renamed all HUF_decompress*X4*() functions into *X2

to underline they generate up to 2 symbols per decoding,
in preparation for a future *X3 variant.
This commit is contained in:
Yann Collet
2018-06-14 15:17:03 -04:00
parent a09af5eb6b
commit 1adf84ccb7
3 changed files with 104 additions and 104 deletions
+14 -14
View File
@@ -165,7 +165,7 @@ typedef U32 HUF_DTable;
#define HUF_DTABLE_SIZE(maxTableLog) (1 + (1<<(maxTableLog)))
#define HUF_CREATE_STATIC_DTABLEX1(DTable, maxTableLog) \
HUF_DTable DTable[HUF_DTABLE_SIZE((maxTableLog)-1)] = { ((U32)((maxTableLog)-1) * 0x01000001) }
#define HUF_CREATE_STATIC_DTABLEX4(DTable, maxTableLog) \
#define HUF_CREATE_STATIC_DTABLEX2(DTable, maxTableLog) \
HUF_DTable DTable[HUF_DTABLE_SIZE(maxTableLog)] = { ((U32)(maxTableLog) * 0x01000001) }
@@ -173,15 +173,15 @@ typedef U32 HUF_DTable;
* Advanced decompression functions
******************************************/
size_t HUF_decompress4X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< single-symbol decoder */
size_t HUF_decompress4X4 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< double-symbols decoder */
size_t HUF_decompress4X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< double-symbols decoder */
size_t HUF_decompress4X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< decodes RLE and uncompressed */
size_t HUF_decompress4X_hufOnly(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< considers RLE and uncompressed as errors */
size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /**< considers RLE and uncompressed as errors */
size_t HUF_decompress4X1_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< single-symbol decoder */
size_t HUF_decompress4X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /**< single-symbol decoder */
size_t HUF_decompress4X4_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< double-symbols decoder */
size_t HUF_decompress4X4_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /**< double-symbols decoder */
size_t HUF_decompress4X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< double-symbols decoder */
size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /**< double-symbols decoder */
/* ****************************************
@@ -252,7 +252,7 @@ U32 HUF_getNbBits(const void* symbolTable, U32 symbolValue);
/*
* HUF_decompress() does the following:
* 1. select the decompression algorithm (X1, X4) based on pre-computed heuristics
* 1. select the decompression algorithm (X1, X2) based on pre-computed heuristics
* 2. build Huffman table from save, using HUF_readDTableX?()
* 3. decode 1 or 4 segments in parallel using HUF_decompress?X?_usingDTable()
*/
@@ -260,13 +260,13 @@ U32 HUF_getNbBits(const void* symbolTable, U32 symbolValue);
/** HUF_selectDecoder() :
* Tells which decoder is likely to decode faster,
* based on a set of pre-computed metrics.
* @return : 0==HUF_decompress4X1, 1==HUF_decompress4X4 .
* @return : 0==HUF_decompress4X1, 1==HUF_decompress4X2 .
* Assumption : 0 < dstSize <= 128 KB */
U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize);
/**
* The minimum workspace size for the `workSpace` used in
* HUF_readDTableX1_wksp() and HUF_readDTableX4_wksp().
* HUF_readDTableX1_wksp() and HUF_readDTableX2_wksp().
*
* The space used depends on HUF_TABLELOG_MAX, ranging from ~1500 bytes when
* HUF_TABLE_LOG_MAX=12 to ~1850 bytes when HUF_TABLE_LOG_MAX=15.
@@ -279,12 +279,12 @@ U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize);
size_t HUF_readDTableX1 (HUF_DTable* DTable, const void* src, size_t srcSize);
size_t HUF_readDTableX1_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
size_t HUF_readDTableX4 (HUF_DTable* DTable, const void* src, size_t srcSize);
size_t HUF_readDTableX4_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
size_t HUF_readDTableX2 (HUF_DTable* DTable, const void* src, size_t srcSize);
size_t HUF_readDTableX2_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
size_t HUF_decompress4X1_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
size_t HUF_decompress4X4_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
size_t HUF_decompress4X2_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
/* ====================== */
@@ -306,18 +306,18 @@ size_t HUF_compress1X_repeat(void* dst, size_t dstSize,
HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2);
size_t HUF_decompress1X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /* single-symbol decoder */
size_t HUF_decompress1X4 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /* double-symbol decoder */
size_t HUF_decompress1X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /* double-symbol decoder */
size_t HUF_decompress1X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);
size_t HUF_decompress1X_DCtx_wksp (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);
size_t HUF_decompress1X1_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< single-symbol decoder */
size_t HUF_decompress1X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /**< single-symbol decoder */
size_t HUF_decompress1X4_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< double-symbols decoder */
size_t HUF_decompress1X4_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /**< double-symbols decoder */
size_t HUF_decompress1X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< double-symbols decoder */
size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /**< double-symbols decoder */
size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable); /**< automatic selection of sing or double symbol decoder, based on DTable */
size_t HUF_decompress1X1_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
size_t HUF_decompress1X4_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
size_t HUF_decompress1X2_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
/* BMI2 variants.
* If the CPU has BMI2 support, pass bmi2=1, otherwise pass bmi2=0.
+89 -89
View File
@@ -140,7 +140,7 @@ size_t HUF_readDTableX1(HUF_DTable* DTable, const void* src, size_t srcSize)
workSpace, sizeof(workSpace));
}
typedef struct { U16 sequence; BYTE nbBits; BYTE length; } HUF_DEltX4; /* double-symbols decoding */
typedef struct { U16 sequence; BYTE nbBits; BYTE length; } HUF_DEltX2; /* double-symbols decoding */
FORCE_INLINE_TEMPLATE BYTE
HUF_decodeSymbolX1(BIT_DStream_t* Dstream, const HUF_DEltX1* dt, const U32 dtLog)
@@ -306,7 +306,7 @@ HUF_decompress4X1_usingDTable_internal_body(
FORCE_INLINE_TEMPLATE U32
HUF_decodeSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const U32 dtLog)
HUF_decodeSymbolX2(void* op, BIT_DStream_t* DStream, const HUF_DEltX2* dt, const U32 dtLog)
{
size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */
memcpy(op, dt+val, 2);
@@ -315,7 +315,7 @@ HUF_decodeSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const
}
FORCE_INLINE_TEMPLATE U32
HUF_decodeLastSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const U32 dtLog)
HUF_decodeLastSymbolX2(void* op, BIT_DStream_t* DStream, const HUF_DEltX2* dt, const U32 dtLog)
{
size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */
memcpy(op, dt+val, 1);
@@ -330,46 +330,46 @@ HUF_decodeLastSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, c
return 1;
}
#define HUF_DECODE_SYMBOLX4_0(ptr, DStreamPtr) \
ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
#define HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) \
ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)
#define HUF_DECODE_SYMBOLX4_1(ptr, DStreamPtr) \
#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr) \
if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)
#define HUF_DECODE_SYMBOLX4_2(ptr, DStreamPtr) \
#define HUF_DECODE_SYMBOLX2_2(ptr, DStreamPtr) \
if (MEM_64bits()) \
ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)
HINT_INLINE size_t
HUF_decodeStreamX4(BYTE* p, BIT_DStream_t* bitDPtr, BYTE* const pEnd,
const HUF_DEltX4* const dt, const U32 dtLog)
HUF_decodeStreamX2(BYTE* p, BIT_DStream_t* bitDPtr, BYTE* const pEnd,
const HUF_DEltX2* const dt, const U32 dtLog)
{
BYTE* const pStart = p;
/* up to 8 symbols at a time */
while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd-(sizeof(bitDPtr->bitContainer)-1))) {
HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
HUF_DECODE_SYMBOLX4_1(p, bitDPtr);
HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
HUF_DECODE_SYMBOLX2_1(p, bitDPtr);
HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
}
/* closer to end : up to 2 symbols at a time */
while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p <= pEnd-2))
HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
while (p <= pEnd-2)
HUF_DECODE_SYMBOLX4_0(p, bitDPtr); /* no need to reload : reached the end of DStream */
HUF_DECODE_SYMBOLX2_0(p, bitDPtr); /* no need to reload : reached the end of DStream */
if (p < pEnd)
p += HUF_decodeLastSymbolX4(p, bitDPtr, dt, dtLog);
p += HUF_decodeLastSymbolX2(p, bitDPtr, dt, dtLog);
return p-pStart;
}
FORCE_INLINE_TEMPLATE size_t
HUF_decompress1X4_usingDTable_internal_body(
HUF_decompress1X2_usingDTable_internal_body(
void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
const HUF_DTable* DTable)
@@ -383,9 +383,9 @@ HUF_decompress1X4_usingDTable_internal_body(
{ BYTE* const ostart = (BYTE*) dst;
BYTE* const oend = ostart + dstSize;
const void* const dtPtr = DTable+1; /* force compiler to not use strict-aliasing */
const HUF_DEltX4* const dt = (const HUF_DEltX4*)dtPtr;
const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr;
DTableDesc const dtd = HUF_getDTableDesc(DTable);
HUF_decodeStreamX4(ostart, &bitD, oend, dt, dtd.tableLog);
HUF_decodeStreamX2(ostart, &bitD, oend, dt, dtd.tableLog);
}
/* check */
@@ -397,7 +397,7 @@ HUF_decompress1X4_usingDTable_internal_body(
FORCE_INLINE_TEMPLATE size_t
HUF_decompress4X4_usingDTable_internal_body(
HUF_decompress4X2_usingDTable_internal_body(
void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
const HUF_DTable* DTable)
@@ -408,7 +408,7 @@ HUF_decompress4X4_usingDTable_internal_body(
BYTE* const ostart = (BYTE*) dst;
BYTE* const oend = ostart + dstSize;
const void* const dtPtr = DTable+1;
const HUF_DEltX4* const dt = (const HUF_DEltX4*)dtPtr;
const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr;
/* Init */
BIT_DStream_t bitD1;
@@ -444,22 +444,22 @@ HUF_decompress4X4_usingDTable_internal_body(
/* 16-32 symbols per loop (4-8 symbols per stream) */
endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
for ( ; (endSignal==BIT_DStream_unfinished) & (op4<(oend-(sizeof(bitD4.bitContainer)-1))) ; ) {
HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
HUF_DECODE_SYMBOLX4_1(op1, &bitD1);
HUF_DECODE_SYMBOLX4_1(op2, &bitD2);
HUF_DECODE_SYMBOLX4_1(op3, &bitD3);
HUF_DECODE_SYMBOLX4_1(op4, &bitD4);
HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
HUF_DECODE_SYMBOLX4_0(op1, &bitD1);
HUF_DECODE_SYMBOLX4_0(op2, &bitD2);
HUF_DECODE_SYMBOLX4_0(op3, &bitD3);
HUF_DECODE_SYMBOLX4_0(op4, &bitD4);
HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
HUF_DECODE_SYMBOLX2_1(op1, &bitD1);
HUF_DECODE_SYMBOLX2_1(op2, &bitD2);
HUF_DECODE_SYMBOLX2_1(op3, &bitD3);
HUF_DECODE_SYMBOLX2_1(op4, &bitD4);
HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
HUF_DECODE_SYMBOLX2_0(op1, &bitD1);
HUF_DECODE_SYMBOLX2_0(op2, &bitD2);
HUF_DECODE_SYMBOLX2_0(op3, &bitD3);
HUF_DECODE_SYMBOLX2_0(op4, &bitD4);
endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
}
@@ -471,10 +471,10 @@ HUF_decompress4X4_usingDTable_internal_body(
/* note : op4 already verified within main loop */
/* finish bitStreams one by one */
HUF_decodeStreamX4(op1, &bitD1, opStart2, dt, dtLog);
HUF_decodeStreamX4(op2, &bitD2, opStart3, dt, dtLog);
HUF_decodeStreamX4(op3, &bitD3, opStart4, dt, dtLog);
HUF_decodeStreamX4(op4, &bitD4, oend, dt, dtLog);
HUF_decodeStreamX2(op1, &bitD1, opStart2, dt, dtLog);
HUF_decodeStreamX2(op2, &bitD2, opStart3, dt, dtLog);
HUF_decodeStreamX2(op3, &bitD3, opStart4, dt, dtLog);
HUF_decodeStreamX2(op4, &bitD4, oend, dt, dtLog);
/* check */
{ U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
@@ -533,8 +533,8 @@ typedef size_t (*HUF_decompress_usingDTable_t)(void *dst, size_t dstSize,
X(HUF_decompress1X1_usingDTable_internal)
X(HUF_decompress4X1_usingDTable_internal)
X(HUF_decompress1X4_usingDTable_internal)
X(HUF_decompress4X4_usingDTable_internal)
X(HUF_decompress1X2_usingDTable_internal)
X(HUF_decompress4X2_usingDTable_internal)
#undef X
@@ -629,14 +629,14 @@ size_t HUF_decompress4X1 (void* dst, size_t dstSize, const void* cSrc, size_t cS
/* *************************/
typedef struct { BYTE symbol; BYTE weight; } sortedSymbol_t;
/* HUF_fillDTableX4Level2() :
/* HUF_fillDTableX2Level2() :
* `rankValOrigin` must be a table of at least (HUF_TABLELOG_MAX + 1) U32 */
static void HUF_fillDTableX4Level2(HUF_DEltX4* DTable, U32 sizeLog, const U32 consumed,
static void HUF_fillDTableX2Level2(HUF_DEltX2* DTable, U32 sizeLog, const U32 consumed,
const U32* rankValOrigin, const int minWeight,
const sortedSymbol_t* sortedSymbols, const U32 sortedListSize,
U32 nbBitsBaseline, U16 baseSeq)
{
HUF_DEltX4 DElt;
HUF_DEltX2 DElt;
U32 rankVal[HUF_TABLELOG_MAX + 1];
/* get pre-calculated rankVal */
@@ -674,7 +674,7 @@ static void HUF_fillDTableX4Level2(HUF_DEltX4* DTable, U32 sizeLog, const U32 co
typedef U32 rankValCol_t[HUF_TABLELOG_MAX + 1];
typedef rankValCol_t rankVal_t[HUF_TABLELOG_MAX];
static void HUF_fillDTableX4(HUF_DEltX4* DTable, const U32 targetLog,
static void HUF_fillDTableX2(HUF_DEltX2* DTable, const U32 targetLog,
const sortedSymbol_t* sortedList, const U32 sortedListSize,
const U32* rankStart, rankVal_t rankValOrigin, const U32 maxWeight,
const U32 nbBitsBaseline)
@@ -699,12 +699,12 @@ static void HUF_fillDTableX4(HUF_DEltX4* DTable, const U32 targetLog,
int minWeight = nbBits + scaleLog;
if (minWeight < 1) minWeight = 1;
sortedRank = rankStart[minWeight];
HUF_fillDTableX4Level2(DTable+start, targetLog-nbBits, nbBits,
HUF_fillDTableX2Level2(DTable+start, targetLog-nbBits, nbBits,
rankValOrigin[nbBits], minWeight,
sortedList+sortedRank, sortedListSize-sortedRank,
nbBitsBaseline, symbol);
} else {
HUF_DEltX4 DElt;
HUF_DEltX2 DElt;
MEM_writeLE16(&(DElt.sequence), symbol);
DElt.nbBits = (BYTE)(nbBits);
DElt.length = 1;
@@ -716,7 +716,7 @@ static void HUF_fillDTableX4(HUF_DEltX4* DTable, const U32 targetLog,
}
}
size_t HUF_readDTableX4_wksp(HUF_DTable* DTable, const void* src,
size_t HUF_readDTableX2_wksp(HUF_DTable* DTable, const void* src,
size_t srcSize, void* workSpace,
size_t wkspSize)
{
@@ -725,7 +725,7 @@ size_t HUF_readDTableX4_wksp(HUF_DTable* DTable, const void* src,
U32 const maxTableLog = dtd.maxTableLog;
size_t iSize;
void* dtPtr = DTable+1; /* force compiler to avoid strict-aliasing */
HUF_DEltX4* const dt = (HUF_DEltX4*)dtPtr;
HUF_DEltX2* const dt = (HUF_DEltX2*)dtPtr;
U32 *rankStart;
rankValCol_t* rankVal;
@@ -751,7 +751,7 @@ size_t HUF_readDTableX4_wksp(HUF_DTable* DTable, const void* src,
rankStart = rankStart0 + 1;
memset(rankStats, 0, sizeof(U32) * (2 * HUF_TABLELOG_MAX + 2 + 1));
DEBUG_STATIC_ASSERT(sizeof(HUF_DEltX4) == sizeof(HUF_DTable)); /* if compiler fails here, assertion is wrong */
DEBUG_STATIC_ASSERT(sizeof(HUF_DEltX2) == sizeof(HUF_DTable)); /* if compiler fails here, assertion is wrong */
if (maxTableLog > HUF_TABLELOG_MAX) return ERROR(tableLog_tooLarge);
/* memset(weightList, 0, sizeof(weightList)); */ /* is not necessary, even though some analyzer complain ... */
@@ -805,7 +805,7 @@ size_t HUF_readDTableX4_wksp(HUF_DTable* DTable, const void* src,
rankValPtr[w] = rankVal0[w] >> consumed;
} } } }
HUF_fillDTableX4(dt, maxTableLog,
HUF_fillDTableX2(dt, maxTableLog,
sortedSymbol, sizeOfSort,
rankStart0, rankVal, maxW,
tableLog+1);
@@ -816,98 +816,98 @@ size_t HUF_readDTableX4_wksp(HUF_DTable* DTable, const void* src,
return iSize;
}
size_t HUF_readDTableX4(HUF_DTable* DTable, const void* src, size_t srcSize)
size_t HUF_readDTableX2(HUF_DTable* DTable, const void* src, size_t srcSize)
{
U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];
return HUF_readDTableX4_wksp(DTable, src, srcSize,
return HUF_readDTableX2_wksp(DTable, src, srcSize,
workSpace, sizeof(workSpace));
}
size_t HUF_decompress1X4_usingDTable(
size_t HUF_decompress1X2_usingDTable(
void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
const HUF_DTable* DTable)
{
DTableDesc dtd = HUF_getDTableDesc(DTable);
if (dtd.tableType != 1) return ERROR(GENERIC);
return HUF_decompress1X4_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
return HUF_decompress1X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
}
size_t HUF_decompress1X4_DCtx_wksp(HUF_DTable* DCtx, void* dst, size_t dstSize,
size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable* DCtx, void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
void* workSpace, size_t wkspSize)
{
const BYTE* ip = (const BYTE*) cSrc;
size_t const hSize = HUF_readDTableX4_wksp(DCtx, cSrc, cSrcSize,
size_t const hSize = HUF_readDTableX2_wksp(DCtx, cSrc, cSrcSize,
workSpace, wkspSize);
if (HUF_isError(hSize)) return hSize;
if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
ip += hSize; cSrcSize -= hSize;
return HUF_decompress1X4_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx, /* bmi2 */ 0);
return HUF_decompress1X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx, /* bmi2 */ 0);
}
size_t HUF_decompress1X4_DCtx(HUF_DTable* DCtx, void* dst, size_t dstSize,
size_t HUF_decompress1X2_DCtx(HUF_DTable* DCtx, void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize)
{
U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];
return HUF_decompress1X4_DCtx_wksp(DCtx, dst, dstSize, cSrc, cSrcSize,
return HUF_decompress1X2_DCtx_wksp(DCtx, dst, dstSize, cSrc, cSrcSize,
workSpace, sizeof(workSpace));
}
size_t HUF_decompress1X4 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
size_t HUF_decompress1X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
{
HUF_CREATE_STATIC_DTABLEX4(DTable, HUF_TABLELOG_MAX);
return HUF_decompress1X4_DCtx(DTable, dst, dstSize, cSrc, cSrcSize);
HUF_CREATE_STATIC_DTABLEX2(DTable, HUF_TABLELOG_MAX);
return HUF_decompress1X2_DCtx(DTable, dst, dstSize, cSrc, cSrcSize);
}
size_t HUF_decompress4X4_usingDTable(
size_t HUF_decompress4X2_usingDTable(
void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
const HUF_DTable* DTable)
{
DTableDesc dtd = HUF_getDTableDesc(DTable);
if (dtd.tableType != 1) return ERROR(GENERIC);
return HUF_decompress4X4_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
return HUF_decompress4X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
}
static size_t HUF_decompress4X4_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize,
static size_t HUF_decompress4X2_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
void* workSpace, size_t wkspSize, int bmi2)
{
const BYTE* ip = (const BYTE*) cSrc;
size_t hSize = HUF_readDTableX4_wksp(dctx, cSrc, cSrcSize,
size_t hSize = HUF_readDTableX2_wksp(dctx, cSrc, cSrcSize,
workSpace, wkspSize);
if (HUF_isError(hSize)) return hSize;
if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
ip += hSize; cSrcSize -= hSize;
return HUF_decompress4X4_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
return HUF_decompress4X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
}
size_t HUF_decompress4X4_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
void* workSpace, size_t wkspSize)
{
return HUF_decompress4X4_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, /* bmi2 */ 0);
return HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, /* bmi2 */ 0);
}
size_t HUF_decompress4X4_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize,
size_t HUF_decompress4X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize)
{
U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];
return HUF_decompress4X4_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize,
return HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize,
workSpace, sizeof(workSpace));
}
size_t HUF_decompress4X4 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
size_t HUF_decompress4X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
{
HUF_CREATE_STATIC_DTABLEX4(DTable, HUF_TABLELOG_MAX);
return HUF_decompress4X4_DCtx(DTable, dst, dstSize, cSrc, cSrcSize);
HUF_CREATE_STATIC_DTABLEX2(DTable, HUF_TABLELOG_MAX);
return HUF_decompress4X2_DCtx(DTable, dst, dstSize, cSrc, cSrcSize);
}
@@ -920,7 +920,7 @@ size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize,
const HUF_DTable* DTable)
{
DTableDesc const dtd = HUF_getDTableDesc(DTable);
return dtd.tableType ? HUF_decompress1X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) :
return dtd.tableType ? HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) :
HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
}
@@ -929,7 +929,7 @@ size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize,
const HUF_DTable* DTable)
{
DTableDesc const dtd = HUF_getDTableDesc(DTable);
return dtd.tableType ? HUF_decompress4X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) :
return dtd.tableType ? HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) :
HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
}
@@ -959,7 +959,7 @@ static const algo_time_t algoTime[16 /* Quantization */][3 /* single, double, qu
/** HUF_selectDecoder() :
* Tells which decoder is likely to decode faster,
* based on a set of pre-computed metrics.
* @return : 0==HUF_decompress4X1, 1==HUF_decompress4X4 .
* @return : 0==HUF_decompress4X1, 1==HUF_decompress4X2 .
* Assumption : 0 < dstSize <= 128 KB */
U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize)
{
@@ -979,7 +979,7 @@ typedef size_t (*decompressionAlgo)(void* dst, size_t dstSize, const void* cSrc,
size_t HUF_decompress (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
{
static const decompressionAlgo decompress[2] = { HUF_decompress4X1, HUF_decompress4X4 };
static const decompressionAlgo decompress[2] = { HUF_decompress4X1, HUF_decompress4X2 };
/* validation checks */
if (dstSize == 0) return ERROR(dstSize_tooSmall);
@@ -1001,7 +1001,7 @@ size_t HUF_decompress4X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const
if (cSrcSize == 1) { memset(dst, *(const BYTE*)cSrc, dstSize); return dstSize; } /* RLE */
{ U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
return algoNb ? HUF_decompress4X4_DCtx(dctx, dst, dstSize, cSrc, cSrcSize) :
return algoNb ? HUF_decompress4X2_DCtx(dctx, dst, dstSize, cSrc, cSrcSize) :
HUF_decompress4X1_DCtx(dctx, dst, dstSize, cSrc, cSrcSize) ;
}
}
@@ -1024,7 +1024,7 @@ size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable* dctx, void* dst,
if (cSrcSize == 0) return ERROR(corruption_detected);
{ U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
return algoNb ? HUF_decompress4X4_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize):
return algoNb ? HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize):
HUF_decompress4X1_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize);
}
}
@@ -1040,7 +1040,7 @@ size_t HUF_decompress1X_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
if (cSrcSize == 1) { memset(dst, *(const BYTE*)cSrc, dstSize); return dstSize; } /* RLE */
{ U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
return algoNb ? HUF_decompress1X4_DCtx_wksp(dctx, dst, dstSize, cSrc,
return algoNb ? HUF_decompress1X2_DCtx_wksp(dctx, dst, dstSize, cSrc,
cSrcSize, workSpace, wkspSize):
HUF_decompress1X1_DCtx_wksp(dctx, dst, dstSize, cSrc,
cSrcSize, workSpace, wkspSize);
@@ -1059,7 +1059,7 @@ size_t HUF_decompress1X_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize,
size_t HUF_decompress1X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2)
{
DTableDesc const dtd = HUF_getDTableDesc(DTable);
return dtd.tableType ? HUF_decompress1X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2) :
return dtd.tableType ? HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2) :
HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
}
@@ -1078,7 +1078,7 @@ size_t HUF_decompress1X1_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstS
size_t HUF_decompress4X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2)
{
DTableDesc const dtd = HUF_getDTableDesc(DTable);
return dtd.tableType ? HUF_decompress4X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2) :
return dtd.tableType ? HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2) :
HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
}
@@ -1089,7 +1089,7 @@ size_t HUF_decompress4X_hufOnly_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t ds
if (cSrcSize == 0) return ERROR(corruption_detected);
{ U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
return algoNb ? HUF_decompress4X4_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2) :
return algoNb ? HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2) :
HUF_decompress4X1_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
}
}
+1 -1
View File
@@ -2193,7 +2193,7 @@ static size_t ZSTD_loadEntropy(ZSTD_entropyDTables_t* entropy, const void* const
dictPtr += 8; /* skip header = magic + dictID */
{ size_t const hSize = HUF_readDTableX4_wksp(
{ size_t const hSize = HUF_readDTableX2_wksp(
entropy->hufTable, dictPtr, dictEnd - dictPtr,
entropy->workspace, sizeof(entropy->workspace));
if (HUF_isError(hSize)) return ERROR(dictionary_corrupted);