branch (468:15): [True: 3.12k, False: 2] ------------------ 469| 3.12k| auto val = implicit_cast(img(row, col) * mult[c]); 470| 3.12k| img(row, col) = implicit_cast(std::min(val, 0xFFFFU)); 471| 3.12k| } 472| 6.25k| for (int c = 0; c < nc; c += 2) ------------------ | Branch (472:27): [True: 3.12k, False: 3.13k] ------------------ 473| 3.12k| mult[c] += mult[c + 1]; 474| 3.13k| } 475| 55.8k| } 476| 58.7k| for (int x = 0; x < wide; x++) ------------------ | Branch (476:23): [True: 57.2k, False: 1.43k] ------------------ 477| 116k| for (int c = 0; c < nc; c += 2) ------------------ | Branch (477:25): [True: 58.7k, False: 57.2k] ------------------ 478| 58.7k| mrow(x, c) += mrow(x, c + 1); 479| 1.43k| } 480| 24.1k| } 481| 266|} IiqDecoder.cpp:_ZZNK8rawspeed10IiqDecoder17PhaseOneFlatFieldENS_10ByteStreamENS0_7IiqCorrEENK3$_0clEv: 415| 516| int nc = [corr]() { 416| 516| switch (corr) { ------------------ | Branch (416:13): [True: 516, False: 0] ------------------ 417| 308| case IiqCorr::LUMA: ------------------ | Branch (417:5): [True: 308, False: 208] ------------------ 418| 308| return 2; 419| 208| case IiqCorr::CHROMA: ------------------ | Branch (419:5): [True: 208, False: 308] ------------------ 420| 208| return 4; 421| 516| } 422| 0| ThrowRDE("Unsupported IIQ correction"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 423| 516| }(); _ZN8rawspeed10IiqDecoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 494| 9|void IiqDecoder::decodeMetaDataInternal(const CameraMetaData* meta) { 495| 9| setMetaData(meta, "", 0); 496| | 497| 9| if (black_level) ------------------ | Branch (497:7): [True: 6, False: 3] ------------------ 498| 6| mRaw->blackLevel = black_level; 499| 9|} _ZNK8rawspeed10IiqDecoder20correctSensorDefectsENS_10ByteStreamE: 501| 54|void IiqDecoder::correctSensorDefects(ByteStream data) const { 502| 145k| while (data.getRemainSize() != 0) { ------------------ | Branch (502:10): [True: 145k, False: 13] ------------------ 503| 145k| const uint16_t col = data.getU16(); 504| 145k| const uint16_t row = data.getU16(); 505| 145k| const uint16_t type = data.getU16(); 506| 145k| data.skipBytes(2); // Ignore unknown/unused bits. 507| | 508| 145k| if (col >= mRaw->dim.x) // Value for col is outside the raw image. ------------------ | Branch (508:9): [True: 56.4k, False: 89.1k] ------------------ 509| 56.4k| continue; 510| 89.1k| switch (type) { 511| 124| case 131: // bad column ------------------ | Branch (511:5): [True: 124, False: 88.9k] ------------------ 512| 510| case 137: // bad column ------------------ | Branch (512:5): [True: 386, False: 88.7k] ------------------ 513| 510| correctBadColumn(col); 514| 510| break; 515| 940| case 129: // bad pixel ------------------ | Branch (515:5): [True: 940, False: 88.1k] ------------------ 516| 940| handleBadPixel(col, row); 517| 940| break; 518| 87.6k| default: // Oooh, a sensor defect not in dcraw! ------------------ | Branch (518:5): [True: 87.6k, False: 1.49k] ------------------ 519| 87.6k| break; 520| 89.1k| } 521| 89.1k| } 522| 54|} _ZNK8rawspeed10IiqDecoder14handleBadPixelEtt: 524| 940|void IiqDecoder::handleBadPixel(const uint16_t col, const uint16_t row) const { 525| 940| MutexLocker guard(&mRaw->mBadPixelMutex); 526| 940| mRaw->mBadPixelPositions.insert(mRaw->mBadPixelPositions.end(), 527| 940| (static_cast(row) << 16) + col); 528| 940|} _ZNK8rawspeed10IiqDecoder16correctBadColumnEt: 530| 510|void IiqDecoder::correctBadColumn(const uint16_t col) const { 531| 510| const Array2DRef img(mRaw->getU16DataAsUncroppedArray2DRef()); 532| | 533| 511| for (int row = 2; row < mRaw->dim.y - 2; row++) { ------------------ | Branch (533:21): [True: 1, False: 510] ------------------ 534| 1| if (mRaw->cfa.getColorAt(col, row) == CFAColor::GREEN) { ------------------ | Branch (534:9): [True: 0, False: 1] ------------------ 535| | /* Do green pixels. Let's pretend we are in "G" pixel, in the middle: 536| | * G=G 537| | * BGB 538| | * G0G 539| | * We accumulate the values 4 "G" pixels form diagonals, then check which 540| | * of 4 values is most distant from the mean of those 4 values, subtract 541| | * it from the sum, average (divide by 3) and round to nearest int. 542| | */ 543| 0| int max = 0; 544| 0| std::array val; 545| 0| std::array dev; 546| 0| int32_t sum = 0; 547| 0| sum += val[0] = img(row - 1, col - 1); 548| 0| sum += val[1] = img(row + 1, col - 1); 549| 0| sum += val[2] = img(row - 1, col + 1); 550| 0| sum += val[3] = img(row + 1, col + 1); 551| 0| for (int i = 0; i < 4; i++) { ------------------ | Branch (551:23): [True: 0, False: 0] ------------------ 552| 0| dev[i] = std::abs((val[i] * 4) - sum); 553| 0| if (dev[max] < dev[i]) ------------------ | Branch (553:13): [True: 0, False: 0] ------------------ 554| 0| max = i; 555| 0| } 556| 0| const int three_pixels = sum - val[max]; 557| | // This is `std::lround(three_pixels / 3.0)`, but without FP. 558| 0| img(row, col) = implicit_cast((three_pixels + 1) / 3); 559| 1| } else { 560| | /* 561| | * Do non-green pixels. Let's pretend we are in "R" pixel, in the middle: 562| | * RG=GR 563| | * GB=BG 564| | * RGRGR 565| | * GB0BG 566| | * RG0GR 567| | * We have 6 other "R" pixels - 2 by horizontal, 4 by diagonals. 568| | * We need to combine them, to get the value of the pixel we are in. 569| | */ 570| 1| uint32_t diags = img(row + 2, col - 2) + img(row - 2, col - 2) + 571| 1| img(row + 2, col + 2) + img(row - 2, col + 2); 572| 1| uint32_t horiz = img(row, col - 2) + img(row, col + 2); 573| | // But this is not just averaging, we bias towards the horizontal pixels. 574| 1| img(row, col) = implicit_cast( 575| 1| std::lround((diags * 0.0732233) + (horiz * 0.3535534))); 576| 1| } 577| 1| } 578| 510|} _ZN8rawspeed10IiqDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 57| 1.15k| : AbstractTiffDecoder(std::move(rootIFD), file) {} _ZN8rawspeed10IiqDecoder9IiqOffsetC2Ejj: 45| 105k| IiqOffset(uint32_t block, uint32_t offset_) : n(block), offset(offset_) {} _ZN8rawspeed10KdcDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 48| 829| [[maybe_unused]] Buffer file) { 49| 829| const auto id = rootIFD->getID(); 50| 829| const std::string& make = id.make; 51| | 52| | // FIXME: magic 53| | 54| 829| return make == "EASTMAN KODAK COMPANY"; 55| 829|} _ZNK8rawspeed10KdcDecoder14getInputBufferEv: 57| 34|Buffer KdcDecoder::getInputBuffer() const { 58| 34| const TiffEntry* offset = 59| 34| mRootIFD->getEntryRecursive(TiffTag::KODAK_KDC_OFFSET); 60| 34| if (!offset || offset->count < 13) ------------------ | Branch (60:7): [True: 1, False: 33] | Branch (60:18): [True: 1, False: 32] ------------------ 61| 34| ThrowRDE("Couldn't find the KDC offset"); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 62| | 63| 34| assert(offset != nullptr); ------------------ | Branch (63:3): [True: 32, False: 0] ------------------ 64| 32| uint64_t off = uint64_t(offset->getU32(4)) + uint64_t(offset->getU32(12)); 65| 32| if (off > std::numeric_limits::max()) ------------------ | Branch (65:7): [True: 3, False: 29] ------------------ 66| 32| ThrowRDE("Offset is too large."); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 67| | 68| | // Offset hardcoding gotten from dcraw 69| 29| if (hints.contains("easyshare_offset_hack")) ------------------ | Branch (69:7): [True: 0, False: 29] ------------------ 70| 0| off = off < 0x15000 ? 0x15000 : 0x17000; ------------------ | Branch (70:11): [True: 0, False: 0] ------------------ 71| | 72| 29| return mFile.getSubView(implicit_cast(off)); 73| 32|} _ZN8rawspeed10KdcDecoder17decodeRawInternalEv: 75| 87|RawImage KdcDecoder::decodeRawInternal() { 76| 87| if (!mRootIFD->hasEntryRecursive(TiffTag::COMPRESSION)) ------------------ | Branch (76:7): [True: 3, False: 84] ------------------ 77| 87| ThrowRDE("Couldn't find compression setting"); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 78| | 79| 84| if (auto compression = 80| 84| mRootIFD->getEntryRecursive(TiffTag::COMPRESSION)->getU32(); 81| 84| 7 != compression) ------------------ | Branch (81:7): [True: 25, False: 59] ------------------ 82| 84| ThrowRDE("Unsupported compression %u", compression); ------------------ | | 38| 25| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 25| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 25| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 25| __VA_ARGS__) | | ------------------ ------------------ 83| | 84| 59| const TiffEntry* ifdoffset = mRootIFD->getEntryRecursive(TiffTag::KODAK_IFD2); 85| 59| if (!ifdoffset) ------------------ | Branch (85:7): [True: 1, False: 58] ------------------ 86| 59| ThrowRDE("Couldn't find the Kodak IFD offset"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 87| | 88| 58| NORangesSet ifds; 89| | 90| 58| assert(ifdoffset != nullptr); ------------------ | Branch (90:3): [True: 58, False: 0] ------------------ 91| 58| TiffRootIFD kodakifd(nullptr, &ifds, ifdoffset->getRootIfdData(), 92| 58| ifdoffset->getU32()); 93| | 94| 58| const TiffEntry* ew = 95| 58| kodakifd.getEntryRecursive(TiffTag::KODAK_KDC_SENSOR_WIDTH); 96| 58| const TiffEntry* eh = 97| 58| kodakifd.getEntryRecursive(TiffTag::KODAK_KDC_SENSOR_HEIGHT); 98| 58| if (!ew || !eh) ------------------ | Branch (98:7): [True: 2, False: 56] | Branch (98:14): [True: 2, False: 54] ------------------ 99| 58| ThrowRDE("Unable to retrieve image size"); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 100| | 101| 55| uint32_t width = ew->getU32(); 102| 55| uint32_t height = eh->getU32(); 103| | 104| 55| mRaw->dim = iPoint2D(width, height); 105| | 106| 55| if (!mRaw->dim.hasPositiveArea() || !(mRaw->dim <= iPoint2D(4304, 3221))) ------------------ | Branch (106:7): [True: 7, False: 48] | Branch (106:7): [True: 19, False: 36] | Branch (106:39): [True: 14, False: 34] ------------------ 107| 55| ThrowRDE("Unexpected image dimensions found: (%d; %d)", mRaw->dim.x, ------------------ | | 38| 19| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 19| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 19| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 19| __VA_ARGS__) | | ------------------ ------------------ 108| 36| mRaw->dim.y); 109| | 110| 36| const Buffer inputBuffer = KdcDecoder::getInputBuffer(); 111| | 112| 36| UncompressedDecompressor u( 113| 36| ByteStream(DataBuffer(inputBuffer, Endianness::little)), mRaw, 114| 36| iRectangle2D({0, 0}, iPoint2D(width, height)), 12 * width / 8, 12, 115| 36| BitOrder::MSB); 116| 36| mRaw->createData(); 117| 36| u.readUncompressedRaw(); 118| | 119| 36| return mRaw; 120| 55|} _ZN8rawspeed10KdcDecoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 122| 11|void KdcDecoder::decodeMetaDataInternal(const CameraMetaData* meta) { 123| 11| setMetaData(meta, "", 0); 124| | 125| | // Try the kodak hidden IFD for WB 126| 11| if (mRootIFD->hasEntryRecursive(TiffTag::KODAK_IFD2)) { ------------------ | Branch (126:7): [True: 11, False: 0] ------------------ 127| 11| const TiffEntry* ifdoffset = 128| 11| mRootIFD->getEntryRecursive(TiffTag::KODAK_IFD2); 129| 11| try { 130| 11| NORangesSet ifds; 131| | 132| 11| TiffRootIFD kodakifd(nullptr, &ifds, ifdoffset->getRootIfdData(), 133| 11| ifdoffset->getU32()); 134| | 135| 11| if (kodakifd.hasEntryRecursive(TiffTag::KODAK_KDC_WB)) { ------------------ | Branch (135:11): [True: 3, False: 8] ------------------ 136| 3| const TiffEntry* wb = kodakifd.getEntryRecursive(TiffTag::KODAK_KDC_WB); 137| 3| if (wb->count == 3) { ------------------ | Branch (137:13): [True: 2, False: 1] ------------------ 138| 2| std::array wbCoeffs = {}; 139| 2| wbCoeffs[0] = wb->getFloat(0); 140| 2| wbCoeffs[1] = wb->getFloat(1); 141| 2| wbCoeffs[2] = wb->getFloat(2); 142| 2| mRaw->metadata.wbCoeffs = wbCoeffs; 143| 2| } 144| 3| } 145| 11| } catch (const TiffParserException& e) { 146| 2| mRaw->setError(e.what()); 147| 2| } 148| 11| } 149| | 150| | // Use the normal WB if available 151| 11| if (mRootIFD->hasEntryRecursive(TiffTag::KODAKWB)) { ------------------ | Branch (151:7): [True: 8, False: 3] ------------------ 152| 8| const TiffEntry* wb = mRootIFD->getEntryRecursive(TiffTag::KODAKWB); 153| 8| if (wb->count == 734 || wb->count == 1502) { ------------------ | Branch (153:9): [True: 1, False: 7] | Branch (153:29): [True: 1, False: 6] ------------------ 154| 2| std::array wbCoeffs = {}; 155| 2| wbCoeffs[0] = 156| 2| static_cast(((static_cast(wb->getByte(148))) << 8) | 157| 2| wb->getByte(149)) / 158| 2| 256.0F; 159| 2| wbCoeffs[1] = 1.0F; 160| 2| wbCoeffs[2] = 161| 2| static_cast(((static_cast(wb->getByte(150))) << 8) | 162| 2| wb->getByte(151)) / 163| 2| 256.0F; 164| 2| mRaw->metadata.wbCoeffs = wbCoeffs; 165| 2| } 166| 8| } 167| 11|} _ZN8rawspeed10KdcDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 40| 87| : AbstractTiffDecoder(std::move(root), file) {} _ZN8rawspeed10MefDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 38| 1.26k| [[maybe_unused]] Buffer file) { 39| 1.26k| const auto id = rootIFD->getID(); 40| 1.26k| const std::string& make = id.make; 41| | 42| | // FIXME: magic 43| | 44| 1.26k| return make == "Mamiya-OP Co.,Ltd."; 45| 1.26k|} _ZN8rawspeed10MefDecoder20checkImageDimensionsEv: 47| 58|void MefDecoder::checkImageDimensions() { 48| 58| if (width > 4016 || height > 5344) ------------------ | Branch (48:7): [True: 20, False: 38] | Branch (48:23): [True: 25, False: 13] ------------------ 49| 58| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 45| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 45| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 45| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 45| __VA_ARGS__) | | ------------------ ------------------ 50| 58|} _ZN8rawspeed10MefDecoder17decodeRawInternalEv: 52| 61|RawImage MefDecoder::decodeRawInternal() { 53| 61| SimpleTiffDecoder::prepareForRawDecoding(); 54| | 55| 61| UncompressedDecompressor u( 56| 61| ByteStream(DataBuffer(mFile.getSubView(off), Endianness::little)), mRaw, 57| 61| iRectangle2D({0, 0}, iPoint2D(width, height)), 12 * width / 8, 12, 58| 61| BitOrder::MSB); 59| 61| mRaw->createData(); 60| 61| u.readUncompressedRaw(); 61| | 62| 61| return mRaw; 63| 61|} _ZN8rawspeed10MefDecoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 65| 6|void MefDecoder::decodeMetaDataInternal(const CameraMetaData* meta) { 66| 6| setMetaData(meta, "", 0); 67| 6|} _ZN8rawspeed10MefDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 41| 61| : SimpleTiffDecoder(std::move(root), file) {} _ZN8rawspeed10MosDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 55| 8.86k|bool MosDecoder::isAppropriateDecoder(const TiffRootIFD* rootIFD, Buffer file) { 56| 8.86k| try { 57| 8.86k| const auto id = rootIFD->getID(); 58| 8.86k| const std::string& make = id.make; 59| | 60| | // This is messy. see https://github.com/darktable-org/rawspeed/issues/116 61| | // Old Leafs are MOS, new ones are IIQ. Use IIQ's magic to differentiate. 62| 8.86k| return make == "Leaf" && !IiqDecoder::isAppropriateDecoder(file); ------------------ | Branch (62:12): [True: 656, False: 8.20k] | Branch (62:30): [True: 497, False: 159] ------------------ 63| 8.86k| } catch (const TiffParserException&) { 64| | // Last ditch effort to identify Leaf cameras that don't have a Tiff Make 65| | // set 66| 785| const TiffEntry* softwareIFD = 67| 785| rootIFD->getEntryRecursive(TiffTag::SOFTWARE); 68| 785| if (!softwareIFD) ------------------ | Branch (68:9): [True: 517, False: 268] ------------------ 69| 517| return false; 70| | 71| 268| const std::string software = trimSpaces(softwareIFD->getString()); 72| 268| return software == "Camera Library"; 73| 785| } 74| 8.86k|} _ZN8rawspeed10MosDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 77| 581| : AbstractTiffDecoder(std::move(rootIFD), file) { 78| 581| if (mRootIFD->getEntryRecursive(TiffTag::MAKE)) { ------------------ | Branch (78:7): [True: 498, False: 83] ------------------ 79| 498| auto id = mRootIFD->getID(); 80| 498| make = id.make; 81| 498| model = id.model; 82| 498| } else { 83| 83| const TiffEntry* xmp = mRootIFD->getEntryRecursive(TiffTag::XMP); 84| 83| if (!xmp) ------------------ | Branch (84:9): [True: 1, False: 82] ------------------ 85| 83| ThrowRDE("Couldn't find the XMP"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 86| | 87| 83| assert(xmp != nullptr); ------------------ | Branch (87:5): [True: 82, False: 0] ------------------ 88| 82| std::string xmpText = xmp->getString(); 89| 82| make = getXMPTag(xmpText, "Make"); 90| 82| model = getXMPTag(xmpText, "Model"); 91| 82| } 92| 581|} _ZN8rawspeed10MosDecoder9getXMPTagENSt3__117basic_string_viewIcNS1_11char_traitsIcEEEES5_: 94| 140|std::string MosDecoder::getXMPTag(std::string_view xmp, std::string_view tag) { 95| 140| std::string::size_type start = xmp.find(""); 96| 140| std::string::size_type end = xmp.find(""); 97| 140| if (start == std::string::npos || end == std::string::npos || end <= start) ------------------ | Branch (97:7): [True: 42, False: 98] | Branch (97:37): [True: 13, False: 85] | Branch (97:65): [True: 1, False: 84] ------------------ 98| 140| ThrowRDE("Couldn't find tag '%s' in the XMP", tag.data()); ------------------ | | 38| 56| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 56| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 56| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 56| __VA_ARGS__) | | ------------------ ------------------ 99| 84| auto startlen = implicit_cast(tag.size() + 7); 100| 84| return std::string(xmp.substr(start + startlen, end - start - startlen)); 101| 140|} _ZN8rawspeed10MosDecoder17decodeRawInternalEv: 103| 522|RawImage MosDecoder::decodeRawInternal() { 104| 522| uint32_t off = 0; 105| | 106| 522| const TiffIFD* raw = nullptr; 107| | 108| 522| if (mRootIFD->hasEntryRecursive(TiffTag::TILEOFFSETS)) { ------------------ | Branch (108:7): [True: 452, False: 70] ------------------ 109| 452| raw = mRootIFD->getIFDWithTag(TiffTag::TILEOFFSETS); 110| 452| off = raw->getEntry(TiffTag::TILEOFFSETS)->getU32(); 111| 452| } else { 112| 70| raw = mRootIFD->getIFDWithTag(TiffTag::CFAPATTERN); 113| 70| off = raw->getEntry(TiffTag::STRIPOFFSETS)->getU32(); 114| 70| } 115| | 116| 522| uint32_t width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 117| 522| uint32_t height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 118| | 119| | // FIXME: could be wrong. max "active pixels" - "80 MP" 120| 522| if (width == 0 || height == 0 || width > 10328 || height > 7760) ------------------ | Branch (120:7): [True: 72, False: 450] | Branch (120:21): [True: 1, False: 449] | Branch (120:36): [True: 28, False: 421] | Branch (120:53): [True: 16, False: 405] ------------------ 121| 522| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 46| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 46| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 46| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 46| __VA_ARGS__) | | ------------------ ------------------ 122| | 123| 476| mRaw->dim = iPoint2D(width, height); 124| | 125| 476| const ByteStream bs(DataBuffer(mFile.getSubView(off), Endianness::little)); 126| 476| if (bs.getRemainSize() == 0) ------------------ | Branch (126:7): [True: 1, False: 475] ------------------ 127| 476| ThrowRDE("Input buffer is empty"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 128| | 129| 475| if (int compression = raw->getEntry(TiffTag::COMPRESSION)->getU32(); 130| 475| 1 == compression) { ------------------ | Branch (130:7): [True: 332, False: 143] ------------------ 131| 332| const Endianness endianness = 132| 332| getTiffByteOrder(ByteStream(DataBuffer(mFile, Endianness::little)), 0); 133| 332| UncompressedDecompressor u( 134| 332| bs, mRaw, iRectangle2D({0, 0}, iPoint2D(width, height)), 2 * width, 16, 135| 332| endianness == Endianness::big ? BitOrder::MSB : BitOrder::LSB); ------------------ | Branch (135:9): [True: 11, False: 321] ------------------ 136| 332| mRaw->createData(); 137| 332| u.readUncompressedRaw(); 138| 332| } else if (99 == compression || 7 == compression) { ------------------ | Branch (138:14): [True: 107, False: 36] | Branch (138:35): [True: 1, False: 35] ------------------ 139| 2| ThrowRDE("Leaf LJpeg not yet supported"); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 140| | // LJpegPlain l(mFile, mRaw); 141| | // l.startDecoder(off, mFile.getSize()-off, 0, 0); 142| 141| } else { 143| 141| ThrowRDE("Unsupported compression: %d", compression); ------------------ | | 38| 141| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 141| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 141| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 141| __VA_ARGS__) | | ------------------ ------------------ 144| 141| } 145| | 146| 332| return mRaw; 147| 475|} _ZN8rawspeed10MosDecoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 153| 327|void MosDecoder::decodeMetaDataInternal(const CameraMetaData* meta) { 154| 327| RawDecoder::setMetaData(meta, make, model, "", 0); 155| | 156| | // Fetch the white balance (see dcraw.c parse_mos for more metadata that can 157| | // be gotten) 158| 327| if (mRootIFD->hasEntryRecursive(TiffTag::LEAFMETADATA)) { ------------------ | Branch (158:7): [True: 322, False: 5] ------------------ 159| 322| ByteStream bs = 160| 322| mRootIFD->getEntryRecursive(TiffTag::LEAFMETADATA)->getData(); 161| | 162| | // We need at least a couple of bytes: 163| | // "NeutObj_neutrals" + 28 bytes binary + 4x uint as strings + 3x space + \0 164| 322| const uint32_t minSize = 16 + 28 + 4 + 3 + 1; 165| | 166| | // dcraw does actual parsing, since we just want one field we bruteforce it 167| 845k| while (bs.getRemainSize() > minSize) { ------------------ | Branch (167:12): [True: 845k, False: 51] ------------------ 168| 845k| if (bs.skipPrefix("NeutObj_neutrals")) { ------------------ | Branch (168:11): [True: 271, False: 845k] ------------------ 169| 271| bs.skipBytes(28); 170| | // check for nulltermination of string inside bounds 171| 271| if (!memchr(bs.peekData(bs.getRemainSize()), 0, bs.getRemainSize())) ------------------ | Branch (171:13): [True: 1, False: 270] ------------------ 172| 1| break; 173| 270| std::array tmp = {{}}; 174| 270| const std::string tmpString(bs.peekString()); 175| 270| std::istringstream iss(tmpString); 176| 270| iss >> tmp[0] >> tmp[1] >> tmp[2] >> tmp[3]; 177| 270| if (!iss.fail() && tmp[0] > 0 && tmp[1] > 0 && tmp[2] > 0 && ------------------ | Branch (177:13): [True: 96, False: 174] | Branch (177:28): [True: 89, False: 7] | Branch (177:42): [True: 79, False: 10] | Branch (177:56): [True: 68, False: 11] ------------------ 178| 68| tmp[3] > 0) { ------------------ | Branch (178:13): [True: 54, False: 14] ------------------ 179| 54| std::array wbCoeffs = {}; 180| 54| wbCoeffs[0] = 181| 54| static_cast(tmp[0]) / implicit_cast(tmp[1]); 182| 54| wbCoeffs[1] = 183| 54| static_cast(tmp[0]) / implicit_cast(tmp[2]); 184| 54| wbCoeffs[2] = 185| 54| static_cast(tmp[0]) / implicit_cast(tmp[3]); 186| 54| mRaw->metadata.wbCoeffs = wbCoeffs; 187| 54| } 188| 270| break; 189| 271| } 190| 845k| bs.skipBytes(1); 191| 845k| } 192| 322| } 193| 327|} _ZN8rawspeed10NefDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 64| 5.34k| [[maybe_unused]] Buffer file) { 65| 5.34k| const auto id = rootIFD->getID(); 66| 5.34k| const std::string& make = id.make; 67| | 68| | // FIXME: magic 69| | 70| 5.34k| return make == "NIKON CORPORATION" || make == "NIKON"; ------------------ | Branch (70:10): [True: 33, False: 5.31k] | Branch (70:41): [True: 830, False: 4.48k] ------------------ 71| 5.34k|} _ZN8rawspeed10NefDecoder17decodeRawInternalEv: 73| 863|RawImage NefDecoder::decodeRawInternal() { 74| 863| const auto* raw = mRootIFD->getIFDWithTag(TiffTag::CFAPATTERN); 75| 863| auto compression = raw->getEntry(TiffTag::COMPRESSION)->getU32(); 76| | 77| 863| const TiffEntry* offsets = raw->getEntry(TiffTag::STRIPOFFSETS); 78| 863| const TiffEntry* counts = raw->getEntry(TiffTag::STRIPBYTECOUNTS); 79| | 80| 863| if (mRootIFD->getEntryRecursive(TiffTag::MODEL)->getString() == ------------------ | Branch (80:7): [True: 42, False: 821] ------------------ 81| 863| "NIKON D100 ") { /**Sigh**/ 82| 42| if (!mFile.isValid(offsets->getU32())) ------------------ | Branch (82:9): [True: 18, False: 24] ------------------ 83| 42| ThrowRDE("Image data outside of file."); ------------------ | | 38| 18| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 18| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 18| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 18| __VA_ARGS__) | | ------------------ ------------------ 84| 24| if (!D100IsCompressed(offsets->getU32())) { ------------------ | Branch (84:9): [True: 7, False: 17] ------------------ 85| 7| DecodeD100Uncompressed(); 86| 7| return mRaw; 87| 7| } 88| 24| } 89| | 90| 838| if (compression == 1 || (hints.contains("force_uncompressed")) || ------------------ | Branch (90:7): [True: 114, False: 724] | Branch (90:7): [True: 207, False: 631] | Branch (90:27): [True: 0, False: 724] ------------------ 91| 724| NEFIsUncompressed(raw)) { ------------------ | Branch (91:7): [True: 120, False: 604] ------------------ 92| 207| DecodeUncompressed(); 93| 207| return mRaw; 94| 207| } 95| | 96| 631| if (NEFIsUncompressedRGB(raw)) { ------------------ | Branch (96:7): [True: 268, False: 363] ------------------ 97| 268| DecodeSNefUncompressed(); 98| 268| return mRaw; 99| 268| } 100| | 101| 363| if (offsets->count != 1) { ------------------ | Branch (101:7): [True: 13, False: 350] ------------------ 102| 13| ThrowRDE("Multiple Strips found: %u", offsets->count); ------------------ | | 38| 13| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 13| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 13| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 13| __VA_ARGS__) | | ------------------ ------------------ 103| 13| } 104| 350| if (counts->count != offsets->count) { ------------------ | Branch (104:7): [True: 68, False: 282] ------------------ 105| 68| ThrowRDE( ------------------ | | 38| 68| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 68| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 68| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 68| __VA_ARGS__) | | ------------------ ------------------ 106| 68| "Byte count number does not match strip size: count:%u, strips:%u ", 107| 68| counts->count, offsets->count); 108| 68| } 109| 282| if (!mFile.isValid(offsets->getU32(), counts->getU32())) ------------------ | Branch (109:7): [True: 32, False: 250] ------------------ 110| 282| ThrowRDE("Invalid strip byte count. File probably truncated."); ------------------ | | 38| 32| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 32| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 32| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 32| __VA_ARGS__) | | ------------------ ------------------ 111| | 112| 250| if (34713 != compression) ------------------ | Branch (112:7): [True: 4, False: 246] ------------------ 113| 250| ThrowRDE("Unsupported compression"); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 114| | 115| 246| uint32_t width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 116| 246| uint32_t height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 117| 246| uint32_t bitPerPixel = raw->getEntry(TiffTag::BITSPERSAMPLE)->getU32(); 118| | 119| 246| mRaw->dim = iPoint2D(width, height); 120| | 121| 246| const TiffEntry* meta = 122| 246| mRootIFD->getEntryRecursive(static_cast(0x96)); 123| 246| if (!meta) { ------------------ | Branch (123:7): [True: 100, False: 146] ------------------ 124| 100| meta = mRootIFD->getEntryRecursive(static_cast(0x8c)); // Fall back 125| 100| if (!meta) { ------------------ | Branch (125:9): [True: 1, False: 99] ------------------ 126| 1| ThrowRDE("Missing linearization table."); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 127| 1| } 128| 100| } 129| | 130| 245| const auto input = 131| 245| mFile.getSubView(offsets->getU32(), counts->getU32()).getAsArray1DRef(); 132| | 133| 245| NikonDecompressor n(mRaw, meta->getData(), bitPerPixel); 134| 245| mRaw->createData(); 135| 245| n.decompress(input, uncorrectedRawValues); 136| | 137| 245| return mRaw; 138| 246|} _ZNK8rawspeed10NefDecoder16D100IsCompressedEj: 145| 24|bool NefDecoder::D100IsCompressed(uint32_t offset) const { 146| 24| const auto test = mFile.getSubView(offset, 256); 147| 269| for (int i = 15; i < 256; i += 16) ------------------ | Branch (147:20): [True: 262, False: 7] ------------------ 148| 262| if (test[i]) ------------------ | Branch (148:9): [True: 17, False: 245] ------------------ 149| 17| return true; 150| | 151| 7| return false; 152| 24|} _ZN8rawspeed10NefDecoder17NEFIsUncompressedEPKNS_7TiffIFDE: 157| 821|bool NefDecoder::NEFIsUncompressed(const TiffIFD* raw) { 158| 821| const TiffEntry* counts = raw->getEntry(TiffTag::STRIPBYTECOUNTS); 159| 821| uint32_t width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 160| 821| uint32_t height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 161| 821| uint32_t bitPerPixel = raw->getEntry(TiffTag::BITSPERSAMPLE)->getU32(); 162| | 163| 821| if (!width || !height || !bitPerPixel) ------------------ | Branch (163:7): [True: 26, False: 795] | Branch (163:17): [True: 4, False: 791] | Branch (163:28): [True: 137, False: 654] ------------------ 164| 144| return false; 165| | 166| 677| const auto avaliableInputBytes = counts->getU32(0); 167| 677| const auto requiredPixels = iPoint2D(width, height).area(); 168| | 169| | // Now, there can be three situations. 170| | 171| | // We might have not enough input to produce the requested image size. 172| 677| const uint64_t avaliableInputBits = uint64_t(8) * avaliableInputBytes; 173| 677| const auto avaliablePixels = avaliableInputBits / bitPerPixel; // round down! 174| 677| if (avaliablePixels < requiredPixels) ------------------ | Branch (174:7): [True: 380, False: 297] ------------------ 175| 380| return false; 176| | 177| | // We might have exactly enough input with no padding whatsoever. 178| 297| if (avaliablePixels == requiredPixels) ------------------ | Branch (178:7): [True: 71, False: 226] ------------------ 179| 71| return true; 180| | 181| | // Or, we might have too much input. And sadly this is the worst case. 182| | // We can't just accept this. Some *compressed* NEF's also pass this check :( 183| | // Thus, let's accept *some* *small* padding. 184| 226| const auto requiredInputBits = bitPerPixel * requiredPixels; 185| 226| const auto requiredInputBytes = roundUpDivisionSafe(requiredInputBits, 8); 186| | // While we might have more *pixels* than needed, it does not nessesairly mean 187| | // that we have more input *bytes*. We might be off by a few pixels, and with 188| | // small image dimensions and bpp, we might still be in the same byte. 189| 226| assert(avaliableInputBytes >= requiredInputBytes); ------------------ | Branch (189:3): [True: 203, False: 23] ------------------ 190| 203| const auto totalPadding = avaliableInputBytes - requiredInputBytes; 191| 203| if (totalPadding % height != 0) ------------------ | Branch (191:7): [True: 62, False: 141] ------------------ 192| 62| return false; // Inconsistent padding makes no sense here. 193| 141| const auto perRowPadding = totalPadding / height; 194| 141| return perRowPadding < 16; 195| 203|} _ZN8rawspeed10NefDecoder20NEFIsUncompressedRGBEPKNS_7TiffIFDE: 200| 797|bool NefDecoder::NEFIsUncompressedRGB(const TiffIFD* raw) { 201| 797| uint32_t byteCount = raw->getEntry(TiffTag::STRIPBYTECOUNTS)->getU32(0); 202| 797| uint32_t width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 203| 797| uint32_t height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 204| | 205| 797| if (byteCount % 3 != 0) ------------------ | Branch (205:7): [True: 199, False: 598] ------------------ 206| 199| return false; 207| | 208| 598| return byteCount / 3 == iPoint2D(width, height).area(); 209| 797|} _ZNK8rawspeed10NefDecoder18DecodeUncompressedEv: 211| 207|void NefDecoder::DecodeUncompressed() const { 212| 207| const auto* raw = getIFDWithLargestImage(TiffTag::CFAPATTERN); 213| 207| const TiffEntry* offsets = raw->getEntry(TiffTag::STRIPOFFSETS); 214| 207| const TiffEntry* counts = raw->getEntry(TiffTag::STRIPBYTECOUNTS); 215| 207| uint32_t yPerSlice = raw->getEntry(TiffTag::ROWSPERSTRIP)->getU32(); 216| 207| uint32_t width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 217| 207| uint32_t height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 218| 207| uint32_t bitPerPixel = raw->getEntry(TiffTag::BITSPERSAMPLE)->getU32(); 219| | 220| 207| mRaw->dim = iPoint2D(width, height); 221| | 222| 207| if (width == 0 || height == 0 || width > 8288 || height > 5520) ------------------ | Branch (222:7): [True: 4, False: 203] | Branch (222:21): [True: 1, False: 202] | Branch (222:36): [True: 2, False: 200] | Branch (222:52): [True: 20, False: 180] ------------------ 223| 207| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 24| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 24| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 24| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 24| __VA_ARGS__) | | ------------------ ------------------ 224| | 225| 183| if (counts->count != offsets->count) { ------------------ | Branch (225:7): [True: 1, False: 182] ------------------ 226| 1| ThrowRDE("Byte count number does not match strip size: " ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 227| 1| "count:%u, stips:%u ", 228| 1| counts->count, offsets->count); 229| 1| } 230| | 231| 182| if (yPerSlice == 0 || yPerSlice > static_cast(mRaw->dim.y) || ------------------ | Branch (231:7): [True: 4, False: 178] | Branch (231:25): [True: 42, False: 136] ------------------ 232| 136| roundUpDivisionSafe(mRaw->dim.y, yPerSlice) != counts->count) { ------------------ | Branch (232:7): [True: 6, False: 130] ------------------ 233| 49| ThrowRDE("Invalid y per slice %u or strip count %u (height = %i)", ------------------ | | 38| 49| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 49| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 49| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 49| __VA_ARGS__) | | ------------------ ------------------ 234| 49| yPerSlice, counts->count, mRaw->dim.y); 235| 49| } 236| | 237| 133| vector slices; 238| 133| slices.reserve(counts->count); 239| 133| uint32_t offY = 0; 240| | 241| 255| for (uint32_t s = 0; s < counts->count; s++) { ------------------ | Branch (241:24): [True: 170, False: 85] ------------------ 242| 170| NefSlice slice; 243| 170| slice.offset = offsets->getU32(s); 244| 170| slice.count = counts->getU32(s); 245| | 246| 170| if (slice.count < 1) ------------------ | Branch (246:9): [True: 1, False: 169] ------------------ 247| 170| ThrowRDE("Slice %u is empty", s); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 248| | 249| 169| if (offY + yPerSlice > height) ------------------ | Branch (249:9): [True: 3, False: 166] ------------------ 250| 3| slice.h = height - offY; 251| 166| else 252| 166| slice.h = yPerSlice; 253| | 254| 169| offY = min(height, offY + yPerSlice); 255| | 256| 169| if (!mFile.isValid(slice.offset, slice.count)) ------------------ | Branch (256:9): [True: 47, False: 122] ------------------ 257| 169| ThrowRDE("Slice offset/count invalid"); ------------------ | | 38| 47| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 47| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 47| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 47| __VA_ARGS__) | | ------------------ ------------------ 258| | 259| 122| slices.push_back(slice); 260| 122| } 261| | 262| 85| if (slices.empty()) ------------------ | Branch (262:7): [True: 0, False: 85] ------------------ 263| 85| ThrowRDE("No valid slices found. File probably truncated."); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 264| | 265| 85| assert(height == offY); ------------------ | Branch (265:3): [True: 81, False: 4] ------------------ 266| 85| assert(slices.size() == counts->count); ------------------ | Branch (266:3): [True: 81, False: 0] ------------------ 267| | 268| 81| if (bitPerPixel == 14 && width * slices[0].h * 2 == slices[0].count) ------------------ | Branch (268:7): [True: 60, False: 21] | Branch (268:28): [True: 42, False: 18] ------------------ 269| 42| bitPerPixel = 16; // D3 & D810 270| | 271| 81| mRaw->createData(); 272| 81| bitPerPixel = hints.get("real_bpp", bitPerPixel); 273| | 274| 81| switch (bitPerPixel) { 275| 5| case 12: ------------------ | Branch (275:3): [True: 5, False: 76] ------------------ 276| 23| case 14: ------------------ | Branch (276:3): [True: 18, False: 63] ------------------ 277| 80| case 16: ------------------ | Branch (277:3): [True: 57, False: 24] ------------------ 278| 80| break; 279| 1| default: ------------------ | Branch (279:3): [True: 1, False: 80] ------------------ 280| 1| ThrowRDE("Invalid bpp found: %u", bitPerPixel); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 281| 81| } 282| | 283| 80| offY = 0; 284| 97| for (const NefSlice& slice : slices) { ------------------ | Branch (284:30): [True: 97, False: 78] ------------------ 285| 97| ByteStream in(DataBuffer(mFile.getSubView(slice.offset, slice.count), 286| 97| Endianness::little)); 287| 97| iPoint2D size(width, slice.h); 288| 97| iPoint2D pos(0, offY); 289| | 290| 97| if (hints.contains("coolpixmangled")) { ------------------ | Branch (290:9): [True: 0, False: 97] ------------------ 291| 0| UncompressedDecompressor u(in, mRaw, iRectangle2D(pos, size), 292| 0| width * bitPerPixel / 8, 12, BitOrder::MSB32); 293| 0| u.readUncompressedRaw(); 294| 97| } else { 295| 97| if (hints.contains("coolpixsplit")) { ------------------ | Branch (295:11): [True: 0, False: 97] ------------------ 296| 0| readCoolpixSplitRaw(in, size, pos, width * bitPerPixel / 8); 297| 97| } else { 298| 97| if (in.getSize() % size.y != 0) ------------------ | Branch (298:13): [True: 2, False: 95] ------------------ 299| 97| ThrowRDE("Inconsistent row size"); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 300| 95| const auto inputPitchBytes = in.getSize() / size.y; 301| 95| BitOrder bo = (mRootIFD->rootBuffer.getByteOrder() == Endianness::big) ^ ------------------ | Branch (301:23): [True: 11, False: 84] ------------------ 302| 95| hints.contains("msb_override") 303| 95| ? BitOrder::MSB 304| 95| : BitOrder::LSB; 305| 95| UncompressedDecompressor u(in, mRaw, iRectangle2D(pos, size), 306| 95| inputPitchBytes, bitPerPixel, bo); 307| 95| u.readUncompressedRaw(); 308| 95| } 309| 97| } 310| | 311| 95| offY += slice.h; 312| 95| } 313| 80|} _ZNK8rawspeed10NefDecoder22DecodeD100UncompressedEv: 360| 7|void NefDecoder::DecodeD100Uncompressed() const { 361| 7| const auto* ifd = mRootIFD->getIFDWithTag(TiffTag::STRIPOFFSETS, 1); 362| | 363| 7| uint32_t offset = ifd->getEntry(TiffTag::STRIPOFFSETS)->getU32(); 364| | // Hardcode the sizes as at least the width is not correctly reported 365| 7| uint32_t width = 3040; 366| 7| uint32_t height = 2024; 367| | 368| 7| mRaw->dim = iPoint2D(width, height); 369| | 370| 7| if (ByteStream bs(DataBuffer(mFile.getSubView(offset), Endianness::little)); 371| 7| bs.getRemainSize() == 0) ------------------ | Branch (371:7): [True: 1, False: 6] ------------------ 372| 7| ThrowRDE("No input to decode!"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 373| | 374| 6| UncompressedDecompressor u( 375| 6| ByteStream(DataBuffer(mFile.getSubView(offset), Endianness::little)), 376| 6| mRaw, iRectangle2D({0, 0}, iPoint2D(width, height)), 377| 6| (12 * width / 8) + ((width + 2) / 10), 12, BitOrder::MSB); 378| 6| mRaw->createData(); 379| | 380| 6| u.decode12BitRawWithControl(); 381| 6|} _ZNK8rawspeed10NefDecoder22DecodeSNefUncompressedEv: 383| 268|void NefDecoder::DecodeSNefUncompressed() const { 384| 268| const auto* raw = getIFDWithLargestImage(TiffTag::CFAPATTERN); 385| 268| uint32_t offset = raw->getEntry(TiffTag::STRIPOFFSETS)->getU32(); 386| 268| uint32_t width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 387| 268| uint32_t height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 388| | 389| 268| if (width == 0 || height == 0 || width % 2 != 0 || width > 3680 || ------------------ | Branch (389:7): [True: 3, False: 265] | Branch (389:21): [True: 1, False: 264] | Branch (389:36): [True: 13, False: 251] | Branch (389:54): [True: 31, False: 220] ------------------ 390| 220| height > 2456) ------------------ | Branch (390:7): [True: 15, False: 205] ------------------ 391| 268| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 61| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 61| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 61| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 61| __VA_ARGS__) | | ------------------ ------------------ 392| | 393| 207| mRaw->dim = iPoint2D(width, height); 394| 207| mRaw->setCpp(3); 395| 207| mRaw->isCFA = false; 396| 207| mRaw->createData(); 397| | 398| 207| ByteStream in(DataBuffer(mFile.getSubView(offset), Endianness::little)); 399| 207| DecodeNikonSNef(in); 400| 207|} _ZNK8rawspeed10NefDecoder15getBitPerSampleEv: 413| 75|int NefDecoder::getBitPerSample() const { 414| 75| const auto* raw = getIFDWithLargestImage(TiffTag::CFAPATTERN); 415| 75| return raw->getEntry(TiffTag::BITSPERSAMPLE)->getU32(); 416| 75|} _ZNK8rawspeed10NefDecoder7getModeEv: 418| 228|std::string NefDecoder::getMode() const { 419| 228| ostringstream mode; 420| 228| const auto* raw = getIFDWithLargestImage(TiffTag::CFAPATTERN); 421| 228| int compression = raw->getEntry(TiffTag::COMPRESSION)->getU32(); 422| 228| uint32_t bitPerPixel = raw->getEntry(TiffTag::BITSPERSAMPLE)->getU32(); 423| | 424| 228| if (NEFIsUncompressedRGB(raw)) ------------------ | Branch (424:7): [True: 81, False: 147] ------------------ 425| 81| mode << "sNEF-uncompressed"; 426| 147| else { 427| 147| if (1 == compression || NEFIsUncompressed(raw)) ------------------ | Branch (427:9): [True: 50, False: 97] | Branch (427:29): [True: 33, False: 64] ------------------ 428| 45| mode << bitPerPixel << "bit-uncompressed"; 429| 102| else 430| 102| mode << bitPerPixel << "bit-compressed"; 431| 147| } 432| 228| return mode.str(); 433| 228|} _ZNK8rawspeed10NefDecoder15getExtendedModeERKNSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEE: 435| 190|std::string NefDecoder::getExtendedMode(const std::string& mode) const { 436| 190| ostringstream extended_mode; 437| | 438| 190| const auto* ifd = mRootIFD->getIFDWithTag(TiffTag::CFAPATTERN); 439| 190| uint32_t width = ifd->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 440| 190| uint32_t height = ifd->getEntry(TiffTag::IMAGELENGTH)->getU32(); 441| | 442| 190| extended_mode << width << "x" << height << "-" << mode; 443| 190| return extended_mode.str(); 444| 190|} _ZNK8rawspeed10NefDecoder17parseWhiteBalanceEv: 494| 247|void NefDecoder::parseWhiteBalance() const { 495| | // Read the whitebalance 496| | 497| 247| if (mRootIFD->hasEntryRecursive(static_cast(12))) { ------------------ | Branch (497:7): [True: 162, False: 85] ------------------ 498| 162| const TiffEntry* wb = mRootIFD->getEntryRecursive(static_cast(12)); 499| 162| if (wb->count == 4) { ------------------ | Branch (499:9): [True: 160, False: 2] ------------------ 500| 160| std::array wbCoeffs = {}; 501| 160| wbCoeffs[0] = wb->getFloat(0); 502| 160| wbCoeffs[1] = wb->getFloat(2); 503| 160| wbCoeffs[2] = wb->getFloat(1); 504| 160| if (wbCoeffs[1] <= 0.0F) ------------------ | Branch (504:11): [True: 2, False: 158] ------------------ 505| 2| wbCoeffs[1] = 1.0F; 506| 160| mRaw->metadata.wbCoeffs = wbCoeffs; 507| 160| } 508| 162| } else if (mRootIFD->hasEntryRecursive(static_cast(0x0097))) { ------------------ | Branch (508:14): [True: 41, False: 44] ------------------ 509| 41| const TiffEntry* wb = 510| 41| mRootIFD->getEntryRecursive(static_cast(0x0097)); 511| 41| if (wb->count > 4) { ------------------ | Branch (511:9): [True: 40, False: 1] ------------------ 512| 40| uint32_t version = 0; 513| 179| for (uint32_t i = 0; i < 4; i++) { ------------------ | Branch (513:28): [True: 146, False: 33] ------------------ 514| 146| const auto v = wb->getByte(i); 515| 146| if (v < '0' || v > '9') ------------------ | Branch (515:13): [True: 4, False: 142] | Branch (515:24): [True: 4, False: 138] ------------------ 516| 146| ThrowRDE("Bad version component: %c - not a digit", v); ------------------ | | 38| 7| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 7| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 7| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 7| __VA_ARGS__) | | ------------------ ------------------ 517| 139| version = (version << 4) + v - '0'; 518| 139| } 519| | 520| 33| if (version == 0x100 && wb->count >= 80 && ------------------ | Branch (520:11): [True: 3, False: 30] | Branch (520:31): [True: 2, False: 1] ------------------ 521| 2| wb->type == TiffDataType::UNDEFINED) { ------------------ | Branch (521:11): [True: 1, False: 1] ------------------ 522| 1| std::array wbCoeffs = {}; 523| 1| wbCoeffs[0] = static_cast(wb->getU16(36)); 524| 1| wbCoeffs[2] = static_cast(wb->getU16(37)); 525| 1| wbCoeffs[1] = static_cast(wb->getU16(38)); 526| 1| mRaw->metadata.wbCoeffs = wbCoeffs; 527| 32| } else if (version == 0x103 && wb->count >= 26 && ------------------ | Branch (527:18): [True: 3, False: 29] | Branch (527:38): [True: 2, False: 1] ------------------ 528| 2| wb->type == TiffDataType::UNDEFINED) { ------------------ | Branch (528:18): [True: 1, False: 1] ------------------ 529| 1| std::array wbCoeffs = {}; 530| 1| wbCoeffs[0] = static_cast(wb->getU16(10)); 531| 1| wbCoeffs[1] = static_cast(wb->getU16(11)); 532| 1| wbCoeffs[2] = static_cast(wb->getU16(12)); 533| 1| mRaw->metadata.wbCoeffs = wbCoeffs; 534| 31| } else if (((version == 0x204 && wb->count >= 564) || ------------------ | Branch (534:20): [True: 1, False: 30] | Branch (534:40): [True: 0, False: 1] ------------------ 535| 30| (version == 0x205 && wb->count >= 284)) && ------------------ | Branch (535:20): [True: 18, False: 12] | Branch (535:40): [True: 17, False: 1] ------------------ 536| 17| mRootIFD->hasEntryRecursive(static_cast(0x001d)) && ------------------ | Branch (536:18): [True: 16, False: 1] ------------------ 537| 16| mRootIFD->hasEntryRecursive(static_cast(0x00a7))) { ------------------ | Branch (537:18): [True: 15, False: 1] ------------------ 538| | // Get the serial number 539| 15| std::string serial = 540| 15| mRootIFD->getEntryRecursive(static_cast(0x001d)) 541| 15| ->getString(); 542| 15| if (serial.length() > 9) ------------------ | Branch (542:13): [True: 2, False: 13] ------------------ 543| 15| ThrowRDE("Serial number is too long (%zu)", serial.length()); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 544| 13| uint32_t serialno = 0; 545| 72| for (unsigned char c : serial) { ------------------ | Branch (545:30): [True: 72, False: 13] ------------------ 546| 72| if (c >= '0' && c <= '9') ------------------ | Branch (546:15): [True: 44, False: 28] | Branch (546:27): [True: 21, False: 23] ------------------ 547| 21| serialno = serialno * 10 + c - '0'; 548| 51| else 549| 51| serialno = serialno * 10 + c % 10; 550| 72| } 551| | 552| | // Get the decryption key 553| 13| const TiffEntry* key = 554| 13| mRootIFD->getEntryRecursive(static_cast(0x00a7)); 555| 13| const auto keydata = key->getData().getBuffer(4); 556| 13| uint32_t keyno = keydata[0] ^ keydata[1] ^ keydata[2] ^ keydata[3]; 557| | 558| | // "Decrypt" the block using the serial and key 559| 13| uint8_t ci = serialmap[serialno & 0xff]; 560| 13| uint8_t cj = keymap[keyno & 0xff]; 561| 13| uint8_t ck = 0x60; 562| | 563| 13| ByteStream bs = wb->getData(); 564| 13| bs.skipBytes(version == 0x204 ? 284 : 4); ------------------ | Branch (564:22): [True: 0, False: 13] ------------------ 565| | 566| 13| std::array buf; 567| 88| for (unsigned char& i : buf) { ------------------ | Branch (567:31): [True: 88, False: 13] ------------------ 568| 88| cj = uint8_t(cj + (ci * ck)); // modulo arithmetics. 569| 88| i = bs.getByte() ^ cj; 570| 88| ck++; 571| 88| } 572| | 573| | // Finally set the WB coeffs 574| 13| uint32_t off = (version == 0x204) ? 6 : 14; ------------------ | Branch (574:24): [True: 0, False: 13] ------------------ 575| 13| std::array wbCoeffs = {}; 576| | 577| 13| wbCoeffs[0] = static_cast(getU16BE(&buf[off + 0])); 578| 13| wbCoeffs[1] = static_cast(getU16BE(&buf[off + 2])); 579| 13| wbCoeffs[2] = static_cast(getU16BE(&buf[off + 6])); 580| 13| mRaw->metadata.wbCoeffs = wbCoeffs; 581| 13| } 582| 33| } 583| 44| } else if (mRootIFD->hasEntryRecursive(static_cast(0x0014))) { ------------------ | Branch (583:14): [True: 13, False: 31] ------------------ 584| 13| const TiffEntry* wb = 585| 13| mRootIFD->getEntryRecursive(static_cast(0x0014)); 586| 13| ByteStream bs = wb->getData(); 587| 13| if (wb->count == 2560 && wb->type == TiffDataType::UNDEFINED) { ------------------ | Branch (587:9): [True: 3, False: 10] | Branch (587:30): [True: 1, False: 2] ------------------ 588| 1| bs.skipBytes(1248); 589| 1| bs.setByteOrder(Endianness::big); 590| 1| std::array wbCoeffs = {}; 591| 1| wbCoeffs[0] = static_cast(bs.getU16()) / 256.0F; 592| 1| wbCoeffs[1] = 1.0F; 593| 1| wbCoeffs[2] = static_cast(bs.getU16()) / 256.0F; 594| 1| mRaw->metadata.wbCoeffs = wbCoeffs; 595| 12| } else if (bs.hasPatternAt("NRW ", 0)) { ------------------ | Branch (595:16): [True: 3, False: 9] ------------------ 596| 3| uint32_t offset = 0; 597| 3| if (!bs.hasPatternAt("0100", 4) && wb->count > 72) ------------------ | Branch (597:11): [True: 3, False: 0] | Branch (597:42): [True: 1, False: 2] ------------------ 598| 1| offset = 56; 599| 2| else if (wb->count > 1572) ------------------ | Branch (599:16): [True: 0, False: 2] ------------------ 600| 0| offset = 1556; 601| | 602| 3| if (offset) { ------------------ | Branch (602:11): [True: 1, False: 2] ------------------ 603| 1| bs.skipBytes(offset); 604| 1| bs.setByteOrder(Endianness::little); 605| 1| std::array wbCoeffs = {}; 606| 1| wbCoeffs[0] = 4.0F * implicit_cast(bs.getU32()); 607| 1| wbCoeffs[1] = implicit_cast(bs.getU32()); 608| 1| wbCoeffs[1] += implicit_cast(bs.getU32()); 609| 1| wbCoeffs[2] = 4.0F * implicit_cast(bs.getU32()); 610| 1| mRaw->metadata.wbCoeffs = wbCoeffs; 611| 1| } 612| 3| } 613| 13| } 614| | 615| 238| if (hints.contains("nikon_wb_adjustment") && mRaw->metadata.wbCoeffs) { ------------------ | Branch (615:7): [True: 0, False: 238] | Branch (615:7): [True: 0, False: 238] | Branch (615:48): [True: 0, False: 0] ------------------ 616| 0| (*mRaw->metadata.wbCoeffs)[0] *= 256.0F / 527.0F; 617| 0| (*mRaw->metadata.wbCoeffs)[2] *= 256.0F / 317.0F; 618| 0| } 619| 238|} _ZN8rawspeed10NefDecoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 621| 247|void NefDecoder::decodeMetaDataInternal(const CameraMetaData* meta) { 622| 247| int iso = 0; 623| 247| mRaw->cfa.setCFA(iPoint2D(2, 2), CFAColor::RED, CFAColor::GREEN, 624| 247| CFAColor::GREEN, CFAColor::BLUE); 625| | 626| 247| auto makernotesWhite = mRaw->whitePoint; 627| 247| int makernotesBlack = mRaw->blackLevel; 628| | 629| 247| if (mRootIFD->hasEntryRecursive(TiffTag::ISOSPEEDRATINGS)) ------------------ | Branch (629:7): [True: 2, False: 245] ------------------ 630| 2| iso = mRootIFD->getEntryRecursive(TiffTag::ISOSPEEDRATINGS)->getU32(); 631| | 632| 247| parseWhiteBalance(); 633| | 634| 247| auto id = mRootIFD->getID(); 635| 247| std::string mode = getMode(); 636| 247| std::string extended_mode = getExtendedMode(mode); 637| | 638| | // Read black levels (seem to be recorded for 14bit always) 639| 247| if (mRootIFD->hasEntryRecursive(TiffTag::NIKON_BLACKLEVEL)) { ------------------ | Branch (639:7): [True: 84, False: 163] ------------------ 640| 84| const TiffEntry* bl = 641| 84| mRootIFD->getEntryRecursive(TiffTag::NIKON_BLACKLEVEL); 642| 84| if (bl->count != 4) ------------------ | Branch (642:9): [True: 9, False: 75] ------------------ 643| 84| ThrowRDE("BlackLevel has %u entries instead of 4", bl->count); ------------------ | | 38| 9| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 9| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 9| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 9| __VA_ARGS__) | | ------------------ ------------------ 644| 75| uint32_t bitPerPixel = getBitPerSample(); 645| 75| if (bitPerPixel != 12 && bitPerPixel != 14) ------------------ | Branch (645:9): [True: 74, False: 1] | Branch (645:30): [True: 71, False: 3] ------------------ 646| 75| ThrowRDE("Bad bit per pixel: %u", bitPerPixel); ------------------ | | 38| 71| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 71| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 71| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 71| __VA_ARGS__) | | ------------------ ------------------ 647| 4| const int sh = 14 - bitPerPixel; 648| 4| mRaw->blackLevelSeparate = 649| 4| Array2DRef(mRaw->blackLevelSeparateStorage.data(), 2, 2); 650| 4| auto blackLevelSeparate1D = *mRaw->blackLevelSeparate->getAsArray1DRef(); 651| 4| blackLevelSeparate1D(0) = bl->getU16(0) >> sh; 652| 4| blackLevelSeparate1D(1) = bl->getU16(1) >> sh; 653| 4| blackLevelSeparate1D(2) = bl->getU16(2) >> sh; 654| 4| blackLevelSeparate1D(3) = bl->getU16(3) >> sh; 655| 4| } 656| | 657| 167| if (meta->hasCamera(id.make, id.model, extended_mode)) { ------------------ | Branch (657:7): [True: 0, False: 167] ------------------ 658| 0| setMetaData(meta, id, extended_mode, iso); 659| 167| } else if (meta->hasCamera(id.make, id.model, mode)) { ------------------ | Branch (659:14): [True: 0, False: 167] ------------------ 660| 0| setMetaData(meta, id, mode, iso); 661| 167| } else { 662| 167| setMetaData(meta, id, "", iso); 663| 167| } 664| | 665| 167| if (makernotesWhite) ------------------ | Branch (665:7): [True: 0, False: 167] ------------------ 666| 0| mRaw->whitePoint = *makernotesWhite; 667| 167| if (makernotesBlack != -1) ------------------ | Branch (667:7): [True: 0, False: 167] ------------------ 668| 0| mRaw->blackLevel = makernotesBlack; 669| 167|} _ZNK8rawspeed10NefDecoder15DecodeNikonSNefENS_10ByteStreamE: 675| 174|void NefDecoder::DecodeNikonSNef(ByteStream input) const { 676| 174| if (mRaw->dim.x < 6) ------------------ | Branch (676:7): [True: 1, False: 173] ------------------ 677| 174| ThrowIOE("got a %i wide sNEF, aborting", mRaw->dim.x); ------------------ | | 37| 1|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 678| | 679| | // We need to read the applied whitebalance, since we should return 680| | // data before whitebalance, so we "unapply" it. 681| 173| const TiffEntry* wb = mRootIFD->getEntryRecursive(static_cast(12)); 682| 173| if (!wb) ------------------ | Branch (682:7): [True: 1, False: 172] ------------------ 683| 173| ThrowRDE("Unable to locate whitebalance needed for decompression"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 684| | 685| 173| assert(wb != nullptr); ------------------ | Branch (685:3): [True: 172, False: 0] ------------------ 686| 172| if (wb->count != 4 || wb->type != TiffDataType::RATIONAL) ------------------ | Branch (686:7): [True: 1, False: 171] | Branch (686:25): [True: 1, False: 170] ------------------ 687| 172| ThrowRDE("Whitebalance has unknown count or type"); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 688| | 689| 170| float wb_r = wb->getFloat(0); 690| 170| float wb_b = wb->getFloat(1); 691| | 692| | // ((1024/x)*((1<<16)-1)+(1<<9))<=((1<<31)-1), x>0 gives: (0.0312495) 693| 170| if (const auto lower_limit = 694| 170| implicit_cast(13'421'568.0 / 429'496'627.0); 695| 170| wb_r < lower_limit || wb_b < lower_limit || wb_r > 10.0F || wb_b > 10.0F) ------------------ | Branch (695:7): [True: 1, False: 169] | Branch (695:29): [True: 1, False: 168] | Branch (695:51): [True: 1, False: 167] | Branch (695:67): [True: 1, False: 166] ------------------ 696| 170| ThrowRDE("Whitebalance has bad values (%f, %f)", ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 697| 166| implicit_cast(wb_r), implicit_cast(wb_b)); 698| | 699| 166| std::array wbCoeffs = {}; 700| 166| wbCoeffs[0] = wb_r; 701| 166| wbCoeffs[1] = 1.0F; 702| 166| wbCoeffs[2] = wb_b; 703| 166| mRaw->metadata.wbCoeffs = wbCoeffs; 704| | 705| 166| auto inv_wb_r = static_cast(1024.0F / wb_r); 706| 166| auto inv_wb_b = static_cast(1024.0F / wb_b); 707| | 708| 166| auto curve = gammaCurve(1 / 2.4, 12.92, 4095); 709| | 710| | // Scale output values to 16 bits. 711| 680k| for (int i = 0; i < 4096; i++) { ------------------ | Branch (711:19): [True: 679k, False: 166] ------------------ 712| 679k| curve[i] = clampBits(static_cast(curve[i]) << 2, 16); 713| 679k| } 714| | 715| 166| curve.resize(4095); 716| | 717| 166| RawImageCurveGuard curveHandler(&mRaw, curve, false); 718| | 719| 166| uint16_t tmp; 720| 166| auto* tmpch = reinterpret_cast(&tmp); 721| | 722| 166| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 723| 166| const auto in = Array2DRef(input.peekData(out.width() * out.height()), 724| 166| out.width(), out.height()); 725| | 726| 7.56k| for (int row = 0; row < out.height(); row++) { ------------------ | Branch (726:21): [True: 7.39k, False: 166] ------------------ 727| 7.39k| uint32_t random = in(row, 0) + (in(row, 1) << 8) + (in(row, 2) << 16); 728| 107k| for (int col = 0; col < out.width(); col += 6) { ------------------ | Branch (728:23): [True: 99.7k, False: 7.39k] ------------------ 729| 99.7k| uint32_t g1 = in(row, col + 0); 730| 99.7k| uint32_t g2 = in(row, col + 1); 731| 99.7k| uint32_t g3 = in(row, col + 2); 732| 99.7k| uint32_t g4 = in(row, col + 3); 733| 99.7k| uint32_t g5 = in(row, col + 4); 734| 99.7k| uint32_t g6 = in(row, col + 5); 735| | 736| 99.7k| auto y1 = static_cast(g1 | ((g2 & 0x0f) << 8)); 737| 99.7k| auto y2 = static_cast((g2 >> 4) | (g3 << 4)); 738| 99.7k| auto cb = static_cast(g4 | ((g5 & 0x0f) << 8)); 739| 99.7k| auto cr = static_cast((g5 >> 4) | (g6 << 4)); 740| | 741| 99.7k| float cb2 = cb; 742| 99.7k| float cr2 = cr; 743| | // Interpolate right pixel. We assume the sample is aligned with left 744| | // pixel. 745| 99.7k| if ((col + 6) < out.width()) { ------------------ | Branch (745:11): [True: 92.3k, False: 7.39k] ------------------ 746| 92.3k| g4 = in(row, col + 6 + 3); 747| 92.3k| g5 = in(row, col + 6 + 4); 748| 92.3k| g6 = in(row, col + 6 + 5); 749| 92.3k| cb2 = (static_cast((g4 | ((g5 & 0x0f) << 8))) + cb) * 0.5F; 750| 92.3k| cr2 = (static_cast(((g5 >> 4) | (g6 << 4))) + cr) * 0.5F; 751| 92.3k| } 752| | 753| 99.7k| cb -= 2048; 754| 99.7k| cr -= 2048; 755| 99.7k| cb2 -= 2048; 756| 99.7k| cr2 -= 2048; 757| | 758| 99.7k| mRaw->setWithLookUp( 759| 99.7k| clampBits(static_cast(implicit_cast(y1) + 760| 99.7k| (1.370705 * implicit_cast(cr))), 761| 99.7k| 12), 762| 99.7k| tmpch, &random); 763| 99.7k| out(row, col) = clampBits((inv_wb_r * tmp + (1 << 9)) >> 10, 15); 764| | 765| 99.7k| mRaw->setWithLookUp( 766| 99.7k| clampBits(static_cast(implicit_cast(y1) - 767| 99.7k| (0.337633 * implicit_cast(cb)) - 768| 99.7k| (0.698001 * implicit_cast(cr))), 769| 99.7k| 12), 770| 99.7k| reinterpret_cast(&out(row, col + 1)), &random); 771| | 772| 99.7k| mRaw->setWithLookUp( 773| 99.7k| clampBits( 774| 99.7k| static_cast(implicit_cast(y1) + 775| 99.7k| (1.732446 // NOLINT(modernize-use-std-numbers) 776| 99.7k| * implicit_cast(cb))), 777| 99.7k| 12), 778| 99.7k| tmpch, &random); 779| 99.7k| out(row, col + 2) = clampBits((inv_wb_b * tmp + (1 << 9)) >> 10, 15); 780| | 781| 99.7k| mRaw->setWithLookUp( 782| 99.7k| clampBits(static_cast(implicit_cast(y2) + 783| 99.7k| (1.370705 * implicit_cast(cr2))), 784| 99.7k| 12), 785| 99.7k| tmpch, &random); 786| 99.7k| out(row, col + 3) = clampBits((inv_wb_r * tmp + (1 << 9)) >> 10, 15); 787| | 788| 99.7k| mRaw->setWithLookUp( 789| 99.7k| clampBits(static_cast(implicit_cast(y2) - 790| 99.7k| (0.337633 * implicit_cast(cb2)) - 791| 99.7k| (0.698001 * implicit_cast(cr2))), 792| 99.7k| 12), 793| 99.7k| reinterpret_cast(&out(row, col + 4)), &random); 794| | 795| 99.7k| mRaw->setWithLookUp( 796| 99.7k| clampBits( 797| 99.7k| static_cast(implicit_cast(y2) + 798| 99.7k| (1.732446 // NOLINT(modernize-use-std-numbers) 799| 99.7k| * implicit_cast(cb2))), 800| 99.7k| 12), 801| 99.7k| tmpch, &random); 802| 99.7k| out(row, col + 5) = clampBits((inv_wb_b * tmp + (1 << 9)) >> 10, 15); 803| 99.7k| } 804| 7.39k| } 805| 166|} _ZN8rawspeed10NefDecoder10gammaCurveEddi: 809| 166|std::vector NefDecoder::gammaCurve(double pwr, double ts, int imax) { 810| 166| std::vector curve(65536); 811| | 812| 166| int i; 813| 166| std::array g; 814| 166| std::array bnd = {{}}; 815| 166| g[0] = pwr; 816| 166| g[1] = ts; 817| 166| g[2] = g[3] = g[4] = 0; 818| 166| bnd[g[1] >= 1] = 1; 819| 166| if (std::abs(g[1]) > 0 && (g[1] - 1) * (g[0] - 1) <= 0) { ------------------ | Branch (819:7): [True: 166, False: 0] | Branch (819:29): [True: 166, False: 0] ------------------ 820| 8.13k| for (i = 0; i < 48; i++) { ------------------ | Branch (820:17): [True: 7.96k, False: 166] ------------------ 821| 7.96k| g[2] = (bnd[0] + bnd[1]) / 2; 822| 7.96k| if (std::abs(g[0]) > 0) ------------------ | Branch (822:11): [True: 7.96k, False: 0] ------------------ 823| 7.96k| bnd[(pow(g[2] / g[1], -g[0]) - 1) / g[0] - 1 / g[2] > -1] = g[2]; 824| 0| else 825| 0| bnd[g[2] / exp(1 - (1 / g[2])) < g[1]] = g[2]; 826| 7.96k| } 827| 166| g[3] = g[2] / g[1]; 828| 166| if (std::abs(g[0]) > 0) ------------------ | Branch (828:9): [True: 166, False: 0] ------------------ 829| 166| g[4] = g[2] * (1 / g[0] - 1); 830| 166| } 831| 166| if (std::abs(g[0]) > 0) { ------------------ | Branch (831:7): [True: 166, False: 0] ------------------ 832| 166| g[5] = 1 / (g[1] * SQR(g[3]) / 2 - g[4] * (1 - g[3]) + ------------------ | | 808| 166|#define SQR(x) ((x) * (x)) ------------------ 833| 166| (1 - pow(g[3], 1 + g[0])) * (1 + g[4]) / (1 + g[0])) - 834| 166| 1; 835| 166| } else { 836| 0| g[5] = 1 / (g[1] * SQR(g[3]) / 2 + 1 - g[2] - g[3] - ------------------ | | 808| 0|#define SQR(x) ((x) * (x)) ------------------ 837| 0| g[2] * g[3] * (log(g[3]) - 1)) - 838| 0| 1; 839| 0| } 840| | 841| 10.8M| for (i = 0; i < 0x10000; i++) { ------------------ | Branch (841:15): [True: 10.8M, False: 166] ------------------ 842| 10.8M| curve[i] = 0xffff; 843| 10.8M| const double r = static_cast(i) / imax; 844| 10.8M| if (r >= 1) ------------------ | Branch (844:9): [True: 10.1M, False: 679k] ------------------ 845| 10.1M| continue; 846| 679k| double v; 847| 679k| if (r < g[2]) { ------------------ | Branch (847:9): [True: 26.7k, False: 653k] ------------------ 848| 26.7k| v = r / g[1]; 849| 653k| } else { 850| 653k| if (std::abs(g[0]) > 0) { ------------------ | Branch (850:11): [True: 653k, False: 0] ------------------ 851| 653k| v = pow((r + g[4]) / (1 + g[4]), 1 / g[0]); 852| 653k| } else { 853| 0| v = exp((r - 1) / g[2]); 854| 0| } 855| 653k| } 856| 679k| curve[i] = static_cast(0x10000 * v); 857| 679k| } 858| | 859| 166| assert(curve.size() == 65536); ------------------ | Branch (859:3): [True: 166, False: 0] ------------------ 860| | 861| 166| return curve; 862| 166|} _ZN8rawspeed10NefDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 45| 863| : AbstractTiffDecoder(std::move(root), file) {} _ZN8rawspeed10OrfDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 54| 4.48k| [[maybe_unused]] Buffer file) { 55| 4.48k| const auto id = rootIFD->getID(); 56| 4.48k| const std::string& make = id.make; 57| | 58| | // FIXME: magic 59| | 60| 4.48k| return make == "OLYMPUS IMAGING CORP." || make == "OLYMPUS CORPORATION" || ------------------ | Branch (60:10): [True: 18, False: 4.46k] | Branch (60:45): [True: 119, False: 4.34k] ------------------ 61| 4.34k| make == "OLYMPUS OPTICAL CO.,LTD" || make == "OM Digital Solutions"; ------------------ | Branch (61:10): [True: 192, False: 4.15k] | Branch (61:47): [True: 0, False: 4.15k] ------------------ 62| 4.48k|} _ZNK8rawspeed10OrfDecoder12handleSlicesEv: 64| 277|ByteStream OrfDecoder::handleSlices() const { 65| 277| const auto* raw = mRootIFD->getIFDWithTag(TiffTag::STRIPOFFSETS); 66| | 67| 277| const TiffEntry* offsets = raw->getEntry(TiffTag::STRIPOFFSETS); 68| 277| const TiffEntry* counts = raw->getEntry(TiffTag::STRIPBYTECOUNTS); 69| | 70| 277| if (counts->count != offsets->count) { ------------------ | Branch (70:7): [True: 10, False: 267] ------------------ 71| 10| ThrowRDE( ------------------ | | 38| 10| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 10| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 10| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 10| __VA_ARGS__) | | ------------------ ------------------ 72| 10| "Byte count number does not match strip size: count:%u, strips:%u ", 73| 10| counts->count, offsets->count); 74| 10| } 75| | 76| 267| const uint32_t off = offsets->getU32(0); 77| 267| uint32_t size = counts->getU32(0); 78| 267| auto end = [&off, &size]() { return off + size; }; 79| | 80| 560| for (uint32_t i = 0; i < counts->count; i++) { ------------------ | Branch (80:24): [True: 333, False: 227] ------------------ 81| 333| const auto offset = offsets->getU32(i); 82| 333| const auto count = counts->getU32(i); 83| 333| if (!mFile.isValid(offset, count)) ------------------ | Branch (83:9): [True: 37, False: 296] ------------------ 84| 333| ThrowRDE("Truncated file"); ------------------ | | 38| 37| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 37| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 37| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 37| __VA_ARGS__) | | ------------------ ------------------ 85| | 86| 296| if (count < 1) ------------------ | Branch (86:9): [True: 2, False: 294] ------------------ 87| 296| ThrowRDE("Empty slice"); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 88| | 89| 294| if (i == 0) ------------------ | Branch (89:9): [True: 251, False: 43] ------------------ 90| 251| continue; 91| | 92| 43| if (offset < end()) ------------------ | Branch (92:9): [True: 1, False: 42] ------------------ 93| 43| ThrowRDE("Slices overlap"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 94| | 95| | // Now, everything would be great, but some uncompressed raws 96| | // (packed_with_control i believe) have "padding" between at least 97| | // the first two slices, and we need to account for it. 98| 42| const uint32_t padding = offset - end(); 99| | 100| 42| size += padding; 101| 42| size += count; 102| 42| } 103| | 104| 227| ByteStream input(offsets->getRootIfdData()); 105| 227| input.setPosition(off); 106| | 107| 227| return input.getStream(size); 108| 267|} OrfDecoder.cpp:_ZZNK8rawspeed10OrfDecoder12handleSlicesEvENK3$_0clEv: 78| 85| auto end = [&off, &size]() { return off + size; }; _ZN8rawspeed10OrfDecoder17decodeRawInternalEv: 110| 329|RawImage OrfDecoder::decodeRawInternal() { 111| 329| const auto* raw = mRootIFD->getIFDWithTag(TiffTag::STRIPOFFSETS); 112| | 113| 329| if (int compression = raw->getEntry(TiffTag::COMPRESSION)->getU32(); 114| 329| 1 != compression) ------------------ | Branch (114:7): [True: 3, False: 326] ------------------ 115| 329| ThrowRDE("Unsupported compression"); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 116| | 117| 326| uint32_t width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 118| 326| uint32_t height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 119| | 120| 326| if (!width || !height || width % 2 != 0 || width > 10400 || height > 7796) ------------------ | Branch (120:7): [True: 6, False: 320] | Branch (120:17): [True: 1, False: 319] | Branch (120:28): [True: 3, False: 316] | Branch (120:46): [True: 14, False: 302] | Branch (120:63): [True: 25, False: 277] ------------------ 121| 326| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 44| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 44| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 44| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 44| __VA_ARGS__) | | ------------------ ------------------ 122| | 123| 282| mRaw->dim = iPoint2D(width, height); 124| | 125| 282| ByteStream input(handleSlices()); 126| | 127| 282| if (decodeUncompressed(input, width, height, input.getSize())) ------------------ | Branch (127:7): [True: 61, False: 221] ------------------ 128| 61| return mRaw; 129| | 130| 221| if (raw->getEntry(TiffTag::STRIPOFFSETS)->count != 1) ------------------ | Branch (130:7): [True: 1, False: 220] ------------------ 131| 221| ThrowRDE("%u stripes, and not uncompressed. Unsupported.", ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 132| 220| raw->getEntry(TiffTag::STRIPOFFSETS)->count); 133| | 134| 220| if (const int bitsPerPixel = getBitsPerPixel(); bitsPerPixel == 12) { ------------------ | Branch (134:51): [True: 142, False: 78] ------------------ 135| 142| input.skipBytes(7); 136| 142| } else if (bitsPerPixel == 14) { ------------------ | Branch (136:14): [True: 0, False: 78] ------------------ 137| 0| input.skipBytes(8); 138| 78| } else { 139| 78| ThrowRDE("%i-bit images are not supported currently.", bitsPerPixel); ------------------ | | 38| 78| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 78| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 78| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 78| __VA_ARGS__) | | ------------------ ------------------ 140| 78| } 141| | 142| 142| const OlympusDecompressor o(mRaw); 143| 142| mRaw->createData(); 144| 142| o.decompress(input); 145| | 146| 142| return mRaw; 147| 220|} _ZNK8rawspeed10OrfDecoder15getBitsPerPixelEv: 149| 145|int OrfDecoder::getBitsPerPixel() const { 150| 145| int result = 12; 151| 145| if (mRootIFD->hasEntryRecursive(TiffTag::OLYMPUSIMAGEPROCESSING)) { ------------------ | Branch (151:7): [True: 11, False: 134] ------------------ 152| | // Newer cameras process the Image Processing SubIFD in the makernote 153| 11| const TiffEntry* img_entry = 154| 11| mRootIFD->getEntryRecursive(TiffTag::OLYMPUSIMAGEPROCESSING); 155| | // get makernote ifd with containing Buffer 156| 11| NORangesSet ifds; 157| | 158| 11| TiffRootIFD image_processing(nullptr, &ifds, img_entry->getRootIfdData(), 159| 11| img_entry->getU32()); 160| | 161| 11| if (image_processing.hasEntry(static_cast(0x0611))) { ------------------ | Branch (161:9): [True: 1, False: 10] ------------------ 162| 1| const TiffEntry* validBits = 163| 1| image_processing.getEntry(static_cast(0x0611)); 164| 1| result = validBits->getU16(); 165| 1| } 166| 11| } 167| 145| return result; 168| 145|} _ZNK8rawspeed10OrfDecoder29decodeUncompressedInterleavedENS_10ByteStreamEjjj: 172| 46| uint32_t size) const { 173| 46| int inputPitchBits = 12 * w; 174| 46| assert(inputPitchBits % 8 == 0); ------------------ | Branch (174:3): [True: 46, False: 0] ------------------ 175| | 176| 46| int inputPitchBytes = inputPitchBits / 8; 177| | 178| 46| const auto numEvenLines = implicit_cast(roundUpDivisionSafe(h, 2)); 179| 46| const auto evenLinesInput = s.getStream(numEvenLines, inputPitchBytes) 180| 46| .peekRemainingBuffer() 181| 46| .getAsArray1DRef(); 182| | 183| 46| const auto oddLinesInputBegin = 184| 46| implicit_cast(roundUp(evenLinesInput.size(), 1U << 11U)); 185| 46| assert(oddLinesInputBegin >= evenLinesInput.size()); ------------------ | Branch (185:3): [True: 46, False: 0] ------------------ 186| 46| int padding = oddLinesInputBegin - evenLinesInput.size(); 187| 46| assert(padding >= 0); ------------------ | Branch (187:3): [True: 46, False: 0] ------------------ 188| 46| s.skipBytes(padding); 189| | 190| 46| const int numOddLines = h - numEvenLines; 191| 46| const auto oddLinesInput = s.getStream(numOddLines, inputPitchBytes) 192| 46| .peekRemainingBuffer() 193| 46| .getAsArray1DRef(); 194| | 195| | // By now we know we have enough input to produce the image. 196| 46| mRaw->createData(); 197| | 198| 46| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 199| 46| { 200| 46| BitStreamerMSB bs(evenLinesInput); 201| 8.43k| for (int i = 0; i != numEvenLines; ++i) { ------------------ | Branch (201:21): [True: 8.39k, False: 46] ------------------ 202| 203k| for (unsigned col = 0; col != w; ++col) { ------------------ | Branch (202:30): [True: 195k, False: 8.39k] ------------------ 203| 195k| int row = 2 * i; 204| 195k| out(row, col) = implicit_cast(bs.getBits(12)); 205| 195k| } 206| 8.39k| } 207| 46| } 208| 46| { 209| 46| BitStreamerMSB bs(oddLinesInput); 210| 8.41k| for (int i = 0; i != numOddLines; ++i) { ------------------ | Branch (210:21): [True: 8.36k, False: 46] ------------------ 211| 197k| for (unsigned col = 0; col != w; ++col) { ------------------ | Branch (211:30): [True: 189k, False: 8.36k] ------------------ 212| 189k| int row = 1 + (2 * i); 213| 189k| out(row, col) = implicit_cast(bs.getBits(12)); 214| 189k| } 215| 8.36k| } 216| 46| } 217| 46|} _ZNK8rawspeed10OrfDecoder18decodeUncompressedENS_10ByteStreamEjjj: 220| 225| uint32_t size) const { 221| | // FIXME: most of this logic should be in UncompressedDecompressor, 222| | // one way or another. 223| | 224| 225| if (size == h * ((w * 12 / 8) + ((w + 2) / 10))) { ------------------ | Branch (224:7): [True: 15, False: 210] ------------------ 225| | // 12-bit packed 'with control' raw 226| 15| UncompressedDecompressor u(s, mRaw, iRectangle2D({0, 0}, iPoint2D(w, h)), 227| 15| (12 * w / 8) + ((w + 2) / 10), 12, 228| 15| BitOrder::LSB); 229| 15| mRaw->createData(); 230| 15| u.decode12BitRawWithControl(); 231| 15| return true; 232| 15| } 233| | 234| 210| iPoint2D dimensions(w, h); 235| 210| iPoint2D pos(0, 0); 236| 210| if (size == w * h * 12 / 8) { // We're in a 12-bit packed raw ------------------ | Branch (236:7): [True: 18, False: 192] ------------------ 237| 18| UncompressedDecompressor u(s, mRaw, iRectangle2D(pos, dimensions), 238| 18| w * 12 / 8, 12, BitOrder::MSB32); 239| 18| mRaw->createData(); 240| 18| u.readUncompressedRaw(); 241| 18| return true; 242| 18| } 243| | 244| 192| if (size == w * h * 2) { // We're in an unpacked raw ------------------ | Branch (244:7): [True: 1, False: 191] ------------------ 245| | // FIXME: seems fishy 246| 1| if (s.getByteOrder() == getHostEndianness()) { ------------------ | Branch (246:9): [True: 1, False: 0] ------------------ 247| 1| UncompressedDecompressor u(s, mRaw, iRectangle2D({0, 0}, iPoint2D(w, h)), 248| 1| 16 * w / 8, 16, BitOrder::LSB); 249| 1| mRaw->createData(); 250| 1| u.decode12BitRawUnpackedLeftAligned(); 251| 1| } else { 252| 0| UncompressedDecompressor u(s, mRaw, iRectangle2D({0, 0}, iPoint2D(w, h)), 253| 0| 16 * w / 8, 16, BitOrder::MSB); 254| 0| mRaw->createData(); 255| 0| u.decode12BitRawUnpackedLeftAligned(); 256| 0| } 257| 1| return true; 258| 1| } 259| | 260| 191| if (size > w * h * 3 / 2) { ------------------ | Branch (260:7): [True: 46, False: 145] ------------------ 261| | // We're in one of those weird interlaced packed raws 262| 46| decodeUncompressedInterleaved(s, w, h, size); 263| 46| return true; 264| 46| } 265| | 266| | // Does not appear to be uncomporessed. Maybe it's compressed? 267| 145| return false; 268| 191|} _ZNK8rawspeed10OrfDecoder8parseCFAEv: 270| 63|void OrfDecoder::parseCFA() const { 271| 63| if (!mRootIFD->hasEntryRecursive(TiffTag::EXIFCFAPATTERN)) ------------------ | Branch (271:7): [True: 37, False: 26] ------------------ 272| 63| ThrowRDE("No EXIFCFAPATTERN entry found!"); ------------------ | | 38| 37| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 37| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 37| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 37| __VA_ARGS__) | | ------------------ ------------------ 273| | 274| 26| const TiffEntry* CFA = mRootIFD->getEntryRecursive(TiffTag::EXIFCFAPATTERN); 275| 26| if (CFA->type != TiffDataType::UNDEFINED || CFA->count != 8) { ------------------ | Branch (275:7): [True: 1, False: 25] | Branch (275:47): [True: 1, False: 24] ------------------ 276| 2| ThrowRDE("Bad EXIFCFAPATTERN entry (type %u, count %u).", ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 277| 2| static_cast(CFA->type), CFA->count); 278| 2| } 279| | 280| 24| iPoint2D cfaSize(CFA->getU16(0), CFA->getU16(1)); 281| 24| if (cfaSize != iPoint2D{2, 2}) ------------------ | Branch (281:7): [True: 2, False: 22] ------------------ 282| 24| ThrowRDE("Bad CFA size: (%i, %i)", cfaSize.x, cfaSize.y); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 283| | 284| 22| mRaw->cfa.setSize(cfaSize); 285| | 286| 22| auto int2enum = [](uint8_t i) { 287| 22| switch (i) { 288| 22| using enum CFAColor; 289| 22| case 0: 290| 22| return RED; 291| 22| case 1: 292| 22| return GREEN; 293| 22| case 2: 294| 22| return BLUE; 295| 22| default: 296| 22| ThrowRDE("Unexpected CFA color: %u", i); 297| 22| } 298| 22| }; 299| | 300| 66| for (int y = 0; y < cfaSize.y; y++) { ------------------ | Branch (300:19): [True: 44, False: 22] ------------------ 301| 131| for (int x = 0; x < cfaSize.x; x++) { ------------------ | Branch (301:21): [True: 87, False: 44] ------------------ 302| 87| uint8_t c1 = CFA->getByte(4 + x + (y * cfaSize.x)); 303| 87| CFAColor c2 = int2enum(c1); 304| 87| mRaw->cfa.setColorAt(iPoint2D(x, y), c2); 305| 87| } 306| 44| } 307| 22|} OrfDecoder.cpp:_ZZNK8rawspeed10OrfDecoder8parseCFAEvENK3$_0clEh: 286| 87| auto int2enum = [](uint8_t i) { 287| 87| switch (i) { 288| 0| using enum CFAColor; 289| 57| case 0: ------------------ | Branch (289:5): [True: 57, False: 30] ------------------ 290| 57| return RED; 291| 19| case 1: ------------------ | Branch (291:5): [True: 19, False: 68] ------------------ 292| 19| return GREEN; 293| 10| case 2: ------------------ | Branch (293:5): [True: 10, False: 77] ------------------ 294| 10| return BLUE; 295| 1| default: ------------------ | Branch (295:5): [True: 1, False: 86] ------------------ 296| 1| ThrowRDE("Unexpected CFA color: %u", i); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 297| 87| } 298| 87| }; _ZN8rawspeed10OrfDecoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 309| 63|void OrfDecoder::decodeMetaDataInternal(const CameraMetaData* meta) { 310| 63| mRaw->whitePoint = (1U << 12) - 1; 311| | 312| 63| int iso = 0; 313| | 314| 63| if (mRootIFD->hasEntryRecursive(TiffTag::ISOSPEEDRATINGS)) ------------------ | Branch (314:7): [True: 1, False: 62] ------------------ 315| 1| iso = mRootIFD->getEntryRecursive(TiffTag::ISOSPEEDRATINGS)->getU32(); 316| | 317| 63| parseCFA(); 318| | 319| 63| setMetaData(meta, "", iso); 320| | 321| 63| if (mRootIFD->hasEntryRecursive(TiffTag::OLYMPUSREDMULTIPLIER) && ------------------ | Branch (321:7): [True: 5, False: 58] ------------------ 322| 5| mRootIFD->hasEntryRecursive(TiffTag::OLYMPUSBLUEMULTIPLIER)) { ------------------ | Branch (322:7): [True: 0, False: 5] ------------------ 323| 0| std::array wbCoeffs = {}; 324| 0| wbCoeffs[0] = static_cast( 325| 0| mRootIFD->getEntryRecursive(TiffTag::OLYMPUSREDMULTIPLIER)->getU16()); 326| 0| wbCoeffs[1] = 256.0F; 327| 0| wbCoeffs[2] = static_cast( 328| 0| mRootIFD->getEntryRecursive(TiffTag::OLYMPUSBLUEMULTIPLIER)->getU16()); 329| 0| mRaw->metadata.wbCoeffs = wbCoeffs; 330| 63| } else if (mRootIFD->hasEntryRecursive(TiffTag::OLYMPUSIMAGEPROCESSING)) { ------------------ | Branch (330:14): [True: 20, False: 43] ------------------ 331| | // Newer cameras process the Image Processing SubIFD in the makernote 332| 20| const TiffEntry* img_entry = 333| 20| mRootIFD->getEntryRecursive(TiffTag::OLYMPUSIMAGEPROCESSING); 334| | // get makernote ifd with containing Buffer 335| 20| NORangesSet ifds; 336| | 337| 20| TiffRootIFD image_processing(nullptr, &ifds, img_entry->getRootIfdData(), 338| 20| img_entry->getU32()); 339| | 340| | // Get the WB 341| 20| if (image_processing.hasEntry(static_cast(0x0100))) { ------------------ | Branch (341:9): [True: 9, False: 11] ------------------ 342| 9| const TiffEntry* wb = 343| 9| image_processing.getEntry(static_cast(0x0100)); 344| 9| if (wb->count == 2 || wb->count == 4) { ------------------ | Branch (344:11): [True: 2, False: 7] | Branch (344:29): [True: 2, False: 5] ------------------ 345| 4| std::array wbCoeffs = {}; 346| 4| wbCoeffs[0] = wb->getFloat(0); 347| 4| wbCoeffs[1] = 256.0F; 348| 4| wbCoeffs[2] = wb->getFloat(1); 349| 4| mRaw->metadata.wbCoeffs = wbCoeffs; 350| 4| } 351| 9| } 352| | 353| | // Get the black levels 354| 20| if (image_processing.hasEntry(static_cast(0x0600))) { ------------------ | Branch (354:9): [True: 6, False: 14] ------------------ 355| 6| const TiffEntry* blackEntry = 356| 6| image_processing.getEntry(static_cast(0x0600)); 357| | // Order is assumed to be RGGB 358| 6| if (blackEntry->count == 4) { ------------------ | Branch (358:11): [True: 5, False: 1] ------------------ 359| 5| mRaw->blackLevelSeparate = 360| 5| Array2DRef(mRaw->blackLevelSeparateStorage.data(), 2, 2); 361| 5| auto blackLevelSeparate1D = 362| 5| *mRaw->blackLevelSeparate->getAsArray1DRef(); 363| 19| for (int i = 0; i < 4; i++) { ------------------ | Branch (363:25): [True: 14, False: 5] ------------------ 364| 14| auto c = mRaw->cfa.getColorAt(i & 1, i >> 1); 365| 14| int j; 366| 14| switch (c) { 367| 0| using enum CFAColor; 368| 8| case RED: ------------------ | Branch (368:11): [True: 8, False: 6] ------------------ 369| 8| j = 0; 370| 8| break; 371| 3| case GREEN: ------------------ | Branch (371:11): [True: 3, False: 11] ------------------ 372| 3| j = i < 2 ? 1 : 2; ------------------ | Branch (372:17): [True: 2, False: 1] ------------------ 373| 3| break; 374| 3| case BLUE: ------------------ | Branch (374:11): [True: 3, False: 11] ------------------ 375| 3| j = 3; 376| 3| break; 377| 0| default: ------------------ | Branch (377:11): [True: 0, False: 14] ------------------ 378| 0| ThrowRDE("Unexpected CFA color: %u", static_cast(c)); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 379| 14| } 380| | 381| 14| blackLevelSeparate1D(i) = blackEntry->getU16(j); 382| 14| } 383| | // Adjust whitelevel based on the read black (we assume the dynamic 384| | // range is the same) 385| 5| mRaw->whitePoint = 386| 5| *mRaw->whitePoint - (mRaw->blackLevel - blackLevelSeparate1D(0)); 387| 5| if (getBitsPerPixel() == 14) { ------------------ | Branch (387:13): [True: 0, False: 5] ------------------ 388| 0| mRaw->whitePoint = *mRaw->whitePoint * 4; 389| 0| mRaw->blackLevel = mRaw->blackLevel * 4; 390| 0| for (int i = 0; i < 4; i++) { ------------------ | Branch (390:27): [True: 0, False: 0] ------------------ 391| 0| blackLevelSeparate1D(i) = blackLevelSeparate1D(i) * 4; 392| 0| } 393| 0| } 394| 5| } 395| 6| } 396| 20| } 397| 63|} _ZN8rawspeed10OrfDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 43| 329| : AbstractTiffDecoder(std::move(root), file) {} _ZN8rawspeed10PefDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 48| 2.82k| [[maybe_unused]] Buffer file) { 49| 2.82k| const auto id = rootIFD->getID(); 50| 2.82k| const std::string& make = id.make; 51| | 52| | // FIXME: magic 53| | 54| 2.82k| return make == "PENTAX Corporation" || ------------------ | Branch (54:10): [True: 23, False: 2.80k] ------------------ 55| 2.80k| make == "RICOH IMAGING COMPANY, LTD." || make == "PENTAX" || ------------------ | Branch (55:10): [True: 1, False: 2.80k] | Branch (55:51): [True: 281, False: 2.52k] ------------------ 56| 2.52k| make == "SAMSUNG TECHWIN"; ------------------ | Branch (56:10): [True: 1, False: 2.52k] ------------------ 57| 2.82k|} _ZN8rawspeed10PefDecoder17decodeRawInternalEv: 59| 306|RawImage PefDecoder::decodeRawInternal() { 60| 306| const auto* raw = mRootIFD->getIFDWithTag(TiffTag::STRIPOFFSETS); 61| | 62| 306| int compression = raw->getEntry(TiffTag::COMPRESSION)->getU32(); 63| | 64| 306| if (1 == compression || compression == 32773) { ------------------ | Branch (64:7): [True: 90, False: 216] | Branch (64:27): [True: 1, False: 215] ------------------ 65| 87| decodeUncompressed(raw, BitOrder::MSB); 66| 87| return mRaw; 67| 87| } 68| | 69| 219| if (65535 != compression) ------------------ | Branch (69:7): [True: 30, False: 189] ------------------ 70| 219| ThrowRDE("Unsupported compression"); ------------------ | | 38| 30| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 30| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 30| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 30| __VA_ARGS__) | | ------------------ ------------------ 71| | 72| 189| if (raw->hasEntry(TiffTag::PHOTOMETRICINTERPRETATION)) { ------------------ | Branch (72:7): [True: 2, False: 187] ------------------ 73| 2| mRaw->isCFA = 74| 2| (raw->getEntry(TiffTag::PHOTOMETRICINTERPRETATION)->getU16() != 34892); 75| 2| } 76| | 77| 189| if (mRaw->isCFA) ------------------ | Branch (77:7): [True: 185, False: 4] ------------------ 78| 185| writeLog(DEBUG_PRIO::EXTRA, "This is a CFA image"); 79| 4| else { 80| 4| writeLog(DEBUG_PRIO::EXTRA, "This is NOT a CFA image"); 81| 4| } 82| | 83| 189| const TiffEntry* offsets = raw->getEntry(TiffTag::STRIPOFFSETS); 84| 189| const TiffEntry* counts = raw->getEntry(TiffTag::STRIPBYTECOUNTS); 85| | 86| 189| if (offsets->count != 1) { ------------------ | Branch (86:7): [True: 2, False: 187] ------------------ 87| 2| ThrowRDE("Multiple Strips found: %u", offsets->count); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 88| 2| } 89| 187| if (counts->count != offsets->count) { ------------------ | Branch (89:7): [True: 1, False: 186] ------------------ 90| 1| ThrowRDE( ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 91| 1| "Byte count number does not match strip size: count:%u, strips:%u ", 92| 1| counts->count, offsets->count); 93| 1| } 94| 186| ByteStream bs( 95| 186| DataBuffer(mFile.getSubView(offsets->getU32(), counts->getU32()), 96| 186| Endianness::little)); 97| | 98| 186| uint32_t width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 99| 186| uint32_t height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 100| | 101| 186| mRaw->dim = iPoint2D(width, height); 102| | 103| 186| Optional metaData; 104| 186| if (getRootIFD()->hasEntryRecursive(static_cast(0x220))) { ------------------ | Branch (104:7): [True: 51, False: 135] ------------------ 105| | /* Attempt to read huffman table, if found in makernote */ 106| 51| const TiffEntry* t = 107| 51| getRootIFD()->getEntryRecursive(static_cast(0x220)); 108| 51| if (t->type != TiffDataType::UNDEFINED) ------------------ | Branch (108:9): [True: 1, False: 50] ------------------ 109| 51| ThrowRDE("Unknown Huffman table type."); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 110| | 111| 50| metaData = t->getData(); 112| 50| } 113| | 114| 185| PentaxDecompressor p(mRaw, metaData); 115| 185| mRaw->createData(); 116| 185| p.decompress(bs); 117| | 118| 185| return mRaw; 119| 186|} _ZN8rawspeed10PefDecoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 121| 10|void PefDecoder::decodeMetaDataInternal(const CameraMetaData* meta) { 122| 10| int iso = 0; 123| 10| mRaw->cfa.setCFA(iPoint2D(2, 2), CFAColor::RED, CFAColor::GREEN, 124| 10| CFAColor::GREEN, CFAColor::BLUE); 125| | 126| 10| if (mRootIFD->hasEntryRecursive(TiffTag::ISOSPEEDRATINGS)) ------------------ | Branch (126:7): [True: 1, False: 9] ------------------ 127| 1| iso = mRootIFD->getEntryRecursive(TiffTag::ISOSPEEDRATINGS)->getU32(); 128| | 129| 10| setMetaData(meta, "", iso); 130| | 131| | // Read black level 132| 10| if (mRootIFD->hasEntryRecursive(static_cast(0x200))) { ------------------ | Branch (132:7): [True: 3, False: 7] ------------------ 133| 3| const TiffEntry* black = 134| 3| mRootIFD->getEntryRecursive(static_cast(0x200)); 135| 3| if (black->count == 4) { ------------------ | Branch (135:9): [True: 2, False: 1] ------------------ 136| 2| mRaw->blackLevelSeparate = 137| 2| Array2DRef(mRaw->blackLevelSeparateStorage.data(), 2, 2); 138| 2| auto blackLevelSeparate1D = *mRaw->blackLevelSeparate->getAsArray1DRef(); 139| 4| for (int i = 0; i < 4; i++) ------------------ | Branch (139:23): [True: 2, False: 2] ------------------ 140| 2| blackLevelSeparate1D(i) = black->getU32(i); 141| 2| } 142| 3| } 143| | 144| | // Set the whitebalance 145| 10| if (mRootIFD->hasEntryRecursive(static_cast(0x0201))) { ------------------ | Branch (145:7): [True: 2, False: 8] ------------------ 146| 2| const TiffEntry* wb = 147| 2| mRootIFD->getEntryRecursive(static_cast(0x0201)); 148| 2| if (wb->count == 4) { ------------------ | Branch (148:9): [True: 1, False: 1] ------------------ 149| 1| std::array wbCoeffs = {}; 150| 1| wbCoeffs[0] = implicit_cast(wb->getU32(0)); 151| 1| wbCoeffs[1] = implicit_cast(wb->getU32(1)); 152| 1| wbCoeffs[2] = implicit_cast(wb->getU32(3)); 153| 1| mRaw->metadata.wbCoeffs = wbCoeffs; 154| 1| } 155| 2| } 156| 10|} _ZN8rawspeed10PefDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 38| 306| : AbstractTiffDecoder(std::move(root), file) {} _ZN8rawspeed10RawDecoderC2ENS_6BufferE: 54| 11.6k|RawDecoder::RawDecoder(Buffer file) : mFile(file) {} _ZNK8rawspeed10RawDecoder18decodeUncompressedEPKNS_7TiffIFDENS_8BitOrderE: 57| 96| BitOrder order) const { 58| 96| const TiffEntry* offsets = rawIFD->getEntry(TiffTag::STRIPOFFSETS); 59| 96| const TiffEntry* counts = rawIFD->getEntry(TiffTag::STRIPBYTECOUNTS); 60| 96| uint32_t yPerSlice = rawIFD->getEntry(TiffTag::ROWSPERSTRIP)->getU32(); 61| 96| uint32_t width = rawIFD->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 62| 96| uint32_t height = rawIFD->getEntry(TiffTag::IMAGELENGTH)->getU32(); 63| 96| uint32_t bitPerPixel = rawIFD->getEntry(TiffTag::BITSPERSAMPLE)->getU32(); 64| | 65| 96| if (width == 0 || height == 0 || width > 5632 || height > 3720) ------------------ | Branch (65:7): [True: 12, False: 84] | Branch (65:21): [True: 1, False: 83] | Branch (65:36): [True: 3, False: 80] | Branch (65:52): [True: 5, False: 75] ------------------ 66| 96| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 10| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 10| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 10| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 10| __VA_ARGS__) | | ------------------ ------------------ 67| | 68| 86| mRaw->dim = iPoint2D(width, height); 69| | 70| 86| if (counts->count != offsets->count) { ------------------ | Branch (70:7): [True: 1, False: 85] ------------------ 71| 1| ThrowRDE("Byte count number does not match strip size: " ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 72| 1| "count:%u, stips:%u ", 73| 1| counts->count, offsets->count); 74| 1| } 75| | 76| 85| if (yPerSlice == 0 || yPerSlice > static_cast(mRaw->dim.y) || ------------------ | Branch (76:7): [True: 12, False: 73] | Branch (76:25): [True: 20, False: 53] ------------------ 77| 53| roundUpDivisionSafe(mRaw->dim.y, yPerSlice) != counts->count) { ------------------ | Branch (77:7): [True: 7, False: 46] ------------------ 78| 28| ThrowRDE("Invalid y per slice %u or strip count %u (height = %d)", ------------------ | | 38| 28| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 28| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 28| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 28| __VA_ARGS__) | | ------------------ ------------------ 79| 28| yPerSlice, counts->count, mRaw->dim.y); 80| 28| } 81| | 82| 57| switch (bitPerPixel) { 83| 20| case 12: ------------------ | Branch (83:3): [True: 20, False: 37] ------------------ 84| 45| case 14: ------------------ | Branch (84:3): [True: 25, False: 32] ------------------ 85| 45| break; 86| 1| default: ------------------ | Branch (86:3): [True: 1, False: 56] ------------------ 87| 1| ThrowRDE("Unexpected bits per pixel: %u.", bitPerPixel); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 88| 57| } 89| | 90| 45| vector slices; 91| 45| slices.reserve(counts->count); 92| 45| uint32_t offY = 0; 93| | 94| 323| for (uint32_t s = 0; s < counts->count; s++) { ------------------ | Branch (94:24): [True: 311, False: 12] ------------------ 95| 311| RawSlice slice; 96| 311| slice.offset = offsets->getU32(s); 97| 311| slice.count = counts->getU32(s); 98| | 99| 311| if (slice.count < 1) ------------------ | Branch (99:9): [True: 2, False: 309] ------------------ 100| 311| ThrowRDE("Slice %u is empty", s); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 101| | 102| 309| if (offY + yPerSlice > height) ------------------ | Branch (102:9): [True: 5, False: 304] ------------------ 103| 5| slice.h = height - offY; 104| 304| else 105| 304| slice.h = yPerSlice; 106| | 107| 309| offY += yPerSlice; 108| | 109| 309| if (!mFile.isValid(slice.offset, slice.count)) ------------------ | Branch (109:9): [True: 31, False: 278] ------------------ 110| 309| ThrowRDE("Slice offset/count invalid"); ------------------ | | 38| 31| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 31| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 31| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 31| __VA_ARGS__) | | ------------------ ------------------ 111| | 112| 278| slices.push_back(slice); 113| 278| } 114| | 115| 12| if (slices.empty()) ------------------ | Branch (115:7): [True: 0, False: 12] ------------------ 116| 12| ThrowRDE("No valid slices found. File probably truncated."); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 117| | 118| 12| assert(height <= offY); ------------------ | Branch (118:3): [True: 11, False: 1] ------------------ 119| 12| assert(slices.size() == counts->count); ------------------ | Branch (119:3): [True: 11, False: 0] ------------------ 120| | 121| 11| mRaw->createData(); 122| | 123| | // Default white level is (2 ** BitsPerSample) - 1 124| 11| mRaw->whitePoint = implicit_cast((1UL << bitPerPixel) - 1UL); 125| | 126| 11| offY = 0; 127| 107| for (const RawSlice& slice : slices) { ------------------ | Branch (127:30): [True: 107, False: 6] ------------------ 128| 107| iPoint2D size(width, slice.h); 129| 107| iPoint2D pos(0, offY); 130| 107| bitPerPixel = implicit_cast( 131| 107| (static_cast(slice.count) * 8U) / (slice.h * width)); 132| 107| const auto inputPitch = width * bitPerPixel / 8; 133| 107| if (!inputPitch) ------------------ | Branch (133:9): [True: 5, False: 102] ------------------ 134| 107| ThrowRDE("Bad input pitch. Can not decode anything."); ------------------ | | 38| 5| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 5| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 5| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 5| __VA_ARGS__) | | ------------------ ------------------ 135| | 136| 102| UncompressedDecompressor u( 137| 102| ByteStream(DataBuffer(mFile.getSubView(slice.offset, slice.count), 138| 102| Endianness::little)), 139| 102| mRaw, iRectangle2D(pos, size), inputPitch, bitPerPixel, order); 140| 102| u.readUncompressedRaw(); 141| | 142| 102| offY += slice.h; 143| 102| } 144| 11|} _ZN8rawspeed10RawDecoder19handleCameraSupportEPKNS_14CameraMetaDataERKNSt3__112basic_stringIcNS4_11char_traitsIcEENS4_9allocatorIcEEEESC_SC_: 149| 787| const std::string& mode) { 150| 787| Camera::SupportStatus supportStatus = Camera::SupportStatus::UnknownCamera; 151| 787| const Camera* cam = meta->getCamera(make, model, mode); 152| 787| if (cam) ------------------ | Branch (152:7): [True: 0, False: 787] ------------------ 153| 0| supportStatus = cam->supportStatus; 154| | 155| | // Sample beggary block. 156| 787| switch (supportStatus) { ------------------ | Branch (156:11): [True: 787, False: 0] ------------------ 157| 0| using enum Camera::SupportStatus; 158| 787| case UnknownCamera: ------------------ | Branch (158:3): [True: 787, False: 0] ------------------ 159| 787| if ("dng" != mode) { ------------------ | Branch (159:9): [True: 787, False: 0] ------------------ 160| 787| noSamples = true; 161| 787| writeLog(DEBUG_PRIO::WARNING, 162| 787| "Unable to find camera in database: '%s' '%s' '%s'\nPlease " 163| 787| "consider providing samples on , thanks!", 164| 787| make.c_str(), model.c_str(), mode.c_str()); 165| 787| } 166| 787| break; 167| 0| case UnknownNoSamples: ------------------ | Branch (167:3): [True: 0, False: 787] ------------------ 168| 0| case SupportedNoSamples: ------------------ | Branch (168:3): [True: 0, False: 787] ------------------ 169| 0| noSamples = true; 170| 0| writeLog(DEBUG_PRIO::WARNING, 171| 0| "Camera support status is unknown: '%s' '%s' '%s'\n" 172| 0| "Please consider providing samples on " 173| 0| "if you wish for the support to not be discontinued, thanks!", 174| 0| make.c_str(), model.c_str(), mode.c_str()); 175| 0| break; // WYSIWYG. 176| 0| case Supported: ------------------ | Branch (176:3): [True: 0, False: 787] ------------------ 177| 0| case Unknown: ------------------ | Branch (177:3): [True: 0, False: 787] ------------------ 178| 0| case Unsupported: ------------------ | Branch (178:3): [True: 0, False: 787] ------------------ 179| 0| break; // All these imply existence of a sample on RPU. 180| 787| } 181| | 182| | // Actual support handling. 183| 787| switch (supportStatus) { ------------------ | Branch (183:11): [True: 787, False: 0] ------------------ 184| 0| using enum Camera::SupportStatus; 185| 0| case Supported: ------------------ | Branch (185:3): [True: 0, False: 787] ------------------ 186| 0| case SupportedNoSamples: ------------------ | Branch (186:3): [True: 0, False: 787] ------------------ 187| 0| return true; // Explicitly supported. 188| 0| case Unsupported: ------------------ | Branch (188:3): [True: 0, False: 787] ------------------ 189| 0| ThrowRDE("Camera not supported (explicit). Sorry."); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 190| 787| case UnknownCamera: ------------------ | Branch (190:3): [True: 787, False: 0] ------------------ 191| 787| case UnknownNoSamples: ------------------ | Branch (191:3): [True: 0, False: 787] ------------------ 192| 787| case Unknown: ------------------ | Branch (192:3): [True: 0, False: 787] ------------------ 193| 787| if (failOnUnknown) { ------------------ | Branch (193:9): [True: 0, False: 787] ------------------ 194| 0| ThrowRDE("Camera '%s' '%s', mode '%s' not supported, and not allowed to " ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 195| 0| "guess. Sorry.", 196| 0| make.c_str(), model.c_str(), mode.c_str()); 197| 0| } 198| 787| return cam; // Might be implicitly supported. 199| 787| } 200| | 201| 0| return true; 202| 787|} _ZN8rawspeed10RawDecoder20checkCameraSupportedEPKNS_14CameraMetaDataERKNSt3__112basic_stringIcNS4_11char_traitsIcEENS4_9allocatorIcEEEESC_SC_: 207| 39| const std::string& mode) { 208| 39| mRaw->metadata.make = make; 209| 39| mRaw->metadata.model = model; 210| | 211| 39| if (!handleCameraSupport(meta, make, model, mode)) ------------------ | Branch (211:7): [True: 39, False: 0] ------------------ 212| 39| return false; 213| | 214| 0| const Camera* cam = meta->getCamera(make, model, mode); 215| 0| assert(cam); ------------------ | Branch (215:3): [True: 0, False: 0] ------------------ 216| | 217| 0| if (cam->decoderVersion > getDecoderVersion()) ------------------ | Branch (217:7): [True: 0, False: 0] ------------------ 218| 0| ThrowRDE( ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 219| 0| "Camera not supported in this version. Update RawSpeed for support."); 220| | 221| 0| hints = cam->hints; 222| 0| return true; 223| 0|} _ZN8rawspeed10RawDecoder11setMetaDataEPKNS_14CameraMetaDataERKNSt3__112basic_stringIcNS4_11char_traitsIcEENS4_9allocatorIcEEEESC_SC_i: 227| 748| const std::string& mode, int iso_speed) { 228| 748| mRaw->metadata.isoSpeed = iso_speed; 229| | 230| 748| if (!handleCameraSupport(meta, make, model, mode)) ------------------ | Branch (230:7): [True: 748, False: 0] ------------------ 231| 748| return; 232| | 233| 0| const Camera* cam = meta->getCamera(make, model, mode); 234| 0| assert(cam); ------------------ | Branch (234:3): [True: 0, False: 0] ------------------ 235| | 236| | // Only final CFA with the data from cameras.xml if it actually contained 237| | // the CFA. 238| 0| if (cam->cfa.getSize().area() > 0) ------------------ | Branch (238:7): [True: 0, False: 0] ------------------ 239| 0| mRaw->cfa = cam->cfa; 240| | 241| 0| if (!cam->color_matrix.empty()) ------------------ | Branch (241:7): [True: 0, False: 0] ------------------ 242| 0| mRaw->metadata.colorMatrix = cam->color_matrix; 243| | 244| 0| mRaw->metadata.canonical_make = cam->canonical_make; 245| 0| mRaw->metadata.canonical_model = cam->canonical_model; 246| 0| mRaw->metadata.canonical_alias = cam->canonical_alias; 247| 0| mRaw->metadata.canonical_id = cam->canonical_id; 248| 0| mRaw->metadata.make = make; 249| 0| mRaw->metadata.model = model; 250| 0| mRaw->metadata.mode = mode; 251| | 252| 0| if (applyCrop) { ------------------ | Branch (252:7): [True: 0, False: 0] ------------------ 253| 0| if (cam->cropAvailable) { ------------------ | Branch (253:9): [True: 0, False: 0] ------------------ 254| 0| iPoint2D new_size = cam->cropSize; 255| | 256| | // If crop size is negative, use relative cropping 257| 0| if (new_size.x <= 0) ------------------ | Branch (257:11): [True: 0, False: 0] ------------------ 258| 0| new_size.x = mRaw->dim.x - cam->cropPos.x + new_size.x; 259| | 260| 0| if (new_size.y <= 0) ------------------ | Branch (260:11): [True: 0, False: 0] ------------------ 261| 0| new_size.y = mRaw->dim.y - cam->cropPos.y + new_size.y; 262| | 263| 0| mRaw->subFrame(iRectangle2D(cam->cropPos, new_size)); 264| 0| } else { 265| 0| mRaw->subFrame(getDefaultCrop()); 266| 0| } 267| 0| } 268| | 269| 0| mRaw->blackAreas = cam->blackAreas; 270| 0| if (const CameraSensorInfo* sensor = cam->getSensorInfo(iso_speed)) { ------------------ | Branch (270:31): [True: 0, False: 0] ------------------ 271| 0| mRaw->blackLevel = sensor->mBlackLevel; 272| 0| mRaw->whitePoint = sensor->mWhiteLevel; 273| 0| if (mRaw->blackAreas.empty() && !sensor->mBlackLevelSeparate.empty()) { ------------------ | Branch (273:9): [True: 0, False: 0] | Branch (273:37): [True: 0, False: 0] ------------------ 274| 0| auto cfaArea = implicit_cast(mRaw->cfa.getSize().area()); 275| 0| if (mRaw->isCFA && ------------------ | Branch (275:11): [True: 0, False: 0] ------------------ 276| 0| cfaArea <= implicit_cast(sensor->mBlackLevelSeparate.size())) { ------------------ | Branch (276:11): [True: 0, False: 0] ------------------ 277| 0| mRaw->blackLevelSeparate = 278| 0| Array2DRef(mRaw->blackLevelSeparateStorage.data(), 2, 2); 279| 0| auto blackLevelSeparate1D = 280| 0| *mRaw->blackLevelSeparate->getAsArray1DRef(); 281| 0| for (int i = 0; i < cfaArea; i++) { ------------------ | Branch (281:25): [True: 0, False: 0] ------------------ 282| 0| blackLevelSeparate1D(i) = sensor->mBlackLevelSeparate[i]; 283| 0| } 284| 0| } else if (!mRaw->isCFA && ------------------ | Branch (284:18): [True: 0, False: 0] ------------------ 285| 0| mRaw->getCpp() <= sensor->mBlackLevelSeparate.size()) { ------------------ | Branch (285:18): [True: 0, False: 0] ------------------ 286| 0| mRaw->blackLevelSeparate = 287| 0| Array2DRef(mRaw->blackLevelSeparateStorage.data(), 2, 2); 288| 0| auto blackLevelSeparate1D = 289| 0| *mRaw->blackLevelSeparate->getAsArray1DRef(); 290| 0| for (uint32_t i = 0; i < mRaw->getCpp(); i++) { ------------------ | Branch (290:30): [True: 0, False: 0] ------------------ 291| 0| blackLevelSeparate1D(i) = sensor->mBlackLevelSeparate[i]; 292| 0| } 293| 0| } 294| 0| } 295| 0| } 296| | 297| | // Allow overriding individual blacklevels. Values are in CFA order 298| | // (the same order as the in the CFA tag) 299| | // A hint could be: 300| | // 301| 0| std::string cfa_black = hints.get("final_cfa_black", std::string()); 302| 0| if (!cfa_black.empty()) { ------------------ | Branch (302:7): [True: 0, False: 0] ------------------ 303| 0| vector v = splitString(cfa_black, ','); 304| 0| if (v.size() != 4) { ------------------ | Branch (304:9): [True: 0, False: 0] ------------------ 305| 0| mRaw->setError("Expected 4 values '10,20,30,20' as values for " 306| 0| "final_cfa_black hint."); 307| 0| } else { 308| 0| auto blackLevelSeparate1D = *mRaw->blackLevelSeparate->getAsArray1DRef(); 309| 0| for (int i = 0; i < 4; i++) { ------------------ | Branch (309:23): [True: 0, False: 0] ------------------ 310| 0| blackLevelSeparate1D(i) = stoi(v[i]); 311| 0| } 312| 0| } 313| 0| } 314| 0|} _ZN8rawspeed10RawDecoder9decodeRawEv: 320| 11.6k|rawspeed::RawImage RawDecoder::decodeRaw() { 321| 11.6k| try { 322| 11.6k| RawImage raw = decodeRawInternal(); 323| 11.6k| MSan::CheckMemIsInitialized(raw->getByteDataAsUncroppedArray2DRef()); 324| | 325| 11.6k| raw->metadata.pixelAspectRatio = 326| 11.6k| hints.get("pixel_aspect_ratio", raw->metadata.pixelAspectRatio); 327| 11.6k| if (interpolateBadPixels) { ------------------ | Branch (327:9): [True: 0, False: 11.6k] ------------------ 328| 0| raw->fixBadPixels(); 329| 0| MSan::CheckMemIsInitialized(raw->getByteDataAsUncroppedArray2DRef()); 330| 0| } 331| | 332| 11.6k| return raw; 333| 11.6k| } catch (const TiffParserException& e) { 334| 372| ThrowRDE("%s", e.what()); ------------------ | | 38| 372| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 372| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 372| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 372| __VA_ARGS__) | | ------------------ ------------------ 335| 372| } catch (const FileIOException& e) { 336| 0| ThrowRDE("%s", e.what()); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 337| 1.85k| } catch (const IOException& e) { 338| 1.85k| ThrowRDE("%s", e.what()); ------------------ | | 38| 1.85k| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1.85k| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1.85k| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1.85k| __VA_ARGS__) | | ------------------ ------------------ 339| 1.85k| } 340| 11.6k|} _ZN8rawspeed10RawDecoder14decodeMetaDataEPKNS_14CameraMetaDataE: 342| 3.07k|void RawDecoder::decodeMetaData(const CameraMetaData* meta) { 343| 3.07k| try { 344| 3.07k| decodeMetaDataInternal(meta); 345| 3.07k| } catch (const TiffParserException& e) { 346| 75| ThrowRDE("%s", e.what()); ------------------ | | 38| 75| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 75| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 75| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 75| __VA_ARGS__) | | ------------------ ------------------ 347| 75| } catch (const FileIOException& e) { 348| 0| ThrowRDE("%s", e.what()); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 349| 36| } catch (const IOException& e) { 350| 36| ThrowRDE("%s", e.what()); ------------------ | | 38| 36| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 36| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 36| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 36| __VA_ARGS__) | | ------------------ ------------------ 351| 36| } 352| 3.07k|} _ZNK8rawspeed10RawDecoderUt_cvbEv: 112| 326| explicit operator bool() const { return quadrantMultipliers /*|| ...*/; } _ZN8rawspeed10RawDecoderD2Ev: 45| 11.6k| virtual ~RawDecoder() = default; _ZN8rawspeed10Rw2Decoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 63| 2.52k| [[maybe_unused]] Buffer file) { 64| 2.52k| const auto id = rootIFD->getID(); 65| 2.52k| const std::string& make = id.make; 66| | 67| | // FIXME: magic 68| | 69| 2.52k| return make == "Panasonic" || make == "LEICA" || make == "LEICA CAMERA AG"; ------------------ | Branch (69:10): [True: 707, False: 1.81k] | Branch (69:33): [True: 2, False: 1.81k] | Branch (69:52): [True: 0, False: 1.81k] ------------------ 70| 2.52k|} _ZNK8rawspeed10Rw2Decoder11decodeRawV8ERKNS_7TiffIFDE: 243| 6|RawImage Rw2Decoder::decodeRawV8(const TiffIFD& raw) const { 244| 6| parseCFA(); 245| 6| if (getAsBayerPhase(mRaw->cfa) != BayerPhase::RGGB) ------------------ | Branch (245:7): [True: 3, False: 3] ------------------ 246| 6| ThrowRDE("Unexpected CFA, only RGGB is supported"); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 247| | 248| 3| const DecompressorV8Params mParams(raw); 249| 3| populateGammaLUT(mParams, raw); 250| 3| const std::vector> mStrips = 251| 3| getInputStrips(mParams, mFile); 252| | 253| 3| PanasonicV8Decompressor::DecompressorParamsBuilder b( 254| 3| mRaw->dim, mParams.initialPrediction, getAsArray1DRef(mStrips), 255| 3| getAsArray1DRef(mParams.stripLineOffsets), 256| 3| getAsArray1DRef(mParams.stripWidths), 257| 3| getAsArray1DRef(mParams.stripHeights), 258| 3| raw.getEntry(TiffTag::PANASONIC_V8_HUF_TABLE)->getData()); 259| | 260| 3| PanasonicV8Decompressor v8(mRaw, b.getDecompressorParams()); 261| 3| mRaw->createData(); 262| 3| v8.decompress(); 263| 3| return mRaw; 264| 6|} Rw2Decoder.cpp:_ZN8rawspeed12_GLOBAL__N_120DecompressorV8ParamsC2ERKNS_7TiffIFDE: 145| 2|DecompressorV8Params::DecompressorV8Params(const TiffIFD& ifd) { 146| | // NOLINTBEGIN(cppcoreguidelines-prefer-member-initializer) 147| 2| horizontalStripCount = 148| 2| ifd.getEntry(TiffTag::PANASONIC_V8_NUMBER_OF_STRIPS_H)->getU16(); 149| 2| verticalStripCount = 150| 2| ifd.getEntry(TiffTag::PANASONIC_V8_NUMBER_OF_STRIPS_V)->getU16(); 151| | 152| 2| getPanasonicTiffVector(ifd, TiffTag::PANASONIC_V8_STRIP_BYTE_OFFSETS, 153| 2| stripByteOffsets); 154| 2| getPanasonicTiffVector(ifd, TiffTag::PANASONIC_V8_STRIP_LINE_OFFSETS, 155| 2| stripLineOffsets); 156| 2| getPanasonicTiffVector(ifd, TiffTag::PANASONIC_V8_STRIP_DATA_SIZE, 157| 2| stripBitLengths); 158| 2| getPanasonicTiffVector(ifd, TiffTag::PANASONIC_V8_STRIP_WIDTHS, stripWidths); 159| 2| getPanasonicTiffVector(ifd, TiffTag::PANASONIC_V8_STRIP_HEIGHTS, 160| 2| stripHeights); 161| | 162| | // Get decoder's initial prediction value: 163| | // Note, the positions of the green samples are swapped. This is intentional, 164| | // the original implementation did this each swap redundantly during decoding 165| | // of each tile. 166| 2| initialPrediction[0] = 167| 2| ifd.getEntry(TiffTag::PANASONIC_V8_INIT_PRED_RED)->getU16(); 168| 2| initialPrediction[2] = 169| 2| ifd.getEntry(TiffTag::PANASONIC_V8_INIT_PRED_GREEN1)->getU16(); 170| 2| initialPrediction[1] = 171| 2| ifd.getEntry(TiffTag::PANASONIC_V8_INIT_PRED_GREEN2)->getU16(); 172| 2| initialPrediction[3] = 173| 2| ifd.getEntry(TiffTag::PANASONIC_V8_INIT_PRED_BLUE)->getU16(); 174| | 175| 2| getPanasonicTiffVector(ifd, TiffTag::PANASONIC_V8_HUF_SHIFT_DOWN, shiftDown); 176| | 177| 2| gammaClipVal = ifd.getEntry(TiffTag::PANASONIC_V8_CLIP_VAL)->getU16(); 178| | // NOLINTEND(cppcoreguidelines-prefer-member-initializer) 179| | 180| 2| validate(); 181| 2|} Rw2Decoder.cpp:_ZN8rawspeed12_GLOBAL__N_122getPanasonicTiffVectorIjEEvRKNS_7TiffIFDENS_7TiffTagERNSt3__16vectorIT_NS6_9allocatorIS8_EEEE: 77| 1| std::vector& output) { 78| 1| ByteStream bs = ifd.getEntry(tag)->getData(); 79| 1| output.resize(bs.getU16()); 80| | 81| | // Note: Relying on ByteStream and its parent classes to prevent out-of-bounds 82| | // reading. 83| 1| for (T& v : output) ------------------ | Branch (83:13): [True: 0, False: 1] ------------------ 84| 0| v = bs.get(); 85| 1|} _ZN8rawspeed10Rw2Decoder17decodeRawInternalEv: 266| 709|RawImage Rw2Decoder::decodeRawInternal() { 267| | 268| 709| const TiffIFD* raw = nullptr; 269| 709| bool isOldPanasonic = 270| 709| !mRootIFD->hasEntryRecursive(TiffTag::PANASONIC_STRIPOFFSET); 271| | 272| 709| if (!isOldPanasonic) ------------------ | Branch (272:7): [True: 365, False: 344] ------------------ 273| 365| raw = mRootIFD->getIFDWithTag(TiffTag::PANASONIC_STRIPOFFSET); 274| 344| else 275| 344| raw = mRootIFD->getIFDWithTag(TiffTag::STRIPOFFSETS); 276| | 277| 709| uint32_t height = raw->getEntry(static_cast(3))->getU16(); 278| 709| uint32_t width = raw->getEntry(static_cast(2))->getU16(); 279| | 280| 709| if (isOldPanasonic) { ------------------ | Branch (280:7): [True: 341, False: 368] ------------------ 281| 341| if (width == 0 || height == 0 || width > 4330 || height > 2751) ------------------ | Branch (281:9): [True: 1, False: 340] | Branch (281:23): [True: 2, False: 338] | Branch (281:38): [True: 10, False: 328] | Branch (281:54): [True: 9, False: 319] ------------------ 282| 341| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 22| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 22| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 22| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 22| __VA_ARGS__) | | ------------------ ------------------ 283| | 284| 319| const TiffEntry* offsets = raw->getEntry(TiffTag::STRIPOFFSETS); 285| | 286| 319| if (offsets->count != 1) { ------------------ | Branch (286:9): [True: 2, False: 317] ------------------ 287| 2| ThrowRDE("Multiple Strips found: %u", offsets->count); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 288| 2| } 289| 317| uint32_t offset = offsets->getU32(); 290| 317| if (!mFile.isValid(offset)) ------------------ | Branch (290:9): [True: 42, False: 275] ------------------ 291| 317| ThrowRDE("Invalid image data offset, cannot decode."); ------------------ | | 38| 42| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 42| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 42| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 42| __VA_ARGS__) | | ------------------ ------------------ 292| | 293| 275| mRaw->dim = iPoint2D(width, height); 294| | 295| 275| uint32_t size = mFile.getSize() - offset; 296| | 297| 275| if (size >= width * height * 2) { ------------------ | Branch (297:9): [True: 72, False: 203] ------------------ 298| | // It's completely unpacked little-endian 299| 72| UncompressedDecompressor u( 300| 72| ByteStream(DataBuffer(mFile.getSubView(offset), Endianness::little)), 301| 72| mRaw, iRectangle2D({0, 0}, iPoint2D(width, height)), 16 * width / 8, 302| 72| 16, BitOrder::LSB); 303| 72| mRaw->createData(); 304| 72| u.decode12BitRawUnpackedLeftAligned(); 305| 203| } else if (size >= width * height * 3 / 2) { ------------------ | Branch (305:16): [True: 51, False: 152] ------------------ 306| | // It's a packed format 307| 51| UncompressedDecompressor u( 308| 51| ByteStream(DataBuffer(mFile.getSubView(offset), Endianness::little)), 309| 51| mRaw, iRectangle2D({0, 0}, iPoint2D(width, height)), 310| 51| (12 * width / 8) + ((width + 2) / 10), 12, BitOrder::LSB); 311| 51| mRaw->createData(); 312| 51| u.decode12BitRawWithControl(); 313| 152| } else { 314| 152| uint32_t section_split_offset = 0; 315| 152| PanasonicV4Decompressor p( 316| 152| mRaw, 317| 152| ByteStream(DataBuffer(mFile.getSubView(offset), Endianness::little)), 318| 152| hints.contains("zero_is_not_bad"), section_split_offset); 319| 152| mRaw->createData(); 320| 152| p.decompress(); 321| 152| } 322| 368| } else { 323| 368| mRaw->dim = iPoint2D(width, height); 324| | 325| 368| const TiffEntry* offsets = raw->getEntry(TiffTag::PANASONIC_STRIPOFFSET); 326| | 327| 368| if (offsets->count != 1) { ------------------ | Branch (327:9): [True: 1, False: 367] ------------------ 328| 1| ThrowRDE("Multiple Strips found: %u", offsets->count); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 329| 1| } 330| | 331| 367| uint32_t offset = offsets->getU32(); 332| | 333| 367| ByteStream bs(DataBuffer(mFile.getSubView(offset), Endianness::little)); 334| | 335| 367| uint16_t bitsPerSample = 12; 336| 367| if (raw->hasEntry(TiffTag::PANASONIC_BITSPERSAMPLE)) ------------------ | Branch (336:9): [True: 165, False: 202] ------------------ 337| 165| bitsPerSample = raw->getEntry(TiffTag::PANASONIC_BITSPERSAMPLE)->getU16(); 338| | 339| 367| switch (uint16_t version = 340| 367| raw->getEntry(TiffTag::PANASONIC_RAWFORMAT)->getU16()) { 341| 31| case 4: { ------------------ | Branch (341:5): [True: 31, False: 336] ------------------ 342| 31| uint32_t section_split_offset = 0x1FF8; 343| 31| PanasonicV4Decompressor p(mRaw, bs, hints.contains("zero_is_not_bad"), 344| 31| section_split_offset); 345| 31| mRaw->createData(); 346| 31| p.decompress(); 347| 31| return mRaw; 348| 0| } 349| 116| case 5: { ------------------ | Branch (349:5): [True: 116, False: 251] ------------------ 350| 116| PanasonicV5Decompressor v5(mRaw, bs, bitsPerSample); 351| 116| mRaw->createData(); 352| 116| v5.decompress(); 353| 116| return mRaw; 354| 0| } 355| 122| case 6: { ------------------ | Branch (355:5): [True: 122, False: 245] ------------------ 356| 122| if (bitsPerSample != 14 && bitsPerSample != 12) ------------------ | Branch (356:11): [True: 75, False: 47] | Branch (356:34): [True: 1, False: 74] ------------------ 357| 122| ThrowRDE("Version %i: unexpected bits per sample: %i", version, ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 358| 121| bitsPerSample); 359| | 360| 121| PanasonicV6Decompressor v6(mRaw, bs, bitsPerSample); 361| 121| mRaw->createData(); 362| 121| v6.decompress(); 363| 121| return mRaw; 364| 122| } 365| 73| case 7: { ------------------ | Branch (365:5): [True: 73, False: 294] ------------------ 366| 73| if (bitsPerSample != 14) ------------------ | Branch (366:11): [True: 9, False: 64] ------------------ 367| 73| ThrowRDE("Version %i: unexpected bits per sample: %i", version, ------------------ | | 38| 9| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 9| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 9| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 9| __VA_ARGS__) | | ------------------ ------------------ 368| 64| bitsPerSample); 369| 64| PanasonicV7Decompressor v7(mRaw, bs); 370| 64| mRaw->createData(); 371| 64| v7.decompress(); 372| 64| return mRaw; 373| 73| } 374| 7| case 8: { ------------------ | Branch (374:5): [True: 7, False: 360] ------------------ 375| | // Known values are 12, 14, and 16. Other less than 16 should decompress 376| | // fine. 377| 7| if (bitsPerSample > 16) ------------------ | Branch (377:11): [True: 1, False: 6] ------------------ 378| 7| ThrowRDE("Version %i: unexpected bits per sample: %i", version, ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 379| 6| bitsPerSample); 380| 6| return decodeRawV8(*raw); 381| 7| } 382| 1| default: ------------------ | Branch (382:5): [True: 1, False: 366] ------------------ 383| 1| ThrowRDE("Version %i is unsupported", version); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 384| 367| } 385| 367| } 386| | 387| 275| return mRaw; 388| 709|} _ZNK8rawspeed10Rw2Decoder8parseCFAEv: 396| 488|void Rw2Decoder::parseCFA() const { 397| 488| if (!mRootIFD->hasEntryRecursive(TiffTag::PANASONIC_CFAPATTERN)) ------------------ | Branch (397:7): [True: 438, False: 50] ------------------ 398| 488| ThrowRDE("No PANASONIC_CFAPATTERN entry found!"); ------------------ | | 38| 438| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 438| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 438| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 438| __VA_ARGS__) | | ------------------ ------------------ 399| | 400| 50| const TiffEntry* CFA = 401| 50| mRootIFD->getEntryRecursive(TiffTag::PANASONIC_CFAPATTERN); 402| 50| if (CFA->type != TiffDataType::SHORT || CFA->count != 1) { ------------------ | Branch (402:7): [True: 3, False: 47] | Branch (402:43): [True: 1, False: 46] ------------------ 403| 4| ThrowRDE("Bad PANASONIC_CFAPATTERN entry (type %u, count %u).", ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 404| 4| static_cast(CFA->type), CFA->count); 405| 4| } 406| | 407| 46| switch (auto i = CFA->getU16()) { 408| 0| using enum CFAColor; 409| 13| case 1: ------------------ | Branch (409:3): [True: 13, False: 33] ------------------ 410| 13| mRaw->cfa.setCFA(iPoint2D(2, 2), RED, GREEN, GREEN, BLUE); 411| 13| break; 412| 8| case 2: ------------------ | Branch (412:3): [True: 8, False: 38] ------------------ 413| 8| mRaw->cfa.setCFA(iPoint2D(2, 2), GREEN, RED, BLUE, GREEN); 414| 8| break; 415| 19| case 3: ------------------ | Branch (415:3): [True: 19, False: 27] ------------------ 416| 19| mRaw->cfa.setCFA(iPoint2D(2, 2), GREEN, BLUE, RED, GREEN); 417| 19| break; 418| 5| case 4: ------------------ | Branch (418:3): [True: 5, False: 41] ------------------ 419| 5| mRaw->cfa.setCFA(iPoint2D(2, 2), BLUE, GREEN, GREEN, RED); 420| 5| break; 421| 1| default: ------------------ | Branch (421:3): [True: 1, False: 45] ------------------ 422| 1| ThrowRDE("Unexpected CFA pattern: %u", i); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 423| 46| } 424| 46|} _ZNK8rawspeed10Rw2Decoder6getRawEv: 426| 39|const TiffIFD* Rw2Decoder::getRaw() const { 427| 39| return mRootIFD->hasEntryRecursive(TiffTag::PANASONIC_STRIPOFFSET) ------------------ | Branch (427:10): [True: 1, False: 38] ------------------ 428| 39| ? mRootIFD->getIFDWithTag(TiffTag::PANASONIC_STRIPOFFSET) 429| 39| : mRootIFD->getIFDWithTag(TiffTag::STRIPOFFSETS); 430| 39|} _ZN8rawspeed10Rw2Decoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 432| 482|void Rw2Decoder::decodeMetaDataInternal(const CameraMetaData* meta) { 433| 482| parseCFA(); 434| | 435| 482| auto id = mRootIFD->getID(); 436| 482| std::string mode = guessMode(); 437| 482| int iso = 0; 438| 482| if (mRootIFD->hasEntryRecursive(TiffTag::PANASONIC_ISO_SPEED)) ------------------ | Branch (438:7): [True: 3, False: 479] ------------------ 439| 3| iso = mRootIFD->getEntryRecursive(TiffTag::PANASONIC_ISO_SPEED)->getU32(); 440| | 441| 482| if (this->checkCameraSupported(meta, id, mode)) { ------------------ | Branch (441:7): [True: 0, False: 482] ------------------ 442| 0| setMetaData(meta, id, mode, iso); 443| 482| } else { 444| 482| mRaw->metadata.mode = mode; 445| 482| writeLog(DEBUG_PRIO::EXTRA, "Mode not found in DB: %s", mode.c_str()); 446| 482| setMetaData(meta, id, "", iso); 447| 482| } 448| | 449| 482| const TiffIFD* raw = getRaw(); 450| | 451| | // Read blacklevels 452| 482| if (raw->hasEntry(static_cast(0x1c)) && ------------------ | Branch (452:7): [True: 20, False: 462] ------------------ 453| 20| raw->hasEntry(static_cast(0x1d)) && ------------------ | Branch (453:7): [True: 11, False: 9] ------------------ 454| 11| raw->hasEntry(static_cast(0x1e))) { ------------------ | Branch (454:7): [True: 6, False: 5] ------------------ 455| 6| auto blackLevelsNeedOffsetting = [&]() { 456| 6| bool isOldPanasonic = 457| 6| !mRootIFD->hasEntryRecursive(TiffTag::PANASONIC_STRIPOFFSET); 458| 6| if (isOldPanasonic) 459| 6| return true; 460| 6| const uint16_t version = 461| 6| raw->getEntry(TiffTag::PANASONIC_RAWFORMAT)->getU16(); 462| | // After version 4 the black levels appears to be correct. 463| 6| return version <= 4; 464| 6| }; 465| 6| const auto getBlack = [&raw, blackLevelsNeedOffsetting](TiffTag t) { 466| 6| const int val = raw->getEntry(t)->getU16(); 467| 6| if (!blackLevelsNeedOffsetting()) 468| 6| return val; 469| | // Continue adding 15 for older raw versions. 470| 6| int out; 471| 6| if (__builtin_sadd_overflow(val, 15, &out)) 472| 6| ThrowRDE("Integer overflow when calculating black level"); 473| 6| return out; 474| 6| }; 475| | 476| 6| const int blackRed = getBlack(static_cast(0x1c)); 477| 6| const int blackGreen = getBlack(static_cast(0x1d)); 478| 6| const int blackBlue = getBlack(static_cast(0x1e)); 479| | 480| 6| mRaw->blackLevelSeparate = 481| 6| Array2DRef(mRaw->blackLevelSeparateStorage.data(), 2, 2); 482| 6| auto blackLevelSeparate1D = *mRaw->blackLevelSeparate->getAsArray1DRef(); 483| 12| for (int i = 0; i < 2; i++) { ------------------ | Branch (483:21): [True: 6, False: 6] ------------------ 484| 18| for (int j = 0; j < 2; j++) { ------------------ | Branch (484:23): [True: 12, False: 6] ------------------ 485| 12| const int k = i + (2 * j); 486| 12| const CFAColor c = mRaw->cfa.getColorAt(i, j); 487| 12| switch (c) { 488| 3| case CFAColor::RED: ------------------ | Branch (488:9): [True: 3, False: 9] ------------------ 489| 3| blackLevelSeparate1D(k) = blackRed; 490| 3| break; 491| 6| case CFAColor::GREEN: ------------------ | Branch (491:9): [True: 6, False: 6] ------------------ 492| 6| blackLevelSeparate1D(k) = blackGreen; 493| 6| break; 494| 3| case CFAColor::BLUE: ------------------ | Branch (494:9): [True: 3, False: 9] ------------------ 495| 3| blackLevelSeparate1D(k) = blackBlue; 496| 3| break; 497| 0| default: ------------------ | Branch (497:9): [True: 0, False: 12] ------------------ 498| 0| ThrowRDE("Unexpected CFA color %s.", ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 499| 12| ColorFilterArray::colorToString(c).c_str()); 500| 12| } 501| 12| } 502| 6| } 503| 6| } 504| | 505| | // Read WB levels 506| 482| if (raw->hasEntry(static_cast(0x0024)) && ------------------ | Branch (506:7): [True: 10, False: 472] ------------------ 507| 10| raw->hasEntry(static_cast(0x0025)) && ------------------ | Branch (507:7): [True: 6, False: 4] ------------------ 508| 6| raw->hasEntry(static_cast(0x0026))) { ------------------ | Branch (508:7): [True: 1, False: 5] ------------------ 509| 1| std::array wbCoeffs = {}; 510| 1| wbCoeffs[0] = static_cast( 511| 1| raw->getEntry(static_cast(0x0024))->getU16()); 512| 1| wbCoeffs[1] = static_cast( 513| 1| raw->getEntry(static_cast(0x0025))->getU16()); 514| 1| wbCoeffs[2] = static_cast( 515| 1| raw->getEntry(static_cast(0x0026))->getU16()); 516| 1| mRaw->metadata.wbCoeffs = wbCoeffs; 517| 481| } else if (raw->hasEntry(static_cast(0x0011)) && ------------------ | Branch (517:14): [True: 7, False: 474] ------------------ 518| 7| raw->hasEntry(static_cast(0x0012))) { ------------------ | Branch (518:14): [True: 3, False: 4] ------------------ 519| 3| std::array wbCoeffs = {}; 520| 3| wbCoeffs[0] = static_cast( 521| 3| raw->getEntry(static_cast(0x0011))->getU16()); 522| 3| wbCoeffs[1] = 256.0F; 523| 3| wbCoeffs[2] = static_cast( 524| 3| raw->getEntry(static_cast(0x0012))->getU16()); 525| 3| mRaw->metadata.wbCoeffs = wbCoeffs; 526| 3| } 527| 482|} Rw2Decoder.cpp:_ZZN8rawspeed10Rw2Decoder22decodeMetaDataInternalEPKNS_14CameraMetaDataEENK3$_1clENS_7TiffTagE: 465| 15| const auto getBlack = [&raw, blackLevelsNeedOffsetting](TiffTag t) { 466| 15| const int val = raw->getEntry(t)->getU16(); 467| 15| if (!blackLevelsNeedOffsetting()) ------------------ | Branch (467:11): [True: 0, False: 15] ------------------ 468| 0| return val; 469| | // Continue adding 15 for older raw versions. 470| 15| int out; 471| 15| if (__builtin_sadd_overflow(val, 15, &out)) ------------------ | Branch (471:11): [True: 0, False: 15] ------------------ 472| 15| ThrowRDE("Integer overflow when calculating black level"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 473| 15| return out; 474| 15| }; Rw2Decoder.cpp:_ZZN8rawspeed10Rw2Decoder22decodeMetaDataInternalEPKNS_14CameraMetaDataEENK3$_0clEv: 455| 12| auto blackLevelsNeedOffsetting = [&]() { 456| 12| bool isOldPanasonic = 457| 12| !mRootIFD->hasEntryRecursive(TiffTag::PANASONIC_STRIPOFFSET); 458| 12| if (isOldPanasonic) ------------------ | Branch (458:11): [True: 12, False: 0] ------------------ 459| 12| return true; 460| 0| const uint16_t version = 461| 0| raw->getEntry(TiffTag::PANASONIC_RAWFORMAT)->getU16(); 462| | // After version 4 the black levels appears to be correct. 463| 0| return version <= 4; 464| 12| }; _ZNK8rawspeed10Rw2Decoder9guessModeEv: 529| 40|std::string Rw2Decoder::guessMode() const { 530| 40| float ratio = 3.0F / 2.0F; // Default 531| | 532| 40| if (!mRaw->isAllocated()) ------------------ | Branch (532:7): [True: 0, False: 40] ------------------ 533| 0| return ""; 534| | 535| 40| ratio = static_cast(mRaw->dim.x) / static_cast(mRaw->dim.y); 536| | 537| 40| float min_diff = fabs(ratio - (16.0F / 9.0F)); 538| 40| std::string closest_match = "16:9"; 539| | 540| 40| float t = fabs(ratio - (3.0F / 2.0F)); 541| 40| if (t < min_diff) { ------------------ | Branch (541:7): [True: 14, False: 26] ------------------ 542| 14| closest_match = "3:2"; 543| 14| min_diff = t; 544| 14| } 545| | 546| 40| t = fabs(ratio - (4.0F / 3.0F)); 547| 40| if (t < min_diff) { ------------------ | Branch (547:7): [True: 14, False: 26] ------------------ 548| 14| closest_match = "4:3"; 549| 14| min_diff = t; 550| 14| } 551| | 552| 40| t = fabs(ratio - 1.0F); 553| 40| if (t < min_diff) { ------------------ | Branch (553:7): [True: 13, False: 27] ------------------ 554| 13| closest_match = "1:1"; 555| 13| min_diff = t; 556| 13| } 557| 40| writeLog(DEBUG_PRIO::EXTRA, "Mode guess: '%s'", closest_match.c_str()); 558| 40| return closest_match; 559| 40|} _ZN8rawspeed10Rw2DecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 41| 709| : AbstractTiffDecoder(std::move(root), file) {} _ZN8rawspeed17SimpleTiffDecoder21prepareForRawDecodingEv: 39| 492|void SimpleTiffDecoder::prepareForRawDecoding() { 40| 492| raw = getIFDWithLargestImage(); 41| 492| width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 42| 492| height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 43| 492| off = raw->getEntry(TiffTag::STRIPOFFSETS)->getU32(); 44| 492| c2 = raw->getEntry(TiffTag::STRIPBYTECOUNTS)->getU32(); 45| | 46| 492| if (!mFile.isValid(off, c2)) ------------------ | Branch (46:7): [True: 39, False: 453] ------------------ 47| 492| ThrowRDE("Image is truncated."); ------------------ | | 38| 39| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 39| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 39| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 39| __VA_ARGS__) | | ------------------ ------------------ 48| | 49| 453| if (c2 == 0) ------------------ | Branch (49:7): [True: 1, False: 452] ------------------ 50| 453| ThrowRDE("No image data found."); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 51| | 52| 452| if (0 == width || 0 == height) ------------------ | Branch (52:7): [True: 18, False: 434] | Branch (52:21): [True: 1, False: 433] ------------------ 53| 452| ThrowRDE("Image has zero size."); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 54| | 55| 450| checkImageDimensions(); 56| | 57| 450| mRaw->dim = iPoint2D(width, height); 58| 450|} _ZN8rawspeed17SimpleTiffDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 42| 492| : AbstractTiffDecoder(std::move(root), file) {} _ZN8rawspeed10SrwDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 47| 1.81k| [[maybe_unused]] Buffer file) { 48| 1.81k| const auto id = rootIFD->getID(); 49| 1.81k| const std::string& make = id.make; 50| | 51| | // FIXME: magic 52| | 53| 1.81k| return make == "SAMSUNG"; 54| 1.81k|} _ZN8rawspeed10SrwDecoder17decodeRawInternalEv: 56| 544|RawImage SrwDecoder::decodeRawInternal() { 57| 544| const auto* raw = mRootIFD->getIFDWithTag(TiffTag::STRIPOFFSETS); 58| | 59| 544| int compression = raw->getEntry(TiffTag::COMPRESSION)->getU32(); 60| 544| const int bits = raw->getEntry(TiffTag::BITSPERSAMPLE)->getU32(); 61| | 62| 544| if (12 != bits && 14 != bits) ------------------ | Branch (62:7): [True: 232, False: 312] | Branch (62:21): [True: 30, False: 202] ------------------ 63| 544| ThrowRDE("Unsupported bits per sample"); ------------------ | | 38| 30| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 30| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 30| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 30| __VA_ARGS__) | | ------------------ ------------------ 64| | 65| 514| if (32769 != compression && 32770 != compression && 32772 != compression && ------------------ | Branch (65:7): [True: 508, False: 6] | Branch (65:31): [True: 342, False: 166] | Branch (65:55): [True: 244, False: 98] ------------------ 66| 244| 32773 != compression) ------------------ | Branch (66:7): [True: 27, False: 217] ------------------ 67| 514| ThrowRDE("Unsupported compression"); ------------------ | | 38| 27| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 27| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 27| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 27| __VA_ARGS__) | | ------------------ ------------------ 68| | 69| 487| if (uint32_t nslices = raw->getEntry(TiffTag::STRIPOFFSETS)->count; 70| 487| nslices != 1) ------------------ | Branch (70:7): [True: 1, False: 486] ------------------ 71| 487| ThrowRDE("Only one slice supported, found %u", nslices); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 72| | 73| 486| if (const auto wrongComp = 74| 486| 32770 == compression && !raw->hasEntry(static_cast(40976)); ------------------ | Branch (74:11): [True: 166, False: 320] | Branch (74:35): [True: 8, False: 158] ------------------ 75| 486| 32769 == compression || wrongComp) { ------------------ | Branch (75:7): [True: 6, False: 480] | Branch (75:31): [True: 8, False: 472] ------------------ 76| 9| bool bit_order = hints.get("msb_override", wrongComp ? bits == 12 : false); ------------------ | Branch (76:48): [True: 8, False: 1] ------------------ 77| 9| this->decodeUncompressed(raw, bit_order ? BitOrder::MSB : BitOrder::LSB); ------------------ | Branch (77:35): [True: 5, False: 4] ------------------ 78| 9| return mRaw; 79| 9| } 80| | 81| 477| const uint32_t width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 82| 477| const uint32_t height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 83| 477| mRaw->dim = iPoint2D(width, height); 84| | 85| 477| if (32770 == compression) { ------------------ | Branch (85:7): [True: 158, False: 319] ------------------ 86| 158| const TiffEntry* sliceOffsets = raw->getEntry(static_cast(40976)); 87| 158| if (sliceOffsets->type != TiffDataType::LONG || sliceOffsets->count != 1) ------------------ | Branch (87:9): [True: 1, False: 157] | Branch (87:53): [True: 1, False: 156] ------------------ 88| 158| ThrowRDE("Entry 40976 is corrupt"); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 89| | 90| 156| ByteStream bso(DataBuffer(mFile, Endianness::little)); 91| 156| bso.skipBytes(sliceOffsets->getU32()); 92| 156| bso = bso.getStream(height, 4); 93| | 94| 156| const uint32_t offset = raw->getEntry(TiffTag::STRIPOFFSETS)->getU32(); 95| 156| const uint32_t count = raw->getEntry(TiffTag::STRIPBYTECOUNTS)->getU32(); 96| 156| Buffer rbuf(mFile.getSubView(offset, count)); 97| 156| ByteStream bsr(DataBuffer(rbuf, Endianness::little)); 98| | 99| 156| SamsungV0Decompressor s0(mRaw, bso, bsr); 100| | 101| 156| mRaw->createData(); 102| | 103| 156| s0.decompress(); 104| | 105| 156| return mRaw; 106| 158| } 107| 319| if (32772 == compression) { ------------------ | Branch (107:7): [True: 98, False: 221] ------------------ 108| 98| uint32_t offset = raw->getEntry(TiffTag::STRIPOFFSETS)->getU32(); 109| 98| uint32_t count = raw->getEntry(TiffTag::STRIPBYTECOUNTS)->getU32(); 110| 98| const ByteStream bs( 111| 98| DataBuffer(mFile.getSubView(offset, count), Endianness::little)); 112| | 113| 98| SamsungV1Decompressor s1(mRaw, bs, bits); 114| | 115| 98| mRaw->createData(); 116| | 117| 98| s1.decompress(); 118| | 119| 98| return mRaw; 120| 98| } 121| 221| if (32773 == compression) { ------------------ | Branch (121:7): [True: 216, False: 5] ------------------ 122| 216| uint32_t offset = raw->getEntry(TiffTag::STRIPOFFSETS)->getU32(); 123| 216| uint32_t count = raw->getEntry(TiffTag::STRIPBYTECOUNTS)->getU32(); 124| 216| const ByteStream bs( 125| 216| DataBuffer(mFile.getSubView(offset, count), Endianness::little)); 126| | 127| 216| SamsungV2Decompressor s2(mRaw, bs, bits); 128| | 129| 216| mRaw->createData(); 130| | 131| 216| s2.decompress(); 132| | 133| 216| return mRaw; 134| 216| } 135| 5| ThrowRDE("Unsupported compression"); ------------------ | | 38| 5| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 5| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 5| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 5| __VA_ARGS__) | | ------------------ ------------------ 136| 221|} _ZNK8rawspeed10SrwDecoder7getModeEv: 138| 35|std::string SrwDecoder::getMode() const { 139| 35| std::vector data = 140| 35| mRootIFD->getIFDsWithTag(TiffTag::CFAPATTERN); 141| 35| std::ostringstream mode; 142| 35| if (!data.empty() && data[0]->hasEntryRecursive(TiffTag::BITSPERSAMPLE)) { ------------------ | Branch (142:7): [True: 5, False: 30] | Branch (142:24): [True: 3, False: 2] ------------------ 143| 3| mode << data[0]->getEntryRecursive(TiffTag::BITSPERSAMPLE)->getU32() 144| 3| << "bit"; 145| 3| return mode.str(); 146| 3| } 147| 32| return ""; 148| 35|} _ZN8rawspeed10SrwDecoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 159| 35|void SrwDecoder::decodeMetaDataInternal(const CameraMetaData* meta) { 160| 35| int iso = 0; 161| 35| if (mRootIFD->hasEntryRecursive(TiffTag::ISOSPEEDRATINGS)) ------------------ | Branch (161:7): [True: 1, False: 34] ------------------ 162| 1| iso = mRootIFD->getEntryRecursive(TiffTag::ISOSPEEDRATINGS)->getU32(); 163| | 164| 35| auto id = mRootIFD->getID(); 165| 35| if (std::string mode = getMode(); meta->hasCamera(id.make, id.model, mode)) ------------------ | Branch (165:37): [True: 0, False: 35] ------------------ 166| 0| setMetaData(meta, id, mode, iso); 167| 35| else 168| 35| setMetaData(meta, id, "", iso); 169| | 170| | // Set the whitebalance 171| 35| if (mRootIFD->hasEntryRecursive(TiffTag::SAMSUNG_WB_RGGBLEVELSUNCORRECTED) && ------------------ | Branch (171:7): [True: 8, False: 27] ------------------ 172| 8| mRootIFD->hasEntryRecursive(TiffTag::SAMSUNG_WB_RGGBLEVELSBLACK)) { ------------------ | Branch (172:7): [True: 5, False: 3] ------------------ 173| 5| const TiffEntry* wb_levels = 174| 5| mRootIFD->getEntryRecursive(TiffTag::SAMSUNG_WB_RGGBLEVELSUNCORRECTED); 175| 5| const TiffEntry* wb_black = 176| 5| mRootIFD->getEntryRecursive(TiffTag::SAMSUNG_WB_RGGBLEVELSBLACK); 177| 5| if (wb_levels->count == 4 && wb_black->count == 4) { ------------------ | Branch (177:9): [True: 3, False: 2] | Branch (177:34): [True: 1, False: 2] ------------------ 178| 1| std::array wbCoeffs = {}; 179| 1| wbCoeffs[0] = wb_levels->getFloat(0) - wb_black->getFloat(0); 180| 1| wbCoeffs[1] = wb_levels->getFloat(1) - wb_black->getFloat(1); 181| 1| wbCoeffs[2] = wb_levels->getFloat(3) - wb_black->getFloat(3); 182| 1| mRaw->metadata.wbCoeffs = wbCoeffs; 183| 1| } 184| 5| } 185| 35|} _ZN8rawspeed10SrwDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 39| 544| : AbstractTiffDecoder(std::move(root), file) {} _ZN8rawspeed10StiDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 43| 689| [[maybe_unused]] Buffer file) { 44| 689| const auto id = rootIFD->getID(); 45| 689| const std::string& make = id.make; 46| | 47| | // FIXME: magic 48| | 49| 689| return make == "Sinar AG"; 50| 689|} _ZN8rawspeed10StiDecoder17decodeRawInternalEv: 52| 26|RawImage StiDecoder::decodeRawInternal() { 53| 26| const auto* raw = mRootIFD->getIFDWithTag(TiffTag::TILEOFFSETS, 0); 54| 26| uint32_t width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 55| 26| uint32_t height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 56| 26| uint32_t compression = raw->getEntry(TiffTag::COMPRESSION)->getU32(); 57| | 58| 26| mRaw->dim = iPoint2D(width, height); 59| | 60| 26| if (1 != compression) ------------------ | Branch (60:7): [True: 3, False: 23] ------------------ 61| 26| ThrowRDE("Unexpected compression type."); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 62| | 63| 23| DecodeUncompressed(raw); 64| 23| return mRaw; 65| 26|} _ZNK8rawspeed10StiDecoder18DecodeUncompressedEPKNS_7TiffIFDE: 67| 22|void StiDecoder::DecodeUncompressed(const TiffIFD* raw) const { 68| 22| if (mRaw->getDataType() != RawImageType::UINT16) ------------------ | Branch (68:7): [True: 0, False: 22] ------------------ 69| 22| ThrowRDE("Unexpected data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 70| | 71| 22| if (mRaw->getCpp() != 1 || mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (71:7): [True: 0, False: 22] | Branch (71:30): [True: 0, False: 22] ------------------ 72| 22| ThrowRDE("Unexpected cpp: %u", mRaw->getCpp()); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 73| | 74| | // FIXME: could be wrong. 75| 22| if (!mRaw->dim.hasPositiveArea() || mRaw->dim.x % 2 != 0 || ------------------ | Branch (75:7): [True: 12, False: 10] | Branch (75:39): [True: 1, False: 9] ------------------ 76| 20| mRaw->dim.y % 2 != 0 || mRaw->dim.x > 4992 || mRaw->dim.y > 6668) { ------------------ | Branch (76:7): [True: 1, False: 8] | Branch (76:31): [True: 5, False: 3] | Branch (76:53): [True: 1, False: 2] ------------------ 77| 20| ThrowRDE("Unexpected image dimensions found: (%d; %d)", mRaw->dim.x, ------------------ | | 38| 20| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 20| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 20| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 20| __VA_ARGS__) | | ------------------ ------------------ 78| 20| mRaw->dim.y); 79| 20| } 80| | 81| 2| uint32_t off = raw->getEntry(TiffTag::TILEOFFSETS)->getU32(); 82| 2| uint32_t count = raw->getEntry(TiffTag::TILEBYTECOUNTS)->getU32(); 83| | 84| 2| const ByteStream bs( 85| 2| DataBuffer(mFile.getSubView(off, count), Endianness::little)); 86| | 87| 2| UncompressedDecompressor u(bs, mRaw, iRectangle2D({0, 0}, mRaw->dim), 88| 2| 2 * mRaw->dim.x, 16, BitOrder::MSB); 89| 2| mRaw->createData(); 90| 2| u.readUncompressedRaw(); 91| 2|} _ZN8rawspeed10StiDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 38| 26| : AbstractTiffDecoder(std::move(root), file) {} _ZN8rawspeed14ThreefrDecoder20isAppropriateDecoderEPKNS_11TiffRootIFDENS_6BufferE: 47| 663| [[maybe_unused]] Buffer file) { 48| 663| const auto id = rootIFD->getID(); 49| 663| const std::string& make = id.make; 50| | 51| | // FIXME: magic 52| | 53| 663| return make == "Hasselblad"; 54| 663|} _ZN8rawspeed14ThreefrDecoder17decodeRawInternalEv: 56| 230|RawImage ThreefrDecoder::decodeRawInternal() { 57| 230| const auto* raw = mRootIFD->getIFDWithTag(TiffTag::STRIPOFFSETS, 1); 58| 230| uint32_t width = raw->getEntry(TiffTag::IMAGEWIDTH)->getU32(); 59| 230| uint32_t height = raw->getEntry(TiffTag::IMAGELENGTH)->getU32(); 60| 230| uint32_t compression = raw->getEntry(TiffTag::COMPRESSION)->getU32(); 61| | 62| 230| mRaw->dim = iPoint2D(width, height); 63| | 64| 230| if (1 == compression) { ------------------ | Branch (64:7): [True: 48, False: 182] ------------------ 65| 48| DecodeUncompressed(raw); 66| 48| return mRaw; 67| 48| } 68| | 69| 182| if (compression != 7) // LJpeg ------------------ | Branch (69:7): [True: 4, False: 178] ------------------ 70| 182| ThrowRDE("Unexpected compression type."); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 71| | 72| 178| uint32_t off = raw->getEntry(TiffTag::STRIPOFFSETS)->getU32(); 73| | // STRIPBYTECOUNTS is strange/invalid for the existing (compressed?) 3FR 74| | // samples... 75| | 76| 178| const ByteStream bs(DataBuffer(mFile.getSubView(off), Endianness::little)); 77| | 78| 178| HasselbladLJpegDecoder l(bs, mRaw); 79| 178| mRaw->createData(); 80| | 81| 178| l.decode(); 82| | 83| 178| return mRaw; 84| 182|} _ZNK8rawspeed14ThreefrDecoder18DecodeUncompressedEPKNS_7TiffIFDE: 86| 48|void ThreefrDecoder::DecodeUncompressed(const TiffIFD* raw) const { 87| 48| if (mRaw->getDataType() != RawImageType::UINT16) ------------------ | Branch (87:7): [True: 0, False: 48] ------------------ 88| 48| ThrowRDE("Unexpected data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 89| | 90| 48| if (mRaw->getCpp() != 1 || mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (90:7): [True: 0, False: 48] | Branch (90:30): [True: 0, False: 48] ------------------ 91| 48| ThrowRDE("Unexpected cpp: %u", mRaw->getCpp()); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 92| | 93| | // FIXME: could be wrong. max "active pixels" - "100 MP" 94| 48| if (!mRaw->dim.hasPositiveArea() || mRaw->dim.x % 2 != 0 || ------------------ | Branch (94:7): [True: 25, False: 23] | Branch (94:39): [True: 1, False: 22] ------------------ 95| 36| mRaw->dim.x > 12000 || mRaw->dim.y > 8842) { ------------------ | Branch (95:7): [True: 1, False: 21] | Branch (95:30): [True: 9, False: 12] ------------------ 96| 36| ThrowRDE("Unexpected image dimensions found: (%d; %d)", mRaw->dim.x, ------------------ | | 38| 36| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 36| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 36| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 36| __VA_ARGS__) | | ------------------ ------------------ 97| 36| mRaw->dim.y); 98| 36| } 99| | 100| 12| uint32_t off = raw->getEntry(TiffTag::STRIPOFFSETS)->getU32(); 101| | // STRIPBYTECOUNTS looks valid for the existing uncompressed 3FR samples 102| 12| uint32_t count = raw->getEntry(TiffTag::STRIPBYTECOUNTS)->getU32(); 103| | 104| 12| const ByteStream bs( 105| 12| DataBuffer(mFile.getSubView(off, count), Endianness::little)); 106| | 107| 12| UncompressedDecompressor u(bs, mRaw, iRectangle2D({0, 0}, mRaw->dim), 108| 12| 2 * mRaw->dim.x, 16, BitOrder::LSB); 109| 12| mRaw->createData(); 110| 12| u.readUncompressedRaw(); 111| 12|} _ZN8rawspeed14ThreefrDecoder22decodeMetaDataInternalEPKNS_14CameraMetaDataE: 113| 8|void ThreefrDecoder::decodeMetaDataInternal(const CameraMetaData* meta) { 114| 8| mRaw->cfa.setCFA(iPoint2D(2, 2), CFAColor::RED, CFAColor::GREEN, 115| 8| CFAColor::GREEN, CFAColor::BLUE); 116| | 117| 8| setMetaData(meta, "", 0); 118| | 119| 8| if (mRootIFD->hasEntryRecursive(TiffTag::BLACKLEVEL)) { ------------------ | Branch (119:7): [True: 2, False: 6] ------------------ 120| 2| const TiffEntry* bl = mRootIFD->getEntryRecursive(TiffTag::BLACKLEVEL); 121| 2| if (bl->count == 1) ------------------ | Branch (121:9): [True: 1, False: 1] ------------------ 122| 1| mRaw->blackLevel = implicit_cast(bl->getFloat()); 123| 2| } 124| | 125| 8| if (mRootIFD->hasEntryRecursive(TiffTag::WHITELEVEL)) { ------------------ | Branch (125:7): [True: 2, False: 6] ------------------ 126| 2| const TiffEntry* wl = mRootIFD->getEntryRecursive(TiffTag::WHITELEVEL); 127| 2| if (wl->count == 1) ------------------ | Branch (127:9): [True: 1, False: 1] ------------------ 128| 1| mRaw->whitePoint = implicit_cast(wl->getFloat()); 129| 2| } 130| | 131| | // Fetch the white balance 132| 8| if (mRootIFD->hasEntryRecursive(TiffTag::ASSHOTNEUTRAL)) { ------------------ | Branch (132:7): [True: 5, False: 3] ------------------ 133| 5| const TiffEntry* wb = mRootIFD->getEntryRecursive(TiffTag::ASSHOTNEUTRAL); 134| 5| if (wb->count == 3) { ------------------ | Branch (134:9): [True: 5, False: 0] ------------------ 135| 5| std::array wbCoeffs = {}; 136| 12| for (uint32_t i = 0; i < 3; i++) { ------------------ | Branch (136:28): [True: 11, False: 1] ------------------ 137| 11| const float div = wb->getFloat(i); 138| 11| if (div == 0.0F) ------------------ | Branch (138:13): [True: 4, False: 7] ------------------ 139| 11| ThrowRDE("Can not decode WB, multiplier is zero/"); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 140| | 141| 7| wbCoeffs[i] = 1.0F / div; 142| 7| } 143| 1| mRaw->metadata.wbCoeffs = wbCoeffs; 144| 1| } 145| 5| } 146| 8|} _ZN8rawspeed14ThreefrDecoderC2EONSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 39| 230| : AbstractTiffDecoder(std::move(root), file) {} _ZNK8rawspeed23AbstractDngDecompressor16decompressThreadILi1EEEvv: 54| 29.6k|template <> void AbstractDngDecompressor::decompressThread<1>() const noexcept { 55| 29.6k|#ifdef HAVE_OPENMP 56| 29.6k|#pragma omp for schedule(static) 57| 29.6k|#endif 58| 29.6k| for (const auto& e : 59| 29.6k| Array1DRef(slices.data(), implicit_cast(slices.size()))) { 60| 29.6k| try { 61| 29.6k| iPoint2D tileSize(e.width, e.height); 62| 29.6k| iPoint2D pos(e.offX, e.offY); 63| | 64| 29.6k| bool big_endian = e.bs.getByteOrder() == Endianness::big; 65| | 66| | // DNG spec says that if not 8/16/32 bit/sample, always use big endian. 67| | // It's not very obvious, but that does not appear to apply to FP. 68| 29.6k| switch (mBps) { 69| 29.6k| case 8: 70| 29.6k| case 16: 71| 29.6k| case 32: 72| 29.6k| break; 73| 29.6k| default: 74| 29.6k| if (mRaw->getDataType() == RawImageType::UINT16) 75| 29.6k| big_endian = true; 76| 29.6k| break; 77| 29.6k| } 78| | 79| 29.6k| const uint32_t inputPixelBits = mRaw->getCpp() * mBps; 80| | 81| 29.6k| if (e.dsc.tileW > std::numeric_limits::max() / inputPixelBits) 82| 29.6k| ThrowIOE("Integer overflow when calculating input pitch"); 83| | 84| 29.6k| const int inputPitchBits = inputPixelBits * e.dsc.tileW; 85| 29.6k| invariant(inputPitchBits > 0); 86| | 87| 29.6k| if (inputPitchBits % 8 != 0) { 88| 29.6k| ThrowRDE("Bad combination of cpp (%u), bps (%u) and width (%u), the " 89| 29.6k| "pitch is %d bits, which is not a multiple of 8 (1 byte)", 90| 29.6k| mRaw->getCpp(), mBps, e.width, inputPitchBits); 91| 29.6k| } 92| | 93| 29.6k| const int inputPitch = inputPitchBits / 8; 94| 29.6k| if (inputPitch == 0) 95| 29.6k| ThrowRDE("Data input pitch is too short. Can not decode!"); 96| | 97| 29.6k| UncompressedDecompressor decompressor( 98| 29.6k| e.bs, mRaw, iRectangle2D(pos, tileSize), inputPitch, mBps, 99| 29.6k| big_endian ? BitOrder::MSB : BitOrder::LSB); 100| 29.6k| decompressor.readUncompressedRaw(); 101| 29.6k| } catch (const RawDecoderException& err) { 102| 29.6k| mRaw->setError(err.what()); 103| 29.6k| } catch (const IOException& err) { 104| 29.6k| mRaw->setError(err.what()); 105| 29.6k| } catch (...) { 106| | // We should not get any other exception type here. 107| 29.6k| __builtin_unreachable(); 108| 29.6k| } 109| 29.6k| } 110| 29.6k|} _ZNK8rawspeed23AbstractDngDecompressor16decompressThreadILi7EEEvv: 112| 11.9k|template <> void AbstractDngDecompressor::decompressThread<7>() const noexcept { 113| 11.9k|#ifdef HAVE_OPENMP 114| 11.9k|#pragma omp for schedule(static) 115| 11.9k|#endif 116| 11.9k| for (const auto& e : 117| 11.9k| Array1DRef(slices.data(), implicit_cast(slices.size()))) { 118| 11.9k| try { 119| 11.9k| LJpegDecoder d(e.bs, mRaw); 120| 11.9k| d.decode(e.offX, e.offY, e.width, e.height, 121| 11.9k| iPoint2D(e.dsc.tileW, e.dsc.tileH), mFixLjpeg); 122| 11.9k| } catch (const RawDecoderException& err) { 123| 11.9k| mRaw->setError(err.what()); 124| 11.9k| } catch (const IOException& err) { 125| 11.9k| mRaw->setError(err.what()); 126| 11.9k| } catch (...) { 127| | // We should not get any other exception type here. 128| 11.9k| __builtin_unreachable(); 129| 11.9k| } 130| 11.9k| } 131| 11.9k|} _ZNK8rawspeed23AbstractDngDecompressor16decompressThreadILi9EEEvv: 161| 9.27k|template <> void AbstractDngDecompressor::decompressThread<9>() const noexcept { 162| 9.27k|#ifdef HAVE_OPENMP 163| 9.27k|#pragma omp for schedule(static) 164| 9.27k|#endif 165| 9.27k| for (const auto& e : 166| 9.27k| Array1DRef(slices.data(), implicit_cast(slices.size()))) { 167| 9.27k| try { 168| 9.27k| VC5Decompressor d(e.bs, mRaw); 169| 9.27k| d.decode(e.offX, e.offY, e.width, e.height); 170| 9.27k| } catch (const RawDecoderException& err) { 171| 9.27k| mRaw->setError(err.what()); 172| 9.27k| } catch (const IOException& err) { 173| 9.27k| mRaw->setError(err.what()); 174| 9.27k| } catch (...) { 175| | // We should not get any other exception type here. 176| 9.27k| __builtin_unreachable(); 177| 9.27k| } 178| 9.27k| } 179| 9.27k|} _ZNK8rawspeed23AbstractDngDecompressor16decompressThreadEv: 204| 14.1k|void AbstractDngDecompressor::decompressThread() const noexcept { 205| 14.1k| invariant(mRaw->dim.x > 0); ------------------ | | 27| 14.1k|#define invariant(expr) assert(expr) ------------------ | Branch (205:3): [True: 14.1k, False: 31] ------------------ 206| 14.1k| invariant(mRaw->dim.y > 0); ------------------ | | 27| 14.1k|#define invariant(expr) assert(expr) ------------------ | Branch (206:3): [True: 13.8k, False: 233] ------------------ 207| 13.8k| invariant(mRaw->getCpp() > 0 && mRaw->getCpp() <= 4); ------------------ | | 27| 13.8k|#define invariant(expr) assert(expr) ------------------ | Branch (207:3): [True: 13.7k, False: 98] | Branch (207:3): [True: 13.6k, False: 134] | Branch (207:3): [True: 13.6k, False: 225] ------------------ 208| 13.6k| invariant(mBps > 0 && mBps <= 32); ------------------ | | 27| 13.6k|#define invariant(expr) assert(expr) ------------------ | Branch (208:3): [True: 13.6k, False: 18.4E] | Branch (208:3): [True: 13.6k, False: 18.4E] | Branch (208:3): [True: 13.6k, False: 18.4E] ------------------ 209| | 210| 13.6k| if (compression == 1) { ------------------ | Branch (210:7): [True: 5.34k, False: 8.31k] ------------------ 211| | /* Uncompressed */ 212| 5.34k| decompressThread<1>(); 213| 8.31k| } else if (compression == 7) { ------------------ | Branch (213:14): [True: 4.55k, False: 3.76k] ------------------ 214| | /* Lossless JPEG */ 215| 4.55k| decompressThread<7>(); 216| 4.55k| } else if (compression == 8) { ------------------ | Branch (216:14): [True: 0, False: 3.76k] ------------------ 217| | /* Deflate compression */ 218| |#ifdef HAVE_ZLIB 219| | decompressThread<8>(); 220| |#else 221| 0|#pragma message \ 222| 0| "ZLIB is not present! Deflate compression will not be supported!" 223| 0| mRaw->setError("deflate support is disabled."); 224| 0|#endif 225| 3.76k| } else if (compression == 9) { ------------------ | Branch (225:14): [True: 3.76k, False: 5] ------------------ 226| | /* GOPRO VC-5 */ 227| 3.76k| decompressThread<9>(); 228| 3.76k| } else if (compression == 0x884c) { ------------------ | Branch (228:14): [True: 0, False: 5] ------------------ 229| | /* Lossy DNG */ 230| |#ifdef HAVE_JPEG 231| | decompressThread<0x884c>(); 232| |#else 233| 0|#pragma message "JPEG is not present! Lossy JPEG DNG will not be supported!" 234| 0| mRaw->setError("jpeg support is disabled."); 235| 0|#endif 236| 5| } else { 237| 5| mRaw->setError("AbstractDngDecompressor: Unknown compression"); 238| 5| } 239| 13.6k|} _ZNK8rawspeed23AbstractDngDecompressor10decompressEv: 241| 3.36k|void AbstractDngDecompressor::decompress() const { 242| 3.36k|#ifdef HAVE_OPENMP 243| 3.36k|#pragma omp parallel default(none) num_threads( \ 244| 3.36k| rawspeed_get_number_of_processor_cores()) if (slices.size() > 1) 245| 3.36k|#endif 246| 3.36k| decompressThread(); 247| | 248| 3.36k| std::string firstErr; 249| 3.36k| if (mRaw->isTooManyErrors(1, &firstErr)) { ------------------ | Branch (249:7): [True: 824, False: 2.54k] ------------------ 250| 824| ThrowRDE("Too many errors encountered. Giving up. First Error:\n%s", ------------------ | | 38| 824| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 824| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 824| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 824| __VA_ARGS__) | | ------------------ ------------------ 251| 824| firstErr.c_str()); 252| 824| } 253| 3.36k|} AbstractDngDecompressor.cpp:_ZNK8rawspeed23AbstractDngDecompressor10decompressEv.omp_outlined_debug__: 246| 14.1k| decompressThread(); _ZN8rawspeed20DngTilingDescriptionC2ERKNS_8iPoint2DEjj: 57| 3.41k| : dim(dim_), tileW(tileW_), tileH(tileH_), 58| 3.41k| tilesX(implicit_cast(roundUpDivisionSafe(dim.x, tileW))), 59| 3.41k| tilesY(implicit_cast(roundUpDivisionSafe(dim.y, tileH))), 60| 3.41k| numTiles(tilesX * tilesY) { 61| 3.41k| invariant(dim.area() > 0); ------------------ | | 27| 3.41k|#define invariant(expr) assert(expr) ------------------ | Branch (61:5): [True: 3.41k, False: 0] ------------------ 62| 3.41k| invariant(tileW > 0); ------------------ | | 27| 3.41k|#define invariant(expr) assert(expr) ------------------ | Branch (62:5): [True: 3.41k, False: 0] ------------------ 63| 3.41k| invariant(tileH > 0); ------------------ | | 27| 3.41k|#define invariant(expr) assert(expr) ------------------ | Branch (63:5): [True: 3.41k, False: 0] ------------------ 64| 3.41k| invariant(tilesX > 0); ------------------ | | 27| 3.41k|#define invariant(expr) assert(expr) ------------------ | Branch (64:5): [True: 3.41k, False: 0] ------------------ 65| 3.41k| invariant(tilesY > 0); ------------------ | | 27| 3.41k|#define invariant(expr) assert(expr) ------------------ | Branch (65:5): [True: 3.41k, False: 0] ------------------ 66| 3.41k| invariant(tileW * tilesX >= static_cast(dim.x)); ------------------ | | 27| 3.41k|#define invariant(expr) assert(expr) ------------------ | Branch (66:5): [True: 3.41k, False: 0] ------------------ 67| 3.41k| invariant(tileH * tilesY >= static_cast(dim.y)); ------------------ | | 27| 3.41k|#define invariant(expr) assert(expr) ------------------ | Branch (67:5): [True: 3.41k, False: 0] ------------------ 68| 3.41k| invariant(tileW * (tilesX - 1) < static_cast(dim.x)); ------------------ | | 27| 3.41k|#define invariant(expr) assert(expr) ------------------ | Branch (68:5): [True: 3.41k, False: 0] ------------------ 69| 3.41k| invariant(tileH * (tilesY - 1) < static_cast(dim.y)); ------------------ | | 27| 3.41k|#define invariant(expr) assert(expr) ------------------ | Branch (69:5): [True: 3.41k, False: 0] ------------------ 70| 3.41k| invariant(numTiles > 0); ------------------ | | 27| 3.41k|#define invariant(expr) assert(expr) ------------------ | Branch (70:5): [True: 3.41k, False: 0] ------------------ 71| 3.41k| } _ZN8rawspeed23AbstractDngDecompressorC2ENS_8RawImageERKNS_20DngTilingDescriptionEibjj: 138| 3.41k| : mRaw(std::move(img)), dsc(dsc_), compression(compression_), 139| 3.41k| mFixLjpeg(mFixLjpeg_), mBps(mBps_), mPredictor(mPredictor_) {} _ZN8rawspeed15DngSliceElementC2ERKNS_20DngTilingDescriptionEjNS_10ByteStreamE: 105| 4.52k| : dsc(dsc_), n(n_), bs(bs_), column(n % dsc.tilesX), row(n / dsc.tilesX), 106| 4.52k| lastColumn((column + 1) == dsc.tilesX), 107| 4.52k| lastRow((row + 1) == dsc.tilesY), offX(dsc.tileW * column), 108| 4.52k| offY(dsc.tileH * row), 109| 4.52k| width(!lastColumn ? dsc.tileW : dsc.dim.x - offX), ------------------ | Branch (109:15): [True: 238, False: 4.28k] ------------------ 110| 4.52k| height(!lastRow ? dsc.tileH : dsc.dim.y - offY) { ------------------ | Branch (110:16): [True: 1.03k, False: 3.49k] ------------------ 111| 4.52k| invariant(n < dsc.numTiles); ------------------ | | 27| 4.52k|#define invariant(expr) assert(expr) ------------------ | Branch (111:5): [True: 4.52k, False: 0] ------------------ 112| 4.52k| invariant(bs.getRemainSize() > 0); ------------------ | | 27| 4.52k|#define invariant(expr) assert(expr) ------------------ | Branch (112:5): [True: 4.52k, False: 0] ------------------ 113| 4.52k| invariant(column < dsc.tilesX); ------------------ | | 27| 4.52k|#define invariant(expr) assert(expr) ------------------ | Branch (113:5): [True: 4.52k, False: 0] ------------------ 114| 4.52k| invariant(row < dsc.tilesY); ------------------ | | 27| 4.52k|#define invariant(expr) assert(expr) ------------------ | Branch (114:5): [True: 4.52k, False: 0] ------------------ 115| 4.52k| invariant(offX < static_cast(dsc.dim.x)); ------------------ | | 27| 4.52k|#define invariant(expr) assert(expr) ------------------ | Branch (115:5): [True: 4.52k, False: 0] ------------------ 116| 4.52k| invariant(offY < static_cast(dsc.dim.y)); ------------------ | | 27| 4.52k|#define invariant(expr) assert(expr) ------------------ | Branch (116:5): [True: 4.52k, False: 0] ------------------ 117| 4.52k| invariant(width > 0); ------------------ | | 27| 4.52k|#define invariant(expr) assert(expr) ------------------ | Branch (117:5): [True: 4.52k, False: 0] ------------------ 118| 4.52k| invariant(height > 0); ------------------ | | 27| 4.52k|#define invariant(expr) assert(expr) ------------------ | Branch (118:5): [True: 4.52k, False: 0] ------------------ 119| 4.52k| invariant(offX + width <= static_cast(dsc.dim.x)); ------------------ | | 27| 4.52k|#define invariant(expr) assert(expr) ------------------ | Branch (119:5): [True: 4.52k, False: 0] ------------------ 120| 4.52k| invariant(offY + height <= static_cast(dsc.dim.y)); ------------------ | | 27| 4.52k|#define invariant(expr) assert(expr) ------------------ | Branch (120:5): [True: 4.52k, False: 0] ------------------ 121| 4.52k| invariant(!lastColumn || ------------------ | | 27| 4.52k|#define invariant(expr) assert(expr) ------------------ | Branch (121:5): [True: 238, False: 4.28k] | Branch (121:5): [True: 4.28k, False: 0] | Branch (121:5): [True: 4.52k, False: 0] ------------------ 122| 4.52k| (offX + width == static_cast(dsc.dim.x))); 123| 4.52k| invariant(!lastRow || (offY + height == static_cast(dsc.dim.y))); ------------------ | | 27| 4.52k|#define invariant(expr) assert(expr) ------------------ | Branch (123:5): [True: 1.03k, False: 3.49k] | Branch (123:5): [True: 3.49k, False: 0] | Branch (123:5): [True: 4.52k, False: 0] ------------------ 124| 4.52k| } _ZN8rawspeed20AbstractLJpegDecoderC2ENS_10ByteStreamENS_8RawImageE: 50| 2.08k| : input(bs), mRaw(std::move(img)) { 51| 2.08k| input.setByteOrder(Endianness::big); 52| | 53| 2.08k| if (!mRaw->dim.hasPositiveArea()) ------------------ | Branch (53:7): [True: 29, False: 2.05k] ------------------ 54| 2.08k| ThrowRDE("Image has zero size"); ------------------ | | 38| 29| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 29| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 29| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 29| __VA_ARGS__) | | ------------------ ------------------ 55| | 56| 2.05k|#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION 57| | // Yeah, sure, here it would be just dumb to leave this for production :) 58| 2.05k| if (mRaw->dim.x > 19440 || mRaw->dim.y > 8842) { ------------------ | Branch (58:7): [True: 13, False: 2.04k] | Branch (58:30): [True: 20, False: 2.02k] ------------------ 59| 23| ThrowRDE("Unexpected image dimensions found: (%i; %i)", mRaw->dim.x, ------------------ | | 38| 23| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 23| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 23| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 23| __VA_ARGS__) | | ------------------ ------------------ 60| 23| mRaw->dim.y); 61| 23| } 62| 2.05k|#endif 63| 2.05k|} _ZN8rawspeed20AbstractLJpegDecoder9decodeSOIEv: 65| 1.93k|void AbstractLJpegDecoder::decodeSOI() { 66| 1.93k| if (getNextMarker(false) != JpegMarker::SOI) ------------------ | Branch (66:7): [True: 27, False: 1.90k] ------------------ 67| 1.93k| ThrowRDE("Image did not start with SOI. Probably not an LJPEG"); ------------------ | | 38| 27| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 27| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 27| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 27| __VA_ARGS__) | | ------------------ ------------------ 68| | 69| 1.90k| struct { 70| 1.90k| bool DRI = false; 71| 1.90k| bool DHT = false; 72| 1.90k| bool SOF = false; 73| 1.90k| bool SOS = false; 74| 1.90k| } FoundMarkers; 75| | 76| 5.80k| for (JpegMarker m; (m = getNextMarker(true)) != JpegMarker::EOI;) { ------------------ | Branch (76:22): [True: 5.25k, False: 557] ------------------ 77| 5.25k| ByteStream data(input.getStream(input.peekU16())); 78| 5.25k| data.skipBytes(2); // headerLength 79| | 80| 5.25k| switch (m) { 81| 2.12k| case JpegMarker::DHT: ------------------ | Branch (81:5): [True: 2.12k, False: 3.12k] ------------------ 82| 2.12k| if (FoundMarkers.SOS) ------------------ | Branch (82:11): [True: 3, False: 2.12k] ------------------ 83| 2.12k| ThrowRDE("Found second DHT marker after SOS"); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 84| | // there can be more than one DHT markers. 85| | // FIXME: do we really want to reparse and use the last one? 86| 2.12k| parseDHT(data); 87| 2.12k| FoundMarkers.DHT = true; 88| 2.12k| break; 89| 1.26k| case JpegMarker::SOF3: ------------------ | Branch (89:5): [True: 1.26k, False: 3.98k] ------------------ 90| 1.26k| if (FoundMarkers.SOS) ------------------ | Branch (90:11): [True: 6, False: 1.26k] ------------------ 91| 1.26k| ThrowRDE("Found second SOF marker after SOS"); ------------------ | | 38| 6| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 6| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 6| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 6| __VA_ARGS__) | | ------------------ ------------------ 92| 1.26k| if (FoundMarkers.SOF) ------------------ | Branch (92:11): [True: 1, False: 1.26k] ------------------ 93| 1.26k| ThrowRDE("Found second SOF marker"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 94| | // SOF is not required to be after DHT 95| 1.26k| parseSOF(data, &frame); 96| 1.26k| FoundMarkers.SOF = true; 97| 1.26k| break; 98| 1.15k| case JpegMarker::SOS: ------------------ | Branch (98:5): [True: 1.15k, False: 4.09k] ------------------ 99| 1.15k| if (FoundMarkers.SOS) ------------------ | Branch (99:11): [True: 1, False: 1.15k] ------------------ 100| 1.15k| ThrowRDE("Found second SOS marker"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 101| 1.15k| if (!FoundMarkers.DHT) ------------------ | Branch (101:11): [True: 1, False: 1.15k] ------------------ 102| 1.15k| ThrowRDE("Did not find DHT marker before SOS."); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 103| 1.15k| if (!FoundMarkers.SOF) ------------------ | Branch (103:11): [True: 4, False: 1.15k] ------------------ 104| 1.15k| ThrowRDE("Did not find SOF marker before SOS."); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 105| 1.15k| parseSOS(data); 106| 1.15k| FoundMarkers.SOS = true; 107| 1.15k| if (erratumImplicitEOIMarkerAfterScan()) ------------------ | Branch (107:11): [True: 0, False: 1.15k] ------------------ 108| 0| return; 109| 1.15k| break; 110| 1.15k| case JpegMarker::DQT: ------------------ | Branch (110:5): [True: 2, False: 5.25k] ------------------ 111| 2| ThrowRDE("Not a valid RAW file."); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 112| 22| case JpegMarker::DRI: ------------------ | Branch (112:5): [True: 22, False: 5.23k] ------------------ 113| 22| if (FoundMarkers.DRI) ------------------ | Branch (113:11): [True: 1, False: 21] ------------------ 114| 22| ThrowRDE("Found second DRI marker"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 115| 21| parseDRI(data); 116| 21| FoundMarkers.DRI = true; 117| 21| break; 118| 605| default: // Just let it skip to next marker ------------------ | Branch (118:5): [True: 605, False: 4.64k] ------------------ 119| 605| break; 120| 5.25k| } 121| 5.25k| } 122| | 123| 557| if (!FoundMarkers.SOS) ------------------ | Branch (123:7): [True: 7, False: 550] ------------------ 124| 557| ThrowRDE("Did not find SOS marker."); ------------------ | | 38| 7| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 7| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 7| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 7| __VA_ARGS__) | | ------------------ ------------------ 125| 557|} _ZN8rawspeed20AbstractLJpegDecoder8parseSOFENS_10ByteStreamEPNS_7SOFInfoE: 127| 1.26k|void AbstractLJpegDecoder::parseSOF(ByteStream sofInput, SOFInfo* sof) { 128| 1.26k| sof->prec = sofInput.getByte(); 129| 1.26k| sof->h = sofInput.getU16(); 130| 1.26k| sof->w = sofInput.getU16(); 131| 1.26k| sof->cps = sofInput.getByte(); 132| | 133| 1.26k| if (sof->prec < 2 || sof->prec > 16) ------------------ | Branch (133:7): [True: 3, False: 1.25k] | Branch (133:24): [True: 4, False: 1.25k] ------------------ 134| 1.26k| ThrowRDE("Invalid precision (%u).", sof->prec); ------------------ | | 38| 5| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 5| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 5| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 5| __VA_ARGS__) | | ------------------ ------------------ 135| | 136| 1.25k| if (sof->h == 0 || sof->w == 0) ------------------ | Branch (136:7): [True: 3, False: 1.25k] | Branch (136:22): [True: 1, False: 1.25k] ------------------ 137| 1.25k| ThrowRDE("Frame width or height set to zero"); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 138| | 139| 1.25k| if (sof->cps > 4 || sof->cps < 1) ------------------ | Branch (139:7): [True: 8, False: 1.24k] | Branch (139:23): [True: 1, False: 1.24k] ------------------ 140| 1.25k| ThrowRDE("Only from 1 to 4 components are supported."); ------------------ | | 38| 7| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 7| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 7| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 7| __VA_ARGS__) | | ------------------ ------------------ 141| | 142| 1.24k| if (sof->cps < mRaw->getCpp()) { ------------------ | Branch (142:7): [True: 1, False: 1.24k] ------------------ 143| 1| ThrowRDE("Component count should be no less than sample count (%u vs %u).", ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 144| 1| sof->cps, mRaw->getCpp()); 145| 1| } 146| | 147| 1.24k| if (sof->cps > static_cast(mRaw->dim.x)) { ------------------ | Branch (147:7): [True: 1, False: 1.24k] ------------------ 148| 1| ThrowRDE("Component count should be no greater than row length (%u vs %d).", ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 149| 1| sof->cps, mRaw->dim.x); 150| 1| } 151| | 152| 1.24k| if (sofInput.getRemainSize() != 3 * sof->cps) ------------------ | Branch (152:7): [True: 8, False: 1.23k] ------------------ 153| 1.24k| ThrowRDE("Header size mismatch."); ------------------ | | 38| 8| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 8| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 8| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 8| __VA_ARGS__) | | ------------------ ------------------ 154| | 155| 4.48k| for (uint32_t i = 0; i < sof->cps; i++) { ------------------ | Branch (155:24): [True: 3.26k, False: 1.21k] ------------------ 156| 3.26k| sof->compInfo[i].componentId = sofInput.getByte(); 157| | 158| 3.26k| uint32_t subs = sofInput.getByte(); 159| 3.26k| frame.compInfo[i].superV = subs & 0xf; 160| 3.26k| frame.compInfo[i].superH = subs >> 4; 161| | 162| 3.26k| if (frame.compInfo[i].superV < 1 || frame.compInfo[i].superV > 4) ------------------ | Branch (162:9): [True: 4, False: 3.26k] | Branch (162:41): [True: 8, False: 3.25k] ------------------ 163| 3.26k| ThrowRDE("Horizontal sampling factor is invalid."); ------------------ | | 38| 10| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 10| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 10| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 10| __VA_ARGS__) | | ------------------ ------------------ 164| | 165| 3.25k| if (frame.compInfo[i].superH < 1 || frame.compInfo[i].superH > 4) ------------------ | Branch (165:9): [True: 3, False: 3.25k] | Branch (165:41): [True: 4, False: 3.25k] ------------------ 166| 3.25k| ThrowRDE("Horizontal sampling factor is invalid."); ------------------ | | 38| 5| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 5| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 5| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 5| __VA_ARGS__) | | ------------------ ------------------ 167| | 168| 3.25k| uint32_t Tq = sofInput.getByte(); 169| 3.25k| if (Tq != 0) ------------------ | Branch (169:9): [True: 8, False: 3.24k] ------------------ 170| 3.25k| ThrowRDE("Quantized components not supported."); ------------------ | | 38| 8| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 8| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 8| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 8| __VA_ARGS__) | | ------------------ ------------------ 171| 3.25k| } 172| | 173| 1.21k| if (static_cast(sof->compInfo[0].superH) != ------------------ | Branch (173:7): [True: 4, False: 1.21k] ------------------ 174| 1.21k| mRaw->metadata.subsampling.x || 175| 1.21k| static_cast(sof->compInfo[0].superV) != mRaw->metadata.subsampling.y) ------------------ | Branch (175:7): [True: 2, False: 1.20k] ------------------ 176| 1.21k| ThrowRDE("LJpeg's subsampling does not match image's subsampling."); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 177| | 178| 1.21k| sof->initialized = true; 179| 1.21k|} _ZN8rawspeed20AbstractLJpegDecoder8parseSOSENS_10ByteStreamE: 181| 1.15k|void AbstractLJpegDecoder::parseSOS(ByteStream sos) { 182| 1.15k| invariant(frame.initialized); ------------------ | | 27| 1.15k|#define invariant(expr) assert(expr) ------------------ | Branch (182:3): [True: 1.15k, False: 0] ------------------ 183| | 184| 1.15k| if (sos.getRemainSize() != 1 + 2 * frame.cps + 3) ------------------ | Branch (184:7): [True: 15, False: 1.13k] ------------------ 185| 1.15k| ThrowRDE("Invalid SOS header length."); ------------------ | | 38| 15| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 15| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 15| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 15| __VA_ARGS__) | | ------------------ ------------------ 186| | 187| 1.13k| if (uint32_t soscps = sos.getByte(); frame.cps != soscps) ------------------ | Branch (187:40): [True: 2, False: 1.13k] ------------------ 188| 1.13k| ThrowRDE("Component number mismatch."); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 189| | 190| 4.07k| for (uint32_t i = 0; i < frame.cps; i++) { ------------------ | Branch (190:24): [True: 2.96k, False: 1.10k] ------------------ 191| 2.96k| uint32_t cs = sos.getByte(); 192| 2.96k| uint32_t td = sos.getByte() >> 4; 193| | 194| 2.96k| if (td >= huff.size() || !huff[td]) ------------------ | Branch (194:9): [True: 2, False: 2.96k] | Branch (194:30): [True: 2, False: 2.95k] ------------------ 195| 2.96k| ThrowRDE("Invalid Huffman table selection."); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 196| | 197| 2.95k| int ciIndex = -1; 198| 12.0k| for (uint32_t j = 0; j < frame.cps; ++j) { ------------------ | Branch (198:26): [True: 9.12k, False: 2.95k] ------------------ 199| 9.12k| if (frame.compInfo[j].componentId == cs) ------------------ | Branch (199:11): [True: 2.94k, False: 6.17k] ------------------ 200| 2.94k| ciIndex = j; 201| 9.12k| } 202| | 203| 2.95k| if (ciIndex == -1) ------------------ | Branch (203:9): [True: 20, False: 2.93k] ------------------ 204| 2.95k| ThrowRDE("Invalid Component Selector"); ------------------ | | 38| 20| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 20| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 20| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 20| __VA_ARGS__) | | ------------------ ------------------ 205| | 206| 2.93k| frame.compInfo[ciIndex].dcTblNo = td; 207| 2.93k| } 208| | 209| | // Get predictor, see table H.1 from the JPEG spec 210| 1.10k| predictorMode = sos.getByte(); 211| | // The spec says predictoreMode is in [0..7], but Hasselblad uses '8'. 212| 1.10k| if (predictorMode > 8) ------------------ | Branch (212:7): [True: 1, False: 1.10k] ------------------ 213| 1.10k| ThrowRDE("Invalid predictor mode."); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 214| | 215| | // Se + Ah Not used in LJPEG 216| 1.10k| if (sos.getByte() != 0) ------------------ | Branch (216:7): [True: 1, False: 1.10k] ------------------ 217| 1.10k| ThrowRDE("Se/Ah not zero."); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 218| | 219| 1.10k| Pt = sos.getByte(); // Point Transform 220| 1.10k| if (Pt > 15) ------------------ | Branch (220:7): [True: 2, False: 1.10k] ------------------ 221| 1.10k| ThrowRDE("Invalid Point transform."); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 222| 1.10k| if (Pt != 0) ------------------ | Branch (222:7): [True: 2, False: 1.10k] ------------------ 223| 1.10k| ThrowRDE("Point transform not supported."); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 224| | 225| 1.10k| const auto scanLength = decodeScan(); 226| | // FIXME: invariant(scanLength != 0); 227| 1.10k| input.skipBytes(scanLength); 228| 1.10k|} _ZN8rawspeed20AbstractLJpegDecoder8parseDHTENS_10ByteStreamE: 230| 2.12k|void AbstractLJpegDecoder::parseDHT(ByteStream dht) { 231| 4.00k| while (dht.getRemainSize() > 0) { ------------------ | Branch (231:10): [True: 2.03k, False: 1.97k] ------------------ 232| 2.03k| uint32_t b = dht.getByte(); 233| | 234| 2.03k| if (uint32_t htClass = b >> 4; htClass != 0) ------------------ | Branch (234:36): [True: 17, False: 2.01k] ------------------ 235| 2.03k| ThrowRDE("Unsupported Table class."); ------------------ | | 38| 17| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 17| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 17| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 17| __VA_ARGS__) | | ------------------ ------------------ 236| | 237| 2.01k| uint32_t htIndex = b & 0xf; 238| 2.01k| if (htIndex >= huff.size()) ------------------ | Branch (238:9): [True: 10, False: 2.00k] ------------------ 239| 2.01k| ThrowRDE("Invalid huffman table destination id."); ------------------ | | 38| 10| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 10| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 10| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 10| __VA_ARGS__) | | ------------------ ------------------ 240| | 241| 2.00k| if (huff[htIndex] != nullptr) ------------------ | Branch (241:9): [True: 41, False: 1.96k] ------------------ 242| 2.00k| ThrowRDE("Duplicate table definition"); ------------------ | | 38| 41| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 41| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 41| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 41| __VA_ARGS__) | | ------------------ ------------------ 243| | 244| | // Temporary table, used during parsing LJpeg. 245| 1.96k| HuffmanCode hc; 246| | 247| | // copy 16 bytes from input stream to number of codes per length table 248| 1.96k| uint32_t nCodes = hc.setNCodesPerLength(dht.getBuffer(16)); 249| | 250| | // spec says 16 different codes is max but Hasselblad violates that -> 17 251| 1.96k| if (nCodes > 17) ------------------ | Branch (251:9): [True: 13, False: 1.95k] ------------------ 252| 1.96k| ThrowRDE("Invalid DHT table."); ------------------ | | 38| 13| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 13| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 13| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 13| __VA_ARGS__) | | ------------------ ------------------ 253| | 254| | // copy nCodes bytes from input stream to code values table 255| 1.95k| const auto codesBuf = dht.getBuffer(nCodes); 256| 1.95k| hc.setCodeValues(codesBuf.getAsArray1DRef()); 257| | 258| | // see if we already have a PrefixCodeDecoder with the same codes 259| 1.95k| assert(PrefixCodeDecoderStore.size() == huffmanCodeStore.size()); ------------------ | Branch (259:5): [True: 1.88k, False: 64] ------------------ 260| 2.63k| for (unsigned index = 0; index != PrefixCodeDecoderStore.size(); ++index) { ------------------ | Branch (260:30): [True: 746, False: 1.88k] ------------------ 261| 746| if (*huffmanCodeStore[index] == hc) ------------------ | Branch (261:11): [True: 122, False: 624] ------------------ 262| 122| huff[htIndex] = PrefixCodeDecoderStore[index].get(); 263| 746| } 264| | 265| 1.88k| if (!huff[htIndex]) { ------------------ | Branch (265:9): [True: 1.76k, False: 124] ------------------ 266| 1.76k| huffmanCodeStore.emplace_back(std::make_unique(hc)); 267| | // setup new hc and put it into the store 268| 1.76k| auto dHT = std::make_unique>(std::move(hc)); 269| 1.76k| dHT->setup(fullDecodeHT, fixDng16Bug); 270| 1.76k| huff[htIndex] = dHT.get(); 271| 1.76k| PrefixCodeDecoderStore.emplace_back(std::move(dHT)); 272| 1.76k| } 273| 1.88k| } 274| 2.12k|} _ZN8rawspeed20AbstractLJpegDecoder8parseDRIENS_10ByteStreamE: 276| 21|void AbstractLJpegDecoder::parseDRI(ByteStream dri) { 277| 21| if (dri.getRemainSize() != 2) ------------------ | Branch (277:7): [True: 3, False: 18] ------------------ 278| 21| ThrowRDE("Invalid DRI header length."); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 279| 18| numMCUsPerRestartInterval = dri.getU16(); 280| 18|} _ZN8rawspeed20AbstractLJpegDecoder13getNextMarkerEb: 282| 7.38k|JpegMarker AbstractLJpegDecoder::getNextMarker(bool allowskip) { 283| 7.38k| if (Optional markerPos = advanceToNextMarker(input, allowskip)) ------------------ | Branch (283:28): [True: 6.88k, False: 496] ------------------ 284| 6.88k| input = *markerPos; 285| 496| else 286| 7.38k| ThrowRDE("(Noskip) Expected marker not found. Probably corrupt file."); ------------------ | | 38| 496| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 496| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 496| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 496| __VA_ARGS__) | | ------------------ ------------------ 287| | 288| 6.88k| JpegMarker m = *peekMarker(input); 289| 6.88k| input.skipBytes(2); // Skip the bytes we've just consumed. 290| 6.88k| return m; 291| 7.38k|} _ZN8rawspeed20AbstractLJpegDecoderD2Ev: 92| 2.03k| virtual ~AbstractLJpegDecoder() = default; _ZNK8rawspeed20AbstractLJpegDecoder18getSamplePrecisionEv: 90| 41| [[nodiscard]] int getSamplePrecision() const { return frame.prec; } _ZNK8rawspeed20AbstractLJpegDecoder33erratumImplicitEOIMarkerAfterScanEv: 101| 126| [[nodiscard]] virtual bool erratumImplicitEOIMarkerAfterScan() const { 102| 126| return false; 103| 126| } _ZNK8rawspeed20AbstractLJpegDecoder21getPrefixCodeDecodersEi: 113| 1.07k| getPrefixCodeDecoders(int N_COMP) const { 114| 1.07k| std::vector*> ht(N_COMP); 115| 3.62k| for (int i = 0; i < N_COMP; ++i) { ------------------ | Branch (115:21): [True: 2.55k, False: 1.07k] ------------------ 116| 2.55k| const unsigned dcTblNo = frame.compInfo[i].dcTblNo; 117| 2.55k| if (const auto dcTbls = implicit_cast(huff.size()); 118| 2.55k| dcTblNo >= dcTbls) { ------------------ | Branch (118:11): [True: 6, False: 2.54k] ------------------ 119| 6| ThrowRDE("Decoding table %u for comp %i does not exist (tables = %u)", ------------------ | | 38| 6| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 6| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 6| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 6| __VA_ARGS__) | | ------------------ ------------------ 120| 6| dcTblNo, i, dcTbls); 121| 6| } 122| 2.54k| ht[i] = huff[dcTblNo]; 123| 2.54k| } 124| | 125| 1.07k| return ht; 126| 1.07k| } _ZNK8rawspeed20AbstractLJpegDecoder20getInitialPredictorsEi: 128| 1.07k| [[nodiscard]] std::vector getInitialPredictors(int N_COMP) const { 129| 1.07k| std::vector pred(N_COMP); 130| 1.07k| if (frame.prec < (Pt + 1)) { ------------------ | Branch (130:9): [True: 0, False: 1.07k] ------------------ 131| 0| ThrowRDE("Invalid precision (%u) and point transform (%u) combination!", ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 132| 0| frame.prec, Pt); 133| 0| } 134| 1.07k| std::fill(pred.begin(), pred.end(), 1 << (frame.prec - Pt - 1)); 135| 1.07k| return pred; 136| 1.07k| } _ZN8rawspeed27AbstractSamsungDecompressorC2ENS_8RawImageE: 34| 460| explicit AbstractSamsungDecompressor(RawImage raw) : mRaw(std::move(raw)) {} _ZN8rawspeed14Cr2SliceWidthsC2Ettt: 65| 948| : numSlices(numSlices_), sliceWidth(sliceWidth_), 66| 948| lastSliceWidth(lastSliceWidth_) { 67| 948| if (numSlices < 1) ------------------ | Branch (67:9): [True: 1, False: 947] ------------------ 68| 948| ThrowRDE("Bad slice count: %d", numSlices); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 69| 948| } _ZN8rawspeed14Cr2SliceWidthsC2Ev: 61| 1.06k| Cr2SliceWidths() = default; _ZNK8rawspeed14Cr2SliceWidths5emptyEv: 71| 592| [[nodiscard]] bool empty() const { 72| 592| return 0 == numSlices && 0 == sliceWidth && 0 == lastSliceWidth; ------------------ | Branch (72:12): [True: 12, False: 580] | Branch (72:30): [True: 12, False: 0] | Branch (72:49): [True: 12, False: 0] ------------------ 73| 592| } _ZNK8rawspeed14Cr2SliceWidths12widthOfSliceEi: 75| 3.11M| [[nodiscard]] int widthOfSlice(int sliceId) const { 76| 3.11M| invariant(sliceId >= 0 && sliceId < numSlices); ------------------ | | 27| 3.11M|#define invariant(expr) assert(expr) ------------------ | Branch (76:5): [True: 3.11M, False: 0] | Branch (76:5): [True: 3.11M, False: 0] | Branch (76:5): [True: 3.11M, False: 0] ------------------ 77| 3.11M| if ((sliceId + 1) == numSlices) ------------------ | Branch (77:9): [True: 380k, False: 2.73M] ------------------ 78| 380k| return lastSliceWidth; 79| 2.73M| return sliceWidth; 80| 3.11M| } _ZN8rawspeed21Cr2SliceWidthIteratorC2ERKNS_14Cr2SliceWidthsEi: 98| 2.16k| : slicing(slicing_), sliceId(sliceId_) { 99| | invariant(sliceId >= 0 && sliceId <= slicing.numSlices && ------------------ | | 27| 2.16k|#define invariant(expr) assert(expr) ------------------ | Branch (99:5): [True: 2.16k, False: 0] | Branch (99:5): [True: 2.16k, False: 0] | Branch (99:5): [True: 2.16k, Folded] | Branch (99:5): [True: 2.16k, False: 0] ------------------ 100| 2.16k| "Iterator overflow"); 101| 2.16k| } _ZNK8rawspeed21Cr2SliceWidthIteratordeEv: 103| 858k| value_type operator*() const { 104| 858k| invariant(sliceId >= 0 && sliceId < slicing.numSlices && ------------------ | | 27| 858k|#define invariant(expr) assert(expr) ------------------ | Branch (104:5): [True: 858k, False: 0] | Branch (104:5): [True: 858k, False: 0] | Branch (104:5): [True: 858k, Folded] | Branch (104:5): [True: 858k, False: 0] ------------------ 105| 858k| "Iterator overflow"); 106| 858k| return slicing.widthOfSlice(sliceId); 107| 858k| } _ZN8rawspeed21Cr2SliceWidthIteratorppEv: 108| 66.7k| Cr2SliceWidthIterator& operator++() { 109| 66.7k| ++sliceId; 110| 66.7k| return *this; 111| 66.7k| } _ZN8rawspeedeqERKNS_21Cr2SliceWidthIteratorES2_: 113| 163k| const Cr2SliceWidthIterator& b) { 114| 163k| invariant(&a.slicing == &b.slicing && "Comparing unrelated iterators."); ------------------ | | 27| 163k|#define invariant(expr) assert(expr) ------------------ | Branch (114:5): [True: 163k, False: 0] | Branch (114:5): [True: 163k, Folded] | Branch (114:5): [True: 163k, False: 0] ------------------ 115| 163k| return a.sliceId == b.sliceId; 116| 163k| } _ZNK8rawspeed14Cr2SliceWidths5beginEv: 119| 1.08k|inline Cr2SliceWidthIterator Cr2SliceWidths::begin() const { 120| 1.08k| return {*this, 0}; 121| 1.08k|} _ZNK8rawspeed14Cr2SliceWidths3endEv: 122| 1.08k|inline Cr2SliceWidthIterator Cr2SliceWidths::end() const { 123| 1.08k| return {*this, numSlices}; 124| 1.08k|} _ZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE9getSlicesEv: 209| 1.08k|Cr2Decompressor::getSlices() const { 210| 1.08k| return {Cr2SliceIterator(slicing.begin(), frame), 211| 1.08k| Cr2SliceIterator(slicing.end(), frame)}; 212| 1.08k|} _ZN8rawspeed16Cr2SliceIteratorC2ERKNS_21Cr2SliceWidthIteratorERKNS_8iPoint2DE: 89| 2.16k| : frameHeight(frame.y), widthIter(sliceWidthIter_) {} _ZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE17getAllOutputTilesEv: 216| 1.08k|Cr2Decompressor::getAllOutputTiles() const { 217| 1.08k| auto slices = getSlices(); 218| 1.08k| return {Cr2OutputTileIterator(std::begin(slices), dim), 219| 1.08k| Cr2OutputTileIterator(std::end(slices), dim)}; 220| 1.08k|} _ZN8rawspeed21Cr2OutputTileIteratorC2ERKNS_16Cr2SliceIteratorERKNS_8iPoint2DE: 119| 2.16k| : imgDim(imgDim_), sliceIter(sliceIter_) {} _ZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE14getOutputTilesEv: 224| 512|Cr2Decompressor::getOutputTiles() const { 225| 512| auto allOutputTiles = getAllOutputTiles(); 226| 512| auto first = allOutputTiles.begin(); 227| 512| auto end = allOutputTiles.end(); 228| 512| invariant(first != end && "No tiles?"); ------------------ | | 27| 512|#define invariant(expr) assert(expr) ------------------ | Branch (228:3): [True: 512, False: 0] | Branch (228:3): [True: 512, Folded] | Branch (228:3): [True: 512, False: 0] ------------------ 229| 512| auto last = first; 230| 75.8k| while (std::next(last) != end && (*last).getBottomRight() != dim) ------------------ | Branch (230:10): [True: 75.8k, False: 2] | Branch (230:10): [True: 75.3k, False: 512] | Branch (230:36): [True: 75.3k, False: 510] ------------------ 231| 75.3k| ++last; 232| 512| assert((*last).getBottomRight() == dim && "Bad tiling"); ------------------ | Branch (232:3): [True: 512, False: 0] | Branch (232:3): [True: 512, Folded] | Branch (232:3): [True: 512, False: 0] ------------------ 233| 512| return {first, ++last}; 234| 512|} _ZN8rawspeedeqERKNS_21Cr2OutputTileIteratorES2_: 149| 163k| const Cr2OutputTileIterator& b) { 150| 163k| invariant(&a.imgDim == &b.imgDim && "Unrelated iterators."); ------------------ | | 27| 163k|#define invariant(expr) assert(expr) ------------------ | Branch (150:5): [True: 163k, False: 0] | Branch (150:5): [True: 163k, Folded] | Branch (150:5): [True: 163k, False: 0] ------------------ 151| | // NOTE: outPos is correctly omitted here. 152| 163k| return a.sliceIter == b.sliceIter && a.sliceRow == b.sliceRow; ------------------ | Branch (152:12): [True: 5.52k, False: 157k] | Branch (152:42): [True: 2.38k, False: 3.13k] ------------------ 153| 163k| } _ZN8rawspeedeqERKNS_16Cr2SliceIteratorES2_: 98| 163k| friend bool operator==(const Cr2SliceIterator& a, const Cr2SliceIterator& b) { 99| 163k| invariant(a.frameHeight == b.frameHeight && "Unrelated iterators."); ------------------ | | 27| 163k|#define invariant(expr) assert(expr) ------------------ | Branch (99:5): [True: 163k, False: 0] | Branch (99:5): [True: 163k, Folded] | Branch (99:5): [True: 163k, False: 0] ------------------ 100| 163k| return a.widthIter == b.widthIter; 101| 163k| } _ZNK8rawspeed21Cr2OutputTileIteratordeEv: 121| 392k| value_type operator*() const { 122| | // Positioning 123| 392k| iRectangle2D tile = {outPos, *sliceIter}; 124| | // Clamping 125| 392k| int outRowsRemaining = imgDim.y - tile.getTop(); 126| 392k| invariant(outRowsRemaining >= 0); ------------------ | | 27| 392k|#define invariant(expr) assert(expr) ------------------ | Branch (126:5): [True: 392k, False: 0] ------------------ 127| 392k| int tileRowsRemaining = tile.getHeight() - sliceRow; 128| 392k| invariant(tileRowsRemaining >= 0); ------------------ | | 27| 392k|#define invariant(expr) assert(expr) ------------------ | Branch (128:5): [True: 392k, False: 0] ------------------ 129| 392k| const int tileHeight = std::min(outRowsRemaining, tileRowsRemaining); 130| 392k| tile.dim.y = tileHeight; 131| 392k| return tile; 132| 392k| } _ZNK8rawspeed16Cr2SliceIteratordeEv: 91| 858k| value_type RAWSPEED_READONLY operator*() const { 92| 858k| return {*widthIter, frameHeight}; 93| 858k| } _ZN8rawspeed21Cr2OutputTileIteratorppEv: 133| 233k| Cr2OutputTileIterator& operator++() { 134| 233k| const iRectangle2D currTile = operator*(); 135| 233k| sliceRow += currTile.getHeight(); 136| 233k| outPos = currTile.getBottomLeft(); 137| 233k| invariant(sliceRow >= 0 && sliceRow <= (*sliceIter).y && "Overflow"); ------------------ | | 27| 233k|#define invariant(expr) assert(expr) ------------------ | Branch (137:5): [True: 233k, False: 0] | Branch (137:5): [True: 233k, False: 0] | Branch (137:5): [True: 233k, Folded] | Branch (137:5): [True: 233k, False: 0] ------------------ 138| 233k| if (sliceRow == (*sliceIter).y) { ------------------ | Branch (138:9): [True: 66.7k, False: 166k] ------------------ 139| 66.7k| ++sliceIter; 140| 66.7k| sliceRow = 0; 141| 66.7k| } 142| 233k| if (outPos.y == imgDim.y) { ------------------ | Branch (142:9): [True: 175k, False: 57.4k] ------------------ 143| 175k| outPos.y = 0; 144| 175k| outPos.x += currTile.getWidth(); 145| 175k| } 146| 233k| return *this; 147| 233k| } _ZN8rawspeed16Cr2SliceIteratorppEv: 94| 66.7k| Cr2SliceIterator& operator++() { 95| 66.7k| ++widthIter; 96| 66.7k| return *this; 97| 66.7k| } _ZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE23getVerticalOutputStripsEv: 238| 512|Cr2Decompressor::getVerticalOutputStrips() const { 239| 512| auto outputTiles = getOutputTiles(); 240| 512| return {Cr2VerticalOutputStripIterator(std::begin(outputTiles), 241| 512| std::end(outputTiles)), 242| 512| Cr2VerticalOutputStripIterator(std::end(outputTiles), 243| 512| std::end(outputTiles))}; 244| 512|} _ZN8rawspeed30Cr2VerticalOutputStripIteratorC2ERKNS_21Cr2OutputTileIteratorES3_: 191| 1.02k| : outputTileIterator(outputTileIterator_), 192| 1.02k| outputTileIterator_end(outputTileIterator_end_) {} _ZN8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEEC2ENS_8RawImageENSt3__15tupleIJiiiEEENS_8iPoint2DENS_14Cr2SliceWidthsENS8_6vectorINS6_18PerComponentRecipeENS8_9allocatorISE_EEEENS_10Array1DRefIKhEE: 285| 576| : mRaw(std::move(mRaw_)), format(std::move(format_)), frame(frame_), 286| 576| slicing(slicing_), rec(std::move(rec_)), input(input_) { 287| 576| if (mRaw->getDataType() != RawImageType::UINT16) ------------------ | Branch (287:7): [True: 0, False: 576] ------------------ 288| 576| ThrowRDE("Unexpected data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 289| | 290| 576| if (mRaw->getCpp() != 1 || mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (290:7): [True: 0, False: 576] | Branch (290:30): [True: 0, False: 576] ------------------ 291| 576| ThrowRDE("Unexpected cpp: %u", mRaw->getCpp()); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 292| | 293| 576| if (!((std::make_tuple(3, 2, 2) == format) || ------------------ | Branch (293:7): [True: 0, False: 576] | Branch (293:9): [True: 0, False: 576] ------------------ 294| 576| (std::make_tuple(3, 2, 1) == format) || ------------------ | Branch (294:9): [True: 0, False: 576] ------------------ 295| 576| (std::make_tuple(2, 1, 1) == format) || ------------------ | Branch (295:9): [True: 323, False: 253] ------------------ 296| 253| (std::make_tuple(4, 1, 1) == format))) ------------------ | Branch (296:9): [True: 253, False: 0] ------------------ 297| 576| ThrowRDE("Unknown format <%i,%i,%i>", std::get<0>(format), ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 298| 576| std::get<1>(format), std::get<2>(format)); 299| | 300| 576| const Dsc dsc(format); 301| | 302| 576| dim = mRaw->dim; 303| 576| if (!dim.hasPositiveArea() || dim.x % dsc.groupSize != 0) ------------------ | Branch (303:7): [True: 0, False: 576] | Branch (303:33): [True: 1, False: 575] ------------------ 304| 576| ThrowRDE("Unexpected image dimension multiplicity"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 305| 575| dim.x /= dsc.groupSize; 306| | 307| 575| if (!frame.hasPositiveArea() || frame.x % dsc.X_S_F != 0 || ------------------ | Branch (307:7): [True: 0, False: 575] | Branch (307:35): [True: 0, False: 575] ------------------ 308| 575| frame.y % dsc.Y_S_F != 0) ------------------ | Branch (308:7): [True: 0, False: 575] ------------------ 309| 575| ThrowRDE("Unexpected LJpeg frame dimension multiplicity"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 310| 575| frame.x /= dsc.X_S_F; 311| 575| frame.y /= dsc.Y_S_F; 312| | 313| 575| if (mRaw->dim.x > 19440 || mRaw->dim.y > 5920) { ------------------ | Branch (313:7): [True: 0, False: 575] | Branch (313:30): [True: 0, False: 575] ------------------ 314| 0| ThrowRDE("Unexpected image dimensions found: (%d; %d)", mRaw->dim.x, ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 315| 0| mRaw->dim.y); 316| 0| } 317| | 318| 1.10M| for (auto sliceId = 0; sliceId < slicing.numSlices; sliceId++) { ------------------ | Branch (318:26): [True: 1.10M, False: 575] ------------------ 319| 1.10M| const auto sliceWidth = slicing.widthOfSlice(sliceId); 320| 1.10M| if (sliceWidth <= 0) ------------------ | Branch (320:9): [True: 0, False: 1.10M] ------------------ 321| 1.10M| ThrowRDE("Bad slice width: %i", sliceWidth); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 322| 1.10M| } 323| | 324| 575| if (dsc.subSampled == mRaw->isCFA) ------------------ | Branch (324:7): [True: 0, False: 575] ------------------ 325| 575| ThrowRDE("Cannot decode subsampled image to CFA data or vice versa"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 326| | 327| 575| if (static_cast(rec.size()) != dsc.N_COMP) ------------------ | Branch (327:7): [True: 0, False: 575] ------------------ 328| 575| ThrowRDE("HT/Initial predictor count does not match component count"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 329| | 330| 1.65k| for (const auto& recip : rec) { ------------------ | Branch (330:26): [True: 1.65k, False: 575] ------------------ 331| 1.65k| if (!recip.ht.isFullDecode()) ------------------ | Branch (331:9): [True: 0, False: 1.65k] ------------------ 332| 1.65k| ThrowRDE("Huffman table is not of a full decoding variety"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 333| 1.65k| } 334| | 335| 1.14k| for (auto* width : {&slicing.sliceWidth, &slicing.lastSliceWidth}) { ------------------ | Branch (335:20): [True: 1.14k, False: 572] ------------------ 336| 1.14k| if (*width % dsc.sliceColStep != 0) { ------------------ | Branch (336:9): [True: 3, False: 1.14k] ------------------ 337| 3| ThrowRDE("Slice width (%d) should be multiple of pixel group size (%d)", ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 338| 3| *width, dsc.sliceColStep); 339| 3| } 340| 1.14k| *width /= dsc.sliceColStep; 341| 1.14k| } 342| | 343| 572| if (frame.area() < dim.area()) ------------------ | Branch (343:7): [True: 2, False: 570] ------------------ 344| 572| ThrowRDE("Frame area smaller than the image area"); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 345| | 346| 570| Optional lastTile; 347| 76.5k| for (iRectangle2D output : getAllOutputTiles()) { ------------------ | Branch (347:28): [True: 76.5k, False: 9] ------------------ 348| 76.5k| if (lastTile && evaluateConsecutiveTiles(*lastTile, output) == ------------------ | Branch (348:9): [True: 75.9k, False: 570] | Branch (348:21): [True: 1, False: 75.9k] ------------------ 349| 75.9k| TileSequenceStatus::Invalid) 350| 76.5k| ThrowRDE("Invalid tiling - slice width change mid-output row?"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 351| 76.5k| if (output.getBottomRight() <= dim) { ------------------ | Branch (351:9): [True: 75.9k, False: 560] ------------------ 352| 75.9k| lastTile = output; 353| 75.9k| continue; // Tile still inbounds of image. 354| 75.9k| } 355| 560| if (output.getTopLeft() < dim) ------------------ | Branch (355:9): [True: 11, False: 549] ------------------ 356| 560| ThrowRDE("Output tile partially outside of image"); ------------------ | | 38| 11| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 11| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 11| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 11| __VA_ARGS__) | | ------------------ ------------------ 357| 549| break; // Skip the rest of the tiles - they do not contribute to the image. 358| 560| } 359| 558| if (!lastTile) ------------------ | Branch (359:7): [True: 0, False: 558] ------------------ 360| 558| ThrowRDE("No tiles are provided"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 361| 558| if (lastTile->getBottomRight() != dim) ------------------ | Branch (361:7): [True: 5, False: 553] ------------------ 362| 558| ThrowRDE("Tiles do not cover the entire image area."); ------------------ | | 38| 5| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 5| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 5| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 5| __VA_ARGS__) | | ------------------ ------------------ 363| 558|} Cr2Decompressor.cpp:_ZN8rawspeed12_GLOBAL__N_13DscC2ENSt3__15tupleIJiiiEEE: 270| 576| : N_COMP(std::get<0>(format)), X_S_F(std::get<1>(format)), 271| 576| Y_S_F(std::get<2>(format)), subSampled(X_S_F != 1 || Y_S_F != 1), ------------------ | Branch (271:48): [True: 0, False: 576] | Branch (271:62): [True: 0, False: 576] ------------------ 272| 576| sliceColStep(N_COMP * X_S_F), pixelsPerGroup(X_S_F * Y_S_F), 273| 576| groupSize(!subSampled ? N_COMP : 2 + pixelsPerGroup), ------------------ | Branch (273:19): [True: 576, False: 0] ------------------ 274| 576| cpp(!subSampled ? 1 : 3), colsPerGroup(!subSampled ? cpp : groupSize) {} ------------------ | Branch (274:13): [True: 576, False: 0] | Branch (274:48): [True: 576, False: 0] ------------------ Cr2Decompressor.cpp:_ZN8rawspeed12_GLOBAL__N_124evaluateConsecutiveTilesERKNS_12iRectangle2DES3_: 62| 79.2k| const iRectangle2D& nextRect) { 63| 79.2k| using enum TileSequenceStatus; 64| | // Are these two are verically-adjacent rectangles of same width? 65| 79.2k| if (rect.getBottomLeft() == nextRect.getTopLeft() && ------------------ | Branch (65:7): [True: 19.6k, False: 59.5k] | Branch (65:7): [True: 19.6k, False: 59.5k] ------------------ 66| 19.6k| rect.getBottomRight() == nextRect.getTopRight()) ------------------ | Branch (66:7): [True: 19.6k, False: 1] ------------------ 67| 19.6k| return ContinuesColumn; 68| | // Otherwise, the next rectangle should be the first row of next column. 69| 59.5k| if (nextRect.getTop() == 0 && nextRect.getLeft() == rect.getRight()) ------------------ | Branch (69:7): [True: 59.5k, False: 1] | Branch (69:33): [True: 59.5k, False: 0] ------------------ 70| 59.5k| return BeginsNewColumn; 71| 1| return Invalid; 72| 59.5k|} _ZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE10decompressEv: 471| 553|ByteStream::size_type Cr2Decompressor::decompress() const { 472| 553| if (std::make_tuple(3, 2, 2) == format) { ------------------ | Branch (472:7): [True: 0, False: 553] ------------------ 473| 0| return decompressN_X_Y<3, 2, 2>(); // Cr2 sRaw1/mRaw 474| 0| } 475| 553| if (std::make_tuple(3, 2, 1) == format) { ------------------ | Branch (475:7): [True: 0, False: 553] ------------------ 476| 0| return decompressN_X_Y<3, 2, 1>(); // Cr2 sRaw2/sRaw 477| 0| } 478| 553| if (std::make_tuple(2, 1, 1) == format) { ------------------ | Branch (478:7): [True: 308, False: 245] ------------------ 479| 308| return decompressN_X_Y<2, 1, 1>(); 480| 308| } 481| 245| if (std::make_tuple(4, 1, 1) == format) { ------------------ | Branch (481:7): [True: 245, False: 0] ------------------ 482| 245| return decompressN_X_Y<4, 1, 1>(); 483| 245| } 484| 0| __builtin_unreachable(); 485| 245|} _ZN8rawspeedeqERKNS_30Cr2VerticalOutputStripIteratorES2_: 200| 1.74k| const Cr2VerticalOutputStripIterator& b) { 201| 1.74k| invariant(a.outputTileIterator_end == b.outputTileIterator_end && ------------------ | | 27| 1.74k|#define invariant(expr) assert(expr) ------------------ | Branch (201:5): [True: 1.74k, False: 0] | Branch (201:5): [True: 1.74k, Folded] | Branch (201:5): [True: 1.74k, False: 0] ------------------ 202| 1.74k| "Comparing unrelated iterators."); 203| 1.74k| return a.outputTileIterator == b.outputTileIterator; 204| 1.74k| } _ZNK8rawspeed30Cr2VerticalOutputStripIteratordeEv: 194| 1.66k| value_type operator*() const { return coalesce().first; } _ZNK8rawspeed30Cr2VerticalOutputStripIterator8coalesceEv: 160| 2.89k| [[nodiscard]] std::pair coalesce() const { 161| 2.89k| Cr2OutputTileIterator tmpIter = outputTileIterator; 162| 2.89k| invariant(tmpIter != outputTileIterator_end && "Iterator overflow."); ------------------ | | 27| 2.89k|#define invariant(expr) assert(expr) ------------------ | Branch (162:5): [True: 2.89k, False: 0] | Branch (162:5): [True: 2.89k, Folded] | Branch (162:5): [True: 2.89k, False: 0] ------------------ 163| | 164| 2.89k| iRectangle2D rect = *tmpIter; 165| 2.89k| int num = 1; 166| | 167| 3.84k| for (++tmpIter; tmpIter != outputTileIterator_end; ++tmpIter) { ------------------ | Branch (167:21): [True: 3.28k, False: 558] ------------------ 168| 3.28k| iRectangle2D nextRect = *tmpIter; 169| 3.28k| const TileSequenceStatus s = evaluateConsecutiveTiles(rect, nextRect); 170| 3.28k| invariant(s != TileSequenceStatus::Invalid && "Bad tiling."); ------------------ | | 27| 3.28k|#define invariant(expr) assert(expr) ------------------ | Branch (170:7): [True: 3.28k, False: 0] | Branch (170:7): [True: 3.28k, Folded] | Branch (170:7): [True: 3.28k, False: 0] ------------------ 171| 3.28k| if (s == TileSequenceStatus::BeginsNewColumn) ------------------ | Branch (171:11): [True: 2.33k, False: 950] ------------------ 172| 2.33k| break; 173| 950| invariant(s == TileSequenceStatus::ContinuesColumn); ------------------ | | 27| 950|#define invariant(expr) assert(expr) ------------------ | Branch (173:7): [True: 950, False: 0] ------------------ 174| 950| rect.dim.y += nextRect.dim.y; 175| 950| ++num; 176| 950| } 177| | 178| 2.89k| return {rect, num}; 179| 2.89k| } _ZN8rawspeed30Cr2VerticalOutputStripIteratorppEv: 195| 1.23k| Cr2VerticalOutputStripIterator& operator++() { 196| 1.23k| std::advance(outputTileIterator, coalesce().second); 197| 1.23k| return *this; 198| 1.23k| } _ZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE15decompressN_X_YILi2ELi1ELi1EEEjv: 399| 308|Cr2Decompressor::decompressN_X_Y() const { 400| 308| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 401| | 402| | // To understand the CR2 slice handling and sampling factor behavior, see 403| | // https://github.com/lclevy/libcraw2/blob/master/docs/cr2_lossless.pdf?raw=true 404| | 405| 308| constexpr Dsc dsc({N_COMP, X_S_F, Y_S_F}); 406| | 407| | // inner loop decodes one group of pixels at a time 408| | // * for : N = N*1*1 (full raw) 409| | // * for <3,2,1>: 6 = 3*2*1 410| | // * for <3,2,2>: 12 = 3*2*2 411| | // and advances x by N_COMP*X_S_F and y by Y_S_F 412| | 413| 308| auto ht = getPrefixCodeDecoders(); 414| 308| auto pred = getInitialPreds(); 415| 308| auto predNext = out[/*row=*/0] 416| 308| .getCrop(/*offset=*/0, /*size=*/dsc.groupSize) 417| 308| .getAsArray1DRef(); 418| | 419| 308| BitStreamerJPEG bs(input); 420| | 421| 308| int globalFrameCol = 0; 422| 308| int globalFrameRow = 0; 423| 308| (void)globalFrameRow; 424| | 425| 308| auto frameColsRemaining = [&]() { 426| 308| int r = frame.x - globalFrameCol; 427| 308| invariant(r >= 0); 428| 308| return r; 429| 308| }; 430| | 431| 1.32k| for (iRectangle2D output : getVerticalOutputStrips()) { ------------------ | Branch (431:28): [True: 1.32k, False: 308] ------------------ 432| 8.70k| for (int row = output.getTop(), rowEnd = output.getBottom(); row != rowEnd; ------------------ | Branch (432:66): [True: 7.37k, False: 1.32k] ------------------ 433| 7.37k| ++row) { 434| 7.37k| for (int col = output.getLeft(), colEnd = output.getRight(); 435| 78.3k| col != colEnd;) { ------------------ | Branch (435:12): [True: 70.9k, False: 7.37k] ------------------ 436| | // check if we processed one full raw row worth of pixels 437| 70.9k| if (frameColsRemaining() == 0) { ------------------ | Branch (437:13): [True: 64.3k, False: 6.60k] ------------------ 438| | // if yes -> update predictor by going back exactly one row, 439| | // no matter where we are right now. 440| | // makes no sense from an image compression point of view, ask 441| | // Canon. 442| 193k| for (int c = 0; c < N_COMP; ++c) ------------------ | Branch (442:27): [True: 128k, False: 64.3k] ------------------ 443| 128k| pred[c] = predNext(c == 0 ? c : dsc.groupSize - (N_COMP - c)); ------------------ | Branch (443:32): [True: 64.3k, False: 64.3k] ------------------ 444| 64.3k| predNext = out[row] 445| 64.3k| .getBlock(/*size=*/dsc.groupSize, 446| 64.3k| /*index=*/col) 447| 64.3k| .getAsArray1DRef(); 448| 64.3k| ++globalFrameRow; 449| 64.3k| globalFrameCol = 0; 450| 64.3k| invariant(globalFrameRow < frame.y && "Run out of frame"); ------------------ | | 27| 64.3k|#define invariant(expr) assert(expr) ------------------ | Branch (450:11): [True: 64.3k, False: 0] | Branch (450:11): [True: 64.3k, Folded] | Branch (450:11): [True: 64.3k, False: 0] ------------------ 451| 64.3k| } 452| | 453| | // How many pixel can we decode until we finish the row of either 454| | // the frame (i.e. predictor change time), or of the current slice? 455| 70.9k| for (int colFrameEnd = std::min(colEnd, col + frameColsRemaining()); 456| 2.08M| col != colFrameEnd; ++col, ++globalFrameCol) { ------------------ | Branch (456:14): [True: 2.01M, False: 70.9k] ------------------ 457| 6.03M| for (int p = 0; p < dsc.groupSize; ++p) { ------------------ | Branch (457:27): [True: 4.02M, False: 2.01M] ------------------ 458| 4.02M| int c = p < dsc.pixelsPerGroup ? 0 : p - dsc.pixelsPerGroup + 1; ------------------ | Branch (458:21): [True: 2.01M, False: 2.01M] ------------------ 459| 4.02M| out(row, (dsc.groupSize * col) + p) = pred[c] += 460| 4.02M| (static_cast(ht[c])) 461| 4.02M| .decodeDifference(bs); 462| 4.02M| } 463| 2.01M| } 464| 70.9k| } 465| 7.37k| } 466| 1.32k| } 467| 308| return bs.getStreamPosition(); 468| 308|} _ZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE21getPrefixCodeDecodersILi2EEENSt3__15arrayINS8_17reference_wrapperIKS5_EEXT_EEEv: 377| 308|Cr2Decompressor::getPrefixCodeDecoders() const { 378| 308| return getPrefixCodeDecodersImpl(std::make_index_sequence{}); 379| 308|} _ZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE25getPrefixCodeDecodersImplILi2EJLm0ELm1EEEENSt3__15arrayINS8_17reference_wrapperIKS5_EEXT_EEENS8_16integer_sequenceImJXspT0_EEEE: 369| 308| std::index_sequence /*unused*/) const { 370| 308| return std::array, N_COMP>{ 371| 308| std::cref(rec[I].ht)...}; 372| 308|} _ZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE15getInitialPredsILi2EEENSt3__15arrayItXT_EEEv: 384| 308|Cr2Decompressor::getInitialPreds() const { 385| 308| std::array preds; 386| 308| std::transform( 387| 308| rec.begin(), rec.end(), preds.begin(), 388| 308| [](const PerComponentRecipe& compRec) { return compRec.initPred; }); 389| 308| return preds; 390| 308|} _ZZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE15getInitialPredsILi2EEENSt3__15arrayItXT_EEEvENKUlRKNS6_18PerComponentRecipeEE_clESD_: 388| 616| [](const PerComponentRecipe& compRec) { return compRec.initPred; }); _ZZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE15decompressN_X_YILi2ELi1ELi1EEEjvENKUlvE_clEv: 425| 141k| auto frameColsRemaining = [&]() { 426| 141k| int r = frame.x - globalFrameCol; 427| 141k| invariant(r >= 0); ------------------ | | 27| 141k|#define invariant(expr) assert(expr) ------------------ | Branch (427:5): [True: 141k, False: 0] ------------------ 428| 141k| return r; 429| 141k| }; _ZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE15decompressN_X_YILi4ELi1ELi1EEEjv: 399| 245|Cr2Decompressor::decompressN_X_Y() const { 400| 245| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 401| | 402| | // To understand the CR2 slice handling and sampling factor behavior, see 403| | // https://github.com/lclevy/libcraw2/blob/master/docs/cr2_lossless.pdf?raw=true 404| | 405| 245| constexpr Dsc dsc({N_COMP, X_S_F, Y_S_F}); 406| | 407| | // inner loop decodes one group of pixels at a time 408| | // * for : N = N*1*1 (full raw) 409| | // * for <3,2,1>: 6 = 3*2*1 410| | // * for <3,2,2>: 12 = 3*2*2 411| | // and advances x by N_COMP*X_S_F and y by Y_S_F 412| | 413| 245| auto ht = getPrefixCodeDecoders(); 414| 245| auto pred = getInitialPreds(); 415| 245| auto predNext = out[/*row=*/0] 416| 245| .getCrop(/*offset=*/0, /*size=*/dsc.groupSize) 417| 245| .getAsArray1DRef(); 418| | 419| 245| BitStreamerJPEG bs(input); 420| | 421| 245| int globalFrameCol = 0; 422| 245| int globalFrameRow = 0; 423| 245| (void)globalFrameRow; 424| | 425| 245| auto frameColsRemaining = [&]() { 426| 245| int r = frame.x - globalFrameCol; 427| 245| invariant(r >= 0); 428| 245| return r; 429| 245| }; 430| | 431| 340| for (iRectangle2D output : getVerticalOutputStrips()) { ------------------ | Branch (431:28): [True: 340, False: 245] ------------------ 432| 2.36k| for (int row = output.getTop(), rowEnd = output.getBottom(); row != rowEnd; ------------------ | Branch (432:66): [True: 2.02k, False: 340] ------------------ 433| 2.02k| ++row) { 434| 2.02k| for (int col = output.getLeft(), colEnd = output.getRight(); 435| 38.6k| col != colEnd;) { ------------------ | Branch (435:12): [True: 36.6k, False: 2.02k] ------------------ 436| | // check if we processed one full raw row worth of pixels 437| 36.6k| if (frameColsRemaining() == 0) { ------------------ | Branch (437:13): [True: 34.7k, False: 1.94k] ------------------ 438| | // if yes -> update predictor by going back exactly one row, 439| | // no matter where we are right now. 440| | // makes no sense from an image compression point of view, ask 441| | // Canon. 442| 173k| for (int c = 0; c < N_COMP; ++c) ------------------ | Branch (442:27): [True: 138k, False: 34.7k] ------------------ 443| 138k| pred[c] = predNext(c == 0 ? c : dsc.groupSize - (N_COMP - c)); ------------------ | Branch (443:32): [True: 34.7k, False: 104k] ------------------ 444| 34.7k| predNext = out[row] 445| 34.7k| .getBlock(/*size=*/dsc.groupSize, 446| 34.7k| /*index=*/col) 447| 34.7k| .getAsArray1DRef(); 448| 34.7k| ++globalFrameRow; 449| 34.7k| globalFrameCol = 0; 450| 34.7k| invariant(globalFrameRow < frame.y && "Run out of frame"); ------------------ | | 27| 34.7k|#define invariant(expr) assert(expr) ------------------ | Branch (450:11): [True: 34.7k, False: 0] | Branch (450:11): [True: 34.7k, Folded] | Branch (450:11): [True: 34.7k, False: 0] ------------------ 451| 34.7k| } 452| | 453| | // How many pixel can we decode until we finish the row of either 454| | // the frame (i.e. predictor change time), or of the current slice? 455| 36.6k| for (int colFrameEnd = std::min(colEnd, col + frameColsRemaining()); 456| 730k| col != colFrameEnd; ++col, ++globalFrameCol) { ------------------ | Branch (456:14): [True: 694k, False: 36.6k] ------------------ 457| 3.47M| for (int p = 0; p < dsc.groupSize; ++p) { ------------------ | Branch (457:27): [True: 2.77M, False: 694k] ------------------ 458| 2.77M| int c = p < dsc.pixelsPerGroup ? 0 : p - dsc.pixelsPerGroup + 1; ------------------ | Branch (458:21): [True: 694k, False: 2.08M] ------------------ 459| 2.77M| out(row, (dsc.groupSize * col) + p) = pred[c] += 460| 2.77M| (static_cast(ht[c])) 461| 2.77M| .decodeDifference(bs); 462| 2.77M| } 463| 694k| } 464| 36.6k| } 465| 2.02k| } 466| 340| } 467| 245| return bs.getStreamPosition(); 468| 245|} _ZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE21getPrefixCodeDecodersILi4EEENSt3__15arrayINS8_17reference_wrapperIKS5_EEXT_EEEv: 377| 245|Cr2Decompressor::getPrefixCodeDecoders() const { 378| 245| return getPrefixCodeDecodersImpl(std::make_index_sequence{}); 379| 245|} _ZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE25getPrefixCodeDecodersImplILi4EJLm0ELm1ELm2ELm3EEEENSt3__15arrayINS8_17reference_wrapperIKS5_EEXT_EEENS8_16integer_sequenceImJXspT0_EEEE: 369| 245| std::index_sequence /*unused*/) const { 370| 245| return std::array, N_COMP>{ 371| 245| std::cref(rec[I].ht)...}; 372| 245|} _ZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE15getInitialPredsILi4EEENSt3__15arrayItXT_EEEv: 384| 245|Cr2Decompressor::getInitialPreds() const { 385| 245| std::array preds; 386| 245| std::transform( 387| 245| rec.begin(), rec.end(), preds.begin(), 388| 245| [](const PerComponentRecipe& compRec) { return compRec.initPred; }); 389| 245| return preds; 390| 245|} _ZZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE15getInitialPredsILi4EEENSt3__15arrayItXT_EEEvENKUlRKNS6_18PerComponentRecipeEE_clESD_: 388| 980| [](const PerComponentRecipe& compRec) { return compRec.initPred; }); _ZZNK8rawspeed15Cr2DecompressorINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS2_EEEEE15decompressN_X_YILi4ELi1ELi1EEEjvENKUlvE_clEv: 425| 73.3k| auto frameColsRemaining = [&]() { 426| 73.3k| int r = frame.x - globalFrameCol; 427| 73.3k| invariant(r >= 0); ------------------ | | 27| 73.3k|#define invariant(expr) assert(expr) ------------------ | Branch (427:5): [True: 73.3k, False: 0] ------------------ 428| 73.3k| return r; 429| 73.3k| }; _ZN8rawspeed15Cr2LJpegDecoderC2ENS_10ByteStreamERKNS_8RawImageE: 44| 977| : AbstractLJpegDecoder(bs, img) { 45| 977| if (mRaw->getDataType() != RawImageType::UINT16) ------------------ | Branch (45:7): [True: 0, False: 977] ------------------ 46| 977| ThrowRDE("Unexpected data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 47| | 48| 977| if (mRaw->getCpp() != 1 || mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (48:7): [True: 16, False: 961] | Branch (48:30): [True: 0, False: 961] ------------------ 49| 977| ThrowRDE("Unexpected cpp: %u", mRaw->getCpp()); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 50| | 51| 977| if (!mRaw->dim.x || !mRaw->dim.y || mRaw->dim.x > 19440 || ------------------ | Branch (51:7): [True: 16, False: 961] | Branch (51:23): [True: 0, False: 961] | Branch (51:39): [True: 0, False: 961] ------------------ 52| 961| mRaw->dim.y > 5920) { ------------------ | Branch (52:7): [True: 8, False: 953] ------------------ 53| 8| ThrowRDE("Unexpected image dimensions found: (%d; %d)", mRaw->dim.x, ------------------ | | 38| 8| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 8| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 8| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 8| __VA_ARGS__) | | ------------------ ------------------ 54| 8| mRaw->dim.y); 55| 8| } 56| 977|} _ZN8rawspeed15Cr2LJpegDecoder10decodeScanEv: 58| 595|Buffer::size_type Cr2LJpegDecoder::decodeScan() { 59| 595| if (numMCUsPerRestartInterval != 0) ------------------ | Branch (59:7): [True: 2, False: 593] ------------------ 60| 595| ThrowRDE("Non-zero restart interval not supported."); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 61| | 62| 593| if (predictorMode != 1) ------------------ | Branch (62:7): [True: 1, False: 592] ------------------ 63| 593| ThrowRDE("Unsupported predictor mode."); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 64| | 65| 592| if (slicing.empty()) { ------------------ | Branch (65:7): [True: 12, False: 580] ------------------ 66| 12| const int slicesWidth = frame.w * frame.cps; 67| 12| if (slicesWidth > mRaw->dim.x) ------------------ | Branch (67:9): [True: 7, False: 5] ------------------ 68| 12| ThrowRDE("Don't know slicing pattern, and failed to guess it."); ------------------ | | 38| 7| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 7| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 7| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 7| __VA_ARGS__) | | ------------------ ------------------ 69| | 70| 5| slicing = 71| 5| Cr2SliceWidths(/*numSlices=*/1, /*sliceWidth=don't care*/ 0, 72| 5| /*lastSliceWidth=*/implicit_cast(slicesWidth)); 73| 5| } 74| | 75| 585| bool isSubSampled = false; 76| 2.26k| for (uint32_t i = 0; i < frame.cps; i++) ------------------ | Branch (76:24): [True: 1.68k, False: 585] ------------------ 77| 1.68k| isSubSampled = isSubSampled || frame.compInfo[i].superH != 1 || ------------------ | Branch (77:20): [True: 3, False: 1.68k] | Branch (77:36): [True: 3, False: 1.67k] ------------------ 78| 1.67k| frame.compInfo[i].superV != 1; ------------------ | Branch (78:20): [True: 1, False: 1.67k] ------------------ 79| | 80| 585| if (frame.cps != 3 && frame.w * frame.cps > 2 * frame.h) { ------------------ | Branch (80:7): [True: 584, False: 1] | Branch (80:25): [True: 241, False: 343] ------------------ 81| | // Fix Canon double height issue where Canon doubled the width and halfed 82| | // the height (e.g. with 5Ds), ask Canon. frame.w needs to stay as is here 83| | // because the number of pixels after which the predictor gets updated is 84| | // still the doubled width. 85| | // see: FIX_CANON_HALF_HEIGHT_DOUBLE_WIDTH 86| 241| frame.h *= 2; 87| 241| } 88| | 89| 585| std::tuple format; 90| | 91| 585| if (isSubSampled) { ------------------ | Branch (91:7): [True: 4, False: 581] ------------------ 92| 4| if (mRaw->isCFA) ------------------ | Branch (92:9): [True: 4, False: 0] ------------------ 93| 4| ThrowRDE("Cannot decode subsampled image to CFA data"); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 94| | 95| 0| if (frame.cps != 3) ------------------ | Branch (95:9): [True: 0, False: 0] ------------------ 96| 0| ThrowRDE("Unsupported number of subsampled components: %u", frame.cps); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 97| | 98| | // see http://lclevy.free.fr/cr2/#sraw for overview table 99| 0| bool isSupported = frame.compInfo[0].superH == 2; 100| | 101| 0| isSupported = isSupported && (frame.compInfo[0].superV == 1 || ------------------ | Branch (101:19): [True: 0, False: 0] | Branch (101:35): [True: 0, False: 0] ------------------ 102| 0| frame.compInfo[0].superV == 2); ------------------ | Branch (102:35): [True: 0, False: 0] ------------------ 103| | 104| 0| for (uint32_t i = 1; i < frame.cps; i++) ------------------ | Branch (104:26): [True: 0, False: 0] ------------------ 105| 0| isSupported = isSupported && frame.compInfo[i].superH == 1 && ------------------ | Branch (105:21): [True: 0, False: 0] | Branch (105:36): [True: 0, False: 0] ------------------ 106| 0| frame.compInfo[i].superV == 1; ------------------ | Branch (106:21): [True: 0, False: 0] ------------------ 107| | 108| 0| if (!isSupported) { ------------------ | Branch (108:9): [True: 0, False: 0] ------------------ 109| 0| ThrowRDE("Unsupported subsampling ([[%u, %u], [%u, %u], [%u, %u]])", ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 110| 0| frame.compInfo[0].superH, frame.compInfo[0].superV, 111| 0| frame.compInfo[1].superH, frame.compInfo[1].superV, 112| 0| frame.compInfo[2].superH, frame.compInfo[2].superV); 113| 0| } 114| | 115| 0| if (frame.compInfo[0].superV == 2) ------------------ | Branch (115:9): [True: 0, False: 0] ------------------ 116| 0| format = {3, 2, 2}; // Cr2 sRaw1/mRaw 117| 0| else { 118| 0| assert(frame.compInfo[0].superV == 1); ------------------ | Branch (118:7): [True: 0, False: 0] ------------------ 119| | // fix the inconsistent slice width in sRaw mode, ask Canon. 120| 0| for (auto* width : {&slicing.sliceWidth, &slicing.lastSliceWidth}) ------------------ | Branch (120:24): [True: 0, False: 0] ------------------ 121| 0| *width = (*width) * 3 / 2; 122| 0| format = {3, 2, 1}; // Cr2 sRaw2/sRaw 123| 0| } 124| 581| } else { 125| 581| switch (frame.cps) { 126| 324| case 2: ------------------ | Branch (126:5): [True: 324, False: 257] ------------------ 127| 324| format = {2, 1, 1}; 128| 324| break; 129| 255| case 4: ------------------ | Branch (129:5): [True: 255, False: 326] ------------------ 130| 255| format = {4, 1, 1}; 131| 255| break; 132| 2| default: ------------------ | Branch (132:5): [True: 2, False: 579] ------------------ 133| 2| ThrowRDE("Unsupported number of components: %u", frame.cps); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 134| 581| } 135| 581| } 136| | 137| 579| int N_COMP = std::get<0>(format); 138| | 139| 579| std::vector>::PerComponentRecipe> rec; 140| 579| rec.reserve(N_COMP); 141| 579| std::generate_n( 142| 579| std::back_inserter(rec), N_COMP, 143| 579| [&rec, hts = getPrefixCodeDecoders(N_COMP), 144| 579| initPred = getInitialPredictors(N_COMP)]() 145| 579| -> Cr2Decompressor>::PerComponentRecipe { 146| 579| const auto i = implicit_cast(rec.size()); 147| 579| return {*hts[i], initPred[i]}; 148| 579| }); 149| | 150| 579| Cr2Decompressor> d( 151| 579| mRaw, format, iPoint2D(frame.w, frame.h), slicing, rec, 152| 579| input.peekRemainingBuffer().getAsArray1DRef()); 153| 579| return d.decompress(); 154| 585|} Cr2LJpegDecoder.cpp:_ZZN8rawspeed15Cr2LJpegDecoder10decodeScanEvENK3$_0clEv: 145| 1.65k| -> Cr2Decompressor>::PerComponentRecipe { 146| 1.65k| const auto i = implicit_cast(rec.size()); 147| 1.65k| return {*hts[i], initPred[i]}; 148| 1.65k| }); _ZN8rawspeed15Cr2LJpegDecoder6decodeERKNS_14Cr2SliceWidthsE: 156| 953|void Cr2LJpegDecoder::decode(const Cr2SliceWidths& slicing_) { 157| 953| slicing = slicing_; 158| 1.14M| for (auto sliceId = 0; sliceId < slicing.numSlices; sliceId++) { ------------------ | Branch (158:26): [True: 1.14M, False: 953] ------------------ 159| 1.14M| const auto sliceWidth = slicing.widthOfSlice(sliceId); 160| 1.14M| if (sliceWidth <= 0) ------------------ | Branch (160:9): [True: 0, False: 1.14M] ------------------ 161| 1.14M| ThrowRDE("Bad slice width: %i", sliceWidth); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 162| 1.14M| } 163| | 164| 953| AbstractLJpegDecoder::decodeSOI(); 165| 953|} _ZN8rawspeed22HasselbladDecompressorC2ENS_8RawImageERKNS0_18PerComponentRecipeENS_10Array1DRefIKhEE: 41| 89| : mRaw(std::move(mRaw_)), rec(rec_), input(input_) { 42| 89| if (mRaw->getDataType() != RawImageType::UINT16) ------------------ | Branch (42:7): [True: 0, False: 89] ------------------ 43| 89| ThrowRDE("Unexpected data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 44| | 45| 89| if (mRaw->getCpp() != 1 || mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (45:7): [True: 0, False: 89] | Branch (45:30): [True: 0, False: 89] ------------------ 46| 89| ThrowRDE("Unexpected cpp: %u", mRaw->getCpp()); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 47| | 48| | // FIXME: could be wrong. max "active pixels" - "100 MP" 49| 89| if (!mRaw->dim.hasPositiveArea() || mRaw->dim.x % 2 != 0 || ------------------ | Branch (49:7): [True: 0, False: 89] | Branch (49:39): [True: 0, False: 89] ------------------ 50| 89| mRaw->dim.x > 12000 || mRaw->dim.y > 8842) { ------------------ | Branch (50:7): [True: 0, False: 89] | Branch (50:30): [True: 0, False: 89] ------------------ 51| 0| ThrowRDE("Unexpected image dimensions found: (%d; %d)", mRaw->dim.x, ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 52| 0| mRaw->dim.y); 53| 0| } 54| | 55| 89| if (rec.ht.isFullDecode()) ------------------ | Branch (55:7): [True: 0, False: 89] ------------------ 56| 89| ThrowRDE("Huffman table is of a full decoding variety"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 57| 89|} _ZN8rawspeed22HasselbladDecompressor10decompressEv: 71| 89|ByteStream::size_type HasselbladDecompressor::decompress() { 72| 89| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 73| | 74| 89| invariant(out.height() > 0); ------------------ | | 27| 89|#define invariant(expr) assert(expr) ------------------ | Branch (74:3): [True: 89, False: 0] ------------------ 75| 89| invariant(out.width() > 0); ------------------ | | 27| 89|#define invariant(expr) assert(expr) ------------------ | Branch (75:3): [True: 89, False: 0] ------------------ 76| 89| invariant(out.width() % 2 == 0); ------------------ | | 27| 89|#define invariant(expr) assert(expr) ------------------ | Branch (76:3): [True: 89, False: 0] ------------------ 77| | 78| 89| const auto ht = rec.ht; 79| 89| ht.verifyCodeValuesAsDiffLengths(); 80| | 81| 89| BitStreamerMSB32 bitStreamer(input); 82| | // Pixels are packed two at a time, not like LJPEG: 83| | // [p1_length_as_huffman][p2_length_as_huffman][p0_diff_with_length][p1_diff_with_length]|NEXT 84| | // PIXELS 85| 224| for (int row = 0; row < out.height(); row++) { ------------------ | Branch (85:21): [True: 135, False: 89] ------------------ 86| 135| int p1 = rec.initPred; 87| 135| int p2 = rec.initPred; 88| 10.6k| for (int col = 0; col < out.width(); col += 2) { ------------------ | Branch (88:23): [True: 10.5k, False: 135] ------------------ 89| 10.5k| int len1 = ht.decodeCodeValue(bitStreamer); 90| 10.5k| int len2 = ht.decodeCodeValue(bitStreamer); 91| 10.5k| p1 += getBits(bitStreamer, len1); 92| 10.5k| p2 += getBits(bitStreamer, len2); 93| | // NOTE: this is rather unusual and weird, but appears to be correct. 94| | // clampBits(p, 16) results in completely garbled images. 95| 10.5k| out(row, col) = uint16_t(p1); 96| 10.5k| out(row, col + 1) = uint16_t(p2); 97| 10.5k| } 98| 135| } 99| 89| return bitStreamer.getStreamPosition(); 100| 89|} _ZN8rawspeed22HasselbladDecompressor7getBitsERNS_16BitStreamerMSB32Ei: 61| 20.8k|inline int HasselbladDecompressor::getBits(BitStreamerMSB32& bs, int len) { 62| 20.8k| if (!len) ------------------ | Branch (62:7): [True: 11.1k, False: 9.70k] ------------------ 63| 11.1k| return 0; 64| 9.70k| int diff = bs.getBits(len); 65| 9.70k| diff = PrefixCodeDecoder<>::extend(diff, len); 66| 9.70k| if (diff == 65535) ------------------ | Branch (66:7): [True: 56, False: 9.64k] ------------------ 67| 56| return -32768; 68| 9.64k| return diff; 69| 9.70k|} _ZN8rawspeed22HasselbladLJpegDecoderC2ENS_10ByteStreamERKNS_8RawImageE: 37| 151| : AbstractLJpegDecoder(bs, img) { 38| 151| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (38:7): [True: 36, False: 115] | Branch (38:30): [True: 0, False: 115] ------------------ 39| 115| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (39:7): [True: 0, False: 115] ------------------ 40| 151| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 41| | 42| | // FIXME: could be wrong. max "active pixels" - "100 MP" 43| 151| if (!mRaw->dim.hasPositiveArea() || mRaw->dim.x % 2 != 0 || ------------------ | Branch (43:7): [True: 36, False: 115] | Branch (43:39): [True: 5, False: 110] ------------------ 44| 110| mRaw->dim.x > 12000 || mRaw->dim.y > 8842) { ------------------ | Branch (44:7): [True: 1, False: 109] | Branch (44:30): [True: 0, False: 109] ------------------ 45| 6| ThrowRDE("Unexpected image dimensions found: (%d; %d)", mRaw->dim.x, ------------------ | | 38| 6| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 6| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 6| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 6| __VA_ARGS__) | | ------------------ ------------------ 46| 6| mRaw->dim.y); 47| 6| } 48| 151|} _ZN8rawspeed22HasselbladLJpegDecoder10decodeScanEv: 50| 94|Buffer::size_type HasselbladLJpegDecoder::decodeScan() { 51| 94| if (numMCUsPerRestartInterval != 0) ------------------ | Branch (51:7): [True: 1, False: 93] ------------------ 52| 94| ThrowRDE("Non-zero restart interval not supported."); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 53| | 54| 93| if (frame.w != static_cast(mRaw->dim.x) || ------------------ | Branch (54:7): [True: 1, False: 92] ------------------ 55| 92| frame.h != static_cast(mRaw->dim.y)) { ------------------ | Branch (55:7): [True: 2, False: 90] ------------------ 56| 3| ThrowRDE("LJPEG frame does not match EXIF dimensions: (%u; %u) vs (%i; %i)", ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 57| 3| frame.w, frame.h, mRaw->dim.x, mRaw->dim.y); 58| 3| } 59| | 60| 90| const HasselbladDecompressor::PerComponentRecipe rec = { 61| 90| *getPrefixCodeDecoders(1)[0], getInitialPredictors(1)[0]}; 62| | 63| 90| HasselbladDecompressor d(mRaw, rec, 64| 90| input.peekRemainingBuffer().getAsArray1DRef()); 65| 90| return d.decompress(); 66| 93|} _ZN8rawspeed22HasselbladLJpegDecoder6decodeEv: 68| 109|void HasselbladLJpegDecoder::decode() { 69| | // We cannot use fully decoding huffman table, 70| | // because values are packed two pixels at the time. 71| 109| fullDecodeHT = false; 72| | 73| 109| AbstractLJpegDecoder::decodeSOI(); 74| 109|} _ZN8rawspeed10peekMarkerENS_10ByteStreamE: 110| 4.21M|inline Optional peekMarker(ByteStream input) { 111| 4.21M| uint8_t c0 = input.peekByte(0); 112| 4.21M| uint8_t c1 = input.peekByte(1); 113| | 114| 4.21M| if (c0 == 0xFF && c1 != 0 && c1 != 0xFF) ------------------ | Branch (114:7): [True: 23.0k, False: 4.18M] | Branch (114:21): [True: 21.8k, False: 1.18k] | Branch (114:32): [True: 13.7k, False: 8.03k] ------------------ 115| 13.7k| return static_cast(c1); 116| 4.19M| return {}; 117| 4.21M|} _ZN8rawspeed19advanceToNextMarkerENS_10ByteStreamEb: 120| 7.38k| bool skipPadding) { 121| 4.20M| while (input.getRemainSize() >= 2) { ------------------ | Branch (121:10): [True: 4.20M, False: 158] ------------------ 122| 4.20M| if (Optional m = peekMarker(input)) ------------------ | Branch (122:30): [True: 6.88k, False: 4.19M] ------------------ 123| 6.88k| return input; 124| | 125| | // Marker not found. Might there be leading padding bytes? 126| 4.19M| if (!skipPadding) ------------------ | Branch (126:9): [True: 343, False: 4.19M] ------------------ 127| 343| break; // Nope, give up. 128| | 129| | // Advance by a single(!) byte and try again. 130| 4.19M| input.skipBytes(1); 131| 4.19M| } 132| | 133| 501| return std::nullopt; 134| 7.38k|} _ZN8rawspeed22getRestartMarkerNumberENS_10JpegMarkerE: 137| 8|inline Optional getRestartMarkerNumber(JpegMarker m) { 138| 8| switch (m) { 139| 0| using enum JpegMarker; 140| 6| case RST0: ------------------ | Branch (140:3): [True: 6, False: 2] ------------------ 141| 6| case RST1: ------------------ | Branch (141:3): [True: 0, False: 8] ------------------ 142| 6| case RST2: ------------------ | Branch (142:3): [True: 0, False: 8] ------------------ 143| 6| case RST3: ------------------ | Branch (143:3): [True: 0, False: 8] ------------------ 144| 6| case RST4: ------------------ | Branch (144:3): [True: 0, False: 8] ------------------ 145| 6| case RST5: ------------------ | Branch (145:3): [True: 0, False: 8] ------------------ 146| 6| case RST6: ------------------ | Branch (146:3): [True: 0, False: 8] ------------------ 147| 7| case RST7: ------------------ | Branch (147:3): [True: 1, False: 7] ------------------ 148| 7| return static_cast(m) - static_cast(RST0); 149| 1| default: ------------------ | Branch (149:3): [True: 1, False: 7] ------------------ 150| 1| return std::nullopt; // Not a restart marker. 151| 8| } 152| 8|} _ZN8rawspeed17KodakDecompressorC2ENS_8RawImageENS_10ByteStreamEib: 48| 70| : mRaw(std::move(img)), input(bs), bps(bps_), 49| 70| uncorrectedRawValues(uncorrectedRawValues_) { 50| 70| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (50:7): [True: 0, False: 70] | Branch (50:30): [True: 0, False: 70] ------------------ 51| 70| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (51:7): [True: 0, False: 70] ------------------ 52| 70| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 53| | 54| 70| if (!mRaw->dim.hasPositiveArea() || mRaw->dim.x % 4 != 0 || ------------------ | Branch (54:7): [True: 0, False: 70] | Branch (54:39): [True: 1, False: 69] ------------------ 55| 69| mRaw->dim.x > 4516 || mRaw->dim.y > 3012) ------------------ | Branch (55:7): [True: 0, False: 69] | Branch (55:29): [True: 0, False: 69] ------------------ 56| 70| ThrowRDE("Unexpected image dimensions found: (%d; %d)", mRaw->dim.x, ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 57| 69| mRaw->dim.y); 58| | 59| 69| if (bps != 10 && bps != 12) ------------------ | Branch (59:7): [True: 1, False: 68] | Branch (59:20): [True: 0, False: 1] ------------------ 60| 69| ThrowRDE("Unexpected bits per sample: %i", bps); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 61| | 62| | // Lower estimate: this decompressor requires *at least* half a byte 63| | // per output pixel 64| 69| (void)input.check(implicit_cast(mRaw->dim.area() / 2ULL)); 65| 69|} _ZN8rawspeed17KodakDecompressor13decodeSegmentEj: 68| 1.21k|KodakDecompressor::decodeSegment(const uint32_t bsize) { 69| 1.21k| invariant(bsize > 0); ------------------ | | 27| 1.21k|#define invariant(expr) assert(expr) ------------------ | Branch (69:3): [True: 1.21k, False: 0] ------------------ 70| 1.21k| invariant(bsize % 4 == 0); ------------------ | | 27| 1.21k|#define invariant(expr) assert(expr) ------------------ | Branch (70:3): [True: 1.21k, False: 0] ------------------ 71| 1.21k| invariant(bsize <= segment_size); ------------------ | | 27| 1.21k|#define invariant(expr) assert(expr) ------------------ | Branch (71:3): [True: 1.21k, False: 0] ------------------ 72| | 73| 1.21k| segment out; 74| 1.21k| static_assert(out.size() == segment_size, "Wrong segment size"); 75| | 76| |#if __has_feature(address_sanitizer) || defined(__SANITIZE_ADDRESS__) 77| | // We are to produce only bsize pixels. 78| | __sanitizer_annotate_contiguous_container(out.begin(), out.end(), out.end(), 79| | &out[bsize]); 80| |#endif 81| | 82| 1.21k| std::array blen; 83| 1.21k| uint64_t bitbuf = 0; 84| 1.21k| uint32_t bits = 0; 85| | 86| 68.2k| for (uint32_t i = 0; i < bsize; i += 2) { ------------------ | Branch (86:24): [True: 67.0k, False: 1.21k] ------------------ 87| | // One byte per two pixels 88| 67.0k| blen[i] = input.peekByte() & 15; 89| 67.0k| blen[i + 1] = input.getByte() >> 4; 90| 67.0k| } 91| 1.21k| if ((bsize & 7) == 4) { ------------------ | Branch (91:7): [True: 495, False: 720] ------------------ 92| 495| bitbuf = (static_cast(input.getByte())) << 8UL; 93| 495| bitbuf += (static_cast(input.getByte())); 94| 495| bits = 16; 95| 495| } 96| 134k| for (uint32_t i = 0; i < bsize; i++) { ------------------ | Branch (96:24): [True: 133k, False: 1.21k] ------------------ 97| 133k| uint32_t len = blen[i]; 98| 133k| invariant(len < 16); ------------------ | | 27| 133k|#define invariant(expr) assert(expr) ------------------ | Branch (98:5): [True: 133k, False: 0] ------------------ 99| | 100| 133k| if (bits < len) { ------------------ | Branch (100:9): [True: 1.16k, False: 132k] ------------------ 101| 5.79k| for (uint32_t j = 0; j < 32; j += 8) { ------------------ | Branch (101:28): [True: 4.63k, False: 1.16k] ------------------ 102| 4.63k| bitbuf += static_cast(static_cast(input.getByte())) 103| 4.63k| << (bits + (j ^ 8)); 104| 4.63k| } 105| 1.16k| bits += 32; 106| 1.16k| } 107| | 108| 133k| uint32_t diff = static_cast(bitbuf) & 109| 133k| extractHighBits(0xffffU, len, /*effectiveBitwidth=*/16); 110| 133k| bitbuf >>= len; 111| 133k| bits -= len; 112| | 113| 133k| out[i] = implicit_cast( 114| 133k| len != 0 ? PrefixCodeDecoder<>::extend(diff, len) : int(diff)); ------------------ | Branch (114:9): [True: 4.97k, False: 128k] ------------------ 115| 133k| } 116| | 117| 1.21k| return out; 118| 1.21k|} _ZN8rawspeed17KodakDecompressor10decompressEv: 120| 67|void KodakDecompressor::decompress() { 121| 67| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 122| | 123| 67| uint32_t random = 0; 124| 1.12k| for (int row = 0; row < out.height(); row++) { ------------------ | Branch (124:21): [True: 1.11k, False: 8] ------------------ 125| 2.27k| for (int col = 0; col < out.width();) { ------------------ | Branch (125:23): [True: 1.21k, False: 1.05k] ------------------ 126| 1.21k| const int len = std::min(segment_size, mRaw->dim.x - col); 127| | 128| 1.21k| const segment buf = decodeSegment(len); 129| | 130| 1.21k| std::array pred; 131| 1.21k| pred.fill(0); 132| | 133| 126k| for (int i = 0; i < len; ++i, ++col) { ------------------ | Branch (133:23): [True: 125k, False: 1.15k] ------------------ 134| 125k| pred[i & 1] += buf[i]; 135| | 136| 125k| int value = pred[i & 1]; 137| 125k| if (!isIntN(value, bps)) ------------------ | Branch (137:13): [True: 59, False: 125k] ------------------ 138| 125k| ThrowRDE("Value out of bounds %d (bps = %i)", value, bps); ------------------ | | 38| 59| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 59| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 59| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 59| __VA_ARGS__) | | ------------------ ------------------ 139| | 140| 125k| if (uncorrectedRawValues) ------------------ | Branch (140:13): [True: 0, False: 125k] ------------------ 141| 0| out(row, col) = implicit_cast(value); 142| 125k| else { 143| 125k| mRaw->setWithLookUp(implicit_cast(value), 144| 125k| reinterpret_cast(&out(row, col)), 145| 125k| &random); 146| 125k| } 147| 125k| } 148| 1.21k| } 149| 1.11k| } 150| 67|} _ZN8rawspeed12LJpegDecoderC2ENS_10ByteStreamERKNS_8RawImageE: 44| 952| : AbstractLJpegDecoder(bs, img) { 45| 952| if (mRaw->getDataType() != RawImageType::UINT16) ------------------ | Branch (45:7): [True: 0, False: 952] ------------------ 46| 952| ThrowRDE("Unexpected data type (%u)", ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 47| 952| static_cast(mRaw->getDataType())); 48| | 49| 952| if ((mRaw->getCpp() != 1 || mRaw->getBpp() != sizeof(uint16_t)) && ------------------ | Branch (49:8): [True: 624, False: 328] | Branch (49:31): [True: 0, False: 328] ------------------ 50| 616| (mRaw->getCpp() != 2 || mRaw->getBpp() != 2 * sizeof(uint16_t)) && ------------------ | Branch (50:8): [True: 480, False: 136] | Branch (50:31): [True: 0, False: 136] ------------------ 51| 475| (mRaw->getCpp() != 3 || mRaw->getBpp() != 3 * sizeof(uint16_t))) ------------------ | Branch (51:8): [True: 17, False: 458] | Branch (51:31): [True: 3, False: 455] ------------------ 52| 952| ThrowRDE("Unexpected component count (%u)", mRaw->getCpp()); ------------------ | | 38| 10| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 10| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 10| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 10| __VA_ARGS__) | | ------------------ ------------------ 53| | 54| 942| if (!mRaw->dim.hasPositiveArea()) ------------------ | Branch (54:7): [True: 0, False: 942] ------------------ 55| 942| ThrowRDE("Image has zero size"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 56| | 57| 942|#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION 58| | // Yeah, sure, here it would be just dumb to leave this for production :) 59| 942| if (mRaw->dim.x > 9728 || mRaw->dim.y > 6656) { ------------------ | Branch (59:7): [True: 22, False: 920] | Branch (59:29): [True: 13, False: 907] ------------------ 60| 11| ThrowRDE("Unexpected image dimensions found: (%i; %i)", mRaw->dim.x, ------------------ | | 38| 11| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 11| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 11| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 11| __VA_ARGS__) | | ------------------ ------------------ 61| 11| mRaw->dim.y); 62| 11| } 63| 942|#endif 64| 942|} _ZN8rawspeed12LJpegDecoder6decodeEjjjjNS_8iPoint2DEb: 68| 909| bool fixDng16Bug_) { 69| 909| if (offsetX >= static_cast(mRaw->dim.x)) ------------------ | Branch (69:7): [True: 0, False: 909] ------------------ 70| 909| ThrowRDE("X offset outside of image"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 71| 909| if (offsetY >= static_cast(mRaw->dim.y)) ------------------ | Branch (71:7): [True: 0, False: 909] ------------------ 72| 909| ThrowRDE("Y offset outside of image"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 73| | 74| 909| if (width > static_cast(mRaw->dim.x)) ------------------ | Branch (74:7): [True: 0, False: 909] ------------------ 75| 909| ThrowRDE("Tile wider than image"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 76| 909| if (height > static_cast(mRaw->dim.y)) ------------------ | Branch (76:7): [True: 0, False: 909] ------------------ 77| 909| ThrowRDE("Tile taller than image"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 78| | 79| 909| if (offsetX + width > static_cast(mRaw->dim.x)) ------------------ | Branch (79:7): [True: 0, False: 909] ------------------ 80| 909| ThrowRDE("Tile overflows image horizontally"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 81| 909| if (offsetY + height > static_cast(mRaw->dim.y)) ------------------ | Branch (81:7): [True: 0, False: 909] ------------------ 82| 909| ThrowRDE("Tile overflows image vertically"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 83| | 84| 914| if (width == 0 || height == 0) ------------------ | Branch (84:7): [True: 18.4E, False: 914] | Branch (84:21): [True: 18.4E, False: 915] ------------------ 85| 0| return; // We do not need anything from this tile. 86| | 87| 909| if (!maxDim_.hasPositiveArea() || ------------------ | Branch (87:7): [True: 34, False: 875] ------------------ 88| 875| implicit_cast(maxDim_.x) < width || ------------------ | Branch (88:7): [True: 3, False: 872] ------------------ 89| 873| implicit_cast(maxDim_.y) < height) ------------------ | Branch (89:7): [True: 2, False: 871] ------------------ 90| 909| ThrowRDE("Requested tile is larger than tile's maximal dimensions"); ------------------ | | 38| 44| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 44| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 44| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 44| __VA_ARGS__) | | ------------------ ------------------ 91| | 92| 865| offX = offsetX; 93| 865| offY = offsetY; 94| 865| w = width; 95| 865| h = height; 96| | 97| 865| maxDim = maxDim_; 98| | 99| 865| fixDng16Bug = fixDng16Bug_; 100| | 101| 865| AbstractLJpegDecoder::decodeSOI(); 102| 865|} _ZN8rawspeed12LJpegDecoder10decodeScanEv: 104| 414|Buffer::size_type LJpegDecoder::decodeScan() { 105| 414| invariant(frame.cps > 0); ------------------ | | 27| 414|#define invariant(expr) assert(expr) ------------------ | Branch (105:3): [True: 414, False: 0] ------------------ 106| | 107| 414| if (predictorMode != 1) ------------------ | Branch (107:7): [True: 4, False: 410] ------------------ 108| 414| ThrowRDE("Unsupported predictor mode: %u", predictorMode); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 109| | 110| 1.21k| for (uint32_t i = 0; i < frame.cps; i++) ------------------ | Branch (110:24): [True: 804, False: 407] ------------------ 111| 804| if (frame.compInfo[i].superH != 1 || frame.compInfo[i].superV != 1) ------------------ | Branch (111:9): [True: 2, False: 802] | Branch (111:42): [True: 1, False: 801] ------------------ 112| 410| ThrowRDE("Unsupported subsampling"); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 413| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 413| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 413| __VA_ARGS__) | | ------------------ ------------------ 113| | 114| 407| int N_COMP = frame.cps; 115| | 116| 407| std::vector rec; 117| 407| rec.reserve(N_COMP); 118| 407| std::generate_n(std::back_inserter(rec), N_COMP, 119| 407| [&rec, hts = getPrefixCodeDecoders(N_COMP), 120| 407| initPred = getInitialPredictors( 121| 407| N_COMP)]() -> LJpegDecompressor::PerComponentRecipe { 122| 407| const auto i = implicit_cast(rec.size()); 123| 407| return {*hts[i], initPred[i]}; 124| 407| }); 125| | 126| 407| const iRectangle2D imgFrame = { 127| 407| {static_cast(offX), static_cast(offY)}, 128| 407| {static_cast(w), static_cast(h)}}; 129| 407| const auto jpegFrameDim = iPoint2D(frame.w, frame.h); 130| | 131| 407| if (implicit_cast(maxDim.x) * implicit_cast(mRaw->getCpp()) > ------------------ | Branch (131:7): [True: 0, False: 407] ------------------ 132| 407| std::numeric_limits::max()) 133| 407| ThrowRDE("Maximal output tile is too large"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 134| | 135| 407| auto maxRes = 136| 407| iPoint2D(implicit_cast(mRaw->getCpp()) * maxDim.x, maxDim.y); 137| 407| if (maxRes.area() != N_COMP * jpegFrameDim.area()) ------------------ | Branch (137:7): [True: 42, False: 365] ------------------ 138| 407| ThrowRDE("LJpeg frame area does not match maximal tile area"); ------------------ | | 38| 42| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 42| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 42| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 42| __VA_ARGS__) | | ------------------ ------------------ 139| | 140| 365| if (maxRes.x % jpegFrameDim.x != 0 || maxRes.y % jpegFrameDim.y != 0) ------------------ | Branch (140:7): [True: 4, False: 361] | Branch (140:41): [True: 1, False: 360] ------------------ 141| 365| ThrowRDE("Maximal output tile size is not a multiple of LJpeg frame size"); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 142| | 143| 363| auto MCUSize = iPoint2D{maxRes.x / jpegFrameDim.x, maxRes.y / jpegFrameDim.y}; 144| 363| if (MCUSize.area() != implicit_cast(N_COMP)) ------------------ | Branch (144:7): [True: 0, False: 363] ------------------ 145| 363| ThrowRDE("Unexpected MCU size, does not match LJpeg component count"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 146| | 147| 363| const LJpegDecompressor::Frame jpegFrame = {MCUSize, jpegFrameDim}; 148| | 149| 363| int numLJpegRowsPerRestartInterval; 150| 363| if (numMCUsPerRestartInterval == 0) { ------------------ | Branch (150:7): [True: 348, False: 15] ------------------ 151| | // Restart interval not enabled, so all of the rows 152| | // are contained in the first (implicit) restart interval. 153| 348| numLJpegRowsPerRestartInterval = jpegFrameDim.y; 154| 348| } else { 155| 15| const int numMCUsPerRow = jpegFrameDim.x; 156| 15| if (numMCUsPerRestartInterval % numMCUsPerRow != 0) ------------------ | Branch (156:9): [True: 1, False: 14] ------------------ 157| 15| ThrowRDE("Restart interval is not a multiple of frame row size"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 158| 14| numLJpegRowsPerRestartInterval = numMCUsPerRestartInterval / numMCUsPerRow; 159| 14| } 160| | 161| 362| LJpegDecompressor d(mRaw, imgFrame, jpegFrame, rec, 162| 362| numLJpegRowsPerRestartInterval, 163| 362| input.peekRemainingBuffer().getAsArray1DRef()); 164| 362| return d.decode(); 165| 363|} LJpegDecoder.cpp:_ZZN8rawspeed12LJpegDecoder10decodeScanEvENK3$_0clEv: 121| 790| N_COMP)]() -> LJpegDecompressor::PerComponentRecipe { 122| 790| const auto i = implicit_cast(rec.size()); 123| 790| return {*hts[i], initPred[i]}; 124| 790| }); _ZN8rawspeed17LJpegDecompressorC2ENS_8RawImageENS_12iRectangle2DENS0_5FrameENSt3__16vectorINS0_18PerComponentRecipeENS4_9allocatorIS6_EEEEiNS_10Array1DRefIKhEE: 57| 360| : mRaw(std::move(img)), input(input_), imgFrame(imgFrame_), 58| 360| frame(std::move(frame_)), rec(std::move(rec_)), 59| 360| numLJpegRowsPerRestartInterval(numLJpegRowsPerRestartInterval_) { 60| | 61| 360| if (mRaw->getDataType() != RawImageType::UINT16) ------------------ | Branch (61:7): [True: 0, False: 360] ------------------ 62| 360| ThrowRDE("Unexpected data type (%u)", ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 63| 360| static_cast(mRaw->getDataType())); 64| | 65| 360| if ((mRaw->getCpp() != 1 || mRaw->getBpp() != sizeof(uint16_t)) && ------------------ | Branch (65:8): [True: 130, False: 230] | Branch (65:31): [True: 0, False: 230] ------------------ 66| 130| (mRaw->getCpp() != 2 || mRaw->getBpp() != 2 * sizeof(uint16_t)) && ------------------ | Branch (66:8): [True: 128, False: 2] | Branch (66:31): [True: 0, False: 2] ------------------ 67| 127| (mRaw->getCpp() != 3 || mRaw->getBpp() != 3 * sizeof(uint16_t))) ------------------ | Branch (67:8): [True: 0, False: 127] | Branch (67:31): [True: 0, False: 127] ------------------ 68| 360| ThrowRDE("Unexpected component count (%u)", mRaw->getCpp()); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 69| | 70| 360| if (!mRaw->dim.hasPositiveArea()) ------------------ | Branch (70:7): [True: 0, False: 360] ------------------ 71| 360| ThrowRDE("Image has zero size"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 72| | 73| 360| if (!imgFrame.hasPositiveArea()) ------------------ | Branch (73:7): [True: 0, False: 360] ------------------ 74| 360| ThrowRDE("Tile has zero size"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 75| | 76| 360|#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION 77| | // Yeah, sure, here it would be just dumb to leave this for production :) 78| 360| if (mRaw->dim.x > 9728 || mRaw->dim.y > 6656) { ------------------ | Branch (78:7): [True: 1, False: 359] | Branch (78:29): [True: 0, False: 359] ------------------ 79| 0| ThrowRDE("Unexpected image dimensions found: (%i; %i)", mRaw->dim.x, ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 80| 0| mRaw->dim.y); 81| 0| } 82| 360|#endif 83| | 84| 360| if (imgFrame.pos.x >= mRaw->dim.x) ------------------ | Branch (84:7): [True: 0, False: 360] ------------------ 85| 360| ThrowRDE("X offset outside of image"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 86| 360| if (imgFrame.pos.y >= mRaw->dim.y) ------------------ | Branch (86:7): [True: 0, False: 360] ------------------ 87| 360| ThrowRDE("Y offset outside of image"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 88| | 89| 360| if (imgFrame.dim.x > mRaw->dim.x) ------------------ | Branch (89:7): [True: 0, False: 360] ------------------ 90| 360| ThrowRDE("Tile wider than image"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 91| 360| if (imgFrame.dim.y > mRaw->dim.y) ------------------ | Branch (91:7): [True: 0, False: 360] ------------------ 92| 360| ThrowRDE("Tile taller than image"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 93| | 94| 360| if (imgFrame.pos.x + imgFrame.dim.x > mRaw->dim.x) ------------------ | Branch (94:7): [True: 0, False: 360] ------------------ 95| 360| ThrowRDE("Tile overflows image horizontally"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 96| 360| if (imgFrame.pos.y + imgFrame.dim.y > mRaw->dim.y) ------------------ | Branch (96:7): [True: 0, False: 360] ------------------ 97| 360| ThrowRDE("Tile overflows image vertically"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 98| | 99| 360| if (!frame.dim.hasPositiveArea()) ------------------ | Branch (99:7): [True: 0, False: 360] ------------------ 100| 360| ThrowRDE("Frame has zero size"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 101| | 102| 360| if (iPoint2D{1, 1} != frame.mcu && iPoint2D{2, 1} != frame.mcu && ------------------ | Branch (102:7): [True: 222, False: 138] | Branch (102:7): [True: 1, False: 359] | Branch (102:38): [True: 129, False: 93] ------------------ 103| 129| iPoint2D{3, 1} != frame.mcu && iPoint2D{4, 1} != frame.mcu && ------------------ | Branch (103:7): [True: 1, False: 128] | Branch (103:38): [True: 1, False: 0] ------------------ 104| 1| iPoint2D{2, 2} != frame.mcu) ------------------ | Branch (104:7): [True: 1, False: 0] ------------------ 105| 360| ThrowRDE("Unexpected MCU size: {%i, %i}", frame.mcu.x, frame.mcu.y); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 106| | 107| 359| if (rec.size() != static_cast(frame.mcu.area())) ------------------ | Branch (107:7): [True: 0, False: 359] ------------------ 108| 359| ThrowRDE("Must have exactly one recepie per component"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 109| | 110| 706| for (const auto& recip : rec) { ------------------ | Branch (110:26): [True: 706, False: 359] ------------------ 111| 706| if (!recip.ht.isFullDecode()) ------------------ | Branch (111:9): [True: 0, False: 706] ------------------ 112| 706| ThrowRDE("Huffman table is not of a full decoding variety"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 113| 706| } 114| | 115| 359| if (numLJpegRowsPerRestartInterval < 1) ------------------ | Branch (115:7): [True: 0, False: 359] ------------------ 116| 359| ThrowRDE("Number of rows per restart interval must be positives"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 117| | 118| 359| if (static_cast(frame.mcu.x) * frame.dim.x > ------------------ | Branch (118:7): [True: 0, False: 359] ------------------ 119| 359| std::numeric_limits::max() || 120| 359| static_cast(frame.mcu.y) * frame.dim.y > ------------------ | Branch (120:7): [True: 0, False: 359] ------------------ 121| 359| std::numeric_limits::max()) 122| 359| ThrowRDE("LJpeg frame is too big"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 123| | 124| 359| if (static_cast(mRaw->getCpp()) * imgFrame.dim.x > ------------------ | Branch (124:7): [True: 0, False: 359] ------------------ 125| 359| std::numeric_limits::max()) 126| 359| ThrowRDE("Img frame is too big"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 127| | 128| 359| if (imgFrame.dim.x < frame.mcu.x || imgFrame.dim.y < frame.mcu.y) ------------------ | Branch (128:7): [True: 0, False: 359] | Branch (128:39): [True: 0, False: 359] ------------------ 129| 359| ThrowRDE("Tile size is smaller than a single frame MCU"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 130| | 131| 359| if (imgFrame.dim.y % frame.mcu.y != 0) ------------------ | Branch (131:7): [True: 0, False: 359] ------------------ 132| 359| ThrowRDE("Output row count is not a multiple of MCU row count"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 133| | 134| 359| const int tileRequiredWidth = 135| 359| static_cast(mRaw->getCpp()) * imgFrame.dim.x; 136| | 137| | // How many full pixel MCUs do we need to consume for that? 138| 359| if (const auto mcusToConsume = implicit_cast( 139| 359| roundUpDivisionSafe(tileRequiredWidth, frame.mcu.x)); 140| 359| frame.dim.x < mcusToConsume || ------------------ | Branch (140:7): [True: 1, False: 358] ------------------ 141| 358| frame.mcu.y * frame.dim.y < imgFrame.dim.y || ------------------ | Branch (141:7): [True: 18.4E, False: 359] ------------------ 142| 359| frame.mcu.x * frame.dim.x < tileRequiredWidth) { ------------------ | Branch (142:7): [True: 0, False: 359] ------------------ 143| 0| ThrowRDE("LJpeg frame (%d, %d) is smaller than expected (%d, %d)", ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 144| 0| frame.mcu.x * frame.dim.x, frame.mcu.y * frame.dim.y, 145| 0| tileRequiredWidth, imgFrame.dim.y); 146| 0| } 147| | 148| | // How many full pixel MCUs will we produce? 149| 359| numFullMCUs = tileRequiredWidth / frame.mcu.x; // Truncating division! 150| | // Do we need to also produce part of a MCU? 151| 359| trailingPixels = tileRequiredWidth % frame.mcu.x; 152| 359|} _ZNK8rawspeed17LJpegDecompressor6decodeEv: 341| 359|ByteStream::size_type LJpegDecompressor::decode() const { 342| 359| switch (frame.mcu.area()) { 343| 138| case 1: ------------------ | Branch (343:3): [True: 138, False: 221] ------------------ 344| 138| if (frame.mcu == MCU<1, 1>) { ------------------ | Branch (344:9): [True: 138, False: 0] ------------------ 345| 138| return decodeN>(); 346| 138| } 347| 0| break; 348| 93| case 2: ------------------ | Branch (348:3): [True: 93, False: 266] ------------------ 349| 93| if (frame.mcu == MCU<2, 1>) { ------------------ | Branch (349:9): [True: 93, False: 0] ------------------ 350| 93| return decodeN>(); 351| 93| } 352| 0| break; 353| 127| case 3: ------------------ | Branch (353:3): [True: 127, False: 232] ------------------ 354| 127| if (frame.mcu == MCU<3, 1>) { ------------------ | Branch (354:9): [True: 127, False: 0] ------------------ 355| 127| return decodeN>(); 356| 127| } 357| 0| break; 358| 0| case 4: ------------------ | Branch (358:3): [True: 0, False: 359] ------------------ 359| 0| if (frame.mcu == MCU<4, 1>) { ------------------ | Branch (359:9): [True: 0, False: 0] ------------------ 360| 0| return decodeN>(); 361| 0| } 362| 0| if (frame.mcu == MCU<2, 2>) { ------------------ | Branch (362:9): [True: 0, False: 0] ------------------ 363| 0| return decodeN>(); 364| 0| } 365| 0| break; 366| 0| default: ------------------ | Branch (366:3): [True: 0, False: 359] ------------------ 367| 0| __builtin_unreachable(); 368| 359| } 369| 0| __builtin_unreachable(); 370| 359|} LJpegDecompressor.cpp:_ZNK8rawspeed17LJpegDecompressor7decodeNIL_ZNS_12_GLOBAL__N_13MCUILi1ELi1EEEEEEjv: 255| 138|ByteStream::size_type LJpegDecompressor::decodeN() const { 256| 138| invariant(MCU == this->frame.mcu); ------------------ | | 27| 138|#define invariant(expr) assert(expr) ------------------ | Branch (256:3): [True: 138, False: 0] ------------------ 257| | 258| 138| invariant(MCU.hasPositiveArea()); ------------------ | | 27| 138|#define invariant(expr) assert(expr) ------------------ | Branch (258:3): [True: 138, False: 0] ------------------ 259| | // FIXME: workarounding lack of constexpr std::abs() :( 260| 138| constexpr int N_COMP = MCU.x * MCU.y; 261| | 262| 138| invariant(mRaw->getCpp() > 0); ------------------ | | 27| 138|#define invariant(expr) assert(expr) ------------------ | Branch (262:3): [True: 138, False: 0] ------------------ 263| | 264| 138| const auto img = 265| 138| CroppedArray2DRef(mRaw->getU16DataAsUncroppedArray2DRef(), 266| 138| mRaw->getCpp() * imgFrame.pos.x, imgFrame.pos.y, 267| 138| mRaw->getCpp() * imgFrame.dim.x, imgFrame.dim.y) 268| 138| .getAsArray2DRef(); 269| | 270| 138| const auto ht = getPrefixCodeDecoders(); 271| | 272| | // A recoded DNG might be split up into tiles of self contained LJpeg blobs. 273| | // The tiles at the bottom and the right may extend beyond the dimension of 274| | // the raw image buffer. The excessive content has to be ignored. 275| | 276| 138| invariant(imgFrame.dim.y % frame.mcu.y == 0); ------------------ | | 27| 138|#define invariant(expr) assert(expr) ------------------ | Branch (276:3): [True: 138, False: 0] ------------------ 277| 138| const auto numRestartIntervals = implicit_cast(roundUpDivisionSafe( 278| 138| imgFrame.dim.y / frame.mcu.y, numLJpegRowsPerRestartInterval)); 279| 138| invariant(numRestartIntervals >= 0); ------------------ | | 27| 138|#define invariant(expr) assert(expr) ------------------ | Branch (279:3): [True: 138, False: 0] ------------------ 280| 138| invariant(numRestartIntervals != 0); ------------------ | | 27| 138|#define invariant(expr) assert(expr) ------------------ | Branch (280:3): [True: 138, False: 0] ------------------ 281| | 282| 138| ByteStream inputStream(DataBuffer(input, Endianness::little)); 283| 138| for (int restartIntervalIndex = 0; 284| 282| restartIntervalIndex != numRestartIntervals; ++restartIntervalIndex) { ------------------ | Branch (284:8): [True: 150, False: 132] ------------------ 285| 150| auto predStorage = getInitialPreds(); 286| 150| auto pred = Array2DRef(predStorage.data(), MCU.x, MCU.y); 287| | 288| 150| if (restartIntervalIndex != 0) { ------------------ | Branch (288:9): [True: 12, False: 138] ------------------ 289| 12| auto marker = peekMarker(inputStream); 290| 12| if (!marker) // FIXME: can there be padding bytes before the marker? ------------------ | Branch (290:11): [True: 4, False: 8] ------------------ 291| 12| ThrowRDE("Jpeg marker not encountered"); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 292| 8| Optional number = getRestartMarkerNumber(*marker); 293| 8| if (!number) ------------------ | Branch (293:11): [True: 1, False: 7] ------------------ 294| 8| ThrowRDE("Not a restart marker!"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 295| 7| if (*number != ((restartIntervalIndex - 1) % 8)) ------------------ | Branch (295:11): [True: 1, False: 6] ------------------ 296| 7| ThrowRDE("Unexpected restart marker found"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 297| 6| inputStream.skipBytes(2); // Good restart marker. 298| 6| } 299| | 300| 144| BitStreamerJPEG bs(inputStream.peekRemainingBuffer().getAsArray1DRef()); 301| | 302| 144| for (int ljpegRowOfRestartInterval = 0; 303| 10.7k| ljpegRowOfRestartInterval != numLJpegRowsPerRestartInterval; ------------------ | Branch (303:10): [True: 10.6k, False: 143] ------------------ 304| 10.6k| ++ljpegRowOfRestartInterval) { 305| 10.6k| const int row = 306| 10.6k| frame.mcu.y * (numLJpegRowsPerRestartInterval * restartIntervalIndex + 307| 10.6k| ljpegRowOfRestartInterval); 308| 10.6k| invariant(row >= 0); ------------------ | | 27| 10.6k|#define invariant(expr) assert(expr) ------------------ | Branch (308:7): [True: 10.6k, False: 0] ------------------ 309| 10.6k| invariant(row <= imgFrame.dim.y); ------------------ | | 27| 10.6k|#define invariant(expr) assert(expr) ------------------ | Branch (309:7): [True: 10.6k, False: 0] ------------------ 310| | 311| | // For y, we can simply stop decoding when we reached the border. 312| 10.6k| if (row == imgFrame.dim.y) { ------------------ | Branch (312:11): [True: 1, False: 10.6k] ------------------ 313| 1| invariant((restartIntervalIndex + 1) == numRestartIntervals); ------------------ | | 27| 1|#define invariant(expr) assert(expr) ------------------ | Branch (313:9): [True: 1, False: 0] ------------------ 314| 1| break; 315| 1| } 316| | 317| 10.6k| const auto outStripe = CroppedArray2DRef(img, 318| 10.6k| /*offsetCols=*/0, 319| 10.6k| /*offsetRows=*/row, 320| 10.6k| /*croppedWidth=*/img.width(), 321| 10.6k| /*croppedHeight=*/frame.mcu.y) 322| 10.6k| .getAsArray2DRef(); 323| | 324| 10.6k| decodeRowN(outStripe, pred, ht, bs); 325| | 326| | // The predictor for the next line is the start of this line. 327| 10.6k| pred = CroppedArray2DRef(outStripe, 328| 10.6k| /*offsetCols=*/0, 329| 10.6k| /*offsetRows=*/0, 330| 10.6k| /*croppedWidth=*/MCU.x, 331| 10.6k| /*croppedHeight=*/MCU.y) 332| 10.6k| .getAsArray2DRef(); 333| 10.6k| } 334| | 335| 144| inputStream.skipBytes(bs.getStreamPosition()); 336| 144| } 337| | 338| 132| return inputStream.getPosition(); 339| 138|} _ZNK8rawspeed17LJpegDecompressor21getPrefixCodeDecodersILi1EEENSt3__15arrayINS2_17reference_wrapperIKNS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS6_EEEEEEXT_EEEv: 164| 138|LJpegDecompressor::getPrefixCodeDecoders() const { 165| 138| return getPrefixCodeDecodersImpl(std::make_index_sequence{}); 166| 138|} _ZNK8rawspeed17LJpegDecompressor25getPrefixCodeDecodersImplILi1EJLm0EEEENSt3__15arrayINS2_17reference_wrapperIKNS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS6_EEEEEEXT_EEENS2_16integer_sequenceImJXspT0_EEEE: 157| 138| std::index_sequence /*unused*/) const { 158| 138| return std::array>, N_COMP>{ 159| 138| std::cref(rec[I].ht)...}; 160| 138|} _ZNK8rawspeed17LJpegDecompressor15getInitialPredsILi1EEENSt3__15arrayItXT_EEEv: 169| 150|std::array LJpegDecompressor::getInitialPreds() const { 170| 150| std::array preds; 171| 150| std::transform( 172| 150| rec.begin(), rec.end(), preds.begin(), 173| 150| [](const PerComponentRecipe& compRec) { return compRec.initPred; }); 174| 150| return preds; 175| 150|} _ZZNK8rawspeed17LJpegDecompressor15getInitialPredsILi1EEENSt3__15arrayItXT_EEEvENKUlRKNS0_18PerComponentRecipeEE_clES7_: 173| 150| [](const PerComponentRecipe& compRec) { return compRec.initPred; }); LJpegDecompressor.cpp:_ZNK8rawspeed17LJpegDecompressor10decodeRowNIL_ZNS_12_GLOBAL__N_13MCUILi1ELi1EEEELi1EEEvNS_10Array2DRefItEENS4_IKtEENSt3__15arrayINS8_17reference_wrapperIKNS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderISC_EEEEEEXT0_EEERNS_15BitStreamerJPEGE: 188| 10.6k| BitStreamerJPEG& bs) const { 189| 10.6k| invariant(MCUSize.area() == N_COMP); ------------------ | | 27| 10.6k|#define invariant(expr) assert(expr) ------------------ | Branch (189:3): [True: 10.6k, False: 0] ------------------ 190| 10.6k| invariant(outStripe.width() >= MCUSize.x); ------------------ | | 27| 10.6k|#define invariant(expr) assert(expr) ------------------ | Branch (190:3): [True: 10.6k, False: 0] ------------------ 191| 10.6k| invariant(outStripe.height() == MCUSize.y); ------------------ | | 27| 10.6k|#define invariant(expr) assert(expr) ------------------ | Branch (191:3): [True: 10.6k, False: 0] ------------------ 192| 10.6k| invariant(pred.width() == MCUSize.x); ------------------ | | 27| 10.6k|#define invariant(expr) assert(expr) ------------------ | Branch (192:3): [True: 10.6k, False: 0] ------------------ 193| 10.6k| invariant(pred.height() == MCUSize.y); ------------------ | | 27| 10.6k|#define invariant(expr) assert(expr) ------------------ | Branch (193:3): [True: 10.6k, False: 0] ------------------ 194| | 195| | // FIXME: predictor may have value outside of the uint16_t. 196| | // https://github.com/darktable-org/rawspeed/issues/175 197| | 198| 10.6k| int mcuIdx = 0; 199| | // For x, we first process all full pixel MCUs within the image buffer ... 200| 26.4k| for (; mcuIdx < numFullMCUs; ++mcuIdx) { ------------------ | Branch (200:10): [True: 15.8k, False: 10.6k] ------------------ 201| 15.8k| const auto outTile = CroppedArray2DRef(outStripe, 202| 15.8k| /*offsetCols=*/MCUSize.x * mcuIdx, 203| 15.8k| /*offsetRows=*/0, 204| 15.8k| /*croppedWidth=*/MCUSize.x, 205| 15.8k| /*croppedHeight=*/MCUSize.y) 206| 15.8k| .getAsArray2DRef(); 207| 31.6k| for (int MCURow = 0; MCURow != MCUSize.y; ++MCURow) { ------------------ | Branch (207:26): [True: 15.8k, False: 15.8k] ------------------ 208| 31.6k| for (int MCUСol = 0; MCUСol != MCUSize.x; ++MCUСol) { ------------------ | Branch (208:29): [True: 15.8k, False: 15.8k] ------------------ 209| 15.8k| int c = (MCUSize.x * MCURow) + MCUСol; 210| 15.8k| int prediction = pred(MCURow, MCUСol); 211| 15.8k| int diff = (static_cast&>(ht[c])) 212| 15.8k| .decodeDifference(bs); 213| 15.8k| int pix = prediction + diff; 214| 15.8k| outTile(MCURow, MCUСol) = uint16_t(pix); 215| 15.8k| } 216| 15.8k| } 217| | // The predictor for the next MCU of the row is the just-decoded MCU. 218| 15.8k| pred = outTile; 219| 15.8k| } 220| | 221| | // Sometimes we also need to consume one more MCU, and produce part of it. 222| 10.6k| if (trailingPixels != 0) { ------------------ | Branch (222:7): [True: 0, False: 10.6k] ------------------ 223| | // Some rather esoteric DNG's have odd dimensions, e.g. width % 2 = 1. 224| | // We may end up needing just part of last N_COMP pixels. 225| 0| invariant(trailingPixels > 0); ------------------ | | 27| 0|#define invariant(expr) assert(expr) ------------------ | Branch (225:5): [True: 0, False: 0] ------------------ 226| 0| invariant(trailingPixels < N_COMP); ------------------ | | 27| 0|#define invariant(expr) assert(expr) ------------------ | Branch (226:5): [True: 0, False: 0] ------------------ 227| 0| invariant(N_COMP > 1 && "can't want part of 1-pixel-wide block"); ------------------ | | 27| 0|#define invariant(expr) assert(expr) ------------------ | Branch (227:5): [Folded, False: 0] | Branch (227:5): [True: 0, Folded] | Branch (227:5): [Folded, False: 0] ------------------ 228| | // Some rather esoteric DNG's have odd dimensions, e.g. width % 2 = 1. 229| | // We may end up needing just part of last N_COMP pixels. 230| 0| for (int MCURow = 0; MCURow != MCUSize.y; ++MCURow) { ------------------ | Branch (230:26): [True: 0, False: 0] ------------------ 231| 0| for (int MCUСol = 0; MCUСol != MCUSize.x; ++MCUСol) { ------------------ | Branch (231:29): [True: 0, False: 0] ------------------ 232| 0| int c = (MCUSize.x * MCURow) + MCUСol; 233| 0| int prediction = pred(MCURow, MCUСol); 234| 0| int diff = (static_cast&>(ht[c])) 235| 0| .decodeDifference(bs); 236| 0| int pix = prediction + diff; 237| 0| int stripeRow = MCURow; 238| 0| int stripeCol = (MCUSize.x * mcuIdx) + MCUСol; 239| 0| if (stripeCol < outStripe.width()) ------------------ | Branch (239:13): [True: 0, False: 0] ------------------ 240| 0| outStripe(stripeRow, stripeCol) = uint16_t(pix); 241| 0| } 242| 0| } 243| 0| ++mcuIdx; // We did just process one more MCU. 244| 0| } 245| | 246| | // ... and discard the rest. 247| 10.6k| for (; mcuIdx < frame.dim.x; ++mcuIdx) { ------------------ | Branch (247:10): [True: 0, False: 10.6k] ------------------ 248| 0| for (int i = 0; i != N_COMP; ++i) ------------------ | Branch (248:21): [True: 0, False: 0] ------------------ 249| 0| (static_cast&>(ht[i])).decodeDifference(bs); 250| 0| } 251| 10.6k|} LJpegDecompressor.cpp:_ZNK8rawspeed17LJpegDecompressor7decodeNIL_ZNS_12_GLOBAL__N_13MCUILi2ELi1EEEEEEjv: 255| 93|ByteStream::size_type LJpegDecompressor::decodeN() const { 256| 93| invariant(MCU == this->frame.mcu); ------------------ | | 27| 93|#define invariant(expr) assert(expr) ------------------ | Branch (256:3): [True: 93, False: 0] ------------------ 257| | 258| 93| invariant(MCU.hasPositiveArea()); ------------------ | | 27| 93|#define invariant(expr) assert(expr) ------------------ | Branch (258:3): [True: 93, False: 0] ------------------ 259| | // FIXME: workarounding lack of constexpr std::abs() :( 260| 93| constexpr int N_COMP = MCU.x * MCU.y; 261| | 262| 93| invariant(mRaw->getCpp() > 0); ------------------ | | 27| 93|#define invariant(expr) assert(expr) ------------------ | Branch (262:3): [True: 93, False: 0] ------------------ 263| | 264| 93| const auto img = 265| 93| CroppedArray2DRef(mRaw->getU16DataAsUncroppedArray2DRef(), 266| 93| mRaw->getCpp() * imgFrame.pos.x, imgFrame.pos.y, 267| 93| mRaw->getCpp() * imgFrame.dim.x, imgFrame.dim.y) 268| 93| .getAsArray2DRef(); 269| | 270| 93| const auto ht = getPrefixCodeDecoders(); 271| | 272| | // A recoded DNG might be split up into tiles of self contained LJpeg blobs. 273| | // The tiles at the bottom and the right may extend beyond the dimension of 274| | // the raw image buffer. The excessive content has to be ignored. 275| | 276| 93| invariant(imgFrame.dim.y % frame.mcu.y == 0); ------------------ | | 27| 93|#define invariant(expr) assert(expr) ------------------ | Branch (276:3): [True: 93, False: 0] ------------------ 277| 93| const auto numRestartIntervals = implicit_cast(roundUpDivisionSafe( 278| 93| imgFrame.dim.y / frame.mcu.y, numLJpegRowsPerRestartInterval)); 279| 93| invariant(numRestartIntervals >= 0); ------------------ | | 27| 93|#define invariant(expr) assert(expr) ------------------ | Branch (279:3): [True: 93, False: 0] ------------------ 280| 93| invariant(numRestartIntervals != 0); ------------------ | | 27| 93|#define invariant(expr) assert(expr) ------------------ | Branch (280:3): [True: 93, False: 0] ------------------ 281| | 282| 93| ByteStream inputStream(DataBuffer(input, Endianness::little)); 283| 93| for (int restartIntervalIndex = 0; 284| 186| restartIntervalIndex != numRestartIntervals; ++restartIntervalIndex) { ------------------ | Branch (284:8): [True: 93, False: 93] ------------------ 285| 93| auto predStorage = getInitialPreds(); 286| 93| auto pred = Array2DRef(predStorage.data(), MCU.x, MCU.y); 287| | 288| 93| if (restartIntervalIndex != 0) { ------------------ | Branch (288:9): [True: 0, False: 93] ------------------ 289| 0| auto marker = peekMarker(inputStream); 290| 0| if (!marker) // FIXME: can there be padding bytes before the marker? ------------------ | Branch (290:11): [True: 0, False: 0] ------------------ 291| 0| ThrowRDE("Jpeg marker not encountered"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 292| 0| Optional number = getRestartMarkerNumber(*marker); 293| 0| if (!number) ------------------ | Branch (293:11): [True: 0, False: 0] ------------------ 294| 0| ThrowRDE("Not a restart marker!"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 295| 0| if (*number != ((restartIntervalIndex - 1) % 8)) ------------------ | Branch (295:11): [True: 0, False: 0] ------------------ 296| 0| ThrowRDE("Unexpected restart marker found"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 297| 0| inputStream.skipBytes(2); // Good restart marker. 298| 0| } 299| | 300| 93| BitStreamerJPEG bs(inputStream.peekRemainingBuffer().getAsArray1DRef()); 301| | 302| 93| for (int ljpegRowOfRestartInterval = 0; 303| 974| ljpegRowOfRestartInterval != numLJpegRowsPerRestartInterval; ------------------ | Branch (303:10): [True: 881, False: 93] ------------------ 304| 881| ++ljpegRowOfRestartInterval) { 305| 881| const int row = 306| 881| frame.mcu.y * (numLJpegRowsPerRestartInterval * restartIntervalIndex + 307| 881| ljpegRowOfRestartInterval); 308| 881| invariant(row >= 0); ------------------ | | 27| 881|#define invariant(expr) assert(expr) ------------------ | Branch (308:7): [True: 881, False: 0] ------------------ 309| 881| invariant(row <= imgFrame.dim.y); ------------------ | | 27| 881|#define invariant(expr) assert(expr) ------------------ | Branch (309:7): [True: 881, False: 0] ------------------ 310| | 311| | // For y, we can simply stop decoding when we reached the border. 312| 881| if (row == imgFrame.dim.y) { ------------------ | Branch (312:11): [True: 0, False: 881] ------------------ 313| 0| invariant((restartIntervalIndex + 1) == numRestartIntervals); ------------------ | | 27| 0|#define invariant(expr) assert(expr) ------------------ | Branch (313:9): [True: 0, False: 0] ------------------ 314| 0| break; 315| 0| } 316| | 317| 881| const auto outStripe = CroppedArray2DRef(img, 318| 881| /*offsetCols=*/0, 319| 881| /*offsetRows=*/row, 320| 881| /*croppedWidth=*/img.width(), 321| 881| /*croppedHeight=*/frame.mcu.y) 322| 881| .getAsArray2DRef(); 323| | 324| 881| decodeRowN(outStripe, pred, ht, bs); 325| | 326| | // The predictor for the next line is the start of this line. 327| 881| pred = CroppedArray2DRef(outStripe, 328| 881| /*offsetCols=*/0, 329| 881| /*offsetRows=*/0, 330| 881| /*croppedWidth=*/MCU.x, 331| 881| /*croppedHeight=*/MCU.y) 332| 881| .getAsArray2DRef(); 333| 881| } 334| | 335| 93| inputStream.skipBytes(bs.getStreamPosition()); 336| 93| } 337| | 338| 93| return inputStream.getPosition(); 339| 93|} _ZNK8rawspeed17LJpegDecompressor21getPrefixCodeDecodersILi2EEENSt3__15arrayINS2_17reference_wrapperIKNS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS6_EEEEEEXT_EEEv: 164| 93|LJpegDecompressor::getPrefixCodeDecoders() const { 165| 93| return getPrefixCodeDecodersImpl(std::make_index_sequence{}); 166| 93|} _ZNK8rawspeed17LJpegDecompressor25getPrefixCodeDecodersImplILi2EJLm0ELm1EEEENSt3__15arrayINS2_17reference_wrapperIKNS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS6_EEEEEEXT_EEENS2_16integer_sequenceImJXspT0_EEEE: 157| 93| std::index_sequence /*unused*/) const { 158| 93| return std::array>, N_COMP>{ 159| 93| std::cref(rec[I].ht)...}; 160| 93|} _ZNK8rawspeed17LJpegDecompressor15getInitialPredsILi2EEENSt3__15arrayItXT_EEEv: 169| 93|std::array LJpegDecompressor::getInitialPreds() const { 170| 93| std::array preds; 171| 93| std::transform( 172| 93| rec.begin(), rec.end(), preds.begin(), 173| 93| [](const PerComponentRecipe& compRec) { return compRec.initPred; }); 174| 93| return preds; 175| 93|} _ZZNK8rawspeed17LJpegDecompressor15getInitialPredsILi2EEENSt3__15arrayItXT_EEEvENKUlRKNS0_18PerComponentRecipeEE_clES7_: 173| 186| [](const PerComponentRecipe& compRec) { return compRec.initPred; }); LJpegDecompressor.cpp:_ZNK8rawspeed17LJpegDecompressor10decodeRowNIL_ZNS_12_GLOBAL__N_13MCUILi2ELi1EEEELi2EEEvNS_10Array2DRefItEENS4_IKtEENSt3__15arrayINS8_17reference_wrapperIKNS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderISC_EEEEEEXT0_EEERNS_15BitStreamerJPEGE: 188| 881| BitStreamerJPEG& bs) const { 189| 881| invariant(MCUSize.area() == N_COMP); ------------------ | | 27| 881|#define invariant(expr) assert(expr) ------------------ | Branch (189:3): [True: 881, False: 0] ------------------ 190| 881| invariant(outStripe.width() >= MCUSize.x); ------------------ | | 27| 881|#define invariant(expr) assert(expr) ------------------ | Branch (190:3): [True: 881, False: 0] ------------------ 191| 881| invariant(outStripe.height() == MCUSize.y); ------------------ | | 27| 881|#define invariant(expr) assert(expr) ------------------ | Branch (191:3): [True: 881, False: 0] ------------------ 192| 881| invariant(pred.width() == MCUSize.x); ------------------ | | 27| 881|#define invariant(expr) assert(expr) ------------------ | Branch (192:3): [True: 881, False: 0] ------------------ 193| 881| invariant(pred.height() == MCUSize.y); ------------------ | | 27| 881|#define invariant(expr) assert(expr) ------------------ | Branch (193:3): [True: 881, False: 0] ------------------ 194| | 195| | // FIXME: predictor may have value outside of the uint16_t. 196| | // https://github.com/darktable-org/rawspeed/issues/175 197| | 198| 881| int mcuIdx = 0; 199| | // For x, we first process all full pixel MCUs within the image buffer ... 200| 7.30k| for (; mcuIdx < numFullMCUs; ++mcuIdx) { ------------------ | Branch (200:10): [True: 6.42k, False: 881] ------------------ 201| 6.42k| const auto outTile = CroppedArray2DRef(outStripe, 202| 6.42k| /*offsetCols=*/MCUSize.x * mcuIdx, 203| 6.42k| /*offsetRows=*/0, 204| 6.42k| /*croppedWidth=*/MCUSize.x, 205| 6.42k| /*croppedHeight=*/MCUSize.y) 206| 6.42k| .getAsArray2DRef(); 207| 12.8k| for (int MCURow = 0; MCURow != MCUSize.y; ++MCURow) { ------------------ | Branch (207:26): [True: 6.42k, False: 6.42k] ------------------ 208| 19.2k| for (int MCUСol = 0; MCUСol != MCUSize.x; ++MCUСol) { ------------------ | Branch (208:29): [True: 12.8k, False: 6.42k] ------------------ 209| 12.8k| int c = (MCUSize.x * MCURow) + MCUСol; 210| 12.8k| int prediction = pred(MCURow, MCUСol); 211| 12.8k| int diff = (static_cast&>(ht[c])) 212| 12.8k| .decodeDifference(bs); 213| 12.8k| int pix = prediction + diff; 214| 12.8k| outTile(MCURow, MCUСol) = uint16_t(pix); 215| 12.8k| } 216| 6.42k| } 217| | // The predictor for the next MCU of the row is the just-decoded MCU. 218| 6.42k| pred = outTile; 219| 6.42k| } 220| | 221| | // Sometimes we also need to consume one more MCU, and produce part of it. 222| 881| if (trailingPixels != 0) { ------------------ | Branch (222:7): [True: 0, False: 881] ------------------ 223| | // Some rather esoteric DNG's have odd dimensions, e.g. width % 2 = 1. 224| | // We may end up needing just part of last N_COMP pixels. 225| 0| invariant(trailingPixels > 0); ------------------ | | 27| 0|#define invariant(expr) assert(expr) ------------------ | Branch (225:5): [True: 0, False: 0] ------------------ 226| 0| invariant(trailingPixels < N_COMP); ------------------ | | 27| 0|#define invariant(expr) assert(expr) ------------------ | Branch (226:5): [True: 0, False: 0] ------------------ 227| 0| invariant(N_COMP > 1 && "can't want part of 1-pixel-wide block"); ------------------ | | 27| 0|#define invariant(expr) assert(expr) ------------------ | Branch (227:5): [True: 0, Folded] | Branch (227:5): [True: 0, Folded] | Branch (227:5): [True: 0, Folded] ------------------ 228| | // Some rather esoteric DNG's have odd dimensions, e.g. width % 2 = 1. 229| | // We may end up needing just part of last N_COMP pixels. 230| 0| for (int MCURow = 0; MCURow != MCUSize.y; ++MCURow) { ------------------ | Branch (230:26): [True: 0, False: 0] ------------------ 231| 0| for (int MCUСol = 0; MCUСol != MCUSize.x; ++MCUСol) { ------------------ | Branch (231:29): [True: 0, False: 0] ------------------ 232| 0| int c = (MCUSize.x * MCURow) + MCUСol; 233| 0| int prediction = pred(MCURow, MCUСol); 234| 0| int diff = (static_cast&>(ht[c])) 235| 0| .decodeDifference(bs); 236| 0| int pix = prediction + diff; 237| 0| int stripeRow = MCURow; 238| 0| int stripeCol = (MCUSize.x * mcuIdx) + MCUСol; 239| 0| if (stripeCol < outStripe.width()) ------------------ | Branch (239:13): [True: 0, False: 0] ------------------ 240| 0| outStripe(stripeRow, stripeCol) = uint16_t(pix); 241| 0| } 242| 0| } 243| 0| ++mcuIdx; // We did just process one more MCU. 244| 0| } 245| | 246| | // ... and discard the rest. 247| 881| for (; mcuIdx < frame.dim.x; ++mcuIdx) { ------------------ | Branch (247:10): [True: 0, False: 881] ------------------ 248| 0| for (int i = 0; i != N_COMP; ++i) ------------------ | Branch (248:21): [True: 0, False: 0] ------------------ 249| 0| (static_cast&>(ht[i])).decodeDifference(bs); 250| 0| } 251| 881|} LJpegDecompressor.cpp:_ZNK8rawspeed17LJpegDecompressor7decodeNIL_ZNS_12_GLOBAL__N_13MCUILi3ELi1EEEEEEjv: 255| 128|ByteStream::size_type LJpegDecompressor::decodeN() const { 256| 128| invariant(MCU == this->frame.mcu); ------------------ | | 27| 128|#define invariant(expr) assert(expr) ------------------ | Branch (256:3): [True: 128, False: 0] ------------------ 257| | 258| 128| invariant(MCU.hasPositiveArea()); ------------------ | | 27| 128|#define invariant(expr) assert(expr) ------------------ | Branch (258:3): [True: 128, False: 0] ------------------ 259| | // FIXME: workarounding lack of constexpr std::abs() :( 260| 128| constexpr int N_COMP = MCU.x * MCU.y; 261| | 262| 128| invariant(mRaw->getCpp() > 0); ------------------ | | 27| 128|#define invariant(expr) assert(expr) ------------------ | Branch (262:3): [True: 128, False: 0] ------------------ 263| | 264| 128| const auto img = 265| 128| CroppedArray2DRef(mRaw->getU16DataAsUncroppedArray2DRef(), 266| 128| mRaw->getCpp() * imgFrame.pos.x, imgFrame.pos.y, 267| 128| mRaw->getCpp() * imgFrame.dim.x, imgFrame.dim.y) 268| 128| .getAsArray2DRef(); 269| | 270| 128| const auto ht = getPrefixCodeDecoders(); 271| | 272| | // A recoded DNG might be split up into tiles of self contained LJpeg blobs. 273| | // The tiles at the bottom and the right may extend beyond the dimension of 274| | // the raw image buffer. The excessive content has to be ignored. 275| | 276| 128| invariant(imgFrame.dim.y % frame.mcu.y == 0); ------------------ | | 27| 128|#define invariant(expr) assert(expr) ------------------ | Branch (276:3): [True: 127, False: 1] ------------------ 277| 127| const auto numRestartIntervals = implicit_cast(roundUpDivisionSafe( 278| 127| imgFrame.dim.y / frame.mcu.y, numLJpegRowsPerRestartInterval)); 279| 127| invariant(numRestartIntervals >= 0); ------------------ | | 27| 127|#define invariant(expr) assert(expr) ------------------ | Branch (279:3): [True: 126, False: 1] ------------------ 280| 126| invariant(numRestartIntervals != 0); ------------------ | | 27| 126|#define invariant(expr) assert(expr) ------------------ | Branch (280:3): [True: 126, False: 0] ------------------ 281| | 282| 126| ByteStream inputStream(DataBuffer(input, Endianness::little)); 283| 126| for (int restartIntervalIndex = 0; 284| 254| restartIntervalIndex != numRestartIntervals; ++restartIntervalIndex) { ------------------ | Branch (284:8): [True: 128, False: 126] ------------------ 285| 128| auto predStorage = getInitialPreds(); 286| 128| auto pred = Array2DRef(predStorage.data(), MCU.x, MCU.y); 287| | 288| 128| if (restartIntervalIndex != 0) { ------------------ | Branch (288:9): [True: 0, False: 128] ------------------ 289| 0| auto marker = peekMarker(inputStream); 290| 0| if (!marker) // FIXME: can there be padding bytes before the marker? ------------------ | Branch (290:11): [True: 0, False: 0] ------------------ 291| 0| ThrowRDE("Jpeg marker not encountered"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 292| 0| Optional number = getRestartMarkerNumber(*marker); 293| 0| if (!number) ------------------ | Branch (293:11): [True: 0, False: 0] ------------------ 294| 0| ThrowRDE("Not a restart marker!"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 295| 0| if (*number != ((restartIntervalIndex - 1) % 8)) ------------------ | Branch (295:11): [True: 0, False: 0] ------------------ 296| 0| ThrowRDE("Unexpected restart marker found"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 297| 0| inputStream.skipBytes(2); // Good restart marker. 298| 0| } 299| | 300| 128| BitStreamerJPEG bs(inputStream.peekRemainingBuffer().getAsArray1DRef()); 301| | 302| 128| for (int ljpegRowOfRestartInterval = 0; 303| 7.00k| ljpegRowOfRestartInterval != numLJpegRowsPerRestartInterval; ------------------ | Branch (303:10): [True: 6.88k, False: 121] ------------------ 304| 6.88k| ++ljpegRowOfRestartInterval) { 305| 6.88k| const int row = 306| 6.88k| frame.mcu.y * (numLJpegRowsPerRestartInterval * restartIntervalIndex + 307| 6.88k| ljpegRowOfRestartInterval); 308| 6.88k| invariant(row >= 0); ------------------ | | 27| 6.88k|#define invariant(expr) assert(expr) ------------------ | Branch (308:7): [True: 6.88k, False: 0] ------------------ 309| 6.88k| invariant(row <= imgFrame.dim.y); ------------------ | | 27| 6.88k|#define invariant(expr) assert(expr) ------------------ | Branch (309:7): [True: 6.88k, False: 0] ------------------ 310| | 311| | // For y, we can simply stop decoding when we reached the border. 312| 6.88k| if (row == imgFrame.dim.y) { ------------------ | Branch (312:11): [True: 7, False: 6.87k] ------------------ 313| 7| invariant((restartIntervalIndex + 1) == numRestartIntervals); ------------------ | | 27| 7|#define invariant(expr) assert(expr) ------------------ | Branch (313:9): [True: 7, False: 0] ------------------ 314| 7| break; 315| 7| } 316| | 317| 6.87k| const auto outStripe = CroppedArray2DRef(img, 318| 6.87k| /*offsetCols=*/0, 319| 6.87k| /*offsetRows=*/row, 320| 6.87k| /*croppedWidth=*/img.width(), 321| 6.87k| /*croppedHeight=*/frame.mcu.y) 322| 6.87k| .getAsArray2DRef(); 323| | 324| 6.87k| decodeRowN(outStripe, pred, ht, bs); 325| | 326| | // The predictor for the next line is the start of this line. 327| 6.87k| pred = CroppedArray2DRef(outStripe, 328| 6.87k| /*offsetCols=*/0, 329| 6.87k| /*offsetRows=*/0, 330| 6.87k| /*croppedWidth=*/MCU.x, 331| 6.87k| /*croppedHeight=*/MCU.y) 332| 6.87k| .getAsArray2DRef(); 333| 6.87k| } 334| | 335| 128| inputStream.skipBytes(bs.getStreamPosition()); 336| 128| } 337| | 338| 126| return inputStream.getPosition(); 339| 126|} _ZNK8rawspeed17LJpegDecompressor21getPrefixCodeDecodersILi3EEENSt3__15arrayINS2_17reference_wrapperIKNS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS6_EEEEEEXT_EEEv: 164| 126|LJpegDecompressor::getPrefixCodeDecoders() const { 165| 126| return getPrefixCodeDecodersImpl(std::make_index_sequence{}); 166| 126|} _ZNK8rawspeed17LJpegDecompressor25getPrefixCodeDecodersImplILi3EJLm0ELm1ELm2EEEENSt3__15arrayINS2_17reference_wrapperIKNS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS6_EEEEEEXT_EEENS2_16integer_sequenceImJXspT0_EEEE: 157| 127| std::index_sequence /*unused*/) const { 158| 127| return std::array>, N_COMP>{ 159| 127| std::cref(rec[I].ht)...}; 160| 127|} _ZNK8rawspeed17LJpegDecompressor15getInitialPredsILi3EEENSt3__15arrayItXT_EEEv: 169| 128|std::array LJpegDecompressor::getInitialPreds() const { 170| 128| std::array preds; 171| 128| std::transform( 172| 128| rec.begin(), rec.end(), preds.begin(), 173| 128| [](const PerComponentRecipe& compRec) { return compRec.initPred; }); 174| 128| return preds; 175| 128|} _ZZNK8rawspeed17LJpegDecompressor15getInitialPredsILi3EEENSt3__15arrayItXT_EEEvENKUlRKNS0_18PerComponentRecipeEE_clES7_: 173| 384| [](const PerComponentRecipe& compRec) { return compRec.initPred; }); LJpegDecompressor.cpp:_ZNK8rawspeed17LJpegDecompressor10decodeRowNIL_ZNS_12_GLOBAL__N_13MCUILi3ELi1EEEELi3EEEvNS_10Array2DRefItEENS4_IKtEENSt3__15arrayINS8_17reference_wrapperIKNS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderISC_EEEEEEXT0_EEERNS_15BitStreamerJPEGE: 188| 6.87k| BitStreamerJPEG& bs) const { 189| 6.87k| invariant(MCUSize.area() == N_COMP); ------------------ | | 27| 6.87k|#define invariant(expr) assert(expr) ------------------ | Branch (189:3): [True: 6.87k, False: 0] ------------------ 190| 6.87k| invariant(outStripe.width() >= MCUSize.x); ------------------ | | 27| 6.87k|#define invariant(expr) assert(expr) ------------------ | Branch (190:3): [True: 6.87k, False: 0] ------------------ 191| 6.87k| invariant(outStripe.height() == MCUSize.y); ------------------ | | 27| 6.87k|#define invariant(expr) assert(expr) ------------------ | Branch (191:3): [True: 6.87k, False: 1] ------------------ 192| 6.87k| invariant(pred.width() == MCUSize.x); ------------------ | | 27| 6.87k|#define invariant(expr) assert(expr) ------------------ | Branch (192:3): [True: 6.87k, False: 18.4E] ------------------ 193| 6.87k| invariant(pred.height() == MCUSize.y); ------------------ | | 27| 6.87k|#define invariant(expr) assert(expr) ------------------ | Branch (193:3): [True: 6.87k, False: 0] ------------------ 194| | 195| | // FIXME: predictor may have value outside of the uint16_t. 196| | // https://github.com/darktable-org/rawspeed/issues/175 197| | 198| 6.87k| int mcuIdx = 0; 199| | // For x, we first process all full pixel MCUs within the image buffer ... 200| 1.24M| for (; mcuIdx < numFullMCUs; ++mcuIdx) { ------------------ | Branch (200:10): [True: 1.23M, False: 6.87k] ------------------ 201| 1.23M| const auto outTile = CroppedArray2DRef(outStripe, 202| 1.23M| /*offsetCols=*/MCUSize.x * mcuIdx, 203| 1.23M| /*offsetRows=*/0, 204| 1.23M| /*croppedWidth=*/MCUSize.x, 205| 1.23M| /*croppedHeight=*/MCUSize.y) 206| 1.23M| .getAsArray2DRef(); 207| 2.46M| for (int MCURow = 0; MCURow != MCUSize.y; ++MCURow) { ------------------ | Branch (207:26): [True: 1.23M, False: 1.23M] ------------------ 208| 4.92M| for (int MCUСol = 0; MCUСol != MCUSize.x; ++MCUСol) { ------------------ | Branch (208:29): [True: 3.69M, False: 1.23M] ------------------ 209| 3.69M| int c = (MCUSize.x * MCURow) + MCUСol; 210| 3.69M| int prediction = pred(MCURow, MCUСol); 211| 3.69M| int diff = (static_cast&>(ht[c])) 212| 3.69M| .decodeDifference(bs); 213| 3.69M| int pix = prediction + diff; 214| 3.69M| outTile(MCURow, MCUСol) = uint16_t(pix); 215| 3.69M| } 216| 1.23M| } 217| | // The predictor for the next MCU of the row is the just-decoded MCU. 218| 1.23M| pred = outTile; 219| 1.23M| } 220| | 221| | // Sometimes we also need to consume one more MCU, and produce part of it. 222| 6.87k| if (trailingPixels != 0) { ------------------ | Branch (222:7): [True: 0, False: 6.87k] ------------------ 223| | // Some rather esoteric DNG's have odd dimensions, e.g. width % 2 = 1. 224| | // We may end up needing just part of last N_COMP pixels. 225| 0| invariant(trailingPixels > 0); ------------------ | | 27| 0|#define invariant(expr) assert(expr) ------------------ | Branch (225:5): [True: 0, False: 0] ------------------ 226| 0| invariant(trailingPixels < N_COMP); ------------------ | | 27| 0|#define invariant(expr) assert(expr) ------------------ | Branch (226:5): [True: 0, False: 0] ------------------ 227| 0| invariant(N_COMP > 1 && "can't want part of 1-pixel-wide block"); ------------------ | | 27| 0|#define invariant(expr) assert(expr) ------------------ | Branch (227:5): [True: 0, Folded] | Branch (227:5): [True: 0, Folded] | Branch (227:5): [True: 0, Folded] ------------------ 228| | // Some rather esoteric DNG's have odd dimensions, e.g. width % 2 = 1. 229| | // We may end up needing just part of last N_COMP pixels. 230| 0| for (int MCURow = 0; MCURow != MCUSize.y; ++MCURow) { ------------------ | Branch (230:26): [True: 0, False: 0] ------------------ 231| 0| for (int MCUСol = 0; MCUСol != MCUSize.x; ++MCUСol) { ------------------ | Branch (231:29): [True: 0, False: 0] ------------------ 232| 0| int c = (MCUSize.x * MCURow) + MCUСol; 233| 0| int prediction = pred(MCURow, MCUСol); 234| 0| int diff = (static_cast&>(ht[c])) 235| 0| .decodeDifference(bs); 236| 0| int pix = prediction + diff; 237| 0| int stripeRow = MCURow; 238| 0| int stripeCol = (MCUSize.x * mcuIdx) + MCUСol; 239| 0| if (stripeCol < outStripe.width()) ------------------ | Branch (239:13): [True: 0, False: 0] ------------------ 240| 0| outStripe(stripeRow, stripeCol) = uint16_t(pix); 241| 0| } 242| 0| } 243| 0| ++mcuIdx; // We did just process one more MCU. 244| 0| } 245| | 246| | // ... and discard the rest. 247| 83.2k| for (; mcuIdx < frame.dim.x; ++mcuIdx) { ------------------ | Branch (247:10): [True: 76.3k, False: 6.87k] ------------------ 248| 305k| for (int i = 0; i != N_COMP; ++i) ------------------ | Branch (248:21): [True: 229k, False: 76.3k] ------------------ 249| 229k| (static_cast&>(ht[i])).decodeDifference(bs); 250| 76.3k| } 251| 6.87k|} _ZN8rawspeed17NikonDecompressor11createCurveERNS_10ByteStreamEjjjPj: 384| 164| uint32_t* split) { 385| | // Nikon Z7 12/14 bit compressed hack. 386| 164| if (v0 == 68 && v1 == 64) ------------------ | Branch (386:7): [True: 89, False: 75] | Branch (386:19): [True: 75, False: 14] ------------------ 387| 75| bitsPS -= 2; 388| | 389| | // 'curve' will hold a peace wise linearly interpolated function. 390| | // there are 'csize' segments, each is 'step' values long. 391| | // the very last value is not part of the used table but necessary 392| | // to linearly interpolate the last segment, therefore the '+1/-1' 393| | // size adjustments of 'curve'. 394| 164| std::vector curve((1 << bitsPS & 0x7fff) + 1); 395| 164| assert(curve.size() > 1); ------------------ | Branch (395:3): [True: 164, False: 0] ------------------ 396| | 397| 1.13M| for (size_t i = 0; i < curve.size(); i++) ------------------ | Branch (397:22): [True: 1.13M, False: 164] ------------------ 398| 1.13M| curve[i] = implicit_cast(i); 399| | 400| 164| uint32_t step = 0; 401| 164| uint32_t csize = metadata.getU16(); 402| 164| if (csize > 1) ------------------ | Branch (402:7): [True: 145, False: 19] ------------------ 403| 145| step = implicit_cast(curve.size() / (csize - 1)); 404| | 405| 164| if (v0 == 68 && (v1 == 32 || v1 == 64) && step > 0) { ------------------ | Branch (405:7): [True: 89, False: 75] | Branch (405:20): [True: 10, False: 79] | Branch (405:32): [True: 75, False: 4] | Branch (405:45): [True: 80, False: 5] ------------------ 406| 80| if ((csize - 1) * step != curve.size() - 1) ------------------ | Branch (406:9): [True: 1, False: 79] ------------------ 407| 80| ThrowRDE("Bad curve segment count (%u)", csize); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 408| | 409| 1.09k| for (size_t i = 0; i < csize; i++) ------------------ | Branch (409:24): [True: 1.01k, False: 79] ------------------ 410| 1.01k| curve[i * step] = metadata.getU16(); 411| 300k| for (size_t i = 0; i < curve.size() - 1; i++) { ------------------ | Branch (411:24): [True: 300k, False: 79] ------------------ 412| 300k| const uint32_t b_scale = i % step; 413| | 414| 300k| const auto a_pos = implicit_cast(i - b_scale); 415| 300k| const uint32_t b_pos = a_pos + step; 416| 300k| assert(a_pos < curve.size()); ------------------ | Branch (416:7): [True: 300k, False: 0] ------------------ 417| 300k| assert(b_pos > 0); ------------------ | Branch (417:7): [True: 300k, False: 0] ------------------ 418| 300k| assert(b_pos < curve.size()); ------------------ | Branch (418:7): [True: 300k, False: 0] ------------------ 419| 300k| assert(a_pos < b_pos); ------------------ | Branch (419:7): [True: 300k, False: 0] ------------------ 420| | 421| 300k| const uint32_t a_scale = step - b_scale; 422| 300k| curve[i] = implicit_cast( 423| 300k| (a_scale * curve[a_pos] + b_scale * curve[b_pos]) / step); 424| 300k| } 425| | 426| 79| metadata.setPosition(562); 427| 79| *split = metadata.getU16(); 428| 84| } else if (v0 != 70) { ------------------ | Branch (428:14): [True: 69, False: 15] ------------------ 429| 69| if (csize == 0 || csize > 0x4001) ------------------ | Branch (429:9): [True: 2, False: 67] | Branch (429:23): [True: 2, False: 65] ------------------ 430| 69| ThrowRDE("Don't know how to compute curve! csize = %u", csize); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 431| | 432| 65| curve.resize(csize + 1UL); 433| 65| assert(curve.size() > 1); ------------------ | Branch (433:5): [True: 65, False: 0] ------------------ 434| | 435| 35.4k| for (uint32_t i = 0; i < csize; i++) { ------------------ | Branch (435:26): [True: 35.4k, False: 65] ------------------ 436| 35.4k| curve[i] = metadata.getU16(); 437| 35.4k| } 438| 65| } 439| | 440| | // and drop the last value 441| 159| curve.resize(curve.size() - 1); 442| 159| assert(!curve.empty()); ------------------ | Branch (442:3): [True: 130, False: 29] ------------------ 443| | 444| 130| return curve; 445| 159|} _ZN8rawspeed17NikonDecompressor23createPrefixCodeDecoderINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS3_EEEEEET_j: 461| 128| uint32_t huffSelect) { 462| 128| HuffmanCode hc; 463| 128| uint32_t count = 464| 128| hc.setNCodesPerLength(Buffer(nikon_tree[huffSelect][0].data(), 16)); 465| 128| hc.setCodeValues( 466| 128| Array1DRef(nikon_tree[huffSelect][1].data(), count)); 467| | 468| 128| PrefixCodeDecoder<> ht(std::move(hc)); 469| 128| ht.setup(true, false); 470| 128| return ht; 471| 128|} _ZN8rawspeed17NikonDecompressorC2ENS_8RawImageENS_10ByteStreamEj: 475| 218| : mRaw(std::move(raw)), bitsPS(bitsPS_) { 476| 218| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (476:7): [True: 0, False: 218] | Branch (476:30): [True: 0, False: 218] ------------------ 477| 218| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (477:7): [True: 0, False: 218] ------------------ 478| 218| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 479| | 480| 218| if (!mRaw->dim.hasPositiveArea() || mRaw->dim.x % 2 != 0 || ------------------ | Branch (480:7): [True: 35, False: 183] | Branch (480:39): [True: 2, False: 181] ------------------ 481| 181| mRaw->dim.x > 8288 || mRaw->dim.y > 5520) ------------------ | Branch (481:7): [True: 4, False: 177] | Branch (481:29): [True: 7, False: 170] ------------------ 482| 218| ThrowRDE("Unexpected image dimensions found: (%d; %d)", mRaw->dim.x, ------------------ | | 38| 48| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 48| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 48| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 48| __VA_ARGS__) | | ------------------ ------------------ 483| 170| mRaw->dim.y); 484| | 485| 170| switch (bitsPS) { 486| 87| case 12: ------------------ | Branch (486:3): [True: 87, False: 83] ------------------ 487| 168| case 14: ------------------ | Branch (487:3): [True: 81, False: 89] ------------------ 488| 168| break; 489| 2| default: ------------------ | Branch (489:3): [True: 2, False: 168] ------------------ 490| 2| ThrowRDE("Invalid bpp found: %u", bitsPS); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 491| 170| } 492| | 493| 168| uint32_t v0 = metadata.getByte(); 494| 168| uint32_t v1 = metadata.getByte(); 495| | 496| 168| writeLog(DEBUG_PRIO::EXTRA, "Nef version v0:%u, v1:%u", v0, v1); 497| | 498| 168| if (v0 == 73 || v1 == 88) ------------------ | Branch (498:7): [True: 15, False: 153] | Branch (498:19): [True: 1, False: 152] ------------------ 499| 15| metadata.skipBytes(2110); 500| | 501| 168| if (v0 == 70) ------------------ | Branch (501:7): [True: 15, False: 153] ------------------ 502| 15| huffSelect = 2; 503| 168| if (bitsPS == 14) ------------------ | Branch (503:7): [True: 79, False: 89] ------------------ 504| 79| huffSelect += 3; 505| | 506| 168| pUp[0][0] = metadata.getU16(); 507| 168| pUp[1][0] = metadata.getU16(); 508| 168| pUp[0][1] = metadata.getU16(); 509| 168| pUp[1][1] = metadata.getU16(); 510| | 511| 168| curve = createCurve(metadata, bitsPS, v0, v1, &split); 512| | 513| | // If the 'split' happens outside of the image, it does not actually happen. 514| 168| if (split >= static_cast(mRaw->dim.y)) ------------------ | Branch (514:7): [True: 15, False: 153] ------------------ 515| 15| split = 0; 516| 168|} _ZN8rawspeed17NikonDecompressor10decompressENS_10Array1DRefIKhEEb: 544| 130| bool uncorrectedRawValues) { 545| 130| RawImageCurveGuard curveHandler(&mRaw, curve, uncorrectedRawValues); 546| | 547| 130| BitStreamerMSB bits(input); 548| | 549| 130| random = bits.peekBits(24); 550| | 551| 130| invariant(split == 0 || split < static_cast(mRaw->dim.y)); ------------------ | | 27| 130|#define invariant(expr) assert(expr) ------------------ | Branch (551:3): [True: 77, False: 53] | Branch (551:3): [True: 53, False: 0] | Branch (551:3): [True: 128, False: 2] ------------------ 552| | 553| 128| if (!split) { ------------------ | Branch (553:7): [True: 75, False: 53] ------------------ 554| 75| decompress>(bits, 0, mRaw->dim.y); 555| 75| } else { 556| 53| decompress>(bits, 0, split); 557| 53| huffSelect += 1; 558| 53| decompress(bits, split, mRaw->dim.y); 559| 53| } 560| 128|} _ZN8rawspeed17NikonDecompressor10decompressINS_20PrefixCodeLUTDecoderINS_15BaselineCodeTagENS_23PrefixCodeLookupDecoderIS3_EEEEEEvRNS_14BitStreamerMSBEii: 520| 128| int end_y) { 521| 128| auto ht = createPrefixCodeDecoder(huffSelect); 522| | 523| 128| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 524| | 525| | // allow gcc to devirtualize the calls below 526| 128| auto* rawdata = reinterpret_cast(mRaw.get()); 527| | 528| 128| invariant(out.width() % 2 == 0); ------------------ | | 27| 128|#define invariant(expr) assert(expr) ------------------ | Branch (528:3): [True: 128, False: 0] ------------------ 529| 128| invariant(out.width() >= 2); ------------------ | | 27| 128|#define invariant(expr) assert(expr) ------------------ | Branch (529:3): [True: 128, False: 0] ------------------ 530| 3.39k| for (int row = start_y; row < end_y; row++) { ------------------ | Branch (530:27): [True: 3.27k, False: 128] ------------------ 531| 3.27k| std::array pred = pUp[row & 1]; 532| 379k| for (int col = 0; col < out.width(); col++) { ------------------ | Branch (532:23): [True: 375k, False: 3.27k] ------------------ 533| 375k| pred[col & 1] += ht.decodeDifference(bits); 534| 375k| if (col < 2) ------------------ | Branch (534:11): [True: 6.52k, False: 369k] ------------------ 535| 6.52k| pUp[row & 1][col & 1] = pred[col & 1]; 536| 375k| rawdata->setWithLookUp(clampBits(pred[col & 1], 15), 537| 375k| reinterpret_cast(&out(row, col)), 538| 375k| &random); 539| 375k| } 540| 3.27k| } 541| 128|} NikonDecompressor.cpp:_ZN8rawspeed17NikonDecompressor10decompressINS_12_GLOBAL__N_120NikonLASDecompressorEEEvRNS_14BitStreamerMSBEii: 520| 49| int end_y) { 521| 49| auto ht = createPrefixCodeDecoder(huffSelect); 522| | 523| 49| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 524| | 525| | // allow gcc to devirtualize the calls below 526| 49| auto* rawdata = reinterpret_cast(mRaw.get()); 527| | 528| 49| invariant(out.width() % 2 == 0); ------------------ | | 27| 49|#define invariant(expr) assert(expr) ------------------ | Branch (528:3): [True: 49, False: 0] ------------------ 529| 49| invariant(out.width() >= 2); ------------------ | | 27| 49|#define invariant(expr) assert(expr) ------------------ | Branch (529:3): [True: 49, False: 0] ------------------ 530| 1.58k| for (int row = start_y; row < end_y; row++) { ------------------ | Branch (530:27): [True: 1.53k, False: 49] ------------------ 531| 1.53k| std::array pred = pUp[row & 1]; 532| 455k| for (int col = 0; col < out.width(); col++) { ------------------ | Branch (532:23): [True: 454k, False: 1.53k] ------------------ 533| 454k| pred[col & 1] += ht.decodeDifference(bits); 534| 454k| if (col < 2) ------------------ | Branch (534:11): [True: 3.06k, False: 451k] ------------------ 535| 3.06k| pUp[row & 1][col & 1] = pred[col & 1]; 536| 454k| rawdata->setWithLookUp(clampBits(pred[col & 1], 15), 537| 454k| reinterpret_cast(&out(row, col)), 538| 454k| &random); 539| 454k| } 540| 1.53k| } 541| 49|} NikonDecompressor.cpp:_ZN8rawspeed17NikonDecompressor23createPrefixCodeDecoderINS_12_GLOBAL__N_120NikonLASDecompressorEEET_j: 448| 49|Huffman NikonDecompressor::createPrefixCodeDecoder(uint32_t huffSelect) { 449| 49| Huffman ht; 450| 49| uint32_t count = 451| 49| ht.setNCodesPerLength(Buffer(nikon_tree[huffSelect][0].data(), 16)); 452| 49| ht.setCodeValues( 453| 49| Array1DRef(nikon_tree[huffSelect][1].data(), count)); 454| 49| ht.setup(true, false); 455| 49| return ht; 456| 49|} NikonDecompressor.cpp:_ZN8rawspeed12_GLOBAL__N_120NikonLASDecompressor18setNCodesPerLengthENS_6BufferE: 289| 49| uint32_t setNCodesPerLength(Buffer data) { 290| 49| uint32_t acc = 0; 291| 833| for (uint32_t i = 0; i < 16; i++) { ------------------ | Branch (291:26): [True: 784, False: 49] ------------------ 292| 784| dctbl1.bits[i + 1] = data[i]; 293| 784| acc += dctbl1.bits[i + 1]; 294| 784| } 295| 49| dctbl1.bits[0] = 0; 296| 49| return acc; 297| 49| } NikonDecompressor.cpp:_ZN8rawspeed12_GLOBAL__N_120NikonLASDecompressor13setCodeValuesENS_10Array1DRefIKhEE: 299| 49| void setCodeValues(Array1DRef data) { 300| 755| for (int i = 0; i < data.size(); i++) ------------------ | Branch (300:21): [True: 706, False: 49] ------------------ 301| 706| dctbl1.huffval[i] = data(i); 302| 49| } NikonDecompressor.cpp:_ZN8rawspeed12_GLOBAL__N_120NikonLASDecompressor5setupEbb: 305| 49| [[maybe_unused]] bool fixDNGBug16_) { 306| 49| createPrefixCodeDecoder(); 307| 49| } NikonDecompressor.cpp:_ZN8rawspeed12_GLOBAL__N_120NikonLASDecompressor23createPrefixCodeDecoderEv: 106| 49| void createPrefixCodeDecoder() { 107| 49| int p; 108| 49| int i; 109| 49| int l; 110| 49| int lastp; 111| 49| int si; 112| 49| std::array huffsize; 113| 49| std::array huffcode; 114| 49| uint16_t code; 115| 49| int size; 116| 49| int value; 117| 49| int ll; 118| 49| int ul; 119| | 120| | /* 121| | * Figure C.1: make table of Huffman code length for each symbol 122| | * Note that this is in code-length order. 123| | */ 124| 49| p = 0; 125| 833| for (l = 1; l <= 16; l++) { ------------------ | Branch (125:17): [True: 784, False: 49] ------------------ 126| 1.49k| for (i = 1; i <= static_cast(dctbl1.bits[l]); i++) { ------------------ | Branch (126:19): [True: 706, False: 784] ------------------ 127| 706| huffsize[p] = static_cast(l); 128| 706| ++p; 129| 706| if (p > 256) ------------------ | Branch (129:13): [True: 0, False: 706] ------------------ 130| 706| ThrowRDE("LJpegDecoder::createPrefixCodeDecoder: Code length too " ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 131| 706| "long. Corrupt data."); 132| 706| } 133| 784| } 134| 49| huffsize[p] = 0; 135| 49| lastp = p; 136| | 137| | /* 138| | * Figure C.2: generate the codes themselves 139| | * Note that this is in code-length order. 140| | */ 141| 49| code = 0; 142| 49| si = huffsize[0]; 143| 49| p = 0; 144| 510| while (huffsize[p]) { ------------------ | Branch (144:12): [True: 461, False: 49] ------------------ 145| 1.16k| while ((static_cast(huffsize[p])) == si) { ------------------ | Branch (145:14): [True: 706, False: 461] ------------------ 146| 706| huffcode[p] = code; 147| 706| ++p; 148| 706| code++; 149| 706| } 150| 461| code <<= 1; 151| 461| si++; 152| 461| if (p > 256) ------------------ | Branch (152:11): [True: 0, False: 461] ------------------ 153| 461| ThrowRDE( ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 154| 461| "createPrefixCodeDecoder: Code length too long. Corrupt data."); 155| 461| } 156| | 157| | /* 158| | * Figure F.15: generate decoding tables 159| | */ 160| 49| dctbl1.mincode[0] = 0; 161| 49| dctbl1.maxcode[0] = 0; 162| 49| p = 0; 163| 833| for (l = 1; l <= 16; l++) { ------------------ | Branch (163:17): [True: 784, False: 49] ------------------ 164| 784| if (dctbl1.bits[l]) { ------------------ | Branch (164:11): [True: 461, False: 323] ------------------ 165| 461| dctbl1.valptr[l] = implicit_cast(p); 166| 461| dctbl1.mincode[l] = huffcode[p]; 167| 461| p += dctbl1.bits[l]; 168| 461| dctbl1.maxcode[l] = huffcode[p - 1]; 169| 461| } else { 170| 323| dctbl1.valptr[l] = 171| 323| 0xff; // This check must be present to avoid crash on junk 172| 323| dctbl1.maxcode[l] = -1; 173| 323| } 174| 784| if (p > 256) ------------------ | Branch (174:11): [True: 0, False: 784] ------------------ 175| 784| ThrowRDE( ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 176| 784| "createPrefixCodeDecoder: Code length too long. Corrupt data."); 177| 784| } 178| | 179| | /* 180| | * We put in this value to ensure HuffDecode terminates. 181| | */ 182| 49| dctbl1.maxcode[17] = 0xFFFFFL; 183| | 184| | /* 185| | * Build the numbits, value lookup tables. 186| | * These table allow us to gather 8 bits from the bits stream, 187| | * and immediately lookup the size and value of the huffman codes. 188| | * If size is zero, it means that more than 8 bits are in the huffman 189| | * code (this happens about 3-4% of the time). 190| | */ 191| 49| dctbl1.numbits.fill(0); 192| 755| for (p = 0; p < lastp; p++) { ------------------ | Branch (192:17): [True: 706, False: 49] ------------------ 193| 706| size = huffsize[p]; 194| 706| if (size <= 8) { ------------------ | Branch (194:11): [True: 539, False: 167] ------------------ 195| 539| value = dctbl1.huffval[p]; 196| 539| code = huffcode[p]; 197| 539| ll = code << (8 - size); 198| 539| if (size < 8) { ------------------ | Branch (198:13): [True: 490, False: 49] ------------------ 199| 490| ul = ll | bitMask[24 + size]; 200| 490| } else { 201| 49| ul = ll; 202| 49| } 203| 539| if (ul > 256 || ll > ul) ------------------ | Branch (203:13): [True: 0, False: 539] | Branch (203:25): [True: 0, False: 539] ------------------ 204| 539| ThrowRDE( ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 205| 539| "createPrefixCodeDecoder: Code length too long. Corrupt data."); 206| 13.0k| for (i = ll; i <= ul; i++) { ------------------ | Branch (206:22): [True: 12.4k, False: 539] ------------------ 207| 12.4k| dctbl1.numbits[i] = size | (value << 4); 208| 12.4k| } 209| 539| } 210| 706| } 211| 49| if (mUseBigtable) ------------------ | Branch (211:9): [True: 49, False: 0] ------------------ 212| 49| createBigTable(); 213| 49| dctbl1.initialized = true; 214| 49| } NikonDecompressor.cpp:_ZN8rawspeed12_GLOBAL__N_120NikonLASDecompressor14createBigTableEv: 228| 49| void createBigTable() { 229| 49| const uint32_t bits = 230| 49| 14; // HuffDecode functions must be changed, if this is modified. 231| 49| const uint32_t size = 1 << bits; 232| 49| int rv = 0; 233| 49| int temp; 234| 49| uint32_t l; 235| | 236| 49| dctbl1.bigTable.resize(size); 237| 802k| for (uint32_t i = 0; i < size; i++) { ------------------ | Branch (237:26): [True: 802k, False: 49] ------------------ 238| 802k| auto input = implicit_cast(i << 2); // Calculate input value 239| 802k| int code = input >> 8; // Get 8 bits 240| 802k| uint32_t val = dctbl1.numbits[code]; 241| 802k| l = val & 15; 242| 802k| if (l) { ------------------ | Branch (242:11): [True: 799k, False: 3.13k] ------------------ 243| 799k| rv = val >> 4; 244| 799k| } else { 245| 3.13k| l = 8; 246| 8.16k| while (code > dctbl1.maxcode[l]) { ------------------ | Branch (246:16): [True: 5.02k, False: 3.13k] ------------------ 247| 5.02k| temp = extractHighBits(input, l, /*effectiveBitwidth=*/15) & 1; 248| 5.02k| code = (code << 1) | temp; 249| 5.02k| l++; 250| 5.02k| } 251| | 252| | /* 253| | * With garbage input we may reach the sentinel value l = 17. 254| | */ 255| | 256| 3.13k| if (l > 16 || dctbl1.valptr[l] == 0xff) { ------------------ | Branch (256:13): [True: 0, False: 3.13k] | Branch (256:23): [True: 0, False: 3.13k] ------------------ 257| 0| dctbl1.bigTable[i] = 0xff; 258| 0| continue; 259| 0| } 260| 3.13k| rv = dctbl1.huffval[dctbl1.valptr[l] + (code - dctbl1.mincode[l])]; 261| 3.13k| } 262| | 263| 802k| if (rv == 16) { ------------------ | Branch (263:11): [True: 0, False: 802k] ------------------ 264| 0| if (mDNGCompatible) ------------------ | Branch (264:13): [True: 0, False: 0] ------------------ 265| 0| dctbl1.bigTable[i] = (-(32768 << 8)) | (16 + l); 266| 0| else 267| 0| dctbl1.bigTable[i] = (-(32768 << 8)) | l; 268| 0| continue; 269| 0| } 270| | 271| 802k| if (rv + l > bits) { ------------------ | Branch (271:11): [True: 522k, False: 280k] ------------------ 272| 522k| dctbl1.bigTable[i] = 0xff; 273| 522k| continue; 274| 522k| } 275| | 276| 280k| if (rv) { ------------------ | Branch (276:11): [True: 278k, False: 2.17k] ------------------ 277| 278k| int x = extractHighBits(input, l + rv) & ((1 << rv) - 1); 278| 278k| if ((x & (1 << (rv - 1))) == 0) ------------------ | Branch (278:13): [True: 139k, False: 139k] ------------------ 279| 139k| x -= (1 << rv) - 1; 280| 278k| dctbl1.bigTable[i] = 281| 278k| static_cast((static_cast(x) << 8) | (l + rv)); 282| 278k| } else { 283| 2.17k| dctbl1.bigTable[i] = l; 284| 2.17k| } 285| 280k| } 286| 49| } NikonDecompressor.cpp:_ZN8rawspeed12_GLOBAL__N_120NikonLASDecompressor16decodeDifferenceERNS_14BitStreamerMSBE: 325| 454k| int decodeDifference(BitStreamerMSB& bits) { 326| 454k| int rv; 327| 454k| int l; 328| 454k| int temp; 329| 454k| int code; 330| 454k| unsigned val; 331| | 332| 454k| bits.fill(); 333| 454k| code = bits.peekBitsNoFill(14); 334| 454k| val = static_cast(dctbl1.bigTable[code]); 335| 454k| if ((val & 0xff) != 0xff) { ------------------ | Branch (335:9): [True: 292k, False: 161k] ------------------ 336| 292k| bits.skipBitsNoFill(val & 0xff); 337| 292k| return static_cast(val) >> 8; 338| 292k| } 339| | 340| 161k| rv = 0; 341| 161k| code = bits.peekBitsNoFill(8); 342| 161k| val = dctbl1.numbits[code]; 343| 161k| l = val & 15; 344| 161k| if (l) { ------------------ | Branch (344:9): [True: 152k, False: 9.09k] ------------------ 345| 152k| bits.skipBitsNoFill(l); 346| 152k| rv = static_cast(val) >> 4; 347| 152k| } else { 348| 9.09k| bits.skipBitsNoFill(8); 349| 9.09k| l = 8; 350| 36.0k| while (code > dctbl1.maxcode[l]) { ------------------ | Branch (350:14): [True: 26.9k, False: 9.09k] ------------------ 351| 26.9k| temp = bits.getBitsNoFill(1); 352| 26.9k| code = (code << 1) | temp; 353| 26.9k| l++; 354| 26.9k| } 355| | 356| 9.09k| if (l > 16) { ------------------ | Branch (356:11): [True: 0, False: 9.09k] ------------------ 357| 0| ThrowRDE("Corrupt JPEG data: bad Huffman code:%d\n", l); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 358| 9.09k| } else { 359| 9.09k| rv = dctbl1.huffval[dctbl1.valptr[l] + (code - dctbl1.mincode[l])]; 360| 9.09k| } 361| 9.09k| } 362| | 363| 161k| if (rv == 16) ------------------ | Branch (363:9): [True: 0, False: 161k] ------------------ 364| 0| return -32768; 365| | 366| | /* 367| | * Section F.2.2.1: decode the difference and 368| | * Figure F.12: extend sign bit 369| | */ 370| 161k| uint32_t len = rv & 15; 371| 161k| uint32_t shl = rv >> 4; 372| 161k| int diff = ((bits.getBits(len - shl) << 1) + 1) << shl >> 1; 373| 161k| if ((diff & (1 << (len - 1))) == 0) ------------------ | Branch (373:9): [True: 89.0k, False: 72.6k] ------------------ 374| 89.0k| diff -= (1 << len) - !shl; 375| 161k| return diff; 376| 161k| } _ZN8rawspeed19OlympusDecompressorC2ENS_8RawImageE: 218| 142|OlympusDecompressor::OlympusDecompressor(RawImage img) : mRaw(std::move(img)) { 219| 142| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (219:7): [True: 0, False: 142] | Branch (219:30): [True: 0, False: 142] ------------------ 220| 142| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (220:7): [True: 0, False: 142] ------------------ 221| 142| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 222| | 223| 142| if (!mRaw->dim.hasPositiveArea() || mRaw->dim.x % 2 != 0 || ------------------ | Branch (223:7): [True: 0, False: 142] | Branch (223:39): [True: 0, False: 142] ------------------ 224| 142| mRaw->dim.y % 2 != 0 || mRaw->dim.x > 10400 || mRaw->dim.y > 7792) ------------------ | Branch (224:7): [True: 7, False: 135] | Branch (224:31): [True: 0, False: 135] | Branch (224:54): [True: 2, False: 133] ------------------ 225| 142| ThrowRDE("Unexpected image dimensions found: (%d; %d)", mRaw->dim.x, ------------------ | | 38| 9| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 9| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 9| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 9| __VA_ARGS__) | | ------------------ ------------------ 226| 142| mRaw->dim.y); 227| 142|} _ZNK8rawspeed19OlympusDecompressor10decompressERKNS_10ByteStreamE: 229| 133|void OlympusDecompressor::decompress(const ByteStream& input) const { 230| 133| OlympusDecompressorImpl(mRaw).decompress(input); 231| 133|} OlympusDecompressor.cpp:_ZN8rawspeed12_GLOBAL__N_123OlympusDecompressorImplC2ENS_8RawImageE: 118| 133| explicit OlympusDecompressorImpl(RawImage img) : mRaw(std::move(img)) {} OlympusDecompressor.cpp:_ZNK8rawspeed12_GLOBAL__N_123OlympusDecompressorImpl5numLZMUlmjE_clEmj: 104| 544k| [](size_t i, [[maybe_unused]] unsigned tableSize) { 105| 544k| return 12 - numActiveBits(i); 106| 544k| }}; OlympusDecompressor.cpp:_ZNK8rawspeed12_GLOBAL__N_123OlympusDecompressorImpl10decompressENS_10ByteStreamE: 205| 133|void OlympusDecompressorImpl::decompress(ByteStream input) const { 206| 133| invariant(mRaw->dim.y > 0); ------------------ | | 27| 133|#define invariant(expr) assert(expr) ------------------ | Branch (206:3): [True: 133, False: 0] ------------------ 207| 133| invariant(mRaw->dim.x > 0); ------------------ | | 27| 133|#define invariant(expr) assert(expr) ------------------ | Branch (207:3): [True: 133, False: 0] ------------------ 208| 133| invariant(mRaw->dim.x % 2 == 0); ------------------ | | 27| 133|#define invariant(expr) assert(expr) ------------------ | Branch (208:3): [True: 133, False: 0] ------------------ 209| | 210| 133| BitStreamerMSB bits(input.peekRemainingBuffer().getAsArray1DRef()); 211| | 212| 12.7k| for (int y = 0; y < mRaw->dim.y; y++) ------------------ | Branch (212:19): [True: 12.6k, False: 133] ------------------ 213| 12.6k| decompressRow(bits, y); 214| 133|} OlympusDecompressor.cpp:_ZNK8rawspeed12_GLOBAL__N_123OlympusDecompressorImpl13decompressRowERNS_14BitStreamerMSBEi: 185| 12.6k| int row) const { 186| 12.6k| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 187| | 188| 12.6k| invariant(out.width() > 0); ------------------ | | 27| 12.6k|#define invariant(expr) assert(expr) ------------------ | Branch (188:3): [True: 12.6k, False: 0] ------------------ 189| 12.6k| invariant(out.width() % 2 == 0); ------------------ | | 27| 12.6k|#define invariant(expr) assert(expr) ------------------ | Branch (189:3): [True: 12.6k, False: 0] ------------------ 190| | 191| 12.6k| std::array acarry{numLZ, numLZ}; 192| | 193| 12.6k| const int numGroups = out.width() / 2; 194| 12.6k| int group = 0; 195| 12.6k| { 196| | // Process first group separately, allows to unswitch predictor calculation. 197| 12.6k| decompressGroup(acarry, bits, row, group); 198| 12.6k| ++group; 199| 12.6k| } 200| 728k| for (; group != numGroups; ++group) { ------------------ | Branch (200:10): [True: 716k, False: 12.6k] ------------------ 201| 716k| decompressGroup(acarry, bits, row, group); 202| 716k| } 203| 12.6k|} OlympusDecompressor.cpp:_ZN8rawspeed12_GLOBAL__N_124OlympusDifferenceDecoderC2ERKNS_9SimpleLUTIaLi12EEE: 56| 25.2k| : numLZ(numLZ_) {} OlympusDecompressor.cpp:_ZNK8rawspeed12_GLOBAL__N_123OlympusDecompressorImpl15decompressGroupERNSt3__15arrayINS0_24OlympusDifferenceDecoderELm2EEERNS_14BitStreamerMSBEii: 170| 728k| int row, int group) const { 171| 728k| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 172| | 173| 2.18M| for (int c = 0; c != 2; ++c) { ------------------ | Branch (173:19): [True: 1.45M, False: 728k] ------------------ 174| 1.45M| const int col = (2 * group) + c; 175| 1.45M| OlympusDifferenceDecoder& carry = acarry[c]; 176| | 177| 1.45M| int diff = carry.getDiff(bits); 178| 1.45M| int pred = getPred(out, row, col); 179| | 180| 1.45M| out(row, col) = implicit_cast(pred + diff); 181| 1.45M| } 182| 728k|} OlympusDecompressor.cpp:_ZN8rawspeed12_GLOBAL__N_124OlympusDifferenceDecoder7getDiffERNS_14BitStreamerMSBE: 62| 1.45M|OlympusDifferenceDecoder::getDiff(BitStreamerMSB& bits) { 63| 1.45M| bits.fill(); 64| | 65| 1.45M| int numLowBitsBias = (carry[2] < 3) ? 2 : 0; ------------------ | Branch (65:24): [True: 996k, False: 460k] ------------------ 66| 1.45M| int numLowBits = numActiveBits(implicit_cast(carry[0])); 67| 1.45M| numLowBits -= numLowBitsBias; 68| 1.45M| numLowBits = std::max(numLowBits, 2 + numLowBitsBias); 69| 1.45M| assert(numLowBits >= 2); ------------------ | Branch (69:3): [True: 1.45M, False: 129] ------------------ 70| 1.45M| assert(numLowBits <= 14); ------------------ | Branch (70:3): [True: 1.45M, False: 0] ------------------ 71| | 72| 1.45M| int b = bits.peekBitsNoFill(15); 73| 1.45M| int sign = (b >> 14) * -1; 74| 1.45M| int low = (b >> 12) & 3; 75| 1.45M| int numLeadingZeros = numLZ[b & 4095]; 76| | 77| 1.45M| int highBits; 78| | // Skip bytes used above or read bits 79| 1.45M| if (numLeadingZeros != 12) [[likely]] { // Happens in 99.9% of cases. ------------------ | Branch (79:7): [True: 1.35M, False: 102k] ------------------ 80| 1.35M| bits.skipBitsNoFill(numLeadingZeros + 1 + 3); 81| 1.35M| highBits = numLeadingZeros; 82| 1.35M| } else { 83| 102k| bits.skipBitsNoFill(15); 84| 102k| int numHighBits = 15 - numLowBits; 85| 102k| assert(numHighBits >= 1); ------------------ | Branch (85:5): [True: 102k, False: 0] ------------------ 86| 102k| assert(numHighBits <= 13); ------------------ | Branch (86:5): [True: 102k, False: 0] ------------------ 87| 102k| highBits = bits.peekBitsNoFill(numHighBits); 88| 102k| bits.skipBitsNoFill(1 + numHighBits); 89| 102k| } 90| | 91| 1.45M| carry[0] = (highBits << numLowBits) | bits.getBitsNoFill(numLowBits); 92| 1.45M| int diff = (carry[0] ^ sign) + carry[1]; 93| 1.45M| carry[1] = (diff * 3 + carry[1]) >> 5; 94| 1.45M| carry[2] = carry[0] > 16 ? 0 : carry[2] + 1; ------------------ | Branch (94:14): [True: 641k, False: 815k] ------------------ 95| | 96| 1.45M| return (diff * 4) | low; 97| 1.45M|} OlympusDecompressor.cpp:_ZN8rawspeed12_GLOBAL__N_123OlympusDecompressorImpl7getPredENS_10Array2DRefItEEii: 132| 1.45M| int col) { 133| 1.45M| auto getLeft = [&]() { return out(row, col - 2); }; 134| 1.45M| auto getUp = [&]() { return out(row - 2, col); }; 135| 1.45M| auto getLeftUp = [&]() { return out(row - 2, col - 2); }; 136| | 137| 1.45M| int pred; 138| 1.45M| if (row < 2 && col < 2) ------------------ | Branch (138:7): [True: 93.2k, False: 1.36M] | Branch (138:18): [True: 467, False: 92.8k] ------------------ 139| 467| pred = 0; 140| 1.45M| else if (row < 2) ------------------ | Branch (140:12): [True: 92.8k, False: 1.36M] ------------------ 141| 92.8k| pred = getLeft(); 142| 1.36M| else if (col < 2) ------------------ | Branch (142:12): [True: 24.6k, False: 1.33M] ------------------ 143| 24.6k| pred = getUp(); 144| 1.33M| else { 145| 1.33M| int left = getLeft(); 146| 1.33M| int up = getUp(); 147| 1.33M| int leftUp = getLeftUp(); 148| | 149| 1.33M| int leftMinusNw = left - leftUp; 150| 1.33M| int upMinusNw = up - leftUp; 151| | 152| | // Check if sign is different, and they are both not zero 153| 1.33M| if ((std::signbit(leftMinusNw) != std::signbit(upMinusNw)) && ------------------ | Branch (153:9): [True: 641k, False: 698k] ------------------ 154| 641k| (leftMinusNw != 0 && upMinusNw != 0)) { ------------------ | Branch (154:10): [True: 622k, False: 18.3k] | Branch (154:30): [True: 619k, False: 3.01k] ------------------ 155| 619k| if (std::abs(leftMinusNw) > 32 || std::abs(upMinusNw) > 32) ------------------ | Branch (155:11): [True: 471k, False: 148k] | Branch (155:41): [True: 142k, False: 5.44k] ------------------ 156| 614k| pred = left + upMinusNw; 157| 5.44k| else 158| 5.44k| pred = (left + up) >> 1; 159| 719k| } else { 160| 719k| pred = std::abs(leftMinusNw) > std::abs(upMinusNw) ? left : up; ------------------ | Branch (160:14): [True: 128k, False: 590k] ------------------ 161| 719k| } 162| 1.33M| } 163| | 164| 1.45M| return pred; 165| 1.45M|} OlympusDecompressor.cpp:_ZZN8rawspeed12_GLOBAL__N_123OlympusDecompressorImpl7getPredENS_10Array2DRefItEEiiENKUlvE_clEv: 133| 1.43M| auto getLeft = [&]() { return out(row, col - 2); }; OlympusDecompressor.cpp:_ZZN8rawspeed12_GLOBAL__N_123OlympusDecompressorImpl7getPredENS_10Array2DRefItEEiiENKUlvE0_clEv: 134| 1.36M| auto getUp = [&]() { return out(row - 2, col); }; OlympusDecompressor.cpp:_ZZN8rawspeed12_GLOBAL__N_123OlympusDecompressorImpl7getPredENS_10Array2DRefItEEiiENKUlvE1_clEv: 135| 1.33M| auto getLeftUp = [&]() { return out(row - 2, col - 2); }; _ZN8rawspeed23PanasonicV4DecompressorC2ENS_8RawImageENS_10ByteStreamEbj: 53| 183| : mRaw(std::move(img)), zero_is_bad(!zero_is_not_bad), 54| 183| section_split_offset(section_split_offset_) { 55| 183| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (55:7): [True: 0, False: 183] | Branch (55:30): [True: 0, False: 183] ------------------ 56| 183| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (56:7): [True: 0, False: 183] ------------------ 57| 183| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 58| | 59| 183| if (!mRaw->dim.hasPositiveArea() || mRaw->dim.x % PixelsPerPacket != 0) { ------------------ | Branch (59:7): [True: 2, False: 181] | Branch (59:39): [True: 8, False: 173] ------------------ 60| 10| ThrowRDE("Unexpected image dimensions found: (%i; %i)", mRaw->dim.x, ------------------ | | 38| 10| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 10| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 10| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 10| __VA_ARGS__) | | ------------------ ------------------ 61| 10| mRaw->dim.y); 62| 10| } 63| | 64| 173| if (BlockSize < section_split_offset) ------------------ | Branch (64:7): [True: 0, False: 173] ------------------ 65| 173| ThrowRDE("Bad section_split_offset: %u, less than BlockSize (%u)", ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 66| 173| section_split_offset, BlockSize); 67| | 68| | // Naive count of bytes that given pixel count requires. 69| 173| invariant(mRaw->dim.area() % PixelsPerPacket == 0); ------------------ | | 27| 173|#define invariant(expr) assert(expr) ------------------ | Branch (69:3): [True: 173, False: 0] ------------------ 70| 173| const auto bytesTotal = (mRaw->dim.area() / PixelsPerPacket) * BytesPerPacket; 71| 173| invariant(bytesTotal > 0); ------------------ | | 27| 173|#define invariant(expr) assert(expr) ------------------ | Branch (71:3): [True: 173, False: 0] ------------------ 72| | 73| | // If section_split_offset is zero, then that we need to read the normal 74| | // amount of bytes. But if it is not, then we need to round up to multiple of 75| | // BlockSize, because of splitting&rotation of each BlockSize's slice in half 76| | // at section_split_offset bytes. 77| 173| const auto bufSize = 78| 173| section_split_offset == 0 ? bytesTotal : roundUp(bytesTotal, BlockSize); ------------------ | Branch (78:7): [True: 144, False: 29] ------------------ 79| | 80| 173| if (bufSize > std::numeric_limits::max()) ------------------ | Branch (80:7): [True: 1, False: 172] ------------------ 81| 173| ThrowRDE("Raw dimensions require input buffer larger than supported"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 82| | 83| 172| input = input_.peekStream(implicit_cast(bufSize)); 84| | 85| 172| chopInputIntoBlocks(); 86| 172|} _ZN8rawspeed23PanasonicV4Decompressor19chopInputIntoBlocksEv: 88| 151|void PanasonicV4Decompressor::chopInputIntoBlocks() { 89| 151| auto pixelToCoordinate = [width = mRaw->dim.x](unsigned pixel) { 90| 151| return iPoint2D(pixel % width, pixel / width); 91| 151| }; 92| | 93| | // If section_split_offset == 0, last block may not be full. 94| 151| const auto blocksTotal = 95| 151| roundUpDivisionSafe(input.getRemainSize(), BlockSize); 96| 151| invariant(blocksTotal > 0); ------------------ | | 27| 151|#define invariant(expr) assert(expr) ------------------ | Branch (96:3): [True: 151, False: 0] ------------------ 97| 151| invariant(blocksTotal * PixelsPerBlock >= mRaw->dim.area()); ------------------ | | 27| 151|#define invariant(expr) assert(expr) ------------------ | Branch (97:3): [True: 151, False: 0] ------------------ 98| 151| assert(blocksTotal <= std::numeric_limits::max()); ------------------ | Branch (98:3): [True: 151, False: 0] ------------------ 99| 151| assert(blocksTotal <= std::numeric_limits::max()); ------------------ | Branch (99:3): [True: 151, False: 0] ------------------ 100| 151| blocks.reserve(implicit_cast(blocksTotal)); 101| | 102| 151| unsigned currPixel = 0; 103| 151| std::generate_n( 104| 151| std::back_inserter(blocks), blocksTotal, [&, pixelToCoordinate]() { 105| 151| invariant(input.getRemainSize() != 0); 106| 151| const auto blockSize = std::min(input.getRemainSize(), BlockSize); 107| 151| invariant(blockSize > 0); 108| 151| invariant(blockSize % BytesPerPacket == 0); 109| 151| const auto packets = blockSize / BytesPerPacket; 110| 151| invariant(packets > 0); 111| 151| const auto pixels = packets * PixelsPerPacket; 112| 151| invariant(pixels > 0); 113| | 114| 151| ByteStream bs = input.getStream(blockSize); 115| 151| iPoint2D beginCoord = pixelToCoordinate(currPixel); 116| 151| currPixel += pixels; 117| 151| iPoint2D endCoord = pixelToCoordinate(currPixel); 118| 151| return Block(bs, beginCoord, endCoord); 119| 151| }); 120| 151| assert(blocks.size() == blocksTotal); ------------------ | Branch (120:3): [True: 151, False: 0] ------------------ 121| 151| assert(currPixel >= mRaw->dim.area()); ------------------ | Branch (121:3): [True: 151, False: 0] ------------------ 122| 151| assert(input.getRemainSize() == 0); ------------------ | Branch (122:3): [True: 151, False: 0] ------------------ 123| | 124| | // Clamp the end coordinate for the last block. 125| 151| blocks.back().endCoord = mRaw->dim; 126| 151| blocks.back().endCoord.y -= 1; 127| 151|} PanasonicV4Decompressor.cpp:_ZZN8rawspeed23PanasonicV4Decompressor19chopInputIntoBlocksEvENK3$_0clEv: 104| 1.05k| std::back_inserter(blocks), blocksTotal, [&, pixelToCoordinate]() { 105| 1.05k| invariant(input.getRemainSize() != 0); ------------------ | | 27| 1.05k|#define invariant(expr) assert(expr) ------------------ | Branch (105:9): [True: 1.05k, False: 0] ------------------ 106| 1.05k| const auto blockSize = std::min(input.getRemainSize(), BlockSize); 107| 1.05k| invariant(blockSize > 0); ------------------ | | 27| 1.05k|#define invariant(expr) assert(expr) ------------------ | Branch (107:9): [True: 1.05k, False: 0] ------------------ 108| 1.05k| invariant(blockSize % BytesPerPacket == 0); ------------------ | | 27| 1.05k|#define invariant(expr) assert(expr) ------------------ | Branch (108:9): [True: 1.05k, False: 0] ------------------ 109| 1.05k| const auto packets = blockSize / BytesPerPacket; 110| 1.05k| invariant(packets > 0); ------------------ | | 27| 1.05k|#define invariant(expr) assert(expr) ------------------ | Branch (110:9): [True: 1.05k, False: 0] ------------------ 111| 1.05k| const auto pixels = packets * PixelsPerPacket; 112| 1.05k| invariant(pixels > 0); ------------------ | | 27| 1.05k|#define invariant(expr) assert(expr) ------------------ | Branch (112:9): [True: 1.05k, False: 0] ------------------ 113| | 114| 1.05k| ByteStream bs = input.getStream(blockSize); 115| 1.05k| iPoint2D beginCoord = pixelToCoordinate(currPixel); 116| 1.05k| currPixel += pixels; 117| 1.05k| iPoint2D endCoord = pixelToCoordinate(currPixel); 118| 1.05k| return Block(bs, beginCoord, endCoord); 119| 1.05k| }); PanasonicV4Decompressor.cpp:_ZZN8rawspeed23PanasonicV4Decompressor19chopInputIntoBlocksEvENK3$_1clEj: 89| 2.10k| auto pixelToCoordinate = [width = mRaw->dim.x](unsigned pixel) { 90| 2.10k| return iPoint2D(pixel % width, pixel / width); 91| 2.10k| }; _ZNK8rawspeed23PanasonicV4Decompressor12processBlockERKNS0_5BlockEPNSt3__16vectorIjNS4_9allocatorIjEEEE: 221| 1.03k| const Block& block, std::vector* zero_pos) const noexcept { 222| 1.03k| ProxyStream bits(block.bs, section_split_offset); 223| | 224| 689k| for (int row = block.beginCoord.y; row <= block.endCoord.y; row++) { ------------------ | Branch (224:38): [True: 687k, False: 1.50k] ------------------ 225| 687k| int col = 0; 226| | // First row may not begin at the first column. 227| 687k| if (block.beginCoord.y == row) ------------------ | Branch (227:9): [True: 1.03k, False: 686k] ------------------ 228| 1.03k| col = block.beginCoord.x; 229| | 230| 687k| int endCol = mRaw->dim.x; 231| | // Last row may end before the last column. 232| 687k| if (block.endCoord.y == row) ------------------ | Branch (232:9): [True: 1.04k, False: 686k] ------------------ 233| 1.04k| endCol = block.endCoord.x; 234| | 235| 687k| invariant(col % PixelsPerPacket == 0); ------------------ | | 27| 687k|#define invariant(expr) assert(expr) ------------------ | Branch (235:5): [True: 688k, False: 18.4E] ------------------ 236| 688k| invariant(endCol % PixelsPerPacket == 0); ------------------ | | 27| 688k|#define invariant(expr) assert(expr) ------------------ | Branch (236:5): [True: 688k, False: 18.4E] ------------------ 237| | 238| 1.59M| for (; col < endCol; col += PixelsPerPacket) ------------------ | Branch (238:12): [True: 908k, False: 688k] ------------------ 239| 908k| processPixelPacket(bits, row, col, zero_pos); 240| 688k| } 241| 1.03k|} _ZN8rawspeed23PanasonicV4Decompressor11ProxyStreamC2ENS_10ByteStreamEi: 162| 1.04k| : block(block_), section_split_offset(section_split_offset_) { 163| 1.04k| parseBlock(); 164| 1.04k| } _ZN8rawspeed23PanasonicV4Decompressor11ProxyStream10parseBlockEv: 136| 1.02k| void parseBlock() { 137| 1.02k| assert(buf.empty()); ------------------ | Branch (137:5): [True: 1.03k, False: 18.4E] ------------------ 138| 1.03k| invariant(block.getRemainSize() <= BlockSize); ------------------ | | 27| 1.03k|#define invariant(expr) assert(expr) ------------------ | Branch (138:5): [True: 1.03k, False: 2] ------------------ 139| 1.03k| invariant(section_split_offset <= BlockSize); ------------------ | | 27| 1.03k|#define invariant(expr) assert(expr) ------------------ | Branch (139:5): [True: 1.03k, False: 0] ------------------ 140| | 141| 1.03k| Buffer FirstSection = block.getBuffer(section_split_offset); 142| 1.03k| Buffer SecondSection = block.getBuffer(block.getRemainSize()); 143| | 144| | // get one more byte, so the return statement of getBits does not have 145| | // to special case for accessing the last byte 146| 1.03k| buf.reserve(BlockSize + 1UL); 147| | 148| | // First copy the second section. This makes it the first section. 149| 1.03k| buf.insert(buf.end(), SecondSection.begin(), SecondSection.end()); 150| | // Now append the original 1'st section right after the new 1'st section. 151| 1.03k| buf.insert(buf.end(), FirstSection.begin(), FirstSection.end()); 152| | 153| 1.03k| invariant(block.getRemainSize() == 0); ------------------ | | 27| 1.03k|#define invariant(expr) assert(expr) ------------------ | Branch (153:5): [True: 1.04k, False: 18.4E] ------------------ 154| | 155| | // get one more byte, so the return statement of getBits does not have 156| | // to special case for accessing the last byte 157| 1.04k| buf.emplace_back(0); 158| 1.04k| } _ZNK8rawspeed23PanasonicV4Decompressor18processPixelPacketERNS0_11ProxyStreamEiiPNSt3__16vectorIjNS3_9allocatorIjEEEE: 175| 908k| std::vector* zero_pos) const noexcept { 176| 908k| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 177| | 178| 908k| int sh = 0; 179| | 180| 908k| std::array pred; 181| 908k| pred.fill(0); 182| | 183| 908k| std::array nonz; 184| 908k| nonz.fill(0); 185| | 186| 908k| int u = 0; 187| | 188| 9.62M| for (int p = 0; p < PixelsPerPacket; ++p, ++col) { ------------------ | Branch (188:19): [True: 8.71M, False: 908k] ------------------ 189| 8.71M| const int c = p & 1; 190| | 191| | // FIXME: this is actually just `p % 3 == 2`, cleanup after perf is good. 192| 8.71M| if (u == 2) { ------------------ | Branch (192:9): [True: 3.21M, False: 5.50M] ------------------ 193| 3.21M| sh = extractHighBits(4U, bits.getBits(2), /*effectiveBitwidth=*/3); 194| 3.21M| u = -1; 195| 3.21M| } 196| | 197| 8.71M| if (nonz[c]) { ------------------ | Branch (197:9): [True: 3.35M, False: 5.36M] ------------------ 198| 3.35M| int j = bits.getBits(8); 199| 3.35M| if (j) { ------------------ | Branch (199:11): [True: 2.95M, False: 397k] ------------------ 200| 2.95M| pred[c] -= 0x80 << sh; 201| 2.95M| if (pred[c] < 0 || sh == 4) ------------------ | Branch (201:13): [True: 463k, False: 2.49M] | Branch (201:28): [True: 1.08M, False: 1.40M] ------------------ 202| 1.79M| pred[c] &= ~(-(1 << sh)); 203| 2.95M| pred[c] += j << sh; 204| 2.95M| } 205| 5.36M| } else { 206| 5.36M| nonz[c] = bits.getBits(8); 207| 6.11M| if (nonz[c] || p > 11) ------------------ | Branch (207:11): [True: 18.4E, False: 6.11M] | Branch (207:22): [True: 605k, False: 5.50M] ------------------ 208| 1.76M| pred[c] = nonz[c] << 4 | bits.getBits(4); 209| 5.36M| } 210| | 211| 8.71M| out(row, col) = implicit_cast(pred[c]); 212| | 213| 8.71M| if (zero_is_bad && 0 == pred[c]) ------------------ | Branch (213:9): [True: 8.47M, False: 242k] | Branch (213:24): [True: 5.86M, False: 2.61M] ------------------ 214| 5.86M| zero_pos->push_back((row << 16) | col); 215| | 216| 8.71M| u++; 217| 8.71M| } 218| 908k|} _ZN8rawspeed23PanasonicV4Decompressor11ProxyStream7getBitsEi: 166| 11.6M| uint32_t getBits(int nbits) noexcept { 167| 11.6M| vbits = (vbits - nbits) & 0x1ffff; 168| 11.6M| int byte = vbits >> 3 ^ 0x3ff0; 169| 11.6M| return (buf[byte] | buf[byte + 1UL] << 8) >> (vbits & 7) & ~(-(1 << nbits)); 170| 11.6M| } _ZNK8rawspeed23PanasonicV4Decompressor16decompressThreadEv: 243| 10.5k|void PanasonicV4Decompressor::decompressThread() const noexcept { 244| 10.5k| std::vector zero_pos; 245| | 246| 10.5k| assert(!blocks.empty()); ------------------ | Branch (246:3): [True: 4.78k, False: 5.72k] ------------------ 247| | 248| 4.78k|#ifdef HAVE_OPENMP 249| 4.78k|#pragma omp for schedule(static) 250| 4.78k|#endif 251| 4.78k| for (const auto& block : 252| 4.78k| Array1DRef(blocks.data(), implicit_cast(blocks.size()))) { 253| 4.78k| try { 254| 4.78k| processBlock(block, &zero_pos); 255| 4.78k| } catch (...) { 256| | // We should not get any exceptions here. 257| 4.78k| __builtin_unreachable(); 258| 4.78k| } 259| 4.78k| } 260| | 261| 4.78k| if (zero_is_bad && !zero_pos.empty()) { ------------------ | Branch (261:7): [True: 4.70k, False: 82] | Branch (261:22): [True: 789, False: 3.91k] ------------------ 262| 789| MutexLocker guard(&mRaw->mBadPixelMutex); 263| 789| mRaw->mBadPixelPositions.insert(mRaw->mBadPixelPositions.end(), 264| 789| zero_pos.begin(), zero_pos.end()); 265| 789| } 266| 4.78k|} _ZNK8rawspeed23PanasonicV4Decompressor10decompressEv: 268| 151|void PanasonicV4Decompressor::decompress() const noexcept { 269| 151| assert(!blocks.empty()); ------------------ | Branch (269:3): [True: 151, False: 0] ------------------ 270| 151|#ifdef HAVE_OPENMP 271| 151|#pragma omp parallel default(none) \ 272| 151| num_threads(rawspeed_get_number_of_processor_cores()) 273| 151|#endif 274| 151| decompressThread(); 275| 151|} PanasonicV4Decompressor.cpp:_ZNK8rawspeed23PanasonicV4Decompressor10decompressEv.omp_outlined_debug__: 274| 4.75k| decompressThread(); _ZN8rawspeed23PanasonicV4Decompressor5BlockC2ENS_10ByteStreamENS_8iPoint2DES3_: 66| 1.05k| : bs(bs_), beginCoord(beginCoord_), endCoord(endCoord_) {} _ZN8rawspeed23PanasonicV5DecompressorC2ENS_8RawImageENS_10ByteStreamEj: 73| 116| : mRaw(std::move(img)), bps(bps_) { 74| 116| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (74:7): [True: 0, False: 116] | Branch (74:30): [True: 0, False: 116] ------------------ 75| 116| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (75:7): [True: 0, False: 116] ------------------ 76| 116| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 77| | 78| 116| const PacketDsc* dsc = nullptr; 79| 116| switch (bps) { 80| 72| case 12: ------------------ | Branch (80:3): [True: 72, False: 44] ------------------ 81| 72| dsc = &TwelveBitPacket; 82| 72| break; 83| 43| case 14: ------------------ | Branch (83:3): [True: 43, False: 73] ------------------ 84| 43| dsc = &FourteenBitPacket; 85| 43| break; 86| 1| default: ------------------ | Branch (86:3): [True: 1, False: 115] ------------------ 87| 1| ThrowRDE("Unsupported bps: %u", bps); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 88| 116| } 89| | 90| 115| if (!mRaw->dim.hasPositiveArea() || mRaw->dim.x % dsc->pixelsPerPacket != 0) { ------------------ | Branch (90:7): [True: 2, False: 113] | Branch (90:39): [True: 7, False: 106] ------------------ 91| 9| ThrowRDE("Unexpected image dimensions found: (%i; %i)", mRaw->dim.x, ------------------ | | 38| 9| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 9| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 9| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 9| __VA_ARGS__) | | ------------------ ------------------ 92| 9| mRaw->dim.y); 93| 9| } 94| | 95| | // How many pixel packets does the specified pixel count require? 96| 106| invariant(mRaw->dim.area() % dsc->pixelsPerPacket == 0); ------------------ | | 27| 106|#define invariant(expr) assert(expr) ------------------ | Branch (96:3): [True: 106, False: 0] ------------------ 97| 106| const auto numPackets = mRaw->dim.area() / dsc->pixelsPerPacket; 98| 106| invariant(numPackets > 0); ------------------ | | 27| 106|#define invariant(expr) assert(expr) ------------------ | Branch (98:3): [True: 106, False: 0] ------------------ 99| | 100| | // And how many blocks that would be? Last block may not be full, pad it. 101| 106| numBlocks = roundUpDivisionSafe(numPackets, PacketsPerBlock); 102| 106| invariant(numBlocks > 0); ------------------ | | 27| 106|#define invariant(expr) assert(expr) ------------------ | Branch (102:3): [True: 106, False: 0] ------------------ 103| | 104| | // Does the input contain enough blocks? 105| | // How many full blocks does the input contain? This is truncating division. 106| 106| if (const auto haveBlocks = input_.getRemainSize() / BlockSize; 107| 106| haveBlocks < numBlocks) ------------------ | Branch (107:7): [True: 28, False: 78] ------------------ 108| 106| ThrowRDE("Insufficient count of input blocks for a given image"); ------------------ | | 38| 28| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 28| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 28| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 28| __VA_ARGS__) | | ------------------ ------------------ 109| | 110| | // We only want those blocks we need, no extras. 111| 78| input = 112| 78| input_.peekStream(implicit_cast(numBlocks), BlockSize); 113| | 114| 78| chopInputIntoBlocks(*dsc); 115| 78|} _ZN8rawspeed23PanasonicV5Decompressor19chopInputIntoBlocksERKNS0_9PacketDscE: 117| 78|void PanasonicV5Decompressor::chopInputIntoBlocks(const PacketDsc& dsc) { 118| 78| auto pixelToCoordinate = [width = mRaw->dim.x](unsigned pixel) { 119| 78| return iPoint2D(pixel % width, pixel / width); 120| 78| }; 121| | 122| 78| invariant(numBlocks * BlockSize == input.getRemainSize()); ------------------ | | 27| 78|#define invariant(expr) assert(expr) ------------------ | Branch (122:3): [True: 78, False: 0] ------------------ 123| 78| assert(numBlocks <= std::numeric_limits::max()); ------------------ | Branch (123:3): [True: 78, False: 0] ------------------ 124| 78| assert(numBlocks <= std::numeric_limits::max()); ------------------ | Branch (124:3): [True: 78, False: 0] ------------------ 125| 78| blocks.reserve(implicit_cast(numBlocks)); 126| | 127| 78| const auto pixelsPerBlock = dsc.pixelsPerPacket * PacketsPerBlock; 128| 78| invariant((numBlocks - 1U) * pixelsPerBlock < mRaw->dim.area()); ------------------ | | 27| 78|#define invariant(expr) assert(expr) ------------------ | Branch (128:3): [True: 78, False: 0] ------------------ 129| 78| invariant(numBlocks * pixelsPerBlock >= mRaw->dim.area()); ------------------ | | 27| 78|#define invariant(expr) assert(expr) ------------------ | Branch (129:3): [True: 78, False: 0] ------------------ 130| | 131| 78| unsigned currPixel = 0; 132| 78| std::generate_n(std::back_inserter(blocks), numBlocks, 133| 78| [&, pixelToCoordinate, pixelsPerBlock]() { 134| 78| ByteStream bs = input.getStream(BlockSize); 135| 78| iPoint2D beginCoord = pixelToCoordinate(currPixel); 136| 78| currPixel += pixelsPerBlock; 137| 78| iPoint2D endCoord = pixelToCoordinate(currPixel); 138| 78| return Block(bs, beginCoord, endCoord); 139| 78| }); 140| 78| assert(blocks.size() == numBlocks); ------------------ | Branch (140:3): [True: 78, False: 0] ------------------ 141| 78| invariant(currPixel >= mRaw->dim.area()); ------------------ | | 27| 78|#define invariant(expr) assert(expr) ------------------ | Branch (141:3): [True: 78, False: 0] ------------------ 142| 78| invariant(input.getRemainSize() == 0); ------------------ | | 27| 78|#define invariant(expr) assert(expr) ------------------ | Branch (142:3): [True: 78, False: 0] ------------------ 143| | 144| | // Clamp the end coordinate for the last block. 145| 78| blocks.back().endCoord = mRaw->dim; 146| 78| blocks.back().endCoord.y -= 1; 147| 78|} PanasonicV5Decompressor.cpp:_ZZN8rawspeed23PanasonicV5Decompressor19chopInputIntoBlocksERKNS0_9PacketDscEENK3$_0clEv: 133| 674| [&, pixelToCoordinate, pixelsPerBlock]() { 134| 674| ByteStream bs = input.getStream(BlockSize); 135| 674| iPoint2D beginCoord = pixelToCoordinate(currPixel); 136| 674| currPixel += pixelsPerBlock; 137| 674| iPoint2D endCoord = pixelToCoordinate(currPixel); 138| 674| return Block(bs, beginCoord, endCoord); 139| 674| }); PanasonicV5Decompressor.cpp:_ZZN8rawspeed23PanasonicV5Decompressor19chopInputIntoBlocksERKNS0_9PacketDscEENK3$_1clEj: 118| 1.34k| auto pixelToCoordinate = [width = mRaw->dim.x](unsigned pixel) { 119| 1.34k| return iPoint2D(pixel % width, pixel / width); 120| 1.34k| }; _ZNK8rawspeed23PanasonicV5Decompressor10decompressEv: 253| 78|void PanasonicV5Decompressor::decompress() const noexcept { 254| 78| switch (bps) { 255| 41| case 12: ------------------ | Branch (255:3): [True: 41, False: 37] ------------------ 256| 41| decompressInternal(); 257| 41| break; 258| 37| case 14: ------------------ | Branch (258:3): [True: 37, False: 41] ------------------ 259| 37| decompressInternal(); 260| 37| break; 261| 0| default: ------------------ | Branch (261:3): [True: 0, False: 78] ------------------ 262| 0| __builtin_unreachable(); 263| 78| } 264| 78|} _ZNK8rawspeed23PanasonicV5Decompressor18decompressInternalIL_ZNS0_15TwelveBitPacketEEEEvv: 237| 41|void PanasonicV5Decompressor::decompressInternal() const noexcept { 238| 41|#ifdef HAVE_OPENMP 239| 41|#pragma omp parallel for num_threads(rawspeed_get_number_of_processor_cores()) \ 240| 41| schedule(static) default(none) 241| 41|#endif 242| 41| for (const auto& block : 243| 41| Array1DRef(blocks.data(), implicit_cast(blocks.size()))) { 244| 41| try { 245| 41| processBlock(block); 246| 41| } catch (...) { 247| | // We should not get any exceptions here. 248| 41| __builtin_unreachable(); 249| 41| } 250| 41| } 251| 41|} PanasonicV5Decompressor.cpp:_ZNK8rawspeed23PanasonicV5Decompressor18decompressInternalIL_ZNS0_15TwelveBitPacketEEEEvv.omp_outlined_debug__: 242| 1.27k| for (const auto& block : ------------------ | Branch (242:26): [True: 1.27k, False: 367] ------------------ 243| 1.27k| Array1DRef(blocks.data(), implicit_cast(blocks.size()))) { 244| 1.27k| try { 245| 1.27k| processBlock(block); 246| 1.27k| } catch (...) { 247| | // We should not get any exceptions here. 248| 0| __builtin_unreachable(); 249| 0| } 250| 1.27k| } _ZNK8rawspeed23PanasonicV5Decompressor12processBlockIL_ZNS0_15TwelveBitPacketEEEEvRKNS0_5BlockE: 210| 359|void PanasonicV5Decompressor::processBlock(const Block& block) const { 211| 359| static_assert(dsc.pixelsPerPacket > 0, "dsc should be compile-time const"); 212| 359| static_assert(BlockSize % bytesPerPacket == 0); 213| | 214| 359| ProxyStream proxy(block.bs); 215| 359| BitStreamerLSB bs(proxy.getStream().peekRemainingBuffer().getAsArray1DRef()); 216| | 217| 332k| for (int row = block.beginCoord.y; row <= block.endCoord.y; row++) { ------------------ | Branch (217:38): [True: 332k, False: 403] ------------------ 218| 332k| int col = 0; 219| | // First row may not begin at the first column. 220| 332k| if (block.beginCoord.y == row) ------------------ | Branch (220:9): [True: 339, False: 332k] ------------------ 221| 339| col = block.beginCoord.x; 222| | 223| 332k| int endx = mRaw->dim.x; 224| | // Last row may end before the last column. 225| 332k| if (block.endCoord.y == row) ------------------ | Branch (225:9): [True: 367, False: 332k] ------------------ 226| 367| endx = block.endCoord.x; 227| | 228| 332k| invariant(col % dsc.pixelsPerPacket == 0); ------------------ | | 27| 332k|#define invariant(expr) assert(expr) ------------------ | Branch (228:5): [True: 332k, False: 18.4E] ------------------ 229| 332k| invariant(endx % dsc.pixelsPerPacket == 0); ------------------ | | 27| 332k|#define invariant(expr) assert(expr) ------------------ | Branch (229:5): [True: 332k, False: 17] ------------------ 230| | 231| 678k| for (; col < endx; col += dsc.pixelsPerPacket) ------------------ | Branch (231:12): [True: 346k, False: 332k] ------------------ 232| 346k| processPixelPacket(bs, row, col); 233| 332k| } 234| 359|} _ZN8rawspeed23PanasonicV5Decompressor11ProxyStreamC2ENS_10ByteStreamE: 182| 655| explicit ProxyStream(ByteStream block_) : block(block_) {} _ZN8rawspeed23PanasonicV5Decompressor11ProxyStream9getStreamEv: 184| 658| ByteStream& getStream() { 185| 658| parseBlock(); 186| 658| return input; 187| 658| } _ZN8rawspeed23PanasonicV5Decompressor11ProxyStream10parseBlockEv: 154| 655| void parseBlock() { 155| 655| assert(buf.empty()); ------------------ | Branch (155:5): [True: 653, False: 2] ------------------ 156| 653| invariant(block.getRemainSize() == BlockSize); ------------------ | | 27| 653|#define invariant(expr) assert(expr) ------------------ | Branch (156:5): [True: 651, False: 2] ------------------ 157| | 158| 651| static_assert(BlockSize > sectionSplitOffset); 159| | 160| 651| Buffer FirstSection = block.getBuffer(sectionSplitOffset); 161| 651| Buffer SecondSection = block.getBuffer(block.getRemainSize()); 162| 651| invariant(FirstSection.getSize() < SecondSection.getSize()); ------------------ | | 27| 651|#define invariant(expr) assert(expr) ------------------ | Branch (162:5): [True: 643, False: 8] ------------------ 163| | 164| 643| buf.reserve(BlockSize); 165| | 166| | // First copy the second section. This makes it the first section. 167| 643| buf.insert(buf.end(), SecondSection.begin(), SecondSection.end()); 168| | // Now append the original 1'st section right after the new 1'st section. 169| 643| buf.insert(buf.end(), FirstSection.begin(), FirstSection.end()); 170| | 171| 643| assert(buf.size() == BlockSize); ------------------ | Branch (171:5): [True: 673, False: 18.4E] ------------------ 172| 673| invariant(block.getRemainSize() == 0); ------------------ | | 27| 673|#define invariant(expr) assert(expr) ------------------ | Branch (172:5): [True: 665, False: 8] ------------------ 173| | 174| | // And reset the clock. 175| 665| input = ByteStream(DataBuffer( 176| 665| Buffer(buf.data(), implicit_cast(buf.size())), 177| 665| Endianness::little)); 178| | // input.setByteOrder(Endianness::big); // does not seem to matter?! 179| 665| } _ZNK8rawspeed23PanasonicV5Decompressor18processPixelPacketIL_ZNS0_15TwelveBitPacketEEEEvRNS_14BitStreamerLSBEii: 193| 345k| int col) const { 194| 345k| static_assert(dsc.pixelsPerPacket > 0, "dsc should be compile-time const"); 195| 345k| static_assert(dsc.bps > 0 && dsc.bps <= 16); 196| | 197| 345k| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 198| | 199| 345k| invariant(bs.getFillLevel() == 0); ------------------ | | 27| 345k|#define invariant(expr) assert(expr) ------------------ | Branch (199:3): [True: 345k, False: 510] ------------------ 200| | 201| 1.65M| for (int p = 0; p < dsc.pixelsPerPacket;) { ------------------ | Branch (201:19): [True: 1.30M, False: 345k] ------------------ 202| 1.30M| bs.fill(); 203| 4.29M| for (; bs.getFillLevel() >= implicit_cast(dsc.bps); ++p, ++col) ------------------ | Branch (203:12): [True: 2.98M, False: 1.30M] ------------------ 204| 2.98M| out(row, col) = implicit_cast(bs.getBitsNoFill(dsc.bps)); 205| 1.30M| } 206| 345k| bs.skipBitsNoFill(bs.getFillLevel()); // get rid of padding. 207| 345k|} _ZNK8rawspeed23PanasonicV5Decompressor18decompressInternalIL_ZNS0_17FourteenBitPacketEEEEvv: 237| 37|void PanasonicV5Decompressor::decompressInternal() const noexcept { 238| 37|#ifdef HAVE_OPENMP 239| 37|#pragma omp parallel for num_threads(rawspeed_get_number_of_processor_cores()) \ 240| 37| schedule(static) default(none) 241| 37|#endif 242| 37| for (const auto& block : 243| 37| Array1DRef(blocks.data(), implicit_cast(blocks.size()))) { 244| 37| try { 245| 37| processBlock(block); 246| 37| } catch (...) { 247| | // We should not get any exceptions here. 248| 37| __builtin_unreachable(); 249| 37| } 250| 37| } 251| 37|} PanasonicV5Decompressor.cpp:_ZNK8rawspeed23PanasonicV5Decompressor18decompressInternalIL_ZNS0_17FourteenBitPacketEEEEvv.omp_outlined_debug__: 242| 1.15k| for (const auto& block : ------------------ | Branch (242:26): [True: 1.15k, False: 307] ------------------ 243| 1.15k| Array1DRef(blocks.data(), implicit_cast(blocks.size()))) { 244| 1.15k| try { 245| 1.15k| processBlock(block); 246| 1.15k| } catch (...) { 247| | // We should not get any exceptions here. 248| 0| __builtin_unreachable(); 249| 0| } 250| 1.15k| } _ZNK8rawspeed23PanasonicV5Decompressor12processBlockIL_ZNS0_17FourteenBitPacketEEEEvRKNS0_5BlockE: 210| 293|void PanasonicV5Decompressor::processBlock(const Block& block) const { 211| 293| static_assert(dsc.pixelsPerPacket > 0, "dsc should be compile-time const"); 212| 293| static_assert(BlockSize % bytesPerPacket == 0); 213| | 214| 293| ProxyStream proxy(block.bs); 215| 293| BitStreamerLSB bs(proxy.getStream().peekRemainingBuffer().getAsArray1DRef()); 216| | 217| 282k| for (int row = block.beginCoord.y; row <= block.endCoord.y; row++) { ------------------ | Branch (217:38): [True: 282k, False: 313] ------------------ 218| 282k| int col = 0; 219| | // First row may not begin at the first column. 220| 282k| if (block.beginCoord.y == row) ------------------ | Branch (220:9): [True: 279, False: 282k] ------------------ 221| 279| col = block.beginCoord.x; 222| | 223| 282k| int endx = mRaw->dim.x; 224| | // Last row may end before the last column. 225| 282k| if (block.endCoord.y == row) ------------------ | Branch (225:9): [True: 307, False: 282k] ------------------ 226| 307| endx = block.endCoord.x; 227| | 228| 282k| invariant(col % dsc.pixelsPerPacket == 0); ------------------ | | 27| 282k|#define invariant(expr) assert(expr) ------------------ | Branch (228:5): [True: 282k, False: 18.4E] ------------------ 229| 282k| invariant(endx % dsc.pixelsPerPacket == 0); ------------------ | | 27| 282k|#define invariant(expr) assert(expr) ------------------ | Branch (229:5): [True: 282k, False: 23] ------------------ 230| | 231| 570k| for (; col < endx; col += dsc.pixelsPerPacket) ------------------ | Branch (231:12): [True: 287k, False: 282k] ------------------ 232| 287k| processPixelPacket(bs, row, col); 233| 282k| } 234| 293|} _ZNK8rawspeed23PanasonicV5Decompressor18processPixelPacketIL_ZNS0_17FourteenBitPacketEEEEvRNS_14BitStreamerLSBEii: 193| 287k| int col) const { 194| 287k| static_assert(dsc.pixelsPerPacket > 0, "dsc should be compile-time const"); 195| 287k| static_assert(dsc.bps > 0 && dsc.bps <= 16); 196| | 197| 287k| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 198| | 199| 287k| invariant(bs.getFillLevel() == 0); ------------------ | | 27| 287k|#define invariant(expr) assert(expr) ------------------ | Branch (199:3): [True: 286k, False: 487] ------------------ 200| | 201| 1.37M| for (int p = 0; p < dsc.pixelsPerPacket;) { ------------------ | Branch (201:19): [True: 1.08M, False: 286k] ------------------ 202| 1.08M| bs.fill(); 203| 3.37M| for (; bs.getFillLevel() >= implicit_cast(dsc.bps); ++p, ++col) ------------------ | Branch (203:12): [True: 2.28M, False: 1.08M] ------------------ 204| 2.28M| out(row, col) = implicit_cast(bs.getBitsNoFill(dsc.bps)); 205| 1.08M| } 206| 286k| bs.skipBitsNoFill(bs.getFillLevel()); // get rid of padding. 207| 286k|} _ZN8rawspeed23PanasonicV5Decompressor5BlockC2ENS_10ByteStreamENS_8iPoint2DES3_: 82| 674| : bs(bs_), beginCoord(beginCoord_), endCoord(endCoord_) {} _ZN8rawspeed23PanasonicV6DecompressorC2ENS_8RawImageENS_10ByteStreamEj: 140| 121| : mRaw(std::move(img)), bps(bps_) { 141| 121| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (141:7): [True: 0, False: 121] | Branch (141:30): [True: 0, False: 121] ------------------ 142| 121| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (142:7): [True: 0, False: 121] ------------------ 143| 121| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 144| | 145| 121| const BlockDsc* dsc = nullptr; 146| 121| switch (bps) { 147| 74| case 12: ------------------ | Branch (147:3): [True: 74, False: 47] ------------------ 148| 74| dsc = &TwelveBitBlock; 149| 74| break; 150| 47| case 14: ------------------ | Branch (150:3): [True: 47, False: 74] ------------------ 151| 47| dsc = &FourteenBitBlock; 152| 47| break; 153| 0| default: ------------------ | Branch (153:3): [True: 0, False: 121] ------------------ 154| 0| ThrowRDE("Unsupported bps: %u", bps); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 155| 121| } 156| | 157| 121| if (!mRaw->dim.hasPositiveArea() || mRaw->dim.x % dsc->PixelsPerBlock != 0) { ------------------ | Branch (157:7): [True: 2, False: 119] | Branch (157:39): [True: 4, False: 115] ------------------ 158| 6| ThrowRDE("Unexpected image dimensions found: (%i; %i)", mRaw->dim.x, ------------------ | | 38| 6| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 6| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 6| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 6| __VA_ARGS__) | | ------------------ ------------------ 159| 6| mRaw->dim.y); 160| 6| } 161| | 162| | // How many blocks are needed for the given image size? 163| 115| const auto numBlocks = mRaw->dim.area() / dsc->PixelsPerBlock; 164| | 165| | // Does the input contain enough blocks? 166| | // How many full blocks does the input contain? This is truncating division. 167| 115| if (const auto haveBlocks = input_.getRemainSize() / dsc->BytesPerBlock; 168| 115| haveBlocks < numBlocks) ------------------ | Branch (168:7): [True: 16, False: 99] ------------------ 169| 115| ThrowRDE("Insufficient count of input blocks for a given image"); ------------------ | | 38| 16| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 16| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 16| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 16| __VA_ARGS__) | | ------------------ ------------------ 170| | 171| | // We only want those blocks we need, no extras. 172| 99| input = input_.peekStream(implicit_cast(numBlocks), 173| 99| dsc->BytesPerBlock); 174| 99|} _ZNK8rawspeed23PanasonicV6Decompressor10decompressEv: 250| 99|void PanasonicV6Decompressor::decompress() const noexcept { 251| 99| switch (bps) { 252| 54| case 12: ------------------ | Branch (252:3): [True: 54, False: 45] ------------------ 253| 54| decompressInternal(); 254| 54| break; 255| 45| case 14: ------------------ | Branch (255:3): [True: 45, False: 54] ------------------ 256| 45| decompressInternal(); 257| 45| break; 258| 0| default: ------------------ | Branch (258:3): [True: 0, False: 99] ------------------ 259| 0| __builtin_unreachable(); 260| 99| } 261| 99|} _ZNK8rawspeed23PanasonicV6Decompressor18decompressInternalIL_ZNS0_14TwelveBitBlockEEEEvv: 235| 54|void PanasonicV6Decompressor::decompressInternal() const noexcept { 236| 54|#ifdef HAVE_OPENMP 237| 54|#pragma omp parallel for num_threads(rawspeed_get_number_of_processor_cores()) \ 238| 54| schedule(static) default(none) 239| 54|#endif 240| 54| for (int row = 0; row < mRaw->dim.y; ++row) { 241| 54| try { 242| 54| decompressRow(row); 243| 54| } catch (...) { 244| | // We should not get any exceptions here. 245| 54| __builtin_unreachable(); 246| 54| } 247| 54| } 248| 54|} PanasonicV6Decompressor.cpp:_ZNK8rawspeed23PanasonicV6Decompressor18decompressInternalIL_ZNS0_14TwelveBitBlockEEEEvv.omp_outlined_debug__: 240| 1.70k| for (int row = 0; row < mRaw->dim.y; ++row) { ------------------ | Branch (240:21): [True: 1.70k, False: 0] ------------------ 241| 1.70k| try { 242| 1.70k| decompressRow(row); 243| 1.70k| } catch (...) { 244| | // We should not get any exceptions here. 245| 0| __builtin_unreachable(); 246| 0| } 247| 1.70k| } _ZNK8rawspeed23PanasonicV6Decompressor13decompressRowIL_ZNS0_14TwelveBitBlockEEEEvi: 223| 3.07k|void PanasonicV6Decompressor::decompressRow(int row) const noexcept { 224| 3.07k| invariant(mRaw->dim.x % dsc.PixelsPerBlock == 0); ------------------ | | 27| 3.07k|#define invariant(expr) assert(expr) ------------------ | Branch (224:3): [True: 3.07k, False: 2] ------------------ 225| 3.07k| const int blocksperrow = mRaw->dim.x / dsc.PixelsPerBlock; 226| 3.07k| const int bytesPerRow = dsc.BytesPerBlock * blocksperrow; 227| | 228| 3.07k| ByteStream rowInput = input.getSubStream(bytesPerRow * row, bytesPerRow); 229| 59.8k| for (int rblock = 0, col = 0; rblock < blocksperrow; ------------------ | Branch (229:33): [True: 56.7k, False: 3.07k] ------------------ 230| 56.7k| rblock++, col += dsc.PixelsPerBlock) 231| 56.7k| decompressBlock(rowInput, row, col); 232| 3.07k|} _ZNK8rawspeed23PanasonicV6Decompressor15decompressBlockIL_ZNS0_14TwelveBitBlockEEEEvRNS_10ByteStreamEii: 179| 56.7k| int col) const noexcept { 180| 56.7k| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 181| | 182| 56.7k| pana_cs6_page_decoder page( 183| 56.7k| rowInput.getStream(dsc.BytesPerBlock)); 184| | 185| 56.7k| std::array oddeven = {0, 0}; 186| 56.7k| std::array nonzero = {0, 0}; 187| 56.7k| unsigned pmul = 0; 188| 56.7k| unsigned pixel_base = 0; 189| 838k| for (int pix = 0; pix < dsc.PixelsPerBlock; pix++, col++) { ------------------ | Branch (189:21): [True: 782k, False: 56.7k] ------------------ 190| 782k| if (pix % 3 == 2) { ------------------ | Branch (190:9): [True: 226k, False: 556k] ------------------ 191| 226k| uint16_t base = page.nextpixel(); 192| 226k| if (base == 3) ------------------ | Branch (192:11): [True: 35.8k, False: 190k] ------------------ 193| 35.8k| base = 4; 194| 226k| pixel_base = dsc.PixelbaseZero << base; 195| 226k| pmul = 1 << base; 196| 226k| } 197| 782k| uint16_t epixel = page.nextpixel(); 198| 782k| if (oddeven[pix % 2]) { ------------------ | Branch (198:9): [True: 488k, False: 293k] ------------------ 199| 488k| epixel *= pmul; 200| 488k| if (pixel_base < dsc.PixelbaseCompare && nonzero[pix % 2] > pixel_base) ------------------ | Branch (200:11): [True: 382k, False: 105k] | Branch (200:48): [True: 255k, False: 127k] ------------------ 201| 255k| epixel += nonzero[pix % 2] - pixel_base; 202| 488k| nonzero[pix % 2] = epixel; 203| 488k| } else { 204| 293k| oddeven[pix % 2] = epixel; 205| 293k| if (epixel) ------------------ | Branch (205:11): [True: 86.8k, False: 207k] ------------------ 206| 86.8k| nonzero[pix % 2] = epixel; 207| 207k| else 208| 207k| epixel = implicit_cast(nonzero[pix % 2]); 209| 293k| } 210| 782k| auto spix = static_cast(static_cast(epixel) - 0xf); 211| 782k| if (spix <= dsc.SpixCompare) ------------------ | Branch (211:9): [True: 502k, False: 279k] ------------------ 212| 502k| out(row, col) = spix & dsc.SpixCompare; 213| 279k| else { 214| | // FIXME: this is a convoluted way to compute zero. 215| | // What was this code trying to do, actually? 216| 279k| epixel = static_cast(epixel + 0x7ffffff1) >> 0x1f; 217| 279k| out(row, col) = epixel & dsc.PixelMask; 218| 279k| } 219| 782k| } 220| 56.7k|} PanasonicV6Decompressor.cpp:_ZN8rawspeed12_GLOBAL__N_121pana_cs6_page_decoderILi12EEC2ENS_10ByteStreamE: 79| 56.7k| explicit pana_cs6_page_decoder(ByteStream bs) { fillBuffer(bs); } PanasonicV6Decompressor.cpp:_ZN8rawspeed12_GLOBAL__N_121pana_cs6_page_decoderILi12EE10fillBufferENS_10ByteStreamE: 90| 56.7k|pana_cs6_page_decoder<12>::fillBuffer(ByteStream bs_) noexcept { 91| 56.7k| BitStreamerLSB bs(bs_.peekRemainingBuffer().getAsArray1DRef()); 92| 56.7k| bs.fill(32); 93| 56.7k| pixelbuffer[17] = implicit_cast(bs.getBits(8)); 94| 56.7k| pixelbuffer[16] = implicit_cast(bs.getBits(8)); 95| 56.7k| pixelbuffer[15] = implicit_cast(bs.getBits(8)); 96| 56.7k| pixelbuffer[14] = implicit_cast(bs.getBits(2)); 97| 56.7k| pixelbuffer[13] = implicit_cast(bs.getBits(8)); 98| 56.7k| pixelbuffer[12] = implicit_cast(bs.getBits(8)); 99| 56.7k| pixelbuffer[11] = implicit_cast(bs.getBits(8)); 100| 56.7k| pixelbuffer[10] = implicit_cast(bs.getBits(2)); 101| 56.7k| pixelbuffer[9] = implicit_cast(bs.getBits(8)); 102| 56.7k| pixelbuffer[8] = implicit_cast(bs.getBits(8)); 103| 56.7k| pixelbuffer[7] = implicit_cast(bs.getBits(8)); 104| 56.7k| pixelbuffer[6] = implicit_cast(bs.getBits(2)); 105| 56.7k| pixelbuffer[5] = implicit_cast(bs.getBits(8)); 106| 56.7k| pixelbuffer[4] = implicit_cast(bs.getBits(8)); 107| 56.7k| pixelbuffer[3] = implicit_cast(bs.getBits(8)); 108| 56.7k| pixelbuffer[2] = implicit_cast(bs.getBits(2)); 109| 56.7k| pixelbuffer[1] = implicit_cast(bs.getBits(12)); 110| 56.7k| pixelbuffer[0] = implicit_cast(bs.getBits(12)); 111| 56.7k|} PanasonicV6Decompressor.cpp:_ZN8rawspeed12_GLOBAL__N_121pana_cs6_page_decoderILi12EE9nextpixelEv: 81| 1.00M| uint16_t nextpixel() { 82| 1.00M| uint16_t currPixel = pixelbuffer[current]; 83| 1.00M| ++current; 84| 1.00M| return currPixel; 85| 1.00M| } _ZNK8rawspeed23PanasonicV6Decompressor18decompressInternalIL_ZNS0_16FourteenBitBlockEEEEvv: 235| 45|void PanasonicV6Decompressor::decompressInternal() const noexcept { 236| 45|#ifdef HAVE_OPENMP 237| 45|#pragma omp parallel for num_threads(rawspeed_get_number_of_processor_cores()) \ 238| 45| schedule(static) default(none) 239| 45|#endif 240| 45| for (int row = 0; row < mRaw->dim.y; ++row) { 241| 45| try { 242| 45| decompressRow(row); 243| 45| } catch (...) { 244| | // We should not get any exceptions here. 245| 45| __builtin_unreachable(); 246| 45| } 247| 45| } 248| 45|} PanasonicV6Decompressor.cpp:_ZNK8rawspeed23PanasonicV6Decompressor18decompressInternalIL_ZNS0_16FourteenBitBlockEEEEvv.omp_outlined_debug__: 240| 1.42k| for (int row = 0; row < mRaw->dim.y; ++row) { ------------------ | Branch (240:21): [True: 1.42k, False: 0] ------------------ 241| 1.42k| try { 242| 1.42k| decompressRow(row); 243| 1.42k| } catch (...) { 244| | // We should not get any exceptions here. 245| 0| __builtin_unreachable(); 246| 0| } 247| 1.42k| } _ZNK8rawspeed23PanasonicV6Decompressor13decompressRowIL_ZNS0_16FourteenBitBlockEEEEvi: 223| 89.1k|void PanasonicV6Decompressor::decompressRow(int row) const noexcept { 224| 89.1k| invariant(mRaw->dim.x % dsc.PixelsPerBlock == 0); ------------------ | | 27| 89.1k|#define invariant(expr) assert(expr) ------------------ | Branch (224:3): [True: 89.1k, False: 20] ------------------ 225| 89.1k| const int blocksperrow = mRaw->dim.x / dsc.PixelsPerBlock; 226| 89.1k| const int bytesPerRow = dsc.BytesPerBlock * blocksperrow; 227| | 228| 89.1k| ByteStream rowInput = input.getSubStream(bytesPerRow * row, bytesPerRow); 229| 178k| for (int rblock = 0, col = 0; rblock < blocksperrow; ------------------ | Branch (229:33): [True: 89.8k, False: 89.1k] ------------------ 230| 89.8k| rblock++, col += dsc.PixelsPerBlock) 231| 89.8k| decompressBlock(rowInput, row, col); 232| 89.1k|} _ZNK8rawspeed23PanasonicV6Decompressor15decompressBlockIL_ZNS0_16FourteenBitBlockEEEEvRNS_10ByteStreamEii: 179| 89.8k| int col) const noexcept { 180| 89.8k| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 181| | 182| 89.8k| pana_cs6_page_decoder page( 183| 89.8k| rowInput.getStream(dsc.BytesPerBlock)); 184| | 185| 89.8k| std::array oddeven = {0, 0}; 186| 89.8k| std::array nonzero = {0, 0}; 187| 89.8k| unsigned pmul = 0; 188| 89.8k| unsigned pixel_base = 0; 189| 1.02M| for (int pix = 0; pix < dsc.PixelsPerBlock; pix++, col++) { ------------------ | Branch (189:21): [True: 933k, False: 89.8k] ------------------ 190| 933k| if (pix % 3 == 2) { ------------------ | Branch (190:9): [True: 265k, False: 668k] ------------------ 191| 265k| uint16_t base = page.nextpixel(); 192| 265k| if (base == 3) ------------------ | Branch (192:11): [True: 51.9k, False: 213k] ------------------ 193| 51.9k| base = 4; 194| 265k| pixel_base = dsc.PixelbaseZero << base; 195| 265k| pmul = 1 << base; 196| 265k| } 197| 933k| uint16_t epixel = page.nextpixel(); 198| 933k| if (oddeven[pix % 2]) { ------------------ | Branch (198:9): [True: 674k, False: 258k] ------------------ 199| 674k| epixel *= pmul; 200| 674k| if (pixel_base < dsc.PixelbaseCompare && nonzero[pix % 2] > pixel_base) ------------------ | Branch (200:11): [True: 531k, False: 142k] | Branch (200:48): [True: 462k, False: 69.5k] ------------------ 201| 462k| epixel += nonzero[pix % 2] - pixel_base; 202| 674k| nonzero[pix % 2] = epixel; 203| 674k| } else { 204| 258k| oddeven[pix % 2] = epixel; 205| 258k| if (epixel) ------------------ | Branch (205:11): [True: 156k, False: 102k] ------------------ 206| 156k| nonzero[pix % 2] = epixel; 207| 102k| else 208| 102k| epixel = implicit_cast(nonzero[pix % 2]); 209| 258k| } 210| 933k| auto spix = static_cast(static_cast(epixel) - 0xf); 211| 933k| if (spix <= dsc.SpixCompare) ------------------ | Branch (211:9): [True: 800k, False: 133k] ------------------ 212| 800k| out(row, col) = spix & dsc.SpixCompare; 213| 133k| else { 214| | // FIXME: this is a convoluted way to compute zero. 215| | // What was this code trying to do, actually? 216| 133k| epixel = static_cast(epixel + 0x7ffffff1) >> 0x1f; 217| 133k| out(row, col) = epixel & dsc.PixelMask; 218| 133k| } 219| 933k| } 220| 89.8k|} PanasonicV6Decompressor.cpp:_ZN8rawspeed12_GLOBAL__N_121pana_cs6_page_decoderILi14EEC2ENS_10ByteStreamE: 79| 89.9k| explicit pana_cs6_page_decoder(ByteStream bs) { fillBuffer(bs); } PanasonicV6Decompressor.cpp:_ZN8rawspeed12_GLOBAL__N_121pana_cs6_page_decoderILi14EE10fillBufferENS_10ByteStreamE: 115| 89.9k|pana_cs6_page_decoder<14>::fillBuffer(ByteStream bs_) noexcept { 116| 89.9k| BitStreamerLSB bs(bs_.peekRemainingBuffer().getAsArray1DRef()); 117| 89.9k| bs.fill(32); 118| 89.9k| bs.skipBitsNoFill(4); 119| 89.9k| pixelbuffer[13] = implicit_cast(bs.getBits(10)); 120| 89.9k| pixelbuffer[12] = implicit_cast(bs.getBits(10)); 121| 89.9k| pixelbuffer[11] = implicit_cast(bs.getBits(10)); 122| 89.9k| pixelbuffer[10] = implicit_cast(bs.getBits(2)); 123| 89.9k| pixelbuffer[9] = implicit_cast(bs.getBits(10)); 124| 89.9k| pixelbuffer[8] = implicit_cast(bs.getBits(10)); 125| 89.9k| pixelbuffer[7] = implicit_cast(bs.getBits(10)); 126| 89.9k| pixelbuffer[6] = implicit_cast(bs.getBits(2)); 127| 89.9k| pixelbuffer[5] = implicit_cast(bs.getBits(10)); 128| 89.9k| pixelbuffer[4] = implicit_cast(bs.getBits(10)); 129| 89.9k| pixelbuffer[3] = implicit_cast(bs.getBits(10)); 130| 89.9k| pixelbuffer[2] = implicit_cast(bs.getBits(2)); 131| 89.9k| pixelbuffer[1] = implicit_cast(bs.getBits(14)); 132| 89.9k| pixelbuffer[0] = implicit_cast(bs.getBits(14)); 133| 89.9k|} PanasonicV6Decompressor.cpp:_ZN8rawspeed12_GLOBAL__N_121pana_cs6_page_decoderILi14EE9nextpixelEv: 81| 1.17M| uint16_t nextpixel() { 82| 1.17M| uint16_t currPixel = pixelbuffer[current]; 83| 1.17M| ++current; 84| 1.17M| return currPixel; 85| 1.17M| } _ZN8rawspeed23PanasonicV7DecompressorC2ENS_8RawImageENS_10ByteStreamE: 43| 64| : mRaw(std::move(img)) { 44| 64| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (44:7): [True: 0, False: 64] | Branch (44:30): [True: 0, False: 64] ------------------ 45| 64| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (45:7): [True: 0, False: 64] ------------------ 46| 64| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 47| | 48| 64| if (!mRaw->dim.hasPositiveArea() || mRaw->dim.x % PixelsPerBlock != 0) { ------------------ | Branch (48:7): [True: 2, False: 62] | Branch (48:39): [True: 1, False: 61] ------------------ 49| 3| ThrowRDE("Unexpected image dimensions found: (%i; %i)", mRaw->dim.x, ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 50| 3| mRaw->dim.y); 51| 3| } 52| | 53| | // How many blocks are needed for the given image size? 54| 61| const auto numBlocks = mRaw->dim.area() / PixelsPerBlock; 55| | 56| | // Does the input contain enough blocks? 57| | // How many full blocks does the input contain? This is truncating division. 58| 61| if (const auto haveBlocks = input_.getRemainSize() / BytesPerBlock; 59| 61| haveBlocks < numBlocks) ------------------ | Branch (59:7): [True: 27, False: 34] ------------------ 60| 61| ThrowRDE("Insufficient count of input blocks for a given image"); ------------------ | | 38| 27| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 27| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 27| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 27| __VA_ARGS__) | | ------------------ ------------------ 61| | 62| | // We only want those blocks we need, no extras. 63| 34| input = input_.peekStream(implicit_cast(numBlocks), 64| 34| BytesPerBlock); 65| 34|} _ZNK8rawspeed23PanasonicV7Decompressor13decompressRowEi: 76| 229k|void PanasonicV7Decompressor::decompressRow(int row) const noexcept { 77| 229k| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 78| 229k| Array1DRef outRow = out[row]; 79| | 80| 229k| invariant(outRow.size() % PixelsPerBlock == 0); ------------------ | | 27| 229k|#define invariant(expr) assert(expr) ------------------ | Branch (80:3): [True: 231k, False: 18.4E] ------------------ 81| 231k| const int blocksperrow = outRow.size() / PixelsPerBlock; 82| 231k| const int bytesPerRow = BytesPerBlock * blocksperrow; 83| | 84| 231k| ByteStream rowInput = input.getSubStream(bytesPerRow * row, bytesPerRow); 85| 465k| for (int rblock = 0; rblock < blocksperrow; rblock++) { ------------------ | Branch (85:24): [True: 233k, False: 231k] ------------------ 86| 233k| ByteStream block = rowInput.getStream(BytesPerBlock); 87| 233k| decompressBlock(block, outRow.getBlock(PixelsPerBlock, rblock)); 88| 233k| } 89| 231k|} _ZN8rawspeed23PanasonicV7Decompressor15decompressBlockENS_10ByteStreamENS_17CroppedArray1DRefItEE: 69| 230k| ByteStream block, CroppedArray1DRef out) noexcept { 70| 230k| invariant(out.size() == PixelsPerBlock); ------------------ | | 27| 230k|#define invariant(expr) assert(expr) ------------------ | Branch (70:3): [True: 230k, False: 672] ------------------ 71| 230k| BitStreamerLSB pump(block.peekRemainingBuffer().getAsArray1DRef()); 72| 2.03M| for (int pix = 0; pix < PixelsPerBlock; pix++) ------------------ | Branch (72:21): [True: 1.80M, False: 230k] ------------------ 73| 1.80M| out(pix) = implicit_cast(pump.getBits(BitsPerSample)); 74| 230k|} _ZNK8rawspeed23PanasonicV7Decompressor10decompressEv: 91| 34|void PanasonicV7Decompressor::decompress() const { 92| 34|#ifdef HAVE_OPENMP 93| 34|#pragma omp parallel for num_threads(rawspeed_get_number_of_processor_cores()) \ 94| 34| schedule(static) default(none) 95| 34|#endif 96| 34| for (int row = 0; row < mRaw->dim.y; ++row) { 97| 34| try { 98| 34| decompressRow(row); 99| 34| } catch (...) { 100| | // We should not get any exceptions here. 101| 34| __builtin_unreachable(); 102| 34| } 103| 34| } 104| 34|} PanasonicV7Decompressor.cpp:_ZNK8rawspeed23PanasonicV7Decompressor10decompressEv.omp_outlined_debug__: 96| 1.07k| for (int row = 0; row < mRaw->dim.y; ++row) { ------------------ | Branch (96:21): [True: 1.07k, False: 0] ------------------ 97| 1.07k| try { 98| 1.07k| decompressRow(row); 99| 1.07k| } catch (...) { 100| | // We should not get any exceptions here. 101| 0| __builtin_unreachable(); 102| 0| } 103| 1.07k| } _ZN8rawspeed18PentaxDecompressorC2ENS_8RawImageENS_8OptionalINS_10ByteStreamEEE: 55| 174| : mRaw(std::move(img)), ht(SetupPrefixCodeDecoder(metaData)) { 56| 174| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (56:7): [True: 12, False: 162] | Branch (56:30): [True: 0, False: 162] ------------------ 57| 162| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (57:7): [True: 0, False: 162] ------------------ 58| 174| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 59| | 60| 174| if (!mRaw->dim.x || !mRaw->dim.y || mRaw->dim.x % 2 != 0 || ------------------ | Branch (60:7): [True: 13, False: 161] | Branch (60:23): [True: 1, False: 160] | Branch (60:39): [True: 1, False: 159] ------------------ 61| 159| mRaw->dim.x > 8384 || mRaw->dim.y > 6208) { ------------------ | Branch (61:7): [True: 8, False: 151] | Branch (61:29): [True: 11, False: 140] ------------------ 62| 22| ThrowRDE("Unexpected image dimensions found: (%d; %d)", mRaw->dim.x, ------------------ | | 38| 22| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 22| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 22| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 22| __VA_ARGS__) | | ------------------ ------------------ 63| 22| mRaw->dim.y); 64| 22| } 65| 174|} _ZN8rawspeed18PentaxDecompressor29SetupPrefixCodeDecoder_LegacyEv: 68| 124|PentaxDecompressor::SetupPrefixCodeDecoder_Legacy() { 69| | // Temporary table, used during parsing LJpeg. 70| 124| HuffmanCode hc; 71| | 72| | /* Initialize with legacy data */ 73| 124| auto nCodes = hc.setNCodesPerLength(Buffer(pentax_tree[0][0].data(), 16)); 74| 124| invariant(nCodes == 13); // see pentax_tree definition ------------------ | | 27| 124|#define invariant(expr) assert(expr) ------------------ | Branch (74:3): [True: 124, False: 0] ------------------ 75| 124| hc.setCodeValues(Array1DRef(pentax_tree[0][1].data(), nCodes)); 76| | 77| 124| return hc; 78| 124|} _ZN8rawspeed18PentaxDecompressor29SetupPrefixCodeDecoder_ModernENS_10ByteStreamE: 81| 50|PentaxDecompressor::SetupPrefixCodeDecoder_Modern(ByteStream stream) { 82| | // Temporary table, used during parsing LJpeg. 83| 50| HuffmanCode hc; 84| | 85| 50| const uint32_t depth = stream.getU16() + 12; 86| 50| if (depth > 15) ------------------ | Branch (86:7): [True: 1, False: 49] ------------------ 87| 50| ThrowRDE("Depth of huffman table is too great (%u).", depth); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 88| | 89| 49| stream.skipBytes(12); 90| | 91| 49| std::array v0; 92| 49| std::array v1; 93| 650| for (uint32_t i = 0; i < depth; i++) ------------------ | Branch (93:24): [True: 601, False: 49] ------------------ 94| 601| v0[i] = stream.getU16(); 95| 566| for (uint32_t i = 0; i < depth; i++) { ------------------ | Branch (95:24): [True: 521, False: 45] ------------------ 96| 521| v1[i] = stream.getByte(); 97| | 98| 521| if (v1[i] == 0 || v1[i] > 12) ------------------ | Branch (98:9): [True: 3, False: 518] | Branch (98:23): [True: 3, False: 515] ------------------ 99| 521| ThrowRDE("Data corrupt: v1[%u]=%u, expected [1..12]", depth, v1[i]); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 100| 521| } 101| | 102| 45| std::vector nCodesPerLength; 103| 45| nCodesPerLength.resize(17); 104| | 105| 45| std::array v2; 106| | /* Calculate codes and store bitcounts */ 107| 550| for (uint32_t c = 0; c < depth; c++) { ------------------ | Branch (107:24): [True: 505, False: 45] ------------------ 108| 505| v2[c] = extractHighBits(v0[c], v1[c], /*effectiveBitwidth=*/12); 109| 505| nCodesPerLength.at(v1[c])++; 110| 505| } 111| | 112| 45| assert(nCodesPerLength.size() == 17); ------------------ | Branch (112:3): [True: 39, False: 6] ------------------ 113| 45| assert(nCodesPerLength[0] == 0); ------------------ | Branch (113:3): [True: 39, False: 0] ------------------ 114| 39| auto nCodes = hc.setNCodesPerLength(Buffer(&nCodesPerLength[1], 16)); 115| 39| invariant(nCodes == depth); ------------------ | | 27| 39|#define invariant(expr) assert(expr) ------------------ | Branch (115:3): [True: 38, False: 1] ------------------ 116| | 117| 38| std::vector codeValues; 118| 38| codeValues.reserve(nCodes); 119| | 120| | /* Find smallest */ 121| 531| for (uint32_t i = 0; i < depth; i++) { ------------------ | Branch (121:24): [True: 493, False: 38] ------------------ 122| 493| uint32_t sm_val = 0xfffffff; 123| 493| uint32_t sm_num = 0xff; 124| 6.92k| for (uint32_t j = 0; j < depth; j++) { ------------------ | Branch (124:26): [True: 6.43k, False: 493] ------------------ 125| 6.43k| if (v2[j] <= sm_val) { ------------------ | Branch (125:11): [True: 1.43k, False: 5.00k] ------------------ 126| 1.43k| sm_num = j; 127| 1.43k| sm_val = v2[j]; 128| 1.43k| } 129| 6.43k| } 130| 493| invariant(sm_num < 16); ------------------ | | 27| 493|#define invariant(expr) assert(expr) ------------------ | Branch (130:5): [True: 493, False: 0] ------------------ 131| 493| codeValues.push_back(implicit_cast(sm_num)); 132| 493| v2[sm_num] = 0xffffffff; 133| 493| } 134| | 135| 38| assert(codeValues.size() == nCodes); ------------------ | Branch (135:3): [True: 38, False: 0] ------------------ 136| 38| hc.setCodeValues(Array1DRef(codeValues.data(), nCodes)); 137| | 138| 38| return hc; 139| 38|} _ZN8rawspeed18PentaxDecompressor22SetupPrefixCodeDecoderENS_8OptionalINS_10ByteStreamEEE: 142| 174|PentaxDecompressor::SetupPrefixCodeDecoder(Optional metaData) { 143| 174| Optional> hc; 144| | 145| 174| if (metaData) ------------------ | Branch (145:7): [True: 50, False: 124] ------------------ 146| 50| hc = SetupPrefixCodeDecoder_Modern(*metaData); 147| 124| else 148| 124| hc = SetupPrefixCodeDecoder_Legacy(); 149| | 150| 174| PrefixCodeDecoder<> ht(std::move(*hc)); 151| 174| ht.setup(true, false); 152| | 153| 174| return ht; 154| 174|} _ZNK8rawspeed18PentaxDecompressor10decompressENS_10ByteStreamE: 156| 86|void PentaxDecompressor::decompress(ByteStream data) const { 157| 86| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 158| | 159| 86| invariant(out.height() > 0); ------------------ | | 27| 86|#define invariant(expr) assert(expr) ------------------ | Branch (159:3): [True: 86, False: 0] ------------------ 160| 86| invariant(out.width() > 0); ------------------ | | 27| 86|#define invariant(expr) assert(expr) ------------------ | Branch (160:3): [True: 86, False: 0] ------------------ 161| 86| invariant(out.width() % 2 == 0); ------------------ | | 27| 86|#define invariant(expr) assert(expr) ------------------ | Branch (161:3): [True: 86, False: 0] ------------------ 162| | 163| 86| BitStreamerMSB bs(data.peekRemainingBuffer().getAsArray1DRef()); 164| 1.12k| for (int row = 0; row < out.height(); row++) { ------------------ | Branch (164:21): [True: 1.07k, False: 45] ------------------ 165| 1.07k| std::array pred = {{}}; 166| 1.07k| if (row >= 2) ------------------ | Branch (166:9): [True: 959, False: 117] ------------------ 167| 959| pred = {out(row - 2, 0), out(row - 2, 1)}; 168| | 169| 9.54k| for (int col = 0; col < out.width(); col++) { ------------------ | Branch (169:23): [True: 8.51k, False: 1.03k] ------------------ 170| 8.51k| pred[col & 1] += ht.decodeDifference(bs); 171| 8.51k| int value = pred[col & 1]; 172| 8.51k| if (!isIntN(value, 16)) ------------------ | Branch (172:11): [True: 41, False: 8.47k] ------------------ 173| 8.51k| ThrowRDE("decoded value out of bounds at %d:%d", col, row); ------------------ | | 38| 41| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 41| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 41| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 41| __VA_ARGS__) | | ------------------ ------------------ 174| 8.47k| out(row, col) = implicit_cast(value); 175| 8.47k| } 176| 1.07k| } 177| 86|} _ZN8rawspeed20PhaseOneDecompressorC2ENS_8RawImageEONSt3__16vectorINS_13PhaseOneStripENS2_9allocatorIS4_EEEE: 45| 488| : mRaw(std::move(img)), strips(std::move(strips_)) { 46| 488| if (mRaw->getDataType() != RawImageType::UINT16) ------------------ | Branch (46:7): [True: 0, False: 488] ------------------ 47| 488| ThrowRDE("Unexpected data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 48| | 49| 488| if (mRaw->getCpp() != 1 || mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (49:7): [True: 0, False: 488] | Branch (49:30): [True: 0, False: 488] ------------------ 50| 488| ThrowRDE("Unexpected cpp: %u", mRaw->getCpp()); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 51| | 52| 488| if (!mRaw->dim.hasPositiveArea() || mRaw->dim.x % 2 != 0 || ------------------ | Branch (52:7): [True: 0, False: 488] | Branch (52:39): [True: 2, False: 486] ------------------ 53| 486| mRaw->dim.x > 11976 || mRaw->dim.y > 8854) { ------------------ | Branch (53:7): [True: 0, False: 486] | Branch (53:30): [True: 0, False: 486] ------------------ 54| 2| ThrowRDE("Unexpected image dimensions found: (%d; %d)", mRaw->dim.x, ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 55| 2| mRaw->dim.y); 56| 2| } 57| | 58| 486| prepareStrips(); 59| 486|} _ZN8rawspeed20PhaseOneDecompressor13prepareStripsEv: 61| 486|void PhaseOneDecompressor::prepareStrips() { 62| | // The 'strips' vector should contain exactly one element per row of image. 63| | 64| | // If the length is different, then the 'strips' vector is clearly incorrect. 65| 486| if (strips.size() != static_cast(mRaw->dim.y)) { ------------------ | Branch (65:7): [True: 0, False: 486] ------------------ 66| 0| ThrowRDE("Height (%d) vs strip count %zu mismatch", mRaw->dim.y, ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 67| 0| strips.size()); 68| 0| } 69| | 70| | // Now, the strips in 'strips' vector aren't in order. 71| | // The 'decltype(strips)::value_type::n' is the row number of a strip. 72| | // We need to make sure that we have every row (0..mRaw->dim.y-1), once. 73| | // For that, first let's sort them to have monothonically increasting `n`. 74| | // This will also serialize the per-line outputting. 75| 486| std::sort( 76| 486| strips.begin(), strips.end(), 77| 486| [](const PhaseOneStrip& a, const PhaseOneStrip& b) { return a.n < b.n; }); 78| | // And now ensure that slice number matches the slice's row. 79| 3.78k| for (decltype(strips)::size_type i = 0; i < strips.size(); ++i) ------------------ | Branch (79:43): [True: 3.29k, False: 486] ------------------ 80| 3.29k| if (static_cast(strips[i].n) != i) ------------------ | Branch (80:9): [True: 0, False: 3.29k] ------------------ 81| 486| ThrowRDE("Strips validation issue."); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 486| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 486| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 486| __VA_ARGS__) | | ------------------ ------------------ 82| | // All good. 83| 486|} PhaseOneDecompressor.cpp:_ZZN8rawspeed20PhaseOneDecompressor13prepareStripsEvENK3$_0clERKNS_13PhaseOneStripES4_: 77| 9.79k| [](const PhaseOneStrip& a, const PhaseOneStrip& b) { return a.n < b.n; }); _ZNK8rawspeed20PhaseOneDecompressor15decompressStripERKNS_13PhaseOneStripE: 85| 3.26k|void PhaseOneDecompressor::decompressStrip(const PhaseOneStrip& strip) const { 86| 3.26k| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 87| | 88| 3.26k| invariant(out.width() > 0); ------------------ | | 27| 3.26k|#define invariant(expr) assert(expr) ------------------ | Branch (88:3): [True: 3.19k, False: 62] ------------------ 89| 3.19k| invariant(out.width() % 2 == 0); ------------------ | | 27| 3.19k|#define invariant(expr) assert(expr) ------------------ | Branch (89:3): [True: 3.18k, False: 14] ------------------ 90| | 91| 3.18k| static constexpr std::array length = {8, 7, 6, 9, 11, 92| 3.18k| 10, 5, 12, 14, 13}; 93| | 94| 3.18k| BitStreamerMSB32 pump(strip.bs.peekRemainingBuffer().getAsArray1DRef()); 95| | 96| 3.18k| std::array pred; 97| 3.18k| pred.fill(0); 98| 3.18k| std::array len; 99| 3.18k| const int row = strip.n; 100| 192k| for (int col = 0; col < out.width(); col++) { ------------------ | Branch (100:21): [True: 189k, False: 2.79k] ------------------ 101| 189k| pump.fill(32); 102| 189k| if (static_cast(col) >= ------------------ | Branch (102:9): [True: 8.09k, False: 181k] ------------------ 103| 189k| (out.width() & ~7U)) // last 'width % 8' pixels. 104| 8.09k| len[0] = len[1] = 14; 105| 181k| else if ((col & 7) == 0) { ------------------ | Branch (105:14): [True: 23.1k, False: 158k] ------------------ 106| 45.9k| for (int& i : len) { ------------------ | Branch (106:19): [True: 45.9k, False: 22.7k] ------------------ 107| 45.9k| int j = 0; 108| | 109| 151k| for (; j < 5; j++) { ------------------ | Branch (109:16): [True: 137k, False: 14.8k] ------------------ 110| 137k| if (pump.getBitsNoFill(1) != 0) { ------------------ | Branch (110:15): [True: 31.1k, False: 105k] ------------------ 111| 31.1k| if (col == 0) ------------------ | Branch (111:17): [True: 393, False: 30.7k] ------------------ 112| 31.1k| ThrowRDE("Can not initialize lengths. Data is corrupt."); ------------------ | | 38| 393| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 393| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 393| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 393| __VA_ARGS__) | | ------------------ ------------------ 113| | 114| | // else, we have previously initialized lengths, so we are fine 115| 30.7k| break; 116| 31.1k| } 117| 137k| } 118| | 119| 45.5k| invariant((col == 0 && j > 0) || col != 0); ------------------ | | 27| 45.5k|#define invariant(expr) assert(expr) ------------------ | Branch (119:9): [True: 829, False: 44.7k] | Branch (119:9): [True: 829, False: 0] | Branch (119:9): [True: 44.7k, False: 0] | Branch (119:9): [True: 45.5k, False: 18.4E] ------------------ 120| 45.5k| if (j > 0) ------------------ | Branch (120:13): [True: 31.3k, False: 14.2k] ------------------ 121| 31.3k| i = length[(2 * (j - 1)) + pump.getBitsNoFill(1)]; 122| 45.5k| } 123| 23.1k| } 124| | 125| 189k| int i = len[col & 1]; 126| 189k| if (i == 14) { ------------------ | Branch (126:9): [True: 77.9k, False: 111k] ------------------ 127| 77.9k| pred[col & 1] = pump.getBitsNoFill(16); 128| 77.9k| out(row, col) = implicit_cast(pred[col & 1]); 129| 111k| } else { 130| 111k| pred[col & 1] += 131| 111k| static_cast(pump.getBitsNoFill(i)) + 1 - (1 << (i - 1)); 132| | // FIXME: is the truncation the right solution here? 133| 111k| out(row, col) = uint16_t(pred[col & 1]); 134| 111k| } 135| 189k| } 136| 3.18k|} _ZNK8rawspeed20PhaseOneDecompressor16decompressThreadEv: 138| 37.7k|void PhaseOneDecompressor::decompressThread() const noexcept { 139| 37.7k|#ifdef HAVE_OPENMP 140| 37.7k|#pragma omp for schedule(static) 141| 37.7k|#endif 142| 37.7k| for (const auto& strip : 143| 37.7k| Array1DRef(strips.data(), implicit_cast(strips.size()))) { 144| 37.7k| try { 145| 37.7k| decompressStrip(strip); 146| 37.7k| } catch (const RawspeedException& err) { 147| | // Propagate the exception out of OpenMP magic. 148| 37.7k| mRaw->setError(err.what()); 149| 37.7k| } catch (...) { 150| | // We should not get any other exception type here. 151| 37.7k| __builtin_unreachable(); 152| 37.7k| } 153| 37.7k| } 154| 37.7k|} _ZNK8rawspeed20PhaseOneDecompressor10decompressEv: 156| 486|void PhaseOneDecompressor::decompress() const { 157| 486|#ifdef HAVE_OPENMP 158| 486|#pragma omp parallel default(none) \ 159| 486| num_threads(rawspeed_get_number_of_processor_cores()) 160| 486|#endif 161| 486| decompressThread(); 162| | 163| 486| std::string firstErr; 164| 486| if (mRaw->isTooManyErrors(1, &firstErr)) { ------------------ | Branch (164:7): [True: 152, False: 334] ------------------ 165| 152| ThrowRDE("Too many errors encountered. Giving up. First Error:\n%s", ------------------ | | 38| 152| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 152| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 152| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 152| __VA_ARGS__) | | ------------------ ------------------ 166| 152| firstErr.c_str()); 167| 152| } 168| 486|} PhaseOneDecompressor.cpp:_ZNK8rawspeed20PhaseOneDecompressor10decompressEv.omp_outlined_debug__: 161| 15.2k| decompressThread(); _ZN8rawspeed13PhaseOneStripC2EiNS_10ByteStreamE: 37| 3.67k| PhaseOneStrip(int block, ByteStream bs_) : n(block), bs(bs_) {} _ZN8rawspeed21SamsungV0DecompressorC2ERKNS_8RawImageENS_10ByteStreamES4_: 46| 146| : AbstractSamsungDecompressor(image) { 47| 146| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (47:7): [True: 0, False: 146] | Branch (47:30): [True: 0, False: 146] ------------------ 48| 146| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (48:7): [True: 0, False: 146] ------------------ 49| 146| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 50| | 51| 146| const uint32_t width = mRaw->dim.x; 52| 146| const uint32_t height = mRaw->dim.y; 53| | 54| 146| if (width == 0 || height == 0 || width < 16 || width > 5546 || height > 3714) ------------------ | Branch (54:7): [True: 1, False: 145] | Branch (54:21): [True: 1, False: 144] | Branch (54:36): [True: 1, False: 143] | Branch (54:50): [True: 2, False: 141] | Branch (54:66): [True: 1, False: 140] ------------------ 55| 146| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 6| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 6| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 6| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 6| __VA_ARGS__) | | ------------------ ------------------ 56| | 57| 140| computeStripes(bso.peekStream(height, 4), bsr); 58| 140|} _ZN8rawspeed21SamsungV0Decompressor14computeStripesENS_10ByteStreamES1_: 61| 140|void SamsungV0Decompressor::computeStripes(ByteStream bso, ByteStream bsr) { 62| 140| const uint32_t height = mRaw->dim.y; 63| | 64| 140| std::vector offsets; 65| 140| offsets.reserve(1 + height); 66| 5.22k| for (uint32_t y = 0; y < height; y++) ------------------ | Branch (66:24): [True: 5.08k, False: 140] ------------------ 67| 5.08k| offsets.emplace_back(bso.getU32()); 68| 140| offsets.emplace_back(bsr.getSize()); 69| | 70| 140| stripes.reserve(height); 71| | 72| 140| auto offset_iterator = std::begin(offsets); 73| 140| bsr.skipBytes(*offset_iterator); 74| | 75| 140| auto next_offset_iterator = std::next(offset_iterator); 76| 389| while (next_offset_iterator < std::end(offsets)) { ------------------ | Branch (76:10): [True: 250, False: 139] ------------------ 77| 250| if (*offset_iterator >= *next_offset_iterator) ------------------ | Branch (77:9): [True: 1, False: 249] ------------------ 78| 250| ThrowRDE("Line offsets are out of sequence or slice is empty."); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 79| | 80| 249| const auto size = *next_offset_iterator - *offset_iterator; 81| 249| invariant(size > 0); ------------------ | | 27| 249|#define invariant(expr) assert(expr) ------------------ | Branch (81:5): [True: 249, False: 0] ------------------ 82| | 83| 249| stripes.emplace_back(bsr.getStream(size)); 84| | 85| 249| std::advance(offset_iterator, 1); 86| 249| std::advance(next_offset_iterator, 1); 87| 249| } 88| | 89| 140| assert(stripes.size() == height); ------------------ | Branch (89:3): [True: 129, False: 10] ------------------ 90| 139|} _ZNK8rawspeed21SamsungV0Decompressor10decompressEv: 92| 129|void SamsungV0Decompressor::decompress() const { 93| 349| for (int row = 0; row < mRaw->dim.y; row++) ------------------ | Branch (93:21): [True: 220, False: 129] ------------------ 94| 220| decompressStrip(row, stripes[row]); 95| | 96| | // Swap red and blue pixels to get the final CFA pattern 97| 129| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 98| 155| for (int row = 0; row < out.height() - 1; row += 2) { ------------------ | Branch (98:21): [True: 26, False: 129] ------------------ 99| 1.94k| for (int col = 0; col < out.width() - 1; col += 2) ------------------ | Branch (99:23): [True: 1.91k, False: 26] ------------------ 100| 1.91k| std::swap(out(row, col + 1), out(row + 1, col)); 101| 26| } 102| 129|} _ZN8rawspeed21SamsungV0Decompressor7calcAdjERNS_16BitStreamerMSB32Ei: 104| 54.7k|int32_t SamsungV0Decompressor::calcAdj(BitStreamerMSB32& bits, int nbits) { 105| 54.7k| if (!nbits) ------------------ | Branch (105:7): [True: 14.4k, False: 40.3k] ------------------ 106| 14.4k| return 0; 107| 40.3k| return signExtend(bits.getBits(nbits), nbits); 108| 54.7k|} _ZNK8rawspeed21SamsungV0Decompressor15decompressStripEiNS_10ByteStreamE: 110| 220|void SamsungV0Decompressor::decompressStrip(int row, ByteStream bs) const { 111| 220| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 112| 220| invariant(out.width() > 0); ------------------ | | 27| 220|#define invariant(expr) assert(expr) ------------------ | Branch (112:3): [True: 220, False: 0] ------------------ 113| | 114| 220| BitStreamerMSB32 bits(bs.peekRemainingBuffer().getAsArray1DRef()); 115| | 116| 220| std::array len; 117| 220| for (int& i : len) ------------------ | Branch (117:15): [True: 876, False: 220] ------------------ 118| 876| i = row < 2 ? 7 : 4; ------------------ | Branch (118:9): [True: 652, False: 224] ------------------ 119| | 120| | // Image is arranged in groups of 16 pixels horizontally 121| 3.65k| for (int col = 0; col < out.width(); col += 16) { ------------------ | Branch (121:21): [True: 3.50k, False: 145] ------------------ 122| 3.50k| bits.fill(); 123| 3.50k| bool dir = !!bits.getBitsNoFill(1); 124| | 125| 3.50k| std::array op; 126| 3.50k| for (int& i : op) ------------------ | Branch (126:17): [True: 14.0k, False: 3.50k] ------------------ 127| 14.0k| i = bits.getBitsNoFill(2); 128| | 129| 17.5k| for (int i = 0; i < 4; i++) { ------------------ | Branch (129:21): [True: 14.0k, False: 3.49k] ------------------ 130| 14.0k| assert(op[i] >= 0 && op[i] <= 3); ------------------ | Branch (130:7): [True: 14.0k, False: 0] | Branch (130:7): [True: 14.0k, False: 0] | Branch (130:7): [True: 14.0k, False: 0] ------------------ 131| | 132| 14.0k| switch (op[i]) { 133| 1.20k| case 3: ------------------ | Branch (133:7): [True: 1.20k, False: 12.7k] ------------------ 134| 1.20k| len[i] = bits.getBits(4); 135| 1.20k| break; 136| 934| case 2: ------------------ | Branch (136:7): [True: 934, False: 13.0k] ------------------ 137| 934| len[i]--; 138| 934| break; 139| 1.14k| case 1: ------------------ | Branch (139:7): [True: 1.14k, False: 12.8k] ------------------ 140| 1.14k| len[i]++; 141| 1.14k| break; 142| 10.7k| default: ------------------ | Branch (142:7): [True: 10.7k, False: 3.29k] ------------------ 143| | // FIXME: it can be zero too. 144| 10.7k| break; 145| 14.0k| } 146| | 147| 14.0k| if (len[i] < 0) ------------------ | Branch (147:11): [True: 8, False: 13.9k] ------------------ 148| 14.0k| ThrowRDE("Bit length less than 0."); ------------------ | | 38| 8| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 8| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 8| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 8| __VA_ARGS__) | | ------------------ ------------------ 149| 13.9k| if (len[i] > 16) ------------------ | Branch (149:11): [True: 1, False: 13.9k] ------------------ 150| 13.9k| ThrowRDE("Bit Length more than 16."); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 151| 13.9k| } 152| | 153| 3.49k| if (dir) { ------------------ | Branch (153:9): [True: 152, False: 3.34k] ------------------ 154| | // Upward prediction 155| | 156| 152| if (row < 2) ------------------ | Branch (156:11): [True: 62, False: 90] ------------------ 157| 152| ThrowRDE("Upward prediction for the first two rows. Raw corrupt"); ------------------ | | 38| 62| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 62| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 62| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 62| __VA_ARGS__) | | ------------------ ------------------ 158| | 159| 90| if (col + 16 >= out.width()) ------------------ | Branch (159:11): [True: 4, False: 86] ------------------ 160| 90| ThrowRDE("Upward prediction for the last block of pixels. Raw corrupt"); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 161| | 162| | // First we decode even pixels 163| 770| for (int c = 0; c < 16; c += 2) { ------------------ | Branch (163:23): [True: 684, False: 86] ------------------ 164| 684| int b = len[c >> 3]; 165| 684| int32_t adj = calcAdj(bits, b); 166| | 167| 684| out(row, col + c) = 168| 684| implicit_cast(adj + out(row - 1, col + c)); 169| 684| } 170| | 171| | // Now we decode odd pixels 172| | // Why on earth upward prediction only looks up 1 line above 173| | // is beyond me, it will hurt compression a deal. 174| 751| for (int c = 1; c < 16; c += 2) { ------------------ | Branch (174:23): [True: 665, False: 86] ------------------ 175| 665| int b = len[2 | (c >> 3)]; 176| 665| int32_t adj = calcAdj(bits, b); 177| | 178| 665| out(row, col + c) = 179| 665| implicit_cast(adj + out(row - 2, col + c)); 180| 665| } 181| 3.34k| } else { 182| | // Left to right prediction 183| | // First we decode even pixels 184| 3.34k| int pred_left = col != 0 ? out(row, col - 2) : 128; ------------------ | Branch (184:23): [True: 3.13k, False: 211] ------------------ 185| 30.0k| for (int c = 0; c < 16; c += 2) { ------------------ | Branch (185:23): [True: 26.7k, False: 3.34k] ------------------ 186| 26.7k| int b = len[c >> 3]; 187| 26.7k| int32_t adj = calcAdj(bits, b); 188| | 189| 26.7k| if (col + c < out.width()) ------------------ | Branch (189:13): [True: 26.4k, False: 298] ------------------ 190| 26.4k| out(row, col + c) = implicit_cast(adj + pred_left); 191| 26.7k| } 192| | 193| | // Now we decode odd pixels 194| 3.34k| pred_left = col != 0 ? out(row, col - 1) : 128; ------------------ | Branch (194:19): [True: 3.12k, False: 218] ------------------ 195| 30.0k| for (int c = 1; c < 16; c += 2) { ------------------ | Branch (195:23): [True: 26.6k, False: 3.34k] ------------------ 196| 26.6k| int b = len[2 | (c >> 3)]; 197| 26.6k| int32_t adj = calcAdj(bits, b); 198| | 199| 26.6k| if (col + c < out.width()) ------------------ | Branch (199:13): [True: 26.2k, False: 369] ------------------ 200| 26.2k| out(row, col + c) = implicit_cast(adj + pred_left); 201| 26.6k| } 202| 3.34k| } 203| 3.49k| } 204| 220|} _ZN8rawspeed21SamsungV1DecompressorC2ERKNS_8RawImageENS_10ByteStreamEi: 47| 98| : AbstractSamsungDecompressor(image), bs(bs_) { 48| 98| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (48:7): [True: 0, False: 98] | Branch (48:30): [True: 0, False: 98] ------------------ 49| 98| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (49:7): [True: 0, False: 98] ------------------ 50| 98| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 51| | 52| 98| if (bit != 12) ------------------ | Branch (52:7): [True: 1, False: 97] ------------------ 53| 98| ThrowRDE("Unexpected bit per pixel (%d)", bit); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 54| | 55| 97| const uint32_t width = mRaw->dim.x; 56| 97| const uint32_t height = mRaw->dim.y; 57| | 58| 97| if (width == 0 || height == 0 || width % 32 != 0 || height % 2 != 0 || ------------------ | Branch (58:7): [True: 1, False: 96] | Branch (58:21): [True: 1, False: 95] | Branch (58:36): [True: 5, False: 90] | Branch (58:55): [True: 0, False: 90] ------------------ 59| 90| width > 5664 || height > 3714) ------------------ | Branch (59:7): [True: 18, False: 72] | Branch (59:23): [True: 23, False: 49] ------------------ 60| 97| ThrowRDE("Unexpected image dimensions found: (%u; %u)", width, height); ------------------ | | 38| 48| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 48| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 48| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 48| __VA_ARGS__) | | ------------------ ------------------ 61| 97|} _ZNK8rawspeed21SamsungV1Decompressor10decompressEv: 81| 49|void SamsungV1Decompressor::decompress() const { 82| | // This format has a variable length encoding of how many bits are needed 83| | // to encode the difference between pixels, we use a table to process it 84| | // that has two values, the first the number of bits that were used to 85| | // encode, the second the number of bits that come after with the difference 86| | // The table has 14 entries because the difference can have between 0 (no 87| | // difference) and 13 bits (differences between 12 bits numbers can need 13) 88| 49| static const std::array, 14> tab = {{{3, 4}, 89| 49| {3, 7}, 90| 49| {2, 6}, 91| 49| {2, 5}, 92| 49| {4, 3}, 93| 49| {6, 0}, 94| 49| {7, 9}, 95| 49| {8, 10}, 96| 49| {9, 11}, 97| 49| {10, 12}, 98| 49| {10, 13}, 99| 49| {5, 1}, 100| 49| {4, 8}, 101| 49| {4, 2}}}; 102| 49| std::vector tbl(1024); 103| | 104| | // We generate a 1024 entry table (to be addressed by reading 10 bits) by 105| | // consecutively filling in 2^(10-N) positions where N is the variable number 106| | // of bits of the encoding. So for example 4 is encoded with 3 bits so the 107| | // first 2^(10-3)=128 positions are set with 3,4 so that any time we read 000 108| | // we know the next 4 bits are the difference. We read 10 bits because that is 109| | // the maximum number of bits used in the variable encoding (for the 12 and 110| | // 13 cases) 111| 49| uint32_t n = 0; 112| 686| for (auto i : tab) { ------------------ | Branch (112:15): [True: 686, False: 49] ------------------ 113| 50.8k| for (int32_t c = 0; c < (1024 >> i[0]); c++) { ------------------ | Branch (113:25): [True: 50.1k, False: 686] ------------------ 114| 50.1k| tbl[n].encLen = i[0]; 115| 50.1k| tbl[n].diffLen = i[1]; 116| 50.1k| n++; 117| 50.1k| } 118| 686| } 119| | 120| 49| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 121| 49| invariant(out.width() % 32 == 0 && ------------------ | | 27| 49|#define invariant(expr) assert(expr) ------------------ | Branch (121:3): [True: 49, False: 0] | Branch (121:3): [True: 49, Folded] | Branch (121:3): [True: 49, False: 0] ------------------ 122| 49| "Should have even count of pixels per row."); 123| 49| invariant(out.height() % 2 == 0 && "Should have even row count."); ------------------ | | 27| 49|#define invariant(expr) assert(expr) ------------------ | Branch (123:3): [True: 49, False: 0] | Branch (123:3): [True: 49, Folded] | Branch (123:3): [True: 49, False: 0] ------------------ 124| 49| BitStreamerMSB pump(bs.peekRemainingBuffer().getAsArray1DRef()); 125| 410| for (int row = 0; row < out.height(); row++) { ------------------ | Branch (125:21): [True: 397, False: 13] ------------------ 126| 397| std::array pred = {{}}; 127| 397| if (row >= 2) ------------------ | Branch (127:9): [True: 333, False: 64] ------------------ 128| 333| pred = {out(row - 2, 0), out(row - 2, 1)}; 129| | 130| 84.0k| for (int col = 0; col < out.width(); col++) { ------------------ | Branch (130:23): [True: 83.7k, False: 361] ------------------ 131| 83.7k| int32_t diff = samsungDiff(pump, tbl); 132| 83.7k| pred[col & 1] += diff; 133| | 134| 83.7k| int value = pred[col & 1]; 135| 83.7k| if (!isIntN(value, bits)) ------------------ | Branch (135:11): [True: 36, False: 83.6k] ------------------ 136| 83.7k| ThrowRDE("decoded value out of bounds"); ------------------ | | 38| 36| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 36| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 36| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 36| __VA_ARGS__) | | ------------------ ------------------ 137| 83.6k| out(row, col) = implicit_cast(value); 138| 83.6k| } 139| 397| } 140| 49|} _ZN8rawspeed21SamsungV1Decompressor11samsungDiffERNS_14BitStreamerMSBERKNSt3__16vectorINS0_12encTableItemENS3_9allocatorIS5_EEEE: 65| 83.7k| const std::vector& tbl) { 66| 83.7k| pump.fill(23); // That is the maximal number of bits we will need here. 67| | // We read 10 bits to index into our table 68| 83.7k| uint32_t c = pump.peekBitsNoFill(10); 69| | // Skip the bits that were used to encode this case 70| 83.7k| pump.skipBitsNoFill(tbl[c].encLen); 71| | // Read the number of bits the table tells me 72| 83.7k| int32_t len = tbl[c].diffLen; 73| 83.7k| if (len == 0) ------------------ | Branch (73:7): [True: 412, False: 83.2k] ------------------ 74| 412| return 0; 75| 83.2k| int32_t diff = pump.getBitsNoFill(len); 76| | // If the first bit is 0 we need to turn this into a negative number 77| 83.2k| diff = PrefixCodeDecoder<>::extend(diff, len); 78| 83.2k| return diff; 79| 83.7k|} _ZN8rawspeed21SamsungV2DecompressorC2ERKNS_8RawImageENS_10ByteStreamEj: 92| 216| : AbstractSamsungDecompressor(image) { 93| 216| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (93:7): [True: 0, False: 216] | Branch (93:30): [True: 0, False: 216] ------------------ 94| 216| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (94:7): [True: 0, False: 216] ------------------ 95| 216| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 96| | 97| 216| switch (bits) { 98| 157| case 12: ------------------ | Branch (98:3): [True: 157, False: 59] ------------------ 99| 216| case 14: ------------------ | Branch (99:3): [True: 59, False: 157] ------------------ 100| 216| break; 101| 0| default: ------------------ | Branch (101:3): [True: 0, False: 216] ------------------ 102| 0| ThrowRDE("Unexpected bit per pixel (%u)", bits); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 103| 216| } 104| | 105| 216| static constexpr const auto headerSize = 16; 106| 216| (void)bs.check(headerSize); 107| | 108| 216| BitStreamerMSB32 startpump(bs.peekRemainingBuffer().getAsArray1DRef()); 109| | 110| | // Process the initial metadata bits, we only really use initVal, width and 111| | // height (the last two match the TIFF values anyway) 112| 216| startpump.getBits(16); // NLCVersion 113| 216| startpump.getBits(4); // ImgFormat 114| 216| bitDepth = startpump.getBits(4) + 1; 115| 216| if (bitDepth != bits) ------------------ | Branch (115:7): [True: 1, False: 215] ------------------ 116| 216| ThrowRDE("Bit depth mismatch with container, %u vs %u", bitDepth, bits); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 117| 215| startpump.getBits(4); // NumBlkInRCUnit 118| 215| startpump.getBits(4); // CompressionRatio 119| 215| width = startpump.getBits(16); 120| 215| height = startpump.getBits(16); 121| 215| startpump.getBits(16); // TileWidth 122| 215| startpump.getBits(4); // reserved 123| | 124| | // The format includes an optimization code that sets 3 flags to change the 125| | // decoding parameters 126| 215| const uint32_t _flags = startpump.getBits(4); 127| 215| if (_flags > static_cast(OptFlags::ALL)) ------------------ | Branch (127:7): [True: 1, False: 214] ------------------ 128| 215| ThrowRDE("Invalid opt flags %x", _flags); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 129| 214| optflags = static_cast(_flags); 130| | 131| 214| startpump.getBits(8); // OverlapWidth 132| 214| startpump.getBits(8); // reserved 133| 214| startpump.getBits(8); // Inc 134| 214| startpump.getBits(2); // reserved 135| 214| initVal = implicit_cast(startpump.getBits(14)); 136| | 137| 214| invariant(startpump.getInputPosition() == headerSize); ------------------ | | 27| 214|#define invariant(expr) assert(expr) ------------------ | Branch (137:3): [True: 213, False: 1] ------------------ 138| | 139| 213| if (width == 0 || height == 0 || width % 16 != 0 || width > 6496 || ------------------ | Branch (139:7): [True: 1, False: 212] | Branch (139:21): [True: 1, False: 211] | Branch (139:36): [True: 4, False: 207] | Branch (139:55): [True: 2, False: 205] ------------------ 140| 205| height > 4336) ------------------ | Branch (140:7): [True: 2, False: 203] ------------------ 141| 213| ThrowRDE("Unexpected image dimensions found: (%i; %i)", width, height); ------------------ | | 38| 10| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 10| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 10| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 10| __VA_ARGS__) | | ------------------ ------------------ 142| | 143| 203| if (width != mRaw->dim.x || height != mRaw->dim.y) ------------------ | Branch (143:7): [True: 6, False: 197] | Branch (143:31): [True: 8, False: 189] ------------------ 144| 203| ThrowRDE("EXIF image dimensions do not match dimensions from raw header"); ------------------ | | 38| 14| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 14| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 14| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 14| __VA_ARGS__) | | ------------------ ------------------ 145| | 146| 189| data = bs.getSubStream(startpump.getInputPosition(), 147| 189| startpump.getRemainingSize()); 148| 189|} _ZN8rawspeed21SamsungV2Decompressor13decompressRowEi: 335| 7.58k|void SamsungV2Decompressor::decompressRow(int row) { 336| | // Align pump to 16byte boundary 337| 7.58k| if (const auto line_offset = data.getPosition(); (line_offset & 0xf) != 0) ------------------ | Branch (337:52): [True: 5.83k, False: 1.75k] ------------------ 338| 5.83k| data.skipBytes(16 - (line_offset & 0xf)); 339| | 340| 7.58k| BitStreamerMSB32 pump(data.peekRemainingBuffer().getAsArray1DRef()); 341| | 342| | // Initialize the motion and diff modes at the start of the line 343| 7.58k| motion = 7; 344| | // By default we are not scaling values at all 345| 7.58k| scale = 0; 346| | 347| 7.58k| for (auto& i : diffBitsMode) ------------------ | Branch (347:16): [True: 22.7k, False: 7.58k] ------------------ 348| 22.7k| i[0] = i[1] = (row == 0 || row == 1) ? 7 : 4; ------------------ | Branch (348:20): [True: 567, False: 22.1k] | Branch (348:32): [True: 357, False: 21.7k] ------------------ 349| | 350| 7.58k| invariant(width >= 16); ------------------ | | 27| 7.58k|#define invariant(expr) assert(expr) ------------------ | Branch (350:3): [True: 7.57k, False: 15] ------------------ 351| 7.57k| invariant(width % 16 == 0); ------------------ | | 27| 7.57k|#define invariant(expr) assert(expr) ------------------ | Branch (351:3): [True: 7.57k, False: 0] ------------------ 352| 34.7k| for (int col = 0; col < width; col += 16) ------------------ | Branch (352:21): [True: 27.1k, False: 7.57k] ------------------ 353| 27.1k| processBlock(pump, row, col); 354| | 355| 7.57k| data.skipBytes(pump.getStreamPosition()); 356| 7.57k|} _ZN8rawspeed21SamsungV2Decompressor12processBlockERNS_16BitStreamerMSB32Eii: 321| 27.1k|SamsungV2Decompressor::processBlock(BitStreamerMSB32& pump, int row, int col) { 322| 27.1k| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 323| | 324| 27.1k| const std::array baseline = 325| 27.1k| prepareBaselineValues(pump, row, col); 326| | 327| | // Figure out how many difference bits we have to read for each pixel 328| 27.1k| const std::array diffs = decodeDifferences(pump, row); 329| | 330| | // Actually apply the differences and write them to the pixels 331| 459k| for (int i = 0; i < 16; ++i, ++col) ------------------ | Branch (331:19): [True: 432k, False: 27.1k] ------------------ 332| 432k| out(row, col) = clampBits(baseline[i] + diffs[i], bitDepth); 333| 27.1k|} _ZN8rawspeed21SamsungV2Decompressor21prepareBaselineValuesERNS_16BitStreamerMSB32Eii: 159| 27.1k| int col) { 160| 27.1k| const Array2DRef img(mRaw->getU16DataAsUncroppedArray2DRef()); 161| | 162| 27.1k| std::array baseline; 163| | 164| 27.1k| if (!(optflags & OptFlags::QP) && (col % 64) == 0) { ------------------ | Branch (164:7): [True: 23.4k, False: 3.76k] | Branch (164:37): [True: 10.2k, False: 13.1k] ------------------ 165| | // Change scale every four 16-pixel blocks. 166| 10.2k| static constexpr std::array scalevals = {{0, -2, 2}}; 167| 10.2k| uint32_t i = pump.getBits(2); 168| 10.2k| scale = i < 3 ? scale + scalevals[i] : pump.getBits(12); ------------------ | Branch (168:13): [True: 5.89k, False: 4.35k] ------------------ 169| 10.2k| } 170| | 171| | // First we figure out which reference pixels mode we're in 172| 27.1k| if (optflags & OptFlags::MV) ------------------ | Branch (172:7): [True: 14.2k, False: 12.8k] ------------------ 173| 14.2k| motion = pump.getBits(1) ? 3 : 7; ------------------ | Branch (173:14): [True: 5.09k, False: 9.19k] ------------------ 174| 12.8k| else if (!pump.getBits(1)) ------------------ | Branch (174:12): [True: 2.02k, False: 10.8k] ------------------ 175| 2.02k| motion = pump.getBits(3); 176| | 177| 27.1k| if ((row == 0 || row == 1) && (motion != 7)) ------------------ | Branch (177:8): [True: 6.17k, False: 21.0k] | Branch (177:20): [True: 3.47k, False: 17.5k] | Branch (177:33): [True: 47, False: 9.60k] ------------------ 178| 27.1k| ThrowRDE("At start of image and motion isn't 7. File corrupted?"); ------------------ | | 38| 47| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 47| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 47| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 47| __VA_ARGS__) | | ------------------ ------------------ 179| | 180| 27.1k| if (motion == 7) { ------------------ | Branch (180:7): [True: 18.0k, False: 9.09k] ------------------ 181| | // The base case. 182| | // If we're at the left edge we just start at the initial value. 183| 18.0k| if (col == 0) { ------------------ | Branch (183:9): [True: 4.36k, False: 13.6k] ------------------ 184| 4.36k| baseline.fill(initVal); 185| 4.36k| return baseline; 186| 4.36k| } 187| | // Else just set all pixels to the previous ones on the same line. 188| 13.6k| std::array prev; 189| 41.0k| for (int i = 0; i < 2; i++) ------------------ | Branch (189:21): [True: 27.3k, False: 13.6k] ------------------ 190| 27.3k| prev[i] = img(row, col + i - 2); 191| 232k| for (int i = 0; i < 16; i++) ------------------ | Branch (191:21): [True: 218k, False: 13.6k] ------------------ 192| 218k| baseline[i] = prev[i & 1]; 193| 13.6k| return baseline; 194| 18.0k| } 195| | 196| | // The complex case, we now need to actually lookup one or two lines above 197| 9.09k| if (row < 2) ------------------ | Branch (197:7): [True: 0, False: 9.09k] ------------------ 198| 9.09k| ThrowRDE("Got a previous line lookup on first two lines. File corrupted?"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 199| | 200| 9.09k| static constexpr std::array motionOffset = {-4, -2, -2, 0, 201| 9.09k| 0, 2, 4}; 202| 9.09k| static constexpr std::array motionDoAverage = {0, 0, 1, 0, 203| 9.09k| 1, 0, 0}; 204| | 205| 9.09k| int32_t slideOffset = motionOffset[motion]; 206| 9.09k| int32_t doAverage = motionDoAverage[motion]; 207| | 208| 154k| for (int i = 0; i < 16; i++) { ------------------ | Branch (208:19): [True: 145k, False: 9.07k] ------------------ 209| 145k| int refRow = row; 210| 145k| int refCol = col + i + slideOffset; 211| | 212| 145k| if ((row + i) & 1) { // Red or blue pixels use same color two lines up ------------------ | Branch (212:9): [True: 72.6k, False: 72.6k] ------------------ 213| 72.6k| refRow -= 2; 214| 72.6k| } else { // Green pixel N uses Green pixel N from row above 215| 72.6k| refRow -= 1; 216| 72.6k| refCol += (i & 1) ? -1 : 1; // (top left or top right) ------------------ | Branch (216:17): [True: 32.4k, False: 40.2k] ------------------ 217| 72.6k| } 218| | 219| 145k| if (refCol < 0) ------------------ | Branch (219:9): [True: 6, False: 145k] ------------------ 220| 145k| ThrowRDE("Bad motion %d at the beginning of the row", motion); ------------------ | | 38| 6| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 6| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 6| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 6| __VA_ARGS__) | | ------------------ ------------------ 221| 145k| if ((refCol >= width) || (doAverage && (refCol + 2 >= width))) ------------------ | Branch (221:9): [True: 5, False: 145k] | Branch (221:31): [True: 3.42k, False: 141k] | Branch (221:44): [True: 3, False: 3.41k] ------------------ 222| 145k| ThrowRDE("Bad motion %d at the end of the row", motion); ------------------ | | 38| 8| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 8| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 8| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 8| __VA_ARGS__) | | ------------------ ------------------ 223| | 224| | // In some cases we use as reference interpolation of this pixel and 225| | // the next 226| 145k| if (doAverage) { ------------------ | Branch (226:9): [True: 3.41k, False: 141k] ------------------ 227| 3.41k| baseline[i] = (img(refRow, refCol) + img(refRow, refCol + 2) + 1) >> 1; 228| 141k| } else { 229| 141k| baseline[i] = img(refRow, refCol); 230| 141k| } 231| 145k| } 232| | 233| 9.07k| return baseline; 234| 9.09k|} SamsungV2Decompressor.cpp:_ZN8rawspeed12_GLOBAL__N_1anENS_21SamsungV2Decompressor8OptFlagsES2_: 71| 81.4k| SamsungV2Decompressor::OptFlags rhs) { 72| 81.4k| return SamsungV2Decompressor::OptFlags::NONE != 73| 81.4k| static_cast( 74| 81.4k| static_cast< 75| 81.4k| std::underlying_type_t>(lhs) & 76| 81.4k| static_cast< 77| 81.4k| std::underlying_type_t>(rhs)); 78| 81.4k|} _ZN8rawspeed21SamsungV2Decompressor17decodeDifferencesERNS_16BitStreamerMSB32Ei: 284| 27.1k|SamsungV2Decompressor::decodeDifferences(BitStreamerMSB32& pump, int row) { 285| | // Figure out how many difference bits we have to read for each pixel 286| 27.1k| const std::array diffBits = decodeDiffLengths(pump, row); 287| | 288| | // Actually read the differences. We know these fit into 15-bit ints. 289| 27.1k| std::array diffs; 290| 459k| for (int i = 0; i < 16; i++) { ------------------ | Branch (290:19): [True: 432k, False: 27.1k] ------------------ 291| 432k| uint32_t len = diffBits[i >> 2]; 292| 432k| int16_t diff = getDiff(pump, len); 293| 432k| diffs[i] = diff; 294| 432k| } 295| | 296| | // Reshuffle the differences, while they still are only 16-bit. 297| 27.1k| std::array shuffled; 298| 459k| for (int i = 0; i < 16; i++) { ------------------ | Branch (298:19): [True: 432k, False: 27.1k] ------------------ 299| 432k| int p; 300| | // The differences are stored interlaced: 301| | // 0 2 4 6 8 10 12 14 1 3 5 7 9 11 13 15 302| 432k| if (row % 2) ------------------ | Branch (302:9): [True: 182k, False: 249k] ------------------ 303| 182k| p = ((i % 8) << 1) - (i >> 3) + 1; 304| 249k| else 305| 249k| p = ((i % 8) << 1) + (i >> 3); 306| | 307| 432k| shuffled[p] = diffs[i]; 308| 432k| } 309| | 310| | // And finally widen and scale the differences. 311| 27.1k| std::array scaled; 312| 459k| for (int i = 0; i < 16; i++) { ------------------ | Branch (312:19): [True: 432k, False: 27.1k] ------------------ 313| 432k| int scaledDiff = (int(shuffled[i]) * (scale * 2 + 1)) + scale; 314| 432k| scaled[i] = scaledDiff; 315| 432k| } 316| | 317| 27.1k| return scaled; 318| 27.1k|} _ZN8rawspeed21SamsungV2Decompressor17decodeDiffLengthsERNS_16BitStreamerMSB32Ei: 237| 27.1k|SamsungV2Decompressor::decodeDiffLengths(BitStreamerMSB32& pump, int row) { 238| 27.1k| if (!(optflags & OptFlags::SKIP) && pump.getBits(1)) ------------------ | Branch (238:7): [True: 21.3k, False: 5.78k] | Branch (238:39): [True: 17.4k, False: 3.86k] ------------------ 239| 17.4k| return {}; 240| | 241| 9.64k| std::array diffBits; 242| | 243| | // Figure out how many difference bits we have to read for each pixel 244| 9.64k| std::array flags; 245| 9.64k| for (unsigned int& flag : flags) ------------------ | Branch (245:27): [True: 38.5k, False: 9.64k] ------------------ 246| 38.5k| flag = pump.getBits(2); 247| | 248| 48.1k| for (int i = 0; i < 4; i++) { ------------------ | Branch (248:19): [True: 38.4k, False: 9.61k] ------------------ 249| | // The color is 0-Green 1-Blue 2-Red 250| 38.4k| uint32_t colornum = (row % 2 != 0) ? i >> 1 : ((i >> 1) + 2) % 3; ------------------ | Branch (250:25): [True: 16.7k, False: 21.7k] ------------------ 251| | 252| 38.4k| assert(flags[i] <= 3); ------------------ | Branch (252:5): [True: 38.4k, False: 0] ------------------ 253| 38.4k| switch (flags[i]) { 254| 17.3k| case 0: ------------------ | Branch (254:5): [True: 17.3k, False: 21.1k] ------------------ 255| 17.3k| diffBits[i] = diffBitsMode[colornum][0]; 256| 17.3k| break; 257| 6.30k| case 1: ------------------ | Branch (257:5): [True: 6.30k, False: 32.1k] ------------------ 258| 6.30k| diffBits[i] = diffBitsMode[colornum][0] + 1; 259| 6.30k| break; 260| 5.42k| case 2: ------------------ | Branch (260:5): [True: 5.42k, False: 33.0k] ------------------ 261| 5.42k| if (diffBitsMode[colornum][0] == 0) ------------------ | Branch (261:11): [True: 7, False: 5.42k] ------------------ 262| 5.42k| ThrowRDE("Difference bits underflow. File corrupted?"); ------------------ | | 38| 7| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 7| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 7| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 7| __VA_ARGS__) | | ------------------ ------------------ 263| 5.42k| diffBits[i] = diffBitsMode[colornum][0] - 1; 264| 5.42k| break; 265| 9.37k| case 3: ------------------ | Branch (265:5): [True: 9.37k, False: 29.1k] ------------------ 266| 9.37k| diffBits[i] = pump.getBits(4); 267| 9.37k| break; 268| 0| default: ------------------ | Branch (268:5): [True: 0, False: 38.4k] ------------------ 269| 0| __builtin_unreachable(); 270| 38.4k| } 271| | 272| 38.4k| diffBitsMode[colornum][0] = diffBitsMode[colornum][1]; 273| 38.4k| diffBitsMode[colornum][1] = diffBits[i]; 274| | 275| 38.4k| if (diffBits[i] > bitDepth + 1) ------------------ | Branch (275:9): [True: 20, False: 38.4k] ------------------ 276| 38.4k| ThrowRDE("Too many difference bits (%u). File corrupted?", diffBits[i]); ------------------ | | 38| 20| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 20| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 20| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 20| __VA_ARGS__) | | ------------------ ------------------ 277| 38.4k| assert(diffBits[i] <= 15 && "So any difference fits within uint16_t"); ------------------ | Branch (277:5): [True: 38.4k, False: 0] | Branch (277:5): [True: 38.4k, Folded] | Branch (277:5): [True: 38.4k, False: 0] ------------------ 278| 38.4k| } 279| | 280| 9.61k| return diffBits; 281| 9.64k|} _ZN8rawspeed21SamsungV2Decompressor7getDiffERNS_16BitStreamerMSB32Ej: 83| 432k|SamsungV2Decompressor::getDiff(BitStreamerMSB32& pump, uint32_t len) { 84| 432k| if (len == 0) ------------------ | Branch (84:7): [True: 290k, False: 141k] ------------------ 85| 290k| return 0; 86| 141k| invariant(len <= 15 && "Difference occupies at most 15 bits."); ------------------ | | 27| 141k|#define invariant(expr) assert(expr) ------------------ | Branch (86:3): [True: 141k, False: 0] | Branch (86:3): [True: 141k, Folded] | Branch (86:3): [True: 141k, False: 0] ------------------ 87| 141k| return implicit_cast(signExtend(pump.getBits(len), len)); 88| 141k|} _ZN8rawspeed21SamsungV2Decompressor10decompressEv: 358| 189|void SamsungV2Decompressor::decompress() { 359| 7.77k| for (int row = 0; row < height; row++) ------------------ | Branch (359:21): [True: 7.58k, False: 189] ------------------ 360| 7.58k| decompressRow(row); 361| 189|} _ZN8rawspeed20SonyArw1DecompressorC2ENS_8RawImageE: 40| 826| : mRaw(std::move(img)) { 41| 826| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (41:7): [True: 0, False: 826] | Branch (41:30): [True: 0, False: 826] ------------------ 42| 826| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (42:7): [True: 0, False: 826] ------------------ 43| 826| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 44| | 45| 826| const uint32_t w = mRaw->dim.x; 46| 826| const uint32_t h = mRaw->dim.y; 47| | 48| 826| if (w == 0 || h == 0 || h % 2 != 0 || w > 4600 || h > 3072) ------------------ | Branch (48:7): [True: 0, False: 826] | Branch (48:17): [True: 0, False: 826] | Branch (48:27): [True: 0, False: 826] | Branch (48:41): [True: 8, False: 818] | Branch (48:53): [True: 2, False: 816] ------------------ 49| 826| ThrowRDE("Unexpected image dimensions found: (%u; %u)", w, h); ------------------ | | 38| 10| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 10| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 10| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 10| __VA_ARGS__) | | ------------------ ------------------ 50| 826|} _ZNK8rawspeed20SonyArw1Decompressor10decompressENS_10ByteStreamE: 59| 816|void SonyArw1Decompressor::decompress(ByteStream input) const { 60| 816| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 61| 816| invariant(out.width() > 0); ------------------ | | 27| 816|#define invariant(expr) assert(expr) ------------------ | Branch (61:3): [True: 816, False: 0] ------------------ 62| 816| invariant(out.height() > 0); ------------------ | | 27| 816|#define invariant(expr) assert(expr) ------------------ | Branch (62:3): [True: 816, False: 0] ------------------ 63| 816| invariant(out.height() % 2 == 0); ------------------ | | 27| 816|#define invariant(expr) assert(expr) ------------------ | Branch (63:3): [True: 816, False: 0] ------------------ 64| | 65| 816| BitStreamerMSB bits(input.peekRemainingBuffer().getAsArray1DRef()); 66| 816| int pred = 0; 67| 1.18k| for (int col = out.width() - 1; col >= 0; col--) { ------------------ | Branch (67:35): [True: 1.17k, False: 9] ------------------ 68| 438k| for (int row = 0; row < out.height() + 1; row += 2) { ------------------ | Branch (68:23): [True: 438k, False: 366] ------------------ 69| 438k| bits.fill(32); 70| | 71| 438k| if (row == out.height()) ------------------ | Branch (71:11): [True: 377, False: 437k] ------------------ 72| 377| row = 1; 73| | 74| 438k| uint32_t len = 4 - bits.getBitsNoFill(2); 75| | 76| 438k| if (len == 3 && bits.getBitsNoFill(1)) ------------------ | Branch (76:11): [True: 42.2k, False: 396k] | Branch (76:23): [True: 20.9k, False: 21.3k] ------------------ 77| 20.9k| len = 0; 78| | 79| 438k| if (len == 4) ------------------ | Branch (79:11): [True: 47.7k, False: 390k] ------------------ 80| 116k| while (len < 17 && !bits.getBitsNoFill(1)) ------------------ | Branch (80:16): [True: 115k, False: 493] | Branch (80:28): [True: 68.2k, False: 47.2k] ------------------ 81| 68.2k| len++; 82| | 83| 438k| int diff = getDiff(bits, len); 84| 438k| pred += diff; 85| | 86| 438k| if (!isIntN(pred, 12)) ------------------ | Branch (86:11): [True: 807, False: 437k] ------------------ 87| 438k| ThrowRDE("Error decompressing"); ------------------ | | 38| 807| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 807| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 807| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 807| __VA_ARGS__) | | ------------------ ------------------ 88| | 89| 437k| out(row, col) = implicit_cast(pred); 90| 437k| } 91| 1.17k| } 92| 816|} _ZN8rawspeed20SonyArw1Decompressor7getDiffERNS_14BitStreamerMSBEj: 52| 438k|inline int SonyArw1Decompressor::getDiff(BitStreamerMSB& bs, uint32_t len) { 53| 438k| if (len == 0) ------------------ | Branch (53:7): [True: 20.9k, False: 417k] ------------------ 54| 20.9k| return 0; 55| 417k| int diff = bs.getBitsNoFill(len); 56| 417k| return PrefixCodeDecoder<>::extend(diff, len); 57| 438k|} _ZN8rawspeed20SonyArw2DecompressorC2ENS_8RawImageENS_10ByteStreamE: 43| 92| : mRaw(std::move(img)) { 44| 92| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (44:7): [True: 0, False: 92] | Branch (44:30): [True: 0, False: 92] ------------------ 45| 92| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (45:7): [True: 0, False: 92] ------------------ 46| 92| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 47| | 48| 92| if (!mRaw->dim.hasPositiveArea() || mRaw->dim.x % 32 != 0 || ------------------ | Branch (48:7): [True: 0, False: 92] | Branch (48:39): [True: 2, False: 90] ------------------ 49| 90| mRaw->dim.x > 9600 || mRaw->dim.y > 6376) ------------------ | Branch (49:7): [True: 0, False: 90] | Branch (49:29): [True: 0, False: 90] ------------------ 50| 92| ThrowRDE("Unexpected image dimensions found: (%d; %d)", mRaw->dim.x, ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 51| 90| mRaw->dim.y); 52| | 53| | // 1 byte per pixel 54| 90| input = input_.peekStream(mRaw->dim.x * mRaw->dim.y); 55| 90|} _ZNK8rawspeed20SonyArw2Decompressor13decompressRowEi: 57| 18.2k|void SonyArw2Decompressor::decompressRow(int row) const { 58| 18.2k| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 59| 18.2k| invariant(out.width() > 0); ------------------ | | 27| 18.2k|#define invariant(expr) assert(expr) ------------------ | Branch (59:3): [True: 18.1k, False: 161] ------------------ 60| 18.1k| invariant(out.width() % 32 == 0); ------------------ | | 27| 18.1k|#define invariant(expr) assert(expr) ------------------ | Branch (60:3): [True: 18.0k, False: 36] ------------------ 61| | 62| | // Allow compiler to devirtualize the calls below. 63| 18.0k| auto& rawdata = reinterpret_cast(*mRaw); 64| | 65| 18.0k| ByteStream rowBs = input; 66| 18.0k| rowBs.skipBytes(row * out.width()); 67| 18.0k| rowBs = rowBs.peekStream(out.width()); 68| | 69| 18.0k| BitStreamerLSB bits(rowBs.peekRemainingBuffer().getAsArray1DRef()); 70| | 71| 18.0k| uint32_t random = bits.peekBits(24); 72| | 73| | // Each loop iteration processes 16 pixels, consuming 128 bits of input. 74| 39.3k| for (int col = 0; col < out.width(); col += ((col & 1) != 0) ? 31 : 1) { ------------------ | Branch (74:21): [True: 35.3k, False: 4.05k] | Branch (74:47): [True: 9.03k, False: 12.2k] ------------------ 75| | // 30 bits. 76| 35.3k| int _max = bits.getBits(11); 77| 35.3k| int _min = bits.getBits(11); 78| 35.3k| int _imax = bits.getBits(4); 79| 35.3k| int _imin = bits.getBits(4); 80| | 81| | // 128-30 = 98 bits remaining, still need to decode 16 pixels... 82| | // Each full pixel consumes 7 bits, thus we can only have 14 full pixels. 83| | // So we lack 2 pixels. That is where _imin and _imax come into play, 84| | // values of those pixels were already specified in _min and _max. 85| | // But what that means is, _imin and _imax must not be equal! 86| 35.3k| if (_imax == _imin) ------------------ | Branch (86:9): [True: 14.0k, False: 21.2k] ------------------ 87| 35.3k| ThrowRDE("ARW2 invariant failed, same pixel is both min and max"); ------------------ | | 38| 14.0k| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 14.0k| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 14.0k| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 14.0k| __VA_ARGS__) | | ------------------ ------------------ 88| | 89| 21.2k| int sh = 0; 90| 49.8k| while ((sh < 4) && ((0x80 << sh) <= (_max - _min))) ------------------ | Branch (90:12): [True: 47.5k, False: 2.27k] | Branch (90:24): [True: 28.5k, False: 19.0k] ------------------ 91| 28.5k| sh++; 92| | 93| 331k| for (int i = 0; i < 16; i++) { ------------------ | Branch (93:21): [True: 310k, False: 21.2k] ------------------ 94| 310k| int p; 95| 310k| if (i == _imax) ------------------ | Branch (95:11): [True: 21.4k, False: 288k] ------------------ 96| 21.4k| p = _max; 97| 288k| else { 98| 288k| if (i == _imin) ------------------ | Branch (98:13): [True: 21.4k, False: 267k] ------------------ 99| 21.4k| p = _min; 100| 267k| else { 101| 267k| p = (bits.getBits(7) << sh) + _min; 102| 267k| p = std::min(p, 0x7ff); 103| 267k| } 104| 288k| } 105| 310k| rawdata.setWithLookUp( 106| 310k| implicit_cast(p << 1), 107| 310k| reinterpret_cast(&out(row, col + (i * 2))), &random); 108| 310k| } 109| 21.2k| } 110| 18.0k|} _ZNK8rawspeed20SonyArw2Decompressor16decompressThreadEv: 112| 55.6k|void SonyArw2Decompressor::decompressThread() const noexcept { 113| 55.6k| invariant(mRaw->dim.x > 0); ------------------ | | 27| 55.6k|#define invariant(expr) assert(expr) ------------------ | Branch (113:3): [True: 2.63k, False: 53.0k] ------------------ 114| 2.63k| invariant(mRaw->dim.x % 32 == 0); ------------------ | | 27| 2.63k|#define invariant(expr) assert(expr) ------------------ | Branch (114:3): [True: 2.60k, False: 30] ------------------ 115| 2.60k| invariant(mRaw->dim.y > 0); ------------------ | | 27| 2.60k|#define invariant(expr) assert(expr) ------------------ | Branch (115:3): [True: 2.59k, False: 9] ------------------ 116| | 117| 2.59k|#ifdef HAVE_OPENMP 118| 2.59k|#pragma omp for schedule(static) 119| 2.59k|#endif 120| 2.59k| for (int y = 0; y < mRaw->dim.y; y++) { 121| 2.59k| try { 122| 2.59k| decompressRow(y); 123| 2.59k| } catch (const RawspeedException& err) { 124| | // Propagate the exception out of OpenMP magic. 125| 2.59k| mRaw->setError(err.what()); 126| 2.59k|#ifdef HAVE_OPENMP 127| 2.59k|#pragma omp cancel for 128| 2.59k|#endif 129| 2.59k| } catch (...) { 130| | // We should not get any other exception type here. 131| 2.59k| __builtin_unreachable(); 132| 2.59k| } 133| 2.59k| } 134| 2.59k|} _ZNK8rawspeed20SonyArw2Decompressor10decompressEv: 136| 84|void SonyArw2Decompressor::decompress() const { 137| 84|#ifdef HAVE_OPENMP 138| 84|#pragma omp parallel default(none) \ 139| 84| num_threads(rawspeed_get_number_of_processor_cores()) 140| 84|#endif 141| 84| decompressThread(); 142| | 143| 84| std::string firstErr; 144| 84| if (mRaw->isTooManyErrors(1, &firstErr)) { ------------------ | Branch (144:7): [True: 83, False: 1] ------------------ 145| 83| ThrowRDE("Too many errors encountered. Giving up. First Error:\n%s", ------------------ | | 38| 83| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 83| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 83| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 83| __VA_ARGS__) | | ------------------ ------------------ 146| 83| firstErr.c_str()); 147| 83| } 148| 84|} SonyArw2Decompressor.cpp:_ZNK8rawspeed20SonyArw2Decompressor10decompressEv.omp_outlined_debug__: 141| 2.65k| decompressThread(); _ZN8rawspeed24UncompressedDecompressor13decode8BitRawILb0EEEvv: 271| 17|void UncompressedDecompressor::decode8BitRaw() { 272| 17| uint32_t w = size.x; 273| 17| uint32_t h = size.y; 274| 17| sanityCheck(w, &h, 1); 275| | 276| 17| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 277| | 278| 17| const auto in = Array2DRef(input.getData(w * h), w, h); 279| 17| uint32_t random = 0; 280| 370| for (uint32_t row = 0; row < h; row++) { ------------------ | Branch (280:26): [True: 353, False: 17] ------------------ 281| 2.06k| for (uint32_t col = 0; col < w; col++) { ------------------ | Branch (281:28): [True: 1.71k, False: 353] ------------------ 282| | if constexpr (uncorrectedRawValues) 283| | out(row, col) = in(row, col); 284| 1.71k| else { 285| 1.71k| mRaw->setWithLookUp(in(row, col), 286| 1.71k| reinterpret_cast(&out(row, col)), 287| 1.71k| &random); 288| 1.71k| } 289| 1.71k| } 290| 353| } 291| 17|} _ZN8rawspeed24UncompressedDecompressor25decode12BitRawWithControlILNS_10EndiannessE57005EEEvv: 297| 63|void UncompressedDecompressor::decode12BitRawWithControl() { 298| 63| uint32_t w = size.x; 299| 63| uint32_t h = size.y; 300| 63| static constexpr const auto bits = 12; 301| | 302| 63| static_assert(e == Endianness::little || e == Endianness::big, 303| 63| "unknown endianness"); 304| | 305| 63| static constexpr const auto shift = 16 - bits; 306| 63| static constexpr const auto pack = 8 - shift; 307| 63| static constexpr const auto mask = (1 << pack) - 1; 308| | 309| 63| static_assert(bits == 12 && pack == 4, "wrong pack"); 310| | 311| 63| static_assert(bits == 12 && mask == 0x0f, "wrong mask"); 312| | 313| 63| uint32_t perline = bytesPerLine(w, /*skips=*/true); 314| | 315| 63| sanityCheck(&h, perline); 316| | 317| 63| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 318| | 319| 63| const auto in = Array2DRef(input.getData(perline * h), perline, h); 320| 1.66k| for (uint32_t row = 0; row < h; row++) { ------------------ | Branch (320:26): [True: 1.60k, False: 63] ------------------ 321| 1.60k| uint32_t col = 0; 322| 407k| for (uint32_t x = 0; x < w; x += 2) { ------------------ | Branch (322:26): [True: 405k, False: 1.60k] ------------------ 323| 405k| uint32_t g1 = in(row, col + 0); 324| 405k| uint32_t g2 = in(row, col + 1); 325| | 326| 405k| auto process = [out, row](uint32_t i, bool invert, uint32_t p1, 327| 405k| uint32_t p2) { 328| 405k| uint16_t pix; 329| 405k| if (!(invert ^ (e == Endianness::little))) 330| 405k| pix = implicit_cast((p1 << pack) | (p2 >> pack)); 331| 405k| else 332| 405k| pix = implicit_cast(((p2 & mask) << 8) | p1); 333| 405k| out(row, i) = pix; 334| 405k| }; 335| | 336| 405k| process(x, false, g1, g2); 337| | 338| 405k| g1 = in(row, col + 2); 339| | 340| 405k| process(x + 1, true, g1, g2); 341| | 342| 405k| col += 3; 343| | 344| 405k| if ((x % 10) == 8) ------------------ | Branch (344:11): [True: 80.4k, False: 325k] ------------------ 345| 80.4k| col++; 346| 405k| } 347| 1.60k| } 348| 63| input.skipBytes(input.getRemainSize()); 349| 63|} _ZZN8rawspeed24UncompressedDecompressor25decode12BitRawWithControlILNS_10EndiannessE57005EEEvvENKUljbjjE_clEjbjj: 327| 811k| uint32_t p2) { 328| 811k| uint16_t pix; 329| 811k| if (!(invert ^ (e == Endianness::little))) ------------------ | Branch (329:13): [True: 405k, False: 405k] ------------------ 330| 405k| pix = implicit_cast((p1 << pack) | (p2 >> pack)); 331| 405k| else 332| 405k| pix = implicit_cast(((p2 & mask) << 8) | p1); 333| 811k| out(row, i) = pix; 334| 811k| }; _ZN8rawspeed24UncompressedDecompressor25decode12BitRawWithControlILNS_10EndiannessE48879EEEvv: 297| 22|void UncompressedDecompressor::decode12BitRawWithControl() { 298| 22| uint32_t w = size.x; 299| 22| uint32_t h = size.y; 300| 22| static constexpr const auto bits = 12; 301| | 302| 22| static_assert(e == Endianness::little || e == Endianness::big, 303| 22| "unknown endianness"); 304| | 305| 22| static constexpr const auto shift = 16 - bits; 306| 22| static constexpr const auto pack = 8 - shift; 307| 22| static constexpr const auto mask = (1 << pack) - 1; 308| | 309| 22| static_assert(bits == 12 && pack == 4, "wrong pack"); 310| | 311| 22| static_assert(bits == 12 && mask == 0x0f, "wrong mask"); 312| | 313| 22| uint32_t perline = bytesPerLine(w, /*skips=*/true); 314| | 315| 22| sanityCheck(&h, perline); 316| | 317| 22| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 318| | 319| 22| const auto in = Array2DRef(input.getData(perline * h), perline, h); 320| 331| for (uint32_t row = 0; row < h; row++) { ------------------ | Branch (320:26): [True: 309, False: 22] ------------------ 321| 309| uint32_t col = 0; 322| 3.94k| for (uint32_t x = 0; x < w; x += 2) { ------------------ | Branch (322:26): [True: 3.63k, False: 309] ------------------ 323| 3.63k| uint32_t g1 = in(row, col + 0); 324| 3.63k| uint32_t g2 = in(row, col + 1); 325| | 326| 3.63k| auto process = [out, row](uint32_t i, bool invert, uint32_t p1, 327| 3.63k| uint32_t p2) { 328| 3.63k| uint16_t pix; 329| 3.63k| if (!(invert ^ (e == Endianness::little))) 330| 3.63k| pix = implicit_cast((p1 << pack) | (p2 >> pack)); 331| 3.63k| else 332| 3.63k| pix = implicit_cast(((p2 & mask) << 8) | p1); 333| 3.63k| out(row, i) = pix; 334| 3.63k| }; 335| | 336| 3.63k| process(x, false, g1, g2); 337| | 338| 3.63k| g1 = in(row, col + 2); 339| | 340| 3.63k| process(x + 1, true, g1, g2); 341| | 342| 3.63k| col += 3; 343| | 344| 3.63k| if ((x % 10) == 8) ------------------ | Branch (344:11): [True: 653, False: 2.97k] ------------------ 345| 653| col++; 346| 3.63k| } 347| 309| } 348| 22| input.skipBytes(input.getRemainSize()); 349| 22|} _ZZN8rawspeed24UncompressedDecompressor25decode12BitRawWithControlILNS_10EndiannessE48879EEEvvENKUljbjjE_clEjbjj: 327| 7.26k| uint32_t p2) { 328| 7.26k| uint16_t pix; 329| 7.26k| if (!(invert ^ (e == Endianness::little))) ------------------ | Branch (329:13): [True: 3.63k, False: 3.63k] ------------------ 330| 3.63k| pix = implicit_cast((p1 << pack) | (p2 >> pack)); 331| 3.63k| else 332| 3.63k| pix = implicit_cast(((p2 & mask) << 8) | p1); 333| 7.26k| out(row, i) = pix; 334| 7.26k| }; _ZN8rawspeed24UncompressedDecompressor33decode12BitRawUnpackedLeftAlignedILNS_10EndiannessE57005EEEvv: 357| 73|void UncompressedDecompressor::decode12BitRawUnpackedLeftAligned() { 358| 73| uint32_t w = size.x; 359| 73| uint32_t h = size.y; 360| 73| sanityCheck(w, &h, 2); 361| | 362| 73| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 363| 73| const auto in = Array2DRef(input.getData(w * h * 2), 2 * w, h); 364| | 365| 11.8k| for (int row = 0; row < static_cast(h); row++) { ------------------ | Branch (365:21): [True: 11.8k, False: 73] ------------------ 366| 310k| for (int col = 0; col < static_cast(w); ++col) { ------------------ | Branch (366:23): [True: 298k, False: 11.8k] ------------------ 367| 298k| uint32_t g1 = in(row, (2 * col) + 0); 368| 298k| uint32_t g2 = in(row, (2 * col) + 1); 369| | 370| 298k| uint16_t pix; 371| | if constexpr (e == Endianness::little) 372| 298k| pix = implicit_cast((g2 << 8) | g1); 373| | else 374| | pix = implicit_cast((g1 << 8) | g2); 375| 298k| out(row, col) = pix >> 4; 376| 298k| } 377| 11.8k| } 378| 73|} _ZNK8rawspeed24UncompressedDecompressor11sanityCheckEPKji: 53| 3.79k| int bytesPerLine) const { 54| 3.79k| invariant(h != nullptr); ------------------ | | 27| 3.79k|#define invariant(expr) assert(expr) ------------------ | Branch (54:3): [True: 3.79k, False: 18.4E] ------------------ 55| 3.79k| invariant(*h > 0); ------------------ | | 27| 3.79k|#define invariant(expr) assert(expr) ------------------ | Branch (55:3): [True: 3.79k, False: 0] ------------------ 56| 3.79k| invariant(bytesPerLine > 0); ------------------ | | 27| 3.79k|#define invariant(expr) assert(expr) ------------------ | Branch (56:3): [True: 3.79k, False: 0] ------------------ 57| 3.79k| invariant(input.getSize() > 0); ------------------ | | 27| 3.79k|#define invariant(expr) assert(expr) ------------------ | Branch (57:3): [True: 3.79k, False: 0] ------------------ 58| | 59| | // How many multiples of bpl are there in the input buffer? 60| | // The remainder is ignored/discarded. 61| 3.79k| const auto fullRows = input.getRemainSize() / bytesPerLine; 62| | 63| | // If more than the output height, we are all good. 64| 3.79k| if (fullRows >= *h) ------------------ | Branch (64:7): [True: 3.79k, False: 1] ------------------ 65| 3.79k| return; // all good! 66| | 67| 1| if (fullRows == 0) ------------------ | Branch (67:7): [True: 0, False: 1] ------------------ 68| 1| ThrowIOE("Not enough data to decode a single line. Image file truncated."); ------------------ | | 37| 0|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 69| | 70| 1| ThrowIOE("Image truncated, only %u of %u lines found", fullRows, *h); ------------------ | | 37| 1|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 71| | 72| | // FIXME: need to come up with some common variable to allow proceeding here 73| | // *h = min_h; 74| 1|} _ZNK8rawspeed24UncompressedDecompressor11sanityCheckEjPKji: 77| 90| int bpp) const { 78| 90| invariant(w > 0); ------------------ | | 27| 90|#define invariant(expr) assert(expr) ------------------ | Branch (78:3): [True: 90, False: 0] ------------------ 79| 90| invariant(bpp > 0); ------------------ | | 27| 90|#define invariant(expr) assert(expr) ------------------ | Branch (79:3): [True: 90, False: 0] ------------------ 80| | 81| | // bytes per line 82| 90| const auto bpl = bpp * w; 83| 90| invariant(bpl > 0); ------------------ | | 27| 90|#define invariant(expr) assert(expr) ------------------ | Branch (83:3): [True: 90, False: 0] ------------------ 84| | 85| 90| sanityCheck(h, bpl); 86| 90|} _ZN8rawspeed24UncompressedDecompressor12bytesPerLineEib: 88| 85|int UncompressedDecompressor::bytesPerLine(int w, bool skips) { 89| 85| invariant(w > 0); ------------------ | | 27| 85|#define invariant(expr) assert(expr) ------------------ | Branch (89:3): [True: 85, False: 0] ------------------ 90| | 91| 85| if ((12 * w) % 8 != 0) ------------------ | Branch (91:7): [True: 0, False: 85] ------------------ 92| 85| ThrowIOE("Bad image width"); ------------------ | | 37| 0|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 93| | 94| | // Calculate expected bytes per line. 95| 85| auto perline = (12 * w) / 8; 96| | 97| 85| if (!skips) ------------------ | Branch (97:7): [True: 0, False: 85] ------------------ 98| 0| return perline; 99| | 100| | // Add skips every 10 pixels 101| 85| perline += ((w + 2) / 10); 102| | 103| 85| return perline; 104| 85|} _ZN8rawspeed24UncompressedDecompressorC2ENS_10ByteStreamENS_8RawImageERKNS_12iRectangle2DEiiNS_8BitOrderE: 109| 3.76k| : input(input_.getStream(crop.dim.y, inputPitchBytes_)), 110| 3.76k| mRaw(std::move(img_)), size(crop.dim), offset(crop.pos), 111| 3.76k| inputPitchBytes(inputPitchBytes_), bitPerPixel(bitPerPixel_), 112| 3.76k| order(order_) { 113| 3.76k| if (!size.hasPositiveArea()) ------------------ | Branch (113:7): [True: 0, False: 3.76k] ------------------ 114| 3.76k| ThrowRDE("Empty tile."); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 115| | 116| 3.76k| if (inputPitchBytes < 1) ------------------ | Branch (116:7): [True: 0, False: 3.76k] ------------------ 117| 3.76k| ThrowRDE("Input pitch is non-positive"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 118| | 119| 3.76k| switch (order) { ------------------ | Branch (119:11): [True: 3.64k, False: 115] ------------------ 120| 1.29k| case BitOrder::LSB: ------------------ | Branch (120:3): [True: 1.29k, False: 2.46k] ------------------ 121| 3.63k| case BitOrder::MSB: ------------------ | Branch (121:3): [True: 2.33k, False: 1.42k] ------------------ 122| 3.63k| case BitOrder::MSB16: ------------------ | Branch (122:3): [True: 0, False: 3.76k] ------------------ 123| 3.64k| case BitOrder::MSB32: ------------------ | Branch (123:3): [True: 18, False: 3.74k] ------------------ 124| 3.64k| break; 125| 0| case BitOrder::JPEG: ------------------ | Branch (125:3): [True: 0, False: 3.76k] ------------------ 126| 0| ThrowRDE("JPEG bit order not supported."); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 127| 3.76k| } 128| | 129| 3.64k| uint32_t w = size.x; 130| 3.64k| uint32_t h = size.y; 131| 3.64k| uint32_t cpp = mRaw->getCpp(); 132| 3.64k| uint64_t ox = offset.x; 133| 3.64k| uint64_t oy = offset.y; 134| | 135| 3.64k| if (cpp < 1 || cpp > 3) ------------------ | Branch (135:7): [True: 18.4E, False: 3.64k] | Branch (135:18): [True: 10, False: 3.63k] ------------------ 136| 3.64k| ThrowRDE("Unsupported number of components per pixel: %u", cpp); ------------------ | | 38| 10| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 10| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 10| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 10| __VA_ARGS__) | | ------------------ ------------------ 137| | 138| 3.63k| if (bitPerPixel < 1 || bitPerPixel > 32 || ------------------ | Branch (138:7): [True: 18.4E, False: 3.63k] | Branch (138:26): [True: 2, False: 3.63k] ------------------ 139| 3.63k| (bitPerPixel > 16 && mRaw->getDataType() == RawImageType::UINT16)) ------------------ | Branch (139:8): [True: 176, False: 3.46k] | Branch (139:28): [True: 2, False: 174] ------------------ 140| 3.63k| ThrowRDE("Unsupported bit depth"); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 141| | 142| 3.63k| const auto outPixelBits = static_cast(w) * cpp * bitPerPixel; 143| 3.63k| invariant(outPixelBits > 0); ------------------ | | 27| 3.63k|#define invariant(expr) assert(expr) ------------------ | Branch (143:3): [True: 3.63k, False: 0] ------------------ 144| | 145| 3.63k| if (outPixelBits % 8 != 0) { ------------------ | Branch (145:7): [True: 5, False: 3.62k] ------------------ 146| 5| ThrowRDE("Bad combination of cpp (%u), bps (%d) and width (%u), the " ------------------ | | 38| 5| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 5| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 5| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 5| __VA_ARGS__) | | ------------------ ------------------ 147| 5| "pitch is %" PRIu64 " bits, which is not a multiple of 8 (1 byte)", 148| 5| cpp, bitPerPixel, w, outPixelBits); 149| 5| } 150| | 151| 3.62k| const auto outPixelBytes = outPixelBits / 8; 152| | 153| | // The input pitch might be larger than needed (i.e. have some padding), 154| | // but it can *not* be smaller than needed. 155| 3.62k| if (static_cast(inputPitchBytes) < outPixelBytes) ------------------ | Branch (155:7): [True: 14, False: 3.61k] ------------------ 156| 3.62k| ThrowRDE("Specified pitch is smaller than minimally-required pitch"); ------------------ | | 38| 14| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 14| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 14| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 14| __VA_ARGS__) | | ------------------ ------------------ 157| | 158| | // Check the specified pitch, not the minimally-required pitch. 159| 3.61k| sanityCheck(&h, inputPitchBytes); 160| | 161| 3.61k| invariant(static_cast(inputPitchBytes) >= outPixelBytes); ------------------ | | 27| 3.61k|#define invariant(expr) assert(expr) ------------------ | Branch (161:3): [True: 3.61k, False: 0] ------------------ 162| 3.61k| skipBytes = 163| 3.61k| implicit_cast(inputPitchBytes - outPixelBytes); // Skip per line 164| | 165| 3.61k| if (oy > static_cast(mRaw->dim.y)) ------------------ | Branch (165:7): [True: 0, False: 3.61k] ------------------ 166| 3.61k| ThrowRDE("Invalid y offset"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 167| 3.61k| if (ox + size.x > static_cast(mRaw->dim.x)) ------------------ | Branch (167:7): [True: 0, False: 3.61k] ------------------ 168| 3.61k| ThrowRDE("Invalid x offset"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 169| 3.61k|} _ZN8rawspeed24UncompressedDecompressor19readUncompressedRawEv: 202| 3.44k|void UncompressedDecompressor::readUncompressedRaw() { 203| 3.44k| uint32_t outPitch = mRaw->pitch; 204| 3.44k| uint32_t w = size.x; 205| 3.44k| uint32_t h = size.y; 206| 3.44k| uint32_t cpp = mRaw->getCpp(); 207| 3.44k| uint64_t oy = offset.y; 208| | 209| 3.44k| uint64_t y = oy; 210| 3.44k| h = implicit_cast(min(h + oy, static_cast(mRaw->dim.y))); 211| | 212| 3.44k| if (mRaw->getDataType() == RawImageType::F32) { ------------------ | Branch (212:7): [True: 340, False: 3.10k] ------------------ 213| 340| if (bitPerPixel == 32) { ------------------ | Branch (213:9): [True: 28, False: 312] ------------------ 214| 28| const Array2DRef out(mRaw->getF32DataAsUncroppedArray2DRef()); 215| 28| copyPixels( 216| 28| reinterpret_cast( 217| 28| &out(implicit_cast(y), offset.x * cpp)), 218| 28| outPitch, 219| 28| reinterpret_cast(input.getData( 220| 28| implicit_cast(inputPitchBytes * (h - y)))), 221| 28| inputPitchBytes, w * mRaw->getBpp(), implicit_cast(h - y)); 222| 28| return; 223| 28| } 224| 312| if (BitOrder::MSB == order && bitPerPixel == 16) { ------------------ | Branch (224:9): [True: 130, False: 182] | Branch (224:35): [True: 68, False: 62] ------------------ 225| 68| decodePackedFP( 226| 68| h, implicit_cast(y)); 227| 68| return; 228| 68| } 229| 244| if (BitOrder::LSB == order && bitPerPixel == 16) { ------------------ | Branch (229:9): [True: 182, False: 62] | Branch (229:35): [True: 99, False: 83] ------------------ 230| 99| decodePackedFP( 231| 99| h, implicit_cast(y)); 232| 99| return; 233| 99| } 234| 145| if (BitOrder::MSB == order && bitPerPixel == 24) { ------------------ | Branch (234:9): [True: 62, False: 83] | Branch (234:35): [True: 62, False: 0] ------------------ 235| 62| decodePackedFP( 236| 62| h, implicit_cast(y)); 237| 62| return; 238| 62| } 239| 83| if (BitOrder::LSB == order && bitPerPixel == 24) { ------------------ | Branch (239:9): [True: 83, False: 0] | Branch (239:35): [True: 83, False: 0] ------------------ 240| 83| decodePackedFP( 241| 83| h, implicit_cast(y)); 242| 83| return; 243| 83| } 244| 83| ThrowRDE("Unsupported floating-point input bitwidth/bit packing: %d / %u", ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 245| 83| bitPerPixel, static_cast(order)); 246| 83| } 247| | 248| 3.10k| if (BitOrder::MSB == order) { ------------------ | Branch (248:7): [True: 2.15k, False: 942] ------------------ 249| 2.15k| decodePackedInt(h, implicit_cast(y)); 250| 2.15k| } else if (BitOrder::MSB16 == order) { ------------------ | Branch (250:14): [True: 0, False: 942] ------------------ 251| 0| decodePackedInt(h, implicit_cast(y)); 252| 942| } else if (BitOrder::MSB32 == order) { ------------------ | Branch (252:14): [True: 18, False: 924] ------------------ 253| 18| decodePackedInt(h, implicit_cast(y)); 254| 924| } else { 255| 924| if (bitPerPixel == 16 && getHostEndianness() == Endianness::little) { ------------------ | Branch (255:9): [True: 679, False: 245] | Branch (255:30): [True: 679, False: 0] ------------------ 256| 679| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 257| 679| copyPixels( 258| 679| reinterpret_cast( 259| 679| &out(implicit_cast(y), offset.x * cpp)), 260| 679| outPitch, 261| 679| reinterpret_cast(input.getData( 262| 679| implicit_cast(inputPitchBytes * (h - y)))), 263| 679| inputPitchBytes, w * mRaw->getBpp(), implicit_cast(h - y)); 264| 679| return; 265| 679| } 266| 245| decodePackedInt(h, implicit_cast(y)); 267| 245| } 268| 3.10k|} _ZNK8rawspeed24UncompressedDecompressor14decodePackedFPINS_14BitStreamerMSBENS_13ieee_754_20088Binary16EEEvii: 172| 68|void UncompressedDecompressor::decodePackedFP(int rows, int row) const { 173| 68| const Array2DRef out(mRaw->getF32DataAsUncroppedArray2DRef()); 174| 68| Pump bits(input.peekRemainingBuffer().getAsArray1DRef()); 175| | 176| 68| int cols = size.x * mRaw->getCpp(); 177| 9.38k| for (; row < rows; row++) { ------------------ | Branch (177:10): [True: 9.32k, False: 68] ------------------ 178| 32.8k| for (int col = 0; col < cols; col++) { ------------------ | Branch (178:23): [True: 23.5k, False: 9.32k] ------------------ 179| 23.5k| uint32_t b = bits.getBits(NarrowFpType::StorageWidth); 180| 23.5k| uint32_t f = 181| 23.5k| extendBinaryFloatingPoint(b); 182| 23.5k| out(row, offset.x + col) = std::bit_cast(f); 183| 23.5k| } 184| 9.32k| bits.skipBytes(skipBytes); 185| 9.32k| } 186| 68|} _ZNK8rawspeed24UncompressedDecompressor14decodePackedFPINS_14BitStreamerLSBENS_13ieee_754_20088Binary16EEEvii: 172| 99|void UncompressedDecompressor::decodePackedFP(int rows, int row) const { 173| 99| const Array2DRef out(mRaw->getF32DataAsUncroppedArray2DRef()); 174| 99| Pump bits(input.peekRemainingBuffer().getAsArray1DRef()); 175| | 176| 99| int cols = size.x * mRaw->getCpp(); 177| 11.2k| for (; row < rows; row++) { ------------------ | Branch (177:10): [True: 11.1k, False: 99] ------------------ 178| 102k| for (int col = 0; col < cols; col++) { ------------------ | Branch (178:23): [True: 91.3k, False: 11.1k] ------------------ 179| 91.3k| uint32_t b = bits.getBits(NarrowFpType::StorageWidth); 180| 91.3k| uint32_t f = 181| 91.3k| extendBinaryFloatingPoint(b); 182| 91.3k| out(row, offset.x + col) = std::bit_cast(f); 183| 91.3k| } 184| 11.1k| bits.skipBytes(skipBytes); 185| 11.1k| } 186| 99|} _ZNK8rawspeed24UncompressedDecompressor14decodePackedFPINS_14BitStreamerMSBENS_13ieee_754_20088Binary24EEEvii: 172| 62|void UncompressedDecompressor::decodePackedFP(int rows, int row) const { 173| 62| const Array2DRef out(mRaw->getF32DataAsUncroppedArray2DRef()); 174| 62| Pump bits(input.peekRemainingBuffer().getAsArray1DRef()); 175| | 176| 62| int cols = size.x * mRaw->getCpp(); 177| 21.1k| for (; row < rows; row++) { ------------------ | Branch (177:10): [True: 21.0k, False: 62] ------------------ 178| 77.9k| for (int col = 0; col < cols; col++) { ------------------ | Branch (178:23): [True: 56.8k, False: 21.0k] ------------------ 179| 56.8k| uint32_t b = bits.getBits(NarrowFpType::StorageWidth); 180| 56.8k| uint32_t f = 181| 56.8k| extendBinaryFloatingPoint(b); 182| 56.8k| out(row, offset.x + col) = std::bit_cast(f); 183| 56.8k| } 184| 21.0k| bits.skipBytes(skipBytes); 185| 21.0k| } 186| 62|} _ZNK8rawspeed24UncompressedDecompressor14decodePackedFPINS_14BitStreamerLSBENS_13ieee_754_20088Binary24EEEvii: 172| 83|void UncompressedDecompressor::decodePackedFP(int rows, int row) const { 173| 83| const Array2DRef out(mRaw->getF32DataAsUncroppedArray2DRef()); 174| 83| Pump bits(input.peekRemainingBuffer().getAsArray1DRef()); 175| | 176| 83| int cols = size.x * mRaw->getCpp(); 177| 6.36k| for (; row < rows; row++) { ------------------ | Branch (177:10): [True: 6.28k, False: 83] ------------------ 178| 36.4k| for (int col = 0; col < cols; col++) { ------------------ | Branch (178:23): [True: 30.1k, False: 6.28k] ------------------ 179| 30.1k| uint32_t b = bits.getBits(NarrowFpType::StorageWidth); 180| 30.1k| uint32_t f = 181| 30.1k| extendBinaryFloatingPoint(b); 182| 30.1k| out(row, offset.x + col) = std::bit_cast(f); 183| 30.1k| } 184| 6.28k| bits.skipBytes(skipBytes); 185| 6.28k| } 186| 83|} _ZNK8rawspeed24UncompressedDecompressor15decodePackedIntINS_14BitStreamerMSBEEEvii: 189| 2.15k|void UncompressedDecompressor::decodePackedInt(int rows, int row) const { 190| 2.15k| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 191| 2.15k| Pump bits(input.peekRemainingBuffer().getAsArray1DRef()); 192| | 193| 2.15k| int cols = size.x * mRaw->getCpp(); 194| 76.2k| for (; row < rows; row++) { ------------------ | Branch (194:10): [True: 74.0k, False: 2.15k] ------------------ 195| 3.84M| for (int x = 0; x < cols; x++) { ------------------ | Branch (195:21): [True: 3.77M, False: 74.0k] ------------------ 196| 3.77M| out(row, x) = implicit_cast(bits.getBits(bitPerPixel)); 197| 3.77M| } 198| 74.0k| bits.skipBytes(skipBytes); 199| 74.0k| } 200| 2.15k|} _ZNK8rawspeed24UncompressedDecompressor15decodePackedIntINS_16BitStreamerMSB32EEEvii: 189| 18|void UncompressedDecompressor::decodePackedInt(int rows, int row) const { 190| 18| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 191| 18| Pump bits(input.peekRemainingBuffer().getAsArray1DRef()); 192| | 193| 18| int cols = size.x * mRaw->getCpp(); 194| 56| for (; row < rows; row++) { ------------------ | Branch (194:10): [True: 38, False: 18] ------------------ 195| 1.30k| for (int x = 0; x < cols; x++) { ------------------ | Branch (195:21): [True: 1.26k, False: 38] ------------------ 196| 1.26k| out(row, x) = implicit_cast(bits.getBits(bitPerPixel)); 197| 1.26k| } 198| 38| bits.skipBytes(skipBytes); 199| 38| } 200| 18|} _ZNK8rawspeed24UncompressedDecompressor15decodePackedIntINS_14BitStreamerLSBEEEvii: 189| 245|void UncompressedDecompressor::decodePackedInt(int rows, int row) const { 190| 245| const Array2DRef out(mRaw->getU16DataAsUncroppedArray2DRef()); 191| 245| Pump bits(input.peekRemainingBuffer().getAsArray1DRef()); 192| | 193| 245| int cols = size.x * mRaw->getCpp(); 194| 6.19k| for (; row < rows; row++) { ------------------ | Branch (194:10): [True: 5.94k, False: 245] ------------------ 195| 547k| for (int x = 0; x < cols; x++) { ------------------ | Branch (195:21): [True: 541k, False: 5.94k] ------------------ 196| 541k| out(row, x) = implicit_cast(bits.getBits(bitPerPixel)); 197| 541k| } 198| 5.94k| bits.skipBytes(skipBytes); 199| 5.94k| } 200| 245|} _ZN8rawspeed15VC5Decompressor7Wavelet12setBandValidEi: 126| 23|void VC5Decompressor::Wavelet::setBandValid(const int band) { 127| 23| mDecodedBandMask |= (1 << band); 128| 23|} _ZNK8rawspeed15VC5Decompressor7Wavelet11isBandValidEi: 130| 10.4k|bool VC5Decompressor::Wavelet::isBandValid(const int band) const { 131| 10.4k| return mDecodedBandMask & (1 << band); 132| 10.4k|} _ZNK8rawspeed15VC5Decompressor7Wavelet13allBandsValidEv: 134| 23|bool VC5Decompressor::Wavelet::allBandsValid() const { 135| 23| return mDecodedBandMask == static_cast((1 << maxBands) - 1); 136| 23|} _ZN8rawspeed15VC5DecompressorC2ENS_10ByteStreamERKNS_8RawImageE: 385| 577| : mRaw(img), mBs(bs) { 386| 577| if (mRaw->getCpp() != 1 || mRaw->getDataType() != RawImageType::UINT16 || ------------------ | Branch (386:7): [True: 14, False: 563] | Branch (386:30): [True: 14, False: 549] ------------------ 387| 551| mRaw->getBpp() != sizeof(uint16_t)) ------------------ | Branch (387:7): [True: 5, False: 546] ------------------ 388| 577| ThrowRDE("Unexpected component count / data type"); ------------------ | | 38| 25| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 25| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 25| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 25| __VA_ARGS__) | | ------------------ ------------------ 389| | 390| 552| if (!mRaw->dim.hasPositiveArea()) ------------------ | Branch (390:7): [True: 0, False: 552] ------------------ 391| 552| ThrowRDE("Bad image dimensions."); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 392| | 393| 552| if (mRaw->dim.x % mVC5.patternWidth != 0) ------------------ | Branch (393:7): [True: 41, False: 511] ------------------ 394| 552| ThrowRDE("Width %i is not a multiple of %i", mRaw->dim.x, ------------------ | | 38| 41| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 41| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 41| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 41| __VA_ARGS__) | | ------------------ ------------------ 395| 511| mVC5.patternWidth); 396| | 397| 511| if (mRaw->dim.y % mVC5.patternHeight != 0) ------------------ | Branch (397:7): [True: 28, False: 483] ------------------ 398| 511| ThrowRDE("Height %i is not a multiple of %i", mRaw->dim.y, ------------------ | | 38| 28| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 28| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 28| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 28| __VA_ARGS__) | | ------------------ ------------------ 399| 483| mVC5.patternHeight); 400| | 401| 483| Optional p = getAsBayerPhase(mRaw->cfa); 402| 483| if (!p) ------------------ | Branch (402:7): [True: 102, False: 381] ------------------ 403| 483| ThrowRDE("Image has invalid CFA."); ------------------ | | 38| 102| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 102| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 102| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 102| __VA_ARGS__) | | ------------------ ------------------ 404| 381| phase = *p; 405| 381| if (phase != BayerPhase::RGGB && phase != BayerPhase::GBRG) ------------------ | Branch (405:7): [True: 227, False: 154] | Branch (405:36): [True: 13, False: 214] ------------------ 406| 381| ThrowRDE("Unexpected bayer phase, please file a bug."); ------------------ | | 38| 13| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 13| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 13| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 13| __VA_ARGS__) | | ------------------ ------------------ 407| | 408| | // Initialize wavelet sizes. 409| 1.46k| for (Channel& channel : channels) { ------------------ | Branch (409:25): [True: 1.46k, False: 368] ------------------ 410| 1.46k| auto waveletWidth = implicit_cast(mRaw->dim.x); 411| 1.46k| auto waveletHeight = implicit_cast(mRaw->dim.y); 412| 5.82k| for (Wavelet& wavelet : channel.wavelets) { ------------------ | Branch (412:27): [True: 5.82k, False: 1.46k] ------------------ 413| | // Pad dimensions as necessary and divide them by two for the next wavelet 414| 5.82k| for (auto* dimension : {&waveletWidth, &waveletHeight}) ------------------ | Branch (414:28): [True: 11.5k, False: 5.82k] ------------------ 415| 11.5k| *dimension = 416| 11.5k| implicit_cast(roundUpDivisionSafe(*dimension, 2)); 417| 5.82k| wavelet.width = waveletWidth; 418| 5.82k| wavelet.height = waveletHeight; 419| | 420| 5.82k| wavelet.bands.resize( 421| 5.82k| &wavelet == &*channel.wavelets.begin() ? 1 : Wavelet::maxBands); ------------------ | Branch (421:11): [True: 1.43k, False: 4.38k] ------------------ 422| 5.82k| } 423| 1.46k| } 424| | 425| 368| if (*img->whitePoint <= 0 || *img->whitePoint > int(((1U << 16U) - 1U))) ------------------ | Branch (425:7): [True: 72, False: 296] | Branch (425:32): [True: 21, False: 275] ------------------ 426| 368| ThrowRDE("Bad white level %i", *img->whitePoint); ------------------ | | 38| 94| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 94| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 94| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 94| __VA_ARGS__) | | ------------------ ------------------ 427| | 428| 274| outputBits = 0; 429| 2.17k| for (int wp = *img->whitePoint; wp != 0; wp >>= 1) ------------------ | Branch (429:35): [True: 1.90k, False: 274] ------------------ 430| 1.90k| ++outputBits; 431| 274| invariant(outputBits <= 16); ------------------ | | 27| 274|#define invariant(expr) assert(expr) ------------------ | Branch (431:3): [True: 275, False: 18.4E] ------------------ 432| | 433| 275| parseVC5(); 434| 275|} _ZN8rawspeed15VC5Decompressor8parseVC5Ev: 492| 275|void VC5Decompressor::parseVC5() { 493| 275| mBs.setByteOrder(Endianness::big); 494| | 495| 275| invariant(mRaw->dim.x > 0); ------------------ | | 27| 275|#define invariant(expr) assert(expr) ------------------ | Branch (495:3): [True: 274, False: 1] ------------------ 496| 274| invariant(mRaw->dim.y > 0); ------------------ | | 27| 274|#define invariant(expr) assert(expr) ------------------ | Branch (496:3): [True: 275, False: 18.4E] ------------------ 497| | 498| | // All VC-5 data must start with "VC-%" (0x56432d35) 499| 275| if (mBs.getU32() != 0x56432d35) ------------------ | Branch (499:7): [True: 102, False: 173] ------------------ 500| 275| ThrowRDE("not a valid VC-5 datablock"); ------------------ | | 38| 102| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 102| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 102| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 102| __VA_ARGS__) | | ------------------ ------------------ 501| | 502| 173| bool done = false; 503| 10.6k| while (!done) { ------------------ | Branch (503:10): [True: 10.5k, False: 54] ------------------ 504| 10.5k| auto tag = static_cast(mBs.getU16()); 505| 10.5k| uint16_t val = mBs.getU16(); 506| | 507| 10.5k| bool optional = matches(tag, VC5Tag::Optional); 508| 10.5k| if (optional) ------------------ | Branch (508:9): [True: 6.46k, False: 4.10k] ------------------ 509| 6.46k| tag = -tag; 510| | 511| 10.5k| switch (tag) { 512| 0| using enum VC5Tag; 513| 141| case ChannelCount: ------------------ | Branch (513:5): [True: 141, False: 10.4k] ------------------ 514| 141| if (val != numChannels) ------------------ | Branch (514:11): [True: 3, False: 138] ------------------ 515| 141| ThrowRDE("Bad channel count %u, expected %i", val, numChannels); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 516| 138| break; 517| 138| case ImageWidth: ------------------ | Branch (517:5): [True: 86, False: 10.4k] ------------------ 518| 86| if (val != mRaw->dim.x) ------------------ | Branch (518:11): [True: 10, False: 76] ------------------ 519| 86| ThrowRDE("Image width mismatch: %u vs %i", val, mRaw->dim.x); ------------------ | | 38| 10| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 10| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 10| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 10| __VA_ARGS__) | | ------------------ ------------------ 520| 76| break; 521| 76| case ImageHeight: ------------------ | Branch (521:5): [True: 21, False: 10.5k] ------------------ 522| 21| if (val != mRaw->dim.y) ------------------ | Branch (522:11): [True: 5, False: 16] ------------------ 523| 21| ThrowRDE("Image height mismatch: %u vs %i", val, mRaw->dim.y); ------------------ | | 38| 5| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 5| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 5| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 5| __VA_ARGS__) | | ------------------ ------------------ 524| 16| break; 525| 24| case LowpassPrecision: ------------------ | Branch (525:5): [True: 24, False: 10.5k] ------------------ 526| 24| if (val < PRECISION_MIN || val > PRECISION_MAX) ------------------ | Branch (526:11): [True: 0, False: 24] | Branch (526:34): [True: 1, False: 23] ------------------ 527| 24| ThrowRDE("Invalid precision %i", val); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 528| 23| mVC5.lowpassPrecision = val; 529| 23| break; 530| 117| case ChannelNumber: ------------------ | Branch (530:5): [True: 117, False: 10.4k] ------------------ 531| 117| if (val >= numChannels) ------------------ | Branch (531:11): [True: 5, False: 112] ------------------ 532| 117| ThrowRDE("Bad channel number (%u)", val); ------------------ | | 38| 5| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 5| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 5| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 5| __VA_ARGS__) | | ------------------ ------------------ 533| 112| mVC5.iChannel = val; 534| 112| break; 535| 11| case ImageFormat: ------------------ | Branch (535:5): [True: 11, False: 10.5k] ------------------ 536| 11| if (val != mVC5.imgFormat) ------------------ | Branch (536:11): [True: 2, False: 9] ------------------ 537| 11| ThrowRDE("Image format %i is not 4(RAW)", val); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 538| 9| break; 539| 9| case SubbandCount: ------------------ | Branch (539:5): [True: 9, False: 10.5k] ------------------ 540| 9| if (val != numSubbands) ------------------ | Branch (540:11): [True: 1, False: 8] ------------------ 541| 9| ThrowRDE("Unexpected subband count %u, expected %i", val, numSubbands); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 542| 8| break; 543| 34| case MaxBitsPerComponent: ------------------ | Branch (543:5): [True: 34, False: 10.5k] ------------------ 544| 34| if (val != VC5_LOG_TABLE_BITWIDTH) { ------------------ | Branch (544:11): [True: 1, False: 33] ------------------ 545| 1| ThrowRDE("Bad bits per componend %u, not %i", val, ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 546| 1| VC5_LOG_TABLE_BITWIDTH); 547| 1| } 548| 33| break; 549| 82| case PatternWidth: ------------------ | Branch (549:5): [True: 82, False: 10.4k] ------------------ 550| 82| if (val != mVC5.patternWidth) ------------------ | Branch (550:11): [True: 3, False: 79] ------------------ 551| 82| ThrowRDE("Bad pattern width %u, not %u", val, mVC5.patternWidth); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 552| 79| break; 553| 91| case PatternHeight: ------------------ | Branch (553:5): [True: 91, False: 10.4k] ------------------ 554| 91| if (val != mVC5.patternHeight) ------------------ | Branch (554:11): [True: 1, False: 90] ------------------ 555| 91| ThrowRDE("Bad pattern height %u, not %u", val, mVC5.patternHeight); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 556| 90| break; 557| 90| case SubbandNumber: ------------------ | Branch (557:5): [True: 38, False: 10.5k] ------------------ 558| 38| if (val >= numSubbands) ------------------ | Branch (558:11): [True: 1, False: 37] ------------------ 559| 38| ThrowRDE("Bad subband number %u", val); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 560| 37| mVC5.iSubband = val; 561| 37| break; 562| 300| case Quantization: ------------------ | Branch (562:5): [True: 300, False: 10.2k] ------------------ 563| 300| mVC5.quantization = static_cast(val); 564| 300| break; 565| 14| case ComponentsPerSample: ------------------ | Branch (565:5): [True: 14, False: 10.5k] ------------------ 566| 14| if (val != mVC5.cps) ------------------ | Branch (566:11): [True: 4, False: 10] ------------------ 567| 14| ThrowRDE("Bad component per sample count %u, not %u", val, mVC5.cps); ------------------ | | 38| 4| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 4| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 4| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 4| __VA_ARGS__) | | ------------------ ------------------ 568| 10| break; 569| 83| case PrescaleShift: ------------------ | Branch (569:5): [True: 83, False: 10.4k] ------------------ 570| | // FIXME: something is wrong. We get this before ChannelNumber. 571| | // Defaulting to 'mVC5.iChannel=0' seems to work *for existing samples*. 572| 332| for (int iWavelet = 0; iWavelet < numWaveletLevels; ++iWavelet) { ------------------ | Branch (572:30): [True: 249, False: 83] ------------------ 573| 249| auto& channel = channels[mVC5.iChannel]; 574| 249| auto& wavelet = channel.wavelets[1 + iWavelet]; 575| 249| wavelet.prescale = 576| 249| extractHighBits(val, 2 * iWavelet, /*effectiveBitwidth=*/14) & 0x03; 577| 249| } 578| 83| break; 579| 9.50k| default: { // A chunk. ------------------ | Branch (579:5): [True: 9.50k, False: 1.06k] ------------------ 580| 9.50k| unsigned int chunkSize = 0; 581| 9.50k| if (matches(tag, LARGE_CHUNK)) { ------------------ | Branch (581:11): [True: 4.03k, False: 5.47k] ------------------ 582| 4.03k| chunkSize = static_cast( 583| 4.03k| ((static_cast>(tag) & 0xff) << 16) | 584| 4.03k| (val & 0xffff)); 585| 5.47k| } else if (matches(tag, SMALL_CHUNK)) { ------------------ | Branch (585:18): [True: 203, False: 5.26k] ------------------ 586| 203| chunkSize = (val & 0xffff); 587| 203| } 588| | 589| 9.50k| if (is(tag, LargeCodeblock)) { ------------------ | Branch (589:11): [True: 46, False: 9.45k] ------------------ 590| 46| parseLargeCodeblock(mBs.getStream(chunkSize, 4)); 591| 46| break; 592| 46| } 593| | 594| | // And finally, we got here if we didn't handle this tag/maybe-chunk. 595| | 596| | // Magic, all the other 'large' chunks are actually optional, 597| | // and don't specify any chunk bytes-to-be-skipped. 598| 9.45k| if (matches(tag, LARGE_CHUNK)) { ------------------ | Branch (598:11): [True: 3.98k, False: 5.47k] ------------------ 599| 3.98k| optional = true; 600| 3.98k| chunkSize = 0; 601| 3.98k| } 602| | 603| 9.45k| if (!optional) { ------------------ | Branch (603:11): [True: 36, False: 9.42k] ------------------ 604| 36| ThrowRDE("Unknown (unhandled) non-optional Tag 0x%04hx", ------------------ | | 38| 36| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 36| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 36| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 36| __VA_ARGS__) | | ------------------ ------------------ 605| 36| static_cast>(tag)); 606| 36| } 607| | 608| 9.42k| if (chunkSize) ------------------ | Branch (608:11): [True: 115, False: 9.30k] ------------------ 609| 115| mBs.skipBytes(chunkSize, 4); 610| | 611| 9.42k| break; 612| 9.45k| } 613| 10.5k| } 614| | 615| 10.4k| done = std::all_of(channels.begin(), channels.end(), 616| 10.4k| [](const Channel& channel) { 617| 10.4k| return channel.wavelets[0].isBandValid(0); 618| 10.4k| }); 619| 10.4k| } 620| 173|} VC5Decompressor.cpp:_ZZN8rawspeed15VC5Decompressor8parseVC5EvENK3$_0clERKNS0_7ChannelE: 616| 10.4k| [](const Channel& channel) { 617| 10.4k| return channel.wavelets[0].isBandValid(0); 618| 10.4k| }); _ZN8rawspeed15VC5Decompressor7Wavelet11LowPassBandC2ERS1_NS_10ByteStreamEt: 654| 2| : AbstractDecodeableBand(wavelet_, bs.getAsArray1DRef()), 655| 2| lowpassPrecision(lowpassPrecision_) { 656| | // Low-pass band is a uncompressed version of the image, hugely downscaled. 657| | // It consists of width * height pixels, `lowpassPrecision` each. 658| | // We can easily check that we have sufficient amount of bits to decode it. 659| 2| const auto waveletArea = iPoint2D(wavelet.width, wavelet.height).area(); 660| 2| const auto bitsTotal = waveletArea * lowpassPrecision; 661| 2| constexpr int bytesPerChunk = 8; // FIXME: or is it 4? 662| 2| constexpr int bitsPerChunk = 8 * bytesPerChunk; 663| 2| const auto chunksTotal = roundUpDivisionSafe(bitsTotal, bitsPerChunk); 664| 2| const auto bytesTotal = bytesPerChunk * chunksTotal; 665| | // And clamp the size / verify sufficient input while we are at it. 666| | // NOTE: this might fail (and should throw, not assert). 667| 2| input = bs.getStream(implicit_cast(bytesTotal)) 668| 2| .getAsArray1DRef(); 669| 2|} _ZN8rawspeed15VC5Decompressor19parseLargeCodeblockENS_10ByteStreamE: 746| 31|void VC5Decompressor::parseLargeCodeblock(ByteStream bs) { 747| 31| static const auto subband_wavelet_index = []() { 748| 31| std::array wavelets; 749| 31| int wavelet = 0; 750| 31| for (auto i = wavelets.size() - 1; i > 0;) { 751| 31| for (auto t = 0; t < numWaveletLevels; t++) { 752| 31| wavelets[i] = wavelet; 753| 31| i--; 754| 31| } 755| 31| if (i > 0) 756| 31| wavelet++; 757| 31| } 758| 31| wavelets.front() = wavelet; 759| 31| return wavelets; 760| 31| }(); 761| 31| static const auto subband_band_index = []() { 762| 31| std::array bands; 763| 31| bands.front() = 0; 764| 31| for (auto i = 1U; i < bands.size();) { 765| 31| for (int t = 1; t <= numWaveletLevels;) { 766| 31| bands[i] = t; 767| 31| t++; 768| 31| i++; 769| 31| } 770| 31| } 771| 31| return bands; 772| 31| }(); 773| | 774| 31| if (!mVC5.iSubband.has_value()) ------------------ | Branch (774:7): [True: 3, False: 28] ------------------ 775| 31| ThrowRDE("Did not see VC5Tag::SubbandNumber yet"); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 776| | 777| 28| const int idx = subband_wavelet_index[*mVC5.iSubband]; 778| 28| const int band = subband_band_index[*mVC5.iSubband]; 779| | 780| 28| auto& wavelets = channels[mVC5.iChannel].wavelets; 781| | 782| 28| Wavelet& wavelet = wavelets[1 + idx]; 783| 28| if (wavelet.isBandValid(band)) { ------------------ | Branch (783:7): [True: 1, False: 27] ------------------ 784| 1| ThrowRDE("Band %i for wavelet %i on channel %u was already seen", band, idx, ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 785| 1| mVC5.iChannel); 786| 1| } 787| | 788| 27| std::unique_ptr& dstBand = wavelet.bands[band]; 789| 27| if (mVC5.iSubband == 0) { ------------------ | Branch (789:7): [True: 4, False: 23] ------------------ 790| 4| assert(band == 0); ------------------ | Branch (790:5): [True: 4, False: 0] ------------------ 791| | // low-pass band, only one, for the smallest wavelet, per channel per image 792| 4| if (!mVC5.lowpassPrecision.has_value()) ------------------ | Branch (792:9): [True: 2, False: 2] ------------------ 793| 4| ThrowRDE("Did not see VC5Tag::LowpassPrecision yet"); ------------------ | | 38| 2| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 794| 2| dstBand = std::make_unique(wavelet, bs, 795| 2| *mVC5.lowpassPrecision); 796| 2| mVC5.lowpassPrecision.reset(); 797| 23| } else { 798| 23| if (!mVC5.quantization.has_value()) ------------------ | Branch (798:9): [True: 1, False: 22] ------------------ 799| 23| ThrowRDE("Did not see VC5Tag::Quantization yet"); ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 800| 22| dstBand = std::make_unique( 801| 22| wavelet, bs.getAsArray1DRef(), codeDecoder, *mVC5.quantization); 802| 22| mVC5.quantization.reset(); 803| 22| } 804| 24| wavelet.setBandValid(band); 805| | 806| | // If this wavelet is fully specified, mark the low-pass band of the 807| | // next lower wavelet as specified. 808| 24| if (wavelet.allBandsValid()) { ------------------ | Branch (808:7): [True: 0, False: 24] ------------------ 809| 0| Wavelet& nextWavelet = wavelets[idx]; 810| 0| assert(!nextWavelet.isBandValid(0)); ------------------ | Branch (810:5): [True: 0, False: 0] ------------------ 811| 0| bool finalWavelet = idx == 0; 812| 0| nextWavelet.bands[0] = std::make_unique( 813| 0| wavelet, /*clampUint=*/finalWavelet, finalWavelet); 814| 0| nextWavelet.setBandValid(0); 815| 0| } 816| | 817| 24| mVC5.iSubband.reset(); 818| 24|} _ZN8rawspeed15VC5Decompressor7Wavelet12AbstractBandC2ERS1_: 150| 24| explicit AbstractBand(Wavelet& wavelet_) : wavelet(wavelet_) {} _ZN8rawspeed15VC5Decompressor7Wavelet22AbstractDecodeableBandC2ERS1_NS_10Array1DRefIKhEE: 178| 24| : AbstractBand(wavelet_), input(input_) {} _ZN8rawspeed15VC5Decompressor7Wavelet12HighPassBandC2ERS1_NS_10Array1DRefIKhEERKNS_8OptionalINS_20PrefixCodeLUTDecoderINS_10VC5CodeTagENS_23PrefixCodeVectorDecoderIS9_EEEEEEs: 193| 22| : AbstractDecodeableBand(wavelet_, input_), decoder(decoder_), 194| 22| quant(quant_) {} _ZN8rawspeedanENS_6VC5TagES0_: 76| 44.4k|inline VC5Tag operator&(VC5Tag LHS, VC5Tag RHS) { 77| 44.4k| using value_type = std::underlying_type_t; 78| 44.4k| return static_cast(static_cast(LHS) & 79| 44.4k| static_cast(RHS)); 80| 44.4k|} _ZN8rawspeed7matchesENS_6VC5TagES0_: 81| 34.9k|inline bool matches(VC5Tag LHS, VC5Tag RHS) { 82| | // Are there any common bit set? 83| 34.9k| return (LHS & RHS) != VC5Tag::NoTag; 84| 34.9k|} _ZN8rawspeed2isENS_6VC5TagES0_: 85| 9.50k|inline bool is(VC5Tag LHS, VC5Tag RHS) { 86| | // Does LHS have all the RHS bits set? 87| 9.50k| return (LHS & RHS) == RHS; 88| 9.50k|} _ZN8rawspeedngENS_6VC5TagE: 89| 6.46k|inline VC5Tag operator-(VC5Tag tag) { 90| 6.46k| using value_type = std::underlying_type_t; 91| | // Negate 92| 6.46k| return static_cast(-static_cast(tag)); 93| 6.46k|} _ZN8rawspeed15VC5Decompressor7Wavelet12AbstractBandD2Ev: 151| 24| virtual ~AbstractBand() = default; _ZN8rawspeed6BufferC2EPKhj: 68| 14.8k| : Buffer(Array1DRef(data_, implicit_cast(size_))) {} _ZN8rawspeed6BufferC2ENS_10Array1DRefIKhEE: 62| 26.5M| : data(data_.begin()), size(data_.size()) { 63| 26.5M| assert(data); ------------------ | Branch (63:5): [True: 26.5M, False: 2.74k] ------------------ 64| 26.5M| assert(!ASan::RegionIsPoisoned(data, size)); ------------------ | Branch (64:5): [True: 26.5M, False: 2.10k] ------------------ 65| 26.5M| } _ZNK8rawspeed6Buffer15getAsArray1DRefEv: 70| 53.8M| [[nodiscard]] Array1DRef getAsArray1DRef() const { 71| 53.8M| return {data, implicit_cast(size)}; 72| 53.8M| } _ZNK8rawspeed6Buffer10getSubViewEjj: 78| 26.6M| [[nodiscard]] Buffer getSubView(size_type offset, size_type size_) const { 79| 26.6M| if (!isValid(offset, size_)) ------------------ | Branch (79:9): [True: 103k, False: 26.5M] ------------------ 80| 26.6M| ThrowIOE("Buffer overflow: image file may be truncated"); ------------------ | | 37| 103k|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 103k| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 103k| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 103k| __VA_ARGS__) | | ------------------ ------------------ 81| | 82| 26.5M| return getAsArray1DRef().getCrop(offset, size_).getAsArray1DRef(); 83| 26.6M| } _ZNK8rawspeed6Buffer10getSubViewEj: 85| 6.14k| [[nodiscard]] Buffer getSubView(size_type offset) const { 86| 6.14k| if (!isValid(0, offset)) ------------------ | Branch (86:9): [True: 209, False: 5.93k] ------------------ 87| 6.14k| ThrowIOE("Buffer overflow: image file may be truncated"); ------------------ | | 37| 209|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 209| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 209| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 209| __VA_ARGS__) | | ------------------ ------------------ 88| | 89| 5.93k| size_type newSize = getSize() - offset; 90| 5.93k| return getSubView(offset, newSize); 91| 6.14k| } _ZNK8rawspeed6BufferixEj: 94| 12.9k| uint8_t operator[](size_type offset) const { 95| 12.9k| return getAsArray1DRef()(offset); 96| 12.9k| } _ZNK8rawspeed6Buffer5beginEv: 99| 25.5M| [[nodiscard]] const uint8_t* begin() const { 100| 25.5M| return getAsArray1DRef().begin(); 101| 25.5M| } _ZNK8rawspeed6Buffer3endEv: 102| 1.26M| [[nodiscard]] const uint8_t* end() const { return getAsArray1DRef().end(); } _ZNK8rawspeed6Buffer7getSizeEv: 115| 77.3M| [[nodiscard]] size_type RAWSPEED_READONLY getSize() const { return size; } _ZNK8rawspeed6Buffer7isValidEjj: 117| 39.9M| [[nodiscard]] bool isValid(size_type offset, size_type count = 1) const { 118| 39.9M| return static_cast(offset) + count <= 119| 39.9M| static_cast(getSize()); 120| 39.9M| } _ZN8rawspeed10DataBufferC2ENS_6BufferENS_10EndiannessE: 142| 921k| : Buffer(data_), endianness(endianness_) {} _ZNK8rawspeed10DataBuffer12getByteOrderEv: 154| 893k| [[nodiscard]] Endianness getByteOrder() const { return endianness; } _ZN8rawspeed10DataBuffer12setByteOrderENS_10EndiannessE: 156| 21.5k| Endianness setByteOrder(Endianness endianness_) { 157| 21.5k| std::swap(endianness, endianness_); 158| 21.5k| return endianness_; 159| 21.5k| } _ZN8rawspeedltENS_6BufferES0_: 124| 43.3k|inline bool operator<(Buffer lhs, Buffer rhs) { 125| 43.3k| return std::pair(lhs.begin(), lhs.end()) < std::pair(rhs.begin(), rhs.end()); 126| 43.3k|} _ZNK8rawspeed10DataBuffer3getIhEET_jj: 147| 8.58M| [[nodiscard]] T get(size_type offset, size_type index = 0) const { 148| 8.58M| assert(Endianness::unknown != endianness); ------------------ | Branch (148:5): [True: 8.58M, False: 18.4E] ------------------ 149| 8.58M| assert(Endianness::little == endianness || Endianness::big == endianness); ------------------ | Branch (149:5): [True: 142k, False: 8.44M] | Branch (149:5): [True: 8.44M, False: 0] | Branch (149:5): [True: 8.58M, False: 18.4E] ------------------ 150| | 151| 8.58M| return Buffer::get(getHostEndianness() == endianness, offset, index); 152| 8.58M| } _ZNK8rawspeed6Buffer3getIhEET_bjj: 108| 8.58M| size_type index = 0) const { 109| 8.58M| const Buffer buf = 110| 8.58M| getSubView(offset + (index * static_cast(sizeof(T))), 111| 8.58M| static_cast(sizeof(T))); 112| 8.58M| return getByteSwapped(buf.begin(), !inNativeByteOrder); 113| 8.58M| } _ZNK8rawspeed10DataBuffer3getItEET_jj: 147| 1.85M| [[nodiscard]] T get(size_type offset, size_type index = 0) const { 148| 1.85M| assert(Endianness::unknown != endianness); ------------------ | Branch (148:5): [True: 1.85M, False: 0] ------------------ 149| 1.85M| assert(Endianness::little == endianness || Endianness::big == endianness); ------------------ | Branch (149:5): [True: 1.55M, False: 291k] | Branch (149:5): [True: 291k, False: 0] | Branch (149:5): [True: 1.85M, False: 18.4E] ------------------ 150| | 151| 1.85M| return Buffer::get(getHostEndianness() == endianness, offset, index); 152| 1.85M| } _ZNK8rawspeed6Buffer3getItEET_bjj: 108| 1.85M| size_type index = 0) const { 109| 1.85M| const Buffer buf = 110| 1.85M| getSubView(offset + (index * static_cast(sizeof(T))), 111| 1.85M| static_cast(sizeof(T))); 112| 1.85M| return getByteSwapped(buf.begin(), !inNativeByteOrder); 113| 1.85M| } _ZNK8rawspeed10DataBuffer3getIjEET_jj: 147| 1.24M| [[nodiscard]] T get(size_type offset, size_type index = 0) const { 148| 1.24M| assert(Endianness::unknown != endianness); ------------------ | Branch (148:5): [True: 1.24M, False: 0] ------------------ 149| 1.24M| assert(Endianness::little == endianness || Endianness::big == endianness); ------------------ | Branch (149:5): [True: 1.11M, False: 130k] | Branch (149:5): [True: 130k, False: 0] | Branch (149:5): [True: 1.24M, False: 1] ------------------ 150| | 151| 1.24M| return Buffer::get(getHostEndianness() == endianness, offset, index); 152| 1.24M| } _ZNK8rawspeed6Buffer3getIjEET_bjj: 108| 1.24M| size_type index = 0) const { 109| 1.24M| const Buffer buf = 110| 1.24M| getSubView(offset + (index * static_cast(sizeof(T))), 111| 1.24M| static_cast(sizeof(T))); 112| 1.24M| return getByteSwapped(buf.begin(), !inNativeByteOrder); 113| 1.24M| } _ZNK8rawspeed10DataBuffer3getIiEET_jj: 147| 145k| [[nodiscard]] T get(size_type offset, size_type index = 0) const { 148| 145k| assert(Endianness::unknown != endianness); ------------------ | Branch (148:5): [True: 145k, False: 0] ------------------ 149| 145k| assert(Endianness::little == endianness || Endianness::big == endianness); ------------------ | Branch (149:5): [True: 145k, False: 9] | Branch (149:5): [True: 9, False: 0] | Branch (149:5): [True: 145k, False: 0] ------------------ 150| | 151| 145k| return Buffer::get(getHostEndianness() == endianness, offset, index); 152| 145k| } _ZNK8rawspeed6Buffer3getIiEET_bjj: 108| 145k| size_type index = 0) const { 109| 145k| const Buffer buf = 110| 145k| getSubView(offset + (index * static_cast(sizeof(T))), 111| 145k| static_cast(sizeof(T))); 112| 145k| return getByteSwapped(buf.begin(), !inNativeByteOrder); 113| 145k| } _ZNK8rawspeed10DataBuffer3getIfEET_jj: 147| 49.6k| [[nodiscard]] T get(size_type offset, size_type index = 0) const { 148| 49.6k| assert(Endianness::unknown != endianness); ------------------ | Branch (148:5): [True: 49.6k, False: 0] ------------------ 149| 49.6k| assert(Endianness::little == endianness || Endianness::big == endianness); ------------------ | Branch (149:5): [True: 390, False: 49.2k] | Branch (149:5): [True: 49.2k, False: 0] | Branch (149:5): [True: 49.6k, False: 0] ------------------ 150| | 151| 49.6k| return Buffer::get(getHostEndianness() == endianness, offset, index); 152| 49.6k| } _ZNK8rawspeed6Buffer3getIfEET_bjj: 108| 49.6k| size_type index = 0) const { 109| 49.6k| const Buffer buf = 110| 49.6k| getSubView(offset + (index * static_cast(sizeof(T))), 111| 49.6k| static_cast(sizeof(T))); 112| 49.6k| return getByteSwapped(buf.begin(), !inNativeByteOrder); 113| 49.6k| } _ZN8rawspeed10DataBufferC2Ev: 139| 277k| DataBuffer() = default; _ZN8rawspeed6BufferC2Ev: 58| 277k| Buffer() = default; _ZNK8rawspeed10DataBuffer3getIsEET_jj: 147| 1.09k| [[nodiscard]] T get(size_type offset, size_type index = 0) const { 148| 1.09k| assert(Endianness::unknown != endianness); ------------------ | Branch (148:5): [True: 1.09k, False: 0] ------------------ 149| 1.09k| assert(Endianness::little == endianness || Endianness::big == endianness); ------------------ | Branch (149:5): [True: 1.09k, False: 3] | Branch (149:5): [True: 3, False: 0] | Branch (149:5): [True: 1.09k, False: 0] ------------------ 150| | 151| 1.09k| return Buffer::get(getHostEndianness() == endianness, offset, index); 152| 1.09k| } _ZNK8rawspeed6Buffer3getIsEET_bjj: 108| 1.09k| size_type index = 0) const { 109| 1.09k| const Buffer buf = 110| 1.09k| getSubView(offset + (index * static_cast(sizeof(T))), 111| 1.09k| static_cast(sizeof(T))); 112| 1.09k| return getByteSwapped(buf.begin(), !inNativeByteOrder); 113| 1.09k| } _ZNK8rawspeed10DataBuffer3getIdEET_jj: 147| 326| [[nodiscard]] T get(size_type offset, size_type index = 0) const { 148| 326| assert(Endianness::unknown != endianness); ------------------ | Branch (148:5): [True: 326, False: 0] ------------------ 149| 326| assert(Endianness::little == endianness || Endianness::big == endianness); ------------------ | Branch (149:5): [True: 244, False: 82] | Branch (149:5): [True: 82, False: 0] | Branch (149:5): [True: 326, False: 0] ------------------ 150| | 151| 326| return Buffer::get(getHostEndianness() == endianness, offset, index); 152| 326| } _ZNK8rawspeed6Buffer3getIdEET_bjj: 108| 326| size_type index = 0) const { 109| 326| const Buffer buf = 110| 326| getSubView(offset + (index * static_cast(sizeof(T))), 111| 326| static_cast(sizeof(T))); 112| 326| return getByteSwapped(buf.begin(), !inNativeByteOrder); 113| 326| } _ZN8rawspeed10ByteStreamC2ENS_10DataBufferE: 50| 942k| explicit ByteStream(DataBuffer buffer) : DataBuffer(buffer) {} _ZNK8rawspeed10ByteStream12getSubStreamEjj: 55| 978k| size_type size_) const { 56| 978k| return ByteStream(DataBuffer(getSubView(offset, size_), getByteOrder())); 57| 978k| } _ZNK8rawspeed10ByteStream12getSubStreamEj: 59| 647| [[nodiscard]] ByteStream getSubStream(size_type offset) const { 60| 647| return ByteStream(DataBuffer(getSubView(offset), getByteOrder())); 61| 647| } _ZNK8rawspeed10ByteStream5checkEj: 63| 12.2M| [[nodiscard]] size_type check(size_type bytes) const { 64| 12.2M| if (!isValid(pos, bytes)) ------------------ | Branch (64:9): [True: 16.2k, False: 12.1M] ------------------ 65| 12.2M| ThrowIOE("Out of bounds access in ByteStream"); ------------------ | | 37| 16.2k|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 16.2k| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 16.2k| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 16.2k| __VA_ARGS__) | | ------------------ ------------------ 66| 12.1M| [[maybe_unused]] Buffer tmp = getSubView(pos, bytes); 67| 12.1M| assert(tmp.getSize() == bytes); ------------------ | Branch (67:5): [True: 12.1M, False: 18.4E] ------------------ 68| 12.1M| assert(!ASan::RegionIsPoisoned(tmp.begin(), tmp.getSize())); ------------------ | Branch (68:5): [True: 12.1M, False: 18.4E] ------------------ 69| 12.1M| return bytes; 70| 12.1M| } _ZNK8rawspeed10ByteStream5checkEjj: 72| 1.20k| [[nodiscard]] size_type check(size_type nmemb, size_type size_) const { 73| 1.20k| if (size_ && nmemb > std::numeric_limits::max() / size_) ------------------ | Branch (73:9): [True: 1.20k, False: 0] | Branch (73:18): [True: 23, False: 1.17k] ------------------ 74| 1.20k| ThrowIOE("Integer overflow when calculating stream length"); ------------------ | | 37| 23|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 23| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 23| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 23| __VA_ARGS__) | | ------------------ ------------------ 75| 1.17k| return check(nmemb * size_); 76| 1.20k| } _ZNK8rawspeed10ByteStream11getPositionEv: 78| 485k| [[nodiscard]] size_type getPosition() const { 79| 485k| invariant(getSize() >= pos); ------------------ | | 27| 485k|#define invariant(expr) assert(expr) ------------------ | Branch (79:5): [True: 485k, False: 0] ------------------ 80| 485k| (void)check(0); 81| 485k| return pos; 82| 485k| } _ZN8rawspeed10ByteStream11setPositionEj: 83| 183k| void setPosition(size_type newPos) { 84| 183k| pos = newPos; 85| 183k| (void)check(0); 86| 183k| } _ZNK8rawspeed10ByteStream13getRemainSizeEv: 87| 5.74M| [[nodiscard]] size_type RAWSPEED_READONLY getRemainSize() const { 88| 5.74M| invariant(getSize() >= pos); ------------------ | | 27| 5.74M|#define invariant(expr) assert(expr) ------------------ | Branch (88:5): [True: 5.74M, False: 18.4E] ------------------ 89| 5.74M| (void)check(0); 90| 5.74M| return getSize() - pos; 91| 5.74M| } _ZNK8rawspeed10ByteStream8peekDataEj: 92| 1.36k| [[nodiscard]] const uint8_t* peekData(size_type count) const { 93| 1.36k| return Buffer::getSubView(pos, count).begin(); 94| 1.36k| } _ZN8rawspeed10ByteStream7getDataEj: 95| 929| const uint8_t* getData(size_type count) { 96| 929| const uint8_t* ret = peekData(count); 97| 929| pos += count; 98| 929| return ret; 99| 929| } _ZNK8rawspeed10ByteStream10peekBufferEj: 100| 117k| [[nodiscard]] Buffer peekBuffer(size_type size_) const { 101| 117k| return getSubView(pos, size_); 102| 117k| } _ZN8rawspeed10ByteStream9getBufferEj: 103| 11.1k| Buffer getBuffer(size_type size_) { 104| 11.1k| Buffer ret = peekBuffer(size_); 105| 11.1k| pos += size_; 106| 11.1k| return ret; 107| 11.1k| } _ZNK8rawspeed10ByteStream19peekRemainingBufferEv: 108| 412k| [[nodiscard]] Buffer peekRemainingBuffer() const { 109| 412k| return getSubView(pos, getRemainSize()); 110| 412k| } _ZNK8rawspeed10ByteStream10peekStreamEj: 111| 417k| [[nodiscard]] ByteStream peekStream(size_type size_) const { 112| 417k| return getSubStream(pos, size_); 113| 417k| } _ZNK8rawspeed10ByteStream10peekStreamEjj: 114| 351| [[nodiscard]] ByteStream peekStream(size_type nmemb, size_type size_) const { 115| 351| if (size_ && nmemb > std::numeric_limits::max() / size_) ------------------ | Branch (115:9): [True: 351, False: 0] | Branch (115:18): [True: 0, False: 351] ------------------ 116| 351| ThrowIOE("Integer overflow when calculating stream length"); ------------------ | | 37| 0|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 117| 351| return peekStream(nmemb * size_); 118| 351| } _ZN8rawspeed10ByteStream9getStreamEj: 119| 398k| ByteStream getStream(size_type size_) { 120| 398k| ByteStream ret = peekStream(size_); 121| 398k| pos += size_; 122| 398k| return ret; 123| 398k| } _ZN8rawspeed10ByteStream9getStreamEjj: 124| 6.40k| ByteStream getStream(size_type nmemb, size_type size_) { 125| 6.40k| if (size_ && nmemb > std::numeric_limits::max() / size_) ------------------ | Branch (125:9): [True: 6.40k, False: 0] | Branch (125:18): [True: 14, False: 6.39k] ------------------ 126| 6.40k| ThrowIOE("Integer overflow when calculating stream length"); ------------------ | | 37| 14|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 14| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 14| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 14| __VA_ARGS__) | | ------------------ ------------------ 127| 6.39k| return getStream(nmemb * size_); 128| 6.40k| } _ZN8rawspeed10ByteStream9skipBytesEj: 130| 5.76M| void skipBytes(size_type nbytes) { pos += check(nbytes); } _ZN8rawspeed10ByteStream9skipBytesEjj: 131| 908| void skipBytes(size_type nmemb, size_type size_) { 132| 908| pos += check(nmemb, size_); 133| 908| } _ZNK8rawspeed10ByteStream12hasPatternAtENSt3__117basic_string_viewIcNS1_11char_traitsIcEEEEj: 136| 1.08M| size_type relPos) const { 137| 1.08M| if (!isValid(pos + relPos, implicit_cast(pattern.size()))) ------------------ | Branch (137:9): [True: 27.5k, False: 1.05M] ------------------ 138| 27.5k| return false; 139| 1.05M| auto tmp = 140| 1.05M| getSubView(pos + relPos, implicit_cast(pattern.size())); 141| 1.05M| assert(tmp.getSize() == pattern.size()); ------------------ | Branch (141:5): [True: 1.05M, False: 0] ------------------ 142| 1.05M| return std::equal(tmp.begin(), tmp.end(), pattern.begin()); 143| 1.05M| } _ZNK8rawspeed10ByteStream9hasPrefixENSt3__117basic_string_viewIcNS1_11char_traitsIcEEEE: 145| 1.04M| [[nodiscard]] bool hasPrefix(std::string_view prefix) const { 146| 1.04M| return hasPatternAt(prefix, /*relPos=*/0); 147| 1.04M| } _ZN8rawspeed10ByteStream10skipPrefixENSt3__117basic_string_viewIcNS1_11char_traitsIcEEEE: 149| 877k| bool skipPrefix(std::string_view prefix) { 150| 877k| bool has_prefix = hasPrefix(prefix); 151| 877k| if (has_prefix) ------------------ | Branch (151:9): [True: 2.30k, False: 874k] ------------------ 152| 2.30k| pos += prefix.size(); 153| 877k| return has_prefix; 154| 877k| } _ZNK8rawspeed10ByteStream8peekByteEj: 183| 8.49M| [[nodiscard]] uint8_t peekByte(size_type i = 0) const { 184| 8.49M| return peek(i); 185| 8.49M| } _ZN8rawspeed10ByteStream7getByteEv: 186| 98.3k| uint8_t getByte() { return get(); } _ZNK8rawspeed10ByteStream7peekU16Ev: 188| 5.25k| [[nodiscard]] uint16_t peekU16() const { return peek(); } _ZN8rawspeed10ByteStream6getU16Ev: 194| 1.34M| uint16_t getU16() { return get(); } _ZN8rawspeed10ByteStream6getU32Ev: 196| 1.18M| uint32_t getU32() { return get(); } _ZN8rawspeed10ByteStream8getFloatEv: 197| 127| float getFloat() { return get(); } _ZNK8rawspeed10ByteStream10peekStringEv: 199| 270| [[nodiscard]] std::string_view peekString() const { 200| 270| Buffer tmp = peekBuffer(getRemainSize()); 201| 270| const auto* termIter = std::find(tmp.begin(), tmp.end(), '\0'); 202| 270| if (termIter == tmp.end()) ------------------ | Branch (202:9): [True: 0, False: 270] ------------------ 203| 270| ThrowIOE("String is not null-terminated"); ------------------ | | 37| 0|#define ThrowIOE(...) ThrowExceptionHelper(rawspeed::IOException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 204| 270| std::string_view::size_type strlen = std::distance(tmp.begin(), termIter); 205| 270| return {reinterpret_cast(tmp.begin()), strlen}; 206| 270| } _ZNK8rawspeed10ByteStream4peekIhEET_j: 156| 8.58M| template [[nodiscard]] T peek(size_type i = 0) const { 157| 8.58M| return DataBuffer::get(pos, i); 158| 8.58M| } _ZN8rawspeed10ByteStream3getIhEET_v: 159| 98.3k| template T get() { 160| 98.3k| auto ret = peek(); 161| 98.3k| pos += sizeof(T); 162| 98.3k| return ret; 163| 98.3k| } _ZNK8rawspeed10ByteStream4peekItEET_j: 156| 1.85M| template [[nodiscard]] T peek(size_type i = 0) const { 157| 1.85M| return DataBuffer::get(pos, i); 158| 1.85M| } _ZNK8rawspeed10ByteStream4peekIjEET_j: 156| 1.23M| template [[nodiscard]] T peek(size_type i = 0) const { 157| 1.23M| return DataBuffer::get(pos, i); 158| 1.23M| } _ZN8rawspeed10ByteStream3getItEET_v: 159| 1.34M| template T get() { 160| 1.34M| auto ret = peek(); 161| 1.34M| pos += sizeof(T); 162| 1.34M| return ret; 163| 1.34M| } _ZNK8rawspeed10ByteStream4peekIiEET_j: 156| 145k| template [[nodiscard]] T peek(size_type i = 0) const { 157| 145k| return DataBuffer::get(pos, i); 158| 145k| } _ZN8rawspeed10ByteStream3getIjEET_v: 159| 1.18M| template T get() { 160| 1.18M| auto ret = peek(); 161| 1.18M| pos += sizeof(T); 162| 1.18M| return ret; 163| 1.18M| } _ZN8rawspeed10ByteStream3getIfEET_v: 159| 49.3k| template T get() { 160| 49.3k| auto ret = peek(); 161| 49.3k| pos += sizeof(T); 162| 49.3k| return ret; 163| 49.3k| } _ZNK8rawspeed10ByteStream4peekIfEET_j: 156| 49.6k| template [[nodiscard]] T peek(size_type i = 0) const { 157| 49.6k| return DataBuffer::get(pos, i); 158| 49.6k| } _ZN8rawspeed10ByteStreamC2Ev: 48| 277k| ByteStream() = default; _ZNK8rawspeed10ByteStream4peekIsEET_j: 156| 1.09k| template [[nodiscard]] T peek(size_type i = 0) const { 157| 1.09k| return DataBuffer::get(pos, i); 158| 1.09k| } _ZNK8rawspeed10ByteStream4peekIdEET_j: 156| 326| template [[nodiscard]] T peek(size_type i = 0) const { 157| 326| return DataBuffer::get(pos, i); 158| 326| } _ZN8rawspeed10ByteStream3getIdEET_v: 159| 82| template T get() { 160| 82| auto ret = peek(); 161| 82| pos += sizeof(T); 162| 82| return ret; 163| 82| } _ZN8rawspeed17getHostEndiannessEv: 63| 18.8M|inline Endianness getHostEndianness() { 64| 18.8M|#if defined(__BYTE_ORDER__) && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ 65| 18.8M| return Endianness::little; 66| |#elif defined(__BYTE_ORDER__) && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ 67| | return Endianness::big; 68| |#elif defined(__BYTE_ORDER__) 69| |#error "uhm, __BYTE_ORDER__ has some strange value" 70| |#else 71| | return getHostEndiannessRuntime(); 72| |#endif 73| 18.8M|} _ZN8rawspeed14getByteSwappedEh: 76| 8.44M|inline uint8_t getByteSwapped(uint8_t v) { return v; } _ZN8rawspeed14getByteSwappedEs: 77| 3|inline int16_t getByteSwapped(int16_t v) { 78| 3| return static_cast(BSWAP16(static_cast(v))); ------------------ | | 36| 3|#define BSWAP16(A) __builtin_bswap16(A) ------------------ 79| 3|} _ZN8rawspeed14getByteSwappedEt: 80| 291k|inline uint16_t getByteSwapped(uint16_t v) { return BSWAP16(v); } ------------------ | | 36| 291k|#define BSWAP16(A) __builtin_bswap16(A) ------------------ _ZN8rawspeed14getByteSwappedEi: 81| 9|inline int32_t getByteSwapped(int32_t v) { 82| 9| return static_cast(BSWAP32(static_cast(v))); ------------------ | | 37| 9|#define BSWAP32(A) __builtin_bswap32(A) ------------------ 83| 9|} _ZN8rawspeed14getByteSwappedEj: 84| 2.89M|inline uint32_t getByteSwapped(uint32_t v) { return BSWAP32(v); } ------------------ | | 37| 2.89M|#define BSWAP32(A) __builtin_bswap32(A) ------------------ _ZN8rawspeed14getByteSwappedEm: 85| 82|inline uint64_t getByteSwapped(uint64_t v) { return BSWAP64(v); } ------------------ | | 38| 82|#define BSWAP64(A) __builtin_bswap64(A) ------------------ _ZN8rawspeed14getByteSwappedEf: 89| 49.2k|inline float getByteSwapped(float f) { 90| 49.2k| auto i = std::bit_cast(f); 91| 49.2k| i = getByteSwapped(i); 92| 49.2k| return std::bit_cast(i); 93| 49.2k|} _ZN8rawspeed14getByteSwappedEd: 94| 82|inline double getByteSwapped(double d) { 95| 82| auto i = std::bit_cast(d); 96| 82| i = getByteSwapped(i); 97| 82| return std::bit_cast(i); 98| 82|} _ZN8rawspeed8getU16BEEPKv: 124| 12|inline uint16_t getU16BE(const void* data) { return getBE(data); } _ZN8rawspeed8getU32BEEPKv: 126| 2.41k|inline uint32_t getU32BE(const void* data) { return getBE(data); } _ZN8rawspeed8getU32LEEPKv: 127| 47|inline uint32_t getU32LE(const void* data) { return getLE(data); } _ZN8rawspeed5getBEItEET_PKv: 116| 12|template inline T getBE(const void* data) { 117| 12| return getByteSwapped(data, getHostEndianness() == Endianness::little); 118| 12|} _ZN8rawspeed14getByteSwappedItEET_PKvb: 100| 1.84M|template inline T getByteSwapped(const void* data, bool bswap) { 101| 1.84M| T ret; 102| | // all interesting compilers optimize this memcpy into a single move 103| | // this is the most effective way to load some bytes without running into 104| | // alignment or aliasing issues 105| 1.84M| memcpy(&ret, data, sizeof(T)); 106| 1.84M| return bswap ? getByteSwapped(ret) : ret; ------------------ | Branch (106:10): [True: 291k, False: 1.55M] ------------------ 107| 1.84M|} _ZN8rawspeed5getBEIjEET_PKv: 116| 2.41k|template inline T getBE(const void* data) { 117| 2.41k| return getByteSwapped(data, getHostEndianness() == Endianness::little); 118| 2.41k|} _ZN8rawspeed14getByteSwappedIjEET_PKvb: 100| 8.21M|template inline T getByteSwapped(const void* data, bool bswap) { 101| 8.21M| T ret; 102| | // all interesting compilers optimize this memcpy into a single move 103| | // this is the most effective way to load some bytes without running into 104| | // alignment or aliasing issues 105| 8.21M| memcpy(&ret, data, sizeof(T)); 106| 8.21M| return bswap ? getByteSwapped(ret) : ret; ------------------ | Branch (106:10): [True: 2.84M, False: 5.36M] ------------------ 107| 8.21M|} _ZN8rawspeed5getLEIjEET_PKv: 120| 47|template inline T getLE(const void* data) { 121| 47| return getByteSwapped(data, getHostEndianness() == Endianness::big); 122| 47|} _ZN8rawspeed14getByteSwappedIhEET_PKvb: 100| 8.58M|template inline T getByteSwapped(const void* data, bool bswap) { 101| 8.58M| T ret; 102| | // all interesting compilers optimize this memcpy into a single move 103| | // this is the most effective way to load some bytes without running into 104| | // alignment or aliasing issues 105| 8.58M| memcpy(&ret, data, sizeof(T)); 106| 8.58M| return bswap ? getByteSwapped(ret) : ret; ------------------ | Branch (106:10): [True: 8.44M, False: 142k] ------------------ 107| 8.58M|} _ZN8rawspeed14getByteSwappedIiEET_PKvb: 100| 145k|template inline T getByteSwapped(const void* data, bool bswap) { 101| 145k| T ret; 102| | // all interesting compilers optimize this memcpy into a single move 103| | // this is the most effective way to load some bytes without running into 104| | // alignment or aliasing issues 105| 145k| memcpy(&ret, data, sizeof(T)); 106| 145k| return bswap ? getByteSwapped(ret) : ret; ------------------ | Branch (106:10): [True: 9, False: 145k] ------------------ 107| 145k|} _ZN8rawspeed14getByteSwappedIfEET_PKvb: 100| 49.5k|template inline T getByteSwapped(const void* data, bool bswap) { 101| 49.5k| T ret; 102| | // all interesting compilers optimize this memcpy into a single move 103| | // this is the most effective way to load some bytes without running into 104| | // alignment or aliasing issues 105| 49.5k| memcpy(&ret, data, sizeof(T)); 106| 49.5k| return bswap ? getByteSwapped(ret) : ret; ------------------ | Branch (106:10): [True: 49.2k, False: 290] ------------------ 107| 49.5k|} _ZN8rawspeed14getByteSwappedIsEET_PKvb: 100| 1.09k|template inline T getByteSwapped(const void* data, bool bswap) { 101| 1.09k| T ret; 102| | // all interesting compilers optimize this memcpy into a single move 103| | // this is the most effective way to load some bytes without running into 104| | // alignment or aliasing issues 105| 1.09k| memcpy(&ret, data, sizeof(T)); 106| 1.09k| return bswap ? getByteSwapped(ret) : ret; ------------------ | Branch (106:10): [True: 3, False: 1.09k] ------------------ 107| 1.09k|} _ZN8rawspeed14getByteSwappedIdEET_PKvb: 100| 326|template inline T getByteSwapped(const void* data, bool bswap) { 101| 326| T ret; 102| | // all interesting compilers optimize this memcpy into a single move 103| | // this is the most effective way to load some bytes without running into 104| | // alignment or aliasing issues 105| 326| memcpy(&ret, data, sizeof(T)); 106| 326| return bswap ? getByteSwapped(ret) : ret; ------------------ | Branch (106:10): [True: 82, False: 244] ------------------ 107| 326|} _ZN8rawspeed9BlackAreaC2Eiib: 30| 313| : offset(offset_), size(size_), isVertical(isVertical_) {} _ZNK8rawspeed5Hints8containsERKNSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEE: 55| 1.56k| [[nodiscard]] bool contains(const std::string& key) const { 56| 1.56k| return data.contains(key); 57| 1.56k| } _ZNK8rawspeed5Hints3getERKNSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEEb: 69| 9| [[nodiscard]] bool get(const std::string& key, bool defaultValue) const { 70| 9| auto hint = data.find(key); 71| 9| if (hint == data.end()) ------------------ | Branch (71:9): [True: 9, False: 0] ------------------ 72| 9| return defaultValue; 73| 0| return "true" == hint->second; 74| 9| } _ZNK8rawspeed5Hints3getIjEET_RKNSt3__112basic_stringIcNS3_11char_traitsIcEENS3_9allocatorIcEEEES2_: 60| 81| [[nodiscard]] T get(const std::string& key, T defaultValue) const { 61| 81| if (auto hint = data.find(key); 62| 81| hint != data.end() && !hint->second.empty()) { ------------------ | Branch (62:9): [True: 0, False: 81] | Branch (62:9): [True: 0, False: 81] | Branch (62:31): [True: 0, False: 0] ------------------ 63| 0| std::istringstream iss(hint->second); 64| 0| iss >> defaultValue; 65| 0| } 66| 81| return defaultValue; 67| 81| } _ZNK8rawspeed5Hints3getIdEET_RKNSt3__112basic_stringIcNS3_11char_traitsIcEENS3_9allocatorIcEEEES2_: 60| 3.07k| [[nodiscard]] T get(const std::string& key, T defaultValue) const { 61| 3.07k| if (auto hint = data.find(key); 62| 3.07k| hint != data.end() && !hint->second.empty()) { ------------------ | Branch (62:9): [True: 0, False: 3.07k] | Branch (62:9): [True: 0, False: 3.07k] | Branch (62:31): [True: 0, False: 0] ------------------ 63| 0| std::istringstream iss(hint->second); 64| 0| iss >> defaultValue; 65| 0| } 66| 3.07k| return defaultValue; 67| 3.07k| } _ZNK8rawspeed14CameraMetaData9getCameraERKNSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEES9_S9_: 94| 4.42k| const std::string& mode) const { 95| 4.42k| auto camera = cameras.find(getId(make, model, mode)); 96| 4.42k| return camera == cameras.end() ? nullptr : camera->second.get(); ------------------ | Branch (96:10): [True: 4.42k, False: 0] ------------------ 97| 4.42k|} _ZNK8rawspeed14CameraMetaData9getCameraERKNSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEES9_: 100| 1.69k| const std::string& model) const { 101| 1.69k| auto id = getId(make, model, ""); 102| | 103| 1.69k| auto iter = find_if( 104| 1.69k| cameras.cbegin(), cameras.cend(), [&id](decltype(*cameras.cbegin())& i) { 105| 1.69k| const auto& cid = i.first; 106| 1.69k| return tie(id.make, id.model) == tie(cid.make, cid.model); 107| 1.69k| }); 108| | 109| 1.69k| if (iter == cameras.end()) ------------------ | Branch (109:7): [True: 1.69k, False: 0] ------------------ 110| 1.69k| return nullptr; 111| | 112| 0| return iter->second.get(); 113| 1.69k|} _ZNK8rawspeed14CameraMetaData9hasCameraERKNSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEES9_S9_: 117| 247| const std::string& mode) const { 118| 247| return getCamera(make, model, mode); 119| 247|} CameraMetaData.cpp:_ZN8rawspeed12_GLOBAL__N_15getIdERKNSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEES9_S9_: 81| 6.11k| const std::string& mode) { 82| 6.11k| CameraId id; 83| 6.11k| id.make = trimSpaces(make); 84| 6.11k| id.model = trimSpaces(model); 85| 6.11k| id.mode = trimSpaces(mode); 86| | 87| 6.11k| return id; 88| 6.11k|} _ZN8rawspeed14CameraMetaDataC2Ev: 49| 2| CameraMetaData() = default; _ZN8rawspeed16ColorFilterArray7setSizeERKNS_8iPoint2DE: 50| 710|void ColorFilterArray::setSize(const iPoint2D& _size) { 51| 710| if (_size == iPoint2D(0, 0)) ------------------ | Branch (51:7): [True: 0, False: 710] ------------------ 52| 0| return; 53| | 54| 710| assert(_size.hasPositiveArea() && "CFA must have positive size."); ------------------ | Branch (54:3): [True: 710, False: 0] | Branch (54:3): [True: 710, Folded] | Branch (54:3): [True: 710, False: 0] ------------------ 55| | 56| 710| size = _size; 57| | 58| 710| if (size.area() > 36) { ------------------ | Branch (58:7): [True: 1, False: 709] ------------------ 59| | // Bayer, FC() supports 2x8 pattern 60| | // X-Trans is 6x6 pattern 61| | // is there anything bigger? 62| 1| ThrowRDE("if your CFA pattern is really %" PRIu64 " pixels " ------------------ | | 38| 1| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 1| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1| __VA_ARGS__) | | ------------------ ------------------ 63| 1| "in area we may as well give up now", 64| 1| size.area()); 65| 1| } 66| 709| if (size.area() <= 0) ------------------ | Branch (66:7): [True: 0, False: 709] ------------------ 67| 0| return; 68| 709| cfa.resize(implicit_cast(size.area())); 69| 709| fill(cfa.begin(), cfa.end(), CFAColor::UNKNOWN); 70| 709|} _ZNK8rawspeed16ColorFilterArray10getColorAtEii: 72| 2.72k|CFAColor ColorFilterArray::getColorAt(int x, int y) const { 73| 2.72k| if (cfa.empty()) ------------------ | Branch (73:7): [True: 3, False: 2.71k] ------------------ 74| 2.72k| ThrowRDE("No CFA size set"); ------------------ | | 38| 3| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 3| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 3| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 3| __VA_ARGS__) | | ------------------ ------------------ 75| | 76| | // calculate the positive modulo [0 .. size-1] 77| 2.71k| invariant(size.hasPositiveArea()); ------------------ | | 27| 2.71k|#define invariant(expr) assert(expr) ------------------ | Branch (77:3): [True: 2.71k, False: 4] ------------------ 78| 2.71k| x = (x % size.x + size.x) % size.x; 79| 2.71k| y = (y % size.y + size.y) % size.y; 80| | 81| 2.71k| return cfa[x + (static_cast(y) * size.x)]; 82| 2.71k|} _ZN8rawspeed16ColorFilterArray6setCFAENS_8iPoint2DEz: 84| 446|void ColorFilterArray::setCFA(iPoint2D in_size, ...) { 85| 446| if (in_size != size) { ------------------ | Branch (85:7): [True: 446, False: 0] ------------------ 86| 446| setSize(in_size); 87| 446| } 88| 446| va_list arguments; 89| 446| va_start(arguments, in_size); 90| 2.23k| for (auto i = 0UL; i < size.area(); i++) { ------------------ | Branch (90:22): [True: 1.78k, False: 446] ------------------ 91| 1.78k| cfa[i] = static_cast(va_arg(arguments, int)); 92| 1.78k| } 93| | va_end(arguments); 94| 446|} _ZN8rawspeed16ColorFilterArray10shiftRightEi: 96| 26|void ColorFilterArray::shiftRight(int n) { 97| 26| if (cfa.empty()) ------------------ | Branch (97:7): [True: 0, False: 26] ------------------ 98| 26| ThrowRDE("No CFA size set (or set to zero)"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 99| | 100| 26| writeLog(DEBUG_PRIO::EXTRA, "Shift right:%d", n); 101| 26| n %= size.x; 102| 26| if (n == 0) ------------------ | Branch (102:7): [True: 20, False: 6] ------------------ 103| 20| return; 104| | 105| 6| vector tmp(implicit_cast(size.area())); 106| 20| for (int y = 0; y < size.y; ++y) { ------------------ | Branch (106:19): [True: 14, False: 6] ------------------ 107| 66| for (int x = 0; x < size.x; ++x) { ------------------ | Branch (107:21): [True: 52, False: 14] ------------------ 108| 52| tmp[x + (static_cast(y) * size.x)] = getColorAt(x + n, y); 109| 52| } 110| 14| } 111| 6| cfa = tmp; 112| 6|} _ZN8rawspeed16ColorFilterArray9shiftDownEi: 114| 26|void ColorFilterArray::shiftDown(int n) { 115| 26| if (cfa.empty()) ------------------ | Branch (115:7): [True: 0, False: 26] ------------------ 116| 26| ThrowRDE("No CFA size set (or set to zero)"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 117| | 118| 26| writeLog(DEBUG_PRIO::EXTRA, "Shift down:%d", n); 119| 26| n %= size.y; 120| 26| if (n == 0) ------------------ | Branch (120:7): [True: 17, False: 9] ------------------ 121| 17| return; 122| | 123| 9| vector tmp(implicit_cast(size.area())); 124| 66| for (int y = 0; y < size.y; ++y) { ------------------ | Branch (124:19): [True: 57, False: 9] ------------------ 125| 147| for (int x = 0; x < size.x; ++x) { ------------------ | Branch (125:21): [True: 90, False: 57] ------------------ 126| 90| tmp[x + (static_cast(y) * size.x)] = getColorAt(x, y + n); 127| 90| } 128| 57| } 129| 9| cfa = tmp; 130| 9|} _ZNK8rawspeed16ColorFilterArray8asStringEv: 132| 36|std::string ColorFilterArray::asString() const { 133| 36| std::string dst; 134| 154| for (int y = 0; y < size.y; y++) { ------------------ | Branch (134:19): [True: 118, False: 36] ------------------ 135| 244| for (int x = 0; x < size.x; x++) { ------------------ | Branch (135:21): [True: 126, False: 118] ------------------ 136| 126| dst += colorToString(getColorAt(x, y)); 137| 126| dst += (x == size.x - 1) ? "\n" : ","; ------------------ | Branch (137:14): [True: 118, False: 8] ------------------ 138| 126| } 139| 118| } 140| 36| return dst; 141| 36|} _ZN8rawspeed16ColorFilterArray13colorToStringENS_8CFAColorE: 202| 126|std::string ColorFilterArray::colorToString(CFAColor c) { 203| 126| auto s = getColorAsString(c); 204| 126| if (!s) ------------------ | Branch (204:7): [True: 0, False: 126] ------------------ 205| 126| ThrowRDE("Unsupported CFA Color: %u", static_cast(c)); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 206| 126| return std::string(*s); 207| 126|} _ZN8rawspeed16ColorFilterArray10setColorAtENS_8iPoint2DENS_8CFAColorE: 209| 1.26k|void ColorFilterArray::setColorAt(iPoint2D pos, CFAColor c) { 210| 1.26k| if (pos.x >= size.x || pos.x < 0) ------------------ | Branch (210:7): [True: 0, False: 1.26k] | Branch (210:26): [True: 0, False: 1.26k] ------------------ 211| 1.26k| ThrowRDE("position out of CFA pattern"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 212| 1.26k| if (pos.y >= size.y || pos.y < 0) ------------------ | Branch (212:7): [True: 0, False: 1.26k] | Branch (212:26): [True: 0, False: 1.26k] ------------------ 213| 1.26k| ThrowRDE("position out of CFA pattern"); ------------------ | | 38| 0| ThrowExceptionHelper(rawspeed::RawDecoderException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 214| 1.26k| cfa[pos.x + (static_cast(pos.y) * size.x)] = c; 215| 1.26k|} _ZNK8rawspeed16ColorFilterArray14getDcrawFilterEv: 242| 37|uint32_t ColorFilterArray::getDcrawFilter() const { 243| | // dcraw magic 244| 37| if (size.x == 6 && size.y == 6) ------------------ | Branch (244:7): [True: 0, False: 37] | Branch (244:22): [True: 0, False: 0] ------------------ 245| 0| return 9; 246| | 247| 37| if (cfa.empty() || size.x > 2 || size.y > 8 || !isPowerOfTwo(size.y)) ------------------ | Branch (247:7): [True: 0, False: 37] | Branch (247:22): [True: 1, False: 36] | Branch (247:36): [True: 0, False: 36] | Branch (247:50): [True: 0, False: 36] ------------------ 248| 1| return 1; 249| | 250| | // FIXME: the idea here is that for a given CFA, there are at most 4 unique 251| | // colors in CFA, *AND* `toDcrawColor()` returns an unique `0b??` pattern 252| | // for each of these 4 *IN THE GIVEN CFA*. 253| | // We don't validate that invariant presently. 254| | 255| 36| uint32_t ret = 0; 256| 108| for (int x = 0; x < 2; x++) { ------------------ | Branch (256:19): [True: 72, False: 36] ------------------ 257| 648| for (int y = 0; y < 8; y++) { ------------------ | Branch (257:21): [True: 576, False: 72] ------------------ 258| 576| uint32_t c = toDcrawColor(getColorAt(x, y)); 259| 576| int g = (x >> 1) * 8; 260| 576| ret |= c << ((x & 1) * 2 + y * 4 + g); 261| 576| } 262| 72| } 263| | 264| 36| writeLog(DEBUG_PRIO::EXTRA, "%s", asString().c_str()); 265| 36| writeLog(DEBUG_PRIO::EXTRA, "DCRAW filter:%x", ret); 266| | 267| 36| return ret; 268| 37|} ColorFilterArray.cpp:_ZN8rawspeed12_GLOBAL__N_116getColorAsStringENS_8CFAColorE: 174| 126|Optional getColorAsString(CFAColor c) { 175| 126| switch (c) { 176| 0| using enum CFAColor; 177| 26| case RED: ------------------ | Branch (177:3): [True: 26, False: 100] ------------------ 178| 26| return "RED"; 179| 16| case GREEN: ------------------ | Branch (179:3): [True: 16, False: 110] ------------------ 180| 16| return "GREEN"; 181| 18| case BLUE: ------------------ | Branch (181:3): [True: 18, False: 108] ------------------ 182| 18| return "BLUE"; 183| 12| case CYAN: ------------------ | Branch (183:3): [True: 12, False: 114] ------------------ 184| 12| return "CYAN"; 185| 14| case MAGENTA: ------------------ | Branch (185:3): [True: 14, False: 112] ------------------ 186| 14| return "MAGENTA"; 187| 18| case YELLOW: ------------------ | Branch (187:3): [True: 18, False: 108] ------------------ 188| 18| return "YELLOW"; 189| 22| case WHITE: ------------------ | Branch (189:3): [True: 22, False: 104] ------------------ 190| 22| return "WHITE"; 191| 0| case FUJI_GREEN: ------------------ | Branch (191:3): [True: 0, False: 126] ------------------ 192| 0| return "FUJIGREEN"; 193| 0| case UNKNOWN: ------------------ | Branch (193:3): [True: 0, False: 126] ------------------ 194| 0| return "UNKNOWN"; 195| 0| default: ------------------ | Branch (195:3): [True: 0, False: 126] ------------------ 196| 0| return std::nullopt; 197| 126| } 198| 126|} ColorFilterArray.cpp:_ZN8rawspeed12_GLOBAL__N_112toDcrawColorENS_8CFAColorE: 218| 576|uint32_t toDcrawColor(CFAColor c) { 219| 576| switch (c) { ------------------ | Branch (219:11): [True: 576, False: 0] ------------------ 220| 0| using enum CFAColor; 221| 0| case FUJI_GREEN: ------------------ | Branch (221:3): [True: 0, False: 576] ------------------ 222| 128| case RED: ------------------ | Branch (222:3): [True: 128, False: 448] ------------------ 223| 128| return 0; 224| 56| case MAGENTA: ------------------ | Branch (224:3): [True: 56, False: 520] ------------------ 225| 128| case GREEN: ------------------ | Branch (225:3): [True: 72, False: 504] ------------------ 226| 128| return 1; 227| 48| case CYAN: ------------------ | Branch (227:3): [True: 48, False: 528] ------------------ 228| 160| case BLUE: ------------------ | Branch (228:3): [True: 112, False: 464] ------------------ 229| 160| return 2; 230| 72| case YELLOW: ------------------ | Branch (230:3): [True: 72, False: 504] ------------------ 231| 160| case WHITE: ------------------ | Branch (231:3): [True: 88, False: 488] ------------------ 232| 160| return 3; 233| 0| case UNKNOWN: ------------------ | Branch (233:3): [True: 0, False: 576] ------------------ 234| 0| case END: ------------------ | Branch (234:3): [True: 0, False: 576] ------------------ 235| 0| throw out_of_range(ColorFilterArray::colorToString(c)); 236| 576| } 237| 0| __builtin_unreachable(); 238| 576|} _ZNK8rawspeed16ColorFilterArray7getSizeEv: 66| 485| [[nodiscard]] iPoint2D getSize() const { return size; } _ZN8rawspeed16ColorFilterArrayC2Ev: 51| 15.4k| ColorFilterArray() = default; _ZN8rawspeed9RawParserC2ENS_6BufferE: 33| 13.7k| explicit RawParser(Buffer inputData) : mInput(inputData) {} _ZN8rawspeed10TiffParserC2ENS_6BufferE: 60| 13.7k|TiffParser::TiffParser(Buffer file) : RawParser(file) {} _ZN8rawspeed10TiffParser10getDecoderEPKNS_14CameraMetaDataE: 62| 13.7k|std::unique_ptr TiffParser::getDecoder(const CameraMetaData* meta) { 63| 13.7k| return TiffParser::makeDecoder(TiffParser::parse(nullptr, mInput), mInput); 64| 13.7k|} _ZN8rawspeed10TiffParser5parseEPNS_7TiffIFDENS_6BufferE: 66| 13.7k|TiffRootIFDOwner TiffParser::parse(TiffIFD* parent, Buffer data) { 67| 13.7k| ByteStream bs(DataBuffer(data, Endianness::unknown)); 68| 13.7k| bs.setByteOrder(getTiffByteOrder(bs, 0, "TIFF header")); 69| 13.7k| bs.skipBytes(2); 70| | 71| 13.7k| if (uint16_t magic = bs.getU16(); 72| 13.7k| magic != 42 && magic != 0x4f52 && magic != 0x5352 && ------------------ | Branch (72:7): [True: 751, False: 13.0k] | Branch (72:22): [True: 499, False: 252] | Branch (72:41): [True: 466, False: 33] ------------------ 73| 466| magic != 0x55) // ORF has 0x4f52/0x5352, RW2 0x55 - Brilliant! ------------------ | Branch (73:7): [True: 44, False: 422] ------------------ 74| 13.7k| ThrowTPE("Not a TIFF file (magic 42)"); ------------------ | | 39| 44| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 44| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 44| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 44| __VA_ARGS__) | | ------------------ ------------------ 75| | 76| 13.7k| auto root = std::make_unique( 77| 13.7k| parent, nullptr, bs, 78| 13.7k| UINT32_MAX); // tell TiffIFD constructor not to parse bs as IFD 79| | 80| 13.7k| NORangesSet ifds; 81| | 82| 31.1k| for (uint32_t IFDOffset = bs.getU32(); IFDOffset; ------------------ | Branch (82:42): [True: 19.3k, False: 11.8k] ------------------ 83| 19.3k| IFDOffset = root->getSubIFDs().back()->getNextIFD()) { 84| 19.3k| std::unique_ptr subIFD; 85| 19.3k| try { 86| 19.3k| subIFD = std::make_unique(root.get(), &ifds, bs, IFDOffset); 87| 19.3k| } catch (const TiffParserException&) { 88| | // This IFD may fail to parse, in which case exit the loop, 89| | // because the offset to the next IFD is last 4 bytes of an IFD, 90| | // and we didn't get them because the IFD failed to parse. 91| | // BUT: don't discard the IFD's that did succeed to parse! 92| 964| break; 93| 964| } 94| 19.3k| assert(subIFD.get()); ------------------ | Branch (94:5): [True: 17.4k, False: 0] ------------------ 95| 17.4k| root->add(std::move(subIFD)); 96| 17.4k| } 97| | 98| 12.8k| return root; 99| 13.7k|} _ZN8rawspeed10TiffParser11makeDecoderENSt3__110unique_ptrINS_11TiffRootIFDENS1_14default_deleteIS3_EEEENS_6BufferE: 102| 12.8k| Buffer data) { 103| 12.8k| if (!root) ------------------ | Branch (103:7): [True: 0, False: 12.8k] ------------------ 104| 12.8k| ThrowTPE("TiffIFD is null."); ------------------ | | 39| 0| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 105| | 106| 63.8k| for (const auto& decoder : Map) { ------------------ | Branch (106:28): [True: 63.8k, False: 433] ------------------ 107| 63.8k| checker_t dChecker = nullptr; 108| 63.8k| constructor_t dConstructor = nullptr; 109| | 110| 63.8k| std::tie(dChecker, dConstructor) = decoder; 111| | 112| 63.8k| assert(dChecker); ------------------ | Branch (112:5): [True: 63.8k, False: 0] ------------------ 113| 63.8k| assert(dConstructor); ------------------ | Branch (113:5): [True: 63.8k, False: 0] ------------------ 114| | 115| 63.8k| if (!dChecker(root.get(), data)) ------------------ | Branch (115:9): [True: 51.4k, False: 12.3k] ------------------ 116| 51.4k| continue; 117| | 118| 12.3k| return dConstructor(std::move(root), data); 119| 63.8k| } 120| | 121| 433| ThrowTPE("No decoder found. Sorry."); ------------------ | | 39| 433| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 433| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 433| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 433| __VA_ARGS__) | | ------------------ ------------------ 122| 12.8k|} _ZN8rawspeed10TiffParser11constructorINS_10DngDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 3.94k| Buffer data) { 127| 3.94k| return std::make_unique(std::move(root), data); 128| 3.94k|} _ZN8rawspeed10TiffParser11constructorINS_10MosDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 581| Buffer data) { 127| 581| return std::make_unique(std::move(root), data); 128| 581|} _ZN8rawspeed10TiffParser11constructorINS_10IiqDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 1.15k| Buffer data) { 127| 1.15k| return std::make_unique(std::move(root), data); 128| 1.15k|} _ZN8rawspeed10TiffParser11constructorINS_10Cr2DecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 1.07k| Buffer data) { 127| 1.07k| return std::make_unique(std::move(root), data); 128| 1.07k|} _ZN8rawspeed10TiffParser11constructorINS_10NefDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 863| Buffer data) { 127| 863| return std::make_unique(std::move(root), data); 128| 863|} _ZN8rawspeed10TiffParser11constructorINS_10OrfDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 329| Buffer data) { 127| 329| return std::make_unique(std::move(root), data); 128| 329|} _ZN8rawspeed10TiffParser11constructorINS_10ArwDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 1.32k| Buffer data) { 127| 1.32k| return std::make_unique(std::move(root), data); 128| 1.32k|} _ZN8rawspeed10TiffParser11constructorINS_10PefDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 306| Buffer data) { 127| 306| return std::make_unique(std::move(root), data); 128| 306|} _ZN8rawspeed10TiffParser11constructorINS_10Rw2DecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 709| Buffer data) { 127| 709| return std::make_unique(std::move(root), data); 128| 709|} _ZN8rawspeed10TiffParser11constructorINS_10SrwDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 544| Buffer data) { 127| 544| return std::make_unique(std::move(root), data); 128| 544|} _ZN8rawspeed10TiffParser11constructorINS_10MefDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 61| Buffer data) { 127| 61| return std::make_unique(std::move(root), data); 128| 61|} _ZN8rawspeed10TiffParser11constructorINS_10DcrDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 278| Buffer data) { 127| 278| return std::make_unique(std::move(root), data); 128| 278|} _ZN8rawspeed10TiffParser11constructorINS_10DcsDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 100| Buffer data) { 127| 100| return std::make_unique(std::move(root), data); 128| 100|} _ZN8rawspeed10TiffParser11constructorINS_10KdcDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 87| Buffer data) { 127| 87| return std::make_unique(std::move(root), data); 128| 87|} _ZN8rawspeed10TiffParser11constructorINS_10ErfDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 53| Buffer data) { 127| 53| return std::make_unique(std::move(root), data); 128| 53|} _ZN8rawspeed10TiffParser11constructorINS_10StiDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 26| Buffer data) { 127| 26| return std::make_unique(std::move(root), data); 128| 26|} _ZN8rawspeed10TiffParser11constructorINS_14ThreefrDecoderEEENSt3__110unique_ptrINS_10RawDecoderENS3_14default_deleteIS5_EEEEONS4_INS_11TiffRootIFDENS6_IS9_EEEENS_6BufferE: 126| 230| Buffer data) { 127| 230| return std::make_unique(std::move(root), data); 128| 230|} _ZN8rawspeed9TiffEntryC2EPNS_7TiffIFDERNS_10ByteStreamE: 58| 272k| : parent(parent_), tag(static_cast(bs.getU16())) { 59| 272k| const uint16_t numType = bs.getU16(); 60| 272k| if (numType > static_cast(TiffDataType::OFFSET)) ------------------ | Branch (60:7): [True: 1.04k, False: 271k] ------------------ 61| 272k| ThrowTPE("Error reading TIFF structure. Unknown Type 0x%x encountered.", ------------------ | | 39| 1.04k| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 1.04k| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1.04k| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1.04k| __VA_ARGS__) | | ------------------ ------------------ 62| 271k| numType); 63| 271k| type = static_cast(numType); 64| 271k| count = bs.getU32(); 65| | 66| | // check for count << datashift overflow 67| 271k| if (count > UINT32_MAX >> datashifts[numType]) ------------------ | Branch (67:7): [True: 83, False: 271k] ------------------ 68| 271k| ThrowTPE("integer overflow in size calculation."); ------------------ | | 39| 83| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 83| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 83| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 83| __VA_ARGS__) | | ------------------ ------------------ 69| | 70| 271k| uint32_t byte_size = count << datashifts[numType]; 71| 271k| uint32_t data_offset = UINT32_MAX; 72| | 73| 271k| if (byte_size <= 4) { ------------------ | Branch (73:7): [True: 88.7k, False: 183k] ------------------ 74| 88.7k| data_offset = bs.getPosition(); 75| 88.7k| data = bs.getSubStream(bs.getPosition(), byte_size); 76| 88.7k| bs.skipBytes(4); 77| 183k| } else { 78| 183k| data_offset = bs.getU32(); 79| 183k| if (type == TiffDataType::OFFSET || ------------------ | Branch (79:9): [True: 4.28k, False: 178k] | Branch (79:9): [True: 39.4k, False: 143k] ------------------ 80| 178k| isIn(tag, {TiffTag::DNGPRIVATEDATA, TiffTag::MAKERNOTE, ------------------ | Branch (80:9): [True: 35.1k, False: 143k] ------------------ 81| 178k| TiffTag::MAKERNOTE_ALT, TiffTag::FUJI_RAW_IFD, 82| 178k| TiffTag::SUBIFDS, TiffTag::EXIFIFDPOINTER})) { 83| | // preserve offset for SUB_IFD/EXIF/MAKER_NOTE data 84| | if constexpr ((false)) { 85| | // limit access to range from 0 to data_offset+byte_size 86| | data = bs.getSubStream(data_offset, byte_size); 87| 39.4k| } else { 88| | // allow access to whole file, necessary if offsets inside the maker 89| | // note point to outside data, which is forbidden due to the TIFF/DNG 90| | // spec but may happen none the less (see e.g. "old" ORF files like 91| | // EX-1, note: the tags outside of the maker note area are currently not 92| | // used anyway) 93| 39.4k| data = bs; 94| 39.4k| data.setPosition(data_offset); 95| 39.4k| (void)data.check(byte_size); 96| 39.4k| } 97| 143k| } else { 98| 143k| data = bs.getSubStream(data_offset, byte_size); 99| 143k| } 100| 183k| } 101| 271k|} _ZNK8rawspeed9TiffEntry5isIntEv: 129| 578|bool RAWSPEED_READONLY TiffEntry::isInt() const { 130| 578| using enum TiffDataType; 131| 578| return type == LONG || type == SHORT || type == BYTE; ------------------ | Branch (131:10): [True: 367, False: 211] | Branch (131:26): [True: 159, False: 52] | Branch (131:43): [True: 24, False: 28] ------------------ 132| 578|} _ZNK8rawspeed9TiffEntry7isFloatEv: 138| 6.05k|bool RAWSPEED_READONLY TiffEntry::isFloat() const { 139| 6.05k| switch (type) { 140| 0| using enum TiffDataType; 141| 263| case FLOAT: ------------------ | Branch (141:3): [True: 263, False: 5.79k] ------------------ 142| 507| case DOUBLE: ------------------ | Branch (142:3): [True: 244, False: 5.81k] ------------------ 143| 3.00k| case RATIONAL: ------------------ | Branch (143:3): [True: 2.49k, False: 3.56k] ------------------ 144| 4.88k| case SRATIONAL: ------------------ | Branch (144:3): [True: 1.88k, False: 4.17k] ------------------ 145| 5.17k| case LONG: ------------------ | Branch (145:3): [True: 284, False: 5.77k] ------------------ 146| 5.53k| case SLONG: ------------------ | Branch (146:3): [True: 365, False: 5.69k] ------------------ 147| 5.79k| case SHORT: ------------------ | Branch (147:3): [True: 255, False: 5.80k] ------------------ 148| 6.03k| case SSHORT: ------------------ | Branch (148:3): [True: 247, False: 5.80k] ------------------ 149| 6.03k| return true; 150| 19| default: ------------------ | Branch (150:3): [True: 19, False: 6.03k] ------------------ 151| 19| return false; 152| 6.05k| } 153| 6.05k|} _ZNK8rawspeed9TiffEntry10isRationalEv: 155| 3.56k|bool RAWSPEED_READONLY TiffEntry::isRational() const { 156| 3.56k| switch (type) { 157| 0| using enum TiffDataType; 158| 86| case SHORT: ------------------ | Branch (158:3): [True: 86, False: 3.47k] ------------------ 159| 775| case LONG: ------------------ | Branch (159:3): [True: 689, False: 2.87k] ------------------ 160| 3.55k| case RATIONAL: ------------------ | Branch (160:3): [True: 2.77k, False: 786] ------------------ 161| 3.55k| return true; 162| 11| default: ------------------ | Branch (162:3): [True: 11, False: 3.55k] ------------------ 163| 11| return false; 164| 3.56k| } 165| 3.56k|} _ZNK8rawspeed9TiffEntry11isSRationalEv: 167| 74.0k|bool RAWSPEED_READONLY TiffEntry::isSRational() const { 168| 74.0k| switch (type) { 169| 0| using enum TiffDataType; 170| 852| case SSHORT: ------------------ | Branch (170:3): [True: 852, False: 73.1k] ------------------ 171| 1.72k| case SLONG: ------------------ | Branch (171:3): [True: 873, False: 73.1k] ------------------ 172| 74.0k| case SRATIONAL: ------------------ | Branch (172:3): [True: 72.3k, False: 1.72k] ------------------ 173| 74.0k| return true; 174| 2| default: ------------------ | Branch (174:3): [True: 2, False: 74.0k] ------------------ 175| 2| return false; 176| 74.0k| } 177| 74.0k|} _ZNK8rawspeed9TiffEntry7getByteEj: 179| 1.43k|uint8_t TiffEntry::getByte(uint32_t index) const { 180| 1.43k| if (type != TiffDataType::BYTE && type != TiffDataType::UNDEFINED) ------------------ | Branch (180:7): [True: 1.01k, False: 413] | Branch (180:37): [True: 5, False: 1.01k] ------------------ 181| 1.43k| ThrowTPE("Wrong type %u encountered. Expected Byte on 0x%x", ------------------ | | 39| 5| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 5| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 5| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 5| __VA_ARGS__) | | ------------------ ------------------ 182| 1.42k| static_cast(type), static_cast(tag)); 183| | 184| 1.42k| return data.peekByte(index); 185| 1.43k|} _ZNK8rawspeed9TiffEntry6getU16Ej: 187| 499k|uint16_t TiffEntry::getU16(uint32_t index) const { 188| 499k| if (type != TiffDataType::SHORT && type != TiffDataType::UNDEFINED) ------------------ | Branch (188:7): [True: 5.64k, False: 493k] | Branch (188:38): [True: 16, False: 5.62k] ------------------ 189| 499k| ThrowTPE("Wrong type %u encountered. Expected Short or Undefined on 0x%x", ------------------ | | 39| 16| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 16| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 16| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 16| __VA_ARGS__) | | ------------------ ------------------ 190| 499k| static_cast(type), static_cast(tag)); 191| | 192| 499k| return data.peek(index); 193| 499k|} _ZNK8rawspeed9TiffEntry6getI16Ej: 195| 1.09k|int16_t TiffEntry::getI16(uint32_t index) const { 196| 1.09k| if (type != TiffDataType::SSHORT && type != TiffDataType::UNDEFINED) ------------------ | Branch (196:7): [True: 0, False: 1.09k] | Branch (196:39): [True: 0, False: 0] ------------------ 197| 1.09k| ThrowTPE("Wrong type %u encountered. Expected Short or Undefined on 0x%x", ------------------ | | 39| 0| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 198| 1.09k| static_cast(type), static_cast(tag)); 199| | 200| 1.09k| return data.peek(index); 201| 1.09k|} _ZNK8rawspeed9TiffEntry6getU32Ej: 203| 187k|uint32_t TiffEntry::getU32(uint32_t index) const { 204| 187k| using enum TiffDataType; 205| 187k| if (type == SHORT) ------------------ | Branch (205:7): [True: 139k, False: 47.4k] ------------------ 206| 139k| return getU16(index); 207| | 208| 47.4k| switch (type) { 209| 21.1k| case LONG: ------------------ | Branch (209:3): [True: 21.1k, False: 26.3k] ------------------ 210| 24.3k| case OFFSET: ------------------ | Branch (210:3): [True: 3.17k, False: 44.2k] ------------------ 211| 32.4k| case BYTE: ------------------ | Branch (211:3): [True: 8.09k, False: 39.3k] ------------------ 212| 38.9k| case UNDEFINED: ------------------ | Branch (212:3): [True: 6.50k, False: 40.9k] ------------------ 213| 46.5k| case RATIONAL: ------------------ | Branch (213:3): [True: 7.66k, False: 39.7k] ------------------ 214| 46.5k| break; 215| 874| default: ------------------ | Branch (215:3): [True: 874, False: 46.5k] ------------------ 216| 874| ThrowTPE("Wrong type %u encountered. Expected Long, Offset, Rational or " ------------------ | | 39| 874| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 874| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 874| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 874| __VA_ARGS__) | | ------------------ ------------------ 217| 47.4k| "Undefined on 0x%x", 218| 47.4k| static_cast(type), static_cast(tag)); 219| 47.4k| } 220| | 221| 46.5k| return data.peek(index); 222| 47.4k|} _ZNK8rawspeed9TiffEntry6getI32Ej: 224| 146k|int32_t TiffEntry::getI32(uint32_t index) const { 225| 146k| using enum TiffDataType; 226| 146k| if (type == SSHORT) ------------------ | Branch (226:7): [True: 1.09k, False: 145k] ------------------ 227| 1.09k| return getI16(index); 228| 145k| if (type != SLONG && type != SRATIONAL && type != UNDEFINED) ------------------ | Branch (228:7): [True: 144k, False: 1.23k] | Branch (228:24): [True: 0, False: 144k] | Branch (228:45): [True: 0, False: 0] ------------------ 229| 145k| ThrowTPE("Wrong type %u encountered. Expected SLong or Undefined on 0x%x", ------------------ | | 39| 0| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 230| 145k| static_cast(type), static_cast(tag)); 231| | 232| 145k| return data.peek(index); 233| 145k|} _ZNK8rawspeed9TiffEntry11getRationalEj: 235| 3.56k|NotARational TiffEntry::getRational(uint32_t index) const { 236| 3.56k| if (!isRational()) { ------------------ | Branch (236:7): [True: 11, False: 3.55k] ------------------ 237| 11| ThrowTPE("Wrong type 0x%x encountered. Expected Rational", ------------------ | | 39| 11| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 11| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 11| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 11| __VA_ARGS__) | | ------------------ ------------------ 238| 11| static_cast(type)); 239| 11| } 240| | 241| 3.55k| if (type != TiffDataType::RATIONAL) ------------------ | Branch (241:7): [True: 775, False: 2.77k] ------------------ 242| 775| return {getU32(index), 1}; 243| | 244| 2.77k| auto a = getU32(index * 2); 245| 2.77k| auto b = getU32((index * 2) + 1); 246| 2.77k| return {a, b}; 247| 3.55k|} _ZNK8rawspeed9TiffEntry12getSRationalEj: 249| 74.0k|NotARational TiffEntry::getSRational(uint32_t index) const { 250| 74.0k| if (!isSRational()) { ------------------ | Branch (250:7): [True: 2, False: 74.0k] ------------------ 251| 2| ThrowTPE("Wrong type 0x%x encountered. Expected SRational", ------------------ | | 39| 2| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 2| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2| __VA_ARGS__) | | ------------------ ------------------ 252| 2| static_cast(type)); 253| 2| } 254| | 255| 74.0k| if (type != TiffDataType::SRATIONAL) ------------------ | Branch (255:7): [True: 1.72k, False: 72.3k] ------------------ 256| 1.72k| return {getI32(index), 1}; 257| | 258| 72.3k| auto a = getI32(index * 2); 259| 72.3k| auto b = getI32((index * 2) + 1); 260| 72.3k| return {a, b}; 261| 74.0k|} _ZNK8rawspeed9TiffEntry8getFloatEj: 263| 6.05k|float TiffEntry::getFloat(uint32_t index) const { 264| 6.05k| if (!isFloat()) { ------------------ | Branch (264:7): [True: 19, False: 6.03k] ------------------ 265| 19| ThrowTPE("Wrong type 0x%x encountered. Expected Float or something " ------------------ | | 39| 19| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 19| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 19| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 19| __VA_ARGS__) | | ------------------ ------------------ 266| 19| "convertible on 0x%x", 267| 19| static_cast(type), static_cast(tag)); 268| 19| } 269| | 270| 6.03k| switch (type) { 271| 0| using enum TiffDataType; 272| 244| case DOUBLE: ------------------ | Branch (272:3): [True: 244, False: 5.79k] ------------------ 273| 244| return implicit_cast(data.peek(index)); 274| 263| case FLOAT: ------------------ | Branch (274:3): [True: 263, False: 5.77k] ------------------ 275| 263| return data.peek(index); 276| 284| case LONG: ------------------ | Branch (276:3): [True: 284, False: 5.75k] ------------------ 277| 539| case SHORT: ------------------ | Branch (277:3): [True: 255, False: 5.78k] ------------------ 278| 539| return static_cast(getU32(index)); 279| 365| case SLONG: ------------------ | Branch (279:3): [True: 365, False: 5.67k] ------------------ 280| 612| case SSHORT: ------------------ | Branch (280:3): [True: 247, False: 5.79k] ------------------ 281| 612| return static_cast(getI32(index)); 282| 2.49k| case RATIONAL: { ------------------ | Branch (282:3): [True: 2.49k, False: 3.54k] ------------------ 283| 2.49k| auto r = getRational(index); 284| 2.49k| return r.den ? static_cast(r) : 0.0F; ------------------ | Branch (284:12): [True: 2.15k, False: 340] ------------------ 285| 365| } 286| 1.88k| case SRATIONAL: { ------------------ | Branch (286:3): [True: 1.88k, False: 4.15k] ------------------ 287| 1.88k| auto r = getSRational(index); 288| 1.88k| return r.den ? static_cast(r) : 0.0F; ------------------ | Branch (288:12): [True: 1.42k, False: 457] ------------------ 289| 365| } 290| 0| default: ------------------ | Branch (290:3): [True: 0, False: 6.03k] ------------------ 291| | // unreachable 292| 0| return 0.0F; 293| 6.03k| } 294| 6.03k|} _ZNK8rawspeed9TiffEntry9getStringEv: 296| 106k|std::string TiffEntry::getString() const { 297| 106k| if (type != TiffDataType::ASCII && type != TiffDataType::BYTE) ------------------ | Branch (297:7): [True: 33.5k, False: 73.1k] | Branch (297:38): [True: 284, False: 33.3k] ------------------ 298| 106k| ThrowTPE("Wrong type 0x%x encountered. Expected Ascii or Byte", ------------------ | | 39| 284| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 284| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 284| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 284| __VA_ARGS__) | | ------------------ ------------------ 299| 106k| static_cast(type)); 300| | 301| | // *NOT* ByteStream::peekString() ! 302| 106k| Buffer tmp = data.peekBuffer(data.getRemainSize()); 303| 106k| const auto* termIter = std::find(tmp.begin(), tmp.end(), '\0'); 304| 106k| return {reinterpret_cast(tmp.begin()), 305| 106k| reinterpret_cast(termIter)}; 306| 106k|} _ZNK8rawspeed9TiffEntry14getRootIfdDataEv: 308| 658|DataBuffer TiffEntry::getRootIfdData() const { 309| 658| const TiffIFD* p = parent; 310| 658| const TiffRootIFD* r = nullptr; 311| 1.28k| while (p) { ------------------ | Branch (311:10): [True: 1.28k, False: 0] ------------------ 312| 1.28k| r = dynamic_cast(p); 313| 1.28k| if (r) ------------------ | Branch (313:9): [True: 658, False: 629] ------------------ 314| 658| break; 315| 629| p = p->parent; 316| 629| } 317| 658| if (!r) ------------------ | Branch (317:7): [True: 0, False: 658] ------------------ 318| 658| ThrowTPE("Internal error in TiffIFD data structure."); ------------------ | | 39| 0| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 0| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 0| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 0| __VA_ARGS__) | | ------------------ ------------------ 319| | 320| 658| assert(r != nullptr); ------------------ | Branch (320:3): [True: 658, False: 0] ------------------ 321| 658| return r->rootBuffer; 322| 658|} _ZNK8rawspeed9TiffEntry11getU16ArrayEj: 107| 167| [[nodiscard]] std::vector getU16Array(uint32_t count_) const { 108| 167| return getArray(count_); 109| 167| } _ZNK8rawspeed9TiffEntry11getU32ArrayEj: 111| 443| [[nodiscard]] std::vector getU32Array(uint32_t count_) const { 112| 443| return getArray(count_); 113| 443| } _ZNK8rawspeed9TiffEntry16getRationalArrayEj: 120| 543| getRationalArray(uint32_t count_) const { 121| 543| return getArray, &TiffEntry::getRational>(count_); 122| 543| } _ZNK8rawspeed9TiffEntry17getSRationalArrayEj: 125| 26| getSRationalArray(uint32_t count_) const { 126| 26| return getArray, &TiffEntry::getSRational>(count_); 127| 26| } _ZNK8rawspeed9TiffEntry7getDataEv: 129| 27.3k| [[nodiscard]] ByteStream getData() const { return data; } _ZNK8rawspeed9TiffEntry8getArrayItTnMS0_KFT_jEXadL_ZNKS0_6getU16EjEEEENSt3__16vectorIS2_NS5_9allocatorIS2_EEEEj: 71| 167| [[nodiscard]] std::vector getArray(uint32_t count_) const { 72| 167| std::vector res(count_); 73| 343k| for (uint32_t i = 0; i < count_; ++i) ------------------ | Branch (73:26): [True: 343k, False: 167] ------------------ 74| 343k| res[i] = (this->*getter)(i); 75| 167| return res; 76| 167| } _ZNK8rawspeed9TiffEntry8getArrayIjTnMS0_KFT_jEXadL_ZNKS0_6getU32EjEEEENSt3__16vectorIS2_NS5_9allocatorIS2_EEEEj: 71| 443| [[nodiscard]] std::vector getArray(uint32_t count_) const { 72| 443| std::vector res(count_); 73| 112k| for (uint32_t i = 0; i < count_; ++i) ------------------ | Branch (73:26): [True: 111k, False: 443] ------------------ 74| 111k| res[i] = (this->*getter)(i); 75| 443| return res; 76| 443| } _ZNK8rawspeed9TiffEntry8getArrayINS_12NotARationalIjEETnMS0_KFT_jEXadL_ZNKS0_11getRationalEjEEEENSt3__16vectorIS4_NS7_9allocatorIS4_EEEEj: 71| 543| [[nodiscard]] std::vector getArray(uint32_t count_) const { 72| 543| std::vector res(count_); 73| 1.61k| for (uint32_t i = 0; i < count_; ++i) ------------------ | Branch (73:26): [True: 1.07k, False: 543] ------------------ 74| 1.07k| res[i] = (this->*getter)(i); 75| 543| return res; 76| 543| } _ZNK8rawspeed9TiffEntry8getArrayINS_12NotARationalIiEETnMS0_KFT_jEXadL_ZNKS0_12getSRationalEjEEEENSt3__16vectorIS4_NS7_9allocatorIS4_EEEEj: 71| 26| [[nodiscard]] std::vector getArray(uint32_t count_) const { 72| 26| std::vector res(count_); 73| 72.1k| for (uint32_t i = 0; i < count_; ++i) ------------------ | Branch (73:26): [True: 72.1k, False: 26] ------------------ 74| 72.1k| res[i] = (this->*getter)(i); 75| 26| return res; 76| 26| } _ZN8rawspeed9TiffEntryD2Ev: 90| 161k| virtual ~TiffEntry() = default; _ZN8rawspeed7TiffIFD13parseIFDEntryEPNS_11NORangesSetINS_6BufferEEERNS_10ByteStreamE: 53| 272k|void TiffIFD::parseIFDEntry(NORangesSet* ifds, ByteStream& bs) { 54| 272k| assert(ifds); ------------------ | Branch (54:3): [True: 272k, False: 0] ------------------ 55| | 56| 272k| TiffEntryOwner t; 57| | 58| 272k| auto origPos = bs.getPosition(); 59| | 60| 272k| try { 61| 272k| t = std::make_unique(this, bs); 62| 272k| } catch (const IOException&) { // Ignore unparsable entry 63| | // fix probably broken position due to interruption by exception 64| | // i.e. setting it to the next entry. 65| 110k| bs.setPosition(origPos + 12); 66| 110k| return; 67| 110k| } 68| | 69| 161k| try { 70| 161k| switch (t->tag) { 71| 0| using enum TiffTag; 72| 420| case DNGPRIVATEDATA: ------------------ | Branch (72:5): [True: 420, False: 161k] ------------------ 73| | // These are arbitrarily 'rebased', to preserve the offsets, but as it is 74| | // implemented right now, that could trigger UB (pointer arithmetics, 75| | // creating pointer to unowned memory, etc). And since this is not even 76| | // used anywhere right now, let's not 77| | // add(parseDngPrivateData(ifds, t.get())); 78| | // but just add them as entries. (e.g. ArwDecoder uses WB from them) 79| 420| add(std::move(t)); 80| 420| break; 81| | 82| 2.96k| case MAKERNOTE: ------------------ | Branch (82:5): [True: 2.96k, False: 158k] ------------------ 83| 21.9k| case MAKERNOTE_ALT: ------------------ | Branch (83:5): [True: 18.9k, False: 142k] ------------------ 84| 21.9k| add(parseMakerNote(ifds, t.get())); 85| 21.9k| break; 86| | 87| 1.76k| case FUJI_RAW_IFD: ------------------ | Branch (87:5): [True: 1.76k, False: 160k] ------------------ 88| 8.21k| case SUBIFDS: ------------------ | Branch (88:5): [True: 6.45k, False: 155k] ------------------ 89| 9.64k| case EXIFIFDPOINTER: ------------------ | Branch (89:5): [True: 1.43k, False: 160k] ------------------ 90| 20.2k| for (uint32_t j = 0; j < t->count; j++) ------------------ | Branch (90:28): [True: 10.5k, False: 9.64k] ------------------ 91| 10.5k| add(std::make_unique(this, ifds, bs, t->getU32(j))); 92| 9.64k| break; 93| | 94| 129k| default: ------------------ | Branch (94:5): [True: 129k, False: 31.9k] ------------------ 95| 129k| add(std::move(t)); 96| 161k| } 97| 161k| } catch (const RawspeedException&) { // Unparsable private data are added as 98| | // entries 99| 27.6k| add(std::move(t)); 100| 27.6k| } 101| 161k|} _ZN8rawspeed7TiffIFDC2EPS0_: 103| 62.1k|TiffIFD::TiffIFD(TiffIFD* parent_) : parent(parent_) { 104| 62.1k| recursivelyCheckSubIFDs(1); 105| | // If we are good (can add this IFD without violating the limits), 106| | // we are still here. However, due to the way we add parsed sub-IFD's (lazy), 107| | // we need to count this IFD right *NOW*, not when adding it at the end. 108| 62.1k| recursivelyIncrementSubIFDCount(); 109| 62.1k|} _ZN8rawspeed7TiffIFDC2EPS0_PNS_11NORangesSetINS_6BufferEEENS_10DataBufferEj: 113| 62.1k| : TiffIFD(parent_) { 114| | // see TiffParser::parse: UINT32_MAX is used to mark the "virtual" top level 115| | // TiffRootIFD in a tiff file 116| 62.1k| if (offset == UINT32_MAX) ------------------ | Branch (116:7): [True: 17.3k, False: 44.8k] ------------------ 117| 17.3k| return; 118| | 119| 62.1k| assert(ifds); ------------------ | Branch (119:3): [True: 29.3k, False: 15.5k] ------------------ 120| | 121| 29.3k| ByteStream bs(data); 122| 29.3k| bs.setPosition(offset); 123| | 124| | // Directory entries in this IFD 125| 29.3k| auto numEntries = bs.getU16(); 126| | 127| | // 2 bytes for entry count 128| | // each entry is 12 bytes 129| | // 4-byte offset to the next IFD at the end 130| 29.3k| const auto IFDFullSize = 2 + 4 + (12 * numEntries); 131| 29.3k| if (const Buffer IFDBuf(data.getSubView(offset, IFDFullSize)); 132| 29.3k| !ifds->insert(IFDBuf)) ------------------ | Branch (132:7): [True: 2.33k, False: 26.9k] ------------------ 133| 29.3k| ThrowTPE("Two IFD's overlap. Raw corrupt!"); ------------------ | | 39| 2.33k| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 2.33k| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2.33k| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2.33k| __VA_ARGS__) | | ------------------ ------------------ 134| | 135| 299k| for (uint32_t i = 0; i < numEntries; i++) ------------------ | Branch (135:24): [True: 272k, False: 26.9k] ------------------ 136| 272k| parseIFDEntry(ifds, bs); 137| | 138| 26.9k| nextIFD = bs.getU32(); 139| 26.9k|} _ZN8rawspeed7TiffIFD14parseMakerNoteEPNS_11NORangesSetINS_6BufferEEEPKNS_9TiffEntryE: 143| 21.9k| const TiffEntry* t) { 144| 21.9k| assert(ifds); ------------------ | Branch (144:3): [True: 21.9k, False: 0] ------------------ 145| | 146| | // go up the IFD tree and try to find the MAKE entry on each level. 147| | // we can not go all the way to the top first because this partial tree 148| | // is not yet added to the TiffRootIFD. 149| 21.9k| const TiffIFD* p = this; 150| 21.9k| const TiffEntry* makeEntry = nullptr; 151| 81.7k| while (p && !makeEntry) { ------------------ | Branch (151:10): [True: 62.6k, False: 19.1k] | Branch (151:15): [True: 59.8k, False: 2.77k] ------------------ 152| 59.8k| makeEntry = p->getEntryRecursive(TiffTag::MAKE); 153| 59.8k| p = p->parent; 154| 59.8k| } 155| 21.9k| std::string make = 156| 21.9k| makeEntry != nullptr ? trimSpaces(makeEntry->getString()) : ""; ------------------ | Branch (156:7): [True: 3.21k, False: 18.7k] ------------------ 157| | 158| 21.9k| ByteStream bs = t->getData(); 159| | 160| | // helper function for easy setup of ByteStream buffer for the different maker 161| | // note types 'rebase' means position 0 of new stream equals current position 162| | // 'newPosition' is the position where the IFD starts 163| | // 'byteOrderOffset' is the position where the 2 magic bytes (II/MM) may be 164| | // found 'context' is a string providing error information in case the byte 165| | // order parsing should fail 166| 21.9k| auto setup = [&bs](bool rebase, uint32_t newPosition, 167| 21.9k| uint32_t byteOrderOffset = 0, 168| 21.9k| const char* context = nullptr) { 169| 21.9k| if (rebase) 170| 21.9k| bs = bs.getSubStream(bs.getPosition(), bs.getRemainSize()); 171| 21.9k| if (context) 172| 21.9k| bs.setByteOrder(getTiffByteOrder(bs, byteOrderOffset, context)); 173| 21.9k| bs.skipBytes(newPosition); 174| 21.9k| }; 175| | 176| 21.9k| if (bs.hasPrefix(std::string_view("AOC\0", 4))) { ------------------ | Branch (176:7): [True: 559, False: 21.3k] ------------------ 177| 559| setup(false, 6, 4, "Pentax makernote"); 178| 21.3k| } else if (bs.hasPrefix("PENTAX")) { ------------------ | Branch (178:14): [True: 447, False: 20.9k] ------------------ 179| 447| setup(true, 10, 8, "Pentax makernote"); 180| 20.9k| } else if (bs.hasPrefix(std::string_view("FUJIFILM\x0c\x00\x00\x00", 12))) { ------------------ | Branch (180:14): [True: 353, False: 20.5k] ------------------ 181| 353| bs.setByteOrder(Endianness::little); 182| 353| setup(true, 12); 183| 20.5k| } else if (bs.hasPrefix(std::string_view("Nikon\x00\x02", 7))) { ------------------ | Branch (183:14): [True: 1.73k, False: 18.8k] ------------------ 184| | // this is Nikon type 3 maker note format 185| | // TODO: implement Nikon type 1 maker note format 186| | // see http://www.ozhiker.com/electronics/pjmt/jpeg_info/nikon_mn.html 187| 1.73k| bs.skipBytes(10); 188| 1.73k| setup(true, 8, 0, "Nikon makernote"); 189| 18.8k| } else if (bs.hasPrefix("OLYMPUS")) { // new Olympus ------------------ | Branch (189:14): [True: 295, False: 18.5k] ------------------ 190| 295| setup(true, 12); 191| 18.5k| } else if (bs.hasPrefix("OLYMP")) { // old Olympus ------------------ | Branch (191:14): [True: 918, False: 17.6k] ------------------ 192| 918| setup(true, 8); 193| 17.6k| } else if (bs.hasPrefix("OM SYSTEM")) { // ex Olympus ------------------ | Branch (193:14): [True: 568, False: 17.0k] ------------------ 194| 568| setup(true, 16); 195| 17.0k| } else if (bs.hasPrefix("EPSON")) { ------------------ | Branch (195:14): [True: 267, False: 16.7k] ------------------ 196| 267| setup(false, 8); 197| 16.7k| } else if (bs.hasPrefix("Apple iOS")) { ------------------ | Branch (197:14): [True: 0, False: 16.7k] ------------------ 198| 0| setup(true, 14, 12, "Apple makernote"); 199| 16.7k| } else if (bs.hasPatternAt("Exif", 6)) { ------------------ | Branch (199:14): [True: 121, False: 16.6k] ------------------ 200| | // TODO: for none of the rawsamples.ch files from Panasonic is this true, 201| | // instead their MakerNote start with "Panasonic" Panasonic has the word 202| | // Exif at byte 6, a complete Tiff header starts at byte 12 This TIFF is 0 203| | // offset based 204| 121| setup(false, 20, 12, "Panosonic makernote"); 205| 16.6k| } else if (make == "SAMSUNG") { ------------------ | Branch (205:14): [True: 284, False: 16.3k] ------------------ 206| | // Samsung has no identification in its MakerNote but starts with the IFD 207| | // right away 208| 284| setup(true, 0); 209| 16.3k| } else { 210| | // cerr << "default MakerNote from " << make << endl; // Canon, Nikon (type 211| | // 2), Sony, Minolta, Ricoh, Leica, Hasselblad, etc. 212| | 213| | // At least one MAKE has not been handled explicitly and starts its 214| | // MakerNote with an endian prefix: Kodak 215| 16.3k| if (bs.skipPrefix("II")) { ------------------ | Branch (215:9): [True: 615, False: 15.7k] ------------------ 216| 615| bs.setByteOrder(Endianness::little); 217| 15.7k| } else if (bs.skipPrefix("MM")) { ------------------ | Branch (217:16): [True: 1.41k, False: 14.3k] ------------------ 218| 1.41k| bs.setByteOrder(Endianness::big); 219| 1.41k| } 220| 16.3k| } 221| | 222| | // Attempt to parse the rest as an IFD 223| 21.9k| return std::make_unique(this, ifds, bs, bs.getPosition()); 224| 21.9k|} _ZNK8rawspeed7TiffIFD14getIFDsWithTagENS_7TiffTagE: 226| 30.2k|std::vector TiffIFD::getIFDsWithTag(TiffTag tag) const { 227| 30.2k| vector matchingIFDs; 228| 30.2k| if (entries.contains(tag)) { ------------------ | Branch (228:7): [True: 12.4k, False: 17.8k] ------------------ 229| 12.4k| matchingIFDs.push_back(this); 230| 12.4k| } 231| 30.2k| for (const auto& i : subIFDs) { ------------------ | Branch (231:22): [True: 18.5k, False: 30.2k] ------------------ 232| 18.5k| vector t = i->getIFDsWithTag(tag); 233| 18.5k| matchingIFDs.insert(matchingIFDs.end(), t.begin(), t.end()); 234| 18.5k| } 235| 30.2k| return matchingIFDs; 236| 30.2k|} _ZNK8rawspeed7TiffIFD13getIFDWithTagENS_7TiffTagEj: 238| 4.85k|const TiffIFD* TiffIFD::getIFDWithTag(TiffTag tag, uint32_t index) const { 239| 4.85k| auto ifds = getIFDsWithTag(tag); 240| 4.85k| if (index >= ifds.size()) ------------------ | Branch (240:7): [True: 116, False: 4.74k] ------------------ 241| 4.85k| ThrowTPE("failed to find %u ifd with tag 0x%04x", index + 1, ------------------ | | 39| 116| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 116| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 116| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 116| __VA_ARGS__) | | ------------------ ------------------ 242| 4.74k| static_cast(tag)); 243| 4.74k| return ifds[index]; 244| 4.85k|} _ZNK8rawspeed7TiffIFD17getEntryRecursiveENS_7TiffTagE: 246| 396k|TiffEntry* RAWSPEED_READONLY TiffIFD::getEntryRecursive(TiffTag tag) const { 247| 396k| if (auto i = entries.find(tag); i != entries.end()) { ------------------ | Branch (247:35): [True: 121k, False: 274k] ------------------ 248| 121k| return i->second.get(); 249| 121k| } 250| 274k| for (const auto& j : subIFDs) { ------------------ | Branch (250:22): [True: 185k, False: 149k] ------------------ 251| 185k| TiffEntry* entry = j->getEntryRecursive(tag); 252| 185k| if (entry) ------------------ | Branch (252:9): [True: 125k, False: 60.0k] ------------------ 253| 125k| return entry; 254| 185k| } 255| 149k| return nullptr; 256| 274k|} _ZN8rawspeed7TiffIFD31recursivelyIncrementSubIFDCountEv: 258| 46.6k|void TiffIFD::recursivelyIncrementSubIFDCount() { 259| 46.6k| TiffIFD* p = this->parent; 260| 46.6k| if (!p) ------------------ | Branch (260:7): [True: 14.0k, False: 32.5k] ------------------ 261| 14.0k| return; 262| | 263| 32.5k| p->subIFDCount++; 264| | 265| 86.7k| for (; p != nullptr; p = p->parent) ------------------ | Branch (265:10): [True: 54.1k, False: 32.5k] ------------------ 266| 54.1k| p->subIFDCountRecursive++; 267| 32.5k|} _ZNK8rawspeed7TiffIFD12checkSubIFDsEi: 269| 192k|void TiffIFD::checkSubIFDs(int headroom) const { 270| 192k| int count = headroom + subIFDCount; 271| 192k| if (!headroom) ------------------ | Branch (271:7): [True: 53.2k, False: 139k] ------------------ 272| 192k| assert(count <= TiffIFD::Limits::SubIFDCount); ------------------ | Branch (272:5): [True: 53.2k, False: 0] ------------------ 273| 139k| else if (count > TiffIFD::Limits::SubIFDCount) ------------------ | Branch (273:12): [True: 12.4k, False: 127k] ------------------ 274| 12.4k| ThrowTPE("TIFF IFD has %d SubIFDs", count); ------------------ | | 39| 12.4k| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 12.4k| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 12.4k| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 12.4k| __VA_ARGS__) | | ------------------ ------------------ 275| | 276| 180k| count = headroom + subIFDCountRecursive; 277| 180k| if (!headroom) ------------------ | Branch (277:7): [True: 53.2k, False: 127k] ------------------ 278| 180k| assert(count <= TiffIFD::Limits::RecursiveSubIFDCount); ------------------ | Branch (278:5): [True: 53.2k, False: 0] ------------------ 279| 127k| else if (count > TiffIFD::Limits::RecursiveSubIFDCount) ------------------ | Branch (279:12): [True: 1.98k, False: 125k] ------------------ 280| 1.98k| ThrowTPE("TIFF IFD file has %d SubIFDs (recursively)", count); ------------------ | | 39| 1.98k| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 1.98k| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1.98k| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1.98k| __VA_ARGS__) | | ------------------ ------------------ 281| 180k|} _ZNK8rawspeed7TiffIFD23recursivelyCheckSubIFDsEi: 283| 84.5k|void TiffIFD::recursivelyCheckSubIFDs(int headroom) const { 284| 84.5k| int depth = 0; 285| 277k| for (const TiffIFD* p = this; p != nullptr;) { ------------------ | Branch (285:33): [True: 193k, False: 83.3k] ------------------ 286| 193k| if (!headroom) ------------------ | Branch (286:9): [True: 53.2k, False: 140k] ------------------ 287| 193k| assert(depth <= TiffIFD::Limits::Depth); ------------------ | Branch (287:7): [True: 53.2k, False: 0] ------------------ 288| 140k| else if (depth > TiffIFD::Limits::Depth) ------------------ | Branch (288:14): [True: 1.12k, False: 139k] ------------------ 289| 1.12k| ThrowTPE("TiffIFD cascading overflow, found %d level IFD", depth); ------------------ | | 39| 1.12k| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 1.12k| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1.12k| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1.12k| __VA_ARGS__) | | ------------------ ------------------ 290| | 291| 192k| p->checkSubIFDs(headroom); 292| | 293| | // And step up 294| 192k| p = p->parent; 295| 192k| depth++; 296| 192k| } 297| 84.5k|} _ZN8rawspeed7TiffIFD3addENSt3__110unique_ptrIS0_NS1_14default_deleteIS0_EEEE: 299| 22.3k|void TiffIFD::add(TiffIFDOwner subIFD) { 300| 22.3k| assert(subIFD->parent == this); ------------------ | Branch (300:3): [True: 22.3k, False: 0] ------------------ 301| | 302| | // We are good, and actually can add this sub-IFD, right? 303| 22.3k| subIFD->recursivelyCheckSubIFDs(0); 304| | 305| 22.3k| subIFDs.push_back(std::move(subIFD)); 306| 22.3k|} _ZN8rawspeed7TiffIFD3addENSt3__110unique_ptrINS_9TiffEntryENS1_14default_deleteIS3_EEEE: 308| 157k|void TiffIFD::add(TiffEntryOwner entry) { 309| 157k| entry->parent = this; 310| 157k| entries[entry->tag] = std::move(entry); 311| 157k|} _ZNK8rawspeed7TiffIFD8getEntryENS_7TiffTagE: 313| 79.2k|TiffEntry* TiffIFD::getEntry(TiffTag tag) const { 314| 79.2k| auto i = entries.find(tag); 315| 79.2k| if (i == entries.end()) ------------------ | Branch (315:7): [True: 235, False: 78.9k] ------------------ 316| 79.2k| ThrowTPE("Entry 0x%x not found.", static_cast(tag)); ------------------ | | 39| 235| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 235| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 235| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 235| __VA_ARGS__) | | ------------------ ------------------ 317| 78.9k| return i->second.get(); 318| 79.2k|} _ZNK8rawspeed11TiffRootIFD5getIDEv: 325| 53.7k|TiffID TiffRootIFD::getID() const { 326| 53.7k| TiffID id; 327| 53.7k| const auto* makeE = getEntryRecursive(TiffTag::MAKE); 328| 53.7k| const auto* modelE = getEntryRecursive(TiffTag::MODEL); 329| | 330| 53.7k| if (!makeE) ------------------ | Branch (330:7): [True: 2.93k, False: 50.8k] ------------------ 331| 53.7k| ThrowTPE("Failed to find MAKE entry."); ------------------ | | 39| 2.93k| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 2.93k| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 2.93k| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 2.93k| __VA_ARGS__) | | ------------------ ------------------ 332| 50.8k| if (!modelE) ------------------ | Branch (332:7): [True: 86, False: 50.7k] ------------------ 333| 50.8k| ThrowTPE("Failed to find MODEL entry."); ------------------ | | 39| 86| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 86| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 86| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 86| __VA_ARGS__) | | ------------------ ------------------ 334| | 335| 50.7k| id.make = trimSpaces(makeE->getString()); 336| 50.7k| id.model = trimSpaces(modelE->getString()); 337| | 338| 50.7k| return id; 339| 50.8k|} TiffIFD.cpp:_ZZN8rawspeed7TiffIFD14parseMakerNoteEPNS_11NORangesSetINS_6BufferEEEPKNS_9TiffEntryEENK3$_0clEbjjPKc: 168| 5.50k| const char* context = nullptr) { 169| 5.50k| if (rebase) ------------------ | Branch (169:9): [True: 4.55k, False: 947] ------------------ 170| 4.55k| bs = bs.getSubStream(bs.getPosition(), bs.getRemainSize()); 171| 5.50k| if (context) ------------------ | Branch (171:9): [True: 2.82k, False: 2.68k] ------------------ 172| 2.82k| bs.setByteOrder(getTiffByteOrder(bs, byteOrderOffset, context)); 173| 5.50k| bs.skipBytes(newPosition); 174| 5.50k| }; _ZNK8rawspeed7TiffIFD10getNextIFDEv: 112| 17.4k| [[nodiscard]] uint32_t getNextIFD() const { return nextIFD; } _ZNK8rawspeed7TiffIFD8hasEntryENS_7TiffTagE: 119| 43.3k| [[nodiscard]] bool RAWSPEED_READONLY hasEntry(TiffTag tag) const { 120| 43.3k| return entries.contains(tag); 121| 43.3k| } _ZNK8rawspeed7TiffIFD17hasEntryRecursiveENS_7TiffTagE: 122| 29.2k| [[nodiscard]] bool hasEntryRecursive(TiffTag tag) const { 123| 29.2k| return getEntryRecursive(tag) != nullptr; 124| 29.2k| } _ZNK8rawspeed7TiffIFD10getSubIFDsEv: 126| 19.0k| [[nodiscard]] const std::vector& getSubIFDs() const { 127| 19.0k| return subIFDs; 128| 19.0k| } _ZN8rawspeed11TiffRootIFDC2EPNS_7TiffIFDEPNS_11NORangesSetINS_6BufferEEENS_10DataBufferEj: 146| 34.5k| : TiffIFD(parent_, ifds, data, offset), rootBuffer(data) {} _ZN8rawspeed16getTiffByteOrderENS_10ByteStreamEjPKc: 154| 16.9k| const char* context = "") { 155| 16.9k| if (bs.hasPatternAt("II", pos)) ------------------ | Branch (155:7): [True: 14.4k, False: 2.50k] ------------------ 156| 14.4k| return Endianness::little; 157| 2.50k| if (bs.hasPatternAt("MM", pos)) ------------------ | Branch (157:7): [True: 1.41k, False: 1.08k] ------------------ 158| 1.41k| return Endianness::big; 159| | 160| 1.08k| ThrowTPE("Failed to parse TIFF endianness information in %s.", context); ------------------ | | 39| 1.08k| ThrowExceptionHelper(rawspeed::TiffParserException, __VA_ARGS__) | | ------------------ | | | | 84| 1.08k| rawspeed::ThrowException("%s, line " STR(__LINE__) ": " fmt, \ | | | | 85| 1.08k| __PRETTY_FUNCTION__ __VA_OPT__(, ) \ | | | | 86| 1.08k| __VA_ARGS__) | | ------------------ ------------------ 161| 2.50k|} _ZN8rawspeed7TiffIFDD2Ev: 105| 46.6k| virtual ~TiffIFD() = default; _ZNKSt3__16__lessIvvEclB8ne180100ImmEEbRKT_RKT0_: 40| 500| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 bool operator()(const _Tp& __lhs, const _Up& __rhs) const { 41| 500| return __lhs < __rhs; 42| 500| } _ZNKSt3__16__lessIvvEclB8ne180100IPKcS4_EEbRKT_RKT0_: 40| 280| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 bool operator()(const _Tp& __lhs, const _Up& __rhs) const { 41| 280| return __lhs < __rhs; 42| 280| } _ZNKSt3__16__lessIvvEclB8ne180100IllEEbRKT_RKT0_: 40| 1.25k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 bool operator()(const _Tp& __lhs, const _Up& __rhs) const { 41| 1.25k| return __lhs < __rhs; 42| 1.25k| } _ZNKSt3__16__lessIvvEclB8ne180100IPcS3_EEbRKT_RKT0_: 40| 438| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 bool operator()(const _Tp& __lhs, const _Up& __rhs) const { 41| 438| return __lhs < __rhs; 42| 438| } _ZNKSt3__114__copy_trivialclB8ne180100IKccTnNS_9enable_ifIXsr38__can_lower_copy_assignment_to_memmoveIT_T0_EE5valueEiE4typeELi0EEENS_4pairIPS4_PS5_EES9_S9_SA_: 103| 21.3k| operator()(_In* __first, _In* __last, _Out* __result) const { 104| 21.3k| return std::__copy_trivial_impl(__first, __last, __result); 105| 21.3k| } _ZNSt3__16__copyB8ne180100INS_17_ClassicAlgPolicyEPKcS3_PcEENS_4pairIT0_T2_EES6_T1_S7_: 110| 21.3k|__copy(_InIter __first, _Sent __last, _OutIter __result) { 111| 21.3k| return std::__dispatch_copy_or_move<_AlgPolicy, __copy_loop<_AlgPolicy>, __copy_trivial>( 112| 21.3k| std::move(__first), std::move(__last), std::move(__result)); 113| 21.3k|} _ZNSt3__14copyB8ne180100IPKcPcEET0_T_S5_S4_: 117| 21.3k|copy(_InputIterator __first, _InputIterator __last, _OutputIterator __result) { 118| 21.3k| return std::__copy<_ClassicAlgPolicy>(__first, __last, __result).second; 119| 21.3k|} _ZNSt3__119__copy_trivial_implB8ne180100IKccEENS_4pairIPT_PT0_EES4_S4_S6_: 63| 21.3k|__copy_trivial_impl(_In* __first, _In* __last, _Out* __result) { 64| 21.3k| const size_t __n = static_cast(__last - __first); 65| | 66| 21.3k| std::__constexpr_memmove(__result, __first, __element_count(__n)); 67| | 68| 21.3k| return std::make_pair(__last, __result + __n); 69| 21.3k|} _ZNSt3__121__unwrap_and_dispatchB8ne180100INS_10__overloadINS_11__copy_loopINS_17_ClassicAlgPolicyEEENS_14__copy_trivialEEEPKcS8_PcTnNS_9enable_ifIXsr12__can_rewrapIT0_T1_T2_EE5valueEiE4typeELi0EEENS_4pairISB_SD_EESB_SC_SD_: 107| 21.3k|__unwrap_and_dispatch(_InIter __first, _Sent __last, _OutIter __out_first) { 108| 21.3k| auto __range = std::__unwrap_range(__first, std::move(__last)); 109| 21.3k| auto __result = _Algorithm()(std::move(__range.first), std::move(__range.second), std::__unwrap_iter(__out_first)); 110| 21.3k| return std::make_pair(std::__rewrap_range<_Sent>(std::move(__first), std::move(__result.first)), 111| 21.3k| std::__rewrap_iter(std::move(__out_first), std::move(__result.second))); 112| 21.3k|} _ZNSt3__123__dispatch_copy_or_moveB8ne180100INS_17_ClassicAlgPolicyENS_11__copy_loopIS1_EENS_14__copy_trivialEPKcS6_PcEENS_4pairIT2_T4_EES9_T3_SA_: 131| 21.3k|__dispatch_copy_or_move(_InIter __first, _Sent __last, _OutIter __out_first) { 132| 21.3k| using _Algorithm = __overload<_NaiveAlgorithm, _OptimizedAlgorithm>; 133| 21.3k| return std::__unwrap_and_dispatch<_Algorithm>(std::move(__first), std::move(__last), std::move(__out_first)); 134| 21.3k|} _ZNSt3__16copy_nB8ne180100IPKcmPcTnNS_9enable_ifIXsr37__has_random_access_iterator_categoryIT_EE5valueEiE4typeELi0EEET1_S5_T0_S8_: 51| 21.3k|copy_n(_InputIterator __first, _Size __orig_n, _OutputIterator __result) { 52| 21.3k| typedef typename iterator_traits<_InputIterator>::difference_type difference_type; 53| 21.3k| typedef decltype(std::__convert_to_integral(__orig_n)) _IntegralSize; 54| 21.3k| _IntegralSize __n = __orig_n; 55| 21.3k| return std::copy(__first, __first + difference_type(__n), __result); 56| 21.3k|} _ZNSt3__18__fill_nB8ne180100IPcmcEET_S2_T0_RKT1_: 26| 1.84k|__fill_n(_OutputIterator __first, _Size __n, const _Tp& __value) { 27| 453k| for (; __n > 0; ++__first, (void)--__n) ------------------ | Branch (27:10): [True: 451k, False: 1.84k] ------------------ 28| 451k| *__first = __value; 29| 1.84k| return __first; 30| 1.84k|} _ZNSt3__16fill_nB8ne180100IPcmcEET_S2_T0_RKT1_: 34| 1.84k|fill_n(_OutputIterator __first, _Size __n, const _Tp& __value) { 35| 1.84k| return std::__fill_n(__first, std::__convert_to_integral(__n), __value); 36| 1.84k|} _ZNSt3__111__find_implB8ne180100IccNS_10__identityETnNS_9enable_ifIXaaaasr13__is_identityIT1_EE5valuesr41__libcpp_is_trivially_equality_comparableIT_T0_EE5valueeqstS4_Li1EEiE4typeELi0EEEPS4_S8_S8_RKS5_RS3_: 61| 237k|__find_impl(_Tp* __first, _Tp* __last, const _Up& __value, _Proj&) { 62| 237k| if (auto __ret = std::__constexpr_memchr(__first, __value, __last - __first)) ------------------ | Branch (62:12): [True: 237k, False: 215] ------------------ 63| 237k| return __ret; 64| 215| return __last; 65| 237k|} _ZNSt3__14findB8ne180100IPccEET_S2_S2_RKT0_: 173| 237k|find(_InputIterator __first, _InputIterator __last, const _Tp& __value) { 174| 237k| __identity __proj; 175| 237k| return std::__rewrap_iter( 176| 237k| __first, std::__find_impl(std::__unwrap_iter(__first), std::__unwrap_iter(__last), __value, __proj)); 177| 237k|} _ZNSt3__13maxB8ne180100ImNS_6__lessIvvEEEERKT_S5_S5_T0_: 29| 499|max(_LIBCPP_LIFETIMEBOUND const _Tp& __a, _LIBCPP_LIFETIMEBOUND const _Tp& __b, _Compare __comp) { 30| 499| return __comp(__a, __b) ? __b : __a; ------------------ | Branch (30:10): [True: 98, False: 401] ------------------ 31| 499|} _ZNSt3__13maxB8ne180100ImEERKT_S3_S3_: 35| 499|max(_LIBCPP_LIFETIMEBOUND const _Tp& __a, _LIBCPP_LIFETIMEBOUND const _Tp& __b) { 36| 499| return std::max(__a, __b, __less<>()); 37| 499|} _ZNSt3__13maxB8ne180100IPcEERKT_S4_S4_: 35| 438|max(_LIBCPP_LIFETIMEBOUND const _Tp& __a, _LIBCPP_LIFETIMEBOUND const _Tp& __b) { 36| 438| return std::max(__a, __b, __less<>()); 37| 438|} _ZNSt3__13maxB8ne180100IPcNS_6__lessIvvEEEERKT_S6_S6_T0_: 29| 438|max(_LIBCPP_LIFETIMEBOUND const _Tp& __a, _LIBCPP_LIFETIMEBOUND const _Tp& __b, _Compare __comp) { 30| 438| return __comp(__a, __b) ? __b : __a; ------------------ | Branch (30:10): [True: 0, False: 438] ------------------ 31| 438|} _ZNSt3__13minB8ne180100ImEERKT_S3_S3_: 35| 1|min(_LIBCPP_LIFETIMEBOUND const _Tp& __a, _LIBCPP_LIFETIMEBOUND const _Tp& __b) { 36| 1| return std::min(__a, __b, __less<>()); 37| 1|} _ZNSt3__13minB8ne180100ImNS_6__lessIvvEEEERKT_S5_S5_T0_: 29| 1|min(_LIBCPP_LIFETIMEBOUND const _Tp& __a, _LIBCPP_LIFETIMEBOUND const _Tp& __b, _Compare __comp) { 30| 1| return __comp(__b, __a) ? __b : __a; ------------------ | Branch (30:10): [True: 0, False: 1] ------------------ 31| 1|} _ZNSt3__13minB8ne180100IlNS_6__lessIvvEEEERKT_S5_S5_T0_: 29| 1.25k|min(_LIBCPP_LIFETIMEBOUND const _Tp& __a, _LIBCPP_LIFETIMEBOUND const _Tp& __b, _Compare __comp) { 30| 1.25k| return __comp(__b, __a) ? __b : __a; ------------------ | Branch (30:10): [True: 1.07k, False: 178] ------------------ 31| 1.25k|} _ZNSt3__13minB8ne180100IlEERKT_S3_S3_: 35| 1.25k|min(_LIBCPP_LIFETIMEBOUND const _Tp& __a, _LIBCPP_LIFETIMEBOUND const _Tp& __b) { 36| 1.25k| return std::min(__a, __b, __less<>()); 37| 1.25k|} _ZNSt3__118__unwrap_iter_implIPKcLb1EE8__unwrapB8ne180100ES2_: 55| 42.7k| static _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR _ToAddressT __unwrap(_Iter __i) _NOEXCEPT { 56| 42.7k| return std::__to_address(__i); 57| 42.7k| } _ZNSt3__113__unwrap_iterB8ne180100IPKcNS_18__unwrap_iter_implIS2_Lb1EEETnNS_9enable_ifIXsr21is_copy_constructibleIT_EE5valueEiE4typeELi0EEEDTclsrT0_8__unwrapclsr3stdE7declvalIS6_EEEES6_: 64| 42.7k|__unwrap_iter(_Iter __i) _NOEXCEPT { 65| 42.7k| return _Impl::__unwrap(__i); 66| 42.7k|} _ZNSt3__118__unwrap_iter_implIPcLb1EE8__unwrapB8ne180100ES1_: 55| 495k| static _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR _ToAddressT __unwrap(_Iter __i) _NOEXCEPT { 56| 495k| return std::__to_address(__i); 57| 495k| } _ZNSt3__113__unwrap_iterB8ne180100IPcNS_18__unwrap_iter_implIS1_Lb1EEETnNS_9enable_ifIXsr21is_copy_constructibleIT_EE5valueEiE4typeELi0EEEDTclsrT0_8__unwrapclsr3stdE7declvalIS5_EEEES5_: 64| 495k|__unwrap_iter(_Iter __i) _NOEXCEPT { 65| 495k| return _Impl::__unwrap(__i); 66| 495k|} _ZNSt3__118__unwrap_iter_implIPKcLb1EE8__rewrapB8ne180100ES2_S2_: 51| 21.3k| static _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR _Iter __rewrap(_Iter __orig_iter, _ToAddressT __unwrapped_iter) { 52| 21.3k| return __orig_iter + (__unwrapped_iter - std::__to_address(__orig_iter)); 53| 21.3k| } _ZNSt3__113__rewrap_iterB8ne180100IPKcS2_NS_18__unwrap_iter_implIS2_Lb1EEEEET_S5_T0_: 77| 21.3k|_LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR _OrigIter __rewrap_iter(_OrigIter __orig_iter, _Iter __iter) _NOEXCEPT { 78| 21.3k| return _Impl::__rewrap(std::move(__orig_iter), std::move(__iter)); 79| 21.3k|} _ZNSt3__118__unwrap_iter_implIPcLb1EE8__rewrapB8ne180100ES1_S1_: 51| 258k| static _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR _Iter __rewrap(_Iter __orig_iter, _ToAddressT __unwrapped_iter) { 52| 258k| return __orig_iter + (__unwrapped_iter - std::__to_address(__orig_iter)); 53| 258k| } _ZNSt3__113__rewrap_iterB8ne180100IPcS1_NS_18__unwrap_iter_implIS1_Lb1EEEEET_S4_T0_: 77| 258k|_LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR _OrigIter __rewrap_iter(_OrigIter __orig_iter, _Iter __iter) _NOEXCEPT { 78| 258k| return _Impl::__rewrap(std::move(__orig_iter), std::move(__iter)); 79| 258k|} _ZNSt3__119__unwrap_range_implIPKcS2_E8__unwrapB8ne180100ES2_S2_: 64| 21.3k| _LIBCPP_HIDE_FROM_ABI static constexpr auto __unwrap(_Iter __first, _Iter __last) { 65| 21.3k| return pair{std::__unwrap_iter(std::move(__first)), std::__unwrap_iter(std::move(__last))}; 66| 21.3k| } _ZNSt3__114__unwrap_rangeB8ne180100IPKcS2_EEDaT_T0_: 75| 21.3k|_LIBCPP_HIDE_FROM_ABI constexpr auto __unwrap_range(_Iter __first, _Sent __last) { 76| 21.3k| return __unwrap_range_impl<_Iter, _Sent>::__unwrap(std::move(__first), std::move(__last)); 77| 21.3k|} _ZNSt3__119__unwrap_range_implIPKcS2_E8__rewrapB8ne180100ES2_S2_: 69| 21.3k| __rewrap(_Iter __orig_iter, decltype(std::__unwrap_iter(__orig_iter)) __iter) { 70| 21.3k| return std::__rewrap_iter(std::move(__orig_iter), std::move(__iter)); 71| 21.3k| } _ZNSt3__114__rewrap_rangeB8ne180100IPKcS2_S2_EET0_S3_T1_: 80| 21.3k|_LIBCPP_HIDE_FROM_ABI constexpr _Iter __rewrap_range(_Iter __orig_iter, _Unwrapped __iter) { 81| 21.3k| return __unwrap_range_impl<_Iter, _Sent>::__rewrap(std::move(__orig_iter), std::move(__iter)); 82| 21.3k|} _ZNSt9exceptionC2B8ne180100Ev: 74| 172k| _LIBCPP_HIDE_FROM_ABI exception() _NOEXCEPT {} _ZNSt3__119istreambuf_iteratorIcNS_11char_traitsIcEEEC2B8ne180100Ev: 67| 1.32k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR istreambuf_iterator() _NOEXCEPT : __sbuf_(nullptr) {} _ZNSt3__119istreambuf_iteratorIcNS_11char_traitsIcEEEC2B8ne180100ERNS_13basic_istreamIcS2_EE: 71| 1.32k| _LIBCPP_HIDE_FROM_ABI istreambuf_iterator(istream_type& __s) _NOEXCEPT : __sbuf_(__s.rdbuf()) {} _ZNSt3__1eqB8ne180100IcNS_11char_traitsIcEEEEbRKNS_19istreambuf_iteratorIT_T0_EES8_: 95| 254k|operator==(const istreambuf_iterator<_CharT, _Traits>& __a, const istreambuf_iterator<_CharT, _Traits>& __b) { 96| 254k| return __a.equal(__b); 97| 254k|} _ZNKSt3__119istreambuf_iteratorIcNS_11char_traitsIcEEE5equalB8ne180100ERKS3_: 82| 254k| _LIBCPP_HIDE_FROM_ABI bool equal(const istreambuf_iterator& __b) const { 83| 254k| return __test_for_eof() == __b.__test_for_eof(); 84| 254k| } _ZNKSt3__119istreambuf_iteratorIcNS_11char_traitsIcEEE14__test_for_eofB8ne180100Ev: 60| 508k| _LIBCPP_HIDE_FROM_ABI bool __test_for_eof() const { 61| 508k| if (__sbuf_ && traits_type::eq_int_type(__sbuf_->sgetc(), traits_type::eof())) ------------------ | Branch (61:9): [True: 254k, False: 254k] | Branch (61:20): [True: 122, False: 254k] ------------------ 62| 122| __sbuf_ = nullptr; 63| 508k| return __sbuf_ == nullptr; 64| 508k| } _ZNSt3__119istreambuf_iteratorIcNS_11char_traitsIcEEEppB8ne180100Ev: 76| 251k| _LIBCPP_HIDE_FROM_ABI istreambuf_iterator& operator++() { 77| 251k| __sbuf_->sbumpc(); 78| 251k| return *this; 79| 251k| } _ZNKSt3__119istreambuf_iteratorIcNS_11char_traitsIcEEEdeB8ne180100Ev: 75| 252k| _LIBCPP_HIDE_FROM_ABI char_type operator*() const { return static_cast(__sbuf_->sgetc()); } _ZNSt3__119ostreambuf_iteratorIcNS_11char_traitsIcEEEC2B8ne180100ERNS_13basic_ostreamIcS2_EE: 53| 492| _LIBCPP_HIDE_FROM_ABI ostreambuf_iterator(ostream_type& __s) _NOEXCEPT : __sbuf_(__s.rdbuf()) {} _ZNKSt3__119ostreambuf_iteratorIcNS_11char_traitsIcEEE6failedB8ne180100Ev: 63| 492| _LIBCPP_HIDE_FROM_ABI bool failed() const _NOEXCEPT { return __sbuf_ == nullptr; } _ZNSt3__16localeC2B8ne180100ENS0_13__private_tagEPNS0_5__impE: 122| 1| _LIBCPP_HIDE_FROM_ABI explicit locale(__private_tag, __imp* __loc) : __locale_(__loc) {} _ZNSt3__16locale5facetC2B8ne180100Em: 137| 31| _LIBCPP_HIDE_FROM_ABI explicit facet(size_t __refs = 0) : __shared_count(static_cast(__refs) - 1) {} _ZNSt3__110ctype_baseC2B8ne180100Ev: 460| 2| _LIBCPP_HIDE_FROM_ABI ctype_base() {} _ZNSt3__15ctypeIwEC2B8ne180100Em: 476| 1| _LIBCPP_HIDE_FROM_ABI explicit ctype(size_t __refs = 0) : locale::facet(__refs) {} _ZNKSt3__15ctypeIcE2isB8ne180100Etc: 547| 15.2k| _LIBCPP_HIDE_FROM_ABI bool is(mask __m, char_type __c) const { 548| 15.2k| return isascii(__c) ? (__tab_[static_cast(__c)] & __m) != 0 : false; ------------------ | Branch (548:12): [True: 15.2k, False: 34] ------------------ 549| 15.2k| } _ZNKSt3__15ctypeIcE5widenB8ne180100Ec: 583| 302| _LIBCPP_HIDE_FROM_ABI char_type widen(char __c) const { return do_widen(__c); } _ZNKSt3__15ctypeIcE5widenB8ne180100EPKcS3_Pc: 585| 1.15k| _LIBCPP_HIDE_FROM_ABI const char* widen(const char* __low, const char* __high, char_type* __to) const { 586| 1.15k| return do_widen(__low, __high, __to); 587| 1.15k| } _ZNSt3__112codecvt_baseC2B8ne180100Ev: 752| 6| _LIBCPP_HIDE_FROM_ABI codecvt_base() {} _ZNSt3__17codecvtIcc11__mbstate_tEC2B8ne180100Em: 770| 1| _LIBCPP_HIDE_FROM_ABI explicit codecvt(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__17codecvtIDsc11__mbstate_tEC2B8ne180100Em: 933| 1| _LIBCPP_HIDE_FROM_ABI explicit codecvt(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__17codecvtIDsDu11__mbstate_tEC2B8ne180100Em: 1014| 1| _LIBCPP_HIDE_FROM_ABI explicit codecvt(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__17codecvtIDic11__mbstate_tEC2B8ne180100Em: 1096| 1| _LIBCPP_HIDE_FROM_ABI explicit codecvt(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__17codecvtIDiDu11__mbstate_tEC2B8ne180100Em: 1177| 1| _LIBCPP_HIDE_FROM_ABI explicit codecvt(size_t __refs = 0) : locale::facet(__refs) {} _ZNKSt3__18numpunctIcE13thousands_sepB8ne180100Ev: 1439| 661| _LIBCPP_HIDE_FROM_ABI char_type thousands_sep() const { return do_thousands_sep(); } _ZNKSt3__18numpunctIcE8groupingB8ne180100Ev: 1440| 1.15k| _LIBCPP_HIDE_FROM_ABI string grouping() const { return do_grouping(); } _ZNSt3__19use_facetB8ne180100INS_5ctypeIcEEEERKT_RKNS_6localeE: 185| 2.12k|inline _LIBCPP_HIDE_FROM_ABI const _Facet& use_facet(const locale& __l) { 186| 2.12k| return static_cast(*__l.use_facet(_Facet::id)); 187| 2.12k|} _ZNSt3__19use_facetB8ne180100INS_7num_getIcNS_19istreambuf_iteratorIcNS_11char_traitsIcEEEEEEEERKT_RKNS_6localeE: 185| 661|inline _LIBCPP_HIDE_FROM_ABI const _Facet& use_facet(const locale& __l) { 186| 661| return static_cast(*__l.use_facet(_Facet::id)); 187| 661|} _ZNSt3__19use_facetB8ne180100INS_7num_putIcNS_19ostreambuf_iteratorIcNS_11char_traitsIcEEEEEEEERKT_RKNS_6localeE: 185| 492|inline _LIBCPP_HIDE_FROM_ABI const _Facet& use_facet(const locale& __l) { 186| 492| return static_cast(*__l.use_facet(_Facet::id)); 187| 492|} _ZNSt3__17collateIcEC2B8ne180100Em: 197| 1| _LIBCPP_HIDE_FROM_ABI explicit collate(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__17collateIwEC2B8ne180100Em: 197| 1| _LIBCPP_HIDE_FROM_ABI explicit collate(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__19use_facetB8ne180100INS_8numpunctIcEEEERKT_RKNS_6localeE: 185| 1.15k|inline _LIBCPP_HIDE_FROM_ABI const _Facet& use_facet(const locale& __l) { 186| 1.15k| return static_cast(*__l.use_facet(_Facet::id)); 187| 1.15k|} _ZNSt3__119__libcpp_snprintf_lEPcmP15__locale_structPKcz: 95| 492| char* __s, size_t __n, locale_t __l, const char* __format, ...) { 96| 492| va_list __va; 97| 492| va_start(__va, __format); 98| 492| __libcpp_locale_guard __current(__l); 99| 492| int __res = vsnprintf(__s, __n, __format, __va); 100| | va_end(__va); 101| 492| return __res; 102| 492|} _ZNSt3__121__libcpp_locale_guardC2B8ne180100ERP15__locale_struct: 24| 492| _LIBCPP_HIDE_FROM_ABI __libcpp_locale_guard(locale_t& __loc) : __old_loc_(uselocale(__loc)) {} _ZNSt3__121__libcpp_locale_guardD2B8ne180100Ev: 26| 492| _LIBCPP_HIDE_FROM_ABI ~__libcpp_locale_guard() { 27| 492| if (__old_loc_) ------------------ | Branch (27:9): [True: 492, False: 0] ------------------ 28| 492| uselocale(__old_loc_); 29| 492| } _ZNSt3__122__libcpp_aligned_allocB8ne180100Emm: 30| 180k|inline _LIBCPP_HIDE_FROM_ABI void* __libcpp_aligned_alloc(std::size_t __alignment, std::size_t __size) { 31| |# if defined(_LIBCPP_MSVCRT_LIKE) 32| | return ::_aligned_malloc(__size, __alignment); 33| |# elif _LIBCPP_STD_VER >= 17 && !defined(_LIBCPP_HAS_NO_C11_ALIGNED_ALLOC) 34| | // aligned_alloc() requires that __size is a multiple of __alignment, 35| | // but for C++ [new.delete.general], only states "if the value of an 36| | // alignment argument passed to any of these functions is not a valid 37| | // alignment value, the behavior is undefined". 38| | // To handle calls such as ::operator new(1, std::align_val_t(128)), we 39| | // round __size up to the next multiple of __alignment. 40| 180k| size_t __rounded_size = (__size + __alignment - 1) & ~(__alignment - 1); 41| | // Rounding up could have wrapped around to zero, so we have to add another 42| | // max() ternary to the actual call site to avoid succeeded in that case. 43| 180k| return ::aligned_alloc(__alignment, __size > __rounded_size ? __size : __rounded_size); ------------------ | Branch (43:39): [True: 0, False: 180k] ------------------ 44| |# else 45| | void* __result = nullptr; 46| | (void)::posix_memalign(&__result, __alignment, __size); 47| | // If posix_memalign fails, __result is unmodified so we still return `nullptr`. 48| | return __result; 49| |# endif 50| 180k|} _ZNSt3__121__libcpp_aligned_freeB8ne180100EPv: 52| 180k|inline _LIBCPP_HIDE_FROM_ABI void __libcpp_aligned_free(void* __ptr) { 53| |# if defined(_LIBCPP_MSVCRT_LIKE) 54| | ::_aligned_free(__ptr); 55| |# else 56| 180k| ::free(__ptr); 57| 180k|# endif 58| 180k|} _ZNSt3__117allocate_at_leastB8ne180100INS_9allocatorIcEEEENS_17allocation_resultINS_16allocator_traitsIT_E7pointerEEERS5_m: 32| 19.7k|allocate_at_least(_Alloc& __alloc, size_t __n) { 33| 19.7k| if constexpr (requires { __alloc.allocate_at_least(__n); }) { 34| 19.7k| return __alloc.allocate_at_least(__n); 35| | } else { 36| | return {__alloc.allocate(__n), __n}; 37| | } 38| 19.7k|} _ZNSt3__119__allocate_at_leastB8ne180100INS_9allocatorIcEEEEDaRT_m: 41| 19.7k|[[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr auto __allocate_at_least(_Alloc& __alloc, size_t __n) { 42| 19.7k| return std::allocate_at_least(__alloc, __n); 43| 19.7k|} _ZNSt3__119__allocate_at_leastB8ne180100INS_15__sso_allocatorIPNS_6locale5facetELm30EEEEEDaRT_m: 41| 1|[[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr auto __allocate_at_least(_Alloc& __alloc, size_t __n) { 42| 1| return std::allocate_at_least(__alloc, __n); 43| 1|} _ZNSt3__117allocate_at_leastB8ne180100INS_15__sso_allocatorIPNS_6locale5facetELm30EEEEENS_17allocation_resultINS_16allocator_traitsIT_E7pointerEEERS8_m: 32| 1|allocate_at_least(_Alloc& __alloc, size_t __n) { 33| | if constexpr (requires { __alloc.allocate_at_least(__n); }) { 34| | return __alloc.allocate_at_least(__n); 35| 1| } else { 36| 1| return {__alloc.allocate(__n), __n}; 37| 1| } 38| 1|} _ZNSt3__19allocatorIcE8allocateB8ne180100Em: 119| 19.7k| _LIBCPP_NODISCARD_AFTER_CXX17 _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 _Tp* allocate(size_t __n) { 120| 19.7k| if (__n > allocator_traits::max_size(*this)) ------------------ | Branch (120:9): [True: 0, False: 19.7k] ------------------ 121| 0| __throw_bad_array_new_length(); 122| 19.7k| if (__libcpp_is_constant_evaluated()) { ------------------ | Branch (122:9): [Folded, False: 19.7k] ------------------ 123| 0| return static_cast<_Tp*>(::operator new(__n * sizeof(_Tp))); 124| 19.7k| } else { 125| 19.7k| return static_cast<_Tp*>(std::__libcpp_allocate(__n * sizeof(_Tp), _LIBCPP_ALIGNOF(_Tp))); ------------------ | | 616| 19.7k|# define _LIBCPP_ALIGNOF(_Tp) alignof(_Tp) ------------------ 126| 19.7k| } 127| 19.7k| } _ZNSt3__19allocatorIcE17allocate_at_leastB8ne180100Em: 130| 19.7k| [[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr allocation_result<_Tp*> allocate_at_least(size_t __n) { 131| 19.7k| return {allocate(__n), __n}; 132| 19.7k| } _ZNSt3__19allocatorIcE10deallocateB8ne180100EPcm: 135| 441| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 void deallocate(_Tp* __p, size_t __n) _NOEXCEPT { 136| 441| if (__libcpp_is_constant_evaluated()) { ------------------ | Branch (136:9): [Folded, False: 441] ------------------ 137| 0| ::operator delete(__p); 138| 441| } else { 139| 441| std::__libcpp_deallocate((void*)__p, __n * sizeof(_Tp), _LIBCPP_ALIGNOF(_Tp)); ------------------ | | 616| 441|# define _LIBCPP_ALIGNOF(_Tp) alignof(_Tp) ------------------ 140| 441| } 141| 441| } _ZNSt3__19allocatorIcEC2B8ne180100Ev: 114| 19.9k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 allocator() _NOEXCEPT = default; _ZNSt3__116__non_trivial_ifILb1ENS_9allocatorIcEEEC2B8ne180100Ev: 93| 19.9k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR __non_trivial_if() _NOEXCEPT {} _ZNSt3__116allocator_traitsINS_9allocatorIcEEE8max_sizeB8ne180100IS2_vvEEmRKS2_: 326| 39.4k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 static size_type max_size(const allocator_type&) _NOEXCEPT { 327| 39.4k| return numeric_limits::max() / sizeof(value_type); 328| 39.4k| } _ZNSt3__116allocator_traitsINS_9allocatorIcEEE10deallocateB8ne180100ERS2_Pcm: 288| 441| deallocate(allocator_type& __a, pointer __p, size_type __n) _NOEXCEPT { 289| 441| __a.deallocate(__p, __n); 290| 441| } _ZNSt3__116allocator_traitsINS_9allocatorIcEEE37select_on_container_copy_constructionB8ne180100IS2_vvEES2_RKS2_: 339| 1.15k| select_on_container_copy_construction(const allocator_type& __a) { 340| 1.15k| return __a; 341| 1.15k| } _ZNSt3__116allocator_traitsINS_15__sso_allocatorIPNS_6locale5facetELm30EEEE8max_sizeB8ne180100IS5_vEEmRKS5_: 320| 1| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 static size_type max_size(const allocator_type& __a) _NOEXCEPT { 321| 1| _LIBCPP_SUPPRESS_DEPRECATED_PUSH 322| 1| return __a.max_size(); 323| 1| _LIBCPP_SUPPRESS_DEPRECATED_POP 324| 1| } _ZNSt3__116allocator_traitsINS_15__sso_allocatorIPNS_6locale5facetELm30EEEE9constructB8ne180100IS4_JEvvEEvRS5_PT_DpOT0_: 304| 60| construct(allocator_type&, _Tp* __p, _Args&&... __args) { 305| 60| std::__construct_at(__p, std::forward<_Args>(__args)...); 306| 60| } _ZNSt3__116allocator_traitsINS_15__sso_allocatorIPNS_6locale5facetELm30EEEE7destroyB8ne180100IS4_vvEEvRS5_PT_: 315| 30| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 static void destroy(allocator_type&, _Tp* __p) { 316| 30| std::__destroy_at(__p); 317| 30| } _ZNSt3__122__compressed_pair_elemINS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE5__repELi0ELb0EE5__getB8ne180100Ev: 63| 90.3k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 reference __get() _NOEXCEPT { return __value_; } _ZNSt3__117__compressed_pairINS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE5__repES5_E5firstB8ne180100Ev: 129| 90.3k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 typename _Base1::reference first() _NOEXCEPT { 130| 90.3k| return static_cast<_Base1&>(*this).__get(); 131| 90.3k| } _ZNKSt3__122__compressed_pair_elemINS_9allocatorIcEELi1ELb1EE5__getB8ne180100Ev: 93| 20.9k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR const_reference __get() const _NOEXCEPT { return *this; } _ZNKSt3__117__compressed_pairINS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE5__repES5_E6secondB8ne180100Ev: 141| 20.9k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR typename _Base2::const_reference second() const _NOEXCEPT { 142| 20.9k| return static_cast<_Base2 const&>(*this).__get(); 143| 20.9k| } _ZNSt3__122__compressed_pair_elemINS_9allocatorIcEELi1ELb1EE5__getB8ne180100Ev: 92| 20.1k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 reference __get() _NOEXCEPT { return *this; } _ZNSt3__117__compressed_pairINS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE5__repES5_E6secondB8ne180100Ev: 137| 20.1k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 typename _Base2::reference second() _NOEXCEPT { 138| 20.1k| return static_cast<_Base2&>(*this).__get(); 139| 20.1k| } _ZNKSt3__122__compressed_pair_elemINS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE5__repELi0ELb0EE5__getB8ne180100Ev: 64| 988k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR const_reference __get() const _NOEXCEPT { return __value_; } _ZNKSt3__117__compressed_pairINS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE5__repES5_E5firstB8ne180100Ev: 133| 988k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR typename _Base1::const_reference first() const _NOEXCEPT { 134| 988k| return static_cast<_Base1 const&>(*this).__get(); 135| 988k| } _ZNSt3__117__compressed_pairINS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE5__repES5_EC2B8ne180100INS_18__default_init_tagES5_EEOT_OT0_: 119| 1.15k| : _Base1(std::forward<_U1>(__t1)), _Base2(std::forward<_U2>(__t2)) {} _ZNSt3__122__compressed_pair_elemINS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE5__repELi0ELb0EEC2B8ne180100ENS_18__default_init_tagE: 49| 20.4k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR explicit __compressed_pair_elem(__default_init_tag) {} _ZNSt3__122__compressed_pair_elemINS_9allocatorIcEELi1ELb1EEC2B8ne180100IS2_vEEOT_: 83| 1.15k| : __value_type(std::forward<_Up>(__u)) {} _ZNSt3__117__compressed_pairINS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE5__repES5_EC2B8ne180100INS_18__default_init_tagESA_EEOT_OT0_: 119| 19.2k| : _Base1(std::forward<_U1>(__t1)), _Base2(std::forward<_U2>(__t2)) {} _ZNSt3__122__compressed_pair_elemINS_9allocatorIcEELi1ELb1EEC2B8ne180100ENS_18__default_init_tagE: 78| 19.9k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR explicit __compressed_pair_elem(__default_init_tag) {} _ZNSt3__117__compressed_pairINS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE5__repES5_EC2B8ne180100INS_16__value_init_tagENS_18__default_init_tagEEEOT_OT0_: 119| 663| : _Base1(std::forward<_U1>(__t1)), _Base2(std::forward<_U2>(__t2)) {} _ZNSt3__122__compressed_pair_elemINS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE5__repELi0ELb0EEC2B8ne180100ENS_16__value_init_tagE: 50| 663| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR explicit __compressed_pair_elem(__value_init_tag) : __value_() {} _ZNSt3__117__compressed_pairIPPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEEC2B8ne180100IDnNS_18__default_init_tagEEEOT_OT0_: 119| 1| : _Base1(std::forward<_U1>(__t1)), _Base2(std::forward<_U2>(__t2)) {} _ZNSt3__122__compressed_pair_elemIPPNS_6locale5facetELi0ELb0EEC2B8ne180100IDnvEEOT_: 54| 1| : __value_(std::forward<_Up>(__u)) {} _ZNSt3__122__compressed_pair_elemINS_15__sso_allocatorIPNS_6locale5facetELm30EEELi1ELb0EEC2B8ne180100ENS_18__default_init_tagE: 49| 1| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR explicit __compressed_pair_elem(__default_init_tag) {} _ZNKSt3__117__compressed_pairIPPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE6secondB8ne180100Ev: 141| 1| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR typename _Base2::const_reference second() const _NOEXCEPT { 142| 1| return static_cast<_Base2 const&>(*this).__get(); 143| 1| } _ZNKSt3__122__compressed_pair_elemINS_15__sso_allocatorIPNS_6locale5facetELm30EEELi1ELb0EE5__getB8ne180100Ev: 64| 1| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR const_reference __get() const _NOEXCEPT { return __value_; } _ZNSt3__117__compressed_pairIPPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE6secondB8ne180100Ev: 137| 91| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 typename _Base2::reference second() _NOEXCEPT { 138| 91| return static_cast<_Base2&>(*this).__get(); 139| 91| } _ZNSt3__122__compressed_pair_elemINS_15__sso_allocatorIPNS_6locale5facetELm30EEELi1ELb0EE5__getB8ne180100Ev: 63| 91| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 reference __get() _NOEXCEPT { return __value_; } _ZNSt3__117__compressed_pairIPPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE5firstB8ne180100Ev: 129| 31| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 typename _Base1::reference first() _NOEXCEPT { 130| 31| return static_cast<_Base1&>(*this).__get(); 131| 31| } _ZNSt3__122__compressed_pair_elemIPPNS_6locale5facetELi0ELb0EE5__getB8ne180100Ev: 63| 31| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 reference __get() _NOEXCEPT { return __value_; } locale.cpp:_ZNSt3__117__compressed_pairIPNS_6locale5facetENS_12_GLOBAL__N_18releaserEEC2B8ne180100IRS3_NS_16__value_init_tagEEEOT_OT0_: 119| 30| : _Base1(std::forward<_U1>(__t1)), _Base2(std::forward<_U2>(__t2)) {} _ZNSt3__122__compressed_pair_elemIPNS_6locale5facetELi0ELb0EEC2B8ne180100IRS3_vEEOT_: 54| 30| : __value_(std::forward<_Up>(__u)) {} locale.cpp:_ZNSt3__122__compressed_pair_elemINS_12_GLOBAL__N_18releaserELi1ELb1EEC2B8ne180100ENS_16__value_init_tagE: 79| 30| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR explicit __compressed_pair_elem(__value_init_tag) : __value_type() {} locale.cpp:_ZNSt3__117__compressed_pairIPNS_6locale5facetENS_12_GLOBAL__N_18releaserEE5firstB8ne180100Ev: 129| 120| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 typename _Base1::reference first() _NOEXCEPT { 130| 120| return static_cast<_Base1&>(*this).__get(); 131| 120| } _ZNSt3__122__compressed_pair_elemIPNS_6locale5facetELi0ELb0EE5__getB8ne180100Ev: 63| 120| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 reference __get() _NOEXCEPT { return __value_; } _ZNSt3__114__construct_atB8ne180100IPNS_6locale5facetEJEPS3_EEPT_S6_DpOT0_: 47| 60|_LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR _Tp* __construct_at(_Tp* __location, _Args&&... __args) { 48| 60|#if _LIBCPP_STD_VER >= 20 49| 60| return std::construct_at(__location, std::forward<_Args>(__args)...); 50| |#else 51| | return _LIBCPP_ASSERT_NON_NULL(__location != nullptr, "null pointer given to construct_at"), 52| | ::new (std::__voidify(*__location)) _Tp(std::forward<_Args>(__args)...); 53| |#endif 54| 60|} _ZNSt3__112construct_atB8ne180100IPNS_6locale5facetEJEPS3_EEPT_S6_DpOT0_: 39| 60|_LIBCPP_HIDE_FROM_ABI constexpr _Tp* construct_at(_Tp* __location, _Args&&... __args) { 40| 60| _LIBCPP_ASSERT_NON_NULL(__location != nullptr, "null pointer given to construct_at"); ------------------ | | 409| 60|# define _LIBCPP_ASSERT_NON_NULL(expression, message) _LIBCPP_ASSUME(expression) | | ------------------ | | | | 34| 60|# define _LIBCPP_ASSUME(expression) ((void)0) | | ------------------ ------------------ 41| 60| return ::new (std::__voidify(*__location)) _Tp(std::forward<_Args>(__args)...); 42| 60|} _ZNSt3__112__destroy_atB8ne180100IPNS_6locale5facetETnNS_9enable_ifIXntsr8is_arrayIT_EE5valueEiE4typeELi0EEEvPS5_: 65| 30|_LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 void __destroy_at(_Tp* __loc) { 66| 30| _LIBCPP_ASSERT_NON_NULL(__loc != nullptr, "null pointer given to destroy_at"); ------------------ | | 409| 30|# define _LIBCPP_ASSERT_NON_NULL(expression, message) _LIBCPP_ASSUME(expression) | | ------------------ | | | | 34| 30|# define _LIBCPP_ASSUME(expression) ((void)0) | | ------------------ ------------------ 67| 30| __loc->~_Tp(); 68| 30|} _ZNSt3__112__to_addressB8ne180100IKcEEPT_S3_: 180| 63.9k|_LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR _Tp* __to_address(_Tp* __p) _NOEXCEPT { 181| 63.9k| static_assert(!is_function<_Tp>::value, "_Tp is a function type"); 182| 63.9k| return __p; 183| 63.9k|} _ZNSt3__112__to_addressB8ne180100IcEEPT_S2_: 180| 777k|_LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR _Tp* __to_address(_Tp* __p) _NOEXCEPT { 181| 777k| static_assert(!is_function<_Tp>::value, "_Tp is a function type"); 182| 777k| return __p; 183| 777k|} _ZNSt3__114pointer_traitsIPcE10pointer_toB8ne180100ERc: 161| 3.59k| pointer_to(__conditional_t::value, __nat, element_type>& __r) _NOEXCEPT { 162| 3.59k| return std::addressof(__r); 163| 3.59k| } _ZNSt3__112__to_addressB8ne180100IPNS_6locale5facetEEEPT_S5_: 180| 90|_LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR _Tp* __to_address(_Tp* __p) _NOEXCEPT { 181| 90| static_assert(!is_function<_Tp>::value, "_Tp is a function type"); 182| 90| return __p; 183| 90|} _ZNSt3__114__shared_countC2B8ne180100El: 148| 31| _LIBCPP_HIDE_FROM_ABI explicit __shared_count(long __refs = 0) _NOEXCEPT : __shared_owners_(__refs) {} _ZNSt3__114__shared_count12__add_sharedB8ne180100Ev: 154| 30| _LIBCPP_HIDE_FROM_ABI void __add_shared() _NOEXCEPT { __libcpp_atomic_refcount_increment(__shared_owners_); } _ZNSt3__134__libcpp_atomic_refcount_incrementB8ne180100IlEET_RS1_: 99| 30|inline _LIBCPP_HIDE_FROM_ABI _Tp __libcpp_atomic_refcount_increment(_Tp& __t) _NOEXCEPT { 100| 30|#if defined(_LIBCPP_HAS_BUILTIN_ATOMIC_SUPPORT) && !defined(_LIBCPP_HAS_NO_THREADS) 101| 30| return __atomic_add_fetch(&__t, 1, __ATOMIC_RELAXED); 102| |#else 103| | return __t += 1; 104| |#endif 105| 30|} _ZNSt3__121__libcpp_acquire_loadB8ne180100ImEET_PKS1_: 89| 4.53k|inline _LIBCPP_HIDE_FROM_ABI _ValueType __libcpp_acquire_load(_ValueType const* __value) { 90| 4.53k|#if !defined(_LIBCPP_HAS_NO_THREADS) && defined(__ATOMIC_ACQUIRE) && \ 91| 4.53k| (__has_builtin(__atomic_load_n) || defined(_LIBCPP_COMPILER_GCC)) 92| 4.53k| return __atomic_load_n(__value, __ATOMIC_ACQUIRE); 93| |#else 94| | return *__value; 95| |#endif 96| 4.53k|} locale.cpp:_ZNSt3__110unique_ptrINS_6locale5facetENS_12_GLOBAL__N_18releaserEEC2B8ne180100ILb1EvEEPS2_: 175| 30| : __ptr_(__p, __value_init_tag()) {} locale.cpp:_ZNSt3__110unique_ptrINS_6locale5facetENS_12_GLOBAL__N_18releaserEED2B8ne180100Ev: 236| 30| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX23 ~unique_ptr() { reset(); } locale.cpp:_ZNSt3__110unique_ptrINS_6locale5facetENS_12_GLOBAL__N_18releaserEE5resetB8ne180100EPS2_: 262| 30| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX23 void reset(pointer __p = pointer()) _NOEXCEPT { 263| 30| pointer __tmp = __ptr_.first(); 264| 30| __ptr_.first() = __p; 265| 30| if (__tmp) ------------------ | Branch (265:9): [True: 0, False: 30] ------------------ 266| 0| __ptr_.second()(__tmp); 267| 30| } locale.cpp:_ZNSt3__110unique_ptrINS_6locale5facetENS_12_GLOBAL__N_18releaserEE7releaseB8ne180100Ev: 256| 30| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX23 pointer release() _NOEXCEPT { 257| 30| pointer __t = __ptr_.first(); 258| 30| __ptr_.first() = pointer(); 259| 30| return __t; 260| 30| } _ZNSt3__19__voidifyB8ne180100IPNS_6locale5facetEEEPvRT_: 23| 60|_LIBCPP_ALWAYS_INLINE _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 void* __voidify(_Tp& __from) { 24| | // Cast away cv-qualifiers to allow modifying elements of a range through const iterators. 25| 60| return const_cast(static_cast(std::addressof(__from))); 26| 60|} locale.cpp:_ZNSt3__19call_onceB8ne180100IZNS_6locale2id5__getEvE3$_0JEEEvRNS_9once_flagEOT_DpOT0_: 126| 4.53k|inline _LIBCPP_HIDE_FROM_ABI void call_once(once_flag& __flag, _Callable&& __func, _Args&&... __args) { 127| 4.53k| if (__libcpp_acquire_load(&__flag.__state_) != once_flag::_Complete) { ------------------ | Branch (127:7): [True: 30, False: 4.50k] ------------------ 128| 30| typedef tuple<_Callable&&, _Args&&...> _Gp; 129| 30| _Gp __f(std::forward<_Callable>(__func), std::forward<_Args>(__args)...); 130| 30| __call_once_param<_Gp> __p(__f); 131| 30| std::__call_once(__flag.__state_, &__p, &__call_once_proxy<_Gp>); 132| 30| } 133| 4.53k|} locale.cpp:_ZNSt3__117__call_once_paramINS_5tupleIJOZNS_6locale2id5__getEvE3$_0EEEEC2B8ne180100ERS6_: 87| 30| _LIBCPP_HIDE_FROM_ABI explicit __call_once_param(_Fp& __f) : __f_(__f) {} locale.cpp:_ZNSt3__117__call_once_proxyB8ne180100INS_5tupleIJOZNS_6locale2id5__getEvE3$_0EEEEEvPv: 116| 30|void _LIBCPP_HIDE_FROM_ABI __call_once_proxy(void* __vp) { 117| 30| __call_once_param<_Fp>* __p = static_cast<__call_once_param<_Fp>*>(__vp); 118| 30| (*__p)(); 119| 30|} locale.cpp:_ZNSt3__117__call_once_paramINS_5tupleIJOZNS_6locale2id5__getEvE3$_0EEEEclB8ne180100Ev: 89| 30| _LIBCPP_HIDE_FROM_ABI void operator()() { 90| 30| typedef typename __make_tuple_indices::value, 1>::type _Index; 91| 30| __execute(_Index()); 92| 30| } locale.cpp:_ZNSt3__117__call_once_paramINS_5tupleIJOZNS_6locale2id5__getEvE3$_0EEEE9__executeB8ne180100ITpTnmJEEEvNS_15__tuple_indicesIJXspT_EEEE: 96| 30| _LIBCPP_HIDE_FROM_ABI void __execute(__tuple_indices<_Indices...>) { 97| 30| std::__invoke(std::get<0>(std::move(__f_)), std::get<_Indices>(std::move(__f_))...); 98| 30| } _ZNSt3__111char_traitsIcE6lengthB8ne180100EPKc: 226| 19.2k| static inline _LIBCPP_HIDE_FROM_ABI size_t _LIBCPP_CONSTEXPR_SINCE_CXX17 length(const char_type* __s) _NOEXCEPT { 227| 19.2k| return std::__constexpr_strlen(__s); 228| 19.2k| } _ZNSt3__111char_traitsIcE6assignB8ne180100ERcRKc: 189| 22.5k| assign(char_type& __c1, const char_type& __c2) _NOEXCEPT { 190| 22.5k| __c1 = __c2; 191| 22.5k| } _ZNSt3__111char_traitsIcE4moveB8ne180100EPcPKcm: 238| 560| move(char_type* __s1, const char_type* __s2, size_t __n) _NOEXCEPT { 239| 560| return std::__constexpr_memmove(__s1, __s2, __element_count(__n)); 240| 560| } _ZNSt3__111char_traitsIcE4copyB8ne180100EPcPKcm: 243| 21.3k| copy(char_type* __s1, const char_type* __s2, size_t __n) _NOEXCEPT { 244| 21.3k| _LIBCPP_ASSERT_NON_OVERLAPPING_RANGES(!std::__is_pointer_in_range(__s1, __s1 + __n, __s2), ------------------ | | 410| 21.3k|# define _LIBCPP_ASSERT_NON_OVERLAPPING_RANGES(expression, message) _LIBCPP_ASSUME(expression) | | ------------------ | | | | 34| 21.3k|# define _LIBCPP_ASSUME(expression) ((void)0) | | ------------------ ------------------ 245| 21.3k| "char_traits::copy: source and destination ranges overlap"); 246| 21.3k| std::copy_n(__s2, __n, __s1); 247| 21.3k| return __s1; 248| 21.3k| } _ZNSt3__111char_traitsIcE6assignB8ne180100EPcmc: 251| 1.84k| assign(char_type* __s, size_t __n, char_type __a) _NOEXCEPT { 252| 1.84k| std::fill_n(__s, __n, __a); 253| 1.84k| return __s; 254| 1.84k| } _ZNSt3__111char_traitsIcE12to_char_typeB8ne180100Ei: 259| 438| static inline _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR char_type to_char_type(int_type __c) _NOEXCEPT { 260| 438| return char_type(__c); 261| 438| } _ZNSt3__111char_traitsIcE11to_int_typeB8ne180100Ec: 262| 759k| static inline _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR int_type to_int_type(char_type __c) _NOEXCEPT { 263| 759k| return int_type((unsigned char)__c); 264| 759k| } _ZNSt3__111char_traitsIcE11eq_int_typeB8ne180100Eii: 265| 255k| static inline _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR bool eq_int_type(int_type __c1, int_type __c2) _NOEXCEPT { 266| 255k| return __c1 == __c2; 267| 255k| } _ZNSt3__111char_traitsIcE3eofB8ne180100Ev: 268| 256k| static inline _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR int_type eof() _NOEXCEPT { return int_type(EOF); } _ZNSt3__118__constexpr_strlenB8ne180100EPKc: 38| 19.2k|inline _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 size_t __constexpr_strlen(const char* __str) { 39| | // GCC currently doesn't support __builtin_strlen for heap-allocated memory during constant evaluation. 40| | // https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70816 41| |#ifdef _LIBCPP_COMPILER_GCC 42| | if (__libcpp_is_constant_evaluated()) { 43| | size_t __i = 0; 44| | for (; __str[__i] != '\0'; ++__i) 45| | ; 46| | return __i; 47| | } 48| |#endif 49| 19.2k| return __builtin_strlen(__str); 50| 19.2k|} _ZNSt3__119__constexpr_memmoveB8ne180100IcKcTnNS_9enable_ifIXsr23__is_always_bitcastableIT0_T_EE5valueEiE4typeELi0EEEPS4_S7_PS3_NS_15__element_countE: 195| 21.9k|__constexpr_memmove(_Tp* __dest, _Up* __src, __element_count __n) { 196| 21.9k| size_t __count = static_cast(__n); 197| 21.9k| if (__libcpp_is_constant_evaluated()) { ------------------ | Branch (197:7): [Folded, False: 21.9k] ------------------ 198| 0|#ifdef _LIBCPP_COMPILER_CLANG_BASED 199| 0| if (is_same<__remove_cv_t<_Tp>, __remove_cv_t<_Up> >::value) { ------------------ | Branch (199:9): [True: 0, Folded] ------------------ 200| 0| ::__builtin_memmove(__dest, __src, __count * sizeof(_Tp)); 201| 0| return __dest; 202| 0| } 203| 0|#endif 204| 0| if (std::__is_pointer_in_range(__src, __src + __count, __dest)) { ------------------ | Branch (204:9): [True: 0, False: 0] ------------------ 205| 0| for (; __count > 0; --__count) ------------------ | Branch (205:14): [True: 0, False: 0] ------------------ 206| 0| std::__assign_trivially_copyable(__dest[__count - 1], __src[__count - 1]); 207| 0| } else { 208| 0| for (size_t __i = 0; __i != __count; ++__i) ------------------ | Branch (208:28): [True: 0, False: 0] ------------------ 209| 0| std::__assign_trivially_copyable(__dest[__i], __src[__i]); 210| 0| } 211| 21.9k| } else if (__count > 0) { ------------------ | Branch (211:14): [True: 21.9k, False: 18.4E] ------------------ 212| 21.9k| ::__builtin_memmove(__dest, __src, (__count - 1) * sizeof(_Tp) + __libcpp_datasizeof<_Tp>::value); 213| 21.9k| } 214| 21.9k| return __dest; 215| 21.9k|} _ZNSt3__118__constexpr_memchrB8ne180100IccEEPT_S2_T0_m: 116| 237k|_LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 _Tp* __constexpr_memchr(_Tp* __str, _Up __value, size_t __count) { 117| 237k| static_assert(sizeof(_Tp) == 1 && __libcpp_is_trivially_equality_comparable<_Tp, _Up>::value, 118| 237k| "Calling memchr on non-trivially equality comparable types is unsafe."); 119| | 120| 237k| if (__libcpp_is_constant_evaluated()) { ------------------ | Branch (120:7): [Folded, False: 237k] ------------------ 121| |// use __builtin_char_memchr to optimize constexpr evaluation if we can 122| 0|#if _LIBCPP_STD_VER >= 17 && __has_builtin(__builtin_char_memchr) 123| | if constexpr (is_same_v, char> && is_same_v, char>) 124| 0| return __builtin_char_memchr(__str, __value, __count); 125| 0|#endif 126| | 127| 0| for (; __count; --__count) { ------------------ | Branch (127:12): [True: 0, False: 0] ------------------ 128| 0| if (*__str == __value) ------------------ | Branch (128:11): [True: 0, False: 0] ------------------ 129| 0| return __str; 130| 0| ++__str; 131| 0| } 132| 0| return nullptr; 133| 237k| } else { 134| 237k| char __value_buffer = 0; 135| 237k| __builtin_memcpy(&__value_buffer, &__value, sizeof(char)); 136| 237k| return static_cast<_Tp*>(__builtin_memchr(__str, __value_buffer, __count)); 137| 237k| } 138| 237k|} _ZNSt3__116__libcpp_tls_setB8ne180100EjPv: 352| 63|int __libcpp_tls_set(__libcpp_tls_key __key, void* __p) { return pthread_setspecific(__key, __p); } _ZNSt3__121__libcpp_execute_onceB8ne180100EPiPFvvE: 302| 521k|int __libcpp_execute_once(__libcpp_exec_once_flag* __flag, void (*__init_routine)()) { 303| 521k| return pthread_once(__flag, __init_routine); 304| 521k|} _ZNSt3__119__libcpp_tls_createB8ne180100EPjPFvPvE: 346| 1|int __libcpp_tls_create(__libcpp_tls_key* __key, void (*__at_exit)(void*)) { 347| 1| return pthread_key_create(__key, __at_exit); 348| 1|} _ZNSt3__116__libcpp_tls_getB8ne180100Ej: 350| 521k|void* __libcpp_tls_get(__libcpp_tls_key __key) { return pthread_getspecific(__key); } _ZNSt3__119__libcpp_mutex_lockB8ne180100EP15pthread_mutex_t: 280| 66|int __libcpp_mutex_lock(__libcpp_mutex_t* __m) { return pthread_mutex_lock(__m); } _ZNSt3__121__libcpp_mutex_unlockB8ne180100EP15pthread_mutex_t: 284| 66|int __libcpp_mutex_unlock(__libcpp_mutex_t* __m) { return pthread_mutex_unlock(__m); } _ZNSt3__126__libcpp_condvar_broadcastB8ne180100EP14pthread_cond_t: 291| 30|int __libcpp_condvar_broadcast(__libcpp_condvar_t* __cv) { return pthread_cond_broadcast(__cv); } locale.cpp:_ZNSt3__18__invokeB8ne180100IZNS_6locale2id5__getEvE3$_0JEEEDTclclsr3stdE7declvalIT_EEspclsr3stdE7declvalIT0_EEEEOS4_DpOS5_: 344| 30| { return static_cast<_Fp&&>(__f)(static_cast<_Args&&>(__args)...); } _ZNSt3__121__convert_to_integralB8ne180100Em: 30| 1.84k|inline _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR unsigned long __convert_to_integral(unsigned long __val) { 31| 1.84k| return __val; 32| 1.84k|} _ZNSt3__122__make_exception_guardB8ne180100INS_6vectorIPNS_6locale5facetENS_15__sso_allocatorIS4_Lm30EEEE16__destroy_vectorEEENS_28__exception_guard_exceptionsIT_EESA_: 136| 1|_LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR __exception_guard<_Rollback> __make_exception_guard(_Rollback __rollback) { 137| 1| return __exception_guard<_Rollback>(std::move(__rollback)); 138| 1|} _ZNSt3__128__exception_guard_exceptionsINS_6vectorIPNS_6locale5facetENS_15__sso_allocatorIS4_Lm30EEEE16__destroy_vectorEEC2B8ne180100ES8_: 69| 1| : __rollback_(std::move(__rollback)), __completed_(false) {} _ZNSt3__128__exception_guard_exceptionsINS_6vectorIPNS_6locale5facetENS_15__sso_allocatorIS4_Lm30EEEE16__destroy_vectorEE10__completeB8ne180100Ev: 82| 1| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 void __complete() _NOEXCEPT { __completed_ = true; } _ZNSt3__128__exception_guard_exceptionsINS_6vectorIPNS_6locale5facetENS_15__sso_allocatorIS4_Lm30EEEE16__destroy_vectorEED2B8ne180100Ev: 84| 1| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 ~__exception_guard_exceptions() { 85| 1| if (!__completed_) ------------------ | Branch (85:9): [True: 0, False: 1] ------------------ 86| 0| __rollback_(); 87| 1| } call_once.cpp:_ZNSt3__122__make_exception_guardB8ne180100IZNS_11__call_onceERVmPvPFvS3_EE3$_0EENS_28__exception_guard_exceptionsIT_EES8_: 136| 30|_LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR __exception_guard<_Rollback> __make_exception_guard(_Rollback __rollback) { 137| 30| return __exception_guard<_Rollback>(std::move(__rollback)); 138| 30|} call_once.cpp:_ZNSt3__128__exception_guard_exceptionsIZNS_11__call_onceERVmPvPFvS3_EE3$_0EC2B8ne180100ES6_: 69| 30| : __rollback_(std::move(__rollback)), __completed_(false) {} call_once.cpp:_ZNSt3__128__exception_guard_exceptionsIZNS_11__call_onceERVmPvPFvS3_EE3$_0E10__completeB8ne180100Ev: 82| 30| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 void __complete() _NOEXCEPT { __completed_ = true; } call_once.cpp:_ZNSt3__128__exception_guard_exceptionsIZNS_11__call_onceERVmPvPFvS3_EE3$_0ED2B8ne180100Ev: 84| 30| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 ~__exception_guard_exceptions() { 85| 30| if (!__completed_) ------------------ | Branch (85:9): [True: 0, False: 30] ------------------ 86| 0| __rollback_(); 87| 30| } _ZNSt3__121__is_pointer_in_rangeB8ne180100IccTnNS_9enable_ifIXsr25__is_less_than_comparableIPKT_PKT0_EE5valueEiE4typeELi0EEEbS4_S4_S7_: 36| 280| const _Tp* __begin, const _Tp* __end, const _Up* __ptr) { 37| 280| if (__libcpp_is_constant_evaluated()) { ------------------ | Branch (37:7): [Folded, False: 280] ------------------ 38| 0| _LIBCPP_ASSERT_VALID_INPUT_RANGE(__builtin_constant_p(__begin <= __end), "__begin and __end do not form a range"); ------------------ | | 407| 0|# define _LIBCPP_ASSERT_VALID_INPUT_RANGE(expression, message) _LIBCPP_ASSUME(expression) | | ------------------ | | | | 34| 0|# define _LIBCPP_ASSUME(expression) ((void)0) | | ------------------ ------------------ 39| | 40| | // If this is not a constant during constant evaluation we know that __ptr is not part of the allocation where 41| | // [__begin, __end) is. 42| 0| if (!__builtin_constant_p(__begin <= __ptr && __ptr < __end)) ------------------ | Branch (42:9): [True: 0, False: 0] | Branch (42:31): [True: 0, False: 0] | Branch (42:51): [True: 0, False: 0] ------------------ 43| 0| return false; 44| 0| } 45| | 46| | // Checking this for unrelated pointers is technically UB, but no compiler optimizes based on it (currently). 47| 280| return !__less<>()(__ptr, __begin) && __less<>()(__ptr, __end); ------------------ | Branch (47:10): [True: 0, False: 280] | Branch (47:41): [True: 0, False: 0] ------------------ 48| 280|} _ZNSt3__112__no_destroyINS_6locale5__impEE9__emplaceB8ne180100IJjEEERS2_DpOT_: 46| 1| _LIBCPP_CONSTEXPR_SINCE_CXX14 _LIBCPP_HIDE_FROM_ABI _Tp& __emplace(_Args&&... __args) { 47| 1| new (&__obj_) _Tp(std::forward<_Args>(__args)...); 48| 1| return __obj_; 49| 1| } _ZNSt3__112__no_destroyINS_6localeEEC2B8ne180100IJNS1_13__private_tagEPNS1_5__impEEEEDpOT_: 43| 1| : __obj_(std::forward<_Args>(__args)...) {} _ZNKSt3__112__no_destroyINS_6localeEE5__getB8ne180100Ev: 52| 1| _LIBCPP_CONSTEXPR_SINCE_CXX14 _LIBCPP_HIDE_FROM_ABI _Tp const& __get() const { return __obj_; } _ZNSt3__112__no_destroyINS_6localeEEC2B8ne180100IJRKS1_EEEDpOT_: 43| 1| : __obj_(std::forward<_Args>(__args)...) {} _ZNSt3__112__no_destroyINS_6localeEE5__getB8ne180100Ev: 51| 1.44k| _LIBCPP_CONSTEXPR_SINCE_CXX14 _LIBCPP_HIDE_FROM_ABI _Tp& __get() { return __obj_; } _ZNSt3__112__no_destroyINS_6locale5__impEE5__getB8ne180100Ev: 51| 9.60k| _LIBCPP_CONSTEXPR_SINCE_CXX14 _LIBCPP_HIDE_FROM_ABI _Tp& __get() { return __obj_; } _ZNSt3__112__no_destroyINS_6localeEED2B8ne180100Ev: 37| 2| _LIBCPP_HIDE_FROM_ABI ~__no_destroy() { 38| | // nothing 39| 2| } _ZNSt3__19make_pairB8ne180100IRPKcPcEENS_4pairINS_18__unwrap_ref_decayIT_E4typeENS6_IT0_E4typeEEEOS7_OSA_: 599| 21.3k| make_pair(_T1&& __t1, _T2&& __t2) { 600| 21.3k| return pair::type, typename __unwrap_ref_decay<_T2>::type>( 601| 21.3k| std::forward<_T1>(__t1), std::forward<_T2>(__t2)); 602| 21.3k|} _ZNSt3__19make_pairB8ne180100IPKcPcEENS_4pairINS_18__unwrap_ref_decayIT_E4typeENS5_IT0_E4typeEEEOS6_OS9_: 599| 21.3k| make_pair(_T1&& __t1, _T2&& __t2) { 600| 21.3k| return pair::type, typename __unwrap_ref_decay<_T2>::type>( 601| 21.3k| std::forward<_T1>(__t1), std::forward<_T2>(__t2)); 602| 21.3k|} _ZNSt3__14pairIPKcS2_EC2B8ne180100IS2_S2_TnNS_9enable_ifIXclsr10_CheckArgsE17__enable_implicitIT_T0_EEEiE4typeELi0EEEOS6_OS7_: 207| 21.3k| : first(std::forward<_U1>(__u1)), second(std::forward<_U2>(__u2)) { 208| 21.3k| } _ZNSt3__14pairIPKcPcEC2B8ne180100IRS2_S3_TnNS_9enable_ifIXclsr10_CheckArgsE17__enable_implicitIT_T0_EEEiE4typeELi0EEEOS8_OS9_: 207| 21.3k| : first(std::forward<_U1>(__u1)), second(std::forward<_U2>(__u2)) { 208| 21.3k| } _ZNSt3__14pairIPKcPcEC2B8ne180100IS2_S3_TnNS_9enable_ifIXclsr10_CheckArgsE17__enable_implicitIT_T0_EEEiE4typeELi0EEEOS7_OS8_: 207| 21.3k| : first(std::forward<_U1>(__u1)), second(std::forward<_U2>(__u2)) { 208| 21.3k| } _ZNKSt3__18ios_base5rdbufB8ne180100Ev: 367| 3.07k| _LIBCPP_HIDE_FROM_ABI void* rdbuf() const { return __rdbuf_; } _ZNKSt3__18ios_base5flagsB8ne180100Ev: 455| 3.56k|inline _LIBCPP_HIDE_FROM_ABI ios_base::fmtflags ios_base::flags() const { return __fmtflags_; } _ZNKSt3__18ios_base5widthB8ne180100Ev: 490| 1.25k|inline _LIBCPP_HIDE_FROM_ABI streamsize ios_base::width() const { return __width_; } _ZNSt3__18ios_base5widthB8ne180100El: 492| 1.25k|inline _LIBCPP_HIDE_FROM_ABI streamsize ios_base::width(streamsize __wide) { 493| 1.25k| streamsize __r = __width_; 494| 1.25k| __width_ = __wide; 495| 1.25k| return __r; 496| 1.25k|} _ZNSt3__18ios_base8setstateB8ne180100Ej: 502| 1.08k|inline _LIBCPP_HIDE_FROM_ABI void ios_base::setstate(iostate __state) { clear(__rdstate_ | __state); } _ZNKSt3__18ios_base4goodB8ne180100Ev: 504| 4.25k|inline _LIBCPP_HIDE_FROM_ABI bool ios_base::good() const { return __rdstate_ == 0; } _ZNSt3__19basic_iosIcNS_11char_traitsIcEEED2Ev: 595| 723|basic_ios<_CharT, _Traits>::~basic_ios() {} _ZNKSt3__19basic_iosIcNS_11char_traitsIcEEE4goodB8ne180100Ev: 545| 4.25k| _LIBCPP_HIDE_FROM_ABI bool good() const { return ios_base::good(); } _ZNKSt3__19basic_iosIcNS_11char_traitsIcEEE3tieB8ne180100Ev: 605| 1.92k|inline _LIBCPP_HIDE_FROM_ABI basic_ostream<_CharT, _Traits>* basic_ios<_CharT, _Traits>::tie() const { 606| 1.92k| return __tie_; 607| 1.92k|} _ZNSt3__19basic_iosIcNS_11char_traitsIcEEE8setstateB8ne180100Ej: 544| 1.08k| _LIBCPP_HIDE_FROM_ABI void setstate(iostate __state) { ios_base::setstate(__state); } _ZNKSt3__19basic_iosIcNS_11char_traitsIcEEE5rdbufB8ne180100Ev: 618| 3.07k|inline _LIBCPP_HIDE_FROM_ABI basic_streambuf<_CharT, _Traits>* basic_ios<_CharT, _Traits>::rdbuf() const { 619| 3.07k| return static_cast*>(ios_base::rdbuf()); 620| 3.07k|} _ZNKSt3__19basic_iosIcNS_11char_traitsIcEEE5widenB8ne180100Ec: 645| 302|inline _LIBCPP_HIDE_FROM_ABI _CharT basic_ios<_CharT, _Traits>::widen(char __c) const { 646| 302| return std::use_facet >(getloc()).widen(__c); 647| 302|} _ZNKSt3__19basic_iosIcNS_11char_traitsIcEEE4fillB8ne180100Ev: 650| 492|inline _LIBCPP_HIDE_FROM_ABI _CharT basic_ios<_CharT, _Traits>::fill() const { 651| 492| if (traits_type::eq_int_type(traits_type::eof(), __fill_)) ------------------ | Branch (651:7): [True: 302, False: 190] ------------------ 652| 302| __fill_ = widen(' '); 653| 492| return __fill_; 654| 492|} _ZNSt3__113basic_istreamIcNS_11char_traitsIcEEED2Ev: 346| 270|basic_istream<_CharT, _Traits>::~basic_istream() {} _ZNSt3__113basic_istreamIcNS_11char_traitsIcEEE6sentryC2ERS3_b: 313| 1.08k|basic_istream<_CharT, _Traits>::sentry::sentry(basic_istream<_CharT, _Traits>& __is, bool __noskipws) : __ok_(false) { 314| 1.08k| if (__is.good()) { ------------------ | Branch (314:7): [True: 667, False: 413] ------------------ 315| 667| if (__is.tie()) ------------------ | Branch (315:9): [True: 0, False: 667] ------------------ 316| 0| __is.tie()->flush(); 317| 667| if (!__noskipws && (__is.flags() & ios_base::skipws)) { ------------------ | Branch (317:9): [True: 667, False: 0] | Branch (317:24): [True: 667, False: 0] ------------------ 318| 667| typedef istreambuf_iterator<_CharT, _Traits> _Ip; 319| 667| const ctype<_CharT>& __ct = std::use_facet >(__is.getloc()); 320| 667| _Ip __i(__is); 321| 667| _Ip __eof; 322| 15.3k| for (; __i != __eof; ++__i) ------------------ | Branch (322:14): [True: 15.2k, False: 6] ------------------ 323| 15.2k| if (!__ct.is(__ct.space, *__i)) ------------------ | Branch (323:13): [True: 661, False: 14.6k] ------------------ 324| 661| break; 325| 667| if (__i == __eof) ------------------ | Branch (325:11): [True: 6, False: 661] ------------------ 326| 6| __is.setstate(ios_base::failbit | ios_base::eofbit); 327| 667| } 328| 667| __ok_ = __is.good(); 329| 667| } else 330| 413| __is.setstate(ios_base::failbit); 331| 1.08k|} _ZNSt3__113basic_istreamIcNS_11char_traitsIcEEErsERj: 380| 1.08k|basic_istream<_CharT, _Traits>& basic_istream<_CharT, _Traits>::operator>>(unsigned int& __n) { 381| 1.08k| return std::__input_arithmetic(*this, __n); 382| 1.08k|} _ZNKSt3__113basic_istreamIcNS_11char_traitsIcEEE6sentrycvbB8ne180100Ev: 306| 1.08k| _LIBCPP_HIDE_FROM_ABI explicit operator bool() const { return __ok_; } _ZNSt3__118__input_arithmeticB8ne180100IjcNS_11char_traitsIcEEEERNS_13basic_istreamIT0_T1_EES7_RT_: 350| 1.08k|__input_arithmetic(basic_istream<_CharT, _Traits>& __is, _Tp& __n) { 351| 1.08k| ios_base::iostate __state = ios_base::goodbit; 352| 1.08k| typename basic_istream<_CharT, _Traits>::sentry __s(__is); 353| 1.08k| if (__s) { ------------------ | Branch (353:7): [True: 661, False: 419] ------------------ 354| 661|#ifndef _LIBCPP_HAS_NO_EXCEPTIONS 355| 661| try { 356| 661|#endif // _LIBCPP_HAS_NO_EXCEPTIONS 357| 661| typedef istreambuf_iterator<_CharT, _Traits> _Ip; 358| 661| typedef num_get<_CharT, _Ip> _Fp; 359| 661| std::use_facet<_Fp>(__is.getloc()).get(_Ip(__is), _Ip(), __is, __state, __n); 360| 661|#ifndef _LIBCPP_HAS_NO_EXCEPTIONS 361| 661| } catch (...) { 362| 0| __state |= ios_base::badbit; 363| 0| __is.__setstate_nothrow(__state); 364| 0| if (__is.exceptions() & ios_base::badbit) { ------------------ | Branch (364:11): [True: 0, False: 0] ------------------ 365| 0| throw; 366| 0| } 367| 0| } 368| 0|#endif 369| 661| __is.setstate(__state); 370| 661| } 371| 1.08k| return __is; 372| 1.08k|} _ZNSt3__123__libcpp_numeric_limitsIlLb1EE3maxB8ne180100Ev: 203| 1| _LIBCPP_HIDE_FROM_ABI static _LIBCPP_CONSTEXPR type max() _NOEXCEPT { return __max; } _ZNSt3__114numeric_limitsIlE3maxB8ne180100Ev: 446| 1| _LIBCPP_HIDE_FROM_ABI static _LIBCPP_CONSTEXPR type max() _NOEXCEPT { return __base::max(); } _ZNSt3__114numeric_limitsIjE3maxB8ne180100Ev: 446| 627| _LIBCPP_HIDE_FROM_ABI static _LIBCPP_CONSTEXPR type max() _NOEXCEPT { return __base::max(); } _ZNSt3__123__libcpp_numeric_limitsIjLb1EE3maxB8ne180100Ev: 203| 627| _LIBCPP_HIDE_FROM_ABI static _LIBCPP_CONSTEXPR type max() _NOEXCEPT { return __max; } _ZNSt3__123__libcpp_numeric_limitsImLb1EE3maxB8ne180100Ev: 203| 59.1k| _LIBCPP_HIDE_FROM_ABI static _LIBCPP_CONSTEXPR type max() _NOEXCEPT { return __max; } _ZNSt3__114numeric_limitsImE3maxB8ne180100Ev: 446| 59.1k| _LIBCPP_HIDE_FROM_ABI static _LIBCPP_CONSTEXPR type max() _NOEXCEPT { return __base::max(); } _ZNSt3__110__time_putC2B8ne180100Ev: 2076| 2| _LIBCPP_HIDE_FROM_ABI __time_put() : __loc_(_LIBCPP_GET_C_LOCALE) {} ------------------ | | 246| 2|# define _LIBCPP_GET_C_LOCALE __cloc() ------------------ _ZNSt3__110money_baseC2B8ne180100Ev: 2188| 4| _LIBCPP_HIDE_FROM_ABI money_base() {} _ZNSt3__113messages_baseC2B8ne180100Ev: 3012| 2| _LIBCPP_HIDE_FROM_ABI messages_base() {} _ZNKSt3__17num_getIcNS_19istreambuf_iteratorIcNS_11char_traitsIcEEEEE3getB8ne180100ES4_S4_RNS_8ios_baseERjS8_: 611| 661| get(iter_type __b, iter_type __e, ios_base& __iob, ios_base::iostate& __err, unsigned int& __v) const { 612| 661| return do_get(__b, __e, __iob, __err, __v); 613| 661| } _ZNKSt3__17num_putIcNS_19ostreambuf_iteratorIcNS_11char_traitsIcEEEEE3putB8ne180100ES4_RNS_8ios_baseEcm: 1174| 492| _LIBCPP_HIDE_FROM_ABI iter_type put(iter_type __s, ios_base& __iob, char_type __fl, unsigned long __v) const { 1175| 492| return do_put(__s, __iob, __fl, __v); 1176| 492| } _ZNKSt3__17num_getIcNS_19istreambuf_iteratorIcNS_11char_traitsIcEEEEE6do_getES4_S4_RNS_8ios_baseERjS8_: 679| 661| do_get(iter_type __b, iter_type __e, ios_base& __iob, ios_base::iostate& __err, unsigned int& __v) const { 680| 661| return this->__do_get_unsigned(__b, __e, __iob, __err, __v); 681| 661| } _ZNKSt3__17num_getIcNS_19istreambuf_iteratorIcNS_11char_traitsIcEEEEE17__do_get_unsignedIjEES4_S4_S4_RNS_8ios_baseERjRT_: 893| 661| iter_type __b, iter_type __e, ios_base& __iob, ios_base::iostate& __err, _Unsigned& __v) const { 894| | // Stage 1 895| 661| int __base = this->__get_base(__iob); 896| | // Stage 2 897| 661| char_type __thousands_sep; 898| 661| const int __atoms_size = 26; 899| |#ifdef _LIBCPP_ABI_OPTIMIZED_LOCALE_NUM_GET 900| | char_type __atoms1[__atoms_size]; 901| | const char_type* __atoms = this->__do_widen(__iob, __atoms1); 902| | string __grouping = this->__stage2_int_prep(__iob, __thousands_sep); 903| |#else 904| 661| char_type __atoms[__atoms_size]; 905| 661| string __grouping = this->__stage2_int_prep(__iob, __atoms, __thousands_sep); 906| 661|#endif 907| 661| string __buf; 908| 661| __buf.resize(__buf.capacity()); 909| 661| char* __a = &__buf[0]; 910| 661| char* __a_end = __a; 911| 661| unsigned __g[__num_get_base::__num_get_buf_sz]; 912| 661| unsigned* __g_end = __g; 913| 661| unsigned __dc = 0; 914| 237k| for (; __b != __e; ++__b) { ------------------ | Branch (914:10): [True: 237k, False: 116] ------------------ 915| 237k| if (__a_end == __a + __buf.size()) { ------------------ | Branch (915:9): [True: 375, False: 237k] ------------------ 916| 375| size_t __tmp = __buf.size(); 917| 375| __buf.resize(2 * __buf.size()); 918| 375| __buf.resize(__buf.capacity()); 919| 375| __a = &__buf[0]; 920| 375| __a_end = __a + __tmp; 921| 375| } 922| 237k| if (this->__stage2_int_loop(*__b, __base, __a, __a_end, __dc, __thousands_sep, __grouping, __g, __g_end, __atoms)) ------------------ | Branch (922:9): [True: 545, False: 237k] ------------------ 923| 545| break; 924| 237k| } 925| 661| if (__grouping.size() != 0 && __g_end - __g < __num_get_base::__num_get_buf_sz) ------------------ | Branch (925:7): [True: 0, False: 661] | Branch (925:33): [True: 0, False: 0] ------------------ 926| 0| *__g_end++ = __dc; 927| | // Stage 3 928| 661| __v = std::__num_get_unsigned_integral<_Unsigned>(__a, __a_end, __err, __base); 929| | // Digit grouping checked 930| 661| __check_grouping(__grouping, __g, __g_end, __err); 931| | // EOF checked 932| 661| if (__b == __e) ------------------ | Branch (932:7): [True: 116, False: 545] ------------------ 933| 116| __err |= ios_base::eofbit; 934| 661| return __b; 935| 661|} _ZNSt3__127__num_get_unsigned_integralB8ne180100IjEET_PKcS3_Rji: 741| 661|__num_get_unsigned_integral(const char* __a, const char* __a_end, ios_base::iostate& __err, int __base) { 742| 661| if (__a != __a_end) { ------------------ | Branch (742:7): [True: 586, False: 75] ------------------ 743| 586| const bool __negate = *__a == '-'; 744| 586| if (__negate && ++__a == __a_end) { ------------------ | Branch (744:9): [True: 275, False: 311] | Branch (744:21): [True: 4, False: 271] ------------------ 745| 4| __err = ios_base::failbit; 746| 4| return 0; 747| 4| } 748| 582| __libcpp_remove_reference_t __save_errno = errno; 749| 582| errno = 0; 750| 582| char* __p2; 751| 582| unsigned long long __ll = strtoull_l(__a, &__p2, __base, _LIBCPP_GET_C_LOCALE); ------------------ | | 246| 582|# define _LIBCPP_GET_C_LOCALE __cloc() ------------------ 752| 582| __libcpp_remove_reference_t __current_errno = errno; 753| 582| if (__current_errno == 0) ------------------ | Branch (753:9): [True: 558, False: 24] ------------------ 754| 582| errno = __save_errno; 755| 582| if (__p2 != __a_end) { ------------------ | Branch (755:9): [True: 2, False: 580] ------------------ 756| 2| __err = ios_base::failbit; 757| 2| return 0; 758| 580| } else if (__current_errno == ERANGE || numeric_limits<_Tp>::max() < __ll) { ------------------ | Branch (758:16): [True: 24, False: 556] | Branch (758:45): [True: 47, False: 509] ------------------ 759| 71| __err = ios_base::failbit; 760| 71| return numeric_limits<_Tp>::max(); 761| 71| } 762| 509| _Tp __res = static_cast<_Tp>(__ll); 763| 509| if (__negate) ------------------ | Branch (763:9): [True: 269, False: 240] ------------------ 764| 269| __res = -__res; 765| 509| return __res; 766| 582| } 767| 75| __err = ios_base::failbit; 768| 75| return 0; 769| 661|} _ZNSt3__19__num_getIcE17__stage2_int_prepERNS_8ios_baseEPcRc: 448| 661|string __num_get<_CharT>::__stage2_int_prep(ios_base& __iob, _CharT* __atoms, _CharT& __thousands_sep) { 449| 661| locale __loc = __iob.getloc(); 450| 661| std::use_facet >(__loc).widen(__src, __src + 26, __atoms); 451| 661| const numpunct<_CharT>& __np = std::use_facet >(__loc); 452| 661| __thousands_sep = __np.thousands_sep(); 453| 661| return __np.grouping(); 454| 661|} _ZNSt3__19__num_getIcE17__stage2_int_loopEciPcRS2_RjcRKNS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEEPjRSD_S2_: 480| 237k|{ 481| 237k| if (__a_end == __a && (__ct == __atoms[24] || __ct == __atoms[25])) { ------------------ | Branch (481:7): [True: 661, False: 236k] | Branch (481:26): [True: 47, False: 614] | Branch (481:49): [True: 275, False: 339] ------------------ 482| 322| *__a_end++ = __ct == __atoms[24] ? '+' : '-'; ------------------ | Branch (482:18): [True: 47, False: 275] ------------------ 483| 322| __dc = 0; 484| 322| return 0; 485| 322| } 486| 237k| if (__grouping.size() != 0 && __ct == __thousands_sep) { ------------------ | Branch (486:7): [True: 0, False: 237k] | Branch (486:33): [True: 0, False: 0] ------------------ 487| 0| if (__g_end - __g < __num_get_buf_sz) { ------------------ | Branch (487:9): [True: 0, False: 0] ------------------ 488| 0| *__g_end++ = __dc; 489| 0| __dc = 0; 490| 0| } 491| 0| return 0; 492| 0| } 493| 237k| ptrdiff_t __f = std::find(__atoms, __atoms + 26, __ct) - __atoms; 494| 237k| if (__f >= 24) ------------------ | Branch (494:7): [True: 533, False: 236k] ------------------ 495| 533| return -1; 496| 236k| switch (__base) { ------------------ | Branch (496:11): [True: 236k, False: 0] ------------------ 497| 0| case 8: ------------------ | Branch (497:3): [True: 0, False: 236k] ------------------ 498| 236k| case 10: ------------------ | Branch (498:3): [True: 236k, False: 0] ------------------ 499| 236k| if (__f >= __base) ------------------ | Branch (499:9): [True: 12, False: 236k] ------------------ 500| 12| return -1; 501| 236k| break; 502| 236k| case 16: ------------------ | Branch (502:3): [True: 0, False: 236k] ------------------ 503| 0| if (__f < 22) ------------------ | Branch (503:9): [True: 0, False: 0] ------------------ 504| 0| break; 505| 0| if (__a_end != __a && __a_end - __a <= 2 && __a_end[-1] == '0') { ------------------ | Branch (505:9): [True: 0, False: 0] | Branch (505:27): [True: 0, False: 0] | Branch (505:49): [True: 0, False: 0] ------------------ 506| 0| __dc = 0; 507| 0| *__a_end++ = __src[__f]; 508| 0| return 0; 509| 0| } 510| 0| return -1; 511| 236k| } 512| 236k| *__a_end++ = __src[__f]; 513| 236k| ++__dc; 514| 236k| return 0; 515| 236k|} _ZNKSt3__17num_putIcNS_19ostreambuf_iteratorIcNS_11char_traitsIcEEEEE6do_putES4_RNS_8ios_baseEcm: 1339| 492|num_put<_CharT, _OutputIterator>::do_put(iter_type __s, ios_base& __iob, char_type __fl, unsigned long __v) const { 1340| 492| return this->__do_put_integral(__s, __iob, __fl, __v, "l"); 1341| 492|} _ZNSt3__19__num_putIcE21__widen_and_group_intEPcS2_S2_S2_RS2_S3_RKNS_6localeE: 1053| 492| char* __nb, char* __np, char* __ne, _CharT* __ob, _CharT*& __op, _CharT*& __oe, const locale& __loc) { 1054| 492| const ctype<_CharT>& __ct = std::use_facet >(__loc); 1055| 492| const numpunct<_CharT>& __npt = std::use_facet >(__loc); 1056| 492| string __grouping = __npt.grouping(); 1057| 492| if (__grouping.empty()) { ------------------ | Branch (1057:7): [True: 492, False: 0] ------------------ 1058| 492| __ct.widen(__nb, __ne, __ob); 1059| 492| __oe = __ob + (__ne - __nb); 1060| 492| } else { 1061| 0| __oe = __ob; 1062| 0| char* __nf = __nb; 1063| 0| if (*__nf == '-' || *__nf == '+') ------------------ | Branch (1063:9): [True: 0, False: 0] | Branch (1063:25): [True: 0, False: 0] ------------------ 1064| 0| *__oe++ = __ct.widen(*__nf++); 1065| 0| if (__ne - __nf >= 2 && __nf[0] == '0' && (__nf[1] == 'x' || __nf[1] == 'X')) { ------------------ | Branch (1065:9): [True: 0, False: 0] | Branch (1065:29): [True: 0, False: 0] | Branch (1065:48): [True: 0, False: 0] | Branch (1065:66): [True: 0, False: 0] ------------------ 1066| 0| *__oe++ = __ct.widen(*__nf++); 1067| 0| *__oe++ = __ct.widen(*__nf++); 1068| 0| } 1069| 0| std::reverse(__nf, __ne); 1070| 0| _CharT __thousands_sep = __npt.thousands_sep(); 1071| 0| unsigned __dc = 0; 1072| 0| unsigned __dg = 0; 1073| 0| for (char* __p = __nf; __p < __ne; ++__p) { ------------------ | Branch (1073:28): [True: 0, False: 0] ------------------ 1074| 0| if (static_cast(__grouping[__dg]) > 0 && __dc == static_cast(__grouping[__dg])) { ------------------ | Branch (1074:11): [True: 0, False: 0] | Branch (1074:58): [True: 0, False: 0] ------------------ 1075| 0| *__oe++ = __thousands_sep; 1076| 0| __dc = 0; 1077| 0| if (__dg < __grouping.size() - 1) ------------------ | Branch (1077:13): [True: 0, False: 0] ------------------ 1078| 0| ++__dg; 1079| 0| } 1080| 0| *__oe++ = __ct.widen(*__p); 1081| 0| ++__dc; 1082| 0| } 1083| 0| std::reverse(__ob + (__nf - __nb), __oe); 1084| 0| } 1085| 492| if (__np == __ne) ------------------ | Branch (1085:7): [True: 0, False: 492] ------------------ 1086| 0| __op = __oe; 1087| 492| else 1088| 492| __op = __ob + (__np - __nb); 1089| 492|} _ZNSt3__17num_getIcNS_19istreambuf_iteratorIcNS_11char_traitsIcEEEEEC2B8ne180100Em: 588| 1| _LIBCPP_HIDE_FROM_ABI explicit num_get(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__17num_getIwNS_19istreambuf_iteratorIwNS_11char_traitsIwEEEEEC2B8ne180100Em: 588| 1| _LIBCPP_HIDE_FROM_ABI explicit num_get(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__17num_putIcNS_19ostreambuf_iteratorIcNS_11char_traitsIcEEEEEC2B8ne180100Em: 1160| 1| _LIBCPP_HIDE_FROM_ABI explicit num_put(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__17num_putIwNS_19ostreambuf_iteratorIwNS_11char_traitsIwEEEEEC2B8ne180100Em: 1160| 1| _LIBCPP_HIDE_FROM_ABI explicit num_put(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__110moneypunctIcLb0EEC2B8ne180100Em: 2199| 1| _LIBCPP_HIDE_FROM_ABI explicit moneypunct(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__110moneypunctIcLb1EEC2B8ne180100Em: 2199| 1| _LIBCPP_HIDE_FROM_ABI explicit moneypunct(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__110moneypunctIwLb0EEC2B8ne180100Em: 2199| 1| _LIBCPP_HIDE_FROM_ABI explicit moneypunct(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__110moneypunctIwLb1EEC2B8ne180100Em: 2199| 1| _LIBCPP_HIDE_FROM_ABI explicit moneypunct(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__19money_getIcNS_19istreambuf_iteratorIcNS_11char_traitsIcEEEEEC2B8ne180100Em: 2379| 1| _LIBCPP_HIDE_FROM_ABI explicit money_get(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__111__money_getIcEC2B8ne180100Ev: 2317| 1| _LIBCPP_HIDE_FROM_ABI __money_get() {} _ZNSt3__19money_getIwNS_19istreambuf_iteratorIwNS_11char_traitsIwEEEEEC2B8ne180100Em: 2379| 1| _LIBCPP_HIDE_FROM_ABI explicit money_get(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__111__money_getIwEC2B8ne180100Ev: 2317| 1| _LIBCPP_HIDE_FROM_ABI __money_get() {} _ZNSt3__19money_putIcNS_19ostreambuf_iteratorIcNS_11char_traitsIcEEEEEC2B8ne180100Em: 2867| 1| _LIBCPP_HIDE_FROM_ABI explicit money_put(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__111__money_putIcEC2B8ne180100Ev: 2690| 1| _LIBCPP_HIDE_FROM_ABI __money_put() {} _ZNSt3__19money_putIwNS_19ostreambuf_iteratorIwNS_11char_traitsIwEEEEEC2B8ne180100Em: 2867| 1| _LIBCPP_HIDE_FROM_ABI explicit money_put(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__111__money_putIwEC2B8ne180100Ev: 2690| 1| _LIBCPP_HIDE_FROM_ABI __money_put() {} _ZNSt3__18time_getIcNS_19istreambuf_iteratorIcNS_11char_traitsIcEEEEEC2B8ne180100Em: 1529| 1| _LIBCPP_HIDE_FROM_ABI explicit time_get(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__18time_getIwNS_19istreambuf_iteratorIwNS_11char_traitsIwEEEEEC2B8ne180100Em: 1529| 1| _LIBCPP_HIDE_FROM_ABI explicit time_get(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__18time_putIcNS_19ostreambuf_iteratorIcNS_11char_traitsIcEEEEEC2B8ne180100Em: 2092| 1| _LIBCPP_HIDE_FROM_ABI explicit time_put(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__18time_putIwNS_19ostreambuf_iteratorIwNS_11char_traitsIwEEEEEC2B8ne180100Em: 2092| 1| _LIBCPP_HIDE_FROM_ABI explicit time_put(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__18messagesIcEC2B8ne180100Em: 3021| 1| _LIBCPP_HIDE_FROM_ABI explicit messages(size_t __refs = 0) : locale::facet(__refs) {} _ZNSt3__18messagesIwEC2B8ne180100Em: 3021| 1| _LIBCPP_HIDE_FROM_ABI explicit messages(size_t __refs = 0) : locale::facet(__refs) {} _ZNKSt3__17num_putIcNS_19ostreambuf_iteratorIcNS_11char_traitsIcEEEEE17__do_put_integralB8ne180100ImEES4_S4_RNS_8ios_baseEcT_PKc: 1296| 492| iter_type __s, ios_base& __iob, char_type __fl, _Integral __v, char const* __len) const { 1297| | // Stage 1 - Get number in narrow char 1298| 492| char __fmt[8] = {'%', 0}; 1299| 492| this->__format_int(__fmt + 1, __len, is_signed<_Integral>::value, __iob.flags()); 1300| | // Worst case is octal, with showbase enabled. Note that octal is always 1301| | // printed as an unsigned value. 1302| 492| using _Unsigned = typename make_unsigned<_Integral>::type; 1303| 492| _LIBCPP_CONSTEXPR const unsigned __nbuf = ------------------ | | 622| 492|# define _LIBCPP_CONSTEXPR constexpr ------------------ 1304| 492| (numeric_limits<_Unsigned>::digits / 3) // 1 char per 3 bits 1305| 492| + ((numeric_limits<_Unsigned>::digits % 3) != 0) // round up 1306| 492| + 2; // base prefix + terminating null character 1307| 492| char __nar[__nbuf]; 1308| 492| _LIBCPP_DIAGNOSTIC_PUSH 1309| 492| _LIBCPP_CLANG_DIAGNOSTIC_IGNORED("-Wformat-nonliteral") 1310| 492| _LIBCPP_GCC_DIAGNOSTIC_IGNORED("-Wformat-nonliteral") 1311| 492| int __nc = __libcpp_snprintf_l(__nar, sizeof(__nar), _LIBCPP_GET_C_LOCALE, __fmt, __v); ------------------ | | 246| 492|# define _LIBCPP_GET_C_LOCALE __cloc() ------------------ 1312| 492| _LIBCPP_DIAGNOSTIC_POP 1313| 492| char* __ne = __nar + __nc; 1314| 492| char* __np = this->__identify_padding(__nar, __ne, __iob); 1315| | // Stage 2 - Widen __nar while adding thousands separators 1316| 492| char_type __o[2 * (__nbuf - 1) - 1]; 1317| 492| char_type* __op; // pad here 1318| 492| char_type* __oe; // end of output 1319| 492| this->__widen_and_group_int(__nar, __np, __ne, __o, __op, __oe, __iob.getloc()); 1320| | // [__o, __oe) contains thousands_sep'd wide number 1321| | // Stage 3 & 4 1322| 492| return std::__pad_and_output(__s, __o, __op, __oe, __iob, __fl); 1323| 492|} _ZNSt3__116__pad_and_outputB8ne180100IcNS_11char_traitsIcEEEENS_19ostreambuf_iteratorIT_T0_EES6_PKS4_S8_S8_RNS_8ios_baseES4_: 1246| 1.25k| _CharT __fl) { 1247| 1.25k| if (__s.__sbuf_ == nullptr) ------------------ | Branch (1247:7): [True: 0, False: 1.25k] ------------------ 1248| 0| return __s; 1249| 1.25k| streamsize __sz = __oe - __ob; 1250| 1.25k| streamsize __ns = __iob.width(); 1251| 1.25k| if (__ns > __sz) ------------------ | Branch (1251:7): [True: 0, False: 1.25k] ------------------ 1252| 0| __ns -= __sz; 1253| 1.25k| else 1254| 1.25k| __ns = 0; 1255| 1.25k| streamsize __np = __op - __ob; 1256| 1.25k| if (__np > 0) { ------------------ | Branch (1256:7): [True: 0, False: 1.25k] ------------------ 1257| 0| if (__s.__sbuf_->sputn(__ob, __np) != __np) { ------------------ | Branch (1257:9): [True: 0, False: 0] ------------------ 1258| 0| __s.__sbuf_ = nullptr; 1259| 0| return __s; 1260| 0| } 1261| 0| } 1262| 1.25k| if (__ns > 0) { ------------------ | Branch (1262:7): [True: 0, False: 1.25k] ------------------ 1263| 0| basic_string<_CharT, _Traits> __sp(__ns, __fl); 1264| 0| if (__s.__sbuf_->sputn(__sp.data(), __ns) != __ns) { ------------------ | Branch (1264:9): [True: 0, False: 0] ------------------ 1265| 0| __s.__sbuf_ = nullptr; 1266| 0| return __s; 1267| 0| } 1268| 0| } 1269| 1.25k| __np = __oe - __op; 1270| 1.25k| if (__np > 0) { ------------------ | Branch (1270:7): [True: 1.25k, False: 0] ------------------ 1271| 1.25k| if (__s.__sbuf_->sputn(__op, __np) != __np) { ------------------ | Branch (1271:9): [True: 0, False: 1.25k] ------------------ 1272| 0| __s.__sbuf_ = nullptr; 1273| 0| return __s; 1274| 0| } 1275| 1.25k| } 1276| 1.25k| __iob.width(0); 1277| 1.25k| return __s; 1278| 1.25k|} _ZNSt3__124__is_overaligned_for_newB8ne180100Em: 260| 20.1k|_LIBCPP_CONSTEXPR inline _LIBCPP_HIDE_FROM_ABI bool __is_overaligned_for_new(size_t __align) _NOEXCEPT { 261| 20.1k|#ifdef __STDCPP_DEFAULT_NEW_ALIGNMENT__ 262| 20.1k| return __align > __STDCPP_DEFAULT_NEW_ALIGNMENT__; 263| |#else 264| | return __align > _LIBCPP_ALIGNOF(max_align_t); 265| |#endif 266| 20.1k|} _ZNSt3__117__libcpp_allocateB8ne180100Emm: 286| 19.7k|inline _LIBCPP_HIDE_FROM_ABI void* __libcpp_allocate(size_t __size, size_t __align) { 287| 19.7k|#ifndef _LIBCPP_HAS_NO_ALIGNED_ALLOCATION 288| 19.7k| if (__is_overaligned_for_new(__align)) { ------------------ | Branch (288:7): [True: 0, False: 19.7k] ------------------ 289| 0| const align_val_t __align_val = static_cast(__align); 290| 0| return __libcpp_operator_new(__size, __align_val); 291| 0| } 292| 19.7k|#endif 293| | 294| 19.7k| (void)__align; 295| 19.7k| return __libcpp_operator_new(__size); 296| 19.7k|} _ZNSt3__119__libcpp_deallocateB8ne180100EPvmm: 308| 441|inline _LIBCPP_HIDE_FROM_ABI void __libcpp_deallocate(void* __ptr, size_t __size, size_t __align) { 309| |#if defined(_LIBCPP_HAS_NO_ALIGNED_ALLOCATION) 310| | (void)__align; 311| | return __do_deallocate_handle_size(__ptr, __size); 312| |#else 313| 441| if (__is_overaligned_for_new(__align)) { ------------------ | Branch (313:7): [True: 0, False: 441] ------------------ 314| 0| const align_val_t __align_val = static_cast(__align); 315| 0| return __do_deallocate_handle_size(__ptr, __size, __align_val); 316| 441| } else { 317| 441| return __do_deallocate_handle_size(__ptr, __size); 318| 441| } 319| 441|#endif 320| 441|} _ZNSt3__121__libcpp_operator_newB8ne180100IJmEEEPvDpT_: 269| 19.7k|_LIBCPP_HIDE_FROM_ABI void* __libcpp_operator_new(_Args... __args) { 270| 19.7k|#if __has_builtin(__builtin_operator_new) && __has_builtin(__builtin_operator_delete) 271| 19.7k| return __builtin_operator_new(__args...); 272| |#else 273| | return ::operator new(__args...); 274| |#endif 275| 19.7k|} _ZNSt3__127__do_deallocate_handle_sizeB8ne180100IJEEEvPvmDpT_: 299| 441|_LIBCPP_HIDE_FROM_ABI void __do_deallocate_handle_size(void* __ptr, size_t __size, _Args... __args) { 300| |#ifdef _LIBCPP_HAS_NO_SIZED_DEALLOCATION 301| | (void)__size; 302| | return std::__libcpp_operator_delete(__ptr, __args...); 303| |#else 304| 441| return std::__libcpp_operator_delete(__ptr, __size, __args...); 305| 441|#endif 306| 441|} _ZNSt3__124__libcpp_operator_deleteB8ne180100IJPvmEEEvDpT_: 278| 441|_LIBCPP_HIDE_FROM_ABI void __libcpp_operator_delete(_Args... __args) { 279| 441|#if __has_builtin(__builtin_operator_new) && __has_builtin(__builtin_operator_delete) 280| 441| __builtin_operator_delete(__args...); 281| |#else 282| | ::operator delete(__args...); 283| |#endif 284| 441|} _ZNSt3__113basic_ostreamIcNS_11char_traitsIcEEED2Ev: 348| 453|basic_ostream<_CharT, _Traits>::~basic_ostream() {} _ZNSt3__113basic_ostreamIcNS_11char_traitsIcEEE6sentryC2ERS3_: 313| 1.25k|basic_ostream<_CharT, _Traits>::sentry::sentry(basic_ostream<_CharT, _Traits>& __os) : __ok_(false), __os_(__os) { 314| 1.25k| if (__os.good()) { ------------------ | Branch (314:7): [True: 1.25k, False: 0] ------------------ 315| 1.25k| if (__os.tie()) ------------------ | Branch (315:9): [True: 0, False: 1.25k] ------------------ 316| 0| __os.tie()->flush(); 317| 1.25k| __ok_ = true; 318| 1.25k| } 319| 1.25k|} _ZNSt3__113basic_ostreamIcNS_11char_traitsIcEEE6sentryD2Ev: 322| 1.25k|basic_ostream<_CharT, _Traits>::sentry::~sentry() { 323| 1.25k| if (__os_.rdbuf() && __os_.good() && (__os_.flags() & ios_base::unitbuf) && !uncaught_exception()) { ------------------ | Branch (323:7): [True: 1.25k, False: 0] | Branch (323:24): [True: 1.25k, False: 0] | Branch (323:40): [True: 0, False: 1.25k] | Branch (323:79): [True: 0, False: 0] ------------------ 324| 0|#ifndef _LIBCPP_HAS_NO_EXCEPTIONS 325| 0| try { 326| 0|#endif // _LIBCPP_HAS_NO_EXCEPTIONS 327| 0| if (__os_.rdbuf()->pubsync() == -1) ------------------ | Branch (327:11): [True: 0, False: 0] ------------------ 328| 0| __os_.setstate(ios_base::badbit); 329| 0|#ifndef _LIBCPP_HAS_NO_EXCEPTIONS 330| 0| } catch (...) { 331| 0| } 332| 0|#endif // _LIBCPP_HAS_NO_EXCEPTIONS 333| 0| } 334| 1.25k|} _ZNSt3__113basic_ostreamIcNS_11char_traitsIcEEElsEj: 486| 492|basic_ostream<_CharT, _Traits>& basic_ostream<_CharT, _Traits>::operator<<(unsigned int __n) { 487| 492|#ifndef _LIBCPP_HAS_NO_EXCEPTIONS 488| 492| try { 489| 492|#endif // _LIBCPP_HAS_NO_EXCEPTIONS 490| 492| sentry __s(*this); 491| 492| if (__s) { ------------------ | Branch (491:9): [True: 492, False: 0] ------------------ 492| 492| typedef num_put > _Fp; 493| 492| const _Fp& __f = std::use_facet<_Fp>(this->getloc()); 494| 492| if (__f.put(*this, *this, this->fill(), static_cast(__n)).failed()) ------------------ | Branch (494:11): [True: 0, False: 492] ------------------ 495| 0| this->setstate(ios_base::badbit | ios_base::failbit); 496| 492| } 497| 492|#ifndef _LIBCPP_HAS_NO_EXCEPTIONS 498| 492| } catch (...) { 499| 0| this->__set_badbit_and_consider_rethrow(); 500| 0| } 501| 492|#endif // _LIBCPP_HAS_NO_EXCEPTIONS 502| 492| return *this; 503| 492|} _ZNKSt3__113basic_ostreamIcNS_11char_traitsIcEEE6sentrycvbB8ne180100Ev: 309| 492| _LIBCPP_HIDE_FROM_ABI explicit operator bool() const { return __ok_; } _ZNSt3__115basic_stringbufIcNS_11char_traitsIcEENS_9allocatorIcEEE9underflowEv: 665| 122|basic_stringbuf<_CharT, _Traits, _Allocator>::underflow() { 666| 122| if (__hm_ < this->pptr()) ------------------ | Branch (666:7): [True: 0, False: 122] ------------------ 667| 0| __hm_ = this->pptr(); 668| 122| if (__mode_ & ios_base::in) { ------------------ | Branch (668:7): [True: 122, False: 0] ------------------ 669| 122| if (this->egptr() < __hm_) ------------------ | Branch (669:9): [True: 0, False: 122] ------------------ 670| 0| this->setg(this->eback(), this->gptr(), __hm_); 671| 122| if (this->gptr() < this->egptr()) ------------------ | Branch (671:9): [True: 0, False: 122] ------------------ 672| 0| return traits_type::to_int_type(*this->gptr()); 673| 122| } 674| 122| return traits_type::eof(); 675| 122|} _ZNSt3__115basic_stringbufIcNS_11char_traitsIcEENS_9allocatorIcEEE8overflowEi: 698| 438|basic_stringbuf<_CharT, _Traits, _Allocator>::overflow(int_type __c) { 699| 438| if (!traits_type::eq_int_type(__c, traits_type::eof())) { ------------------ | Branch (699:7): [True: 438, False: 0] ------------------ 700| 438| ptrdiff_t __ninp = this->gptr() - this->eback(); 701| 438| if (this->pptr() == this->epptr()) { ------------------ | Branch (701:9): [True: 438, False: 0] ------------------ 702| 438| if (!(__mode_ & ios_base::out)) ------------------ | Branch (702:11): [True: 0, False: 438] ------------------ 703| 0| return traits_type::eof(); 704| 438|#ifndef _LIBCPP_HAS_NO_EXCEPTIONS 705| 438| try { 706| 438|#endif // _LIBCPP_HAS_NO_EXCEPTIONS 707| 438| ptrdiff_t __nout = this->pptr() - this->pbase(); 708| 438| ptrdiff_t __hm = __hm_ - this->pbase(); 709| 438| __str_.push_back(char_type()); 710| 438| __str_.resize(__str_.capacity()); 711| 438| char_type* __p = const_cast(__str_.data()); 712| 438| this->setp(__p, __p + __str_.size()); 713| 438| this->__pbump(__nout); 714| 438| __hm_ = this->pbase() + __hm; 715| 438|#ifndef _LIBCPP_HAS_NO_EXCEPTIONS 716| 438| } catch (...) { 717| 0| return traits_type::eof(); 718| 0| } 719| 438|#endif // _LIBCPP_HAS_NO_EXCEPTIONS 720| 438| } 721| 438| __hm_ = std::max(this->pptr() + 1, __hm_); 722| 438| if (__mode_ & ios_base::in) { ------------------ | Branch (722:9): [True: 0, False: 438] ------------------ 723| 0| char_type* __p = const_cast(__str_.data()); 724| 0| this->setg(__p, __p + __ninp, __hm_); 725| 0| } 726| 438| return this->sputc(traits_type::to_char_type(__c)); 727| 438| } 728| 0| return traits_type::not_eof(__c); 729| 438|} _ZNKSt3__118__libcpp_refstring5c_strB8ne180100Ev: 68| 21.5k| _LIBCPP_HIDE_FROM_ABI const char* c_str() const _NOEXCEPT { return __imp_; } _ZNSt3__115basic_streambufIcNS_11char_traitsIcEEED2Ev: 290| 723|basic_streambuf<_CharT, _Traits>::~basic_streambuf() {} _ZNSt3__115basic_streambufIcNS_11char_traitsIcEEE6sbumpcEv: 183| 251k| inline _LIBCPP_HIDE_FROM_ABI_AFTER_V1 int_type sbumpc() { 184| 251k| if (__ninp_ == __einp_) ------------------ | Branch (184:9): [True: 0, False: 251k] ------------------ 185| 0| return uflow(); 186| 251k| return traits_type::to_int_type(*__ninp_++); 187| 251k| } _ZNSt3__115basic_streambufIcNS_11char_traitsIcEEE5sgetcEv: 189| 507k| inline _LIBCPP_HIDE_FROM_ABI_AFTER_V1 int_type sgetc() { 190| 507k| if (__ninp_ == __einp_) ------------------ | Branch (190:9): [True: 122, False: 507k] ------------------ 191| 122| return underflow(); 192| 507k| return traits_type::to_int_type(*__ninp_); 193| 507k| } _ZNSt3__115basic_streambufIcNS_11char_traitsIcEEE5sputcEc: 211| 438| inline _LIBCPP_HIDE_FROM_ABI_AFTER_V1 int_type sputc(char_type __c) { 212| 438| if (__nout_ == __eout_) ------------------ | Branch (212:9): [True: 0, False: 438] ------------------ 213| 0| return overflow(traits_type::to_int_type(__c)); 214| 438| *__nout_++ = __c; 215| 438| return traits_type::to_int_type(__c); 216| 438| } _ZNSt3__115basic_streambufIcNS_11char_traitsIcEEE5sputnEPKcl: 218| 1.25k| inline _LIBCPP_HIDE_FROM_ABI_AFTER_V1 streamsize sputn(const char_type* __s, streamsize __n) { 219| 1.25k| return xsputn(__s, __n); 220| 1.25k| } _ZNSt3__115basic_streambufIcNS_11char_traitsIcEEEC2Ev: 294| 723| : __binp_(nullptr), __ninp_(nullptr), __einp_(nullptr), __bout_(nullptr), __nout_(nullptr), __eout_(nullptr) {} _ZNSt3__115basic_streambufIcNS_11char_traitsIcEEE4setpEPcS4_: 250| 438| inline _LIBCPP_HIDE_FROM_ABI_AFTER_V1 void setp(char_type* __pbeg, char_type* __pend) { 251| 438| __bout_ = __nout_ = __pbeg; 252| 438| __eout_ = __pend; 253| 438| } _ZNSt3__115basic_streambufIcNS_11char_traitsIcEEE6xsputnEPKcl: 399| 1.25k|streamsize basic_streambuf<_CharT, _Traits>::xsputn(const char_type* __s, streamsize __n) { 400| 1.25k| streamsize __i = 0; 401| 1.25k| int_type __eof = traits_type::eof(); 402| 2.94k| while (__i < __n) { ------------------ | Branch (402:10): [True: 1.68k, False: 1.25k] ------------------ 403| 1.68k| if (__nout_ >= __eout_) { ------------------ | Branch (403:9): [True: 438, False: 1.25k] ------------------ 404| 438| if (overflow(traits_type::to_int_type(*__s)) == __eof) ------------------ | Branch (404:11): [True: 0, False: 438] ------------------ 405| 0| break; 406| 438| ++__s; 407| 438| ++__i; 408| 1.25k| } else { 409| 1.25k| streamsize __chunk_size = std::min(__eout_ - __nout_, __n - __i); 410| 1.25k| traits_type::copy(__nout_, __s, __chunk_size); 411| 1.25k| __nout_ += __chunk_size; 412| 1.25k| __s += __chunk_size; 413| 1.25k| __i += __chunk_size; 414| 1.25k| } 415| 1.68k| } 416| 1.25k| return __i; 417| 1.25k|} _ZNKSt3__115basic_streambufIcNS_11char_traitsIcEEE5ebackB8ne180100Ev: 229| 438| _LIBCPP_HIDE_FROM_ABI char_type* eback() const { return __binp_; } _ZNKSt3__115basic_streambufIcNS_11char_traitsIcEEE4gptrB8ne180100Ev: 230| 560| _LIBCPP_HIDE_FROM_ABI char_type* gptr() const { return __ninp_; } _ZNKSt3__115basic_streambufIcNS_11char_traitsIcEEE5egptrB8ne180100Ev: 231| 244| _LIBCPP_HIDE_FROM_ABI char_type* egptr() const { return __einp_; } _ZNKSt3__115basic_streambufIcNS_11char_traitsIcEEE5pbaseB8ne180100Ev: 242| 1.31k| _LIBCPP_HIDE_FROM_ABI char_type* pbase() const { return __bout_; } _ZNKSt3__115basic_streambufIcNS_11char_traitsIcEEE4pptrB8ne180100Ev: 243| 1.43k| _LIBCPP_HIDE_FROM_ABI char_type* pptr() const { return __nout_; } _ZNKSt3__115basic_streambufIcNS_11char_traitsIcEEE5epptrB8ne180100Ev: 244| 438| _LIBCPP_HIDE_FROM_ABI char_type* epptr() const { return __eout_; } _ZNSt3__115basic_streambufIcNS_11char_traitsIcEEE7__pbumpB8ne180100El: 248| 438| _LIBCPP_HIDE_FROM_ABI void __pbump(streamsize __n) { __nout_ += __n; } _ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE7__allocB8ne180100Ev: 1827| 20.9k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR const allocator_type& __alloc() const _NOEXCEPT { return __r_.second(); } _ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE8max_sizeB8ne180100Ev: 1181| 19.7k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 size_type max_size() const _NOEXCEPT { 1182| 19.7k| size_type __m = __alloc_traits::max_size(__alloc()); 1183| 19.7k| if (__m <= std::numeric_limits::max() / 2) { ------------------ | Branch (1183:9): [True: 0, False: 19.7k] ------------------ 1184| 0| return __m - __alignment; 1185| 19.7k| } else { 1186| 19.7k| bool __uses_lsb = __endian_factor == 2; 1187| 19.7k| return __uses_lsb ? __m - __alignment : (__m / 2) - __alignment; ------------------ | Branch (1187:14): [True: 19.7k, False: 10] ------------------ 1188| 19.7k| } 1189| 19.7k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE13__fits_in_ssoB8ne180100Em: 1788| 20.2k| _LIBCPP_CONSTEXPR _LIBCPP_HIDE_FROM_ABI static bool __fits_in_sso(size_type __sz) { return __sz < __min_cap; } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE16__set_short_sizeB8ne180100Em: 1830| 1.99k| __set_short_size(size_type __s) _NOEXCEPT { 1831| 1.99k| _LIBCPP_ASSERT_INTERNAL(__s < __min_cap, "__s should never be greater than or equal to the short string capacity"); ------------------ | | 417| 1.99k|# define _LIBCPP_ASSERT_INTERNAL(expression, message) _LIBCPP_ASSUME(expression) | | ------------------ | | | | 34| 1.99k|# define _LIBCPP_ASSUME(expression) ((void)0) | | ------------------ ------------------ 1832| 1.99k| __r_.first().__s.__size_ = __s; 1833| 1.99k| __r_.first().__s.__is_long_ = false; 1834| 1.99k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE19__get_short_pointerB8ne180100Ev: 1873| 3.59k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 pointer __get_short_pointer() _NOEXCEPT { 1874| 3.59k| return pointer_traits::pointer_to(__r_.first().__s.__data_[0]); 1875| 3.59k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE7__allocB8ne180100Ev: 1826| 20.1k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 allocator_type& __alloc() _NOEXCEPT { return __r_.second(); } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE10__align_itB8ne180100ILm8EEEmm: 1936| 19.7k| static _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 size_type __align_it(size_type __s) _NOEXCEPT { 1937| 19.7k| return (__s + (__a - 1)) & ~(__a - 1); 1938| 19.7k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE11__recommendB8ne180100Em: 1942| 19.7k| static _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 size_type __recommend(size_type __s) _NOEXCEPT { 1943| 19.7k| if (__s < __min_cap) { ------------------ | Branch (1943:9): [True: 0, False: 19.7k] ------------------ 1944| 0| return static_cast(__min_cap) - 1; 1945| 0| } 1946| 19.7k| const size_type __boundary = sizeof(value_type) < __alignment ? __alignment / sizeof(value_type) : __endian_factor; ------------------ | Branch (1946:34): [True: 0, Folded] ------------------ 1947| 19.7k| size_type __guess = __align_it<__boundary>(__s + 1) - 1; 1948| 19.7k| if (__guess == __min_cap) ------------------ | Branch (1948:9): [True: 0, False: 19.7k] ------------------ 1949| 0| __guess += __endian_factor; 1950| 19.7k| return __guess; 1951| 19.7k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE16__begin_lifetimeB8ne180100EPcm: 1776| 19.7k| static _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 void __begin_lifetime(pointer __begin, size_type __n) { 1777| 19.7k|#if _LIBCPP_STD_VER >= 20 1778| 19.7k| if (__libcpp_is_constant_evaluated()) { ------------------ | Branch (1778:9): [Folded, False: 19.7k] ------------------ 1779| 0| for (size_type __i = 0; __i != __n; ++__i) ------------------ | Branch (1779:31): [True: 0, False: 0] ------------------ 1780| 0| std::construct_at(std::addressof(__begin[__i])); 1781| 0| } 1782| |#else 1783| | (void)__begin; 1784| | (void)__n; 1785| |#endif // _LIBCPP_STD_VER >= 20 1786| 19.7k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE18__set_long_pointerB8ne180100EPc: 1864| 19.7k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 void __set_long_pointer(pointer __p) _NOEXCEPT { 1865| 19.7k| __r_.first().__l.__data_ = __p; 1866| 19.7k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE14__set_long_capB8ne180100Em: 1855| 19.7k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 void __set_long_cap(size_type __s) _NOEXCEPT { 1856| 19.7k| __r_.first().__l.__cap_ = __s / __endian_factor; 1857| 19.7k| __r_.first().__l.__is_long_ = true; 1858| 19.7k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE15__set_long_sizeB8ne180100Em: 1842| 20.6k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 void __set_long_size(size_type __s) _NOEXCEPT { 1843| 20.6k| __r_.first().__l.__size_ = __s; 1844| 20.6k| } _ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE14__annotate_newB8ne180100Em: 1904| 21.5k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 void __annotate_new(size_type __current_size) const _NOEXCEPT { 1905| 21.5k| (void) __current_size; 1906| |#if !defined(_LIBCPP_HAS_NO_ASAN) && defined(_LIBCPP_INSTRUMENTED_WITH_ASAN) 1907| | if (!__libcpp_is_constant_evaluated() && (__asan_short_string_is_annotated() || __is_long())) 1908| | __annotate_contiguous_container(data() + capacity() + 1, data() + __current_size + 1); 1909| |#endif 1910| 21.5k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE6__initEPKcm: 2202| 19.2k|basic_string<_CharT, _Traits, _Allocator>::__init(const value_type* __s, size_type __sz) { 2203| 19.2k| if (__libcpp_is_constant_evaluated()) ------------------ | Branch (2203:7): [Folded, False: 19.2k] ------------------ 2204| 0| __r_.first() = __rep(); 2205| 19.2k| if (__sz > max_size()) ------------------ | Branch (2205:7): [True: 0, False: 19.2k] ------------------ 2206| 0| __throw_length_error(); 2207| 19.2k| pointer __p; 2208| 19.2k| if (__fits_in_sso(__sz)) { ------------------ | Branch (2208:7): [True: 1, False: 19.2k] ------------------ 2209| 1| __set_short_size(__sz); 2210| 1| __p = __get_short_pointer(); 2211| 19.2k| } else { 2212| 19.2k| auto __allocation = std::__allocate_at_least(__alloc(), __recommend(__sz) + 1); 2213| 19.2k| __p = __allocation.ptr; 2214| 19.2k| __begin_lifetime(__p, __allocation.count); 2215| 19.2k| __set_long_pointer(__p); 2216| 19.2k| __set_long_cap(__allocation.count); 2217| 19.2k| __set_long_size(__sz); 2218| 19.2k| } 2219| 19.2k| traits_type::copy(std::__to_address(__p), __s, __sz); 2220| 19.2k| traits_type::assign(__p[__sz], value_type()); 2221| 19.2k| __annotate_new(__sz); 2222| 19.2k|} _ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE17__annotate_deleteB8ne180100Ev: 1912| 2.35k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 void __annotate_delete() const _NOEXCEPT { 1913| |#if !defined(_LIBCPP_HAS_NO_ASAN) && defined(_LIBCPP_INSTRUMENTED_WITH_ASAN) 1914| | if (!__libcpp_is_constant_evaluated() && (__asan_short_string_is_annotated() || __is_long())) 1915| | __annotate_contiguous_container(data() + size() + 1, data() + capacity() + 1); 1916| |#endif 1917| 2.35k| } _ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE9__is_longB8ne180100Ev: 1769| 500k| __is_long() const _NOEXCEPT { 1770| 500k| if (__libcpp_is_constant_evaluated() && __builtin_constant_p(__r_.first().__l.__is_long_)) { ------------------ | Branch (1770:9): [Folded, False: 500k] | Branch (1770:45): [Folded, False: 0] ------------------ 1771| 0| return __r_.first().__l.__is_long_; 1772| 0| } 1773| 500k| return __r_.first().__s.__is_long_; 1774| 500k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE18__get_long_pointerB8ne180100Ev: 1867| 1.79k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 pointer __get_long_pointer() _NOEXCEPT { 1868| 1.79k| return __r_.first().__l.__data_; 1869| 1.79k| } _ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE14__get_long_capB8ne180100Ev: 1860| 1.31k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 size_type __get_long_cap() const _NOEXCEPT { 1861| 1.31k| return __r_.first().__l.__cap_ * __endian_factor; 1862| 1.31k| } _ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE8capacityB8ne180100Ev: 1191| 5.56k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 size_type capacity() const _NOEXCEPT { 1192| 5.56k| return (__is_long() ? __get_long_cap() : static_cast(__min_cap)) - 1; ------------------ | Branch (1192:13): [True: 1.19k, False: 4.36k] ------------------ 1193| 5.56k| } _ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE15__get_long_sizeB8ne180100Ev: 1845| 233k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 size_type __get_long_size() const _NOEXCEPT { 1846| 233k| return __r_.first().__l.__size_; 1847| 233k| } _ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE16__get_short_sizeB8ne180100Ev: 1837| 251k| __get_short_size() const _NOEXCEPT { 1838| 251k| _LIBCPP_ASSERT_INTERNAL(!__r_.first().__s.__is_long_, "String has to be short when trying to get the short size"); ------------------ | | 417| 251k|# define _LIBCPP_ASSERT_INTERNAL(expression, message) _LIBCPP_ASSUME(expression) | | ------------------ | | | | 34| 251k|# define _LIBCPP_ASSUME(expression) ((void)0) | | ------------------ ------------------ 1839| 251k| return __r_.first().__s.__size_; 1840| 251k| } _ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE4sizeB8ne180100Ev: 1176| 483k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 size_type size() const _NOEXCEPT { 1177| 483k| return __is_long() ? __get_long_size() : __get_short_size(); ------------------ | Branch (1177:12): [True: 233k, False: 250k] ------------------ 1178| 483k| } _ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE19__annotate_increaseB8ne180100Em: 1919| 2.86k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 void __annotate_increase(size_type __n) const _NOEXCEPT { 1920| 2.86k| (void) __n; 1921| |#if !defined(_LIBCPP_HAS_NO_ASAN) && defined(_LIBCPP_INSTRUMENTED_WITH_ASAN) 1922| | if (!__libcpp_is_constant_evaluated() && (__asan_short_string_is_annotated() || __is_long())) 1923| | __annotate_contiguous_container(data() + size() + 1, data() + size() + 1 + __n); 1924| |#endif 1925| 2.86k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE13__get_pointerB8ne180100Ev: 1879| 4.83k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 pointer __get_pointer() _NOEXCEPT { 1880| 4.83k| return __is_long() ? __get_long_pointer() : __get_short_pointer(); ------------------ | Branch (1880:12): [True: 1.62k, False: 3.21k] ------------------ 1881| 4.83k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE10__set_sizeB8ne180100Em: 1848| 2.42k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 void __set_size(size_type __s) _NOEXCEPT { 1849| 2.42k| if (__is_long()) ------------------ | Branch (1849:9): [True: 816, False: 1.60k] ------------------ 1850| 816| __set_long_size(__s); 1851| 1.60k| else 1852| 1.60k| __set_short_size(__s); 1853| 2.42k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE21__grow_by_and_replaceEmmmmmmPKc: 2357| 69| const value_type* __p_new_stuff) { 2358| 69| size_type __ms = max_size(); 2359| 69| if (__delta_cap > __ms - __old_cap - 1) ------------------ | Branch (2359:7): [True: 0, False: 69] ------------------ 2360| 0| __throw_length_error(); 2361| 69| pointer __old_p = __get_pointer(); 2362| 69| size_type __cap = 2363| 69| __old_cap < __ms / 2 - __alignment ? __recommend(std::max(__old_cap + __delta_cap, 2 * __old_cap)) : __ms - 1; ------------------ | Branch (2363:7): [True: 69, False: 0] ------------------ 2364| 69| __annotate_delete(); 2365| 69| auto __allocation = std::__allocate_at_least(__alloc(), __cap + 1); 2366| 69| pointer __p = __allocation.ptr; 2367| 69| __begin_lifetime(__p, __allocation.count); 2368| 69| if (__n_copy != 0) ------------------ | Branch (2368:7): [True: 69, False: 0] ------------------ 2369| 69| traits_type::copy(std::__to_address(__p), std::__to_address(__old_p), __n_copy); 2370| 69| if (__n_add != 0) ------------------ | Branch (2370:7): [True: 69, False: 0] ------------------ 2371| 69| traits_type::copy(std::__to_address(__p) + __n_copy, __p_new_stuff, __n_add); 2372| 69| size_type __sec_cp_sz = __old_sz - __n_del - __n_copy; 2373| 69| if (__sec_cp_sz != 0) ------------------ | Branch (2373:7): [True: 0, False: 69] ------------------ 2374| 0| traits_type::copy( 2375| 0| std::__to_address(__p) + __n_copy + __n_add, std::__to_address(__old_p) + __n_copy + __n_del, __sec_cp_sz); 2376| 69| if (__old_cap + 1 != __min_cap) ------------------ | Branch (2376:7): [True: 66, False: 3] ------------------ 2377| 66| __alloc_traits::deallocate(__alloc(), __old_p, __old_cap + 1); 2378| 69| __set_long_pointer(__p); 2379| 69| __set_long_cap(__allocation.count); 2380| 69| __old_sz = __n_copy + __n_add + __sec_cp_sz; 2381| 69| __set_long_size(__old_sz); 2382| 69| traits_type::assign(__p[__old_sz], value_type()); 2383| 69| __annotate_new(__old_cap + __delta_cap); 2384| 69|} _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE6appendEPKcm: 2711| 1.95k|basic_string<_CharT, _Traits, _Allocator>::append(const value_type* __s, size_type __n) { 2712| 1.95k| _LIBCPP_ASSERT_NON_NULL(__n == 0 || __s != nullptr, "string::append received nullptr"); ------------------ | | 409| 1.95k|# define _LIBCPP_ASSERT_NON_NULL(expression, message) _LIBCPP_ASSUME(expression) | | ------------------ | | | | 34| 1.95k|# define _LIBCPP_ASSUME(expression) ((void)0) | | ------------------ ------------------ 2713| 1.95k| size_type __cap = capacity(); 2714| 1.95k| size_type __sz = size(); 2715| 1.95k| if (__cap - __sz >= __n) { ------------------ | Branch (2715:7): [True: 1.88k, False: 69] ------------------ 2716| 1.88k| if (__n) { ------------------ | Branch (2716:9): [True: 294, False: 1.59k] ------------------ 2717| 294| __annotate_increase(__n); 2718| 294| value_type* __p = std::__to_address(__get_pointer()); 2719| 294| traits_type::copy(__p + __sz, __s, __n); 2720| 294| __sz += __n; 2721| 294| __set_size(__sz); 2722| 294| traits_type::assign(__p[__sz], value_type()); 2723| 294| } 2724| 1.88k| } else 2725| 69| __grow_by_and_replace(__cap, __sz + __n - __cap, __sz, __sz, 0, __n, __s); 2726| 1.95k| return *this; 2727| 1.95k|} _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE9__grow_byEmmmmmm: 2400| 430| size_type __n_add) { 2401| 430| size_type __ms = max_size(); 2402| 430| if (__delta_cap > __ms - __old_cap) ------------------ | Branch (2402:7): [True: 0, False: 430] ------------------ 2403| 0| __throw_length_error(); 2404| 430| pointer __old_p = __get_pointer(); 2405| 430| size_type __cap = 2406| 430| __old_cap < __ms / 2 - __alignment ? __recommend(std::max(__old_cap + __delta_cap, 2 * __old_cap)) : __ms - 1; ------------------ | Branch (2406:7): [True: 430, False: 0] ------------------ 2407| 430| __annotate_delete(); 2408| 430| auto __allocation = std::__allocate_at_least(__alloc(), __cap + 1); 2409| 430| pointer __p = __allocation.ptr; 2410| 430| __begin_lifetime(__p, __allocation.count); 2411| 430| if (__n_copy != 0) ------------------ | Branch (2411:7): [True: 430, False: 0] ------------------ 2412| 430| traits_type::copy(std::__to_address(__p), std::__to_address(__old_p), __n_copy); 2413| 430| size_type __sec_cp_sz = __old_sz - __n_del - __n_copy; 2414| 430| if (__sec_cp_sz != 0) ------------------ | Branch (2414:7): [True: 0, False: 430] ------------------ 2415| 0| traits_type::copy( 2416| 0| std::__to_address(__p) + __n_copy + __n_add, std::__to_address(__old_p) + __n_copy + __n_del, __sec_cp_sz); 2417| 430| if (__old_cap + 1 != __min_cap) ------------------ | Branch (2417:7): [True: 253, False: 177] ------------------ 2418| 253| __alloc_traits::deallocate(__alloc(), __old_p, __old_cap + 1); 2419| 430| __set_long_pointer(__p); 2420| 430| __set_long_cap(__allocation.count); 2421| 430|} _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE25__grow_by_without_replaceB8ne180100Emmmmmm: 2431| 430| size_type __n_add) { 2432| 430| _LIBCPP_SUPPRESS_DEPRECATED_PUSH 2433| 430| __grow_by(__old_cap, __delta_cap, __old_sz, __n_copy, __n_del, __n_add); 2434| 430| _LIBCPP_SUPPRESS_DEPRECATED_POP 2435| 430| __set_long_size(__old_sz - __n_del + __n_add); 2436| 430| __annotate_new(__old_sz - __n_del + __n_add); 2437| 430|} _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE6appendEmc: 2731| 1.84k|basic_string<_CharT, _Traits, _Allocator>::append(size_type __n, value_type __c) { 2732| 1.84k| if (__n) { ------------------ | Branch (2732:7): [True: 1.84k, False: 0] ------------------ 2733| 1.84k| size_type __cap = capacity(); 2734| 1.84k| size_type __sz = size(); 2735| 1.84k| if (__cap - __sz < __n) ------------------ | Branch (2735:9): [True: 375, False: 1.47k] ------------------ 2736| 375| __grow_by_without_replace(__cap, __sz + __n - __cap, __sz, __sz, 0); 2737| 1.84k| __annotate_increase(__n); 2738| 1.84k| pointer __p = __get_pointer(); 2739| 1.84k| traits_type::assign(std::__to_address(__p) + __sz, __n, __c); 2740| 1.84k| __sz += __n; 2741| 1.84k| __set_size(__sz); 2742| 1.84k| traits_type::assign(__p[__sz], value_type()); 2743| 1.84k| } 2744| 1.84k| return *this; 2745| 1.84k|} _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE6resizeEmc: 3179| 1.84k|_LIBCPP_CONSTEXPR_SINCE_CXX20 void basic_string<_CharT, _Traits, _Allocator>::resize(size_type __n, value_type __c) { 3180| 1.84k| size_type __sz = size(); 3181| 1.84k| if (__n > __sz) ------------------ | Branch (3181:7): [True: 1.84k, False: 0] ------------------ 3182| 1.84k| append(__n - __sz, __c); 3183| 0| else 3184| 0| __erase_to_end(__n); 3185| 1.84k|} _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE6resizeB8ne180100Em: 1196| 1.84k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 void resize(size_type __n) { resize(__n, value_type()); } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEixB8ne180100Em: 1228| 1.03k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 reference operator[](size_type __pos) _NOEXCEPT { 1229| 1.03k| _LIBCPP_ASSERT_VALID_ELEMENT_ACCESS(__pos <= size(), "string index out of bounds"); ------------------ | | 408| 1.03k|# define _LIBCPP_ASSERT_VALID_ELEMENT_ACCESS(expression, message) _LIBCPP_ASSUME(expression) | | ------------------ | | | | 34| 1.03k|# define _LIBCPP_ASSUME(expression) ((void)0) | | ------------------ ------------------ 1230| 1.03k| if (__builtin_constant_p(__pos) && !__fits_in_sso(__pos)) { ------------------ | Branch (1230:9): [True: 1.03k, False: 0] | Branch (1230:40): [True: 0, False: 1.03k] ------------------ 1231| 0| return *(__get_long_pointer() + __pos); 1232| 0| } 1233| 1.03k| return *(__get_pointer() + __pos); 1234| 1.03k| } _ZNKSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE5emptyB8ne180100Ev: 1216| 492| _LIBCPP_NODISCARD_AFTER_CXX17 _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 bool empty() const _NOEXCEPT { 1217| 492| return size() == 0; 1218| 492| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE9push_backEc: 2764| 438|_LIBCPP_CONSTEXPR_SINCE_CXX20 void basic_string<_CharT, _Traits, _Allocator>::push_back(value_type __c) { 2765| 438| bool __is_short = !__is_long(); 2766| 438| size_type __cap; 2767| 438| size_type __sz; 2768| 438| if (__is_short) { ------------------ | Branch (2768:7): [True: 438, False: 0] ------------------ 2769| 438| __cap = __min_cap - 1; 2770| 438| __sz = __get_short_size(); 2771| 438| } else { 2772| 0| __cap = __get_long_cap() - 1; 2773| 0| __sz = __get_long_size(); 2774| 0| } 2775| 438| if (__sz == __cap) { ------------------ | Branch (2775:7): [True: 55, False: 383] ------------------ 2776| 55| __grow_by_without_replace(__cap, 1, __sz, __sz, 0); 2777| 55| __annotate_increase(1); 2778| 55| __is_short = false; // the string is always long after __grow_by 2779| 55| } else 2780| 383| __annotate_increase(1); 2781| 438| pointer __p = __get_pointer(); 2782| 438| if (__is_short) { ------------------ | Branch (2782:7): [True: 383, False: 55] ------------------ 2783| 383| __p = __get_short_pointer() + __sz; 2784| 383| __set_short_size(__sz + 1); 2785| 383| } else { 2786| 55| __p = __get_long_pointer() + __sz; 2787| 55| __set_long_size(__sz + 1); 2788| 55| } 2789| 438| traits_type::assign(*__p, __c); 2790| 438| traits_type::assign(*++__p, value_type()); 2791| 438|} _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE4dataB8ne180100Ev: 1600| 438| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX20 value_type* data() _NOEXCEPT { 1601| 438| return std::__to_address(__get_pointer()); 1602| 438| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEC2ERKS5_: 897| 1.15k| : __r_(__default_init_tag(), __alloc_traits::select_on_container_copy_construction(__str.__alloc())) { 898| 1.15k| if (!__str.__is_long()) { ------------------ | Branch (898:9): [True: 1.15k, False: 0] ------------------ 899| 1.15k| __r_.first() = __str.__r_.first(); 900| 1.15k| __annotate_new(__get_short_size()); 901| 1.15k| } else 902| 0| __init_copy_ctor_external(std::__to_address(__str.__get_long_pointer()), __str.__get_long_size()); 903| 1.15k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEED2Ev: 1102| 1.85k| inline _LIBCPP_CONSTEXPR_SINCE_CXX20 ~basic_string() { 1103| 1.85k| __annotate_delete(); 1104| 1.85k| if (__is_long()) ------------------ | Branch (1104:9): [True: 122, False: 1.73k] ------------------ 1105| 122| __alloc_traits::deallocate(__alloc(), __get_long_pointer(), __get_long_cap()); 1106| 1.85k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEE6insertEmPKcm: 2852| 280|basic_string<_CharT, _Traits, _Allocator>::insert(size_type __pos, const value_type* __s, size_type __n) { 2853| 280| _LIBCPP_ASSERT_NON_NULL(__n == 0 || __s != nullptr, "string::insert received nullptr"); ------------------ | | 409| 280|# define _LIBCPP_ASSERT_NON_NULL(expression, message) _LIBCPP_ASSUME(expression) | | ------------------ | | | | 34| 280|# define _LIBCPP_ASSUME(expression) ((void)0) | | ------------------ ------------------ 2854| 280| size_type __sz = size(); 2855| 280| if (__pos > __sz) ------------------ | Branch (2855:7): [True: 0, False: 280] ------------------ 2856| 0| __throw_out_of_range(); 2857| 280| size_type __cap = capacity(); 2858| 280| if (__cap - __sz >= __n) { ------------------ | Branch (2858:7): [True: 280, False: 0] ------------------ 2859| 280| if (__n) { ------------------ | Branch (2859:9): [True: 280, False: 0] ------------------ 2860| 280| __annotate_increase(__n); 2861| 280| value_type* __p = std::__to_address(__get_pointer()); 2862| 280| size_type __n_move = __sz - __pos; 2863| 280| if (__n_move != 0) { ------------------ | Branch (2863:11): [True: 280, False: 0] ------------------ 2864| 280| if (std::__is_pointer_in_range(__p + __pos, __p + __sz, __s)) ------------------ | Branch (2864:13): [True: 0, False: 280] ------------------ 2865| 0| __s += __n; 2866| 280| traits_type::move(__p + __pos + __n, __p + __pos, __n_move); 2867| 280| } 2868| 280| traits_type::move(__p + __pos, __s, __n); 2869| 280| __sz += __n; 2870| 280| __set_size(__sz); 2871| 280| traits_type::assign(__p[__sz], value_type()); 2872| 280| } 2873| 280| } else 2874| 0| __grow_by_and_replace(__cap, __sz + __n - __cap, __sz, __pos, 0, __n, __s); 2875| 280| return *this; 2876| 280|} _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEC2B8ne180100ILi0EEEPKc: 952| 19.2k| : __r_(__default_init_tag(), __default_init_tag()) { 953| 19.2k| _LIBCPP_ASSERT_NON_NULL(__s != nullptr, "basic_string(const char*) detected nullptr"); ------------------ | | 409| 19.2k|# define _LIBCPP_ASSERT_NON_NULL(expression, message) _LIBCPP_ASSUME(expression) | | ------------------ | | | | 34| 19.2k|# define _LIBCPP_ASSUME(expression) ((void)0) | | ------------------ ------------------ 954| 19.2k| __init(__s, traits_type::length(__s)); 955| 19.2k| } _ZNSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEC2B8ne180100Ev: 882| 663| : __r_(__value_init_tag(), __default_init_tag()) { 883| 663| __annotate_new(0); 884| 663| } locale.cpp:_ZNSt3__15tupleIJOZNS_6locale2id5__getEvE3$_0EEC2B8ne180100IJS3_ETnNS_9enable_ifIXsr4_AndINS_17integral_constantIbXeqsZT_sZT_EEENS5_17_EnableUTypesCtorIJDpT_EEEEE5valueEiE4typeELi0EEEDpOSB_: 620| 30| : __base_(typename __make_tuple_indices::type(), 621| 30| typename __make_tuple_types::type(), 622| 30| typename __make_tuple_indices::type(), 623| 30| typename __make_tuple_types::type(), 624| 30| std::forward<_Up>(__u)...) {} locale.cpp:_ZNSt3__112__tuple_implINS_15__tuple_indicesIJLm0EEEEJOZNS_6locale2id5__getEvE3$_0EEC2B8ne180100IJLm0EEJS6_ETpTnmJEJEJS5_EEENS1_IJXspT_EEEENS_13__tuple_typesIJDpT0_EEENS1_IJXspT1_EEEENSA_IJDpT2_EEEDpOT3_: 473| 30| : __tuple_leaf<_Uf, _Tf>(std::forward<_Up>(__u))..., __tuple_leaf<_Ul, _Tl>()... {} locale.cpp:_ZNSt3__112__tuple_leafILm0EOZNS_6locale2id5__getEvE3$_0Lb0EEC2B8ne180100IS3_vEEOT_: 345| 30| : __value_(std::forward<_Tp>(__t)) { 346| 30| static_assert(__can_bind_reference<_Tp&&>(), 347| 30| "Attempted construction of reference element binds to a temporary whose lifetime has ended"); 348| 30| } locale.cpp:_ZNSt3__13getB8ne180100ILm0EJOZNS_6locale2id5__getEvE3$_0EEEONS_13tuple_elementIXT_ENS_5tupleIJDpT0_EEEE4typeEOS9_: 1084| 30|get(tuple<_Tp...>&& __t) _NOEXCEPT { 1085| 30| typedef _LIBCPP_NODEBUG typename tuple_element<_Ip, tuple<_Tp...> >::type type; 1086| 30| return static_cast(static_cast<__tuple_leaf<_Ip, type>&&>(__t.__base_).get()); 1087| 30|} locale.cpp:_ZNSt3__112__tuple_leafILm0EOZNS_6locale2id5__getEvE3$_0Lb0EE3getB8ne180100Ev: 387| 30| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 _Hp& get() _NOEXCEPT { return __value_; } _ZNSt27__type_info_implementations13__unique_impl4__eqB8ne180100EPKcS2_: 203| 910k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_ALWAYS_INLINE static bool __eq(__type_name_t __lhs, __type_name_t __rhs) _NOEXCEPT { 204| 910k| return __lhs == __rhs; 205| 910k| } _ZNKSt9type_infoeqB8ne180100ERKS_: 302| 910k| _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX23 bool operator==(const type_info& __arg) const _NOEXCEPT { 303| | // When evaluated in a constant expression, both type infos simply can't come 304| | // from different translation units, so it is sufficient to compare their addresses. 305| 910k| if (__libcpp_is_constant_evaluated()) { ------------------ | Branch (305:9): [Folded, False: 910k] ------------------ 306| 0| return this == &__arg; 307| 0| } 308| 910k| return __impl::__eq(__type_name, __arg.__type_name); 309| 910k| } _ZNSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEEC2Em: 1118| 1|_LIBCPP_CONSTEXPR_SINCE_CXX20 vector<_Tp, _Allocator>::vector(size_type __n) { 1119| 1| auto __guard = std::__make_exception_guard(__destroy_vector(*this)); 1120| 1| if (__n > 0) { ------------------ | Branch (1120:7): [True: 1, False: 0] ------------------ 1121| 1| __vallocate(__n); 1122| 1| __construct_at_end(__n); 1123| 1| } 1124| 1| __guard.__complete(); 1125| 1|} _ZNSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE16__destroy_vectorC2B8ne180100ERS6_: 486| 1| _LIBCPP_CONSTEXPR _LIBCPP_HIDE_FROM_ABI __destroy_vector(vector& __vec) : __vec_(__vec) {} _ZNSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE11__vallocateB8ne180100Em: 738| 1| _LIBCPP_CONSTEXPR_SINCE_CXX20 _LIBCPP_HIDE_FROM_ABI void __vallocate(size_type __n) { 739| 1| if (__n > max_size()) ------------------ | Branch (739:9): [True: 0, False: 1] ------------------ 740| 0| __throw_length_error(); 741| 1| auto __allocation = std::__allocate_at_least(__alloc(), __n); 742| 1| __begin_ = __allocation.ptr; 743| 1| __end_ = __allocation.ptr; 744| 1| __end_cap() = __begin_ + __allocation.count; 745| 1| __annotate_new(0); 746| 1| } _ZNKSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE8max_sizeEv: 1030| 1|vector<_Tp, _Allocator>::max_size() const _NOEXCEPT { 1031| 1| return std::min(__alloc_traits::max_size(this->__alloc()), numeric_limits::max()); 1032| 1|} _ZNKSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE7__allocB8ne180100Ev: 909| 1| _LIBCPP_CONSTEXPR_SINCE_CXX20 _LIBCPP_HIDE_FROM_ABI const allocator_type& __alloc() const _NOEXCEPT { 910| 1| return this->__end_cap_.second(); 911| 1| } _ZNSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE7__allocB8ne180100Ev: 906| 91| _LIBCPP_CONSTEXPR_SINCE_CXX20 _LIBCPP_HIDE_FROM_ABI allocator_type& __alloc() _NOEXCEPT { 907| 91| return this->__end_cap_.second(); 908| 91| } _ZNSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE9__end_capB8ne180100Ev: 912| 31| _LIBCPP_CONSTEXPR_SINCE_CXX20 _LIBCPP_HIDE_FROM_ABI pointer& __end_cap() _NOEXCEPT { 913| 31| return this->__end_cap_.first(); 914| 31| } _ZNKSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE14__annotate_newB8ne180100Em: 847| 1| _LIBCPP_CONSTEXPR_SINCE_CXX20 _LIBCPP_HIDE_FROM_ABI void __annotate_new(size_type __current_size) const _NOEXCEPT { 848| 1| (void)__current_size; 849| |#ifndef _LIBCPP_HAS_NO_ASAN 850| | __annotate_contiguous_container(data() + capacity(), data() + __current_size); 851| |#endif 852| 1| } _ZNSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE18__construct_at_endEm: 1053| 31|_LIBCPP_CONSTEXPR_SINCE_CXX20 void vector<_Tp, _Allocator>::__construct_at_end(size_type __n) { 1054| 31| _ConstructTransaction __tx(*this, __n); 1055| 31| const_pointer __new_end = __tx.__new_end_; 1056| 91| for (pointer __pos = __tx.__pos_; __pos != __new_end; __tx.__pos_ = ++__pos) { ------------------ | Branch (1056:37): [True: 60, False: 31] ------------------ 1057| 60| __alloc_traits::construct(this->__alloc(), std::__to_address(__pos)); 1058| 60| } 1059| 31|} _ZNSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE21_ConstructTransactionC2B8ne180100ERS6_m: 876| 31| : __v_(__v), __pos_(__v.__end_), __new_end_(__v.__end_ + __n) { 877| |#ifndef _LIBCPP_HAS_NO_ASAN 878| | __v_.__annotate_increase(__n); 879| |#endif 880| 31| } _ZNSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE21_ConstructTransactionD2B8ne180100Ev: 881| 31| _LIBCPP_CONSTEXPR_SINCE_CXX20 _LIBCPP_HIDE_FROM_ABI ~_ConstructTransaction() { 882| 31| __v_.__end_ = __pos_; 883| |#ifndef _LIBCPP_HAS_NO_ASAN 884| | if (__pos_ != __new_end_) { 885| | __v_.__annotate_shrink(__new_end_ - __v_.__begin_); 886| | } 887| |#endif 888| 31| } _ZNSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE7__clearB8ne180100Ev: 919| 1| _LIBCPP_CONSTEXPR_SINCE_CXX20 _LIBCPP_HIDE_FROM_ABI void __clear() _NOEXCEPT { 920| 1| __base_destruct_at_end(this->__begin_); 921| 1| } _ZNSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE22__base_destruct_at_endB8ne180100EPS3_: 923| 1| _LIBCPP_CONSTEXPR_SINCE_CXX20 _LIBCPP_HIDE_FROM_ABI void __base_destruct_at_end(pointer __new_last) _NOEXCEPT { 924| 1| pointer __soon_to_be_end = this->__end_; 925| 31| while (__new_last != __soon_to_be_end) ------------------ | Branch (925:12): [True: 30, False: 1] ------------------ 926| 30| __alloc_traits::destroy(__alloc(), std::__to_address(--__soon_to_be_end)); 927| 1| this->__end_ = __new_last; 928| 1| } _ZNSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE5clearB8ne180100Ev: 707| 1| _LIBCPP_CONSTEXPR_SINCE_CXX20 _LIBCPP_HIDE_FROM_ABI void clear() _NOEXCEPT { 708| 1| size_type __old_size = size(); 709| 1| __clear(); 710| 1| __annotate_shrink(__old_size); 711| 1| } _ZNKSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE17__annotate_shrinkB8ne180100Em: 867| 1| _LIBCPP_CONSTEXPR_SINCE_CXX20 _LIBCPP_HIDE_FROM_ABI void __annotate_shrink(size_type __old_size) const _NOEXCEPT { 868| 1| (void)__old_size; 869| |#ifndef _LIBCPP_HAS_NO_ASAN 870| | __annotate_contiguous_container(data() + __old_size, data() + size()); 871| |#endif 872| 1| } _ZNKSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE4sizeB8ne180100Ev: 594| 4.57k| _LIBCPP_CONSTEXPR_SINCE_CXX20 _LIBCPP_HIDE_FROM_ABI size_type size() const _NOEXCEPT { 595| 4.57k| return static_cast(this->__end_ - this->__begin_); 596| 4.57k| } _ZNSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEEixB8ne180100Em: 1392| 60|vector<_Tp, _Allocator>::operator[](size_type __n) _NOEXCEPT { 1393| 60| _LIBCPP_ASSERT_VALID_ELEMENT_ACCESS(__n < size(), "vector[] index out of bounds"); ------------------ | | 408| 60|# define _LIBCPP_ASSERT_VALID_ELEMENT_ACCESS(expression, message) _LIBCPP_ASSUME(expression) | | ------------------ | | | | 34| 60|# define _LIBCPP_ASSUME(expression) ((void)0) | | ------------------ ------------------ 1394| 60| return this->__begin_[__n]; 1395| 60|} _ZNSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE6resizeEm: 1747| 30|_LIBCPP_CONSTEXPR_SINCE_CXX20 void vector<_Tp, _Allocator>::resize(size_type __sz) { 1748| 30| size_type __cs = size(); 1749| 30| if (__cs < __sz) ------------------ | Branch (1749:7): [True: 30, False: 0] ------------------ 1750| 30| this->__append(__sz - __cs); 1751| 0| else if (__cs > __sz) ------------------ | Branch (1751:12): [True: 0, False: 0] ------------------ 1752| 0| this->__destruct_at_end(this->__begin_ + __sz); 1753| 30|} _ZNSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEE8__appendEm: 1090| 30|_LIBCPP_CONSTEXPR_SINCE_CXX20 void vector<_Tp, _Allocator>::__append(size_type __n) { 1091| 30| if (static_cast(this->__end_cap() - this->__end_) >= __n) ------------------ | Branch (1091:7): [True: 30, False: 0] ------------------ 1092| 30| this->__construct_at_end(__n); 1093| 0| else { 1094| 0| allocator_type& __a = this->__alloc(); 1095| 0| __split_buffer __v(__recommend(size() + __n), size(), __a); 1096| 0| __v.__construct_at_end(__n); 1097| 0| __swap_out_circular_buffer(__v); 1098| 0| } 1099| 30|} _ZNKSt3__16vectorIPNS_6locale5facetENS_15__sso_allocatorIS3_Lm30EEEEixB8ne180100Em: 1399| 9.01k|vector<_Tp, _Allocator>::operator[](size_type __n) const _NOEXCEPT { 1400| 9.01k| _LIBCPP_ASSERT_VALID_ELEMENT_ACCESS(__n < size(), "vector[] index out of bounds"); ------------------ | | 408| 9.01k|# define _LIBCPP_ASSERT_VALID_ELEMENT_ACCESS(expression, message) _LIBCPP_ASSUME(expression) | | ------------------ | | | | 34| 9.01k|# define _LIBCPP_ASSUME(expression) ((void)0) | | ------------------ ------------------ 1401| 9.01k| return this->__begin_[__n]; 1402| 9.01k|} _ZNSt3__111__call_onceERVmPvPFvS2_E: 31| 30|void __call_once(volatile once_flag::_State_type& flag, void* arg, void (*func)(void*)) { 32| |#if defined(_LIBCPP_HAS_NO_THREADS) 33| | 34| | if (flag == once_flag::_Unset) { 35| | auto guard = std::__make_exception_guard([&flag] { flag = once_flag::_Unset; }); 36| | flag = once_flag::_Pending; 37| | func(arg); 38| | flag = once_flag::_Complete; 39| | guard.__complete(); 40| | } 41| | 42| |#else // !_LIBCPP_HAS_NO_THREADS 43| | 44| 30| __libcpp_mutex_lock(&mut); 45| 30| while (flag == once_flag::_Pending) ------------------ | Branch (45:10): [True: 0, False: 30] ------------------ 46| 0| __libcpp_condvar_wait(&cv, &mut); 47| 30| if (flag == once_flag::_Unset) { ------------------ | Branch (47:7): [True: 30, False: 0] ------------------ 48| 30| auto guard = std::__make_exception_guard([&flag] { 49| 30| __libcpp_mutex_lock(&mut); 50| 30| __libcpp_relaxed_store(&flag, once_flag::_Unset); 51| 30| __libcpp_mutex_unlock(&mut); 52| 30| __libcpp_condvar_broadcast(&cv); 53| 30| }); 54| | 55| 30| __libcpp_relaxed_store(&flag, once_flag::_Pending); 56| 30| __libcpp_mutex_unlock(&mut); 57| 30| func(arg); 58| 30| __libcpp_mutex_lock(&mut); 59| 30| __libcpp_atomic_store(&flag, once_flag::_Complete, _AO_Release); 60| 30| __libcpp_mutex_unlock(&mut); 61| 30| __libcpp_condvar_broadcast(&cv); 62| 30| guard.__complete(); 63| 30| } else { 64| 0| __libcpp_mutex_unlock(&mut); 65| 0| } 66| | 67| 30|#endif // !_LIBCPP_HAS_NO_THREADS 68| 30|} cxa_handlers.cpp:_ZNSt3__112_GLOBAL__N_120__libcpp_atomic_loadB8ne180100IPFvvEEET_PKS4_i: 58| 344k|inline _LIBCPP_HIDE_FROM_ABI _ValueType __libcpp_atomic_load(_ValueType const* __val, int __order = _AO_Seq) { 59| 344k| return __atomic_load_n(__val, __order); 60| 344k|} cxa_exception.cpp:_ZNSt3__112_GLOBAL__N_119__libcpp_atomic_addB8ne180100ImmEET_PS2_T0_i: 63| 172k|inline _LIBCPP_HIDE_FROM_ABI _ValueType __libcpp_atomic_add(_ValueType* __val, _AddType __a, int __order = _AO_Seq) { 64| 172k| return __atomic_add_fetch(__val, __a, __order); 65| 172k|} memory.cpp:_ZNSt3__112_GLOBAL__N_120__libcpp_atomic_loadB8ne180100IlEET_PKS2_i: 58| 15.4k|inline _LIBCPP_HIDE_FROM_ABI _ValueType __libcpp_atomic_load(_ValueType const* __val, int __order = _AO_Seq) { 59| 15.4k| return __atomic_load_n(__val, __order); 60| 15.4k|} stdlib_stdexcept.cpp:_ZNSt3__112_GLOBAL__N_119__libcpp_atomic_addB8ne180100IiiEET_PS2_T0_i: 63| 172k|inline _LIBCPP_HIDE_FROM_ABI _ValueType __libcpp_atomic_add(_ValueType* __val, _AddType __a, int __order = _AO_Seq) { 64| 172k| return __atomic_add_fetch(__val, __a, __order); 65| 172k|} cxa_guard.cpp:_ZNSt3__112_GLOBAL__N_120__libcpp_atomic_loadB8ne180100IhEET_PKS2_i: 58| 3|inline _LIBCPP_HIDE_FROM_ABI _ValueType __libcpp_atomic_load(_ValueType const* __val, int __order = _AO_Seq) { 59| 3| return __atomic_load_n(__val, __order); 60| 3|} cxa_guard.cpp:_ZNSt3__112_GLOBAL__N_121__libcpp_atomic_storeB8ne180100IhhEEvPT_T0_i: 48| 3|inline _LIBCPP_HIDE_FROM_ABI void __libcpp_atomic_store(_ValueType* __dest, _FromType __val, int __order = _AO_Seq) { 49| 3| __atomic_store_n(__dest, __val, __order); 50| 3|} locale.cpp:_ZNSt3__112_GLOBAL__N_119__libcpp_atomic_addB8ne180100IiiEET_PS2_T0_i: 63| 30|inline _LIBCPP_HIDE_FROM_ABI _ValueType __libcpp_atomic_add(_ValueType* __val, _AddType __a, int __order = _AO_Seq) { 64| 30| return __atomic_add_fetch(__val, __a, __order); 65| 30|} call_once.cpp:_ZNSt3__112_GLOBAL__N_122__libcpp_relaxed_storeB8ne180100IVmmEEvPT_T0_: 53| 30|inline _LIBCPP_HIDE_FROM_ABI void __libcpp_relaxed_store(_ValueType* __dest, _FromType __val) { 54| 30| __atomic_store_n(__dest, __val, _AO_Relaxed); 55| 30|} call_once.cpp:_ZNSt3__112_GLOBAL__N_121__libcpp_atomic_storeB8ne180100IVmmEEvPT_T0_i: 48| 30|inline _LIBCPP_HIDE_FROM_ABI void __libcpp_atomic_store(_ValueType* __dest, _FromType __val, int __order = _AO_Seq) { 49| 30| __atomic_store_n(__dest, __val, __order); 50| 30|} _ZNSt3__118__libcpp_refstringD2Ev: 108| 172k|inline __libcpp_refstring::~__libcpp_refstring() { 109| 172k| if (__uses_refcount()) { ------------------ | Branch (109:7): [True: 172k, False: 0] ------------------ 110| 172k| _Rep_base* rep = rep_from_data(__imp_); 111| 172k| if (__libcpp_atomic_add(&rep->count, count_t(-1)) < 0) { ------------------ | Branch (111:9): [True: 172k, False: 3] ------------------ 112| 172k| ::operator delete(rep); 113| 172k| } 114| 172k| } 115| 172k|} _ZNKSt3__118__libcpp_refstring15__uses_refcountEv: 117| 172k|inline bool __libcpp_refstring::__uses_refcount() const { 118| |#if defined(_LIBCPP_CHECK_FOR_GCC_EMPTY_STRING_STORAGE) 119| | return __imp_ != get_gcc_empty_string_storage(); 120| |#else 121| 172k| return true; 122| 172k|#endif 123| 172k|} stdlib_stdexcept.cpp:_ZNSt3__115__refstring_imp12_GLOBAL__N_113rep_from_dataEPKc: 46| 172k|inline _Rep_base* rep_from_data(const char* data_) noexcept { 47| 172k| char* data = const_cast(data_); 48| 172k| return reinterpret_cast<_Rep_base*>(data - sizeof(_Rep_base)); 49| 172k|} _ZNSt3__118__libcpp_refstringC2EPKc: 78| 172k|inline __libcpp_refstring::__libcpp_refstring(const char* msg) { 79| 172k| std::size_t len = strlen(msg); 80| 172k| _Rep_base* rep = static_cast<_Rep_base*>(::operator new(sizeof(*rep) + len + 1)); 81| 172k| rep->len = len; 82| 172k| rep->cap = len; 83| 172k| rep->count = 0; 84| 172k| char* data = data_from_rep(rep); 85| 172k| std::memcpy(data, msg, len + 1); 86| 172k| __imp_ = data; 87| 172k|} stdexcept.cpp:_ZNSt3__115__refstring_imp12_GLOBAL__N_113data_from_repEPNS1_9_Rep_baseE: 51| 172k|inline char* data_from_rep(_Rep_base* rep) noexcept { 52| 172k| char* data = reinterpret_cast(rep); 53| 172k| return data + sizeof(*rep); 54| 172k|} _ZNSt3__115__sso_allocatorIPNS_6locale5facetELm30EEC2B8ne180100Ev: 50| 1| _LIBCPP_HIDE_FROM_ABI __sso_allocator() throw() : __allocated_(false) {} _ZNKSt3__115__sso_allocatorIPNS_6locale5facetELm30EE8max_sizeB8ne180100Ev: 72| 1| _LIBCPP_HIDE_FROM_ABI size_type max_size() const throw() { return size_type(~0) / sizeof(_Tp); } _ZNSt3__115__sso_allocatorIPNS_6locale5facetELm30EE8allocateB8ne180100EmPKv: 59| 1| _LIBCPP_HIDE_FROM_ABI pointer allocate(size_type __n, typename __sso_allocator::const_pointer = nullptr) { 60| 1| if (!__allocated_ && __n <= _Np) { ------------------ | Branch (60:9): [True: 1, False: 0] | Branch (60:26): [True: 1, False: 0] ------------------ 61| 1| __allocated_ = true; 62| 1| return (pointer)&buf_; 63| 1| } 64| 0| return allocator<_Tp>().allocate(__n); 65| 1| } _ZNSt3__18ios_base16__call_callbacksENS0_5eventE: 95| 723|void ios_base::__call_callbacks(event ev) { 96| 723| for (size_t i = __event_size_; i;) { ------------------ | Branch (96:34): [True: 0, False: 723] ------------------ 97| 0| --i; 98| 0| __fn_[i](ev, *this, __index_[i]); 99| 0| } 100| 723|} _ZNKSt3__18ios_base6getlocEv: 113| 3.35k|locale ios_base::getloc() const { 114| 3.35k| const locale& loc_storage = *reinterpret_cast(&__loc_); 115| 3.35k| return loc_storage; 116| 3.35k|} _ZNSt3__18ios_baseD2Ev: 197| 723|ios_base::~ios_base() { 198| 723| __call_callbacks(erase_event); 199| 723| locale& loc_storage = *reinterpret_cast(&__loc_); 200| 723| loc_storage.~locale(); 201| 723| free(__fn_); 202| 723| free(__index_); 203| 723| free(__iarray_); 204| 723| free(__parray_); 205| 723|} _ZNSt3__18ios_base5clearEj: 209| 1.08k|void ios_base::clear(iostate state) { 210| 1.08k| if (__rdbuf_) ------------------ | Branch (210:7): [True: 1.08k, False: 0] ------------------ 211| 1.08k| __rdstate_ = state; 212| 0| else 213| 0| __rdstate_ = state | badbit; 214| | 215| 1.08k| if (((state | (__rdbuf_ ? goodbit : badbit)) & __exceptions_) != 0) ------------------ | Branch (215:7): [True: 0, False: 1.08k] | Branch (215:18): [True: 1.08k, False: 0] ------------------ 216| 0| __throw_failure("ios_base::clear"); 217| 1.08k|} _ZNSt3__18ios_base4initEPv: 221| 723|void ios_base::init(void* sb) { 222| 723| __rdbuf_ = sb; 223| 723| __rdstate_ = __rdbuf_ ? goodbit : badbit; ------------------ | Branch (223:20): [True: 723, False: 0] ------------------ 224| 723| __exceptions_ = goodbit; 225| 723| __fmtflags_ = skipws | dec; 226| 723| __width_ = 0; 227| 723| __precision_ = 6; 228| 723| __fn_ = 0; 229| 723| __index_ = 0; 230| 723| __event_size_ = 0; 231| 723| __event_cap_ = 0; 232| 723| __iarray_ = 0; 233| 723| __iarray_size_ = 0; 234| 723| __iarray_cap_ = 0; 235| 723| __parray_ = 0; 236| 723| __parray_size_ = 0; 237| 723| __parray_cap_ = 0; 238| 723| ::new (&__loc_) locale; 239| 723|} _ZNSt3__16__clocEv: 75| 1.07k|locale_t __cloc() { 76| | // In theory this could create a race condition. In practice 77| | // the race condition is non-fatal since it will just create 78| | // a little resource leak. Better approach would be appreciated. 79| | static locale_t result = newlocale(LC_ALL_MASK, "C", 0); 80| 1.07k| return result; 81| 1.07k|} _ZNSt3__16locale5__impC2Em: 161| 1|locale::__imp::__imp(size_t refs) : facet(refs), facets_(N), name_("C") { 162| 1| facets_.clear(); 163| 1| install(&make >(1u)); 164| 1|#ifndef _LIBCPP_HAS_NO_WIDE_CHARACTERS 165| 1| install(&make >(1u)); 166| 1|#endif 167| 1| install(&make >(nullptr, false, 1u)); 168| 1|#ifndef _LIBCPP_HAS_NO_WIDE_CHARACTERS 169| 1| install(&make >(1u)); 170| 1|#endif 171| 1| install(&make >(1u)); 172| 1|#ifndef _LIBCPP_HAS_NO_WIDE_CHARACTERS 173| 1| install(&make >(1u)); 174| 1|#endif 175| 1| _LIBCPP_SUPPRESS_DEPRECATED_PUSH 176| 1| install(&make >(1u)); 177| 1| install(&make >(1u)); 178| 1| _LIBCPP_SUPPRESS_DEPRECATED_POP 179| 1|#ifndef _LIBCPP_HAS_NO_CHAR8_T 180| 1| install(&make >(1u)); 181| 1| install(&make >(1u)); 182| 1|#endif 183| 1| install(&make >(1u)); 184| 1|#ifndef _LIBCPP_HAS_NO_WIDE_CHARACTERS 185| 1| install(&make >(1u)); 186| 1|#endif 187| 1| install(&make >(1u)); 188| 1|#ifndef _LIBCPP_HAS_NO_WIDE_CHARACTERS 189| 1| install(&make >(1u)); 190| 1|#endif 191| 1| install(&make >(1u)); 192| 1|#ifndef _LIBCPP_HAS_NO_WIDE_CHARACTERS 193| 1| install(&make >(1u)); 194| 1|#endif 195| 1| install(&make >(1u)); 196| 1| install(&make >(1u)); 197| 1|#ifndef _LIBCPP_HAS_NO_WIDE_CHARACTERS 198| 1| install(&make >(1u)); 199| 1| install(&make >(1u)); 200| 1|#endif 201| 1| install(&make >(1u)); 202| 1|#ifndef _LIBCPP_HAS_NO_WIDE_CHARACTERS 203| 1| install(&make >(1u)); 204| 1|#endif 205| 1| install(&make >(1u)); 206| 1|#ifndef _LIBCPP_HAS_NO_WIDE_CHARACTERS 207| 1| install(&make >(1u)); 208| 1|#endif 209| 1| install(&make >(1u)); 210| 1|#ifndef _LIBCPP_HAS_NO_WIDE_CHARACTERS 211| 1| install(&make >(1u)); 212| 1|#endif 213| 1| install(&make >(1u)); 214| 1|#ifndef _LIBCPP_HAS_NO_WIDE_CHARACTERS 215| 1| install(&make >(1u)); 216| 1|#endif 217| 1| install(&make >(1u)); 218| 1|#ifndef _LIBCPP_HAS_NO_WIDE_CHARACTERS 219| 1| install(&make >(1u)); 220| 1|#endif 221| 1|} _ZNSt3__16locale5__imp7installEPNS0_5facetEl: 475| 30|void locale::__imp::install(facet* f, long id) { 476| 30| f->__add_shared(); 477| 30| unique_ptr hold(f); 478| 30| if (static_cast(id) >= facets_.size()) ------------------ | Branch (478:7): [True: 30, False: 0] ------------------ 479| 30| facets_.resize(static_cast(id + 1)); 480| 30| if (facets_[static_cast(id)]) ------------------ | Branch (480:7): [True: 0, False: 30] ------------------ 481| 0| facets_[static_cast(id)]->__release_shared(); 482| 30| facets_[static_cast(id)] = hold.release(); 483| 30|} _ZNKSt3__16locale5__imp9use_facetEl: 485| 4.50k|const locale::facet* locale::__imp::use_facet(long id) const { 486| 4.50k| if (!has_facet(id)) ------------------ | Branch (486:7): [True: 0, False: 4.50k] ------------------ 487| 0| __throw_bad_cast(); 488| 4.50k| return facets_[static_cast(id)]; 489| 4.50k|} _ZNSt3__16locale7classicEv: 501| 1|const locale& locale::classic() { 502| 1| static const __no_destroy classic_locale(__private_tag{}, [] { 503| | // executed exactly once on first initialization of `classic_locale` 504| 1| locale::__imp::classic_locale_imp_.__emplace(1u); 505| 1| return &locale::__imp::classic_locale_imp_.__get(); 506| 1| }()); 507| 1| return classic_locale.__get(); 508| 1|} _ZNSt3__16locale8__globalEv: 510| 1.44k|locale& locale::__global() { 511| 1.44k| static __no_destroy g(locale::classic()); 512| 1.44k| return g.__get(); 513| 1.44k|} _ZNSt3__16locale5__imp7acquireEv: 515| 4.80k|void locale::__imp::acquire() { 516| 4.80k| if (this != &locale::__imp::classic_locale_imp_.__get()) ------------------ | Branch (516:7): [True: 0, False: 4.80k] ------------------ 517| 0| __add_shared(); 518| 4.80k|} _ZNSt3__16locale5__imp7releaseEv: 520| 4.80k|void locale::__imp::release() { 521| 4.80k| if (this != &locale::__imp::classic_locale_imp_.__get()) ------------------ | Branch (521:7): [True: 0, False: 4.80k] ------------------ 522| 0| __release_shared(); 523| 4.80k|} _ZNSt3__16localeC2Ev: 525| 1.44k|locale::locale() noexcept : __locale_(__global().__locale_) { __locale_->acquire(); } _ZNSt3__16localeC2ERKS0_: 527| 3.35k|locale::locale(const locale& l) noexcept : __locale_(l.__locale_) { __locale_->acquire(); } _ZNSt3__16localeD2Ev: 529| 4.80k|locale::~locale() { __locale_->release(); } _ZNKSt3__16locale9use_facetERNS0_2idE: 581| 4.50k|const locale::facet* locale::use_facet(id& x) const { return __locale_->use_facet(x.__get()); } _ZNSt3__16locale2id5__getEv: 597| 4.53k|long locale::id::__get() { 598| 4.53k| call_once(__flag_, [&] { __id_ = __libcpp_atomic_add(&__next_id, 1); }); 599| 4.53k| return __id_ - 1; 600| 4.53k|} _ZNSt3__15ctypeIcEC2EPKtbm: 809| 1|ctype::ctype(const mask* tab, bool del, size_t refs) : locale::facet(refs), __tab_(tab), __del_(del) { 810| 1| if (__tab_ == 0) ------------------ | Branch (810:7): [True: 1, False: 0] ------------------ 811| 1| __tab_ = classic_table(); 812| 1|} _ZNKSt3__15ctypeIcE8do_widenEc: 871| 383|char ctype::do_widen(char c) const { return c; } _ZNKSt3__15ctypeIcE8do_widenEPKcS3_Pc: 873| 1.15k|const char* ctype::do_widen(const char* low, const char* high, char_type* dest) const { 874| 19.2k| for (; low != high; ++low, ++dest) ------------------ | Branch (874:10): [True: 18.1k, False: 1.15k] ------------------ 875| 18.1k| *dest = *low; 876| 1.15k| return low; 877| 1.15k|} _ZNSt3__15ctypeIcE13classic_tableEv: 981| 1|const ctype::mask* ctype::classic_table() noexcept { 982| |# if defined(__APPLE__) || defined(__FreeBSD__) 983| | return _DefaultRuneLocale.__runetype; 984| |# elif defined(__NetBSD__) 985| | return _C_ctype_tab_ + 1; 986| |# elif defined(__GLIBC__) 987| 1| return _LIBCPP_GET_C_LOCALE->__ctype_b; ------------------ | | 246| 1|# define _LIBCPP_GET_C_LOCALE __cloc() ------------------ 988| |# elif defined(_LIBCPP_MSVCRT) || defined(__MINGW32__) 989| | return __pctype_func(); 990| |# elif defined(__EMSCRIPTEN__) 991| | return *__ctype_b_loc(); 992| |# elif defined(_NEWLIB_VERSION) 993| | // Newlib has a 257-entry table in ctype_.c, where (char)0 starts at [1]. 994| | return _ctype_ + 1; 995| |# elif defined(_AIX) 996| | return (const unsigned int*)__lc_ctype_ptr->obj->mask; 997| |# elif defined(__MVS__) 998| |# if defined(__NATIVE_ASCII_F) 999| | return const_cast::mask*>(__OBJ_DATA(__lc_ctype_a)->mask); 1000| |# else 1001| | return const_cast::mask*>(__ctypec); 1002| |# endif 1003| |# else 1004| | // Platform not supported: abort so the person doing the port knows what to 1005| | // fix 1006| |# warning ctype::classic_table() is not implemented 1007| | printf("ctype::classic_table() is not implemented\n"); 1008| | abort(); 1009| | return NULL; 1010| |# endif 1011| 1|} _ZNSt3__17codecvtIwc11__mbstate_tEC2Em: 1339| 1|codecvt::codecvt(size_t refs) : locale::facet(refs), __l_(_LIBCPP_GET_C_LOCALE) {} ------------------ | | 246| 1|# define _LIBCPP_GET_C_LOCALE __cloc() ------------------ _ZNSt3__18numpunctIcEC2Em: 4011| 1|numpunct::numpunct(size_t refs) : locale::facet(refs), __decimal_point_('.'), __thousands_sep_(',') {} _ZNSt3__18numpunctIwEC2Em: 4014| 1|numpunct::numpunct(size_t refs) : locale::facet(refs), __decimal_point_(L'.'), __thousands_sep_(L',') {} _ZNKSt3__18numpunctIcE16do_thousands_sepEv: 4028| 661|char numpunct< char >::do_thousands_sep() const { return __thousands_sep_; } _ZNKSt3__18numpunctIcE11do_groupingEv: 4033| 1.15k|string numpunct< char >::do_grouping() const { return __grouping_; } _ZNSt3__114__num_get_base10__get_baseERNS_8ios_baseE: 4109| 661|int __num_get_base::__get_base(ios_base& iob) { 4110| 661| ios_base::fmtflags __basefield = iob.flags() & ios_base::basefield; 4111| 661| if (__basefield == ios_base::oct) ------------------ | Branch (4111:7): [True: 0, False: 661] ------------------ 4112| 0| return 8; 4113| 661| else if (__basefield == ios_base::hex) ------------------ | Branch (4113:12): [True: 0, False: 661] ------------------ 4114| 0| return 16; 4115| 661| else if (__basefield == 0) ------------------ | Branch (4115:12): [True: 0, False: 661] ------------------ 4116| 0| return 0; 4117| 661| return 10; 4118| 661|} _ZNSt3__116__check_groupingERKNS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEEPjS8_Rj: 4122| 661|void __check_grouping(const string& __grouping, unsigned* __g, unsigned* __g_end, ios_base::iostate& __err) { 4123| | // if the grouping pattern is empty _or_ there are no grouping bits, then do nothing 4124| | // we always have at least a single entry in [__g, __g_end); the end of the input sequence 4125| 661| if (__grouping.size() != 0 && __g_end - __g > 1) { ------------------ | Branch (4125:7): [True: 0, False: 661] | Branch (4125:33): [True: 0, False: 0] ------------------ 4126| 0| reverse(__g, __g_end); 4127| 0| const char* __ig = __grouping.data(); 4128| 0| const char* __eg = __ig + __grouping.size(); 4129| 0| for (unsigned* __r = __g; __r < __g_end - 1; ++__r) { ------------------ | Branch (4129:31): [True: 0, False: 0] ------------------ 4130| 0| if (0 < *__ig && *__ig < numeric_limits::max()) { ------------------ | Branch (4130:11): [True: 0, False: 0] | Branch (4130:24): [True: 0, False: 0] ------------------ 4131| 0| if (static_cast(*__ig) != *__r) { ------------------ | Branch (4131:13): [True: 0, False: 0] ------------------ 4132| 0| __err = ios_base::failbit; 4133| 0| return; 4134| 0| } 4135| 0| } 4136| 0| if (__eg - __ig > 1) ------------------ | Branch (4136:11): [True: 0, False: 0] ------------------ 4137| 0| ++__ig; 4138| 0| } 4139| 0| if (0 < *__ig && *__ig < numeric_limits::max()) { ------------------ | Branch (4139:9): [True: 0, False: 0] | Branch (4139:22): [True: 0, False: 0] ------------------ 4140| 0| if (static_cast(*__ig) < __g_end[-1] || __g_end[-1] == 0) ------------------ | Branch (4140:11): [True: 0, False: 0] | Branch (4140:57): [True: 0, False: 0] ------------------ 4141| 0| __err = ios_base::failbit; 4142| 0| } 4143| 0| } 4144| 661|} _ZNSt3__114__num_put_base12__format_intEPcPKcbj: 4146| 492|void __num_put_base::__format_int(char* __fmtp, const char* __len, bool __signd, ios_base::fmtflags __flags) { 4147| 492| if ((__flags & ios_base::showpos) && (__flags & ios_base::basefield) != ios_base::oct && ------------------ | Branch (4147:7): [True: 0, False: 492] | Branch (4147:40): [True: 0, False: 0] ------------------ 4148| 0| (__flags & ios_base::basefield) != ios_base::hex && __signd) ------------------ | Branch (4148:7): [True: 0, False: 0] | Branch (4148:59): [True: 0, False: 0] ------------------ 4149| 0| *__fmtp++ = '+'; 4150| 492| if (__flags & ios_base::showbase) ------------------ | Branch (4150:7): [True: 0, False: 492] ------------------ 4151| 0| *__fmtp++ = '#'; 4152| 984| while (*__len) ------------------ | Branch (4152:10): [True: 492, False: 492] ------------------ 4153| 492| *__fmtp++ = *__len++; 4154| 492| if ((__flags & ios_base::basefield) == ios_base::oct) ------------------ | Branch (4154:7): [True: 0, False: 492] ------------------ 4155| 0| *__fmtp = 'o'; 4156| 492| else if ((__flags & ios_base::basefield) == ios_base::hex) { ------------------ | Branch (4156:12): [True: 0, False: 492] ------------------ 4157| 0| if (__flags & ios_base::uppercase) ------------------ | Branch (4157:9): [True: 0, False: 0] ------------------ 4158| 0| *__fmtp = 'X'; 4159| 0| else 4160| 0| *__fmtp = 'x'; 4161| 492| } else if (__signd) ------------------ | Branch (4161:14): [True: 0, False: 492] ------------------ 4162| 0| *__fmtp = 'd'; 4163| 492| else 4164| 492| *__fmtp = 'u'; 4165| 492|} _ZNSt3__114__num_put_base18__identify_paddingEPcS1_RKNS_8ios_baseE: 4207| 492|char* __num_put_base::__identify_padding(char* __nb, char* __ne, const ios_base& __iob) { 4208| 492| switch (__iob.flags() & ios_base::adjustfield) { 4209| 0| case ios_base::internal: ------------------ | Branch (4209:3): [True: 0, False: 492] ------------------ 4210| 0| if (__nb[0] == '-' || __nb[0] == '+') ------------------ | Branch (4210:9): [True: 0, False: 0] | Branch (4210:27): [True: 0, False: 0] ------------------ 4211| 0| return __nb + 1; 4212| 0| if (__ne - __nb >= 2 && __nb[0] == '0' && (__nb[1] == 'x' || __nb[1] == 'X')) ------------------ | Branch (4212:9): [True: 0, False: 0] | Branch (4212:29): [True: 0, False: 0] | Branch (4212:48): [True: 0, False: 0] | Branch (4212:66): [True: 0, False: 0] ------------------ 4213| 0| return __nb + 2; 4214| 0| break; 4215| 0| case ios_base::left: ------------------ | Branch (4215:3): [True: 0, False: 492] ------------------ 4216| 0| return __ne; 4217| 0| case ios_base::right: ------------------ | Branch (4217:3): [True: 0, False: 492] ------------------ 4218| 492| default: ------------------ | Branch (4218:3): [True: 492, False: 0] ------------------ 4219| 492| break; 4220| 492| } 4221| 492| return __nb; 4222| 492|} _ZNKSt3__16locale5__imp9has_facetEl: 144| 4.50k| bool has_facet(long id) const { return static_cast(id) < facets_.size() && facets_[static_cast(id)]; } ------------------ | Branch (144:42): [True: 4.50k, False: 0] | Branch (144:86): [True: 4.50k, False: 0] ------------------ locale.cpp:_ZZNSt3__16locale7classicEvENK3$_0clEv: 502| 1| static const __no_destroy classic_locale(__private_tag{}, [] { 503| | // executed exactly once on first initialization of `classic_locale` 504| 1| locale::__imp::classic_locale_imp_.__emplace(1u); 505| 1| return &locale::__imp::classic_locale_imp_.__get(); 506| 1| }()); locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_7collateIcEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_7collateIcEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_7collateIwEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_7collateIwEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_5ctypeIcEEJDnbjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_5ctypeIcEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_5ctypeIwEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_5ctypeIwEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_7codecvtIcc11__mbstate_tEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_7codecvtIcc11__mbstate_tEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_7codecvtIwc11__mbstate_tEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_7codecvtIwc11__mbstate_tEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_7codecvtIDsc11__mbstate_tEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_7codecvtIDsc11__mbstate_tEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_7codecvtIDic11__mbstate_tEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_7codecvtIDic11__mbstate_tEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_7codecvtIDsDu11__mbstate_tEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_7codecvtIDsDu11__mbstate_tEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_7codecvtIDiDu11__mbstate_tEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_7codecvtIDiDu11__mbstate_tEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_8numpunctIcEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_8numpunctIcEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_8numpunctIwEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_8numpunctIwEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_7num_getIcNS_19istreambuf_iteratorIcNS_11char_traitsIcEEEEEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_7num_getIcNS_19istreambuf_iteratorIcNS_11char_traitsIcEEEEEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_7num_getIwNS_19istreambuf_iteratorIwNS_11char_traitsIwEEEEEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_7num_getIwNS_19istreambuf_iteratorIwNS_11char_traitsIwEEEEEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_7num_putIcNS_19ostreambuf_iteratorIcNS_11char_traitsIcEEEEEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_7num_putIcNS_19ostreambuf_iteratorIcNS_11char_traitsIcEEEEEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_7num_putIwNS_19ostreambuf_iteratorIwNS_11char_traitsIwEEEEEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_7num_putIwNS_19ostreambuf_iteratorIwNS_11char_traitsIwEEEEEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_10moneypunctIcLb0EEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_10moneypunctIcLb0EEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_10moneypunctIcLb1EEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_10moneypunctIcLb1EEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_10moneypunctIwLb0EEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_10moneypunctIwLb0EEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_10moneypunctIwLb1EEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_10moneypunctIwLb1EEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_9money_getIcNS_19istreambuf_iteratorIcNS_11char_traitsIcEEEEEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_9money_getIcNS_19istreambuf_iteratorIcNS_11char_traitsIcEEEEEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_9money_getIwNS_19istreambuf_iteratorIwNS_11char_traitsIwEEEEEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_9money_getIwNS_19istreambuf_iteratorIwNS_11char_traitsIwEEEEEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_9money_putIcNS_19ostreambuf_iteratorIcNS_11char_traitsIcEEEEEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_9money_putIcNS_19ostreambuf_iteratorIcNS_11char_traitsIcEEEEEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_9money_putIwNS_19ostreambuf_iteratorIwNS_11char_traitsIwEEEEEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_9money_putIwNS_19ostreambuf_iteratorIwNS_11char_traitsIwEEEEEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_8time_getIcNS_19istreambuf_iteratorIcNS_11char_traitsIcEEEEEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_8time_getIcNS_19istreambuf_iteratorIcNS_11char_traitsIcEEEEEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_8time_getIwNS_19istreambuf_iteratorIwNS_11char_traitsIwEEEEEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_8time_getIwNS_19istreambuf_iteratorIwNS_11char_traitsIwEEEEEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_8time_putIcNS_19ostreambuf_iteratorIcNS_11char_traitsIcEEEEEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_8time_putIcNS_19ostreambuf_iteratorIcNS_11char_traitsIcEEEEEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_8time_putIwNS_19ostreambuf_iteratorIwNS_11char_traitsIwEEEEEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_8time_putIwNS_19ostreambuf_iteratorIwNS_11char_traitsIwEEEEEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_8messagesIcEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_8messagesIcEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZNSt3__112_GLOBAL__N_14makeINS_8messagesIwEEJjEEERT_DpT0_: 91| 1|T& make(Args... args) { 92| 1| alignas(T) static std::byte buf[sizeof(T)]; 93| 1| auto* obj = ::new (&buf) T(args...); 94| 1| return *obj; 95| 1|} _ZNSt3__16locale5__imp7installINS_8messagesIwEEEEvPT_: 154| 1| void install(F* f) { 155| 1| install(f, f->id.__get()); 156| 1| } locale.cpp:_ZZNSt3__16locale2id5__getEvENK3$_0clEv: 598| 30| call_once(__flag_, [&] { __id_ = __libcpp_atomic_add(&__next_id, 1); }); _ZNSt3__119__shared_weak_count14__release_weakEv: 57| 15.4k|void __shared_weak_count::__release_weak() noexcept { 58| | // NOTE: The acquire load here is an optimization of the very 59| | // common case where a shared pointer is being destructed while 60| | // having no other contended references. 61| | // 62| | // BENEFIT: We avoid expensive atomic stores like XADD and STREX 63| | // in a common case. Those instructions are slow and do nasty 64| | // things to caches. 65| | // 66| | // IS THIS SAFE? Yes. During weak destruction, if we see that we 67| | // are the last reference, we know that no-one else is accessing 68| | // us. If someone were accessing us, then they would be doing so 69| | // while the last shared / weak_ptr was being destructed, and 70| | // that's undefined anyway. 71| | // 72| | // If we see anything other than a 0, then we have possible 73| | // contention, and need to use an atomicrmw primitive. 74| | // The same arguments don't apply for increment, where it is legal 75| | // (though inadvisable) to share shared_ptr references between 76| | // threads, and have them all get copied at once. The argument 77| | // also doesn't apply for __release_shared, because an outstanding 78| | // weak_ptr::lock() could read / modify the shared count. 79| 15.4k| if (__libcpp_atomic_load(&__shared_weak_owners_, _AO_Acquire) == 0) { ------------------ | Branch (79:7): [True: 15.4k, False: 0] ------------------ 80| | // no need to do this store, because we are about 81| | // to destroy everything. 82| | //__libcpp_atomic_store(&__shared_weak_owners_, -1, _AO_Release); 83| 15.4k| __on_zero_shared_weak(); 84| 15.4k| } else if (__libcpp_atomic_refcount_decrement(__shared_weak_owners_) == -1) ------------------ | Branch (84:14): [True: 0, False: 0] ------------------ 85| 0| __on_zero_shared_weak(); 86| 15.4k|} _ZSt19uncaught_exceptionsv: 18| 4.78k|int uncaught_exceptions() noexcept { 19| 4.78k|#if _LIBCPPABI_VERSION > 1001 20| 4.78k| return __cxa_uncaught_exceptions(); 21| |#else 22| | return __cxa_uncaught_exception() ? 1 : 0; 23| |#endif 24| 4.78k|} _ZNSt13runtime_errorC2EPKc: 34| 172k|runtime_error::runtime_error(const char* msg) : __imp_(msg) {} _ZN10__cxxabiv119__setExceptionClassEP17_Unwind_Exceptionm: 81| 172k|void __setExceptionClass(_Unwind_Exception* unwind_exception, uint64_t newValue) { 82| 172k| ::memcpy(&unwind_exception->exception_class, &newValue, sizeof(newValue)); 83| 172k|} _ZN10__cxxabiv119__getExceptionClassEPK17_Unwind_Exception: 95| 750k|uint64_t __getExceptionClass(const _Unwind_Exception* unwind_exception) { 96| | // On x86 and some ARM unwinders, unwind_exception->exception_class is 97| | // a uint64_t. On other ARM unwinders, it is a char[8]. 98| | // See: http://infocenter.arm.com/help/topic/com.arm.doc.ihi0038b/IHI0038B_ehabi.pdf 99| | // So we just copy it into a uint64_t to be sure. 100| 750k| uint64_t exClass; 101| 750k| ::memcpy(&exClass, &unwind_exception->exception_class, sizeof(exClass)); 102| 750k| return exClass; 103| 750k|} _ZN10__cxxabiv121__isOurExceptionClassEPK17_Unwind_Exception: 105| 344k|bool __isOurExceptionClass(const _Unwind_Exception* unwind_exception) { 106| 344k| return (__getExceptionClass(unwind_exception) & get_vendor_and_language) == 107| 344k| (kOurExceptionClass & get_vendor_and_language); 108| 344k|} __cxa_allocate_exception: 183| 172k|void *__cxa_allocate_exception(size_t thrown_size) throw() { 184| 172k| size_t actual_size = cxa_exception_size_from_exception_thrown_size(thrown_size); 185| | 186| | // Allocate extra space before the __cxa_exception header to ensure the 187| | // start of the thrown object is sufficiently aligned. 188| 172k| size_t header_offset = get_cxa_exception_offset(); 189| 172k| char *raw_buffer = 190| 172k| (char *)__aligned_malloc_with_fallback(header_offset + actual_size); 191| 172k| if (NULL == raw_buffer) ------------------ | Branch (191:9): [True: 0, False: 172k] ------------------ 192| 0| std::terminate(); 193| 172k| __cxa_exception *exception_header = 194| 172k| static_cast<__cxa_exception *>((void *)(raw_buffer + header_offset)); 195| 172k| ::memset(exception_header, 0, actual_size); 196| 172k| return thrown_object_from_cxa_exception(exception_header); 197| 172k|} __cxa_free_exception: 201| 172k|void __cxa_free_exception(void *thrown_object) throw() { 202| | // Compute the size of the padding before the header. 203| 172k| size_t header_offset = get_cxa_exception_offset(); 204| 172k| char *raw_buffer = 205| 172k| ((char *)cxa_exception_from_thrown_object(thrown_object)) - header_offset; 206| 172k| __aligned_free_with_fallback((void *)raw_buffer); 207| 172k|} __cxa_init_primary_exception: 210| 172k| void(_LIBCXXABI_DTOR_FUNC* dest)(void*)) throw() { 211| 172k| __cxa_exception* exception_header = cxa_exception_from_thrown_object(object); 212| 172k| exception_header->referenceCount = 0; 213| 172k| exception_header->unexpectedHandler = std::get_unexpected(); 214| 172k| exception_header->terminateHandler = std::get_terminate(); 215| 172k| exception_header->exceptionType = tinfo; 216| 172k| exception_header->exceptionDestructor = dest; 217| 172k| setOurExceptionClass(&exception_header->unwindHeader); 218| 172k| exception_header->unwindHeader.exception_cleanup = exception_cleanup_func; 219| | 220| 172k| return exception_header; 221| 172k|} __cxa_throw: 274| 172k|__cxa_throw(void *thrown_object, std::type_info *tinfo, void (_LIBCXXABI_DTOR_FUNC *dest)(void *)) { 275| 172k|#endif 276| 172k| __cxa_eh_globals* globals = __cxa_get_globals(); 277| 172k| globals->uncaughtExceptions += 1; // Not atomically, since globals are thread-local 278| | 279| 172k| __cxa_exception* exception_header = __cxa_init_primary_exception(thrown_object, tinfo, dest); 280| 172k| exception_header->referenceCount = 1; // This is a newly allocated exception, no need for thread safety. 281| | 282| |#if __has_feature(address_sanitizer) 283| | // Inform the ASan runtime that now might be a good time to clean stuff up. 284| | __asan_handle_no_return(); 285| |#endif 286| | 287| |#ifdef __USING_SJLJ_EXCEPTIONS__ 288| | _Unwind_SjLj_RaiseException(&exception_header->unwindHeader); 289| |#else 290| 172k| _Unwind_RaiseException(&exception_header->unwindHeader); 291| 172k|#endif 292| | // This only happens when there is no handler, or some unexpected unwinding 293| | // error happens. 294| 172k| failed_throw(exception_header); 295| 172k|} __cxa_begin_catch: 446| 172k|{ 447| 172k| _Unwind_Exception* unwind_exception = static_cast<_Unwind_Exception*>(unwind_arg); 448| 172k| bool native_exception = __isOurExceptionClass(unwind_exception); 449| 172k| __cxa_eh_globals* globals = __cxa_get_globals(); 450| | // exception_header is a hackish offset from a foreign exception, but it 451| | // works as long as we're careful not to try to access any __cxa_exception 452| | // parts. 453| 172k| __cxa_exception* exception_header = 454| 172k| cxa_exception_from_exception_unwind_exception 455| 172k| ( 456| 172k| static_cast<_Unwind_Exception*>(unwind_exception) 457| 172k| ); 458| | 459| |#if defined(__MVS__) 460| | // Remove the exception object from the linked list of exceptions that the z/OS unwinder 461| | // maintains before adding it to the libc++abi list of caught exceptions. 462| | // The libc++abi will manage the lifetime of the exception from this point forward. 463| | _UnwindZOS_PopException(); 464| |#endif 465| | 466| 172k| if (native_exception) ------------------ | Branch (466:9): [True: 172k, False: 1] ------------------ 467| 172k| { 468| | // Increment the handler count, removing the flag about being rethrown 469| 172k| exception_header->handlerCount = exception_header->handlerCount < 0 ? ------------------ | Branch (469:42): [True: 0, False: 172k] ------------------ 470| 172k| -exception_header->handlerCount + 1 : exception_header->handlerCount + 1; 471| | // place the exception on the top of the stack if it's not already 472| | // there by a previous rethrow 473| 172k| if (exception_header != globals->caughtExceptions) ------------------ | Branch (473:13): [True: 172k, False: 0] ------------------ 474| 172k| { 475| 172k| exception_header->nextException = globals->caughtExceptions; 476| 172k| globals->caughtExceptions = exception_header; 477| 172k| } 478| 172k| globals->uncaughtExceptions -= 1; // Not atomically, since globals are thread-local 479| |#if defined(_LIBCXXABI_ARM_EHABI) 480| | return reinterpret_cast(exception_header->unwindHeader.barrier_cache.bitpattern[0]); 481| |#else 482| 172k| return exception_header->adjustedPtr; 483| 172k|#endif 484| 172k| } 485| | // Else this is a foreign exception 486| | // If the caughtExceptions stack is not empty, terminate 487| 1| if (globals->caughtExceptions != 0) ------------------ | Branch (487:9): [True: 0, False: 1] ------------------ 488| 0| std::terminate(); 489| | // Push the foreign exception on to the stack 490| 1| globals->caughtExceptions = exception_header; 491| 1| return unwind_exception + 1; 492| 172k|} __cxa_end_catch: 513| 172k|void __cxa_end_catch() { 514| 172k| static_assert(sizeof(__cxa_exception) == sizeof(__cxa_dependent_exception), 515| 172k| "sizeof(__cxa_exception) must be equal to " 516| 172k| "sizeof(__cxa_dependent_exception)"); 517| 172k| static_assert(__builtin_offsetof(__cxa_exception, referenceCount) == 518| 172k| __builtin_offsetof(__cxa_dependent_exception, 519| 172k| primaryException), 520| 172k| "the layout of __cxa_exception must match the layout of " 521| 172k| "__cxa_dependent_exception"); 522| 172k| static_assert(__builtin_offsetof(__cxa_exception, handlerCount) == 523| 172k| __builtin_offsetof(__cxa_dependent_exception, handlerCount), 524| 172k| "the layout of __cxa_exception must match the layout of " 525| 172k| "__cxa_dependent_exception"); 526| 172k| __cxa_eh_globals* globals = __cxa_get_globals_fast(); // __cxa_get_globals called in __cxa_begin_catch 527| 172k| __cxa_exception* exception_header = globals->caughtExceptions; 528| | // If we've rethrown a foreign exception, then globals->caughtExceptions 529| | // will have been made an empty stack by __cxa_rethrow() and there is 530| | // nothing more to be done. Do nothing! 531| 172k| if (NULL != exception_header) ------------------ | Branch (531:9): [True: 172k, False: 0] ------------------ 532| 172k| { 533| 172k| bool native_exception = __isOurExceptionClass(&exception_header->unwindHeader); 534| 172k| if (native_exception) ------------------ | Branch (534:13): [True: 172k, False: 0] ------------------ 535| 172k| { 536| | // This is a native exception 537| 172k| if (exception_header->handlerCount < 0) ------------------ | Branch (537:17): [True: 0, False: 172k] ------------------ 538| 0| { 539| | // The exception has been rethrown by __cxa_rethrow, so don't delete it 540| 0| if (0 == incrementHandlerCount(exception_header)) ------------------ | Branch (540:21): [True: 0, False: 0] ------------------ 541| 0| { 542| | // Remove from the chain of uncaught exceptions 543| 0| globals->caughtExceptions = exception_header->nextException; 544| | // but don't destroy 545| 0| } 546| | // Keep handlerCount negative in case there are nested catch's 547| | // that need to be told that this exception is rethrown. Don't 548| | // erase this rethrow flag until the exception is recaught. 549| 0| } 550| 172k| else 551| 172k| { 552| | // The native exception has not been rethrown 553| 172k| if (0 == decrementHandlerCount(exception_header)) ------------------ | Branch (553:21): [True: 172k, False: 0] ------------------ 554| 172k| { 555| | // Remove from the chain of uncaught exceptions 556| 172k| globals->caughtExceptions = exception_header->nextException; 557| | // Destroy this exception, being careful to distinguish 558| | // between dependent and primary exceptions 559| 172k| if (isDependentException(&exception_header->unwindHeader)) ------------------ | Branch (559:25): [True: 0, False: 172k] ------------------ 560| 0| { 561| | // Reset exception_header to primaryException and deallocate the dependent exception 562| 0| __cxa_dependent_exception* dep_exception_header = 563| 0| reinterpret_cast<__cxa_dependent_exception*>(exception_header); 564| 0| exception_header = 565| 0| cxa_exception_from_thrown_object(dep_exception_header->primaryException); 566| 0| __cxa_free_dependent_exception(dep_exception_header); 567| 0| } 568| | // Destroy the primary exception only if its referenceCount goes to 0 569| | // (this decrement must be atomic) 570| 172k| __cxa_decrement_exception_refcount(thrown_object_from_cxa_exception(exception_header)); 571| 172k| } 572| 172k| } 573| 172k| } 574| 0| else 575| 0| { 576| | // The foreign exception has not been rethrown. Pop the stack 577| | // and delete it. If there are nested catch's and they try 578| | // to touch a foreign exception in any way, that is undefined 579| | // behavior. They likely can't since the only way to catch 580| | // a foreign exception is with catch (...)! 581| 0| _Unwind_DeleteException(&globals->caughtExceptions->unwindHeader); 582| 0| globals->caughtExceptions = 0; 583| 0| } 584| 172k| } 585| 172k|} __cxa_decrement_exception_refcount: 678| 172k|void __cxa_decrement_exception_refcount(void *thrown_object) throw() { 679| 172k| if (thrown_object != NULL ) ------------------ | Branch (679:9): [True: 172k, False: 0] ------------------ 680| 172k| { 681| 172k| __cxa_exception* exception_header = cxa_exception_from_thrown_object(thrown_object); 682| 172k| if (std::__libcpp_atomic_add(&exception_header->referenceCount, size_t(-1)) == 0) ------------------ | Branch (682:13): [True: 172k, False: 0] ------------------ 683| 172k| { 684| 172k| if (NULL != exception_header->exceptionDestructor) ------------------ | Branch (684:17): [True: 172k, False: 0] ------------------ 685| 172k| exception_header->exceptionDestructor(thrown_object); 686| 172k| __cxa_free_exception(thrown_object); 687| 172k| } 688| 172k| } 689| 172k|} __cxa_uncaught_exceptions: 777| 4.78k|{ 778| | // This does not report foreign exceptions in flight 779| 4.78k| __cxa_eh_globals* globals = __cxa_get_globals_fast(); 780| 4.78k| if (globals == 0) ------------------ | Branch (780:9): [True: 0, False: 4.78k] ------------------ 781| 0| return 0; 782| 4.78k| return globals->uncaughtExceptions; 783| 4.78k|} cxa_exception.cpp:_ZN10__cxxabiv1L45cxa_exception_size_from_exception_thrown_sizeEm: 76| 172k|size_t cxa_exception_size_from_exception_thrown_size(size_t size) { 77| 172k| return aligned_allocation_size(size + sizeof (__cxa_exception), 78| 172k| alignof(__cxa_exception)); 79| 172k|} cxa_exception.cpp:_ZN10__cxxabiv1L23aligned_allocation_sizeEmm: 71| 172k|size_t aligned_allocation_size(size_t s, size_t a) { 72| 172k| return (s + a - 1) & ~(a - 1); 73| 172k|} cxa_exception.cpp:_ZN10__cxxabiv1L24get_cxa_exception_offsetEv: 161| 344k|static size_t get_cxa_exception_offset() { 162| 344k| struct S { 163| 344k| } __attribute__((aligned)); 164| | 165| | // Compute the maximum alignment for the target machine. 166| 344k| constexpr size_t alignment = alignof(S); 167| 344k| constexpr size_t excp_size = sizeof(__cxa_exception); 168| 344k| constexpr size_t aligned_size = 169| 344k| (excp_size + alignment - 1) / alignment * alignment; 170| 344k| constexpr size_t offset = aligned_size - excp_size; 171| 344k| static_assert((offset == 0 || alignof(_Unwind_Exception) < alignment), 172| 344k| "offset is non-zero only if _Unwind_Exception isn't aligned"); 173| 344k| return offset; 174| 344k|} cxa_exception.cpp:_ZN10__cxxabiv1L32thrown_object_from_cxa_exceptionEPNS_15__cxa_exceptionE: 54| 344k|{ 55| 344k| return static_cast(exception_header + 1); 56| 344k|} cxa_exception.cpp:_ZN10__cxxabiv1L32cxa_exception_from_thrown_objectEPv: 44| 689k|{ 45| 689k| return static_cast<__cxa_exception*>(thrown_object) - 1; 46| 689k|} cxa_exception.cpp:_ZN10__cxxabiv1L20setOurExceptionClassEP17_Unwind_Exception: 86| 172k|static void setOurExceptionClass(_Unwind_Exception* unwind_exception) { 87| 172k| __setExceptionClass(unwind_exception, kOurExceptionClass); 88| 172k|} cxa_exception.cpp:_ZN10__cxxabiv1L45cxa_exception_from_exception_unwind_exceptionEP17_Unwind_Exception: 65| 172k|{ 66| 172k| return cxa_exception_from_thrown_object(unwind_exception + 1 ); 67| 172k|} cxa_exception.cpp:_ZN10__cxxabiv1L21decrementHandlerCountEPNS_15__cxa_exceptionE: 120| 172k|static inline int decrementHandlerCount(__cxa_exception *exception) { 121| 172k| return --exception->handlerCount; 122| 172k|} cxa_exception.cpp:_ZN10__cxxabiv1L20isDependentExceptionEP17_Unwind_Exception: 110| 172k|static bool isDependentException(_Unwind_Exception* unwind_exception) { 111| 172k| return (__getExceptionClass(unwind_exception) & 0xFF) == 0x01; 112| 172k|} __cxa_get_globals: 74| 344k| __cxa_eh_globals *__cxa_get_globals() { 75| | // Try to get the globals for this thread 76| 344k| __cxa_eh_globals *retVal = __cxa_get_globals_fast(); 77| | 78| | // If this is the first time we've been asked for these globals, create them 79| 344k| if (NULL == retVal) { ------------------ | Branch (79:13): [True: 32, False: 344k] ------------------ 80| 32| retVal = static_cast<__cxa_eh_globals*>( 81| 32| __calloc_with_fallback(1, sizeof(__cxa_eh_globals))); 82| 32| if (NULL == retVal) ------------------ | Branch (82:17): [True: 0, False: 32] ------------------ 83| 0| abort_message("cannot allocate __cxa_eh_globals"); 84| 32| if (0 != std::__libcpp_tls_set(key_, retVal)) ------------------ | Branch (84:17): [True: 0, False: 32] ------------------ 85| 0| abort_message("std::__libcpp_tls_set failure in __cxa_get_globals()"); 86| 32| } 87| 344k| return retVal; 88| 344k| } __cxa_get_globals_fast: 94| 521k| __cxa_eh_globals *__cxa_get_globals_fast() { 95| | // First time through, create the key. 96| 521k| if (0 != std::__libcpp_execute_once(&flag_, construct_)) ------------------ | Branch (96:13): [True: 0, False: 521k] ------------------ 97| 0| abort_message("execute once failure in __cxa_get_globals_fast()"); 98| 521k| return static_cast<__cxa_eh_globals*>(std::__libcpp_tls_get(key_)); 99| 521k| } cxa_exception_storage.cpp:_ZN10__cxxabiv112_GLOBAL__N_110construct_Ev: 67| 1| void construct_() { 68| 1| if (0 != std::__libcpp_tls_create(&key_, destruct_)) ------------------ | Branch (68:13): [True: 0, False: 1] ------------------ 69| 0| abort_message("cannot create thread specific key for __cxa_get_globals()"); 70| 1| } cxa_exception_storage.cpp:_ZN10__cxxabiv112_GLOBAL__N_19destruct_EPv: 61| 31| void _LIBCPP_TLS_DESTRUCTOR_CC destruct_(void *p) { 62| 31| __free_with_fallback(p); 63| 31| if (0 != std::__libcpp_tls_set(key_, NULL)) ------------------ | Branch (63:13): [True: 0, False: 31] ------------------ 64| 0| abort_message("cannot zero out thread value for __cxa_get_globals()"); 65| 31| } __cxa_guard_acquire: 37| 3|_LIBCXXABI_FUNC_VIS int __cxa_guard_acquire(guard_type* raw_guard_object) { 38| 3| SelectedImplementation imp(raw_guard_object); 39| 3| return static_cast(imp.cxa_guard_acquire()); 40| 3|} __cxa_guard_release: 42| 3|_LIBCXXABI_FUNC_VIS void __cxa_guard_release(guard_type *raw_guard_object) { 43| 3| SelectedImplementation imp(raw_guard_object); 44| 3| imp.cxa_guard_release(); 45| 3|} cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_111GuardObjectINS0_19InitByteGlobalMutexINS0_11LibcppMutexENS0_13LibcppCondVarEL_ZNS0_12GlobalStaticIS3_E8instanceEEL_ZNS5_IS4_E8instanceEEXadL_ZNS0_16PlatformThreadIDEvEEEEEC2EPm: 581| 6| : guard_byte(reinterpret_cast(raw_guard_object)), 582| 6| init_byte(reinterpret_cast(raw_guard_object) + 1, reinterpret_cast(raw_guard_object) + 1) { 583| 6| } cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_19GuardByteC2EPh: 188| 6| explicit GuardByte(uint8_t* const guard_byte_address) : guard_byte(guard_byte_address) {} cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_19AtomicIntIhEC2EPh: 140| 6| explicit AtomicInt(IntType* b) : b_(b) {} cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_119InitByteGlobalMutexINS0_11LibcppMutexENS0_13LibcppCondVarEL_ZNS0_12GlobalStaticIS2_E8instanceEEL_ZNS4_IS3_E8instanceEEXadL_ZNS0_16PlatformThreadIDEvEEEC2EPhPj: 320| 6| : init_byte_address(_init_byte_address), thread_id_address(_thread_id_address), 321| 6| has_thread_id_support(_thread_id_address != nullptr && GetThreadID != nullptr) {} ------------------ | Branch (321:31): [True: 6, False: 0] | Branch (321:64): [True: 6, Folded] ------------------ cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_116PlatformThreadIDEv: 165| 3|uint32_t PlatformThreadID() { 166| 3| static_assert(sizeof(pid_t) == sizeof(uint32_t), ""); 167| | return static_cast(syscall(SYS_gettid)); 168| 3|} cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_19LazyValueIjXadL_ZNS0_16PlatformThreadIDEvEEEC2Ev: 120| 6| LazyValue() : is_init(false) {} cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_111GuardObjectINS0_19InitByteGlobalMutexINS0_11LibcppMutexENS0_13LibcppCondVarEL_ZNS0_12GlobalStaticIS3_E8instanceEEL_ZNS5_IS4_E8instanceEEXadL_ZNS0_16PlatformThreadIDEvEEEEE17cxa_guard_acquireEv: 586| 3| AcquireResult cxa_guard_acquire() { 587| | // Use short-circuit evaluation to avoid calling init_byte.acquire when 588| | // guard_byte.acquire returns true. (i.e. don't call it when we know from 589| | // the guard byte that initialization has already been completed) 590| 3| if (guard_byte.acquire() || init_byte.acquire()) ------------------ | Branch (590:9): [True: 0, False: 3] | Branch (590:33): [True: 0, False: 3] ------------------ 591| 0| return INIT_IS_DONE; 592| 3| return INIT_IS_PENDING; 593| 3| } cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_19GuardByte7acquireEv: 193| 3| bool acquire() { 194| | // if guard_byte is non-zero, we have already completed initialization 195| | // (i.e. release has been called) 196| 3| return guard_byte.load(std::_AO_Acquire) != UNSET; 197| 3| } cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_19AtomicIntIhE4loadENSt3__112_GLOBAL__N_121__libcpp_atomic_orderE: 144| 3| IntType load(MemoryOrder ord) { return std::__libcpp_atomic_load(b_, ord); } cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_119InitByteGlobalMutexINS0_11LibcppMutexENS0_13LibcppCondVarEL_ZNS0_12GlobalStaticIS2_E8instanceEEL_ZNS4_IS3_E8instanceEEXadL_ZNS0_16PlatformThreadIDEvEEE7acquireEv: 326| 3| bool acquire() { 327| 3| LockGuard g("__cxa_guard_acquire"); 328| | // Check for possible recursive initialization. 329| 3| if (has_thread_id_support && (*init_byte_address & PENDING_BIT)) { ------------------ | Branch (329:9): [True: 3, False: 0] | Branch (329:34): [True: 0, False: 3] ------------------ 330| 0| if (*thread_id_address == current_thread_id.get()) ------------------ | Branch (330:11): [True: 0, False: 0] ------------------ 331| 0| ABORT_WITH_MESSAGE("__cxa_guard_acquire detected recursive initialization: do you have a function-local static variable whose initialization depends on that function?"); ------------------ | | 94| 0|# define ABORT_WITH_MESSAGE(...) ::abort_message(__VA_ARGS__) ------------------ 332| 0| } 333| | 334| | // Wait until the pending bit is not set. 335| 3| while (*init_byte_address & PENDING_BIT) { ------------------ | Branch (335:12): [True: 0, False: 3] ------------------ 336| 0| *init_byte_address |= WAITING_BIT; 337| 0| global_cond.wait(global_mutex); 338| 0| } 339| | 340| 3| if (*init_byte_address == COMPLETE_BIT) ------------------ | Branch (340:9): [True: 0, False: 3] ------------------ 341| 0| return true; 342| | 343| 3| if (has_thread_id_support) ------------------ | Branch (343:9): [True: 3, False: 0] ------------------ 344| 3| *thread_id_address = current_thread_id.get(); 345| | 346| 3| *init_byte_address = PENDING_BIT; 347| 3| return false; 348| 3| } cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_119InitByteGlobalMutexINS0_11LibcppMutexENS0_13LibcppCondVarEL_ZNS0_12GlobalStaticIS2_E8instanceEEL_ZNS4_IS3_E8instanceEEXadL_ZNS0_16PlatformThreadIDEvEEE9LockGuardC2EPKc: 398| 6| explicit LockGuard(const char* calling_func) : calling_func_(calling_func) { 399| 6| if (global_mutex.lock()) ------------------ | Branch (399:11): [True: 0, False: 6] ------------------ 400| 0| ABORT_WITH_MESSAGE("%s failed to acquire mutex", calling_func_); ------------------ | | 94| 0|# define ABORT_WITH_MESSAGE(...) ::abort_message(__VA_ARGS__) ------------------ 401| 6| } cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_111LibcppMutex4lockEv: 289| 6| bool lock() { return std::__libcpp_mutex_lock(&mutex); } cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_19LazyValueIjXadL_ZNS0_16PlatformThreadIDEvEEE3getEv: 122| 3| T& get() { 123| 3| if (!is_init) { ------------------ | Branch (123:9): [True: 3, False: 0] ------------------ 124| 3| value = Init(); 125| 3| is_init = true; 126| 3| } 127| 3| return value; 128| 3| } cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_119InitByteGlobalMutexINS0_11LibcppMutexENS0_13LibcppCondVarEL_ZNS0_12GlobalStaticIS2_E8instanceEEL_ZNS4_IS3_E8instanceEEXadL_ZNS0_16PlatformThreadIDEvEEE9LockGuardD2Ev: 403| 6| ~LockGuard() { 404| 6| if (global_mutex.unlock()) ------------------ | Branch (404:11): [True: 0, False: 6] ------------------ 405| 0| ABORT_WITH_MESSAGE("%s failed to release mutex", calling_func_); ------------------ | | 94| 0|# define ABORT_WITH_MESSAGE(...) ::abort_message(__VA_ARGS__) ------------------ 406| 6| } cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_111LibcppMutex6unlockEv: 290| 6| bool unlock() { return std::__libcpp_mutex_unlock(&mutex); } cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_111GuardObjectINS0_19InitByteGlobalMutexINS0_11LibcppMutexENS0_13LibcppCondVarEL_ZNS0_12GlobalStaticIS3_E8instanceEEL_ZNS5_IS4_E8instanceEEXadL_ZNS0_16PlatformThreadIDEvEEEEE17cxa_guard_releaseEv: 596| 3| void cxa_guard_release() { 597| | // Update guard byte first, so if somebody is woken up by init_byte.release 598| | // and comes all the way back around to __cxa_guard_acquire again, they see 599| | // it as having completed initialization. 600| 3| guard_byte.release(); 601| 3| init_byte.release(); 602| 3| } cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_19GuardByte7releaseEv: 200| 3| void release() { guard_byte.store(COMPLETE_BIT, std::_AO_Release); } cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_19AtomicIntIhE5storeEhNSt3__112_GLOBAL__N_121__libcpp_atomic_orderE: 145| 3| void store(IntType val, MemoryOrder ord) { std::__libcpp_atomic_store(b_, val, ord); } cxa_guard.cpp:_ZN10__cxxabiv112_GLOBAL__N_119InitByteGlobalMutexINS0_11LibcppMutexENS0_13LibcppCondVarEL_ZNS0_12GlobalStaticIS2_E8instanceEEL_ZNS4_IS3_E8instanceEEXadL_ZNS0_16PlatformThreadIDEvEEE7releaseEv: 351| 3| void release() { 352| 3| bool has_waiting; 353| 3| { 354| 3| LockGuard g("__cxa_guard_release"); 355| 3| has_waiting = *init_byte_address & WAITING_BIT; 356| 3| *init_byte_address = COMPLETE_BIT; 357| 3| } 358| 3| if (has_waiting) { ------------------ | Branch (358:9): [True: 0, False: 3] ------------------ 359| 0| if (global_cond.broadcast()) { ------------------ | Branch (359:11): [True: 0, False: 0] ------------------ 360| 0| ABORT_WITH_MESSAGE("%s failed to broadcast", "__cxa_guard_release"); ------------------ | | 94| 0|# define ABORT_WITH_MESSAGE(...) ::abort_message(__VA_ARGS__) ------------------ 361| 0| } 362| 0| } 363| 3| } _ZSt14get_unexpectedv: 27| 172k|{ 28| 172k| return __libcpp_atomic_load(&__cxa_unexpected_handler, _AO_Acquire); 29| 172k|} _ZSt13get_terminatev: 48| 172k|{ 49| 172k| return __libcpp_atomic_load(&__cxa_terminate_handler, _AO_Acquire); 50| 172k|} __gxx_personality_v0: 924| 923k|{ 925| 923k| if (version != 1 || unwind_exception == 0 || context == 0) ------------------ | Branch (925:9): [True: 0, False: 923k] | Branch (925:25): [True: 0, False: 923k] | Branch (925:50): [True: 0, False: 923k] ------------------ 926| 0| return _URC_FATAL_PHASE1_ERROR; 927| | 928| 923k| bool native_exception = (exceptionClass & get_vendor_and_language) == 929| 923k| (kOurExceptionClass & get_vendor_and_language); 930| 923k| scan_results results; 931| | // Process a catch handler for a native exception first. 932| 923k| if (actions == (_UA_CLEANUP_PHASE | _UA_HANDLER_FRAME) && ------------------ | Branch (932:9): [True: 172k, False: 751k] ------------------ 933| 172k| native_exception) { ------------------ | Branch (933:9): [True: 172k, False: 0] ------------------ 934| | // Reload the results from the phase 1 cache. 935| 172k| __cxa_exception* exception_header = 936| 172k| (__cxa_exception*)(unwind_exception + 1) - 1; 937| 172k| results.ttypeIndex = exception_header->handlerSwitchValue; 938| 172k| results.actionRecord = exception_header->actionRecord; 939| 172k| results.languageSpecificData = exception_header->languageSpecificData; 940| 172k| results.landingPad = 941| 172k| reinterpret_cast(exception_header->catchTemp); 942| 172k| results.adjustedPtr = exception_header->adjustedPtr; 943| | 944| | // Jump to the handler. 945| 172k| set_registers(unwind_exception, context, results); 946| | // Cache base for calculating the address of ttype in 947| | // __cxa_call_unexpected. 948| 172k| if (results.ttypeIndex < 0) { ------------------ | Branch (948:13): [True: 0, False: 172k] ------------------ 949| |#if defined(_AIX) 950| | exception_header->catchTemp = (void *)_Unwind_GetDataRelBase(context); 951| |#else 952| 0| exception_header->catchTemp = 0; 953| 0|#endif 954| 0| } 955| 172k| return _URC_INSTALL_CONTEXT; 956| 172k| } 957| | 958| | // In other cases we need to scan LSDA. 959| 751k| scan_eh_tab(results, actions, native_exception, unwind_exception, context); 960| 751k| if (results.reason == _URC_CONTINUE_UNWIND || ------------------ | Branch (960:9): [True: 509k, False: 241k] ------------------ 961| 241k| results.reason == _URC_FATAL_PHASE1_ERROR) ------------------ | Branch (961:9): [True: 18.4E, False: 241k] ------------------ 962| 509k| return results.reason; 963| | 964| 241k| if (actions & _UA_SEARCH_PHASE) ------------------ | Branch (964:9): [True: 172k, False: 69.2k] ------------------ 965| 172k| { 966| | // Phase 1 search: All we're looking for in phase 1 is a handler that 967| | // halts unwinding 968| 172k| assert(results.reason == _URC_HANDLER_FOUND); ------------------ | Branch (968:9): [True: 172k, False: 0] ------------------ 969| 172k| if (native_exception) { ------------------ | Branch (969:13): [True: 172k, False: 0] ------------------ 970| | // For a native exception, cache the LSDA result. 971| 172k| __cxa_exception* exc = (__cxa_exception*)(unwind_exception + 1) - 1; 972| 172k| exc->handlerSwitchValue = static_cast(results.ttypeIndex); 973| 172k| exc->actionRecord = results.actionRecord; 974| 172k| exc->languageSpecificData = results.languageSpecificData; 975| 172k| exc->catchTemp = reinterpret_cast(results.landingPad); 976| 172k| exc->adjustedPtr = results.adjustedPtr; 977| |#ifdef __USING_WASM_EXCEPTIONS__ 978| | // Wasm only uses a single phase (_UA_SEARCH_PHASE), so save the 979| | // results here. 980| | set_registers(unwind_exception, context, results); 981| |#endif 982| 172k| } 983| 172k| return _URC_HANDLER_FOUND; 984| 172k| } 985| | 986| 241k| assert(actions & _UA_CLEANUP_PHASE); ------------------ | Branch (986:5): [True: 69.1k, False: 13] ------------------ 987| 69.2k| assert(results.reason == _URC_HANDLER_FOUND); ------------------ | Branch (987:5): [True: 69.1k, False: 0] ------------------ 988| 69.1k| set_registers(unwind_exception, context, results); 989| | // Cache base for calculating the address of ttype in __cxa_call_unexpected. 990| 69.1k| if (results.ttypeIndex < 0) { ------------------ | Branch (990:9): [True: 0, False: 69.1k] ------------------ 991| 0| __cxa_exception* exception_header = 992| 0| (__cxa_exception*)(unwind_exception + 1) - 1; 993| |#if defined(_AIX) 994| | exception_header->catchTemp = (void *)_Unwind_GetDataRelBase(context); 995| |#else 996| 0| exception_header->catchTemp = 0; 997| 0|#endif 998| 0| } 999| 69.1k| return _URC_INSTALL_CONTEXT; 1000| 69.1k|} cxa_personality.cpp:_ZN10__cxxabiv1L13set_registersEP17_Unwind_ExceptionP15_Unwind_ContextRKNS_12_GLOBAL__N_112scan_resultsE: 549| 241k|{ 550| |#if defined(__USING_SJLJ_EXCEPTIONS__) || defined(__USING_WASM_EXCEPTIONS__) 551| |#define __builtin_eh_return_data_regno(regno) regno 552| |#elif defined(__ibmxl__) 553| |// IBM xlclang++ compiler does not support __builtin_eh_return_data_regno. 554| |#define __builtin_eh_return_data_regno(regno) regno + 3 555| |#endif 556| 241k| _Unwind_SetGR(context, __builtin_eh_return_data_regno(0), 557| 241k| reinterpret_cast(unwind_exception)); 558| 241k| _Unwind_SetGR(context, __builtin_eh_return_data_regno(1), 559| 241k| static_cast(results.ttypeIndex)); 560| 241k| _Unwind_SetIP(context, results.landingPad); 561| 241k|} cxa_personality.cpp:_ZN10__cxxabiv1L11scan_eh_tabERNS_12_GLOBAL__N_112scan_resultsE14_Unwind_ActionbP17_Unwind_ExceptionP15_Unwind_Context: 587| 751k| _Unwind_Context *context) { 588| | // Initialize results to found nothing but an error 589| 751k| results.ttypeIndex = 0; 590| 751k| results.actionRecord = 0; 591| 751k| results.languageSpecificData = 0; 592| 751k| results.landingPad = 0; 593| 751k| results.adjustedPtr = 0; 594| 751k| results.reason = _URC_FATAL_PHASE1_ERROR; 595| | // Check for consistent actions 596| 751k| if (actions & _UA_SEARCH_PHASE) ------------------ | Branch (596:9): [True: 427k, False: 323k] ------------------ 597| 427k| { 598| | // Do Phase 1 599| 427k| if (actions & (_UA_CLEANUP_PHASE | _UA_HANDLER_FRAME | _UA_FORCE_UNWIND)) ------------------ | Branch (599:13): [True: 0, False: 427k] ------------------ 600| 0| { 601| | // None of these flags should be set during Phase 1 602| | // Client error 603| 0| results.reason = _URC_FATAL_PHASE1_ERROR; 604| 0| return; 605| 0| } 606| 427k| } 607| 323k| else if (actions & _UA_CLEANUP_PHASE) ------------------ | Branch (607:14): [True: 323k, False: 18.4E] ------------------ 608| 323k| { 609| 323k| if ((actions & _UA_HANDLER_FRAME) && (actions & _UA_FORCE_UNWIND)) ------------------ | Branch (609:13): [True: 0, False: 323k] | Branch (609:46): [True: 0, False: 0] ------------------ 610| 0| { 611| | // _UA_HANDLER_FRAME should only be set if phase 1 found a handler. 612| | // If _UA_FORCE_UNWIND is set, phase 1 shouldn't have happened. 613| | // Client error 614| 0| results.reason = _URC_FATAL_PHASE2_ERROR; 615| 0| return; 616| 0| } 617| 323k| } 618| 18.4E| else // Neither _UA_SEARCH_PHASE nor _UA_CLEANUP_PHASE is set 619| 18.4E| { 620| | // One of these should be set. 621| | // Client error 622| 18.4E| results.reason = _URC_FATAL_PHASE1_ERROR; 623| 18.4E| return; 624| 18.4E| } 625| | // Start scan by getting exception table address. 626| 751k| const uint8_t *lsda = (const uint8_t *)_Unwind_GetLanguageSpecificData(context); 627| 751k| if (lsda == 0) ------------------ | Branch (627:9): [True: 0, False: 751k] ------------------ 628| 0| { 629| | // There is no exception table 630| 0| results.reason = _URC_CONTINUE_UNWIND; 631| 0| return; 632| 0| } 633| 751k| results.languageSpecificData = lsda; 634| |#if defined(_AIX) 635| | uintptr_t base = _Unwind_GetDataRelBase(context); 636| |#else 637| 751k| uintptr_t base = 0; 638| 751k|#endif 639| | // Get the current instruction pointer and offset it before next 640| | // instruction in the current frame which threw the exception. 641| 751k| uintptr_t ip = _Unwind_GetIP(context) - 1; 642| | // Get beginning current frame's code (as defined by the 643| | // emitted dwarf code) 644| 751k| uintptr_t funcStart = _Unwind_GetRegionStart(context); 645| |#if defined(__USING_SJLJ_EXCEPTIONS__) || defined(__USING_WASM_EXCEPTIONS__) 646| | if (ip == uintptr_t(-1)) 647| | { 648| | // no action 649| | results.reason = _URC_CONTINUE_UNWIND; 650| | return; 651| | } 652| | else if (ip == 0) 653| | call_terminate(native_exception, unwind_exception); 654| | // ip is 1-based index into call site table 655| |#else // !__USING_SJLJ_EXCEPTIONS__ && !__USING_WASM_EXCEPTIONS__ 656| 751k| uintptr_t ipOffset = ip - funcStart; 657| 751k|#endif // !__USING_SJLJ_EXCEPTIONS__ && !__USING_WASM_EXCEPTIONS__ 658| 751k| const uint8_t* classInfo = NULL; 659| | // Note: See JITDwarfEmitter::EmitExceptionTable(...) for corresponding 660| | // dwarf emission 661| | // Parse LSDA header. 662| 751k| uint8_t lpStartEncoding = *lsda++; 663| 751k| const uint8_t* lpStart = lpStartEncoding == DW_EH_PE_omit ------------------ | Branch (663:30): [True: 751k, False: 4] ------------------ 664| 751k| ? (const uint8_t*)funcStart 665| 751k| : (const uint8_t*)readEncodedPointer(&lsda, lpStartEncoding, base); 666| 751k| uint8_t ttypeEncoding = *lsda++; 667| 751k| if (ttypeEncoding != DW_EH_PE_omit) ------------------ | Branch (667:9): [True: 201k, False: 549k] ------------------ 668| 201k| { 669| | // Calculate type info locations in emitted dwarf code which 670| | // were flagged by type info arguments to llvm.eh.selector 671| | // intrinsic 672| 201k| uintptr_t classInfoOffset = readULEB128(&lsda); 673| 201k| classInfo = lsda + classInfoOffset; 674| 201k| } 675| | // Walk call-site table looking for range that 676| | // includes current PC. 677| 751k| uint8_t callSiteEncoding = *lsda++; 678| |#if defined(__USING_SJLJ_EXCEPTIONS__) || defined(__USING_WASM_EXCEPTIONS__) 679| | (void)callSiteEncoding; // When using SjLj/Wasm exceptions, callSiteEncoding is never used 680| |#endif 681| 751k| uint32_t callSiteTableLength = static_cast(readULEB128(&lsda)); 682| 751k| const uint8_t* callSiteTableStart = lsda; 683| 751k| const uint8_t* callSiteTableEnd = callSiteTableStart + callSiteTableLength; 684| 751k| const uint8_t* actionTableStart = callSiteTableEnd; 685| 751k| const uint8_t* callSitePtr = callSiteTableStart; 686| 3.01M| while (callSitePtr < callSiteTableEnd) ------------------ | Branch (686:12): [True: 3.01M, False: 17] ------------------ 687| 3.01M| { 688| | // There is one entry per call site. 689| 3.01M|#if !defined(__USING_SJLJ_EXCEPTIONS__) && !defined(__USING_WASM_EXCEPTIONS__) 690| | // The call sites are non-overlapping in [start, start+length) 691| | // The call sites are ordered in increasing value of start 692| 3.01M| uintptr_t start = readEncodedPointer(&callSitePtr, callSiteEncoding); 693| 3.01M| uintptr_t length = readEncodedPointer(&callSitePtr, callSiteEncoding); 694| 3.01M| uintptr_t landingPad = readEncodedPointer(&callSitePtr, callSiteEncoding); 695| 3.01M| uintptr_t actionEntry = readULEB128(&callSitePtr); 696| 3.01M| if ((start <= ipOffset) && (ipOffset < (start + length))) ------------------ | Branch (696:13): [True: 3.01M, False: 15] | Branch (696:36): [True: 751k, False: 2.26M] ------------------ 697| |#else // __USING_SJLJ_EXCEPTIONS__ || __USING_WASM_EXCEPTIONS__ 698| | // ip is 1-based index into this table 699| | uintptr_t landingPad = readULEB128(&callSitePtr); 700| | uintptr_t actionEntry = readULEB128(&callSitePtr); 701| | if (--ip == 0) 702| |#endif // __USING_SJLJ_EXCEPTIONS__ || __USING_WASM_EXCEPTIONS__ 703| 751k| { 704| | // Found the call site containing ip. 705| 751k|#if !defined(__USING_SJLJ_EXCEPTIONS__) && !defined(__USING_WASM_EXCEPTIONS__) 706| 751k| if (landingPad == 0) ------------------ | Branch (706:17): [True: 426k, False: 324k] ------------------ 707| 426k| { 708| | // No handler here 709| 426k| results.reason = _URC_CONTINUE_UNWIND; 710| 426k| return; 711| 426k| } 712| 324k| landingPad = (uintptr_t)lpStart + landingPad; 713| |#else // __USING_SJLJ_EXCEPTIONS__ || __USING_WASM_EXCEPTIONS__ 714| | ++landingPad; 715| |#endif // __USING_SJLJ_EXCEPTIONS__ || __USING_WASM_EXCEPTIONS__ 716| 324k| results.landingPad = landingPad; 717| 324k| if (actionEntry == 0) ------------------ | Branch (717:17): [True: 134k, False: 190k] ------------------ 718| 134k| { 719| | // Found a cleanup 720| 134k| results.reason = actions & _UA_SEARCH_PHASE ------------------ | Branch (720:34): [True: 67.0k, False: 67.0k] ------------------ 721| 134k| ? _URC_CONTINUE_UNWIND 722| 134k| : _URC_HANDLER_FOUND; 723| 134k| return; 724| 134k| } 725| | // Convert 1-based byte offset into 726| 190k| const uint8_t* action = actionTableStart + (actionEntry - 1); 727| 190k| bool hasCleanup = false; 728| | // Scan action entries until you find a matching handler, cleanup, or the end of action list 729| 237k| while (true) ------------------ | Branch (729:20): [True: 237k, Folded] ------------------ 730| 237k| { 731| 237k| const uint8_t* actionRecord = action; 732| 237k| int64_t ttypeIndex = readSLEB128(&action); 733| 237k| if (ttypeIndex > 0) ------------------ | Branch (733:21): [True: 233k, False: 4.30k] ------------------ 734| 233k| { 735| | // Found a catch, does it actually catch? 736| | // First check for catch (...) 737| 233k| const __shim_type_info* catchType = 738| 233k| get_shim_type_info(static_cast(ttypeIndex), 739| 233k| classInfo, ttypeEncoding, 740| 233k| native_exception, unwind_exception, 741| 233k| base); 742| 233k| if (catchType == 0) ------------------ | Branch (742:25): [True: 0, False: 233k] ------------------ 743| 0| { 744| | // Found catch (...) catches everything, including 745| | // foreign exceptions. This is search phase, cleanup 746| | // phase with foreign exception, or forced unwinding. 747| 0| assert(actions & (_UA_SEARCH_PHASE | _UA_HANDLER_FRAME | ------------------ | Branch (747:25): [True: 0, False: 0] ------------------ 748| 0| _UA_FORCE_UNWIND)); 749| 0| results.ttypeIndex = ttypeIndex; 750| 0| results.actionRecord = actionRecord; 751| 0| results.adjustedPtr = 752| 0| get_thrown_object_ptr(unwind_exception); 753| 0| results.reason = _URC_HANDLER_FOUND; 754| 0| return; 755| 0| } 756| | // Else this is a catch (T) clause and will never 757| | // catch a foreign exception 758| 233k| else if (native_exception) ------------------ | Branch (758:30): [True: 233k, False: 1] ------------------ 759| 233k| { 760| 233k| __cxa_exception* exception_header = (__cxa_exception*)(unwind_exception+1) - 1; 761| 233k| void* adjustedPtr = get_thrown_object_ptr(unwind_exception); 762| 233k| const __shim_type_info* excpType = 763| 233k| static_cast(exception_header->exceptionType); 764| 233k| if (adjustedPtr == 0 || excpType == 0) ------------------ | Branch (764:29): [True: 0, False: 233k] | Branch (764:49): [True: 0, False: 233k] ------------------ 765| 0| { 766| | // Something very bad happened 767| 0| call_terminate(native_exception, unwind_exception); 768| 0| } 769| 233k| if (catchType->can_catch(excpType, adjustedPtr)) ------------------ | Branch (769:29): [True: 172k, False: 60.8k] ------------------ 770| 172k| { 771| | // Found a matching handler. This is either search 772| | // phase or forced unwinding. 773| 172k| assert(actions & ------------------ | Branch (773:29): [True: 172k, False: 18.4E] ------------------ 774| 172k| (_UA_SEARCH_PHASE | _UA_FORCE_UNWIND)); 775| 172k| results.ttypeIndex = ttypeIndex; 776| 172k| results.actionRecord = actionRecord; 777| 172k| results.adjustedPtr = adjustedPtr; 778| 172k| results.reason = _URC_HANDLER_FOUND; 779| 172k| return; 780| 172k| } 781| 233k| } 782| | // Scan next action ... 783| 233k| } 784| 4.30k| else if (ttypeIndex < 0) ------------------ | Branch (784:26): [True: 0, False: 4.30k] ------------------ 785| 0| { 786| | // Found an exception specification. 787| 0| if (actions & _UA_FORCE_UNWIND) { ------------------ | Branch (787:25): [True: 0, False: 0] ------------------ 788| | // Skip if forced unwinding. 789| 0| } else if (native_exception) { ------------------ | Branch (789:32): [True: 0, False: 0] ------------------ 790| | // Does the exception spec catch this native exception? 791| 0| __cxa_exception* exception_header = (__cxa_exception*)(unwind_exception+1) - 1; 792| 0| void* adjustedPtr = get_thrown_object_ptr(unwind_exception); 793| 0| const __shim_type_info* excpType = 794| 0| static_cast(exception_header->exceptionType); 795| 0| if (adjustedPtr == 0 || excpType == 0) ------------------ | Branch (795:29): [True: 0, False: 0] | Branch (795:49): [True: 0, False: 0] ------------------ 796| 0| { 797| | // Something very bad happened 798| 0| call_terminate(native_exception, unwind_exception); 799| 0| } 800| 0| if (exception_spec_can_catch(ttypeIndex, classInfo, ------------------ | Branch (800:29): [True: 0, False: 0] ------------------ 801| 0| ttypeEncoding, excpType, 802| 0| adjustedPtr, 803| 0| unwind_exception, base)) 804| 0| { 805| | // Native exception caught by exception 806| | // specification. 807| 0| assert(actions & _UA_SEARCH_PHASE); ------------------ | Branch (807:29): [True: 0, False: 0] ------------------ 808| 0| results.ttypeIndex = ttypeIndex; 809| 0| results.actionRecord = actionRecord; 810| 0| results.adjustedPtr = adjustedPtr; 811| 0| results.reason = _URC_HANDLER_FOUND; 812| 0| return; 813| 0| } 814| 0| } else { 815| | // foreign exception caught by exception spec 816| 0| results.ttypeIndex = ttypeIndex; 817| 0| results.actionRecord = actionRecord; 818| 0| results.adjustedPtr = 819| 0| get_thrown_object_ptr(unwind_exception); 820| 0| results.reason = _URC_HANDLER_FOUND; 821| 0| return; 822| 0| } 823| | // Scan next action ... 824| 4.30k| } else { 825| 4.30k| hasCleanup = true; 826| 4.30k| } 827| 65.1k| const uint8_t* temp = action; 828| 65.1k| int64_t actionOffset = readSLEB128(&temp); 829| 65.1k| if (actionOffset == 0) ------------------ | Branch (829:21): [True: 18.0k, False: 47.0k] ------------------ 830| 18.0k| { 831| | // End of action list. If this is phase 2 and we have found 832| | // a cleanup (ttypeIndex=0), return _URC_HANDLER_FOUND; 833| | // otherwise return _URC_CONTINUE_UNWIND. 834| 18.0k| results.reason = hasCleanup && actions & _UA_CLEANUP_PHASE ------------------ | Branch (834:38): [True: 4.30k, False: 13.7k] | Branch (834:52): [True: 2.15k, False: 2.15k] ------------------ 835| 18.0k| ? _URC_HANDLER_FOUND 836| 18.0k| : _URC_CONTINUE_UNWIND; 837| 18.0k| return; 838| 18.0k| } 839| | // Go to next action 840| 47.0k| action += actionOffset; 841| 47.0k| } // there is no break out of this loop, only return 842| 190k| } 843| 2.26M|#if !defined(__USING_SJLJ_EXCEPTIONS__) && !defined(__USING_WASM_EXCEPTIONS__) 844| 2.26M| else if (ipOffset < start) ------------------ | Branch (844:18): [True: 0, False: 2.26M] ------------------ 845| 0| { 846| | // There is no call site for this ip 847| | // Something bad has happened. We should never get here. 848| | // Possible stack corruption. 849| 0| call_terminate(native_exception, unwind_exception); 850| 0| } 851| 3.01M|#endif // !__USING_SJLJ_EXCEPTIONS__ && !__USING_WASM_EXCEPTIONS__ 852| 3.01M| } // there might be some tricky cases which break out of this loop 853| | 854| | // It is possible that no eh table entry specify how to handle 855| | // this exception. By spec, terminate it immediately. 856| 17| call_terminate(native_exception, unwind_exception); 857| 17|} cxa_personality.cpp:_ZN10__cxxabiv1L11readSLEB128EPPKh: 231| 302k|{ 232| 302k| uintptr_t result = 0; 233| 302k| uintptr_t shift = 0; 234| 302k| unsigned char byte; 235| 302k| const uint8_t *p = *data; 236| 302k| do 237| 302k| { 238| 302k| byte = *p++; 239| 302k| result |= static_cast(byte & 0x7F) << shift; 240| 302k| shift += 7; 241| 302k| } while (byte & 0x80); ------------------ | Branch (241:14): [True: 0, False: 302k] ------------------ 242| 302k| *data = p; 243| 302k| if ((byte & 0x40) && (shift < (sizeof(result) << 3))) ------------------ | Branch (243:9): [True: 47.0k, False: 255k] | Branch (243:26): [True: 47.0k, False: 0] ------------------ 244| 47.0k| result |= static_cast(~0) << shift; 245| 302k| return static_cast(result); 246| 302k|} cxa_personality.cpp:_ZN10__cxxabiv1L18get_shim_type_infoEmPKhhbP17_Unwind_Exceptionm: 389| 233k|{ 390| 233k| if (classInfo == 0) ------------------ | Branch (390:9): [True: 0, False: 233k] ------------------ 391| 0| { 392| | // this should not happen. Indicates corrupted eh_table. 393| 0| call_terminate(native_exception, unwind_exception); 394| 0| } 395| 233k| switch (ttypeEncoding & 0x0F) 396| 233k| { 397| 0| case DW_EH_PE_absptr: ------------------ | Branch (397:5): [True: 0, False: 233k] ------------------ 398| 0| ttypeIndex *= sizeof(void*); 399| 0| break; 400| 0| case DW_EH_PE_udata2: ------------------ | Branch (400:5): [True: 0, False: 233k] ------------------ 401| 0| case DW_EH_PE_sdata2: ------------------ | Branch (401:5): [True: 0, False: 233k] ------------------ 402| 0| ttypeIndex *= 2; 403| 0| break; 404| 0| case DW_EH_PE_udata4: ------------------ | Branch (404:5): [True: 0, False: 233k] ------------------ 405| 233k| case DW_EH_PE_sdata4: ------------------ | Branch (405:5): [True: 233k, False: 18.4E] ------------------ 406| 233k| ttypeIndex *= 4; 407| 233k| break; 408| 0| case DW_EH_PE_udata8: ------------------ | Branch (408:5): [True: 0, False: 233k] ------------------ 409| 0| case DW_EH_PE_sdata8: ------------------ | Branch (409:5): [True: 0, False: 233k] ------------------ 410| 0| ttypeIndex *= 8; 411| 0| break; 412| 0| default: ------------------ | Branch (412:5): [True: 0, False: 233k] ------------------ 413| | // this should not happen. Indicates corrupted eh_table. 414| 0| call_terminate(native_exception, unwind_exception); 415| 233k| } 416| 233k| classInfo -= ttypeIndex; 417| 233k| return (const __shim_type_info*)readEncodedPointer(&classInfo, 418| 233k| ttypeEncoding, base); 419| 233k|} cxa_personality.cpp:_ZN10__cxxabiv1L21get_thrown_object_ptrEP17_Unwind_Exception: 518| 233k|{ 519| | // Even for foreign exceptions, the exception object is *probably* at unwind_exception + 1 520| | // Regardless, this library is prohibited from touching a foreign exception 521| 233k| void* adjustedPtr = unwind_exception + 1; 522| 233k| if (__getExceptionClass(unwind_exception) == kOurDependentExceptionClass) ------------------ | Branch (522:9): [True: 0, False: 233k] ------------------ 523| 0| adjustedPtr = ((__cxa_dependent_exception*)adjustedPtr - 1)->primaryException; 524| 233k| return adjustedPtr; 525| 233k|} cxa_personality.cpp:_ZN10__cxxabiv1L18readEncodedPointerEPPKhhm: 258| 9.28M|{ 259| 9.28M| uintptr_t result = 0; 260| 9.28M| if (encoding == DW_EH_PE_omit) ------------------ | Branch (260:9): [True: 0, False: 9.28M] ------------------ 261| 0| return result; 262| 9.28M| const uint8_t* p = *data; 263| | // first get value 264| 9.28M| switch (encoding & 0x0F) 265| 9.28M| { 266| 0| case DW_EH_PE_absptr: ------------------ | Branch (266:5): [True: 0, False: 9.28M] ------------------ 267| 0| result = readPointerHelper(p); 268| 0| break; 269| 9.04M| case DW_EH_PE_uleb128: ------------------ | Branch (269:5): [True: 9.04M, False: 233k] ------------------ 270| 9.04M| result = readULEB128(&p); 271| 9.04M| break; 272| 0| case DW_EH_PE_sleb128: ------------------ | Branch (272:5): [True: 0, False: 9.28M] ------------------ 273| 0| result = static_cast(readSLEB128(&p)); 274| 0| break; 275| 0| case DW_EH_PE_udata2: ------------------ | Branch (275:5): [True: 0, False: 9.28M] ------------------ 276| 0| result = readPointerHelper(p); 277| 0| break; 278| 0| case DW_EH_PE_udata4: ------------------ | Branch (278:5): [True: 0, False: 9.28M] ------------------ 279| 0| result = readPointerHelper(p); 280| 0| break; 281| 0| case DW_EH_PE_udata8: ------------------ | Branch (281:5): [True: 0, False: 9.28M] ------------------ 282| 0| result = readPointerHelper(p); 283| 0| break; 284| 0| case DW_EH_PE_sdata2: ------------------ | Branch (284:5): [True: 0, False: 9.28M] ------------------ 285| 0| result = readPointerHelper(p); 286| 0| break; 287| 233k| case DW_EH_PE_sdata4: ------------------ | Branch (287:5): [True: 233k, False: 9.04M] ------------------ 288| 233k| result = readPointerHelper(p); 289| 233k| break; 290| 0| case DW_EH_PE_sdata8: ------------------ | Branch (290:5): [True: 0, False: 9.28M] ------------------ 291| 0| result = readPointerHelper(p); 292| 0| break; 293| 0| default: ------------------ | Branch (293:5): [True: 0, False: 9.28M] ------------------ 294| | // not supported 295| 0| abort(); 296| 0| break; 297| 9.28M| } 298| | // then add relative offset 299| 9.28M| switch (encoding & 0x70) 300| 9.28M| { 301| 9.04M| case DW_EH_PE_absptr: ------------------ | Branch (301:5): [True: 9.04M, False: 233k] ------------------ 302| | // do nothing 303| 9.04M| break; 304| 233k| case DW_EH_PE_pcrel: ------------------ | Branch (304:5): [True: 233k, False: 9.04M] ------------------ 305| 233k| if (result) ------------------ | Branch (305:13): [True: 233k, False: 1] ------------------ 306| 233k| result += (uintptr_t)(*data); 307| 233k| break; 308| 0| case DW_EH_PE_datarel: ------------------ | Branch (308:5): [True: 0, False: 9.28M] ------------------ 309| 0| assert((base != 0) && "DW_EH_PE_datarel is invalid with a base of 0"); ------------------ | Branch (309:9): [True: 0, False: 0] | Branch (309:9): [True: 0, Folded] | Branch (309:9): [True: 0, False: 0] ------------------ 310| 0| if (result) ------------------ | Branch (310:13): [True: 0, False: 0] ------------------ 311| 0| result += base; 312| 0| break; 313| 0| case DW_EH_PE_textrel: ------------------ | Branch (313:5): [True: 0, False: 9.28M] ------------------ 314| 0| case DW_EH_PE_funcrel: ------------------ | Branch (314:5): [True: 0, False: 9.28M] ------------------ 315| 0| case DW_EH_PE_aligned: ------------------ | Branch (315:5): [True: 0, False: 9.28M] ------------------ 316| 0| default: ------------------ | Branch (316:5): [True: 0, False: 9.28M] ------------------ 317| | // not supported 318| 0| abort(); 319| 0| break; 320| 9.28M| } 321| | // then apply indirection 322| 9.28M| if (result && (encoding & DW_EH_PE_indirect)) ------------------ | Branch (322:9): [True: 7.49M, False: 1.79M] | Branch (322:19): [True: 233k, False: 7.25M] ------------------ 323| 233k| result = *((uintptr_t*)result); 324| 9.28M| *data = p; 325| 9.28M| return result; 326| 9.28M|} cxa_personality.cpp:_ZN10__cxxabiv112_GLOBAL__N_117readPointerHelperIiEEmRPKh: 163| 233k|uintptr_t readPointerHelper(const uint8_t*& p) { 164| 233k| AsType value; 165| 233k| memcpy(&value, p, sizeof(AsType)); 166| 233k| p += sizeof(AsType); 167| 233k| return static_cast(value); 168| 233k|} cxa_personality.cpp:_ZN10__cxxabiv1L11readULEB128EPPKh: 208| 13.0M|{ 209| 13.0M| uintptr_t result = 0; 210| 13.0M| uintptr_t shift = 0; 211| 13.0M| unsigned char byte; 212| 13.0M| const uint8_t *p = *data; 213| 13.0M| do 214| 17.6M| { 215| 17.6M| byte = *p++; 216| 17.6M| result |= static_cast(byte & 0x7F) << shift; 217| 17.6M| shift += 7; 218| 17.6M| } while (byte & 0x80); ------------------ | Branch (218:14): [True: 4.58M, False: 13.0M] ------------------ 219| 13.0M| *data = p; 220| 13.0M| return result; 221| 13.0M|} _ZN10__cxxabiv130__aligned_malloc_with_fallbackEm: 258| 172k|void* __aligned_malloc_with_fallback(size_t size) { 259| |#if defined(_WIN32) 260| | if (void* dest = std::__libcpp_aligned_alloc(alignof(__aligned_type), size)) 261| | return dest; 262| |#elif defined(_LIBCPP_HAS_NO_LIBRARY_ALIGNED_ALLOCATION) 263| | if (void* dest = ::malloc(size)) 264| | return dest; 265| |#else 266| 172k| if (size == 0) ------------------ | Branch (266:7): [True: 0, False: 172k] ------------------ 267| 0| size = 1; 268| 172k| if (void* dest = std::__libcpp_aligned_alloc(__alignof(__aligned_type), size)) ------------------ | Branch (268:13): [True: 172k, False: 2] ------------------ 269| 172k| return dest; 270| 2|#endif 271| 2| return fallback_malloc(size); 272| 172k|} _ZN10__cxxabiv122__calloc_with_fallbackEmm: 274| 32|void* __calloc_with_fallback(size_t count, size_t size) { 275| 32| void* ptr = ::calloc(count, size); 276| 32| if (NULL != ptr) ------------------ | Branch (276:7): [True: 32, False: 0] ------------------ 277| 32| return ptr; 278| | // if calloc fails, fall back to emergency stash 279| 0| ptr = fallback_malloc(size * count); 280| 0| if (NULL != ptr) ------------------ | Branch (280:7): [True: 0, False: 0] ------------------ 281| 0| ::memset(ptr, 0, size * count); 282| 0| return ptr; 283| 32|} _ZN10__cxxabiv128__aligned_free_with_fallbackEPv: 285| 172k|void __aligned_free_with_fallback(void* ptr) { 286| 172k| if (is_fallback_ptr(ptr)) ------------------ | Branch (286:7): [True: 0, False: 172k] ------------------ 287| 0| fallback_free(ptr); 288| 172k| else { 289| |#if defined(_LIBCPP_HAS_NO_LIBRARY_ALIGNED_ALLOCATION) 290| | ::free(ptr); 291| |#else 292| 172k| std::__libcpp_aligned_free(ptr); 293| 172k|#endif 294| 172k| } 295| 172k|} _ZN10__cxxabiv120__free_with_fallbackEPv: 297| 31|void __free_with_fallback(void* ptr) { 298| 31| if (is_fallback_ptr(ptr)) ------------------ | Branch (298:7): [True: 0, False: 31] ------------------ 299| 0| fallback_free(ptr); 300| 31| else 301| 31| ::free(ptr); 302| 31|} fallback_malloc.cpp:_ZN12_GLOBAL__N_115is_fallback_ptrEPv: 128| 172k|bool is_fallback_ptr(void* ptr) { 129| 172k| return ptr >= heap && ptr < (heap + HEAP_SIZE); ------------------ | Branch (129:10): [True: 172k, False: 0] | Branch (129:25): [True: 0, False: 172k] ------------------ 130| 172k|} _ZNK10__cxxabiv117__class_type_info9can_catchEPKNS_16__shim_type_infoERPv: 464| 233k|{ 465| | // bullet 1 466| 233k| if (is_equal(this, thrown_type, false)) ------------------ | Branch (466:9): [True: 126k, False: 106k] ------------------ 467| 126k| return true; 468| 106k| const __class_type_info* thrown_class_type = 469| 106k| dynamic_cast(thrown_type); 470| 106k| if (thrown_class_type == 0) ------------------ | Branch (470:9): [True: 0, False: 106k] ------------------ 471| 0| return false; 472| | // bullet 2 473| 106k| assert(adjustedPtr && "catching a class without an object?"); ------------------ | Branch (473:5): [True: 106k, False: 8] | Branch (473:5): [True: 106k, Folded] | Branch (473:5): [True: 106k, False: 8] ------------------ 474| 106k| __dynamic_cast_info info = {thrown_class_type, 0, this, -1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, true, nullptr}; 475| 106k| info.number_of_dst_type = 1; 476| 106k| thrown_class_type->has_unambiguous_public_base(&info, adjustedPtr, public_path); 477| 106k| if (info.path_dst_ptr_to_static_ptr == public_path) ------------------ | Branch (477:9): [True: 45.5k, False: 60.8k] ------------------ 478| 45.5k| { 479| 45.5k| adjustedPtr = const_cast(info.dst_ptr_leading_to_static_ptr); 480| 45.5k| return true; 481| 45.5k| } 482| 60.8k| return false; 483| 106k|} _ZNK10__cxxabiv117__class_type_info24process_found_base_classEPNS_19__dynamic_cast_infoEPvi: 509| 45.5k|{ 510| 45.5k| if (info->number_to_static_ptr == 0) { ------------------ | Branch (510:7): [True: 45.5k, False: 0] ------------------ 511| | // First time we found this base 512| 45.5k| info->dst_ptr_leading_to_static_ptr = adjustedPtr; 513| 45.5k| info->path_dst_ptr_to_static_ptr = path_below; 514| | // stash the virtual base cookie. 515| 45.5k| info->dst_ptr_not_leading_to_static_ptr = info->vbase_cookie; 516| 45.5k| info->number_to_static_ptr = 1; 517| 45.5k| } else if (info->dst_ptr_not_leading_to_static_ptr == info->vbase_cookie && ------------------ | Branch (517:14): [True: 0, False: 0] ------------------ 518| 0| info->dst_ptr_leading_to_static_ptr == adjustedPtr) { ------------------ | Branch (518:14): [True: 0, False: 0] ------------------ 519| | // We've been here before. Update path to "most public" 520| 0| if (info->path_dst_ptr_to_static_ptr == not_public_path) ------------------ | Branch (520:9): [True: 0, False: 0] ------------------ 521| 0| info->path_dst_ptr_to_static_ptr = path_below; 522| 0| } else { 523| | // We've detected an ambiguous cast from (thrown_class_type, adjustedPtr) 524| | // to a static_type. 525| 0| info->number_to_static_ptr += 1; 526| 0| info->path_dst_ptr_to_static_ptr = not_public_path; 527| 0| info->search_done = true; 528| 0| } 529| 45.5k|} _ZNK10__cxxabiv117__class_type_info27has_unambiguous_public_baseEPNS_19__dynamic_cast_infoEPvi: 535| 60.8k|{ 536| 60.8k| if (is_equal(this, info->static_type, false)) ------------------ | Branch (536:9): [True: 0, False: 60.8k] ------------------ 537| 0| process_found_base_class(info, adjustedPtr, path_below); 538| 60.8k|} _ZNK10__cxxabiv120__si_class_type_info27has_unambiguous_public_baseEPNS_19__dynamic_cast_infoEPvi: 544| 297k|{ 545| 297k| if (is_equal(this, info->static_type, false)) ------------------ | Branch (545:9): [True: 45.5k, False: 252k] ------------------ 546| 45.5k| process_found_base_class(info, adjustedPtr, path_below); 547| 252k| else 548| 252k| __base_type->has_unambiguous_public_base(info, adjustedPtr, path_below); 549| 297k|} __dynamic_cast: 899| 106k| std::ptrdiff_t src2dst_offset) { 900| | // Get (dynamic_ptr, dynamic_type) from static_ptr 901| 106k| derived_object_info derived_info; 902| 106k| dyn_cast_get_derived_info(&derived_info, static_ptr); 903| | 904| | // Initialize answer to nullptr. This will be changed from the search 905| | // results if a non-null answer is found. Regardless, this is what will 906| | // be returned. 907| 106k| const void* dst_ptr = 0; 908| | 909| | // Find out if we can use a giant short cut in the search 910| 106k| if (is_equal(derived_info.dynamic_type, dst_type, false)) ------------------ | Branch (910:9): [True: 0, False: 106k] ------------------ 911| 0| { 912| 0| dst_ptr = dyn_cast_to_derived(static_ptr, 913| 0| derived_info.dynamic_ptr, 914| 0| static_type, 915| 0| dst_type, 916| 0| derived_info.offset_to_derived, 917| 0| src2dst_offset); 918| 0| } 919| 106k| else 920| 106k| { 921| | // Optimize toward downcasting: let's first try to do a downcast before 922| | // falling back to the slow path. 923| 106k| dst_ptr = dyn_cast_try_downcast(static_ptr, 924| 106k| derived_info.dynamic_ptr, 925| 106k| dst_type, 926| 106k| derived_info.dynamic_type, 927| 106k| src2dst_offset); 928| | 929| 106k| if (!dst_ptr) ------------------ | Branch (929:13): [True: 0, False: 106k] ------------------ 930| 0| { 931| 0| dst_ptr = dyn_cast_slow(static_ptr, 932| 0| derived_info.dynamic_ptr, 933| 0| static_type, 934| 0| dst_type, 935| 0| derived_info.dynamic_type, 936| 0| src2dst_offset); 937| 0| } 938| 106k| } 939| | 940| 106k| return const_cast(dst_ptr); 941| 106k|} _ZNK10__cxxabiv117__class_type_info29process_static_type_above_dstEPNS_19__dynamic_cast_infoEPKvS4_i: 961| 106k|{ 962| | // Record that we found a static_type 963| 106k| info->found_any_static_type = true; 964| 106k| if (current_ptr == info->static_ptr) ------------------ | Branch (964:9): [True: 106k, False: 2] ------------------ 965| 106k| { 966| | // Record that we found (static_ptr, static_type) 967| 106k| info->found_our_static_ptr = true; 968| 106k| if (info->dst_ptr_leading_to_static_ptr == 0) ------------------ | Branch (968:13): [True: 106k, False: 1] ------------------ 969| 106k| { 970| | // First time here 971| 106k| info->dst_ptr_leading_to_static_ptr = dst_ptr; 972| 106k| info->path_dst_ptr_to_static_ptr = path_below; 973| 106k| info->number_to_static_ptr = 1; 974| | // If there is only one dst_type in the entire tree and the path from 975| | // there to here is public then we are done! 976| 106k| if (info->number_of_dst_type == 1 && info->path_dst_ptr_to_static_ptr == public_path) ------------------ | Branch (976:17): [True: 106k, False: 18.4E] | Branch (976:50): [True: 106k, False: 0] ------------------ 977| 106k| info->search_done = true; 978| 106k| } 979| 1| else if (info->dst_ptr_leading_to_static_ptr == dst_ptr) ------------------ | Branch (979:18): [True: 0, False: 1] ------------------ 980| 0| { 981| | // We've been here before. Update path to "most public" 982| 0| if (info->path_dst_ptr_to_static_ptr == not_public_path) ------------------ | Branch (982:17): [True: 0, False: 0] ------------------ 983| 0| info->path_dst_ptr_to_static_ptr = path_below; 984| | // If there is only one dst_type in the entire tree and the path from 985| | // there to here is public then we are done! 986| 0| if (info->number_of_dst_type == 1 && info->path_dst_ptr_to_static_ptr == public_path) ------------------ | Branch (986:17): [True: 0, False: 0] | Branch (986:50): [True: 0, False: 0] ------------------ 987| 0| info->search_done = true; 988| 0| } 989| 1| else 990| 1| { 991| | // We've detected an ambiguous cast from (static_ptr, static_type) 992| | // to a dst_type 993| 1| info->number_to_static_ptr += 1; 994| 1| info->search_done = true; 995| 1| } 996| 106k| } 997| 106k|} _ZNK10__cxxabiv120__si_class_type_info16search_above_dstEPNS_19__dynamic_cast_infoEPKvS4_ib: 1469| 212k|{ 1470| 212k| if (is_equal(this, info->static_type, use_strcmp)) ------------------ | Branch (1470:9): [True: 106k, False: 106k] ------------------ 1471| 106k| process_static_type_above_dst(info, dst_ptr, current_ptr, path_below); 1472| 106k| else 1473| 106k| __base_type->search_above_dst(info, dst_ptr, current_ptr, path_below, use_strcmp); 1474| 212k|} private_typeinfo.cpp:_ZL8is_equalPKSt9type_infoS1_b: 57| 910k|{ 58| | // Use std::type_info's default comparison unless we've explicitly asked 59| | // for strcmp. 60| 910k| if (!use_strcmp) ------------------ | Branch (60:9): [True: 910k, False: 18.4E] ------------------ 61| 910k| return *x == *y; 62| | // Still allow pointer equality to short circut. 63| 18.4E| return x == y || strcmp(x->name(), y->name()) == 0; ------------------ | Branch (63:12): [True: 18.4E, False: 0] | Branch (63:22): [True: 0, False: 0] ------------------ 64| 910k|} private_typeinfo.cpp:_ZN10__cxxabiv112_GLOBAL__N_125dyn_cast_get_derived_infoEPNS0_19derived_object_infoEPKv: 89| 106k|{ 90| |#if __has_feature(cxx_abi_relative_vtable) 91| | // The vtable address will point to the first virtual function, which is 8 92| | // bytes after the start of the vtable (4 for the offset from top + 4 for 93| | // the typeinfo component). 94| | const int32_t* vtable = 95| | *reinterpret_cast(static_ptr); 96| | info->offset_to_derived = static_cast(vtable[-2]); 97| | info->dynamic_ptr = static_cast(static_ptr) + info->offset_to_derived; 98| | 99| | // The typeinfo component is now a relative offset to a proxy. 100| | int32_t offset_to_ti_proxy = vtable[-1]; 101| | const uint8_t* ptr_to_ti_proxy = 102| | reinterpret_cast(vtable) + offset_to_ti_proxy; 103| | info->dynamic_type = *(reinterpret_cast(ptr_to_ti_proxy)); 104| |#else 105| 106k| void **vtable = *static_cast(static_ptr); 106| 106k| info->offset_to_derived = reinterpret_cast(vtable[-2]); 107| 106k| info->dynamic_ptr = static_cast(static_ptr) + info->offset_to_derived; 108| 106k| info->dynamic_type = static_cast(vtable[-1]); 109| 106k|#endif 110| 106k|} private_typeinfo.cpp:_ZN10__cxxabiv112_GLOBAL__N_121dyn_cast_try_downcastEPKvS2_PKNS_17__class_type_infoES5_l: 205| 106k|{ 206| 106k| if (src2dst_offset < 0) ------------------ | Branch (206:9): [True: 0, False: 106k] ------------------ 207| 0| { 208| | // We can only optimize the case if the static type is a unique public 209| | // base of dst_type. Give up. 210| 0| return nullptr; 211| 0| } 212| | 213| | // Pretend there is a dst_type object that leads to static_ptr. Later we 214| | // will check whether this imagined dst_type object exists. If it exists 215| | // then it will be the casting result. 216| 106k| const void* dst_ptr_to_static = reinterpret_cast(static_ptr) - src2dst_offset; 217| | 218| 106k| if (reinterpret_cast(dst_ptr_to_static) < reinterpret_cast(dynamic_ptr)) ------------------ | Branch (218:9): [True: 0, False: 106k] ------------------ 219| 0| { 220| | // The imagined dst_type object does not exist. Bail-out quickly. 221| 0| return nullptr; 222| 0| } 223| | 224| | // Try to search a path from dynamic_type to dst_type. 225| 106k| __dynamic_cast_info dynamic_to_dst_info = {dynamic_type, 226| 106k| dst_ptr_to_static, 227| 106k| dst_type, 228| 106k| src2dst_offset, 229| 106k| 0, 230| 106k| 0, 231| 106k| 0, 232| 106k| 0, 233| 106k| 0, 234| 106k| 0, 235| 106k| 0, 236| 106k| 0, 237| 106k| 1, // number_of_dst_type 238| 106k| false, 239| 106k| false, 240| 106k| false, 241| 106k| true, 242| 106k| nullptr}; 243| 106k| dynamic_type->search_above_dst(&dynamic_to_dst_info, dynamic_ptr, dynamic_ptr, public_path, false); 244| 106k| if (dynamic_to_dst_info.path_dst_ptr_to_static_ptr != unknown) { ------------------ | Branch (244:9): [True: 106k, False: 18.4E] ------------------ 245| | // We have found at least one path from dynamic_ptr to dst_ptr. The 246| | // downcast can succeed. 247| 106k| return dst_ptr_to_static; 248| 106k| } 249| | 250| 18.4E| return nullptr; 251| 106k|} _ZNSt9exceptionD2Ev: 18| 172k|{ 19| 172k|} _Znwm: 66| 863k|_LIBCPP_MAKE_OVERRIDABLE_FUNCTION_DETECTABLE _LIBCPP_WEAK void* operator new(std::size_t size) _THROW_BAD_ALLOC { 67| 863k| void* p = operator_new_impl(size); 68| 863k| if (p == nullptr) ------------------ | Branch (68:7): [True: 0, False: 863k] ------------------ 69| 0| __throw_bad_alloc_shim(); 70| 863k| return p; 71| 863k|} _ZdlPv: 125| 863k|_LIBCPP_WEAK void operator delete(void* ptr) noexcept { std::free(ptr); } _ZdlPvm: 129| 691k|_LIBCPP_WEAK void operator delete(void* ptr, size_t) noexcept { ::operator delete(ptr); } _ZnwmSt11align_val_t: 160| 7.93k|operator new(std::size_t size, std::align_val_t alignment) _THROW_BAD_ALLOC { 161| 7.93k| void* p = operator_new_aligned_impl(size, alignment); 162| 7.93k| if (p == nullptr) ------------------ | Branch (162:7): [True: 0, False: 7.93k] ------------------ 163| 0| __throw_bad_alloc_shim(); 164| 7.93k| return p; 165| 7.93k|} _ZdlPvSt11align_val_t: 221| 7.93k|_LIBCPP_WEAK void operator delete(void* ptr, std::align_val_t) noexcept { std::__libcpp_aligned_free(ptr); } stdlib_new_delete.cpp:_ZL17operator_new_implm: 50| 863k|static void* operator_new_impl(std::size_t size) { 51| 863k| if (size == 0) ------------------ | Branch (51:7): [True: 0, False: 863k] ------------------ 52| 0| size = 1; 53| 863k| void* p; 54| 863k| while ((p = std::malloc(size)) == nullptr) { ------------------ | Branch (54:10): [True: 0, False: 863k] ------------------ 55| | // If malloc fails and there is a new_handler, 56| | // call it to try free up memory. 57| 0| std::new_handler nh = std::get_new_handler(); 58| 0| if (nh) ------------------ | Branch (58:9): [True: 0, False: 0] ------------------ 59| 0| nh(); 60| 0| else 61| 0| break; 62| 0| } 63| 863k| return p; 64| 863k|} stdlib_new_delete.cpp:_ZL25operator_new_aligned_implmSt11align_val_t: 139| 7.93k|static void* operator_new_aligned_impl(std::size_t size, std::align_val_t alignment) { 140| 7.93k| if (size == 0) ------------------ | Branch (140:7): [True: 0, False: 7.93k] ------------------ 141| 0| size = 1; 142| 7.93k| if (static_cast(alignment) < sizeof(void*)) ------------------ | Branch (142:7): [True: 0, False: 7.93k] ------------------ 143| 0| alignment = std::align_val_t(sizeof(void*)); 144| | 145| | // Try allocating memory. If allocation fails and there is a new_handler, 146| | // call it to try free up memory, and try again until it succeeds, or until 147| | // the new_handler decides to terminate. 148| 7.93k| void* p; 149| 7.93k| while ((p = std::__libcpp_aligned_alloc(static_cast(alignment), size)) == nullptr) { ------------------ | Branch (149:10): [True: 0, False: 7.93k] ------------------ 150| 0| std::new_handler nh = std::get_new_handler(); 151| 0| if (nh) ------------------ | Branch (151:9): [True: 0, False: 0] ------------------ 152| 0| nh(); 153| 0| else 154| 0| break; 155| 0| } 156| 7.93k| return p; 157| 7.93k|} _ZNSt13runtime_errorD2Ev: 30| 172k|runtime_error::~runtime_error() noexcept {} _ZNKSt13runtime_error4whatEv: 34| 21.5k|{ 35| 21.5k| return __imp_.c_str(); 36| 21.5k|} kmp_csupport.cpp:_ZL23__kmp_assert_valid_gtidi: 3510| 47.3k|static inline void __kmp_assert_valid_gtid(kmp_int32 gtid) { 3511| 47.3k| if (UNLIKELY(gtid < 0 || gtid >= __kmp_threads_capacity)) ------------------ | | 119| 94.7k|#define UNLIKELY(x) __builtin_expect(!!(x), 0) | | ------------------ | | | Branch (119:21): [True: 0, False: 47.3k] | | | Branch (119:41): [True: 26, False: 47.3k] | | | Branch (119:41): [True: 18.4E, False: 47.3k] | | ------------------ ------------------ 3512| 0| KMP_FATAL(ThreadIdentInvalid); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 3513| 47.3k|} kmp_csupport.cpp:_ZL31__kmp_clear_x87_fpu_status_wordv: 1365| 3.01k|static inline void __kmp_clear_x87_fpu_status_word() { 1366| |#if KMP_MIC 1367| | // 32-bit protected mode x87 FPU state 1368| | struct x87_fpu_state { 1369| | unsigned cw; 1370| | unsigned sw; 1371| | unsigned tw; 1372| | unsigned fip; 1373| | unsigned fips; 1374| | unsigned fdp; 1375| | unsigned fds; 1376| | }; 1377| | struct x87_fpu_state fpu_state = {0, 0, 0, 0, 0, 0, 0}; 1378| | __asm__ __volatile__("fstenv %0\n\t" // store FP env 1379| | "andw $0x7f00, %1\n\t" // clear 0-7,15 bits of FP SW 1380| | "fldenv %0\n\t" // load FP env back 1381| | : "+m"(fpu_state), "+m"(fpu_state.sw)); 1382| |#else 1383| 3.01k| __asm__ __volatile__("fnclex"); 1384| 3.01k|#endif // KMP_MIC 1385| 3.01k|} kmp_csupport.cpp:_ZL31__kmp_load_x87_fpu_control_wordPKs: 1358| 3.01k|static inline void __kmp_load_x87_fpu_control_word(const kmp_int16 *p) { 1359| 3.01k| __asm__ __volatile__("fldcw %0" : : "m"(*p)); 1360| 3.01k|} kmp_csupport.cpp:_ZL16__kmp_load_mxcsrPKj: 1387| 3.01k|static inline void __kmp_load_mxcsr(const kmp_uint32 *p) { _mm_setcsr(*p); } _ZN11KMPAffinity4MaskD2Ev: 724| 68| virtual ~Mask() {} _ZNK11KMPAffinity4Mask11get_max_cpuEv: 756| 1| int get_max_cpu() const { 757| 1| int cpu; 758| 1| int max_cpu = -1; 759| 32| KMP_CPU_SET_ITERATE(cpu, this) { ------------------ | | 688| 33| for (i = (mask)->begin(); (int)i != (mask)->end(); i = (mask)->next(i)) | | ------------------ | | | Branch (688:29): [True: 32, False: 1] | | ------------------ ------------------ 760| 32| if (cpu > max_cpu) ------------------ | Branch (760:13): [True: 32, False: 0] ------------------ 761| 32| max_cpu = cpu; 762| 32| } 763| 1| return max_cpu; 764| 1| } _ZN19kmp_affinity_raii_tC2EPKN11KMPAffinity4MaskE: 807| 2| : restored(false) { 808| 2| if (KMP_AFFINITY_CAPABLE()) { ------------------ | | 684| 2|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) | | ------------------ | | | Branch (684:32): [True: 2, False: 0] | | ------------------ ------------------ 809| 2| KMP_CPU_ALLOC(mask); ------------------ | | 697| 2|#define KMP_CPU_ALLOC(ptr) (ptr = __kmp_affinity_dispatch->allocate_mask()) ------------------ 810| 2| KMP_ASSERT(mask != NULL); ------------------ | | 50| 2| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 2] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 811| 2| __kmp_get_system_affinity(mask, /*abort_on_error=*/true); ------------------ | | 711| 2| (mask)->get_system_affinity(abort_bool) ------------------ 812| 2| if (new_mask) ------------------ | Branch (812:11): [True: 1, False: 1] ------------------ 813| 1| __kmp_set_system_affinity(new_mask, /*abort_on_error=*/true); ------------------ | | 713| 1| (mask)->set_system_affinity(abort_bool) ------------------ 814| 2| } 815| 2| } _ZN19kmp_affinity_raii_t7restoreEv: 816| 3| void restore() { 817| 3| if (!restored && KMP_AFFINITY_CAPABLE()) { ------------------ | | 684| 2|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) | | ------------------ | | | Branch (684:32): [True: 2, False: 0] | | ------------------ ------------------ | Branch (817:9): [True: 2, False: 1] ------------------ 818| 2| __kmp_set_system_affinity(mask, /*abort_on_error=*/true); ------------------ | | 713| 2| (mask)->set_system_affinity(abort_bool) ------------------ 819| 2| KMP_CPU_FREE(mask); ------------------ | | 698| 2|#define KMP_CPU_FREE(ptr) __kmp_affinity_dispatch->deallocate_mask(ptr) ------------------ 820| 2| } 821| 3| restored = true; 822| 3| } _ZN19kmp_affinity_raii_tD2Ev: 823| 2| ~kmp_affinity_raii_t() { restore(); } _ZNK18kmp_affinity_ids_tixEm: 875| 384| int operator[](size_t idx) const { return ids[idx]; } _ZN18kmp_affinity_ids_tixEm: 876| 650| int &operator[](size_t idx) { return ids[idx]; } _ZN18kmp_affinity_ids_taSERKS_: 877| 32| kmp_affinity_ids_t &operator=(const kmp_affinity_ids_t &rhs) { 878| 416| for (int i = 0; i < KMP_HW_LAST; ++i) ------------------ | Branch (878:21): [True: 384, False: 32] ------------------ 879| 384| ids[i] = rhs[i]; 880| 32| return *this; 881| 32| } _ZN11kmp_convertImiLb0ELb0ELb0ELb1EE2toEm: 4667| 460| static TargetType to(SourceType src) { 4668| 460| KMP_ASSERT(src <= static_cast( ------------------ | | 50| 460| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 460] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4669| 460| (std::numeric_limits::max)())); 4670| 460| return (TargetType)src; 4671| 460| } kmp_runtime.cpp:_ZL22__kmp_gtid_from_threadPK8kmp_info: 3495| 32|static inline int __kmp_gtid_from_thread(const kmp_info_t *thr) { 3496| 32| KMP_DEBUG_ASSERT(thr); 3497| 32| return thr->th.th_info.ds.ds_gtid; 3498| 32|} kmp_runtime.cpp:_ZL13KMP_UBER_GTIDi: 3478| 4|static inline bool KMP_UBER_GTID(int gtid) { 3479| 4| KMP_DEBUG_ASSERT(gtid >= KMP_GTID_MIN); 3480| 4| KMP_DEBUG_ASSERT(gtid < __kmp_threads_capacity); 3481| 4| return (gtid >= 0 && __kmp_root[gtid] && __kmp_threads[gtid] && ------------------ | Branch (3481:11): [True: 4, False: 0] | Branch (3481:24): [True: 4, False: 0] | Branch (3481:44): [True: 4, False: 0] ------------------ 3482| 4| __kmp_threads[gtid] == __kmp_root[gtid]->r.r_uber_thread); ------------------ | Branch (3482:11): [True: 4, False: 0] ------------------ 3483| 4|} kmp_runtime.cpp:_ZL19__kmp_tid_from_gtidi: 3485| 45.2k|static inline int __kmp_tid_from_gtid(int gtid) { 3486| 45.2k| KMP_DEBUG_ASSERT(gtid >= 0); 3487| 45.2k| return __kmp_threads[gtid]->th.th_info.ds.ds_tid; 3488| 45.2k|} kmp_runtime.cpp:_ZL9copy_icvsP20kmp_internal_controlS0_: 2111| 4.42k| kmp_internal_control_t *src) { 2112| 4.42k| *dst = *src; 2113| 4.42k|} kmp_runtime.cpp:_ZL27__kmp_assign_root_init_maskv: 3785| 1.35k|static inline void __kmp_assign_root_init_mask() { 3786| 1.35k| int gtid = __kmp_entry_gtid(); ------------------ | | 3467| 1.35k|#define __kmp_entry_gtid() __kmp_get_global_thread_id_reg() ------------------ 3787| 1.35k| kmp_root_t *r = __kmp_threads[gtid]->th.th_root; 3788| 1.35k| if (r->r.r_uber_thread == __kmp_threads[gtid] && !r->r.r_affinity_assigned) { ------------------ | Branch (3788:7): [True: 1.35k, False: 0] | Branch (3788:52): [True: 0, False: 1.35k] ------------------ 3789| 0| __kmp_affinity_set_init_mask(gtid, TRUE); ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 3790| 0| r->r.r_affinity_assigned = TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 3791| 0| } 3792| 1.35k|} kmp_runtime.cpp:_ZL19__kmp_gtid_from_tidiPK8kmp_team: 3490| 32|static inline int __kmp_gtid_from_tid(int tid, const kmp_team_t *team) { 3491| 32| KMP_DEBUG_ASSERT(tid >= 0 && team); 3492| 32| return team->t.t_threads[tid]->th.th_info.ds.ds_gtid; 3493| 32|} kmp_runtime.cpp:_ZL15__kmp_x86_cpuidiiP9kmp_cpuid: 1352| 2|static inline void __kmp_x86_cpuid(int leaf, int subleaf, struct kmp_cpuid *p) { 1353| 2| __asm__ __volatile__("cpuid" 1354| 2| : "=a"(p->eax), "=b"(p->ebx), "=c"(p->ecx), "=d"(p->edx) 1355| 2| : "a"(leaf), "c"(subleaf)); 1356| 2|} kmp_runtime.cpp:_ZL27__kmp_resume_if_hard_pausedv: 4380| 1|static inline void __kmp_resume_if_hard_paused() { 4381| 1| if (__kmp_pause_status == kmp_hard_paused) { ------------------ | Branch (4381:7): [True: 0, False: 1] ------------------ 4382| 0| __kmp_pause_status = kmp_not_paused; 4383| 0| } 4384| 1|} kmp_runtime.cpp:_ZL32__kmp_store_x87_fpu_control_wordPs: 1362| 46.9k|static inline void __kmp_store_x87_fpu_control_word(kmp_int16 *p) { 1363| 46.9k| __asm__ __volatile__("fstcw %0" : "=m"(*p)); 1364| 46.9k|} kmp_runtime.cpp:_ZL17__kmp_store_mxcsrPj: 1388| 46.9k|static inline void __kmp_store_mxcsr(kmp_uint32 *p) { *p = _mm_getcsr(); } kmp_runtime.cpp:_ZL19__kmp_is_hybrid_cpuv: 3226| 1|static inline bool __kmp_is_hybrid_cpu() { return __kmp_cpuinfo.flags.hybrid; } kmp_runtime.cpp:_ZL18__kmp_type_convertIliEvT_PT0_: 4696| 31|static inline void __kmp_type_convert(T1 src, T2 *dest) { 4697| 31| *dest = kmp_convert::to(src); 4698| 31|} _ZN11kmp_convertIliLb0ELb0ELb1ELb1EE2toEl: 4612| 33| static TargetType to(SourceType src) { 4613| 33| KMP_ASSERT(src <= static_cast( ------------------ | | 50| 33| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 33] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4614| 33| (std::numeric_limits::max)())); 4615| 33| KMP_ASSERT(src >= static_cast( ------------------ | | 50| 33| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 33] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4616| 33| (std::numeric_limits::min)())); 4617| 33| return (TargetType)src; 4618| 33| } kmp_str.cpp:_ZL18__kmp_type_convertImiEvT_PT0_: 4696| 68|static inline void __kmp_type_convert(T1 src, T2 *dest) { 4697| 68| *dest = kmp_convert::to(src); 4698| 68|} kmp_utility.cpp:_ZL15__kmp_x86_cpuidiiP9kmp_cpuid: 1352| 6|static inline void __kmp_x86_cpuid(int leaf, int subleaf, struct kmp_cpuid *p) { 1353| 6| __asm__ __volatile__("cpuid" 1354| 6| : "=a"(p->eax), "=b"(p->ebx), "=c"(p->ecx), "=d"(p->edx) 1355| 6| : "a"(leaf), "c"(subleaf)); 1356| 6|} kmp_barrier.cpp:_ZL19__kmp_tid_from_gtidi: 3485| 194k|static inline int __kmp_tid_from_gtid(int gtid) { 3486| 194k| KMP_DEBUG_ASSERT(gtid >= 0); 3487| 194k| return __kmp_threads[gtid]->th.th_info.ds.ds_tid; 3488| 194k|} kmp_barrier.cpp:_ZL15__kmp_x86_pausev: 1496| 426M|static inline void __kmp_x86_pause(void) { _mm_pause(); } kmp_barrier.cpp:_ZL9copy_icvsP20kmp_internal_controlS0_: 2111| 81.9k| kmp_internal_control_t *src) { 2112| 81.9k| *dst = *src; 2113| 81.9k|} kmp_affinity.cpp:_ZL19__kmp_is_hybrid_cpuv: 3226| 64|static inline bool __kmp_is_hybrid_cpu() { return __kmp_cpuinfo.flags.hybrid; } _ZN11KMPAffinityD2Ev: 769| 1| virtual ~KMPAffinity() = default; kmp_affinity.cpp:_ZL18__kmp_type_convertImiEvT_PT0_: 4696| 392|static inline void __kmp_type_convert(T1 src, T2 *dest) { 4697| 392| *dest = kmp_convert::to(src); 4698| 392|} kmp_affinity.cpp:_ZL15__kmp_x86_cpuidiiP9kmp_cpuid: 1352| 134|static inline void __kmp_x86_cpuid(int leaf, int subleaf, struct kmp_cpuid *p) { 1353| 134| __asm__ __volatile__("cpuid" 1354| 134| : "=a"(p->eax), "=b"(p->ebx), "=c"(p->ecx), "=d"(p->edx) 1355| 134| : "a"(leaf), "c"(subleaf)); 1356| 134|} kmp_dispatch.cpp:_ZL15__kmp_x86_pausev: 1496| 11.9M|static inline void __kmp_x86_pause(void) { _mm_pause(); } kmp_lock.cpp:_ZL15__kmp_x86_pausev: 1496| 64.4k|static inline void __kmp_x86_pause(void) { _mm_pause(); } kmp_lock.cpp:_ZL22__kmp_thread_from_gtidi: 3500| 29.0k|static inline kmp_info_t *__kmp_thread_from_gtid(int gtid) { 3501| 29.0k| KMP_DEBUG_ASSERT(gtid >= 0); 3502| 29.0k| return __kmp_threads[gtid]; 3503| 29.0k|} kmp_sched.cpp:_ZL23__kmp_assert_valid_gtidi: 3510| 45.0k|static inline void __kmp_assert_valid_gtid(kmp_int32 gtid) { 3511| 45.0k| if (UNLIKELY(gtid < 0 || gtid >= __kmp_threads_capacity)) ------------------ | | 119| 90.2k|#define UNLIKELY(x) __builtin_expect(!!(x), 0) | | ------------------ | | | Branch (119:21): [True: 0, False: 45.0k] | | | Branch (119:41): [True: 18.4E, False: 45.1k] | | | Branch (119:41): [True: 18.4E, False: 45.1k] | | ------------------ ------------------ 3512| 0| KMP_FATAL(ThreadIdentInvalid); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 3513| 45.0k|} kmp_sched.cpp:_ZL19__kmp_tid_from_gtidi: 3485| 45.4k|static inline int __kmp_tid_from_gtid(int gtid) { 3486| 45.4k| KMP_DEBUG_ASSERT(gtid >= 0); 3487| 45.4k| return __kmp_threads[gtid]->th.th_info.ds.ds_tid; 3488| 45.4k|} z_Linux_util.cpp:_ZL13KMP_UBER_GTIDi: 3478| 64|static inline bool KMP_UBER_GTID(int gtid) { 3479| 64| KMP_DEBUG_ASSERT(gtid >= KMP_GTID_MIN); 3480| 64| KMP_DEBUG_ASSERT(gtid < __kmp_threads_capacity); 3481| 64| return (gtid >= 0 && __kmp_root[gtid] && __kmp_threads[gtid] && ------------------ | Branch (3481:11): [True: 64, False: 0] | Branch (3481:24): [True: 2, False: 62] | Branch (3481:44): [True: 2, False: 0] ------------------ 3482| 2| __kmp_threads[gtid] == __kmp_root[gtid]->r.r_uber_thread); ------------------ | Branch (3482:11): [True: 2, False: 0] ------------------ 3483| 64|} z_Linux_util.cpp:_ZL31__kmp_clear_x87_fpu_status_wordv: 1365| 31|static inline void __kmp_clear_x87_fpu_status_word() { 1366| |#if KMP_MIC 1367| | // 32-bit protected mode x87 FPU state 1368| | struct x87_fpu_state { 1369| | unsigned cw; 1370| | unsigned sw; 1371| | unsigned tw; 1372| | unsigned fip; 1373| | unsigned fips; 1374| | unsigned fdp; 1375| | unsigned fds; 1376| | }; 1377| | struct x87_fpu_state fpu_state = {0, 0, 0, 0, 0, 0, 0}; 1378| | __asm__ __volatile__("fstenv %0\n\t" // store FP env 1379| | "andw $0x7f00, %1\n\t" // clear 0-7,15 bits of FP SW 1380| | "fldenv %0\n\t" // load FP env back 1381| | : "+m"(fpu_state), "+m"(fpu_state.sw)); 1382| |#else 1383| 31| __asm__ __volatile__("fnclex"); 1384| 31|#endif // KMP_MIC 1385| 31|} z_Linux_util.cpp:_ZL31__kmp_load_x87_fpu_control_wordPKs: 1358| 31|static inline void __kmp_load_x87_fpu_control_word(const kmp_int16 *p) { 1359| 31| __asm__ __volatile__("fldcw %0" : : "m"(*p)); 1360| 31|} z_Linux_util.cpp:_ZL16__kmp_load_mxcsrPKj: 1387| 31|static inline void __kmp_load_mxcsr(const kmp_uint32 *p) { _mm_setcsr(*p); } z_Linux_util.cpp:_ZL18__kmp_type_convertIliEvT_PT0_: 4696| 2|static inline void __kmp_type_convert(T1 src, T2 *dest) { 4697| 2| *dest = kmp_convert::to(src); 4698| 2|} kmp_ftn_cdecl.cpp:_ZL27__kmp_assign_root_init_maskv: 3785| 4.36k|static inline void __kmp_assign_root_init_mask() { 3786| 4.36k| int gtid = __kmp_entry_gtid(); ------------------ | | 3467| 4.36k|#define __kmp_entry_gtid() __kmp_get_global_thread_id_reg() ------------------ 3787| 4.36k| kmp_root_t *r = __kmp_threads[gtid]->th.th_root; 3788| 4.36k| if (r->r.r_uber_thread == __kmp_threads[gtid] && !r->r.r_affinity_assigned) { ------------------ | Branch (3788:7): [True: 4.36k, False: 0] | Branch (3788:52): [True: 1, False: 4.36k] ------------------ 3789| 1| __kmp_affinity_set_init_mask(gtid, TRUE); ------------------ | | 1280| 1|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1|#define FALSE 0 | | ------------------ ------------------ 3790| 1| r->r.r_affinity_assigned = TRUE; ------------------ | | 1280| 1|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1|#define FALSE 0 | | ------------------ ------------------ 3791| 1| } 3792| 4.36k|} ompt-general.cpp:_ZL22__kmp_thread_from_gtidi: 3500| 1|static inline kmp_info_t *__kmp_thread_from_gtid(int gtid) { 3501| 1| KMP_DEBUG_ASSERT(gtid >= 0); 3502| 1| return __kmp_threads[gtid]; 3503| 1|} __kmp_cleanup_hierarchy: 39| 1|void __kmp_cleanup_hierarchy() { machine_hierarchy.fini(); } _ZN15kmp_hw_thread_t11compare_idsEPKvS1_: 154| 95|int kmp_hw_thread_t::compare_ids(const void *a, const void *b) { 155| 95| const kmp_hw_thread_t *ahwthread = (const kmp_hw_thread_t *)a; 156| 95| const kmp_hw_thread_t *bhwthread = (const kmp_hw_thread_t *)b; 157| 95| int depth = __kmp_topology->get_depth(); 158| 206| for (int level = 0; level < depth; ++level) { ------------------ | Branch (158:23): [True: 206, False: 0] ------------------ 159| 206| if (ahwthread->ids[level] < bhwthread->ids[level]) ------------------ | Branch (159:9): [True: 80, False: 126] ------------------ 160| 80| return -1; 161| 126| else if (ahwthread->ids[level] > bhwthread->ids[level]) ------------------ | Branch (161:14): [True: 15, False: 111] ------------------ 162| 15| return 1; 163| 206| } 164| 0| if (ahwthread->os_id < bhwthread->os_id) ------------------ | Branch (164:7): [True: 0, False: 0] ------------------ 165| 0| return -1; 166| 0| else if (ahwthread->os_id > bhwthread->os_id) ------------------ | Branch (166:12): [True: 0, False: 0] ------------------ 167| 0| return 1; 168| 0| return 0; 169| 0|} _ZN14kmp_topology_t21_remove_radix1_layersEv: 286| 1|void kmp_topology_t::_remove_radix1_layers() { 287| 1| int preference[KMP_HW_LAST]; 288| 1| int top_index1, top_index2; 289| | // Set up preference associative array 290| 1| preference[KMP_HW_SOCKET] = 110; 291| 1| preference[KMP_HW_PROC_GROUP] = 100; 292| 1| preference[KMP_HW_CORE] = 95; 293| 1| preference[KMP_HW_THREAD] = 90; 294| 1| preference[KMP_HW_NUMA] = 85; 295| 1| preference[KMP_HW_DIE] = 80; 296| 1| preference[KMP_HW_TILE] = 75; 297| 1| preference[KMP_HW_MODULE] = 73; 298| 1| preference[KMP_HW_L3] = 70; 299| 1| preference[KMP_HW_L2] = 65; 300| 1| preference[KMP_HW_L1] = 60; 301| 1| preference[KMP_HW_LLC] = 5; 302| 1| top_index1 = 0; 303| 1| top_index2 = 1; 304| 3| while (top_index1 < depth - 1 && top_index2 < depth) { ------------------ | Branch (304:10): [True: 2, False: 1] | Branch (304:36): [True: 2, False: 0] ------------------ 305| 2| kmp_hw_t type1 = types[top_index1]; 306| 2| kmp_hw_t type2 = types[top_index2]; 307| 2| KMP_ASSERT_VALID_HW_TYPE(type1); ------------------ | | 637| 2| KMP_ASSERT(type >= (kmp_hw_t)0 && type < KMP_HW_LAST) | | ------------------ | | | | 50| 4| if (!(cond)) { \ | | | | ------------------ | | | | | Branch (50:9): [True: 2, False: 0] | | | | | Branch (50:9): [True: 2, False: 0] | | | | ------------------ | | | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | | | 52| 0| } | | ------------------ ------------------ 308| 2| KMP_ASSERT_VALID_HW_TYPE(type2); ------------------ | | 637| 2| KMP_ASSERT(type >= (kmp_hw_t)0 && type < KMP_HW_LAST) | | ------------------ | | | | 50| 4| if (!(cond)) { \ | | | | ------------------ | | | | | Branch (50:9): [True: 2, False: 0] | | | | | Branch (50:9): [True: 2, False: 0] | | | | ------------------ | | | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | | | 52| 0| } | | ------------------ ------------------ 309| | // Do not allow the three main topology levels (sockets, cores, threads) to 310| | // be compacted down 311| 2| if ((type1 == KMP_HW_THREAD || type1 == KMP_HW_CORE || ------------------ | Branch (311:10): [True: 0, False: 2] | Branch (311:36): [True: 1, False: 1] ------------------ 312| 1| type1 == KMP_HW_SOCKET) && ------------------ | Branch (312:10): [True: 1, False: 0] ------------------ 313| 2| (type2 == KMP_HW_THREAD || type2 == KMP_HW_CORE || ------------------ | Branch (313:10): [True: 1, False: 1] | Branch (313:36): [True: 1, False: 0] ------------------ 314| 2| type2 == KMP_HW_SOCKET)) { ------------------ | Branch (314:10): [True: 0, False: 0] ------------------ 315| 2| top_index1 = top_index2++; 316| 2| continue; 317| 2| } 318| 0| bool radix1 = true; 319| 0| bool all_same = true; 320| 0| int id1 = hw_threads[0].ids[top_index1]; 321| 0| int id2 = hw_threads[0].ids[top_index2]; 322| 0| int pref1 = preference[type1]; 323| 0| int pref2 = preference[type2]; 324| 0| for (int hwidx = 1; hwidx < num_hw_threads; ++hwidx) { ------------------ | Branch (324:25): [True: 0, False: 0] ------------------ 325| 0| if (hw_threads[hwidx].ids[top_index1] == id1 && ------------------ | Branch (325:11): [True: 0, False: 0] ------------------ 326| 0| hw_threads[hwidx].ids[top_index2] != id2) { ------------------ | Branch (326:11): [True: 0, False: 0] ------------------ 327| 0| radix1 = false; 328| 0| break; 329| 0| } 330| 0| if (hw_threads[hwidx].ids[top_index2] != id2) ------------------ | Branch (330:11): [True: 0, False: 0] ------------------ 331| 0| all_same = false; 332| 0| id1 = hw_threads[hwidx].ids[top_index1]; 333| 0| id2 = hw_threads[hwidx].ids[top_index2]; 334| 0| } 335| 0| if (radix1) { ------------------ | Branch (335:9): [True: 0, False: 0] ------------------ 336| | // Select the layer to remove based on preference 337| 0| kmp_hw_t remove_type, keep_type; 338| 0| int remove_layer, remove_layer_ids; 339| 0| if (pref1 > pref2) { ------------------ | Branch (339:11): [True: 0, False: 0] ------------------ 340| 0| remove_type = type2; 341| 0| remove_layer = remove_layer_ids = top_index2; 342| 0| keep_type = type1; 343| 0| } else { 344| 0| remove_type = type1; 345| 0| remove_layer = remove_layer_ids = top_index1; 346| 0| keep_type = type2; 347| 0| } 348| | // If all the indexes for the second (deeper) layer are the same. 349| | // e.g., all are zero, then make sure to keep the first layer's ids 350| 0| if (all_same) ------------------ | Branch (350:11): [True: 0, False: 0] ------------------ 351| 0| remove_layer_ids = top_index2; 352| | // Remove radix one type by setting the equivalence, removing the id from 353| | // the hw threads and removing the layer from types and depth 354| 0| set_equivalent_type(remove_type, keep_type); 355| 0| for (int idx = 0; idx < num_hw_threads; ++idx) { ------------------ | Branch (355:25): [True: 0, False: 0] ------------------ 356| 0| kmp_hw_thread_t &hw_thread = hw_threads[idx]; 357| 0| for (int d = remove_layer_ids; d < depth - 1; ++d) ------------------ | Branch (357:40): [True: 0, False: 0] ------------------ 358| 0| hw_thread.ids[d] = hw_thread.ids[d + 1]; 359| 0| } 360| 0| for (int idx = remove_layer; idx < depth - 1; ++idx) ------------------ | Branch (360:36): [True: 0, False: 0] ------------------ 361| 0| types[idx] = types[idx + 1]; 362| 0| depth--; 363| 0| } else { 364| 0| top_index1 = top_index2++; 365| 0| } 366| 0| } 367| 1| KMP_ASSERT(depth > 0); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 368| 1|} _ZN14kmp_topology_t21_set_last_level_cacheEv: 370| 1|void kmp_topology_t::_set_last_level_cache() { 371| 1| if (get_equivalent_type(KMP_HW_L3) != KMP_HW_UNKNOWN) ------------------ | Branch (371:7): [True: 0, False: 1] ------------------ 372| 0| set_equivalent_type(KMP_HW_LLC, KMP_HW_L3); 373| 1| else if (get_equivalent_type(KMP_HW_L2) != KMP_HW_UNKNOWN) ------------------ | Branch (373:12): [True: 0, False: 1] ------------------ 374| 0| set_equivalent_type(KMP_HW_LLC, KMP_HW_L2); 375| 1|#if KMP_MIC_SUPPORTED 376| 1| else if (__kmp_mic_type == mic3) { ------------------ | Branch (376:12): [True: 0, False: 1] ------------------ 377| 0| if (get_equivalent_type(KMP_HW_L2) != KMP_HW_UNKNOWN) ------------------ | Branch (377:9): [True: 0, False: 0] ------------------ 378| 0| set_equivalent_type(KMP_HW_LLC, KMP_HW_L2); 379| 0| else if (get_equivalent_type(KMP_HW_TILE) != KMP_HW_UNKNOWN) ------------------ | Branch (379:14): [True: 0, False: 0] ------------------ 380| 0| set_equivalent_type(KMP_HW_LLC, KMP_HW_TILE); 381| | // L2/Tile wasn't detected so just say L1 382| 0| else 383| 0| set_equivalent_type(KMP_HW_LLC, KMP_HW_L1); 384| 0| } 385| 1|#endif 386| 1| else if (get_equivalent_type(KMP_HW_L1) != KMP_HW_UNKNOWN) ------------------ | Branch (386:12): [True: 0, False: 1] ------------------ 387| 0| set_equivalent_type(KMP_HW_LLC, KMP_HW_L1); 388| | // Fallback is to set last level cache to socket or core 389| 1| if (get_equivalent_type(KMP_HW_LLC) == KMP_HW_UNKNOWN) { ------------------ | Branch (389:7): [True: 1, False: 0] ------------------ 390| 1| if (get_equivalent_type(KMP_HW_SOCKET) != KMP_HW_UNKNOWN) ------------------ | Branch (390:9): [True: 1, False: 0] ------------------ 391| 1| set_equivalent_type(KMP_HW_LLC, KMP_HW_SOCKET); 392| 0| else if (get_equivalent_type(KMP_HW_CORE) != KMP_HW_UNKNOWN) ------------------ | Branch (392:14): [True: 0, False: 0] ------------------ 393| 0| set_equivalent_type(KMP_HW_LLC, KMP_HW_CORE); 394| 1| } 395| 1| KMP_ASSERT(get_equivalent_type(KMP_HW_LLC) != KMP_HW_UNKNOWN); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 396| 1|} _ZN14kmp_topology_t31_gather_enumeration_informationEv: 399| 1|void kmp_topology_t::_gather_enumeration_information() { 400| 1| int previous_id[KMP_HW_LAST]; 401| 1| int max[KMP_HW_LAST]; 402| | 403| 4| for (int i = 0; i < depth; ++i) { ------------------ | Branch (403:19): [True: 3, False: 1] ------------------ 404| 3| previous_id[i] = kmp_hw_thread_t::UNKNOWN_ID; 405| 3| max[i] = 0; 406| 3| count[i] = 0; 407| 3| ratio[i] = 0; 408| 3| } 409| 1| int core_level = get_level(KMP_HW_CORE); 410| 33| for (int i = 0; i < num_hw_threads; ++i) { ------------------ | Branch (410:19): [True: 32, False: 1] ------------------ 411| 32| kmp_hw_thread_t &hw_thread = hw_threads[i]; 412| 79| for (int layer = 0; layer < depth; ++layer) { ------------------ | Branch (412:25): [True: 79, False: 0] ------------------ 413| 79| int id = hw_thread.ids[layer]; 414| 79| if (id != previous_id[layer]) { ------------------ | Branch (414:11): [True: 32, False: 47] ------------------ 415| | // Add an additional increment to each count 416| 81| for (int l = layer; l < depth; ++l) ------------------ | Branch (416:29): [True: 49, False: 32] ------------------ 417| 49| count[l]++; 418| | // Keep track of topology layer ratio statistics 419| 32| max[layer]++; 420| 49| for (int l = layer + 1; l < depth; ++l) { ------------------ | Branch (420:33): [True: 17, False: 32] ------------------ 421| 17| if (max[l] > ratio[l]) ------------------ | Branch (421:15): [True: 1, False: 16] ------------------ 422| 1| ratio[l] = max[l]; 423| 17| max[l] = 1; 424| 17| } 425| | // Figure out the number of different core types 426| | // and efficiencies for hybrid CPUs 427| 32| if (__kmp_is_hybrid_cpu() && core_level >= 0 && layer <= core_level) { ------------------ | Branch (427:13): [True: 0, False: 32] | Branch (427:38): [True: 0, False: 0] | Branch (427:57): [True: 0, False: 0] ------------------ 428| 0| if (hw_thread.attrs.is_core_eff_valid() && ------------------ | Branch (428:15): [True: 0, False: 0] ------------------ 429| 0| hw_thread.attrs.core_eff >= num_core_efficiencies) { ------------------ | Branch (429:15): [True: 0, False: 0] ------------------ 430| | // Because efficiencies can range from 0 to max efficiency - 1, 431| | // the number of efficiencies is max efficiency + 1 432| 0| num_core_efficiencies = hw_thread.attrs.core_eff + 1; 433| 0| } 434| 0| if (hw_thread.attrs.is_core_type_valid()) { ------------------ | Branch (434:15): [True: 0, False: 0] ------------------ 435| 0| bool found = false; 436| 0| for (int j = 0; j < num_core_types; ++j) { ------------------ | Branch (436:29): [True: 0, False: 0] ------------------ 437| 0| if (hw_thread.attrs.get_core_type() == core_types[j]) { ------------------ | Branch (437:19): [True: 0, False: 0] ------------------ 438| 0| found = true; 439| 0| break; 440| 0| } 441| 0| } 442| 0| if (!found) { ------------------ | Branch (442:17): [True: 0, False: 0] ------------------ 443| 0| KMP_ASSERT(num_core_types < KMP_HW_MAX_NUM_CORE_TYPES); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 444| 0| core_types[num_core_types++] = hw_thread.attrs.get_core_type(); 445| 0| } 446| 0| } 447| 0| } 448| 32| break; 449| 32| } 450| 79| } 451| 128| for (int layer = 0; layer < depth; ++layer) { ------------------ | Branch (451:25): [True: 96, False: 32] ------------------ 452| 96| previous_id[layer] = hw_thread.ids[layer]; 453| 96| } 454| 32| } 455| 4| for (int layer = 0; layer < depth; ++layer) { ------------------ | Branch (455:23): [True: 3, False: 1] ------------------ 456| 3| if (max[layer] > ratio[layer]) ------------------ | Branch (456:9): [True: 2, False: 1] ------------------ 457| 2| ratio[layer] = max[layer]; 458| 3| } 459| 1|} _ZN14kmp_topology_t20_discover_uniformityEv: 498| 1|void kmp_topology_t::_discover_uniformity() { 499| 1| int num = 1; 500| 4| for (int level = 0; level < depth; ++level) ------------------ | Branch (500:23): [True: 3, False: 1] ------------------ 501| 3| num *= ratio[level]; 502| 1| flags.uniform = (num == count[depth - 1]); 503| 1|} _ZN14kmp_topology_t12_set_sub_idsEv: 506| 1|void kmp_topology_t::_set_sub_ids() { 507| 1| int previous_id[KMP_HW_LAST]; 508| 1| int sub_id[KMP_HW_LAST]; 509| | 510| 4| for (int i = 0; i < depth; ++i) { ------------------ | Branch (510:19): [True: 3, False: 1] ------------------ 511| 3| previous_id[i] = -1; 512| 3| sub_id[i] = -1; 513| 3| } 514| 33| for (int i = 0; i < num_hw_threads; ++i) { ------------------ | Branch (514:19): [True: 32, False: 1] ------------------ 515| 32| kmp_hw_thread_t &hw_thread = hw_threads[i]; 516| | // Setup the sub_id 517| 79| for (int j = 0; j < depth; ++j) { ------------------ | Branch (517:21): [True: 79, False: 0] ------------------ 518| 79| if (hw_thread.ids[j] != previous_id[j]) { ------------------ | Branch (518:11): [True: 32, False: 47] ------------------ 519| 32| sub_id[j]++; 520| 49| for (int k = j + 1; k < depth; ++k) { ------------------ | Branch (520:29): [True: 17, False: 32] ------------------ 521| 17| sub_id[k] = 0; 522| 17| } 523| 32| break; 524| 32| } 525| 79| } 526| | // Set previous_id 527| 128| for (int j = 0; j < depth; ++j) { ------------------ | Branch (527:21): [True: 96, False: 32] ------------------ 528| 96| previous_id[j] = hw_thread.ids[j]; 529| 96| } 530| | // Set the sub_ids field 531| 128| for (int j = 0; j < depth; ++j) { ------------------ | Branch (531:21): [True: 96, False: 32] ------------------ 532| 96| hw_thread.sub_ids[j] = sub_id[j]; 533| 96| } 534| 32| } 535| 1|} _ZN14kmp_topology_t12_set_globalsEv: 537| 1|void kmp_topology_t::_set_globals() { 538| | // Set nCoresPerPkg, nPackages, __kmp_nThreadsPerCore, __kmp_ncores 539| 1| int core_level, thread_level, package_level; 540| 1| package_level = get_level(KMP_HW_SOCKET); 541| |#if KMP_GROUP_AFFINITY 542| | if (package_level == -1) 543| | package_level = get_level(KMP_HW_PROC_GROUP); 544| |#endif 545| 1| core_level = get_level(KMP_HW_CORE); 546| 1| thread_level = get_level(KMP_HW_THREAD); 547| | 548| 1| KMP_ASSERT(core_level != -1); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 549| 1| KMP_ASSERT(thread_level != -1); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 550| | 551| 1| __kmp_nThreadsPerCore = calculate_ratio(thread_level, core_level); 552| 1| if (package_level != -1) { ------------------ | Branch (552:7): [True: 1, False: 0] ------------------ 553| 1| nCoresPerPkg = calculate_ratio(core_level, package_level); 554| 1| nPackages = get_count(package_level); 555| 1| } else { 556| | // assume one socket 557| 0| nCoresPerPkg = get_count(core_level); 558| 0| nPackages = 1; 559| 0| } 560| 1|#ifndef KMP_DFLT_NTH_CORES 561| 1| __kmp_ncores = get_count(core_level); 562| 1|#endif 563| 1|} _ZN14kmp_topology_t8allocateEiiPK8kmp_hw_t: 566| 1| const kmp_hw_t *types) { 567| 1| kmp_topology_t *retval; 568| | // Allocate all data in one large allocation 569| 1| size_t size = sizeof(kmp_topology_t) + sizeof(kmp_hw_thread_t) * nproc + 570| 1| sizeof(int) * (size_t)KMP_HW_LAST * 3; 571| 1| char *bytes = (char *)__kmp_allocate(size); ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 572| 1| retval = (kmp_topology_t *)bytes; 573| 1| if (nproc > 0) { ------------------ | Branch (573:7): [True: 1, False: 0] ------------------ 574| 1| retval->hw_threads = (kmp_hw_thread_t *)(bytes + sizeof(kmp_topology_t)); 575| 1| } else { 576| 0| retval->hw_threads = nullptr; 577| 0| } 578| 1| retval->num_hw_threads = nproc; 579| 1| retval->depth = ndepth; 580| 1| int *arr = 581| 1| (int *)(bytes + sizeof(kmp_topology_t) + sizeof(kmp_hw_thread_t) * nproc); 582| 1| retval->types = (kmp_hw_t *)arr; 583| 1| retval->ratio = arr + (size_t)KMP_HW_LAST; 584| 1| retval->count = arr + 2 * (size_t)KMP_HW_LAST; 585| 1| retval->num_core_efficiencies = 0; 586| 1| retval->num_core_types = 0; 587| 1| retval->compact = 0; 588| 4| for (int i = 0; i < KMP_HW_MAX_NUM_CORE_TYPES; ++i) ------------------ | Branch (588:19): [True: 3, False: 1] ------------------ 589| 3| retval->core_types[i] = KMP_HW_CORE_TYPE_UNKNOWN; 590| 12| KMP_FOREACH_HW_TYPE(type) { retval->equivalent[type] = KMP_HW_UNKNOWN; } ------------------ | | 640| 13| for (kmp_hw_t type = (kmp_hw_t)0; type < KMP_HW_LAST; \ | | ------------------ | | | Branch (640:37): [True: 12, False: 1] | | ------------------ | | 641| 12| type = (kmp_hw_t)((int)type + 1)) ------------------ 591| 4| for (int i = 0; i < ndepth; ++i) { ------------------ | Branch (591:19): [True: 3, False: 1] ------------------ 592| 3| retval->types[i] = types[i]; 593| 3| retval->equivalent[types[i]] = types[i]; 594| 3| } 595| 1| return retval; 596| 1|} _ZN14kmp_topology_t10deallocateEPS_: 598| 1|void kmp_topology_t::deallocate(kmp_topology_t *topology) { 599| 1| if (topology) ------------------ | Branch (599:7): [True: 1, False: 0] ------------------ 600| 1| __kmp_free(topology); ------------------ | | 3619| 1|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 601| 1|} _ZNK14kmp_topology_t9check_idsEv: 603| 1|bool kmp_topology_t::check_ids() const { 604| | // Assume ids have been sorted 605| 1| if (num_hw_threads == 0) ------------------ | Branch (605:7): [True: 0, False: 1] ------------------ 606| 0| return true; 607| 32| for (int i = 1; i < num_hw_threads; ++i) { ------------------ | Branch (607:19): [True: 31, False: 1] ------------------ 608| 31| kmp_hw_thread_t ¤t_thread = hw_threads[i]; 609| 31| kmp_hw_thread_t &previous_thread = hw_threads[i - 1]; 610| 31| bool unique = false; 611| 78| for (int j = 0; j < depth; ++j) { ------------------ | Branch (611:21): [True: 78, False: 0] ------------------ 612| 78| if (previous_thread.ids[j] != current_thread.ids[j]) { ------------------ | Branch (612:11): [True: 31, False: 47] ------------------ 613| 31| unique = true; 614| 31| break; 615| 31| } 616| 78| } 617| 31| if (unique) ------------------ | Branch (617:9): [True: 31, False: 0] ------------------ 618| 31| continue; 619| 0| return false; 620| 31| } 621| 1| return true; 622| 1|} _ZN14kmp_topology_t12canonicalizeEv: 847| 1|void kmp_topology_t::canonicalize() { 848| |#if KMP_GROUP_AFFINITY 849| | _insert_windows_proc_groups(); 850| |#endif 851| 1| _remove_radix1_layers(); 852| 1| _gather_enumeration_information(); 853| 1| _discover_uniformity(); 854| 1| _set_sub_ids(); 855| 1| _set_globals(); 856| 1| _set_last_level_cache(); 857| | 858| 1|#if KMP_MIC_SUPPORTED 859| | // Manually Add L2 = Tile equivalence 860| 1| if (__kmp_mic_type == mic3) { ------------------ | Branch (860:7): [True: 0, False: 1] ------------------ 861| 0| if (get_level(KMP_HW_L2) != -1) ------------------ | Branch (861:9): [True: 0, False: 0] ------------------ 862| 0| set_equivalent_type(KMP_HW_TILE, KMP_HW_L2); 863| 0| else if (get_level(KMP_HW_TILE) != -1) ------------------ | Branch (863:14): [True: 0, False: 0] ------------------ 864| 0| set_equivalent_type(KMP_HW_L2, KMP_HW_TILE); 865| 0| } 866| 1|#endif 867| | 868| | // Perform post canonicalization checking 869| 1| KMP_ASSERT(depth > 0); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 870| 4| for (int level = 0; level < depth; ++level) { ------------------ | Branch (870:23): [True: 3, False: 1] ------------------ 871| | // All counts, ratios, and types must be valid 872| 3| KMP_ASSERT(count[level] > 0 && ratio[level] > 0); ------------------ | | 50| 6| if (!(cond)) { \ | | ------------------ | | | Branch (50:9): [True: 3, False: 0] | | | Branch (50:9): [True: 3, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 873| 3| KMP_ASSERT_VALID_HW_TYPE(types[level]); ------------------ | | 637| 3| KMP_ASSERT(type >= (kmp_hw_t)0 && type < KMP_HW_LAST) | | ------------------ | | | | 50| 6| if (!(cond)) { \ | | | | ------------------ | | | | | Branch (50:9): [True: 3, False: 0] | | | | | Branch (50:9): [True: 3, False: 0] | | | | ------------------ | | | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | | | 52| 0| } | | ------------------ ------------------ 874| | // Detected types must point to themselves 875| 3| KMP_ASSERT(equivalent[types[level]] == types[level]); ------------------ | | 50| 3| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 3] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 876| 3| } 877| 1|} _ZN14kmp_topology_t16filter_hw_subsetEv: 958| 1|bool kmp_topology_t::filter_hw_subset() { 959| | // If KMP_HW_SUBSET wasn't requested, then do nothing. 960| 1| if (!__kmp_hw_subset) ------------------ | Branch (960:7): [True: 1, False: 0] ------------------ 961| 1| return false; 962| | 963| | // First, sort the KMP_HW_SUBSET items by the machine topology 964| 0| __kmp_hw_subset->sort(); 965| | 966| | // Check to see if KMP_HW_SUBSET is a valid subset of the detected topology 967| 0| bool using_core_types = false; 968| 0| bool using_core_effs = false; 969| 0| int hw_subset_depth = __kmp_hw_subset->get_depth(); 970| 0| kmp_hw_t specified[KMP_HW_LAST]; 971| 0| int *topology_levels = (int *)KMP_ALLOCA(sizeof(int) * hw_subset_depth); ------------------ | | 59| 0|#define KMP_ALLOCA alloca ------------------ 972| 0| KMP_ASSERT(hw_subset_depth > 0); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 973| 0| KMP_FOREACH_HW_TYPE(i) { specified[i] = KMP_HW_UNKNOWN; } ------------------ | | 640| 0| for (kmp_hw_t type = (kmp_hw_t)0; type < KMP_HW_LAST; \ | | ------------------ | | | Branch (640:37): [True: 0, False: 0] | | ------------------ | | 641| 0| type = (kmp_hw_t)((int)type + 1)) ------------------ 974| 0| int core_level = get_level(KMP_HW_CORE); 975| 0| for (int i = 0; i < hw_subset_depth; ++i) { ------------------ | Branch (975:19): [True: 0, False: 0] ------------------ 976| 0| int max_count; 977| 0| const kmp_hw_subset_t::item_t &item = __kmp_hw_subset->at(i); 978| 0| int num = item.num[0]; 979| 0| int offset = item.offset[0]; 980| 0| kmp_hw_t type = item.type; 981| 0| kmp_hw_t equivalent_type = equivalent[type]; 982| 0| int level = get_level(type); 983| 0| topology_levels[i] = level; 984| | 985| | // Check to see if current layer is in detected machine topology 986| 0| if (equivalent_type != KMP_HW_UNKNOWN) { ------------------ | Branch (986:9): [True: 0, False: 0] ------------------ 987| 0| __kmp_hw_subset->at(i).type = equivalent_type; 988| 0| } else { 989| 0| KMP_AFF_WARNING(__kmp_affinity, AffHWSubsetNotExistGeneric, ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 990| 0| __kmp_hw_get_catalog_string(type)); 991| 0| return false; 992| 0| } 993| | 994| | // Check to see if current layer has already been 995| | // specified either directly or through an equivalent type 996| 0| if (specified[equivalent_type] != KMP_HW_UNKNOWN) { ------------------ | Branch (996:9): [True: 0, False: 0] ------------------ 997| 0| KMP_AFF_WARNING(__kmp_affinity, AffHWSubsetEqvLayers, ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 998| 0| __kmp_hw_get_catalog_string(type), 999| 0| __kmp_hw_get_catalog_string(specified[equivalent_type])); 1000| 0| return false; 1001| 0| } 1002| 0| specified[equivalent_type] = type; 1003| | 1004| | // Check to see if each layer's num & offset parameters are valid 1005| 0| max_count = get_ratio(level); 1006| 0| if (max_count < 0 || ------------------ | Branch (1006:9): [True: 0, False: 0] ------------------ 1007| 0| (num != kmp_hw_subset_t::USE_ALL && num + offset > max_count)) { ------------------ | Branch (1007:10): [True: 0, False: 0] | Branch (1007:45): [True: 0, False: 0] ------------------ 1008| 0| bool plural = (num > 1); 1009| 0| KMP_AFF_WARNING(__kmp_affinity, AffHWSubsetManyGeneric, ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 1010| 0| __kmp_hw_get_catalog_string(type, plural)); 1011| 0| return false; 1012| 0| } 1013| | 1014| | // Check to see if core attributes are consistent 1015| 0| if (core_level == level) { ------------------ | Branch (1015:9): [True: 0, False: 0] ------------------ 1016| | // Determine which core attributes are specified 1017| 0| for (int j = 0; j < item.num_attrs; ++j) { ------------------ | Branch (1017:23): [True: 0, False: 0] ------------------ 1018| 0| if (item.attr[j].is_core_type_valid()) ------------------ | Branch (1018:13): [True: 0, False: 0] ------------------ 1019| 0| using_core_types = true; 1020| 0| if (item.attr[j].is_core_eff_valid()) ------------------ | Branch (1020:13): [True: 0, False: 0] ------------------ 1021| 0| using_core_effs = true; 1022| 0| } 1023| | 1024| | // Check if using a single core attribute on non-hybrid arch. 1025| | // Do not ignore all of KMP_HW_SUBSET, just ignore the attribute. 1026| | // 1027| | // Check if using multiple core attributes on non-hyrbid arch. 1028| | // Ignore all of KMP_HW_SUBSET if this is the case. 1029| 0| if ((using_core_effs || using_core_types) && !__kmp_is_hybrid_cpu()) { ------------------ | Branch (1029:12): [True: 0, False: 0] | Branch (1029:31): [True: 0, False: 0] | Branch (1029:52): [True: 0, False: 0] ------------------ 1030| 0| if (item.num_attrs == 1) { ------------------ | Branch (1030:13): [True: 0, False: 0] ------------------ 1031| 0| if (using_core_effs) { ------------------ | Branch (1031:15): [True: 0, False: 0] ------------------ 1032| 0| KMP_AFF_WARNING(__kmp_affinity, AffHWSubsetIgnoringAttr, ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 1033| 0| "efficiency"); 1034| 0| } else { 1035| 0| KMP_AFF_WARNING(__kmp_affinity, AffHWSubsetIgnoringAttr, ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 1036| 0| "core_type"); 1037| 0| } 1038| 0| using_core_effs = false; 1039| 0| using_core_types = false; 1040| 0| } else { 1041| 0| KMP_AFF_WARNING(__kmp_affinity, AffHWSubsetAttrsNonHybrid); ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 1042| 0| return false; 1043| 0| } 1044| 0| } 1045| | 1046| | // Check if using both core types and core efficiencies together 1047| 0| if (using_core_types && using_core_effs) { ------------------ | Branch (1047:11): [True: 0, False: 0] | Branch (1047:31): [True: 0, False: 0] ------------------ 1048| 0| KMP_AFF_WARNING(__kmp_affinity, AffHWSubsetIncompat, "core_type", ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 1049| 0| "efficiency"); 1050| 0| return false; 1051| 0| } 1052| | 1053| | // Check that core efficiency values are valid 1054| 0| if (using_core_effs) { ------------------ | Branch (1054:11): [True: 0, False: 0] ------------------ 1055| 0| for (int j = 0; j < item.num_attrs; ++j) { ------------------ | Branch (1055:25): [True: 0, False: 0] ------------------ 1056| 0| if (item.attr[j].is_core_eff_valid()) { ------------------ | Branch (1056:15): [True: 0, False: 0] ------------------ 1057| 0| int core_eff = item.attr[j].get_core_eff(); 1058| 0| if (core_eff < 0 || core_eff >= num_core_efficiencies) { ------------------ | Branch (1058:17): [True: 0, False: 0] | Branch (1058:33): [True: 0, False: 0] ------------------ 1059| 0| kmp_str_buf_t buf; 1060| 0| __kmp_str_buf_init(&buf); ------------------ | | 41| 0| { \ | | 42| 0| (b)->str = (b)->bulk; \ | | 43| 0| (b)->size = sizeof((b)->bulk); \ | | 44| 0| (b)->used = 0; \ | | 45| 0| (b)->bulk[0] = 0; \ | | 46| 0| } ------------------ 1061| 0| __kmp_str_buf_print(&buf, "%d", item.attr[j].get_core_eff()); 1062| 0| __kmp_msg(kmp_ms_warning, 1063| 0| KMP_MSG(AffHWSubsetAttrInvalid, "efficiency", buf.str), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ 1064| 0| KMP_HNT(ValidValuesRange, 0, num_core_efficiencies - 1), ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 1065| 0| __kmp_msg_null); 1066| 0| __kmp_str_buf_free(&buf); 1067| 0| return false; 1068| 0| } 1069| 0| } 1070| 0| } 1071| 0| } 1072| | 1073| | // Check that the number of requested cores with attributes is valid 1074| 0| if (using_core_types || using_core_effs) { ------------------ | Branch (1074:11): [True: 0, False: 0] | Branch (1074:31): [True: 0, False: 0] ------------------ 1075| 0| for (int j = 0; j < item.num_attrs; ++j) { ------------------ | Branch (1075:25): [True: 0, False: 0] ------------------ 1076| 0| int num = item.num[j]; 1077| 0| int offset = item.offset[j]; 1078| 0| int level_above = core_level - 1; 1079| 0| if (level_above >= 0) { ------------------ | Branch (1079:15): [True: 0, False: 0] ------------------ 1080| 0| max_count = get_ncores_with_attr_per(item.attr[j], level_above); 1081| 0| if (max_count <= 0 || ------------------ | Branch (1081:17): [True: 0, False: 0] ------------------ 1082| 0| (num != kmp_hw_subset_t::USE_ALL && num + offset > max_count)) { ------------------ | Branch (1082:18): [True: 0, False: 0] | Branch (1082:53): [True: 0, False: 0] ------------------ 1083| 0| kmp_str_buf_t buf; 1084| 0| __kmp_hw_get_catalog_core_string(item.attr[j], &buf, num > 0); 1085| 0| KMP_AFF_WARNING(__kmp_affinity, AffHWSubsetManyGeneric, buf.str); ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 1086| 0| __kmp_str_buf_free(&buf); 1087| 0| return false; 1088| 0| } 1089| 0| } 1090| 0| } 1091| 0| } 1092| | 1093| 0| if ((using_core_types || using_core_effs) && item.num_attrs > 1) { ------------------ | Branch (1093:12): [True: 0, False: 0] | Branch (1093:32): [True: 0, False: 0] | Branch (1093:52): [True: 0, False: 0] ------------------ 1094| 0| for (int j = 0; j < item.num_attrs; ++j) { ------------------ | Branch (1094:25): [True: 0, False: 0] ------------------ 1095| | // Ambiguous use of specific core attribute + generic core 1096| | // e.g., 4c & 3c:intel_core or 4c & 3c:eff1 1097| 0| if (!item.attr[j]) { ------------------ | Branch (1097:15): [True: 0, False: 0] ------------------ 1098| 0| kmp_hw_attr_t other_attr; 1099| 0| for (int k = 0; k < item.num_attrs; ++k) { ------------------ | Branch (1099:29): [True: 0, False: 0] ------------------ 1100| 0| if (item.attr[k] != item.attr[j]) { ------------------ | Branch (1100:19): [True: 0, False: 0] ------------------ 1101| 0| other_attr = item.attr[k]; 1102| 0| break; 1103| 0| } 1104| 0| } 1105| 0| kmp_str_buf_t buf; 1106| 0| __kmp_hw_get_catalog_core_string(other_attr, &buf, item.num[j] > 0); 1107| 0| KMP_AFF_WARNING(__kmp_affinity, AffHWSubsetIncompat, ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 1108| 0| __kmp_hw_get_catalog_string(KMP_HW_CORE), buf.str); 1109| 0| __kmp_str_buf_free(&buf); 1110| 0| return false; 1111| 0| } 1112| | // Allow specifying a specific core type or core eff exactly once 1113| 0| for (int k = 0; k < j; ++k) { ------------------ | Branch (1113:27): [True: 0, False: 0] ------------------ 1114| 0| if (!item.attr[j] || !item.attr[k]) ------------------ | Branch (1114:17): [True: 0, False: 0] | Branch (1114:34): [True: 0, False: 0] ------------------ 1115| 0| continue; 1116| 0| if (item.attr[k] == item.attr[j]) { ------------------ | Branch (1116:17): [True: 0, False: 0] ------------------ 1117| 0| kmp_str_buf_t buf; 1118| 0| __kmp_hw_get_catalog_core_string(item.attr[j], &buf, 1119| 0| item.num[j] > 0); 1120| 0| KMP_AFF_WARNING(__kmp_affinity, AffHWSubsetAttrRepeat, buf.str); ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 1121| 0| __kmp_str_buf_free(&buf); 1122| 0| return false; 1123| 0| } 1124| 0| } 1125| 0| } 1126| 0| } 1127| 0| } 1128| 0| } 1129| | 1130| 0| struct core_type_indexer { 1131| 0| int operator()(const kmp_hw_thread_t &t) const { 1132| 0| switch (t.attrs.get_core_type()) { 1133| 0|#if KMP_ARCH_X86 || KMP_ARCH_X86_64 1134| 0| case KMP_HW_CORE_TYPE_ATOM: 1135| 0| return 1; 1136| 0| case KMP_HW_CORE_TYPE_CORE: 1137| 0| return 2; 1138| 0|#endif 1139| 0| case KMP_HW_CORE_TYPE_UNKNOWN: 1140| 0| return 0; 1141| 0| } 1142| 0| KMP_ASSERT(0); 1143| 0| return 0; 1144| 0| } 1145| 0| }; 1146| 0| struct core_eff_indexer { 1147| 0| int operator()(const kmp_hw_thread_t &t) const { 1148| 0| return t.attrs.get_core_eff(); 1149| 0| } 1150| 0| }; 1151| | 1152| 0| kmp_sub_ids_t core_type_sub_ids( 1153| 0| core_level); 1154| 0| kmp_sub_ids_t core_eff_sub_ids( 1155| 0| core_level); 1156| | 1157| | // Determine which hardware threads should be filtered. 1158| 0| int num_filtered = 0; 1159| 0| bool *filtered = (bool *)__kmp_allocate(sizeof(bool) * num_hw_threads); ------------------ | | 3617| 0|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 1160| 0| for (int i = 0; i < num_hw_threads; ++i) { ------------------ | Branch (1160:19): [True: 0, False: 0] ------------------ 1161| 0| kmp_hw_thread_t &hw_thread = hw_threads[i]; 1162| | // Update type_sub_id 1163| 0| if (using_core_types) ------------------ | Branch (1163:9): [True: 0, False: 0] ------------------ 1164| 0| core_type_sub_ids.update(hw_thread); 1165| 0| if (using_core_effs) ------------------ | Branch (1165:9): [True: 0, False: 0] ------------------ 1166| 0| core_eff_sub_ids.update(hw_thread); 1167| | 1168| | // Check to see if this hardware thread should be filtered 1169| 0| bool should_be_filtered = false; 1170| 0| for (int hw_subset_index = 0; hw_subset_index < hw_subset_depth; ------------------ | Branch (1170:35): [True: 0, False: 0] ------------------ 1171| 0| ++hw_subset_index) { 1172| 0| const auto &hw_subset_item = __kmp_hw_subset->at(hw_subset_index); 1173| 0| int level = topology_levels[hw_subset_index]; 1174| 0| if (level == -1) ------------------ | Branch (1174:11): [True: 0, False: 0] ------------------ 1175| 0| continue; 1176| 0| if ((using_core_effs || using_core_types) && level == core_level) { ------------------ | Branch (1176:12): [True: 0, False: 0] | Branch (1176:31): [True: 0, False: 0] | Branch (1176:52): [True: 0, False: 0] ------------------ 1177| | // Look for the core attribute in KMP_HW_SUBSET which corresponds 1178| | // to this hardware thread's core attribute. Use this num,offset plus 1179| | // the running sub_id for the particular core attribute of this hardware 1180| | // thread to determine if the hardware thread should be filtered or not. 1181| 0| int attr_idx; 1182| 0| kmp_hw_core_type_t core_type = hw_thread.attrs.get_core_type(); 1183| 0| int core_eff = hw_thread.attrs.get_core_eff(); 1184| 0| for (attr_idx = 0; attr_idx < hw_subset_item.num_attrs; ++attr_idx) { ------------------ | Branch (1184:28): [True: 0, False: 0] ------------------ 1185| 0| if (using_core_types && ------------------ | Branch (1185:15): [True: 0, False: 0] ------------------ 1186| 0| hw_subset_item.attr[attr_idx].get_core_type() == core_type) ------------------ | Branch (1186:15): [True: 0, False: 0] ------------------ 1187| 0| break; 1188| 0| if (using_core_effs && ------------------ | Branch (1188:15): [True: 0, False: 0] ------------------ 1189| 0| hw_subset_item.attr[attr_idx].get_core_eff() == core_eff) ------------------ | Branch (1189:15): [True: 0, False: 0] ------------------ 1190| 0| break; 1191| 0| } 1192| | // This core attribute isn't in the KMP_HW_SUBSET so always filter it. 1193| 0| if (attr_idx == hw_subset_item.num_attrs) { ------------------ | Branch (1193:13): [True: 0, False: 0] ------------------ 1194| 0| should_be_filtered = true; 1195| 0| break; 1196| 0| } 1197| 0| int sub_id; 1198| 0| int num = hw_subset_item.num[attr_idx]; 1199| 0| int offset = hw_subset_item.offset[attr_idx]; 1200| 0| if (using_core_types) ------------------ | Branch (1200:13): [True: 0, False: 0] ------------------ 1201| 0| sub_id = core_type_sub_ids.get_sub_id(hw_thread); 1202| 0| else 1203| 0| sub_id = core_eff_sub_ids.get_sub_id(hw_thread); 1204| 0| if (sub_id < offset || ------------------ | Branch (1204:13): [True: 0, False: 0] ------------------ 1205| 0| (num != kmp_hw_subset_t::USE_ALL && sub_id >= offset + num)) { ------------------ | Branch (1205:14): [True: 0, False: 0] | Branch (1205:49): [True: 0, False: 0] ------------------ 1206| 0| should_be_filtered = true; 1207| 0| break; 1208| 0| } 1209| 0| } else { 1210| 0| int num = hw_subset_item.num[0]; 1211| 0| int offset = hw_subset_item.offset[0]; 1212| 0| if (hw_thread.sub_ids[level] < offset || ------------------ | Branch (1212:13): [True: 0, False: 0] ------------------ 1213| 0| (num != kmp_hw_subset_t::USE_ALL && ------------------ | Branch (1213:14): [True: 0, False: 0] ------------------ 1214| 0| hw_thread.sub_ids[level] >= offset + num)) { ------------------ | Branch (1214:14): [True: 0, False: 0] ------------------ 1215| 0| should_be_filtered = true; 1216| 0| break; 1217| 0| } 1218| 0| } 1219| 0| } 1220| | // Collect filtering information 1221| 0| filtered[i] = should_be_filtered; 1222| 0| if (should_be_filtered) ------------------ | Branch (1222:9): [True: 0, False: 0] ------------------ 1223| 0| num_filtered++; 1224| 0| } 1225| | 1226| | // One last check that we shouldn't allow filtering entire machine 1227| 0| if (num_filtered == num_hw_threads) { ------------------ | Branch (1227:7): [True: 0, False: 0] ------------------ 1228| 0| KMP_AFF_WARNING(__kmp_affinity, AffHWSubsetAllFiltered); ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 1229| 0| __kmp_free(filtered); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 1230| 0| return false; 1231| 0| } 1232| | 1233| | // Apply the filter 1234| 0| int new_index = 0; 1235| 0| for (int i = 0; i < num_hw_threads; ++i) { ------------------ | Branch (1235:19): [True: 0, False: 0] ------------------ 1236| 0| if (!filtered[i]) { ------------------ | Branch (1236:9): [True: 0, False: 0] ------------------ 1237| 0| if (i != new_index) ------------------ | Branch (1237:11): [True: 0, False: 0] ------------------ 1238| 0| hw_threads[new_index] = hw_threads[i]; 1239| 0| new_index++; 1240| 0| } else { 1241| 0|#if KMP_AFFINITY_SUPPORTED 1242| 0| KMP_CPU_CLR(hw_threads[i].os_id, __kmp_affin_fullMask); ------------------ | | 691| 0|#define KMP_CPU_CLR(i, mask) (mask)->clear(i) ------------------ 1243| 0|#endif 1244| 0| __kmp_avail_proc--; 1245| 0| } 1246| 0| } 1247| | 1248| 0| KMP_DEBUG_ASSERT(new_index <= num_hw_threads); 1249| 0| num_hw_threads = new_index; 1250| | 1251| | // Post hardware subset canonicalization 1252| 0| _gather_enumeration_information(); 1253| 0| _discover_uniformity(); 1254| 0| _set_globals(); 1255| 0| _set_last_level_cache(); 1256| 0| __kmp_free(filtered); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 1257| 0| return true; 1258| 0|} _ZN11KMPAffinity4MasknwEm: 1279| 68|void *KMPAffinity::Mask::operator new(size_t n) { return __kmp_allocate(n); } ------------------ | | 3617| 68|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ _ZN11KMPAffinity4MasknaEm: 1280| 1|void *KMPAffinity::Mask::operator new[](size_t n) { return __kmp_allocate(n); } ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ _ZN11KMPAffinity4MaskdlEPv: 1281| 67|void KMPAffinity::Mask::operator delete(void *p) { __kmp_free(p); } ------------------ | | 3619| 67|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ _ZN11KMPAffinity4MaskdaEPv: 1282| 1|void KMPAffinity::Mask::operator delete[](void *p) { __kmp_free(p); } ------------------ | | 3619| 1|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ _ZN11KMPAffinitynwEm: 1283| 1|void *KMPAffinity::operator new(size_t n) { return __kmp_allocate(n); } ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ _ZN11KMPAffinitydlEPv: 1284| 1|void KMPAffinity::operator delete(void *p) { __kmp_free(p); } ------------------ | | 3619| 1|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ _ZN11KMPAffinity8pick_apiEv: 1286| 1|void KMPAffinity::pick_api() { 1287| 1| KMPAffinity *affinity_dispatch; 1288| 1| if (picked_api) ------------------ | Branch (1288:7): [True: 0, False: 1] ------------------ 1289| 0| return; 1290| |#if KMP_USE_HWLOC 1291| | // Only use Hwloc if affinity isn't explicitly disabled and 1292| | // user requests Hwloc topology method 1293| | if (__kmp_affinity_top_method == affinity_top_method_hwloc && 1294| | __kmp_affinity.type != affinity_disabled) { 1295| | affinity_dispatch = new KMPHwlocAffinity(); 1296| | } else 1297| |#endif 1298| 1| { 1299| 1| affinity_dispatch = new KMPNativeAffinity(); 1300| 1| } 1301| 1| __kmp_affinity_dispatch = affinity_dispatch; 1302| 1| picked_api = true; 1303| 1|} _ZN11KMPAffinity11destroy_apiEv: 1305| 2|void KMPAffinity::destroy_api() { 1306| 2| if (__kmp_affinity_dispatch != NULL) { ------------------ | Branch (1306:7): [True: 1, False: 1] ------------------ 1307| 1| delete __kmp_affinity_dispatch; 1308| | __kmp_affinity_dispatch = NULL; 1309| 1| picked_api = false; 1310| 1| } 1311| 2|} __kmp_affinity_initialize: 4608| 1|void __kmp_affinity_initialize(kmp_affinity_t &affinity) { 4609| | // Much of the code above was written assuming that if a machine was not 4610| | // affinity capable, then affinity type == affinity_none. 4611| | // We now explicitly represent this as affinity type == affinity_disabled. 4612| | // There are too many checks for affinity type == affinity_none in this code. 4613| | // Instead of trying to change them all, check if 4614| | // affinity type == affinity_disabled, and if so, slam it with affinity_none, 4615| | // call the real initialization routine, then restore affinity type to 4616| | // affinity_disabled. 4617| 1| int disabled = (affinity.type == affinity_disabled); 4618| 1| if (!KMP_AFFINITY_CAPABLE()) ------------------ | | 684| 1|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) ------------------ | Branch (4618:7): [True: 0, False: 1] ------------------ 4619| 0| KMP_ASSERT(disabled); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4620| 1| if (disabled) ------------------ | Branch (4620:7): [True: 0, False: 1] ------------------ 4621| 0| affinity.type = affinity_none; 4622| 1| __kmp_aux_affinity_initialize(affinity); 4623| 1| if (disabled) ------------------ | Branch (4623:7): [True: 0, False: 1] ------------------ 4624| 0| affinity.type = affinity_disabled; 4625| 1|} __kmp_affinity_uninitialize: 4627| 2|void __kmp_affinity_uninitialize(void) { 4628| 4| for (kmp_affinity_t *affinity : __kmp_affinities) { ------------------ | Branch (4628:33): [True: 4, False: 2] ------------------ 4629| 4| if (affinity->masks != NULL) ------------------ | Branch (4629:9): [True: 1, False: 3] ------------------ 4630| 1| KMP_CPU_FREE_ARRAY(affinity->masks, affinity->num_masks); ------------------ | | 707| 1| __kmp_affinity_dispatch->deallocate_mask_array(arr) ------------------ 4631| 4| if (affinity->os_id_masks != NULL) ------------------ | Branch (4631:9): [True: 0, False: 4] ------------------ 4632| 0| KMP_CPU_FREE_ARRAY(affinity->os_id_masks, affinity->num_os_id_masks); ------------------ | | 707| 0| __kmp_affinity_dispatch->deallocate_mask_array(arr) ------------------ 4633| 4| if (affinity->proclist != NULL) ------------------ | Branch (4633:9): [True: 0, False: 4] ------------------ 4634| 0| __kmp_free(affinity->proclist); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 4635| 4| if (affinity->ids != NULL) ------------------ | Branch (4635:9): [True: 1, False: 3] ------------------ 4636| 1| __kmp_free(affinity->ids); ------------------ | | 3619| 1|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 4637| 4| if (affinity->attrs != NULL) ------------------ | Branch (4637:9): [True: 1, False: 3] ------------------ 4638| 1| __kmp_free(affinity->attrs); ------------------ | | 3619| 1|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 4639| 4| *affinity = KMP_AFFINITY_INIT(affinity->env_var); ------------------ | | 911| 4| { \ | | 912| 4| nullptr, affinity_default, KMP_HW_UNKNOWN, -1, 0, 0, \ | | 913| 4| {TRUE, FALSE, TRUE, affinity_respect_mask_default, FALSE, FALSE}, 0, \ | | ------------------ | | | | 1280| 4|#define TRUE (!FALSE) | | | | ------------------ | | | | | | 1279| 4|#define FALSE 0 | | | | ------------------ | | ------------------ | | {TRUE, FALSE, TRUE, affinity_respect_mask_default, FALSE, FALSE}, 0, \ | | ------------------ | | | | 1279| 4|#define FALSE 0 | | ------------------ | | {TRUE, FALSE, TRUE, affinity_respect_mask_default, FALSE, FALSE}, 0, \ | | ------------------ | | | | 1280| 4|#define TRUE (!FALSE) | | | | ------------------ | | | | | | 1279| 4|#define FALSE 0 | | | | ------------------ | | ------------------ | | {TRUE, FALSE, TRUE, affinity_respect_mask_default, FALSE, FALSE}, 0, \ | | ------------------ | | | | 860| 4|#define affinity_respect_mask_default (2) | | ------------------ | | {TRUE, FALSE, TRUE, affinity_respect_mask_default, FALSE, FALSE}, 0, \ | | ------------------ | | | | 1279| 4|#define FALSE 0 | | ------------------ | | {TRUE, FALSE, TRUE, affinity_respect_mask_default, FALSE, FALSE}, 0, \ | | ------------------ | | | | 1279| 4|#define FALSE 0 | | ------------------ | | 914| 4| nullptr, nullptr, nullptr, 0, nullptr, env \ | | 915| 4| } ------------------ 4640| 4| } 4641| 2| if (__kmp_affin_origMask != NULL) { ------------------ | Branch (4641:7): [True: 1, False: 1] ------------------ 4642| 1| if (KMP_AFFINITY_CAPABLE()) { ------------------ | | 684| 1|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) | | ------------------ | | | Branch (684:32): [True: 1, False: 0] | | ------------------ ------------------ 4643| 1| __kmp_set_system_affinity(__kmp_affin_origMask, FALSE); ------------------ | | 713| 1| (mask)->set_system_affinity(abort_bool) ------------------ 4644| 1| } 4645| 1| KMP_CPU_FREE(__kmp_affin_origMask); ------------------ | | 698| 1|#define KMP_CPU_FREE(ptr) __kmp_affinity_dispatch->deallocate_mask(ptr) ------------------ 4646| 1| __kmp_affin_origMask = NULL; 4647| 1| } 4648| 2| __kmp_affinity_num_places = 0; 4649| 2| if (procarr != NULL) { ------------------ | Branch (4649:7): [True: 0, False: 2] ------------------ 4650| 0| __kmp_free(procarr); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 4651| 0| procarr = NULL; 4652| 0| } 4653| 2| if (__kmp_osid_to_hwthread_map) { ------------------ | Branch (4653:7): [True: 1, False: 1] ------------------ 4654| 1| __kmp_free(__kmp_osid_to_hwthread_map); ------------------ | | 3619| 1|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 4655| 1| __kmp_osid_to_hwthread_map = NULL; 4656| 1| } 4657| |#if KMP_USE_HWLOC 4658| | if (__kmp_hwloc_topology != NULL) { 4659| | hwloc_topology_destroy(__kmp_hwloc_topology); 4660| | __kmp_hwloc_topology = NULL; 4661| | } 4662| |#endif 4663| 2| if (__kmp_hw_subset) { ------------------ | Branch (4663:7): [True: 0, False: 2] ------------------ 4664| 0| kmp_hw_subset_t::deallocate(__kmp_hw_subset); 4665| 0| __kmp_hw_subset = nullptr; 4666| 0| } 4667| 2| if (__kmp_topology) { ------------------ | Branch (4667:7): [True: 1, False: 1] ------------------ 4668| 1| kmp_topology_t::deallocate(__kmp_topology); 4669| 1| __kmp_topology = nullptr; 4670| 1| } 4671| 2| KMPAffinity::destroy_api(); 4672| 2|} __kmp_affinity_set_init_mask: 4691| 32|void __kmp_affinity_set_init_mask(int gtid, int isa_root) { 4692| | 4693| 32| kmp_info_t *th = (kmp_info_t *)TCR_SYNC_PTR(__kmp_threads[gtid]); ------------------ | | 1158| 32|#define TCR_SYNC_PTR(a) ((void *)TCR_SYNC_8(a)) | | ------------------ | | | | 1140| 32|#define TCR_SYNC_8(a) (a) | | ------------------ ------------------ 4694| | 4695| | // Set the thread topology information to default of unknown 4696| 416| for (int id = 0; id < KMP_HW_LAST; ++id) ------------------ | Branch (4696:20): [True: 384, False: 32] ------------------ 4697| 384| th->th.th_topology_ids[id] = kmp_hw_thread_t::UNKNOWN_ID; 4698| 32| th->th.th_topology_attrs = KMP_AFFINITY_ATTRS_UNKNOWN; ------------------ | | 891| 32| { KMP_HW_CORE_TYPE_UNKNOWN, kmp_hw_attr_t::UNKNOWN_CORE_EFF, 0, 0 } ------------------ 4699| | 4700| 32| if (!KMP_AFFINITY_CAPABLE()) { ------------------ | | 684| 32|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) ------------------ | Branch (4700:7): [True: 0, False: 32] ------------------ 4701| 0| return; 4702| 0| } 4703| | 4704| 32| if (th->th.th_affin_mask == NULL) { ------------------ | Branch (4704:7): [True: 32, False: 0] ------------------ 4705| 32| KMP_CPU_ALLOC(th->th.th_affin_mask); ------------------ | | 697| 32|#define KMP_CPU_ALLOC(ptr) (ptr = __kmp_affinity_dispatch->allocate_mask()) ------------------ 4706| 32| } else { 4707| 0| KMP_CPU_ZERO(th->th.th_affin_mask); ------------------ | | 692| 0|#define KMP_CPU_ZERO(mask) (mask)->zero() ------------------ 4708| 0| } 4709| | 4710| | // Copy the thread mask to the kmp_info_t structure. If 4711| | // __kmp_affinity.type == affinity_none, copy the "full" mask, i.e. 4712| | // one that has all of the OS proc ids set, or if 4713| | // __kmp_affinity.flags.respect is set, then the full mask is the 4714| | // same as the mask of the initialization thread. 4715| 32| kmp_affin_mask_t *mask; 4716| 32| int i; 4717| 32| const kmp_affinity_t *affinity; 4718| 32| const char *env_var; 4719| 32| bool is_hidden_helper = KMP_HIDDEN_HELPER_THREAD(gtid); ------------------ | | 4417| 32| ((gtid) >= 1 && (gtid) <= __kmp_hidden_helper_threads_num) | | ------------------ | | | Branch (4417:4): [True: 31, False: 1] | | | Branch (4417:19): [True: 0, False: 31] | | ------------------ ------------------ 4720| | 4721| 32| if (is_hidden_helper) ------------------ | Branch (4721:7): [True: 0, False: 32] ------------------ 4722| 0| affinity = &__kmp_hh_affinity; 4723| 32| else 4724| 32| affinity = &__kmp_affinity; 4725| 32| env_var = affinity->env_var; 4726| | 4727| 32| if (KMP_AFFINITY_NON_PROC_BIND || is_hidden_helper) { ------------------ | | 963| 64| ((__kmp_nested_proc_bind.bind_types[0] == proc_bind_false || \ | | ------------------ | | | Branch (963:5): [True: 32, False: 0] | | ------------------ | | 964| 32| __kmp_nested_proc_bind.bind_types[0] == proc_bind_intel) && \ | | ------------------ | | | Branch (964:5): [True: 0, False: 0] | | ------------------ | | 965| 64| (__kmp_affinity.num_masks > 0 || __kmp_affinity.type == affinity_balanced)) | | ------------------ | | | Branch (965:5): [True: 32, False: 0] | | | Branch (965:37): [True: 0, False: 0] | | ------------------ ------------------ | Branch (4727:37): [True: 0, False: 0] ------------------ 4728| 32| if ((affinity->type == affinity_none) || ------------------ | Branch (4728:9): [True: 32, False: 0] ------------------ 4729| 0| (affinity->type == affinity_balanced) || ------------------ | Branch (4729:9): [True: 0, False: 0] ------------------ 4730| 32| KMP_HIDDEN_HELPER_MAIN_THREAD(gtid)) { ------------------ | | 4423| 0| ((gtid) == 1 && (gtid) <= __kmp_hidden_helper_threads_num) | | ------------------ | | | Branch (4423:4): [True: 0, False: 0] | | | Branch (4423:19): [True: 0, False: 0] | | ------------------ ------------------ 4731| |#if KMP_GROUP_AFFINITY 4732| | if (__kmp_num_proc_groups > 1) { 4733| | return; 4734| | } 4735| |#endif 4736| 32| KMP_ASSERT(__kmp_affin_fullMask != NULL); ------------------ | | 50| 32| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 32] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4737| 32| i = 0; 4738| 32| mask = __kmp_affin_fullMask; 4739| 32| } else { 4740| 0| __kmp_select_mask_by_gtid(gtid, affinity, &i, &mask); 4741| 0| } 4742| 32| } else { 4743| 0| if (!isa_root || __kmp_nested_proc_bind.bind_types[0] == proc_bind_false) { ------------------ | Branch (4743:9): [True: 0, False: 0] | Branch (4743:22): [True: 0, False: 0] ------------------ 4744| |#if KMP_GROUP_AFFINITY 4745| | if (__kmp_num_proc_groups > 1) { 4746| | return; 4747| | } 4748| |#endif 4749| 0| KMP_ASSERT(__kmp_affin_fullMask != NULL); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4750| 0| i = KMP_PLACE_ALL; ------------------ | | 959| 0|#define KMP_PLACE_ALL (-1) ------------------ 4751| 0| mask = __kmp_affin_fullMask; 4752| 0| } else { 4753| 0| __kmp_select_mask_by_gtid(gtid, affinity, &i, &mask); 4754| 0| } 4755| 0| } 4756| | 4757| 32| th->th.th_current_place = i; 4758| 32| if (isa_root && !is_hidden_helper) { ------------------ | Branch (4758:7): [True: 1, False: 31] | Branch (4758:19): [True: 1, False: 0] ------------------ 4759| 1| th->th.th_new_place = i; 4760| 1| th->th.th_first_place = 0; 4761| 1| th->th.th_last_place = affinity->num_masks - 1; 4762| 31| } else if (KMP_AFFINITY_NON_PROC_BIND) { ------------------ | | 963| 31| ((__kmp_nested_proc_bind.bind_types[0] == proc_bind_false || \ | | ------------------ | | | Branch (963:5): [True: 31, False: 0] | | ------------------ | | 964| 31| __kmp_nested_proc_bind.bind_types[0] == proc_bind_intel) && \ | | ------------------ | | | Branch (964:5): [True: 0, False: 0] | | ------------------ | | 965| 31| (__kmp_affinity.num_masks > 0 || __kmp_affinity.type == affinity_balanced)) | | ------------------ | | | Branch (965:5): [True: 31, False: 0] | | | Branch (965:37): [True: 0, False: 0] | | ------------------ ------------------ 4763| | // When using a Non-OMP_PROC_BIND affinity method, 4764| | // set all threads' place-partition-var to the entire place list 4765| 31| th->th.th_first_place = 0; 4766| 31| th->th.th_last_place = affinity->num_masks - 1; 4767| 31| } 4768| | // Copy topology information associated with the place 4769| 32| if (i >= 0) { ------------------ | Branch (4769:7): [True: 32, False: 0] ------------------ 4770| 32| th->th.th_topology_ids = __kmp_affinity.ids[i]; 4771| 32| th->th.th_topology_attrs = __kmp_affinity.attrs[i]; 4772| 32| } 4773| | 4774| 32| if (i == KMP_PLACE_ALL) { ------------------ | | 959| 32|#define KMP_PLACE_ALL (-1) ------------------ | Branch (4774:7): [True: 0, False: 32] ------------------ 4775| 0| KA_TRACE(100, ("__kmp_affinity_set_init_mask: binding T#%d to all places\n", 4776| 0| gtid)); 4777| 32| } else { 4778| 32| KA_TRACE(100, ("__kmp_affinity_set_init_mask: binding T#%d to place %d\n", 4779| 32| gtid, i)); 4780| 32| } 4781| | 4782| 32| KMP_CPU_COPY(th->th.th_affin_mask, mask); ------------------ | | 693| 32|#define KMP_CPU_COPY(dest, src) (dest)->copy(src) ------------------ 4783| | 4784| | /* to avoid duplicate printing (will be correctly printed on barrier) */ 4785| 32| if (affinity->flags.verbose && ------------------ | Branch (4785:7): [True: 0, False: 32] ------------------ 4786| 0| (affinity->type == affinity_none || ------------------ | Branch (4786:8): [True: 0, False: 0] ------------------ 4787| 0| (i != KMP_PLACE_ALL && affinity->type != affinity_balanced)) && ------------------ | | 959| 0|#define KMP_PLACE_ALL (-1) ------------------ | Branch (4787:9): [True: 0, False: 0] | Branch (4787:31): [True: 0, False: 0] ------------------ 4788| 0| !KMP_HIDDEN_HELPER_MAIN_THREAD(gtid)) { ------------------ | | 4423| 0| ((gtid) == 1 && (gtid) <= __kmp_hidden_helper_threads_num) | | ------------------ | | | Branch (4423:4): [True: 0, False: 0] | | | Branch (4423:19): [True: 0, False: 0] | | ------------------ ------------------ 4789| 0| char buf[KMP_AFFIN_MASK_PRINT_LEN]; 4790| 0| __kmp_affinity_print_mask(buf, KMP_AFFIN_MASK_PRINT_LEN, ------------------ | | 828| 0|#define KMP_AFFIN_MASK_PRINT_LEN 1024 ------------------ 4791| 0| th->th.th_affin_mask); 4792| 0| KMP_INFORM(BoundToOSProcSet, env_var, (kmp_int32)getpid(), __kmp_gettid(), ------------------ | | 143| 0| __kmp_msg(kmp_ms_inform, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 4793| 0| gtid, buf); 4794| 0| } 4795| | 4796| |#if KMP_OS_WINDOWS 4797| | // On Windows* OS, the process affinity mask might have changed. If the user 4798| | // didn't request affinity and this call fails, just continue silently. 4799| | // See CQ171393. 4800| | if (affinity->type == affinity_none) { 4801| | __kmp_set_system_affinity(th->th.th_affin_mask, FALSE); 4802| | } else 4803| |#endif 4804| 32| __kmp_set_system_affinity(th->th.th_affin_mask, TRUE); ------------------ | | 713| 32| (mask)->set_system_affinity(abort_bool) ------------------ 4805| 32|} kmp_affinity.cpp:_ZL29__kmp_aux_affinity_initializeR14kmp_affinity_t: 4398| 1|static void __kmp_aux_affinity_initialize(kmp_affinity_t &affinity) { 4399| 1| bool is_regular_affinity = (&affinity == &__kmp_affinity); 4400| 1| bool is_hidden_helper_affinity = (&affinity == &__kmp_hh_affinity); 4401| 1| const char *env_var = affinity.env_var; 4402| | 4403| 1| if (affinity.flags.initialized) { ------------------ | Branch (4403:7): [True: 0, False: 1] ------------------ 4404| 0| KMP_ASSERT(__kmp_affin_fullMask != NULL); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4405| 0| return; 4406| 0| } 4407| | 4408| 1| if (is_regular_affinity && (!__kmp_affin_fullMask || !__kmp_affin_origMask)) ------------------ | Branch (4408:7): [True: 1, False: 0] | Branch (4408:31): [True: 1, False: 0] | Branch (4408:56): [True: 0, False: 0] ------------------ 4409| 1| __kmp_aux_affinity_initialize_masks(affinity); 4410| | 4411| 1| if (is_regular_affinity && !__kmp_topology) { ------------------ | Branch (4411:7): [True: 1, False: 0] | Branch (4411:30): [True: 1, False: 0] ------------------ 4412| 1| bool success = __kmp_aux_affinity_initialize_topology(affinity); 4413| 1| if (success) { ------------------ | Branch (4413:9): [True: 1, False: 0] ------------------ 4414| | // Initialize other data structures which depend on the topology 4415| 1| machine_hierarchy.init(__kmp_topology->get_num_hw_threads()); 4416| 1| KMP_ASSERT(__kmp_avail_proc == __kmp_topology->get_num_hw_threads()); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4417| 1| } else { 4418| 0| affinity.type = affinity_none; 4419| 0| KMP_AFFINITY_DISABLE(); ------------------ | | 685| 0|#define KMP_AFFINITY_DISABLE() (__kmp_affin_mask_size = 0) ------------------ 4420| 0| } 4421| 1| } 4422| | 4423| | // If KMP_AFFINITY=none, then only create the single "none" place 4424| | // which is the process's initial affinity mask or the number of 4425| | // hardware threads depending on respect,norespect 4426| 1| if (affinity.type == affinity_none) { ------------------ | Branch (4426:7): [True: 1, False: 0] ------------------ 4427| 1| __kmp_create_affinity_none_places(affinity); 4428| |#if KMP_USE_HIER_SCHED 4429| | __kmp_dispatch_set_hierarchy_values(); 4430| |#endif 4431| 1| affinity.flags.initialized = TRUE; ------------------ | | 1280| 1|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1|#define FALSE 0 | | ------------------ ------------------ 4432| 1| return; 4433| 1| } 4434| | 4435| 0| __kmp_topology->set_granularity(affinity); 4436| 0| int depth = __kmp_topology->get_depth(); 4437| | 4438| | // Create the table of masks, indexed by thread Id. 4439| 0| unsigned numUnique; 4440| 0| __kmp_create_os_id_masks(&numUnique, affinity); 4441| 0| if (affinity.gran_levels == 0) { ------------------ | Branch (4441:7): [True: 0, False: 0] ------------------ 4442| 0| KMP_DEBUG_ASSERT((int)numUnique == __kmp_avail_proc); 4443| 0| } 4444| | 4445| 0| switch (affinity.type) { 4446| | 4447| 0| case affinity_explicit: ------------------ | Branch (4447:3): [True: 0, False: 0] ------------------ 4448| 0| KMP_DEBUG_ASSERT(affinity.proclist != NULL); 4449| 0| if (is_hidden_helper_affinity || ------------------ | Branch (4449:9): [True: 0, False: 0] ------------------ 4450| 0| __kmp_nested_proc_bind.bind_types[0] == proc_bind_intel) { ------------------ | Branch (4450:9): [True: 0, False: 0] ------------------ 4451| 0| __kmp_affinity_process_proclist(affinity); 4452| 0| } else { 4453| 0| __kmp_affinity_process_placelist(affinity); 4454| 0| } 4455| 0| if (affinity.num_masks == 0) { ------------------ | Branch (4455:9): [True: 0, False: 0] ------------------ 4456| 0| KMP_AFF_WARNING(affinity, AffNoValidProcID); ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 4457| 0| affinity.type = affinity_none; 4458| 0| __kmp_create_affinity_none_places(affinity); 4459| 0| affinity.flags.initialized = TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 4460| 0| return; 4461| 0| } 4462| 0| break; 4463| | 4464| | // The other affinity types rely on sorting the hardware threads according to 4465| | // some permutation of the machine topology tree. Set affinity.compact 4466| | // and affinity.offset appropriately, then jump to a common code 4467| | // fragment to do the sort and create the array of affinity masks. 4468| 0| case affinity_logical: ------------------ | Branch (4468:3): [True: 0, False: 0] ------------------ 4469| 0| affinity.compact = 0; 4470| 0| if (affinity.offset) { ------------------ | Branch (4470:9): [True: 0, False: 0] ------------------ 4471| 0| affinity.offset = 4472| 0| __kmp_nThreadsPerCore * affinity.offset % __kmp_avail_proc; 4473| 0| } 4474| 0| goto sortTopology; 4475| | 4476| 0| case affinity_physical: ------------------ | Branch (4476:3): [True: 0, False: 0] ------------------ 4477| 0| if (__kmp_nThreadsPerCore > 1) { ------------------ | Branch (4477:9): [True: 0, False: 0] ------------------ 4478| 0| affinity.compact = 1; 4479| 0| if (affinity.compact >= depth) { ------------------ | Branch (4479:11): [True: 0, False: 0] ------------------ 4480| 0| affinity.compact = 0; 4481| 0| } 4482| 0| } else { 4483| 0| affinity.compact = 0; 4484| 0| } 4485| 0| if (affinity.offset) { ------------------ | Branch (4485:9): [True: 0, False: 0] ------------------ 4486| 0| affinity.offset = 4487| 0| __kmp_nThreadsPerCore * affinity.offset % __kmp_avail_proc; 4488| 0| } 4489| 0| goto sortTopology; 4490| | 4491| 0| case affinity_scatter: ------------------ | Branch (4491:3): [True: 0, False: 0] ------------------ 4492| 0| if (affinity.compact >= depth) { ------------------ | Branch (4492:9): [True: 0, False: 0] ------------------ 4493| 0| affinity.compact = 0; 4494| 0| } else { 4495| 0| affinity.compact = depth - 1 - affinity.compact; 4496| 0| } 4497| 0| goto sortTopology; 4498| | 4499| 0| case affinity_compact: ------------------ | Branch (4499:3): [True: 0, False: 0] ------------------ 4500| 0| if (affinity.compact >= depth) { ------------------ | Branch (4500:9): [True: 0, False: 0] ------------------ 4501| 0| affinity.compact = depth - 1; 4502| 0| } 4503| 0| goto sortTopology; 4504| | 4505| 0| case affinity_balanced: ------------------ | Branch (4505:3): [True: 0, False: 0] ------------------ 4506| 0| if (depth <= 1 || is_hidden_helper_affinity) { ------------------ | Branch (4506:9): [True: 0, False: 0] | Branch (4506:23): [True: 0, False: 0] ------------------ 4507| 0| KMP_AFF_WARNING(affinity, AffBalancedNotAvail, env_var); ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 4508| 0| affinity.type = affinity_none; 4509| 0| __kmp_create_affinity_none_places(affinity); 4510| 0| affinity.flags.initialized = TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 4511| 0| return; 4512| 0| } else if (!__kmp_topology->is_uniform()) { ------------------ | Branch (4512:16): [True: 0, False: 0] ------------------ 4513| | // Save the depth for further usage 4514| 0| __kmp_aff_depth = depth; 4515| | 4516| 0| int core_level = 4517| 0| __kmp_affinity_find_core_level(__kmp_avail_proc, depth - 1); 4518| 0| int ncores = __kmp_affinity_compute_ncores(__kmp_avail_proc, depth - 1, 4519| 0| core_level); 4520| 0| int maxprocpercore = __kmp_affinity_max_proc_per_core( 4521| 0| __kmp_avail_proc, depth - 1, core_level); 4522| | 4523| 0| int nproc = ncores * maxprocpercore; 4524| 0| if ((nproc < 2) || (nproc < __kmp_avail_proc)) { ------------------ | Branch (4524:11): [True: 0, False: 0] | Branch (4524:26): [True: 0, False: 0] ------------------ 4525| 0| KMP_AFF_WARNING(affinity, AffBalancedNotAvail, env_var); ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 4526| 0| affinity.type = affinity_none; 4527| 0| __kmp_create_affinity_none_places(affinity); 4528| 0| affinity.flags.initialized = TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 4529| 0| return; 4530| 0| } 4531| | 4532| 0| procarr = (int *)__kmp_allocate(sizeof(int) * nproc); ------------------ | | 3617| 0|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 4533| 0| for (int i = 0; i < nproc; i++) { ------------------ | Branch (4533:23): [True: 0, False: 0] ------------------ 4534| 0| procarr[i] = -1; 4535| 0| } 4536| | 4537| 0| int lastcore = -1; 4538| 0| int inlastcore = 0; 4539| 0| for (int i = 0; i < __kmp_avail_proc; i++) { ------------------ | Branch (4539:23): [True: 0, False: 0] ------------------ 4540| 0| int proc = __kmp_topology->at(i).os_id; 4541| 0| int core = __kmp_affinity_find_core(i, depth - 1, core_level); 4542| | 4543| 0| if (core == lastcore) { ------------------ | Branch (4543:13): [True: 0, False: 0] ------------------ 4544| 0| inlastcore++; 4545| 0| } else { 4546| 0| inlastcore = 0; 4547| 0| } 4548| 0| lastcore = core; 4549| | 4550| 0| procarr[core * maxprocpercore + inlastcore] = proc; 4551| 0| } 4552| 0| } 4553| 0| if (affinity.compact >= depth) { ------------------ | Branch (4553:9): [True: 0, False: 0] ------------------ 4554| 0| affinity.compact = depth - 1; 4555| 0| } 4556| | 4557| 0| sortTopology: 4558| | // Allocate the gtid->affinity mask table. 4559| 0| if (affinity.flags.dups) { ------------------ | Branch (4559:9): [True: 0, False: 0] ------------------ 4560| 0| affinity.num_masks = __kmp_avail_proc; 4561| 0| } else { 4562| 0| affinity.num_masks = numUnique; 4563| 0| } 4564| | 4565| 0| if ((__kmp_nested_proc_bind.bind_types[0] != proc_bind_intel) && ------------------ | Branch (4565:9): [True: 0, False: 0] ------------------ 4566| 0| (__kmp_affinity_num_places > 0) && ------------------ | Branch (4566:9): [True: 0, False: 0] ------------------ 4567| 0| ((unsigned)__kmp_affinity_num_places < affinity.num_masks) && ------------------ | Branch (4567:9): [True: 0, False: 0] ------------------ 4568| 0| !is_hidden_helper_affinity) { ------------------ | Branch (4568:9): [True: 0, False: 0] ------------------ 4569| 0| affinity.num_masks = __kmp_affinity_num_places; 4570| 0| } 4571| | 4572| 0| KMP_CPU_ALLOC_ARRAY(affinity.masks, affinity.num_masks); ------------------ | | 705| 0| (arr = __kmp_affinity_dispatch->allocate_mask_array(n)) ------------------ 4573| | 4574| | // Sort the topology table according to the current setting of 4575| | // affinity.compact, then fill out affinity.masks. 4576| 0| __kmp_topology->sort_compact(affinity); 4577| 0| { 4578| 0| int i; 4579| 0| unsigned j; 4580| 0| int num_hw_threads = __kmp_topology->get_num_hw_threads(); 4581| 0| for (i = 0, j = 0; i < num_hw_threads; i++) { ------------------ | Branch (4581:26): [True: 0, False: 0] ------------------ 4582| 0| if ((!affinity.flags.dups) && (!__kmp_topology->at(i).leader)) { ------------------ | Branch (4582:13): [True: 0, False: 0] | Branch (4582:39): [True: 0, False: 0] ------------------ 4583| 0| continue; 4584| 0| } 4585| 0| int osId = __kmp_topology->at(i).os_id; 4586| | 4587| 0| kmp_affin_mask_t *src = KMP_CPU_INDEX(affinity.os_id_masks, osId); ------------------ | | 703| 0|#define KMP_CPU_INDEX(arr, i) __kmp_affinity_dispatch->index_mask_array(arr, i) ------------------ 4588| 0| kmp_affin_mask_t *dest = KMP_CPU_INDEX(affinity.masks, j); ------------------ | | 703| 0|#define KMP_CPU_INDEX(arr, i) __kmp_affinity_dispatch->index_mask_array(arr, i) ------------------ 4589| 0| KMP_ASSERT(KMP_CPU_ISSET(osId, src)); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4590| 0| KMP_CPU_COPY(dest, src); ------------------ | | 693| 0|#define KMP_CPU_COPY(dest, src) (dest)->copy(src) ------------------ 4591| 0| if (++j >= affinity.num_masks) { ------------------ | Branch (4591:13): [True: 0, False: 0] ------------------ 4592| 0| break; 4593| 0| } 4594| 0| } 4595| 0| KMP_DEBUG_ASSERT(j == affinity.num_masks); 4596| 0| } 4597| | // Sort the topology back using ids 4598| 0| __kmp_topology->sort_ids(); 4599| 0| break; 4600| | 4601| 0| default: ------------------ | Branch (4601:3): [True: 0, False: 0] ------------------ 4602| 0| KMP_ASSERT2(0, "Unexpected affinity setting"); ------------------ | | 53| 0|#define KMP_ASSERT2(cond, msg) KMP_ASSERT(cond) | | ------------------ | | | | 50| 0| if (!(cond)) { \ | | | | ------------------ | | | | | Branch (50:7): [True: 0, Folded] | | | | ------------------ | | | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | | | 52| 0| } | | ------------------ ------------------ 4603| 0| } 4604| 0| __kmp_affinity_get_topology_info(affinity); 4605| 0| affinity.flags.initialized = TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 4606| 0|} kmp_affinity.cpp:_ZL35__kmp_aux_affinity_initialize_masksR14kmp_affinity_t: 4160| 1|static void __kmp_aux_affinity_initialize_masks(kmp_affinity_t &affinity) { 4161| | // Create the "full" mask - this defines all of the processors that we 4162| | // consider to be in the machine model. If respect is set, then it is the 4163| | // initialization thread's affinity mask. Otherwise, it is all processors that 4164| | // we know about on the machine. 4165| 1| int verbose = affinity.flags.verbose; 4166| 1| const char *env_var = affinity.env_var; 4167| | 4168| | // Already initialized 4169| 1| if (__kmp_affin_fullMask && __kmp_affin_origMask) ------------------ | Branch (4169:7): [True: 0, False: 1] | Branch (4169:31): [True: 0, False: 0] ------------------ 4170| 0| return; 4171| | 4172| 1| if (__kmp_affin_fullMask == NULL) { ------------------ | Branch (4172:7): [True: 1, False: 0] ------------------ 4173| 1| KMP_CPU_ALLOC(__kmp_affin_fullMask); ------------------ | | 697| 1|#define KMP_CPU_ALLOC(ptr) (ptr = __kmp_affinity_dispatch->allocate_mask()) ------------------ 4174| 1| } 4175| 1| if (__kmp_affin_origMask == NULL) { ------------------ | Branch (4175:7): [True: 1, False: 0] ------------------ 4176| 1| KMP_CPU_ALLOC(__kmp_affin_origMask); ------------------ | | 697| 1|#define KMP_CPU_ALLOC(ptr) (ptr = __kmp_affinity_dispatch->allocate_mask()) ------------------ 4177| 1| } 4178| 1| if (KMP_AFFINITY_CAPABLE()) { ------------------ | | 684| 1|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) | | ------------------ | | | Branch (684:32): [True: 1, False: 0] | | ------------------ ------------------ 4179| 1| __kmp_get_system_affinity(__kmp_affin_fullMask, TRUE); ------------------ | | 711| 1| (mask)->get_system_affinity(abort_bool) ------------------ 4180| | // Make a copy before possible expanding to the entire machine mask 4181| 1| __kmp_affin_origMask->copy(__kmp_affin_fullMask); 4182| 1| if (affinity.flags.respect) { ------------------ | Branch (4182:9): [True: 1, False: 0] ------------------ 4183| | // Count the number of available processors. 4184| 1| unsigned i; 4185| 1| __kmp_avail_proc = 0; 4186| 32| KMP_CPU_SET_ITERATE(i, __kmp_affin_fullMask) { ------------------ | | 688| 33| for (i = (mask)->begin(); (int)i != (mask)->end(); i = (mask)->next(i)) | | ------------------ | | | Branch (688:29): [True: 32, False: 1] | | ------------------ ------------------ 4187| 32| if (!KMP_CPU_ISSET(i, __kmp_affin_fullMask)) { ------------------ | | 690| 32|#define KMP_CPU_ISSET(i, mask) (mask)->is_set(i) ------------------ | Branch (4187:13): [True: 0, False: 32] ------------------ 4188| 0| continue; 4189| 0| } 4190| 32| __kmp_avail_proc++; 4191| 32| } 4192| 1| if (__kmp_avail_proc > __kmp_xproc) { ------------------ | Branch (4192:11): [True: 0, False: 1] ------------------ 4193| 0| KMP_AFF_WARNING(affinity, ErrorInitializeAffinity); ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 4194| 0| affinity.type = affinity_none; 4195| 0| KMP_AFFINITY_DISABLE(); ------------------ | | 685| 0|#define KMP_AFFINITY_DISABLE() (__kmp_affin_mask_size = 0) ------------------ 4196| 0| return; 4197| 0| } 4198| | 4199| 1| if (verbose) { ------------------ | Branch (4199:11): [True: 0, False: 1] ------------------ 4200| 0| char buf[KMP_AFFIN_MASK_PRINT_LEN]; 4201| 0| __kmp_affinity_print_mask(buf, KMP_AFFIN_MASK_PRINT_LEN, ------------------ | | 828| 0|#define KMP_AFFIN_MASK_PRINT_LEN 1024 ------------------ 4202| 0| __kmp_affin_fullMask); 4203| 0| KMP_INFORM(InitOSProcSetRespect, env_var, buf); ------------------ | | 143| 0| __kmp_msg(kmp_ms_inform, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 4204| 0| } 4205| 1| } else { 4206| 0| if (verbose) { ------------------ | Branch (4206:11): [True: 0, False: 0] ------------------ 4207| 0| char buf[KMP_AFFIN_MASK_PRINT_LEN]; 4208| 0| __kmp_affinity_print_mask(buf, KMP_AFFIN_MASK_PRINT_LEN, ------------------ | | 828| 0|#define KMP_AFFIN_MASK_PRINT_LEN 1024 ------------------ 4209| 0| __kmp_affin_fullMask); 4210| 0| KMP_INFORM(InitOSProcSetNotRespect, env_var, buf); ------------------ | | 143| 0| __kmp_msg(kmp_ms_inform, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 4211| 0| } 4212| 0| __kmp_avail_proc = 4213| 0| __kmp_affinity_entire_machine_mask(__kmp_affin_fullMask); 4214| |#if KMP_OS_WINDOWS 4215| | if (__kmp_num_proc_groups <= 1) { 4216| | // Copy expanded full mask if topology has single processor group 4217| | __kmp_affin_origMask->copy(__kmp_affin_fullMask); 4218| | } 4219| | // Set the process affinity mask since threads' affinity 4220| | // masks must be subset of process mask in Windows* OS 4221| | __kmp_affin_fullMask->set_process_affinity(true); 4222| |#endif 4223| 0| } 4224| 1| } 4225| 1|} kmp_affinity.cpp:_ZL38__kmp_aux_affinity_initialize_topologyR14kmp_affinity_t: 4227| 1|static bool __kmp_aux_affinity_initialize_topology(kmp_affinity_t &affinity) { 4228| 1| bool success = false; 4229| 1| const char *env_var = affinity.env_var; 4230| 1| kmp_i18n_id_t msg_id = kmp_i18n_null; 4231| 1| int verbose = affinity.flags.verbose; 4232| | 4233| | // For backward compatibility, setting KMP_CPUINFO_FILE => 4234| | // KMP_TOPOLOGY_METHOD=cpuinfo 4235| 1| if ((__kmp_cpuinfo_file != NULL) && ------------------ | Branch (4235:7): [True: 0, False: 1] ------------------ 4236| 0| (__kmp_affinity_top_method == affinity_top_method_all)) { ------------------ | Branch (4236:7): [True: 0, False: 0] ------------------ 4237| 0| __kmp_affinity_top_method = affinity_top_method_cpuinfo; 4238| 0| } 4239| | 4240| 1| if (__kmp_affinity_top_method == affinity_top_method_all) { ------------------ | Branch (4240:7): [True: 1, False: 0] ------------------ 4241| |// In the default code path, errors are not fatal - we just try using 4242| |// another method. We only emit a warning message if affinity is on, or the 4243| |// verbose flag is set, an the nowarnings flag was not set. 4244| |#if KMP_USE_HWLOC 4245| | if (!success && 4246| | __kmp_affinity_dispatch->get_api_type() == KMPAffinity::HWLOC) { 4247| | if (!__kmp_hwloc_error) { 4248| | success = __kmp_affinity_create_hwloc_map(&msg_id); 4249| | if (!success && verbose) { 4250| | KMP_INFORM(AffIgnoringHwloc, env_var); 4251| | } 4252| | } else if (verbose) { 4253| | KMP_INFORM(AffIgnoringHwloc, env_var); 4254| | } 4255| | } 4256| |#endif 4257| | 4258| 1|#if KMP_ARCH_X86 || KMP_ARCH_X86_64 4259| 1| if (!success) { ------------------ | Branch (4259:9): [True: 1, False: 0] ------------------ 4260| 1| success = __kmp_affinity_create_x2apicid_map(&msg_id); 4261| 1| if (!success && verbose && msg_id != kmp_i18n_null) { ------------------ | Branch (4261:11): [True: 0, False: 1] | Branch (4261:23): [True: 0, False: 0] | Branch (4261:34): [True: 0, False: 0] ------------------ 4262| 0| KMP_INFORM(AffInfoStr, env_var, __kmp_i18n_catgets(msg_id)); ------------------ | | 143| 0| __kmp_msg(kmp_ms_inform, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 4263| 0| } 4264| 1| } 4265| 1| if (!success) { ------------------ | Branch (4265:9): [True: 0, False: 1] ------------------ 4266| 0| success = __kmp_affinity_create_apicid_map(&msg_id); 4267| 0| if (!success && verbose && msg_id != kmp_i18n_null) { ------------------ | Branch (4267:11): [True: 0, False: 0] | Branch (4267:23): [True: 0, False: 0] | Branch (4267:34): [True: 0, False: 0] ------------------ 4268| 0| KMP_INFORM(AffInfoStr, env_var, __kmp_i18n_catgets(msg_id)); ------------------ | | 143| 0| __kmp_msg(kmp_ms_inform, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 4269| 0| } 4270| 0| } 4271| 1|#endif /* KMP_ARCH_X86 || KMP_ARCH_X86_64 */ 4272| | 4273| 1|#if KMP_OS_LINUX 4274| 1| if (!success) { ------------------ | Branch (4274:9): [True: 0, False: 1] ------------------ 4275| 0| int line = 0; 4276| 0| success = __kmp_affinity_create_cpuinfo_map(&line, &msg_id); 4277| 0| if (!success && verbose && msg_id != kmp_i18n_null) { ------------------ | Branch (4277:11): [True: 0, False: 0] | Branch (4277:23): [True: 0, False: 0] | Branch (4277:34): [True: 0, False: 0] ------------------ 4278| 0| KMP_INFORM(AffInfoStr, env_var, __kmp_i18n_catgets(msg_id)); ------------------ | | 143| 0| __kmp_msg(kmp_ms_inform, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 4279| 0| } 4280| 0| } 4281| 1|#endif /* KMP_OS_LINUX */ 4282| | 4283| |#if KMP_GROUP_AFFINITY 4284| | if (!success && (__kmp_num_proc_groups > 1)) { 4285| | success = __kmp_affinity_create_proc_group_map(&msg_id); 4286| | if (!success && verbose && msg_id != kmp_i18n_null) { 4287| | KMP_INFORM(AffInfoStr, env_var, __kmp_i18n_catgets(msg_id)); 4288| | } 4289| | } 4290| |#endif /* KMP_GROUP_AFFINITY */ 4291| | 4292| 1| if (!success) { ------------------ | Branch (4292:9): [True: 0, False: 1] ------------------ 4293| 0| success = __kmp_affinity_create_flat_map(&msg_id); 4294| 0| if (!success && verbose && msg_id != kmp_i18n_null) { ------------------ | Branch (4294:11): [True: 0, False: 0] | Branch (4294:23): [True: 0, False: 0] | Branch (4294:34): [True: 0, False: 0] ------------------ 4295| 0| KMP_INFORM(AffInfoStr, env_var, __kmp_i18n_catgets(msg_id)); ------------------ | | 143| 0| __kmp_msg(kmp_ms_inform, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 4296| 0| } 4297| 0| KMP_ASSERT(success); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4298| 0| } 4299| 1| } 4300| | 4301| |// If the user has specified that a paricular topology discovery method is to be 4302| |// used, then we abort if that method fails. The exception is group affinity, 4303| |// which might have been implicitly set. 4304| |#if KMP_USE_HWLOC 4305| | else if (__kmp_affinity_top_method == affinity_top_method_hwloc) { 4306| | KMP_ASSERT(__kmp_affinity_dispatch->get_api_type() == KMPAffinity::HWLOC); 4307| | success = __kmp_affinity_create_hwloc_map(&msg_id); 4308| | if (!success) { 4309| | KMP_ASSERT(msg_id != kmp_i18n_null); 4310| | KMP_FATAL(MsgExiting, __kmp_i18n_catgets(msg_id)); 4311| | } 4312| | } 4313| |#endif // KMP_USE_HWLOC 4314| | 4315| 0|#if KMP_ARCH_X86 || KMP_ARCH_X86_64 4316| 0| else if (__kmp_affinity_top_method == affinity_top_method_x2apicid || ------------------ | Branch (4316:12): [True: 0, False: 0] ------------------ 4317| 0| __kmp_affinity_top_method == affinity_top_method_x2apicid_1f) { ------------------ | Branch (4317:12): [True: 0, False: 0] ------------------ 4318| 0| success = __kmp_affinity_create_x2apicid_map(&msg_id); 4319| 0| if (!success) { ------------------ | Branch (4319:9): [True: 0, False: 0] ------------------ 4320| 0| KMP_ASSERT(msg_id != kmp_i18n_null); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4321| 0| KMP_FATAL(MsgExiting, __kmp_i18n_catgets(msg_id)); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 4322| 0| } 4323| 0| } else if (__kmp_affinity_top_method == affinity_top_method_apicid) { ------------------ | Branch (4323:14): [True: 0, False: 0] ------------------ 4324| 0| success = __kmp_affinity_create_apicid_map(&msg_id); 4325| 0| if (!success) { ------------------ | Branch (4325:9): [True: 0, False: 0] ------------------ 4326| 0| KMP_ASSERT(msg_id != kmp_i18n_null); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4327| 0| KMP_FATAL(MsgExiting, __kmp_i18n_catgets(msg_id)); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 4328| 0| } 4329| 0| } 4330| 0|#endif /* KMP_ARCH_X86 || KMP_ARCH_X86_64 */ 4331| | 4332| 0| else if (__kmp_affinity_top_method == affinity_top_method_cpuinfo) { ------------------ | Branch (4332:12): [True: 0, False: 0] ------------------ 4333| 0| int line = 0; 4334| 0| success = __kmp_affinity_create_cpuinfo_map(&line, &msg_id); 4335| 0| if (!success) { ------------------ | Branch (4335:9): [True: 0, False: 0] ------------------ 4336| 0| KMP_ASSERT(msg_id != kmp_i18n_null); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4337| 0| const char *filename = __kmp_cpuinfo_get_filename(); 4338| 0| if (line > 0) { ------------------ | Branch (4338:11): [True: 0, False: 0] ------------------ 4339| 0| KMP_FATAL(FileLineMsgExiting, filename, line, ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 4340| 0| __kmp_i18n_catgets(msg_id)); 4341| 0| } else { 4342| 0| KMP_FATAL(FileMsgExiting, filename, __kmp_i18n_catgets(msg_id)); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 4343| 0| } 4344| 0| } 4345| 0| } 4346| | 4347| |#if KMP_GROUP_AFFINITY 4348| | else if (__kmp_affinity_top_method == affinity_top_method_group) { 4349| | success = __kmp_affinity_create_proc_group_map(&msg_id); 4350| | KMP_ASSERT(success); 4351| | if (!success) { 4352| | KMP_ASSERT(msg_id != kmp_i18n_null); 4353| | KMP_FATAL(MsgExiting, __kmp_i18n_catgets(msg_id)); 4354| | } 4355| | } 4356| |#endif /* KMP_GROUP_AFFINITY */ 4357| | 4358| 0| else if (__kmp_affinity_top_method == affinity_top_method_flat) { ------------------ | Branch (4358:12): [True: 0, False: 0] ------------------ 4359| 0| success = __kmp_affinity_create_flat_map(&msg_id); 4360| | // should not fail 4361| 0| KMP_ASSERT(success); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4362| 0| } 4363| | 4364| | // Early exit if topology could not be created 4365| 1| if (!__kmp_topology) { ------------------ | Branch (4365:7): [True: 0, False: 1] ------------------ 4366| 0| if (KMP_AFFINITY_CAPABLE()) { ------------------ | | 684| 0|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) | | ------------------ | | | Branch (684:32): [True: 0, False: 0] | | ------------------ ------------------ 4367| 0| KMP_AFF_WARNING(affinity, ErrorInitializeAffinity); ------------------ | | 145| 0| if (s.flags.verbose || (s.flags.warnings && (s.type != affinity_none))) { \ | | ------------------ | | | Branch (145:7): [True: 0, False: 0] | | | Branch (145:27): [True: 0, False: 0] | | | Branch (145:47): [True: 0, False: 0] | | ------------------ | | 146| 0| KMP_WARNING(__VA_ARGS__); \ | | ------------------ | | | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | ------------------ | | 147| 0| } ------------------ 4368| 0| } 4369| 0| if (nPackages > 0 && nCoresPerPkg > 0 && __kmp_nThreadsPerCore > 0 && ------------------ | Branch (4369:9): [True: 0, False: 0] | Branch (4369:26): [True: 0, False: 0] | Branch (4369:46): [True: 0, False: 0] ------------------ 4370| 0| __kmp_ncores > 0) { ------------------ | Branch (4370:9): [True: 0, False: 0] ------------------ 4371| 0| __kmp_topology = kmp_topology_t::allocate(0, 0, NULL); 4372| 0| __kmp_topology->canonicalize(nPackages, nCoresPerPkg, 4373| 0| __kmp_nThreadsPerCore, __kmp_ncores); 4374| 0| if (verbose) { ------------------ | Branch (4374:11): [True: 0, False: 0] ------------------ 4375| 0| __kmp_topology->print(env_var); 4376| 0| } 4377| 0| } 4378| 0| return false; 4379| 0| } 4380| | 4381| | // Canonicalize, print (if requested), apply KMP_HW_SUBSET 4382| 1| __kmp_topology->canonicalize(); 4383| 1| if (verbose) ------------------ | Branch (4383:7): [True: 0, False: 1] ------------------ 4384| 0| __kmp_topology->print(env_var); 4385| 1| bool filtered = __kmp_topology->filter_hw_subset(); 4386| 1| if (filtered) { ------------------ | Branch (4386:7): [True: 0, False: 1] ------------------ 4387| |#if KMP_OS_WINDOWS 4388| | // Copy filtered full mask if topology has single processor group 4389| | if (__kmp_num_proc_groups <= 1) 4390| |#endif 4391| 0| __kmp_affin_origMask->copy(__kmp_affin_fullMask); 4392| 0| } 4393| 1| if (filtered && verbose) ------------------ | Branch (4393:7): [True: 0, False: 1] | Branch (4393:19): [True: 0, False: 0] ------------------ 4394| 0| __kmp_topology->print("KMP_HW_SUBSET"); 4395| 1| return success; 4396| 1|} kmp_affinity.cpp:_ZL34__kmp_affinity_create_x2apicid_mapP11kmp_i18n_id: 2510| 1|static bool __kmp_affinity_create_x2apicid_map(kmp_i18n_id_t *const msg_id) { 2511| | 2512| 1| cpuid_level_info_t levels[INTEL_LEVEL_TYPE_LAST]; 2513| 1| kmp_hw_t types[INTEL_LEVEL_TYPE_LAST]; 2514| 1| unsigned levels_index; 2515| 1| kmp_cpuid buf; 2516| 1| kmp_uint64 known_levels; 2517| 1| int topology_leaf, highest_leaf, apic_id; 2518| 1| int num_leaves; 2519| 1| static int leaves[] = {0, 0}; 2520| | 2521| 1| kmp_i18n_id_t leaf_message_id; 2522| | 2523| 1| KMP_BUILD_ASSERT(sizeof(known_levels) * CHAR_BIT > KMP_HW_LAST); ------------------ | | 26| 1|#define KMP_BUILD_ASSERT(expr) static_assert(expr, "Build condition error") ------------------ 2524| | 2525| 1| *msg_id = kmp_i18n_null; 2526| 1| if (__kmp_affinity.flags.verbose) { ------------------ | Branch (2526:7): [True: 0, False: 1] ------------------ 2527| 0| KMP_INFORM(AffInfoStr, "KMP_AFFINITY", KMP_I18N_STR(Decodingx2APIC)); ------------------ | | 143| 0| __kmp_msg(kmp_ms_inform, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 2528| 0| } 2529| | 2530| | // Figure out the known topology levels 2531| 1| known_levels = 0ull; 2532| 7| for (int i = 0; i < INTEL_LEVEL_TYPE_LAST; ++i) { ------------------ | Branch (2532:19): [True: 6, False: 1] ------------------ 2533| 6| if (__kmp_intel_type_2_topology_type(i) != KMP_HW_UNKNOWN) { ------------------ | Branch (2533:9): [True: 6, False: 0] ------------------ 2534| 6| known_levels |= (1ull << i); 2535| 6| } 2536| 6| } 2537| | 2538| | // Get the highest cpuid leaf supported 2539| 1| __kmp_x86_cpuid(0, 0, &buf); 2540| 1| highest_leaf = buf.eax; 2541| | 2542| | // If a specific topology method was requested, only allow that specific leaf 2543| | // otherwise, try both leaves 31 and 11 in that order 2544| 1| num_leaves = 0; 2545| 1| if (__kmp_affinity_top_method == affinity_top_method_x2apicid) { ------------------ | Branch (2545:7): [True: 0, False: 1] ------------------ 2546| 0| num_leaves = 1; 2547| 0| leaves[0] = 11; 2548| 0| leaf_message_id = kmp_i18n_str_NoLeaf11Support; 2549| 1| } else if (__kmp_affinity_top_method == affinity_top_method_x2apicid_1f) { ------------------ | Branch (2549:14): [True: 0, False: 1] ------------------ 2550| 0| num_leaves = 1; 2551| 0| leaves[0] = 31; 2552| 0| leaf_message_id = kmp_i18n_str_NoLeaf31Support; 2553| 1| } else { 2554| 1| num_leaves = 2; 2555| 1| leaves[0] = 31; 2556| 1| leaves[1] = 11; 2557| 1| leaf_message_id = kmp_i18n_str_NoLeaf11Support; 2558| 1| } 2559| | 2560| | // Check to see if cpuid leaf 31 or 11 is supported. 2561| 1| __kmp_nThreadsPerCore = nCoresPerPkg = nPackages = 1; 2562| 1| topology_leaf = -1; 2563| 2| for (int i = 0; i < num_leaves; ++i) { ------------------ | Branch (2563:19): [True: 2, False: 0] ------------------ 2564| 2| int leaf = leaves[i]; 2565| 2| if (highest_leaf < leaf) ------------------ | Branch (2565:9): [True: 1, False: 1] ------------------ 2566| 1| continue; 2567| 1| __kmp_x86_cpuid(leaf, 0, &buf); 2568| 1| if (buf.ebx == 0) ------------------ | Branch (2568:9): [True: 0, False: 1] ------------------ 2569| 0| continue; 2570| 1| topology_leaf = leaf; 2571| 1| levels_index = __kmp_x2apicid_get_levels(leaf, levels, known_levels); 2572| 1| if (levels_index == 0) ------------------ | Branch (2572:9): [True: 0, False: 1] ------------------ 2573| 0| continue; 2574| 1| break; 2575| 1| } 2576| 1| if (topology_leaf == -1 || levels_index == 0) { ------------------ | Branch (2576:7): [True: 0, False: 1] | Branch (2576:30): [True: 0, False: 1] ------------------ 2577| 0| *msg_id = leaf_message_id; 2578| 0| return false; 2579| 0| } 2580| 1| KMP_ASSERT(levels_index <= INTEL_LEVEL_TYPE_LAST); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 2581| | 2582| | // The algorithm used starts by setting the affinity to each available thread 2583| | // and retrieving info from the cpuid instruction, so if we are not capable of 2584| | // calling __kmp_get_system_affinity() and __kmp_get_system_affinity(), then 2585| | // we need to do something else - use the defaults that we calculated from 2586| | // issuing cpuid without binding to each proc. 2587| 1| if (!KMP_AFFINITY_CAPABLE()) { ------------------ | | 684| 1|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) ------------------ | Branch (2587:7): [True: 0, False: 1] ------------------ 2588| | // Hack to try and infer the machine topology using only the data 2589| | // available from cpuid on the current thread, and __kmp_xproc. 2590| 0| KMP_ASSERT(__kmp_affinity.type == affinity_none); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 2591| 0| for (unsigned i = 0; i < levels_index; ++i) { ------------------ | Branch (2591:26): [True: 0, False: 0] ------------------ 2592| 0| if (levels[i].level_type == INTEL_LEVEL_TYPE_SMT) { ------------------ | Branch (2592:11): [True: 0, False: 0] ------------------ 2593| 0| __kmp_nThreadsPerCore = levels[i].nitems; 2594| 0| } else if (levels[i].level_type == INTEL_LEVEL_TYPE_CORE) { ------------------ | Branch (2594:18): [True: 0, False: 0] ------------------ 2595| 0| nCoresPerPkg = levels[i].nitems; 2596| 0| } 2597| 0| } 2598| 0| __kmp_ncores = __kmp_xproc / __kmp_nThreadsPerCore; 2599| 0| nPackages = (__kmp_xproc + nCoresPerPkg - 1) / nCoresPerPkg; 2600| 0| return true; 2601| 0| } 2602| | 2603| | // Allocate the data structure to be returned. 2604| 1| int depth = levels_index; 2605| 4| for (int i = depth - 1, j = 0; i >= 0; --i, ++j) ------------------ | Branch (2605:34): [True: 3, False: 1] ------------------ 2606| 3| types[j] = __kmp_intel_type_2_topology_type(levels[i].level_type); 2607| 1| __kmp_topology = 2608| 1| kmp_topology_t::allocate(__kmp_avail_proc, levels_index, types); 2609| | 2610| | // Insert equivalent cache types if they exist 2611| 1| kmp_cache_info_t cache_info; 2612| 1| for (size_t i = 0; i < cache_info.get_depth(); ++i) { ------------------ | Branch (2612:22): [True: 0, False: 1] ------------------ 2613| 0| const kmp_cache_info_t::info_t &info = cache_info[i]; 2614| 0| unsigned cache_mask = info.mask; 2615| 0| unsigned cache_level = info.level; 2616| 0| for (unsigned j = 0; j < levels_index; ++j) { ------------------ | Branch (2616:26): [True: 0, False: 0] ------------------ 2617| 0| unsigned hw_cache_mask = levels[j].cache_mask; 2618| 0| kmp_hw_t cache_type = kmp_cache_info_t::get_topology_type(cache_level); 2619| 0| if (hw_cache_mask == cache_mask && j < levels_index - 1) { ------------------ | Branch (2619:11): [True: 0, False: 0] | Branch (2619:42): [True: 0, False: 0] ------------------ 2620| 0| kmp_hw_t type = 2621| 0| __kmp_intel_type_2_topology_type(levels[j + 1].level_type); 2622| 0| __kmp_topology->set_equivalent_type(cache_type, type); 2623| 0| } 2624| 0| } 2625| 0| } 2626| | 2627| | // From here on, we can assume that it is safe to call 2628| | // __kmp_get_system_affinity() and __kmp_set_system_affinity(), even if 2629| | // __kmp_affinity.type = affinity_none. 2630| | 2631| | // Save the affinity mask for the current thread. 2632| 1| kmp_affinity_raii_t previous_affinity; 2633| | 2634| | // Run through each of the available contexts, binding the current thread 2635| | // to it, and obtaining the pertinent information using the cpuid instr. 2636| 1| unsigned int proc; 2637| 1| int hw_thread_index = 0; 2638| 32| KMP_CPU_SET_ITERATE(proc, __kmp_affin_fullMask) { ------------------ | | 688| 33| for (i = (mask)->begin(); (int)i != (mask)->end(); i = (mask)->next(i)) | | ------------------ | | | Branch (688:29): [True: 32, False: 1] | | ------------------ ------------------ 2639| 32| cpuid_level_info_t my_levels[INTEL_LEVEL_TYPE_LAST]; 2640| 32| unsigned my_levels_index; 2641| | 2642| | // Skip this proc if it is not included in the machine model. 2643| 32| if (!KMP_CPU_ISSET(proc, __kmp_affin_fullMask)) { ------------------ | | 690| 32|#define KMP_CPU_ISSET(i, mask) (mask)->is_set(i) ------------------ | Branch (2643:9): [True: 0, False: 32] ------------------ 2644| 0| continue; 2645| 0| } 2646| 32| KMP_DEBUG_ASSERT(hw_thread_index < __kmp_avail_proc); 2647| | 2648| 32| __kmp_affinity_dispatch->bind_thread(proc); 2649| | 2650| | // New algorithm 2651| 32| __kmp_x86_cpuid(topology_leaf, 0, &buf); 2652| 32| apic_id = buf.edx; 2653| 32| kmp_hw_thread_t &hw_thread = __kmp_topology->at(hw_thread_index); 2654| 32| my_levels_index = 2655| 32| __kmp_x2apicid_get_levels(topology_leaf, my_levels, known_levels); 2656| 32| if (my_levels_index == 0 || my_levels_index != levels_index) { ------------------ | Branch (2656:9): [True: 0, False: 32] | Branch (2656:33): [True: 0, False: 32] ------------------ 2657| 0| *msg_id = kmp_i18n_str_InvalidCpuidInfo; 2658| 0| return false; 2659| 0| } 2660| 32| hw_thread.clear(); 2661| 32| hw_thread.os_id = proc; 2662| | // Put in topology information 2663| 128| for (unsigned j = 0, idx = depth - 1; j < my_levels_index; ++j, --idx) { ------------------ | Branch (2663:43): [True: 96, False: 32] ------------------ 2664| 96| hw_thread.ids[idx] = apic_id & my_levels[j].mask; 2665| 96| if (j > 0) { ------------------ | Branch (2665:11): [True: 64, False: 32] ------------------ 2666| 64| hw_thread.ids[idx] >>= my_levels[j - 1].mask_width; 2667| 64| } 2668| 96| } 2669| | // Hybrid information 2670| 32| if (__kmp_is_hybrid_cpu() && highest_leaf >= 0x1a) { ------------------ | Branch (2670:9): [True: 0, False: 32] | Branch (2670:34): [True: 0, False: 0] ------------------ 2671| 0| kmp_hw_core_type_t type; 2672| 0| unsigned native_model_id; 2673| 0| int efficiency; 2674| 0| __kmp_get_hybrid_info(&type, &efficiency, &native_model_id); 2675| 0| hw_thread.attrs.set_core_type(type); 2676| 0| hw_thread.attrs.set_core_eff(efficiency); 2677| 0| } 2678| 32| hw_thread_index++; 2679| 32| } 2680| 1| KMP_ASSERT(hw_thread_index > 0); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 2681| 1| __kmp_topology->sort_ids(); 2682| 1| if (!__kmp_topology->check_ids()) { ------------------ | Branch (2682:7): [True: 0, False: 1] ------------------ 2683| 0| kmp_topology_t::deallocate(__kmp_topology); 2684| 0| __kmp_topology = nullptr; 2685| 0| *msg_id = kmp_i18n_str_x2ApicIDsNotUnique; 2686| 0| return false; 2687| 0| } 2688| 1| return true; 2689| 1|} kmp_affinity.cpp:_ZL32__kmp_intel_type_2_topology_typei: 2428| 9|static kmp_hw_t __kmp_intel_type_2_topology_type(int intel_type) { 2429| 9| switch (intel_type) { ------------------ | Branch (2429:11): [True: 9, False: 0] ------------------ 2430| 2| case INTEL_LEVEL_TYPE_INVALID: ------------------ | Branch (2430:3): [True: 2, False: 7] ------------------ 2431| 2| return KMP_HW_SOCKET; 2432| 2| case INTEL_LEVEL_TYPE_SMT: ------------------ | Branch (2432:3): [True: 2, False: 7] ------------------ 2433| 2| return KMP_HW_THREAD; 2434| 2| case INTEL_LEVEL_TYPE_CORE: ------------------ | Branch (2434:3): [True: 2, False: 7] ------------------ 2435| 2| return KMP_HW_CORE; 2436| 1| case INTEL_LEVEL_TYPE_TILE: ------------------ | Branch (2436:3): [True: 1, False: 8] ------------------ 2437| 1| return KMP_HW_TILE; 2438| 1| case INTEL_LEVEL_TYPE_MODULE: ------------------ | Branch (2438:3): [True: 1, False: 8] ------------------ 2439| 1| return KMP_HW_MODULE; 2440| 1| case INTEL_LEVEL_TYPE_DIE: ------------------ | Branch (2440:3): [True: 1, False: 8] ------------------ 2441| 1| return KMP_HW_DIE; 2442| 9| } 2443| 0| return KMP_HW_UNKNOWN; 2444| 9|} kmp_affinity.cpp:_ZL25__kmp_x2apicid_get_levelsiP18cpuid_level_info_ty: 2452| 33| kmp_uint64 known_levels) { 2453| 33| unsigned level, levels_index; 2454| 33| unsigned level_type, mask_width, nitems; 2455| 33| kmp_cpuid buf; 2456| | 2457| | // New algorithm has known topology layers act as highest unknown topology 2458| | // layers when unknown topology layers exist. 2459| | // e.g., Suppose layers were SMT CORE PACKAGE, where 2460| | // are unknown topology layers, Then SMT will take the characteristics of 2461| | // (SMT x ) and CORE will take the characteristics of (CORE x x ). 2462| | // This eliminates unknown portions of the topology while still keeping the 2463| | // correct structure. 2464| 33| level = levels_index = 0; 2465| 99| do { 2466| 99| __kmp_x86_cpuid(leaf, level, &buf); 2467| 99| level_type = __kmp_extract_bits<8, 15>(buf.ecx); 2468| 99| mask_width = __kmp_extract_bits<0, 4>(buf.eax); 2469| 99| nitems = __kmp_extract_bits<0, 15>(buf.ebx); 2470| 99| if (level_type != INTEL_LEVEL_TYPE_INVALID && nitems == 0) ------------------ | Branch (2470:9): [True: 66, False: 33] | Branch (2470:51): [True: 0, False: 66] ------------------ 2471| 0| return 0; 2472| | 2473| 99| if (known_levels & (1ull << level_type)) { ------------------ | Branch (2473:9): [True: 99, False: 0] ------------------ 2474| | // Add a new level to the topology 2475| 99| KMP_ASSERT(levels_index < INTEL_LEVEL_TYPE_LAST); ------------------ | | 50| 99| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 99] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 2476| 99| levels[levels_index].level_type = level_type; 2477| 99| levels[levels_index].mask_width = mask_width; 2478| 99| levels[levels_index].nitems = nitems; 2479| 99| levels_index++; 2480| 99| } else { 2481| | // If it is an unknown level, then logically move the previous layer up 2482| 0| if (levels_index > 0) { ------------------ | Branch (2482:11): [True: 0, False: 0] ------------------ 2483| 0| levels[levels_index - 1].mask_width = mask_width; 2484| 0| levels[levels_index - 1].nitems = nitems; 2485| 0| } 2486| 0| } 2487| 99| level++; 2488| 99| } while (level_type != INTEL_LEVEL_TYPE_INVALID); ------------------ | Branch (2488:12): [True: 66, False: 33] ------------------ 2489| | 2490| | // Ensure the INTEL_LEVEL_TYPE_INVALID (Socket) layer isn't first 2491| 33| if (levels_index == 0 || levels[0].level_type == INTEL_LEVEL_TYPE_INVALID) ------------------ | Branch (2491:7): [True: 0, False: 33] | Branch (2491:28): [True: 0, False: 33] ------------------ 2492| 0| return 0; 2493| | 2494| | // Set the masks to & with apicid 2495| 132| for (unsigned i = 0; i < levels_index; ++i) { ------------------ | Branch (2495:24): [True: 99, False: 33] ------------------ 2496| 99| if (levels[i].level_type != INTEL_LEVEL_TYPE_INVALID) { ------------------ | Branch (2496:9): [True: 66, False: 33] ------------------ 2497| 66| levels[i].mask = ~((-1) << levels[i].mask_width); 2498| 66| levels[i].cache_mask = (-1) << levels[i].mask_width; 2499| 99| for (unsigned j = 0; j < i; ++j) ------------------ | Branch (2499:28): [True: 33, False: 66] ------------------ 2500| 33| levels[i].mask ^= levels[j].mask; 2501| 66| } else { 2502| 33| KMP_DEBUG_ASSERT(i > 0); 2503| 33| levels[i].mask = (-1) << levels[i - 1].mask_width; 2504| 33| levels[i].cache_mask = 0; 2505| 33| } 2506| 99| } 2507| 33| return levels_index; 2508| 33|} kmp_affinity.cpp:_ZL18__kmp_extract_bitsILj8ELj15EEjj: 1956| 99|static inline unsigned __kmp_extract_bits(kmp_uint32 v) { 1957| 99| const kmp_uint32 SHIFT_LEFT = sizeof(kmp_uint32) * 8 - 1 - MSB; 1958| 99| const kmp_uint32 SHIFT_RIGHT = LSB; 1959| 99| kmp_uint32 retval = v; 1960| 99| retval <<= SHIFT_LEFT; 1961| 99| retval >>= (SHIFT_LEFT + SHIFT_RIGHT); 1962| 99| return retval; 1963| 99|} kmp_affinity.cpp:_ZL18__kmp_extract_bitsILj0ELj4EEjj: 1956| 100|static inline unsigned __kmp_extract_bits(kmp_uint32 v) { 1957| 100| const kmp_uint32 SHIFT_LEFT = sizeof(kmp_uint32) * 8 - 1 - MSB; 1958| 100| const kmp_uint32 SHIFT_RIGHT = LSB; 1959| 100| kmp_uint32 retval = v; 1960| 100| retval <<= SHIFT_LEFT; 1961| 100| retval >>= (SHIFT_LEFT + SHIFT_RIGHT); 1962| 100| return retval; 1963| 100|} kmp_affinity.cpp:_ZL18__kmp_extract_bitsILj0ELj15EEjj: 1956| 99|static inline unsigned __kmp_extract_bits(kmp_uint32 v) { 1957| 99| const kmp_uint32 SHIFT_LEFT = sizeof(kmp_uint32) * 8 - 1 - MSB; 1958| 99| const kmp_uint32 SHIFT_RIGHT = LSB; 1959| 99| kmp_uint32 retval = v; 1960| 99| retval <<= SHIFT_LEFT; 1961| 99| retval >>= (SHIFT_LEFT + SHIFT_RIGHT); 1962| 99| return retval; 1963| 99|} _ZN16kmp_cache_info_tC2Ev: 2008| 1| kmp_cache_info_t() : depth(0) { get_leaf4_levels(); } _ZN16kmp_cache_info_t16get_leaf4_levelsEv: 2032| 1| void get_leaf4_levels() { 2033| 1| unsigned level = 0; 2034| 1| while (depth < MAX_CACHE_LEVEL) { ------------------ | Branch (2034:12): [True: 1, False: 0] ------------------ 2035| 1| unsigned cache_type, max_threads_sharing; 2036| 1| unsigned cache_level, cache_mask_width; 2037| 1| kmp_cpuid buf2; 2038| 1| __kmp_x86_cpuid(4, level, &buf2); 2039| 1| cache_type = __kmp_extract_bits<0, 4>(buf2.eax); 2040| 1| if (!cache_type) ------------------ | Branch (2040:11): [True: 1, False: 0] ------------------ 2041| 1| break; 2042| | // Skip instruction caches 2043| 0| if (cache_type == 2) { ------------------ | Branch (2043:11): [True: 0, False: 0] ------------------ 2044| 0| level++; 2045| 0| continue; 2046| 0| } 2047| 0| max_threads_sharing = __kmp_extract_bits<14, 25>(buf2.eax) + 1; 2048| 0| cache_mask_width = __kmp_cpuid_mask_width(max_threads_sharing); 2049| 0| cache_level = __kmp_extract_bits<5, 7>(buf2.eax); 2050| 0| table[depth].level = cache_level; 2051| 0| table[depth].mask = ((-1) << cache_mask_width); 2052| 0| depth++; 2053| 0| level++; 2054| 0| } 2055| 1| } _ZNK16kmp_cache_info_t9get_depthEv: 2009| 1| size_t get_depth() const { return depth; } kmp_affinity.cpp:_ZL33__kmp_create_affinity_none_placesR14kmp_affinity_t: 4150| 1|static void __kmp_create_affinity_none_places(kmp_affinity_t &affinity) { 4151| 1| KMP_ASSERT(__kmp_affin_fullMask != NULL); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4152| 1| KMP_ASSERT(affinity.type == affinity_none); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4153| 1| affinity.num_masks = 1; 4154| 1| KMP_CPU_ALLOC_ARRAY(affinity.masks, affinity.num_masks); ------------------ | | 705| 1| (arr = __kmp_affinity_dispatch->allocate_mask_array(n)) ------------------ 4155| 1| kmp_affin_mask_t *dest = KMP_CPU_INDEX(affinity.masks, 0); ------------------ | | 703| 1|#define KMP_CPU_INDEX(arr, i) __kmp_affinity_dispatch->index_mask_array(arr, i) ------------------ 4156| 1| KMP_CPU_COPY(dest, __kmp_affin_fullMask); ------------------ | | 693| 1|#define KMP_CPU_COPY(dest, src) (dest)->copy(src) ------------------ 4157| 1| __kmp_affinity_get_topology_info(affinity); 4158| 1|} kmp_affinity.cpp:_ZL32__kmp_affinity_get_topology_infoR14kmp_affinity_t: 4107| 1|static void __kmp_affinity_get_topology_info(kmp_affinity_t &affinity) { 4108| 1| if (!KMP_AFFINITY_CAPABLE()) ------------------ | | 684| 1|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) ------------------ | Branch (4108:7): [True: 0, False: 1] ------------------ 4109| 0| return; 4110| 1| if (affinity.type != affinity_none) { ------------------ | Branch (4110:7): [True: 0, False: 1] ------------------ 4111| 0| KMP_ASSERT(affinity.num_os_id_masks); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4112| 0| KMP_ASSERT(affinity.os_id_masks); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4113| 0| } 4114| 1| KMP_ASSERT(affinity.num_masks); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4115| 1| KMP_ASSERT(affinity.masks); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4116| 1| KMP_ASSERT(__kmp_affin_fullMask); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4117| | 4118| 1| int max_cpu = __kmp_affin_fullMask->get_max_cpu(); 4119| 1| int num_hw_threads = __kmp_topology->get_num_hw_threads(); 4120| | 4121| | // Allocate thread topology information 4122| 1| if (!affinity.ids) { ------------------ | Branch (4122:7): [True: 1, False: 0] ------------------ 4123| 1| affinity.ids = (kmp_affinity_ids_t *)__kmp_allocate( ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 4124| 1| sizeof(kmp_affinity_ids_t) * affinity.num_masks); 4125| 1| } 4126| 1| if (!affinity.attrs) { ------------------ | Branch (4126:7): [True: 1, False: 0] ------------------ 4127| 1| affinity.attrs = (kmp_affinity_attrs_t *)__kmp_allocate( ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 4128| 1| sizeof(kmp_affinity_attrs_t) * affinity.num_masks); 4129| 1| } 4130| 1| if (!__kmp_osid_to_hwthread_map) { ------------------ | Branch (4130:7): [True: 1, False: 0] ------------------ 4131| | // Want the +1 because max_cpu should be valid index into map 4132| 1| __kmp_osid_to_hwthread_map = 4133| 1| (int *)__kmp_allocate(sizeof(int) * (max_cpu + 1)); ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 4134| 1| } 4135| | 4136| | // Create the OS proc to hardware thread map 4137| 33| for (int hw_thread = 0; hw_thread < num_hw_threads; ++hw_thread) ------------------ | Branch (4137:27): [True: 32, False: 1] ------------------ 4138| 32| __kmp_osid_to_hwthread_map[__kmp_topology->at(hw_thread).os_id] = hw_thread; 4139| | 4140| 2| for (unsigned i = 0; i < affinity.num_masks; ++i) { ------------------ | Branch (4140:24): [True: 1, False: 1] ------------------ 4141| 1| kmp_affinity_ids_t &ids = affinity.ids[i]; 4142| 1| kmp_affinity_attrs_t &attrs = affinity.attrs[i]; 4143| 1| kmp_affin_mask_t *mask = KMP_CPU_INDEX(affinity.masks, i); ------------------ | | 703| 1|#define KMP_CPU_INDEX(arr, i) __kmp_affinity_dispatch->index_mask_array(arr, i) ------------------ 4144| 1| __kmp_affinity_get_mask_topology_info(mask, ids, attrs); 4145| 1| } 4146| 1|} kmp_affinity.cpp:_ZL37__kmp_affinity_get_mask_topology_infoPKN11KMPAffinity4MaskER18kmp_affinity_ids_tR20kmp_affinity_attrs_t: 4050| 1| kmp_affinity_attrs_t &attrs) { 4051| 1| if (!KMP_AFFINITY_CAPABLE()) ------------------ | | 684| 1|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) ------------------ | Branch (4051:7): [True: 0, False: 1] ------------------ 4052| 0| return; 4053| | 4054| | // Initiailze ids and attrs thread data 4055| 13| for (int i = 0; i < KMP_HW_LAST; ++i) ------------------ | Branch (4055:19): [True: 12, False: 1] ------------------ 4056| 12| ids[i] = kmp_hw_thread_t::UNKNOWN_ID; 4057| 1| attrs = KMP_AFFINITY_ATTRS_UNKNOWN; ------------------ | | 891| 1| { KMP_HW_CORE_TYPE_UNKNOWN, kmp_hw_attr_t::UNKNOWN_CORE_EFF, 0, 0 } ------------------ 4058| | 4059| | // Iterate through each os id within the mask and determine 4060| | // the topology id and attribute information 4061| 1| int cpu; 4062| 1| int depth = __kmp_topology->get_depth(); 4063| 32| KMP_CPU_SET_ITERATE(cpu, mask) { ------------------ | | 688| 33| for (i = (mask)->begin(); (int)i != (mask)->end(); i = (mask)->next(i)) | | ------------------ | | | Branch (688:29): [True: 32, False: 1] | | ------------------ ------------------ 4064| 32| int osid_idx = __kmp_osid_to_hwthread_map[cpu]; 4065| 32| const kmp_hw_thread_t &hw_thread = __kmp_topology->at(osid_idx); 4066| 97| for (int level = 0; level < depth; ++level) { ------------------ | Branch (4066:25): [True: 65, False: 32] ------------------ 4067| 65| kmp_hw_t type = __kmp_topology->get_type(level); 4068| 65| int id = hw_thread.sub_ids[level]; 4069| 65| if (ids[type] == kmp_hw_thread_t::UNKNOWN_ID || ids[type] == id) { ------------------ | Branch (4069:11): [True: 3, False: 62] | Branch (4069:55): [True: 31, False: 31] ------------------ 4070| 34| ids[type] = id; 4071| 34| } else { 4072| | // This mask spans across multiple topology units, set it as such 4073| | // and mark every level below as such as well. 4074| 31| ids[type] = kmp_hw_thread_t::MULTIPLE_ID; 4075| 93| for (; level < depth; ++level) { ------------------ | Branch (4075:16): [True: 62, False: 31] ------------------ 4076| 62| kmp_hw_t type = __kmp_topology->get_type(level); 4077| 62| ids[type] = kmp_hw_thread_t::MULTIPLE_ID; 4078| 62| } 4079| 31| } 4080| 65| } 4081| 32| if (!attrs.valid) { ------------------ | Branch (4081:9): [True: 1, False: 31] ------------------ 4082| 1| attrs.core_type = hw_thread.attrs.get_core_type(); 4083| 1| attrs.core_eff = hw_thread.attrs.get_core_eff(); 4084| 1| attrs.valid = 1; 4085| 31| } else { 4086| | // This mask spans across multiple attributes, set it as such 4087| 31| if (attrs.core_type != hw_thread.attrs.get_core_type()) ------------------ | Branch (4087:11): [True: 0, False: 31] ------------------ 4088| 0| attrs.core_type = KMP_HW_CORE_TYPE_UNKNOWN; 4089| 31| if (attrs.core_eff != hw_thread.attrs.get_core_eff()) ------------------ | Branch (4089:11): [True: 0, False: 31] ------------------ 4090| 0| attrs.core_eff = kmp_hw_attr_t::UNKNOWN_CORE_EFF; 4091| 31| } 4092| 32| } 4093| 1|} _ZNK17KMPNativeAffinity4Mask18get_num_mask_typesEv: 297| 458| mask_size_type get_num_mask_types() const { 298| 458| return __kmp_affin_mask_size / sizeof(mask_t); 299| 458| } _ZN17KMPNativeAffinity4MaskC2Ev: 303| 69| Mask() { mask = (mask_t *)__kmp_allocate(__kmp_affin_mask_size); } ------------------ | | 3617| 69|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ _ZN17KMPNativeAffinity4MaskD2Ev: 304| 68| ~Mask() { 305| 68| if (mask) ------------------ | Branch (305:11): [True: 68, False: 0] ------------------ 306| 68| __kmp_free(mask); ------------------ | | 3619| 68|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 307| 68| } _ZN17KMPNativeAffinity4Mask3setEi: 308| 32| void set(int i) override { 309| 32| mask[i / BITS_PER_MASK_T] |= (ONE << (i % BITS_PER_MASK_T)); 310| 32| } _ZNK17KMPNativeAffinity4Mask6is_setEi: 311| 320| bool is_set(int i) const override { 312| 320| return (mask[i / BITS_PER_MASK_T] & (ONE << (i % BITS_PER_MASK_T))); 313| 320| } _ZN17KMPNativeAffinity4Mask4zeroEv: 317| 32| void zero() override { 318| 32| mask_size_type e = get_num_mask_types(); 319| 64| for (mask_size_type i = 0; i < e; ++i) ------------------ | Branch (319:34): [True: 32, False: 32] ------------------ 320| 32| mask[i] = (mask_t)0; 321| 32| } _ZN17KMPNativeAffinity4Mask4copyEPKN11KMPAffinity4MaskE: 322| 34| void copy(const KMPAffinity::Mask *src) override { 323| 34| const Mask *convert = static_cast(src); 324| 34| mask_size_type e = get_num_mask_types(); 325| 68| for (mask_size_type i = 0; i < e; ++i) ------------------ | Branch (325:34): [True: 34, False: 34] ------------------ 326| 34| mask[i] = convert->mask[i]; 327| 34| } _ZNK17KMPNativeAffinity4Mask5beginEv: 345| 4| int begin() const override { 346| 4| int retval = 0; 347| 4| while (retval < end() && !is_set(retval)) ------------------ | Branch (347:14): [True: 4, False: 0] | Branch (347:32): [True: 0, False: 4] ------------------ 348| 0| ++retval; 349| 4| return retval; 350| 4| } _ZNK17KMPNativeAffinity4Mask3endEv: 351| 392| int end() const override { 352| 392| int e; 353| 392| __kmp_type_convert(get_num_mask_types() * BITS_PER_MASK_T, &e); 354| 392| return e; 355| 392| } _ZNK17KMPNativeAffinity4Mask4nextEi: 356| 128| int next(int previous) const override { 357| 128| int retval = previous + 1; 358| 256| while (retval < end() && !is_set(retval)) ------------------ | Branch (358:14): [True: 252, False: 4] | Branch (358:32): [True: 128, False: 124] ------------------ 359| 128| ++retval; 360| 128| return retval; 361| 128| } _ZN17KMPNativeAffinity4Mask19get_system_affinityEb: 362| 3| int get_system_affinity(bool abort_on_error) override { 363| 3| KMP_ASSERT2(KMP_AFFINITY_CAPABLE(), ------------------ | | 53| 3|#define KMP_ASSERT2(cond, msg) KMP_ASSERT(cond) | | ------------------ | | | | 50| 3| if (!(cond)) { \ | | | | ------------------ | | | | | Branch (50:7): [True: 0, False: 3] | | | | ------------------ | | | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | | | 52| 0| } | | ------------------ ------------------ 364| 3| "Illegal get affinity operation when not capable"); 365| 3|#if KMP_OS_LINUX 366| 3| long retval = 367| 3| syscall(__NR_sched_getaffinity, 0, __kmp_affin_mask_size, mask); 368| |#elif KMP_OS_FREEBSD 369| | int r = pthread_getaffinity_np(pthread_self(), __kmp_affin_mask_size, 370| | reinterpret_cast(mask)); 371| | int retval = (r == 0 ? 0 : -1); 372| |#endif 373| 3| if (retval >= 0) { ------------------ | Branch (373:11): [True: 3, False: 0] ------------------ 374| 3| return 0; 375| 3| } 376| 0| int error = errno; 377| 0| if (abort_on_error) { ------------------ | Branch (377:11): [True: 0, False: 0] ------------------ 378| 0| __kmp_fatal(KMP_MSG(FunctionError, "pthread_getaffinity_np()"), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ 379| 0| KMP_ERR(error), __kmp_msg_null); ------------------ | | 125| 0|#define KMP_ERR KMP_SYSERRCODE | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ ------------------ 380| 0| } 381| 0| return error; 382| 3| } _ZNK17KMPNativeAffinity4Mask19set_system_affinityEb: 383| 68| int set_system_affinity(bool abort_on_error) const override { 384| 68| KMP_ASSERT2(KMP_AFFINITY_CAPABLE(), ------------------ | | 53| 68|#define KMP_ASSERT2(cond, msg) KMP_ASSERT(cond) | | ------------------ | | | | 50| 68| if (!(cond)) { \ | | | | ------------------ | | | | | Branch (50:7): [True: 0, False: 68] | | | | ------------------ | | | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | | | 52| 0| } | | ------------------ ------------------ 385| 68| "Illegal set affinity operation when not capable"); 386| 68|#if KMP_OS_LINUX 387| 68| long retval = 388| 68| syscall(__NR_sched_setaffinity, 0, __kmp_affin_mask_size, mask); 389| |#elif KMP_OS_FREEBSD 390| | int r = pthread_setaffinity_np(pthread_self(), __kmp_affin_mask_size, 391| | reinterpret_cast(mask)); 392| | int retval = (r == 0 ? 0 : -1); 393| |#endif 394| 68| if (retval >= 0) { ------------------ | Branch (394:11): [True: 68, False: 0] ------------------ 395| 68| return 0; 396| 68| } 397| 0| int error = errno; 398| 0| if (abort_on_error) { ------------------ | Branch (398:11): [True: 0, False: 0] ------------------ 399| 0| __kmp_fatal(KMP_MSG(FunctionError, "pthread_setaffinity_np()"), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ 400| 0| KMP_ERR(error), __kmp_msg_null); ------------------ | | 125| 0|#define KMP_ERR KMP_SYSERRCODE | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ ------------------ 401| 0| } 402| 0| return error; 403| 68| } _ZN17KMPNativeAffinity17determine_capableEPKc: 405| 1| void determine_capable(const char *env_var) override { 406| 1| __kmp_affinity_determine_capable(env_var); 407| 1| } _ZN17KMPNativeAffinity11bind_threadEi: 408| 32| void bind_thread(int which) override { __kmp_affinity_bind_thread(which); } _ZN17KMPNativeAffinity13allocate_maskEv: 409| 68| KMPAffinity::Mask *allocate_mask() override { 410| 68| KMPNativeAffinity::Mask *retval = new Mask(); 411| 68| return retval; 412| 68| } _ZN17KMPNativeAffinity15deallocate_maskEPN11KMPAffinity4MaskE: 413| 67| void deallocate_mask(KMPAffinity::Mask *m) override { 414| 67| KMPNativeAffinity::Mask *native_mask = 415| 67| static_cast(m); 416| 67| delete native_mask; 417| 67| } _ZN17KMPNativeAffinity19allocate_mask_arrayEi: 418| 1| KMPAffinity::Mask *allocate_mask_array(int num) override { 419| 1| return new Mask[num]; 420| 1| } _ZN17KMPNativeAffinity21deallocate_mask_arrayEPN11KMPAffinity4MaskE: 421| 1| void deallocate_mask_array(KMPAffinity::Mask *array) override { 422| 1| Mask *linux_array = static_cast(array); 423| 1| delete[] linux_array; 424| 1| } _ZN17KMPNativeAffinity16index_mask_arrayEPN11KMPAffinity4MaskEi: 426| 2| int index) override { 427| 2| Mask *linux_array = static_cast(array); 428| 2| return &(linux_array[index]); 429| 2| } _ZNK13kmp_hw_attr_t13get_core_typeEv: 652| 32| kmp_hw_core_type_t get_core_type() const { 653| 32| return (kmp_hw_core_type_t)core_type; 654| 32| } _ZNK13kmp_hw_attr_t12get_core_effEv: 655| 32| int get_core_eff() const { return core_eff; } _ZN13kmp_hw_attr_t5clearEv: 661| 32| void clear() { 662| 32| core_type = KMP_HW_CORE_TYPE_UNKNOWN; 663| 32| core_eff = UNKNOWN_CORE_EFF; 664| 32| valid = 0; 665| 32| } _ZN15kmp_hw_thread_t5clearEv: 706| 32| void clear() { 707| 416| for (int i = 0; i < (int)KMP_HW_LAST; ++i) ------------------ | Branch (707:21): [True: 384, False: 32] ------------------ 708| 384| ids[i] = UNKNOWN_ID; 709| 32| leader = false; 710| 32| attrs.clear(); 711| 32| } _ZN14kmp_topology_t2atEi: 811| 96| kmp_hw_thread_t &at(int index) { 812| 96| KMP_DEBUG_ASSERT(index >= 0 && index < num_hw_threads); 813| 96| return hw_threads[index]; 814| 96| } _ZNK14kmp_topology_t18get_num_hw_threadsEv: 819| 3| int get_num_hw_threads() const { return num_hw_threads; } _ZN14kmp_topology_t8sort_idsEv: 820| 1| void sort_ids() { 821| 1| qsort(hw_threads, num_hw_threads, sizeof(kmp_hw_thread_t), 822| 1| kmp_hw_thread_t::compare_ids); 823| 1| } _ZNK14kmp_topology_t19get_equivalent_typeE8kmp_hw_t: 843| 6| kmp_hw_t get_equivalent_type(kmp_hw_t type) const { return equivalent[type]; } _ZN14kmp_topology_t19set_equivalent_typeE8kmp_hw_tS0_: 845| 1| void set_equivalent_type(kmp_hw_t type1, kmp_hw_t type2) { 846| 1| KMP_DEBUG_ASSERT_VALID_HW_TYPE(type1); 847| 1| KMP_DEBUG_ASSERT_VALID_HW_TYPE(type2); 848| 1| kmp_hw_t real_type2 = equivalent[type2]; 849| 1| if (real_type2 == KMP_HW_UNKNOWN) ------------------ | Branch (849:9): [True: 0, False: 1] ------------------ 850| 0| real_type2 = type2; 851| 1| equivalent[type1] = real_type2; 852| | // This loop is required since any of the types may have been set to 853| | // be equivalent to type1. They all must be checked and reset to type2. 854| 12| KMP_FOREACH_HW_TYPE(type) { ------------------ | | 640| 13| for (kmp_hw_t type = (kmp_hw_t)0; type < KMP_HW_LAST; \ | | ------------------ | | | Branch (640:37): [True: 12, False: 1] | | ------------------ | | 641| 12| type = (kmp_hw_t)((int)type + 1)) ------------------ 855| 12| if (equivalent[type] == type1) { ------------------ | Branch (855:11): [True: 0, False: 12] ------------------ 856| 0| equivalent[type] = real_type2; 857| 0| } 858| 12| } 859| 1| } _ZNK14kmp_topology_t15calculate_ratioEii: 862| 2| int calculate_ratio(int level1, int level2) const { 863| 2| KMP_DEBUG_ASSERT(level1 >= 0 && level1 < depth); 864| 2| KMP_DEBUG_ASSERT(level2 >= 0 && level2 < depth); 865| 2| int r = 1; 866| 4| for (int level = level1; level > level2; --level) ------------------ | Branch (866:30): [True: 2, False: 2] ------------------ 867| 2| r *= ratio[level]; 868| 2| return r; 869| 2| } _ZNK14kmp_topology_t9get_ratioEi: 870| 3| int get_ratio(int level) const { 871| 3| KMP_DEBUG_ASSERT(level >= 0 && level < depth); 872| 3| return ratio[level]; 873| 3| } _ZNK14kmp_topology_t9get_depthEv: 874| 98| int get_depth() const { return depth; }; _ZNK14kmp_topology_t8get_typeEi: 875| 127| kmp_hw_t get_type(int level) const { 876| 127| KMP_DEBUG_ASSERT(level >= 0 && level < depth); 877| 127| return types[level]; 878| 127| } _ZNK14kmp_topology_t9get_levelE8kmp_hw_t: 879| 4| int get_level(kmp_hw_t type) const { 880| 4| KMP_DEBUG_ASSERT_VALID_HW_TYPE(type); 881| 4| int eq_type = equivalent[type]; 882| 4| if (eq_type == KMP_HW_UNKNOWN) ------------------ | Branch (882:9): [True: 0, False: 4] ------------------ 883| 0| return -1; 884| 8| for (int i = 0; i < depth; ++i) ------------------ | Branch (884:21): [True: 8, False: 0] ------------------ 885| 8| if (types[i] == eq_type) ------------------ | Branch (885:11): [True: 4, False: 4] ------------------ 886| 4| return i; 887| 0| return -1; 888| 4| } _ZNK14kmp_topology_t9get_countEi: 889| 2| int get_count(int level) const { 890| 2| KMP_DEBUG_ASSERT(level >= 0 && level < depth); 891| 2| return count[level]; 892| 2| } _ZN14hierarchy_info12deriveLevelsEv: 1090| 1| void deriveLevels() { 1091| 1| int hier_depth = __kmp_topology->get_depth(); 1092| 4| for (int i = hier_depth - 1, level = 0; i >= 0; --i, ++level) { ------------------ | Branch (1092:45): [True: 3, False: 1] ------------------ 1093| 3| numPerLevel[level] = __kmp_topology->get_ratio(i); 1094| 3| } 1095| 1| } _ZN14hierarchy_infoC2Ev: 1098| 2| : maxLevels(7), depth(1), uninitialized(not_initialized), resizing(0) {} _ZN14hierarchy_info4finiEv: 1100| 1| void fini() { 1101| 1| if (!uninitialized && numPerLevel) { ------------------ | Branch (1101:9): [True: 1, False: 0] | Branch (1101:27): [True: 1, False: 0] ------------------ 1102| 1| __kmp_free(numPerLevel); ------------------ | | 3619| 1|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 1103| | numPerLevel = NULL; 1104| 1| uninitialized = not_initialized; 1105| 1| } 1106| 1| } _ZN14hierarchy_info4initEi: 1108| 1| void init(int num_addrs) { 1109| 1| kmp_int8 bool_result = KMP_COMPARE_AND_STORE_ACQ8( ------------------ | | 795| 1| __sync_bool_compare_and_swap((volatile kmp_uint8 *)(p), (kmp_uint8)(cv), \ | | 796| 1| (kmp_uint8)(sv)) ------------------ 1110| 1| &uninitialized, not_initialized, initializing); 1111| 1| if (bool_result == 0) { // Wait for initialization ------------------ | Branch (1111:9): [True: 0, False: 1] ------------------ 1112| 0| while (TCR_1(uninitialized) != initialized) ------------------ | | 1125| 0|#define TCR_1(a) (a) ------------------ | Branch (1112:14): [True: 0, False: 0] ------------------ 1113| 0| KMP_CPU_PAUSE(); ------------------ | | 1498| 0|#define KMP_CPU_PAUSE() __kmp_x86_pause() ------------------ 1114| 0| return; 1115| 0| } 1116| 1| KMP_DEBUG_ASSERT(bool_result == 1); 1117| | 1118| | /* Added explicit initialization of the data fields here to prevent usage of 1119| | dirty value observed when static library is re-initialized multiple times 1120| | (e.g. when non-OpenMP thread repeatedly launches/joins thread that uses 1121| | OpenMP). */ 1122| 1| depth = 1; 1123| 1| resizing = 0; 1124| 1| maxLevels = 7; 1125| 1| numPerLevel = 1126| 1| (kmp_uint32 *)__kmp_allocate(maxLevels * 2 * sizeof(kmp_uint32)); ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 1127| 1| skipPerLevel = &(numPerLevel[maxLevels]); 1128| 8| for (kmp_uint32 i = 0; i < maxLevels; ------------------ | Branch (1128:28): [True: 7, False: 1] ------------------ 1129| 7| ++i) { // init numPerLevel[*] to 1 item per level 1130| 7| numPerLevel[i] = 1; 1131| 7| skipPerLevel[i] = 1; 1132| 7| } 1133| | 1134| | // Sort table by physical ID 1135| 1| if (__kmp_topology && __kmp_topology->get_depth() > 0) { ------------------ | Branch (1135:9): [True: 1, False: 0] | Branch (1135:27): [True: 1, False: 0] ------------------ 1136| 1| deriveLevels(); 1137| 1| } else { 1138| 0| numPerLevel[0] = maxLeaves; 1139| 0| numPerLevel[1] = num_addrs / maxLeaves; 1140| 0| if (num_addrs % maxLeaves) ------------------ | Branch (1140:11): [True: 0, False: 0] ------------------ 1141| 0| numPerLevel[1]++; 1142| 0| } 1143| | 1144| 1| base_num_threads = num_addrs; 1145| 8| for (int i = maxLevels - 1; i >= 0; ------------------ | Branch (1145:33): [True: 7, False: 1] ------------------ 1146| 7| --i) // count non-empty levels to get depth 1147| 7| if (numPerLevel[i] != 1 || depth > 1) // only count one top-level '1' ------------------ | Branch (1147:11): [True: 2, False: 5] | Branch (1147:34): [True: 0, False: 5] ------------------ 1148| 2| depth++; 1149| | 1150| 1| kmp_uint32 branch = minBranch; 1151| 1| if (numPerLevel[0] == 1) ------------------ | Branch (1151:9): [True: 0, False: 1] ------------------ 1152| 0| branch = num_addrs / maxLeaves; 1153| 1| if (branch < minBranch) ------------------ | Branch (1153:9): [True: 0, False: 1] ------------------ 1154| 0| branch = minBranch; 1155| 4| for (kmp_uint32 d = 0; d < depth - 1; ++d) { // optimize hierarchy width ------------------ | Branch (1155:28): [True: 3, False: 1] ------------------ 1156| 5| while (numPerLevel[d] > branch || ------------------ | Branch (1156:14): [True: 2, False: 3] ------------------ 1157| 3| (d == 0 && numPerLevel[d] > maxLeaves)) { // max 4 on level 0! ------------------ | Branch (1157:15): [True: 1, False: 2] | Branch (1157:25): [True: 0, False: 1] ------------------ 1158| 2| if (numPerLevel[d] & 1) ------------------ | Branch (1158:13): [True: 0, False: 2] ------------------ 1159| 0| numPerLevel[d]++; 1160| 2| numPerLevel[d] = numPerLevel[d] >> 1; 1161| 2| if (numPerLevel[d + 1] == 1) ------------------ | Branch (1161:13): [True: 1, False: 1] ------------------ 1162| 1| depth++; 1163| 2| numPerLevel[d + 1] = numPerLevel[d + 1] << 1; 1164| 2| } 1165| 3| if (numPerLevel[0] == 1) { ------------------ | Branch (1165:11): [True: 0, False: 3] ------------------ 1166| 0| branch = branch >> 1; 1167| 0| if (branch < 4) ------------------ | Branch (1167:13): [True: 0, False: 0] ------------------ 1168| 0| branch = minBranch; 1169| 0| } 1170| 3| } 1171| | 1172| 4| for (kmp_uint32 i = 1; i < depth; ++i) ------------------ | Branch (1172:28): [True: 3, False: 1] ------------------ 1173| 3| skipPerLevel[i] = numPerLevel[i - 1] * skipPerLevel[i - 1]; 1174| | // Fill in hierarchy in the case of oversubscription 1175| 4| for (kmp_uint32 i = depth; i < maxLevels; ++i) ------------------ | Branch (1175:32): [True: 3, False: 1] ------------------ 1176| 3| skipPerLevel[i] = 2 * skipPerLevel[i - 1]; 1177| | 1178| 1| uninitialized = initialized; // One writer 1179| 1| } __kmp_initialize_bget: 992| 32|void __kmp_initialize_bget(kmp_info_t *th) { 993| 32| KMP_DEBUG_ASSERT(SizeQuant >= sizeof(void *) && (th != 0)); 994| | 995| 32| set_thr_data(th); 996| | 997| 32| bectl(th, (bget_compact_t)0, (bget_acquire_t)malloc, (bget_release_t)free, 998| 32| (bufsize)__kmp_malloc_pool_incr); 999| 32|} __kmp_finalize_bget: 1001| 32|void __kmp_finalize_bget(kmp_info_t *th) { 1002| 32| thr_data_t *thr; 1003| 32| bfhead_t *b; 1004| | 1005| 32| KMP_DEBUG_ASSERT(th != 0); 1006| | 1007| 32|#if BufStats 1008| 32| thr = (thr_data_t *)th->th.th_local.bget_data; 1009| 32| KMP_DEBUG_ASSERT(thr != NULL); 1010| 32| b = thr->last_pool; 1011| | 1012| | /* If a block-release function is defined, and this free buffer constitutes 1013| | the entire block, release it. Note that pool_len is defined in such a way 1014| | that the test will fail unless all pool blocks are the same size. */ 1015| | 1016| | // Deallocate the last pool if one exists because we no longer do it in brel() 1017| 32| if (thr->relfcn != 0 && b != 0 && thr->numpblk != 0 && ------------------ | Branch (1017:7): [True: 32, False: 0] | Branch (1017:27): [True: 0, False: 32] | Branch (1017:37): [True: 0, False: 0] ------------------ 1018| 0| b->bh.bb.bsize == (bufsize)(thr->pool_len - sizeof(bhead_t))) { ------------------ | Branch (1018:7): [True: 0, False: 0] ------------------ 1019| 0| KMP_DEBUG_ASSERT(b->bh.bb.prevfree == 0); 1020| 0| KMP_DEBUG_ASSERT(BH((char *)b + b->bh.bb.bsize)->bb.bsize == ESent); 1021| 0| KMP_DEBUG_ASSERT(BH((char *)b + b->bh.bb.bsize)->bb.prevfree == 1022| 0| b->bh.bb.bsize); 1023| | 1024| | /* Unlink the buffer from the free list */ 1025| 0| __kmp_bget_remove_from_freelist(b); 1026| | 1027| 0| KE_TRACE(10, ("%%%%%% FREE( %p )\n", (void *)b)); 1028| | 1029| 0| (*thr->relfcn)(b); 1030| 0| thr->numprel++; /* Nr of expansion block releases */ 1031| 0| thr->numpblk--; /* Total number of blocks */ 1032| 0| KMP_DEBUG_ASSERT(thr->numpblk == thr->numpget - thr->numprel); 1033| 0| } 1034| 32|#endif /* BufStats */ 1035| | 1036| | /* Deallocate bget_data */ 1037| 32| if (th->th.th_local.bget_data != NULL) { ------------------ | Branch (1037:7): [True: 32, False: 0] ------------------ 1038| 32| __kmp_free(th->th.th_local.bget_data); ------------------ | | 3619| 32|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 1039| | th->th.th_local.bget_data = NULL; 1040| 32| } 1041| 32|} __kmp_init_memkind: 1272| 1|void __kmp_init_memkind() { 1273| |// as of 2018-07-31 memkind does not support Windows*, exclude it for now 1274| |#if KMP_OS_UNIX && KMP_DYNAMIC_LIB && !KMP_OS_DARWIN 1275| | // use of statically linked memkind is problematic, as it depends on libnuma 1276| | kmp_mk_lib_name = "libmemkind.so"; 1277| | h_memkind = dlopen(kmp_mk_lib_name, RTLD_LAZY); 1278| | if (h_memkind) { 1279| | kmp_mk_check = (int (*)(void *))dlsym(h_memkind, "memkind_check_available"); 1280| | kmp_mk_alloc = 1281| | (void *(*)(void *, size_t))dlsym(h_memkind, "memkind_malloc"); 1282| | kmp_mk_free = (void (*)(void *, void *))dlsym(h_memkind, "memkind_free"); 1283| | mk_default = (void **)dlsym(h_memkind, "MEMKIND_DEFAULT"); 1284| | if (kmp_mk_check && kmp_mk_alloc && kmp_mk_free && mk_default && 1285| | !kmp_mk_check(*mk_default)) { 1286| | __kmp_memkind_available = 1; 1287| | mk_interleave = (void **)dlsym(h_memkind, "MEMKIND_INTERLEAVE"); 1288| | chk_kind(&mk_interleave); 1289| | mk_hbw = (void **)dlsym(h_memkind, "MEMKIND_HBW"); 1290| | chk_kind(&mk_hbw); 1291| | mk_hbw_interleave = (void **)dlsym(h_memkind, "MEMKIND_HBW_INTERLEAVE"); 1292| | chk_kind(&mk_hbw_interleave); 1293| | mk_hbw_preferred = (void **)dlsym(h_memkind, "MEMKIND_HBW_PREFERRED"); 1294| | chk_kind(&mk_hbw_preferred); 1295| | mk_hugetlb = (void **)dlsym(h_memkind, "MEMKIND_HUGETLB"); 1296| | chk_kind(&mk_hugetlb); 1297| | mk_hbw_hugetlb = (void **)dlsym(h_memkind, "MEMKIND_HBW_HUGETLB"); 1298| | chk_kind(&mk_hbw_hugetlb); 1299| | mk_hbw_preferred_hugetlb = 1300| | (void **)dlsym(h_memkind, "MEMKIND_HBW_PREFERRED_HUGETLB"); 1301| | chk_kind(&mk_hbw_preferred_hugetlb); 1302| | mk_dax_kmem = (void **)dlsym(h_memkind, "MEMKIND_DAX_KMEM"); 1303| | chk_kind(&mk_dax_kmem); 1304| | mk_dax_kmem_all = (void **)dlsym(h_memkind, "MEMKIND_DAX_KMEM_ALL"); 1305| | chk_kind(&mk_dax_kmem_all); 1306| | mk_dax_kmem_preferred = 1307| | (void **)dlsym(h_memkind, "MEMKIND_DAX_KMEM_PREFERRED"); 1308| | chk_kind(&mk_dax_kmem_preferred); 1309| | KE_TRACE(25, ("__kmp_init_memkind: memkind library initialized\n")); 1310| | return; // success 1311| | } 1312| | dlclose(h_memkind); // failure 1313| | } 1314| |#else // !(KMP_OS_UNIX && KMP_DYNAMIC_LIB) 1315| 1| kmp_mk_lib_name = ""; 1316| 1|#endif // !(KMP_OS_UNIX && KMP_DYNAMIC_LIB) 1317| 1| h_memkind = NULL; 1318| 1| kmp_mk_check = NULL; 1319| 1| kmp_mk_alloc = NULL; 1320| 1| kmp_mk_free = NULL; 1321| 1| mk_default = NULL; 1322| 1| mk_interleave = NULL; 1323| 1| mk_hbw = NULL; 1324| 1| mk_hbw_interleave = NULL; 1325| 1| mk_hbw_preferred = NULL; 1326| 1| mk_hugetlb = NULL; 1327| 1| mk_hbw_hugetlb = NULL; 1328| 1| mk_hbw_preferred_hugetlb = NULL; 1329| 1| mk_dax_kmem = NULL; 1330| 1| mk_dax_kmem_all = NULL; 1331| | mk_dax_kmem_preferred = NULL; 1332| 1|} __kmp_fini_memkind: 1334| 1|void __kmp_fini_memkind() { 1335| |#if KMP_OS_UNIX && KMP_DYNAMIC_LIB 1336| | if (__kmp_memkind_available) 1337| | KE_TRACE(25, ("__kmp_fini_memkind: finalize memkind library\n")); 1338| | if (h_memkind) { 1339| | dlclose(h_memkind); 1340| | h_memkind = NULL; 1341| | } 1342| | kmp_mk_check = NULL; 1343| | kmp_mk_alloc = NULL; 1344| | kmp_mk_free = NULL; 1345| | mk_default = NULL; 1346| | mk_interleave = NULL; 1347| | mk_hbw = NULL; 1348| | mk_hbw_interleave = NULL; 1349| | mk_hbw_preferred = NULL; 1350| | mk_hugetlb = NULL; 1351| | mk_hbw_hugetlb = NULL; 1352| | mk_hbw_preferred_hugetlb = NULL; 1353| | mk_dax_kmem = NULL; 1354| | mk_dax_kmem_all = NULL; 1355| | mk_dax_kmem_preferred = NULL; 1356| |#endif 1357| 1|} __kmp_init_target_mem: 1359| 1|void __kmp_init_target_mem() { 1360| 1| *(void **)(&kmp_target_alloc_host) = KMP_DLSYM("llvm_omp_target_alloc_host"); ------------------ | | 1289| 1|#define KMP_DLSYM(name) dlsym(RTLD_DEFAULT, name) ------------------ 1361| 1| *(void **)(&kmp_target_alloc_shared) = 1362| 1| KMP_DLSYM("llvm_omp_target_alloc_shared"); ------------------ | | 1289| 1|#define KMP_DLSYM(name) dlsym(RTLD_DEFAULT, name) ------------------ 1363| 1| *(void **)(&kmp_target_alloc_device) = 1364| 1| KMP_DLSYM("llvm_omp_target_alloc_device"); ------------------ | | 1289| 1|#define KMP_DLSYM(name) dlsym(RTLD_DEFAULT, name) ------------------ 1365| 1| *(void **)(&kmp_target_free_host) = KMP_DLSYM("llvm_omp_target_free_host"); ------------------ | | 1289| 1|#define KMP_DLSYM(name) dlsym(RTLD_DEFAULT, name) ------------------ 1366| 1| *(void **)(&kmp_target_free_shared) = 1367| 1| KMP_DLSYM("llvm_omp_target_free_shared"); ------------------ | | 1289| 1|#define KMP_DLSYM(name) dlsym(RTLD_DEFAULT, name) ------------------ 1368| 1| *(void **)(&kmp_target_free_device) = 1369| 1| KMP_DLSYM("llvm_omp_target_free_device"); ------------------ | | 1289| 1|#define KMP_DLSYM(name) dlsym(RTLD_DEFAULT, name) ------------------ 1370| 1| __kmp_target_mem_available = 1371| 1| kmp_target_alloc_host && kmp_target_alloc_shared && ------------------ | Branch (1371:7): [True: 0, False: 1] | Branch (1371:32): [True: 0, False: 0] ------------------ 1372| 0| kmp_target_alloc_device && kmp_target_free_host && ------------------ | Branch (1372:7): [True: 0, False: 0] | Branch (1372:34): [True: 0, False: 0] ------------------ 1373| 0| kmp_target_free_shared && kmp_target_free_device; ------------------ | Branch (1373:7): [True: 0, False: 0] | Branch (1373:33): [True: 0, False: 0] ------------------ 1374| | // lock/pin and unlock/unpin target calls 1375| 1| *(void **)(&kmp_target_lock_mem) = KMP_DLSYM("llvm_omp_target_lock_mem"); ------------------ | | 1289| 1|#define KMP_DLSYM(name) dlsym(RTLD_DEFAULT, name) ------------------ 1376| | *(void **)(&kmp_target_unlock_mem) = KMP_DLSYM("llvm_omp_target_unlock_mem"); ------------------ | | 1289| 1|#define KMP_DLSYM(name) dlsym(RTLD_DEFAULT, name) ------------------ 1377| 1|} ___kmp_allocate: 1997| 3.49k|void *___kmp_allocate(size_t size KMP_SRC_LOC_DECL) { 1998| 3.49k| void *ptr; 1999| 3.49k| KE_TRACE(25, ("-> __kmp_allocate( %d ) called from %s:%d\n", 2000| 3.49k| (int)size KMP_SRC_LOC_PARM)); 2001| 3.49k| ptr = ___kmp_allocate_align(size, __kmp_align_alloc KMP_SRC_LOC_PARM); 2002| 3.49k| KE_TRACE(25, ("<- __kmp_allocate() returns %p\n", ptr)); 2003| 3.49k| return ptr; 2004| 3.49k|} // func ___kmp_allocate ___kmp_free: 2023| 3.49k|void ___kmp_free(void *ptr KMP_SRC_LOC_DECL) { 2024| 3.49k| kmp_mem_descr_t descr; 2025| |#if KMP_DEBUG 2026| | kmp_uintptr_t addr_allocated; // Address returned by malloc(). 2027| | kmp_uintptr_t addr_aligned; // Aligned address passed by caller. 2028| |#endif 2029| 3.49k| KE_TRACE(25, 2030| 3.49k| ("-> __kmp_free( %p ) called from %s:%d\n", ptr KMP_SRC_LOC_PARM)); 2031| 3.49k| KMP_ASSERT(ptr != NULL); ------------------ | | 50| 3.49k| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 3.49k] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 2032| | 2033| 3.49k| descr = *(kmp_mem_descr_t *)((kmp_uintptr_t)ptr - sizeof(kmp_mem_descr_t)); 2034| | 2035| 3.49k| KE_TRACE(26, (" __kmp_free: " 2036| 3.49k| "ptr_allocated=%p, size_allocated=%d, " 2037| 3.49k| "ptr_aligned=%p, size_aligned=%d\n", 2038| 3.49k| descr.ptr_allocated, (int)descr.size_allocated, 2039| 3.49k| descr.ptr_aligned, (int)descr.size_aligned)); 2040| |#if KMP_DEBUG 2041| | addr_allocated = (kmp_uintptr_t)descr.ptr_allocated; 2042| | addr_aligned = (kmp_uintptr_t)descr.ptr_aligned; 2043| | KMP_DEBUG_ASSERT(addr_aligned % CACHE_LINE == 0); 2044| | KMP_DEBUG_ASSERT(descr.ptr_aligned == ptr); 2045| | KMP_DEBUG_ASSERT(addr_allocated + sizeof(kmp_mem_descr_t) <= addr_aligned); 2046| | KMP_DEBUG_ASSERT(descr.size_aligned < descr.size_allocated); 2047| | KMP_DEBUG_ASSERT(addr_aligned + descr.size_aligned <= 2048| | addr_allocated + descr.size_allocated); 2049| | memset(descr.ptr_allocated, 0xEF, descr.size_allocated); 2050| |// Fill memory block with 0xEF, it helps catch using freed memory. 2051| |#endif 2052| | 2053| 3.49k|#ifndef LEAK_MEMORY 2054| 3.49k| KE_TRACE(10, (" free( %p )\n", descr.ptr_allocated)); 2055| |#ifdef KMP_DEBUG 2056| | _free_src_loc(descr.ptr_allocated, _file_, _line_); 2057| |#else 2058| 3.49k| free_src_loc(descr.ptr_allocated KMP_SRC_LOC_PARM); ------------------ | | 145| 3.49k|#define free_src_loc(args) _free_src_loc(args) | | ------------------ | | | | 190| 3.49k|#define _free_src_loc(ptr) free((ptr)) | | ------------------ ------------------ 2059| 3.49k|#endif 2060| 3.49k|#endif 2061| 3.49k| KMP_MB(); 2062| 3.49k| KE_TRACE(25, ("<- __kmp_free() returns\n")); 2063| 3.49k|} // func ___kmp_free __kmp_initialize_fast_memory: 2276| 32|void __kmp_initialize_fast_memory(kmp_info_t *this_thr) { 2277| 32| KE_TRACE(10, ("__kmp_initialize_fast_memory: Called from th %p\n", this_thr)); 2278| | 2279| 32| memset(this_thr->th.th_free_lists, 0, NUM_LISTS * sizeof(kmp_free_list_t)); ------------------ | | 2926| 32|#define NUM_LISTS 4 ------------------ 2280| 32|} __kmp_free_fast_memory: 2284| 32|void __kmp_free_fast_memory(kmp_info_t *th) { 2285| | // Suppose we use BGET underlying allocator, walk through its structures... 2286| 32| int bin; 2287| 32| thr_data_t *thr = get_thr_data(th); 2288| 32| void **lst = NULL; 2289| | 2290| 32| KE_TRACE( 2291| 32| 5, ("__kmp_free_fast_memory: Called T#%d\n", __kmp_gtid_from_thread(th))); 2292| | 2293| 32| __kmp_bget_dequeue(th); // Release any queued buffers 2294| | 2295| | // Dig through free lists and extract all allocated blocks 2296| 672| for (bin = 0; bin < MAX_BGET_BINS; ++bin) { ------------------ | | 124| 672|#define MAX_BGET_BINS (int)(sizeof(bget_bin_size) / sizeof(bufsize)) ------------------ | Branch (2296:17): [True: 640, False: 32] ------------------ 2297| 640| bfhead_t *b = thr->freelist[bin].ql.flink; 2298| 640| while (b != &thr->freelist[bin]) { ------------------ | Branch (2298:12): [True: 0, False: 640] ------------------ 2299| 0| if ((kmp_uintptr_t)b->bh.bb.bthr & 1) { // the buffer is allocated address ------------------ | Branch (2299:11): [True: 0, False: 0] ------------------ 2300| 0| *((void **)b) = 2301| 0| lst; // link the list (override bthr, but keep flink yet) 2302| 0| lst = (void **)b; // push b into lst 2303| 0| } 2304| 0| b = b->ql.flink; // get next buffer 2305| 0| } 2306| 640| } 2307| 32| while (lst != NULL) { ------------------ | Branch (2307:10): [True: 0, False: 32] ------------------ 2308| 0| void *next = *lst; 2309| 0| KE_TRACE(10, ("__kmp_free_fast_memory: freeing %p, next=%p th %p (%d)\n", 2310| 0| lst, next, th, __kmp_gtid_from_thread(th))); 2311| 0| (*thr->relfcn)(lst); 2312| 0|#if BufStats 2313| | // count blocks to prevent problems in __kmp_finalize_bget() 2314| 0| thr->numprel++; /* Nr of expansion block releases */ 2315| 0| thr->numpblk--; /* Total number of blocks */ 2316| 0|#endif 2317| 0| lst = (void **)next; 2318| 0| } 2319| | 2320| 32| KE_TRACE( 2321| 32| 5, ("__kmp_free_fast_memory: Freed T#%d\n", __kmp_gtid_from_thread(th))); 2322| 32|} kmp_alloc.cpp:_ZL12set_thr_dataP8kmp_info: 227| 32|static void set_thr_data(kmp_info_t *th) { 228| 32| int i; 229| 32| thr_data_t *data; 230| | 231| 32| data = (thr_data_t *)((!th->th.th_local.bget_data) ------------------ | Branch (231:25): [True: 32, False: 0] ------------------ 232| 32| ? __kmp_allocate(sizeof(*data)) ------------------ | | 3617| 32|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 233| 32| : th->th.th_local.bget_data); 234| | 235| 32| memset(data, '\0', sizeof(*data)); 236| | 237| 672| for (i = 0; i < MAX_BGET_BINS; ++i) { ------------------ | | 124| 672|#define MAX_BGET_BINS (int)(sizeof(bget_bin_size) / sizeof(bufsize)) ------------------ | Branch (237:15): [True: 640, False: 32] ------------------ 238| 640| data->freelist[i].ql.flink = &data->freelist[i]; 239| 640| data->freelist[i].ql.blink = &data->freelist[i]; 240| 640| } 241| | 242| 32| th->th.th_local.bget_data = data; 243| 32| th->th.th_local.bget_list = 0; 244| |#if !USE_CMP_XCHG_FOR_BGET 245| |#ifdef USE_QUEUING_LOCK_FOR_BGET 246| | __kmp_init_lock(&th->th.th_local.bget_lock); 247| |#else 248| | __kmp_init_bootstrap_lock(&th->th.th_local.bget_lock); 249| |#endif /* USE_LOCK_FOR_BGET */ 250| |#endif /* ! USE_CMP_XCHG_FOR_BGET */ 251| 32|} kmp_alloc.cpp:_ZL5bectlP8kmp_infoPFimiEPFPvmEPFvS3_El: 867| 32| bufsize pool_incr) { 868| 32| thr_data_t *thr = get_thr_data(th); 869| | 870| 32| thr->compfcn = compact; 871| 32| thr->acqfcn = acquire; 872| 32| thr->relfcn = release; 873| 32| thr->exp_incr = pool_incr; 874| 32|} kmp_alloc.cpp:_ZL12get_thr_dataP8kmp_info: 253| 64|static thr_data_t *get_thr_data(kmp_info_t *th) { 254| 64| thr_data_t *data; 255| | 256| 64| data = (thr_data_t *)th->th.th_local.bget_data; 257| | 258| 64| KMP_DEBUG_ASSERT(data != 0); 259| | 260| 64| return data; 261| 64|} kmp_alloc.cpp:_ZL18__kmp_bget_dequeueP8kmp_info: 264| 32|static void __kmp_bget_dequeue(kmp_info_t *th) { 265| 32| void *p = TCR_SYNC_PTR(th->th.th_local.bget_list); ------------------ | | 1158| 32|#define TCR_SYNC_PTR(a) ((void *)TCR_SYNC_8(a)) | | ------------------ | | | | 1140| 32|#define TCR_SYNC_8(a) (a) | | ------------------ ------------------ 266| | 267| 32| if (p != 0) { ------------------ | Branch (267:7): [True: 0, False: 32] ------------------ 268| 0|#if USE_CMP_XCHG_FOR_BGET 269| 0| { 270| 0| volatile void *old_value = TCR_SYNC_PTR(th->th.th_local.bget_list); ------------------ | | 1158| 0|#define TCR_SYNC_PTR(a) ((void *)TCR_SYNC_8(a)) | | ------------------ | | | | 1140| 0|#define TCR_SYNC_8(a) (a) | | ------------------ ------------------ 271| 0| while (!KMP_COMPARE_AND_STORE_PTR(&th->th.th_local.bget_list, ------------------ | | 813| 0| __sync_bool_compare_and_swap((void *volatile *)(p), (void *)(cv), \ | | 814| 0| (void *)(sv)) ------------------ | Branch (271:14): [True: 0, False: 0] ------------------ 272| 0| CCAST(void *, old_value), nullptr)) { 273| 0| KMP_CPU_PAUSE(); ------------------ | | 1498| 0|#define KMP_CPU_PAUSE() __kmp_x86_pause() ------------------ 274| 0| old_value = TCR_SYNC_PTR(th->th.th_local.bget_list); ------------------ | | 1158| 0|#define TCR_SYNC_PTR(a) ((void *)TCR_SYNC_8(a)) | | ------------------ | | | | 1140| 0|#define TCR_SYNC_8(a) (a) | | ------------------ ------------------ 275| 0| } 276| 0| p = CCAST(void *, old_value); ------------------ | | 229| 0|#define CCAST(type, var) const_cast(var) ------------------ 277| 0| } 278| |#else /* ! USE_CMP_XCHG_FOR_BGET */ 279| |#ifdef USE_QUEUING_LOCK_FOR_BGET 280| | __kmp_acquire_lock(&th->th.th_local.bget_lock, __kmp_gtid_from_thread(th)); 281| |#else 282| | __kmp_acquire_bootstrap_lock(&th->th.th_local.bget_lock); 283| |#endif /* USE_QUEUING_LOCK_FOR_BGET */ 284| | 285| | p = (void *)th->th.th_local.bget_list; 286| | th->th.th_local.bget_list = 0; 287| | 288| |#ifdef USE_QUEUING_LOCK_FOR_BGET 289| | __kmp_release_lock(&th->th.th_local.bget_lock, __kmp_gtid_from_thread(th)); 290| |#else 291| | __kmp_release_bootstrap_lock(&th->th.th_local.bget_lock); 292| |#endif 293| |#endif /* USE_CMP_XCHG_FOR_BGET */ 294| | 295| | /* Check again to make sure the list is not empty */ 296| 0| while (p != 0) { ------------------ | Branch (296:12): [True: 0, False: 0] ------------------ 297| 0| void *buf = p; 298| 0| bfhead_t *b = BFH(((char *)p) - sizeof(bhead_t)); ------------------ | | 165| 0|#define BFH(p) ((bfhead_t *)(p)) ------------------ 299| | 300| 0| KMP_DEBUG_ASSERT(b->bh.bb.bsize != 0); 301| 0| KMP_DEBUG_ASSERT(((kmp_uintptr_t)TCR_PTR(b->bh.bb.bthr) & ~1) == 302| 0| (kmp_uintptr_t)th); // clear possible mark 303| 0| KMP_DEBUG_ASSERT(b->ql.blink == 0); 304| | 305| 0| p = (void *)b->ql.flink; 306| | 307| 0| brel(th, buf); 308| 0| } 309| 0| } 310| 32|} kmp_alloc.cpp:_ZL21___kmp_allocate_alignmm: 1913| 3.49k| size_t alignment KMP_SRC_LOC_DECL) { 1914| | /* __kmp_allocate() allocates (by call to malloc()) bigger memory block than 1915| | requested to return properly aligned pointer. Original pointer returned 1916| | by malloc() and size of allocated block is saved in descriptor just 1917| | before the aligned pointer. This information used by __kmp_free() -- it 1918| | has to pass to free() original pointer, not aligned one. 1919| | 1920| | +---------+------------+-----------------------------------+---------+ 1921| | | padding | descriptor | aligned block | padding | 1922| | +---------+------------+-----------------------------------+---------+ 1923| | ^ ^ 1924| | | | 1925| | | +- Aligned pointer returned to caller 1926| | +- Pointer returned by malloc() 1927| | 1928| | Aligned block is filled with zeros, paddings are filled with 0xEF. */ 1929| | 1930| 3.49k| kmp_mem_descr_t descr; 1931| 3.49k| kmp_uintptr_t addr_allocated; // Address returned by malloc(). 1932| 3.49k| kmp_uintptr_t addr_aligned; // Aligned address to return to caller. 1933| 3.49k| kmp_uintptr_t addr_descr; // Address of memory block descriptor. 1934| | 1935| 3.49k| KE_TRACE(25, ("-> ___kmp_allocate_align( %d, %d ) called from %s:%d\n", 1936| 3.49k| (int)size, (int)alignment KMP_SRC_LOC_PARM)); 1937| | 1938| 3.49k| KMP_DEBUG_ASSERT(alignment < 32 * 1024); // Alignment should not be too 1939| 3.49k| KMP_DEBUG_ASSERT(sizeof(void *) <= sizeof(kmp_uintptr_t)); 1940| | // Make sure kmp_uintptr_t is enough to store addresses. 1941| | 1942| 3.49k| descr.size_aligned = size; 1943| 3.49k| descr.size_allocated = 1944| 3.49k| descr.size_aligned + sizeof(kmp_mem_descr_t) + alignment; 1945| | 1946| |#if KMP_DEBUG 1947| | descr.ptr_allocated = _malloc_src_loc(descr.size_allocated, _file_, _line_); 1948| |#else 1949| 3.49k| descr.ptr_allocated = malloc_src_loc(descr.size_allocated KMP_SRC_LOC_PARM); ------------------ | | 144| 3.49k|#define malloc_src_loc(args) _malloc_src_loc(args) | | ------------------ | | | | 189| 3.49k|#define _malloc_src_loc(size) malloc((size)) | | ------------------ ------------------ 1950| 3.49k|#endif 1951| 3.49k| KE_TRACE(10, (" malloc( %d ) returned %p\n", (int)descr.size_allocated, 1952| 3.49k| descr.ptr_allocated)); 1953| 3.49k| if (descr.ptr_allocated == NULL) { ------------------ | Branch (1953:7): [True: 0, False: 3.49k] ------------------ 1954| 0| KMP_FATAL(OutOfHeapMemory); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 1955| 0| } 1956| | 1957| 3.49k| addr_allocated = (kmp_uintptr_t)descr.ptr_allocated; 1958| 3.49k| addr_aligned = 1959| 3.49k| (addr_allocated + sizeof(kmp_mem_descr_t) + alignment) & ~(alignment - 1); 1960| 3.49k| addr_descr = addr_aligned - sizeof(kmp_mem_descr_t); 1961| | 1962| 3.49k| descr.ptr_aligned = (void *)addr_aligned; 1963| | 1964| 3.49k| KE_TRACE(26, (" ___kmp_allocate_align: " 1965| 3.49k| "ptr_allocated=%p, size_allocated=%d, " 1966| 3.49k| "ptr_aligned=%p, size_aligned=%d\n", 1967| 3.49k| descr.ptr_allocated, (int)descr.size_allocated, 1968| 3.49k| descr.ptr_aligned, (int)descr.size_aligned)); 1969| | 1970| 3.49k| KMP_DEBUG_ASSERT(addr_allocated <= addr_descr); 1971| 3.49k| KMP_DEBUG_ASSERT(addr_descr + sizeof(kmp_mem_descr_t) == addr_aligned); 1972| 3.49k| KMP_DEBUG_ASSERT(addr_aligned + descr.size_aligned <= 1973| 3.49k| addr_allocated + descr.size_allocated); 1974| 3.49k| KMP_DEBUG_ASSERT(addr_aligned % alignment == 0); 1975| |#ifdef KMP_DEBUG 1976| | memset(descr.ptr_allocated, 0xEF, descr.size_allocated); 1977| |// Fill allocated memory block with 0xEF. 1978| |#endif 1979| 3.49k| memset(descr.ptr_aligned, 0x00, descr.size_aligned); 1980| | // Fill the aligned memory block (which is intended for using by caller) with 1981| | // 0x00. Do not 1982| | // put this filling under KMP_DEBUG condition! Many callers expect zeroed 1983| | // memory. (Padding 1984| | // bytes remain filled with 0xEF in debugging library.) 1985| 3.49k| *((kmp_mem_descr_t *)addr_descr) = descr; 1986| | 1987| 3.49k| KMP_MB(); 1988| | 1989| 3.49k| KE_TRACE(25, ("<- ___kmp_allocate_align() returns %p\n", descr.ptr_aligned)); 1990| 3.49k| return descr.ptr_aligned; 1991| 3.49k|} // func ___kmp_allocate_align kmp_runtime.cpp:_ZL22__kmp_init_atomic_lockP16kmp_queuing_lock: 405| 13|static inline void __kmp_init_atomic_lock(kmp_atomic_lock_t *lck) { 406| 13| __kmp_init_queuing_lock(lck); 407| 13|} __kmp_barrier: 2084| 37.1k| void (*reduce)(void *, void *)) { 2085| 37.1k| return __kmp_barrier_template<>(bt, gtid, is_split, reduce_size, reduce_data, 2086| 37.1k| reduce); 2087| 37.1k|} _Z18__kmp_join_barrieri: 2158| 42.3k|void __kmp_join_barrier(int gtid) { 2159| 42.3k| KMP_TIME_PARTITIONED_BLOCK(OMP_join_barrier); ------------------ | | 1013| 42.3k|#define KMP_TIME_PARTITIONED_BLOCK(name) ((void)0) ------------------ 2160| 42.3k| KMP_SET_THREAD_STATE_BLOCK(FORK_JOIN_BARRIER); ------------------ | | 1018| 42.3k|#define KMP_SET_THREAD_STATE_BLOCK(state_name) ((void)0) ------------------ 2161| | 2162| 42.3k| KMP_DEBUG_ASSERT(__kmp_threads && __kmp_threads[gtid]); 2163| | 2164| 42.3k| kmp_info_t *this_thr = __kmp_threads[gtid]; 2165| 42.3k| kmp_team_t *team; 2166| 42.3k| int tid; 2167| |#ifdef KMP_DEBUG 2168| | int team_id; 2169| |#endif /* KMP_DEBUG */ 2170| |#if USE_ITT_BUILD 2171| | void *itt_sync_obj = NULL; 2172| |#if USE_ITT_NOTIFY 2173| | if (__itt_sync_create_ptr || KMP_ITT_DEBUG) // Don't call routine without need 2174| | // Get object created at fork_barrier 2175| | itt_sync_obj = __kmp_itt_barrier_object(gtid, bs_forkjoin_barrier); 2176| |#endif 2177| |#endif /* USE_ITT_BUILD */ 2178| |#if ((USE_ITT_BUILD && USE_ITT_NOTIFY) || defined KMP_DEBUG) 2179| | int nproc = this_thr->th.th_team_nproc; 2180| |#endif 2181| 42.3k| KMP_MB(); 2182| | 2183| | // Get current info 2184| 42.3k| team = this_thr->th.th_team; 2185| 42.3k| KMP_DEBUG_ASSERT(nproc == team->t.t_nproc); 2186| 42.3k| tid = __kmp_tid_from_gtid(gtid); 2187| |#ifdef KMP_DEBUG 2188| | team_id = team->t.t_id; 2189| | kmp_info_t *master_thread = this_thr->th.th_team_master; 2190| | if (master_thread != team->t.t_threads[0]) { 2191| | __kmp_print_structure(); 2192| | } 2193| |#endif /* KMP_DEBUG */ 2194| 42.3k| KMP_DEBUG_ASSERT(master_thread == team->t.t_threads[0]); 2195| 42.3k| KMP_MB(); 2196| | 2197| | // Verify state 2198| 42.3k| KMP_DEBUG_ASSERT(TCR_PTR(this_thr->th.th_team)); 2199| 42.3k| KMP_DEBUG_ASSERT(TCR_PTR(this_thr->th.th_root)); 2200| 42.3k| KMP_DEBUG_ASSERT(this_thr == team->t.t_threads[tid]); 2201| 42.3k| KA_TRACE(10, ("__kmp_join_barrier: T#%d(%d:%d) arrived at join barrier\n", 2202| 42.3k| gtid, team_id, tid)); 2203| | 2204| 42.3k|#if OMPT_SUPPORT 2205| 42.3k| if (ompt_enabled.enabled) { ------------------ | Branch (2205:7): [True: 0, False: 42.3k] ------------------ 2206| 0|#if OMPT_OPTIONAL 2207| 0| ompt_data_t *my_task_data; 2208| 0| ompt_data_t *my_parallel_data; 2209| 0| void *codeptr = NULL; 2210| 0| int ds_tid = this_thr->th.th_info.ds.ds_tid; 2211| 0| if (KMP_MASTER_TID(ds_tid) && ------------------ | | 1271| 0|#define KMP_MASTER_TID(tid) (0 == (tid)) | | ------------------ | | | Branch (1271:29): [True: 0, False: 0] | | ------------------ ------------------ 2212| 0| (ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait) || ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ | Branch (2212:10): [True: 0, False: 0] ------------------ 2213| 0| ompt_callbacks.ompt_callback(ompt_callback_sync_region))) ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ | Branch (2213:10): [True: 0, False: 0] ------------------ 2214| 0| codeptr = team->t.ompt_team_info.master_return_address; 2215| 0| my_task_data = OMPT_CUR_TASK_DATA(this_thr); ------------------ | | 70| 0| (&((thr)->th.th_current_task->ompt_task_info.task_data)) ------------------ 2216| 0| my_parallel_data = OMPT_CUR_TEAM_DATA(this_thr); ------------------ | | 73| 0| (&((thr)->th.th_team->t.ompt_team_info.parallel_data)) ------------------ 2217| 0| if (ompt_enabled.ompt_callback_sync_region) { ------------------ | Branch (2217:9): [True: 0, False: 0] ------------------ 2218| 0| ompt_callbacks.ompt_callback(ompt_callback_sync_region)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 2219| 0| ompt_sync_region_barrier_implicit, ompt_scope_begin, my_parallel_data, 2220| 0| my_task_data, codeptr); 2221| 0| } 2222| 0| if (ompt_enabled.ompt_callback_sync_region_wait) { ------------------ | Branch (2222:9): [True: 0, False: 0] ------------------ 2223| 0| ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 2224| 0| ompt_sync_region_barrier_implicit, ompt_scope_begin, my_parallel_data, 2225| 0| my_task_data, codeptr); 2226| 0| } 2227| 0| if (!KMP_MASTER_TID(ds_tid)) ------------------ | | 1271| 0|#define KMP_MASTER_TID(tid) (0 == (tid)) ------------------ | Branch (2227:9): [True: 0, False: 0] ------------------ 2228| 0| this_thr->th.ompt_thread_info.task_data = *OMPT_CUR_TASK_DATA(this_thr); ------------------ | | 70| 0| (&((thr)->th.th_current_task->ompt_task_info.task_data)) ------------------ 2229| 0|#endif 2230| 0| this_thr->th.ompt_thread_info.state = ompt_state_wait_barrier_implicit; 2231| 0| } 2232| 42.3k|#endif 2233| | 2234| 42.3k| if (__kmp_tasking_mode == tskm_extra_barrier) { ------------------ | Branch (2234:7): [True: 0, False: 42.3k] ------------------ 2235| 0| __kmp_tasking_barrier(team, this_thr, gtid); 2236| 0| KA_TRACE(10, ("__kmp_join_barrier: T#%d(%d:%d) past tasking barrier\n", 2237| 0| gtid, team_id, tid)); 2238| 0| } 2239| |#ifdef KMP_DEBUG 2240| | if (__kmp_tasking_mode != tskm_immediate_exec) { 2241| | KA_TRACE(20, ("__kmp_join_barrier: T#%d, old team = %d, old task_team = " 2242| | "%p, th_task_team = %p\n", 2243| | __kmp_gtid_from_thread(this_thr), team_id, 2244| | team->t.t_task_team[this_thr->th.th_task_state], 2245| | this_thr->th.th_task_team)); 2246| | if (this_thr->th.th_task_team) 2247| | KMP_DEBUG_ASSERT(this_thr->th.th_task_team == 2248| | team->t.t_task_team[this_thr->th.th_task_state]); 2249| | } 2250| |#endif /* KMP_DEBUG */ 2251| | 2252| | /* Copy the blocktime info to the thread, where __kmp_wait_template() can 2253| | access it when the team struct is not guaranteed to exist. Doing these 2254| | loads causes a cache miss slows down EPCC parallel by 2x. As a workaround, 2255| | we do not perform the copy if blocktime=infinite, since the values are not 2256| | used by __kmp_wait_template() in that case. */ 2257| 42.3k| if (__kmp_dflt_blocktime != KMP_MAX_BLOCKTIME) { ------------------ | | 1184| 42.3k| (INT_MAX) /* Must be this for "infinite" setting the work */ ------------------ | Branch (2257:7): [True: 42.3k, False: 18.4E] ------------------ 2258| |#if KMP_USE_MONITOR 2259| | this_thr->th.th_team_bt_intervals = 2260| | team->t.t_implicit_task_taskdata[tid].td_icvs.bt_intervals; 2261| | this_thr->th.th_team_bt_set = 2262| | team->t.t_implicit_task_taskdata[tid].td_icvs.bt_set; 2263| |#else 2264| 42.3k| this_thr->th.th_team_bt_intervals = KMP_BLOCKTIME_INTERVAL(team, tid); ------------------ | | 1221| 42.3k| (KMP_BLOCKTIME(team, tid) * __kmp_ticks_per_msec) | | ------------------ | | | | 1210| 42.3k| (get__bt_set(team, tid) ? get__blocktime(team, tid) : __kmp_dflt_blocktime) | | | | ------------------ | | | | | | 2289| 42.3k| ((xteam)->t.t_threads[(xtid)]->th.th_current_task->td_icvs.bt_set) | | | | | | ------------------ | | | | | | | Branch (2289:3): [True: 0, False: 42.3k] | | | | | | ------------------ | | | | ------------------ | | | | (get__bt_set(team, tid) ? get__blocktime(team, tid) : __kmp_dflt_blocktime) | | | | ------------------ | | | | | | 2287| 0| ((xteam)->t.t_threads[(xtid)]->th.th_current_task->td_icvs.blocktime) | | | | ------------------ | | ------------------ ------------------ 2265| 42.3k|#endif 2266| 42.3k| } 2267| | 2268| |#if USE_ITT_BUILD 2269| | if (__itt_sync_create_ptr || KMP_ITT_DEBUG) 2270| | __kmp_itt_barrier_starting(gtid, itt_sync_obj); 2271| |#endif /* USE_ITT_BUILD */ 2272| | 2273| 42.3k| switch (__kmp_barrier_gather_pattern[bs_forkjoin_barrier]) { 2274| 0| case bp_dist_bar: { ------------------ | Branch (2274:3): [True: 0, False: 42.3k] ------------------ 2275| 0| __kmp_dist_barrier_gather(bs_forkjoin_barrier, this_thr, gtid, tid, 2276| 0| NULL USE_ITT_BUILD_ARG(itt_sync_obj)); 2277| 0| break; 2278| 0| } 2279| 42.4k| case bp_hyper_bar: { ------------------ | Branch (2279:3): [True: 42.4k, False: 18.4E] ------------------ 2280| 42.4k| KMP_ASSERT(__kmp_barrier_gather_branch_bits[bs_forkjoin_barrier]); ------------------ | | 50| 42.4k| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 42.4k] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 2281| 42.4k| __kmp_hyper_barrier_gather(bs_forkjoin_barrier, this_thr, gtid, tid, 2282| 42.4k| NULL USE_ITT_BUILD_ARG(itt_sync_obj)); 2283| 42.4k| break; 2284| 0| } 2285| 0| case bp_hierarchical_bar: { ------------------ | Branch (2285:3): [True: 0, False: 42.3k] ------------------ 2286| 0| __kmp_hierarchical_barrier_gather(bs_forkjoin_barrier, this_thr, gtid, tid, 2287| 0| NULL USE_ITT_BUILD_ARG(itt_sync_obj)); 2288| 0| break; 2289| 0| } 2290| 0| case bp_tree_bar: { ------------------ | Branch (2290:3): [True: 0, False: 42.3k] ------------------ 2291| 0| KMP_ASSERT(__kmp_barrier_gather_branch_bits[bs_forkjoin_barrier]); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 2292| 0| __kmp_tree_barrier_gather(bs_forkjoin_barrier, this_thr, gtid, tid, 2293| 0| NULL USE_ITT_BUILD_ARG(itt_sync_obj)); 2294| 0| break; 2295| 0| } 2296| 0| default: { ------------------ | Branch (2296:3): [True: 0, False: 42.3k] ------------------ 2297| 0| __kmp_linear_barrier_gather(bs_forkjoin_barrier, this_thr, gtid, tid, 2298| 0| NULL USE_ITT_BUILD_ARG(itt_sync_obj)); 2299| 0| } 2300| 42.3k| } 2301| | 2302| | /* From this point on, the team data structure may be deallocated at any time 2303| | by the primary thread - it is unsafe to reference it in any of the worker 2304| | threads. Any per-team data items that need to be referenced before the 2305| | end of the barrier should be moved to the kmp_task_team_t structs. */ 2306| 43.1k| if (KMP_MASTER_TID(tid)) { ------------------ | | 1271| 43.1k|#define KMP_MASTER_TID(tid) (0 == (tid)) | | ------------------ | | | Branch (1271:29): [True: 1.35k, False: 41.8k] | | ------------------ ------------------ 2307| 1.35k| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (2307:9): [True: 1.35k, False: 0] ------------------ 2308| 1.35k| __kmp_task_team_wait(this_thr, team USE_ITT_BUILD_ARG(itt_sync_obj)); 2309| 1.35k| } 2310| 1.35k| if (__kmp_display_affinity) { ------------------ | Branch (2310:9): [True: 0, False: 1.35k] ------------------ 2311| 0| KMP_CHECK_UPDATE(team->t.t_display_affinity, 0); ------------------ | | 2280| 0| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 0] | | ------------------ | | 2281| 0| (a) = (b) ------------------ 2312| 0| } 2313| |#if KMP_STATS_ENABLED 2314| | // Have primary thread flag the workers to indicate they are now waiting for 2315| | // next parallel region, Also wake them up so they switch their timers to 2316| | // idle. 2317| | for (int i = 0; i < team->t.t_nproc; ++i) { 2318| | kmp_info_t *team_thread = team->t.t_threads[i]; 2319| | if (team_thread == this_thr) 2320| | continue; 2321| | team_thread->th.th_stats->setIdleFlag(); 2322| | if (__kmp_dflt_blocktime != KMP_MAX_BLOCKTIME && 2323| | team_thread->th.th_sleep_loc != NULL) 2324| | __kmp_null_resume_wrapper(team_thread); 2325| | } 2326| |#endif 2327| |#if USE_ITT_BUILD 2328| | if (__itt_sync_create_ptr || KMP_ITT_DEBUG) 2329| | __kmp_itt_barrier_middle(gtid, itt_sync_obj); 2330| |#endif /* USE_ITT_BUILD */ 2331| | 2332| |#if USE_ITT_BUILD && USE_ITT_NOTIFY 2333| | // Join barrier - report frame end 2334| | if ((__itt_frame_submit_v3_ptr || KMP_ITT_DEBUG) && 2335| | __kmp_forkjoin_frames_mode && 2336| | (this_thr->th.th_teams_microtask == NULL || // either not in teams 2337| | this_thr->th.th_teams_size.nteams == 1) && // or inside single team 2338| | team->t.t_active_level == 1) { 2339| | kmp_uint64 cur_time = __itt_get_timestamp(); 2340| | ident_t *loc = team->t.t_ident; 2341| | kmp_info_t **other_threads = team->t.t_threads; 2342| | switch (__kmp_forkjoin_frames_mode) { 2343| | case 1: 2344| | __kmp_itt_frame_submit(gtid, this_thr->th.th_frame_time, cur_time, 0, 2345| | loc, nproc); 2346| | break; 2347| | case 2: 2348| | __kmp_itt_frame_submit(gtid, this_thr->th.th_bar_min_time, cur_time, 1, 2349| | loc, nproc); 2350| | break; 2351| | case 3: 2352| | if (__itt_metadata_add_ptr) { 2353| | // Initialize with primary thread's wait time 2354| | kmp_uint64 delta = cur_time - this_thr->th.th_bar_arrive_time; 2355| | // Set arrive time to zero to be able to check it in 2356| | // __kmp_invoke_task(); the same is done inside the loop below 2357| | this_thr->th.th_bar_arrive_time = 0; 2358| | for (int i = 1; i < nproc; ++i) { 2359| | delta += (cur_time - other_threads[i]->th.th_bar_arrive_time); 2360| | other_threads[i]->th.th_bar_arrive_time = 0; 2361| | } 2362| | __kmp_itt_metadata_imbalance(gtid, this_thr->th.th_frame_time, 2363| | cur_time, delta, 0); 2364| | } 2365| | __kmp_itt_frame_submit(gtid, this_thr->th.th_frame_time, cur_time, 0, 2366| | loc, nproc); 2367| | this_thr->th.th_frame_time = cur_time; 2368| | break; 2369| | } 2370| | } 2371| |#endif /* USE_ITT_BUILD */ 2372| 1.35k| } 2373| |#if USE_ITT_BUILD 2374| | else { 2375| | if (__itt_sync_create_ptr || KMP_ITT_DEBUG) 2376| | __kmp_itt_barrier_middle(gtid, itt_sync_obj); 2377| | } 2378| |#endif /* USE_ITT_BUILD */ 2379| | 2380| |#if KMP_DEBUG 2381| | if (KMP_MASTER_TID(tid)) { 2382| | KA_TRACE( 2383| | 15, 2384| | ("__kmp_join_barrier: T#%d(%d:%d) says all %d team threads arrived\n", 2385| | gtid, team_id, tid, nproc)); 2386| | } 2387| |#endif /* KMP_DEBUG */ 2388| | 2389| | // TODO now, mark worker threads as done so they may be disbanded 2390| 43.1k| KMP_MB(); // Flush all pending memory write invalidates. 2391| 43.1k| KA_TRACE(10, 2392| 43.1k| ("__kmp_join_barrier: T#%d(%d:%d) leaving\n", gtid, team_id, tid)); 2393| | 2394| 43.1k|} _Z18__kmp_fork_barrierii: 2398| 43.2k|void __kmp_fork_barrier(int gtid, int tid) { 2399| 43.2k| KMP_TIME_PARTITIONED_BLOCK(OMP_fork_barrier); ------------------ | | 1013| 43.2k|#define KMP_TIME_PARTITIONED_BLOCK(name) ((void)0) ------------------ 2400| 43.2k| KMP_SET_THREAD_STATE_BLOCK(FORK_JOIN_BARRIER); ------------------ | | 1018| 43.2k|#define KMP_SET_THREAD_STATE_BLOCK(state_name) ((void)0) ------------------ 2401| 43.2k| kmp_info_t *this_thr = __kmp_threads[gtid]; 2402| 43.2k| kmp_team_t *team = (tid == 0) ? this_thr->th.th_team : NULL; ------------------ | Branch (2402:22): [True: 1.35k, False: 41.8k] ------------------ 2403| |#if USE_ITT_BUILD 2404| | void *itt_sync_obj = NULL; 2405| |#endif /* USE_ITT_BUILD */ 2406| 43.2k| if (team) ------------------ | Branch (2406:7): [True: 1.35k, False: 41.8k] ------------------ 2407| | 2408| 1.35k| KA_TRACE(10, ("__kmp_fork_barrier: T#%d(%d:%d) has arrived\n", gtid, 2409| 1.35k| (team != NULL) ? team->t.t_id : -1, tid)); 2410| | 2411| | // th_team pointer only valid for primary thread here 2412| 43.2k| if (KMP_MASTER_TID(tid)) { ------------------ | | 1271| 43.2k|#define KMP_MASTER_TID(tid) (0 == (tid)) | | ------------------ | | | Branch (1271:29): [True: 1.35k, False: 41.8k] | | ------------------ ------------------ 2413| |#if USE_ITT_BUILD && USE_ITT_NOTIFY 2414| | if (__itt_sync_create_ptr || KMP_ITT_DEBUG) { 2415| | // Create itt barrier object 2416| | itt_sync_obj = __kmp_itt_barrier_object(gtid, bs_forkjoin_barrier, 1); 2417| | __kmp_itt_barrier_middle(gtid, itt_sync_obj); // Call acquired/releasing 2418| | } 2419| |#endif /* USE_ITT_BUILD && USE_ITT_NOTIFY */ 2420| | 2421| |#ifdef KMP_DEBUG 2422| | KMP_DEBUG_ASSERT(team); 2423| | kmp_info_t **other_threads = team->t.t_threads; 2424| | int i; 2425| | 2426| | // Verify state 2427| | KMP_MB(); 2428| | 2429| | for (i = 1; i < team->t.t_nproc; ++i) { 2430| | KA_TRACE(500, 2431| | ("__kmp_fork_barrier: T#%d(%d:0) checking T#%d(%d:%d) fork go " 2432| | "== %u.\n", 2433| | gtid, team->t.t_id, other_threads[i]->th.th_info.ds.ds_gtid, 2434| | team->t.t_id, other_threads[i]->th.th_info.ds.ds_tid, 2435| | other_threads[i]->th.th_bar[bs_forkjoin_barrier].bb.b_go)); 2436| | KMP_DEBUG_ASSERT( 2437| | (TCR_4(other_threads[i]->th.th_bar[bs_forkjoin_barrier].bb.b_go) & 2438| | ~(KMP_BARRIER_SLEEP_STATE)) == KMP_INIT_BARRIER_STATE); 2439| | KMP_DEBUG_ASSERT(other_threads[i]->th.th_team == team); 2440| | } 2441| |#endif 2442| | 2443| 1.35k| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (2443:9): [True: 1.35k, False: 0] ------------------ 2444| | // 0 indicates setup current task team if nthreads > 1 2445| 1.35k| __kmp_task_team_setup(this_thr, team, 0); 2446| 1.35k| } 2447| | 2448| | /* The primary thread may have changed its blocktime between join barrier 2449| | and fork barrier. Copy the blocktime info to the thread, where 2450| | __kmp_wait_template() can access it when the team struct is not 2451| | guaranteed to exist. */ 2452| | // See note about the corresponding code in __kmp_join_barrier() being 2453| | // performance-critical 2454| 1.35k| if (__kmp_dflt_blocktime != KMP_MAX_BLOCKTIME) { ------------------ | | 1184| 1.35k| (INT_MAX) /* Must be this for "infinite" setting the work */ ------------------ | Branch (2454:9): [True: 1.35k, False: 0] ------------------ 2455| |#if KMP_USE_MONITOR 2456| | this_thr->th.th_team_bt_intervals = 2457| | team->t.t_implicit_task_taskdata[tid].td_icvs.bt_intervals; 2458| | this_thr->th.th_team_bt_set = 2459| | team->t.t_implicit_task_taskdata[tid].td_icvs.bt_set; 2460| |#else 2461| 1.35k| this_thr->th.th_team_bt_intervals = KMP_BLOCKTIME_INTERVAL(team, tid); ------------------ | | 1221| 1.35k| (KMP_BLOCKTIME(team, tid) * __kmp_ticks_per_msec) | | ------------------ | | | | 1210| 1.35k| (get__bt_set(team, tid) ? get__blocktime(team, tid) : __kmp_dflt_blocktime) | | | | ------------------ | | | | | | 2289| 1.35k| ((xteam)->t.t_threads[(xtid)]->th.th_current_task->td_icvs.bt_set) | | | | | | ------------------ | | | | | | | Branch (2289:3): [True: 0, False: 1.35k] | | | | | | ------------------ | | | | ------------------ | | | | (get__bt_set(team, tid) ? get__blocktime(team, tid) : __kmp_dflt_blocktime) | | | | ------------------ | | | | | | 2287| 0| ((xteam)->t.t_threads[(xtid)]->th.th_current_task->td_icvs.blocktime) | | | | ------------------ | | ------------------ ------------------ 2462| 1.35k|#endif 2463| 1.35k| } 2464| 1.35k| } // primary thread 2465| | 2466| 43.2k| switch (__kmp_barrier_release_pattern[bs_forkjoin_barrier]) { 2467| 0| case bp_dist_bar: { ------------------ | Branch (2467:3): [True: 0, False: 43.2k] ------------------ 2468| 0| __kmp_dist_barrier_release(bs_forkjoin_barrier, this_thr, gtid, tid, 2469| 0| TRUE USE_ITT_BUILD_ARG(NULL)); ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 2470| 0| break; 2471| 0| } 2472| 43.1k| case bp_hyper_bar: { ------------------ | Branch (2472:3): [True: 43.1k, False: 5] ------------------ 2473| 43.1k| KMP_ASSERT(__kmp_barrier_release_branch_bits[bs_forkjoin_barrier]); ------------------ | | 50| 43.1k| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 43.1k] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 2474| 43.1k| __kmp_hyper_barrier_release(bs_forkjoin_barrier, this_thr, gtid, tid, 2475| 43.1k| TRUE USE_ITT_BUILD_ARG(itt_sync_obj)); ------------------ | | 1280| 43.1k|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 43.1k|#define FALSE 0 | | ------------------ ------------------ 2476| 43.1k| break; 2477| 0| } 2478| 0| case bp_hierarchical_bar: { ------------------ | Branch (2478:3): [True: 0, False: 43.2k] ------------------ 2479| 0| __kmp_hierarchical_barrier_release(bs_forkjoin_barrier, this_thr, gtid, tid, 2480| 0| TRUE USE_ITT_BUILD_ARG(itt_sync_obj)); ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 2481| 0| break; 2482| 0| } 2483| 0| case bp_tree_bar: { ------------------ | Branch (2483:3): [True: 0, False: 43.2k] ------------------ 2484| 0| KMP_ASSERT(__kmp_barrier_release_branch_bits[bs_forkjoin_barrier]); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 2485| 0| __kmp_tree_barrier_release(bs_forkjoin_barrier, this_thr, gtid, tid, 2486| 0| TRUE USE_ITT_BUILD_ARG(itt_sync_obj)); ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 2487| 0| break; 2488| 0| } 2489| 0| default: { ------------------ | Branch (2489:3): [True: 0, False: 43.2k] ------------------ 2490| 0| __kmp_linear_barrier_release(bs_forkjoin_barrier, this_thr, gtid, tid, 2491| 0| TRUE USE_ITT_BUILD_ARG(itt_sync_obj)); ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 2492| 0| } 2493| 43.2k| } 2494| | 2495| 42.2k|#if OMPT_SUPPORT 2496| 42.2k| if (ompt_enabled.enabled && ------------------ | Branch (2496:7): [True: 0, False: 42.2k] ------------------ 2497| 0| this_thr->th.ompt_thread_info.state == ompt_state_wait_barrier_implicit) { ------------------ | Branch (2497:7): [True: 0, False: 0] ------------------ 2498| 0| int ds_tid = this_thr->th.th_info.ds.ds_tid; 2499| 0| ompt_data_t *task_data = (team) ------------------ | Branch (2499:30): [True: 0, False: 0] ------------------ 2500| 0| ? OMPT_CUR_TASK_DATA(this_thr) ------------------ | | 70| 0| (&((thr)->th.th_current_task->ompt_task_info.task_data)) ------------------ 2501| 0| : &(this_thr->th.ompt_thread_info.task_data); 2502| 0| this_thr->th.ompt_thread_info.state = ompt_state_overhead; 2503| 0|#if OMPT_OPTIONAL 2504| 0| void *codeptr = NULL; 2505| 0| if (KMP_MASTER_TID(ds_tid) && ------------------ | | 1271| 0|#define KMP_MASTER_TID(tid) (0 == (tid)) | | ------------------ | | | Branch (1271:29): [True: 0, False: 0] | | ------------------ ------------------ 2506| 0| (ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait) || ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ | Branch (2506:10): [True: 0, False: 0] ------------------ 2507| 0| ompt_callbacks.ompt_callback(ompt_callback_sync_region))) ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ | Branch (2507:10): [True: 0, False: 0] ------------------ 2508| 0| codeptr = team ? team->t.ompt_team_info.master_return_address : NULL; ------------------ | Branch (2508:17): [True: 0, False: 0] ------------------ 2509| 0| if (ompt_enabled.ompt_callback_sync_region_wait) { ------------------ | Branch (2509:9): [True: 0, False: 0] ------------------ 2510| 0| ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 2511| 0| ompt_sync_region_barrier_implicit, ompt_scope_end, NULL, task_data, 2512| 0| codeptr); 2513| 0| } 2514| 0| if (ompt_enabled.ompt_callback_sync_region) { ------------------ | Branch (2514:9): [True: 0, False: 0] ------------------ 2515| 0| ompt_callbacks.ompt_callback(ompt_callback_sync_region)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 2516| 0| ompt_sync_region_barrier_implicit, ompt_scope_end, NULL, task_data, 2517| 0| codeptr); 2518| 0| } 2519| 0|#endif 2520| 0| if (!KMP_MASTER_TID(ds_tid) && ompt_enabled.ompt_callback_implicit_task) { ------------------ | | 1271| 0|#define KMP_MASTER_TID(tid) (0 == (tid)) ------------------ | Branch (2520:9): [True: 0, False: 0] | Branch (2520:36): [True: 0, False: 0] ------------------ 2521| 0| ompt_callbacks.ompt_callback(ompt_callback_implicit_task)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 2522| 0| ompt_scope_end, NULL, task_data, 0, ds_tid, 2523| 0| ompt_task_implicit); // TODO: Can this be ompt_task_initial? 2524| 0| } 2525| 0| } 2526| 42.2k|#endif 2527| | 2528| | // Early exit for reaping threads releasing forkjoin barrier 2529| 42.2k| if (TCR_4(__kmp_global.g.g_done)) { ------------------ | | 1127| 42.2k|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 31, False: 42.1k] | | ------------------ ------------------ 2530| 31| this_thr->th.th_task_team = NULL; 2531| | 2532| |#if USE_ITT_BUILD && USE_ITT_NOTIFY 2533| | if (__itt_sync_create_ptr || KMP_ITT_DEBUG) { 2534| | if (!KMP_MASTER_TID(tid)) { 2535| | itt_sync_obj = __kmp_itt_barrier_object(gtid, bs_forkjoin_barrier); 2536| | if (itt_sync_obj) 2537| | __kmp_itt_barrier_finished(gtid, itt_sync_obj); 2538| | } 2539| | } 2540| |#endif /* USE_ITT_BUILD && USE_ITT_NOTIFY */ 2541| 31| KA_TRACE(10, ("__kmp_fork_barrier: T#%d is leaving early\n", gtid)); 2542| 31| return; 2543| 31| } 2544| | 2545| | /* We can now assume that a valid team structure has been allocated by the 2546| | primary thread and propagated to all worker threads. The current thread, 2547| | however, may not be part of the team, so we can't blindly assume that the 2548| | team pointer is non-null. */ 2549| 42.1k| team = (kmp_team_t *)TCR_PTR(this_thr->th.th_team); ------------------ | | 1156| 42.1k|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 42.1k|#define TCR_8(a) (a) | | ------------------ ------------------ 2550| 42.1k| KMP_DEBUG_ASSERT(team != NULL); 2551| 42.1k| tid = __kmp_tid_from_gtid(gtid); 2552| | 2553| |#if KMP_BARRIER_ICV_PULL 2554| | /* Primary thread's copy of the ICVs was set up on the implicit taskdata in 2555| | __kmp_reinitialize_team. __kmp_fork_call() assumes the primary thread's 2556| | implicit task has this data before this function is called. We cannot 2557| | modify __kmp_fork_call() to look at the fixed ICVs in the primary thread's 2558| | thread struct, because it is not always the case that the threads arrays 2559| | have been allocated when __kmp_fork_call() is executed. */ 2560| | { 2561| | KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(USER_icv_copy); 2562| | if (!KMP_MASTER_TID(tid)) { // primary thread already has ICVs 2563| | // Copy the initial ICVs from the primary thread's thread struct to the 2564| | // implicit task for this tid. 2565| | KA_TRACE(10, 2566| | ("__kmp_fork_barrier: T#%d(%d) is PULLing ICVs\n", gtid, tid)); 2567| | __kmp_init_implicit_task(team->t.t_ident, team->t.t_threads[tid], team, 2568| | tid, FALSE); 2569| | copy_icvs(&team->t.t_implicit_task_taskdata[tid].td_icvs, 2570| | &team->t.t_threads[0] 2571| | ->th.th_bar[bs_forkjoin_barrier] 2572| | .bb.th_fixed_icvs); 2573| | } 2574| | } 2575| |#endif // KMP_BARRIER_ICV_PULL 2576| | 2577| 42.2k| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (2577:7): [True: 42.2k, False: 18.4E] ------------------ 2578| 42.2k| __kmp_task_team_sync(this_thr, team); 2579| 42.2k| } 2580| | 2581| 42.1k|#if KMP_AFFINITY_SUPPORTED 2582| 42.1k| kmp_proc_bind_t proc_bind = team->t.t_proc_bind; 2583| 42.1k| if (proc_bind == proc_bind_intel) { ------------------ | Branch (2583:7): [True: 0, False: 42.1k] ------------------ 2584| | // Call dynamic affinity settings 2585| 0| if (__kmp_affinity.type == affinity_balanced && team->t.t_size_changed) { ------------------ | Branch (2585:9): [True: 0, False: 0] | Branch (2585:53): [True: 0, False: 0] ------------------ 2586| 0| __kmp_balanced_affinity(this_thr, team->t.t_nproc); 2587| 0| } 2588| 42.1k| } else if (proc_bind != proc_bind_false) { ------------------ | Branch (2588:14): [True: 0, False: 42.1k] ------------------ 2589| 0| if (this_thr->th.th_new_place == this_thr->th.th_current_place) { ------------------ | Branch (2589:9): [True: 0, False: 0] ------------------ 2590| 0| KA_TRACE(100, ("__kmp_fork_barrier: T#%d already in correct place %d\n", 2591| 0| __kmp_gtid_from_thread(this_thr), 2592| 0| this_thr->th.th_current_place)); 2593| 0| } else { 2594| 0| __kmp_affinity_set_place(gtid); 2595| 0| } 2596| 0| } 2597| 42.1k|#endif // KMP_AFFINITY_SUPPORTED 2598| | // Perform the display affinity functionality 2599| 42.1k| if (__kmp_display_affinity) { ------------------ | Branch (2599:7): [True: 0, False: 42.1k] ------------------ 2600| 0| if (team->t.t_display_affinity ------------------ | Branch (2600:9): [True: 0, False: 0] ------------------ 2601| 0|#if KMP_AFFINITY_SUPPORTED 2602| 0| || (__kmp_affinity.type == affinity_balanced && team->t.t_size_changed) ------------------ | Branch (2602:13): [True: 0, False: 0] | Branch (2602:57): [True: 0, False: 0] ------------------ 2603| 0|#endif 2604| 0| ) { 2605| | // NULL means use the affinity-format-var ICV 2606| 0| __kmp_aux_display_affinity(gtid, NULL); 2607| 0| this_thr->th.th_prev_num_threads = team->t.t_nproc; 2608| 0| this_thr->th.th_prev_level = team->t.t_level; 2609| 0| } 2610| 0| } 2611| 42.1k| if (!KMP_MASTER_TID(tid)) ------------------ | | 1271| 42.1k|#define KMP_MASTER_TID(tid) (0 == (tid)) ------------------ | Branch (2611:7): [True: 41.1k, False: 1.07k] ------------------ 2612| 41.1k| KMP_CHECK_UPDATE(this_thr->th.th_def_allocator, team->t.t_def_allocator); ------------------ | | 2280| 41.1k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 41.1k] | | ------------------ | | 2281| 41.1k| (a) = (b) ------------------ 2613| | 2614| |#if USE_ITT_BUILD && USE_ITT_NOTIFY 2615| | if (__itt_sync_create_ptr || KMP_ITT_DEBUG) { 2616| | if (!KMP_MASTER_TID(tid)) { 2617| | // Get correct barrier object 2618| | itt_sync_obj = __kmp_itt_barrier_object(gtid, bs_forkjoin_barrier); 2619| | __kmp_itt_barrier_finished(gtid, itt_sync_obj); // Workers call acquired 2620| | } // (prepare called inside barrier_release) 2621| | } 2622| |#endif /* USE_ITT_BUILD && USE_ITT_NOTIFY */ 2623| 42.1k| KA_TRACE(10, ("__kmp_fork_barrier: T#%d(%d:%d) is leaving\n", gtid, 2624| 42.1k| team->t.t_id, tid)); 2625| 42.1k|} _Z20__kmp_setup_icv_copyP8kmp_teamiP20kmp_internal_controlP5ident: 2628| 1.35k| kmp_internal_control_t *new_icvs, ident_t *loc) { 2629| 1.35k| KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(KMP_setup_icv_copy); ------------------ | | 1008| 1.35k|#define KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(n) ((void)0) ------------------ 2630| | 2631| 1.35k| KMP_DEBUG_ASSERT(team && new_nproc && new_icvs); 2632| 1.35k| KMP_DEBUG_ASSERT((!TCR_4(__kmp_init_parallel)) || new_icvs->nproc); 2633| | 2634| |/* Primary thread's copy of the ICVs was set up on the implicit taskdata in 2635| | __kmp_reinitialize_team. __kmp_fork_call() assumes the primary thread's 2636| | implicit task has this data before this function is called. */ 2637| |#if KMP_BARRIER_ICV_PULL 2638| | /* Copy ICVs to primary thread's thread structure into th_fixed_icvs (which 2639| | remains untouched), where all of the worker threads can access them and 2640| | make their own copies after the barrier. */ 2641| | KMP_DEBUG_ASSERT(team->t.t_threads[0]); // The threads arrays should be 2642| | // allocated at this point 2643| | copy_icvs( 2644| | &team->t.t_threads[0]->th.th_bar[bs_forkjoin_barrier].bb.th_fixed_icvs, 2645| | new_icvs); 2646| | KF_TRACE(10, ("__kmp_setup_icv_copy: PULL: T#%d this_thread=%p team=%p\n", 0, 2647| | team->t.t_threads[0], team)); 2648| |#elif KMP_BARRIER_ICV_PUSH 2649| | // The ICVs will be propagated in the fork barrier, so nothing needs to be 2650| | // done here. 2651| 1.35k| KF_TRACE(10, ("__kmp_setup_icv_copy: PUSH: T#%d this_thread=%p team=%p\n", 0, 2652| 1.35k| team->t.t_threads[0], team)); 2653| |#else 2654| | // Copy the ICVs to each of the non-primary threads. This takes O(nthreads) 2655| | // time. 2656| | ngo_load(new_icvs); 2657| | KMP_DEBUG_ASSERT(team->t.t_threads[0]); // The threads arrays should be 2658| | // allocated at this point 2659| | for (int f = 1; f < new_nproc; ++f) { // Skip the primary thread 2660| | // TODO: GEH - pass in better source location info since usually NULL here 2661| | KF_TRACE(10, ("__kmp_setup_icv_copy: LINEAR: T#%d this_thread=%p team=%p\n", 2662| | f, team->t.t_threads[f], team)); 2663| | __kmp_init_implicit_task(loc, team->t.t_threads[f], team, f, FALSE); 2664| | ngo_store_icvs(&team->t.t_implicit_task_taskdata[f].td_icvs, new_icvs); 2665| | KF_TRACE(10, ("__kmp_setup_icv_copy: LINEAR: T#%d this_thread=%p team=%p\n", 2666| | f, team->t.t_threads[f], team)); 2667| | } 2668| | ngo_sync(); 2669| |#endif // KMP_BARRIER_ICV_PULL 2670| 1.35k|} kmp_barrier.cpp:_ZL27__kmp_hyper_barrier_release12barrier_typeP8kmp_infoiii: 1127| 77.5k| int propagate_icvs USE_ITT_BUILD_ARG(void *itt_sync_obj)) { 1128| 77.5k| KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(KMP_hyper_release); ------------------ | | 1008| 77.5k|#define KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(n) ((void)0) ------------------ 1129| 77.5k| kmp_team_t *team; 1130| 77.5k| kmp_bstate_t *thr_bar = &this_thr->th.th_bar[bt].bb; 1131| 77.5k| kmp_info_t **other_threads; 1132| 77.5k| kmp_uint32 num_threads; 1133| 77.5k| kmp_uint32 branch_bits = __kmp_barrier_release_branch_bits[bt]; 1134| 77.5k| kmp_uint32 branch_factor = 1 << branch_bits; 1135| 77.5k| kmp_uint32 child; 1136| 77.5k| kmp_uint32 child_tid; 1137| 77.5k| kmp_uint32 offset; 1138| 77.5k| kmp_uint32 level; 1139| | 1140| | /* Perform a hypercube-embedded tree release for all of the threads that have 1141| | been gathered. If KMP_REVERSE_HYPER_BAR is defined (default) the threads 1142| | are released in the reverse order of the corresponding gather, otherwise 1143| | threads are released in the same order. */ 1144| 77.5k| if (KMP_MASTER_TID(tid)) { // primary thread ------------------ | | 1271| 77.5k|#define KMP_MASTER_TID(tid) (0 == (tid)) | | ------------------ | | | Branch (1271:29): [True: 2.42k, False: 75.0k] | | ------------------ ------------------ 1145| 2.42k| team = __kmp_threads[gtid]->th.th_team; 1146| 2.42k| KMP_DEBUG_ASSERT(team != NULL); 1147| 2.42k| KA_TRACE(20, ("__kmp_hyper_barrier_release: T#%d(%d:%d) primary enter for " 1148| 2.42k| "barrier type %d\n", 1149| 2.42k| gtid, team->t.t_id, tid, bt)); 1150| 2.42k|#if KMP_BARRIER_ICV_PUSH 1151| 2.42k| if (propagate_icvs) { // primary already has ICVs in final destination; copy ------------------ | Branch (1151:9): [True: 1.35k, False: 1.07k] ------------------ 1152| 1.35k| copy_icvs(&thr_bar->th_fixed_icvs, 1153| 1.35k| &team->t.t_implicit_task_taskdata[tid].td_icvs); 1154| 1.35k| } 1155| 2.42k|#endif 1156| 75.0k| } else { // Handle fork barrier workers who aren't part of a team yet 1157| 75.0k| KA_TRACE(20, ("__kmp_hyper_barrier_release: T#%d wait go(%p) == %u\n", gtid, 1158| 75.0k| &thr_bar->b_go, KMP_BARRIER_STATE_BUMP)); 1159| | // Wait for parent thread to release us 1160| 75.0k| kmp_flag_64<> flag(&thr_bar->b_go, KMP_BARRIER_STATE_BUMP); ------------------ | | 2025| 75.0k|#define KMP_BARRIER_STATE_BUMP (1 << KMP_BARRIER_BUMP_BIT) | | ------------------ | | | | 2021| 75.0k|#define KMP_BARRIER_BUMP_BIT 2 /* lsb used for bump of go/arrived state */ | | ------------------ ------------------ 1161| 75.0k| flag.wait(this_thr, TRUE USE_ITT_BUILD_ARG(itt_sync_obj)); ------------------ | | 1280| 75.0k|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 75.0k|#define FALSE 0 | | ------------------ ------------------ 1162| |#if USE_ITT_BUILD && USE_ITT_NOTIFY 1163| | if ((__itt_sync_create_ptr && itt_sync_obj == NULL) || KMP_ITT_DEBUG) { 1164| | // In fork barrier where we could not get the object reliably 1165| | itt_sync_obj = __kmp_itt_barrier_object(gtid, bs_forkjoin_barrier, 0, -1); 1166| | // Cancel wait on previous parallel region... 1167| | __kmp_itt_task_starting(itt_sync_obj); 1168| | 1169| | if (bt == bs_forkjoin_barrier && TCR_4(__kmp_global.g.g_done)) 1170| | return; 1171| | 1172| | itt_sync_obj = __kmp_itt_barrier_object(gtid, bs_forkjoin_barrier); 1173| | if (itt_sync_obj != NULL) 1174| | // Call prepare as early as possible for "new" barrier 1175| | __kmp_itt_task_finished(itt_sync_obj); 1176| | } else 1177| |#endif /* USE_ITT_BUILD && USE_ITT_NOTIFY */ 1178| | // Early exit for reaping threads releasing forkjoin barrier 1179| 75.0k| if (bt == bs_forkjoin_barrier && TCR_4(__kmp_global.g.g_done)) ------------------ | | 1127| 41.7k|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 31, False: 41.7k] | | ------------------ ------------------ | Branch (1179:13): [True: 41.7k, False: 33.3k] ------------------ 1180| 31| return; 1181| | 1182| | // The worker thread may now assume that the team is valid. 1183| 75.0k| team = __kmp_threads[gtid]->th.th_team; 1184| 75.0k| KMP_DEBUG_ASSERT(team != NULL); 1185| 75.0k| tid = __kmp_tid_from_gtid(gtid); 1186| | 1187| 75.0k| TCW_4(thr_bar->b_go, KMP_INIT_BARRIER_STATE); ------------------ | | 1128| 75.0k|#define TCW_4(a, b) (a) = (b) ------------------ 1188| 75.0k| KA_TRACE(20, 1189| 75.0k| ("__kmp_hyper_barrier_release: T#%d(%d:%d) set go(%p) = %u\n", 1190| 75.0k| gtid, team->t.t_id, tid, &thr_bar->b_go, KMP_INIT_BARRIER_STATE)); 1191| 75.0k| KMP_MB(); // Flush all pending memory write invalidates. 1192| 75.0k| } 1193| 77.4k| num_threads = this_thr->th.th_team_nproc; 1194| 77.4k| other_threads = team->t.t_threads; 1195| | 1196| 77.4k|#ifdef KMP_REVERSE_HYPER_BAR 1197| | // Count up to correct level for parent 1198| 77.4k| for (level = 0, offset = 1; 1199| 103k| offset < num_threads && (((tid >> level) & (branch_factor - 1)) == 0); ------------------ | Branch (1199:8): [True: 100k, False: 3.13k] | Branch (1199:32): [True: 26.5k, False: 74.2k] ------------------ 1200| 77.4k| level += branch_bits, offset <<= branch_bits) 1201| 26.4k| ; 1202| | 1203| | // Now go down from there 1204| 103k| for (level -= branch_bits, offset >>= branch_bits; offset != 0; ------------------ | Branch (1204:54): [True: 26.0k, False: 77.4k] ------------------ 1205| 77.4k| level -= branch_bits, offset >>= branch_bits) 1206| |#else 1207| | // Go down the tree, level by level 1208| | for (level = 0, offset = 1; offset < num_threads; 1209| | level += branch_bits, offset <<= branch_bits) 1210| |#endif // KMP_REVERSE_HYPER_BAR 1211| 26.0k| { 1212| 26.0k|#ifdef KMP_REVERSE_HYPER_BAR 1213| | /* Now go in reverse order through the children, highest to lowest. 1214| | Initial setting of child is conservative here. */ 1215| 26.0k| child = num_threads >> ((level == 0) ? level : level - 1); ------------------ | Branch (1215:29): [True: 18.9k, False: 7.09k] ------------------ 1216| 26.0k| for (child = (child < branch_factor - 1) ? child : branch_factor - 1, ------------------ | Branch (1216:18): [True: 0, False: 26.0k] ------------------ 1217| 26.0k| child_tid = tid + (child << level); 1218| 103k| child >= 1; child--, child_tid -= (1 << level)) ------------------ | Branch (1218:10): [True: 77.8k, False: 26.0k] ------------------ 1219| |#else 1220| | if (((tid >> level) & (branch_factor - 1)) != 0) 1221| | // No need to go lower than this, since this is the level parent would be 1222| | // notified 1223| | break; 1224| | // Iterate through children on this level of the tree 1225| | for (child = 1, child_tid = tid + (1 << level); 1226| | child < branch_factor && child_tid < num_threads; 1227| | child++, child_tid += (1 << level)) 1228| |#endif // KMP_REVERSE_HYPER_BAR 1229| 77.8k| { 1230| 77.8k| if (child_tid >= num_threads) ------------------ | Branch (1230:11): [True: 4.85k, False: 73.0k] ------------------ 1231| 4.85k| continue; // Child doesn't exist so keep going 1232| 73.0k| else { 1233| 73.0k| kmp_info_t *child_thr = other_threads[child_tid]; 1234| 73.0k| kmp_bstate_t *child_bar = &child_thr->th.th_bar[bt].bb; 1235| |#if KMP_CACHE_MANAGE 1236| | kmp_uint32 next_child_tid = child_tid - (1 << level); 1237| |// Prefetch next thread's go count 1238| |#ifdef KMP_REVERSE_HYPER_BAR 1239| | if (child - 1 >= 1 && next_child_tid < num_threads) 1240| |#else 1241| | if (child + 1 < branch_factor && next_child_tid < num_threads) 1242| |#endif // KMP_REVERSE_HYPER_BAR 1243| | KMP_CACHE_PREFETCH( 1244| | &other_threads[next_child_tid]->th.th_bar[bt].bb.b_go); 1245| |#endif /* KMP_CACHE_MANAGE */ 1246| | 1247| 73.0k|#if KMP_BARRIER_ICV_PUSH 1248| 73.0k| if (propagate_icvs) // push my fixed ICVs to my child ------------------ | Branch (1248:13): [True: 40.7k, False: 32.2k] ------------------ 1249| 40.7k| copy_icvs(&child_bar->th_fixed_icvs, &thr_bar->th_fixed_icvs); 1250| 73.0k|#endif // KMP_BARRIER_ICV_PUSH 1251| | 1252| 73.0k| KA_TRACE( 1253| 73.0k| 20, 1254| 73.0k| ("__kmp_hyper_barrier_release: T#%d(%d:%d) releasing T#%d(%d:%u)" 1255| 73.0k| "go(%p): %u => %u\n", 1256| 73.0k| gtid, team->t.t_id, tid, __kmp_gtid_from_tid(child_tid, team), 1257| 73.0k| team->t.t_id, child_tid, &child_bar->b_go, child_bar->b_go, 1258| 73.0k| child_bar->b_go + KMP_BARRIER_STATE_BUMP)); 1259| | // Release child from barrier 1260| 73.0k| kmp_flag_64<> flag(&child_bar->b_go, child_thr); 1261| 73.0k| flag.release(); 1262| 73.0k| } 1263| 77.8k| } 1264| 26.0k| } 1265| 77.4k|#if KMP_BARRIER_ICV_PUSH 1266| 77.4k| if (propagate_icvs && ------------------ | Branch (1266:7): [True: 42.5k, False: 34.9k] ------------------ 1267| 42.5k| !KMP_MASTER_TID(tid)) { // copy ICVs locally to final dest ------------------ | | 1271| 42.5k|#define KMP_MASTER_TID(tid) (0 == (tid)) ------------------ | Branch (1267:7): [True: 41.3k, False: 1.26k] ------------------ 1268| 41.1k| __kmp_init_implicit_task(team->t.t_ident, team->t.t_threads[tid], team, tid, 1269| 41.1k| FALSE); ------------------ | | 1279| 41.1k|#define FALSE 0 ------------------ 1270| 41.1k| copy_icvs(&team->t.t_implicit_task_taskdata[tid].td_icvs, 1271| 41.1k| &thr_bar->th_fixed_icvs); 1272| 41.1k| } 1273| 77.4k|#endif 1274| 77.4k| KA_TRACE( 1275| 77.4k| 20, 1276| 77.4k| ("__kmp_hyper_barrier_release: T#%d(%d:%d) exit for barrier type %d\n", 1277| 77.4k| gtid, team->t.t_id, tid, bt)); 1278| 77.4k|} kmp_barrier.cpp:_ZL26__kmp_hyper_barrier_gather12barrier_typeP8kmp_infoiiPFvPvS2_E: 1008| 76.5k| void (*reduce)(void *, void *) USE_ITT_BUILD_ARG(void *itt_sync_obj)) { 1009| 76.5k| KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(KMP_hyper_gather); ------------------ | | 1008| 76.5k|#define KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(n) ((void)0) ------------------ 1010| 76.5k| kmp_team_t *team = this_thr->th.th_team; 1011| 76.5k| kmp_bstate_t *thr_bar = &this_thr->th.th_bar[bt].bb; 1012| 76.5k| kmp_info_t **other_threads = team->t.t_threads; 1013| 76.5k| kmp_uint64 new_state = KMP_BARRIER_UNUSED_STATE; ------------------ | | 2024| 76.5k|#define KMP_BARRIER_UNUSED_STATE (1 << KMP_BARRIER_UNUSED_BIT) | | ------------------ | | | | 2020| 76.5k|#define KMP_BARRIER_UNUSED_BIT 1 // bit that must never be set for valid state | | ------------------ ------------------ 1014| 76.5k| kmp_uint32 num_threads = this_thr->th.th_team_nproc; 1015| 76.5k| kmp_uint32 branch_bits = __kmp_barrier_gather_branch_bits[bt]; 1016| 76.5k| kmp_uint32 branch_factor = 1 << branch_bits; 1017| 76.5k| kmp_uint32 offset; 1018| 76.5k| kmp_uint32 level; 1019| | 1020| 76.5k| KA_TRACE( 1021| 76.5k| 20, 1022| 76.5k| ("__kmp_hyper_barrier_gather: T#%d(%d:%d) enter for barrier type %d\n", 1023| 76.5k| gtid, team->t.t_id, tid, bt)); 1024| 76.5k| KMP_DEBUG_ASSERT(this_thr == other_threads[this_thr->th.th_info.ds.ds_tid]); 1025| | 1026| |#if USE_ITT_BUILD && USE_ITT_NOTIFY 1027| | // Barrier imbalance - save arrive time to the thread 1028| | if (__kmp_forkjoin_frames_mode == 3 || __kmp_forkjoin_frames_mode == 2) { 1029| | this_thr->th.th_bar_arrive_time = this_thr->th.th_bar_min_time = 1030| | __itt_get_timestamp(); 1031| | } 1032| |#endif 1033| | /* Perform a hypercube-embedded tree gather to wait until all of the threads 1034| | have arrived, and reduce any required data as we go. */ 1035| 76.5k| kmp_flag_64<> p_flag(&thr_bar->b_arrived); 1036| 102k| for (level = 0, offset = 1; offset < num_threads; ------------------ | Branch (1036:31): [True: 100k, False: 1.52k] ------------------ 1037| 100k| level += branch_bits, offset <<= branch_bits) { 1038| 100k| kmp_uint32 child; 1039| 100k| kmp_uint32 child_tid; 1040| | 1041| 100k| if (((tid >> level) & (branch_factor - 1)) != 0) { ------------------ | Branch (1041:9): [True: 75.0k, False: 25.6k] ------------------ 1042| 75.0k| kmp_int32 parent_tid = tid & ~((1 << (level + branch_bits)) - 1); 1043| | 1044| 75.0k| KMP_MB(); // Synchronize parent and child threads. 1045| 75.0k| KA_TRACE(20, 1046| 75.0k| ("__kmp_hyper_barrier_gather: T#%d(%d:%d) releasing T#%d(%d:%d) " 1047| 75.0k| "arrived(%p): %llu => %llu\n", 1048| 75.0k| gtid, team->t.t_id, tid, __kmp_gtid_from_tid(parent_tid, team), 1049| 75.0k| team->t.t_id, parent_tid, &thr_bar->b_arrived, 1050| 75.0k| thr_bar->b_arrived, 1051| 75.0k| thr_bar->b_arrived + KMP_BARRIER_STATE_BUMP)); 1052| | // Mark arrival to parent thread 1053| | /* After performing this write (in the last iteration of the enclosing for 1054| | loop), a worker thread may not assume that the team is valid any more 1055| | - it could be deallocated by the primary thread at any time. */ 1056| 75.0k| p_flag.set_waiter(other_threads[parent_tid]); 1057| 75.0k| p_flag.release(); 1058| 75.0k| break; 1059| 75.0k| } 1060| | 1061| | // Parent threads wait for children to arrive 1062| 25.6k| if (new_state == KMP_BARRIER_UNUSED_STATE) ------------------ | | 2024| 25.6k|#define KMP_BARRIER_UNUSED_STATE (1 << KMP_BARRIER_UNUSED_BIT) | | ------------------ | | | | 2020| 25.6k|#define KMP_BARRIER_UNUSED_BIT 1 // bit that must never be set for valid state | | ------------------ ------------------ | Branch (1062:9): [True: 19.1k, False: 6.47k] ------------------ 1063| 19.1k| new_state = team->t.t_bar[bt].b_arrived + KMP_BARRIER_STATE_BUMP; ------------------ | | 2025| 19.1k|#define KMP_BARRIER_STATE_BUMP (1 << KMP_BARRIER_BUMP_BIT) | | ------------------ | | | | 2021| 19.1k|#define KMP_BARRIER_BUMP_BIT 2 /* lsb used for bump of go/arrived state */ | | ------------------ ------------------ 1064| 25.6k| for (child = 1, child_tid = tid + (1 << level); 1065| 100k| child < branch_factor && child_tid < num_threads; ------------------ | Branch (1065:10): [True: 77.4k, False: 23.2k] | Branch (1065:35): [True: 74.9k, False: 2.42k] ------------------ 1066| 74.9k| child++, child_tid += (1 << level)) { 1067| 74.9k| kmp_info_t *child_thr = other_threads[child_tid]; 1068| 74.9k| kmp_bstate_t *child_bar = &child_thr->th.th_bar[bt].bb; 1069| |#if KMP_CACHE_MANAGE 1070| | kmp_uint32 next_child_tid = child_tid + (1 << level); 1071| | // Prefetch next thread's arrived count 1072| | if (child + 1 < branch_factor && next_child_tid < num_threads) 1073| | KMP_CACHE_PREFETCH( 1074| | &other_threads[next_child_tid]->th.th_bar[bt].bb.b_arrived); 1075| |#endif /* KMP_CACHE_MANAGE */ 1076| 74.9k| KA_TRACE(20, 1077| 74.9k| ("__kmp_hyper_barrier_gather: T#%d(%d:%d) wait T#%d(%d:%u) " 1078| 74.9k| "arrived(%p) == %llu\n", 1079| 74.9k| gtid, team->t.t_id, tid, __kmp_gtid_from_tid(child_tid, team), 1080| 74.9k| team->t.t_id, child_tid, &child_bar->b_arrived, new_state)); 1081| | // Wait for child to arrive 1082| 74.9k| kmp_flag_64<> c_flag(&child_bar->b_arrived, new_state); 1083| 74.9k| c_flag.wait(this_thr, FALSE USE_ITT_BUILD_ARG(itt_sync_obj)); ------------------ | | 1279| 74.9k|#define FALSE 0 ------------------ 1084| 74.9k| KMP_MB(); // Synchronize parent and child threads. 1085| |#if USE_ITT_BUILD && USE_ITT_NOTIFY 1086| | // Barrier imbalance - write min of the thread time and a child time to 1087| | // the thread. 1088| | if (__kmp_forkjoin_frames_mode == 2) { 1089| | this_thr->th.th_bar_min_time = KMP_MIN(this_thr->th.th_bar_min_time, 1090| | child_thr->th.th_bar_min_time); 1091| | } 1092| |#endif 1093| 74.9k| if (reduce) { ------------------ | Branch (1093:11): [True: 0, False: 74.9k] ------------------ 1094| 0| KA_TRACE(100, 1095| 0| ("__kmp_hyper_barrier_gather: T#%d(%d:%d) += T#%d(%d:%u)\n", 1096| 0| gtid, team->t.t_id, tid, __kmp_gtid_from_tid(child_tid, team), 1097| 0| team->t.t_id, child_tid)); 1098| 0| OMPT_REDUCTION_DECL(this_thr, gtid); ------------------ | | 157| 0| ompt_data_t *my_task_data = OMPT_CUR_TASK_DATA(this_thr); \ | | ------------------ | | | | 70| 0| (&((thr)->th.th_current_task->ompt_task_info.task_data)) | | ------------------ | | 158| 0| ompt_data_t *my_parallel_data = OMPT_CUR_TEAM_DATA(this_thr); \ | | ------------------ | | | | 73| 0| (&((thr)->th.th_team->t.ompt_team_info.parallel_data)) | | ------------------ | | 159| 0| void *return_address = OMPT_LOAD_RETURN_ADDRESS(gtid); | | ------------------ | | | | 93| 0|#define OMPT_LOAD_RETURN_ADDRESS(gtid) __ompt_load_return_address(gtid) | | ------------------ ------------------ 1099| 0| OMPT_REDUCTION_BEGIN; ------------------ | | 161| 0| if (ompt_enabled.enabled && ompt_enabled.ompt_callback_reduction) { \ | | ------------------ | | | Branch (161:7): [True: 0, False: 0] | | | Branch (161:31): [True: 0, False: 0] | | ------------------ | | 162| 0| ompt_callbacks.ompt_callback(ompt_callback_reduction)( \ | | ------------------ | | | | 27| 0|#define ompt_callback(e) e##_callback | | ------------------ | | 163| 0| ompt_sync_region_reduction, ompt_scope_begin, my_parallel_data, \ | | 164| 0| my_task_data, return_address); \ | | 165| 0| } ------------------ 1100| 0| (*reduce)(this_thr->th.th_local.reduce_data, 1101| 0| child_thr->th.th_local.reduce_data); 1102| 0| OMPT_REDUCTION_END; ------------------ | | 167| 0| if (ompt_enabled.enabled && ompt_enabled.ompt_callback_reduction) { \ | | ------------------ | | | Branch (167:7): [True: 0, False: 0] | | | Branch (167:31): [True: 0, False: 0] | | ------------------ | | 168| 0| ompt_callbacks.ompt_callback(ompt_callback_reduction)( \ | | ------------------ | | | | 27| 0|#define ompt_callback(e) e##_callback | | ------------------ | | 169| 0| ompt_sync_region_reduction, ompt_scope_end, my_parallel_data, \ | | 170| 0| my_task_data, return_address); \ | | 171| 0| } ------------------ 1103| 0| } 1104| 74.9k| } 1105| 25.6k| } 1106| | 1107| 76.5k| if (KMP_MASTER_TID(tid)) { ------------------ | | 1271| 76.5k|#define KMP_MASTER_TID(tid) (0 == (tid)) | | ------------------ | | | Branch (1271:29): [True: 2.42k, False: 74.0k] | | ------------------ ------------------ 1108| | // Need to update the team arrived pointer if we are the primary thread 1109| 2.42k| if (new_state == KMP_BARRIER_UNUSED_STATE) ------------------ | | 2024| 2.42k|#define KMP_BARRIER_UNUSED_STATE (1 << KMP_BARRIER_UNUSED_BIT) | | ------------------ | | | | 2020| 2.42k|#define KMP_BARRIER_UNUSED_BIT 1 // bit that must never be set for valid state | | ------------------ ------------------ | Branch (1109:9): [True: 0, False: 2.42k] ------------------ 1110| 0| team->t.t_bar[bt].b_arrived += KMP_BARRIER_STATE_BUMP; ------------------ | | 2025| 0|#define KMP_BARRIER_STATE_BUMP (1 << KMP_BARRIER_BUMP_BIT) | | ------------------ | | | | 2021| 0|#define KMP_BARRIER_BUMP_BIT 2 /* lsb used for bump of go/arrived state */ | | ------------------ ------------------ 1111| 2.42k| else 1112| 2.42k| team->t.t_bar[bt].b_arrived = new_state; 1113| 2.42k| KA_TRACE(20, ("__kmp_hyper_barrier_gather: T#%d(%d:%d) set team %d " 1114| 2.42k| "arrived(%p) = %llu\n", 1115| 2.42k| gtid, team->t.t_id, tid, team->t.t_id, 1116| 2.42k| &team->t.t_bar[bt].b_arrived, team->t.t_bar[bt].b_arrived)); 1117| 2.42k| } 1118| 76.5k| KA_TRACE( 1119| 76.5k| 20, ("__kmp_hyper_barrier_gather: T#%d(%d:%d) exit for barrier type %d\n", 1120| 76.5k| gtid, team->t.t_id, tid, bt)); 1121| 76.5k|} kmp_barrier.cpp:_ZL22__kmp_barrier_templateILb0EEi12barrier_typeiimPvPFvS1_S1_E: 1769| 37.1k| void (*reduce)(void *, void *)) { 1770| 37.1k| KMP_TIME_PARTITIONED_BLOCK(OMP_plain_barrier); ------------------ | | 1013| 37.1k|#define KMP_TIME_PARTITIONED_BLOCK(name) ((void)0) ------------------ 1771| 37.1k| KMP_SET_THREAD_STATE_BLOCK(PLAIN_BARRIER); ------------------ | | 1018| 37.1k|#define KMP_SET_THREAD_STATE_BLOCK(state_name) ((void)0) ------------------ 1772| 37.1k| int tid = __kmp_tid_from_gtid(gtid); 1773| 37.1k| kmp_info_t *this_thr = __kmp_threads[gtid]; 1774| 37.1k| kmp_team_t *team = this_thr->th.th_team; 1775| 37.1k| int status = 0; 1776| 37.1k| is_cancellable cancelled; 1777| 37.1k|#if OMPT_SUPPORT && OMPT_OPTIONAL 1778| 37.1k| ompt_data_t *my_task_data; 1779| 37.1k| ompt_data_t *my_parallel_data; 1780| 37.1k| void *return_address; 1781| 37.1k| ompt_sync_region_t barrier_kind; 1782| 37.1k|#endif 1783| | 1784| 37.1k| KA_TRACE(15, ("__kmp_barrier: T#%d(%d:%d) has arrived\n", gtid, 1785| 37.1k| __kmp_team_from_gtid(gtid)->t.t_id, __kmp_tid_from_gtid(gtid))); 1786| | 1787| 37.1k|#if OMPT_SUPPORT 1788| 37.1k| if (ompt_enabled.enabled) { ------------------ | Branch (1788:7): [True: 0, False: 37.1k] ------------------ 1789| 0|#if OMPT_OPTIONAL 1790| 0| my_task_data = OMPT_CUR_TASK_DATA(this_thr); ------------------ | | 70| 0| (&((thr)->th.th_current_task->ompt_task_info.task_data)) ------------------ 1791| 0| my_parallel_data = OMPT_CUR_TEAM_DATA(this_thr); ------------------ | | 73| 0| (&((thr)->th.th_team->t.ompt_team_info.parallel_data)) ------------------ 1792| 0| return_address = OMPT_LOAD_RETURN_ADDRESS(gtid); ------------------ | | 93| 0|#define OMPT_LOAD_RETURN_ADDRESS(gtid) __ompt_load_return_address(gtid) ------------------ 1793| 0| barrier_kind = __ompt_get_barrier_kind(bt, this_thr); 1794| 0| if (ompt_enabled.ompt_callback_sync_region) { ------------------ | Branch (1794:9): [True: 0, False: 0] ------------------ 1795| 0| ompt_callbacks.ompt_callback(ompt_callback_sync_region)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 1796| 0| barrier_kind, ompt_scope_begin, my_parallel_data, my_task_data, 1797| 0| return_address); 1798| 0| } 1799| 0| if (ompt_enabled.ompt_callback_sync_region_wait) { ------------------ | Branch (1799:9): [True: 0, False: 0] ------------------ 1800| 0| ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 1801| 0| barrier_kind, ompt_scope_begin, my_parallel_data, my_task_data, 1802| 0| return_address); 1803| 0| } 1804| 0|#endif 1805| | // It is OK to report the barrier state after the barrier begin callback. 1806| | // According to the OMPT specification, a compliant implementation may 1807| | // even delay reporting this state until the barrier begins to wait. 1808| 0| this_thr->th.ompt_thread_info.state = ompt_state_wait_barrier; 1809| 0| } 1810| 37.1k|#endif 1811| | 1812| 37.1k| if (!team->t.t_serialized) { ------------------ | Branch (1812:7): [True: 34.1k, False: 2.99k] ------------------ 1813| |#if USE_ITT_BUILD 1814| | // This value will be used in itt notify events below. 1815| | void *itt_sync_obj = NULL; 1816| |#if USE_ITT_NOTIFY 1817| | if (__itt_sync_create_ptr || KMP_ITT_DEBUG) 1818| | itt_sync_obj = __kmp_itt_barrier_object(gtid, bt, 1); 1819| |#endif 1820| |#endif /* USE_ITT_BUILD */ 1821| 34.1k| if (__kmp_tasking_mode == tskm_extra_barrier) { ------------------ | Branch (1821:9): [True: 0, False: 34.1k] ------------------ 1822| 0| __kmp_tasking_barrier(team, this_thr, gtid); 1823| 0| KA_TRACE(15, 1824| 0| ("__kmp_barrier: T#%d(%d:%d) past tasking barrier\n", gtid, 1825| 0| __kmp_team_from_gtid(gtid)->t.t_id, __kmp_tid_from_gtid(gtid))); 1826| 0| } 1827| | 1828| | /* Copy the blocktime info to the thread, where __kmp_wait_template() can 1829| | access it when the team struct is not guaranteed to exist. */ 1830| | // See note about the corresponding code in __kmp_join_barrier() being 1831| | // performance-critical. 1832| 34.1k| if (__kmp_dflt_blocktime != KMP_MAX_BLOCKTIME) { ------------------ | | 1184| 34.1k| (INT_MAX) /* Must be this for "infinite" setting the work */ ------------------ | Branch (1832:9): [True: 34.1k, False: 16] ------------------ 1833| |#if KMP_USE_MONITOR 1834| | this_thr->th.th_team_bt_intervals = 1835| | team->t.t_implicit_task_taskdata[tid].td_icvs.bt_intervals; 1836| | this_thr->th.th_team_bt_set = 1837| | team->t.t_implicit_task_taskdata[tid].td_icvs.bt_set; 1838| |#else 1839| 34.1k| this_thr->th.th_team_bt_intervals = KMP_BLOCKTIME_INTERVAL(team, tid); ------------------ | | 1221| 34.1k| (KMP_BLOCKTIME(team, tid) * __kmp_ticks_per_msec) | | ------------------ | | | | 1210| 34.1k| (get__bt_set(team, tid) ? get__blocktime(team, tid) : __kmp_dflt_blocktime) | | | | ------------------ | | | | | | 2289| 34.1k| ((xteam)->t.t_threads[(xtid)]->th.th_current_task->td_icvs.bt_set) | | | | | | ------------------ | | | | | | | Branch (2289:3): [True: 0, False: 34.1k] | | | | | | ------------------ | | | | ------------------ | | | | (get__bt_set(team, tid) ? get__blocktime(team, tid) : __kmp_dflt_blocktime) | | | | ------------------ | | | | | | 2287| 0| ((xteam)->t.t_threads[(xtid)]->th.th_current_task->td_icvs.blocktime) | | | | ------------------ | | ------------------ ------------------ 1840| 34.1k|#endif 1841| 34.1k| } 1842| | 1843| |#if USE_ITT_BUILD 1844| | if (__itt_sync_create_ptr || KMP_ITT_DEBUG) 1845| | __kmp_itt_barrier_starting(gtid, itt_sync_obj); 1846| |#endif /* USE_ITT_BUILD */ 1847| |#if USE_DEBUGGER 1848| | // Let the debugger know: the thread arrived to the barrier and waiting. 1849| | if (KMP_MASTER_TID(tid)) { // Primary thread counter stored in team struct 1850| | team->t.t_bar[bt].b_master_arrived += 1; 1851| | } else { 1852| | this_thr->th.th_bar[bt].bb.b_worker_arrived += 1; 1853| | } // if 1854| |#endif /* USE_DEBUGGER */ 1855| 34.1k| if (reduce != NULL) { ------------------ | Branch (1855:9): [True: 0, False: 34.1k] ------------------ 1856| | // KMP_DEBUG_ASSERT( is_split == TRUE ); // #C69956 1857| 0| this_thr->th.th_local.reduce_data = reduce_data; 1858| 0| } 1859| | 1860| 34.1k| if (KMP_MASTER_TID(tid) && __kmp_tasking_mode != tskm_immediate_exec) ------------------ | | 1271| 68.2k|#define KMP_MASTER_TID(tid) (0 == (tid)) | | ------------------ | | | Branch (1271:29): [True: 1.07k, False: 33.0k] | | ------------------ ------------------ | Branch (1860:32): [True: 1.07k, False: 0] ------------------ 1861| | // use 0 to only setup the current team if nthreads > 1 1862| 1.07k| __kmp_task_team_setup(this_thr, team, 0); 1863| | 1864| 34.1k| if (cancellable) { ------------------ | Branch (1864:9): [Folded, False: 34.1k] ------------------ 1865| 0| cancelled = __kmp_linear_barrier_gather_cancellable( 1866| 0| bt, this_thr, gtid, tid, reduce USE_ITT_BUILD_ARG(itt_sync_obj)); 1867| 34.1k| } else { 1868| 34.1k| switch (__kmp_barrier_gather_pattern[bt]) { 1869| 0| case bp_dist_bar: { ------------------ | Branch (1869:7): [True: 0, False: 34.1k] ------------------ 1870| 0| __kmp_dist_barrier_gather(bt, this_thr, gtid, tid, 1871| 0| reduce USE_ITT_BUILD_ARG(itt_sync_obj)); 1872| 0| break; 1873| 0| } 1874| 34.1k| case bp_hyper_bar: { ------------------ | Branch (1874:7): [True: 34.1k, False: 3] ------------------ 1875| | // don't set branch bits to 0; use linear 1876| 34.1k| KMP_ASSERT(__kmp_barrier_gather_branch_bits[bt]); ------------------ | | 50| 34.1k| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 34.1k] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 1877| 34.1k| __kmp_hyper_barrier_gather(bt, this_thr, gtid, tid, 1878| 34.1k| reduce USE_ITT_BUILD_ARG(itt_sync_obj)); 1879| 34.1k| break; 1880| 0| } 1881| 0| case bp_hierarchical_bar: { ------------------ | Branch (1881:7): [True: 0, False: 34.1k] ------------------ 1882| 0| __kmp_hierarchical_barrier_gather( 1883| 0| bt, this_thr, gtid, tid, reduce USE_ITT_BUILD_ARG(itt_sync_obj)); 1884| 0| break; 1885| 0| } 1886| 0| case bp_tree_bar: { ------------------ | Branch (1886:7): [True: 0, False: 34.1k] ------------------ 1887| | // don't set branch bits to 0; use linear 1888| 0| KMP_ASSERT(__kmp_barrier_gather_branch_bits[bt]); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 1889| 0| __kmp_tree_barrier_gather(bt, this_thr, gtid, tid, 1890| 0| reduce USE_ITT_BUILD_ARG(itt_sync_obj)); 1891| 0| break; 1892| 0| } 1893| 0| default: { ------------------ | Branch (1893:7): [True: 0, False: 34.1k] ------------------ 1894| 0| __kmp_linear_barrier_gather(bt, this_thr, gtid, tid, 1895| 0| reduce USE_ITT_BUILD_ARG(itt_sync_obj)); 1896| 0| } 1897| 34.1k| } 1898| 34.1k| } 1899| | 1900| 34.3k| KMP_MB(); 1901| | 1902| 34.3k| if (KMP_MASTER_TID(tid)) { ------------------ | | 1271| 34.3k|#define KMP_MASTER_TID(tid) (0 == (tid)) | | ------------------ | | | Branch (1271:29): [True: 1.07k, False: 33.2k] | | ------------------ ------------------ 1903| 1.07k| status = 0; 1904| 1.07k| if (__kmp_tasking_mode != tskm_immediate_exec && !cancelled) { ------------------ | Branch (1904:11): [True: 1.07k, False: 0] | Branch (1904:56): [True: 0, Folded] ------------------ 1905| 1.07k| __kmp_task_team_wait(this_thr, team USE_ITT_BUILD_ARG(itt_sync_obj)); 1906| 1.07k| } 1907| |#if USE_DEBUGGER 1908| | // Let the debugger know: All threads are arrived and starting leaving the 1909| | // barrier. 1910| | team->t.t_bar[bt].b_team_arrived += 1; 1911| |#endif 1912| | 1913| 1.07k| if (__kmp_omp_cancellation) { ------------------ | Branch (1913:11): [True: 0, False: 1.07k] ------------------ 1914| 0| kmp_int32 cancel_request = KMP_ATOMIC_LD_RLX(&team->t.t_cancel_request); ------------------ | | 1250| 0|#define KMP_ATOMIC_LD_RLX(p) KMP_ATOMIC_LD(p, relaxed) | | ------------------ | | | | 1245| 0|#define KMP_ATOMIC_LD(p, order) (p)->load(std::memory_order_##order) | | ------------------ ------------------ 1915| | // Reset cancellation flag for worksharing constructs 1916| 0| if (cancel_request == cancel_loop || ------------------ | Branch (1916:13): [True: 0, False: 0] ------------------ 1917| 0| cancel_request == cancel_sections) { ------------------ | Branch (1917:13): [True: 0, False: 0] ------------------ 1918| 0| KMP_ATOMIC_ST_RLX(&team->t.t_cancel_request, cancel_noreq); ------------------ | | 1252| 0|#define KMP_ATOMIC_ST_RLX(p, v) KMP_ATOMIC_OP(store, p, v, relaxed) | | ------------------ | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 1919| 0| } 1920| 0| } 1921| |#if USE_ITT_BUILD 1922| | /* TODO: In case of split reduction barrier, primary thread may send 1923| | acquired event early, before the final summation into the shared 1924| | variable is done (final summation can be a long operation for array 1925| | reductions). */ 1926| | if (__itt_sync_create_ptr || KMP_ITT_DEBUG) 1927| | __kmp_itt_barrier_middle(gtid, itt_sync_obj); 1928| |#endif /* USE_ITT_BUILD */ 1929| |#if USE_ITT_BUILD && USE_ITT_NOTIFY 1930| | // Barrier - report frame end (only if active_level == 1) 1931| | if ((__itt_frame_submit_v3_ptr || KMP_ITT_DEBUG) && 1932| | __kmp_forkjoin_frames_mode && 1933| | (this_thr->th.th_teams_microtask == NULL || // either not in teams 1934| | this_thr->th.th_teams_size.nteams == 1) && // or inside single team 1935| | team->t.t_active_level == 1) { 1936| | ident_t *loc = __kmp_threads[gtid]->th.th_ident; 1937| | kmp_uint64 cur_time = __itt_get_timestamp(); 1938| | kmp_info_t **other_threads = team->t.t_threads; 1939| | int nproc = this_thr->th.th_team_nproc; 1940| | int i; 1941| | switch (__kmp_forkjoin_frames_mode) { 1942| | case 1: 1943| | __kmp_itt_frame_submit(gtid, this_thr->th.th_frame_time, cur_time, 0, 1944| | loc, nproc); 1945| | this_thr->th.th_frame_time = cur_time; 1946| | break; 1947| | case 2: // AC 2015-01-19: currently does not work for hierarchical (to 1948| | // be fixed) 1949| | __kmp_itt_frame_submit(gtid, this_thr->th.th_bar_min_time, cur_time, 1950| | 1, loc, nproc); 1951| | break; 1952| | case 3: 1953| | if (__itt_metadata_add_ptr) { 1954| | // Initialize with primary thread's wait time 1955| | kmp_uint64 delta = cur_time - this_thr->th.th_bar_arrive_time; 1956| | // Set arrive time to zero to be able to check it in 1957| | // __kmp_invoke_task(); the same is done inside the loop below 1958| | this_thr->th.th_bar_arrive_time = 0; 1959| | for (i = 1; i < nproc; ++i) { 1960| | delta += (cur_time - other_threads[i]->th.th_bar_arrive_time); 1961| | other_threads[i]->th.th_bar_arrive_time = 0; 1962| | } 1963| | __kmp_itt_metadata_imbalance(gtid, this_thr->th.th_frame_time, 1964| | cur_time, delta, 1965| | (kmp_uint64)(reduce != NULL)); 1966| | } 1967| | __kmp_itt_frame_submit(gtid, this_thr->th.th_frame_time, cur_time, 0, 1968| | loc, nproc); 1969| | this_thr->th.th_frame_time = cur_time; 1970| | break; 1971| | } 1972| | } 1973| |#endif /* USE_ITT_BUILD */ 1974| 33.2k| } else { 1975| 33.2k| status = 1; 1976| |#if USE_ITT_BUILD 1977| | if (__itt_sync_create_ptr || KMP_ITT_DEBUG) 1978| | __kmp_itt_barrier_middle(gtid, itt_sync_obj); 1979| |#endif /* USE_ITT_BUILD */ 1980| 33.2k| } 1981| 34.3k| if ((status == 1 || !is_split) && !cancelled) { ------------------ | Branch (1981:10): [True: 33.2k, False: 1.07k] | Branch (1981:25): [True: 1.07k, False: 0] | Branch (1981:39): [True: 0, Folded] ------------------ 1982| 34.3k| if (cancellable) { ------------------ | Branch (1982:11): [Folded, False: 34.3k] ------------------ 1983| 0| cancelled = __kmp_linear_barrier_release_cancellable( 1984| 0| bt, this_thr, gtid, tid, FALSE USE_ITT_BUILD_ARG(itt_sync_obj)); ------------------ | | 1279| 0|#define FALSE 0 ------------------ 1985| 34.3k| } else { 1986| 34.3k| switch (__kmp_barrier_release_pattern[bt]) { 1987| 0| case bp_dist_bar: { ------------------ | Branch (1987:9): [True: 0, False: 34.3k] ------------------ 1988| 0| KMP_ASSERT(__kmp_barrier_release_branch_bits[bt]); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 1989| 0| __kmp_dist_barrier_release(bt, this_thr, gtid, tid, 1990| 0| FALSE USE_ITT_BUILD_ARG(itt_sync_obj)); ------------------ | | 1279| 0|#define FALSE 0 ------------------ 1991| 0| break; 1992| 0| } 1993| 34.3k| case bp_hyper_bar: { ------------------ | Branch (1993:9): [True: 34.3k, False: 1] ------------------ 1994| 34.3k| KMP_ASSERT(__kmp_barrier_release_branch_bits[bt]); ------------------ | | 50| 34.3k| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 34.3k] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 1995| 34.3k| __kmp_hyper_barrier_release(bt, this_thr, gtid, tid, 1996| 34.3k| FALSE USE_ITT_BUILD_ARG(itt_sync_obj)); ------------------ | | 1279| 34.3k|#define FALSE 0 ------------------ 1997| 34.3k| break; 1998| 0| } 1999| 0| case bp_hierarchical_bar: { ------------------ | Branch (1999:9): [True: 0, False: 34.3k] ------------------ 2000| 0| __kmp_hierarchical_barrier_release( 2001| 0| bt, this_thr, gtid, tid, FALSE USE_ITT_BUILD_ARG(itt_sync_obj)); ------------------ | | 1279| 0|#define FALSE 0 ------------------ 2002| 0| break; 2003| 0| } 2004| 0| case bp_tree_bar: { ------------------ | Branch (2004:9): [True: 0, False: 34.3k] ------------------ 2005| 0| KMP_ASSERT(__kmp_barrier_release_branch_bits[bt]); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 2006| 0| __kmp_tree_barrier_release(bt, this_thr, gtid, tid, 2007| 0| FALSE USE_ITT_BUILD_ARG(itt_sync_obj)); ------------------ | | 1279| 0|#define FALSE 0 ------------------ 2008| 0| break; 2009| 0| } 2010| 0| default: { ------------------ | Branch (2010:9): [True: 0, False: 34.3k] ------------------ 2011| 0| __kmp_linear_barrier_release(bt, this_thr, gtid, tid, 2012| 0| FALSE USE_ITT_BUILD_ARG(itt_sync_obj)); ------------------ | | 1279| 0|#define FALSE 0 ------------------ 2013| 0| } 2014| 34.3k| } 2015| 34.3k| } 2016| 33.7k| if (__kmp_tasking_mode != tskm_immediate_exec && !cancelled) { ------------------ | Branch (2016:11): [True: 33.1k, False: 561] | Branch (2016:56): [True: 0, Folded] ------------------ 2017| 33.1k| __kmp_task_team_sync(this_thr, team); 2018| 33.1k| } 2019| 33.7k| } 2020| | 2021| |#if USE_ITT_BUILD 2022| | /* GEH: TODO: Move this under if-condition above and also include in 2023| | __kmp_end_split_barrier(). This will more accurately represent the actual 2024| | release time of the threads for split barriers. */ 2025| | if (__itt_sync_create_ptr || KMP_ITT_DEBUG) 2026| | __kmp_itt_barrier_finished(gtid, itt_sync_obj); 2027| |#endif /* USE_ITT_BUILD */ 2028| 34.3k| } else { // Team is serialized. 2029| 2.99k| status = 0; 2030| 3.01k| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (2030:9): [True: 3.01k, False: 18.4E] ------------------ 2031| 3.01k| if (this_thr->th.th_task_team != NULL) { ------------------ | Branch (2031:11): [True: 0, False: 3.01k] ------------------ 2032| |#if USE_ITT_NOTIFY 2033| | void *itt_sync_obj = NULL; 2034| | if (__itt_sync_create_ptr || KMP_ITT_DEBUG) { 2035| | itt_sync_obj = __kmp_itt_barrier_object(gtid, bt, 1); 2036| | __kmp_itt_barrier_starting(gtid, itt_sync_obj); 2037| | } 2038| |#endif 2039| | 2040| 0| KMP_DEBUG_ASSERT( 2041| 0| this_thr->th.th_task_team->tt.tt_found_proxy_tasks == TRUE || 2042| 0| this_thr->th.th_task_team->tt.tt_hidden_helper_task_encountered == 2043| 0| TRUE); 2044| 0| __kmp_task_team_wait(this_thr, team USE_ITT_BUILD_ARG(itt_sync_obj)); 2045| 0| __kmp_task_team_setup(this_thr, team, 0); 2046| | 2047| |#if USE_ITT_BUILD 2048| | if (__itt_sync_create_ptr || KMP_ITT_DEBUG) 2049| | __kmp_itt_barrier_finished(gtid, itt_sync_obj); 2050| |#endif /* USE_ITT_BUILD */ 2051| 0| } 2052| 3.01k| } 2053| 2.99k| } 2054| 36.7k| KA_TRACE(15, ("__kmp_barrier: T#%d(%d:%d) is leaving with return value %d\n", 2055| 36.7k| gtid, __kmp_team_from_gtid(gtid)->t.t_id, 2056| 36.7k| __kmp_tid_from_gtid(gtid), status)); 2057| | 2058| 36.7k|#if OMPT_SUPPORT 2059| 36.7k| if (ompt_enabled.enabled) { ------------------ | Branch (2059:7): [True: 0, False: 36.7k] ------------------ 2060| 0|#if OMPT_OPTIONAL 2061| 0| if (ompt_enabled.ompt_callback_sync_region_wait) { ------------------ | Branch (2061:9): [True: 0, False: 0] ------------------ 2062| 0| ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 2063| 0| barrier_kind, ompt_scope_end, my_parallel_data, my_task_data, 2064| 0| return_address); 2065| 0| } 2066| 0| if (ompt_enabled.ompt_callback_sync_region) { ------------------ | Branch (2066:9): [True: 0, False: 0] ------------------ 2067| 0| ompt_callbacks.ompt_callback(ompt_callback_sync_region)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 2068| 0| barrier_kind, ompt_scope_end, my_parallel_data, my_task_data, 2069| 0| return_address); 2070| 0| } 2071| 0|#endif 2072| 0| this_thr->th.ompt_thread_info.state = ompt_state_work_parallel; 2073| 0| } 2074| 36.7k|#endif 2075| | 2076| 36.7k| if (cancellable) ------------------ | Branch (2076:7): [Folded, False: 36.7k] ------------------ 2077| 0| return (int)cancelled; 2078| 36.7k| return status; 2079| 36.7k|} __kmpc_cancel: 29| 14.1k|kmp_int32 __kmpc_cancel(ident_t *loc_ref, kmp_int32 gtid, kmp_int32 cncl_kind) { 30| 14.1k| kmp_info_t *this_thr = __kmp_threads[gtid]; 31| | 32| 14.1k| KC_TRACE(10, ("__kmpc_cancel: T#%d request %d OMP_CANCELLATION=%d\n", gtid, 33| 14.1k| cncl_kind, __kmp_omp_cancellation)); 34| | 35| 14.1k| KMP_DEBUG_ASSERT(cncl_kind != cancel_noreq); 36| 14.1k| KMP_DEBUG_ASSERT(cncl_kind == cancel_parallel || cncl_kind == cancel_loop || 37| 14.1k| cncl_kind == cancel_sections || 38| 14.1k| cncl_kind == cancel_taskgroup); 39| 14.1k| KMP_DEBUG_ASSERT(__kmp_get_gtid() == gtid); 40| | 41| 14.1k| if (__kmp_omp_cancellation) { ------------------ | Branch (41:7): [True: 0, False: 14.1k] ------------------ 42| 0| switch (cncl_kind) { 43| 0| case cancel_parallel: ------------------ | Branch (43:5): [True: 0, False: 0] ------------------ 44| 0| case cancel_loop: ------------------ | Branch (44:5): [True: 0, False: 0] ------------------ 45| 0| case cancel_sections: ------------------ | Branch (45:5): [True: 0, False: 0] ------------------ 46| | // cancellation requests for parallel and worksharing constructs 47| | // are handled through the team structure 48| 0| { 49| 0| kmp_team_t *this_team = this_thr->th.th_team; 50| 0| KMP_DEBUG_ASSERT(this_team); 51| 0| kmp_int32 old = cancel_noreq; 52| 0| this_team->t.t_cancel_request.compare_exchange_strong(old, cncl_kind); 53| 0| if (old == cancel_noreq || old == cncl_kind) { ------------------ | Branch (53:13): [True: 0, False: 0] | Branch (53:36): [True: 0, False: 0] ------------------ 54| |// we do not have a cancellation request in this team or we do have 55| |// one that matches the current request -> cancel 56| 0|#if OMPT_SUPPORT && OMPT_OPTIONAL 57| 0| if (ompt_enabled.ompt_callback_cancel) { ------------------ | Branch (57:15): [True: 0, False: 0] ------------------ 58| 0| ompt_data_t *task_data; 59| 0| __ompt_get_task_info_internal(0, NULL, &task_data, NULL, NULL, 60| 0| NULL); 61| 0| ompt_cancel_flag_t type = ompt_cancel_parallel; 62| 0| if (cncl_kind == cancel_parallel) ------------------ | Branch (62:17): [True: 0, False: 0] ------------------ 63| 0| type = ompt_cancel_parallel; 64| 0| else if (cncl_kind == cancel_loop) ------------------ | Branch (64:22): [True: 0, False: 0] ------------------ 65| 0| type = ompt_cancel_loop; 66| 0| else if (cncl_kind == cancel_sections) ------------------ | Branch (66:22): [True: 0, False: 0] ------------------ 67| 0| type = ompt_cancel_sections; 68| 0| ompt_callbacks.ompt_callback(ompt_callback_cancel)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 69| 0| task_data, type | ompt_cancel_activated, 70| 0| OMPT_GET_RETURN_ADDRESS(0)); ------------------ | | 108| 0|#define OMPT_GET_RETURN_ADDRESS(level) __builtin_return_address(level) ------------------ 71| 0| } 72| 0|#endif // OMPT_SUPPORT && OMPT_OPTIONAL 73| 0| return 1 /* true */; 74| 0| } 75| 0| break; 76| 0| } 77| 0| case cancel_taskgroup: ------------------ | Branch (77:5): [True: 0, False: 0] ------------------ 78| | // cancellation requests for a task group 79| | // are handled through the taskgroup structure 80| 0| { 81| 0| kmp_taskdata_t *task; 82| 0| kmp_taskgroup_t *taskgroup; 83| | 84| 0| task = this_thr->th.th_current_task; 85| 0| KMP_DEBUG_ASSERT(task); 86| | 87| 0| taskgroup = task->td_taskgroup; 88| 0| if (taskgroup) { ------------------ | Branch (88:13): [True: 0, False: 0] ------------------ 89| 0| kmp_int32 old = cancel_noreq; 90| 0| taskgroup->cancel_request.compare_exchange_strong(old, cncl_kind); 91| 0| if (old == cancel_noreq || old == cncl_kind) { ------------------ | Branch (91:15): [True: 0, False: 0] | Branch (91:38): [True: 0, False: 0] ------------------ 92| |// we do not have a cancellation request in this taskgroup or we do 93| |// have one that matches the current request -> cancel 94| 0|#if OMPT_SUPPORT && OMPT_OPTIONAL 95| 0| if (ompt_enabled.ompt_callback_cancel) { ------------------ | Branch (95:17): [True: 0, False: 0] ------------------ 96| 0| ompt_data_t *task_data; 97| 0| __ompt_get_task_info_internal(0, NULL, &task_data, NULL, NULL, 98| 0| NULL); 99| 0| ompt_callbacks.ompt_callback(ompt_callback_cancel)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 100| 0| task_data, ompt_cancel_taskgroup | ompt_cancel_activated, 101| 0| OMPT_GET_RETURN_ADDRESS(0)); ------------------ | | 108| 0|#define OMPT_GET_RETURN_ADDRESS(level) __builtin_return_address(level) ------------------ 102| 0| } 103| 0|#endif 104| 0| return 1 /* true */; 105| 0| } 106| 0| } else { 107| | // TODO: what needs to happen here? 108| | // the specification disallows cancellation w/o taskgroups 109| | // so we might do anything here, let's abort for now 110| 0| KMP_ASSERT(0 /* false */); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, Folded] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 111| 0| } 112| 0| } 113| 0| break; 114| 0| default: ------------------ | Branch (114:5): [True: 0, False: 0] ------------------ 115| 0| KMP_ASSERT(0 /* false */); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, Folded] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 116| 0| } 117| 0| } 118| | 119| | // ICV OMP_CANCELLATION=false, so we ignored this cancel request 120| 14.1k| KMP_DEBUG_ASSERT(!__kmp_omp_cancellation); 121| 14.1k| return 0 /* false */; 122| 14.1k|} __kmpc_global_thread_num: 99| 40.7k|kmp_int32 __kmpc_global_thread_num(ident_t *loc) { 100| 40.7k| kmp_int32 gtid = __kmp_entry_gtid(); ------------------ | | 3467| 40.7k|#define __kmp_entry_gtid() __kmp_get_global_thread_id_reg() ------------------ 101| | 102| 40.7k| KC_TRACE(10, ("__kmpc_global_thread_num: T#%d\n", gtid)); 103| | 104| 40.7k| return gtid; 105| 40.7k|} __kmpc_push_num_threads: 232| 4.36k| kmp_int32 num_threads) { 233| 4.36k| KA_TRACE(20, ("__kmpc_push_num_threads: enter T#%d num_threads=%d\n", 234| 4.36k| global_tid, num_threads)); 235| 4.36k| __kmp_assert_valid_gtid(global_tid); 236| 4.36k| __kmp_push_num_threads(loc, global_tid, num_threads); 237| 4.36k|} __kmpc_fork_call: 262| 1.35k|void __kmpc_fork_call(ident_t *loc, kmp_int32 argc, kmpc_micro microtask, ...) { 263| 1.35k| int gtid = __kmp_entry_gtid(); ------------------ | | 3467| 1.35k|#define __kmp_entry_gtid() __kmp_get_global_thread_id_reg() ------------------ 264| | 265| |#if (KMP_STATS_ENABLED) 266| | // If we were in a serial region, then stop the serial timer, record 267| | // the event, and start parallel region timer 268| | stats_state_e previous_state = KMP_GET_THREAD_STATE(); 269| | if (previous_state == stats_state_e::SERIAL_REGION) { 270| | KMP_EXCHANGE_PARTITIONED_TIMER(OMP_parallel_overhead); 271| | } else { 272| | KMP_PUSH_PARTITIONED_TIMER(OMP_parallel_overhead); 273| | } 274| | int inParallel = __kmpc_in_parallel(loc); 275| | if (inParallel) { 276| | KMP_COUNT_BLOCK(OMP_NESTED_PARALLEL); 277| | } else { 278| | KMP_COUNT_BLOCK(OMP_PARALLEL); 279| | } 280| |#endif 281| | 282| | // maybe to save thr_state is enough here 283| 1.35k| { 284| 1.35k| va_list ap; 285| 1.35k| va_start(ap, microtask); 286| | 287| 1.35k|#if OMPT_SUPPORT 288| 1.35k| ompt_frame_t *ompt_frame; 289| 1.35k| if (ompt_enabled.enabled) { ------------------ | Branch (289:9): [True: 0, False: 1.35k] ------------------ 290| 0| kmp_info_t *master_th = __kmp_threads[gtid]; 291| 0| ompt_frame = &master_th->th.th_current_task->ompt_task_info.frame; 292| 0| ompt_frame->enter_frame.ptr = OMPT_GET_FRAME_ADDRESS(0); ------------------ | | 109| 0|#define OMPT_GET_FRAME_ADDRESS(level) __builtin_frame_address(level) ------------------ 293| 0| } 294| 1.35k| OMPT_STORE_RETURN_ADDRESS(gtid); ------------------ | | 92| 1.35k| OmptReturnAddressGuard ReturnAddressGuard{gtid, __builtin_return_address(0)}; ------------------ 295| 1.35k|#endif 296| | 297| |#if INCLUDE_SSC_MARKS 298| | SSC_MARK_FORKING(); 299| |#endif 300| 1.35k| __kmp_fork_call(loc, gtid, fork_context_intel, argc, 301| 1.35k| VOLATILE_CAST(microtask_t) microtask, // "wrapped" task ------------------ | | 1180| 1.35k|#define VOLATILE_CAST(x) (x) ------------------ 302| 1.35k| VOLATILE_CAST(launch_t) __kmp_invoke_task_func, ------------------ | | 1180| 1.35k|#define VOLATILE_CAST(x) (x) ------------------ 303| 1.35k| kmp_va_addr_of(ap)); ------------------ | | 220| 1.35k|#define kmp_va_addr_of(ap) (&(ap)) ------------------ 304| |#if INCLUDE_SSC_MARKS 305| | SSC_MARK_JOINING(); 306| |#endif 307| 1.35k| __kmp_join_call(loc, gtid 308| 1.35k|#if OMPT_SUPPORT 309| 1.35k| , 310| 1.35k| fork_context_intel 311| 1.35k|#endif 312| 1.35k| ); 313| | 314| 1.35k| va_end(ap); 315| | 316| 1.35k|#if OMPT_SUPPORT 317| 1.35k| if (ompt_enabled.enabled) { ------------------ | Branch (317:9): [True: 0, False: 1.35k] ------------------ 318| 0| ompt_frame->enter_frame = ompt_data_none; ------------------ | | 1408| 0|#define ompt_data_none {0} ------------------ 319| 0| } 320| 1.35k|#endif 321| 1.35k| } 322| | 323| |#if KMP_STATS_ENABLED 324| | if (previous_state == stats_state_e::SERIAL_REGION) { 325| | KMP_EXCHANGE_PARTITIONED_TIMER(OMP_serial); 326| | KMP_SET_THREAD_STATE(previous_state); 327| | } else { 328| | KMP_POP_PARTITIONED_TIMER(); 329| | } 330| |#endif // KMP_STATS_ENABLED 331| 1.35k|} __kmpc_serialized_parallel: 522| 3.01k|void __kmpc_serialized_parallel(ident_t *loc, kmp_int32 global_tid) { 523| | // The implementation is now in kmp_runtime.cpp so that it can share static 524| | // functions with kmp_fork_call since the tasks to be done are similar in 525| | // each case. 526| 3.01k| __kmp_assert_valid_gtid(global_tid); 527| 3.01k|#if OMPT_SUPPORT 528| 3.01k| OMPT_STORE_RETURN_ADDRESS(global_tid); ------------------ | | 92| 3.01k| OmptReturnAddressGuard ReturnAddressGuard{gtid, __builtin_return_address(0)}; ------------------ 529| 3.01k|#endif 530| 3.01k| __kmp_serialized_parallel(loc, global_tid); 531| 3.01k|} __kmpc_end_serialized_parallel: 540| 3.01k|void __kmpc_end_serialized_parallel(ident_t *loc, kmp_int32 global_tid) { 541| 3.01k| kmp_internal_control_t *top; 542| 3.01k| kmp_info_t *this_thr; 543| 3.01k| kmp_team_t *serial_team; 544| | 545| 3.01k| KC_TRACE(10, 546| 3.01k| ("__kmpc_end_serialized_parallel: called by T#%d\n", global_tid)); 547| | 548| | /* skip all this code for autopar serialized loops since it results in 549| | unacceptable overhead */ 550| 3.01k| if (loc != NULL && (loc->flags & KMP_IDENT_AUTOPAR)) ------------------ | Branch (550:7): [True: 3.01k, False: 0] | Branch (550:22): [True: 0, False: 3.01k] ------------------ 551| 0| return; 552| | 553| | // Not autopar code 554| 3.01k| __kmp_assert_valid_gtid(global_tid); 555| 3.01k| if (!TCR_4(__kmp_init_parallel)) ------------------ | | 1127| 3.01k|#define TCR_4(a) (a) ------------------ | Branch (555:7): [True: 0, False: 3.01k] ------------------ 556| 0| __kmp_parallel_initialize(); 557| | 558| 3.01k| __kmp_resume_if_soft_paused(); 559| | 560| 3.01k| this_thr = __kmp_threads[global_tid]; 561| 3.01k| serial_team = this_thr->th.th_serial_team; 562| | 563| 3.01k| kmp_task_team_t *task_team = this_thr->th.th_task_team; 564| | // we need to wait for the proxy tasks before finishing the thread 565| 3.01k| if (task_team != NULL && (task_team->tt.tt_found_proxy_tasks || ------------------ | Branch (565:7): [True: 0, False: 3.01k] | Branch (565:29): [True: 0, False: 0] ------------------ 566| 0| task_team->tt.tt_hidden_helper_task_encountered)) ------------------ | Branch (566:29): [True: 0, False: 0] ------------------ 567| 0| __kmp_task_team_wait(this_thr, serial_team USE_ITT_BUILD_ARG(NULL)); 568| | 569| 3.01k| KMP_MB(); 570| 3.01k| KMP_DEBUG_ASSERT(serial_team); 571| 3.01k| KMP_ASSERT(serial_team->t.t_serialized); ------------------ | | 50| 3.01k| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 3.01k] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 572| 3.01k| KMP_DEBUG_ASSERT(this_thr->th.th_team == serial_team); 573| 3.01k| KMP_DEBUG_ASSERT(serial_team != this_thr->th.th_root->r.r_root_team); 574| 3.01k| KMP_DEBUG_ASSERT(serial_team->t.t_threads); 575| 3.01k| KMP_DEBUG_ASSERT(serial_team->t.t_threads[0] == this_thr); 576| | 577| 3.01k|#if OMPT_SUPPORT 578| 3.01k| if (ompt_enabled.enabled && ------------------ | Branch (578:7): [True: 0, False: 3.01k] ------------------ 579| 0| this_thr->th.ompt_thread_info.state != ompt_state_overhead) { ------------------ | Branch (579:7): [True: 0, False: 0] ------------------ 580| 0| OMPT_CUR_TASK_INFO(this_thr)->frame.exit_frame = ompt_data_none; ------------------ | | 68| 0|#define OMPT_CUR_TASK_INFO(thr) (&((thr)->th.th_current_task->ompt_task_info)) ------------------ OMPT_CUR_TASK_INFO(this_thr)->frame.exit_frame = ompt_data_none; ------------------ | | 1408| 0|#define ompt_data_none {0} ------------------ 581| 0| if (ompt_enabled.ompt_callback_implicit_task) { ------------------ | Branch (581:9): [True: 0, False: 0] ------------------ 582| 0| ompt_callbacks.ompt_callback(ompt_callback_implicit_task)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 583| 0| ompt_scope_end, NULL, OMPT_CUR_TASK_DATA(this_thr), 1, ------------------ | | 70| 0| (&((thr)->th.th_current_task->ompt_task_info.task_data)) ------------------ 584| 0| OMPT_CUR_TASK_INFO(this_thr)->thread_num, ompt_task_implicit); ------------------ | | 68| 0|#define OMPT_CUR_TASK_INFO(thr) (&((thr)->th.th_current_task->ompt_task_info)) ------------------ 585| 0| } 586| | 587| | // reset clear the task id only after unlinking the task 588| 0| ompt_data_t *parent_task_data; 589| 0| __ompt_get_task_info_internal(1, NULL, &parent_task_data, NULL, NULL, NULL); 590| | 591| 0| if (ompt_enabled.ompt_callback_parallel_end) { ------------------ | Branch (591:9): [True: 0, False: 0] ------------------ 592| 0| ompt_callbacks.ompt_callback(ompt_callback_parallel_end)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 593| 0| &(serial_team->t.ompt_team_info.parallel_data), parent_task_data, 594| 0| ompt_parallel_invoker_program | ompt_parallel_team, 595| 0| OMPT_LOAD_RETURN_ADDRESS(global_tid)); ------------------ | | 93| 0|#define OMPT_LOAD_RETURN_ADDRESS(gtid) __ompt_load_return_address(gtid) ------------------ 596| 0| } 597| 0| __ompt_lw_taskteam_unlink(this_thr); 598| 0| this_thr->th.ompt_thread_info.state = ompt_state_overhead; 599| 0| } 600| 3.01k|#endif 601| | 602| | /* If necessary, pop the internal control stack values and replace the team 603| | * values */ 604| 3.01k| top = serial_team->t.t_control_stack_top; 605| 3.01k| if (top && top->serial_nesting_level == serial_team->t.t_serialized) { ------------------ | Branch (605:7): [True: 0, False: 3.01k] | Branch (605:14): [True: 0, False: 0] ------------------ 606| 0| copy_icvs(&serial_team->t.t_threads[0]->th.th_current_task->td_icvs, top); 607| 0| serial_team->t.t_control_stack_top = top->next; 608| 0| __kmp_free(top); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 609| 0| } 610| | 611| | /* pop dispatch buffers stack */ 612| 3.01k| KMP_DEBUG_ASSERT(serial_team->t.t_dispatch->th_disp_buffer); 613| 3.01k| { 614| 3.01k| dispatch_private_info_t *disp_buffer = 615| 3.01k| serial_team->t.t_dispatch->th_disp_buffer; 616| 3.01k| serial_team->t.t_dispatch->th_disp_buffer = 617| 3.01k| serial_team->t.t_dispatch->th_disp_buffer->next; 618| 3.01k| __kmp_free(disp_buffer); ------------------ | | 3619| 3.01k|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 619| 3.01k| } 620| 3.01k| this_thr->th.th_def_allocator = serial_team->t.t_def_allocator; // restore 621| | 622| 3.01k| --serial_team->t.t_serialized; 623| 3.01k| if (serial_team->t.t_serialized == 0) { ------------------ | Branch (623:7): [True: 3.01k, False: 0] ------------------ 624| | 625| | /* return to the parallel section */ 626| | 627| 3.01k|#if KMP_ARCH_X86 || KMP_ARCH_X86_64 628| 3.01k| if (__kmp_inherit_fp_control && serial_team->t.t_fp_control_saved) { ------------------ | Branch (628:9): [True: 3.01k, False: 0] | Branch (628:37): [True: 3.01k, False: 0] ------------------ 629| 3.01k| __kmp_clear_x87_fpu_status_word(); 630| 3.01k| __kmp_load_x87_fpu_control_word(&serial_team->t.t_x87_fpu_control_word); 631| 3.01k| __kmp_load_mxcsr(&serial_team->t.t_mxcsr); 632| 3.01k| } 633| 3.01k|#endif /* KMP_ARCH_X86 || KMP_ARCH_X86_64 */ 634| | 635| 3.01k| __kmp_pop_current_task_from_thread(this_thr); 636| 3.01k|#if OMPD_SUPPORT 637| 3.01k| if (ompd_state & OMPD_ENABLE_BP) ------------------ | | 35| 3.01k|#define OMPD_ENABLE_BP 0x1 ------------------ | Branch (637:9): [True: 0, False: 3.01k] ------------------ 638| 0| ompd_bp_parallel_end(); 639| 3.01k|#endif 640| | 641| 3.01k| this_thr->th.th_team = serial_team->t.t_parent; 642| 3.01k| this_thr->th.th_info.ds.ds_tid = serial_team->t.t_master_tid; 643| | 644| | /* restore values cached in the thread */ 645| 3.01k| this_thr->th.th_team_nproc = serial_team->t.t_parent->t.t_nproc; /* JPH */ 646| 3.01k| this_thr->th.th_team_master = 647| 3.01k| serial_team->t.t_parent->t.t_threads[0]; /* JPH */ 648| 3.01k| this_thr->th.th_team_serialized = this_thr->th.th_team->t.t_serialized; 649| | 650| | /* TODO the below shouldn't need to be adjusted for serialized teams */ 651| 3.01k| this_thr->th.th_dispatch = 652| 3.01k| &this_thr->th.th_team->t.t_dispatch[serial_team->t.t_master_tid]; 653| | 654| 3.01k| KMP_ASSERT(this_thr->th.th_current_task->td_flags.executing == 0); ------------------ | | 50| 3.01k| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 3.01k] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 655| 3.01k| this_thr->th.th_current_task->td_flags.executing = 1; 656| | 657| 3.01k| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (657:9): [True: 3.01k, False: 0] ------------------ 658| | // Copy the task team from the new child / old parent team to the thread. 659| 3.01k| this_thr->th.th_task_team = 660| 3.01k| this_thr->th.th_team->t.t_task_team[this_thr->th.th_task_state]; 661| 3.01k| KA_TRACE(20, 662| 3.01k| ("__kmpc_end_serialized_parallel: T#%d restoring task_team %p / " 663| 3.01k| "team %p\n", 664| 3.01k| global_tid, this_thr->th.th_task_team, this_thr->th.th_team)); 665| 3.01k| } 666| 3.01k|#if KMP_AFFINITY_SUPPORTED 667| 3.01k| if (this_thr->th.th_team->t.t_level == 0 && __kmp_affinity.flags.reset) { ------------------ | Branch (667:9): [True: 3.01k, False: 0] | Branch (667:49): [True: 0, False: 3.01k] ------------------ 668| 0| __kmp_reset_root_init_mask(global_tid); 669| 0| } 670| 3.01k|#endif 671| 3.01k| } else { 672| 0| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (672:9): [True: 0, False: 0] ------------------ 673| 0| KA_TRACE(20, ("__kmpc_end_serialized_parallel: T#%d decreasing nesting " 674| 0| "depth of serial team %p to %d\n", 675| 0| global_tid, serial_team, serial_team->t.t_serialized)); 676| 0| } 677| 0| } 678| | 679| 3.01k| serial_team->t.t_level--; 680| 3.01k| if (__kmp_env_consistency_check) ------------------ | Branch (680:7): [True: 0, False: 3.01k] ------------------ 681| 0| __kmp_pop_parallel(global_tid, NULL); 682| 3.01k|#if OMPT_SUPPORT 683| 3.01k| if (ompt_enabled.enabled) ------------------ | Branch (683:7): [True: 0, False: 3.01k] ------------------ 684| 0| this_thr->th.ompt_thread_info.state = 685| 0| ((this_thr->th.th_team_serialized) ? ompt_state_work_serial ------------------ | Branch (685:10): [True: 0, False: 0] ------------------ 686| 0| : ompt_state_work_parallel); 687| 3.01k|#endif 688| 3.01k|} __kmpc_barrier: 720| 37.0k|void __kmpc_barrier(ident_t *loc, kmp_int32 global_tid) { 721| 37.0k| KMP_COUNT_BLOCK(OMP_BARRIER); ------------------ | | 1001| 37.0k|#define KMP_COUNT_BLOCK(n) ((void)0) ------------------ 722| 37.0k| KC_TRACE(10, ("__kmpc_barrier: called T#%d\n", global_tid)); 723| 37.0k| __kmp_assert_valid_gtid(global_tid); 724| | 725| 37.0k| if (!TCR_4(__kmp_init_parallel)) ------------------ | | 1127| 37.0k|#define TCR_4(a) (a) ------------------ | Branch (725:7): [True: 0, False: 37.0k] ------------------ 726| 0| __kmp_parallel_initialize(); 727| | 728| 37.0k| __kmp_resume_if_soft_paused(); 729| | 730| 37.0k| if (__kmp_env_consistency_check) { ------------------ | Branch (730:7): [True: 0, False: 37.0k] ------------------ 731| 0| if (loc == 0) { ------------------ | Branch (731:9): [True: 0, False: 0] ------------------ 732| 0| KMP_WARNING(ConstructIdentInvalid); // ??? What does it mean for the user? ------------------ | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 733| 0| } 734| 0| __kmp_check_barrier(global_tid, ct_barrier, loc); 735| 0| } 736| | 737| 37.0k|#if OMPT_SUPPORT 738| 37.0k| ompt_frame_t *ompt_frame; 739| 37.0k| if (ompt_enabled.enabled) { ------------------ | Branch (739:7): [True: 0, False: 37.0k] ------------------ 740| 0| __ompt_get_task_info_internal(0, NULL, NULL, &ompt_frame, NULL, NULL); 741| 0| if (ompt_frame->enter_frame.ptr == NULL) ------------------ | Branch (741:9): [True: 0, False: 0] ------------------ 742| 0| ompt_frame->enter_frame.ptr = OMPT_GET_FRAME_ADDRESS(0); ------------------ | | 109| 0|#define OMPT_GET_FRAME_ADDRESS(level) __builtin_frame_address(level) ------------------ 743| 0| } 744| 37.0k| OMPT_STORE_RETURN_ADDRESS(global_tid); ------------------ | | 92| 37.0k| OmptReturnAddressGuard ReturnAddressGuard{gtid, __builtin_return_address(0)}; ------------------ 745| 37.0k|#endif 746| 37.0k| __kmp_threads[global_tid]->th.th_ident = loc; 747| | // TODO: explicit barrier_wait_id: 748| | // this function is called when 'barrier' directive is present or 749| | // implicit barrier at the end of a worksharing construct. 750| | // 1) better to add a per-thread barrier counter to a thread data structure 751| | // 2) set to 0 when a new team is created 752| | // 4) no sync is required 753| | 754| 37.0k| __kmp_barrier(bs_plain_barrier, global_tid, FALSE, 0, NULL, NULL); ------------------ | | 1279| 37.0k|#define FALSE 0 ------------------ 755| 37.0k|#if OMPT_SUPPORT && OMPT_OPTIONAL 756| 37.0k| if (ompt_enabled.enabled) { ------------------ | Branch (756:7): [True: 0, False: 37.0k] ------------------ 757| 0| ompt_frame->enter_frame = ompt_data_none; ------------------ | | 1408| 0|#define ompt_data_none {0} ------------------ 758| 0| } 759| 37.0k|#endif 760| 37.0k|} __kmpc_for_static_fini: 1910| 45.8k|void __kmpc_for_static_fini(ident_t *loc, kmp_int32 global_tid) { 1911| 45.8k| KMP_POP_PARTITIONED_TIMER(); ------------------ | | 1015| 45.8k|#define KMP_POP_PARTITIONED_TIMER() ((void)0) ------------------ 1912| 45.8k| KE_TRACE(10, ("__kmpc_for_static_fini called T#%d\n", global_tid)); 1913| | 1914| 45.8k|#if OMPT_SUPPORT && OMPT_OPTIONAL 1915| 45.8k| if (ompt_enabled.ompt_callback_work) { ------------------ | Branch (1915:7): [True: 0, False: 45.8k] ------------------ 1916| 0| ompt_work_t ompt_work_type = ompt_work_loop; 1917| 0| ompt_team_info_t *team_info = __ompt_get_teaminfo(0, NULL); 1918| 0| ompt_task_info_t *task_info = __ompt_get_task_info_object(0); 1919| | // Determine workshare type 1920| 0| if (loc != NULL) { ------------------ | Branch (1920:9): [True: 0, False: 0] ------------------ 1921| 0| if ((loc->flags & KMP_IDENT_WORK_LOOP) != 0) { ------------------ | Branch (1921:11): [True: 0, False: 0] ------------------ 1922| 0| ompt_work_type = ompt_work_loop; 1923| 0| } else if ((loc->flags & KMP_IDENT_WORK_SECTIONS) != 0) { ------------------ | Branch (1923:18): [True: 0, False: 0] ------------------ 1924| 0| ompt_work_type = ompt_work_sections; 1925| 0| } else if ((loc->flags & KMP_IDENT_WORK_DISTRIBUTE) != 0) { ------------------ | Branch (1925:18): [True: 0, False: 0] ------------------ 1926| 0| ompt_work_type = ompt_work_distribute; 1927| 0| } else { 1928| | // use default set above. 1929| | // a warning about this case is provided in __kmpc_for_static_init 1930| 0| } 1931| 0| KMP_DEBUG_ASSERT(ompt_work_type); 1932| 0| } 1933| 0| ompt_callbacks.ompt_callback(ompt_callback_work)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 1934| 0| ompt_work_type, ompt_scope_end, &(team_info->parallel_data), 1935| 0| &(task_info->task_data), 0, OMPT_GET_RETURN_ADDRESS(0)); ------------------ | | 108| 0|#define OMPT_GET_RETURN_ADDRESS(level) __builtin_return_address(level) ------------------ 1936| 0| } 1937| 45.8k|#endif 1938| 45.8k| if (__kmp_env_consistency_check) ------------------ | Branch (1938:7): [True: 0, False: 45.8k] ------------------ 1939| 0| __kmp_pop_workshare(global_tid, ct_pdo, loc); 1940| 45.8k|} __kmpc_init_lock: 2419| 30.8k|void __kmpc_init_lock(ident_t *loc, kmp_int32 gtid, void **user_lock) { 2420| 30.8k|#if KMP_USE_DYNAMIC_LOCK 2421| | 2422| 30.8k| KMP_DEBUG_ASSERT(__kmp_init_serial); 2423| 30.8k| if (__kmp_env_consistency_check && user_lock == NULL) { ------------------ | Branch (2423:7): [True: 0, False: 30.8k] | Branch (2423:38): [True: 0, False: 0] ------------------ 2424| 0| KMP_FATAL(LockIsUninitialized, "omp_init_lock"); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 2425| 0| } 2426| 30.8k| __kmp_init_lock_with_hint(loc, user_lock, __kmp_user_lock_seq); 2427| | 2428| 30.8k|#if OMPT_SUPPORT && OMPT_OPTIONAL 2429| | // This is the case, if called from omp_init_lock_with_hint: 2430| 30.8k| void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid); ------------------ | | 93| 30.8k|#define OMPT_LOAD_RETURN_ADDRESS(gtid) __ompt_load_return_address(gtid) ------------------ 2431| 30.8k| if (!codeptr) ------------------ | Branch (2431:7): [True: 30.8k, False: 0] ------------------ 2432| 30.8k| codeptr = OMPT_GET_RETURN_ADDRESS(0); ------------------ | | 108| 30.8k|#define OMPT_GET_RETURN_ADDRESS(level) __builtin_return_address(level) ------------------ 2433| 30.8k| if (ompt_enabled.ompt_callback_lock_init) { ------------------ | Branch (2433:7): [True: 0, False: 30.8k] ------------------ 2434| 0| ompt_callbacks.ompt_callback(ompt_callback_lock_init)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 2435| 0| ompt_mutex_lock, omp_lock_hint_none, 2436| 0| __ompt_get_mutex_impl_type(user_lock), 2437| 0| (ompt_wait_id_t)(uintptr_t)user_lock, codeptr); 2438| 0| } 2439| 30.8k|#endif 2440| | 2441| |#else // KMP_USE_DYNAMIC_LOCK 2442| | 2443| | static char const *const func = "omp_init_lock"; 2444| | kmp_user_lock_p lck; 2445| | KMP_DEBUG_ASSERT(__kmp_init_serial); 2446| | 2447| | if (__kmp_env_consistency_check) { 2448| | if (user_lock == NULL) { 2449| | KMP_FATAL(LockIsUninitialized, func); 2450| | } 2451| | } 2452| | 2453| | KMP_CHECK_USER_LOCK_INIT(); 2454| | 2455| | if ((__kmp_user_lock_kind == lk_tas) && 2456| | (sizeof(lck->tas.lk.poll) <= OMP_LOCK_T_SIZE)) { 2457| | lck = (kmp_user_lock_p)user_lock; 2458| | } 2459| |#if KMP_USE_FUTEX 2460| | else if ((__kmp_user_lock_kind == lk_futex) && 2461| | (sizeof(lck->futex.lk.poll) <= OMP_LOCK_T_SIZE)) { 2462| | lck = (kmp_user_lock_p)user_lock; 2463| | } 2464| |#endif 2465| | else { 2466| | lck = __kmp_user_lock_allocate(user_lock, gtid, 0); 2467| | } 2468| | INIT_LOCK(lck); 2469| | __kmp_set_user_lock_location(lck, loc); 2470| | 2471| |#if OMPT_SUPPORT && OMPT_OPTIONAL 2472| | // This is the case, if called from omp_init_lock_with_hint: 2473| | void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid); 2474| | if (!codeptr) 2475| | codeptr = OMPT_GET_RETURN_ADDRESS(0); 2476| | if (ompt_enabled.ompt_callback_lock_init) { 2477| | ompt_callbacks.ompt_callback(ompt_callback_lock_init)( 2478| | ompt_mutex_lock, omp_lock_hint_none, __ompt_get_mutex_impl_type(), 2479| | (ompt_wait_id_t)(uintptr_t)user_lock, codeptr); 2480| | } 2481| |#endif 2482| | 2483| |#if USE_ITT_BUILD 2484| | __kmp_itt_lock_creating(lck); 2485| |#endif /* USE_ITT_BUILD */ 2486| | 2487| |#endif // KMP_USE_DYNAMIC_LOCK 2488| 30.8k|} // __kmpc_init_lock __kmpc_destroy_lock: 2565| 30.8k|void __kmpc_destroy_lock(ident_t *loc, kmp_int32 gtid, void **user_lock) { 2566| 30.8k|#if KMP_USE_DYNAMIC_LOCK 2567| | 2568| |#if USE_ITT_BUILD 2569| | kmp_user_lock_p lck; 2570| | if (KMP_EXTRACT_D_TAG(user_lock) == 0) { 2571| | lck = ((kmp_indirect_lock_t *)KMP_LOOKUP_I_LOCK(user_lock))->lock; 2572| | } else { 2573| | lck = (kmp_user_lock_p)user_lock; 2574| | } 2575| | __kmp_itt_lock_destroyed(lck); 2576| |#endif 2577| 30.8k|#if OMPT_SUPPORT && OMPT_OPTIONAL 2578| | // This is the case, if called from omp_init_lock_with_hint: 2579| 30.8k| void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid); ------------------ | | 93| 30.8k|#define OMPT_LOAD_RETURN_ADDRESS(gtid) __ompt_load_return_address(gtid) ------------------ 2580| 30.8k| if (!codeptr) ------------------ | Branch (2580:7): [True: 30.8k, False: 0] ------------------ 2581| 30.8k| codeptr = OMPT_GET_RETURN_ADDRESS(0); ------------------ | | 108| 30.8k|#define OMPT_GET_RETURN_ADDRESS(level) __builtin_return_address(level) ------------------ 2582| 30.8k| if (ompt_enabled.ompt_callback_lock_destroy) { ------------------ | Branch (2582:7): [True: 0, False: 30.8k] ------------------ 2583| 0| ompt_callbacks.ompt_callback(ompt_callback_lock_destroy)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 2584| 0| ompt_mutex_lock, (ompt_wait_id_t)(uintptr_t)user_lock, codeptr); 2585| 0| } 2586| 30.8k|#endif 2587| 30.8k| KMP_D_LOCK_FUNC(user_lock, destroy)((kmp_dyna_lock_t *)user_lock); ------------------ | | 1149| 30.8k|#define KMP_D_LOCK_FUNC(l, op) __kmp_direct_##op[KMP_EXTRACT_D_TAG(l)] | | ------------------ | | | | 1141| 30.8k| (*((kmp_dyna_lock_t *)(l)) & ((1 << KMP_LOCK_SHIFT) - 1) & \ | | | | ------------------ | | | | | | 1076| 30.8k| 8 // number of low bits to be used as tag for direct locks | | | | ------------------ | | | | 1142| 30.8k| -(*((kmp_dyna_lock_t *)(l)) & 1)) | | ------------------ ------------------ 2588| |#else 2589| | kmp_user_lock_p lck; 2590| | 2591| | if ((__kmp_user_lock_kind == lk_tas) && 2592| | (sizeof(lck->tas.lk.poll) <= OMP_LOCK_T_SIZE)) { 2593| | lck = (kmp_user_lock_p)user_lock; 2594| | } 2595| |#if KMP_USE_FUTEX 2596| | else if ((__kmp_user_lock_kind == lk_futex) && 2597| | (sizeof(lck->futex.lk.poll) <= OMP_LOCK_T_SIZE)) { 2598| | lck = (kmp_user_lock_p)user_lock; 2599| | } 2600| |#endif 2601| | else { 2602| | lck = __kmp_lookup_user_lock(user_lock, "omp_destroy_lock"); 2603| | } 2604| | 2605| |#if OMPT_SUPPORT && OMPT_OPTIONAL 2606| | // This is the case, if called from omp_init_lock_with_hint: 2607| | void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid); 2608| | if (!codeptr) 2609| | codeptr = OMPT_GET_RETURN_ADDRESS(0); 2610| | if (ompt_enabled.ompt_callback_lock_destroy) { 2611| | ompt_callbacks.ompt_callback(ompt_callback_lock_destroy)( 2612| | ompt_mutex_lock, (ompt_wait_id_t)(uintptr_t)user_lock, codeptr); 2613| | } 2614| |#endif 2615| | 2616| |#if USE_ITT_BUILD 2617| | __kmp_itt_lock_destroyed(lck); 2618| |#endif /* USE_ITT_BUILD */ 2619| | DESTROY_LOCK(lck); 2620| | 2621| | if ((__kmp_user_lock_kind == lk_tas) && 2622| | (sizeof(lck->tas.lk.poll) <= OMP_LOCK_T_SIZE)) { 2623| | ; 2624| | } 2625| |#if KMP_USE_FUTEX 2626| | else if ((__kmp_user_lock_kind == lk_futex) && 2627| | (sizeof(lck->futex.lk.poll) <= OMP_LOCK_T_SIZE)) { 2628| | ; 2629| | } 2630| |#endif 2631| | else { 2632| | __kmp_user_lock_free(user_lock, gtid, lck); 2633| | } 2634| |#endif // KMP_USE_DYNAMIC_LOCK 2635| 30.8k|} // __kmpc_destroy_lock __kmpc_set_lock: 2712| 25.1k|void __kmpc_set_lock(ident_t *loc, kmp_int32 gtid, void **user_lock) { 2713| 25.1k| KMP_COUNT_BLOCK(OMP_set_lock); ------------------ | | 1001| 25.1k|#define KMP_COUNT_BLOCK(n) ((void)0) ------------------ 2714| 25.1k|#if KMP_USE_DYNAMIC_LOCK 2715| 25.1k| int tag = KMP_EXTRACT_D_TAG(user_lock); ------------------ | | 1141| 25.1k| (*((kmp_dyna_lock_t *)(l)) & ((1 << KMP_LOCK_SHIFT) - 1) & \ | | ------------------ | | | | 1076| 25.1k| 8 // number of low bits to be used as tag for direct locks | | ------------------ | | 1142| 25.1k| -(*((kmp_dyna_lock_t *)(l)) & 1)) ------------------ 2716| |#if USE_ITT_BUILD 2717| | __kmp_itt_lock_acquiring( 2718| | (kmp_user_lock_p) 2719| | user_lock); // itt function will get to the right lock object. 2720| |#endif 2721| 25.1k|#if OMPT_SUPPORT && OMPT_OPTIONAL 2722| | // This is the case, if called from omp_init_lock_with_hint: 2723| 25.1k| void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid); ------------------ | | 93| 25.1k|#define OMPT_LOAD_RETURN_ADDRESS(gtid) __ompt_load_return_address(gtid) ------------------ 2724| 25.1k| if (!codeptr) ------------------ | Branch (2724:7): [True: 25.1k, False: 1] ------------------ 2725| 25.1k| codeptr = OMPT_GET_RETURN_ADDRESS(0); ------------------ | | 108| 25.1k|#define OMPT_GET_RETURN_ADDRESS(level) __builtin_return_address(level) ------------------ 2726| 25.1k| if (ompt_enabled.ompt_callback_mutex_acquire) { ------------------ | Branch (2726:7): [True: 0, False: 25.1k] ------------------ 2727| 0| ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 2728| 0| ompt_mutex_lock, omp_lock_hint_none, 2729| 0| __ompt_get_mutex_impl_type(user_lock), 2730| 0| (ompt_wait_id_t)(uintptr_t)user_lock, codeptr); 2731| 0| } 2732| 25.1k|#endif 2733| 25.1k|#if KMP_USE_INLINED_TAS 2734| 25.1k| if (tag == locktag_tas && !__kmp_env_consistency_check) { ------------------ | Branch (2734:7): [True: 0, False: 25.1k] | Branch (2734:29): [True: 0, False: 0] ------------------ 2735| 0| KMP_ACQUIRE_TAS_LOCK(user_lock, gtid); ------------------ | | 1080| 0| { \ | | 1081| 0| kmp_tas_lock_t *l = (kmp_tas_lock_t *)lock; \ | | 1082| 0| kmp_int32 tas_free = KMP_LOCK_FREE(tas); \ | | ------------------ | | | | 1165| 0|#define KMP_LOCK_FREE(type) (locktag_##type) | | ------------------ | | 1083| 0| kmp_int32 tas_busy = KMP_LOCK_BUSY(gtid + 1, tas); \ | | ------------------ | | | | 1168| 0|#define KMP_LOCK_BUSY(v, type) ((v) << KMP_LOCK_SHIFT | locktag_##type) | | | | ------------------ | | | | | | 1076| 0| 8 // number of low bits to be used as tag for direct locks | | | | ------------------ | | ------------------ | | 1084| 0| if (KMP_ATOMIC_LD_RLX(&l->lk.poll) != tas_free || \ | | ------------------ | | | | 1250| 0|#define KMP_ATOMIC_LD_RLX(p) KMP_ATOMIC_LD(p, relaxed) | | | | ------------------ | | | | | | 1245| 0|#define KMP_ATOMIC_LD(p, order) (p)->load(std::memory_order_##order) | | | | ------------------ | | ------------------ | | | Branch (1084:9): [True: 0, False: 0] | | ------------------ | | 1085| 0| !__kmp_atomic_compare_store_acq(&l->lk.poll, tas_free, tas_busy)) { \ | | ------------------ | | | Branch (1085:9): [True: 0, False: 0] | | ------------------ | | 1086| 0| kmp_uint32 spins; \ | | 1087| 0| KMP_FSYNC_PREPARE(l); \ | | ------------------ | | | | 332| 0|#define KMP_FSYNC_PREPARE(obj) ((void)0) | | ------------------ | | 1088| 0| KMP_INIT_YIELD(spins); \ | | ------------------ | | | | 1514| 0| { (count) = __kmp_yield_init; } | | ------------------ | | 1089| 0| kmp_backoff_t backoff = __kmp_spin_backoff_params; \ | | 1090| 0| do { \ | | 1091| 0| if (TCR_4(__kmp_nth) > \ | | ------------------ | | | | 1127| 0|#define TCR_4(a) (a) | | ------------------ | | | Branch (1091:13): [True: 0, False: 0] | | ------------------ | | 1092| 0| (__kmp_avail_proc ? __kmp_avail_proc : __kmp_xproc)) { \ | | ------------------ | | | Branch (1092:14): [True: 0, False: 0] | | ------------------ | | 1093| 0| KMP_YIELD(TRUE); \ | | ------------------ | | | | 1529| 0| { \ | | | | 1530| 0| KMP_CPU_PAUSE(); \ | | | | ------------------ | | | | | | 1498| 0|#define KMP_CPU_PAUSE() __kmp_x86_pause() | | | | ------------------ | | | | 1531| 0| if ((cond) && (KMP_TRY_YIELD)) \ | | | | ------------------ | | | | | | 1523| 0| ((__kmp_use_yield == 1) || (__kmp_use_yield == 2 && (KMP_OVERSUBSCRIBED))) | | | | | | ------------------ | | | | | | | | 1520| 0| (TCR_4(__kmp_nth) > (__kmp_avail_proc ? __kmp_avail_proc : __kmp_xproc)) | | | | | | | | ------------------ | | | | | | | | | | 1127| 0|#define TCR_4(a) (a) | | | | | | | | ------------------ | | | | | | | | | Branch (1520:24): [True: 0, False: 0] | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | | Branch (1523:4): [True: 0, False: 0] | | | | | | | Branch (1523:31): [True: 0, False: 0] | | | | | | | Branch (1523:55): [True: 0, False: 0] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (1531:9): [True: 0, Folded] | | | | ------------------ | | | | 1532| 0| __kmp_yield(); \ | | | | 1533| 0| } | | ------------------ | | 1094| 0| } else { \ | | 1095| 0| KMP_YIELD_SPIN(spins); \ | | ------------------ | | | | 1545| 0| { \ | | | | 1546| 0| KMP_CPU_PAUSE(); \ | | | | ------------------ | | | | | | 1498| 0|#define KMP_CPU_PAUSE() __kmp_x86_pause() | | | | ------------------ | | | | 1547| 0| if (KMP_TRY_YIELD) { \ | | | | ------------------ | | | | | | 1523| 0| ((__kmp_use_yield == 1) || (__kmp_use_yield == 2 && (KMP_OVERSUBSCRIBED))) | | | | | | ------------------ | | | | | | | | 1520| 0| (TCR_4(__kmp_nth) > (__kmp_avail_proc ? __kmp_avail_proc : __kmp_xproc)) | | | | | | | | ------------------ | | | | | | | | | | 1127| 0|#define TCR_4(a) (a) | | | | | | | | ------------------ | | | | | | | | | Branch (1520:24): [True: 0, False: 0] | | | | | | | | ------------------ | | | | | | ------------------ | | | | | | | Branch (1523:4): [True: 0, False: 0] | | | | | | | Branch (1523:31): [True: 0, False: 0] | | | | | | | Branch (1523:55): [True: 0, False: 0] | | | | | | ------------------ | | | | ------------------ | | | | 1548| 0| (count) -= 2; \ | | | | 1549| 0| if (!(count)) { \ | | | | ------------------ | | | | | Branch (1549:11): [True: 0, False: 0] | | | | ------------------ | | | | 1550| 0| __kmp_yield(); \ | | | | 1551| 0| (count) = __kmp_yield_next; \ | | | | 1552| 0| } \ | | | | 1553| 0| } \ | | | | 1554| 0| } | | ------------------ | | 1096| 0| } \ | | 1097| 0| __kmp_spin_backoff(&backoff); \ | | 1098| 0| } while ( \ | | 1099| 0| KMP_ATOMIC_LD_RLX(&l->lk.poll) != tas_free || \ | | ------------------ | | | | 1250| 0|#define KMP_ATOMIC_LD_RLX(p) KMP_ATOMIC_LD(p, relaxed) | | | | ------------------ | | | | | | 1245| 0|#define KMP_ATOMIC_LD(p, order) (p)->load(std::memory_order_##order) | | | | ------------------ | | ------------------ | | | Branch (1099:11): [True: 0, False: 0] | | ------------------ | | 1100| 0| !__kmp_atomic_compare_store_acq(&l->lk.poll, tas_free, tas_busy)); \ | | ------------------ | | | Branch (1100:11): [True: 0, False: 0] | | ------------------ | | 1101| 0| } \ | | 1102| 0| KMP_FSYNC_ACQUIRED(l); \ | | ------------------ | | | | 334| 0|#define KMP_FSYNC_ACQUIRED(obj) ((void)0) | | ------------------ | | 1103| 0| } ------------------ 2736| 0| } else 2737| |#elif KMP_USE_INLINED_FUTEX 2738| | if (tag == locktag_futex && !__kmp_env_consistency_check) { 2739| | KMP_ACQUIRE_FUTEX_LOCK(user_lock, gtid); 2740| | } else 2741| |#endif 2742| 25.1k| { 2743| 25.1k| __kmp_direct_set[tag]((kmp_dyna_lock_t *)user_lock, gtid); 2744| 25.1k| } 2745| |#if USE_ITT_BUILD 2746| | __kmp_itt_lock_acquired((kmp_user_lock_p)user_lock); 2747| |#endif 2748| 25.1k|#if OMPT_SUPPORT && OMPT_OPTIONAL 2749| 25.1k| if (ompt_enabled.ompt_callback_mutex_acquired) { ------------------ | Branch (2749:7): [True: 0, False: 25.1k] ------------------ 2750| 0| ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 2751| 0| ompt_mutex_lock, (ompt_wait_id_t)(uintptr_t)user_lock, codeptr); 2752| 0| } 2753| 25.1k|#endif 2754| | 2755| |#else // KMP_USE_DYNAMIC_LOCK 2756| | 2757| | kmp_user_lock_p lck; 2758| | 2759| | if ((__kmp_user_lock_kind == lk_tas) && 2760| | (sizeof(lck->tas.lk.poll) <= OMP_LOCK_T_SIZE)) { 2761| | lck = (kmp_user_lock_p)user_lock; 2762| | } 2763| |#if KMP_USE_FUTEX 2764| | else if ((__kmp_user_lock_kind == lk_futex) && 2765| | (sizeof(lck->futex.lk.poll) <= OMP_LOCK_T_SIZE)) { 2766| | lck = (kmp_user_lock_p)user_lock; 2767| | } 2768| |#endif 2769| | else { 2770| | lck = __kmp_lookup_user_lock(user_lock, "omp_set_lock"); 2771| | } 2772| | 2773| |#if USE_ITT_BUILD 2774| | __kmp_itt_lock_acquiring(lck); 2775| |#endif /* USE_ITT_BUILD */ 2776| |#if OMPT_SUPPORT && OMPT_OPTIONAL 2777| | // This is the case, if called from omp_init_lock_with_hint: 2778| | void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid); 2779| | if (!codeptr) 2780| | codeptr = OMPT_GET_RETURN_ADDRESS(0); 2781| | if (ompt_enabled.ompt_callback_mutex_acquire) { 2782| | ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)( 2783| | ompt_mutex_lock, omp_lock_hint_none, __ompt_get_mutex_impl_type(), 2784| | (ompt_wait_id_t)(uintptr_t)lck, codeptr); 2785| | } 2786| |#endif 2787| | 2788| | ACQUIRE_LOCK(lck, gtid); 2789| | 2790| |#if USE_ITT_BUILD 2791| | __kmp_itt_lock_acquired(lck); 2792| |#endif /* USE_ITT_BUILD */ 2793| | 2794| |#if OMPT_SUPPORT && OMPT_OPTIONAL 2795| | if (ompt_enabled.ompt_callback_mutex_acquired) { 2796| | ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)( 2797| | ompt_mutex_lock, (ompt_wait_id_t)(uintptr_t)lck, codeptr); 2798| | } 2799| |#endif 2800| | 2801| |#endif // KMP_USE_DYNAMIC_LOCK 2802| 25.1k|} __kmpc_unset_lock: 2915| 25.2k|void __kmpc_unset_lock(ident_t *loc, kmp_int32 gtid, void **user_lock) { 2916| 25.2k|#if KMP_USE_DYNAMIC_LOCK 2917| | 2918| 25.2k| int tag = KMP_EXTRACT_D_TAG(user_lock); ------------------ | | 1141| 25.2k| (*((kmp_dyna_lock_t *)(l)) & ((1 << KMP_LOCK_SHIFT) - 1) & \ | | ------------------ | | | | 1076| 25.2k| 8 // number of low bits to be used as tag for direct locks | | ------------------ | | 1142| 25.2k| -(*((kmp_dyna_lock_t *)(l)) & 1)) ------------------ 2919| |#if USE_ITT_BUILD 2920| | __kmp_itt_lock_releasing((kmp_user_lock_p)user_lock); 2921| |#endif 2922| 25.2k|#if KMP_USE_INLINED_TAS 2923| 25.2k| if (tag == locktag_tas && !__kmp_env_consistency_check) { ------------------ | Branch (2923:7): [True: 0, False: 25.2k] | Branch (2923:29): [True: 0, False: 0] ------------------ 2924| 0| KMP_RELEASE_TAS_LOCK(user_lock, gtid); ------------------ | | 1117| 0| { KMP_ATOMIC_ST_REL(&((kmp_tas_lock_t *)lock)->lk.poll, KMP_LOCK_FREE(tas)); } | | ------------------ | | | | 1251| 0|#define KMP_ATOMIC_ST_REL(p, v) KMP_ATOMIC_OP(store, p, v, release) | | | | ------------------ | | | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | | | ------------------ | | ------------------ ------------------ 2925| 0| } else 2926| |#elif KMP_USE_INLINED_FUTEX 2927| | if (tag == locktag_futex && !__kmp_env_consistency_check) { 2928| | KMP_RELEASE_FUTEX_LOCK(user_lock, gtid); 2929| | } else 2930| |#endif 2931| 25.2k| { 2932| 25.2k| __kmp_direct_unset[tag]((kmp_dyna_lock_t *)user_lock, gtid); 2933| 25.2k| } 2934| | 2935| 25.2k|#if OMPT_SUPPORT && OMPT_OPTIONAL 2936| | // This is the case, if called from omp_init_lock_with_hint: 2937| 25.2k| void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid); ------------------ | | 93| 25.2k|#define OMPT_LOAD_RETURN_ADDRESS(gtid) __ompt_load_return_address(gtid) ------------------ 2938| 25.2k| if (!codeptr) ------------------ | Branch (2938:7): [True: 25.2k, False: 0] ------------------ 2939| 25.2k| codeptr = OMPT_GET_RETURN_ADDRESS(0); ------------------ | | 108| 25.2k|#define OMPT_GET_RETURN_ADDRESS(level) __builtin_return_address(level) ------------------ 2940| 25.2k| if (ompt_enabled.ompt_callback_mutex_released) { ------------------ | Branch (2940:7): [True: 0, False: 25.2k] ------------------ 2941| 0| ompt_callbacks.ompt_callback(ompt_callback_mutex_released)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 2942| 0| ompt_mutex_lock, (ompt_wait_id_t)(uintptr_t)user_lock, codeptr); 2943| 0| } 2944| 25.2k|#endif 2945| | 2946| |#else // KMP_USE_DYNAMIC_LOCK 2947| | 2948| | kmp_user_lock_p lck; 2949| | 2950| | /* Can't use serial interval since not block structured */ 2951| | /* release the lock */ 2952| | 2953| | if ((__kmp_user_lock_kind == lk_tas) && 2954| | (sizeof(lck->tas.lk.poll) <= OMP_LOCK_T_SIZE)) { 2955| |#if KMP_OS_LINUX && \ 2956| | (KMP_ARCH_X86 || KMP_ARCH_X86_64 || KMP_ARCH_ARM || KMP_ARCH_AARCH64) 2957| |// "fast" path implemented to fix customer performance issue 2958| |#if USE_ITT_BUILD 2959| | __kmp_itt_lock_releasing((kmp_user_lock_p)user_lock); 2960| |#endif /* USE_ITT_BUILD */ 2961| | TCW_4(((kmp_user_lock_p)user_lock)->tas.lk.poll, 0); 2962| | KMP_MB(); 2963| | 2964| |#if OMPT_SUPPORT && OMPT_OPTIONAL 2965| | // This is the case, if called from omp_init_lock_with_hint: 2966| | void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid); 2967| | if (!codeptr) 2968| | codeptr = OMPT_GET_RETURN_ADDRESS(0); 2969| | if (ompt_enabled.ompt_callback_mutex_released) { 2970| | ompt_callbacks.ompt_callback(ompt_callback_mutex_released)( 2971| | ompt_mutex_lock, (ompt_wait_id_t)(uintptr_t)lck, codeptr); 2972| | } 2973| |#endif 2974| | 2975| | return; 2976| |#else 2977| | lck = (kmp_user_lock_p)user_lock; 2978| |#endif 2979| | } 2980| |#if KMP_USE_FUTEX 2981| | else if ((__kmp_user_lock_kind == lk_futex) && 2982| | (sizeof(lck->futex.lk.poll) <= OMP_LOCK_T_SIZE)) { 2983| | lck = (kmp_user_lock_p)user_lock; 2984| | } 2985| |#endif 2986| | else { 2987| | lck = __kmp_lookup_user_lock(user_lock, "omp_unset_lock"); 2988| | } 2989| | 2990| |#if USE_ITT_BUILD 2991| | __kmp_itt_lock_releasing(lck); 2992| |#endif /* USE_ITT_BUILD */ 2993| | 2994| | RELEASE_LOCK(lck, gtid); 2995| | 2996| |#if OMPT_SUPPORT && OMPT_OPTIONAL 2997| | // This is the case, if called from omp_init_lock_with_hint: 2998| | void *codeptr = OMPT_LOAD_RETURN_ADDRESS(gtid); 2999| | if (!codeptr) 3000| | codeptr = OMPT_GET_RETURN_ADDRESS(0); 3001| | if (ompt_enabled.ompt_callback_mutex_released) { 3002| | ompt_callbacks.ompt_callback(ompt_callback_mutex_released)( 3003| | ompt_mutex_lock, (ompt_wait_id_t)(uintptr_t)lck, codeptr); 3004| | } 3005| |#endif 3006| | 3007| |#endif // KMP_USE_DYNAMIC_LOCK 3008| 25.2k|} kmp_csupport.cpp:_ZL25__kmp_init_lock_with_hintP5identPPv18kmp_dyna_lockseq_t: 2315| 30.8k| kmp_dyna_lockseq_t seq) { 2316| 30.8k| if (KMP_IS_D_LOCK(seq)) { ------------------ | | 1104| 30.8k| ((seq) >= KMP_FIRST_D_LOCK && (seq) <= KMP_LAST_D_LOCK) | | ------------------ | | | | 1077| 61.6k|#define KMP_FIRST_D_LOCK lockseq_tas | | ------------------ | | ((seq) >= KMP_FIRST_D_LOCK && (seq) <= KMP_LAST_D_LOCK) | | ------------------ | | | | 1057| 30.8k|#define KMP_LAST_D_LOCK lockseq_rtm_spin | | ------------------ | | | Branch (1104:4): [True: 30.8k, False: 0] | | | Branch (1104:33): [True: 0, False: 30.8k] | | ------------------ ------------------ 2317| 0| KMP_INIT_D_LOCK(lock, seq); ------------------ | | 1158| 0| __kmp_direct_init[KMP_GET_D_TAG(seq)]((kmp_dyna_lock_t *)l, seq) | | ------------------ | | | | 1108| 0|#define KMP_GET_D_TAG(seq) ((seq) << 1 | 1) | | ------------------ ------------------ 2318| |#if USE_ITT_BUILD 2319| | __kmp_itt_lock_creating((kmp_user_lock_p)lock, NULL); 2320| |#endif 2321| 30.8k| } else { 2322| 30.8k| KMP_INIT_I_LOCK(lock, seq); ------------------ | | 1162| 30.8k| __kmp_direct_init[0]((kmp_dyna_lock_t *)(l), seq) ------------------ 2323| |#if USE_ITT_BUILD 2324| | kmp_indirect_lock_t *ilk = KMP_LOOKUP_I_LOCK(lock); 2325| | __kmp_itt_lock_creating(ilk->lock, loc); 2326| |#endif 2327| 30.8k| } 2328| 30.8k|} __kmp_eq_4: 2854| 11.9M|kmp_uint32 __kmp_eq_4(kmp_uint32 value, kmp_uint32 checker) { 2855| 11.9M| return value == checker; 2856| 11.9M|} __kmp_neq_4: 2858| 1.21k|kmp_uint32 __kmp_neq_4(kmp_uint32 value, kmp_uint32 checker) { 2859| 1.21k| return value != checker; 2860| 1.21k|} __kmp_wait_4: 2878| 2.71k|) { 2879| | // note: we may not belong to a team at this point 2880| 2.71k| volatile kmp_uint32 *spin = spinner; 2881| 2.71k| kmp_uint32 check = checker; 2882| 2.71k| kmp_uint32 spins; 2883| 2.71k| kmp_uint32 (*f)(kmp_uint32, kmp_uint32) = pred; 2884| 2.71k| kmp_uint32 r; 2885| 2.71k| kmp_uint64 time; 2886| | 2887| 2.71k| KMP_FSYNC_SPIN_INIT(obj, CCAST(kmp_uint32 *, spin)); ------------------ | | 337| 2.71k|#define KMP_FSYNC_SPIN_INIT(obj, spin) ((void)0) ------------------ 2888| 2.71k| KMP_INIT_YIELD(spins); ------------------ | | 1514| 2.71k| { (count) = __kmp_yield_init; } ------------------ 2889| 2.71k| KMP_INIT_BACKOFF(time); ------------------ | | 1517| 2.71k| { (time) = __kmp_pause_init; } ------------------ 2890| | // main wait spin loop 2891| 11.9M| while (!f(r = TCR_4(*spin), check)) { ------------------ | | 1127| 11.9M|#define TCR_4(a) (a) ------------------ | Branch (2891:10): [True: 11.9M, False: 2.71k] ------------------ 2892| 11.9M| KMP_FSYNC_SPIN_PREPARE(obj); ------------------ | | 338| 11.9M|#define KMP_FSYNC_SPIN_PREPARE(obj) ((void)0) ------------------ 2893| | /* GEH - remove this since it was accidentally introduced when kmp_wait was 2894| | split. It causes problems with infinite recursion because of exit lock */ 2895| | /* if ( TCR_4(__kmp_global.g.g_done) && __kmp_global.g.g_abort) 2896| | __kmp_abort_thread(); */ 2897| 11.9M| KMP_YIELD_OVERSUB_ELSE_SPIN(spins, time); ------------------ | | 1564| 11.9M| { \ | | 1565| 11.9M| if (__kmp_tpause_enabled) { \ | | ------------------ | | | Branch (1565:9): [True: 0, False: 11.9M] | | ------------------ | | 1566| 0| if (KMP_OVERSUBSCRIBED) { \ | | ------------------ | | | | 1520| 0| (TCR_4(__kmp_nth) > (__kmp_avail_proc ? __kmp_avail_proc : __kmp_xproc)) | | | | ------------------ | | | | | | 1127| 0|#define TCR_4(a) (a) | | | | ------------------ | | | | | Branch (1520:3): [True: 0, False: 0] | | | | | Branch (1520:24): [True: 0, False: 0] | | | | ------------------ | | ------------------ | | 1567| 0| __kmp_tpause(0, (time)); \ | | 1568| 0| } else { \ | | 1569| 0| __kmp_tpause(__kmp_tpause_hint, (time)); \ | | 1570| 0| } \ | | 1571| 0| (time) = (time << 1 | 1) & KMP_TPAUSE_MAX_MASK; \ | | ------------------ | | | | 1562| 0|#define KMP_TPAUSE_MAX_MASK ((kmp_uint64)0xFFFF) | | ------------------ | | 1572| 11.9M| } else { \ | | 1573| 11.9M| KMP_CPU_PAUSE(); \ | | ------------------ | | | | 1498| 11.9M|#define KMP_CPU_PAUSE() __kmp_x86_pause() | | ------------------ | | 1574| 11.9M| if ((KMP_TRY_YIELD_OVERSUB)) { \ | | ------------------ | | | | 1526| 11.9M| ((__kmp_use_yield == 1 || __kmp_use_yield == 2) && (KMP_OVERSUBSCRIBED)) | | | | ------------------ | | | | | | 1520| 18.4E| (TCR_4(__kmp_nth) > (__kmp_avail_proc ? __kmp_avail_proc : __kmp_xproc)) | | | | | | ------------------ | | | | | | | | 1127| 11.9M|#define TCR_4(a) (a) | | | | | | ------------------ | | | | | | | Branch (1520:24): [True: 12.0M, False: 18.4E] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (1526:5): [True: 11.9M, False: 0] | | | | | Branch (1526:29): [True: 0, False: 0] | | | | | Branch (1526:54): [True: 0, False: 11.9M] | | | | ------------------ | | ------------------ | | 1575| 0| __kmp_yield(); \ | | 1576| 12.1M| } else if (__kmp_use_yield == 1) { \ | | ------------------ | | | Branch (1576:18): [True: 12.1M, False: 18.4E] | | ------------------ | | 1577| 12.1M| (count) -= 2; \ | | 1578| 12.1M| if (!(count)) { \ | | ------------------ | | | Branch (1578:13): [True: 74.7k, False: 12.0M] | | ------------------ | | 1579| 74.7k| __kmp_yield(); \ | | 1580| 74.7k| (count) = __kmp_yield_next; \ | | 1581| 74.7k| } \ | | 1582| 12.1M| } \ | | 1583| 11.9M| } \ | | 1584| 11.9M| } ------------------ 2898| 11.9M| } 2899| 2.71k| KMP_FSYNC_SPIN_ACQUIRED(obj); ------------------ | | 339| 2.71k|#define KMP_FSYNC_SPIN_ACQUIRED(obj) ((void)0) ------------------ 2900| 2.71k| return r; 2901| 2.71k|} __kmp_env_get: 84| 3|char *__kmp_env_get(char const *name) { 85| | 86| 3| char *result = NULL; 87| | 88| 3|#if KMP_OS_UNIX 89| 3| char const *value = getenv(name); 90| 3| if (value != NULL) { ------------------ | Branch (90:7): [True: 2, False: 1] ------------------ 91| 2| size_t len = KMP_STRLEN(value) + 1; ------------------ | | 68| 2|#define KMP_STRLEN strlen ------------------ 92| 2| result = (char *)KMP_INTERNAL_MALLOC(len); ------------------ | | 110| 2|#define KMP_INTERNAL_MALLOC(sz) malloc(sz) ------------------ 93| 2| if (result == NULL) { ------------------ | Branch (93:9): [True: 0, False: 2] ------------------ 94| 0| KMP_FATAL(MemoryAllocFailed); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 95| 0| } 96| 2| KMP_STRNCPY_S(result, len, value, len); ------------------ | | 64| 2|#define KMP_STRNCPY_S(dst, bsz, src, cnt) strncpy(dst, src, cnt) ------------------ 97| 2| } 98| |#elif KMP_OS_WINDOWS 99| | /* We use GetEnvironmentVariable for Windows* OS instead of getenv because the 100| | act of loading a DLL on Windows* OS makes any user-set environment 101| | variables (i.e. with putenv()) unavailable. getenv() apparently gets a 102| | clean copy of the env variables as they existed at the start of the run. 103| | JH 12/23/2002 */ 104| | DWORD rc; 105| | rc = GetEnvironmentVariable(name, NULL, 0); 106| | if (!rc) { 107| | DWORD error = GetLastError(); 108| | if (error != ERROR_ENVVAR_NOT_FOUND) { 109| | __kmp_fatal(KMP_MSG(CantGetEnvVar, name), KMP_ERR(error), __kmp_msg_null); 110| | } 111| | // Variable is not found, it's ok, just continue. 112| | } else { 113| | DWORD len = rc; 114| | result = (char *)KMP_INTERNAL_MALLOC(len); 115| | if (result == NULL) { 116| | KMP_FATAL(MemoryAllocFailed); 117| | } 118| | rc = GetEnvironmentVariable(name, result, len); 119| | if (!rc) { 120| | // GetEnvironmentVariable() may return 0 if variable is empty. 121| | // In such a case GetLastError() returns ERROR_SUCCESS. 122| | DWORD error = GetLastError(); 123| | if (error != ERROR_SUCCESS) { 124| | // Unexpected error. The variable should be in the environment, 125| | // and buffer should be large enough. 126| | __kmp_fatal(KMP_MSG(CantGetEnvVar, name), KMP_ERR(error), 127| | __kmp_msg_null); 128| | KMP_INTERNAL_FREE((void *)result); 129| | result = NULL; 130| | } 131| | } 132| | } 133| |#else 134| |#error Unknown or unsupported OS. 135| |#endif 136| | 137| 3| return result; 138| | 139| 3|} // func __kmp_env_get __kmp_env_set: 173| 1|void __kmp_env_set(char const *name, char const *value, int overwrite) { 174| | 175| 1|#if KMP_OS_UNIX 176| 1| int rc = setenv(name, value, overwrite); 177| 1| if (rc != 0) { ------------------ | Branch (177:7): [True: 0, False: 1] ------------------ 178| | // Dead code. I tried to put too many variables into Linux* OS 179| | // environment on IA-32 architecture. When application consumes 180| | // more than ~2.5 GB of memory, entire system feels bad. Sometimes 181| | // application is killed (by OS?), sometimes system stops 182| | // responding... But this error message never appears. --ln 183| 0| __kmp_fatal(KMP_MSG(CantSetEnvVar, name), KMP_HNT(NotEnoughMemory), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ __kmp_fatal(KMP_MSG(CantSetEnvVar, name), KMP_HNT(NotEnoughMemory), ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 184| 0| __kmp_msg_null); 185| 0| } 186| |#elif KMP_OS_WINDOWS 187| | BOOL rc; 188| | if (!overwrite) { 189| | rc = GetEnvironmentVariable(name, NULL, 0); 190| | if (rc) { 191| | // Variable exists, do not overwrite. 192| | return; 193| | } 194| | DWORD error = GetLastError(); 195| | if (error != ERROR_ENVVAR_NOT_FOUND) { 196| | __kmp_fatal(KMP_MSG(CantGetEnvVar, name), KMP_ERR(error), __kmp_msg_null); 197| | } 198| | } 199| | rc = SetEnvironmentVariable(name, value); 200| | if (!rc) { 201| | DWORD error = GetLastError(); 202| | __kmp_fatal(KMP_MSG(CantSetEnvVar, name), KMP_ERR(error), __kmp_msg_null); 203| | } 204| |#else 205| |#error Unknown or unsupported OS. 206| |#endif 207| | 208| 1|} // func __kmp_env_set __kmp_env_unset: 210| 1|void __kmp_env_unset(char const *name) { 211| | 212| 1|#if KMP_OS_UNIX 213| 1| unsetenv(name); 214| |#elif KMP_OS_WINDOWS 215| | BOOL rc = SetEnvironmentVariable(name, NULL); 216| | if (!rc) { 217| | DWORD error = GetLastError(); 218| | __kmp_fatal(KMP_MSG(CantSetEnvVar, name), KMP_ERR(error), __kmp_msg_null); 219| | } 220| |#else 221| |#error Unknown or unsupported OS. 222| |#endif 223| | 224| 1|} // func __kmp_env_unset __kmp_env_blk_init: 431| 2|) { 432| | 433| 2| if (bulk != NULL) { ------------------ | Branch (433:7): [True: 0, False: 2] ------------------ 434| 0| ___kmp_env_blk_parse_string(block, bulk); 435| 2| } else { 436| 2|#if KMP_OS_UNIX 437| 2| ___kmp_env_blk_parse_unix(block, environ); 438| |#elif KMP_OS_WINDOWS 439| | { 440| | char *mem = GetEnvironmentStrings(); 441| | if (mem == NULL) { 442| | DWORD error = GetLastError(); 443| | __kmp_fatal(KMP_MSG(CantGetEnvironment), KMP_ERR(error), 444| | __kmp_msg_null); 445| | } 446| | ___kmp_env_blk_parse_windows(block, mem); 447| | FreeEnvironmentStrings(mem); 448| | } 449| |#else 450| |#error Unknown or unsupported OS. 451| |#endif 452| 2| } 453| | 454| 2|} // __kmp_env_blk_init __kmp_env_blk_sort: 463| 1|) { 464| | 465| 1| qsort(CCAST(kmp_env_var_t *, block->vars), block->count, ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 466| 1| sizeof(kmp_env_var_t), 467| 1| (int (*)(void const *, void const *)) & ___kmp_env_var_cmp); 468| | 469| 1|} // __kmp_env_block_sort __kmp_env_blk_free: 473| 2|) { 474| | 475| 2| KMP_INTERNAL_FREE(CCAST(kmp_env_var_t *, block->vars)); ------------------ | | 111| 2|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 476| 2| __kmp_str_free(&(block->bulk)); 477| | 478| 2| block->count = 0; 479| 2| block->vars = NULL; 480| | 481| 2|} // __kmp_env_blk_free __kmp_env_blk_var: 486| 3|) { 487| | 488| 3| int i; 489| 105| for (i = 0; i < block->count; ++i) { ------------------ | Branch (489:15): [True: 102, False: 3] ------------------ 490| 102| if (strcmp(block->vars[i].name, name) == 0) { ------------------ | Branch (490:9): [True: 0, False: 102] ------------------ 491| 0| return block->vars[i].value; 492| 0| } 493| 102| } 494| 3| return NULL; 495| | 496| 3|} // __kmp_env_block_var kmp_environment.cpp:_ZL8allocatem: 76| 4|static inline void *allocate(size_t size) { 77| 4| void *ptr = KMP_INTERNAL_MALLOC(size); ------------------ | | 110| 4|#define KMP_INTERNAL_MALLOC(sz) malloc(sz) ------------------ 78| 4| if (ptr == NULL) { ------------------ | Branch (78:7): [True: 0, False: 4] ------------------ 79| 0| KMP_FATAL(MemoryAllocFailed); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 80| 0| } 81| 4| return ptr; 82| 4|} // allocate kmp_environment.cpp:_ZL25___kmp_env_blk_parse_unixP13__kmp_env_blkPPc: 383| 2|) { 384| 2| char *bulk = NULL; 385| 2| kmp_env_var_t *vars = NULL; 386| 2| int count = 0; 387| 2| size_t size = 0; // Size of bulk. 388| | 389| | // Count number of variables and length of required bulk. 390| 2| { 391| 69| while (env[count] != NULL) { ------------------ | Branch (391:12): [True: 67, False: 2] ------------------ 392| 67| size += KMP_STRLEN(env[count]) + 1; ------------------ | | 68| 67|#define KMP_STRLEN strlen ------------------ 393| 67| ++count; 394| 67| } 395| 2| } 396| | 397| | // Allocate memory. 398| 2| bulk = (char *)allocate(size); 399| 2| vars = (kmp_env_var_t *)allocate(count * sizeof(kmp_env_var_t)); 400| | 401| | // Loop thru all the vars. 402| 2| { 403| 2| char *var; // Pointer to beginning of var. 404| 2| char *name; // Pointer to name of variable. 405| 2| char *value; // Pointer to value. 406| 2| size_t len; // Length of variable. 407| 2| int i; 408| 2| var = bulk; 409| 69| for (i = 0; i < count; ++i) { ------------------ | Branch (409:17): [True: 67, False: 2] ------------------ 410| | // Copy variable to bulk. 411| 67| len = KMP_STRLEN(env[i]); ------------------ | | 68| 67|#define KMP_STRLEN strlen ------------------ 412| 67| KMP_MEMCPY_S(var, size, env[i], len + 1); ------------------ | | 60| 67|#define KMP_MEMCPY_S(dst, bsz, src, cnt) memcpy(dst, src, cnt) ------------------ 413| | // Save found variable in vars array. 414| 67| __kmp_str_split(var, '=', &name, &value); 415| 67| vars[i].name = name; 416| 67| vars[i].value = value; 417| | // Move pointer. 418| 67| var += len + 1; 419| 67| } 420| 2| } 421| | 422| | // Fill out result. 423| 2| block->bulk = bulk; 424| 2| block->vars = vars; 425| 2| block->count = count; 426| 2|} kmp_environment.cpp:_ZL18___kmp_env_var_cmpPK13__kmp_env_varS1_: 457| 122| kmp_env_var_t const *lhs, kmp_env_var_t const *rhs) { 458| 122| return strcmp(lhs->name, rhs->name); 459| 122|} omp_get_max_threads: 357| 4.36k|int FTN_STDCALL KMP_EXPAND_NAME(FTN_GET_MAX_THREADS)(void) { 358| |#ifdef KMP_STUB 359| | return 1; 360| |#else 361| 4.36k| int gtid; 362| 4.36k| kmp_info_t *thread; 363| 4.36k| if (!TCR_4(__kmp_init_middle)) { ------------------ | | 1127| 4.36k|#define TCR_4(a) (a) ------------------ | Branch (363:7): [True: 1, False: 4.36k] ------------------ 364| 1| __kmp_middle_initialize(); 365| 1| } 366| 4.36k| gtid = __kmp_entry_gtid(); ------------------ | | 3467| 4.36k|#define __kmp_entry_gtid() __kmp_get_global_thread_id_reg() ------------------ 367| 4.36k| thread = __kmp_threads[gtid]; 368| 4.36k|#if KMP_AFFINITY_SUPPORTED 369| 4.36k| if (thread->th.th_team->t.t_level == 0 && !__kmp_affinity.flags.reset) { ------------------ | Branch (369:7): [True: 4.36k, False: 0] | Branch (369:45): [True: 4.36k, False: 0] ------------------ 370| 4.36k| __kmp_assign_root_init_mask(); 371| 4.36k| } 372| 4.36k|#endif 373| | // return thread -> th.th_team -> t.t_current_task[ 374| | // thread->th.th_info.ds.ds_tid ] -> icvs.nproc; 375| 4.36k| return thread->th.th_current_task->td_icvs.nproc; 376| 4.36k|#endif 377| 4.36k|} omp_init_lock: 1136| 30.8k|void FTN_STDCALL KMP_EXPAND_NAME(FTN_INIT_LOCK)(void **user_lock) { 1137| |#ifdef KMP_STUB 1138| | *((kmp_stub_lock_t *)user_lock) = UNLOCKED; 1139| |#else 1140| 30.8k| int gtid = __kmp_entry_gtid(); ------------------ | | 3467| 30.8k|#define __kmp_entry_gtid() __kmp_get_global_thread_id_reg() ------------------ 1141| 30.8k|#if OMPT_SUPPORT && OMPT_OPTIONAL 1142| 30.8k| OMPT_STORE_RETURN_ADDRESS(gtid); ------------------ | | 92| 30.8k| OmptReturnAddressGuard ReturnAddressGuard{gtid, __builtin_return_address(0)}; ------------------ 1143| 30.8k|#endif 1144| | __kmpc_init_lock(NULL, gtid, user_lock); 1145| 30.8k|#endif 1146| 30.8k|} omp_destroy_lock: 1161| 30.8k|void FTN_STDCALL KMP_EXPAND_NAME(FTN_DESTROY_LOCK)(void **user_lock) { 1162| |#ifdef KMP_STUB 1163| | *((kmp_stub_lock_t *)user_lock) = UNINIT; 1164| |#else 1165| 30.8k| int gtid = __kmp_entry_gtid(); ------------------ | | 3467| 30.8k|#define __kmp_entry_gtid() __kmp_get_global_thread_id_reg() ------------------ 1166| 30.8k|#if OMPT_SUPPORT && OMPT_OPTIONAL 1167| 30.8k| OMPT_STORE_RETURN_ADDRESS(gtid); ------------------ | | 92| 30.8k| OmptReturnAddressGuard ReturnAddressGuard{gtid, __builtin_return_address(0)}; ------------------ 1168| 30.8k|#endif 1169| | __kmpc_destroy_lock(NULL, gtid, user_lock); 1170| 30.8k|#endif 1171| 30.8k|} omp_set_lock: 1185| 25.1k|void FTN_STDCALL KMP_EXPAND_NAME(FTN_SET_LOCK)(void **user_lock) { 1186| |#ifdef KMP_STUB 1187| | if (*((kmp_stub_lock_t *)user_lock) == UNINIT) { 1188| | // TODO: Issue an error. 1189| | } 1190| | if (*((kmp_stub_lock_t *)user_lock) != UNLOCKED) { 1191| | // TODO: Issue an error. 1192| | } 1193| | *((kmp_stub_lock_t *)user_lock) = LOCKED; 1194| |#else 1195| 25.1k| int gtid = __kmp_entry_gtid(); ------------------ | | 3467| 25.1k|#define __kmp_entry_gtid() __kmp_get_global_thread_id_reg() ------------------ 1196| 25.1k|#if OMPT_SUPPORT && OMPT_OPTIONAL 1197| 25.1k| OMPT_STORE_RETURN_ADDRESS(gtid); ------------------ | | 92| 25.1k| OmptReturnAddressGuard ReturnAddressGuard{gtid, __builtin_return_address(0)}; ------------------ 1198| 25.1k|#endif 1199| | __kmpc_set_lock(NULL, gtid, user_lock); 1200| 25.1k|#endif 1201| 25.1k|} omp_unset_lock: 1218| 25.2k|void FTN_STDCALL KMP_EXPAND_NAME(FTN_UNSET_LOCK)(void **user_lock) { 1219| |#ifdef KMP_STUB 1220| | if (*((kmp_stub_lock_t *)user_lock) == UNINIT) { 1221| | // TODO: Issue an error. 1222| | } 1223| | if (*((kmp_stub_lock_t *)user_lock) == UNLOCKED) { 1224| | // TODO: Issue an error. 1225| | } 1226| | *((kmp_stub_lock_t *)user_lock) = UNLOCKED; 1227| |#else 1228| 25.2k| int gtid = __kmp_entry_gtid(); ------------------ | | 3467| 25.2k|#define __kmp_entry_gtid() __kmp_get_global_thread_id_reg() ------------------ 1229| 25.2k|#if OMPT_SUPPORT && OMPT_OPTIONAL 1230| 25.2k| OMPT_STORE_RETURN_ADDRESS(gtid); ------------------ | | 92| 25.2k| OmptReturnAddressGuard ReturnAddressGuard{gtid, __builtin_return_address(0)}; ------------------ 1231| 25.2k|#endif 1232| | __kmpc_unset_lock(NULL, gtid, user_lock); 1233| 25.2k|#endif 1234| 25.2k|} __kmp_i18n_catopen: 62| 1|void __kmp_i18n_catopen() { 63| 1| if (status == KMP_I18N_CLOSED) { ------------------ | Branch (63:7): [True: 1, False: 0] ------------------ 64| 1| __kmp_acquire_bootstrap_lock(&lock); 65| 1| if (status == KMP_I18N_CLOSED) { ------------------ | Branch (65:9): [True: 1, False: 0] ------------------ 66| 1| __kmp_i18n_do_catopen(); 67| 1| } 68| 1| __kmp_release_bootstrap_lock(&lock); 69| 1| } 70| 1|} // func __kmp_i18n_catopen __kmp_i18n_catclose: 178| 1|void __kmp_i18n_catclose() { 179| 1| if (status == KMP_I18N_OPENED) { ------------------ | Branch (179:7): [True: 0, False: 1] ------------------ 180| 0| KMP_DEBUG_ASSERT(cat != KMP_I18N_NULLCAT); 181| 0| catclose(cat); 182| 0| cat = KMP_I18N_NULLCAT; ------------------ | | 78| 0|#define KMP_I18N_NULLCAT ((nl_catd)(-1)) ------------------ 183| 0| } 184| 1| status = KMP_I18N_CLOSED; 185| 1|} // func __kmp_i18n_catclose __kmp_i18n_catgets: 187| 10|char const *__kmp_i18n_catgets(kmp_i18n_id_t id) { 188| | 189| 10| int section = get_section(id); ------------------ | | 33| 10|#define get_section(id) ((id) >> 16) ------------------ 190| 10| int number = get_number(id); ------------------ | | 34| 10|#define get_number(id) ((id)&0xFFFF) ------------------ 191| 10| char const *message = NULL; 192| | 193| 10| if (1 <= section && section <= __kmp_i18n_default_table.size) { ------------------ | Branch (193:7): [True: 10, False: 0] | Branch (193:23): [True: 10, False: 0] ------------------ 194| 10| if (1 <= number && number <= __kmp_i18n_default_table.sect[section].size) { ------------------ | Branch (194:9): [True: 10, False: 0] | Branch (194:24): [True: 10, False: 0] ------------------ 195| 10| if (status == KMP_I18N_CLOSED) { ------------------ | Branch (195:11): [True: 1, False: 9] ------------------ 196| 1| __kmp_i18n_catopen(); 197| 1| } 198| 10| if (status == KMP_I18N_OPENED) { ------------------ | Branch (198:11): [True: 0, False: 10] ------------------ 199| 0| message = catgets(cat, section, number, 200| 0| __kmp_i18n_default_table.sect[section].str[number]); 201| 0| } 202| 10| if (message == NULL) { ------------------ | Branch (202:11): [True: 10, False: 0] ------------------ 203| 10| message = __kmp_i18n_default_table.sect[section].str[number]; 204| 10| } 205| 10| } 206| 10| } 207| 10| if (message == NULL) { ------------------ | Branch (207:7): [True: 0, False: 10] ------------------ 208| 0| message = no_message_available; 209| 0| } 210| 10| return message; 211| | 212| 10|} // func __kmp_i18n_catgets __kmp_msg_format: 624| 1|kmp_msg_t __kmp_msg_format(unsigned id_arg, ...) { 625| | 626| 1| kmp_msg_t msg; 627| 1| va_list args; 628| 1| kmp_str_buf_t buffer; 629| 1| __kmp_str_buf_init(&buffer); ------------------ | | 41| 1| { \ | | 42| 1| (b)->str = (b)->bulk; \ | | 43| 1| (b)->size = sizeof((b)->bulk); \ | | 44| 1| (b)->used = 0; \ | | 45| 1| (b)->bulk[0] = 0; \ | | 46| 1| } ------------------ 630| | 631| 1| va_start(args, id_arg); 632| | 633| | // We use unsigned for the ID argument and explicitly cast it here to the 634| | // right enumerator because variadic functions are not compatible with 635| | // default promotions. 636| 1| kmp_i18n_id_t id = (kmp_i18n_id_t)id_arg; 637| | 638| 1|#if KMP_OS_UNIX 639| | // On Linux* OS and OS X*, printf() family functions process parameter 640| | // numbers, for example: "%2$s %1$s". 641| 1| __kmp_str_buf_vprint(&buffer, __kmp_i18n_catgets(id), args); 642| |#elif KMP_OS_WINDOWS 643| | // On Windows, printf() family functions does not recognize GNU style 644| | // parameter numbers, so we have to use FormatMessage() instead. It recognizes 645| | // parameter numbers, e. g.: "%2!s! "%1!s!". 646| | { 647| | LPTSTR str = NULL; 648| | int len; 649| | FormatMessage(FORMAT_MESSAGE_FROM_STRING | FORMAT_MESSAGE_ALLOCATE_BUFFER, 650| | __kmp_i18n_catgets(id), 0, 0, (LPTSTR)(&str), 0, &args); 651| | len = ___strip_crs(str); 652| | __kmp_str_buf_cat(&buffer, str, len); 653| | LocalFree(str); 654| | } 655| |#else 656| |#error 657| |#endif 658| 1| va_end(args); 659| 1| __kmp_str_buf_detach(&buffer); 660| | 661| 1| msg.type = (kmp_msg_type_t)(id >> 16); 662| 1| msg.num = id & 0xFFFF; 663| 1| msg.str = buffer.str; 664| 1| msg.len = buffer.used; 665| | 666| 1| return msg; 667| | 668| 1|} // __kmp_msg_format kmp_i18n.cpp:_ZL21__kmp_i18n_do_catopenv: 89| 1|void __kmp_i18n_do_catopen() { 90| 1| int english = 0; 91| 1| char *lang = __kmp_env_get("LANG"); 92| | // TODO: What about LC_ALL or LC_MESSAGES? 93| | 94| 1| KMP_DEBUG_ASSERT(status == KMP_I18N_CLOSED); 95| 1| KMP_DEBUG_ASSERT(cat == KMP_I18N_NULLCAT); 96| | 97| 1| english = lang == NULL || // In all these cases English language is used. ------------------ | Branch (97:13): [True: 1, False: 0] ------------------ 98| 0| strcmp(lang, "") == 0 || strcmp(lang, " ") == 0 || ------------------ | Branch (98:13): [True: 0, False: 0] | Branch (98:38): [True: 0, False: 0] ------------------ 99| | // Workaround for Fortran RTL bug DPD200137873 "Fortran runtime 100| | // resets LANG env var to space if it is not set". 101| 0| strcmp(lang, "C") == 0 || strcmp(lang, "POSIX") == 0; ------------------ | Branch (101:13): [True: 0, False: 0] | Branch (101:39): [True: 0, False: 0] ------------------ 102| | 103| 1| if (!english) { // English language is not yet detected, let us continue. ------------------ | Branch (103:7): [True: 0, False: 1] ------------------ 104| | // Format of LANG is: [language[_territory][.codeset][@modifier]] 105| | // Strip all parts except language. 106| 0| char *tail = NULL; 107| 0| __kmp_str_split(lang, '@', &lang, &tail); 108| 0| __kmp_str_split(lang, '.', &lang, &tail); 109| 0| __kmp_str_split(lang, '_', &lang, &tail); 110| 0| english = (strcmp(lang, "en") == 0); 111| 0| } 112| | 113| 1| KMP_INTERNAL_FREE(lang); ------------------ | | 111| 1|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 114| | 115| | // Do not try to open English catalog because internal messages are 116| | // exact copy of messages in English catalog. 117| 1| if (english) { ------------------ | Branch (117:7): [True: 1, False: 0] ------------------ 118| 1| status = KMP_I18N_ABSENT; // mark catalog as absent so it will not 119| | // be re-opened. 120| 1| return; 121| 1| } 122| | 123| 0| cat = catopen(name, 0); 124| | // TODO: Why do we pass 0 in flags? 125| 0| status = (cat == KMP_I18N_NULLCAT ? KMP_I18N_ABSENT : KMP_I18N_OPENED); ------------------ | | 78| 0|#define KMP_I18N_NULLCAT ((nl_catd)(-1)) ------------------ | Branch (125:13): [True: 0, False: 0] ------------------ 126| | 127| 0| if (status == KMP_I18N_ABSENT) { ------------------ | Branch (127:7): [True: 0, False: 0] ------------------ 128| 0| if (__kmp_generate_warnings > kmp_warnings_low) { ------------------ | Branch (128:9): [True: 0, False: 0] ------------------ 129| | // AC: only issue warning in case explicitly asked to 130| 0| int error = errno; // Save errno immediately. 131| 0| char *nlspath = __kmp_env_get("NLSPATH"); 132| 0| char *lang = __kmp_env_get("LANG"); 133| | 134| | // Infinite recursion will not occur -- status is KMP_I18N_ABSENT now, so 135| | // __kmp_i18n_catgets() will not try to open catalog, but will return 136| | // default message. 137| 0| kmp_msg_t err_code = KMP_ERR(error); ------------------ | | 125| 0|#define KMP_ERR KMP_SYSERRCODE | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ ------------------ 138| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(CantOpenMessageCatalog, name), err_code, ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ 139| 0| KMP_HNT(CheckEnvVar, "NLSPATH", nlspath), ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 140| 0| KMP_HNT(CheckEnvVar, "LANG", lang), __kmp_msg_null); ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 141| 0| if (__kmp_generate_warnings == kmp_warnings_off) { ------------------ | Branch (141:11): [True: 0, False: 0] ------------------ 142| 0| __kmp_str_free(&err_code.str); 143| 0| } 144| | 145| 0| KMP_INFORM(WillUseDefaultMessages); ------------------ | | 143| 0| __kmp_msg(kmp_ms_inform, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 146| 0| KMP_INTERNAL_FREE(nlspath); ------------------ | | 111| 0|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 147| 0| KMP_INTERNAL_FREE(lang); ------------------ | | 111| 0|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 148| 0| } 149| 0| } else { // status == KMP_I18N_OPENED 150| 0| int section = get_section(kmp_i18n_prp_Version); ------------------ | | 33| 0|#define get_section(id) ((id) >> 16) ------------------ 151| 0| int number = get_number(kmp_i18n_prp_Version); ------------------ | | 34| 0|#define get_number(id) ((id)&0xFFFF) ------------------ 152| 0| char const *expected = __kmp_i18n_default_table.sect[section].str[number]; 153| | // Expected version of the catalog. 154| 0| kmp_str_buf_t version; // Actual version of the catalog. 155| 0| __kmp_str_buf_init(&version); ------------------ | | 41| 0| { \ | | 42| 0| (b)->str = (b)->bulk; \ | | 43| 0| (b)->size = sizeof((b)->bulk); \ | | 44| 0| (b)->used = 0; \ | | 45| 0| (b)->bulk[0] = 0; \ | | 46| 0| } ------------------ 156| 0| __kmp_str_buf_print(&version, "%s", catgets(cat, section, number, NULL)); 157| | 158| | // String returned by catgets is invalid after closing catalog, so copy it. 159| 0| if (strcmp(version.str, expected) != 0) { ------------------ | Branch (159:9): [True: 0, False: 0] ------------------ 160| 0| __kmp_i18n_catclose(); // Close bad catalog. 161| 0| status = KMP_I18N_ABSENT; // And mark it as absent. 162| 0| if (__kmp_generate_warnings > kmp_warnings_low) { ------------------ | Branch (162:11): [True: 0, False: 0] ------------------ 163| | // AC: only issue warning in case explicitly asked to 164| | // And now print a warning using default messages. 165| 0| char const *name = "NLSPATH"; 166| 0| char const *nlspath = __kmp_env_get(name); 167| 0| __kmp_msg(kmp_ms_warning, 168| 0| KMP_MSG(WrongMessageCatalog, name, version.str, expected), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ 169| 0| KMP_HNT(CheckEnvVar, name, nlspath), __kmp_msg_null); ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 170| 0| KMP_INFORM(WillUseDefaultMessages); ------------------ | | 143| 0| __kmp_msg(kmp_ms_inform, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 171| 0| KMP_INTERNAL_FREE(CCAST(char *, nlspath)); ------------------ | | 111| 0|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 172| 0| } // __kmp_generate_warnings 173| 0| } 174| 0| __kmp_str_buf_free(&version); 175| 0| } 176| 0|} // func __kmp_i18n_do_catopen __kmp_validate_locks: 43| 1|void __kmp_validate_locks(void) { 44| 1| int i; 45| 1| kmp_uint32 x, y; 46| | 47| | /* Check to make sure unsigned arithmetic does wraps properly */ 48| 1| x = ~((kmp_uint32)0) - 2; 49| 1| y = x - 2; 50| | 51| 9| for (i = 0; i < 8; ++i, ++x, ++y) { ------------------ | Branch (51:15): [True: 8, False: 1] ------------------ 52| 8| kmp_uint32 z = (x - y); 53| 8| KMP_ASSERT(z == 2); ------------------ | | 50| 8| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 8] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 54| 8| } 55| | 56| 1| KMP_ASSERT(offsetof(kmp_base_queuing_lock, tail_id) % 8 == 0); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [Folded, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 57| 1|} __kmp_acquire_ticket_lock: 644| 64.3k|int __kmp_acquire_ticket_lock(kmp_ticket_lock_t *lck, kmp_int32 gtid) { 645| 64.3k| int retval = __kmp_acquire_ticket_lock_timed_template(lck, gtid); 646| 64.3k| return retval; 647| 64.3k|} __kmp_release_ticket_lock: 714| 64.3k|int __kmp_release_ticket_lock(kmp_ticket_lock_t *lck, kmp_int32 gtid) { 715| 64.3k| kmp_uint32 distance = std::atomic_load_explicit(&lck->lk.next_ticket, 716| 64.3k| std::memory_order_relaxed) - 717| 64.3k| std::atomic_load_explicit(&lck->lk.now_serving, 718| 64.3k| std::memory_order_relaxed); 719| | 720| 64.3k| std::atomic_fetch_add_explicit(&lck->lk.now_serving, 1U, 721| 64.3k| std::memory_order_release); 722| | 723| 64.3k| KMP_YIELD(distance > ------------------ | | 1529| 64.3k| { \ | | 1530| 64.3k| KMP_CPU_PAUSE(); \ | | ------------------ | | | | 1498| 64.3k|#define KMP_CPU_PAUSE() __kmp_x86_pause() | | ------------------ | | 1531| 128k| if ((cond) && (KMP_TRY_YIELD)) \ | | ------------------ | | | | 1523| 1| ((__kmp_use_yield == 1) || (__kmp_use_yield == 2 && (KMP_OVERSUBSCRIBED))) | | | | ------------------ | | | | | | 1520| 0| (TCR_4(__kmp_nth) > (__kmp_avail_proc ? __kmp_avail_proc : __kmp_xproc)) | | | | | | ------------------ | | | | | | | | 1127| 0|#define TCR_4(a) (a) | | | | | | ------------------ | | | | | | | Branch (1520:24): [True: 0, False: 0] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (1523:4): [True: 1, False: 0] | | | | | Branch (1523:31): [True: 0, False: 0] | | | | | Branch (1523:55): [True: 0, False: 0] | | | | ------------------ | | ------------------ | | | Branch (1531:9): [True: 1, False: 64.3k] | | | Branch (1531:10): [True: 58.9k, False: 5.37k] | | ------------------ | | 1532| 64.3k| __kmp_yield(); \ | | 1533| 64.3k| } ------------------ 724| 64.3k| (kmp_uint32)(__kmp_avail_proc ? __kmp_avail_proc : __kmp_xproc)); 725| 64.3k| return KMP_LOCK_RELEASED; ------------------ | | 156| 64.3k|#define KMP_LOCK_RELEASED 1 ------------------ 726| 64.3k|} __kmp_init_ticket_lock: 753| 10|void __kmp_init_ticket_lock(kmp_ticket_lock_t *lck) { 754| | lck->lk.location = NULL; 755| 10| lck->lk.self = lck; 756| 10| std::atomic_store_explicit(&lck->lk.next_ticket, 0U, 757| 10| std::memory_order_relaxed); 758| 10| std::atomic_store_explicit(&lck->lk.now_serving, 0U, 759| 10| std::memory_order_relaxed); 760| 10| std::atomic_store_explicit( 761| 10| &lck->lk.owner_id, 0, 762| 10| std::memory_order_relaxed); // no thread owns the lock. 763| 10| std::atomic_store_explicit( 764| 10| &lck->lk.depth_locked, -1, 765| 10| std::memory_order_relaxed); // -1 => not a nested lock. 766| 10| std::atomic_store_explicit(&lck->lk.initialized, true, 767| 10| std::memory_order_release); 768| 10|} __kmp_acquire_queuing_lock: 1277| 25.1k|int __kmp_acquire_queuing_lock(kmp_queuing_lock_t *lck, kmp_int32 gtid) { 1278| 25.1k| KMP_DEBUG_ASSERT(gtid >= 0); 1279| | 1280| 25.1k| int retval = __kmp_acquire_queuing_lock_timed_template(lck, gtid); 1281| 25.1k| return retval; 1282| 25.1k|} __kmp_release_queuing_lock: 1353| 25.2k|int __kmp_release_queuing_lock(kmp_queuing_lock_t *lck, kmp_int32 gtid) { 1354| 25.2k| volatile kmp_int32 *head_id_p = &lck->lk.head_id; 1355| 25.2k| volatile kmp_int32 *tail_id_p = &lck->lk.tail_id; 1356| | 1357| 25.2k| KA_TRACE(1000, 1358| 25.2k| ("__kmp_release_queuing_lock: lck:%p, T#%d entering\n", lck, gtid)); 1359| 25.2k| KMP_DEBUG_ASSERT(gtid >= 0); 1360| |#if KMP_DEBUG || DEBUG_QUEUING_LOCKS 1361| | kmp_info_t *this_thr = __kmp_thread_from_gtid(gtid); 1362| |#endif 1363| 25.2k| KMP_DEBUG_ASSERT(this_thr != NULL); 1364| |#ifdef DEBUG_QUEUING_LOCKS 1365| | TRACE_LOCK(gtid + 1, "rel ent"); 1366| | 1367| | if (this_thr->th.th_spin_here) 1368| | __kmp_dump_queuing_lock(this_thr, gtid, lck, *head_id_p, *tail_id_p); 1369| | if (this_thr->th.th_next_waiting != 0) 1370| | __kmp_dump_queuing_lock(this_thr, gtid, lck, *head_id_p, *tail_id_p); 1371| |#endif 1372| 25.2k| KMP_DEBUG_ASSERT(!this_thr->th.th_spin_here); 1373| 25.2k| KMP_DEBUG_ASSERT(this_thr->th.th_next_waiting == 0); 1374| | 1375| 25.2k| KMP_FSYNC_RELEASING(lck); ------------------ | | 335| 25.2k|#define KMP_FSYNC_RELEASING(obj) ((void)0) ------------------ 1376| | 1377| 25.2k| while (1) { ------------------ | Branch (1377:10): [True: 25.2k, Folded] ------------------ 1378| 25.2k| kmp_int32 dequeued; 1379| 25.2k| kmp_int32 head; 1380| 25.2k| kmp_int32 tail; 1381| | 1382| 25.2k| head = *head_id_p; 1383| | 1384| |#ifdef DEBUG_QUEUING_LOCKS 1385| | tail = *tail_id_p; 1386| | TRACE_LOCK_HT(gtid + 1, "rel read: ", head, tail); 1387| | if (head == 0) 1388| | __kmp_dump_queuing_lock(this_thr, gtid, lck, head, tail); 1389| |#endif 1390| 25.2k| KMP_DEBUG_ASSERT(head != 1391| 25.2k| 0); /* holding the lock, head must be -1 or queue head */ 1392| | 1393| 25.2k| if (head == -1) { /* nobody on queue */ ------------------ | Branch (1393:9): [True: 23.7k, False: 1.51k] ------------------ 1394| | /* try (-1,0)->(0,0) */ 1395| 23.7k| if (KMP_COMPARE_AND_STORE_REL32(head_id_p, -1, 0)) { ------------------ | | 810| 23.7k| __sync_bool_compare_and_swap((volatile kmp_uint32 *)(p), (kmp_uint32)(cv), \ | | ------------------ | | | Branch (810:3): [True: 23.6k, False: 13] | | ------------------ | | 811| 23.7k| (kmp_uint32)(sv)) ------------------ 1396| 23.6k| KA_TRACE( 1397| 23.6k| 1000, 1398| 23.6k| ("__kmp_release_queuing_lock: lck:%p, T#%d exiting: queue empty\n", 1399| 23.6k| lck, gtid)); 1400| |#ifdef DEBUG_QUEUING_LOCKS 1401| | TRACE_LOCK_HT(gtid + 1, "rel exit: ", 0, 0); 1402| |#endif 1403| | 1404| 23.6k|#if OMPT_SUPPORT 1405| |/* nothing to do - no other thread is trying to shift blame */ 1406| 23.6k|#endif 1407| 23.6k| return KMP_LOCK_RELEASED; ------------------ | | 156| 23.6k|#define KMP_LOCK_RELEASED 1 ------------------ 1408| 23.6k| } 1409| 13| dequeued = FALSE; ------------------ | | 1279| 13|#define FALSE 0 ------------------ 1410| 1.51k| } else { 1411| 1.51k| KMP_MB(); 1412| 1.51k| tail = *tail_id_p; 1413| 1.51k| if (head == tail) { /* only one thread on the queue */ ------------------ | Branch (1413:11): [True: 302, False: 1.21k] ------------------ 1414| |#ifdef DEBUG_QUEUING_LOCKS 1415| | if (head <= 0) 1416| | __kmp_dump_queuing_lock(this_thr, gtid, lck, head, tail); 1417| |#endif 1418| 302| KMP_DEBUG_ASSERT(head > 0); 1419| | 1420| | /* try (h,h)->(-1,0) */ 1421| 302| dequeued = KMP_COMPARE_AND_STORE_REL64( ------------------ | | 853| 302| __sync_bool_compare_and_swap((volatile kmp_uint64 *)(p), (kmp_uint64)(cv), \ | | 854| 302| (kmp_uint64)(sv)) ------------------ 1422| 302| RCAST(volatile kmp_int64 *, tail_id_p), KMP_PACK_64(head, head), 1423| 302| KMP_PACK_64(-1, 0)); 1424| |#ifdef DEBUG_QUEUING_LOCKS 1425| | TRACE_LOCK(gtid + 1, "rel deq: (h,h)->(-1,0)"); 1426| |#endif 1427| | 1428| 1.21k| } else { 1429| 1.21k| volatile kmp_int32 *waiting_id_p; 1430| 1.21k| kmp_info_t *head_thr = __kmp_thread_from_gtid(head - 1); 1431| 1.21k| KMP_DEBUG_ASSERT(head_thr != NULL); 1432| 1.21k| waiting_id_p = &head_thr->th.th_next_waiting; 1433| | 1434| |/* Does this require synchronous reads? */ 1435| |#ifdef DEBUG_QUEUING_LOCKS 1436| | if (head <= 0 || tail <= 0) 1437| | __kmp_dump_queuing_lock(this_thr, gtid, lck, head, tail); 1438| |#endif 1439| 1.21k| KMP_DEBUG_ASSERT(head > 0 && tail > 0); 1440| | 1441| | /* try (h,t)->(h',t) or (t,t) */ 1442| 1.21k| KMP_MB(); 1443| | /* make sure enqueuing thread has time to update next waiting thread 1444| | * field */ 1445| 1.21k| *head_id_p = 1446| 1.21k| KMP_WAIT((volatile kmp_uint32 *)waiting_id_p, 0, KMP_NEQ, NULL); ------------------ | | 1183| 1.21k|#define KMP_WAIT __kmp_wait_4 ------------------ KMP_WAIT((volatile kmp_uint32 *)waiting_id_p, 0, KMP_NEQ, NULL); ------------------ | | 1186| 1.21k|#define KMP_NEQ __kmp_neq_4 ------------------ 1447| |#ifdef DEBUG_QUEUING_LOCKS 1448| | TRACE_LOCK(gtid + 1, "rel deq: (h,t)->(h',t)"); 1449| |#endif 1450| 1.21k| dequeued = TRUE; ------------------ | | 1280| 1.21k|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1.21k|#define FALSE 0 | | ------------------ ------------------ 1451| 1.21k| } 1452| 1.51k| } 1453| | 1454| 1.52k| if (dequeued) { ------------------ | Branch (1454:9): [True: 1.51k, False: 13] ------------------ 1455| 1.51k| kmp_info_t *head_thr = __kmp_thread_from_gtid(head - 1); 1456| 1.51k| KMP_DEBUG_ASSERT(head_thr != NULL); 1457| | 1458| |/* Does this require synchronous reads? */ 1459| |#ifdef DEBUG_QUEUING_LOCKS 1460| | if (head <= 0 || tail <= 0) 1461| | __kmp_dump_queuing_lock(this_thr, gtid, lck, head, tail); 1462| |#endif 1463| 1.51k| KMP_DEBUG_ASSERT(head > 0 && tail > 0); 1464| | 1465| | /* For clean code only. Thread not released until next statement prevents 1466| | race with acquire code. */ 1467| 1.51k| head_thr->th.th_next_waiting = 0; 1468| |#ifdef DEBUG_QUEUING_LOCKS 1469| | TRACE_LOCK_T(gtid + 1, "rel nw=0 for t=", head); 1470| |#endif 1471| | 1472| 1.51k| KMP_MB(); 1473| | /* reset spin value */ 1474| 1.51k| head_thr->th.th_spin_here = FALSE; ------------------ | | 1279| 1.51k|#define FALSE 0 ------------------ 1475| | 1476| 1.51k| KA_TRACE(1000, ("__kmp_release_queuing_lock: lck:%p, T#%d exiting: after " 1477| 1.51k| "dequeuing\n", 1478| 1.51k| lck, gtid)); 1479| |#ifdef DEBUG_QUEUING_LOCKS 1480| | TRACE_LOCK(gtid + 1, "rel exit 2"); 1481| |#endif 1482| 1.51k| return KMP_LOCK_RELEASED; ------------------ | | 156| 1.51k|#define KMP_LOCK_RELEASED 1 ------------------ 1483| 1.51k| } 1484| | /* KMP_CPU_PAUSE(); don't want to make releasing thread hold up acquiring 1485| | threads */ 1486| | 1487| |#ifdef DEBUG_QUEUING_LOCKS 1488| | TRACE_LOCK(gtid + 1, "rel retry"); 1489| |#endif 1490| | 1491| 1.52k| } /* while */ 1492| 0| KMP_ASSERT2(0, "should not get here"); ------------------ | | 53| 0|#define KMP_ASSERT2(cond, msg) KMP_ASSERT(cond) | | ------------------ | | | | 50| 0| if (!(cond)) { \ | | | | ------------------ | | | | | Branch (50:7): [True: 0, Folded] | | | | ------------------ | | | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | | | 52| 0| } | | ------------------ ------------------ 1493| 0| return KMP_LOCK_RELEASED; ------------------ | | 156| 0|#define KMP_LOCK_RELEASED 1 ------------------ 1494| 25.2k|} __kmp_init_queuing_lock: 1516| 30.8k|void __kmp_init_queuing_lock(kmp_queuing_lock_t *lck) { 1517| 30.8k| lck->lk.location = NULL; 1518| 30.8k| lck->lk.head_id = 0; 1519| 30.8k| lck->lk.tail_id = 0; 1520| 30.8k| lck->lk.next_ticket = 0; 1521| 30.8k| lck->lk.now_serving = 0; 1522| 30.8k| lck->lk.owner_id = 0; // no thread owns the lock. 1523| 30.8k| lck->lk.depth_locked = -1; // >= 0 for nestable locks, -1 for simple locks. 1524| 30.8k| lck->lk.initialized = lck; 1525| | 1526| 30.8k| KA_TRACE(1000, ("__kmp_init_queuing_lock: lock %p initialized\n", lck)); 1527| 30.8k|} __kmp_destroy_queuing_lock: 1529| 30.8k|void __kmp_destroy_queuing_lock(kmp_queuing_lock_t *lck) { 1530| 30.8k| lck->lk.initialized = NULL; 1531| | lck->lk.location = NULL; 1532| 30.8k| lck->lk.head_id = 0; 1533| 30.8k| lck->lk.tail_id = 0; 1534| 30.8k| lck->lk.next_ticket = 0; 1535| 30.8k| lck->lk.now_serving = 0; 1536| 30.8k| lck->lk.owner_id = 0; 1537| 30.8k| lck->lk.depth_locked = -1; 1538| 30.8k|} __kmp_allocate_indirect_lock: 3122| 30.8k| kmp_indirect_locktag_t tag) { 3123| 30.8k| kmp_indirect_lock_t *lck; 3124| 30.8k| kmp_lock_index_t idx, table_idx; 3125| | 3126| 30.8k| __kmp_acquire_lock(&__kmp_global_lock, gtid); 3127| | 3128| 30.8k| if (__kmp_indirect_lock_pool[tag] != NULL) { ------------------ | Branch (3128:7): [True: 30.8k, False: 4] ------------------ 3129| | // Reuse the allocated and destroyed lock object 3130| 30.8k| lck = __kmp_indirect_lock_pool[tag]; 3131| 30.8k| if (OMP_LOCK_T_SIZE < sizeof(void *)) ------------------ | | 54| 30.8k|#define OMP_LOCK_T_SIZE sizeof(int) ------------------ | Branch (3131:9): [True: 30.8k, Folded] ------------------ 3132| 30.8k| idx = lck->lock->pool.index; 3133| 30.8k| __kmp_indirect_lock_pool[tag] = (kmp_indirect_lock_t *)lck->lock->pool.next; 3134| 30.8k| KA_TRACE(20, ("__kmp_allocate_indirect_lock: reusing an existing lock %p\n", 3135| 30.8k| lck)); 3136| 30.8k| } else { 3137| 4| kmp_uint32 row, col; 3138| 4| kmp_indirect_lock_table_t *lock_table = &__kmp_i_lock_table; 3139| 4| idx = 0; 3140| | // Find location in list of lock tables to put new lock 3141| 4| while (1) { ------------------ | Branch (3141:12): [True: 4, Folded] ------------------ 3142| 4| table_idx = lock_table->next; // index within this table 3143| 4| idx += lock_table->next; // global index within list of tables 3144| 4| if (table_idx < lock_table->nrow_ptrs * KMP_I_LOCK_CHUNK) { ------------------ | | 1222| 4|#define KMP_I_LOCK_CHUNK 1024 ------------------ | Branch (3144:11): [True: 4, False: 0] ------------------ 3145| 4| row = table_idx / KMP_I_LOCK_CHUNK; ------------------ | | 1222| 4|#define KMP_I_LOCK_CHUNK 1024 ------------------ 3146| 4| col = table_idx % KMP_I_LOCK_CHUNK; ------------------ | | 1222| 4|#define KMP_I_LOCK_CHUNK 1024 ------------------ 3147| | // Allocate a new row of locks if necessary 3148| 4| if (!lock_table->table[row]) { ------------------ | Branch (3148:13): [True: 0, False: 4] ------------------ 3149| 0| lock_table->table[row] = (kmp_indirect_lock_t *)__kmp_allocate( ------------------ | | 3617| 0|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3150| 0| sizeof(kmp_indirect_lock_t) * KMP_I_LOCK_CHUNK); 3151| 0| } 3152| 4| break; 3153| 4| } 3154| | // Allocate a new lock table if necessary with double the capacity 3155| 0| if (!lock_table->next_table) { ------------------ | Branch (3155:11): [True: 0, False: 0] ------------------ 3156| 0| kmp_indirect_lock_table_t *next_table = 3157| 0| (kmp_indirect_lock_table_t *)__kmp_allocate( ------------------ | | 3617| 0|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3158| 0| sizeof(kmp_indirect_lock_table_t)); 3159| 0| next_table->table = (kmp_indirect_lock_t **)__kmp_allocate( ------------------ | | 3617| 0|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3160| 0| sizeof(kmp_indirect_lock_t *) * 2 * lock_table->nrow_ptrs); 3161| 0| next_table->nrow_ptrs = 2 * lock_table->nrow_ptrs; 3162| 0| next_table->next = 0; 3163| 0| next_table->next_table = nullptr; 3164| 0| lock_table->next_table = next_table; 3165| 0| } 3166| 0| lock_table = lock_table->next_table; 3167| 0| KMP_ASSERT(lock_table); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 3168| 0| } 3169| 4| lock_table->next++; 3170| | 3171| 4| lck = &lock_table->table[row][col]; 3172| | // Allocate a new base lock object 3173| 4| lck->lock = (kmp_user_lock_p)__kmp_allocate(__kmp_indirect_lock_size[tag]); ------------------ | | 3617| 4|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3174| 4| KA_TRACE(20, 3175| 4| ("__kmp_allocate_indirect_lock: allocated a new lock %p\n", lck)); 3176| 4| } 3177| | 3178| 30.8k| __kmp_release_lock(&__kmp_global_lock, gtid); 3179| | 3180| 30.8k| lck->type = tag; 3181| | 3182| 30.8k| if (OMP_LOCK_T_SIZE < sizeof(void *)) { ------------------ | | 54| 30.8k|#define OMP_LOCK_T_SIZE sizeof(int) ------------------ | Branch (3182:7): [True: 30.8k, Folded] ------------------ 3183| 30.8k| *((kmp_lock_index_t *)user_lock) = idx 3184| 30.8k| << 1; // indirect lock word must be even 3185| 30.8k| } else { 3186| 0| *((kmp_indirect_lock_t **)user_lock) = lck; 3187| 0| } 3188| | 3189| 30.8k| return lck; 3190| 30.8k|} __kmp_init_dynamic_user_locks: 3327| 1|void __kmp_init_dynamic_user_locks() { 3328| | // Initialize jump table for the lock functions 3329| 1| if (__kmp_env_consistency_check) { ------------------ | Branch (3329:7): [True: 0, False: 1] ------------------ 3330| 0| __kmp_direct_set = direct_set_check; 3331| 0| __kmp_direct_unset = direct_unset_check; 3332| 0| __kmp_direct_test = direct_test_check; 3333| 0| __kmp_direct_destroy = direct_destroy_check; 3334| 0| __kmp_indirect_set = indirect_set_check; 3335| 0| __kmp_indirect_unset = indirect_unset_check; 3336| 0| __kmp_indirect_test = indirect_test_check; 3337| 0| __kmp_indirect_destroy = indirect_destroy_check; 3338| 1| } else { 3339| 1| __kmp_direct_set = direct_set; 3340| 1| __kmp_direct_unset = direct_unset; 3341| 1| __kmp_direct_test = direct_test; 3342| 1| __kmp_direct_destroy = direct_destroy; 3343| 1| __kmp_indirect_set = indirect_set; 3344| 1| __kmp_indirect_unset = indirect_unset; 3345| 1| __kmp_indirect_test = indirect_test; 3346| 1| __kmp_indirect_destroy = indirect_destroy; 3347| 1| } 3348| | // If the user locks have already been initialized, then return. Allow the 3349| | // switch between different KMP_CONSISTENCY_CHECK values, but do not allocate 3350| | // new lock tables if they have already been allocated. 3351| 1| if (__kmp_init_user_locks) ------------------ | Branch (3351:7): [True: 0, False: 1] ------------------ 3352| 0| return; 3353| | 3354| | // Initialize lock index table 3355| 1| __kmp_i_lock_table.nrow_ptrs = KMP_I_LOCK_TABLE_INIT_NROW_PTRS; ------------------ | | 1226| 1|#define KMP_I_LOCK_TABLE_INIT_NROW_PTRS 8 ------------------ 3356| 1| __kmp_i_lock_table.table = (kmp_indirect_lock_t **)__kmp_allocate( ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3357| 1| sizeof(kmp_indirect_lock_t *) * KMP_I_LOCK_TABLE_INIT_NROW_PTRS); 3358| 1| *(__kmp_i_lock_table.table) = (kmp_indirect_lock_t *)__kmp_allocate( ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3359| 1| KMP_I_LOCK_CHUNK * sizeof(kmp_indirect_lock_t)); 3360| 1| __kmp_i_lock_table.next = 0; 3361| 1| __kmp_i_lock_table.next_table = nullptr; 3362| | 3363| | // Indirect lock size 3364| 1| __kmp_indirect_lock_size[locktag_ticket] = sizeof(kmp_ticket_lock_t); 3365| 1| __kmp_indirect_lock_size[locktag_queuing] = sizeof(kmp_queuing_lock_t); 3366| 1|#if KMP_USE_ADAPTIVE_LOCKS 3367| 1| __kmp_indirect_lock_size[locktag_adaptive] = sizeof(kmp_adaptive_lock_t); 3368| 1|#endif 3369| 1| __kmp_indirect_lock_size[locktag_drdpa] = sizeof(kmp_drdpa_lock_t); 3370| 1|#if KMP_USE_TSX 3371| 1| __kmp_indirect_lock_size[locktag_rtm_queuing] = sizeof(kmp_queuing_lock_t); 3372| 1|#endif 3373| 1| __kmp_indirect_lock_size[locktag_nested_tas] = sizeof(kmp_tas_lock_t); 3374| 1|#if KMP_USE_FUTEX 3375| 1| __kmp_indirect_lock_size[locktag_nested_futex] = sizeof(kmp_futex_lock_t); 3376| 1|#endif 3377| 1| __kmp_indirect_lock_size[locktag_nested_ticket] = sizeof(kmp_ticket_lock_t); 3378| 1| __kmp_indirect_lock_size[locktag_nested_queuing] = sizeof(kmp_queuing_lock_t); 3379| 1| __kmp_indirect_lock_size[locktag_nested_drdpa] = sizeof(kmp_drdpa_lock_t); 3380| | 3381| |// Initialize lock accessor/modifier 3382| 1|#define fill_jumps(table, expand, sep) \ 3383| 1| { \ 3384| 1| table[locktag##sep##ticket] = expand(ticket); \ 3385| 1| table[locktag##sep##queuing] = expand(queuing); \ 3386| 1| table[locktag##sep##drdpa] = expand(drdpa); \ 3387| 1| } 3388| | 3389| 1|#if KMP_USE_ADAPTIVE_LOCKS 3390| 1|#define fill_table(table, expand) \ 3391| 1| { \ 3392| 1| fill_jumps(table, expand, _); \ 3393| 1| table[locktag_adaptive] = expand(queuing); \ 3394| 1| fill_jumps(table, expand, _nested_); \ 3395| 1| } 3396| |#else 3397| |#define fill_table(table, expand) \ 3398| | { \ 3399| | fill_jumps(table, expand, _); \ 3400| | fill_jumps(table, expand, _nested_); \ 3401| | } 3402| |#endif // KMP_USE_ADAPTIVE_LOCKS 3403| | 3404| 1|#define expand(l) \ 3405| 1| (void (*)(kmp_user_lock_p, const ident_t *)) __kmp_set_##l##_lock_location 3406| 1| fill_table(__kmp_indirect_set_location, expand); ------------------ | | 3391| 1| { \ | | 3392| 1| fill_jumps(table, expand, _); \ | | ------------------ | | | | 3383| 1| { \ | | | | 3384| 1| table[locktag##sep##ticket] = expand(ticket); \ | | | | ------------------ | | | | | | 3406| 1| fill_table(__kmp_indirect_set_location, expand); | | | | | | ------------------ | | | | | | | | 3405| 1| (void (*)(kmp_user_lock_p, const ident_t *)) __kmp_set_##l##_lock_location | | | | | | ------------------ | | | | ------------------ | | | | 3385| 1| table[locktag##sep##queuing] = expand(queuing); \ | | | | ------------------ | | | | | | 3406| 1| fill_table(__kmp_indirect_set_location, expand); | | | | | | ------------------ | | | | | | | | 3405| 1| (void (*)(kmp_user_lock_p, const ident_t *)) __kmp_set_##l##_lock_location | | | | | | ------------------ | | | | ------------------ | | | | 3386| 1| table[locktag##sep##drdpa] = expand(drdpa); \ | | | | ------------------ | | | | | | 3406| 1| fill_table(__kmp_indirect_set_location, expand); | | | | | | ------------------ | | | | | | | | 3405| 1| (void (*)(kmp_user_lock_p, const ident_t *)) __kmp_set_##l##_lock_location | | | | | | ------------------ | | | | ------------------ | | | | 3387| 1| } | | ------------------ | | 3393| 1| table[locktag_adaptive] = expand(queuing); \ | | ------------------ | | | | 3406| 1| fill_table(__kmp_indirect_set_location, expand); | | | | ------------------ | | | | | | 3405| 1| (void (*)(kmp_user_lock_p, const ident_t *)) __kmp_set_##l##_lock_location | | | | ------------------ | | ------------------ | | 3394| 1| fill_jumps(table, expand, _nested_); \ | | ------------------ | | | | 3383| 1| { \ | | | | 3384| 1| table[locktag##sep##ticket] = expand(ticket); \ | | | | ------------------ | | | | | | 3406| 1| fill_table(__kmp_indirect_set_location, expand); | | | | | | ------------------ | | | | | | | | 3405| 1| (void (*)(kmp_user_lock_p, const ident_t *)) __kmp_set_##l##_lock_location | | | | | | ------------------ | | | | ------------------ | | | | 3385| 1| table[locktag##sep##queuing] = expand(queuing); \ | | | | ------------------ | | | | | | 3406| 1| fill_table(__kmp_indirect_set_location, expand); | | | | | | ------------------ | | | | | | | | 3405| 1| (void (*)(kmp_user_lock_p, const ident_t *)) __kmp_set_##l##_lock_location | | | | | | ------------------ | | | | ------------------ | | | | 3386| 1| table[locktag##sep##drdpa] = expand(drdpa); \ | | | | ------------------ | | | | | | 3406| 1| fill_table(__kmp_indirect_set_location, expand); | | | | | | ------------------ | | | | | | | | 3405| 1| (void (*)(kmp_user_lock_p, const ident_t *)) __kmp_set_##l##_lock_location | | | | | | ------------------ | | | | ------------------ | | | | 3387| 1| } | | ------------------ | | 3395| 1| } ------------------ 3407| 1|#undef expand 3408| 1|#define expand(l) \ 3409| 1| (void (*)(kmp_user_lock_p, kmp_lock_flags_t)) __kmp_set_##l##_lock_flags 3410| 1| fill_table(__kmp_indirect_set_flags, expand); ------------------ | | 3391| 1| { \ | | 3392| 1| fill_jumps(table, expand, _); \ | | ------------------ | | | | 3383| 1| { \ | | | | 3384| 1| table[locktag##sep##ticket] = expand(ticket); \ | | | | ------------------ | | | | | | 3410| 1| fill_table(__kmp_indirect_set_flags, expand); | | | | | | ------------------ | | | | | | | | 3409| 1| (void (*)(kmp_user_lock_p, kmp_lock_flags_t)) __kmp_set_##l##_lock_flags | | | | | | ------------------ | | | | ------------------ | | | | 3385| 1| table[locktag##sep##queuing] = expand(queuing); \ | | | | ------------------ | | | | | | 3410| 1| fill_table(__kmp_indirect_set_flags, expand); | | | | | | ------------------ | | | | | | | | 3409| 1| (void (*)(kmp_user_lock_p, kmp_lock_flags_t)) __kmp_set_##l##_lock_flags | | | | | | ------------------ | | | | ------------------ | | | | 3386| 1| table[locktag##sep##drdpa] = expand(drdpa); \ | | | | ------------------ | | | | | | 3410| 1| fill_table(__kmp_indirect_set_flags, expand); | | | | | | ------------------ | | | | | | | | 3409| 1| (void (*)(kmp_user_lock_p, kmp_lock_flags_t)) __kmp_set_##l##_lock_flags | | | | | | ------------------ | | | | ------------------ | | | | 3387| 1| } | | ------------------ | | 3393| 1| table[locktag_adaptive] = expand(queuing); \ | | ------------------ | | | | 3410| 1| fill_table(__kmp_indirect_set_flags, expand); | | | | ------------------ | | | | | | 3409| 1| (void (*)(kmp_user_lock_p, kmp_lock_flags_t)) __kmp_set_##l##_lock_flags | | | | ------------------ | | ------------------ | | 3394| 1| fill_jumps(table, expand, _nested_); \ | | ------------------ | | | | 3383| 1| { \ | | | | 3384| 1| table[locktag##sep##ticket] = expand(ticket); \ | | | | ------------------ | | | | | | 3410| 1| fill_table(__kmp_indirect_set_flags, expand); | | | | | | ------------------ | | | | | | | | 3409| 1| (void (*)(kmp_user_lock_p, kmp_lock_flags_t)) __kmp_set_##l##_lock_flags | | | | | | ------------------ | | | | ------------------ | | | | 3385| 1| table[locktag##sep##queuing] = expand(queuing); \ | | | | ------------------ | | | | | | 3410| 1| fill_table(__kmp_indirect_set_flags, expand); | | | | | | ------------------ | | | | | | | | 3409| 1| (void (*)(kmp_user_lock_p, kmp_lock_flags_t)) __kmp_set_##l##_lock_flags | | | | | | ------------------ | | | | ------------------ | | | | 3386| 1| table[locktag##sep##drdpa] = expand(drdpa); \ | | | | ------------------ | | | | | | 3410| 1| fill_table(__kmp_indirect_set_flags, expand); | | | | | | ------------------ | | | | | | | | 3409| 1| (void (*)(kmp_user_lock_p, kmp_lock_flags_t)) __kmp_set_##l##_lock_flags | | | | | | ------------------ | | | | ------------------ | | | | 3387| 1| } | | ------------------ | | 3395| 1| } ------------------ 3411| 1|#undef expand 3412| 1|#define expand(l) \ 3413| 1| (const ident_t *(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_location 3414| 1| fill_table(__kmp_indirect_get_location, expand); ------------------ | | 3391| 1| { \ | | 3392| 1| fill_jumps(table, expand, _); \ | | ------------------ | | | | 3383| 1| { \ | | | | 3384| 1| table[locktag##sep##ticket] = expand(ticket); \ | | | | ------------------ | | | | | | 3414| 1| fill_table(__kmp_indirect_get_location, expand); | | | | | | ------------------ | | | | | | | | 3413| 1| (const ident_t *(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_location | | | | | | ------------------ | | | | ------------------ | | | | 3385| 1| table[locktag##sep##queuing] = expand(queuing); \ | | | | ------------------ | | | | | | 3414| 1| fill_table(__kmp_indirect_get_location, expand); | | | | | | ------------------ | | | | | | | | 3413| 1| (const ident_t *(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_location | | | | | | ------------------ | | | | ------------------ | | | | 3386| 1| table[locktag##sep##drdpa] = expand(drdpa); \ | | | | ------------------ | | | | | | 3414| 1| fill_table(__kmp_indirect_get_location, expand); | | | | | | ------------------ | | | | | | | | 3413| 1| (const ident_t *(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_location | | | | | | ------------------ | | | | ------------------ | | | | 3387| 1| } | | ------------------ | | 3393| 1| table[locktag_adaptive] = expand(queuing); \ | | ------------------ | | | | 3414| 1| fill_table(__kmp_indirect_get_location, expand); | | | | ------------------ | | | | | | 3413| 1| (const ident_t *(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_location | | | | ------------------ | | ------------------ | | 3394| 1| fill_jumps(table, expand, _nested_); \ | | ------------------ | | | | 3383| 1| { \ | | | | 3384| 1| table[locktag##sep##ticket] = expand(ticket); \ | | | | ------------------ | | | | | | 3414| 1| fill_table(__kmp_indirect_get_location, expand); | | | | | | ------------------ | | | | | | | | 3413| 1| (const ident_t *(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_location | | | | | | ------------------ | | | | ------------------ | | | | 3385| 1| table[locktag##sep##queuing] = expand(queuing); \ | | | | ------------------ | | | | | | 3414| 1| fill_table(__kmp_indirect_get_location, expand); | | | | | | ------------------ | | | | | | | | 3413| 1| (const ident_t *(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_location | | | | | | ------------------ | | | | ------------------ | | | | 3386| 1| table[locktag##sep##drdpa] = expand(drdpa); \ | | | | ------------------ | | | | | | 3414| 1| fill_table(__kmp_indirect_get_location, expand); | | | | | | ------------------ | | | | | | | | 3413| 1| (const ident_t *(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_location | | | | | | ------------------ | | | | ------------------ | | | | 3387| 1| } | | ------------------ | | 3395| 1| } ------------------ 3415| 1|#undef expand 3416| 1|#define expand(l) \ 3417| 1| (kmp_lock_flags_t(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_flags 3418| 1| fill_table(__kmp_indirect_get_flags, expand); ------------------ | | 3391| 1| { \ | | 3392| 1| fill_jumps(table, expand, _); \ | | ------------------ | | | | 3383| 1| { \ | | | | 3384| 1| table[locktag##sep##ticket] = expand(ticket); \ | | | | ------------------ | | | | | | 3418| 1| fill_table(__kmp_indirect_get_flags, expand); | | | | | | ------------------ | | | | | | | | 3417| 1| (kmp_lock_flags_t(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_flags | | | | | | ------------------ | | | | ------------------ | | | | 3385| 1| table[locktag##sep##queuing] = expand(queuing); \ | | | | ------------------ | | | | | | 3418| 1| fill_table(__kmp_indirect_get_flags, expand); | | | | | | ------------------ | | | | | | | | 3417| 1| (kmp_lock_flags_t(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_flags | | | | | | ------------------ | | | | ------------------ | | | | 3386| 1| table[locktag##sep##drdpa] = expand(drdpa); \ | | | | ------------------ | | | | | | 3418| 1| fill_table(__kmp_indirect_get_flags, expand); | | | | | | ------------------ | | | | | | | | 3417| 1| (kmp_lock_flags_t(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_flags | | | | | | ------------------ | | | | ------------------ | | | | 3387| 1| } | | ------------------ | | 3393| 1| table[locktag_adaptive] = expand(queuing); \ | | ------------------ | | | | 3418| 1| fill_table(__kmp_indirect_get_flags, expand); | | | | ------------------ | | | | | | 3417| 1| (kmp_lock_flags_t(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_flags | | | | ------------------ | | ------------------ | | 3394| 1| fill_jumps(table, expand, _nested_); \ | | ------------------ | | | | 3383| 1| { \ | | | | 3384| 1| table[locktag##sep##ticket] = expand(ticket); \ | | | | ------------------ | | | | | | 3418| 1| fill_table(__kmp_indirect_get_flags, expand); | | | | | | ------------------ | | | | | | | | 3417| 1| (kmp_lock_flags_t(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_flags | | | | | | ------------------ | | | | ------------------ | | | | 3385| 1| table[locktag##sep##queuing] = expand(queuing); \ | | | | ------------------ | | | | | | 3418| 1| fill_table(__kmp_indirect_get_flags, expand); | | | | | | ------------------ | | | | | | | | 3417| 1| (kmp_lock_flags_t(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_flags | | | | | | ------------------ | | | | ------------------ | | | | 3386| 1| table[locktag##sep##drdpa] = expand(drdpa); \ | | | | ------------------ | | | | | | 3418| 1| fill_table(__kmp_indirect_get_flags, expand); | | | | | | ------------------ | | | | | | | | 3417| 1| (kmp_lock_flags_t(*)(kmp_user_lock_p)) __kmp_get_##l##_lock_flags | | | | | | ------------------ | | | | ------------------ | | | | 3387| 1| } | | ------------------ | | 3395| 1| } ------------------ 3419| 1|#undef expand 3420| | 3421| 1| __kmp_init_user_locks = TRUE; ------------------ | | 1280| 1|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1|#define FALSE 0 | | ------------------ ------------------ 3422| 1|} __kmp_cleanup_indirect_user_locks: 3425| 1|void __kmp_cleanup_indirect_user_locks() { 3426| 1| int k; 3427| | 3428| | // Clean up locks in the pools first (they were already destroyed before going 3429| | // into the pools). 3430| 11| for (k = 0; k < KMP_NUM_I_LOCKS; ++k) { ------------------ | | 1081| 11| (locktag_nested_drdpa + 1) // number of indirect lock types ------------------ | Branch (3430:15): [True: 10, False: 1] ------------------ 3431| 10| kmp_indirect_lock_t *l = __kmp_indirect_lock_pool[k]; 3432| 14| while (l != NULL) { ------------------ | Branch (3432:12): [True: 4, False: 10] ------------------ 3433| 4| kmp_indirect_lock_t *ll = l; 3434| 4| l = (kmp_indirect_lock_t *)l->lock->pool.next; 3435| 4| KA_TRACE(20, ("__kmp_cleanup_indirect_user_locks: freeing %p from pool\n", 3436| 4| ll)); 3437| 4| __kmp_free(ll->lock); ------------------ | | 3619| 4|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 3438| 4| ll->lock = NULL; 3439| 4| } 3440| 10| __kmp_indirect_lock_pool[k] = NULL; 3441| 10| } 3442| | // Clean up the remaining undestroyed locks. 3443| 1| kmp_indirect_lock_table_t *ptr = &__kmp_i_lock_table; 3444| 2| while (ptr) { ------------------ | Branch (3444:10): [True: 1, False: 1] ------------------ 3445| 9| for (kmp_uint32 row = 0; row < ptr->nrow_ptrs; ++row) { ------------------ | Branch (3445:30): [True: 8, False: 1] ------------------ 3446| 8| if (!ptr->table[row]) ------------------ | Branch (3446:11): [True: 7, False: 1] ------------------ 3447| 7| continue; 3448| 1.02k| for (kmp_uint32 col = 0; col < KMP_I_LOCK_CHUNK; ++col) { ------------------ | | 1222| 1.02k|#define KMP_I_LOCK_CHUNK 1024 ------------------ | Branch (3448:32): [True: 1.02k, False: 1] ------------------ 3449| 1.02k| kmp_indirect_lock_t *l = &ptr->table[row][col]; 3450| 1.02k| if (l->lock) { ------------------ | Branch (3450:13): [True: 0, False: 1.02k] ------------------ 3451| | // Locks not destroyed explicitly need to be destroyed here. 3452| 0| KMP_I_LOCK_FUNC(l, destroy)(l->lock); ------------------ | | 1154| 0| __kmp_indirect_##op[((kmp_indirect_lock_t *)(l))->type] ------------------ 3453| 0| KA_TRACE(20, ("__kmp_cleanup_indirect_user_locks: destroy/freeing %p " 3454| 0| "from table\n", 3455| 0| l)); 3456| 0| __kmp_free(l->lock); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 3457| 0| } 3458| 1.02k| } 3459| 1| __kmp_free(ptr->table[row]); ------------------ | | 3619| 1|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 3460| 1| } 3461| 1| kmp_indirect_lock_table_t *next_table = ptr->next_table; 3462| 1| if (ptr != &__kmp_i_lock_table) ------------------ | Branch (3462:9): [True: 0, False: 1] ------------------ 3463| 0| __kmp_free(ptr); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 3464| 1| ptr = next_table; 3465| 1| } 3466| | 3467| 1| __kmp_init_user_locks = FALSE; ------------------ | | 1279| 1|#define FALSE 0 ------------------ 3468| 1|} kmp_lock.cpp:_ZL40__kmp_acquire_ticket_lock_timed_templateP15kmp_ticket_locki: 625| 64.3k| kmp_int32 gtid) { 626| 64.3k| kmp_uint32 my_ticket = std::atomic_fetch_add_explicit( 627| 64.3k| &lck->lk.next_ticket, 1U, std::memory_order_relaxed); 628| | 629| |#ifdef USE_LOCK_PROFILE 630| | if (std::atomic_load_explicit(&lck->lk.now_serving, 631| | std::memory_order_relaxed) != my_ticket) 632| | __kmp_printf("LOCK CONTENTION: %p\n", lck); 633| |/* else __kmp_printf( "." );*/ 634| |#endif /* USE_LOCK_PROFILE */ 635| | 636| 64.3k| if (std::atomic_load_explicit(&lck->lk.now_serving, ------------------ | Branch (636:7): [True: 64.3k, False: 0] ------------------ 637| 64.3k| std::memory_order_acquire) == my_ticket) { 638| 64.3k| return KMP_LOCK_ACQUIRED_FIRST; ------------------ | | 158| 64.3k|#define KMP_LOCK_ACQUIRED_FIRST 1 ------------------ 639| 64.3k| } 640| 0| KMP_WAIT_PTR(&lck->lk.now_serving, my_ticket, __kmp_bakery_check, lck); ------------------ | | 1184| 0|#define KMP_WAIT_PTR __kmp_wait_4_ptr ------------------ 641| 0| return KMP_LOCK_ACQUIRED_FIRST; ------------------ | | 158| 0|#define KMP_LOCK_ACQUIRED_FIRST 1 ------------------ 642| 64.3k|} kmp_lock.cpp:_ZL23__kmp_init_queuing_lockP13kmp_user_lock: 2953| 30.8k| static void __kmp_##op##_##lk##_##lock(kmp_user_lock_p lock) { \ 2954| 30.8k| __kmp_##op##_##lk##_##lock(&lock->lk); \ 2955| 30.8k| } kmp_lock.cpp:_ZL24__kmp_init_indirect_lockPj18kmp_dyna_lockseq_t: 3220| 30.8k| kmp_dyna_lockseq_t seq) { 3221| 30.8k|#if KMP_USE_ADAPTIVE_LOCKS 3222| 30.8k| if (seq == lockseq_adaptive && !__kmp_cpuinfo.flags.rtm) { ------------------ | Branch (3222:7): [True: 0, False: 30.8k] | Branch (3222:34): [True: 0, False: 0] ------------------ 3223| 0| KMP_WARNING(AdaptiveNotSupported, "kmp_lockseq_t", "adaptive"); ------------------ | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 3224| 0| seq = lockseq_queuing; 3225| 0| } 3226| 30.8k|#endif 3227| 30.8k|#if KMP_USE_TSX 3228| 30.8k| if (seq == lockseq_rtm_queuing && !__kmp_cpuinfo.flags.rtm) { ------------------ | Branch (3228:7): [True: 0, False: 30.8k] | Branch (3228:37): [True: 0, False: 0] ------------------ 3229| 0| seq = lockseq_queuing; 3230| 0| } 3231| 30.8k|#endif 3232| 30.8k| kmp_indirect_locktag_t tag = KMP_GET_I_TAG(seq); ------------------ | | 1107| 30.8k|#define KMP_GET_I_TAG(seq) (kmp_indirect_locktag_t)((seq)-KMP_FIRST_I_LOCK) | | ------------------ | | | | 1078| 30.8k|#define KMP_FIRST_I_LOCK lockseq_ticket | | ------------------ ------------------ 3233| 30.8k| kmp_indirect_lock_t *l = 3234| 30.8k| __kmp_allocate_indirect_lock((void **)lock, __kmp_entry_gtid(), tag); ------------------ | | 3467| 30.8k|#define __kmp_entry_gtid() __kmp_get_global_thread_id_reg() ------------------ 3235| 30.8k| KMP_I_LOCK_FUNC(l, init)(l->lock); ------------------ | | 1154| 30.8k| __kmp_indirect_##op[((kmp_indirect_lock_t *)(l))->type] ------------------ 3236| 30.8k| KA_TRACE( 3237| 30.8k| 20, ("__kmp_init_indirect_lock: initialized indirect lock with type#%d\n", 3238| 30.8k| seq)); 3239| 30.8k|} kmp_lock.cpp:_ZL26__kmp_lookup_indirect_lockPPvPKc: 3194| 30.8k|__kmp_lookup_indirect_lock(void **user_lock, const char *func) { 3195| 30.8k| if (__kmp_env_consistency_check) { ------------------ | Branch (3195:7): [True: 0, False: 30.8k] ------------------ 3196| 0| kmp_indirect_lock_t *lck = NULL; 3197| 0| if (user_lock == NULL) { ------------------ | Branch (3197:9): [True: 0, False: 0] ------------------ 3198| 0| KMP_FATAL(LockIsUninitialized, func); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 3199| 0| } 3200| 0| if (OMP_LOCK_T_SIZE < sizeof(void *)) { ------------------ | | 54| 0|#define OMP_LOCK_T_SIZE sizeof(int) ------------------ | Branch (3200:9): [True: 0, Folded] ------------------ 3201| 0| kmp_lock_index_t idx = KMP_EXTRACT_I_INDEX(user_lock); ------------------ | | 1145| 0|#define KMP_EXTRACT_I_INDEX(l) (*(kmp_lock_index_t *)(l) >> 1) ------------------ 3202| 0| lck = __kmp_get_i_lock(idx); 3203| 0| } else { 3204| 0| lck = *((kmp_indirect_lock_t **)user_lock); 3205| 0| } 3206| 0| if (lck == NULL) { ------------------ | Branch (3206:9): [True: 0, False: 0] ------------------ 3207| 0| KMP_FATAL(LockIsUninitialized, func); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 3208| 0| } 3209| 0| return lck; 3210| 30.8k| } else { 3211| 30.8k| if (OMP_LOCK_T_SIZE < sizeof(void *)) { ------------------ | | 54| 30.8k|#define OMP_LOCK_T_SIZE sizeof(int) ------------------ | Branch (3211:9): [True: 30.8k, Folded] ------------------ 3212| 30.8k| return __kmp_get_i_lock(KMP_EXTRACT_I_INDEX(user_lock)); ------------------ | | 1145| 30.8k|#define KMP_EXTRACT_I_INDEX(l) (*(kmp_lock_index_t *)(l) >> 1) ------------------ 3213| 30.8k| } else { 3214| 0| return *((kmp_indirect_lock_t **)user_lock); 3215| 0| } 3216| 30.8k| } 3217| 30.8k|} kmp_lock.cpp:_ZL27__kmp_destroy_indirect_lockPj: 3241| 30.8k|static void __kmp_destroy_indirect_lock(kmp_dyna_lock_t *lock) { 3242| 30.8k| kmp_uint32 gtid = __kmp_entry_gtid(); ------------------ | | 3467| 30.8k|#define __kmp_entry_gtid() __kmp_get_global_thread_id_reg() ------------------ 3243| 30.8k| kmp_indirect_lock_t *l = 3244| 30.8k| __kmp_lookup_indirect_lock((void **)lock, "omp_destroy_lock"); 3245| 30.8k| KMP_I_LOCK_FUNC(l, destroy)(l->lock); ------------------ | | 1154| 30.8k| __kmp_indirect_##op[((kmp_indirect_lock_t *)(l))->type] ------------------ 3246| 30.8k| kmp_indirect_locktag_t tag = l->type; 3247| | 3248| 30.8k| __kmp_acquire_lock(&__kmp_global_lock, gtid); 3249| | 3250| | // Use the base lock's space to keep the pool chain. 3251| 30.8k| l->lock->pool.next = (kmp_user_lock_p)__kmp_indirect_lock_pool[tag]; 3252| 30.8k| if (OMP_LOCK_T_SIZE < sizeof(void *)) { ------------------ | | 54| 30.8k|#define OMP_LOCK_T_SIZE sizeof(int) ------------------ | Branch (3252:7): [True: 30.8k, Folded] ------------------ 3253| 30.8k| l->lock->pool.index = KMP_EXTRACT_I_INDEX(lock); ------------------ | | 1145| 30.8k|#define KMP_EXTRACT_I_INDEX(l) (*(kmp_lock_index_t *)(l) >> 1) ------------------ 3254| 30.8k| } 3255| 30.8k| __kmp_indirect_lock_pool[tag] = l; 3256| | 3257| 30.8k| __kmp_release_lock(&__kmp_global_lock, gtid); 3258| 30.8k|} kmp_lock.cpp:_ZL23__kmp_set_indirect_lockPji: 3260| 25.1k|static int __kmp_set_indirect_lock(kmp_dyna_lock_t *lock, kmp_int32 gtid) { 3261| 25.1k| kmp_indirect_lock_t *l = KMP_LOOKUP_I_LOCK(lock); ------------------ | | 1264| 25.1k| ((OMP_LOCK_T_SIZE < sizeof(void *)) \ | | ------------------ | | | | 54| 25.1k|#define OMP_LOCK_T_SIZE sizeof(int) | | ------------------ | | | Branch (1264:4): [True: 25.1k, Folded] | | ------------------ | | 1265| 25.1k| ? __kmp_get_i_lock(KMP_EXTRACT_I_INDEX(l)) \ | | ------------------ | | | | 1145| 25.1k|#define KMP_EXTRACT_I_INDEX(l) (*(kmp_lock_index_t *)(l) >> 1) | | ------------------ | | 1266| 25.1k| : *((kmp_indirect_lock_t **)(l))) ------------------ 3262| 25.1k| return KMP_I_LOCK_FUNC(l, set)(l->lock, gtid); ------------------ | | 1154| 25.1k| __kmp_indirect_##op[((kmp_indirect_lock_t *)(l))->type] ------------------ 3263| 25.1k|} kmp_lock.cpp:_ZL25__kmp_unset_indirect_lockPji: 3265| 25.2k|static int __kmp_unset_indirect_lock(kmp_dyna_lock_t *lock, kmp_int32 gtid) { 3266| 25.2k| kmp_indirect_lock_t *l = KMP_LOOKUP_I_LOCK(lock); ------------------ | | 1264| 25.2k| ((OMP_LOCK_T_SIZE < sizeof(void *)) \ | | ------------------ | | | | 54| 25.2k|#define OMP_LOCK_T_SIZE sizeof(int) | | ------------------ | | | Branch (1264:4): [True: 25.2k, Folded] | | ------------------ | | 1265| 25.2k| ? __kmp_get_i_lock(KMP_EXTRACT_I_INDEX(l)) \ | | ------------------ | | | | 1145| 25.2k|#define KMP_EXTRACT_I_INDEX(l) (*(kmp_lock_index_t *)(l) >> 1) | | ------------------ | | 1266| 25.2k| : *((kmp_indirect_lock_t **)(l))) ------------------ 3267| 25.2k| return KMP_I_LOCK_FUNC(l, unset)(l->lock, gtid); ------------------ | | 1154| 25.2k| __kmp_indirect_##op[((kmp_indirect_lock_t *)(l))->type] ------------------ 3268| 25.2k|} kmp_lock.cpp:_ZL26__kmp_acquire_queuing_lockP13kmp_user_locki: 2958| 25.1k| kmp_int32 gtid) { \ 2959| 25.1k| return __kmp_##op##_##lk##_##lock(&lock->lk, gtid); \ 2960| 25.1k| } kmp_lock.cpp:_ZL26__kmp_release_queuing_lockP13kmp_user_locki: 2958| 25.2k| kmp_int32 gtid) { \ 2959| 25.2k| return __kmp_##op##_##lk##_##lock(&lock->lk, gtid); \ 2960| 25.2k| } kmp_lock.cpp:_ZL26__kmp_destroy_queuing_lockP13kmp_user_lock: 2953| 30.8k| static void __kmp_##op##_##lk##_##lock(kmp_user_lock_p lock) { \ 2954| 30.8k| __kmp_##op##_##lk##_##lock(&lock->lk); \ 2955| 30.8k| } kmp_lock.cpp:_ZL41__kmp_acquire_queuing_lock_timed_templateILb0EEiP16kmp_queuing_locki: 1085| 25.1k| kmp_int32 gtid) { 1086| 25.1k| kmp_info_t *this_thr = __kmp_thread_from_gtid(gtid); 1087| 25.1k| volatile kmp_int32 *head_id_p = &lck->lk.head_id; 1088| 25.1k| volatile kmp_int32 *tail_id_p = &lck->lk.tail_id; 1089| 25.1k| volatile kmp_uint32 *spin_here_p; 1090| | 1091| 25.1k|#if OMPT_SUPPORT 1092| 25.1k| ompt_state_t prev_state = ompt_state_undefined; 1093| 25.1k|#endif 1094| | 1095| 25.1k| KA_TRACE(1000, 1096| 25.1k| ("__kmp_acquire_queuing_lock: lck:%p, T#%d entering\n", lck, gtid)); 1097| | 1098| 25.1k| KMP_FSYNC_PREPARE(lck); ------------------ | | 332| 25.1k|#define KMP_FSYNC_PREPARE(obj) ((void)0) ------------------ 1099| 25.1k| KMP_DEBUG_ASSERT(this_thr != NULL); 1100| 25.1k| spin_here_p = &this_thr->th.th_spin_here; 1101| | 1102| |#ifdef DEBUG_QUEUING_LOCKS 1103| | TRACE_LOCK(gtid + 1, "acq ent"); 1104| | if (*spin_here_p) 1105| | __kmp_dump_queuing_lock(this_thr, gtid, lck, *head_id_p, *tail_id_p); 1106| | if (this_thr->th.th_next_waiting != 0) 1107| | __kmp_dump_queuing_lock(this_thr, gtid, lck, *head_id_p, *tail_id_p); 1108| |#endif 1109| 25.1k| KMP_DEBUG_ASSERT(!*spin_here_p); 1110| 25.1k| KMP_DEBUG_ASSERT(this_thr->th.th_next_waiting == 0); 1111| | 1112| | /* The following st.rel to spin_here_p needs to precede the cmpxchg.acq to 1113| | head_id_p that may follow, not just in execution order, but also in 1114| | visibility order. This way, when a releasing thread observes the changes to 1115| | the queue by this thread, it can rightly assume that spin_here_p has 1116| | already been set to TRUE, so that when it sets spin_here_p to FALSE, it is 1117| | not premature. If the releasing thread sets spin_here_p to FALSE before 1118| | this thread sets it to TRUE, this thread will hang. */ 1119| 25.1k| *spin_here_p = TRUE; /* before enqueuing to prevent race */ ------------------ | | 1280| 25.1k|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 25.1k|#define FALSE 0 | | ------------------ ------------------ 1120| | 1121| 25.2k| while (1) { ------------------ | Branch (1121:10): [True: 25.2k, Folded] ------------------ 1122| 25.2k| kmp_int32 enqueued; 1123| 25.2k| kmp_int32 head; 1124| 25.2k| kmp_int32 tail; 1125| | 1126| 25.2k| head = *head_id_p; 1127| | 1128| 25.2k| switch (head) { 1129| | 1130| 318| case -1: { ------------------ | Branch (1130:5): [True: 318, False: 24.9k] ------------------ 1131| |#ifdef DEBUG_QUEUING_LOCKS 1132| | tail = *tail_id_p; 1133| | TRACE_LOCK_HT(gtid + 1, "acq read: ", head, tail); 1134| |#endif 1135| 318| tail = 0; /* to make sure next link asynchronously read is not set 1136| | accidentally; this assignment prevents us from entering the 1137| | if ( t > 0 ) condition in the enqueued case below, which is not 1138| | necessary for this state transition */ 1139| | 1140| | /* try (-1,0)->(tid,tid) */ 1141| 318| enqueued = KMP_COMPARE_AND_STORE_ACQ64((volatile kmp_int64 *)tail_id_p, ------------------ | | 850| 318| __sync_bool_compare_and_swap((volatile kmp_uint64 *)(p), (kmp_uint64)(cv), \ | | 851| 318| (kmp_uint64)(sv)) ------------------ 1142| 318| KMP_PACK_64(-1, 0), 1143| 318| KMP_PACK_64(gtid + 1, gtid + 1)); 1144| |#ifdef DEBUG_QUEUING_LOCKS 1145| | if (enqueued) 1146| | TRACE_LOCK(gtid + 1, "acq enq: (-1,0)->(tid,tid)"); 1147| |#endif 1148| 318| } break; 1149| | 1150| 1.23k| default: { ------------------ | Branch (1150:5): [True: 1.23k, False: 24.0k] ------------------ 1151| 1.23k| tail = *tail_id_p; 1152| 1.23k| KMP_DEBUG_ASSERT(tail != gtid + 1); 1153| | 1154| |#ifdef DEBUG_QUEUING_LOCKS 1155| | TRACE_LOCK_HT(gtid + 1, "acq read: ", head, tail); 1156| |#endif 1157| | 1158| 1.23k| if (tail == 0) { ------------------ | Branch (1158:11): [True: 0, False: 1.23k] ------------------ 1159| 0| enqueued = FALSE; ------------------ | | 1279| 0|#define FALSE 0 ------------------ 1160| 1.23k| } else { 1161| | /* try (h,t) or (h,h)->(h,tid) */ 1162| 1.23k| enqueued = KMP_COMPARE_AND_STORE_ACQ32(tail_id_p, tail, gtid + 1); ------------------ | | 807| 1.23k| __sync_bool_compare_and_swap((volatile kmp_uint32 *)(p), (kmp_uint32)(cv), \ | | 808| 1.23k| (kmp_uint32)(sv)) ------------------ 1163| | 1164| |#ifdef DEBUG_QUEUING_LOCKS 1165| | if (enqueued) 1166| | TRACE_LOCK(gtid + 1, "acq enq: (h,t)->(h,tid)"); 1167| |#endif 1168| 1.23k| } 1169| 1.23k| } break; 1170| | 1171| 23.7k| case 0: /* empty queue */ ------------------ | Branch (1171:5): [True: 23.7k, False: 1.53k] ------------------ 1172| 23.7k| { 1173| 23.7k| kmp_int32 grabbed_lock; 1174| | 1175| |#ifdef DEBUG_QUEUING_LOCKS 1176| | tail = *tail_id_p; 1177| | TRACE_LOCK_HT(gtid + 1, "acq read: ", head, tail); 1178| |#endif 1179| | /* try (0,0)->(-1,0) */ 1180| | 1181| | /* only legal transition out of head = 0 is head = -1 with no change to 1182| | * tail */ 1183| 23.7k| grabbed_lock = KMP_COMPARE_AND_STORE_ACQ32(head_id_p, 0, -1); ------------------ | | 807| 23.7k| __sync_bool_compare_and_swap((volatile kmp_uint32 *)(p), (kmp_uint32)(cv), \ | | 808| 23.7k| (kmp_uint32)(sv)) ------------------ 1184| | 1185| 23.7k| if (grabbed_lock) { ------------------ | Branch (1185:11): [True: 23.6k, False: 29] ------------------ 1186| | 1187| 23.6k| *spin_here_p = FALSE; ------------------ | | 1279| 23.6k|#define FALSE 0 ------------------ 1188| | 1189| 23.6k| KA_TRACE( 1190| 23.6k| 1000, 1191| 23.6k| ("__kmp_acquire_queuing_lock: lck:%p, T#%d exiting: no queuing\n", 1192| 23.6k| lck, gtid)); 1193| |#ifdef DEBUG_QUEUING_LOCKS 1194| | TRACE_LOCK_HT(gtid + 1, "acq exit: ", head, 0); 1195| |#endif 1196| | 1197| 23.6k|#if OMPT_SUPPORT 1198| 23.6k| if (ompt_enabled.enabled && prev_state != ompt_state_undefined) { ------------------ | Branch (1198:13): [True: 0, False: 23.6k] | Branch (1198:37): [True: 0, False: 0] ------------------ 1199| | /* change the state before clearing wait_id */ 1200| 0| this_thr->th.ompt_thread_info.state = prev_state; 1201| 0| this_thr->th.ompt_thread_info.wait_id = 0; 1202| 0| } 1203| 23.6k|#endif 1204| | 1205| 23.6k| KMP_FSYNC_ACQUIRED(lck); ------------------ | | 334| 23.6k|#define KMP_FSYNC_ACQUIRED(obj) ((void)0) ------------------ 1206| 23.6k| return KMP_LOCK_ACQUIRED_FIRST; /* lock holder cannot be on queue */ ------------------ | | 158| 23.6k|#define KMP_LOCK_ACQUIRED_FIRST 1 ------------------ 1207| 23.6k| } 1208| 29| enqueued = FALSE; ------------------ | | 1279| 29|#define FALSE 0 ------------------ 1209| 29| } break; 1210| 25.2k| } 1211| | 1212| 1.62k|#if OMPT_SUPPORT 1213| 1.62k| if (ompt_enabled.enabled && prev_state == ompt_state_undefined) { ------------------ | Branch (1213:9): [True: 0, False: 1.62k] | Branch (1213:33): [True: 0, False: 0] ------------------ 1214| | /* this thread will spin; set wait_id before entering wait state */ 1215| 0| prev_state = this_thr->th.ompt_thread_info.state; 1216| 0| this_thr->th.ompt_thread_info.wait_id = (uint64_t)lck; 1217| 0| this_thr->th.ompt_thread_info.state = ompt_state_wait_lock; 1218| 0| } 1219| 1.62k|#endif 1220| | 1221| 1.62k| if (enqueued) { ------------------ | Branch (1221:9): [True: 1.51k, False: 113] ------------------ 1222| 1.51k| if (tail > 0) { ------------------ | Branch (1222:11): [True: 1.20k, False: 304] ------------------ 1223| 1.20k| kmp_info_t *tail_thr = __kmp_thread_from_gtid(tail - 1); 1224| 1.20k| KMP_ASSERT(tail_thr != NULL); ------------------ | | 50| 1.20k| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1.20k] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 1225| 1.20k| tail_thr->th.th_next_waiting = gtid + 1; 1226| | /* corresponding wait for this write in release code */ 1227| 1.20k| } 1228| 1.51k| KA_TRACE(1000, 1229| 1.51k| ("__kmp_acquire_queuing_lock: lck:%p, T#%d waiting for lock\n", 1230| 1.51k| lck, gtid)); 1231| | 1232| 1.51k| KMP_MB(); 1233| | // ToDo: Use __kmp_wait_sleep or similar when blocktime != inf 1234| 1.51k| KMP_WAIT(spin_here_p, FALSE, KMP_EQ, lck); ------------------ | | 1183| 1.51k|#define KMP_WAIT __kmp_wait_4 ------------------ KMP_WAIT(spin_here_p, FALSE, KMP_EQ, lck); ------------------ | | 1279| 1.51k|#define FALSE 0 ------------------ KMP_WAIT(spin_here_p, FALSE, KMP_EQ, lck); ------------------ | | 1185| 1.51k|#define KMP_EQ __kmp_eq_4 ------------------ 1235| | // Synchronize writes to both runtime thread structures 1236| | // and writes in user code. 1237| 1.51k| KMP_MB(); 1238| | 1239| |#ifdef DEBUG_QUEUING_LOCKS 1240| | TRACE_LOCK(gtid + 1, "acq spin"); 1241| | 1242| | if (this_thr->th.th_next_waiting != 0) 1243| | __kmp_dump_queuing_lock(this_thr, gtid, lck, *head_id_p, *tail_id_p); 1244| |#endif 1245| 1.51k| KMP_DEBUG_ASSERT(this_thr->th.th_next_waiting == 0); 1246| 1.51k| KA_TRACE(1000, ("__kmp_acquire_queuing_lock: lck:%p, T#%d exiting: after " 1247| 1.51k| "waiting on queue\n", 1248| 1.51k| lck, gtid)); 1249| | 1250| |#ifdef DEBUG_QUEUING_LOCKS 1251| | TRACE_LOCK(gtid + 1, "acq exit 2"); 1252| |#endif 1253| | 1254| 1.51k|#if OMPT_SUPPORT 1255| | /* change the state before clearing wait_id */ 1256| 1.51k| this_thr->th.ompt_thread_info.state = prev_state; 1257| 1.51k| this_thr->th.ompt_thread_info.wait_id = 0; 1258| 1.51k|#endif 1259| | 1260| | /* got lock, we were dequeued by the thread that released lock */ 1261| 1.51k| return KMP_LOCK_ACQUIRED_FIRST; ------------------ | | 158| 1.51k|#define KMP_LOCK_ACQUIRED_FIRST 1 ------------------ 1262| 1.51k| } 1263| | 1264| | /* Yield if number of threads > number of logical processors */ 1265| | /* ToDo: Not sure why this should only be in oversubscription case, 1266| | maybe should be traditional YIELD_INIT/YIELD_WHEN loop */ 1267| 113| KMP_YIELD_OVERSUB(); ------------------ | | 1536| 113| { \ | | 1537| 113| KMP_CPU_PAUSE(); \ | | ------------------ | | | | 1498| 113|#define KMP_CPU_PAUSE() __kmp_x86_pause() | | ------------------ | | 1538| 113| if ((KMP_TRY_YIELD_OVERSUB)) \ | | ------------------ | | | | 1526| 113| ((__kmp_use_yield == 1 || __kmp_use_yield == 2) && (KMP_OVERSUBSCRIBED)) | | | | ------------------ | | | | | | 1520| 18.4E| (TCR_4(__kmp_nth) > (__kmp_avail_proc ? __kmp_avail_proc : __kmp_xproc)) | | | | | | ------------------ | | | | | | | | 1127| 109|#define TCR_4(a) (a) | | | | | | ------------------ | | | | | | | Branch (1520:24): [True: 111, False: 18.4E] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (1526:5): [True: 113, False: 0] | | | | | Branch (1526:29): [True: 0, False: 0] | | | | | Branch (1526:54): [True: 0, False: 109] | | | | ------------------ | | ------------------ | | 1539| 113| __kmp_yield(); \ | | 1540| 113| } ------------------ 1268| | 1269| |#ifdef DEBUG_QUEUING_LOCKS 1270| | TRACE_LOCK(gtid + 1, "acq retry"); 1271| |#endif 1272| 113| } 1273| 8| KMP_ASSERT2(0, "should not get here"); ------------------ | | 53| 8|#define KMP_ASSERT2(cond, msg) KMP_ASSERT(cond) | | ------------------ | | | | 50| 8| if (!(cond)) { \ | | | | ------------------ | | | | | Branch (50:7): [True: 0, Folded] | | | | ------------------ | | | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | | | 52| 0| } | | ------------------ ------------------ 1274| 8| return KMP_LOCK_ACQUIRED_FIRST; ------------------ | | 158| 8|#define KMP_LOCK_ACQUIRED_FIRST 1 ------------------ 1275| 25.1k|} kmp_i18n.cpp:_ZL28__kmp_acquire_bootstrap_lockP15kmp_ticket_lock: 519| 1|static inline int __kmp_acquire_bootstrap_lock(kmp_bootstrap_lock_t *lck) { 520| 1| return __kmp_acquire_ticket_lock(lck, KMP_GTID_DNE); ------------------ | | 34| 1|#define KMP_GTID_DNE (-2) ------------------ 521| 1|} kmp_i18n.cpp:_ZL28__kmp_release_bootstrap_lockP15kmp_ticket_lock: 527| 1|static inline void __kmp_release_bootstrap_lock(kmp_bootstrap_lock_t *lck) { 528| 1| __kmp_release_ticket_lock(lck, KMP_GTID_DNE); ------------------ | | 34| 1|#define KMP_GTID_DNE (-2) ------------------ 529| 1|} kmp_runtime.cpp:_ZL28__kmp_acquire_bootstrap_lockP15kmp_ticket_lock: 519| 2.71k|static inline int __kmp_acquire_bootstrap_lock(kmp_bootstrap_lock_t *lck) { 520| 2.71k| return __kmp_acquire_ticket_lock(lck, KMP_GTID_DNE); ------------------ | | 34| 2.71k|#define KMP_GTID_DNE (-2) ------------------ 521| 2.71k|} kmp_runtime.cpp:_ZL28__kmp_release_bootstrap_lockP15kmp_ticket_lock: 527| 2.71k|static inline void __kmp_release_bootstrap_lock(kmp_bootstrap_lock_t *lck) { 528| 2.71k| __kmp_release_ticket_lock(lck, KMP_GTID_DNE); ------------------ | | 34| 2.71k|#define KMP_GTID_DNE (-2) ------------------ 529| 2.71k|} kmp_runtime.cpp:_ZL15__kmp_init_lockP15kmp_ticket_lock: 563| 3|static inline void __kmp_init_lock(kmp_lock_t *lck) { 564| 3| __kmp_init_ticket_lock(lck); 565| 3|} kmp_runtime.cpp:_ZL25__kmp_init_bootstrap_lockP15kmp_ticket_lock: 531| 3|static inline void __kmp_init_bootstrap_lock(kmp_bootstrap_lock_t *lck) { 532| 3| __kmp_init_ticket_lock(lck); 533| 3|} kmp_tasking.cpp:_ZL25__kmp_init_bootstrap_lockP15kmp_ticket_lock: 531| 4|static inline void __kmp_init_bootstrap_lock(kmp_bootstrap_lock_t *lck) { 532| 4| __kmp_init_ticket_lock(lck); 533| 4|} kmp_tasking.cpp:_ZL28__kmp_acquire_bootstrap_lockP15kmp_ticket_lock: 519| 3|static inline int __kmp_acquire_bootstrap_lock(kmp_bootstrap_lock_t *lck) { 520| 3| return __kmp_acquire_ticket_lock(lck, KMP_GTID_DNE); ------------------ | | 34| 3|#define KMP_GTID_DNE (-2) ------------------ 521| 3|} kmp_tasking.cpp:_ZL28__kmp_release_bootstrap_lockP15kmp_ticket_lock: 527| 3|static inline void __kmp_release_bootstrap_lock(kmp_bootstrap_lock_t *lck) { 528| 3| __kmp_release_ticket_lock(lck, KMP_GTID_DNE); ------------------ | | 34| 3|#define KMP_GTID_DNE (-2) ------------------ 529| 3|} kmp_lock.cpp:_ZL18__kmp_acquire_lockP15kmp_ticket_locki: 551| 61.6k|static inline int __kmp_acquire_lock(kmp_lock_t *lck, kmp_int32 gtid) { 552| 61.6k| return __kmp_acquire_ticket_lock(lck, gtid); 553| 61.6k|} kmp_lock.cpp:_ZL18__kmp_release_lockP15kmp_ticket_locki: 559| 61.6k|static inline void __kmp_release_lock(kmp_lock_t *lck, kmp_int32 gtid) { 560| 61.6k| __kmp_release_ticket_lock(lck, gtid); 561| 61.6k|} kmp_lock.cpp:_ZL16__kmp_get_i_lockj: 1240| 81.1k|static inline kmp_indirect_lock_t *__kmp_get_i_lock(kmp_lock_index_t idx) { 1241| 81.1k| kmp_indirect_lock_table_t *lock_table = &__kmp_i_lock_table; 1242| 81.1k| while (lock_table) { ------------------ | Branch (1242:10): [True: 81.1k, False: 18.4E] ------------------ 1243| 81.1k| kmp_lock_index_t max_locks = lock_table->nrow_ptrs * KMP_I_LOCK_CHUNK; ------------------ | | 1222| 81.1k|#define KMP_I_LOCK_CHUNK 1024 ------------------ 1244| 81.1k| if (idx < max_locks) { ------------------ | Branch (1244:9): [True: 81.1k, False: 18.4E] ------------------ 1245| 81.1k| kmp_lock_index_t row = idx / KMP_I_LOCK_CHUNK; ------------------ | | 1222| 81.1k|#define KMP_I_LOCK_CHUNK 1024 ------------------ 1246| 81.1k| kmp_lock_index_t col = idx % KMP_I_LOCK_CHUNK; ------------------ | | 1222| 81.1k|#define KMP_I_LOCK_CHUNK 1024 ------------------ 1247| 81.1k| if (!lock_table->table[row] || idx >= lock_table->next) ------------------ | Branch (1247:11): [True: 0, False: 81.1k] | Branch (1247:38): [True: 18.4E, False: 81.1k] ------------------ 1248| 0| break; 1249| 81.1k| return &lock_table->table[row][col]; 1250| 81.1k| } 1251| 18.4E| idx -= max_locks; 1252| 18.4E| lock_table = lock_table->next_table; 1253| 18.4E| } 1254| 18.4E| return nullptr; 1255| 81.1k|} _Z26__kmp_atomic_compare_storeIiEbPNSt3__16atomicIT_EES2_S2_: 1266| 32|bool __kmp_atomic_compare_store(std::atomic *p, T expected, T desired) { 1267| 32| return p->compare_exchange_strong( 1268| 32| expected, desired, std::memory_order_acq_rel, std::memory_order_relaxed); 1269| 32|} __kmp_get_global_thread_id: 117| 2|int __kmp_get_global_thread_id() { 118| 2| int i; 119| 2| kmp_info_t **other_threads; 120| 2| size_t stack_data; 121| 2| char *stack_addr; 122| 2| size_t stack_size; 123| 2| char *stack_base; 124| | 125| 2| KA_TRACE( 126| 2| 1000, 127| 2| ("*** __kmp_get_global_thread_id: entering, nproc=%d all_nproc=%d\n", 128| 2| __kmp_nth, __kmp_all_nth)); 129| | 130| | /* JPH - to handle the case where __kmpc_end(0) is called immediately prior to 131| | a parallel region, made it return KMP_GTID_DNE to force serial_initialize 132| | by caller. Had to handle KMP_GTID_DNE at all call-sites, or else guarantee 133| | __kmp_init_gtid for this to work. */ 134| | 135| 2| if (!TCR_4(__kmp_init_gtid)) ------------------ | | 1127| 2|#define TCR_4(a) (a) ------------------ | Branch (135:7): [True: 0, False: 2] ------------------ 136| 0| return KMP_GTID_DNE; ------------------ | | 1000| 0|#define KMP_GTID_DNE (-2) /* Does not exist */ ------------------ 137| | 138| 2|#ifdef KMP_TDATA_GTID 139| 2| if (TCR_4(__kmp_gtid_mode) >= 3) { ------------------ | | 1127| 2|#define TCR_4(a) (a) ------------------ | Branch (139:7): [True: 2, False: 0] ------------------ 140| 2| KA_TRACE(1000, ("*** __kmp_get_global_thread_id: using TDATA\n")); 141| 2| return __kmp_gtid; 142| 2| } 143| 0|#endif 144| 0| if (TCR_4(__kmp_gtid_mode) >= 2) { ------------------ | | 1127| 0|#define TCR_4(a) (a) ------------------ | Branch (144:7): [True: 0, False: 0] ------------------ 145| 0| KA_TRACE(1000, ("*** __kmp_get_global_thread_id: using keyed TLS\n")); 146| 0| return __kmp_gtid_get_specific(); 147| 0| } 148| 0| KA_TRACE(1000, ("*** __kmp_get_global_thread_id: using internal alg.\n")); 149| | 150| 0| stack_addr = (char *)&stack_data; 151| 0| other_threads = __kmp_threads; 152| | 153| | /* ATT: The code below is a source of potential bugs due to unsynchronized 154| | access to __kmp_threads array. For example: 155| | 1. Current thread loads other_threads[i] to thr and checks it, it is 156| | non-NULL. 157| | 2. Current thread is suspended by OS. 158| | 3. Another thread unregisters and finishes (debug versions of free() 159| | may fill memory with something like 0xEF). 160| | 4. Current thread is resumed. 161| | 5. Current thread reads junk from *thr. 162| | TODO: Fix it. --ln */ 163| | 164| 0| for (i = 0; i < __kmp_threads_capacity; i++) { ------------------ | Branch (164:15): [True: 0, False: 0] ------------------ 165| | 166| 0| kmp_info_t *thr = (kmp_info_t *)TCR_SYNC_PTR(other_threads[i]); ------------------ | | 1158| 0|#define TCR_SYNC_PTR(a) ((void *)TCR_SYNC_8(a)) | | ------------------ | | | | 1140| 0|#define TCR_SYNC_8(a) (a) | | ------------------ ------------------ 167| 0| if (!thr) ------------------ | Branch (167:9): [True: 0, False: 0] ------------------ 168| 0| continue; 169| | 170| 0| stack_size = (size_t)TCR_PTR(thr->th.th_info.ds.ds_stacksize); ------------------ | | 1156| 0|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 0|#define TCR_8(a) (a) | | ------------------ ------------------ 171| 0| stack_base = (char *)TCR_PTR(thr->th.th_info.ds.ds_stackbase); ------------------ | | 1156| 0|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 0|#define TCR_8(a) (a) | | ------------------ ------------------ 172| | 173| | /* stack grows down -- search through all of the active threads */ 174| | 175| 0| if (stack_addr <= stack_base) { ------------------ | Branch (175:9): [True: 0, False: 0] ------------------ 176| 0| size_t stack_diff = stack_base - stack_addr; 177| | 178| 0| if (stack_diff <= stack_size) { ------------------ | Branch (178:11): [True: 0, False: 0] ------------------ 179| | /* The only way we can be closer than the allocated */ 180| | /* stack size is if we are running on this thread. */ 181| 0| KMP_DEBUG_ASSERT(__kmp_gtid_get_specific() == i); 182| 0| return i; 183| 0| } 184| 0| } 185| 0| } 186| | 187| | /* get specific to try and determine our gtid */ 188| 0| KA_TRACE(1000, 189| 0| ("*** __kmp_get_global_thread_id: internal alg. failed to find " 190| 0| "thread, using TLS\n")); 191| 0| i = __kmp_gtid_get_specific(); 192| | 193| | /*fprintf( stderr, "=== %d\n", i ); */ /* GROO */ 194| | 195| | /* if we havn't been assigned a gtid, then return code */ 196| 0| if (i < 0) ------------------ | Branch (196:7): [True: 0, False: 0] ------------------ 197| 0| return i; 198| | 199| | /* dynamically updated stack window for uber threads to avoid get_specific 200| | call */ 201| 0| if (!TCR_4(other_threads[i]->th.th_info.ds.ds_stackgrow)) { ------------------ | | 1127| 0|#define TCR_4(a) (a) ------------------ | Branch (201:7): [True: 0, False: 0] ------------------ 202| 0| KMP_FATAL(StackOverflow, i); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 203| 0| } 204| | 205| 0| stack_base = (char *)other_threads[i]->th.th_info.ds.ds_stackbase; 206| 0| if (stack_addr > stack_base) { ------------------ | Branch (206:7): [True: 0, False: 0] ------------------ 207| 0| TCW_PTR(other_threads[i]->th.th_info.ds.ds_stackbase, stack_addr); ------------------ | | 1157| 0|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 0|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 208| 0| TCW_PTR(other_threads[i]->th.th_info.ds.ds_stacksize, ------------------ | | 1157| 0|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 0|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 209| 0| other_threads[i]->th.th_info.ds.ds_stacksize + stack_addr - 210| 0| stack_base); 211| 0| } else { 212| 0| TCW_PTR(other_threads[i]->th.th_info.ds.ds_stacksize, ------------------ | | 1157| 0|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 0|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 213| 0| stack_base - stack_addr); 214| 0| } 215| | 216| | /* Reprint stack bounds for ubermaster since they have been refined */ 217| 0| if (__kmp_storage_map) { ------------------ | Branch (217:7): [True: 0, False: 0] ------------------ 218| 0| char *stack_end = (char *)other_threads[i]->th.th_info.ds.ds_stackbase; 219| 0| char *stack_beg = stack_end - other_threads[i]->th.th_info.ds.ds_stacksize; 220| 0| __kmp_print_storage_map_gtid(i, stack_beg, stack_end, 221| 0| other_threads[i]->th.th_info.ds.ds_stacksize, 222| 0| "th_%d stack (refinement)", i); 223| 0| } 224| 0| return i; 225| 0|} __kmp_get_global_thread_id_reg: 227| 225k|int __kmp_get_global_thread_id_reg() { 228| 225k| int gtid; 229| | 230| 225k| if (!__kmp_init_serial) { ------------------ | Branch (230:7): [True: 1, False: 225k] ------------------ 231| 1| gtid = KMP_GTID_DNE; ------------------ | | 1000| 1|#define KMP_GTID_DNE (-2) /* Does not exist */ ------------------ 232| 1| } else 233| 225k|#ifdef KMP_TDATA_GTID 234| 225k| if (TCR_4(__kmp_gtid_mode) >= 3) { ------------------ | | 1127| 225k|#define TCR_4(a) (a) ------------------ | Branch (234:11): [True: 225k, False: 18.4E] ------------------ 235| 225k| KA_TRACE(1000, ("*** __kmp_get_global_thread_id_reg: using TDATA\n")); 236| 225k| gtid = __kmp_gtid; 237| 225k| } else 238| 18.4E|#endif 239| 18.4E| if (TCR_4(__kmp_gtid_mode) >= 2) { ------------------ | | 1127| 18.4E|#define TCR_4(a) (a) ------------------ | Branch (239:11): [True: 0, False: 18.4E] ------------------ 240| 0| KA_TRACE(1000, ("*** __kmp_get_global_thread_id_reg: using keyed TLS\n")); 241| 0| gtid = __kmp_gtid_get_specific(); 242| 18.4E| } else { 243| 18.4E| KA_TRACE(1000, 244| 18.4E| ("*** __kmp_get_global_thread_id_reg: using internal alg.\n")); 245| 18.4E| gtid = __kmp_get_global_thread_id(); 246| 18.4E| } 247| | 248| | /* we must be a new uber master sibling thread */ 249| 225k| if (gtid == KMP_GTID_DNE) { ------------------ | | 1000| 225k|#define KMP_GTID_DNE (-2) /* Does not exist */ ------------------ | Branch (249:7): [True: 1, False: 225k] ------------------ 250| 1| KA_TRACE(10, 251| 1| ("__kmp_get_global_thread_id_reg: Encountered new root thread. " 252| 1| "Registering a new gtid.\n")); 253| 1| __kmp_acquire_bootstrap_lock(&__kmp_initz_lock); 254| 1| if (!__kmp_init_serial) { ------------------ | Branch (254:9): [True: 1, False: 0] ------------------ 255| 1| __kmp_do_serial_initialize(); 256| 1| gtid = __kmp_gtid_get_specific(); 257| 1| } else { 258| 0| gtid = __kmp_register_root(FALSE); ------------------ | | 1279| 0|#define FALSE 0 ------------------ 259| 0| } 260| 1| __kmp_release_bootstrap_lock(&__kmp_initz_lock); 261| | /*__kmp_printf( "+++ %d\n", gtid ); */ /* GROO */ 262| 1| } 263| | 264| 225k| KMP_DEBUG_ASSERT(gtid >= 0); 265| | 266| 225k| return gtid; 267| 225k|} __kmp_check_stack_overlap: 270| 32|void __kmp_check_stack_overlap(kmp_info_t *th) { 271| 32| int f; 272| 32| char *stack_beg = NULL; 273| 32| char *stack_end = NULL; 274| 32| int gtid; 275| | 276| 32| KA_TRACE(10, ("__kmp_check_stack_overlap: called\n")); 277| 32| if (__kmp_storage_map) { ------------------ | Branch (277:7): [True: 0, False: 32] ------------------ 278| 0| stack_end = (char *)th->th.th_info.ds.ds_stackbase; 279| 0| stack_beg = stack_end - th->th.th_info.ds.ds_stacksize; 280| | 281| 0| gtid = __kmp_gtid_from_thread(th); 282| | 283| 0| if (gtid == KMP_GTID_MONITOR) { ------------------ | | 1002| 0|#define KMP_GTID_MONITOR (-4) /* Monitor thread ID */ ------------------ | Branch (283:9): [True: 0, False: 0] ------------------ 284| 0| __kmp_print_storage_map_gtid( 285| 0| gtid, stack_beg, stack_end, th->th.th_info.ds.ds_stacksize, 286| 0| "th_%s stack (%s)", "mon", 287| 0| (th->th.th_info.ds.ds_stackgrow) ? "initial" : "actual"); ------------------ | Branch (287:11): [True: 0, False: 0] ------------------ 288| 0| } else { 289| 0| __kmp_print_storage_map_gtid( 290| 0| gtid, stack_beg, stack_end, th->th.th_info.ds.ds_stacksize, 291| 0| "th_%d stack (%s)", gtid, 292| 0| (th->th.th_info.ds.ds_stackgrow) ? "initial" : "actual"); ------------------ | Branch (292:11): [True: 0, False: 0] ------------------ 293| 0| } 294| 0| } 295| | 296| | /* No point in checking ubermaster threads since they use refinement and 297| | * cannot overlap */ 298| 32| gtid = __kmp_gtid_from_thread(th); 299| 32| if (__kmp_env_checks == TRUE && !KMP_UBER_GTID(gtid)) { ------------------ | | 1280| 64|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 32|#define FALSE 0 | | ------------------ ------------------ | Branch (299:7): [True: 0, False: 32] | Branch (299:35): [True: 0, False: 0] ------------------ 300| 0| KA_TRACE(10, 301| 0| ("__kmp_check_stack_overlap: performing extensive checking\n")); 302| 0| if (stack_beg == NULL) { ------------------ | Branch (302:9): [True: 0, False: 0] ------------------ 303| 0| stack_end = (char *)th->th.th_info.ds.ds_stackbase; 304| 0| stack_beg = stack_end - th->th.th_info.ds.ds_stacksize; 305| 0| } 306| | 307| 0| for (f = 0; f < __kmp_threads_capacity; f++) { ------------------ | Branch (307:17): [True: 0, False: 0] ------------------ 308| 0| kmp_info_t *f_th = (kmp_info_t *)TCR_SYNC_PTR(__kmp_threads[f]); ------------------ | | 1158| 0|#define TCR_SYNC_PTR(a) ((void *)TCR_SYNC_8(a)) | | ------------------ | | | | 1140| 0|#define TCR_SYNC_8(a) (a) | | ------------------ ------------------ 309| | 310| 0| if (f_th && f_th != th) { ------------------ | Branch (310:11): [True: 0, False: 0] | Branch (310:19): [True: 0, False: 0] ------------------ 311| 0| char *other_stack_end = 312| 0| (char *)TCR_PTR(f_th->th.th_info.ds.ds_stackbase); ------------------ | | 1156| 0|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 0|#define TCR_8(a) (a) | | ------------------ ------------------ 313| 0| char *other_stack_beg = 314| 0| other_stack_end - (size_t)TCR_PTR(f_th->th.th_info.ds.ds_stacksize); ------------------ | | 1156| 0|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 0|#define TCR_8(a) (a) | | ------------------ ------------------ 315| 0| if ((stack_beg > other_stack_beg && stack_beg < other_stack_end) || ------------------ | Branch (315:14): [True: 0, False: 0] | Branch (315:45): [True: 0, False: 0] ------------------ 316| 0| (stack_end > other_stack_beg && stack_end < other_stack_end)) { ------------------ | Branch (316:14): [True: 0, False: 0] | Branch (316:45): [True: 0, False: 0] ------------------ 317| | 318| | /* Print the other stack values before the abort */ 319| 0| if (__kmp_storage_map) ------------------ | Branch (319:15): [True: 0, False: 0] ------------------ 320| 0| __kmp_print_storage_map_gtid( 321| 0| -1, other_stack_beg, other_stack_end, 322| 0| (size_t)TCR_PTR(f_th->th.th_info.ds.ds_stacksize), ------------------ | | 1156| 0|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 0|#define TCR_8(a) (a) | | ------------------ ------------------ 323| 0| "th_%d stack (overlapped)", __kmp_gtid_from_thread(f_th)); 324| | 325| 0| __kmp_fatal(KMP_MSG(StackOverlap), KMP_HNT(ChangeStackLimit), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ __kmp_fatal(KMP_MSG(StackOverlap), KMP_HNT(ChangeStackLimit), ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 326| 0| __kmp_msg_null); 327| 0| } 328| 0| } 329| 0| } 330| 0| } 331| 32| KA_TRACE(10, ("__kmp_check_stack_overlap: returning\n")); 332| 32|} __kmp_serialized_parallel: 1113| 3.01k|void __kmp_serialized_parallel(ident_t *loc, kmp_int32 global_tid) { 1114| 3.01k| kmp_info_t *this_thr; 1115| 3.01k| kmp_team_t *serial_team; 1116| | 1117| 3.01k| KC_TRACE(10, ("__kmpc_serialized_parallel: called by T#%d\n", global_tid)); 1118| | 1119| | /* Skip all this code for autopar serialized loops since it results in 1120| | unacceptable overhead */ 1121| 3.01k| if (loc != NULL && (loc->flags & KMP_IDENT_AUTOPAR)) ------------------ | Branch (1121:7): [True: 3.01k, False: 0] | Branch (1121:22): [True: 0, False: 3.01k] ------------------ 1122| 0| return; 1123| | 1124| 3.01k| if (!TCR_4(__kmp_init_parallel)) ------------------ | | 1127| 3.01k|#define TCR_4(a) (a) ------------------ | Branch (1124:7): [True: 0, False: 3.01k] ------------------ 1125| 0| __kmp_parallel_initialize(); 1126| 3.01k| __kmp_resume_if_soft_paused(); 1127| | 1128| 3.01k| this_thr = __kmp_threads[global_tid]; 1129| 3.01k| serial_team = this_thr->th.th_serial_team; 1130| | 1131| | /* utilize the serialized team held by this thread */ 1132| 3.01k| KMP_DEBUG_ASSERT(serial_team); 1133| 3.01k| KMP_MB(); 1134| | 1135| 3.01k| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (1135:7): [True: 3.01k, False: 0] ------------------ 1136| 3.01k| KMP_DEBUG_ASSERT( 1137| 3.01k| this_thr->th.th_task_team == 1138| 3.01k| this_thr->th.th_team->t.t_task_team[this_thr->th.th_task_state]); 1139| 3.01k| KMP_DEBUG_ASSERT(serial_team->t.t_task_team[this_thr->th.th_task_state] == 1140| 3.01k| NULL); 1141| 3.01k| KA_TRACE(20, ("__kmpc_serialized_parallel: T#%d pushing task_team %p / " 1142| 3.01k| "team %p, new task_team = NULL\n", 1143| 3.01k| global_tid, this_thr->th.th_task_team, this_thr->th.th_team)); 1144| 3.01k| this_thr->th.th_task_team = NULL; 1145| 3.01k| } 1146| | 1147| 3.01k| kmp_proc_bind_t proc_bind = this_thr->th.th_set_proc_bind; 1148| 3.01k| if (this_thr->th.th_current_task->td_icvs.proc_bind == proc_bind_false) { ------------------ | Branch (1148:7): [True: 3.01k, False: 0] ------------------ 1149| 3.01k| proc_bind = proc_bind_false; 1150| 3.01k| } else if (proc_bind == proc_bind_default) { ------------------ | Branch (1150:14): [True: 0, False: 0] ------------------ 1151| | // No proc_bind clause was specified, so use the current value 1152| | // of proc-bind-var for this parallel region. 1153| 0| proc_bind = this_thr->th.th_current_task->td_icvs.proc_bind; 1154| 0| } 1155| | // Reset for next parallel region 1156| 3.01k| this_thr->th.th_set_proc_bind = proc_bind_default; 1157| | 1158| | // Reset num_threads for next parallel region 1159| 3.01k| this_thr->th.th_set_nproc = 0; 1160| | 1161| 3.01k|#if OMPT_SUPPORT 1162| 3.01k| ompt_data_t ompt_parallel_data = ompt_data_none; ------------------ | | 1408| 3.01k|#define ompt_data_none {0} ------------------ 1163| 3.01k| void *codeptr = OMPT_LOAD_RETURN_ADDRESS(global_tid); ------------------ | | 93| 3.01k|#define OMPT_LOAD_RETURN_ADDRESS(gtid) __ompt_load_return_address(gtid) ------------------ 1164| 3.01k| if (ompt_enabled.enabled && ------------------ | Branch (1164:7): [True: 0, False: 3.01k] ------------------ 1165| 0| this_thr->th.ompt_thread_info.state != ompt_state_overhead) { ------------------ | Branch (1165:7): [True: 0, False: 0] ------------------ 1166| | 1167| 0| ompt_task_info_t *parent_task_info; 1168| 0| parent_task_info = OMPT_CUR_TASK_INFO(this_thr); ------------------ | | 68| 0|#define OMPT_CUR_TASK_INFO(thr) (&((thr)->th.th_current_task->ompt_task_info)) ------------------ 1169| | 1170| 0| parent_task_info->frame.enter_frame.ptr = OMPT_GET_FRAME_ADDRESS(0); ------------------ | | 109| 0|#define OMPT_GET_FRAME_ADDRESS(level) __builtin_frame_address(level) ------------------ 1171| 0| if (ompt_enabled.ompt_callback_parallel_begin) { ------------------ | Branch (1171:9): [True: 0, False: 0] ------------------ 1172| 0| int team_size = 1; 1173| | 1174| 0| ompt_callbacks.ompt_callback(ompt_callback_parallel_begin)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 1175| 0| &(parent_task_info->task_data), &(parent_task_info->frame), 1176| 0| &ompt_parallel_data, team_size, 1177| 0| ompt_parallel_invoker_program | ompt_parallel_team, codeptr); 1178| 0| } 1179| 0| } 1180| 3.01k|#endif // OMPT_SUPPORT 1181| | 1182| 3.01k| if (this_thr->th.th_team != serial_team) { ------------------ | Branch (1182:7): [True: 3.01k, False: 0] ------------------ 1183| | // Nested level will be an index in the nested nthreads array 1184| 3.01k| int level = this_thr->th.th_team->t.t_level; 1185| | 1186| 3.01k| if (serial_team->t.t_serialized) { ------------------ | Branch (1186:9): [True: 0, False: 3.01k] ------------------ 1187| | /* this serial team was already used 1188| | TODO increase performance by making this locks more specific */ 1189| 0| kmp_team_t *new_team; 1190| | 1191| 0| __kmp_acquire_bootstrap_lock(&__kmp_forkjoin_lock); 1192| | 1193| 0| new_team = 1194| 0| __kmp_allocate_team(this_thr->th.th_root, 1, 1, 1195| 0|#if OMPT_SUPPORT 1196| 0| ompt_parallel_data, 1197| 0|#endif 1198| 0| proc_bind, &this_thr->th.th_current_task->td_icvs, 1199| 0| 0 USE_NESTED_HOT_ARG(NULL)); ------------------ | | 165| 0|#define USE_NESTED_HOT_ARG(x) , x ------------------ 1200| 0| __kmp_release_bootstrap_lock(&__kmp_forkjoin_lock); 1201| 0| KMP_ASSERT(new_team); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 1202| | 1203| | /* setup new serialized team and install it */ 1204| 0| new_team->t.t_threads[0] = this_thr; 1205| 0| new_team->t.t_parent = this_thr->th.th_team; 1206| 0| serial_team = new_team; 1207| 0| this_thr->th.th_serial_team = serial_team; 1208| | 1209| 0| KF_TRACE( 1210| 0| 10, 1211| 0| ("__kmpc_serialized_parallel: T#%d allocated new serial team %p\n", 1212| 0| global_tid, serial_team)); 1213| | 1214| | /* TODO the above breaks the requirement that if we run out of resources, 1215| | then we can still guarantee that serialized teams are ok, since we may 1216| | need to allocate a new one */ 1217| 3.01k| } else { 1218| 3.01k| KF_TRACE( 1219| 3.01k| 10, 1220| 3.01k| ("__kmpc_serialized_parallel: T#%d reusing cached serial team %p\n", 1221| 3.01k| global_tid, serial_team)); 1222| 3.01k| } 1223| | 1224| | /* we have to initialize this serial team */ 1225| 3.01k| KMP_DEBUG_ASSERT(serial_team->t.t_threads); 1226| 3.01k| KMP_DEBUG_ASSERT(serial_team->t.t_threads[0] == this_thr); 1227| 3.01k| KMP_DEBUG_ASSERT(this_thr->th.th_team != serial_team); 1228| 3.01k| serial_team->t.t_ident = loc; 1229| 3.01k| serial_team->t.t_serialized = 1; 1230| 3.01k| serial_team->t.t_nproc = 1; 1231| 3.01k| serial_team->t.t_parent = this_thr->th.th_team; 1232| 3.01k| serial_team->t.t_sched.sched = this_thr->th.th_team->t.t_sched.sched; 1233| 3.01k| this_thr->th.th_team = serial_team; 1234| 3.01k| serial_team->t.t_master_tid = this_thr->th.th_info.ds.ds_tid; 1235| | 1236| 3.01k| KF_TRACE(10, ("__kmpc_serialized_parallel: T#%d curtask=%p\n", global_tid, 1237| 3.01k| this_thr->th.th_current_task)); 1238| 3.01k| KMP_ASSERT(this_thr->th.th_current_task->td_flags.executing == 1); ------------------ | | 50| 3.01k| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 3.01k] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 1239| 3.01k| this_thr->th.th_current_task->td_flags.executing = 0; 1240| | 1241| 3.01k| __kmp_push_current_task_to_thread(this_thr, serial_team, 0); 1242| | 1243| | /* TODO: GEH: do ICVs work for nested serialized teams? Don't we need an 1244| | implicit task for each serialized task represented by 1245| | team->t.t_serialized? */ 1246| 3.01k| copy_icvs(&this_thr->th.th_current_task->td_icvs, 1247| 3.01k| &this_thr->th.th_current_task->td_parent->td_icvs); 1248| | 1249| | // Thread value exists in the nested nthreads array for the next nested 1250| | // level 1251| 3.01k| if (__kmp_nested_nth.used && (level + 1 < __kmp_nested_nth.used)) { ------------------ | Branch (1251:9): [True: 0, False: 3.01k] | Branch (1251:34): [True: 0, False: 0] ------------------ 1252| 0| this_thr->th.th_current_task->td_icvs.nproc = 1253| 0| __kmp_nested_nth.nth[level + 1]; 1254| 0| } 1255| | 1256| 3.01k| if (__kmp_nested_proc_bind.used && ------------------ | Branch (1256:9): [True: 3.01k, False: 0] ------------------ 1257| 3.01k| (level + 1 < __kmp_nested_proc_bind.used)) { ------------------ | Branch (1257:9): [True: 0, False: 3.01k] ------------------ 1258| 0| this_thr->th.th_current_task->td_icvs.proc_bind = 1259| 0| __kmp_nested_proc_bind.bind_types[level + 1]; 1260| 0| } 1261| | 1262| |#if USE_DEBUGGER 1263| | serial_team->t.t_pkfn = (microtask_t)(~0); // For the debugger. 1264| |#endif 1265| 3.01k| this_thr->th.th_info.ds.ds_tid = 0; 1266| | 1267| | /* set thread cache values */ 1268| 3.01k| this_thr->th.th_team_nproc = 1; 1269| 3.01k| this_thr->th.th_team_master = this_thr; 1270| 3.01k| this_thr->th.th_team_serialized = 1; 1271| | 1272| 3.01k| serial_team->t.t_level = serial_team->t.t_parent->t.t_level + 1; 1273| 3.01k| serial_team->t.t_active_level = serial_team->t.t_parent->t.t_active_level; 1274| 3.01k| serial_team->t.t_def_allocator = this_thr->th.th_def_allocator; // save 1275| | 1276| 3.01k| propagateFPControl(serial_team); 1277| | 1278| | /* check if we need to allocate dispatch buffers stack */ 1279| 3.01k| KMP_DEBUG_ASSERT(serial_team->t.t_dispatch); 1280| 3.01k| if (!serial_team->t.t_dispatch->th_disp_buffer) { ------------------ | Branch (1280:9): [True: 3.01k, False: 0] ------------------ 1281| 3.01k| serial_team->t.t_dispatch->th_disp_buffer = 1282| 3.01k| (dispatch_private_info_t *)__kmp_allocate( ------------------ | | 3617| 3.01k|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 1283| 3.01k| sizeof(dispatch_private_info_t)); 1284| 3.01k| } 1285| 3.01k| this_thr->th.th_dispatch = serial_team->t.t_dispatch; 1286| | 1287| 3.01k| KMP_MB(); 1288| | 1289| 3.01k| } else { 1290| | /* this serialized team is already being used, 1291| | * that's fine, just add another nested level */ 1292| 0| KMP_DEBUG_ASSERT(this_thr->th.th_team == serial_team); 1293| 0| KMP_DEBUG_ASSERT(serial_team->t.t_threads); 1294| 0| KMP_DEBUG_ASSERT(serial_team->t.t_threads[0] == this_thr); 1295| 0| ++serial_team->t.t_serialized; 1296| 0| this_thr->th.th_team_serialized = serial_team->t.t_serialized; 1297| | 1298| | // Nested level will be an index in the nested nthreads array 1299| 0| int level = this_thr->th.th_team->t.t_level; 1300| | // Thread value exists in the nested nthreads array for the next nested 1301| | // level 1302| 0| if (__kmp_nested_nth.used && (level + 1 < __kmp_nested_nth.used)) { ------------------ | Branch (1302:9): [True: 0, False: 0] | Branch (1302:34): [True: 0, False: 0] ------------------ 1303| 0| this_thr->th.th_current_task->td_icvs.nproc = 1304| 0| __kmp_nested_nth.nth[level + 1]; 1305| 0| } 1306| 0| serial_team->t.t_level++; 1307| 0| KF_TRACE(10, ("__kmpc_serialized_parallel: T#%d increasing nesting level " 1308| 0| "of serial team %p to %d\n", 1309| 0| global_tid, serial_team, serial_team->t.t_level)); 1310| | 1311| | /* allocate/push dispatch buffers stack */ 1312| 0| KMP_DEBUG_ASSERT(serial_team->t.t_dispatch); 1313| 0| { 1314| 0| dispatch_private_info_t *disp_buffer = 1315| 0| (dispatch_private_info_t *)__kmp_allocate( ------------------ | | 3617| 0|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 1316| 0| sizeof(dispatch_private_info_t)); 1317| 0| disp_buffer->next = serial_team->t.t_dispatch->th_disp_buffer; 1318| 0| serial_team->t.t_dispatch->th_disp_buffer = disp_buffer; 1319| 0| } 1320| 0| this_thr->th.th_dispatch = serial_team->t.t_dispatch; 1321| | 1322| 0| KMP_MB(); 1323| 0| } 1324| 3.01k| KMP_CHECK_UPDATE(serial_team->t.t_cancel_request, cancel_noreq); ------------------ | | 2280| 3.01k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 3.01k] | | ------------------ | | 2281| 3.01k| (a) = (b) ------------------ 1325| | 1326| | // Perform the display affinity functionality for 1327| | // serialized parallel regions 1328| 3.01k| if (__kmp_display_affinity) { ------------------ | Branch (1328:7): [True: 0, False: 3.01k] ------------------ 1329| 0| if (this_thr->th.th_prev_level != serial_team->t.t_level || ------------------ | Branch (1329:9): [True: 0, False: 0] ------------------ 1330| 0| this_thr->th.th_prev_num_threads != 1) { ------------------ | Branch (1330:9): [True: 0, False: 0] ------------------ 1331| | // NULL means use the affinity-format-var ICV 1332| 0| __kmp_aux_display_affinity(global_tid, NULL); 1333| 0| this_thr->th.th_prev_level = serial_team->t.t_level; 1334| 0| this_thr->th.th_prev_num_threads = 1; 1335| 0| } 1336| 0| } 1337| | 1338| 3.01k| if (__kmp_env_consistency_check) ------------------ | Branch (1338:7): [True: 0, False: 3.01k] ------------------ 1339| 0| __kmp_push_parallel(global_tid, NULL); 1340| 3.01k|#if OMPT_SUPPORT 1341| 3.01k| serial_team->t.ompt_team_info.master_return_address = codeptr; 1342| 3.01k| if (ompt_enabled.enabled && ------------------ | Branch (1342:7): [True: 0, False: 3.01k] ------------------ 1343| 0| this_thr->th.ompt_thread_info.state != ompt_state_overhead) { ------------------ | Branch (1343:7): [True: 0, False: 0] ------------------ 1344| 0| OMPT_CUR_TASK_INFO(this_thr)->frame.exit_frame.ptr = ------------------ | | 68| 0|#define OMPT_CUR_TASK_INFO(thr) (&((thr)->th.th_current_task->ompt_task_info)) ------------------ 1345| 0| OMPT_GET_FRAME_ADDRESS(0); ------------------ | | 109| 0|#define OMPT_GET_FRAME_ADDRESS(level) __builtin_frame_address(level) ------------------ 1346| | 1347| 0| ompt_lw_taskteam_t lw_taskteam; 1348| 0| __ompt_lw_taskteam_init(&lw_taskteam, this_thr, global_tid, 1349| 0| &ompt_parallel_data, codeptr); 1350| | 1351| 0| __ompt_lw_taskteam_link(&lw_taskteam, this_thr, 1); 1352| | // don't use lw_taskteam after linking. content was swaped 1353| | 1354| | /* OMPT implicit task begin */ 1355| 0| if (ompt_enabled.ompt_callback_implicit_task) { ------------------ | Branch (1355:9): [True: 0, False: 0] ------------------ 1356| 0| ompt_callbacks.ompt_callback(ompt_callback_implicit_task)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 1357| 0| ompt_scope_begin, OMPT_CUR_TEAM_DATA(this_thr), ------------------ | | 73| 0| (&((thr)->th.th_team->t.ompt_team_info.parallel_data)) ------------------ 1358| 0| OMPT_CUR_TASK_DATA(this_thr), 1, __kmp_tid_from_gtid(global_tid), ------------------ | | 70| 0| (&((thr)->th.th_current_task->ompt_task_info.task_data)) ------------------ 1359| 0| ompt_task_implicit); // TODO: Can this be ompt_task_initial? 1360| 0| OMPT_CUR_TASK_INFO(this_thr)->thread_num = ------------------ | | 68| 0|#define OMPT_CUR_TASK_INFO(thr) (&((thr)->th.th_current_task->ompt_task_info)) ------------------ 1361| 0| __kmp_tid_from_gtid(global_tid); 1362| 0| } 1363| | 1364| | /* OMPT state */ 1365| 0| this_thr->th.ompt_thread_info.state = ompt_state_work_parallel; 1366| 0| OMPT_CUR_TASK_INFO(this_thr)->frame.exit_frame.ptr = ------------------ | | 68| 0|#define OMPT_CUR_TASK_INFO(thr) (&((thr)->th.th_current_task->ompt_task_info)) ------------------ 1367| 0| OMPT_GET_FRAME_ADDRESS(0); ------------------ | | 109| 0|#define OMPT_GET_FRAME_ADDRESS(level) __builtin_frame_address(level) ------------------ 1368| 0| } 1369| 3.01k|#endif 1370| 3.01k|} __kmp_fork_call: 1863| 1.35k| kmp_va_list ap) { 1864| 1.35k| void **argv; 1865| 1.35k| int i; 1866| 1.35k| int master_tid; 1867| 1.35k| int master_this_cons; 1868| 1.35k| kmp_team_t *team; 1869| 1.35k| kmp_team_t *parent_team; 1870| 1.35k| kmp_info_t *master_th; 1871| 1.35k| kmp_root_t *root; 1872| 1.35k| int nthreads; 1873| 1.35k| int master_active; 1874| 1.35k| int master_set_numthreads; 1875| 1.35k| int level; 1876| 1.35k| int active_level; 1877| 1.35k| int teams_level; 1878| 1.35k|#if KMP_NESTED_HOT_TEAMS 1879| 1.35k| kmp_hot_team_ptr_t **p_hot_teams; 1880| 1.35k|#endif 1881| 1.35k| { // KMP_TIME_BLOCK 1882| 1.35k| KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(KMP_fork_call); ------------------ | | 1008| 1.35k|#define KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(n) ((void)0) ------------------ 1883| 1.35k| KMP_COUNT_VALUE(OMP_PARALLEL_args, argc); ------------------ | | 1000| 1.35k|#define KMP_COUNT_VALUE(n, v) ((void)0) ------------------ 1884| | 1885| 1.35k| KA_TRACE(20, ("__kmp_fork_call: enter T#%d\n", gtid)); 1886| 1.35k| if (__kmp_stkpadding > 0 && __kmp_root[gtid] != NULL) { ------------------ | Branch (1886:9): [True: 0, False: 1.35k] | Branch (1886:33): [True: 0, False: 0] ------------------ 1887| | /* Some systems prefer the stack for the root thread(s) to start with */ 1888| | /* some gap from the parent stack to prevent false sharing. */ 1889| 0| void *dummy = KMP_ALLOCA(__kmp_stkpadding); ------------------ | | 59| 0|#define KMP_ALLOCA alloca ------------------ 1890| | /* These 2 lines below are so this does not get optimized out */ 1891| 0| if (__kmp_stkpadding > KMP_MAX_STKPADDING) ------------------ | | 1178| 0|#define KMP_MAX_STKPADDING (2 * 1024 * 1024) ------------------ | Branch (1891:11): [True: 0, False: 0] ------------------ 1892| 0| __kmp_stkpadding += (short)((kmp_int64)dummy); 1893| 0| } 1894| | 1895| | /* initialize if needed */ 1896| 1.35k| KMP_DEBUG_ASSERT( 1897| 1.35k| __kmp_init_serial); // AC: potentially unsafe, not in sync with shutdown 1898| 1.35k| if (!TCR_4(__kmp_init_parallel)) ------------------ | | 1127| 1.35k|#define TCR_4(a) (a) ------------------ | Branch (1898:9): [True: 1, False: 1.35k] ------------------ 1899| 1| __kmp_parallel_initialize(); 1900| 1.35k| __kmp_resume_if_soft_paused(); 1901| | 1902| | /* setup current data */ 1903| | // AC: potentially unsafe, not in sync with library shutdown, 1904| | // __kmp_threads can be freed 1905| 1.35k| master_th = __kmp_threads[gtid]; 1906| | 1907| 1.35k| parent_team = master_th->th.th_team; 1908| 1.35k| master_tid = master_th->th.th_info.ds.ds_tid; 1909| 1.35k| master_this_cons = master_th->th.th_local.this_construct; 1910| 1.35k| root = master_th->th.th_root; 1911| 1.35k| master_active = root->r.r_active; 1912| 1.35k| master_set_numthreads = master_th->th.th_set_nproc; 1913| | 1914| 1.35k|#if OMPT_SUPPORT 1915| 1.35k| ompt_data_t ompt_parallel_data = ompt_data_none; ------------------ | | 1408| 1.35k|#define ompt_data_none {0} ------------------ 1916| 1.35k| ompt_data_t *parent_task_data; 1917| 1.35k| ompt_frame_t *ompt_frame; 1918| 1.35k| void *return_address = NULL; 1919| | 1920| 1.35k| if (ompt_enabled.enabled) { ------------------ | Branch (1920:9): [True: 0, False: 1.35k] ------------------ 1921| 0| __ompt_get_task_info_internal(0, NULL, &parent_task_data, &ompt_frame, 1922| 0| NULL, NULL); 1923| 0| return_address = OMPT_LOAD_RETURN_ADDRESS(gtid); ------------------ | | 93| 0|#define OMPT_LOAD_RETURN_ADDRESS(gtid) __ompt_load_return_address(gtid) ------------------ 1924| 0| } 1925| 1.35k|#endif 1926| | 1927| | // Assign affinity to root thread if it hasn't happened yet 1928| 1.35k| __kmp_assign_root_init_mask(); 1929| | 1930| | // Nested level will be an index in the nested nthreads array 1931| 1.35k| level = parent_team->t.t_level; 1932| | // used to launch non-serial teams even if nested is not allowed 1933| 1.35k| active_level = parent_team->t.t_active_level; 1934| | // needed to check nesting inside the teams 1935| 1.35k| teams_level = master_th->th.th_teams_level; 1936| 1.35k|#if KMP_NESTED_HOT_TEAMS 1937| 1.35k| p_hot_teams = &master_th->th.th_hot_teams; 1938| 1.35k| if (*p_hot_teams == NULL && __kmp_hot_teams_max_level > 0) { ------------------ | Branch (1938:9): [True: 1, False: 1.35k] | Branch (1938:33): [True: 1, False: 0] ------------------ 1939| 1| *p_hot_teams = (kmp_hot_team_ptr_t *)__kmp_allocate( ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 1940| 1| sizeof(kmp_hot_team_ptr_t) * __kmp_hot_teams_max_level); 1941| 1| (*p_hot_teams)[0].hot_team = root->r.r_hot_team; 1942| | // it is either actual or not needed (when active_level > 0) 1943| 1| (*p_hot_teams)[0].hot_team_nth = 1; 1944| 1| } 1945| 1.35k|#endif 1946| | 1947| 1.35k|#if OMPT_SUPPORT 1948| 1.35k| if (ompt_enabled.enabled) { ------------------ | Branch (1948:9): [True: 0, False: 1.35k] ------------------ 1949| 0| if (ompt_enabled.ompt_callback_parallel_begin) { ------------------ | Branch (1949:11): [True: 0, False: 0] ------------------ 1950| 0| int team_size = master_set_numthreads ------------------ | Branch (1950:25): [True: 0, False: 0] ------------------ 1951| 0| ? master_set_numthreads 1952| 0| : get__nproc_2(parent_team, master_tid); ------------------ | | 2298| 0| ((xteam)->t.t_threads[(xtid)]->th.th_current_task->td_icvs.nproc) ------------------ 1953| 0| int flags = OMPT_INVOKER(call_context) | ------------------ | | 24| 0| ((x == fork_context_gnu) ? ompt_parallel_invoker_program \ | | ------------------ | | | Branch (24:4): [True: 0, False: 0] | | ------------------ | | 25| 0| : ompt_parallel_invoker_runtime) ------------------ 1954| 0| ((microtask == (microtask_t)__kmp_teams_master) ------------------ | Branch (1954:22): [True: 0, False: 0] ------------------ 1955| 0| ? ompt_parallel_league 1956| 0| : ompt_parallel_team); 1957| 0| ompt_callbacks.ompt_callback(ompt_callback_parallel_begin)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 1958| 0| parent_task_data, ompt_frame, &ompt_parallel_data, team_size, flags, 1959| 0| return_address); 1960| 0| } 1961| 0| master_th->th.ompt_thread_info.state = ompt_state_overhead; 1962| 0| } 1963| 1.35k|#endif 1964| | 1965| 1.35k| master_th->th.th_ident = loc; 1966| | 1967| | // Parallel closely nested in teams construct: 1968| 1.35k| if (__kmp_is_fork_in_teams(master_th, microtask, level, teams_level, ap)) { ------------------ | Branch (1968:9): [True: 0, False: 1.35k] ------------------ 1969| 0| return __kmp_fork_in_teams(loc, gtid, parent_team, argc, master_th, root, 1970| 0| call_context, microtask, invoker, 1971| 0| master_set_numthreads, level, 1972| 0|#if OMPT_SUPPORT 1973| 0| ompt_parallel_data, return_address, 1974| 0|#endif 1975| 0| ap); 1976| 0| } // End parallel closely nested in teams construct 1977| | 1978| |#if KMP_DEBUG 1979| | if (__kmp_tasking_mode != tskm_immediate_exec) { 1980| | KMP_DEBUG_ASSERT(master_th->th.th_task_team == 1981| | parent_team->t.t_task_team[master_th->th.th_task_state]); 1982| | } 1983| |#endif 1984| | 1985| | // Need this to happen before we determine the number of threads, not while 1986| | // we are allocating the team 1987| | //__kmp_push_current_task_to_thread(master_th, parent_team, 0); 1988| | 1989| | // Determine the number of threads 1990| 1.35k| int enter_teams = 1991| 1.35k| __kmp_is_entering_teams(active_level, level, teams_level, ap); 1992| 1.35k| if ((!enter_teams && ------------------ | Branch (1992:10): [True: 1.35k, False: 0] ------------------ 1993| 1.35k| (parent_team->t.t_active_level >= ------------------ | Branch (1993:10): [True: 0, False: 1.35k] ------------------ 1994| 1.35k| master_th->th.th_current_task->td_icvs.max_active_levels)) || 1995| 1.35k| (__kmp_library == library_serial)) { ------------------ | Branch (1995:9): [True: 0, False: 1.35k] ------------------ 1996| 0| KC_TRACE(10, ("__kmp_fork_call: T#%d serializing team\n", gtid)); 1997| 0| nthreads = 1; 1998| 1.35k| } else { 1999| 1.35k| nthreads = master_set_numthreads ------------------ | Branch (1999:18): [True: 1.35k, False: 0] ------------------ 2000| 1.35k| ? master_set_numthreads 2001| | // TODO: get nproc directly from current task 2002| 1.35k| : get__nproc_2(parent_team, master_tid); ------------------ | | 2298| 1.35k| ((xteam)->t.t_threads[(xtid)]->th.th_current_task->td_icvs.nproc) ------------------ 2003| | // Check if we need to take forkjoin lock? (no need for serialized 2004| | // parallel out of teams construct). 2005| 1.35k| if (nthreads > 1) { ------------------ | Branch (2005:11): [True: 1.35k, False: 0] ------------------ 2006| | /* determine how many new threads we can use */ 2007| 1.35k| __kmp_acquire_bootstrap_lock(&__kmp_forkjoin_lock); 2008| | /* AC: If we execute teams from parallel region (on host), then teams 2009| | should be created but each can only have 1 thread if nesting is 2010| | disabled. If teams called from serial region, then teams and their 2011| | threads should be created regardless of the nesting setting. */ 2012| 1.35k| nthreads = __kmp_reserve_threads(root, parent_team, master_tid, 2013| 1.35k| nthreads, enter_teams); 2014| 1.35k| if (nthreads == 1) { ------------------ | Branch (2014:13): [True: 0, False: 1.35k] ------------------ 2015| | // Free lock for single thread execution here; for multi-thread 2016| | // execution it will be freed later after team of threads created 2017| | // and initialized 2018| 0| __kmp_release_bootstrap_lock(&__kmp_forkjoin_lock); 2019| 0| } 2020| 1.35k| } 2021| 1.35k| } 2022| 1.35k| KMP_DEBUG_ASSERT(nthreads > 0); 2023| | 2024| | // If we temporarily changed the set number of threads then restore it now 2025| 1.35k| master_th->th.th_set_nproc = 0; 2026| | 2027| 1.35k| if (nthreads == 1) { ------------------ | Branch (2027:9): [True: 0, False: 1.35k] ------------------ 2028| 0| return __kmp_serial_fork_call(loc, gtid, call_context, argc, microtask, 2029| 0| invoker, master_th, parent_team, 2030| 0|#if OMPT_SUPPORT 2031| 0| &ompt_parallel_data, &return_address, 2032| 0| &parent_task_data, 2033| 0|#endif 2034| 0| ap); 2035| 0| } // if (nthreads == 1) 2036| | 2037| | // GEH: only modify the executing flag in the case when not serialized 2038| | // serialized case is handled in kmpc_serialized_parallel 2039| 1.35k| KF_TRACE(10, ("__kmp_fork_call: parent_team_aclevel=%d, master_th=%p, " 2040| 1.35k| "curtask=%p, curtask_max_aclevel=%d\n", 2041| 1.35k| parent_team->t.t_active_level, master_th, 2042| 1.35k| master_th->th.th_current_task, 2043| 1.35k| master_th->th.th_current_task->td_icvs.max_active_levels)); 2044| | // TODO: GEH - cannot do this assertion because root thread not set up as 2045| | // executing 2046| | // KMP_ASSERT( master_th->th.th_current_task->td_flags.executing == 1 ); 2047| 1.35k| master_th->th.th_current_task->td_flags.executing = 0; 2048| | 2049| 1.35k| if (!master_th->th.th_teams_microtask || level > teams_level) { ------------------ | Branch (2049:9): [True: 1.35k, False: 0] | Branch (2049:46): [True: 0, False: 0] ------------------ 2050| | /* Increment our nested depth level */ 2051| 1.35k| KMP_ATOMIC_INC(&root->r.r_in_parallel); ------------------ | | 1259| 1.35k|#define KMP_ATOMIC_INC(p) KMP_ATOMIC_OP(fetch_add, p, 1, acq_rel) | | ------------------ | | | | 1246| 1.35k|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 2052| 1.35k| } 2053| | 2054| | // See if we need to make a copy of the ICVs. 2055| 1.35k| int nthreads_icv = master_th->th.th_current_task->td_icvs.nproc; 2056| 1.35k| if ((level + 1 < __kmp_nested_nth.used) && ------------------ | Branch (2056:9): [True: 0, False: 1.35k] ------------------ 2057| 0| (__kmp_nested_nth.nth[level + 1] != nthreads_icv)) { ------------------ | Branch (2057:9): [True: 0, False: 0] ------------------ 2058| 0| nthreads_icv = __kmp_nested_nth.nth[level + 1]; 2059| 1.35k| } else { 2060| 1.35k| nthreads_icv = 0; // don't update 2061| 1.35k| } 2062| | 2063| | // Figure out the proc_bind_policy for the new team. 2064| 1.35k| kmp_proc_bind_t proc_bind = master_th->th.th_set_proc_bind; 2065| | // proc_bind_default means don't update 2066| 1.35k| kmp_proc_bind_t proc_bind_icv = proc_bind_default; 2067| 1.35k| if (master_th->th.th_current_task->td_icvs.proc_bind == proc_bind_false) { ------------------ | Branch (2067:9): [True: 1.35k, False: 0] ------------------ 2068| 1.35k| proc_bind = proc_bind_false; 2069| 1.35k| } else { 2070| | // No proc_bind clause specified; use current proc-bind-var for this 2071| | // parallel region 2072| 0| if (proc_bind == proc_bind_default) { ------------------ | Branch (2072:11): [True: 0, False: 0] ------------------ 2073| 0| proc_bind = master_th->th.th_current_task->td_icvs.proc_bind; 2074| 0| } 2075| | // Have teams construct take proc_bind value from KMP_TEAMS_PROC_BIND 2076| 0| if (master_th->th.th_teams_microtask && ------------------ | Branch (2076:11): [True: 0, False: 0] ------------------ 2077| 0| microtask == (microtask_t)__kmp_teams_master) { ------------------ | Branch (2077:11): [True: 0, False: 0] ------------------ 2078| 0| proc_bind = __kmp_teams_proc_bind; 2079| 0| } 2080| | /* else: The proc_bind policy was specified explicitly on parallel clause. 2081| | This overrides proc-bind-var for this parallel region, but does not 2082| | change proc-bind-var. */ 2083| | // Figure the value of proc-bind-var for the child threads. 2084| 0| if ((level + 1 < __kmp_nested_proc_bind.used) && ------------------ | Branch (2084:11): [True: 0, False: 0] ------------------ 2085| 0| (__kmp_nested_proc_bind.bind_types[level + 1] != ------------------ | Branch (2085:11): [True: 0, False: 0] ------------------ 2086| 0| master_th->th.th_current_task->td_icvs.proc_bind)) { 2087| | // Do not modify the proc bind icv for the two teams construct forks 2088| | // They just let the proc bind icv pass through 2089| 0| if (!master_th->th.th_teams_microtask || ------------------ | Branch (2089:13): [True: 0, False: 0] ------------------ 2090| 0| !(microtask == (microtask_t)__kmp_teams_master || ap == NULL)) ------------------ | Branch (2090:15): [True: 0, False: 0] | Branch (2090:63): [True: 0, False: 0] ------------------ 2091| 0| proc_bind_icv = __kmp_nested_proc_bind.bind_types[level + 1]; 2092| 0| } 2093| 0| } 2094| | 2095| | // Reset for next parallel region 2096| 1.35k| master_th->th.th_set_proc_bind = proc_bind_default; 2097| | 2098| 1.35k| if ((nthreads_icv > 0) || (proc_bind_icv != proc_bind_default)) { ------------------ | Branch (2098:9): [True: 0, False: 1.35k] | Branch (2098:31): [True: 0, False: 1.35k] ------------------ 2099| 0| kmp_internal_control_t new_icvs; 2100| 0| copy_icvs(&new_icvs, &master_th->th.th_current_task->td_icvs); 2101| 0| new_icvs.next = NULL; 2102| 0| if (nthreads_icv > 0) { ------------------ | Branch (2102:11): [True: 0, False: 0] ------------------ 2103| 0| new_icvs.nproc = nthreads_icv; 2104| 0| } 2105| 0| if (proc_bind_icv != proc_bind_default) { ------------------ | Branch (2105:11): [True: 0, False: 0] ------------------ 2106| 0| new_icvs.proc_bind = proc_bind_icv; 2107| 0| } 2108| | 2109| | /* allocate a new parallel team */ 2110| 0| KF_TRACE(10, ("__kmp_fork_call: before __kmp_allocate_team\n")); 2111| 0| team = __kmp_allocate_team(root, nthreads, nthreads, 2112| 0|#if OMPT_SUPPORT 2113| 0| ompt_parallel_data, 2114| 0|#endif 2115| 0| proc_bind, &new_icvs, 2116| 0| argc USE_NESTED_HOT_ARG(master_th)); ------------------ | | 165| 0|#define USE_NESTED_HOT_ARG(x) , x ------------------ 2117| 0| if (__kmp_barrier_release_pattern[bs_forkjoin_barrier] == bp_dist_bar) ------------------ | Branch (2117:11): [True: 0, False: 0] ------------------ 2118| 0| copy_icvs((kmp_internal_control_t *)team->t.b->team_icvs, &new_icvs); 2119| 1.35k| } else { 2120| | /* allocate a new parallel team */ 2121| 1.35k| KF_TRACE(10, ("__kmp_fork_call: before __kmp_allocate_team\n")); 2122| 1.35k| team = __kmp_allocate_team(root, nthreads, nthreads, 2123| 1.35k|#if OMPT_SUPPORT 2124| 1.35k| ompt_parallel_data, 2125| 1.35k|#endif 2126| 1.35k| proc_bind, 2127| 1.35k| &master_th->th.th_current_task->td_icvs, 2128| 1.35k| argc USE_NESTED_HOT_ARG(master_th)); ------------------ | | 165| 1.35k|#define USE_NESTED_HOT_ARG(x) , x ------------------ 2129| 1.35k| if (__kmp_barrier_release_pattern[bs_forkjoin_barrier] == bp_dist_bar) ------------------ | Branch (2129:11): [True: 0, False: 1.35k] ------------------ 2130| 0| copy_icvs((kmp_internal_control_t *)team->t.b->team_icvs, 2131| 0| &master_th->th.th_current_task->td_icvs); 2132| 1.35k| } 2133| 1.35k| KF_TRACE( 2134| 1.35k| 10, ("__kmp_fork_call: after __kmp_allocate_team - team = %p\n", team)); 2135| | 2136| | /* setup the new team */ 2137| 1.35k| KMP_CHECK_UPDATE(team->t.t_master_tid, master_tid); ------------------ | | 2280| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 1.35k] | | ------------------ | | 2281| 1.35k| (a) = (b) ------------------ 2138| 1.35k| KMP_CHECK_UPDATE(team->t.t_master_this_cons, master_this_cons); ------------------ | | 2280| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 1.35k] | | ------------------ | | 2281| 1.35k| (a) = (b) ------------------ 2139| 1.35k| KMP_CHECK_UPDATE(team->t.t_ident, loc); ------------------ | | 2280| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 1.35k] | | ------------------ | | 2281| 1.35k| (a) = (b) ------------------ 2140| 1.35k| KMP_CHECK_UPDATE(team->t.t_parent, parent_team); ------------------ | | 2280| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 1.35k] | | ------------------ | | 2281| 1.35k| (a) = (b) ------------------ 2141| 1.35k| KMP_CHECK_UPDATE_SYNC(team->t.t_pkfn, microtask); ------------------ | | 2283| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2283:7): [True: 1.35k, False: 0] | | ------------------ | | 2284| 1.35k| TCW_SYNC_PTR((a), (b)) | | ------------------ | | | | 1159| 1.35k|#define TCW_SYNC_PTR(a, b) TCW_SYNC_8((a), (b)) | | | | ------------------ | | | | | | 1141| 1.35k|#define TCW_SYNC_8(a, b) (a) = (b) | | | | ------------------ | | ------------------ ------------------ 2142| 1.35k|#if OMPT_SUPPORT 2143| 1.35k| KMP_CHECK_UPDATE_SYNC(team->t.ompt_team_info.master_return_address, ------------------ | | 2283| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2283:7): [True: 0, False: 1.35k] | | ------------------ | | 2284| 1.35k| TCW_SYNC_PTR((a), (b)) | | ------------------ | | | | 1159| 0|#define TCW_SYNC_PTR(a, b) TCW_SYNC_8((a), (b)) | | | | ------------------ | | | | | | 1141| 0|#define TCW_SYNC_8(a, b) (a) = (b) | | | | ------------------ | | ------------------ ------------------ 2144| 1.35k| return_address); 2145| 1.35k|#endif 2146| 1.35k| KMP_CHECK_UPDATE(team->t.t_invoke, invoker); // TODO move to root, maybe ------------------ | | 2280| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 1, False: 1.35k] | | ------------------ | | 2281| 1.35k| (a) = (b) ------------------ 2147| | // TODO: parent_team->t.t_level == INT_MAX ??? 2148| 1.35k| if (!master_th->th.th_teams_microtask || level > teams_level) { ------------------ | Branch (2148:9): [True: 1.35k, False: 0] | Branch (2148:46): [True: 0, False: 0] ------------------ 2149| 1.35k| int new_level = parent_team->t.t_level + 1; 2150| 1.35k| KMP_CHECK_UPDATE(team->t.t_level, new_level); ------------------ | | 2280| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 1, False: 1.35k] | | ------------------ | | 2281| 1.35k| (a) = (b) ------------------ 2151| 1.35k| new_level = parent_team->t.t_active_level + 1; 2152| 1.35k| KMP_CHECK_UPDATE(team->t.t_active_level, new_level); ------------------ | | 2280| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 1, False: 1.35k] | | ------------------ | | 2281| 1.35k| (a) = (b) ------------------ 2153| 1.35k| } else { 2154| | // AC: Do not increase parallel level at start of the teams construct 2155| 0| int new_level = parent_team->t.t_level; 2156| 0| KMP_CHECK_UPDATE(team->t.t_level, new_level); ------------------ | | 2280| 0| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 0] | | ------------------ | | 2281| 0| (a) = (b) ------------------ 2157| 0| new_level = parent_team->t.t_active_level; 2158| 0| KMP_CHECK_UPDATE(team->t.t_active_level, new_level); ------------------ | | 2280| 0| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 0] | | ------------------ | | 2281| 0| (a) = (b) ------------------ 2159| 0| } 2160| 1.35k| kmp_r_sched_t new_sched = get__sched_2(parent_team, master_tid); ------------------ | | 2300| 1.35k| ((xteam)->t.t_threads[(xtid)]->th.th_current_task->td_icvs.sched) ------------------ 2161| | // set primary thread's schedule as new run-time schedule 2162| 1.35k| KMP_CHECK_UPDATE(team->t.t_sched.sched, new_sched.sched); ------------------ | | 2280| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 1.35k] | | ------------------ | | 2281| 1.35k| (a) = (b) ------------------ 2163| | 2164| 1.35k| KMP_CHECK_UPDATE(team->t.t_cancel_request, cancel_noreq); ------------------ | | 2280| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 1.35k] | | ------------------ | | 2281| 1.35k| (a) = (b) ------------------ 2165| 1.35k| KMP_CHECK_UPDATE(team->t.t_def_allocator, master_th->th.th_def_allocator); ------------------ | | 2280| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 1, False: 1.35k] | | ------------------ | | 2281| 1.35k| (a) = (b) ------------------ 2166| | 2167| | // Update the floating point rounding in the team if required. 2168| 1.35k| propagateFPControl(team); 2169| 1.35k|#if OMPD_SUPPORT 2170| 1.35k| if (ompd_state & OMPD_ENABLE_BP) ------------------ | | 35| 1.35k|#define OMPD_ENABLE_BP 0x1 ------------------ | Branch (2170:9): [True: 0, False: 1.35k] ------------------ 2171| 0| ompd_bp_parallel_begin(); 2172| 1.35k|#endif 2173| | 2174| 1.35k| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (2174:9): [True: 1.35k, False: 0] ------------------ 2175| | // Set primary thread's task team to team's task team. Unless this is hot 2176| | // team, it should be NULL. 2177| 1.35k| KMP_DEBUG_ASSERT(master_th->th.th_task_team == 2178| 1.35k| parent_team->t.t_task_team[master_th->th.th_task_state]); 2179| 1.35k| KA_TRACE(20, ("__kmp_fork_call: Primary T#%d pushing task_team %p / team " 2180| 1.35k| "%p, new task_team %p / team %p\n", 2181| 1.35k| __kmp_gtid_from_thread(master_th), 2182| 1.35k| master_th->th.th_task_team, parent_team, 2183| 1.35k| team->t.t_task_team[master_th->th.th_task_state], team)); 2184| | 2185| 1.35k| if (active_level || master_th->th.th_task_team) { ------------------ | Branch (2185:11): [True: 0, False: 1.35k] | Branch (2185:27): [True: 0, False: 1.35k] ------------------ 2186| | // Take a memo of primary thread's task_state 2187| 0| KMP_DEBUG_ASSERT(master_th->th.th_task_state_memo_stack); 2188| 0| if (master_th->th.th_task_state_top >= ------------------ | Branch (2188:13): [True: 0, False: 0] ------------------ 2189| 0| master_th->th.th_task_state_stack_sz) { // increase size 2190| 0| kmp_uint32 new_size = 2 * master_th->th.th_task_state_stack_sz; 2191| 0| kmp_uint8 *old_stack, *new_stack; 2192| 0| kmp_uint32 i; 2193| 0| new_stack = (kmp_uint8 *)__kmp_allocate(new_size); ------------------ | | 3617| 0|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 2194| 0| for (i = 0; i < master_th->th.th_task_state_stack_sz; ++i) { ------------------ | Branch (2194:23): [True: 0, False: 0] ------------------ 2195| 0| new_stack[i] = master_th->th.th_task_state_memo_stack[i]; 2196| 0| } 2197| 0| for (i = master_th->th.th_task_state_stack_sz; i < new_size; ------------------ | Branch (2197:58): [True: 0, False: 0] ------------------ 2198| 0| ++i) { // zero-init rest of stack 2199| 0| new_stack[i] = 0; 2200| 0| } 2201| 0| old_stack = master_th->th.th_task_state_memo_stack; 2202| 0| master_th->th.th_task_state_memo_stack = new_stack; 2203| 0| master_th->th.th_task_state_stack_sz = new_size; 2204| 0| __kmp_free(old_stack); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 2205| 0| } 2206| | // Store primary thread's task_state on stack 2207| 0| master_th->th 2208| 0| .th_task_state_memo_stack[master_th->th.th_task_state_top] = 2209| 0| master_th->th.th_task_state; 2210| 0| master_th->th.th_task_state_top++; 2211| 0|#if KMP_NESTED_HOT_TEAMS 2212| 0| if (master_th->th.th_hot_teams && ------------------ | Branch (2212:13): [True: 0, False: 0] ------------------ 2213| 0| active_level < __kmp_hot_teams_max_level && ------------------ | Branch (2213:13): [True: 0, False: 0] ------------------ 2214| 0| team == master_th->th.th_hot_teams[active_level].hot_team) { ------------------ | Branch (2214:13): [True: 0, False: 0] ------------------ 2215| | // Restore primary thread's nested state if nested hot team 2216| 0| master_th->th.th_task_state = 2217| 0| master_th->th 2218| 0| .th_task_state_memo_stack[master_th->th.th_task_state_top]; 2219| 0| } else { 2220| 0|#endif 2221| 0| master_th->th.th_task_state = 0; 2222| 0|#if KMP_NESTED_HOT_TEAMS 2223| 0| } 2224| 0|#endif 2225| 0| } 2226| |#if !KMP_NESTED_HOT_TEAMS 2227| | KMP_DEBUG_ASSERT((master_th->th.th_task_team == NULL) || 2228| | (team == root->r.r_hot_team)); 2229| |#endif 2230| 1.35k| } 2231| | 2232| 1.35k| KA_TRACE( 2233| 1.35k| 20, 2234| 1.35k| ("__kmp_fork_call: T#%d(%d:%d)->(%d:0) created a team of %d threads\n", 2235| 1.35k| gtid, parent_team->t.t_id, team->t.t_master_tid, team->t.t_id, 2236| 1.35k| team->t.t_nproc)); 2237| 1.35k| KMP_DEBUG_ASSERT(team != root->r.r_hot_team || 2238| 1.35k| (team->t.t_master_tid == 0 && 2239| 1.35k| (team->t.t_parent == root->r.r_root_team || 2240| 1.35k| team->t.t_parent->t.t_serialized))); 2241| 1.35k| KMP_MB(); 2242| | 2243| | /* now, setup the arguments */ 2244| 1.35k| argv = (void **)team->t.t_argv; 2245| 1.35k| if (ap) { ------------------ | Branch (2245:9): [True: 1.35k, False: 0] ------------------ 2246| 2.97k| for (i = argc - 1; i >= 0; --i) { ------------------ | Branch (2246:26): [True: 1.61k, False: 1.35k] ------------------ 2247| 1.61k| void *new_argv = va_arg(kmp_va_deref(ap), void *); 2248| 1.61k| KMP_CHECK_UPDATE(*argv, new_argv); ------------------ | | 2280| 1.61k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 333, False: 1.28k] | | ------------------ | | 2281| 1.61k| (a) = (b) ------------------ 2249| 1.61k| argv++; 2250| 1.61k| } 2251| 1.35k| } else { 2252| 0| for (i = 0; i < argc; ++i) { ------------------ | Branch (2252:19): [True: 0, False: 0] ------------------ 2253| | // Get args from parent team for teams construct 2254| 0| KMP_CHECK_UPDATE(argv[i], team->t.t_parent->t.t_argv[i]); ------------------ | | 2280| 0| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 0] | | ------------------ | | 2281| 0| (a) = (b) ------------------ 2255| 0| } 2256| 0| } 2257| | 2258| | /* now actually fork the threads */ 2259| 1.35k| KMP_CHECK_UPDATE(team->t.t_master_active, master_active); ------------------ | | 2280| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 1.35k] | | ------------------ | | 2281| 1.35k| (a) = (b) ------------------ 2260| 1.35k| if (!root->r.r_active) // Only do assignment if it prevents cache ping-pong ------------------ | Branch (2260:9): [True: 1.35k, False: 0] ------------------ 2261| 1.35k| root->r.r_active = TRUE; ------------------ | | 1280| 1.35k|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1.35k|#define FALSE 0 | | ------------------ ------------------ 2262| | 2263| 1.35k| __kmp_fork_team_threads(root, team, master_th, gtid, !ap); 2264| 1.35k| __kmp_setup_icv_copy(team, nthreads, 2265| 1.35k| &master_th->th.th_current_task->td_icvs, loc); 2266| | 2267| 1.35k|#if OMPT_SUPPORT 2268| 1.35k| master_th->th.ompt_thread_info.state = ompt_state_work_parallel; 2269| 1.35k|#endif 2270| | 2271| 1.35k| __kmp_release_bootstrap_lock(&__kmp_forkjoin_lock); 2272| | 2273| |#if USE_ITT_BUILD 2274| | if (team->t.t_active_level == 1 // only report frames at level 1 2275| | && !master_th->th.th_teams_microtask) { // not in teams construct 2276| |#if USE_ITT_NOTIFY 2277| | if ((__itt_frame_submit_v3_ptr || KMP_ITT_DEBUG) && 2278| | (__kmp_forkjoin_frames_mode == 3 || 2279| | __kmp_forkjoin_frames_mode == 1)) { 2280| | kmp_uint64 tmp_time = 0; 2281| | if (__itt_get_timestamp_ptr) 2282| | tmp_time = __itt_get_timestamp(); 2283| | // Internal fork - report frame begin 2284| | master_th->th.th_frame_time = tmp_time; 2285| | if (__kmp_forkjoin_frames_mode == 3) 2286| | team->t.t_region_time = tmp_time; 2287| | } else 2288| |// only one notification scheme (either "submit" or "forking/joined", not both) 2289| |#endif /* USE_ITT_NOTIFY */ 2290| | if ((__itt_frame_begin_v3_ptr || KMP_ITT_DEBUG) && 2291| | __kmp_forkjoin_frames && !__kmp_forkjoin_frames_mode) { 2292| | // Mark start of "parallel" region for Intel(R) VTune(TM) analyzer. 2293| | __kmp_itt_region_forking(gtid, team->t.t_nproc, 0); 2294| | } 2295| | } 2296| |#endif /* USE_ITT_BUILD */ 2297| | 2298| | /* now go on and do the work */ 2299| 1.35k| KMP_DEBUG_ASSERT(team == __kmp_threads[gtid]->th.th_team); 2300| 1.35k| KMP_MB(); 2301| 1.35k| KF_TRACE(10, 2302| 1.35k| ("__kmp_internal_fork : root=%p, team=%p, master_th=%p, gtid=%d\n", 2303| 1.35k| root, team, master_th, gtid)); 2304| | 2305| |#if USE_ITT_BUILD 2306| | if (__itt_stack_caller_create_ptr) { 2307| | // create new stack stitching id before entering fork barrier 2308| | if (!enter_teams) { 2309| | KMP_DEBUG_ASSERT(team->t.t_stack_id == NULL); 2310| | team->t.t_stack_id = __kmp_itt_stack_caller_create(); 2311| | } else if (parent_team->t.t_serialized) { 2312| | // keep stack stitching id in the serialized parent_team; 2313| | // current team will be used for parallel inside the teams; 2314| | // if parent_team is active, then it already keeps stack stitching id 2315| | // for the league of teams 2316| | KMP_DEBUG_ASSERT(parent_team->t.t_stack_id == NULL); 2317| | parent_team->t.t_stack_id = __kmp_itt_stack_caller_create(); 2318| | } 2319| | } 2320| |#endif /* USE_ITT_BUILD */ 2321| | 2322| | // AC: skip __kmp_internal_fork at teams construct, let only primary 2323| | // threads execute 2324| 1.35k| if (ap) { ------------------ | Branch (2324:9): [True: 1.35k, False: 0] ------------------ 2325| 1.35k| __kmp_internal_fork(loc, gtid, team); 2326| 1.35k| KF_TRACE(10, ("__kmp_internal_fork : after : root=%p, team=%p, " 2327| 1.35k| "master_th=%p, gtid=%d\n", 2328| 1.35k| root, team, master_th, gtid)); 2329| 1.35k| } 2330| | 2331| 1.35k| if (call_context == fork_context_gnu) { ------------------ | Branch (2331:9): [True: 0, False: 1.35k] ------------------ 2332| 0| KA_TRACE(20, ("__kmp_fork_call: parallel exit T#%d\n", gtid)); 2333| 0| return TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 2334| 0| } 2335| | 2336| | /* Invoke microtask for PRIMARY thread */ 2337| 1.35k| KA_TRACE(20, ("__kmp_fork_call: T#%d(%d:0) invoke microtask = %p\n", gtid, 2338| 1.35k| team->t.t_id, team->t.t_pkfn)); 2339| 1.35k| } // END of timer KMP_fork_call block 2340| | 2341| |#if KMP_STATS_ENABLED 2342| | // If beginning a teams construct, then change thread state 2343| | stats_state_e previous_state = KMP_GET_THREAD_STATE(); 2344| | if (!ap) { 2345| | KMP_SET_THREAD_STATE(stats_state_e::TEAMS_REGION); 2346| | } 2347| |#endif 2348| | 2349| 1.35k| if (!team->t.t_invoke(gtid)) { ------------------ | Branch (2349:7): [True: 0, False: 1.35k] ------------------ 2350| 0| KMP_ASSERT2(0, "cannot invoke microtask for PRIMARY thread"); ------------------ | | 53| 0|#define KMP_ASSERT2(cond, msg) KMP_ASSERT(cond) | | ------------------ | | | | 50| 0| if (!(cond)) { \ | | | | ------------------ | | | | | Branch (50:7): [True: 0, Folded] | | | | ------------------ | | | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | | | 52| 0| } | | ------------------ ------------------ 2351| 0| } 2352| | 2353| |#if KMP_STATS_ENABLED 2354| | // If was beginning of a teams construct, then reset thread state 2355| | if (!ap) { 2356| | KMP_SET_THREAD_STATE(previous_state); 2357| | } 2358| |#endif 2359| | 2360| 1.35k| KA_TRACE(20, ("__kmp_fork_call: T#%d(%d:0) done microtask = %p\n", gtid, 2361| 1.35k| team->t.t_id, team->t.t_pkfn)); 2362| 1.35k| KMP_MB(); /* Flush all pending memory write invalidates. */ 2363| | 2364| 1.35k| KA_TRACE(20, ("__kmp_fork_call: parallel exit T#%d\n", gtid)); 2365| 1.35k|#if OMPT_SUPPORT 2366| 1.35k| if (ompt_enabled.enabled) { ------------------ | Branch (2366:7): [True: 0, False: 1.35k] ------------------ 2367| 0| master_th->th.ompt_thread_info.state = ompt_state_overhead; 2368| 0| } 2369| 1.35k|#endif 2370| | 2371| 1.35k| return TRUE; ------------------ | | 1280| 1.35k|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1.35k|#define FALSE 0 | | ------------------ ------------------ 2372| 1.35k|} __kmp_join_call: 2403| 1.35k| int exit_teams) { 2404| 1.35k| KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(KMP_join_call); ------------------ | | 1008| 1.35k|#define KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(n) ((void)0) ------------------ 2405| 1.35k| kmp_team_t *team; 2406| 1.35k| kmp_team_t *parent_team; 2407| 1.35k| kmp_info_t *master_th; 2408| 1.35k| kmp_root_t *root; 2409| 1.35k| int master_active; 2410| | 2411| 1.35k| KA_TRACE(20, ("__kmp_join_call: enter T#%d\n", gtid)); 2412| | 2413| | /* setup current data */ 2414| 1.35k| master_th = __kmp_threads[gtid]; 2415| 1.35k| root = master_th->th.th_root; 2416| 1.35k| team = master_th->th.th_team; 2417| 1.35k| parent_team = team->t.t_parent; 2418| | 2419| 1.35k| master_th->th.th_ident = loc; 2420| | 2421| 1.35k|#if OMPT_SUPPORT 2422| 1.35k| void *team_microtask = (void *)team->t.t_pkfn; 2423| | // For GOMP interface with serialized parallel, need the 2424| | // __kmpc_end_serialized_parallel to call hooks for OMPT end-implicit-task 2425| | // and end-parallel events. 2426| 1.35k| if (ompt_enabled.enabled && ------------------ | Branch (2426:7): [True: 0, False: 1.35k] ------------------ 2427| 0| !(team->t.t_serialized && fork_context == fork_context_gnu)) { ------------------ | Branch (2427:9): [True: 0, False: 0] | Branch (2427:33): [True: 0, False: 0] ------------------ 2428| 0| master_th->th.ompt_thread_info.state = ompt_state_overhead; 2429| 0| } 2430| 1.35k|#endif 2431| | 2432| |#if KMP_DEBUG 2433| | if (__kmp_tasking_mode != tskm_immediate_exec && !exit_teams) { 2434| | KA_TRACE(20, ("__kmp_join_call: T#%d, old team = %p old task_team = %p, " 2435| | "th_task_team = %p\n", 2436| | __kmp_gtid_from_thread(master_th), team, 2437| | team->t.t_task_team[master_th->th.th_task_state], 2438| | master_th->th.th_task_team)); 2439| | KMP_DEBUG_ASSERT(master_th->th.th_task_team == 2440| | team->t.t_task_team[master_th->th.th_task_state]); 2441| | } 2442| |#endif 2443| | 2444| 1.35k| if (team->t.t_serialized) { ------------------ | Branch (2444:7): [True: 0, False: 1.35k] ------------------ 2445| 0| if (master_th->th.th_teams_microtask) { ------------------ | Branch (2445:9): [True: 0, False: 0] ------------------ 2446| | // We are in teams construct 2447| 0| int level = team->t.t_level; 2448| 0| int tlevel = master_th->th.th_teams_level; 2449| 0| if (level == tlevel) { ------------------ | Branch (2449:11): [True: 0, False: 0] ------------------ 2450| | // AC: we haven't incremented it earlier at start of teams construct, 2451| | // so do it here - at the end of teams construct 2452| 0| team->t.t_level++; 2453| 0| } else if (level == tlevel + 1) { ------------------ | Branch (2453:18): [True: 0, False: 0] ------------------ 2454| | // AC: we are exiting parallel inside teams, need to increment 2455| | // serialization in order to restore it in the next call to 2456| | // __kmpc_end_serialized_parallel 2457| 0| team->t.t_serialized++; 2458| 0| } 2459| 0| } 2460| 0| __kmpc_end_serialized_parallel(loc, gtid); 2461| | 2462| 0|#if OMPT_SUPPORT 2463| 0| if (ompt_enabled.enabled) { ------------------ | Branch (2463:9): [True: 0, False: 0] ------------------ 2464| 0| if (fork_context == fork_context_gnu) { ------------------ | Branch (2464:11): [True: 0, False: 0] ------------------ 2465| 0| __ompt_lw_taskteam_unlink(master_th); 2466| 0| } 2467| 0| __kmp_join_restore_state(master_th, parent_team); 2468| 0| } 2469| 0|#endif 2470| | 2471| 0| return; 2472| 0| } 2473| | 2474| 1.35k| master_active = team->t.t_master_active; 2475| | 2476| 1.35k| if (!exit_teams) { ------------------ | Branch (2476:7): [True: 1.35k, False: 0] ------------------ 2477| | // AC: No barrier for internal teams at exit from teams construct. 2478| | // But there is barrier for external team (league). 2479| 1.35k| __kmp_internal_join(loc, gtid, team); 2480| |#if USE_ITT_BUILD 2481| | if (__itt_stack_caller_create_ptr) { 2482| | KMP_DEBUG_ASSERT(team->t.t_stack_id != NULL); 2483| | // destroy the stack stitching id after join barrier 2484| | __kmp_itt_stack_caller_destroy((__itt_caller)team->t.t_stack_id); 2485| | team->t.t_stack_id = NULL; 2486| | } 2487| |#endif 2488| 1.35k| } else { 2489| 0| master_th->th.th_task_state = 2490| 0| 0; // AC: no tasking in teams (out of any parallel) 2491| |#if USE_ITT_BUILD 2492| | if (__itt_stack_caller_create_ptr && parent_team->t.t_serialized) { 2493| | KMP_DEBUG_ASSERT(parent_team->t.t_stack_id != NULL); 2494| | // destroy the stack stitching id on exit from the teams construct 2495| | // if parent_team is active, then the id will be destroyed later on 2496| | // by master of the league of teams 2497| | __kmp_itt_stack_caller_destroy((__itt_caller)parent_team->t.t_stack_id); 2498| | parent_team->t.t_stack_id = NULL; 2499| | } 2500| |#endif 2501| 0| } 2502| | 2503| 1.35k| KMP_MB(); 2504| | 2505| 1.35k|#if OMPT_SUPPORT 2506| 1.35k| ompt_data_t *parallel_data = &(team->t.ompt_team_info.parallel_data); 2507| 1.35k| void *codeptr = team->t.ompt_team_info.master_return_address; 2508| 1.35k|#endif 2509| | 2510| |#if USE_ITT_BUILD 2511| | // Mark end of "parallel" region for Intel(R) VTune(TM) analyzer. 2512| | if (team->t.t_active_level == 1 && 2513| | (!master_th->th.th_teams_microtask || /* not in teams construct */ 2514| | master_th->th.th_teams_size.nteams == 1)) { 2515| | master_th->th.th_ident = loc; 2516| | // only one notification scheme (either "submit" or "forking/joined", not 2517| | // both) 2518| | if ((__itt_frame_submit_v3_ptr || KMP_ITT_DEBUG) && 2519| | __kmp_forkjoin_frames_mode == 3) 2520| | __kmp_itt_frame_submit(gtid, team->t.t_region_time, 2521| | master_th->th.th_frame_time, 0, loc, 2522| | master_th->th.th_team_nproc, 1); 2523| | else if ((__itt_frame_end_v3_ptr || KMP_ITT_DEBUG) && 2524| | !__kmp_forkjoin_frames_mode && __kmp_forkjoin_frames) 2525| | __kmp_itt_region_joined(gtid); 2526| | } // active_level == 1 2527| |#endif /* USE_ITT_BUILD */ 2528| | 2529| 1.35k|#if KMP_AFFINITY_SUPPORTED 2530| 1.35k| if (!exit_teams) { ------------------ | Branch (2530:7): [True: 1.35k, False: 0] ------------------ 2531| | // Restore master thread's partition. 2532| 1.35k| master_th->th.th_first_place = team->t.t_first_place; 2533| 1.35k| master_th->th.th_last_place = team->t.t_last_place; 2534| 1.35k| } 2535| 1.35k|#endif // KMP_AFFINITY_SUPPORTED 2536| | 2537| 1.35k| if (master_th->th.th_teams_microtask && !exit_teams && ------------------ | Branch (2537:7): [True: 0, False: 1.35k] | Branch (2537:43): [True: 0, False: 0] ------------------ 2538| 0| team->t.t_pkfn != (microtask_t)__kmp_teams_master && ------------------ | Branch (2538:7): [True: 0, False: 0] ------------------ 2539| 0| team->t.t_level == master_th->th.th_teams_level + 1) { ------------------ | Branch (2539:7): [True: 0, False: 0] ------------------ 2540| |// AC: We need to leave the team structure intact at the end of parallel 2541| |// inside the teams construct, so that at the next parallel same (hot) team 2542| |// works, only adjust nesting levels 2543| 0|#if OMPT_SUPPORT 2544| 0| ompt_data_t ompt_parallel_data = ompt_data_none; ------------------ | | 1408| 0|#define ompt_data_none {0} ------------------ 2545| 0| if (ompt_enabled.enabled) { ------------------ | Branch (2545:9): [True: 0, False: 0] ------------------ 2546| 0| ompt_task_info_t *task_info = __ompt_get_task_info_object(0); 2547| 0| if (ompt_enabled.ompt_callback_implicit_task) { ------------------ | Branch (2547:11): [True: 0, False: 0] ------------------ 2548| 0| int ompt_team_size = team->t.t_nproc; 2549| 0| ompt_callbacks.ompt_callback(ompt_callback_implicit_task)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 2550| 0| ompt_scope_end, NULL, &(task_info->task_data), ompt_team_size, 2551| 0| OMPT_CUR_TASK_INFO(master_th)->thread_num, ompt_task_implicit); ------------------ | | 68| 0|#define OMPT_CUR_TASK_INFO(thr) (&((thr)->th.th_current_task->ompt_task_info)) ------------------ 2552| 0| } 2553| 0| task_info->frame.exit_frame = ompt_data_none; ------------------ | | 1408| 0|#define ompt_data_none {0} ------------------ 2554| 0| task_info->task_data = ompt_data_none; ------------------ | | 1408| 0|#define ompt_data_none {0} ------------------ 2555| 0| ompt_parallel_data = *OMPT_CUR_TEAM_DATA(master_th); ------------------ | | 73| 0| (&((thr)->th.th_team->t.ompt_team_info.parallel_data)) ------------------ 2556| 0| __ompt_lw_taskteam_unlink(master_th); 2557| 0| } 2558| 0|#endif 2559| | /* Decrement our nested depth level */ 2560| 0| team->t.t_level--; 2561| 0| team->t.t_active_level--; 2562| 0| KMP_ATOMIC_DEC(&root->r.r_in_parallel); ------------------ | | 1260| 0|#define KMP_ATOMIC_DEC(p) KMP_ATOMIC_OP(fetch_sub, p, 1, acq_rel) | | ------------------ | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 2563| | 2564| | // Restore number of threads in the team if needed. This code relies on 2565| | // the proper adjustment of th_teams_size.nth after the fork in 2566| | // __kmp_teams_master on each teams primary thread in the case that 2567| | // __kmp_reserve_threads reduced it. 2568| 0| if (master_th->th.th_team_nproc < master_th->th.th_teams_size.nth) { ------------------ | Branch (2568:9): [True: 0, False: 0] ------------------ 2569| 0| int old_num = master_th->th.th_team_nproc; 2570| 0| int new_num = master_th->th.th_teams_size.nth; 2571| 0| kmp_info_t **other_threads = team->t.t_threads; 2572| 0| team->t.t_nproc = new_num; 2573| 0| for (int i = 0; i < old_num; ++i) { ------------------ | Branch (2573:23): [True: 0, False: 0] ------------------ 2574| 0| other_threads[i]->th.th_team_nproc = new_num; 2575| 0| } 2576| | // Adjust states of non-used threads of the team 2577| 0| for (int i = old_num; i < new_num; ++i) { ------------------ | Branch (2577:29): [True: 0, False: 0] ------------------ 2578| | // Re-initialize thread's barrier data. 2579| 0| KMP_DEBUG_ASSERT(other_threads[i]); 2580| 0| kmp_balign_t *balign = other_threads[i]->th.th_bar; 2581| 0| for (int b = 0; b < bs_last_barrier; ++b) { ------------------ | Branch (2581:25): [True: 0, False: 0] ------------------ 2582| 0| balign[b].bb.b_arrived = team->t.t_bar[b].b_arrived; 2583| 0| KMP_DEBUG_ASSERT(balign[b].bb.wait_flag != KMP_BARRIER_PARENT_FLAG); 2584| |#if USE_DEBUGGER 2585| | balign[b].bb.b_worker_arrived = team->t.t_bar[b].b_team_arrived; 2586| |#endif 2587| 0| } 2588| 0| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (2588:13): [True: 0, False: 0] ------------------ 2589| | // Synchronize thread's task state 2590| 0| other_threads[i]->th.th_task_state = master_th->th.th_task_state; 2591| 0| } 2592| 0| } 2593| 0| } 2594| | 2595| 0|#if OMPT_SUPPORT 2596| 0| if (ompt_enabled.enabled) { ------------------ | Branch (2596:9): [True: 0, False: 0] ------------------ 2597| 0| __kmp_join_ompt(gtid, master_th, parent_team, &ompt_parallel_data, 2598| 0| OMPT_INVOKER(fork_context) | ompt_parallel_team, codeptr); ------------------ | | 24| 0| ((x == fork_context_gnu) ? ompt_parallel_invoker_program \ | | ------------------ | | | Branch (24:4): [True: 0, False: 0] | | ------------------ | | 25| 0| : ompt_parallel_invoker_runtime) ------------------ 2599| 0| } 2600| 0|#endif 2601| | 2602| 0| return; 2603| 0| } 2604| | 2605| | /* do cleanup and restore the parent team */ 2606| 1.35k| master_th->th.th_info.ds.ds_tid = team->t.t_master_tid; 2607| 1.35k| master_th->th.th_local.this_construct = team->t.t_master_this_cons; 2608| | 2609| 1.35k| master_th->th.th_dispatch = &parent_team->t.t_dispatch[team->t.t_master_tid]; 2610| | 2611| | /* jc: The following lock has instructions with REL and ACQ semantics, 2612| | separating the parallel user code called in this parallel region 2613| | from the serial user code called after this function returns. */ 2614| 1.35k| __kmp_acquire_bootstrap_lock(&__kmp_forkjoin_lock); 2615| | 2616| 1.35k| if (!master_th->th.th_teams_microtask || ------------------ | Branch (2616:7): [True: 1.35k, False: 0] ------------------ 2617| 1.35k| team->t.t_level > master_th->th.th_teams_level) { ------------------ | Branch (2617:7): [True: 0, False: 0] ------------------ 2618| | /* Decrement our nested depth level */ 2619| 1.35k| KMP_ATOMIC_DEC(&root->r.r_in_parallel); ------------------ | | 1260| 1.35k|#define KMP_ATOMIC_DEC(p) KMP_ATOMIC_OP(fetch_sub, p, 1, acq_rel) | | ------------------ | | | | 1246| 1.35k|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 2620| 1.35k| } 2621| 1.35k| KMP_DEBUG_ASSERT(root->r.r_in_parallel >= 0); 2622| | 2623| 1.35k|#if OMPT_SUPPORT 2624| 1.35k| if (ompt_enabled.enabled) { ------------------ | Branch (2624:7): [True: 0, False: 1.35k] ------------------ 2625| 0| ompt_task_info_t *task_info = __ompt_get_task_info_object(0); 2626| 0| if (ompt_enabled.ompt_callback_implicit_task) { ------------------ | Branch (2626:9): [True: 0, False: 0] ------------------ 2627| 0| int flags = (team_microtask == (void *)__kmp_teams_master) ------------------ | Branch (2627:19): [True: 0, False: 0] ------------------ 2628| 0| ? ompt_task_initial 2629| 0| : ompt_task_implicit; 2630| 0| int ompt_team_size = (flags == ompt_task_initial) ? 0 : team->t.t_nproc; ------------------ | Branch (2630:28): [True: 0, False: 0] ------------------ 2631| 0| ompt_callbacks.ompt_callback(ompt_callback_implicit_task)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 2632| 0| ompt_scope_end, NULL, &(task_info->task_data), ompt_team_size, 2633| 0| OMPT_CUR_TASK_INFO(master_th)->thread_num, flags); ------------------ | | 68| 0|#define OMPT_CUR_TASK_INFO(thr) (&((thr)->th.th_current_task->ompt_task_info)) ------------------ 2634| 0| } 2635| 0| task_info->frame.exit_frame = ompt_data_none; ------------------ | | 1408| 0|#define ompt_data_none {0} ------------------ 2636| 0| task_info->task_data = ompt_data_none; ------------------ | | 1408| 0|#define ompt_data_none {0} ------------------ 2637| 0| } 2638| 1.35k|#endif 2639| | 2640| 1.35k| KF_TRACE(10, ("__kmp_join_call1: T#%d, this_thread=%p team=%p\n", 0, 2641| 1.35k| master_th, team)); 2642| 1.35k| __kmp_pop_current_task_from_thread(master_th); 2643| | 2644| 1.35k| master_th->th.th_def_allocator = team->t.t_def_allocator; 2645| | 2646| 1.35k|#if OMPD_SUPPORT 2647| 1.35k| if (ompd_state & OMPD_ENABLE_BP) ------------------ | | 35| 1.35k|#define OMPD_ENABLE_BP 0x1 ------------------ | Branch (2647:7): [True: 0, False: 1.35k] ------------------ 2648| 0| ompd_bp_parallel_end(); 2649| 1.35k|#endif 2650| 1.35k| updateHWFPControl(team); 2651| | 2652| 1.35k| if (root->r.r_active != master_active) ------------------ | Branch (2652:7): [True: 1.35k, False: 0] ------------------ 2653| 1.35k| root->r.r_active = master_active; 2654| | 2655| 1.35k| __kmp_free_team(root, team USE_NESTED_HOT_ARG( ------------------ | | 165| 1.35k|#define USE_NESTED_HOT_ARG(x) , x ------------------ 2656| 1.35k| master_th)); // this will free worker threads 2657| | 2658| | /* this race was fun to find. make sure the following is in the critical 2659| | region otherwise assertions may fail occasionally since the old team may be 2660| | reallocated and the hierarchy appears inconsistent. it is actually safe to 2661| | run and won't cause any bugs, but will cause those assertion failures. it's 2662| | only one deref&assign so might as well put this in the critical region */ 2663| 1.35k| master_th->th.th_team = parent_team; 2664| 1.35k| master_th->th.th_team_nproc = parent_team->t.t_nproc; 2665| 1.35k| master_th->th.th_team_master = parent_team->t.t_threads[0]; 2666| 1.35k| master_th->th.th_team_serialized = parent_team->t.t_serialized; 2667| | 2668| | /* restore serialized team, if need be */ 2669| 1.35k| if (parent_team->t.t_serialized && ------------------ | Branch (2669:7): [True: 1.35k, False: 0] ------------------ 2670| 1.35k| parent_team != master_th->th.th_serial_team && ------------------ | Branch (2670:7): [True: 1.35k, False: 0] ------------------ 2671| 1.35k| parent_team != root->r.r_root_team) { ------------------ | Branch (2671:7): [True: 0, False: 1.35k] ------------------ 2672| 0| __kmp_free_team(root, 2673| 0| master_th->th.th_serial_team USE_NESTED_HOT_ARG(NULL)); ------------------ | | 165| 0|#define USE_NESTED_HOT_ARG(x) , x ------------------ 2674| 0| master_th->th.th_serial_team = parent_team; 2675| 0| } 2676| | 2677| 1.35k| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (2677:7): [True: 1.35k, False: 0] ------------------ 2678| 1.35k| if (master_th->th.th_task_state_top > ------------------ | Branch (2678:9): [True: 0, False: 1.35k] ------------------ 2679| 1.35k| 0) { // Restore task state from memo stack 2680| 0| KMP_DEBUG_ASSERT(master_th->th.th_task_state_memo_stack); 2681| | // Remember primary thread's state if we re-use this nested hot team 2682| 0| master_th->th.th_task_state_memo_stack[master_th->th.th_task_state_top] = 2683| 0| master_th->th.th_task_state; 2684| 0| --master_th->th.th_task_state_top; // pop 2685| | // Now restore state at this level 2686| 0| master_th->th.th_task_state = 2687| 0| master_th->th 2688| 0| .th_task_state_memo_stack[master_th->th.th_task_state_top]; 2689| 1.35k| } else if (team != root->r.r_hot_team) { ------------------ | Branch (2689:16): [True: 0, False: 1.35k] ------------------ 2690| | // Reset the task state of primary thread if we are not hot team because 2691| | // in this case all the worker threads will be free, and their task state 2692| | // will be reset. If not reset the primary's, the task state will be 2693| | // inconsistent. 2694| 0| master_th->th.th_task_state = 0; 2695| 0| } 2696| | // Copy the task team from the parent team to the primary thread 2697| 1.35k| master_th->th.th_task_team = 2698| 1.35k| parent_team->t.t_task_team[master_th->th.th_task_state]; 2699| 1.35k| KA_TRACE(20, 2700| 1.35k| ("__kmp_join_call: Primary T#%d restoring task_team %p, team %p\n", 2701| 1.35k| __kmp_gtid_from_thread(master_th), master_th->th.th_task_team, 2702| 1.35k| parent_team)); 2703| 1.35k| } 2704| | 2705| | // TODO: GEH - cannot do this assertion because root thread not set up as 2706| | // executing 2707| | // KMP_ASSERT( master_th->th.th_current_task->td_flags.executing == 0 ); 2708| 1.35k| master_th->th.th_current_task->td_flags.executing = 1; 2709| | 2710| 1.35k| __kmp_release_bootstrap_lock(&__kmp_forkjoin_lock); 2711| | 2712| 1.35k|#if KMP_AFFINITY_SUPPORTED 2713| 1.35k| if (master_th->th.th_team->t.t_level == 0 && __kmp_affinity.flags.reset) { ------------------ | Branch (2713:7): [True: 1.35k, False: 0] | Branch (2713:48): [True: 0, False: 1.35k] ------------------ 2714| 0| __kmp_reset_root_init_mask(gtid); 2715| 0| } 2716| 1.35k|#endif 2717| 1.35k|#if OMPT_SUPPORT 2718| 1.35k| int flags = 2719| 1.35k| OMPT_INVOKER(fork_context) | ------------------ | | 24| 1.35k| ((x == fork_context_gnu) ? ompt_parallel_invoker_program \ | | ------------------ | | | Branch (24:4): [True: 0, False: 1.35k] | | ------------------ | | 25| 1.35k| : ompt_parallel_invoker_runtime) ------------------ 2720| 1.35k| ((team_microtask == (void *)__kmp_teams_master) ? ompt_parallel_league ------------------ | Branch (2720:8): [True: 0, False: 1.35k] ------------------ 2721| 1.35k| : ompt_parallel_team); 2722| 1.35k| if (ompt_enabled.enabled) { ------------------ | Branch (2722:7): [True: 0, False: 1.35k] ------------------ 2723| 0| __kmp_join_ompt(gtid, master_th, parent_team, parallel_data, flags, 2724| 0| codeptr); 2725| 0| } 2726| 1.35k|#endif 2727| | 2728| 1.35k| KMP_MB(); 2729| 1.35k| KA_TRACE(20, ("__kmp_join_call: exit T#%d\n", gtid)); 2730| 1.35k|} __kmp_get_schedule_global: 3134| 3|kmp_r_sched_t __kmp_get_schedule_global() { 3135| | // This routine created because pairs (__kmp_sched, __kmp_chunk) and 3136| | // (__kmp_static, __kmp_guided) may be changed by kmp_set_defaults 3137| | // independently. So one can get the updated schedule here. 3138| | 3139| 3| kmp_r_sched_t r_sched; 3140| | 3141| | // create schedule from 4 globals: __kmp_sched, __kmp_chunk, __kmp_static, 3142| | // __kmp_guided. __kmp_sched should keep original value, so that user can set 3143| | // KMP_SCHEDULE multiple times, and thus have different run-time schedules in 3144| | // different roots (even in OMP 2.5) 3145| 3| enum sched_type s = SCHEDULE_WITHOUT_MODIFIERS(__kmp_sched); ------------------ | | 451| 3| (enum sched_type)( \ | | 452| 3| (s) & ~(kmp_sch_modifier_nonmonotonic | kmp_sch_modifier_monotonic)) ------------------ 3146| 3| enum sched_type sched_modifiers = SCHEDULE_GET_MODIFIERS(__kmp_sched); ------------------ | | 458| 3| ((enum sched_type)( \ | | 459| 3| (s) & (kmp_sch_modifier_nonmonotonic | kmp_sch_modifier_monotonic))) ------------------ 3147| 3| if (s == kmp_sch_static) { ------------------ | Branch (3147:7): [True: 3, False: 0] ------------------ 3148| | // replace STATIC with more detailed schedule (balanced or greedy) 3149| 3| r_sched.r_sched_type = __kmp_static; 3150| 3| } else if (s == kmp_sch_guided_chunked) { ------------------ | Branch (3150:14): [True: 0, False: 0] ------------------ 3151| | // replace GUIDED with more detailed schedule (iterative or analytical) 3152| 0| r_sched.r_sched_type = __kmp_guided; 3153| 0| } else { // (STATIC_CHUNKED), or (DYNAMIC_CHUNKED), or other 3154| 0| r_sched.r_sched_type = __kmp_sched; 3155| 0| } 3156| 3| SCHEDULE_SET_MODIFIERS(r_sched.r_sched_type, sched_modifiers); ------------------ | | 461| 3| (s = (enum sched_type)((kmp_int32)s | (kmp_int32)m)) ------------------ 3157| | 3158| 3| if (__kmp_chunk < KMP_DEFAULT_CHUNK) { ------------------ | | 1244| 3|#define KMP_DEFAULT_CHUNK 1 ------------------ | Branch (3158:7): [True: 3, False: 0] ------------------ 3159| | // __kmp_chunk may be wrong here (if it was not ever set) 3160| 3| r_sched.chunk = KMP_DEFAULT_CHUNK; ------------------ | | 1244| 3|#define KMP_DEFAULT_CHUNK 1 ------------------ 3161| 3| } else { 3162| 0| r_sched.chunk = __kmp_chunk; 3163| 0| } 3164| | 3165| 3| return r_sched; 3166| 3|} __kmp_init_random: 3634| 32|void __kmp_init_random(kmp_info_t *thread) { 3635| 32| unsigned seed = thread->th.th_info.ds.ds_tid; 3636| | 3637| 32| thread->th.th_a = 3638| 32| __kmp_primes[seed % (sizeof(__kmp_primes) / sizeof(__kmp_primes[0]))]; 3639| 32| thread->th.th_x = (seed + 1) * thread->th.th_a + 1; 3640| 32| KA_TRACE(30, 3641| 32| ("__kmp_init_random: THREAD: %u; A: %u\n", seed, thread->th.th_a)); 3642| 32|} __kmp_register_root: 3774| 1|int __kmp_register_root(int initial_thread) { 3775| 1| kmp_info_t *root_thread; 3776| 1| kmp_root_t *root; 3777| 1| int gtid; 3778| 1| int capacity; 3779| 1| __kmp_acquire_bootstrap_lock(&__kmp_forkjoin_lock); 3780| 1| KA_TRACE(20, ("__kmp_register_root: entered\n")); 3781| 1| KMP_MB(); 3782| | 3783| | /* 2007-03-02: 3784| | If initial thread did not invoke OpenMP RTL yet, and this thread is not an 3785| | initial one, "__kmp_all_nth >= __kmp_threads_capacity" condition does not 3786| | work as expected -- it may return false (that means there is at least one 3787| | empty slot in __kmp_threads array), but it is possible the only free slot 3788| | is #0, which is reserved for initial thread and so cannot be used for this 3789| | one. Following code workarounds this bug. 3790| | 3791| | However, right solution seems to be not reserving slot #0 for initial 3792| | thread because: 3793| | (1) there is no magic in slot #0, 3794| | (2) we cannot detect initial thread reliably (the first thread which does 3795| | serial initialization may be not a real initial thread). 3796| | */ 3797| 1| capacity = __kmp_threads_capacity; 3798| 1| if (!initial_thread && TCR_PTR(__kmp_threads[0]) == NULL) { ------------------ | | 1156| 0|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 0|#define TCR_8(a) (a) | | ------------------ ------------------ | Branch (3798:7): [True: 0, False: 1] | Branch (3798:26): [True: 0, False: 0] ------------------ 3799| 0| --capacity; 3800| 0| } 3801| | 3802| | // If it is not for initializing the hidden helper team, we need to take 3803| | // __kmp_hidden_helper_threads_num out of the capacity because it is included 3804| | // in __kmp_threads_capacity. 3805| 1| if (__kmp_enable_hidden_helper && !TCR_4(__kmp_init_hidden_helper_threads)) { ------------------ | | 1127| 1|#define TCR_4(a) (a) ------------------ | Branch (3805:7): [True: 1, False: 0] | Branch (3805:37): [True: 1, False: 0] ------------------ 3806| 1| capacity -= __kmp_hidden_helper_threads_num; 3807| 1| } 3808| | 3809| | /* see if there are too many threads */ 3810| 1| if (__kmp_all_nth >= capacity && !__kmp_expand_threads(1)) { ------------------ | Branch (3810:7): [True: 0, False: 1] | Branch (3810:36): [True: 0, False: 0] ------------------ 3811| 0| if (__kmp_tp_cached) { ------------------ | Branch (3811:9): [True: 0, False: 0] ------------------ 3812| 0| __kmp_fatal(KMP_MSG(CantRegisterNewThread), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ 3813| 0| KMP_HNT(Set_ALL_THREADPRIVATE, __kmp_tp_capacity), ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 3814| 0| KMP_HNT(PossibleSystemLimitOnThreads), __kmp_msg_null); ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 3815| 0| } else { 3816| 0| __kmp_fatal(KMP_MSG(CantRegisterNewThread), KMP_HNT(SystemLimitOnThreads), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ __kmp_fatal(KMP_MSG(CantRegisterNewThread), KMP_HNT(SystemLimitOnThreads), ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 3817| 0| __kmp_msg_null); 3818| 0| } 3819| 0| } 3820| | 3821| | // When hidden helper task is enabled, __kmp_threads is organized as follows: 3822| | // 0: initial thread, also a regular OpenMP thread. 3823| | // [1, __kmp_hidden_helper_threads_num]: slots for hidden helper threads. 3824| | // [__kmp_hidden_helper_threads_num + 1, __kmp_threads_capacity): slots for 3825| | // regular OpenMP threads. 3826| 1| if (TCR_4(__kmp_init_hidden_helper_threads)) { ------------------ | | 1127| 1|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 1] | | ------------------ ------------------ 3827| | // Find an available thread slot for hidden helper thread. Slots for hidden 3828| | // helper threads start from 1 to __kmp_hidden_helper_threads_num. 3829| 0| for (gtid = 1; TCR_PTR(__kmp_threads[gtid]) != NULL && ------------------ | | 1156| 0|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 0|#define TCR_8(a) (a) | | ------------------ ------------------ | Branch (3829:20): [True: 0, False: 0] ------------------ 3830| 0| gtid <= __kmp_hidden_helper_threads_num; ------------------ | Branch (3830:20): [True: 0, False: 0] ------------------ 3831| 0| gtid++) 3832| 0| ; 3833| 0| KMP_ASSERT(gtid <= __kmp_hidden_helper_threads_num); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 3834| 0| KA_TRACE(1, ("__kmp_register_root: found slot in threads array for " 3835| 0| "hidden helper thread: T#%d\n", 3836| 0| gtid)); 3837| 1| } else { 3838| | /* find an available thread slot */ 3839| | // Don't reassign the zero slot since we need that to only be used by 3840| | // initial thread. Slots for hidden helper threads should also be skipped. 3841| 1| if (initial_thread && TCR_PTR(__kmp_threads[0]) == NULL) { ------------------ | | 1156| 1|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 1|#define TCR_8(a) (a) | | ------------------ ------------------ | Branch (3841:9): [True: 1, False: 0] | Branch (3841:27): [True: 1, False: 0] ------------------ 3842| 1| gtid = 0; 3843| 1| } else { 3844| 0| for (gtid = __kmp_hidden_helper_threads_num + 1; 3845| 0| TCR_PTR(__kmp_threads[gtid]) != NULL; gtid++) ------------------ | | 1156| 0|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 0|#define TCR_8(a) (a) | | ------------------ ------------------ | Branch (3845:12): [True: 0, False: 0] ------------------ 3846| 0| ; 3847| 0| } 3848| 1| KA_TRACE( 3849| 1| 1, ("__kmp_register_root: found slot in threads array: T#%d\n", gtid)); 3850| 1| KMP_ASSERT(gtid < __kmp_threads_capacity); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 3851| 1| } 3852| | 3853| | /* update global accounting */ 3854| 1| __kmp_all_nth++; 3855| 1| TCW_4(__kmp_nth, __kmp_nth + 1); ------------------ | | 1128| 1|#define TCW_4(a, b) (a) = (b) ------------------ 3856| | 3857| | // if __kmp_adjust_gtid_mode is set, then we use method #1 (sp search) for low 3858| | // numbers of procs, and method #2 (keyed API call) for higher numbers. 3859| 1| if (__kmp_adjust_gtid_mode) { ------------------ | Branch (3859:7): [True: 0, False: 1] ------------------ 3860| 0| if (__kmp_all_nth >= __kmp_tls_gtid_min) { ------------------ | Branch (3860:9): [True: 0, False: 0] ------------------ 3861| 0| if (TCR_4(__kmp_gtid_mode) != 2) { ------------------ | | 1127| 0|#define TCR_4(a) (a) ------------------ | Branch (3861:11): [True: 0, False: 0] ------------------ 3862| 0| TCW_4(__kmp_gtid_mode, 2); ------------------ | | 1128| 0|#define TCW_4(a, b) (a) = (b) ------------------ 3863| 0| } 3864| 0| } else { 3865| 0| if (TCR_4(__kmp_gtid_mode) != 1) { ------------------ | | 1127| 0|#define TCR_4(a) (a) ------------------ | Branch (3865:11): [True: 0, False: 0] ------------------ 3866| 0| TCW_4(__kmp_gtid_mode, 1); ------------------ | | 1128| 0|#define TCW_4(a, b) (a) = (b) ------------------ 3867| 0| } 3868| 0| } 3869| 0| } 3870| | 3871| 1|#ifdef KMP_ADJUST_BLOCKTIME 3872| | /* Adjust blocktime to zero if necessary */ 3873| | /* Middle initialization might not have occurred yet */ 3874| 1| if (!__kmp_env_blocktime && (__kmp_avail_proc > 0)) { ------------------ | Branch (3874:7): [True: 1, False: 0] | Branch (3874:31): [True: 0, False: 1] ------------------ 3875| 0| if (__kmp_nth > __kmp_avail_proc) { ------------------ | Branch (3875:9): [True: 0, False: 0] ------------------ 3876| 0| __kmp_zero_bt = TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 3877| 0| } 3878| 0| } 3879| 1|#endif /* KMP_ADJUST_BLOCKTIME */ 3880| | 3881| | /* setup this new hierarchy */ 3882| 1| if (!(root = __kmp_root[gtid])) { ------------------ | Branch (3882:7): [True: 1, False: 0] ------------------ 3883| 1| root = __kmp_root[gtid] = (kmp_root_t *)__kmp_allocate(sizeof(kmp_root_t)); ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3884| 1| KMP_DEBUG_ASSERT(!root->r.r_root_team); 3885| 1| } 3886| | 3887| |#if KMP_STATS_ENABLED 3888| | // Initialize stats as soon as possible (right after gtid assignment). 3889| | __kmp_stats_thread_ptr = __kmp_stats_list->push_back(gtid); 3890| | __kmp_stats_thread_ptr->startLife(); 3891| | KMP_SET_THREAD_STATE(SERIAL_REGION); 3892| | KMP_INIT_PARTITIONED_TIMERS(OMP_serial); 3893| |#endif 3894| 1| __kmp_initialize_root(root); 3895| | 3896| | /* setup new root thread structure */ 3897| 1| if (root->r.r_uber_thread) { ------------------ | Branch (3897:7): [True: 0, False: 1] ------------------ 3898| 0| root_thread = root->r.r_uber_thread; 3899| 1| } else { 3900| 1| root_thread = (kmp_info_t *)__kmp_allocate(sizeof(kmp_info_t)); ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3901| 1| if (__kmp_storage_map) { ------------------ | Branch (3901:9): [True: 0, False: 1] ------------------ 3902| 0| __kmp_print_thread_storage_map(root_thread, gtid); 3903| 0| } 3904| 1| root_thread->th.th_info.ds.ds_gtid = gtid; 3905| 1|#if OMPT_SUPPORT 3906| 1| root_thread->th.ompt_thread_info.thread_data = ompt_data_none; ------------------ | | 1408| 1|#define ompt_data_none {0} ------------------ 3907| 1|#endif 3908| 1| root_thread->th.th_root = root; 3909| 1| if (__kmp_env_consistency_check) { ------------------ | Branch (3909:9): [True: 0, False: 1] ------------------ 3910| 0| root_thread->th.th_cons = __kmp_allocate_cons_stack(gtid); 3911| 0| } 3912| 1|#if USE_FAST_MEMORY 3913| 1| __kmp_initialize_fast_memory(root_thread); 3914| 1|#endif /* USE_FAST_MEMORY */ 3915| | 3916| 1|#if KMP_USE_BGET 3917| 1| KMP_DEBUG_ASSERT(root_thread->th.th_local.bget_data == NULL); 3918| 1| __kmp_initialize_bget(root_thread); 3919| 1|#endif 3920| 1| __kmp_init_random(root_thread); // Initialize random number generator 3921| 1| } 3922| | 3923| | /* setup the serial team held in reserve by the root thread */ 3924| 1| if (!root_thread->th.th_serial_team) { ------------------ | Branch (3924:7): [True: 1, False: 0] ------------------ 3925| 1| kmp_internal_control_t r_icvs = __kmp_get_global_icvs(); 3926| 1| KF_TRACE(10, ("__kmp_register_root: before serial_team\n")); 3927| 1| root_thread->th.th_serial_team = __kmp_allocate_team( 3928| 1| root, 1, 1, 3929| 1|#if OMPT_SUPPORT 3930| 1| ompt_data_none, // root parallel id ------------------ | | 1408| 1|#define ompt_data_none {0} ------------------ 3931| 1|#endif 3932| 1| proc_bind_default, &r_icvs, 0 USE_NESTED_HOT_ARG(NULL)); ------------------ | | 165| 1|#define USE_NESTED_HOT_ARG(x) , x ------------------ 3933| 1| } 3934| 1| KMP_ASSERT(root_thread->th.th_serial_team); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 3935| 1| KF_TRACE(10, ("__kmp_register_root: after serial_team = %p\n", 3936| 1| root_thread->th.th_serial_team)); 3937| | 3938| | /* drop root_thread into place */ 3939| 1| TCW_SYNC_PTR(__kmp_threads[gtid], root_thread); ------------------ | | 1159| 1|#define TCW_SYNC_PTR(a, b) TCW_SYNC_8((a), (b)) | | ------------------ | | | | 1141| 1|#define TCW_SYNC_8(a, b) (a) = (b) | | ------------------ ------------------ 3940| | 3941| 1| root->r.r_root_team->t.t_threads[0] = root_thread; 3942| 1| root->r.r_hot_team->t.t_threads[0] = root_thread; 3943| 1| root_thread->th.th_serial_team->t.t_threads[0] = root_thread; 3944| | // AC: the team created in reserve, not for execution (it is unused for now). 3945| 1| root_thread->th.th_serial_team->t.t_serialized = 0; 3946| 1| root->r.r_uber_thread = root_thread; 3947| | 3948| | /* initialize the thread, get it ready to go */ 3949| 1| __kmp_initialize_info(root_thread, root->r.r_root_team, 0, gtid); 3950| 1| TCW_4(__kmp_init_gtid, TRUE); ------------------ | | 1128| 1|#define TCW_4(a, b) (a) = (b) ------------------ 3951| | 3952| | /* prepare the primary thread for get_gtid() */ 3953| 1| __kmp_gtid_set_specific(gtid); 3954| | 3955| |#if USE_ITT_BUILD 3956| | __kmp_itt_thread_name(gtid); 3957| |#endif /* USE_ITT_BUILD */ 3958| | 3959| 1|#ifdef KMP_TDATA_GTID 3960| 1| __kmp_gtid = gtid; 3961| 1|#endif 3962| 1| __kmp_create_worker(gtid, root_thread, __kmp_stksize); 3963| 1| KMP_DEBUG_ASSERT(__kmp_gtid_get_specific() == gtid); 3964| | 3965| 1| KA_TRACE(20, ("__kmp_register_root: T#%d init T#%d(%d:%d) arrived: join=%u, " 3966| 1| "plain=%u\n", 3967| 1| gtid, __kmp_gtid_from_tid(0, root->r.r_hot_team), 3968| 1| root->r.r_hot_team->t.t_id, 0, KMP_INIT_BARRIER_STATE, 3969| 1| KMP_INIT_BARRIER_STATE)); 3970| 1| { // Initialize barrier data. 3971| 1| int b; 3972| 4| for (b = 0; b < bs_last_barrier; ++b) { ------------------ | Branch (3972:17): [True: 3, False: 1] ------------------ 3973| 3| root_thread->th.th_bar[b].bb.b_arrived = KMP_INIT_BARRIER_STATE; ------------------ | | 2018| 3|#define KMP_INIT_BARRIER_STATE 0 /* should probably start from zero */ ------------------ 3974| |#if USE_DEBUGGER 3975| | root_thread->th.th_bar[b].bb.b_worker_arrived = 0; 3976| |#endif 3977| 3| } 3978| 1| } 3979| 1| KMP_DEBUG_ASSERT(root->r.r_hot_team->t.t_bar[bs_forkjoin_barrier].b_arrived == 3980| 1| KMP_INIT_BARRIER_STATE); 3981| | 3982| 1|#if KMP_AFFINITY_SUPPORTED 3983| 1| root_thread->th.th_current_place = KMP_PLACE_UNDEFINED; ------------------ | | 960| 1|#define KMP_PLACE_UNDEFINED (-2) ------------------ 3984| 1| root_thread->th.th_new_place = KMP_PLACE_UNDEFINED; ------------------ | | 960| 1|#define KMP_PLACE_UNDEFINED (-2) ------------------ 3985| 1| root_thread->th.th_first_place = KMP_PLACE_UNDEFINED; ------------------ | | 960| 1|#define KMP_PLACE_UNDEFINED (-2) ------------------ 3986| 1| root_thread->th.th_last_place = KMP_PLACE_UNDEFINED; ------------------ | | 960| 1|#define KMP_PLACE_UNDEFINED (-2) ------------------ 3987| 1|#endif /* KMP_AFFINITY_SUPPORTED */ 3988| 1| root_thread->th.th_def_allocator = __kmp_def_allocator; 3989| 1| root_thread->th.th_prev_level = 0; 3990| 1| root_thread->th.th_prev_num_threads = 1; 3991| | 3992| 1| kmp_cg_root_t *tmp = (kmp_cg_root_t *)__kmp_allocate(sizeof(kmp_cg_root_t)); ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3993| 1| tmp->cg_root = root_thread; 3994| 1| tmp->cg_thread_limit = __kmp_cg_max_nth; 3995| 1| tmp->cg_nthreads = 1; 3996| 1| KA_TRACE(100, ("__kmp_register_root: Thread %p created node %p with" 3997| 1| " cg_nthreads init to 1\n", 3998| 1| root_thread, tmp)); 3999| 1| tmp->up = NULL; 4000| 1| root_thread->th.th_cg_roots = tmp; 4001| | 4002| 1| __kmp_root_counter++; 4003| | 4004| 1|#if OMPT_SUPPORT 4005| 1| if (!initial_thread && ompt_enabled.enabled) { ------------------ | Branch (4005:7): [True: 0, False: 1] | Branch (4005:26): [True: 0, False: 0] ------------------ 4006| | 4007| 0| kmp_info_t *root_thread = ompt_get_thread(); 4008| | 4009| 0| ompt_set_thread_state(root_thread, ompt_state_overhead); 4010| | 4011| 0| if (ompt_enabled.ompt_callback_thread_begin) { ------------------ | Branch (4011:9): [True: 0, False: 0] ------------------ 4012| 0| ompt_callbacks.ompt_callback(ompt_callback_thread_begin)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 4013| 0| ompt_thread_initial, __ompt_get_thread_data_internal()); 4014| 0| } 4015| 0| ompt_data_t *task_data; 4016| 0| ompt_data_t *parallel_data; 4017| 0| __ompt_get_task_info_internal(0, NULL, &task_data, NULL, ¶llel_data, 4018| 0| NULL); 4019| 0| if (ompt_enabled.ompt_callback_implicit_task) { ------------------ | Branch (4019:9): [True: 0, False: 0] ------------------ 4020| 0| ompt_callbacks.ompt_callback(ompt_callback_implicit_task)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 4021| 0| ompt_scope_begin, parallel_data, task_data, 1, 1, ompt_task_initial); 4022| 0| } 4023| | 4024| 0| ompt_set_thread_state(root_thread, ompt_state_work_serial); 4025| 0| } 4026| 1|#endif 4027| 1|#if OMPD_SUPPORT 4028| 1| if (ompd_state & OMPD_ENABLE_BP) ------------------ | | 35| 1|#define OMPD_ENABLE_BP 0x1 ------------------ | Branch (4028:7): [True: 0, False: 1] ------------------ 4029| 0| ompd_bp_thread_begin(); 4030| 1|#endif 4031| | 4032| 1| KMP_MB(); 4033| 1| __kmp_release_bootstrap_lock(&__kmp_forkjoin_lock); 4034| | 4035| 1| return gtid; 4036| 1|} _Z36__kmp_unregister_root_current_threadi: 4160| 1|void __kmp_unregister_root_current_thread(int gtid) { 4161| 1| KA_TRACE(1, ("__kmp_unregister_root_current_thread: enter T#%d\n", gtid)); 4162| | /* this lock should be ok, since unregister_root_current_thread is never 4163| | called during an abort, only during a normal close. furthermore, if you 4164| | have the forkjoin lock, you should never try to get the initz lock */ 4165| 1| __kmp_acquire_bootstrap_lock(&__kmp_forkjoin_lock); 4166| 1| if (TCR_4(__kmp_global.g.g_done) || !__kmp_init_serial) { ------------------ | | 1127| 2|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 1] | | ------------------ ------------------ | Branch (4166:39): [True: 0, False: 1] ------------------ 4167| 0| KC_TRACE(10, ("__kmp_unregister_root_current_thread: already finished, " 4168| 0| "exiting T#%d\n", 4169| 0| gtid)); 4170| 0| __kmp_release_bootstrap_lock(&__kmp_forkjoin_lock); 4171| 0| return; 4172| 0| } 4173| 1| kmp_root_t *root = __kmp_root[gtid]; 4174| | 4175| 1| KMP_DEBUG_ASSERT(__kmp_threads && __kmp_threads[gtid]); 4176| 1| KMP_ASSERT(KMP_UBER_GTID(gtid)); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4177| 1| KMP_ASSERT(root == __kmp_threads[gtid]->th.th_root); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4178| 1| KMP_ASSERT(root->r.r_active == FALSE); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4179| | 4180| 1| KMP_MB(); 4181| | 4182| 1| kmp_info_t *thread = __kmp_threads[gtid]; 4183| 1| kmp_team_t *team = thread->th.th_team; 4184| 1| kmp_task_team_t *task_team = thread->th.th_task_team; 4185| | 4186| | // we need to wait for the proxy tasks before finishing the thread 4187| 1| if (task_team != NULL && (task_team->tt.tt_found_proxy_tasks || ------------------ | Branch (4187:7): [True: 0, False: 1] | Branch (4187:29): [True: 0, False: 0] ------------------ 4188| 0| task_team->tt.tt_hidden_helper_task_encountered)) { ------------------ | Branch (4188:29): [True: 0, False: 0] ------------------ 4189| 0|#if OMPT_SUPPORT 4190| | // the runtime is shutting down so we won't report any events 4191| 0| thread->th.ompt_thread_info.state = ompt_state_undefined; 4192| 0|#endif 4193| 0| __kmp_task_team_wait(thread, team USE_ITT_BUILD_ARG(NULL)); 4194| 0| } 4195| | 4196| 1| __kmp_reset_root(gtid, root); 4197| | 4198| 1| KMP_MB(); 4199| 1| KC_TRACE(10, 4200| 1| ("__kmp_unregister_root_current_thread: T#%d unregistered\n", gtid)); 4201| | 4202| 1| __kmp_release_bootstrap_lock(&__kmp_forkjoin_lock); 4203| 1|} __kmp_allocate_thread: 4405| 31| int new_tid) { 4406| 31| kmp_team_t *serial_team; 4407| 31| kmp_info_t *new_thr; 4408| 31| int new_gtid; 4409| | 4410| 31| KA_TRACE(20, ("__kmp_allocate_thread: T#%d\n", __kmp_get_gtid())); 4411| 31| KMP_DEBUG_ASSERT(root && team); 4412| |#if !KMP_NESTED_HOT_TEAMS 4413| | KMP_DEBUG_ASSERT(KMP_MASTER_GTID(__kmp_get_gtid())); 4414| |#endif 4415| 31| KMP_MB(); 4416| | 4417| | /* first, try to get one from the thread pool */ 4418| 31| if (__kmp_thread_pool) { ------------------ | Branch (4418:7): [True: 0, False: 31] ------------------ 4419| 0| new_thr = CCAST(kmp_info_t *, __kmp_thread_pool); ------------------ | | 229| 0|#define CCAST(type, var) const_cast(var) ------------------ 4420| 0| __kmp_thread_pool = (volatile kmp_info_t *)new_thr->th.th_next_pool; 4421| 0| if (new_thr == __kmp_thread_pool_insert_pt) { ------------------ | Branch (4421:9): [True: 0, False: 0] ------------------ 4422| 0| __kmp_thread_pool_insert_pt = NULL; 4423| 0| } 4424| 0| TCW_4(new_thr->th.th_in_pool, FALSE); ------------------ | | 1128| 0|#define TCW_4(a, b) (a) = (b) ------------------ 4425| 0| __kmp_suspend_initialize_thread(new_thr); 4426| 0| __kmp_lock_suspend_mx(new_thr); 4427| 0| if (new_thr->th.th_active_in_pool == TRUE) { ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ | Branch (4427:9): [True: 0, False: 0] ------------------ 4428| 0| KMP_DEBUG_ASSERT(new_thr->th.th_active == TRUE); 4429| 0| KMP_ATOMIC_DEC(&__kmp_thread_pool_active_nth); ------------------ | | 1260| 0|#define KMP_ATOMIC_DEC(p) KMP_ATOMIC_OP(fetch_sub, p, 1, acq_rel) | | ------------------ | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 4430| 0| new_thr->th.th_active_in_pool = FALSE; ------------------ | | 1279| 0|#define FALSE 0 ------------------ 4431| 0| } 4432| 0| __kmp_unlock_suspend_mx(new_thr); 4433| | 4434| 0| KA_TRACE(20, ("__kmp_allocate_thread: T#%d using thread T#%d\n", 4435| 0| __kmp_get_gtid(), new_thr->th.th_info.ds.ds_gtid)); 4436| 0| KMP_ASSERT(!new_thr->th.th_team); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4437| 0| KMP_DEBUG_ASSERT(__kmp_nth < __kmp_threads_capacity); 4438| | 4439| | /* setup the thread structure */ 4440| 0| __kmp_initialize_info(new_thr, team, new_tid, 4441| 0| new_thr->th.th_info.ds.ds_gtid); 4442| 0| KMP_DEBUG_ASSERT(new_thr->th.th_serial_team); 4443| | 4444| 0| TCW_4(__kmp_nth, __kmp_nth + 1); ------------------ | | 1128| 0|#define TCW_4(a, b) (a) = (b) ------------------ 4445| | 4446| 0| new_thr->th.th_task_state = 0; 4447| 0| new_thr->th.th_task_state_top = 0; 4448| 0| new_thr->th.th_task_state_stack_sz = 4; 4449| | 4450| 0| if (__kmp_barrier_gather_pattern[bs_forkjoin_barrier] == bp_dist_bar) { ------------------ | Branch (4450:9): [True: 0, False: 0] ------------------ 4451| | // Make sure pool thread has transitioned to waiting on own thread struct 4452| 0| KMP_DEBUG_ASSERT(new_thr->th.th_used_in_team.load() == 0); 4453| | // Thread activated in __kmp_allocate_team when increasing team size 4454| 0| } 4455| | 4456| 0|#ifdef KMP_ADJUST_BLOCKTIME 4457| | /* Adjust blocktime back to zero if necessary */ 4458| | /* Middle initialization might not have occurred yet */ 4459| 0| if (!__kmp_env_blocktime && (__kmp_avail_proc > 0)) { ------------------ | Branch (4459:9): [True: 0, False: 0] | Branch (4459:33): [True: 0, False: 0] ------------------ 4460| 0| if (__kmp_nth > __kmp_avail_proc) { ------------------ | Branch (4460:11): [True: 0, False: 0] ------------------ 4461| 0| __kmp_zero_bt = TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 4462| 0| } 4463| 0| } 4464| 0|#endif /* KMP_ADJUST_BLOCKTIME */ 4465| | 4466| |#if KMP_DEBUG 4467| | // If thread entered pool via __kmp_free_thread, wait_flag should != 4468| | // KMP_BARRIER_PARENT_FLAG. 4469| | int b; 4470| | kmp_balign_t *balign = new_thr->th.th_bar; 4471| | for (b = 0; b < bs_last_barrier; ++b) 4472| | KMP_DEBUG_ASSERT(balign[b].bb.wait_flag != KMP_BARRIER_PARENT_FLAG); 4473| |#endif 4474| | 4475| 0| KF_TRACE(10, ("__kmp_allocate_thread: T#%d using thread %p T#%d\n", 4476| 0| __kmp_get_gtid(), new_thr, new_thr->th.th_info.ds.ds_gtid)); 4477| | 4478| 0| KMP_MB(); 4479| 0| return new_thr; 4480| 0| } 4481| | 4482| | /* no, well fork a new one */ 4483| 31| KMP_ASSERT(__kmp_nth == __kmp_all_nth); ------------------ | | 50| 31| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 31] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4484| 31| KMP_ASSERT(__kmp_all_nth < __kmp_threads_capacity); ------------------ | | 50| 31| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 31] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4485| | 4486| |#if KMP_USE_MONITOR 4487| | // If this is the first worker thread the RTL is creating, then also 4488| | // launch the monitor thread. We try to do this as early as possible. 4489| | if (!TCR_4(__kmp_init_monitor)) { 4490| | __kmp_acquire_bootstrap_lock(&__kmp_monitor_lock); 4491| | if (!TCR_4(__kmp_init_monitor)) { 4492| | KF_TRACE(10, ("before __kmp_create_monitor\n")); 4493| | TCW_4(__kmp_init_monitor, 1); 4494| | __kmp_create_monitor(&__kmp_monitor); 4495| | KF_TRACE(10, ("after __kmp_create_monitor\n")); 4496| |#if KMP_OS_WINDOWS 4497| | // AC: wait until monitor has started. This is a fix for CQ232808. 4498| | // The reason is that if the library is loaded/unloaded in a loop with 4499| | // small (parallel) work in between, then there is high probability that 4500| | // monitor thread started after the library shutdown. At shutdown it is 4501| | // too late to cope with the problem, because when the primary thread is 4502| | // in DllMain (process detach) the monitor has no chances to start (it is 4503| | // blocked), and primary thread has no means to inform the monitor that 4504| | // the library has gone, because all the memory which the monitor can 4505| | // access is going to be released/reset. 4506| | while (TCR_4(__kmp_init_monitor) < 2) { 4507| | KMP_YIELD(TRUE); 4508| | } 4509| | KF_TRACE(10, ("after monitor thread has started\n")); 4510| |#endif 4511| | } 4512| | __kmp_release_bootstrap_lock(&__kmp_monitor_lock); 4513| | } 4514| |#endif 4515| | 4516| 31| KMP_MB(); 4517| | 4518| 31| { 4519| 31| int new_start_gtid = TCR_4(__kmp_init_hidden_helper_threads) ------------------ | | 1127| 31|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 31] | | ------------------ ------------------ 4520| 31| ? 1 4521| 31| : __kmp_hidden_helper_threads_num + 1; 4522| | 4523| 496| for (new_gtid = new_start_gtid; TCR_PTR(__kmp_threads[new_gtid]) != NULL; ------------------ | | 1156| 496|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 496|#define TCR_8(a) (a) | | ------------------ ------------------ | Branch (4523:37): [True: 465, False: 31] ------------------ 4524| 465| ++new_gtid) { 4525| 465| KMP_DEBUG_ASSERT(new_gtid < __kmp_threads_capacity); 4526| 465| } 4527| | 4528| 31| if (TCR_4(__kmp_init_hidden_helper_threads)) { ------------------ | | 1127| 31|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 31] | | ------------------ ------------------ 4529| 0| KMP_DEBUG_ASSERT(new_gtid <= __kmp_hidden_helper_threads_num); 4530| 0| } 4531| 31| } 4532| | 4533| | /* allocate space for it. */ 4534| 31| new_thr = (kmp_info_t *)__kmp_allocate(sizeof(kmp_info_t)); ------------------ | | 3617| 31|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 4535| | 4536| 31| TCW_SYNC_PTR(__kmp_threads[new_gtid], new_thr); ------------------ | | 1159| 31|#define TCW_SYNC_PTR(a, b) TCW_SYNC_8((a), (b)) | | ------------------ | | | | 1141| 31|#define TCW_SYNC_8(a, b) (a) = (b) | | ------------------ ------------------ 4537| | 4538| |#if USE_ITT_BUILD && USE_ITT_NOTIFY && KMP_DEBUG 4539| | // suppress race conditions detection on synchronization flags in debug mode 4540| | // this helps to analyze library internals eliminating false positives 4541| | __itt_suppress_mark_range( 4542| | __itt_suppress_range, __itt_suppress_threading_errors, 4543| | &new_thr->th.th_sleep_loc, sizeof(new_thr->th.th_sleep_loc)); 4544| | __itt_suppress_mark_range( 4545| | __itt_suppress_range, __itt_suppress_threading_errors, 4546| | &new_thr->th.th_reap_state, sizeof(new_thr->th.th_reap_state)); 4547| |#if KMP_OS_WINDOWS 4548| | __itt_suppress_mark_range( 4549| | __itt_suppress_range, __itt_suppress_threading_errors, 4550| | &new_thr->th.th_suspend_init, sizeof(new_thr->th.th_suspend_init)); 4551| |#else 4552| | __itt_suppress_mark_range(__itt_suppress_range, 4553| | __itt_suppress_threading_errors, 4554| | &new_thr->th.th_suspend_init_count, 4555| | sizeof(new_thr->th.th_suspend_init_count)); 4556| |#endif 4557| | // TODO: check if we need to also suppress b_arrived flags 4558| | __itt_suppress_mark_range(__itt_suppress_range, 4559| | __itt_suppress_threading_errors, 4560| | CCAST(kmp_uint64 *, &new_thr->th.th_bar[0].bb.b_go), 4561| | sizeof(new_thr->th.th_bar[0].bb.b_go)); 4562| | __itt_suppress_mark_range(__itt_suppress_range, 4563| | __itt_suppress_threading_errors, 4564| | CCAST(kmp_uint64 *, &new_thr->th.th_bar[1].bb.b_go), 4565| | sizeof(new_thr->th.th_bar[1].bb.b_go)); 4566| | __itt_suppress_mark_range(__itt_suppress_range, 4567| | __itt_suppress_threading_errors, 4568| | CCAST(kmp_uint64 *, &new_thr->th.th_bar[2].bb.b_go), 4569| | sizeof(new_thr->th.th_bar[2].bb.b_go)); 4570| |#endif /* USE_ITT_BUILD && USE_ITT_NOTIFY && KMP_DEBUG */ 4571| 31| if (__kmp_storage_map) { ------------------ | Branch (4571:7): [True: 0, False: 31] ------------------ 4572| 0| __kmp_print_thread_storage_map(new_thr, new_gtid); 4573| 0| } 4574| | 4575| | // add the reserve serialized team, initialized from the team's primary thread 4576| 31| { 4577| 31| kmp_internal_control_t r_icvs = __kmp_get_x_global_icvs(team); 4578| 31| KF_TRACE(10, ("__kmp_allocate_thread: before th_serial/serial_team\n")); 4579| 31| new_thr->th.th_serial_team = serial_team = 4580| 31| (kmp_team_t *)__kmp_allocate_team(root, 1, 1, 4581| 31|#if OMPT_SUPPORT 4582| 31| ompt_data_none, // root parallel id ------------------ | | 1408| 31|#define ompt_data_none {0} ------------------ 4583| 31|#endif 4584| 31| proc_bind_default, &r_icvs, 4585| 31| 0 USE_NESTED_HOT_ARG(NULL)); ------------------ | | 165| 31|#define USE_NESTED_HOT_ARG(x) , x ------------------ 4586| 31| } 4587| 31| KMP_ASSERT(serial_team); ------------------ | | 50| 31| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 31] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4588| 31| serial_team->t.t_serialized = 0; // AC: the team created in reserve, not for 4589| | // execution (it is unused for now). 4590| 31| serial_team->t.t_threads[0] = new_thr; 4591| 31| KF_TRACE(10, 4592| 31| ("__kmp_allocate_thread: after th_serial/serial_team : new_thr=%p\n", 4593| 31| new_thr)); 4594| | 4595| | /* setup the thread structures */ 4596| 31| __kmp_initialize_info(new_thr, team, new_tid, new_gtid); 4597| | 4598| 31|#if USE_FAST_MEMORY 4599| 31| __kmp_initialize_fast_memory(new_thr); 4600| 31|#endif /* USE_FAST_MEMORY */ 4601| | 4602| 31|#if KMP_USE_BGET 4603| 31| KMP_DEBUG_ASSERT(new_thr->th.th_local.bget_data == NULL); 4604| 31| __kmp_initialize_bget(new_thr); 4605| 31|#endif 4606| | 4607| 31| __kmp_init_random(new_thr); // Initialize random number generator 4608| | 4609| | /* Initialize these only once when thread is grabbed for a team allocation */ 4610| 31| KA_TRACE(20, 4611| 31| ("__kmp_allocate_thread: T#%d init go fork=%u, plain=%u\n", 4612| 31| __kmp_get_gtid(), KMP_INIT_BARRIER_STATE, KMP_INIT_BARRIER_STATE)); 4613| | 4614| 31| int b; 4615| 31| kmp_balign_t *balign = new_thr->th.th_bar; 4616| 124| for (b = 0; b < bs_last_barrier; ++b) { ------------------ | Branch (4616:15): [True: 93, False: 31] ------------------ 4617| 93| balign[b].bb.b_go = KMP_INIT_BARRIER_STATE; ------------------ | | 2018| 93|#define KMP_INIT_BARRIER_STATE 0 /* should probably start from zero */ ------------------ 4618| 93| balign[b].bb.team = NULL; 4619| 93| balign[b].bb.wait_flag = KMP_BARRIER_NOT_WAITING; ------------------ | | 2035| 93|#define KMP_BARRIER_NOT_WAITING 0 // Normal state; worker not in wait_sleep ------------------ 4620| 93| balign[b].bb.use_oncore_barrier = 0; 4621| 93| } 4622| | 4623| 31| TCW_PTR(new_thr->th.th_sleep_loc, NULL); ------------------ | | 1157| 31|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 31|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 4624| 31| new_thr->th.th_sleep_loc_type = flag_unset; 4625| | 4626| 31| new_thr->th.th_spin_here = FALSE; ------------------ | | 1279| 31|#define FALSE 0 ------------------ 4627| 31| new_thr->th.th_next_waiting = 0; 4628| 31|#if KMP_OS_UNIX 4629| 31| new_thr->th.th_blocking = false; 4630| 31|#endif 4631| | 4632| 31|#if KMP_AFFINITY_SUPPORTED 4633| 31| new_thr->th.th_current_place = KMP_PLACE_UNDEFINED; ------------------ | | 960| 31|#define KMP_PLACE_UNDEFINED (-2) ------------------ 4634| 31| new_thr->th.th_new_place = KMP_PLACE_UNDEFINED; ------------------ | | 960| 31|#define KMP_PLACE_UNDEFINED (-2) ------------------ 4635| 31| new_thr->th.th_first_place = KMP_PLACE_UNDEFINED; ------------------ | | 960| 31|#define KMP_PLACE_UNDEFINED (-2) ------------------ 4636| 31| new_thr->th.th_last_place = KMP_PLACE_UNDEFINED; ------------------ | | 960| 31|#define KMP_PLACE_UNDEFINED (-2) ------------------ 4637| 31|#endif 4638| 31| new_thr->th.th_def_allocator = __kmp_def_allocator; 4639| 31| new_thr->th.th_prev_level = 0; 4640| 31| new_thr->th.th_prev_num_threads = 1; 4641| | 4642| 31| TCW_4(new_thr->th.th_in_pool, FALSE); ------------------ | | 1128| 31|#define TCW_4(a, b) (a) = (b) ------------------ 4643| 31| new_thr->th.th_active_in_pool = FALSE; ------------------ | | 1279| 31|#define FALSE 0 ------------------ 4644| 31| TCW_4(new_thr->th.th_active, TRUE); ------------------ | | 1128| 31|#define TCW_4(a, b) (a) = (b) ------------------ 4645| | 4646| | /* adjust the global counters */ 4647| 31| __kmp_all_nth++; 4648| 31| __kmp_nth++; 4649| | 4650| | // if __kmp_adjust_gtid_mode is set, then we use method #1 (sp search) for low 4651| | // numbers of procs, and method #2 (keyed API call) for higher numbers. 4652| 31| if (__kmp_adjust_gtid_mode) { ------------------ | Branch (4652:7): [True: 0, False: 31] ------------------ 4653| 0| if (__kmp_all_nth >= __kmp_tls_gtid_min) { ------------------ | Branch (4653:9): [True: 0, False: 0] ------------------ 4654| 0| if (TCR_4(__kmp_gtid_mode) != 2) { ------------------ | | 1127| 0|#define TCR_4(a) (a) ------------------ | Branch (4654:11): [True: 0, False: 0] ------------------ 4655| 0| TCW_4(__kmp_gtid_mode, 2); ------------------ | | 1128| 0|#define TCW_4(a, b) (a) = (b) ------------------ 4656| 0| } 4657| 0| } else { 4658| 0| if (TCR_4(__kmp_gtid_mode) != 1) { ------------------ | | 1127| 0|#define TCR_4(a) (a) ------------------ | Branch (4658:11): [True: 0, False: 0] ------------------ 4659| 0| TCW_4(__kmp_gtid_mode, 1); ------------------ | | 1128| 0|#define TCW_4(a, b) (a) = (b) ------------------ 4660| 0| } 4661| 0| } 4662| 0| } 4663| | 4664| 31|#ifdef KMP_ADJUST_BLOCKTIME 4665| | /* Adjust blocktime back to zero if necessary */ 4666| | /* Middle initialization might not have occurred yet */ 4667| 31| if (!__kmp_env_blocktime && (__kmp_avail_proc > 0)) { ------------------ | Branch (4667:7): [True: 31, False: 0] | Branch (4667:31): [True: 31, False: 0] ------------------ 4668| 31| if (__kmp_nth > __kmp_avail_proc) { ------------------ | Branch (4668:9): [True: 0, False: 31] ------------------ 4669| 0| __kmp_zero_bt = TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 4670| 0| } 4671| 31| } 4672| 31|#endif /* KMP_ADJUST_BLOCKTIME */ 4673| | 4674| | /* actually fork it and create the new worker thread */ 4675| 31| KF_TRACE( 4676| 31| 10, ("__kmp_allocate_thread: before __kmp_create_worker: %p\n", new_thr)); 4677| 31| __kmp_create_worker(new_gtid, new_thr, __kmp_stksize); 4678| 31| KF_TRACE(10, 4679| 31| ("__kmp_allocate_thread: after __kmp_create_worker: %p\n", new_thr)); 4680| | 4681| 31| KA_TRACE(20, ("__kmp_allocate_thread: T#%d forked T#%d\n", __kmp_get_gtid(), 4682| 31| new_gtid)); 4683| 31| KMP_MB(); 4684| 31| return new_thr; 4685| 31|} __kmp_allocate_team: 5130| 1.38k| int argc USE_NESTED_HOT_ARG(kmp_info_t *master)) { 5131| 1.38k| KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(KMP_allocate_team); ------------------ | | 1008| 1.38k|#define KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(n) ((void)0) ------------------ 5132| 1.38k| int f; 5133| 1.38k| kmp_team_t *team; 5134| 1.38k| int use_hot_team = !root->r.r_active; 5135| 1.38k| int level = 0; 5136| 1.38k| int do_place_partition = 1; 5137| | 5138| 1.38k| KA_TRACE(20, ("__kmp_allocate_team: called\n")); 5139| 1.38k| KMP_DEBUG_ASSERT(new_nproc >= 1 && argc >= 0); 5140| 1.38k| KMP_DEBUG_ASSERT(max_nproc >= new_nproc); 5141| 1.38k| KMP_MB(); 5142| | 5143| 1.38k|#if KMP_NESTED_HOT_TEAMS 5144| 1.38k| kmp_hot_team_ptr_t *hot_teams; 5145| 1.38k| if (master) { ------------------ | Branch (5145:7): [True: 1.35k, False: 34] ------------------ 5146| 1.35k| team = master->th.th_team; 5147| 1.35k| level = team->t.t_active_level; 5148| 1.35k| if (master->th.th_teams_microtask) { // in teams construct? ------------------ | Branch (5148:9): [True: 0, False: 1.35k] ------------------ 5149| 0| if (master->th.th_teams_size.nteams > 1 && ------------------ | Branch (5149:11): [True: 0, False: 0] ------------------ 5150| 0| ( // #teams > 1 5151| 0| team->t.t_pkfn == ------------------ | Branch (5151:15): [True: 0, False: 0] ------------------ 5152| 0| (microtask_t)__kmp_teams_master || // inner fork of the teams 5153| 0| master->th.th_teams_level < ------------------ | Branch (5153:15): [True: 0, False: 0] ------------------ 5154| 0| team->t.t_level)) { // or nested parallel inside the teams 5155| 0| ++level; // not increment if #teams==1, or for outer fork of the teams; 5156| | // increment otherwise 5157| 0| } 5158| | // Do not perform the place partition if inner fork of the teams 5159| | // Wait until nested parallel region encountered inside teams construct 5160| 0| if ((master->th.th_teams_size.nteams == 1 && ------------------ | Branch (5160:12): [True: 0, False: 0] ------------------ 5161| 0| master->th.th_teams_level >= team->t.t_level) || ------------------ | Branch (5161:12): [True: 0, False: 0] ------------------ 5162| 0| (team->t.t_pkfn == (microtask_t)__kmp_teams_master)) ------------------ | Branch (5162:11): [True: 0, False: 0] ------------------ 5163| 0| do_place_partition = 0; 5164| 0| } 5165| 1.35k| hot_teams = master->th.th_hot_teams; 5166| 1.35k| if (level < __kmp_hot_teams_max_level && hot_teams && ------------------ | Branch (5166:9): [True: 1.35k, False: 0] | Branch (5166:46): [True: 1.35k, False: 0] ------------------ 5167| 1.35k| hot_teams[level].hot_team) { ------------------ | Branch (5167:9): [True: 1.35k, False: 0] ------------------ 5168| | // hot team has already been allocated for given level 5169| 1.35k| use_hot_team = 1; 5170| 1.35k| } else { 5171| 0| use_hot_team = 0; 5172| 0| } 5173| 1.35k| } else { 5174| | // check we won't access uninitialized hot_teams, just in case 5175| 34| KMP_DEBUG_ASSERT(new_nproc == 1); 5176| 34| } 5177| 1.38k|#endif 5178| | // Optimization to use a "hot" team 5179| 1.38k| if (use_hot_team && new_nproc > 1) { ------------------ | Branch (5179:7): [True: 1.38k, False: 0] | Branch (5179:23): [True: 1.35k, False: 34] ------------------ 5180| 1.35k| KMP_DEBUG_ASSERT(new_nproc <= max_nproc); 5181| 1.35k|#if KMP_NESTED_HOT_TEAMS 5182| 1.35k| team = hot_teams[level].hot_team; 5183| |#else 5184| | team = root->r.r_hot_team; 5185| |#endif 5186| |#if KMP_DEBUG 5187| | if (__kmp_tasking_mode != tskm_immediate_exec) { 5188| | KA_TRACE(20, ("__kmp_allocate_team: hot team task_team[0] = %p " 5189| | "task_team[1] = %p before reinit\n", 5190| | team->t.t_task_team[0], team->t.t_task_team[1])); 5191| | } 5192| |#endif 5193| | 5194| 1.35k| if (team->t.t_nproc != new_nproc && ------------------ | Branch (5194:9): [True: 1, False: 1.35k] ------------------ 5195| 1| __kmp_barrier_release_pattern[bs_forkjoin_barrier] == bp_dist_bar) { ------------------ | Branch (5195:9): [True: 0, False: 1] ------------------ 5196| | // Distributed barrier may need a resize 5197| 0| int old_nthr = team->t.t_nproc; 5198| 0| __kmp_resize_dist_barrier(team, old_nthr, new_nproc); 5199| 0| } 5200| | 5201| | // If not doing the place partition, then reset the team's proc bind 5202| | // to indicate that partitioning of all threads still needs to take place 5203| 1.35k| if (do_place_partition == 0) ------------------ | Branch (5203:9): [True: 0, False: 1.35k] ------------------ 5204| 0| team->t.t_proc_bind = proc_bind_default; 5205| | // Has the number of threads changed? 5206| | /* Let's assume the most common case is that the number of threads is 5207| | unchanged, and put that case first. */ 5208| 1.35k| if (team->t.t_nproc == new_nproc) { // Check changes in number of threads ------------------ | Branch (5208:9): [True: 1.35k, False: 1] ------------------ 5209| 1.35k| KA_TRACE(20, ("__kmp_allocate_team: reusing hot team\n")); 5210| | // This case can mean that omp_set_num_threads() was called and the hot 5211| | // team size was already reduced, so we check the special flag 5212| 1.35k| if (team->t.t_size_changed == -1) { ------------------ | Branch (5212:11): [True: 0, False: 1.35k] ------------------ 5213| 0| team->t.t_size_changed = 1; 5214| 1.35k| } else { 5215| 1.35k| KMP_CHECK_UPDATE(team->t.t_size_changed, 0); ------------------ | | 2280| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 1, False: 1.35k] | | ------------------ | | 2281| 1.35k| (a) = (b) ------------------ 5216| 1.35k| } 5217| | 5218| | // TODO???: team->t.t_max_active_levels = new_max_active_levels; 5219| 1.35k| kmp_r_sched_t new_sched = new_icvs->sched; 5220| | // set primary thread's schedule as new run-time schedule 5221| 1.35k| KMP_CHECK_UPDATE(team->t.t_sched.sched, new_sched.sched); ------------------ | | 2280| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 1.35k] | | ------------------ | | 2281| 1.35k| (a) = (b) ------------------ 5222| | 5223| 1.35k| __kmp_reinitialize_team(team, new_icvs, 5224| 1.35k| root->r.r_uber_thread->th.th_ident); 5225| | 5226| 1.35k| KF_TRACE(10, ("__kmp_allocate_team2: T#%d, this_thread=%p team=%p\n", 0, 5227| 1.35k| team->t.t_threads[0], team)); 5228| 1.35k| __kmp_push_current_task_to_thread(team->t.t_threads[0], team, 0); 5229| | 5230| 1.35k|#if KMP_AFFINITY_SUPPORTED 5231| 1.35k| if ((team->t.t_size_changed == 0) && ------------------ | Branch (5231:11): [True: 1.35k, False: 0] ------------------ 5232| 1.35k| (team->t.t_proc_bind == new_proc_bind)) { ------------------ | Branch (5232:11): [True: 1.35k, False: 0] ------------------ 5233| 1.35k| if (new_proc_bind == proc_bind_spread) { ------------------ | Branch (5233:13): [True: 0, False: 1.35k] ------------------ 5234| 0| if (do_place_partition) { ------------------ | Branch (5234:15): [True: 0, False: 0] ------------------ 5235| | // add flag to update only master for spread 5236| 0| __kmp_partition_places(team, 1); 5237| 0| } 5238| 0| } 5239| 1.35k| KA_TRACE(200, ("__kmp_allocate_team: reusing hot team #%d bindings: " 5240| 1.35k| "proc_bind = %d, partition = [%d,%d]\n", 5241| 1.35k| team->t.t_id, new_proc_bind, team->t.t_first_place, 5242| 1.35k| team->t.t_last_place)); 5243| 1.35k| } else { 5244| 0| if (do_place_partition) { ------------------ | Branch (5244:13): [True: 0, False: 0] ------------------ 5245| 0| KMP_CHECK_UPDATE(team->t.t_proc_bind, new_proc_bind); ------------------ | | 2280| 0| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 0] | | ------------------ | | 2281| 0| (a) = (b) ------------------ 5246| 0| __kmp_partition_places(team); 5247| 0| } 5248| 0| } 5249| |#else 5250| | KMP_CHECK_UPDATE(team->t.t_proc_bind, new_proc_bind); 5251| |#endif /* KMP_AFFINITY_SUPPORTED */ 5252| 1.35k| } else if (team->t.t_nproc > new_nproc) { ------------------ | Branch (5252:16): [True: 0, False: 1] ------------------ 5253| 0| KA_TRACE(20, 5254| 0| ("__kmp_allocate_team: decreasing hot team thread count to %d\n", 5255| 0| new_nproc)); 5256| | 5257| 0| team->t.t_size_changed = 1; 5258| 0| if (__kmp_barrier_release_pattern[bs_forkjoin_barrier] == bp_dist_bar) { ------------------ | Branch (5258:11): [True: 0, False: 0] ------------------ 5259| | // Barrier size already reduced earlier in this function 5260| | // Activate team threads via th_used_in_team 5261| 0| __kmp_add_threads_to_team(team, new_nproc); 5262| 0| } 5263| 0|#if KMP_NESTED_HOT_TEAMS 5264| 0| if (__kmp_hot_teams_mode == 0) { ------------------ | Branch (5264:11): [True: 0, False: 0] ------------------ 5265| | // AC: saved number of threads should correspond to team's value in this 5266| | // mode, can be bigger in mode 1, when hot team has threads in reserve 5267| 0| KMP_DEBUG_ASSERT(hot_teams[level].hot_team_nth == team->t.t_nproc); 5268| 0| hot_teams[level].hot_team_nth = new_nproc; 5269| 0|#endif // KMP_NESTED_HOT_TEAMS 5270| | /* release the extra threads we don't need any more */ 5271| 0| for (f = new_nproc; f < team->t.t_nproc; f++) { ------------------ | Branch (5271:29): [True: 0, False: 0] ------------------ 5272| 0| KMP_DEBUG_ASSERT(team->t.t_threads[f]); 5273| 0| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (5273:15): [True: 0, False: 0] ------------------ 5274| | // When decreasing team size, threads no longer in the team should 5275| | // unref task team. 5276| 0| team->t.t_threads[f]->th.th_task_team = NULL; 5277| 0| } 5278| 0| __kmp_free_thread(team->t.t_threads[f]); 5279| 0| team->t.t_threads[f] = NULL; 5280| 0| } 5281| 0|#if KMP_NESTED_HOT_TEAMS 5282| 0| } // (__kmp_hot_teams_mode == 0) 5283| 0| else { 5284| | // When keeping extra threads in team, switch threads to wait on own 5285| | // b_go flag 5286| 0| for (f = new_nproc; f < team->t.t_nproc; ++f) { ------------------ | Branch (5286:29): [True: 0, False: 0] ------------------ 5287| 0| KMP_DEBUG_ASSERT(team->t.t_threads[f]); 5288| 0| kmp_balign_t *balign = team->t.t_threads[f]->th.th_bar; 5289| 0| for (int b = 0; b < bs_last_barrier; ++b) { ------------------ | Branch (5289:27): [True: 0, False: 0] ------------------ 5290| 0| if (balign[b].bb.wait_flag == KMP_BARRIER_PARENT_FLAG) { ------------------ | | 2039| 0| 2 // Special state; worker waiting on parent's b_go flag in release ------------------ | Branch (5290:17): [True: 0, False: 0] ------------------ 5291| 0| balign[b].bb.wait_flag = KMP_BARRIER_SWITCH_TO_OWN_FLAG; ------------------ | | 2041| 0| 3 // Special state; tells worker to shift from parent to own b_go ------------------ 5292| 0| } 5293| 0| KMP_CHECK_UPDATE(balign[b].bb.leaf_kids, 0); ------------------ | | 2280| 0| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 0] | | ------------------ | | 2281| 0| (a) = (b) ------------------ 5294| 0| } 5295| 0| } 5296| 0| } 5297| 0|#endif // KMP_NESTED_HOT_TEAMS 5298| 0| team->t.t_nproc = new_nproc; 5299| | // TODO???: team->t.t_max_active_levels = new_max_active_levels; 5300| 0| KMP_CHECK_UPDATE(team->t.t_sched.sched, new_icvs->sched.sched); ------------------ | | 2280| 0| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 0] | | ------------------ | | 2281| 0| (a) = (b) ------------------ 5301| 0| __kmp_reinitialize_team(team, new_icvs, 5302| 0| root->r.r_uber_thread->th.th_ident); 5303| | 5304| | // Update remaining threads 5305| 0| for (f = 0; f < new_nproc; ++f) { ------------------ | Branch (5305:19): [True: 0, False: 0] ------------------ 5306| 0| team->t.t_threads[f]->th.th_team_nproc = new_nproc; 5307| 0| } 5308| | 5309| | // restore the current task state of the primary thread: should be the 5310| | // implicit task 5311| 0| KF_TRACE(10, ("__kmp_allocate_team: T#%d, this_thread=%p team=%p\n", 0, 5312| 0| team->t.t_threads[0], team)); 5313| | 5314| 0| __kmp_push_current_task_to_thread(team->t.t_threads[0], team, 0); 5315| | 5316| |#ifdef KMP_DEBUG 5317| | for (f = 0; f < team->t.t_nproc; f++) { 5318| | KMP_DEBUG_ASSERT(team->t.t_threads[f] && 5319| | team->t.t_threads[f]->th.th_team_nproc == 5320| | team->t.t_nproc); 5321| | } 5322| |#endif 5323| | 5324| 0| if (do_place_partition) { ------------------ | Branch (5324:11): [True: 0, False: 0] ------------------ 5325| 0| KMP_CHECK_UPDATE(team->t.t_proc_bind, new_proc_bind); ------------------ | | 2280| 0| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 0] | | ------------------ | | 2281| 0| (a) = (b) ------------------ 5326| 0|#if KMP_AFFINITY_SUPPORTED 5327| 0| __kmp_partition_places(team); 5328| 0|#endif 5329| 0| } 5330| 1| } else { // team->t.t_nproc < new_nproc 5331| | 5332| 1| KA_TRACE(20, 5333| 1| ("__kmp_allocate_team: increasing hot team thread count to %d\n", 5334| 1| new_nproc)); 5335| 1| int old_nproc = team->t.t_nproc; // save old value and use to update only 5336| 1| team->t.t_size_changed = 1; 5337| | 5338| 1|#if KMP_NESTED_HOT_TEAMS 5339| 1| int avail_threads = hot_teams[level].hot_team_nth; 5340| 1| if (new_nproc < avail_threads) ------------------ | Branch (5340:11): [True: 0, False: 1] ------------------ 5341| 0| avail_threads = new_nproc; 5342| 1| kmp_info_t **other_threads = team->t.t_threads; 5343| 1| for (f = team->t.t_nproc; f < avail_threads; ++f) { ------------------ | Branch (5343:33): [True: 0, False: 1] ------------------ 5344| | // Adjust barrier data of reserved threads (if any) of the team 5345| | // Other data will be set in __kmp_initialize_info() below. 5346| 0| int b; 5347| 0| kmp_balign_t *balign = other_threads[f]->th.th_bar; 5348| 0| for (b = 0; b < bs_last_barrier; ++b) { ------------------ | Branch (5348:21): [True: 0, False: 0] ------------------ 5349| 0| balign[b].bb.b_arrived = team->t.t_bar[b].b_arrived; 5350| 0| KMP_DEBUG_ASSERT(balign[b].bb.wait_flag != KMP_BARRIER_PARENT_FLAG); 5351| |#if USE_DEBUGGER 5352| | balign[b].bb.b_worker_arrived = team->t.t_bar[b].b_team_arrived; 5353| |#endif 5354| 0| } 5355| 0| } 5356| 1| if (hot_teams[level].hot_team_nth >= new_nproc) { ------------------ | Branch (5356:11): [True: 0, False: 1] ------------------ 5357| | // we have all needed threads in reserve, no need to allocate any 5358| | // this only possible in mode 1, cannot have reserved threads in mode 0 5359| 0| KMP_DEBUG_ASSERT(__kmp_hot_teams_mode == 1); 5360| 0| team->t.t_nproc = new_nproc; // just get reserved threads involved 5361| 1| } else { 5362| | // We may have some threads in reserve, but not enough; 5363| | // get reserved threads involved if any. 5364| 1| team->t.t_nproc = hot_teams[level].hot_team_nth; 5365| 1| hot_teams[level].hot_team_nth = new_nproc; // adjust hot team max size 5366| 1|#endif // KMP_NESTED_HOT_TEAMS 5367| 1| if (team->t.t_max_nproc < new_nproc) { ------------------ | Branch (5367:13): [True: 0, False: 1] ------------------ 5368| | /* reallocate larger arrays */ 5369| 0| __kmp_reallocate_team_arrays(team, new_nproc); 5370| 0| __kmp_reinitialize_team(team, new_icvs, NULL); 5371| 0| } 5372| | 5373| 1|#if (KMP_OS_LINUX || KMP_OS_FREEBSD) && KMP_AFFINITY_SUPPORTED 5374| | /* Temporarily set full mask for primary thread before creation of 5375| | workers. The reason is that workers inherit the affinity from the 5376| | primary thread, so if a lot of workers are created on the single 5377| | core quickly, they don't get a chance to set their own affinity for 5378| | a long time. */ 5379| 1| kmp_affinity_raii_t new_temp_affinity{__kmp_affin_fullMask}; 5380| 1|#endif 5381| | 5382| | /* allocate new threads for the hot team */ 5383| 32| for (f = team->t.t_nproc; f < new_nproc; f++) { ------------------ | Branch (5383:35): [True: 31, False: 1] ------------------ 5384| 31| kmp_info_t *new_worker = __kmp_allocate_thread(root, team, f); 5385| 31| KMP_DEBUG_ASSERT(new_worker); 5386| 31| team->t.t_threads[f] = new_worker; 5387| | 5388| 31| KA_TRACE(20, 5389| 31| ("__kmp_allocate_team: team %d init T#%d arrived: " 5390| 31| "join=%llu, plain=%llu\n", 5391| 31| team->t.t_id, __kmp_gtid_from_tid(f, team), team->t.t_id, f, 5392| 31| team->t.t_bar[bs_forkjoin_barrier].b_arrived, 5393| 31| team->t.t_bar[bs_plain_barrier].b_arrived)); 5394| | 5395| 31| { // Initialize barrier data for new threads. 5396| 31| int b; 5397| 31| kmp_balign_t *balign = new_worker->th.th_bar; 5398| 124| for (b = 0; b < bs_last_barrier; ++b) { ------------------ | Branch (5398:25): [True: 93, False: 31] ------------------ 5399| 93| balign[b].bb.b_arrived = team->t.t_bar[b].b_arrived; 5400| 93| KMP_DEBUG_ASSERT(balign[b].bb.wait_flag != 5401| 93| KMP_BARRIER_PARENT_FLAG); 5402| |#if USE_DEBUGGER 5403| | balign[b].bb.b_worker_arrived = team->t.t_bar[b].b_team_arrived; 5404| |#endif 5405| 93| } 5406| 31| } 5407| 31| } 5408| | 5409| 1|#if (KMP_OS_LINUX || KMP_OS_FREEBSD) && KMP_AFFINITY_SUPPORTED 5410| | /* Restore initial primary thread's affinity mask */ 5411| 1| new_temp_affinity.restore(); 5412| 1|#endif 5413| 1|#if KMP_NESTED_HOT_TEAMS 5414| 1| } // end of check of t_nproc vs. new_nproc vs. hot_team_nth 5415| 1|#endif // KMP_NESTED_HOT_TEAMS 5416| 1| if (__kmp_barrier_release_pattern[bs_forkjoin_barrier] == bp_dist_bar) { ------------------ | Branch (5416:11): [True: 0, False: 1] ------------------ 5417| | // Barrier size already increased earlier in this function 5418| | // Activate team threads via th_used_in_team 5419| 0| __kmp_add_threads_to_team(team, new_nproc); 5420| 0| } 5421| | /* make sure everyone is syncronized */ 5422| | // new threads below 5423| 1| __kmp_initialize_team(team, new_nproc, new_icvs, 5424| 1| root->r.r_uber_thread->th.th_ident); 5425| | 5426| | /* reinitialize the threads */ 5427| 1| KMP_DEBUG_ASSERT(team->t.t_nproc == new_nproc); 5428| 33| for (f = 0; f < team->t.t_nproc; ++f) ------------------ | Branch (5428:19): [True: 32, False: 1] ------------------ 5429| 32| __kmp_initialize_info(team->t.t_threads[f], team, f, 5430| 32| __kmp_gtid_from_tid(f, team)); 5431| | 5432| | // set th_task_state for new threads in hot team with older thread's state 5433| 1| kmp_uint8 old_state = team->t.t_threads[old_nproc - 1]->th.th_task_state; 5434| 32| for (f = old_nproc; f < team->t.t_nproc; ++f) ------------------ | Branch (5434:27): [True: 31, False: 1] ------------------ 5435| 31| team->t.t_threads[f]->th.th_task_state = old_state; 5436| | 5437| |#ifdef KMP_DEBUG 5438| | for (f = 0; f < team->t.t_nproc; ++f) { 5439| | KMP_DEBUG_ASSERT(team->t.t_threads[f] && 5440| | team->t.t_threads[f]->th.th_team_nproc == 5441| | team->t.t_nproc); 5442| | } 5443| |#endif 5444| | 5445| 1| if (do_place_partition) { ------------------ | Branch (5445:11): [True: 1, False: 0] ------------------ 5446| 1| KMP_CHECK_UPDATE(team->t.t_proc_bind, new_proc_bind); ------------------ | | 2280| 1| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 1] | | ------------------ | | 2281| 1| (a) = (b) ------------------ 5447| 1|#if KMP_AFFINITY_SUPPORTED 5448| 1| __kmp_partition_places(team); 5449| 1|#endif 5450| 1| } 5451| 1| } // Check changes in number of threads 5452| | 5453| 1.35k| kmp_info_t *master = team->t.t_threads[0]; 5454| 1.35k| if (master->th.th_teams_microtask) { ------------------ | Branch (5454:9): [True: 0, False: 1.35k] ------------------ 5455| 0| for (f = 1; f < new_nproc; ++f) { ------------------ | Branch (5455:19): [True: 0, False: 0] ------------------ 5456| | // propagate teams construct specific info to workers 5457| 0| kmp_info_t *thr = team->t.t_threads[f]; 5458| 0| thr->th.th_teams_microtask = master->th.th_teams_microtask; 5459| 0| thr->th.th_teams_level = master->th.th_teams_level; 5460| 0| thr->th.th_teams_size = master->th.th_teams_size; 5461| 0| } 5462| 0| } 5463| 1.35k|#if KMP_NESTED_HOT_TEAMS 5464| 1.35k| if (level) { ------------------ | Branch (5464:9): [True: 0, False: 1.35k] ------------------ 5465| | // Sync barrier state for nested hot teams, not needed for outermost hot 5466| | // team. 5467| 0| for (f = 1; f < new_nproc; ++f) { ------------------ | Branch (5467:19): [True: 0, False: 0] ------------------ 5468| 0| kmp_info_t *thr = team->t.t_threads[f]; 5469| 0| int b; 5470| 0| kmp_balign_t *balign = thr->th.th_bar; 5471| 0| for (b = 0; b < bs_last_barrier; ++b) { ------------------ | Branch (5471:21): [True: 0, False: 0] ------------------ 5472| 0| balign[b].bb.b_arrived = team->t.t_bar[b].b_arrived; 5473| 0| KMP_DEBUG_ASSERT(balign[b].bb.wait_flag != KMP_BARRIER_PARENT_FLAG); 5474| |#if USE_DEBUGGER 5475| | balign[b].bb.b_worker_arrived = team->t.t_bar[b].b_team_arrived; 5476| |#endif 5477| 0| } 5478| 0| } 5479| 0| } 5480| 1.35k|#endif // KMP_NESTED_HOT_TEAMS 5481| | 5482| | /* reallocate space for arguments if necessary */ 5483| 1.35k| __kmp_alloc_argv_entries(argc, team, TRUE); ------------------ | | 1280| 1.35k|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1.35k|#define FALSE 0 | | ------------------ ------------------ 5484| 1.35k| KMP_CHECK_UPDATE(team->t.t_argc, argc); ------------------ | | 2280| 1.35k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 69, False: 1.28k] | | ------------------ | | 2281| 1.35k| (a) = (b) ------------------ 5485| | // The hot team re-uses the previous task team, 5486| | // if untouched during the previous release->gather phase. 5487| | 5488| 1.35k| KF_TRACE(10, (" hot_team = %p\n", team)); 5489| | 5490| |#if KMP_DEBUG 5491| | if (__kmp_tasking_mode != tskm_immediate_exec) { 5492| | KA_TRACE(20, ("__kmp_allocate_team: hot team task_team[0] = %p " 5493| | "task_team[1] = %p after reinit\n", 5494| | team->t.t_task_team[0], team->t.t_task_team[1])); 5495| | } 5496| |#endif 5497| | 5498| 1.35k|#if OMPT_SUPPORT 5499| 1.35k| __ompt_team_assign_id(team, ompt_parallel_data); 5500| 1.35k|#endif 5501| | 5502| 1.35k| KMP_MB(); 5503| | 5504| 1.35k| return team; 5505| 1.35k| } 5506| | 5507| | /* next, let's try to take one from the team pool */ 5508| 34| KMP_MB(); 5509| 34| for (team = CCAST(kmp_team_t *, __kmp_team_pool); (team);) { ------------------ | | 229| 34|#define CCAST(type, var) const_cast(var) ------------------ | Branch (5509:53): [True: 0, False: 34] ------------------ 5510| | /* TODO: consider resizing undersized teams instead of reaping them, now 5511| | that we have a resizing mechanism */ 5512| 0| if (team->t.t_max_nproc >= max_nproc) { ------------------ | Branch (5512:9): [True: 0, False: 0] ------------------ 5513| | /* take this team from the team pool */ 5514| 0| __kmp_team_pool = team->t.t_next_pool; 5515| | 5516| 0| if (max_nproc > 1 && ------------------ | Branch (5516:11): [True: 0, False: 0] ------------------ 5517| 0| __kmp_barrier_gather_pattern[bs_forkjoin_barrier] == bp_dist_bar) { ------------------ | Branch (5517:11): [True: 0, False: 0] ------------------ 5518| 0| if (!team->t.b) { // Allocate barrier structure ------------------ | Branch (5518:13): [True: 0, False: 0] ------------------ 5519| 0| team->t.b = distributedBarrier::allocate(__kmp_dflt_team_nth_ub); 5520| 0| } 5521| 0| } 5522| | 5523| | /* setup the team for fresh use */ 5524| 0| __kmp_initialize_team(team, new_nproc, new_icvs, NULL); 5525| | 5526| 0| KA_TRACE(20, ("__kmp_allocate_team: setting task_team[0] %p and " 5527| 0| "task_team[1] %p to NULL\n", 5528| 0| &team->t.t_task_team[0], &team->t.t_task_team[1])); 5529| 0| team->t.t_task_team[0] = NULL; 5530| 0| team->t.t_task_team[1] = NULL; 5531| | 5532| | /* reallocate space for arguments if necessary */ 5533| 0| __kmp_alloc_argv_entries(argc, team, TRUE); ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 5534| 0| KMP_CHECK_UPDATE(team->t.t_argc, argc); ------------------ | | 2280| 0| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 0] | | ------------------ | | 2281| 0| (a) = (b) ------------------ 5535| | 5536| 0| KA_TRACE( 5537| 0| 20, ("__kmp_allocate_team: team %d init arrived: join=%u, plain=%u\n", 5538| 0| team->t.t_id, KMP_INIT_BARRIER_STATE, KMP_INIT_BARRIER_STATE)); 5539| 0| { // Initialize barrier data. 5540| 0| int b; 5541| 0| for (b = 0; b < bs_last_barrier; ++b) { ------------------ | Branch (5541:21): [True: 0, False: 0] ------------------ 5542| 0| team->t.t_bar[b].b_arrived = KMP_INIT_BARRIER_STATE; ------------------ | | 2018| 0|#define KMP_INIT_BARRIER_STATE 0 /* should probably start from zero */ ------------------ 5543| |#if USE_DEBUGGER 5544| | team->t.t_bar[b].b_master_arrived = 0; 5545| | team->t.t_bar[b].b_team_arrived = 0; 5546| |#endif 5547| 0| } 5548| 0| } 5549| | 5550| 0| team->t.t_proc_bind = new_proc_bind; 5551| | 5552| 0| KA_TRACE(20, ("__kmp_allocate_team: using team from pool %d.\n", 5553| 0| team->t.t_id)); 5554| | 5555| 0|#if OMPT_SUPPORT 5556| 0| __ompt_team_assign_id(team, ompt_parallel_data); 5557| 0|#endif 5558| | 5559| 0| KMP_MB(); 5560| | 5561| 0| return team; 5562| 0| } 5563| | 5564| | /* reap team if it is too small, then loop back and check the next one */ 5565| | // not sure if this is wise, but, will be redone during the hot-teams 5566| | // rewrite. 5567| | /* TODO: Use technique to find the right size hot-team, don't reap them */ 5568| 0| team = __kmp_reap_team(team); 5569| 0| __kmp_team_pool = team; 5570| 0| } 5571| | 5572| | /* nothing available in the pool, no matter, make a new team! */ 5573| 34| KMP_MB(); 5574| 34| team = (kmp_team_t *)__kmp_allocate(sizeof(kmp_team_t)); ------------------ | | 3617| 34|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 5575| | 5576| | /* and set it up */ 5577| 34| team->t.t_max_nproc = max_nproc; 5578| 34| if (max_nproc > 1 && ------------------ | Branch (5578:7): [True: 1, False: 33] ------------------ 5579| 1| __kmp_barrier_gather_pattern[bs_forkjoin_barrier] == bp_dist_bar) { ------------------ | Branch (5579:7): [True: 0, False: 1] ------------------ 5580| | // Allocate barrier structure 5581| 0| team->t.b = distributedBarrier::allocate(__kmp_dflt_team_nth_ub); 5582| 0| } 5583| | 5584| | /* NOTE well, for some reason allocating one big buffer and dividing it up 5585| | seems to really hurt performance a lot on the P4, so, let's not use this */ 5586| 34| __kmp_allocate_team_arrays(team, max_nproc); 5587| | 5588| 34| KA_TRACE(20, ("__kmp_allocate_team: making a new team\n")); 5589| 34| __kmp_initialize_team(team, new_nproc, new_icvs, NULL); 5590| | 5591| 34| KA_TRACE(20, ("__kmp_allocate_team: setting task_team[0] %p and task_team[1] " 5592| 34| "%p to NULL\n", 5593| 34| &team->t.t_task_team[0], &team->t.t_task_team[1])); 5594| 34| team->t.t_task_team[0] = NULL; // to be removed, as __kmp_allocate zeroes 5595| | // memory, no need to duplicate 5596| 34| team->t.t_task_team[1] = NULL; // to be removed, as __kmp_allocate zeroes 5597| | // memory, no need to duplicate 5598| | 5599| 34| if (__kmp_storage_map) { ------------------ | Branch (5599:7): [True: 0, False: 34] ------------------ 5600| 0| __kmp_print_team_storage_map("team", team, team->t.t_id, new_nproc); 5601| 0| } 5602| | 5603| | /* allocate space for arguments */ 5604| 34| __kmp_alloc_argv_entries(argc, team, FALSE); ------------------ | | 1279| 34|#define FALSE 0 ------------------ 5605| 34| team->t.t_argc = argc; 5606| | 5607| 34| KA_TRACE(20, 5608| 34| ("__kmp_allocate_team: team %d init arrived: join=%u, plain=%u\n", 5609| 34| team->t.t_id, KMP_INIT_BARRIER_STATE, KMP_INIT_BARRIER_STATE)); 5610| 34| { // Initialize barrier data. 5611| 34| int b; 5612| 136| for (b = 0; b < bs_last_barrier; ++b) { ------------------ | Branch (5612:17): [True: 102, False: 34] ------------------ 5613| 102| team->t.t_bar[b].b_arrived = KMP_INIT_BARRIER_STATE; ------------------ | | 2018| 102|#define KMP_INIT_BARRIER_STATE 0 /* should probably start from zero */ ------------------ 5614| |#if USE_DEBUGGER 5615| | team->t.t_bar[b].b_master_arrived = 0; 5616| | team->t.t_bar[b].b_team_arrived = 0; 5617| |#endif 5618| 102| } 5619| 34| } 5620| | 5621| 34| team->t.t_proc_bind = new_proc_bind; 5622| | 5623| 34|#if OMPT_SUPPORT 5624| 34| __ompt_team_assign_id(team, ompt_parallel_data); 5625| 34| team->t.ompt_serialized_team_info = NULL; 5626| 34|#endif 5627| | 5628| 34| KMP_MB(); 5629| | 5630| 34| KA_TRACE(20, ("__kmp_allocate_team: done creating a new team %d.\n", 5631| 34| team->t.t_id)); 5632| | 5633| 34| return team; 5634| 34|} __kmp_free_team: 5642| 1.35k| kmp_team_t *team USE_NESTED_HOT_ARG(kmp_info_t *master)) { 5643| 1.35k| int f; 5644| 1.35k| KA_TRACE(20, ("__kmp_free_team: T#%d freeing team %d\n", __kmp_get_gtid(), 5645| 1.35k| team->t.t_id)); 5646| | 5647| | /* verify state */ 5648| 1.35k| KMP_DEBUG_ASSERT(root); 5649| 1.35k| KMP_DEBUG_ASSERT(team); 5650| 1.35k| KMP_DEBUG_ASSERT(team->t.t_nproc <= team->t.t_max_nproc); 5651| 1.35k| KMP_DEBUG_ASSERT(team->t.t_threads); 5652| | 5653| 1.35k| int use_hot_team = team == root->r.r_hot_team; 5654| 1.35k|#if KMP_NESTED_HOT_TEAMS 5655| 1.35k| int level; 5656| 1.35k| if (master) { ------------------ | Branch (5656:7): [True: 1.35k, False: 2] ------------------ 5657| 1.35k| level = team->t.t_active_level - 1; 5658| 1.35k| if (master->th.th_teams_microtask) { // in teams construct? ------------------ | Branch (5658:9): [True: 0, False: 1.35k] ------------------ 5659| 0| if (master->th.th_teams_size.nteams > 1) { ------------------ | Branch (5659:11): [True: 0, False: 0] ------------------ 5660| 0| ++level; // level was not increased in teams construct for 5661| | // team_of_masters 5662| 0| } 5663| 0| if (team->t.t_pkfn != (microtask_t)__kmp_teams_master && ------------------ | Branch (5663:11): [True: 0, False: 0] ------------------ 5664| 0| master->th.th_teams_level == team->t.t_level) { ------------------ | Branch (5664:11): [True: 0, False: 0] ------------------ 5665| 0| ++level; // level was not increased in teams construct for 5666| | // team_of_workers before the parallel 5667| 0| } // team->t.t_level will be increased inside parallel 5668| 0| } 5669| |#if KMP_DEBUG 5670| | kmp_hot_team_ptr_t *hot_teams = master->th.th_hot_teams; 5671| |#endif 5672| 1.35k| if (level < __kmp_hot_teams_max_level) { ------------------ | Branch (5672:9): [True: 1.35k, False: 0] ------------------ 5673| 1.35k| KMP_DEBUG_ASSERT(team == hot_teams[level].hot_team); 5674| 1.35k| use_hot_team = 1; 5675| 1.35k| } 5676| 1.35k| } 5677| 1.35k|#endif // KMP_NESTED_HOT_TEAMS 5678| | 5679| | /* team is done working */ 5680| 1.35k| TCW_SYNC_PTR(team->t.t_pkfn, ------------------ | | 1159| 1.35k|#define TCW_SYNC_PTR(a, b) TCW_SYNC_8((a), (b)) | | ------------------ | | | | 1141| 1.35k|#define TCW_SYNC_8(a, b) (a) = (b) | | ------------------ ------------------ 5681| 1.35k| NULL); // Important for Debugging Support Library. 5682| |#if KMP_OS_WINDOWS 5683| | team->t.t_copyin_counter = 0; // init counter for possible reuse 5684| |#endif 5685| | // Do not reset pointer to parent team to NULL for hot teams. 5686| | 5687| | /* if we are non-hot team, release our threads */ 5688| 1.35k| if (!use_hot_team) { ------------------ | Branch (5688:7): [True: 2, False: 1.35k] ------------------ 5689| 2| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (5689:9): [True: 2, False: 0] ------------------ 5690| | // Wait for threads to reach reapable state 5691| 33| for (f = 1; f < team->t.t_nproc; ++f) { ------------------ | Branch (5691:19): [True: 31, False: 2] ------------------ 5692| 31| KMP_DEBUG_ASSERT(team->t.t_threads[f]); 5693| 31| kmp_info_t *th = team->t.t_threads[f]; 5694| 31| volatile kmp_uint32 *state = &th->th.th_reap_state; 5695| 31| while (*state != KMP_SAFE_TO_REAP) { ------------------ | | 2047| 31|#define KMP_SAFE_TO_REAP 1 // Thread th_reap_state: safe to reap (not tasking) ------------------ | Branch (5695:16): [True: 0, False: 31] ------------------ 5696| |#if KMP_OS_WINDOWS 5697| | // On Windows a thread can be killed at any time, check this 5698| | DWORD ecode; 5699| | if (!__kmp_is_thread_alive(th, &ecode)) { 5700| | *state = KMP_SAFE_TO_REAP; // reset the flag for dead thread 5701| | break; 5702| | } 5703| |#endif 5704| | // first check if thread is sleeping 5705| 0| kmp_flag_64<> fl(&th->th.th_bar[bs_forkjoin_barrier].bb.b_go, th); 5706| 0| if (fl.is_sleeping()) ------------------ | Branch (5706:15): [True: 0, False: 0] ------------------ 5707| 0| fl.resume(__kmp_gtid_from_thread(th)); 5708| 0| KMP_CPU_PAUSE(); ------------------ | | 1498| 0|#define KMP_CPU_PAUSE() __kmp_x86_pause() ------------------ 5709| 0| } 5710| 31| } 5711| | 5712| | // Delete task teams 5713| 2| int tt_idx; 5714| 6| for (tt_idx = 0; tt_idx < 2; ++tt_idx) { ------------------ | Branch (5714:24): [True: 4, False: 2] ------------------ 5715| 4| kmp_task_team_t *task_team = team->t.t_task_team[tt_idx]; 5716| 4| if (task_team != NULL) { ------------------ | Branch (5716:13): [True: 2, False: 2] ------------------ 5717| 66| for (f = 0; f < team->t.t_nproc; ++f) { // threads unref task teams ------------------ | Branch (5717:23): [True: 64, False: 2] ------------------ 5718| 64| KMP_DEBUG_ASSERT(team->t.t_threads[f]); 5719| 64| team->t.t_threads[f]->th.th_task_team = NULL; 5720| 64| } 5721| 2| KA_TRACE( 5722| 2| 20, 5723| 2| ("__kmp_free_team: T#%d deactivating task_team %p on team %d\n", 5724| 2| __kmp_get_gtid(), task_team, team->t.t_id)); 5725| 2|#if KMP_NESTED_HOT_TEAMS 5726| 2| __kmp_free_task_team(master, task_team); 5727| 2|#endif 5728| 2| team->t.t_task_team[tt_idx] = NULL; 5729| 2| } 5730| 4| } 5731| 2| } 5732| | 5733| | // Reset pointer to parent team only for non-hot teams. 5734| 2| team->t.t_parent = NULL; 5735| 2| team->t.t_level = 0; 5736| 2| team->t.t_active_level = 0; 5737| | 5738| | /* free the worker threads */ 5739| 33| for (f = 1; f < team->t.t_nproc; ++f) { ------------------ | Branch (5739:17): [True: 31, False: 2] ------------------ 5740| 31| KMP_DEBUG_ASSERT(team->t.t_threads[f]); 5741| 31| if (__kmp_barrier_gather_pattern[bs_forkjoin_barrier] == bp_dist_bar) { ------------------ | Branch (5741:11): [True: 0, False: 31] ------------------ 5742| 0| KMP_COMPARE_AND_STORE_ACQ32(&(team->t.t_threads[f]->th.th_used_in_team), ------------------ | | 807| 0| __sync_bool_compare_and_swap((volatile kmp_uint32 *)(p), (kmp_uint32)(cv), \ | | 808| 0| (kmp_uint32)(sv)) ------------------ 5743| 0| 1, 2); 5744| 0| } 5745| 31| __kmp_free_thread(team->t.t_threads[f]); 5746| 31| } 5747| | 5748| 2| if (__kmp_barrier_gather_pattern[bs_forkjoin_barrier] == bp_dist_bar) { ------------------ | Branch (5748:9): [True: 0, False: 2] ------------------ 5749| 0| if (team->t.b) { ------------------ | Branch (5749:11): [True: 0, False: 0] ------------------ 5750| | // wake up thread at old location 5751| 0| team->t.b->go_release(); 5752| 0| if (__kmp_dflt_blocktime != KMP_MAX_BLOCKTIME) { ------------------ | | 1184| 0| (INT_MAX) /* Must be this for "infinite" setting the work */ ------------------ | Branch (5752:13): [True: 0, False: 0] ------------------ 5753| 0| for (f = 1; f < team->t.t_nproc; ++f) { ------------------ | Branch (5753:23): [True: 0, False: 0] ------------------ 5754| 0| if (team->t.b->sleep[f].sleep) { ------------------ | Branch (5754:17): [True: 0, False: 0] ------------------ 5755| 0| __kmp_atomic_resume_64( 5756| 0| team->t.t_threads[f]->th.th_info.ds.ds_gtid, 5757| 0| (kmp_atomic_flag_64<> *)NULL); 5758| 0| } 5759| 0| } 5760| 0| } 5761| | // Wait for threads to be removed from team 5762| 0| for (int f = 1; f < team->t.t_nproc; ++f) { ------------------ | Branch (5762:25): [True: 0, False: 0] ------------------ 5763| 0| while (team->t.t_threads[f]->th.th_used_in_team.load() != 0) ------------------ | Branch (5763:18): [True: 0, False: 0] ------------------ 5764| 0| KMP_CPU_PAUSE(); ------------------ | | 1498| 0|#define KMP_CPU_PAUSE() __kmp_x86_pause() ------------------ 5765| 0| } 5766| 0| } 5767| 0| } 5768| | 5769| 33| for (f = 1; f < team->t.t_nproc; ++f) { ------------------ | Branch (5769:17): [True: 31, False: 2] ------------------ 5770| 31| team->t.t_threads[f] = NULL; 5771| 31| } 5772| | 5773| 2| if (team->t.t_max_nproc > 1 && ------------------ | Branch (5773:9): [True: 1, False: 1] ------------------ 5774| 1| __kmp_barrier_gather_pattern[bs_forkjoin_barrier] == bp_dist_bar) { ------------------ | Branch (5774:9): [True: 0, False: 1] ------------------ 5775| 0| distributedBarrier::deallocate(team->t.b); 5776| 0| team->t.b = NULL; 5777| 0| } 5778| | /* put the team back in the team pool */ 5779| | /* TODO limit size of team pool, call reap_team if pool too large */ 5780| 2| team->t.t_next_pool = CCAST(kmp_team_t *, __kmp_team_pool); ------------------ | | 229| 2|#define CCAST(type, var) const_cast(var) ------------------ 5781| 2| __kmp_team_pool = (volatile kmp_team_t *)team; 5782| 1.35k| } else { // Check if team was created for primary threads in teams construct 5783| | // See if first worker is a CG root 5784| 1.35k| KMP_DEBUG_ASSERT(team->t.t_threads[1] && 5785| 1.35k| team->t.t_threads[1]->th.th_cg_roots); 5786| 1.35k| if (team->t.t_threads[1]->th.th_cg_roots->cg_root == team->t.t_threads[1]) { ------------------ | Branch (5786:9): [True: 0, False: 1.35k] ------------------ 5787| | // Clean up the CG root nodes on workers so that this team can be re-used 5788| 0| for (f = 1; f < team->t.t_nproc; ++f) { ------------------ | Branch (5788:19): [True: 0, False: 0] ------------------ 5789| 0| kmp_info_t *thr = team->t.t_threads[f]; 5790| 0| KMP_DEBUG_ASSERT(thr && thr->th.th_cg_roots && 5791| 0| thr->th.th_cg_roots->cg_root == thr); 5792| | // Pop current CG root off list 5793| 0| kmp_cg_root_t *tmp = thr->th.th_cg_roots; 5794| 0| thr->th.th_cg_roots = tmp->up; 5795| 0| KA_TRACE(100, ("__kmp_free_team: Thread %p popping node %p and moving" 5796| 0| " up to node %p. cg_nthreads was %d\n", 5797| 0| thr, tmp, thr->th.th_cg_roots, tmp->cg_nthreads)); 5798| 0| int i = tmp->cg_nthreads--; 5799| 0| if (i == 1) { ------------------ | Branch (5799:13): [True: 0, False: 0] ------------------ 5800| 0| __kmp_free(tmp); // free CG if we are the last thread in it ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 5801| 0| } 5802| | // Restore current task's thread_limit from CG root 5803| 0| if (thr->th.th_cg_roots) ------------------ | Branch (5803:13): [True: 0, False: 0] ------------------ 5804| 0| thr->th.th_current_task->td_icvs.thread_limit = 5805| 0| thr->th.th_cg_roots->cg_thread_limit; 5806| 0| } 5807| 0| } 5808| 1.35k| } 5809| | 5810| 1.35k| KMP_MB(); 5811| 1.35k|} __kmp_reap_team: 5814| 34|kmp_team_t *__kmp_reap_team(kmp_team_t *team) { 5815| 34| kmp_team_t *next_pool = team->t.t_next_pool; 5816| | 5817| 34| KMP_DEBUG_ASSERT(team); 5818| 34| KMP_DEBUG_ASSERT(team->t.t_dispatch); 5819| 34| KMP_DEBUG_ASSERT(team->t.t_disp_buffer); 5820| 34| KMP_DEBUG_ASSERT(team->t.t_threads); 5821| 34| KMP_DEBUG_ASSERT(team->t.t_argv); 5822| | 5823| | /* TODO clean the threads that are a part of this? */ 5824| | 5825| | /* free stuff */ 5826| 34| __kmp_free_team_arrays(team); 5827| 34| if (team->t.t_argv != &team->t.t_inline_argv[0]) ------------------ | Branch (5827:7): [True: 0, False: 34] ------------------ 5828| 0| __kmp_free((void *)team->t.t_argv); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 5829| 34| __kmp_free(team); ------------------ | | 3619| 34|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 5830| | 5831| 34| KMP_MB(); 5832| 34| return next_pool; 5833| 34|} __kmp_free_thread: 5860| 31|void __kmp_free_thread(kmp_info_t *this_th) { 5861| 31| int gtid; 5862| 31| kmp_info_t **scan; 5863| | 5864| 31| KA_TRACE(20, ("__kmp_free_thread: T#%d putting T#%d back on free pool.\n", 5865| 31| __kmp_get_gtid(), this_th->th.th_info.ds.ds_gtid)); 5866| | 5867| 31| KMP_DEBUG_ASSERT(this_th); 5868| | 5869| | // When moving thread to pool, switch thread to wait on own b_go flag, and 5870| | // uninitialized (NULL team). 5871| 31| int b; 5872| 31| kmp_balign_t *balign = this_th->th.th_bar; 5873| 124| for (b = 0; b < bs_last_barrier; ++b) { ------------------ | Branch (5873:15): [True: 93, False: 31] ------------------ 5874| 93| if (balign[b].bb.wait_flag == KMP_BARRIER_PARENT_FLAG) ------------------ | | 2039| 93| 2 // Special state; worker waiting on parent's b_go flag in release ------------------ | Branch (5874:9): [True: 0, False: 93] ------------------ 5875| 0| balign[b].bb.wait_flag = KMP_BARRIER_SWITCH_TO_OWN_FLAG; ------------------ | | 2041| 0| 3 // Special state; tells worker to shift from parent to own b_go ------------------ 5876| 93| balign[b].bb.team = NULL; 5877| 93| balign[b].bb.leaf_kids = 0; 5878| 93| } 5879| 31| this_th->th.th_task_state = 0; 5880| 31| this_th->th.th_reap_state = KMP_SAFE_TO_REAP; ------------------ | | 2047| 31|#define KMP_SAFE_TO_REAP 1 // Thread th_reap_state: safe to reap (not tasking) ------------------ 5881| | 5882| | /* put thread back on the free pool */ 5883| 31| TCW_PTR(this_th->th.th_team, NULL); ------------------ | | 1157| 31|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 31|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 5884| 31| TCW_PTR(this_th->th.th_root, NULL); ------------------ | | 1157| 31|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 31|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 5885| 31| TCW_PTR(this_th->th.th_dispatch, NULL); /* NOT NEEDED */ ------------------ | | 1157| 31|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 31|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 5886| | 5887| 31| while (this_th->th.th_cg_roots) { ------------------ | Branch (5887:10): [True: 31, False: 0] ------------------ 5888| 31| this_th->th.th_cg_roots->cg_nthreads--; 5889| 31| KA_TRACE(100, ("__kmp_free_thread: Thread %p decrement cg_nthreads on node" 5890| 31| " %p of thread %p to %d\n", 5891| 31| this_th, this_th->th.th_cg_roots, 5892| 31| this_th->th.th_cg_roots->cg_root, 5893| 31| this_th->th.th_cg_roots->cg_nthreads)); 5894| 31| kmp_cg_root_t *tmp = this_th->th.th_cg_roots; 5895| 31| if (tmp->cg_root == this_th) { // Thread is a cg_root ------------------ | Branch (5895:9): [True: 0, False: 31] ------------------ 5896| 0| KMP_DEBUG_ASSERT(tmp->cg_nthreads == 0); 5897| 0| KA_TRACE( 5898| 0| 5, ("__kmp_free_thread: Thread %p freeing node %p\n", this_th, tmp)); 5899| 0| this_th->th.th_cg_roots = tmp->up; 5900| 0| __kmp_free(tmp); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 5901| 31| } else { // Worker thread 5902| 31| if (tmp->cg_nthreads == 0) { // last thread leaves contention group ------------------ | Branch (5902:11): [True: 0, False: 31] ------------------ 5903| 0| __kmp_free(tmp); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 5904| 0| } 5905| 31| this_th->th.th_cg_roots = NULL; 5906| 31| break; 5907| 31| } 5908| 31| } 5909| | 5910| | /* If the implicit task assigned to this thread can be used by other threads 5911| | * -> multiple threads can share the data and try to free the task at 5912| | * __kmp_reap_thread at exit. This duplicate use of the task data can happen 5913| | * with higher probability when hot team is disabled but can occurs even when 5914| | * the hot team is enabled */ 5915| 31| __kmp_free_implicit_task(this_th); 5916| 31| this_th->th.th_current_task = NULL; 5917| | 5918| | // If the __kmp_thread_pool_insert_pt is already past the new insert 5919| | // point, then we need to re-scan the entire list. 5920| 31| gtid = this_th->th.th_info.ds.ds_gtid; 5921| 31| if (__kmp_thread_pool_insert_pt != NULL) { ------------------ | Branch (5921:7): [True: 30, False: 1] ------------------ 5922| 30| KMP_DEBUG_ASSERT(__kmp_thread_pool != NULL); 5923| 30| if (__kmp_thread_pool_insert_pt->th.th_info.ds.ds_gtid > gtid) { ------------------ | Branch (5923:9): [True: 0, False: 30] ------------------ 5924| 0| __kmp_thread_pool_insert_pt = NULL; 5925| 0| } 5926| 30| } 5927| | 5928| | // Scan down the list to find the place to insert the thread. 5929| | // scan is the address of a link in the list, possibly the address of 5930| | // __kmp_thread_pool itself. 5931| | // 5932| | // In the absence of nested parallelism, the for loop will have 0 iterations. 5933| 31| if (__kmp_thread_pool_insert_pt != NULL) { ------------------ | Branch (5933:7): [True: 30, False: 1] ------------------ 5934| 30| scan = &(__kmp_thread_pool_insert_pt->th.th_next_pool); 5935| 30| } else { 5936| 1| scan = CCAST(kmp_info_t **, &__kmp_thread_pool); ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5937| 1| } 5938| 31| for (; (*scan != NULL) && ((*scan)->th.th_info.ds.ds_gtid < gtid); ------------------ | Branch (5938:10): [True: 0, False: 31] | Branch (5938:29): [True: 0, False: 0] ------------------ 5939| 31| scan = &((*scan)->th.th_next_pool)) 5940| 0| ; 5941| | 5942| | // Insert the new element on the list, and set __kmp_thread_pool_insert_pt 5943| | // to its address. 5944| 31| TCW_PTR(this_th->th.th_next_pool, *scan); ------------------ | | 1157| 31|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 31|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 5945| 31| __kmp_thread_pool_insert_pt = *scan = this_th; 5946| 31| KMP_DEBUG_ASSERT((this_th->th.th_next_pool == NULL) || 5947| 31| (this_th->th.th_info.ds.ds_gtid < 5948| 31| this_th->th.th_next_pool->th.th_info.ds.ds_gtid)); 5949| 31| TCW_4(this_th->th.th_in_pool, TRUE); ------------------ | | 1128| 31|#define TCW_4(a, b) (a) = (b) ------------------ 5950| 31| __kmp_suspend_initialize_thread(this_th); 5951| 31| __kmp_lock_suspend_mx(this_th); 5952| 31| if (this_th->th.th_active == TRUE) { ------------------ | | 1280| 31|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 31|#define FALSE 0 | | ------------------ ------------------ | Branch (5952:7): [True: 0, False: 31] ------------------ 5953| 0| KMP_ATOMIC_INC(&__kmp_thread_pool_active_nth); ------------------ | | 1259| 0|#define KMP_ATOMIC_INC(p) KMP_ATOMIC_OP(fetch_add, p, 1, acq_rel) | | ------------------ | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 5954| 0| this_th->th.th_active_in_pool = TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 5955| 0| } 5956| |#if KMP_DEBUG 5957| | else { 5958| | KMP_DEBUG_ASSERT(this_th->th.th_active_in_pool == FALSE); 5959| | } 5960| |#endif 5961| 31| __kmp_unlock_suspend_mx(this_th); 5962| | 5963| 31| TCW_4(__kmp_nth, __kmp_nth - 1); ------------------ | | 1128| 31|#define TCW_4(a, b) (a) = (b) ------------------ 5964| | 5965| 31|#ifdef KMP_ADJUST_BLOCKTIME 5966| | /* Adjust blocktime back to user setting or default if necessary */ 5967| | /* Middle initialization might never have occurred */ 5968| 31| if (!__kmp_env_blocktime && (__kmp_avail_proc > 0)) { ------------------ | Branch (5968:7): [True: 31, False: 0] | Branch (5968:31): [True: 31, False: 0] ------------------ 5969| 31| KMP_DEBUG_ASSERT(__kmp_avail_proc > 0); 5970| 31| if (__kmp_nth <= __kmp_avail_proc) { ------------------ | Branch (5970:9): [True: 31, False: 0] ------------------ 5971| 31| __kmp_zero_bt = FALSE; ------------------ | | 1279| 31|#define FALSE 0 ------------------ 5972| 31| } 5973| 31| } 5974| 31|#endif /* KMP_ADJUST_BLOCKTIME */ 5975| | 5976| 31| KMP_MB(); 5977| 31|} __kmp_launch_thread: 5981| 31|void *__kmp_launch_thread(kmp_info_t *this_thr) { 5982| |#if OMP_PROFILING_SUPPORT 5983| | ProfileTraceFile = getenv("LIBOMPTARGET_PROFILE"); 5984| | // TODO: add a configuration option for time granularity 5985| | if (ProfileTraceFile) 5986| | llvm::timeTraceProfilerInitialize(500 /* us */, "libomptarget"); 5987| |#endif 5988| | 5989| 31| int gtid = this_thr->th.th_info.ds.ds_gtid; 5990| | /* void *stack_data;*/ 5991| 31| kmp_team_t **volatile pteam; 5992| | 5993| 31| KMP_MB(); 5994| 31| KA_TRACE(10, ("__kmp_launch_thread: T#%d start\n", gtid)); 5995| | 5996| 31| if (__kmp_env_consistency_check) { ------------------ | Branch (5996:7): [True: 0, False: 31] ------------------ 5997| 0| this_thr->th.th_cons = __kmp_allocate_cons_stack(gtid); // ATT: Memory leak? 5998| 0| } 5999| | 6000| 31|#if OMPD_SUPPORT 6001| 31| if (ompd_state & OMPD_ENABLE_BP) ------------------ | | 35| 31|#define OMPD_ENABLE_BP 0x1 ------------------ | Branch (6001:7): [True: 0, False: 31] ------------------ 6002| 0| ompd_bp_thread_begin(); 6003| 31|#endif 6004| | 6005| 31|#if OMPT_SUPPORT 6006| 31| ompt_data_t *thread_data = nullptr; 6007| 31| if (ompt_enabled.enabled) { ------------------ | Branch (6007:7): [True: 0, False: 31] ------------------ 6008| 0| thread_data = &(this_thr->th.ompt_thread_info.thread_data); 6009| 0| *thread_data = ompt_data_none; ------------------ | | 1408| 0|#define ompt_data_none {0} ------------------ 6010| | 6011| 0| this_thr->th.ompt_thread_info.state = ompt_state_overhead; 6012| 0| this_thr->th.ompt_thread_info.wait_id = 0; 6013| 0| this_thr->th.ompt_thread_info.idle_frame = OMPT_GET_FRAME_ADDRESS(0); ------------------ | | 109| 0|#define OMPT_GET_FRAME_ADDRESS(level) __builtin_frame_address(level) ------------------ 6014| 0| this_thr->th.ompt_thread_info.parallel_flags = 0; 6015| 0| if (ompt_enabled.ompt_callback_thread_begin) { ------------------ | Branch (6015:9): [True: 0, False: 0] ------------------ 6016| 0| ompt_callbacks.ompt_callback(ompt_callback_thread_begin)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 6017| 0| ompt_thread_worker, thread_data); 6018| 0| } 6019| 0| this_thr->th.ompt_thread_info.state = ompt_state_idle; 6020| 0| } 6021| 31|#endif 6022| | 6023| | /* This is the place where threads wait for work */ 6024| 41.9k| while (!TCR_4(__kmp_global.g.g_done)) { ------------------ | | 1127| 41.9k|#define TCR_4(a) (a) ------------------ | Branch (6024:10): [True: 41.8k, False: 31] ------------------ 6025| 41.8k| KMP_DEBUG_ASSERT(this_thr == __kmp_threads[gtid]); 6026| 41.8k| KMP_MB(); 6027| | 6028| | /* wait for work to do */ 6029| 41.8k| KA_TRACE(20, ("__kmp_launch_thread: T#%d waiting for work\n", gtid)); 6030| | 6031| | /* No tid yet since not part of a team */ 6032| 41.8k| __kmp_fork_barrier(gtid, KMP_GTID_DNE); ------------------ | | 1000| 41.8k|#define KMP_GTID_DNE (-2) /* Does not exist */ ------------------ 6033| | 6034| 41.8k|#if OMPT_SUPPORT 6035| 41.8k| if (ompt_enabled.enabled) { ------------------ | Branch (6035:9): [True: 0, False: 41.8k] ------------------ 6036| 0| this_thr->th.ompt_thread_info.state = ompt_state_overhead; 6037| 0| } 6038| 41.8k|#endif 6039| | 6040| 41.8k| pteam = &this_thr->th.th_team; 6041| | 6042| | /* have we been allocated? */ 6043| 41.8k| if (TCR_SYNC_PTR(*pteam) && !TCR_4(__kmp_global.g.g_done)) { ------------------ | | 1158| 83.7k|#define TCR_SYNC_PTR(a) ((void *)TCR_SYNC_8(a)) | | ------------------ | | | | 1140| 41.8k|#define TCR_SYNC_8(a) (a) | | ------------------ | | | Branch (1158:25): [True: 41.2k, False: 604] | | ------------------ ------------------ if (TCR_SYNC_PTR(*pteam) && !TCR_4(__kmp_global.g.g_done)) { ------------------ | | 1127| 41.2k|#define TCR_4(a) (a) ------------------ | Branch (6043:33): [True: 41.2k, False: 12] ------------------ 6044| | /* we were just woken up, so run our new task */ 6045| 41.2k| if (TCR_SYNC_PTR((*pteam)->t.t_pkfn) != NULL) { ------------------ | | 1158| 41.2k|#define TCR_SYNC_PTR(a) ((void *)TCR_SYNC_8(a)) | | ------------------ | | | | 1140| 41.2k|#define TCR_SYNC_8(a) (a) | | ------------------ ------------------ | Branch (6045:11): [True: 41.2k, False: 18.4E] ------------------ 6046| 41.2k| int rc; 6047| 41.2k| KA_TRACE(20, 6048| 41.2k| ("__kmp_launch_thread: T#%d(%d:%d) invoke microtask = %p\n", 6049| 41.2k| gtid, (*pteam)->t.t_id, __kmp_tid_from_gtid(gtid), 6050| 41.2k| (*pteam)->t.t_pkfn)); 6051| | 6052| 41.2k| updateHWFPControl(*pteam); 6053| | 6054| 41.2k|#if OMPT_SUPPORT 6055| 41.2k| if (ompt_enabled.enabled) { ------------------ | Branch (6055:13): [True: 0, False: 41.2k] ------------------ 6056| 0| this_thr->th.ompt_thread_info.state = ompt_state_work_parallel; 6057| 0| } 6058| 41.2k|#endif 6059| | 6060| 41.2k| rc = (*pteam)->t.t_invoke(gtid); 6061| 41.2k| KMP_ASSERT(rc); ------------------ | | 50| 41.2k| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 41.2k] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 6062| | 6063| 41.2k| KMP_MB(); 6064| 41.2k| KA_TRACE(20, ("__kmp_launch_thread: T#%d(%d:%d) done microtask = %p\n", 6065| 41.2k| gtid, (*pteam)->t.t_id, __kmp_tid_from_gtid(gtid), 6066| 41.2k| (*pteam)->t.t_pkfn)); 6067| 41.2k| } 6068| 41.2k|#if OMPT_SUPPORT 6069| 41.2k| if (ompt_enabled.enabled) { ------------------ | Branch (6069:11): [True: 0, False: 41.2k] ------------------ 6070| | /* no frame set while outside task */ 6071| 0| __ompt_get_task_info_object(0)->frame.exit_frame = ompt_data_none; ------------------ | | 1408| 0|#define ompt_data_none {0} ------------------ 6072| | 6073| 0| this_thr->th.ompt_thread_info.state = ompt_state_overhead; 6074| 0| } 6075| 41.2k|#endif 6076| | /* join barrier after parallel region */ 6077| 41.2k| __kmp_join_barrier(gtid); 6078| 41.2k| } 6079| 41.8k| } 6080| | 6081| 31|#if OMPD_SUPPORT 6082| 31| if (ompd_state & OMPD_ENABLE_BP) ------------------ | | 35| 31|#define OMPD_ENABLE_BP 0x1 ------------------ | Branch (6082:7): [True: 0, False: 31] ------------------ 6083| 0| ompd_bp_thread_end(); 6084| 31|#endif 6085| | 6086| 31|#if OMPT_SUPPORT 6087| 31| if (ompt_enabled.ompt_callback_thread_end) { ------------------ | Branch (6087:7): [True: 0, False: 31] ------------------ 6088| 0| ompt_callbacks.ompt_callback(ompt_callback_thread_end)(thread_data); ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 6089| 0| } 6090| 31|#endif 6091| | 6092| 31| this_thr->th.th_task_team = NULL; 6093| | /* run the destructors for the threadprivate data for this thread */ 6094| 31| __kmp_common_destroy_gtid(gtid); 6095| | 6096| 31| KA_TRACE(10, ("__kmp_launch_thread: T#%d done\n", gtid)); 6097| 31| KMP_MB(); 6098| | 6099| |#if OMP_PROFILING_SUPPORT 6100| | llvm::timeTraceProfilerFinishThread(); 6101| |#endif 6102| 31| return this_thr; 6103| 31|} __kmp_internal_end_dest: 6107| 31|void __kmp_internal_end_dest(void *specific_gtid) { 6108| | // Make sure no significant bits are lost 6109| 31| int gtid; 6110| 31| __kmp_type_convert((kmp_intptr_t)specific_gtid - 1, >id); 6111| | 6112| 31| KA_TRACE(30, ("__kmp_internal_end_dest: T#%d\n", gtid)); 6113| | /* NOTE: the gtid is stored as gitd+1 in the thread-local-storage 6114| | * this is because 0 is reserved for the nothing-stored case */ 6115| | 6116| 31| __kmp_internal_end_thread(gtid); 6117| 31|} __kmp_internal_end_atexit: 6129| 1|void __kmp_internal_end_atexit(void) { 6130| 1| KA_TRACE(30, ("__kmp_internal_end_atexit\n")); 6131| | /* [Windows] 6132| | josh: ideally, we want to completely shutdown the library in this atexit 6133| | handler, but stat code that depends on thread specific data for gtid fails 6134| | because that data becomes unavailable at some point during the shutdown, so 6135| | we call __kmp_internal_end_thread instead. We should eventually remove the 6136| | dependency on __kmp_get_specific_gtid in the stat code and use 6137| | __kmp_internal_end_library to cleanly shutdown the library. 6138| | 6139| | // TODO: Can some of this comment about GVS be removed? 6140| | I suspect that the offending stat code is executed when the calling thread 6141| | tries to clean up a dead root thread's data structures, resulting in GVS 6142| | code trying to close the GVS structures for that thread, but since the stat 6143| | code uses __kmp_get_specific_gtid to get the gtid with the assumption that 6144| | the calling thread is cleaning up itself instead of another thread, it get 6145| | confused. This happens because allowing a thread to unregister and cleanup 6146| | another thread is a recent modification for addressing an issue. 6147| | Based on the current design (20050722), a thread may end up 6148| | trying to unregister another thread only if thread death does not trigger 6149| | the calling of __kmp_internal_end_thread. For Linux* OS, there is the 6150| | thread specific data destructor function to detect thread death. For 6151| | Windows dynamic, there is DllMain(THREAD_DETACH). For Windows static, there 6152| | is nothing. Thus, the workaround is applicable only for Windows static 6153| | stat library. */ 6154| 1| __kmp_internal_end_library(-1); 6155| |#if KMP_OS_WINDOWS 6156| | __kmp_close_console(); 6157| |#endif 6158| 1|} __kmp_internal_end_library: 6436| 1|void __kmp_internal_end_library(int gtid_req) { 6437| | /* if we have already cleaned up, don't try again, it wouldn't be pretty */ 6438| | /* this shouldn't be a race condition because __kmp_internal_end() is the 6439| | only place to clear __kmp_serial_init */ 6440| | /* we'll check this later too, after we get the lock */ 6441| | // 2009-09-06: We do not set g_abort without setting g_done. This check looks 6442| | // redundant, because the next check will work in any case. 6443| 1| if (__kmp_global.g.g_abort) { ------------------ | Branch (6443:7): [True: 0, False: 1] ------------------ 6444| 0| KA_TRACE(11, ("__kmp_internal_end_library: abort, exiting\n")); 6445| | /* TODO abort? */ 6446| 0| return; 6447| 0| } 6448| 1| if (TCR_4(__kmp_global.g.g_done) || !__kmp_init_serial) { ------------------ | | 1127| 2|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 1] | | ------------------ ------------------ | Branch (6448:39): [True: 0, False: 1] ------------------ 6449| 0| KA_TRACE(10, ("__kmp_internal_end_library: already finished\n")); 6450| 0| return; 6451| 0| } 6452| | 6453| | // If hidden helper team has been initialized, we need to deinit it 6454| 1| if (TCR_4(__kmp_init_hidden_helper) && ------------------ | | 1127| 2|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 1] | | ------------------ ------------------ 6455| 0| !TCR_4(__kmp_hidden_helper_team_done)) { ------------------ | | 1127| 0|#define TCR_4(a) (a) ------------------ | Branch (6455:7): [True: 0, False: 0] ------------------ 6456| 0| TCW_SYNC_4(__kmp_hidden_helper_team_done, TRUE); ------------------ | | 1136| 0|#define TCW_SYNC_4(a, b) (a) = (b) ------------------ 6457| | // First release the main thread to let it continue its work 6458| 0| __kmp_hidden_helper_main_thread_release(); 6459| | // Wait until the hidden helper team has been destroyed 6460| 0| __kmp_hidden_helper_threads_deinitz_wait(); 6461| 0| } 6462| | 6463| 1| KMP_MB(); /* Flush all pending memory write invalidates. */ 6464| | /* find out who we are and what we should do */ 6465| 1| { 6466| 1| int gtid = (gtid_req >= 0) ? gtid_req : __kmp_gtid_get_specific(); ------------------ | Branch (6466:16): [True: 0, False: 1] ------------------ 6467| 1| KA_TRACE( 6468| 1| 10, ("__kmp_internal_end_library: enter T#%d (%d)\n", gtid, gtid_req)); 6469| 1| if (gtid == KMP_GTID_SHUTDOWN) { ------------------ | | 1001| 1|#define KMP_GTID_SHUTDOWN (-3) /* Library is shutting down */ ------------------ | Branch (6469:9): [True: 0, False: 1] ------------------ 6470| 0| KA_TRACE(10, ("__kmp_internal_end_library: !__kmp_init_runtime, system " 6471| 0| "already shutdown\n")); 6472| 0| return; 6473| 1| } else if (gtid == KMP_GTID_MONITOR) { ------------------ | | 1002| 1|#define KMP_GTID_MONITOR (-4) /* Monitor thread ID */ ------------------ | Branch (6473:16): [True: 0, False: 1] ------------------ 6474| 0| KA_TRACE(10, ("__kmp_internal_end_library: monitor thread, gtid not " 6475| 0| "registered, or system shutdown\n")); 6476| 0| return; 6477| 1| } else if (gtid == KMP_GTID_DNE) { ------------------ | | 1000| 1|#define KMP_GTID_DNE (-2) /* Does not exist */ ------------------ | Branch (6477:16): [True: 0, False: 1] ------------------ 6478| 0| KA_TRACE(10, ("__kmp_internal_end_library: gtid not registered or system " 6479| 0| "shutdown\n")); 6480| | /* we don't know who we are, but we may still shutdown the library */ 6481| 1| } else if (KMP_UBER_GTID(gtid)) { ------------------ | Branch (6481:16): [True: 1, False: 0] ------------------ 6482| | /* unregister ourselves as an uber thread. gtid is no longer valid */ 6483| 1| if (__kmp_root[gtid]->r.r_active) { ------------------ | Branch (6483:11): [True: 0, False: 1] ------------------ 6484| 0| __kmp_global.g.g_abort = -1; 6485| 0| TCW_SYNC_4(__kmp_global.g.g_done, TRUE); ------------------ | | 1136| 0|#define TCW_SYNC_4(a, b) (a) = (b) ------------------ 6486| 0| __kmp_unregister_library(); 6487| 0| KA_TRACE(10, 6488| 0| ("__kmp_internal_end_library: root still active, abort T#%d\n", 6489| 0| gtid)); 6490| 0| return; 6491| 1| } else { 6492| 1| __kmp_itthash_clean(__kmp_threads[gtid]); 6493| 1| KA_TRACE( 6494| 1| 10, 6495| 1| ("__kmp_internal_end_library: unregistering sibling T#%d\n", gtid)); 6496| 1| __kmp_unregister_root_current_thread(gtid); 6497| 1| } 6498| 1| } else { 6499| |/* worker threads may call this function through the atexit handler, if they 6500| | * call exit() */ 6501| |/* For now, skip the usual subsequent processing and just dump the debug buffer. 6502| | TODO: do a thorough shutdown instead */ 6503| |#ifdef DUMP_DEBUG_ON_EXIT 6504| | if (__kmp_debug_buf) 6505| | __kmp_dump_debug_buffer(); 6506| |#endif 6507| | // added unregister library call here when we switch to shm linux 6508| | // if we don't, it will leave lots of files in /dev/shm 6509| | // cleanup shared memory file before exiting. 6510| 0| __kmp_unregister_library(); 6511| 0| return; 6512| 0| } 6513| 1| } 6514| | /* synchronize the termination process */ 6515| 1| __kmp_acquire_bootstrap_lock(&__kmp_initz_lock); 6516| | 6517| | /* have we already finished */ 6518| 1| if (__kmp_global.g.g_abort) { ------------------ | Branch (6518:7): [True: 0, False: 1] ------------------ 6519| 0| KA_TRACE(10, ("__kmp_internal_end_library: abort, exiting\n")); 6520| | /* TODO abort? */ 6521| 0| __kmp_release_bootstrap_lock(&__kmp_initz_lock); 6522| 0| return; 6523| 0| } 6524| 1| if (TCR_4(__kmp_global.g.g_done) || !__kmp_init_serial) { ------------------ | | 1127| 2|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 1] | | ------------------ ------------------ | Branch (6524:39): [True: 0, False: 1] ------------------ 6525| 0| __kmp_release_bootstrap_lock(&__kmp_initz_lock); 6526| 0| return; 6527| 0| } 6528| | 6529| | /* We need this lock to enforce mutex between this reading of 6530| | __kmp_threads_capacity and the writing by __kmp_register_root. 6531| | Alternatively, we can use a counter of roots that is atomically updated by 6532| | __kmp_get_global_thread_id_reg, __kmp_do_serial_initialize and 6533| | __kmp_internal_end_*. */ 6534| 1| __kmp_acquire_bootstrap_lock(&__kmp_forkjoin_lock); 6535| | 6536| | /* now we can safely conduct the actual termination */ 6537| 1| __kmp_internal_end(); 6538| | 6539| 1| __kmp_release_bootstrap_lock(&__kmp_forkjoin_lock); 6540| 1| __kmp_release_bootstrap_lock(&__kmp_initz_lock); 6541| | 6542| 1| KA_TRACE(10, ("__kmp_internal_end_library: exit\n")); 6543| | 6544| |#ifdef DUMP_DEBUG_ON_EXIT 6545| | if (__kmp_debug_buf) 6546| | __kmp_dump_debug_buffer(); 6547| |#endif 6548| | 6549| |#if KMP_OS_WINDOWS 6550| | __kmp_close_console(); 6551| |#endif 6552| | 6553| 1| __kmp_fini_allocator(); 6554| | 6555| 1|} // __kmp_internal_end_library __kmp_internal_end_thread: 6557| 31|void __kmp_internal_end_thread(int gtid_req) { 6558| 31| int i; 6559| | 6560| | /* if we have already cleaned up, don't try again, it wouldn't be pretty */ 6561| | /* this shouldn't be a race condition because __kmp_internal_end() is the 6562| | * only place to clear __kmp_serial_init */ 6563| | /* we'll check this later too, after we get the lock */ 6564| | // 2009-09-06: We do not set g_abort without setting g_done. This check looks 6565| | // redundant, because the next check will work in any case. 6566| 31| if (__kmp_global.g.g_abort) { ------------------ | Branch (6566:7): [True: 0, False: 31] ------------------ 6567| 0| KA_TRACE(11, ("__kmp_internal_end_thread: abort, exiting\n")); 6568| | /* TODO abort? */ 6569| 0| return; 6570| 0| } 6571| 31| if (TCR_4(__kmp_global.g.g_done) || !__kmp_init_serial) { ------------------ | | 1127| 62|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 31, False: 0] | | ------------------ ------------------ | Branch (6571:39): [True: 0, False: 0] ------------------ 6572| 31| KA_TRACE(10, ("__kmp_internal_end_thread: already finished\n")); 6573| 31| return; 6574| 31| } 6575| | 6576| | // If hidden helper team has been initialized, we need to deinit it 6577| 0| if (TCR_4(__kmp_init_hidden_helper) && ------------------ | | 1127| 0|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 0] | | ------------------ ------------------ 6578| 0| !TCR_4(__kmp_hidden_helper_team_done)) { ------------------ | | 1127| 0|#define TCR_4(a) (a) ------------------ | Branch (6578:7): [True: 0, False: 0] ------------------ 6579| 0| TCW_SYNC_4(__kmp_hidden_helper_team_done, TRUE); ------------------ | | 1136| 0|#define TCW_SYNC_4(a, b) (a) = (b) ------------------ 6580| | // First release the main thread to let it continue its work 6581| 0| __kmp_hidden_helper_main_thread_release(); 6582| | // Wait until the hidden helper team has been destroyed 6583| 0| __kmp_hidden_helper_threads_deinitz_wait(); 6584| 0| } 6585| | 6586| 0| KMP_MB(); /* Flush all pending memory write invalidates. */ 6587| | 6588| | /* find out who we are and what we should do */ 6589| 0| { 6590| 0| int gtid = (gtid_req >= 0) ? gtid_req : __kmp_gtid_get_specific(); ------------------ | Branch (6590:16): [True: 0, False: 0] ------------------ 6591| 0| KA_TRACE(10, 6592| 0| ("__kmp_internal_end_thread: enter T#%d (%d)\n", gtid, gtid_req)); 6593| 0| if (gtid == KMP_GTID_SHUTDOWN) { ------------------ | | 1001| 0|#define KMP_GTID_SHUTDOWN (-3) /* Library is shutting down */ ------------------ | Branch (6593:9): [True: 0, False: 0] ------------------ 6594| 0| KA_TRACE(10, ("__kmp_internal_end_thread: !__kmp_init_runtime, system " 6595| 0| "already shutdown\n")); 6596| 0| return; 6597| 0| } else if (gtid == KMP_GTID_MONITOR) { ------------------ | | 1002| 0|#define KMP_GTID_MONITOR (-4) /* Monitor thread ID */ ------------------ | Branch (6597:16): [True: 0, False: 0] ------------------ 6598| 0| KA_TRACE(10, ("__kmp_internal_end_thread: monitor thread, gtid not " 6599| 0| "registered, or system shutdown\n")); 6600| 0| return; 6601| 0| } else if (gtid == KMP_GTID_DNE) { ------------------ | | 1000| 0|#define KMP_GTID_DNE (-2) /* Does not exist */ ------------------ | Branch (6601:16): [True: 0, False: 0] ------------------ 6602| 0| KA_TRACE(10, ("__kmp_internal_end_thread: gtid not registered or system " 6603| 0| "shutdown\n")); 6604| 0| return; 6605| | /* we don't know who we are */ 6606| 0| } else if (KMP_UBER_GTID(gtid)) { ------------------ | Branch (6606:16): [True: 0, False: 0] ------------------ 6607| | /* unregister ourselves as an uber thread. gtid is no longer valid */ 6608| 0| if (__kmp_root[gtid]->r.r_active) { ------------------ | Branch (6608:11): [True: 0, False: 0] ------------------ 6609| 0| __kmp_global.g.g_abort = -1; 6610| 0| TCW_SYNC_4(__kmp_global.g.g_done, TRUE); ------------------ | | 1136| 0|#define TCW_SYNC_4(a, b) (a) = (b) ------------------ 6611| 0| KA_TRACE(10, 6612| 0| ("__kmp_internal_end_thread: root still active, abort T#%d\n", 6613| 0| gtid)); 6614| 0| return; 6615| 0| } else { 6616| 0| KA_TRACE(10, ("__kmp_internal_end_thread: unregistering sibling T#%d\n", 6617| 0| gtid)); 6618| 0| __kmp_unregister_root_current_thread(gtid); 6619| 0| } 6620| 0| } else { 6621| | /* just a worker thread, let's leave */ 6622| 0| KA_TRACE(10, ("__kmp_internal_end_thread: worker thread T#%d\n", gtid)); 6623| | 6624| 0| if (gtid >= 0) { ------------------ | Branch (6624:11): [True: 0, False: 0] ------------------ 6625| 0| __kmp_threads[gtid]->th.th_task_team = NULL; 6626| 0| } 6627| | 6628| 0| KA_TRACE(10, 6629| 0| ("__kmp_internal_end_thread: worker thread done, exiting T#%d\n", 6630| 0| gtid)); 6631| 0| return; 6632| 0| } 6633| 0| } 6634| |#if KMP_DYNAMIC_LIB 6635| | if (__kmp_pause_status != kmp_hard_paused) 6636| | // AC: lets not shutdown the dynamic library at the exit of uber thread, 6637| | // because we will better shutdown later in the library destructor. 6638| | { 6639| | KA_TRACE(10, ("__kmp_internal_end_thread: exiting T#%d\n", gtid_req)); 6640| | return; 6641| | } 6642| |#endif 6643| | /* synchronize the termination process */ 6644| 0| __kmp_acquire_bootstrap_lock(&__kmp_initz_lock); 6645| | 6646| | /* have we already finished */ 6647| 0| if (__kmp_global.g.g_abort) { ------------------ | Branch (6647:7): [True: 0, False: 0] ------------------ 6648| 0| KA_TRACE(10, ("__kmp_internal_end_thread: abort, exiting\n")); 6649| | /* TODO abort? */ 6650| 0| __kmp_release_bootstrap_lock(&__kmp_initz_lock); 6651| 0| return; 6652| 0| } 6653| 0| if (TCR_4(__kmp_global.g.g_done) || !__kmp_init_serial) { ------------------ | | 1127| 0|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 0] | | ------------------ ------------------ | Branch (6653:39): [True: 0, False: 0] ------------------ 6654| 0| __kmp_release_bootstrap_lock(&__kmp_initz_lock); 6655| 0| return; 6656| 0| } 6657| | 6658| | /* We need this lock to enforce mutex between this reading of 6659| | __kmp_threads_capacity and the writing by __kmp_register_root. 6660| | Alternatively, we can use a counter of roots that is atomically updated by 6661| | __kmp_get_global_thread_id_reg, __kmp_do_serial_initialize and 6662| | __kmp_internal_end_*. */ 6663| | 6664| | /* should we finish the run-time? are all siblings done? */ 6665| 0| __kmp_acquire_bootstrap_lock(&__kmp_forkjoin_lock); 6666| | 6667| 0| for (i = 0; i < __kmp_threads_capacity; ++i) { ------------------ | Branch (6667:15): [True: 0, False: 0] ------------------ 6668| 0| if (KMP_UBER_GTID(i)) { ------------------ | Branch (6668:9): [True: 0, False: 0] ------------------ 6669| 0| KA_TRACE( 6670| 0| 10, 6671| 0| ("__kmp_internal_end_thread: remaining sibling task: gtid==%d\n", i)); 6672| 0| __kmp_release_bootstrap_lock(&__kmp_forkjoin_lock); 6673| 0| __kmp_release_bootstrap_lock(&__kmp_initz_lock); 6674| 0| return; 6675| 0| } 6676| 0| } 6677| | 6678| | /* now we can safely conduct the actual termination */ 6679| | 6680| 0| __kmp_internal_end(); 6681| | 6682| 0| __kmp_release_bootstrap_lock(&__kmp_forkjoin_lock); 6683| 0| __kmp_release_bootstrap_lock(&__kmp_initz_lock); 6684| | 6685| 0| KA_TRACE(10, ("__kmp_internal_end_thread: exit T#%d\n", gtid_req)); 6686| | 6687| |#ifdef DUMP_DEBUG_ON_EXIT 6688| | if (__kmp_debug_buf) 6689| | __kmp_dump_debug_buffer(); 6690| |#endif 6691| 0|} // __kmp_internal_end_thread _Z30__kmp_register_library_startupv: 6720| 1|void __kmp_register_library_startup(void) { 6721| | 6722| 1| char *name = __kmp_reg_status_name(); // Name of the environment variable. 6723| 1| int done = 0; 6724| 1| union { 6725| 1| double dtime; 6726| 1| long ltime; 6727| 1| } time; 6728| 1|#if KMP_ARCH_X86 || KMP_ARCH_X86_64 6729| 1| __kmp_initialize_system_tick(); 6730| 1|#endif 6731| 1| __kmp_read_system_time(&time.dtime); 6732| 1| __kmp_registration_flag = 0xCAFE0000L | (time.ltime & 0x0000FFFFL); 6733| 1| __kmp_registration_str = 6734| 1| __kmp_str_format("%p-%lx-%s", &__kmp_registration_flag, 6735| 1| __kmp_registration_flag, KMP_LIBRARY_FILE); ------------------ | | 70| 1|#define KMP_LIBRARY_FILE "libomp.a" ------------------ 6736| | 6737| 1| KA_TRACE(50, ("__kmp_register_library_startup: %s=\"%s\"\n", name, 6738| 1| __kmp_registration_str)); 6739| | 6740| 2| while (!done) { ------------------ | Branch (6740:10): [True: 1, False: 1] ------------------ 6741| | 6742| 1| char *value = NULL; // Actual value of the environment variable. 6743| | 6744| |#if defined(KMP_USE_SHM) 6745| | char *shm_name = __kmp_str_format("/%s", name); 6746| | int shm_preexist = 0; 6747| | char *data1; 6748| | int fd1 = shm_open(shm_name, O_CREAT | O_EXCL | O_RDWR, 0666); 6749| | if ((fd1 == -1) && (errno == EEXIST)) { 6750| | // file didn't open because it already exists. 6751| | // try opening existing file 6752| | fd1 = shm_open(shm_name, O_RDWR, 0666); 6753| | if (fd1 == -1) { // file didn't open 6754| | // error out here 6755| | __kmp_fatal(KMP_MSG(FunctionError, "Can't open SHM"), KMP_ERR(0), 6756| | __kmp_msg_null); 6757| | } else { 6758| | // able to open existing file 6759| | shm_preexist = 1; 6760| | } 6761| | } else if (fd1 == -1) { 6762| | // SHM didn't open; it was due to error other than already exists. Try to 6763| | // create a temp file under /tmp. 6764| | // TODO: /tmp might not always be the temporary directory. For now we will 6765| | // not consider TMPDIR. If /tmp is not accessible, we simply error out. 6766| | char *temp_file_name = __kmp_str_format("/tmp/%sXXXXXX", name); 6767| | fd1 = mkstemp(temp_file_name); 6768| | if (fd1 == -1) { 6769| | // error out here. 6770| | __kmp_fatal(KMP_MSG(FunctionError, "Can't open TEMP"), KMP_ERR(errno), 6771| | __kmp_msg_null); 6772| | } 6773| | temp_reg_status_file_name = temp_file_name; 6774| | } 6775| | if (shm_preexist == 0) { 6776| | // we created SHM now set size 6777| | if (ftruncate(fd1, SHM_SIZE) == -1) { 6778| | // error occured setting size; 6779| | __kmp_fatal(KMP_MSG(FunctionError, "Can't set size of SHM"), 6780| | KMP_ERR(errno), __kmp_msg_null); 6781| | } 6782| | } 6783| | data1 = 6784| | (char *)mmap(0, SHM_SIZE, PROT_READ | PROT_WRITE, MAP_SHARED, fd1, 0); 6785| | if (data1 == MAP_FAILED) { 6786| | // failed to map shared memory 6787| | __kmp_fatal(KMP_MSG(FunctionError, "Can't map SHM"), KMP_ERR(errno), 6788| | __kmp_msg_null); 6789| | } 6790| | if (shm_preexist == 0) { // set data to SHM, set value 6791| | KMP_STRCPY_S(data1, SHM_SIZE, __kmp_registration_str); 6792| | } 6793| | // Read value from either what we just wrote or existing file. 6794| | value = __kmp_str_format("%s", data1); // read value from SHM 6795| | munmap(data1, SHM_SIZE); 6796| | close(fd1); 6797| |#else // Windows and unix with static library 6798| | // Set environment variable, but do not overwrite if it is exist. 6799| 1| __kmp_env_set(name, __kmp_registration_str, 0); 6800| | // read value to see if it got set 6801| 1| value = __kmp_env_get(name); 6802| 1|#endif 6803| | 6804| 1| if (value != NULL && strcmp(value, __kmp_registration_str) == 0) { ------------------ | Branch (6804:9): [True: 1, False: 0] | Branch (6804:26): [True: 1, False: 0] ------------------ 6805| 1| done = 1; // Ok, environment variable set successfully, exit the loop. 6806| 1| } else { 6807| | // Oops. Write failed. Another copy of OpenMP RTL is in memory. 6808| | // Check whether it alive or dead. 6809| 0| int neighbor = 0; // 0 -- unknown status, 1 -- alive, 2 -- dead. 6810| 0| char *tail = value; 6811| 0| char *flag_addr_str = NULL; 6812| 0| char *flag_val_str = NULL; 6813| 0| char const *file_name = NULL; 6814| 0| __kmp_str_split(tail, '-', &flag_addr_str, &tail); 6815| 0| __kmp_str_split(tail, '-', &flag_val_str, &tail); 6816| 0| file_name = tail; 6817| 0| if (tail != NULL) { ------------------ | Branch (6817:11): [True: 0, False: 0] ------------------ 6818| 0| unsigned long *flag_addr = 0; 6819| 0| unsigned long flag_val = 0; 6820| 0| KMP_SSCANF(flag_addr_str, "%p", RCAST(void **, &flag_addr)); ------------------ | | 62| 0|#define KMP_SSCANF sscanf ------------------ KMP_SSCANF(flag_addr_str, "%p", RCAST(void **, &flag_addr)); ------------------ | | 230| 0|#define RCAST(type, var) reinterpret_cast(var) ------------------ 6821| 0| KMP_SSCANF(flag_val_str, "%lx", &flag_val); ------------------ | | 62| 0|#define KMP_SSCANF sscanf ------------------ 6822| 0| if (flag_addr != 0 && flag_val != 0 && strcmp(file_name, "") != 0) { ------------------ | Branch (6822:13): [True: 0, False: 0] | Branch (6822:31): [True: 0, False: 0] | Branch (6822:48): [True: 0, False: 0] ------------------ 6823| | // First, check whether environment-encoded address is mapped into 6824| | // addr space. 6825| | // If so, dereference it to see if it still has the right value. 6826| 0| if (__kmp_is_address_mapped(flag_addr) && *flag_addr == flag_val) { ------------------ | Branch (6826:15): [True: 0, False: 0] | Branch (6826:53): [True: 0, False: 0] ------------------ 6827| 0| neighbor = 1; 6828| 0| } else { 6829| | // If not, then we know the other copy of the library is no longer 6830| | // running. 6831| 0| neighbor = 2; 6832| 0| } 6833| 0| } 6834| 0| } 6835| 0| switch (neighbor) { 6836| 0| case 0: // Cannot parse environment variable -- neighbor status unknown. ------------------ | Branch (6836:7): [True: 0, False: 0] ------------------ 6837| | // Assume it is the incompatible format of future version of the 6838| | // library. Assume the other library is alive. 6839| | // WARN( ... ); // TODO: Issue a warning. 6840| 0| file_name = "unknown library"; 6841| 0| KMP_FALLTHROUGH(); ------------------ | | 347| 0|#define KMP_FALLTHROUGH() [[fallthrough]] ------------------ 6842| | // Attention! Falling to the next case. That's intentional. 6843| 0| case 1: { // Neighbor is alive. ------------------ | Branch (6843:7): [True: 0, False: 0] ------------------ 6844| | // Check it is allowed. 6845| 0| char *duplicate_ok = __kmp_env_get("KMP_DUPLICATE_LIB_OK"); 6846| 0| if (!__kmp_str_match_true(duplicate_ok)) { ------------------ | Branch (6846:13): [True: 0, False: 0] ------------------ 6847| | // That's not allowed. Issue fatal error. 6848| 0| __kmp_fatal(KMP_MSG(DuplicateLibrary, KMP_LIBRARY_FILE, file_name), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ | | | | 6848| 0| __kmp_fatal(KMP_MSG(DuplicateLibrary, KMP_LIBRARY_FILE, file_name), | | | | ------------------ | | | | | | 70| 0|#define KMP_LIBRARY_FILE "libomp.a" | | | | ------------------ | | ------------------ ------------------ 6849| 0| KMP_HNT(DuplicateLibrary), __kmp_msg_null); ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 6850| 0| } 6851| 0| KMP_INTERNAL_FREE(duplicate_ok); ------------------ | | 111| 0|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 6852| 0| __kmp_duplicate_library_ok = 1; 6853| 0| done = 1; // Exit the loop. 6854| 0| } break; 6855| 0| case 2: { // Neighbor is dead. ------------------ | Branch (6855:7): [True: 0, False: 0] ------------------ 6856| | 6857| |#if defined(KMP_USE_SHM) 6858| | // close shared memory. 6859| | shm_unlink(shm_name); // this removes file in /dev/shm 6860| |#else 6861| | // Clear the variable and try to register library again. 6862| 0| __kmp_env_unset(name); 6863| 0|#endif 6864| 0| } break; 6865| 0| default: { ------------------ | Branch (6865:7): [True: 0, False: 0] ------------------ 6866| 0| KMP_DEBUG_ASSERT(0); 6867| 0| } break; 6868| 0| } 6869| 0| } 6870| 1| KMP_INTERNAL_FREE((void *)value); ------------------ | | 111| 1|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 6871| |#if defined(KMP_USE_SHM) 6872| | KMP_INTERNAL_FREE((void *)shm_name); 6873| |#endif 6874| 1| } // while 6875| 1| KMP_INTERNAL_FREE((void *)name); ------------------ | | 111| 1|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 6876| | 6877| 1|} // func __kmp_register_library_startup __kmp_unregister_library: 6879| 1|void __kmp_unregister_library(void) { 6880| | 6881| 1| char *name = __kmp_reg_status_name(); 6882| 1| char *value = NULL; 6883| | 6884| |#if defined(KMP_USE_SHM) 6885| | bool use_shm = true; 6886| | char *shm_name = __kmp_str_format("/%s", name); 6887| | int fd1 = shm_open(shm_name, O_RDONLY, 0666); 6888| | if (fd1 == -1) { 6889| | // File did not open. Try the temporary file. 6890| | use_shm = false; 6891| | KMP_DEBUG_ASSERT(temp_reg_status_file_name); 6892| | fd1 = open(temp_reg_status_file_name, O_RDONLY); 6893| | if (fd1 == -1) { 6894| | // give it up now. 6895| | return; 6896| | } 6897| | } 6898| | char *data1 = (char *)mmap(0, SHM_SIZE, PROT_READ, MAP_SHARED, fd1, 0); 6899| | if (data1 != MAP_FAILED) { 6900| | value = __kmp_str_format("%s", data1); // read value from SHM 6901| | munmap(data1, SHM_SIZE); 6902| | } 6903| | close(fd1); 6904| |#else 6905| 1| value = __kmp_env_get(name); 6906| 1|#endif 6907| | 6908| 1| KMP_DEBUG_ASSERT(__kmp_registration_flag != 0); 6909| 1| KMP_DEBUG_ASSERT(__kmp_registration_str != NULL); 6910| 1| if (value != NULL && strcmp(value, __kmp_registration_str) == 0) { ------------------ | Branch (6910:7): [True: 1, False: 0] | Branch (6910:24): [True: 1, False: 0] ------------------ 6911| |// Ok, this is our variable. Delete it. 6912| |#if defined(KMP_USE_SHM) 6913| | if (use_shm) { 6914| | shm_unlink(shm_name); // this removes file in /dev/shm 6915| | } else { 6916| | KMP_DEBUG_ASSERT(temp_reg_status_file_name); 6917| | unlink(temp_reg_status_file_name); // this removes the temp file 6918| | } 6919| |#else 6920| 1| __kmp_env_unset(name); 6921| 1|#endif 6922| 1| } 6923| | 6924| |#if defined(KMP_USE_SHM) 6925| | KMP_INTERNAL_FREE(shm_name); 6926| | if (!use_shm) { 6927| | KMP_DEBUG_ASSERT(temp_reg_status_file_name); 6928| | KMP_INTERNAL_FREE(temp_reg_status_file_name); 6929| | } 6930| |#endif 6931| | 6932| 1| KMP_INTERNAL_FREE(__kmp_registration_str); ------------------ | | 111| 1|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 6933| 1| KMP_INTERNAL_FREE(value); ------------------ | | 111| 1|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 6934| 1| KMP_INTERNAL_FREE(name); ------------------ | | 111| 1|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 6935| | 6936| 1| __kmp_registration_flag = 0; 6937| 1| __kmp_registration_str = NULL; 6938| | 6939| 1|} // __kmp_unregister_library __kmp_middle_initialize: 7431| 1|void __kmp_middle_initialize(void) { 7432| 1| if (__kmp_init_middle) { ------------------ | Branch (7432:7): [True: 0, False: 1] ------------------ 7433| 0| return; 7434| 0| } 7435| 1| __kmp_acquire_bootstrap_lock(&__kmp_initz_lock); 7436| 1| if (__kmp_init_middle) { ------------------ | Branch (7436:7): [True: 0, False: 1] ------------------ 7437| 0| __kmp_release_bootstrap_lock(&__kmp_initz_lock); 7438| 0| return; 7439| 0| } 7440| 1| __kmp_do_middle_initialize(); 7441| 1| __kmp_release_bootstrap_lock(&__kmp_initz_lock); 7442| 1|} __kmp_parallel_initialize: 7444| 1|void __kmp_parallel_initialize(void) { 7445| 1| int gtid = __kmp_entry_gtid(); // this might be a new root ------------------ | | 3467| 1|#define __kmp_entry_gtid() __kmp_get_global_thread_id_reg() ------------------ 7446| | 7447| | /* synchronize parallel initialization (for sibling) */ 7448| 1| if (TCR_4(__kmp_init_parallel)) ------------------ | | 1127| 1|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 1] | | ------------------ ------------------ 7449| 0| return; 7450| 1| __kmp_acquire_bootstrap_lock(&__kmp_initz_lock); 7451| 1| if (TCR_4(__kmp_init_parallel)) { ------------------ | | 1127| 1|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 1] | | ------------------ ------------------ 7452| 0| __kmp_release_bootstrap_lock(&__kmp_initz_lock); 7453| 0| return; 7454| 0| } 7455| | 7456| | /* TODO reinitialization after we have already shut down */ 7457| 1| if (TCR_4(__kmp_global.g.g_done)) { ------------------ | | 1127| 1|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 1] | | ------------------ ------------------ 7458| 0| KA_TRACE( 7459| 0| 10, 7460| 0| ("__kmp_parallel_initialize: attempt to init while shutting down\n")); 7461| 0| __kmp_infinite_loop(); 7462| 0| } 7463| | 7464| | /* jc: The lock __kmp_initz_lock is already held, so calling 7465| | __kmp_serial_initialize would cause a deadlock. So we call 7466| | __kmp_do_serial_initialize directly. */ 7467| 1| if (!__kmp_init_middle) { ------------------ | Branch (7467:7): [True: 0, False: 1] ------------------ 7468| 0| __kmp_do_middle_initialize(); 7469| 0| } 7470| 1| __kmp_assign_root_init_mask(); 7471| 1| __kmp_resume_if_hard_paused(); 7472| | 7473| | /* begin initialization */ 7474| 1| KA_TRACE(10, ("__kmp_parallel_initialize: enter\n")); 7475| 1| KMP_ASSERT(KMP_UBER_GTID(gtid)); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 7476| | 7477| 1|#if KMP_ARCH_X86 || KMP_ARCH_X86_64 7478| | // Save the FP control regs. 7479| | // Worker threads will set theirs to these values at thread startup. 7480| 1| __kmp_store_x87_fpu_control_word(&__kmp_init_x87_fpu_control_word); 7481| 1| __kmp_store_mxcsr(&__kmp_init_mxcsr); 7482| 1| __kmp_init_mxcsr &= KMP_X86_MXCSR_MASK; ------------------ | | 1403| 1|#define KMP_X86_MXCSR_MASK 0xffffffc0 /* ignore status flags (6 lsb) */ ------------------ 7483| 1|#endif /* KMP_ARCH_X86 || KMP_ARCH_X86_64 */ 7484| | 7485| 1|#if KMP_OS_UNIX 7486| 1|#if KMP_HANDLE_SIGNALS 7487| | /* must be after __kmp_serial_initialize */ 7488| 1| __kmp_install_signals(TRUE); ------------------ | | 1280| 1|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1|#define FALSE 0 | | ------------------ ------------------ 7489| 1|#endif 7490| 1|#endif 7491| | 7492| 1| __kmp_suspend_initialize(); 7493| | 7494| 1|#if defined(USE_LOAD_BALANCE) 7495| 1| if (__kmp_global.g.g_dynamic_mode == dynamic_default) { ------------------ | Branch (7495:7): [True: 1, False: 0] ------------------ 7496| 1| __kmp_global.g.g_dynamic_mode = dynamic_load_balance; 7497| 1| } 7498| |#else 7499| | if (__kmp_global.g.g_dynamic_mode == dynamic_default) { 7500| | __kmp_global.g.g_dynamic_mode = dynamic_thread_limit; 7501| | } 7502| |#endif 7503| | 7504| 1| if (__kmp_version) { ------------------ | Branch (7504:7): [True: 0, False: 1] ------------------ 7505| 0| __kmp_print_version_2(); 7506| 0| } 7507| | 7508| | /* we have finished parallel initialization */ 7509| 1| TCW_SYNC_4(__kmp_init_parallel, TRUE); ------------------ | | 1136| 1|#define TCW_SYNC_4(a, b) (a) = (b) ------------------ 7510| | 7511| 1| KMP_MB(); 7512| 1| KA_TRACE(10, ("__kmp_parallel_initialize: exit\n")); 7513| | 7514| 1| __kmp_release_bootstrap_lock(&__kmp_initz_lock); 7515| 1|} __kmp_run_before_invoked_task: 7563| 42.4k| kmp_team_t *team) { 7564| 42.4k| kmp_disp_t *dispatch; 7565| | 7566| 42.4k| KMP_MB(); 7567| | 7568| | /* none of the threads have encountered any constructs, yet. */ 7569| 42.4k| this_thr->th.th_local.this_construct = 0; 7570| |#if KMP_CACHE_MANAGE 7571| | KMP_CACHE_PREFETCH(&this_thr->th.th_bar[bs_forkjoin_barrier].bb.b_arrived); 7572| |#endif /* KMP_CACHE_MANAGE */ 7573| 42.4k| dispatch = (kmp_disp_t *)TCR_PTR(this_thr->th.th_dispatch); ------------------ | | 1156| 42.4k|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 42.4k|#define TCR_8(a) (a) | | ------------------ ------------------ 7574| 42.4k| KMP_DEBUG_ASSERT(dispatch); 7575| 42.4k| KMP_DEBUG_ASSERT(team->t.t_dispatch); 7576| | // KMP_DEBUG_ASSERT( this_thr->th.th_dispatch == &team->t.t_dispatch[ 7577| | // this_thr->th.th_info.ds.ds_tid ] ); 7578| | 7579| 42.4k| dispatch->th_disp_index = 0; /* reset the dispatch buffer counter */ 7580| 42.4k| dispatch->th_doacross_buf_idx = 0; // reset doacross dispatch buffer counter 7581| 42.4k| if (__kmp_env_consistency_check) ------------------ | Branch (7581:7): [True: 0, False: 42.4k] ------------------ 7582| 0| __kmp_push_parallel(gtid, team->t.t_ident); 7583| | 7584| 42.4k| KMP_MB(); /* Flush all pending memory write invalidates. */ 7585| 42.4k|} __kmp_run_after_invoked_task: 7588| 42.2k| kmp_team_t *team) { 7589| 42.2k| if (__kmp_env_consistency_check) ------------------ | Branch (7589:7): [True: 0, False: 42.2k] ------------------ 7590| 0| __kmp_pop_parallel(gtid, team->t.t_ident); 7591| | 7592| 42.2k| __kmp_finish_implicit_task(this_thr); 7593| 42.2k|} __kmp_invoke_task_func: 7595| 42.5k|int __kmp_invoke_task_func(int gtid) { 7596| 42.5k| int rc; 7597| 42.5k| int tid = __kmp_tid_from_gtid(gtid); 7598| 42.5k| kmp_info_t *this_thr = __kmp_threads[gtid]; 7599| 42.5k| kmp_team_t *team = this_thr->th.th_team; 7600| | 7601| 42.5k| __kmp_run_before_invoked_task(gtid, tid, this_thr, team); 7602| |#if USE_ITT_BUILD 7603| | if (__itt_stack_caller_create_ptr) { 7604| | // inform ittnotify about entering user's code 7605| | if (team->t.t_stack_id != NULL) { 7606| | __kmp_itt_stack_callee_enter((__itt_caller)team->t.t_stack_id); 7607| | } else { 7608| | KMP_DEBUG_ASSERT(team->t.t_parent->t.t_stack_id != NULL); 7609| | __kmp_itt_stack_callee_enter( 7610| | (__itt_caller)team->t.t_parent->t.t_stack_id); 7611| | } 7612| | } 7613| |#endif /* USE_ITT_BUILD */ 7614| |#if INCLUDE_SSC_MARKS 7615| | SSC_MARK_INVOKING(); 7616| |#endif 7617| | 7618| 42.5k|#if OMPT_SUPPORT 7619| 42.5k| void *dummy; 7620| 42.5k| void **exit_frame_p; 7621| 42.5k| ompt_data_t *my_task_data; 7622| 42.5k| ompt_data_t *my_parallel_data; 7623| 42.5k| int ompt_team_size; 7624| | 7625| 42.5k| if (ompt_enabled.enabled) { ------------------ | Branch (7625:7): [True: 0, False: 42.5k] ------------------ 7626| 0| exit_frame_p = &(team->t.t_implicit_task_taskdata[tid] 7627| 0| .ompt_task_info.frame.exit_frame.ptr); 7628| 42.5k| } else { 7629| 42.5k| exit_frame_p = &dummy; 7630| 42.5k| } 7631| | 7632| 42.5k| my_task_data = 7633| 42.5k| &(team->t.t_implicit_task_taskdata[tid].ompt_task_info.task_data); 7634| 42.5k| my_parallel_data = &(team->t.ompt_team_info.parallel_data); 7635| 42.5k| if (ompt_enabled.ompt_callback_implicit_task) { ------------------ | Branch (7635:7): [True: 0, False: 42.5k] ------------------ 7636| 0| ompt_team_size = team->t.t_nproc; 7637| 0| ompt_callbacks.ompt_callback(ompt_callback_implicit_task)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 7638| 0| ompt_scope_begin, my_parallel_data, my_task_data, ompt_team_size, 7639| 0| __kmp_tid_from_gtid(gtid), ompt_task_implicit); 7640| 0| OMPT_CUR_TASK_INFO(this_thr)->thread_num = __kmp_tid_from_gtid(gtid); ------------------ | | 68| 0|#define OMPT_CUR_TASK_INFO(thr) (&((thr)->th.th_current_task->ompt_task_info)) ------------------ 7641| 0| } 7642| 42.5k|#endif 7643| | 7644| |#if KMP_STATS_ENABLED 7645| | stats_state_e previous_state = KMP_GET_THREAD_STATE(); 7646| | if (previous_state == stats_state_e::TEAMS_REGION) { 7647| | KMP_PUSH_PARTITIONED_TIMER(OMP_teams); 7648| | } else { 7649| | KMP_PUSH_PARTITIONED_TIMER(OMP_parallel); 7650| | } 7651| | KMP_SET_THREAD_STATE(IMPLICIT_TASK); 7652| |#endif 7653| | 7654| 42.5k| rc = __kmp_invoke_microtask((microtask_t)TCR_SYNC_PTR(team->t.t_pkfn), gtid, ------------------ | | 1158| 42.5k|#define TCR_SYNC_PTR(a) ((void *)TCR_SYNC_8(a)) | | ------------------ | | | | 1140| 42.5k|#define TCR_SYNC_8(a) (a) | | ------------------ ------------------ 7655| 42.5k| tid, (int)team->t.t_argc, (void **)team->t.t_argv 7656| 42.5k|#if OMPT_SUPPORT 7657| 42.5k| , 7658| 42.5k| exit_frame_p 7659| 42.5k|#endif 7660| 42.5k| ); 7661| 42.5k|#if OMPT_SUPPORT 7662| 42.5k| *exit_frame_p = NULL; 7663| 42.5k| this_thr->th.ompt_thread_info.parallel_flags |= ompt_parallel_team; 7664| 42.5k|#endif 7665| | 7666| |#if KMP_STATS_ENABLED 7667| | if (previous_state == stats_state_e::TEAMS_REGION) { 7668| | KMP_SET_THREAD_STATE(previous_state); 7669| | } 7670| | KMP_POP_PARTITIONED_TIMER(); 7671| |#endif 7672| | 7673| |#if USE_ITT_BUILD 7674| | if (__itt_stack_caller_create_ptr) { 7675| | // inform ittnotify about leaving user's code 7676| | if (team->t.t_stack_id != NULL) { 7677| | __kmp_itt_stack_callee_leave((__itt_caller)team->t.t_stack_id); 7678| | } else { 7679| | KMP_DEBUG_ASSERT(team->t.t_parent->t.t_stack_id != NULL); 7680| | __kmp_itt_stack_callee_leave( 7681| | (__itt_caller)team->t.t_parent->t.t_stack_id); 7682| | } 7683| | } 7684| |#endif /* USE_ITT_BUILD */ 7685| 42.5k| __kmp_run_after_invoked_task(gtid, tid, this_thr, team); 7686| | 7687| 42.5k| return rc; 7688| 42.5k|} __kmp_push_num_threads: 7772| 4.36k|void __kmp_push_num_threads(ident_t *id, int gtid, int num_threads) { 7773| 4.36k| kmp_info_t *thr = __kmp_threads[gtid]; 7774| | 7775| 4.36k| if (num_threads > 0) ------------------ | Branch (7775:7): [True: 4.36k, False: 0] ------------------ 7776| 4.36k| thr->th.th_set_nproc = num_threads; 7777| 4.36k|} __kmp_internal_fork: 7942| 1.35k|void __kmp_internal_fork(ident_t *id, int gtid, kmp_team_t *team) { 7943| 1.35k| kmp_info_t *this_thr = __kmp_threads[gtid]; 7944| | 7945| |#ifdef KMP_DEBUG 7946| | int f; 7947| |#endif /* KMP_DEBUG */ 7948| | 7949| 1.35k| KMP_DEBUG_ASSERT(team); 7950| 1.35k| KMP_DEBUG_ASSERT(this_thr->th.th_team == team); 7951| 1.35k| KMP_ASSERT(KMP_MASTER_GTID(gtid)); ------------------ | | 50| 1.35k| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1.35k] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 7952| 1.35k| KMP_MB(); /* Flush all pending memory write invalidates. */ 7953| | 7954| 1.35k| team->t.t_construct = 0; /* no single directives seen yet */ 7955| 1.35k| team->t.t_ordered.dt.t_value = 7956| 1.35k| 0; /* thread 0 enters the ordered section first */ 7957| | 7958| | /* Reset the identifiers on the dispatch buffer */ 7959| 1.35k| KMP_DEBUG_ASSERT(team->t.t_disp_buffer); 7960| 1.35k| if (team->t.t_max_nproc > 1) { ------------------ | Branch (7960:7): [True: 1.35k, False: 0] ------------------ 7961| 1.35k| int i; 7962| 10.8k| for (i = 0; i < __kmp_dispatch_num_buffers; ++i) { ------------------ | Branch (7962:17): [True: 9.47k, False: 1.35k] ------------------ 7963| 9.47k| team->t.t_disp_buffer[i].buffer_index = i; 7964| 9.47k| team->t.t_disp_buffer[i].doacross_buf_idx = i; 7965| 9.47k| } 7966| 1.35k| } else { 7967| 0| team->t.t_disp_buffer[0].buffer_index = 0; 7968| 0| team->t.t_disp_buffer[0].doacross_buf_idx = 0; 7969| 0| } 7970| | 7971| 1.35k| KMP_MB(); /* Flush all pending memory write invalidates. */ 7972| 1.35k| KMP_ASSERT(this_thr->th.th_team == team); ------------------ | | 50| 1.35k| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1.35k] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 7973| | 7974| |#ifdef KMP_DEBUG 7975| | for (f = 0; f < team->t.t_nproc; f++) { 7976| | KMP_DEBUG_ASSERT(team->t.t_threads[f] && 7977| | team->t.t_threads[f]->th.th_team_nproc == team->t.t_nproc); 7978| | } 7979| |#endif /* KMP_DEBUG */ 7980| | 7981| | /* release the worker threads so they may begin working */ 7982| 1.35k| __kmp_fork_barrier(gtid, 0); 7983| 1.35k|} __kmp_internal_join: 7985| 1.35k|void __kmp_internal_join(ident_t *id, int gtid, kmp_team_t *team) { 7986| 1.35k| kmp_info_t *this_thr = __kmp_threads[gtid]; 7987| | 7988| 1.35k| KMP_DEBUG_ASSERT(team); 7989| 1.35k| KMP_DEBUG_ASSERT(this_thr->th.th_team == team); 7990| 1.35k| KMP_ASSERT(KMP_MASTER_GTID(gtid)); ------------------ | | 50| 1.35k| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1.35k] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 7991| 1.35k| KMP_MB(); /* Flush all pending memory write invalidates. */ 7992| | 7993| | /* Join barrier after fork */ 7994| | 7995| |#ifdef KMP_DEBUG 7996| | if (__kmp_threads[gtid] && 7997| | __kmp_threads[gtid]->th.th_team_nproc != team->t.t_nproc) { 7998| | __kmp_printf("GTID: %d, __kmp_threads[%d]=%p\n", gtid, gtid, 7999| | __kmp_threads[gtid]); 8000| | __kmp_printf("__kmp_threads[%d]->th.th_team_nproc=%d, TEAM: %p, " 8001| | "team->t.t_nproc=%d\n", 8002| | gtid, __kmp_threads[gtid]->th.th_team_nproc, team, 8003| | team->t.t_nproc); 8004| | __kmp_print_structure(); 8005| | } 8006| | KMP_DEBUG_ASSERT(__kmp_threads[gtid] && 8007| | __kmp_threads[gtid]->th.th_team_nproc == team->t.t_nproc); 8008| |#endif /* KMP_DEBUG */ 8009| | 8010| 1.35k| __kmp_join_barrier(gtid); /* wait for everyone */ 8011| 1.35k|#if OMPT_SUPPORT 8012| 1.35k| if (ompt_enabled.enabled && ------------------ | Branch (8012:7): [True: 0, False: 1.35k] ------------------ 8013| 0| this_thr->th.ompt_thread_info.state == ompt_state_wait_barrier_implicit) { ------------------ | Branch (8013:7): [True: 0, False: 0] ------------------ 8014| 0| int ds_tid = this_thr->th.th_info.ds.ds_tid; 8015| 0| ompt_data_t *task_data = OMPT_CUR_TASK_DATA(this_thr); ------------------ | | 70| 0| (&((thr)->th.th_current_task->ompt_task_info.task_data)) ------------------ 8016| 0| this_thr->th.ompt_thread_info.state = ompt_state_overhead; 8017| 0|#if OMPT_OPTIONAL 8018| 0| void *codeptr = NULL; 8019| 0| if (KMP_MASTER_TID(ds_tid) && ------------------ | | 1271| 0|#define KMP_MASTER_TID(tid) (0 == (tid)) | | ------------------ | | | Branch (1271:29): [True: 0, False: 0] | | ------------------ ------------------ 8020| 0| (ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait) || ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ | Branch (8020:10): [True: 0, False: 0] ------------------ 8021| 0| ompt_callbacks.ompt_callback(ompt_callback_sync_region))) ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ | Branch (8021:10): [True: 0, False: 0] ------------------ 8022| 0| codeptr = OMPT_CUR_TEAM_INFO(this_thr)->master_return_address; ------------------ | | 71| 0|#define OMPT_CUR_TEAM_INFO(thr) (&((thr)->th.th_team->t.ompt_team_info)) ------------------ 8023| | 8024| 0| if (ompt_enabled.ompt_callback_sync_region_wait) { ------------------ | Branch (8024:9): [True: 0, False: 0] ------------------ 8025| 0| ompt_callbacks.ompt_callback(ompt_callback_sync_region_wait)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 8026| 0| ompt_sync_region_barrier_implicit, ompt_scope_end, NULL, task_data, 8027| 0| codeptr); 8028| 0| } 8029| 0| if (ompt_enabled.ompt_callback_sync_region) { ------------------ | Branch (8029:9): [True: 0, False: 0] ------------------ 8030| 0| ompt_callbacks.ompt_callback(ompt_callback_sync_region)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 8031| 0| ompt_sync_region_barrier_implicit, ompt_scope_end, NULL, task_data, 8032| 0| codeptr); 8033| 0| } 8034| 0|#endif 8035| 0| if (!KMP_MASTER_TID(ds_tid) && ompt_enabled.ompt_callback_implicit_task) { ------------------ | | 1271| 0|#define KMP_MASTER_TID(tid) (0 == (tid)) ------------------ | Branch (8035:9): [True: 0, False: 0] | Branch (8035:36): [True: 0, False: 0] ------------------ 8036| 0| ompt_callbacks.ompt_callback(ompt_callback_implicit_task)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 8037| 0| ompt_scope_end, NULL, task_data, 0, ds_tid, 8038| 0| ompt_task_implicit); // TODO: Can this be ompt_task_initial? 8039| 0| } 8040| 0| } 8041| 1.35k|#endif 8042| | 8043| 1.35k| KMP_MB(); /* Flush all pending memory write invalidates. */ 8044| 1.35k| KMP_ASSERT(this_thr->th.th_team == team); ------------------ | | 50| 1.35k| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1.35k] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 8045| 1.35k|} __kmp_cleanup: 8157| 1|void __kmp_cleanup(void) { 8158| 1| int f; 8159| | 8160| 1| KA_TRACE(10, ("__kmp_cleanup: enter\n")); 8161| | 8162| 1| if (TCR_4(__kmp_init_parallel)) { ------------------ | | 1127| 1|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 1, False: 0] | | ------------------ ------------------ 8163| 1|#if KMP_HANDLE_SIGNALS 8164| 1| __kmp_remove_signals(); 8165| 1|#endif 8166| 1| TCW_4(__kmp_init_parallel, FALSE); ------------------ | | 1128| 1|#define TCW_4(a, b) (a) = (b) ------------------ 8167| 1| } 8168| | 8169| 1| if (TCR_4(__kmp_init_middle)) { ------------------ | | 1127| 1|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 1, False: 0] | | ------------------ ------------------ 8170| 1|#if KMP_AFFINITY_SUPPORTED 8171| 1| __kmp_affinity_uninitialize(); 8172| 1|#endif /* KMP_AFFINITY_SUPPORTED */ 8173| 1| __kmp_cleanup_hierarchy(); 8174| 1| TCW_4(__kmp_init_middle, FALSE); ------------------ | | 1128| 1|#define TCW_4(a, b) (a) = (b) ------------------ 8175| 1| } 8176| | 8177| 1| KA_TRACE(10, ("__kmp_cleanup: go serial cleanup\n")); 8178| | 8179| 1| if (__kmp_init_serial) { ------------------ | Branch (8179:7): [True: 1, False: 0] ------------------ 8180| 1| __kmp_runtime_destroy(); 8181| 1| __kmp_init_serial = FALSE; ------------------ | | 1279| 1|#define FALSE 0 ------------------ 8182| 1| } 8183| | 8184| 1| __kmp_cleanup_threadprivate_caches(); 8185| | 8186| 137| for (f = 0; f < __kmp_threads_capacity; f++) { ------------------ | Branch (8186:15): [True: 136, False: 1] ------------------ 8187| 136| if (__kmp_root[f] != NULL) { ------------------ | Branch (8187:9): [True: 1, False: 135] ------------------ 8188| 1| __kmp_free(__kmp_root[f]); ------------------ | | 3619| 1|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 8189| 1| __kmp_root[f] = NULL; 8190| 1| } 8191| 136| } 8192| 1| __kmp_free(__kmp_threads); ------------------ | | 3619| 1|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 8193| | // __kmp_threads and __kmp_root were allocated at once, as single block, so 8194| | // there is no need in freeing __kmp_root. 8195| 1| __kmp_threads = NULL; 8196| 1| __kmp_root = NULL; 8197| 1| __kmp_threads_capacity = 0; 8198| | 8199| | // Free old __kmp_threads arrays if they exist. 8200| 1| kmp_old_threads_list_t *ptr = __kmp_old_threads_list; 8201| 1| while (ptr) { ------------------ | Branch (8201:10): [True: 0, False: 1] ------------------ 8202| 0| kmp_old_threads_list_t *next = ptr->next; 8203| 0| __kmp_free(ptr->threads); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 8204| 0| __kmp_free(ptr); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 8205| 0| ptr = next; 8206| 0| } 8207| | 8208| 1|#if KMP_USE_DYNAMIC_LOCK 8209| 1| __kmp_cleanup_indirect_user_locks(); 8210| |#else 8211| | __kmp_cleanup_user_locks(); 8212| |#endif 8213| 1|#if OMPD_SUPPORT 8214| 1| if (ompd_state) { ------------------ | Branch (8214:7): [True: 0, False: 1] ------------------ 8215| 0| __kmp_free(ompd_env_block); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 8216| 0| ompd_env_block = NULL; 8217| 0| ompd_env_block_size = 0; 8218| 0| } 8219| 1|#endif 8220| | 8221| 1|#if KMP_AFFINITY_SUPPORTED 8222| 1| KMP_INTERNAL_FREE(CCAST(char *, __kmp_cpuinfo_file)); ------------------ | | 111| 1|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 8223| 1| __kmp_cpuinfo_file = NULL; 8224| 1|#endif /* KMP_AFFINITY_SUPPORTED */ 8225| | 8226| 1|#if KMP_USE_ADAPTIVE_LOCKS 8227| |#if KMP_DEBUG_ADAPTIVE_LOCKS 8228| | __kmp_print_speculative_stats(); 8229| |#endif 8230| 1|#endif 8231| 1| KMP_INTERNAL_FREE(__kmp_nested_nth.nth); ------------------ | | 111| 1|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 8232| 1| __kmp_nested_nth.nth = NULL; 8233| 1| __kmp_nested_nth.size = 0; 8234| 1| __kmp_nested_nth.used = 0; 8235| 1| KMP_INTERNAL_FREE(__kmp_nested_proc_bind.bind_types); ------------------ | | 111| 1|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 8236| 1| __kmp_nested_proc_bind.bind_types = NULL; 8237| 1| __kmp_nested_proc_bind.size = 0; 8238| 1| __kmp_nested_proc_bind.used = 0; 8239| 1| if (__kmp_affinity_format) { ------------------ | Branch (8239:7): [True: 1, False: 0] ------------------ 8240| 1| KMP_INTERNAL_FREE(__kmp_affinity_format); ------------------ | | 111| 1|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 8241| 1| __kmp_affinity_format = NULL; 8242| 1| } 8243| | 8244| 1| __kmp_i18n_catclose(); 8245| | 8246| |#if KMP_USE_HIER_SCHED 8247| | __kmp_hier_scheds.deallocate(); 8248| |#endif 8249| | 8250| |#if KMP_STATS_ENABLED 8251| | __kmp_stats_fini(); 8252| |#endif 8253| | 8254| 1| KA_TRACE(10, ("__kmp_cleanup: exit\n")); 8255| 1|} __kmp_resume_if_soft_paused: 8960| 44.4k|void __kmp_resume_if_soft_paused() { 8961| 44.4k| if (__kmp_pause_status == kmp_soft_paused) { ------------------ | Branch (8961:7): [True: 0, False: 44.4k] ------------------ 8962| 0| __kmp_pause_status = kmp_not_paused; 8963| | 8964| 0| for (int gtid = 1; gtid < __kmp_threads_capacity; ++gtid) { ------------------ | Branch (8964:24): [True: 0, False: 0] ------------------ 8965| 0| kmp_info_t *thread = __kmp_threads[gtid]; 8966| 0| if (thread) { // Wake it if sleeping ------------------ | Branch (8966:11): [True: 0, False: 0] ------------------ 8967| 0| kmp_flag_64<> fl(&thread->th.th_bar[bs_forkjoin_barrier].bb.b_go, 8968| 0| thread); 8969| 0| if (fl.is_sleeping()) ------------------ | Branch (8969:13): [True: 0, False: 0] ------------------ 8970| 0| fl.resume(gtid); 8971| 0| else if (__kmp_try_suspend_mx(thread)) { // got suspend lock ------------------ | Branch (8971:18): [True: 0, False: 0] ------------------ 8972| 0| __kmp_unlock_suspend_mx(thread); // unlock it; it won't sleep 8973| 0| } else { // thread holds the lock and may sleep soon 8974| 0| do { // until either the thread sleeps, or we can get the lock 8975| 0| if (fl.is_sleeping()) { ------------------ | Branch (8975:17): [True: 0, False: 0] ------------------ 8976| 0| fl.resume(gtid); 8977| 0| break; 8978| 0| } else if (__kmp_try_suspend_mx(thread)) { ------------------ | Branch (8978:24): [True: 0, False: 0] ------------------ 8979| 0| __kmp_unlock_suspend_mx(thread); 8980| 0| break; 8981| 0| } 8982| 0| } while (1); ------------------ | Branch (8982:20): [True: 0, Folded] ------------------ 8983| 0| } 8984| 0| } 8985| 0| } 8986| 0| } 8987| 44.4k|} _Z23__kmp_init_nesting_modev: 9193| 1|void __kmp_init_nesting_mode() { 9194| 1| int levels = KMP_HW_LAST; 9195| 1| __kmp_nesting_mode_nlevels = levels; 9196| 1| __kmp_nesting_nth_level = (int *)KMP_INTERNAL_MALLOC(levels * sizeof(int)); ------------------ | | 110| 1|#define KMP_INTERNAL_MALLOC(sz) malloc(sz) ------------------ 9197| 13| for (int i = 0; i < levels; ++i) ------------------ | Branch (9197:19): [True: 12, False: 1] ------------------ 9198| 12| __kmp_nesting_nth_level[i] = 0; 9199| 1| if (__kmp_nested_nth.size < levels) { ------------------ | Branch (9199:7): [True: 1, False: 0] ------------------ 9200| 1| __kmp_nested_nth.nth = 9201| 1| (int *)KMP_INTERNAL_REALLOC(__kmp_nested_nth.nth, levels * sizeof(int)); ------------------ | | 112| 1|#define KMP_INTERNAL_REALLOC(p, sz) realloc((p), (sz)) ------------------ 9202| 1| __kmp_nested_nth.size = levels; 9203| 1| } 9204| 1|} kmp_runtime.cpp:_ZL18propagateFPControlP8kmp_team: 1050| 4.36k|inline static void propagateFPControl(kmp_team_t *team) { 1051| 4.36k| if (__kmp_inherit_fp_control) { ------------------ | Branch (1051:7): [True: 4.36k, False: 0] ------------------ 1052| 4.36k| kmp_int16 x87_fpu_control_word; 1053| 4.36k| kmp_uint32 mxcsr; 1054| | 1055| | // Get primary thread's values of FPU control flags (both X87 and vector) 1056| 4.36k| __kmp_store_x87_fpu_control_word(&x87_fpu_control_word); 1057| 4.36k| __kmp_store_mxcsr(&mxcsr); 1058| 4.36k| mxcsr &= KMP_X86_MXCSR_MASK; ------------------ | | 1403| 4.36k|#define KMP_X86_MXCSR_MASK 0xffffffc0 /* ignore status flags (6 lsb) */ ------------------ 1059| | 1060| | // There is no point looking at t_fp_control_saved here. 1061| | // If it is TRUE, we still have to update the values if they are different 1062| | // from those we now have. If it is FALSE we didn't save anything yet, but 1063| | // our objective is the same. We have to ensure that the values in the team 1064| | // are the same as those we have. 1065| | // So, this code achieves what we need whether or not t_fp_control_saved is 1066| | // true. By checking whether the value needs updating we avoid unnecessary 1067| | // writes that would put the cache-line into a written state, causing all 1068| | // threads in the team to have to read it again. 1069| 4.36k| KMP_CHECK_UPDATE(team->t.t_x87_fpu_control_word, x87_fpu_control_word); ------------------ | | 2280| 4.36k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 2, False: 4.36k] | | ------------------ | | 2281| 4.36k| (a) = (b) ------------------ 1070| 4.36k| KMP_CHECK_UPDATE(team->t.t_mxcsr, mxcsr); ------------------ | | 2280| 4.36k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 2, False: 4.36k] | | ------------------ | | 2281| 4.36k| (a) = (b) ------------------ 1071| | // Although we don't use this value, other code in the runtime wants to know 1072| | // whether it should restore them. So we must ensure it is correct. 1073| 4.36k| KMP_CHECK_UPDATE(team->t.t_fp_control_saved, TRUE); ------------------ | | 2280| 4.36k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 2, False: 4.36k] | | ------------------ | | 2281| 4.36k| (a) = (b) ------------------ 1074| 4.36k| } else { 1075| | // Similarly here. Don't write to this cache-line in the team structure 1076| | // unless we have to. 1077| 0| KMP_CHECK_UPDATE(team->t.t_fp_control_saved, FALSE); ------------------ | | 2280| 0| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 0] | | ------------------ | | 2281| 0| (a) = (b) ------------------ 1078| 0| } 1079| 4.36k|} kmp_runtime.cpp:_ZL22__kmp_is_fork_in_teamsP8kmp_infoPFvPiS1_zEiiPA1_13__va_list_tag: 1375| 1.35k| int teams_level, kmp_va_list ap) { 1376| 1.35k| return (master_th->th.th_teams_microtask && ap && ------------------ | Branch (1376:11): [True: 0, False: 1.35k] | Branch (1376:47): [True: 0, False: 0] ------------------ 1377| 0| microtask != (microtask_t)__kmp_teams_master && level == teams_level); ------------------ | Branch (1377:11): [True: 0, False: 0] | Branch (1377:59): [True: 0, False: 0] ------------------ 1378| 1.35k|} kmp_runtime.cpp:_ZL23__kmp_is_entering_teamsiiiPA1_13__va_list_tag: 1383| 1.35k| int teams_level, kmp_va_list ap) { 1384| 1.35k| return ((ap == NULL && active_level == 0) || ------------------ | Branch (1384:12): [True: 0, False: 1.35k] | Branch (1384:26): [True: 0, False: 0] ------------------ 1385| 1.35k| (ap && teams_level > 0 && teams_level == level)); ------------------ | Branch (1385:12): [True: 1.35k, False: 0] | Branch (1385:18): [True: 0, False: 1.35k] | Branch (1385:37): [True: 0, False: 0] ------------------ 1386| 1.35k|} kmp_runtime.cpp:_ZL21__kmp_reserve_threadsP8kmp_rootP8kmp_teamiii: 740| 1.35k| int enter_teams) { 741| 1.35k| int capacity; 742| 1.35k| int new_nthreads; 743| 1.35k| KMP_DEBUG_ASSERT(__kmp_init_serial); 744| 1.35k| KMP_DEBUG_ASSERT(root && parent_team); 745| 1.35k| kmp_info_t *this_thr = parent_team->t.t_threads[master_tid]; 746| | 747| | // If dyn-var is set, dynamically adjust the number of desired threads, 748| | // according to the method specified by dynamic_mode. 749| 1.35k| new_nthreads = set_nthreads; 750| 1.35k| if (!get__dynamic_2(parent_team, master_tid)) { ------------------ | | 2296| 1.35k| ((xteam)->t.t_threads[(xtid)]->th.th_current_task->td_icvs.dynamic) ------------------ | Branch (750:7): [True: 1.35k, False: 0] ------------------ 751| 1.35k| ; 752| 1.35k| } 753| 0|#ifdef USE_LOAD_BALANCE 754| 0| else if (__kmp_global.g.g_dynamic_mode == dynamic_load_balance) { ------------------ | Branch (754:12): [True: 0, False: 0] ------------------ 755| 0| new_nthreads = __kmp_load_balance_nproc(root, set_nthreads); 756| 0| if (new_nthreads == 1) { ------------------ | Branch (756:9): [True: 0, False: 0] ------------------ 757| 0| KC_TRACE(10, ("__kmp_reserve_threads: T#%d load balance reduced " 758| 0| "reservation to 1 thread\n", 759| 0| master_tid)); 760| 0| return 1; 761| 0| } 762| 0| if (new_nthreads < set_nthreads) { ------------------ | Branch (762:9): [True: 0, False: 0] ------------------ 763| 0| KC_TRACE(10, ("__kmp_reserve_threads: T#%d load balance reduced " 764| 0| "reservation to %d threads\n", 765| 0| master_tid, new_nthreads)); 766| 0| } 767| 0| } 768| 0|#endif /* USE_LOAD_BALANCE */ 769| 0| else if (__kmp_global.g.g_dynamic_mode == dynamic_thread_limit) { ------------------ | Branch (769:12): [True: 0, False: 0] ------------------ 770| 0| new_nthreads = __kmp_avail_proc - __kmp_nth + 771| 0| (root->r.r_active ? 1 : root->r.r_hot_team->t.t_nproc); ------------------ | Branch (771:21): [True: 0, False: 0] ------------------ 772| 0| if (new_nthreads <= 1) { ------------------ | Branch (772:9): [True: 0, False: 0] ------------------ 773| 0| KC_TRACE(10, ("__kmp_reserve_threads: T#%d thread limit reduced " 774| 0| "reservation to 1 thread\n", 775| 0| master_tid)); 776| 0| return 1; 777| 0| } 778| 0| if (new_nthreads < set_nthreads) { ------------------ | Branch (778:9): [True: 0, False: 0] ------------------ 779| 0| KC_TRACE(10, ("__kmp_reserve_threads: T#%d thread limit reduced " 780| 0| "reservation to %d threads\n", 781| 0| master_tid, new_nthreads)); 782| 0| } else { 783| 0| new_nthreads = set_nthreads; 784| 0| } 785| 0| } else if (__kmp_global.g.g_dynamic_mode == dynamic_random) { ------------------ | Branch (785:14): [True: 0, False: 0] ------------------ 786| 0| if (set_nthreads > 2) { ------------------ | Branch (786:9): [True: 0, False: 0] ------------------ 787| 0| new_nthreads = __kmp_get_random(parent_team->t.t_threads[master_tid]); 788| 0| new_nthreads = (new_nthreads % set_nthreads) + 1; 789| 0| if (new_nthreads == 1) { ------------------ | Branch (789:11): [True: 0, False: 0] ------------------ 790| 0| KC_TRACE(10, ("__kmp_reserve_threads: T#%d dynamic random reduced " 791| 0| "reservation to 1 thread\n", 792| 0| master_tid)); 793| 0| return 1; 794| 0| } 795| 0| if (new_nthreads < set_nthreads) { ------------------ | Branch (795:11): [True: 0, False: 0] ------------------ 796| 0| KC_TRACE(10, ("__kmp_reserve_threads: T#%d dynamic random reduced " 797| 0| "reservation to %d threads\n", 798| 0| master_tid, new_nthreads)); 799| 0| } 800| 0| } 801| 0| } else { 802| 0| KMP_ASSERT(0); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, Folded] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 803| 0| } 804| | 805| | // Respect KMP_ALL_THREADS/KMP_DEVICE_THREAD_LIMIT. 806| 1.35k| if (__kmp_nth + new_nthreads - ------------------ | Branch (806:7): [True: 0, False: 1.35k] ------------------ 807| 1.35k| (root->r.r_active ? 1 : root->r.r_hot_team->t.t_nproc) > ------------------ | Branch (807:12): [True: 0, False: 1.35k] ------------------ 808| 1.35k| __kmp_max_nth) { 809| 0| int tl_nthreads = __kmp_max_nth - __kmp_nth + 810| 0| (root->r.r_active ? 1 : root->r.r_hot_team->t.t_nproc); ------------------ | Branch (810:24): [True: 0, False: 0] ------------------ 811| 0| if (tl_nthreads <= 0) { ------------------ | Branch (811:9): [True: 0, False: 0] ------------------ 812| 0| tl_nthreads = 1; 813| 0| } 814| | 815| | // If dyn-var is false, emit a 1-time warning. 816| 0| if (!get__dynamic_2(parent_team, master_tid) && (!__kmp_reserve_warn)) { ------------------ | | 2296| 0| ((xteam)->t.t_threads[(xtid)]->th.th_current_task->td_icvs.dynamic) ------------------ | Branch (816:9): [True: 0, False: 0] | Branch (816:53): [True: 0, False: 0] ------------------ 817| 0| __kmp_reserve_warn = 1; 818| 0| __kmp_msg(kmp_ms_warning, 819| 0| KMP_MSG(CantFormThrTeam, set_nthreads, tl_nthreads), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ 820| 0| KMP_HNT(Unset_ALL_THREADS), __kmp_msg_null); ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 821| 0| } 822| 0| if (tl_nthreads == 1) { ------------------ | Branch (822:9): [True: 0, False: 0] ------------------ 823| 0| KC_TRACE(10, ("__kmp_reserve_threads: T#%d KMP_DEVICE_THREAD_LIMIT " 824| 0| "reduced reservation to 1 thread\n", 825| 0| master_tid)); 826| 0| return 1; 827| 0| } 828| 0| KC_TRACE(10, ("__kmp_reserve_threads: T#%d KMP_DEVICE_THREAD_LIMIT reduced " 829| 0| "reservation to %d threads\n", 830| 0| master_tid, tl_nthreads)); 831| 0| new_nthreads = tl_nthreads; 832| 0| } 833| | 834| | // Respect OMP_THREAD_LIMIT 835| 1.35k| int cg_nthreads = this_thr->th.th_cg_roots->cg_nthreads; 836| 1.35k| int max_cg_threads = this_thr->th.th_cg_roots->cg_thread_limit; 837| 1.35k| if (cg_nthreads + new_nthreads - ------------------ | Branch (837:7): [True: 0, False: 1.35k] ------------------ 838| 1.35k| (root->r.r_active ? 1 : root->r.r_hot_team->t.t_nproc) > ------------------ | Branch (838:12): [True: 0, False: 1.35k] ------------------ 839| 1.35k| max_cg_threads) { 840| 0| int tl_nthreads = max_cg_threads - cg_nthreads + 841| 0| (root->r.r_active ? 1 : root->r.r_hot_team->t.t_nproc); ------------------ | Branch (841:24): [True: 0, False: 0] ------------------ 842| 0| if (tl_nthreads <= 0) { ------------------ | Branch (842:9): [True: 0, False: 0] ------------------ 843| 0| tl_nthreads = 1; 844| 0| } 845| | 846| | // If dyn-var is false, emit a 1-time warning. 847| 0| if (!get__dynamic_2(parent_team, master_tid) && (!__kmp_reserve_warn)) { ------------------ | | 2296| 0| ((xteam)->t.t_threads[(xtid)]->th.th_current_task->td_icvs.dynamic) ------------------ | Branch (847:9): [True: 0, False: 0] | Branch (847:53): [True: 0, False: 0] ------------------ 848| 0| __kmp_reserve_warn = 1; 849| 0| __kmp_msg(kmp_ms_warning, 850| 0| KMP_MSG(CantFormThrTeam, set_nthreads, tl_nthreads), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ 851| 0| KMP_HNT(Unset_ALL_THREADS), __kmp_msg_null); ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 852| 0| } 853| 0| if (tl_nthreads == 1) { ------------------ | Branch (853:9): [True: 0, False: 0] ------------------ 854| 0| KC_TRACE(10, ("__kmp_reserve_threads: T#%d OMP_THREAD_LIMIT " 855| 0| "reduced reservation to 1 thread\n", 856| 0| master_tid)); 857| 0| return 1; 858| 0| } 859| 0| KC_TRACE(10, ("__kmp_reserve_threads: T#%d OMP_THREAD_LIMIT reduced " 860| 0| "reservation to %d threads\n", 861| 0| master_tid, tl_nthreads)); 862| 0| new_nthreads = tl_nthreads; 863| 0| } 864| | 865| | // Check if the threads array is large enough, or needs expanding. 866| | // See comment in __kmp_register_root() about the adjustment if 867| | // __kmp_threads[0] == NULL. 868| 1.35k| capacity = __kmp_threads_capacity; 869| 1.35k| if (TCR_PTR(__kmp_threads[0]) == NULL) { ------------------ | | 1156| 1.35k|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 1.35k|#define TCR_8(a) (a) | | ------------------ ------------------ | Branch (869:7): [True: 0, False: 1.35k] ------------------ 870| 0| --capacity; 871| 0| } 872| | // If it is not for initializing the hidden helper team, we need to take 873| | // __kmp_hidden_helper_threads_num out of the capacity because it is included 874| | // in __kmp_threads_capacity. 875| 1.35k| if (__kmp_enable_hidden_helper && !TCR_4(__kmp_init_hidden_helper_threads)) { ------------------ | | 1127| 1.35k|#define TCR_4(a) (a) ------------------ | Branch (875:7): [True: 1.35k, False: 0] | Branch (875:37): [True: 1.35k, False: 0] ------------------ 876| 1.35k| capacity -= __kmp_hidden_helper_threads_num; 877| 1.35k| } 878| 1.35k| if (__kmp_nth + new_nthreads - ------------------ | Branch (878:7): [True: 0, False: 1.35k] ------------------ 879| 1.35k| (root->r.r_active ? 1 : root->r.r_hot_team->t.t_nproc) > ------------------ | Branch (879:12): [True: 0, False: 1.35k] ------------------ 880| 1.35k| capacity) { 881| | // Expand the threads array. 882| 0| int slotsRequired = __kmp_nth + new_nthreads - 883| 0| (root->r.r_active ? 1 : root->r.r_hot_team->t.t_nproc) - ------------------ | Branch (883:26): [True: 0, False: 0] ------------------ 884| 0| capacity; 885| 0| int slotsAdded = __kmp_expand_threads(slotsRequired); 886| 0| if (slotsAdded < slotsRequired) { ------------------ | Branch (886:9): [True: 0, False: 0] ------------------ 887| | // The threads array was not expanded enough. 888| 0| new_nthreads -= (slotsRequired - slotsAdded); 889| 0| KMP_ASSERT(new_nthreads >= 1); ------------------ | | 50| 0| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 0] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 890| | 891| | // If dyn-var is false, emit a 1-time warning. 892| 0| if (!get__dynamic_2(parent_team, master_tid) && (!__kmp_reserve_warn)) { ------------------ | | 2296| 0| ((xteam)->t.t_threads[(xtid)]->th.th_current_task->td_icvs.dynamic) ------------------ | Branch (892:11): [True: 0, False: 0] | Branch (892:55): [True: 0, False: 0] ------------------ 893| 0| __kmp_reserve_warn = 1; 894| 0| if (__kmp_tp_cached) { ------------------ | Branch (894:13): [True: 0, False: 0] ------------------ 895| 0| __kmp_msg(kmp_ms_warning, 896| 0| KMP_MSG(CantFormThrTeam, set_nthreads, new_nthreads), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ 897| 0| KMP_HNT(Set_ALL_THREADPRIVATE, __kmp_tp_capacity), ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 898| 0| KMP_HNT(PossibleSystemLimitOnThreads), __kmp_msg_null); ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 899| 0| } else { 900| 0| __kmp_msg(kmp_ms_warning, 901| 0| KMP_MSG(CantFormThrTeam, set_nthreads, new_nthreads), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ 902| 0| KMP_HNT(SystemLimitOnThreads), __kmp_msg_null); ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 903| 0| } 904| 0| } 905| 0| } 906| 0| } 907| | 908| |#ifdef KMP_DEBUG 909| | if (new_nthreads == 1) { 910| | KC_TRACE(10, 911| | ("__kmp_reserve_threads: T#%d serializing team after reclaiming " 912| | "dead roots and rechecking; requested %d threads\n", 913| | __kmp_get_gtid(), set_nthreads)); 914| | } else { 915| | KC_TRACE(10, ("__kmp_reserve_threads: T#%d allocating %d threads; requested" 916| | " %d threads\n", 917| | __kmp_get_gtid(), new_nthreads, set_nthreads)); 918| | } 919| |#endif // KMP_DEBUG 920| 1.35k| return new_nthreads; 921| 1.35k|} kmp_runtime.cpp:_ZL23__kmp_fork_team_threadsP8kmp_rootP8kmp_teamP8kmp_infoii: 928| 1.35k| int fork_teams_workers) { 929| 1.35k| int i; 930| 1.35k| int use_hot_team; 931| | 932| 1.35k| KA_TRACE(10, ("__kmp_fork_team_threads: new_nprocs = %d\n", team->t.t_nproc)); 933| 1.35k| KMP_DEBUG_ASSERT(master_gtid == __kmp_get_gtid()); 934| 1.35k| KMP_MB(); 935| | 936| | /* first, let's setup the primary thread */ 937| 1.35k| master_th->th.th_info.ds.ds_tid = 0; 938| 1.35k| master_th->th.th_team = team; 939| 1.35k| master_th->th.th_team_nproc = team->t.t_nproc; 940| 1.35k| master_th->th.th_team_master = master_th; 941| 1.35k| master_th->th.th_team_serialized = FALSE; ------------------ | | 1279| 1.35k|#define FALSE 0 ------------------ 942| 1.35k| master_th->th.th_dispatch = &team->t.t_dispatch[0]; 943| | 944| |/* make sure we are not the optimized hot team */ 945| 1.35k|#if KMP_NESTED_HOT_TEAMS 946| 1.35k| use_hot_team = 0; 947| 1.35k| kmp_hot_team_ptr_t *hot_teams = master_th->th.th_hot_teams; 948| 1.35k| if (hot_teams) { // hot teams array is not allocated if ------------------ | Branch (948:7): [True: 1.35k, False: 0] ------------------ 949| | // KMP_HOT_TEAMS_MAX_LEVEL=0 950| 1.35k| int level = team->t.t_active_level - 1; // index in array of hot teams 951| 1.35k| if (master_th->th.th_teams_microtask) { // are we inside the teams? ------------------ | Branch (951:9): [True: 0, False: 1.35k] ------------------ 952| 0| if (master_th->th.th_teams_size.nteams > 1) { ------------------ | Branch (952:11): [True: 0, False: 0] ------------------ 953| 0| ++level; // level was not increased in teams construct for 954| | // team_of_masters 955| 0| } 956| 0| if (team->t.t_pkfn != (microtask_t)__kmp_teams_master && ------------------ | Branch (956:11): [True: 0, False: 0] ------------------ 957| 0| master_th->th.th_teams_level == team->t.t_level) { ------------------ | Branch (957:11): [True: 0, False: 0] ------------------ 958| 0| ++level; // level was not increased in teams construct for 959| | // team_of_workers before the parallel 960| 0| } // team->t.t_level will be increased inside parallel 961| 0| } 962| 1.35k| if (level < __kmp_hot_teams_max_level) { ------------------ | Branch (962:9): [True: 1.35k, False: 0] ------------------ 963| 1.35k| if (hot_teams[level].hot_team) { ------------------ | Branch (963:11): [True: 1.35k, False: 0] ------------------ 964| | // hot team has already been allocated for given level 965| 1.35k| KMP_DEBUG_ASSERT(hot_teams[level].hot_team == team); 966| 1.35k| use_hot_team = 1; // the team is ready to use 967| 1.35k| } else { 968| 0| use_hot_team = 0; // AC: threads are not allocated yet 969| 0| hot_teams[level].hot_team = team; // remember new hot team 970| 0| hot_teams[level].hot_team_nth = team->t.t_nproc; 971| 0| } 972| 1.35k| } else { 973| 0| use_hot_team = 0; 974| 0| } 975| 1.35k| } 976| |#else 977| | use_hot_team = team == root->r.r_hot_team; 978| |#endif 979| 1.35k| if (!use_hot_team) { ------------------ | Branch (979:7): [True: 0, False: 1.35k] ------------------ 980| | 981| | /* install the primary thread */ 982| 0| team->t.t_threads[0] = master_th; 983| 0| __kmp_initialize_info(master_th, team, 0, master_gtid); 984| | 985| | /* now, install the worker threads */ 986| 0| for (i = 1; i < team->t.t_nproc; i++) { ------------------ | Branch (986:17): [True: 0, False: 0] ------------------ 987| | 988| | /* fork or reallocate a new thread and install it in team */ 989| 0| kmp_info_t *thr = __kmp_allocate_thread(root, team, i); 990| 0| team->t.t_threads[i] = thr; 991| 0| KMP_DEBUG_ASSERT(thr); 992| 0| KMP_DEBUG_ASSERT(thr->th.th_team == team); 993| | /* align team and thread arrived states */ 994| 0| KA_TRACE(20, ("__kmp_fork_team_threads: T#%d(%d:%d) init arrived " 995| 0| "T#%d(%d:%d) join =%llu, plain=%llu\n", 996| 0| __kmp_gtid_from_tid(0, team), team->t.t_id, 0, 997| 0| __kmp_gtid_from_tid(i, team), team->t.t_id, i, 998| 0| team->t.t_bar[bs_forkjoin_barrier].b_arrived, 999| 0| team->t.t_bar[bs_plain_barrier].b_arrived)); 1000| 0| thr->th.th_teams_microtask = master_th->th.th_teams_microtask; 1001| 0| thr->th.th_teams_level = master_th->th.th_teams_level; 1002| 0| thr->th.th_teams_size = master_th->th.th_teams_size; 1003| 0| { // Initialize threads' barrier data. 1004| 0| int b; 1005| 0| kmp_balign_t *balign = team->t.t_threads[i]->th.th_bar; 1006| 0| for (b = 0; b < bs_last_barrier; ++b) { ------------------ | Branch (1006:21): [True: 0, False: 0] ------------------ 1007| 0| balign[b].bb.b_arrived = team->t.t_bar[b].b_arrived; 1008| 0| KMP_DEBUG_ASSERT(balign[b].bb.wait_flag != KMP_BARRIER_PARENT_FLAG); 1009| |#if USE_DEBUGGER 1010| | balign[b].bb.b_worker_arrived = team->t.t_bar[b].b_team_arrived; 1011| |#endif 1012| 0| } 1013| 0| } 1014| 0| } 1015| | 1016| 0|#if KMP_AFFINITY_SUPPORTED 1017| | // Do not partition the places list for teams construct workers who 1018| | // haven't actually been forked to do real work yet. This partitioning 1019| | // will take place in the parallel region nested within the teams construct. 1020| 0| if (!fork_teams_workers) { ------------------ | Branch (1020:9): [True: 0, False: 0] ------------------ 1021| 0| __kmp_partition_places(team); 1022| 0| } 1023| 0|#endif 1024| | 1025| 0| if (team->t.t_nproc > 1 && ------------------ | Branch (1025:9): [True: 0, False: 0] ------------------ 1026| 0| __kmp_barrier_gather_pattern[bs_forkjoin_barrier] == bp_dist_bar) { ------------------ | Branch (1026:9): [True: 0, False: 0] ------------------ 1027| 0| team->t.b->update_num_threads(team->t.t_nproc); 1028| 0| __kmp_add_threads_to_team(team, team->t.t_nproc); 1029| 0| } 1030| 0| } 1031| | 1032| 1.35k| if (__kmp_display_affinity && team->t.t_display_affinity != 1) { ------------------ | Branch (1032:7): [True: 0, False: 1.35k] | Branch (1032:33): [True: 0, False: 0] ------------------ 1033| 0| for (i = 0; i < team->t.t_nproc; i++) { ------------------ | Branch (1033:17): [True: 0, False: 0] ------------------ 1034| 0| kmp_info_t *thr = team->t.t_threads[i]; 1035| 0| if (thr->th.th_prev_num_threads != team->t.t_nproc || ------------------ | Branch (1035:11): [True: 0, False: 0] ------------------ 1036| 0| thr->th.th_prev_level != team->t.t_level) { ------------------ | Branch (1036:11): [True: 0, False: 0] ------------------ 1037| 0| team->t.t_display_affinity = 1; 1038| 0| break; 1039| 0| } 1040| 0| } 1041| 0| } 1042| | 1043| 1.35k| KMP_MB(); 1044| 1.35k|} kmp_runtime.cpp:_ZL17updateHWFPControlP8kmp_team: 1083| 42.6k|inline static void updateHWFPControl(kmp_team_t *team) { 1084| 42.6k| if (__kmp_inherit_fp_control && team->t.t_fp_control_saved) { ------------------ | Branch (1084:7): [True: 42.6k, False: 13] | Branch (1084:35): [True: 42.5k, False: 30] ------------------ 1085| | // Only reset the fp control regs if they have been changed in the team. 1086| | // the parallel region that we are exiting. 1087| 42.5k| kmp_int16 x87_fpu_control_word; 1088| 42.5k| kmp_uint32 mxcsr; 1089| 42.5k| __kmp_store_x87_fpu_control_word(&x87_fpu_control_word); 1090| 42.5k| __kmp_store_mxcsr(&mxcsr); 1091| 42.5k| mxcsr &= KMP_X86_MXCSR_MASK; ------------------ | | 1403| 42.5k|#define KMP_X86_MXCSR_MASK 0xffffffc0 /* ignore status flags (6 lsb) */ ------------------ 1092| | 1093| 42.5k| if (team->t.t_x87_fpu_control_word != x87_fpu_control_word) { ------------------ | Branch (1093:9): [True: 0, False: 42.5k] ------------------ 1094| 0| __kmp_clear_x87_fpu_status_word(); 1095| 0| __kmp_load_x87_fpu_control_word(&team->t.t_x87_fpu_control_word); 1096| 0| } 1097| | 1098| 42.5k| if (team->t.t_mxcsr != mxcsr) { ------------------ | Branch (1098:9): [True: 0, False: 42.5k] ------------------ 1099| 0| __kmp_load_mxcsr(&team->t.t_mxcsr); 1100| 0| } 1101| 42.5k| } 1102| 42.6k|} kmp_runtime.cpp:_ZL21__kmp_initialize_rootP8kmp_root: 3317| 1|static void __kmp_initialize_root(kmp_root_t *root) { 3318| 1| int f; 3319| 1| kmp_team_t *root_team; 3320| 1| kmp_team_t *hot_team; 3321| 1| int hot_team_max_nth; 3322| 1| kmp_r_sched_t r_sched = 3323| 1| __kmp_get_schedule_global(); // get current state of scheduling globals 3324| 1| kmp_internal_control_t r_icvs = __kmp_get_global_icvs(); 3325| 1| KMP_DEBUG_ASSERT(root); 3326| 1| KMP_ASSERT(!root->r.r_begin); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 3327| | 3328| | /* setup the root state structure */ 3329| 1| __kmp_init_lock(&root->r.r_begin_lock); 3330| 1| root->r.r_begin = FALSE; ------------------ | | 1279| 1|#define FALSE 0 ------------------ 3331| 1| root->r.r_active = FALSE; ------------------ | | 1279| 1|#define FALSE 0 ------------------ 3332| 1| root->r.r_in_parallel = 0; 3333| 1| root->r.r_blocktime = __kmp_dflt_blocktime; 3334| 1|#if KMP_AFFINITY_SUPPORTED 3335| 1| root->r.r_affinity_assigned = FALSE; ------------------ | | 1279| 1|#define FALSE 0 ------------------ 3336| 1|#endif 3337| | 3338| | /* setup the root team for this task */ 3339| | /* allocate the root team structure */ 3340| 1| KF_TRACE(10, ("__kmp_initialize_root: before root_team\n")); 3341| | 3342| 1| root_team = 3343| 1| __kmp_allocate_team(root, 3344| 1| 1, // new_nproc 3345| 1| 1, // max_nproc 3346| 1|#if OMPT_SUPPORT 3347| 1| ompt_data_none, // root parallel id ------------------ | | 1408| 1|#define ompt_data_none {0} ------------------ 3348| 1|#endif 3349| 1| __kmp_nested_proc_bind.bind_types[0], &r_icvs, 3350| 1| 0 // argc 3351| 1| USE_NESTED_HOT_ARG(NULL) // primary thread is unknown ------------------ | | 165| 1|#define USE_NESTED_HOT_ARG(x) , x ------------------ 3352| 1| ); 3353| |#if USE_DEBUGGER 3354| | // Non-NULL value should be assigned to make the debugger display the root 3355| | // team. 3356| | TCW_SYNC_PTR(root_team->t.t_pkfn, (microtask_t)(~0)); 3357| |#endif 3358| | 3359| 1| KF_TRACE(10, ("__kmp_initialize_root: after root_team = %p\n", root_team)); 3360| | 3361| 1| root->r.r_root_team = root_team; 3362| 1| root_team->t.t_control_stack_top = NULL; 3363| | 3364| | /* initialize root team */ 3365| 1| root_team->t.t_threads[0] = NULL; 3366| 1| root_team->t.t_nproc = 1; 3367| 1| root_team->t.t_serialized = 1; 3368| | // TODO???: root_team->t.t_max_active_levels = __kmp_dflt_max_active_levels; 3369| 1| root_team->t.t_sched.sched = r_sched.sched; 3370| 1| KA_TRACE( 3371| 1| 20, 3372| 1| ("__kmp_initialize_root: init root team %d arrived: join=%u, plain=%u\n", 3373| 1| root_team->t.t_id, KMP_INIT_BARRIER_STATE, KMP_INIT_BARRIER_STATE)); 3374| | 3375| | /* setup the hot team for this task */ 3376| | /* allocate the hot team structure */ 3377| 1| KF_TRACE(10, ("__kmp_initialize_root: before hot_team\n")); 3378| | 3379| 1| hot_team = 3380| 1| __kmp_allocate_team(root, 3381| 1| 1, // new_nproc 3382| 1| __kmp_dflt_team_nth_ub * 2, // max_nproc 3383| 1|#if OMPT_SUPPORT 3384| 1| ompt_data_none, // root parallel id ------------------ | | 1408| 1|#define ompt_data_none {0} ------------------ 3385| 1|#endif 3386| 1| __kmp_nested_proc_bind.bind_types[0], &r_icvs, 3387| 1| 0 // argc 3388| 1| USE_NESTED_HOT_ARG(NULL) // primary thread is unknown ------------------ | | 165| 1|#define USE_NESTED_HOT_ARG(x) , x ------------------ 3389| 1| ); 3390| 1| KF_TRACE(10, ("__kmp_initialize_root: after hot_team = %p\n", hot_team)); 3391| | 3392| 1| root->r.r_hot_team = hot_team; 3393| 1| root_team->t.t_control_stack_top = NULL; 3394| | 3395| | /* first-time initialization */ 3396| 1| hot_team->t.t_parent = root_team; 3397| | 3398| | /* initialize hot team */ 3399| 1| hot_team_max_nth = hot_team->t.t_max_nproc; 3400| 65| for (f = 0; f < hot_team_max_nth; ++f) { ------------------ | Branch (3400:15): [True: 64, False: 1] ------------------ 3401| 64| hot_team->t.t_threads[f] = NULL; 3402| 64| } 3403| 1| hot_team->t.t_nproc = 1; 3404| | // TODO???: hot_team->t.t_max_active_levels = __kmp_dflt_max_active_levels; 3405| 1| hot_team->t.t_sched.sched = r_sched.sched; 3406| 1| hot_team->t.t_size_changed = 0; 3407| 1|} kmp_runtime.cpp:_ZL21__kmp_get_global_icvsv: 3272| 2|static kmp_internal_control_t __kmp_get_global_icvs(void) { 3273| | 3274| 2| kmp_r_sched_t r_sched = 3275| 2| __kmp_get_schedule_global(); // get current state of scheduling globals 3276| | 3277| 2| KMP_DEBUG_ASSERT(__kmp_nested_proc_bind.used > 0); 3278| | 3279| 2| kmp_internal_control_t g_icvs = { 3280| 2| 0, // int serial_nesting_level; //corresponds to value of th_team_serialized 3281| 2| (kmp_int8)__kmp_global.g.g_dynamic, // internal control for dynamic 3282| | // adjustment of threads (per thread) 3283| 2| (kmp_int8)__kmp_env_blocktime, // int bt_set; //internal control for 3284| | // whether blocktime is explicitly set 3285| 2| __kmp_dflt_blocktime, // int blocktime; //internal control for blocktime 3286| |#if KMP_USE_MONITOR 3287| | __kmp_bt_intervals, // int bt_intervals; //internal control for blocktime 3288| |// intervals 3289| |#endif 3290| 2| __kmp_dflt_team_nth, // int nproc; //internal control for # of threads for 3291| | // next parallel region (per thread) 3292| | // (use a max ub on value if __kmp_parallel_initialize not called yet) 3293| 2| __kmp_cg_max_nth, // int thread_limit; 3294| 2| __kmp_dflt_max_active_levels, // int max_active_levels; //internal control 3295| | // for max_active_levels 3296| 2| r_sched, // kmp_r_sched_t sched; //internal control for runtime schedule 3297| | // {sched,chunk} pair 3298| 2| __kmp_nested_proc_bind.bind_types[0], 3299| 2| __kmp_default_device, 3300| | NULL // struct kmp_internal_control *next; 3301| 2| }; 3302| | 3303| 2| return g_icvs; 3304| 2|} kmp_runtime.cpp:_ZL16__kmp_reset_rootiP8kmp_root: 4067| 1|static int __kmp_reset_root(int gtid, kmp_root_t *root) { 4068| 1| kmp_team_t *root_team = root->r.r_root_team; 4069| 1| kmp_team_t *hot_team = root->r.r_hot_team; 4070| 1| int n = hot_team->t.t_nproc; 4071| 1| int i; 4072| | 4073| 1| KMP_DEBUG_ASSERT(!root->r.r_active); 4074| | 4075| 1| root->r.r_root_team = NULL; 4076| 1| root->r.r_hot_team = NULL; 4077| | // __kmp_free_team() does not free hot teams, so we have to clear r_hot_team 4078| | // before call to __kmp_free_team(). 4079| 1| __kmp_free_team(root, root_team USE_NESTED_HOT_ARG(NULL)); ------------------ | | 165| 1|#define USE_NESTED_HOT_ARG(x) , x ------------------ 4080| 1|#if KMP_NESTED_HOT_TEAMS 4081| 1| if (__kmp_hot_teams_max_level > ------------------ | Branch (4081:7): [True: 1, False: 0] ------------------ 4082| 1| 0) { // need to free nested hot teams and their threads if any 4083| 33| for (i = 0; i < hot_team->t.t_nproc; ++i) { ------------------ | Branch (4083:17): [True: 32, False: 1] ------------------ 4084| 32| kmp_info_t *th = hot_team->t.t_threads[i]; 4085| 32| if (__kmp_hot_teams_max_level > 1) { ------------------ | Branch (4085:11): [True: 0, False: 32] ------------------ 4086| 0| n += __kmp_free_hot_teams(root, th, 1, __kmp_hot_teams_max_level); 4087| 0| } 4088| 32| if (th->th.th_hot_teams) { ------------------ | Branch (4088:11): [True: 1, False: 31] ------------------ 4089| 1| __kmp_free(th->th.th_hot_teams); ------------------ | | 3619| 1|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 4090| 1| th->th.th_hot_teams = NULL; 4091| 1| } 4092| 32| } 4093| 1| } 4094| 1|#endif 4095| 1| __kmp_free_team(root, hot_team USE_NESTED_HOT_ARG(NULL)); ------------------ | | 165| 1|#define USE_NESTED_HOT_ARG(x) , x ------------------ 4096| | 4097| | // Before we can reap the thread, we need to make certain that all other 4098| | // threads in the teams that had this root as ancestor have stopped trying to 4099| | // steal tasks. 4100| 1| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (4100:7): [True: 1, False: 0] ------------------ 4101| 1| __kmp_wait_to_unref_task_teams(); 4102| 1| } 4103| | 4104| |#if KMP_OS_WINDOWS 4105| | /* Close Handle of root duplicated in __kmp_create_worker (tr #62919) */ 4106| | KA_TRACE( 4107| | 10, ("__kmp_reset_root: free handle, th = %p, handle = %" KMP_UINTPTR_SPEC 4108| | "\n", 4109| | (LPVOID) & (root->r.r_uber_thread->th), 4110| | root->r.r_uber_thread->th.th_info.ds.ds_thread)); 4111| | __kmp_free_handle(root->r.r_uber_thread->th.th_info.ds.ds_thread); 4112| |#endif /* KMP_OS_WINDOWS */ 4113| | 4114| 1|#if OMPD_SUPPORT 4115| 1| if (ompd_state & OMPD_ENABLE_BP) ------------------ | | 35| 1|#define OMPD_ENABLE_BP 0x1 ------------------ | Branch (4115:7): [True: 0, False: 1] ------------------ 4116| 0| ompd_bp_thread_end(); 4117| 1|#endif 4118| | 4119| 1|#if OMPT_SUPPORT 4120| 1| ompt_data_t *task_data; 4121| 1| ompt_data_t *parallel_data; 4122| 1| __ompt_get_task_info_internal(0, NULL, &task_data, NULL, ¶llel_data, 4123| 1| NULL); 4124| 1| if (ompt_enabled.ompt_callback_implicit_task) { ------------------ | Branch (4124:7): [True: 0, False: 1] ------------------ 4125| 0| ompt_callbacks.ompt_callback(ompt_callback_implicit_task)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 4126| 0| ompt_scope_end, parallel_data, task_data, 0, 1, ompt_task_initial); 4127| 0| } 4128| 1| if (ompt_enabled.ompt_callback_thread_end) { ------------------ | Branch (4128:7): [True: 0, False: 1] ------------------ 4129| 0| ompt_callbacks.ompt_callback(ompt_callback_thread_end)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 4130| 0| &(root->r.r_uber_thread->th.ompt_thread_info.thread_data)); 4131| 0| } 4132| 1|#endif 4133| | 4134| 1| TCW_4(__kmp_nth, ------------------ | | 1128| 1|#define TCW_4(a, b) (a) = (b) ------------------ 4135| 1| __kmp_nth - 1); // __kmp_reap_thread will decrement __kmp_all_nth. 4136| 1| i = root->r.r_uber_thread->th.th_cg_roots->cg_nthreads--; 4137| 1| KA_TRACE(100, ("__kmp_reset_root: Thread %p decrement cg_nthreads on node %p" 4138| 1| " to %d\n", 4139| 1| root->r.r_uber_thread, root->r.r_uber_thread->th.th_cg_roots, 4140| 1| root->r.r_uber_thread->th.th_cg_roots->cg_nthreads)); 4141| 1| if (i == 1) { ------------------ | Branch (4141:7): [True: 1, False: 0] ------------------ 4142| | // need to free contention group structure 4143| 1| KMP_DEBUG_ASSERT(root->r.r_uber_thread == 4144| 1| root->r.r_uber_thread->th.th_cg_roots->cg_root); 4145| 1| KMP_DEBUG_ASSERT(root->r.r_uber_thread->th.th_cg_roots->up == NULL); 4146| 1| __kmp_free(root->r.r_uber_thread->th.th_cg_roots); ------------------ | | 3619| 1|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 4147| 1| root->r.r_uber_thread->th.th_cg_roots = NULL; 4148| 1| } 4149| 1| __kmp_reap_thread(root->r.r_uber_thread, 1); 4150| | 4151| | // We canot put root thread to __kmp_thread_pool, so we have to reap it 4152| | // instead of freeing. 4153| 1| root->r.r_uber_thread = NULL; 4154| | /* mark root as no longer in use */ 4155| 1| root->r.r_begin = FALSE; ------------------ | | 1279| 1|#define FALSE 0 ------------------ 4156| | 4157| 1| return n; 4158| 1|} kmp_runtime.cpp:_ZL17__kmp_reap_threadP8kmp_infoi: 6160| 32|static void __kmp_reap_thread(kmp_info_t *thread, int is_root) { 6161| | // It is assumed __kmp_forkjoin_lock is acquired. 6162| | 6163| 32| int gtid; 6164| | 6165| 32| KMP_DEBUG_ASSERT(thread != NULL); 6166| | 6167| 32| gtid = thread->th.th_info.ds.ds_gtid; 6168| | 6169| 32| if (!is_root) { ------------------ | Branch (6169:7): [True: 31, False: 1] ------------------ 6170| 31| if (__kmp_dflt_blocktime != KMP_MAX_BLOCKTIME) { ------------------ | | 1184| 31| (INT_MAX) /* Must be this for "infinite" setting the work */ ------------------ | Branch (6170:9): [True: 31, False: 0] ------------------ 6171| | /* Assume the threads are at the fork barrier here */ 6172| 31| KA_TRACE( 6173| 31| 20, ("__kmp_reap_thread: releasing T#%d from fork barrier for reap\n", 6174| 31| gtid)); 6175| 31| if (__kmp_barrier_gather_pattern[bs_forkjoin_barrier] == bp_dist_bar) { ------------------ | Branch (6175:11): [True: 0, False: 31] ------------------ 6176| 0| while ( 6177| 0| !KMP_COMPARE_AND_STORE_ACQ32(&(thread->th.th_used_in_team), 0, 3)) ------------------ | | 807| 0| __sync_bool_compare_and_swap((volatile kmp_uint32 *)(p), (kmp_uint32)(cv), \ | | 808| 0| (kmp_uint32)(sv)) ------------------ | Branch (6177:13): [True: 0, False: 0] ------------------ 6178| 0| KMP_CPU_PAUSE(); ------------------ | | 1498| 0|#define KMP_CPU_PAUSE() __kmp_x86_pause() ------------------ 6179| 0| __kmp_resume_32(gtid, (kmp_flag_32 *)NULL); 6180| 31| } else { 6181| | /* Need release fence here to prevent seg faults for tree forkjoin 6182| | barrier (GEH) */ 6183| 31| kmp_flag_64<> flag(&thread->th.th_bar[bs_forkjoin_barrier].bb.b_go, 6184| 31| thread); 6185| 31| __kmp_release_64(&flag); 6186| 31| } 6187| 31| } 6188| | 6189| | // Terminate OS thread. 6190| 31| __kmp_reap_worker(thread); 6191| | 6192| | // The thread was killed asynchronously. If it was actively 6193| | // spinning in the thread pool, decrement the global count. 6194| | // 6195| | // There is a small timing hole here - if the worker thread was just waking 6196| | // up after sleeping in the pool, had reset it's th_active_in_pool flag but 6197| | // not decremented the global counter __kmp_thread_pool_active_nth yet, then 6198| | // the global counter might not get updated. 6199| | // 6200| | // Currently, this can only happen as the library is unloaded, 6201| | // so there are no harmful side effects. 6202| 31| if (thread->th.th_active_in_pool) { ------------------ | Branch (6202:9): [True: 0, False: 31] ------------------ 6203| 0| thread->th.th_active_in_pool = FALSE; ------------------ | | 1279| 0|#define FALSE 0 ------------------ 6204| 0| KMP_ATOMIC_DEC(&__kmp_thread_pool_active_nth); ------------------ | | 1260| 0|#define KMP_ATOMIC_DEC(p) KMP_ATOMIC_OP(fetch_sub, p, 1, acq_rel) | | ------------------ | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 6205| 0| KMP_DEBUG_ASSERT(__kmp_thread_pool_active_nth >= 0); 6206| 0| } 6207| 31| } 6208| | 6209| 32| __kmp_free_implicit_task(thread); 6210| | 6211| |// Free the fast memory for tasking 6212| 32|#if USE_FAST_MEMORY 6213| 32| __kmp_free_fast_memory(thread); 6214| 32|#endif /* USE_FAST_MEMORY */ 6215| | 6216| 32| __kmp_suspend_uninitialize_thread(thread); 6217| | 6218| 32| KMP_DEBUG_ASSERT(__kmp_threads[gtid] == thread); 6219| 32| TCW_SYNC_PTR(__kmp_threads[gtid], NULL); ------------------ | | 1159| 32|#define TCW_SYNC_PTR(a, b) TCW_SYNC_8((a), (b)) | | ------------------ | | | | 1141| 32|#define TCW_SYNC_8(a, b) (a) = (b) | | ------------------ ------------------ 6220| | 6221| 32| --__kmp_all_nth; 6222| | // __kmp_nth was decremented when thread is added to the pool. 6223| | 6224| 32|#ifdef KMP_ADJUST_BLOCKTIME 6225| | /* Adjust blocktime back to user setting or default if necessary */ 6226| | /* Middle initialization might never have occurred */ 6227| 32| if (!__kmp_env_blocktime && (__kmp_avail_proc > 0)) { ------------------ | Branch (6227:7): [True: 32, False: 0] | Branch (6227:31): [True: 32, False: 0] ------------------ 6228| 32| KMP_DEBUG_ASSERT(__kmp_avail_proc > 0); 6229| 32| if (__kmp_nth <= __kmp_avail_proc) { ------------------ | Branch (6229:9): [True: 32, False: 0] ------------------ 6230| 32| __kmp_zero_bt = FALSE; ------------------ | | 1279| 32|#define FALSE 0 ------------------ 6231| 32| } 6232| 32| } 6233| 32|#endif /* KMP_ADJUST_BLOCKTIME */ 6234| | 6235| | /* free the memory being used */ 6236| 32| if (__kmp_env_consistency_check) { ------------------ | Branch (6236:7): [True: 0, False: 32] ------------------ 6237| 0| if (thread->th.th_cons) { ------------------ | Branch (6237:9): [True: 0, False: 0] ------------------ 6238| 0| __kmp_free_cons_stack(thread->th.th_cons); 6239| 0| thread->th.th_cons = NULL; 6240| 0| } 6241| 0| } 6242| | 6243| 32| if (thread->th.th_pri_common != NULL) { ------------------ | Branch (6243:7): [True: 32, False: 0] ------------------ 6244| 32| __kmp_free(thread->th.th_pri_common); ------------------ | | 3619| 32|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 6245| 32| thread->th.th_pri_common = NULL; 6246| 32| } 6247| | 6248| 32| if (thread->th.th_task_state_memo_stack != NULL) { ------------------ | Branch (6248:7): [True: 32, False: 0] ------------------ 6249| 32| __kmp_free(thread->th.th_task_state_memo_stack); ------------------ | | 3619| 32|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 6250| 32| thread->th.th_task_state_memo_stack = NULL; 6251| 32| } 6252| | 6253| 32|#if KMP_USE_BGET 6254| 32| if (thread->th.th_local.bget_data != NULL) { ------------------ | Branch (6254:7): [True: 32, False: 0] ------------------ 6255| 32| __kmp_finalize_bget(thread); 6256| 32| } 6257| 32|#endif 6258| | 6259| 32|#if KMP_AFFINITY_SUPPORTED 6260| 32| if (thread->th.th_affin_mask != NULL) { ------------------ | Branch (6260:7): [True: 32, False: 0] ------------------ 6261| 32| KMP_CPU_FREE(thread->th.th_affin_mask); ------------------ | | 698| 32|#define KMP_CPU_FREE(ptr) __kmp_affinity_dispatch->deallocate_mask(ptr) ------------------ 6262| 32| thread->th.th_affin_mask = NULL; 6263| 32| } 6264| 32|#endif /* KMP_AFFINITY_SUPPORTED */ 6265| | 6266| |#if KMP_USE_HIER_SCHED 6267| | if (thread->th.th_hier_bar_data != NULL) { 6268| | __kmp_free(thread->th.th_hier_bar_data); 6269| | thread->th.th_hier_bar_data = NULL; 6270| | } 6271| |#endif 6272| | 6273| 32| __kmp_reap_team(thread->th.th_serial_team); 6274| 32| thread->th.th_serial_team = NULL; 6275| 32| __kmp_free(thread); ------------------ | | 3619| 32|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 6276| | 6277| 32| KMP_MB(); 6278| | 6279| 32|} // __kmp_reap_thread kmp_runtime.cpp:_ZL21__kmp_initialize_infoP8kmp_infoP8kmp_teamii: 4247| 64| int tid, int gtid) { 4248| | /* this_thr->th.th_info.ds.ds_gtid is setup in 4249| | kmp_allocate_thread/create_worker. 4250| | this_thr->th.th_serial_team is setup in __kmp_allocate_thread */ 4251| 64| KMP_DEBUG_ASSERT(this_thr != NULL); 4252| 64| KMP_DEBUG_ASSERT(this_thr->th.th_serial_team); 4253| 64| KMP_DEBUG_ASSERT(team); 4254| 64| KMP_DEBUG_ASSERT(team->t.t_threads); 4255| 64| KMP_DEBUG_ASSERT(team->t.t_dispatch); 4256| 64| kmp_info_t *master = team->t.t_threads[0]; 4257| 64| KMP_DEBUG_ASSERT(master); 4258| 64| KMP_DEBUG_ASSERT(master->th.th_root); 4259| | 4260| 64| KMP_MB(); 4261| | 4262| 64| TCW_SYNC_PTR(this_thr->th.th_team, team); ------------------ | | 1159| 64|#define TCW_SYNC_PTR(a, b) TCW_SYNC_8((a), (b)) | | ------------------ | | | | 1141| 64|#define TCW_SYNC_8(a, b) (a) = (b) | | ------------------ ------------------ 4263| | 4264| 64| this_thr->th.th_info.ds.ds_tid = tid; 4265| 64| this_thr->th.th_set_nproc = 0; 4266| 64| if (__kmp_tasking_mode != tskm_immediate_exec) ------------------ | Branch (4266:7): [True: 64, False: 0] ------------------ 4267| | // When tasking is possible, threads are not safe to reap until they are 4268| | // done tasking; this will be set when tasking code is exited in wait 4269| 64| this_thr->th.th_reap_state = KMP_NOT_SAFE_TO_REAP; ------------------ | | 2046| 64| 0 // Thread th_reap_state: not safe to reap (tasking) ------------------ 4270| 0| else // no tasking --> always safe to reap 4271| 0| this_thr->th.th_reap_state = KMP_SAFE_TO_REAP; ------------------ | | 2047| 0|#define KMP_SAFE_TO_REAP 1 // Thread th_reap_state: safe to reap (not tasking) ------------------ 4272| 64| this_thr->th.th_set_proc_bind = proc_bind_default; 4273| 64|#if KMP_AFFINITY_SUPPORTED 4274| 64| this_thr->th.th_new_place = this_thr->th.th_current_place; 4275| 64|#endif 4276| 64| this_thr->th.th_root = master->th.th_root; 4277| | 4278| | /* setup the thread's cache of the team structure */ 4279| 64| this_thr->th.th_team_nproc = team->t.t_nproc; 4280| 64| this_thr->th.th_team_master = master; 4281| 64| this_thr->th.th_team_serialized = team->t.t_serialized; 4282| | 4283| 64| KMP_DEBUG_ASSERT(team->t.t_implicit_task_taskdata); 4284| | 4285| 64| KF_TRACE(10, ("__kmp_initialize_info1: T#%d:%d this_thread=%p curtask=%p\n", 4286| 64| tid, gtid, this_thr, this_thr->th.th_current_task)); 4287| | 4288| 64| __kmp_init_implicit_task(this_thr->th.th_team_master->th.th_ident, this_thr, 4289| 64| team, tid, TRUE); ------------------ | | 1280| 64|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 64|#define FALSE 0 | | ------------------ ------------------ 4290| | 4291| 64| KF_TRACE(10, ("__kmp_initialize_info2: T#%d:%d this_thread=%p curtask=%p\n", 4292| 64| tid, gtid, this_thr, this_thr->th.th_current_task)); 4293| | // TODO: Initialize ICVs from parent; GEH - isn't that already done in 4294| | // __kmp_initialize_team()? 4295| | 4296| | /* TODO no worksharing in speculative threads */ 4297| 64| this_thr->th.th_dispatch = &team->t.t_dispatch[tid]; 4298| | 4299| 64| this_thr->th.th_local.this_construct = 0; 4300| | 4301| 64| if (!this_thr->th.th_pri_common) { ------------------ | Branch (4301:7): [True: 32, False: 32] ------------------ 4302| 32| this_thr->th.th_pri_common = 4303| 32| (struct common_table *)__kmp_allocate(sizeof(struct common_table)); ------------------ | | 3617| 32|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 4304| 32| if (__kmp_storage_map) { ------------------ | Branch (4304:9): [True: 0, False: 32] ------------------ 4305| 0| __kmp_print_storage_map_gtid( 4306| 0| gtid, this_thr->th.th_pri_common, this_thr->th.th_pri_common + 1, 4307| 0| sizeof(struct common_table), "th_%d.th_pri_common\n", gtid); 4308| 0| } 4309| 32| this_thr->th.th_pri_head = NULL; 4310| 32| } 4311| | 4312| 64| if (this_thr != master && // Primary thread's CG root is initialized elsewhere ------------------ | Branch (4312:7): [True: 62, False: 2] ------------------ 4313| 62| this_thr->th.th_cg_roots != master->th.th_cg_roots) { // CG root not set ------------------ | Branch (4313:7): [True: 31, False: 31] ------------------ 4314| | // Make new thread's CG root same as primary thread's 4315| 31| KMP_DEBUG_ASSERT(master->th.th_cg_roots); 4316| 31| kmp_cg_root_t *tmp = this_thr->th.th_cg_roots; 4317| 31| if (tmp) { ------------------ | Branch (4317:9): [True: 0, False: 31] ------------------ 4318| | // worker changes CG, need to check if old CG should be freed 4319| 0| int i = tmp->cg_nthreads--; 4320| 0| KA_TRACE(100, ("__kmp_initialize_info: Thread %p decrement cg_nthreads" 4321| 0| " on node %p of thread %p to %d\n", 4322| 0| this_thr, tmp, tmp->cg_root, tmp->cg_nthreads)); 4323| 0| if (i == 1) { ------------------ | Branch (4323:11): [True: 0, False: 0] ------------------ 4324| 0| __kmp_free(tmp); // last thread left CG --> free it ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 4325| 0| } 4326| 0| } 4327| 31| this_thr->th.th_cg_roots = master->th.th_cg_roots; 4328| | // Increment new thread's CG root's counter to add the new thread 4329| 31| this_thr->th.th_cg_roots->cg_nthreads++; 4330| 31| KA_TRACE(100, ("__kmp_initialize_info: Thread %p increment cg_nthreads on" 4331| 31| " node %p of thread %p to %d\n", 4332| 31| this_thr, this_thr->th.th_cg_roots, 4333| 31| this_thr->th.th_cg_roots->cg_root, 4334| 31| this_thr->th.th_cg_roots->cg_nthreads)); 4335| 31| this_thr->th.th_current_task->td_icvs.thread_limit = 4336| 31| this_thr->th.th_cg_roots->cg_thread_limit; 4337| 31| } 4338| | 4339| | /* Initialize dynamic dispatch */ 4340| 64| { 4341| 64| volatile kmp_disp_t *dispatch = this_thr->th.th_dispatch; 4342| | // Use team max_nproc since this will never change for the team. 4343| 64| size_t disp_size = 4344| 64| sizeof(dispatch_private_info_t) * 4345| 64| (team->t.t_max_nproc == 1 ? 1 : __kmp_dispatch_num_buffers); ------------------ | Branch (4345:10): [True: 1, False: 63] ------------------ 4346| 64| KD_TRACE(10, ("__kmp_initialize_info: T#%d max_nproc: %d\n", gtid, 4347| 64| team->t.t_max_nproc)); 4348| 64| KMP_ASSERT(dispatch); ------------------ | | 50| 64| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 64] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 4349| 64| KMP_DEBUG_ASSERT(team->t.t_dispatch); 4350| 64| KMP_DEBUG_ASSERT(dispatch == &team->t.t_dispatch[tid]); 4351| | 4352| 64| dispatch->th_disp_index = 0; 4353| 64| dispatch->th_doacross_buf_idx = 0; 4354| 64| if (!dispatch->th_disp_buffer) { ------------------ | Branch (4354:9): [True: 33, False: 31] ------------------ 4355| 33| dispatch->th_disp_buffer = 4356| 33| (dispatch_private_info_t *)__kmp_allocate(disp_size); ------------------ | | 3617| 33|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 4357| | 4358| 33| if (__kmp_storage_map) { ------------------ | Branch (4358:11): [True: 0, False: 33] ------------------ 4359| 0| __kmp_print_storage_map_gtid( 4360| 0| gtid, &dispatch->th_disp_buffer[0], 4361| 0| &dispatch->th_disp_buffer[team->t.t_max_nproc == 1 ------------------ | Branch (4361:39): [True: 0, False: 0] ------------------ 4362| 0| ? 1 4363| 0| : __kmp_dispatch_num_buffers], 4364| 0| disp_size, 4365| 0| "th_%d.th_dispatch.th_disp_buffer " 4366| 0| "(team_%d.t_dispatch[%d].th_disp_buffer)", 4367| 0| gtid, team->t.t_id, gtid); 4368| 0| } 4369| 33| } else { 4370| 31| memset(&dispatch->th_disp_buffer[0], '\0', disp_size); 4371| 31| } 4372| | 4373| 64| dispatch->th_dispatch_pr_current = 0; 4374| 64| dispatch->th_dispatch_sh_current = 0; 4375| | 4376| 64| dispatch->th_deo_fcn = 0; /* ORDERED */ 4377| 64| dispatch->th_dxo_fcn = 0; /* END ORDERED */ 4378| 64| } 4379| | 4380| 64| this_thr->th.th_next_pool = NULL; 4381| | 4382| 64| if (!this_thr->th.th_task_state_memo_stack) { ------------------ | Branch (4382:7): [True: 32, False: 32] ------------------ 4383| 32| size_t i; 4384| 32| this_thr->th.th_task_state_memo_stack = 4385| 32| (kmp_uint8 *)__kmp_allocate(4 * sizeof(kmp_uint8)); ------------------ | | 3617| 32|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 4386| 32| this_thr->th.th_task_state_top = 0; 4387| 32| this_thr->th.th_task_state_stack_sz = 4; 4388| 160| for (i = 0; i < this_thr->th.th_task_state_stack_sz; ------------------ | Branch (4388:17): [True: 128, False: 32] ------------------ 4389| 128| ++i) // zero init the stack 4390| 128| this_thr->th.th_task_state_memo_stack[i] = 0; 4391| 32| } 4392| | 4393| 64| KMP_DEBUG_ASSERT(!this_thr->th.th_spin_here); 4394| 64| KMP_DEBUG_ASSERT(this_thr->th.th_next_waiting == 0); 4395| | 4396| 64| KMP_MB(); 4397| 64|} kmp_runtime.cpp:_ZL23__kmp_get_x_global_icvsPK8kmp_team: 3306| 31|static kmp_internal_control_t __kmp_get_x_global_icvs(const kmp_team_t *team) { 3307| | 3308| 31| kmp_internal_control_t gx_icvs; 3309| 31| gx_icvs.serial_nesting_level = 3310| 31| 0; // probably =team->t.t_serial like in save_inter_controls 3311| 31| copy_icvs(&gx_icvs, &team->t.t_threads[0]->th.th_current_task->td_icvs); 3312| 31| gx_icvs.next = NULL; 3313| | 3314| 31| return gx_icvs; 3315| 31|} kmp_runtime.cpp:_ZL23__kmp_reinitialize_teamP8kmp_teamP20kmp_internal_controlP5ident: 4694| 1.38k| ident_t *loc) { 4695| 1.38k| KF_TRACE(10, ("__kmp_reinitialize_team: enter this_thread=%p team=%p\n", 4696| 1.38k| team->t.t_threads[0], team)); 4697| 1.38k| KMP_DEBUG_ASSERT(team && new_icvs); 4698| 1.38k| KMP_DEBUG_ASSERT((!TCR_4(__kmp_init_parallel)) || new_icvs->nproc); 4699| 1.38k| KMP_CHECK_UPDATE(team->t.t_ident, loc); ------------------ | | 2280| 1.38k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 321, False: 1.06k] | | ------------------ | | 2281| 1.38k| (a) = (b) ------------------ 4700| | 4701| 1.38k| KMP_CHECK_UPDATE(team->t.t_id, KMP_GEN_TEAM_ID()); ------------------ | | 2280| 1.38k| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 34, False: 1.35k] | | ------------------ | | 2281| 1.38k| (a) = (b) ------------------ 4702| | // Copy ICVs to the primary thread's implicit taskdata 4703| 1.38k| __kmp_init_implicit_task(loc, team->t.t_threads[0], team, 0, FALSE); ------------------ | | 1279| 1.38k|#define FALSE 0 ------------------ 4704| 1.38k| copy_icvs(&team->t.t_implicit_task_taskdata[0].td_icvs, new_icvs); 4705| | 4706| 1.38k| KF_TRACE(10, ("__kmp_reinitialize_team: exit this_thread=%p team=%p\n", 4707| 1.38k| team->t.t_threads[0], team)); 4708| 1.38k|} kmp_runtime.cpp:_ZL22__kmp_partition_placesP8kmp_teami: 4772| 1|static void __kmp_partition_places(kmp_team_t *team, int update_master_only) { 4773| | // Do not partition places for the hidden helper team 4774| 1| if (KMP_HIDDEN_HELPER_TEAM(team)) ------------------ | | 4426| 1| (team->t.t_threads[0] == __kmp_hidden_helper_main_thread) | | ------------------ | | | Branch (4426:3): [True: 0, False: 1] | | ------------------ ------------------ 4775| 0| return; 4776| | // Copy the primary thread's place partition to the team struct 4777| 1| kmp_info_t *master_th = team->t.t_threads[0]; 4778| 1| KMP_DEBUG_ASSERT(master_th != NULL); 4779| 1| kmp_proc_bind_t proc_bind = team->t.t_proc_bind; 4780| 1| int first_place = master_th->th.th_first_place; 4781| 1| int last_place = master_th->th.th_last_place; 4782| 1| int masters_place = master_th->th.th_current_place; 4783| 1| int num_masks = __kmp_affinity.num_masks; 4784| 1| team->t.t_first_place = first_place; 4785| 1| team->t.t_last_place = last_place; 4786| | 4787| 1| KA_TRACE(20, ("__kmp_partition_places: enter: proc_bind = %d T#%d(%d:0) " 4788| 1| "bound to place %d partition = [%d,%d]\n", 4789| 1| proc_bind, __kmp_gtid_from_thread(team->t.t_threads[0]), 4790| 1| team->t.t_id, masters_place, first_place, last_place)); 4791| | 4792| 1| switch (proc_bind) { 4793| | 4794| 0| case proc_bind_default: ------------------ | Branch (4794:3): [True: 0, False: 1] ------------------ 4795| | // Serial teams might have the proc_bind policy set to proc_bind_default. 4796| | // Not an issue -- we don't rebind primary thread for any proc_bind policy. 4797| 0| KMP_DEBUG_ASSERT(team->t.t_nproc == 1); 4798| 0| break; 4799| | 4800| 0| case proc_bind_primary: { ------------------ | Branch (4800:3): [True: 0, False: 1] ------------------ 4801| 0| int f; 4802| 0| int n_th = team->t.t_nproc; 4803| 0| for (f = 1; f < n_th; f++) { ------------------ | Branch (4803:17): [True: 0, False: 0] ------------------ 4804| 0| kmp_info_t *th = team->t.t_threads[f]; 4805| 0| KMP_DEBUG_ASSERT(th != NULL); 4806| 0| th->th.th_first_place = first_place; 4807| 0| th->th.th_last_place = last_place; 4808| 0| th->th.th_new_place = masters_place; 4809| 0| if (__kmp_display_affinity && masters_place != th->th.th_current_place && ------------------ | Branch (4809:11): [True: 0, False: 0] | Branch (4809:37): [True: 0, False: 0] ------------------ 4810| 0| team->t.t_display_affinity != 1) { ------------------ | Branch (4810:11): [True: 0, False: 0] ------------------ 4811| 0| team->t.t_display_affinity = 1; 4812| 0| } 4813| | 4814| 0| KA_TRACE(100, ("__kmp_partition_places: primary: T#%d(%d:%d) place %d " 4815| 0| "partition = [%d,%d]\n", 4816| 0| __kmp_gtid_from_thread(team->t.t_threads[f]), team->t.t_id, 4817| 0| f, masters_place, first_place, last_place)); 4818| 0| } 4819| 0| } break; 4820| | 4821| 0| case proc_bind_close: { ------------------ | Branch (4821:3): [True: 0, False: 1] ------------------ 4822| 0| int f; 4823| 0| int n_th = team->t.t_nproc; 4824| 0| int n_places; 4825| 0| if (first_place <= last_place) { ------------------ | Branch (4825:9): [True: 0, False: 0] ------------------ 4826| 0| n_places = last_place - first_place + 1; 4827| 0| } else { 4828| 0| n_places = num_masks - first_place + last_place + 1; 4829| 0| } 4830| 0| if (n_th <= n_places) { ------------------ | Branch (4830:9): [True: 0, False: 0] ------------------ 4831| 0| int place = masters_place; 4832| 0| for (f = 1; f < n_th; f++) { ------------------ | Branch (4832:19): [True: 0, False: 0] ------------------ 4833| 0| kmp_info_t *th = team->t.t_threads[f]; 4834| 0| KMP_DEBUG_ASSERT(th != NULL); 4835| | 4836| 0| if (place == last_place) { ------------------ | Branch (4836:13): [True: 0, False: 0] ------------------ 4837| 0| place = first_place; 4838| 0| } else if (place == (num_masks - 1)) { ------------------ | Branch (4838:20): [True: 0, False: 0] ------------------ 4839| 0| place = 0; 4840| 0| } else { 4841| 0| place++; 4842| 0| } 4843| 0| th->th.th_first_place = first_place; 4844| 0| th->th.th_last_place = last_place; 4845| 0| th->th.th_new_place = place; 4846| 0| if (__kmp_display_affinity && place != th->th.th_current_place && ------------------ | Branch (4846:13): [True: 0, False: 0] | Branch (4846:39): [True: 0, False: 0] ------------------ 4847| 0| team->t.t_display_affinity != 1) { ------------------ | Branch (4847:13): [True: 0, False: 0] ------------------ 4848| 0| team->t.t_display_affinity = 1; 4849| 0| } 4850| | 4851| 0| KA_TRACE(100, ("__kmp_partition_places: close: T#%d(%d:%d) place %d " 4852| 0| "partition = [%d,%d]\n", 4853| 0| __kmp_gtid_from_thread(team->t.t_threads[f]), 4854| 0| team->t.t_id, f, place, first_place, last_place)); 4855| 0| } 4856| 0| } else { 4857| 0| int S, rem, gap, s_count; 4858| 0| S = n_th / n_places; 4859| 0| s_count = 0; 4860| 0| rem = n_th - (S * n_places); 4861| 0| gap = rem > 0 ? n_places / rem : n_places; ------------------ | Branch (4861:13): [True: 0, False: 0] ------------------ 4862| 0| int place = masters_place; 4863| 0| int gap_ct = gap; 4864| 0| for (f = 0; f < n_th; f++) { ------------------ | Branch (4864:19): [True: 0, False: 0] ------------------ 4865| 0| kmp_info_t *th = team->t.t_threads[f]; 4866| 0| KMP_DEBUG_ASSERT(th != NULL); 4867| | 4868| 0| th->th.th_first_place = first_place; 4869| 0| th->th.th_last_place = last_place; 4870| 0| th->th.th_new_place = place; 4871| 0| if (__kmp_display_affinity && place != th->th.th_current_place && ------------------ | Branch (4871:13): [True: 0, False: 0] | Branch (4871:39): [True: 0, False: 0] ------------------ 4872| 0| team->t.t_display_affinity != 1) { ------------------ | Branch (4872:13): [True: 0, False: 0] ------------------ 4873| 0| team->t.t_display_affinity = 1; 4874| 0| } 4875| 0| s_count++; 4876| | 4877| 0| if ((s_count == S) && rem && (gap_ct == gap)) { ------------------ | Branch (4877:13): [True: 0, False: 0] | Branch (4877:31): [True: 0, False: 0] | Branch (4877:38): [True: 0, False: 0] ------------------ 4878| | // do nothing, add an extra thread to place on next iteration 4879| 0| } else if ((s_count == S + 1) && rem && (gap_ct == gap)) { ------------------ | Branch (4879:20): [True: 0, False: 0] | Branch (4879:42): [True: 0, False: 0] | Branch (4879:49): [True: 0, False: 0] ------------------ 4880| | // we added an extra thread to this place; move to next place 4881| 0| if (place == last_place) { ------------------ | Branch (4881:15): [True: 0, False: 0] ------------------ 4882| 0| place = first_place; 4883| 0| } else if (place == (num_masks - 1)) { ------------------ | Branch (4883:22): [True: 0, False: 0] ------------------ 4884| 0| place = 0; 4885| 0| } else { 4886| 0| place++; 4887| 0| } 4888| 0| s_count = 0; 4889| 0| gap_ct = 1; 4890| 0| rem--; 4891| 0| } else if (s_count == S) { // place full; don't add extra ------------------ | Branch (4891:20): [True: 0, False: 0] ------------------ 4892| 0| if (place == last_place) { ------------------ | Branch (4892:15): [True: 0, False: 0] ------------------ 4893| 0| place = first_place; 4894| 0| } else if (place == (num_masks - 1)) { ------------------ | Branch (4894:22): [True: 0, False: 0] ------------------ 4895| 0| place = 0; 4896| 0| } else { 4897| 0| place++; 4898| 0| } 4899| 0| gap_ct++; 4900| 0| s_count = 0; 4901| 0| } 4902| | 4903| 0| KA_TRACE(100, 4904| 0| ("__kmp_partition_places: close: T#%d(%d:%d) place %d " 4905| 0| "partition = [%d,%d]\n", 4906| 0| __kmp_gtid_from_thread(team->t.t_threads[f]), team->t.t_id, f, 4907| 0| th->th.th_new_place, first_place, last_place)); 4908| 0| } 4909| 0| KMP_DEBUG_ASSERT(place == masters_place); 4910| 0| } 4911| 0| } break; 4912| | 4913| 0| case proc_bind_spread: { ------------------ | Branch (4913:3): [True: 0, False: 1] ------------------ 4914| 0| int f; 4915| 0| int n_th = team->t.t_nproc; 4916| 0| int n_places; 4917| 0| int thidx; 4918| 0| if (first_place <= last_place) { ------------------ | Branch (4918:9): [True: 0, False: 0] ------------------ 4919| 0| n_places = last_place - first_place + 1; 4920| 0| } else { 4921| 0| n_places = num_masks - first_place + last_place + 1; 4922| 0| } 4923| 0| if (n_th <= n_places) { ------------------ | Branch (4923:9): [True: 0, False: 0] ------------------ 4924| 0| int place = -1; 4925| | 4926| 0| if (n_places != num_masks) { ------------------ | Branch (4926:11): [True: 0, False: 0] ------------------ 4927| 0| int S = n_places / n_th; 4928| 0| int s_count, rem, gap, gap_ct; 4929| | 4930| 0| place = masters_place; 4931| 0| rem = n_places - n_th * S; 4932| 0| gap = rem ? n_th / rem : 1; ------------------ | Branch (4932:15): [True: 0, False: 0] ------------------ 4933| 0| gap_ct = gap; 4934| 0| thidx = n_th; 4935| 0| if (update_master_only == 1) ------------------ | Branch (4935:13): [True: 0, False: 0] ------------------ 4936| 0| thidx = 1; 4937| 0| for (f = 0; f < thidx; f++) { ------------------ | Branch (4937:21): [True: 0, False: 0] ------------------ 4938| 0| kmp_info_t *th = team->t.t_threads[f]; 4939| 0| KMP_DEBUG_ASSERT(th != NULL); 4940| | 4941| 0| th->th.th_first_place = place; 4942| 0| th->th.th_new_place = place; 4943| 0| if (__kmp_display_affinity && place != th->th.th_current_place && ------------------ | Branch (4943:15): [True: 0, False: 0] | Branch (4943:41): [True: 0, False: 0] ------------------ 4944| 0| team->t.t_display_affinity != 1) { ------------------ | Branch (4944:15): [True: 0, False: 0] ------------------ 4945| 0| team->t.t_display_affinity = 1; 4946| 0| } 4947| 0| s_count = 1; 4948| 0| while (s_count < S) { ------------------ | Branch (4948:18): [True: 0, False: 0] ------------------ 4949| 0| if (place == last_place) { ------------------ | Branch (4949:17): [True: 0, False: 0] ------------------ 4950| 0| place = first_place; 4951| 0| } else if (place == (num_masks - 1)) { ------------------ | Branch (4951:24): [True: 0, False: 0] ------------------ 4952| 0| place = 0; 4953| 0| } else { 4954| 0| place++; 4955| 0| } 4956| 0| s_count++; 4957| 0| } 4958| 0| if (rem && (gap_ct == gap)) { ------------------ | Branch (4958:15): [True: 0, False: 0] | Branch (4958:22): [True: 0, False: 0] ------------------ 4959| 0| if (place == last_place) { ------------------ | Branch (4959:17): [True: 0, False: 0] ------------------ 4960| 0| place = first_place; 4961| 0| } else if (place == (num_masks - 1)) { ------------------ | Branch (4961:24): [True: 0, False: 0] ------------------ 4962| 0| place = 0; 4963| 0| } else { 4964| 0| place++; 4965| 0| } 4966| 0| rem--; 4967| 0| gap_ct = 0; 4968| 0| } 4969| 0| th->th.th_last_place = place; 4970| 0| gap_ct++; 4971| | 4972| 0| if (place == last_place) { ------------------ | Branch (4972:15): [True: 0, False: 0] ------------------ 4973| 0| place = first_place; 4974| 0| } else if (place == (num_masks - 1)) { ------------------ | Branch (4974:22): [True: 0, False: 0] ------------------ 4975| 0| place = 0; 4976| 0| } else { 4977| 0| place++; 4978| 0| } 4979| | 4980| 0| KA_TRACE(100, 4981| 0| ("__kmp_partition_places: spread: T#%d(%d:%d) place %d " 4982| 0| "partition = [%d,%d], num_masks: %u\n", 4983| 0| __kmp_gtid_from_thread(team->t.t_threads[f]), team->t.t_id, 4984| 0| f, th->th.th_new_place, th->th.th_first_place, 4985| 0| th->th.th_last_place, num_masks)); 4986| 0| } 4987| 0| } else { 4988| | /* Having uniform space of available computation places I can create 4989| | T partitions of round(P/T) size and put threads into the first 4990| | place of each partition. */ 4991| 0| double current = static_cast(masters_place); 4992| 0| double spacing = 4993| 0| (static_cast(n_places + 1) / static_cast(n_th)); 4994| 0| int first, last; 4995| 0| kmp_info_t *th; 4996| | 4997| 0| thidx = n_th + 1; 4998| 0| if (update_master_only == 1) ------------------ | Branch (4998:13): [True: 0, False: 0] ------------------ 4999| 0| thidx = 1; 5000| 0| for (f = 0; f < thidx; f++) { ------------------ | Branch (5000:21): [True: 0, False: 0] ------------------ 5001| 0| first = static_cast(current); 5002| 0| last = static_cast(current + spacing) - 1; 5003| 0| KMP_DEBUG_ASSERT(last >= first); 5004| 0| if (first >= n_places) { ------------------ | Branch (5004:15): [True: 0, False: 0] ------------------ 5005| 0| if (masters_place) { ------------------ | Branch (5005:17): [True: 0, False: 0] ------------------ 5006| 0| first -= n_places; 5007| 0| last -= n_places; 5008| 0| if (first == (masters_place + 1)) { ------------------ | Branch (5008:19): [True: 0, False: 0] ------------------ 5009| 0| KMP_DEBUG_ASSERT(f == n_th); 5010| 0| first--; 5011| 0| } 5012| 0| if (last == masters_place) { ------------------ | Branch (5012:19): [True: 0, False: 0] ------------------ 5013| 0| KMP_DEBUG_ASSERT(f == (n_th - 1)); 5014| 0| last--; 5015| 0| } 5016| 0| } else { 5017| 0| KMP_DEBUG_ASSERT(f == n_th); 5018| 0| first = 0; 5019| 0| last = 0; 5020| 0| } 5021| 0| } 5022| 0| if (last >= n_places) { ------------------ | Branch (5022:15): [True: 0, False: 0] ------------------ 5023| 0| last = (n_places - 1); 5024| 0| } 5025| 0| place = first; 5026| 0| current += spacing; 5027| 0| if (f < n_th) { ------------------ | Branch (5027:15): [True: 0, False: 0] ------------------ 5028| 0| KMP_DEBUG_ASSERT(0 <= first); 5029| 0| KMP_DEBUG_ASSERT(n_places > first); 5030| 0| KMP_DEBUG_ASSERT(0 <= last); 5031| 0| KMP_DEBUG_ASSERT(n_places > last); 5032| 0| KMP_DEBUG_ASSERT(last_place >= first_place); 5033| 0| th = team->t.t_threads[f]; 5034| 0| KMP_DEBUG_ASSERT(th); 5035| 0| th->th.th_first_place = first; 5036| 0| th->th.th_new_place = place; 5037| 0| th->th.th_last_place = last; 5038| 0| if (__kmp_display_affinity && place != th->th.th_current_place && ------------------ | Branch (5038:17): [True: 0, False: 0] | Branch (5038:43): [True: 0, False: 0] ------------------ 5039| 0| team->t.t_display_affinity != 1) { ------------------ | Branch (5039:17): [True: 0, False: 0] ------------------ 5040| 0| team->t.t_display_affinity = 1; 5041| 0| } 5042| 0| KA_TRACE(100, 5043| 0| ("__kmp_partition_places: spread: T#%d(%d:%d) place %d " 5044| 0| "partition = [%d,%d], spacing = %.4f\n", 5045| 0| __kmp_gtid_from_thread(team->t.t_threads[f]), 5046| 0| team->t.t_id, f, th->th.th_new_place, 5047| 0| th->th.th_first_place, th->th.th_last_place, spacing)); 5048| 0| } 5049| 0| } 5050| 0| } 5051| 0| KMP_DEBUG_ASSERT(update_master_only || place == masters_place); 5052| 0| } else { 5053| 0| int S, rem, gap, s_count; 5054| 0| S = n_th / n_places; 5055| 0| s_count = 0; 5056| 0| rem = n_th - (S * n_places); 5057| 0| gap = rem > 0 ? n_places / rem : n_places; ------------------ | Branch (5057:13): [True: 0, False: 0] ------------------ 5058| 0| int place = masters_place; 5059| 0| int gap_ct = gap; 5060| 0| thidx = n_th; 5061| 0| if (update_master_only == 1) ------------------ | Branch (5061:11): [True: 0, False: 0] ------------------ 5062| 0| thidx = 1; 5063| 0| for (f = 0; f < thidx; f++) { ------------------ | Branch (5063:19): [True: 0, False: 0] ------------------ 5064| 0| kmp_info_t *th = team->t.t_threads[f]; 5065| 0| KMP_DEBUG_ASSERT(th != NULL); 5066| | 5067| 0| th->th.th_first_place = place; 5068| 0| th->th.th_last_place = place; 5069| 0| th->th.th_new_place = place; 5070| 0| if (__kmp_display_affinity && place != th->th.th_current_place && ------------------ | Branch (5070:13): [True: 0, False: 0] | Branch (5070:39): [True: 0, False: 0] ------------------ 5071| 0| team->t.t_display_affinity != 1) { ------------------ | Branch (5071:13): [True: 0, False: 0] ------------------ 5072| 0| team->t.t_display_affinity = 1; 5073| 0| } 5074| 0| s_count++; 5075| | 5076| 0| if ((s_count == S) && rem && (gap_ct == gap)) { ------------------ | Branch (5076:13): [True: 0, False: 0] | Branch (5076:31): [True: 0, False: 0] | Branch (5076:38): [True: 0, False: 0] ------------------ 5077| | // do nothing, add an extra thread to place on next iteration 5078| 0| } else if ((s_count == S + 1) && rem && (gap_ct == gap)) { ------------------ | Branch (5078:20): [True: 0, False: 0] | Branch (5078:42): [True: 0, False: 0] | Branch (5078:49): [True: 0, False: 0] ------------------ 5079| | // we added an extra thread to this place; move on to next place 5080| 0| if (place == last_place) { ------------------ | Branch (5080:15): [True: 0, False: 0] ------------------ 5081| 0| place = first_place; 5082| 0| } else if (place == (num_masks - 1)) { ------------------ | Branch (5082:22): [True: 0, False: 0] ------------------ 5083| 0| place = 0; 5084| 0| } else { 5085| 0| place++; 5086| 0| } 5087| 0| s_count = 0; 5088| 0| gap_ct = 1; 5089| 0| rem--; 5090| 0| } else if (s_count == S) { // place is full; don't add extra thread ------------------ | Branch (5090:20): [True: 0, False: 0] ------------------ 5091| 0| if (place == last_place) { ------------------ | Branch (5091:15): [True: 0, False: 0] ------------------ 5092| 0| place = first_place; 5093| 0| } else if (place == (num_masks - 1)) { ------------------ | Branch (5093:22): [True: 0, False: 0] ------------------ 5094| 0| place = 0; 5095| 0| } else { 5096| 0| place++; 5097| 0| } 5098| 0| gap_ct++; 5099| 0| s_count = 0; 5100| 0| } 5101| | 5102| 0| KA_TRACE(100, ("__kmp_partition_places: spread: T#%d(%d:%d) place %d " 5103| 0| "partition = [%d,%d]\n", 5104| 0| __kmp_gtid_from_thread(team->t.t_threads[f]), 5105| 0| team->t.t_id, f, th->th.th_new_place, 5106| 0| th->th.th_first_place, th->th.th_last_place)); 5107| 0| } 5108| 0| KMP_DEBUG_ASSERT(update_master_only || place == masters_place); 5109| 0| } 5110| 0| } break; 5111| | 5112| 1| default: ------------------ | Branch (5112:3): [True: 1, False: 0] ------------------ 5113| 1| break; 5114| 1| } 5115| | 5116| 1| KA_TRACE(20, ("__kmp_partition_places: exit T#%d\n", team->t.t_id)); 5117| 1|} kmp_runtime.cpp:_ZL21__kmp_initialize_teamP8kmp_teamiP20kmp_internal_controlP5ident: 4715| 35| ident_t *loc) { 4716| 35| KF_TRACE(10, ("__kmp_initialize_team: enter: team=%p\n", team)); 4717| | 4718| | /* verify */ 4719| 35| KMP_DEBUG_ASSERT(team); 4720| 35| KMP_DEBUG_ASSERT(new_nproc <= team->t.t_max_nproc); 4721| 35| KMP_DEBUG_ASSERT(team->t.t_threads); 4722| 35| KMP_MB(); 4723| | 4724| 35| team->t.t_master_tid = 0; /* not needed */ 4725| | /* team->t.t_master_bar; not needed */ 4726| 35| team->t.t_serialized = new_nproc > 1 ? 0 : 1; ------------------ | Branch (4726:26): [True: 1, False: 34] ------------------ 4727| 35| team->t.t_nproc = new_nproc; 4728| | 4729| | /* team->t.t_parent = NULL; TODO not needed & would mess up hot team */ 4730| 35| team->t.t_next_pool = NULL; 4731| | /* memset( team->t.t_threads, 0, sizeof(kmp_info_t*)*new_nproc ); would mess 4732| | * up hot team */ 4733| | 4734| 35| TCW_SYNC_PTR(team->t.t_pkfn, NULL); /* not needed */ ------------------ | | 1159| 35|#define TCW_SYNC_PTR(a, b) TCW_SYNC_8((a), (b)) | | ------------------ | | | | 1141| 35|#define TCW_SYNC_8(a, b) (a) = (b) | | ------------------ ------------------ 4735| 35| team->t.t_invoke = NULL; /* not needed */ 4736| | 4737| | // TODO???: team->t.t_max_active_levels = new_max_active_levels; 4738| 35| team->t.t_sched.sched = new_icvs->sched.sched; 4739| | 4740| 35|#if KMP_ARCH_X86 || KMP_ARCH_X86_64 4741| 35| team->t.t_fp_control_saved = FALSE; /* not needed */ ------------------ | | 1279| 35|#define FALSE 0 ------------------ 4742| 35| team->t.t_x87_fpu_control_word = 0; /* not needed */ 4743| 35| team->t.t_mxcsr = 0; /* not needed */ 4744| 35|#endif /* KMP_ARCH_X86 || KMP_ARCH_X86_64 */ 4745| | 4746| 35| team->t.t_construct = 0; 4747| | 4748| 35| team->t.t_ordered.dt.t_value = 0; 4749| 35| team->t.t_master_active = FALSE; ------------------ | | 1279| 35|#define FALSE 0 ------------------ 4750| | 4751| |#ifdef KMP_DEBUG 4752| | team->t.t_copypriv_data = NULL; /* not necessary, but nice for debugging */ 4753| |#endif 4754| |#if KMP_OS_WINDOWS 4755| | team->t.t_copyin_counter = 0; /* for barrier-free copyin implementation */ 4756| |#endif 4757| | 4758| 35| team->t.t_control_stack_top = NULL; 4759| | 4760| 35| __kmp_reinitialize_team(team, new_icvs, loc); 4761| | 4762| 35| KMP_MB(); 4763| 35| KF_TRACE(10, ("__kmp_initialize_team: exit: team=%p\n", team)); 4764| 35|} kmp_runtime.cpp:_ZL24__kmp_alloc_argv_entriesiP8kmp_teami: 3170| 1.38k|static void __kmp_alloc_argv_entries(int argc, kmp_team_t *team, int realloc) { 3171| | 3172| 1.38k| KMP_DEBUG_ASSERT(team); 3173| 1.38k| if (!realloc || argc > team->t.t_max_argc) { ------------------ | Branch (3173:7): [True: 34, False: 1.35k] | Branch (3173:19): [True: 0, False: 1.35k] ------------------ 3174| | 3175| 34| KA_TRACE(100, ("__kmp_alloc_argv_entries: team %d: needed entries=%d, " 3176| 34| "current entries=%d\n", 3177| 34| team->t.t_id, argc, (realloc) ? team->t.t_max_argc : 0)); 3178| | /* if previously allocated heap space for args, free them */ 3179| 34| if (realloc && team->t.t_argv != &team->t.t_inline_argv[0]) ------------------ | Branch (3179:9): [True: 0, False: 34] | Branch (3179:20): [True: 0, False: 0] ------------------ 3180| 0| __kmp_free((void *)team->t.t_argv); ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 3181| | 3182| 34| if (argc <= KMP_INLINE_ARGV_ENTRIES) { ------------------ | | 2998| 34|#define KMP_INLINE_ARGV_ENTRIES (int)(KMP_INLINE_ARGV_BYTES / KMP_PTR_SKIP) | | ------------------ | | | | 2990| 34| (4 * CACHE_LINE - \ | | | | ------------------ | | | | | | 108| 34|# define CACHE_LINE 64 | | | | ------------------ | | | | 2991| 34| ((3 * KMP_PTR_SKIP + 2 * sizeof(int) + 2 * sizeof(kmp_int8) + \ | | | | ------------------ | | | | | | 30| 34|#define KMP_PTR_SKIP (sizeof(void *)) | | | | ------------------ | | | | 2992| 34| sizeof(kmp_int16) + sizeof(kmp_uint32)) % \ | | | | 2993| 34| CACHE_LINE)) | | | | ------------------ | | | | | | 108| 34|# define CACHE_LINE 64 | | | | ------------------ | | ------------------ | | #define KMP_INLINE_ARGV_ENTRIES (int)(KMP_INLINE_ARGV_BYTES / KMP_PTR_SKIP) | | ------------------ | | | | 30| 34|#define KMP_PTR_SKIP (sizeof(void *)) | | ------------------ ------------------ | Branch (3182:9): [True: 34, False: 0] ------------------ 3183| | /* use unused space in the cache line for arguments */ 3184| 34| team->t.t_max_argc = KMP_INLINE_ARGV_ENTRIES; ------------------ | | 2998| 34|#define KMP_INLINE_ARGV_ENTRIES (int)(KMP_INLINE_ARGV_BYTES / KMP_PTR_SKIP) | | ------------------ | | | | 2990| 34| (4 * CACHE_LINE - \ | | | | ------------------ | | | | | | 108| 34|# define CACHE_LINE 64 | | | | ------------------ | | | | 2991| 34| ((3 * KMP_PTR_SKIP + 2 * sizeof(int) + 2 * sizeof(kmp_int8) + \ | | | | ------------------ | | | | | | 30| 34|#define KMP_PTR_SKIP (sizeof(void *)) | | | | ------------------ | | | | 2992| 34| sizeof(kmp_int16) + sizeof(kmp_uint32)) % \ | | | | 2993| 34| CACHE_LINE)) | | | | ------------------ | | | | | | 108| 34|# define CACHE_LINE 64 | | | | ------------------ | | ------------------ | | #define KMP_INLINE_ARGV_ENTRIES (int)(KMP_INLINE_ARGV_BYTES / KMP_PTR_SKIP) | | ------------------ | | | | 30| 34|#define KMP_PTR_SKIP (sizeof(void *)) | | ------------------ ------------------ 3185| 34| KA_TRACE(100, ("__kmp_alloc_argv_entries: team %d: inline allocate %d " 3186| 34| "argv entries\n", 3187| 34| team->t.t_id, team->t.t_max_argc)); 3188| 34| team->t.t_argv = &team->t.t_inline_argv[0]; 3189| 34| if (__kmp_storage_map) { ------------------ | Branch (3189:11): [True: 0, False: 34] ------------------ 3190| 0| __kmp_print_storage_map_gtid( 3191| 0| -1, &team->t.t_inline_argv[0], 3192| 0| &team->t.t_inline_argv[KMP_INLINE_ARGV_ENTRIES], ------------------ | | 2998| 0|#define KMP_INLINE_ARGV_ENTRIES (int)(KMP_INLINE_ARGV_BYTES / KMP_PTR_SKIP) | | ------------------ | | | | 2990| 0| (4 * CACHE_LINE - \ | | | | ------------------ | | | | | | 108| 0|# define CACHE_LINE 64 | | | | ------------------ | | | | 2991| 0| ((3 * KMP_PTR_SKIP + 2 * sizeof(int) + 2 * sizeof(kmp_int8) + \ | | | | ------------------ | | | | | | 30| 0|#define KMP_PTR_SKIP (sizeof(void *)) | | | | ------------------ | | | | 2992| 0| sizeof(kmp_int16) + sizeof(kmp_uint32)) % \ | | | | 2993| 0| CACHE_LINE)) | | | | ------------------ | | | | | | 108| 0|# define CACHE_LINE 64 | | | | ------------------ | | ------------------ | | #define KMP_INLINE_ARGV_ENTRIES (int)(KMP_INLINE_ARGV_BYTES / KMP_PTR_SKIP) | | ------------------ | | | | 30| 0|#define KMP_PTR_SKIP (sizeof(void *)) | | ------------------ ------------------ 3193| 0| (sizeof(void *) * KMP_INLINE_ARGV_ENTRIES), "team_%d.t_inline_argv", ------------------ | | 2998| 0|#define KMP_INLINE_ARGV_ENTRIES (int)(KMP_INLINE_ARGV_BYTES / KMP_PTR_SKIP) | | ------------------ | | | | 2990| 0| (4 * CACHE_LINE - \ | | | | ------------------ | | | | | | 108| 0|# define CACHE_LINE 64 | | | | ------------------ | | | | 2991| 0| ((3 * KMP_PTR_SKIP + 2 * sizeof(int) + 2 * sizeof(kmp_int8) + \ | | | | ------------------ | | | | | | 30| 0|#define KMP_PTR_SKIP (sizeof(void *)) | | | | ------------------ | | | | 2992| 0| sizeof(kmp_int16) + sizeof(kmp_uint32)) % \ | | | | 2993| 0| CACHE_LINE)) | | | | ------------------ | | | | | | 108| 0|# define CACHE_LINE 64 | | | | ------------------ | | ------------------ | | #define KMP_INLINE_ARGV_ENTRIES (int)(KMP_INLINE_ARGV_BYTES / KMP_PTR_SKIP) | | ------------------ | | | | 30| 0|#define KMP_PTR_SKIP (sizeof(void *)) | | ------------------ ------------------ 3194| 0| team->t.t_id); 3195| 0| } 3196| 34| } else { 3197| | /* allocate space for arguments in the heap */ 3198| 0| team->t.t_max_argc = (argc <= (KMP_MIN_MALLOC_ARGV_ENTRIES >> 1)) ------------------ | | 2982| 0|#define KMP_MIN_MALLOC_ARGV_ENTRIES 100 ------------------ | Branch (3198:28): [True: 0, False: 0] ------------------ 3199| 0| ? KMP_MIN_MALLOC_ARGV_ENTRIES ------------------ | | 2982| 0|#define KMP_MIN_MALLOC_ARGV_ENTRIES 100 ------------------ 3200| 0| : 2 * argc; 3201| 0| KA_TRACE(100, ("__kmp_alloc_argv_entries: team %d: dynamic allocate %d " 3202| 0| "argv entries\n", 3203| 0| team->t.t_id, team->t.t_max_argc)); 3204| 0| team->t.t_argv = 3205| 0| (void **)__kmp_page_allocate(sizeof(void *) * team->t.t_max_argc); ------------------ | | 3618| 0|#define __kmp_page_allocate(size) ___kmp_page_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3206| 0| if (__kmp_storage_map) { ------------------ | Branch (3206:11): [True: 0, False: 0] ------------------ 3207| 0| __kmp_print_storage_map_gtid(-1, &team->t.t_argv[0], 3208| 0| &team->t.t_argv[team->t.t_max_argc], 3209| 0| sizeof(void *) * team->t.t_max_argc, 3210| 0| "team_%d.t_argv", team->t.t_id); 3211| 0| } 3212| 0| } 3213| 34| } 3214| 1.38k|} kmp_runtime.cpp:_ZL26__kmp_allocate_team_arraysP8kmp_teami: 3216| 34|static void __kmp_allocate_team_arrays(kmp_team_t *team, int max_nth) { 3217| 34| int i; 3218| 34| int num_disp_buff = max_nth > 1 ? __kmp_dispatch_num_buffers : 2; ------------------ | Branch (3218:23): [True: 1, False: 33] ------------------ 3219| 34| team->t.t_threads = 3220| 34| (kmp_info_t **)__kmp_allocate(sizeof(kmp_info_t *) * max_nth); ------------------ | | 3617| 34|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3221| 34| team->t.t_disp_buffer = (dispatch_shared_info_t *)__kmp_allocate( ------------------ | | 3617| 34|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3222| 34| sizeof(dispatch_shared_info_t) * num_disp_buff); 3223| 34| team->t.t_dispatch = 3224| 34| (kmp_disp_t *)__kmp_allocate(sizeof(kmp_disp_t) * max_nth); ------------------ | | 3617| 34|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3225| 34| team->t.t_implicit_task_taskdata = 3226| 34| (kmp_taskdata_t *)__kmp_allocate(sizeof(kmp_taskdata_t) * max_nth); ------------------ | | 3617| 34|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3227| 34| team->t.t_max_nproc = max_nth; 3228| | 3229| | /* setup dispatch buffers */ 3230| 107| for (i = 0; i < num_disp_buff; ++i) { ------------------ | Branch (3230:15): [True: 73, False: 34] ------------------ 3231| 73| team->t.t_disp_buffer[i].buffer_index = i; 3232| 73| team->t.t_disp_buffer[i].doacross_buf_idx = i; 3233| 73| } 3234| 34|} kmp_runtime.cpp:_ZL22__kmp_free_team_arraysP8kmp_team: 3236| 34|static void __kmp_free_team_arrays(kmp_team_t *team) { 3237| | /* Note: this does not free the threads in t_threads (__kmp_free_threads) */ 3238| 34| int i; 3239| 131| for (i = 0; i < team->t.t_max_nproc; ++i) { ------------------ | Branch (3239:15): [True: 97, False: 34] ------------------ 3240| 97| if (team->t.t_dispatch[i].th_disp_buffer != NULL) { ------------------ | Branch (3240:9): [True: 33, False: 64] ------------------ 3241| 33| __kmp_free(team->t.t_dispatch[i].th_disp_buffer); ------------------ | | 3619| 33|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 3242| 33| team->t.t_dispatch[i].th_disp_buffer = NULL; 3243| 33| } 3244| 97| } 3245| |#if KMP_USE_HIER_SCHED 3246| | __kmp_dispatch_free_hierarchies(team); 3247| |#endif 3248| 34| __kmp_free(team->t.t_threads); ------------------ | | 3619| 34|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 3249| 34| __kmp_free(team->t.t_disp_buffer); ------------------ | | 3619| 34|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 3250| 34| __kmp_free(team->t.t_dispatch); ------------------ | | 3619| 34|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 3251| 34| __kmp_free(team->t.t_implicit_task_taskdata); ------------------ | | 3619| 34|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 3252| 34| team->t.t_threads = NULL; 3253| 34| team->t.t_disp_buffer = NULL; 3254| | team->t.t_dispatch = NULL; 3255| 34| team->t.t_implicit_task_taskdata = 0; 3256| 34|} kmp_runtime.cpp:_ZL19__kmp_itthash_cleanP8kmp_info: 6281| 1|static void __kmp_itthash_clean(kmp_info_t *th) { 6282| |#if USE_ITT_NOTIFY 6283| | if (__kmp_itt_region_domains.count > 0) { 6284| | for (int i = 0; i < KMP_MAX_FRAME_DOMAINS; ++i) { 6285| | kmp_itthash_entry_t *bucket = __kmp_itt_region_domains.buckets[i]; 6286| | while (bucket) { 6287| | kmp_itthash_entry_t *next = bucket->next_in_bucket; 6288| | __kmp_thread_free(th, bucket); 6289| | bucket = next; 6290| | } 6291| | } 6292| | } 6293| | if (__kmp_itt_barrier_domains.count > 0) { 6294| | for (int i = 0; i < KMP_MAX_FRAME_DOMAINS; ++i) { 6295| | kmp_itthash_entry_t *bucket = __kmp_itt_barrier_domains.buckets[i]; 6296| | while (bucket) { 6297| | kmp_itthash_entry_t *next = bucket->next_in_bucket; 6298| | __kmp_thread_free(th, bucket); 6299| | bucket = next; 6300| | } 6301| | } 6302| | } 6303| |#endif 6304| 1|} kmp_runtime.cpp:_ZL18__kmp_internal_endv: 6306| 1|static void __kmp_internal_end(void) { 6307| 1| int i; 6308| | 6309| | /* First, unregister the library */ 6310| 1| __kmp_unregister_library(); 6311| | 6312| |#if KMP_OS_WINDOWS 6313| | /* In Win static library, we can't tell when a root actually dies, so we 6314| | reclaim the data structures for any root threads that have died but not 6315| | unregistered themselves, in order to shut down cleanly. 6316| | In Win dynamic library we also can't tell when a thread dies. */ 6317| | __kmp_reclaim_dead_roots(); // AC: moved here to always clean resources of 6318| |// dead roots 6319| |#endif 6320| | 6321| 137| for (i = 0; i < __kmp_threads_capacity; i++) ------------------ | Branch (6321:15): [True: 136, False: 1] ------------------ 6322| 136| if (__kmp_root[i]) ------------------ | Branch (6322:9): [True: 1, False: 135] ------------------ 6323| 1| if (__kmp_root[i]->r.r_active) ------------------ | Branch (6323:11): [True: 0, False: 1] ------------------ 6324| 0| break; 6325| 1| KMP_MB(); /* Flush all pending memory write invalidates. */ 6326| 1| TCW_SYNC_4(__kmp_global.g.g_done, TRUE); ------------------ | | 1136| 1|#define TCW_SYNC_4(a, b) (a) = (b) ------------------ 6327| | 6328| 1| if (i < __kmp_threads_capacity) { ------------------ | Branch (6328:7): [True: 0, False: 1] ------------------ 6329| |#if KMP_USE_MONITOR 6330| | // 2009-09-08 (lev): Other alive roots found. Why do we kill the monitor?? 6331| | KMP_MB(); /* Flush all pending memory write invalidates. */ 6332| | 6333| | // Need to check that monitor was initialized before reaping it. If we are 6334| | // called form __kmp_atfork_child (which sets __kmp_init_parallel = 0), then 6335| | // __kmp_monitor will appear to contain valid data, but it is only valid in 6336| | // the parent process, not the child. 6337| | // New behavior (201008): instead of keying off of the flag 6338| | // __kmp_init_parallel, the monitor thread creation is keyed off 6339| | // of the new flag __kmp_init_monitor. 6340| | __kmp_acquire_bootstrap_lock(&__kmp_monitor_lock); 6341| | if (TCR_4(__kmp_init_monitor)) { 6342| | __kmp_reap_monitor(&__kmp_monitor); 6343| | TCW_4(__kmp_init_monitor, 0); 6344| | } 6345| | __kmp_release_bootstrap_lock(&__kmp_monitor_lock); 6346| | KA_TRACE(10, ("__kmp_internal_end: monitor reaped\n")); 6347| |#endif // KMP_USE_MONITOR 6348| 1| } else { 6349| |/* TODO move this to cleanup code */ 6350| |#ifdef KMP_DEBUG 6351| | /* make sure that everything has properly ended */ 6352| | for (i = 0; i < __kmp_threads_capacity; i++) { 6353| | if (__kmp_root[i]) { 6354| | // KMP_ASSERT( ! KMP_UBER_GTID( i ) ); // AC: 6355| | // there can be uber threads alive here 6356| | KMP_ASSERT(!__kmp_root[i]->r.r_active); // TODO: can they be active? 6357| | } 6358| | } 6359| |#endif 6360| | 6361| 1| KMP_MB(); 6362| | 6363| | // Reap the worker threads. 6364| | // This is valid for now, but be careful if threads are reaped sooner. 6365| 32| while (__kmp_thread_pool != NULL) { // Loop thru all the thread in the pool. ------------------ | Branch (6365:12): [True: 31, False: 1] ------------------ 6366| | // Get the next thread from the pool. 6367| 31| kmp_info_t *thread = CCAST(kmp_info_t *, __kmp_thread_pool); ------------------ | | 229| 31|#define CCAST(type, var) const_cast(var) ------------------ 6368| 31| __kmp_thread_pool = thread->th.th_next_pool; 6369| | // Reap it. 6370| 31| KMP_DEBUG_ASSERT(thread->th.th_reap_state == KMP_SAFE_TO_REAP); 6371| 31| thread->th.th_next_pool = NULL; 6372| 31| thread->th.th_in_pool = FALSE; ------------------ | | 1279| 31|#define FALSE 0 ------------------ 6373| 31| __kmp_reap_thread(thread, 0); 6374| 31| } 6375| 1| __kmp_thread_pool_insert_pt = NULL; 6376| | 6377| | // Reap teams. 6378| 3| while (__kmp_team_pool != NULL) { // Loop thru all the teams in the pool. ------------------ | Branch (6378:12): [True: 2, False: 1] ------------------ 6379| | // Get the next team from the pool. 6380| 2| kmp_team_t *team = CCAST(kmp_team_t *, __kmp_team_pool); ------------------ | | 229| 2|#define CCAST(type, var) const_cast(var) ------------------ 6381| 2| __kmp_team_pool = team->t.t_next_pool; 6382| | // Reap it. 6383| 2| team->t.t_next_pool = NULL; 6384| 2| __kmp_reap_team(team); 6385| 2| } 6386| | 6387| 1| __kmp_reap_task_teams(); 6388| | 6389| 1|#if KMP_OS_UNIX 6390| | // Threads that are not reaped should not access any resources since they 6391| | // are going to be deallocated soon, so the shutdown sequence should wait 6392| | // until all threads either exit the final spin-waiting loop or begin 6393| | // sleeping after the given blocktime. 6394| 137| for (i = 0; i < __kmp_threads_capacity; i++) { ------------------ | Branch (6394:17): [True: 136, False: 1] ------------------ 6395| 136| kmp_info_t *thr = __kmp_threads[i]; 6396| 136| while (thr && KMP_ATOMIC_LD_ACQ(&thr->th.th_blocking)) ------------------ | | 1249| 0|#define KMP_ATOMIC_LD_ACQ(p) KMP_ATOMIC_LD(p, acquire) | | ------------------ | | | | 1245| 0|#define KMP_ATOMIC_LD(p, order) (p)->load(std::memory_order_##order) | | | | ------------------ | | | | | Branch (1245:33): [True: 0, False: 0] | | | | ------------------ | | ------------------ ------------------ | Branch (6396:14): [True: 0, False: 136] ------------------ 6397| 0| KMP_CPU_PAUSE(); ------------------ | | 1498| 0|#define KMP_CPU_PAUSE() __kmp_x86_pause() ------------------ 6398| 136| } 6399| 1|#endif 6400| | 6401| 137| for (i = 0; i < __kmp_threads_capacity; ++i) { ------------------ | Branch (6401:17): [True: 136, False: 1] ------------------ 6402| | // TBD: Add some checking... 6403| | // Something like KMP_DEBUG_ASSERT( __kmp_thread[ i ] == NULL ); 6404| 136| } 6405| | 6406| | /* Make sure all threadprivate destructors get run by joining with all 6407| | worker threads before resetting this flag */ 6408| 1| TCW_SYNC_4(__kmp_init_common, FALSE); ------------------ | | 1136| 1|#define TCW_SYNC_4(a, b) (a) = (b) ------------------ 6409| | 6410| 1| KA_TRACE(10, ("__kmp_internal_end: all workers reaped\n")); 6411| 1| KMP_MB(); 6412| | 6413| |#if KMP_USE_MONITOR 6414| | // See note above: One of the possible fixes for CQ138434 / CQ140126 6415| | // 6416| | // FIXME: push both code fragments down and CSE them? 6417| | // push them into __kmp_cleanup() ? 6418| | __kmp_acquire_bootstrap_lock(&__kmp_monitor_lock); 6419| | if (TCR_4(__kmp_init_monitor)) { 6420| | __kmp_reap_monitor(&__kmp_monitor); 6421| | TCW_4(__kmp_init_monitor, 0); 6422| | } 6423| | __kmp_release_bootstrap_lock(&__kmp_monitor_lock); 6424| | KA_TRACE(10, ("__kmp_internal_end: monitor reaped\n")); 6425| |#endif 6426| 1| } /* else !__kmp_global.t_active */ 6427| 1| TCW_4(__kmp_init_gtid, FALSE); ------------------ | | 1128| 1|#define TCW_4(a, b) (a) = (b) ------------------ 6428| 1| KMP_MB(); /* Flush all pending memory write invalidates. */ 6429| | 6430| 1| __kmp_cleanup(); 6431| 1|#if OMPT_SUPPORT 6432| 1| ompt_fini(); 6433| 1|#endif 6434| 1|} kmp_runtime.cpp:_ZL20__kmp_fini_allocatorv: 554| 1|static void __kmp_fini_allocator() { __kmp_fini_memkind(); } kmp_runtime.cpp:_ZL21__kmp_reg_status_namev: 6701| 2|static inline char *__kmp_reg_status_name() { 6702| |/* On RHEL 3u5 if linked statically, getpid() returns different values in 6703| | each thread. If registration and unregistration go in different threads 6704| | (omp_misc_other_root_exit.cpp test case), the name of registered_lib_env 6705| | env var can not be found, because the name will contain different pid. */ 6706| |// macOS* complains about name being too long with additional getuid() 6707| |#if KMP_OS_UNIX && !KMP_OS_DARWIN && KMP_DYNAMIC_LIB 6708| | return __kmp_str_format("__KMP_REGISTERED_LIB_%d_%d", (int)getpid(), 6709| | (int)getuid()); 6710| |#else 6711| 2| return __kmp_str_format("__KMP_REGISTERED_LIB_%d", (int)getpid()); 6712| 2|#endif 6713| 2|} // __kmp_reg_status_get kmp_runtime.cpp:_ZL26__kmp_do_serial_initializev: 7006| 1|static void __kmp_do_serial_initialize(void) { 7007| 1| int i, gtid; 7008| 1| size_t size; 7009| | 7010| 1| KA_TRACE(10, ("__kmp_do_serial_initialize: enter\n")); 7011| | 7012| 1| KMP_DEBUG_ASSERT(sizeof(kmp_int32) == 4); 7013| 1| KMP_DEBUG_ASSERT(sizeof(kmp_uint32) == 4); 7014| 1| KMP_DEBUG_ASSERT(sizeof(kmp_int64) == 8); 7015| 1| KMP_DEBUG_ASSERT(sizeof(kmp_uint64) == 8); 7016| 1| KMP_DEBUG_ASSERT(sizeof(kmp_intptr_t) == sizeof(void *)); 7017| | 7018| 1|#if OMPT_SUPPORT 7019| 1| ompt_pre_init(); 7020| 1|#endif 7021| 1|#if OMPD_SUPPORT 7022| 1| __kmp_env_dump(); 7023| 1| ompd_init(); 7024| 1|#endif 7025| | 7026| 1| __kmp_validate_locks(); 7027| | 7028| |#if ENABLE_LIBOMPTARGET 7029| | /* Initialize functions from libomptarget */ 7030| | __kmp_init_omptarget(); 7031| |#endif 7032| | 7033| | /* Initialize internal memory allocator */ 7034| 1| __kmp_init_allocator(); 7035| | 7036| | /* Register the library startup via an environment variable or via mapped 7037| | shared memory file and check to see whether another copy of the library is 7038| | already registered. Since forked child process is often terminated, we 7039| | postpone the registration till middle initialization in the child */ 7040| 1| if (__kmp_need_register_serial) ------------------ | Branch (7040:7): [True: 1, False: 0] ------------------ 7041| 1| __kmp_register_library_startup(); 7042| | 7043| | /* TODO reinitialization of library */ 7044| 1| if (TCR_4(__kmp_global.g.g_done)) { ------------------ | | 1127| 1|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 1] | | ------------------ ------------------ 7045| 0| KA_TRACE(10, ("__kmp_do_serial_initialize: reinitialization of library\n")); 7046| 0| } 7047| | 7048| 1| __kmp_global.g.g_abort = 0; 7049| 1| TCW_SYNC_4(__kmp_global.g.g_done, FALSE); ------------------ | | 1136| 1|#define TCW_SYNC_4(a, b) (a) = (b) ------------------ 7050| | 7051| |/* initialize the locks */ 7052| 1|#if KMP_USE_ADAPTIVE_LOCKS 7053| |#if KMP_DEBUG_ADAPTIVE_LOCKS 7054| | __kmp_init_speculative_stats(); 7055| |#endif 7056| 1|#endif 7057| |#if KMP_STATS_ENABLED 7058| | __kmp_stats_init(); 7059| |#endif 7060| 1| __kmp_init_lock(&__kmp_global_lock); 7061| 1| __kmp_init_queuing_lock(&__kmp_dispatch_lock); 7062| 1| __kmp_init_lock(&__kmp_debug_lock); 7063| 1| __kmp_init_atomic_lock(&__kmp_atomic_lock); 7064| 1| __kmp_init_atomic_lock(&__kmp_atomic_lock_1i); 7065| 1| __kmp_init_atomic_lock(&__kmp_atomic_lock_2i); 7066| 1| __kmp_init_atomic_lock(&__kmp_atomic_lock_4i); 7067| 1| __kmp_init_atomic_lock(&__kmp_atomic_lock_4r); 7068| 1| __kmp_init_atomic_lock(&__kmp_atomic_lock_8i); 7069| 1| __kmp_init_atomic_lock(&__kmp_atomic_lock_8r); 7070| 1| __kmp_init_atomic_lock(&__kmp_atomic_lock_8c); 7071| 1| __kmp_init_atomic_lock(&__kmp_atomic_lock_10r); 7072| 1| __kmp_init_atomic_lock(&__kmp_atomic_lock_16r); 7073| 1| __kmp_init_atomic_lock(&__kmp_atomic_lock_16c); 7074| 1| __kmp_init_atomic_lock(&__kmp_atomic_lock_20c); 7075| 1| __kmp_init_atomic_lock(&__kmp_atomic_lock_32c); 7076| 1| __kmp_init_bootstrap_lock(&__kmp_forkjoin_lock); 7077| 1| __kmp_init_bootstrap_lock(&__kmp_exit_lock); 7078| |#if KMP_USE_MONITOR 7079| | __kmp_init_bootstrap_lock(&__kmp_monitor_lock); 7080| |#endif 7081| 1| __kmp_init_bootstrap_lock(&__kmp_tp_cached_lock); 7082| | 7083| | /* conduct initialization and initial setup of configuration */ 7084| | 7085| 1| __kmp_runtime_initialize(); 7086| | 7087| 1|#if KMP_MIC_SUPPORTED 7088| 1| __kmp_check_mic_type(); 7089| 1|#endif 7090| | 7091| |// Some global variable initialization moved here from kmp_env_initialize() 7092| |#ifdef KMP_DEBUG 7093| | kmp_diag = 0; 7094| |#endif 7095| 1| __kmp_abort_delay = 0; 7096| | 7097| | // From __kmp_init_dflt_team_nth() 7098| | /* assume the entire machine will be used */ 7099| 1| __kmp_dflt_team_nth_ub = __kmp_xproc; 7100| 1| if (__kmp_dflt_team_nth_ub < KMP_MIN_NTH) { ------------------ | | 1137| 1|#define KMP_MIN_NTH 1 ------------------ | Branch (7100:7): [True: 0, False: 1] ------------------ 7101| 0| __kmp_dflt_team_nth_ub = KMP_MIN_NTH; ------------------ | | 1137| 0|#define KMP_MIN_NTH 1 ------------------ 7102| 0| } 7103| 1| if (__kmp_dflt_team_nth_ub > __kmp_sys_max_nth) { ------------------ | Branch (7103:7): [True: 0, False: 1] ------------------ 7104| 0| __kmp_dflt_team_nth_ub = __kmp_sys_max_nth; 7105| 0| } 7106| 1| __kmp_max_nth = __kmp_sys_max_nth; 7107| 1| __kmp_cg_max_nth = __kmp_sys_max_nth; 7108| 1| __kmp_teams_max_nth = __kmp_xproc; // set a "reasonable" default 7109| 1| if (__kmp_teams_max_nth > __kmp_sys_max_nth) { ------------------ | Branch (7109:7): [True: 0, False: 1] ------------------ 7110| 0| __kmp_teams_max_nth = __kmp_sys_max_nth; 7111| 0| } 7112| | 7113| | // Three vars below moved here from __kmp_env_initialize() "KMP_BLOCKTIME" 7114| | // part 7115| 1| __kmp_dflt_blocktime = KMP_DEFAULT_BLOCKTIME; ------------------ | | 1187| 1|#define KMP_DEFAULT_BLOCKTIME (__kmp_is_hybrid_cpu() ? (0) : (200)) | | ------------------ | | | Branch (1187:32): [True: 0, False: 1] | | ------------------ ------------------ 7116| |#if KMP_USE_MONITOR 7117| | __kmp_monitor_wakeups = 7118| | KMP_WAKEUPS_FROM_BLOCKTIME(__kmp_dflt_blocktime, __kmp_monitor_wakeups); 7119| | __kmp_bt_intervals = 7120| | KMP_INTERVALS_FROM_BLOCKTIME(__kmp_dflt_blocktime, __kmp_monitor_wakeups); 7121| |#endif 7122| | // From "KMP_LIBRARY" part of __kmp_env_initialize() 7123| 1| __kmp_library = library_throughput; 7124| | // From KMP_SCHEDULE initialization 7125| 1| __kmp_static = kmp_sch_static_balanced; 7126| |// AC: do not use analytical here, because it is non-monotonous 7127| |//__kmp_guided = kmp_sch_guided_iterative_chunked; 7128| |//__kmp_auto = kmp_sch_guided_analytical_chunked; // AC: it is the default, no 7129| |// need to repeat assignment 7130| |// Barrier initialization. Moved here from __kmp_env_initialize() Barrier branch 7131| |// bit control and barrier method control parts 7132| 1|#if KMP_FAST_REDUCTION_BARRIER 7133| 1|#define kmp_reduction_barrier_gather_bb ((int)1) 7134| 1|#define kmp_reduction_barrier_release_bb ((int)1) 7135| 1|#define kmp_reduction_barrier_gather_pat __kmp_barrier_gather_pat_dflt 7136| 1|#define kmp_reduction_barrier_release_pat __kmp_barrier_release_pat_dflt 7137| 1|#endif // KMP_FAST_REDUCTION_BARRIER 7138| 4| for (i = bs_plain_barrier; i < bs_last_barrier; i++) { ------------------ | Branch (7138:30): [True: 3, False: 1] ------------------ 7139| 3| __kmp_barrier_gather_branch_bits[i] = __kmp_barrier_gather_bb_dflt; 7140| 3| __kmp_barrier_release_branch_bits[i] = __kmp_barrier_release_bb_dflt; 7141| 3| __kmp_barrier_gather_pattern[i] = __kmp_barrier_gather_pat_dflt; 7142| 3| __kmp_barrier_release_pattern[i] = __kmp_barrier_release_pat_dflt; 7143| 3|#if KMP_FAST_REDUCTION_BARRIER 7144| 3| if (i == bs_reduction_barrier) { // tested and confirmed on ALTIX only ( ------------------ | Branch (7144:9): [True: 1, False: 2] ------------------ 7145| | // lin_64 ): hyper,1 7146| 1| __kmp_barrier_gather_branch_bits[i] = kmp_reduction_barrier_gather_bb; ------------------ | | 7133| 1|#define kmp_reduction_barrier_gather_bb ((int)1) ------------------ 7147| 1| __kmp_barrier_release_branch_bits[i] = kmp_reduction_barrier_release_bb; ------------------ | | 7134| 1|#define kmp_reduction_barrier_release_bb ((int)1) ------------------ 7148| 1| __kmp_barrier_gather_pattern[i] = kmp_reduction_barrier_gather_pat; ------------------ | | 7135| 1|#define kmp_reduction_barrier_gather_pat __kmp_barrier_gather_pat_dflt ------------------ 7149| 1| __kmp_barrier_release_pattern[i] = kmp_reduction_barrier_release_pat; ------------------ | | 7136| 1|#define kmp_reduction_barrier_release_pat __kmp_barrier_release_pat_dflt ------------------ 7150| 1| } 7151| 3|#endif // KMP_FAST_REDUCTION_BARRIER 7152| 3| } 7153| 1|#if KMP_FAST_REDUCTION_BARRIER 7154| 1|#undef kmp_reduction_barrier_release_pat 7155| 1|#undef kmp_reduction_barrier_gather_pat 7156| 1|#undef kmp_reduction_barrier_release_bb 7157| 1|#undef kmp_reduction_barrier_gather_bb 7158| 1|#endif // KMP_FAST_REDUCTION_BARRIER 7159| 1|#if KMP_MIC_SUPPORTED 7160| 1| if (__kmp_mic_type == mic2) { // KNC ------------------ | Branch (7160:7): [True: 0, False: 1] ------------------ 7161| | // AC: plane=3,2, forkjoin=2,1 are optimal for 240 threads on KNC 7162| 0| __kmp_barrier_gather_branch_bits[bs_plain_barrier] = 3; // plain gather 7163| 0| __kmp_barrier_release_branch_bits[bs_forkjoin_barrier] = 7164| 0| 1; // forkjoin release 7165| 0| __kmp_barrier_gather_pattern[bs_forkjoin_barrier] = bp_hierarchical_bar; 7166| 0| __kmp_barrier_release_pattern[bs_forkjoin_barrier] = bp_hierarchical_bar; 7167| 0| } 7168| 1|#if KMP_FAST_REDUCTION_BARRIER 7169| 1| if (__kmp_mic_type == mic2) { // KNC ------------------ | Branch (7169:7): [True: 0, False: 1] ------------------ 7170| 0| __kmp_barrier_gather_pattern[bs_reduction_barrier] = bp_hierarchical_bar; 7171| 0| __kmp_barrier_release_pattern[bs_reduction_barrier] = bp_hierarchical_bar; 7172| 0| } 7173| 1|#endif // KMP_FAST_REDUCTION_BARRIER 7174| 1|#endif // KMP_MIC_SUPPORTED 7175| | 7176| |// From KMP_CHECKS initialization 7177| |#ifdef KMP_DEBUG 7178| | __kmp_env_checks = TRUE; /* development versions have the extra checks */ 7179| |#else 7180| 1| __kmp_env_checks = FALSE; /* port versions do not have the extra checks */ ------------------ | | 1279| 1|#define FALSE 0 ------------------ 7181| 1|#endif 7182| | 7183| | // From "KMP_FOREIGN_THREADS_THREADPRIVATE" initialization 7184| 1| __kmp_foreign_tp = TRUE; ------------------ | | 1280| 1|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1|#define FALSE 0 | | ------------------ ------------------ 7185| | 7186| 1| __kmp_global.g.g_dynamic = FALSE; ------------------ | | 1279| 1|#define FALSE 0 ------------------ 7187| 1| __kmp_global.g.g_dynamic_mode = dynamic_default; 7188| | 7189| 1| __kmp_init_nesting_mode(); 7190| | 7191| 1| __kmp_env_initialize(NULL); 7192| | 7193| 1|#if KMP_HAVE_MWAIT || KMP_HAVE_UMWAIT 7194| 1| __kmp_user_level_mwait_init(); 7195| 1|#endif 7196| |// Print all messages in message catalog for testing purposes. 7197| |#ifdef KMP_DEBUG 7198| | char const *val = __kmp_env_get("KMP_DUMP_CATALOG"); 7199| | if (__kmp_str_match_true(val)) { 7200| | kmp_str_buf_t buffer; 7201| | __kmp_str_buf_init(&buffer); 7202| | __kmp_i18n_dump_catalog(&buffer); 7203| | __kmp_printf("%s", buffer.str); 7204| | __kmp_str_buf_free(&buffer); 7205| | } 7206| | __kmp_env_free(&val); 7207| |#endif 7208| | 7209| 1| __kmp_threads_capacity = 7210| 1| __kmp_initial_threads_capacity(__kmp_dflt_team_nth_ub); 7211| | // Moved here from __kmp_env_initialize() "KMP_ALL_THREADPRIVATE" part 7212| 1| __kmp_tp_capacity = __kmp_default_tp_capacity( 7213| 1| __kmp_dflt_team_nth_ub, __kmp_max_nth, __kmp_allThreadsSpecified); 7214| | 7215| | // If the library is shut down properly, both pools must be NULL. Just in 7216| | // case, set them to NULL -- some memory may leak, but subsequent code will 7217| | // work even if pools are not freed. 7218| 1| KMP_DEBUG_ASSERT(__kmp_thread_pool == NULL); 7219| 1| KMP_DEBUG_ASSERT(__kmp_thread_pool_insert_pt == NULL); 7220| 1| KMP_DEBUG_ASSERT(__kmp_team_pool == NULL); 7221| 1| __kmp_thread_pool = NULL; 7222| 1| __kmp_thread_pool_insert_pt = NULL; 7223| 1| __kmp_team_pool = NULL; 7224| | 7225| | /* Allocate all of the variable sized records */ 7226| | /* NOTE: __kmp_threads_capacity entries are allocated, but the arrays are 7227| | * expandable */ 7228| | /* Since allocation is cache-aligned, just add extra padding at the end */ 7229| 1| size = 7230| 1| (sizeof(kmp_info_t *) + sizeof(kmp_root_t *)) * __kmp_threads_capacity + 7231| 1| CACHE_LINE; ------------------ | | 108| 1|# define CACHE_LINE 64 ------------------ 7232| 1| __kmp_threads = (kmp_info_t **)__kmp_allocate(size); ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 7233| 1| __kmp_root = (kmp_root_t **)((char *)__kmp_threads + 7234| 1| sizeof(kmp_info_t *) * __kmp_threads_capacity); 7235| | 7236| | /* init thread counts */ 7237| 1| KMP_DEBUG_ASSERT(__kmp_all_nth == 7238| 1| 0); // Asserts fail if the library is reinitializing and 7239| 1| KMP_DEBUG_ASSERT(__kmp_nth == 0); // something was wrong in termination. 7240| 1| __kmp_all_nth = 0; 7241| 1| __kmp_nth = 0; 7242| | 7243| | /* setup the uber master thread and hierarchy */ 7244| 1| gtid = __kmp_register_root(TRUE); ------------------ | | 1280| 1|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1|#define FALSE 0 | | ------------------ ------------------ 7245| 1| KA_TRACE(10, ("__kmp_do_serial_initialize T#%d\n", gtid)); 7246| 1| KMP_ASSERT(KMP_UBER_GTID(gtid)); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 7247| 1| KMP_ASSERT(KMP_INITIAL_GTID(gtid)); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 7248| | 7249| 1| KMP_MB(); /* Flush all pending memory write invalidates. */ 7250| | 7251| 1| __kmp_common_initialize(); 7252| | 7253| 1|#if KMP_OS_UNIX 7254| | /* invoke the child fork handler */ 7255| 1| __kmp_register_atfork(); 7256| 1|#endif 7257| | 7258| 1|#if !KMP_DYNAMIC_LIB || \ 7259| 1| ((KMP_COMPILER_ICC || KMP_COMPILER_ICX) && KMP_OS_DARWIN) 7260| 1| { 7261| | /* Invoke the exit handler when the program finishes, only for static 7262| | library and macOS* dynamic. For other dynamic libraries, we already 7263| | have _fini and DllMain. */ 7264| 1| int rc = atexit(__kmp_internal_end_atexit); 7265| 1| if (rc != 0) { ------------------ | Branch (7265:9): [True: 0, False: 1] ------------------ 7266| 0| __kmp_fatal(KMP_MSG(FunctionError, "atexit()"), KMP_ERR(rc), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ __kmp_fatal(KMP_MSG(FunctionError, "atexit()"), KMP_ERR(rc), ------------------ | | 125| 0|#define KMP_ERR KMP_SYSERRCODE | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ ------------------ 7267| 0| __kmp_msg_null); 7268| 0| } 7269| 1| } 7270| 1|#endif 7271| | 7272| 1|#if KMP_HANDLE_SIGNALS 7273| 1|#if KMP_OS_UNIX 7274| | /* NOTE: make sure that this is called before the user installs their own 7275| | signal handlers so that the user handlers are called first. this way they 7276| | can return false, not call our handler, avoid terminating the library, and 7277| | continue execution where they left off. */ 7278| 1| __kmp_install_signals(FALSE); ------------------ | | 1279| 1|#define FALSE 0 ------------------ 7279| 1|#endif /* KMP_OS_UNIX */ 7280| |#if KMP_OS_WINDOWS 7281| | __kmp_install_signals(TRUE); 7282| |#endif /* KMP_OS_WINDOWS */ 7283| 1|#endif 7284| | 7285| | /* we have finished the serial initialization */ 7286| 1| __kmp_init_counter++; 7287| | 7288| 1| __kmp_init_serial = TRUE; ------------------ | | 1280| 1|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1|#define FALSE 0 | | ------------------ ------------------ 7289| | 7290| 1| if (__kmp_version) { ------------------ | Branch (7290:7): [True: 0, False: 1] ------------------ 7291| 0| __kmp_print_version_1(); 7292| 0| } 7293| | 7294| 1| if (__kmp_settings) { ------------------ | Branch (7294:7): [True: 0, False: 1] ------------------ 7295| 0| __kmp_env_print(); 7296| 0| } 7297| | 7298| 1| if (__kmp_display_env || __kmp_display_env_verbose) { ------------------ | Branch (7298:7): [True: 0, False: 1] | Branch (7298:28): [True: 0, False: 1] ------------------ 7299| 0| __kmp_env_print_2(); 7300| 0| } 7301| | 7302| 1|#if OMPT_SUPPORT 7303| 1| ompt_post_init(); 7304| 1|#endif 7305| | 7306| 1| KMP_MB(); 7307| | 7308| 1| KA_TRACE(10, ("__kmp_do_serial_initialize: exit\n")); 7309| 1|} kmp_runtime.cpp:_ZL20__kmp_init_allocatorv: 550| 1|static void __kmp_init_allocator() { 551| 1| __kmp_init_memkind(); 552| 1| __kmp_init_target_mem(); 553| 1|} kmp_runtime.cpp:_ZL20__kmp_check_mic_typev: 6946| 1|static void __kmp_check_mic_type() { 6947| 1| kmp_cpuid_t cpuid_state = {0}; 6948| 1| kmp_cpuid_t *cs_p = &cpuid_state; 6949| 1| __kmp_x86_cpuid(1, 0, cs_p); 6950| | // We don't support mic1 at the moment 6951| 1| if ((cs_p->eax & 0xff0) == 0xB10) { ------------------ | Branch (6951:7): [True: 0, False: 1] ------------------ 6952| 0| __kmp_mic_type = mic2; 6953| 1| } else if ((cs_p->eax & 0xf0ff0) == 0x50670) { ------------------ | Branch (6953:14): [True: 0, False: 1] ------------------ 6954| 0| __kmp_mic_type = mic3; 6955| 1| } else { 6956| 1| __kmp_mic_type = non_mic; 6957| 1| } 6958| 1|} kmp_runtime.cpp:_ZL27__kmp_user_level_mwait_initv: 6963| 1|static void __kmp_user_level_mwait_init() { 6964| 1| struct kmp_cpuid buf; 6965| 1| __kmp_x86_cpuid(7, 0, &buf); 6966| 1| __kmp_waitpkg_enabled = ((buf.ecx >> 5) & 1); 6967| 1| __kmp_umwait_enabled = __kmp_waitpkg_enabled && __kmp_user_level_mwait; ------------------ | Branch (6967:26): [True: 0, False: 1] | Branch (6967:51): [True: 0, False: 0] ------------------ 6968| 1| __kmp_tpause_enabled = __kmp_waitpkg_enabled && (__kmp_tpause_state > 0); ------------------ | Branch (6968:26): [True: 0, False: 1] | Branch (6968:51): [True: 0, False: 0] ------------------ 6969| 1| KF_TRACE(30, ("__kmp_user_level_mwait_init: __kmp_umwait_enabled = %d\n", 6970| 1| __kmp_umwait_enabled)); 6971| 1|} kmp_runtime.cpp:_ZL26__kmp_do_middle_initializev: 7324| 1|static void __kmp_do_middle_initialize(void) { 7325| 1| int i, j; 7326| 1| int prev_dflt_team_nth; 7327| | 7328| 1| if (!__kmp_init_serial) { ------------------ | Branch (7328:7): [True: 0, False: 1] ------------------ 7329| 0| __kmp_do_serial_initialize(); 7330| 0| } 7331| | 7332| 1| KA_TRACE(10, ("__kmp_middle_initialize: enter\n")); 7333| | 7334| 1| if (UNLIKELY(!__kmp_need_register_serial)) { ------------------ | | 119| 1|#define UNLIKELY(x) __builtin_expect(!!(x), 0) | | ------------------ | | | Branch (119:21): [True: 0, False: 1] | | ------------------ ------------------ 7335| | // We are in a forked child process. The registration was skipped during 7336| | // serial initialization in __kmp_atfork_child handler. Do it here. 7337| 0| __kmp_register_library_startup(); 7338| 0| } 7339| | 7340| | // Save the previous value for the __kmp_dflt_team_nth so that 7341| | // we can avoid some reinitialization if it hasn't changed. 7342| 1| prev_dflt_team_nth = __kmp_dflt_team_nth; 7343| | 7344| 1|#if KMP_AFFINITY_SUPPORTED 7345| | // __kmp_affinity_initialize() will try to set __kmp_ncores to the 7346| | // number of cores on the machine. 7347| 1| __kmp_affinity_initialize(__kmp_affinity); 7348| | 7349| 1|#endif /* KMP_AFFINITY_SUPPORTED */ 7350| | 7351| 1| KMP_ASSERT(__kmp_xproc > 0); ------------------ | | 50| 1| if (!(cond)) { \ | | ------------------ | | | Branch (50:7): [True: 0, False: 1] | | ------------------ | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | 52| 0| } ------------------ 7352| 1| if (__kmp_avail_proc == 0) { ------------------ | Branch (7352:7): [True: 0, False: 1] ------------------ 7353| 0| __kmp_avail_proc = __kmp_xproc; 7354| 0| } 7355| | 7356| | // If there were empty places in num_threads list (OMP_NUM_THREADS=,,2,3), 7357| | // correct them now 7358| 1| j = 0; 7359| 1| while ((j < __kmp_nested_nth.used) && !__kmp_nested_nth.nth[j]) { ------------------ | Branch (7359:10): [True: 0, False: 1] | Branch (7359:41): [True: 0, False: 0] ------------------ 7360| 0| __kmp_nested_nth.nth[j] = __kmp_dflt_team_nth = __kmp_dflt_team_nth_ub = 7361| 0| __kmp_avail_proc; 7362| 0| j++; 7363| 0| } 7364| | 7365| 1| if (__kmp_dflt_team_nth == 0) { ------------------ | Branch (7365:7): [True: 1, False: 0] ------------------ 7366| |#ifdef KMP_DFLT_NTH_CORES 7367| | // Default #threads = #cores 7368| | __kmp_dflt_team_nth = __kmp_ncores; 7369| | KA_TRACE(20, ("__kmp_middle_initialize: setting __kmp_dflt_team_nth = " 7370| | "__kmp_ncores (%d)\n", 7371| | __kmp_dflt_team_nth)); 7372| |#else 7373| | // Default #threads = #available OS procs 7374| 1| __kmp_dflt_team_nth = __kmp_avail_proc; 7375| 1| KA_TRACE(20, ("__kmp_middle_initialize: setting __kmp_dflt_team_nth = " 7376| 1| "__kmp_avail_proc(%d)\n", 7377| 1| __kmp_dflt_team_nth)); 7378| 1|#endif /* KMP_DFLT_NTH_CORES */ 7379| 1| } 7380| | 7381| 1| if (__kmp_dflt_team_nth < KMP_MIN_NTH) { ------------------ | | 1137| 1|#define KMP_MIN_NTH 1 ------------------ | Branch (7381:7): [True: 0, False: 1] ------------------ 7382| 0| __kmp_dflt_team_nth = KMP_MIN_NTH; ------------------ | | 1137| 0|#define KMP_MIN_NTH 1 ------------------ 7383| 0| } 7384| 1| if (__kmp_dflt_team_nth > __kmp_sys_max_nth) { ------------------ | Branch (7384:7): [True: 0, False: 1] ------------------ 7385| 0| __kmp_dflt_team_nth = __kmp_sys_max_nth; 7386| 0| } 7387| | 7388| 1| if (__kmp_nesting_mode > 0) ------------------ | Branch (7388:7): [True: 0, False: 1] ------------------ 7389| 0| __kmp_set_nesting_mode_threads(); 7390| | 7391| | // There's no harm in continuing if the following check fails, 7392| | // but it indicates an error in the previous logic. 7393| 1| KMP_DEBUG_ASSERT(__kmp_dflt_team_nth <= __kmp_dflt_team_nth_ub); 7394| | 7395| 1| if (__kmp_dflt_team_nth != prev_dflt_team_nth) { ------------------ | Branch (7395:7): [True: 1, False: 0] ------------------ 7396| | // Run through the __kmp_threads array and set the num threads icv for each 7397| | // root thread that is currently registered with the RTL (which has not 7398| | // already explicitly set its nthreads-var with a call to 7399| | // omp_set_num_threads()). 7400| 137| for (i = 0; i < __kmp_threads_capacity; i++) { ------------------ | Branch (7400:17): [True: 136, False: 1] ------------------ 7401| 136| kmp_info_t *thread = __kmp_threads[i]; 7402| 136| if (thread == NULL) ------------------ | Branch (7402:11): [True: 135, False: 1] ------------------ 7403| 135| continue; 7404| 1| if (thread->th.th_current_task->td_icvs.nproc != 0) ------------------ | Branch (7404:11): [True: 0, False: 1] ------------------ 7405| 0| continue; 7406| | 7407| 1| set__nproc(__kmp_threads[i], __kmp_dflt_team_nth); ------------------ | | 2321| 1| (((xthread)->th.th_current_task->td_icvs.nproc) = (xval)) ------------------ 7408| 1| } 7409| 1| } 7410| 1| KA_TRACE( 7411| 1| 20, 7412| 1| ("__kmp_middle_initialize: final value for __kmp_dflt_team_nth = %d\n", 7413| 1| __kmp_dflt_team_nth)); 7414| | 7415| 1|#ifdef KMP_ADJUST_BLOCKTIME 7416| | /* Adjust blocktime to zero if necessary now that __kmp_avail_proc is set */ 7417| 1| if (!__kmp_env_blocktime && (__kmp_avail_proc > 0)) { ------------------ | Branch (7417:7): [True: 1, False: 0] | Branch (7417:31): [True: 1, False: 0] ------------------ 7418| 1| KMP_DEBUG_ASSERT(__kmp_avail_proc > 0); 7419| 1| if (__kmp_nth > __kmp_avail_proc) { ------------------ | Branch (7419:9): [True: 0, False: 1] ------------------ 7420| 0| __kmp_zero_bt = TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 7421| 0| } 7422| 1| } 7423| 1|#endif /* KMP_ADJUST_BLOCKTIME */ 7424| | 7425| | /* we have finished middle initialization */ 7426| 1| TCW_SYNC_4(__kmp_init_middle, TRUE); ------------------ | | 1136| 1|#define TCW_SYNC_4(a, b) (a) = (b) ------------------ 7427| | 7428| 1| KA_TRACE(10, ("__kmp_do_middle_initialize: exit\n")); 7429| 1|} __kmpc_for_static_init_4: 875| 8.87k| kmp_int32 incr, kmp_int32 chunk) { 876| 8.87k| __kmp_for_static_init(loc, gtid, schedtype, plastiter, plower, 877| 8.87k| pupper, pstride, incr, chunk 878| 8.87k|#if OMPT_SUPPORT && OMPT_OPTIONAL 879| 8.87k| , 880| 8.87k| OMPT_GET_RETURN_ADDRESS(0) ------------------ | | 108| 8.87k|#define OMPT_GET_RETURN_ADDRESS(level) __builtin_return_address(level) ------------------ 881| 8.87k|#endif 882| 8.87k| ); 883| 8.87k|} __kmpc_for_static_init_8: 908| 36.2k| kmp_int64 incr, kmp_int64 chunk) { 909| 36.2k| __kmp_for_static_init(loc, gtid, schedtype, plastiter, plower, 910| 36.2k| pupper, pstride, incr, chunk 911| 36.2k|#if OMPT_SUPPORT && OMPT_OPTIONAL 912| 36.2k| , 913| 36.2k| OMPT_GET_RETURN_ADDRESS(0) ------------------ | | 108| 36.2k|#define OMPT_GET_RETURN_ADDRESS(level) __builtin_return_address(level) ------------------ 914| 36.2k|#endif 915| 36.2k| ); 916| 36.2k|} kmp_sched.cpp:_ZL21__kmp_for_static_initIiEvP5identiiPiPT_S4_PN8traits_tIS3_E8signed_tES7_S7_Pv: 83| 8.87k|) { 84| 8.87k| KMP_COUNT_BLOCK(OMP_LOOP_STATIC); ------------------ | | 1001| 8.87k|#define KMP_COUNT_BLOCK(n) ((void)0) ------------------ 85| 8.87k| KMP_PUSH_PARTITIONED_TIMER(OMP_loop_static); ------------------ | | 1014| 8.87k|#define KMP_PUSH_PARTITIONED_TIMER(name) ((void)0) ------------------ 86| 8.87k| KMP_PUSH_PARTITIONED_TIMER(OMP_loop_static_scheduling); ------------------ | | 1014| 8.87k|#define KMP_PUSH_PARTITIONED_TIMER(name) ((void)0) ------------------ 87| | 88| | // Clear monotonic/nonmonotonic bits (ignore it) 89| 8.87k| schedtype = SCHEDULE_WITHOUT_MODIFIERS(schedtype); ------------------ | | 451| 8.87k| (enum sched_type)( \ | | 452| 8.87k| (s) & ~(kmp_sch_modifier_nonmonotonic | kmp_sch_modifier_monotonic)) ------------------ 90| | 91| 8.87k| typedef typename traits_t::unsigned_t UT; 92| 8.87k| typedef typename traits_t::signed_t ST; 93| | /* this all has to be changed back to TID and such.. */ 94| 8.87k| kmp_int32 gtid = global_tid; 95| 8.87k| kmp_uint32 tid; 96| 8.87k| kmp_uint32 nth; 97| 8.87k| UT trip_count; 98| 8.87k| kmp_team_t *team; 99| 8.87k| __kmp_assert_valid_gtid(gtid); 100| 8.87k| kmp_info_t *th = __kmp_threads[gtid]; 101| | 102| 8.87k|#if OMPT_SUPPORT && OMPT_OPTIONAL 103| 8.87k| ompt_team_info_t *team_info = NULL; 104| 8.87k| ompt_task_info_t *task_info = NULL; 105| 8.87k| ompt_work_t ompt_work_type = ompt_work_loop; 106| | 107| 8.87k| static kmp_int8 warn = 0; 108| | 109| 8.87k| if (ompt_enabled.ompt_callback_work || ompt_enabled.ompt_callback_dispatch) { ------------------ | Branch (109:7): [True: 5, False: 8.87k] | Branch (109:42): [True: 6, False: 8.86k] ------------------ 110| | // Only fully initialize variables needed by OMPT if OMPT is enabled. 111| 0| team_info = __ompt_get_teaminfo(0, NULL); 112| 0| task_info = __ompt_get_task_info_object(0); 113| | // Determine workshare type 114| 0| if (loc != NULL) { ------------------ | Branch (114:9): [True: 0, False: 0] ------------------ 115| 0| if ((loc->flags & KMP_IDENT_WORK_LOOP) != 0) { ------------------ | Branch (115:11): [True: 0, False: 0] ------------------ 116| 0| ompt_work_type = ompt_work_loop; 117| 0| } else if ((loc->flags & KMP_IDENT_WORK_SECTIONS) != 0) { ------------------ | Branch (117:18): [True: 0, False: 0] ------------------ 118| 0| ompt_work_type = ompt_work_sections; 119| 0| } else if ((loc->flags & KMP_IDENT_WORK_DISTRIBUTE) != 0) { ------------------ | Branch (119:18): [True: 0, False: 0] ------------------ 120| 0| ompt_work_type = ompt_work_distribute; 121| 0| } else { 122| 0| kmp_int8 bool_res = 123| 0| KMP_COMPARE_AND_STORE_ACQ8(&warn, (kmp_int8)0, (kmp_int8)1); ------------------ | | 795| 0| __sync_bool_compare_and_swap((volatile kmp_uint8 *)(p), (kmp_uint8)(cv), \ | | 796| 0| (kmp_uint8)(sv)) ------------------ 124| 0| if (bool_res) ------------------ | Branch (124:13): [True: 0, False: 0] ------------------ 125| 0| KMP_WARNING(OmptOutdatedWorkshare); ------------------ | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 126| 0| } 127| 0| KMP_DEBUG_ASSERT(ompt_work_type); 128| 0| } 129| 0| } 130| 8.87k|#endif 131| | 132| 8.87k| KMP_DEBUG_ASSERT(plastiter && plower && pupper && pstride); 133| 8.87k| KE_TRACE(10, ("__kmpc_for_static_init called (%d)\n", global_tid)); 134| |#ifdef KMP_DEBUG 135| | { 136| | char *buff; 137| | // create format specifiers before the debug output 138| | buff = __kmp_str_format( 139| | "__kmpc_for_static_init: T#%%d sched=%%d liter=%%d iter=(%%%s," 140| | " %%%s, %%%s) incr=%%%s chunk=%%%s signed?<%s>\n", 141| | traits_t::spec, traits_t::spec, traits_t::spec, 142| | traits_t::spec, traits_t::spec, traits_t::spec); 143| | KD_TRACE(100, (buff, global_tid, schedtype, *plastiter, *plower, *pupper, 144| | *pstride, incr, chunk)); 145| | __kmp_str_free(&buff); 146| | } 147| |#endif 148| | 149| 8.87k| if (__kmp_env_consistency_check) { ------------------ | Branch (149:7): [True: 0, False: 8.87k] ------------------ 150| 0| __kmp_push_workshare(global_tid, ct_pdo, loc); 151| 0| if (incr == 0) { ------------------ | Branch (151:9): [True: 0, False: 0] ------------------ 152| 0| __kmp_error_construct(kmp_i18n_msg_CnsLoopIncrZeroProhibited, ct_pdo, 153| 0| loc); 154| 0| } 155| 0| } 156| | /* special handling for zero-trip loops */ 157| 18.4E| if (incr > 0 ? (*pupper < *plower) : (*plower < *pupper)) { ------------------ | Branch (157:7): [True: 8.89k, False: 18.4E] | Branch (157:7): [True: 0, False: 8.87k] ------------------ 158| 0| if (plastiter != NULL) ------------------ | Branch (158:9): [True: 0, False: 0] ------------------ 159| 0| *plastiter = FALSE; ------------------ | | 1279| 0|#define FALSE 0 ------------------ 160| | /* leave pupper and plower set to entire iteration space */ 161| 0| *pstride = incr; /* value should never be used */ 162| |// *plower = *pupper - incr; 163| |// let compiler bypass the illegal loop (like for(i=1;i<10;i--)) 164| |// THE LINE COMMENTED ABOVE CAUSED shape2F/h_tests_1.f TO HAVE A FAILURE 165| |// ON A ZERO-TRIP LOOP (lower=1, upper=0,stride=1) - JPH June 23, 2009. 166| |#ifdef KMP_DEBUG 167| | { 168| | char *buff; 169| | // create format specifiers before the debug output 170| | buff = __kmp_str_format("__kmpc_for_static_init:(ZERO TRIP) liter=%%d " 171| | "lower=%%%s upper=%%%s stride = %%%s " 172| | "signed?<%s>, loc = %%s\n", 173| | traits_t::spec, traits_t::spec, 174| | traits_t::spec, traits_t::spec); 175| | check_loc(loc); 176| | KD_TRACE(100, 177| | (buff, *plastiter, *plower, *pupper, *pstride, loc->psource)); 178| | __kmp_str_free(&buff); 179| | } 180| |#endif 181| 0| KE_TRACE(10, ("__kmpc_for_static_init: T#%d return\n", global_tid)); 182| | 183| 0|#if OMPT_SUPPORT && OMPT_OPTIONAL 184| 0| if (ompt_enabled.ompt_callback_work) { ------------------ | Branch (184:9): [True: 0, False: 0] ------------------ 185| 0| ompt_callbacks.ompt_callback(ompt_callback_work)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 186| 0| ompt_work_type, ompt_scope_begin, &(team_info->parallel_data), 187| 0| &(task_info->task_data), 0, codeptr); 188| 0| } 189| 0|#endif 190| 0| KMP_STATS_LOOP_END(OMP_loop_static_iterations); 191| 0| return; 192| 0| } 193| | 194| | // Although there are schedule enumerations above kmp_ord_upper which are not 195| | // schedules for "distribute", the only ones which are useful are dynamic, so 196| | // cannot be seen here, since this codepath is only executed for static 197| | // schedules. 198| 8.87k| if (schedtype > kmp_ord_upper) { ------------------ | Branch (198:7): [True: 0, False: 8.87k] ------------------ 199| | // we are in DISTRIBUTE construct 200| 0| schedtype += kmp_sch_static - 201| 0| kmp_distribute_static; // AC: convert to usual schedule type 202| 0| if (th->th.th_team->t.t_serialized > 1) { ------------------ | Branch (202:9): [True: 0, False: 0] ------------------ 203| 0| tid = 0; 204| 0| team = th->th.th_team; 205| 0| } else { 206| 0| tid = th->th.th_team->t.t_master_tid; 207| 0| team = th->th.th_team->t.t_parent; 208| 0| } 209| 8.87k| } else { 210| 8.87k| tid = __kmp_tid_from_gtid(global_tid); 211| 8.87k| team = th->th.th_team; 212| 8.87k| } 213| | 214| | /* determine if "for" loop is an active worksharing construct */ 215| 8.87k| if (team->t.t_serialized) { ------------------ | Branch (215:7): [True: 0, False: 8.87k] ------------------ 216| | /* serialized parallel, each thread executes whole iteration space */ 217| 0| if (plastiter != NULL) ------------------ | Branch (217:9): [True: 0, False: 0] ------------------ 218| 0| *plastiter = TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 219| | /* leave pupper and plower set to entire iteration space */ 220| 0| *pstride = 221| 0| (incr > 0) ? (*pupper - *plower + 1) : (-(*plower - *pupper + 1)); ------------------ | Branch (221:9): [True: 0, False: 0] ------------------ 222| | 223| |#ifdef KMP_DEBUG 224| | { 225| | char *buff; 226| | // create format specifiers before the debug output 227| | buff = __kmp_str_format("__kmpc_for_static_init: (serial) liter=%%d " 228| | "lower=%%%s upper=%%%s stride = %%%s\n", 229| | traits_t::spec, traits_t::spec, 230| | traits_t::spec); 231| | KD_TRACE(100, (buff, *plastiter, *plower, *pupper, *pstride)); 232| | __kmp_str_free(&buff); 233| | } 234| |#endif 235| 0| KE_TRACE(10, ("__kmpc_for_static_init: T#%d return\n", global_tid)); 236| | 237| 0|#if OMPT_SUPPORT && OMPT_OPTIONAL 238| 0| if (ompt_enabled.ompt_callback_work) { ------------------ | Branch (238:9): [True: 0, False: 0] ------------------ 239| 0| ompt_callbacks.ompt_callback(ompt_callback_work)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 240| 0| ompt_work_type, ompt_scope_begin, &(team_info->parallel_data), 241| 0| &(task_info->task_data), *pstride, codeptr); 242| 0| } 243| 0|#endif 244| 0| KMP_STATS_LOOP_END(OMP_loop_static_iterations); 245| 0| return; 246| 0| } 247| 8.87k| nth = team->t.t_nproc; 248| 8.87k| if (nth == 1) { ------------------ | Branch (248:7): [True: 0, False: 8.87k] ------------------ 249| 0| if (plastiter != NULL) ------------------ | Branch (249:9): [True: 0, False: 0] ------------------ 250| 0| *plastiter = TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 251| 0| *pstride = 252| 0| (incr > 0) ? (*pupper - *plower + 1) : (-(*plower - *pupper + 1)); ------------------ | Branch (252:9): [True: 0, False: 0] ------------------ 253| |#ifdef KMP_DEBUG 254| | { 255| | char *buff; 256| | // create format specifiers before the debug output 257| | buff = __kmp_str_format("__kmpc_for_static_init: (serial) liter=%%d " 258| | "lower=%%%s upper=%%%s stride = %%%s\n", 259| | traits_t::spec, traits_t::spec, 260| | traits_t::spec); 261| | KD_TRACE(100, (buff, *plastiter, *plower, *pupper, *pstride)); 262| | __kmp_str_free(&buff); 263| | } 264| |#endif 265| 0| KE_TRACE(10, ("__kmpc_for_static_init: T#%d return\n", global_tid)); 266| | 267| 0|#if OMPT_SUPPORT && OMPT_OPTIONAL 268| 0| if (ompt_enabled.ompt_callback_work) { ------------------ | Branch (268:9): [True: 0, False: 0] ------------------ 269| 0| ompt_callbacks.ompt_callback(ompt_callback_work)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 270| 0| ompt_work_type, ompt_scope_begin, &(team_info->parallel_data), 271| 0| &(task_info->task_data), *pstride, codeptr); 272| 0| } 273| 0|#endif 274| 0| KMP_STATS_LOOP_END(OMP_loop_static_iterations); 275| 0| return; 276| 0| } 277| | 278| | /* compute trip count */ 279| 8.92k| if (incr == 1) { ------------------ | Branch (279:7): [True: 8.92k, False: 18.4E] ------------------ 280| 8.92k| trip_count = *pupper - *plower + 1; 281| 18.4E| } else if (incr == -1) { ------------------ | Branch (281:14): [True: 0, False: 18.4E] ------------------ 282| 0| trip_count = *plower - *pupper + 1; 283| 18.4E| } else if (incr > 0) { ------------------ | Branch (283:14): [True: 0, False: 18.4E] ------------------ 284| | // upper-lower can exceed the limit of signed type 285| 0| trip_count = (UT)(*pupper - *plower) / incr + 1; 286| 18.4E| } else { 287| 18.4E| trip_count = (UT)(*plower - *pupper) / (-incr) + 1; 288| 18.4E| } 289| | 290| |#if KMP_STATS_ENABLED 291| | if (KMP_MASTER_GTID(gtid)) { 292| | KMP_COUNT_VALUE(OMP_loop_static_total_iterations, trip_count); 293| | } 294| |#endif 295| | 296| 8.87k| if (__kmp_env_consistency_check) { ------------------ | Branch (296:7): [True: 0, False: 8.87k] ------------------ 297| | /* tripcount overflow? */ 298| 0| if (trip_count == 0 && *pupper != *plower) { ------------------ | Branch (298:9): [True: 0, False: 0] | Branch (298:28): [True: 0, False: 0] ------------------ 299| 0| __kmp_error_construct(kmp_i18n_msg_CnsIterationRangeTooLarge, ct_pdo, 300| 0| loc); 301| 0| } 302| 0| } 303| | 304| | /* compute remaining parameters */ 305| 8.87k| switch (schedtype) { 306| 8.93k| case kmp_sch_static: { ------------------ | Branch (306:3): [True: 8.93k, False: 18.4E] ------------------ 307| 8.93k| if (trip_count < nth) { ------------------ | Branch (307:9): [True: 4.48k, False: 4.44k] ------------------ 308| 4.48k| KMP_DEBUG_ASSERT( 309| 4.48k| __kmp_static == kmp_sch_static_greedy || 310| 4.48k| __kmp_static == 311| 4.48k| kmp_sch_static_balanced); // Unknown static scheduling type. 312| 4.48k| if (tid < trip_count) { ------------------ | Branch (312:11): [True: 916, False: 3.57k] ------------------ 313| 916| *pupper = *plower = *plower + tid * incr; 314| 3.57k| } else { 315| | // set bounds so non-active threads execute no iterations 316| 18.4E| *plower = *pupper + (incr > 0 ? 1 : -1); ------------------ | Branch (316:30): [True: 3.58k, False: 18.4E] ------------------ 317| 3.57k| } 318| 4.48k| if (plastiter != NULL) ------------------ | Branch (318:11): [True: 4.48k, False: 0] ------------------ 319| 4.48k| *plastiter = (tid == trip_count - 1); 320| 4.48k| } else { 321| 4.45k| if (__kmp_static == kmp_sch_static_balanced) { ------------------ | Branch (321:11): [True: 4.45k, False: 18.4E] ------------------ 322| 4.45k| UT small_chunk = trip_count / nth; 323| 4.45k| UT extras = trip_count % nth; 324| 4.45k| *plower += incr * (tid * small_chunk + (tid < extras ? tid : extras)); ------------------ | Branch (324:49): [True: 642, False: 3.81k] ------------------ 325| 4.45k| *pupper = *plower + small_chunk * incr - (tid < extras ? 0 : incr); ------------------ | Branch (325:51): [True: 642, False: 3.81k] ------------------ 326| 4.45k| if (plastiter != NULL) ------------------ | Branch (326:13): [True: 4.44k, False: 8] ------------------ 327| 4.44k| *plastiter = (tid == nth - 1); 328| 18.4E| } else { 329| 18.4E| T big_chunk_inc_count = 330| 18.4E| (trip_count / nth + ((trip_count % nth) ? 1 : 0)) * incr; ------------------ | Branch (330:34): [True: 0, False: 18.4E] ------------------ 331| 18.4E| T old_upper = *pupper; 332| | 333| 18.4E| KMP_DEBUG_ASSERT(__kmp_static == kmp_sch_static_greedy); 334| | // Unknown static scheduling type. 335| | 336| 18.4E| *plower += tid * big_chunk_inc_count; 337| 18.4E| *pupper = *plower + big_chunk_inc_count - incr; 338| 18.4E| if (incr > 0) { ------------------ | Branch (338:13): [True: 0, False: 18.4E] ------------------ 339| 0| if (*pupper < *plower) ------------------ | Branch (339:15): [True: 0, False: 0] ------------------ 340| 0| *pupper = traits_t::max_value; 341| 0| if (plastiter != NULL) ------------------ | Branch (341:15): [True: 0, False: 0] ------------------ 342| 0| *plastiter = *plower <= old_upper && *pupper > old_upper - incr; ------------------ | Branch (342:26): [True: 0, False: 0] | Branch (342:50): [True: 0, False: 0] ------------------ 343| 0| if (*pupper > old_upper) ------------------ | Branch (343:15): [True: 0, False: 0] ------------------ 344| 0| *pupper = old_upper; // tracker C73258 345| 18.4E| } else { 346| 18.4E| if (*pupper > *plower) ------------------ | Branch (346:15): [True: 0, False: 18.4E] ------------------ 347| 0| *pupper = traits_t::min_value; 348| 18.4E| if (plastiter != NULL) ------------------ | Branch (348:15): [True: 0, False: 18.4E] ------------------ 349| 0| *plastiter = *plower >= old_upper && *pupper < old_upper - incr; ------------------ | Branch (349:26): [True: 0, False: 0] | Branch (349:50): [True: 0, False: 0] ------------------ 350| 18.4E| if (*pupper < old_upper) ------------------ | Branch (350:15): [True: 0, False: 18.4E] ------------------ 351| 0| *pupper = old_upper; // tracker C73258 352| 18.4E| } 353| 18.4E| } 354| 4.44k| } 355| 8.93k| *pstride = trip_count; 356| 8.93k| break; 357| 0| } 358| 0| case kmp_sch_static_chunked: { ------------------ | Branch (358:3): [True: 0, False: 8.87k] ------------------ 359| 0| ST span; 360| 0| UT nchunks; 361| 0| if (chunk < 1) ------------------ | Branch (361:9): [True: 0, False: 0] ------------------ 362| 0| chunk = 1; 363| 0| else if ((UT)chunk > trip_count) ------------------ | Branch (363:14): [True: 0, False: 0] ------------------ 364| 0| chunk = trip_count; 365| 0| nchunks = (trip_count) / (UT)chunk + (trip_count % (UT)chunk ? 1 : 0); ------------------ | Branch (365:43): [True: 0, False: 0] ------------------ 366| 0| span = chunk * incr; 367| 0| if (nchunks < nth) { ------------------ | Branch (367:9): [True: 0, False: 0] ------------------ 368| 0| *pstride = span * nchunks; 369| 0| if (tid < nchunks) { ------------------ | Branch (369:11): [True: 0, False: 0] ------------------ 370| 0| *plower = *plower + (span * tid); 371| 0| *pupper = *plower + span - incr; 372| 0| } else { 373| 0| *plower = *pupper + (incr > 0 ? 1 : -1); ------------------ | Branch (373:30): [True: 0, False: 0] ------------------ 374| 0| } 375| 0| } else { 376| 0| *pstride = span * nth; 377| 0| *plower = *plower + (span * tid); 378| 0| *pupper = *plower + span - incr; 379| 0| } 380| 0| if (plastiter != NULL) ------------------ | Branch (380:9): [True: 0, False: 0] ------------------ 381| 0| *plastiter = (tid == (nchunks - 1) % nth); 382| 0| break; 383| 0| } 384| 0| case kmp_sch_static_balanced_chunked: { ------------------ | Branch (384:3): [True: 0, False: 8.87k] ------------------ 385| 0| T old_upper = *pupper; 386| | // round up to make sure the chunk is enough to cover all iterations 387| 0| UT span = (trip_count + nth - 1) / nth; 388| | 389| | // perform chunk adjustment 390| 0| chunk = (span + chunk - 1) & ~(chunk - 1); 391| | 392| 0| span = chunk * incr; 393| 0| *plower = *plower + (span * tid); 394| 0| *pupper = *plower + span - incr; 395| 0| if (incr > 0) { ------------------ | Branch (395:9): [True: 0, False: 0] ------------------ 396| 0| if (*pupper > old_upper) ------------------ | Branch (396:11): [True: 0, False: 0] ------------------ 397| 0| *pupper = old_upper; 398| 0| } else if (*pupper < old_upper) ------------------ | Branch (398:16): [True: 0, False: 0] ------------------ 399| 0| *pupper = old_upper; 400| | 401| 0| if (plastiter != NULL) ------------------ | Branch (401:9): [True: 0, False: 0] ------------------ 402| 0| *plastiter = (tid == ((trip_count - 1) / (UT)chunk)); 403| 0| break; 404| 0| } 405| 0| default: ------------------ | Branch (405:3): [True: 0, False: 8.87k] ------------------ 406| 0| KMP_ASSERT2(0, "__kmpc_for_static_init: unknown scheduling type"); ------------------ | | 53| 0|#define KMP_ASSERT2(cond, msg) KMP_ASSERT(cond) | | ------------------ | | | | 50| 0| if (!(cond)) { \ | | | | ------------------ | | | | | Branch (50:7): [True: 0, Folded] | | | | ------------------ | | | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | | | 52| 0| } | | ------------------ ------------------ 407| 0| break; 408| 8.87k| } 409| | 410| |#if USE_ITT_BUILD 411| | // Report loop metadata 412| | if (KMP_MASTER_TID(tid) && __itt_metadata_add_ptr && 413| | __kmp_forkjoin_frames_mode == 3 && th->th.th_teams_microtask == NULL && 414| | team->t.t_active_level == 1) { 415| | kmp_uint64 cur_chunk = chunk; 416| | check_loc(loc); 417| | // Calculate chunk in case it was not specified; it is specified for 418| | // kmp_sch_static_chunked 419| | if (schedtype == kmp_sch_static) { 420| | cur_chunk = trip_count / nth + ((trip_count % nth) ? 1 : 0); 421| | } 422| | // 0 - "static" schedule 423| | __kmp_itt_metadata_loop(loc, 0, trip_count, cur_chunk); 424| | } 425| |#endif 426| |#ifdef KMP_DEBUG 427| | { 428| | char *buff; 429| | // create format specifiers before the debug output 430| | buff = __kmp_str_format("__kmpc_for_static_init: liter=%%d lower=%%%s " 431| | "upper=%%%s stride = %%%s signed?<%s>\n", 432| | traits_t::spec, traits_t::spec, 433| | traits_t::spec, traits_t::spec); 434| | KD_TRACE(100, (buff, *plastiter, *plower, *pupper, *pstride)); 435| | __kmp_str_free(&buff); 436| | } 437| |#endif 438| 8.92k| KE_TRACE(10, ("__kmpc_for_static_init: T#%d return\n", global_tid)); 439| | 440| 8.92k|#if OMPT_SUPPORT && OMPT_OPTIONAL 441| 8.92k| if (ompt_enabled.ompt_callback_work) { ------------------ | Branch (441:7): [True: 0, False: 8.92k] ------------------ 442| 0| ompt_callbacks.ompt_callback(ompt_callback_work)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 443| 0| ompt_work_type, ompt_scope_begin, &(team_info->parallel_data), 444| 0| &(task_info->task_data), trip_count, codeptr); 445| 0| } 446| 8.92k| if (ompt_enabled.ompt_callback_dispatch) { ------------------ | Branch (446:7): [True: 0, False: 8.92k] ------------------ 447| 0| ompt_dispatch_t dispatch_type; 448| 0| ompt_data_t instance = ompt_data_none; ------------------ | | 1408| 0|#define ompt_data_none {0} ------------------ 449| 0| ompt_dispatch_chunk_t dispatch_chunk; 450| 0| if (ompt_work_type == ompt_work_sections) { ------------------ | Branch (450:9): [True: 0, False: 0] ------------------ 451| 0| dispatch_type = ompt_dispatch_section; 452| 0| instance.ptr = codeptr; 453| 0| } else { 454| 0| OMPT_GET_DISPATCH_CHUNK(dispatch_chunk, *plower, *pupper, incr); ------------------ | | 101| 0| do { \ | | 102| 0| if (incr > 0) { \ | | ------------------ | | | Branch (102:9): [True: 0, False: 0] | | ------------------ | | 103| 0| chunk.start = static_cast(lb); \ | | 104| 0| chunk.iterations = static_cast(((ub) - (lb)) / (incr) + 1); \ | | 105| 0| } else { \ | | 106| 0| chunk.start = static_cast(ub); \ | | 107| 0| chunk.iterations = static_cast(((lb) - (ub)) / -(incr) + 1); \ | | 108| 0| } \ | | 109| 0| } while (0) | | ------------------ | | | Branch (109:12): [Folded, False: 0] | | ------------------ ------------------ 455| 0| dispatch_type = (ompt_work_type == ompt_work_distribute) ------------------ | Branch (455:23): [True: 0, False: 0] ------------------ 456| 0| ? ompt_dispatch_distribute_chunk 457| 0| : ompt_dispatch_ws_loop_chunk; 458| 0| instance.ptr = &dispatch_chunk; 459| 0| } 460| 0| ompt_callbacks.ompt_callback(ompt_callback_dispatch)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 461| 0| &(team_info->parallel_data), &(task_info->task_data), dispatch_type, 462| 0| instance); 463| 0| } 464| 8.92k|#endif 465| | 466| 8.92k| KMP_STATS_LOOP_END(OMP_loop_static_iterations); 467| 8.92k| return; 468| 8.87k|} kmp_sched.cpp:_ZL21__kmp_for_static_initIxEvP5identiiPiPT_S4_PN8traits_tIS3_E8signed_tES7_S7_Pv: 83| 36.2k|) { 84| 36.2k| KMP_COUNT_BLOCK(OMP_LOOP_STATIC); ------------------ | | 1001| 36.2k|#define KMP_COUNT_BLOCK(n) ((void)0) ------------------ 85| 36.2k| KMP_PUSH_PARTITIONED_TIMER(OMP_loop_static); ------------------ | | 1014| 36.2k|#define KMP_PUSH_PARTITIONED_TIMER(name) ((void)0) ------------------ 86| 36.2k| KMP_PUSH_PARTITIONED_TIMER(OMP_loop_static_scheduling); ------------------ | | 1014| 36.2k|#define KMP_PUSH_PARTITIONED_TIMER(name) ((void)0) ------------------ 87| | 88| | // Clear monotonic/nonmonotonic bits (ignore it) 89| 36.2k| schedtype = SCHEDULE_WITHOUT_MODIFIERS(schedtype); ------------------ | | 451| 36.2k| (enum sched_type)( \ | | 452| 36.2k| (s) & ~(kmp_sch_modifier_nonmonotonic | kmp_sch_modifier_monotonic)) ------------------ 90| | 91| 36.2k| typedef typename traits_t::unsigned_t UT; 92| 36.2k| typedef typename traits_t::signed_t ST; 93| | /* this all has to be changed back to TID and such.. */ 94| 36.2k| kmp_int32 gtid = global_tid; 95| 36.2k| kmp_uint32 tid; 96| 36.2k| kmp_uint32 nth; 97| 36.2k| UT trip_count; 98| 36.2k| kmp_team_t *team; 99| 36.2k| __kmp_assert_valid_gtid(gtid); 100| 36.2k| kmp_info_t *th = __kmp_threads[gtid]; 101| | 102| 36.2k|#if OMPT_SUPPORT && OMPT_OPTIONAL 103| 36.2k| ompt_team_info_t *team_info = NULL; 104| 36.2k| ompt_task_info_t *task_info = NULL; 105| 36.2k| ompt_work_t ompt_work_type = ompt_work_loop; 106| | 107| 36.2k| static kmp_int8 warn = 0; 108| | 109| 36.2k| if (ompt_enabled.ompt_callback_work || ompt_enabled.ompt_callback_dispatch) { ------------------ | Branch (109:7): [True: 23, False: 36.2k] | Branch (109:42): [True: 18.4E, False: 36.2k] ------------------ 110| | // Only fully initialize variables needed by OMPT if OMPT is enabled. 111| 0| team_info = __ompt_get_teaminfo(0, NULL); 112| 0| task_info = __ompt_get_task_info_object(0); 113| | // Determine workshare type 114| 0| if (loc != NULL) { ------------------ | Branch (114:9): [True: 0, False: 0] ------------------ 115| 0| if ((loc->flags & KMP_IDENT_WORK_LOOP) != 0) { ------------------ | Branch (115:11): [True: 0, False: 0] ------------------ 116| 0| ompt_work_type = ompt_work_loop; 117| 0| } else if ((loc->flags & KMP_IDENT_WORK_SECTIONS) != 0) { ------------------ | Branch (117:18): [True: 0, False: 0] ------------------ 118| 0| ompt_work_type = ompt_work_sections; 119| 0| } else if ((loc->flags & KMP_IDENT_WORK_DISTRIBUTE) != 0) { ------------------ | Branch (119:18): [True: 0, False: 0] ------------------ 120| 0| ompt_work_type = ompt_work_distribute; 121| 0| } else { 122| 0| kmp_int8 bool_res = 123| 0| KMP_COMPARE_AND_STORE_ACQ8(&warn, (kmp_int8)0, (kmp_int8)1); ------------------ | | 795| 0| __sync_bool_compare_and_swap((volatile kmp_uint8 *)(p), (kmp_uint8)(cv), \ | | 796| 0| (kmp_uint8)(sv)) ------------------ 124| 0| if (bool_res) ------------------ | Branch (124:13): [True: 0, False: 0] ------------------ 125| 0| KMP_WARNING(OmptOutdatedWorkshare); ------------------ | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 126| 0| } 127| 0| KMP_DEBUG_ASSERT(ompt_work_type); 128| 0| } 129| 0| } 130| 36.2k|#endif 131| | 132| 36.2k| KMP_DEBUG_ASSERT(plastiter && plower && pupper && pstride); 133| 36.2k| KE_TRACE(10, ("__kmpc_for_static_init called (%d)\n", global_tid)); 134| |#ifdef KMP_DEBUG 135| | { 136| | char *buff; 137| | // create format specifiers before the debug output 138| | buff = __kmp_str_format( 139| | "__kmpc_for_static_init: T#%%d sched=%%d liter=%%d iter=(%%%s," 140| | " %%%s, %%%s) incr=%%%s chunk=%%%s signed?<%s>\n", 141| | traits_t::spec, traits_t::spec, traits_t::spec, 142| | traits_t::spec, traits_t::spec, traits_t::spec); 143| | KD_TRACE(100, (buff, global_tid, schedtype, *plastiter, *plower, *pupper, 144| | *pstride, incr, chunk)); 145| | __kmp_str_free(&buff); 146| | } 147| |#endif 148| | 149| 36.2k| if (__kmp_env_consistency_check) { ------------------ | Branch (149:7): [True: 0, False: 36.2k] ------------------ 150| 0| __kmp_push_workshare(global_tid, ct_pdo, loc); 151| 0| if (incr == 0) { ------------------ | Branch (151:9): [True: 0, False: 0] ------------------ 152| 0| __kmp_error_construct(kmp_i18n_msg_CnsLoopIncrZeroProhibited, ct_pdo, 153| 0| loc); 154| 0| } 155| 0| } 156| | /* special handling for zero-trip loops */ 157| 18.4E| if (incr > 0 ? (*pupper < *plower) : (*plower < *pupper)) { ------------------ | Branch (157:7): [True: 36.3k, False: 18.4E] | Branch (157:7): [True: 0, False: 36.2k] ------------------ 158| 0| if (plastiter != NULL) ------------------ | Branch (158:9): [True: 0, False: 0] ------------------ 159| 0| *plastiter = FALSE; ------------------ | | 1279| 0|#define FALSE 0 ------------------ 160| | /* leave pupper and plower set to entire iteration space */ 161| 0| *pstride = incr; /* value should never be used */ 162| |// *plower = *pupper - incr; 163| |// let compiler bypass the illegal loop (like for(i=1;i<10;i--)) 164| |// THE LINE COMMENTED ABOVE CAUSED shape2F/h_tests_1.f TO HAVE A FAILURE 165| |// ON A ZERO-TRIP LOOP (lower=1, upper=0,stride=1) - JPH June 23, 2009. 166| |#ifdef KMP_DEBUG 167| | { 168| | char *buff; 169| | // create format specifiers before the debug output 170| | buff = __kmp_str_format("__kmpc_for_static_init:(ZERO TRIP) liter=%%d " 171| | "lower=%%%s upper=%%%s stride = %%%s " 172| | "signed?<%s>, loc = %%s\n", 173| | traits_t::spec, traits_t::spec, 174| | traits_t::spec, traits_t::spec); 175| | check_loc(loc); 176| | KD_TRACE(100, 177| | (buff, *plastiter, *plower, *pupper, *pstride, loc->psource)); 178| | __kmp_str_free(&buff); 179| | } 180| |#endif 181| 0| KE_TRACE(10, ("__kmpc_for_static_init: T#%d return\n", global_tid)); 182| | 183| 0|#if OMPT_SUPPORT && OMPT_OPTIONAL 184| 0| if (ompt_enabled.ompt_callback_work) { ------------------ | Branch (184:9): [True: 0, False: 0] ------------------ 185| 0| ompt_callbacks.ompt_callback(ompt_callback_work)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 186| 0| ompt_work_type, ompt_scope_begin, &(team_info->parallel_data), 187| 0| &(task_info->task_data), 0, codeptr); 188| 0| } 189| 0|#endif 190| 0| KMP_STATS_LOOP_END(OMP_loop_static_iterations); 191| 0| return; 192| 0| } 193| | 194| | // Although there are schedule enumerations above kmp_ord_upper which are not 195| | // schedules for "distribute", the only ones which are useful are dynamic, so 196| | // cannot be seen here, since this codepath is only executed for static 197| | // schedules. 198| 36.2k| if (schedtype > kmp_ord_upper) { ------------------ | Branch (198:7): [True: 0, False: 36.2k] ------------------ 199| | // we are in DISTRIBUTE construct 200| 0| schedtype += kmp_sch_static - 201| 0| kmp_distribute_static; // AC: convert to usual schedule type 202| 0| if (th->th.th_team->t.t_serialized > 1) { ------------------ | Branch (202:9): [True: 0, False: 0] ------------------ 203| 0| tid = 0; 204| 0| team = th->th.th_team; 205| 0| } else { 206| 0| tid = th->th.th_team->t.t_master_tid; 207| 0| team = th->th.th_team->t.t_parent; 208| 0| } 209| 36.2k| } else { 210| 36.2k| tid = __kmp_tid_from_gtid(global_tid); 211| 36.2k| team = th->th.th_team; 212| 36.2k| } 213| | 214| | /* determine if "for" loop is an active worksharing construct */ 215| 36.2k| if (team->t.t_serialized) { ------------------ | Branch (215:7): [True: 3.01k, False: 33.2k] ------------------ 216| | /* serialized parallel, each thread executes whole iteration space */ 217| 3.01k| if (plastiter != NULL) ------------------ | Branch (217:9): [True: 3.01k, False: 0] ------------------ 218| 3.01k| *plastiter = TRUE; ------------------ | | 1280| 3.01k|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 3.01k|#define FALSE 0 | | ------------------ ------------------ 219| | /* leave pupper and plower set to entire iteration space */ 220| 3.01k| *pstride = 221| 3.01k| (incr > 0) ? (*pupper - *plower + 1) : (-(*plower - *pupper + 1)); ------------------ | Branch (221:9): [True: 3.01k, False: 0] ------------------ 222| | 223| |#ifdef KMP_DEBUG 224| | { 225| | char *buff; 226| | // create format specifiers before the debug output 227| | buff = __kmp_str_format("__kmpc_for_static_init: (serial) liter=%%d " 228| | "lower=%%%s upper=%%%s stride = %%%s\n", 229| | traits_t::spec, traits_t::spec, 230| | traits_t::spec); 231| | KD_TRACE(100, (buff, *plastiter, *plower, *pupper, *pstride)); 232| | __kmp_str_free(&buff); 233| | } 234| |#endif 235| 3.01k| KE_TRACE(10, ("__kmpc_for_static_init: T#%d return\n", global_tid)); 236| | 237| 3.01k|#if OMPT_SUPPORT && OMPT_OPTIONAL 238| 3.01k| if (ompt_enabled.ompt_callback_work) { ------------------ | Branch (238:9): [True: 0, False: 3.01k] ------------------ 239| 0| ompt_callbacks.ompt_callback(ompt_callback_work)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 240| 0| ompt_work_type, ompt_scope_begin, &(team_info->parallel_data), 241| 0| &(task_info->task_data), *pstride, codeptr); 242| 0| } 243| 3.01k|#endif 244| 3.01k| KMP_STATS_LOOP_END(OMP_loop_static_iterations); 245| 3.01k| return; 246| 3.01k| } 247| 33.2k| nth = team->t.t_nproc; 248| 33.2k| if (nth == 1) { ------------------ | Branch (248:7): [True: 0, False: 33.2k] ------------------ 249| 0| if (plastiter != NULL) ------------------ | Branch (249:9): [True: 0, False: 0] ------------------ 250| 0| *plastiter = TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 251| 0| *pstride = 252| 0| (incr > 0) ? (*pupper - *plower + 1) : (-(*plower - *pupper + 1)); ------------------ | Branch (252:9): [True: 0, False: 0] ------------------ 253| |#ifdef KMP_DEBUG 254| | { 255| | char *buff; 256| | // create format specifiers before the debug output 257| | buff = __kmp_str_format("__kmpc_for_static_init: (serial) liter=%%d " 258| | "lower=%%%s upper=%%%s stride = %%%s\n", 259| | traits_t::spec, traits_t::spec, 260| | traits_t::spec); 261| | KD_TRACE(100, (buff, *plastiter, *plower, *pupper, *pstride)); 262| | __kmp_str_free(&buff); 263| | } 264| |#endif 265| 0| KE_TRACE(10, ("__kmpc_for_static_init: T#%d return\n", global_tid)); 266| | 267| 0|#if OMPT_SUPPORT && OMPT_OPTIONAL 268| 0| if (ompt_enabled.ompt_callback_work) { ------------------ | Branch (268:9): [True: 0, False: 0] ------------------ 269| 0| ompt_callbacks.ompt_callback(ompt_callback_work)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 270| 0| ompt_work_type, ompt_scope_begin, &(team_info->parallel_data), 271| 0| &(task_info->task_data), *pstride, codeptr); 272| 0| } 273| 0|#endif 274| 0| KMP_STATS_LOOP_END(OMP_loop_static_iterations); 275| 0| return; 276| 0| } 277| | 278| | /* compute trip count */ 279| 33.5k| if (incr == 1) { ------------------ | Branch (279:7): [True: 33.5k, False: 18.4E] ------------------ 280| 33.5k| trip_count = *pupper - *plower + 1; 281| 18.4E| } else if (incr == -1) { ------------------ | Branch (281:14): [True: 0, False: 18.4E] ------------------ 282| 0| trip_count = *plower - *pupper + 1; 283| 18.4E| } else if (incr > 0) { ------------------ | Branch (283:14): [True: 0, False: 18.4E] ------------------ 284| | // upper-lower can exceed the limit of signed type 285| 0| trip_count = (UT)(*pupper - *plower) / incr + 1; 286| 18.4E| } else { 287| 18.4E| trip_count = (UT)(*plower - *pupper) / (-incr) + 1; 288| 18.4E| } 289| | 290| |#if KMP_STATS_ENABLED 291| | if (KMP_MASTER_GTID(gtid)) { 292| | KMP_COUNT_VALUE(OMP_loop_static_total_iterations, trip_count); 293| | } 294| |#endif 295| | 296| 33.2k| if (__kmp_env_consistency_check) { ------------------ | Branch (296:7): [True: 0, False: 33.2k] ------------------ 297| | /* tripcount overflow? */ 298| 0| if (trip_count == 0 && *pupper != *plower) { ------------------ | Branch (298:9): [True: 0, False: 0] | Branch (298:28): [True: 0, False: 0] ------------------ 299| 0| __kmp_error_construct(kmp_i18n_msg_CnsIterationRangeTooLarge, ct_pdo, 300| 0| loc); 301| 0| } 302| 0| } 303| | 304| | /* compute remaining parameters */ 305| 33.2k| switch (schedtype) { 306| 33.5k| case kmp_sch_static: { ------------------ | Branch (306:3): [True: 33.5k, False: 18.4E] ------------------ 307| 33.5k| if (trip_count < nth) { ------------------ | Branch (307:9): [True: 32.8k, False: 648] ------------------ 308| 32.8k| KMP_DEBUG_ASSERT( 309| 32.8k| __kmp_static == kmp_sch_static_greedy || 310| 32.8k| __kmp_static == 311| 32.8k| kmp_sch_static_balanced); // Unknown static scheduling type. 312| 32.8k| if (tid < trip_count) { ------------------ | Branch (312:11): [True: 5.53k, False: 27.3k] ------------------ 313| 5.53k| *pupper = *plower = *plower + tid * incr; 314| 27.3k| } else { 315| | // set bounds so non-active threads execute no iterations 316| 18.4E| *plower = *pupper + (incr > 0 ? 1 : -1); ------------------ | Branch (316:30): [True: 27.5k, False: 18.4E] ------------------ 317| 27.3k| } 318| 32.8k| if (plastiter != NULL) ------------------ | Branch (318:11): [True: 32.9k, False: 18.4E] ------------------ 319| 32.9k| *plastiter = (tid == trip_count - 1); 320| 32.8k| } else { 321| 648| if (__kmp_static == kmp_sch_static_balanced) { ------------------ | Branch (321:11): [True: 645, False: 3] ------------------ 322| 645| UT small_chunk = trip_count / nth; 323| 645| UT extras = trip_count % nth; 324| 645| *plower += incr * (tid * small_chunk + (tid < extras ? tid : extras)); ------------------ | Branch (324:49): [True: 79, False: 566] ------------------ 325| 645| *pupper = *plower + small_chunk * incr - (tid < extras ? 0 : incr); ------------------ | Branch (325:51): [True: 79, False: 566] ------------------ 326| 645| if (plastiter != NULL) ------------------ | Branch (326:13): [True: 642, False: 3] ------------------ 327| 642| *plastiter = (tid == nth - 1); 328| 645| } else { 329| 3| T big_chunk_inc_count = 330| 3| (trip_count / nth + ((trip_count % nth) ? 1 : 0)) * incr; ------------------ | Branch (330:34): [True: 0, False: 3] ------------------ 331| 3| T old_upper = *pupper; 332| | 333| 3| KMP_DEBUG_ASSERT(__kmp_static == kmp_sch_static_greedy); 334| | // Unknown static scheduling type. 335| | 336| 3| *plower += tid * big_chunk_inc_count; 337| 3| *pupper = *plower + big_chunk_inc_count - incr; 338| 3| if (incr > 0) { ------------------ | Branch (338:13): [True: 0, False: 3] ------------------ 339| 0| if (*pupper < *plower) ------------------ | Branch (339:15): [True: 0, False: 0] ------------------ 340| 0| *pupper = traits_t::max_value; 341| 0| if (plastiter != NULL) ------------------ | Branch (341:15): [True: 0, False: 0] ------------------ 342| 0| *plastiter = *plower <= old_upper && *pupper > old_upper - incr; ------------------ | Branch (342:26): [True: 0, False: 0] | Branch (342:50): [True: 0, False: 0] ------------------ 343| 0| if (*pupper > old_upper) ------------------ | Branch (343:15): [True: 0, False: 0] ------------------ 344| 0| *pupper = old_upper; // tracker C73258 345| 3| } else { 346| 3| if (*pupper > *plower) ------------------ | Branch (346:15): [True: 0, False: 3] ------------------ 347| 0| *pupper = traits_t::min_value; 348| 3| if (plastiter != NULL) ------------------ | Branch (348:15): [True: 0, False: 3] ------------------ 349| 0| *plastiter = *plower >= old_upper && *pupper < old_upper - incr; ------------------ | Branch (349:26): [True: 0, False: 0] | Branch (349:50): [True: 0, False: 0] ------------------ 350| 3| if (*pupper < old_upper) ------------------ | Branch (350:15): [True: 0, False: 3] ------------------ 351| 0| *pupper = old_upper; // tracker C73258 352| 3| } 353| 3| } 354| 648| } 355| 33.5k| *pstride = trip_count; 356| 33.5k| break; 357| 0| } 358| 0| case kmp_sch_static_chunked: { ------------------ | Branch (358:3): [True: 0, False: 33.2k] ------------------ 359| 0| ST span; 360| 0| UT nchunks; 361| 0| if (chunk < 1) ------------------ | Branch (361:9): [True: 0, False: 0] ------------------ 362| 0| chunk = 1; 363| 0| else if ((UT)chunk > trip_count) ------------------ | Branch (363:14): [True: 0, False: 0] ------------------ 364| 0| chunk = trip_count; 365| 0| nchunks = (trip_count) / (UT)chunk + (trip_count % (UT)chunk ? 1 : 0); ------------------ | Branch (365:43): [True: 0, False: 0] ------------------ 366| 0| span = chunk * incr; 367| 0| if (nchunks < nth) { ------------------ | Branch (367:9): [True: 0, False: 0] ------------------ 368| 0| *pstride = span * nchunks; 369| 0| if (tid < nchunks) { ------------------ | Branch (369:11): [True: 0, False: 0] ------------------ 370| 0| *plower = *plower + (span * tid); 371| 0| *pupper = *plower + span - incr; 372| 0| } else { 373| 0| *plower = *pupper + (incr > 0 ? 1 : -1); ------------------ | Branch (373:30): [True: 0, False: 0] ------------------ 374| 0| } 375| 0| } else { 376| 0| *pstride = span * nth; 377| 0| *plower = *plower + (span * tid); 378| 0| *pupper = *plower + span - incr; 379| 0| } 380| 0| if (plastiter != NULL) ------------------ | Branch (380:9): [True: 0, False: 0] ------------------ 381| 0| *plastiter = (tid == (nchunks - 1) % nth); 382| 0| break; 383| 0| } 384| 0| case kmp_sch_static_balanced_chunked: { ------------------ | Branch (384:3): [True: 0, False: 33.2k] ------------------ 385| 0| T old_upper = *pupper; 386| | // round up to make sure the chunk is enough to cover all iterations 387| 0| UT span = (trip_count + nth - 1) / nth; 388| | 389| | // perform chunk adjustment 390| 0| chunk = (span + chunk - 1) & ~(chunk - 1); 391| | 392| 0| span = chunk * incr; 393| 0| *plower = *plower + (span * tid); 394| 0| *pupper = *plower + span - incr; 395| 0| if (incr > 0) { ------------------ | Branch (395:9): [True: 0, False: 0] ------------------ 396| 0| if (*pupper > old_upper) ------------------ | Branch (396:11): [True: 0, False: 0] ------------------ 397| 0| *pupper = old_upper; 398| 0| } else if (*pupper < old_upper) ------------------ | Branch (398:16): [True: 0, False: 0] ------------------ 399| 0| *pupper = old_upper; 400| | 401| 0| if (plastiter != NULL) ------------------ | Branch (401:9): [True: 0, False: 0] ------------------ 402| 0| *plastiter = (tid == ((trip_count - 1) / (UT)chunk)); 403| 0| break; 404| 0| } 405| 0| default: ------------------ | Branch (405:3): [True: 0, False: 33.2k] ------------------ 406| 0| KMP_ASSERT2(0, "__kmpc_for_static_init: unknown scheduling type"); ------------------ | | 53| 0|#define KMP_ASSERT2(cond, msg) KMP_ASSERT(cond) | | ------------------ | | | | 50| 0| if (!(cond)) { \ | | | | ------------------ | | | | | Branch (50:7): [True: 0, Folded] | | | | ------------------ | | | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | | | 52| 0| } | | ------------------ ------------------ 407| 0| break; 408| 33.2k| } 409| | 410| |#if USE_ITT_BUILD 411| | // Report loop metadata 412| | if (KMP_MASTER_TID(tid) && __itt_metadata_add_ptr && 413| | __kmp_forkjoin_frames_mode == 3 && th->th.th_teams_microtask == NULL && 414| | team->t.t_active_level == 1) { 415| | kmp_uint64 cur_chunk = chunk; 416| | check_loc(loc); 417| | // Calculate chunk in case it was not specified; it is specified for 418| | // kmp_sch_static_chunked 419| | if (schedtype == kmp_sch_static) { 420| | cur_chunk = trip_count / nth + ((trip_count % nth) ? 1 : 0); 421| | } 422| | // 0 - "static" schedule 423| | __kmp_itt_metadata_loop(loc, 0, trip_count, cur_chunk); 424| | } 425| |#endif 426| |#ifdef KMP_DEBUG 427| | { 428| | char *buff; 429| | // create format specifiers before the debug output 430| | buff = __kmp_str_format("__kmpc_for_static_init: liter=%%d lower=%%%s " 431| | "upper=%%%s stride = %%%s signed?<%s>\n", 432| | traits_t::spec, traits_t::spec, 433| | traits_t::spec, traits_t::spec); 434| | KD_TRACE(100, (buff, *plastiter, *plower, *pupper, *pstride)); 435| | __kmp_str_free(&buff); 436| | } 437| |#endif 438| 33.5k| KE_TRACE(10, ("__kmpc_for_static_init: T#%d return\n", global_tid)); 439| | 440| 33.5k|#if OMPT_SUPPORT && OMPT_OPTIONAL 441| 33.5k| if (ompt_enabled.ompt_callback_work) { ------------------ | Branch (441:7): [True: 0, False: 33.5k] ------------------ 442| 0| ompt_callbacks.ompt_callback(ompt_callback_work)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 443| 0| ompt_work_type, ompt_scope_begin, &(team_info->parallel_data), 444| 0| &(task_info->task_data), trip_count, codeptr); 445| 0| } 446| 33.5k| if (ompt_enabled.ompt_callback_dispatch) { ------------------ | Branch (446:7): [True: 0, False: 33.5k] ------------------ 447| 0| ompt_dispatch_t dispatch_type; 448| 0| ompt_data_t instance = ompt_data_none; ------------------ | | 1408| 0|#define ompt_data_none {0} ------------------ 449| 0| ompt_dispatch_chunk_t dispatch_chunk; 450| 0| if (ompt_work_type == ompt_work_sections) { ------------------ | Branch (450:9): [True: 0, False: 0] ------------------ 451| 0| dispatch_type = ompt_dispatch_section; 452| 0| instance.ptr = codeptr; 453| 0| } else { 454| 0| OMPT_GET_DISPATCH_CHUNK(dispatch_chunk, *plower, *pupper, incr); ------------------ | | 101| 0| do { \ | | 102| 0| if (incr > 0) { \ | | ------------------ | | | Branch (102:9): [True: 0, False: 0] | | ------------------ | | 103| 0| chunk.start = static_cast(lb); \ | | 104| 0| chunk.iterations = static_cast(((ub) - (lb)) / (incr) + 1); \ | | 105| 0| } else { \ | | 106| 0| chunk.start = static_cast(ub); \ | | 107| 0| chunk.iterations = static_cast(((lb) - (ub)) / -(incr) + 1); \ | | 108| 0| } \ | | 109| 0| } while (0) | | ------------------ | | | Branch (109:12): [Folded, False: 0] | | ------------------ ------------------ 455| 0| dispatch_type = (ompt_work_type == ompt_work_distribute) ------------------ | Branch (455:23): [True: 0, False: 0] ------------------ 456| 0| ? ompt_dispatch_distribute_chunk 457| 0| : ompt_dispatch_ws_loop_chunk; 458| 0| instance.ptr = &dispatch_chunk; 459| 0| } 460| 0| ompt_callbacks.ompt_callback(ompt_callback_dispatch)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 461| 0| &(team_info->parallel_data), &(task_info->task_data), dispatch_type, 462| 0| instance); 463| 0| } 464| 33.5k|#endif 465| | 466| 33.5k| KMP_STATS_LOOP_END(OMP_loop_static_iterations); 467| 33.5k| return; 468| 33.2k|} __kmp_check_stksize: 320| 1|void __kmp_check_stksize(size_t *val) { 321| | // if system stack size is too big then limit the size for worker threads 322| 1| if (*val > KMP_DEFAULT_STKSIZE * 16) // just a heuristics... ------------------ | | 1158| 1|#define KMP_DEFAULT_STKSIZE ((size_t)(4 * 1024 * 1024)) ------------------ | Branch (322:7): [True: 0, False: 1] ------------------ 323| 0| *val = KMP_DEFAULT_STKSIZE * 16; ------------------ | | 1158| 0|#define KMP_DEFAULT_STKSIZE ((size_t)(4 * 1024 * 1024)) ------------------ 324| 1| if (*val < __kmp_sys_min_stksize) ------------------ | Branch (324:7): [True: 0, False: 1] ------------------ 325| 0| *val = __kmp_sys_min_stksize; 326| 1| if (*val > KMP_MAX_STKSIZE) ------------------ | | 1153| 1|#define KMP_MAX_STKSIZE (~((size_t)1 << ((sizeof(size_t) * (1 << 3)) - 1))) ------------------ | Branch (326:7): [True: 0, False: 1] ------------------ 327| 0| *val = KMP_MAX_STKSIZE; // dead code currently, but may work in future ------------------ | | 1153| 0|#define KMP_MAX_STKSIZE (~((size_t)1 << ((sizeof(size_t) * (1 << 3)) - 1))) ------------------ 328| |#if KMP_OS_DARWIN 329| | *val = __kmp_round4k(*val); 330| |#endif // KMP_OS_DARWIN 331| 1|} _Z30__kmp_initial_threads_capacityi: 536| 2|int __kmp_initial_threads_capacity(int req_nproc) { 537| 2| int nth = 32; 538| | 539| | /* MIN( MAX( 32, 4 * $OMP_NUM_THREADS, 4 * omp_get_num_procs() ), 540| | * __kmp_max_nth) */ 541| 2| if (nth < (4 * req_nproc)) ------------------ | Branch (541:7): [True: 2, False: 0] ------------------ 542| 2| nth = (4 * req_nproc); 543| 2| if (nth < (4 * __kmp_xproc)) ------------------ | Branch (543:7): [True: 0, False: 2] ------------------ 544| 0| nth = (4 * __kmp_xproc); 545| | 546| | // If hidden helper task is enabled, we initialize the thread capacity with 547| | // extra __kmp_hidden_helper_threads_num. 548| 2| if (__kmp_enable_hidden_helper) { ------------------ | Branch (548:7): [True: 2, False: 0] ------------------ 549| 2| nth += __kmp_hidden_helper_threads_num; 550| 2| } 551| | 552| 2| if (nth > __kmp_max_nth) ------------------ | Branch (552:7): [True: 0, False: 2] ------------------ 553| 0| nth = __kmp_max_nth; 554| | 555| 2| return nth; 556| 2|} _Z25__kmp_default_tp_capacityiii: 559| 1| int all_threads_specified) { 560| 1| int nth = 128; 561| | 562| 1| if (all_threads_specified) ------------------ | Branch (562:7): [True: 0, False: 1] ------------------ 563| 0| return max_nth; 564| | /* MIN( MAX (128, 4 * $OMP_NUM_THREADS, 4 * omp_get_num_procs() ), 565| | * __kmp_max_nth ) */ 566| 1| if (nth < (4 * req_nproc)) ------------------ | Branch (566:7): [True: 0, False: 1] ------------------ 567| 0| nth = (4 * req_nproc); 568| 1| if (nth < (4 * __kmp_xproc)) ------------------ | Branch (568:7): [True: 0, False: 1] ------------------ 569| 0| nth = (4 * __kmp_xproc); 570| | 571| 1| if (nth > __kmp_max_nth) ------------------ | Branch (571:7): [True: 0, False: 1] ------------------ 572| 0| nth = __kmp_max_nth; 573| | 574| 1| return nth; 575| 1|} _Z20__kmp_env_initializePKc: 6003| 1|void __kmp_env_initialize(char const *string) { 6004| | 6005| 1| kmp_env_blk_t block; 6006| 1| int i; 6007| | 6008| 1| __kmp_stg_init(); 6009| | 6010| | // Hack!!! 6011| 1| if (string == NULL) { ------------------ | Branch (6011:7): [True: 1, False: 0] ------------------ 6012| | // __kmp_max_nth = __kmp_sys_max_nth; 6013| 1| __kmp_threads_capacity = 6014| 1| __kmp_initial_threads_capacity(__kmp_dflt_team_nth_ub); 6015| 1| } 6016| 1| __kmp_env_blk_init(&block, string); 6017| | 6018| | // update the set flag on all entries that have an env var 6019| 35| for (i = 0; i < block.count; ++i) { ------------------ | Branch (6019:15): [True: 34, False: 1] ------------------ 6020| 34| if ((block.vars[i].name == NULL) || (*block.vars[i].name == '\0')) { ------------------ | Branch (6020:9): [True: 0, False: 34] | Branch (6020:41): [True: 0, False: 34] ------------------ 6021| 0| continue; 6022| 0| } 6023| 34| if (block.vars[i].value == NULL) { ------------------ | Branch (6023:9): [True: 0, False: 34] ------------------ 6024| 0| continue; 6025| 0| } 6026| 34| kmp_setting_t *setting = __kmp_stg_find(block.vars[i].name); 6027| 34| if (setting != NULL) { ------------------ | Branch (6027:9): [True: 0, False: 34] ------------------ 6028| 0| setting->set = 1; 6029| 0| } 6030| 34| } 6031| | 6032| | // We need to know if blocktime was set when processing OMP_WAIT_POLICY 6033| 1| blocktime_str = __kmp_env_blk_var(&block, "KMP_BLOCKTIME"); 6034| | 6035| | // Special case. If we parse environment, not a string, process KMP_WARNINGS 6036| | // first. 6037| 1| if (string == NULL) { ------------------ | Branch (6037:7): [True: 1, False: 0] ------------------ 6038| 1| char const *name = "KMP_WARNINGS"; 6039| 1| char const *value = __kmp_env_blk_var(&block, name); 6040| 1| __kmp_stg_parse(name, value); 6041| 1| } 6042| | 6043| 1|#if KMP_AFFINITY_SUPPORTED 6044| | // Special case. KMP_AFFINITY is not a rival to other affinity env vars 6045| | // if no affinity type is specified. We want to allow 6046| | // KMP_AFFINITY=[no],verbose/[no]warnings/etc. to be enabled when 6047| | // specifying the affinity type via GOMP_CPU_AFFINITY or the OMP 4.0 6048| | // affinity mechanism. 6049| 1| __kmp_affinity_notype = NULL; 6050| 1| char const *aff_str = __kmp_env_blk_var(&block, "KMP_AFFINITY"); 6051| 1| if (aff_str != NULL) { ------------------ | Branch (6051:7): [True: 0, False: 1] ------------------ 6052| | // Check if the KMP_AFFINITY type is specified in the string. 6053| | // We just search the string for "compact", "scatter", etc. 6054| | // without really parsing the string. The syntax of the 6055| | // KMP_AFFINITY env var is such that none of the affinity 6056| | // type names can appear anywhere other that the type 6057| | // specifier, even as substrings. 6058| | // 6059| | // I can't find a case-insensitive version of strstr on Windows* OS. 6060| | // Use the case-sensitive version for now. 6061| | 6062| |#if KMP_OS_WINDOWS 6063| |#define FIND strstr 6064| |#else 6065| 0|#define FIND strcasestr 6066| 0|#endif 6067| | 6068| 0| if ((FIND(aff_str, "none") == NULL) && ------------------ | | 6065| 0|#define FIND strcasestr ------------------ | Branch (6068:9): [True: 0, False: 0] ------------------ 6069| 0| (FIND(aff_str, "physical") == NULL) && ------------------ | | 6065| 0|#define FIND strcasestr ------------------ | Branch (6069:9): [True: 0, False: 0] ------------------ 6070| 0| (FIND(aff_str, "logical") == NULL) && ------------------ | | 6065| 0|#define FIND strcasestr ------------------ | Branch (6070:9): [True: 0, False: 0] ------------------ 6071| 0| (FIND(aff_str, "compact") == NULL) && ------------------ | | 6065| 0|#define FIND strcasestr ------------------ | Branch (6071:9): [True: 0, False: 0] ------------------ 6072| 0| (FIND(aff_str, "scatter") == NULL) && ------------------ | | 6065| 0|#define FIND strcasestr ------------------ | Branch (6072:9): [True: 0, False: 0] ------------------ 6073| 0| (FIND(aff_str, "explicit") == NULL) && ------------------ | | 6065| 0|#define FIND strcasestr ------------------ | Branch (6073:9): [True: 0, False: 0] ------------------ 6074| 0| (FIND(aff_str, "balanced") == NULL) && ------------------ | | 6065| 0|#define FIND strcasestr ------------------ | Branch (6074:9): [True: 0, False: 0] ------------------ 6075| 0| (FIND(aff_str, "disabled") == NULL)) { ------------------ | | 6065| 0|#define FIND strcasestr ------------------ | Branch (6075:9): [True: 0, False: 0] ------------------ 6076| 0| __kmp_affinity_notype = __kmp_stg_find("KMP_AFFINITY"); 6077| 0| } else { 6078| | // A new affinity type is specified. 6079| | // Reset the affinity flags to their default values, 6080| | // in case this is called from kmp_set_defaults(). 6081| 0| __kmp_affinity.type = affinity_default; 6082| 0| __kmp_affinity.gran = KMP_HW_UNKNOWN; 6083| 0| __kmp_affinity_top_method = affinity_top_method_default; 6084| 0| __kmp_affinity.flags.respect = affinity_respect_mask_default; ------------------ | | 860| 0|#define affinity_respect_mask_default (2) ------------------ 6085| 0| } 6086| 0|#undef FIND 6087| | 6088| | // Also reset the affinity flags if OMP_PROC_BIND is specified. 6089| 0| aff_str = __kmp_env_blk_var(&block, "OMP_PROC_BIND"); 6090| 0| if (aff_str != NULL) { ------------------ | Branch (6090:9): [True: 0, False: 0] ------------------ 6091| 0| __kmp_affinity.type = affinity_default; 6092| 0| __kmp_affinity.gran = KMP_HW_UNKNOWN; 6093| 0| __kmp_affinity_top_method = affinity_top_method_default; 6094| 0| __kmp_affinity.flags.respect = affinity_respect_mask_default; ------------------ | | 860| 0|#define affinity_respect_mask_default (2) ------------------ 6095| 0| } 6096| 0| } 6097| | 6098| 1|#endif /* KMP_AFFINITY_SUPPORTED */ 6099| | 6100| | // Set up the nested proc bind type vector. 6101| 1| if (__kmp_nested_proc_bind.bind_types == NULL) { ------------------ | Branch (6101:7): [True: 1, False: 0] ------------------ 6102| 1| __kmp_nested_proc_bind.bind_types = 6103| 1| (kmp_proc_bind_t *)KMP_INTERNAL_MALLOC(sizeof(kmp_proc_bind_t)); ------------------ | | 110| 1|#define KMP_INTERNAL_MALLOC(sz) malloc(sz) ------------------ 6104| 1| if (__kmp_nested_proc_bind.bind_types == NULL) { ------------------ | Branch (6104:9): [True: 0, False: 1] ------------------ 6105| 0| KMP_FATAL(MemoryAllocFailed); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 6106| 0| } 6107| 1| __kmp_nested_proc_bind.size = 1; 6108| 1| __kmp_nested_proc_bind.used = 1; 6109| 1|#if KMP_AFFINITY_SUPPORTED 6110| 1| __kmp_nested_proc_bind.bind_types[0] = proc_bind_default; 6111| |#else 6112| | // default proc bind is false if affinity not supported 6113| | __kmp_nested_proc_bind.bind_types[0] = proc_bind_false; 6114| |#endif 6115| 1| } 6116| | 6117| | // Set up the affinity format ICV 6118| | // Grab the default affinity format string from the message catalog 6119| 1| kmp_msg_t m = 6120| 1| __kmp_msg_format(kmp_i18n_msg_AffFormatDefault, "%P", "%i", "%n", "%A"); 6121| 1| KMP_DEBUG_ASSERT(KMP_STRLEN(m.str) < KMP_AFFINITY_FORMAT_SIZE); 6122| | 6123| 1| if (__kmp_affinity_format == NULL) { ------------------ | Branch (6123:7): [True: 1, False: 0] ------------------ 6124| 1| __kmp_affinity_format = 6125| 1| (char *)KMP_INTERNAL_MALLOC(sizeof(char) * KMP_AFFINITY_FORMAT_SIZE); ------------------ | | 110| 1|#define KMP_INTERNAL_MALLOC(sz) malloc(sz) ------------------ 6126| 1| } 6127| 1| KMP_STRCPY_S(__kmp_affinity_format, KMP_AFFINITY_FORMAT_SIZE, m.str); ------------------ | | 63| 1|#define KMP_STRCPY_S(dst, bsz, src) strcpy(dst, src) ------------------ 6128| 1| __kmp_str_free(&m.str); 6129| | 6130| | // Now process all of the settings. 6131| 35| for (i = 0; i < block.count; ++i) { ------------------ | Branch (6131:15): [True: 34, False: 1] ------------------ 6132| 34| __kmp_stg_parse(block.vars[i].name, block.vars[i].value); 6133| 34| } 6134| | 6135| | // If user locks have been allocated yet, don't reset the lock vptr table. 6136| 1| if (!__kmp_init_user_locks) { ------------------ | Branch (6136:7): [True: 1, False: 0] ------------------ 6137| 1| if (__kmp_user_lock_kind == lk_default) { ------------------ | Branch (6137:9): [True: 1, False: 0] ------------------ 6138| 1| __kmp_user_lock_kind = lk_queuing; 6139| 1| } 6140| 1|#if KMP_USE_DYNAMIC_LOCK 6141| 1| __kmp_init_dynamic_user_locks(); 6142| |#else 6143| | __kmp_set_user_lock_vptrs(__kmp_user_lock_kind); 6144| |#endif 6145| 1| } else { 6146| 0| KMP_DEBUG_ASSERT(string != NULL); // kmp_set_defaults() was called 6147| 0| KMP_DEBUG_ASSERT(__kmp_user_lock_kind != lk_default); 6148| |// Binds lock functions again to follow the transition between different 6149| |// KMP_CONSISTENCY_CHECK values. Calling this again is harmless as long 6150| |// as we do not allow lock kind changes after making a call to any 6151| |// user lock functions (true). 6152| 0|#if KMP_USE_DYNAMIC_LOCK 6153| 0| __kmp_init_dynamic_user_locks(); 6154| |#else 6155| | __kmp_set_user_lock_vptrs(__kmp_user_lock_kind); 6156| |#endif 6157| 0| } 6158| | 6159| 1|#if KMP_AFFINITY_SUPPORTED 6160| | 6161| 1| if (!TCR_4(__kmp_init_middle)) { ------------------ | | 1127| 1|#define TCR_4(a) (a) ------------------ | Branch (6161:7): [True: 1, False: 0] ------------------ 6162| |#if KMP_USE_HWLOC 6163| | // Force using hwloc when either tiles or numa nodes requested within 6164| | // KMP_HW_SUBSET or granularity setting and no other topology method 6165| | // is requested 6166| | if (__kmp_hw_subset && 6167| | __kmp_affinity_top_method == affinity_top_method_default) 6168| | if (__kmp_hw_subset->specified(KMP_HW_NUMA) || 6169| | __kmp_hw_subset->specified(KMP_HW_TILE) || 6170| | __kmp_affinity.gran == KMP_HW_TILE || 6171| | __kmp_affinity.gran == KMP_HW_NUMA) 6172| | __kmp_affinity_top_method = affinity_top_method_hwloc; 6173| | // Force using hwloc when tiles or numa nodes requested for OMP_PLACES 6174| | if (__kmp_affinity.gran == KMP_HW_NUMA || 6175| | __kmp_affinity.gran == KMP_HW_TILE) 6176| | __kmp_affinity_top_method = affinity_top_method_hwloc; 6177| |#endif 6178| | // Determine if the machine/OS is actually capable of supporting 6179| | // affinity. 6180| 1| const char *var = "KMP_AFFINITY"; 6181| 1| KMPAffinity::pick_api(); 6182| |#if KMP_USE_HWLOC 6183| | // If Hwloc topology discovery was requested but affinity was also disabled, 6184| | // then tell user that Hwloc request is being ignored and use default 6185| | // topology discovery method. 6186| | if (__kmp_affinity_top_method == affinity_top_method_hwloc && 6187| | __kmp_affinity_dispatch->get_api_type() != KMPAffinity::HWLOC) { 6188| | KMP_WARNING(AffIgnoringHwloc, var); 6189| | __kmp_affinity_top_method = affinity_top_method_all; 6190| | } 6191| |#endif 6192| 1| if (__kmp_affinity.type == affinity_disabled) { ------------------ | Branch (6192:9): [True: 0, False: 1] ------------------ 6193| 0| KMP_AFFINITY_DISABLE(); ------------------ | | 685| 0|#define KMP_AFFINITY_DISABLE() (__kmp_affin_mask_size = 0) ------------------ 6194| 1| } else if (!KMP_AFFINITY_CAPABLE()) { ------------------ | | 684| 1|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) ------------------ | Branch (6194:16): [True: 1, False: 0] ------------------ 6195| 1| __kmp_affinity_dispatch->determine_capable(var); 6196| 1| if (!KMP_AFFINITY_CAPABLE()) { ------------------ | | 684| 1|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) ------------------ | Branch (6196:11): [True: 0, False: 1] ------------------ 6197| 0| if (__kmp_affinity.flags.verbose || ------------------ | Branch (6197:13): [True: 0, False: 0] ------------------ 6198| 0| (__kmp_affinity.flags.warnings && ------------------ | Branch (6198:14): [True: 0, False: 0] ------------------ 6199| 0| (__kmp_affinity.type != affinity_default) && ------------------ | Branch (6199:14): [True: 0, False: 0] ------------------ 6200| 0| (__kmp_affinity.type != affinity_none) && ------------------ | Branch (6200:14): [True: 0, False: 0] ------------------ 6201| 0| (__kmp_affinity.type != affinity_disabled))) { ------------------ | Branch (6201:14): [True: 0, False: 0] ------------------ 6202| 0| KMP_WARNING(AffNotSupported, var); ------------------ | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 6203| 0| } 6204| 0| __kmp_affinity.type = affinity_disabled; 6205| 0| __kmp_affinity.flags.respect = FALSE; ------------------ | | 1279| 0|#define FALSE 0 ------------------ 6206| 0| __kmp_affinity.gran = KMP_HW_THREAD; 6207| 0| } 6208| 1| } 6209| | 6210| 1| if (__kmp_affinity.type == affinity_disabled) { ------------------ | Branch (6210:9): [True: 0, False: 1] ------------------ 6211| 0| __kmp_nested_proc_bind.bind_types[0] = proc_bind_false; 6212| 1| } else if (__kmp_nested_proc_bind.bind_types[0] == proc_bind_true) { ------------------ | Branch (6212:16): [True: 0, False: 1] ------------------ 6213| | // OMP_PROC_BIND=true maps to OMP_PROC_BIND=spread. 6214| 0| __kmp_nested_proc_bind.bind_types[0] = proc_bind_spread; 6215| 0| } 6216| | 6217| 1| if (KMP_AFFINITY_CAPABLE()) { ------------------ | | 684| 1|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) | | ------------------ | | | Branch (684:32): [True: 1, False: 0] | | ------------------ ------------------ 6218| | 6219| |#if KMP_GROUP_AFFINITY 6220| | // This checks to see if the initial affinity mask is equal 6221| | // to a single windows processor group. If it is, then we do 6222| | // not respect the initial affinity mask and instead, use the 6223| | // entire machine. 6224| | bool exactly_one_group = false; 6225| | if (__kmp_num_proc_groups > 1) { 6226| | int group; 6227| | bool within_one_group; 6228| | // Get the initial affinity mask and determine if it is 6229| | // contained within a single group. 6230| | kmp_affin_mask_t *init_mask; 6231| | KMP_CPU_ALLOC(init_mask); 6232| | __kmp_get_system_affinity(init_mask, TRUE); 6233| | group = __kmp_get_proc_group(init_mask); 6234| | within_one_group = (group >= 0); 6235| | // If the initial affinity is within a single group, 6236| | // then determine if it is equal to that single group. 6237| | if (within_one_group) { 6238| | DWORD num_bits_in_group = __kmp_GetActiveProcessorCount(group); 6239| | DWORD num_bits_in_mask = 0; 6240| | for (int bit = init_mask->begin(); bit != init_mask->end(); 6241| | bit = init_mask->next(bit)) 6242| | num_bits_in_mask++; 6243| | exactly_one_group = (num_bits_in_group == num_bits_in_mask); 6244| | } 6245| | KMP_CPU_FREE(init_mask); 6246| | } 6247| | 6248| | // Handle the Win 64 group affinity stuff if there are multiple 6249| | // processor groups, or if the user requested it, and OMP 4.0 6250| | // affinity is not in effect. 6251| | if (__kmp_num_proc_groups > 1 && 6252| | __kmp_affinity.type == affinity_default && 6253| | __kmp_nested_proc_bind.bind_types[0] == proc_bind_default) { 6254| | // Do not respect the initial processor affinity mask if it is assigned 6255| | // exactly one Windows Processor Group since this is interpreted as the 6256| | // default OS assignment. Not respecting the mask allows the runtime to 6257| | // use all the logical processors in all groups. 6258| | if (__kmp_affinity.flags.respect == affinity_respect_mask_default && 6259| | exactly_one_group) { 6260| | __kmp_affinity.flags.respect = FALSE; 6261| | } 6262| | // Use compact affinity with anticipation of pinning to at least the 6263| | // group granularity since threads can only be bound to one group. 6264| | if (__kmp_affinity.type == affinity_default) { 6265| | __kmp_affinity.type = affinity_compact; 6266| | __kmp_nested_proc_bind.bind_types[0] = proc_bind_intel; 6267| | } 6268| | if (__kmp_hh_affinity.type == affinity_default) 6269| | __kmp_hh_affinity.type = affinity_compact; 6270| | if (__kmp_affinity_top_method == affinity_top_method_default) 6271| | __kmp_affinity_top_method = affinity_top_method_all; 6272| | if (__kmp_affinity.gran == KMP_HW_UNKNOWN) 6273| | __kmp_affinity.gran = KMP_HW_PROC_GROUP; 6274| | if (__kmp_hh_affinity.gran == KMP_HW_UNKNOWN) 6275| | __kmp_hh_affinity.gran = KMP_HW_PROC_GROUP; 6276| | } else 6277| | 6278| |#endif /* KMP_GROUP_AFFINITY */ 6279| | 6280| 1| { 6281| 1| if (__kmp_affinity.flags.respect == affinity_respect_mask_default) { ------------------ | | 860| 1|#define affinity_respect_mask_default (2) ------------------ | Branch (6281:13): [True: 1, False: 0] ------------------ 6282| |#if KMP_GROUP_AFFINITY 6283| | if (__kmp_num_proc_groups > 1 && exactly_one_group) { 6284| | __kmp_affinity.flags.respect = FALSE; 6285| | } else 6286| |#endif /* KMP_GROUP_AFFINITY */ 6287| 1| { 6288| 1| __kmp_affinity.flags.respect = TRUE; ------------------ | | 1280| 1|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1|#define FALSE 0 | | ------------------ ------------------ 6289| 1| } 6290| 1| } 6291| 1| if ((__kmp_nested_proc_bind.bind_types[0] != proc_bind_intel) && ------------------ | Branch (6291:13): [True: 1, False: 0] ------------------ 6292| 1| (__kmp_nested_proc_bind.bind_types[0] != proc_bind_default)) { ------------------ | Branch (6292:13): [True: 0, False: 1] ------------------ 6293| 0| if (__kmp_affinity.type == affinity_default) { ------------------ | Branch (6293:15): [True: 0, False: 0] ------------------ 6294| 0| __kmp_affinity.type = affinity_compact; 6295| 0| __kmp_affinity.flags.dups = FALSE; ------------------ | | 1279| 0|#define FALSE 0 ------------------ 6296| 0| } 6297| 1| } else if (__kmp_affinity.type == affinity_default) { ------------------ | Branch (6297:20): [True: 1, False: 0] ------------------ 6298| 1|#if KMP_MIC_SUPPORTED 6299| 1| if (__kmp_mic_type != non_mic) { ------------------ | Branch (6299:15): [True: 0, False: 1] ------------------ 6300| 0| __kmp_nested_proc_bind.bind_types[0] = proc_bind_intel; 6301| 0| } else 6302| 1|#endif 6303| 1| { 6304| 1| __kmp_nested_proc_bind.bind_types[0] = proc_bind_false; 6305| 1| } 6306| 1|#if KMP_MIC_SUPPORTED 6307| 1| if (__kmp_mic_type != non_mic) { ------------------ | Branch (6307:15): [True: 0, False: 1] ------------------ 6308| 0| __kmp_affinity.type = affinity_scatter; 6309| 0| } else 6310| 1|#endif 6311| 1| { 6312| 1| __kmp_affinity.type = affinity_none; 6313| 1| } 6314| 1| } 6315| 1| if (__kmp_hh_affinity.type == affinity_default) ------------------ | Branch (6315:13): [True: 1, False: 0] ------------------ 6316| 1| __kmp_hh_affinity.type = affinity_none; 6317| 1| if ((__kmp_affinity.gran == KMP_HW_UNKNOWN) && ------------------ | Branch (6317:13): [True: 1, False: 0] ------------------ 6318| 1| (__kmp_affinity.gran_levels < 0)) { ------------------ | Branch (6318:13): [True: 1, False: 0] ------------------ 6319| 1|#if KMP_MIC_SUPPORTED 6320| 1| if (__kmp_mic_type != non_mic) { ------------------ | Branch (6320:15): [True: 0, False: 1] ------------------ 6321| 0| __kmp_affinity.gran = KMP_HW_THREAD; 6322| 0| } else 6323| 1|#endif 6324| 1| { 6325| 1| __kmp_affinity.gran = KMP_HW_CORE; 6326| 1| } 6327| 1| } 6328| 1| if ((__kmp_hh_affinity.gran == KMP_HW_UNKNOWN) && ------------------ | Branch (6328:13): [True: 1, False: 0] ------------------ 6329| 1| (__kmp_hh_affinity.gran_levels < 0)) { ------------------ | Branch (6329:13): [True: 1, False: 0] ------------------ 6330| 1|#if KMP_MIC_SUPPORTED 6331| 1| if (__kmp_mic_type != non_mic) { ------------------ | Branch (6331:15): [True: 0, False: 1] ------------------ 6332| 0| __kmp_hh_affinity.gran = KMP_HW_THREAD; 6333| 0| } else 6334| 1|#endif 6335| 1| { 6336| 1| __kmp_hh_affinity.gran = KMP_HW_CORE; 6337| 1| } 6338| 1| } 6339| 1| if (__kmp_affinity_top_method == affinity_top_method_default) { ------------------ | Branch (6339:13): [True: 1, False: 0] ------------------ 6340| 1| __kmp_affinity_top_method = affinity_top_method_all; 6341| 1| } 6342| 1| } 6343| 1| } else { 6344| | // If affinity is disabled, then still need to assign topology method 6345| | // to attempt machine detection and affinity types 6346| 0| if (__kmp_affinity_top_method == affinity_top_method_default) ------------------ | Branch (6346:11): [True: 0, False: 0] ------------------ 6347| 0| __kmp_affinity_top_method = affinity_top_method_all; 6348| 0| if (__kmp_affinity.type == affinity_default) ------------------ | Branch (6348:11): [True: 0, False: 0] ------------------ 6349| 0| __kmp_affinity.type = affinity_disabled; 6350| 0| if (__kmp_hh_affinity.type == affinity_default) ------------------ | Branch (6350:11): [True: 0, False: 0] ------------------ 6351| 0| __kmp_hh_affinity.type = affinity_disabled; 6352| 0| } 6353| | 6354| |#ifdef KMP_DEBUG 6355| | for (const kmp_affinity_t *affinity : __kmp_affinities) 6356| | __kmp_print_affinity_settings(affinity); 6357| | KMP_DEBUG_ASSERT(__kmp_nested_proc_bind.bind_types[0] != proc_bind_default); 6358| | K_DIAG(1, ("__kmp_nested_proc_bind.bind_types[0] == %d\n", 6359| | __kmp_nested_proc_bind.bind_types[0])); 6360| |#endif 6361| 1| } 6362| | 6363| 1|#endif /* KMP_AFFINITY_SUPPORTED */ 6364| | 6365| | // Post-initialization step: some env. vars need their value's further 6366| | // processing 6367| 1| if (string != NULL) { // kmp_set_defaults() was called ------------------ | Branch (6367:7): [True: 0, False: 1] ------------------ 6368| 0| __kmp_aux_env_initialize(&block); 6369| 0| } 6370| | 6371| 1| __kmp_env_blk_free(&block); 6372| | 6373| 1| KMP_MB(); 6374| | 6375| 1|} // __kmp_env_initialize _Z14__kmp_env_dumpv: 6464| 1|void __kmp_env_dump() { 6465| | 6466| 1| kmp_env_blk_t block; 6467| 1| kmp_str_buf_t buffer, env, notdefined; 6468| | 6469| 1| __kmp_stg_init(); 6470| 1| __kmp_str_buf_init(&buffer); ------------------ | | 41| 1| { \ | | 42| 1| (b)->str = (b)->bulk; \ | | 43| 1| (b)->size = sizeof((b)->bulk); \ | | 44| 1| (b)->used = 0; \ | | 45| 1| (b)->bulk[0] = 0; \ | | 46| 1| } ------------------ 6471| 1| __kmp_str_buf_init(&env); ------------------ | | 41| 1| { \ | | 42| 1| (b)->str = (b)->bulk; \ | | 43| 1| (b)->size = sizeof((b)->bulk); \ | | 44| 1| (b)->used = 0; \ | | 45| 1| (b)->bulk[0] = 0; \ | | 46| 1| } ------------------ 6472| 1| __kmp_str_buf_init(¬defined); ------------------ | | 41| 1| { \ | | 42| 1| (b)->str = (b)->bulk; \ | | 43| 1| (b)->size = sizeof((b)->bulk); \ | | 44| 1| (b)->used = 0; \ | | 45| 1| (b)->bulk[0] = 0; \ | | 46| 1| } ------------------ 6473| | 6474| 1| __kmp_env_blk_init(&block, NULL); 6475| 1| __kmp_env_blk_sort(&block); 6476| | 6477| 1| __kmp_str_buf_print(¬defined, ": %s", KMP_I18N_STR(NotDefined)); ------------------ | | 46| 1|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) ------------------ 6478| | 6479| 88| for (int i = 0; i < __kmp_stg_count; ++i) { ------------------ | Branch (6479:19): [True: 87, False: 1] ------------------ 6480| 87| if (__kmp_stg_table[i].print == NULL) ------------------ | Branch (6480:9): [True: 4, False: 83] ------------------ 6481| 4| continue; 6482| 83| __kmp_str_buf_clear(&env); 6483| 83| __kmp_stg_table[i].print(&env, __kmp_stg_table[i].name, 6484| 83| __kmp_stg_table[i].data); 6485| 83| if (env.used < 4) // valid definition must have indents (3) and a new line ------------------ | Branch (6485:9): [True: 8, False: 75] ------------------ 6486| 8| continue; 6487| 75| if (strstr(env.str, notdefined.str)) ------------------ | Branch (6487:9): [True: 7, False: 68] ------------------ 6488| | // normalize the string 6489| 7| __kmp_str_buf_print(&buffer, "%s=undefined\n", __kmp_stg_table[i].name); 6490| 68| else 6491| 68| __kmp_str_buf_cat(&buffer, env.str + 3, env.used - 3); 6492| 75| } 6493| | 6494| 1| ompd_env_block = (char *)__kmp_allocate(buffer.used + 1); ------------------ | | 3617| 1|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 6495| 1| KMP_MEMCPY(ompd_env_block, buffer.str, buffer.used + 1); ------------------ | | 69| 1|#define KMP_MEMCPY memcpy ------------------ 6496| 1| ompd_env_block_size = (ompd_size_t)KMP_STRLEN(ompd_env_block); ------------------ | | 68| 1|#define KMP_STRLEN strlen ------------------ 6497| | 6498| 1| __kmp_env_blk_free(&block); 6499| 1| __kmp_str_buf_free(&buffer); 6500| 1| __kmp_str_buf_free(&env); 6501| 1| __kmp_str_buf_free(¬defined); 6502| 1|} kmp_settings.cpp:_ZL14__kmp_stg_initv: 5695| 2|static void __kmp_stg_init(void) { 5696| | 5697| 2| static int initialized = 0; 5698| | 5699| 2| if (!initialized) { ------------------ | Branch (5699:7): [True: 1, False: 1] ------------------ 5700| | 5701| | // Sort table. 5702| 1| qsort(__kmp_stg_table, __kmp_stg_count - 1, sizeof(kmp_setting_t), 5703| 1| __kmp_stg_cmp); 5704| | 5705| 1| { // Initialize *_STACKSIZE data. 5706| 1| kmp_setting_t *kmp_stacksize = 5707| 1| __kmp_stg_find("KMP_STACKSIZE"); // 1st priority. 5708| 1|#ifdef KMP_GOMP_COMPAT 5709| 1| kmp_setting_t *gomp_stacksize = 5710| 1| __kmp_stg_find("GOMP_STACKSIZE"); // 2nd priority. 5711| 1|#endif 5712| 1| kmp_setting_t *omp_stacksize = 5713| 1| __kmp_stg_find("OMP_STACKSIZE"); // 3rd priority. 5714| | 5715| | // !!! volatile keyword is Intel(R) C Compiler bug CQ49908 workaround. 5716| | // !!! Compiler does not understand rivals is used and optimizes out 5717| | // assignments 5718| | // !!! rivals[ i ++ ] = ...; 5719| 1| static kmp_setting_t *volatile rivals[4]; 5720| 1| static kmp_stg_ss_data_t kmp_data = {1, CCAST(kmp_setting_t **, rivals)}; ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5721| 1|#ifdef KMP_GOMP_COMPAT 5722| 1| static kmp_stg_ss_data_t gomp_data = {1024, 5723| 1| CCAST(kmp_setting_t **, rivals)}; ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5724| 1|#endif 5725| 1| static kmp_stg_ss_data_t omp_data = {1024, 5726| 1| CCAST(kmp_setting_t **, rivals)}; ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5727| 1| int i = 0; 5728| | 5729| 1| rivals[i++] = kmp_stacksize; 5730| 1|#ifdef KMP_GOMP_COMPAT 5731| 1| if (gomp_stacksize != NULL) { ------------------ | Branch (5731:11): [True: 1, False: 0] ------------------ 5732| 1| rivals[i++] = gomp_stacksize; 5733| 1| } 5734| 1|#endif 5735| 1| rivals[i++] = omp_stacksize; 5736| 1| rivals[i++] = NULL; 5737| | 5738| 1| kmp_stacksize->data = &kmp_data; 5739| 1|#ifdef KMP_GOMP_COMPAT 5740| 1| if (gomp_stacksize != NULL) { ------------------ | Branch (5740:11): [True: 1, False: 0] ------------------ 5741| 1| gomp_stacksize->data = &gomp_data; 5742| 1| } 5743| 1|#endif 5744| 1| omp_stacksize->data = &omp_data; 5745| 1| } 5746| | 5747| 1| { // Initialize KMP_LIBRARY and OMP_WAIT_POLICY data. 5748| 1| kmp_setting_t *kmp_library = 5749| 1| __kmp_stg_find("KMP_LIBRARY"); // 1st priority. 5750| 1| kmp_setting_t *omp_wait_policy = 5751| 1| __kmp_stg_find("OMP_WAIT_POLICY"); // 2nd priority. 5752| | 5753| | // !!! volatile keyword is Intel(R) C Compiler bug CQ49908 workaround. 5754| 1| static kmp_setting_t *volatile rivals[3]; 5755| 1| static kmp_stg_wp_data_t kmp_data = {0, CCAST(kmp_setting_t **, rivals)}; ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5756| 1| static kmp_stg_wp_data_t omp_data = {1, CCAST(kmp_setting_t **, rivals)}; ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5757| 1| int i = 0; 5758| | 5759| 1| rivals[i++] = kmp_library; 5760| 1| if (omp_wait_policy != NULL) { ------------------ | Branch (5760:11): [True: 1, False: 0] ------------------ 5761| 1| rivals[i++] = omp_wait_policy; 5762| 1| } 5763| 1| rivals[i++] = NULL; 5764| | 5765| 1| kmp_library->data = &kmp_data; 5766| 1| if (omp_wait_policy != NULL) { ------------------ | Branch (5766:11): [True: 1, False: 0] ------------------ 5767| 1| omp_wait_policy->data = &omp_data; 5768| 1| } 5769| 1| } 5770| | 5771| 1| { // Initialize KMP_DEVICE_THREAD_LIMIT and KMP_ALL_THREADS 5772| 1| kmp_setting_t *kmp_device_thread_limit = 5773| 1| __kmp_stg_find("KMP_DEVICE_THREAD_LIMIT"); // 1st priority. 5774| 1| kmp_setting_t *kmp_all_threads = 5775| 1| __kmp_stg_find("KMP_ALL_THREADS"); // 2nd priority. 5776| | 5777| | // !!! volatile keyword is Intel(R) C Compiler bug CQ49908 workaround. 5778| 1| static kmp_setting_t *volatile rivals[3]; 5779| 1| int i = 0; 5780| | 5781| 1| rivals[i++] = kmp_device_thread_limit; 5782| 1| rivals[i++] = kmp_all_threads; 5783| 1| rivals[i++] = NULL; 5784| | 5785| 1| kmp_device_thread_limit->data = CCAST(kmp_setting_t **, rivals); ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5786| 1| kmp_all_threads->data = CCAST(kmp_setting_t **, rivals); ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5787| 1| } 5788| | 5789| 1| { // Initialize KMP_HW_SUBSET and KMP_PLACE_THREADS 5790| | // 1st priority 5791| 1| kmp_setting_t *kmp_hw_subset = __kmp_stg_find("KMP_HW_SUBSET"); 5792| | // 2nd priority 5793| 1| kmp_setting_t *kmp_place_threads = __kmp_stg_find("KMP_PLACE_THREADS"); 5794| | 5795| | // !!! volatile keyword is Intel(R) C Compiler bug CQ49908 workaround. 5796| 1| static kmp_setting_t *volatile rivals[3]; 5797| 1| int i = 0; 5798| | 5799| 1| rivals[i++] = kmp_hw_subset; 5800| 1| rivals[i++] = kmp_place_threads; 5801| 1| rivals[i++] = NULL; 5802| | 5803| 1| kmp_hw_subset->data = CCAST(kmp_setting_t **, rivals); ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5804| 1| kmp_place_threads->data = CCAST(kmp_setting_t **, rivals); ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5805| 1| } 5806| | 5807| 1|#if KMP_AFFINITY_SUPPORTED 5808| 1| { // Initialize KMP_AFFINITY, GOMP_CPU_AFFINITY, and OMP_PROC_BIND data. 5809| 1| kmp_setting_t *kmp_affinity = 5810| 1| __kmp_stg_find("KMP_AFFINITY"); // 1st priority. 5811| 1| KMP_DEBUG_ASSERT(kmp_affinity != NULL); 5812| | 5813| 1|#ifdef KMP_GOMP_COMPAT 5814| 1| kmp_setting_t *gomp_cpu_affinity = 5815| 1| __kmp_stg_find("GOMP_CPU_AFFINITY"); // 2nd priority. 5816| 1| KMP_DEBUG_ASSERT(gomp_cpu_affinity != NULL); 5817| 1|#endif 5818| | 5819| 1| kmp_setting_t *omp_proc_bind = 5820| 1| __kmp_stg_find("OMP_PROC_BIND"); // 3rd priority. 5821| 1| KMP_DEBUG_ASSERT(omp_proc_bind != NULL); 5822| | 5823| | // !!! volatile keyword is Intel(R) C Compiler bug CQ49908 workaround. 5824| 1| static kmp_setting_t *volatile rivals[4]; 5825| 1| int i = 0; 5826| | 5827| 1| rivals[i++] = kmp_affinity; 5828| | 5829| 1|#ifdef KMP_GOMP_COMPAT 5830| 1| rivals[i++] = gomp_cpu_affinity; 5831| 1| gomp_cpu_affinity->data = CCAST(kmp_setting_t **, rivals); ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5832| 1|#endif 5833| | 5834| 1| rivals[i++] = omp_proc_bind; 5835| 1| omp_proc_bind->data = CCAST(kmp_setting_t **, rivals); ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5836| 1| rivals[i++] = NULL; 5837| | 5838| 1| static kmp_setting_t *volatile places_rivals[4]; 5839| 1| i = 0; 5840| | 5841| 1| kmp_setting_t *omp_places = __kmp_stg_find("OMP_PLACES"); // 3rd priority. 5842| 1| KMP_DEBUG_ASSERT(omp_places != NULL); 5843| | 5844| 1| places_rivals[i++] = kmp_affinity; 5845| 1|#ifdef KMP_GOMP_COMPAT 5846| 1| places_rivals[i++] = gomp_cpu_affinity; 5847| 1|#endif 5848| 1| places_rivals[i++] = omp_places; 5849| 1| omp_places->data = CCAST(kmp_setting_t **, places_rivals); ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5850| 1| places_rivals[i++] = NULL; 5851| 1| } 5852| |#else 5853| |// KMP_AFFINITY not supported, so OMP_PROC_BIND has no rivals. 5854| |// OMP_PLACES not supported yet. 5855| |#endif // KMP_AFFINITY_SUPPORTED 5856| | 5857| 1| { // Initialize KMP_DETERMINISTIC_REDUCTION and KMP_FORCE_REDUCTION data. 5858| 1| kmp_setting_t *kmp_force_red = 5859| 1| __kmp_stg_find("KMP_FORCE_REDUCTION"); // 1st priority. 5860| 1| kmp_setting_t *kmp_determ_red = 5861| 1| __kmp_stg_find("KMP_DETERMINISTIC_REDUCTION"); // 2nd priority. 5862| | 5863| | // !!! volatile keyword is Intel(R) C Compiler bug CQ49908 workaround. 5864| 1| static kmp_setting_t *volatile rivals[3]; 5865| 1| static kmp_stg_fr_data_t force_data = {1, 5866| 1| CCAST(kmp_setting_t **, rivals)}; ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5867| 1| static kmp_stg_fr_data_t determ_data = {0, 5868| 1| CCAST(kmp_setting_t **, rivals)}; ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ 5869| 1| int i = 0; 5870| | 5871| 1| rivals[i++] = kmp_force_red; 5872| 1| if (kmp_determ_red != NULL) { ------------------ | Branch (5872:11): [True: 1, False: 0] ------------------ 5873| 1| rivals[i++] = kmp_determ_red; 5874| 1| } 5875| 1| rivals[i++] = NULL; 5876| | 5877| 1| kmp_force_red->data = &force_data; 5878| 1| if (kmp_determ_red != NULL) { ------------------ | Branch (5878:11): [True: 1, False: 0] ------------------ 5879| 1| kmp_determ_red->data = &determ_data; 5880| 1| } 5881| 1| } 5882| | 5883| 1| initialized = 1; 5884| 1| } 5885| | 5886| | // Reset flags. 5887| 2| int i; 5888| 176| for (i = 0; i < __kmp_stg_count; ++i) { ------------------ | Branch (5888:15): [True: 174, False: 2] ------------------ 5889| 174| __kmp_stg_table[i].set = 0; 5890| 174| } 5891| | 5892| 2|} // __kmp_stg_init kmp_settings.cpp:_ZL13__kmp_stg_cmpPKvS0_: 5678| 431|static int __kmp_stg_cmp(void const *_a, void const *_b) { 5679| 431| const kmp_setting_t *a = RCAST(const kmp_setting_t *, _a); ------------------ | | 230| 431|#define RCAST(type, var) reinterpret_cast(var) ------------------ 5680| 431| const kmp_setting_t *b = RCAST(const kmp_setting_t *, _b); ------------------ | | 230| 431|#define RCAST(type, var) reinterpret_cast(var) ------------------ 5681| | 5682| | // Process KMP_AFFINITY last. 5683| | // It needs to come after OMP_PLACES and GOMP_CPU_AFFINITY. 5684| 431| if (strcmp(a->name, "KMP_AFFINITY") == 0) { ------------------ | Branch (5684:7): [True: 7, False: 424] ------------------ 5685| 7| if (strcmp(b->name, "KMP_AFFINITY") == 0) { ------------------ | Branch (5685:9): [True: 0, False: 7] ------------------ 5686| 0| return 0; 5687| 0| } 5688| 7| return 1; 5689| 424| } else if (strcmp(b->name, "KMP_AFFINITY") == 0) { ------------------ | Branch (5689:14): [True: 1, False: 423] ------------------ 5690| 1| return -1; 5691| 1| } 5692| 423| return strcmp(a->name, b->name); 5693| 431|} // __kmp_stg_cmp kmp_settings.cpp:_ZL14__kmp_stg_findPKc: 5664| 83|static inline kmp_setting_t *__kmp_stg_find(char const *name) { 5665| | 5666| 83| int i; 5667| 83| if (name != NULL) { ------------------ | Branch (5667:7): [True: 83, False: 0] ------------------ 5668| 6.58k| for (i = 0; i < __kmp_stg_count; ++i) { ------------------ | Branch (5668:17): [True: 6.51k, False: 68] ------------------ 5669| 6.51k| if (strcmp(__kmp_stg_table[i].name, name) == 0) { ------------------ | Branch (5669:11): [True: 15, False: 6.50k] ------------------ 5670| 15| return &__kmp_stg_table[i]; 5671| 15| } 5672| 6.51k| } 5673| 83| } 5674| 68| return NULL; 5675| | 5676| 83|} // __kmp_stg_find kmp_settings.cpp:_ZL15__kmp_stg_parsePKcS0_: 5894| 35|static void __kmp_stg_parse(char const *name, char const *value) { 5895| | // On Windows* OS there are some nameless variables like "C:=C:\" (yeah, 5896| | // really nameless, they are presented in environment block as 5897| | // "=C:=C\\\x00=D:=D:\\\x00...", so let us skip them. 5898| 35| if (name[0] == 0) { ------------------ | Branch (5898:7): [True: 0, False: 35] ------------------ 5899| 0| return; 5900| 0| } 5901| | 5902| 35| if (value != NULL) { ------------------ | Branch (5902:7): [True: 34, False: 1] ------------------ 5903| 34| kmp_setting_t *setting = __kmp_stg_find(name); 5904| 34| if (setting != NULL) { ------------------ | Branch (5904:9): [True: 0, False: 34] ------------------ 5905| 0| setting->parse(name, value, setting->data); 5906| 0| setting->defined = 1; 5907| 0| } 5908| 34| } 5909| | 5910| 35|} // __kmp_stg_parse kmp_settings.cpp:_ZL25__kmp_stg_print_blocktimeP11kmp_str_bufPKcPv: 770| 1| void *data) { 771| 1| __kmp_stg_print_int(buffer, name, __kmp_dflt_blocktime); 772| 1|} // __kmp_stg_print_blocktime kmp_settings.cpp:_ZL19__kmp_stg_print_intP11kmp_str_bufPKci: 590| 24| int value) { 591| 24| if (__kmp_env_format) { ------------------ | Branch (591:7): [True: 0, False: 24] ------------------ 592| 0| KMP_STR_BUF_PRINT_INT; ------------------ | | 60| 0| __kmp_str_buf_print(buffer, " %s %s='%d'\n", KMP_I18N_STR(Host), name, value) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 593| 24| } else { 594| 24| __kmp_str_buf_print(buffer, " %s=%d\n", name, value); 595| 24| } 596| 24|} // __kmp_stg_print_int kmp_settings.cpp:_ZL25__kmp_stg_print_use_yieldP11kmp_str_bufPKcPv: 725| 1| void *data) { 726| 1| __kmp_stg_print_int(buffer, name, __kmp_use_yield); 727| 1|} // __kmp_stg_print_use_yield kmp_settings.cpp:_ZL32__kmp_stg_print_duplicate_lib_okP11kmp_str_bufPKcPv: 785| 1| char const *name, void *data) { 786| 1| __kmp_stg_print_bool(buffer, name, __kmp_duplicate_library_ok); 787| 1|} // __kmp_stg_print_duplicate_lib_ok kmp_settings.cpp:_ZL20__kmp_stg_print_boolP11kmp_str_bufPKci: 581| 17| int value) { 582| 17| if (__kmp_env_format) { ------------------ | Branch (582:7): [True: 0, False: 17] ------------------ 583| 0| KMP_STR_BUF_PRINT_BOOL; ------------------ | | 58| 0| KMP_STR_BUF_PRINT_BOOL_EX(name, value, "TRUE", "FALSE") | | ------------------ | | | | 55| 0| __kmp_str_buf_print(buffer, " %s %s='%s'\n", KMP_I18N_STR(Host), n, \ | | | | ------------------ | | | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | | | ------------------ | | | | 56| 0| (v) ? t : f) | | | | ------------------ | | | | | Branch (56:23): [True: 0, False: 0] | | | | ------------------ | | ------------------ ------------------ 584| 17| } else { 585| 17| __kmp_str_buf_print(buffer, " %s=%s\n", name, value ? "true" : "false"); ------------------ | Branch (585:53): [True: 5, False: 12] ------------------ 586| 17| } 587| 17|} // __kmp_stg_print_bool kmp_settings.cpp:_ZL27__kmp_stg_print_wait_policyP11kmp_str_bufPKcPv: 868| 2| void *data) { 869| | 870| 2| kmp_stg_wp_data_t *wait = (kmp_stg_wp_data_t *)data; 871| 2| char const *value = NULL; 872| | 873| 2| if (wait->omp) { ------------------ | Branch (873:7): [True: 1, False: 1] ------------------ 874| 1| switch (__kmp_library) { ------------------ | Branch (874:13): [True: 0, False: 1] ------------------ 875| 0| case library_turnaround: { ------------------ | Branch (875:5): [True: 0, False: 1] ------------------ 876| 0| value = "ACTIVE"; 877| 0| } break; 878| 0| case library_throughput: { ------------------ | Branch (878:5): [True: 0, False: 1] ------------------ 879| 0| value = "PASSIVE"; 880| 0| } break; 881| 1| } 882| 1| } else { 883| 1| switch (__kmp_library) { ------------------ | Branch (883:13): [True: 0, False: 1] ------------------ 884| 0| case library_serial: { ------------------ | Branch (884:5): [True: 0, False: 1] ------------------ 885| 0| value = "serial"; 886| 0| } break; 887| 0| case library_turnaround: { ------------------ | Branch (887:5): [True: 0, False: 1] ------------------ 888| 0| value = "turnaround"; 889| 0| } break; 890| 0| case library_throughput: { ------------------ | Branch (890:5): [True: 0, False: 1] ------------------ 891| 0| value = "throughput"; 892| 0| } break; 893| 1| } 894| 1| } 895| 2| if (value != NULL) { ------------------ | Branch (895:7): [True: 0, False: 2] ------------------ 896| 0| __kmp_stg_print_str(buffer, name, value); 897| 0| } 898| | 899| 2|} // __kmp_stg_print_wait_policy kmp_settings.cpp:_ZL19__kmp_stg_print_strP11kmp_str_bufPKcS2_: 608| 4| char const *value) { 609| 4| if (__kmp_env_format) { ------------------ | Branch (609:7): [True: 0, False: 4] ------------------ 610| 0| KMP_STR_BUF_PRINT_STR; ------------------ | | 65| 0| __kmp_str_buf_print(buffer, " %s %s='%s'\n", KMP_I18N_STR(Host), name, value) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 611| 4| } else { 612| 4| __kmp_str_buf_print(buffer, " %s=%s\n", name, value); 613| 4| } 614| 4|} // __kmp_stg_print_str kmp_settings.cpp:_ZL35__kmp_stg_print_device_thread_limitP11kmp_str_bufPKcPv: 659| 1| char const *name, void *data) { 660| 1| __kmp_stg_print_int(buffer, name, __kmp_max_nth); 661| 1|} // __kmp_stg_print_device_thread_limit kmp_settings.cpp:_ZL24__kmp_stg_print_settingsP11kmp_str_bufPKcPv: 941| 1| void *data) { 942| 1| __kmp_stg_print_bool(buffer, name, __kmp_settings); 943| 1|} // __kmp_stg_print_settings kmp_settings.cpp:_ZL27__kmp_stg_print_stackoffsetP11kmp_str_bufPKcPv: 978| 1| void *data) { 979| 1| __kmp_stg_print_size(buffer, name, __kmp_stkoffset); 980| 1|} // __kmp_stg_print_stackoffset kmp_settings.cpp:_ZL20__kmp_stg_print_sizeP11kmp_str_bufPKcm: 617| 3| size_t value) { 618| 3| if (__kmp_env_format) { ------------------ | Branch (618:7): [True: 0, False: 3] ------------------ 619| 0| KMP_STR_BUF_PRINT_NAME_EX(name); ------------------ | | 53| 0| __kmp_str_buf_print(buffer, " %s %s='", KMP_I18N_STR(Host), x) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 620| 0| __kmp_str_buf_print_size(buffer, value); 621| 0| __kmp_str_buf_print(buffer, "'\n"); 622| 3| } else { 623| 3| __kmp_str_buf_print(buffer, " %s=", name); 624| 3| __kmp_str_buf_print_size(buffer, value); 625| 3| __kmp_str_buf_print(buffer, "\n"); 626| 3| return; 627| 3| } 628| 3|} // __kmp_stg_print_size kmp_settings.cpp:_ZL25__kmp_stg_print_stacksizeP11kmp_str_bufPKcPv: 1010| 2| void *data) { 1011| 2| kmp_stg_ss_data_t *stacksize = (kmp_stg_ss_data_t *)data; 1012| 2| if (__kmp_env_format) { ------------------ | Branch (1012:7): [True: 0, False: 2] ------------------ 1013| 0| KMP_STR_BUF_PRINT_NAME_EX(name); ------------------ | | 53| 0| __kmp_str_buf_print(buffer, " %s %s='", KMP_I18N_STR(Host), x) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 1014| 0| __kmp_str_buf_print_size(buffer, (__kmp_stksize % 1024) ------------------ | Branch (1014:38): [True: 0, False: 0] ------------------ 1015| 0| ? __kmp_stksize / stacksize->factor 1016| 0| : __kmp_stksize); 1017| 0| __kmp_str_buf_print(buffer, "'\n"); 1018| 2| } else { 1019| 2| __kmp_str_buf_print(buffer, " %s=", name); 1020| 2| __kmp_str_buf_print_size(buffer, (__kmp_stksize % 1024) ------------------ | Branch (1020:38): [True: 0, False: 2] ------------------ 1021| 2| ? __kmp_stksize / stacksize->factor 1022| 2| : __kmp_stksize); 1023| 2| __kmp_str_buf_print(buffer, "\n"); 1024| 2| } 1025| 2|} // __kmp_stg_print_stacksize kmp_settings.cpp:_ZL24__kmp_stg_print_stackpadP11kmp_str_bufPKcPv: 959| 1| void *data) { 960| 1| __kmp_stg_print_int(buffer, name, __kmp_stkpadding); 961| 1|} // __kmp_stg_print_stackpad kmp_settings.cpp:_ZL23__kmp_stg_print_versionP11kmp_str_bufPKcPv: 1036| 1| void *data) { 1037| 1| __kmp_stg_print_bool(buffer, name, __kmp_version); 1038| 1|} // __kmp_stg_print_version kmp_settings.cpp:_ZL24__kmp_stg_print_warningsP11kmp_str_bufPKcPv: 1054| 1| void *data) { 1055| | // AC: TODO: change to print_int? (needs documentation change) 1056| 1| __kmp_stg_print_bool(buffer, name, __kmp_generate_warnings); 1057| 1|} // __kmp_stg_print_warnings kmp_settings.cpp:_ZL28__kmp_stg_print_nesting_modeP11kmp_str_bufPKcPv: 1072| 1| char const *name, void *data) { 1073| 1| if (__kmp_env_format) { ------------------ | Branch (1073:7): [True: 0, False: 1] ------------------ 1074| 0| KMP_STR_BUF_PRINT_NAME; ------------------ | | 51| 0| __kmp_str_buf_print(buffer, " %s %s", KMP_I18N_STR(Host), name) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 1075| 1| } else { 1076| 1| __kmp_str_buf_print(buffer, " %s", name); 1077| 1| } 1078| 1| __kmp_str_buf_print(buffer, "=%d\n", __kmp_nesting_mode); 1079| 1|} // __kmp_stg_print_nesting_mode kmp_settings.cpp:_ZL22__kmp_stg_print_nestedP11kmp_str_bufPKcPv: 1099| 1| void *data) { 1100| 1| if (__kmp_env_format) { ------------------ | Branch (1100:7): [True: 0, False: 1] ------------------ 1101| 0| KMP_STR_BUF_PRINT_NAME; ------------------ | | 51| 0| __kmp_str_buf_print(buffer, " %s %s", KMP_I18N_STR(Host), name) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 1102| 1| } else { 1103| 1| __kmp_str_buf_print(buffer, " %s", name); 1104| 1| } 1105| 1| __kmp_str_buf_print(buffer, ": deprecated; max-active-levels-var=%d\n", 1106| 1| __kmp_dflt_max_active_levels); 1107| 1|} // __kmp_stg_print_nested kmp_settings.cpp:_ZL27__kmp_stg_print_num_threadsP11kmp_str_bufPKcPv: 1328| 1| void *data) { 1329| 1| if (__kmp_env_format) { ------------------ | Branch (1329:7): [True: 0, False: 1] ------------------ 1330| 0| KMP_STR_BUF_PRINT_NAME; ------------------ | | 51| 0| __kmp_str_buf_print(buffer, " %s %s", KMP_I18N_STR(Host), name) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 1331| 1| } else { 1332| 1| __kmp_str_buf_print(buffer, " %s", name); 1333| 1| } 1334| 1| if (__kmp_nested_nth.used) { ------------------ | Branch (1334:7): [True: 0, False: 1] ------------------ 1335| 0| kmp_str_buf_t buf; 1336| 0| __kmp_str_buf_init(&buf); ------------------ | | 41| 0| { \ | | 42| 0| (b)->str = (b)->bulk; \ | | 43| 0| (b)->size = sizeof((b)->bulk); \ | | 44| 0| (b)->used = 0; \ | | 45| 0| (b)->bulk[0] = 0; \ | | 46| 0| } ------------------ 1337| 0| for (int i = 0; i < __kmp_nested_nth.used; i++) { ------------------ | Branch (1337:21): [True: 0, False: 0] ------------------ 1338| 0| __kmp_str_buf_print(&buf, "%d", __kmp_nested_nth.nth[i]); 1339| 0| if (i < __kmp_nested_nth.used - 1) { ------------------ | Branch (1339:11): [True: 0, False: 0] ------------------ 1340| 0| __kmp_str_buf_print(&buf, ","); 1341| 0| } 1342| 0| } 1343| 0| __kmp_str_buf_print(buffer, "='%s'\n", buf.str); 1344| 0| __kmp_str_buf_free(&buf); 1345| 1| } else { 1346| 1| __kmp_str_buf_print(buffer, ": %s\n", KMP_I18N_STR(NotDefined)); ------------------ | | 46| 1|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) ------------------ 1347| 1| } 1348| 1|} // __kmp_stg_print_num_threads kmp_settings.cpp:_ZL23__kmp_stg_print_taskingP11kmp_str_bufPKcPv: 1360| 1| void *data) { 1361| 1| __kmp_stg_print_int(buffer, name, __kmp_tasking_mode); 1362| 1|} // __kmp_stg_print_tasking kmp_settings.cpp:_ZL29__kmp_stg_print_task_stealingP11kmp_str_bufPKcPv: 1371| 1| char const *name, void *data) { 1372| 1| __kmp_stg_print_int(buffer, name, __kmp_task_stealing_constraint); 1373| 1|} // __kmp_stg_print_task_stealing kmp_settings.cpp:_ZL33__kmp_stg_print_max_active_levelsP11kmp_str_bufPKcPv: 1396| 1| char const *name, void *data) { 1397| 1| __kmp_stg_print_int(buffer, name, __kmp_dflt_max_active_levels); 1398| 1|} // __kmp_stg_print_max_active_levels kmp_settings.cpp:_ZL30__kmp_stg_print_default_deviceP11kmp_str_bufPKcPv: 1409| 1| char const *name, void *data) { 1410| 1| __kmp_stg_print_int(buffer, name, __kmp_default_device); 1411| 1|} // __kmp_stg_print_default_device kmp_settings.cpp:_ZL30__kmp_stg_print_target_offloadP11kmp_str_bufPKcPv: 1436| 1| char const *name, void *data) { 1437| 1| const char *value = NULL; 1438| 1| if (__kmp_target_offload == tgt_default) ------------------ | Branch (1438:7): [True: 1, False: 0] ------------------ 1439| 1| value = "DEFAULT"; 1440| 0| else if (__kmp_target_offload == tgt_mandatory) ------------------ | Branch (1440:12): [True: 0, False: 0] ------------------ 1441| 0| value = "MANDATORY"; 1442| 0| else if (__kmp_target_offload == tgt_disabled) ------------------ | Branch (1442:12): [True: 0, False: 0] ------------------ 1443| 0| value = "DISABLED"; 1444| 1| KMP_DEBUG_ASSERT(value); 1445| 1| if (__kmp_env_format) { ------------------ | Branch (1445:7): [True: 0, False: 1] ------------------ 1446| 0| KMP_STR_BUF_PRINT_NAME; ------------------ | | 51| 0| __kmp_str_buf_print(buffer, " %s %s", KMP_I18N_STR(Host), name) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 1447| 1| } else { 1448| 1| __kmp_str_buf_print(buffer, " %s", name); 1449| 1| } 1450| 1| __kmp_str_buf_print(buffer, "=%s\n", value); 1451| 1|} // __kmp_stg_print_target_offload kmp_settings.cpp:_ZL33__kmp_stg_print_max_task_priorityP11kmp_str_bufPKcPv: 1462| 1| char const *name, void *data) { 1463| 1| __kmp_stg_print_int(buffer, name, __kmp_max_task_priority); 1464| 1|} // __kmp_stg_print_max_task_priority kmp_settings.cpp:_ZL34__kmp_stg_print_taskloop_min_tasksP11kmp_str_bufPKcPv: 1476| 1| char const *name, void *data) { 1477| 1| __kmp_stg_print_uint64(buffer, name, __kmp_taskloop_min_tasks); 1478| 1|} // __kmp_stg_print_taskloop_min_tasks kmp_settings.cpp:_ZL22__kmp_stg_print_uint64P11kmp_str_bufPKcy: 599| 1| kmp_uint64 value) { 600| 1| if (__kmp_env_format) { ------------------ | Branch (600:7): [True: 0, False: 1] ------------------ 601| 0| KMP_STR_BUF_PRINT_UINT64; ------------------ | | 62| 0| __kmp_str_buf_print(buffer, " %s %s='%" KMP_UINT64_SPEC "'\n", \ | | 63| 0| KMP_I18N_STR(Host), name, value); | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 602| 1| } else { 603| 1| __kmp_str_buf_print(buffer, " %s=%" KMP_UINT64_SPEC "\n", name, value); 604| 1| } 605| 1|} // __kmp_stg_print_uint64 kmp_settings.cpp:_ZL28__kmp_stg_print_thread_limitP11kmp_str_bufPKcPv: 673| 1| char const *name, void *data) { 674| 1| __kmp_stg_print_int(buffer, name, __kmp_cg_max_nth); 675| 1|} // __kmp_stg_print_thread_limit kmp_settings.cpp:_ZL34__kmp_stg_print_teams_thread_limitP11kmp_str_bufPKcPv: 712| 1| char const *name, void *data) { 713| 1| __kmp_stg_print_int(buffer, name, __kmp_teams_max_nth); 714| 1|} // __kmp_stg_print_teams_thread_limit kmp_settings.cpp:_ZL22__kmp_stg_print_nteamsP11kmp_str_bufPKcPv: 686| 1| void *data) { 687| 1| __kmp_stg_print_int(buffer, name, __kmp_nteams); 688| 1|} // __kmp_stg_print_nteams kmp_settings.cpp:_ZL30__kmp_stg_print_teams_th_limitP11kmp_str_bufPKcPv: 700| 1| char const *name, void *data) { 701| 1| __kmp_stg_print_int(buffer, name, __kmp_teams_thread_limit); 702| 1|} // __kmp_stg_print_teams_th_limit kmp_settings.cpp:_ZL28__kmp_stg_print_disp_buffersP11kmp_str_bufPKcPv: 1493| 1| char const *name, void *data) { 1494| 1| __kmp_stg_print_int(buffer, name, __kmp_dispatch_num_buffers); 1495| 1|} // __kmp_stg_print_disp_buffers kmp_settings.cpp:_ZL31__kmp_stg_print_hot_teams_levelP11kmp_str_bufPKcPv: 1512| 1| char const *name, void *data) { 1513| 1| __kmp_stg_print_int(buffer, name, __kmp_hot_teams_max_level); 1514| 1|} // __kmp_stg_print_hot_teams_level kmp_settings.cpp:_ZL30__kmp_stg_print_hot_teams_modeP11kmp_str_bufPKcPv: 1527| 1| char const *name, void *data) { 1528| 1| __kmp_stg_print_int(buffer, name, __kmp_hot_teams_mode); 1529| 1|} // __kmp_stg_print_hot_teams_mode kmp_settings.cpp:_ZL30__kmp_stg_print_handle_signalsP11kmp_str_bufPKcPv: 1544| 1| char const *name, void *data) { 1545| 1| __kmp_stg_print_bool(buffer, name, __kmp_handle_signals); 1546| 1|} // __kmp_stg_print_handle_signals kmp_settings.cpp:_ZL34__kmp_stg_print_inherit_fp_controlP11kmp_str_bufPKcPv: 800| 1| char const *name, void *data) { 801| |#if KMP_DEBUG 802| | __kmp_stg_print_bool(buffer, name, __kmp_inherit_fp_control); 803| |#endif /* KMP_DEBUG */ 804| 1|} // __kmp_stg_print_inherit_fp_control kmp_settings.cpp:_ZL27__kmp_stg_print_align_allocP11kmp_str_bufPKcPv: 1677| 1| void *data) { 1678| 1| __kmp_stg_print_size(buffer, name, __kmp_align_alloc); 1679| 1|} // __kmp_stg_print_align_alloc kmp_settings.cpp:_ZL34__kmp_stg_print_barrier_branch_bitP11kmp_str_bufPKcPv: 1727| 3| char const *name, void *data) { 1728| 3| const char *var; 1729| 12| for (int i = bs_plain_barrier; i < bs_last_barrier; i++) { ------------------ | Branch (1729:34): [True: 9, False: 3] ------------------ 1730| 9| var = __kmp_barrier_branch_bit_env_name[i]; 1731| 9| if (strcmp(var, name) == 0) { ------------------ | Branch (1731:9): [True: 3, False: 6] ------------------ 1732| 3| if (__kmp_env_format) { ------------------ | Branch (1732:11): [True: 0, False: 3] ------------------ 1733| 0| KMP_STR_BUF_PRINT_NAME_EX(__kmp_barrier_branch_bit_env_name[i]); ------------------ | | 53| 0| __kmp_str_buf_print(buffer, " %s %s='", KMP_I18N_STR(Host), x) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 1734| 3| } else { 1735| 3| __kmp_str_buf_print(buffer, " %s='", 1736| 3| __kmp_barrier_branch_bit_env_name[i]); 1737| 3| } 1738| 3| __kmp_str_buf_print(buffer, "%d,%d'\n", 1739| 3| __kmp_barrier_gather_branch_bits[i], 1740| 3| __kmp_barrier_release_branch_bits[i]); 1741| 3| } 1742| 9| } 1743| 3|} // __kmp_stg_print_barrier_branch_bit kmp_settings.cpp:_ZL31__kmp_stg_print_barrier_patternP11kmp_str_bufPKcPv: 1825| 3| char const *name, void *data) { 1826| 3| const char *var; 1827| 12| for (int i = bs_plain_barrier; i < bs_last_barrier; i++) { ------------------ | Branch (1827:34): [True: 9, False: 3] ------------------ 1828| 9| var = __kmp_barrier_pattern_env_name[i]; 1829| 9| if (strcmp(var, name) == 0) { ------------------ | Branch (1829:9): [True: 3, False: 6] ------------------ 1830| 3| int j = __kmp_barrier_gather_pattern[i]; 1831| 3| int k = __kmp_barrier_release_pattern[i]; 1832| 3| if (__kmp_env_format) { ------------------ | Branch (1832:11): [True: 0, False: 3] ------------------ 1833| 0| KMP_STR_BUF_PRINT_NAME_EX(__kmp_barrier_pattern_env_name[i]); ------------------ | | 53| 0| __kmp_str_buf_print(buffer, " %s %s='", KMP_I18N_STR(Host), x) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 1834| 3| } else { 1835| 3| __kmp_str_buf_print(buffer, " %s='", 1836| 3| __kmp_barrier_pattern_env_name[i]); 1837| 3| } 1838| 3| KMP_DEBUG_ASSERT(j < bp_last_bar && k < bp_last_bar); 1839| 3| __kmp_str_buf_print(buffer, "%s,%s'\n", __kmp_barrier_pattern_name[j], 1840| 3| __kmp_barrier_pattern_name[k]); 1841| 3| } 1842| 9| } 1843| 3|} // __kmp_stg_print_barrier_pattern kmp_settings.cpp:_ZL27__kmp_stg_print_abort_delayP11kmp_str_bufPKcPv: 1858| 1| void *data) { 1859| 1| __kmp_stg_print_int(buffer, name, __kmp_abort_delay); 1860| 1|} // __kmp_stg_print_abort_delay kmp_settings.cpp:_ZL28__kmp_stg_print_cpuinfo_fileP11kmp_str_bufPKcPv: 1874| 1| char const *name, void *data) { 1875| 1|#if KMP_AFFINITY_SUPPORTED 1876| 1| if (__kmp_env_format) { ------------------ | Branch (1876:7): [True: 0, False: 1] ------------------ 1877| 0| KMP_STR_BUF_PRINT_NAME; ------------------ | | 51| 0| __kmp_str_buf_print(buffer, " %s %s", KMP_I18N_STR(Host), name) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 1878| 1| } else { 1879| 1| __kmp_str_buf_print(buffer, " %s", name); 1880| 1| } 1881| 1| if (__kmp_cpuinfo_file) { ------------------ | Branch (1881:7): [True: 0, False: 1] ------------------ 1882| 0| __kmp_str_buf_print(buffer, "='%s'\n", __kmp_cpuinfo_file); 1883| 1| } else { 1884| 1| __kmp_str_buf_print(buffer, ": %s\n", KMP_I18N_STR(NotDefined)); ------------------ | | 46| 1|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) ------------------ 1885| 1| } 1886| 1|#endif 1887| 1|} //__kmp_stg_print_cpuinfo_file kmp_settings.cpp:_ZL31__kmp_stg_print_force_reductionP11kmp_str_bufPKcPv: 1926| 2| char const *name, void *data) { 1927| | 1928| 2| kmp_stg_fr_data_t *reduction = (kmp_stg_fr_data_t *)data; 1929| 2| if (reduction->force) { ------------------ | Branch (1929:7): [True: 1, False: 1] ------------------ 1930| 1| if (__kmp_force_reduction_method == critical_reduce_block) { ------------------ | Branch (1930:9): [True: 0, False: 1] ------------------ 1931| 0| __kmp_stg_print_str(buffer, name, "critical"); 1932| 1| } else if (__kmp_force_reduction_method == atomic_reduce_block) { ------------------ | Branch (1932:16): [True: 0, False: 1] ------------------ 1933| 0| __kmp_stg_print_str(buffer, name, "atomic"); 1934| 1| } else if (__kmp_force_reduction_method == tree_reduce_block) { ------------------ | Branch (1934:16): [True: 0, False: 1] ------------------ 1935| 0| __kmp_stg_print_str(buffer, name, "tree"); 1936| 1| } else { 1937| 1| if (__kmp_env_format) { ------------------ | Branch (1937:11): [True: 0, False: 1] ------------------ 1938| 0| KMP_STR_BUF_PRINT_NAME; ------------------ | | 51| 0| __kmp_str_buf_print(buffer, " %s %s", KMP_I18N_STR(Host), name) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 1939| 1| } else { 1940| 1| __kmp_str_buf_print(buffer, " %s", name); 1941| 1| } 1942| 1| __kmp_str_buf_print(buffer, ": %s\n", KMP_I18N_STR(NotDefined)); ------------------ | | 46| 1|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) ------------------ 1943| 1| } 1944| 1| } else { 1945| 1| __kmp_stg_print_bool(buffer, name, __kmp_determ_red); 1946| 1| } 1947| | 1948| 2|} // __kmp_stg_print_force_reduction kmp_settings.cpp:_ZL27__kmp_stg_print_storage_mapP11kmp_str_bufPKcPv: 1967| 1| void *data) { 1968| 1| if (__kmp_storage_map_verbose || __kmp_storage_map_verbose_specified) { ------------------ | Branch (1968:7): [True: 0, False: 1] | Branch (1968:36): [True: 0, False: 1] ------------------ 1969| 0| __kmp_stg_print_str(buffer, name, "verbose"); 1970| 1| } else { 1971| 1| __kmp_stg_print_bool(buffer, name, __kmp_storage_map); 1972| 1| } 1973| 1|} // __kmp_stg_print_storage_map kmp_settings.cpp:_ZL33__kmp_stg_print_all_threadprivateP11kmp_str_bufPKcPv: 1986| 1| char const *name, void *data) { 1987| 1| __kmp_stg_print_int(buffer, name, __kmp_tp_capacity); 1988| 1|} kmp_settings.cpp:_ZL45__kmp_stg_print_foreign_threads_threadprivateP11kmp_str_bufPKcPv: 2001| 1| void *data) { 2002| 1| __kmp_stg_print_bool(buffer, name, __kmp_foreign_tp); 2003| 1|} // __kmp_stg_print_foreign_threads_threadprivate kmp_settings.cpp:_ZL24__kmp_stg_print_affinityP11kmp_str_bufPKcPv: 2675| 1| void *data) { 2676| 1| __kmp_print_affinity_env(buffer, name, __kmp_affinity); 2677| 1|} kmp_settings.cpp:_ZL24__kmp_print_affinity_envP11kmp_str_bufPKcRK14kmp_affinity_t: 2597| 2| const kmp_affinity_t &affinity) { 2598| 2| bool is_hh_affinity = (&affinity == &__kmp_hh_affinity); 2599| 2| if (__kmp_env_format) { ------------------ | Branch (2599:7): [True: 0, False: 2] ------------------ 2600| 0| KMP_STR_BUF_PRINT_NAME_EX(name); ------------------ | | 53| 0| __kmp_str_buf_print(buffer, " %s %s='", KMP_I18N_STR(Host), x) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 2601| 2| } else { 2602| 2| __kmp_str_buf_print(buffer, " %s='", name); 2603| 2| } 2604| 2| if (affinity.flags.verbose) { ------------------ | Branch (2604:7): [True: 0, False: 2] ------------------ 2605| 0| __kmp_str_buf_print(buffer, "%s,", "verbose"); 2606| 2| } else { 2607| 2| __kmp_str_buf_print(buffer, "%s,", "noverbose"); 2608| 2| } 2609| 2| if (affinity.flags.warnings) { ------------------ | Branch (2609:7): [True: 2, False: 0] ------------------ 2610| 2| __kmp_str_buf_print(buffer, "%s,", "warnings"); 2611| 2| } else { 2612| 0| __kmp_str_buf_print(buffer, "%s,", "nowarnings"); 2613| 0| } 2614| 2| if (KMP_AFFINITY_CAPABLE()) { ------------------ | | 684| 2|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) | | ------------------ | | | Branch (684:32): [True: 0, False: 2] | | ------------------ ------------------ 2615| | // Hidden helper affinity does not affect global reset 2616| | // or respect flags. That is still solely controlled by KMP_AFFINITY. 2617| 0| if (!is_hh_affinity) { ------------------ | Branch (2617:9): [True: 0, False: 0] ------------------ 2618| 0| if (affinity.flags.respect) { ------------------ | Branch (2618:11): [True: 0, False: 0] ------------------ 2619| 0| __kmp_str_buf_print(buffer, "%s,", "respect"); 2620| 0| } else { 2621| 0| __kmp_str_buf_print(buffer, "%s,", "norespect"); 2622| 0| } 2623| 0| if (affinity.flags.reset) { ------------------ | Branch (2623:11): [True: 0, False: 0] ------------------ 2624| 0| __kmp_str_buf_print(buffer, "%s,", "reset"); 2625| 0| } else { 2626| 0| __kmp_str_buf_print(buffer, "%s,", "noreset"); 2627| 0| } 2628| 0| } 2629| 0| __kmp_str_buf_print(buffer, "granularity=%s,", 2630| 0| __kmp_hw_get_keyword(affinity.gran, false)); 2631| 0| } 2632| 2| if (!KMP_AFFINITY_CAPABLE()) { ------------------ | | 684| 2|#define KMP_AFFINITY_CAPABLE() (__kmp_affin_mask_size > 0) ------------------ | Branch (2632:7): [True: 2, False: 0] ------------------ 2633| 2| __kmp_str_buf_print(buffer, "%s", "disabled"); 2634| 2| } else { 2635| 0| int compact = affinity.compact; 2636| 0| int offset = affinity.offset; 2637| 0| switch (affinity.type) { 2638| 0| case affinity_none: ------------------ | Branch (2638:5): [True: 0, False: 0] ------------------ 2639| 0| __kmp_str_buf_print(buffer, "%s", "none"); 2640| 0| break; 2641| 0| case affinity_physical: ------------------ | Branch (2641:5): [True: 0, False: 0] ------------------ 2642| 0| __kmp_str_buf_print(buffer, "%s,%d", "physical", offset); 2643| 0| break; 2644| 0| case affinity_logical: ------------------ | Branch (2644:5): [True: 0, False: 0] ------------------ 2645| 0| __kmp_str_buf_print(buffer, "%s,%d", "logical", offset); 2646| 0| break; 2647| 0| case affinity_compact: ------------------ | Branch (2647:5): [True: 0, False: 0] ------------------ 2648| 0| __kmp_str_buf_print(buffer, "%s,%d,%d", "compact", compact, offset); 2649| 0| break; 2650| 0| case affinity_scatter: ------------------ | Branch (2650:5): [True: 0, False: 0] ------------------ 2651| 0| __kmp_str_buf_print(buffer, "%s,%d,%d", "scatter", compact, offset); 2652| 0| break; 2653| 0| case affinity_explicit: ------------------ | Branch (2653:5): [True: 0, False: 0] ------------------ 2654| 0| __kmp_str_buf_print(buffer, "%s=[%s],%s", "proclist", affinity.proclist, 2655| 0| "explicit"); 2656| 0| break; 2657| 0| case affinity_balanced: ------------------ | Branch (2657:5): [True: 0, False: 0] ------------------ 2658| 0| __kmp_str_buf_print(buffer, "%s,%d,%d", "balanced", compact, offset); 2659| 0| break; 2660| 0| case affinity_disabled: ------------------ | Branch (2660:5): [True: 0, False: 0] ------------------ 2661| 0| __kmp_str_buf_print(buffer, "%s", "disabled"); 2662| 0| break; 2663| 0| case affinity_default: ------------------ | Branch (2663:5): [True: 0, False: 0] ------------------ 2664| 0| __kmp_str_buf_print(buffer, "%s", "default"); 2665| 0| break; 2666| 0| default: ------------------ | Branch (2666:5): [True: 0, False: 0] ------------------ 2667| 0| __kmp_str_buf_print(buffer, "%s", ""); 2668| 0| break; 2669| 0| } 2670| 0| } 2671| 2| __kmp_str_buf_print(buffer, "'\n"); 2672| 2|} //__kmp_stg_print_affinity kmp_settings.cpp:_ZL27__kmp_stg_print_hh_affinityP11kmp_str_bufPKcPv: 2679| 1| void *data) { 2680| 1| __kmp_print_affinity_env(buffer, name, __kmp_hh_affinity); 2681| 1|} kmp_settings.cpp:_ZL25__kmp_stg_print_proc_bindP11kmp_str_bufPKcPv: 3471| 1| void *data) { 3472| 1| int nelem = __kmp_nested_proc_bind.used; 3473| 1| if (__kmp_env_format) { ------------------ | Branch (3473:7): [True: 0, False: 1] ------------------ 3474| 0| KMP_STR_BUF_PRINT_NAME; ------------------ | | 51| 0| __kmp_str_buf_print(buffer, " %s %s", KMP_I18N_STR(Host), name) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 3475| 1| } else { 3476| 1| __kmp_str_buf_print(buffer, " %s", name); 3477| 1| } 3478| 1| if (nelem == 0) { ------------------ | Branch (3478:7): [True: 1, False: 0] ------------------ 3479| 1| __kmp_str_buf_print(buffer, ": %s\n", KMP_I18N_STR(NotDefined)); ------------------ | | 46| 1|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) ------------------ 3480| 1| } else { 3481| 0| int i; 3482| 0| __kmp_str_buf_print(buffer, "='", name); 3483| 0| for (i = 0; i < nelem; i++) { ------------------ | Branch (3483:17): [True: 0, False: 0] ------------------ 3484| 0| switch (__kmp_nested_proc_bind.bind_types[i]) { ------------------ | Branch (3484:15): [True: 0, False: 0] ------------------ 3485| 0| case proc_bind_false: ------------------ | Branch (3485:7): [True: 0, False: 0] ------------------ 3486| 0| __kmp_str_buf_print(buffer, "false"); 3487| 0| break; 3488| | 3489| 0| case proc_bind_true: ------------------ | Branch (3489:7): [True: 0, False: 0] ------------------ 3490| 0| __kmp_str_buf_print(buffer, "true"); 3491| 0| break; 3492| | 3493| 0| case proc_bind_primary: ------------------ | Branch (3493:7): [True: 0, False: 0] ------------------ 3494| 0| __kmp_str_buf_print(buffer, "primary"); 3495| 0| break; 3496| | 3497| 0| case proc_bind_close: ------------------ | Branch (3497:7): [True: 0, False: 0] ------------------ 3498| 0| __kmp_str_buf_print(buffer, "close"); 3499| 0| break; 3500| | 3501| 0| case proc_bind_spread: ------------------ | Branch (3501:7): [True: 0, False: 0] ------------------ 3502| 0| __kmp_str_buf_print(buffer, "spread"); 3503| 0| break; 3504| | 3505| 0| case proc_bind_intel: ------------------ | Branch (3505:7): [True: 0, False: 0] ------------------ 3506| 0| __kmp_str_buf_print(buffer, "intel"); 3507| 0| break; 3508| | 3509| 0| case proc_bind_default: ------------------ | Branch (3509:7): [True: 0, False: 0] ------------------ 3510| 0| __kmp_str_buf_print(buffer, "default"); 3511| 0| break; 3512| 0| } 3513| 0| if (i < nelem - 1) { ------------------ | Branch (3513:11): [True: 0, False: 0] ------------------ 3514| 0| __kmp_str_buf_print(buffer, ","); 3515| 0| } 3516| 0| } 3517| 0| __kmp_str_buf_print(buffer, "'\n"); 3518| 0| } 3519| 1|} kmp_settings.cpp:_ZL31__kmp_stg_print_teams_proc_bindP11kmp_str_bufPKcPv: 3321| 1| char const *name, void *data) { 3322| 1| const char *value = KMP_I18N_STR(NotDefined); ------------------ | | 46| 1|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) ------------------ 3323| 1| for (size_t i = 0; i < sizeof(proc_bind_table) / sizeof(proc_bind_table[0]); ------------------ | Branch (3323:22): [True: 1, False: 0] ------------------ 3324| 1| ++i) { 3325| 1| if (__kmp_teams_proc_bind == proc_bind_table[i].proc_bind) { ------------------ | Branch (3325:9): [True: 1, False: 0] ------------------ 3326| 1| value = proc_bind_table[i].name; 3327| 1| break; 3328| 1| } 3329| 1| } 3330| 1| __kmp_stg_print_str(buffer, name, value); 3331| 1|} kmp_settings.cpp:_ZL22__kmp_stg_print_placesP11kmp_str_bufPKcPv: 3108| 1| void *data) { 3109| 1| enum affinity_type type = __kmp_affinity.type; 3110| 1| const char *proclist = __kmp_affinity.proclist; 3111| 1| kmp_hw_t gran = __kmp_affinity.gran; 3112| | 3113| 1| if (__kmp_env_format) { ------------------ | Branch (3113:7): [True: 0, False: 1] ------------------ 3114| 0| KMP_STR_BUF_PRINT_NAME; ------------------ | | 51| 0| __kmp_str_buf_print(buffer, " %s %s", KMP_I18N_STR(Host), name) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 3115| 1| } else { 3116| 1| __kmp_str_buf_print(buffer, " %s", name); 3117| 1| } 3118| 1| if ((__kmp_nested_proc_bind.used == 0) || ------------------ | Branch (3118:7): [True: 1, False: 0] ------------------ 3119| 0| (__kmp_nested_proc_bind.bind_types == NULL) || ------------------ | Branch (3119:7): [True: 0, False: 0] ------------------ 3120| 1| (__kmp_nested_proc_bind.bind_types[0] == proc_bind_false)) { ------------------ | Branch (3120:7): [True: 0, False: 0] ------------------ 3121| 1| __kmp_str_buf_print(buffer, ": %s\n", KMP_I18N_STR(NotDefined)); ------------------ | | 46| 1|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) ------------------ 3122| 1| } else if (type == affinity_explicit) { ------------------ | Branch (3122:14): [True: 0, False: 0] ------------------ 3123| 0| if (proclist != NULL) { ------------------ | Branch (3123:9): [True: 0, False: 0] ------------------ 3124| 0| __kmp_str_buf_print(buffer, "='%s'\n", proclist); 3125| 0| } else { 3126| 0| __kmp_str_buf_print(buffer, ": %s\n", KMP_I18N_STR(NotDefined)); ------------------ | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) ------------------ 3127| 0| } 3128| 0| } else if (type == affinity_compact) { ------------------ | Branch (3128:14): [True: 0, False: 0] ------------------ 3129| 0| int num; 3130| 0| if (__kmp_affinity.num_masks > 0) { ------------------ | Branch (3130:9): [True: 0, False: 0] ------------------ 3131| 0| num = __kmp_affinity.num_masks; 3132| 0| } else if (__kmp_affinity_num_places > 0) { ------------------ | Branch (3132:16): [True: 0, False: 0] ------------------ 3133| 0| num = __kmp_affinity_num_places; 3134| 0| } else { 3135| 0| num = 0; 3136| 0| } 3137| 0| if (gran != KMP_HW_UNKNOWN) { ------------------ | Branch (3137:9): [True: 0, False: 0] ------------------ 3138| 0| const char *name = __kmp_hw_get_keyword(gran, true); 3139| 0| if (num > 0) { ------------------ | Branch (3139:11): [True: 0, False: 0] ------------------ 3140| 0| __kmp_str_buf_print(buffer, "='%s(%d)'\n", name, num); 3141| 0| } else { 3142| 0| __kmp_str_buf_print(buffer, "='%s'\n", name); 3143| 0| } 3144| 0| } else { 3145| 0| __kmp_str_buf_print(buffer, ": %s\n", KMP_I18N_STR(NotDefined)); ------------------ | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) ------------------ 3146| 0| } 3147| 0| } else { 3148| 0| __kmp_str_buf_print(buffer, ": %s\n", KMP_I18N_STR(NotDefined)); ------------------ | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) ------------------ 3149| 0| } 3150| 1|} kmp_settings.cpp:_ZL31__kmp_stg_print_topology_methodP11kmp_str_bufPKcPv: 3239| 1| char const *name, void *data) { 3240| 1| char const *value = NULL; 3241| | 3242| 1| switch (__kmp_affinity_top_method) { ------------------ | Branch (3242:11): [True: 1, False: 0] ------------------ 3243| 1| case affinity_top_method_default: ------------------ | Branch (3243:3): [True: 1, False: 0] ------------------ 3244| 1| value = "default"; 3245| 1| break; 3246| | 3247| 0| case affinity_top_method_all: ------------------ | Branch (3247:3): [True: 0, False: 1] ------------------ 3248| 0| value = "all"; 3249| 0| break; 3250| | 3251| 0|#if KMP_ARCH_X86 || KMP_ARCH_X86_64 3252| 0| case affinity_top_method_x2apicid_1f: ------------------ | Branch (3252:3): [True: 0, False: 1] ------------------ 3253| 0| value = "x2APIC id leaf 0x1f"; 3254| 0| break; 3255| | 3256| 0| case affinity_top_method_x2apicid: ------------------ | Branch (3256:3): [True: 0, False: 1] ------------------ 3257| 0| value = "x2APIC id leaf 0xb"; 3258| 0| break; 3259| | 3260| 0| case affinity_top_method_apicid: ------------------ | Branch (3260:3): [True: 0, False: 1] ------------------ 3261| 0| value = "APIC id"; 3262| 0| break; 3263| 0|#endif /* KMP_ARCH_X86 || KMP_ARCH_X86_64 */ 3264| | 3265| |#if KMP_USE_HWLOC 3266| | case affinity_top_method_hwloc: 3267| | value = "hwloc"; 3268| | break; 3269| |#endif 3270| | 3271| 0| case affinity_top_method_cpuinfo: ------------------ | Branch (3271:3): [True: 0, False: 1] ------------------ 3272| 0| value = "cpuinfo"; 3273| 0| break; 3274| | 3275| |#if KMP_GROUP_AFFINITY 3276| | case affinity_top_method_group: 3277| | value = "group"; 3278| | break; 3279| |#endif /* KMP_GROUP_AFFINITY */ 3280| | 3281| 0| case affinity_top_method_flat: ------------------ | Branch (3281:3): [True: 0, False: 1] ------------------ 3282| 0| value = "flat"; 3283| 0| break; 3284| 1| } 3285| | 3286| 1| if (value != NULL) { ------------------ | Branch (3286:7): [True: 1, False: 0] ------------------ 3287| 1| __kmp_stg_print_str(buffer, name, value); 3288| 1| } 3289| 1|} // __kmp_stg_print_topology_method kmp_settings.cpp:_ZL32__kmp_stg_print_display_affinityP11kmp_str_bufPKcPv: 3526| 1| char const *name, void *data) { 3527| 1| __kmp_stg_print_bool(buffer, name, __kmp_display_affinity); 3528| 1|} kmp_settings.cpp:_ZL31__kmp_stg_print_affinity_formatP11kmp_str_bufPKcPv: 3536| 1| char const *name, void *data) { 3537| 1| if (__kmp_env_format) { ------------------ | Branch (3537:7): [True: 0, False: 1] ------------------ 3538| 0| KMP_STR_BUF_PRINT_NAME_EX(name); ------------------ | | 53| 0| __kmp_str_buf_print(buffer, " %s %s='", KMP_I18N_STR(Host), x) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 3539| 1| } else { 3540| 1| __kmp_str_buf_print(buffer, " %s='", name); 3541| 1| } 3542| 1| __kmp_str_buf_print(buffer, "%s'\n", __kmp_affinity_format); 3543| 1|} kmp_settings.cpp:_ZL28__kmp_stg_print_init_at_forkP11kmp_str_bufPKcPv: 3999| 1| char const *name, void *data) { 4000| 1| __kmp_stg_print_bool(buffer, name, __kmp_need_register_atfork_specified); 4001| 1|} // __kmp_stg_print_init_at_fork kmp_settings.cpp:_ZL24__kmp_stg_print_scheduleP11kmp_str_bufPKcPv: 4060| 1| void *data) { 4061| 1| if (__kmp_env_format) { ------------------ | Branch (4061:7): [True: 0, False: 1] ------------------ 4062| 0| KMP_STR_BUF_PRINT_NAME_EX(name); ------------------ | | 53| 0| __kmp_str_buf_print(buffer, " %s %s='", KMP_I18N_STR(Host), x) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 4063| 1| } else { 4064| 1| __kmp_str_buf_print(buffer, " %s='", name); 4065| 1| } 4066| 1| if (__kmp_static == kmp_sch_static_greedy) { ------------------ | Branch (4066:7): [True: 1, False: 0] ------------------ 4067| 1| __kmp_str_buf_print(buffer, "%s", "static,greedy"); 4068| 1| } else if (__kmp_static == kmp_sch_static_balanced) { ------------------ | Branch (4068:14): [True: 0, False: 0] ------------------ 4069| 0| __kmp_str_buf_print(buffer, "%s", "static,balanced"); 4070| 0| } 4071| 1| if (__kmp_guided == kmp_sch_guided_iterative_chunked) { ------------------ | Branch (4071:7): [True: 1, False: 0] ------------------ 4072| 1| __kmp_str_buf_print(buffer, ";%s'\n", "guided,iterative"); 4073| 1| } else if (__kmp_guided == kmp_sch_guided_analytical_chunked) { ------------------ | Branch (4073:14): [True: 0, False: 0] ------------------ 4074| 0| __kmp_str_buf_print(buffer, ";%s'\n", "guided,analytical"); 4075| 0| } 4076| 1|} // __kmp_stg_print_schedule kmp_settings.cpp:_ZL28__kmp_stg_print_omp_scheduleP11kmp_str_bufPKcPv: 4272| 1| char const *name, void *data) { 4273| 1| if (__kmp_env_format) { ------------------ | Branch (4273:7): [True: 0, False: 1] ------------------ 4274| 0| KMP_STR_BUF_PRINT_NAME_EX(name); ------------------ | | 53| 0| __kmp_str_buf_print(buffer, " %s %s='", KMP_I18N_STR(Host), x) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 4275| 1| } else { 4276| 1| __kmp_str_buf_print(buffer, " %s='", name); 4277| 1| } 4278| 1| enum sched_type sched = SCHEDULE_WITHOUT_MODIFIERS(__kmp_sched); ------------------ | | 451| 1| (enum sched_type)( \ | | 452| 1| (s) & ~(kmp_sch_modifier_nonmonotonic | kmp_sch_modifier_monotonic)) ------------------ 4279| 1| if (SCHEDULE_HAS_MONOTONIC(__kmp_sched)) { ------------------ | | 453| 1|#define SCHEDULE_HAS_MONOTONIC(s) (((s)&kmp_sch_modifier_monotonic) != 0) | | ------------------ | | | Branch (453:35): [True: 0, False: 1] | | ------------------ ------------------ 4280| 0| __kmp_str_buf_print(buffer, "monotonic:"); 4281| 1| } else if (SCHEDULE_HAS_NONMONOTONIC(__kmp_sched)) { ------------------ | | 454| 1|#define SCHEDULE_HAS_NONMONOTONIC(s) (((s)&kmp_sch_modifier_nonmonotonic) != 0) | | ------------------ | | | Branch (454:38): [True: 0, False: 1] | | ------------------ ------------------ 4282| 0| __kmp_str_buf_print(buffer, "nonmonotonic:"); 4283| 0| } 4284| 1| if (__kmp_chunk) { ------------------ | Branch (4284:7): [True: 0, False: 1] ------------------ 4285| 0| switch (sched) { ------------------ | Branch (4285:13): [True: 0, False: 0] ------------------ 4286| 0| case kmp_sch_dynamic_chunked: ------------------ | Branch (4286:5): [True: 0, False: 0] ------------------ 4287| 0| __kmp_str_buf_print(buffer, "%s,%d'\n", "dynamic", __kmp_chunk); 4288| 0| break; 4289| 0| case kmp_sch_guided_iterative_chunked: ------------------ | Branch (4289:5): [True: 0, False: 0] ------------------ 4290| 0| case kmp_sch_guided_analytical_chunked: ------------------ | Branch (4290:5): [True: 0, False: 0] ------------------ 4291| 0| __kmp_str_buf_print(buffer, "%s,%d'\n", "guided", __kmp_chunk); 4292| 0| break; 4293| 0| case kmp_sch_trapezoidal: ------------------ | Branch (4293:5): [True: 0, False: 0] ------------------ 4294| 0| __kmp_str_buf_print(buffer, "%s,%d'\n", "trapezoidal", __kmp_chunk); 4295| 0| break; 4296| 0| case kmp_sch_static: ------------------ | Branch (4296:5): [True: 0, False: 0] ------------------ 4297| 0| case kmp_sch_static_chunked: ------------------ | Branch (4297:5): [True: 0, False: 0] ------------------ 4298| 0| case kmp_sch_static_balanced: ------------------ | Branch (4298:5): [True: 0, False: 0] ------------------ 4299| 0| case kmp_sch_static_greedy: ------------------ | Branch (4299:5): [True: 0, False: 0] ------------------ 4300| 0| __kmp_str_buf_print(buffer, "%s,%d'\n", "static", __kmp_chunk); 4301| 0| break; 4302| 0| case kmp_sch_static_steal: ------------------ | Branch (4302:5): [True: 0, False: 0] ------------------ 4303| 0| __kmp_str_buf_print(buffer, "%s,%d'\n", "static_steal", __kmp_chunk); 4304| 0| break; 4305| 0| case kmp_sch_auto: ------------------ | Branch (4305:5): [True: 0, False: 0] ------------------ 4306| 0| __kmp_str_buf_print(buffer, "%s,%d'\n", "auto", __kmp_chunk); 4307| 0| break; 4308| 0| } 4309| 1| } else { 4310| 1| switch (sched) { ------------------ | Branch (4310:13): [True: 1, False: 0] ------------------ 4311| 0| case kmp_sch_dynamic_chunked: ------------------ | Branch (4311:5): [True: 0, False: 1] ------------------ 4312| 0| __kmp_str_buf_print(buffer, "%s'\n", "dynamic"); 4313| 0| break; 4314| 0| case kmp_sch_guided_iterative_chunked: ------------------ | Branch (4314:5): [True: 0, False: 1] ------------------ 4315| 0| case kmp_sch_guided_analytical_chunked: ------------------ | Branch (4315:5): [True: 0, False: 1] ------------------ 4316| 0| __kmp_str_buf_print(buffer, "%s'\n", "guided"); 4317| 0| break; 4318| 0| case kmp_sch_trapezoidal: ------------------ | Branch (4318:5): [True: 0, False: 1] ------------------ 4319| 0| __kmp_str_buf_print(buffer, "%s'\n", "trapezoidal"); 4320| 0| break; 4321| 1| case kmp_sch_static: ------------------ | Branch (4321:5): [True: 1, False: 0] ------------------ 4322| 1| case kmp_sch_static_chunked: ------------------ | Branch (4322:5): [True: 0, False: 1] ------------------ 4323| 1| case kmp_sch_static_balanced: ------------------ | Branch (4323:5): [True: 0, False: 1] ------------------ 4324| 1| case kmp_sch_static_greedy: ------------------ | Branch (4324:5): [True: 0, False: 1] ------------------ 4325| 1| __kmp_str_buf_print(buffer, "%s'\n", "static"); 4326| 1| break; 4327| 0| case kmp_sch_static_steal: ------------------ | Branch (4327:5): [True: 0, False: 1] ------------------ 4328| 0| __kmp_str_buf_print(buffer, "%s'\n", "static_steal"); 4329| 0| break; 4330| 0| case kmp_sch_auto: ------------------ | Branch (4330:5): [True: 0, False: 1] ------------------ 4331| 0| __kmp_str_buf_print(buffer, "%s'\n", "auto"); 4332| 0| break; 4333| 1| } 4334| 1| } 4335| 1|} // __kmp_stg_print_omp_schedule kmp_settings.cpp:_ZL35__kmp_stg_print_kmp_force_monotonicP11kmp_str_bufPKcPv: 4359| 1| char const *name, void *data) { 4360| 1| __kmp_stg_print_bool(buffer, name, __kmp_force_monotonic); 4361| 1|} // __kmp_stg_print_kmp_force_monotonic kmp_settings.cpp:_ZL27__kmp_stg_print_atomic_modeP11kmp_str_bufPKcPv: 4385| 1| void *data) { 4386| 1| __kmp_stg_print_int(buffer, name, __kmp_atomic_mode); 4387| 1|} // __kmp_stg_print_atomic_mode kmp_settings.cpp:_ZL33__kmp_stg_print_consistency_checkP11kmp_str_bufPKcPv: 4409| 1| char const *name, void *data) { 4410| |#if KMP_DEBUG 4411| | const char *value = NULL; 4412| | 4413| | if (__kmp_env_consistency_check) { 4414| | value = "all"; 4415| | } else { 4416| | value = "none"; 4417| | } 4418| | 4419| | if (value != NULL) { 4420| | __kmp_stg_print_str(buffer, name, value); 4421| | } 4422| |#endif /* KMP_DEBUG */ 4423| 1|} // __kmp_stg_print_consistency_check kmp_settings.cpp:_ZL32__kmp_stg_print_malloc_pool_incrP11kmp_str_bufPKcPv: 4460| 1| char const *name, void *data) { 4461| 1| __kmp_stg_print_size(buffer, name, __kmp_malloc_pool_incr); 4462| | 4463| 1|} // _kmp_stg_print_malloc_pool_incr kmp_settings.cpp:_ZL25__kmp_stg_print_gtid_modeP11kmp_str_bufPKcPv: 4514| 1| void *data) { 4515| 1| if (__kmp_adjust_gtid_mode) { ------------------ | Branch (4515:7): [True: 0, False: 1] ------------------ 4516| 0| __kmp_stg_print_int(buffer, name, 0); 4517| 1| } else { 4518| 1| __kmp_stg_print_int(buffer, name, __kmp_gtid_mode); 4519| 1| } 4520| 1|} // __kmp_stg_print_gtid_mode kmp_settings.cpp:_ZL27__kmp_stg_print_omp_dynamicP11kmp_str_bufPKcPv: 3908| 1| void *data) { 3909| 1| __kmp_stg_print_bool(buffer, name, __kmp_global.g.g_dynamic); 3910| 1|} // __kmp_stg_print_omp_dynamic kmp_settings.cpp:_ZL32__kmp_stg_print_kmp_dynamic_modeP11kmp_str_bufPKcPv: 3942| 1| char const *name, void *data) { 3943| |#if KMP_DEBUG 3944| | if (__kmp_global.g.g_dynamic_mode == dynamic_default) { 3945| | __kmp_str_buf_print(buffer, " %s: %s \n", name, KMP_I18N_STR(NotDefined)); 3946| | } 3947| |#ifdef USE_LOAD_BALANCE 3948| | else if (__kmp_global.g.g_dynamic_mode == dynamic_load_balance) { 3949| | __kmp_stg_print_str(buffer, name, "load balance"); 3950| | } 3951| |#endif /* USE_LOAD_BALANCE */ 3952| | else if (__kmp_global.g.g_dynamic_mode == dynamic_thread_limit) { 3953| | __kmp_stg_print_str(buffer, name, "thread limit"); 3954| | } else if (__kmp_global.g.g_dynamic_mode == dynamic_random) { 3955| | __kmp_stg_print_str(buffer, name, "random"); 3956| | } else { 3957| | KMP_ASSERT(0); 3958| | } 3959| |#endif /* KMP_DEBUG */ 3960| 1|} // __kmp_stg_print_kmp_dynamic_mode kmp_settings.cpp:_ZL35__kmp_stg_print_ld_balance_intervalP11kmp_str_bufPKcPv: 3978| 1| char const *name, void *data) { 3979| |#if KMP_DEBUG 3980| | __kmp_str_buf_print(buffer, " %s=%8.6f\n", name, 3981| | __kmp_load_balance_interval); 3982| |#endif /* KMP_DEBUG */ 3983| 1|} // __kmp_stg_print_load_balance_interval kmp_settings.cpp:_ZL26__kmp_stg_print_lock_blockP11kmp_str_bufPKcPv: 4531| 1| void *data) { 4532| 1| __kmp_stg_print_int(buffer, name, __kmp_num_locks_in_block); 4533| 1|} // __kmp_stg_print_lock_block kmp_settings.cpp:_ZL25__kmp_stg_print_lock_kindP11kmp_str_bufPKcPv: 4632| 1| void *data) { 4633| 1| const char *value = NULL; 4634| | 4635| 1| switch (__kmp_user_lock_kind) { ------------------ | Branch (4635:11): [True: 1, False: 0] ------------------ 4636| 1| case lk_default: ------------------ | Branch (4636:3): [True: 1, False: 0] ------------------ 4637| 1| value = "default"; 4638| 1| break; 4639| | 4640| 0| case lk_tas: ------------------ | Branch (4640:3): [True: 0, False: 1] ------------------ 4641| 0| value = "tas"; 4642| 0| break; 4643| | 4644| 0|#if KMP_USE_FUTEX 4645| 0| case lk_futex: ------------------ | Branch (4645:3): [True: 0, False: 1] ------------------ 4646| 0| value = "futex"; 4647| 0| break; 4648| 0|#endif 4649| | 4650| 0|#if KMP_USE_DYNAMIC_LOCK && KMP_USE_TSX 4651| 0| case lk_rtm_queuing: ------------------ | Branch (4651:3): [True: 0, False: 1] ------------------ 4652| 0| value = "rtm_queuing"; 4653| 0| break; 4654| | 4655| 0| case lk_rtm_spin: ------------------ | Branch (4655:3): [True: 0, False: 1] ------------------ 4656| 0| value = "rtm_spin"; 4657| 0| break; 4658| | 4659| 0| case lk_hle: ------------------ | Branch (4659:3): [True: 0, False: 1] ------------------ 4660| 0| value = "hle"; 4661| 0| break; 4662| 0|#endif 4663| | 4664| 0| case lk_ticket: ------------------ | Branch (4664:3): [True: 0, False: 1] ------------------ 4665| 0| value = "ticket"; 4666| 0| break; 4667| | 4668| 0| case lk_queuing: ------------------ | Branch (4668:3): [True: 0, False: 1] ------------------ 4669| 0| value = "queuing"; 4670| 0| break; 4671| | 4672| 0| case lk_drdpa: ------------------ | Branch (4672:3): [True: 0, False: 1] ------------------ 4673| 0| value = "drdpa"; 4674| 0| break; 4675| 0|#if KMP_USE_ADAPTIVE_LOCKS 4676| 0| case lk_adaptive: ------------------ | Branch (4676:3): [True: 0, False: 1] ------------------ 4677| 0| value = "adaptive"; 4678| 0| break; 4679| 1|#endif 4680| 1| } 4681| | 4682| 1| if (value != NULL) { ------------------ | Branch (4682:7): [True: 1, False: 0] ------------------ 4683| 1| __kmp_stg_print_str(buffer, name, value); 4684| 1| } 4685| 1|} kmp_settings.cpp:_ZL35__kmp_stg_print_spin_backoff_paramsP11kmp_str_bufPKcPv: 4773| 1| char const *name, void *data) { 4774| 1| if (__kmp_env_format) { ------------------ | Branch (4774:7): [True: 0, False: 1] ------------------ 4775| 0| KMP_STR_BUF_PRINT_NAME_EX(name); ------------------ | | 53| 0| __kmp_str_buf_print(buffer, " %s %s='", KMP_I18N_STR(Host), x) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 4776| 1| } else { 4777| 1| __kmp_str_buf_print(buffer, " %s='", name); 4778| 1| } 4779| 1| __kmp_str_buf_print(buffer, "%d,%d'\n", __kmp_spin_backoff_params.max_backoff, 4780| 1| __kmp_spin_backoff_params.min_tick); 4781| 1|} kmp_settings.cpp:_ZL35__kmp_stg_print_adaptive_lock_propsP11kmp_str_bufPKcPv: 4872| 1| char const *name, void *data) { 4873| 1| if (__kmp_env_format) { ------------------ | Branch (4873:7): [True: 0, False: 1] ------------------ 4874| 0| KMP_STR_BUF_PRINT_NAME_EX(name); ------------------ | | 53| 0| __kmp_str_buf_print(buffer, " %s %s='", KMP_I18N_STR(Host), x) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 4875| 1| } else { 4876| 1| __kmp_str_buf_print(buffer, " %s='", name); 4877| 1| } 4878| 1| __kmp_str_buf_print(buffer, "%d,%d'\n", 4879| 1| __kmp_adaptive_backoff_params.max_soft_retries, 4880| 1| __kmp_adaptive_backoff_params.max_badness); 4881| 1|} // __kmp_stg_print_adaptive_lock_props kmp_settings.cpp:_ZL25__kmp_stg_print_hw_subsetP11kmp_str_bufPKcPv: 5158| 2| void *data) { 5159| 2| kmp_str_buf_t buf; 5160| 2| int depth; 5161| 2| if (!__kmp_hw_subset) ------------------ | Branch (5161:7): [True: 2, False: 0] ------------------ 5162| 2| return; 5163| 0| __kmp_str_buf_init(&buf); ------------------ | | 41| 0| { \ | | 42| 0| (b)->str = (b)->bulk; \ | | 43| 0| (b)->size = sizeof((b)->bulk); \ | | 44| 0| (b)->used = 0; \ | | 45| 0| (b)->bulk[0] = 0; \ | | 46| 0| } ------------------ 5164| 0| if (__kmp_env_format) ------------------ | Branch (5164:7): [True: 0, False: 0] ------------------ 5165| 0| KMP_STR_BUF_PRINT_NAME_EX(name); ------------------ | | 53| 0| __kmp_str_buf_print(buffer, " %s %s='", KMP_I18N_STR(Host), x) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 5166| 0| else 5167| 0| __kmp_str_buf_print(buffer, " %s='", name); 5168| | 5169| 0| depth = __kmp_hw_subset->get_depth(); 5170| 0| for (int i = 0; i < depth; ++i) { ------------------ | Branch (5170:19): [True: 0, False: 0] ------------------ 5171| 0| const auto &item = __kmp_hw_subset->at(i); 5172| 0| if (i > 0) ------------------ | Branch (5172:9): [True: 0, False: 0] ------------------ 5173| 0| __kmp_str_buf_print(&buf, "%c", ','); 5174| 0| for (int j = 0; j < item.num_attrs; ++j) { ------------------ | Branch (5174:21): [True: 0, False: 0] ------------------ 5175| 0| __kmp_str_buf_print(&buf, "%s%d%s", (j > 0 ? "&" : ""), item.num[j], ------------------ | Branch (5175:44): [True: 0, False: 0] ------------------ 5176| 0| __kmp_hw_get_keyword(item.type)); 5177| 0| if (item.attr[j].is_core_type_valid()) ------------------ | Branch (5177:11): [True: 0, False: 0] ------------------ 5178| 0| __kmp_str_buf_print( 5179| 0| &buf, ":%s", 5180| 0| __kmp_hw_get_core_type_keyword(item.attr[j].get_core_type())); 5181| 0| if (item.attr[j].is_core_eff_valid()) ------------------ | Branch (5181:11): [True: 0, False: 0] ------------------ 5182| 0| __kmp_str_buf_print(&buf, ":eff%d", item.attr[j].get_core_eff()); 5183| 0| if (item.offset[j]) ------------------ | Branch (5183:11): [True: 0, False: 0] ------------------ 5184| 0| __kmp_str_buf_print(&buf, "@%d", item.offset[j]); 5185| 0| } 5186| 0| } 5187| 0| __kmp_str_buf_print(buffer, "%s'\n", buf.str); 5188| 0| __kmp_str_buf_free(&buf); 5189| 0|} kmp_settings.cpp:_ZL31__kmp_stg_print_task_throttlingP11kmp_str_bufPKcPv: 5229| 1| char const *name, void *data) { 5230| 1| __kmp_stg_print_bool(buffer, name, __kmp_enable_task_throttling); 5231| 1|} // __kmp_stg_print_task_throttling kmp_settings.cpp:_ZL31__kmp_stg_print_omp_display_envP11kmp_str_bufPKcPv: 5296| 1| char const *name, void *data) { 5297| 1| if (__kmp_display_env_verbose) { ------------------ | Branch (5297:7): [True: 0, False: 1] ------------------ 5298| 0| __kmp_stg_print_str(buffer, name, "VERBOSE"); 5299| 1| } else { 5300| 1| __kmp_stg_print_bool(buffer, name, __kmp_display_env); 5301| 1| } 5302| 1|} // __kmp_stg_print_omp_display_env kmp_settings.cpp:_ZL32__kmp_stg_print_omp_cancellationP11kmp_str_bufPKcPv: 5314| 1| char const *name, void *data) { 5315| 1| __kmp_stg_print_bool(buffer, name, __kmp_omp_cancellation); 5316| 1|} // __kmp_stg_print_omp_cancellation kmp_settings.cpp:_ZL25__kmp_stg_print_allocatorP11kmp_str_bufPKcPv: 3879| 1| void *data) { 3880| 1| if (__kmp_def_allocator == omp_default_mem_alloc) { ------------------ | Branch (3880:7): [True: 1, False: 0] ------------------ 3881| 1| __kmp_stg_print_str(buffer, name, "omp_default_mem_alloc"); 3882| 1| } else if (__kmp_def_allocator == omp_high_bw_mem_alloc) { ------------------ | Branch (3882:14): [True: 0, False: 0] ------------------ 3883| 0| __kmp_stg_print_str(buffer, name, "omp_high_bw_mem_alloc"); 3884| 0| } else if (__kmp_def_allocator == omp_large_cap_mem_alloc) { ------------------ | Branch (3884:14): [True: 0, False: 0] ------------------ 3885| 0| __kmp_stg_print_str(buffer, name, "omp_large_cap_mem_alloc"); 3886| 0| } else if (__kmp_def_allocator == omp_const_mem_alloc) { ------------------ | Branch (3886:14): [True: 0, False: 0] ------------------ 3887| 0| __kmp_stg_print_str(buffer, name, "omp_const_mem_alloc"); 3888| 0| } else if (__kmp_def_allocator == omp_low_lat_mem_alloc) { ------------------ | Branch (3888:14): [True: 0, False: 0] ------------------ 3889| 0| __kmp_stg_print_str(buffer, name, "omp_low_lat_mem_alloc"); 3890| 0| } else if (__kmp_def_allocator == omp_cgroup_mem_alloc) { ------------------ | Branch (3890:14): [True: 0, False: 0] ------------------ 3891| 0| __kmp_stg_print_str(buffer, name, "omp_cgroup_mem_alloc"); 3892| 0| } else if (__kmp_def_allocator == omp_pteam_mem_alloc) { ------------------ | Branch (3892:14): [True: 0, False: 0] ------------------ 3893| 0| __kmp_stg_print_str(buffer, name, "omp_pteam_mem_alloc"); 3894| 0| } else if (__kmp_def_allocator == omp_thread_mem_alloc) { ------------------ | Branch (3894:14): [True: 0, False: 0] ------------------ 3895| 0| __kmp_stg_print_str(buffer, name, "omp_thread_mem_alloc"); 3896| 0| } 3897| 1|} kmp_settings.cpp:_ZL33__kmp_stg_print_use_hidden_helperP11kmp_str_bufPKcPv: 1323| 1| char const *name, void *data) { 1324| 1| __kmp_stg_print_bool(buffer, name, __kmp_enable_hidden_helper); 1325| 1|} // __kmp_stg_print_use_hidden_helper kmp_settings.cpp:_ZL41__kmp_stg_print_num_hidden_helper_threadsP11kmp_str_bufPKcPv: 1300| 1| void *data) { 1301| 1| if (__kmp_hidden_helper_threads_num == 0) { ------------------ | Branch (1301:7): [True: 0, False: 1] ------------------ 1302| 0| __kmp_stg_print_int(buffer, name, __kmp_hidden_helper_threads_num); 1303| 1| } else { 1304| 1| KMP_DEBUG_ASSERT(__kmp_hidden_helper_threads_num > 1); 1305| | // Let's exclude the main thread of hidden helper team and print 1306| | // number of worker threads those do actual job. 1307| 1| __kmp_stg_print_int(buffer, name, __kmp_hidden_helper_threads_num - 1); 1308| 1| } 1309| 1|} // __kmp_stg_print_num_hidden_helper_threads kmp_settings.cpp:_ZL24__kmp_stg_print_omp_toolP11kmp_str_bufPKcPv: 5327| 1| void *data) { 5328| 1| if (__kmp_env_format) { ------------------ | Branch (5328:7): [True: 0, False: 1] ------------------ 5329| 0| KMP_STR_BUF_PRINT_BOOL_EX(name, __kmp_tool, "enabled", "disabled"); ------------------ | | 55| 0| __kmp_str_buf_print(buffer, " %s %s='%s'\n", KMP_I18N_STR(Host), n, \ | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ | | 56| 0| (v) ? t : f) | | ------------------ | | | Branch (56:23): [True: 0, False: 0] | | ------------------ ------------------ 5330| 1| } else { 5331| 1| __kmp_str_buf_print(buffer, " %s=%s\n", name, 5332| 1| __kmp_tool ? "enabled" : "disabled"); ------------------ | Branch (5332:25): [True: 1, False: 0] ------------------ 5333| 1| } 5334| 1|} // __kmp_stg_print_omp_tool kmp_settings.cpp:_ZL34__kmp_stg_print_omp_tool_librariesP11kmp_str_bufPKcPv: 5344| 1| char const *name, void *data) { 5345| 1| if (__kmp_tool_libraries) ------------------ | Branch (5345:7): [True: 0, False: 1] ------------------ 5346| 0| __kmp_stg_print_str(buffer, name, __kmp_tool_libraries); 5347| 1| else { 5348| 1| if (__kmp_env_format) { ------------------ | Branch (5348:9): [True: 0, False: 1] ------------------ 5349| 0| KMP_STR_BUF_PRINT_NAME; ------------------ | | 51| 0| __kmp_str_buf_print(buffer, " %s %s", KMP_I18N_STR(Host), name) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 5350| 1| } else { 5351| 1| __kmp_str_buf_print(buffer, " %s", name); 5352| 1| } 5353| 1| __kmp_str_buf_print(buffer, ": %s\n", KMP_I18N_STR(NotDefined)); ------------------ | | 46| 1|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) ------------------ 5354| 1| } 5355| 1|} // __kmp_stg_print_omp_tool_libraries kmp_settings.cpp:_ZL37__kmp_stg_print_omp_tool_verbose_initP11kmp_str_bufPKcPv: 5367| 1| void *data) { 5368| 1| if (__kmp_tool_verbose_init) ------------------ | Branch (5368:7): [True: 0, False: 1] ------------------ 5369| 0| __kmp_stg_print_str(buffer, name, __kmp_tool_verbose_init); 5370| 1| else { 5371| 1| if (__kmp_env_format) { ------------------ | Branch (5371:9): [True: 0, False: 1] ------------------ 5372| 0| KMP_STR_BUF_PRINT_NAME; ------------------ | | 51| 0| __kmp_str_buf_print(buffer, " %s %s", KMP_I18N_STR(Host), name) | | ------------------ | | | | 46| 0|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) | | ------------------ ------------------ 5373| 1| } else { 5374| 1| __kmp_str_buf_print(buffer, " %s", name); 5375| 1| } 5376| 1| __kmp_str_buf_print(buffer, ": %s\n", KMP_I18N_STR(NotDefined)); ------------------ | | 46| 1|#define KMP_I18N_STR(id) __kmp_i18n_catgets(kmp_i18n_str_##id) ------------------ 5377| 1| } 5378| 1|} // __kmp_stg_print_omp_tool_verbose_init kmp_settings.cpp:_ZL32__kmp_stg_print_user_level_mwaitP11kmp_str_bufPKcPv: 5243| 1| char const *name, void *data) { 5244| 1| __kmp_stg_print_bool(buffer, name, __kmp_user_level_mwait); 5245| 1|} // __kmp_stg_print_user_level_mwait kmp_settings.cpp:_ZL27__kmp_stg_print_mwait_hintsP11kmp_str_bufPKcPv: 5256| 1| void *data) { 5257| 1| __kmp_stg_print_int(buffer, name, __kmp_mwait_hints); 5258| 1|} // __kmp_stg_print_mwait_hints kmp_settings.cpp:_ZL22__kmp_stg_print_tpauseP11kmp_str_bufPKcPv: 5278| 1| void *data) { 5279| 1| __kmp_stg_print_int(buffer, name, __kmp_tpause_state); 5280| 1|} // __kmp_stg_print_tpause __kmp_str_buf_clear: 71| 83|void __kmp_str_buf_clear(kmp_str_buf_t *buffer) { 72| 83| KMP_STR_BUF_INVARIANT(buffer); ------------------ | | 60| 83| { \ | | 61| 83| KMP_DEBUG_ASSERT((b)->str != NULL); \ | | 62| 83| KMP_DEBUG_ASSERT((b)->size >= sizeof((b)->bulk)); \ | | 63| 83| KMP_DEBUG_ASSERT((b)->size % sizeof((b)->bulk) == 0); \ | | 64| 83| KMP_DEBUG_ASSERT((unsigned)(b)->used < (b)->size); \ | | 65| 83| KMP_DEBUG_ASSERT( \ | | 66| 83| (b)->size == sizeof((b)->bulk) ? (b)->str == &(b)->bulk[0] : 1); \ | | 67| 83| KMP_DEBUG_ASSERT((b)->size > sizeof((b)->bulk) ? (b)->str != &(b)->bulk[0] \ | | 68| 83| : 1); \ | | 69| 83| } ------------------ 73| 83| if (buffer->used > 0) { ------------------ | Branch (73:7): [True: 74, False: 9] ------------------ 74| 74| buffer->used = 0; 75| 74| buffer->str[0] = 0; 76| 74| } 77| 83| KMP_STR_BUF_INVARIANT(buffer); ------------------ | | 60| 83| { \ | | 61| 83| KMP_DEBUG_ASSERT((b)->str != NULL); \ | | 62| 83| KMP_DEBUG_ASSERT((b)->size >= sizeof((b)->bulk)); \ | | 63| 83| KMP_DEBUG_ASSERT((b)->size % sizeof((b)->bulk) == 0); \ | | 64| 83| KMP_DEBUG_ASSERT((unsigned)(b)->used < (b)->size); \ | | 65| 83| KMP_DEBUG_ASSERT( \ | | 66| 83| (b)->size == sizeof((b)->bulk) ? (b)->str == &(b)->bulk[0] : 1); \ | | 67| 83| KMP_DEBUG_ASSERT((b)->size > sizeof((b)->bulk) ? (b)->str != &(b)->bulk[0] \ | | 68| 83| : 1); \ | | 69| 83| } ------------------ 78| 83|} // __kmp_str_buf_clear __kmp_str_buf_reserve: 80| 68|void __kmp_str_buf_reserve(kmp_str_buf_t *buffer, size_t size) { 81| 68| KMP_STR_BUF_INVARIANT(buffer); ------------------ | | 60| 68| { \ | | 61| 68| KMP_DEBUG_ASSERT((b)->str != NULL); \ | | 62| 68| KMP_DEBUG_ASSERT((b)->size >= sizeof((b)->bulk)); \ | | 63| 68| KMP_DEBUG_ASSERT((b)->size % sizeof((b)->bulk) == 0); \ | | 64| 68| KMP_DEBUG_ASSERT((unsigned)(b)->used < (b)->size); \ | | 65| 68| KMP_DEBUG_ASSERT( \ | | 66| 68| (b)->size == sizeof((b)->bulk) ? (b)->str == &(b)->bulk[0] : 1); \ | | 67| 68| KMP_DEBUG_ASSERT((b)->size > sizeof((b)->bulk) ? (b)->str != &(b)->bulk[0] \ | | 68| 68| : 1); \ | | 69| 68| } ------------------ 82| 68| KMP_DEBUG_ASSERT(size >= 0); 83| | 84| 68| if (buffer->size < (unsigned int)size) { ------------------ | Branch (84:7): [True: 2, False: 66] ------------------ 85| | // Calculate buffer size. 86| 2| do { 87| 2| buffer->size *= 2; 88| 2| } while (buffer->size < (unsigned int)size); ------------------ | Branch (88:14): [True: 0, False: 2] ------------------ 89| | 90| | // Enlarge buffer. 91| 2| if (buffer->str == &buffer->bulk[0]) { ------------------ | Branch (91:9): [True: 1, False: 1] ------------------ 92| 1| buffer->str = (char *)KMP_INTERNAL_MALLOC(buffer->size); ------------------ | | 110| 1|#define KMP_INTERNAL_MALLOC(sz) malloc(sz) ------------------ 93| 1| if (buffer->str == NULL) { ------------------ | Branch (93:11): [True: 0, False: 1] ------------------ 94| 0| KMP_FATAL(MemoryAllocFailed); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 95| 0| } 96| 1| KMP_MEMCPY_S(buffer->str, buffer->size, buffer->bulk, buffer->used + 1); ------------------ | | 60| 1|#define KMP_MEMCPY_S(dst, bsz, src, cnt) memcpy(dst, src, cnt) ------------------ 97| 1| } else { 98| 1| buffer->str = (char *)KMP_INTERNAL_REALLOC(buffer->str, buffer->size); ------------------ | | 112| 1|#define KMP_INTERNAL_REALLOC(p, sz) realloc((p), (sz)) ------------------ 99| 1| if (buffer->str == NULL) { ------------------ | Branch (99:11): [True: 0, False: 1] ------------------ 100| 0| KMP_FATAL(MemoryAllocFailed); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 101| 0| } 102| 1| } 103| 2| } 104| | 105| 68| KMP_DEBUG_ASSERT(buffer->size > 0); 106| 68| KMP_DEBUG_ASSERT(buffer->size >= (unsigned)size); 107| 68| KMP_STR_BUF_INVARIANT(buffer); ------------------ | | 60| 68| { \ | | 61| 68| KMP_DEBUG_ASSERT((b)->str != NULL); \ | | 62| 68| KMP_DEBUG_ASSERT((b)->size >= sizeof((b)->bulk)); \ | | 63| 68| KMP_DEBUG_ASSERT((b)->size % sizeof((b)->bulk) == 0); \ | | 64| 68| KMP_DEBUG_ASSERT((unsigned)(b)->used < (b)->size); \ | | 65| 68| KMP_DEBUG_ASSERT( \ | | 66| 68| (b)->size == sizeof((b)->bulk) ? (b)->str == &(b)->bulk[0] : 1); \ | | 67| 68| KMP_DEBUG_ASSERT((b)->size > sizeof((b)->bulk) ? (b)->str != &(b)->bulk[0] \ | | 68| 68| : 1); \ | | 69| 68| } ------------------ 108| 68|} // __kmp_str_buf_reserve __kmp_str_buf_detach: 110| 1|void __kmp_str_buf_detach(kmp_str_buf_t *buffer) { 111| 1| KMP_STR_BUF_INVARIANT(buffer); ------------------ | | 60| 1| { \ | | 61| 1| KMP_DEBUG_ASSERT((b)->str != NULL); \ | | 62| 1| KMP_DEBUG_ASSERT((b)->size >= sizeof((b)->bulk)); \ | | 63| 1| KMP_DEBUG_ASSERT((b)->size % sizeof((b)->bulk) == 0); \ | | 64| 1| KMP_DEBUG_ASSERT((unsigned)(b)->used < (b)->size); \ | | 65| 1| KMP_DEBUG_ASSERT( \ | | 66| 1| (b)->size == sizeof((b)->bulk) ? (b)->str == &(b)->bulk[0] : 1); \ | | 67| 1| KMP_DEBUG_ASSERT((b)->size > sizeof((b)->bulk) ? (b)->str != &(b)->bulk[0] \ | | 68| 1| : 1); \ | | 69| 1| } ------------------ 112| | 113| | // If internal bulk is used, allocate memory and copy it. 114| 1| if (buffer->size <= sizeof(buffer->bulk)) { ------------------ | Branch (114:7): [True: 1, False: 0] ------------------ 115| 1| buffer->str = (char *)KMP_INTERNAL_MALLOC(buffer->size); ------------------ | | 110| 1|#define KMP_INTERNAL_MALLOC(sz) malloc(sz) ------------------ 116| 1| if (buffer->str == NULL) { ------------------ | Branch (116:9): [True: 0, False: 1] ------------------ 117| 0| KMP_FATAL(MemoryAllocFailed); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 118| 0| } 119| 1| KMP_MEMCPY_S(buffer->str, buffer->size, buffer->bulk, buffer->used + 1); ------------------ | | 60| 1|#define KMP_MEMCPY_S(dst, bsz, src, cnt) memcpy(dst, src, cnt) ------------------ 120| 1| } 121| 1|} // __kmp_str_buf_detach __kmp_str_buf_free: 123| 3|void __kmp_str_buf_free(kmp_str_buf_t *buffer) { 124| 3| KMP_STR_BUF_INVARIANT(buffer); ------------------ | | 60| 3| { \ | | 61| 3| KMP_DEBUG_ASSERT((b)->str != NULL); \ | | 62| 3| KMP_DEBUG_ASSERT((b)->size >= sizeof((b)->bulk)); \ | | 63| 3| KMP_DEBUG_ASSERT((b)->size % sizeof((b)->bulk) == 0); \ | | 64| 3| KMP_DEBUG_ASSERT((unsigned)(b)->used < (b)->size); \ | | 65| 3| KMP_DEBUG_ASSERT( \ | | 66| 3| (b)->size == sizeof((b)->bulk) ? (b)->str == &(b)->bulk[0] : 1); \ | | 67| 3| KMP_DEBUG_ASSERT((b)->size > sizeof((b)->bulk) ? (b)->str != &(b)->bulk[0] \ | | 68| 3| : 1); \ | | 69| 3| } ------------------ 125| 3| if (buffer->size > sizeof(buffer->bulk)) { ------------------ | Branch (125:7): [True: 1, False: 2] ------------------ 126| 1| KMP_INTERNAL_FREE(buffer->str); ------------------ | | 111| 1|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 127| 1| } 128| 3| buffer->str = buffer->bulk; 129| 3| buffer->size = sizeof(buffer->bulk); 130| 3| buffer->used = 0; 131| 3| KMP_STR_BUF_INVARIANT(buffer); ------------------ | | 60| 3| { \ | | 61| 3| KMP_DEBUG_ASSERT((b)->str != NULL); \ | | 62| 3| KMP_DEBUG_ASSERT((b)->size >= sizeof((b)->bulk)); \ | | 63| 3| KMP_DEBUG_ASSERT((b)->size % sizeof((b)->bulk) == 0); \ | | 64| 3| KMP_DEBUG_ASSERT((unsigned)(b)->used < (b)->size); \ | | 65| 3| KMP_DEBUG_ASSERT( \ | | 66| 3| (b)->size == sizeof((b)->bulk) ? (b)->str == &(b)->bulk[0] : 1); \ | | 67| 3| KMP_DEBUG_ASSERT((b)->size > sizeof((b)->bulk) ? (b)->str != &(b)->bulk[0] \ | | 68| 3| : 1); \ | | 69| 3| } ------------------ 132| 3|} // __kmp_str_buf_free __kmp_str_buf_cat: 134| 68|void __kmp_str_buf_cat(kmp_str_buf_t *buffer, char const *str, size_t len) { 135| 68| KMP_STR_BUF_INVARIANT(buffer); ------------------ | | 60| 68| { \ | | 61| 68| KMP_DEBUG_ASSERT((b)->str != NULL); \ | | 62| 68| KMP_DEBUG_ASSERT((b)->size >= sizeof((b)->bulk)); \ | | 63| 68| KMP_DEBUG_ASSERT((b)->size % sizeof((b)->bulk) == 0); \ | | 64| 68| KMP_DEBUG_ASSERT((unsigned)(b)->used < (b)->size); \ | | 65| 68| KMP_DEBUG_ASSERT( \ | | 66| 68| (b)->size == sizeof((b)->bulk) ? (b)->str == &(b)->bulk[0] : 1); \ | | 67| 68| KMP_DEBUG_ASSERT((b)->size > sizeof((b)->bulk) ? (b)->str != &(b)->bulk[0] \ | | 68| 68| : 1); \ | | 69| 68| } ------------------ 136| 68| KMP_DEBUG_ASSERT(str != NULL); 137| 68| KMP_DEBUG_ASSERT(len >= 0); 138| | 139| 68| __kmp_str_buf_reserve(buffer, buffer->used + len + 1); 140| 68| buffer->str[buffer->used] = '\0'; 141| 68| KMP_STRNCAT_S(buffer->str + buffer->used, len + 1, str, len); ------------------ | | 65| 68|#define KMP_STRNCAT_S(dst, bsz, src, cnt) strncat(dst, src, cnt) ------------------ 142| 68| __kmp_type_convert(buffer->used + len, &(buffer->used)); 143| 68| KMP_STR_BUF_INVARIANT(buffer); ------------------ | | 60| 68| { \ | | 61| 68| KMP_DEBUG_ASSERT((b)->str != NULL); \ | | 62| 68| KMP_DEBUG_ASSERT((b)->size >= sizeof((b)->bulk)); \ | | 63| 68| KMP_DEBUG_ASSERT((b)->size % sizeof((b)->bulk) == 0); \ | | 64| 68| KMP_DEBUG_ASSERT((unsigned)(b)->used < (b)->size); \ | | 65| 68| KMP_DEBUG_ASSERT( \ | | 66| 68| (b)->size == sizeof((b)->bulk) ? (b)->str == &(b)->bulk[0] : 1); \ | | 67| 68| KMP_DEBUG_ASSERT((b)->size > sizeof((b)->bulk) ? (b)->str != &(b)->bulk[0] \ | | 68| 68| : 1); \ | | 69| 68| } ------------------ 144| 68|} // __kmp_str_buf_cat __kmp_str_buf_vprint: 162| 124| va_list args) { 163| 124| int rc; 164| 124| KMP_STR_BUF_INVARIANT(buffer); ------------------ | | 60| 124| { \ | | 61| 124| KMP_DEBUG_ASSERT((b)->str != NULL); \ | | 62| 124| KMP_DEBUG_ASSERT((b)->size >= sizeof((b)->bulk)); \ | | 63| 124| KMP_DEBUG_ASSERT((b)->size % sizeof((b)->bulk) == 0); \ | | 64| 124| KMP_DEBUG_ASSERT((unsigned)(b)->used < (b)->size); \ | | 65| 124| KMP_DEBUG_ASSERT( \ | | 66| 124| (b)->size == sizeof((b)->bulk) ? (b)->str == &(b)->bulk[0] : 1); \ | | 67| 124| KMP_DEBUG_ASSERT((b)->size > sizeof((b)->bulk) ? (b)->str != &(b)->bulk[0] \ | | 68| 124| : 1); \ | | 69| 124| } ------------------ 165| | 166| 124| for (;;) { 167| 124| int const free = buffer->size - buffer->used; 168| 124| int size; 169| | 170| | // Try to format string. 171| 124| { 172| | /* On Linux* OS Intel(R) 64, vsnprintf() modifies args argument, so 173| | vsnprintf() crashes if it is called for the second time with the same 174| | args. To prevent the crash, we have to pass a fresh intact copy of args 175| | to vsnprintf() on each iteration. 176| | 177| | Unfortunately, standard va_copy() macro is not available on Windows* 178| | OS. However, it seems vsnprintf() does not modify args argument on 179| | Windows* OS. 180| | */ 181| | 182| 124|#if !KMP_OS_WINDOWS 183| 124| va_list _args; 184| 124| va_copy(_args, args); // Make copy of args. 185| 124|#define args _args // Substitute args with its copy, _args. 186| 124|#endif // KMP_OS_WINDOWS 187| 124| rc = KMP_VSNPRINTF(buffer->str + buffer->used, free, format, args); ------------------ | | 66| 124|#define KMP_VSNPRINTF vsnprintf ------------------ rc = KMP_VSNPRINTF(buffer->str + buffer->used, free, format, args); ------------------ | | 185| 124|#define args _args // Substitute args with its copy, _args. ------------------ 188| 124|#if !KMP_OS_WINDOWS 189| 124|#undef args // Remove substitution. 190| 124| va_end(_args); 191| 124|#endif // KMP_OS_WINDOWS 192| 124| } 193| | 194| | // No errors, string has been formatted. 195| 124| if (rc >= 0 && rc < free) { ------------------ | Branch (195:9): [True: 124, False: 0] | Branch (195:20): [True: 124, False: 0] ------------------ 196| 124| buffer->used += rc; 197| 124| break; 198| 124| } 199| | 200| | // Error occurred, buffer is too small. 201| 0| if (rc >= 0) { ------------------ | Branch (201:9): [True: 0, False: 0] ------------------ 202| | // C99-conforming implementation of vsnprintf returns required buffer size 203| 0| size = buffer->used + rc + 1; 204| 0| } else { 205| | // Older implementations just return -1. Double buffer size. 206| 0| size = buffer->size * 2; 207| 0| } 208| | 209| | // Enlarge buffer. 210| 0| __kmp_str_buf_reserve(buffer, size); 211| | 212| | // And try again. 213| 0| } 214| | 215| 124| KMP_DEBUG_ASSERT(buffer->size > 0); 216| 124| KMP_STR_BUF_INVARIANT(buffer); ------------------ | | 60| 124| { \ | | 61| 124| KMP_DEBUG_ASSERT((b)->str != NULL); \ | | 62| 124| KMP_DEBUG_ASSERT((b)->size >= sizeof((b)->bulk)); \ | | 63| 124| KMP_DEBUG_ASSERT((b)->size % sizeof((b)->bulk) == 0); \ | | 64| 124| KMP_DEBUG_ASSERT((unsigned)(b)->used < (b)->size); \ | | 65| 124| KMP_DEBUG_ASSERT( \ | | 66| 124| (b)->size == sizeof((b)->bulk) ? (b)->str == &(b)->bulk[0] : 1); \ | | 67| 124| KMP_DEBUG_ASSERT((b)->size > sizeof((b)->bulk) ? (b)->str != &(b)->bulk[0] \ | | 68| 124| : 1); \ | | 69| 124| } ------------------ 217| 124| return rc; 218| 124|} // __kmp_str_buf_vprint __kmp_str_buf_print: 221| 123|int __kmp_str_buf_print(kmp_str_buf_t *buffer, char const *format, ...) { 222| 123| int rc; 223| 123| va_list args; 224| 123| va_start(args, format); 225| 123| rc = __kmp_str_buf_vprint(buffer, format, args); 226| | va_end(args); 227| 123| return rc; 228| 123|} // __kmp_str_buf_print __kmp_str_buf_print_size: 232| 5|void __kmp_str_buf_print_size(kmp_str_buf_t *buf, size_t size) { 233| 5| char const *names[] = {"", "k", "M", "G", "T", "P", "E", "Z", "Y"}; 234| 5| int const units = sizeof(names) / sizeof(char const *); 235| 5| int u = 0; 236| 5| if (size > 0) { ------------------ | Branch (236:7): [True: 5, False: 0] ------------------ 237| 11| while ((size % 1024 == 0) && (u + 1 < units)) { ------------------ | Branch (237:12): [True: 6, False: 5] | Branch (237:34): [True: 6, False: 0] ------------------ 238| 6| size = size / 1024; 239| 6| ++u; 240| 6| } 241| 5| } 242| | 243| 5| __kmp_str_buf_print(buf, "%" KMP_SIZE_T_SPEC "%s", size, names[u]); 244| 5|} // __kmp_str_buf_print_size __kmp_str_format: 451| 3|) { 452| 3| va_list args; 453| 3| int size = 512; 454| 3| char *buffer = NULL; 455| 3| int rc; 456| | 457| | // Allocate buffer. 458| 3| buffer = (char *)KMP_INTERNAL_MALLOC(size); ------------------ | | 110| 3|#define KMP_INTERNAL_MALLOC(sz) malloc(sz) ------------------ 459| 3| if (buffer == NULL) { ------------------ | Branch (459:7): [True: 0, False: 3] ------------------ 460| 0| KMP_FATAL(MemoryAllocFailed); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 461| 0| } 462| | 463| 3| for (;;) { 464| | // Try to format string. 465| 3| va_start(args, format); 466| 3| rc = KMP_VSNPRINTF(buffer, size, format, args); ------------------ | | 66| 3|#define KMP_VSNPRINTF vsnprintf ------------------ 467| 3| va_end(args); 468| | 469| | // No errors, string has been formatted. 470| 3| if (rc >= 0 && rc < size) { ------------------ | Branch (470:9): [True: 3, False: 0] | Branch (470:20): [True: 3, False: 0] ------------------ 471| 3| break; 472| 3| } 473| | 474| | // Error occurred, buffer is too small. 475| 0| if (rc >= 0) { ------------------ | Branch (475:9): [True: 0, False: 0] ------------------ 476| | // C99-conforming implementation of vsnprintf returns required buffer 477| | // size. 478| 0| size = rc + 1; 479| 0| } else { 480| | // Older implementations just return -1. 481| 0| size = size * 2; 482| 0| } 483| | 484| | // Enlarge buffer and try again. 485| 0| buffer = (char *)KMP_INTERNAL_REALLOC(buffer, size); ------------------ | | 112| 0|#define KMP_INTERNAL_REALLOC(p, sz) realloc((p), (sz)) ------------------ 486| 0| if (buffer == NULL) { ------------------ | Branch (486:9): [True: 0, False: 0] ------------------ 487| 0| KMP_FATAL(MemoryAllocFailed); ------------------ | | 146| 0|#define KMP_FATAL(...) __kmp_fatal(KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 488| 0| } 489| 0| } 490| | 491| 3| return buffer; 492| 3|} // func __kmp_str_format __kmp_str_free: 494| 3|void __kmp_str_free(char **str) { 495| 3| KMP_DEBUG_ASSERT(str != NULL); 496| 3| KMP_INTERNAL_FREE(*str); ------------------ | | 111| 3|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 497| | *str = NULL; 498| 3|} // func __kmp_str_free __kmp_str_split: 575| 67|) { 576| 67| char *h = str; 577| 67| char *t = NULL; 578| 67| if (str != NULL) { ------------------ | Branch (578:7): [True: 67, False: 0] ------------------ 579| 67| char *ptr = strchr(str, delim); 580| 67| if (ptr != NULL) { ------------------ | Branch (580:9): [True: 67, False: 0] ------------------ 581| 67| *ptr = 0; 582| 67| t = ptr + 1; 583| 67| } 584| 67| } 585| 67| if (head != NULL) { ------------------ | Branch (585:7): [True: 67, False: 0] ------------------ 586| 67| *head = h; 587| 67| } 588| 67| if (tail != NULL) { ------------------ | Branch (588:7): [True: 67, False: 0] ------------------ 589| 67| *tail = t; 590| 67| } 591| 67|} // __kmp_str_split __kmp_pop_current_task_from_thread: 612| 4.36k|void __kmp_pop_current_task_from_thread(kmp_info_t *this_thr) { 613| 4.36k| KF_TRACE(10, ("__kmp_pop_current_task_from_thread(enter): T#%d " 614| 4.36k| "this_thread=%p, curtask=%p, " 615| 4.36k| "curtask_parent=%p\n", 616| 4.36k| 0, this_thr, this_thr->th.th_current_task, 617| 4.36k| this_thr->th.th_current_task->td_parent)); 618| | 619| 4.36k| this_thr->th.th_current_task = this_thr->th.th_current_task->td_parent; 620| | 621| 4.36k| KF_TRACE(10, ("__kmp_pop_current_task_from_thread(exit): T#%d " 622| 4.36k| "this_thread=%p, curtask=%p, " 623| 4.36k| "curtask_parent=%p\n", 624| 4.36k| 0, this_thr, this_thr->th.th_current_task, 625| 4.36k| this_thr->th.th_current_task->td_parent)); 626| 4.36k|} __kmp_push_current_task_to_thread: 635| 4.42k| int tid) { 636| | // current task of the thread is a parent of the new just created implicit 637| | // tasks of new team 638| 4.42k| KF_TRACE(10, ("__kmp_push_current_task_to_thread(enter): T#%d this_thread=%p " 639| 4.42k| "curtask=%p " 640| 4.42k| "parent_task=%p\n", 641| 4.42k| tid, this_thr, this_thr->th.th_current_task, 642| 4.42k| team->t.t_implicit_task_taskdata[tid].td_parent)); 643| | 644| 4.42k| KMP_DEBUG_ASSERT(this_thr != NULL); 645| | 646| 4.42k| if (tid == 0) { ------------------ | Branch (646:7): [True: 4.36k, False: 62] ------------------ 647| 4.36k| if (this_thr->th.th_current_task != &team->t.t_implicit_task_taskdata[0]) { ------------------ | Branch (647:9): [True: 4.36k, False: 0] ------------------ 648| 4.36k| team->t.t_implicit_task_taskdata[0].td_parent = 649| 4.36k| this_thr->th.th_current_task; 650| 4.36k| this_thr->th.th_current_task = &team->t.t_implicit_task_taskdata[0]; 651| 4.36k| } 652| 4.36k| } else { 653| 62| team->t.t_implicit_task_taskdata[tid].td_parent = 654| 62| team->t.t_implicit_task_taskdata[0].td_parent; 655| 62| this_thr->th.th_current_task = &team->t.t_implicit_task_taskdata[tid]; 656| 62| } 657| | 658| 4.42k| KF_TRACE(10, ("__kmp_push_current_task_to_thread(exit): T#%d this_thread=%p " 659| 4.42k| "curtask=%p " 660| 4.42k| "parent_task=%p\n", 661| 4.42k| tid, this_thr, this_thr->th.th_current_task, 662| 4.42k| team->t.t_implicit_task_taskdata[tid].td_parent)); 663| 4.42k|} __kmp_init_implicit_task: 1320| 42.5k| kmp_team_t *team, int tid, int set_curr_task) { 1321| 42.5k| kmp_taskdata_t *task = &team->t.t_implicit_task_taskdata[tid]; 1322| | 1323| 42.5k| KF_TRACE( 1324| 42.5k| 10, 1325| 42.5k| ("__kmp_init_implicit_task(enter): T#:%d team=%p task=%p, reinit=%s\n", 1326| 42.5k| tid, team, task, set_curr_task ? "TRUE" : "FALSE")); 1327| | 1328| 42.5k| task->td_task_id = KMP_GEN_TASK_ID(); ------------------ | | 3546| 42.5k|#define KMP_GEN_TASK_ID() _KMP_GEN_ID(__kmp_task_counter) | | ------------------ | | | | 3543| 42.5k|#define _KMP_GEN_ID(counter) (~0) | | ------------------ ------------------ 1329| 42.5k| task->td_team = team; 1330| | // task->td_parent = NULL; // fix for CQ230101 (broken parent task info 1331| | // in debugger) 1332| 42.5k| task->td_ident = loc_ref; 1333| 42.5k| task->td_taskwait_ident = NULL; 1334| 42.5k| task->td_taskwait_counter = 0; 1335| 42.5k| task->td_taskwait_thread = 0; 1336| | 1337| 42.5k| task->td_flags.tiedness = TASK_TIED; ------------------ | | 45| 42.5k|#define TASK_TIED 1 ------------------ 1338| 42.5k| task->td_flags.tasktype = TASK_IMPLICIT; ------------------ | | 48| 42.5k|#define TASK_IMPLICIT 0 ------------------ 1339| 42.5k| task->td_flags.proxy = TASK_FULL; ------------------ | | 50| 42.5k|#define TASK_FULL 0 ------------------ 1340| | 1341| | // All implicit tasks are executed immediately, not deferred 1342| 42.5k| task->td_flags.task_serial = 1; 1343| 42.5k| task->td_flags.tasking_ser = (__kmp_tasking_mode == tskm_immediate_exec); 1344| 42.5k| task->td_flags.team_serial = (team->t.t_serialized) ? 1 : 0; ------------------ | Branch (1344:32): [True: 66, False: 42.5k] ------------------ 1345| | 1346| 42.5k| task->td_flags.started = 1; 1347| 42.5k| task->td_flags.executing = 1; 1348| 42.5k| task->td_flags.complete = 0; 1349| 42.5k| task->td_flags.freed = 0; 1350| |#if OMPX_TASKGRAPH 1351| | task->td_flags.onced = 0; 1352| |#endif 1353| | 1354| 42.5k| task->td_depnode = NULL; 1355| 42.5k| task->td_last_tied = task; 1356| 42.5k| task->td_allow_completion_event.type = KMP_EVENT_UNINITIALIZED; 1357| | 1358| 42.5k| if (set_curr_task) { // only do this init first time thread is created ------------------ | Branch (1358:7): [True: 64, False: 42.5k] ------------------ 1359| 64| KMP_ATOMIC_ST_REL(&task->td_incomplete_child_tasks, 0); ------------------ | | 1251| 64|#define KMP_ATOMIC_ST_REL(p, v) KMP_ATOMIC_OP(store, p, v, release) | | ------------------ | | | | 1246| 64|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 1360| | // Not used: don't need to deallocate implicit task 1361| 64| KMP_ATOMIC_ST_REL(&task->td_allocated_child_tasks, 0); ------------------ | | 1251| 64|#define KMP_ATOMIC_ST_REL(p, v) KMP_ATOMIC_OP(store, p, v, release) | | ------------------ | | | | 1246| 64|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 1362| 64| task->td_taskgroup = NULL; // An implicit task does not have taskgroup 1363| 64| task->td_dephash = NULL; 1364| 64| __kmp_push_current_task_to_thread(this_thr, team, tid); 1365| 42.5k| } else { 1366| 42.5k| KMP_DEBUG_ASSERT(task->td_incomplete_child_tasks == 0); 1367| 42.5k| KMP_DEBUG_ASSERT(task->td_allocated_child_tasks == 0); 1368| 42.5k| } 1369| | 1370| 42.5k|#if OMPT_SUPPORT 1371| 42.5k| if (UNLIKELY(ompt_enabled.enabled)) ------------------ | | 119| 42.5k|#define UNLIKELY(x) __builtin_expect(!!(x), 0) | | ------------------ | | | Branch (119:21): [True: 0, False: 42.5k] | | ------------------ ------------------ 1372| 0| __ompt_task_init(task, tid); 1373| 42.5k|#endif 1374| | 1375| 42.5k| KF_TRACE(10, ("__kmp_init_implicit_task(exit): T#:%d team=%p task=%p\n", tid, 1376| 42.5k| team, task)); 1377| 42.5k|} __kmp_finish_implicit_task: 1384| 42.2k|void __kmp_finish_implicit_task(kmp_info_t *thread) { 1385| 42.2k| kmp_taskdata_t *task = thread->th.th_current_task; 1386| 42.2k| if (task->td_dephash) { ------------------ | Branch (1386:7): [True: 0, False: 42.2k] ------------------ 1387| 0| int children; 1388| 0| task->td_flags.complete = 1; 1389| |#if OMPX_TASKGRAPH 1390| | task->td_flags.onced = 1; 1391| |#endif 1392| 0| children = KMP_ATOMIC_LD_ACQ(&task->td_incomplete_child_tasks); ------------------ | | 1249| 0|#define KMP_ATOMIC_LD_ACQ(p) KMP_ATOMIC_LD(p, acquire) | | ------------------ | | | | 1245| 0|#define KMP_ATOMIC_LD(p, order) (p)->load(std::memory_order_##order) | | ------------------ ------------------ 1393| 0| kmp_tasking_flags_t flags_old = task->td_flags; 1394| 0| if (children == 0 && flags_old.complete == 1) { ------------------ | Branch (1394:9): [True: 0, False: 0] | Branch (1394:26): [True: 0, False: 0] ------------------ 1395| 0| kmp_tasking_flags_t flags_new = flags_old; 1396| 0| flags_new.complete = 0; 1397| 0| if (KMP_COMPARE_AND_STORE_ACQ32(RCAST(kmp_int32 *, &task->td_flags), ------------------ | | 807| 0| __sync_bool_compare_and_swap((volatile kmp_uint32 *)(p), (kmp_uint32)(cv), \ | | ------------------ | | | Branch (807:3): [True: 0, False: 0] | | ------------------ | | 808| 0| (kmp_uint32)(sv)) ------------------ 1398| 0| *RCAST(kmp_int32 *, &flags_old), 1399| 0| *RCAST(kmp_int32 *, &flags_new))) { 1400| 0| KA_TRACE(100, ("__kmp_finish_implicit_task: T#%d cleans " 1401| 0| "dephash of implicit task %p\n", 1402| 0| thread->th.th_info.ds.ds_gtid, task)); 1403| 0| __kmp_dephash_free_entries(thread, task->td_dephash); 1404| 0| } 1405| 0| } 1406| 0| } 1407| 42.2k|} __kmp_free_implicit_task: 1413| 63|void __kmp_free_implicit_task(kmp_info_t *thread) { 1414| 63| kmp_taskdata_t *task = thread->th.th_current_task; 1415| 63| if (task && task->td_dephash) { ------------------ | Branch (1415:7): [True: 32, False: 31] | Branch (1415:15): [True: 0, False: 32] ------------------ 1416| 0| __kmp_dephash_free(thread, task->td_dephash); 1417| | task->td_dephash = NULL; 1418| 0| } 1419| 63|} __kmp_free_task_team: 3995| 2|void __kmp_free_task_team(kmp_info_t *thread, kmp_task_team_t *task_team) { 3996| 2| KA_TRACE(20, ("__kmp_free_task_team: T#%d task_team = %p\n", 3997| 2| thread ? __kmp_gtid_from_thread(thread) : -1, task_team)); 3998| | 3999| | // Put task team back on free list 4000| 2| __kmp_acquire_bootstrap_lock(&__kmp_task_team_lock); 4001| | 4002| 2| KMP_DEBUG_ASSERT(task_team->tt.tt_next == NULL); 4003| 2| task_team->tt.tt_next = __kmp_free_task_teams; 4004| 2| TCW_PTR(__kmp_free_task_teams, task_team); ------------------ | | 1157| 2|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 2|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 4005| | 4006| 2| __kmp_release_bootstrap_lock(&__kmp_task_team_lock); 4007| 2|} __kmp_reap_task_teams: 4014| 1|void __kmp_reap_task_teams(void) { 4015| 1| kmp_task_team_t *task_team; 4016| | 4017| 1| if (TCR_PTR(__kmp_free_task_teams) != NULL) { ------------------ | | 1156| 1|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 1|#define TCR_8(a) (a) | | ------------------ ------------------ | Branch (4017:7): [True: 1, False: 0] ------------------ 4018| | // Free all task_teams on the free list 4019| 1| __kmp_acquire_bootstrap_lock(&__kmp_task_team_lock); 4020| 3| while ((task_team = __kmp_free_task_teams) != NULL) { ------------------ | Branch (4020:12): [True: 2, False: 1] ------------------ 4021| 2| __kmp_free_task_teams = task_team->tt.tt_next; 4022| 2| task_team->tt.tt_next = NULL; 4023| | 4024| | // Free threads_data if necessary 4025| 2| if (task_team->tt.tt_threads_data != NULL) { ------------------ | Branch (4025:11): [True: 0, False: 2] ------------------ 4026| 0| __kmp_free_task_threads_data(task_team); 4027| 0| } 4028| 2| if (task_team->tt.tt_task_pri_list != NULL) { ------------------ | Branch (4028:11): [True: 0, False: 2] ------------------ 4029| 0| __kmp_free_task_pri_list(task_team); 4030| 0| } 4031| 2| __kmp_free(task_team); ------------------ | | 3619| 2|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 4032| 2| } 4033| 1| __kmp_release_bootstrap_lock(&__kmp_task_team_lock); 4034| 1| } 4035| 1|} __kmp_wait_to_unref_task_teams: 4040| 1|void __kmp_wait_to_unref_task_teams(void) { 4041| 1| kmp_info_t *thread; 4042| 1| kmp_uint32 spins; 4043| 1| kmp_uint64 time; 4044| 1| int done; 4045| | 4046| 1| KMP_INIT_YIELD(spins); ------------------ | | 1514| 1| { (count) = __kmp_yield_init; } ------------------ 4047| 1| KMP_INIT_BACKOFF(time); ------------------ | | 1517| 1| { (time) = __kmp_pause_init; } ------------------ 4048| | 4049| 1| for (;;) { 4050| 1| done = TRUE; ------------------ | | 1280| 1|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1|#define FALSE 0 | | ------------------ ------------------ 4051| | 4052| | // TODO: GEH - this may be is wrong because some sync would be necessary 4053| | // in case threads are added to the pool during the traversal. Need to 4054| | // verify that lock for thread pool is held when calling this routine. 4055| 32| for (thread = CCAST(kmp_info_t *, __kmp_thread_pool); thread != NULL; ------------------ | | 229| 1|#define CCAST(type, var) const_cast(var) ------------------ | Branch (4055:59): [True: 31, False: 1] ------------------ 4056| 31| thread = thread->th.th_next_pool) { 4057| |#if KMP_OS_WINDOWS 4058| | DWORD exit_val; 4059| |#endif 4060| 31| if (TCR_PTR(thread->th.th_task_team) == NULL) { ------------------ | | 1156| 31|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 31|#define TCR_8(a) (a) | | ------------------ ------------------ | Branch (4060:11): [True: 31, False: 0] ------------------ 4061| 31| KA_TRACE(10, ("__kmp_wait_to_unref_task_team: T#%d task_team == NULL\n", 4062| 31| __kmp_gtid_from_thread(thread))); 4063| 31| continue; 4064| 31| } 4065| |#if KMP_OS_WINDOWS 4066| | // TODO: GEH - add this check for Linux* OS / OS X* as well? 4067| | if (!__kmp_is_thread_alive(thread, &exit_val)) { 4068| | thread->th.th_task_team = NULL; 4069| | continue; 4070| | } 4071| |#endif 4072| | 4073| 0| done = FALSE; // Because th_task_team pointer is not NULL for this thread ------------------ | | 1279| 0|#define FALSE 0 ------------------ 4074| | 4075| 0| KA_TRACE(10, ("__kmp_wait_to_unref_task_team: Waiting for T#%d to " 4076| 0| "unreference task_team\n", 4077| 0| __kmp_gtid_from_thread(thread))); 4078| | 4079| 0| if (__kmp_dflt_blocktime != KMP_MAX_BLOCKTIME) { ------------------ | | 1184| 0| (INT_MAX) /* Must be this for "infinite" setting the work */ ------------------ | Branch (4079:11): [True: 0, False: 0] ------------------ 4080| 0| void *sleep_loc; 4081| | // If the thread is sleeping, awaken it. 4082| 0| if ((sleep_loc = TCR_PTR(CCAST(void *, thread->th.th_sleep_loc))) != ------------------ | | 1156| 0|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 0|#define TCR_8(a) (a) | | ------------------ ------------------ | Branch (4082:13): [True: 0, False: 0] ------------------ 4083| 0| NULL) { 4084| 0| KA_TRACE( 4085| 0| 10, 4086| 0| ("__kmp_wait_to_unref_task_team: T#%d waking up thread T#%d\n", 4087| 0| __kmp_gtid_from_thread(thread), __kmp_gtid_from_thread(thread))); 4088| 0| __kmp_null_resume_wrapper(thread); 4089| 0| } 4090| 0| } 4091| 0| } 4092| 1| if (done) { ------------------ | Branch (4092:9): [True: 1, False: 0] ------------------ 4093| 1| break; 4094| 1| } 4095| | 4096| | // If oversubscribed or have waited a bit, yield. 4097| 0| KMP_YIELD_OVERSUB_ELSE_SPIN(spins, time); ------------------ | | 1564| 0| { \ | | 1565| 0| if (__kmp_tpause_enabled) { \ | | ------------------ | | | Branch (1565:9): [True: 0, False: 0] | | ------------------ | | 1566| 0| if (KMP_OVERSUBSCRIBED) { \ | | ------------------ | | | | 1520| 0| (TCR_4(__kmp_nth) > (__kmp_avail_proc ? __kmp_avail_proc : __kmp_xproc)) | | | | ------------------ | | | | | | 1127| 0|#define TCR_4(a) (a) | | | | ------------------ | | | | | Branch (1520:3): [True: 0, False: 0] | | | | | Branch (1520:24): [True: 0, False: 0] | | | | ------------------ | | ------------------ | | 1567| 0| __kmp_tpause(0, (time)); \ | | 1568| 0| } else { \ | | 1569| 0| __kmp_tpause(__kmp_tpause_hint, (time)); \ | | 1570| 0| } \ | | 1571| 0| (time) = (time << 1 | 1) & KMP_TPAUSE_MAX_MASK; \ | | ------------------ | | | | 1562| 0|#define KMP_TPAUSE_MAX_MASK ((kmp_uint64)0xFFFF) | | ------------------ | | 1572| 0| } else { \ | | 1573| 0| KMP_CPU_PAUSE(); \ | | ------------------ | | | | 1498| 0|#define KMP_CPU_PAUSE() __kmp_x86_pause() | | ------------------ | | 1574| 0| if ((KMP_TRY_YIELD_OVERSUB)) { \ | | ------------------ | | | | 1526| 0| ((__kmp_use_yield == 1 || __kmp_use_yield == 2) && (KMP_OVERSUBSCRIBED)) | | | | ------------------ | | | | | | 1520| 0| (TCR_4(__kmp_nth) > (__kmp_avail_proc ? __kmp_avail_proc : __kmp_xproc)) | | | | | | ------------------ | | | | | | | | 1127| 0|#define TCR_4(a) (a) | | | | | | ------------------ | | | | | | | Branch (1520:24): [True: 0, False: 0] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (1526:5): [True: 0, False: 0] | | | | | Branch (1526:29): [True: 0, False: 0] | | | | | Branch (1526:54): [True: 0, False: 0] | | | | ------------------ | | ------------------ | | 1575| 0| __kmp_yield(); \ | | 1576| 0| } else if (__kmp_use_yield == 1) { \ | | ------------------ | | | Branch (1576:18): [True: 0, False: 0] | | ------------------ | | 1577| 0| (count) -= 2; \ | | 1578| 0| if (!(count)) { \ | | ------------------ | | | Branch (1578:13): [True: 0, False: 0] | | ------------------ | | 1579| 0| __kmp_yield(); \ | | 1580| 0| (count) = __kmp_yield_next; \ | | 1581| 0| } \ | | 1582| 0| } \ | | 1583| 0| } \ | | 1584| 0| } ------------------ 4098| 0| } 4099| 1|} __kmp_task_team_setup: 4141| 2.42k|void __kmp_task_team_setup(kmp_info_t *this_thr, kmp_team_t *team, int always) { 4142| 2.42k| KMP_DEBUG_ASSERT(__kmp_tasking_mode != tskm_immediate_exec); 4143| | 4144| | // If this task_team hasn't been created yet, allocate it. It will be used in 4145| | // the region after the next. 4146| | // If it exists, it is the current task team and shouldn't be touched yet as 4147| | // it may still be in use. 4148| 2.42k| if (team->t.t_task_team[this_thr->th.th_task_state] == NULL && ------------------ | Branch (4148:7): [True: 1, False: 2.42k] ------------------ 4149| 1| (always || team->t.t_nproc > 1)) { ------------------ | Branch (4149:8): [True: 0, False: 1] | Branch (4149:18): [True: 1, False: 0] ------------------ 4150| 1| team->t.t_task_team[this_thr->th.th_task_state] = 4151| 1| __kmp_allocate_task_team(this_thr, team); 4152| 1| KA_TRACE(20, ("__kmp_task_team_setup: Primary T#%d created new task_team %p" 4153| 1| " for team %d at parity=%d\n", 4154| 1| __kmp_gtid_from_thread(this_thr), 4155| 1| team->t.t_task_team[this_thr->th.th_task_state], team->t.t_id, 4156| 1| this_thr->th.th_task_state)); 4157| 1| } 4158| 2.42k| if (this_thr->th.th_task_state == 1 && always && team->t.t_nproc == 1) { ------------------ | Branch (4158:7): [True: 1.21k, False: 1.21k] | Branch (4158:42): [True: 0, False: 1.21k] | Branch (4158:52): [True: 0, False: 0] ------------------ 4159| | // fix task state stack to adjust for proxy and helper tasks 4160| 0| KA_TRACE(20, ("__kmp_task_team_setup: Primary T#%d needs to shift stack" 4161| 0| " for team %d at parity=%d\n", 4162| 0| __kmp_gtid_from_thread(this_thr), team->t.t_id, 4163| 0| this_thr->th.th_task_state)); 4164| 0| __kmp_shift_task_state_stack(this_thr, this_thr->th.th_task_state); 4165| 0| } 4166| | 4167| | // After threads exit the release, they will call sync, and then point to this 4168| | // other task_team; make sure it is allocated and properly initialized. As 4169| | // threads spin in the barrier release phase, they will continue to use the 4170| | // previous task_team struct(above), until they receive the signal to stop 4171| | // checking for tasks (they can't safely reference the kmp_team_t struct, 4172| | // which could be reallocated by the primary thread). No task teams are formed 4173| | // for serialized teams. 4174| 2.42k| if (team->t.t_nproc > 1) { ------------------ | Branch (4174:7): [True: 2.42k, False: 0] ------------------ 4175| 2.42k| int other_team = 1 - this_thr->th.th_task_state; 4176| 2.42k| KMP_DEBUG_ASSERT(other_team >= 0 && other_team < 2); 4177| 2.42k| if (team->t.t_task_team[other_team] == NULL) { // setup other team as well ------------------ | Branch (4177:9): [True: 1, False: 2.42k] ------------------ 4178| 1| team->t.t_task_team[other_team] = 4179| 1| __kmp_allocate_task_team(this_thr, team); 4180| 1| KA_TRACE(20, ("__kmp_task_team_setup: Primary T#%d created second new " 4181| 1| "task_team %p for team %d at parity=%d\n", 4182| 1| __kmp_gtid_from_thread(this_thr), 4183| 1| team->t.t_task_team[other_team], team->t.t_id, other_team)); 4184| 2.42k| } else { // Leave the old task team struct in place for the upcoming region; 4185| | // adjust as needed 4186| 2.42k| kmp_task_team_t *task_team = team->t.t_task_team[other_team]; 4187| 2.42k| if (!task_team->tt.tt_active || ------------------ | Branch (4187:11): [True: 0, False: 2.42k] ------------------ 4188| 2.42k| team->t.t_nproc != task_team->tt.tt_nproc) { ------------------ | Branch (4188:11): [True: 0, False: 2.42k] ------------------ 4189| 0| TCW_4(task_team->tt.tt_nproc, team->t.t_nproc); ------------------ | | 1128| 0|#define TCW_4(a, b) (a) = (b) ------------------ 4190| 0| TCW_4(task_team->tt.tt_found_tasks, FALSE); ------------------ | | 1128| 0|#define TCW_4(a, b) (a) = (b) ------------------ 4191| 0| TCW_4(task_team->tt.tt_found_proxy_tasks, FALSE); ------------------ | | 1128| 0|#define TCW_4(a, b) (a) = (b) ------------------ 4192| 0| TCW_4(task_team->tt.tt_hidden_helper_task_encountered, FALSE); ------------------ | | 1128| 0|#define TCW_4(a, b) (a) = (b) ------------------ 4193| 0| KMP_ATOMIC_ST_REL(&task_team->tt.tt_unfinished_threads, ------------------ | | 1251| 0|#define KMP_ATOMIC_ST_REL(p, v) KMP_ATOMIC_OP(store, p, v, release) | | ------------------ | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 4194| 0| team->t.t_nproc); 4195| 0| TCW_4(task_team->tt.tt_active, TRUE); ------------------ | | 1128| 0|#define TCW_4(a, b) (a) = (b) ------------------ 4196| 0| } 4197| | // if team size has changed, the first thread to enable tasking will 4198| | // realloc threads_data if necessary 4199| 2.42k| KA_TRACE(20, ("__kmp_task_team_setup: Primary T#%d reset next task_team " 4200| 2.42k| "%p for team %d at parity=%d\n", 4201| 2.42k| __kmp_gtid_from_thread(this_thr), 4202| 2.42k| team->t.t_task_team[other_team], team->t.t_id, other_team)); 4203| 2.42k| } 4204| 2.42k| } 4205| | 4206| | // For regular thread, task enabling should be called when the task is going 4207| | // to be pushed to a dequeue. However, for the hidden helper thread, we need 4208| | // it ahead of time so that some operations can be performed without race 4209| | // condition. 4210| 2.42k| if (this_thr == __kmp_hidden_helper_main_thread) { ------------------ | Branch (4210:7): [True: 0, False: 2.42k] ------------------ 4211| 0| for (int i = 0; i < 2; ++i) { ------------------ | Branch (4211:21): [True: 0, False: 0] ------------------ 4212| 0| kmp_task_team_t *task_team = team->t.t_task_team[i]; 4213| 0| if (KMP_TASKING_ENABLED(task_team)) { ------------------ | | 2368| 0| (TRUE == TCR_SYNC_4((task_team)->tt.tt_found_tasks)) | | ------------------ | | | | 1280| 0|#define TRUE (!FALSE) | | | | ------------------ | | | | | | 1279| 0|#define FALSE 0 | | | | ------------------ | | ------------------ | | (TRUE == TCR_SYNC_4((task_team)->tt.tt_found_tasks)) | | ------------------ | | | | 1135| 0|#define TCR_SYNC_4(a) (a) | | ------------------ | | | Branch (2368:3): [True: 0, False: 0] | | ------------------ ------------------ 4214| 0| continue; 4215| 0| } 4216| 0| __kmp_enable_tasking(task_team, this_thr); 4217| 0| for (int j = 0; j < task_team->tt.tt_nproc; ++j) { ------------------ | Branch (4217:23): [True: 0, False: 0] ------------------ 4218| 0| kmp_thread_data_t *thread_data = &task_team->tt.tt_threads_data[j]; 4219| 0| if (thread_data->td.td_deque == NULL) { ------------------ | Branch (4219:13): [True: 0, False: 0] ------------------ 4220| 0| __kmp_alloc_task_deque(__kmp_hidden_helper_threads[j], thread_data); 4221| 0| } 4222| 0| } 4223| 0| } 4224| 0| } 4225| 2.42k|} __kmp_task_team_sync: 4230| 75.4k|void __kmp_task_team_sync(kmp_info_t *this_thr, kmp_team_t *team) { 4231| 75.4k| KMP_DEBUG_ASSERT(__kmp_tasking_mode != tskm_immediate_exec); 4232| | 4233| | // Toggle the th_task_state field, to switch which task_team this thread 4234| | // refers to 4235| 75.4k| this_thr->th.th_task_state = (kmp_uint8)(1 - this_thr->th.th_task_state); 4236| | 4237| | // It is now safe to propagate the task team pointer from the team struct to 4238| | // the current thread. 4239| 75.4k| TCW_PTR(this_thr->th.th_task_team, ------------------ | | 1157| 75.4k|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 75.4k|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 4240| 75.4k| team->t.t_task_team[this_thr->th.th_task_state]); 4241| 75.4k| KA_TRACE(20, 4242| 75.4k| ("__kmp_task_team_sync: Thread T#%d task team switched to task_team " 4243| 75.4k| "%p from Team #%d (parity=%d)\n", 4244| 75.4k| __kmp_gtid_from_thread(this_thr), this_thr->th.th_task_team, 4245| 75.4k| team->t.t_id, this_thr->th.th_task_state)); 4246| 75.4k|} __kmp_task_team_wait: 4257| 2.42k| kmp_team_t *team USE_ITT_BUILD_ARG(void *itt_sync_obj), int wait) { 4258| 2.42k| kmp_task_team_t *task_team = team->t.t_task_team[this_thr->th.th_task_state]; 4259| | 4260| 2.42k| KMP_DEBUG_ASSERT(__kmp_tasking_mode != tskm_immediate_exec); 4261| 2.42k| KMP_DEBUG_ASSERT(task_team == this_thr->th.th_task_team); 4262| | 4263| 2.42k| if ((task_team != NULL) && KMP_TASKING_ENABLED(task_team)) { ------------------ | | 2368| 2.42k| (TRUE == TCR_SYNC_4((task_team)->tt.tt_found_tasks)) | | ------------------ | | | | 1280| 2.42k|#define TRUE (!FALSE) | | | | ------------------ | | | | | | 1279| 2.42k|#define FALSE 0 | | | | ------------------ | | ------------------ | | (TRUE == TCR_SYNC_4((task_team)->tt.tt_found_tasks)) | | ------------------ | | | | 1135| 2.42k|#define TCR_SYNC_4(a) (a) | | ------------------ | | | Branch (2368:3): [True: 0, False: 2.42k] | | ------------------ ------------------ | Branch (4263:7): [True: 2.42k, False: 0] ------------------ 4264| 0| if (wait) { ------------------ | Branch (4264:9): [True: 0, False: 0] ------------------ 4265| 0| KA_TRACE(20, ("__kmp_task_team_wait: Primary T#%d waiting for all tasks " 4266| 0| "(for unfinished_threads to reach 0) on task_team = %p\n", 4267| 0| __kmp_gtid_from_thread(this_thr), task_team)); 4268| | // Worker threads may have dropped through to release phase, but could 4269| | // still be executing tasks. Wait here for tasks to complete. To avoid 4270| | // memory contention, only primary thread checks termination condition. 4271| 0| kmp_flag_32 flag( 4272| 0| RCAST(std::atomic *, ------------------ | | 230| 0|#define RCAST(type, var) reinterpret_cast(var) ------------------ 4273| 0| &task_team->tt.tt_unfinished_threads), 4274| 0| 0U); 4275| 0| flag.wait(this_thr, TRUE USE_ITT_BUILD_ARG(itt_sync_obj)); ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 4276| 0| } 4277| | // Deactivate the old task team, so that the worker threads will stop 4278| | // referencing it while spinning. 4279| 0| KA_TRACE( 4280| 0| 20, 4281| 0| ("__kmp_task_team_wait: Primary T#%d deactivating task_team %p: " 4282| 0| "setting active to false, setting local and team's pointer to NULL\n", 4283| 0| __kmp_gtid_from_thread(this_thr), task_team)); 4284| 0| KMP_DEBUG_ASSERT(task_team->tt.tt_nproc > 1 || 4285| 0| task_team->tt.tt_found_proxy_tasks == TRUE || 4286| 0| task_team->tt.tt_hidden_helper_task_encountered == TRUE); 4287| 0| TCW_SYNC_4(task_team->tt.tt_found_proxy_tasks, FALSE); ------------------ | | 1136| 0|#define TCW_SYNC_4(a, b) (a) = (b) ------------------ 4288| 0| TCW_SYNC_4(task_team->tt.tt_hidden_helper_task_encountered, FALSE); ------------------ | | 1136| 0|#define TCW_SYNC_4(a, b) (a) = (b) ------------------ 4289| 0| KMP_CHECK_UPDATE(task_team->tt.tt_untied_task_encountered, 0); ------------------ | | 2280| 0| if ((a) != (b)) \ | | ------------------ | | | Branch (2280:7): [True: 0, False: 0] | | ------------------ | | 2281| 0| (a) = (b) ------------------ 4290| 0| TCW_SYNC_4(task_team->tt.tt_active, FALSE); ------------------ | | 1136| 0|#define TCW_SYNC_4(a, b) (a) = (b) ------------------ 4291| 0| KMP_MB(); 4292| | 4293| 0| TCW_PTR(this_thr->th.th_task_team, NULL); ------------------ | | 1157| 0|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 0|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 4294| 0| } 4295| 2.42k|} kmp_tasking.cpp:_ZL24__kmp_allocate_task_teamP8kmp_infoP8kmp_team: 3932| 2| kmp_team_t *team) { 3933| 2| kmp_task_team_t *task_team = NULL; 3934| 2| int nthreads; 3935| | 3936| 2| KA_TRACE(20, ("__kmp_allocate_task_team: T#%d entering; team = %p\n", 3937| 2| (thread ? __kmp_gtid_from_thread(thread) : -1), team)); 3938| | 3939| 2| if (TCR_PTR(__kmp_free_task_teams) != NULL) { ------------------ | | 1156| 2|#define TCR_PTR(a) ((void *)TCR_8(a)) | | ------------------ | | | | 1131| 2|#define TCR_8(a) (a) | | ------------------ ------------------ | Branch (3939:7): [True: 0, False: 2] ------------------ 3940| | // Take a task team from the task team pool 3941| 0| __kmp_acquire_bootstrap_lock(&__kmp_task_team_lock); 3942| 0| if (__kmp_free_task_teams != NULL) { ------------------ | Branch (3942:9): [True: 0, False: 0] ------------------ 3943| 0| task_team = __kmp_free_task_teams; 3944| 0| TCW_PTR(__kmp_free_task_teams, task_team->tt.tt_next); ------------------ | | 1157| 0|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 0|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 3945| 0| task_team->tt.tt_next = NULL; 3946| 0| } 3947| 0| __kmp_release_bootstrap_lock(&__kmp_task_team_lock); 3948| 0| } 3949| | 3950| 2| if (task_team == NULL) { ------------------ | Branch (3950:7): [True: 2, False: 0] ------------------ 3951| 2| KE_TRACE(10, ("__kmp_allocate_task_team: T#%d allocating " 3952| 2| "task team for team %p\n", 3953| 2| __kmp_gtid_from_thread(thread), team)); 3954| | // Allocate a new task team if one is not available. Cannot use 3955| | // __kmp_thread_malloc because threads not around for kmp_reap_task_team. 3956| 2| task_team = (kmp_task_team_t *)__kmp_allocate(sizeof(kmp_task_team_t)); ------------------ | | 3617| 2|#define __kmp_allocate(size) ___kmp_allocate((size)KMP_SRC_LOC_CURR) ------------------ 3957| 2| __kmp_init_bootstrap_lock(&task_team->tt.tt_threads_lock); 3958| 2| __kmp_init_bootstrap_lock(&task_team->tt.tt_task_pri_lock); 3959| |#if USE_ITT_BUILD && USE_ITT_NOTIFY && KMP_DEBUG 3960| | // suppress race conditions detection on synchronization flags in debug mode 3961| | // this helps to analyze library internals eliminating false positives 3962| | __itt_suppress_mark_range( 3963| | __itt_suppress_range, __itt_suppress_threading_errors, 3964| | &task_team->tt.tt_found_tasks, sizeof(task_team->tt.tt_found_tasks)); 3965| | __itt_suppress_mark_range(__itt_suppress_range, 3966| | __itt_suppress_threading_errors, 3967| | CCAST(kmp_uint32 *, &task_team->tt.tt_active), 3968| | sizeof(task_team->tt.tt_active)); 3969| |#endif /* USE_ITT_BUILD && USE_ITT_NOTIFY && KMP_DEBUG */ 3970| | // Note: __kmp_allocate zeroes returned memory, othewise we would need: 3971| | // task_team->tt.tt_threads_data = NULL; 3972| | // task_team->tt.tt_max_threads = 0; 3973| | // task_team->tt.tt_next = NULL; 3974| 2| } 3975| | 3976| 2| TCW_4(task_team->tt.tt_found_tasks, FALSE); ------------------ | | 1128| 2|#define TCW_4(a, b) (a) = (b) ------------------ 3977| 2| TCW_4(task_team->tt.tt_found_proxy_tasks, FALSE); ------------------ | | 1128| 2|#define TCW_4(a, b) (a) = (b) ------------------ 3978| 2| TCW_4(task_team->tt.tt_hidden_helper_task_encountered, FALSE); ------------------ | | 1128| 2|#define TCW_4(a, b) (a) = (b) ------------------ 3979| 2| task_team->tt.tt_nproc = nthreads = team->t.t_nproc; 3980| | 3981| 2| KMP_ATOMIC_ST_REL(&task_team->tt.tt_unfinished_threads, nthreads); ------------------ | | 1251| 2|#define KMP_ATOMIC_ST_REL(p, v) KMP_ATOMIC_OP(store, p, v, release) | | ------------------ | | | | 1246| 2|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 3982| 2| TCW_4(task_team->tt.tt_hidden_helper_task_encountered, FALSE); ------------------ | | 1128| 2|#define TCW_4(a, b) (a) = (b) ------------------ 3983| 2| TCW_4(task_team->tt.tt_active, TRUE); ------------------ | | 1128| 2|#define TCW_4(a, b) (a) = (b) ------------------ 3984| | 3985| 2| KA_TRACE(20, ("__kmp_allocate_task_team: T#%d exiting; task_team = %p " 3986| 2| "unfinished_threads init'd to %d\n", 3987| 2| (thread ? __kmp_gtid_from_thread(thread) : -1), task_team, 3988| 2| KMP_ATOMIC_LD_RLX(&task_team->tt.tt_unfinished_threads))); 3989| 2| return task_team; 3990| 2|} __kmp_common_initialize: 129| 1|void __kmp_common_initialize(void) { 130| 1| if (!TCR_4(__kmp_init_common)) { ------------------ | | 1127| 1|#define TCR_4(a) (a) ------------------ | Branch (130:7): [True: 1, False: 0] ------------------ 131| 1| int q; 132| |#ifdef KMP_DEBUG 133| | int gtid; 134| |#endif 135| | 136| 1| __kmp_threadpriv_cache_list = NULL; 137| | 138| |#ifdef KMP_DEBUG 139| | /* verify the uber masters were initialized */ 140| | for (gtid = 0; gtid < __kmp_threads_capacity; gtid++) 141| | if (__kmp_root[gtid]) { 142| | KMP_DEBUG_ASSERT(__kmp_root[gtid]->r.r_uber_thread); 143| | for (q = 0; q < KMP_HASH_TABLE_SIZE; ++q) 144| | KMP_DEBUG_ASSERT( 145| | !__kmp_root[gtid]->r.r_uber_thread->th.th_pri_common->data[q]); 146| | /* __kmp_root[ gitd ]-> r.r_uber_thread -> 147| | * th.th_pri_common -> data[ q ] = 0;*/ 148| | } 149| |#endif /* KMP_DEBUG */ 150| | 151| 513| for (q = 0; q < KMP_HASH_TABLE_SIZE; ++q) ------------------ | | 1796| 513| (1 << KMP_HASH_TABLE_LOG2) /* size of the hash table */ | | ------------------ | | | | 1794| 513|#define KMP_HASH_TABLE_LOG2 9 /* log2 of the hash table size */ | | ------------------ ------------------ | Branch (151:17): [True: 512, False: 1] ------------------ 152| 512| __kmp_threadprivate_d_table.data[q] = 0; 153| | 154| 1| TCW_4(__kmp_init_common, TRUE); ------------------ | | 1128| 1|#define TCW_4(a, b) (a) = (b) ------------------ 155| 1| } 156| 1|} __kmp_common_destroy_gtid: 223| 31|void __kmp_common_destroy_gtid(int gtid) { 224| 31| struct private_common *tn; 225| 31| struct shared_common *d_tn; 226| | 227| 31| if (!TCR_4(__kmp_init_gtid)) { ------------------ | | 1127| 31|#define TCR_4(a) (a) ------------------ | Branch (227:7): [True: 0, False: 31] ------------------ 228| | // This is possible when one of multiple roots initiates early library 229| | // termination in a sequential region while other teams are active, and its 230| | // child threads are about to end. 231| 0| return; 232| 0| } 233| | 234| 31| KC_TRACE(10, ("__kmp_common_destroy_gtid: T#%d called\n", gtid)); 235| 31| if ((__kmp_foreign_tp) ? (!KMP_INITIAL_GTID(gtid)) : (!KMP_UBER_GTID(gtid))) { ------------------ | | 1276| 31|#define KMP_INITIAL_GTID(gtid) (0 == (gtid)) ------------------ | Branch (235:7): [True: 31, False: 0] | Branch (235:7): [True: 31, False: 0] ------------------ 236| | 237| 31| if (TCR_4(__kmp_init_common)) { ------------------ | | 1127| 31|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 31, False: 0] | | ------------------ ------------------ 238| | 239| | /* Cannot do this here since not all threads have destroyed their data */ 240| | /* TCW_4(__kmp_init_common, FALSE); */ 241| | 242| 31| for (tn = __kmp_threads[gtid]->th.th_pri_head; tn; tn = tn->link) { ------------------ | Branch (242:54): [True: 0, False: 31] ------------------ 243| | 244| 0| d_tn = __kmp_find_shared_task_common(&__kmp_threadprivate_d_table, gtid, 245| 0| tn->gbl_addr); 246| 0| if (d_tn == NULL) ------------------ | Branch (246:13): [True: 0, False: 0] ------------------ 247| 0| continue; 248| 0| if (d_tn->is_vec) { ------------------ | Branch (248:13): [True: 0, False: 0] ------------------ 249| 0| if (d_tn->dt.dtorv != 0) { ------------------ | Branch (249:15): [True: 0, False: 0] ------------------ 250| 0| (void)(*d_tn->dt.dtorv)(tn->par_addr, d_tn->vec_len); 251| 0| } 252| 0| if (d_tn->obj_init != 0) { ------------------ | Branch (252:15): [True: 0, False: 0] ------------------ 253| 0| (void)(*d_tn->dt.dtorv)(d_tn->obj_init, d_tn->vec_len); 254| 0| } 255| 0| } else { 256| 0| if (d_tn->dt.dtor != 0) { ------------------ | Branch (256:15): [True: 0, False: 0] ------------------ 257| 0| (void)(*d_tn->dt.dtor)(tn->par_addr); 258| 0| } 259| 0| if (d_tn->obj_init != 0) { ------------------ | Branch (259:15): [True: 0, False: 0] ------------------ 260| 0| (void)(*d_tn->dt.dtor)(d_tn->obj_init); 261| 0| } 262| 0| } 263| 0| } 264| 31| KC_TRACE(30, ("__kmp_common_destroy_gtid: T#%d threadprivate destructors " 265| 31| "complete\n", 266| 31| gtid)); 267| 31| } 268| 31| } 269| 31|} __kmp_cleanup_threadprivate_caches: 781| 1|void __kmp_cleanup_threadprivate_caches() { 782| 1| kmp_cached_addr_t *ptr = __kmp_threadpriv_cache_list; 783| | 784| 1| while (ptr) { ------------------ | Branch (784:10): [True: 0, False: 1] ------------------ 785| 0| void **cache = ptr->addr; 786| 0| __kmp_threadpriv_cache_list = ptr->next; 787| 0| if (*ptr->compiler_cache) ------------------ | Branch (787:9): [True: 0, False: 0] ------------------ 788| 0| *ptr->compiler_cache = NULL; 789| 0| ptr->compiler_cache = NULL; 790| 0| ptr->data = NULL; 791| 0| ptr->addr = NULL; 792| 0| ptr->next = NULL; 793| | // Threadprivate data pointed at by cache entries are destroyed at end of 794| | // __kmp_launch_thread with __kmp_common_destroy_gtid. 795| 0| __kmp_free(cache); // implicitly frees ptr too ------------------ | | 3619| 0|#define __kmp_free(ptr) ___kmp_free((ptr)KMP_SRC_LOC_CURR) ------------------ 796| 0| ptr = __kmp_threadpriv_cache_list; 797| 0| } 798| 1|} _Z21__kmp_get_physical_idii: 35| 1|int __kmp_get_physical_id(int log_per_phy, int apic_id) { 36| 1| int index_lsb, index_msb, temp; 37| | 38| 1| if (log_per_phy > 1) { ------------------ | Branch (38:7): [True: 1, False: 0] ------------------ 39| 1| index_lsb = 0; 40| 1| index_msb = 31; 41| | 42| 1| temp = log_per_phy; 43| 6| while ((temp & 1) == 0) { ------------------ | Branch (43:12): [True: 5, False: 1] ------------------ 44| 5| temp >>= 1; 45| 5| index_lsb++; 46| 5| } 47| | 48| 1| temp = log_per_phy; 49| 27| while ((temp & 0x80000000) == 0) { ------------------ | Branch (49:12): [True: 26, False: 1] ------------------ 50| 26| temp <<= 1; 51| 26| index_msb--; 52| 26| } 53| | 54| | /* If >1 bits were set in log_per_phy, choose next higher power of 2 */ 55| 1| if (index_lsb != index_msb) ------------------ | Branch (55:9): [True: 0, False: 1] ------------------ 56| 0| index_msb++; 57| | 58| 1| return ((int)(apic_id >> index_msb)); 59| 1| } 60| | 61| 0| return apic_id; 62| 1|} _Z20__kmp_get_logical_idii: 69| 1|int __kmp_get_logical_id(int log_per_phy, int apic_id) { 70| 1| unsigned current_bit; 71| 1| int bits_seen; 72| | 73| 1| if (log_per_phy <= 1) ------------------ | Branch (73:7): [True: 0, False: 1] ------------------ 74| 0| return (0); 75| | 76| 1| bits_seen = 0; 77| | 78| 7| for (current_bit = 1; log_per_phy != 0; current_bit <<= 1) { ------------------ | Branch (78:25): [True: 6, False: 1] ------------------ 79| 6| if (log_per_phy & current_bit) { ------------------ | Branch (79:9): [True: 1, False: 5] ------------------ 80| 1| log_per_phy &= ~current_bit; 81| 1| bits_seen++; 82| 1| } 83| 6| } 84| | 85| | /* If exactly 1 bit was set in log_per_phy, choose next lower power of 2 */ 86| 1| if (bits_seen == 1) { ------------------ | Branch (86:7): [True: 1, False: 0] ------------------ 87| 1| current_bit >>= 1; 88| 1| } 89| | 90| 1| return ((int)((current_bit - 1) & apic_id)); 91| 1|} __kmp_query_cpuid: 122| 1|void __kmp_query_cpuid(kmp_cpuinfo_t *p) { 123| 1| struct kmp_cpuid buf; 124| 1| int max_arg; 125| 1| int log_per_phy; 126| |#ifdef KMP_DEBUG 127| | int cflush_size; 128| |#endif 129| | 130| 1| p->initialized = 1; 131| | 132| 1| p->flags.sse2 = 1; // Assume SSE2 by default. 133| | 134| 1| __kmp_x86_cpuid(0, 0, &buf); 135| | 136| 1| KA_TRACE(trace_level, 137| 1| ("INFO: CPUID %d: EAX=0x%08X EBX=0x%08X ECX=0x%08X EDX=0x%08X\n", 0, 138| 1| buf.eax, buf.ebx, buf.ecx, buf.edx)); 139| | 140| 1| max_arg = buf.eax; 141| | 142| 1| p->apic_id = -1; 143| | 144| 1| if (max_arg >= 1) { ------------------ | Branch (144:7): [True: 1, False: 0] ------------------ 145| 1| int i; 146| 1| kmp_uint32 t, data[4]; 147| | 148| 1| __kmp_x86_cpuid(1, 0, &buf); 149| 1| KA_TRACE(trace_level, 150| 1| ("INFO: CPUID %d: EAX=0x%08X EBX=0x%08X ECX=0x%08X EDX=0x%08X\n", 151| 1| 1, buf.eax, buf.ebx, buf.ecx, buf.edx)); 152| | 153| 1| { 154| 1|#define get_value(reg, lo, mask) (((reg) >> (lo)) & (mask)) 155| | 156| 1| p->signature = buf.eax; 157| 1| p->family = get_value(buf.eax, 20, 0xff) + get_value(buf.eax, 8, 0x0f); ------------------ | | 154| 1|#define get_value(reg, lo, mask) (((reg) >> (lo)) & (mask)) ------------------ p->family = get_value(buf.eax, 20, 0xff) + get_value(buf.eax, 8, 0x0f); ------------------ | | 154| 1|#define get_value(reg, lo, mask) (((reg) >> (lo)) & (mask)) ------------------ 158| 1| p->model = 159| 1| (get_value(buf.eax, 16, 0x0f) << 4) + get_value(buf.eax, 4, 0x0f); ------------------ | | 154| 1|#define get_value(reg, lo, mask) (((reg) >> (lo)) & (mask)) ------------------ (get_value(buf.eax, 16, 0x0f) << 4) + get_value(buf.eax, 4, 0x0f); ------------------ | | 154| 1|#define get_value(reg, lo, mask) (((reg) >> (lo)) & (mask)) ------------------ 160| 1| p->stepping = get_value(buf.eax, 0, 0x0f); ------------------ | | 154| 1|#define get_value(reg, lo, mask) (((reg) >> (lo)) & (mask)) ------------------ 161| | 162| 1|#undef get_value 163| | 164| 1| KA_TRACE(trace_level, (" family = %d, model = %d, stepping = %d\n", 165| 1| p->family, p->model, p->stepping)); 166| 1| } 167| | 168| 5| for (t = buf.ebx, i = 0; i < 4; t >>= 8, ++i) { ------------------ | Branch (168:30): [True: 4, False: 1] ------------------ 169| 4| data[i] = (t & 0xff); 170| 4| } 171| | 172| 1| p->flags.sse2 = (buf.edx >> 26) & 1; 173| | 174| |#ifdef KMP_DEBUG 175| | 176| | if ((buf.edx >> 4) & 1) { 177| | /* TSC - Timestamp Counter Available */ 178| | KA_TRACE(trace_level, (" TSC")); 179| | } 180| | if ((buf.edx >> 8) & 1) { 181| | /* CX8 - CMPXCHG8B Instruction Available */ 182| | KA_TRACE(trace_level, (" CX8")); 183| | } 184| | if ((buf.edx >> 9) & 1) { 185| | /* APIC - Local APIC Present (multi-processor operation support */ 186| | KA_TRACE(trace_level, (" APIC")); 187| | } 188| | if ((buf.edx >> 15) & 1) { 189| | /* CMOV - Conditional MOVe Instruction Available */ 190| | KA_TRACE(trace_level, (" CMOV")); 191| | } 192| | if ((buf.edx >> 18) & 1) { 193| | /* PSN - Processor Serial Number Available */ 194| | KA_TRACE(trace_level, (" PSN")); 195| | } 196| | if ((buf.edx >> 19) & 1) { 197| | /* CLFLUSH - Cache Flush Instruction Available */ 198| | cflush_size = 199| | data[1] * 8; /* Bits 15-08: CLFLUSH line size = 8 (64 bytes) */ 200| | KA_TRACE(trace_level, (" CLFLUSH(%db)", cflush_size)); 201| | } 202| | if ((buf.edx >> 21) & 1) { 203| | /* DTES - Debug Trace & EMON Store */ 204| | KA_TRACE(trace_level, (" DTES")); 205| | } 206| | if ((buf.edx >> 22) & 1) { 207| | /* ACPI - ACPI Support Available */ 208| | KA_TRACE(trace_level, (" ACPI")); 209| | } 210| | if ((buf.edx >> 23) & 1) { 211| | /* MMX - Multimedia Extensions */ 212| | KA_TRACE(trace_level, (" MMX")); 213| | } 214| | if ((buf.edx >> 25) & 1) { 215| | /* SSE - SSE Instructions */ 216| | KA_TRACE(trace_level, (" SSE")); 217| | } 218| | if ((buf.edx >> 26) & 1) { 219| | /* SSE2 - SSE2 Instructions */ 220| | KA_TRACE(trace_level, (" SSE2")); 221| | } 222| | if ((buf.edx >> 27) & 1) { 223| | /* SLFSNP - Self-Snooping Cache */ 224| | KA_TRACE(trace_level, (" SLFSNP")); 225| | } 226| |#endif /* KMP_DEBUG */ 227| | 228| 1| if ((buf.edx >> 28) & 1) { ------------------ | Branch (228:9): [True: 1, False: 0] ------------------ 229| | /* Bits 23-16: Logical Processors per Physical Processor (1 for P4) */ 230| 1| log_per_phy = data[2]; 231| 1| p->apic_id = data[3]; /* Bits 31-24: Processor Initial APIC ID (X) */ 232| 1| KA_TRACE(trace_level, (" HT(%d TPUs)", log_per_phy)); 233| 1| p->physical_id = __kmp_get_physical_id(log_per_phy, p->apic_id); 234| 1| p->logical_id = __kmp_get_logical_id(log_per_phy, p->apic_id); 235| 1| } 236| |#ifdef KMP_DEBUG 237| | if ((buf.edx >> 29) & 1) { 238| | /* ATHROTL - Automatic Throttle Control */ 239| | KA_TRACE(trace_level, (" ATHROTL")); 240| | } 241| | KA_TRACE(trace_level, (" ]\n")); 242| | 243| | for (i = 2; i <= max_arg; ++i) { 244| | __kmp_x86_cpuid(i, 0, &buf); 245| | KA_TRACE(trace_level, 246| | ("INFO: CPUID %d: EAX=0x%08X EBX=0x%08X ECX=0x%08X EDX=0x%08X\n", 247| | i, buf.eax, buf.ebx, buf.ecx, buf.edx)); 248| | } 249| |#endif 250| 1| p->flags.rtm = 0; 251| 1| p->flags.hybrid = 0; 252| 1| if (max_arg > 7) { ------------------ | Branch (252:9): [True: 1, False: 0] ------------------ 253| | /* RTM bit CPUID.07:EBX, bit 11 */ 254| | /* HYRBID bit CPUID.07:EDX, bit 15 */ 255| 1| __kmp_x86_cpuid(7, 0, &buf); 256| 1| p->flags.rtm = (buf.ebx >> 11) & 1; 257| 1| p->flags.hybrid = (buf.edx >> 15) & 1; 258| 1| if (p->flags.rtm) { ------------------ | Branch (258:11): [True: 0, False: 1] ------------------ 259| 0| KA_TRACE(trace_level, (" RTM")); 260| 0| } 261| 1| if (p->flags.hybrid) { ------------------ | Branch (261:11): [True: 0, False: 1] ------------------ 262| 0| KA_TRACE(trace_level, (" HYBRID")); 263| 0| } 264| 1| } 265| 1| } 266| | 267| 1| { // Parse CPU brand string for frequency, saving the string for later. 268| 1| int i; 269| 1| kmp_cpuid_t *base = (kmp_cpuid_t *)&p->name[0]; 270| | 271| | // Get CPU brand string. 272| 4| for (i = 0; i < 3; ++i) { ------------------ | Branch (272:17): [True: 3, False: 1] ------------------ 273| 3| __kmp_x86_cpuid(0x80000002 + i, 0, base + i); 274| 3| } 275| 1| p->name[sizeof(p->name) - 1] = 0; // Just in case. ;-) 276| 1| KA_TRACE(trace_level, ("cpu brand string: \"%s\"\n", &p->name[0])); 277| | 278| | // Parse frequency. 279| 1| p->frequency = __kmp_parse_frequency(strrchr(&p->name[0], ' ')); 280| 1| KA_TRACE(trace_level, 281| 1| ("cpu frequency from brand string: %" KMP_UINT64_SPEC "\n", 282| 1| p->frequency)); 283| 1| } 284| 1|} kmp_utility.cpp:_ZL21__kmp_parse_frequencyPKc: 95| 1|) { 96| | 97| 1| double value = 0.0; 98| 1| char *unit = NULL; 99| 1| kmp_uint64 result = 0; /* Zero is a better unknown value than all ones. */ 100| | 101| 1| if (frequency == NULL) { ------------------ | Branch (101:7): [True: 0, False: 1] ------------------ 102| 0| return result; 103| 0| } 104| 1| value = strtod(frequency, &unit); 105| 1| if (0 < value && ------------------ | Branch (105:7): [True: 1, False: 0] ------------------ 106| 1| value <= DBL_MAX) { // Good value (not overflow, underflow, etc). ------------------ | Branch (106:7): [True: 1, False: 0] ------------------ 107| 1| if (strcmp(unit, "MHz") == 0) { ------------------ | Branch (107:9): [True: 0, False: 1] ------------------ 108| 0| value = value * 1.0E+6; 109| 1| } else if (strcmp(unit, "GHz") == 0) { ------------------ | Branch (109:16): [True: 0, False: 1] ------------------ 110| 0| value = value * 1.0E+9; 111| 1| } else if (strcmp(unit, "THz") == 0) { ------------------ | Branch (111:16): [True: 0, False: 1] ------------------ 112| 0| value = value * 1.0E+12; 113| 1| } else { // Wrong unit. 114| 1| return result; 115| 1| } 116| 0| result = (kmp_uint64)value; // rounds down 117| 0| } 118| 0| return result; 119| | 120| 1|} // func __kmp_parse_cpu_frequency __kmp_release_64: 25| 31|void __kmp_release_64(kmp_flag_64<> *flag) { __kmp_release_template(flag); } _ZN15kmp_flag_nativeIyL9flag_type1ELb1EED2Ev: 157| 297k| virtual ~kmp_flag_native() {} _ZN11kmp_flag_64ILb0ELb1EEC2EPVyP8kmp_info: 861| 72.8k| : kmp_flag_native(p, thr) {} _ZN15kmp_flag_nativeIyL9flag_type1ELb1EEC2EPVyP8kmp_info: 150| 72.9k| : kmp_flag(1), loc(p) { 151| 72.9k| this->waiting_threads[0] = thr; 152| 72.9k| } _ZN8kmp_flagIL9flag_type1EEC2Ei: 115| 72.8k| : t({FlagType, 0U}), num_waiting_threads(nwaiters), sleepLoc(nullptr) {} _ZN15kmp_flag_nativeIyL9flag_type1ELb1EE10done_checkEv: 166| 150k| virtual bool done_check() { 167| 150k| if (Sleepable && !(this->sleepLoc)) ------------------ | Branch (167:9): [True: 149k, Folded] | Branch (167:22): [True: 150k, False: 18.4E] ------------------ 168| 149k| return (traits_type::tcr(*(this->get())) & ~KMP_BARRIER_SLEEP_STATE) == ------------------ | | 2023| 149k|#define KMP_BARRIER_SLEEP_STATE (1 << KMP_BARRIER_SLEEP_BIT) | | ------------------ | | | | 2019| 149k|#define KMP_BARRIER_SLEEP_BIT 0 /* bit used for suspend/sleep part of state */ | | ------------------ ------------------ 169| 149k| checker; 170| 233| else 171| 233| return traits_type::tcr(*(this->get())) == checker; 172| 150k| } _ZN11flag_traitsIL9flag_type1EE3tcrEy: 76| 423M| static inline flag_t tcr(flag_t f) { return TCR_8(f); } ------------------ | | 1131| 423M|#define TCR_8(a) (a) ------------------ _ZN15kmp_flag_nativeIyL9flag_type1ELb1EE3getEv: 160| 422M| volatile PtrType *get() { return loc; } _ZN15kmp_flag_nativeIyL9flag_type1ELb1EE14done_check_valEy: 175| 1.62k| virtual bool done_check_val(PtrType old_loc) { return old_loc == checker; } _ZN15kmp_flag_nativeIyL9flag_type1ELb1EE13notdone_checkEv: 181| 418M| virtual bool notdone_check() { 182| 418M| return traits_type::tcr(*(this->get())) != checker; 183| 418M| } _ZN15kmp_flag_nativeIyL9flag_type1ELb1EE11is_sleepingEv: 219| 6.31k| bool is_sleeping() { 220| 6.31k| if (this->sleepLoc) ------------------ | Branch (220:9): [True: 0, False: 6.31k] ------------------ 221| 0| return this->sleepLoc->load(); 222| 6.31k| return is_sleeping_val(*(this->get())); 223| 6.31k| } _ZN15kmp_flag_nativeIyL9flag_type1ELb1EE15is_sleeping_valEy: 213| 155k| bool is_sleeping_val(PtrType old_loc) { 214| 155k| if (this->sleepLoc) ------------------ | Branch (214:9): [True: 0, False: 155k] ------------------ 215| 0| return this->sleepLoc->load(); 216| 155k| return old_loc & KMP_BARRIER_SLEEP_STATE; ------------------ | | 2023| 155k|#define KMP_BARRIER_SLEEP_STATE (1 << KMP_BARRIER_SLEEP_BIT) | | ------------------ | | | | 2019| 155k|#define KMP_BARRIER_SLEEP_BIT 0 /* bit used for suspend/sleep part of state */ | | ------------------ ------------------ 217| 155k| } _ZN11kmp_flag_64ILb0ELb1EE6resumeEi: 870| 1.55k| void resume(int th_gtid) { __kmp_resume_64(th_gtid, this); } _ZN11flag_traitsIL9flag_type1EE14test_then_add4EPVy: 77| 147k| static inline flag_t test_then_add4(volatile flag_t *f) { 78| 147k| return KMP_TEST_THEN_ADD4_64(RCAST(volatile kmp_int64 *, f)); ------------------ | | 747| 147k| __sync_fetch_and_add((volatile kmp_int64 *)(p), 4LL) ------------------ 79| 147k| } _ZN11flag_traitsIL9flag_type1EE12test_then_orEPVyy: 80| 1.62k| static inline flag_t test_then_or(volatile flag_t *f, flag_t v) { 81| 1.62k| return KMP_TEST_THEN_OR64(f, v); ------------------ | | 789| 1.62k| __sync_fetch_and_or((volatile kmp_uint64 *)(p), (kmp_uint64)(v)) ------------------ 82| 1.62k| } _ZN11flag_traitsIL9flag_type1EE13test_then_andEPVyy: 83| 1.62k| static inline flag_t test_then_and(volatile flag_t *f, flag_t v) { 84| 1.62k| return KMP_TEST_THEN_AND64(f, v); ------------------ | | 791| 1.62k| __sync_fetch_and_and((volatile kmp_uint64 *)(p), (kmp_uint64)(v)) ------------------ 85| 1.62k| } _ZN11kmp_flag_64ILb0ELb1EEC2EPVyy: 863| 148k| : kmp_flag_native(p, c) {} _ZN15kmp_flag_nativeIyL9flag_type1ELb1EEC2EPVyy: 154| 148k| : kmp_flag(), loc(p), checker(c) {} _ZN8kmp_flagIL9flag_type1EEC2Ev: 113| 223k| kmp_flag() : t({FlagType, 0U}), num_waiting_threads(0), sleepLoc(nullptr) {} _ZN11kmp_flag_64ILb0ELb1EE4waitEP8kmp_infoi: 879| 148k| int final_spin USE_ITT_BUILD_ARG(void *itt_sync_obj)) { 880| 148k| if (final_spin) ------------------ | Branch (880:9): [True: 75.0k, False: 73.2k] ------------------ 881| 75.0k| return __kmp_wait_template( 882| 75.0k| this_thr, this USE_ITT_BUILD_ARG(itt_sync_obj)); 883| 73.2k| else 884| 73.2k| return __kmp_wait_template( 885| 73.2k| this_thr, this USE_ITT_BUILD_ARG(itt_sync_obj)); 886| 148k| } kmp_barrier.cpp:_ZL19__kmp_wait_templateI11kmp_flag_64ILb0ELb1EELb1ELb0ELb1EEbP8kmp_infoPT_: 367| 75.1k| C *flag USE_ITT_BUILD_ARG(void *itt_sync_obj)) { 368| |#if USE_ITT_BUILD && USE_ITT_NOTIFY 369| | volatile void *spin = flag->get(); 370| |#endif 371| 75.1k| kmp_uint32 spins; 372| 75.1k| int th_gtid; 373| 75.1k| int tasks_completed = FALSE; ------------------ | | 1279| 75.1k|#define FALSE 0 ------------------ 374| 75.1k|#if !KMP_USE_MONITOR 375| 75.1k| kmp_uint64 poll_count; 376| 75.1k| kmp_uint64 hibernate_goal; 377| |#else 378| | kmp_uint32 hibernate; 379| |#endif 380| 75.1k| kmp_uint64 time; 381| | 382| 75.1k| KMP_FSYNC_SPIN_INIT(spin, NULL); ------------------ | | 337| 75.1k|#define KMP_FSYNC_SPIN_INIT(obj, spin) ((void)0) ------------------ 383| 75.1k| if (flag->done_check()) { ------------------ | Branch (383:7): [True: 58, False: 75.0k] ------------------ 384| 58| KMP_FSYNC_SPIN_ACQUIRED(CCAST(void *, spin)); ------------------ | | 339| 58|#define KMP_FSYNC_SPIN_ACQUIRED(obj) ((void)0) ------------------ 385| 58| return false; 386| 58| } 387| 75.0k| th_gtid = this_thr->th.th_info.ds.ds_gtid; 388| 75.0k| if (Cancellable) { ------------------ | Branch (388:7): [Folded, False: 75.0k] ------------------ 389| 0| kmp_team_t *team = this_thr->th.th_team; 390| 0| if (team && team->t.t_cancel_request == cancel_parallel) ------------------ | Branch (390:9): [True: 0, False: 0] | Branch (390:17): [True: 0, False: 0] ------------------ 391| 0| return true; 392| 0| } 393| 75.0k|#if KMP_OS_UNIX 394| 75.0k| if (final_spin) ------------------ | Branch (394:7): [True: 74.9k, Folded] ------------------ 395| 75.0k| KMP_ATOMIC_ST_REL(&this_thr->th.th_blocking, true); ------------------ | | 1251| 74.9k|#define KMP_ATOMIC_ST_REL(p, v) KMP_ATOMIC_OP(store, p, v, release) | | ------------------ | | | | 1246| 74.9k|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 396| 75.0k|#endif 397| 75.0k| KA_TRACE(20, 398| 75.0k| ("__kmp_wait_sleep: T#%d waiting for flag(%p)\n", th_gtid, flag)); 399| |#if KMP_STATS_ENABLED 400| | stats_state_e thread_state = KMP_GET_THREAD_STATE(); 401| |#endif 402| | 403| |/* OMPT Behavior: 404| |THIS function is called from 405| | __kmp_barrier (2 times) (implicit or explicit barrier in parallel regions) 406| | these have join / fork behavior 407| | 408| | In these cases, we don't change the state or trigger events in THIS 409| |function. 410| | Events are triggered in the calling code (__kmp_barrier): 411| | 412| | state := ompt_state_overhead 413| | barrier-begin 414| | barrier-wait-begin 415| | state := ompt_state_wait_barrier 416| | call join-barrier-implementation (finally arrive here) 417| | {} 418| | call fork-barrier-implementation (finally arrive here) 419| | {} 420| | state := ompt_state_overhead 421| | barrier-wait-end 422| | barrier-end 423| | state := ompt_state_work_parallel 424| | 425| | 426| | __kmp_fork_barrier (after thread creation, before executing implicit task) 427| | call fork-barrier-implementation (finally arrive here) 428| | {} // worker arrive here with state = ompt_state_idle 429| | 430| | 431| | __kmp_join_barrier (implicit barrier at end of parallel region) 432| | state := ompt_state_barrier_implicit 433| | barrier-begin 434| | barrier-wait-begin 435| | call join-barrier-implementation (finally arrive here 436| |final_spin=FALSE) 437| | { 438| | } 439| | __kmp_fork_barrier (implicit barrier at end of parallel region) 440| | call fork-barrier-implementation (finally arrive here final_spin=TRUE) 441| | 442| | Worker after task-team is finished: 443| | barrier-wait-end 444| | barrier-end 445| | implicit-task-end 446| | idle-begin 447| | state := ompt_state_idle 448| | 449| | Before leaving, if state = ompt_state_idle 450| | idle-end 451| | state := ompt_state_overhead 452| |*/ 453| 75.0k|#if OMPT_SUPPORT 454| 75.0k| ompt_state_t ompt_entry_state; 455| 75.0k| ompt_data_t *tId; 456| 75.0k| if (ompt_enabled.enabled) { ------------------ | Branch (456:7): [True: 0, False: 75.0k] ------------------ 457| 0| ompt_entry_state = this_thr->th.ompt_thread_info.state; 458| 0| if (!final_spin || ompt_entry_state != ompt_state_wait_barrier_implicit || ------------------ | Branch (458:9): [Folded, False: 0] | Branch (458:24): [True: 0, False: 0] ------------------ 459| 0| KMP_MASTER_TID(this_thr->th.th_info.ds.ds_tid)) { ------------------ | | 1271| 0|#define KMP_MASTER_TID(tid) (0 == (tid)) | | ------------------ | | | Branch (1271:29): [True: 0, False: 0] | | ------------------ ------------------ 460| 0| ompt_lw_taskteam_t *team = NULL; 461| 0| if (this_thr->th.th_team) ------------------ | Branch (461:11): [True: 0, False: 0] ------------------ 462| 0| team = this_thr->th.th_team->t.ompt_serialized_team_info; 463| 0| if (team) { ------------------ | Branch (463:11): [True: 0, False: 0] ------------------ 464| 0| tId = &(team->ompt_task_info.task_data); 465| 0| } else { 466| 0| tId = OMPT_CUR_TASK_DATA(this_thr); ------------------ | | 70| 0| (&((thr)->th.th_current_task->ompt_task_info.task_data)) ------------------ 467| 0| } 468| 0| } else { 469| 0| tId = &(this_thr->th.ompt_thread_info.task_data); 470| 0| } 471| 0| if (final_spin && (__kmp_tasking_mode == tskm_immediate_exec || ------------------ | Branch (471:9): [True: 0, Folded] | Branch (471:24): [True: 0, False: 0] ------------------ 472| 0| this_thr->th.th_task_team == NULL)) { ------------------ | Branch (472:24): [True: 0, False: 0] ------------------ 473| | // implicit task is done. Either no taskqueue, or task-team finished 474| 0| __ompt_implicit_task_end(this_thr, ompt_entry_state, tId); 475| 0| } 476| 0| } 477| 75.0k|#endif 478| | 479| 75.0k| KMP_INIT_YIELD(spins); // Setup for waiting ------------------ | | 1514| 75.0k| { (count) = __kmp_yield_init; } ------------------ 480| 75.0k| KMP_INIT_BACKOFF(time); ------------------ | | 1517| 75.0k| { (time) = __kmp_pause_init; } ------------------ 481| | 482| 75.0k| if (__kmp_dflt_blocktime != KMP_MAX_BLOCKTIME || ------------------ | | 1184| 150k| (INT_MAX) /* Must be this for "infinite" setting the work */ ------------------ | Branch (482:7): [True: 75.0k, False: 0] ------------------ 483| 75.0k| __kmp_pause_status == kmp_soft_paused) { ------------------ | Branch (483:7): [True: 0, False: 0] ------------------ 484| |#if KMP_USE_MONITOR 485| |// The worker threads cannot rely on the team struct existing at this point. 486| |// Use the bt values cached in the thread struct instead. 487| |#ifdef KMP_ADJUST_BLOCKTIME 488| | if (__kmp_pause_status == kmp_soft_paused || 489| | (__kmp_zero_bt && !this_thr->th.th_team_bt_set)) 490| | // Force immediate suspend if not set by user and more threads than 491| | // available procs 492| | hibernate = 0; 493| | else 494| | hibernate = this_thr->th.th_team_bt_intervals; 495| |#else 496| | hibernate = this_thr->th.th_team_bt_intervals; 497| |#endif /* KMP_ADJUST_BLOCKTIME */ 498| | 499| | /* If the blocktime is nonzero, we want to make sure that we spin wait for 500| | the entirety of the specified #intervals, plus up to one interval more. 501| | This increment make certain that this thread doesn't go to sleep too 502| | soon. */ 503| | if (hibernate != 0) 504| | hibernate++; 505| | 506| | // Add in the current time value. 507| | hibernate += TCR_4(__kmp_global.g.g_time.dt.t_value); 508| | KF_TRACE(20, ("__kmp_wait_sleep: T#%d now=%d, hibernate=%d, intervals=%d\n", 509| | th_gtid, __kmp_global.g.g_time.dt.t_value, hibernate, 510| | hibernate - __kmp_global.g.g_time.dt.t_value)); 511| |#else 512| 75.0k| if (__kmp_pause_status == kmp_soft_paused) { ------------------ | Branch (512:9): [True: 0, False: 75.0k] ------------------ 513| | // Force immediate suspend 514| 0| hibernate_goal = KMP_NOW(); ------------------ | | 1217| 0|#define KMP_NOW() __kmp_hardware_timestamp() ------------------ 515| 0| } else 516| 75.0k| hibernate_goal = KMP_NOW() + this_thr->th.th_team_bt_intervals; ------------------ | | 1217| 75.0k|#define KMP_NOW() __kmp_hardware_timestamp() ------------------ 517| 75.0k| poll_count = 0; 518| 75.0k| (void)poll_count; 519| 75.0k|#endif // KMP_USE_MONITOR 520| 75.0k| } 521| | 522| 75.0k| KMP_MB(); 523| | 524| | // Main wait spin loop 525| 370M| while (flag->notdone_check()) { ------------------ | Branch (525:10): [True: 370M, False: 75.0k] ------------------ 526| 370M| kmp_task_team_t *task_team = NULL; 527| 370M| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (527:9): [True: 370M, False: 18.4E] ------------------ 528| 370M| task_team = this_thr->th.th_task_team; 529| | /* If the thread's task team pointer is NULL, it means one of 3 things: 530| | 1) A newly-created thread is first being released by 531| | __kmp_fork_barrier(), and its task team has not been set up yet. 532| | 2) All tasks have been executed to completion. 533| | 3) Tasking is off for this region. This could be because we are in a 534| | serialized region (perhaps the outer one), or else tasking was manually 535| | disabled (KMP_TASKING=0). */ 536| 370M| if (task_team != NULL) { ------------------ | Branch (536:11): [True: 368M, False: 1.86M] ------------------ 537| 369M| if (TCR_SYNC_4(task_team->tt.tt_active)) { ------------------ | | 1135| 368M|#define TCR_SYNC_4(a) (a) | | ------------------ | | | Branch (1135:23): [True: 369M, False: 18.4E] | | ------------------ ------------------ 538| 369M| if (KMP_TASKING_ENABLED(task_team)) { ------------------ | | 2368| 369M| (TRUE == TCR_SYNC_4((task_team)->tt.tt_found_tasks)) | | ------------------ | | | | 1280| 369M|#define TRUE (!FALSE) | | | | ------------------ | | | | | | 1279| 369M|#define FALSE 0 | | | | ------------------ | | ------------------ | | (TRUE == TCR_SYNC_4((task_team)->tt.tt_found_tasks)) | | ------------------ | | | | 1135| 369M|#define TCR_SYNC_4(a) (a) | | ------------------ | | | Branch (2368:3): [True: 0, False: 369M] | | ------------------ ------------------ 539| 0| flag->execute_tasks( 540| 0| this_thr, th_gtid, final_spin, 541| 0| &tasks_completed USE_ITT_BUILD_ARG(itt_sync_obj), 0); 542| 0| } else 543| 369M| this_thr->th.th_reap_state = KMP_SAFE_TO_REAP; ------------------ | | 2047| 369M|#define KMP_SAFE_TO_REAP 1 // Thread th_reap_state: safe to reap (not tasking) ------------------ 544| 18.4E| } else { 545| 18.4E| KMP_DEBUG_ASSERT(!KMP_MASTER_TID(this_thr->th.th_info.ds.ds_tid)); 546| 18.4E|#if OMPT_SUPPORT 547| | // task-team is done now, other cases should be catched above 548| 18.4E| if (final_spin && ompt_enabled.enabled) ------------------ | Branch (548:15): [True: 0, Folded] | Branch (548:29): [True: 0, False: 0] ------------------ 549| 0| __ompt_implicit_task_end(this_thr, ompt_entry_state, tId); 550| 18.4E|#endif 551| 18.4E| this_thr->th.th_task_team = NULL; 552| 18.4E| this_thr->th.th_reap_state = KMP_SAFE_TO_REAP; ------------------ | | 2047| 18.4E|#define KMP_SAFE_TO_REAP 1 // Thread th_reap_state: safe to reap (not tasking) ------------------ 553| 18.4E| } 554| 368M| } else { 555| 1.86M| this_thr->th.th_reap_state = KMP_SAFE_TO_REAP; ------------------ | | 2047| 1.86M|#define KMP_SAFE_TO_REAP 1 // Thread th_reap_state: safe to reap (not tasking) ------------------ 556| 1.86M| } // if 557| 370M| } // if 558| | 559| 370M| KMP_FSYNC_SPIN_PREPARE(CCAST(void *, spin)); ------------------ | | 338| 370M|#define KMP_FSYNC_SPIN_PREPARE(obj) ((void)0) ------------------ 560| 370M| if (TCR_4(__kmp_global.g.g_done)) { ------------------ | | 1127| 370M|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 370M] | | ------------------ ------------------ 561| 0| if (__kmp_global.g.g_abort) ------------------ | Branch (561:11): [True: 0, False: 0] ------------------ 562| 0| __kmp_abort_thread(); 563| 0| break; 564| 0| } 565| | 566| | // If we are oversubscribed, or have waited a bit (and 567| | // KMP_LIBRARY=throughput), then yield 568| 370M| KMP_YIELD_OVERSUB_ELSE_SPIN(spins, time); ------------------ | | 1564| 370M| { \ | | 1565| 370M| if (__kmp_tpause_enabled) { \ | | ------------------ | | | Branch (1565:9): [True: 0, False: 370M] | | ------------------ | | 1566| 0| if (KMP_OVERSUBSCRIBED) { \ | | ------------------ | | | | 1520| 0| (TCR_4(__kmp_nth) > (__kmp_avail_proc ? __kmp_avail_proc : __kmp_xproc)) | | | | ------------------ | | | | | | 1127| 0|#define TCR_4(a) (a) | | | | ------------------ | | | | | Branch (1520:3): [True: 0, False: 0] | | | | | Branch (1520:24): [True: 0, False: 0] | | | | ------------------ | | ------------------ | | 1567| 0| __kmp_tpause(0, (time)); \ | | 1568| 0| } else { \ | | 1569| 0| __kmp_tpause(__kmp_tpause_hint, (time)); \ | | 1570| 0| } \ | | 1571| 0| (time) = (time << 1 | 1) & KMP_TPAUSE_MAX_MASK; \ | | ------------------ | | | | 1562| 0|#define KMP_TPAUSE_MAX_MASK ((kmp_uint64)0xFFFF) | | ------------------ | | 1572| 370M| } else { \ | | 1573| 370M| KMP_CPU_PAUSE(); \ | | ------------------ | | | | 1498| 370M|#define KMP_CPU_PAUSE() __kmp_x86_pause() | | ------------------ | | 1574| 370M| if ((KMP_TRY_YIELD_OVERSUB)) { \ | | ------------------ | | | | 1526| 372M| ((__kmp_use_yield == 1 || __kmp_use_yield == 2) && (KMP_OVERSUBSCRIBED)) | | | | ------------------ | | | | | | 1520| 18.4E| (TCR_4(__kmp_nth) > (__kmp_avail_proc ? __kmp_avail_proc : __kmp_xproc)) | | | | | | ------------------ | | | | | | | | 1127| 372M|#define TCR_4(a) (a) | | | | | | ------------------ | | | | | | | Branch (1520:24): [True: 373M, False: 18.4E] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (1526:5): [True: 370M, False: 0] | | | | | Branch (1526:29): [True: 0, False: 0] | | | | | Branch (1526:54): [True: 0, False: 372M] | | | | ------------------ | | ------------------ | | 1575| 0| __kmp_yield(); \ | | 1576| 371M| } else if (__kmp_use_yield == 1) { \ | | ------------------ | | | Branch (1576:18): [True: 371M, False: 18.4E] | | ------------------ | | 1577| 371M| (count) -= 2; \ | | 1578| 371M| if (!(count)) { \ | | ------------------ | | | Branch (1578:13): [True: 1.67M, False: 369M] | | ------------------ | | 1579| 1.67M| __kmp_yield(); \ | | 1580| 1.67M| (count) = __kmp_yield_next; \ | | 1581| 1.67M| } \ | | 1582| 371M| } \ | | 1583| 370M| } \ | | 1584| 370M| } ------------------ 569| | 570| |#if KMP_STATS_ENABLED 571| | // Check if thread has been signalled to idle state 572| | // This indicates that the logical "join-barrier" has finished 573| | if (this_thr->th.th_stats->isIdle() && 574| | KMP_GET_THREAD_STATE() == FORK_JOIN_BARRIER) { 575| | KMP_SET_THREAD_STATE(IDLE); 576| | KMP_PUSH_PARTITIONED_TIMER(OMP_idle); 577| | } 578| |#endif 579| | // Check if the barrier surrounding this wait loop has been cancelled 580| 370M| if (Cancellable) { ------------------ | Branch (580:9): [Folded, False: 370M] ------------------ 581| 0| kmp_team_t *team = this_thr->th.th_team; 582| 0| if (team && team->t.t_cancel_request == cancel_parallel) ------------------ | Branch (582:11): [True: 0, False: 0] | Branch (582:19): [True: 0, False: 0] ------------------ 583| 0| break; 584| 0| } 585| | 586| | // For hidden helper thread, if task_team is nullptr, it means the main 587| | // thread has not released the barrier. We cannot wait here because once the 588| | // main thread releases all children barriers, all hidden helper threads are 589| | // still sleeping. This leads to a problem that following configuration, 590| | // such as task team sync, will not be performed such that this thread does 591| | // not have task team. Usually it is not bad. However, a corner case is, 592| | // when the first task encountered is an untied task, the check in 593| | // __kmp_task_alloc will crash because it uses the task team pointer without 594| | // checking whether it is nullptr. It is probably under some kind of 595| | // assumption. 596| 373M| if (task_team && KMP_HIDDEN_HELPER_WORKER_THREAD(th_gtid) && ------------------ | | 4420| 743M| ((gtid) > 1 && (gtid) <= __kmp_hidden_helper_threads_num) | | ------------------ | | | Branch (4420:4): [True: 368M, False: 4.46M] | | | Branch (4420:18): [True: 0, False: 368M] | | ------------------ ------------------ | Branch (596:9): [True: 373M, False: 18.4E] ------------------ 597| 0| !TCR_4(__kmp_hidden_helper_team_done)) { ------------------ | | 1127| 0|#define TCR_4(a) (a) ------------------ | Branch (597:9): [True: 0, False: 0] ------------------ 598| | // If there is still hidden helper tasks to be executed, the hidden helper 599| | // thread will not enter a waiting status. 600| 0| if (KMP_ATOMIC_LD_ACQ(&__kmp_unexecuted_hidden_helper_tasks) == 0) { ------------------ | | 1249| 0|#define KMP_ATOMIC_LD_ACQ(p) KMP_ATOMIC_LD(p, acquire) | | ------------------ | | | | 1245| 0|#define KMP_ATOMIC_LD(p, order) (p)->load(std::memory_order_##order) | | ------------------ ------------------ | Branch (600:11): [True: 0, False: 0] ------------------ 601| 0| __kmp_hidden_helper_worker_thread_wait(); 602| 0| } 603| 0| continue; 604| 0| } 605| | 606| | // Don't suspend if KMP_BLOCKTIME is set to "infinite" 607| 370M| if (__kmp_dflt_blocktime == KMP_MAX_BLOCKTIME && ------------------ | | 1184| 741M| (INT_MAX) /* Must be this for "infinite" setting the work */ ------------------ | Branch (607:9): [True: 0, False: 370M] ------------------ 608| 0| __kmp_pause_status != kmp_soft_paused) ------------------ | Branch (608:9): [True: 0, False: 0] ------------------ 609| 0| continue; 610| | 611| | // Don't suspend if there is a likelihood of new tasks being spawned. 612| 370M| if (task_team != NULL && TCR_4(task_team->tt.tt_found_tasks) && ------------------ | | 1127| 739M|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 369M] | | ------------------ ------------------ | Branch (612:9): [True: 369M, False: 1.27M] ------------------ 613| 0| !__kmp_wpolicy_passive) ------------------ | Branch (613:9): [True: 0, False: 0] ------------------ 614| 0| continue; 615| | 616| |#if KMP_USE_MONITOR 617| | // If we have waited a bit more, fall asleep 618| | if (TCR_4(__kmp_global.g.g_time.dt.t_value) < hibernate) 619| | continue; 620| |#else 621| 370M| if (KMP_BLOCKING(hibernate_goal, poll_count++)) ------------------ | | 1222| 370M|#define KMP_BLOCKING(goal, count) ((goal) > KMP_NOW()) | | ------------------ | | | | 1217| 370M|#define KMP_NOW() __kmp_hardware_timestamp() | | ------------------ | | | Branch (1222:35): [True: 361M, False: 8.82M] | | ------------------ ------------------ 622| 361M| continue; 623| 8.82M|#endif 624| | // Don't suspend if wait loop designated non-sleepable 625| | // in template parameters 626| 8.82M| if (!Sleepable) ------------------ | Branch (626:9): [Folded, False: 8.82M] ------------------ 627| 0| continue; 628| | 629| 8.82M|#if KMP_HAVE_MWAIT || KMP_HAVE_UMWAIT 630| 8.82M| if (__kmp_mwait_enabled || __kmp_umwait_enabled) { ------------------ | Branch (630:9): [True: 8.82M, False: 1.60k] | Branch (630:32): [True: 0, False: 1.60k] ------------------ 631| 0| KF_TRACE(50, ("__kmp_wait_sleep: T#%d using monitor/mwait\n", th_gtid)); 632| 0| flag->mwait(th_gtid); 633| 8.82M| } else { 634| 8.82M|#endif 635| 8.82M| KF_TRACE(50, ("__kmp_wait_sleep: T#%d suspend time reached\n", th_gtid)); 636| 8.82M|#if KMP_OS_UNIX 637| 8.82M| if (final_spin) ------------------ | Branch (637:11): [True: 1.60k, Folded] ------------------ 638| 8.82M| KMP_ATOMIC_ST_REL(&this_thr->th.th_blocking, false); ------------------ | | 1251| 1.60k|#define KMP_ATOMIC_ST_REL(p, v) KMP_ATOMIC_OP(store, p, v, release) | | ------------------ | | | | 1246| 1.60k|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 639| 8.82M|#endif 640| 8.82M| flag->suspend(th_gtid); 641| 8.82M|#if KMP_OS_UNIX 642| 8.82M| if (final_spin) ------------------ | Branch (642:11): [True: 1.60k, Folded] ------------------ 643| 8.82M| KMP_ATOMIC_ST_REL(&this_thr->th.th_blocking, true); ------------------ | | 1251| 1.60k|#define KMP_ATOMIC_ST_REL(p, v) KMP_ATOMIC_OP(store, p, v, release) | | ------------------ | | | | 1246| 1.60k|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 644| 8.82M|#endif 645| 8.82M|#if KMP_HAVE_MWAIT || KMP_HAVE_UMWAIT 646| 8.82M| } 647| 8.82M|#endif 648| | 649| 8.82M| if (TCR_4(__kmp_global.g.g_done)) { ------------------ | | 1127| 8.82M|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 31, False: 8.82M] | | ------------------ ------------------ 650| 31| if (__kmp_global.g.g_abort) ------------------ | Branch (650:11): [True: 0, False: 31] ------------------ 651| 0| __kmp_abort_thread(); 652| 31| break; 653| 8.82M| } else if (__kmp_tasking_mode != tskm_immediate_exec && ------------------ | Branch (653:16): [True: 1.57k, False: 8.82M] ------------------ 654| 1.57k| this_thr->th.th_reap_state == KMP_SAFE_TO_REAP) { ------------------ | | 2047| 1.57k|#define KMP_SAFE_TO_REAP 1 // Thread th_reap_state: safe to reap (not tasking) ------------------ | Branch (654:16): [True: 1.56k, False: 8] ------------------ 655| 1.56k| this_thr->th.th_reap_state = KMP_NOT_SAFE_TO_REAP; ------------------ | | 2046| 1.56k| 0 // Thread th_reap_state: not safe to reap (tasking) ------------------ 656| 1.56k| } 657| | // TODO: If thread is done with work and times out, disband/free 658| 8.82M| } 659| | 660| 75.0k|#if OMPT_SUPPORT 661| 75.0k| ompt_state_t ompt_exit_state = this_thr->th.ompt_thread_info.state; 662| 75.0k| if (ompt_enabled.enabled && ompt_exit_state != ompt_state_undefined) { ------------------ | Branch (662:7): [True: 0, False: 75.0k] | Branch (662:31): [True: 0, False: 0] ------------------ 663| 0|#if OMPT_OPTIONAL 664| 0| if (final_spin) { ------------------ | Branch (664:9): [True: 0, Folded] ------------------ 665| 0| __ompt_implicit_task_end(this_thr, ompt_exit_state, tId); 666| 0| ompt_exit_state = this_thr->th.ompt_thread_info.state; 667| 0| } 668| 0|#endif 669| 0| if (ompt_exit_state == ompt_state_idle) { ------------------ | Branch (669:9): [True: 0, False: 0] ------------------ 670| 0| this_thr->th.ompt_thread_info.state = ompt_state_overhead; 671| 0| } 672| 0| } 673| 75.0k|#endif 674| |#if KMP_STATS_ENABLED 675| | // If we were put into idle state, pop that off the state stack 676| | if (KMP_GET_THREAD_STATE() == IDLE) { 677| | KMP_POP_PARTITIONED_TIMER(); 678| | KMP_SET_THREAD_STATE(thread_state); 679| | this_thr->th.th_stats->resetIdleFlag(); 680| | } 681| |#endif 682| | 683| 75.0k|#if KMP_OS_UNIX 684| 75.0k| if (final_spin) ------------------ | Branch (684:7): [True: 74.8k, Folded] ------------------ 685| 75.0k| KMP_ATOMIC_ST_REL(&this_thr->th.th_blocking, false); ------------------ | | 1251| 74.8k|#define KMP_ATOMIC_ST_REL(p, v) KMP_ATOMIC_OP(store, p, v, release) | | ------------------ | | | | 1246| 74.8k|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 686| 75.0k|#endif 687| 75.0k| KMP_FSYNC_SPIN_ACQUIRED(CCAST(void *, spin)); ------------------ | | 339| 75.0k|#define KMP_FSYNC_SPIN_ACQUIRED(obj) ((void)0) ------------------ 688| 75.0k| if (Cancellable) { ------------------ | Branch (688:7): [Folded, False: 75.0k] ------------------ 689| 0| kmp_team_t *team = this_thr->th.th_team; 690| 0| if (team && team->t.t_cancel_request == cancel_parallel) { ------------------ | Branch (690:9): [True: 0, False: 0] | Branch (690:17): [True: 0, False: 0] ------------------ 691| 0| if (tasks_completed) { ------------------ | Branch (691:11): [True: 0, False: 0] ------------------ 692| | // undo the previous decrement of unfinished_threads so that the 693| | // thread can decrement at the join barrier with no problem 694| 0| kmp_task_team_t *task_team = this_thr->th.th_task_team; 695| 0| std::atomic *unfinished_threads = 696| 0| &(task_team->tt.tt_unfinished_threads); 697| 0| KMP_ATOMIC_INC(unfinished_threads); ------------------ | | 1259| 0|#define KMP_ATOMIC_INC(p) KMP_ATOMIC_OP(fetch_add, p, 1, acq_rel) | | ------------------ | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 698| 0| } 699| 0| return true; 700| 0| } 701| 0| } 702| 75.0k| return false; 703| 75.0k|} _ZN11kmp_flag_64ILb0ELb1EE7suspendEi: 866| 1.61k| void suspend(int th_gtid) { __kmp_suspend_64(th_gtid, this); } kmp_barrier.cpp:_ZL19__kmp_wait_templateI11kmp_flag_64ILb0ELb1EELb0ELb0ELb1EEbP8kmp_infoPT_: 367| 74.9k| C *flag USE_ITT_BUILD_ARG(void *itt_sync_obj)) { 368| |#if USE_ITT_BUILD && USE_ITT_NOTIFY 369| | volatile void *spin = flag->get(); 370| |#endif 371| 74.9k| kmp_uint32 spins; 372| 74.9k| int th_gtid; 373| 74.9k| int tasks_completed = FALSE; ------------------ | | 1279| 74.9k|#define FALSE 0 ------------------ 374| 74.9k|#if !KMP_USE_MONITOR 375| 74.9k| kmp_uint64 poll_count; 376| 74.9k| kmp_uint64 hibernate_goal; 377| |#else 378| | kmp_uint32 hibernate; 379| |#endif 380| 74.9k| kmp_uint64 time; 381| | 382| 74.9k| KMP_FSYNC_SPIN_INIT(spin, NULL); ------------------ | | 337| 74.9k|#define KMP_FSYNC_SPIN_INIT(obj, spin) ((void)0) ------------------ 383| 74.9k| if (flag->done_check()) { ------------------ | Branch (383:7): [True: 38.5k, False: 36.3k] ------------------ 384| 38.5k| KMP_FSYNC_SPIN_ACQUIRED(CCAST(void *, spin)); ------------------ | | 339| 38.5k|#define KMP_FSYNC_SPIN_ACQUIRED(obj) ((void)0) ------------------ 385| 38.5k| return false; 386| 38.5k| } 387| 36.3k| th_gtid = this_thr->th.th_info.ds.ds_gtid; 388| 36.3k| if (Cancellable) { ------------------ | Branch (388:7): [Folded, False: 36.3k] ------------------ 389| 0| kmp_team_t *team = this_thr->th.th_team; 390| 0| if (team && team->t.t_cancel_request == cancel_parallel) ------------------ | Branch (390:9): [True: 0, False: 0] | Branch (390:17): [True: 0, False: 0] ------------------ 391| 0| return true; 392| 0| } 393| 36.3k|#if KMP_OS_UNIX 394| 36.3k| if (final_spin) ------------------ | Branch (394:7): [Folded, False: 36.3k] ------------------ 395| 36.3k| KMP_ATOMIC_ST_REL(&this_thr->th.th_blocking, true); ------------------ | | 1251| 0|#define KMP_ATOMIC_ST_REL(p, v) KMP_ATOMIC_OP(store, p, v, release) | | ------------------ | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 396| 36.3k|#endif 397| 36.3k| KA_TRACE(20, 398| 36.3k| ("__kmp_wait_sleep: T#%d waiting for flag(%p)\n", th_gtid, flag)); 399| |#if KMP_STATS_ENABLED 400| | stats_state_e thread_state = KMP_GET_THREAD_STATE(); 401| |#endif 402| | 403| |/* OMPT Behavior: 404| |THIS function is called from 405| | __kmp_barrier (2 times) (implicit or explicit barrier in parallel regions) 406| | these have join / fork behavior 407| | 408| | In these cases, we don't change the state or trigger events in THIS 409| |function. 410| | Events are triggered in the calling code (__kmp_barrier): 411| | 412| | state := ompt_state_overhead 413| | barrier-begin 414| | barrier-wait-begin 415| | state := ompt_state_wait_barrier 416| | call join-barrier-implementation (finally arrive here) 417| | {} 418| | call fork-barrier-implementation (finally arrive here) 419| | {} 420| | state := ompt_state_overhead 421| | barrier-wait-end 422| | barrier-end 423| | state := ompt_state_work_parallel 424| | 425| | 426| | __kmp_fork_barrier (after thread creation, before executing implicit task) 427| | call fork-barrier-implementation (finally arrive here) 428| | {} // worker arrive here with state = ompt_state_idle 429| | 430| | 431| | __kmp_join_barrier (implicit barrier at end of parallel region) 432| | state := ompt_state_barrier_implicit 433| | barrier-begin 434| | barrier-wait-begin 435| | call join-barrier-implementation (finally arrive here 436| |final_spin=FALSE) 437| | { 438| | } 439| | __kmp_fork_barrier (implicit barrier at end of parallel region) 440| | call fork-barrier-implementation (finally arrive here final_spin=TRUE) 441| | 442| | Worker after task-team is finished: 443| | barrier-wait-end 444| | barrier-end 445| | implicit-task-end 446| | idle-begin 447| | state := ompt_state_idle 448| | 449| | Before leaving, if state = ompt_state_idle 450| | idle-end 451| | state := ompt_state_overhead 452| |*/ 453| 36.3k|#if OMPT_SUPPORT 454| 36.3k| ompt_state_t ompt_entry_state; 455| 36.3k| ompt_data_t *tId; 456| 36.3k| if (ompt_enabled.enabled) { ------------------ | Branch (456:7): [True: 0, False: 36.3k] ------------------ 457| 0| ompt_entry_state = this_thr->th.ompt_thread_info.state; 458| 0| if (!final_spin || ompt_entry_state != ompt_state_wait_barrier_implicit || ------------------ | Branch (458:9): [True: 0, Folded] | Branch (458:24): [True: 0, False: 0] ------------------ 459| 0| KMP_MASTER_TID(this_thr->th.th_info.ds.ds_tid)) { ------------------ | | 1271| 0|#define KMP_MASTER_TID(tid) (0 == (tid)) | | ------------------ | | | Branch (1271:29): [True: 0, False: 0] | | ------------------ ------------------ 460| 0| ompt_lw_taskteam_t *team = NULL; 461| 0| if (this_thr->th.th_team) ------------------ | Branch (461:11): [True: 0, False: 0] ------------------ 462| 0| team = this_thr->th.th_team->t.ompt_serialized_team_info; 463| 0| if (team) { ------------------ | Branch (463:11): [True: 0, False: 0] ------------------ 464| 0| tId = &(team->ompt_task_info.task_data); 465| 0| } else { 466| 0| tId = OMPT_CUR_TASK_DATA(this_thr); ------------------ | | 70| 0| (&((thr)->th.th_current_task->ompt_task_info.task_data)) ------------------ 467| 0| } 468| 0| } else { 469| 0| tId = &(this_thr->th.ompt_thread_info.task_data); 470| 0| } 471| 0| if (final_spin && (__kmp_tasking_mode == tskm_immediate_exec || ------------------ | Branch (471:9): [Folded, False: 0] | Branch (471:24): [True: 0, False: 0] ------------------ 472| 0| this_thr->th.th_task_team == NULL)) { ------------------ | Branch (472:24): [True: 0, False: 0] ------------------ 473| | // implicit task is done. Either no taskqueue, or task-team finished 474| 0| __ompt_implicit_task_end(this_thr, ompt_entry_state, tId); 475| 0| } 476| 0| } 477| 36.3k|#endif 478| | 479| 36.3k| KMP_INIT_YIELD(spins); // Setup for waiting ------------------ | | 1514| 36.3k| { (count) = __kmp_yield_init; } ------------------ 480| 36.3k| KMP_INIT_BACKOFF(time); ------------------ | | 1517| 36.3k| { (time) = __kmp_pause_init; } ------------------ 481| | 482| 36.3k| if (__kmp_dflt_blocktime != KMP_MAX_BLOCKTIME || ------------------ | | 1184| 72.7k| (INT_MAX) /* Must be this for "infinite" setting the work */ ------------------ | Branch (482:7): [True: 36.3k, False: 0] ------------------ 483| 36.4k| __kmp_pause_status == kmp_soft_paused) { ------------------ | Branch (483:7): [True: 0, False: 0] ------------------ 484| |#if KMP_USE_MONITOR 485| |// The worker threads cannot rely on the team struct existing at this point. 486| |// Use the bt values cached in the thread struct instead. 487| |#ifdef KMP_ADJUST_BLOCKTIME 488| | if (__kmp_pause_status == kmp_soft_paused || 489| | (__kmp_zero_bt && !this_thr->th.th_team_bt_set)) 490| | // Force immediate suspend if not set by user and more threads than 491| | // available procs 492| | hibernate = 0; 493| | else 494| | hibernate = this_thr->th.th_team_bt_intervals; 495| |#else 496| | hibernate = this_thr->th.th_team_bt_intervals; 497| |#endif /* KMP_ADJUST_BLOCKTIME */ 498| | 499| | /* If the blocktime is nonzero, we want to make sure that we spin wait for 500| | the entirety of the specified #intervals, plus up to one interval more. 501| | This increment make certain that this thread doesn't go to sleep too 502| | soon. */ 503| | if (hibernate != 0) 504| | hibernate++; 505| | 506| | // Add in the current time value. 507| | hibernate += TCR_4(__kmp_global.g.g_time.dt.t_value); 508| | KF_TRACE(20, ("__kmp_wait_sleep: T#%d now=%d, hibernate=%d, intervals=%d\n", 509| | th_gtid, __kmp_global.g.g_time.dt.t_value, hibernate, 510| | hibernate - __kmp_global.g.g_time.dt.t_value)); 511| |#else 512| 36.4k| if (__kmp_pause_status == kmp_soft_paused) { ------------------ | Branch (512:9): [True: 0, False: 36.4k] ------------------ 513| | // Force immediate suspend 514| 0| hibernate_goal = KMP_NOW(); ------------------ | | 1217| 0|#define KMP_NOW() __kmp_hardware_timestamp() ------------------ 515| 0| } else 516| 36.4k| hibernate_goal = KMP_NOW() + this_thr->th.th_team_bt_intervals; ------------------ | | 1217| 36.4k|#define KMP_NOW() __kmp_hardware_timestamp() ------------------ 517| 36.4k| poll_count = 0; 518| 36.4k| (void)poll_count; 519| 36.4k|#endif // KMP_USE_MONITOR 520| 36.4k| } 521| | 522| 36.3k| KMP_MB(); 523| | 524| | // Main wait spin loop 525| 88.0M| while (flag->notdone_check()) { ------------------ | Branch (525:10): [True: 87.9M, False: 36.3k] ------------------ 526| 87.9M| kmp_task_team_t *task_team = NULL; 527| 87.9M| if (__kmp_tasking_mode != tskm_immediate_exec) { ------------------ | Branch (527:9): [True: 87.9M, False: 18.4E] ------------------ 528| 87.9M| task_team = this_thr->th.th_task_team; 529| | /* If the thread's task team pointer is NULL, it means one of 3 things: 530| | 1) A newly-created thread is first being released by 531| | __kmp_fork_barrier(), and its task team has not been set up yet. 532| | 2) All tasks have been executed to completion. 533| | 3) Tasking is off for this region. This could be because we are in a 534| | serialized region (perhaps the outer one), or else tasking was manually 535| | disabled (KMP_TASKING=0). */ 536| 88.0M| if (task_team != NULL) { ------------------ | Branch (536:11): [True: 88.0M, False: 18.4E] ------------------ 537| 88.0M| if (TCR_SYNC_4(task_team->tt.tt_active)) { ------------------ | | 1135| 88.0M|#define TCR_SYNC_4(a) (a) | | ------------------ | | | Branch (1135:23): [True: 87.9M, False: 8.47k] | | ------------------ ------------------ 538| 87.9M| if (KMP_TASKING_ENABLED(task_team)) { ------------------ | | 2368| 87.9M| (TRUE == TCR_SYNC_4((task_team)->tt.tt_found_tasks)) | | ------------------ | | | | 1280| 87.9M|#define TRUE (!FALSE) | | | | ------------------ | | | | | | 1279| 87.9M|#define FALSE 0 | | | | ------------------ | | ------------------ | | (TRUE == TCR_SYNC_4((task_team)->tt.tt_found_tasks)) | | ------------------ | | | | 1135| 87.9M|#define TCR_SYNC_4(a) (a) | | ------------------ | | | Branch (2368:3): [True: 0, False: 87.9M] | | ------------------ ------------------ 539| 0| flag->execute_tasks( 540| 0| this_thr, th_gtid, final_spin, 541| 0| &tasks_completed USE_ITT_BUILD_ARG(itt_sync_obj), 0); 542| 0| } else 543| 87.9M| this_thr->th.th_reap_state = KMP_SAFE_TO_REAP; ------------------ | | 2047| 87.9M|#define KMP_SAFE_TO_REAP 1 // Thread th_reap_state: safe to reap (not tasking) ------------------ 544| 87.9M| } else { 545| 8.47k| KMP_DEBUG_ASSERT(!KMP_MASTER_TID(this_thr->th.th_info.ds.ds_tid)); 546| 8.47k|#if OMPT_SUPPORT 547| | // task-team is done now, other cases should be catched above 548| 8.47k| if (final_spin && ompt_enabled.enabled) ------------------ | Branch (548:15): [Folded, False: 8.47k] | Branch (548:29): [True: 0, False: 0] ------------------ 549| 0| __ompt_implicit_task_end(this_thr, ompt_entry_state, tId); 550| 8.47k|#endif 551| 8.47k| this_thr->th.th_task_team = NULL; 552| 8.47k| this_thr->th.th_reap_state = KMP_SAFE_TO_REAP; ------------------ | | 2047| 8.47k|#define KMP_SAFE_TO_REAP 1 // Thread th_reap_state: safe to reap (not tasking) ------------------ 553| 8.47k| } 554| 18.4E| } else { 555| 18.4E| this_thr->th.th_reap_state = KMP_SAFE_TO_REAP; ------------------ | | 2047| 18.4E|#define KMP_SAFE_TO_REAP 1 // Thread th_reap_state: safe to reap (not tasking) ------------------ 556| 18.4E| } // if 557| 87.9M| } // if 558| | 559| 87.9M| KMP_FSYNC_SPIN_PREPARE(CCAST(void *, spin)); ------------------ | | 338| 87.9M|#define KMP_FSYNC_SPIN_PREPARE(obj) ((void)0) ------------------ 560| 87.9M| if (TCR_4(__kmp_global.g.g_done)) { ------------------ | | 1127| 87.9M|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 87.9M] | | ------------------ ------------------ 561| 0| if (__kmp_global.g.g_abort) ------------------ | Branch (561:11): [True: 0, False: 0] ------------------ 562| 0| __kmp_abort_thread(); 563| 0| break; 564| 0| } 565| | 566| | // If we are oversubscribed, or have waited a bit (and 567| | // KMP_LIBRARY=throughput), then yield 568| 87.9M| KMP_YIELD_OVERSUB_ELSE_SPIN(spins, time); ------------------ | | 1564| 87.9M| { \ | | 1565| 87.9M| if (__kmp_tpause_enabled) { \ | | ------------------ | | | Branch (1565:9): [True: 0, False: 87.9M] | | ------------------ | | 1566| 0| if (KMP_OVERSUBSCRIBED) { \ | | ------------------ | | | | 1520| 0| (TCR_4(__kmp_nth) > (__kmp_avail_proc ? __kmp_avail_proc : __kmp_xproc)) | | | | ------------------ | | | | | | 1127| 0|#define TCR_4(a) (a) | | | | ------------------ | | | | | Branch (1520:3): [True: 0, False: 0] | | | | | Branch (1520:24): [True: 0, False: 0] | | | | ------------------ | | ------------------ | | 1567| 0| __kmp_tpause(0, (time)); \ | | 1568| 0| } else { \ | | 1569| 0| __kmp_tpause(__kmp_tpause_hint, (time)); \ | | 1570| 0| } \ | | 1571| 0| (time) = (time << 1 | 1) & KMP_TPAUSE_MAX_MASK; \ | | ------------------ | | | | 1562| 0|#define KMP_TPAUSE_MAX_MASK ((kmp_uint64)0xFFFF) | | ------------------ | | 1572| 87.9M| } else { \ | | 1573| 87.9M| KMP_CPU_PAUSE(); \ | | ------------------ | | | | 1498| 87.9M|#define KMP_CPU_PAUSE() __kmp_x86_pause() | | ------------------ | | 1574| 87.9M| if ((KMP_TRY_YIELD_OVERSUB)) { \ | | ------------------ | | | | 1526| 88.2M| ((__kmp_use_yield == 1 || __kmp_use_yield == 2) && (KMP_OVERSUBSCRIBED)) | | | | ------------------ | | | | | | 1520| 88.2M| (TCR_4(__kmp_nth) > (__kmp_avail_proc ? __kmp_avail_proc : __kmp_xproc)) | | | | | | ------------------ | | | | | | | | 1127| 88.2M|#define TCR_4(a) (a) | | | | | | ------------------ | | | | | | | Branch (1520:24): [True: 88.2M, False: 8.37k] | | | | | | ------------------ | | | | ------------------ | | | | | Branch (1526:5): [True: 87.9M, False: 0] | | | | | Branch (1526:29): [True: 0, False: 0] | | | | | Branch (1526:54): [True: 0, False: 88.2M] | | | | ------------------ | | ------------------ | | 1575| 0| __kmp_yield(); \ | | 1576| 88.1M| } else if (__kmp_use_yield == 1) { \ | | ------------------ | | | Branch (1576:18): [True: 88.1M, False: 18.4E] | | ------------------ | | 1577| 88.1M| (count) -= 2; \ | | 1578| 88.1M| if (!(count)) { \ | | ------------------ | | | Branch (1578:13): [True: 326k, False: 87.8M] | | ------------------ | | 1579| 326k| __kmp_yield(); \ | | 1580| 326k| (count) = __kmp_yield_next; \ | | 1581| 326k| } \ | | 1582| 88.1M| } \ | | 1583| 87.9M| } \ | | 1584| 87.9M| } ------------------ 569| | 570| |#if KMP_STATS_ENABLED 571| | // Check if thread has been signalled to idle state 572| | // This indicates that the logical "join-barrier" has finished 573| | if (this_thr->th.th_stats->isIdle() && 574| | KMP_GET_THREAD_STATE() == FORK_JOIN_BARRIER) { 575| | KMP_SET_THREAD_STATE(IDLE); 576| | KMP_PUSH_PARTITIONED_TIMER(OMP_idle); 577| | } 578| |#endif 579| | // Check if the barrier surrounding this wait loop has been cancelled 580| 87.9M| if (Cancellable) { ------------------ | Branch (580:9): [Folded, False: 87.9M] ------------------ 581| 0| kmp_team_t *team = this_thr->th.th_team; 582| 0| if (team && team->t.t_cancel_request == cancel_parallel) ------------------ | Branch (582:11): [True: 0, False: 0] | Branch (582:19): [True: 0, False: 0] ------------------ 583| 0| break; 584| 0| } 585| | 586| | // For hidden helper thread, if task_team is nullptr, it means the main 587| | // thread has not released the barrier. We cannot wait here because once the 588| | // main thread releases all children barriers, all hidden helper threads are 589| | // still sleeping. This leads to a problem that following configuration, 590| | // such as task team sync, will not be performed such that this thread does 591| | // not have task team. Usually it is not bad. However, a corner case is, 592| | // when the first task encountered is an untied task, the check in 593| | // __kmp_task_alloc will crash because it uses the task team pointer without 594| | // checking whether it is nullptr. It is probably under some kind of 595| | // assumption. 596| 88.1M| if (task_team && KMP_HIDDEN_HELPER_WORKER_THREAD(th_gtid) && ------------------ | | 4420| 176M| ((gtid) > 1 && (gtid) <= __kmp_hidden_helper_threads_num) | | ------------------ | | | Branch (4420:4): [True: 63.3M, False: 24.8M] | | | Branch (4420:18): [True: 0, False: 63.3M] | | ------------------ ------------------ | Branch (596:9): [True: 88.1M, False: 18.4E] ------------------ 597| 0| !TCR_4(__kmp_hidden_helper_team_done)) { ------------------ | | 1127| 0|#define TCR_4(a) (a) ------------------ | Branch (597:9): [True: 0, False: 0] ------------------ 598| | // If there is still hidden helper tasks to be executed, the hidden helper 599| | // thread will not enter a waiting status. 600| 0| if (KMP_ATOMIC_LD_ACQ(&__kmp_unexecuted_hidden_helper_tasks) == 0) { ------------------ | | 1249| 0|#define KMP_ATOMIC_LD_ACQ(p) KMP_ATOMIC_LD(p, acquire) | | ------------------ | | | | 1245| 0|#define KMP_ATOMIC_LD(p, order) (p)->load(std::memory_order_##order) | | ------------------ ------------------ | Branch (600:11): [True: 0, False: 0] ------------------ 601| 0| __kmp_hidden_helper_worker_thread_wait(); 602| 0| } 603| 0| continue; 604| 0| } 605| | 606| | // Don't suspend if KMP_BLOCKTIME is set to "infinite" 607| 87.9M| if (__kmp_dflt_blocktime == KMP_MAX_BLOCKTIME && ------------------ | | 1184| 175M| (INT_MAX) /* Must be this for "infinite" setting the work */ ------------------ | Branch (607:9): [True: 0, False: 87.9M] ------------------ 608| 0| __kmp_pause_status != kmp_soft_paused) ------------------ | Branch (608:9): [True: 0, False: 0] ------------------ 609| 0| continue; 610| | 611| | // Don't suspend if there is a likelihood of new tasks being spawned. 612| 88.1M| if (task_team != NULL && TCR_4(task_team->tt.tt_found_tasks) && ------------------ | | 1127| 176M|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 88.1M] | | ------------------ ------------------ | Branch (612:9): [True: 88.1M, False: 18.4E] ------------------ 613| 0| !__kmp_wpolicy_passive) ------------------ | Branch (613:9): [True: 0, False: 0] ------------------ 614| 0| continue; 615| | 616| |#if KMP_USE_MONITOR 617| | // If we have waited a bit more, fall asleep 618| | if (TCR_4(__kmp_global.g.g_time.dt.t_value) < hibernate) 619| | continue; 620| |#else 621| 87.9M| if (KMP_BLOCKING(hibernate_goal, poll_count++)) ------------------ | | 1222| 87.9M|#define KMP_BLOCKING(goal, count) ((goal) > KMP_NOW()) | | ------------------ | | | | 1217| 87.9M|#define KMP_NOW() __kmp_hardware_timestamp() | | ------------------ | | | Branch (1222:35): [True: 87.7M, False: 269k] | | ------------------ ------------------ 622| 87.7M| continue; 623| 269k|#endif 624| | // Don't suspend if wait loop designated non-sleepable 625| | // in template parameters 626| 269k| if (!Sleepable) ------------------ | Branch (626:9): [Folded, False: 269k] ------------------ 627| 0| continue; 628| | 629| 269k|#if KMP_HAVE_MWAIT || KMP_HAVE_UMWAIT 630| 269k| if (__kmp_mwait_enabled || __kmp_umwait_enabled) { ------------------ | Branch (630:9): [True: 269k, False: 18] | Branch (630:32): [True: 0, False: 18] ------------------ 631| 0| KF_TRACE(50, ("__kmp_wait_sleep: T#%d using monitor/mwait\n", th_gtid)); 632| 0| flag->mwait(th_gtid); 633| 269k| } else { 634| 269k|#endif 635| 269k| KF_TRACE(50, ("__kmp_wait_sleep: T#%d suspend time reached\n", th_gtid)); 636| 269k|#if KMP_OS_UNIX 637| 269k| if (final_spin) ------------------ | Branch (637:11): [Folded, False: 269k] ------------------ 638| 269k| KMP_ATOMIC_ST_REL(&this_thr->th.th_blocking, false); ------------------ | | 1251| 0|#define KMP_ATOMIC_ST_REL(p, v) KMP_ATOMIC_OP(store, p, v, release) | | ------------------ | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 639| 269k|#endif 640| 269k| flag->suspend(th_gtid); 641| 269k|#if KMP_OS_UNIX 642| 269k| if (final_spin) ------------------ | Branch (642:11): [Folded, False: 269k] ------------------ 643| 269k| KMP_ATOMIC_ST_REL(&this_thr->th.th_blocking, true); ------------------ | | 1251| 0|#define KMP_ATOMIC_ST_REL(p, v) KMP_ATOMIC_OP(store, p, v, release) | | ------------------ | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 644| 269k|#endif 645| 269k|#if KMP_HAVE_MWAIT || KMP_HAVE_UMWAIT 646| 269k| } 647| 269k|#endif 648| | 649| 269k| if (TCR_4(__kmp_global.g.g_done)) { ------------------ | | 1127| 269k|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 269k] | | ------------------ ------------------ 650| 0| if (__kmp_global.g.g_abort) ------------------ | Branch (650:11): [True: 0, False: 0] ------------------ 651| 0| __kmp_abort_thread(); 652| 0| break; 653| 269k| } else if (__kmp_tasking_mode != tskm_immediate_exec && ------------------ | Branch (653:16): [True: 18, False: 269k] ------------------ 654| 18| this_thr->th.th_reap_state == KMP_SAFE_TO_REAP) { ------------------ | | 2047| 18|#define KMP_SAFE_TO_REAP 1 // Thread th_reap_state: safe to reap (not tasking) ------------------ | Branch (654:16): [True: 18, False: 0] ------------------ 655| 18| this_thr->th.th_reap_state = KMP_NOT_SAFE_TO_REAP; ------------------ | | 2046| 18| 0 // Thread th_reap_state: not safe to reap (tasking) ------------------ 656| 18| } 657| | // TODO: If thread is done with work and times out, disband/free 658| 269k| } 659| | 660| 36.3k|#if OMPT_SUPPORT 661| 36.3k| ompt_state_t ompt_exit_state = this_thr->th.ompt_thread_info.state; 662| 36.3k| if (ompt_enabled.enabled && ompt_exit_state != ompt_state_undefined) { ------------------ | Branch (662:7): [True: 0, False: 36.3k] | Branch (662:31): [True: 0, False: 0] ------------------ 663| 0|#if OMPT_OPTIONAL 664| 0| if (final_spin) { ------------------ | Branch (664:9): [Folded, False: 0] ------------------ 665| 0| __ompt_implicit_task_end(this_thr, ompt_exit_state, tId); 666| 0| ompt_exit_state = this_thr->th.ompt_thread_info.state; 667| 0| } 668| 0|#endif 669| 0| if (ompt_exit_state == ompt_state_idle) { ------------------ | Branch (669:9): [True: 0, False: 0] ------------------ 670| 0| this_thr->th.ompt_thread_info.state = ompt_state_overhead; 671| 0| } 672| 0| } 673| 36.3k|#endif 674| |#if KMP_STATS_ENABLED 675| | // If we were put into idle state, pop that off the state stack 676| | if (KMP_GET_THREAD_STATE() == IDLE) { 677| | KMP_POP_PARTITIONED_TIMER(); 678| | KMP_SET_THREAD_STATE(thread_state); 679| | this_thr->th.th_stats->resetIdleFlag(); 680| | } 681| |#endif 682| | 683| 36.3k|#if KMP_OS_UNIX 684| 36.3k| if (final_spin) ------------------ | Branch (684:7): [Folded, False: 36.3k] ------------------ 685| 36.3k| KMP_ATOMIC_ST_REL(&this_thr->th.th_blocking, false); ------------------ | | 1251| 0|#define KMP_ATOMIC_ST_REL(p, v) KMP_ATOMIC_OP(store, p, v, release) | | ------------------ | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 686| 36.3k|#endif 687| 36.3k| KMP_FSYNC_SPIN_ACQUIRED(CCAST(void *, spin)); ------------------ | | 339| 36.3k|#define KMP_FSYNC_SPIN_ACQUIRED(obj) ((void)0) ------------------ 688| 36.3k| if (Cancellable) { ------------------ | Branch (688:7): [Folded, False: 36.3k] ------------------ 689| 0| kmp_team_t *team = this_thr->th.th_team; 690| 0| if (team && team->t.t_cancel_request == cancel_parallel) { ------------------ | Branch (690:9): [True: 0, False: 0] | Branch (690:17): [True: 0, False: 0] ------------------ 691| 0| if (tasks_completed) { ------------------ | Branch (691:11): [True: 0, False: 0] ------------------ 692| | // undo the previous decrement of unfinished_threads so that the 693| | // thread can decrement at the join barrier with no problem 694| 0| kmp_task_team_t *task_team = this_thr->th.th_task_team; 695| 0| std::atomic *unfinished_threads = 696| 0| &(task_team->tt.tt_unfinished_threads); 697| 0| KMP_ATOMIC_INC(unfinished_threads); ------------------ | | 1259| 0|#define KMP_ATOMIC_INC(p) KMP_ATOMIC_OP(fetch_add, p, 1, acq_rel) | | ------------------ | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 698| 0| } 699| 0| return true; 700| 0| } 701| 0| } 702| 36.3k| return false; 703| 36.3k|} _ZN11kmp_flag_64ILb0ELb1EE7releaseEv: 887| 147k| void release() { __kmp_release_template(this); } kmp_barrier.cpp:_ZL22__kmp_release_templateI11kmp_flag_64ILb0ELb1EEEvPT_: 788| 147k|template static inline void __kmp_release_template(C *flag) { 789| |#ifdef KMP_DEBUG 790| | int gtid = TCR_4(__kmp_init_gtid) ? __kmp_get_gtid() : -1; 791| |#endif 792| 147k| KF_TRACE(20, ("__kmp_release: T#%d releasing flag(%x)\n", gtid, flag->get())); 793| 147k| KMP_DEBUG_ASSERT(flag->get()); 794| 147k| KMP_FSYNC_RELEASING(flag->get_void_p()); ------------------ | | 335| 147k|#define KMP_FSYNC_RELEASING(obj) ((void)0) ------------------ 795| | 796| 147k| flag->internal_release(); 797| | 798| 147k| KF_TRACE(100, ("__kmp_release: T#%d set new spin=%d\n", gtid, flag->get(), 799| 147k| flag->load())); 800| | 801| 150k| if (__kmp_dflt_blocktime != KMP_MAX_BLOCKTIME) { ------------------ | | 1184| 147k| (INT_MAX) /* Must be this for "infinite" setting the work */ ------------------ | Branch (801:7): [True: 150k, False: 18.4E] ------------------ 802| | // Only need to check sleep stuff if infinite block time not set. 803| | // Are *any* threads waiting on flag sleeping? 804| 150k| if (flag->is_any_sleeping()) { ------------------ | Branch (804:9): [True: 1.54k, False: 148k] ------------------ 805| 3.07k| for (unsigned int i = 0; i < flag->get_num_waiters(); ++i) { ------------------ | Branch (805:32): [True: 1.53k, False: 1.54k] ------------------ 806| | // if sleeping waiter exists at i, sets current_waiter to i inside flag 807| 1.53k| kmp_info_t *waiter = flag->get_waiter(i); 808| 1.53k| if (waiter) { ------------------ | Branch (808:13): [True: 1.52k, False: 6] ------------------ 809| 1.52k| int wait_gtid = waiter->th.th_info.ds.ds_gtid; 810| | // Wake up thread if needed 811| 1.52k| KF_TRACE(50, ("__kmp_release: T#%d waking up thread T#%d since sleep " 812| 1.52k| "flag(%p) set\n", 813| 1.52k| gtid, wait_gtid, flag->get())); 814| 1.52k| flag->resume(wait_gtid); // unsets flag's current_waiter when done 815| 1.52k| } 816| 1.53k| } 817| 1.54k| } 818| 150k| } 819| 147k|} _ZN15kmp_flag_nativeIyL9flag_type1ELb1EE16internal_releaseEv: 186| 147k| void internal_release() { 187| 147k| (void)traits_type::test_then_add4((volatile PtrType *)this->get()); 188| 147k| } _ZN15kmp_flag_nativeIyL9flag_type1ELb1EE15is_any_sleepingEv: 224| 150k| bool is_any_sleeping() { 225| 150k| if (this->sleepLoc) ------------------ | Branch (225:9): [True: 0, False: 150k] ------------------ 226| 0| return this->sleepLoc->load(); 227| 150k| return is_sleeping_val(*(this->get())); 228| 150k| } _ZN8kmp_flagIL9flag_type1EE15get_num_waitersEv: 128| 3.14k| kmp_uint32 get_num_waiters() { return num_waiting_threads; } _ZN8kmp_flagIL9flag_type1EE10get_waiterEj: 123| 1.55k| kmp_info_t *get_waiter(kmp_uint32 i) { 124| 1.55k| KMP_DEBUG_ASSERT(i < num_waiting_threads); 125| 1.55k| return waiting_threads[i]; 126| 1.55k| } _ZN11kmp_flag_64ILb0ELb1EEC2EPVy: 859| 76.5k| : kmp_flag_native(p) {} _ZN15kmp_flag_nativeIyL9flag_type1ELb1EEC2EPVy: 148| 76.4k| kmp_flag_native(volatile PtrType *p) : kmp_flag(), loc(p) {} _ZN8kmp_flagIL9flag_type1EE10set_waiterEP8kmp_info: 131| 75.0k| void set_waiter(kmp_info_t *thr) { 132| 75.0k| waiting_threads[0] = thr; 133| 75.0k| num_waiting_threads = 1; 134| 75.0k| } kmp_wait_release.cpp:_ZL22__kmp_release_templateI11kmp_flag_64ILb0ELb1EEEvPT_: 788| 31|template static inline void __kmp_release_template(C *flag) { 789| |#ifdef KMP_DEBUG 790| | int gtid = TCR_4(__kmp_init_gtid) ? __kmp_get_gtid() : -1; 791| |#endif 792| 31| KF_TRACE(20, ("__kmp_release: T#%d releasing flag(%x)\n", gtid, flag->get())); 793| 31| KMP_DEBUG_ASSERT(flag->get()); 794| 31| KMP_FSYNC_RELEASING(flag->get_void_p()); ------------------ | | 335| 31|#define KMP_FSYNC_RELEASING(obj) ((void)0) ------------------ 795| | 796| 31| flag->internal_release(); 797| | 798| 31| KF_TRACE(100, ("__kmp_release: T#%d set new spin=%d\n", gtid, flag->get(), 799| 31| flag->load())); 800| | 801| 31| if (__kmp_dflt_blocktime != KMP_MAX_BLOCKTIME) { ------------------ | | 1184| 31| (INT_MAX) /* Must be this for "infinite" setting the work */ ------------------ | Branch (801:7): [True: 31, False: 0] ------------------ 802| | // Only need to check sleep stuff if infinite block time not set. 803| | // Are *any* threads waiting on flag sleeping? 804| 31| if (flag->is_any_sleeping()) { ------------------ | Branch (804:9): [True: 31, False: 0] ------------------ 805| 62| for (unsigned int i = 0; i < flag->get_num_waiters(); ++i) { ------------------ | Branch (805:32): [True: 31, False: 31] ------------------ 806| | // if sleeping waiter exists at i, sets current_waiter to i inside flag 807| 31| kmp_info_t *waiter = flag->get_waiter(i); 808| 31| if (waiter) { ------------------ | Branch (808:13): [True: 31, False: 0] ------------------ 809| 31| int wait_gtid = waiter->th.th_info.ds.ds_gtid; 810| | // Wake up thread if needed 811| 31| KF_TRACE(50, ("__kmp_release: T#%d waking up thread T#%d since sleep " 812| 31| "flag(%p) set\n", 813| 31| gtid, wait_gtid, flag->get())); 814| 31| flag->resume(wait_gtid); // unsets flag's current_waiter when done 815| 31| } 816| 31| } 817| 31| } 818| 31| } 819| 31|} _ZN15kmp_flag_nativeIyL9flag_type1ELb1EE12set_sleepingEv: 192| 1.62k| PtrType set_sleeping() { 193| 1.62k| if (this->sleepLoc) { ------------------ | Branch (193:9): [True: 0, False: 1.62k] ------------------ 194| 0| this->sleepLoc->store(true); 195| 0| return *(this->get()); 196| 0| } 197| 1.62k| return traits_type::test_then_or((volatile PtrType *)this->get(), 198| 1.62k| KMP_BARRIER_SLEEP_STATE); ------------------ | | 2023| 1.62k|#define KMP_BARRIER_SLEEP_STATE (1 << KMP_BARRIER_SLEEP_BIT) | | ------------------ | | | | 2019| 1.62k|#define KMP_BARRIER_SLEEP_BIT 0 /* bit used for suspend/sleep part of state */ | | ------------------ ------------------ 199| 1.62k| } _ZN15kmp_flag_nativeIyL9flag_type1ELb1EE14unset_sleepingEv: 203| 1.62k| void unset_sleeping() { 204| 1.62k| if (this->sleepLoc) { ------------------ | Branch (204:9): [True: 0, False: 1.62k] ------------------ 205| 0| this->sleepLoc->store(false); 206| 0| return; 207| 0| } 208| 1.62k| traits_type::test_then_and((volatile PtrType *)this->get(), 209| 1.62k| ~KMP_BARRIER_SLEEP_STATE); ------------------ | | 2023| 1.62k|#define KMP_BARRIER_SLEEP_STATE (1 << KMP_BARRIER_SLEEP_BIT) | | ------------------ | | | | 2019| 1.62k|#define KMP_BARRIER_SLEEP_BIT 0 /* bit used for suspend/sleep part of state */ | | ------------------ ------------------ 210| 1.62k| } _ZN8kmp_flagIL9flag_type1EE8get_typeEv: 119| 3.20k| flag_type get_type() { return (flag_type)(t.type); } _Z9ompd_initv: 43| 1|void ompd_init() { 44| | 45| 1| static int ompd_initialized = 0; 46| | 47| 1| if (ompd_initialized) ------------------ | Branch (47:7): [True: 0, False: 1] ------------------ 48| 0| return; 49| | 50| | /** 51| | * Calculate member offsets for structs and unions 52| | */ 53| | 54| 1|#define ompd_init_access(t, m) \ 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) ------------------ | | 38| 1| OMPD_ACCESS(kmp_base_info_t, th_current_task) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 39| 1| OMPD_ACCESS(kmp_base_info_t, th_team) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 40| 1| OMPD_ACCESS(kmp_base_info_t, th_info) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 41| 1| OMPD_ACCESS(kmp_base_info_t, ompt_thread_info) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 42| 1| \ | | 43| 1| OMPD_ACCESS(kmp_base_root_t, r_in_parallel) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 44| 1| \ | | 45| 1| OMPD_ACCESS(kmp_base_team_t, ompt_team_info) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 46| 1| OMPD_ACCESS(kmp_base_team_t, ompt_serialized_team_info) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 47| 1| OMPD_ACCESS(kmp_base_team_t, t_active_level) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 48| 1| OMPD_ACCESS(kmp_base_team_t, t_implicit_task_taskdata) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 49| 1| OMPD_ACCESS(kmp_base_team_t, t_master_tid) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 50| 1| OMPD_ACCESS(kmp_base_team_t, t_nproc) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 51| 1| OMPD_ACCESS(kmp_base_team_t, t_level) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 52| 1| OMPD_ACCESS(kmp_base_team_t, t_parent) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 53| 1| OMPD_ACCESS(kmp_base_team_t, t_pkfn) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 54| 1| OMPD_ACCESS(kmp_base_team_t, t_threads) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 55| 1| \ | | 56| 1| OMPD_ACCESS(kmp_desc_t, ds) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 57| 1| \ | | 58| 1| OMPD_ACCESS(kmp_desc_base_t, ds_thread) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 59| 1| OMPD_ACCESS(kmp_desc_base_t, ds_tid) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 60| 1| \ | | 61| 1| OMPD_ACCESS(kmp_info_t, th) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 62| 1| \ | | 63| 1| OMPD_ACCESS(kmp_r_sched_t, r_sched_type) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 64| 1| OMPD_ACCESS(kmp_r_sched_t, chunk) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 65| 1| \ | | 66| 1| OMPD_ACCESS(kmp_root_t, r) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 67| 1| \ | | 68| 1| OMPD_ACCESS(kmp_internal_control_t, dynamic) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 69| 1| OMPD_ACCESS(kmp_internal_control_t, max_active_levels) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 70| 1| OMPD_ACCESS(kmp_internal_control_t, nproc) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 71| 1| OMPD_ACCESS(kmp_internal_control_t, proc_bind) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 72| 1| OMPD_ACCESS(kmp_internal_control_t, sched) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 73| 1| OMPD_ACCESS(kmp_internal_control_t, default_device) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 74| 1| OMPD_ACCESS(kmp_internal_control_t, thread_limit) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 75| 1| \ | | 76| 1| OMPD_ACCESS(kmp_taskdata_t, ompt_task_info) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 77| 1| OMPD_ACCESS(kmp_taskdata_t, td_flags) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 78| 1| OMPD_ACCESS(kmp_taskdata_t, td_icvs) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 79| 1| OMPD_ACCESS(kmp_taskdata_t, td_parent) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 80| 1| OMPD_ACCESS(kmp_taskdata_t, td_team) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 81| 1| \ | | 82| 1| OMPD_ACCESS(kmp_task_t, routine) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 83| 1| \ | | 84| 1| OMPD_ACCESS(kmp_team_p, t) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 85| 1| \ | | 86| 1| OMPD_ACCESS(kmp_nested_nthreads_t, used) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 87| 1| OMPD_ACCESS(kmp_nested_nthreads_t, nth) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 88| 1| \ | | 89| 1| OMPD_ACCESS(kmp_nested_proc_bind_t, used) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 90| 1| OMPD_ACCESS(kmp_nested_proc_bind_t, bind_types) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 91| 1| \ | | 92| 1| OMPD_ACCESS(ompt_task_info_t, frame) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 93| 1| OMPD_ACCESS(ompt_task_info_t, scheduling_parent) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 94| 1| OMPD_ACCESS(ompt_task_info_t, task_data) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 95| 1| \ | | 96| 1| OMPD_ACCESS(ompt_team_info_t, parallel_data) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 97| 1| \ | | 98| 1| OMPD_ACCESS(ompt_thread_info_t, state) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 99| 1| OMPD_ACCESS(ompt_thread_info_t, wait_id) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 100| 1| OMPD_ACCESS(ompt_thread_info_t, thread_data) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 101| 1| \ | | 102| 1| OMPD_ACCESS(ompt_data_t, value) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 103| 1| OMPD_ACCESS(ompt_data_t, ptr) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 104| 1| \ | | 105| 1| OMPD_ACCESS(ompt_frame_t, exit_frame) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 106| 1| OMPD_ACCESS(ompt_frame_t, enter_frame) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 107| 1| \ | | 108| 1| OMPD_ACCESS(ompt_lw_taskteam_t, parent) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 109| 1| OMPD_ACCESS(ompt_lw_taskteam_t, ompt_team_info) \ | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ | | 110| 1| OMPD_ACCESS(ompt_lw_taskteam_t, ompt_task_info) | | ------------------ | | | | 56| 1| OMPD_FOREACH_ACCESS(ompd_init_access) | | | | ------------------ | | | | | | 55| 1| ompd_access__##t##__##m = (uint64_t) & (((t *)0)->m); | | | | ------------------ | | ------------------ ------------------ 57| 1|#undef ompd_init_access 58| | 59| | /** 60| | * Create bit mask for bitfield access 61| | */ 62| | 63| 1|#define ompd_init_bitfield(t, m) \ 64| 1| ompd_bitfield__##t##__##m = 0; \ 65| 1| ((t *)(&ompd_bitfield__##t##__##m))->m = 1; 66| 1| OMPD_FOREACH_BITFIELD(ompd_init_bitfield) ------------------ | | 113| 1| OMPD_BITFIELD(kmp_tasking_flags_t, final) \ | | ------------------ | | | | 66| 1| OMPD_FOREACH_BITFIELD(ompd_init_bitfield) | | | | ------------------ | | | | | | 64| 1| ompd_bitfield__##t##__##m = 0; \ | | | | | | 65| 1| ((t *)(&ompd_bitfield__##t##__##m))->m = 1; | | | | ------------------ | | ------------------ | | 114| 1| OMPD_BITFIELD(kmp_tasking_flags_t, tiedness) \ | | ------------------ | | | | 66| 1| OMPD_FOREACH_BITFIELD(ompd_init_bitfield) | | | | ------------------ | | | | | | 64| 1| ompd_bitfield__##t##__##m = 0; \ | | | | | | 65| 1| ((t *)(&ompd_bitfield__##t##__##m))->m = 1; | | | | ------------------ | | ------------------ | | 115| 1| OMPD_BITFIELD(kmp_tasking_flags_t, tasktype) \ | | ------------------ | | | | 66| 1| OMPD_FOREACH_BITFIELD(ompd_init_bitfield) | | | | ------------------ | | | | | | 64| 1| ompd_bitfield__##t##__##m = 0; \ | | | | | | 65| 1| ((t *)(&ompd_bitfield__##t##__##m))->m = 1; | | | | ------------------ | | ------------------ | | 116| 1| OMPD_BITFIELD(kmp_tasking_flags_t, task_serial) \ | | ------------------ | | | | 66| 1| OMPD_FOREACH_BITFIELD(ompd_init_bitfield) | | | | ------------------ | | | | | | 64| 1| ompd_bitfield__##t##__##m = 0; \ | | | | | | 65| 1| ((t *)(&ompd_bitfield__##t##__##m))->m = 1; | | | | ------------------ | | ------------------ | | 117| 1| OMPD_BITFIELD(kmp_tasking_flags_t, tasking_ser) \ | | ------------------ | | | | 66| 1| OMPD_FOREACH_BITFIELD(ompd_init_bitfield) | | | | ------------------ | | | | | | 64| 1| ompd_bitfield__##t##__##m = 0; \ | | | | | | 65| 1| ((t *)(&ompd_bitfield__##t##__##m))->m = 1; | | | | ------------------ | | ------------------ | | 118| 1| OMPD_BITFIELD(kmp_tasking_flags_t, team_serial) \ | | ------------------ | | | | 66| 1| OMPD_FOREACH_BITFIELD(ompd_init_bitfield) | | | | ------------------ | | | | | | 64| 1| ompd_bitfield__##t##__##m = 0; \ | | | | | | 65| 1| ((t *)(&ompd_bitfield__##t##__##m))->m = 1; | | | | ------------------ | | ------------------ | | 119| 1| OMPD_BITFIELD(kmp_tasking_flags_t, started) \ | | ------------------ | | | | 66| 1| OMPD_FOREACH_BITFIELD(ompd_init_bitfield) | | | | ------------------ | | | | | | 64| 1| ompd_bitfield__##t##__##m = 0; \ | | | | | | 65| 1| ((t *)(&ompd_bitfield__##t##__##m))->m = 1; | | | | ------------------ | | ------------------ | | 120| 1| OMPD_BITFIELD(kmp_tasking_flags_t, executing) \ | | ------------------ | | | | 66| 1| OMPD_FOREACH_BITFIELD(ompd_init_bitfield) | | | | ------------------ | | | | | | 64| 1| ompd_bitfield__##t##__##m = 0; \ | | | | | | 65| 1| ((t *)(&ompd_bitfield__##t##__##m))->m = 1; | | | | ------------------ | | ------------------ | | 121| 1| OMPD_BITFIELD(kmp_tasking_flags_t, complete) \ | | ------------------ | | | | 66| 1| OMPD_FOREACH_BITFIELD(ompd_init_bitfield) | | | | ------------------ | | | | | | 64| 1| ompd_bitfield__##t##__##m = 0; \ | | | | | | 65| 1| ((t *)(&ompd_bitfield__##t##__##m))->m = 1; | | | | ------------------ | | ------------------ | | 122| 1| OMPD_BITFIELD(kmp_tasking_flags_t, freed) \ | | ------------------ | | | | 66| 1| OMPD_FOREACH_BITFIELD(ompd_init_bitfield) | | | | ------------------ | | | | | | 64| 1| ompd_bitfield__##t##__##m = 0; \ | | | | | | 65| 1| ((t *)(&ompd_bitfield__##t##__##m))->m = 1; | | | | ------------------ | | ------------------ | | 123| 1| OMPD_BITFIELD(kmp_tasking_flags_t, native) | | ------------------ | | | | 66| 1| OMPD_FOREACH_BITFIELD(ompd_init_bitfield) | | | | ------------------ | | | | | | 64| 1| ompd_bitfield__##t##__##m = 0; \ | | | | | | 65| 1| ((t *)(&ompd_bitfield__##t##__##m))->m = 1; | | | | ------------------ | | ------------------ ------------------ 67| 1|#undef ompd_init_bitfield 68| | 69| | /** 70| | * Calculate type size information 71| | */ 72| | 73| 1|#define ompd_init_sizeof_member(t, m) \ 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) ------------------ | | 38| 1| OMPD_ACCESS(kmp_base_info_t, th_current_task) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 39| 1| OMPD_ACCESS(kmp_base_info_t, th_team) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 40| 1| OMPD_ACCESS(kmp_base_info_t, th_info) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 41| 1| OMPD_ACCESS(kmp_base_info_t, ompt_thread_info) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 42| 1| \ | | 43| 1| OMPD_ACCESS(kmp_base_root_t, r_in_parallel) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 44| 1| \ | | 45| 1| OMPD_ACCESS(kmp_base_team_t, ompt_team_info) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 46| 1| OMPD_ACCESS(kmp_base_team_t, ompt_serialized_team_info) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 47| 1| OMPD_ACCESS(kmp_base_team_t, t_active_level) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 48| 1| OMPD_ACCESS(kmp_base_team_t, t_implicit_task_taskdata) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 49| 1| OMPD_ACCESS(kmp_base_team_t, t_master_tid) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 50| 1| OMPD_ACCESS(kmp_base_team_t, t_nproc) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 51| 1| OMPD_ACCESS(kmp_base_team_t, t_level) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 52| 1| OMPD_ACCESS(kmp_base_team_t, t_parent) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 53| 1| OMPD_ACCESS(kmp_base_team_t, t_pkfn) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 54| 1| OMPD_ACCESS(kmp_base_team_t, t_threads) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 55| 1| \ | | 56| 1| OMPD_ACCESS(kmp_desc_t, ds) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 57| 1| \ | | 58| 1| OMPD_ACCESS(kmp_desc_base_t, ds_thread) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 59| 1| OMPD_ACCESS(kmp_desc_base_t, ds_tid) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 60| 1| \ | | 61| 1| OMPD_ACCESS(kmp_info_t, th) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 62| 1| \ | | 63| 1| OMPD_ACCESS(kmp_r_sched_t, r_sched_type) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 64| 1| OMPD_ACCESS(kmp_r_sched_t, chunk) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 65| 1| \ | | 66| 1| OMPD_ACCESS(kmp_root_t, r) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 67| 1| \ | | 68| 1| OMPD_ACCESS(kmp_internal_control_t, dynamic) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 69| 1| OMPD_ACCESS(kmp_internal_control_t, max_active_levels) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 70| 1| OMPD_ACCESS(kmp_internal_control_t, nproc) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 71| 1| OMPD_ACCESS(kmp_internal_control_t, proc_bind) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 72| 1| OMPD_ACCESS(kmp_internal_control_t, sched) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 73| 1| OMPD_ACCESS(kmp_internal_control_t, default_device) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 74| 1| OMPD_ACCESS(kmp_internal_control_t, thread_limit) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 75| 1| \ | | 76| 1| OMPD_ACCESS(kmp_taskdata_t, ompt_task_info) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 77| 1| OMPD_ACCESS(kmp_taskdata_t, td_flags) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 78| 1| OMPD_ACCESS(kmp_taskdata_t, td_icvs) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 79| 1| OMPD_ACCESS(kmp_taskdata_t, td_parent) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 80| 1| OMPD_ACCESS(kmp_taskdata_t, td_team) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 81| 1| \ | | 82| 1| OMPD_ACCESS(kmp_task_t, routine) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 83| 1| \ | | 84| 1| OMPD_ACCESS(kmp_team_p, t) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 85| 1| \ | | 86| 1| OMPD_ACCESS(kmp_nested_nthreads_t, used) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 87| 1| OMPD_ACCESS(kmp_nested_nthreads_t, nth) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 88| 1| \ | | 89| 1| OMPD_ACCESS(kmp_nested_proc_bind_t, used) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 90| 1| OMPD_ACCESS(kmp_nested_proc_bind_t, bind_types) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 91| 1| \ | | 92| 1| OMPD_ACCESS(ompt_task_info_t, frame) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 93| 1| OMPD_ACCESS(ompt_task_info_t, scheduling_parent) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 94| 1| OMPD_ACCESS(ompt_task_info_t, task_data) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 95| 1| \ | | 96| 1| OMPD_ACCESS(ompt_team_info_t, parallel_data) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 97| 1| \ | | 98| 1| OMPD_ACCESS(ompt_thread_info_t, state) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 99| 1| OMPD_ACCESS(ompt_thread_info_t, wait_id) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 100| 1| OMPD_ACCESS(ompt_thread_info_t, thread_data) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 101| 1| \ | | 102| 1| OMPD_ACCESS(ompt_data_t, value) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 103| 1| OMPD_ACCESS(ompt_data_t, ptr) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 104| 1| \ | | 105| 1| OMPD_ACCESS(ompt_frame_t, exit_frame) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 106| 1| OMPD_ACCESS(ompt_frame_t, enter_frame) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 107| 1| \ | | 108| 1| OMPD_ACCESS(ompt_lw_taskteam_t, parent) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 109| 1| OMPD_ACCESS(ompt_lw_taskteam_t, ompt_team_info) \ | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ | | 110| 1| OMPD_ACCESS(ompt_lw_taskteam_t, ompt_task_info) | | ------------------ | | | | 75| 1| OMPD_FOREACH_ACCESS(ompd_init_sizeof_member) | | | | ------------------ | | | | | | 74| 1| ompd_sizeof__##t##__##m = sizeof(((t *)0)->m); | | | | ------------------ | | ------------------ ------------------ 76| 1|#undef ompd_init_sizeof_member 77| | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) ------------------ | | 126| 1| OMPD_SIZEOF(kmp_info_t) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 127| 1| OMPD_SIZEOF(kmp_taskdata_t) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 128| 1| OMPD_SIZEOF(kmp_task_t) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 129| 1| OMPD_SIZEOF(kmp_tasking_flags_t) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 130| 1| OMPD_SIZEOF(kmp_thread_t) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 131| 1| OMPD_SIZEOF(ompt_data_t) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 132| 1| OMPD_SIZEOF(ompt_id_t) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 133| 1| OMPD_SIZEOF(__kmp_avail_proc) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 134| 1| OMPD_SIZEOF(__kmp_max_nth) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 135| 1| OMPD_SIZEOF(__kmp_stksize) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 136| 1| OMPD_SIZEOF(__kmp_omp_cancellation) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 137| 1| OMPD_SIZEOF(__kmp_max_task_priority) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 138| 1| OMPD_SIZEOF(__kmp_display_affinity) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 139| 1| OMPD_SIZEOF(__kmp_affinity_format) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 140| 1| OMPD_SIZEOF(__kmp_tool_libraries) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 141| 1| OMPD_SIZEOF(__kmp_tool_verbose_init) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 142| 1| OMPD_SIZEOF(__kmp_tool) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 143| 1| OMPD_SIZEOF(ompd_state) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 144| 1| OMPD_SIZEOF(kmp_nested_nthreads_t) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 145| 1| OMPD_SIZEOF(__kmp_nested_nth) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 146| 1| OMPD_SIZEOF(kmp_nested_proc_bind_t) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 147| 1| OMPD_SIZEOF(__kmp_nested_proc_bind) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 148| 1| OMPD_SIZEOF(int) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 149| 1| OMPD_SIZEOF(char) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 150| 1| OMPD_SIZEOF(__kmp_gtid) \ | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ | | 151| 1| OMPD_SIZEOF(__kmp_nth) | | ------------------ | | | | 79| 1| OMPD_FOREACH_SIZEOF(ompd_init_sizeof) | | | | ------------------ | | | | | | 78| 1|#define ompd_init_sizeof(t) ompd_sizeof__##t = sizeof(t); | | | | ------------------ | | ------------------ ------------------ 80| 1|#undef ompd_init_sizeof 81| | 82| 1| char *libname = NULL; 83| | 84| 1|#if KMP_OS_UNIX 85| | // Find the location of libomp.so thru dladdr and replace the libomp with 86| | // libompd to get the full path of libompd 87| 1| Dl_info dl_info; 88| 1| int ret = dladdr((void *)ompd_init, &dl_info); 89| 1| if (!ret) { ------------------ | Branch (89:7): [True: 0, False: 1] ------------------ 90| 0| fprintf(stderr, "%s\n", dlerror()); 91| 0| } 92| 1| int lib_path_length; 93| 1| if (strrchr(dl_info.dli_fname, '/')) { ------------------ | Branch (93:7): [True: 1, False: 0] ------------------ 94| 1| lib_path_length = strrchr(dl_info.dli_fname, '/') - dl_info.dli_fname; 95| 1| libname = 96| 1| (char *)malloc(lib_path_length + 12 /*for '/libompd.so' and '\0'*/); 97| 1| strncpy(libname, dl_info.dli_fname, lib_path_length); 98| 1| memcpy(libname + lib_path_length, "/libompd.so\0", 12); 99| 1| } 100| 1|#endif 101| | 102| 1| const char *ompd_env_var = getenv("OMP_DEBUG"); 103| 1| if (ompd_env_var && !strcmp(ompd_env_var, "enabled")) { ------------------ | Branch (103:7): [True: 0, False: 1] | Branch (103:23): [True: 0, False: 0] ------------------ 104| 0| fprintf(stderr, "OMP_OMPD active\n"); 105| 0| ompt_enabled.enabled = 1; 106| 0| ompd_state |= OMPD_ENABLE_BP; ------------------ | | 35| 0|#define OMPD_ENABLE_BP 0x1 ------------------ 107| 0| } 108| | 109| 1| ompd_initialized = 1; 110| 1| ompd_dll_locations = (volatile const char **)malloc(3 * sizeof(const char *)); 111| 1| ompd_dll_locations[0] = "libompd.so"; 112| 1| ompd_dll_locations[1] = libname; 113| | ompd_dll_locations[2] = NULL; 114| 1| ompd_dll_locations_valid(); 115| 1|} _Z24ompd_dll_locations_validv: 117| 1|void __attribute__((noinline)) ompd_dll_locations_valid(void) { 118| | /* naive way of implementing hard to opt-out empty function 119| | we might want to use a separate object file? */ 120| 1| asm(""); 121| 1|} ompt_start_tool: 161| 1|ompt_start_tool(unsigned int omp_version, const char *runtime_version) { 162| 1| ompt_start_tool_result_t *ret = NULL; 163| | // Search next symbol in the current address space. This can happen if the 164| | // runtime library is linked before the tool. Since glibc 2.2 strong symbols 165| | // don't override weak symbols that have been found before unless the user 166| | // sets the environment variable LD_DYNAMIC_WEAK. 167| 1| ompt_start_tool_t next_tool = 168| 1| (ompt_start_tool_t)dlsym(RTLD_NEXT, "ompt_start_tool"); 169| 1| if (next_tool) { ------------------ | Branch (169:7): [True: 0, False: 1] ------------------ 170| 0| ret = next_tool(omp_version, runtime_version); 171| 0| } 172| 1| return ret; 173| 1|} ompt_pre_init: 392| 1|void ompt_pre_init() { 393| | //-------------------------------------------------- 394| | // Execute the pre-initialization logic only once. 395| | //-------------------------------------------------- 396| 1| static int ompt_pre_initialized = 0; 397| | 398| 1| if (ompt_pre_initialized) ------------------ | Branch (398:7): [True: 0, False: 1] ------------------ 399| 0| return; 400| | 401| 1| ompt_pre_initialized = 1; 402| | 403| | //-------------------------------------------------- 404| | // Use a tool iff a tool is enabled and available. 405| | //-------------------------------------------------- 406| 1| const char *ompt_env_var = getenv("OMP_TOOL"); 407| 1| tool_setting_e tool_setting = omp_tool_error; 408| | 409| 1| if (!ompt_env_var || !strcmp(ompt_env_var, "")) ------------------ | Branch (409:7): [True: 1, False: 0] | Branch (409:24): [True: 0, False: 0] ------------------ 410| 1| tool_setting = omp_tool_unset; 411| 0| else if (OMPT_STR_MATCH(ompt_env_var, "disabled")) ------------------ | | 77| 0|#define OMPT_STR_MATCH(haystack, needle) __kmp_str_match(haystack, 0, needle) | | ------------------ | | | Branch (77:42): [True: 0, False: 0] | | ------------------ ------------------ 412| 0| tool_setting = omp_tool_disabled; 413| 0| else if (OMPT_STR_MATCH(ompt_env_var, "enabled")) ------------------ | | 77| 0|#define OMPT_STR_MATCH(haystack, needle) __kmp_str_match(haystack, 0, needle) | | ------------------ | | | Branch (77:42): [True: 0, False: 0] | | ------------------ ------------------ 414| 0| tool_setting = omp_tool_enabled; 415| | 416| 1| const char *ompt_env_verbose_init = getenv("OMP_TOOL_VERBOSE_INIT"); 417| | // possible options: disabled | stdout | stderr | 418| | // if set, not empty and not disabled -> prepare for logging 419| 1| if (ompt_env_verbose_init && strcmp(ompt_env_verbose_init, "") && ------------------ | Branch (419:7): [True: 0, False: 1] | Branch (419:32): [True: 0, False: 0] ------------------ 420| 0| !OMPT_STR_MATCH(ompt_env_verbose_init, "disabled")) { ------------------ | | 77| 0|#define OMPT_STR_MATCH(haystack, needle) __kmp_str_match(haystack, 0, needle) ------------------ | Branch (420:7): [True: 0, False: 0] ------------------ 421| 0| verbose_init = 1; 422| 0| if (OMPT_STR_MATCH(ompt_env_verbose_init, "STDERR")) ------------------ | | 77| 0|#define OMPT_STR_MATCH(haystack, needle) __kmp_str_match(haystack, 0, needle) | | ------------------ | | | Branch (77:42): [True: 0, False: 0] | | ------------------ ------------------ 423| 0| verbose_file = stderr; 424| 0| else if (OMPT_STR_MATCH(ompt_env_verbose_init, "STDOUT")) ------------------ | | 77| 0|#define OMPT_STR_MATCH(haystack, needle) __kmp_str_match(haystack, 0, needle) | | ------------------ | | | Branch (77:42): [True: 0, False: 0] | | ------------------ ------------------ 425| 0| verbose_file = stdout; 426| 0| else 427| 0| verbose_file = fopen(ompt_env_verbose_init, "w"); 428| 0| } else 429| 1| verbose_init = 0; 430| | 431| |#if OMPT_DEBUG 432| | printf("ompt_pre_init(): tool_setting = %d\n", tool_setting); 433| |#endif 434| 1| switch (tool_setting) { ------------------ | Branch (434:11): [True: 1, False: 0] ------------------ 435| 0| case omp_tool_disabled: ------------------ | Branch (435:3): [True: 0, False: 1] ------------------ 436| 0| OMPT_VERBOSE_INIT_PRINT("OMP tool disabled. \n"); ------------------ | | 53| 0| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 0] | | ------------------ | | 54| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 437| 0| break; 438| | 439| 1| case omp_tool_unset: ------------------ | Branch (439:3): [True: 1, False: 0] ------------------ 440| 1| case omp_tool_enabled: ------------------ | Branch (440:3): [True: 0, False: 1] ------------------ 441| | 442| | //-------------------------------------------------- 443| | // Load tool iff specified in environment variable 444| | //-------------------------------------------------- 445| 1| ompt_start_tool_result = 446| 1| ompt_try_start_tool(__kmp_openmp_version, ompt_get_runtime_version()); 447| | 448| 1| memset(&ompt_enabled, 0, sizeof(ompt_enabled)); 449| 1| break; 450| | 451| 0| case omp_tool_error: ------------------ | Branch (451:3): [True: 0, False: 1] ------------------ 452| 0| fprintf(stderr, 453| 0| "Warning: OMP_TOOL has invalid value \"%s\".\n" 454| 0| " legal values are (NULL,\"\",\"disabled\"," 455| 0| "\"enabled\").\n", 456| 0| ompt_env_var); 457| 0| break; 458| 1| } 459| 1| if (verbose_init && verbose_file != stderr && verbose_file != stdout) ------------------ | Branch (459:7): [True: 0, False: 1] | Branch (459:23): [True: 0, False: 0] | Branch (459:49): [True: 0, False: 0] ------------------ 460| 0| fclose(verbose_file); 461| |#if OMPT_DEBUG 462| | printf("ompt_pre_init(): ompt_enabled = %d\n", ompt_enabled.enabled); 463| |#endif 464| 1|} ompt_post_init: 468| 1|void ompt_post_init() { 469| | //-------------------------------------------------- 470| | // Execute the post-initialization logic only once. 471| | //-------------------------------------------------- 472| 1| static int ompt_post_initialized = 0; 473| | 474| 1| if (ompt_post_initialized) ------------------ | Branch (474:7): [True: 0, False: 1] ------------------ 475| 0| return; 476| | 477| 1| ompt_post_initialized = 1; 478| | 479| | //-------------------------------------------------- 480| | // Initialize the tool if so indicated. 481| | //-------------------------------------------------- 482| 1| if (ompt_start_tool_result) { ------------------ | Branch (482:7): [True: 0, False: 1] ------------------ 483| 0| ompt_enabled.enabled = !!ompt_start_tool_result->initialize( 484| 0| ompt_fn_lookup, omp_get_initial_device(), 485| 0| &(ompt_start_tool_result->tool_data)); 486| | 487| 0| if (!ompt_enabled.enabled) { ------------------ | Branch (487:9): [True: 0, False: 0] ------------------ 488| | // tool not enabled, zero out the bitmap, and done 489| 0| memset(&ompt_enabled, 0, sizeof(ompt_enabled)); 490| 0| return; 491| 0| } 492| | 493| 0| kmp_info_t *root_thread = ompt_get_thread(); 494| | 495| 0| ompt_set_thread_state(root_thread, ompt_state_overhead); 496| | 497| 0| if (ompt_enabled.ompt_callback_thread_begin) { ------------------ | Branch (497:9): [True: 0, False: 0] ------------------ 498| 0| ompt_callbacks.ompt_callback(ompt_callback_thread_begin)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 499| 0| ompt_thread_initial, __ompt_get_thread_data_internal()); 500| 0| } 501| 0| ompt_data_t *task_data = nullptr; 502| 0| ompt_data_t *parallel_data = nullptr; 503| 0| __ompt_get_task_info_internal(0, NULL, &task_data, NULL, ¶llel_data, 504| 0| NULL); 505| 0| if (ompt_enabled.ompt_callback_implicit_task) { ------------------ | Branch (505:9): [True: 0, False: 0] ------------------ 506| 0| ompt_callbacks.ompt_callback(ompt_callback_implicit_task)( ------------------ | | 27| 0|#define ompt_callback(e) e##_callback ------------------ 507| 0| ompt_scope_begin, parallel_data, task_data, 1, 1, ompt_task_initial); 508| 0| } 509| | 510| 0| ompt_set_thread_state(root_thread, ompt_state_work_serial); 511| 0| } 512| 1|} ompt_fini: 514| 1|void ompt_fini() { 515| 1| if (ompt_enabled.enabled) { ------------------ | Branch (515:7): [True: 0, False: 1] ------------------ 516| 0| if (ompt_start_tool_result && ompt_start_tool_result->finalize) { ------------------ | Branch (516:9): [True: 0, False: 0] | Branch (516:35): [True: 0, False: 0] ------------------ 517| 0| ompt_start_tool_result->finalize(&(ompt_start_tool_result->tool_data)); 518| 0| } 519| 0| if (libomptarget_ompt_result && libomptarget_ompt_result->finalize) { ------------------ | Branch (519:9): [True: 0, False: 0] | Branch (519:37): [True: 0, False: 0] ------------------ 520| 0| libomptarget_ompt_result->finalize(NULL); 521| 0| } 522| 0| } 523| | 524| 1| if (ompt_tool_module) ------------------ | Branch (524:7): [True: 0, False: 1] ------------------ 525| 0| OMPT_DLCLOSE(ompt_tool_module); ------------------ | | 110| 0|#define OMPT_DLCLOSE(Lib) dlclose(Lib) ------------------ 526| 1| memset(&ompt_enabled, 0, sizeof(ompt_enabled)); 527| 1|} ompt-general.cpp:_ZL19ompt_try_start_tooljPKc: 248| 1|ompt_try_start_tool(unsigned int omp_version, const char *runtime_version) { 249| 1| ompt_start_tool_result_t *ret = NULL; 250| 1| ompt_start_tool_t start_tool = NULL; 251| |#if KMP_OS_WINDOWS 252| | // Cannot use colon to describe a list of absolute paths on Windows 253| | const char *sep = ";"; 254| |#else 255| 1| const char *sep = ":"; 256| 1|#endif 257| | 258| 1| OMPT_VERBOSE_INIT_PRINT("----- START LOGGING OF TOOL REGISTRATION -----\n"); ------------------ | | 53| 1| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 1] | | ------------------ | | 54| 1| fprintf(verbose_file, __VA_ARGS__) ------------------ 259| 1| OMPT_VERBOSE_INIT_PRINT("Search for OMP tool in current address space... "); ------------------ | | 53| 1| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 1] | | ------------------ | | 54| 1| fprintf(verbose_file, __VA_ARGS__) ------------------ 260| | 261| |#if KMP_OS_DARWIN 262| | // Try in the current address space 263| | ret = ompt_tool_darwin(omp_version, runtime_version); 264| |#elif OMPT_HAVE_WEAK_ATTRIBUTE 265| | ret = ompt_start_tool(omp_version, runtime_version); 266| |#elif OMPT_HAVE_PSAPI 267| | ret = ompt_tool_windows(omp_version, runtime_version); 268| |#else 269| |#error Activation of OMPT is not supported on this platform. 270| |#endif 271| 1| if (ret) { ------------------ | Branch (271:7): [True: 0, False: 1] ------------------ 272| 0| OMPT_VERBOSE_INIT_CONTINUED_PRINT("Success.\n"); ------------------ | | 56| 0| if (verbose_init) \ | | ------------------ | | | Branch (56:7): [True: 0, False: 0] | | ------------------ | | 57| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 273| 0| OMPT_VERBOSE_INIT_PRINT( ------------------ | | 53| 0| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 0] | | ------------------ | | 54| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 274| 0| "Tool was started and is using the OMPT interface.\n"); 275| 0| OMPT_VERBOSE_INIT_PRINT("----- END LOGGING OF TOOL REGISTRATION -----\n"); ------------------ | | 53| 0| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 0] | | ------------------ | | 54| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 276| 0| return ret; 277| 0| } 278| | 279| | // Try tool-libraries-var ICV 280| 1| OMPT_VERBOSE_INIT_CONTINUED_PRINT("Failed.\n"); ------------------ | | 56| 1| if (verbose_init) \ | | ------------------ | | | Branch (56:7): [True: 0, False: 1] | | ------------------ | | 57| 1| fprintf(verbose_file, __VA_ARGS__) ------------------ 281| 1| const char *tool_libs = getenv("OMP_TOOL_LIBRARIES"); 282| 1| if (tool_libs) { ------------------ | Branch (282:7): [True: 0, False: 1] ------------------ 283| 0| OMPT_VERBOSE_INIT_PRINT("Searching tool libraries...\n"); ------------------ | | 53| 0| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 0] | | ------------------ | | 54| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 284| 0| OMPT_VERBOSE_INIT_PRINT("OMP_TOOL_LIBRARIES = %s\n", tool_libs); ------------------ | | 53| 0| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 0] | | ------------------ | | 54| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 285| 0| char *libs = __kmp_str_format("%s", tool_libs); 286| 0| char *buf; 287| 0| char *fname = __kmp_str_token(libs, sep, &buf); 288| | // Reset dl-error 289| 0| dlerror(); 290| | 291| 0| while (fname) { ------------------ | Branch (291:12): [True: 0, False: 0] ------------------ 292| 0|#if KMP_OS_UNIX 293| 0| OMPT_VERBOSE_INIT_PRINT("Opening %s... ", fname); ------------------ | | 53| 0| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 0] | | ------------------ | | 54| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 294| 0| void *h = dlopen(fname, RTLD_LAZY); 295| 0| if (!h) { ------------------ | Branch (295:11): [True: 0, False: 0] ------------------ 296| 0| OMPT_VERBOSE_INIT_CONTINUED_PRINT("Failed: %s\n", dlerror()); ------------------ | | 56| 0| if (verbose_init) \ | | ------------------ | | | Branch (56:7): [True: 0, False: 0] | | ------------------ | | 57| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 297| 0| } else { 298| 0| OMPT_VERBOSE_INIT_CONTINUED_PRINT("Success. \n"); ------------------ | | 56| 0| if (verbose_init) \ | | ------------------ | | | Branch (56:7): [True: 0, False: 0] | | ------------------ | | 57| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 299| 0| OMPT_VERBOSE_INIT_PRINT("Searching for ompt_start_tool in %s... ", ------------------ | | 53| 0| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 0] | | ------------------ | | 54| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 300| 0| fname); 301| 0| dlerror(); // Clear any existing error 302| 0| start_tool = (ompt_start_tool_t)dlsym(h, "ompt_start_tool"); 303| 0| if (!start_tool) { ------------------ | Branch (303:13): [True: 0, False: 0] ------------------ 304| 0| char *error = dlerror(); 305| 0| if (error != NULL) { ------------------ | Branch (305:15): [True: 0, False: 0] ------------------ 306| 0| OMPT_VERBOSE_INIT_CONTINUED_PRINT("Failed: %s\n", error); ------------------ | | 56| 0| if (verbose_init) \ | | ------------------ | | | Branch (56:7): [True: 0, False: 0] | | ------------------ | | 57| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 307| 0| } else { 308| 0| OMPT_VERBOSE_INIT_CONTINUED_PRINT("Failed: %s\n", ------------------ | | 56| 0| if (verbose_init) \ | | ------------------ | | | Branch (56:7): [True: 0, False: 0] | | ------------------ | | 57| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 309| 0| "ompt_start_tool = NULL"); 310| 0| } 311| 0| } else 312| |#elif KMP_OS_WINDOWS 313| | OMPT_VERBOSE_INIT_PRINT("Opening %s... ", fname); 314| | HMODULE h = LoadLibrary(fname); 315| | if (!h) { 316| | OMPT_VERBOSE_INIT_CONTINUED_PRINT("Failed: Error %u\n", 317| | (unsigned)GetLastError()); 318| | } else { 319| | OMPT_VERBOSE_INIT_CONTINUED_PRINT("Success. \n"); 320| | OMPT_VERBOSE_INIT_PRINT("Searching for ompt_start_tool in %s... ", 321| | fname); 322| | start_tool = (ompt_start_tool_t)GetProcAddress(h, "ompt_start_tool"); 323| | if (!start_tool) { 324| | OMPT_VERBOSE_INIT_CONTINUED_PRINT("Failed: Error %u\n", 325| | (unsigned)GetLastError()); 326| | } else 327| |#else 328| |#error Activation of OMPT is not supported on this platform. 329| |#endif 330| 0| { // if (start_tool) 331| 0| ret = (*start_tool)(omp_version, runtime_version); 332| 0| if (ret) { ------------------ | Branch (332:15): [True: 0, False: 0] ------------------ 333| 0| OMPT_VERBOSE_INIT_CONTINUED_PRINT("Success.\n"); ------------------ | | 56| 0| if (verbose_init) \ | | ------------------ | | | Branch (56:7): [True: 0, False: 0] | | ------------------ | | 57| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 334| 0| OMPT_VERBOSE_INIT_PRINT( ------------------ | | 53| 0| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 0] | | ------------------ | | 54| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 335| 0| "Tool was started and is using the OMPT interface.\n"); 336| 0| ompt_tool_module = h; 337| 0| break; 338| 0| } 339| 0| OMPT_VERBOSE_INIT_CONTINUED_PRINT( ------------------ | | 56| 0| if (verbose_init) \ | | ------------------ | | | Branch (56:7): [True: 0, False: 0] | | ------------------ | | 57| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 340| 0| "Found but not using the OMPT interface.\n"); 341| 0| OMPT_VERBOSE_INIT_PRINT("Continuing search...\n"); ------------------ | | 53| 0| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 0] | | ------------------ | | 54| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 342| 0| } 343| 0| OMPT_DLCLOSE(h); ------------------ | | 110| 0|#define OMPT_DLCLOSE(Lib) dlclose(Lib) ------------------ 344| 0| } 345| 0| fname = __kmp_str_token(NULL, sep, &buf); 346| 0| } 347| 0| __kmp_str_free(&libs); 348| 1| } else { 349| 1| OMPT_VERBOSE_INIT_PRINT("No OMP_TOOL_LIBRARIES defined.\n"); ------------------ | | 53| 1| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 1] | | ------------------ | | 54| 1| fprintf(verbose_file, __VA_ARGS__) ------------------ 350| 1| } 351| | 352| | // usable tool found in tool-libraries 353| 1| if (ret) { ------------------ | Branch (353:7): [True: 0, False: 1] ------------------ 354| 0| OMPT_VERBOSE_INIT_PRINT("----- END LOGGING OF TOOL REGISTRATION -----\n"); ------------------ | | 53| 0| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 0] | | ------------------ | | 54| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 355| 0| return ret; 356| 0| } 357| | 358| 1|#if KMP_OS_UNIX 359| 1| { // Non-standard: load archer tool if application is built with TSan 360| 1| const char *fname = "libarcher.so"; 361| 1| OMPT_VERBOSE_INIT_PRINT( ------------------ | | 53| 1| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 1] | | ------------------ | | 54| 1| fprintf(verbose_file, __VA_ARGS__) ------------------ 362| 1| "...searching tool libraries failed. Using archer tool.\n"); 363| 1| OMPT_VERBOSE_INIT_PRINT("Opening %s... ", fname); ------------------ | | 53| 1| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 1] | | ------------------ | | 54| 1| fprintf(verbose_file, __VA_ARGS__) ------------------ 364| 1| void *h = dlopen(fname, RTLD_LAZY); 365| 1| if (h) { ------------------ | Branch (365:9): [True: 0, False: 1] ------------------ 366| 0| OMPT_VERBOSE_INIT_CONTINUED_PRINT("Success.\n"); ------------------ | | 56| 0| if (verbose_init) \ | | ------------------ | | | Branch (56:7): [True: 0, False: 0] | | ------------------ | | 57| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 367| 0| OMPT_VERBOSE_INIT_PRINT("Searching for ompt_start_tool in %s... ", fname); ------------------ | | 53| 0| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 0] | | ------------------ | | 54| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 368| 0| start_tool = (ompt_start_tool_t)dlsym(h, "ompt_start_tool"); 369| 0| if (start_tool) { ------------------ | Branch (369:11): [True: 0, False: 0] ------------------ 370| 0| ret = (*start_tool)(omp_version, runtime_version); 371| 0| if (ret) { ------------------ | Branch (371:13): [True: 0, False: 0] ------------------ 372| 0| OMPT_VERBOSE_INIT_CONTINUED_PRINT("Success.\n"); ------------------ | | 56| 0| if (verbose_init) \ | | ------------------ | | | Branch (56:7): [True: 0, False: 0] | | ------------------ | | 57| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 373| 0| OMPT_VERBOSE_INIT_PRINT( ------------------ | | 53| 0| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 0] | | ------------------ | | 54| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 374| 0| "Tool was started and is using the OMPT interface.\n"); 375| 0| OMPT_VERBOSE_INIT_PRINT( ------------------ | | 53| 0| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 0] | | ------------------ | | 54| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 376| 0| "----- END LOGGING OF TOOL REGISTRATION -----\n"); 377| 0| return ret; 378| 0| } 379| 0| OMPT_VERBOSE_INIT_CONTINUED_PRINT( ------------------ | | 56| 0| if (verbose_init) \ | | ------------------ | | | Branch (56:7): [True: 0, False: 0] | | ------------------ | | 57| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 380| 0| "Found but not using the OMPT interface.\n"); 381| 0| } else { 382| 0| OMPT_VERBOSE_INIT_CONTINUED_PRINT("Failed: %s\n", dlerror()); ------------------ | | 56| 0| if (verbose_init) \ | | ------------------ | | | Branch (56:7): [True: 0, False: 0] | | ------------------ | | 57| 0| fprintf(verbose_file, __VA_ARGS__) ------------------ 383| 0| } 384| 0| } 385| 1| } 386| 1|#endif 387| 1| OMPT_VERBOSE_INIT_PRINT("No OMP tool loaded.\n"); ------------------ | | 53| 1| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 1] | | ------------------ | | 54| 1| fprintf(verbose_file, __VA_ARGS__) ------------------ 388| 1| OMPT_VERBOSE_INIT_PRINT("----- END LOGGING OF TOOL REGISTRATION -----\n"); ------------------ | | 53| 1| if (verbose_init) \ | | ------------------ | | | Branch (53:7): [True: 0, False: 1] | | ------------------ | | 54| 1| fprintf(verbose_file, __VA_ARGS__) ------------------ 389| 1| return ret; 390| 1|} _Z29__ompt_get_task_info_internaliPiPP11ompt_data_tPP12ompt_frame_tS2_S_: 361| 1| int *thread_num) { 362| 1| if (__kmp_get_gtid() < 0) ------------------ | | 3466| 1|#define __kmp_get_gtid() __kmp_get_global_thread_id() ------------------ | Branch (362:7): [True: 0, False: 1] ------------------ 363| 0| return 0; 364| | 365| 1| if (ancestor_level < 0) ------------------ | Branch (365:7): [True: 0, False: 1] ------------------ 366| 0| return 0; 367| | 368| | // copied from __ompt_get_scheduling_taskinfo 369| 1| ompt_task_info_t *info = NULL; 370| 1| ompt_team_info_t *team_info = NULL; 371| 1| kmp_info_t *thr = ompt_get_thread(); 372| 1| int level = ancestor_level; 373| | 374| 1| if (thr) { ------------------ | Branch (374:7): [True: 1, False: 0] ------------------ 375| 1| kmp_taskdata_t *taskdata = thr->th.th_current_task; 376| 1| if (taskdata == NULL) ------------------ | Branch (376:9): [True: 0, False: 1] ------------------ 377| 0| return 0; 378| 1| kmp_team *team = thr->th.th_team, *prev_team = NULL; 379| 1| if (team == NULL) ------------------ | Branch (379:9): [True: 0, False: 1] ------------------ 380| 0| return 0; 381| 1| ompt_lw_taskteam_t *lwt = NULL, 382| 1| *next_lwt = LWT_FROM_TEAM(taskdata->td_team); ------------------ | | 36| 1|#define LWT_FROM_TEAM(team) (team)->t.ompt_serialized_team_info ------------------ 383| | 384| 1| while (ancestor_level > 0) { ------------------ | Branch (384:12): [True: 0, False: 1] ------------------ 385| | // next lightweight team (if any) 386| 0| if (lwt) ------------------ | Branch (386:11): [True: 0, False: 0] ------------------ 387| 0| lwt = lwt->parent; 388| | 389| | // next heavyweight team (if any) after 390| | // lightweight teams are exhausted 391| 0| if (!lwt && taskdata) { ------------------ | Branch (391:11): [True: 0, False: 0] | Branch (391:19): [True: 0, False: 0] ------------------ 392| | // first try scheduling parent (for explicit task scheduling) 393| 0| if (taskdata->ompt_task_info.scheduling_parent) { ------------------ | Branch (393:13): [True: 0, False: 0] ------------------ 394| 0| taskdata = taskdata->ompt_task_info.scheduling_parent; 395| 0| } else if (next_lwt) { ------------------ | Branch (395:20): [True: 0, False: 0] ------------------ 396| 0| lwt = next_lwt; 397| 0| next_lwt = NULL; 398| 0| } else { 399| | // then go for implicit tasks 400| 0| taskdata = taskdata->td_parent; 401| 0| if (team == NULL) ------------------ | Branch (401:15): [True: 0, False: 0] ------------------ 402| 0| return 0; 403| 0| prev_team = team; 404| 0| team = team->t.t_parent; 405| 0| if (taskdata) { ------------------ | Branch (405:15): [True: 0, False: 0] ------------------ 406| 0| next_lwt = LWT_FROM_TEAM(taskdata->td_team); ------------------ | | 36| 0|#define LWT_FROM_TEAM(team) (team)->t.ompt_serialized_team_info ------------------ 407| 0| } 408| 0| } 409| 0| } 410| 0| ancestor_level--; 411| 0| } 412| | 413| 1| if (lwt) { ------------------ | Branch (413:9): [True: 0, False: 1] ------------------ 414| 0| info = &lwt->ompt_task_info; 415| 0| team_info = &lwt->ompt_team_info; 416| 0| if (type) { ------------------ | Branch (416:11): [True: 0, False: 0] ------------------ 417| 0| *type = ompt_task_implicit; 418| 0| } 419| 1| } else if (taskdata) { ------------------ | Branch (419:16): [True: 1, False: 0] ------------------ 420| 1| info = &taskdata->ompt_task_info; 421| 1| team_info = &team->t.ompt_team_info; 422| 1| if (type) { ------------------ | Branch (422:11): [True: 0, False: 1] ------------------ 423| 0| if (taskdata->td_parent) { ------------------ | Branch (423:13): [True: 0, False: 0] ------------------ 424| 0| *type = (taskdata->td_flags.tasktype ? ompt_task_explicit ------------------ | Branch (424:20): [True: 0, False: 0] ------------------ 425| 0| : ompt_task_implicit) | 426| 0| TASK_TYPE_DETAILS_FORMAT(taskdata); ------------------ | | 48| 0| ((info->td_flags.task_serial || info->td_flags.tasking_ser) \ | | ------------------ | | | Branch (48:5): [True: 0, False: 0] | | | Branch (48:35): [True: 0, False: 0] | | ------------------ | | 49| 0| ? ompt_task_undeferred \ | | 50| 0| : 0x0) | \ | | 51| 0| ((!(info->td_flags.tiedness)) ? ompt_task_untied : 0x0) | \ | | ------------------ | | | Branch (51:8): [True: 0, False: 0] | | ------------------ | | 52| 0| (info->td_flags.final ? ompt_task_final : 0x0) | \ | | ------------------ | | | Branch (52:8): [True: 0, False: 0] | | ------------------ | | 53| 0| (info->td_flags.merged_if0 ? ompt_task_mergeable : 0x0) | | ------------------ | | | Branch (53:8): [True: 0, False: 0] | | ------------------ ------------------ 427| 0| } else { 428| 0| *type = ompt_task_initial; 429| 0| } 430| 0| } 431| 1| } 432| 1| if (task_data) { ------------------ | Branch (432:9): [True: 1, False: 0] ------------------ 433| 1| *task_data = info ? &info->task_data : NULL; ------------------ | Branch (433:20): [True: 1, False: 0] ------------------ 434| 1| } 435| 1| if (task_frame) { ------------------ | Branch (435:9): [True: 0, False: 1] ------------------ 436| | // OpenMP spec asks for the scheduling task to be returned. 437| 0| *task_frame = info ? &info->frame : NULL; ------------------ | Branch (437:21): [True: 0, False: 0] ------------------ 438| 0| } 439| 1| if (parallel_data) { ------------------ | Branch (439:9): [True: 1, False: 0] ------------------ 440| 1| *parallel_data = team_info ? &(team_info->parallel_data) : NULL; ------------------ | Branch (440:24): [True: 1, False: 0] ------------------ 441| 1| } 442| 1| if (thread_num) { ------------------ | Branch (442:9): [True: 0, False: 1] ------------------ 443| 0| if (level == 0) ------------------ | Branch (443:11): [True: 0, False: 0] ------------------ 444| 0| *thread_num = __kmp_get_tid(); ------------------ | | 3468| 0|#define __kmp_get_tid() (__kmp_tid_from_gtid(__kmp_get_gtid())) | | ------------------ | | | | 3466| 0|#define __kmp_get_gtid() __kmp_get_global_thread_id() | | ------------------ ------------------ 445| 0| else if (lwt) ------------------ | Branch (445:16): [True: 0, False: 0] ------------------ 446| 0| *thread_num = 0; 447| 0| else if (!prev_team) { ------------------ | Branch (447:16): [True: 0, False: 0] ------------------ 448| | // The innermost parallel region contains at least one explicit task. 449| | // The task at level > 0 is either an implicit task that 450| | // corresponds to the mentioned region or one of the explicit tasks 451| | // nested inside the same region. Note that the task isn't the 452| | // innermost explicit tasks (because of condition level > 0). 453| | // Since the task at this level still belongs to the innermost parallel 454| | // region, thread_num is determined the same way as for level==0. 455| 0| *thread_num = __kmp_get_tid(); ------------------ | | 3468| 0|#define __kmp_get_tid() (__kmp_tid_from_gtid(__kmp_get_gtid())) | | ------------------ | | | | 3466| 0|#define __kmp_get_gtid() __kmp_get_global_thread_id() | | ------------------ ------------------ 456| 0| } else 457| 0| *thread_num = prev_team->t.t_master_tid; 458| | // *thread_num = team->t.t_master_tid; 459| 0| } 460| 1| return info ? 2 : 0; ------------------ | Branch (460:12): [True: 1, False: 0] ------------------ 461| 1| } 462| 0| return 0; 463| 1|} _Z21__ompt_team_assign_idP8kmp_team11ompt_data_t: 501| 1.38k|void __ompt_team_assign_id(kmp_team_t *team, ompt_data_t ompt_pid) { 502| 1.38k| team->t.ompt_team_info.parallel_data = ompt_pid; 503| 1.38k|} _ZN22OmptReturnAddressGuardC2EiPv: 139| 153k| OmptReturnAddressGuard(int Gtid, void *ReturnAddress) : Gtid(Gtid) { 140| 153k| if (ompt_enabled.enabled && Gtid >= 0 && __kmp_threads[Gtid] && ------------------ | Branch (140:9): [True: 0, False: 153k] | Branch (140:33): [True: 0, False: 0] | Branch (140:46): [True: 0, False: 0] ------------------ 141| 0| !__kmp_threads[Gtid]->th.ompt_thread_info.return_address) { ------------------ | Branch (141:9): [True: 0, False: 0] ------------------ 142| 0| SetAddress = true; 143| 0| __kmp_threads[Gtid]->th.ompt_thread_info.return_address = ReturnAddress; 144| 0| } 145| 153k| } _ZN22OmptReturnAddressGuardD2Ev: 146| 152k| ~OmptReturnAddressGuard() { 147| 152k| if (SetAddress) ------------------ | Branch (147:9): [True: 0, False: 152k] ------------------ 148| 0| __kmp_threads[Gtid]->th.ompt_thread_info.return_address = NULL; 149| 152k| } _Z26__ompt_load_return_addressi: 79| 115k|inline void *__ompt_load_return_address(int gtid) { 80| 115k| kmp_info_t *thr = __kmp_threads[gtid]; 81| 115k| void *return_address = thr->th.ompt_thread_info.return_address; 82| | thr->th.ompt_thread_info.return_address = NULL; 83| 115k| return return_address; 84| 115k|} _Z20ompt_get_thread_gtidi: 115| 1|inline kmp_info_t *ompt_get_thread_gtid(int gtid) { 116| 1| return (gtid >= 0) ? __kmp_thread_from_gtid(gtid) : NULL; ------------------ | Branch (116:10): [True: 1, False: 0] ------------------ 117| 1|} _Z15ompt_get_threadv: 119| 1|inline kmp_info_t *ompt_get_thread() { 120| 1| int gtid = __kmp_get_gtid(); ------------------ | | 3466| 1|#define __kmp_get_gtid() __kmp_get_global_thread_id() ------------------ 121| 1| return ompt_get_thread_gtid(gtid); 122| 1|} _Z24ompt_get_runtime_versionv: 129| 1|inline const char *ompt_get_runtime_version() { 130| 1| return &__kmp_version_lib_ver[KMP_VERSION_MAGIC_LEN]; ------------------ | | 32| 1|#define KMP_VERSION_MAGIC_LEN 6 // Length of KMP_VERSION_MAGIC_STR. ------------------ 131| 1|} __kmp_affinity_bind_thread: 109| 32|void __kmp_affinity_bind_thread(int which) { 110| 32| KMP_ASSERT2(KMP_AFFINITY_CAPABLE(), ------------------ | | 53| 32|#define KMP_ASSERT2(cond, msg) KMP_ASSERT(cond) | | ------------------ | | | | 50| 32| if (!(cond)) { \ | | | | ------------------ | | | | | Branch (50:7): [True: 0, False: 32] | | | | ------------------ | | | | 51| 0| __kmp_debug_assert("assertion failure", __FILE__, __LINE__); \ | | | | 52| 0| } | | ------------------ ------------------ 111| 32| "Illegal set affinity operation when not capable"); 112| | 113| 32| kmp_affin_mask_t *mask; 114| 32| KMP_CPU_ALLOC_ON_STACK(mask); ------------------ | | 699| 32|#define KMP_CPU_ALLOC_ON_STACK(ptr) KMP_CPU_ALLOC(ptr) | | ------------------ | | | | 697| 32|#define KMP_CPU_ALLOC(ptr) (ptr = __kmp_affinity_dispatch->allocate_mask()) | | ------------------ ------------------ 115| 32| KMP_CPU_ZERO(mask); ------------------ | | 692| 32|#define KMP_CPU_ZERO(mask) (mask)->zero() ------------------ 116| 32| KMP_CPU_SET(which, mask); ------------------ | | 689| 32|#define KMP_CPU_SET(i, mask) (mask)->set(i) ------------------ 117| 32| __kmp_set_system_affinity(mask, TRUE); ------------------ | | 713| 32| (mask)->set_system_affinity(abort_bool) ------------------ 118| 32| KMP_CPU_FREE_FROM_STACK(mask); ------------------ | | 700| 32|#define KMP_CPU_FREE_FROM_STACK(ptr) KMP_CPU_FREE(ptr) | | ------------------ | | | | 698| 32|#define KMP_CPU_FREE(ptr) __kmp_affinity_dispatch->deallocate_mask(ptr) | | ------------------ ------------------ 119| 32|} __kmp_affinity_determine_capable: 124| 1|void __kmp_affinity_determine_capable(const char *env_var) { 125| | // Check and see if the OS supports thread affinity. 126| | 127| 1|#if KMP_OS_LINUX 128| 1|#define KMP_CPU_SET_SIZE_LIMIT (1024 * 1024) 129| 1|#define KMP_CPU_SET_TRY_SIZE CACHE_LINE 130| |#elif KMP_OS_FREEBSD 131| |#define KMP_CPU_SET_SIZE_LIMIT (sizeof(cpuset_t)) 132| |#endif 133| | 134| 1| int verbose = __kmp_affinity.flags.verbose; 135| 1| int warnings = __kmp_affinity.flags.warnings; 136| 1| enum affinity_type type = __kmp_affinity.type; 137| | 138| 1|#if KMP_OS_LINUX 139| 1| long gCode; 140| 1| unsigned char *buf; 141| 1| buf = (unsigned char *)KMP_INTERNAL_MALLOC(KMP_CPU_SET_SIZE_LIMIT); ------------------ | | 110| 1|#define KMP_INTERNAL_MALLOC(sz) malloc(sz) ------------------ 142| | 143| | // If the syscall returns a suggestion for the size, 144| | // then we don't have to search for an appropriate size. 145| 1| gCode = syscall(__NR_sched_getaffinity, 0, KMP_CPU_SET_TRY_SIZE, buf); ------------------ | | 129| 1|#define KMP_CPU_SET_TRY_SIZE CACHE_LINE | | ------------------ | | | | 108| 1|# define CACHE_LINE 64 | | ------------------ ------------------ 146| 1| KA_TRACE(30, ("__kmp_affinity_determine_capable: " 147| 1| "initial getaffinity call returned %ld errno = %d\n", 148| 1| gCode, errno)); 149| | 150| 1| if (gCode < 0 && errno != EINVAL) { ------------------ | Branch (150:7): [True: 0, False: 1] | Branch (150:20): [True: 0, False: 0] ------------------ 151| | // System call not supported 152| 0| if (verbose || ------------------ | Branch (152:9): [True: 0, False: 0] ------------------ 153| 0| (warnings && (type != affinity_none) && (type != affinity_default) && ------------------ | Branch (153:10): [True: 0, False: 0] | Branch (153:22): [True: 0, False: 0] | Branch (153:49): [True: 0, False: 0] ------------------ 154| 0| (type != affinity_disabled))) { ------------------ | Branch (154:10): [True: 0, False: 0] ------------------ 155| 0| int error = errno; 156| 0| kmp_msg_t err_code = KMP_ERR(error); ------------------ | | 125| 0|#define KMP_ERR KMP_SYSERRCODE | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ ------------------ 157| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(GetAffSysCallNotSupported, env_var), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ 158| 0| err_code, __kmp_msg_null); 159| 0| if (__kmp_generate_warnings == kmp_warnings_off) { ------------------ | Branch (159:11): [True: 0, False: 0] ------------------ 160| 0| __kmp_str_free(&err_code.str); 161| 0| } 162| 0| } 163| 0| KMP_AFFINITY_DISABLE(); ------------------ | | 685| 0|#define KMP_AFFINITY_DISABLE() (__kmp_affin_mask_size = 0) ------------------ 164| 0| KMP_INTERNAL_FREE(buf); ------------------ | | 111| 0|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 165| 0| return; 166| 1| } else if (gCode > 0) { ------------------ | Branch (166:14): [True: 1, False: 0] ------------------ 167| | // The optimal situation: the OS returns the size of the buffer it expects. 168| 1| KMP_AFFINITY_ENABLE(gCode); ------------------ | | 686| 1|#define KMP_AFFINITY_ENABLE(mask_size) (__kmp_affin_mask_size = mask_size) ------------------ 169| 1| KA_TRACE(10, ("__kmp_affinity_determine_capable: " 170| 1| "affinity supported (mask size %d)\n", 171| 1| (int)__kmp_affin_mask_size)); 172| 1| KMP_INTERNAL_FREE(buf); ------------------ | | 111| 1|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 173| 1| return; 174| 1| } 175| | 176| | // Call the getaffinity system call repeatedly with increasing set sizes 177| | // until we succeed, or reach an upper bound on the search. 178| 0| KA_TRACE(30, ("__kmp_affinity_determine_capable: " 179| 0| "searching for proper set size\n")); 180| 0| int size; 181| 0| for (size = 1; size <= KMP_CPU_SET_SIZE_LIMIT; size *= 2) { ------------------ | | 128| 0|#define KMP_CPU_SET_SIZE_LIMIT (1024 * 1024) ------------------ | Branch (181:18): [True: 0, False: 0] ------------------ 182| 0| gCode = syscall(__NR_sched_getaffinity, 0, size, buf); 183| 0| KA_TRACE(30, ("__kmp_affinity_determine_capable: " 184| 0| "getaffinity for mask size %ld returned %ld errno = %d\n", 185| 0| size, gCode, errno)); 186| | 187| 0| if (gCode < 0) { ------------------ | Branch (187:9): [True: 0, False: 0] ------------------ 188| 0| if (errno == ENOSYS) { ------------------ | Branch (188:11): [True: 0, False: 0] ------------------ 189| | // We shouldn't get here 190| 0| KA_TRACE(30, ("__kmp_affinity_determine_capable: " 191| 0| "inconsistent OS call behavior: errno == ENOSYS for mask " 192| 0| "size %d\n", 193| 0| size)); 194| 0| if (verbose || ------------------ | Branch (194:13): [True: 0, False: 0] ------------------ 195| 0| (warnings && (type != affinity_none) && ------------------ | Branch (195:14): [True: 0, False: 0] | Branch (195:26): [True: 0, False: 0] ------------------ 196| 0| (type != affinity_default) && (type != affinity_disabled))) { ------------------ | Branch (196:14): [True: 0, False: 0] | Branch (196:44): [True: 0, False: 0] ------------------ 197| 0| int error = errno; 198| 0| kmp_msg_t err_code = KMP_ERR(error); ------------------ | | 125| 0|#define KMP_ERR KMP_SYSERRCODE | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ ------------------ 199| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(GetAffSysCallNotSupported, env_var), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ 200| 0| err_code, __kmp_msg_null); 201| 0| if (__kmp_generate_warnings == kmp_warnings_off) { ------------------ | Branch (201:15): [True: 0, False: 0] ------------------ 202| 0| __kmp_str_free(&err_code.str); 203| 0| } 204| 0| } 205| 0| KMP_AFFINITY_DISABLE(); ------------------ | | 685| 0|#define KMP_AFFINITY_DISABLE() (__kmp_affin_mask_size = 0) ------------------ 206| 0| KMP_INTERNAL_FREE(buf); ------------------ | | 111| 0|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 207| 0| return; 208| 0| } 209| 0| continue; 210| 0| } 211| | 212| 0| KMP_AFFINITY_ENABLE(gCode); ------------------ | | 686| 0|#define KMP_AFFINITY_ENABLE(mask_size) (__kmp_affin_mask_size = mask_size) ------------------ 213| 0| KA_TRACE(10, ("__kmp_affinity_determine_capable: " 214| 0| "affinity supported (mask size %d)\n", 215| 0| (int)__kmp_affin_mask_size)); 216| 0| KMP_INTERNAL_FREE(buf); ------------------ | | 111| 0|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 217| 0| return; 218| 0| } 219| |#elif KMP_OS_FREEBSD 220| | long gCode; 221| | unsigned char *buf; 222| | buf = (unsigned char *)KMP_INTERNAL_MALLOC(KMP_CPU_SET_SIZE_LIMIT); 223| | gCode = pthread_getaffinity_np(pthread_self(), KMP_CPU_SET_SIZE_LIMIT, 224| | reinterpret_cast(buf)); 225| | KA_TRACE(30, ("__kmp_affinity_determine_capable: " 226| | "initial getaffinity call returned %d errno = %d\n", 227| | gCode, errno)); 228| | if (gCode == 0) { 229| | KMP_AFFINITY_ENABLE(KMP_CPU_SET_SIZE_LIMIT); 230| | KA_TRACE(10, ("__kmp_affinity_determine_capable: " 231| | "affinity supported (mask size %d)\n", 232| | (int)__kmp_affin_mask_size)); 233| | KMP_INTERNAL_FREE(buf); 234| | return; 235| | } 236| |#endif 237| 0| KMP_INTERNAL_FREE(buf); ------------------ | | 111| 0|#define KMP_INTERNAL_FREE(p) free(p) ------------------ 238| | 239| | // Affinity is not supported 240| 0| KMP_AFFINITY_DISABLE(); ------------------ | | 685| 0|#define KMP_AFFINITY_DISABLE() (__kmp_affin_mask_size = 0) ------------------ 241| 0| KA_TRACE(10, ("__kmp_affinity_determine_capable: " 242| 0| "cannot determine mask size - affinity not supported\n")); 243| 0| if (verbose || (warnings && (type != affinity_none) && ------------------ | Branch (243:7): [True: 0, False: 0] | Branch (243:19): [True: 0, False: 0] | Branch (243:31): [True: 0, False: 0] ------------------ 244| 0| (type != affinity_default) && (type != affinity_disabled))) { ------------------ | Branch (244:19): [True: 0, False: 0] | Branch (244:49): [True: 0, False: 0] ------------------ 245| 0| KMP_WARNING(AffCantGetMaskSize, env_var); ------------------ | | 145| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(__VA_ARGS__), __kmp_msg_null) | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ ------------------ 246| 0| } 247| 0|} __kmp_create_worker: 710| 32|void __kmp_create_worker(int gtid, kmp_info_t *th, size_t stack_size) { 711| 32| pthread_t handle; 712| 32| pthread_attr_t thread_attr; 713| 32| int status; 714| | 715| 32| th->th.th_info.ds.ds_gtid = gtid; 716| | 717| |#if KMP_STATS_ENABLED 718| | // sets up worker thread stats 719| | __kmp_acquire_tas_lock(&__kmp_stats_lock, gtid); 720| | 721| | // th->th.th_stats is used to transfer thread-specific stats-pointer to 722| | // __kmp_launch_worker. So when thread is created (goes into 723| | // __kmp_launch_worker) it will set its thread local pointer to 724| | // th->th.th_stats 725| | if (!KMP_UBER_GTID(gtid)) { 726| | th->th.th_stats = __kmp_stats_list->push_back(gtid); 727| | } else { 728| | // For root threads, __kmp_stats_thread_ptr is set in __kmp_register_root(), 729| | // so set the th->th.th_stats field to it. 730| | th->th.th_stats = __kmp_stats_thread_ptr; 731| | } 732| | __kmp_release_tas_lock(&__kmp_stats_lock, gtid); 733| | 734| |#endif // KMP_STATS_ENABLED 735| | 736| 32| if (KMP_UBER_GTID(gtid)) { ------------------ | Branch (736:7): [True: 1, False: 31] ------------------ 737| 1| KA_TRACE(10, ("__kmp_create_worker: uber thread (%d)\n", gtid)); 738| 1| th->th.th_info.ds.ds_thread = pthread_self(); 739| 1| __kmp_set_stack_info(gtid, th); 740| 1| __kmp_check_stack_overlap(th); 741| 1| return; 742| 1| } 743| | 744| 31| KA_TRACE(10, ("__kmp_create_worker: try to create thread (%d)\n", gtid)); 745| | 746| 31| KMP_MB(); /* Flush all pending memory write invalidates. */ 747| | 748| 31|#ifdef KMP_THREAD_ATTR 749| 31| status = pthread_attr_init(&thread_attr); 750| 31| if (status != 0) { ------------------ | Branch (750:7): [True: 0, False: 31] ------------------ 751| 0| __kmp_fatal(KMP_MSG(CantInitThreadAttrs), KMP_ERR(status), __kmp_msg_null); ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ __kmp_fatal(KMP_MSG(CantInitThreadAttrs), KMP_ERR(status), __kmp_msg_null); ------------------ | | 125| 0|#define KMP_ERR KMP_SYSERRCODE | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ ------------------ 752| 0| } 753| 31| status = pthread_attr_setdetachstate(&thread_attr, PTHREAD_CREATE_JOINABLE); 754| 31| if (status != 0) { ------------------ | Branch (754:7): [True: 0, False: 31] ------------------ 755| 0| __kmp_fatal(KMP_MSG(CantSetWorkerState), KMP_ERR(status), __kmp_msg_null); ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ __kmp_fatal(KMP_MSG(CantSetWorkerState), KMP_ERR(status), __kmp_msg_null); ------------------ | | 125| 0|#define KMP_ERR KMP_SYSERRCODE | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ ------------------ 756| 0| } 757| | 758| | /* Set stack size for this thread now. 759| | The multiple of 2 is there because on some machines, requesting an unusual 760| | stacksize causes the thread to have an offset before the dummy alloca() 761| | takes place to create the offset. Since we want the user to have a 762| | sufficient stacksize AND support a stack offset, we alloca() twice the 763| | offset so that the upcoming alloca() does not eliminate any premade offset, 764| | and also gives the user the stack space they requested for all threads */ 765| 31| stack_size += gtid * __kmp_stkoffset * 2; 766| | 767| 31| KA_TRACE(10, ("__kmp_create_worker: T#%d, default stacksize = %lu bytes, " 768| 31| "__kmp_stksize = %lu bytes, final stacksize = %lu bytes\n", 769| 31| gtid, KMP_DEFAULT_STKSIZE, __kmp_stksize, stack_size)); 770| | 771| 31|#ifdef _POSIX_THREAD_ATTR_STACKSIZE 772| 31| status = pthread_attr_setstacksize(&thread_attr, stack_size); 773| 31|#ifdef KMP_BACKUP_STKSIZE 774| 31| if (status != 0) { ------------------ | Branch (774:7): [True: 0, False: 31] ------------------ 775| 0| if (!__kmp_env_stksize) { ------------------ | Branch (775:9): [True: 0, False: 0] ------------------ 776| 0| stack_size = KMP_BACKUP_STKSIZE + gtid * __kmp_stkoffset; ------------------ | | 1159| 0|#define KMP_BACKUP_STKSIZE ((size_t)(2 * 1024 * 1024)) ------------------ 777| 0| __kmp_stksize = KMP_BACKUP_STKSIZE; ------------------ | | 1159| 0|#define KMP_BACKUP_STKSIZE ((size_t)(2 * 1024 * 1024)) ------------------ 778| 0| KA_TRACE(10, ("__kmp_create_worker: T#%d, default stacksize = %lu bytes, " 779| 0| "__kmp_stksize = %lu bytes, (backup) final stacksize = %lu " 780| 0| "bytes\n", 781| 0| gtid, KMP_DEFAULT_STKSIZE, __kmp_stksize, stack_size)); 782| 0| status = pthread_attr_setstacksize(&thread_attr, stack_size); 783| 0| } 784| 0| } 785| 31|#endif /* KMP_BACKUP_STKSIZE */ 786| 31| if (status != 0) { ------------------ | Branch (786:7): [True: 0, False: 31] ------------------ 787| 0| __kmp_fatal(KMP_MSG(CantSetWorkerStackSize, stack_size), KMP_ERR(status), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ __kmp_fatal(KMP_MSG(CantSetWorkerStackSize, stack_size), KMP_ERR(status), ------------------ | | 125| 0|#define KMP_ERR KMP_SYSERRCODE | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ ------------------ 788| 0| KMP_HNT(ChangeWorkerStackSize), __kmp_msg_null); ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 789| 0| } 790| 31|#endif /* _POSIX_THREAD_ATTR_STACKSIZE */ 791| | 792| 31|#endif /* KMP_THREAD_ATTR */ 793| | 794| 31| status = 795| 31| pthread_create(&handle, &thread_attr, __kmp_launch_worker, (void *)th); 796| 31| if (status != 0 || !handle) { // ??? Why do we check handle?? ------------------ | Branch (796:7): [True: 0, False: 31] | Branch (796:22): [True: 0, False: 31] ------------------ 797| 0|#ifdef _POSIX_THREAD_ATTR_STACKSIZE 798| 0| if (status == EINVAL) { ------------------ | Branch (798:9): [True: 0, False: 0] ------------------ 799| 0| __kmp_fatal(KMP_MSG(CantSetWorkerStackSize, stack_size), KMP_ERR(status), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ __kmp_fatal(KMP_MSG(CantSetWorkerStackSize, stack_size), KMP_ERR(status), ------------------ | | 125| 0|#define KMP_ERR KMP_SYSERRCODE | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ ------------------ 800| 0| KMP_HNT(IncreaseWorkerStackSize), __kmp_msg_null); ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 801| 0| } 802| 0| if (status == ENOMEM) { ------------------ | Branch (802:9): [True: 0, False: 0] ------------------ 803| 0| __kmp_fatal(KMP_MSG(CantSetWorkerStackSize, stack_size), KMP_ERR(status), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ __kmp_fatal(KMP_MSG(CantSetWorkerStackSize, stack_size), KMP_ERR(status), ------------------ | | 125| 0|#define KMP_ERR KMP_SYSERRCODE | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ ------------------ 804| 0| KMP_HNT(DecreaseWorkerStackSize), __kmp_msg_null); ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 805| 0| } 806| 0|#endif /* _POSIX_THREAD_ATTR_STACKSIZE */ 807| 0| if (status == EAGAIN) { ------------------ | Branch (807:9): [True: 0, False: 0] ------------------ 808| 0| __kmp_fatal(KMP_MSG(NoResourcesForWorkerThread), KMP_ERR(status), ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ __kmp_fatal(KMP_MSG(NoResourcesForWorkerThread), KMP_ERR(status), ------------------ | | 125| 0|#define KMP_ERR KMP_SYSERRCODE | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ ------------------ 809| 0| KMP_HNT(Decrease_NUM_THREADS), __kmp_msg_null); ------------------ | | 122| 0|#define KMP_HNT(...) __kmp_msg_format(kmp_i18n_hnt_##__VA_ARGS__) ------------------ 810| 0| } 811| 0| KMP_SYSFAIL("pthread_create", status); ------------------ | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ | | 149| 0| __kmp_msg_null) ------------------ 812| 0| } 813| | 814| 31| th->th.th_info.ds.ds_thread = handle; 815| | 816| 31|#ifdef KMP_THREAD_ATTR 817| 31| status = pthread_attr_destroy(&thread_attr); 818| 31| if (status) { ------------------ | Branch (818:7): [True: 0, False: 31] ------------------ 819| 0| kmp_msg_t err_code = KMP_ERR(status); ------------------ | | 125| 0|#define KMP_ERR KMP_SYSERRCODE | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ ------------------ 820| 0| __kmp_msg(kmp_ms_warning, KMP_MSG(CantDestroyThreadAttrs), err_code, ------------------ | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) ------------------ 821| 0| __kmp_msg_null); 822| 0| if (__kmp_generate_warnings == kmp_warnings_off) { ------------------ | Branch (822:9): [True: 0, False: 0] ------------------ 823| 0| __kmp_str_free(&err_code.str); 824| 0| } 825| 0| } 826| 31|#endif /* KMP_THREAD_ATTR */ 827| | 828| 31| KMP_MB(); /* Flush all pending memory write invalidates. */ 829| | 830| 31| KA_TRACE(10, ("__kmp_create_worker: done creating thread (%d)\n", gtid)); 831| | 832| 31|} // __kmp_create_worker __kmp_reap_worker: 1032| 31|void __kmp_reap_worker(kmp_info_t *th) { 1033| 31| int status; 1034| 31| void *exit_val; 1035| | 1036| 31| KMP_MB(); /* Flush all pending memory write invalidates. */ 1037| | 1038| 31| KA_TRACE( 1039| 31| 10, ("__kmp_reap_worker: try to reap T#%d\n", th->th.th_info.ds.ds_gtid)); 1040| | 1041| 31| status = pthread_join(th->th.th_info.ds.ds_thread, &exit_val); 1042| |#ifdef KMP_DEBUG 1043| | /* Don't expose these to the user until we understand when they trigger */ 1044| | if (status != 0) { 1045| | __kmp_fatal(KMP_MSG(ReapWorkerError), KMP_ERR(status), __kmp_msg_null); 1046| | } 1047| | if (exit_val != th) { 1048| | KA_TRACE(10, ("__kmp_reap_worker: worker T#%d did not reap properly, " 1049| | "exit_val = %p\n", 1050| | th->th.th_info.ds.ds_gtid, exit_val)); 1051| | } 1052| |#else 1053| 31| (void)status; // unused variable 1054| 31|#endif /* KMP_DEBUG */ 1055| | 1056| 31| KA_TRACE(10, ("__kmp_reap_worker: done reaping T#%d\n", 1057| 31| th->th.th_info.ds.ds_gtid)); 1058| | 1059| 31| KMP_MB(); /* Flush all pending memory write invalidates. */ 1060| 31|} __kmp_install_signals: 1156| 2|void __kmp_install_signals(int parallel_init) { 1157| 2| KB_TRACE(10, ("__kmp_install_signals( %d )\n", parallel_init)); 1158| 2| if (__kmp_handle_signals || !parallel_init) { ------------------ | Branch (1158:7): [True: 0, False: 2] | Branch (1158:31): [True: 1, False: 1] ------------------ 1159| | // If ! parallel_init, we do not install handlers, just save original 1160| | // handlers. Let us do it even __handle_signals is 0. 1161| 1| sigemptyset(&__kmp_sigset); 1162| 1| __kmp_install_one_handler(SIGHUP, __kmp_team_handler, parallel_init); 1163| 1| __kmp_install_one_handler(SIGINT, __kmp_team_handler, parallel_init); 1164| 1| __kmp_install_one_handler(SIGQUIT, __kmp_team_handler, parallel_init); 1165| 1| __kmp_install_one_handler(SIGILL, __kmp_team_handler, parallel_init); 1166| 1| __kmp_install_one_handler(SIGABRT, __kmp_team_handler, parallel_init); 1167| 1| __kmp_install_one_handler(SIGFPE, __kmp_team_handler, parallel_init); 1168| 1| __kmp_install_one_handler(SIGBUS, __kmp_team_handler, parallel_init); 1169| 1| __kmp_install_one_handler(SIGSEGV, __kmp_team_handler, parallel_init); 1170| 1|#ifdef SIGSYS 1171| 1| __kmp_install_one_handler(SIGSYS, __kmp_team_handler, parallel_init); 1172| 1|#endif // SIGSYS 1173| 1| __kmp_install_one_handler(SIGTERM, __kmp_team_handler, parallel_init); 1174| 1|#ifdef SIGPIPE 1175| | __kmp_install_one_handler(SIGPIPE, __kmp_team_handler, parallel_init); 1176| 1|#endif // SIGPIPE 1177| 1| } 1178| 2|} // __kmp_install_signals __kmp_remove_signals: 1180| 1|void __kmp_remove_signals(void) { 1181| 1| int sig; 1182| 1| KB_TRACE(10, ("__kmp_remove_signals()\n")); 1183| 65| for (sig = 1; sig < NSIG; ++sig) { ------------------ | Branch (1183:17): [True: 64, False: 1] ------------------ 1184| 64| __kmp_remove_one_handler(sig); 1185| 64| } 1186| 1|} // __kmp_remove_signals __kmp_register_atfork: 1318| 1|void __kmp_register_atfork(void) { 1319| 1| if (__kmp_need_register_atfork) { ------------------ | Branch (1319:7): [True: 1, False: 0] ------------------ 1320| 1| int status = pthread_atfork(__kmp_atfork_prepare, __kmp_atfork_parent, 1321| 1| __kmp_atfork_child); 1322| 1| KMP_CHECK_SYSFAIL("pthread_atfork", status); ------------------ | | 153| 1| { \ | | 154| 1| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 1] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 1| } ------------------ 1323| 1| __kmp_need_register_atfork = FALSE; ------------------ | | 1279| 1|#define FALSE 0 ------------------ 1324| 1| } 1325| 1|} __kmp_suspend_initialize: 1327| 1|void __kmp_suspend_initialize(void) { 1328| 1| int status; 1329| 1| status = pthread_mutexattr_init(&__kmp_suspend_mutex_attr); 1330| 1| KMP_CHECK_SYSFAIL("pthread_mutexattr_init", status); ------------------ | | 153| 1| { \ | | 154| 1| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 1] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 1| } ------------------ 1331| 1| status = pthread_condattr_init(&__kmp_suspend_cond_attr); 1332| 1| KMP_CHECK_SYSFAIL("pthread_condattr_init", status); ------------------ | | 153| 1| { \ | | 154| 1| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 1] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 1| } ------------------ 1333| 1|} __kmp_suspend_initialize_thread: 1335| 3.20k|void __kmp_suspend_initialize_thread(kmp_info_t *th) { 1336| 3.20k| int old_value = KMP_ATOMIC_LD_RLX(&th->th.th_suspend_init_count); ------------------ | | 1250| 3.20k|#define KMP_ATOMIC_LD_RLX(p) KMP_ATOMIC_LD(p, relaxed) | | ------------------ | | | | 1245| 3.20k|#define KMP_ATOMIC_LD(p, order) (p)->load(std::memory_order_##order) | | ------------------ ------------------ 1337| 3.20k| int new_value = __kmp_fork_count + 1; 1338| | // Return if already initialized 1339| 3.20k| if (old_value == new_value) ------------------ | Branch (1339:7): [True: 3.17k, False: 27] ------------------ 1340| 3.17k| return; 1341| | // Wait, then return if being initialized 1342| 32| if (old_value == -1 || !__kmp_atomic_compare_store( ------------------ | Branch (1342:7): [True: 18.4E, False: 32] | Branch (1342:26): [True: 0, False: 32] ------------------ 1343| 32| &th->th.th_suspend_init_count, old_value, -1)) { 1344| 0| while (KMP_ATOMIC_LD_ACQ(&th->th.th_suspend_init_count) != new_value) { ------------------ | | 1249| 0|#define KMP_ATOMIC_LD_ACQ(p) KMP_ATOMIC_LD(p, acquire) | | ------------------ | | | | 1245| 0|#define KMP_ATOMIC_LD(p, order) (p)->load(std::memory_order_##order) | | ------------------ ------------------ | Branch (1344:12): [True: 0, False: 0] ------------------ 1345| 0| KMP_CPU_PAUSE(); ------------------ | | 1498| 0|#define KMP_CPU_PAUSE() __kmp_x86_pause() ------------------ 1346| 0| } 1347| 27| } else { 1348| | // Claim to be the initializer and do initializations 1349| 27| int status; 1350| 27| status = pthread_cond_init(&th->th.th_suspend_cv.c_cond, 1351| 27| &__kmp_suspend_cond_attr); 1352| 27| KMP_CHECK_SYSFAIL("pthread_cond_init", status); ------------------ | | 153| 27| { \ | | 154| 27| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 27] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 27| } ------------------ 1353| 27| status = pthread_mutex_init(&th->th.th_suspend_mx.m_mutex, 1354| 27| &__kmp_suspend_mutex_attr); 1355| 27| KMP_CHECK_SYSFAIL("pthread_mutex_init", status); ------------------ | | 153| 27| { \ | | 154| 27| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 27] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 27| } ------------------ 1356| 27| KMP_ATOMIC_ST_REL(&th->th.th_suspend_init_count, new_value); ------------------ | | 1251| 27|#define KMP_ATOMIC_ST_REL(p, v) KMP_ATOMIC_OP(store, p, v, release) | | ------------------ | | | | 1246| 27|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 1357| 27| } 1358| 27|} __kmp_suspend_uninitialize_thread: 1360| 32|void __kmp_suspend_uninitialize_thread(kmp_info_t *th) { 1361| 32| if (KMP_ATOMIC_LD_ACQ(&th->th.th_suspend_init_count) > __kmp_fork_count) { ------------------ | | 1249| 32|#define KMP_ATOMIC_LD_ACQ(p) KMP_ATOMIC_LD(p, acquire) | | ------------------ | | | | 1245| 32|#define KMP_ATOMIC_LD(p, order) (p)->load(std::memory_order_##order) | | ------------------ ------------------ | Branch (1361:7): [True: 32, False: 0] ------------------ 1362| | /* this means we have initialize the suspension pthread objects for this 1363| | thread in this instance of the process */ 1364| 32| int status; 1365| | 1366| 32| status = pthread_cond_destroy(&th->th.th_suspend_cv.c_cond); 1367| 32| if (status != 0 && status != EBUSY) { ------------------ | Branch (1367:9): [True: 0, False: 32] | Branch (1367:24): [True: 0, False: 0] ------------------ 1368| 0| KMP_SYSFAIL("pthread_cond_destroy", status); ------------------ | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ | | 149| 0| __kmp_msg_null) ------------------ 1369| 0| } 1370| 32| status = pthread_mutex_destroy(&th->th.th_suspend_mx.m_mutex); 1371| 32| if (status != 0 && status != EBUSY) { ------------------ | Branch (1371:9): [True: 0, False: 32] | Branch (1371:24): [True: 0, False: 0] ------------------ 1372| 0| KMP_SYSFAIL("pthread_mutex_destroy", status); ------------------ | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ | | 149| 0| __kmp_msg_null) ------------------ 1373| 0| } 1374| 32| --th->th.th_suspend_init_count; 1375| 32| KMP_DEBUG_ASSERT(KMP_ATOMIC_LD_RLX(&th->th.th_suspend_init_count) == 1376| 32| __kmp_fork_count); 1377| 32| } 1378| 32|} __kmp_lock_suspend_mx: 1385| 3.21k|void __kmp_lock_suspend_mx(kmp_info_t *th) { 1386| 3.21k| int status = pthread_mutex_lock(&th->th.th_suspend_mx.m_mutex); 1387| 3.21k| KMP_CHECK_SYSFAIL("pthread_mutex_lock", status); ------------------ | | 153| 3.21k| { \ | | 154| 3.21k| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 3.21k] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 3.21k| } ------------------ 1388| 3.21k|} __kmp_unlock_suspend_mx: 1390| 3.23k|void __kmp_unlock_suspend_mx(kmp_info_t *th) { 1391| 3.23k| int status = pthread_mutex_unlock(&th->th.th_suspend_mx.m_mutex); 1392| 3.23k| KMP_CHECK_SYSFAIL("pthread_mutex_unlock", status); ------------------ | | 153| 3.23k| { \ | | 154| 3.23k| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 3.23k] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 3.23k| } ------------------ 1393| 3.23k|} _Z16__kmp_suspend_64ILb0ELb1EEviP11kmp_flag_64IXT_EXT0_EE: 1561| 1.62k|void __kmp_suspend_64(int th_gtid, kmp_flag_64 *flag) { 1562| 1.62k| __kmp_suspend_template(th_gtid, flag); 1563| 1.62k|} _Z15__kmp_resume_64ILb0ELb1EEviP11kmp_flag_64IXT_EXT0_EE: 1667| 1.55k|void __kmp_resume_64(int target_gtid, kmp_flag_64 *flag) { 1668| 1.55k| __kmp_resume_template(target_gtid, flag); 1669| 1.55k|} __kmp_yield: 1714| 2.07M|void __kmp_yield() { sched_yield(); } __kmp_gtid_set_specific: 1716| 32|void __kmp_gtid_set_specific(int gtid) { 1717| 32| if (__kmp_init_gtid) { ------------------ | Branch (1717:7): [True: 32, False: 0] ------------------ 1718| 32| int status; 1719| 32| status = pthread_setspecific(__kmp_gtid_threadprivate_key, 1720| 32| (void *)(intptr_t)(gtid + 1)); 1721| 32| KMP_CHECK_SYSFAIL("pthread_setspecific", status); ------------------ | | 153| 32| { \ | | 154| 32| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 32] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 32| } ------------------ 1722| 32| } else { 1723| 0| KA_TRACE(50, ("__kmp_gtid_set_specific: runtime shutdown, returning\n")); 1724| 0| } 1725| 32|} __kmp_gtid_get_specific: 1727| 2|int __kmp_gtid_get_specific() { 1728| 2| int gtid; 1729| 2| if (!__kmp_init_gtid) { ------------------ | Branch (1729:7): [True: 0, False: 2] ------------------ 1730| 0| KA_TRACE(50, ("__kmp_gtid_get_specific: runtime shutdown, returning " 1731| 0| "KMP_GTID_SHUTDOWN\n")); 1732| 0| return KMP_GTID_SHUTDOWN; ------------------ | | 1001| 0|#define KMP_GTID_SHUTDOWN (-3) /* Library is shutting down */ ------------------ 1733| 0| } 1734| 2| gtid = (int)(size_t)pthread_getspecific(__kmp_gtid_threadprivate_key); 1735| 2| if (gtid == 0) { ------------------ | Branch (1735:7): [True: 0, False: 2] ------------------ 1736| 0| gtid = KMP_GTID_DNE; ------------------ | | 1000| 0|#define KMP_GTID_DNE (-2) /* Does not exist */ ------------------ 1737| 2| } else { 1738| 2| gtid--; 1739| 2| } 1740| 2| KA_TRACE(50, ("__kmp_gtid_get_specific: key:%d gtid:%d\n", 1741| 2| __kmp_gtid_threadprivate_key, gtid)); 1742| 2| return gtid; 1743| 2|} __kmp_read_system_time: 1784| 1|void __kmp_read_system_time(double *delta) { 1785| 1| double t_ns; 1786| 1| struct timeval tval; 1787| 1| struct timespec stop; 1788| 1| int status; 1789| | 1790| 1| status = gettimeofday(&tval, NULL); 1791| 1| KMP_CHECK_SYSFAIL_ERRNO("gettimeofday", status); ------------------ | | 161| 1| { \ | | 162| 1| if (status != 0) { \ | | ------------------ | | | Branch (162:9): [True: 0, False: 1] | | ------------------ | | 163| 0| int error = errno; \ | | 164| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 165| 0| } \ | | 166| 1| } ------------------ 1792| 1| TIMEVAL_TO_TIMESPEC(&tval, &stop); 1793| 1| t_ns = (double)(TS2NS(stop) - TS2NS(__kmp_sys_timer_data.start)); ------------------ | | 75| 1| (((timespec).tv_sec * (long int)1e9) + (timespec).tv_nsec) ------------------ t_ns = (double)(TS2NS(stop) - TS2NS(__kmp_sys_timer_data.start)); ------------------ | | 75| 1| (((timespec).tv_sec * (long int)1e9) + (timespec).tv_nsec) ------------------ 1794| 1| *delta = (t_ns * 1e-9); 1795| 1|} __kmp_runtime_initialize: 1863| 1|void __kmp_runtime_initialize(void) { 1864| 1| int status; 1865| 1| pthread_mutexattr_t mutex_attr; 1866| 1| pthread_condattr_t cond_attr; 1867| | 1868| 1| if (__kmp_init_runtime) { ------------------ | Branch (1868:7): [True: 0, False: 1] ------------------ 1869| 0| return; 1870| 0| } 1871| | 1872| 1|#if (KMP_ARCH_X86 || KMP_ARCH_X86_64) 1873| 1| if (!__kmp_cpuinfo.initialized) { ------------------ | Branch (1873:7): [True: 1, False: 0] ------------------ 1874| 1| __kmp_query_cpuid(&__kmp_cpuinfo); 1875| 1| } 1876| 1|#endif /* KMP_ARCH_X86 || KMP_ARCH_X86_64 */ 1877| | 1878| 1| __kmp_xproc = __kmp_get_xproc(); 1879| | 1880| 1|#if !KMP_32_BIT_ARCH 1881| 1| struct rlimit rlim; 1882| | // read stack size of calling thread, save it as default for worker threads; 1883| | // this should be done before reading environment variables 1884| 1| status = getrlimit(RLIMIT_STACK, &rlim); 1885| 1| if (status == 0) { // success? ------------------ | Branch (1885:7): [True: 1, False: 0] ------------------ 1886| 1| __kmp_stksize = rlim.rlim_cur; 1887| 1| __kmp_check_stksize(&__kmp_stksize); // check value and adjust if needed 1888| 1| } 1889| 1|#endif /* KMP_32_BIT_ARCH */ 1890| | 1891| 1| if (sysconf(_SC_THREADS)) { ------------------ | Branch (1891:7): [True: 1, False: 0] ------------------ 1892| | 1893| | /* Query the maximum number of threads */ 1894| 1| __kmp_type_convert(sysconf(_SC_THREAD_THREADS_MAX), &(__kmp_sys_max_nth)); 1895| 1| if (__kmp_sys_max_nth == -1) { ------------------ | Branch (1895:9): [True: 1, False: 0] ------------------ 1896| | /* Unlimited threads for NPTL */ 1897| 1| __kmp_sys_max_nth = INT_MAX; 1898| 1| } else if (__kmp_sys_max_nth <= 1) { ------------------ | Branch (1898:16): [True: 0, False: 0] ------------------ 1899| | /* Can't tell, just use PTHREAD_THREADS_MAX */ 1900| 0| __kmp_sys_max_nth = KMP_MAX_NTH; ------------------ | | 1143| 0|#define KMP_MAX_NTH INT_MAX ------------------ 1901| 0| } 1902| | 1903| | /* Query the minimum stack size */ 1904| 1| __kmp_sys_min_stksize = sysconf(_SC_THREAD_STACK_MIN); 1905| 1| if (__kmp_sys_min_stksize <= 1) { ------------------ | Branch (1905:9): [True: 0, False: 1] ------------------ 1906| 0| __kmp_sys_min_stksize = KMP_MIN_STKSIZE; ------------------ | | 1148| 0|#define KMP_MIN_STKSIZE ((size_t)PTHREAD_STACK_MIN) ------------------ 1907| 0| } 1908| 1| } 1909| | 1910| | /* Set up minimum number of threads to switch to TLS gtid */ 1911| 1| __kmp_tls_gtid_min = KMP_TLS_GTID_MIN; ------------------ | | 1266| 1|#define KMP_TLS_GTID_MIN 5 ------------------ 1912| | 1913| 1| status = pthread_key_create(&__kmp_gtid_threadprivate_key, 1914| 1| __kmp_internal_end_dest); 1915| 1| KMP_CHECK_SYSFAIL("pthread_key_create", status); ------------------ | | 153| 1| { \ | | 154| 1| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 1] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 1| } ------------------ 1916| 1| status = pthread_mutexattr_init(&mutex_attr); 1917| 1| KMP_CHECK_SYSFAIL("pthread_mutexattr_init", status); ------------------ | | 153| 1| { \ | | 154| 1| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 1] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 1| } ------------------ 1918| 1| status = pthread_mutex_init(&__kmp_wait_mx.m_mutex, &mutex_attr); 1919| 1| KMP_CHECK_SYSFAIL("pthread_mutex_init", status); ------------------ | | 153| 1| { \ | | 154| 1| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 1] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 1| } ------------------ 1920| 1| status = pthread_mutexattr_destroy(&mutex_attr); 1921| 1| KMP_CHECK_SYSFAIL("pthread_mutexattr_destroy", status); ------------------ | | 153| 1| { \ | | 154| 1| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 1] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 1| } ------------------ 1922| 1| status = pthread_condattr_init(&cond_attr); 1923| 1| KMP_CHECK_SYSFAIL("pthread_condattr_init", status); ------------------ | | 153| 1| { \ | | 154| 1| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 1] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 1| } ------------------ 1924| 1| status = pthread_cond_init(&__kmp_wait_cv.c_cond, &cond_attr); 1925| 1| KMP_CHECK_SYSFAIL("pthread_cond_init", status); ------------------ | | 153| 1| { \ | | 154| 1| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 1] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 1| } ------------------ 1926| 1| status = pthread_condattr_destroy(&cond_attr); 1927| 1| KMP_CHECK_SYSFAIL("pthread_condattr_destroy", status); ------------------ | | 153| 1| { \ | | 154| 1| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 1] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 1| } ------------------ 1928| |#if USE_ITT_BUILD 1929| | __kmp_itt_initialize(); 1930| |#endif /* USE_ITT_BUILD */ 1931| | 1932| 1| __kmp_init_runtime = TRUE; ------------------ | | 1280| 1|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1|#define FALSE 0 | | ------------------ ------------------ 1933| 1|} __kmp_runtime_destroy: 1935| 1|void __kmp_runtime_destroy(void) { 1936| 1| int status; 1937| | 1938| 1| if (!__kmp_init_runtime) { ------------------ | Branch (1938:7): [True: 0, False: 1] ------------------ 1939| 0| return; // Nothing to do. 1940| 0| } 1941| | 1942| |#if USE_ITT_BUILD 1943| | __kmp_itt_destroy(); 1944| |#endif /* USE_ITT_BUILD */ 1945| | 1946| 1| status = pthread_key_delete(__kmp_gtid_threadprivate_key); 1947| 1| KMP_CHECK_SYSFAIL("pthread_key_delete", status); ------------------ | | 153| 1| { \ | | 154| 1| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 1] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 1| } ------------------ 1948| | 1949| 1| status = pthread_mutex_destroy(&__kmp_wait_mx.m_mutex); 1950| 1| if (status != 0 && status != EBUSY) { ------------------ | Branch (1950:7): [True: 0, False: 1] | Branch (1950:22): [True: 0, False: 0] ------------------ 1951| 0| KMP_SYSFAIL("pthread_mutex_destroy", status); ------------------ | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ | | 149| 0| __kmp_msg_null) ------------------ 1952| 0| } 1953| 1| status = pthread_cond_destroy(&__kmp_wait_cv.c_cond); 1954| 1| if (status != 0 && status != EBUSY) { ------------------ | Branch (1954:7): [True: 0, False: 1] | Branch (1954:22): [True: 0, False: 0] ------------------ 1955| 0| KMP_SYSFAIL("pthread_cond_destroy", status); ------------------ | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ | | 149| 0| __kmp_msg_null) ------------------ 1956| 0| } 1957| 1|#if KMP_AFFINITY_SUPPORTED 1958| 1| __kmp_affinity_uninitialize(); 1959| 1|#endif 1960| | 1961| 1| __kmp_init_runtime = FALSE; ------------------ | | 1279| 1|#define FALSE 0 ------------------ 1962| 1|} _Z14__kmp_now_nsecv: 1992| 2|kmp_uint64 __kmp_now_nsec() { 1993| 2| struct timeval t; 1994| 2| gettimeofday(&t, NULL); 1995| 2| kmp_uint64 nsec = (kmp_uint64)KMP_NSEC_PER_SEC * (kmp_uint64)t.tv_sec + ------------------ | | 181| 2|#define KMP_NSEC_PER_SEC 1000000000L ------------------ 1996| 2| (kmp_uint64)1000 * (kmp_uint64)t.tv_usec; 1997| 2| return nsec; 1998| 2|} __kmp_initialize_system_tick: 2002| 1|void __kmp_initialize_system_tick() { 2003| 1| kmp_uint64 now, nsec2, diff; 2004| 1| kmp_uint64 delay = 100000; // 50~100 usec on most machines. 2005| 1| kmp_uint64 nsec = __kmp_now_nsec(); 2006| 1| kmp_uint64 goal = __kmp_hardware_timestamp() + delay; 2007| 3.83k| while ((now = __kmp_hardware_timestamp()) < goal) ------------------ | Branch (2007:10): [True: 3.83k, False: 1] ------------------ 2008| 3.83k| ; 2009| 1| nsec2 = __kmp_now_nsec(); 2010| 1| diff = nsec2 - nsec; 2011| 1| if (diff > 0) { ------------------ | Branch (2011:7): [True: 1, False: 0] ------------------ 2012| 1| kmp_uint64 tpms = ((kmp_uint64)1e6 * (delay + (now - goal)) / diff); 2013| 1| if (tpms > 0) ------------------ | Branch (2013:9): [True: 1, False: 0] ------------------ 2014| 1| __kmp_ticks_per_msec = tpms; 2015| 1| } 2016| 1|} z_Linux_util.cpp:_ZL20__kmp_set_stack_infoiP8kmp_info: 408| 32|static kmp_int32 __kmp_set_stack_info(int gtid, kmp_info_t *th) { 409| 32| int stack_data; 410| 32|#if KMP_OS_LINUX || KMP_OS_DRAGONFLY || KMP_OS_FREEBSD || KMP_OS_NETBSD || \ 411| 32| KMP_OS_HURD 412| 32| pthread_attr_t attr; 413| 32| int status; 414| 32| size_t size = 0; 415| 32| void *addr = 0; 416| | 417| | /* Always do incremental stack refinement for ubermaster threads since the 418| | initial thread stack range can be reduced by sibling thread creation so 419| | pthread_attr_getstack may cause thread gtid aliasing */ 420| 32| if (!KMP_UBER_GTID(gtid)) { ------------------ | Branch (420:7): [True: 31, False: 1] ------------------ 421| | 422| | /* Fetch the real thread attributes */ 423| 31| status = pthread_attr_init(&attr); 424| 31| KMP_CHECK_SYSFAIL("pthread_attr_init", status); ------------------ | | 153| 31| { \ | | 154| 31| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 31] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 31| } ------------------ 425| |#if KMP_OS_DRAGONFLY || KMP_OS_FREEBSD || KMP_OS_NETBSD 426| | status = pthread_attr_get_np(pthread_self(), &attr); 427| | KMP_CHECK_SYSFAIL("pthread_attr_get_np", status); 428| |#else 429| 31| status = pthread_getattr_np(pthread_self(), &attr); 430| 31| KMP_CHECK_SYSFAIL("pthread_getattr_np", status); ------------------ | | 153| 31| { \ | | 154| 31| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 31] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 31| } ------------------ 431| 31|#endif 432| 31| status = pthread_attr_getstack(&attr, &addr, &size); 433| 31| KMP_CHECK_SYSFAIL("pthread_attr_getstack", status); ------------------ | | 153| 31| { \ | | 154| 31| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 31] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 31| } ------------------ 434| 31| KA_TRACE(60, 435| 31| ("__kmp_set_stack_info: T#%d pthread_attr_getstack returned size:" 436| 31| " %lu, low addr: %p\n", 437| 31| gtid, size, addr)); 438| 31| status = pthread_attr_destroy(&attr); 439| 31| KMP_CHECK_SYSFAIL("pthread_attr_destroy", status); ------------------ | | 153| 31| { \ | | 154| 31| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 31] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 31| } ------------------ 440| 31| } 441| | 442| 32| if (size != 0 && addr != 0) { // was stack parameter determination successful? ------------------ | Branch (442:7): [True: 31, False: 1] | Branch (442:20): [True: 31, False: 0] ------------------ 443| | /* Store the correct base and size */ 444| 31| TCW_PTR(th->th.th_info.ds.ds_stackbase, (((char *)addr) + size)); ------------------ | | 1157| 31|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 31|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 445| 31| TCW_PTR(th->th.th_info.ds.ds_stacksize, size); ------------------ | | 1157| 31|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 31|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 446| 31| TCW_4(th->th.th_info.ds.ds_stackgrow, FALSE); ------------------ | | 1128| 31|#define TCW_4(a, b) (a) = (b) ------------------ 447| 31| return TRUE; ------------------ | | 1280| 31|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 31|#define FALSE 0 | | ------------------ ------------------ 448| 31| } 449| 1|#endif /* KMP_OS_LINUX || KMP_OS_DRAGONFLY || KMP_OS_FREEBSD || KMP_OS_NETBSD \ 450| | || KMP_OS_HURD */ 451| | /* Use incremental refinement starting from initial conservative estimate */ 452| 1| TCW_PTR(th->th.th_info.ds.ds_stacksize, 0); ------------------ | | 1157| 1|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 1|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 453| 1| TCW_PTR(th->th.th_info.ds.ds_stackbase, &stack_data); ------------------ | | 1157| 1|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 1|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 454| 1| TCW_4(th->th.th_info.ds.ds_stackgrow, TRUE); ------------------ | | 1128| 1|#define TCW_4(a, b) (a) = (b) ------------------ 455| 1| return FALSE; ------------------ | | 1279| 1|#define FALSE 0 ------------------ 456| 32|} z_Linux_util.cpp:_ZL19__kmp_launch_workerPv: 458| 31|static void *__kmp_launch_worker(void *thr) { 459| 31| int status, old_type, old_state; 460| |#ifdef KMP_BLOCK_SIGNALS 461| | sigset_t new_set, old_set; 462| |#endif /* KMP_BLOCK_SIGNALS */ 463| 31| void *exit_val; 464| 31|#if KMP_OS_LINUX || KMP_OS_DRAGONFLY || KMP_OS_FREEBSD || KMP_OS_NETBSD || \ 465| 31| KMP_OS_OPENBSD || KMP_OS_HURD 466| 31| void *volatile padding = 0; 467| 31|#endif 468| 31| int gtid; 469| | 470| 31| gtid = ((kmp_info_t *)thr)->th.th_info.ds.ds_gtid; 471| 31| __kmp_gtid_set_specific(gtid); 472| 31|#ifdef KMP_TDATA_GTID 473| 31| __kmp_gtid = gtid; 474| 31|#endif 475| |#if KMP_STATS_ENABLED 476| | // set thread local index to point to thread-specific stats 477| | __kmp_stats_thread_ptr = ((kmp_info_t *)thr)->th.th_stats; 478| | __kmp_stats_thread_ptr->startLife(); 479| | KMP_SET_THREAD_STATE(IDLE); 480| | KMP_INIT_PARTITIONED_TIMERS(OMP_idle); 481| |#endif 482| | 483| |#if USE_ITT_BUILD 484| | __kmp_itt_thread_name(gtid); 485| |#endif /* USE_ITT_BUILD */ 486| | 487| 31|#if KMP_AFFINITY_SUPPORTED 488| 31| __kmp_affinity_set_init_mask(gtid, FALSE); ------------------ | | 1279| 31|#define FALSE 0 ------------------ 489| 31|#endif 490| | 491| 31|#ifdef KMP_CANCEL_THREADS 492| 31| status = pthread_setcanceltype(PTHREAD_CANCEL_ASYNCHRONOUS, &old_type); 493| 31| KMP_CHECK_SYSFAIL("pthread_setcanceltype", status); ------------------ | | 153| 31| { \ | | 154| 31| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 31] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 31| } ------------------ 494| | // josh todo: isn't PTHREAD_CANCEL_ENABLE default for newly-created threads? 495| 31| status = pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, &old_state); 496| 31| KMP_CHECK_SYSFAIL("pthread_setcancelstate", status); ------------------ | | 153| 31| { \ | | 154| 31| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 31] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 31| } ------------------ 497| 31|#endif 498| | 499| 31|#if KMP_ARCH_X86 || KMP_ARCH_X86_64 500| | // Set FP control regs to be a copy of the parallel initialization thread's. 501| 31| __kmp_clear_x87_fpu_status_word(); 502| 31| __kmp_load_x87_fpu_control_word(&__kmp_init_x87_fpu_control_word); 503| 31| __kmp_load_mxcsr(&__kmp_init_mxcsr); 504| 31|#endif /* KMP_ARCH_X86 || KMP_ARCH_X86_64 */ 505| | 506| |#ifdef KMP_BLOCK_SIGNALS 507| | status = sigfillset(&new_set); 508| | KMP_CHECK_SYSFAIL_ERRNO("sigfillset", status); 509| | status = pthread_sigmask(SIG_BLOCK, &new_set, &old_set); 510| | KMP_CHECK_SYSFAIL("pthread_sigmask", status); 511| |#endif /* KMP_BLOCK_SIGNALS */ 512| | 513| 31|#if KMP_OS_LINUX || KMP_OS_DRAGONFLY || KMP_OS_FREEBSD || KMP_OS_NETBSD || \ 514| 31| KMP_OS_OPENBSD 515| 31| if (__kmp_stkoffset > 0 && gtid > 0) { ------------------ | Branch (515:7): [True: 31, False: 0] | Branch (515:30): [True: 31, False: 0] ------------------ 516| 31| padding = KMP_ALLOCA(gtid * __kmp_stkoffset); ------------------ | | 59| 31|#define KMP_ALLOCA alloca ------------------ 517| 31| (void)padding; 518| 31| } 519| 31|#endif 520| | 521| 31| KMP_MB(); 522| 31| __kmp_set_stack_info(gtid, (kmp_info_t *)thr); 523| | 524| 31| __kmp_check_stack_overlap((kmp_info_t *)thr); 525| | 526| 31| exit_val = __kmp_launch_thread((kmp_info_t *)thr); 527| | 528| |#ifdef KMP_BLOCK_SIGNALS 529| | status = pthread_sigmask(SIG_SETMASK, &old_set, NULL); 530| | KMP_CHECK_SYSFAIL("pthread_sigmask", status); 531| |#endif /* KMP_BLOCK_SIGNALS */ 532| | 533| 31| return exit_val; 534| 31|} z_Linux_util.cpp:_ZL25__kmp_install_one_handleriPFviEi: 1113| 11| int parallel_init) { 1114| 11| KMP_MB(); // Flush all pending memory write invalidates. 1115| 11| KB_TRACE(60, 1116| 11| ("__kmp_install_one_handler( %d, ..., %d )\n", sig, parallel_init)); 1117| 11| if (parallel_init) { ------------------ | Branch (1117:7): [True: 0, False: 11] ------------------ 1118| 0| struct sigaction new_action; 1119| 0| struct sigaction old_action; 1120| 0| new_action.sa_handler = handler_func; 1121| 0| new_action.sa_flags = 0; 1122| 0| sigfillset(&new_action.sa_mask); 1123| 0| __kmp_sigaction(sig, &new_action, &old_action); 1124| 0| if (old_action.sa_handler == __kmp_sighldrs[sig].sa_handler) { ------------------ | Branch (1124:9): [True: 0, False: 0] ------------------ 1125| 0| sigaddset(&__kmp_sigset, sig); 1126| 0| } else { 1127| | // Restore/keep user's handler if one previously installed. 1128| 0| __kmp_sigaction(sig, &old_action, NULL); 1129| 0| } 1130| 11| } else { 1131| | // Save initial/system signal handlers to see if user handlers installed. 1132| | __kmp_sigaction(sig, NULL, &__kmp_sighldrs[sig]); 1133| 11| } 1134| 11| KMP_MB(); // Flush all pending memory write invalidates. 1135| 11|} // __kmp_install_one_handler z_Linux_util.cpp:_ZL15__kmp_sigactioniPK9sigactionPS_: 1107| 11| struct sigaction *oldact) { 1108| 11| int rc = sigaction(signum, act, oldact); 1109| | KMP_CHECK_SYSFAIL_ERRNO("sigaction", rc); ------------------ | | 161| 11| { \ | | 162| 11| if (status != 0) { \ | | ------------------ | | | Branch (162:9): [True: 0, False: 11] | | ------------------ | | 163| 0| int error = errno; \ | | 164| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 165| 0| } \ | | 166| 11| } ------------------ 1110| 11|} z_Linux_util.cpp:_ZL24__kmp_remove_one_handleri: 1137| 64|static void __kmp_remove_one_handler(int sig) { 1138| 64| KB_TRACE(60, ("__kmp_remove_one_handler( %d )\n", sig)); 1139| 64| if (sigismember(&__kmp_sigset, sig)) { ------------------ | Branch (1139:7): [True: 0, False: 64] ------------------ 1140| 0| struct sigaction old; 1141| 0| KMP_MB(); // Flush all pending memory write invalidates. 1142| 0| __kmp_sigaction(sig, &__kmp_sighldrs[sig], &old); 1143| 0| if ((old.sa_handler != __kmp_team_handler) && ------------------ | Branch (1143:9): [True: 0, False: 0] ------------------ 1144| 0| (old.sa_handler != __kmp_null_handler)) { ------------------ | Branch (1144:9): [True: 0, False: 0] ------------------ 1145| | // Restore the users signal handler. 1146| 0| KB_TRACE(10, ("__kmp_remove_one_handler: oops, not our handler, " 1147| 0| "restoring: sig=%d\n", 1148| 0| sig)); 1149| | __kmp_sigaction(sig, &old, NULL); 1150| 0| } 1151| 0| sigdelset(&__kmp_sigset, sig); 1152| 0| KMP_MB(); // Flush all pending memory write invalidates. 1153| 0| } 1154| 64|} // __kmp_remove_one_handler z_Linux_util.cpp:_ZL15__kmp_get_xprocv: 1805| 1|static int __kmp_get_xproc(void) { 1806| | 1807| 1| int r = 0; 1808| | 1809| 1|#if KMP_OS_LINUX 1810| | 1811| 1| __kmp_type_convert(sysconf(_SC_NPROCESSORS_CONF), &(r)); 1812| | 1813| |#elif KMP_OS_DRAGONFLY || KMP_OS_FREEBSD || KMP_OS_NETBSD || KMP_OS_OPENBSD || \ 1814| | KMP_OS_HURD 1815| | 1816| | __kmp_type_convert(sysconf(_SC_NPROCESSORS_ONLN), &(r)); 1817| | 1818| |#elif KMP_OS_DARWIN 1819| | 1820| | // Bug C77011 High "OpenMP Threads and number of active cores". 1821| | 1822| | // Find the number of available CPUs. 1823| | kern_return_t rc; 1824| | host_basic_info_data_t info; 1825| | mach_msg_type_number_t num = HOST_BASIC_INFO_COUNT; 1826| | rc = host_info(mach_host_self(), HOST_BASIC_INFO, (host_info_t)&info, &num); 1827| | if (rc == 0 && num == HOST_BASIC_INFO_COUNT) { 1828| | // Cannot use KA_TRACE() here because this code works before trace support 1829| | // is initialized. 1830| | r = info.avail_cpus; 1831| | } else { 1832| | KMP_WARNING(CantGetNumAvailCPU); 1833| | KMP_INFORM(AssumedNumCPU); 1834| | } 1835| | 1836| |#else 1837| | 1838| |#error "Unknown or unsupported OS." 1839| | 1840| |#endif 1841| | 1842| 1| return r > 0 ? r : 2; /* guess value of 2 if OS told us 0 */ ------------------ | Branch (1842:10): [True: 1, False: 0] ------------------ 1843| | 1844| 1|} // __kmp_get_xproc z_Linux_util.cpp:_ZL22__kmp_suspend_templateI11kmp_flag_64ILb0ELb1EEEviPT_: 1398| 1.62k|static inline void __kmp_suspend_template(int th_gtid, C *flag) { 1399| 1.62k| KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(USER_suspend); ------------------ | | 1008| 1.62k|#define KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(n) ((void)0) ------------------ 1400| 1.62k| kmp_info_t *th = __kmp_threads[th_gtid]; 1401| 1.62k| int status; 1402| 1.62k| typename C::flag_t old_spin; 1403| | 1404| 1.62k| KF_TRACE(30, ("__kmp_suspend_template: T#%d enter for flag = %p\n", th_gtid, 1405| 1.62k| flag->get())); 1406| | 1407| 1.62k| __kmp_suspend_initialize_thread(th); 1408| | 1409| 1.62k| __kmp_lock_suspend_mx(th); 1410| | 1411| 1.62k| KF_TRACE(10, ("__kmp_suspend_template: T#%d setting sleep bit for spin(%p)\n", 1412| 1.62k| th_gtid, flag->get())); 1413| | 1414| | /* TODO: shouldn't this use release semantics to ensure that 1415| | __kmp_suspend_initialize_thread gets called first? */ 1416| 1.62k| old_spin = flag->set_sleeping(); 1417| 1.62k| TCW_PTR(th->th.th_sleep_loc, (void *)flag); ------------------ | | 1157| 1.62k|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 1.62k|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 1418| 1.62k| th->th.th_sleep_loc_type = flag->get_type(); 1419| 1.62k| if (__kmp_dflt_blocktime == KMP_MAX_BLOCKTIME && ------------------ | | 1184| 3.24k| (INT_MAX) /* Must be this for "infinite" setting the work */ ------------------ | Branch (1419:7): [True: 0, False: 1.62k] ------------------ 1420| 0| __kmp_pause_status != kmp_soft_paused) { ------------------ | Branch (1420:7): [True: 0, False: 0] ------------------ 1421| 0| flag->unset_sleeping(); 1422| 0| TCW_PTR(th->th.th_sleep_loc, NULL); ------------------ | | 1157| 0|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 0|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 1423| 0| th->th.th_sleep_loc_type = flag_unset; 1424| 0| __kmp_unlock_suspend_mx(th); 1425| 0| return; 1426| 0| } 1427| 1.62k| KF_TRACE(5, ("__kmp_suspend_template: T#%d set sleep bit for spin(%p)==%x," 1428| 1.62k| " was %x\n", 1429| 1.62k| th_gtid, flag->get(), flag->load(), old_spin)); 1430| | 1431| 1.62k| if (flag->done_check_val(old_spin) || flag->done_check()) { ------------------ | Branch (1431:7): [True: 41, False: 1.58k] | Branch (1431:41): [True: 2, False: 1.57k] ------------------ 1432| 47| flag->unset_sleeping(); 1433| 47| TCW_PTR(th->th.th_sleep_loc, NULL); ------------------ | | 1157| 47|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 47|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 1434| 47| th->th.th_sleep_loc_type = flag_unset; 1435| 47| KF_TRACE(5, ("__kmp_suspend_template: T#%d false alarm, reset sleep bit " 1436| 47| "for spin(%p)\n", 1437| 47| th_gtid, flag->get())); 1438| 1.57k| } else { 1439| | /* Encapsulate in a loop as the documentation states that this may 1440| | "with low probability" return when the condition variable has 1441| | not been signaled or broadcast */ 1442| 1.57k| int deactivated = FALSE; ------------------ | | 1279| 1.57k|#define FALSE 0 ------------------ 1443| | 1444| 3.14k| while (flag->is_sleeping()) { ------------------ | Branch (1444:12): [True: 1.57k, False: 1.57k] ------------------ 1445| |#ifdef DEBUG_SUSPEND 1446| | char buffer[128]; 1447| | __kmp_suspend_count++; 1448| | __kmp_print_cond(buffer, &th->th.th_suspend_cv); 1449| | __kmp_printf("__kmp_suspend_template: suspending T#%d: %s\n", th_gtid, 1450| | buffer); 1451| |#endif 1452| | // Mark the thread as no longer active (only in the first iteration of the 1453| | // loop). 1454| 1.57k| if (!deactivated) { ------------------ | Branch (1454:11): [True: 1.57k, False: 0] ------------------ 1455| 1.57k| th->th.th_active = FALSE; ------------------ | | 1279| 1.57k|#define FALSE 0 ------------------ 1456| 1.57k| if (th->th.th_active_in_pool) { ------------------ | Branch (1456:13): [True: 0, False: 1.57k] ------------------ 1457| 0| th->th.th_active_in_pool = FALSE; ------------------ | | 1279| 0|#define FALSE 0 ------------------ 1458| 0| KMP_ATOMIC_DEC(&__kmp_thread_pool_active_nth); ------------------ | | 1260| 0|#define KMP_ATOMIC_DEC(p) KMP_ATOMIC_OP(fetch_sub, p, 1, acq_rel) | | ------------------ | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 1459| 0| KMP_DEBUG_ASSERT(TCR_4(__kmp_thread_pool_active_nth) >= 0); 1460| 0| } 1461| 1.57k| deactivated = TRUE; ------------------ | | 1280| 1.57k|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1.57k|#define FALSE 0 | | ------------------ ------------------ 1462| 1.57k| } 1463| | 1464| 1.57k| KMP_DEBUG_ASSERT(th->th.th_sleep_loc); 1465| 1.57k| KMP_DEBUG_ASSERT(flag->get_type() == th->th.th_sleep_loc_type); 1466| | 1467| |#if USE_SUSPEND_TIMEOUT 1468| | struct timespec now; 1469| | struct timeval tval; 1470| | int msecs; 1471| | 1472| | status = gettimeofday(&tval, NULL); 1473| | KMP_CHECK_SYSFAIL_ERRNO("gettimeofday", status); 1474| | TIMEVAL_TO_TIMESPEC(&tval, &now); 1475| | 1476| | msecs = (4 * __kmp_dflt_blocktime) + 200; 1477| | now.tv_sec += msecs / 1000; 1478| | now.tv_nsec += (msecs % 1000) * 1000; 1479| | 1480| | KF_TRACE(15, ("__kmp_suspend_template: T#%d about to perform " 1481| | "pthread_cond_timedwait\n", 1482| | th_gtid)); 1483| | status = pthread_cond_timedwait(&th->th.th_suspend_cv.c_cond, 1484| | &th->th.th_suspend_mx.m_mutex, &now); 1485| |#else 1486| 1.57k| KF_TRACE(15, ("__kmp_suspend_template: T#%d about to perform" 1487| 1.57k| " pthread_cond_wait\n", 1488| 1.57k| th_gtid)); 1489| 1.57k| status = pthread_cond_wait(&th->th.th_suspend_cv.c_cond, 1490| 1.57k| &th->th.th_suspend_mx.m_mutex); 1491| 1.57k|#endif // USE_SUSPEND_TIMEOUT 1492| | 1493| 1.57k| if ((status != 0) && (status != EINTR) && (status != ETIMEDOUT)) { ------------------ | Branch (1493:11): [True: 0, False: 1.57k] | Branch (1493:28): [True: 0, False: 0] | Branch (1493:49): [True: 0, False: 0] ------------------ 1494| 0| KMP_SYSFAIL("pthread_cond_wait", status); ------------------ | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | ------------------ | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | ------------------ | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | ------------------ | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | ------------------ | | 149| 0| __kmp_msg_null) ------------------ 1495| 0| } 1496| | 1497| 1.57k| KMP_DEBUG_ASSERT(flag->get_type() == flag->get_ptr_type()); 1498| | 1499| 1.57k| if (!flag->is_sleeping() && ------------------ | Branch (1499:11): [True: 1.57k, False: 18.4E] ------------------ 1500| 1.57k| ((status == EINTR) || (status == ETIMEDOUT))) { ------------------ | Branch (1500:12): [True: 1, False: 1.57k] | Branch (1500:33): [True: 18.4E, False: 1.57k] ------------------ 1501| | // if interrupt or timeout, and thread is no longer sleeping, we need to 1502| | // make sure sleep_loc gets reset; however, this shouldn't be needed if 1503| | // we woke up with resume 1504| 0| flag->unset_sleeping(); 1505| 0| TCW_PTR(th->th.th_sleep_loc, NULL); ------------------ | | 1157| 0|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 0|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 1506| 0| th->th.th_sleep_loc_type = flag_unset; 1507| 0| } 1508| |#ifdef KMP_DEBUG 1509| | if (status == ETIMEDOUT) { 1510| | if (flag->is_sleeping()) { 1511| | KF_TRACE(100, 1512| | ("__kmp_suspend_template: T#%d timeout wakeup\n", th_gtid)); 1513| | } else { 1514| | KF_TRACE(2, ("__kmp_suspend_template: T#%d timeout wakeup, sleep bit " 1515| | "not set!\n", 1516| | th_gtid)); 1517| | TCW_PTR(th->th.th_sleep_loc, NULL); 1518| | th->th.th_sleep_loc_type = flag_unset; 1519| | } 1520| | } else if (flag->is_sleeping()) { 1521| | KF_TRACE(100, 1522| | ("__kmp_suspend_template: T#%d spurious wakeup\n", th_gtid)); 1523| | } 1524| |#endif 1525| 1.57k| } // while 1526| | 1527| | // Mark the thread as active again (if it was previous marked as inactive) 1528| 1.57k| if (deactivated) { ------------------ | Branch (1528:9): [True: 1.57k, False: 0] ------------------ 1529| 1.57k| th->th.th_active = TRUE; ------------------ | | 1280| 1.57k|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 1.57k|#define FALSE 0 | | ------------------ ------------------ 1530| 1.57k| if (TCR_4(th->th.th_in_pool)) { ------------------ | | 1127| 1.57k|#define TCR_4(a) (a) | | ------------------ | | | Branch (1127:18): [True: 0, False: 1.57k] | | ------------------ ------------------ 1531| 0| KMP_ATOMIC_INC(&__kmp_thread_pool_active_nth); ------------------ | | 1259| 0|#define KMP_ATOMIC_INC(p) KMP_ATOMIC_OP(fetch_add, p, 1, acq_rel) | | ------------------ | | | | 1246| 0|#define KMP_ATOMIC_OP(op, p, v, order) (p)->op(v, std::memory_order_##order) | | ------------------ ------------------ 1532| 0| th->th.th_active_in_pool = TRUE; ------------------ | | 1280| 0|#define TRUE (!FALSE) | | ------------------ | | | | 1279| 0|#define FALSE 0 | | ------------------ ------------------ 1533| 0| } 1534| 1.57k| } 1535| 1.57k| } 1536| | // We may have had the loop variable set before entering the loop body; 1537| | // so we need to reset sleep_loc. 1538| 1.62k| TCW_PTR(th->th.th_sleep_loc, NULL); ------------------ | | 1157| 1.62k|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 1.62k|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 1539| 1.62k| th->th.th_sleep_loc_type = flag_unset; 1540| | 1541| 1.62k| KMP_DEBUG_ASSERT(!flag->is_sleeping()); 1542| 1.62k| KMP_DEBUG_ASSERT(!th->th.th_sleep_loc); 1543| |#ifdef DEBUG_SUSPEND 1544| | { 1545| | char buffer[128]; 1546| | __kmp_print_cond(buffer, &th->th.th_suspend_cv); 1547| | __kmp_printf("__kmp_suspend_template: T#%d has awakened: %s\n", th_gtid, 1548| | buffer); 1549| | } 1550| |#endif 1551| | 1552| 1.62k| __kmp_unlock_suspend_mx(th); 1553| 1.62k| KF_TRACE(30, ("__kmp_suspend_template: T#%d exit\n", th_gtid)); 1554| 1.62k|} z_Linux_util.cpp:_ZL21__kmp_resume_templateI11kmp_flag_64ILb0ELb1EEEviPT_: 1584| 1.55k|static inline void __kmp_resume_template(int target_gtid, C *flag) { 1585| 1.55k| KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(USER_resume); ------------------ | | 1008| 1.55k|#define KMP_TIME_DEVELOPER_PARTITIONED_BLOCK(n) ((void)0) ------------------ 1586| 1.55k| kmp_info_t *th = __kmp_threads[target_gtid]; 1587| 1.55k| int status; 1588| | 1589| |#ifdef KMP_DEBUG 1590| | int gtid = TCR_4(__kmp_init_gtid) ? __kmp_get_gtid() : -1; 1591| |#endif 1592| | 1593| 1.55k| KF_TRACE(30, ("__kmp_resume_template: T#%d wants to wakeup T#%d enter\n", 1594| 1.55k| gtid, target_gtid)); 1595| 1.55k| KMP_DEBUG_ASSERT(gtid != target_gtid); 1596| | 1597| 1.55k| __kmp_suspend_initialize_thread(th); 1598| | 1599| 1.55k| __kmp_lock_suspend_mx(th); 1600| | 1601| 1.58k| if (!flag || flag != th->th.th_sleep_loc) { ------------------ | Branch (1601:7): [True: 18.4E, False: 1.58k] | Branch (1601:16): [True: 1.58k, False: 0] ------------------ 1602| | // coming from __kmp_null_resume_wrapper, or thread is now sleeping on a 1603| | // different location; wake up at new location 1604| 1.58k| flag = (C *)CCAST(void *, th->th.th_sleep_loc); ------------------ | | 229| 1.58k|#define CCAST(type, var) const_cast(var) ------------------ 1605| 1.58k| } 1606| | 1607| | // First, check if the flag is null or its type has changed. If so, someone 1608| | // else woke it up. 1609| 1.55k| if (!flag) { // Thread doesn't appear to be sleeping on anything ------------------ | Branch (1609:7): [True: 0, False: 1.55k] ------------------ 1610| 0| KF_TRACE(5, ("__kmp_resume_template: T#%d exiting, thread T#%d already " 1611| 0| "awake: flag(%p)\n", 1612| 0| gtid, target_gtid, (void *)NULL)); 1613| 0| __kmp_unlock_suspend_mx(th); 1614| 0| return; 1615| 1.55k| } else if (flag->get_type() != th->th.th_sleep_loc_type) { ------------------ | Branch (1615:14): [True: 0, False: 1.55k] ------------------ 1616| | // Flag type does not appear to match this function template; possibly the 1617| | // thread is sleeping on something else. Try null resume again. 1618| 0| KF_TRACE( 1619| 0| 5, 1620| 0| ("__kmp_resume_template: T#%d retrying, thread T#%d Mismatch flag(%p), " 1621| 0| "spin(%p) type=%d ptr_type=%d\n", 1622| 0| gtid, target_gtid, flag, flag->get(), flag->get_type(), 1623| 0| th->th.th_sleep_loc_type)); 1624| 0| __kmp_unlock_suspend_mx(th); 1625| 0| __kmp_null_resume_wrapper(th); 1626| 0| return; 1627| 1.55k| } else { // if multiple threads are sleeping, flag should be internally 1628| | // referring to a specific thread here 1629| 1.55k| if (!flag->is_sleeping()) { ------------------ | Branch (1629:9): [True: 0, False: 1.55k] ------------------ 1630| 0| KF_TRACE(5, ("__kmp_resume_template: T#%d exiting, thread T#%d already " 1631| 0| "awake: flag(%p): %u\n", 1632| 0| gtid, target_gtid, flag->get(), (unsigned int)flag->load())); 1633| 0| __kmp_unlock_suspend_mx(th); 1634| 0| return; 1635| 0| } 1636| 1.55k| } 1637| 1.55k| KMP_DEBUG_ASSERT(flag); 1638| 1.55k| flag->unset_sleeping(); 1639| 1.55k| TCW_PTR(th->th.th_sleep_loc, NULL); ------------------ | | 1157| 1.55k|#define TCW_PTR(a, b) TCW_8((a), (b)) | | ------------------ | | | | 1132| 1.55k|#define TCW_8(a, b) (a) = (b) | | ------------------ ------------------ 1640| 1.55k| th->th.th_sleep_loc_type = flag_unset; 1641| | 1642| 1.55k| KF_TRACE(5, ("__kmp_resume_template: T#%d about to wakeup T#%d, reset " 1643| 1.55k| "sleep bit for flag's loc(%p): %u\n", 1644| 1.55k| gtid, target_gtid, flag->get(), (unsigned int)flag->load())); 1645| | 1646| |#ifdef DEBUG_SUSPEND 1647| | { 1648| | char buffer[128]; 1649| | __kmp_print_cond(buffer, &th->th.th_suspend_cv); 1650| | __kmp_printf("__kmp_resume_template: T#%d resuming T#%d: %s\n", gtid, 1651| | target_gtid, buffer); 1652| | } 1653| |#endif 1654| 1.55k| status = pthread_cond_signal(&th->th.th_suspend_cv.c_cond); 1655| 1.55k| KMP_CHECK_SYSFAIL("pthread_cond_signal", status); ------------------ | | 153| 1.55k| { \ | | 154| 1.55k| if (error) { \ | | ------------------ | | | Branch (154:9): [True: 0, False: 1.55k] | | ------------------ | | 155| 0| KMP_SYSFAIL(func, error); \ | | ------------------ | | | | 148| 0| __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 121| 0|#define KMP_MSG(...) __kmp_msg_format(kmp_i18n_msg_##__VA_ARGS__) | | | | ------------------ | | | | __kmp_fatal(KMP_MSG(FunctionError, func), KMP_SYSERRCODE(error), \ | | | | ------------------ | | | | | | 123| 0|#define KMP_SYSERRCODE(code) __kmp_msg_error_code(code) | | | | ------------------ | | | | 149| 0| __kmp_msg_null) | | ------------------ | | 156| 0| } \ | | 157| 1.55k| } ------------------ 1656| 1.55k| __kmp_unlock_suspend_mx(th); 1657| 1.55k| KF_TRACE(30, ("__kmp_resume_template: T#%d exiting after signaling wake up" 1658| 1.55k| " for T#%d\n", 1659| 1.55k| gtid, target_gtid)); 1660| 1.55k|}