/src/FreeRDP/libfreerdp/primitives/prim_copy.c
Line | Count | Source (jump to first uncovered line) |
1 | | /* FreeRDP: A Remote Desktop Protocol Client |
2 | | * Copy operations. |
3 | | * vi:ts=4 sw=4: |
4 | | * |
5 | | * (c) Copyright 2012 Hewlett-Packard Development Company, L.P. |
6 | | * Licensed under the Apache License, Version 2.0 (the "License"); you may |
7 | | * not use this file except in compliance with the License. You may obtain |
8 | | * a copy of the License at http://www.apache.org/licenses/LICENSE-2.0. |
9 | | * Unless required by applicable law or agreed to in writing, software |
10 | | * distributed under the License is distributed on an "AS IS" BASIS, |
11 | | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express |
12 | | * or implied. See the License for the specific language governing |
13 | | * permissions and limitations under the License. |
14 | | */ |
15 | | |
16 | | #include <freerdp/config.h> |
17 | | |
18 | | #include <string.h> |
19 | | #include <freerdp/types.h> |
20 | | #include <freerdp/primitives.h> |
21 | | #include <freerdp/log.h> |
22 | | |
23 | | #include "prim_internal.h" |
24 | | #include "prim_copy.h" |
25 | | #include "../codec/color.h" |
26 | | |
27 | | #include <freerdp/codec/color.h> |
28 | | |
29 | | #define TAG FREERDP_TAG("primitives.copy") |
30 | | |
31 | | static primitives_t* generic = NULL; |
32 | | |
33 | | /* ------------------------------------------------------------------------- */ |
34 | | /*static inline BOOL memory_regions_overlap_1d(*/ |
35 | | static BOOL memory_regions_overlap_1d(const BYTE* p1, const BYTE* p2, size_t bytes) |
36 | 0 | { |
37 | 0 | const ULONG_PTR p1m = (const ULONG_PTR)p1; |
38 | 0 | const ULONG_PTR p2m = (const ULONG_PTR)p2; |
39 | |
|
40 | 0 | if (p1m <= p2m) |
41 | 0 | { |
42 | 0 | if (p1m + bytes > p2m) |
43 | 0 | return TRUE; |
44 | 0 | } |
45 | 0 | else |
46 | 0 | { |
47 | 0 | if (p2m + bytes > p1m) |
48 | 0 | return TRUE; |
49 | 0 | } |
50 | | |
51 | | /* else */ |
52 | 0 | return FALSE; |
53 | 0 | } |
54 | | |
55 | | /* ------------------------------------------------------------------------- */ |
56 | | /*static inline BOOL memory_regions_overlap_2d( */ |
57 | | static BOOL memory_regions_overlap_2d(const BYTE* p1, int p1Step, int p1Size, const BYTE* p2, |
58 | | int p2Step, int p2Size, int width, int height) |
59 | 0 | { |
60 | 0 | ULONG_PTR p1m = (ULONG_PTR)p1; |
61 | 0 | ULONG_PTR p2m = (ULONG_PTR)p2; |
62 | |
|
63 | 0 | if (p1m <= p2m) |
64 | 0 | { |
65 | 0 | ULONG_PTR p1mEnd = p1m + 1ull * (height - 1) * p1Step + 1ull * width * p1Size; |
66 | |
|
67 | 0 | if (p1mEnd > p2m) |
68 | 0 | return TRUE; |
69 | 0 | } |
70 | 0 | else |
71 | 0 | { |
72 | 0 | ULONG_PTR p2mEnd = p2m + 1ull * (height - 1) * p2Step + 1ull * width * p2Size; |
73 | |
|
74 | 0 | if (p2mEnd > p1m) |
75 | 0 | return TRUE; |
76 | 0 | } |
77 | | |
78 | | /* else */ |
79 | 0 | return FALSE; |
80 | 0 | } |
81 | | |
82 | | /* ------------------------------------------------------------------------- */ |
83 | | static pstatus_t general_copy_8u(const BYTE* pSrc, BYTE* pDst, INT32 len) |
84 | 0 | { |
85 | 0 | if (memory_regions_overlap_1d(pSrc, pDst, (size_t)len)) |
86 | 0 | { |
87 | 0 | memmove((void*)pDst, (const void*)pSrc, (size_t)len); |
88 | 0 | } |
89 | 0 | else |
90 | 0 | { |
91 | 0 | memcpy((void*)pDst, (const void*)pSrc, (size_t)len); |
92 | 0 | } |
93 | |
|
94 | 0 | return PRIMITIVES_SUCCESS; |
95 | 0 | } |
96 | | |
97 | | /* ------------------------------------------------------------------------- */ |
98 | | /* Copy a block of pixels from one buffer to another. |
99 | | * The addresses are assumed to have been already offset to the upper-left |
100 | | * corners of the source and destination region of interest. |
101 | | */ |
102 | | static pstatus_t general_copy_8u_AC4r(const BYTE* pSrc, INT32 srcStep, BYTE* pDst, INT32 dstStep, |
103 | | INT32 width, INT32 height) |
104 | 0 | { |
105 | 0 | const BYTE* src = pSrc; |
106 | 0 | BYTE* dst = pDst; |
107 | 0 | int rowbytes = width * sizeof(UINT32); |
108 | |
|
109 | 0 | if ((width == 0) || (height == 0)) |
110 | 0 | return PRIMITIVES_SUCCESS; |
111 | | |
112 | 0 | if (memory_regions_overlap_2d(pSrc, srcStep, sizeof(UINT32), pDst, dstStep, sizeof(UINT32), |
113 | 0 | width, height)) |
114 | 0 | { |
115 | 0 | do |
116 | 0 | { |
117 | 0 | generic->copy(src, dst, rowbytes); |
118 | 0 | src += srcStep; |
119 | 0 | dst += dstStep; |
120 | 0 | } while (--height); |
121 | 0 | } |
122 | 0 | else |
123 | 0 | { |
124 | | /* TODO: do it in one operation when the rowdata is adjacent. */ |
125 | 0 | do |
126 | 0 | { |
127 | | /* If we find a replacement for memcpy that is consistently |
128 | | * faster, this could be replaced with that. |
129 | | */ |
130 | 0 | memcpy(dst, src, rowbytes); |
131 | 0 | src += srcStep; |
132 | 0 | dst += dstStep; |
133 | 0 | } while (--height); |
134 | 0 | } |
135 | |
|
136 | 0 | return PRIMITIVES_SUCCESS; |
137 | 0 | } |
138 | | |
139 | | static INLINE pstatus_t generic_image_copy_bgr24_bgrx32(BYTE* WINPR_RESTRICT pDstData, |
140 | | UINT32 nDstStep, UINT32 nXDst, UINT32 nYDst, |
141 | | UINT32 nWidth, UINT32 nHeight, |
142 | | const BYTE* WINPR_RESTRICT pSrcData, |
143 | | UINT32 nSrcStep, UINT32 nXSrc, UINT32 nYSrc, |
144 | | SSIZE_T srcVMultiplier, SSIZE_T srcVOffset, |
145 | | SSIZE_T dstVMultiplier, SSIZE_T dstVOffset) |
146 | 0 | { |
147 | |
|
148 | 0 | const SSIZE_T srcByte = 3; |
149 | 0 | const SSIZE_T dstByte = 4; |
150 | |
|
151 | 0 | for (SSIZE_T y = 0; y < nHeight; y++) |
152 | 0 | { |
153 | 0 | const BYTE* WINPR_RESTRICT srcLine = |
154 | 0 | &pSrcData[srcVMultiplier * (y + nYSrc) * nSrcStep + srcVOffset]; |
155 | 0 | BYTE* WINPR_RESTRICT dstLine = |
156 | 0 | &pDstData[dstVMultiplier * (y + nYDst) * nDstStep + dstVOffset]; |
157 | |
|
158 | 0 | for (SSIZE_T x = 0; x < nWidth; x++) |
159 | 0 | { |
160 | 0 | dstLine[(x + nXDst) * dstByte + 0] = srcLine[(x + nXSrc) * srcByte + 0]; |
161 | 0 | dstLine[(x + nXDst) * dstByte + 1] = srcLine[(x + nXSrc) * srcByte + 1]; |
162 | 0 | dstLine[(x + nXDst) * dstByte + 2] = srcLine[(x + nXSrc) * srcByte + 2]; |
163 | 0 | } |
164 | 0 | } |
165 | |
|
166 | 0 | return PRIMITIVES_SUCCESS; |
167 | 0 | } |
168 | | |
169 | | static INLINE pstatus_t generic_image_copy_bgrx32_bgrx32( |
170 | | BYTE* WINPR_RESTRICT pDstData, UINT32 nDstStep, UINT32 nXDst, UINT32 nYDst, UINT32 nWidth, |
171 | | UINT32 nHeight, const BYTE* WINPR_RESTRICT pSrcData, UINT32 nSrcStep, UINT32 nXSrc, |
172 | | UINT32 nYSrc, SSIZE_T srcVMultiplier, SSIZE_T srcVOffset, SSIZE_T dstVMultiplier, |
173 | | SSIZE_T dstVOffset) |
174 | 0 | { |
175 | |
|
176 | 0 | const SSIZE_T srcByte = 4; |
177 | 0 | const SSIZE_T dstByte = 4; |
178 | |
|
179 | 0 | for (SSIZE_T y = 0; y < nHeight; y++) |
180 | 0 | { |
181 | 0 | const BYTE* WINPR_RESTRICT srcLine = |
182 | 0 | &pSrcData[srcVMultiplier * (y + nYSrc) * nSrcStep + srcVOffset]; |
183 | 0 | BYTE* WINPR_RESTRICT dstLine = |
184 | 0 | &pDstData[dstVMultiplier * (y + nYDst) * nDstStep + dstVOffset]; |
185 | |
|
186 | 0 | for (SSIZE_T x = 0; x < nWidth; x++) |
187 | 0 | { |
188 | 0 | dstLine[(x + nXDst) * dstByte + 0] = srcLine[(x + nXSrc) * srcByte + 0]; |
189 | 0 | dstLine[(x + nXDst) * dstByte + 1] = srcLine[(x + nXSrc) * srcByte + 1]; |
190 | 0 | dstLine[(x + nXDst) * dstByte + 2] = srcLine[(x + nXSrc) * srcByte + 2]; |
191 | 0 | } |
192 | 0 | } |
193 | |
|
194 | 0 | return PRIMITIVES_SUCCESS; |
195 | 0 | } |
196 | | |
197 | | pstatus_t generic_image_copy_no_overlap_convert( |
198 | | BYTE* WINPR_RESTRICT pDstData, DWORD DstFormat, UINT32 nDstStep, UINT32 nXDst, UINT32 nYDst, |
199 | | UINT32 nWidth, UINT32 nHeight, const BYTE* WINPR_RESTRICT pSrcData, DWORD SrcFormat, |
200 | | UINT32 nSrcStep, UINT32 nXSrc, UINT32 nYSrc, const gdiPalette* WINPR_RESTRICT palette, |
201 | | SSIZE_T srcVMultiplier, SSIZE_T srcVOffset, SSIZE_T dstVMultiplier, SSIZE_T dstVOffset) |
202 | 0 | { |
203 | 0 | const SSIZE_T srcByte = FreeRDPGetBytesPerPixel(SrcFormat); |
204 | 0 | const SSIZE_T dstByte = FreeRDPGetBytesPerPixel(DstFormat); |
205 | |
|
206 | 0 | const UINT32 width = nWidth - nWidth % 8; |
207 | 0 | for (SSIZE_T y = 0; y < nHeight; y++) |
208 | 0 | { |
209 | 0 | const BYTE* WINPR_RESTRICT srcLine = |
210 | 0 | &pSrcData[srcVMultiplier * (y + nYSrc) * nSrcStep + srcVOffset]; |
211 | 0 | BYTE* WINPR_RESTRICT dstLine = |
212 | 0 | &pDstData[dstVMultiplier * (y + nYDst) * nDstStep + dstVOffset]; |
213 | |
|
214 | 0 | SSIZE_T x = 0; |
215 | 0 | WINPR_PRAGMA_UNROLL_LOOP |
216 | 0 | for (; x < width; x++) |
217 | 0 | { |
218 | 0 | const UINT32 color = FreeRDPReadColor_int(&srcLine[(x + nXSrc) * srcByte], SrcFormat); |
219 | 0 | const UINT32 dstColor = FreeRDPConvertColor(color, SrcFormat, DstFormat, palette); |
220 | 0 | FreeRDPWriteColor_int(&dstLine[(x + nXDst) * dstByte], DstFormat, dstColor); |
221 | 0 | } |
222 | 0 | for (; x < nWidth; x++) |
223 | 0 | { |
224 | 0 | const UINT32 color = FreeRDPReadColor_int(&srcLine[(x + nXSrc) * srcByte], SrcFormat); |
225 | 0 | const UINT32 dstColor = FreeRDPConvertColor(color, SrcFormat, DstFormat, palette); |
226 | 0 | FreeRDPWriteColor_int(&dstLine[(x + nXDst) * dstByte], DstFormat, dstColor); |
227 | 0 | } |
228 | 0 | } |
229 | 0 | return PRIMITIVES_SUCCESS; |
230 | 0 | } |
231 | | |
232 | | pstatus_t generic_image_copy_no_overlap_memcpy( |
233 | | BYTE* WINPR_RESTRICT pDstData, DWORD DstFormat, UINT32 nDstStep, UINT32 nXDst, UINT32 nYDst, |
234 | | UINT32 nWidth, UINT32 nHeight, const BYTE* WINPR_RESTRICT pSrcData, DWORD SrcFormat, |
235 | | UINT32 nSrcStep, UINT32 nXSrc, UINT32 nYSrc, const gdiPalette* WINPR_RESTRICT palette, |
236 | | SSIZE_T srcVMultiplier, SSIZE_T srcVOffset, SSIZE_T dstVMultiplier, SSIZE_T dstVOffset, |
237 | | UINT32 flags) |
238 | 0 | { |
239 | 0 | const SSIZE_T dstByte = FreeRDPGetBytesPerPixel(DstFormat); |
240 | 0 | const SSIZE_T srcByte = FreeRDPGetBytesPerPixel(SrcFormat); |
241 | 0 | const SSIZE_T copyDstWidth = nWidth * dstByte; |
242 | 0 | const SSIZE_T xSrcOffset = nXSrc * srcByte; |
243 | 0 | const SSIZE_T xDstOffset = nXDst * dstByte; |
244 | |
|
245 | 0 | for (SSIZE_T y = 0; y < nHeight; y++) |
246 | 0 | { |
247 | 0 | const BYTE* WINPR_RESTRICT srcLine = |
248 | 0 | &pSrcData[srcVMultiplier * (y + nYSrc) * nSrcStep + srcVOffset]; |
249 | 0 | BYTE* WINPR_RESTRICT dstLine = |
250 | 0 | &pDstData[dstVMultiplier * (y + nYDst) * nDstStep + dstVOffset]; |
251 | 0 | memcpy(&dstLine[xDstOffset], &srcLine[xSrcOffset], copyDstWidth); |
252 | 0 | } |
253 | |
|
254 | 0 | return PRIMITIVES_SUCCESS; |
255 | 0 | } |
256 | | |
257 | | static INLINE pstatus_t generic_image_copy_no_overlap_dst_alpha( |
258 | | BYTE* WINPR_RESTRICT pDstData, DWORD DstFormat, UINT32 nDstStep, UINT32 nXDst, UINT32 nYDst, |
259 | | UINT32 nWidth, UINT32 nHeight, const BYTE* WINPR_RESTRICT pSrcData, DWORD SrcFormat, |
260 | | UINT32 nSrcStep, UINT32 nXSrc, UINT32 nYSrc, const gdiPalette* WINPR_RESTRICT palette, |
261 | | SSIZE_T srcVMultiplier, SSIZE_T srcVOffset, SSIZE_T dstVMultiplier, SSIZE_T dstVOffset) |
262 | 0 | { |
263 | 0 | WINPR_ASSERT(pDstData); |
264 | 0 | WINPR_ASSERT(pSrcData); |
265 | | |
266 | 0 | switch (SrcFormat) |
267 | 0 | { |
268 | 0 | case PIXEL_FORMAT_BGR24: |
269 | 0 | switch (DstFormat) |
270 | 0 | { |
271 | 0 | case PIXEL_FORMAT_BGRX32: |
272 | 0 | case PIXEL_FORMAT_BGRA32: |
273 | 0 | return generic_image_copy_bgr24_bgrx32( |
274 | 0 | pDstData, nDstStep, nXDst, nYDst, nWidth, nHeight, pSrcData, nSrcStep, |
275 | 0 | nXSrc, nYSrc, srcVMultiplier, srcVOffset, dstVMultiplier, dstVOffset); |
276 | 0 | default: |
277 | 0 | break; |
278 | 0 | } |
279 | 0 | break; |
280 | 0 | case PIXEL_FORMAT_BGRX32: |
281 | 0 | case PIXEL_FORMAT_BGRA32: |
282 | 0 | switch (DstFormat) |
283 | 0 | { |
284 | 0 | case PIXEL_FORMAT_BGRX32: |
285 | 0 | case PIXEL_FORMAT_BGRA32: |
286 | 0 | return generic_image_copy_bgrx32_bgrx32( |
287 | 0 | pDstData, nDstStep, nXDst, nYDst, nWidth, nHeight, pSrcData, nSrcStep, |
288 | 0 | nXSrc, nYSrc, srcVMultiplier, srcVOffset, dstVMultiplier, dstVOffset); |
289 | 0 | case PIXEL_FORMAT_BGR24: |
290 | 0 | return generic_image_copy_bgr24_bgrx32( |
291 | 0 | pDstData, nDstStep, nXDst, nYDst, nWidth, nHeight, pSrcData, nSrcStep, |
292 | 0 | nXSrc, nYSrc, srcVMultiplier, srcVOffset, dstVMultiplier, dstVOffset); |
293 | 0 | default: |
294 | 0 | break; |
295 | 0 | } |
296 | 0 | break; |
297 | 0 | default: |
298 | 0 | break; |
299 | 0 | } |
300 | | |
301 | 0 | return generic_image_copy_no_overlap_convert( |
302 | 0 | pDstData, DstFormat, nDstStep, nXDst, nYDst, nWidth, nHeight, pSrcData, SrcFormat, nSrcStep, |
303 | 0 | nXSrc, nYSrc, palette, srcVMultiplier, srcVOffset, dstVMultiplier, dstVOffset); |
304 | 0 | } |
305 | | |
306 | | static INLINE pstatus_t generic_image_copy_no_overlap_no_alpha( |
307 | | BYTE* WINPR_RESTRICT pDstData, DWORD DstFormat, UINT32 nDstStep, UINT32 nXDst, UINT32 nYDst, |
308 | | UINT32 nWidth, UINT32 nHeight, const BYTE* WINPR_RESTRICT pSrcData, DWORD SrcFormat, |
309 | | UINT32 nSrcStep, UINT32 nXSrc, UINT32 nYSrc, const gdiPalette* WINPR_RESTRICT palette, |
310 | | SSIZE_T srcVMultiplier, SSIZE_T srcVOffset, SSIZE_T dstVMultiplier, SSIZE_T dstVOffset, |
311 | | UINT32 flags) |
312 | 0 | { |
313 | 0 | if (FreeRDPAreColorFormatsEqualNoAlpha(SrcFormat, DstFormat)) |
314 | 0 | return generic_image_copy_no_overlap_memcpy(pDstData, DstFormat, nDstStep, nXDst, nYDst, |
315 | 0 | nWidth, nHeight, pSrcData, SrcFormat, nSrcStep, |
316 | 0 | nXSrc, nYSrc, palette, srcVMultiplier, |
317 | 0 | srcVOffset, dstVMultiplier, dstVOffset, flags); |
318 | 0 | else |
319 | 0 | return generic_image_copy_no_overlap_convert(pDstData, DstFormat, nDstStep, nXDst, nYDst, |
320 | 0 | nWidth, nHeight, pSrcData, SrcFormat, nSrcStep, |
321 | 0 | nXSrc, nYSrc, palette, srcVMultiplier, |
322 | 0 | srcVOffset, dstVMultiplier, dstVOffset); |
323 | 0 | } |
324 | | |
325 | | static pstatus_t generic_image_copy_no_overlap(BYTE* WINPR_RESTRICT pDstData, DWORD DstFormat, |
326 | | UINT32 nDstStep, UINT32 nXDst, UINT32 nYDst, |
327 | | UINT32 nWidth, UINT32 nHeight, |
328 | | const BYTE* WINPR_RESTRICT pSrcData, DWORD SrcFormat, |
329 | | UINT32 nSrcStep, UINT32 nXSrc, UINT32 nYSrc, |
330 | | const gdiPalette* WINPR_RESTRICT palette, |
331 | | UINT32 flags) |
332 | 0 | { |
333 | 0 | const BOOL vSrcVFlip = (flags & FREERDP_FLIP_VERTICAL) ? TRUE : FALSE; |
334 | 0 | SSIZE_T srcVOffset = 0; |
335 | 0 | SSIZE_T srcVMultiplier = 1; |
336 | 0 | SSIZE_T dstVOffset = 0; |
337 | 0 | SSIZE_T dstVMultiplier = 1; |
338 | |
|
339 | 0 | if ((nWidth == 0) || (nHeight == 0)) |
340 | 0 | return PRIMITIVES_SUCCESS; |
341 | | |
342 | 0 | if ((nHeight > INT32_MAX) || (nWidth > INT32_MAX)) |
343 | 0 | return -1; |
344 | | |
345 | 0 | if (!pDstData || !pSrcData) |
346 | 0 | return -1; |
347 | | |
348 | 0 | if (nDstStep == 0) |
349 | 0 | nDstStep = nWidth * FreeRDPGetBytesPerPixel(DstFormat); |
350 | |
|
351 | 0 | if (nSrcStep == 0) |
352 | 0 | nSrcStep = nWidth * FreeRDPGetBytesPerPixel(SrcFormat); |
353 | |
|
354 | 0 | if (vSrcVFlip) |
355 | 0 | { |
356 | 0 | srcVOffset = (nHeight - 1ll) * nSrcStep; |
357 | 0 | srcVMultiplier = -1; |
358 | 0 | } |
359 | |
|
360 | 0 | if (((flags & FREERDP_KEEP_DST_ALPHA) != 0) && FreeRDPColorHasAlpha(DstFormat)) |
361 | 0 | return generic_image_copy_no_overlap_dst_alpha( |
362 | 0 | pDstData, DstFormat, nDstStep, nXDst, nYDst, nWidth, nHeight, pSrcData, SrcFormat, |
363 | 0 | nSrcStep, nXSrc, nYSrc, palette, srcVMultiplier, srcVOffset, dstVMultiplier, |
364 | 0 | dstVOffset); |
365 | 0 | else |
366 | 0 | return generic_image_copy_no_overlap_no_alpha( |
367 | 0 | pDstData, DstFormat, nDstStep, nXDst, nYDst, nWidth, nHeight, pSrcData, SrcFormat, |
368 | 0 | nSrcStep, nXSrc, nYSrc, palette, srcVMultiplier, srcVOffset, dstVMultiplier, dstVOffset, |
369 | 0 | flags); |
370 | | |
371 | 0 | return PRIMITIVES_SUCCESS; |
372 | 0 | } |
373 | | |
374 | | /* ------------------------------------------------------------------------- */ |
375 | | void primitives_init_copy(primitives_t* prims) |
376 | 0 | { |
377 | | /* Start with the default. */ |
378 | 0 | prims->copy_8u = general_copy_8u; |
379 | 0 | prims->copy_8u_AC4r = general_copy_8u_AC4r; |
380 | 0 | prims->copy = WINPR_FUNC_PTR_CAST(prims->copy_8u, __copy_t); |
381 | 0 | prims->copy_no_overlap = generic_image_copy_no_overlap; |
382 | 0 | } |
383 | | |
384 | | void primitives_init_copy_opt(primitives_t* prims) |
385 | 0 | { |
386 | 0 | primitives_init_copy_sse41(prims); |
387 | 0 | primitives_init_copy_avx2(prims); |
388 | 0 | } |