diff options
Diffstat (limited to 'lib/decompress')
| -rw-r--r-- | lib/decompress/huf_decompress.c | 654 | ||||
| -rw-r--r-- | lib/decompress/zstd_decompress.c | 385 |
2 files changed, 572 insertions, 467 deletions
diff --git a/lib/decompress/huf_decompress.c b/lib/decompress/huf_decompress.c index 73f5c46c06152..83ecaff01e848 100644 --- a/lib/decompress/huf_decompress.c +++ b/lib/decompress/huf_decompress.c @@ -1,6 +1,7 @@ /* ****************************************************************** - Huffman decoder, part of New Generation Entropy library - Copyright (C) 2013-2016, Yann Collet. + huff0 huffman decoder, + part of Finite State Entropy library + Copyright (C) 2013-present, Yann Collet. BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) @@ -29,16 +30,15 @@ You can contact the author at : - FSE+HUF source repository : https://github.com/Cyan4973/FiniteStateEntropy - - Public forum : https://groups.google.com/forum/#!forum/lz4c ****************************************************************** */ /* ************************************************************** * Dependencies ****************************************************************/ #include <string.h> /* memcpy, memset */ -#include "bitstream.h" /* BIT_* */ #include "compiler.h" -#include "fse.h" /* header compression */ +#include "bitstream.h" /* BIT_* */ +#include "fse.h" /* to compress headers */ #define HUF_STATIC_LINKING_ONLY #include "huf.h" #include "error_private.h" @@ -48,7 +48,6 @@ * Error Management ****************************************************************/ #define HUF_isError ERR_isError -#define HUF_STATIC_ASSERT(c) { enum { HUF_static_assert = 1/(int)(!!(c)) }; } /* use only *after* variable declarations */ #define CHECK_F(f) { size_t const err_ = (f); if (HUF_isError(err_)) return err_; } @@ -75,15 +74,15 @@ static DTableDesc HUF_getDTableDesc(const HUF_DTable* table) /*-***************************/ /* single-symbol decoding */ /*-***************************/ -typedef struct { BYTE byte; BYTE nbBits; } HUF_DEltX2; /* single-symbol decoding */ +typedef struct { BYTE byte; BYTE nbBits; } HUF_DEltX1; /* single-symbol decoding */ -size_t HUF_readDTableX2_wksp(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize) +size_t HUF_readDTableX1_wksp(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize) { U32 tableLog = 0; U32 nbSymbols = 0; size_t iSize; void* const dtPtr = DTable + 1; - HUF_DEltX2* const dt = (HUF_DEltX2*)dtPtr; + HUF_DEltX1* const dt = (HUF_DEltX1*)dtPtr; U32* rankVal; BYTE* huffWeight; @@ -96,7 +95,7 @@ size_t HUF_readDTableX2_wksp(HUF_DTable* DTable, const void* src, size_t srcSize if ((spaceUsed32 << 2) > wkspSize) return ERROR(tableLog_tooLarge); - HUF_STATIC_ASSERT(sizeof(DTableDesc) == sizeof(HUF_DTable)); + DEBUG_STATIC_ASSERT(sizeof(DTableDesc) == sizeof(HUF_DTable)); /* memset(huffWeight, 0, sizeof(huffWeight)); */ /* is not necessary, even though some analyzer complain ... */ iSize = HUF_readStats(huffWeight, HUF_SYMBOLVALUE_MAX + 1, rankVal, &nbSymbols, &tableLog, src, srcSize); @@ -124,7 +123,7 @@ size_t HUF_readDTableX2_wksp(HUF_DTable* DTable, const void* src, size_t srcSize U32 const w = huffWeight[n]; U32 const length = (1 << w) >> 1; U32 u; - HUF_DEltX2 D; + HUF_DEltX1 D; D.byte = (BYTE)n; D.nbBits = (BYTE)(tableLog + 1 - w); for (u = rankVal[w]; u < rankVal[w] + length; u++) dt[u] = D; @@ -134,17 +133,15 @@ size_t HUF_readDTableX2_wksp(HUF_DTable* DTable, const void* src, size_t srcSize return iSize; } -size_t HUF_readDTableX2(HUF_DTable* DTable, const void* src, size_t srcSize) +size_t HUF_readDTableX1(HUF_DTable* DTable, const void* src, size_t srcSize) { U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32]; - return HUF_readDTableX2_wksp(DTable, src, srcSize, + return HUF_readDTableX1_wksp(DTable, src, srcSize, workSpace, sizeof(workSpace)); } -typedef struct { U16 sequence; BYTE nbBits; BYTE length; } HUF_DEltX4; /* double-symbols decoding */ - FORCE_INLINE_TEMPLATE BYTE -HUF_decodeSymbolX2(BIT_DStream_t* Dstream, const HUF_DEltX2* dt, const U32 dtLog) +HUF_decodeSymbolX1(BIT_DStream_t* Dstream, const HUF_DEltX1* dt, const U32 dtLog) { size_t const val = BIT_lookBitsFast(Dstream, dtLog); /* note : dtLog >= 1 */ BYTE const c = dt[val].byte; @@ -152,44 +149,44 @@ HUF_decodeSymbolX2(BIT_DStream_t* Dstream, const HUF_DEltX2* dt, const U32 dtLog return c; } -#define HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) \ - *ptr++ = HUF_decodeSymbolX2(DStreamPtr, dt, dtLog) +#define HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr) \ + *ptr++ = HUF_decodeSymbolX1(DStreamPtr, dt, dtLog) -#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr) \ +#define HUF_DECODE_SYMBOLX1_1(ptr, DStreamPtr) \ if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \ - HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) + HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr) -#define HUF_DECODE_SYMBOLX2_2(ptr, DStreamPtr) \ +#define HUF_DECODE_SYMBOLX1_2(ptr, DStreamPtr) \ if (MEM_64bits()) \ - HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) + HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr) HINT_INLINE size_t -HUF_decodeStreamX2(BYTE* p, BIT_DStream_t* const bitDPtr, BYTE* const pEnd, const HUF_DEltX2* const dt, const U32 dtLog) +HUF_decodeStreamX1(BYTE* p, BIT_DStream_t* const bitDPtr, BYTE* const pEnd, const HUF_DEltX1* const dt, const U32 dtLog) { BYTE* const pStart = p; /* up to 4 symbols at a time */ while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd-3)) { - HUF_DECODE_SYMBOLX2_2(p, bitDPtr); - HUF_DECODE_SYMBOLX2_1(p, bitDPtr); - HUF_DECODE_SYMBOLX2_2(p, bitDPtr); - HUF_DECODE_SYMBOLX2_0(p, bitDPtr); + HUF_DECODE_SYMBOLX1_2(p, bitDPtr); + HUF_DECODE_SYMBOLX1_1(p, bitDPtr); + HUF_DECODE_SYMBOLX1_2(p, bitDPtr); + HUF_DECODE_SYMBOLX1_0(p, bitDPtr); } /* [0-3] symbols remaining */ if (MEM_32bits()) while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd)) - HUF_DECODE_SYMBOLX2_0(p, bitDPtr); + HUF_DECODE_SYMBOLX1_0(p, bitDPtr); /* no more data to retrieve from bitstream, no need to reload */ while (p < pEnd) - HUF_DECODE_SYMBOLX2_0(p, bitDPtr); + HUF_DECODE_SYMBOLX1_0(p, bitDPtr); return pEnd-pStart; } FORCE_INLINE_TEMPLATE size_t -HUF_decompress1X2_usingDTable_internal_body( +HUF_decompress1X1_usingDTable_internal_body( void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable) @@ -197,14 +194,14 @@ HUF_decompress1X2_usingDTable_internal_body( BYTE* op = (BYTE*)dst; BYTE* const oend = op + dstSize; const void* dtPtr = DTable + 1; - const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr; + const HUF_DEltX1* const dt = (const HUF_DEltX1*)dtPtr; BIT_DStream_t bitD; DTableDesc const dtd = HUF_getDTableDesc(DTable); U32 const dtLog = dtd.tableLog; CHECK_F( BIT_initDStream(&bitD, cSrc, cSrcSize) ); - HUF_decodeStreamX2(op, &bitD, oend, dt, dtLog); + HUF_decodeStreamX1(op, &bitD, oend, dt, dtLog); if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected); @@ -212,7 +209,7 @@ HUF_decompress1X2_usingDTable_internal_body( } FORCE_INLINE_TEMPLATE size_t -HUF_decompress4X2_usingDTable_internal_body( +HUF_decompress4X1_usingDTable_internal_body( void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable) @@ -224,7 +221,7 @@ HUF_decompress4X2_usingDTable_internal_body( BYTE* const ostart = (BYTE*) dst; BYTE* const oend = ostart + dstSize; const void* const dtPtr = DTable + 1; - const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr; + const HUF_DEltX1* const dt = (const HUF_DEltX1*)dtPtr; /* Init */ BIT_DStream_t bitD1; @@ -260,22 +257,22 @@ HUF_decompress4X2_usingDTable_internal_body( /* up to 16 symbols per loop (4 symbols per stream) in 64-bit mode */ endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4); while ( (endSignal==BIT_DStream_unfinished) && (op4<(oend-3)) ) { - 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); + HUF_DECODE_SYMBOLX1_2(op1, &bitD1); + HUF_DECODE_SYMBOLX1_2(op2, &bitD2); + HUF_DECODE_SYMBOLX1_2(op3, &bitD3); + HUF_DECODE_SYMBOLX1_2(op4, &bitD4); + HUF_DECODE_SYMBOLX1_1(op1, &bitD1); + HUF_DECODE_SYMBOLX1_1(op2, &bitD2); + HUF_DECODE_SYMBOLX1_1(op3, &bitD3); + HUF_DECODE_SYMBOLX1_1(op4, &bitD4); + HUF_DECODE_SYMBOLX1_2(op1, &bitD1); + HUF_DECODE_SYMBOLX1_2(op2, &bitD2); + HUF_DECODE_SYMBOLX1_2(op3, &bitD3); + HUF_DECODE_SYMBOLX1_2(op4, &bitD4); + HUF_DECODE_SYMBOLX1_0(op1, &bitD1); + HUF_DECODE_SYMBOLX1_0(op2, &bitD2); + HUF_DECODE_SYMBOLX1_0(op3, &bitD3); + HUF_DECODE_SYMBOLX1_0(op4, &bitD4); BIT_reloadDStream(&bitD1); BIT_reloadDStream(&bitD2); BIT_reloadDStream(&bitD3); @@ -291,191 +288,10 @@ HUF_decompress4X2_usingDTable_internal_body( /* note : op4 supposed already verified within main loop */ /* finish bitStreams one by one */ - 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); - if (!endCheck) return ERROR(corruption_detected); } - - /* decoded size */ - return dstSize; - } -} - - -FORCE_INLINE_TEMPLATE U32 -HUF_decodeSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const U32 dtLog) -{ - size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */ - memcpy(op, dt+val, 2); - BIT_skipBits(DStream, dt[val].nbBits); - return dt[val].length; -} - -FORCE_INLINE_TEMPLATE U32 -HUF_decodeLastSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const U32 dtLog) -{ - size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */ - memcpy(op, dt+val, 1); - if (dt[val].length==1) BIT_skipBits(DStream, dt[val].nbBits); - else { - if (DStream->bitsConsumed < (sizeof(DStream->bitContainer)*8)) { - BIT_skipBits(DStream, dt[val].nbBits); - if (DStream->bitsConsumed > (sizeof(DStream->bitContainer)*8)) - /* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */ - DStream->bitsConsumed = (sizeof(DStream->bitContainer)*8); - } } - return 1; -} - -#define HUF_DECODE_SYMBOLX4_0(ptr, DStreamPtr) \ - ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog) - -#define HUF_DECODE_SYMBOLX4_1(ptr, DStreamPtr) \ - if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \ - ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog) - -#define HUF_DECODE_SYMBOLX4_2(ptr, DStreamPtr) \ - if (MEM_64bits()) \ - ptr += HUF_decodeSymbolX4(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) -{ - 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); - } - - /* 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); - - while (p <= pEnd-2) - HUF_DECODE_SYMBOLX4_0(p, bitDPtr); /* no need to reload : reached the end of DStream */ - - if (p < pEnd) - p += HUF_decodeLastSymbolX4(p, bitDPtr, dt, dtLog); - - return p-pStart; -} - -FORCE_INLINE_TEMPLATE size_t -HUF_decompress1X4_usingDTable_internal_body( - void* dst, size_t dstSize, - const void* cSrc, size_t cSrcSize, - const HUF_DTable* DTable) -{ - BIT_DStream_t bitD; - - /* Init */ - CHECK_F( BIT_initDStream(&bitD, cSrc, cSrcSize) ); - - /* decode */ - { 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; - DTableDesc const dtd = HUF_getDTableDesc(DTable); - HUF_decodeStreamX4(ostart, &bitD, oend, dt, dtd.tableLog); - } - - /* check */ - if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected); - - /* decoded size */ - return dstSize; -} - - -FORCE_INLINE_TEMPLATE size_t -HUF_decompress4X4_usingDTable_internal_body( - void* dst, size_t dstSize, - const void* cSrc, size_t cSrcSize, - const HUF_DTable* DTable) -{ - if (cSrcSize < 10) return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */ - - { const BYTE* const istart = (const BYTE*) cSrc; - 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; - - /* Init */ - BIT_DStream_t bitD1; - BIT_DStream_t bitD2; - BIT_DStream_t bitD3; - BIT_DStream_t bitD4; - size_t const length1 = MEM_readLE16(istart); - size_t const length2 = MEM_readLE16(istart+2); - size_t const length3 = MEM_readLE16(istart+4); - size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6); - const BYTE* const istart1 = istart + 6; /* jumpTable */ - const BYTE* const istart2 = istart1 + length1; - const BYTE* const istart3 = istart2 + length2; - const BYTE* const istart4 = istart3 + length3; - size_t const segmentSize = (dstSize+3) / 4; - BYTE* const opStart2 = ostart + segmentSize; - BYTE* const opStart3 = opStart2 + segmentSize; - BYTE* const opStart4 = opStart3 + segmentSize; - BYTE* op1 = ostart; - BYTE* op2 = opStart2; - BYTE* op3 = opStart3; - BYTE* op4 = opStart4; - U32 endSignal; - DTableDesc const dtd = HUF_getDTableDesc(DTable); - U32 const dtLog = dtd.tableLog; - - if (length4 > cSrcSize) return ERROR(corruption_detected); /* overflow */ - CHECK_F( BIT_initDStream(&bitD1, istart1, length1) ); - CHECK_F( BIT_initDStream(&bitD2, istart2, length2) ); - CHECK_F( BIT_initDStream(&bitD3, istart3, length3) ); - CHECK_F( BIT_initDStream(&bitD4, istart4, length4) ); - - /* 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); - - endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4); - } - - /* check corruption */ - if (op1 > opStart2) return ERROR(corruption_detected); - if (op2 > opStart3) return ERROR(corruption_detected); - if (op3 > opStart4) return ERROR(corruption_detected); - /* 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_decodeStreamX1(op1, &bitD1, opStart2, dt, dtLog); + HUF_decodeStreamX1(op2, &bitD2, opStart3, dt, dtLog); + HUF_decodeStreamX1(op3, &bitD3, opStart4, dt, dtLog); + HUF_decodeStreamX1(op4, &bitD4, oend, dt, dtLog); /* check */ { U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4); @@ -493,7 +309,7 @@ typedef size_t (*HUF_decompress_usingDTable_t)(void *dst, size_t dstSize, const HUF_DTable *DTable); #if DYNAMIC_BMI2 -#define X(fn) \ +#define HUF_DGEN(fn) \ \ static size_t fn##_default( \ void* dst, size_t dstSize, \ @@ -522,7 +338,7 @@ typedef size_t (*HUF_decompress_usingDTable_t)(void *dst, size_t dstSize, #else -#define X(fn) \ +#define HUF_DGEN(fn) \ static size_t fn(void* dst, size_t dstSize, void const* cSrc, \ size_t cSrcSize, HUF_DTable const* DTable, int bmi2) \ { \ @@ -532,112 +348,114 @@ typedef size_t (*HUF_decompress_usingDTable_t)(void *dst, size_t dstSize, #endif -X(HUF_decompress1X2_usingDTable_internal) -X(HUF_decompress4X2_usingDTable_internal) -X(HUF_decompress1X4_usingDTable_internal) -X(HUF_decompress4X4_usingDTable_internal) +HUF_DGEN(HUF_decompress1X1_usingDTable_internal) +HUF_DGEN(HUF_decompress4X1_usingDTable_internal) -#undef X -size_t HUF_decompress1X2_usingDTable( +size_t HUF_decompress1X1_usingDTable( void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable) { DTableDesc dtd = HUF_getDTableDesc(DTable); if (dtd.tableType != 0) return ERROR(GENERIC); - return HUF_decompress1X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0); + return HUF_decompress1X1_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0); } -size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable* DCtx, void* dst, size_t dstSize, +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) { const BYTE* ip = (const BYTE*) cSrc; - size_t const hSize = HUF_readDTableX2_wksp(DCtx, cSrc, cSrcSize, workSpace, wkspSize); + size_t const hSize = HUF_readDTableX1_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_decompress1X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx, /* bmi2 */ 0); + return HUF_decompress1X1_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx, /* bmi2 */ 0); } -size_t HUF_decompress1X2_DCtx(HUF_DTable* DCtx, void* dst, size_t dstSize, +size_t HUF_decompress1X1_DCtx(HUF_DTable* DCtx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize) { U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32]; - return HUF_decompress1X2_DCtx_wksp(DCtx, dst, dstSize, cSrc, cSrcSize, + return HUF_decompress1X1_DCtx_wksp(DCtx, dst, dstSize, cSrc, cSrcSize, workSpace, sizeof(workSpace)); } -size_t HUF_decompress1X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize) +size_t HUF_decompress1X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize) { - HUF_CREATE_STATIC_DTABLEX2(DTable, HUF_TABLELOG_MAX); - return HUF_decompress1X2_DCtx (DTable, dst, dstSize, cSrc, cSrcSize); + HUF_CREATE_STATIC_DTABLEX1(DTable, HUF_TABLELOG_MAX); + return HUF_decompress1X1_DCtx (DTable, dst, dstSize, cSrc, cSrcSize); } -size_t HUF_decompress4X2_usingDTable( +size_t HUF_decompress4X1_usingDTable( void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable) { DTableDesc dtd = HUF_getDTableDesc(DTable); if (dtd.tableType != 0) return ERROR(GENERIC); - return HUF_decompress4X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0); + return HUF_decompress4X1_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0); } -static size_t HUF_decompress4X2_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, +static size_t HUF_decompress4X1_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 const hSize = HUF_readDTableX2_wksp (dctx, cSrc, cSrcSize, + size_t const hSize = HUF_readDTableX1_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_decompress4X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2); + return HUF_decompress4X1_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2); } -size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, +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) { - return HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, 0); + return HUF_decompress4X1_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, 0); } -size_t HUF_decompress4X2_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize) +size_t HUF_decompress4X1_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize) { U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32]; - return HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, + return HUF_decompress4X1_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, sizeof(workSpace)); } -size_t HUF_decompress4X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize) +size_t HUF_decompress4X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize) { - HUF_CREATE_STATIC_DTABLEX2(DTable, HUF_TABLELOG_MAX); - return HUF_decompress4X2_DCtx(DTable, dst, dstSize, cSrc, cSrcSize); + HUF_CREATE_STATIC_DTABLEX1(DTable, HUF_TABLELOG_MAX); + return HUF_decompress4X1_DCtx(DTable, dst, dstSize, cSrc, cSrcSize); } /* *************************/ /* double-symbols decoding */ /* *************************/ + +typedef struct { U16 sequence; BYTE nbBits; BYTE length; } HUF_DEltX2; /* double-symbols decoding */ typedef struct { BYTE symbol; BYTE weight; } sortedSymbol_t; +typedef U32 rankValCol_t[HUF_TABLELOG_MAX + 1]; +typedef rankValCol_t rankVal_t[HUF_TABLELOG_MAX]; -/* 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 */ @@ -672,10 +490,8 @@ 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) @@ -700,12 +516,12 @@ 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; @@ -717,16 +533,16 @@ static void HUF_fillDTableX4(HUF_DEltX4* DTable, const U32 targetLog, } } -size_t HUF_readDTableX4_wksp(HUF_DTable* DTable, const void* src, - size_t srcSize, void* workSpace, - size_t wkspSize) +size_t HUF_readDTableX2_wksp(HUF_DTable* DTable, + const void* src, size_t srcSize, + void* workSpace, size_t wkspSize) { U32 tableLog, maxW, sizeOfSort, nbSymbols; DTableDesc dtd = HUF_getDTableDesc(DTable); 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; @@ -752,7 +568,7 @@ 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)); - HUF_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 ... */ @@ -806,7 +622,7 @@ 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); @@ -817,112 +633,296 @@ 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( + +FORCE_INLINE_TEMPLATE U32 +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); + BIT_skipBits(DStream, dt[val].nbBits); + return dt[val].length; +} + +FORCE_INLINE_TEMPLATE U32 +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); + if (dt[val].length==1) BIT_skipBits(DStream, dt[val].nbBits); + else { + if (DStream->bitsConsumed < (sizeof(DStream->bitContainer)*8)) { + BIT_skipBits(DStream, dt[val].nbBits); + if (DStream->bitsConsumed > (sizeof(DStream->bitContainer)*8)) + /* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */ + DStream->bitsConsumed = (sizeof(DStream->bitContainer)*8); + } } + return 1; +} + +#define HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) \ + ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog) + +#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr) \ + if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \ + ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog) + +#define HUF_DECODE_SYMBOLX2_2(ptr, DStreamPtr) \ + if (MEM_64bits()) \ + ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog) + +HINT_INLINE size_t +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_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_SYMBOLX2_0(p, bitDPtr); + + while (p <= pEnd-2) + HUF_DECODE_SYMBOLX2_0(p, bitDPtr); /* no need to reload : reached the end of DStream */ + + if (p < pEnd) + p += HUF_decodeLastSymbolX2(p, bitDPtr, dt, dtLog); + + return p-pStart; +} + +FORCE_INLINE_TEMPLATE size_t +HUF_decompress1X2_usingDTable_internal_body( + void* dst, size_t dstSize, + const void* cSrc, size_t cSrcSize, + const HUF_DTable* DTable) +{ + BIT_DStream_t bitD; + + /* Init */ + CHECK_F( BIT_initDStream(&bitD, cSrc, cSrcSize) ); + + /* decode */ + { 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_DEltX2* const dt = (const HUF_DEltX2*)dtPtr; + DTableDesc const dtd = HUF_getDTableDesc(DTable); + HUF_decodeStreamX2(ostart, &bitD, oend, dt, dtd.tableLog); + } + + /* check */ + if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected); + + /* decoded size */ + return dstSize; +} + + +FORCE_INLINE_TEMPLATE size_t +HUF_decompress4X2_usingDTable_internal_body( + void* dst, size_t dstSize, + const void* cSrc, size_t cSrcSize, + const HUF_DTable* DTable) +{ + if (cSrcSize < 10) return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */ + + { const BYTE* const istart = (const BYTE*) cSrc; + BYTE* const ostart = (BYTE*) dst; + BYTE* const oend = ostart + dstSize; + const void* const dtPtr = DTable+1; + const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr; + + /* Init */ + BIT_DStream_t bitD1; + BIT_DStream_t bitD2; + BIT_DStream_t bitD3; + BIT_DStream_t bitD4; + size_t const length1 = MEM_readLE16(istart); + size_t const length2 = MEM_readLE16(istart+2); + size_t const length3 = MEM_readLE16(istart+4); + size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6); + const BYTE* const istart1 = istart + 6; /* jumpTable */ + const BYTE* const istart2 = istart1 + length1; + const BYTE* const istart3 = istart2 + length2; + const BYTE* const istart4 = istart3 + length3; + size_t const segmentSize = (dstSize+3) / 4; + BYTE* const opStart2 = ostart + segmentSize; + BYTE* const opStart3 = opStart2 + segmentSize; + BYTE* const opStart4 = opStart3 + segmentSize; + BYTE* op1 = ostart; + BYTE* op2 = opStart2; + BYTE* op3 = opStart3; + BYTE* op4 = opStart4; + U32 endSignal; + DTableDesc const dtd = HUF_getDTableDesc(DTable); + U32 const dtLog = dtd.tableLog; + + if (length4 > cSrcSize) return ERROR(corruption_detected); /* overflow */ + CHECK_F( BIT_initDStream(&bitD1, istart1, length1) ); + CHECK_F( BIT_initDStream(&bitD2, istart2, length2) ); + CHECK_F( BIT_initDStream(&bitD3, istart3, length3) ); + CHECK_F( BIT_initDStream(&bitD4, istart4, length4) ); + + /* 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_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); + } + + /* check corruption */ + if (op1 > opStart2) return ERROR(corruption_detected); + if (op2 > opStart3) return ERROR(corruption_detected); + if (op3 > opStart4) return ERROR(corruption_detected); + /* note : op4 already verified within main loop */ + + /* finish bitStreams one by one */ + 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); + if (!endCheck) return ERROR(corruption_detected); } + + /* decoded size */ + return dstSize; + } +} + +HUF_DGEN(HUF_decompress1X2_usingDTable_internal) +HUF_DGEN(HUF_decompress4X2_usingDTable_internal) + +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); } -/* ********************************/ -/* Generic decompression selector */ -/* ********************************/ +/* ***********************************/ +/* Universal decompression selectors */ +/* ***********************************/ size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable) { DTableDesc const dtd = HUF_getDTableDesc(DTable); - return dtd.tableType ? HUF_decompress1X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) : - HUF_decompress1X2_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); } size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize, @@ -930,8 +930,8 @@ 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) : - HUF_decompress4X2_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); } @@ -960,12 +960,12 @@ 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_decompress4X2, 1==HUF_decompress4X4 . + * @return : 0==HUF_decompress4X1, 1==HUF_decompress4X2 . * Assumption : 0 < dstSize <= 128 KB */ U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize) { assert(dstSize > 0); - assert(dstSize <= 128 KB); + assert(dstSize <= 128*1024); /* decoder timing evaluation */ { U32 const Q = (cSrcSize >= dstSize) ? 15 : (U32)(cSrcSize * 16 / dstSize); /* Q < 16 */ U32 const D256 = (U32)(dstSize >> 8); @@ -980,7 +980,7 @@ 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_decompress4X2, HUF_decompress4X4 }; + static const decompressionAlgo decompress[2] = { HUF_decompress4X1, HUF_decompress4X2 }; /* validation checks */ if (dstSize == 0) return ERROR(dstSize_tooSmall); @@ -1002,8 +1002,8 @@ 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) : - HUF_decompress4X2_DCtx(dctx, dst, dstSize, cSrc, cSrcSize) ; + return algoNb ? HUF_decompress4X2_DCtx(dctx, dst, dstSize, cSrc, cSrcSize) : + HUF_decompress4X1_DCtx(dctx, dst, dstSize, cSrc, cSrcSize) ; } } @@ -1025,8 +1025,8 @@ 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): - HUF_decompress4X2_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); } } @@ -1041,9 +1041,9 @@ 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_decompress1X2_DCtx_wksp(dctx, dst, dstSize, cSrc, + HUF_decompress1X1_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize); } } @@ -1060,27 +1060,27 @@ 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) : - HUF_decompress1X2_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); } -size_t HUF_decompress1X2_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2) +size_t HUF_decompress1X1_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 const hSize = HUF_readDTableX2_wksp(dctx, cSrc, cSrcSize, workSpace, wkspSize); + size_t const hSize = HUF_readDTableX1_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_decompress1X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2); + return HUF_decompress1X1_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2); } 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) : - HUF_decompress4X2_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); } size_t HUF_decompress4X_hufOnly_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2) @@ -1090,7 +1090,7 @@ 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) : - HUF_decompress4X2_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); } } diff --git a/lib/decompress/zstd_decompress.c b/lib/decompress/zstd_decompress.c index 3ec6a1cb328bd..711b5b6d7acad 100644 --- a/lib/decompress/zstd_decompress.c +++ b/lib/decompress/zstd_decompress.c @@ -40,12 +40,24 @@ # define ZSTD_MAXWINDOWSIZE_DEFAULT (((U32)1 << ZSTD_WINDOWLOG_DEFAULTMAX) + 1) #endif +/*! + * NO_FORWARD_PROGRESS_MAX : + * maximum allowed nb of calls to ZSTD_decompressStream() and ZSTD_decompress_generic() + * without any forward progress + * (defined as: no byte read from input, and no byte flushed to output) + * before triggering an error. + */ +#ifndef ZSTD_NO_FORWARD_PROGRESS_MAX +# define ZSTD_NO_FORWARD_PROGRESS_MAX 16 +#endif + /*-******************************************************* * Dependencies *********************************************************/ #include <string.h> /* memcpy, memmove, memset */ -#include "cpu.h" +#include "compiler.h" /* prefetch */ +#include "cpu.h" /* bmi2 */ #include "mem.h" /* low level memory routines */ #define FSE_STATIC_LINKING_ONLY #include "fse.h" @@ -57,6 +69,9 @@ # include "zstd_legacy.h" #endif +static const void* ZSTD_DDictDictContent(const ZSTD_DDict* ddict); +static size_t ZSTD_DDictDictSize(const ZSTD_DDict* ddict); + /*-************************************* * Errors @@ -99,11 +114,10 @@ typedef struct { #define SEQSYMBOL_TABLE_SIZE(log) (1 + (1 << (log))) typedef struct { - ZSTD_seqSymbol LLTable[SEQSYMBOL_TABLE_SIZE(LLFSELog)]; - ZSTD_seqSymbol OFTable[SEQSYMBOL_TABLE_SIZE(OffFSELog)]; - ZSTD_seqSymbol MLTable[SEQSYMBOL_TABLE_SIZE(MLFSELog)]; + ZSTD_seqSymbol LLTable[SEQSYMBOL_TABLE_SIZE(LLFSELog)]; /* Note : Space reserved for FSE Tables */ + ZSTD_seqSymbol OFTable[SEQSYMBOL_TABLE_SIZE(OffFSELog)]; /* is also used as temporary workspace while building hufTable during DDict creation */ + ZSTD_seqSymbol MLTable[SEQSYMBOL_TABLE_SIZE(MLFSELog)]; /* and therefore must be at least HUF_DECOMPRESS_WORKSPACE_SIZE large */ HUF_DTable hufTable[HUF_DTABLE_SIZE(HufLog)]; /* can accommodate HUF_decompress4X */ - U32 workspace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32]; U32 rep[ZSTD_REP_NUM]; } ZSTD_entropyDTables_t; @@ -114,9 +128,10 @@ struct ZSTD_DCtx_s const ZSTD_seqSymbol* OFTptr; const HUF_DTable* HUFptr; ZSTD_entropyDTables_t entropy; + U32 workspace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32]; /* space needed when building huffman tables */ const void* previousDstEnd; /* detect continuity */ - const void* base; /* start of current segment */ - const void* vBase; /* virtual start of previous segment if it was just before current one */ + const void* prefixStart; /* start of current segment */ + const void* virtualStart; /* virtual start of previous segment if it was just before current one */ const void* dictEnd; /* end of previous segment */ size_t expected; ZSTD_frameHeader fParams; @@ -127,7 +142,6 @@ struct ZSTD_DCtx_s U32 fseEntropy; XXH64_state_t xxhState; size_t headerSize; - U32 dictID; ZSTD_format_e format; const BYTE* litPtr; ZSTD_customMem customMem; @@ -136,9 +150,13 @@ struct ZSTD_DCtx_s size_t staticSize; int bmi2; /* == 1 if the CPU supports BMI2 and 0 otherwise. CPU support is determined dynamically once per context lifetime. */ - /* streaming */ + /* dictionary */ ZSTD_DDict* ddictLocal; - const ZSTD_DDict* ddict; + const ZSTD_DDict* ddict; /* set by ZSTD_initDStream_usingDDict(), or ZSTD_DCtx_refDDict() */ + U32 dictID; + int ddictIsCold; /* if == 1 : dictionary is "new" for working context, and presumed "cold" (not in cpu cache) */ + + /* streaming */ ZSTD_dStreamStage streamStage; char* inBuff; size_t inBuffSize; @@ -153,6 +171,7 @@ struct ZSTD_DCtx_s U32 previousLegacyVersion; U32 legacyVersion; U32 hostageByte; + int noForwardProgress; /* workspace */ BYTE litBuffer[ZSTD_BLOCKSIZE_MAX + WILDCOPY_OVERLENGTH]; @@ -173,7 +192,7 @@ size_t ZSTD_estimateDCtxSize(void) { return sizeof(ZSTD_DCtx); } static size_t ZSTD_startingInputLength(ZSTD_format_e format) { size_t const startingInputLength = (format==ZSTD_f_zstd1_magicless) ? - ZSTD_frameHeaderSize_prefix - ZSTD_frameIdSize : + ZSTD_frameHeaderSize_prefix - ZSTD_FRAMEIDSIZE : ZSTD_frameHeaderSize_prefix; ZSTD_STATIC_ASSERT(ZSTD_FRAMEHEADERSIZE_PREFIX >= ZSTD_FRAMEIDSIZE); /* only supports formats ZSTD_f_zstd1 and ZSTD_f_zstd1_magicless */ @@ -188,10 +207,15 @@ static void ZSTD_initDCtx_internal(ZSTD_DCtx* dctx) dctx->maxWindowSize = ZSTD_MAXWINDOWSIZE_DEFAULT; dctx->ddict = NULL; dctx->ddictLocal = NULL; + dctx->dictEnd = NULL; + dctx->ddictIsCold = 0; dctx->inBuff = NULL; dctx->inBuffSize = 0; dctx->outBuffSize = 0; dctx->streamStage = zdss_init; + dctx->legacyContext = NULL; + dctx->previousLegacyVersion = 0; + dctx->noForwardProgress = 0; dctx->bmi2 = ZSTD_cpuid_bmi2(ZSTD_cpuid()); } @@ -215,8 +239,6 @@ ZSTD_DCtx* ZSTD_createDCtx_advanced(ZSTD_customMem customMem) { ZSTD_DCtx* const dctx = (ZSTD_DCtx*)ZSTD_malloc(sizeof(*dctx), customMem); if (!dctx) return NULL; dctx->customMem = customMem; - dctx->legacyContext = NULL; - dctx->previousLegacyVersion = 0; ZSTD_initDCtx_internal(dctx); return dctx; } @@ -265,7 +287,7 @@ void ZSTD_copyDCtx(ZSTD_DCtx* dstDCtx, const ZSTD_DCtx* srcDCtx) * Note 3 : Skippable Frame Identifiers are considered valid. */ unsigned ZSTD_isFrame(const void* buffer, size_t size) { - if (size < ZSTD_frameIdSize) return 0; + if (size < ZSTD_FRAMEIDSIZE) return 0; { U32 const magic = MEM_readLE32(buffer); if (magic == ZSTD_MAGICNUMBER) return 1; if ((magic & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) return 1; @@ -298,25 +320,28 @@ static size_t ZSTD_frameHeaderSize_internal(const void* src, size_t srcSize, ZST /** ZSTD_frameHeaderSize() : * srcSize must be >= ZSTD_frameHeaderSize_prefix. - * @return : size of the Frame Header */ + * @return : size of the Frame Header, + * or an error code (if srcSize is too small) */ size_t ZSTD_frameHeaderSize(const void* src, size_t srcSize) { return ZSTD_frameHeaderSize_internal(src, srcSize, ZSTD_f_zstd1); } -/** ZSTD_getFrameHeader_internal() : +/** ZSTD_getFrameHeader_advanced() : * decode Frame Header, or require larger `srcSize`. * note : only works for formats ZSTD_f_zstd1 and ZSTD_f_zstd1_magicless * @return : 0, `zfhPtr` is correctly filled, * >0, `srcSize` is too small, value is wanted `srcSize` amount, * or an error code, which can be tested using ZSTD_isError() */ -static size_t ZSTD_getFrameHeader_internal(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format) +size_t ZSTD_getFrameHeader_advanced(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format) { const BYTE* ip = (const BYTE*)src; size_t const minInputSize = ZSTD_startingInputLength(format); + memset(zfhPtr, 0, sizeof(*zfhPtr)); /* not strictly necessary, but static analyzer do not understand that zfhPtr is only going to be read only if return value is zero, since they are 2 different signals */ if (srcSize < minInputSize) return minInputSize; + if (src==NULL) return ERROR(GENERIC); /* invalid parameter */ if ( (format != ZSTD_f_zstd1_magicless) && (MEM_readLE32(src) != ZSTD_MAGICNUMBER) ) { @@ -325,7 +350,7 @@ static size_t ZSTD_getFrameHeader_internal(ZSTD_frameHeader* zfhPtr, const void* if (srcSize < ZSTD_skippableHeaderSize) return ZSTD_skippableHeaderSize; /* magic number + frame length */ memset(zfhPtr, 0, sizeof(*zfhPtr)); - zfhPtr->frameContentSize = MEM_readLE32((const char *)src + ZSTD_frameIdSize); + zfhPtr->frameContentSize = MEM_readLE32((const char *)src + ZSTD_FRAMEIDSIZE); zfhPtr->frameType = ZSTD_skippableFrame; return 0; } @@ -394,7 +419,7 @@ static size_t ZSTD_getFrameHeader_internal(ZSTD_frameHeader* zfhPtr, const void* * or an error code, which can be tested using ZSTD_isError() */ size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize) { - return ZSTD_getFrameHeader_internal(zfhPtr, src, srcSize, ZSTD_f_zstd1); + return ZSTD_getFrameHeader_advanced(zfhPtr, src, srcSize, ZSTD_f_zstd1); } @@ -437,7 +462,7 @@ unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize) size_t skippableSize; if (srcSize < ZSTD_skippableHeaderSize) return ERROR(srcSize_wrong); - skippableSize = MEM_readLE32((const BYTE *)src + ZSTD_frameIdSize) + skippableSize = MEM_readLE32((const BYTE *)src + ZSTD_FRAMEIDSIZE) + ZSTD_skippableHeaderSize; if (srcSize < skippableSize) { return ZSTD_CONTENTSIZE_ERROR; @@ -491,7 +516,7 @@ unsigned long long ZSTD_getDecompressedSize(const void* src, size_t srcSize) * @return : 0 if success, or an error code, which can be tested using ZSTD_isError() */ static size_t ZSTD_decodeFrameHeader(ZSTD_DCtx* dctx, const void* src, size_t headerSize) { - size_t const result = ZSTD_getFrameHeader_internal(&(dctx->fParams), src, headerSize, dctx->format); + size_t const result = ZSTD_getFrameHeader_advanced(&(dctx->fParams), src, headerSize, dctx->format); if (ZSTD_isError(result)) return result; /* invalid header */ if (result>0) return ERROR(srcSize_wrong); /* headerSize too small */ if (dctx->fParams.dictID && (dctx->dictID != dctx->fParams.dictID)) @@ -526,6 +551,7 @@ size_t ZSTD_getcBlockSize(const void* src, size_t srcSize, static size_t ZSTD_copyRawBlock(void* dst, size_t dstCapacity, const void* src, size_t srcSize) { + if (dst==NULL) return ERROR(dstSize_tooSmall); if (srcSize > dstCapacity) return ERROR(dstSize_tooSmall); memcpy(dst, src, srcSize); return srcSize; @@ -542,6 +568,9 @@ static size_t ZSTD_setRleBlock(void* dst, size_t dstCapacity, return regenSize; } +/* Hidden declaration for fullbench */ +size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx, + const void* src, size_t srcSize); /*! ZSTD_decodeLiteralsBlock() : * @return : nb of bytes read from src (< srcSize ) * note : symbol not declared but exposed for fullbench */ @@ -558,6 +587,7 @@ size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx, case set_repeat: if (dctx->litEntropy==0) return ERROR(dictionary_corrupted); /* fall-through */ + case set_compressed: if (srcSize < 5) return ERROR(corruption_detected); /* srcSize >= MIN_CBLOCK_SIZE == 3; here we need up to 5 for case 3 */ { size_t lhSize, litSize, litCSize; @@ -589,15 +619,20 @@ size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx, if (litSize > ZSTD_BLOCKSIZE_MAX) return ERROR(corruption_detected); if (litCSize + lhSize > srcSize) return ERROR(corruption_detected); + /* prefetch huffman table if cold */ + if (dctx->ddictIsCold && (litSize > 768 /* heuristic */)) { + PREFETCH_AREA(dctx->HUFptr, sizeof(dctx->entropy.hufTable)); + } + if (HUF_isError((litEncType==set_repeat) ? ( singleStream ? HUF_decompress1X_usingDTable_bmi2(dctx->litBuffer, litSize, istart+lhSize, litCSize, dctx->HUFptr, dctx->bmi2) : HUF_decompress4X_usingDTable_bmi2(dctx->litBuffer, litSize, istart+lhSize, litCSize, dctx->HUFptr, dctx->bmi2) ) : ( singleStream ? - HUF_decompress1X2_DCtx_wksp_bmi2(dctx->entropy.hufTable, dctx->litBuffer, litSize, istart+lhSize, litCSize, - dctx->entropy.workspace, sizeof(dctx->entropy.workspace), dctx->bmi2) : + HUF_decompress1X1_DCtx_wksp_bmi2(dctx->entropy.hufTable, dctx->litBuffer, litSize, istart+lhSize, litCSize, + dctx->workspace, sizeof(dctx->workspace), dctx->bmi2) : HUF_decompress4X_hufOnly_wksp_bmi2(dctx->entropy.hufTable, dctx->litBuffer, litSize, istart+lhSize, litCSize, - dctx->entropy.workspace, sizeof(dctx->entropy.workspace), dctx->bmi2)))) + dctx->workspace, sizeof(dctx->workspace), dctx->bmi2)))) return ERROR(corruption_detected); dctx->litPtr = dctx->litBuffer; @@ -869,7 +904,8 @@ static size_t ZSTD_buildSeqTable(ZSTD_seqSymbol* DTableSpace, const ZSTD_seqSymb symbolEncodingType_e type, U32 max, U32 maxLog, const void* src, size_t srcSize, const U32* baseValue, const U32* nbAdditionalBits, - const ZSTD_seqSymbol* defaultTable, U32 flagRepeatTable) + const ZSTD_seqSymbol* defaultTable, U32 flagRepeatTable, + int ddictIsCold, int nbSeq) { switch(type) { @@ -888,6 +924,12 @@ static size_t ZSTD_buildSeqTable(ZSTD_seqSymbol* DTableSpace, const ZSTD_seqSymb return 0; case set_repeat: if (!flagRepeatTable) return ERROR(corruption_detected); + /* prefetch FSE table if used */ + if (ddictIsCold && (nbSeq > 24 /* heuristic */)) { + const void* const pStart = *DTablePtr; + size_t const pSize = sizeof(ZSTD_seqSymbol) * (SEQSYMBOL_TABLE_SIZE(maxLog)); + PREFETCH_AREA(pStart, pSize); + } return 0; case set_compressed : { U32 tableLog; @@ -933,6 +975,9 @@ static const U32 ML_base[MaxML+1] = { 67, 83, 99, 0x83, 0x103, 0x203, 0x403, 0x803, 0x1003, 0x2003, 0x4003, 0x8003, 0x10003 }; +/* Hidden delcaration for fullbench */ +size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr, + const void* src, size_t srcSize); size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr, const void* src, size_t srcSize) @@ -940,25 +985,25 @@ size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr, const BYTE* const istart = (const BYTE* const)src; const BYTE* const iend = istart + srcSize; const BYTE* ip = istart; + int nbSeq; DEBUGLOG(5, "ZSTD_decodeSeqHeaders"); /* check */ if (srcSize < MIN_SEQUENCES_SIZE) return ERROR(srcSize_wrong); /* SeqHead */ - { int nbSeq = *ip++; - if (!nbSeq) { *nbSeqPtr=0; return 1; } - if (nbSeq > 0x7F) { - if (nbSeq == 0xFF) { - if (ip+2 > iend) return ERROR(srcSize_wrong); - nbSeq = MEM_readLE16(ip) + LONGNBSEQ, ip+=2; - } else { - if (ip >= iend) return ERROR(srcSize_wrong); - nbSeq = ((nbSeq-0x80)<<8) + *ip++; - } + nbSeq = *ip++; + if (!nbSeq) { *nbSeqPtr=0; return 1; } + if (nbSeq > 0x7F) { + if (nbSeq == 0xFF) { + if (ip+2 > iend) return ERROR(srcSize_wrong); + nbSeq = MEM_readLE16(ip) + LONGNBSEQ, ip+=2; + } else { + if (ip >= iend) return ERROR(srcSize_wrong); + nbSeq = ((nbSeq-0x80)<<8) + *ip++; } - *nbSeqPtr = nbSeq; } + *nbSeqPtr = nbSeq; /* FSE table descriptors */ if (ip+4 > iend) return ERROR(srcSize_wrong); /* minimum possible size */ @@ -972,7 +1017,8 @@ size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr, LLtype, MaxLL, LLFSELog, ip, iend-ip, LL_base, LL_bits, - LL_defaultDTable, dctx->fseEntropy); + LL_defaultDTable, dctx->fseEntropy, + dctx->ddictIsCold, nbSeq); if (ZSTD_isError(llhSize)) return ERROR(corruption_detected); ip += llhSize; } @@ -981,7 +1027,8 @@ size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr, OFtype, MaxOff, OffFSELog, ip, iend-ip, OF_base, OF_bits, - OF_defaultDTable, dctx->fseEntropy); + OF_defaultDTable, dctx->fseEntropy, + dctx->ddictIsCold, nbSeq); if (ZSTD_isError(ofhSize)) return ERROR(corruption_detected); ip += ofhSize; } @@ -990,12 +1037,23 @@ size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr, MLtype, MaxML, MLFSELog, ip, iend-ip, ML_base, ML_bits, - ML_defaultDTable, dctx->fseEntropy); + ML_defaultDTable, dctx->fseEntropy, + dctx->ddictIsCold, nbSeq); if (ZSTD_isError(mlhSize)) return ERROR(corruption_detected); ip += mlhSize; } } + /* prefetch dictionary content */ + if (dctx->ddictIsCold) { + size_t const dictSize = (const char*)dctx->prefixStart - (const char*)dctx->virtualStart; + size_t const psmin = MIN(dictSize, (size_t)(64*nbSeq) /* heuristic */ ); + size_t const pSize = MIN(psmin, 128 KB /* protection */ ); + const void* const pStart = (const char*)dctx->dictEnd - pSize; + PREFETCH_AREA(pStart, pSize); + dctx->ddictIsCold = 0; + } + return ip-istart; } @@ -1075,7 +1133,7 @@ HINT_INLINE size_t ZSTD_execSequence(BYTE* op, BYTE* const oend, seq_t sequence, const BYTE** litPtr, const BYTE* const litLimit, - const BYTE* const base, const BYTE* const vBase, const BYTE* const dictEnd) + const BYTE* const prefixStart, const BYTE* const virtualStart, const BYTE* const dictEnd) { BYTE* const oLitEnd = op + sequence.litLength; size_t const sequenceLength = sequence.litLength + sequence.matchLength; @@ -1087,7 +1145,7 @@ size_t ZSTD_execSequence(BYTE* op, /* check */ if (oMatchEnd>oend) return ERROR(dstSize_tooSmall); /* last match must start at a minimum distance of WILDCOPY_OVERLENGTH from oend */ if (iLitEnd > litLimit) return ERROR(corruption_detected); /* over-read beyond lit buffer */ - if (oLitEnd>oend_w) return ZSTD_execSequenceLast7(op, oend, sequence, litPtr, litLimit, base, vBase, dictEnd); + if (oLitEnd>oend_w) return ZSTD_execSequenceLast7(op, oend, sequence, litPtr, litLimit, prefixStart, virtualStart, dictEnd); /* copy Literals */ ZSTD_copy8(op, *litPtr); @@ -1097,11 +1155,11 @@ size_t ZSTD_execSequence(BYTE* op, *litPtr = iLitEnd; /* update for next sequence */ /* copy Match */ - if (sequence.offset > (size_t)(oLitEnd - base)) { + if (sequence.offset > (size_t)(oLitEnd - prefixStart)) { /* offset beyond prefix -> go into extDict */ - if (sequence.offset > (size_t)(oLitEnd - vBase)) + if (sequence.offset > (size_t)(oLitEnd - virtualStart)) return ERROR(corruption_detected); - match = dictEnd + (match - base); + match = dictEnd + (match - prefixStart); if (match + sequence.matchLength <= dictEnd) { memmove(oLitEnd, match, sequence.matchLength); return sequenceLength; @@ -1111,7 +1169,7 @@ size_t ZSTD_execSequence(BYTE* op, memmove(oLitEnd, match, length1); op = oLitEnd + length1; sequence.matchLength -= length1; - match = base; + match = prefixStart; if (op > oend_w || sequence.matchLength < MINMATCH) { U32 i; for (i = 0; i < sequence.matchLength; ++i) op[i] = match[i]; @@ -1354,10 +1412,10 @@ ZSTD_decompressSequences_body( ZSTD_DCtx* dctx, BYTE* op = ostart; const BYTE* litPtr = dctx->litPtr; const BYTE* const litEnd = litPtr + dctx->litSize; - const BYTE* const base = (const BYTE*) (dctx->base); - const BYTE* const vBase = (const BYTE*) (dctx->vBase); + const BYTE* const prefixStart = (const BYTE*) (dctx->prefixStart); + const BYTE* const vBase = (const BYTE*) (dctx->virtualStart); const BYTE* const dictEnd = (const BYTE*) (dctx->dictEnd); - DEBUGLOG(5, "ZSTD_decompressSequences"); + DEBUGLOG(5, "ZSTD_decompressSequences_body"); /* Regen sequences */ if (nbSeq) { @@ -1372,14 +1430,14 @@ ZSTD_decompressSequences_body( ZSTD_DCtx* dctx, for ( ; (BIT_reloadDStream(&(seqState.DStream)) <= BIT_DStream_completed) && nbSeq ; ) { nbSeq--; { seq_t const sequence = ZSTD_decodeSequence(&seqState, isLongOffset); - size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequence, &litPtr, litEnd, base, vBase, dictEnd); + size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequence, &litPtr, litEnd, prefixStart, vBase, dictEnd); DEBUGLOG(6, "regenerated sequence size : %u", (U32)oneSeqSize); if (ZSTD_isError(oneSeqSize)) return oneSeqSize; op += oneSeqSize; } } /* check if reached exact end */ - DEBUGLOG(5, "ZSTD_decompressSequences: after decode loop, remaining nbSeq : %i", nbSeq); + DEBUGLOG(5, "ZSTD_decompressSequences_body: after decode loop, remaining nbSeq : %i", nbSeq); if (nbSeq) return ERROR(corruption_detected); /* save reps for next block */ { U32 i; for (i=0; i<ZSTD_REP_NUM; i++) dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]); } @@ -1498,8 +1556,8 @@ ZSTD_decompressSequencesLong_body( BYTE* op = ostart; const BYTE* litPtr = dctx->litPtr; const BYTE* const litEnd = litPtr + dctx->litSize; - const BYTE* const prefixStart = (const BYTE*) (dctx->base); - const BYTE* const dictStart = (const BYTE*) (dctx->vBase); + const BYTE* const prefixStart = (const BYTE*) (dctx->prefixStart); + const BYTE* const dictStart = (const BYTE*) (dctx->virtualStart); const BYTE* const dictEnd = (const BYTE*) (dctx->dictEnd); /* Regen sequences */ @@ -1662,7 +1720,8 @@ static size_t ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx, /* isLongOffset must be true if there are long offsets. * Offsets are long if they are larger than 2^STREAM_ACCUMULATOR_MIN. * We don't expect that to be the case in 64-bit mode. - * In block mode, window size is not known, so we have to be conservative. (note: but it could be evaluated from current-lowLimit) + * In block mode, window size is not known, so we have to be conservative. + * (note: but it could be evaluated from current-lowLimit) */ ZSTD_longOffset_e const isLongOffset = (ZSTD_longOffset_e)(MEM_32bits() && (!frame || dctx->fParams.windowSize > (1ULL << STREAM_ACCUMULATOR_MIN))); DEBUGLOG(5, "ZSTD_decompressBlock_internal (size : %u)", (U32)srcSize); @@ -1701,8 +1760,8 @@ static void ZSTD_checkContinuity(ZSTD_DCtx* dctx, const void* dst) { if (dst != dctx->previousDstEnd) { /* not contiguous */ dctx->dictEnd = dctx->previousDstEnd; - dctx->vBase = (const char*)dst - ((const char*)(dctx->previousDstEnd) - (const char*)(dctx->base)); - dctx->base = dst; + dctx->virtualStart = (const char*)dst - ((const char*)(dctx->previousDstEnd) - (const char*)(dctx->prefixStart)); + dctx->prefixStart = dst; dctx->previousDstEnd = dst; } } @@ -1729,10 +1788,10 @@ ZSTDLIB_API size_t ZSTD_insertBlock(ZSTD_DCtx* dctx, const void* blockStart, siz } -static size_t ZSTD_generateNxBytes(void* dst, size_t dstCapacity, BYTE byte, size_t length) +static size_t ZSTD_generateNxBytes(void* dst, size_t dstCapacity, BYTE value, size_t length) { if (length > dstCapacity) return ERROR(dstSize_tooSmall); - memset(dst, byte, length); + memset(dst, value, length); return length; } @@ -1749,7 +1808,7 @@ size_t ZSTD_findFrameCompressedSize(const void *src, size_t srcSize) #endif if ( (srcSize >= ZSTD_skippableHeaderSize) && (MEM_readLE32(src) & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START ) { - return ZSTD_skippableHeaderSize + MEM_readLE32((const BYTE*)src + ZSTD_frameIdSize); + return ZSTD_skippableHeaderSize + MEM_readLE32((const BYTE*)src + ZSTD_FRAMEIDSIZE); } else { const BYTE* ip = (const BYTE*)src; const BYTE* const ipstart = ip; @@ -1783,7 +1842,6 @@ size_t ZSTD_findFrameCompressedSize(const void *src, size_t srcSize) if (zfh.checksumFlag) { /* Final frame content checksum */ if (remainingSize < 4) return ERROR(srcSize_wrong); ip += 4; - remainingSize -= 4; } return ip - ipstart; @@ -1871,9 +1929,6 @@ static size_t ZSTD_decompressFrame(ZSTD_DCtx* dctx, return op-ostart; } -static const void* ZSTD_DDictDictContent(const ZSTD_DDict* ddict); -static size_t ZSTD_DDictDictSize(const ZSTD_DDict* ddict); - static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize, @@ -1881,6 +1936,9 @@ static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict) { void* const dststart = dst; + int moreThan1Frame = 0; + + DEBUGLOG(5, "ZSTD_decompressMultiFrame"); assert(dict==NULL || ddict==NULL); /* either dict or ddict set, not both */ if (ddict) { @@ -1889,7 +1947,6 @@ static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx, } while (srcSize >= ZSTD_frameHeaderSize_prefix) { - U32 magicNumber; #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT >= 1) if (ZSTD_isLegacy(src, srcSize)) { @@ -1911,24 +1968,21 @@ static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx, } #endif - magicNumber = MEM_readLE32(src); - DEBUGLOG(4, "reading magic number %08X (expecting %08X)", - (U32)magicNumber, (U32)ZSTD_MAGICNUMBER); - if (magicNumber != ZSTD_MAGICNUMBER) { + { U32 const magicNumber = MEM_readLE32(src); + DEBUGLOG(4, "reading magic number %08X (expecting %08X)", + (U32)magicNumber, (U32)ZSTD_MAGICNUMBER); if ((magicNumber & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) { size_t skippableSize; if (srcSize < ZSTD_skippableHeaderSize) return ERROR(srcSize_wrong); - skippableSize = MEM_readLE32((const BYTE*)src + ZSTD_frameIdSize) + skippableSize = MEM_readLE32((const BYTE*)src + ZSTD_FRAMEIDSIZE) + ZSTD_skippableHeaderSize; if (srcSize < skippableSize) return ERROR(srcSize_wrong); src = (const BYTE *)src + skippableSize; srcSize -= skippableSize; continue; - } - return ERROR(prefix_unknown); - } + } } if (ddict) { /* we were called from ZSTD_decompress_usingDDict */ @@ -1942,11 +1996,25 @@ static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx, { const size_t res = ZSTD_decompressFrame(dctx, dst, dstCapacity, &src, &srcSize); + if ( (ZSTD_getErrorCode(res) == ZSTD_error_prefix_unknown) + && (moreThan1Frame==1) ) { + /* at least one frame successfully completed, + * but following bytes are garbage : + * it's more likely to be a srcSize error, + * specifying more bytes than compressed size of frame(s). + * This error message replaces ERROR(prefix_unknown), + * which would be confusing, as the first header is actually correct. + * Note that one could be unlucky, it might be a corruption error instead, + * happening right at the place where we expect zstd magic bytes. + * But this is _much_ less likely than a srcSize field error. */ + return ERROR(srcSize_wrong); + } if (ZSTD_isError(res)) return res; /* no need to bound check, ZSTD_decompressFrame already has */ dst = (BYTE*)dst + res; dstCapacity -= res; } + moreThan1Frame = 1; } /* while (srcSize >= ZSTD_frameHeaderSize_prefix) */ if (srcSize) return ERROR(srcSize_wrong); /* input not entirely consumed */ @@ -1980,6 +2048,7 @@ size_t ZSTD_decompress(void* dst, size_t dstCapacity, const void* src, size_t sr return regenSize; #else /* stack mode */ ZSTD_DCtx dctx; + ZSTD_initDCtx_internal(&dctx); return ZSTD_decompressDCtx(&dctx, dst, dstCapacity, src, srcSize); #endif } @@ -2031,7 +2100,7 @@ size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, c case ZSTDds_getFrameHeaderSize : assert(src != NULL); if (dctx->format == ZSTD_f_zstd1) { /* allows header */ - assert(srcSize >= ZSTD_frameIdSize); /* to read skippable magic number */ + assert(srcSize >= ZSTD_FRAMEIDSIZE); /* to read skippable magic number */ if ((MEM_readLE32(src) & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) { /* skippable frame */ memcpy(dctx->headerBuffer, src, srcSize); dctx->expected = ZSTD_skippableHeaderSize - srcSize; /* remaining to load to get full skippable frame header */ @@ -2141,7 +2210,7 @@ size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, c assert(src != NULL); assert(srcSize <= ZSTD_skippableHeaderSize); memcpy(dctx->headerBuffer + (ZSTD_skippableHeaderSize - srcSize), src, srcSize); /* complete skippable header */ - dctx->expected = MEM_readLE32(dctx->headerBuffer + ZSTD_frameIdSize); /* note : dctx->expected can grow seriously large, beyond local buffer size */ + dctx->expected = MEM_readLE32(dctx->headerBuffer + ZSTD_FRAMEIDSIZE); /* note : dctx->expected can grow seriously large, beyond local buffer size */ dctx->stage = ZSTDds_skipFrame; return 0; @@ -2159,27 +2228,33 @@ size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, c static size_t ZSTD_refDictContent(ZSTD_DCtx* dctx, const void* dict, size_t dictSize) { dctx->dictEnd = dctx->previousDstEnd; - dctx->vBase = (const char*)dict - ((const char*)(dctx->previousDstEnd) - (const char*)(dctx->base)); - dctx->base = dict; + dctx->virtualStart = (const char*)dict - ((const char*)(dctx->previousDstEnd) - (const char*)(dctx->prefixStart)); + dctx->prefixStart = dict; dctx->previousDstEnd = (const char*)dict + dictSize; return 0; } -/* ZSTD_loadEntropy() : - * dict : must point at beginning of a valid zstd dictionary +/*! ZSTD_loadEntropy() : + * dict : must point at beginning of a valid zstd dictionary. * @return : size of entropy tables read */ -static size_t ZSTD_loadEntropy(ZSTD_entropyDTables_t* entropy, const void* const dict, size_t const dictSize) +static size_t ZSTD_loadEntropy(ZSTD_entropyDTables_t* entropy, + const void* const dict, size_t const dictSize) { const BYTE* dictPtr = (const BYTE*)dict; const BYTE* const dictEnd = dictPtr + dictSize; if (dictSize <= 8) return ERROR(dictionary_corrupted); + assert(MEM_readLE32(dict) == ZSTD_MAGIC_DICTIONARY); /* dict must be valid */ dictPtr += 8; /* skip header = magic + dictID */ - - { size_t const hSize = HUF_readDTableX4_wksp( - entropy->hufTable, dictPtr, dictEnd - dictPtr, - entropy->workspace, sizeof(entropy->workspace)); + ZSTD_STATIC_ASSERT(offsetof(ZSTD_entropyDTables_t, OFTable) == offsetof(ZSTD_entropyDTables_t, LLTable) + sizeof(entropy->LLTable)); + ZSTD_STATIC_ASSERT(offsetof(ZSTD_entropyDTables_t, MLTable) == offsetof(ZSTD_entropyDTables_t, OFTable) + sizeof(entropy->OFTable)); + ZSTD_STATIC_ASSERT(sizeof(entropy->LLTable) + sizeof(entropy->OFTable) + sizeof(entropy->MLTable) >= HUF_DECOMPRESS_WORKSPACE_SIZE); + { void* const workspace = &entropy->LLTable; /* use fse tables as temporary workspace; implies fse tables are grouped together */ + size_t const workspaceSize = sizeof(entropy->LLTable) + sizeof(entropy->OFTable) + sizeof(entropy->MLTable); + size_t const hSize = HUF_readDTableX2_wksp(entropy->hufTable, + dictPtr, dictEnd - dictPtr, + workspace, workspaceSize); if (HUF_isError(hSize)) return ERROR(dictionary_corrupted); dictPtr += hSize; } @@ -2190,7 +2265,7 @@ static size_t ZSTD_loadEntropy(ZSTD_entropyDTables_t* entropy, const void* const if (FSE_isError(offcodeHeaderSize)) return ERROR(dictionary_corrupted); if (offcodeMaxValue > MaxOff) return ERROR(dictionary_corrupted); if (offcodeLog > OffFSELog) return ERROR(dictionary_corrupted); - ZSTD_buildFSETable(entropy->OFTable, + ZSTD_buildFSETable( entropy->OFTable, offcodeNCount, offcodeMaxValue, OF_base, OF_bits, offcodeLog); @@ -2203,7 +2278,7 @@ static size_t ZSTD_loadEntropy(ZSTD_entropyDTables_t* entropy, const void* const if (FSE_isError(matchlengthHeaderSize)) return ERROR(dictionary_corrupted); if (matchlengthMaxValue > MaxML) return ERROR(dictionary_corrupted); if (matchlengthLog > MLFSELog) return ERROR(dictionary_corrupted); - ZSTD_buildFSETable(entropy->MLTable, + ZSTD_buildFSETable( entropy->MLTable, matchlengthNCount, matchlengthMaxValue, ML_base, ML_bits, matchlengthLog); @@ -2216,7 +2291,7 @@ static size_t ZSTD_loadEntropy(ZSTD_entropyDTables_t* entropy, const void* const if (FSE_isError(litlengthHeaderSize)) return ERROR(dictionary_corrupted); if (litlengthMaxValue > MaxLL) return ERROR(dictionary_corrupted); if (litlengthLog > LLFSELog) return ERROR(dictionary_corrupted); - ZSTD_buildFSETable(entropy->LLTable, + ZSTD_buildFSETable( entropy->LLTable, litlengthNCount, litlengthMaxValue, LL_base, LL_bits, litlengthLog); @@ -2242,7 +2317,7 @@ static size_t ZSTD_decompress_insertDictionary(ZSTD_DCtx* dctx, const void* dict if (magic != ZSTD_MAGIC_DICTIONARY) { return ZSTD_refDictContent(dctx, dict, dictSize); /* pure content mode */ } } - dctx->dictID = MEM_readLE32((const char*)dict + ZSTD_frameIdSize); + dctx->dictID = MEM_readLE32((const char*)dict + ZSTD_FRAMEIDSIZE); /* load entropy tables */ { size_t const eSize = ZSTD_loadEntropy(&dctx->entropy, dict, dictSize); @@ -2256,7 +2331,6 @@ static size_t ZSTD_decompress_insertDictionary(ZSTD_DCtx* dctx, const void* dict return ZSTD_refDictContent(dctx, dict, dictSize); } -/* Note : this function cannot fail */ size_t ZSTD_decompressBegin(ZSTD_DCtx* dctx) { assert(dctx != NULL); @@ -2264,8 +2338,8 @@ size_t ZSTD_decompressBegin(ZSTD_DCtx* dctx) dctx->stage = ZSTDds_getFrameHeaderSize; dctx->decodedSize = 0; dctx->previousDstEnd = NULL; - dctx->base = NULL; - dctx->vBase = NULL; + dctx->prefixStart = NULL; + dctx->virtualStart = NULL; dctx->dictEnd = NULL; dctx->entropy.hufTable[0] = (HUF_DTable)((HufLog)*0x1000001); /* cover both little and big endian */ dctx->litEntropy = dctx->fseEntropy = 0; @@ -2302,42 +2376,53 @@ struct ZSTD_DDict_s { static const void* ZSTD_DDictDictContent(const ZSTD_DDict* ddict) { + assert(ddict != NULL); return ddict->dictContent; } static size_t ZSTD_DDictDictSize(const ZSTD_DDict* ddict) { + assert(ddict != NULL); return ddict->dictSize; } -size_t ZSTD_decompressBegin_usingDDict(ZSTD_DCtx* dstDCtx, const ZSTD_DDict* ddict) +size_t ZSTD_decompressBegin_usingDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict) { - CHECK_F( ZSTD_decompressBegin(dstDCtx) ); - if (ddict) { /* support begin on NULL */ - dstDCtx->dictID = ddict->dictID; - dstDCtx->base = ddict->dictContent; - dstDCtx->vBase = ddict->dictContent; - dstDCtx->dictEnd = (const BYTE*)ddict->dictContent + ddict->dictSize; - dstDCtx->previousDstEnd = dstDCtx->dictEnd; + DEBUGLOG(4, "ZSTD_decompressBegin_usingDDict"); + assert(dctx != NULL); + if (ddict) { + dctx->ddictIsCold = (dctx->dictEnd != (const char*)ddict->dictContent + ddict->dictSize); + DEBUGLOG(4, "DDict is %s", + dctx->ddictIsCold ? "~cold~" : "hot!"); + } + CHECK_F( ZSTD_decompressBegin(dctx) ); + if (ddict) { /* NULL ddict is equivalent to no dictionary */ + dctx->dictID = ddict->dictID; + dctx->prefixStart = ddict->dictContent; + dctx->virtualStart = ddict->dictContent; + dctx->dictEnd = (const BYTE*)ddict->dictContent + ddict->dictSize; + dctx->previousDstEnd = dctx->dictEnd; if (ddict->entropyPresent) { - dstDCtx->litEntropy = 1; - dstDCtx->fseEntropy = 1; - dstDCtx->LLTptr = ddict->entropy.LLTable; - dstDCtx->MLTptr = ddict->entropy.MLTable; - dstDCtx->OFTptr = ddict->entropy.OFTable; - dstDCtx->HUFptr = ddict->entropy.hufTable; - dstDCtx->entropy.rep[0] = ddict->entropy.rep[0]; - dstDCtx->entropy.rep[1] = ddict->entropy.rep[1]; - dstDCtx->entropy.rep[2] = ddict->entropy.rep[2]; + dctx->litEntropy = 1; + dctx->fseEntropy = 1; + dctx->LLTptr = ddict->entropy.LLTable; + dctx->MLTptr = ddict->entropy.MLTable; + dctx->OFTptr = ddict->entropy.OFTable; + dctx->HUFptr = ddict->entropy.hufTable; + dctx->entropy.rep[0] = ddict->entropy.rep[0]; + dctx->entropy.rep[1] = ddict->entropy.rep[1]; + dctx->entropy.rep[2] = ddict->entropy.rep[2]; } else { - dstDCtx->litEntropy = 0; - dstDCtx->fseEntropy = 0; + dctx->litEntropy = 0; + dctx->fseEntropy = 0; } } return 0; } -static size_t ZSTD_loadEntropy_inDDict(ZSTD_DDict* ddict, ZSTD_dictContentType_e dictContentType) +static size_t +ZSTD_loadEntropy_inDDict(ZSTD_DDict* ddict, + ZSTD_dictContentType_e dictContentType) { ddict->dictID = 0; ddict->entropyPresent = 0; @@ -2355,10 +2440,12 @@ static size_t ZSTD_loadEntropy_inDDict(ZSTD_DDict* ddict, ZSTD_dictContentType_e return 0; /* pure content mode */ } } - ddict->dictID = MEM_readLE32((const char*)ddict->dictContent + ZSTD_frameIdSize); + ddict->dictID = MEM_readLE32((const char*)ddict->dictContent + ZSTD_FRAMEIDSIZE); /* load entropy tables */ - CHECK_E( ZSTD_loadEntropy(&ddict->entropy, ddict->dictContent, ddict->dictSize), dictionary_corrupted ); + CHECK_E( ZSTD_loadEntropy(&ddict->entropy, + ddict->dictContent, ddict->dictSize), + dictionary_corrupted ); ddict->entropyPresent = 1; return 0; } @@ -2372,6 +2459,7 @@ static size_t ZSTD_initDDict_internal(ZSTD_DDict* ddict, if ((dictLoadMethod == ZSTD_dlm_byRef) || (!dict) || (!dictSize)) { ddict->dictBuffer = NULL; ddict->dictContent = dict; + if (!dict) dictSize = 0; } else { void* const internalBuffer = ZSTD_malloc(dictSize, ddict->cMem); ddict->dictBuffer = internalBuffer; @@ -2396,14 +2484,15 @@ ZSTD_DDict* ZSTD_createDDict_advanced(const void* dict, size_t dictSize, if (!customMem.customAlloc ^ !customMem.customFree) return NULL; { ZSTD_DDict* const ddict = (ZSTD_DDict*) ZSTD_malloc(sizeof(ZSTD_DDict), customMem); - if (!ddict) return NULL; + if (ddict == NULL) return NULL; ddict->cMem = customMem; - - if (ZSTD_isError( ZSTD_initDDict_internal(ddict, dict, dictSize, dictLoadMethod, dictContentType) )) { - ZSTD_freeDDict(ddict); - return NULL; - } - + { size_t const initResult = ZSTD_initDDict_internal(ddict, + dict, dictSize, + dictLoadMethod, dictContentType); + if (ZSTD_isError(initResult)) { + ZSTD_freeDDict(ddict); + return NULL; + } } return ddict; } } @@ -2430,23 +2519,25 @@ ZSTD_DDict* ZSTD_createDDict_byReference(const void* dictBuffer, size_t dictSize const ZSTD_DDict* ZSTD_initStaticDDict( - void* workspace, size_t workspaceSize, + void* sBuffer, size_t sBufferSize, const void* dict, size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod, ZSTD_dictContentType_e dictContentType) { - size_t const neededSpace = - sizeof(ZSTD_DDict) + (dictLoadMethod == ZSTD_dlm_byRef ? 0 : dictSize); - ZSTD_DDict* const ddict = (ZSTD_DDict*)workspace; - assert(workspace != NULL); + size_t const neededSpace = sizeof(ZSTD_DDict) + + (dictLoadMethod == ZSTD_dlm_byRef ? 0 : dictSize); + ZSTD_DDict* const ddict = (ZSTD_DDict*)sBuffer; + assert(sBuffer != NULL); assert(dict != NULL); - if ((size_t)workspace & 7) return NULL; /* 8-aligned */ - if (workspaceSize < neededSpace) return NULL; + if ((size_t)sBuffer & 7) return NULL; /* 8-aligned */ + if (sBufferSize < neededSpace) return NULL; if (dictLoadMethod == ZSTD_dlm_byCopy) { memcpy(ddict+1, dict, dictSize); /* local copy */ dict = ddict+1; } - if (ZSTD_isError( ZSTD_initDDict_internal(ddict, dict, dictSize, ZSTD_dlm_byRef, dictContentType) )) + if (ZSTD_isError( ZSTD_initDDict_internal(ddict, + dict, dictSize, + ZSTD_dlm_byRef, dictContentType) )) return NULL; return ddict; } @@ -2484,7 +2575,7 @@ unsigned ZSTD_getDictID_fromDict(const void* dict, size_t dictSize) { if (dictSize < 8) return 0; if (MEM_readLE32(dict) != ZSTD_MAGIC_DICTIONARY) return 0; - return MEM_readLE32((const char*)dict + ZSTD_frameIdSize); + return MEM_readLE32((const char*)dict + ZSTD_FRAMEIDSIZE); } /*! ZSTD_getDictID_fromDDict() : @@ -2560,12 +2651,15 @@ size_t ZSTD_freeDStream(ZSTD_DStream* zds) } -/* *** Initialization *** */ +/* *** Initialization *** */ size_t ZSTD_DStreamInSize(void) { return ZSTD_BLOCKSIZE_MAX + ZSTD_blockHeaderSize; } size_t ZSTD_DStreamOutSize(void) { return ZSTD_BLOCKSIZE_MAX; } -size_t ZSTD_DCtx_loadDictionary_advanced(ZSTD_DCtx* dctx, const void* dict, size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod, ZSTD_dictContentType_e dictContentType) +size_t ZSTD_DCtx_loadDictionary_advanced(ZSTD_DCtx* dctx, + const void* dict, size_t dictSize, + ZSTD_dictLoadMethod_e dictLoadMethod, + ZSTD_dictContentType_e dictContentType) { if (dctx->streamStage != zdss_init) return ERROR(stage_wrong); ZSTD_freeDDict(dctx->ddictLocal); @@ -2607,6 +2701,7 @@ size_t ZSTD_initDStream_usingDict(ZSTD_DStream* zds, const void* dict, size_t di { DEBUGLOG(4, "ZSTD_initDStream_usingDict"); zds->streamStage = zdss_init; + zds->noForwardProgress = 0; CHECK_F( ZSTD_DCtx_loadDictionary(zds, dict, dictSize) ); return ZSTD_frameHeaderSize_prefix; } @@ -2618,13 +2713,6 @@ size_t ZSTD_initDStream(ZSTD_DStream* zds) return ZSTD_initDStream_usingDict(zds, NULL, 0); } -size_t ZSTD_DCtx_refDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict) -{ - if (dctx->streamStage != zdss_init) return ERROR(stage_wrong); - dctx->ddict = ddict; - return 0; -} - /* ZSTD_initDStream_usingDDict() : * ddict will just be referenced, and must outlive decompression session * this function cannot fail */ @@ -2663,6 +2751,13 @@ size_t ZSTD_setDStreamParameter(ZSTD_DStream* dctx, return 0; } +size_t ZSTD_DCtx_refDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict) +{ + if (dctx->streamStage != zdss_init) return ERROR(stage_wrong); + dctx->ddict = ddict; + return 0; +} + size_t ZSTD_DCtx_setMaxWindowSize(ZSTD_DCtx* dctx, size_t maxWindowSize) { if (dctx->streamStage != zdss_init) return ERROR(stage_wrong); @@ -2767,7 +2862,7 @@ size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inB return hint; } } #endif - { size_t const hSize = ZSTD_getFrameHeader_internal(&zds->fParams, zds->headerBuffer, zds->lhSize, zds->format); + { size_t const hSize = ZSTD_getFrameHeader_advanced(&zds->fParams, zds->headerBuffer, zds->lhSize, zds->format); DEBUGLOG(5, "header size : %u", (U32)hSize); if (ZSTD_isError(hSize)) { #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT>=1) @@ -2828,7 +2923,7 @@ size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inB CHECK_F(ZSTD_decompressBegin_usingDDict(zds, zds->ddict)); if ((MEM_readLE32(zds->headerBuffer) & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) { /* skippable frame */ - zds->expected = MEM_readLE32(zds->headerBuffer + ZSTD_frameIdSize); + zds->expected = MEM_readLE32(zds->headerBuffer + ZSTD_FRAMEIDSIZE); zds->stage = ZSTDds_skipFrame; } else { CHECK_F(ZSTD_decodeFrameHeader(zds, zds->headerBuffer, zds->lhSize)); @@ -2947,8 +3042,18 @@ size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inB } } /* result */ - input->pos += (size_t)(ip-istart); - output->pos += (size_t)(op-ostart); + input->pos = (size_t)(ip - (const char*)(input->src)); + output->pos = (size_t)(op - (char*)(output->dst)); + if ((ip==istart) && (op==ostart)) { /* no forward progress */ + zds->noForwardProgress ++; + if (zds->noForwardProgress >= ZSTD_NO_FORWARD_PROGRESS_MAX) { + if (op==oend) return ERROR(dstSize_tooSmall); + if (ip==iend) return ERROR(srcSize_wrong); + assert(0); + } + } else { + zds->noForwardProgress = 0; + } { size_t nextSrcSizeHint = ZSTD_nextSrcSizeToDecompress(zds); if (!nextSrcSizeHint) { /* frame fully decoded */ if (zds->outEnd == zds->outStart) { /* output fully flushed */ |