/src/zstd/lib/common/fse_decompress.c

Source
/* ******************************************************************
 * FSE : Finite State Entropy decoder
 * Copyright (c) Meta Platforms, Inc. and affiliates.
 *
 *  You can contact the author at :
 *  - FSE source repository : https://github.com/Cyan4973/FiniteStateEntropy
 *  - Public forum : https://groups.google.com/forum/#!forum/lz4c
 *
 * This source code is licensed under both the BSD-style license (found in the
 * LICENSE file in the root directory of this source tree) and the GPLv2 (found
 * in the COPYING file in the root directory of this source tree).
 * You may select, at your option, one of the above-listed licenses.
****************************************************************** */


/* **************************************************************
*  Includes
****************************************************************/
#include "debug.h"      /* assert */
#include "bitstream.h"
#include "compiler.h"
#define FSE_STATIC_LINKING_ONLY
#include "fse.h"
#include "error_private.h"
#include "zstd_deps.h"  /* ZSTD_memcpy */
#include "bits.h"       /* ZSTD_highbit32 */


/* **************************************************************
*  Error Management
****************************************************************/
#define FSE_isError ERR_isError
#define FSE_STATIC_ASSERT(c) DEBUG_STATIC_ASSERT(c)   /* use only *after* variable declarations */


/* **************************************************************
*  Templates
****************************************************************/
/*
  designed to be included
  for type-specific functions (template emulation in C)
  Objective is to write these functions only once, for improved maintenance
*/

/* safety checks */
#ifndef FSE_FUNCTION_EXTENSION
#  error "FSE_FUNCTION_EXTENSION must be defined"
#endif
#ifndef FSE_FUNCTION_TYPE
#  error "FSE_FUNCTION_TYPE must be defined"
#endif

/* Function names */
#define FSE_CAT(X,Y) X##Y
#define FSE_FUNCTION_NAME(X,Y) FSE_CAT(X,Y)
#define FSE_TYPE_NAME(X,Y) FSE_CAT(X,Y)

static size_t FSE_buildDTable_internal(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize)
{
    void* const tdPtr = dt+1;   /* because *dt is unsigned, 32-bits aligned on 32-bits */
    FSE_DECODE_TYPE* const tableDecode = (FSE_DECODE_TYPE*) (tdPtr);
    U16* symbolNext = (U16*)workSpace;
    BYTE* spread = (BYTE*)(symbolNext + maxSymbolValue + 1);

    U32 const maxSV1 = maxSymbolValue + 1;
    U32 const tableSize = 1 << tableLog;
    U32 highThreshold = tableSize-1;

    /* Sanity Checks */
    if (FSE_BUILD_DTABLE_WKSP_SIZE(tableLog, maxSymbolValue) > wkspSize) return ERROR(maxSymbolValue_tooLarge);
    if (maxSymbolValue > FSE_MAX_SYMBOL_VALUE) return ERROR(maxSymbolValue_tooLarge);
    if (tableLog > FSE_MAX_TABLELOG) return ERROR(tableLog_tooLarge);

    /* Init, lay down lowprob symbols */
    {   FSE_DTableHeader DTableH;
        DTableH.tableLog = (U16)tableLog;
        DTableH.fastMode = 1;
        {   S16 const largeLimit= (S16)(1 << (tableLog-1));
            U32 s;
            for (s=0; s<maxSV1; s++) {
                if (normalizedCounter[s]==-1) {
                    tableDecode[highThreshold--].symbol = (FSE_FUNCTION_TYPE)s;
                    symbolNext[s] = 1;
                } else {
                    if (normalizedCounter[s] >= largeLimit) DTableH.fastMode=0;
                    symbolNext[s] = (U16)normalizedCounter[s];
        }   }   }
        ZSTD_memcpy(dt, &DTableH, sizeof(DTableH));
    }

    /* Spread symbols */
    if (highThreshold == tableSize - 1) {
        size_t const tableMask = tableSize-1;
        size_t const step = FSE_TABLESTEP(tableSize);
        /* First lay down the symbols in order.
         * We use a uint64_t to lay down 8 bytes at a time. This reduces branch
         * misses since small blocks generally have small table logs, so nearly
         * all symbols have counts <= 8. We ensure we have 8 bytes at the end of
         * our buffer to handle the over-write.
         */
        {   U64 const add = 0x0101010101010101ull;
            size_t pos = 0;
            U64 sv = 0;
            U32 s;
            for (s=0; s<maxSV1; ++s, sv += add) {
                int i;
                int const n = normalizedCounter[s];
                MEM_write64(spread + pos, sv);
                for (i = 8; i < n; i += 8) {
                    MEM_write64(spread + pos + i, sv);
                }
                pos += (size_t)n;
        }   }
        /* Now we spread those positions across the table.
         * The benefit of doing it in two stages is that we avoid the
         * variable size inner loop, which caused lots of branch misses.
         * Now we can run through all the positions without any branch misses.
         * We unroll the loop twice, since that is what empirically worked best.
         */
        {
            size_t position = 0;
            size_t s;
            size_t const unroll = 2;
            assert(tableSize % unroll == 0); /* FSE_MIN_TABLELOG is 5 */
            for (s = 0; s < (size_t)tableSize; s += unroll) {
                size_t u;
                for (u = 0; u < unroll; ++u) {
                    size_t const uPosition = (position + (u * step)) & tableMask;
                    tableDecode[uPosition].symbol = spread[s + u];
                }
                position = (position + (unroll * step)) & tableMask;
            }
            assert(position == 0);
        }
    } else {
        U32 const tableMask = tableSize-1;
        U32 const step = FSE_TABLESTEP(tableSize);
        U32 s, position = 0;
        for (s=0; s<maxSV1; s++) {
            int i;
            for (i=0; i<normalizedCounter[s]; i++) {
                tableDecode[position].symbol = (FSE_FUNCTION_TYPE)s;
                position = (position + step) & tableMask;
                while (position > highThreshold) position = (position + step) & tableMask;   /* lowprob area */
        }   }
        if (position!=0) return ERROR(GENERIC);   /* position must reach all cells once, otherwise normalizedCounter is incorrect */
    }

    /* Build Decoding table */
    {   U32 u;
        for (u=0; u<tableSize; u++) {
            FSE_FUNCTION_TYPE const symbol = (FSE_FUNCTION_TYPE)(tableDecode[u].symbol);
            U32 const nextState = symbolNext[symbol]++;
            tableDecode[u].nbBits = (BYTE) (tableLog - ZSTD_highbit32(nextState) );
            tableDecode[u].newState = (U16) ( (nextState << tableDecode[u].nbBits) - tableSize);
    }   }

    return 0;
}

size_t FSE_buildDTable_wksp(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize)
{
    return FSE_buildDTable_internal(dt, normalizedCounter, maxSymbolValue, tableLog, workSpace, wkspSize);
}


#ifndef FSE_COMMONDEFS_ONLY

/*-*******************************************************
*  Decompression (Byte symbols)
*********************************************************/

FORCE_INLINE_TEMPLATE size_t FSE_decompress_usingDTable_generic(
          void* dst, size_t maxDstSize,
    const void* cSrc, size_t cSrcSize,
    const FSE_DTable* dt, const unsigned fast)
{
    BYTE* const ostart = (BYTE*) dst;
    BYTE* op = ostart;
    BYTE* const omax = op + maxDstSize;
    BYTE* const olimit = omax-3;

    BIT_DStream_t bitD;
    FSE_DState_t state1;
    FSE_DState_t state2;

    /* Init */
    CHECK_F(BIT_initDStream(&bitD, cSrc, cSrcSize));

    FSE_initDState(&state1, &bitD, dt);
    FSE_initDState(&state2, &bitD, dt);

    RETURN_ERROR_IF(BIT_reloadDStream(&bitD)==BIT_DStream_overflow, corruption_detected, "");

#define FSE_GETSYMBOL(statePtr) fast ? FSE_decodeSymbolFast(statePtr, &bitD) : FSE_decodeSymbol(statePtr, &bitD)

    /* 4 symbols per loop */
    for ( ; (BIT_reloadDStream(&bitD)==BIT_DStream_unfinished) & (op<olimit) ; op+=4) {
        op[0] = FSE_GETSYMBOL(&state1);

        if (FSE_MAX_TABLELOG*2+7 > sizeof(bitD.bitContainer)*8)    /* This test must be static */
            BIT_reloadDStream(&bitD);

        op[1] = FSE_GETSYMBOL(&state2);

        if (FSE_MAX_TABLELOG*4+7 > sizeof(bitD.bitContainer)*8)    /* This test must be static */
            { if (BIT_reloadDStream(&bitD) > BIT_DStream_unfinished) { op+=2; break; } }

        op[2] = FSE_GETSYMBOL(&state1);

        if (FSE_MAX_TABLELOG*2+7 > sizeof(bitD.bitContainer)*8)    /* This test must be static */
            BIT_reloadDStream(&bitD);

        op[3] = FSE_GETSYMBOL(&state2);
    }

    /* tail */
    /* note : BIT_reloadDStream(&bitD) >= FSE_DStream_partiallyFilled; Ends at exactly BIT_DStream_completed */
    while (1) {
        if (op>(omax-2)) return ERROR(dstSize_tooSmall);
        *op++ = FSE_GETSYMBOL(&state1);
        if (BIT_reloadDStream(&bitD)==BIT_DStream_overflow) {
            *op++ = FSE_GETSYMBOL(&state2);
            break;
        }

        if (op>(omax-2)) return ERROR(dstSize_tooSmall);
        *op++ = FSE_GETSYMBOL(&state2);
        if (BIT_reloadDStream(&bitD)==BIT_DStream_overflow) {
            *op++ = FSE_GETSYMBOL(&state1);
            break;
    }   }

    assert(op >= ostart);
    return (size_t)(op-ostart);
}

typedef struct {
    short ncount[FSE_MAX_SYMBOL_VALUE + 1];
} FSE_DecompressWksp;


FORCE_INLINE_TEMPLATE size_t FSE_decompress_wksp_body(
        void* dst, size_t dstCapacity,
        const void* cSrc, size_t cSrcSize,
        unsigned maxLog, void* workSpace, size_t wkspSize,
        int bmi2)
{
    const BYTE* const istart = (const BYTE*)cSrc;
    const BYTE* ip = istart;
    unsigned tableLog;
    unsigned maxSymbolValue = FSE_MAX_SYMBOL_VALUE;
    FSE_DecompressWksp* const wksp = (FSE_DecompressWksp*)workSpace;
    size_t const dtablePos = sizeof(FSE_DecompressWksp) / sizeof(FSE_DTable);
    FSE_DTable* const dtable = (FSE_DTable*)workSpace + dtablePos;

    FSE_STATIC_ASSERT((FSE_MAX_SYMBOL_VALUE + 1) % 2 == 0);
    if (wkspSize < sizeof(*wksp)) return ERROR(GENERIC);

    /* correct offset to dtable depends on this property */
    FSE_STATIC_ASSERT(sizeof(FSE_DecompressWksp) % sizeof(FSE_DTable) == 0);

    /* normal FSE decoding mode */
    {   size_t const NCountLength =
            FSE_readNCount_bmi2(wksp->ncount, &maxSymbolValue, &tableLog, istart, cSrcSize, bmi2);
        if (FSE_isError(NCountLength)) return NCountLength;
        if (tableLog > maxLog) return ERROR(tableLog_tooLarge);
        assert(NCountLength <= cSrcSize);
        ip += NCountLength;
        cSrcSize -= NCountLength;
    }

    if (FSE_DECOMPRESS_WKSP_SIZE(tableLog, maxSymbolValue) > wkspSize) return ERROR(tableLog_tooLarge);
    assert(sizeof(*wksp) + FSE_DTABLE_SIZE(tableLog) <= wkspSize);
    workSpace = (BYTE*)workSpace + sizeof(*wksp) + FSE_DTABLE_SIZE(tableLog);
    wkspSize -= sizeof(*wksp) + FSE_DTABLE_SIZE(tableLog);

    CHECK_F( FSE_buildDTable_internal(dtable, wksp->ncount, maxSymbolValue, tableLog, workSpace, wkspSize) );

    {
        const void* ptr = dtable;
        const FSE_DTableHeader* DTableH = (const FSE_DTableHeader*)ptr;
        const U32 fastMode = DTableH->fastMode;

        /* select fast mode (static) */
        if (fastMode) return FSE_decompress_usingDTable_generic(dst, dstCapacity, ip, cSrcSize, dtable, 1);
        return FSE_decompress_usingDTable_generic(dst, dstCapacity, ip, cSrcSize, dtable, 0);
    }
}

/* Avoids the FORCE_INLINE of the _body() function. */
static size_t FSE_decompress_wksp_body_default(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize)
{
    return FSE_decompress_wksp_body(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize, 0);
}

#if DYNAMIC_BMI2
BMI2_TARGET_ATTRIBUTE static size_t FSE_decompress_wksp_body_bmi2(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize)
{
    return FSE_decompress_wksp_body(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize, 1);
}
#endif

size_t FSE_decompress_wksp_bmi2(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize, int bmi2)
{
#if DYNAMIC_BMI2
    if (bmi2) {
        return FSE_decompress_wksp_body_bmi2(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize);
    }
#endif
    (void)bmi2;
    return FSE_decompress_wksp_body_default(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize);
}

#endif   /* FSE_COMMONDEFS_ONLY */

Coverage Report

Created: 2025-11-16 07:22

Line	Count	Source
1		/* ******************************************************************
2		* FSE : Finite State Entropy decoder
3		* Copyright (c) Meta Platforms, Inc. and affiliates.
4		*
5		* You can contact the author at :
6		* - FSE source repository : https://github.com/Cyan4973/FiniteStateEntropy
7		* - Public forum : https://groups.google.com/forum/#!forum/lz4c
8		*
9		* This source code is licensed under both the BSD-style license (found in the
10		* LICENSE file in the root directory of this source tree) and the GPLv2 (found
11		* in the COPYING file in the root directory of this source tree).
12		* You may select, at your option, one of the above-listed licenses.
13		****************************************************************** */
14
15
16		/* **************************************************************
17		* Includes
18		****************************************************************/
19		#include "debug.h" /* assert */
20		#include "bitstream.h"
21		#include "compiler.h"
22		#define FSE_STATIC_LINKING_ONLY
23		#include "fse.h"
24		#include "error_private.h"
25		#include "zstd_deps.h" /* ZSTD_memcpy */
26		#include "bits.h" /* ZSTD_highbit32 */
27
28
29		/* **************************************************************
30		* Error Management
31		****************************************************************/
32	921	#define FSE_isError ERR_isError
33	1.84k	#define FSE_STATIC_ASSERT(c) DEBUG_STATIC_ASSERT(c) /* use only after variable declarations */
34
35
36		/* **************************************************************
37		* Templates
38		****************************************************************/
39		/*
40		designed to be included
41		for type-specific functions (template emulation in C)
42		Objective is to write these functions only once, for improved maintenance
43		*/
44
45		/* safety checks */
46		#ifndef FSE_FUNCTION_EXTENSION
47		# error "FSE_FUNCTION_EXTENSION must be defined"
48		#endif
49		#ifndef FSE_FUNCTION_TYPE
50		# error "FSE_FUNCTION_TYPE must be defined"
51		#endif
52
53		/* Function names */
54		#define FSE_CAT(X,Y) X##Y
55		#define FSE_FUNCTION_NAME(X,Y) FSE_CAT(X,Y)
56		#define FSE_TYPE_NAME(X,Y) FSE_CAT(X,Y)
57
58		static size_t FSE_buildDTable_internal(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize)
59	820	{
60	820	void* const tdPtr = dt+1; /* because dt is unsigned, 32-bits aligned on 32-bits /
61	820	FSE_DECODE_TYPE* const tableDecode = (FSE_DECODE_TYPE*) (tdPtr);
62	820	U16* symbolNext = (U16*)workSpace;
63	820	BYTE* spread = (BYTE*)(symbolNext + maxSymbolValue + 1);
64
65	820	U32 const maxSV1 = maxSymbolValue + 1;
66	820	U32 const tableSize = 1 << tableLog;
67	820	U32 highThreshold = tableSize-1;
68
69		/* Sanity Checks */
70	820	if (FSE_BUILD_DTABLE_WKSP_SIZE(tableLog, maxSymbolValue) > wkspSize) return ERROR(maxSymbolValue_tooLarge);
71	820	if (maxSymbolValue > FSE_MAX_SYMBOL_VALUE) return ERROR(maxSymbolValue_tooLarge);
72	820	if (tableLog > FSE_MAX_TABLELOG) return ERROR(tableLog_tooLarge);
73
74		/* Init, lay down lowprob symbols */
75	820	{ FSE_DTableHeader DTableH;
76	820	DTableH.tableLog = (U16)tableLog;
77	820	DTableH.fastMode = 1;
78	820	{ S16 const largeLimit= (S16)(1 << (tableLog-1));
79	820	U32 s;
80	8.68k	for (s=0; s<maxSV1; s++) {
81	7.86k	if (normalizedCounter[s]==-1) {
82	3.42k	tableDecode[highThreshold--].symbol = (FSE_FUNCTION_TYPE)s;
83	3.42k	symbolNext[s] = 1;
84	4.43k	} else {
85	4.43k	if (normalizedCounter[s] >= largeLimit) DTableH.fastMode=0;
86	4.43k	symbolNext[s] = (U16)normalizedCounter[s];
87	4.43k	} } }
88	820	ZSTD_memcpy(dt, &DTableH, sizeof(DTableH));
89	820	}
90
91		/* Spread symbols */
92	820	if (highThreshold == tableSize - 1) {
93	105	size_t const tableMask = tableSize-1;
94	105	size_t const step = FSE_TABLESTEP(tableSize);
95		/* First lay down the symbols in order.
96		* We use a uint64_t to lay down 8 bytes at a time. This reduces branch
97		* misses since small blocks generally have small table logs, so nearly
98		* all symbols have counts <= 8. We ensure we have 8 bytes at the end of
99		* our buffer to handle the over-write.
100		*/
101	105	{ U64 const add = 0x0101010101010101ull;
102	105	size_t pos = 0;
103	105	U64 sv = 0;
104	105	U32 s;
105	770	for (s=0; s<maxSV1; ++s, sv += add) {
106	665	int i;
107	665	int const n = normalizedCounter[s];
108	665	MEM_write64(spread + pos, sv);
109	993	for (i = 8; i < n; i += 8) {
110	328	MEM_write64(spread + pos + i, sv);
111	328	}
112	665	pos += (size_t)n;
113	665	} }
114		/* Now we spread those positions across the table.
115		* The benefit of doing it in two stages is that we avoid the
116		* variable size inner loop, which caused lots of branch misses.
117		* Now we can run through all the positions without any branch misses.
118		* We unroll the loop twice, since that is what empirically worked best.
119		*/
120	105	{
121	105	size_t position = 0;
122	105	size_t s;
123	105	size_t const unroll = 2;
124	105	assert(tableSize % unroll == 0); /* FSE_MIN_TABLELOG is 5 */
125	2.02k	for (s = 0; s < (size_t)tableSize; s += unroll) {
126	1.92k	size_t u;
127	5.76k	for (u = 0; u < unroll; ++u) {
128	3.84k	size_t const uPosition = (position + (u * step)) & tableMask;
129	3.84k	tableDecode[uPosition].symbol = spread[s + u];
130	3.84k	}
131	1.92k	position = (position + (unroll * step)) & tableMask;
132	1.92k	}
133	105	assert(position == 0);
134	105	}
135	715	} else {
136	715	U32 const tableMask = tableSize-1;
137	715	U32 const step = FSE_TABLESTEP(tableSize);
138	715	U32 s, position = 0;
139	7.91k	for (s=0; s<maxSV1; s++) {
140	7.19k	int i;
141	27.7k	for (i=0; i<normalizedCounter[s]; i++) {
142	20.5k	tableDecode[position].symbol = (FSE_FUNCTION_TYPE)s;
143	20.5k	position = (position + step) & tableMask;
144	23.8k	while (position > highThreshold) position = (position + step) & tableMask; /* lowprob area */
145	20.5k	} }
146	715	if (position!=0) return ERROR(GENERIC); /* position must reach all cells once, otherwise normalizedCounter is incorrect */
147	715	}
148
149		/* Build Decoding table */
150	820	{ U32 u;
151	28.6k	for (u=0; u<tableSize; u++) {
152	27.8k	FSE_FUNCTION_TYPE const symbol = (FSE_FUNCTION_TYPE)(tableDecode[u].symbol);
153	27.8k	U32 const nextState = symbolNext[symbol]++;
154	27.8k	tableDecode[u].nbBits = (BYTE) (tableLog - ZSTD_highbit32(nextState) );
155	27.8k	tableDecode[u].newState = (U16) ( (nextState << tableDecode[u].nbBits) - tableSize);
156	27.8k	} }
157
158	820	return 0;
159	820	}
160
161		size_t FSE_buildDTable_wksp(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize)
162	0	{
163	0	return FSE_buildDTable_internal(dt, normalizedCounter, maxSymbolValue, tableLog, workSpace, wkspSize);
164	0	}
165
166
167		#ifndef FSE_COMMONDEFS_ONLY
168
169		/-******************************************************
170		* Decompression (Byte symbols)
171		*********************************************************/
172
173		FORCE_INLINE_TEMPLATE size_t FSE_decompress_usingDTable_generic(
174		void* dst, size_t maxDstSize,
175		const void* cSrc, size_t cSrcSize,
176		const FSE_DTable* dt, const unsigned fast)
177	820	{
178	820	BYTE* const ostart = (BYTE*) dst;
179	820	BYTE* op = ostart;
180	820	BYTE* const omax = op + maxDstSize;
181	820	BYTE* const olimit = omax-3;
182
183	820	BIT_DStream_t bitD;
184	820	FSE_DState_t state1;
185	820	FSE_DState_t state2;
186
187		/* Init */
188	820	CHECK_F(BIT_initDStream(&bitD, cSrc, cSrcSize));
189
190	793	FSE_initDState(&state1, &bitD, dt);
191	793	FSE_initDState(&state2, &bitD, dt);
192
193	793	RETURN_ERROR_IF(BIT_reloadDStream(&bitD)==BIT_DStream_overflow, corruption_detected, "");
194
195	100k	#define FSE_GETSYMBOL(statePtr) fast ? FSE_decodeSymbolFast(statePtr, &bitD) : FSE_decodeSymbol(statePtr, &bitD)
196
197		/* 4 symbols per loop */
198	17.0k	for ( ; (BIT_reloadDStream(&bitD)==BIT_DStream_unfinished) & (op<olimit) ; op+=4) {
199	16.2k	op[0] = FSE_GETSYMBOL(&state1);
200
201	16.2k	if (FSE_MAX_TABLELOG2+7 > sizeof(bitD.bitContainer)8) /* This test must be static */
202	0	BIT_reloadDStream(&bitD);
203
204	16.2k	op[1] = FSE_GETSYMBOL(&state2);
205
206	16.2k	if (FSE_MAX_TABLELOG4+7 > sizeof(bitD.bitContainer)8) /* This test must be static */
207	0	{ if (BIT_reloadDStream(&bitD) > BIT_DStream_unfinished) { op+=2; break; } }
208
209	16.2k	op[2] = FSE_GETSYMBOL(&state1);
210
211	16.2k	if (FSE_MAX_TABLELOG2+7 > sizeof(bitD.bitContainer)8) /* This test must be static */
212	0	BIT_reloadDStream(&bitD);
213
214	16.2k	op[3] = FSE_GETSYMBOL(&state2);
215	16.2k	}
216
217		/* tail */
218		/* note : BIT_reloadDStream(&bitD) >= FSE_DStream_partiallyFilled; Ends at exactly BIT_DStream_completed */
219	17.7k	while (1) {
220	17.7k	if (op>(omax-2)) return ERROR(dstSize_tooSmall);
221	17.7k	*op++ = FSE_GETSYMBOL(&state1);
222	17.7k	if (BIT_reloadDStream(&bitD)==BIT_DStream_overflow) {
223	350	*op++ = FSE_GETSYMBOL(&state2);
224	350	break;
225	350	}
226
227	17.3k	if (op>(omax-2)) return ERROR(dstSize_tooSmall);
228	17.3k	*op++ = FSE_GETSYMBOL(&state2);
229	17.3k	if (BIT_reloadDStream(&bitD)==BIT_DStream_overflow) {
230	415	*op++ = FSE_GETSYMBOL(&state1);
231	415	break;
232	415	} }
233
234	765	assert(op >= ostart);
235	765	return (size_t)(op-ostart);
236	784	}
237
238		typedef struct {
239		short ncount[FSE_MAX_SYMBOL_VALUE + 1];
240		} FSE_DecompressWksp;
241
242
243		FORCE_INLINE_TEMPLATE size_t FSE_decompress_wksp_body(
244		void* dst, size_t dstCapacity,
245		const void* cSrc, size_t cSrcSize,
246		unsigned maxLog, void* workSpace, size_t wkspSize,
247		int bmi2)
248	921	{
249	921	const BYTE* const istart = (const BYTE*)cSrc;
250	921	const BYTE* ip = istart;
251	921	unsigned tableLog;
252	921	unsigned maxSymbolValue = FSE_MAX_SYMBOL_VALUE;
253	921	FSE_DecompressWksp* const wksp = (FSE_DecompressWksp*)workSpace;
254	921	size_t const dtablePos = sizeof(FSE_DecompressWksp) / sizeof(FSE_DTable);
255	921	FSE_DTable* const dtable = (FSE_DTable*)workSpace + dtablePos;
256
257	921	FSE_STATIC_ASSERT((FSE_MAX_SYMBOL_VALUE + 1) % 2 == 0);
258	921	if (wkspSize < sizeof(*wksp)) return ERROR(GENERIC);
259
260		/* correct offset to dtable depends on this property */
261	921	FSE_STATIC_ASSERT(sizeof(FSE_DecompressWksp) % sizeof(FSE_DTable) == 0);
262
263		/* normal FSE decoding mode */
264	921	{ size_t const NCountLength =
265	921	FSE_readNCount_bmi2(wksp->ncount, &maxSymbolValue, &tableLog, istart, cSrcSize, bmi2);
266	921	if (FSE_isError(NCountLength)) return NCountLength;
267	873	if (tableLog > maxLog) return ERROR(tableLog_tooLarge);
268	838	assert(NCountLength <= cSrcSize);
269	838	ip += NCountLength;
270	838	cSrcSize -= NCountLength;
271	838	}
272
273	838	if (FSE_DECOMPRESS_WKSP_SIZE(tableLog, maxSymbolValue) > wkspSize) return ERROR(tableLog_tooLarge);
274	820	assert(sizeof(*wksp) + FSE_DTABLE_SIZE(tableLog) <= wkspSize);
275	820	workSpace = (BYTE)workSpace + sizeof(wksp) + FSE_DTABLE_SIZE(tableLog);
276	820	wkspSize -= sizeof(*wksp) + FSE_DTABLE_SIZE(tableLog);
277
278	820	CHECK_F( FSE_buildDTable_internal(dtable, wksp->ncount, maxSymbolValue, tableLog, workSpace, wkspSize) );
279
280	820	{
281	820	const void* ptr = dtable;
282	820	const FSE_DTableHeader* DTableH = (const FSE_DTableHeader*)ptr;
283	820	const U32 fastMode = DTableH->fastMode;
284
285		/* select fast mode (static) */
286	820	if (fastMode) return FSE_decompress_usingDTable_generic(dst, dstCapacity, ip, cSrcSize, dtable, 1);
287	642	return FSE_decompress_usingDTable_generic(dst, dstCapacity, ip, cSrcSize, dtable, 0);
288	820	}
289	820	}
290
291		/* Avoids the FORCE_INLINE of the _body() function. */
292		static size_t FSE_decompress_wksp_body_default(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize)
293	0	{
294	0	return FSE_decompress_wksp_body(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize, 0);
295	0	}
296
297		#if DYNAMIC_BMI2
298		BMI2_TARGET_ATTRIBUTE static size_t FSE_decompress_wksp_body_bmi2(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize)
299	921	{
300	921	return FSE_decompress_wksp_body(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize, 1);
301	921	}
302		#endif
303
304		size_t FSE_decompress_wksp_bmi2(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize, int bmi2)
305	921	{
306	921	#if DYNAMIC_BMI2
307	921	if (bmi2) {
308	921	return FSE_decompress_wksp_body_bmi2(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize);
309	921	}
310	0	#endif
311	0	(void)bmi2;
312	0	return FSE_decompress_wksp_body_default(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize);
313	921	}
314
315		#endif /* FSE_COMMONDEFS_ONLY */