From b2d406a411b31acce5bbf0246af32a80c22ca834 Mon Sep 17 00:00:00 2001 From: Vicki Pfau Date: Thu, 16 Jan 2020 20:10:20 -0800 Subject: [PATCH] Third-Party: Update LZMA SDK --- src/third-party/lzma/7z.h | 12 +- src/third-party/lzma/7zAlloc.c | 12 +- src/third-party/lzma/7zAlloc.h | 20 +- src/third-party/lzma/7zArcIn.c | 135 +- src/third-party/lzma/7zBuf.c | 12 +- src/third-party/lzma/7zBuf.h | 10 +- src/third-party/lzma/7zBuf2.c | 17 +- src/third-party/lzma/7zCrc.c | 16 +- src/third-party/lzma/7zCrcOpt.c | 50 +- src/third-party/lzma/7zDec.c | 64 +- src/third-party/lzma/7zFile.c | 26 +- src/third-party/lzma/7zFile.h | 8 +- src/third-party/lzma/7zStream.c | 111 +- src/third-party/lzma/7zTypes.h | 227 +- src/third-party/lzma/7zVersion.h | 22 +- src/third-party/lzma/Aes.c | 67 +- src/third-party/lzma/AesOpt.c | 4 +- src/third-party/lzma/Alloc.c | 399 ++- src/third-party/lzma/Alloc.h | 20 +- src/third-party/lzma/Bcj2.c | 13 +- src/third-party/lzma/Bcj2Enc.c | 11 +- src/third-party/lzma/Bra.c | 289 +- src/third-party/lzma/Bra86.c | 4 +- src/third-party/lzma/BraIA64.c | 82 +- src/third-party/lzma/Compiler.h | 3 +- src/third-party/lzma/CpuArch.c | 39 +- src/third-party/lzma/CpuArch.h | 182 +- src/third-party/lzma/DllSecur.c | 108 + src/third-party/lzma/DllSecur.h | 20 + src/third-party/lzma/LzFind.c | 321 +- src/third-party/lzma/LzFind.h | 12 +- src/third-party/lzma/LzFindMt.c | 182 +- src/third-party/lzma/LzFindMt.h | 18 +- src/third-party/lzma/Lzma2Dec.c | 354 ++- src/third-party/lzma/Lzma2Dec.h | 66 +- src/third-party/lzma/Lzma2DecMt.c | 1082 +++++++ src/third-party/lzma/Lzma2DecMt.h | 79 + src/third-party/lzma/Lzma2Enc.c | 695 +++-- src/third-party/lzma/Lzma2Enc.h | 45 +- src/third-party/lzma/Lzma86Dec.c | 6 +- src/third-party/lzma/Lzma86Enc.c | 12 +- src/third-party/lzma/LzmaDec.c | 439 +-- src/third-party/lzma/LzmaDec.h | 47 +- src/third-party/lzma/LzmaEnc.c | 2726 ++++++++++------- src/third-party/lzma/LzmaEnc.h | 48 +- src/third-party/lzma/MtCoder.c | 744 +++-- src/third-party/lzma/MtCoder.h | 153 +- src/third-party/lzma/MtDec.c | 1138 +++++++ src/third-party/lzma/MtDec.h | 201 ++ src/third-party/lzma/Ppmd.h | 6 +- src/third-party/lzma/Ppmd7.c | 34 +- src/third-party/lzma/Ppmd7.h | 30 +- src/third-party/lzma/Ppmd7Dec.c | 36 +- src/third-party/lzma/Ppmd7Enc.c | 4 +- src/third-party/lzma/Sha256.c | 6 +- src/third-party/lzma/Threads.c | 14 +- src/third-party/lzma/Threads.h | 5 +- src/third-party/lzma/Util/7z/7zMain.c | 176 +- src/third-party/lzma/Util/Lzma/LzmaUtil.c | 42 +- src/third-party/lzma/Util/Lzma/LzmaUtil.dsp | 8 +- src/third-party/lzma/Util/SfxSetup/SfxSetup.c | 51 +- .../lzma/Util/SfxSetup/SfxSetup.dsp | 8 + src/third-party/lzma/Util/SfxSetup/makefile | 2 + .../lzma/Util/SfxSetup/makefile_con | 2 + src/third-party/lzma/Xz.c | 18 +- src/third-party/lzma/Xz.h | 239 +- src/third-party/lzma/XzCrc64.c | 22 +- src/third-party/lzma/XzCrc64Opt.c | 30 +- src/third-party/lzma/XzDec.c | 2265 ++++++++++++-- src/third-party/lzma/XzEnc.c | 1355 ++++++-- src/third-party/lzma/XzEnc.h | 27 +- src/third-party/lzma/XzIn.c | 62 +- src/util/vfs/vfs-lzma.c | 18 +- 73 files changed, 11359 insertions(+), 3452 deletions(-) create mode 100644 src/third-party/lzma/DllSecur.c create mode 100644 src/third-party/lzma/DllSecur.h create mode 100644 src/third-party/lzma/Lzma2DecMt.c create mode 100644 src/third-party/lzma/Lzma2DecMt.h create mode 100644 src/third-party/lzma/MtDec.c create mode 100644 src/third-party/lzma/MtDec.h diff --git a/src/third-party/lzma/7z.h b/src/third-party/lzma/7z.h index 216f381ca..82813c298 100644 --- a/src/third-party/lzma/7z.h +++ b/src/third-party/lzma/7z.h @@ -1,5 +1,5 @@ /* 7z.h -- 7z interface -2015-11-18 : Igor Pavlov : Public domain */ +2017-04-03 : Igor Pavlov : Public domain */ #ifndef __7Z_H #define __7Z_H @@ -98,7 +98,7 @@ UInt64 SzAr_GetFolderUnpackSize(const CSzAr *p, UInt32 folderIndex); SRes SzAr_DecodeFolder(const CSzAr *p, UInt32 folderIndex, ILookInStream *stream, UInt64 startPos, Byte *outBuffer, size_t outSize, - ISzAlloc *allocMain); + ISzAllocPtr allocMain); typedef struct { @@ -131,7 +131,7 @@ typedef struct #define SzArEx_GetFileSize(p, i) ((p)->UnpackPositions[(i) + 1] - (p)->UnpackPositions[i]) void SzArEx_Init(CSzArEx *p); -void SzArEx_Free(CSzArEx *p, ISzAlloc *alloc); +void SzArEx_Free(CSzArEx *p, ISzAllocPtr alloc); UInt64 SzArEx_GetFolderStreamPos(const CSzArEx *p, UInt32 folderIndex, UInt32 indexInFolder); int SzArEx_GetFolderFullPackSize(const CSzArEx *p, UInt32 folderIndex, UInt64 *resSize); @@ -179,8 +179,8 @@ SRes SzArEx_Extract( size_t *outBufferSize, /* buffer size for output buffer */ size_t *offset, /* offset of stream for required file in *outBuffer */ size_t *outSizeProcessed, /* size of file in *outBuffer */ - ISzAlloc *allocMain, - ISzAlloc *allocTemp); + ISzAllocPtr allocMain, + ISzAllocPtr allocTemp); /* @@ -195,7 +195,7 @@ SZ_ERROR_FAIL */ SRes SzArEx_Open(CSzArEx *p, ILookInStream *inStream, - ISzAlloc *allocMain, ISzAlloc *allocTemp); + ISzAllocPtr allocMain, ISzAllocPtr allocTemp); EXTERN_C_END diff --git a/src/third-party/lzma/7zAlloc.c b/src/third-party/lzma/7zAlloc.c index 360da0516..ea32809c6 100644 --- a/src/third-party/lzma/7zAlloc.c +++ b/src/third-party/lzma/7zAlloc.c @@ -1,8 +1,10 @@ /* 7zAlloc.c -- Allocation functions -2015-11-09 : Igor Pavlov : Public domain */ +2017-04-03 : Igor Pavlov : Public domain */ #include "Precomp.h" +#include + #include "7zAlloc.h" /* #define _SZ_ALLOC_DEBUG */ @@ -20,7 +22,7 @@ int g_allocCountTemp = 0; #endif -void *SzAlloc(void *p, size_t size) +void *SzAlloc(ISzAllocPtr p, size_t size) { UNUSED_VAR(p); if (size == 0) @@ -32,7 +34,7 @@ void *SzAlloc(void *p, size_t size) return malloc(size); } -void SzFree(void *p, void *address) +void SzFree(ISzAllocPtr p, void *address) { UNUSED_VAR(p); #ifdef _SZ_ALLOC_DEBUG @@ -45,7 +47,7 @@ void SzFree(void *p, void *address) free(address); } -void *SzAllocTemp(void *p, size_t size) +void *SzAllocTemp(ISzAllocPtr p, size_t size) { UNUSED_VAR(p); if (size == 0) @@ -60,7 +62,7 @@ void *SzAllocTemp(void *p, size_t size) return malloc(size); } -void SzFreeTemp(void *p, void *address) +void SzFreeTemp(ISzAllocPtr p, void *address) { UNUSED_VAR(p); #ifdef _SZ_ALLOC_DEBUG diff --git a/src/third-party/lzma/7zAlloc.h b/src/third-party/lzma/7zAlloc.h index 4d7502ffb..c0f89d73c 100644 --- a/src/third-party/lzma/7zAlloc.h +++ b/src/third-party/lzma/7zAlloc.h @@ -1,23 +1,19 @@ /* 7zAlloc.h -- Allocation functions -2013-03-25 : Igor Pavlov : Public domain */ +2017-04-03 : Igor Pavlov : Public domain */ #ifndef __7Z_ALLOC_H #define __7Z_ALLOC_H -#include +#include "7zTypes.h" -#ifdef __cplusplus -extern "C" { -#endif +EXTERN_C_BEGIN -void *SzAlloc(void *p, size_t size); -void SzFree(void *p, void *address); +void *SzAlloc(ISzAllocPtr p, size_t size); +void SzFree(ISzAllocPtr p, void *address); -void *SzAllocTemp(void *p, size_t size); -void SzFreeTemp(void *p, void *address); +void *SzAllocTemp(ISzAllocPtr p, size_t size); +void SzFreeTemp(ISzAllocPtr p, void *address); -#ifdef __cplusplus -} -#endif +EXTERN_C_END #endif diff --git a/src/third-party/lzma/7zArcIn.c b/src/third-party/lzma/7zArcIn.c index 080d8df2a..68cc12ff4 100644 --- a/src/third-party/lzma/7zArcIn.c +++ b/src/third-party/lzma/7zArcIn.c @@ -1,5 +1,5 @@ /* 7zArcIn.c -- 7z Input functions -2015-11-18 : Igor Pavlov : Public domain */ +2018-12-31 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -11,7 +11,7 @@ #include "CpuArch.h" #define MY_ALLOC(T, p, size, alloc) { \ - if ((p = (T *)IAlloc_Alloc(alloc, (size) * sizeof(T))) == NULL) return SZ_ERROR_MEM; } + if ((p = (T *)ISzAlloc_Alloc(alloc, (size) * sizeof(T))) == NULL) return SZ_ERROR_MEM; } #define MY_ALLOC_ZE(T, p, size, alloc) { if ((size) == 0) p = NULL; else MY_ALLOC(T, p, size, alloc) } @@ -19,7 +19,7 @@ { MY_ALLOC(Byte, to, size, alloc); memcpy(to, from, size); } #define MY_ALLOC_ZE_AND_CPY(to, size, from, alloc) \ - { if ((size) == 0) p = NULL; else { MY_ALLOC_AND_CPY(to, size, from, alloc) } } + { if ((size) == 0) to = NULL; else { MY_ALLOC_AND_CPY(to, size, from, alloc) } } #define k7zMajorVersion 0 @@ -60,7 +60,7 @@ const Byte k7zSignature[k7zSignatureSize] = {'7', 'z', 0xBC, 0xAF, 0x27, 0x1C}; #define SzBitUi32s_Init(p) { (p)->Defs = NULL; (p)->Vals = NULL; } -static SRes SzBitUi32s_Alloc(CSzBitUi32s *p, size_t num, ISzAlloc *alloc) +static SRes SzBitUi32s_Alloc(CSzBitUi32s *p, size_t num, ISzAllocPtr alloc) { if (num == 0) { @@ -75,18 +75,18 @@ static SRes SzBitUi32s_Alloc(CSzBitUi32s *p, size_t num, ISzAlloc *alloc) return SZ_OK; } -void SzBitUi32s_Free(CSzBitUi32s *p, ISzAlloc *alloc) +void SzBitUi32s_Free(CSzBitUi32s *p, ISzAllocPtr alloc) { - IAlloc_Free(alloc, p->Defs); p->Defs = NULL; - IAlloc_Free(alloc, p->Vals); p->Vals = NULL; + ISzAlloc_Free(alloc, p->Defs); p->Defs = NULL; + ISzAlloc_Free(alloc, p->Vals); p->Vals = NULL; } #define SzBitUi64s_Init(p) { (p)->Defs = NULL; (p)->Vals = NULL; } -void SzBitUi64s_Free(CSzBitUi64s *p, ISzAlloc *alloc) +void SzBitUi64s_Free(CSzBitUi64s *p, ISzAllocPtr alloc) { - IAlloc_Free(alloc, p->Defs); p->Defs = NULL; - IAlloc_Free(alloc, p->Vals); p->Vals = NULL; + ISzAlloc_Free(alloc, p->Defs); p->Defs = NULL; + ISzAlloc_Free(alloc, p->Vals); p->Vals = NULL; } @@ -107,18 +107,18 @@ static void SzAr_Init(CSzAr *p) p->CodersData = NULL; } -static void SzAr_Free(CSzAr *p, ISzAlloc *alloc) +static void SzAr_Free(CSzAr *p, ISzAllocPtr alloc) { - IAlloc_Free(alloc, p->PackPositions); + ISzAlloc_Free(alloc, p->PackPositions); SzBitUi32s_Free(&p->FolderCRCs, alloc); - IAlloc_Free(alloc, p->FoCodersOffsets); - IAlloc_Free(alloc, p->FoStartPackStreamIndex); - IAlloc_Free(alloc, p->FoToCoderUnpackSizes); - IAlloc_Free(alloc, p->FoToMainUnpackSizeIndex); - IAlloc_Free(alloc, p->CoderUnpackSizes); + ISzAlloc_Free(alloc, p->FoCodersOffsets); + ISzAlloc_Free(alloc, p->FoStartPackStreamIndex); + ISzAlloc_Free(alloc, p->FoToCoderUnpackSizes); + ISzAlloc_Free(alloc, p->FoToMainUnpackSizeIndex); + ISzAlloc_Free(alloc, p->CoderUnpackSizes); - IAlloc_Free(alloc, p->CodersData); + ISzAlloc_Free(alloc, p->CodersData); SzAr_Init(p); } @@ -147,16 +147,16 @@ void SzArEx_Init(CSzArEx *p) SzBitUi64s_Init(&p->CTime); } -void SzArEx_Free(CSzArEx *p, ISzAlloc *alloc) +void SzArEx_Free(CSzArEx *p, ISzAllocPtr alloc) { - IAlloc_Free(alloc, p->UnpackPositions); - IAlloc_Free(alloc, p->IsDirs); + ISzAlloc_Free(alloc, p->UnpackPositions); + ISzAlloc_Free(alloc, p->IsDirs); - IAlloc_Free(alloc, p->FolderToFile); - IAlloc_Free(alloc, p->FileToFolder); + ISzAlloc_Free(alloc, p->FolderToFile); + ISzAlloc_Free(alloc, p->FileToFolder); - IAlloc_Free(alloc, p->FileNameOffsets); - IAlloc_Free(alloc, p->FileNames); + ISzAlloc_Free(alloc, p->FileNameOffsets); + ISzAlloc_Free(alloc, p->FileNames); SzBitUi32s_Free(&p->CRCs, alloc); SzBitUi32s_Free(&p->Attribs, alloc); @@ -305,7 +305,7 @@ static UInt32 CountDefinedBits(const Byte *bits, UInt32 numItems) return sum; } -static MY_NO_INLINE SRes ReadBitVector(CSzData *sd, UInt32 numItems, Byte **v, ISzAlloc *alloc) +static MY_NO_INLINE SRes ReadBitVector(CSzData *sd, UInt32 numItems, Byte **v, ISzAllocPtr alloc) { Byte allAreDefined; Byte *v2; @@ -328,12 +328,12 @@ static MY_NO_INLINE SRes ReadBitVector(CSzData *sd, UInt32 numItems, Byte **v, I { unsigned numBits = (unsigned)numItems & 7; if (numBits != 0) - v2[numBytes - 1] = (Byte)((((UInt32)1 << numBits) - 1) << (8 - numBits)); + v2[(size_t)numBytes - 1] = (Byte)((((UInt32)1 << numBits) - 1) << (8 - numBits)); } return SZ_OK; } -static MY_NO_INLINE SRes ReadUi32s(CSzData *sd2, UInt32 numItems, CSzBitUi32s *crcs, ISzAlloc *alloc) +static MY_NO_INLINE SRes ReadUi32s(CSzData *sd2, UInt32 numItems, CSzBitUi32s *crcs, ISzAllocPtr alloc) { UInt32 i; CSzData sd; @@ -354,7 +354,7 @@ static MY_NO_INLINE SRes ReadUi32s(CSzData *sd2, UInt32 numItems, CSzBitUi32s *c return SZ_OK; } -static SRes ReadBitUi32s(CSzData *sd, UInt32 numItems, CSzBitUi32s *crcs, ISzAlloc *alloc) +static SRes ReadBitUi32s(CSzData *sd, UInt32 numItems, CSzBitUi32s *crcs, ISzAllocPtr alloc) { SzBitUi32s_Free(crcs, alloc); RINOK(ReadBitVector(sd, numItems, &crcs->Defs, alloc)); @@ -380,7 +380,7 @@ static SRes SkipBitUi32s(CSzData *sd, UInt32 numItems) return SZ_OK; } -static SRes ReadPackInfo(CSzAr *p, CSzData *sd, ISzAlloc *alloc) +static SRes ReadPackInfo(CSzAr *p, CSzData *sd, ISzAllocPtr alloc) { RINOK(SzReadNumber32(sd, &p->NumPackStreams)); @@ -635,7 +635,7 @@ static SRes ReadUnpackInfo(CSzAr *p, CSzData *sd2, UInt32 numFoldersMax, const CBuf *tempBufs, UInt32 numTempBufs, - ISzAlloc *alloc) + ISzAllocPtr alloc) { CSzData sd; @@ -666,7 +666,7 @@ static SRes ReadUnpackInfo(CSzAr *p, MY_ALLOC(size_t, p->FoCodersOffsets, (size_t)numFolders + 1, alloc); MY_ALLOC(UInt32, p->FoStartPackStreamIndex, (size_t)numFolders + 1, alloc); MY_ALLOC(UInt32, p->FoToCoderUnpackSizes, (size_t)numFolders + 1, alloc); - MY_ALLOC(Byte, p->FoToMainUnpackSizeIndex, (size_t)numFolders, alloc); + MY_ALLOC_ZE(Byte, p->FoToMainUnpackSizeIndex, (size_t)numFolders, alloc); startBufPtr = sd.Data; @@ -788,13 +788,9 @@ static SRes ReadUnpackInfo(CSzAr *p, numCodersOutStreams += numCoders; if (numCodersOutStreams < numCoders) return SZ_ERROR_UNSUPPORTED; - - packStreamIndex += numPackStreams; - if (packStreamIndex < numPackStreams) - return SZ_ERROR_UNSUPPORTED; - - if (packStreamIndex > p->NumPackStreams) + if (numPackStreams > p->NumPackStreams - packStreamIndex) return SZ_ERROR_ARCHIVE; + packStreamIndex += numPackStreams; } } @@ -938,7 +934,7 @@ static SRes SzReadStreamsInfo(CSzAr *p, UInt32 numFoldersMax, const CBuf *tempBufs, UInt32 numTempBufs, UInt64 *dataOffset, CSubStreamInfo *ssi, - ISzAlloc *alloc) + ISzAllocPtr alloc) { UInt64 type; @@ -980,7 +976,7 @@ static SRes SzReadAndDecodePackedStreams( UInt32 numFoldersMax, UInt64 baseOffset, CSzAr *p, - ISzAlloc *allocTemp) + ISzAllocPtr allocTemp) { UInt64 dataStartPos; UInt32 fo; @@ -1047,7 +1043,7 @@ static SRes SzReadFileNames(const Byte *data, size_t size, UInt32 numFiles, size static MY_NO_INLINE SRes ReadTime(CSzBitUi64s *p, UInt32 num, CSzData *sd2, const CBuf *tempBufs, UInt32 numTempBufs, - ISzAlloc *alloc) + ISzAllocPtr alloc) { CSzData sd; UInt32 i; @@ -1100,17 +1096,15 @@ static SRes SzReadHeader2( CSzData *sd, ILookInStream *inStream, CBuf *tempBufs, UInt32 *numTempBufs, - ISzAlloc *allocMain, - ISzAlloc *allocTemp + ISzAllocPtr allocMain, + ISzAllocPtr allocTemp ) { - UInt64 type; - UInt32 numFiles = 0; - UInt32 numEmptyStreams = 0; CSubStreamInfo ssi; - const Byte *emptyStreams = NULL; - const Byte *emptyFiles = NULL; +{ + UInt64 type; + SzData_Clear(&ssi.sdSizes); SzData_Clear(&ssi.sdCRCs); SzData_Clear(&ssi.sdNumSubStreams); @@ -1124,9 +1118,9 @@ static SRes SzReadHeader2( { for (;;) { - UInt64 type; - RINOK(ReadID(sd, &type)); - if (type == k7zIdEnd) + UInt64 type2; + RINOK(ReadID(sd, &type2)); + if (type2 == k7zIdEnd) break; RINOK(SkipData(sd)); } @@ -1164,6 +1158,13 @@ static SRes SzReadHeader2( if (type != k7zIdFilesInfo) return SZ_ERROR_ARCHIVE; +} + +{ + UInt32 numFiles = 0; + UInt32 numEmptyStreams = 0; + const Byte *emptyStreams = NULL; + const Byte *emptyFiles = NULL; RINOK(SzReadNumber32(sd, &numFiles)); p->NumFiles = numFiles; @@ -1462,7 +1463,7 @@ static SRes SzReadHeader2( if (ssi.sdNumSubStreams.Data && ssi.sdNumSubStreams.Size != 0) return SZ_ERROR_ARCHIVE; } - +} return SZ_OK; } @@ -1471,8 +1472,8 @@ static SRes SzReadHeader( CSzArEx *p, CSzData *sd, ILookInStream *inStream, - ISzAlloc *allocMain, - ISzAlloc *allocTemp) + ISzAllocPtr allocMain, + ISzAllocPtr allocTemp) { UInt32 i; UInt32 numTempBufs = 0; @@ -1500,8 +1501,8 @@ static SRes SzReadHeader( static SRes SzArEx_Open2( CSzArEx *p, ILookInStream *inStream, - ISzAlloc *allocMain, - ISzAlloc *allocTemp) + ISzAllocPtr allocMain, + ISzAllocPtr allocTemp) { Byte header[k7zStartHeaderSize]; Int64 startArcPos; @@ -1512,7 +1513,7 @@ static SRes SzArEx_Open2( SRes res; startArcPos = 0; - RINOK(inStream->Seek(inStream, &startArcPos, SZ_SEEK_CUR)); + RINOK(ILookInStream_Seek(inStream, &startArcPos, SZ_SEEK_CUR)); RINOK(LookInStream_Read2(inStream, header, k7zStartHeaderSize, SZ_ERROR_NO_ARCHIVE)); @@ -1541,7 +1542,7 @@ static SRes SzArEx_Open2( { Int64 pos = 0; - RINOK(inStream->Seek(inStream, &pos, SZ_SEEK_END)); + RINOK(ILookInStream_Seek(inStream, &pos, SZ_SEEK_END)); if ((UInt64)pos < startArcPos + nextHeaderOffset || (UInt64)pos < startArcPos + k7zStartHeaderSize + nextHeaderOffset || (UInt64)pos < startArcPos + k7zStartHeaderSize + nextHeaderOffset + nextHeaderSize) @@ -1622,7 +1623,7 @@ static SRes SzArEx_Open2( SRes SzArEx_Open(CSzArEx *p, ILookInStream *inStream, - ISzAlloc *allocMain, ISzAlloc *allocTemp) + ISzAllocPtr allocMain, ISzAllocPtr allocTemp) { SRes res = SzArEx_Open2(p, inStream, allocMain, allocTemp); if (res != SZ_OK) @@ -1640,8 +1641,8 @@ SRes SzArEx_Extract( size_t *outBufferSize, size_t *offset, size_t *outSizeProcessed, - ISzAlloc *allocMain, - ISzAlloc *allocTemp) + ISzAllocPtr allocMain, + ISzAllocPtr allocTemp) { UInt32 folderIndex = p->FileToFolder[fileIndex]; SRes res = SZ_OK; @@ -1651,7 +1652,7 @@ SRes SzArEx_Extract( if (folderIndex == (UInt32)-1) { - IAlloc_Free(allocMain, *tempBuf); + ISzAlloc_Free(allocMain, *tempBuf); *blockIndex = folderIndex; *tempBuf = NULL; *outBufferSize = 0; @@ -1663,7 +1664,7 @@ SRes SzArEx_Extract( UInt64 unpackSizeSpec = SzAr_GetFolderUnpackSize(&p->db, folderIndex); /* UInt64 unpackSizeSpec = - p->UnpackPositions[p->FolderToFile[folderIndex + 1]] - + p->UnpackPositions[p->FolderToFile[(size_t)folderIndex + 1]] - p->UnpackPositions[p->FolderToFile[folderIndex]]; */ size_t unpackSize = (size_t)unpackSizeSpec; @@ -1671,7 +1672,7 @@ SRes SzArEx_Extract( if (unpackSize != unpackSizeSpec) return SZ_ERROR_MEM; *blockIndex = folderIndex; - IAlloc_Free(allocMain, *tempBuf); + ISzAlloc_Free(allocMain, *tempBuf); *tempBuf = NULL; if (res == SZ_OK) @@ -1679,7 +1680,7 @@ SRes SzArEx_Extract( *outBufferSize = unpackSize; if (unpackSize != 0) { - *tempBuf = (Byte *)IAlloc_Alloc(allocMain, unpackSize); + *tempBuf = (Byte *)ISzAlloc_Alloc(allocMain, unpackSize); if (*tempBuf == NULL) res = SZ_ERROR_MEM; } @@ -1696,7 +1697,7 @@ SRes SzArEx_Extract( { UInt64 unpackPos = p->UnpackPositions[fileIndex]; *offset = (size_t)(unpackPos - p->UnpackPositions[p->FolderToFile[folderIndex]]); - *outSizeProcessed = (size_t)(p->UnpackPositions[fileIndex + 1] - unpackPos); + *outSizeProcessed = (size_t)(p->UnpackPositions[(size_t)fileIndex + 1] - unpackPos); if (*offset + *outSizeProcessed > *outBufferSize) return SZ_ERROR_FAIL; if (SzBitWithVals_Check(&p->CRCs, fileIndex)) @@ -1743,7 +1744,7 @@ size_t SzArEx_GetFullNameLen(const CSzArEx *p, size_t fileIndex) UInt16 *SzArEx_GetFullNameUtf16_Back(const CSzArEx *p, size_t fileIndex, UInt16 *dest) { - Bool needSlash; + BoolInt needSlash; if (!p->FileNameOffsets) { *(--dest) = 0; diff --git a/src/third-party/lzma/7zBuf.c b/src/third-party/lzma/7zBuf.c index b0ac1104b..438bba68b 100644 --- a/src/third-party/lzma/7zBuf.c +++ b/src/third-party/lzma/7zBuf.c @@ -1,5 +1,5 @@ /* 7zBuf.c -- Byte Buffer -2013-01-21 : Igor Pavlov : Public domain */ +2017-04-03 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -11,7 +11,7 @@ void Buf_Init(CBuf *p) p->size = 0; } -int Buf_Create(CBuf *p, size_t size, ISzAlloc *alloc) +int Buf_Create(CBuf *p, size_t size, ISzAllocPtr alloc) { p->size = 0; if (size == 0) @@ -19,8 +19,8 @@ int Buf_Create(CBuf *p, size_t size, ISzAlloc *alloc) p->data = 0; return 1; } - p->data = (Byte *)alloc->Alloc(alloc, size); - if (p->data != 0) + p->data = (Byte *)ISzAlloc_Alloc(alloc, size); + if (p->data) { p->size = size; return 1; @@ -28,9 +28,9 @@ int Buf_Create(CBuf *p, size_t size, ISzAlloc *alloc) return 0; } -void Buf_Free(CBuf *p, ISzAlloc *alloc) +void Buf_Free(CBuf *p, ISzAllocPtr alloc) { - alloc->Free(alloc, p->data); + ISzAlloc_Free(alloc, p->data); p->data = 0; p->size = 0; } diff --git a/src/third-party/lzma/7zBuf.h b/src/third-party/lzma/7zBuf.h index e5f9218be..5942d6e62 100644 --- a/src/third-party/lzma/7zBuf.h +++ b/src/third-party/lzma/7zBuf.h @@ -1,5 +1,5 @@ /* 7zBuf.h -- Byte Buffer -2013-01-18 : Igor Pavlov : Public domain */ +2017-04-03 : Igor Pavlov : Public domain */ #ifndef __7Z_BUF_H #define __7Z_BUF_H @@ -15,8 +15,8 @@ typedef struct } CBuf; void Buf_Init(CBuf *p); -int Buf_Create(CBuf *p, size_t size, ISzAlloc *alloc); -void Buf_Free(CBuf *p, ISzAlloc *alloc); +int Buf_Create(CBuf *p, size_t size, ISzAllocPtr alloc); +void Buf_Free(CBuf *p, ISzAllocPtr alloc); typedef struct { @@ -27,8 +27,8 @@ typedef struct void DynBuf_Construct(CDynBuf *p); void DynBuf_SeekToBeg(CDynBuf *p); -int DynBuf_Write(CDynBuf *p, const Byte *buf, size_t size, ISzAlloc *alloc); -void DynBuf_Free(CDynBuf *p, ISzAlloc *alloc); +int DynBuf_Write(CDynBuf *p, const Byte *buf, size_t size, ISzAllocPtr alloc); +void DynBuf_Free(CDynBuf *p, ISzAllocPtr alloc); EXTERN_C_END diff --git a/src/third-party/lzma/7zBuf2.c b/src/third-party/lzma/7zBuf2.c index 8e3fe9297..49b4343b6 100644 --- a/src/third-party/lzma/7zBuf2.c +++ b/src/third-party/lzma/7zBuf2.c @@ -1,5 +1,5 @@ /* 7zBuf2.c -- Byte Buffer -2014-08-22 : Igor Pavlov : Public domain */ +2017-04-03 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -19,19 +19,20 @@ void DynBuf_SeekToBeg(CDynBuf *p) p->pos = 0; } -int DynBuf_Write(CDynBuf *p, const Byte *buf, size_t size, ISzAlloc *alloc) +int DynBuf_Write(CDynBuf *p, const Byte *buf, size_t size, ISzAllocPtr alloc) { if (size > p->size - p->pos) { size_t newSize = p->pos + size; Byte *data; newSize += newSize / 4; - data = (Byte *)alloc->Alloc(alloc, newSize); - if (data == 0) + data = (Byte *)ISzAlloc_Alloc(alloc, newSize); + if (!data) return 0; p->size = newSize; - memcpy(data, p->data, p->pos); - alloc->Free(alloc, p->data); + if (p->pos != 0) + memcpy(data, p->data, p->pos); + ISzAlloc_Free(alloc, p->data); p->data = data; } if (size != 0) @@ -42,9 +43,9 @@ int DynBuf_Write(CDynBuf *p, const Byte *buf, size_t size, ISzAlloc *alloc) return 1; } -void DynBuf_Free(CDynBuf *p, ISzAlloc *alloc) +void DynBuf_Free(CDynBuf *p, ISzAllocPtr alloc) { - alloc->Free(alloc, p->data); + ISzAlloc_Free(alloc, p->data); p->data = 0; p->size = 0; p->pos = 0; diff --git a/src/third-party/lzma/7zCrc.c b/src/third-party/lzma/7zCrc.c index 607db3404..40ab75952 100644 --- a/src/third-party/lzma/7zCrc.c +++ b/src/third-party/lzma/7zCrc.c @@ -1,5 +1,5 @@ /* 7zCrc.c -- CRC32 init -2015-03-10 : Igor Pavlov : Public domain */ +2017-06-06 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -61,12 +61,12 @@ void MY_FAST_CALL CrcGenerateTable() UInt32 r = i; unsigned j; for (j = 0; j < 8; j++) - r = (r >> 1) ^ (kCrcPoly & ~((r & 1) - 1)); + r = (r >> 1) ^ (kCrcPoly & ((UInt32)0 - (r & 1))); g_CrcTable[i] = r; } - for (; i < 256 * CRC_NUM_TABLES; i++) + for (i = 256; i < 256 * CRC_NUM_TABLES; i++) { - UInt32 r = g_CrcTable[i - 256]; + UInt32 r = g_CrcTable[(size_t)i - 256]; g_CrcTable[i] = g_CrcTable[r & 0xFF] ^ (r >> 8); } @@ -86,8 +86,8 @@ void MY_FAST_CALL CrcGenerateTable() #ifdef MY_CPU_X86_OR_AMD64 if (!CPU_Is_InOrder()) - g_CrcUpdate = CrcUpdateT8; #endif + g_CrcUpdate = CrcUpdateT8; #endif #else @@ -101,7 +101,7 @@ void MY_FAST_CALL CrcGenerateTable() g_CrcUpdate = CrcUpdateT4; #if CRC_NUM_TABLES >= 8 g_CrcUpdateT8 = CrcUpdateT8; - // g_CrcUpdate = CrcUpdateT8; + g_CrcUpdate = CrcUpdateT8; #endif } else if (p[0] != 1 || p[1] != 2) @@ -111,14 +111,14 @@ void MY_FAST_CALL CrcGenerateTable() { for (i = 256 * CRC_NUM_TABLES - 1; i >= 256; i--) { - UInt32 x = g_CrcTable[i - 256]; + UInt32 x = g_CrcTable[(size_t)i - 256]; g_CrcTable[i] = CRC_UINT32_SWAP(x); } g_CrcUpdateT4 = CrcUpdateT1_BeT4; g_CrcUpdate = CrcUpdateT1_BeT4; #if CRC_NUM_TABLES >= 8 g_CrcUpdateT8 = CrcUpdateT1_BeT8; - // g_CrcUpdate = CrcUpdateT1_BeT8; + g_CrcUpdate = CrcUpdateT1_BeT8; #endif } } diff --git a/src/third-party/lzma/7zCrcOpt.c b/src/third-party/lzma/7zCrcOpt.c index 58628efe5..2ee0de845 100644 --- a/src/third-party/lzma/7zCrcOpt.c +++ b/src/third-party/lzma/7zCrcOpt.c @@ -1,5 +1,5 @@ /* 7zCrcOpt.c -- CRC32 calculation -2015-03-01 : Igor Pavlov : Public domain */ +2017-04-03 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -18,10 +18,10 @@ UInt32 MY_FAST_CALL CrcUpdateT4(UInt32 v, const void *data, size_t size, const U { v ^= *(const UInt32 *)p; v = - table[0x300 + ((v ) & 0xFF)] - ^ table[0x200 + ((v >> 8) & 0xFF)] - ^ table[0x100 + ((v >> 16) & 0xFF)] - ^ table[0x000 + ((v >> 24))]; + (table + 0x300)[((v ) & 0xFF)] + ^ (table + 0x200)[((v >> 8) & 0xFF)] + ^ (table + 0x100)[((v >> 16) & 0xFF)] + ^ (table + 0x000)[((v >> 24))]; } for (; size > 0; size--, p++) v = CRC_UPDATE_BYTE_2(v, *p); @@ -38,16 +38,16 @@ UInt32 MY_FAST_CALL CrcUpdateT8(UInt32 v, const void *data, size_t size, const U UInt32 d; v ^= *(const UInt32 *)p; v = - table[0x700 + ((v ) & 0xFF)] - ^ table[0x600 + ((v >> 8) & 0xFF)] - ^ table[0x500 + ((v >> 16) & 0xFF)] - ^ table[0x400 + ((v >> 24))]; + (table + 0x700)[((v ) & 0xFF)] + ^ (table + 0x600)[((v >> 8) & 0xFF)] + ^ (table + 0x500)[((v >> 16) & 0xFF)] + ^ (table + 0x400)[((v >> 24))]; d = *((const UInt32 *)p + 1); v ^= - table[0x300 + ((d ) & 0xFF)] - ^ table[0x200 + ((d >> 8) & 0xFF)] - ^ table[0x100 + ((d >> 16) & 0xFF)] - ^ table[0x000 + ((d >> 24))]; + (table + 0x300)[((d ) & 0xFF)] + ^ (table + 0x200)[((d >> 8) & 0xFF)] + ^ (table + 0x100)[((d >> 16) & 0xFF)] + ^ (table + 0x000)[((d >> 24))]; } for (; size > 0; size--, p++) v = CRC_UPDATE_BYTE_2(v, *p); @@ -74,10 +74,10 @@ UInt32 MY_FAST_CALL CrcUpdateT1_BeT4(UInt32 v, const void *data, size_t size, co { v ^= *(const UInt32 *)p; v = - table[0x000 + ((v ) & 0xFF)] - ^ table[0x100 + ((v >> 8) & 0xFF)] - ^ table[0x200 + ((v >> 16) & 0xFF)] - ^ table[0x300 + ((v >> 24))]; + (table + 0x000)[((v ) & 0xFF)] + ^ (table + 0x100)[((v >> 8) & 0xFF)] + ^ (table + 0x200)[((v >> 16) & 0xFF)] + ^ (table + 0x300)[((v >> 24))]; } for (; size > 0; size--, p++) v = CRC_UPDATE_BYTE_2_BE(v, *p); @@ -96,16 +96,16 @@ UInt32 MY_FAST_CALL CrcUpdateT1_BeT8(UInt32 v, const void *data, size_t size, co UInt32 d; v ^= *(const UInt32 *)p; v = - table[0x400 + ((v ) & 0xFF)] - ^ table[0x500 + ((v >> 8) & 0xFF)] - ^ table[0x600 + ((v >> 16) & 0xFF)] - ^ table[0x700 + ((v >> 24))]; + (table + 0x400)[((v ) & 0xFF)] + ^ (table + 0x500)[((v >> 8) & 0xFF)] + ^ (table + 0x600)[((v >> 16) & 0xFF)] + ^ (table + 0x700)[((v >> 24))]; d = *((const UInt32 *)p + 1); v ^= - table[0x000 + ((d ) & 0xFF)] - ^ table[0x100 + ((d >> 8) & 0xFF)] - ^ table[0x200 + ((d >> 16) & 0xFF)] - ^ table[0x300 + ((d >> 24))]; + (table + 0x000)[((d ) & 0xFF)] + ^ (table + 0x100)[((d >> 8) & 0xFF)] + ^ (table + 0x200)[((d >> 16) & 0xFF)] + ^ (table + 0x300)[((d >> 24))]; } for (; size > 0; size--, p++) v = CRC_UPDATE_BYTE_2_BE(v, *p); diff --git a/src/third-party/lzma/7zDec.c b/src/third-party/lzma/7zDec.c index e39b4ff12..2a7b09030 100644 --- a/src/third-party/lzma/7zDec.c +++ b/src/third-party/lzma/7zDec.c @@ -1,5 +1,5 @@ /* 7zDec.c -- Decoding from 7z folder -2015-11-18 : Igor Pavlov : Public domain */ +2019-02-02 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -39,28 +39,28 @@ typedef struct { - IByteIn p; + IByteIn vt; const Byte *cur; const Byte *end; const Byte *begin; UInt64 processed; - Bool extra; + BoolInt extra; SRes res; - ILookInStream *inStream; + const ILookInStream *inStream; } CByteInToLook; -static Byte ReadByte(void *pp) +static Byte ReadByte(const IByteIn *pp) { - CByteInToLook *p = (CByteInToLook *)pp; + CByteInToLook *p = CONTAINER_FROM_VTBL(pp, CByteInToLook, vt); if (p->cur != p->end) return *p->cur++; if (p->res == SZ_OK) { size_t size = p->cur - p->begin; p->processed += size; - p->res = p->inStream->Skip(p->inStream, size); + p->res = ILookInStream_Skip(p->inStream, size); size = (1 << 25); - p->res = p->inStream->Look(p->inStream, (const void **)&p->begin, &size); + p->res = ILookInStream_Look(p->inStream, (const void **)&p->begin, &size); p->cur = p->begin; p->end = p->begin + size; if (size != 0) @@ -70,14 +70,14 @@ static Byte ReadByte(void *pp) return 0; } -static SRes SzDecodePpmd(const Byte *props, unsigned propsSize, UInt64 inSize, ILookInStream *inStream, - Byte *outBuffer, SizeT outSize, ISzAlloc *allocMain) +static SRes SzDecodePpmd(const Byte *props, unsigned propsSize, UInt64 inSize, const ILookInStream *inStream, + Byte *outBuffer, SizeT outSize, ISzAllocPtr allocMain) { CPpmd7 ppmd; CByteInToLook s; SRes res = SZ_OK; - s.p.Read = ReadByte; + s.vt.Read = ReadByte; s.inStream = inStream; s.begin = s.end = s.cur = NULL; s.extra = False; @@ -103,7 +103,7 @@ static SRes SzDecodePpmd(const Byte *props, unsigned propsSize, UInt64 inSize, I { CPpmd7z_RangeDec rc; Ppmd7z_RangeDec_CreateVTable(&rc); - rc.Stream = &s.p; + rc.Stream = &s.vt; if (!Ppmd7z_RangeDec_Init(&rc)) res = SZ_ERROR_DATA; else if (s.extra) @@ -113,7 +113,7 @@ static SRes SzDecodePpmd(const Byte *props, unsigned propsSize, UInt64 inSize, I SizeT i; for (i = 0; i < outSize; i++) { - int sym = Ppmd7_DecodeSymbol(&ppmd, &rc.p); + int sym = Ppmd7_DecodeSymbol(&ppmd, &rc.vt); if (s.extra || sym < 0) break; outBuffer[i] = (Byte)sym; @@ -132,7 +132,7 @@ static SRes SzDecodePpmd(const Byte *props, unsigned propsSize, UInt64 inSize, I static SRes SzDecodeLzma(const Byte *props, unsigned propsSize, UInt64 inSize, ILookInStream *inStream, - Byte *outBuffer, SizeT outSize, ISzAlloc *allocMain) + Byte *outBuffer, SizeT outSize, ISzAllocPtr allocMain) { CLzmaDec state; SRes res = SZ_OK; @@ -149,14 +149,14 @@ static SRes SzDecodeLzma(const Byte *props, unsigned propsSize, UInt64 inSize, I size_t lookahead = (1 << 18); if (lookahead > inSize) lookahead = (size_t)inSize; - res = inStream->Look(inStream, &inBuf, &lookahead); + res = ILookInStream_Look(inStream, &inBuf, &lookahead); if (res != SZ_OK) break; { SizeT inProcessed = (SizeT)lookahead, dicPos = state.dicPos; ELzmaStatus status; - res = LzmaDec_DecodeToDic(&state, outSize, inBuf, &inProcessed, LZMA_FINISH_END, &status); + res = LzmaDec_DecodeToDic(&state, outSize, (const Byte *)inBuf, &inProcessed, LZMA_FINISH_END, &status); lookahead -= inProcessed; inSize -= inProcessed; if (res != SZ_OK) @@ -178,7 +178,7 @@ static SRes SzDecodeLzma(const Byte *props, unsigned propsSize, UInt64 inSize, I break; } - res = inStream->Skip((void *)inStream, inProcessed); + res = ILookInStream_Skip(inStream, inProcessed); if (res != SZ_OK) break; } @@ -192,7 +192,7 @@ static SRes SzDecodeLzma(const Byte *props, unsigned propsSize, UInt64 inSize, I #ifndef _7Z_NO_METHOD_LZMA2 static SRes SzDecodeLzma2(const Byte *props, unsigned propsSize, UInt64 inSize, ILookInStream *inStream, - Byte *outBuffer, SizeT outSize, ISzAlloc *allocMain) + Byte *outBuffer, SizeT outSize, ISzAllocPtr allocMain) { CLzma2Dec state; SRes res = SZ_OK; @@ -211,14 +211,14 @@ static SRes SzDecodeLzma2(const Byte *props, unsigned propsSize, UInt64 inSize, size_t lookahead = (1 << 18); if (lookahead > inSize) lookahead = (size_t)inSize; - res = inStream->Look(inStream, &inBuf, &lookahead); + res = ILookInStream_Look(inStream, &inBuf, &lookahead); if (res != SZ_OK) break; { SizeT inProcessed = (SizeT)lookahead, dicPos = state.decoder.dicPos; ELzmaStatus status; - res = Lzma2Dec_DecodeToDic(&state, outSize, inBuf, &inProcessed, LZMA_FINISH_END, &status); + res = Lzma2Dec_DecodeToDic(&state, outSize, (const Byte *)inBuf, &inProcessed, LZMA_FINISH_END, &status); lookahead -= inProcessed; inSize -= inProcessed; if (res != SZ_OK) @@ -237,7 +237,7 @@ static SRes SzDecodeLzma2(const Byte *props, unsigned propsSize, UInt64 inSize, break; } - res = inStream->Skip((void *)inStream, inProcessed); + res = ILookInStream_Skip(inStream, inProcessed); if (res != SZ_OK) break; } @@ -258,18 +258,18 @@ static SRes SzDecodeCopy(UInt64 inSize, ILookInStream *inStream, Byte *outBuffer size_t curSize = (1 << 18); if (curSize > inSize) curSize = (size_t)inSize; - RINOK(inStream->Look(inStream, &inBuf, &curSize)); + RINOK(ILookInStream_Look(inStream, &inBuf, &curSize)); if (curSize == 0) return SZ_ERROR_INPUT_EOF; memcpy(outBuffer, inBuf, curSize); outBuffer += curSize; inSize -= curSize; - RINOK(inStream->Skip((void *)inStream, curSize)); + RINOK(ILookInStream_Skip(inStream, curSize)); } return SZ_OK; } -static Bool IS_MAIN_METHOD(UInt32 m) +static BoolInt IS_MAIN_METHOD(UInt32 m) { switch (m) { @@ -286,7 +286,7 @@ static Bool IS_MAIN_METHOD(UInt32 m) return False; } -static Bool IS_SUPPORTED_CODER(const CSzCoderInfo *c) +static BoolInt IS_SUPPORTED_CODER(const CSzCoderInfo *c) { return c->NumStreams == 1 @@ -372,7 +372,7 @@ static SRes SzFolder_Decode2(const CSzFolder *folder, const UInt64 *unpackSizes, const UInt64 *packPositions, ILookInStream *inStream, UInt64 startPos, - Byte *outBuffer, SizeT outSize, ISzAlloc *allocMain, + Byte *outBuffer, SizeT outSize, ISzAllocPtr allocMain, Byte *tempBuf[]) { UInt32 ci; @@ -404,7 +404,7 @@ static SRes SzFolder_Decode2(const CSzFolder *folder, outSizeCur = (SizeT)unpackSize; if (outSizeCur != unpackSize) return SZ_ERROR_MEM; - temp = (Byte *)IAlloc_Alloc(allocMain, outSizeCur); + temp = (Byte *)ISzAlloc_Alloc(allocMain, outSizeCur); if (!temp && outSizeCur != 0) return SZ_ERROR_MEM; outBufCur = tempBuf[1 - ci] = temp; @@ -421,7 +421,7 @@ static SRes SzFolder_Decode2(const CSzFolder *folder, return SZ_ERROR_UNSUPPORTED; } offset = packPositions[si]; - inSize = packPositions[si + 1] - offset; + inSize = packPositions[(size_t)si + 1] - offset; RINOK(LookInStream_SeekTo(inStream, startPos + offset)); if (coder->MethodID == k_Copy) @@ -460,7 +460,7 @@ static SRes SzFolder_Decode2(const CSzFolder *folder, tempSizes[2] = (SizeT)s3Size; if (tempSizes[2] != s3Size) return SZ_ERROR_MEM; - tempBuf[2] = (Byte *)IAlloc_Alloc(allocMain, tempSizes[2]); + tempBuf[2] = (Byte *)ISzAlloc_Alloc(allocMain, tempSizes[2]); if (!tempBuf[2] && tempSizes[2] != 0) return SZ_ERROR_MEM; @@ -549,7 +549,7 @@ static SRes SzFolder_Decode2(const CSzFolder *folder, SRes SzAr_DecodeFolder(const CSzAr *p, UInt32 folderIndex, ILookInStream *inStream, UInt64 startPos, Byte *outBuffer, size_t outSize, - ISzAlloc *allocMain) + ISzAllocPtr allocMain) { SRes res; CSzFolder folder; @@ -557,7 +557,7 @@ SRes SzAr_DecodeFolder(const CSzAr *p, UInt32 folderIndex, const Byte *data = p->CodersData + p->FoCodersOffsets[folderIndex]; sd.Data = data; - sd.Size = p->FoCodersOffsets[folderIndex + 1] - p->FoCodersOffsets[folderIndex]; + sd.Size = p->FoCodersOffsets[(size_t)folderIndex + 1] - p->FoCodersOffsets[folderIndex]; res = SzGetNextFolderItem(&folder, &sd); @@ -579,7 +579,7 @@ SRes SzAr_DecodeFolder(const CSzAr *p, UInt32 folderIndex, outBuffer, (SizeT)outSize, allocMain, tempBuf); for (i = 0; i < 3; i++) - IAlloc_Free(allocMain, tempBuf[i]); + ISzAlloc_Free(allocMain, tempBuf[i]); if (res == SZ_OK) if (SzBitWithVals_Check(&p->FolderCRCs, folderIndex)) diff --git a/src/third-party/lzma/7zFile.c b/src/third-party/lzma/7zFile.c index 98fe7164f..e486901e3 100644 --- a/src/third-party/lzma/7zFile.c +++ b/src/third-party/lzma/7zFile.c @@ -1,5 +1,5 @@ /* 7zFile.c -- File IO -2009-11-24 : Igor Pavlov : Public domain */ +2017-04-03 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -238,49 +238,49 @@ WRes File_GetLength(CSzFile *p, UInt64 *length) /* ---------- FileSeqInStream ---------- */ -static SRes FileSeqInStream_Read(void *pp, void *buf, size_t *size) +static SRes FileSeqInStream_Read(const ISeqInStream *pp, void *buf, size_t *size) { - CFileSeqInStream *p = (CFileSeqInStream *)pp; + CFileSeqInStream *p = CONTAINER_FROM_VTBL(pp, CFileSeqInStream, vt); return File_Read(&p->file, buf, size) == 0 ? SZ_OK : SZ_ERROR_READ; } void FileSeqInStream_CreateVTable(CFileSeqInStream *p) { - p->s.Read = FileSeqInStream_Read; + p->vt.Read = FileSeqInStream_Read; } /* ---------- FileInStream ---------- */ -static SRes FileInStream_Read(void *pp, void *buf, size_t *size) +static SRes FileInStream_Read(const ISeekInStream *pp, void *buf, size_t *size) { - CFileInStream *p = (CFileInStream *)pp; + CFileInStream *p = CONTAINER_FROM_VTBL(pp, CFileInStream, vt); return (File_Read(&p->file, buf, size) == 0) ? SZ_OK : SZ_ERROR_READ; } -static SRes FileInStream_Seek(void *pp, Int64 *pos, ESzSeek origin) +static SRes FileInStream_Seek(const ISeekInStream *pp, Int64 *pos, ESzSeek origin) { - CFileInStream *p = (CFileInStream *)pp; + CFileInStream *p = CONTAINER_FROM_VTBL(pp, CFileInStream, vt); return File_Seek(&p->file, pos, origin); } void FileInStream_CreateVTable(CFileInStream *p) { - p->s.Read = FileInStream_Read; - p->s.Seek = FileInStream_Seek; + p->vt.Read = FileInStream_Read; + p->vt.Seek = FileInStream_Seek; } /* ---------- FileOutStream ---------- */ -static size_t FileOutStream_Write(void *pp, const void *data, size_t size) +static size_t FileOutStream_Write(const ISeqOutStream *pp, const void *data, size_t size) { - CFileOutStream *p = (CFileOutStream *)pp; + CFileOutStream *p = CONTAINER_FROM_VTBL(pp, CFileOutStream, vt); File_Write(&p->file, data, &size); return size; } void FileOutStream_CreateVTable(CFileOutStream *p) { - p->s.Write = FileOutStream_Write; + p->vt.Write = FileOutStream_Write; } diff --git a/src/third-party/lzma/7zFile.h b/src/third-party/lzma/7zFile.h index d62a19260..7e263bea1 100644 --- a/src/third-party/lzma/7zFile.h +++ b/src/third-party/lzma/7zFile.h @@ -1,5 +1,5 @@ /* 7zFile.h -- File IO -2013-01-18 : Igor Pavlov : Public domain */ +2017-04-03 : Igor Pavlov : Public domain */ #ifndef __7Z_FILE_H #define __7Z_FILE_H @@ -54,7 +54,7 @@ WRes File_GetLength(CSzFile *p, UInt64 *length); typedef struct { - ISeqInStream s; + ISeqInStream vt; CSzFile file; } CFileSeqInStream; @@ -63,7 +63,7 @@ void FileSeqInStream_CreateVTable(CFileSeqInStream *p); typedef struct { - ISeekInStream s; + ISeekInStream vt; CSzFile file; } CFileInStream; @@ -72,7 +72,7 @@ void FileInStream_CreateVTable(CFileInStream *p); typedef struct { - ISeqOutStream s; + ISeqOutStream vt; CSzFile file; } CFileOutStream; diff --git a/src/third-party/lzma/7zStream.c b/src/third-party/lzma/7zStream.c index 5a92d532c..579741fad 100644 --- a/src/third-party/lzma/7zStream.c +++ b/src/third-party/lzma/7zStream.c @@ -1,5 +1,5 @@ /* 7zStream.c -- 7z Stream functions -2013-11-12 : Igor Pavlov : Public domain */ +2017-04-03 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -7,12 +7,12 @@ #include "7zTypes.h" -SRes SeqInStream_Read2(ISeqInStream *stream, void *buf, size_t size, SRes errorType) +SRes SeqInStream_Read2(const ISeqInStream *stream, void *buf, size_t size, SRes errorType) { while (size != 0) { size_t processed = size; - RINOK(stream->Read(stream, buf, &processed)); + RINOK(ISeqInStream_Read(stream, buf, &processed)); if (processed == 0) return errorType; buf = (void *)((Byte *)buf + processed); @@ -21,40 +21,42 @@ SRes SeqInStream_Read2(ISeqInStream *stream, void *buf, size_t size, SRes errorT return SZ_OK; } -SRes SeqInStream_Read(ISeqInStream *stream, void *buf, size_t size) +SRes SeqInStream_Read(const ISeqInStream *stream, void *buf, size_t size) { return SeqInStream_Read2(stream, buf, size, SZ_ERROR_INPUT_EOF); } -SRes SeqInStream_ReadByte(ISeqInStream *stream, Byte *buf) +SRes SeqInStream_ReadByte(const ISeqInStream *stream, Byte *buf) { size_t processed = 1; - RINOK(stream->Read(stream, buf, &processed)); + RINOK(ISeqInStream_Read(stream, buf, &processed)); return (processed == 1) ? SZ_OK : SZ_ERROR_INPUT_EOF; } -SRes LookInStream_SeekTo(ILookInStream *stream, UInt64 offset) + + +SRes LookInStream_SeekTo(const ILookInStream *stream, UInt64 offset) { Int64 t = offset; - return stream->Seek(stream, &t, SZ_SEEK_SET); + return ILookInStream_Seek(stream, &t, SZ_SEEK_SET); } -SRes LookInStream_LookRead(ILookInStream *stream, void *buf, size_t *size) +SRes LookInStream_LookRead(const ILookInStream *stream, void *buf, size_t *size) { const void *lookBuf; if (*size == 0) return SZ_OK; - RINOK(stream->Look(stream, &lookBuf, size)); + RINOK(ILookInStream_Look(stream, &lookBuf, size)); memcpy(buf, lookBuf, *size); - return stream->Skip(stream, *size); + return ILookInStream_Skip(stream, *size); } -SRes LookInStream_Read2(ILookInStream *stream, void *buf, size_t size, SRes errorType) +SRes LookInStream_Read2(const ILookInStream *stream, void *buf, size_t size, SRes errorType) { while (size != 0) { size_t processed = size; - RINOK(stream->Read(stream, buf, &processed)); + RINOK(ILookInStream_Read(stream, buf, &processed)); if (processed == 0) return errorType; buf = (void *)((Byte *)buf + processed); @@ -63,61 +65,67 @@ SRes LookInStream_Read2(ILookInStream *stream, void *buf, size_t size, SRes erro return SZ_OK; } -SRes LookInStream_Read(ILookInStream *stream, void *buf, size_t size) +SRes LookInStream_Read(const ILookInStream *stream, void *buf, size_t size) { return LookInStream_Read2(stream, buf, size, SZ_ERROR_INPUT_EOF); } -static SRes LookToRead_Look_Lookahead(void *pp, const void **buf, size_t *size) + + +#define GET_LookToRead2 CLookToRead2 *p = CONTAINER_FROM_VTBL(pp, CLookToRead2, vt); + +static SRes LookToRead2_Look_Lookahead(const ILookInStream *pp, const void **buf, size_t *size) { SRes res = SZ_OK; - CLookToRead *p = (CLookToRead *)pp; + GET_LookToRead2 size_t size2 = p->size - p->pos; - if (size2 == 0 && *size > 0) + if (size2 == 0 && *size != 0) { p->pos = 0; - size2 = LookToRead_BUF_SIZE; - res = p->realStream->Read(p->realStream, p->buf, &size2); + p->size = 0; + size2 = p->bufSize; + res = ISeekInStream_Read(p->realStream, p->buf, &size2); p->size = size2; } - if (size2 < *size) + if (*size > size2) *size = size2; *buf = p->buf + p->pos; return res; } -static SRes LookToRead_Look_Exact(void *pp, const void **buf, size_t *size) +static SRes LookToRead2_Look_Exact(const ILookInStream *pp, const void **buf, size_t *size) { SRes res = SZ_OK; - CLookToRead *p = (CLookToRead *)pp; + GET_LookToRead2 size_t size2 = p->size - p->pos; - if (size2 == 0 && *size > 0) + if (size2 == 0 && *size != 0) { p->pos = 0; - if (*size > LookToRead_BUF_SIZE) - *size = LookToRead_BUF_SIZE; - res = p->realStream->Read(p->realStream, p->buf, size); + p->size = 0; + if (*size > p->bufSize) + *size = p->bufSize; + res = ISeekInStream_Read(p->realStream, p->buf, size); size2 = p->size = *size; } - if (size2 < *size) + if (*size > size2) *size = size2; *buf = p->buf + p->pos; return res; } -static SRes LookToRead_Skip(void *pp, size_t offset) +static SRes LookToRead2_Skip(const ILookInStream *pp, size_t offset) { - CLookToRead *p = (CLookToRead *)pp; + GET_LookToRead2 p->pos += offset; return SZ_OK; } -static SRes LookToRead_Read(void *pp, void *buf, size_t *size) +static SRes LookToRead2_Read(const ILookInStream *pp, void *buf, size_t *size) { - CLookToRead *p = (CLookToRead *)pp; + GET_LookToRead2 size_t rem = p->size - p->pos; if (rem == 0) - return p->realStream->Read(p->realStream, buf, size); + return ISeekInStream_Read(p->realStream, buf, size); if (rem > *size) rem = *size; memcpy(buf, p->buf + p->pos, rem); @@ -126,46 +134,43 @@ static SRes LookToRead_Read(void *pp, void *buf, size_t *size) return SZ_OK; } -static SRes LookToRead_Seek(void *pp, Int64 *pos, ESzSeek origin) +static SRes LookToRead2_Seek(const ILookInStream *pp, Int64 *pos, ESzSeek origin) { - CLookToRead *p = (CLookToRead *)pp; + GET_LookToRead2 p->pos = p->size = 0; - return p->realStream->Seek(p->realStream, pos, origin); + return ISeekInStream_Seek(p->realStream, pos, origin); } -void LookToRead_CreateVTable(CLookToRead *p, int lookahead) +void LookToRead2_CreateVTable(CLookToRead2 *p, int lookahead) { - p->s.Look = lookahead ? - LookToRead_Look_Lookahead : - LookToRead_Look_Exact; - p->s.Skip = LookToRead_Skip; - p->s.Read = LookToRead_Read; - p->s.Seek = LookToRead_Seek; + p->vt.Look = lookahead ? + LookToRead2_Look_Lookahead : + LookToRead2_Look_Exact; + p->vt.Skip = LookToRead2_Skip; + p->vt.Read = LookToRead2_Read; + p->vt.Seek = LookToRead2_Seek; } -void LookToRead_Init(CLookToRead *p) -{ - p->pos = p->size = 0; -} -static SRes SecToLook_Read(void *pp, void *buf, size_t *size) + +static SRes SecToLook_Read(const ISeqInStream *pp, void *buf, size_t *size) { - CSecToLook *p = (CSecToLook *)pp; + CSecToLook *p = CONTAINER_FROM_VTBL(pp, CSecToLook, vt); return LookInStream_LookRead(p->realStream, buf, size); } void SecToLook_CreateVTable(CSecToLook *p) { - p->s.Read = SecToLook_Read; + p->vt.Read = SecToLook_Read; } -static SRes SecToRead_Read(void *pp, void *buf, size_t *size) +static SRes SecToRead_Read(const ISeqInStream *pp, void *buf, size_t *size) { - CSecToRead *p = (CSecToRead *)pp; - return p->realStream->Read(p->realStream, buf, size); + CSecToRead *p = CONTAINER_FROM_VTBL(pp, CSecToRead, vt); + return ILookInStream_Read(p->realStream, buf, size); } void SecToRead_CreateVTable(CSecToRead *p) { - p->s.Read = SecToRead_Read; + p->vt.Read = SecToRead_Read; } diff --git a/src/third-party/lzma/7zTypes.h b/src/third-party/lzma/7zTypes.h index 903047b10..593f5aa25 100644 --- a/src/third-party/lzma/7zTypes.h +++ b/src/third-party/lzma/7zTypes.h @@ -1,5 +1,5 @@ /* 7zTypes.h -- Basic types -2013-11-12 : Igor Pavlov : Public domain */ +2018-08-04 : Igor Pavlov : Public domain */ #ifndef __7Z_TYPES_H #define __7Z_TYPES_H @@ -42,13 +42,23 @@ EXTERN_C_BEGIN typedef int SRes; + #ifdef _WIN32 + /* typedef DWORD WRes; */ typedef unsigned WRes; +#define MY_SRes_HRESULT_FROM_WRes(x) HRESULT_FROM_WIN32(x) + #else + typedef int WRes; +#define MY__FACILITY_WIN32 7 +#define MY__FACILITY__WRes MY__FACILITY_WIN32 +#define MY_SRes_HRESULT_FROM_WRes(x) ((HRESULT)(x) <= 0 ? ((HRESULT)(x)) : ((HRESULT) (((x) & 0x0000FFFF) | (MY__FACILITY__WRes << 16) | 0x80000000))) + #endif + #ifndef RINOK #define RINOK(x) { int __result__ = (x); if (__result__ != 0) return __result__; } #endif @@ -93,7 +103,8 @@ typedef UInt32 SizeT; typedef size_t SizeT; #endif -typedef int Bool; +typedef int BoolInt; +/* typedef BoolInt Bool; */ #define True 1 #define False 0 @@ -112,48 +123,72 @@ typedef int Bool; #define MY_NO_INLINE #endif +#define MY_FORCE_INLINE __forceinline + #define MY_CDECL __cdecl #define MY_FAST_CALL __fastcall #else #define MY_NO_INLINE +#define MY_FORCE_INLINE #define MY_CDECL #define MY_FAST_CALL +/* inline keyword : for C++ / C99 */ + +/* GCC, clang: */ +/* +#if defined (__GNUC__) && (__GNUC__ >= 4) +#define MY_FORCE_INLINE __attribute__((always_inline)) +#define MY_NO_INLINE __attribute__((noinline)) +#endif +*/ + #endif /* The following interfaces use first parameter as pointer to structure */ -typedef struct +typedef struct IByteIn IByteIn; +struct IByteIn { - Byte (*Read)(void *p); /* reads one byte, returns 0 in case of EOF or error */ -} IByteIn; + Byte (*Read)(const IByteIn *p); /* reads one byte, returns 0 in case of EOF or error */ +}; +#define IByteIn_Read(p) (p)->Read(p) -typedef struct -{ - void (*Write)(void *p, Byte b); -} IByteOut; -typedef struct +typedef struct IByteOut IByteOut; +struct IByteOut { - SRes (*Read)(void *p, void *buf, size_t *size); + void (*Write)(const IByteOut *p, Byte b); +}; +#define IByteOut_Write(p, b) (p)->Write(p, b) + + +typedef struct ISeqInStream ISeqInStream; +struct ISeqInStream +{ + SRes (*Read)(const ISeqInStream *p, void *buf, size_t *size); /* if (input(*size) != 0 && output(*size) == 0) means end_of_stream. (output(*size) < input(*size)) is allowed */ -} ISeqInStream; +}; +#define ISeqInStream_Read(p, buf, size) (p)->Read(p, buf, size) /* it can return SZ_ERROR_INPUT_EOF */ -SRes SeqInStream_Read(ISeqInStream *stream, void *buf, size_t size); -SRes SeqInStream_Read2(ISeqInStream *stream, void *buf, size_t size, SRes errorType); -SRes SeqInStream_ReadByte(ISeqInStream *stream, Byte *buf); +SRes SeqInStream_Read(const ISeqInStream *stream, void *buf, size_t size); +SRes SeqInStream_Read2(const ISeqInStream *stream, void *buf, size_t size, SRes errorType); +SRes SeqInStream_ReadByte(const ISeqInStream *stream, Byte *buf); -typedef struct + +typedef struct ISeqOutStream ISeqOutStream; +struct ISeqOutStream { - size_t (*Write)(void *p, const void *buf, size_t size); + size_t (*Write)(const ISeqOutStream *p, const void *buf, size_t size); /* Returns: result - the number of actually written bytes. (result < size) means error */ -} ISeqOutStream; +}; +#define ISeqOutStream_Write(p, buf, size) (p)->Write(p, buf, size) typedef enum { @@ -162,78 +197,162 @@ typedef enum SZ_SEEK_END = 2 } ESzSeek; -typedef struct -{ - SRes (*Read)(void *p, void *buf, size_t *size); /* same as ISeqInStream::Read */ - SRes (*Seek)(void *p, Int64 *pos, ESzSeek origin); -} ISeekInStream; -typedef struct +typedef struct ISeekInStream ISeekInStream; +struct ISeekInStream { - SRes (*Look)(void *p, const void **buf, size_t *size); + SRes (*Read)(const ISeekInStream *p, void *buf, size_t *size); /* same as ISeqInStream::Read */ + SRes (*Seek)(const ISeekInStream *p, Int64 *pos, ESzSeek origin); +}; +#define ISeekInStream_Read(p, buf, size) (p)->Read(p, buf, size) +#define ISeekInStream_Seek(p, pos, origin) (p)->Seek(p, pos, origin) + + +typedef struct ILookInStream ILookInStream; +struct ILookInStream +{ + SRes (*Look)(const ILookInStream *p, const void **buf, size_t *size); /* if (input(*size) != 0 && output(*size) == 0) means end_of_stream. (output(*size) > input(*size)) is not allowed (output(*size) < input(*size)) is allowed */ - SRes (*Skip)(void *p, size_t offset); + SRes (*Skip)(const ILookInStream *p, size_t offset); /* offset must be <= output(*size) of Look */ - SRes (*Read)(void *p, void *buf, size_t *size); + SRes (*Read)(const ILookInStream *p, void *buf, size_t *size); /* reads directly (without buffer). It's same as ISeqInStream::Read */ - SRes (*Seek)(void *p, Int64 *pos, ESzSeek origin); -} ILookInStream; + SRes (*Seek)(const ILookInStream *p, Int64 *pos, ESzSeek origin); +}; -SRes LookInStream_LookRead(ILookInStream *stream, void *buf, size_t *size); -SRes LookInStream_SeekTo(ILookInStream *stream, UInt64 offset); +#define ILookInStream_Look(p, buf, size) (p)->Look(p, buf, size) +#define ILookInStream_Skip(p, offset) (p)->Skip(p, offset) +#define ILookInStream_Read(p, buf, size) (p)->Read(p, buf, size) +#define ILookInStream_Seek(p, pos, origin) (p)->Seek(p, pos, origin) + + +SRes LookInStream_LookRead(const ILookInStream *stream, void *buf, size_t *size); +SRes LookInStream_SeekTo(const ILookInStream *stream, UInt64 offset); /* reads via ILookInStream::Read */ -SRes LookInStream_Read2(ILookInStream *stream, void *buf, size_t size, SRes errorType); -SRes LookInStream_Read(ILookInStream *stream, void *buf, size_t size); +SRes LookInStream_Read2(const ILookInStream *stream, void *buf, size_t size, SRes errorType); +SRes LookInStream_Read(const ILookInStream *stream, void *buf, size_t size); + -#define LookToRead_BUF_SIZE (1 << 14) typedef struct { - ILookInStream s; - ISeekInStream *realStream; + ILookInStream vt; + const ISeekInStream *realStream; + size_t pos; - size_t size; - Byte buf[LookToRead_BUF_SIZE]; -} CLookToRead; + size_t size; /* it's data size */ + + /* the following variables must be set outside */ + Byte *buf; + size_t bufSize; +} CLookToRead2; + +void LookToRead2_CreateVTable(CLookToRead2 *p, int lookahead); + +#define LookToRead2_Init(p) { (p)->pos = (p)->size = 0; } -void LookToRead_CreateVTable(CLookToRead *p, int lookahead); -void LookToRead_Init(CLookToRead *p); typedef struct { - ISeqInStream s; - ILookInStream *realStream; + ISeqInStream vt; + const ILookInStream *realStream; } CSecToLook; void SecToLook_CreateVTable(CSecToLook *p); + + typedef struct { - ISeqInStream s; - ILookInStream *realStream; + ISeqInStream vt; + const ILookInStream *realStream; } CSecToRead; void SecToRead_CreateVTable(CSecToRead *p); -typedef struct + +typedef struct ICompressProgress ICompressProgress; + +struct ICompressProgress { - SRes (*Progress)(void *p, UInt64 inSize, UInt64 outSize); + SRes (*Progress)(const ICompressProgress *p, UInt64 inSize, UInt64 outSize); /* Returns: result. (result != SZ_OK) means break. Value (UInt64)(Int64)-1 for size means unknown value. */ -} ICompressProgress; +}; +#define ICompressProgress_Progress(p, inSize, outSize) (p)->Progress(p, inSize, outSize) -typedef struct + + +typedef struct ISzAlloc ISzAlloc; +typedef const ISzAlloc * ISzAllocPtr; + +struct ISzAlloc { - void *(*Alloc)(void *p, size_t size); - void (*Free)(void *p, void *address); /* address can be 0 */ -} ISzAlloc; + void *(*Alloc)(ISzAllocPtr p, size_t size); + void (*Free)(ISzAllocPtr p, void *address); /* address can be 0 */ +}; + +#define ISzAlloc_Alloc(p, size) (p)->Alloc(p, size) +#define ISzAlloc_Free(p, a) (p)->Free(p, a) + +/* deprecated */ +#define IAlloc_Alloc(p, size) ISzAlloc_Alloc(p, size) +#define IAlloc_Free(p, a) ISzAlloc_Free(p, a) + + + + + +#ifndef MY_offsetof + #ifdef offsetof + #define MY_offsetof(type, m) offsetof(type, m) + /* + #define MY_offsetof(type, m) FIELD_OFFSET(type, m) + */ + #else + #define MY_offsetof(type, m) ((size_t)&(((type *)0)->m)) + #endif +#endif + + + +#ifndef MY_container_of + +/* +#define MY_container_of(ptr, type, m) container_of(ptr, type, m) +#define MY_container_of(ptr, type, m) CONTAINING_RECORD(ptr, type, m) +#define MY_container_of(ptr, type, m) ((type *)((char *)(ptr) - offsetof(type, m))) +#define MY_container_of(ptr, type, m) (&((type *)0)->m == (ptr), ((type *)(((char *)(ptr)) - MY_offsetof(type, m)))) +*/ + +/* + GCC shows warning: "perhaps the 'offsetof' macro was used incorrectly" + GCC 3.4.4 : classes with constructor + GCC 4.8.1 : classes with non-public variable members" +*/ + +#define MY_container_of(ptr, type, m) ((type *)((char *)(1 ? (ptr) : &((type *)0)->m) - MY_offsetof(type, m))) + + +#endif + +#define CONTAINER_FROM_VTBL_SIMPLE(ptr, type, m) ((type *)(ptr)) + +/* +#define CONTAINER_FROM_VTBL(ptr, type, m) CONTAINER_FROM_VTBL_SIMPLE(ptr, type, m) +*/ +#define CONTAINER_FROM_VTBL(ptr, type, m) MY_container_of(ptr, type, m) + +#define CONTAINER_FROM_VTBL_CLS(ptr, type, m) CONTAINER_FROM_VTBL_SIMPLE(ptr, type, m) +/* +#define CONTAINER_FROM_VTBL_CLS(ptr, type, m) CONTAINER_FROM_VTBL(ptr, type, m) +*/ + -#define IAlloc_Alloc(p, size) (p)->Alloc((p), size) -#define IAlloc_Free(p, a) (p)->Free((p), a) #ifdef _WIN32 diff --git a/src/third-party/lzma/7zVersion.h b/src/third-party/lzma/7zVersion.h index 52ad3c1b1..0074c64be 100644 --- a/src/third-party/lzma/7zVersion.h +++ b/src/third-party/lzma/7zVersion.h @@ -1,14 +1,21 @@ -#define MY_VER_MAJOR 15 -#define MY_VER_MINOR 14 +#define MY_VER_MAJOR 19 +#define MY_VER_MINOR 00 #define MY_VER_BUILD 0 -#define MY_VERSION_NUMBERS "15.14" -#define MY_VERSION "15.14" -#define MY_DATE "2015-12-31" +#define MY_VERSION_NUMBERS "19.00" +#define MY_VERSION MY_VERSION_NUMBERS + +#ifdef MY_CPU_NAME + #define MY_VERSION_CPU MY_VERSION " (" MY_CPU_NAME ")" +#else + #define MY_VERSION_CPU MY_VERSION +#endif + +#define MY_DATE "2019-02-21" #undef MY_COPYRIGHT #undef MY_VERSION_COPYRIGHT_DATE #define MY_AUTHOR_NAME "Igor Pavlov" #define MY_COPYRIGHT_PD "Igor Pavlov : Public domain" -#define MY_COPYRIGHT_CR "Copyright (c) 1999-2015 Igor Pavlov" +#define MY_COPYRIGHT_CR "Copyright (c) 1999-2018 Igor Pavlov" #ifdef USE_COPYRIGHT_CR #define MY_COPYRIGHT MY_COPYRIGHT_CR @@ -16,4 +23,5 @@ #define MY_COPYRIGHT MY_COPYRIGHT_PD #endif -#define MY_VERSION_COPYRIGHT_DATE MY_VERSION " : " MY_COPYRIGHT " : " MY_DATE +#define MY_COPYRIGHT_DATE MY_COPYRIGHT " : " MY_DATE +#define MY_VERSION_COPYRIGHT_DATE MY_VERSION_CPU " : " MY_COPYRIGHT " : " MY_DATE diff --git a/src/third-party/lzma/Aes.c b/src/third-party/lzma/Aes.c index b0263c6cc..8f7d50ea2 100644 --- a/src/third-party/lzma/Aes.c +++ b/src/third-party/lzma/Aes.c @@ -1,5 +1,5 @@ /* Aes.c -- AES encryption / decryption -2015-02-23 : Igor Pavlov : Public domain */ +2017-01-24 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -49,7 +49,13 @@ static const Byte Rcon[11] = { 0x00, 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0 #define gb0(x) ( (x) & 0xFF) #define gb1(x) (((x) >> ( 8)) & 0xFF) #define gb2(x) (((x) >> (16)) & 0xFF) -#define gb3(x) (((x) >> (24)) & 0xFF) +#define gb3(x) (((x) >> (24))) + +#define gb(n, x) gb ## n(x) + +#define TT(x) (T + (x << 8)) +#define DD(x) (D + (x << 8)) + void AesGenTables(void) { @@ -63,10 +69,10 @@ void AesGenTables(void) UInt32 a1 = Sbox[i]; UInt32 a2 = xtime(a1); UInt32 a3 = a2 ^ a1; - T[ i] = Ui32(a2, a1, a1, a3); - T[0x100 + i] = Ui32(a3, a2, a1, a1); - T[0x200 + i] = Ui32(a1, a3, a2, a1); - T[0x300 + i] = Ui32(a1, a1, a3, a2); + TT(0)[i] = Ui32(a2, a1, a1, a3); + TT(1)[i] = Ui32(a3, a2, a1, a1); + TT(2)[i] = Ui32(a1, a3, a2, a1); + TT(3)[i] = Ui32(a1, a1, a3, a2); } { UInt32 a1 = InvS[i]; @@ -77,10 +83,10 @@ void AesGenTables(void) UInt32 aB = a8 ^ a2 ^ a1; UInt32 aD = a8 ^ a4 ^ a1; UInt32 aE = a8 ^ a4 ^ a2; - D[ i] = Ui32(aE, a9, aD, aB); - D[0x100 + i] = Ui32(aB, aE, a9, aD); - D[0x200 + i] = Ui32(aD, aB, aE, a9); - D[0x300 + i] = Ui32(a9, aD, aB, aE); + DD(0)[i] = Ui32(aE, a9, aD, aB); + DD(1)[i] = Ui32(aB, aE, a9, aD); + DD(2)[i] = Ui32(aD, aB, aE, a9); + DD(3)[i] = Ui32(a9, aD, aB, aE); } } @@ -99,7 +105,7 @@ void AesGenTables(void) } -#define HT(i, x, s) (T + (x << 8))[gb ## x(s[(i + x) & 3])] +#define HT(i, x, s) TT(x)[gb(x, s[(i + x) & 3])] #define HT4(m, i, s, p) m[i] = \ HT(i, 0, s) ^ \ @@ -113,11 +119,11 @@ void AesGenTables(void) HT4(m, 2, s, p); \ HT4(m, 3, s, p); \ -#define FT(i, x) Sbox[gb ## x(m[(i + x) & 3])] +#define FT(i, x) Sbox[gb(x, m[(i + x) & 3])] #define FT4(i) dest[i] = Ui32(FT(i, 0), FT(i, 1), FT(i, 2), FT(i, 3)) ^ w[i]; -#define HD(i, x, s) (D + (x << 8))[gb ## x(s[(i - x) & 3])] +#define HD(i, x, s) DD(x)[gb(x, s[(i - x) & 3])] #define HD4(m, i, s, p) m[i] = \ HD(i, 0, s) ^ \ @@ -131,7 +137,7 @@ void AesGenTables(void) HD4(m, 2, s, p); \ HD4(m, 3, s, p); \ -#define FD(i, x) InvS[gb ## x(m[(i - x) & 3])] +#define FD(i, x) InvS[gb(x, m[(i - x) & 3])] #define FD4(i) dest[i] = Ui32(FD(i, 0), FD(i, 1), FD(i, 2), FD(i, 3)) ^ w[i]; void MY_FAST_CALL Aes_SetKey_Enc(UInt32 *w, const Byte *key, unsigned keySize) @@ -147,7 +153,7 @@ void MY_FAST_CALL Aes_SetKey_Enc(UInt32 *w, const Byte *key, unsigned keySize) for (; i < wSize; i++) { - UInt32 t = w[i - 1]; + UInt32 t = w[(size_t)i - 1]; unsigned rem = i % keySize; if (rem == 0) t = Ui32(Sbox[gb1(t)] ^ Rcon[i / keySize], Sbox[gb2(t)], Sbox[gb3(t)], Sbox[gb0(t)]); @@ -167,10 +173,10 @@ void MY_FAST_CALL Aes_SetKey_Dec(UInt32 *w, const Byte *key, unsigned keySize) { UInt32 r = w[i]; w[i] = - D[ Sbox[gb0(r)]] ^ - D[0x100 + Sbox[gb1(r)]] ^ - D[0x200 + Sbox[gb2(r)]] ^ - D[0x300 + Sbox[gb3(r)]]; + DD(0)[Sbox[gb0(r)]] ^ + DD(1)[Sbox[gb1(r)]] ^ + DD(2)[Sbox[gb2(r)]] ^ + DD(3)[Sbox[gb3(r)]]; } } @@ -276,20 +282,25 @@ void MY_FAST_CALL AesCtr_Code(UInt32 *p, Byte *data, size_t numBlocks) for (; numBlocks != 0; numBlocks--) { UInt32 temp[4]; - Byte buf[16]; - int i; + unsigned i; if (++p[0] == 0) p[1]++; Aes_Encode(p + 4, temp, p); - SetUi32(buf, temp[0]); - SetUi32(buf + 4, temp[1]); - SetUi32(buf + 8, temp[2]); - SetUi32(buf + 12, temp[3]); - - for (i = 0; i < 16; i++) - *data++ ^= buf[i]; + for (i = 0; i < 4; i++, data += 4) + { + UInt32 t = temp[i]; + + #ifdef MY_CPU_LE_UNALIGN + *((UInt32 *)data) ^= t; + #else + data[0] ^= (t & 0xFF); + data[1] ^= ((t >> 8) & 0xFF); + data[2] ^= ((t >> 16) & 0xFF); + data[3] ^= ((t >> 24)); + #endif + } } } diff --git a/src/third-party/lzma/AesOpt.c b/src/third-party/lzma/AesOpt.c index e5d4d2611..0e7f49a1b 100644 --- a/src/third-party/lzma/AesOpt.c +++ b/src/third-party/lzma/AesOpt.c @@ -1,12 +1,12 @@ /* AesOpt.c -- Intel's AES -2013-11-12 : Igor Pavlov : Public domain */ +2017-06-08 : Igor Pavlov : Public domain */ #include "Precomp.h" #include "CpuArch.h" #ifdef MY_CPU_X86_OR_AMD64 -#if _MSC_VER >= 1500 +#if (_MSC_VER > 1500) || (_MSC_FULL_VER >= 150030729) #define USE_INTEL_AES #endif #endif diff --git a/src/third-party/lzma/Alloc.c b/src/third-party/lzma/Alloc.c index 9f1d036af..30b499e5f 100644 --- a/src/third-party/lzma/Alloc.c +++ b/src/third-party/lzma/Alloc.c @@ -1,8 +1,10 @@ /* Alloc.c -- Memory allocation functions -2015-02-21 : Igor Pavlov : Public domain */ +2018-04-27 : Igor Pavlov : Public domain */ #include "Precomp.h" +#include + #ifdef _WIN32 #include #endif @@ -14,20 +16,127 @@ /* use _SZ_ALLOC_DEBUG to debug alloc/free operations */ #ifdef _SZ_ALLOC_DEBUG + #include int g_allocCount = 0; int g_allocCountMid = 0; int g_allocCountBig = 0; + + +#define CONVERT_INT_TO_STR(charType, tempSize) \ + unsigned char temp[tempSize]; unsigned i = 0; \ + while (val >= 10) { temp[i++] = (unsigned char)('0' + (unsigned)(val % 10)); val /= 10; } \ + *s++ = (charType)('0' + (unsigned)val); \ + while (i != 0) { i--; *s++ = temp[i]; } \ + *s = 0; + +static void ConvertUInt64ToString(UInt64 val, char *s) +{ + CONVERT_INT_TO_STR(char, 24); +} + +#define GET_HEX_CHAR(t) ((char)(((t < 10) ? ('0' + t) : ('A' + (t - 10))))) + +static void ConvertUInt64ToHex(UInt64 val, char *s) +{ + UInt64 v = val; + unsigned i; + for (i = 1;; i++) + { + v >>= 4; + if (v == 0) + break; + } + s[i] = 0; + do + { + unsigned t = (unsigned)(val & 0xF); + val >>= 4; + s[--i] = GET_HEX_CHAR(t); + } + while (i); +} + +#define DEBUG_OUT_STREAM stderr + +static void Print(const char *s) +{ + fputs(s, DEBUG_OUT_STREAM); +} + +static void PrintAligned(const char *s, size_t align) +{ + size_t len = strlen(s); + for(;;) + { + fputc(' ', DEBUG_OUT_STREAM); + if (len >= align) + break; + ++len; + } + Print(s); +} + +static void PrintLn() +{ + Print("\n"); +} + +static void PrintHex(UInt64 v, size_t align) +{ + char s[32]; + ConvertUInt64ToHex(v, s); + PrintAligned(s, align); +} + +static void PrintDec(UInt64 v, size_t align) +{ + char s[32]; + ConvertUInt64ToString(v, s); + PrintAligned(s, align); +} + +static void PrintAddr(void *p) +{ + PrintHex((UInt64)(size_t)(ptrdiff_t)p, 12); +} + + +#define PRINT_ALLOC(name, cnt, size, ptr) \ + Print(name " "); \ + PrintDec(cnt++, 10); \ + PrintHex(size, 10); \ + PrintAddr(ptr); \ + PrintLn(); + +#define PRINT_FREE(name, cnt, ptr) if (ptr) { \ + Print(name " "); \ + PrintDec(--cnt, 10); \ + PrintAddr(ptr); \ + PrintLn(); } + +#else + +#define PRINT_ALLOC(name, cnt, size, ptr) +#define PRINT_FREE(name, cnt, ptr) +#define Print(s) +#define PrintLn() +#define PrintHex(v, align) +#define PrintDec(v, align) +#define PrintAddr(p) + #endif + + void *MyAlloc(size_t size) { if (size == 0) - return 0; + return NULL; #ifdef _SZ_ALLOC_DEBUG { void *p = malloc(size); - fprintf(stderr, "\nAlloc %10d bytes, count = %10d, addr = %8X", size, g_allocCount++, (unsigned)p); + PRINT_ALLOC("Alloc ", g_allocCount, size, p); return p; } #else @@ -37,10 +146,8 @@ void *MyAlloc(size_t size) void MyFree(void *address) { - #ifdef _SZ_ALLOC_DEBUG - if (address != 0) - fprintf(stderr, "\nFree; count = %10d, addr = %8X", --g_allocCount, (unsigned)address); - #endif + PRINT_FREE("Free ", g_allocCount, address); + free(address); } @@ -49,20 +156,18 @@ void MyFree(void *address) void *MidAlloc(size_t size) { if (size == 0) - return 0; - #ifdef _SZ_ALLOC_DEBUG - fprintf(stderr, "\nAlloc_Mid %10d bytes; count = %10d", size, g_allocCountMid++); - #endif - return VirtualAlloc(0, size, MEM_COMMIT, PAGE_READWRITE); + return NULL; + + PRINT_ALLOC("Alloc-Mid", g_allocCountMid, size, NULL); + + return VirtualAlloc(NULL, size, MEM_COMMIT, PAGE_READWRITE); } void MidFree(void *address) { - #ifdef _SZ_ALLOC_DEBUG - if (address != 0) - fprintf(stderr, "\nFree_Mid; count = %10d", --g_allocCountMid); - #endif - if (address == 0) + PRINT_FREE("Free-Mid", g_allocCountMid, address); + + if (!address) return; VirtualFree(address, 0, MEM_RELEASE); } @@ -79,10 +184,10 @@ typedef SIZE_T (WINAPI *GetLargePageMinimumP)(); void SetLargePageSize() { #ifdef _7ZIP_LARGE_PAGES - SIZE_T size = 0; + SIZE_T size; GetLargePageMinimumP largePageMinimum = (GetLargePageMinimumP) GetProcAddress(GetModuleHandle(TEXT("kernel32.dll")), "GetLargePageMinimum"); - if (largePageMinimum == 0) + if (!largePageMinimum) return; size = largePageMinimum(); if (size == 0 || (size & (size - 1)) != 0) @@ -95,31 +200,36 @@ void SetLargePageSize() void *BigAlloc(size_t size) { if (size == 0) - return 0; - #ifdef _SZ_ALLOC_DEBUG - fprintf(stderr, "\nAlloc_Big %10d bytes; count = %10d", size, g_allocCountBig++); - #endif + return NULL; + + PRINT_ALLOC("Alloc-Big", g_allocCountBig, size, NULL); #ifdef _7ZIP_LARGE_PAGES - if (g_LargePageSize != 0 && g_LargePageSize <= (1 << 30) && size >= (1 << 18)) { - void *res = VirtualAlloc(0, (size + g_LargePageSize - 1) & (~(g_LargePageSize - 1)), - MEM_COMMIT | MEM_LARGE_PAGES, PAGE_READWRITE); - if (res != 0) - return res; + SIZE_T ps = g_LargePageSize; + if (ps != 0 && ps <= (1 << 30) && size > (ps / 2)) + { + size_t size2; + ps--; + size2 = (size + ps) & ~ps; + if (size2 >= size) + { + void *res = VirtualAlloc(NULL, size2, MEM_COMMIT | MEM_LARGE_PAGES, PAGE_READWRITE); + if (res) + return res; + } + } } #endif - return VirtualAlloc(0, size, MEM_COMMIT, PAGE_READWRITE); + + return VirtualAlloc(NULL, size, MEM_COMMIT, PAGE_READWRITE); } void BigFree(void *address) { - #ifdef _SZ_ALLOC_DEBUG - if (address != 0) - fprintf(stderr, "\nFree_Big; count = %10d", --g_allocCountBig); - #endif + PRINT_FREE("Free-Big", g_allocCountBig, address); - if (address == 0) + if (!address) return; VirtualFree(address, 0, MEM_RELEASE); } @@ -127,10 +237,219 @@ void BigFree(void *address) #endif -static void *SzAlloc(void *p, size_t size) { UNUSED_VAR(p); return MyAlloc(size); } -static void SzFree(void *p, void *address) { UNUSED_VAR(p); MyFree(address); } -ISzAlloc g_Alloc = { SzAlloc, SzFree }; +static void *SzAlloc(ISzAllocPtr p, size_t size) { UNUSED_VAR(p); return MyAlloc(size); } +static void SzFree(ISzAllocPtr p, void *address) { UNUSED_VAR(p); MyFree(address); } +const ISzAlloc g_Alloc = { SzAlloc, SzFree }; -static void *SzBigAlloc(void *p, size_t size) { UNUSED_VAR(p); return BigAlloc(size); } -static void SzBigFree(void *p, void *address) { UNUSED_VAR(p); BigFree(address); } -ISzAlloc g_BigAlloc = { SzBigAlloc, SzBigFree }; +static void *SzMidAlloc(ISzAllocPtr p, size_t size) { UNUSED_VAR(p); return MidAlloc(size); } +static void SzMidFree(ISzAllocPtr p, void *address) { UNUSED_VAR(p); MidFree(address); } +const ISzAlloc g_MidAlloc = { SzMidAlloc, SzMidFree }; + +static void *SzBigAlloc(ISzAllocPtr p, size_t size) { UNUSED_VAR(p); return BigAlloc(size); } +static void SzBigFree(ISzAllocPtr p, void *address) { UNUSED_VAR(p); BigFree(address); } +const ISzAlloc g_BigAlloc = { SzBigAlloc, SzBigFree }; + + +/* + uintptr_t : C99 (optional) + : unsupported in VS6 +*/ + +#ifdef _WIN32 + typedef UINT_PTR UIntPtr; +#else + /* + typedef uintptr_t UIntPtr; + */ + typedef ptrdiff_t UIntPtr; +#endif + + +#define ADJUST_ALLOC_SIZE 0 +/* +#define ADJUST_ALLOC_SIZE (sizeof(void *) - 1) +*/ +/* + Use (ADJUST_ALLOC_SIZE = (sizeof(void *) - 1)), if + MyAlloc() can return address that is NOT multiple of sizeof(void *). +*/ + + +/* +#define MY_ALIGN_PTR_DOWN(p, align) ((void *)((char *)(p) - ((size_t)(UIntPtr)(p) & ((align) - 1)))) +*/ +#define MY_ALIGN_PTR_DOWN(p, align) ((void *)((((UIntPtr)(p)) & ~((UIntPtr)(align) - 1)))) + +#define MY_ALIGN_PTR_UP_PLUS(p, align) MY_ALIGN_PTR_DOWN(((char *)(p) + (align) + ADJUST_ALLOC_SIZE), align) + + +#if (_POSIX_C_SOURCE >= 200112L) && !defined(_WIN32) + #define USE_posix_memalign +#endif + +/* + This posix_memalign() is for test purposes only. + We also need special Free() function instead of free(), + if this posix_memalign() is used. +*/ + +/* +static int posix_memalign(void **ptr, size_t align, size_t size) +{ + size_t newSize = size + align; + void *p; + void *pAligned; + *ptr = NULL; + if (newSize < size) + return 12; // ENOMEM + p = MyAlloc(newSize); + if (!p) + return 12; // ENOMEM + pAligned = MY_ALIGN_PTR_UP_PLUS(p, align); + ((void **)pAligned)[-1] = p; + *ptr = pAligned; + return 0; +} +*/ + +/* + ALLOC_ALIGN_SIZE >= sizeof(void *) + ALLOC_ALIGN_SIZE >= cache_line_size +*/ + +#define ALLOC_ALIGN_SIZE ((size_t)1 << 7) + +static void *SzAlignedAlloc(ISzAllocPtr pp, size_t size) +{ + #ifndef USE_posix_memalign + + void *p; + void *pAligned; + size_t newSize; + UNUSED_VAR(pp); + + /* also we can allocate additional dummy ALLOC_ALIGN_SIZE bytes after aligned + block to prevent cache line sharing with another allocated blocks */ + + newSize = size + ALLOC_ALIGN_SIZE * 1 + ADJUST_ALLOC_SIZE; + if (newSize < size) + return NULL; + + p = MyAlloc(newSize); + + if (!p) + return NULL; + pAligned = MY_ALIGN_PTR_UP_PLUS(p, ALLOC_ALIGN_SIZE); + + Print(" size="); PrintHex(size, 8); + Print(" a_size="); PrintHex(newSize, 8); + Print(" ptr="); PrintAddr(p); + Print(" a_ptr="); PrintAddr(pAligned); + PrintLn(); + + ((void **)pAligned)[-1] = p; + + return pAligned; + + #else + + void *p; + UNUSED_VAR(pp); + if (posix_memalign(&p, ALLOC_ALIGN_SIZE, size)) + return NULL; + + Print(" posix_memalign="); PrintAddr(p); + PrintLn(); + + return p; + + #endif +} + + +static void SzAlignedFree(ISzAllocPtr pp, void *address) +{ + UNUSED_VAR(pp); + #ifndef USE_posix_memalign + if (address) + MyFree(((void **)address)[-1]); + #else + free(address); + #endif +} + + +const ISzAlloc g_AlignedAlloc = { SzAlignedAlloc, SzAlignedFree }; + + + +#define MY_ALIGN_PTR_DOWN_1(p) MY_ALIGN_PTR_DOWN(p, sizeof(void *)) + +/* we align ptr to support cases where CAlignOffsetAlloc::offset is not multiply of sizeof(void *) */ +#define REAL_BLOCK_PTR_VAR(p) ((void **)MY_ALIGN_PTR_DOWN_1(p))[-1] +/* +#define REAL_BLOCK_PTR_VAR(p) ((void **)(p))[-1] +*/ + +static void *AlignOffsetAlloc_Alloc(ISzAllocPtr pp, size_t size) +{ + CAlignOffsetAlloc *p = CONTAINER_FROM_VTBL(pp, CAlignOffsetAlloc, vt); + void *adr; + void *pAligned; + size_t newSize; + size_t extra; + size_t alignSize = (size_t)1 << p->numAlignBits; + + if (alignSize < sizeof(void *)) + alignSize = sizeof(void *); + + if (p->offset >= alignSize) + return NULL; + + /* also we can allocate additional dummy ALLOC_ALIGN_SIZE bytes after aligned + block to prevent cache line sharing with another allocated blocks */ + extra = p->offset & (sizeof(void *) - 1); + newSize = size + alignSize + extra + ADJUST_ALLOC_SIZE; + if (newSize < size) + return NULL; + + adr = ISzAlloc_Alloc(p->baseAlloc, newSize); + + if (!adr) + return NULL; + + pAligned = (char *)MY_ALIGN_PTR_DOWN((char *)adr + + alignSize - p->offset + extra + ADJUST_ALLOC_SIZE, alignSize) + p->offset; + + PrintLn(); + Print("- Aligned: "); + Print(" size="); PrintHex(size, 8); + Print(" a_size="); PrintHex(newSize, 8); + Print(" ptr="); PrintAddr(adr); + Print(" a_ptr="); PrintAddr(pAligned); + PrintLn(); + + REAL_BLOCK_PTR_VAR(pAligned) = adr; + + return pAligned; +} + + +static void AlignOffsetAlloc_Free(ISzAllocPtr pp, void *address) +{ + if (address) + { + CAlignOffsetAlloc *p = CONTAINER_FROM_VTBL(pp, CAlignOffsetAlloc, vt); + PrintLn(); + Print("- Aligned Free: "); + PrintLn(); + ISzAlloc_Free(p->baseAlloc, REAL_BLOCK_PTR_VAR(address)); + } +} + + +void AlignOffsetAlloc_CreateVTable(CAlignOffsetAlloc *p) +{ + p->vt.Alloc = AlignOffsetAlloc_Alloc; + p->vt.Free = AlignOffsetAlloc_Free; +} diff --git a/src/third-party/lzma/Alloc.h b/src/third-party/lzma/Alloc.h index 73b282a07..3d796e5ee 100644 --- a/src/third-party/lzma/Alloc.h +++ b/src/third-party/lzma/Alloc.h @@ -1,5 +1,5 @@ /* Alloc.h -- Memory allocation functions -2015-02-21 : Igor Pavlov : Public domain */ +2018-02-19 : Igor Pavlov : Public domain */ #ifndef __COMMON_ALLOC_H #define __COMMON_ALLOC_H @@ -29,8 +29,22 @@ void BigFree(void *address); #endif -extern ISzAlloc g_Alloc; -extern ISzAlloc g_BigAlloc; +extern const ISzAlloc g_Alloc; +extern const ISzAlloc g_BigAlloc; +extern const ISzAlloc g_MidAlloc; +extern const ISzAlloc g_AlignedAlloc; + + +typedef struct +{ + ISzAlloc vt; + ISzAllocPtr baseAlloc; + unsigned numAlignBits; /* ((1 << numAlignBits) >= sizeof(void *)) */ + size_t offset; /* (offset == (k * sizeof(void *)) && offset < (1 << numAlignBits) */ +} CAlignOffsetAlloc; + +void AlignOffsetAlloc_CreateVTable(CAlignOffsetAlloc *p); + EXTERN_C_END diff --git a/src/third-party/lzma/Bcj2.c b/src/third-party/lzma/Bcj2.c index 707362a61..da93985cf 100644 --- a/src/third-party/lzma/Bcj2.c +++ b/src/third-party/lzma/Bcj2.c @@ -1,5 +1,5 @@ /* Bcj2.c -- BCJ2 Decoder (Converter for x86 code) -2015-08-01 : Igor Pavlov : Public domain */ +2018-04-28 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -61,7 +61,8 @@ SRes Bcj2Dec_Decode(CBcj2Dec *p) Byte *dest = p->dest; if (dest == p->destLim) return SZ_OK; - *dest = p->temp[p->state++ - BCJ2_DEC_STATE_ORIG_0]; + *dest = p->temp[(size_t)p->state - BCJ2_DEC_STATE_ORIG_0]; + p->state++; p->dest = dest + 1; } } @@ -231,10 +232,10 @@ SRes Bcj2Dec_Decode(CBcj2Dec *p) if (rem < 4) { - SizeT i; - SetUi32(p->temp, val); - for (i = 0; i < rem; i++) - dest[i] = p->temp[i]; + p->temp[0] = (Byte)val; if (rem > 0) dest[0] = (Byte)val; val >>= 8; + p->temp[1] = (Byte)val; if (rem > 1) dest[1] = (Byte)val; val >>= 8; + p->temp[2] = (Byte)val; if (rem > 2) dest[2] = (Byte)val; val >>= 8; + p->temp[3] = (Byte)val; p->dest = dest + rem; p->state = BCJ2_DEC_STATE_ORIG_0 + (unsigned)rem; break; diff --git a/src/third-party/lzma/Bcj2Enc.c b/src/third-party/lzma/Bcj2Enc.c index 6a210151e..7a02ecde2 100644 --- a/src/third-party/lzma/Bcj2Enc.c +++ b/src/third-party/lzma/Bcj2Enc.c @@ -1,5 +1,5 @@ /* Bcj2Enc.c -- BCJ2 Encoder (Converter for x86 code) -2014-11-10 : Igor Pavlov : Public domain */ +2019-02-02 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -12,7 +12,6 @@ #define PRF(x) #endif -#include #include #include "Bcj2.h" @@ -53,7 +52,7 @@ void Bcj2Enc_Init(CBcj2Enc *p) p->probs[i] = kBitModelTotal >> 1; } -static Bool MY_FAST_CALL RangeEnc_ShiftLow(CBcj2Enc *p) +static BoolInt MY_FAST_CALL RangeEnc_ShiftLow(CBcj2Enc *p) { if ((UInt32)p->low < (UInt32)0xFF000000 || (UInt32)(p->low >> 32) != 0) { @@ -166,7 +165,7 @@ static void Bcj2Enc_Encode_2(CBcj2Enc *p) { Byte context = (Byte)(num == 0 ? p->prevByte : src[-1]); - Bool needConvert; + BoolInt needConvert; p->bufs[BCJ2_STREAM_MAIN] = dest + 1; p->ip += (UInt32)num + 1; @@ -254,7 +253,7 @@ void Bcj2Enc_Encode(CBcj2Enc *p) { const Byte *src = p->src; const Byte *srcLim = p->srcLim; - unsigned finishMode = p->finishMode; + EBcj2Enc_FinishMode finishMode = p->finishMode; p->src = p->temp; p->srcLim = p->temp + p->tempPos; @@ -271,7 +270,7 @@ void Bcj2Enc_Encode(CBcj2Enc *p) unsigned i; p->tempPos = tempPos; for (i = 0; i < tempPos; i++) - p->temp[i] = p->temp[i + num]; + p->temp[i] = p->temp[(size_t)i + num]; p->src = src; p->srcLim = srcLim; diff --git a/src/third-party/lzma/Bra.c b/src/third-party/lzma/Bra.c index 976810c96..cbdcb290d 100644 --- a/src/third-party/lzma/Bra.c +++ b/src/third-party/lzma/Bra.c @@ -1,135 +1,230 @@ /* Bra.c -- Converters for RISC code -2010-04-16 : Igor Pavlov : Public domain */ +2017-04-04 : Igor Pavlov : Public domain */ #include "Precomp.h" +#include "CpuArch.h" #include "Bra.h" SizeT ARM_Convert(Byte *data, SizeT size, UInt32 ip, int encoding) { - SizeT i; - if (size < 4) - return 0; - size -= 4; - ip += 8; - for (i = 0; i <= size; i += 4) + Byte *p; + const Byte *lim; + size &= ~(size_t)3; + ip += 4; + p = data; + lim = data + size; + + if (encoding) + + for (;;) { - if (data[i + 3] == 0xEB) + for (;;) { - UInt32 dest; - UInt32 src = ((UInt32)data[i + 2] << 16) | ((UInt32)data[i + 1] << 8) | (data[i + 0]); - src <<= 2; - if (encoding) - dest = ip + (UInt32)i + src; - else - dest = src - (ip + (UInt32)i); - dest >>= 2; - data[i + 2] = (Byte)(dest >> 16); - data[i + 1] = (Byte)(dest >> 8); - data[i + 0] = (Byte)dest; + if (p >= lim) + return p - data; + p += 4; + if (p[-1] == 0xEB) + break; + } + { + UInt32 v = GetUi32(p - 4); + v <<= 2; + v += ip + (UInt32)(p - data); + v >>= 2; + v &= 0x00FFFFFF; + v |= 0xEB000000; + SetUi32(p - 4, v); + } + } + + for (;;) + { + for (;;) + { + if (p >= lim) + return p - data; + p += 4; + if (p[-1] == 0xEB) + break; + } + { + UInt32 v = GetUi32(p - 4); + v <<= 2; + v -= ip + (UInt32)(p - data); + v >>= 2; + v &= 0x00FFFFFF; + v |= 0xEB000000; + SetUi32(p - 4, v); } } - return i; } + SizeT ARMT_Convert(Byte *data, SizeT size, UInt32 ip, int encoding) { - SizeT i; - if (size < 4) - return 0; - size -= 4; - ip += 4; - for (i = 0; i <= size; i += 2) + Byte *p; + const Byte *lim; + size &= ~(size_t)1; + p = data; + lim = data + size - 4; + + if (encoding) + + for (;;) { - if ((data[i + 1] & 0xF8) == 0xF0 && - (data[i + 3] & 0xF8) == 0xF8) + UInt32 b1; + for (;;) { - UInt32 dest; - UInt32 src = - (((UInt32)data[i + 1] & 0x7) << 19) | - ((UInt32)data[i + 0] << 11) | - (((UInt32)data[i + 3] & 0x7) << 8) | - (data[i + 2]); - - src <<= 1; - if (encoding) - dest = ip + (UInt32)i + src; - else - dest = src - (ip + (UInt32)i); - dest >>= 1; - - data[i + 1] = (Byte)(0xF0 | ((dest >> 19) & 0x7)); - data[i + 0] = (Byte)(dest >> 11); - data[i + 3] = (Byte)(0xF8 | ((dest >> 8) & 0x7)); - data[i + 2] = (Byte)dest; - i += 2; + UInt32 b3; + if (p > lim) + return p - data; + b1 = p[1]; + b3 = p[3]; + p += 2; + b1 ^= 8; + if ((b3 & b1) >= 0xF8) + break; + } + { + UInt32 v = + ((UInt32)b1 << 19) + + (((UInt32)p[1] & 0x7) << 8) + + (((UInt32)p[-2] << 11)) + + (p[0]); + + p += 2; + { + UInt32 cur = (ip + (UInt32)(p - data)) >> 1; + v += cur; + } + + p[-4] = (Byte)(v >> 11); + p[-3] = (Byte)(0xF0 | ((v >> 19) & 0x7)); + p[-2] = (Byte)v; + p[-1] = (Byte)(0xF8 | (v >> 8)); + } + } + + for (;;) + { + UInt32 b1; + for (;;) + { + UInt32 b3; + if (p > lim) + return p - data; + b1 = p[1]; + b3 = p[3]; + p += 2; + b1 ^= 8; + if ((b3 & b1) >= 0xF8) + break; + } + { + UInt32 v = + ((UInt32)b1 << 19) + + (((UInt32)p[1] & 0x7) << 8) + + (((UInt32)p[-2] << 11)) + + (p[0]); + + p += 2; + { + UInt32 cur = (ip + (UInt32)(p - data)) >> 1; + v -= cur; + } + + /* + SetUi16(p - 4, (UInt16)(((v >> 11) & 0x7FF) | 0xF000)); + SetUi16(p - 2, (UInt16)(v | 0xF800)); + */ + + p[-4] = (Byte)(v >> 11); + p[-3] = (Byte)(0xF0 | ((v >> 19) & 0x7)); + p[-2] = (Byte)v; + p[-1] = (Byte)(0xF8 | (v >> 8)); } } - return i; } + SizeT PPC_Convert(Byte *data, SizeT size, UInt32 ip, int encoding) { - SizeT i; - if (size < 4) - return 0; - size -= 4; - for (i = 0; i <= size; i += 4) + Byte *p; + const Byte *lim; + size &= ~(size_t)3; + ip -= 4; + p = data; + lim = data + size; + + for (;;) { - if ((data[i] >> 2) == 0x12 && (data[i + 3] & 3) == 1) + for (;;) { - UInt32 src = ((UInt32)(data[i + 0] & 3) << 24) | - ((UInt32)data[i + 1] << 16) | - ((UInt32)data[i + 2] << 8) | - ((UInt32)data[i + 3] & (~3)); - - UInt32 dest; + if (p >= lim) + return p - data; + p += 4; + /* if ((v & 0xFC000003) == 0x48000001) */ + if ((p[-4] & 0xFC) == 0x48 && (p[-1] & 3) == 1) + break; + } + { + UInt32 v = GetBe32(p - 4); if (encoding) - dest = ip + (UInt32)i + src; + v += ip + (UInt32)(p - data); else - dest = src - (ip + (UInt32)i); - data[i + 0] = (Byte)(0x48 | ((dest >> 24) & 0x3)); - data[i + 1] = (Byte)(dest >> 16); - data[i + 2] = (Byte)(dest >> 8); - data[i + 3] &= 0x3; - data[i + 3] |= dest; + v -= ip + (UInt32)(p - data); + v &= 0x03FFFFFF; + v |= 0x48000000; + SetBe32(p - 4, v); } } - return i; } + SizeT SPARC_Convert(Byte *data, SizeT size, UInt32 ip, int encoding) { - UInt32 i; - if (size < 4) - return 0; - size -= 4; - for (i = 0; i <= size; i += 4) - { - if ((data[i] == 0x40 && (data[i + 1] & 0xC0) == 0x00) || - (data[i] == 0x7F && (data[i + 1] & 0xC0) == 0xC0)) - { - UInt32 src = - ((UInt32)data[i + 0] << 24) | - ((UInt32)data[i + 1] << 16) | - ((UInt32)data[i + 2] << 8) | - ((UInt32)data[i + 3]); - UInt32 dest; - - src <<= 2; - if (encoding) - dest = ip + i + src; - else - dest = src - (ip + i); - dest >>= 2; - - dest = (((0 - ((dest >> 22) & 1)) << 22) & 0x3FFFFFFF) | (dest & 0x3FFFFF) | 0x40000000; + Byte *p; + const Byte *lim; + size &= ~(size_t)3; + ip -= 4; + p = data; + lim = data + size; - data[i + 0] = (Byte)(dest >> 24); - data[i + 1] = (Byte)(dest >> 16); - data[i + 2] = (Byte)(dest >> 8); - data[i + 3] = (Byte)dest; + for (;;) + { + for (;;) + { + if (p >= lim) + return p - data; + /* + v = GetBe32(p); + p += 4; + m = v + ((UInt32)5 << 29); + m ^= (UInt32)7 << 29; + m += (UInt32)1 << 22; + if ((m & ((UInt32)0x1FF << 23)) == 0) + break; + */ + p += 4; + if ((p[-4] == 0x40 && (p[-3] & 0xC0) == 0) || + (p[-4] == 0x7F && (p[-3] >= 0xC0))) + break; + } + { + UInt32 v = GetBe32(p - 4); + v <<= 2; + if (encoding) + v += ip + (UInt32)(p - data); + else + v -= ip + (UInt32)(p - data); + + v &= 0x01FFFFFF; + v -= (UInt32)1 << 24; + v ^= 0xFF000000; + v >>= 2; + v |= 0x40000000; + SetBe32(p - 4, v); } } - return i; } diff --git a/src/third-party/lzma/Bra86.c b/src/third-party/lzma/Bra86.c index 8dd3ed48d..a6463c63b 100644 --- a/src/third-party/lzma/Bra86.c +++ b/src/third-party/lzma/Bra86.c @@ -1,5 +1,5 @@ /* Bra86.c -- Converter for x86 code (BCJ) -2013-11-12 : Igor Pavlov : Public domain */ +2017-04-03 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -37,7 +37,7 @@ SizeT x86_Convert(Byte *data, SizeT size, UInt32 ip, UInt32 *state, int encoding else { mask >>= (unsigned)d; - if (mask != 0 && (mask > 4 || mask == 3 || Test86MSByte(p[(mask >> 1) + 1]))) + if (mask != 0 && (mask > 4 || mask == 3 || Test86MSByte(p[(size_t)(mask >> 1) + 1]))) { mask = (mask >> 1) | 4; pos++; diff --git a/src/third-party/lzma/BraIA64.c b/src/third-party/lzma/BraIA64.c index 813830c79..2656907a0 100644 --- a/src/third-party/lzma/BraIA64.c +++ b/src/third-party/lzma/BraIA64.c @@ -1,69 +1,53 @@ /* BraIA64.c -- Converter for IA-64 code -2013-11-12 : Igor Pavlov : Public domain */ +2017-01-26 : Igor Pavlov : Public domain */ #include "Precomp.h" +#include "CpuArch.h" #include "Bra.h" -static const Byte kBranchTable[32] = -{ - 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, - 4, 4, 6, 6, 0, 0, 7, 7, - 4, 4, 0, 0, 4, 4, 0, 0 -}; - SizeT IA64_Convert(Byte *data, SizeT size, UInt32 ip, int encoding) { SizeT i; if (size < 16) return 0; size -= 16; - for (i = 0; i <= size; i += 16) + i = 0; + do { - UInt32 instrTemplate = data[i] & 0x1F; - UInt32 mask = kBranchTable[instrTemplate]; - UInt32 bitPos = 5; - int slot; - for (slot = 0; slot < 3; slot++, bitPos += 41) + unsigned m = ((UInt32)0x334B0000 >> (data[i] & 0x1E)) & 3; + if (m) { - UInt32 bytePos, bitRes; - UInt64 instruction, instNorm; - int j; - if (((mask >> slot) & 1) == 0) - continue; - bytePos = (bitPos >> 3); - bitRes = bitPos & 0x7; - instruction = 0; - for (j = 0; j < 6; j++) - instruction += (UInt64)data[i + j + bytePos] << (8 * j); - - instNorm = instruction >> bitRes; - if (((instNorm >> 37) & 0xF) == 0x5 && ((instNorm >> 9) & 0x7) == 0) + m++; + do { - UInt32 src = (UInt32)((instNorm >> 13) & 0xFFFFF); - UInt32 dest; - src |= ((UInt32)(instNorm >> 36) & 1) << 20; - - src <<= 4; - - if (encoding) - dest = ip + (UInt32)i + src; - else - dest = src - (ip + (UInt32)i); - - dest >>= 4; - - instNorm &= ~((UInt64)(0x8FFFFF) << 13); - instNorm |= ((UInt64)(dest & 0xFFFFF) << 13); - instNorm |= ((UInt64)(dest & 0x100000) << (36 - 20)); - - instruction &= (1 << bitRes) - 1; - instruction |= (instNorm << bitRes); - for (j = 0; j < 6; j++) - data[i + j + bytePos] = (Byte)(instruction >> (8 * j)); + Byte *p = data + (i + (size_t)m * 5 - 8); + if (((p[3] >> m) & 15) == 5 + && (((p[-1] | ((UInt32)p[0] << 8)) >> m) & 0x70) == 0) + { + unsigned raw = GetUi32(p); + unsigned v = raw >> m; + v = (v & 0xFFFFF) | ((v & (1 << 23)) >> 3); + + v <<= 4; + if (encoding) + v += ip + (UInt32)i; + else + v -= ip + (UInt32)i; + v >>= 4; + + v &= 0x1FFFFF; + v += 0x700000; + v &= 0x8FFFFF; + raw &= ~((UInt32)0x8FFFFF << m); + raw |= (v << m); + SetUi32(p, raw); + } } + while (++m <= 4); } + i += 16; } + while (i <= size); return i; } diff --git a/src/third-party/lzma/Compiler.h b/src/third-party/lzma/Compiler.h index de8fab374..c788648cd 100644 --- a/src/third-party/lzma/Compiler.h +++ b/src/third-party/lzma/Compiler.h @@ -1,5 +1,5 @@ /* Compiler.h -2015-08-02 : Igor Pavlov : Public domain */ +2017-04-03 : Igor Pavlov : Public domain */ #ifndef __7Z_COMPILER_H #define __7Z_COMPILER_H @@ -21,6 +21,7 @@ #pragma warning(disable : 4514) // unreferenced inline function has been removed #pragma warning(disable : 4702) // unreachable code #pragma warning(disable : 4710) // not inlined + #pragma warning(disable : 4714) // function marked as __forceinline not inlined #pragma warning(disable : 4786) // identifier was truncated to '255' characters in the debug information #endif diff --git a/src/third-party/lzma/CpuArch.c b/src/third-party/lzma/CpuArch.c index bcb84cb2b..ff1890e7f 100644 --- a/src/third-party/lzma/CpuArch.c +++ b/src/third-party/lzma/CpuArch.c @@ -1,5 +1,5 @@ /* CpuArch.c -- CPU specific code -2015-03-25: Igor Pavlov : Public domain */ +2018-02-18: Igor Pavlov : Public domain */ #include "Precomp.h" @@ -45,7 +45,8 @@ static UInt32 CheckFlag(UInt32 flag) "push %%EDX\n\t" "popf\n\t" "andl %%EAX, %0\n\t": - "=c" (flag) : "c" (flag)); + "=c" (flag) : "c" (flag) : + "%eax", "%edx"); #endif return flag; } @@ -79,7 +80,13 @@ void MyCPUID(UInt32 function, UInt32 *a, UInt32 *b, UInt32 *c, UInt32 *d) #else __asm__ __volatile__ ( - #if defined(MY_CPU_X86) && defined(__PIC__) + #if defined(MY_CPU_AMD64) && defined(__PIC__) + "mov %%rbx, %%rdi;" + "cpuid;" + "xchg %%rbx, %%rdi;" + : "=a" (*a) , + "=D" (*b) , + #elif defined(MY_CPU_X86) && defined(__PIC__) "mov %%ebx, %%edi;" "cpuid;" "xchgl %%ebx, %%edi;" @@ -108,7 +115,7 @@ void MyCPUID(UInt32 function, UInt32 *a, UInt32 *b, UInt32 *c, UInt32 *d) #endif } -Bool x86cpuid_CheckAndRead(Cx86cpuid *p) +BoolInt x86cpuid_CheckAndRead(Cx86cpuid *p) { CHECK_CPUID_IS_SUPPORTED MyCPUID(0, &p->maxFunc, &p->vendor[0], &p->vendor[2], &p->vendor[1]); @@ -137,7 +144,7 @@ int x86cpuid_GetFirm(const Cx86cpuid *p) return -1; } -Bool CPU_Is_InOrder() +BoolInt CPU_Is_InOrder() { Cx86cpuid p; int firm; @@ -168,7 +175,7 @@ Bool CPU_Is_InOrder() #if !defined(MY_CPU_AMD64) && defined(_WIN32) #include -static Bool CPU_Sys_Is_SSE_Supported() +static BoolInt CPU_Sys_Is_SSE_Supported() { OSVERSIONINFO vi; vi.dwOSVersionInfoSize = sizeof(vi); @@ -181,7 +188,7 @@ static Bool CPU_Sys_Is_SSE_Supported() #define CHECK_SYS_SSE_SUPPORT #endif -Bool CPU_Is_Aes_Supported() +BoolInt CPU_Is_Aes_Supported() { Cx86cpuid p; CHECK_SYS_SSE_SUPPORT @@ -190,4 +197,22 @@ Bool CPU_Is_Aes_Supported() return (p.c >> 25) & 1; } +BoolInt CPU_IsSupported_PageGB() +{ + Cx86cpuid cpuid; + if (!x86cpuid_CheckAndRead(&cpuid)) + return False; + { + UInt32 d[4] = { 0 }; + MyCPUID(0x80000000, &d[0], &d[1], &d[2], &d[3]); + if (d[0] < 0x80000001) + return False; + } + { + UInt32 d[4] = { 0 }; + MyCPUID(0x80000001, &d[0], &d[1], &d[2], &d[3]); + return (d[3] >> 26) & 1; + } +} + #endif diff --git a/src/third-party/lzma/CpuArch.h b/src/third-party/lzma/CpuArch.h index b31c2546c..5f74c1c0c 100644 --- a/src/third-party/lzma/CpuArch.h +++ b/src/third-party/lzma/CpuArch.h @@ -1,5 +1,5 @@ /* CpuArch.h -- CPU specific code -2015-12-01: Igor Pavlov : Public domain */ +2018-02-18 : Igor Pavlov : Public domain */ #ifndef __CPU_ARCH_H #define __CPU_ARCH_H @@ -16,48 +16,122 @@ If MY_CPU_LE and MY_CPU_BE are not defined, we don't know about ENDIANNESS of pl MY_CPU_LE_UNALIGN means that CPU is LITTLE ENDIAN and CPU supports unaligned memory accesses. */ -#if defined(_M_X64) \ - || defined(_M_AMD64) \ - || defined(__x86_64__) \ - || defined(__AMD64__) \ - || defined(__amd64__) +#if defined(_M_X64) \ + || defined(_M_AMD64) \ + || defined(__x86_64__) \ + || defined(__AMD64__) \ + || defined(__amd64__) #define MY_CPU_AMD64 -#endif - -#if defined(MY_CPU_AMD64) \ - || defined(_M_IA64) \ - || defined(__AARCH64EL__) \ - || defined(__AARCH64EB__) + #ifdef __ILP32__ + #define MY_CPU_NAME "x32" + #else + #define MY_CPU_NAME "x64" + #endif #define MY_CPU_64BIT #endif -#if defined(_M_IX86) || defined(__i386__) -#define MY_CPU_X86 + +#if defined(_M_IX86) \ + || defined(__i386__) + #define MY_CPU_X86 + #define MY_CPU_NAME "x86" + #define MY_CPU_32BIT #endif + +#if defined(_M_ARM64) \ + || defined(__AARCH64EL__) \ + || defined(__AARCH64EB__) \ + || defined(__aarch64__) + #define MY_CPU_ARM64 + #define MY_CPU_NAME "arm64" + #define MY_CPU_64BIT +#endif + + +#if defined(_M_ARM) \ + || defined(_M_ARM_NT) \ + || defined(_M_ARMT) \ + || defined(__arm__) \ + || defined(__thumb__) \ + || defined(__ARMEL__) \ + || defined(__ARMEB__) \ + || defined(__THUMBEL__) \ + || defined(__THUMBEB__) + #define MY_CPU_ARM + #define MY_CPU_NAME "arm" + #define MY_CPU_32BIT +#endif + + +#if defined(_M_IA64) \ + || defined(__ia64__) + #define MY_CPU_IA64 + #define MY_CPU_NAME "ia64" + #define MY_CPU_64BIT +#endif + + +#if defined(__mips64) \ + || defined(__mips64__) \ + || (defined(__mips) && (__mips == 64 || __mips == 4 || __mips == 3)) + #define MY_CPU_NAME "mips64" + #define MY_CPU_64BIT +#elif defined(__mips__) + #define MY_CPU_NAME "mips" + /* #define MY_CPU_32BIT */ +#endif + + +#if defined(__ppc64__) \ + || defined(__powerpc64__) + #ifdef __ILP32__ + #define MY_CPU_NAME "ppc64-32" + #else + #define MY_CPU_NAME "ppc64" + #endif + #define MY_CPU_64BIT +#elif defined(__ppc__) \ + || defined(__powerpc__) + #define MY_CPU_NAME "ppc" + #define MY_CPU_32BIT +#endif + + +#if defined(__sparc64__) + #define MY_CPU_NAME "sparc64" + #define MY_CPU_64BIT +#elif defined(__sparc__) + #define MY_CPU_NAME "sparc" + /* #define MY_CPU_32BIT */ +#endif + + #if defined(MY_CPU_X86) || defined(MY_CPU_AMD64) #define MY_CPU_X86_OR_AMD64 #endif -#if defined(MY_CPU_X86) \ - || defined(_M_ARM) \ - || defined(__ARMEL__) \ - || defined(__THUMBEL__) \ - || defined(__ARMEB__) \ - || defined(__THUMBEB__) - #define MY_CPU_32BIT + +#ifdef _WIN32 + + #ifdef MY_CPU_ARM + #define MY_CPU_ARM_LE + #endif + + #ifdef MY_CPU_ARM64 + #define MY_CPU_ARM64_LE + #endif + + #ifdef _M_IA64 + #define MY_CPU_IA64_LE + #endif + #endif -#if defined(_WIN32) && defined(_M_ARM) -#define MY_CPU_ARM_LE -#endif - -#if defined(_WIN32) && defined(_M_IA64) -#define MY_CPU_IA64_LE -#endif #if defined(MY_CPU_X86_OR_AMD64) \ || defined(MY_CPU_ARM_LE) \ + || defined(MY_CPU_ARM64_LE) \ || defined(MY_CPU_IA64_LE) \ || defined(__LITTLE_ENDIAN__) \ || defined(__ARMEL__) \ @@ -66,6 +140,7 @@ MY_CPU_LE_UNALIGN means that CPU is LITTLE ENDIAN and CPU supports unaligned mem || defined(__MIPSEL__) \ || defined(__MIPSEL) \ || defined(_MIPSEL) \ + || defined(__BFIN__) \ || (defined(__BYTE_ORDER__) && (__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__)) #define MY_CPU_LE #endif @@ -85,14 +160,37 @@ MY_CPU_LE_UNALIGN means that CPU is LITTLE ENDIAN and CPU supports unaligned mem #define MY_CPU_BE #endif + #if defined(MY_CPU_LE) && defined(MY_CPU_BE) -Stop_Compiling_Bad_Endian + #error Stop_Compiling_Bad_Endian #endif +#if defined(MY_CPU_32BIT) && defined(MY_CPU_64BIT) + #error Stop_Compiling_Bad_32_64_BIT +#endif + + +#ifndef MY_CPU_NAME + #ifdef MY_CPU_LE + #define MY_CPU_NAME "LE" + #elif defined(MY_CPU_BE) + #define MY_CPU_NAME "BE" + #else + /* + #define MY_CPU_NAME "" + */ + #endif +#endif + + + + + #ifdef MY_CPU_LE #if defined(MY_CPU_X86_OR_AMD64) \ - /* || defined(__AARCH64EL__) */ + || defined(MY_CPU_ARM64) \ + || defined(__ARM_FEATURE_UNALIGNED) #define MY_CPU_LE_UNALIGN #endif #endif @@ -138,6 +236,11 @@ Stop_Compiling_Bad_Endian #endif +#ifdef __has_builtin + #define MY__has_builtin(x) __has_builtin(x) +#else + #define MY__has_builtin(x) 0 +#endif #if defined(MY_CPU_LE_UNALIGN) && /* defined(_WIN64) && */ (_MSC_VER >= 1300) @@ -145,15 +248,21 @@ Stop_Compiling_Bad_Endian #include +#pragma intrinsic(_byteswap_ushort) #pragma intrinsic(_byteswap_ulong) #pragma intrinsic(_byteswap_uint64) + +/* #define GetBe16(p) _byteswap_ushort(*(const UInt16 *)(const Byte *)(p)) */ #define GetBe32(p) _byteswap_ulong(*(const UInt32 *)(const Byte *)(p)) #define GetBe64(p) _byteswap_uint64(*(const UInt64 *)(const Byte *)(p)) #define SetBe32(p, v) (*(UInt32 *)(void *)(p)) = _byteswap_ulong(v) -#elif defined(MY_CPU_LE_UNALIGN) && defined (__GNUC__) && (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3)) +#elif defined(MY_CPU_LE_UNALIGN) && ( \ + (defined(__GNUC__) && (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3))) \ + || (defined(__clang__) && MY__has_builtin(__builtin_bswap16)) ) +/* #define GetBe16(p) __builtin_bswap16(*(const UInt16 *)(const Byte *)(p)) */ #define GetBe32(p) __builtin_bswap32(*(const UInt32 *)(const Byte *)(p)) #define GetBe64(p) __builtin_bswap64(*(const UInt64 *)(const Byte *)(p)) @@ -178,10 +287,14 @@ Stop_Compiling_Bad_Endian #endif +#ifndef GetBe16 + #define GetBe16(p) ( (UInt16) ( \ ((UInt16)((const Byte *)(p))[0] << 8) | \ ((const Byte *)(p))[1] )) +#endif + #ifdef MY_CPU_X86_OR_AMD64 @@ -205,15 +318,16 @@ enum void MyCPUID(UInt32 function, UInt32 *a, UInt32 *b, UInt32 *c, UInt32 *d); -Bool x86cpuid_CheckAndRead(Cx86cpuid *p); +BoolInt x86cpuid_CheckAndRead(Cx86cpuid *p); int x86cpuid_GetFirm(const Cx86cpuid *p); #define x86cpuid_GetFamily(ver) (((ver >> 16) & 0xFF0) | ((ver >> 8) & 0xF)) #define x86cpuid_GetModel(ver) (((ver >> 12) & 0xF0) | ((ver >> 4) & 0xF)) #define x86cpuid_GetStepping(ver) (ver & 0xF) -Bool CPU_Is_InOrder(); -Bool CPU_Is_Aes_Supported(); +BoolInt CPU_Is_InOrder(); +BoolInt CPU_Is_Aes_Supported(); +BoolInt CPU_IsSupported_PageGB(); #endif diff --git a/src/third-party/lzma/DllSecur.c b/src/third-party/lzma/DllSecur.c new file mode 100644 index 000000000..19a22a9f0 --- /dev/null +++ b/src/third-party/lzma/DllSecur.c @@ -0,0 +1,108 @@ +/* DllSecur.c -- DLL loading security +2018-02-21 : Igor Pavlov : Public domain */ + +#include "Precomp.h" + +#ifdef _WIN32 + +#include + +#include "DllSecur.h" + +#ifndef UNDER_CE + +typedef BOOL (WINAPI *Func_SetDefaultDllDirectories)(DWORD DirectoryFlags); + +#define MY_LOAD_LIBRARY_SEARCH_USER_DIRS 0x400 +#define MY_LOAD_LIBRARY_SEARCH_SYSTEM32 0x800 + +static const char * const g_Dlls = + #ifndef _CONSOLE + "UXTHEME\0" + #endif + "USERENV\0" + "SETUPAPI\0" + "APPHELP\0" + "PROPSYS\0" + "DWMAPI\0" + "CRYPTBASE\0" + "OLEACC\0" + "CLBCATQ\0" + "VERSION\0" + ; + +#endif + +void My_SetDefaultDllDirectories() +{ + #ifndef UNDER_CE + + OSVERSIONINFO vi; + vi.dwOSVersionInfoSize = sizeof(vi); + GetVersionEx(&vi); + if (!GetVersionEx(&vi) || vi.dwMajorVersion != 6 || vi.dwMinorVersion != 0) + { + Func_SetDefaultDllDirectories setDllDirs = (Func_SetDefaultDllDirectories) + GetProcAddress(GetModuleHandle(TEXT("kernel32.dll")), "SetDefaultDllDirectories"); + if (setDllDirs) + if (setDllDirs(MY_LOAD_LIBRARY_SEARCH_SYSTEM32 | MY_LOAD_LIBRARY_SEARCH_USER_DIRS)) + return; + } + + #endif +} + + +void LoadSecurityDlls() +{ + #ifndef UNDER_CE + + wchar_t buf[MAX_PATH + 100]; + + { + // at Vista (ver 6.0) : CoCreateInstance(CLSID_ShellLink, ...) doesn't work after SetDefaultDllDirectories() : Check it ??? + OSVERSIONINFO vi; + vi.dwOSVersionInfoSize = sizeof(vi); + if (!GetVersionEx(&vi) || vi.dwMajorVersion != 6 || vi.dwMinorVersion != 0) + { + Func_SetDefaultDllDirectories setDllDirs = (Func_SetDefaultDllDirectories) + GetProcAddress(GetModuleHandle(TEXT("kernel32.dll")), "SetDefaultDllDirectories"); + if (setDllDirs) + if (setDllDirs(MY_LOAD_LIBRARY_SEARCH_SYSTEM32 | MY_LOAD_LIBRARY_SEARCH_USER_DIRS)) + return; + } + } + + { + unsigned len = GetSystemDirectoryW(buf, MAX_PATH + 2); + if (len == 0 || len > MAX_PATH) + return; + } + { + const char *dll; + unsigned pos = (unsigned)lstrlenW(buf); + + if (buf[pos - 1] != '\\') + buf[pos++] = '\\'; + + for (dll = g_Dlls; dll[0] != 0;) + { + unsigned k = 0; + for (;;) + { + char c = *dll++; + buf[pos + k] = (Byte)c; + k++; + if (c == 0) + break; + } + + lstrcatW(buf, L".dll"); + LoadLibraryExW(buf, NULL, LOAD_WITH_ALTERED_SEARCH_PATH); + } + } + + #endif +} + +#endif diff --git a/src/third-party/lzma/DllSecur.h b/src/third-party/lzma/DllSecur.h new file mode 100644 index 000000000..4c113568e --- /dev/null +++ b/src/third-party/lzma/DllSecur.h @@ -0,0 +1,20 @@ +/* DllSecur.h -- DLL loading for security +2018-02-19 : Igor Pavlov : Public domain */ + +#ifndef __DLL_SECUR_H +#define __DLL_SECUR_H + +#include "7zTypes.h" + +EXTERN_C_BEGIN + +#ifdef _WIN32 + +void My_SetDefaultDllDirectories(); +void LoadSecurityDlls(); + +#endif + +EXTERN_C_END + +#endif diff --git a/src/third-party/lzma/LzFind.c b/src/third-party/lzma/LzFind.c index c335d363c..4eefc17dd 100644 --- a/src/third-party/lzma/LzFind.c +++ b/src/third-party/lzma/LzFind.c @@ -1,5 +1,5 @@ /* LzFind.c -- Match finder for LZ algorithms -2015-10-15 : Igor Pavlov : Public domain */ +2018-07-08 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -16,18 +16,18 @@ #define kStartMaxLen 3 -static void LzInWindow_Free(CMatchFinder *p, ISzAlloc *alloc) +static void LzInWindow_Free(CMatchFinder *p, ISzAllocPtr alloc) { if (!p->directInput) { - alloc->Free(alloc, p->bufferBase); + ISzAlloc_Free(alloc, p->bufferBase); p->bufferBase = NULL; } } /* keepSizeBefore + keepSizeAfter + keepSizeReserv must be < 4G) */ -static int LzInWindow_Create(CMatchFinder *p, UInt32 keepSizeReserv, ISzAlloc *alloc) +static int LzInWindow_Create(CMatchFinder *p, UInt32 keepSizeReserv, ISzAllocPtr alloc) { UInt32 blockSize = p->keepSizeBefore + p->keepSizeAfter + keepSizeReserv; if (p->directInput) @@ -39,7 +39,7 @@ static int LzInWindow_Create(CMatchFinder *p, UInt32 keepSizeReserv, ISzAlloc *a { LzInWindow_Free(p, alloc); p->blockSize = blockSize; - p->bufferBase = (Byte *)alloc->Alloc(alloc, (size_t)blockSize); + p->bufferBase = (Byte *)ISzAlloc_Alloc(alloc, (size_t)blockSize); } return (p->bufferBase != NULL); } @@ -81,7 +81,7 @@ static void MatchFinder_ReadBlock(CMatchFinder *p) if (size == 0) return; - p->result = p->stream->Read(p->stream, dest, &size); + p->result = ISeqInStream_Read(p->stream, dest, &size); if (p->result != SZ_OK) return; if (size == 0) @@ -138,45 +138,46 @@ static void MatchFinder_SetDefaultSettings(CMatchFinder *p) void MatchFinder_Construct(CMatchFinder *p) { - UInt32 i; + unsigned i; p->bufferBase = NULL; p->directInput = 0; p->hash = NULL; + p->expectedDataSize = (UInt64)(Int64)-1; MatchFinder_SetDefaultSettings(p); for (i = 0; i < 256; i++) { - UInt32 r = i; + UInt32 r = (UInt32)i; unsigned j; for (j = 0; j < 8; j++) - r = (r >> 1) ^ (kCrcPoly & ~((r & 1) - 1)); + r = (r >> 1) ^ (kCrcPoly & ((UInt32)0 - (r & 1))); p->crc[i] = r; } } -static void MatchFinder_FreeThisClassMemory(CMatchFinder *p, ISzAlloc *alloc) +static void MatchFinder_FreeThisClassMemory(CMatchFinder *p, ISzAllocPtr alloc) { - alloc->Free(alloc, p->hash); + ISzAlloc_Free(alloc, p->hash); p->hash = NULL; } -void MatchFinder_Free(CMatchFinder *p, ISzAlloc *alloc) +void MatchFinder_Free(CMatchFinder *p, ISzAllocPtr alloc) { MatchFinder_FreeThisClassMemory(p, alloc); LzInWindow_Free(p, alloc); } -static CLzRef* AllocRefs(size_t num, ISzAlloc *alloc) +static CLzRef* AllocRefs(size_t num, ISzAllocPtr alloc) { size_t sizeInBytes = (size_t)num * sizeof(CLzRef); if (sizeInBytes / sizeof(CLzRef) != num) return NULL; - return (CLzRef *)alloc->Alloc(alloc, sizeInBytes); + return (CLzRef *)ISzAlloc_Alloc(alloc, sizeInBytes); } int MatchFinder_Create(CMatchFinder *p, UInt32 historySize, UInt32 keepAddBufferBefore, UInt32 matchMaxLen, UInt32 keepAddBufferAfter, - ISzAlloc *alloc) + ISzAllocPtr alloc) { UInt32 sizeReserv; @@ -208,7 +209,11 @@ int MatchFinder_Create(CMatchFinder *p, UInt32 historySize, hs = (1 << 16) - 1; else { - hs = historySize - 1; + hs = historySize; + if (hs > p->expectedDataSize) + hs = (UInt32)p->expectedDataSize; + if (hs != 0) + hs--; hs |= (hs >> 1); hs |= (hs >> 2); hs |= (hs >> 4); @@ -292,17 +297,33 @@ static void MatchFinder_SetLimits(CMatchFinder *p) p->posLimit = p->pos + limit; } -void MatchFinder_Init_2(CMatchFinder *p, int readData) + +void MatchFinder_Init_LowHash(CMatchFinder *p) +{ + size_t i; + CLzRef *items = p->hash; + size_t numItems = p->fixedHashSize; + for (i = 0; i < numItems; i++) + items[i] = kEmptyHashValue; +} + + +void MatchFinder_Init_HighHash(CMatchFinder *p) +{ + size_t i; + CLzRef *items = p->hash + p->fixedHashSize; + size_t numItems = (size_t)p->hashMask + 1; + for (i = 0; i < numItems; i++) + items[i] = kEmptyHashValue; +} + + +void MatchFinder_Init_3(CMatchFinder *p, int readData) { - UInt32 i; - UInt32 *hash = p->hash; - UInt32 num = p->hashSizeSum; - for (i = 0; i < num; i++) - hash[i] = kEmptyHashValue; - p->cyclicBufferPos = 0; p->buffer = p->bufferBase; - p->pos = p->streamPos = p->cyclicBufferSize; + p->pos = + p->streamPos = p->cyclicBufferSize; p->result = SZ_OK; p->streamEndWasReached = 0; @@ -312,10 +333,14 @@ void MatchFinder_Init_2(CMatchFinder *p, int readData) MatchFinder_SetLimits(p); } + void MatchFinder_Init(CMatchFinder *p) { - MatchFinder_Init_2(p, True); + MatchFinder_Init_HighHash(p); + MatchFinder_Init_LowHash(p); + MatchFinder_Init_3(p, True); } + static UInt32 MatchFinder_GetSubValue(CMatchFinder *p) { @@ -343,6 +368,8 @@ static void MatchFinder_Normalize(CMatchFinder *p) MatchFinder_ReduceOffsets(p, subValue); } + +MY_NO_INLINE static void MatchFinder_CheckLimits(CMatchFinder *p) { if (p->pos == kMaxValForNormalize) @@ -354,10 +381,16 @@ static void MatchFinder_CheckLimits(CMatchFinder *p) MatchFinder_SetLimits(p); } -static UInt32 * Hc_GetMatchesSpec(UInt32 lenLimit, UInt32 curMatch, UInt32 pos, const Byte *cur, CLzRef *son, + +/* + (lenLimit > maxLen) +*/ +MY_FORCE_INLINE +static UInt32 * Hc_GetMatchesSpec(unsigned lenLimit, UInt32 curMatch, UInt32 pos, const Byte *cur, CLzRef *son, UInt32 _cyclicBufferPos, UInt32 _cyclicBufferSize, UInt32 cutValue, - UInt32 *distances, UInt32 maxLen) + UInt32 *distances, unsigned maxLen) { + /* son[_cyclicBufferPos] = curMatch; for (;;) { @@ -375,7 +408,8 @@ static UInt32 * Hc_GetMatchesSpec(UInt32 lenLimit, UInt32 curMatch, UInt32 pos, break; if (maxLen < len) { - *distances++ = maxLen = len; + maxLen = len; + *distances++ = len; *distances++ = delta - 1; if (len == lenLimit) return distances; @@ -383,15 +417,58 @@ static UInt32 * Hc_GetMatchesSpec(UInt32 lenLimit, UInt32 curMatch, UInt32 pos, } } } + */ + + const Byte *lim = cur + lenLimit; + son[_cyclicBufferPos] = curMatch; + do + { + UInt32 delta = pos - curMatch; + if (delta >= _cyclicBufferSize) + break; + { + ptrdiff_t diff; + curMatch = son[_cyclicBufferPos - delta + ((delta > _cyclicBufferPos) ? _cyclicBufferSize : 0)]; + diff = (ptrdiff_t)0 - delta; + if (cur[maxLen] == cur[maxLen + diff]) + { + const Byte *c = cur; + while (*c == c[diff]) + { + if (++c == lim) + { + distances[0] = (UInt32)(lim - cur); + distances[1] = delta - 1; + return distances + 2; + } + } + { + unsigned len = (unsigned)(c - cur); + if (maxLen < len) + { + maxLen = len; + distances[0] = (UInt32)len; + distances[1] = delta - 1; + distances += 2; + } + } + } + } + } + while (--cutValue); + + return distances; } + +MY_FORCE_INLINE UInt32 * GetMatchesSpec1(UInt32 lenLimit, UInt32 curMatch, UInt32 pos, const Byte *cur, CLzRef *son, UInt32 _cyclicBufferPos, UInt32 _cyclicBufferSize, UInt32 cutValue, UInt32 *distances, UInt32 maxLen) { - CLzRef *ptr0 = son + (_cyclicBufferPos << 1) + 1; - CLzRef *ptr1 = son + (_cyclicBufferPos << 1); - UInt32 len0 = 0, len1 = 0; + CLzRef *ptr0 = son + ((size_t)_cyclicBufferPos << 1) + 1; + CLzRef *ptr1 = son + ((size_t)_cyclicBufferPos << 1); + unsigned len0 = 0, len1 = 0; for (;;) { UInt32 delta = pos - curMatch; @@ -401,9 +478,10 @@ UInt32 * GetMatchesSpec1(UInt32 lenLimit, UInt32 curMatch, UInt32 pos, const Byt return distances; } { - CLzRef *pair = son + ((_cyclicBufferPos - delta + ((delta > _cyclicBufferPos) ? _cyclicBufferSize : 0)) << 1); + CLzRef *pair = son + ((size_t)(_cyclicBufferPos - delta + ((delta > _cyclicBufferPos) ? _cyclicBufferSize : 0)) << 1); const Byte *pb = cur - delta; - UInt32 len = (len0 < len1 ? len0 : len1); + unsigned len = (len0 < len1 ? len0 : len1); + UInt32 pair0 = pair[0]; if (pb[len] == cur[len]) { if (++len != lenLimit && pb[len] == cur[len]) @@ -412,11 +490,12 @@ UInt32 * GetMatchesSpec1(UInt32 lenLimit, UInt32 curMatch, UInt32 pos, const Byt break; if (maxLen < len) { - *distances++ = maxLen = len; + maxLen = (UInt32)len; + *distances++ = (UInt32)len; *distances++ = delta - 1; if (len == lenLimit) { - *ptr1 = pair[0]; + *ptr1 = pair0; *ptr0 = pair[1]; return distances; } @@ -443,9 +522,9 @@ UInt32 * GetMatchesSpec1(UInt32 lenLimit, UInt32 curMatch, UInt32 pos, const Byt static void SkipMatchesSpec(UInt32 lenLimit, UInt32 curMatch, UInt32 pos, const Byte *cur, CLzRef *son, UInt32 _cyclicBufferPos, UInt32 _cyclicBufferSize, UInt32 cutValue) { - CLzRef *ptr0 = son + (_cyclicBufferPos << 1) + 1; - CLzRef *ptr1 = son + (_cyclicBufferPos << 1); - UInt32 len0 = 0, len1 = 0; + CLzRef *ptr0 = son + ((size_t)_cyclicBufferPos << 1) + 1; + CLzRef *ptr1 = son + ((size_t)_cyclicBufferPos << 1); + unsigned len0 = 0, len1 = 0; for (;;) { UInt32 delta = pos - curMatch; @@ -455,9 +534,9 @@ static void SkipMatchesSpec(UInt32 lenLimit, UInt32 curMatch, UInt32 pos, const return; } { - CLzRef *pair = son + ((_cyclicBufferPos - delta + ((delta > _cyclicBufferPos) ? _cyclicBufferSize : 0)) << 1); + CLzRef *pair = son + ((size_t)(_cyclicBufferPos - delta + ((delta > _cyclicBufferPos) ? _cyclicBufferSize : 0)) << 1); const Byte *pb = cur - delta; - UInt32 len = (len0 < len1 ? len0 : len1); + unsigned len = (len0 < len1 ? len0 : len1); if (pb[len] == cur[len]) { while (++len != lenLimit) @@ -495,13 +574,13 @@ static void SkipMatchesSpec(UInt32 lenLimit, UInt32 curMatch, UInt32 pos, const p->buffer++; \ if (++p->pos == p->posLimit) MatchFinder_CheckLimits(p); -#define MOVE_POS_RET MOVE_POS return offset; +#define MOVE_POS_RET MOVE_POS return (UInt32)offset; static void MatchFinder_MovePos(CMatchFinder *p) { MOVE_POS; } #define GET_MATCHES_HEADER2(minLen, ret_op) \ - UInt32 lenLimit; UInt32 hv; const Byte *cur; UInt32 curMatch; \ - lenLimit = p->lenLimit; { if (lenLimit < minLen) { MatchFinder_MovePos(p); ret_op; }} \ + unsigned lenLimit; UInt32 hv; const Byte *cur; UInt32 curMatch; \ + lenLimit = (unsigned)p->lenLimit; { if (lenLimit < minLen) { MatchFinder_MovePos(p); ret_op; }} \ cur = p->buffer; #define GET_MATCHES_HEADER(minLen) GET_MATCHES_HEADER2(minLen, return 0) @@ -510,22 +589,22 @@ static void MatchFinder_MovePos(CMatchFinder *p) { MOVE_POS; } #define MF_PARAMS(p) p->pos, p->buffer, p->son, p->cyclicBufferPos, p->cyclicBufferSize, p->cutValue #define GET_MATCHES_FOOTER(offset, maxLen) \ - offset = (UInt32)(GetMatchesSpec1(lenLimit, curMatch, MF_PARAMS(p), \ - distances + offset, maxLen) - distances); MOVE_POS_RET; + offset = (unsigned)(GetMatchesSpec1((UInt32)lenLimit, curMatch, MF_PARAMS(p), \ + distances + offset, (UInt32)maxLen) - distances); MOVE_POS_RET; #define SKIP_FOOTER \ - SkipMatchesSpec(lenLimit, curMatch, MF_PARAMS(p)); MOVE_POS; + SkipMatchesSpec((UInt32)lenLimit, curMatch, MF_PARAMS(p)); MOVE_POS; #define UPDATE_maxLen { \ ptrdiff_t diff = (ptrdiff_t)0 - d2; \ const Byte *c = cur + maxLen; \ const Byte *lim = cur + lenLimit; \ for (; c != lim; c++) if (*(c + diff) != *c) break; \ - maxLen = (UInt32)(c - cur); } + maxLen = (unsigned)(c - cur); } static UInt32 Bt2_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) { - UInt32 offset; + unsigned offset; GET_MATCHES_HEADER(2) HASH2_CALC; curMatch = p->hash[hv]; @@ -536,7 +615,7 @@ static UInt32 Bt2_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) UInt32 Bt3Zip_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) { - UInt32 offset; + unsigned offset; GET_MATCHES_HEADER(3) HASH_ZIP_CALC; curMatch = p->hash[hv]; @@ -547,7 +626,8 @@ UInt32 Bt3Zip_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) static UInt32 Bt3_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) { - UInt32 h2, d2, maxLen, offset, pos; + UInt32 h2, d2, pos; + unsigned maxLen, offset; UInt32 *hash; GET_MATCHES_HEADER(3) @@ -558,10 +638,10 @@ static UInt32 Bt3_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) d2 = pos - hash[h2]; - curMatch = hash[kFix3HashSize + hv]; + curMatch = (hash + kFix3HashSize)[hv]; hash[h2] = pos; - hash[kFix3HashSize + hv] = pos; + (hash + kFix3HashSize)[hv] = pos; maxLen = 2; offset = 0; @@ -569,12 +649,12 @@ static UInt32 Bt3_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) if (d2 < p->cyclicBufferSize && *(cur - d2) == *cur) { UPDATE_maxLen - distances[0] = maxLen; + distances[0] = (UInt32)maxLen; distances[1] = d2 - 1; offset = 2; if (maxLen == lenLimit) { - SkipMatchesSpec(lenLimit, curMatch, MF_PARAMS(p)); + SkipMatchesSpec((UInt32)lenLimit, curMatch, MF_PARAMS(p)); MOVE_POS_RET; } } @@ -584,7 +664,8 @@ static UInt32 Bt3_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) static UInt32 Bt4_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) { - UInt32 h2, h3, d2, d3, maxLen, offset, pos; + UInt32 h2, h3, d2, d3, pos; + unsigned maxLen, offset; UInt32 *hash; GET_MATCHES_HEADER(4) @@ -593,21 +674,22 @@ static UInt32 Bt4_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) hash = p->hash; pos = p->pos; - d2 = pos - hash[ h2]; - d3 = pos - hash[kFix3HashSize + h3]; + d2 = pos - hash [h2]; + d3 = pos - (hash + kFix3HashSize)[h3]; - curMatch = hash[kFix4HashSize + hv]; + curMatch = (hash + kFix4HashSize)[hv]; - hash[ h2] = pos; - hash[kFix3HashSize + h3] = pos; - hash[kFix4HashSize + hv] = pos; + hash [h2] = pos; + (hash + kFix3HashSize)[h3] = pos; + (hash + kFix4HashSize)[hv] = pos; maxLen = 0; offset = 0; if (d2 < p->cyclicBufferSize && *(cur - d2) == *cur) { - distances[0] = maxLen = 2; + maxLen = 2; + distances[0] = 2; distances[1] = d2 - 1; offset = 2; } @@ -615,7 +697,7 @@ static UInt32 Bt4_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) if (d2 != d3 && d3 < p->cyclicBufferSize && *(cur - d3) == *cur) { maxLen = 3; - distances[offset + 1] = d3 - 1; + distances[(size_t)offset + 1] = d3 - 1; offset += 2; d2 = d3; } @@ -623,10 +705,10 @@ static UInt32 Bt4_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) if (offset != 0) { UPDATE_maxLen - distances[offset - 2] = maxLen; + distances[(size_t)offset - 2] = (UInt32)maxLen; if (maxLen == lenLimit) { - SkipMatchesSpec(lenLimit, curMatch, MF_PARAMS(p)); + SkipMatchesSpec((UInt32)lenLimit, curMatch, MF_PARAMS(p)); MOVE_POS_RET; } } @@ -649,16 +731,16 @@ static UInt32 Bt5_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) hash = p->hash; pos = p->pos; - d2 = pos - hash[ h2]; - d3 = pos - hash[kFix3HashSize + h3]; - d4 = pos - hash[kFix4HashSize + h4]; + d2 = pos - hash [h2]; + d3 = pos - (hash + kFix3HashSize)[h3]; + d4 = pos - (hash + kFix4HashSize)[h4]; - curMatch = hash[kFix5HashSize + hv]; + curMatch = (hash + kFix5HashSize)[hv]; - hash[ h2] = pos; - hash[kFix3HashSize + h3] = pos; - hash[kFix4HashSize + h4] = pos; - hash[kFix5HashSize + hv] = pos; + hash [h2] = pos; + (hash + kFix3HashSize)[h3] = pos; + (hash + kFix4HashSize)[h4] = pos; + (hash + kFix5HashSize)[hv] = pos; maxLen = 0; offset = 0; @@ -691,7 +773,7 @@ static UInt32 Bt5_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) && *(cur - d4 + 3) == *(cur + 3)) { maxLen = 4; - distances[offset + 1] = d4 - 1; + distances[(size_t)offset + 1] = d4 - 1; offset += 2; d2 = d4; } @@ -699,7 +781,7 @@ static UInt32 Bt5_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) if (offset != 0) { UPDATE_maxLen - distances[offset - 2] = maxLen; + distances[(size_t)offset - 2] = maxLen; if (maxLen == lenLimit) { SkipMatchesSpec(lenLimit, curMatch, MF_PARAMS(p)); @@ -716,7 +798,8 @@ static UInt32 Bt5_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) static UInt32 Hc4_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) { - UInt32 h2, h3, d2, d3, maxLen, offset, pos; + UInt32 h2, h3, d2, d3, pos; + unsigned maxLen, offset; UInt32 *hash; GET_MATCHES_HEADER(4) @@ -725,21 +808,21 @@ static UInt32 Hc4_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) hash = p->hash; pos = p->pos; - d2 = pos - hash[ h2]; - d3 = pos - hash[kFix3HashSize + h3]; - - curMatch = hash[kFix4HashSize + hv]; + d2 = pos - hash [h2]; + d3 = pos - (hash + kFix3HashSize)[h3]; + curMatch = (hash + kFix4HashSize)[hv]; - hash[ h2] = pos; - hash[kFix3HashSize + h3] = pos; - hash[kFix4HashSize + hv] = pos; + hash [h2] = pos; + (hash + kFix3HashSize)[h3] = pos; + (hash + kFix4HashSize)[hv] = pos; maxLen = 0; offset = 0; if (d2 < p->cyclicBufferSize && *(cur - d2) == *cur) { - distances[0] = maxLen = 2; + maxLen = 2; + distances[0] = 2; distances[1] = d2 - 1; offset = 2; } @@ -747,7 +830,7 @@ static UInt32 Hc4_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) if (d2 != d3 && d3 < p->cyclicBufferSize && *(cur - d3) == *cur) { maxLen = 3; - distances[offset + 1] = d3 - 1; + distances[(size_t)offset + 1] = d3 - 1; offset += 2; d2 = d3; } @@ -755,7 +838,7 @@ static UInt32 Hc4_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) if (offset != 0) { UPDATE_maxLen - distances[offset - 2] = maxLen; + distances[(size_t)offset - 2] = (UInt32)maxLen; if (maxLen == lenLimit) { p->son[p->cyclicBufferPos] = curMatch; @@ -766,7 +849,7 @@ static UInt32 Hc4_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) if (maxLen < 3) maxLen = 3; - offset = (UInt32)(Hc_GetMatchesSpec(lenLimit, curMatch, MF_PARAMS(p), + offset = (unsigned)(Hc_GetMatchesSpec(lenLimit, curMatch, MF_PARAMS(p), distances + offset, maxLen) - (distances)); MOVE_POS_RET } @@ -783,16 +866,16 @@ static UInt32 Hc5_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) hash = p->hash; pos = p->pos; - d2 = pos - hash[ h2]; - d3 = pos - hash[kFix3HashSize + h3]; - d4 = pos - hash[kFix4HashSize + h4]; + d2 = pos - hash [h2]; + d3 = pos - (hash + kFix3HashSize)[h3]; + d4 = pos - (hash + kFix4HashSize)[h4]; - curMatch = hash[kFix5HashSize + hv]; + curMatch = (hash + kFix5HashSize)[hv]; - hash[ h2] = pos; - hash[kFix3HashSize + h3] = pos; - hash[kFix4HashSize + h4] = pos; - hash[kFix5HashSize + hv] = pos; + hash [h2] = pos; + (hash + kFix3HashSize)[h3] = pos; + (hash + kFix4HashSize)[h4] = pos; + (hash + kFix5HashSize)[hv] = pos; maxLen = 0; offset = 0; @@ -825,7 +908,7 @@ static UInt32 Hc5_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) && *(cur - d4 + 3) == *(cur + 3)) { maxLen = 4; - distances[offset + 1] = d4 - 1; + distances[(size_t)offset + 1] = d4 - 1; offset += 2; d2 = d4; } @@ -833,7 +916,7 @@ static UInt32 Hc5_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) if (offset != 0) { UPDATE_maxLen - distances[offset - 2] = maxLen; + distances[(size_t)offset - 2] = maxLen; if (maxLen == lenLimit) { p->son[p->cyclicBufferPos] = curMatch; @@ -852,12 +935,12 @@ static UInt32 Hc5_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) UInt32 Hc3Zip_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances) { - UInt32 offset; + unsigned offset; GET_MATCHES_HEADER(3) HASH_ZIP_CALC; curMatch = p->hash[hv]; p->hash[hv] = p->pos; - offset = (UInt32)(Hc_GetMatchesSpec(lenLimit, curMatch, MF_PARAMS(p), + offset = (unsigned)(Hc_GetMatchesSpec(lenLimit, curMatch, MF_PARAMS(p), distances, 2) - (distances)); MOVE_POS_RET } @@ -897,9 +980,9 @@ static void Bt3_MatchFinder_Skip(CMatchFinder *p, UInt32 num) SKIP_HEADER(3) HASH3_CALC; hash = p->hash; - curMatch = hash[kFix3HashSize + hv]; + curMatch = (hash + kFix3HashSize)[hv]; hash[h2] = - hash[kFix3HashSize + hv] = p->pos; + (hash + kFix3HashSize)[hv] = p->pos; SKIP_FOOTER } while (--num != 0); @@ -914,10 +997,10 @@ static void Bt4_MatchFinder_Skip(CMatchFinder *p, UInt32 num) SKIP_HEADER(4) HASH4_CALC; hash = p->hash; - curMatch = hash[kFix4HashSize + hv]; - hash[ h2] = - hash[kFix3HashSize + h3] = - hash[kFix4HashSize + hv] = p->pos; + curMatch = (hash + kFix4HashSize)[hv]; + hash [h2] = + (hash + kFix3HashSize)[h3] = + (hash + kFix4HashSize)[hv] = p->pos; SKIP_FOOTER } while (--num != 0); @@ -933,11 +1016,11 @@ static void Bt5_MatchFinder_Skip(CMatchFinder *p, UInt32 num) SKIP_HEADER(5) HASH5_CALC; hash = p->hash; - curMatch = hash[kFix5HashSize + hv]; - hash[ h2] = - hash[kFix3HashSize + h3] = - hash[kFix4HashSize + h4] = - hash[kFix5HashSize + hv] = p->pos; + curMatch = (hash + kFix5HashSize)[hv]; + hash [h2] = + (hash + kFix3HashSize)[h3] = + (hash + kFix4HashSize)[h4] = + (hash + kFix5HashSize)[hv] = p->pos; SKIP_FOOTER } while (--num != 0); @@ -953,10 +1036,10 @@ static void Hc4_MatchFinder_Skip(CMatchFinder *p, UInt32 num) SKIP_HEADER(4) HASH4_CALC; hash = p->hash; - curMatch = hash[kFix4HashSize + hv]; - hash[ h2] = - hash[kFix3HashSize + h3] = - hash[kFix4HashSize + hv] = p->pos; + curMatch = (hash + kFix4HashSize)[hv]; + hash [h2] = + (hash + kFix3HashSize)[h3] = + (hash + kFix4HashSize)[hv] = p->pos; p->son[p->cyclicBufferPos] = curMatch; MOVE_POS } @@ -973,11 +1056,11 @@ static void Hc5_MatchFinder_Skip(CMatchFinder *p, UInt32 num) SKIP_HEADER(5) HASH5_CALC; hash = p->hash; - curMatch = p->hash[kFix5HashSize + hv]; - hash[ h2] = - hash[kFix3HashSize + h3] = - hash[kFix4HashSize + h4] = - hash[kFix5HashSize + hv] = p->pos; + curMatch = hash + kFix5HashSize)[hv]; + hash [h2] = + (hash + kFix3HashSize)[h3] = + (hash + kFix4HashSize)[h4] = + (hash + kFix5HashSize)[hv] = p->pos; p->son[p->cyclicBufferPos] = curMatch; MOVE_POS } diff --git a/src/third-party/lzma/LzFind.h b/src/third-party/lzma/LzFind.h index 2ff667377..c77added7 100644 --- a/src/third-party/lzma/LzFind.h +++ b/src/third-party/lzma/LzFind.h @@ -1,5 +1,5 @@ /* LzFind.h -- Match finder for LZ algorithms -2015-10-15 : Igor Pavlov : Public domain */ +2017-06-10 : Igor Pavlov : Public domain */ #ifndef __LZ_FIND_H #define __LZ_FIND_H @@ -47,6 +47,8 @@ typedef struct _CMatchFinder SRes result; UInt32 crc[256]; size_t numRefs; + + UInt64 expectedDataSize; } CMatchFinder; #define Inline_MatchFinder_GetPointerToCurrentPos(p) ((p)->buffer) @@ -71,8 +73,8 @@ void MatchFinder_Construct(CMatchFinder *p); */ int MatchFinder_Create(CMatchFinder *p, UInt32 historySize, UInt32 keepAddBufferBefore, UInt32 matchMaxLen, UInt32 keepAddBufferAfter, - ISzAlloc *alloc); -void MatchFinder_Free(CMatchFinder *p, ISzAlloc *alloc); + ISzAllocPtr alloc); +void MatchFinder_Free(CMatchFinder *p, ISzAllocPtr alloc); void MatchFinder_Normalize3(UInt32 subValue, CLzRef *items, size_t numItems); void MatchFinder_ReduceOffsets(CMatchFinder *p, UInt32 subValue); @@ -103,7 +105,9 @@ typedef struct _IMatchFinder void MatchFinder_CreateVTable(CMatchFinder *p, IMatchFinder *vTable); -void MatchFinder_Init_2(CMatchFinder *p, int readData); +void MatchFinder_Init_LowHash(CMatchFinder *p); +void MatchFinder_Init_HighHash(CMatchFinder *p); +void MatchFinder_Init_3(CMatchFinder *p, int readData); void MatchFinder_Init(CMatchFinder *p); UInt32 Bt3Zip_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances); diff --git a/src/third-party/lzma/LzFindMt.c b/src/third-party/lzma/LzFindMt.c index cb61e0953..df32146f9 100644 --- a/src/third-party/lzma/LzFindMt.c +++ b/src/third-party/lzma/LzFindMt.c @@ -1,5 +1,5 @@ /* LzFindMt.c -- multithreaded Match finder for LZ algorithms -2015-10-15 : Igor Pavlov : Public domain */ +2018-12-29 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -33,6 +33,8 @@ static void MtSync_GetNextBlock(CMtSync *p) Event_Set(&p->canStart); Event_Wait(&p->wasStarted); + + // if (mt) MatchFinder_Init_LowHash(mt->MatchFinder); } else { @@ -155,6 +157,9 @@ static void HashThreadFunc(CMatchFinderMt *mt) UInt32 numProcessedBlocks = 0; Event_Wait(&p->canStart); Event_Set(&p->wasStarted); + + MatchFinder_Init_HighHash(mt->MatchFinder); + for (;;) { if (p->exit) @@ -205,7 +210,7 @@ static void HashThreadFunc(CMatchFinderMt *mt) if (num > kMtHashBlockSize - 2) num = kMtHashBlockSize - 2; mt->GetHeadsFunc(mf->buffer, mf->pos, mf->hash + mf->fixedHashSize, mf->hashMask, heads + 2, num, mf->crc); - heads[0] += num; + heads[0] = 2 + num; } mf->pos += num; mf->buffer += num; @@ -227,38 +232,57 @@ static void MatchFinderMt_GetNextBlock_Hash(CMatchFinderMt *p) #define kEmptyHashValue 0 -/* #define MFMT_GM_INLINE */ +#define MFMT_GM_INLINE #ifdef MFMT_GM_INLINE -#define NO_INLINE MY_FAST_CALL +/* + we use size_t for _cyclicBufferPos instead of UInt32 + to eliminate "movsx" BUG in old MSVC x64 compiler. +*/ -static Int32 NO_INLINE GetMatchesSpecN(UInt32 lenLimit, UInt32 pos, const Byte *cur, CLzRef *son, - UInt32 _cyclicBufferPos, UInt32 _cyclicBufferSize, UInt32 _cutValue, - UInt32 *_distances, UInt32 _maxLen, const UInt32 *hash, Int32 limit, UInt32 size, UInt32 *posRes) +MY_NO_INLINE +static UInt32 *GetMatchesSpecN(UInt32 lenLimit, UInt32 pos, const Byte *cur, CLzRef *son, + size_t _cyclicBufferPos, UInt32 _cyclicBufferSize, UInt32 _cutValue, + UInt32 *distances, UInt32 _maxLen, const UInt32 *hash, const UInt32 *limit, UInt32 size, UInt32 *posRes) { do { - UInt32 *distances = _distances + 1; - UInt32 curMatch = pos - *hash++; + UInt32 *_distances = ++distances; + UInt32 delta = *hash++; - CLzRef *ptr0 = son + (_cyclicBufferPos << 1) + 1; - CLzRef *ptr1 = son + (_cyclicBufferPos << 1); - UInt32 len0 = 0, len1 = 0; + CLzRef *ptr0 = son + ((size_t)_cyclicBufferPos << 1) + 1; + CLzRef *ptr1 = son + ((size_t)_cyclicBufferPos << 1); + unsigned len0 = 0, len1 = 0; UInt32 cutValue = _cutValue; - UInt32 maxLen = _maxLen; - for (;;) + unsigned maxLen = (unsigned)_maxLen; + + /* + if (size > 1) { - UInt32 delta = pos - curMatch; - if (cutValue-- == 0 || delta >= _cyclicBufferSize) + UInt32 delta = *hash; + if (delta < _cyclicBufferSize) { - *ptr0 = *ptr1 = kEmptyHashValue; - break; + UInt32 cyc1 = _cyclicBufferPos + 1; + CLzRef *pair = son + ((size_t)(cyc1 - delta + ((delta > cyc1) ? _cyclicBufferSize : 0)) << 1); + Byte b = *(cur + 1 - delta); + _distances[0] = pair[0]; + _distances[1] = b; } + } + */ + if (cutValue == 0 || delta >= _cyclicBufferSize) + { + *ptr0 = *ptr1 = kEmptyHashValue; + } + else + for(;;) + { { - CLzRef *pair = son + ((_cyclicBufferPos - delta + ((delta > _cyclicBufferPos) ? _cyclicBufferSize : 0)) << 1); + CLzRef *pair = son + ((size_t)(_cyclicBufferPos - delta + ((_cyclicBufferPos < delta) ? _cyclicBufferSize : 0)) << 1); const Byte *pb = cur - delta; - UInt32 len = (len0 < len1 ? len0 : len1); + unsigned len = (len0 < len1 ? len0 : len1); + UInt32 pair0 = *pair; if (pb[len] == cur[len]) { if (++len != lenLimit && pb[len] == cur[len]) @@ -267,54 +291,66 @@ static Int32 NO_INLINE GetMatchesSpecN(UInt32 lenLimit, UInt32 pos, const Byte * break; if (maxLen < len) { - *distances++ = maxLen = len; + maxLen = len; + *distances++ = (UInt32)len; *distances++ = delta - 1; if (len == lenLimit) { - *ptr1 = pair[0]; - *ptr0 = pair[1]; + UInt32 pair1 = pair[1]; + *ptr1 = pair0; + *ptr0 = pair1; break; } } } - if (pb[len] < cur[len]) { - *ptr1 = curMatch; - ptr1 = pair + 1; - curMatch = *ptr1; - len1 = len; - } - else - { - *ptr0 = curMatch; - ptr0 = pair; - curMatch = *ptr0; - len0 = len; + UInt32 curMatch = pos - delta; + // delta = pos - *pair; + // delta = pos - pair[((UInt32)pb[len] - (UInt32)cur[len]) >> 31]; + if (pb[len] < cur[len]) + { + delta = pos - pair[1]; + *ptr1 = curMatch; + ptr1 = pair + 1; + len1 = len; + } + else + { + delta = pos - *pair; + *ptr0 = curMatch; + ptr0 = pair; + len0 = len; + } } } + if (--cutValue == 0 || delta >= _cyclicBufferSize) + { + *ptr0 = *ptr1 = kEmptyHashValue; + break; + } } pos++; _cyclicBufferPos++; cur++; { UInt32 num = (UInt32)(distances - _distances); - *_distances = num - 1; - _distances += num; - limit -= num; + _distances[-1] = num; } } - while (limit > 0 && --size != 0); + while (distances < limit && --size != 0); *posRes = pos; - return limit; + return distances; } #endif + + static void BtGetMatches(CMatchFinderMt *p, UInt32 *distances) { UInt32 numProcessed = 0; UInt32 curPos = 2; - UInt32 limit = kMtBtBlockSize - (p->matchMaxLen * 2); + UInt32 limit = kMtBtBlockSize - (p->matchMaxLen * 2); // * 2 distances[1] = p->hashNumAvail; @@ -364,8 +400,10 @@ static void BtGetMatches(CMatchFinderMt *p, UInt32 *distances) #else { UInt32 posRes; - curPos = limit - GetMatchesSpecN(lenLimit, pos, p->buffer, p->son, cyclicBufferPos, p->cyclicBufferSize, p->cutValue, - distances + curPos, p->numHashBytes - 1, p->hashBuf + p->hashBufPos, (Int32)(limit - curPos), size, &posRes); + curPos = (UInt32)(GetMatchesSpecN(lenLimit, pos, p->buffer, p->son, cyclicBufferPos, p->cyclicBufferSize, p->cutValue, + distances + curPos, p->numHashBytes - 1, p->hashBuf + p->hashBufPos, + distances + limit, + size, &posRes) - distances); p->hashBufPos += posRes - pos; cyclicBufferPos += posRes - pos; p->buffer += posRes - pos; @@ -443,13 +481,13 @@ void MatchFinderMt_Construct(CMatchFinderMt *p) MtSync_Construct(&p->btSync); } -static void MatchFinderMt_FreeMem(CMatchFinderMt *p, ISzAlloc *alloc) +static void MatchFinderMt_FreeMem(CMatchFinderMt *p, ISzAllocPtr alloc) { - alloc->Free(alloc, p->hashBuf); + ISzAlloc_Free(alloc, p->hashBuf); p->hashBuf = NULL; } -void MatchFinderMt_Destruct(CMatchFinderMt *p, ISzAlloc *alloc) +void MatchFinderMt_Destruct(CMatchFinderMt *p, ISzAllocPtr alloc) { MtSync_Destruct(&p->hashSync); MtSync_Destruct(&p->btSync); @@ -472,7 +510,7 @@ static THREAD_FUNC_RET_TYPE THREAD_FUNC_CALL_TYPE BtThreadFunc2(void *p) } SRes MatchFinderMt_Create(CMatchFinderMt *p, UInt32 historySize, UInt32 keepAddBufferBefore, - UInt32 matchMaxLen, UInt32 keepAddBufferAfter, ISzAlloc *alloc) + UInt32 matchMaxLen, UInt32 keepAddBufferAfter, ISzAllocPtr alloc) { CMatchFinder *mf = p->MatchFinder; p->historySize = historySize; @@ -480,7 +518,7 @@ SRes MatchFinderMt_Create(CMatchFinderMt *p, UInt32 historySize, UInt32 keepAddB return SZ_ERROR_PARAM; if (!p->hashBuf) { - p->hashBuf = (UInt32 *)alloc->Alloc(alloc, (kHashBufferSize + kBtBufferSize) * sizeof(UInt32)); + p->hashBuf = (UInt32 *)ISzAlloc_Alloc(alloc, (kHashBufferSize + kBtBufferSize) * sizeof(UInt32)); if (!p->hashBuf) return SZ_ERROR_MEM; p->btBuf = p->hashBuf + kHashBufferSize; @@ -496,14 +534,18 @@ SRes MatchFinderMt_Create(CMatchFinderMt *p, UInt32 historySize, UInt32 keepAddB } /* Call it after ReleaseStream / SetStream */ -void MatchFinderMt_Init(CMatchFinderMt *p) +static void MatchFinderMt_Init(CMatchFinderMt *p) { CMatchFinder *mf = p->MatchFinder; - p->btBufPos = p->btBufPosLimit = 0; - p->hashBufPos = p->hashBufPosLimit = 0; + + p->btBufPos = + p->btBufPosLimit = 0; + p->hashBufPos = + p->hashBufPosLimit = 0; /* Init without data reading. We don't want to read data in this thread */ - MatchFinder_Init_2(mf, False); + MatchFinder_Init_3(mf, False); + MatchFinder_Init_LowHash(mf); p->pointerToCurPos = Inline_MatchFinder_GetPointerToCurrentPos(mf); p->btNumAvailBytes = 0; @@ -591,10 +633,10 @@ static UInt32 * MixMatches3(CMatchFinderMt *p, UInt32 matchMinPos, UInt32 *dista MT_HASH3_CALC curMatch2 = hash[ h2]; - curMatch3 = hash[kFix3HashSize + h3]; + curMatch3 = (hash + kFix3HashSize)[h3]; hash[ h2] = lzPos; - hash[kFix3HashSize + h3] = lzPos; + (hash + kFix3HashSize)[h3] = lzPos; if (curMatch2 >= matchMinPos && cur[(ptrdiff_t)curMatch2 - lzPos] == cur[0]) { @@ -627,12 +669,12 @@ static UInt32 *MixMatches4(CMatchFinderMt *p, UInt32 matchMinPos, UInt32 *distan MT_HASH4_CALC curMatch2 = hash[ h2]; - curMatch3 = hash[kFix3HashSize + h3]; - curMatch4 = hash[kFix4HashSize + h4]; + curMatch3 = (hash + kFix3HashSize)[h3]; + curMatch4 = (hash + kFix4HashSize)[h4]; hash[ h2] = lzPos; - hash[kFix3HashSize + h3] = lzPos; - hash[kFix4HashSize + h4] = lzPos; + (hash + kFix3HashSize)[h3] = lzPos; + (hash + kFix4HashSize)[h4] = lzPos; if (curMatch2 >= matchMinPos && cur[(ptrdiff_t)curMatch2 - lzPos] == cur[0]) { @@ -684,8 +726,12 @@ static UInt32 MatchFinderMt2_GetMatches(CMatchFinderMt *p, UInt32 *distances) UInt32 i; for (i = 0; i < len; i += 2) { - *distances++ = *btBuf++; - *distances++ = *btBuf++; + UInt32 v0 = btBuf[0]; + UInt32 v1 = btBuf[1]; + btBuf += 2; + distances[0] = v0; + distances[1] = v1; + distances += 2; } } INCREASE_LZ_POS @@ -712,8 +758,12 @@ static UInt32 MatchFinderMt_GetMatches(CMatchFinderMt *p, UInt32 *distances) distances2 = p->MixMatchesFunc(p, p->lzPos - btBuf[1], distances); do { - *distances2++ = *btBuf++; - *distances2++ = *btBuf++; + UInt32 v0 = btBuf[0]; + UInt32 v1 = btBuf[1]; + btBuf += 2; + distances2[0] = v0; + distances2[1] = v1; + distances2 += 2; } while ((len -= 2) != 0); len = (UInt32)(distances2 - (distances)); @@ -746,7 +796,7 @@ static void MatchFinderMt3_Skip(CMatchFinderMt *p, UInt32 num) SKIP_HEADER_MT(3) UInt32 h2, h3; MT_HASH3_CALC - hash[kFix3HashSize + h3] = + (hash + kFix3HashSize)[h3] = hash[ h2] = p->lzPos; SKIP_FOOTER_MT @@ -758,8 +808,8 @@ static void MatchFinderMt4_Skip(CMatchFinderMt *p, UInt32 num) SKIP_HEADER_MT(4) UInt32 h2, h3, h4; MT_HASH4_CALC - hash[kFix4HashSize + h4] = - hash[kFix3HashSize + h3] = + (hash + kFix4HashSize)[h4] = + (hash + kFix3HashSize)[h3] = hash[ h2] = p->lzPos; SKIP_FOOTER_MT @@ -777,7 +827,7 @@ void MatchFinderMt_CreateVTable(CMatchFinderMt *p, IMatchFinder *vTable) { case 2: p->GetHeadsFunc = GetHeads2; - p->MixMatchesFunc = (Mf_Mix_Matches)0; + p->MixMatchesFunc = (Mf_Mix_Matches)NULL; vTable->Skip = (Mf_Skip_Func)MatchFinderMt0_Skip; vTable->GetMatches = (Mf_GetMatches_Func)MatchFinderMt2_GetMatches; break; diff --git a/src/third-party/lzma/LzFindMt.h b/src/third-party/lzma/LzFindMt.h index 46b6924ad..fdd17008c 100644 --- a/src/third-party/lzma/LzFindMt.h +++ b/src/third-party/lzma/LzFindMt.h @@ -1,5 +1,5 @@ /* LzFindMt.h -- multithreaded Match finder for LZ algorithms -2015-05-03 : Igor Pavlov : Public domain */ +2018-07-04 : Igor Pavlov : Public domain */ #ifndef __LZ_FIND_MT_H #define __LZ_FIND_MT_H @@ -19,10 +19,10 @@ EXTERN_C_BEGIN typedef struct _CMtSync { - Bool wasCreated; - Bool needStart; - Bool exit; - Bool stopWriting; + BoolInt wasCreated; + BoolInt needStart; + BoolInt exit; + BoolInt stopWriting; CThread thread; CAutoResetEvent canStart; @@ -30,8 +30,8 @@ typedef struct _CMtSync CAutoResetEvent wasStopped; CSemaphore freeSemaphore; CSemaphore filledSemaphore; - Bool csWasInitialized; - Bool csWasEntered; + BoolInt csWasInitialized; + BoolInt csWasEntered; CCriticalSection cs; UInt32 numProcessedBlocks; } CMtSync; @@ -90,9 +90,9 @@ typedef struct _CMatchFinderMt } CMatchFinderMt; void MatchFinderMt_Construct(CMatchFinderMt *p); -void MatchFinderMt_Destruct(CMatchFinderMt *p, ISzAlloc *alloc); +void MatchFinderMt_Destruct(CMatchFinderMt *p, ISzAllocPtr alloc); SRes MatchFinderMt_Create(CMatchFinderMt *p, UInt32 historySize, UInt32 keepAddBufferBefore, - UInt32 matchMaxLen, UInt32 keepAddBufferAfter, ISzAlloc *alloc); + UInt32 matchMaxLen, UInt32 keepAddBufferAfter, ISzAllocPtr alloc); void MatchFinderMt_CreateVTable(CMatchFinderMt *p, IMatchFinder *vTable); void MatchFinderMt_ReleaseStream(CMatchFinderMt *p); diff --git a/src/third-party/lzma/Lzma2Dec.c b/src/third-party/lzma/Lzma2Dec.c index b6884571c..2e631051b 100644 --- a/src/third-party/lzma/Lzma2Dec.c +++ b/src/third-party/lzma/Lzma2Dec.c @@ -1,5 +1,5 @@ /* Lzma2Dec.c -- LZMA2 Decoder -2015-11-09 : Igor Pavlov : Public domain */ +2019-02-02 : Igor Pavlov : Public domain */ /* #define SHOW_DEBUG_INFO */ @@ -14,28 +14,22 @@ #include "Lzma2Dec.h" /* -00000000 - EOS -00000001 U U - Uncompressed Reset Dic -00000010 U U - Uncompressed No Reset -100uuuuu U U P P - LZMA no reset -101uuuuu U U P P - LZMA reset state -110uuuuu U U P P S - LZMA reset state + new prop -111uuuuu U U P P S - LZMA reset state + new prop + reset dic +00000000 - End of data +00000001 U U - Uncompressed, reset dic, need reset state and set new prop +00000010 U U - Uncompressed, no reset +100uuuuu U U P P - LZMA, no reset +101uuuuu U U P P - LZMA, reset state +110uuuuu U U P P S - LZMA, reset state + set new prop +111uuuuu U U P P S - LZMA, reset state + set new prop, reset dic u, U - Unpack Size P - Pack Size S - Props */ -#define LZMA2_CONTROL_LZMA (1 << 7) -#define LZMA2_CONTROL_COPY_NO_RESET 2 #define LZMA2_CONTROL_COPY_RESET_DIC 1 -#define LZMA2_CONTROL_EOF 0 -#define LZMA2_IS_UNCOMPRESSED_STATE(p) (((p)->control & LZMA2_CONTROL_LZMA) == 0) - -#define LZMA2_GET_LZMA_MODE(p) (((p)->control >> 5) & 3) -#define LZMA2_IS_THERE_PROP(mode) ((mode) >= 2) +#define LZMA2_IS_UNCOMPRESSED_STATE(p) (((p)->control & (1 << 7)) == 0) #define LZMA2_LCLP_MAX 4 #define LZMA2_DIC_SIZE_FROM_PROP(p) (((UInt32)2 | ((p) & 1)) << ((p) / 2 + 11)) @@ -74,14 +68,14 @@ static SRes Lzma2Dec_GetOldProps(Byte prop, Byte *props) return SZ_OK; } -SRes Lzma2Dec_AllocateProbs(CLzma2Dec *p, Byte prop, ISzAlloc *alloc) +SRes Lzma2Dec_AllocateProbs(CLzma2Dec *p, Byte prop, ISzAllocPtr alloc) { Byte props[LZMA_PROPS_SIZE]; RINOK(Lzma2Dec_GetOldProps(prop, props)); return LzmaDec_AllocateProbs(&p->decoder, props, LZMA_PROPS_SIZE, alloc); } -SRes Lzma2Dec_Allocate(CLzma2Dec *p, Byte prop, ISzAlloc *alloc) +SRes Lzma2Dec_Allocate(CLzma2Dec *p, Byte prop, ISzAllocPtr alloc) { Byte props[LZMA_PROPS_SIZE]; RINOK(Lzma2Dec_GetOldProps(prop, props)); @@ -91,9 +85,11 @@ SRes Lzma2Dec_Allocate(CLzma2Dec *p, Byte prop, ISzAlloc *alloc) void Lzma2Dec_Init(CLzma2Dec *p) { p->state = LZMA2_STATE_CONTROL; - p->needInitDic = True; - p->needInitState = True; - p->needInitProp = True; + p->needInitLevel = 0xE0; + p->isExtraMode = False; + p->unpackSize = 0; + + // p->decoder.dicPos = 0; // we can use it instead of full init LzmaDec_Init(&p->decoder); } @@ -102,19 +98,26 @@ static ELzma2State Lzma2Dec_UpdateState(CLzma2Dec *p, Byte b) switch (p->state) { case LZMA2_STATE_CONTROL: + p->isExtraMode = False; p->control = b; - PRF(printf("\n %4X ", (unsigned)p->decoder.dicPos)); - PRF(printf(" %2X", (unsigned)b)); - if (p->control == 0) + PRF(printf("\n %8X", (unsigned)p->decoder.dicPos)); + PRF(printf(" %02X", (unsigned)b)); + if (b == 0) return LZMA2_STATE_FINISHED; if (LZMA2_IS_UNCOMPRESSED_STATE(p)) { - if ((p->control & 0x7F) > 2) + if (b == LZMA2_CONTROL_COPY_RESET_DIC) + p->needInitLevel = 0xC0; + else if (b > 2 || p->needInitLevel == 0xE0) return LZMA2_STATE_ERROR; - p->unpackSize = 0; } else - p->unpackSize = (UInt32)(p->control & 0x1F) << 16; + { + if (b < p->needInitLevel) + return LZMA2_STATE_ERROR; + p->needInitLevel = 0; + p->unpackSize = (UInt32)(b & 0x1F) << 16; + } return LZMA2_STATE_UNPACK0; case LZMA2_STATE_UNPACK0: @@ -124,8 +127,8 @@ static ELzma2State Lzma2Dec_UpdateState(CLzma2Dec *p, Byte b) case LZMA2_STATE_UNPACK1: p->unpackSize |= (UInt32)b; p->unpackSize++; - PRF(printf(" %8u", (unsigned)p->unpackSize)); - return (LZMA2_IS_UNCOMPRESSED_STATE(p)) ? LZMA2_STATE_DATA : LZMA2_STATE_PACK0; + PRF(printf(" %7u", (unsigned)p->unpackSize)); + return LZMA2_IS_UNCOMPRESSED_STATE(p) ? LZMA2_STATE_DATA : LZMA2_STATE_PACK0; case LZMA2_STATE_PACK0: p->packSize = (UInt32)b << 8; @@ -134,9 +137,9 @@ static ELzma2State Lzma2Dec_UpdateState(CLzma2Dec *p, Byte b) case LZMA2_STATE_PACK1: p->packSize |= (UInt32)b; p->packSize++; - PRF(printf(" %8u", (unsigned)p->packSize)); - return LZMA2_IS_THERE_PROP(LZMA2_GET_LZMA_MODE(p)) ? LZMA2_STATE_PROP: - (p->needInitProp ? LZMA2_STATE_ERROR : LZMA2_STATE_DATA); + // if (p->packSize < 5) return LZMA2_STATE_ERROR; + PRF(printf(" %5u", (unsigned)p->packSize)); + return (p->control & 0x40) ? LZMA2_STATE_PROP : LZMA2_STATE_DATA; case LZMA2_STATE_PROP: { @@ -145,13 +148,12 @@ static ELzma2State Lzma2Dec_UpdateState(CLzma2Dec *p, Byte b) return LZMA2_STATE_ERROR; lc = b % 9; b /= 9; - p->decoder.prop.pb = b / 5; + p->decoder.prop.pb = (Byte)(b / 5); lp = b % 5; if (lc + lp > LZMA2_LCLP_MAX) return LZMA2_STATE_ERROR; - p->decoder.prop.lc = lc; - p->decoder.prop.lp = lp; - p->needInitProp = False; + p->decoder.prop.lc = (Byte)lc; + p->decoder.prop.lp = (Byte)lp; return LZMA2_STATE_DATA; } } @@ -167,7 +169,8 @@ static void LzmaDec_UpdateWithUncompressed(CLzmaDec *p, const Byte *src, SizeT s p->processedPos += (UInt32)size; } -void LzmaDec_InitDicAndState(CLzmaDec *p, Bool initDic, Bool initState); +void LzmaDec_InitDicAndState(CLzmaDec *p, BoolInt initDic, BoolInt initState); + SRes Lzma2Dec_DecodeToDic(CLzma2Dec *p, SizeT dicLimit, const Byte *src, SizeT *srcLen, ELzmaFinishMode finishMode, ELzmaStatus *status) @@ -176,12 +179,17 @@ SRes Lzma2Dec_DecodeToDic(CLzma2Dec *p, SizeT dicLimit, *srcLen = 0; *status = LZMA_STATUS_NOT_SPECIFIED; - while (p->state != LZMA2_STATE_FINISHED) + while (p->state != LZMA2_STATE_ERROR) { - SizeT dicPos = p->decoder.dicPos; + SizeT dicPos; + + if (p->state == LZMA2_STATE_FINISHED) + { + *status = LZMA_STATUS_FINISHED_WITH_MARK; + return SZ_OK; + } - if (p->state == LZMA2_STATE_ERROR) - return SZ_ERROR_DATA; + dicPos = p->decoder.dicPos; if (dicPos == dicLimit && finishMode == LZMA_FINISH_ANY) { @@ -198,29 +206,25 @@ SRes Lzma2Dec_DecodeToDic(CLzma2Dec *p, SizeT dicLimit, } (*srcLen)++; p->state = Lzma2Dec_UpdateState(p, *src++); - if (dicPos == dicLimit && p->state != LZMA2_STATE_FINISHED) - { - p->state = LZMA2_STATE_ERROR; - return SZ_ERROR_DATA; - } + break; continue; } { - SizeT destSizeCur = dicLimit - dicPos; - SizeT srcSizeCur = inSize - *srcLen; + SizeT inCur = inSize - *srcLen; + SizeT outCur = dicLimit - dicPos; ELzmaFinishMode curFinishMode = LZMA_FINISH_ANY; - if (p->unpackSize <= destSizeCur) + if (outCur >= p->unpackSize) { - destSizeCur = (SizeT)p->unpackSize; + outCur = (SizeT)p->unpackSize; curFinishMode = LZMA_FINISH_END; } if (LZMA2_IS_UNCOMPRESSED_STATE(p)) { - if (*srcLen == inSize) + if (inCur == 0) { *status = LZMA_STATUS_NEEDS_MORE_INPUT; return SZ_OK; @@ -228,135 +232,241 @@ SRes Lzma2Dec_DecodeToDic(CLzma2Dec *p, SizeT dicLimit, if (p->state == LZMA2_STATE_DATA) { - Bool initDic = (p->control == LZMA2_CONTROL_COPY_RESET_DIC); - if (initDic) - p->needInitProp = p->needInitState = True; - else if (p->needInitDic) - { - p->state = LZMA2_STATE_ERROR; - return SZ_ERROR_DATA; - } - p->needInitDic = False; + BoolInt initDic = (p->control == LZMA2_CONTROL_COPY_RESET_DIC); LzmaDec_InitDicAndState(&p->decoder, initDic, False); } - if (srcSizeCur > destSizeCur) - srcSizeCur = destSizeCur; + if (inCur > outCur) + inCur = outCur; + if (inCur == 0) + break; - if (srcSizeCur == 0) - { - p->state = LZMA2_STATE_ERROR; - return SZ_ERROR_DATA; - } + LzmaDec_UpdateWithUncompressed(&p->decoder, src, inCur); - LzmaDec_UpdateWithUncompressed(&p->decoder, src, srcSizeCur); - - src += srcSizeCur; - *srcLen += srcSizeCur; - p->unpackSize -= (UInt32)srcSizeCur; + src += inCur; + *srcLen += inCur; + p->unpackSize -= (UInt32)inCur; p->state = (p->unpackSize == 0) ? LZMA2_STATE_CONTROL : LZMA2_STATE_DATA_CONT; } else { - SizeT outSizeProcessed; SRes res; if (p->state == LZMA2_STATE_DATA) { - unsigned mode = LZMA2_GET_LZMA_MODE(p); - Bool initDic = (mode == 3); - Bool initState = (mode != 0); - if ((!initDic && p->needInitDic) || (!initState && p->needInitState)) - { - p->state = LZMA2_STATE_ERROR; - return SZ_ERROR_DATA; - } - + BoolInt initDic = (p->control >= 0xE0); + BoolInt initState = (p->control >= 0xA0); LzmaDec_InitDicAndState(&p->decoder, initDic, initState); - p->needInitDic = False; - p->needInitState = False; p->state = LZMA2_STATE_DATA_CONT; } - if (srcSizeCur > p->packSize) - srcSizeCur = (SizeT)p->packSize; - - res = LzmaDec_DecodeToDic(&p->decoder, dicPos + destSizeCur, src, &srcSizeCur, curFinishMode, status); + if (inCur > p->packSize) + inCur = (SizeT)p->packSize; - src += srcSizeCur; - *srcLen += srcSizeCur; - p->packSize -= (UInt32)srcSizeCur; + res = LzmaDec_DecodeToDic(&p->decoder, dicPos + outCur, src, &inCur, curFinishMode, status); - outSizeProcessed = p->decoder.dicPos - dicPos; - p->unpackSize -= (UInt32)outSizeProcessed; + src += inCur; + *srcLen += inCur; + p->packSize -= (UInt32)inCur; + outCur = p->decoder.dicPos - dicPos; + p->unpackSize -= (UInt32)outCur; - RINOK(res); + if (res != 0) + break; + if (*status == LZMA_STATUS_NEEDS_MORE_INPUT) - return res; + { + if (p->packSize == 0) + break; + return SZ_OK; + } - if (srcSizeCur == 0 && outSizeProcessed == 0) + if (inCur == 0 && outCur == 0) { if (*status != LZMA_STATUS_MAYBE_FINISHED_WITHOUT_MARK || p->unpackSize != 0 || p->packSize != 0) - { - p->state = LZMA2_STATE_ERROR; - return SZ_ERROR_DATA; - } + break; p->state = LZMA2_STATE_CONTROL; } - if (*status == LZMA_STATUS_MAYBE_FINISHED_WITHOUT_MARK) - *status = LZMA_STATUS_NOT_FINISHED; + *status = LZMA_STATUS_NOT_SPECIFIED; } } } - *status = LZMA_STATUS_FINISHED_WITH_MARK; - return SZ_OK; + *status = LZMA_STATUS_NOT_SPECIFIED; + p->state = LZMA2_STATE_ERROR; + return SZ_ERROR_DATA; } + + + +ELzma2ParseStatus Lzma2Dec_Parse(CLzma2Dec *p, + SizeT outSize, + const Byte *src, SizeT *srcLen, + int checkFinishBlock) +{ + SizeT inSize = *srcLen; + *srcLen = 0; + + while (p->state != LZMA2_STATE_ERROR) + { + if (p->state == LZMA2_STATE_FINISHED) + return (ELzma2ParseStatus)LZMA_STATUS_FINISHED_WITH_MARK; + + if (outSize == 0 && !checkFinishBlock) + return (ELzma2ParseStatus)LZMA_STATUS_NOT_FINISHED; + + if (p->state != LZMA2_STATE_DATA && p->state != LZMA2_STATE_DATA_CONT) + { + if (*srcLen == inSize) + return (ELzma2ParseStatus)LZMA_STATUS_NEEDS_MORE_INPUT; + (*srcLen)++; + + p->state = Lzma2Dec_UpdateState(p, *src++); + + if (p->state == LZMA2_STATE_UNPACK0) + { + // if (p->decoder.dicPos != 0) + if (p->control == LZMA2_CONTROL_COPY_RESET_DIC || p->control >= 0xE0) + return LZMA2_PARSE_STATUS_NEW_BLOCK; + // if (outSize == 0) return LZMA_STATUS_NOT_FINISHED; + } + + // The following code can be commented. + // It's not big problem, if we read additional input bytes. + // It will be stopped later in LZMA2_STATE_DATA / LZMA2_STATE_DATA_CONT state. + + if (outSize == 0 && p->state != LZMA2_STATE_FINISHED) + { + // checkFinishBlock is true. So we expect that block must be finished, + // We can return LZMA_STATUS_NOT_SPECIFIED or LZMA_STATUS_NOT_FINISHED here + // break; + return (ELzma2ParseStatus)LZMA_STATUS_NOT_FINISHED; + } + + if (p->state == LZMA2_STATE_DATA) + return LZMA2_PARSE_STATUS_NEW_CHUNK; + + continue; + } + + if (outSize == 0) + return (ELzma2ParseStatus)LZMA_STATUS_NOT_FINISHED; + + { + SizeT inCur = inSize - *srcLen; + + if (LZMA2_IS_UNCOMPRESSED_STATE(p)) + { + if (inCur == 0) + return (ELzma2ParseStatus)LZMA_STATUS_NEEDS_MORE_INPUT; + if (inCur > p->unpackSize) + inCur = p->unpackSize; + if (inCur > outSize) + inCur = outSize; + p->decoder.dicPos += inCur; + src += inCur; + *srcLen += inCur; + outSize -= inCur; + p->unpackSize -= (UInt32)inCur; + p->state = (p->unpackSize == 0) ? LZMA2_STATE_CONTROL : LZMA2_STATE_DATA_CONT; + } + else + { + p->isExtraMode = True; + + if (inCur == 0) + { + if (p->packSize != 0) + return (ELzma2ParseStatus)LZMA_STATUS_NEEDS_MORE_INPUT; + } + else if (p->state == LZMA2_STATE_DATA) + { + p->state = LZMA2_STATE_DATA_CONT; + if (*src != 0) + { + // first byte of lzma chunk must be Zero + *srcLen += 1; + p->packSize--; + break; + } + } + + if (inCur > p->packSize) + inCur = (SizeT)p->packSize; + + src += inCur; + *srcLen += inCur; + p->packSize -= (UInt32)inCur; + + if (p->packSize == 0) + { + SizeT rem = outSize; + if (rem > p->unpackSize) + rem = p->unpackSize; + p->decoder.dicPos += rem; + p->unpackSize -= (UInt32)rem; + outSize -= rem; + if (p->unpackSize == 0) + p->state = LZMA2_STATE_CONTROL; + } + } + } + } + + p->state = LZMA2_STATE_ERROR; + return (ELzma2ParseStatus)LZMA_STATUS_NOT_SPECIFIED; +} + + + + SRes Lzma2Dec_DecodeToBuf(CLzma2Dec *p, Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen, ELzmaFinishMode finishMode, ELzmaStatus *status) { SizeT outSize = *destLen, inSize = *srcLen; *srcLen = *destLen = 0; + for (;;) { - SizeT srcSizeCur = inSize, outSizeCur, dicPos; + SizeT inCur = inSize, outCur, dicPos; ELzmaFinishMode curFinishMode; SRes res; + if (p->decoder.dicPos == p->decoder.dicBufSize) p->decoder.dicPos = 0; dicPos = p->decoder.dicPos; - if (outSize > p->decoder.dicBufSize - dicPos) + curFinishMode = LZMA_FINISH_ANY; + outCur = p->decoder.dicBufSize - dicPos; + + if (outCur >= outSize) { - outSizeCur = p->decoder.dicBufSize; - curFinishMode = LZMA_FINISH_ANY; - } - else - { - outSizeCur = dicPos + outSize; + outCur = outSize; curFinishMode = finishMode; } - res = Lzma2Dec_DecodeToDic(p, outSizeCur, src, &srcSizeCur, curFinishMode, status); - src += srcSizeCur; - inSize -= srcSizeCur; - *srcLen += srcSizeCur; - outSizeCur = p->decoder.dicPos - dicPos; - memcpy(dest, p->decoder.dic + dicPos, outSizeCur); - dest += outSizeCur; - outSize -= outSizeCur; - *destLen += outSizeCur; + res = Lzma2Dec_DecodeToDic(p, dicPos + outCur, src, &inCur, curFinishMode, status); + + src += inCur; + inSize -= inCur; + *srcLen += inCur; + outCur = p->decoder.dicPos - dicPos; + memcpy(dest, p->decoder.dic + dicPos, outCur); + dest += outCur; + outSize -= outCur; + *destLen += outCur; if (res != 0) return res; - if (outSizeCur == 0 || outSize == 0) + if (outCur == 0 || outSize == 0) return SZ_OK; } } + SRes Lzma2Decode(Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen, - Byte prop, ELzmaFinishMode finishMode, ELzmaStatus *status, ISzAlloc *alloc) + Byte prop, ELzmaFinishMode finishMode, ELzmaStatus *status, ISzAllocPtr alloc) { CLzma2Dec p; SRes res; diff --git a/src/third-party/lzma/Lzma2Dec.h b/src/third-party/lzma/Lzma2Dec.h index 026cdefe9..da5038725 100644 --- a/src/third-party/lzma/Lzma2Dec.h +++ b/src/third-party/lzma/Lzma2Dec.h @@ -1,5 +1,5 @@ /* Lzma2Dec.h -- LZMA2 Decoder -2015-05-13 : Igor Pavlov : Public domain */ +2018-02-19 : Igor Pavlov : Public domain */ #ifndef __LZMA2_DEC_H #define __LZMA2_DEC_H @@ -12,25 +12,24 @@ EXTERN_C_BEGIN typedef struct { - CLzmaDec decoder; - UInt32 packSize; - UInt32 unpackSize; unsigned state; Byte control; - Bool needInitDic; - Bool needInitState; - Bool needInitProp; + Byte needInitLevel; + Byte isExtraMode; + Byte _pad_; + UInt32 packSize; + UInt32 unpackSize; + CLzmaDec decoder; } CLzma2Dec; #define Lzma2Dec_Construct(p) LzmaDec_Construct(&(p)->decoder) -#define Lzma2Dec_FreeProbs(p, alloc) LzmaDec_FreeProbs(&(p)->decoder, alloc); -#define Lzma2Dec_Free(p, alloc) LzmaDec_Free(&(p)->decoder, alloc); +#define Lzma2Dec_FreeProbs(p, alloc) LzmaDec_FreeProbs(&(p)->decoder, alloc) +#define Lzma2Dec_Free(p, alloc) LzmaDec_Free(&(p)->decoder, alloc) -SRes Lzma2Dec_AllocateProbs(CLzma2Dec *p, Byte prop, ISzAlloc *alloc); -SRes Lzma2Dec_Allocate(CLzma2Dec *p, Byte prop, ISzAlloc *alloc); +SRes Lzma2Dec_AllocateProbs(CLzma2Dec *p, Byte prop, ISzAllocPtr alloc); +SRes Lzma2Dec_Allocate(CLzma2Dec *p, Byte prop, ISzAllocPtr alloc); void Lzma2Dec_Init(CLzma2Dec *p); - /* finishMode: It has meaning only if the decoding reaches output limit (*destLen or dicLimit). @@ -53,6 +52,47 @@ SRes Lzma2Dec_DecodeToBuf(CLzma2Dec *p, Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen, ELzmaFinishMode finishMode, ELzmaStatus *status); +/* ---------- LZMA2 block and chunk parsing ---------- */ + +/* +Lzma2Dec_Parse() parses compressed data stream up to next independent block or next chunk data. +It can return LZMA_STATUS_* code or LZMA2_PARSE_STATUS_* code: + - LZMA2_PARSE_STATUS_NEW_BLOCK - there is new block, and 1 additional byte (control byte of next block header) was read from input. + - LZMA2_PARSE_STATUS_NEW_CHUNK - there is new chunk, and only lzma2 header of new chunk was read. + CLzma2Dec::unpackSize contains unpack size of that chunk +*/ + +typedef enum +{ +/* + LZMA_STATUS_NOT_SPECIFIED // data error + LZMA_STATUS_FINISHED_WITH_MARK + LZMA_STATUS_NOT_FINISHED // + LZMA_STATUS_NEEDS_MORE_INPUT + LZMA_STATUS_MAYBE_FINISHED_WITHOUT_MARK // unused +*/ + LZMA2_PARSE_STATUS_NEW_BLOCK = LZMA_STATUS_MAYBE_FINISHED_WITHOUT_MARK + 1, + LZMA2_PARSE_STATUS_NEW_CHUNK +} ELzma2ParseStatus; + +ELzma2ParseStatus Lzma2Dec_Parse(CLzma2Dec *p, + SizeT outSize, // output size + const Byte *src, SizeT *srcLen, + int checkFinishBlock // set (checkFinishBlock = 1), if it must read full input data, if decoder.dicPos reaches blockMax position. + ); + +/* +LZMA2 parser doesn't decode LZMA chunks, so we must read + full input LZMA chunk to decode some part of LZMA chunk. + +Lzma2Dec_GetUnpackExtra() returns the value that shows + max possible number of output bytes that can be output by decoder + at current input positon. +*/ + +#define Lzma2Dec_GetUnpackExtra(p) ((p)->isExtraMode ? (p)->unpackSize : 0); + + /* ---------- One Call Interface ---------- */ /* @@ -73,7 +113,7 @@ Returns: */ SRes Lzma2Decode(Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen, - Byte prop, ELzmaFinishMode finishMode, ELzmaStatus *status, ISzAlloc *alloc); + Byte prop, ELzmaFinishMode finishMode, ELzmaStatus *status, ISzAllocPtr alloc); EXTERN_C_END diff --git a/src/third-party/lzma/Lzma2DecMt.c b/src/third-party/lzma/Lzma2DecMt.c new file mode 100644 index 000000000..87d5567ad --- /dev/null +++ b/src/third-party/lzma/Lzma2DecMt.c @@ -0,0 +1,1082 @@ +/* Lzma2DecMt.c -- LZMA2 Decoder Multi-thread +2019-02-02 : Igor Pavlov : Public domain */ + +#include "Precomp.h" + +// #define SHOW_DEBUG_INFO + +#ifdef SHOW_DEBUG_INFO +#include +#endif + +#ifdef SHOW_DEBUG_INFO +#define PRF(x) x +#else +#define PRF(x) +#endif + +#define PRF_STR(s) PRF(printf("\n" s "\n")) +#define PRF_STR_INT(s, d) PRF(printf("\n" s " %d\n", (unsigned)d)) +#define PRF_STR_INT_2(s, d1, d2) PRF(printf("\n" s " %d %d\n", (unsigned)d1, (unsigned)d2)) + +// #define _7ZIP_ST + +#include "Alloc.h" + +#include "Lzma2Dec.h" +#include "Lzma2DecMt.h" + +#ifndef _7ZIP_ST +#include "MtDec.h" +#endif + + +#define LZMA2DECMT_OUT_BLOCK_MAX_DEFAULT (1 << 28) + +void Lzma2DecMtProps_Init(CLzma2DecMtProps *p) +{ + p->inBufSize_ST = 1 << 20; + p->outStep_ST = 1 << 20; + + #ifndef _7ZIP_ST + p->numThreads = 1; + p->inBufSize_MT = 1 << 18; + p->outBlockMax = LZMA2DECMT_OUT_BLOCK_MAX_DEFAULT; + p->inBlockMax = p->outBlockMax + p->outBlockMax / 16; + #endif +} + + + +#ifndef _7ZIP_ST + +/* ---------- CLzma2DecMtThread ---------- */ + +typedef struct +{ + CLzma2Dec dec; + Byte dec_created; + Byte needInit; + + Byte *outBuf; + size_t outBufSize; + + EMtDecParseState state; + ELzma2ParseStatus parseStatus; + + size_t inPreSize; + size_t outPreSize; + + size_t inCodeSize; + size_t outCodeSize; + SRes codeRes; + + CAlignOffsetAlloc alloc; + + Byte mtPad[1 << 7]; +} CLzma2DecMtThread; + +#endif + + +/* ---------- CLzma2DecMt ---------- */ + +typedef struct +{ + // ISzAllocPtr alloc; + ISzAllocPtr allocMid; + + CAlignOffsetAlloc alignOffsetAlloc; + CLzma2DecMtProps props; + Byte prop; + + ISeqInStream *inStream; + ISeqOutStream *outStream; + ICompressProgress *progress; + + BoolInt finishMode; + BoolInt outSize_Defined; + UInt64 outSize; + + UInt64 outProcessed; + UInt64 inProcessed; + BoolInt readWasFinished; + SRes readRes; + + Byte *inBuf; + size_t inBufSize; + Byte dec_created; + CLzma2Dec dec; + + size_t inPos; + size_t inLim; + + #ifndef _7ZIP_ST + UInt64 outProcessed_Parse; + BoolInt mtc_WasConstructed; + CMtDec mtc; + CLzma2DecMtThread coders[MTDEC__THREADS_MAX]; + #endif + +} CLzma2DecMt; + + + +CLzma2DecMtHandle Lzma2DecMt_Create(ISzAllocPtr alloc, ISzAllocPtr allocMid) +{ + CLzma2DecMt *p = (CLzma2DecMt *)ISzAlloc_Alloc(alloc, sizeof(CLzma2DecMt)); + if (!p) + return NULL; + + // p->alloc = alloc; + p->allocMid = allocMid; + + AlignOffsetAlloc_CreateVTable(&p->alignOffsetAlloc); + p->alignOffsetAlloc.numAlignBits = 7; + p->alignOffsetAlloc.offset = 0; + p->alignOffsetAlloc.baseAlloc = alloc; + + p->inBuf = NULL; + p->inBufSize = 0; + p->dec_created = False; + + // Lzma2DecMtProps_Init(&p->props); + + #ifndef _7ZIP_ST + p->mtc_WasConstructed = False; + { + unsigned i; + for (i = 0; i < MTDEC__THREADS_MAX; i++) + { + CLzma2DecMtThread *t = &p->coders[i]; + t->dec_created = False; + t->outBuf = NULL; + t->outBufSize = 0; + } + } + #endif + + return p; +} + + +#ifndef _7ZIP_ST + +static void Lzma2DecMt_FreeOutBufs(CLzma2DecMt *p) +{ + unsigned i; + for (i = 0; i < MTDEC__THREADS_MAX; i++) + { + CLzma2DecMtThread *t = &p->coders[i]; + if (t->outBuf) + { + ISzAlloc_Free(p->allocMid, t->outBuf); + t->outBuf = NULL; + t->outBufSize = 0; + } + } +} + +#endif + + +static void Lzma2DecMt_FreeSt(CLzma2DecMt *p) +{ + if (p->dec_created) + { + Lzma2Dec_Free(&p->dec, &p->alignOffsetAlloc.vt); + p->dec_created = False; + } + if (p->inBuf) + { + ISzAlloc_Free(p->allocMid, p->inBuf); + p->inBuf = NULL; + } + p->inBufSize = 0; +} + + +void Lzma2DecMt_Destroy(CLzma2DecMtHandle pp) +{ + CLzma2DecMt *p = (CLzma2DecMt *)pp; + + Lzma2DecMt_FreeSt(p); + + #ifndef _7ZIP_ST + + if (p->mtc_WasConstructed) + { + MtDec_Destruct(&p->mtc); + p->mtc_WasConstructed = False; + } + { + unsigned i; + for (i = 0; i < MTDEC__THREADS_MAX; i++) + { + CLzma2DecMtThread *t = &p->coders[i]; + if (t->dec_created) + { + // we don't need to free dict here + Lzma2Dec_FreeProbs(&t->dec, &t->alloc.vt); // p->alloc !!! + t->dec_created = False; + } + } + } + Lzma2DecMt_FreeOutBufs(p); + + #endif + + ISzAlloc_Free(p->alignOffsetAlloc.baseAlloc, pp); +} + + + +#ifndef _7ZIP_ST + +static void Lzma2DecMt_MtCallback_Parse(void *obj, unsigned coderIndex, CMtDecCallbackInfo *cc) +{ + CLzma2DecMt *me = (CLzma2DecMt *)obj; + CLzma2DecMtThread *t = &me->coders[coderIndex]; + + PRF_STR_INT_2("Parse", coderIndex, cc->srcSize); + + cc->state = MTDEC_PARSE_CONTINUE; + + if (cc->startCall) + { + if (!t->dec_created) + { + Lzma2Dec_Construct(&t->dec); + t->dec_created = True; + AlignOffsetAlloc_CreateVTable(&t->alloc); + { + /* (1 << 12) is expected size of one way in data cache. + We optimize alignment for cache line size of 128 bytes and smaller */ + const unsigned kNumAlignBits = 12; + const unsigned kNumCacheLineBits = 7; /* <= kNumAlignBits */ + t->alloc.numAlignBits = kNumAlignBits; + t->alloc.offset = ((UInt32)coderIndex * ((1 << 11) + (1 << 8) + (1 << 6))) & ((1 << kNumAlignBits) - (1 << kNumCacheLineBits)); + t->alloc.baseAlloc = me->alignOffsetAlloc.baseAlloc; + } + } + Lzma2Dec_Init(&t->dec); + + t->inPreSize = 0; + t->outPreSize = 0; + // t->blockWasFinished = False; + // t->finishedWithMark = False; + t->parseStatus = (ELzma2ParseStatus)LZMA_STATUS_NOT_SPECIFIED; + t->state = MTDEC_PARSE_CONTINUE; + + t->inCodeSize = 0; + t->outCodeSize = 0; + t->codeRes = SZ_OK; + + // (cc->srcSize == 0) is allowed + } + + { + ELzma2ParseStatus status; + BoolInt overflow; + UInt32 unpackRem = 0; + + int checkFinishBlock = True; + size_t limit = me->props.outBlockMax; + if (me->outSize_Defined) + { + UInt64 rem = me->outSize - me->outProcessed_Parse; + if (limit >= rem) + { + limit = (size_t)rem; + if (!me->finishMode) + checkFinishBlock = False; + } + } + + // checkFinishBlock = False, if we want to decode partial data + // that must be finished at position <= outBlockMax. + + { + const SizeT srcOrig = cc->srcSize; + SizeT srcSize_Point = 0; + SizeT dicPos_Point = 0; + + cc->srcSize = 0; + overflow = False; + + for (;;) + { + SizeT srcCur = srcOrig - cc->srcSize; + + status = Lzma2Dec_Parse(&t->dec, + limit - t->dec.decoder.dicPos, + cc->src + cc->srcSize, &srcCur, + checkFinishBlock); + + cc->srcSize += srcCur; + + if (status == LZMA2_PARSE_STATUS_NEW_CHUNK) + { + if (t->dec.unpackSize > me->props.outBlockMax - t->dec.decoder.dicPos) + { + overflow = True; + break; + } + continue; + } + + if (status == LZMA2_PARSE_STATUS_NEW_BLOCK) + { + if (t->dec.decoder.dicPos == 0) + continue; + // we decode small blocks in one thread + if (t->dec.decoder.dicPos >= (1 << 14)) + break; + dicPos_Point = t->dec.decoder.dicPos; + srcSize_Point = cc->srcSize; + continue; + } + + if ((int)status == LZMA_STATUS_NOT_FINISHED && checkFinishBlock + // && limit == t->dec.decoder.dicPos + // && limit == me->props.outBlockMax + ) + { + overflow = True; + break; + } + + unpackRem = Lzma2Dec_GetUnpackExtra(&t->dec); + break; + } + + if (dicPos_Point != 0 + && (int)status != LZMA2_PARSE_STATUS_NEW_BLOCK + && (int)status != LZMA_STATUS_FINISHED_WITH_MARK + && (int)status != LZMA_STATUS_NOT_SPECIFIED) + { + // we revert to latest newBlock state + status = LZMA2_PARSE_STATUS_NEW_BLOCK; + unpackRem = 0; + t->dec.decoder.dicPos = dicPos_Point; + cc->srcSize = srcSize_Point; + overflow = False; + } + } + + t->inPreSize += cc->srcSize; + t->parseStatus = status; + + if (overflow) + cc->state = MTDEC_PARSE_OVERFLOW; + else + { + size_t dicPos = t->dec.decoder.dicPos; + + if ((int)status != LZMA_STATUS_NEEDS_MORE_INPUT) + { + if (status == LZMA2_PARSE_STATUS_NEW_BLOCK) + { + cc->state = MTDEC_PARSE_NEW; + cc->srcSize--; // we don't need control byte of next block + t->inPreSize--; + } + else + { + cc->state = MTDEC_PARSE_END; + if ((int)status != LZMA_STATUS_FINISHED_WITH_MARK) + { + // (status == LZMA_STATUS_NOT_SPECIFIED) + // (status == LZMA_STATUS_NOT_FINISHED) + if (unpackRem != 0) + { + /* we also reserve space for max possible number of output bytes of current LZMA chunk */ + SizeT rem = limit - dicPos; + if (rem > unpackRem) + rem = unpackRem; + dicPos += rem; + } + } + } + + me->outProcessed_Parse += dicPos; + } + + cc->outPos = dicPos; + t->outPreSize = (size_t)dicPos; + } + + t->state = cc->state; + return; + } +} + + +static SRes Lzma2DecMt_MtCallback_PreCode(void *pp, unsigned coderIndex) +{ + CLzma2DecMt *me = (CLzma2DecMt *)pp; + CLzma2DecMtThread *t = &me->coders[coderIndex]; + Byte *dest = t->outBuf; + + if (t->inPreSize == 0) + { + t->codeRes = SZ_ERROR_DATA; + return t->codeRes; + } + + if (!dest || t->outBufSize < t->outPreSize) + { + if (dest) + { + ISzAlloc_Free(me->allocMid, dest); + t->outBuf = NULL; + t->outBufSize = 0; + } + + dest = (Byte *)ISzAlloc_Alloc(me->allocMid, t->outPreSize + // + (1 << 28) + ); + // Sleep(200); + if (!dest) + return SZ_ERROR_MEM; + t->outBuf = dest; + t->outBufSize = t->outPreSize; + } + + t->dec.decoder.dic = dest; + t->dec.decoder.dicBufSize = t->outPreSize; + + t->needInit = True; + + return Lzma2Dec_AllocateProbs(&t->dec, me->prop, &t->alloc.vt); // alloc.vt +} + + +static SRes Lzma2DecMt_MtCallback_Code(void *pp, unsigned coderIndex, + const Byte *src, size_t srcSize, int srcFinished, + // int finished, int blockFinished, + UInt64 *inCodePos, UInt64 *outCodePos, int *stop) +{ + CLzma2DecMt *me = (CLzma2DecMt *)pp; + CLzma2DecMtThread *t = &me->coders[coderIndex]; + + UNUSED_VAR(srcFinished) + + PRF_STR_INT_2("Code", coderIndex, srcSize); + + *inCodePos = t->inCodeSize; + *outCodePos = 0; + *stop = True; + + if (t->needInit) + { + Lzma2Dec_Init(&t->dec); + t->needInit = False; + } + + { + ELzmaStatus status; + size_t srcProcessed = srcSize; + BoolInt blockWasFinished = + ((int)t->parseStatus == LZMA_STATUS_FINISHED_WITH_MARK + || t->parseStatus == LZMA2_PARSE_STATUS_NEW_BLOCK); + + SRes res = Lzma2Dec_DecodeToDic(&t->dec, + t->outPreSize, + src, &srcProcessed, + blockWasFinished ? LZMA_FINISH_END : LZMA_FINISH_ANY, + &status); + + t->codeRes = res; + + t->inCodeSize += srcProcessed; + *inCodePos = t->inCodeSize; + t->outCodeSize = t->dec.decoder.dicPos; + *outCodePos = t->dec.decoder.dicPos; + + if (res != SZ_OK) + return res; + + if (srcProcessed == srcSize) + *stop = False; + + if (blockWasFinished) + { + if (srcSize != srcProcessed) + return SZ_ERROR_FAIL; + + if (t->inPreSize == t->inCodeSize) + { + if (t->outPreSize != t->outCodeSize) + return SZ_ERROR_FAIL; + *stop = True; + } + } + else + { + if (t->outPreSize == t->outCodeSize) + *stop = True; + } + + return SZ_OK; + } +} + + +#define LZMA2DECMT_STREAM_WRITE_STEP (1 << 24) + +static SRes Lzma2DecMt_MtCallback_Write(void *pp, unsigned coderIndex, + BoolInt needWriteToStream, + const Byte *src, size_t srcSize, + BoolInt *needContinue, BoolInt *canRecode) +{ + CLzma2DecMt *me = (CLzma2DecMt *)pp; + const CLzma2DecMtThread *t = &me->coders[coderIndex]; + size_t size = t->outCodeSize; + const Byte *data = t->outBuf; + BoolInt needContinue2 = True; + + PRF_STR_INT_2("Write", coderIndex, srcSize); + + *needContinue = False; + *canRecode = True; + UNUSED_VAR(src) + UNUSED_VAR(srcSize) + + if ( + // t->parseStatus == LZMA_STATUS_FINISHED_WITH_MARK + t->state == MTDEC_PARSE_OVERFLOW + || t->state == MTDEC_PARSE_END) + needContinue2 = False; + + + if (!needWriteToStream) + return SZ_OK; + + me->mtc.inProcessed += t->inCodeSize; + + if (t->codeRes == SZ_OK) + if ((int)t->parseStatus == LZMA_STATUS_FINISHED_WITH_MARK + || t->parseStatus == LZMA2_PARSE_STATUS_NEW_BLOCK) + if (t->outPreSize != t->outCodeSize + || t->inPreSize != t->inCodeSize) + return SZ_ERROR_FAIL; + + *canRecode = False; + + if (me->outStream) + { + for (;;) + { + size_t cur = size; + size_t written; + if (cur > LZMA2DECMT_STREAM_WRITE_STEP) + cur = LZMA2DECMT_STREAM_WRITE_STEP; + + written = ISeqOutStream_Write(me->outStream, data, cur); + + me->outProcessed += written; + // me->mtc.writtenTotal += written; + if (written != cur) + return SZ_ERROR_WRITE; + data += cur; + size -= cur; + if (size == 0) + { + *needContinue = needContinue2; + return SZ_OK; + } + RINOK(MtProgress_ProgressAdd(&me->mtc.mtProgress, 0, 0)); + } + } + + return SZ_ERROR_FAIL; + /* + if (size > me->outBufSize) + return SZ_ERROR_OUTPUT_EOF; + memcpy(me->outBuf, data, size); + me->outBufSize -= size; + me->outBuf += size; + *needContinue = needContinue2; + return SZ_OK; + */ +} + +#endif + + +static SRes Lzma2Dec_Prepare_ST(CLzma2DecMt *p) +{ + if (!p->dec_created) + { + Lzma2Dec_Construct(&p->dec); + p->dec_created = True; + } + + RINOK(Lzma2Dec_Allocate(&p->dec, p->prop, &p->alignOffsetAlloc.vt)); + + if (!p->inBuf || p->inBufSize != p->props.inBufSize_ST) + { + ISzAlloc_Free(p->allocMid, p->inBuf); + p->inBufSize = 0; + p->inBuf = (Byte *)ISzAlloc_Alloc(p->allocMid, p->props.inBufSize_ST); + if (!p->inBuf) + return SZ_ERROR_MEM; + p->inBufSize = p->props.inBufSize_ST; + } + + Lzma2Dec_Init(&p->dec); + + return SZ_OK; +} + + +static SRes Lzma2Dec_Decode_ST(CLzma2DecMt *p + #ifndef _7ZIP_ST + , BoolInt tMode + #endif + ) +{ + SizeT wrPos; + size_t inPos, inLim; + const Byte *inData; + UInt64 inPrev, outPrev; + + CLzma2Dec *dec; + + #ifndef _7ZIP_ST + if (tMode) + { + Lzma2DecMt_FreeOutBufs(p); + tMode = MtDec_PrepareRead(&p->mtc); + } + #endif + + RINOK(Lzma2Dec_Prepare_ST(p)); + + dec = &p->dec; + + inPrev = p->inProcessed; + outPrev = p->outProcessed; + + inPos = 0; + inLim = 0; + inData = NULL; + wrPos = dec->decoder.dicPos; + + for (;;) + { + SizeT dicPos; + SizeT size; + ELzmaFinishMode finishMode; + SizeT inProcessed; + ELzmaStatus status; + SRes res; + + SizeT outProcessed; + BoolInt outFinished; + BoolInt needStop; + + if (inPos == inLim) + { + #ifndef _7ZIP_ST + if (tMode) + { + inData = MtDec_Read(&p->mtc, &inLim); + inPos = 0; + if (inData) + continue; + tMode = False; + inLim = 0; + } + #endif + + if (!p->readWasFinished) + { + inPos = 0; + inLim = p->inBufSize; + inData = p->inBuf; + p->readRes = ISeqInStream_Read(p->inStream, (void *)inData, &inLim); + // p->readProcessed += inLim; + // inLim -= 5; p->readWasFinished = True; // for test + if (inLim == 0 || p->readRes != SZ_OK) + p->readWasFinished = True; + } + } + + dicPos = dec->decoder.dicPos; + { + SizeT next = dec->decoder.dicBufSize; + if (next - wrPos > p->props.outStep_ST) + next = wrPos + p->props.outStep_ST; + size = next - dicPos; + } + + finishMode = LZMA_FINISH_ANY; + if (p->outSize_Defined) + { + const UInt64 rem = p->outSize - p->outProcessed; + if (size >= rem) + { + size = (SizeT)rem; + if (p->finishMode) + finishMode = LZMA_FINISH_END; + } + } + + inProcessed = inLim - inPos; + + res = Lzma2Dec_DecodeToDic(dec, dicPos + size, inData + inPos, &inProcessed, finishMode, &status); + + inPos += inProcessed; + p->inProcessed += inProcessed; + outProcessed = dec->decoder.dicPos - dicPos; + p->outProcessed += outProcessed; + + outFinished = (p->outSize_Defined && p->outSize <= p->outProcessed); + + needStop = (res != SZ_OK + || (inProcessed == 0 && outProcessed == 0) + || status == LZMA_STATUS_FINISHED_WITH_MARK + || (!p->finishMode && outFinished)); + + if (needStop || outProcessed >= size) + { + SRes res2; + { + size_t writeSize = dec->decoder.dicPos - wrPos; + size_t written = ISeqOutStream_Write(p->outStream, dec->decoder.dic + wrPos, writeSize); + res2 = (written == writeSize) ? SZ_OK : SZ_ERROR_WRITE; + } + + if (dec->decoder.dicPos == dec->decoder.dicBufSize) + dec->decoder.dicPos = 0; + wrPos = dec->decoder.dicPos; + + RINOK(res2); + + if (needStop) + { + if (res != SZ_OK) + return res; + + if (status == LZMA_STATUS_FINISHED_WITH_MARK) + { + if (p->finishMode) + { + if (p->outSize_Defined && p->outSize != p->outProcessed) + return SZ_ERROR_DATA; + } + return SZ_OK; + } + + if (!p->finishMode && outFinished) + return SZ_OK; + + if (status == LZMA_STATUS_NEEDS_MORE_INPUT) + return SZ_ERROR_INPUT_EOF; + + return SZ_ERROR_DATA; + } + } + + if (p->progress) + { + UInt64 inDelta = p->inProcessed - inPrev; + UInt64 outDelta = p->outProcessed - outPrev; + if (inDelta >= (1 << 22) || outDelta >= (1 << 22)) + { + RINOK(ICompressProgress_Progress(p->progress, p->inProcessed, p->outProcessed)); + inPrev = p->inProcessed; + outPrev = p->outProcessed; + } + } + } +} + + + +SRes Lzma2DecMt_Decode(CLzma2DecMtHandle pp, + Byte prop, + const CLzma2DecMtProps *props, + ISeqOutStream *outStream, const UInt64 *outDataSize, int finishMode, + // Byte *outBuf, size_t *outBufSize, + ISeqInStream *inStream, + // const Byte *inData, size_t inDataSize, + UInt64 *inProcessed, + // UInt64 *outProcessed, + int *isMT, + ICompressProgress *progress) +{ + CLzma2DecMt *p = (CLzma2DecMt *)pp; + #ifndef _7ZIP_ST + BoolInt tMode; + #endif + + *inProcessed = 0; + + if (prop > 40) + return SZ_ERROR_UNSUPPORTED; + + p->prop = prop; + p->props = *props; + + p->inStream = inStream; + p->outStream = outStream; + p->progress = progress; + + p->outSize = 0; + p->outSize_Defined = False; + if (outDataSize) + { + p->outSize_Defined = True; + p->outSize = *outDataSize; + } + p->finishMode = finishMode; + + p->outProcessed = 0; + p->inProcessed = 0; + + p->readWasFinished = False; + + *isMT = False; + + + #ifndef _7ZIP_ST + + tMode = False; + + // p->mtc.parseRes = SZ_OK; + + // p->mtc.numFilledThreads = 0; + // p->mtc.crossStart = 0; + // p->mtc.crossEnd = 0; + // p->mtc.allocError_for_Read_BlockIndex = 0; + // p->mtc.isAllocError = False; + + if (p->props.numThreads > 1) + { + IMtDecCallback vt; + + Lzma2DecMt_FreeSt(p); + + p->outProcessed_Parse = 0; + + if (!p->mtc_WasConstructed) + { + p->mtc_WasConstructed = True; + MtDec_Construct(&p->mtc); + } + + p->mtc.progress = progress; + p->mtc.inStream = inStream; + + // p->outBuf = NULL; + // p->outBufSize = 0; + /* + if (!outStream) + { + // p->outBuf = outBuf; + // p->outBufSize = *outBufSize; + // *outBufSize = 0; + return SZ_ERROR_PARAM; + } + */ + + // p->mtc.inBlockMax = p->props.inBlockMax; + p->mtc.alloc = &p->alignOffsetAlloc.vt; + // p->alignOffsetAlloc.baseAlloc; + // p->mtc.inData = inData; + // p->mtc.inDataSize = inDataSize; + p->mtc.mtCallback = &vt; + p->mtc.mtCallbackObject = p; + + p->mtc.inBufSize = p->props.inBufSize_MT; + + p->mtc.numThreadsMax = p->props.numThreads; + + *isMT = True; + + vt.Parse = Lzma2DecMt_MtCallback_Parse; + vt.PreCode = Lzma2DecMt_MtCallback_PreCode; + vt.Code = Lzma2DecMt_MtCallback_Code; + vt.Write = Lzma2DecMt_MtCallback_Write; + + { + BoolInt needContinue = False; + + SRes res = MtDec_Code(&p->mtc); + + /* + if (!outStream) + *outBufSize = p->outBuf - outBuf; + */ + + *inProcessed = p->mtc.inProcessed; + + needContinue = False; + + if (res == SZ_OK) + { + if (p->mtc.mtProgress.res != SZ_OK) + res = p->mtc.mtProgress.res; + else + needContinue = p->mtc.needContinue; + } + + if (!needContinue) + { + if (res == SZ_OK) + return p->mtc.readRes; + return res; + } + + tMode = True; + p->readRes = p->mtc.readRes; + p->readWasFinished = p->mtc.readWasFinished; + p->inProcessed = p->mtc.inProcessed; + + PRF_STR("----- decoding ST -----"); + } + } + + #endif + + + *isMT = False; + + { + SRes res = Lzma2Dec_Decode_ST(p + #ifndef _7ZIP_ST + , tMode + #endif + ); + + *inProcessed = p->inProcessed; + + // res = SZ_OK; // for test + if (res == SZ_OK && p->readRes != SZ_OK) + res = p->readRes; + + /* + #ifndef _7ZIP_ST + if (res == SZ_OK && tMode && p->mtc.parseRes != SZ_OK) + res = p->mtc.parseRes; + #endif + */ + + return res; + } +} + + +/* ---------- Read from CLzma2DecMtHandle Interface ---------- */ + +SRes Lzma2DecMt_Init(CLzma2DecMtHandle pp, + Byte prop, + const CLzma2DecMtProps *props, + const UInt64 *outDataSize, int finishMode, + ISeqInStream *inStream) +{ + CLzma2DecMt *p = (CLzma2DecMt *)pp; + + if (prop > 40) + return SZ_ERROR_UNSUPPORTED; + + p->prop = prop; + p->props = *props; + + p->inStream = inStream; + + p->outSize = 0; + p->outSize_Defined = False; + if (outDataSize) + { + p->outSize_Defined = True; + p->outSize = *outDataSize; + } + p->finishMode = finishMode; + + p->outProcessed = 0; + p->inProcessed = 0; + + p->inPos = 0; + p->inLim = 0; + + return Lzma2Dec_Prepare_ST(p); +} + + +SRes Lzma2DecMt_Read(CLzma2DecMtHandle pp, + Byte *data, size_t *outSize, + UInt64 *inStreamProcessed) +{ + CLzma2DecMt *p = (CLzma2DecMt *)pp; + ELzmaFinishMode finishMode; + SRes readRes; + size_t size = *outSize; + + *outSize = 0; + *inStreamProcessed = 0; + + finishMode = LZMA_FINISH_ANY; + if (p->outSize_Defined) + { + const UInt64 rem = p->outSize - p->outProcessed; + if (size >= rem) + { + size = (size_t)rem; + if (p->finishMode) + finishMode = LZMA_FINISH_END; + } + } + + readRes = SZ_OK; + + for (;;) + { + SizeT inCur; + SizeT outCur; + ELzmaStatus status; + SRes res; + + if (p->inPos == p->inLim && readRes == SZ_OK) + { + p->inPos = 0; + p->inLim = p->props.inBufSize_ST; + readRes = ISeqInStream_Read(p->inStream, p->inBuf, &p->inLim); + } + + inCur = p->inLim - p->inPos; + outCur = size; + + res = Lzma2Dec_DecodeToBuf(&p->dec, data, &outCur, + p->inBuf + p->inPos, &inCur, finishMode, &status); + + p->inPos += inCur; + p->inProcessed += inCur; + *inStreamProcessed += inCur; + p->outProcessed += outCur; + *outSize += outCur; + size -= outCur; + data += outCur; + + if (res != 0) + return res; + + /* + if (status == LZMA_STATUS_FINISHED_WITH_MARK) + return readRes; + + if (size == 0 && status != LZMA_STATUS_NEEDS_MORE_INPUT) + { + if (p->finishMode && p->outSize_Defined && p->outProcessed >= p->outSize) + return SZ_ERROR_DATA; + return readRes; + } + */ + + if (inCur == 0 && outCur == 0) + return readRes; + } +} diff --git a/src/third-party/lzma/Lzma2DecMt.h b/src/third-party/lzma/Lzma2DecMt.h new file mode 100644 index 000000000..96f89a3ca --- /dev/null +++ b/src/third-party/lzma/Lzma2DecMt.h @@ -0,0 +1,79 @@ +/* Lzma2DecMt.h -- LZMA2 Decoder Multi-thread +2018-02-17 : Igor Pavlov : Public domain */ + +#ifndef __LZMA2_DEC_MT_H +#define __LZMA2_DEC_MT_H + +#include "7zTypes.h" + +EXTERN_C_BEGIN + +typedef struct +{ + size_t inBufSize_ST; + size_t outStep_ST; + + #ifndef _7ZIP_ST + unsigned numThreads; + size_t inBufSize_MT; + size_t outBlockMax; + size_t inBlockMax; + #endif +} CLzma2DecMtProps; + +/* init to single-thread mode */ +void Lzma2DecMtProps_Init(CLzma2DecMtProps *p); + + +/* ---------- CLzma2DecMtHandle Interface ---------- */ + +/* Lzma2DecMt_ * functions can return the following exit codes: +SRes: + SZ_OK - OK + SZ_ERROR_MEM - Memory allocation error + SZ_ERROR_PARAM - Incorrect paramater in props + SZ_ERROR_WRITE - ISeqOutStream write callback error + // SZ_ERROR_OUTPUT_EOF - output buffer overflow - version with (Byte *) output + SZ_ERROR_PROGRESS - some break from progress callback + SZ_ERROR_THREAD - error in multithreading functions (only for Mt version) +*/ + +typedef void * CLzma2DecMtHandle; + +CLzma2DecMtHandle Lzma2DecMt_Create(ISzAllocPtr alloc, ISzAllocPtr allocMid); +void Lzma2DecMt_Destroy(CLzma2DecMtHandle p); + +SRes Lzma2DecMt_Decode(CLzma2DecMtHandle p, + Byte prop, + const CLzma2DecMtProps *props, + ISeqOutStream *outStream, + const UInt64 *outDataSize, // NULL means undefined + int finishMode, // 0 - partial unpacking is allowed, 1 - if lzma2 stream must be finished + // Byte *outBuf, size_t *outBufSize, + ISeqInStream *inStream, + // const Byte *inData, size_t inDataSize, + + // out variables: + UInt64 *inProcessed, + int *isMT, /* out: (*isMT == 0), if single thread decoding was used */ + + // UInt64 *outProcessed, + ICompressProgress *progress); + + +/* ---------- Read from CLzma2DecMtHandle Interface ---------- */ + +SRes Lzma2DecMt_Init(CLzma2DecMtHandle pp, + Byte prop, + const CLzma2DecMtProps *props, + const UInt64 *outDataSize, int finishMode, + ISeqInStream *inStream); + +SRes Lzma2DecMt_Read(CLzma2DecMtHandle pp, + Byte *data, size_t *outSize, + UInt64 *inStreamProcessed); + + +EXTERN_C_END + +#endif diff --git a/src/third-party/lzma/Lzma2Enc.c b/src/third-party/lzma/Lzma2Enc.c index cba013497..d54147752 100644 --- a/src/third-party/lzma/Lzma2Enc.c +++ b/src/third-party/lzma/Lzma2Enc.c @@ -1,9 +1,8 @@ /* Lzma2Enc.c -- LZMA2 Encoder -2015-10-04 : Igor Pavlov : Public domain */ +2018-07-04 : Igor Pavlov : Public domain */ #include "Precomp.h" -/* #include */ #include /* #define _7ZIP_ST */ @@ -13,7 +12,7 @@ #ifndef _7ZIP_ST #include "MtCoder.h" #else -#define NUM_MT_CODER_THREADS_MAX 1 +#define MTCODER__THREADS_MAX 1 #endif #define LZMA2_CONTROL_LZMA (1 << 7) @@ -35,41 +34,97 @@ #define PRF(x) /* x */ + +/* ---------- CLimitedSeqInStream ---------- */ + +typedef struct +{ + ISeqInStream vt; + ISeqInStream *realStream; + UInt64 limit; + UInt64 processed; + int finished; +} CLimitedSeqInStream; + +static void LimitedSeqInStream_Init(CLimitedSeqInStream *p) +{ + p->limit = (UInt64)(Int64)-1; + p->processed = 0; + p->finished = 0; +} + +static SRes LimitedSeqInStream_Read(const ISeqInStream *pp, void *data, size_t *size) +{ + CLimitedSeqInStream *p = CONTAINER_FROM_VTBL(pp, CLimitedSeqInStream, vt); + size_t size2 = *size; + SRes res = SZ_OK; + + if (p->limit != (UInt64)(Int64)-1) + { + UInt64 rem = p->limit - p->processed; + if (size2 > rem) + size2 = (size_t)rem; + } + if (size2 != 0) + { + res = ISeqInStream_Read(p->realStream, data, &size2); + p->finished = (size2 == 0 ? 1 : 0); + p->processed += size2; + } + *size = size2; + return res; +} + + /* ---------- CLzma2EncInt ---------- */ typedef struct { CLzmaEncHandle enc; + Byte propsAreSet; + Byte propsByte; + Byte needInitState; + Byte needInitProp; UInt64 srcPos; - Byte props; - Bool needInitState; - Bool needInitProp; } CLzma2EncInt; -static SRes Lzma2EncInt_Init(CLzma2EncInt *p, const CLzma2EncProps *props) + +static SRes Lzma2EncInt_InitStream(CLzma2EncInt *p, const CLzma2EncProps *props) { - Byte propsEncoded[LZMA_PROPS_SIZE]; - SizeT propsSize = LZMA_PROPS_SIZE; - RINOK(LzmaEnc_SetProps(p->enc, &props->lzmaProps)); - RINOK(LzmaEnc_WriteProperties(p->enc, propsEncoded, &propsSize)); - p->srcPos = 0; - p->props = propsEncoded[0]; - p->needInitState = True; - p->needInitProp = True; + if (!p->propsAreSet) + { + SizeT propsSize = LZMA_PROPS_SIZE; + Byte propsEncoded[LZMA_PROPS_SIZE]; + RINOK(LzmaEnc_SetProps(p->enc, &props->lzmaProps)); + RINOK(LzmaEnc_WriteProperties(p->enc, propsEncoded, &propsSize)); + p->propsByte = propsEncoded[0]; + p->propsAreSet = True; + } return SZ_OK; } +static void Lzma2EncInt_InitBlock(CLzma2EncInt *p) +{ + p->srcPos = 0; + p->needInitState = True; + p->needInitProp = True; +} + + SRes LzmaEnc_PrepareForLzma2(CLzmaEncHandle pp, ISeqInStream *inStream, UInt32 keepWindowSize, - ISzAlloc *alloc, ISzAlloc *allocBig); + ISzAllocPtr alloc, ISzAllocPtr allocBig); SRes LzmaEnc_MemPrepare(CLzmaEncHandle pp, const Byte *src, SizeT srcLen, - UInt32 keepWindowSize, ISzAlloc *alloc, ISzAlloc *allocBig); -SRes LzmaEnc_CodeOneMemBlock(CLzmaEncHandle pp, Bool reInit, + UInt32 keepWindowSize, ISzAllocPtr alloc, ISzAllocPtr allocBig); +SRes LzmaEnc_CodeOneMemBlock(CLzmaEncHandle pp, BoolInt reInit, Byte *dest, size_t *destLen, UInt32 desiredPackSize, UInt32 *unpackSize); const Byte *LzmaEnc_GetCurBuf(CLzmaEncHandle pp); void LzmaEnc_Finish(CLzmaEncHandle pp); void LzmaEnc_SaveState(CLzmaEncHandle pp); void LzmaEnc_RestoreState(CLzmaEncHandle pp); +/* +UInt32 LzmaEnc_GetNumAvailableBytes(CLzmaEncHandle pp); +*/ static SRes Lzma2EncInt_EncodeSubblock(CLzma2EncInt *p, Byte *outBuf, size_t *packSizeRes, ISeqOutStream *outStream) @@ -78,7 +133,7 @@ static SRes Lzma2EncInt_EncodeSubblock(CLzma2EncInt *p, Byte *outBuf, size_t packSize = packSizeLimit; UInt32 unpackSize = LZMA2_UNPACK_SIZE_MAX; unsigned lzHeaderSize = 5 + (p->needInitProp ? 1 : 0); - Bool useCopyBlock; + BoolInt useCopyBlock; SRes res; *packSizeRes = 0; @@ -126,7 +181,7 @@ static SRes Lzma2EncInt_EncodeSubblock(CLzma2EncInt *p, Byte *outBuf, if (outStream) { *packSizeRes += destPos; - if (outStream->Write(outStream, outBuf, destPos) != destPos) + if (ISeqOutStream_Write(outStream, outBuf, destPos) != destPos) return SZ_ERROR_WRITE; destPos = 0; } @@ -154,7 +209,7 @@ static SRes Lzma2EncInt_EncodeSubblock(CLzma2EncInt *p, Byte *outBuf, outBuf[destPos++] = (Byte)pm; if (p->needInitProp) - outBuf[destPos++] = p->props; + outBuf[destPos++] = p->propsByte; p->needInitProp = False; p->needInitState = False; @@ -162,7 +217,7 @@ static SRes Lzma2EncInt_EncodeSubblock(CLzma2EncInt *p, Byte *outBuf, p->srcPos += unpackSize; if (outStream) - if (outStream->Write(outStream, outBuf, destPos) != destPos) + if (ISeqOutStream_Write(outStream, outBuf, destPos) != destPos) return SZ_ERROR_WRITE; *packSizeRes = destPos; @@ -176,14 +231,16 @@ static SRes Lzma2EncInt_EncodeSubblock(CLzma2EncInt *p, Byte *outBuf, void Lzma2EncProps_Init(CLzma2EncProps *p) { LzmaEncProps_Init(&p->lzmaProps); + p->blockSize = LZMA2_ENC_PROPS__BLOCK_SIZE__AUTO; + p->numBlockThreads_Reduced = -1; + p->numBlockThreads_Max = -1; p->numTotalThreads = -1; - p->numBlockThreads = -1; - p->blockSize = 0; } void Lzma2EncProps_Normalize(CLzma2EncProps *p) { - int t1, t1n, t2, t3; + UInt64 fileSize; + int t1, t1n, t2, t2r, t3; { CLzmaEncProps lzmaProps = p->lzmaProps; LzmaEncProps_Normalize(&lzmaProps); @@ -191,11 +248,11 @@ void Lzma2EncProps_Normalize(CLzma2EncProps *p) } t1 = p->lzmaProps.numThreads; - t2 = p->numBlockThreads; + t2 = p->numBlockThreads_Max; t3 = p->numTotalThreads; - if (t2 > NUM_MT_CODER_THREADS_MAX) - t2 = NUM_MT_CODER_THREADS_MAX; + if (t2 > MTCODER__THREADS_MAX) + t2 = MTCODER__THREADS_MAX; if (t3 <= 0) { @@ -211,8 +268,8 @@ void Lzma2EncProps_Normalize(CLzma2EncProps *p) t1 = 1; t2 = t3; } - if (t2 > NUM_MT_CODER_THREADS_MAX) - t2 = NUM_MT_CODER_THREADS_MAX; + if (t2 > MTCODER__THREADS_MAX) + t2 = MTCODER__THREADS_MAX; } else if (t1 <= 0) { @@ -225,46 +282,71 @@ void Lzma2EncProps_Normalize(CLzma2EncProps *p) p->lzmaProps.numThreads = t1; + t2r = t2; + + fileSize = p->lzmaProps.reduceSize; + + if ( p->blockSize != LZMA2_ENC_PROPS__BLOCK_SIZE__SOLID + && p->blockSize != LZMA2_ENC_PROPS__BLOCK_SIZE__AUTO + && (p->blockSize < fileSize || fileSize == (UInt64)(Int64)-1)) + p->lzmaProps.reduceSize = p->blockSize; + LzmaEncProps_Normalize(&p->lzmaProps); + p->lzmaProps.reduceSize = fileSize; + t1 = p->lzmaProps.numThreads; - if (p->blockSize == 0) + if (p->blockSize == LZMA2_ENC_PROPS__BLOCK_SIZE__SOLID) { - UInt32 dictSize = p->lzmaProps.dictSize; - UInt64 blockSize = (UInt64)dictSize << 2; - const UInt32 kMinSize = (UInt32)1 << 20; - const UInt32 kMaxSize = (UInt32)1 << 28; - if (blockSize < kMinSize) blockSize = kMinSize; - if (blockSize > kMaxSize) blockSize = kMaxSize; - if (blockSize < dictSize) blockSize = dictSize; - p->blockSize = (size_t)blockSize; + t2r = t2 = 1; + t3 = t1; } - - if (t2 > 1 && p->lzmaProps.reduceSize != (UInt64)(Int64)-1) + else if (p->blockSize == LZMA2_ENC_PROPS__BLOCK_SIZE__AUTO && t2 <= 1) { - UInt64 temp = p->lzmaProps.reduceSize + p->blockSize - 1; - if (temp > p->lzmaProps.reduceSize) + /* if there is no block multi-threading, we use SOLID block */ + p->blockSize = LZMA2_ENC_PROPS__BLOCK_SIZE__SOLID; + } + else + { + if (p->blockSize == LZMA2_ENC_PROPS__BLOCK_SIZE__AUTO) { - UInt64 numBlocks = temp / p->blockSize; + const UInt32 kMinSize = (UInt32)1 << 20; + const UInt32 kMaxSize = (UInt32)1 << 28; + const UInt32 dictSize = p->lzmaProps.dictSize; + UInt64 blockSize = (UInt64)dictSize << 2; + if (blockSize < kMinSize) blockSize = kMinSize; + if (blockSize > kMaxSize) blockSize = kMaxSize; + if (blockSize < dictSize) blockSize = dictSize; + blockSize += (kMinSize - 1); + blockSize &= ~(UInt64)(kMinSize - 1); + p->blockSize = blockSize; + } + + if (t2 > 1 && fileSize != (UInt64)(Int64)-1) + { + UInt64 numBlocks = fileSize / p->blockSize; + if (numBlocks * p->blockSize != fileSize) + numBlocks++; if (numBlocks < (unsigned)t2) { - t2 = (unsigned)numBlocks; - if (t2 == 0) - t2 = 1; - t3 = t1 * t2; + t2r = (unsigned)numBlocks; + if (t2r == 0) + t2r = 1; + t3 = t1 * t2r; } } } - p->numBlockThreads = t2; + p->numBlockThreads_Max = t2; + p->numBlockThreads_Reduced = t2r; p->numTotalThreads = t3; } static SRes Progress(ICompressProgress *p, UInt64 inSize, UInt64 outSize) { - return (p && p->Progress(p, inSize, outSize) != SZ_OK) ? SZ_ERROR_PROGRESS : SZ_OK; + return (p && ICompressProgress_Progress(p, inSize, outSize) != SZ_OK) ? SZ_ERROR_PROGRESS : SZ_OK; } @@ -274,180 +356,112 @@ typedef struct { Byte propEncoded; CLzma2EncProps props; + UInt64 expectedDataSize; - Byte *outBuf; + Byte *tempBufLzma; - ISzAlloc *alloc; - ISzAlloc *allocBig; + ISzAllocPtr alloc; + ISzAllocPtr allocBig; - CLzma2EncInt coders[NUM_MT_CODER_THREADS_MAX]; + CLzma2EncInt coders[MTCODER__THREADS_MAX]; #ifndef _7ZIP_ST + + ISeqOutStream *outStream; + Byte *outBuf; + size_t outBuf_Rem; /* remainder in outBuf */ + + size_t outBufSize; /* size of allocated outBufs[i] */ + size_t outBufsDataSizes[MTCODER__BLOCKS_MAX]; + BoolInt mtCoder_WasConstructed; CMtCoder mtCoder; + Byte *outBufs[MTCODER__BLOCKS_MAX]; + #endif } CLzma2Enc; -/* ---------- Lzma2EncThread ---------- */ -static SRes Lzma2Enc_EncodeMt1(CLzma2EncInt *p, CLzma2Enc *mainEncoder, - ISeqOutStream *outStream, ISeqInStream *inStream, ICompressProgress *progress) +CLzma2EncHandle Lzma2Enc_Create(ISzAllocPtr alloc, ISzAllocPtr allocBig) { - UInt64 packTotal = 0; - SRes res = SZ_OK; - - if (!mainEncoder->outBuf) - { - mainEncoder->outBuf = (Byte *)IAlloc_Alloc(mainEncoder->alloc, LZMA2_CHUNK_SIZE_COMPRESSED_MAX); - if (!mainEncoder->outBuf) - return SZ_ERROR_MEM; - } - - RINOK(Lzma2EncInt_Init(p, &mainEncoder->props)); - RINOK(LzmaEnc_PrepareForLzma2(p->enc, inStream, LZMA2_KEEP_WINDOW_SIZE, - mainEncoder->alloc, mainEncoder->allocBig)); - - for (;;) - { - size_t packSize = LZMA2_CHUNK_SIZE_COMPRESSED_MAX; - res = Lzma2EncInt_EncodeSubblock(p, mainEncoder->outBuf, &packSize, outStream); - if (res != SZ_OK) - break; - packTotal += packSize; - res = Progress(progress, p->srcPos, packTotal); - if (res != SZ_OK) - break; - if (packSize == 0) - break; - } - - LzmaEnc_Finish(p->enc); - - if (res == SZ_OK) - { - Byte b = 0; - if (outStream->Write(outStream, &b, 1) != 1) - return SZ_ERROR_WRITE; - } - - return res; -} - - -#ifndef _7ZIP_ST - -typedef struct -{ - IMtCoderCallback funcTable; - CLzma2Enc *lzma2Enc; -} CMtCallbackImp; - -static SRes MtCallbackImp_Code(void *pp, unsigned index, Byte *dest, size_t *destSize, - const Byte *src, size_t srcSize, int finished) -{ - CMtCallbackImp *imp = (CMtCallbackImp *)pp; - CLzma2Enc *mainEncoder = imp->lzma2Enc; - CLzma2EncInt *p = &mainEncoder->coders[index]; - - SRes res = SZ_OK; - { - size_t destLim = *destSize; - *destSize = 0; - - if (srcSize != 0) - { - RINOK(Lzma2EncInt_Init(p, &mainEncoder->props)); - - RINOK(LzmaEnc_MemPrepare(p->enc, src, srcSize, LZMA2_KEEP_WINDOW_SIZE, - mainEncoder->alloc, mainEncoder->allocBig)); - - while (p->srcPos < srcSize) - { - size_t packSize = destLim - *destSize; - res = Lzma2EncInt_EncodeSubblock(p, dest + *destSize, &packSize, NULL); - if (res != SZ_OK) - break; - *destSize += packSize; - - if (packSize == 0) - { - res = SZ_ERROR_FAIL; - break; - } - - if (MtProgress_Set(&mainEncoder->mtCoder.mtProgress, index, p->srcPos, *destSize) != SZ_OK) - { - res = SZ_ERROR_PROGRESS; - break; - } - } - - LzmaEnc_Finish(p->enc); - if (res != SZ_OK) - return res; - } - - if (finished) - { - if (*destSize == destLim) - return SZ_ERROR_OUTPUT_EOF; - dest[(*destSize)++] = 0; - } - } - return res; -} - -#endif - - -/* ---------- Lzma2Enc ---------- */ - -CLzma2EncHandle Lzma2Enc_Create(ISzAlloc *alloc, ISzAlloc *allocBig) -{ - CLzma2Enc *p = (CLzma2Enc *)alloc->Alloc(alloc, sizeof(CLzma2Enc)); + CLzma2Enc *p = (CLzma2Enc *)ISzAlloc_Alloc(alloc, sizeof(CLzma2Enc)); if (!p) return NULL; Lzma2EncProps_Init(&p->props); Lzma2EncProps_Normalize(&p->props); - p->outBuf = 0; + p->expectedDataSize = (UInt64)(Int64)-1; + p->tempBufLzma = NULL; p->alloc = alloc; p->allocBig = allocBig; { unsigned i; - for (i = 0; i < NUM_MT_CODER_THREADS_MAX; i++) - p->coders[i].enc = 0; + for (i = 0; i < MTCODER__THREADS_MAX; i++) + p->coders[i].enc = NULL; } #ifndef _7ZIP_ST - MtCoder_Construct(&p->mtCoder); + p->mtCoder_WasConstructed = False; + { + unsigned i; + for (i = 0; i < MTCODER__BLOCKS_MAX; i++) + p->outBufs[i] = NULL; + p->outBufSize = 0; + } #endif return p; } + +#ifndef _7ZIP_ST + +static void Lzma2Enc_FreeOutBufs(CLzma2Enc *p) +{ + unsigned i; + for (i = 0; i < MTCODER__BLOCKS_MAX; i++) + if (p->outBufs[i]) + { + ISzAlloc_Free(p->alloc, p->outBufs[i]); + p->outBufs[i] = NULL; + } + p->outBufSize = 0; +} + +#endif + + void Lzma2Enc_Destroy(CLzma2EncHandle pp) { CLzma2Enc *p = (CLzma2Enc *)pp; unsigned i; - for (i = 0; i < NUM_MT_CODER_THREADS_MAX; i++) + for (i = 0; i < MTCODER__THREADS_MAX; i++) { CLzma2EncInt *t = &p->coders[i]; if (t->enc) { LzmaEnc_Destroy(t->enc, p->alloc, p->allocBig); - t->enc = 0; + t->enc = NULL; } } + #ifndef _7ZIP_ST - MtCoder_Destruct(&p->mtCoder); + if (p->mtCoder_WasConstructed) + { + MtCoder_Destruct(&p->mtCoder); + p->mtCoder_WasConstructed = False; + } + Lzma2Enc_FreeOutBufs(p); #endif - IAlloc_Free(p->alloc, p->outBuf); - IAlloc_Free(p->alloc, pp); + ISzAlloc_Free(p->alloc, p->tempBufLzma); + p->tempBufLzma = NULL; + + ISzAlloc_Free(p->alloc, pp); } + SRes Lzma2Enc_SetProps(CLzma2EncHandle pp, const CLzma2EncProps *props) { CLzma2Enc *p = (CLzma2Enc *)pp; @@ -460,6 +474,14 @@ SRes Lzma2Enc_SetProps(CLzma2EncHandle pp, const CLzma2EncProps *props) return SZ_OK; } + +void Lzma2Enc_SetDataSize(CLzmaEncHandle pp, UInt64 expectedDataSiize) +{ + CLzma2Enc *p = (CLzma2Enc *)pp; + p->expectedDataSize = expectedDataSiize; +} + + Byte Lzma2Enc_WriteProperties(CLzma2EncHandle pp) { CLzma2Enc *p = (CLzma2Enc *)pp; @@ -471,50 +493,311 @@ Byte Lzma2Enc_WriteProperties(CLzma2EncHandle pp) return (Byte)i; } -SRes Lzma2Enc_Encode(CLzma2EncHandle pp, - ISeqOutStream *outStream, ISeqInStream *inStream, ICompressProgress *progress) -{ - CLzma2Enc *p = (CLzma2Enc *)pp; - int i; - for (i = 0; i < p->props.numBlockThreads; i++) +static SRes Lzma2Enc_EncodeMt1( + CLzma2Enc *me, + CLzma2EncInt *p, + ISeqOutStream *outStream, + Byte *outBuf, size_t *outBufSize, + ISeqInStream *inStream, + const Byte *inData, size_t inDataSize, + int finished, + ICompressProgress *progress) +{ + UInt64 unpackTotal = 0; + UInt64 packTotal = 0; + size_t outLim = 0; + CLimitedSeqInStream limitedInStream; + + if (outBuf) { - CLzma2EncInt *t = &p->coders[(unsigned)i]; - if (!t->enc) + outLim = *outBufSize; + *outBufSize = 0; + } + + if (!p->enc) + { + p->propsAreSet = False; + p->enc = LzmaEnc_Create(me->alloc); + if (!p->enc) + return SZ_ERROR_MEM; + } + + limitedInStream.realStream = inStream; + if (inStream) + { + limitedInStream.vt.Read = LimitedSeqInStream_Read; + } + + if (!outBuf) + { + // outStream version works only in one thread. So we use CLzma2Enc::tempBufLzma + if (!me->tempBufLzma) { - t->enc = LzmaEnc_Create(p->alloc); - if (!t->enc) + me->tempBufLzma = (Byte *)ISzAlloc_Alloc(me->alloc, LZMA2_CHUNK_SIZE_COMPRESSED_MAX); + if (!me->tempBufLzma) return SZ_ERROR_MEM; } } - #ifndef _7ZIP_ST - if (p->props.numBlockThreads > 1) - { - CMtCallbackImp mtCallback; + RINOK(Lzma2EncInt_InitStream(p, &me->props)); - mtCallback.funcTable.Code = MtCallbackImp_Code; - mtCallback.lzma2Enc = p; + for (;;) + { + SRes res = SZ_OK; + size_t inSizeCur = 0; + + Lzma2EncInt_InitBlock(p); + LimitedSeqInStream_Init(&limitedInStream); + limitedInStream.limit = me->props.blockSize; + + if (inStream) + { + UInt64 expected = (UInt64)(Int64)-1; + // inStream version works only in one thread. So we use CLzma2Enc::expectedDataSize + if (me->expectedDataSize != (UInt64)(Int64)-1 + && me->expectedDataSize >= unpackTotal) + expected = me->expectedDataSize - unpackTotal; + if (me->props.blockSize != LZMA2_ENC_PROPS__BLOCK_SIZE__SOLID + && expected > me->props.blockSize) + expected = (size_t)me->props.blockSize; + + LzmaEnc_SetDataSize(p->enc, expected); + + RINOK(LzmaEnc_PrepareForLzma2(p->enc, + &limitedInStream.vt, + LZMA2_KEEP_WINDOW_SIZE, + me->alloc, + me->allocBig)); + } + else + { + inSizeCur = inDataSize - (size_t)unpackTotal; + if (me->props.blockSize != LZMA2_ENC_PROPS__BLOCK_SIZE__SOLID + && inSizeCur > me->props.blockSize) + inSizeCur = (size_t)me->props.blockSize; + + // LzmaEnc_SetDataSize(p->enc, inSizeCur); + + RINOK(LzmaEnc_MemPrepare(p->enc, + inData + (size_t)unpackTotal, inSizeCur, + LZMA2_KEEP_WINDOW_SIZE, + me->alloc, + me->allocBig)); + } + + for (;;) + { + size_t packSize = LZMA2_CHUNK_SIZE_COMPRESSED_MAX; + if (outBuf) + packSize = outLim - (size_t)packTotal; + + res = Lzma2EncInt_EncodeSubblock(p, + outBuf ? outBuf + (size_t)packTotal : me->tempBufLzma, &packSize, + outBuf ? NULL : outStream); + + if (res != SZ_OK) + break; + + packTotal += packSize; + if (outBuf) + *outBufSize = (size_t)packTotal; + + res = Progress(progress, unpackTotal + p->srcPos, packTotal); + if (res != SZ_OK) + break; + + /* + if (LzmaEnc_GetNumAvailableBytes(p->enc) == 0) + break; + */ + + if (packSize == 0) + break; + } + + LzmaEnc_Finish(p->enc); + + unpackTotal += p->srcPos; + + RINOK(res); + + if (p->srcPos != (inStream ? limitedInStream.processed : inSizeCur)) + return SZ_ERROR_FAIL; + + if (inStream ? limitedInStream.finished : (unpackTotal == inDataSize)) + { + if (finished) + { + if (outBuf) + { + size_t destPos = *outBufSize; + if (destPos >= outLim) + return SZ_ERROR_OUTPUT_EOF; + outBuf[destPos] = 0; + *outBufSize = destPos + 1; + } + else + { + Byte b = 0; + if (ISeqOutStream_Write(outStream, &b, 1) != 1) + return SZ_ERROR_WRITE; + } + } + return SZ_OK; + } + } +} + + + +#ifndef _7ZIP_ST + +static SRes Lzma2Enc_MtCallback_Code(void *pp, unsigned coderIndex, unsigned outBufIndex, + const Byte *src, size_t srcSize, int finished) +{ + CLzma2Enc *me = (CLzma2Enc *)pp; + size_t destSize = me->outBufSize; + SRes res; + CMtProgressThunk progressThunk; + + Byte *dest = me->outBufs[outBufIndex]; + + me->outBufsDataSizes[outBufIndex] = 0; + + if (!dest) + { + dest = (Byte *)ISzAlloc_Alloc(me->alloc, me->outBufSize); + if (!dest) + return SZ_ERROR_MEM; + me->outBufs[outBufIndex] = dest; + } + + MtProgressThunk_CreateVTable(&progressThunk); + progressThunk.mtProgress = &me->mtCoder.mtProgress; + progressThunk.inSize = 0; + progressThunk.outSize = 0; + + res = Lzma2Enc_EncodeMt1(me, + &me->coders[coderIndex], + NULL, dest, &destSize, + NULL, src, srcSize, + finished, + &progressThunk.vt); + + me->outBufsDataSizes[outBufIndex] = destSize; + + return res; +} + + +static SRes Lzma2Enc_MtCallback_Write(void *pp, unsigned outBufIndex) +{ + CLzma2Enc *me = (CLzma2Enc *)pp; + size_t size = me->outBufsDataSizes[outBufIndex]; + const Byte *data = me->outBufs[outBufIndex]; + + if (me->outStream) + return ISeqOutStream_Write(me->outStream, data, size) == size ? SZ_OK : SZ_ERROR_WRITE; + + if (size > me->outBuf_Rem) + return SZ_ERROR_OUTPUT_EOF; + memcpy(me->outBuf, data, size); + me->outBuf_Rem -= size; + me->outBuf += size; + return SZ_OK; +} + +#endif + + + +SRes Lzma2Enc_Encode2(CLzma2EncHandle pp, + ISeqOutStream *outStream, + Byte *outBuf, size_t *outBufSize, + ISeqInStream *inStream, + const Byte *inData, size_t inDataSize, + ICompressProgress *progress) +{ + CLzma2Enc *p = (CLzma2Enc *)pp; + + if (inStream && inData) + return SZ_ERROR_PARAM; + + if (outStream && outBuf) + return SZ_ERROR_PARAM; + + { + unsigned i; + for (i = 0; i < MTCODER__THREADS_MAX; i++) + p->coders[i].propsAreSet = False; + } + + #ifndef _7ZIP_ST + + if (p->props.numBlockThreads_Reduced > 1) + { + IMtCoderCallback2 vt; + + if (!p->mtCoder_WasConstructed) + { + p->mtCoder_WasConstructed = True; + MtCoder_Construct(&p->mtCoder); + } + + vt.Code = Lzma2Enc_MtCallback_Code; + vt.Write = Lzma2Enc_MtCallback_Write; + + p->outStream = outStream; + p->outBuf = NULL; + p->outBuf_Rem = 0; + if (!outStream) + { + p->outBuf = outBuf; + p->outBuf_Rem = *outBufSize; + *outBufSize = 0; + } + + p->mtCoder.allocBig = p->allocBig; p->mtCoder.progress = progress; p->mtCoder.inStream = inStream; - p->mtCoder.outStream = outStream; - p->mtCoder.alloc = p->alloc; - p->mtCoder.mtCallback = &mtCallback.funcTable; + p->mtCoder.inData = inData; + p->mtCoder.inDataSize = inDataSize; + p->mtCoder.mtCallback = &vt; + p->mtCoder.mtCallbackObject = p; + + p->mtCoder.blockSize = (size_t)p->props.blockSize; + if (p->mtCoder.blockSize != p->props.blockSize) + return SZ_ERROR_PARAM; /* SZ_ERROR_MEM */ - p->mtCoder.blockSize = p->props.blockSize; - p->mtCoder.destBlockSize = p->props.blockSize + (p->props.blockSize >> 10) + 16; - if (p->mtCoder.destBlockSize < p->props.blockSize) { - p->mtCoder.destBlockSize = (size_t)0 - 1; - if (p->mtCoder.destBlockSize < p->props.blockSize) - return SZ_ERROR_FAIL; + size_t destBlockSize = p->mtCoder.blockSize + (p->mtCoder.blockSize >> 10) + 16; + if (destBlockSize < p->mtCoder.blockSize) + return SZ_ERROR_PARAM; + if (p->outBufSize != destBlockSize) + Lzma2Enc_FreeOutBufs(p); + p->outBufSize = destBlockSize; } - p->mtCoder.numThreads = p->props.numBlockThreads; + + p->mtCoder.numThreadsMax = p->props.numBlockThreads_Max; + p->mtCoder.expectedDataSize = p->expectedDataSize; - return MtCoder_Code(&p->mtCoder); + { + SRes res = MtCoder_Code(&p->mtCoder); + if (!outStream) + *outBufSize = p->outBuf - outBuf; + return res; + } } + #endif - return Lzma2Enc_EncodeMt1(&p->coders[0], p, outStream, inStream, progress); + + return Lzma2Enc_EncodeMt1(p, + &p->coders[0], + outStream, outBuf, outBufSize, + inStream, inData, inDataSize, + True, /* finished */ + progress); } diff --git a/src/third-party/lzma/Lzma2Enc.h b/src/third-party/lzma/Lzma2Enc.h index 061178aaf..65f2dd145 100644 --- a/src/third-party/lzma/Lzma2Enc.h +++ b/src/third-party/lzma/Lzma2Enc.h @@ -1,5 +1,5 @@ /* Lzma2Enc.h -- LZMA2 Encoder -2013-01-18 : Igor Pavlov : Public domain */ +2017-07-27 : Igor Pavlov : Public domain */ #ifndef __LZMA2_ENC_H #define __LZMA2_ENC_H @@ -8,11 +8,15 @@ EXTERN_C_BEGIN +#define LZMA2_ENC_PROPS__BLOCK_SIZE__AUTO 0 +#define LZMA2_ENC_PROPS__BLOCK_SIZE__SOLID ((UInt64)(Int64)-1) + typedef struct { CLzmaEncProps lzmaProps; - size_t blockSize; - int numBlockThreads; + UInt64 blockSize; + int numBlockThreads_Reduced; + int numBlockThreads_Max; int numTotalThreads; } CLzma2EncProps; @@ -22,40 +26,29 @@ void Lzma2EncProps_Normalize(CLzma2EncProps *p); /* ---------- CLzmaEnc2Handle Interface ---------- */ /* Lzma2Enc_* functions can return the following exit codes: -Returns: +SRes: SZ_OK - OK SZ_ERROR_MEM - Memory allocation error SZ_ERROR_PARAM - Incorrect paramater in props - SZ_ERROR_WRITE - Write callback error + SZ_ERROR_WRITE - ISeqOutStream write callback error + SZ_ERROR_OUTPUT_EOF - output buffer overflow - version with (Byte *) output SZ_ERROR_PROGRESS - some break from progress callback - SZ_ERROR_THREAD - errors in multithreading functions (only for Mt version) + SZ_ERROR_THREAD - error in multithreading functions (only for Mt version) */ typedef void * CLzma2EncHandle; -CLzma2EncHandle Lzma2Enc_Create(ISzAlloc *alloc, ISzAlloc *allocBig); +CLzma2EncHandle Lzma2Enc_Create(ISzAllocPtr alloc, ISzAllocPtr allocBig); void Lzma2Enc_Destroy(CLzma2EncHandle p); SRes Lzma2Enc_SetProps(CLzma2EncHandle p, const CLzma2EncProps *props); +void Lzma2Enc_SetDataSize(CLzma2EncHandle p, UInt64 expectedDataSiize); Byte Lzma2Enc_WriteProperties(CLzma2EncHandle p); -SRes Lzma2Enc_Encode(CLzma2EncHandle p, - ISeqOutStream *outStream, ISeqInStream *inStream, ICompressProgress *progress); - -/* ---------- One Call Interface ---------- */ - -/* Lzma2Encode -Return code: - SZ_OK - OK - SZ_ERROR_MEM - Memory allocation error - SZ_ERROR_PARAM - Incorrect paramater - SZ_ERROR_OUTPUT_EOF - output buffer overflow - SZ_ERROR_THREAD - errors in multithreading functions (only for Mt version) -*/ - -/* -SRes Lzma2Encode(Byte *dest, SizeT *destLen, const Byte *src, SizeT srcLen, - const CLzmaEncProps *props, Byte *propsEncoded, int writeEndMark, - ICompressProgress *progress, ISzAlloc *alloc, ISzAlloc *allocBig); -*/ +SRes Lzma2Enc_Encode2(CLzma2EncHandle p, + ISeqOutStream *outStream, + Byte *outBuf, size_t *outBufSize, + ISeqInStream *inStream, + const Byte *inData, size_t inDataSize, + ICompressProgress *progress); EXTERN_C_END diff --git a/src/third-party/lzma/Lzma86Dec.c b/src/third-party/lzma/Lzma86Dec.c index 0e49d7be9..20ac5e7a9 100644 --- a/src/third-party/lzma/Lzma86Dec.c +++ b/src/third-party/lzma/Lzma86Dec.c @@ -1,5 +1,5 @@ /* Lzma86Dec.c -- LZMA + x86 (BCJ) Filter Decoder -2015-11-08 : Igor Pavlov : Public domain */ +2016-05-16 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -9,9 +9,6 @@ #include "Bra.h" #include "LzmaDec.h" -static void *SzAlloc(void *p, size_t size) { UNUSED_VAR(p); return MyAlloc(size); } -static void SzFree(void *p, void *address) { UNUSED_VAR(p); MyFree(address); } - SRes Lzma86_GetUnpackSize(const Byte *src, SizeT srcLen, UInt64 *unpackSize) { unsigned i; @@ -25,7 +22,6 @@ SRes Lzma86_GetUnpackSize(const Byte *src, SizeT srcLen, UInt64 *unpackSize) SRes Lzma86_Decode(Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen) { - ISzAlloc g_Alloc = { SzAlloc, SzFree }; SRes res; int useFilter; SizeT inSizePure; diff --git a/src/third-party/lzma/Lzma86Enc.c b/src/third-party/lzma/Lzma86Enc.c index 868c726d5..8d35e6dc5 100644 --- a/src/third-party/lzma/Lzma86Enc.c +++ b/src/third-party/lzma/Lzma86Enc.c @@ -1,5 +1,5 @@ /* Lzma86Enc.c -- LZMA + x86 (BCJ) Filter Encoder -2015-11-08 : Igor Pavlov : Public domain */ +2018-07-04 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -13,16 +13,12 @@ #define SZE_OUT_OVERFLOW SZE_DATA_ERROR -static void *SzAlloc(void *p, size_t size) { UNUSED_VAR(p); return MyAlloc(size); } -static void SzFree(void *p, void *address) { UNUSED_VAR(p); MyFree(address); } - int Lzma86_Encode(Byte *dest, size_t *destLen, const Byte *src, size_t srcLen, int level, UInt32 dictSize, int filterMode) { - ISzAlloc g_Alloc = { SzAlloc, SzFree }; size_t outSize2 = *destLen; Byte *filteredStream; - Bool useFilter; + BoolInt useFilter; int mainResult = SZ_ERROR_OUTPUT_EOF; CLzmaEncProps props; LzmaEncProps_Init(&props); @@ -60,7 +56,7 @@ int Lzma86_Encode(Byte *dest, size_t *destLen, const Byte *src, size_t srcLen, { size_t minSize = 0; - Bool bestIsFiltered = False; + BoolInt bestIsFiltered = False; /* passes for SZ_FILTER_AUTO: 0 - BCJ + LZMA @@ -75,7 +71,7 @@ int Lzma86_Encode(Byte *dest, size_t *destLen, const Byte *src, size_t srcLen, size_t outSizeProcessed = outSize2 - LZMA86_HEADER_SIZE; size_t outPropsSize = 5; SRes curRes; - Bool curModeIsFiltered = (numPasses > 1 && i == numPasses - 1); + BoolInt curModeIsFiltered = (numPasses > 1 && i == numPasses - 1); if (curModeIsFiltered && !bestIsFiltered) break; if (useFilter && i == 0) diff --git a/src/third-party/lzma/LzmaDec.c b/src/third-party/lzma/LzmaDec.c index 651d1f284..4d1576419 100644 --- a/src/third-party/lzma/LzmaDec.c +++ b/src/third-party/lzma/LzmaDec.c @@ -1,12 +1,13 @@ /* LzmaDec.c -- LZMA Decoder -2015-06-23 : Igor Pavlov : Public domain */ +2018-07-04 : Igor Pavlov : Public domain */ #include "Precomp.h" -#include "LzmaDec.h" - #include +/* #include "CpuArch.h" */ +#include "LzmaDec.h" + #define kNumTopBits 24 #define kTopValue ((UInt32)1 << kNumTopBits) @@ -18,15 +19,22 @@ #define NORMALIZE if (range < kTopValue) { range <<= 8; code = (code << 8) | (*buf++); } -#define IF_BIT_0(p) ttt = *(p); NORMALIZE; bound = (range >> kNumBitModelTotalBits) * ttt; if (code < bound) +#define IF_BIT_0(p) ttt = *(p); NORMALIZE; bound = (range >> kNumBitModelTotalBits) * (UInt32)ttt; if (code < bound) #define UPDATE_0(p) range = bound; *(p) = (CLzmaProb)(ttt + ((kBitModelTotal - ttt) >> kNumMoveBits)); #define UPDATE_1(p) range -= bound; code -= bound; *(p) = (CLzmaProb)(ttt - (ttt >> kNumMoveBits)); #define GET_BIT2(p, i, A0, A1) IF_BIT_0(p) \ { UPDATE_0(p); i = (i + i); A0; } else \ { UPDATE_1(p); i = (i + i) + 1; A1; } -#define GET_BIT(p, i) GET_BIT2(p, i, ; , ;) -#define TREE_GET_BIT(probs, i) { GET_BIT((probs + i), i); } +#define TREE_GET_BIT(probs, i) { GET_BIT2(probs + i, i, ;, ;); } + +#define REV_BIT(p, i, A0, A1) IF_BIT_0(p + i) \ + { UPDATE_0(p + i); A0; } else \ + { UPDATE_1(p + i); A1; } +#define REV_BIT_VAR( p, i, m) REV_BIT(p, i, i += m; m += m, m += m; i += m; ) +#define REV_BIT_CONST(p, i, m) REV_BIT(p, i, i += m; , i += m * 2; ) +#define REV_BIT_LAST( p, i, m) REV_BIT(p, i, i -= m , ; ) + #define TREE_DECODE(probs, limit, i) \ { i = 1; do { TREE_GET_BIT(probs, i); } while (i < limit); i -= limit; } @@ -46,16 +54,19 @@ i -= 0x40; } #endif -#define NORMAL_LITER_DEC GET_BIT(prob + symbol, symbol) +#define NORMAL_LITER_DEC TREE_GET_BIT(prob, symbol) #define MATCHED_LITER_DEC \ - matchByte <<= 1; \ - bit = (matchByte & offs); \ - probLit = prob + offs + bit + symbol; \ - GET_BIT2(probLit, symbol, offs &= ~bit, offs &= bit) + matchByte += matchByte; \ + bit = offs; \ + offs &= matchByte; \ + probLit = prob + (offs + bit + symbol); \ + GET_BIT2(probLit, symbol, offs ^= bit; , ;) + + #define NORMALIZE_CHECK if (range < kTopValue) { if (buf >= bufLimit) return DUMMY_ERROR; range <<= 8; code = (code << 8) | (*buf++); } -#define IF_BIT_0_CHECK(p) ttt = *(p); NORMALIZE_CHECK; bound = (range >> kNumBitModelTotalBits) * ttt; if (code < bound) +#define IF_BIT_0_CHECK(p) ttt = *(p); NORMALIZE_CHECK; bound = (range >> kNumBitModelTotalBits) * (UInt32)ttt; if (code < bound) #define UPDATE_0_CHECK range = bound; #define UPDATE_1_CHECK range -= bound; code -= bound; #define GET_BIT2_CHECK(p, i, A0, A1) IF_BIT_0_CHECK(p) \ @@ -66,25 +77,28 @@ { i = 1; do { GET_BIT_CHECK(probs + i, i) } while (i < limit); i -= limit; } +#define REV_BIT_CHECK(p, i, m) IF_BIT_0_CHECK(p + i) \ + { UPDATE_0_CHECK; i += m; m += m; } else \ + { UPDATE_1_CHECK; m += m; i += m; } + + #define kNumPosBitsMax 4 #define kNumPosStatesMax (1 << kNumPosBitsMax) #define kLenNumLowBits 3 #define kLenNumLowSymbols (1 << kLenNumLowBits) -#define kLenNumMidBits 3 -#define kLenNumMidSymbols (1 << kLenNumMidBits) #define kLenNumHighBits 8 #define kLenNumHighSymbols (1 << kLenNumHighBits) -#define LenChoice 0 -#define LenChoice2 (LenChoice + 1) -#define LenLow (LenChoice2 + 1) -#define LenMid (LenLow + (kNumPosStatesMax << kLenNumLowBits)) -#define LenHigh (LenMid + (kNumPosStatesMax << kLenNumMidBits)) +#define LenLow 0 +#define LenHigh (LenLow + 2 * (kNumPosStatesMax << kLenNumLowBits)) #define kNumLenProbs (LenHigh + kLenNumHighSymbols) +#define LenChoice LenLow +#define LenChoice2 (LenLow + (1 << kLenNumLowBits)) #define kNumStates 12 +#define kNumStates2 16 #define kNumLitStates 7 #define kStartPosModelIndex 4 @@ -98,54 +112,117 @@ #define kAlignTableSize (1 << kNumAlignBits) #define kMatchMinLen 2 -#define kMatchSpecLenStart (kMatchMinLen + kLenNumLowSymbols + kLenNumMidSymbols + kLenNumHighSymbols) +#define kMatchSpecLenStart (kMatchMinLen + kLenNumLowSymbols * 2 + kLenNumHighSymbols) -#define IsMatch 0 -#define IsRep (IsMatch + (kNumStates << kNumPosBitsMax)) +/* External ASM code needs same CLzmaProb array layout. So don't change it. */ + +/* (probs_1664) is faster and better for code size at some platforms */ +/* +#ifdef MY_CPU_X86_OR_AMD64 +*/ +#define kStartOffset 1664 +#define GET_PROBS p->probs_1664 +/* +#define GET_PROBS p->probs + kStartOffset +#else +#define kStartOffset 0 +#define GET_PROBS p->probs +#endif +*/ + +#define SpecPos (-kStartOffset) +#define IsRep0Long (SpecPos + kNumFullDistances) +#define RepLenCoder (IsRep0Long + (kNumStates2 << kNumPosBitsMax)) +#define LenCoder (RepLenCoder + kNumLenProbs) +#define IsMatch (LenCoder + kNumLenProbs) +#define Align (IsMatch + (kNumStates2 << kNumPosBitsMax)) +#define IsRep (Align + kAlignTableSize) #define IsRepG0 (IsRep + kNumStates) #define IsRepG1 (IsRepG0 + kNumStates) #define IsRepG2 (IsRepG1 + kNumStates) -#define IsRep0Long (IsRepG2 + kNumStates) -#define PosSlot (IsRep0Long + (kNumStates << kNumPosBitsMax)) -#define SpecPos (PosSlot + (kNumLenToPosStates << kNumPosSlotBits)) -#define Align (SpecPos + kNumFullDistances - kEndPosModelIndex) -#define LenCoder (Align + kAlignTableSize) -#define RepLenCoder (LenCoder + kNumLenProbs) -#define Literal (RepLenCoder + kNumLenProbs) +#define PosSlot (IsRepG2 + kNumStates) +#define Literal (PosSlot + (kNumLenToPosStates << kNumPosSlotBits)) +#define NUM_BASE_PROBS (Literal + kStartOffset) -#define LZMA_BASE_SIZE 1846 -#define LZMA_LIT_SIZE 0x300 - -#if Literal != LZMA_BASE_SIZE -StopCompilingDueBUG +#if Align != 0 && kStartOffset != 0 + #error Stop_Compiling_Bad_LZMA_kAlign #endif -#define LzmaProps_GetNumProbs(p) (Literal + ((UInt32)LZMA_LIT_SIZE << ((p)->lc + (p)->lp))) +#if NUM_BASE_PROBS != 1984 + #error Stop_Compiling_Bad_LZMA_PROBS +#endif + + +#define LZMA_LIT_SIZE 0x300 + +#define LzmaProps_GetNumProbs(p) (NUM_BASE_PROBS + ((UInt32)LZMA_LIT_SIZE << ((p)->lc + (p)->lp))) + + +#define CALC_POS_STATE(processedPos, pbMask) (((processedPos) & (pbMask)) << 4) +#define COMBINED_PS_STATE (posState + state) +#define GET_LEN_STATE (posState) #define LZMA_DIC_MIN (1 << 12) -/* First LZMA-symbol is always decoded. -And it decodes new LZMA-symbols while (buf < bufLimit), but "buf" is without last normalization +/* +p->remainLen : shows status of LZMA decoder: + < kMatchSpecLenStart : normal remain + = kMatchSpecLenStart : finished + = kMatchSpecLenStart + 1 : need init range coder + = kMatchSpecLenStart + 2 : need init range coder and state +*/ + +/* ---------- LZMA_DECODE_REAL ---------- */ +/* +LzmaDec_DecodeReal_3() can be implemented in external ASM file. +3 - is the code compatibility version of that function for check at link time. +*/ + +#define LZMA_DECODE_REAL LzmaDec_DecodeReal_3 + +/* +LZMA_DECODE_REAL() +In: + RangeCoder is normalized + if (p->dicPos == limit) + { + LzmaDec_TryDummy() was called before to exclude LITERAL and MATCH-REP cases. + So first symbol can be only MATCH-NON-REP. And if that MATCH-NON-REP symbol + is not END_OF_PAYALOAD_MARKER, then function returns error code. + } + +Processing: + first LZMA symbol will be decoded in any case + All checks for limits are at the end of main loop, + It will decode new LZMA-symbols while (p->buf < bufLimit && dicPos < limit), + RangeCoder is still without last normalization when (p->buf < bufLimit) is being checked. + Out: + RangeCoder is normalized Result: SZ_OK - OK SZ_ERROR_DATA - Error p->remainLen: < kMatchSpecLenStart : normal remain = kMatchSpecLenStart : finished - = kMatchSpecLenStart + 1 : Flush marker (unused now) - = kMatchSpecLenStart + 2 : State Init Marker (unused now) */ -static int MY_FAST_CALL LzmaDec_DecodeReal(CLzmaDec *p, SizeT limit, const Byte *bufLimit) -{ - CLzmaProb *probs = p->probs; - unsigned state = p->state; +#ifdef _LZMA_DEC_OPT + +int MY_FAST_CALL LZMA_DECODE_REAL(CLzmaDec *p, SizeT limit, const Byte *bufLimit); + +#else + +static +int MY_FAST_CALL LZMA_DECODE_REAL(CLzmaDec *p, SizeT limit, const Byte *bufLimit) +{ + CLzmaProb *probs = GET_PROBS; + unsigned state = (unsigned)p->state; UInt32 rep0 = p->reps[0], rep1 = p->reps[1], rep2 = p->reps[2], rep3 = p->reps[3]; unsigned pbMask = ((unsigned)1 << (p->prop.pb)) - 1; - unsigned lpMask = ((unsigned)1 << (p->prop.lp)) - 1; unsigned lc = p->prop.lc; + unsigned lpMask = ((unsigned)0x100 << p->prop.lp) - ((unsigned)0x100 >> lc); Byte *dic = p->dic; SizeT dicBufSize = p->dicBufSize; @@ -164,17 +241,16 @@ static int MY_FAST_CALL LzmaDec_DecodeReal(CLzmaDec *p, SizeT limit, const Byte CLzmaProb *prob; UInt32 bound; unsigned ttt; - unsigned posState = processedPos & pbMask; + unsigned posState = CALC_POS_STATE(processedPos, pbMask); - prob = probs + IsMatch + (state << kNumPosBitsMax) + posState; + prob = probs + IsMatch + COMBINED_PS_STATE; IF_BIT_0(prob) { unsigned symbol; UPDATE_0(prob); prob = probs + Literal; if (processedPos != 0 || checkDicSize != 0) - prob += ((UInt32)LZMA_LIT_SIZE * (((processedPos & lpMask) << lc) + - (dic[(dicPos == 0 ? dicBufSize : dicPos) - 1] >> (8 - lc)))); + prob += (UInt32)3 * ((((processedPos << 8) + dic[(dicPos == 0 ? dicBufSize : dicPos) - 1]) & lpMask) << lc); processedPos++; if (state < kNumLitStates) @@ -240,13 +316,16 @@ static int MY_FAST_CALL LzmaDec_DecodeReal(CLzmaDec *p, SizeT limit, const Byte else { UPDATE_1(prob); + /* + // that case was checked before with kBadRepCode if (checkDicSize == 0 && processedPos == 0) return SZ_ERROR_DATA; + */ prob = probs + IsRepG0 + state; IF_BIT_0(prob) { UPDATE_0(prob); - prob = probs + IsRep0Long + (state << kNumPosBitsMax) + posState; + prob = probs + IsRep0Long + COMBINED_PS_STATE; IF_BIT_0(prob) { UPDATE_0(prob); @@ -294,14 +373,14 @@ static int MY_FAST_CALL LzmaDec_DecodeReal(CLzmaDec *p, SizeT limit, const Byte #ifdef _LZMA_SIZE_OPT { - unsigned limit, offset; + unsigned lim, offset; CLzmaProb *probLen = prob + LenChoice; IF_BIT_0(probLen) { UPDATE_0(probLen); - probLen = prob + LenLow + (posState << kLenNumLowBits); + probLen = prob + LenLow + GET_LEN_STATE; offset = 0; - limit = (1 << kLenNumLowBits); + lim = (1 << kLenNumLowBits); } else { @@ -310,19 +389,19 @@ static int MY_FAST_CALL LzmaDec_DecodeReal(CLzmaDec *p, SizeT limit, const Byte IF_BIT_0(probLen) { UPDATE_0(probLen); - probLen = prob + LenMid + (posState << kLenNumMidBits); + probLen = prob + LenLow + GET_LEN_STATE + (1 << kLenNumLowBits); offset = kLenNumLowSymbols; - limit = (1 << kLenNumMidBits); + lim = (1 << kLenNumLowBits); } else { UPDATE_1(probLen); probLen = prob + LenHigh; - offset = kLenNumLowSymbols + kLenNumMidSymbols; - limit = (1 << kLenNumHighBits); + offset = kLenNumLowSymbols * 2; + lim = (1 << kLenNumHighBits); } } - TREE_DECODE(probLen, limit, len); + TREE_DECODE(probLen, lim, len); len += offset; } #else @@ -331,7 +410,7 @@ static int MY_FAST_CALL LzmaDec_DecodeReal(CLzmaDec *p, SizeT limit, const Byte IF_BIT_0(probLen) { UPDATE_0(probLen); - probLen = prob + LenLow + (posState << kLenNumLowBits); + probLen = prob + LenLow + GET_LEN_STATE; len = 1; TREE_GET_BIT(probLen, len); TREE_GET_BIT(probLen, len); @@ -345,7 +424,7 @@ static int MY_FAST_CALL LzmaDec_DecodeReal(CLzmaDec *p, SizeT limit, const Byte IF_BIT_0(probLen) { UPDATE_0(probLen); - probLen = prob + LenMid + (posState << kLenNumMidBits); + probLen = prob + LenLow + GET_LEN_STATE + (1 << kLenNumLowBits); len = 1; TREE_GET_BIT(probLen, len); TREE_GET_BIT(probLen, len); @@ -356,7 +435,7 @@ static int MY_FAST_CALL LzmaDec_DecodeReal(CLzmaDec *p, SizeT limit, const Byte UPDATE_1(probLen); probLen = prob + LenHigh; TREE_DECODE(probLen, (1 << kLenNumHighBits), len); - len += kLenNumLowSymbols + kLenNumMidSymbols; + len += kLenNumLowSymbols * 2; } } } @@ -376,16 +455,16 @@ static int MY_FAST_CALL LzmaDec_DecodeReal(CLzmaDec *p, SizeT limit, const Byte if (posSlot < kEndPosModelIndex) { distance <<= numDirectBits; - prob = probs + SpecPos + distance - posSlot - 1; + prob = probs + SpecPos; { - UInt32 mask = 1; - unsigned i = 1; + UInt32 m = 1; + distance++; do { - GET_BIT2(prob + i, i, ; , distance |= mask); - mask <<= 1; + REV_BIT_VAR(prob, distance, m); } - while (--numDirectBits != 0); + while (--numDirectBits); + distance -= m; } } else @@ -412,19 +491,20 @@ static int MY_FAST_CALL LzmaDec_DecodeReal(CLzmaDec *p, SizeT limit, const Byte } */ } - while (--numDirectBits != 0); + while (--numDirectBits); prob = probs + Align; distance <<= kNumAlignBits; { unsigned i = 1; - GET_BIT2(prob + i, i, ; , distance |= 1); - GET_BIT2(prob + i, i, ; , distance |= 2); - GET_BIT2(prob + i, i, ; , distance |= 4); - GET_BIT2(prob + i, i, ; , distance |= 8); + REV_BIT_CONST(prob, i, 1); + REV_BIT_CONST(prob, i, 2); + REV_BIT_CONST(prob, i, 4); + REV_BIT_LAST (prob, i, 8); + distance |= i; } if (distance == (UInt32)0xFFFFFFFF) { - len += kMatchSpecLenStart; + len = kMatchSpecLenStart; state -= kNumStates; break; } @@ -435,20 +515,12 @@ static int MY_FAST_CALL LzmaDec_DecodeReal(CLzmaDec *p, SizeT limit, const Byte rep2 = rep1; rep1 = rep0; rep0 = distance + 1; - if (checkDicSize == 0) - { - if (distance >= processedPos) - { - p->dicPos = dicPos; - return SZ_ERROR_DATA; - } - } - else if (distance >= checkDicSize) + state = (state < kNumStates + kNumLitStates) ? kNumLitStates : kNumLitStates + 3; + if (distance >= (checkDicSize == 0 ? processedPos: checkDicSize)) { p->dicPos = dicPos; return SZ_ERROR_DATA; } - state = (state < kNumStates + kNumLitStates) ? kNumLitStates : kNumLitStates + 3; } len += kMatchMinLen; @@ -467,7 +539,7 @@ static int MY_FAST_CALL LzmaDec_DecodeReal(CLzmaDec *p, SizeT limit, const Byte curLen = ((rem < len) ? (unsigned)rem : len); pos = dicPos - rep0 + (dicPos < rep0 ? dicBufSize : 0); - processedPos += curLen; + processedPos += (UInt32)curLen; len -= curLen; if (curLen <= dicBufSize - pos) @@ -475,7 +547,7 @@ static int MY_FAST_CALL LzmaDec_DecodeReal(CLzmaDec *p, SizeT limit, const Byte Byte *dest = dic + dicPos; ptrdiff_t src = (ptrdiff_t)pos - (ptrdiff_t)dicPos; const Byte *lim = dest + curLen; - dicPos += curLen; + dicPos += (SizeT)curLen; do *(dest) = (Byte)*(dest + src); while (++dest != lim); @@ -500,17 +572,18 @@ static int MY_FAST_CALL LzmaDec_DecodeReal(CLzmaDec *p, SizeT limit, const Byte p->buf = buf; p->range = range; p->code = code; - p->remainLen = len; + p->remainLen = (UInt32)len; p->dicPos = dicPos; p->processedPos = processedPos; p->reps[0] = rep0; p->reps[1] = rep1; p->reps[2] = rep2; p->reps[3] = rep3; - p->state = state; + p->state = (UInt32)state; return SZ_OK; } +#endif static void MY_FAST_CALL LzmaDec_WriteRem(CLzmaDec *p, SizeT limit) { @@ -519,7 +592,7 @@ static void MY_FAST_CALL LzmaDec_WriteRem(CLzmaDec *p, SizeT limit) Byte *dic = p->dic; SizeT dicPos = p->dicPos; SizeT dicBufSize = p->dicBufSize; - unsigned len = p->remainLen; + unsigned len = (unsigned)p->remainLen; SizeT rep0 = p->reps[0]; /* we use SizeT to avoid the BUG of VC14 for AMD64 */ SizeT rem = limit - dicPos; if (rem < len) @@ -528,8 +601,8 @@ static void MY_FAST_CALL LzmaDec_WriteRem(CLzmaDec *p, SizeT limit) if (p->checkDicSize == 0 && p->prop.dicSize - p->processedPos <= len) p->checkDicSize = p->prop.dicSize; - p->processedPos += len; - p->remainLen -= len; + p->processedPos += (UInt32)len; + p->remainLen -= (UInt32)len; while (len != 0) { len--; @@ -540,6 +613,14 @@ static void MY_FAST_CALL LzmaDec_WriteRem(CLzmaDec *p, SizeT limit) } } + +#define kRange0 0xFFFFFFFF +#define kBound0 ((kRange0 >> kNumBitModelTotalBits) << (kNumBitModelTotalBits - 1)) +#define kBadRepCode (kBound0 + (((kRange0 - kBound0) >> kNumBitModelTotalBits) << (kNumBitModelTotalBits - 1))) +#if kBadRepCode != (0xC0000000 - 0x400) + #error Stop_Compiling_Bad_LZMA_Check +#endif + static int MY_FAST_CALL LzmaDec_DecodeReal2(CLzmaDec *p, SizeT limit, const Byte *bufLimit) { do @@ -550,9 +631,13 @@ static int MY_FAST_CALL LzmaDec_DecodeReal2(CLzmaDec *p, SizeT limit, const Byte UInt32 rem = p->prop.dicSize - p->processedPos; if (limit - p->dicPos > rem) limit2 = p->dicPos + rem; + + if (p->processedPos == 0) + if (p->code >= kBadRepCode) + return SZ_ERROR_DATA; } - - RINOK(LzmaDec_DecodeReal(p, limit2, bufLimit)); + + RINOK(LZMA_DECODE_REAL(p, limit2, bufLimit)); if (p->checkDicSize == 0 && p->processedPos >= p->prop.dicSize) p->checkDicSize = p->prop.dicSize; @@ -561,9 +646,6 @@ static int MY_FAST_CALL LzmaDec_DecodeReal2(CLzmaDec *p, SizeT limit, const Byte } while (p->dicPos < limit && p->buf < bufLimit && p->remainLen < kMatchSpecLenStart); - if (p->remainLen > kMatchSpecLenStart) - p->remainLen = kMatchSpecLenStart; - return 0; } @@ -580,17 +662,17 @@ static ELzmaDummy LzmaDec_TryDummy(const CLzmaDec *p, const Byte *buf, SizeT inS UInt32 range = p->range; UInt32 code = p->code; const Byte *bufLimit = buf + inSize; - const CLzmaProb *probs = p->probs; - unsigned state = p->state; + const CLzmaProb *probs = GET_PROBS; + unsigned state = (unsigned)p->state; ELzmaDummy res; { const CLzmaProb *prob; UInt32 bound; unsigned ttt; - unsigned posState = (p->processedPos) & ((1 << p->prop.pb) - 1); + unsigned posState = CALC_POS_STATE(p->processedPos, (1 << p->prop.pb) - 1); - prob = probs + IsMatch + (state << kNumPosBitsMax) + posState; + prob = probs + IsMatch + COMBINED_PS_STATE; IF_BIT_0_CHECK(prob) { UPDATE_0_CHECK @@ -618,10 +700,11 @@ static ELzmaDummy LzmaDec_TryDummy(const CLzmaDec *p, const Byte *buf, SizeT inS { unsigned bit; const CLzmaProb *probLit; - matchByte <<= 1; - bit = (matchByte & offs); - probLit = prob + offs + bit + symbol; - GET_BIT2_CHECK(probLit, symbol, offs &= ~bit, offs &= bit) + matchByte += matchByte; + bit = offs; + offs &= matchByte; + probLit = prob + (offs + bit + symbol); + GET_BIT2_CHECK(probLit, symbol, offs ^= bit; , ; ) } while (symbol < 0x100); } @@ -648,7 +731,7 @@ static ELzmaDummy LzmaDec_TryDummy(const CLzmaDec *p, const Byte *buf, SizeT inS IF_BIT_0_CHECK(prob) { UPDATE_0_CHECK; - prob = probs + IsRep0Long + (state << kNumPosBitsMax) + posState; + prob = probs + IsRep0Long + COMBINED_PS_STATE; IF_BIT_0_CHECK(prob) { UPDATE_0_CHECK; @@ -691,7 +774,7 @@ static ELzmaDummy LzmaDec_TryDummy(const CLzmaDec *p, const Byte *buf, SizeT inS IF_BIT_0_CHECK(probLen) { UPDATE_0_CHECK; - probLen = prob + LenLow + (posState << kLenNumLowBits); + probLen = prob + LenLow + GET_LEN_STATE; offset = 0; limit = 1 << kLenNumLowBits; } @@ -702,15 +785,15 @@ static ELzmaDummy LzmaDec_TryDummy(const CLzmaDec *p, const Byte *buf, SizeT inS IF_BIT_0_CHECK(probLen) { UPDATE_0_CHECK; - probLen = prob + LenMid + (posState << kLenNumMidBits); + probLen = prob + LenLow + GET_LEN_STATE + (1 << kLenNumLowBits); offset = kLenNumLowSymbols; - limit = 1 << kLenNumMidBits; + limit = 1 << kLenNumLowBits; } else { UPDATE_1_CHECK; probLen = prob + LenHigh; - offset = kLenNumLowSymbols + kLenNumMidSymbols; + offset = kLenNumLowSymbols * 2; limit = 1 << kLenNumHighBits; } } @@ -722,7 +805,7 @@ static ELzmaDummy LzmaDec_TryDummy(const CLzmaDec *p, const Byte *buf, SizeT inS { unsigned posSlot; prob = probs + PosSlot + - ((len < kNumLenToPosStates ? len : kNumLenToPosStates - 1) << + ((len < kNumLenToPosStates - 1 ? len : kNumLenToPosStates - 1) << kNumPosSlotBits); TREE_DECODE_CHECK(prob, 1 << kNumPosSlotBits, posSlot); if (posSlot >= kStartPosModelIndex) @@ -733,7 +816,7 @@ static ELzmaDummy LzmaDec_TryDummy(const CLzmaDec *p, const Byte *buf, SizeT inS if (posSlot < kEndPosModelIndex) { - prob = probs + SpecPos + ((2 | (posSlot & 1)) << numDirectBits) - posSlot - 1; + prob = probs + SpecPos + ((2 | (posSlot & 1)) << numDirectBits); } else { @@ -745,17 +828,18 @@ static ELzmaDummy LzmaDec_TryDummy(const CLzmaDec *p, const Byte *buf, SizeT inS code -= range & (((code - range) >> 31) - 1); /* if (code >= range) code -= range; */ } - while (--numDirectBits != 0); + while (--numDirectBits); prob = probs + Align; numDirectBits = kNumAlignBits; } { unsigned i = 1; + unsigned m = 1; do { - GET_BIT_CHECK(prob + i, i); + REV_BIT_CHECK(prob, i, m); } - while (--numDirectBits != 0); + while (--numDirectBits); } } } @@ -766,20 +850,19 @@ static ELzmaDummy LzmaDec_TryDummy(const CLzmaDec *p, const Byte *buf, SizeT inS } -void LzmaDec_InitDicAndState(CLzmaDec *p, Bool initDic, Bool initState) +void LzmaDec_InitDicAndState(CLzmaDec *p, BoolInt initDic, BoolInt initState) { - p->needFlush = 1; - p->remainLen = 0; + p->remainLen = kMatchSpecLenStart + 1; p->tempBufSize = 0; if (initDic) { p->processedPos = 0; p->checkDicSize = 0; - p->needInitState = 1; + p->remainLen = kMatchSpecLenStart + 2; } if (initState) - p->needInitState = 1; + p->remainLen = kMatchSpecLenStart + 2; } void LzmaDec_Init(CLzmaDec *p) @@ -788,53 +871,54 @@ void LzmaDec_Init(CLzmaDec *p) LzmaDec_InitDicAndState(p, True, True); } -static void LzmaDec_InitStateReal(CLzmaDec *p) -{ - SizeT numProbs = LzmaProps_GetNumProbs(&p->prop); - SizeT i; - CLzmaProb *probs = p->probs; - for (i = 0; i < numProbs; i++) - probs[i] = kBitModelTotal >> 1; - p->reps[0] = p->reps[1] = p->reps[2] = p->reps[3] = 1; - p->state = 0; - p->needInitState = 0; -} SRes LzmaDec_DecodeToDic(CLzmaDec *p, SizeT dicLimit, const Byte *src, SizeT *srcLen, ELzmaFinishMode finishMode, ELzmaStatus *status) { SizeT inSize = *srcLen; (*srcLen) = 0; - LzmaDec_WriteRem(p, dicLimit); *status = LZMA_STATUS_NOT_SPECIFIED; + if (p->remainLen > kMatchSpecLenStart) + { + for (; inSize > 0 && p->tempBufSize < RC_INIT_SIZE; (*srcLen)++, inSize--) + p->tempBuf[p->tempBufSize++] = *src++; + if (p->tempBufSize != 0 && p->tempBuf[0] != 0) + return SZ_ERROR_DATA; + if (p->tempBufSize < RC_INIT_SIZE) + { + *status = LZMA_STATUS_NEEDS_MORE_INPUT; + return SZ_OK; + } + p->code = + ((UInt32)p->tempBuf[1] << 24) + | ((UInt32)p->tempBuf[2] << 16) + | ((UInt32)p->tempBuf[3] << 8) + | ((UInt32)p->tempBuf[4]); + p->range = 0xFFFFFFFF; + p->tempBufSize = 0; + + if (p->remainLen > kMatchSpecLenStart + 1) + { + SizeT numProbs = LzmaProps_GetNumProbs(&p->prop); + SizeT i; + CLzmaProb *probs = p->probs; + for (i = 0; i < numProbs; i++) + probs[i] = kBitModelTotal >> 1; + p->reps[0] = p->reps[1] = p->reps[2] = p->reps[3] = 1; + p->state = 0; + } + + p->remainLen = 0; + } + + LzmaDec_WriteRem(p, dicLimit); + while (p->remainLen != kMatchSpecLenStart) { - int checkEndMarkNow; + int checkEndMarkNow = 0; - if (p->needFlush) - { - for (; inSize > 0 && p->tempBufSize < RC_INIT_SIZE; (*srcLen)++, inSize--) - p->tempBuf[p->tempBufSize++] = *src++; - if (p->tempBufSize < RC_INIT_SIZE) - { - *status = LZMA_STATUS_NEEDS_MORE_INPUT; - return SZ_OK; - } - if (p->tempBuf[0] != 0) - return SZ_ERROR_DATA; - p->code = - ((UInt32)p->tempBuf[1] << 24) - | ((UInt32)p->tempBuf[2] << 16) - | ((UInt32)p->tempBuf[3] << 8) - | ((UInt32)p->tempBuf[4]); - p->range = 0xFFFFFFFF; - p->needFlush = 0; - p->tempBufSize = 0; - } - - checkEndMarkNow = 0; if (p->dicPos >= dicLimit) { if (p->remainLen == 0 && p->code == 0) @@ -855,9 +939,6 @@ SRes LzmaDec_DecodeToDic(CLzmaDec *p, SizeT dicLimit, const Byte *src, SizeT *sr checkEndMarkNow = 1; } - if (p->needInitState) - LzmaDec_InitStateReal(p); - if (p->tempBufSize == 0) { SizeT processed; @@ -898,10 +979,10 @@ SRes LzmaDec_DecodeToDic(CLzmaDec *p, SizeT dicLimit, const Byte *src, SizeT *sr p->tempBufSize = rem; if (rem < LZMA_REQUIRED_INPUT_MAX || checkEndMarkNow) { - int dummyRes = LzmaDec_TryDummy(p, p->tempBuf, rem); + int dummyRes = LzmaDec_TryDummy(p, p->tempBuf, (SizeT)rem); if (dummyRes == DUMMY_ERROR) { - (*srcLen) += lookAhead; + (*srcLen) += (SizeT)lookAhead; *status = LZMA_STATUS_NEEDS_MORE_INPUT; return SZ_OK; } @@ -924,17 +1005,20 @@ SRes LzmaDec_DecodeToDic(CLzmaDec *p, SizeT dicLimit, const Byte *src, SizeT *sr return SZ_ERROR_FAIL; /* some internal error */ lookAhead -= rem; } - (*srcLen) += lookAhead; + (*srcLen) += (SizeT)lookAhead; src += lookAhead; - inSize -= lookAhead; + inSize -= (SizeT)lookAhead; p->tempBufSize = 0; } } - if (p->code == 0) - *status = LZMA_STATUS_FINISHED_WITH_MARK; - return (p->code == 0) ? SZ_OK : SZ_ERROR_DATA; + + if (p->code != 0) + return SZ_ERROR_DATA; + *status = LZMA_STATUS_FINISHED_WITH_MARK; + return SZ_OK; } + SRes LzmaDec_DecodeToBuf(CLzmaDec *p, Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen, ELzmaFinishMode finishMode, ELzmaStatus *status) { SizeT outSize = *destLen; @@ -975,19 +1059,19 @@ SRes LzmaDec_DecodeToBuf(CLzmaDec *p, Byte *dest, SizeT *destLen, const Byte *sr } } -void LzmaDec_FreeProbs(CLzmaDec *p, ISzAlloc *alloc) +void LzmaDec_FreeProbs(CLzmaDec *p, ISzAllocPtr alloc) { - alloc->Free(alloc, p->probs); + ISzAlloc_Free(alloc, p->probs); p->probs = NULL; } -static void LzmaDec_FreeDict(CLzmaDec *p, ISzAlloc *alloc) +static void LzmaDec_FreeDict(CLzmaDec *p, ISzAllocPtr alloc) { - alloc->Free(alloc, p->dic); + ISzAlloc_Free(alloc, p->dic); p->dic = NULL; } -void LzmaDec_Free(CLzmaDec *p, ISzAlloc *alloc) +void LzmaDec_Free(CLzmaDec *p, ISzAllocPtr alloc) { LzmaDec_FreeProbs(p, alloc); LzmaDec_FreeDict(p, alloc); @@ -1011,29 +1095,30 @@ SRes LzmaProps_Decode(CLzmaProps *p, const Byte *data, unsigned size) if (d >= (9 * 5 * 5)) return SZ_ERROR_UNSUPPORTED; - p->lc = d % 9; + p->lc = (Byte)(d % 9); d /= 9; - p->pb = d / 5; - p->lp = d % 5; + p->pb = (Byte)(d / 5); + p->lp = (Byte)(d % 5); return SZ_OK; } -static SRes LzmaDec_AllocateProbs2(CLzmaDec *p, const CLzmaProps *propNew, ISzAlloc *alloc) +static SRes LzmaDec_AllocateProbs2(CLzmaDec *p, const CLzmaProps *propNew, ISzAllocPtr alloc) { UInt32 numProbs = LzmaProps_GetNumProbs(propNew); if (!p->probs || numProbs != p->numProbs) { LzmaDec_FreeProbs(p, alloc); - p->probs = (CLzmaProb *)alloc->Alloc(alloc, numProbs * sizeof(CLzmaProb)); - p->numProbs = numProbs; + p->probs = (CLzmaProb *)ISzAlloc_Alloc(alloc, numProbs * sizeof(CLzmaProb)); if (!p->probs) return SZ_ERROR_MEM; + p->probs_1664 = p->probs + 1664; + p->numProbs = numProbs; } return SZ_OK; } -SRes LzmaDec_AllocateProbs(CLzmaDec *p, const Byte *props, unsigned propsSize, ISzAlloc *alloc) +SRes LzmaDec_AllocateProbs(CLzmaDec *p, const Byte *props, unsigned propsSize, ISzAllocPtr alloc) { CLzmaProps propNew; RINOK(LzmaProps_Decode(&propNew, props, propsSize)); @@ -1042,7 +1127,7 @@ SRes LzmaDec_AllocateProbs(CLzmaDec *p, const Byte *props, unsigned propsSize, I return SZ_OK; } -SRes LzmaDec_Allocate(CLzmaDec *p, const Byte *props, unsigned propsSize, ISzAlloc *alloc) +SRes LzmaDec_Allocate(CLzmaDec *p, const Byte *props, unsigned propsSize, ISzAllocPtr alloc) { CLzmaProps propNew; SizeT dicBufSize; @@ -1062,7 +1147,7 @@ SRes LzmaDec_Allocate(CLzmaDec *p, const Byte *props, unsigned propsSize, ISzAll if (!p->dic || dicBufSize != p->dicBufSize) { LzmaDec_FreeDict(p, alloc); - p->dic = (Byte *)alloc->Alloc(alloc, dicBufSize); + p->dic = (Byte *)ISzAlloc_Alloc(alloc, dicBufSize); if (!p->dic) { LzmaDec_FreeProbs(p, alloc); @@ -1076,7 +1161,7 @@ SRes LzmaDec_Allocate(CLzmaDec *p, const Byte *props, unsigned propsSize, ISzAll SRes LzmaDecode(Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen, const Byte *propData, unsigned propSize, ELzmaFinishMode finishMode, - ELzmaStatus *status, ISzAlloc *alloc) + ELzmaStatus *status, ISzAllocPtr alloc) { CLzmaDec p; SRes res; diff --git a/src/third-party/lzma/LzmaDec.h b/src/third-party/lzma/LzmaDec.h index 2633abeac..28ce60c3e 100644 --- a/src/third-party/lzma/LzmaDec.h +++ b/src/third-party/lzma/LzmaDec.h @@ -1,5 +1,5 @@ /* LzmaDec.h -- LZMA Decoder -2013-01-18 : Igor Pavlov : Public domain */ +2018-04-21 : Igor Pavlov : Public domain */ #ifndef __LZMA_DEC_H #define __LZMA_DEC_H @@ -12,11 +12,13 @@ EXTERN_C_BEGIN /* _LZMA_PROB32 can increase the speed on some CPUs, but memory usage for CLzmaDec::probs will be doubled in that case */ +typedef #ifdef _LZMA_PROB32 -#define CLzmaProb UInt32 + UInt32 #else -#define CLzmaProb UInt16 + UInt16 #endif + CLzmaProb; /* ---------- LZMA Properties ---------- */ @@ -25,7 +27,10 @@ EXTERN_C_BEGIN typedef struct _CLzmaProps { - unsigned lc, lp, pb; + Byte lc; + Byte lp; + Byte pb; + Byte _pad_; UInt32 dicSize; } CLzmaProps; @@ -47,32 +52,34 @@ SRes LzmaProps_Decode(CLzmaProps *p, const Byte *data, unsigned size); typedef struct { + /* Don't change this structure. ASM code can use it. */ CLzmaProps prop; CLzmaProb *probs; + CLzmaProb *probs_1664; Byte *dic; - const Byte *buf; - UInt32 range, code; - SizeT dicPos; SizeT dicBufSize; + SizeT dicPos; + const Byte *buf; + UInt32 range; + UInt32 code; UInt32 processedPos; UInt32 checkDicSize; - unsigned state; UInt32 reps[4]; - unsigned remainLen; - int needFlush; - int needInitState; + UInt32 state; + UInt32 remainLen; + UInt32 numProbs; unsigned tempBufSize; Byte tempBuf[LZMA_REQUIRED_INPUT_MAX]; } CLzmaDec; -#define LzmaDec_Construct(p) { (p)->dic = 0; (p)->probs = 0; } +#define LzmaDec_Construct(p) { (p)->dic = NULL; (p)->probs = NULL; } void LzmaDec_Init(CLzmaDec *p); /* There are two types of LZMA streams: - 0) Stream with end mark. That end mark adds about 6 bytes to compressed size. - 1) Stream without end mark. You must know exact uncompressed size to decompress such stream. */ + - Stream with end mark. That end mark adds about 6 bytes to compressed size. + - Stream without end mark. You must know exact uncompressed size to decompress such stream. */ typedef enum { @@ -129,11 +136,11 @@ LzmaDec_Allocate* can return: SZ_ERROR_UNSUPPORTED - Unsupported properties */ -SRes LzmaDec_AllocateProbs(CLzmaDec *p, const Byte *props, unsigned propsSize, ISzAlloc *alloc); -void LzmaDec_FreeProbs(CLzmaDec *p, ISzAlloc *alloc); +SRes LzmaDec_AllocateProbs(CLzmaDec *p, const Byte *props, unsigned propsSize, ISzAllocPtr alloc); +void LzmaDec_FreeProbs(CLzmaDec *p, ISzAllocPtr alloc); -SRes LzmaDec_Allocate(CLzmaDec *state, const Byte *prop, unsigned propsSize, ISzAlloc *alloc); -void LzmaDec_Free(CLzmaDec *state, ISzAlloc *alloc); +SRes LzmaDec_Allocate(CLzmaDec *p, const Byte *props, unsigned propsSize, ISzAllocPtr alloc); +void LzmaDec_Free(CLzmaDec *p, ISzAllocPtr alloc); /* ---------- Dictionary Interface ---------- */ @@ -142,7 +149,7 @@ void LzmaDec_Free(CLzmaDec *state, ISzAlloc *alloc); You must work with CLzmaDec variables directly in this interface. STEPS: - LzmaDec_Constr() + LzmaDec_Construct() LzmaDec_Allocate() for (each new stream) { @@ -220,7 +227,7 @@ Returns: SRes LzmaDecode(Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen, const Byte *propData, unsigned propSize, ELzmaFinishMode finishMode, - ELzmaStatus *status, ISzAlloc *alloc); + ELzmaStatus *status, ISzAllocPtr alloc); EXTERN_C_END diff --git a/src/third-party/lzma/LzmaEnc.c b/src/third-party/lzma/LzmaEnc.c index 11de42c09..14086fc4f 100644 --- a/src/third-party/lzma/LzmaEnc.c +++ b/src/third-party/lzma/LzmaEnc.c @@ -1,5 +1,5 @@ /* LzmaEnc.c -- LZMA Encoder -2015-11-08 : Igor Pavlov : Public domain */ +2019-01-10: Igor Pavlov : Public domain */ #include "Precomp.h" @@ -23,17 +23,8 @@ static unsigned g_STAT_OFFSET = 0; #endif -#define kMaxHistorySize ((UInt32)3 << 29) -/* #define kMaxHistorySize ((UInt32)7 << 29) */ - -#define kBlockSizeMax ((1 << LZMA_NUM_BLOCK_SIZE_BITS) - 1) - -#define kBlockSize (9 << 10) -#define kUnpackBlockSize (1 << 18) -#define kMatchArraySize (1 << 21) -#define kMatchRecordMaxSize ((LZMA_MATCH_LEN_MAX * 2 + 3) * LZMA_MATCH_LEN_MAX) - -#define kNumMaxDirectBits (31) +#define kLzmaMaxHistorySize ((UInt32)3 << 29) +/* #define kLzmaMaxHistorySize ((UInt32)7 << 29) */ #define kNumTopBits 24 #define kTopValue ((UInt32)1 << kNumTopBits) @@ -47,6 +38,8 @@ static unsigned g_STAT_OFFSET = 0; #define kNumBitPriceShiftBits 4 #define kBitPrice (1 << kNumBitPriceShiftBits) +#define REP_LEN_COUNT 64 + void LzmaEncProps_Init(CLzmaEncProps *p) { p->level = 5; @@ -62,14 +55,15 @@ void LzmaEncProps_Normalize(CLzmaEncProps *p) if (level < 0) level = 5; p->level = level; - if (p->dictSize == 0) p->dictSize = (level <= 5 ? (1 << (level * 2 + 14)) : (level == 6 ? (1 << 25) : (1 << 26))); + if (p->dictSize == 0) p->dictSize = (level <= 5 ? (1 << (level * 2 + 14)) : (level <= 7 ? (1 << 25) : (1 << 26))); if (p->dictSize > p->reduceSize) { unsigned i; + UInt32 reduceSize = (UInt32)p->reduceSize; for (i = 11; i <= 30; i++) { - if ((UInt32)p->reduceSize <= ((UInt32)2 << i)) { p->dictSize = ((UInt32)2 << i); break; } - if ((UInt32)p->reduceSize <= ((UInt32)3 << i)) { p->dictSize = ((UInt32)3 << i); break; } + if (reduceSize <= ((UInt32)2 << i)) { p->dictSize = ((UInt32)2 << i); break; } + if (reduceSize <= ((UInt32)3 << i)) { p->dictSize = ((UInt32)3 << i); break; } } } @@ -108,11 +102,11 @@ UInt32 LzmaEncProps_GetDictSize(const CLzmaEncProps *props2) #define kDicLogSizeMaxCompress 32 -#define BSR2_RET(pos, res) { unsigned long i; _BitScanReverse(&i, (pos)); res = (i + i) + ((pos >> (i - 1)) & 1); } +#define BSR2_RET(pos, res) { unsigned long zz; _BitScanReverse(&zz, (pos)); res = (zz + zz) + ((pos >> (zz - 1)) & 1); } -static UInt32 GetPosSlot1(UInt32 pos) +static unsigned GetPosSlot1(UInt32 pos) { - UInt32 res; + unsigned res; BSR2_RET(pos, res); return res; } @@ -145,19 +139,19 @@ static void LzmaEnc_FastPosInit(Byte *g_FastPos) /* we can use ((limit - pos) >> 31) only if (pos < ((UInt32)1 << 31)) */ /* -#define BSR2_RET(pos, res) { UInt32 i = 6 + ((kNumLogBits - 1) & \ +#define BSR2_RET(pos, res) { unsigned zz = 6 + ((kNumLogBits - 1) & \ (0 - (((((UInt32)1 << (kNumLogBits + 6)) - 1) - pos) >> 31))); \ - res = p->g_FastPos[pos >> i] + (i * 2); } + res = p->g_FastPos[pos >> zz] + (zz * 2); } */ /* -#define BSR2_RET(pos, res) { UInt32 i = 6 + ((kNumLogBits - 1) & \ +#define BSR2_RET(pos, res) { unsigned zz = 6 + ((kNumLogBits - 1) & \ (0 - (((((UInt32)1 << (kNumLogBits)) - 1) - (pos >> 6)) >> 31))); \ - res = p->g_FastPos[pos >> i] + (i * 2); } + res = p->g_FastPos[pos >> zz] + (zz * 2); } */ -#define BSR2_RET(pos, res) { UInt32 i = (pos < (1 << (kNumLogBits + 6))) ? 6 : 6 + kNumLogBits - 1; \ - res = p->g_FastPos[pos >> i] + (i * 2); } +#define BSR2_RET(pos, res) { unsigned zz = (pos < (1 << (kNumLogBits + 6))) ? 6 : 6 + kNumLogBits - 1; \ + res = p->g_FastPos[pos >> zz] + (zz * 2); } /* #define BSR2_RET(pos, res) { res = (pos < (1 << (kNumLogBits + 6))) ? \ @@ -167,32 +161,35 @@ static void LzmaEnc_FastPosInit(Byte *g_FastPos) #define GetPosSlot1(pos) p->g_FastPos[pos] #define GetPosSlot2(pos, res) { BSR2_RET(pos, res); } -#define GetPosSlot(pos, res) { if (pos < kNumFullDistances) res = p->g_FastPos[pos]; else BSR2_RET(pos, res); } +#define GetPosSlot(pos, res) { if (pos < kNumFullDistances) res = p->g_FastPos[pos & (kNumFullDistances - 1)]; else BSR2_RET(pos, res); } #endif #define LZMA_NUM_REPS 4 -typedef unsigned CState; +typedef UInt16 CState; +typedef UInt16 CExtra; typedef struct { UInt32 price; - CState state; - int prev1IsChar; - int prev2; - - UInt32 posPrev2; - UInt32 backPrev2; - - UInt32 posPrev; - UInt32 backPrev; - UInt32 backs[LZMA_NUM_REPS]; + CExtra extra; + // 0 : normal + // 1 : LIT : MATCH + // > 1 : MATCH (extra-1) : LIT : REP0 (len) + UInt32 len; + UInt32 dist; + UInt32 reps[LZMA_NUM_REPS]; } COptimal; -#define kNumOpts (1 << 12) + +// 18.06 +#define kNumOpts (1 << 11) +#define kPackReserve (kNumOpts * 8) +// #define kNumOpts (1 << 12) +// #define kPackReserve (1 + kNumOpts * 2) #define kNumLenToPosStates 4 #define kNumPosSlotBits 6 @@ -200,22 +197,21 @@ typedef struct #define kDicLogSizeMax 32 #define kDistTableSizeMax (kDicLogSizeMax * 2) - #define kNumAlignBits 4 #define kAlignTableSize (1 << kNumAlignBits) #define kAlignMask (kAlignTableSize - 1) #define kStartPosModelIndex 4 #define kEndPosModelIndex 14 -#define kNumPosModels (kEndPosModelIndex - kStartPosModelIndex) - #define kNumFullDistances (1 << (kEndPosModelIndex >> 1)) +typedef #ifdef _LZMA_PROB32 -#define CLzmaProb UInt32 + UInt32 #else -#define CLzmaProb UInt16 + UInt16 #endif + CLzmaProb; #define LZMA_PB_MAX 4 #define LZMA_LC_MAX 8 @@ -223,15 +219,11 @@ typedef struct #define LZMA_NUM_PB_STATES_MAX (1 << LZMA_PB_MAX) - #define kLenNumLowBits 3 #define kLenNumLowSymbols (1 << kLenNumLowBits) -#define kLenNumMidBits 3 -#define kLenNumMidSymbols (1 << kLenNumMidBits) #define kLenNumHighBits 8 #define kLenNumHighSymbols (1 << kLenNumHighBits) - -#define kLenNumSymbolsTotal (kLenNumLowSymbols + kLenNumMidSymbols + kLenNumHighSymbols) +#define kLenNumSymbolsTotal (kLenNumLowSymbols * 2 + kLenNumHighSymbols) #define LZMA_MATCH_LEN_MIN 2 #define LZMA_MATCH_LEN_MAX (LZMA_MATCH_LEN_MIN + kLenNumSymbolsTotal - 1) @@ -241,27 +233,31 @@ typedef struct typedef struct { - CLzmaProb choice; - CLzmaProb choice2; - CLzmaProb low[LZMA_NUM_PB_STATES_MAX << kLenNumLowBits]; - CLzmaProb mid[LZMA_NUM_PB_STATES_MAX << kLenNumMidBits]; + CLzmaProb low[LZMA_NUM_PB_STATES_MAX << (kLenNumLowBits + 1)]; CLzmaProb high[kLenNumHighSymbols]; } CLenEnc; typedef struct { - CLenEnc p; - UInt32 tableSize; + unsigned tableSize; UInt32 prices[LZMA_NUM_PB_STATES_MAX][kLenNumSymbolsTotal]; - UInt32 counters[LZMA_NUM_PB_STATES_MAX]; + // UInt32 prices1[LZMA_NUM_PB_STATES_MAX][kLenNumLowSymbols * 2]; + // UInt32 prices2[kLenNumSymbolsTotal]; } CLenPriceEnc; +#define GET_PRICE_LEN(p, posState, len) \ + ((p)->prices[posState][(size_t)(len) - LZMA_MATCH_LEN_MIN]) + +/* +#define GET_PRICE_LEN(p, posState, len) \ + ((p)->prices2[(size_t)(len) - 2] + ((p)->prices1[posState][((len) - 2) & (kLenNumLowSymbols * 2 - 1)] & (((len) - 2 - kLenNumLowSymbols * 2) >> 9))) +*/ typedef struct { UInt32 range; - Byte cache; + unsigned cache; UInt64 low; UInt64 cacheSize; Byte *buf; @@ -277,69 +273,77 @@ typedef struct { CLzmaProb *litProbs; - UInt32 state; + unsigned state; UInt32 reps[LZMA_NUM_REPS]; - CLzmaProb isMatch[kNumStates][LZMA_NUM_PB_STATES_MAX]; + CLzmaProb posAlignEncoder[1 << kNumAlignBits]; CLzmaProb isRep[kNumStates]; CLzmaProb isRepG0[kNumStates]; CLzmaProb isRepG1[kNumStates]; CLzmaProb isRepG2[kNumStates]; + CLzmaProb isMatch[kNumStates][LZMA_NUM_PB_STATES_MAX]; CLzmaProb isRep0Long[kNumStates][LZMA_NUM_PB_STATES_MAX]; CLzmaProb posSlotEncoder[kNumLenToPosStates][1 << kNumPosSlotBits]; - CLzmaProb posEncoders[kNumFullDistances - kEndPosModelIndex]; - CLzmaProb posAlignEncoder[1 << kNumAlignBits]; + CLzmaProb posEncoders[kNumFullDistances]; - CLenPriceEnc lenEnc; - CLenPriceEnc repLenEnc; + CLenEnc lenProbs; + CLenEnc repLenProbs; + } CSaveState; +typedef UInt32 CProbPrice; + + typedef struct { void *matchFinderObj; IMatchFinder matchFinder; - UInt32 optimumEndIndex; - UInt32 optimumCurrentIndex; + unsigned optCur; + unsigned optEnd; - UInt32 longestMatchLength; - UInt32 numPairs; + unsigned longestMatchLen; + unsigned numPairs; UInt32 numAvail; - UInt32 numFastBytes; - UInt32 additionalOffset; + unsigned state; + unsigned numFastBytes; + unsigned additionalOffset; UInt32 reps[LZMA_NUM_REPS]; - UInt32 state; + unsigned lpMask, pbMask; + CLzmaProb *litProbs; + CRangeEnc rc; + + UInt32 backRes; unsigned lc, lp, pb; - unsigned lpMask, pbMask; unsigned lclp; - CLzmaProb *litProbs; - - Bool fastMode; - Bool writeEndMark; - Bool finished; - Bool multiThread; - Bool needInit; + BoolInt fastMode; + BoolInt writeEndMark; + BoolInt finished; + BoolInt multiThread; + BoolInt needInit; + // BoolInt _maxMode; UInt64 nowPos64; - UInt32 matchPriceCount; - UInt32 alignPriceCount; + unsigned matchPriceCount; + // unsigned alignPriceCount; + int repLenEncCounter; - UInt32 distTableSize; + unsigned distTableSize; UInt32 dictSize; SRes result; - CRangeEnc rc; - #ifndef _7ZIP_ST - Bool mtMode; + BoolInt mtMode; + // begin of CMatchFinderMt is used in LZ thread CMatchFinderMt matchFinderMt; + // end of CMatchFinderMt is used in BT and HASH threads #endif CMatchFinder matchFinderBase; @@ -348,33 +352,37 @@ typedef struct Byte pad[128]; #endif - COptimal opt[kNumOpts]; - - #ifndef LZMA_LOG_BSR - Byte g_FastPos[1 << kNumLogBits]; - #endif + // LZ thread + CProbPrice ProbPrices[kBitModelTotal >> kNumMoveReducingBits]; - UInt32 ProbPrices[kBitModelTotal >> kNumMoveReducingBits]; UInt32 matches[LZMA_MATCH_LEN_MAX * 2 + 2 + 1]; + UInt32 alignPrices[kAlignTableSize]; UInt32 posSlotPrices[kNumLenToPosStates][kDistTableSizeMax]; UInt32 distancesPrices[kNumLenToPosStates][kNumFullDistances]; - UInt32 alignPrices[kAlignTableSize]; - CLzmaProb isMatch[kNumStates][LZMA_NUM_PB_STATES_MAX]; + CLzmaProb posAlignEncoder[1 << kNumAlignBits]; CLzmaProb isRep[kNumStates]; CLzmaProb isRepG0[kNumStates]; CLzmaProb isRepG1[kNumStates]; CLzmaProb isRepG2[kNumStates]; + CLzmaProb isMatch[kNumStates][LZMA_NUM_PB_STATES_MAX]; CLzmaProb isRep0Long[kNumStates][LZMA_NUM_PB_STATES_MAX]; - CLzmaProb posSlotEncoder[kNumLenToPosStates][1 << kNumPosSlotBits]; - CLzmaProb posEncoders[kNumFullDistances - kEndPosModelIndex]; - CLzmaProb posAlignEncoder[1 << kNumAlignBits]; + CLzmaProb posEncoders[kNumFullDistances]; + CLenEnc lenProbs; + CLenEnc repLenProbs; + + #ifndef LZMA_LOG_BSR + Byte g_FastPos[1 << kNumLogBits]; + #endif + CLenPriceEnc lenEnc; CLenPriceEnc repLenEnc; + COptimal opt[kNumOpts]; + CSaveState saveState; #ifndef _7ZIP_ST @@ -383,58 +391,62 @@ typedef struct } CLzmaEnc; + +#define COPY_ARR(dest, src, arr) memcpy(dest->arr, src->arr, sizeof(src->arr)); + void LzmaEnc_SaveState(CLzmaEncHandle pp) { CLzmaEnc *p = (CLzmaEnc *)pp; CSaveState *dest = &p->saveState; - int i; - dest->lenEnc = p->lenEnc; - dest->repLenEnc = p->repLenEnc; + dest->state = p->state; + + dest->lenProbs = p->lenProbs; + dest->repLenProbs = p->repLenProbs; + + COPY_ARR(dest, p, reps); + + COPY_ARR(dest, p, posAlignEncoder); + COPY_ARR(dest, p, isRep); + COPY_ARR(dest, p, isRepG0); + COPY_ARR(dest, p, isRepG1); + COPY_ARR(dest, p, isRepG2); + COPY_ARR(dest, p, isMatch); + COPY_ARR(dest, p, isRep0Long); + COPY_ARR(dest, p, posSlotEncoder); + COPY_ARR(dest, p, posEncoders); - for (i = 0; i < kNumStates; i++) - { - memcpy(dest->isMatch[i], p->isMatch[i], sizeof(p->isMatch[i])); - memcpy(dest->isRep0Long[i], p->isRep0Long[i], sizeof(p->isRep0Long[i])); - } - for (i = 0; i < kNumLenToPosStates; i++) - memcpy(dest->posSlotEncoder[i], p->posSlotEncoder[i], sizeof(p->posSlotEncoder[i])); - memcpy(dest->isRep, p->isRep, sizeof(p->isRep)); - memcpy(dest->isRepG0, p->isRepG0, sizeof(p->isRepG0)); - memcpy(dest->isRepG1, p->isRepG1, sizeof(p->isRepG1)); - memcpy(dest->isRepG2, p->isRepG2, sizeof(p->isRepG2)); - memcpy(dest->posEncoders, p->posEncoders, sizeof(p->posEncoders)); - memcpy(dest->posAlignEncoder, p->posAlignEncoder, sizeof(p->posAlignEncoder)); - memcpy(dest->reps, p->reps, sizeof(p->reps)); memcpy(dest->litProbs, p->litProbs, ((UInt32)0x300 << p->lclp) * sizeof(CLzmaProb)); } + void LzmaEnc_RestoreState(CLzmaEncHandle pp) { CLzmaEnc *dest = (CLzmaEnc *)pp; const CSaveState *p = &dest->saveState; - int i; - dest->lenEnc = p->lenEnc; - dest->repLenEnc = p->repLenEnc; + dest->state = p->state; - for (i = 0; i < kNumStates; i++) - { - memcpy(dest->isMatch[i], p->isMatch[i], sizeof(p->isMatch[i])); - memcpy(dest->isRep0Long[i], p->isRep0Long[i], sizeof(p->isRep0Long[i])); - } - for (i = 0; i < kNumLenToPosStates; i++) - memcpy(dest->posSlotEncoder[i], p->posSlotEncoder[i], sizeof(p->posSlotEncoder[i])); - memcpy(dest->isRep, p->isRep, sizeof(p->isRep)); - memcpy(dest->isRepG0, p->isRepG0, sizeof(p->isRepG0)); - memcpy(dest->isRepG1, p->isRepG1, sizeof(p->isRepG1)); - memcpy(dest->isRepG2, p->isRepG2, sizeof(p->isRepG2)); - memcpy(dest->posEncoders, p->posEncoders, sizeof(p->posEncoders)); - memcpy(dest->posAlignEncoder, p->posAlignEncoder, sizeof(p->posAlignEncoder)); - memcpy(dest->reps, p->reps, sizeof(p->reps)); + dest->lenProbs = p->lenProbs; + dest->repLenProbs = p->repLenProbs; + + COPY_ARR(dest, p, reps); + + COPY_ARR(dest, p, posAlignEncoder); + COPY_ARR(dest, p, isRep); + COPY_ARR(dest, p, isRepG0); + COPY_ARR(dest, p, isRepG1); + COPY_ARR(dest, p, isRepG2); + COPY_ARR(dest, p, isMatch); + COPY_ARR(dest, p, isRep0Long); + COPY_ARR(dest, p, posSlotEncoder); + COPY_ARR(dest, p, posEncoders); + memcpy(dest->litProbs, p->litProbs, ((UInt32)0x300 << dest->lclp) * sizeof(CLzmaProb)); } + + SRes LzmaEnc_SetProps(CLzmaEncHandle pp, const CLzmaEncProps *props2) { CLzmaEnc *p = (CLzmaEnc *)pp; @@ -445,7 +457,7 @@ SRes LzmaEnc_SetProps(CLzmaEncHandle pp, const CLzmaEncProps *props2) || props.lp > LZMA_LP_MAX || props.pb > LZMA_PB_MAX || props.dictSize > ((UInt64)1 << kDicLogSizeMaxCompress) - || props.dictSize > kMaxHistorySize) + || props.dictSize > kLzmaMaxHistorySize) return SZ_ERROR_PARAM; p->dictSize = props.dictSize; @@ -461,9 +473,10 @@ SRes LzmaEnc_SetProps(CLzmaEncHandle pp, const CLzmaEncProps *props2) p->lp = props.lp; p->pb = props.pb; p->fastMode = (props.algo == 0); + // p->_maxMode = True; p->matchFinderBase.btMode = (Byte)(props.btMode ? 1 : 0); { - UInt32 numHashBytes = 4; + unsigned numHashBytes = 4; if (props.btMode) { if (props.numHashBytes < 2) @@ -492,13 +505,27 @@ SRes LzmaEnc_SetProps(CLzmaEncHandle pp, const CLzmaEncProps *props2) return SZ_OK; } -static const int kLiteralNextStates[kNumStates] = {0, 0, 0, 0, 1, 2, 3, 4, 5, 6, 4, 5}; -static const int kMatchNextStates[kNumStates] = {7, 7, 7, 7, 7, 7, 7, 10, 10, 10, 10, 10}; -static const int kRepNextStates[kNumStates] = {8, 8, 8, 8, 8, 8, 8, 11, 11, 11, 11, 11}; -static const int kShortRepNextStates[kNumStates]= {9, 9, 9, 9, 9, 9, 9, 11, 11, 11, 11, 11}; -#define IsCharState(s) ((s) < 7) +void LzmaEnc_SetDataSize(CLzmaEncHandle pp, UInt64 expectedDataSiize) +{ + CLzmaEnc *p = (CLzmaEnc *)pp; + p->matchFinderBase.expectedDataSize = expectedDataSiize; +} + +#define kState_Start 0 +#define kState_LitAfterMatch 4 +#define kState_LitAfterRep 5 +#define kState_MatchAfterLit 7 +#define kState_RepAfterLit 8 + +static const Byte kLiteralNextStates[kNumStates] = {0, 0, 0, 0, 1, 2, 3, 4, 5, 6, 4, 5}; +static const Byte kMatchNextStates[kNumStates] = {7, 7, 7, 7, 7, 7, 7, 10, 10, 10, 10, 10}; +static const Byte kRepNextStates[kNumStates] = {8, 8, 8, 8, 8, 8, 8, 11, 11, 11, 11, 11}; +static const Byte kShortRepNextStates[kNumStates]= {9, 9, 9, 9, 9, 9, 9, 11, 11, 11, 11, 11}; + +#define IsLitState(s) ((s) < 7) +#define GetLenToPosState2(len) (((len) < kNumLenToPosStates - 1) ? (len) : kNumLenToPosStates - 1) #define GetLenToPosState(len) (((len) < kNumLenToPosStates + 1) ? (len) - 2 : kNumLenToPosStates - 1) #define kInfinityPrice (1 << 30) @@ -509,14 +536,16 @@ static void RangeEnc_Construct(CRangeEnc *p) p->bufBase = NULL; } -#define RangeEnc_GetProcessed(p) ((p)->processed + ((p)->buf - (p)->bufBase) + (p)->cacheSize) +#define RangeEnc_GetProcessed(p) ((p)->processed + ((p)->buf - (p)->bufBase) + (p)->cacheSize) +#define RangeEnc_GetProcessed_sizet(p) ((size_t)(p)->processed + ((p)->buf - (p)->bufBase) + (size_t)(p)->cacheSize) #define RC_BUF_SIZE (1 << 16) -static int RangeEnc_Alloc(CRangeEnc *p, ISzAlloc *alloc) + +static int RangeEnc_Alloc(CRangeEnc *p, ISzAllocPtr alloc) { if (!p->bufBase) { - p->bufBase = (Byte *)alloc->Alloc(alloc, RC_BUF_SIZE); + p->bufBase = (Byte *)ISzAlloc_Alloc(alloc, RC_BUF_SIZE); if (!p->bufBase) return 0; p->bufLim = p->bufBase + RC_BUF_SIZE; @@ -524,19 +553,19 @@ static int RangeEnc_Alloc(CRangeEnc *p, ISzAlloc *alloc) return 1; } -static void RangeEnc_Free(CRangeEnc *p, ISzAlloc *alloc) +static void RangeEnc_Free(CRangeEnc *p, ISzAllocPtr alloc) { - alloc->Free(alloc, p->bufBase); + ISzAlloc_Free(alloc, p->bufBase); p->bufBase = 0; } static void RangeEnc_Init(CRangeEnc *p) { /* Stream.Init(); */ - p->low = 0; p->range = 0xFFFFFFFF; - p->cacheSize = 1; p->cache = 0; + p->low = 0; + p->cacheSize = 0; p->buf = p->bufBase; @@ -544,37 +573,48 @@ static void RangeEnc_Init(CRangeEnc *p) p->res = SZ_OK; } -static void RangeEnc_FlushStream(CRangeEnc *p) +MY_NO_INLINE static void RangeEnc_FlushStream(CRangeEnc *p) { size_t num; if (p->res != SZ_OK) return; num = p->buf - p->bufBase; - if (num != p->outStream->Write(p->outStream, p->bufBase, num)) + if (num != ISeqOutStream_Write(p->outStream, p->bufBase, num)) p->res = SZ_ERROR_WRITE; p->processed += num; p->buf = p->bufBase; } -static void MY_FAST_CALL RangeEnc_ShiftLow(CRangeEnc *p) +MY_NO_INLINE static void MY_FAST_CALL RangeEnc_ShiftLow(CRangeEnc *p) { - if ((UInt32)p->low < (UInt32)0xFF000000 || (unsigned)(p->low >> 32) != 0) + UInt32 low = (UInt32)p->low; + unsigned high = (unsigned)(p->low >> 32); + p->low = (UInt32)(low << 8); + if (low < (UInt32)0xFF000000 || high != 0) { - Byte temp = p->cache; - do { Byte *buf = p->buf; - *buf++ = (Byte)(temp + (Byte)(p->low >> 32)); + *buf++ = (Byte)(p->cache + high); + p->cache = (unsigned)(low >> 24); p->buf = buf; if (buf == p->bufLim) RangeEnc_FlushStream(p); - temp = 0xFF; + if (p->cacheSize == 0) + return; + } + high += 0xFF; + for (;;) + { + Byte *buf = p->buf; + *buf++ = (Byte)(high); + p->buf = buf; + if (buf == p->bufLim) + RangeEnc_FlushStream(p); + if (--p->cacheSize == 0) + return; } - while (--p->cacheSize != 0); - p->cache = (Byte)((UInt32)p->low >> 24); } p->cacheSize++; - p->low = (UInt32)p->low << 8; } static void RangeEnc_FlushData(CRangeEnc *p) @@ -584,78 +624,121 @@ static void RangeEnc_FlushData(CRangeEnc *p) RangeEnc_ShiftLow(p); } -static void RangeEnc_EncodeDirectBits(CRangeEnc *p, UInt32 value, unsigned numBits) +#define RC_NORM(p) if (range < kTopValue) { range <<= 8; RangeEnc_ShiftLow(p); } + +#define RC_BIT_PRE(p, prob) \ + ttt = *(prob); \ + newBound = (range >> kNumBitModelTotalBits) * ttt; + +// #define _LZMA_ENC_USE_BRANCH + +#ifdef _LZMA_ENC_USE_BRANCH + +#define RC_BIT(p, prob, bit) { \ + RC_BIT_PRE(p, prob) \ + if (bit == 0) { range = newBound; ttt += (kBitModelTotal - ttt) >> kNumMoveBits; } \ + else { (p)->low += newBound; range -= newBound; ttt -= ttt >> kNumMoveBits; } \ + *(prob) = (CLzmaProb)ttt; \ + RC_NORM(p) \ + } + +#else + +#define RC_BIT(p, prob, bit) { \ + UInt32 mask; \ + RC_BIT_PRE(p, prob) \ + mask = 0 - (UInt32)bit; \ + range &= mask; \ + mask &= newBound; \ + range -= mask; \ + (p)->low += mask; \ + mask = (UInt32)bit - 1; \ + range += newBound & mask; \ + mask &= (kBitModelTotal - ((1 << kNumMoveBits) - 1)); \ + mask += ((1 << kNumMoveBits) - 1); \ + ttt += (Int32)(mask - ttt) >> kNumMoveBits; \ + *(prob) = (CLzmaProb)ttt; \ + RC_NORM(p) \ + } + +#endif + + + + +#define RC_BIT_0_BASE(p, prob) \ + range = newBound; *(prob) = (CLzmaProb)(ttt + ((kBitModelTotal - ttt) >> kNumMoveBits)); + +#define RC_BIT_1_BASE(p, prob) \ + range -= newBound; (p)->low += newBound; *(prob) = (CLzmaProb)(ttt - (ttt >> kNumMoveBits)); \ + +#define RC_BIT_0(p, prob) \ + RC_BIT_0_BASE(p, prob) \ + RC_NORM(p) + +#define RC_BIT_1(p, prob) \ + RC_BIT_1_BASE(p, prob) \ + RC_NORM(p) + +static void RangeEnc_EncodeBit_0(CRangeEnc *p, CLzmaProb *prob) { + UInt32 range, ttt, newBound; + range = p->range; + RC_BIT_PRE(p, prob) + RC_BIT_0(p, prob) + p->range = range; +} + +static void LitEnc_Encode(CRangeEnc *p, CLzmaProb *probs, UInt32 sym) +{ + UInt32 range = p->range; + sym |= 0x100; do { - p->range >>= 1; - p->low += p->range & (0 - ((value >> --numBits) & 1)); - if (p->range < kTopValue) - { - p->range <<= 8; - RangeEnc_ShiftLow(p); - } + UInt32 ttt, newBound; + // RangeEnc_EncodeBit(p, probs + (sym >> 8), (sym >> 7) & 1); + CLzmaProb *prob = probs + (sym >> 8); + UInt32 bit = (sym >> 7) & 1; + sym <<= 1; + RC_BIT(p, prob, bit); } - while (numBits != 0); + while (sym < 0x10000); + p->range = range; } -static void RangeEnc_EncodeBit(CRangeEnc *p, CLzmaProb *prob, UInt32 symbol) -{ - UInt32 ttt = *prob; - UInt32 newBound = (p->range >> kNumBitModelTotalBits) * ttt; - if (symbol == 0) - { - p->range = newBound; - ttt += (kBitModelTotal - ttt) >> kNumMoveBits; - } - else - { - p->low += newBound; - p->range -= newBound; - ttt -= ttt >> kNumMoveBits; - } - *prob = (CLzmaProb)ttt; - if (p->range < kTopValue) - { - p->range <<= 8; - RangeEnc_ShiftLow(p); - } -} - -static void LitEnc_Encode(CRangeEnc *p, CLzmaProb *probs, UInt32 symbol) -{ - symbol |= 0x100; - do - { - RangeEnc_EncodeBit(p, probs + (symbol >> 8), (symbol >> 7) & 1); - symbol <<= 1; - } - while (symbol < 0x10000); -} - -static void LitEnc_EncodeMatched(CRangeEnc *p, CLzmaProb *probs, UInt32 symbol, UInt32 matchByte) +static void LitEnc_EncodeMatched(CRangeEnc *p, CLzmaProb *probs, UInt32 sym, UInt32 matchByte) { + UInt32 range = p->range; UInt32 offs = 0x100; - symbol |= 0x100; + sym |= 0x100; do { + UInt32 ttt, newBound; + CLzmaProb *prob; + UInt32 bit; matchByte <<= 1; - RangeEnc_EncodeBit(p, probs + (offs + (matchByte & offs) + (symbol >> 8)), (symbol >> 7) & 1); - symbol <<= 1; - offs &= ~(matchByte ^ symbol); + // RangeEnc_EncodeBit(p, probs + (offs + (matchByte & offs) + (sym >> 8)), (sym >> 7) & 1); + prob = probs + (offs + (matchByte & offs) + (sym >> 8)); + bit = (sym >> 7) & 1; + sym <<= 1; + offs &= ~(matchByte ^ sym); + RC_BIT(p, prob, bit); } - while (symbol < 0x10000); + while (sym < 0x10000); + p->range = range; } -static void LzmaEnc_InitPriceTables(UInt32 *ProbPrices) + + +static void LzmaEnc_InitPriceTables(CProbPrice *ProbPrices) { UInt32 i; - for (i = (1 << kNumMoveReducingBits) / 2; i < kBitModelTotal; i += (1 << kNumMoveReducingBits)) + for (i = 0; i < (kBitModelTotal >> kNumMoveReducingBits); i++) { - const int kCyclesBits = kNumBitPriceShiftBits; - UInt32 w = i; - UInt32 bitCount = 0; - int j; + const unsigned kCyclesBits = kNumBitPriceShiftBits; + UInt32 w = (i << kNumMoveReducingBits) + (1 << (kNumMoveReducingBits - 1)); + unsigned bitCount = 0; + unsigned j; for (j = 0; j < kCyclesBits; j++) { w = w * w; @@ -666,559 +749,655 @@ static void LzmaEnc_InitPriceTables(UInt32 *ProbPrices) bitCount++; } } - ProbPrices[i >> kNumMoveReducingBits] = ((kNumBitModelTotalBits << kCyclesBits) - 15 - bitCount); + ProbPrices[i] = (CProbPrice)((kNumBitModelTotalBits << kCyclesBits) - 15 - bitCount); + // printf("\n%3d: %5d", i, ProbPrices[i]); } } -#define GET_PRICE(prob, symbol) \ - p->ProbPrices[((prob) ^ (((-(int)(symbol))) & (kBitModelTotal - 1))) >> kNumMoveReducingBits]; +#define GET_PRICE(prob, bit) \ + p->ProbPrices[((prob) ^ (unsigned)(((-(int)(bit))) & (kBitModelTotal - 1))) >> kNumMoveReducingBits]; -#define GET_PRICEa(prob, symbol) \ - ProbPrices[((prob) ^ ((-((int)(symbol))) & (kBitModelTotal - 1))) >> kNumMoveReducingBits]; +#define GET_PRICEa(prob, bit) \ + ProbPrices[((prob) ^ (unsigned)((-((int)(bit))) & (kBitModelTotal - 1))) >> kNumMoveReducingBits]; #define GET_PRICE_0(prob) p->ProbPrices[(prob) >> kNumMoveReducingBits] #define GET_PRICE_1(prob) p->ProbPrices[((prob) ^ (kBitModelTotal - 1)) >> kNumMoveReducingBits] -#define GET_PRICE_0a(prob) ProbPrices[(prob) >> kNumMoveReducingBits] -#define GET_PRICE_1a(prob) ProbPrices[((prob) ^ (kBitModelTotal - 1)) >> kNumMoveReducingBits] +#define GET_PRICEa_0(prob) ProbPrices[(prob) >> kNumMoveReducingBits] +#define GET_PRICEa_1(prob) ProbPrices[((prob) ^ (kBitModelTotal - 1)) >> kNumMoveReducingBits] -static UInt32 LitEnc_GetPrice(const CLzmaProb *probs, UInt32 symbol, const UInt32 *ProbPrices) + +static UInt32 LitEnc_GetPrice(const CLzmaProb *probs, UInt32 sym, const CProbPrice *ProbPrices) { UInt32 price = 0; - symbol |= 0x100; + sym |= 0x100; do { - price += GET_PRICEa(probs[symbol >> 8], (symbol >> 7) & 1); - symbol <<= 1; + unsigned bit = sym & 1; + sym >>= 1; + price += GET_PRICEa(probs[sym], bit); } - while (symbol < 0x10000); + while (sym >= 2); return price; } -static UInt32 LitEnc_GetPriceMatched(const CLzmaProb *probs, UInt32 symbol, UInt32 matchByte, const UInt32 *ProbPrices) + +static UInt32 LitEnc_Matched_GetPrice(const CLzmaProb *probs, UInt32 sym, UInt32 matchByte, const CProbPrice *ProbPrices) { UInt32 price = 0; UInt32 offs = 0x100; - symbol |= 0x100; + sym |= 0x100; do { matchByte <<= 1; - price += GET_PRICEa(probs[offs + (matchByte & offs) + (symbol >> 8)], (symbol >> 7) & 1); - symbol <<= 1; - offs &= ~(matchByte ^ symbol); + price += GET_PRICEa(probs[offs + (matchByte & offs) + (sym >> 8)], (sym >> 7) & 1); + sym <<= 1; + offs &= ~(matchByte ^ sym); } - while (symbol < 0x10000); + while (sym < 0x10000); return price; } -static void RcTree_Encode(CRangeEnc *rc, CLzmaProb *probs, int numBitLevels, UInt32 symbol) +static void RcTree_ReverseEncode(CRangeEnc *rc, CLzmaProb *probs, unsigned numBits, unsigned sym) { - UInt32 m = 1; - int i; - for (i = numBitLevels; i != 0;) + UInt32 range = rc->range; + unsigned m = 1; + do { - UInt32 bit; - i--; - bit = (symbol >> i) & 1; - RangeEnc_EncodeBit(rc, probs + m, bit); + UInt32 ttt, newBound; + unsigned bit = sym & 1; + // RangeEnc_EncodeBit(rc, probs + m, bit); + sym >>= 1; + RC_BIT(rc, probs + m, bit); m = (m << 1) | bit; } + while (--numBits); + rc->range = range; } -static void RcTree_ReverseEncode(CRangeEnc *rc, CLzmaProb *probs, int numBitLevels, UInt32 symbol) -{ - UInt32 m = 1; - int i; - for (i = 0; i < numBitLevels; i++) - { - UInt32 bit = symbol & 1; - RangeEnc_EncodeBit(rc, probs + m, bit); - m = (m << 1) | bit; - symbol >>= 1; - } -} - -static UInt32 RcTree_GetPrice(const CLzmaProb *probs, int numBitLevels, UInt32 symbol, const UInt32 *ProbPrices) -{ - UInt32 price = 0; - symbol |= (1 << numBitLevels); - while (symbol != 1) - { - price += GET_PRICEa(probs[symbol >> 1], symbol & 1); - symbol >>= 1; - } - return price; -} - -static UInt32 RcTree_ReverseGetPrice(const CLzmaProb *probs, int numBitLevels, UInt32 symbol, const UInt32 *ProbPrices) -{ - UInt32 price = 0; - UInt32 m = 1; - int i; - for (i = numBitLevels; i != 0; i--) - { - UInt32 bit = symbol & 1; - symbol >>= 1; - price += GET_PRICEa(probs[m], bit); - m = (m << 1) | bit; - } - return price; -} static void LenEnc_Init(CLenEnc *p) { unsigned i; - p->choice = p->choice2 = kProbInitValue; - for (i = 0; i < (LZMA_NUM_PB_STATES_MAX << kLenNumLowBits); i++) + for (i = 0; i < (LZMA_NUM_PB_STATES_MAX << (kLenNumLowBits + 1)); i++) p->low[i] = kProbInitValue; - for (i = 0; i < (LZMA_NUM_PB_STATES_MAX << kLenNumMidBits); i++) - p->mid[i] = kProbInitValue; for (i = 0; i < kLenNumHighSymbols; i++) p->high[i] = kProbInitValue; } -static void LenEnc_Encode(CLenEnc *p, CRangeEnc *rc, UInt32 symbol, UInt32 posState) +static void LenEnc_Encode(CLenEnc *p, CRangeEnc *rc, unsigned sym, unsigned posState) { - if (symbol < kLenNumLowSymbols) + UInt32 range, ttt, newBound; + CLzmaProb *probs = p->low; + range = rc->range; + RC_BIT_PRE(rc, probs); + if (sym >= kLenNumLowSymbols) { - RangeEnc_EncodeBit(rc, &p->choice, 0); - RcTree_Encode(rc, p->low + (posState << kLenNumLowBits), kLenNumLowBits, symbol); - } - else - { - RangeEnc_EncodeBit(rc, &p->choice, 1); - if (symbol < kLenNumLowSymbols + kLenNumMidSymbols) + RC_BIT_1(rc, probs); + probs += kLenNumLowSymbols; + RC_BIT_PRE(rc, probs); + if (sym >= kLenNumLowSymbols * 2) { - RangeEnc_EncodeBit(rc, &p->choice2, 0); - RcTree_Encode(rc, p->mid + (posState << kLenNumMidBits), kLenNumMidBits, symbol - kLenNumLowSymbols); + RC_BIT_1(rc, probs); + rc->range = range; + // RcTree_Encode(rc, p->high, kLenNumHighBits, sym - kLenNumLowSymbols * 2); + LitEnc_Encode(rc, p->high, sym - kLenNumLowSymbols * 2); + return; } - else + sym -= kLenNumLowSymbols; + } + + // RcTree_Encode(rc, probs + (posState << kLenNumLowBits), kLenNumLowBits, sym); + { + unsigned m; + unsigned bit; + RC_BIT_0(rc, probs); + probs += (posState << (1 + kLenNumLowBits)); + bit = (sym >> 2) ; RC_BIT(rc, probs + 1, bit); m = (1 << 1) + bit; + bit = (sym >> 1) & 1; RC_BIT(rc, probs + m, bit); m = (m << 1) + bit; + bit = sym & 1; RC_BIT(rc, probs + m, bit); + rc->range = range; + } +} + +static void SetPrices_3(const CLzmaProb *probs, UInt32 startPrice, UInt32 *prices, const CProbPrice *ProbPrices) +{ + unsigned i; + for (i = 0; i < 8; i += 2) + { + UInt32 price = startPrice; + UInt32 prob; + price += GET_PRICEa(probs[1 ], (i >> 2)); + price += GET_PRICEa(probs[2 + (i >> 2)], (i >> 1) & 1); + prob = probs[4 + (i >> 1)]; + prices[i ] = price + GET_PRICEa_0(prob); + prices[i + 1] = price + GET_PRICEa_1(prob); + } +} + + +MY_NO_INLINE static void MY_FAST_CALL LenPriceEnc_UpdateTables( + CLenPriceEnc *p, + unsigned numPosStates, + const CLenEnc *enc, + const CProbPrice *ProbPrices) +{ + UInt32 b; + + { + unsigned prob = enc->low[0]; + UInt32 a, c; + unsigned posState; + b = GET_PRICEa_1(prob); + a = GET_PRICEa_0(prob); + c = b + GET_PRICEa_0(enc->low[kLenNumLowSymbols]); + for (posState = 0; posState < numPosStates; posState++) { - RangeEnc_EncodeBit(rc, &p->choice2, 1); - RcTree_Encode(rc, p->high, kLenNumHighBits, symbol - kLenNumLowSymbols - kLenNumMidSymbols); + UInt32 *prices = p->prices[posState]; + const CLzmaProb *probs = enc->low + (posState << (1 + kLenNumLowBits)); + SetPrices_3(probs, a, prices, ProbPrices); + SetPrices_3(probs + kLenNumLowSymbols, c, prices + kLenNumLowSymbols, ProbPrices); + } + } + + /* + { + unsigned i; + UInt32 b; + a = GET_PRICEa_0(enc->low[0]); + for (i = 0; i < kLenNumLowSymbols; i++) + p->prices2[i] = a; + a = GET_PRICEa_1(enc->low[0]); + b = a + GET_PRICEa_0(enc->low[kLenNumLowSymbols]); + for (i = kLenNumLowSymbols; i < kLenNumLowSymbols * 2; i++) + p->prices2[i] = b; + a += GET_PRICEa_1(enc->low[kLenNumLowSymbols]); + } + */ + + // p->counter = numSymbols; + // p->counter = 64; + + { + unsigned i = p->tableSize; + + if (i > kLenNumLowSymbols * 2) + { + const CLzmaProb *probs = enc->high; + UInt32 *prices = p->prices[0] + kLenNumLowSymbols * 2; + i -= kLenNumLowSymbols * 2 - 1; + i >>= 1; + b += GET_PRICEa_1(enc->low[kLenNumLowSymbols]); + do + { + /* + p->prices2[i] = a + + // RcTree_GetPrice(enc->high, kLenNumHighBits, i - kLenNumLowSymbols * 2, ProbPrices); + LitEnc_GetPrice(probs, i - kLenNumLowSymbols * 2, ProbPrices); + */ + // UInt32 price = a + RcTree_GetPrice(probs, kLenNumHighBits - 1, sym, ProbPrices); + unsigned sym = --i + (1 << (kLenNumHighBits - 1)); + UInt32 price = b; + do + { + unsigned bit = sym & 1; + sym >>= 1; + price += GET_PRICEa(probs[sym], bit); + } + while (sym >= 2); + + { + unsigned prob = probs[(size_t)i + (1 << (kLenNumHighBits - 1))]; + prices[(size_t)i * 2 ] = price + GET_PRICEa_0(prob); + prices[(size_t)i * 2 + 1] = price + GET_PRICEa_1(prob); + } + } + while (i); + + { + unsigned posState; + size_t num = (p->tableSize - kLenNumLowSymbols * 2) * sizeof(p->prices[0][0]); + for (posState = 1; posState < numPosStates; posState++) + memcpy(p->prices[posState] + kLenNumLowSymbols * 2, p->prices[0] + kLenNumLowSymbols * 2, num); + } } } } -static void LenEnc_SetPrices(CLenEnc *p, UInt32 posState, UInt32 numSymbols, UInt32 *prices, const UInt32 *ProbPrices) -{ - UInt32 a0 = GET_PRICE_0a(p->choice); - UInt32 a1 = GET_PRICE_1a(p->choice); - UInt32 b0 = a1 + GET_PRICE_0a(p->choice2); - UInt32 b1 = a1 + GET_PRICE_1a(p->choice2); - UInt32 i = 0; - for (i = 0; i < kLenNumLowSymbols; i++) - { - if (i >= numSymbols) - return; - prices[i] = a0 + RcTree_GetPrice(p->low + (posState << kLenNumLowBits), kLenNumLowBits, i, ProbPrices); - } - for (; i < kLenNumLowSymbols + kLenNumMidSymbols; i++) - { - if (i >= numSymbols) - return; - prices[i] = b0 + RcTree_GetPrice(p->mid + (posState << kLenNumMidBits), kLenNumMidBits, i - kLenNumLowSymbols, ProbPrices); - } - for (; i < numSymbols; i++) - prices[i] = b1 + RcTree_GetPrice(p->high, kLenNumHighBits, i - kLenNumLowSymbols - kLenNumMidSymbols, ProbPrices); -} - -static void MY_FAST_CALL LenPriceEnc_UpdateTable(CLenPriceEnc *p, UInt32 posState, const UInt32 *ProbPrices) -{ - LenEnc_SetPrices(&p->p, posState, p->tableSize, p->prices[posState], ProbPrices); - p->counters[posState] = p->tableSize; -} - -static void LenPriceEnc_UpdateTables(CLenPriceEnc *p, UInt32 numPosStates, const UInt32 *ProbPrices) -{ - UInt32 posState; - for (posState = 0; posState < numPosStates; posState++) - LenPriceEnc_UpdateTable(p, posState, ProbPrices); -} - -static void LenEnc_Encode2(CLenPriceEnc *p, CRangeEnc *rc, UInt32 symbol, UInt32 posState, Bool updatePrice, const UInt32 *ProbPrices) -{ - LenEnc_Encode(&p->p, rc, symbol, posState); - if (updatePrice) - if (--p->counters[posState] == 0) - LenPriceEnc_UpdateTable(p, posState, ProbPrices); -} - - - - -static void MovePos(CLzmaEnc *p, UInt32 num) -{ +/* #ifdef SHOW_STAT g_STAT_OFFSET += num; printf("\n MovePos %u", num); #endif +*/ - if (num != 0) - { - p->additionalOffset += num; - p->matchFinder.Skip(p->matchFinderObj, num); - } -} +#define MOVE_POS(p, num) { \ + p->additionalOffset += (num); \ + p->matchFinder.Skip(p->matchFinderObj, (UInt32)(num)); } -static UInt32 ReadMatchDistances(CLzmaEnc *p, UInt32 *numDistancePairsRes) + +static unsigned ReadMatchDistances(CLzmaEnc *p, unsigned *numPairsRes) { - UInt32 lenRes = 0, numPairs; + unsigned numPairs; + + p->additionalOffset++; p->numAvail = p->matchFinder.GetNumAvailableBytes(p->matchFinderObj); numPairs = p->matchFinder.GetMatches(p->matchFinderObj, p->matches); + *numPairsRes = numPairs; #ifdef SHOW_STAT printf("\n i = %u numPairs = %u ", g_STAT_OFFSET, numPairs / 2); g_STAT_OFFSET++; { - UInt32 i; + unsigned i; for (i = 0; i < numPairs; i += 2) printf("%2u %6u | ", p->matches[i], p->matches[i + 1]); } #endif - if (numPairs > 0) + if (numPairs == 0) + return 0; { - lenRes = p->matches[numPairs - 2]; - if (lenRes == p->numFastBytes) + unsigned len = p->matches[(size_t)numPairs - 2]; + if (len != p->numFastBytes) + return len; { UInt32 numAvail = p->numAvail; if (numAvail > LZMA_MATCH_LEN_MAX) numAvail = LZMA_MATCH_LEN_MAX; { - const Byte *pbyCur = p->matchFinder.GetPointerToCurrentPos(p->matchFinderObj) - 1; - const Byte *pby = pbyCur + lenRes; - ptrdiff_t dif = (ptrdiff_t)-1 - p->matches[numPairs - 1]; - const Byte *pbyLim = pbyCur + numAvail; - for (; pby != pbyLim && *pby == pby[dif]; pby++); - lenRes = (UInt32)(pby - pbyCur); + const Byte *p1 = p->matchFinder.GetPointerToCurrentPos(p->matchFinderObj) - 1; + const Byte *p2 = p1 + len; + ptrdiff_t dif = (ptrdiff_t)-1 - p->matches[(size_t)numPairs - 1]; + const Byte *lim = p1 + numAvail; + for (; p2 != lim && *p2 == p2[dif]; p2++) + {} + return (unsigned)(p2 - p1); } } } - p->additionalOffset++; - *numDistancePairsRes = numPairs; - return lenRes; } +#define MARK_LIT ((UInt32)(Int32)-1) -#define MakeAsChar(p) (p)->backPrev = (UInt32)(-1); (p)->prev1IsChar = False; -#define MakeAsShortRep(p) (p)->backPrev = 0; (p)->prev1IsChar = False; -#define IsShortRep(p) ((p)->backPrev == 0) +#define MakeAs_Lit(p) { (p)->dist = MARK_LIT; (p)->extra = 0; } +#define MakeAs_ShortRep(p) { (p)->dist = 0; (p)->extra = 0; } +#define IsShortRep(p) ((p)->dist == 0) -static UInt32 GetRepLen1Price(CLzmaEnc *p, UInt32 state, UInt32 posState) -{ - return - GET_PRICE_0(p->isRepG0[state]) + - GET_PRICE_0(p->isRep0Long[state][posState]); -} -static UInt32 GetPureRepPrice(CLzmaEnc *p, UInt32 repIndex, UInt32 state, UInt32 posState) +#define GetPrice_ShortRep(p, state, posState) \ + ( GET_PRICE_0(p->isRepG0[state]) + GET_PRICE_0(p->isRep0Long[state][posState])) + +#define GetPrice_Rep_0(p, state, posState) ( \ + GET_PRICE_1(p->isMatch[state][posState]) \ + + GET_PRICE_1(p->isRep0Long[state][posState])) \ + + GET_PRICE_1(p->isRep[state]) \ + + GET_PRICE_0(p->isRepG0[state]) + +MY_FORCE_INLINE +static UInt32 GetPrice_PureRep(const CLzmaEnc *p, unsigned repIndex, size_t state, size_t posState) { UInt32 price; + UInt32 prob = p->isRepG0[state]; if (repIndex == 0) { - price = GET_PRICE_0(p->isRepG0[state]); + price = GET_PRICE_0(prob); price += GET_PRICE_1(p->isRep0Long[state][posState]); } else { - price = GET_PRICE_1(p->isRepG0[state]); + price = GET_PRICE_1(prob); + prob = p->isRepG1[state]; if (repIndex == 1) - price += GET_PRICE_0(p->isRepG1[state]); + price += GET_PRICE_0(prob); else { - price += GET_PRICE_1(p->isRepG1[state]); + price += GET_PRICE_1(prob); price += GET_PRICE(p->isRepG2[state], repIndex - 2); } } return price; } -static UInt32 GetRepPrice(CLzmaEnc *p, UInt32 repIndex, UInt32 len, UInt32 state, UInt32 posState) -{ - return p->repLenEnc.prices[posState][len - LZMA_MATCH_LEN_MIN] + - GetPureRepPrice(p, repIndex, state, posState); -} -static UInt32 Backward(CLzmaEnc *p, UInt32 *backRes, UInt32 cur) +static unsigned Backward(CLzmaEnc *p, unsigned cur) { - UInt32 posMem = p->opt[cur].posPrev; - UInt32 backMem = p->opt[cur].backPrev; - p->optimumEndIndex = cur; - do + unsigned wr = cur + 1; + p->optEnd = wr; + + for (;;) { - if (p->opt[cur].prev1IsChar) + UInt32 dist = p->opt[cur].dist; + unsigned len = (unsigned)p->opt[cur].len; + unsigned extra = (unsigned)p->opt[cur].extra; + cur -= len; + + if (extra) { - MakeAsChar(&p->opt[posMem]) - p->opt[posMem].posPrev = posMem - 1; - if (p->opt[cur].prev2) + wr--; + p->opt[wr].len = (UInt32)len; + cur -= extra; + len = extra; + if (extra == 1) { - p->opt[posMem - 1].prev1IsChar = False; - p->opt[posMem - 1].posPrev = p->opt[cur].posPrev2; - p->opt[posMem - 1].backPrev = p->opt[cur].backPrev2; + p->opt[wr].dist = dist; + dist = MARK_LIT; } - } - { - UInt32 posPrev = posMem; - UInt32 backCur = backMem; - - backMem = p->opt[posPrev].backPrev; - posMem = p->opt[posPrev].posPrev; - - p->opt[posPrev].backPrev = backCur; - p->opt[posPrev].posPrev = cur; - cur = posPrev; - } - } - while (cur != 0); - *backRes = p->opt[0].backPrev; - p->optimumCurrentIndex = p->opt[0].posPrev; - return p->optimumCurrentIndex; -} - -#define LIT_PROBS(pos, prevByte) (p->litProbs + ((((pos) & p->lpMask) << p->lc) + ((prevByte) >> (8 - p->lc))) * (UInt32)0x300) - -static UInt32 GetOptimum(CLzmaEnc *p, UInt32 position, UInt32 *backRes) -{ - UInt32 numAvail, mainLen, numPairs, repMaxIndex, i, posState, lenEnd, len, cur; - UInt32 matchPrice, repMatchPrice, normalMatchPrice; - UInt32 reps[LZMA_NUM_REPS], repLens[LZMA_NUM_REPS]; - UInt32 *matches; - const Byte *data; - Byte curByte, matchByte; - if (p->optimumEndIndex != p->optimumCurrentIndex) - { - const COptimal *opt = &p->opt[p->optimumCurrentIndex]; - UInt32 lenRes = opt->posPrev - p->optimumCurrentIndex; - *backRes = opt->backPrev; - p->optimumCurrentIndex = opt->posPrev; - return lenRes; - } - p->optimumCurrentIndex = p->optimumEndIndex = 0; - - if (p->additionalOffset == 0) - mainLen = ReadMatchDistances(p, &numPairs); - else - { - mainLen = p->longestMatchLength; - numPairs = p->numPairs; - } - - numAvail = p->numAvail; - if (numAvail < 2) - { - *backRes = (UInt32)(-1); - return 1; - } - if (numAvail > LZMA_MATCH_LEN_MAX) - numAvail = LZMA_MATCH_LEN_MAX; - - data = p->matchFinder.GetPointerToCurrentPos(p->matchFinderObj) - 1; - repMaxIndex = 0; - for (i = 0; i < LZMA_NUM_REPS; i++) - { - UInt32 lenTest; - const Byte *data2; - reps[i] = p->reps[i]; - data2 = data - reps[i] - 1; - if (data[0] != data2[0] || data[1] != data2[1]) - { - repLens[i] = 0; - continue; - } - for (lenTest = 2; lenTest < numAvail && data[lenTest] == data2[lenTest]; lenTest++); - repLens[i] = lenTest; - if (lenTest > repLens[repMaxIndex]) - repMaxIndex = i; - } - if (repLens[repMaxIndex] >= p->numFastBytes) - { - UInt32 lenRes; - *backRes = repMaxIndex; - lenRes = repLens[repMaxIndex]; - MovePos(p, lenRes - 1); - return lenRes; - } - - matches = p->matches; - if (mainLen >= p->numFastBytes) - { - *backRes = matches[numPairs - 1] + LZMA_NUM_REPS; - MovePos(p, mainLen - 1); - return mainLen; - } - curByte = *data; - matchByte = *(data - (reps[0] + 1)); - - if (mainLen < 2 && curByte != matchByte && repLens[repMaxIndex] < 2) - { - *backRes = (UInt32)-1; - return 1; - } - - p->opt[0].state = (CState)p->state; - - posState = (position & p->pbMask); - - { - const CLzmaProb *probs = LIT_PROBS(position, *(data - 1)); - p->opt[1].price = GET_PRICE_0(p->isMatch[p->state][posState]) + - (!IsCharState(p->state) ? - LitEnc_GetPriceMatched(probs, curByte, matchByte, p->ProbPrices) : - LitEnc_GetPrice(probs, curByte, p->ProbPrices)); - } - - MakeAsChar(&p->opt[1]); - - matchPrice = GET_PRICE_1(p->isMatch[p->state][posState]); - repMatchPrice = matchPrice + GET_PRICE_1(p->isRep[p->state]); - - if (matchByte == curByte) - { - UInt32 shortRepPrice = repMatchPrice + GetRepLen1Price(p, p->state, posState); - if (shortRepPrice < p->opt[1].price) - { - p->opt[1].price = shortRepPrice; - MakeAsShortRep(&p->opt[1]); - } - } - lenEnd = ((mainLen >= repLens[repMaxIndex]) ? mainLen : repLens[repMaxIndex]); - - if (lenEnd < 2) - { - *backRes = p->opt[1].backPrev; - return 1; - } - - p->opt[1].posPrev = 0; - for (i = 0; i < LZMA_NUM_REPS; i++) - p->opt[0].backs[i] = reps[i]; - - len = lenEnd; - do - p->opt[len--].price = kInfinityPrice; - while (len >= 2); - - for (i = 0; i < LZMA_NUM_REPS; i++) - { - UInt32 repLen = repLens[i]; - UInt32 price; - if (repLen < 2) - continue; - price = repMatchPrice + GetPureRepPrice(p, i, p->state, posState); - do - { - UInt32 curAndLenPrice = price + p->repLenEnc.prices[posState][repLen - 2]; - COptimal *opt = &p->opt[repLen]; - if (curAndLenPrice < opt->price) - { - opt->price = curAndLenPrice; - opt->posPrev = 0; - opt->backPrev = i; - opt->prev1IsChar = False; - } - } - while (--repLen >= 2); - } - - normalMatchPrice = matchPrice + GET_PRICE_0(p->isRep[p->state]); - - len = ((repLens[0] >= 2) ? repLens[0] + 1 : 2); - if (len <= mainLen) - { - UInt32 offs = 0; - while (len > matches[offs]) - offs += 2; - for (; ; len++) - { - COptimal *opt; - UInt32 distance = matches[offs + 1]; - - UInt32 curAndLenPrice = normalMatchPrice + p->lenEnc.prices[posState][len - LZMA_MATCH_LEN_MIN]; - UInt32 lenToPosState = GetLenToPosState(len); - if (distance < kNumFullDistances) - curAndLenPrice += p->distancesPrices[lenToPosState][distance]; else { - UInt32 slot; - GetPosSlot2(distance, slot); - curAndLenPrice += p->alignPrices[distance & kAlignMask] + p->posSlotPrices[lenToPosState][slot]; - } - opt = &p->opt[len]; - if (curAndLenPrice < opt->price) - { - opt->price = curAndLenPrice; - opt->posPrev = 0; - opt->backPrev = distance + LZMA_NUM_REPS; - opt->prev1IsChar = False; - } - if (len == matches[offs]) - { - offs += 2; - if (offs == numPairs) - break; + p->opt[wr].dist = 0; + len--; + wr--; + p->opt[wr].dist = MARK_LIT; + p->opt[wr].len = 1; } } - } - cur = 0; + if (cur == 0) + { + p->backRes = dist; + p->optCur = wr; + return len; + } + + wr--; + p->opt[wr].dist = dist; + p->opt[wr].len = (UInt32)len; + } +} + + + +#define LIT_PROBS(pos, prevByte) \ + (p->litProbs + (UInt32)3 * (((((pos) << 8) + (prevByte)) & p->lpMask) << p->lc)) + + +static unsigned GetOptimum(CLzmaEnc *p, UInt32 position) +{ + unsigned last, cur; + UInt32 reps[LZMA_NUM_REPS]; + unsigned repLens[LZMA_NUM_REPS]; + UInt32 *matches; + + { + UInt32 numAvail; + unsigned numPairs, mainLen, repMaxIndex, i, posState; + UInt32 matchPrice, repMatchPrice; + const Byte *data; + Byte curByte, matchByte; + + p->optCur = p->optEnd = 0; + + if (p->additionalOffset == 0) + mainLen = ReadMatchDistances(p, &numPairs); + else + { + mainLen = p->longestMatchLen; + numPairs = p->numPairs; + } + + numAvail = p->numAvail; + if (numAvail < 2) + { + p->backRes = MARK_LIT; + return 1; + } + if (numAvail > LZMA_MATCH_LEN_MAX) + numAvail = LZMA_MATCH_LEN_MAX; + + data = p->matchFinder.GetPointerToCurrentPos(p->matchFinderObj) - 1; + repMaxIndex = 0; + + for (i = 0; i < LZMA_NUM_REPS; i++) + { + unsigned len; + const Byte *data2; + reps[i] = p->reps[i]; + data2 = data - reps[i]; + if (data[0] != data2[0] || data[1] != data2[1]) + { + repLens[i] = 0; + continue; + } + for (len = 2; len < numAvail && data[len] == data2[len]; len++) + {} + repLens[i] = len; + if (len > repLens[repMaxIndex]) + repMaxIndex = i; + } + + if (repLens[repMaxIndex] >= p->numFastBytes) + { + unsigned len; + p->backRes = (UInt32)repMaxIndex; + len = repLens[repMaxIndex]; + MOVE_POS(p, len - 1) + return len; + } + + matches = p->matches; + + if (mainLen >= p->numFastBytes) + { + p->backRes = matches[(size_t)numPairs - 1] + LZMA_NUM_REPS; + MOVE_POS(p, mainLen - 1) + return mainLen; + } + + curByte = *data; + matchByte = *(data - reps[0]); + + last = repLens[repMaxIndex]; + if (last <= mainLen) + last = mainLen; + + if (last < 2 && curByte != matchByte) + { + p->backRes = MARK_LIT; + return 1; + } + + p->opt[0].state = (CState)p->state; + + posState = (position & p->pbMask); + + { + const CLzmaProb *probs = LIT_PROBS(position, *(data - 1)); + p->opt[1].price = GET_PRICE_0(p->isMatch[p->state][posState]) + + (!IsLitState(p->state) ? + LitEnc_Matched_GetPrice(probs, curByte, matchByte, p->ProbPrices) : + LitEnc_GetPrice(probs, curByte, p->ProbPrices)); + } + + MakeAs_Lit(&p->opt[1]); + + matchPrice = GET_PRICE_1(p->isMatch[p->state][posState]); + repMatchPrice = matchPrice + GET_PRICE_1(p->isRep[p->state]); + + // 18.06 + if (matchByte == curByte && repLens[0] == 0) + { + UInt32 shortRepPrice = repMatchPrice + GetPrice_ShortRep(p, p->state, posState); + if (shortRepPrice < p->opt[1].price) + { + p->opt[1].price = shortRepPrice; + MakeAs_ShortRep(&p->opt[1]); + } + if (last < 2) + { + p->backRes = p->opt[1].dist; + return 1; + } + } + + p->opt[1].len = 1; + + p->opt[0].reps[0] = reps[0]; + p->opt[0].reps[1] = reps[1]; + p->opt[0].reps[2] = reps[2]; + p->opt[0].reps[3] = reps[3]; + + // ---------- REP ---------- + + for (i = 0; i < LZMA_NUM_REPS; i++) + { + unsigned repLen = repLens[i]; + UInt32 price; + if (repLen < 2) + continue; + price = repMatchPrice + GetPrice_PureRep(p, i, p->state, posState); + do + { + UInt32 price2 = price + GET_PRICE_LEN(&p->repLenEnc, posState, repLen); + COptimal *opt = &p->opt[repLen]; + if (price2 < opt->price) + { + opt->price = price2; + opt->len = (UInt32)repLen; + opt->dist = (UInt32)i; + opt->extra = 0; + } + } + while (--repLen >= 2); + } + + + // ---------- MATCH ---------- + { + unsigned len = repLens[0] + 1; + if (len <= mainLen) + { + unsigned offs = 0; + UInt32 normalMatchPrice = matchPrice + GET_PRICE_0(p->isRep[p->state]); + + if (len < 2) + len = 2; + else + while (len > matches[offs]) + offs += 2; + + for (; ; len++) + { + COptimal *opt; + UInt32 dist = matches[(size_t)offs + 1]; + UInt32 price = normalMatchPrice + GET_PRICE_LEN(&p->lenEnc, posState, len); + unsigned lenToPosState = GetLenToPosState(len); + + if (dist < kNumFullDistances) + price += p->distancesPrices[lenToPosState][dist & (kNumFullDistances - 1)]; + else + { + unsigned slot; + GetPosSlot2(dist, slot); + price += p->alignPrices[dist & kAlignMask]; + price += p->posSlotPrices[lenToPosState][slot]; + } + + opt = &p->opt[len]; + + if (price < opt->price) + { + opt->price = price; + opt->len = (UInt32)len; + opt->dist = dist + LZMA_NUM_REPS; + opt->extra = 0; + } + + if (len == matches[offs]) + { + offs += 2; + if (offs == numPairs) + break; + } + } + } + } + + + cur = 0; #ifdef SHOW_STAT2 /* if (position >= 0) */ { unsigned i; printf("\n pos = %4X", position); - for (i = cur; i <= lenEnd; i++) + for (i = cur; i <= last; i++) printf("\nprice[%4X] = %u", position - cur + i, p->opt[i].price); } #endif + } + + + + // ---------- Optimal Parsing ---------- for (;;) { - UInt32 numAvailFull, newLen, numPairs, posPrev, state, posState, startLen; - UInt32 curPrice, curAnd1Price, matchPrice, repMatchPrice; - Bool nextIsChar; + unsigned numAvail; + UInt32 numAvailFull; + unsigned newLen, numPairs, prev, state, posState, startLen; + UInt32 litPrice, matchPrice, repMatchPrice; + BoolInt nextIsLit; Byte curByte, matchByte; const Byte *data; - COptimal *curOpt; - COptimal *nextOpt; + COptimal *curOpt, *nextOpt; - cur++; - if (cur == lenEnd) - return Backward(p, backRes, cur); + if (++cur == last) + break; + + // 18.06 + if (cur >= kNumOpts - 64) + { + unsigned j, best; + UInt32 price = p->opt[cur].price; + best = cur; + for (j = cur + 1; j <= last; j++) + { + UInt32 price2 = p->opt[j].price; + if (price >= price2) + { + price = price2; + best = j; + } + } + { + unsigned delta = best - cur; + if (delta != 0) + { + MOVE_POS(p, delta); + } + } + cur = best; + break; + } newLen = ReadMatchDistances(p, &numPairs); + if (newLen >= p->numFastBytes) { p->numPairs = numPairs; - p->longestMatchLength = newLen; - return Backward(p, backRes, cur); + p->longestMatchLen = newLen; + break; } - position++; + curOpt = &p->opt[cur]; - posPrev = curOpt->posPrev; - if (curOpt->prev1IsChar) - { - posPrev--; - if (curOpt->prev2) - { - state = p->opt[curOpt->posPrev2].state; - if (curOpt->backPrev2 < LZMA_NUM_REPS) - state = kRepNextStates[state]; - else - state = kMatchNextStates[state]; - } - else - state = p->opt[posPrev].state; - state = kLiteralNextStates[state]; - } - else - state = p->opt[posPrev].state; - if (posPrev == cur - 1) + + position++; + + // we need that check here, if skip_items in p->opt are possible + /* + if (curOpt->price >= kInfinityPrice) + continue; + */ + + prev = cur - curOpt->len; + + if (curOpt->len == 1) { + state = (unsigned)p->opt[prev].state; if (IsShortRep(curOpt)) state = kShortRepNextStates[state]; else @@ -1226,355 +1405,499 @@ static UInt32 GetOptimum(CLzmaEnc *p, UInt32 position, UInt32 *backRes) } else { - UInt32 pos; const COptimal *prevOpt; - if (curOpt->prev1IsChar && curOpt->prev2) + UInt32 b0; + UInt32 dist = curOpt->dist; + + if (curOpt->extra) { - posPrev = curOpt->posPrev2; - pos = curOpt->backPrev2; - state = kRepNextStates[state]; + prev -= (unsigned)curOpt->extra; + state = kState_RepAfterLit; + if (curOpt->extra == 1) + state = (dist < LZMA_NUM_REPS ? kState_RepAfterLit : kState_MatchAfterLit); } else { - pos = curOpt->backPrev; - if (pos < LZMA_NUM_REPS) + state = (unsigned)p->opt[prev].state; + if (dist < LZMA_NUM_REPS) state = kRepNextStates[state]; else state = kMatchNextStates[state]; } - prevOpt = &p->opt[posPrev]; - if (pos < LZMA_NUM_REPS) + + prevOpt = &p->opt[prev]; + b0 = prevOpt->reps[0]; + + if (dist < LZMA_NUM_REPS) { - UInt32 i; - reps[0] = prevOpt->backs[pos]; - for (i = 1; i <= pos; i++) - reps[i] = prevOpt->backs[i - 1]; - for (; i < LZMA_NUM_REPS; i++) - reps[i] = prevOpt->backs[i]; + if (dist == 0) + { + reps[0] = b0; + reps[1] = prevOpt->reps[1]; + reps[2] = prevOpt->reps[2]; + reps[3] = prevOpt->reps[3]; + } + else + { + reps[1] = b0; + b0 = prevOpt->reps[1]; + if (dist == 1) + { + reps[0] = b0; + reps[2] = prevOpt->reps[2]; + reps[3] = prevOpt->reps[3]; + } + else + { + reps[2] = b0; + reps[0] = prevOpt->reps[dist]; + reps[3] = prevOpt->reps[dist ^ 1]; + } + } } else { - UInt32 i; - reps[0] = (pos - LZMA_NUM_REPS); - for (i = 1; i < LZMA_NUM_REPS; i++) - reps[i] = prevOpt->backs[i - 1]; + reps[0] = (dist - LZMA_NUM_REPS + 1); + reps[1] = b0; + reps[2] = prevOpt->reps[1]; + reps[3] = prevOpt->reps[2]; } } + curOpt->state = (CState)state; + curOpt->reps[0] = reps[0]; + curOpt->reps[1] = reps[1]; + curOpt->reps[2] = reps[2]; + curOpt->reps[3] = reps[3]; - curOpt->backs[0] = reps[0]; - curOpt->backs[1] = reps[1]; - curOpt->backs[2] = reps[2]; - curOpt->backs[3] = reps[3]; - - curPrice = curOpt->price; - nextIsChar = False; data = p->matchFinder.GetPointerToCurrentPos(p->matchFinderObj) - 1; curByte = *data; - matchByte = *(data - (reps[0] + 1)); + matchByte = *(data - reps[0]); posState = (position & p->pbMask); - curAnd1Price = curPrice + GET_PRICE_0(p->isMatch[state][posState]); + /* + The order of Price checks: + < LIT + <= SHORT_REP + < LIT : REP_0 + < REP [ : LIT : REP_0 ] + < MATCH [ : LIT : REP_0 ] + */ + + { + UInt32 curPrice = curOpt->price; + unsigned prob = p->isMatch[state][posState]; + matchPrice = curPrice + GET_PRICE_1(prob); + litPrice = curPrice + GET_PRICE_0(prob); + } + + nextOpt = &p->opt[(size_t)cur + 1]; + nextIsLit = False; + + // here we can allow skip_items in p->opt, if we don't check (nextOpt->price < kInfinityPrice) + // 18.new.06 + if ((nextOpt->price < kInfinityPrice + // && !IsLitState(state) + && matchByte == curByte) + || litPrice > nextOpt->price + ) + litPrice = 0; + else { const CLzmaProb *probs = LIT_PROBS(position, *(data - 1)); - curAnd1Price += - (!IsCharState(state) ? - LitEnc_GetPriceMatched(probs, curByte, matchByte, p->ProbPrices) : + litPrice += (!IsLitState(state) ? + LitEnc_Matched_GetPrice(probs, curByte, matchByte, p->ProbPrices) : LitEnc_GetPrice(probs, curByte, p->ProbPrices)); - } - - nextOpt = &p->opt[cur + 1]; - - if (curAnd1Price < nextOpt->price) - { - nextOpt->price = curAnd1Price; - nextOpt->posPrev = cur; - MakeAsChar(nextOpt); - nextIsChar = True; - } - - matchPrice = curPrice + GET_PRICE_1(p->isMatch[state][posState]); - repMatchPrice = matchPrice + GET_PRICE_1(p->isRep[state]); - - if (matchByte == curByte && !(nextOpt->posPrev < cur && nextOpt->backPrev == 0)) - { - UInt32 shortRepPrice = repMatchPrice + GetRepLen1Price(p, state, posState); - if (shortRepPrice <= nextOpt->price) + + if (litPrice < nextOpt->price) { - nextOpt->price = shortRepPrice; - nextOpt->posPrev = cur; - MakeAsShortRep(nextOpt); - nextIsChar = True; + nextOpt->price = litPrice; + nextOpt->len = 1; + MakeAs_Lit(nextOpt); + nextIsLit = True; } } + + repMatchPrice = matchPrice + GET_PRICE_1(p->isRep[state]); + numAvailFull = p->numAvail; { - UInt32 temp = kNumOpts - 1 - cur; - if (temp < numAvailFull) - numAvailFull = temp; + unsigned temp = kNumOpts - 1 - cur; + if (numAvailFull > temp) + numAvailFull = (UInt32)temp; } + // 18.06 + // ---------- SHORT_REP ---------- + if (IsLitState(state)) // 18.new + if (matchByte == curByte) + if (repMatchPrice < nextOpt->price) // 18.new + // if (numAvailFull < 2 || data[1] != *(data - reps[0] + 1)) + if ( + // nextOpt->price >= kInfinityPrice || + nextOpt->len < 2 // we can check nextOpt->len, if skip items are not allowed in p->opt + || (nextOpt->dist != 0 + // && nextOpt->extra <= 1 // 17.old + ) + ) + { + UInt32 shortRepPrice = repMatchPrice + GetPrice_ShortRep(p, state, posState); + // if (shortRepPrice <= nextOpt->price) // 17.old + if (shortRepPrice < nextOpt->price) // 18.new + { + nextOpt->price = shortRepPrice; + nextOpt->len = 1; + MakeAs_ShortRep(nextOpt); + nextIsLit = False; + } + } + if (numAvailFull < 2) continue; numAvail = (numAvailFull <= p->numFastBytes ? numAvailFull : p->numFastBytes); - if (!nextIsChar && matchByte != curByte) /* speed optimization */ - { - /* try Literal + rep0 */ - UInt32 temp; - UInt32 lenTest2; - const Byte *data2 = data - reps[0] - 1; - UInt32 limit = p->numFastBytes + 1; - if (limit > numAvailFull) - limit = numAvailFull; + // numAvail <= p->numFastBytes - for (temp = 1; temp < limit && data[temp] == data2[temp]; temp++); - lenTest2 = temp - 1; - if (lenTest2 >= 2) + // ---------- LIT : REP_0 ---------- + + if (!nextIsLit + && litPrice != 0 // 18.new + && matchByte != curByte + && numAvailFull > 2) + { + const Byte *data2 = data - reps[0]; + if (data[1] == data2[1] && data[2] == data2[2]) { - UInt32 state2 = kLiteralNextStates[state]; - UInt32 posStateNext = (position + 1) & p->pbMask; - UInt32 nextRepMatchPrice = curAnd1Price + - GET_PRICE_1(p->isMatch[state2][posStateNext]) + - GET_PRICE_1(p->isRep[state2]); - /* for (; lenTest2 >= 2; lenTest2--) */ + unsigned len; + unsigned limit = p->numFastBytes + 1; + if (limit > numAvailFull) + limit = numAvailFull; + for (len = 3; len < limit && data[len] == data2[len]; len++) + {} + { - UInt32 curAndLenPrice; - COptimal *opt; - UInt32 offset = cur + 1 + lenTest2; - while (lenEnd < offset) - p->opt[++lenEnd].price = kInfinityPrice; - curAndLenPrice = nextRepMatchPrice + GetRepPrice(p, 0, lenTest2, state2, posStateNext); - opt = &p->opt[offset]; - if (curAndLenPrice < opt->price) + unsigned state2 = kLiteralNextStates[state]; + unsigned posState2 = (position + 1) & p->pbMask; + UInt32 price = litPrice + GetPrice_Rep_0(p, state2, posState2); { - opt->price = curAndLenPrice; - opt->posPrev = cur + 1; - opt->backPrev = 0; - opt->prev1IsChar = True; - opt->prev2 = False; + unsigned offset = cur + len; + + if (last < offset) + last = offset; + + // do + { + UInt32 price2; + COptimal *opt; + len--; + // price2 = price + GetPrice_Len_Rep_0(p, len, state2, posState2); + price2 = price + GET_PRICE_LEN(&p->repLenEnc, posState2, len); + + opt = &p->opt[offset]; + // offset--; + if (price2 < opt->price) + { + opt->price = price2; + opt->len = (UInt32)len; + opt->dist = 0; + opt->extra = 1; + } + } + // while (len >= 3); } } } } startLen = 2; /* speed optimization */ + { - UInt32 repIndex; - for (repIndex = 0; repIndex < LZMA_NUM_REPS; repIndex++) - { - UInt32 lenTest; - UInt32 lenTestTemp; - UInt32 price; - const Byte *data2 = data - reps[repIndex] - 1; - if (data[0] != data2[0] || data[1] != data2[1]) - continue; - for (lenTest = 2; lenTest < numAvail && data[lenTest] == data2[lenTest]; lenTest++); - while (lenEnd < cur + lenTest) - p->opt[++lenEnd].price = kInfinityPrice; - lenTestTemp = lenTest; - price = repMatchPrice + GetPureRepPrice(p, repIndex, state, posState); - do + // ---------- REP ---------- + unsigned repIndex = 0; // 17.old + // unsigned repIndex = IsLitState(state) ? 0 : 1; // 18.notused + for (; repIndex < LZMA_NUM_REPS; repIndex++) { - UInt32 curAndLenPrice = price + p->repLenEnc.prices[posState][lenTest - 2]; - COptimal *opt = &p->opt[cur + lenTest]; - if (curAndLenPrice < opt->price) - { - opt->price = curAndLenPrice; - opt->posPrev = cur; - opt->backPrev = repIndex; - opt->prev1IsChar = False; - } - } - while (--lenTest >= 2); - lenTest = lenTestTemp; - - if (repIndex == 0) - startLen = lenTest + 1; + unsigned len; + UInt32 price; + const Byte *data2 = data - reps[repIndex]; + if (data[0] != data2[0] || data[1] != data2[1]) + continue; - /* if (_maxMode) */ + for (len = 2; len < numAvail && data[len] == data2[len]; len++) + {} + + // if (len < startLen) continue; // 18.new: speed optimization + { - UInt32 lenTest2 = lenTest + 1; - UInt32 limit = lenTest2 + p->numFastBytes; - if (limit > numAvailFull) - limit = numAvailFull; - for (; lenTest2 < limit && data[lenTest2] == data2[lenTest2]; lenTest2++); - lenTest2 -= lenTest + 1; - if (lenTest2 >= 2) + unsigned offset = cur + len; + if (last < offset) + last = offset; + } + { + unsigned len2 = len; + price = repMatchPrice + GetPrice_PureRep(p, repIndex, state, posState); + do { - UInt32 nextRepMatchPrice; - UInt32 state2 = kRepNextStates[state]; - UInt32 posStateNext = (position + lenTest) & p->pbMask; - UInt32 curAndLenCharPrice = - price + p->repLenEnc.prices[posState][lenTest - 2] + - GET_PRICE_0(p->isMatch[state2][posStateNext]) + - LitEnc_GetPriceMatched(LIT_PROBS(position + lenTest, data[lenTest - 1]), - data[lenTest], data2[lenTest], p->ProbPrices); - state2 = kLiteralNextStates[state2]; - posStateNext = (position + lenTest + 1) & p->pbMask; - nextRepMatchPrice = curAndLenCharPrice + - GET_PRICE_1(p->isMatch[state2][posStateNext]) + - GET_PRICE_1(p->isRep[state2]); - - /* for (; lenTest2 >= 2; lenTest2--) */ + UInt32 price2 = price + GET_PRICE_LEN(&p->repLenEnc, posState, len2); + COptimal *opt = &p->opt[cur + len2]; + if (price2 < opt->price) { - UInt32 curAndLenPrice; - COptimal *opt; - UInt32 offset = cur + lenTest + 1 + lenTest2; - while (lenEnd < offset) - p->opt[++lenEnd].price = kInfinityPrice; - curAndLenPrice = nextRepMatchPrice + GetRepPrice(p, 0, lenTest2, state2, posStateNext); - opt = &p->opt[offset]; - if (curAndLenPrice < opt->price) - { - opt->price = curAndLenPrice; - opt->posPrev = cur + lenTest + 1; - opt->backPrev = 0; - opt->prev1IsChar = True; - opt->prev2 = True; - opt->posPrev2 = cur; - opt->backPrev2 = repIndex; - } + opt->price = price2; + opt->len = (UInt32)len2; + opt->dist = (UInt32)repIndex; + opt->extra = 0; } } + while (--len2 >= 2); } + + if (repIndex == 0) startLen = len + 1; // 17.old + // startLen = len + 1; // 18.new + + /* if (_maxMode) */ + { + // ---------- REP : LIT : REP_0 ---------- + // numFastBytes + 1 + numFastBytes + + unsigned len2 = len + 1; + unsigned limit = len2 + p->numFastBytes; + if (limit > numAvailFull) + limit = numAvailFull; + + len2 += 2; + if (len2 <= limit) + if (data[len2 - 2] == data2[len2 - 2]) + if (data[len2 - 1] == data2[len2 - 1]) + { + unsigned state2 = kRepNextStates[state]; + unsigned posState2 = (position + len) & p->pbMask; + price += GET_PRICE_LEN(&p->repLenEnc, posState, len) + + GET_PRICE_0(p->isMatch[state2][posState2]) + + LitEnc_Matched_GetPrice(LIT_PROBS(position + len, data[(size_t)len - 1]), + data[len], data2[len], p->ProbPrices); + + // state2 = kLiteralNextStates[state2]; + state2 = kState_LitAfterRep; + posState2 = (posState2 + 1) & p->pbMask; + + + price += GetPrice_Rep_0(p, state2, posState2); + + for (; len2 < limit && data[len2] == data2[len2]; len2++) + {} + + len2 -= len; + // if (len2 >= 3) + { + { + unsigned offset = cur + len + len2; + + if (last < offset) + last = offset; + // do + { + UInt32 price2; + COptimal *opt; + len2--; + // price2 = price + GetPrice_Len_Rep_0(p, len2, state2, posState2); + price2 = price + GET_PRICE_LEN(&p->repLenEnc, posState2, len2); + + opt = &p->opt[offset]; + // offset--; + if (price2 < opt->price) + { + opt->price = price2; + opt->len = (UInt32)len2; + opt->extra = (CExtra)(len + 1); + opt->dist = (UInt32)repIndex; + } + } + // while (len2 >= 3); + } + } + } + } + } } - } - /* for (UInt32 lenTest = 2; lenTest <= newLen; lenTest++) */ + + + // ---------- MATCH ---------- + /* for (unsigned len = 2; len <= newLen; len++) */ if (newLen > numAvail) { newLen = numAvail; for (numPairs = 0; newLen > matches[numPairs]; numPairs += 2); - matches[numPairs] = newLen; + matches[numPairs] = (UInt32)newLen; numPairs += 2; } + + // startLen = 2; /* speed optimization */ + if (newLen >= startLen) { UInt32 normalMatchPrice = matchPrice + GET_PRICE_0(p->isRep[state]); - UInt32 offs, curBack, posSlot; - UInt32 lenTest; - while (lenEnd < cur + newLen) - p->opt[++lenEnd].price = kInfinityPrice; + UInt32 dist; + unsigned offs, posSlot, len; + + { + unsigned offset = cur + newLen; + if (last < offset) + last = offset; + } offs = 0; while (startLen > matches[offs]) offs += 2; - curBack = matches[offs + 1]; - GetPosSlot2(curBack, posSlot); - for (lenTest = /*2*/ startLen; ; lenTest++) + dist = matches[(size_t)offs + 1]; + + // if (dist >= kNumFullDistances) + GetPosSlot2(dist, posSlot); + + for (len = /*2*/ startLen; ; len++) { - UInt32 curAndLenPrice = normalMatchPrice + p->lenEnc.prices[posState][lenTest - LZMA_MATCH_LEN_MIN]; - UInt32 lenToPosState = GetLenToPosState(lenTest); - COptimal *opt; - if (curBack < kNumFullDistances) - curAndLenPrice += p->distancesPrices[lenToPosState][curBack]; - else - curAndLenPrice += p->posSlotPrices[lenToPosState][posSlot] + p->alignPrices[curBack & kAlignMask]; - - opt = &p->opt[cur + lenTest]; - if (curAndLenPrice < opt->price) + UInt32 price = normalMatchPrice + GET_PRICE_LEN(&p->lenEnc, posState, len); { - opt->price = curAndLenPrice; - opt->posPrev = cur; - opt->backPrev = curBack + LZMA_NUM_REPS; - opt->prev1IsChar = False; + COptimal *opt; + unsigned lenNorm = len - 2; + lenNorm = GetLenToPosState2(lenNorm); + if (dist < kNumFullDistances) + price += p->distancesPrices[lenNorm][dist & (kNumFullDistances - 1)]; + else + price += p->posSlotPrices[lenNorm][posSlot] + p->alignPrices[dist & kAlignMask]; + + opt = &p->opt[cur + len]; + if (price < opt->price) + { + opt->price = price; + opt->len = (UInt32)len; + opt->dist = dist + LZMA_NUM_REPS; + opt->extra = 0; + } } - if (/*_maxMode && */lenTest == matches[offs]) + if (len == matches[offs]) { - /* Try Match + Literal + Rep0 */ - const Byte *data2 = data - curBack - 1; - UInt32 lenTest2 = lenTest + 1; - UInt32 limit = lenTest2 + p->numFastBytes; + // if (p->_maxMode) { + // MATCH : LIT : REP_0 + + const Byte *data2 = data - dist - 1; + unsigned len2 = len + 1; + unsigned limit = len2 + p->numFastBytes; if (limit > numAvailFull) limit = numAvailFull; - for (; lenTest2 < limit && data[lenTest2] == data2[lenTest2]; lenTest2++); - lenTest2 -= lenTest + 1; - if (lenTest2 >= 2) + + len2 += 2; + if (len2 <= limit) + if (data[len2 - 2] == data2[len2 - 2]) + if (data[len2 - 1] == data2[len2 - 1]) { - UInt32 nextRepMatchPrice; - UInt32 state2 = kMatchNextStates[state]; - UInt32 posStateNext = (position + lenTest) & p->pbMask; - UInt32 curAndLenCharPrice = curAndLenPrice + - GET_PRICE_0(p->isMatch[state2][posStateNext]) + - LitEnc_GetPriceMatched(LIT_PROBS(position + lenTest, data[lenTest - 1]), - data[lenTest], data2[lenTest], p->ProbPrices); - state2 = kLiteralNextStates[state2]; - posStateNext = (posStateNext + 1) & p->pbMask; - nextRepMatchPrice = curAndLenCharPrice + - GET_PRICE_1(p->isMatch[state2][posStateNext]) + - GET_PRICE_1(p->isRep[state2]); - - /* for (; lenTest2 >= 2; lenTest2--) */ + for (; len2 < limit && data[len2] == data2[len2]; len2++) + {} + + len2 -= len; + + // if (len2 >= 3) + { + unsigned state2 = kMatchNextStates[state]; + unsigned posState2 = (position + len) & p->pbMask; + unsigned offset; + price += GET_PRICE_0(p->isMatch[state2][posState2]); + price += LitEnc_Matched_GetPrice(LIT_PROBS(position + len, data[(size_t)len - 1]), + data[len], data2[len], p->ProbPrices); + + // state2 = kLiteralNextStates[state2]; + state2 = kState_LitAfterMatch; + + posState2 = (posState2 + 1) & p->pbMask; + price += GetPrice_Rep_0(p, state2, posState2); + + offset = cur + len + len2; + + if (last < offset) + last = offset; + // do { - UInt32 offset = cur + lenTest + 1 + lenTest2; - UInt32 curAndLenPrice; + UInt32 price2; COptimal *opt; - while (lenEnd < offset) - p->opt[++lenEnd].price = kInfinityPrice; - curAndLenPrice = nextRepMatchPrice + GetRepPrice(p, 0, lenTest2, state2, posStateNext); + len2--; + // price2 = price + GetPrice_Len_Rep_0(p, len2, state2, posState2); + price2 = price + GET_PRICE_LEN(&p->repLenEnc, posState2, len2); opt = &p->opt[offset]; - if (curAndLenPrice < opt->price) + // offset--; + if (price2 < opt->price) { - opt->price = curAndLenPrice; - opt->posPrev = cur + lenTest + 1; - opt->backPrev = 0; - opt->prev1IsChar = True; - opt->prev2 = True; - opt->posPrev2 = cur; - opt->backPrev2 = curBack + LZMA_NUM_REPS; + opt->price = price2; + opt->len = (UInt32)len2; + opt->extra = (CExtra)(len + 1); + opt->dist = dist + LZMA_NUM_REPS; } } + // while (len2 >= 3); } + + } + offs += 2; if (offs == numPairs) break; - curBack = matches[offs + 1]; - if (curBack >= kNumFullDistances) - GetPosSlot2(curBack, posSlot); + dist = matches[(size_t)offs + 1]; + // if (dist >= kNumFullDistances) + GetPosSlot2(dist, posSlot); } } } } + + do + p->opt[last].price = kInfinityPrice; + while (--last); + + return Backward(p, cur); } + + #define ChangePair(smallDist, bigDist) (((bigDist) >> 7) > (smallDist)) -static UInt32 GetOptimumFast(CLzmaEnc *p, UInt32 *backRes) + + +static unsigned GetOptimumFast(CLzmaEnc *p) { - UInt32 numAvail, mainLen, mainDist, numPairs, repIndex, repLen, i; + UInt32 numAvail, mainDist; + unsigned mainLen, numPairs, repIndex, repLen, i; const Byte *data; - const UInt32 *matches; if (p->additionalOffset == 0) mainLen = ReadMatchDistances(p, &numPairs); else { - mainLen = p->longestMatchLength; + mainLen = p->longestMatchLen; numPairs = p->numPairs; } numAvail = p->numAvail; - *backRes = (UInt32)-1; + p->backRes = MARK_LIT; if (numAvail < 2) return 1; + // if (mainLen < 2 && p->state == 0) return 1; // 18.06.notused if (numAvail > LZMA_MATCH_LEN_MAX) numAvail = LZMA_MATCH_LEN_MAX; data = p->matchFinder.GetPointerToCurrentPos(p->matchFinderObj) - 1; - repLen = repIndex = 0; + for (i = 0; i < LZMA_NUM_REPS; i++) { - UInt32 len; - const Byte *data2 = data - p->reps[i] - 1; + unsigned len; + const Byte *data2 = data - p->reps[i]; if (data[0] != data2[0] || data[1] != data2[1]) continue; - for (len = 2; len < numAvail && data[len] == data2[len]; len++); + for (len = 2; len < numAvail && data[len] == data2[len]; len++) + {} if (len >= p->numFastBytes) { - *backRes = i; - MovePos(p, len - 1); + p->backRes = (UInt32)i; + MOVE_POS(p, len - 1) return len; } if (len > repLen) @@ -1584,84 +1907,152 @@ static UInt32 GetOptimumFast(CLzmaEnc *p, UInt32 *backRes) } } - matches = p->matches; if (mainLen >= p->numFastBytes) { - *backRes = matches[numPairs - 1] + LZMA_NUM_REPS; - MovePos(p, mainLen - 1); + p->backRes = p->matches[(size_t)numPairs - 1] + LZMA_NUM_REPS; + MOVE_POS(p, mainLen - 1) return mainLen; } mainDist = 0; /* for GCC */ + if (mainLen >= 2) { - mainDist = matches[numPairs - 1]; - while (numPairs > 2 && mainLen == matches[numPairs - 4] + 1) + mainDist = p->matches[(size_t)numPairs - 1]; + while (numPairs > 2) { - if (!ChangePair(matches[numPairs - 3], mainDist)) + UInt32 dist2; + if (mainLen != p->matches[(size_t)numPairs - 4] + 1) + break; + dist2 = p->matches[(size_t)numPairs - 3]; + if (!ChangePair(dist2, mainDist)) break; numPairs -= 2; - mainLen = matches[numPairs - 2]; - mainDist = matches[numPairs - 1]; + mainLen--; + mainDist = dist2; } if (mainLen == 2 && mainDist >= 0x80) mainLen = 1; } - if (repLen >= 2 && ( - (repLen + 1 >= mainLen) || - (repLen + 2 >= mainLen && mainDist >= (1 << 9)) || - (repLen + 3 >= mainLen && mainDist >= (1 << 15)))) + if (repLen >= 2) + if ( repLen + 1 >= mainLen + || (repLen + 2 >= mainLen && mainDist >= (1 << 9)) + || (repLen + 3 >= mainLen && mainDist >= (1 << 15))) { - *backRes = repIndex; - MovePos(p, repLen - 1); + p->backRes = (UInt32)repIndex; + MOVE_POS(p, repLen - 1) return repLen; } if (mainLen < 2 || numAvail <= 2) return 1; - p->longestMatchLength = ReadMatchDistances(p, &p->numPairs); - if (p->longestMatchLength >= 2) { - UInt32 newDistance = matches[p->numPairs - 1]; - if ((p->longestMatchLength >= mainLen && newDistance < mainDist) || - (p->longestMatchLength == mainLen + 1 && !ChangePair(mainDist, newDistance)) || - (p->longestMatchLength > mainLen + 1) || - (p->longestMatchLength + 1 >= mainLen && mainLen >= 3 && ChangePair(newDistance, mainDist))) - return 1; + unsigned len1 = ReadMatchDistances(p, &p->numPairs); + p->longestMatchLen = len1; + + if (len1 >= 2) + { + UInt32 newDist = p->matches[(size_t)p->numPairs - 1]; + if ( (len1 >= mainLen && newDist < mainDist) + || (len1 == mainLen + 1 && !ChangePair(mainDist, newDist)) + || (len1 > mainLen + 1) + || (len1 + 1 >= mainLen && mainLen >= 3 && ChangePair(newDist, mainDist))) + return 1; + } } data = p->matchFinder.GetPointerToCurrentPos(p->matchFinderObj) - 1; + for (i = 0; i < LZMA_NUM_REPS; i++) { - UInt32 len, limit; - const Byte *data2 = data - p->reps[i] - 1; + unsigned len, limit; + const Byte *data2 = data - p->reps[i]; if (data[0] != data2[0] || data[1] != data2[1]) continue; limit = mainLen - 1; - for (len = 2; len < limit && data[len] == data2[len]; len++); - if (len >= limit) - return 1; + for (len = 2;; len++) + { + if (len >= limit) + return 1; + if (data[len] != data2[len]) + break; + } + } + + p->backRes = mainDist + LZMA_NUM_REPS; + if (mainLen != 2) + { + MOVE_POS(p, mainLen - 2) } - *backRes = mainDist + LZMA_NUM_REPS; - MovePos(p, mainLen - 2); return mainLen; } -static void WriteEndMarker(CLzmaEnc *p, UInt32 posState) + + + +static void WriteEndMarker(CLzmaEnc *p, unsigned posState) { - UInt32 len; - RangeEnc_EncodeBit(&p->rc, &p->isMatch[p->state][posState], 1); - RangeEnc_EncodeBit(&p->rc, &p->isRep[p->state], 0); + UInt32 range; + range = p->rc.range; + { + UInt32 ttt, newBound; + CLzmaProb *prob = &p->isMatch[p->state][posState]; + RC_BIT_PRE(&p->rc, prob) + RC_BIT_1(&p->rc, prob) + prob = &p->isRep[p->state]; + RC_BIT_PRE(&p->rc, prob) + RC_BIT_0(&p->rc, prob) + } p->state = kMatchNextStates[p->state]; - len = LZMA_MATCH_LEN_MIN; - LenEnc_Encode2(&p->lenEnc, &p->rc, len - LZMA_MATCH_LEN_MIN, posState, !p->fastMode, p->ProbPrices); - RcTree_Encode(&p->rc, p->posSlotEncoder[GetLenToPosState(len)], kNumPosSlotBits, (1 << kNumPosSlotBits) - 1); - RangeEnc_EncodeDirectBits(&p->rc, (((UInt32)1 << 30) - 1) >> kNumAlignBits, 30 - kNumAlignBits); - RcTree_ReverseEncode(&p->rc, p->posAlignEncoder, kNumAlignBits, kAlignMask); + + p->rc.range = range; + LenEnc_Encode(&p->lenProbs, &p->rc, 0, posState); + range = p->rc.range; + + { + // RcTree_Encode_PosSlot(&p->rc, p->posSlotEncoder[0], (1 << kNumPosSlotBits) - 1); + CLzmaProb *probs = p->posSlotEncoder[0]; + unsigned m = 1; + do + { + UInt32 ttt, newBound; + RC_BIT_PRE(p, probs + m) + RC_BIT_1(&p->rc, probs + m); + m = (m << 1) + 1; + } + while (m < (1 << kNumPosSlotBits)); + } + { + // RangeEnc_EncodeDirectBits(&p->rc, ((UInt32)1 << (30 - kNumAlignBits)) - 1, 30 - kNumAlignBits); UInt32 range = p->range; + unsigned numBits = 30 - kNumAlignBits; + do + { + range >>= 1; + p->rc.low += range; + RC_NORM(&p->rc) + } + while (--numBits); + } + + { + // RcTree_ReverseEncode(&p->rc, p->posAlignEncoder, kNumAlignBits, kAlignMask); + CLzmaProb *probs = p->posAlignEncoder; + unsigned m = 1; + do + { + UInt32 ttt, newBound; + RC_BIT_PRE(p, probs + m) + RC_BIT_1(&p->rc, probs + m); + m = (m << 1) + 1; + } + while (m < kAlignTableSize); + } + p->rc.range = range; } + static SRes CheckErrors(CLzmaEnc *p) { if (p->result != SZ_OK) @@ -1675,7 +2066,8 @@ static SRes CheckErrors(CLzmaEnc *p) return p->result; } -static SRes Flush(CLzmaEnc *p, UInt32 nowPos) + +MY_NO_INLINE static SRes Flush(CLzmaEnc *p, UInt32 nowPos) { /* ReleaseMFStream(); */ p->finished = True; @@ -1686,48 +2078,126 @@ static SRes Flush(CLzmaEnc *p, UInt32 nowPos) return CheckErrors(p); } -static void FillAlignPrices(CLzmaEnc *p) + +MY_NO_INLINE static void FillAlignPrices(CLzmaEnc *p) { - UInt32 i; - for (i = 0; i < kAlignTableSize; i++) - p->alignPrices[i] = RcTree_ReverseGetPrice(p->posAlignEncoder, kNumAlignBits, i, p->ProbPrices); - p->alignPriceCount = 0; + unsigned i; + const CProbPrice *ProbPrices = p->ProbPrices; + const CLzmaProb *probs = p->posAlignEncoder; + // p->alignPriceCount = 0; + for (i = 0; i < kAlignTableSize / 2; i++) + { + UInt32 price = 0; + unsigned sym = i; + unsigned m = 1; + unsigned bit; + UInt32 prob; + bit = sym & 1; sym >>= 1; price += GET_PRICEa(probs[m], bit); m = (m << 1) + bit; + bit = sym & 1; sym >>= 1; price += GET_PRICEa(probs[m], bit); m = (m << 1) + bit; + bit = sym & 1; sym >>= 1; price += GET_PRICEa(probs[m], bit); m = (m << 1) + bit; + prob = probs[m]; + p->alignPrices[i ] = price + GET_PRICEa_0(prob); + p->alignPrices[i + 8] = price + GET_PRICEa_1(prob); + // p->alignPrices[i] = RcTree_ReverseGetPrice(p->posAlignEncoder, kNumAlignBits, i, p->ProbPrices); + } } -static void FillDistancesPrices(CLzmaEnc *p) -{ - UInt32 tempPrices[kNumFullDistances]; - UInt32 i, lenToPosState; - for (i = kStartPosModelIndex; i < kNumFullDistances; i++) - { - UInt32 posSlot = GetPosSlot1(i); - UInt32 footerBits = ((posSlot >> 1) - 1); - UInt32 base = ((2 | (posSlot & 1)) << footerBits); - tempPrices[i] = RcTree_ReverseGetPrice(p->posEncoders + base - posSlot - 1, footerBits, i - base, p->ProbPrices); - } - for (lenToPosState = 0; lenToPosState < kNumLenToPosStates; lenToPosState++) +MY_NO_INLINE static void FillDistancesPrices(CLzmaEnc *p) +{ + // int y; for (y = 0; y < 100; y++) { + + UInt32 tempPrices[kNumFullDistances]; + unsigned i, lps; + + const CProbPrice *ProbPrices = p->ProbPrices; + p->matchPriceCount = 0; + + for (i = kStartPosModelIndex / 2; i < kNumFullDistances / 2; i++) { - UInt32 posSlot; - const CLzmaProb *encoder = p->posSlotEncoder[lenToPosState]; - UInt32 *posSlotPrices = p->posSlotPrices[lenToPosState]; - for (posSlot = 0; posSlot < p->distTableSize; posSlot++) - posSlotPrices[posSlot] = RcTree_GetPrice(encoder, kNumPosSlotBits, posSlot, p->ProbPrices); - for (posSlot = kEndPosModelIndex; posSlot < p->distTableSize; posSlot++) - posSlotPrices[posSlot] += ((((posSlot >> 1) - 1) - kNumAlignBits) << kNumBitPriceShiftBits); + unsigned posSlot = GetPosSlot1(i); + unsigned footerBits = (posSlot >> 1) - 1; + unsigned base = ((2 | (posSlot & 1)) << footerBits); + const CLzmaProb *probs = p->posEncoders + (size_t)base * 2; + // tempPrices[i] = RcTree_ReverseGetPrice(p->posEncoders + base, footerBits, i - base, p->ProbPrices); + UInt32 price = 0; + unsigned m = 1; + unsigned sym = i; + unsigned offset = (unsigned)1 << footerBits; + base += i; + + if (footerBits) + do + { + unsigned bit = sym & 1; + sym >>= 1; + price += GET_PRICEa(probs[m], bit); + m = (m << 1) + bit; + } + while (--footerBits); { - UInt32 *distancesPrices = p->distancesPrices[lenToPosState]; - UInt32 i; - for (i = 0; i < kStartPosModelIndex; i++) - distancesPrices[i] = posSlotPrices[i]; - for (; i < kNumFullDistances; i++) - distancesPrices[i] = posSlotPrices[GetPosSlot1(i)] + tempPrices[i]; + unsigned prob = probs[m]; + tempPrices[base ] = price + GET_PRICEa_0(prob); + tempPrices[base + offset] = price + GET_PRICEa_1(prob); } } - p->matchPriceCount = 0; + + for (lps = 0; lps < kNumLenToPosStates; lps++) + { + unsigned slot; + unsigned distTableSize2 = (p->distTableSize + 1) >> 1; + UInt32 *posSlotPrices = p->posSlotPrices[lps]; + const CLzmaProb *probs = p->posSlotEncoder[lps]; + + for (slot = 0; slot < distTableSize2; slot++) + { + // posSlotPrices[slot] = RcTree_GetPrice(encoder, kNumPosSlotBits, slot, p->ProbPrices); + UInt32 price; + unsigned bit; + unsigned sym = slot + (1 << (kNumPosSlotBits - 1)); + unsigned prob; + bit = sym & 1; sym >>= 1; price = GET_PRICEa(probs[sym], bit); + bit = sym & 1; sym >>= 1; price += GET_PRICEa(probs[sym], bit); + bit = sym & 1; sym >>= 1; price += GET_PRICEa(probs[sym], bit); + bit = sym & 1; sym >>= 1; price += GET_PRICEa(probs[sym], bit); + bit = sym & 1; sym >>= 1; price += GET_PRICEa(probs[sym], bit); + prob = probs[(size_t)slot + (1 << (kNumPosSlotBits - 1))]; + posSlotPrices[(size_t)slot * 2 ] = price + GET_PRICEa_0(prob); + posSlotPrices[(size_t)slot * 2 + 1] = price + GET_PRICEa_1(prob); + } + + { + UInt32 delta = ((UInt32)((kEndPosModelIndex / 2 - 1) - kNumAlignBits) << kNumBitPriceShiftBits); + for (slot = kEndPosModelIndex / 2; slot < distTableSize2; slot++) + { + posSlotPrices[(size_t)slot * 2 ] += delta; + posSlotPrices[(size_t)slot * 2 + 1] += delta; + delta += ((UInt32)1 << kNumBitPriceShiftBits); + } + } + + { + UInt32 *dp = p->distancesPrices[lps]; + + dp[0] = posSlotPrices[0]; + dp[1] = posSlotPrices[1]; + dp[2] = posSlotPrices[2]; + dp[3] = posSlotPrices[3]; + + for (i = 4; i < kNumFullDistances; i += 2) + { + UInt32 slotPrice = posSlotPrices[GetPosSlot1(i)]; + dp[i ] = slotPrice + tempPrices[i]; + dp[i + 1] = slotPrice + tempPrices[i + 1]; + } + } + } + // } } + + void LzmaEnc_Construct(CLzmaEnc *p) { RangeEnc_Construct(&p->rc); @@ -1751,26 +2221,27 @@ void LzmaEnc_Construct(CLzmaEnc *p) LzmaEnc_InitPriceTables(p->ProbPrices); p->litProbs = NULL; p->saveState.litProbs = NULL; + } -CLzmaEncHandle LzmaEnc_Create(ISzAlloc *alloc) +CLzmaEncHandle LzmaEnc_Create(ISzAllocPtr alloc) { void *p; - p = alloc->Alloc(alloc, sizeof(CLzmaEnc)); + p = ISzAlloc_Alloc(alloc, sizeof(CLzmaEnc)); if (p) LzmaEnc_Construct((CLzmaEnc *)p); return p; } -void LzmaEnc_FreeLits(CLzmaEnc *p, ISzAlloc *alloc) +void LzmaEnc_FreeLits(CLzmaEnc *p, ISzAllocPtr alloc) { - alloc->Free(alloc, p->litProbs); - alloc->Free(alloc, p->saveState.litProbs); + ISzAlloc_Free(alloc, p->litProbs); + ISzAlloc_Free(alloc, p->saveState.litProbs); p->litProbs = NULL; p->saveState.litProbs = NULL; } -void LzmaEnc_Destruct(CLzmaEnc *p, ISzAlloc *alloc, ISzAlloc *allocBig) +void LzmaEnc_Destruct(CLzmaEnc *p, ISzAllocPtr alloc, ISzAllocPtr allocBig) { #ifndef _7ZIP_ST MatchFinderMt_Destruct(&p->matchFinderMt, allocBig); @@ -1781,13 +2252,14 @@ void LzmaEnc_Destruct(CLzmaEnc *p, ISzAlloc *alloc, ISzAlloc *allocBig) RangeEnc_Free(&p->rc, alloc); } -void LzmaEnc_Destroy(CLzmaEncHandle p, ISzAlloc *alloc, ISzAlloc *allocBig) +void LzmaEnc_Destroy(CLzmaEncHandle p, ISzAllocPtr alloc, ISzAllocPtr allocBig) { LzmaEnc_Destruct((CLzmaEnc *)p, alloc, allocBig); - alloc->Free(alloc, p); + ISzAlloc_Free(alloc, p); } -static SRes LzmaEnc_CodeOneBlock(CLzmaEnc *p, Bool useLimits, UInt32 maxPackSize, UInt32 maxUnpackSize) + +static SRes LzmaEnc_CodeOneBlock(CLzmaEnc *p, UInt32 maxPackSize, UInt32 maxUnpackSize) { UInt32 nowPos32, startPos32; if (p->needInit) @@ -1805,13 +2277,13 @@ static SRes LzmaEnc_CodeOneBlock(CLzmaEnc *p, Bool useLimits, UInt32 maxPackSize if (p->nowPos64 == 0) { - UInt32 numPairs; + unsigned numPairs; Byte curByte; if (p->matchFinder.GetNumAvailableBytes(p->matchFinderObj) == 0) return Flush(p, nowPos32); ReadMatchDistances(p, &numPairs); - RangeEnc_EncodeBit(&p->rc, &p->isMatch[p->state][0], 0); - p->state = kLiteralNextStates[p->state]; + RangeEnc_EncodeBit_0(&p->rc, &p->isMatch[kState_Start][0]); + // p->state = kLiteralNextStates[p->state]; curByte = *(p->matchFinder.GetPointerToCurrentPos(p->matchFinderObj) - p->additionalOffset); LitEnc_Encode(&p->rc, p->litProbs, curByte); p->additionalOffset--; @@ -1819,123 +2291,253 @@ static SRes LzmaEnc_CodeOneBlock(CLzmaEnc *p, Bool useLimits, UInt32 maxPackSize } if (p->matchFinder.GetNumAvailableBytes(p->matchFinderObj) != 0) + for (;;) { - UInt32 pos, len, posState; - + UInt32 dist; + unsigned len, posState; + UInt32 range, ttt, newBound; + CLzmaProb *probs; + if (p->fastMode) - len = GetOptimumFast(p, &pos); + len = GetOptimumFast(p); else - len = GetOptimum(p, nowPos32, &pos); + { + unsigned oci = p->optCur; + if (p->optEnd == oci) + len = GetOptimum(p, nowPos32); + else + { + const COptimal *opt = &p->opt[oci]; + len = opt->len; + p->backRes = opt->dist; + p->optCur = oci + 1; + } + } + + posState = (unsigned)nowPos32 & p->pbMask; + range = p->rc.range; + probs = &p->isMatch[p->state][posState]; + + RC_BIT_PRE(&p->rc, probs) + + dist = p->backRes; #ifdef SHOW_STAT2 - printf("\n pos = %4X, len = %u pos = %u", nowPos32, len, pos); + printf("\n pos = %6X, len = %3u pos = %6u", nowPos32, len, dist); #endif - posState = nowPos32 & p->pbMask; - if (len == 1 && pos == (UInt32)-1) + if (dist == MARK_LIT) { Byte curByte; - CLzmaProb *probs; const Byte *data; + unsigned state; - RangeEnc_EncodeBit(&p->rc, &p->isMatch[p->state][posState], 0); + RC_BIT_0(&p->rc, probs); + p->rc.range = range; data = p->matchFinder.GetPointerToCurrentPos(p->matchFinderObj) - p->additionalOffset; - curByte = *data; probs = LIT_PROBS(nowPos32, *(data - 1)); - if (IsCharState(p->state)) + curByte = *data; + state = p->state; + p->state = kLiteralNextStates[state]; + if (IsLitState(state)) LitEnc_Encode(&p->rc, probs, curByte); else - LitEnc_EncodeMatched(&p->rc, probs, curByte, *(data - p->reps[0] - 1)); - p->state = kLiteralNextStates[p->state]; + LitEnc_EncodeMatched(&p->rc, probs, curByte, *(data - p->reps[0])); } else { - RangeEnc_EncodeBit(&p->rc, &p->isMatch[p->state][posState], 1); - if (pos < LZMA_NUM_REPS) + RC_BIT_1(&p->rc, probs); + probs = &p->isRep[p->state]; + RC_BIT_PRE(&p->rc, probs) + + if (dist < LZMA_NUM_REPS) { - RangeEnc_EncodeBit(&p->rc, &p->isRep[p->state], 1); - if (pos == 0) + RC_BIT_1(&p->rc, probs); + probs = &p->isRepG0[p->state]; + RC_BIT_PRE(&p->rc, probs) + if (dist == 0) { - RangeEnc_EncodeBit(&p->rc, &p->isRepG0[p->state], 0); - RangeEnc_EncodeBit(&p->rc, &p->isRep0Long[p->state][posState], ((len == 1) ? 0 : 1)); + RC_BIT_0(&p->rc, probs); + probs = &p->isRep0Long[p->state][posState]; + RC_BIT_PRE(&p->rc, probs) + if (len != 1) + { + RC_BIT_1_BASE(&p->rc, probs); + } + else + { + RC_BIT_0_BASE(&p->rc, probs); + p->state = kShortRepNextStates[p->state]; + } } else { - UInt32 distance = p->reps[pos]; - RangeEnc_EncodeBit(&p->rc, &p->isRepG0[p->state], 1); - if (pos == 1) - RangeEnc_EncodeBit(&p->rc, &p->isRepG1[p->state], 0); + RC_BIT_1(&p->rc, probs); + probs = &p->isRepG1[p->state]; + RC_BIT_PRE(&p->rc, probs) + if (dist == 1) + { + RC_BIT_0_BASE(&p->rc, probs); + dist = p->reps[1]; + } else { - RangeEnc_EncodeBit(&p->rc, &p->isRepG1[p->state], 1); - RangeEnc_EncodeBit(&p->rc, &p->isRepG2[p->state], pos - 2); - if (pos == 3) + RC_BIT_1(&p->rc, probs); + probs = &p->isRepG2[p->state]; + RC_BIT_PRE(&p->rc, probs) + if (dist == 2) + { + RC_BIT_0_BASE(&p->rc, probs); + dist = p->reps[2]; + } + else + { + RC_BIT_1_BASE(&p->rc, probs); + dist = p->reps[3]; p->reps[3] = p->reps[2]; + } p->reps[2] = p->reps[1]; } p->reps[1] = p->reps[0]; - p->reps[0] = distance; + p->reps[0] = dist; } - if (len == 1) - p->state = kShortRepNextStates[p->state]; - else + + RC_NORM(&p->rc) + + p->rc.range = range; + + if (len != 1) { - LenEnc_Encode2(&p->repLenEnc, &p->rc, len - LZMA_MATCH_LEN_MIN, posState, !p->fastMode, p->ProbPrices); + LenEnc_Encode(&p->repLenProbs, &p->rc, len - LZMA_MATCH_LEN_MIN, posState); + --p->repLenEncCounter; p->state = kRepNextStates[p->state]; } } else { - UInt32 posSlot; - RangeEnc_EncodeBit(&p->rc, &p->isRep[p->state], 0); + unsigned posSlot; + RC_BIT_0(&p->rc, probs); + p->rc.range = range; p->state = kMatchNextStates[p->state]; - LenEnc_Encode2(&p->lenEnc, &p->rc, len - LZMA_MATCH_LEN_MIN, posState, !p->fastMode, p->ProbPrices); - pos -= LZMA_NUM_REPS; - GetPosSlot(pos, posSlot); - RcTree_Encode(&p->rc, p->posSlotEncoder[GetLenToPosState(len)], kNumPosSlotBits, posSlot); - - if (posSlot >= kStartPosModelIndex) - { - UInt32 footerBits = ((posSlot >> 1) - 1); - UInt32 base = ((2 | (posSlot & 1)) << footerBits); - UInt32 posReduced = pos - base; - if (posSlot < kEndPosModelIndex) - RcTree_ReverseEncode(&p->rc, p->posEncoders + base - posSlot - 1, footerBits, posReduced); - else - { - RangeEnc_EncodeDirectBits(&p->rc, posReduced >> kNumAlignBits, footerBits - kNumAlignBits); - RcTree_ReverseEncode(&p->rc, p->posAlignEncoder, kNumAlignBits, posReduced & kAlignMask); - p->alignPriceCount++; - } - } + LenEnc_Encode(&p->lenProbs, &p->rc, len - LZMA_MATCH_LEN_MIN, posState); + // --p->lenEnc.counter; + + dist -= LZMA_NUM_REPS; p->reps[3] = p->reps[2]; p->reps[2] = p->reps[1]; p->reps[1] = p->reps[0]; - p->reps[0] = pos; + p->reps[0] = dist + 1; + p->matchPriceCount++; + GetPosSlot(dist, posSlot); + // RcTree_Encode_PosSlot(&p->rc, p->posSlotEncoder[GetLenToPosState(len)], posSlot); + { + UInt32 sym = (UInt32)posSlot + (1 << kNumPosSlotBits); + range = p->rc.range; + probs = p->posSlotEncoder[GetLenToPosState(len)]; + do + { + CLzmaProb *prob = probs + (sym >> kNumPosSlotBits); + UInt32 bit = (sym >> (kNumPosSlotBits - 1)) & 1; + sym <<= 1; + RC_BIT(&p->rc, prob, bit); + } + while (sym < (1 << kNumPosSlotBits * 2)); + p->rc.range = range; + } + + if (dist >= kStartPosModelIndex) + { + unsigned footerBits = ((posSlot >> 1) - 1); + + if (dist < kNumFullDistances) + { + unsigned base = ((2 | (posSlot & 1)) << footerBits); + RcTree_ReverseEncode(&p->rc, p->posEncoders + base, footerBits, (unsigned)(dist /* - base */)); + } + else + { + UInt32 pos2 = (dist | 0xF) << (32 - footerBits); + range = p->rc.range; + // RangeEnc_EncodeDirectBits(&p->rc, posReduced >> kNumAlignBits, footerBits - kNumAlignBits); + /* + do + { + range >>= 1; + p->rc.low += range & (0 - ((dist >> --footerBits) & 1)); + RC_NORM(&p->rc) + } + while (footerBits > kNumAlignBits); + */ + do + { + range >>= 1; + p->rc.low += range & (0 - (pos2 >> 31)); + pos2 += pos2; + RC_NORM(&p->rc) + } + while (pos2 != 0xF0000000); + + + // RcTree_ReverseEncode(&p->rc, p->posAlignEncoder, kNumAlignBits, posReduced & kAlignMask); + + { + unsigned m = 1; + unsigned bit; + bit = dist & 1; dist >>= 1; RC_BIT(&p->rc, p->posAlignEncoder + m, bit); m = (m << 1) + bit; + bit = dist & 1; dist >>= 1; RC_BIT(&p->rc, p->posAlignEncoder + m, bit); m = (m << 1) + bit; + bit = dist & 1; dist >>= 1; RC_BIT(&p->rc, p->posAlignEncoder + m, bit); m = (m << 1) + bit; + bit = dist & 1; RC_BIT(&p->rc, p->posAlignEncoder + m, bit); + p->rc.range = range; + // p->alignPriceCount++; + } + } + } } } + + nowPos32 += (UInt32)len; p->additionalOffset -= len; - nowPos32 += len; + if (p->additionalOffset == 0) { UInt32 processed; + if (!p->fastMode) { - if (p->matchPriceCount >= (1 << 7)) - FillDistancesPrices(p); - if (p->alignPriceCount >= kAlignTableSize) + /* + if (p->alignPriceCount >= 16) // kAlignTableSize FillAlignPrices(p); + if (p->matchPriceCount >= 128) + FillDistancesPrices(p); + if (p->lenEnc.counter <= 0) + LenPriceEnc_UpdateTables(&p->lenEnc, 1 << p->pb, &p->lenProbs, p->ProbPrices); + */ + if (p->matchPriceCount >= 64) + { + FillAlignPrices(p); + // { int y; for (y = 0; y < 100; y++) { + FillDistancesPrices(p); + // }} + LenPriceEnc_UpdateTables(&p->lenEnc, 1 << p->pb, &p->lenProbs, p->ProbPrices); + } + if (p->repLenEncCounter <= 0) + { + p->repLenEncCounter = REP_LEN_COUNT; + LenPriceEnc_UpdateTables(&p->repLenEnc, 1 << p->pb, &p->repLenProbs, p->ProbPrices); + } } + if (p->matchFinder.GetNumAvailableBytes(p->matchFinderObj) == 0) break; processed = nowPos32 - startPos32; - if (useLimits) + + if (maxPackSize) { - if (processed + kNumOpts + 300 >= maxUnpackSize || - RangeEnc_GetProcessed(&p->rc) + kNumOpts * 2 >= maxPackSize) + if (processed + kNumOpts + 300 >= maxUnpackSize + || RangeEnc_GetProcessed_sizet(&p->rc) + kPackReserve >= maxPackSize) break; } else if (processed >= (1 << 17)) @@ -1945,13 +2547,16 @@ static SRes LzmaEnc_CodeOneBlock(CLzmaEnc *p, Bool useLimits, UInt32 maxPackSize } } } + p->nowPos64 += nowPos32 - startPos32; return Flush(p, nowPos32); } + + #define kBigHashDicLimit ((UInt32)1 << 24) -static SRes LzmaEnc_Alloc(CLzmaEnc *p, UInt32 keepWindowSize, ISzAlloc *alloc, ISzAlloc *allocBig) +static SRes LzmaEnc_Alloc(CLzmaEnc *p, UInt32 keepWindowSize, ISzAllocPtr alloc, ISzAllocPtr allocBig) { UInt32 beforeSize = kNumOpts; if (!RangeEnc_Alloc(&p->rc, alloc)) @@ -1966,8 +2571,8 @@ static SRes LzmaEnc_Alloc(CLzmaEnc *p, UInt32 keepWindowSize, ISzAlloc *alloc, I if (!p->litProbs || !p->saveState.litProbs || p->lclp != lclp) { LzmaEnc_FreeLits(p, alloc); - p->litProbs = (CLzmaProb *)alloc->Alloc(alloc, ((UInt32)0x300 << lclp) * sizeof(CLzmaProb)); - p->saveState.litProbs = (CLzmaProb *)alloc->Alloc(alloc, ((UInt32)0x300 << lclp) * sizeof(CLzmaProb)); + p->litProbs = (CLzmaProb *)ISzAlloc_Alloc(alloc, ((UInt32)0x300 << lclp) * sizeof(CLzmaProb)); + p->saveState.litProbs = (CLzmaProb *)ISzAlloc_Alloc(alloc, ((UInt32)0x300 << lclp) * sizeof(CLzmaProb)); if (!p->litProbs || !p->saveState.litProbs) { LzmaEnc_FreeLits(p, alloc); @@ -1985,8 +2590,13 @@ static SRes LzmaEnc_Alloc(CLzmaEnc *p, UInt32 keepWindowSize, ISzAlloc *alloc, I #ifndef _7ZIP_ST if (p->mtMode) { - RINOK(MatchFinderMt_Create(&p->matchFinderMt, p->dictSize, beforeSize, p->numFastBytes, LZMA_MATCH_LEN_MAX, allocBig)); + RINOK(MatchFinderMt_Create(&p->matchFinderMt, p->dictSize, beforeSize, p->numFastBytes, + LZMA_MATCH_LEN_MAX + + 1 /* 18.04 */ + , allocBig)); p->matchFinderObj = &p->matchFinderMt; + p->matchFinderBase.bigHash = (Byte)( + (p->dictSize > kBigHashDicLimit && p->matchFinderBase.hashMask >= 0xFFFFFF) ? 1 : 0); MatchFinderMt_CreateVTable(&p->matchFinderMt, &p->matchFinder); } else @@ -2003,17 +2613,21 @@ static SRes LzmaEnc_Alloc(CLzmaEnc *p, UInt32 keepWindowSize, ISzAlloc *alloc, I void LzmaEnc_Init(CLzmaEnc *p) { - UInt32 i; + unsigned i; p->state = 0; - for (i = 0 ; i < LZMA_NUM_REPS; i++) - p->reps[i] = 0; + p->reps[0] = + p->reps[1] = + p->reps[2] = + p->reps[3] = 1; RangeEnc_Init(&p->rc); + for (i = 0; i < (1 << kNumAlignBits); i++) + p->posAlignEncoder[i] = kProbInitValue; for (i = 0; i < kNumStates; i++) { - UInt32 j; + unsigned j; for (j = 0; j < LZMA_NUM_PB_STATES_MAX; j++) { p->isMatch[i][j] = kProbInitValue; @@ -2025,41 +2639,47 @@ void LzmaEnc_Init(CLzmaEnc *p) p->isRepG2[i] = kProbInitValue; } - { - UInt32 num = (UInt32)0x300 << (p->lp + p->lc); - CLzmaProb *probs = p->litProbs; - for (i = 0; i < num; i++) - probs[i] = kProbInitValue; - } - { for (i = 0; i < kNumLenToPosStates; i++) { CLzmaProb *probs = p->posSlotEncoder[i]; - UInt32 j; + unsigned j; for (j = 0; j < (1 << kNumPosSlotBits); j++) probs[j] = kProbInitValue; } } { - for (i = 0; i < kNumFullDistances - kEndPosModelIndex; i++) + for (i = 0; i < kNumFullDistances; i++) p->posEncoders[i] = kProbInitValue; } - LenEnc_Init(&p->lenEnc.p); - LenEnc_Init(&p->repLenEnc.p); + { + UInt32 num = (UInt32)0x300 << (p->lp + p->lc); + UInt32 k; + CLzmaProb *probs = p->litProbs; + for (k = 0; k < num; k++) + probs[k] = kProbInitValue; + } - for (i = 0; i < (1 << kNumAlignBits); i++) - p->posAlignEncoder[i] = kProbInitValue; - p->optimumEndIndex = 0; - p->optimumCurrentIndex = 0; + LenEnc_Init(&p->lenProbs); + LenEnc_Init(&p->repLenProbs); + + p->optEnd = 0; + p->optCur = 0; + + { + for (i = 0; i < kNumOpts; i++) + p->opt[i].price = kInfinityPrice; + } + p->additionalOffset = 0; p->pbMask = (1 << p->pb) - 1; - p->lpMask = (1 << p->lp) - 1; + p->lpMask = ((UInt32)0x100 << p->lp) - ((unsigned)0x100 >> p->lc); } + void LzmaEnc_InitPrices(CLzmaEnc *p) { if (!p->fastMode) @@ -2071,14 +2691,17 @@ void LzmaEnc_InitPrices(CLzmaEnc *p) p->lenEnc.tableSize = p->repLenEnc.tableSize = p->numFastBytes + 1 - LZMA_MATCH_LEN_MIN; - LenPriceEnc_UpdateTables(&p->lenEnc, 1 << p->pb, p->ProbPrices); - LenPriceEnc_UpdateTables(&p->repLenEnc, 1 << p->pb, p->ProbPrices); + + p->repLenEncCounter = REP_LEN_COUNT; + + LenPriceEnc_UpdateTables(&p->lenEnc, 1 << p->pb, &p->lenProbs, p->ProbPrices); + LenPriceEnc_UpdateTables(&p->repLenEnc, 1 << p->pb, &p->repLenProbs, p->ProbPrices); } -static SRes LzmaEnc_AllocAndInit(CLzmaEnc *p, UInt32 keepWindowSize, ISzAlloc *alloc, ISzAlloc *allocBig) +static SRes LzmaEnc_AllocAndInit(CLzmaEnc *p, UInt32 keepWindowSize, ISzAllocPtr alloc, ISzAllocPtr allocBig) { - UInt32 i; - for (i = 0; i < (UInt32)kDicLogSizeMaxCompress; i++) + unsigned i; + for (i = kEndPosModelIndex / 2; i < kDicLogSizeMax; i++) if (p->dictSize <= ((UInt32)1 << i)) break; p->distTableSize = i * 2; @@ -2093,7 +2716,7 @@ static SRes LzmaEnc_AllocAndInit(CLzmaEnc *p, UInt32 keepWindowSize, ISzAlloc *a } static SRes LzmaEnc_Prepare(CLzmaEncHandle pp, ISeqOutStream *outStream, ISeqInStream *inStream, - ISzAlloc *alloc, ISzAlloc *allocBig) + ISzAllocPtr alloc, ISzAllocPtr allocBig) { CLzmaEnc *p = (CLzmaEnc *)pp; p->matchFinderBase.stream = inStream; @@ -2104,7 +2727,7 @@ static SRes LzmaEnc_Prepare(CLzmaEncHandle pp, ISeqOutStream *outStream, ISeqInS SRes LzmaEnc_PrepareForLzma2(CLzmaEncHandle pp, ISeqInStream *inStream, UInt32 keepWindowSize, - ISzAlloc *alloc, ISzAlloc *allocBig) + ISzAllocPtr alloc, ISzAllocPtr allocBig) { CLzmaEnc *p = (CLzmaEnc *)pp; p->matchFinderBase.stream = inStream; @@ -2120,12 +2743,13 @@ static void LzmaEnc_SetInputBuf(CLzmaEnc *p, const Byte *src, SizeT srcLen) } SRes LzmaEnc_MemPrepare(CLzmaEncHandle pp, const Byte *src, SizeT srcLen, - UInt32 keepWindowSize, ISzAlloc *alloc, ISzAlloc *allocBig) + UInt32 keepWindowSize, ISzAllocPtr alloc, ISzAllocPtr allocBig) { CLzmaEnc *p = (CLzmaEnc *)pp; LzmaEnc_SetInputBuf(p, src, srcLen); p->needInit = 1; + LzmaEnc_SetDataSize(pp, srcLen); return LzmaEnc_AllocAndInit(p, keepWindowSize, alloc, allocBig); } @@ -2143,15 +2767,15 @@ void LzmaEnc_Finish(CLzmaEncHandle pp) typedef struct { - ISeqOutStream funcTable; + ISeqOutStream vt; Byte *data; SizeT rem; - Bool overflow; -} CSeqOutStreamBuf; + BoolInt overflow; +} CLzmaEnc_SeqOutStreamBuf; -static size_t MyWrite(void *pp, const void *data, size_t size) +static size_t SeqOutStreamBuf_Write(const ISeqOutStream *pp, const void *data, size_t size) { - CSeqOutStreamBuf *p = (CSeqOutStreamBuf *)pp; + CLzmaEnc_SeqOutStreamBuf *p = CONTAINER_FROM_VTBL(pp, CLzmaEnc_SeqOutStreamBuf, vt); if (p->rem < size) { size = p->rem; @@ -2178,15 +2802,15 @@ const Byte *LzmaEnc_GetCurBuf(CLzmaEncHandle pp) } -SRes LzmaEnc_CodeOneMemBlock(CLzmaEncHandle pp, Bool reInit, +SRes LzmaEnc_CodeOneMemBlock(CLzmaEncHandle pp, BoolInt reInit, Byte *dest, size_t *destLen, UInt32 desiredPackSize, UInt32 *unpackSize) { CLzmaEnc *p = (CLzmaEnc *)pp; UInt64 nowPos64; SRes res; - CSeqOutStreamBuf outStream; + CLzmaEnc_SeqOutStreamBuf outStream; - outStream.funcTable.Write = MyWrite; + outStream.vt.Write = SeqOutStreamBuf_Write; outStream.data = dest; outStream.rem = *destLen; outStream.overflow = False; @@ -2198,11 +2822,15 @@ SRes LzmaEnc_CodeOneMemBlock(CLzmaEncHandle pp, Bool reInit, if (reInit) LzmaEnc_Init(p); LzmaEnc_InitPrices(p); + nowPos64 = p->nowPos64; RangeEnc_Init(&p->rc); - p->rc.outStream = &outStream.funcTable; + p->rc.outStream = &outStream.vt; - res = LzmaEnc_CodeOneBlock(p, True, desiredPackSize, *unpackSize); + if (desiredPackSize == 0) + return SZ_ERROR_OUTPUT_EOF; + + res = LzmaEnc_CodeOneBlock(p, desiredPackSize, *unpackSize); *unpackSize = (UInt32)(p->nowPos64 - nowPos64); *destLen -= outStream.rem; @@ -2225,12 +2853,12 @@ static SRes LzmaEnc_Encode2(CLzmaEnc *p, ICompressProgress *progress) for (;;) { - res = LzmaEnc_CodeOneBlock(p, False, 0, 0); + res = LzmaEnc_CodeOneBlock(p, 0, 0); if (res != SZ_OK || p->finished) break; if (progress) { - res = progress->Progress(progress, p->nowPos64, RangeEnc_GetProcessed(&p->rc)); + res = ICompressProgress_Progress(progress, p->nowPos64, RangeEnc_GetProcessed(&p->rc)); if (res != SZ_OK) { res = SZ_ERROR_PROGRESS; @@ -2242,7 +2870,7 @@ static SRes LzmaEnc_Encode2(CLzmaEnc *p, ICompressProgress *progress) LzmaEnc_Finish(p); /* - if (res == S_OK && !Inline_MatchFinder_IsFinishedOK(&p->matchFinderBase)) + if (res == SZ_OK && !Inline_MatchFinder_IsFinishedOK(&p->matchFinderBase)) res = SZ_ERROR_FAIL; } */ @@ -2252,7 +2880,7 @@ static SRes LzmaEnc_Encode2(CLzmaEnc *p, ICompressProgress *progress) SRes LzmaEnc_Encode(CLzmaEncHandle pp, ISeqOutStream *outStream, ISeqInStream *inStream, ICompressProgress *progress, - ISzAlloc *alloc, ISzAlloc *allocBig) + ISzAllocPtr alloc, ISzAllocPtr allocBig) { RINOK(LzmaEnc_Prepare(pp, outStream, inStream, alloc, allocBig)); return LzmaEnc_Encode2((CLzmaEnc *)pp, progress); @@ -2287,21 +2915,27 @@ SRes LzmaEnc_WriteProperties(CLzmaEncHandle pp, Byte *props, SizeT *size) } +unsigned LzmaEnc_IsWriteEndMark(CLzmaEncHandle pp) +{ + return ((CLzmaEnc *)pp)->writeEndMark; +} + + SRes LzmaEnc_MemEncode(CLzmaEncHandle pp, Byte *dest, SizeT *destLen, const Byte *src, SizeT srcLen, - int writeEndMark, ICompressProgress *progress, ISzAlloc *alloc, ISzAlloc *allocBig) + int writeEndMark, ICompressProgress *progress, ISzAllocPtr alloc, ISzAllocPtr allocBig) { SRes res; CLzmaEnc *p = (CLzmaEnc *)pp; - CSeqOutStreamBuf outStream; + CLzmaEnc_SeqOutStreamBuf outStream; - outStream.funcTable.Write = MyWrite; + outStream.vt.Write = SeqOutStreamBuf_Write; outStream.data = dest; outStream.rem = *destLen; outStream.overflow = False; p->writeEndMark = writeEndMark; - p->rc.outStream = &outStream.funcTable; + p->rc.outStream = &outStream.vt; res = LzmaEnc_MemPrepare(pp, src, srcLen, 0, alloc, allocBig); @@ -2321,7 +2955,7 @@ SRes LzmaEnc_MemEncode(CLzmaEncHandle pp, Byte *dest, SizeT *destLen, const Byte SRes LzmaEncode(Byte *dest, SizeT *destLen, const Byte *src, SizeT srcLen, const CLzmaEncProps *props, Byte *propsEncoded, SizeT *propsSize, int writeEndMark, - ICompressProgress *progress, ISzAlloc *alloc, ISzAlloc *allocBig) + ICompressProgress *progress, ISzAllocPtr alloc, ISzAllocPtr allocBig) { CLzmaEnc *p = (CLzmaEnc *)LzmaEnc_Create(alloc); SRes res; diff --git a/src/third-party/lzma/LzmaEnc.h b/src/third-party/lzma/LzmaEnc.h index c2806b45f..c9938f04b 100644 --- a/src/third-party/lzma/LzmaEnc.h +++ b/src/third-party/lzma/LzmaEnc.h @@ -1,5 +1,5 @@ /* LzmaEnc.h -- LZMA Encoder -2013-01-18 : Igor Pavlov : Public domain */ +2017-07-27 : Igor Pavlov : Public domain */ #ifndef __LZMA_ENC_H #define __LZMA_ENC_H @@ -12,12 +12,10 @@ EXTERN_C_BEGIN typedef struct _CLzmaEncProps { - int level; /* 0 <= level <= 9 */ + int level; /* 0 <= level <= 9 */ UInt32 dictSize; /* (1 << 12) <= dictSize <= (1 << 27) for 32-bit version - (1 << 12) <= dictSize <= (1 << 30) for 64-bit version - default = (1 << 24) */ - UInt64 reduceSize; /* estimated size of data that will be compressed. default = 0xFFFFFFFF. - Encoder uses this value to reduce dictionary size */ + (1 << 12) <= dictSize <= (3 << 29) for 64-bit version + default = (1 << 24) */ int lc; /* 0 <= lc <= 8, default = 3 */ int lp; /* 0 <= lp <= 4, default = 0 */ int pb; /* 0 <= pb <= 4, default = 2 */ @@ -25,9 +23,12 @@ typedef struct _CLzmaEncProps int fb; /* 5 <= fb <= 273, default = 32 */ int btMode; /* 0 - hashChain Mode, 1 - binTree mode - normal, default = 1 */ int numHashBytes; /* 2, 3 or 4, default = 4 */ - UInt32 mc; /* 1 <= mc <= (1 << 30), default = 32 */ + UInt32 mc; /* 1 <= mc <= (1 << 30), default = 32 */ unsigned writeEndMark; /* 0 - do not write EOPM, 1 - write EOPM, default = 0 */ int numThreads; /* 1 or 2, default = 2 */ + + UInt64 reduceSize; /* estimated size of data that will be compressed. default = (UInt64)(Int64)-1. + Encoder uses this value to reduce dictionary size */ } CLzmaEncProps; void LzmaEncProps_Init(CLzmaEncProps *p); @@ -37,41 +38,38 @@ UInt32 LzmaEncProps_GetDictSize(const CLzmaEncProps *props2); /* ---------- CLzmaEncHandle Interface ---------- */ -/* LzmaEnc_* functions can return the following exit codes: -Returns: +/* LzmaEnc* functions can return the following exit codes: +SRes: SZ_OK - OK SZ_ERROR_MEM - Memory allocation error SZ_ERROR_PARAM - Incorrect paramater in props - SZ_ERROR_WRITE - Write callback error. + SZ_ERROR_WRITE - ISeqOutStream write callback error + SZ_ERROR_OUTPUT_EOF - output buffer overflow - version with (Byte *) output SZ_ERROR_PROGRESS - some break from progress callback - SZ_ERROR_THREAD - errors in multithreading functions (only for Mt version) + SZ_ERROR_THREAD - error in multithreading functions (only for Mt version) */ typedef void * CLzmaEncHandle; -CLzmaEncHandle LzmaEnc_Create(ISzAlloc *alloc); -void LzmaEnc_Destroy(CLzmaEncHandle p, ISzAlloc *alloc, ISzAlloc *allocBig); +CLzmaEncHandle LzmaEnc_Create(ISzAllocPtr alloc); +void LzmaEnc_Destroy(CLzmaEncHandle p, ISzAllocPtr alloc, ISzAllocPtr allocBig); + SRes LzmaEnc_SetProps(CLzmaEncHandle p, const CLzmaEncProps *props); +void LzmaEnc_SetDataSize(CLzmaEncHandle p, UInt64 expectedDataSiize); SRes LzmaEnc_WriteProperties(CLzmaEncHandle p, Byte *properties, SizeT *size); +unsigned LzmaEnc_IsWriteEndMark(CLzmaEncHandle p); + SRes LzmaEnc_Encode(CLzmaEncHandle p, ISeqOutStream *outStream, ISeqInStream *inStream, - ICompressProgress *progress, ISzAlloc *alloc, ISzAlloc *allocBig); + ICompressProgress *progress, ISzAllocPtr alloc, ISzAllocPtr allocBig); SRes LzmaEnc_MemEncode(CLzmaEncHandle p, Byte *dest, SizeT *destLen, const Byte *src, SizeT srcLen, - int writeEndMark, ICompressProgress *progress, ISzAlloc *alloc, ISzAlloc *allocBig); + int writeEndMark, ICompressProgress *progress, ISzAllocPtr alloc, ISzAllocPtr allocBig); + /* ---------- One Call Interface ---------- */ -/* LzmaEncode -Return code: - SZ_OK - OK - SZ_ERROR_MEM - Memory allocation error - SZ_ERROR_PARAM - Incorrect paramater - SZ_ERROR_OUTPUT_EOF - output buffer overflow - SZ_ERROR_THREAD - errors in multithreading functions (only for Mt version) -*/ - SRes LzmaEncode(Byte *dest, SizeT *destLen, const Byte *src, SizeT srcLen, const CLzmaEncProps *props, Byte *propsEncoded, SizeT *propsSize, int writeEndMark, - ICompressProgress *progress, ISzAlloc *alloc, ISzAlloc *allocBig); + ICompressProgress *progress, ISzAllocPtr alloc, ISzAllocPtr allocBig); EXTERN_C_END diff --git a/src/third-party/lzma/MtCoder.c b/src/third-party/lzma/MtCoder.c index 8c0d9b339..5667f2d5b 100644 --- a/src/third-party/lzma/MtCoder.c +++ b/src/third-party/lzma/MtCoder.c @@ -1,168 +1,89 @@ /* MtCoder.c -- Multi-thread Coder -2015-10-13 : Igor Pavlov : Public domain */ +2018-07-04 : Igor Pavlov : Public domain */ #include "Precomp.h" #include "MtCoder.h" -void LoopThread_Construct(CLoopThread *p) -{ - Thread_Construct(&p->thread); - Event_Construct(&p->startEvent); - Event_Construct(&p->finishedEvent); -} +#ifndef _7ZIP_ST -void LoopThread_Close(CLoopThread *p) +SRes MtProgressThunk_Progress(const ICompressProgress *pp, UInt64 inSize, UInt64 outSize) { - Thread_Close(&p->thread); - Event_Close(&p->startEvent); - Event_Close(&p->finishedEvent); -} - -static THREAD_FUNC_RET_TYPE THREAD_FUNC_CALL_TYPE LoopThreadFunc(void *pp) -{ - CLoopThread *p = (CLoopThread *)pp; - for (;;) + CMtProgressThunk *thunk = CONTAINER_FROM_VTBL(pp, CMtProgressThunk, vt); + UInt64 inSize2 = 0; + UInt64 outSize2 = 0; + if (inSize != (UInt64)(Int64)-1) { - if (Event_Wait(&p->startEvent) != 0) - return SZ_ERROR_THREAD; - if (p->stop) - return 0; - p->res = p->func(p->param); - if (Event_Set(&p->finishedEvent) != 0) - return SZ_ERROR_THREAD; + inSize2 = inSize - thunk->inSize; + thunk->inSize = inSize; } + if (outSize != (UInt64)(Int64)-1) + { + outSize2 = outSize - thunk->outSize; + thunk->outSize = outSize; + } + return MtProgress_ProgressAdd(thunk->mtProgress, inSize2, outSize2); } -WRes LoopThread_Create(CLoopThread *p) + +void MtProgressThunk_CreateVTable(CMtProgressThunk *p) { - p->stop = 0; - RINOK(AutoResetEvent_CreateNotSignaled(&p->startEvent)); - RINOK(AutoResetEvent_CreateNotSignaled(&p->finishedEvent)); - return Thread_Create(&p->thread, LoopThreadFunc, p); + p->vt.Progress = MtProgressThunk_Progress; } -WRes LoopThread_StopAndWait(CLoopThread *p) -{ - p->stop = 1; - if (Event_Set(&p->startEvent) != 0) - return SZ_ERROR_THREAD; - return Thread_Wait(&p->thread); -} -WRes LoopThread_StartSubThread(CLoopThread *p) { return Event_Set(&p->startEvent); } -WRes LoopThread_WaitSubThread(CLoopThread *p) { return Event_Wait(&p->finishedEvent); } - -static SRes Progress(ICompressProgress *p, UInt64 inSize, UInt64 outSize) -{ - return (p && p->Progress(p, inSize, outSize) != SZ_OK) ? SZ_ERROR_PROGRESS : SZ_OK; -} - -static void MtProgress_Init(CMtProgress *p, ICompressProgress *progress) -{ - unsigned i; - for (i = 0; i < NUM_MT_CODER_THREADS_MAX; i++) - p->inSizes[i] = p->outSizes[i] = 0; - p->totalInSize = p->totalOutSize = 0; - p->progress = progress; - p->res = SZ_OK; -} - -static void MtProgress_Reinit(CMtProgress *p, unsigned index) -{ - p->inSizes[index] = 0; - p->outSizes[index] = 0; -} - -#define UPDATE_PROGRESS(size, prev, total) \ - if (size != (UInt64)(Int64)-1) { total += size - prev; prev = size; } - -SRes MtProgress_Set(CMtProgress *p, unsigned index, UInt64 inSize, UInt64 outSize) -{ - SRes res; - CriticalSection_Enter(&p->cs); - UPDATE_PROGRESS(inSize, p->inSizes[index], p->totalInSize) - UPDATE_PROGRESS(outSize, p->outSizes[index], p->totalOutSize) - if (p->res == SZ_OK) - p->res = Progress(p->progress, p->totalInSize, p->totalOutSize); - res = p->res; - CriticalSection_Leave(&p->cs); - return res; -} - -static void MtProgress_SetError(CMtProgress *p, SRes res) -{ - CriticalSection_Enter(&p->cs); - if (p->res == SZ_OK) - p->res = res; - CriticalSection_Leave(&p->cs); -} - -static void MtCoder_SetError(CMtCoder* p, SRes res) -{ - CriticalSection_Enter(&p->cs); - if (p->res == SZ_OK) - p->res = res; - CriticalSection_Leave(&p->cs); -} - -/* ---------- MtThread ---------- */ - -void CMtThread_Construct(CMtThread *p, CMtCoder *mtCoder) -{ - p->mtCoder = mtCoder; - p->outBuf = 0; - p->inBuf = 0; - Event_Construct(&p->canRead); - Event_Construct(&p->canWrite); - LoopThread_Construct(&p->thread); -} #define RINOK_THREAD(x) { if ((x) != 0) return SZ_ERROR_THREAD; } -static void CMtThread_CloseEvents(CMtThread *p) + +static WRes ArEvent_OptCreate_And_Reset(CEvent *p) { - Event_Close(&p->canRead); - Event_Close(&p->canWrite); + if (Event_IsCreated(p)) + return Event_Reset(p); + return AutoResetEvent_CreateNotSignaled(p); } -static void CMtThread_Destruct(CMtThread *p) -{ - CMtThread_CloseEvents(p); - if (Thread_WasCreated(&p->thread.thread)) +static THREAD_FUNC_RET_TYPE THREAD_FUNC_CALL_TYPE ThreadFunc(void *pp); + + +static SRes MtCoderThread_CreateAndStart(CMtCoderThread *t) +{ + WRes wres = ArEvent_OptCreate_And_Reset(&t->startEvent); + if (wres == 0) { - LoopThread_StopAndWait(&p->thread); - LoopThread_Close(&p->thread); + t->stop = False; + if (!Thread_WasCreated(&t->thread)) + wres = Thread_Create(&t->thread, ThreadFunc, t); + if (wres == 0) + wres = Event_Set(&t->startEvent); + } + if (wres == 0) + return SZ_OK; + return MY_SRes_HRESULT_FROM_WRes(wres); +} + + +static void MtCoderThread_Destruct(CMtCoderThread *t) +{ + if (Thread_WasCreated(&t->thread)) + { + t->stop = 1; + Event_Set(&t->startEvent); + Thread_Wait(&t->thread); + Thread_Close(&t->thread); } - if (p->mtCoder->alloc) - IAlloc_Free(p->mtCoder->alloc, p->outBuf); - p->outBuf = 0; + Event_Close(&t->startEvent); - if (p->mtCoder->alloc) - IAlloc_Free(p->mtCoder->alloc, p->inBuf); - p->inBuf = 0; + if (t->inBuf) + { + ISzAlloc_Free(t->mtCoder->allocBig, t->inBuf); + t->inBuf = NULL; + } } -#define MY_BUF_ALLOC(buf, size, newSize) \ - if (buf == 0 || size != newSize) \ - { IAlloc_Free(p->mtCoder->alloc, buf); \ - size = newSize; buf = (Byte *)IAlloc_Alloc(p->mtCoder->alloc, size); \ - if (buf == 0) return SZ_ERROR_MEM; } -static SRes CMtThread_Prepare(CMtThread *p) -{ - MY_BUF_ALLOC(p->inBuf, p->inBufSize, p->mtCoder->blockSize) - MY_BUF_ALLOC(p->outBuf, p->outBufSize, p->mtCoder->destBlockSize) - - p->stopReading = False; - p->stopWriting = False; - RINOK_THREAD(AutoResetEvent_CreateNotSignaled(&p->canRead)); - RINOK_THREAD(AutoResetEvent_CreateNotSignaled(&p->canWrite)); - - return SZ_OK; -} static SRes FullRead(ISeqInStream *stream, Byte *data, size_t *processedSize) { @@ -170,158 +91,511 @@ static SRes FullRead(ISeqInStream *stream, Byte *data, size_t *processedSize) *processedSize = 0; while (size != 0) { - size_t curSize = size; - SRes res = stream->Read(stream, data, &curSize); - *processedSize += curSize; - data += curSize; - size -= curSize; + size_t cur = size; + SRes res = ISeqInStream_Read(stream, data, &cur); + *processedSize += cur; + data += cur; + size -= cur; RINOK(res); - if (curSize == 0) + if (cur == 0) return SZ_OK; } return SZ_OK; } -#define GET_NEXT_THREAD(p) &p->mtCoder->threads[p->index == p->mtCoder->numThreads - 1 ? 0 : p->index + 1] -static SRes MtThread_Process(CMtThread *p, Bool *stop) +/* + ThreadFunc2() returns: + SZ_OK - in all normal cases (even for stream error or memory allocation error) + SZ_ERROR_THREAD - in case of failure in system synch function +*/ + +static SRes ThreadFunc2(CMtCoderThread *t) { - CMtThread *next; - *stop = True; - if (Event_Wait(&p->canRead) != 0) - return SZ_ERROR_THREAD; - - next = GET_NEXT_THREAD(p); - - if (p->stopReading) - { - next->stopReading = True; - return Event_Set(&next->canRead) == 0 ? SZ_OK : SZ_ERROR_THREAD; - } + CMtCoder *mtc = t->mtCoder; - { - size_t size = p->mtCoder->blockSize; - size_t destSize = p->outBufSize; - - RINOK(FullRead(p->mtCoder->inStream, p->inBuf, &size)); - next->stopReading = *stop = (size != p->mtCoder->blockSize); - if (Event_Set(&next->canRead) != 0) - return SZ_ERROR_THREAD; - - RINOK(p->mtCoder->mtCallback->Code(p->mtCoder->mtCallback, p->index, - p->outBuf, &destSize, p->inBuf, size, *stop)); - - MtProgress_Reinit(&p->mtCoder->mtProgress, p->index); - - if (Event_Wait(&p->canWrite) != 0) - return SZ_ERROR_THREAD; - if (p->stopWriting) - return SZ_ERROR_FAIL; - if (p->mtCoder->outStream->Write(p->mtCoder->outStream, p->outBuf, destSize) != destSize) - return SZ_ERROR_WRITE; - return Event_Set(&next->canWrite) == 0 ? SZ_OK : SZ_ERROR_THREAD; - } -} - -static THREAD_FUNC_RET_TYPE THREAD_FUNC_CALL_TYPE ThreadFunc(void *pp) -{ - CMtThread *p = (CMtThread *)pp; for (;;) { - Bool stop; - CMtThread *next = GET_NEXT_THREAD(p); - SRes res = MtThread_Process(p, &stop); - if (res != SZ_OK) + unsigned bi; + SRes res; + SRes res2; + BoolInt finished; + unsigned bufIndex; + size_t size; + const Byte *inData; + UInt64 readProcessed = 0; + + RINOK_THREAD(Event_Wait(&mtc->readEvent)) + + /* after Event_Wait(&mtc->readEvent) we must call Event_Set(&mtc->readEvent) in any case to unlock another threads */ + + if (mtc->stopReading) { - MtCoder_SetError(p->mtCoder, res); - MtProgress_SetError(&p->mtCoder->mtProgress, res); - next->stopReading = True; - next->stopWriting = True; - Event_Set(&next->canRead); - Event_Set(&next->canWrite); - return res; + return Event_Set(&mtc->readEvent) == 0 ? SZ_OK : SZ_ERROR_THREAD; } - if (stop) + + res = MtProgress_GetError(&mtc->mtProgress); + + size = 0; + inData = NULL; + finished = True; + + if (res == SZ_OK) + { + size = mtc->blockSize; + if (mtc->inStream) + { + if (!t->inBuf) + { + t->inBuf = (Byte *)ISzAlloc_Alloc(mtc->allocBig, mtc->blockSize); + if (!t->inBuf) + res = SZ_ERROR_MEM; + } + if (res == SZ_OK) + { + res = FullRead(mtc->inStream, t->inBuf, &size); + readProcessed = mtc->readProcessed + size; + mtc->readProcessed = readProcessed; + } + if (res != SZ_OK) + { + mtc->readRes = res; + /* after reading error - we can stop encoding of previous blocks */ + MtProgress_SetError(&mtc->mtProgress, res); + } + else + finished = (size != mtc->blockSize); + } + else + { + size_t rem; + readProcessed = mtc->readProcessed; + rem = mtc->inDataSize - (size_t)readProcessed; + if (size > rem) + size = rem; + inData = mtc->inData + (size_t)readProcessed; + readProcessed += size; + mtc->readProcessed = readProcessed; + finished = (mtc->inDataSize == (size_t)readProcessed); + } + } + + /* we must get some block from blocksSemaphore before Event_Set(&mtc->readEvent) */ + + res2 = SZ_OK; + + if (Semaphore_Wait(&mtc->blocksSemaphore) != 0) + { + res2 = SZ_ERROR_THREAD; + if (res == SZ_OK) + { + res = res2; + // MtProgress_SetError(&mtc->mtProgress, res); + } + } + + bi = mtc->blockIndex; + + if (++mtc->blockIndex >= mtc->numBlocksMax) + mtc->blockIndex = 0; + + bufIndex = (unsigned)(int)-1; + + if (res == SZ_OK) + res = MtProgress_GetError(&mtc->mtProgress); + + if (res != SZ_OK) + finished = True; + + if (!finished) + { + if (mtc->numStartedThreads < mtc->numStartedThreadsLimit + && mtc->expectedDataSize != readProcessed) + { + res = MtCoderThread_CreateAndStart(&mtc->threads[mtc->numStartedThreads]); + if (res == SZ_OK) + mtc->numStartedThreads++; + else + { + MtProgress_SetError(&mtc->mtProgress, res); + finished = True; + } + } + } + + if (finished) + mtc->stopReading = True; + + RINOK_THREAD(Event_Set(&mtc->readEvent)) + + if (res2 != SZ_OK) + return res2; + + if (res == SZ_OK) + { + CriticalSection_Enter(&mtc->cs); + bufIndex = mtc->freeBlockHead; + mtc->freeBlockHead = mtc->freeBlockList[bufIndex]; + CriticalSection_Leave(&mtc->cs); + + res = mtc->mtCallback->Code(mtc->mtCallbackObject, t->index, bufIndex, + mtc->inStream ? t->inBuf : inData, size, finished); + + // MtProgress_Reinit(&mtc->mtProgress, t->index); + + if (res != SZ_OK) + MtProgress_SetError(&mtc->mtProgress, res); + } + + { + CMtCoderBlock *block = &mtc->blocks[bi]; + block->res = res; + block->bufIndex = bufIndex; + block->finished = finished; + } + + #ifdef MTCODER__USE_WRITE_THREAD + RINOK_THREAD(Event_Set(&mtc->writeEvents[bi])) + #else + { + unsigned wi; + { + CriticalSection_Enter(&mtc->cs); + wi = mtc->writeIndex; + if (wi == bi) + mtc->writeIndex = (unsigned)(int)-1; + else + mtc->ReadyBlocks[bi] = True; + CriticalSection_Leave(&mtc->cs); + } + + if (wi != bi) + { + if (res != SZ_OK || finished) + return 0; + continue; + } + + if (mtc->writeRes != SZ_OK) + res = mtc->writeRes; + + for (;;) + { + if (res == SZ_OK && bufIndex != (unsigned)(int)-1) + { + res = mtc->mtCallback->Write(mtc->mtCallbackObject, bufIndex); + if (res != SZ_OK) + { + mtc->writeRes = res; + MtProgress_SetError(&mtc->mtProgress, res); + } + } + + if (++wi >= mtc->numBlocksMax) + wi = 0; + { + BoolInt isReady; + + CriticalSection_Enter(&mtc->cs); + + if (bufIndex != (unsigned)(int)-1) + { + mtc->freeBlockList[bufIndex] = mtc->freeBlockHead; + mtc->freeBlockHead = bufIndex; + } + + isReady = mtc->ReadyBlocks[wi]; + + if (isReady) + mtc->ReadyBlocks[wi] = False; + else + mtc->writeIndex = wi; + + CriticalSection_Leave(&mtc->cs); + + RINOK_THREAD(Semaphore_Release1(&mtc->blocksSemaphore)) + + if (!isReady) + break; + } + + { + CMtCoderBlock *block = &mtc->blocks[wi]; + if (res == SZ_OK && block->res != SZ_OK) + res = block->res; + bufIndex = block->bufIndex; + finished = block->finished; + } + } + } + #endif + + if (finished || res != SZ_OK) return 0; } } -void MtCoder_Construct(CMtCoder* p) + +static THREAD_FUNC_RET_TYPE THREAD_FUNC_CALL_TYPE ThreadFunc(void *pp) +{ + CMtCoderThread *t = (CMtCoderThread *)pp; + for (;;) + { + if (Event_Wait(&t->startEvent) != 0) + return SZ_ERROR_THREAD; + if (t->stop) + return 0; + { + SRes res = ThreadFunc2(t); + CMtCoder *mtc = t->mtCoder; + if (res != SZ_OK) + { + MtProgress_SetError(&mtc->mtProgress, res); + } + + #ifndef MTCODER__USE_WRITE_THREAD + { + unsigned numFinished = (unsigned)InterlockedIncrement(&mtc->numFinishedThreads); + if (numFinished == mtc->numStartedThreads) + if (Event_Set(&mtc->finishedEvent) != 0) + return SZ_ERROR_THREAD; + } + #endif + } + } +} + + + +void MtCoder_Construct(CMtCoder *p) { unsigned i; - p->alloc = 0; - for (i = 0; i < NUM_MT_CODER_THREADS_MAX; i++) + + p->blockSize = 0; + p->numThreadsMax = 0; + p->expectedDataSize = (UInt64)(Int64)-1; + + p->inStream = NULL; + p->inData = NULL; + p->inDataSize = 0; + + p->progress = NULL; + p->allocBig = NULL; + + p->mtCallback = NULL; + p->mtCallbackObject = NULL; + + p->allocatedBufsSize = 0; + + Event_Construct(&p->readEvent); + Semaphore_Construct(&p->blocksSemaphore); + + for (i = 0; i < MTCODER__THREADS_MAX; i++) { - CMtThread *t = &p->threads[i]; + CMtCoderThread *t = &p->threads[i]; + t->mtCoder = p; t->index = i; - CMtThread_Construct(t, p); + t->inBuf = NULL; + t->stop = False; + Event_Construct(&t->startEvent); + Thread_Construct(&t->thread); } + + #ifdef MTCODER__USE_WRITE_THREAD + for (i = 0; i < MTCODER__BLOCKS_MAX; i++) + Event_Construct(&p->writeEvents[i]); + #else + Event_Construct(&p->finishedEvent); + #endif + CriticalSection_Init(&p->cs); CriticalSection_Init(&p->mtProgress.cs); } -void MtCoder_Destruct(CMtCoder* p) + + + +static void MtCoder_Free(CMtCoder *p) { unsigned i; - for (i = 0; i < NUM_MT_CODER_THREADS_MAX; i++) - CMtThread_Destruct(&p->threads[i]); + + /* + p->stopReading = True; + if (Event_IsCreated(&p->readEvent)) + Event_Set(&p->readEvent); + */ + + for (i = 0; i < MTCODER__THREADS_MAX; i++) + MtCoderThread_Destruct(&p->threads[i]); + + Event_Close(&p->readEvent); + Semaphore_Close(&p->blocksSemaphore); + + #ifdef MTCODER__USE_WRITE_THREAD + for (i = 0; i < MTCODER__BLOCKS_MAX; i++) + Event_Close(&p->writeEvents[i]); + #else + Event_Close(&p->finishedEvent); + #endif +} + + +void MtCoder_Destruct(CMtCoder *p) +{ + MtCoder_Free(p); + CriticalSection_Delete(&p->cs); CriticalSection_Delete(&p->mtProgress.cs); } + SRes MtCoder_Code(CMtCoder *p) { - unsigned i, numThreads = p->numThreads; + unsigned numThreads = p->numThreadsMax; + unsigned numBlocksMax; + unsigned i; SRes res = SZ_OK; - p->res = SZ_OK; + + if (numThreads > MTCODER__THREADS_MAX) + numThreads = MTCODER__THREADS_MAX; + numBlocksMax = MTCODER__GET_NUM_BLOCKS_FROM_THREADS(numThreads); + + if (p->blockSize < ((UInt32)1 << 26)) numBlocksMax++; + if (p->blockSize < ((UInt32)1 << 24)) numBlocksMax++; + if (p->blockSize < ((UInt32)1 << 22)) numBlocksMax++; + + if (numBlocksMax > MTCODER__BLOCKS_MAX) + numBlocksMax = MTCODER__BLOCKS_MAX; + + if (p->blockSize != p->allocatedBufsSize) + { + for (i = 0; i < MTCODER__THREADS_MAX; i++) + { + CMtCoderThread *t = &p->threads[i]; + if (t->inBuf) + { + ISzAlloc_Free(p->allocBig, t->inBuf); + t->inBuf = NULL; + } + } + p->allocatedBufsSize = p->blockSize; + } + + p->readRes = SZ_OK; MtProgress_Init(&p->mtProgress, p->progress); - for (i = 0; i < numThreads; i++) + #ifdef MTCODER__USE_WRITE_THREAD + for (i = 0; i < numBlocksMax; i++) + { + RINOK_THREAD(ArEvent_OptCreate_And_Reset(&p->writeEvents[i])); + } + #else + RINOK_THREAD(ArEvent_OptCreate_And_Reset(&p->finishedEvent)); + #endif + { - RINOK(CMtThread_Prepare(&p->threads[i])); + RINOK_THREAD(ArEvent_OptCreate_And_Reset(&p->readEvent)); + + if (Semaphore_IsCreated(&p->blocksSemaphore)) + { + RINOK_THREAD(Semaphore_Close(&p->blocksSemaphore)); + } + RINOK_THREAD(Semaphore_Create(&p->blocksSemaphore, numBlocksMax, numBlocksMax)); } - for (i = 0; i < numThreads; i++) + for (i = 0; i < MTCODER__BLOCKS_MAX - 1; i++) + p->freeBlockList[i] = i + 1; + p->freeBlockList[MTCODER__BLOCKS_MAX - 1] = (unsigned)(int)-1; + p->freeBlockHead = 0; + + p->readProcessed = 0; + p->blockIndex = 0; + p->numBlocksMax = numBlocksMax; + p->stopReading = False; + + #ifndef MTCODER__USE_WRITE_THREAD + p->writeIndex = 0; + p->writeRes = SZ_OK; + for (i = 0; i < MTCODER__BLOCKS_MAX; i++) + p->ReadyBlocks[i] = False; + p->numFinishedThreads = 0; + #endif + + p->numStartedThreadsLimit = numThreads; + p->numStartedThreads = 0; + + // for (i = 0; i < numThreads; i++) { - CMtThread *t = &p->threads[i]; - CLoopThread *lt = &t->thread; + CMtCoderThread *nextThread = &p->threads[p->numStartedThreads++]; + RINOK(MtCoderThread_CreateAndStart(nextThread)); + } - if (!Thread_WasCreated(<->thread)) + RINOK_THREAD(Event_Set(&p->readEvent)) + + #ifdef MTCODER__USE_WRITE_THREAD + { + unsigned bi = 0; + + for (;; bi++) { - lt->func = ThreadFunc; - lt->param = t; + if (bi >= numBlocksMax) + bi = 0; + + RINOK_THREAD(Event_Wait(&p->writeEvents[bi])) - if (LoopThread_Create(lt) != SZ_OK) { - res = SZ_ERROR_THREAD; - break; + const CMtCoderBlock *block = &p->blocks[bi]; + unsigned bufIndex = block->bufIndex; + BoolInt finished = block->finished; + if (res == SZ_OK && block->res != SZ_OK) + res = block->res; + + if (bufIndex != (unsigned)(int)-1) + { + if (res == SZ_OK) + { + res = p->mtCallback->Write(p->mtCallbackObject, bufIndex); + if (res != SZ_OK) + MtProgress_SetError(&p->mtProgress, res); + } + + CriticalSection_Enter(&p->cs); + { + p->freeBlockList[bufIndex] = p->freeBlockHead; + p->freeBlockHead = bufIndex; + } + CriticalSection_Leave(&p->cs); + } + + RINOK_THREAD(Semaphore_Release1(&p->blocksSemaphore)) + + if (finished) + break; } } } + #else + { + WRes wres = Event_Wait(&p->finishedEvent); + res = MY_SRes_HRESULT_FROM_WRes(wres); + } + #endif if (res == SZ_OK) - { - unsigned j; - for (i = 0; i < numThreads; i++) - { - CMtThread *t = &p->threads[i]; - if (LoopThread_StartSubThread(&t->thread) != SZ_OK) - { - res = SZ_ERROR_THREAD; - p->threads[0].stopReading = True; - break; - } - } + res = p->readRes; - Event_Set(&p->threads[0].canWrite); - Event_Set(&p->threads[0].canRead); + if (res == SZ_OK) + res = p->mtProgress.res; - for (j = 0; j < i; j++) - LoopThread_WaitSubThread(&p->threads[j].thread); - } + #ifndef MTCODER__USE_WRITE_THREAD + if (res == SZ_OK) + res = p->writeRes; + #endif - for (i = 0; i < numThreads; i++) - CMtThread_CloseEvents(&p->threads[i]); - return (res == SZ_OK) ? p->res : res; + if (res != SZ_OK) + MtCoder_Free(p); + return res; } + +#endif diff --git a/src/third-party/lzma/MtCoder.h b/src/third-party/lzma/MtCoder.h index 705208ed3..603329d36 100644 --- a/src/third-party/lzma/MtCoder.h +++ b/src/third-party/lzma/MtCoder.h @@ -1,98 +1,141 @@ /* MtCoder.h -- Multi-thread Coder -2009-11-19 : Igor Pavlov : Public domain */ +2018-07-04 : Igor Pavlov : Public domain */ #ifndef __MT_CODER_H #define __MT_CODER_H -#include "Threads.h" +#include "MtDec.h" EXTERN_C_BEGIN -typedef struct -{ - CThread thread; - CAutoResetEvent startEvent; - CAutoResetEvent finishedEvent; - int stop; - - THREAD_FUNC_TYPE func; - LPVOID param; - THREAD_FUNC_RET_TYPE res; -} CLoopThread; - -void LoopThread_Construct(CLoopThread *p); -void LoopThread_Close(CLoopThread *p); -WRes LoopThread_Create(CLoopThread *p); -WRes LoopThread_StopAndWait(CLoopThread *p); -WRes LoopThread_StartSubThread(CLoopThread *p); -WRes LoopThread_WaitSubThread(CLoopThread *p); +/* + if ( defined MTCODER__USE_WRITE_THREAD) : main thread writes all data blocks to output stream + if (not defined MTCODER__USE_WRITE_THREAD) : any coder thread can write data blocks to output stream +*/ +/* #define MTCODER__USE_WRITE_THREAD */ #ifndef _7ZIP_ST -#define NUM_MT_CODER_THREADS_MAX 32 + #define MTCODER__GET_NUM_BLOCKS_FROM_THREADS(numThreads) ((numThreads) + (numThreads) / 8 + 1) + #define MTCODER__THREADS_MAX 64 + #define MTCODER__BLOCKS_MAX (MTCODER__GET_NUM_BLOCKS_FROM_THREADS(MTCODER__THREADS_MAX) + 3) #else -#define NUM_MT_CODER_THREADS_MAX 1 + #define MTCODER__THREADS_MAX 1 + #define MTCODER__BLOCKS_MAX 1 #endif + +#ifndef _7ZIP_ST + + typedef struct { - UInt64 totalInSize; - UInt64 totalOutSize; - ICompressProgress *progress; - SRes res; - CCriticalSection cs; - UInt64 inSizes[NUM_MT_CODER_THREADS_MAX]; - UInt64 outSizes[NUM_MT_CODER_THREADS_MAX]; -} CMtProgress; + ICompressProgress vt; + CMtProgress *mtProgress; + UInt64 inSize; + UInt64 outSize; +} CMtProgressThunk; + +void MtProgressThunk_CreateVTable(CMtProgressThunk *p); + +#define MtProgressThunk_Init(p) { (p)->inSize = 0; (p)->outSize = 0; } -SRes MtProgress_Set(CMtProgress *p, unsigned index, UInt64 inSize, UInt64 outSize); struct _CMtCoder; + typedef struct { struct _CMtCoder *mtCoder; - Byte *outBuf; - size_t outBufSize; - Byte *inBuf; - size_t inBufSize; unsigned index; - CLoopThread thread; + int stop; + Byte *inBuf; + + CAutoResetEvent startEvent; + CThread thread; +} CMtCoderThread; - Bool stopReading; - Bool stopWriting; - CAutoResetEvent canRead; - CAutoResetEvent canWrite; -} CMtThread; typedef struct { - SRes (*Code)(void *p, unsigned index, Byte *dest, size_t *destSize, + SRes (*Code)(void *p, unsigned coderIndex, unsigned outBufIndex, const Byte *src, size_t srcSize, int finished); -} IMtCoderCallback; + SRes (*Write)(void *p, unsigned outBufIndex); +} IMtCoderCallback2; + + +typedef struct +{ + SRes res; + unsigned bufIndex; + BoolInt finished; +} CMtCoderBlock; + typedef struct _CMtCoder { - size_t blockSize; - size_t destBlockSize; - unsigned numThreads; + /* input variables */ - ISeqInStream *inStream; - ISeqOutStream *outStream; - ICompressProgress *progress; - ISzAlloc *alloc; + size_t blockSize; /* size of input block */ + unsigned numThreadsMax; + UInt64 expectedDataSize; + + ISeqInStream *inStream; + const Byte *inData; + size_t inDataSize; + + ICompressProgress *progress; + ISzAllocPtr allocBig; + + IMtCoderCallback2 *mtCallback; + void *mtCallbackObject; + + + /* internal variables */ + + size_t allocatedBufsSize; + + CAutoResetEvent readEvent; + CSemaphore blocksSemaphore; + + BoolInt stopReading; + SRes readRes; + + #ifdef MTCODER__USE_WRITE_THREAD + CAutoResetEvent writeEvents[MTCODER__BLOCKS_MAX]; + #else + CAutoResetEvent finishedEvent; + SRes writeRes; + unsigned writeIndex; + Byte ReadyBlocks[MTCODER__BLOCKS_MAX]; + LONG numFinishedThreads; + #endif + + unsigned numStartedThreadsLimit; + unsigned numStartedThreads; + + unsigned numBlocksMax; + unsigned blockIndex; + UInt64 readProcessed; - IMtCoderCallback *mtCallback; CCriticalSection cs; - SRes res; + + unsigned freeBlockHead; + unsigned freeBlockList[MTCODER__BLOCKS_MAX]; CMtProgress mtProgress; - CMtThread threads[NUM_MT_CODER_THREADS_MAX]; + CMtCoderBlock blocks[MTCODER__BLOCKS_MAX]; + CMtCoderThread threads[MTCODER__THREADS_MAX]; } CMtCoder; -void MtCoder_Construct(CMtCoder* p); -void MtCoder_Destruct(CMtCoder* p); + +void MtCoder_Construct(CMtCoder *p); +void MtCoder_Destruct(CMtCoder *p); SRes MtCoder_Code(CMtCoder *p); + +#endif + + EXTERN_C_END #endif diff --git a/src/third-party/lzma/MtDec.c b/src/third-party/lzma/MtDec.c new file mode 100644 index 000000000..25a8b046d --- /dev/null +++ b/src/third-party/lzma/MtDec.c @@ -0,0 +1,1138 @@ +/* MtDec.c -- Multi-thread Decoder +2019-02-02 : Igor Pavlov : Public domain */ + +#include "Precomp.h" + +// #define SHOW_DEBUG_INFO + +// #include + +#ifdef SHOW_DEBUG_INFO +#include +#endif + +#ifdef SHOW_DEBUG_INFO +#define PRF(x) x +#else +#define PRF(x) +#endif + +#define PRF_STR_INT(s, d) PRF(printf("\n" s " %d\n", (unsigned)d)) + +#include "MtDec.h" + +#ifndef _7ZIP_ST + +void MtProgress_Init(CMtProgress *p, ICompressProgress *progress) +{ + p->progress = progress; + p->res = SZ_OK; + p->totalInSize = 0; + p->totalOutSize = 0; +} + + +SRes MtProgress_Progress_ST(CMtProgress *p) +{ + if (p->res == SZ_OK && p->progress) + if (ICompressProgress_Progress(p->progress, p->totalInSize, p->totalOutSize) != SZ_OK) + p->res = SZ_ERROR_PROGRESS; + return p->res; +} + + +SRes MtProgress_ProgressAdd(CMtProgress *p, UInt64 inSize, UInt64 outSize) +{ + SRes res; + CriticalSection_Enter(&p->cs); + + p->totalInSize += inSize; + p->totalOutSize += outSize; + if (p->res == SZ_OK && p->progress) + if (ICompressProgress_Progress(p->progress, p->totalInSize, p->totalOutSize) != SZ_OK) + p->res = SZ_ERROR_PROGRESS; + res = p->res; + + CriticalSection_Leave(&p->cs); + return res; +} + + +SRes MtProgress_GetError(CMtProgress *p) +{ + SRes res; + CriticalSection_Enter(&p->cs); + res = p->res; + CriticalSection_Leave(&p->cs); + return res; +} + + +void MtProgress_SetError(CMtProgress *p, SRes res) +{ + CriticalSection_Enter(&p->cs); + if (p->res == SZ_OK) + p->res = res; + CriticalSection_Leave(&p->cs); +} + + +#define RINOK_THREAD(x) RINOK(x) + + +static WRes ArEvent_OptCreate_And_Reset(CEvent *p) +{ + if (Event_IsCreated(p)) + return Event_Reset(p); + return AutoResetEvent_CreateNotSignaled(p); +} + + +struct __CMtDecBufLink +{ + struct __CMtDecBufLink *next; + void *pad[3]; +}; + +typedef struct __CMtDecBufLink CMtDecBufLink; + +#define MTDEC__LINK_DATA_OFFSET sizeof(CMtDecBufLink) +#define MTDEC__DATA_PTR_FROM_LINK(link) ((Byte *)(link) + MTDEC__LINK_DATA_OFFSET) + + + +static THREAD_FUNC_RET_TYPE THREAD_FUNC_CALL_TYPE ThreadFunc(void *pp); + + +static WRes MtDecThread_CreateEvents(CMtDecThread *t) +{ + WRes wres = ArEvent_OptCreate_And_Reset(&t->canWrite); + if (wres == 0) + { + wres = ArEvent_OptCreate_And_Reset(&t->canRead); + if (wres == 0) + return SZ_OK; + } + return wres; +} + + +static SRes MtDecThread_CreateAndStart(CMtDecThread *t) +{ + WRes wres = MtDecThread_CreateEvents(t); + // wres = 17; // for test + if (wres == 0) + { + if (Thread_WasCreated(&t->thread)) + return SZ_OK; + wres = Thread_Create(&t->thread, ThreadFunc, t); + if (wres == 0) + return SZ_OK; + } + return MY_SRes_HRESULT_FROM_WRes(wres); +} + + +void MtDecThread_FreeInBufs(CMtDecThread *t) +{ + if (t->inBuf) + { + void *link = t->inBuf; + t->inBuf = NULL; + do + { + void *next = ((CMtDecBufLink *)link)->next; + ISzAlloc_Free(t->mtDec->alloc, link); + link = next; + } + while (link); + } +} + + +static void MtDecThread_CloseThread(CMtDecThread *t) +{ + if (Thread_WasCreated(&t->thread)) + { + Event_Set(&t->canWrite); /* we can disable it. There are no threads waiting canWrite in normal cases */ + Event_Set(&t->canRead); + Thread_Wait(&t->thread); + Thread_Close(&t->thread); + } + + Event_Close(&t->canRead); + Event_Close(&t->canWrite); +} + +static void MtDec_CloseThreads(CMtDec *p) +{ + unsigned i; + for (i = 0; i < MTDEC__THREADS_MAX; i++) + MtDecThread_CloseThread(&p->threads[i]); +} + +static void MtDecThread_Destruct(CMtDecThread *t) +{ + MtDecThread_CloseThread(t); + MtDecThread_FreeInBufs(t); +} + + + +static SRes FullRead(ISeqInStream *stream, Byte *data, size_t *processedSize) +{ + size_t size = *processedSize; + *processedSize = 0; + while (size != 0) + { + size_t cur = size; + SRes res = ISeqInStream_Read(stream, data, &cur); + *processedSize += cur; + data += cur; + size -= cur; + RINOK(res); + if (cur == 0) + return SZ_OK; + } + return SZ_OK; +} + + +static SRes MtDec_GetError_Spec(CMtDec *p, UInt64 interruptIndex, BoolInt *wasInterrupted) +{ + SRes res; + CriticalSection_Enter(&p->mtProgress.cs); + *wasInterrupted = (p->needInterrupt && interruptIndex > p->interruptIndex); + res = p->mtProgress.res; + CriticalSection_Leave(&p->mtProgress.cs); + return res; +} + +static SRes MtDec_Progress_GetError_Spec(CMtDec *p, UInt64 inSize, UInt64 outSize, UInt64 interruptIndex, BoolInt *wasInterrupted) +{ + SRes res; + CriticalSection_Enter(&p->mtProgress.cs); + + p->mtProgress.totalInSize += inSize; + p->mtProgress.totalOutSize += outSize; + if (p->mtProgress.res == SZ_OK && p->mtProgress.progress) + if (ICompressProgress_Progress(p->mtProgress.progress, p->mtProgress.totalInSize, p->mtProgress.totalOutSize) != SZ_OK) + p->mtProgress.res = SZ_ERROR_PROGRESS; + + *wasInterrupted = (p->needInterrupt && interruptIndex > p->interruptIndex); + res = p->mtProgress.res; + + CriticalSection_Leave(&p->mtProgress.cs); + + return res; +} + +static void MtDec_Interrupt(CMtDec *p, UInt64 interruptIndex) +{ + CriticalSection_Enter(&p->mtProgress.cs); + if (!p->needInterrupt || interruptIndex < p->interruptIndex) + { + p->interruptIndex = interruptIndex; + p->needInterrupt = True; + } + CriticalSection_Leave(&p->mtProgress.cs); +} + +Byte *MtDec_GetCrossBuff(CMtDec *p) +{ + Byte *cr = p->crossBlock; + if (!cr) + { + cr = (Byte *)ISzAlloc_Alloc(p->alloc, MTDEC__LINK_DATA_OFFSET + p->inBufSize); + if (!cr) + return NULL; + p->crossBlock = cr; + } + return MTDEC__DATA_PTR_FROM_LINK(cr); +} + + +/* + ThreadFunc2() returns: + 0 - in all normal cases (even for stream error or memory allocation error) + (!= 0) - WRes error return by system threading function +*/ + +// #define MTDEC_ProgessStep (1 << 22) +#define MTDEC_ProgessStep (1 << 0) + +static WRes ThreadFunc2(CMtDecThread *t) +{ + CMtDec *p = t->mtDec; + + PRF_STR_INT("ThreadFunc2", t->index); + + // SetThreadAffinityMask(GetCurrentThread(), 1 << t->index); + + for (;;) + { + SRes res, codeRes; + BoolInt wasInterrupted, isAllocError, overflow, finish; + SRes threadingErrorSRes; + BoolInt needCode, needWrite, needContinue; + + size_t inDataSize_Start; + UInt64 inDataSize; + // UInt64 inDataSize_Full; + + UInt64 blockIndex; + + UInt64 inPrev = 0; + UInt64 outPrev = 0; + UInt64 inCodePos; + UInt64 outCodePos; + + Byte *afterEndData = NULL; + size_t afterEndData_Size = 0; + + BoolInt canCreateNewThread = False; + // CMtDecCallbackInfo parse; + CMtDecThread *nextThread; + + PRF_STR_INT("Event_Wait(&t->canRead)", t->index); + + RINOK_THREAD(Event_Wait(&t->canRead)); + if (p->exitThread) + return 0; + + PRF_STR_INT("after Event_Wait(&t->canRead)", t->index); + + // if (t->index == 3) return 19; // for test + + blockIndex = p->blockIndex++; + + // PRF(printf("\ncanRead\n")) + + res = MtDec_Progress_GetError_Spec(p, 0, 0, blockIndex, &wasInterrupted); + + finish = p->readWasFinished; + needCode = False; + needWrite = False; + isAllocError = False; + overflow = False; + + inDataSize_Start = 0; + inDataSize = 0; + // inDataSize_Full = 0; + + if (res == SZ_OK && !wasInterrupted) + { + // if (p->inStream) + { + CMtDecBufLink *prev = NULL; + CMtDecBufLink *link = (CMtDecBufLink *)t->inBuf; + size_t crossSize = p->crossEnd - p->crossStart; + + PRF(printf("\ncrossSize = %d\n", crossSize)); + + for (;;) + { + if (!link) + { + link = (CMtDecBufLink *)ISzAlloc_Alloc(p->alloc, MTDEC__LINK_DATA_OFFSET + p->inBufSize); + if (!link) + { + finish = True; + // p->allocError_for_Read_BlockIndex = blockIndex; + isAllocError = True; + break; + } + link->next = NULL; + if (prev) + { + // static unsigned g_num = 0; + // printf("\n%6d : %x", ++g_num, (unsigned)(size_t)((Byte *)link - (Byte *)prev)); + prev->next = link; + } + else + t->inBuf = (void *)link; + } + + { + Byte *data = MTDEC__DATA_PTR_FROM_LINK(link); + Byte *parseData = data; + size_t size; + + if (crossSize != 0) + { + inDataSize = crossSize; + // inDataSize_Full = inDataSize; + inDataSize_Start = crossSize; + size = crossSize; + parseData = MTDEC__DATA_PTR_FROM_LINK(p->crossBlock) + p->crossStart; + PRF(printf("\ncross : crossStart = %7d crossEnd = %7d finish = %1d", + (int)p->crossStart, (int)p->crossEnd, (int)finish)); + } + else + { + size = p->inBufSize; + + res = FullRead(p->inStream, data, &size); + + // size = 10; // test + + inDataSize += size; + // inDataSize_Full = inDataSize; + if (!prev) + inDataSize_Start = size; + + p->readProcessed += size; + finish = (size != p->inBufSize); + if (finish) + p->readWasFinished = True; + + // res = E_INVALIDARG; // test + + if (res != SZ_OK) + { + // PRF(printf("\nRead error = %d\n", res)) + // we want to decode all data before error + p->readRes = res; + // p->readError_BlockIndex = blockIndex; + p->readWasFinished = True; + finish = True; + res = SZ_OK; + // break; + } + + if (inDataSize - inPrev >= MTDEC_ProgessStep) + { + res = MtDec_Progress_GetError_Spec(p, 0, 0, blockIndex, &wasInterrupted); + if (res != SZ_OK || wasInterrupted) + break; + inPrev = inDataSize; + } + } + + { + CMtDecCallbackInfo parse; + + parse.startCall = (prev == NULL); + parse.src = parseData; + parse.srcSize = size; + parse.srcFinished = finish; + parse.canCreateNewThread = True; + + // PRF(printf("\nParse size = %d\n", (unsigned)size)) + + p->mtCallback->Parse(p->mtCallbackObject, t->index, &parse); + + needWrite = True; + canCreateNewThread = parse.canCreateNewThread; + + // printf("\n\n%12I64u %12I64u", (UInt64)p->mtProgress.totalInSize, (UInt64)p->mtProgress.totalOutSize); + + if ( + // parseRes != SZ_OK || + // inDataSize - (size - parse.srcSize) > p->inBlockMax + // || + parse.state == MTDEC_PARSE_OVERFLOW + // || wasInterrupted + ) + { + // Overflow or Parse error - switch from MT decoding to ST decoding + finish = True; + overflow = True; + + { + PRF(printf("\n Overflow")); + // PRF(printf("\nisBlockFinished = %d", (unsigned)parse.blockWasFinished)); + PRF(printf("\n inDataSize = %d", (unsigned)inDataSize)); + } + + if (crossSize != 0) + memcpy(data, parseData, size); + p->crossStart = 0; + p->crossEnd = 0; + break; + } + + if (crossSize != 0) + { + memcpy(data, parseData, parse.srcSize); + p->crossStart += parse.srcSize; + } + + if (parse.state != MTDEC_PARSE_CONTINUE || finish) + { + // we don't need to parse in current thread anymore + + if (parse.state == MTDEC_PARSE_END) + finish = True; + + needCode = True; + // p->crossFinished = finish; + + if (parse.srcSize == size) + { + // full parsed - no cross transfer + p->crossStart = 0; + p->crossEnd = 0; + break; + } + + if (parse.state == MTDEC_PARSE_END) + { + p->crossStart = 0; + p->crossEnd = 0; + + if (crossSize != 0) + memcpy(data + parse.srcSize, parseData + parse.srcSize, size - parse.srcSize); // we need all data + afterEndData_Size = size - parse.srcSize; + afterEndData = parseData + parse.srcSize; + + // we reduce data size to required bytes (parsed only) + inDataSize -= (size - parse.srcSize); + if (!prev) + inDataSize_Start = parse.srcSize; + break; + } + + { + // partial parsed - need cross transfer + if (crossSize != 0) + inDataSize = parse.srcSize; // it's only parsed now + else + { + // partial parsed - is not in initial cross block - we need to copy new data to cross block + Byte *cr = MtDec_GetCrossBuff(p); + if (!cr) + { + { + PRF(printf("\ncross alloc error error\n")); + // res = SZ_ERROR_MEM; + finish = True; + // p->allocError_for_Read_BlockIndex = blockIndex; + isAllocError = True; + break; + } + } + + { + size_t crSize = size - parse.srcSize; + inDataSize -= crSize; + p->crossEnd = crSize; + p->crossStart = 0; + memcpy(cr, parseData + parse.srcSize, crSize); + } + } + + // inDataSize_Full = inDataSize; + if (!prev) + inDataSize_Start = parse.srcSize; // it's partial size (parsed only) + + finish = False; + break; + } + } + + if (parse.srcSize != size) + { + res = SZ_ERROR_FAIL; + PRF(printf("\nfinished error SZ_ERROR_FAIL = %d\n", res)); + break; + } + } + } + + prev = link; + link = link->next; + + if (crossSize != 0) + { + crossSize = 0; + p->crossStart = 0; + p->crossEnd = 0; + } + } + } + + if (res == SZ_OK) + res = MtDec_GetError_Spec(p, blockIndex, &wasInterrupted); + } + + codeRes = SZ_OK; + + if (res == SZ_OK && needCode && !wasInterrupted) + { + codeRes = p->mtCallback->PreCode(p->mtCallbackObject, t->index); + if (codeRes != SZ_OK) + { + needCode = False; + finish = True; + // SZ_ERROR_MEM is expected error here. + // if (codeRes == SZ_ERROR_MEM) - we will try single-thread decoding later. + // if (codeRes != SZ_ERROR_MEM) - we can stop decoding or try single-thread decoding. + } + } + + if (res != SZ_OK || wasInterrupted) + finish = True; + + nextThread = NULL; + threadingErrorSRes = SZ_OK; + + if (!finish) + { + if (p->numStartedThreads < p->numStartedThreads_Limit && canCreateNewThread) + { + SRes res2 = MtDecThread_CreateAndStart(&p->threads[p->numStartedThreads]); + if (res2 == SZ_OK) + { + // if (p->numStartedThreads % 1000 == 0) PRF(printf("\n numStartedThreads=%d\n", p->numStartedThreads)); + p->numStartedThreads++; + } + else + { + PRF(printf("\nERROR: numStartedThreads=%d\n", p->numStartedThreads)); + if (p->numStartedThreads == 1) + { + // if only one thread is possible, we leave muti-threading code + finish = True; + needCode = False; + threadingErrorSRes = res2; + } + else + p->numStartedThreads_Limit = p->numStartedThreads; + } + } + + if (!finish) + { + unsigned nextIndex = t->index + 1; + nextThread = &p->threads[nextIndex >= p->numStartedThreads ? 0 : nextIndex]; + RINOK_THREAD(Event_Set(&nextThread->canRead)) + // We have started executing for new iteration (with next thread) + // And that next thread now is responsible for possible exit from decoding (threading_code) + } + } + + // each call of Event_Set(&nextThread->canRead) must be followed by call of Event_Set(&nextThread->canWrite) + // if ( !finish ) we must call Event_Set(&nextThread->canWrite) in any case + // if ( finish ) we switch to single-thread mode and there are 2 ways at the end of current iteration (current block): + // - if (needContinue) after Write(&needContinue), we restore decoding with new iteration + // - otherwise we stop decoding and exit from ThreadFunc2() + + // Don't change (finish) variable in the further code + + + // ---------- CODE ---------- + + inPrev = 0; + outPrev = 0; + inCodePos = 0; + outCodePos = 0; + + if (res == SZ_OK && needCode && codeRes == SZ_OK) + { + BoolInt isStartBlock = True; + CMtDecBufLink *link = (CMtDecBufLink *)t->inBuf; + + for (;;) + { + size_t inSize; + int stop; + + if (isStartBlock) + inSize = inDataSize_Start; + else + { + UInt64 rem = inDataSize - inCodePos; + inSize = p->inBufSize; + if (inSize > rem) + inSize = (size_t)rem; + } + + inCodePos += inSize; + stop = True; + + codeRes = p->mtCallback->Code(p->mtCallbackObject, t->index, + (const Byte *)MTDEC__DATA_PTR_FROM_LINK(link), inSize, + (inCodePos == inDataSize), // srcFinished + &inCodePos, &outCodePos, &stop); + + if (codeRes != SZ_OK) + { + PRF(printf("\nCode Interrupt error = %x\n", codeRes)); + // we interrupt only later blocks + MtDec_Interrupt(p, blockIndex); + break; + } + + if (stop || inCodePos == inDataSize) + break; + + { + const UInt64 inDelta = inCodePos - inPrev; + const UInt64 outDelta = outCodePos - outPrev; + if (inDelta >= MTDEC_ProgessStep || outDelta >= MTDEC_ProgessStep) + { + // Sleep(1); + res = MtDec_Progress_GetError_Spec(p, inDelta, outDelta, blockIndex, &wasInterrupted); + if (res != SZ_OK || wasInterrupted) + break; + inPrev = inCodePos; + outPrev = outCodePos; + } + } + + link = link->next; + isStartBlock = False; + } + } + + + // ---------- WRITE ---------- + + RINOK_THREAD(Event_Wait(&t->canWrite)); + + { + BoolInt isErrorMode = False; + BoolInt canRecode = True; + BoolInt needWriteToStream = needWrite; + + if (p->exitThread) return 0; // it's never executed in normal cases + + if (p->wasInterrupted) + wasInterrupted = True; + else + { + if (codeRes != SZ_OK) // || !needCode // check it !!! + { + p->wasInterrupted = True; + p->codeRes = codeRes; + if (codeRes == SZ_ERROR_MEM) + isAllocError = True; + } + + if (threadingErrorSRes) + { + p->wasInterrupted = True; + p->threadingErrorSRes = threadingErrorSRes; + needWriteToStream = False; + } + if (isAllocError) + { + p->wasInterrupted = True; + p->isAllocError = True; + needWriteToStream = False; + } + if (overflow) + { + p->wasInterrupted = True; + p->overflow = True; + needWriteToStream = False; + } + } + + if (needCode) + { + if (wasInterrupted) + { + inCodePos = 0; + outCodePos = 0; + } + { + const UInt64 inDelta = inCodePos - inPrev; + const UInt64 outDelta = outCodePos - outPrev; + // if (inDelta != 0 || outDelta != 0) + res = MtProgress_ProgressAdd(&p->mtProgress, inDelta, outDelta); + } + } + + needContinue = (!finish); + + // if (res == SZ_OK && needWrite && !wasInterrupted) + if (needWrite) + { + // p->inProcessed += inCodePos; + + res = p->mtCallback->Write(p->mtCallbackObject, t->index, + res == SZ_OK && needWriteToStream && !wasInterrupted, // needWrite + afterEndData, afterEndData_Size, + &needContinue, + &canRecode); + + // res= E_INVALIDARG; // for test + + PRF(printf("\nAfter Write needContinue = %d\n", (unsigned)needContinue)); + PRF(printf("\nprocessed = %d\n", (unsigned)p->inProcessed)); + + if (res != SZ_OK) + { + PRF(printf("\nWrite error = %d\n", res)); + isErrorMode = True; + p->wasInterrupted = True; + } + if (res != SZ_OK + || (!needContinue && !finish)) + { + PRF(printf("\nWrite Interrupt error = %x\n", res)); + MtDec_Interrupt(p, blockIndex); + } + } + + if (canRecode) + if (!needCode + || res != SZ_OK + || p->wasInterrupted + || codeRes != SZ_OK + || wasInterrupted + || p->numFilledThreads != 0 + || isErrorMode) + { + if (p->numFilledThreads == 0) + p->filledThreadStart = t->index; + if (inDataSize != 0 || !finish) + { + t->inDataSize_Start = inDataSize_Start; + t->inDataSize = inDataSize; + p->numFilledThreads++; + } + PRF(printf("\np->numFilledThreads = %d\n", p->numFilledThreads)); + PRF(printf("p->filledThreadStart = %d\n", p->filledThreadStart)); + } + + if (!finish) + { + RINOK_THREAD(Event_Set(&nextThread->canWrite)); + } + else + { + if (needContinue) + { + // we restore decoding with new iteration + RINOK_THREAD(Event_Set(&p->threads[0].canWrite)); + } + else + { + // we exit from decoding + if (t->index == 0) + return SZ_OK; + p->exitThread = True; + } + RINOK_THREAD(Event_Set(&p->threads[0].canRead)); + } + } + } +} + +#ifdef _WIN32 +#define USE_ALLOCA +#endif + +#ifdef USE_ALLOCA +#ifdef _WIN32 +#include +#else +#include +#endif +#endif + + +static THREAD_FUNC_RET_TYPE THREAD_FUNC_CALL_TYPE ThreadFunc1(void *pp) +{ + WRes res; + + CMtDecThread *t = (CMtDecThread *)pp; + CMtDec *p; + + // fprintf(stdout, "\n%d = %p\n", t->index, &t); + + res = ThreadFunc2(t); + p = t->mtDec; + if (res == 0) + return p->exitThreadWRes; + { + // it's unexpected situation for some threading function error + if (p->exitThreadWRes == 0) + p->exitThreadWRes = res; + PRF(printf("\nthread exit error = %d\n", res)); + p->exitThread = True; + Event_Set(&p->threads[0].canRead); + Event_Set(&p->threads[0].canWrite); + MtProgress_SetError(&p->mtProgress, MY_SRes_HRESULT_FROM_WRes(res)); + } + return res; +} + +static MY_NO_INLINE THREAD_FUNC_RET_TYPE THREAD_FUNC_CALL_TYPE ThreadFunc(void *pp) +{ + CMtDecThread *t = (CMtDecThread *)pp; + + // fprintf(stderr, "\n%d = %p - before", t->index, &t); + #ifdef USE_ALLOCA + t->allocaPtr = alloca(t->index * 128); + #endif + return ThreadFunc1(pp); +} + + +int MtDec_PrepareRead(CMtDec *p) +{ + if (p->crossBlock && p->crossStart == p->crossEnd) + { + ISzAlloc_Free(p->alloc, p->crossBlock); + p->crossBlock = NULL; + } + + { + unsigned i; + for (i = 0; i < MTDEC__THREADS_MAX; i++) + if (i > p->numStartedThreads + || p->numFilledThreads <= + (i >= p->filledThreadStart ? + i - p->filledThreadStart : + i + p->numStartedThreads - p->filledThreadStart)) + MtDecThread_FreeInBufs(&p->threads[i]); + } + + return (p->numFilledThreads != 0) || (p->crossStart != p->crossEnd); +} + + +const Byte *MtDec_Read(CMtDec *p, size_t *inLim) +{ + while (p->numFilledThreads != 0) + { + CMtDecThread *t = &p->threads[p->filledThreadStart]; + + if (*inLim != 0) + { + { + void *link = t->inBuf; + void *next = ((CMtDecBufLink *)link)->next; + ISzAlloc_Free(p->alloc, link); + t->inBuf = next; + } + + if (t->inDataSize == 0) + { + MtDecThread_FreeInBufs(t); + if (--p->numFilledThreads == 0) + break; + if (++p->filledThreadStart == p->numStartedThreads) + p->filledThreadStart = 0; + t = &p->threads[p->filledThreadStart]; + } + } + + { + size_t lim = t->inDataSize_Start; + if (lim != 0) + t->inDataSize_Start = 0; + else + { + UInt64 rem = t->inDataSize; + lim = p->inBufSize; + if (lim > rem) + lim = (size_t)rem; + } + t->inDataSize -= lim; + *inLim = lim; + return (const Byte *)MTDEC__DATA_PTR_FROM_LINK(t->inBuf); + } + } + + { + size_t crossSize = p->crossEnd - p->crossStart; + if (crossSize != 0) + { + const Byte *data = MTDEC__DATA_PTR_FROM_LINK(p->crossBlock) + p->crossStart; + *inLim = crossSize; + p->crossStart = 0; + p->crossEnd = 0; + return data; + } + *inLim = 0; + if (p->crossBlock) + { + ISzAlloc_Free(p->alloc, p->crossBlock); + p->crossBlock = NULL; + } + return NULL; + } +} + + +void MtDec_Construct(CMtDec *p) +{ + unsigned i; + + p->inBufSize = (size_t)1 << 18; + + p->numThreadsMax = 0; + + p->inStream = NULL; + + // p->inData = NULL; + // p->inDataSize = 0; + + p->crossBlock = NULL; + p->crossStart = 0; + p->crossEnd = 0; + + p->numFilledThreads = 0; + + p->progress = NULL; + p->alloc = NULL; + + p->mtCallback = NULL; + p->mtCallbackObject = NULL; + + p->allocatedBufsSize = 0; + + for (i = 0; i < MTDEC__THREADS_MAX; i++) + { + CMtDecThread *t = &p->threads[i]; + t->mtDec = p; + t->index = i; + t->inBuf = NULL; + Event_Construct(&t->canRead); + Event_Construct(&t->canWrite); + Thread_Construct(&t->thread); + } + + // Event_Construct(&p->finishedEvent); + + CriticalSection_Init(&p->mtProgress.cs); +} + + +static void MtDec_Free(CMtDec *p) +{ + unsigned i; + + p->exitThread = True; + + for (i = 0; i < MTDEC__THREADS_MAX; i++) + MtDecThread_Destruct(&p->threads[i]); + + // Event_Close(&p->finishedEvent); + + if (p->crossBlock) + { + ISzAlloc_Free(p->alloc, p->crossBlock); + p->crossBlock = NULL; + } +} + + +void MtDec_Destruct(CMtDec *p) +{ + MtDec_Free(p); + + CriticalSection_Delete(&p->mtProgress.cs); +} + + +SRes MtDec_Code(CMtDec *p) +{ + unsigned i; + + p->inProcessed = 0; + + p->blockIndex = 1; // it must be larger than not_defined index (0) + p->isAllocError = False; + p->overflow = False; + p->threadingErrorSRes = SZ_OK; + + p->needContinue = True; + + p->readWasFinished = False; + p->needInterrupt = False; + p->interruptIndex = (UInt64)(Int64)-1; + + p->readProcessed = 0; + p->readRes = SZ_OK; + p->codeRes = SZ_OK; + p->wasInterrupted = False; + + p->crossStart = 0; + p->crossEnd = 0; + + p->filledThreadStart = 0; + p->numFilledThreads = 0; + + { + unsigned numThreads = p->numThreadsMax; + if (numThreads > MTDEC__THREADS_MAX) + numThreads = MTDEC__THREADS_MAX; + p->numStartedThreads_Limit = numThreads; + p->numStartedThreads = 0; + } + + if (p->inBufSize != p->allocatedBufsSize) + { + for (i = 0; i < MTDEC__THREADS_MAX; i++) + { + CMtDecThread *t = &p->threads[i]; + if (t->inBuf) + MtDecThread_FreeInBufs(t); + } + if (p->crossBlock) + { + ISzAlloc_Free(p->alloc, p->crossBlock); + p->crossBlock = NULL; + } + + p->allocatedBufsSize = p->inBufSize; + } + + MtProgress_Init(&p->mtProgress, p->progress); + + // RINOK_THREAD(ArEvent_OptCreate_And_Reset(&p->finishedEvent)); + p->exitThread = False; + p->exitThreadWRes = 0; + + { + WRes wres; + WRes sres; + CMtDecThread *nextThread = &p->threads[p->numStartedThreads++]; + // wres = MtDecThread_CreateAndStart(nextThread); + wres = MtDecThread_CreateEvents(nextThread); + if (wres == 0) { wres = Event_Set(&nextThread->canWrite); + if (wres == 0) { wres = Event_Set(&nextThread->canRead); + if (wres == 0) { wres = ThreadFunc(nextThread); + if (wres != 0) + { + p->needContinue = False; + MtDec_CloseThreads(p); + }}}} + + // wres = 17; // for test + // wres = Event_Wait(&p->finishedEvent); + + sres = MY_SRes_HRESULT_FROM_WRes(wres); + + if (sres != 0) + p->threadingErrorSRes = sres; + + if ( + // wres == 0 + // wres != 0 + // || p->mtc.codeRes == SZ_ERROR_MEM + p->isAllocError + || p->threadingErrorSRes != SZ_OK + || p->overflow) + { + // p->needContinue = True; + } + else + p->needContinue = False; + + if (p->needContinue) + return SZ_OK; + + // if (sres != SZ_OK) + return sres; + // return E_FAIL; + } +} + +#endif diff --git a/src/third-party/lzma/MtDec.h b/src/third-party/lzma/MtDec.h new file mode 100644 index 000000000..9864cc874 --- /dev/null +++ b/src/third-party/lzma/MtDec.h @@ -0,0 +1,201 @@ +/* MtDec.h -- Multi-thread Decoder +2018-07-04 : Igor Pavlov : Public domain */ + +#ifndef __MT_DEC_H +#define __MT_DEC_H + +#include "7zTypes.h" + +#ifndef _7ZIP_ST +#include "Threads.h" +#endif + +EXTERN_C_BEGIN + +#ifndef _7ZIP_ST + +#ifndef _7ZIP_ST + #define MTDEC__THREADS_MAX 32 +#else + #define MTDEC__THREADS_MAX 1 +#endif + + +typedef struct +{ + ICompressProgress *progress; + SRes res; + UInt64 totalInSize; + UInt64 totalOutSize; + CCriticalSection cs; +} CMtProgress; + +void MtProgress_Init(CMtProgress *p, ICompressProgress *progress); +SRes MtProgress_Progress_ST(CMtProgress *p); +SRes MtProgress_ProgressAdd(CMtProgress *p, UInt64 inSize, UInt64 outSize); +SRes MtProgress_GetError(CMtProgress *p); +void MtProgress_SetError(CMtProgress *p, SRes res); + +struct _CMtDec; + +typedef struct +{ + struct _CMtDec *mtDec; + unsigned index; + void *inBuf; + + size_t inDataSize_Start; // size of input data in start block + UInt64 inDataSize; // total size of input data in all blocks + + CThread thread; + CAutoResetEvent canRead; + CAutoResetEvent canWrite; + void *allocaPtr; +} CMtDecThread; + +void MtDecThread_FreeInBufs(CMtDecThread *t); + + +typedef enum +{ + MTDEC_PARSE_CONTINUE, // continue this block with more input data + MTDEC_PARSE_OVERFLOW, // MT buffers overflow, need switch to single-thread + MTDEC_PARSE_NEW, // new block + MTDEC_PARSE_END // end of block threading. But we still can return to threading after Write(&needContinue) +} EMtDecParseState; + +typedef struct +{ + // in + int startCall; + const Byte *src; + size_t srcSize; + // in : (srcSize == 0) is allowed + // out : it's allowed to return less that actually was used ? + int srcFinished; + + // out + EMtDecParseState state; + BoolInt canCreateNewThread; + UInt64 outPos; // check it (size_t) +} CMtDecCallbackInfo; + + +typedef struct +{ + void (*Parse)(void *p, unsigned coderIndex, CMtDecCallbackInfo *ci); + + // PreCode() and Code(): + // (SRes_return_result != SZ_OK) means stop decoding, no need another blocks + SRes (*PreCode)(void *p, unsigned coderIndex); + SRes (*Code)(void *p, unsigned coderIndex, + const Byte *src, size_t srcSize, int srcFinished, + UInt64 *inCodePos, UInt64 *outCodePos, int *stop); + // stop - means stop another Code calls + + + /* Write() must be called, if Parse() was called + set (needWrite) if + { + && (was not interrupted by progress) + && (was not interrupted in previous block) + } + + out: + if (*needContinue), decoder still need to continue decoding with new iteration, + even after MTDEC_PARSE_END + if (*canRecode), we didn't flush current block data, so we still can decode current block later. + */ + SRes (*Write)(void *p, unsigned coderIndex, + BoolInt needWriteToStream, + const Byte *src, size_t srcSize, + // int srcFinished, + BoolInt *needContinue, + BoolInt *canRecode); +} IMtDecCallback; + + + +typedef struct _CMtDec +{ + /* input variables */ + + size_t inBufSize; /* size of input block */ + unsigned numThreadsMax; + // size_t inBlockMax; + unsigned numThreadsMax_2; + + ISeqInStream *inStream; + // const Byte *inData; + // size_t inDataSize; + + ICompressProgress *progress; + ISzAllocPtr alloc; + + IMtDecCallback *mtCallback; + void *mtCallbackObject; + + + /* internal variables */ + + size_t allocatedBufsSize; + + BoolInt exitThread; + WRes exitThreadWRes; + + UInt64 blockIndex; + BoolInt isAllocError; + BoolInt overflow; + SRes threadingErrorSRes; + + BoolInt needContinue; + + // CAutoResetEvent finishedEvent; + + SRes readRes; + SRes codeRes; + + BoolInt wasInterrupted; + + unsigned numStartedThreads_Limit; + unsigned numStartedThreads; + + Byte *crossBlock; + size_t crossStart; + size_t crossEnd; + UInt64 readProcessed; + BoolInt readWasFinished; + UInt64 inProcessed; + + unsigned filledThreadStart; + unsigned numFilledThreads; + + #ifndef _7ZIP_ST + BoolInt needInterrupt; + UInt64 interruptIndex; + CMtProgress mtProgress; + CMtDecThread threads[MTDEC__THREADS_MAX]; + #endif +} CMtDec; + + +void MtDec_Construct(CMtDec *p); +void MtDec_Destruct(CMtDec *p); + +/* +MtDec_Code() returns: + SZ_OK - in most cases + MY_SRes_HRESULT_FROM_WRes(WRes_error) - in case of unexpected error in threading function +*/ + +SRes MtDec_Code(CMtDec *p); +Byte *MtDec_GetCrossBuff(CMtDec *p); + +int MtDec_PrepareRead(CMtDec *p); +const Byte *MtDec_Read(CMtDec *p, size_t *inLim); + +#endif + +EXTERN_C_END + +#endif diff --git a/src/third-party/lzma/Ppmd.h b/src/third-party/lzma/Ppmd.h index 25ce2d28f..4b9941521 100644 --- a/src/third-party/lzma/Ppmd.h +++ b/src/third-party/lzma/Ppmd.h @@ -1,5 +1,5 @@ /* Ppmd.h -- PPMD codec common code -2013-01-18 : Igor Pavlov : Public domain +2017-04-03 : Igor Pavlov : Public domain This code is based on PPMd var.H (2001): Dmitry Shkarin : Public domain */ #ifndef __PPMD_H @@ -77,8 +77,8 @@ typedef CPpmd_Byte_Ref; #define PPMD_SetAllBitsIn256Bytes(p) \ - { unsigned i; for (i = 0; i < 256 / sizeof(p[0]); i += 8) { \ - p[i+7] = p[i+6] = p[i+5] = p[i+4] = p[i+3] = p[i+2] = p[i+1] = p[i+0] = ~(size_t)0; }} + { size_t z; for (z = 0; z < 256 / sizeof(p[0]); z += 8) { \ + p[z+7] = p[z+6] = p[z+5] = p[z+4] = p[z+3] = p[z+2] = p[z+1] = p[z+0] = ~(size_t)0; }} EXTERN_C_END diff --git a/src/third-party/lzma/Ppmd7.c b/src/third-party/lzma/Ppmd7.c index 735651f5d..80e7de9a6 100644 --- a/src/third-party/lzma/Ppmd7.c +++ b/src/third-party/lzma/Ppmd7.c @@ -1,5 +1,5 @@ /* Ppmd7.c -- PPMdH codec -2015-09-28 : Igor Pavlov : Public domain +2018-07-04 : Igor Pavlov : Public domain This code is based on PPMd var.H (2001): Dmitry Shkarin : Public domain */ #include "Precomp.h" @@ -15,7 +15,7 @@ static const UInt16 kInitBinEsc[] = { 0x3CDD, 0x1F3F, 0x59BF, 0x48F3, 0x64A1, 0x #define UNIT_SIZE 12 #define U2B(nu) ((UInt32)(nu) * UNIT_SIZE) -#define U2I(nu) (p->Units2Indx[(nu) - 1]) +#define U2I(nu) (p->Units2Indx[(size_t)(nu) - 1]) #define I2U(indx) (p->Indx2Units[indx]) #ifdef PPMD_32BIT @@ -88,29 +88,31 @@ void Ppmd7_Construct(CPpmd7 *p) memset(p->HB2Flag + 0x40, 8, 0x100 - 0x40); } -void Ppmd7_Free(CPpmd7 *p, ISzAlloc *alloc) +void Ppmd7_Free(CPpmd7 *p, ISzAllocPtr alloc) { - alloc->Free(alloc, p->Base); + ISzAlloc_Free(alloc, p->Base); p->Size = 0; p->Base = 0; } -Bool Ppmd7_Alloc(CPpmd7 *p, UInt32 size, ISzAlloc *alloc) +BoolInt Ppmd7_Alloc(CPpmd7 *p, UInt32 size, ISzAllocPtr alloc) { - if (p->Base == 0 || p->Size != size) + if (!p->Base || p->Size != size) { + size_t size2; Ppmd7_Free(p, alloc); + size2 = 0 + #ifndef PPMD_32BIT + + UNIT_SIZE + #endif + ; p->AlignOffset = #ifdef PPMD_32BIT (4 - size) & 3; #else 4 - (size & 3); #endif - if ((p->Base = (Byte *)alloc->Alloc(alloc, p->AlignOffset + size - #ifndef PPMD_32BIT - + UNIT_SIZE - #endif - )) == 0) + if ((p->Base = (Byte *)ISzAlloc_Alloc(alloc, p->AlignOffset + size + size2)) == 0) return False; p->Size = size; } @@ -340,7 +342,7 @@ void Ppmd7_Init(CPpmd7 *p, unsigned maxOrder) p->DummySee.Count = 64; /* unused */ } -static CTX_PTR CreateSuccessors(CPpmd7 *p, Bool skip) +static CTX_PTR CreateSuccessors(CPpmd7 *p, BoolInt skip) { CPpmd_State upState; CTX_PTR c = p->MinContext; @@ -513,7 +515,7 @@ static void UpdateModel(CPpmd7 *p) /* Expand for one UNIT */ unsigned oldNU = ns1 >> 1; unsigned i = U2I(oldNU); - if (i != U2I(oldNU + 1)) + if (i != U2I((size_t)oldNU + 1)) { void *ptr = AllocUnits(p, i + 1); void *oldPtr; @@ -639,10 +641,10 @@ CPpmd_See *Ppmd7_MakeEscFreq(CPpmd7 *p, unsigned numMasked, UInt32 *escFreq) unsigned nonMasked = p->MinContext->NumStats - numMasked; if (p->MinContext->NumStats != 256) { - see = p->See[p->NS2Indx[nonMasked - 1]] + + see = p->See[(unsigned)p->NS2Indx[(size_t)nonMasked - 1]] + (nonMasked < (unsigned)SUFFIX(p->MinContext)->NumStats - p->MinContext->NumStats) + - 2 * (p->MinContext->SummFreq < 11 * p->MinContext->NumStats) + - 4 * (numMasked > nonMasked) + + 2 * (unsigned)(p->MinContext->SummFreq < 11 * p->MinContext->NumStats) + + 4 * (unsigned)(numMasked > nonMasked) + p->HiBitsFlag; { unsigned r = (see->Summ >> see->Shift); diff --git a/src/third-party/lzma/Ppmd7.h b/src/third-party/lzma/Ppmd7.h index 56e81eb12..cce93f120 100644 --- a/src/third-party/lzma/Ppmd7.h +++ b/src/third-party/lzma/Ppmd7.h @@ -1,5 +1,5 @@ /* Ppmd7.h -- PPMdH compression codec -2010-03-12 : Igor Pavlov : Public domain +2018-07-04 : Igor Pavlov : Public domain This code is based on PPMd var.H (2001): Dmitry Shkarin : Public domain */ /* This code supports virtual RangeDecoder and includes the implementation @@ -60,8 +60,8 @@ typedef struct } CPpmd7; void Ppmd7_Construct(CPpmd7 *p); -Bool Ppmd7_Alloc(CPpmd7 *p, UInt32 size, ISzAlloc *alloc); -void Ppmd7_Free(CPpmd7 *p, ISzAlloc *alloc); +BoolInt Ppmd7_Alloc(CPpmd7 *p, UInt32 size, ISzAllocPtr alloc); +void Ppmd7_Free(CPpmd7 *p, ISzAllocPtr alloc); void Ppmd7_Init(CPpmd7 *p, unsigned maxOrder); #define Ppmd7_WasAllocated(p) ((p)->Base != NULL) @@ -86,10 +86,10 @@ void Ppmd7_Update2(CPpmd7 *p); void Ppmd7_UpdateBin(CPpmd7 *p); #define Ppmd7_GetBinSumm(p) \ - &p->BinSumm[Ppmd7Context_OneState(p->MinContext)->Freq - 1][p->PrevSuccess + \ - p->NS2BSIndx[Ppmd7_GetContext(p, p->MinContext->Suffix)->NumStats - 1] + \ + &p->BinSumm[(size_t)(unsigned)Ppmd7Context_OneState(p->MinContext)->Freq - 1][p->PrevSuccess + \ + p->NS2BSIndx[(size_t)Ppmd7_GetContext(p, p->MinContext->Suffix)->NumStats - 1] + \ (p->HiBitsFlag = p->HB2Flag[p->FoundState->Symbol]) + \ - 2 * p->HB2Flag[Ppmd7Context_OneState(p->MinContext)->Symbol] + \ + 2 * p->HB2Flag[(unsigned)Ppmd7Context_OneState(p->MinContext)->Symbol] + \ ((p->RunLength >> 26) & 0x20)] CPpmd_See *Ppmd7_MakeEscFreq(CPpmd7 *p, unsigned numMasked, UInt32 *scale); @@ -97,26 +97,28 @@ CPpmd_See *Ppmd7_MakeEscFreq(CPpmd7 *p, unsigned numMasked, UInt32 *scale); /* ---------- Decode ---------- */ -typedef struct +typedef struct IPpmd7_RangeDec IPpmd7_RangeDec; + +struct IPpmd7_RangeDec { - UInt32 (*GetThreshold)(void *p, UInt32 total); - void (*Decode)(void *p, UInt32 start, UInt32 size); - UInt32 (*DecodeBit)(void *p, UInt32 size0); -} IPpmd7_RangeDec; + UInt32 (*GetThreshold)(const IPpmd7_RangeDec *p, UInt32 total); + void (*Decode)(const IPpmd7_RangeDec *p, UInt32 start, UInt32 size); + UInt32 (*DecodeBit)(const IPpmd7_RangeDec *p, UInt32 size0); +}; typedef struct { - IPpmd7_RangeDec p; + IPpmd7_RangeDec vt; UInt32 Range; UInt32 Code; IByteIn *Stream; } CPpmd7z_RangeDec; void Ppmd7z_RangeDec_CreateVTable(CPpmd7z_RangeDec *p); -Bool Ppmd7z_RangeDec_Init(CPpmd7z_RangeDec *p); +BoolInt Ppmd7z_RangeDec_Init(CPpmd7z_RangeDec *p); #define Ppmd7z_RangeDec_IsFinishedOK(p) ((p)->Code == 0) -int Ppmd7_DecodeSymbol(CPpmd7 *p, IPpmd7_RangeDec *rc); +int Ppmd7_DecodeSymbol(CPpmd7 *p, const IPpmd7_RangeDec *rc); /* ---------- Encode ---------- */ diff --git a/src/third-party/lzma/Ppmd7Dec.c b/src/third-party/lzma/Ppmd7Dec.c index 3d01d7644..202640710 100644 --- a/src/third-party/lzma/Ppmd7Dec.c +++ b/src/third-party/lzma/Ppmd7Dec.c @@ -1,5 +1,5 @@ /* Ppmd7Dec.c -- PPMdH Decoder -2010-03-12 : Igor Pavlov : Public domain +2018-07-04 : Igor Pavlov : Public domain This code is based on PPMd var.H (2001): Dmitry Shkarin : Public domain */ #include "Precomp.h" @@ -8,49 +8,51 @@ This code is based on PPMd var.H (2001): Dmitry Shkarin : Public domain */ #define kTopValue (1 << 24) -Bool Ppmd7z_RangeDec_Init(CPpmd7z_RangeDec *p) +BoolInt Ppmd7z_RangeDec_Init(CPpmd7z_RangeDec *p) { unsigned i; p->Code = 0; p->Range = 0xFFFFFFFF; - if (p->Stream->Read((void *)p->Stream) != 0) + if (IByteIn_Read(p->Stream) != 0) return False; for (i = 0; i < 4; i++) - p->Code = (p->Code << 8) | p->Stream->Read((void *)p->Stream); + p->Code = (p->Code << 8) | IByteIn_Read(p->Stream); return (p->Code < 0xFFFFFFFF); } -static UInt32 Range_GetThreshold(void *pp, UInt32 total) +#define GET_Ppmd7z_RangeDec CPpmd7z_RangeDec *p = CONTAINER_FROM_VTBL(pp, CPpmd7z_RangeDec, vt); + +static UInt32 Range_GetThreshold(const IPpmd7_RangeDec *pp, UInt32 total) { - CPpmd7z_RangeDec *p = (CPpmd7z_RangeDec *)pp; - return (p->Code) / (p->Range /= total); + GET_Ppmd7z_RangeDec + return p->Code / (p->Range /= total); } static void Range_Normalize(CPpmd7z_RangeDec *p) { if (p->Range < kTopValue) { - p->Code = (p->Code << 8) | p->Stream->Read((void *)p->Stream); + p->Code = (p->Code << 8) | IByteIn_Read(p->Stream); p->Range <<= 8; if (p->Range < kTopValue) { - p->Code = (p->Code << 8) | p->Stream->Read((void *)p->Stream); + p->Code = (p->Code << 8) | IByteIn_Read(p->Stream); p->Range <<= 8; } } } -static void Range_Decode(void *pp, UInt32 start, UInt32 size) +static void Range_Decode(const IPpmd7_RangeDec *pp, UInt32 start, UInt32 size) { - CPpmd7z_RangeDec *p = (CPpmd7z_RangeDec *)pp; + GET_Ppmd7z_RangeDec p->Code -= start * p->Range; p->Range *= size; Range_Normalize(p); } -static UInt32 Range_DecodeBit(void *pp, UInt32 size0) +static UInt32 Range_DecodeBit(const IPpmd7_RangeDec *pp, UInt32 size0) { - CPpmd7z_RangeDec *p = (CPpmd7z_RangeDec *)pp; + GET_Ppmd7z_RangeDec UInt32 newBound = (p->Range >> 14) * size0; UInt32 symbol; if (p->Code < newBound) @@ -70,15 +72,15 @@ static UInt32 Range_DecodeBit(void *pp, UInt32 size0) void Ppmd7z_RangeDec_CreateVTable(CPpmd7z_RangeDec *p) { - p->p.GetThreshold = Range_GetThreshold; - p->p.Decode = Range_Decode; - p->p.DecodeBit = Range_DecodeBit; + p->vt.GetThreshold = Range_GetThreshold; + p->vt.Decode = Range_Decode; + p->vt.DecodeBit = Range_DecodeBit; } #define MASK(sym) ((signed char *)charMask)[sym] -int Ppmd7_DecodeSymbol(CPpmd7 *p, IPpmd7_RangeDec *rc) +int Ppmd7_DecodeSymbol(CPpmd7 *p, const IPpmd7_RangeDec *rc) { size_t charMask[256 / sizeof(size_t)]; if (p->MinContext->NumStats != 1) diff --git a/src/third-party/lzma/Ppmd7Enc.c b/src/third-party/lzma/Ppmd7Enc.c index 9b49e5d44..a74d3002b 100644 --- a/src/third-party/lzma/Ppmd7Enc.c +++ b/src/third-party/lzma/Ppmd7Enc.c @@ -1,5 +1,5 @@ /* Ppmd7Enc.c -- PPMdH Encoder -2015-09-28 : Igor Pavlov : Public domain +2017-04-03 : Igor Pavlov : Public domain This code is based on PPMd var.H (2001): Dmitry Shkarin : Public domain */ #include "Precomp.h" @@ -23,7 +23,7 @@ static void RangeEnc_ShiftLow(CPpmd7z_RangeEnc *p) Byte temp = p->Cache; do { - p->Stream->Write(p->Stream, (Byte)(temp + (Byte)(p->Low >> 32))); + IByteOut_Write(p->Stream, (Byte)(temp + (Byte)(p->Low >> 32))); temp = 0xFF; } while (--p->CacheSize != 0); diff --git a/src/third-party/lzma/Sha256.c b/src/third-party/lzma/Sha256.c index 47e2f42b9..90994e5ab 100644 --- a/src/third-party/lzma/Sha256.c +++ b/src/third-party/lzma/Sha256.c @@ -1,5 +1,5 @@ /* Crypto/Sha256.c -- SHA-256 Hash -2015-11-14 : Igor Pavlov : Public domain +2017-04-03 : Igor Pavlov : Public domain This code is based on public domain code from Wei Dai's Crypto++ library. */ #include "Precomp.h" @@ -45,7 +45,7 @@ void Sha256_Init(CSha256 *p) #ifdef _SHA256_UNROLL2 #define R(a,b,c,d,e,f,g,h, i) \ - h += S1(e) + Ch(e,f,g) + K[(i)+(j)] + (j ? blk2(i) : blk0(i)); \ + h += S1(e) + Ch(e,f,g) + K[(i)+(size_t)(j)] + (j ? blk2(i) : blk0(i)); \ d += h; \ h += S0(a) + Maj(a, b, c) @@ -73,7 +73,7 @@ void Sha256_Init(CSha256 *p) #define h(i) T[(7-(i))&7] #define R(i) \ - h(i) += S1(e(i)) + Ch(e(i),f(i),g(i)) + K[(i)+(j)] + (j ? blk2(i) : blk0(i)); \ + h(i) += S1(e(i)) + Ch(e(i),f(i),g(i)) + K[(i)+(size_t)(j)] + (j ? blk2(i) : blk0(i)); \ d(i) += h(i); \ h(i) += S0(a(i)) + Maj(a(i), b(i), c(i)) \ diff --git a/src/third-party/lzma/Threads.c b/src/third-party/lzma/Threads.c index ece07e618..8fd86f224 100644 --- a/src/third-party/lzma/Threads.c +++ b/src/third-party/lzma/Threads.c @@ -1,5 +1,5 @@ /* Threads.c -- multithreading library -2014-09-21 : Igor Pavlov : Public domain */ +2017-06-26 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -12,18 +12,20 @@ static WRes GetError() { DWORD res = GetLastError(); - return (res) ? (WRes)(res) : 1; + return res ? (WRes)res : 1; } -WRes HandleToWRes(HANDLE h) { return (h != 0) ? 0 : GetError(); } -WRes BOOLToWRes(BOOL v) { return v ? 0 : GetError(); } +static WRes HandleToWRes(HANDLE h) { return (h != NULL) ? 0 : GetError(); } +static WRes BOOLToWRes(BOOL v) { return v ? 0 : GetError(); } WRes HandlePtr_Close(HANDLE *p) { if (*p != NULL) + { if (!CloseHandle(*p)) return GetError(); - *p = NULL; + *p = NULL; + } return 0; } @@ -49,7 +51,7 @@ WRes Thread_Create(CThread *p, THREAD_FUNC_TYPE func, LPVOID param) return HandleToWRes(*p); } -WRes Event_Create(CEvent *p, BOOL manualReset, int signaled) +static WRes Event_Create(CEvent *p, BOOL manualReset, int signaled) { *p = CreateEvent(NULL, manualReset, (signaled ? TRUE : FALSE), NULL); return HandleToWRes(*p); diff --git a/src/third-party/lzma/Threads.h b/src/third-party/lzma/Threads.h index e927208d7..f913241ae 100644 --- a/src/third-party/lzma/Threads.h +++ b/src/third-party/lzma/Threads.h @@ -1,5 +1,5 @@ /* Threads.h -- multithreading library -2013-11-12 : Igor Pavlov : Public domain */ +2017-06-18 : Igor Pavlov : Public domain */ #ifndef __7Z_THREADS_H #define __7Z_THREADS_H @@ -49,7 +49,8 @@ WRes AutoResetEvent_Create(CAutoResetEvent *p, int signaled); WRes AutoResetEvent_CreateNotSignaled(CAutoResetEvent *p); typedef HANDLE CSemaphore; -#define Semaphore_Construct(p) (*p) = NULL +#define Semaphore_Construct(p) *(p) = NULL +#define Semaphore_IsCreated(p) (*(p) != NULL) #define Semaphore_Close(p) HandlePtr_Close(p) #define Semaphore_Wait(p) Handle_WaitObject(*(p)) WRes Semaphore_Create(CSemaphore *p, UInt32 initCount, UInt32 maxCount); diff --git a/src/third-party/lzma/Util/7z/7zMain.c b/src/third-party/lzma/Util/7z/7zMain.c index 00c6dbb6b..1c02b48ec 100644 --- a/src/third-party/lzma/Util/7z/7zMain.c +++ b/src/third-party/lzma/Util/7z/7zMain.c @@ -1,11 +1,13 @@ /* 7zMain.c - Test application for 7z Decoder -2015-08-02 : Igor Pavlov : Public domain */ +2019-02-02 : Igor Pavlov : Public domain */ #include "Precomp.h" #include #include +#include "../../CpuArch.h" + #include "../../7z.h" #include "../../7zAlloc.h" #include "../../7zBuf.h" @@ -23,7 +25,17 @@ #endif #endif -static ISzAlloc g_Alloc = { SzAlloc, SzFree }; + +#define kInputBufSize ((size_t)1 << 18) + +static const ISzAlloc g_Alloc = { SzAlloc, SzFree }; + + +static void Print(const char *s) +{ + fputs(s, stdout); +} + static int Buf_EnsureSize(CBuf *dest, size_t size) { @@ -164,7 +176,7 @@ static SRes Utf16_To_Char(CBuf *buf, const UInt16 *s char defaultChar = '_'; BOOL defUsed; unsigned numChars = 0; - numChars = WideCharToMultiByte(codePage, 0, s, len, (char *)buf->data, size, &defaultChar, &defUsed); + numChars = WideCharToMultiByte(codePage, 0, (LPCWSTR)s, len, (char *)buf->data, size, &defaultChar, &defUsed); if (numChars == 0 || numChars >= size) return SZ_ERROR_FAIL; buf->data[numChars] = 0; @@ -190,7 +202,7 @@ static WRes MyCreateDir(const UInt16 *name) { #ifdef USE_WINDOWS_FILE - return CreateDirectoryW(name, NULL) ? 0 : GetLastError(); + return CreateDirectoryW((LPCWSTR)name, NULL) ? 0 : GetLastError(); #else @@ -215,7 +227,7 @@ static WRes MyCreateDir(const UInt16 *name) static WRes OutFile_OpenUtf16(CSzFile *p, const UInt16 *name) { #ifdef USE_WINDOWS_FILE - return OutFile_OpenW(p, name); + return OutFile_OpenW(p, (LPCWSTR)name); #else CBuf buf; WRes res; @@ -227,6 +239,7 @@ static WRes OutFile_OpenUtf16(CSzFile *p, const UInt16 *name) #endif } + static SRes PrintString(const UInt16 *s) { CBuf buf; @@ -238,12 +251,12 @@ static SRes PrintString(const UInt16 *s) #endif ); if (res == SZ_OK) - fputs((const char *)buf.data, stdout); + Print((const char *)buf.data); Buf_Free(&buf, &g_Alloc); return res; } -static void UInt64ToStr(UInt64 value, char *s) +static void UInt64ToStr(UInt64 value, char *s, int numDigits) { char temp[32]; int pos = 0; @@ -253,6 +266,10 @@ static void UInt64ToStr(UInt64 value, char *s) value /= 10; } while (value != 0); + + for (numDigits -= pos; numDigits > 0; numDigits--) + *s++ = ' '; + do *s++ = temp[--pos]; while (pos); @@ -266,8 +283,10 @@ static char *UIntToStr(char *s, unsigned value, int numDigits) do temp[pos++] = (char)('0' + (value % 10)); while (value /= 10); + for (numDigits -= pos; numDigits > 0; numDigits--) *s++ = '0'; + do *s++ = temp[--pos]; while (pos); @@ -310,10 +329,10 @@ static void ConvertFileTimeToString(const CNtfsFileTime *nt, char *s) ms[1] = 29; for (mon = 0;; mon++) { - unsigned s = ms[mon]; - if (v < s) + unsigned d = ms[mon]; + if (v < d) break; - v -= s; + v -= d; } s = UIntToStr(s, year, 4); *s++ = '-'; UIntToStr_2(s, mon + 1); s[2] = '-'; s += 3; @@ -323,12 +342,19 @@ static void ConvertFileTimeToString(const CNtfsFileTime *nt, char *s) UIntToStr_2(s, sec); s[2] = 0; } -void PrintError(char *sz) +static void PrintLF() { - printf("\nERROR: %s\n", sz); + Print("\n"); } -static void GetAttribString(UInt32 wa, Bool isDir, char *s) +static void PrintError(char *s) +{ + Print("\nERROR: "); + Print(s); + PrintLF(); +} + +static void GetAttribString(UInt32 wa, BoolInt isDir, char *s) { #ifdef USE_WINDOWS_FILE s[0] = (char)(((wa & FILE_ATTRIBUTE_DIRECTORY) != 0 || isDir) ? 'D' : '.'); @@ -343,25 +369,27 @@ static void GetAttribString(UInt32 wa, Bool isDir, char *s) #endif } + // #define NUM_PARENTS_MAX 128 int MY_CDECL main(int numargs, char *args[]) { - CFileInStream archiveStream; - CLookToRead lookStream; - CSzArEx db; - SRes res; ISzAlloc allocImp; ISzAlloc allocTempImp; + + CFileInStream archiveStream; + CLookToRead2 lookStream; + CSzArEx db; + SRes res; UInt16 *temp = NULL; size_t tempSize = 0; // UInt32 parents[NUM_PARENTS_MAX]; - printf("\n7z ANSI-C Decoder " MY_VERSION_COPYRIGHT_DATE "\n\n"); + Print("\n7z Decoder " MY_VERSION_CPU " : " MY_COPYRIGHT_DATE "\n\n"); if (numargs == 1) { - printf( + Print( "Usage: 7zDec \n\n" "\n" " e: Extract files from archive (without using directory names)\n" @@ -370,7 +398,7 @@ int MY_CDECL main(int numargs, char *args[]) " x: eXtract files with full paths\n"); return 0; } - + if (numargs < 3) { PrintError("incorrect command"); @@ -381,11 +409,9 @@ int MY_CDECL main(int numargs, char *args[]) g_FileCodePage = AreFileApisANSI() ? CP_ACP : CP_OEMCP; #endif - allocImp.Alloc = SzAlloc; - allocImp.Free = SzFree; - allocTempImp.Alloc = SzAllocTemp; - allocTempImp.Free = SzFreeTemp; + allocImp = g_Alloc; + allocTempImp = g_Alloc; #ifdef UNDER_CE if (InFile_OpenW(&archiveStream.file, L"\test.7z")) @@ -398,16 +424,31 @@ int MY_CDECL main(int numargs, char *args[]) } FileInStream_CreateVTable(&archiveStream); - LookToRead_CreateVTable(&lookStream, False); - - lookStream.realStream = &archiveStream.s; - LookToRead_Init(&lookStream); + LookToRead2_CreateVTable(&lookStream, False); + lookStream.buf = NULL; + res = SZ_OK; + + { + lookStream.buf = (Byte *)ISzAlloc_Alloc(&allocImp, kInputBufSize); + if (!lookStream.buf) + res = SZ_ERROR_MEM; + else + { + lookStream.bufSize = kInputBufSize; + lookStream.realStream = &archiveStream.vt; + LookToRead2_Init(&lookStream); + } + } + CrcGenerateTable(); - + SzArEx_Init(&db); - - res = SzArEx_Open(&db, &lookStream.s, &allocImp, &allocTempImp); + + if (res == SZ_OK) + { + res = SzArEx_Open(&db, &lookStream.vt, &allocImp, &allocTempImp); + } if (res == SZ_OK) { @@ -477,7 +518,7 @@ int MY_CDECL main(int numargs, char *args[]) GetAttribString(SzBitWithVals_Check(&db.Attribs, i) ? db.Attribs.Vals[i] : 0, isDir, attr); fileSize = SzArEx_GetFileSize(&db, i); - UInt64ToStr(fileSize, s); + UInt64ToStr(fileSize, s, 10); if (SzBitWithVals_Check(&db.MTime, i)) ConvertFileTimeToString(&db.MTime.Vals[i], t); @@ -489,29 +530,33 @@ int MY_CDECL main(int numargs, char *args[]) t[j] = '\0'; } - printf("%s %s %10s ", t, attr, s); + Print(t); + Print(" "); + Print(attr); + Print(" "); + Print(s); + Print(" "); res = PrintString(temp); if (res != SZ_OK) break; if (isDir) - printf("/"); - printf("\n"); + Print("/"); + PrintLF(); continue; } - fputs(testCommand ? + Print(testCommand ? "Testing ": - "Extracting ", - stdout); + "Extracting "); res = PrintString(temp); if (res != SZ_OK) break; if (isDir) - printf("/"); + Print("/"); else { - res = SzArEx_Extract(&db, &lookStream.s, i, + res = SzArEx_Extract(&db, &lookStream.vt, i, &blockIndex, &outBuffer, &outBufferSize, &offset, &outSizeProcessed, &allocImp, &allocTempImp); @@ -543,7 +588,7 @@ int MY_CDECL main(int numargs, char *args[]) if (isDir) { MyCreateDir(destPath); - printf("\n"); + PrintLF(); continue; } else if (OutFile_OpenUtf16(&outFile, destPath)) @@ -561,6 +606,31 @@ int MY_CDECL main(int numargs, char *args[]) res = SZ_ERROR_FAIL; break; } + + #ifdef USE_WINDOWS_FILE + { + FILETIME mtime, ctime; + FILETIME *mtimePtr = NULL; + FILETIME *ctimePtr = NULL; + + if (SzBitWithVals_Check(&db.MTime, i)) + { + const CNtfsFileTime *t = &db.MTime.Vals[i]; + mtime.dwLowDateTime = (DWORD)(t->Low); + mtime.dwHighDateTime = (DWORD)(t->High); + mtimePtr = &mtime; + } + if (SzBitWithVals_Check(&db.CTime, i)) + { + const CNtfsFileTime *t = &db.CTime.Vals[i]; + ctime.dwLowDateTime = (DWORD)(t->Low); + ctime.dwHighDateTime = (DWORD)(t->High); + ctimePtr = &ctime; + } + if (mtimePtr || ctimePtr) + SetFileTime(outFile.handle, ctimePtr, NULL, mtimePtr); + } + #endif if (File_Close(&outFile)) { @@ -571,23 +641,31 @@ int MY_CDECL main(int numargs, char *args[]) #ifdef USE_WINDOWS_FILE if (SzBitWithVals_Check(&db.Attribs, i)) - SetFileAttributesW(destPath, db.Attribs.Vals[i]); + { + UInt32 attrib = db.Attribs.Vals[i]; + /* p7zip stores posix attributes in high 16 bits and adds 0x8000 as marker. + We remove posix bits, if we detect posix mode field */ + if ((attrib & 0xF0000000) != 0) + attrib &= 0x7FFF; + SetFileAttributesW((LPCWSTR)destPath, attrib); + } #endif } - printf("\n"); + PrintLF(); } - IAlloc_Free(&allocImp, outBuffer); + ISzAlloc_Free(&allocImp, outBuffer); } } - SzArEx_Free(&db, &allocImp); SzFree(NULL, temp); + SzArEx_Free(&db, &allocImp); + ISzAlloc_Free(&allocImp, lookStream.buf); File_Close(&archiveStream.file); if (res == SZ_OK) { - printf("\nEverything is Ok\n"); + Print("\nEverything is Ok\n"); return 0; } @@ -598,7 +676,11 @@ int MY_CDECL main(int numargs, char *args[]) else if (res == SZ_ERROR_CRC) PrintError("CRC error"); else - printf("\nERROR #%d\n", res); + { + char s[32]; + UInt64ToStr(res, s, 0); + PrintError(s); + } return 1; } diff --git a/src/third-party/lzma/Util/Lzma/LzmaUtil.c b/src/third-party/lzma/Util/Lzma/LzmaUtil.c index 98331b49d..82130e85d 100644 --- a/src/third-party/lzma/Util/Lzma/LzmaUtil.c +++ b/src/third-party/lzma/Util/Lzma/LzmaUtil.c @@ -1,5 +1,5 @@ /* LzmaUtil.c -- Test application for LZMA compression -2015-11-08 : Igor Pavlov : Public domain */ +2018-07-04 : Igor Pavlov : Public domain */ #include "../../Precomp.h" @@ -7,26 +7,29 @@ #include #include +#include "../../CpuArch.h" + #include "../../Alloc.h" #include "../../7zFile.h" #include "../../7zVersion.h" #include "../../LzmaDec.h" #include "../../LzmaEnc.h" -const char *kCantReadMessage = "Can not read input file"; -const char *kCantWriteMessage = "Can not write output file"; -const char *kCantAllocateMessage = "Can not allocate memory"; -const char *kDataErrorMessage = "Data error"; +static const char * const kCantReadMessage = "Can not read input file"; +static const char * const kCantWriteMessage = "Can not write output file"; +static const char * const kCantAllocateMessage = "Can not allocate memory"; +static const char * const kDataErrorMessage = "Data error"; -void PrintHelp(char *buffer) +static void PrintHelp(char *buffer) { - strcat(buffer, "\nLZMA Utility " MY_VERSION_COPYRIGHT_DATE "\n" - "\nUsage: lzma inputFile outputFile\n" - " e: encode file\n" - " d: decode file\n"); + strcat(buffer, + "\nLZMA-C " MY_VERSION_CPU " : " MY_COPYRIGHT_DATE "\n\n" + "Usage: lzma inputFile outputFile\n" + " e: encode file\n" + " d: decode file\n"); } -int PrintError(char *buffer, const char *message) +static int PrintError(char *buffer, const char *message) { strcat(buffer, "\nError: "); strcat(buffer, message); @@ -34,20 +37,22 @@ int PrintError(char *buffer, const char *message) return 1; } -int PrintErrorNumber(char *buffer, SRes val) +static int PrintErrorNumber(char *buffer, SRes val) { sprintf(buffer + strlen(buffer), "\nError code: %x\n", (unsigned)val); return 1; } -int PrintUserError(char *buffer) +static int PrintUserError(char *buffer) { return PrintError(buffer, "Incorrect command"); } + #define IN_BUF_SIZE (1 << 16) #define OUT_BUF_SIZE (1 << 16) + static SRes Decode2(CLzmaDec *state, ISeqOutStream *outStream, ISeqInStream *inStream, UInt64 unpackSize) { @@ -101,6 +106,7 @@ static SRes Decode2(CLzmaDec *state, ISeqOutStream *outStream, ISeqInStream *inS } } + static SRes Decode(ISeqOutStream *outStream, ISeqInStream *inStream) { UInt64 unpackSize; @@ -163,14 +169,15 @@ static SRes Encode(ISeqOutStream *outStream, ISeqInStream *inStream, UInt64 file return res; } -int main2(int numArgs, const char *args[], char *rs) + +static int main2(int numArgs, const char *args[], char *rs) { CFileSeqInStream inStream; CFileOutStream outStream; char c; int res; int encodeMode; - Bool useOutFile = False; + BoolInt useOutFile = False; FileSeqInStream_CreateVTable(&inStream); File_Construct(&inStream.file); @@ -215,11 +222,11 @@ int main2(int numArgs, const char *args[], char *rs) { UInt64 fileSize; File_GetLength(&inStream.file, &fileSize); - res = Encode(&outStream.s, &inStream.s, fileSize, rs); + res = Encode(&outStream.vt, &inStream.vt, fileSize, rs); } else { - res = Decode(&outStream.s, useOutFile ? &inStream.s : NULL); + res = Decode(&outStream.vt, useOutFile ? &inStream.vt : NULL); } if (useOutFile) @@ -241,6 +248,7 @@ int main2(int numArgs, const char *args[], char *rs) return 0; } + int MY_CDECL main(int numArgs, const char *args[]) { char rs[800] = { 0 }; diff --git a/src/third-party/lzma/Util/Lzma/LzmaUtil.dsp b/src/third-party/lzma/Util/Lzma/LzmaUtil.dsp index df1f0cb00..eedde07d8 100644 --- a/src/third-party/lzma/Util/Lzma/LzmaUtil.dsp +++ b/src/third-party/lzma/Util/Lzma/LzmaUtil.dsp @@ -98,6 +98,10 @@ SOURCE=..\..\7zStream.c # End Source File # Begin Source File +SOURCE=..\..\7zTypes.h +# End Source File +# Begin Source File + SOURCE=..\..\7zVersion.h # End Source File # Begin Source File @@ -160,9 +164,5 @@ SOURCE=..\..\Threads.c SOURCE=..\..\Threads.h # End Source File -# Begin Source File - -SOURCE=..\..\Types.h -# End Source File # End Target # End Project diff --git a/src/third-party/lzma/Util/SfxSetup/SfxSetup.c b/src/third-party/lzma/Util/SfxSetup/SfxSetup.c index b7ad5caf5..394369abd 100644 --- a/src/third-party/lzma/Util/SfxSetup/SfxSetup.c +++ b/src/third-party/lzma/Util/SfxSetup/SfxSetup.c @@ -1,5 +1,5 @@ /* SfxSetup.c - 7z SFX Setup -2015-11-08 : Igor Pavlov : Public domain */ +2019-02-02 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -20,9 +20,12 @@ #include "../../7zCrc.h" #include "../../7zFile.h" #include "../../CpuArch.h" +#include "../../DllSecur.h" #define k_EXE_ExtIndex 2 +#define kInputBufSize ((size_t)1 << 18) + static const char * const kExts[] = { "bat" @@ -124,7 +127,7 @@ static WRes MyCreateDir(const WCHAR *name) #define kSignatureSearchLimit (1 << 22) -static Bool FindSignature(CSzFile *stream, UInt64 *resPos) +static BoolInt FindSignature(CSzFile *stream, UInt64 *resPos) { Byte buf[kBufferSize]; size_t numPrevBytes = 0; @@ -160,7 +163,7 @@ static Bool FindSignature(CSzFile *stream, UInt64 *resPos) } } -static Bool DoesFileOrDirExist(const WCHAR *path) +static BoolInt DoesFileOrDirExist(const WCHAR *path) { WIN32_FIND_DATAW fd; HANDLE handle; @@ -237,7 +240,7 @@ int APIENTRY WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, #endif { CFileInStream archiveStream; - CLookToRead lookStream; + CLookToRead2 lookStream; CSzArEx db; SRes res = SZ_OK; ISzAlloc allocImp; @@ -251,9 +254,11 @@ int APIENTRY WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, DWORD winRes; const wchar_t *cmdLineParams; const char *errorMessage = NULL; - Bool useShellExecute = True; + BoolInt useShellExecute = True; DWORD exitCode = 0; + LoadSecurityDlls(); + #ifdef _CONSOLE SetConsoleCtrlHandler(HandlerRoutine, TRUE); #else @@ -272,7 +277,8 @@ int APIENTRY WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, allocTempImp.Free = SzFreeTemp; FileInStream_CreateVTable(&archiveStream); - LookToRead_CreateVTable(&lookStream, False); + LookToRead2_CreateVTable(&lookStream, False); + lookStream.buf = NULL; winRes = GetModuleFileNameW(NULL, sfxPath, MAX_PATH); if (winRes == 0 || winRes > MAX_PATH) @@ -281,7 +287,7 @@ int APIENTRY WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, cmdLineParams = GetCommandLineW(); #ifndef UNDER_CE { - Bool quoteMode = False; + BoolInt quoteMode = False; for (;; cmdLineParams++) { wchar_t c = *cmdLineParams; @@ -373,14 +379,22 @@ int APIENTRY WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, if (res == SZ_OK) { - lookStream.realStream = &archiveStream.s; - LookToRead_Init(&lookStream); + lookStream.buf = (Byte *)ISzAlloc_Alloc(&allocImp, kInputBufSize); + if (!lookStream.buf) + res = SZ_ERROR_MEM; + else + { + lookStream.bufSize = kInputBufSize; + lookStream.realStream = &archiveStream.vt; + LookToRead2_Init(&lookStream); + } } SzArEx_Init(&db); + if (res == SZ_OK) { - res = SzArEx_Open(&db, &lookStream.s, &allocImp, &allocTempImp); + res = SzArEx_Open(&db, &lookStream.vt, &allocImp, &allocTempImp); } if (res == SZ_OK) @@ -396,11 +410,9 @@ int APIENTRY WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, { size_t offset = 0; size_t outSizeProcessed = 0; - size_t len; WCHAR *temp; - len = SzArEx_GetFileNameUtf16(&db, i, NULL); - - if (len >= MAX_PATH) + + if (SzArEx_GetFileNameUtf16(&db, i, NULL) >= MAX_PATH) { res = SZ_ERROR_FAIL; break; @@ -408,9 +420,9 @@ int APIENTRY WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, temp = path + pathLen; - SzArEx_GetFileNameUtf16(&db, i, temp); + SzArEx_GetFileNameUtf16(&db, i, (UInt16 *)temp); { - res = SzArEx_Extract(&db, &lookStream.s, i, + res = SzArEx_Extract(&db, &lookStream.vt, i, &blockIndex, &outBuffer, &outBufferSize, &offset, &outSizeProcessed, &allocImp, &allocTempImp); @@ -515,16 +527,19 @@ int APIENTRY WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, { WCHAR *temp = path + pathLen; UInt32 j; - SzArEx_GetFileNameUtf16(&db, executeFileIndex, temp); + SzArEx_GetFileNameUtf16(&db, executeFileIndex, (UInt16 *)temp); for (j = 0; temp[j] != 0; j++) if (temp[j] == '/') temp[j] = CHAR_PATH_SEPARATOR; } } - IAlloc_Free(&allocImp, outBuffer); + ISzAlloc_Free(&allocImp, outBuffer); } + SzArEx_Free(&db, &allocImp); + ISzAlloc_Free(&allocImp, lookStream.buf); + File_Close(&archiveStream.file); if (res == SZ_OK) diff --git a/src/third-party/lzma/Util/SfxSetup/SfxSetup.dsp b/src/third-party/lzma/Util/SfxSetup/SfxSetup.dsp index 931c537ba..be9de6dec 100644 --- a/src/third-party/lzma/Util/SfxSetup/SfxSetup.dsp +++ b/src/third-party/lzma/Util/SfxSetup/SfxSetup.dsp @@ -187,6 +187,14 @@ SOURCE=..\..\Delta.h # End Source File # Begin Source File +SOURCE=..\..\DllSecur.c +# End Source File +# Begin Source File + +SOURCE=..\..\DllSecur.h +# End Source File +# Begin Source File + SOURCE=..\..\Lzma2Dec.c # End Source File # Begin Source File diff --git a/src/third-party/lzma/Util/SfxSetup/makefile b/src/third-party/lzma/Util/SfxSetup/makefile index 490b51d9e..c9f59ccd3 100644 --- a/src/third-party/lzma/Util/SfxSetup/makefile +++ b/src/third-party/lzma/Util/SfxSetup/makefile @@ -1,4 +1,5 @@ PROG = 7zS2.sfx +MY_FIXED = 1 C_OBJS = \ $O\7zAlloc.obj \ @@ -16,6 +17,7 @@ C_OBJS = \ $O\BraIA64.obj \ $O\CpuArch.obj \ $O\Delta.obj \ + $O\DllSecur.obj \ $O\Lzma2Dec.obj \ $O\LzmaDec.obj \ diff --git a/src/third-party/lzma/Util/SfxSetup/makefile_con b/src/third-party/lzma/Util/SfxSetup/makefile_con index 173bdbf1d..6f604ed81 100644 --- a/src/third-party/lzma/Util/SfxSetup/makefile_con +++ b/src/third-party/lzma/Util/SfxSetup/makefile_con @@ -1,4 +1,5 @@ PROG = 7zS2con.sfx +MY_FIXED = 1 CFLAGS = $(CFLAGS) -D_CONSOLE C_OBJS = \ @@ -17,6 +18,7 @@ C_OBJS = \ $O\BraIA64.obj \ $O\CpuArch.obj \ $O\Delta.obj \ + $O\DllSecur.obj \ $O\Lzma2Dec.obj \ $O\LzmaDec.obj \ diff --git a/src/third-party/lzma/Xz.c b/src/third-party/lzma/Xz.c index 92144db18..7e061d6e7 100644 --- a/src/third-party/lzma/Xz.c +++ b/src/third-party/lzma/Xz.c @@ -1,5 +1,5 @@ /* Xz.c - Xz -2015-05-01 : Igor Pavlov : Public domain */ +2017-05-12 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -9,7 +9,7 @@ #include "XzCrc64.h" const Byte XZ_SIG[XZ_SIG_SIZE] = { 0xFD, '7', 'z', 'X', 'Z', 0 }; -const Byte XZ_FOOTER_SIG[XZ_FOOTER_SIG_SIZE] = { 'Y', 'Z' }; +/* const Byte XZ_FOOTER_SIG[XZ_FOOTER_SIG_SIZE] = { 'Y', 'Z' }; */ unsigned Xz_WriteVarInt(Byte *buf, UInt64 v) { @@ -20,22 +20,22 @@ unsigned Xz_WriteVarInt(Byte *buf, UInt64 v) v >>= 7; } while (v != 0); - buf[i - 1] &= 0x7F; + buf[(size_t)i - 1] &= 0x7F; return i; } void Xz_Construct(CXzStream *p) { - p->numBlocks = p->numBlocksAllocated = 0; - p->blocks = 0; + p->numBlocks = 0; + p->blocks = NULL; p->flags = 0; } -void Xz_Free(CXzStream *p, ISzAlloc *alloc) +void Xz_Free(CXzStream *p, ISzAllocPtr alloc) { - alloc->Free(alloc, p->blocks); - p->numBlocks = p->numBlocksAllocated = 0; - p->blocks = 0; + ISzAlloc_Free(alloc, p->blocks); + p->numBlocks = 0; + p->blocks = NULL; } unsigned XzFlags_GetCheckSize(CXzStreamFlags f) diff --git a/src/third-party/lzma/Xz.h b/src/third-party/lzma/Xz.h index 6937beaf6..fad56a3fb 100644 --- a/src/third-party/lzma/Xz.h +++ b/src/third-party/lzma/Xz.h @@ -1,5 +1,5 @@ /* Xz.h - Xz interface -2015-05-01 : Igor Pavlov : Public domain */ +2018-07-04 : Igor Pavlov : Public domain */ #ifndef __XZ_H #define __XZ_H @@ -50,9 +50,10 @@ typedef struct #define XzBlock_GetNumFilters(p) (((p)->flags & XZ_BF_NUM_FILTERS_MASK) + 1) #define XzBlock_HasPackSize(p) (((p)->flags & XZ_BF_PACK_SIZE) != 0) #define XzBlock_HasUnpackSize(p) (((p)->flags & XZ_BF_UNPACK_SIZE) != 0) +#define XzBlock_HasUnsupportedFlags(p) (((p)->flags & ~(XZ_BF_NUM_FILTERS_MASK | XZ_BF_PACK_SIZE | XZ_BF_UNPACK_SIZE)) != 0) SRes XzBlock_Parse(CXzBlock *p, const Byte *header); -SRes XzBlock_ReadHeader(CXzBlock *p, ISeqInStream *inStream, Bool *isIndex, UInt32 *headerSizeRes); +SRes XzBlock_ReadHeader(CXzBlock *p, ISeqInStream *inStream, BoolInt *isIndex, UInt32 *headerSizeRes); /* ---------- xz stream ---------- */ @@ -60,7 +61,13 @@ SRes XzBlock_ReadHeader(CXzBlock *p, ISeqInStream *inStream, Bool *isIndex, UInt #define XZ_FOOTER_SIG_SIZE 2 extern const Byte XZ_SIG[XZ_SIG_SIZE]; + +/* extern const Byte XZ_FOOTER_SIG[XZ_FOOTER_SIG_SIZE]; +*/ + +#define XZ_FOOTER_SIG_0 'Y' +#define XZ_FOOTER_SIG_1 'Z' #define XZ_STREAM_FLAGS_SIZE 2 #define XZ_STREAM_CRC_SIZE 4 @@ -106,13 +113,12 @@ typedef struct { CXzStreamFlags flags; size_t numBlocks; - size_t numBlocksAllocated; CXzBlockSizes *blocks; UInt64 startOffset; } CXzStream; void Xz_Construct(CXzStream *p); -void Xz_Free(CXzStream *p, ISzAlloc *alloc); +void Xz_Free(CXzStream *p, ISzAllocPtr alloc); #define XZ_SIZE_OVERFLOW ((UInt64)(Int64)-1) @@ -127,12 +133,15 @@ typedef struct } CXzs; void Xzs_Construct(CXzs *p); -void Xzs_Free(CXzs *p, ISzAlloc *alloc); -SRes Xzs_ReadBackward(CXzs *p, ILookInStream *inStream, Int64 *startOffset, ICompressProgress *progress, ISzAlloc *alloc); +void Xzs_Free(CXzs *p, ISzAllocPtr alloc); +SRes Xzs_ReadBackward(CXzs *p, ILookInStream *inStream, Int64 *startOffset, ICompressProgress *progress, ISzAllocPtr alloc); UInt64 Xzs_GetNumBlocks(const CXzs *p); UInt64 Xzs_GetUnpackSize(const CXzs *p); + +// ECoderStatus values are identical to ELzmaStatus values of LZMA2 decoder + typedef enum { CODER_STATUS_NOT_SPECIFIED, /* use main error code instead */ @@ -141,43 +150,55 @@ typedef enum CODER_STATUS_NEEDS_MORE_INPUT /* you must provide more input bytes */ } ECoderStatus; + +// ECoderFinishMode values are identical to ELzmaFinishMode + typedef enum { CODER_FINISH_ANY, /* finish at any point */ CODER_FINISH_END /* block must be finished at the end */ } ECoderFinishMode; + typedef struct _IStateCoder { void *p; - void (*Free)(void *p, ISzAlloc *alloc); - SRes (*SetProps)(void *p, const Byte *props, size_t propSize, ISzAlloc *alloc); + void (*Free)(void *p, ISzAllocPtr alloc); + SRes (*SetProps)(void *p, const Byte *props, size_t propSize, ISzAllocPtr alloc); void (*Init)(void *p); - SRes (*Code)(void *p, Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen, - int srcWasFinished, ECoderFinishMode finishMode, int *wasFinished); + SRes (*Code2)(void *p, Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen, + int srcWasFinished, ECoderFinishMode finishMode, + // int *wasFinished, + ECoderStatus *status); + SizeT (*Filter)(void *p, Byte *data, SizeT size); } IStateCoder; + + #define MIXCODER_NUM_FILTERS_MAX 4 typedef struct { - ISzAlloc *alloc; + ISzAllocPtr alloc; Byte *buf; unsigned numCoders; + + Byte *outBuf; + size_t outBufSize; + size_t outWritten; // is equal to lzmaDecoder.dicPos (in outBuf mode) + BoolInt wasFinished; + SRes res; + ECoderStatus status; + // BoolInt SingleBufMode; + int finished[MIXCODER_NUM_FILTERS_MAX - 1]; size_t pos[MIXCODER_NUM_FILTERS_MAX - 1]; size_t size[MIXCODER_NUM_FILTERS_MAX - 1]; UInt64 ids[MIXCODER_NUM_FILTERS_MAX]; + SRes results[MIXCODER_NUM_FILTERS_MAX]; IStateCoder coders[MIXCODER_NUM_FILTERS_MAX]; } CMixCoder; -void MixCoder_Construct(CMixCoder *p, ISzAlloc *alloc); -void MixCoder_Free(CMixCoder *p); -void MixCoder_Init(CMixCoder *p); -SRes MixCoder_SetFromMethod(CMixCoder *p, unsigned coderIndex, UInt64 methodId); -SRes MixCoder_Code(CMixCoder *p, Byte *dest, SizeT *destLen, - const Byte *src, SizeT *srcLen, int srcWasFinished, - ECoderFinishMode finishMode, ECoderStatus *status); typedef enum { @@ -191,6 +212,7 @@ typedef enum XZ_STATE_BLOCK_FOOTER } EXzState; + typedef struct { EXzState state; @@ -204,7 +226,7 @@ typedef struct UInt64 packSize; UInt64 unpackSize; - UInt64 numBlocks; + UInt64 numBlocks; // number of finished blocks in current stream UInt64 indexSize; UInt64 indexPos; UInt64 padSize; @@ -218,14 +240,65 @@ typedef struct CXzBlock block; CXzCheck check; CSha256 sha; + + BoolInt parseMode; + BoolInt headerParsedOk; + BoolInt decodeToStreamSignature; + unsigned decodeOnlyOneBlock; + + Byte *outBuf; + size_t outBufSize; + size_t outDataWritten; // the size of data in (outBuf) that were fully unpacked + Byte shaDigest[SHA256_DIGEST_SIZE]; Byte buf[XZ_BLOCK_HEADER_SIZE_MAX]; } CXzUnpacker; -void XzUnpacker_Construct(CXzUnpacker *p, ISzAlloc *alloc); +/* alloc : aligned for cache line allocation is better */ +void XzUnpacker_Construct(CXzUnpacker *p, ISzAllocPtr alloc); void XzUnpacker_Init(CXzUnpacker *p); +void XzUnpacker_SetOutBuf(CXzUnpacker *p, Byte *outBuf, size_t outBufSize); void XzUnpacker_Free(CXzUnpacker *p); +/* + XzUnpacker + The sequence for decoding functions: + { + XzUnpacker_Construct() + [Decoding_Calls] + XzUnpacker_Free() + } + + [Decoding_Calls] + + There are 3 types of interfaces for [Decoding_Calls] calls: + + Interface-1 : Partial output buffers: + { + XzUnpacker_Init() + for() + XzUnpacker_Code(); + } + + Interface-2 : Direct output buffer: + Use it, if you know exact size of decoded data, and you need + whole xz unpacked data in one output buffer. + xz unpacker doesn't allocate additional buffer for lzma2 dictionary in that mode. + { + XzUnpacker_Init() + XzUnpacker_SetOutBufMode(); // to set output buffer and size + for() + XzUnpacker_Code(); // (dest = NULL) in XzUnpacker_Code() + } + + Interface-3 : Direct output buffer : One call full decoding + It unpacks whole input buffer to output buffer in one call. + It uses Interface-2 internally. + { + XzUnpacker_CodeFull() + } +*/ + /* finishMode: It has meaning only if the decoding reaches output limit (*destLen). @@ -255,20 +328,132 @@ Returns: SRes XzUnpacker_Code(CXzUnpacker *p, Byte *dest, SizeT *destLen, - const Byte *src, SizeT *srcLen, ECoderFinishMode finishMode, - ECoderStatus *status); + const Byte *src, SizeT *srcLen, int srcFinished, + ECoderFinishMode finishMode, ECoderStatus *status); -Bool XzUnpacker_IsStreamWasFinished(CXzUnpacker *p); +SRes XzUnpacker_CodeFull(CXzUnpacker *p, Byte *dest, SizeT *destLen, + const Byte *src, SizeT *srcLen, + ECoderFinishMode finishMode, ECoderStatus *status); + +BoolInt XzUnpacker_IsStreamWasFinished(const CXzUnpacker *p); /* -Call XzUnpacker_GetExtraSize after XzUnpacker_Code function to detect real size of -xz stream in two cases: -XzUnpacker_Code() returns: +XzUnpacker_GetExtraSize() returns then number of uncofirmed bytes, + if it's in (XZ_STATE_STREAM_HEADER) state or in (XZ_STATE_STREAM_PADDING) state. +These bytes can be some bytes after xz archive, or +it can be start of new xz stream. + +Call XzUnpacker_GetExtraSize() after XzUnpacker_Code() function to detect real size of +xz stream in two cases, if XzUnpacker_Code() returns: res == SZ_OK && status == CODER_STATUS_NEEDS_MORE_INPUT res == SZ_ERROR_NO_ARCHIVE */ -UInt64 XzUnpacker_GetExtraSize(CXzUnpacker *p); +UInt64 XzUnpacker_GetExtraSize(const CXzUnpacker *p); + + +/* + for random block decoding: + XzUnpacker_Init(); + set CXzUnpacker::streamFlags + XzUnpacker_PrepareToRandomBlockDecoding() + loop + { + XzUnpacker_Code() + XzUnpacker_IsBlockFinished() + } +*/ + +void XzUnpacker_PrepareToRandomBlockDecoding(CXzUnpacker *p); +BoolInt XzUnpacker_IsBlockFinished(const CXzUnpacker *p); + +#define XzUnpacker_GetPackSizeForIndex(p) ((p)->packSize + (p)->blockHeaderSize + XzFlags_GetCheckSize((p)->streamFlags)) + + + +/* ---------- Multi Threading Decoding ---------- */ + + +typedef struct +{ + size_t inBufSize_ST; + size_t outStep_ST; + BoolInt ignoreErrors; + + #ifndef _7ZIP_ST + unsigned numThreads; + size_t inBufSize_MT; + size_t memUseMax; + #endif +} CXzDecMtProps; + +void XzDecMtProps_Init(CXzDecMtProps *p); + + +typedef void * CXzDecMtHandle; + +/* + alloc : XzDecMt uses CAlignOffsetAlloc for addresses allocated by (alloc). + allocMid : for big allocations, aligned allocation is better +*/ + +CXzDecMtHandle XzDecMt_Create(ISzAllocPtr alloc, ISzAllocPtr allocMid); +void XzDecMt_Destroy(CXzDecMtHandle p); + + +typedef struct +{ + Byte UnpackSize_Defined; + Byte NumStreams_Defined; + Byte NumBlocks_Defined; + + Byte DataAfterEnd; + Byte DecodingTruncated; // Decoding was Truncated, we need only partial output data + + UInt64 InSize; // pack size processed + UInt64 OutSize; + + UInt64 NumStreams; + UInt64 NumBlocks; + + SRes DecodeRes; + SRes ReadRes; + SRes ProgressRes; + SRes CombinedRes; + SRes CombinedRes_Type; + +} CXzStatInfo; + +void XzStatInfo_Clear(CXzStatInfo *p); + +/* +XzDecMt_Decode() +SRes: + SZ_OK - OK + SZ_ERROR_MEM - Memory allocation error + SZ_ERROR_NO_ARCHIVE - is not xz archive + SZ_ERROR_ARCHIVE - Headers error + SZ_ERROR_DATA - Data Error + SZ_ERROR_CRC - CRC Error + SZ_ERROR_INPUT_EOF - it needs more input data + SZ_ERROR_WRITE - ISeqOutStream error + (SZ_ERROR_READ) - ISeqInStream errors + (SZ_ERROR_PROGRESS) - ICompressProgress errors + // SZ_ERROR_THREAD - error in multi-threading functions + MY_SRes_HRESULT_FROM_WRes(WRes_error) - error in multi-threading function +*/ + +SRes XzDecMt_Decode(CXzDecMtHandle p, + const CXzDecMtProps *props, + const UInt64 *outDataSize, // NULL means undefined + int finishMode, // 0 - partial unpacking is allowed, 1 - xz stream(s) must be finished + ISeqOutStream *outStream, + // Byte *outBuf, size_t *outBufSize, + ISeqInStream *inStream, + // const Byte *inData, size_t inDataSize, + CXzStatInfo *stat, + int *isMT, // 0 means that ST (Single-Thread) version was used + ICompressProgress *progress); EXTERN_C_END diff --git a/src/third-party/lzma/XzCrc64.c b/src/third-party/lzma/XzCrc64.c index bd0cf5a37..e9ca9ec26 100644 --- a/src/third-party/lzma/XzCrc64.c +++ b/src/third-party/lzma/XzCrc64.c @@ -1,5 +1,5 @@ /* XzCrc64.c -- CRC64 calculation -2015-03-01 : Igor Pavlov : Public domain */ +2017-05-23 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -9,9 +9,9 @@ #define kCrc64Poly UINT64_CONST(0xC96C5795D7870F42) #ifdef MY_CPU_LE - #define CRC_NUM_TABLES 4 + #define CRC64_NUM_TABLES 4 #else - #define CRC_NUM_TABLES 5 + #define CRC64_NUM_TABLES 5 #define CRC_UINT64_SWAP(v) \ ((v >> 56) \ | ((v >> 40) & ((UInt64)0xFF << 8)) \ @@ -29,10 +29,10 @@ UInt64 MY_FAST_CALL XzCrc64UpdateT4(UInt64 v, const void *data, size_t size, const UInt64 *table); #endif -typedef UInt64 (MY_FAST_CALL *CRC_FUNC)(UInt64 v, const void *data, size_t size, const UInt64 *table); +typedef UInt64 (MY_FAST_CALL *CRC64_FUNC)(UInt64 v, const void *data, size_t size, const UInt64 *table); -static CRC_FUNC g_Crc64Update; -UInt64 g_Crc64Table[256 * CRC_NUM_TABLES]; +static CRC64_FUNC g_Crc64Update; +UInt64 g_Crc64Table[256 * CRC64_NUM_TABLES]; UInt64 MY_FAST_CALL Crc64Update(UInt64 v, const void *data, size_t size) { @@ -52,12 +52,12 @@ void MY_FAST_CALL Crc64GenerateTable() UInt64 r = i; unsigned j; for (j = 0; j < 8; j++) - r = (r >> 1) ^ (kCrc64Poly & ~((r & 1) - 1)); + r = (r >> 1) ^ (kCrc64Poly & ((UInt64)0 - (r & 1))); g_Crc64Table[i] = r; } - for (; i < 256 * CRC_NUM_TABLES; i++) + for (i = 256; i < 256 * CRC64_NUM_TABLES; i++) { - UInt64 r = g_Crc64Table[i - 256]; + UInt64 r = g_Crc64Table[(size_t)i - 256]; g_Crc64Table[i] = g_Crc64Table[r & 0xFF] ^ (r >> 8); } @@ -74,9 +74,9 @@ void MY_FAST_CALL Crc64GenerateTable() else #endif { - for (i = 256 * CRC_NUM_TABLES - 1; i >= 256; i--) + for (i = 256 * CRC64_NUM_TABLES - 1; i >= 256; i--) { - UInt64 x = g_Crc64Table[i - 256]; + UInt64 x = g_Crc64Table[(size_t)i - 256]; g_Crc64Table[i] = CRC_UINT64_SWAP(x); } g_Crc64Update = XzCrc64UpdateT1_BeT4; diff --git a/src/third-party/lzma/XzCrc64Opt.c b/src/third-party/lzma/XzCrc64Opt.c index 97d086e6a..9273465d4 100644 --- a/src/third-party/lzma/XzCrc64Opt.c +++ b/src/third-party/lzma/XzCrc64Opt.c @@ -1,5 +1,5 @@ /* XzCrc64Opt.c -- CRC64 calculation -2015-03-01 : Igor Pavlov : Public domain */ +2017-06-30 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -7,24 +7,24 @@ #ifndef MY_CPU_BE -#define CRC_UPDATE_BYTE_2(crc, b) (table[((crc) ^ (b)) & 0xFF] ^ ((crc) >> 8)) +#define CRC64_UPDATE_BYTE_2(crc, b) (table[((crc) ^ (b)) & 0xFF] ^ ((crc) >> 8)) UInt64 MY_FAST_CALL XzCrc64UpdateT4(UInt64 v, const void *data, size_t size, const UInt64 *table) { const Byte *p = (const Byte *)data; for (; size > 0 && ((unsigned)(ptrdiff_t)p & 3) != 0; size--, p++) - v = CRC_UPDATE_BYTE_2(v, *p); + v = CRC64_UPDATE_BYTE_2(v, *p); for (; size >= 4; size -= 4, p += 4) { UInt32 d = (UInt32)v ^ *(const UInt32 *)p; v = (v >> 32) - ^ table[0x300 + ((d ) & 0xFF)] - ^ table[0x200 + ((d >> 8) & 0xFF)] - ^ table[0x100 + ((d >> 16) & 0xFF)] - ^ table[0x000 + ((d >> 24))]; + ^ (table + 0x300)[((d ) & 0xFF)] + ^ (table + 0x200)[((d >> 8) & 0xFF)] + ^ (table + 0x100)[((d >> 16) & 0xFF)] + ^ (table + 0x000)[((d >> 24))]; } for (; size > 0; size--, p++) - v = CRC_UPDATE_BYTE_2(v, *p); + v = CRC64_UPDATE_BYTE_2(v, *p); return v; } @@ -43,7 +43,7 @@ UInt64 MY_FAST_CALL XzCrc64UpdateT4(UInt64 v, const void *data, size_t size, con | ((v << 40) & ((UInt64)0xFF << 48)) \ | ((v << 56))) -#define CRC_UPDATE_BYTE_2_BE(crc, b) (table[(Byte)((crc) >> 56) ^ (b)] ^ ((crc) << 8)) +#define CRC64_UPDATE_BYTE_2_BE(crc, b) (table[(Byte)((crc) >> 56) ^ (b)] ^ ((crc) << 8)) UInt64 MY_FAST_CALL XzCrc64UpdateT1_BeT4(UInt64 v, const void *data, size_t size, const UInt64 *table) { @@ -51,18 +51,18 @@ UInt64 MY_FAST_CALL XzCrc64UpdateT1_BeT4(UInt64 v, const void *data, size_t size table += 0x100; v = CRC_UINT64_SWAP(v); for (; size > 0 && ((unsigned)(ptrdiff_t)p & 3) != 0; size--, p++) - v = CRC_UPDATE_BYTE_2_BE(v, *p); + v = CRC64_UPDATE_BYTE_2_BE(v, *p); for (; size >= 4; size -= 4, p += 4) { UInt32 d = (UInt32)(v >> 32) ^ *(const UInt32 *)p; v = (v << 32) - ^ table[0x000 + ((d ) & 0xFF)] - ^ table[0x100 + ((d >> 8) & 0xFF)] - ^ table[0x200 + ((d >> 16) & 0xFF)] - ^ table[0x300 + ((d >> 24))]; + ^ (table + 0x000)[((d ) & 0xFF)] + ^ (table + 0x100)[((d >> 8) & 0xFF)] + ^ (table + 0x200)[((d >> 16) & 0xFF)] + ^ (table + 0x300)[((d >> 24))]; } for (; size > 0; size--, p++) - v = CRC_UPDATE_BYTE_2_BE(v, *p); + v = CRC64_UPDATE_BYTE_2_BE(v, *p); return CRC_UINT64_SWAP(v); } diff --git a/src/third-party/lzma/XzDec.c b/src/third-party/lzma/XzDec.c index 29831b0ca..4f5327207 100644 --- a/src/third-party/lzma/XzDec.c +++ b/src/third-party/lzma/XzDec.c @@ -1,14 +1,33 @@ /* XzDec.c -- Xz Decode -2015-11-09 : Igor Pavlov : Public domain */ +2019-02-02 : Igor Pavlov : Public domain */ #include "Precomp.h" +// #include + +// #define XZ_DUMP + /* #define XZ_DUMP */ #ifdef XZ_DUMP #include #endif +// #define SHOW_DEBUG_INFO + +#ifdef SHOW_DEBUG_INFO +#include +#endif + +#ifdef SHOW_DEBUG_INFO +#define PRF(x) x +#else +#define PRF(x) +#endif + +#define PRF_STR(s) PRF(printf("\n" s "\n")) +#define PRF_STR_INT(s, d) PRF(printf("\n" s " %d\n", (unsigned)d)) + #include #include @@ -19,16 +38,18 @@ #include "Delta.h" #include "Lzma2Dec.h" +// #define USE_SUBBLOCK + #ifdef USE_SUBBLOCK #include "Bcj3Dec.c" -#include "SbDec.c" +#include "SbDec.h" #endif #include "Xz.h" #define XZ_CHECK_SIZE_MAX 64 -#define CODER_BUF_SIZE (1 << 17) +#define CODER_BUF_SIZE ((size_t)1 << 17) unsigned Xz_ReadVarInt(const Byte *p, size_t maxSize, UInt64 *value) { @@ -56,8 +77,9 @@ typedef struct size_t bufConv; size_t bufTotal; - UInt32 methodId; int encodeMode; + + UInt32 methodId; UInt32 delta; UInt32 ip; UInt32 x86State; @@ -66,12 +88,12 @@ typedef struct Byte buf[BRA_BUF_SIZE]; } CBraState; -static void BraState_Free(void *pp, ISzAlloc *alloc) +static void BraState_Free(void *pp, ISzAllocPtr alloc) { - alloc->Free(alloc, pp); + ISzAlloc_Free(alloc, pp); } -static SRes BraState_SetProps(void *pp, const Byte *props, size_t propSize, ISzAlloc *alloc) +static SRes BraState_SetProps(void *pp, const Byte *props, size_t propSize, ISzAllocPtr alloc) { CBraState *p = ((CBraState *)pp); UNUSED_VAR(alloc); @@ -121,69 +143,84 @@ static void BraState_Init(void *pp) Delta_Init(p->deltaState); } -#define CASE_BRA_CONV(isa) case XZ_ID_ ## isa: p->bufConv = isa ## _Convert(p->buf, p->bufTotal, p->ip, p->encodeMode); break; -static SRes BraState_Code(void *pp, Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen, - int srcWasFinished, ECoderFinishMode finishMode, int *wasFinished) +#define CASE_BRA_CONV(isa) case XZ_ID_ ## isa: size = isa ## _Convert(data, size, p->ip, p->encodeMode); break; + +static SizeT BraState_Filter(void *pp, Byte *data, SizeT size) { CBraState *p = ((CBraState *)pp); - SizeT destLenOrig = *destLen; - SizeT srcLenOrig = *srcLen; + switch (p->methodId) + { + case XZ_ID_Delta: + if (p->encodeMode) + Delta_Encode(p->deltaState, p->delta, data, size); + else + Delta_Decode(p->deltaState, p->delta, data, size); + break; + case XZ_ID_X86: + size = x86_Convert(data, size, p->ip, &p->x86State, p->encodeMode); + break; + CASE_BRA_CONV(PPC) + CASE_BRA_CONV(IA64) + CASE_BRA_CONV(ARM) + CASE_BRA_CONV(ARMT) + CASE_BRA_CONV(SPARC) + } + p->ip += (UInt32)size; + return size; +} + + +static SRes BraState_Code2(void *pp, + Byte *dest, SizeT *destLen, + const Byte *src, SizeT *srcLen, int srcWasFinished, + ECoderFinishMode finishMode, + // int *wasFinished + ECoderStatus *status) +{ + CBraState *p = ((CBraState *)pp); + SizeT destRem = *destLen; + SizeT srcRem = *srcLen; UNUSED_VAR(finishMode); + *destLen = 0; *srcLen = 0; - *wasFinished = 0; - while (destLenOrig > 0) + // *wasFinished = False; + *status = CODER_STATUS_NOT_FINISHED; + + while (destRem > 0) { if (p->bufPos != p->bufConv) { - size_t curSize = p->bufConv - p->bufPos; - if (curSize > destLenOrig) - curSize = destLenOrig; - memcpy(dest, p->buf + p->bufPos, curSize); - p->bufPos += curSize; - *destLen += curSize; - dest += curSize; - destLenOrig -= curSize; + size_t size = p->bufConv - p->bufPos; + if (size > destRem) + size = destRem; + memcpy(dest, p->buf + p->bufPos, size); + p->bufPos += size; + *destLen += size; + dest += size; + destRem -= size; continue; } + p->bufTotal -= p->bufPos; memmove(p->buf, p->buf + p->bufPos, p->bufTotal); p->bufPos = 0; p->bufConv = 0; { - size_t curSize = BRA_BUF_SIZE - p->bufTotal; - if (curSize > srcLenOrig) - curSize = srcLenOrig; - memcpy(p->buf + p->bufTotal, src, curSize); - *srcLen += curSize; - src += curSize; - srcLenOrig -= curSize; - p->bufTotal += curSize; + size_t size = BRA_BUF_SIZE - p->bufTotal; + if (size > srcRem) + size = srcRem; + memcpy(p->buf + p->bufTotal, src, size); + *srcLen += size; + src += size; + srcRem -= size; + p->bufTotal += size; } if (p->bufTotal == 0) break; - switch (p->methodId) - { - case XZ_ID_Delta: - if (p->encodeMode) - Delta_Encode(p->deltaState, p->delta, p->buf, p->bufTotal); - else - Delta_Decode(p->deltaState, p->delta, p->buf, p->bufTotal); - p->bufConv = p->bufTotal; - break; - case XZ_ID_X86: - p->bufConv = x86_Convert(p->buf, p->bufTotal, p->ip, &p->x86State, p->encodeMode); - break; - CASE_BRA_CONV(PPC) - CASE_BRA_CONV(IA64) - CASE_BRA_CONV(ARM) - CASE_BRA_CONV(ARMT) - CASE_BRA_CONV(SPARC) - default: - return SZ_ERROR_UNSUPPORTED; - } - p->ip += (UInt32)p->bufConv; + + p->bufConv = BraState_Filter(pp, p->buf, p->bufTotal); if (p->bufConv == 0) { @@ -192,48 +229,54 @@ static SRes BraState_Code(void *pp, Byte *dest, SizeT *destLen, const Byte *src, p->bufConv = p->bufTotal; } } - if (p->bufTotal == p->bufPos && srcLenOrig == 0 && srcWasFinished) - *wasFinished = 1; + + if (p->bufTotal == p->bufPos && srcRem == 0 && srcWasFinished) + { + *status = CODER_STATUS_FINISHED_WITH_MARK; + // *wasFinished = 1; + } + return SZ_OK; } -SRes BraState_SetFromMethod(IStateCoder *p, UInt64 id, int encodeMode, ISzAlloc *alloc) + +SRes BraState_SetFromMethod(IStateCoder *p, UInt64 id, int encodeMode, ISzAllocPtr alloc) { CBraState *decoder; - if (id != XZ_ID_Delta && - id != XZ_ID_X86 && - id != XZ_ID_PPC && - id != XZ_ID_IA64 && - id != XZ_ID_ARM && - id != XZ_ID_ARMT && - id != XZ_ID_SPARC) + if (id < XZ_ID_Delta || id > XZ_ID_SPARC) return SZ_ERROR_UNSUPPORTED; - p->p = 0; - decoder = (CBraState *)alloc->Alloc(alloc, sizeof(CBraState)); - if (decoder == 0) - return SZ_ERROR_MEM; + decoder = (CBraState *)p->p; + if (!decoder) + { + decoder = (CBraState *)ISzAlloc_Alloc(alloc, sizeof(CBraState)); + if (!decoder) + return SZ_ERROR_MEM; + p->p = decoder; + p->Free = BraState_Free; + p->SetProps = BraState_SetProps; + p->Init = BraState_Init; + p->Code2 = BraState_Code2; + p->Filter = BraState_Filter; + } decoder->methodId = (UInt32)id; decoder->encodeMode = encodeMode; - p->p = decoder; - p->Free = BraState_Free; - p->SetProps = BraState_SetProps; - p->Init = BraState_Init; - p->Code = BraState_Code; return SZ_OK; } + + /* ---------- SbState ---------- */ #ifdef USE_SUBBLOCK -static void SbState_Free(void *pp, ISzAlloc *alloc) +static void SbState_Free(void *pp, ISzAllocPtr alloc) { CSbDec *p = (CSbDec *)pp; SbDec_Free(p); - alloc->Free(alloc, pp); + ISzAlloc_Free(alloc, pp); } -static SRes SbState_SetProps(void *pp, const Byte *props, size_t propSize, ISzAlloc *alloc) +static SRes SbState_SetProps(void *pp, const Byte *props, size_t propSize, ISzAllocPtr alloc) { UNUSED_VAR(pp); UNUSED_VAR(props); @@ -246,8 +289,10 @@ static void SbState_Init(void *pp) SbDec_Init((CSbDec *)pp); } -static SRes SbState_Code(void *pp, Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen, - int srcWasFinished, ECoderFinishMode finishMode, int *wasFinished) +static SRes SbState_Code2(void *pp, Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen, + int srcWasFinished, ECoderFinishMode finishMode, + // int *wasFinished + ECoderStatus *status) { CSbDec *p = (CSbDec *)pp; SRes res; @@ -260,102 +305,185 @@ static SRes SbState_Code(void *pp, Byte *dest, SizeT *destLen, const Byte *src, res = SbDec_Decode((CSbDec *)pp); *destLen -= p->destLen; *srcLen -= p->srcLen; - *wasFinished = (*destLen == 0 && *srcLen == 0); /* change it */ + // *wasFinished = (*destLen == 0 && *srcLen == 0); /* change it */ + *status = (*destLen == 0 && *srcLen == 0) ? + CODER_STATUS_FINISHED_WITH_MARK : + CODER_STATUS_NOT_FINISHED; return res; } -SRes SbState_SetFromMethod(IStateCoder *p, ISzAlloc *alloc) +static SRes SbState_SetFromMethod(IStateCoder *p, ISzAllocPtr alloc) { - CSbDec *decoder; - p->p = 0; - decoder = alloc->Alloc(alloc, sizeof(CSbDec)); - if (decoder == 0) - return SZ_ERROR_MEM; - p->p = decoder; - p->Free = SbState_Free; - p->SetProps = SbState_SetProps; - p->Init = SbState_Init; - p->Code = SbState_Code; + CSbDec *decoder = (CSbDec *)p->p; + if (!decoder) + { + decoder = (CSbDec *)ISzAlloc_Alloc(alloc, sizeof(CSbDec)); + if (!decoder) + return SZ_ERROR_MEM; + p->p = decoder; + p->Free = SbState_Free; + p->SetProps = SbState_SetProps; + p->Init = SbState_Init; + p->Code2 = SbState_Code2; + p->Filter = NULL; + } SbDec_Construct(decoder); SbDec_SetAlloc(decoder, alloc); return SZ_OK; } + #endif -/* ---------- Lzma2State ---------- */ -static void Lzma2State_Free(void *pp, ISzAlloc *alloc) + +/* ---------- Lzma2 ---------- */ + +typedef struct { - Lzma2Dec_Free((CLzma2Dec *)pp, alloc); - alloc->Free(alloc, pp); + CLzma2Dec decoder; + BoolInt outBufMode; +} CLzma2Dec_Spec; + + +static void Lzma2State_Free(void *pp, ISzAllocPtr alloc) +{ + CLzma2Dec_Spec *p = (CLzma2Dec_Spec *)pp; + if (p->outBufMode) + Lzma2Dec_FreeProbs(&p->decoder, alloc); + else + Lzma2Dec_Free(&p->decoder, alloc); + ISzAlloc_Free(alloc, pp); } -static SRes Lzma2State_SetProps(void *pp, const Byte *props, size_t propSize, ISzAlloc *alloc) +static SRes Lzma2State_SetProps(void *pp, const Byte *props, size_t propSize, ISzAllocPtr alloc) { if (propSize != 1) return SZ_ERROR_UNSUPPORTED; - return Lzma2Dec_Allocate((CLzma2Dec *)pp, props[0], alloc); + { + CLzma2Dec_Spec *p = (CLzma2Dec_Spec *)pp; + if (p->outBufMode) + return Lzma2Dec_AllocateProbs(&p->decoder, props[0], alloc); + else + return Lzma2Dec_Allocate(&p->decoder, props[0], alloc); + } } static void Lzma2State_Init(void *pp) { - Lzma2Dec_Init((CLzma2Dec *)pp); + Lzma2Dec_Init(&((CLzma2Dec_Spec *)pp)->decoder); } -static SRes Lzma2State_Code(void *pp, Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen, - int srcWasFinished, ECoderFinishMode finishMode, int *wasFinished) + +/* + if (outBufMode), then (dest) is not used. Use NULL. + Data is unpacked to (spec->decoder.decoder.dic) output buffer. +*/ + +static SRes Lzma2State_Code2(void *pp, Byte *dest, SizeT *destLen, const Byte *src, SizeT *srcLen, + int srcWasFinished, ECoderFinishMode finishMode, + // int *wasFinished, + ECoderStatus *status) { - ELzmaStatus status; + CLzma2Dec_Spec *spec = (CLzma2Dec_Spec *)pp; + ELzmaStatus status2; /* ELzmaFinishMode fm = (finishMode == LZMA_FINISH_ANY) ? LZMA_FINISH_ANY : LZMA_FINISH_END; */ - SRes res = Lzma2Dec_DecodeToBuf((CLzma2Dec *)pp, dest, destLen, src, srcLen, (ELzmaFinishMode)finishMode, &status); + SRes res; UNUSED_VAR(srcWasFinished); - *wasFinished = (status == LZMA_STATUS_FINISHED_WITH_MARK); + if (spec->outBufMode) + { + SizeT dicPos = spec->decoder.decoder.dicPos; + SizeT dicLimit = dicPos + *destLen; + res = Lzma2Dec_DecodeToDic(&spec->decoder, dicLimit, src, srcLen, (ELzmaFinishMode)finishMode, &status2); + *destLen = spec->decoder.decoder.dicPos - dicPos; + } + else + res = Lzma2Dec_DecodeToBuf(&spec->decoder, dest, destLen, src, srcLen, (ELzmaFinishMode)finishMode, &status2); + // *wasFinished = (status2 == LZMA_STATUS_FINISHED_WITH_MARK); + // ECoderStatus values are identical to ELzmaStatus values of LZMA2 decoder + *status = (ECoderStatus)status2; return res; } -static SRes Lzma2State_SetFromMethod(IStateCoder *p, ISzAlloc *alloc) + +static SRes Lzma2State_SetFromMethod(IStateCoder *p, Byte *outBuf, size_t outBufSize, ISzAllocPtr alloc) { - CLzma2Dec *decoder = (CLzma2Dec *)alloc->Alloc(alloc, sizeof(CLzma2Dec)); - p->p = decoder; - if (decoder == 0) - return SZ_ERROR_MEM; - p->Free = Lzma2State_Free; - p->SetProps = Lzma2State_SetProps; - p->Init = Lzma2State_Init; - p->Code = Lzma2State_Code; - Lzma2Dec_Construct(decoder); + CLzma2Dec_Spec *spec = (CLzma2Dec_Spec *)p->p; + if (!spec) + { + spec = (CLzma2Dec_Spec *)ISzAlloc_Alloc(alloc, sizeof(CLzma2Dec_Spec)); + if (!spec) + return SZ_ERROR_MEM; + p->p = spec; + p->Free = Lzma2State_Free; + p->SetProps = Lzma2State_SetProps; + p->Init = Lzma2State_Init; + p->Code2 = Lzma2State_Code2; + p->Filter = NULL; + Lzma2Dec_Construct(&spec->decoder); + } + spec->outBufMode = False; + if (outBuf) + { + spec->outBufMode = True; + spec->decoder.decoder.dic = outBuf; + spec->decoder.decoder.dicBufSize = outBufSize; + } return SZ_OK; } -void MixCoder_Construct(CMixCoder *p, ISzAlloc *alloc) +static SRes Lzma2State_ResetOutBuf(IStateCoder *p, Byte *outBuf, size_t outBufSize) +{ + CLzma2Dec_Spec *spec = (CLzma2Dec_Spec *)p->p; + if ((spec->outBufMode && !outBuf) || (!spec->outBufMode && outBuf)) + return SZ_ERROR_FAIL; + if (outBuf) + { + spec->decoder.decoder.dic = outBuf; + spec->decoder.decoder.dicBufSize = outBufSize; + } + return SZ_OK; +} + + + +static void MixCoder_Construct(CMixCoder *p, ISzAllocPtr alloc) { unsigned i; p->alloc = alloc; p->buf = NULL; p->numCoders = 0; + + p->outBufSize = 0; + p->outBuf = NULL; + // p->SingleBufMode = False; + for (i = 0; i < MIXCODER_NUM_FILTERS_MAX; i++) p->coders[i].p = NULL; } -void MixCoder_Free(CMixCoder *p) + +static void MixCoder_Free(CMixCoder *p) { unsigned i; - for (i = 0; i < p->numCoders; i++) + p->numCoders = 0; + for (i = 0; i < MIXCODER_NUM_FILTERS_MAX; i++) { IStateCoder *sc = &p->coders[i]; - if (p->alloc && sc->p) + if (sc->p) + { sc->Free(sc->p, p->alloc); + sc->p = NULL; + } } - p->numCoders = 0; if (p->buf) { - p->alloc->Free(p->alloc, p->buf); + ISzAlloc_Free(p->alloc, p->buf); p->buf = NULL; /* 9.31: the BUG was fixed */ } } -void MixCoder_Init(CMixCoder *p) +static void MixCoder_Init(CMixCoder *p) { unsigned i; for (i = 0; i < MIXCODER_NUM_FILTERS_MAX - 1; i++) @@ -368,16 +496,22 @@ void MixCoder_Init(CMixCoder *p) { IStateCoder *coder = &p->coders[i]; coder->Init(coder->p); + p->results[i] = SZ_OK; } + p->outWritten = 0; + p->wasFinished = False; + p->res = SZ_OK; + p->status = CODER_STATUS_NOT_SPECIFIED; } -SRes MixCoder_SetFromMethod(CMixCoder *p, unsigned coderIndex, UInt64 methodId) + +static SRes MixCoder_SetFromMethod(CMixCoder *p, unsigned coderIndex, UInt64 methodId, Byte *outBuf, size_t outBufSize) { IStateCoder *sc = &p->coders[coderIndex]; p->ids[coderIndex] = methodId; switch (methodId) { - case XZ_ID_LZMA2: return Lzma2State_SetFromMethod(sc, p->alloc); + case XZ_ID_LZMA2: return Lzma2State_SetFromMethod(sc, outBuf, outBufSize, p->alloc); #ifdef USE_SUBBLOCK case XZ_ID_Subblock: return SbState_SetFromMethod(sc, p->alloc); #endif @@ -387,31 +521,128 @@ SRes MixCoder_SetFromMethod(CMixCoder *p, unsigned coderIndex, UInt64 methodId) return BraState_SetFromMethod(sc, methodId, 0, p->alloc); } -SRes MixCoder_Code(CMixCoder *p, Byte *dest, SizeT *destLen, + +static SRes MixCoder_ResetFromMethod(CMixCoder *p, unsigned coderIndex, UInt64 methodId, Byte *outBuf, size_t outBufSize) +{ + IStateCoder *sc = &p->coders[coderIndex]; + switch (methodId) + { + case XZ_ID_LZMA2: return Lzma2State_ResetOutBuf(sc, outBuf, outBufSize); + } + return SZ_ERROR_UNSUPPORTED; +} + + + +/* + if (destFinish) - then unpack data block is finished at (*destLen) position, + and we can return data that were not processed by filter + +output (status) can be : + CODER_STATUS_NOT_FINISHED + CODER_STATUS_FINISHED_WITH_MARK + CODER_STATUS_NEEDS_MORE_INPUT - not implemented still +*/ + +static SRes MixCoder_Code(CMixCoder *p, + Byte *dest, SizeT *destLen, int destFinish, const Byte *src, SizeT *srcLen, int srcWasFinished, - ECoderFinishMode finishMode, ECoderStatus *status) + ECoderFinishMode finishMode) { SizeT destLenOrig = *destLen; SizeT srcLenOrig = *srcLen; - Bool allFinished = True; + *destLen = 0; *srcLen = 0; - *status = CODER_STATUS_NOT_FINISHED; - if (!p->buf) + if (p->wasFinished) + return p->res; + + p->status = CODER_STATUS_NOT_FINISHED; + + // if (p->SingleBufMode) + if (p->outBuf) { - p->buf = (Byte *)p->alloc->Alloc(p->alloc, CODER_BUF_SIZE * (MIXCODER_NUM_FILTERS_MAX - 1)); - if (!p->buf) - return SZ_ERROR_MEM; + SRes res; + SizeT destLen2, srcLen2; + int wasFinished; + + PRF_STR("------- MixCoder Single ----------"); + + srcLen2 = srcLenOrig; + destLen2 = destLenOrig; + + { + IStateCoder *coder = &p->coders[0]; + res = coder->Code2(coder->p, NULL, &destLen2, src, &srcLen2, srcWasFinished, finishMode, + // &wasFinished, + &p->status); + wasFinished = (p->status == CODER_STATUS_FINISHED_WITH_MARK); + } + + p->res = res; + + /* + if (wasFinished) + p->status = CODER_STATUS_FINISHED_WITH_MARK; + else + { + if (res == SZ_OK) + if (destLen2 != destLenOrig) + p->status = CODER_STATUS_NEEDS_MORE_INPUT; + } + */ + + + *srcLen = srcLen2; + src += srcLen2; + p->outWritten += destLen2; + + if (res != SZ_OK || srcWasFinished || wasFinished) + p->wasFinished = True; + + if (p->numCoders == 1) + *destLen = destLen2; + else if (p->wasFinished) + { + unsigned i; + size_t processed = p->outWritten; + + for (i = 1; i < p->numCoders; i++) + { + IStateCoder *coder = &p->coders[i]; + processed = coder->Filter(coder->p, p->outBuf, processed); + if (wasFinished || (destFinish && p->outWritten == destLenOrig)) + processed = p->outWritten; + PRF_STR_INT("filter", i); + } + *destLen = processed; + } + return res; } + PRF_STR("standard mix"); + if (p->numCoders != 1) + { + if (!p->buf) + { + p->buf = (Byte *)ISzAlloc_Alloc(p->alloc, CODER_BUF_SIZE * (MIXCODER_NUM_FILTERS_MAX - 1)); + if (!p->buf) + return SZ_ERROR_MEM; + } + finishMode = CODER_FINISH_ANY; + } for (;;) { - Bool processed = False; + BoolInt processed = False; + BoolInt allFinished = True; + SRes resMain = SZ_OK; unsigned i; + + p->status = CODER_STATUS_NOT_FINISHED; /* if (p->numCoders == 1 && *destLen == destLenOrig && finishMode == LZMA_FINISH_ANY) break; @@ -421,79 +652,106 @@ SRes MixCoder_Code(CMixCoder *p, Byte *dest, SizeT *destLen, { SRes res; IStateCoder *coder = &p->coders[i]; - Byte *destCur; - SizeT destLenCur, srcLenCur; - const Byte *srcCur; - int srcFinishedCur; + Byte *dest2; + SizeT destLen2, srcLen2; // destLen2_Orig; + const Byte *src2; + int srcFinished2; int encodingWasFinished; + ECoderStatus status2; if (i == 0) { - srcCur = src; - srcLenCur = srcLenOrig - *srcLen; - srcFinishedCur = srcWasFinished; + src2 = src; + srcLen2 = srcLenOrig - *srcLen; + srcFinished2 = srcWasFinished; } else { - srcCur = p->buf + (CODER_BUF_SIZE * (i - 1)) + p->pos[i - 1]; - srcLenCur = p->size[i - 1] - p->pos[i - 1]; - srcFinishedCur = p->finished[i - 1]; + size_t k = i - 1; + src2 = p->buf + (CODER_BUF_SIZE * k) + p->pos[k]; + srcLen2 = p->size[k] - p->pos[k]; + srcFinished2 = p->finished[k]; } if (i == p->numCoders - 1) { - destCur = dest; - destLenCur = destLenOrig - *destLen; + dest2 = dest; + destLen2 = destLenOrig - *destLen; } else { if (p->pos[i] != p->size[i]) continue; - destCur = p->buf + (CODER_BUF_SIZE * i); - destLenCur = CODER_BUF_SIZE; + dest2 = p->buf + (CODER_BUF_SIZE * i); + destLen2 = CODER_BUF_SIZE; } - res = coder->Code(coder->p, destCur, &destLenCur, srcCur, &srcLenCur, srcFinishedCur, finishMode, &encodingWasFinished); + // destLen2_Orig = destLen2; + + if (p->results[i] != SZ_OK) + { + if (resMain == SZ_OK) + resMain = p->results[i]; + continue; + } + res = coder->Code2(coder->p, + dest2, &destLen2, + src2, &srcLen2, srcFinished2, + finishMode, + // &encodingWasFinished, + &status2); + + if (res != SZ_OK) + { + p->results[i] = res; + if (resMain == SZ_OK) + resMain = res; + } + + encodingWasFinished = (status2 == CODER_STATUS_FINISHED_WITH_MARK); + if (!encodingWasFinished) + { allFinished = False; + if (p->numCoders == 1 && res == SZ_OK) + p->status = status2; + } if (i == 0) { - *srcLen += srcLenCur; - src += srcLenCur; + *srcLen += srcLen2; + src += srcLen2; } else - { - p->pos[i - 1] += srcLenCur; - } + p->pos[(size_t)i - 1] += srcLen2; if (i == p->numCoders - 1) { - *destLen += destLenCur; - dest += destLenCur; + *destLen += destLen2; + dest += destLen2; } else { - p->size[i] = destLenCur; + p->size[i] = destLen2; p->pos[i] = 0; p->finished[i] = encodingWasFinished; } - if (res != SZ_OK) - return res; - - if (destLenCur != 0 || srcLenCur != 0) + if (destLen2 != 0 || srcLen2 != 0) processed = True; } + if (!processed) - break; + { + if (allFinished) + p->status = CODER_STATUS_FINISHED_WITH_MARK; + return resMain; + } } - if (allFinished) - *status = CODER_STATUS_FINISHED_WITH_MARK; - return SZ_OK; } + SRes Xz_ParseHeader(CXzStreamFlags *p, const Byte *buf) { *p = (CXzStreamFlags)GetBe16(buf + XZ_SIG_SIZE); @@ -503,13 +761,13 @@ SRes Xz_ParseHeader(CXzStreamFlags *p, const Byte *buf) return XzFlags_IsSupported(*p) ? SZ_OK : SZ_ERROR_UNSUPPORTED; } -static Bool Xz_CheckFooter(CXzStreamFlags flags, UInt64 indexSize, const Byte *buf) +static BoolInt Xz_CheckFooter(CXzStreamFlags flags, UInt64 indexSize, const Byte *buf) { - return - indexSize == (((UInt64)GetUi32(buf + 4) + 1) << 2) && - (GetUi32(buf) == CrcCalc(buf + 4, 6) && - flags == GetBe16(buf + 8) && - memcmp(buf + 10, XZ_FOOTER_SIG, XZ_FOOTER_SIG_SIZE) == 0); + return indexSize == (((UInt64)GetUi32(buf + 4) + 1) << 2) + && GetUi32(buf) == CrcCalc(buf + 4, 6) + && flags == GetBe16(buf + 8) + && buf[10] == XZ_FOOTER_SIG_0 + && buf[11] == XZ_FOOTER_SIG_1; } #define READ_VARINT_AND_CHECK(buf, pos, size, res) \ @@ -517,20 +775,48 @@ static Bool Xz_CheckFooter(CXzStreamFlags flags, UInt64 indexSize, const Byte *b if (s == 0) return SZ_ERROR_ARCHIVE; pos += s; } +static BoolInt XzBlock_AreSupportedFilters(const CXzBlock *p) +{ + unsigned numFilters = XzBlock_GetNumFilters(p) - 1; + unsigned i; + { + const CXzFilter *f = &p->filters[numFilters]; + if (f->id != XZ_ID_LZMA2 || f->propsSize != 1 || f->props[0] > 40) + return False; + } + + for (i = 0; i < numFilters; i++) + { + const CXzFilter *f = &p->filters[i]; + if (f->id == XZ_ID_Delta) + { + if (f->propsSize != 1) + return False; + } + else if (f->id < XZ_ID_Delta + || f->id > XZ_ID_SPARC + || (f->propsSize != 0 && f->propsSize != 4)) + return False; + } + return True; +} + + SRes XzBlock_Parse(CXzBlock *p, const Byte *header) { unsigned pos; unsigned numFilters, i; unsigned headerSize = (unsigned)header[0] << 2; + /* (headerSize != 0) : another code checks */ + if (CrcCalc(header, headerSize) != GetUi32(header + headerSize)) return SZ_ERROR_ARCHIVE; pos = 1; - if (pos == headerSize) - return SZ_ERROR_ARCHIVE; p->flags = header[pos++]; + p->packSize = (UInt64)(Int64)-1; if (XzBlock_HasPackSize(p)) { READ_VARINT_AND_CHECK(header, pos, headerSize, &p->packSize); @@ -538,6 +824,7 @@ SRes XzBlock_Parse(CXzBlock *p, const Byte *header) return SZ_ERROR_ARCHIVE; } + p->unpackSize = (UInt64)(Int64)-1; if (XzBlock_HasUnpackSize(p)) READ_VARINT_AND_CHECK(header, pos, headerSize, &p->unpackSize); @@ -564,37 +851,56 @@ SRes XzBlock_Parse(CXzBlock *p, const Byte *header) #endif } + if (XzBlock_HasUnsupportedFlags(p)) + return SZ_ERROR_UNSUPPORTED; + while (pos < headerSize) if (header[pos++] != 0) return SZ_ERROR_ARCHIVE; return SZ_OK; } -SRes XzDec_Init(CMixCoder *p, const CXzBlock *block) + + + +static SRes XzDecMix_Init(CMixCoder *p, const CXzBlock *block, Byte *outBuf, size_t outBufSize) { unsigned i; - Bool needReInit = True; + BoolInt needReInit = True; unsigned numFilters = XzBlock_GetNumFilters(block); - - if (numFilters == p->numCoders) + + if (numFilters == p->numCoders && ((p->outBuf && outBuf) || (!p->outBuf && !outBuf))) { + needReInit = False; for (i = 0; i < numFilters; i++) if (p->ids[i] != block->filters[numFilters - 1 - i].id) + { + needReInit = True; break; - needReInit = (i != numFilters); + } } + + // p->SingleBufMode = (outBuf != NULL); + p->outBuf = outBuf; + p->outBufSize = outBufSize; + + // p->SingleBufMode = False; + // outBuf = NULL; if (needReInit) { MixCoder_Free(p); - p->numCoders = numFilters; for (i = 0; i < numFilters; i++) { - const CXzFilter *f = &block->filters[numFilters - 1 - i]; - RINOK(MixCoder_SetFromMethod(p, i, f->id)); + RINOK(MixCoder_SetFromMethod(p, i, block->filters[numFilters - 1 - i].id, outBuf, outBufSize)); } + p->numCoders = numFilters; } - + else + { + RINOK(MixCoder_ResetFromMethod(p, 0, block->filters[numFilters - 1].id, outBuf, outBufSize)); + } + for (i = 0; i < numFilters; i++) { const CXzFilter *f = &block->filters[numFilters - 1 - i]; @@ -606,6 +912,8 @@ SRes XzDec_Init(CMixCoder *p, const CXzBlock *block) return SZ_OK; } + + void XzUnpacker_Init(CXzUnpacker *p) { p->state = XZ_STATE_STREAM_HEADER; @@ -614,75 +922,171 @@ void XzUnpacker_Init(CXzUnpacker *p) p->numFinishedStreams = 0; p->numTotalBlocks = 0; p->padSize = 0; + p->decodeOnlyOneBlock = 0; + + p->parseMode = False; + p->decodeToStreamSignature = False; + + // p->outBuf = NULL; + // p->outBufSize = 0; + p->outDataWritten = 0; } -void XzUnpacker_Construct(CXzUnpacker *p, ISzAlloc *alloc) + +void XzUnpacker_SetOutBuf(CXzUnpacker *p, Byte *outBuf, size_t outBufSize) +{ + p->outBuf = outBuf; + p->outBufSize = outBufSize; +} + + +void XzUnpacker_Construct(CXzUnpacker *p, ISzAllocPtr alloc) { MixCoder_Construct(&p->decoder, alloc); + p->outBuf = NULL; + p->outBufSize = 0; XzUnpacker_Init(p); } + void XzUnpacker_Free(CXzUnpacker *p) { MixCoder_Free(&p->decoder); } + +void XzUnpacker_PrepareToRandomBlockDecoding(CXzUnpacker *p) +{ + p->indexSize = 0; + p->numBlocks = 0; + Sha256_Init(&p->sha); + p->state = XZ_STATE_BLOCK_HEADER; + p->pos = 0; + p->decodeOnlyOneBlock = 1; +} + + +static void XzUnpacker_UpdateIndex(CXzUnpacker *p, UInt64 packSize, UInt64 unpackSize) +{ + Byte temp[32]; + unsigned num = Xz_WriteVarInt(temp, packSize); + num += Xz_WriteVarInt(temp + num, unpackSize); + Sha256_Update(&p->sha, temp, num); + p->indexSize += num; + p->numBlocks++; +} + + + SRes XzUnpacker_Code(CXzUnpacker *p, Byte *dest, SizeT *destLen, - const Byte *src, SizeT *srcLen, ECoderFinishMode finishMode, ECoderStatus *status) + const Byte *src, SizeT *srcLen, int srcFinished, + ECoderFinishMode finishMode, ECoderStatus *status) { SizeT destLenOrig = *destLen; SizeT srcLenOrig = *srcLen; *destLen = 0; *srcLen = 0; *status = CODER_STATUS_NOT_SPECIFIED; + for (;;) { - SizeT srcRem = srcLenOrig - *srcLen; + SizeT srcRem; if (p->state == XZ_STATE_BLOCK) { SizeT destLen2 = destLenOrig - *destLen; SizeT srcLen2 = srcLenOrig - *srcLen; SRes res; + + ECoderFinishMode finishMode2 = finishMode; + BoolInt srcFinished2 = srcFinished; + BoolInt destFinish = False; + + if (p->block.packSize != (UInt64)(Int64)-1) + { + UInt64 rem = p->block.packSize - p->packSize; + if (srcLen2 >= rem) + { + srcFinished2 = True; + srcLen2 = (SizeT)rem; + } + if (rem == 0 && p->block.unpackSize == p->unpackSize) + return SZ_ERROR_DATA; + } + + if (p->block.unpackSize != (UInt64)(Int64)-1) + { + UInt64 rem = p->block.unpackSize - p->unpackSize; + if (destLen2 >= rem) + { + destFinish = True; + finishMode2 = CODER_FINISH_END; + destLen2 = (SizeT)rem; + } + } + + /* if (srcLen2 == 0 && destLen2 == 0) { *status = CODER_STATUS_NOT_FINISHED; return SZ_OK; } + */ - res = MixCoder_Code(&p->decoder, dest, &destLen2, src, &srcLen2, False, finishMode, status); - XzCheck_Update(&p->check, dest, destLen2); + { + res = MixCoder_Code(&p->decoder, + (p->outBuf ? NULL : dest), &destLen2, destFinish, + src, &srcLen2, srcFinished2, + finishMode2); + + *status = p->decoder.status; + XzCheck_Update(&p->check, (p->outBuf ? p->outBuf + p->outDataWritten : dest), destLen2); + if (!p->outBuf) + dest += destLen2; + p->outDataWritten += destLen2; + } (*srcLen) += srcLen2; src += srcLen2; p->packSize += srcLen2; - (*destLen) += destLen2; - dest += destLen2; p->unpackSize += destLen2; - + RINOK(res); - - if (*status == CODER_STATUS_FINISHED_WITH_MARK) + + if (*status != CODER_STATUS_FINISHED_WITH_MARK) { - Byte temp[32]; - unsigned num = Xz_WriteVarInt(temp, p->packSize + p->blockHeaderSize + XzFlags_GetCheckSize(p->streamFlags)); - num += Xz_WriteVarInt(temp + num, p->unpackSize); - Sha256_Update(&p->sha, temp, num); - p->indexSize += num; - p->numBlocks++; + if (p->block.packSize == p->packSize + && *status == CODER_STATUS_NEEDS_MORE_INPUT) + { + PRF_STR("CODER_STATUS_NEEDS_MORE_INPUT"); + *status = CODER_STATUS_NOT_SPECIFIED; + return SZ_ERROR_DATA; + } + return SZ_OK; + } + { + XzUnpacker_UpdateIndex(p, XzUnpacker_GetPackSizeForIndex(p), p->unpackSize); p->state = XZ_STATE_BLOCK_FOOTER; p->pos = 0; p->alignPos = 0; + *status = CODER_STATUS_NOT_SPECIFIED; + + if ((p->block.packSize != (UInt64)(Int64)-1 && p->block.packSize != p->packSize) + || (p->block.unpackSize != (UInt64)(Int64)-1 && p->block.unpackSize != p->unpackSize)) + { + PRF_STR("ERROR: block.size mismatch"); + return SZ_ERROR_DATA; + } } - else if (srcLen2 == 0 && destLen2 == 0) - return SZ_OK; - - continue; + // continue; } - if (srcRem == 0) + srcRem = srcLenOrig - *srcLen; + + // XZ_STATE_BLOCK_FOOTER can transit to XZ_STATE_BLOCK_HEADER without input bytes + if (srcRem == 0 && p->state != XZ_STATE_BLOCK_FOOTER) { *status = CODER_STATUS_NEEDS_MORE_INPUT; return SZ_OK; @@ -696,6 +1100,8 @@ SRes XzUnpacker_Code(CXzUnpacker *p, Byte *dest, SizeT *destLen, { if (p->pos < XZ_SIG_SIZE && *src != XZ_SIG[p->pos]) return SZ_ERROR_NO_ARCHIVE; + if (p->decodeToStreamSignature) + return SZ_OK; p->buf[p->pos++] = *src++; (*srcLen)++; } @@ -703,10 +1109,10 @@ SRes XzUnpacker_Code(CXzUnpacker *p, Byte *dest, SizeT *destLen, { RINOK(Xz_ParseHeader(&p->streamFlags, p->buf)); p->numStartedStreams++; - p->state = XZ_STATE_BLOCK_HEADER; - Sha256_Init(&p->sha); p->indexSize = 0; p->numBlocks = 0; + Sha256_Init(&p->sha); + p->state = XZ_STATE_BLOCK_HEADER; p->pos = 0; } break; @@ -720,6 +1126,8 @@ SRes XzUnpacker_Code(CXzUnpacker *p, Byte *dest, SizeT *destLen, (*srcLen)++; if (p->buf[0] == 0) { + if (p->decodeOnlyOneBlock) + return SZ_ERROR_DATA; p->indexPreSize = 1 + Xz_WriteVarInt(p->buf + 1, p->numBlocks); p->indexPos = p->indexPreSize; p->indexSize += p->indexPreSize; @@ -727,10 +1135,13 @@ SRes XzUnpacker_Code(CXzUnpacker *p, Byte *dest, SizeT *destLen, Sha256_Init(&p->sha); p->crc = CrcUpdate(CRC_INIT_VAL, p->buf, p->indexPreSize); p->state = XZ_STATE_STREAM_INDEX; + break; } p->blockHeaderSize = ((UInt32)p->buf[0] << 2) + 4; + break; } - else if (p->pos != p->blockHeaderSize) + + if (p->pos != p->blockHeaderSize) { UInt32 cur = p->blockHeaderSize - p->pos; if (cur > srcRem) @@ -743,20 +1154,32 @@ SRes XzUnpacker_Code(CXzUnpacker *p, Byte *dest, SizeT *destLen, else { RINOK(XzBlock_Parse(&p->block, p->buf)); + if (!XzBlock_AreSupportedFilters(&p->block)) + return SZ_ERROR_UNSUPPORTED; p->numTotalBlocks++; p->state = XZ_STATE_BLOCK; p->packSize = 0; p->unpackSize = 0; XzCheck_Init(&p->check, XzFlags_GetCheckType(p->streamFlags)); - RINOK(XzDec_Init(&p->decoder, &p->block)); + if (p->parseMode) + { + p->headerParsedOk = True; + return SZ_OK; + } + RINOK(XzDecMix_Init(&p->decoder, &p->block, p->outBuf, p->outBufSize)); } break; } case XZ_STATE_BLOCK_FOOTER: { - if (((p->packSize + p->alignPos) & 3) != 0) + if ((((unsigned)p->packSize + p->alignPos) & 3) != 0) { + if (srcRem == 0) + { + *status = CODER_STATUS_NEEDS_MORE_INPUT; + return SZ_OK; + } (*srcLen)++; p->alignPos++; if (*src++ != 0) @@ -768,20 +1191,31 @@ SRes XzUnpacker_Code(CXzUnpacker *p, Byte *dest, SizeT *destLen, UInt32 cur = checkSize - p->pos; if (cur != 0) { + if (srcRem == 0) + { + *status = CODER_STATUS_NEEDS_MORE_INPUT; + return SZ_OK; + } if (cur > srcRem) cur = (UInt32)srcRem; memcpy(p->buf + p->pos, src, cur); p->pos += cur; (*srcLen) += cur; src += cur; + if (checkSize != p->pos) + break; } - else { Byte digest[XZ_CHECK_SIZE_MAX]; p->state = XZ_STATE_BLOCK_HEADER; p->pos = 0; if (XzCheck_Final(&p->check, digest) && memcmp(digest, p->buf, checkSize) != 0) return SZ_ERROR_CRC; + if (p->decodeOnlyOneBlock) + { + *status = CODER_STATUS_FINISHED_WITH_MARK; + return SZ_OK; + } } } break; @@ -897,17 +1331,1436 @@ SRes XzUnpacker_Code(CXzUnpacker *p, Byte *dest, SizeT *destLen, */ } -Bool XzUnpacker_IsStreamWasFinished(CXzUnpacker *p) + +SRes XzUnpacker_CodeFull(CXzUnpacker *p, Byte *dest, SizeT *destLen, + const Byte *src, SizeT *srcLen, + ECoderFinishMode finishMode, ECoderStatus *status) +{ + XzUnpacker_Init(p); + XzUnpacker_SetOutBuf(p, dest, *destLen); + + return XzUnpacker_Code(p, + NULL, destLen, + src, srcLen, True, + finishMode, status); +} + + +BoolInt XzUnpacker_IsBlockFinished(const CXzUnpacker *p) +{ + return (p->state == XZ_STATE_BLOCK_HEADER) && (p->pos == 0); +} + +BoolInt XzUnpacker_IsStreamWasFinished(const CXzUnpacker *p) { return (p->state == XZ_STATE_STREAM_PADDING) && (((UInt32)p->padSize & 3) == 0); } -UInt64 XzUnpacker_GetExtraSize(CXzUnpacker *p) +UInt64 XzUnpacker_GetExtraSize(const CXzUnpacker *p) { UInt64 num = 0; if (p->state == XZ_STATE_STREAM_PADDING) - num += p->padSize; + num = p->padSize; else if (p->state == XZ_STATE_STREAM_HEADER) - num += p->padSize + p->pos; + num = p->padSize + p->pos; return num; } + + + + + + + + + + + + + + + + + + + + + +#ifndef _7ZIP_ST +#include "MtDec.h" +#endif + + +void XzDecMtProps_Init(CXzDecMtProps *p) +{ + p->inBufSize_ST = 1 << 18; + p->outStep_ST = 1 << 20; + p->ignoreErrors = False; + + #ifndef _7ZIP_ST + p->numThreads = 1; + p->inBufSize_MT = 1 << 18; + p->memUseMax = sizeof(size_t) << 28; + #endif +} + + + +#ifndef _7ZIP_ST + +/* ---------- CXzDecMtThread ---------- */ + +typedef struct +{ + Byte *outBuf; + size_t outBufSize; + size_t outPreSize; + size_t inPreSize; + size_t inPreHeaderSize; + size_t blockPackSize_for_Index; // including block header and checksum. + size_t blockPackTotal; // including stream header, block header and checksum. + size_t inCodeSize; + size_t outCodeSize; + ECoderStatus status; + SRes codeRes; + BoolInt skipMode; + // BoolInt finishedWithMark; + EMtDecParseState parseState; + BoolInt parsing_Truncated; + BoolInt atBlockHeader; + CXzStreamFlags streamFlags; + // UInt64 numFinishedStreams + UInt64 numStreams; + UInt64 numTotalBlocks; + UInt64 numBlocks; + + BoolInt dec_created; + CXzUnpacker dec; + + Byte mtPad[1 << 7]; +} CXzDecMtThread; + +#endif + + +/* ---------- CXzDecMt ---------- */ + +typedef struct +{ + CAlignOffsetAlloc alignOffsetAlloc; + ISzAllocPtr allocMid; + + CXzDecMtProps props; + size_t unpackBlockMaxSize; + + ISeqInStream *inStream; + ISeqOutStream *outStream; + ICompressProgress *progress; + // CXzStatInfo *stat; + + BoolInt finishMode; + BoolInt outSize_Defined; + UInt64 outSize; + + UInt64 outProcessed; + UInt64 inProcessed; + UInt64 readProcessed; + BoolInt readWasFinished; + SRes readRes; + SRes writeRes; + + Byte *outBuf; + size_t outBufSize; + Byte *inBuf; + size_t inBufSize; + + CXzUnpacker dec; + + ECoderStatus status; + SRes codeRes; + + #ifndef _7ZIP_ST + BoolInt mainDecoderWasCalled; + // int statErrorDefined; + int finishedDecoderIndex; + + // global values that are used in Parse stage + CXzStreamFlags streamFlags; + // UInt64 numFinishedStreams + UInt64 numStreams; + UInt64 numTotalBlocks; + UInt64 numBlocks; + + // UInt64 numBadBlocks; + SRes mainErrorCode; + + BoolInt isBlockHeaderState_Parse; + BoolInt isBlockHeaderState_Write; + UInt64 outProcessed_Parse; + BoolInt parsing_Truncated; + + BoolInt mtc_WasConstructed; + CMtDec mtc; + CXzDecMtThread coders[MTDEC__THREADS_MAX]; + #endif + +} CXzDecMt; + + + +CXzDecMtHandle XzDecMt_Create(ISzAllocPtr alloc, ISzAllocPtr allocMid) +{ + CXzDecMt *p = (CXzDecMt *)ISzAlloc_Alloc(alloc, sizeof(CXzDecMt)); + if (!p) + return NULL; + + AlignOffsetAlloc_CreateVTable(&p->alignOffsetAlloc); + p->alignOffsetAlloc.baseAlloc = alloc; + p->alignOffsetAlloc.numAlignBits = 7; + p->alignOffsetAlloc.offset = 0; + + p->allocMid = allocMid; + + p->outBuf = NULL; + p->outBufSize = 0; + p->inBuf = NULL; + p->inBufSize = 0; + + XzUnpacker_Construct(&p->dec, &p->alignOffsetAlloc.vt); + + p->unpackBlockMaxSize = 0; + + XzDecMtProps_Init(&p->props); + + #ifndef _7ZIP_ST + p->mtc_WasConstructed = False; + { + unsigned i; + for (i = 0; i < MTDEC__THREADS_MAX; i++) + { + CXzDecMtThread *coder = &p->coders[i]; + coder->dec_created = False; + coder->outBuf = NULL; + coder->outBufSize = 0; + } + } + #endif + + return p; +} + + +#ifndef _7ZIP_ST + +static void XzDecMt_FreeOutBufs(CXzDecMt *p) +{ + unsigned i; + for (i = 0; i < MTDEC__THREADS_MAX; i++) + { + CXzDecMtThread *coder = &p->coders[i]; + if (coder->outBuf) + { + ISzAlloc_Free(p->allocMid, coder->outBuf); + coder->outBuf = NULL; + coder->outBufSize = 0; + } + } + p->unpackBlockMaxSize = 0; +} + +#endif + + + +static void XzDecMt_FreeSt(CXzDecMt *p) +{ + XzUnpacker_Free(&p->dec); + + if (p->outBuf) + { + ISzAlloc_Free(p->allocMid, p->outBuf); + p->outBuf = NULL; + } + p->outBufSize = 0; + + if (p->inBuf) + { + ISzAlloc_Free(p->allocMid, p->inBuf); + p->inBuf = NULL; + } + p->inBufSize = 0; +} + + +void XzDecMt_Destroy(CXzDecMtHandle pp) +{ + CXzDecMt *p = (CXzDecMt *)pp; + + XzDecMt_FreeSt(p); + + #ifndef _7ZIP_ST + + if (p->mtc_WasConstructed) + { + MtDec_Destruct(&p->mtc); + p->mtc_WasConstructed = False; + } + { + unsigned i; + for (i = 0; i < MTDEC__THREADS_MAX; i++) + { + CXzDecMtThread *t = &p->coders[i]; + if (t->dec_created) + { + // we don't need to free dict here + XzUnpacker_Free(&t->dec); + t->dec_created = False; + } + } + } + XzDecMt_FreeOutBufs(p); + + #endif + + ISzAlloc_Free(p->alignOffsetAlloc.baseAlloc, pp); +} + + + +#ifndef _7ZIP_ST + +static void XzDecMt_Callback_Parse(void *obj, unsigned coderIndex, CMtDecCallbackInfo *cc) +{ + CXzDecMt *me = (CXzDecMt *)obj; + CXzDecMtThread *coder = &me->coders[coderIndex]; + size_t srcSize = cc->srcSize; + + cc->srcSize = 0; + cc->outPos = 0; + cc->state = MTDEC_PARSE_CONTINUE; + + cc->canCreateNewThread = True; + + if (cc->startCall) + { + coder->outPreSize = 0; + coder->inPreSize = 0; + coder->inPreHeaderSize = 0; + coder->parseState = MTDEC_PARSE_CONTINUE; + coder->parsing_Truncated = False; + coder->skipMode = False; + coder->codeRes = SZ_OK; + coder->status = CODER_STATUS_NOT_SPECIFIED; + coder->inCodeSize = 0; + coder->outCodeSize = 0; + + coder->numStreams = me->numStreams; + coder->numTotalBlocks = me->numTotalBlocks; + coder->numBlocks = me->numBlocks; + + if (!coder->dec_created) + { + XzUnpacker_Construct(&coder->dec, &me->alignOffsetAlloc.vt); + coder->dec_created = True; + } + + XzUnpacker_Init(&coder->dec); + + if (me->isBlockHeaderState_Parse) + { + coder->dec.streamFlags = me->streamFlags; + coder->atBlockHeader = True; + XzUnpacker_PrepareToRandomBlockDecoding(&coder->dec); + } + else + { + coder->atBlockHeader = False; + me->isBlockHeaderState_Parse = True; + } + + coder->dec.numStartedStreams = me->numStreams; + coder->dec.numTotalBlocks = me->numTotalBlocks; + coder->dec.numBlocks = me->numBlocks; + } + + while (!coder->skipMode) + { + ECoderStatus status; + SRes res; + size_t srcSize2 = srcSize; + size_t destSize = (size_t)0 - 1; + + coder->dec.parseMode = True; + coder->dec.headerParsedOk = False; + + PRF_STR_INT("Parse", srcSize2); + + res = XzUnpacker_Code(&coder->dec, + NULL, &destSize, + cc->src, &srcSize2, cc->srcFinished, + CODER_FINISH_END, &status); + + // PRF(printf(" res = %d, srcSize2 = %d", res, (unsigned)srcSize2)); + + coder->codeRes = res; + coder->status = status; + cc->srcSize += srcSize2; + srcSize -= srcSize2; + coder->inPreHeaderSize += srcSize2; + coder->inPreSize = coder->inPreHeaderSize; + + if (res != SZ_OK) + { + cc->state = + coder->parseState = MTDEC_PARSE_END; + /* + if (res == SZ_ERROR_MEM) + return res; + return SZ_OK; + */ + return; // res; + } + + if (coder->dec.headerParsedOk) + { + const CXzBlock *block = &coder->dec.block; + if (XzBlock_HasUnpackSize(block) + // && block->unpackSize <= me->props.outBlockMax + && XzBlock_HasPackSize(block)) + { + { + if (block->unpackSize * 2 * me->mtc.numStartedThreads > me->props.memUseMax) + { + cc->state = MTDEC_PARSE_OVERFLOW; + return; // SZ_OK; + } + } + { + UInt64 packSize = block->packSize; + UInt64 packSizeAligned = packSize + ((0 - (unsigned)packSize) & 3); + UInt32 checkSize = XzFlags_GetCheckSize(coder->dec.streamFlags); + UInt64 blockPackSum = coder->inPreSize + packSizeAligned + checkSize; + // if (blockPackSum <= me->props.inBlockMax) + // unpackBlockMaxSize + { + coder->blockPackSize_for_Index = (size_t)(coder->dec.blockHeaderSize + packSize + checkSize); + coder->blockPackTotal = (size_t)blockPackSum; + coder->outPreSize = (size_t)block->unpackSize; + coder->streamFlags = coder->dec.streamFlags; + me->streamFlags = coder->dec.streamFlags; + coder->skipMode = True; + break; + } + } + } + } + else + // if (coder->inPreSize <= me->props.inBlockMax) + { + if (!cc->srcFinished) + return; // SZ_OK; + cc->state = + coder->parseState = MTDEC_PARSE_END; + return; // SZ_OK; + } + cc->state = MTDEC_PARSE_OVERFLOW; + return; // SZ_OK; + } + + // ---------- skipMode ---------- + { + UInt64 rem = coder->blockPackTotal - coder->inPreSize; + size_t cur = srcSize; + if (cur > rem) + cur = (size_t)rem; + cc->srcSize += cur; + coder->inPreSize += cur; + srcSize -= cur; + + if (coder->inPreSize == coder->blockPackTotal) + { + if (srcSize == 0) + { + if (!cc->srcFinished) + return; // SZ_OK; + cc->state = MTDEC_PARSE_END; + } + else if ((cc->src)[cc->srcSize] == 0) // we check control byte of next block + cc->state = MTDEC_PARSE_END; + else + { + cc->state = MTDEC_PARSE_NEW; + + { + size_t blockMax = me->unpackBlockMaxSize; + if (blockMax < coder->outPreSize) + blockMax = coder->outPreSize; + { + UInt64 required = (UInt64)blockMax * (me->mtc.numStartedThreads + 1) * 2; + if (me->props.memUseMax < required) + cc->canCreateNewThread = False; + } + } + + if (me->outSize_Defined) + { + // next block can be zero size + const UInt64 rem2 = me->outSize - me->outProcessed_Parse; + if (rem2 < coder->outPreSize) + { + coder->parsing_Truncated = True; + cc->state = MTDEC_PARSE_END; + } + me->outProcessed_Parse += coder->outPreSize; + } + } + } + else if (cc->srcFinished) + cc->state = MTDEC_PARSE_END; + else + return; // SZ_OK; + + coder->parseState = cc->state; + cc->outPos = coder->outPreSize; + + me->numStreams = coder->dec.numStartedStreams; + me->numTotalBlocks = coder->dec.numTotalBlocks; + me->numBlocks = coder->dec.numBlocks + 1; + return; // SZ_OK; + } +} + + +static SRes XzDecMt_Callback_PreCode(void *pp, unsigned coderIndex) +{ + CXzDecMt *me = (CXzDecMt *)pp; + CXzDecMtThread *coder = &me->coders[coderIndex]; + Byte *dest; + + if (!coder->dec.headerParsedOk) + return SZ_OK; + + dest = coder->outBuf; + + if (!dest || coder->outBufSize < coder->outPreSize) + { + if (dest) + { + ISzAlloc_Free(me->allocMid, dest); + coder->outBuf = NULL; + coder->outBufSize = 0; + } + { + size_t outPreSize = coder->outPreSize; + if (outPreSize == 0) + outPreSize = 1; + dest = (Byte *)ISzAlloc_Alloc(me->allocMid, outPreSize); + } + if (!dest) + return SZ_ERROR_MEM; + coder->outBuf = dest; + coder->outBufSize = coder->outPreSize; + + if (coder->outBufSize > me->unpackBlockMaxSize) + me->unpackBlockMaxSize = coder->outBufSize; + } + + // return SZ_ERROR_MEM; + + XzUnpacker_SetOutBuf(&coder->dec, coder->outBuf, coder->outBufSize); + + { + SRes res = XzDecMix_Init(&coder->dec.decoder, &coder->dec.block, coder->outBuf, coder->outBufSize); + // res = SZ_ERROR_UNSUPPORTED; // to test + coder->codeRes = res; + if (res != SZ_OK) + { + // if (res == SZ_ERROR_MEM) return res; + if (me->props.ignoreErrors && res != SZ_ERROR_MEM) + return S_OK; + return res; + } + } + + return SZ_OK; +} + + +static SRes XzDecMt_Callback_Code(void *pp, unsigned coderIndex, + const Byte *src, size_t srcSize, int srcFinished, + // int finished, int blockFinished, + UInt64 *inCodePos, UInt64 *outCodePos, int *stop) +{ + CXzDecMt *me = (CXzDecMt *)pp; + CXzDecMtThread *coder = &me->coders[coderIndex]; + + *inCodePos = coder->inCodeSize; + *outCodePos = coder->outCodeSize; + *stop = True; + + if (coder->inCodeSize < coder->inPreHeaderSize) + { + UInt64 rem = coder->inPreHeaderSize - coder->inCodeSize; + size_t step = srcSize; + if (step > rem) + step = (size_t)rem; + src += step; + srcSize -= step; + coder->inCodeSize += step; + if (coder->inCodeSize < coder->inPreHeaderSize) + { + *stop = False; + return SZ_OK; + } + } + + if (!coder->dec.headerParsedOk) + return SZ_OK; + if (!coder->outBuf) + return SZ_OK; + + if (coder->codeRes == SZ_OK) + { + ECoderStatus status; + SRes res; + size_t srcProcessed = srcSize; + size_t outSizeCur = coder->outPreSize - coder->dec.outDataWritten; + + // PRF(printf("\nCallback_Code: Code %d %d\n", (unsigned)srcSize, (unsigned)outSizeCur)); + + res = XzUnpacker_Code(&coder->dec, + NULL, &outSizeCur, + src, &srcProcessed, srcFinished, + // coder->finishedWithMark ? CODER_FINISH_END : CODER_FINISH_ANY, + CODER_FINISH_END, + &status); + + // PRF(printf(" res = %d, srcSize2 = %d, outSizeCur = %d", res, (unsigned)srcProcessed, (unsigned)outSizeCur)); + + coder->codeRes = res; + coder->status = status; + coder->inCodeSize += srcProcessed; + coder->outCodeSize = coder->dec.outDataWritten; + *inCodePos = coder->inCodeSize; + *outCodePos = coder->outCodeSize; + + if (res == SZ_OK) + { + if (srcProcessed == srcSize) + *stop = False; + return SZ_OK; + } + } + + if (me->props.ignoreErrors && coder->codeRes != SZ_ERROR_MEM) + { + *inCodePos = coder->inPreSize; + *outCodePos = coder->outPreSize; + return S_OK; + } + return coder->codeRes; +} + + +#define XZDECMT_STREAM_WRITE_STEP (1 << 24) + +static SRes XzDecMt_Callback_Write(void *pp, unsigned coderIndex, + BoolInt needWriteToStream, + const Byte *src, size_t srcSize, + // int srcFinished, + BoolInt *needContinue, + BoolInt *canRecode) +{ + CXzDecMt *me = (CXzDecMt *)pp; + const CXzDecMtThread *coder = &me->coders[coderIndex]; + + // PRF(printf("\nWrite processed = %d srcSize = %d\n", (unsigned)me->mtc.inProcessed, (unsigned)srcSize)); + + *needContinue = False; + *canRecode = True; + + if (!needWriteToStream) + return SZ_OK; + + if (!coder->dec.headerParsedOk || !coder->outBuf) + { + if (me->finishedDecoderIndex < 0) + me->finishedDecoderIndex = coderIndex; + return SZ_OK; + } + + if (me->finishedDecoderIndex >= 0) + return SZ_OK; + + me->mtc.inProcessed += coder->inCodeSize; + + *canRecode = False; + + { + SRes res; + size_t size = coder->outCodeSize; + Byte *data = coder->outBuf; + + // we use in me->dec: sha, numBlocks, indexSize + + if (!me->isBlockHeaderState_Write) + { + XzUnpacker_PrepareToRandomBlockDecoding(&me->dec); + me->dec.decodeOnlyOneBlock = False; + me->dec.numStartedStreams = coder->dec.numStartedStreams; + me->dec.streamFlags = coder->streamFlags; + + me->isBlockHeaderState_Write = True; + } + + me->dec.numTotalBlocks = coder->dec.numTotalBlocks; + XzUnpacker_UpdateIndex(&me->dec, coder->blockPackSize_for_Index, coder->outPreSize); + + if (coder->outPreSize != size) + { + if (me->props.ignoreErrors) + { + memset(data + size, 0, coder->outPreSize - size); + size = coder->outPreSize; + } + // me->numBadBlocks++; + if (me->mainErrorCode == SZ_OK) + { + if ((int)coder->status == LZMA_STATUS_NEEDS_MORE_INPUT) + me->mainErrorCode = SZ_ERROR_INPUT_EOF; + else + me->mainErrorCode = SZ_ERROR_DATA; + } + } + + if (me->writeRes != SZ_OK) + return me->writeRes; + + res = SZ_OK; + { + if (me->outSize_Defined) + { + const UInt64 rem = me->outSize - me->outProcessed; + if (size > rem) + size = (SizeT)rem; + } + + for (;;) + { + size_t cur = size; + size_t written; + if (cur > XZDECMT_STREAM_WRITE_STEP) + cur = XZDECMT_STREAM_WRITE_STEP; + + written = ISeqOutStream_Write(me->outStream, data, cur); + + // PRF(printf("\nWritten ask = %d written = %d\n", (unsigned)cur, (unsigned)written)); + + me->outProcessed += written; + if (written != cur) + { + me->writeRes = SZ_ERROR_WRITE; + res = me->writeRes; + break; + } + data += cur; + size -= cur; + // PRF_STR_INT("Written size =", size); + if (size == 0) + break; + res = MtProgress_ProgressAdd(&me->mtc.mtProgress, 0, 0); + if (res != SZ_OK) + break; + } + } + + if (coder->codeRes != SZ_OK) + if (!me->props.ignoreErrors) + { + me->finishedDecoderIndex = coderIndex; + return res; + } + + RINOK(res); + + if (coder->inPreSize != coder->inCodeSize + || coder->blockPackTotal != coder->inCodeSize) + { + me->finishedDecoderIndex = coderIndex; + return SZ_OK; + } + + if (coder->parseState != MTDEC_PARSE_END) + { + *needContinue = True; + return SZ_OK; + } + } + + // (coder->state == MTDEC_PARSE_END) means that there are no other working threads + // so we can use mtc variables without lock + + PRF_STR_INT("Write MTDEC_PARSE_END", me->mtc.inProcessed); + + me->mtc.mtProgress.totalInSize = me->mtc.inProcessed; + { + CXzUnpacker *dec = &me->dec; + + PRF_STR_INT("PostSingle", srcSize); + + { + size_t srcProcessed = srcSize; + ECoderStatus status; + size_t outSizeCur = 0; + SRes res; + + // dec->decodeOnlyOneBlock = False; + dec->decodeToStreamSignature = True; + + me->mainDecoderWasCalled = True; + + if (coder->parsing_Truncated) + { + me->parsing_Truncated = True; + return SZ_OK; + } + + res = XzUnpacker_Code(dec, + NULL, &outSizeCur, + src, &srcProcessed, + me->mtc.readWasFinished, // srcFinished + CODER_FINISH_END, // CODER_FINISH_ANY, + &status); + + me->status = status; + me->codeRes = res; + + me->mtc.inProcessed += srcProcessed; + me->mtc.mtProgress.totalInSize = me->mtc.inProcessed; + + if (res != SZ_OK) + { + return S_OK; + // return res; + } + + if (dec->state == XZ_STATE_STREAM_HEADER) + { + *needContinue = True; + me->isBlockHeaderState_Parse = False; + me->isBlockHeaderState_Write = False; + { + Byte *crossBuf = MtDec_GetCrossBuff(&me->mtc); + if (!crossBuf) + return SZ_ERROR_MEM; + memcpy(crossBuf, src + srcProcessed, srcSize - srcProcessed); + } + me->mtc.crossStart = 0; + me->mtc.crossEnd = srcSize - srcProcessed; + return SZ_OK; + } + + if (status != CODER_STATUS_NEEDS_MORE_INPUT) + { + return E_FAIL; + } + + if (me->mtc.readWasFinished) + { + return SZ_OK; + } + } + + { + size_t inPos; + size_t inLim; + const Byte *inData; + UInt64 inProgressPrev = me->mtc.inProcessed; + + // XzDecMt_Prepare_InBuf_ST(p); + Byte *crossBuf = MtDec_GetCrossBuff(&me->mtc); + if (!crossBuf) + return SZ_ERROR_MEM; + + inPos = 0; + inLim = 0; + // outProcessed = 0; + + inData = crossBuf; + + for (;;) + { + SizeT inProcessed; + SizeT outProcessed; + ECoderStatus status; + SRes res; + + if (inPos == inLim) + { + if (!me->mtc.readWasFinished) + { + inPos = 0; + inLim = me->mtc.inBufSize; + me->mtc.readRes = ISeqInStream_Read(me->inStream, (void *)inData, &inLim); + me->mtc.readProcessed += inLim; + if (inLim == 0 || me->mtc.readRes != SZ_OK) + me->mtc.readWasFinished = True; + } + } + + inProcessed = inLim - inPos; + outProcessed = 0; + + res = XzUnpacker_Code(dec, + NULL, &outProcessed, + inData + inPos, &inProcessed, + (inProcessed == 0), // srcFinished + CODER_FINISH_END, &status); + + me->codeRes = res; + me->status = status; + inPos += inProcessed; + me->mtc.inProcessed += inProcessed; + me->mtc.mtProgress.totalInSize = me->mtc.inProcessed; + + if (res != SZ_OK) + { + return S_OK; + // return res; + } + + if (dec->state == XZ_STATE_STREAM_HEADER) + { + *needContinue = True; + me->mtc.crossStart = inPos; + me->mtc.crossEnd = inLim; + me->isBlockHeaderState_Parse = False; + me->isBlockHeaderState_Write = False; + return SZ_OK; + } + + if (status != CODER_STATUS_NEEDS_MORE_INPUT) + return E_FAIL; + + if (me->mtc.progress) + { + UInt64 inDelta = me->mtc.inProcessed - inProgressPrev; + if (inDelta >= (1 << 22)) + { + RINOK(MtProgress_Progress_ST(&me->mtc.mtProgress)); + inProgressPrev = me->mtc.inProcessed; + } + } + if (me->mtc.readWasFinished) + return SZ_OK; + } + } + } +} + + +#endif + + + +void XzStatInfo_Clear(CXzStatInfo *p) +{ + p->InSize = 0; + p->OutSize = 0; + + p->NumStreams = 0; + p->NumBlocks = 0; + + p->UnpackSize_Defined = False; + + p->NumStreams_Defined = False; + p->NumBlocks_Defined = False; + + // p->IsArc = False; + // p->UnexpectedEnd = False; + // p->Unsupported = False; + // p->HeadersError = False; + // p->DataError = False; + // p->CrcError = False; + + p->DataAfterEnd = False; + p->DecodingTruncated = False; + + p->DecodeRes = SZ_OK; + p->ReadRes = SZ_OK; + p->ProgressRes = SZ_OK; + + p->CombinedRes = SZ_OK; + p->CombinedRes_Type = SZ_OK; +} + + + + +static SRes XzDecMt_Decode_ST(CXzDecMt *p + #ifndef _7ZIP_ST + , BoolInt tMode + #endif + , CXzStatInfo *stat) +{ + size_t outPos; + size_t inPos, inLim; + const Byte *inData; + UInt64 inPrev, outPrev; + + CXzUnpacker *dec; + + #ifndef _7ZIP_ST + if (tMode) + { + XzDecMt_FreeOutBufs(p); + tMode = MtDec_PrepareRead(&p->mtc); + } + #endif + + if (!p->outBuf || p->outBufSize != p->props.outStep_ST) + { + ISzAlloc_Free(p->allocMid, p->outBuf); + p->outBufSize = 0; + p->outBuf = (Byte *)ISzAlloc_Alloc(p->allocMid, p->props.outStep_ST); + if (!p->outBuf) + return SZ_ERROR_MEM; + p->outBufSize = p->props.outStep_ST; + } + + if (!p->inBuf || p->inBufSize != p->props.inBufSize_ST) + { + ISzAlloc_Free(p->allocMid, p->inBuf); + p->inBufSize = 0; + p->inBuf = (Byte *)ISzAlloc_Alloc(p->allocMid, p->props.inBufSize_ST); + if (!p->inBuf) + return SZ_ERROR_MEM; + p->inBufSize = p->props.inBufSize_ST; + } + + dec = &p->dec; + dec->decodeToStreamSignature = False; + // dec->decodeOnlyOneBlock = False; + + XzUnpacker_SetOutBuf(dec, NULL, 0); + + inPrev = p->inProcessed; + outPrev = p->outProcessed; + + inPos = 0; + inLim = 0; + inData = NULL; + outPos = 0; + + for (;;) + { + SizeT outSize; + BoolInt finished; + ECoderFinishMode finishMode; + SizeT inProcessed; + ECoderStatus status; + SRes res; + + SizeT outProcessed; + + + + if (inPos == inLim) + { + #ifndef _7ZIP_ST + if (tMode) + { + inData = MtDec_Read(&p->mtc, &inLim); + inPos = 0; + if (inData) + continue; + tMode = False; + inLim = 0; + } + #endif + + if (!p->readWasFinished) + { + inPos = 0; + inLim = p->inBufSize; + inData = p->inBuf; + p->readRes = ISeqInStream_Read(p->inStream, (void *)inData, &inLim); + p->readProcessed += inLim; + if (inLim == 0 || p->readRes != SZ_OK) + p->readWasFinished = True; + } + } + + outSize = p->props.outStep_ST - outPos; + + finishMode = CODER_FINISH_ANY; + if (p->outSize_Defined) + { + const UInt64 rem = p->outSize - p->outProcessed; + if (outSize >= rem) + { + outSize = (SizeT)rem; + if (p->finishMode) + finishMode = CODER_FINISH_END; + } + } + + inProcessed = inLim - inPos; + outProcessed = outSize; + + res = XzUnpacker_Code(dec, p->outBuf + outPos, &outProcessed, + inData + inPos, &inProcessed, + (inPos == inLim), // srcFinished + finishMode, &status); + + p->codeRes = res; + p->status = status; + + inPos += inProcessed; + outPos += outProcessed; + p->inProcessed += inProcessed; + p->outProcessed += outProcessed; + + finished = ((inProcessed == 0 && outProcessed == 0) || res != SZ_OK); + + if (finished || outProcessed >= outSize) + if (outPos != 0) + { + size_t written = ISeqOutStream_Write(p->outStream, p->outBuf, outPos); + p->outProcessed += written; + if (written != outPos) + { + stat->CombinedRes_Type = SZ_ERROR_WRITE; + return SZ_ERROR_WRITE; + } + outPos = 0; + } + + if (p->progress && res == SZ_OK) + { + UInt64 inDelta = p->inProcessed - inPrev; + UInt64 outDelta = p->outProcessed - outPrev; + if (inDelta >= (1 << 22) || outDelta >= (1 << 22)) + { + res = ICompressProgress_Progress(p->progress, p->inProcessed, p->outProcessed); + if (res != SZ_OK) + { + stat->CombinedRes_Type = SZ_ERROR_PROGRESS; + stat->ProgressRes = res; + return res; + } + inPrev = p->inProcessed; + outPrev = p->outProcessed; + } + } + + if (finished) + return res; + } +} + +static SRes XzStatInfo_SetStat(const CXzUnpacker *dec, + int finishMode, + UInt64 readProcessed, UInt64 inProcessed, + SRes res, ECoderStatus status, + BoolInt decodingTruncated, + CXzStatInfo *stat) +{ + UInt64 extraSize; + + stat->DecodingTruncated = (Byte)(decodingTruncated ? 1 : 0); + stat->InSize = inProcessed; + stat->NumStreams = dec->numStartedStreams; + stat->NumBlocks = dec->numTotalBlocks; + + stat->UnpackSize_Defined = True; + stat->NumStreams_Defined = True; + stat->NumBlocks_Defined = True; + + extraSize = XzUnpacker_GetExtraSize(dec); + + if (res == SZ_OK) + { + if (status == CODER_STATUS_NEEDS_MORE_INPUT) + { + // CODER_STATUS_NEEDS_MORE_INPUT is expected status for correct xz streams + extraSize = 0; + if (!XzUnpacker_IsStreamWasFinished(dec)) + res = SZ_ERROR_INPUT_EOF; + } + else if (!decodingTruncated || finishMode) // (status == CODER_STATUS_NOT_FINISHED) + res = SZ_ERROR_DATA; + } + else if (res == SZ_ERROR_NO_ARCHIVE) + { + /* + SZ_ERROR_NO_ARCHIVE is possible for 2 states: + XZ_STATE_STREAM_HEADER - if bad signature or bad CRC + XZ_STATE_STREAM_PADDING - if non-zero padding data + extraSize / inProcessed don't include "bad" byte + */ + if (inProcessed != extraSize) // if good streams before error + if (extraSize != 0 || readProcessed != inProcessed) + { + stat->DataAfterEnd = True; + // there is some good xz stream before. So we set SZ_OK + res = SZ_OK; + } + } + + stat->DecodeRes = res; + + stat->InSize -= extraSize; + return res; +} + + +SRes XzDecMt_Decode(CXzDecMtHandle pp, + const CXzDecMtProps *props, + const UInt64 *outDataSize, int finishMode, + ISeqOutStream *outStream, + // Byte *outBuf, size_t *outBufSize, + ISeqInStream *inStream, + // const Byte *inData, size_t inDataSize, + CXzStatInfo *stat, + int *isMT, + ICompressProgress *progress) +{ + CXzDecMt *p = (CXzDecMt *)pp; + #ifndef _7ZIP_ST + BoolInt tMode; + #endif + + XzStatInfo_Clear(stat); + + p->props = *props; + + p->inStream = inStream; + p->outStream = outStream; + p->progress = progress; + // p->stat = stat; + + p->outSize = 0; + p->outSize_Defined = False; + if (outDataSize) + { + p->outSize_Defined = True; + p->outSize = *outDataSize; + } + + p->finishMode = finishMode; + + // p->outSize = 457; p->outSize_Defined = True; p->finishMode = False; // for test + + p->writeRes = SZ_OK; + p->outProcessed = 0; + p->inProcessed = 0; + p->readProcessed = 0; + p->readWasFinished = False; + + p->codeRes = 0; + p->status = CODER_STATUS_NOT_SPECIFIED; + + XzUnpacker_Init(&p->dec); + + *isMT = False; + + /* + p->outBuf = NULL; + p->outBufSize = 0; + if (!outStream) + { + p->outBuf = outBuf; + p->outBufSize = *outBufSize; + *outBufSize = 0; + } + */ + + + #ifndef _7ZIP_ST + + p->isBlockHeaderState_Parse = False; + p->isBlockHeaderState_Write = False; + // p->numBadBlocks = 0; + p->mainErrorCode = SZ_OK; + p->mainDecoderWasCalled = False; + + tMode = False; + + if (p->props.numThreads > 1) + { + IMtDecCallback vt; + + // we just free ST buffers here + // but we still keep state variables, that was set in XzUnpacker_Init() + XzDecMt_FreeSt(p); + + p->outProcessed_Parse = 0; + p->parsing_Truncated = False; + + p->numStreams = 0; + p->numTotalBlocks = 0; + p->numBlocks = 0; + p->finishedDecoderIndex = -1; + + if (!p->mtc_WasConstructed) + { + p->mtc_WasConstructed = True; + MtDec_Construct(&p->mtc); + } + + p->mtc.mtCallback = &vt; + p->mtc.mtCallbackObject = p; + + p->mtc.progress = progress; + p->mtc.inStream = inStream; + p->mtc.alloc = &p->alignOffsetAlloc.vt; + // p->mtc.inData = inData; + // p->mtc.inDataSize = inDataSize; + p->mtc.inBufSize = p->props.inBufSize_MT; + // p->mtc.inBlockMax = p->props.inBlockMax; + p->mtc.numThreadsMax = p->props.numThreads; + + *isMT = True; + + vt.Parse = XzDecMt_Callback_Parse; + vt.PreCode = XzDecMt_Callback_PreCode; + vt.Code = XzDecMt_Callback_Code; + vt.Write = XzDecMt_Callback_Write; + + { + BoolInt needContinue; + + SRes res = MtDec_Code(&p->mtc); + + stat->InSize = p->mtc.inProcessed; + + p->inProcessed = p->mtc.inProcessed; + p->readRes = p->mtc.readRes; + p->readWasFinished = p->mtc.readWasFinished; + p->readProcessed = p->mtc.readProcessed; + + tMode = True; + needContinue = False; + + if (res == SZ_OK) + { + if (p->mtc.mtProgress.res != SZ_OK) + { + res = p->mtc.mtProgress.res; + stat->ProgressRes = res; + stat->CombinedRes_Type = SZ_ERROR_PROGRESS; + } + else + needContinue = p->mtc.needContinue; + } + + if (!needContinue) + { + SRes codeRes; + BoolInt truncated = False; + ECoderStatus status; + CXzUnpacker *dec; + + stat->OutSize = p->outProcessed; + + if (p->finishedDecoderIndex >= 0) + { + CXzDecMtThread *coder = &p->coders[(unsigned)p->finishedDecoderIndex]; + codeRes = coder->codeRes; + dec = &coder->dec; + status = coder->status; + } + else if (p->mainDecoderWasCalled) + { + codeRes = p->codeRes; + dec = &p->dec; + status = p->status; + truncated = p->parsing_Truncated; + } + else + return E_FAIL; + + XzStatInfo_SetStat(dec, p->finishMode, + p->mtc.readProcessed, p->mtc.inProcessed, + codeRes, status, + truncated, + stat); + + if (res == SZ_OK) + { + if (p->writeRes != SZ_OK) + { + res = p->writeRes; + stat->CombinedRes_Type = SZ_ERROR_WRITE; + } + else if (p->mtc.readRes != SZ_OK && p->mtc.inProcessed == p->mtc.readProcessed) + { + res = p->mtc.readRes; + stat->ReadRes = res; + stat->CombinedRes_Type = SZ_ERROR_READ; + } + else if (p->mainErrorCode != SZ_OK) + { + res = p->mainErrorCode; + } + } + + stat->CombinedRes = res; + if (stat->CombinedRes_Type == SZ_OK) + stat->CombinedRes_Type = res; + return res; + } + + PRF_STR("----- decoding ST -----"); + } + } + + #endif + + + *isMT = False; + + { + SRes res = XzDecMt_Decode_ST(p + #ifndef _7ZIP_ST + , tMode + #endif + , stat + ); + + XzStatInfo_SetStat(&p->dec, + p->finishMode, + p->readProcessed, p->inProcessed, + p->codeRes, p->status, + False, // truncated + stat); + + if (res == SZ_OK) + { + /* + if (p->writeRes != SZ_OK) + { + res = p->writeRes; + stat->CombinedRes_Type = SZ_ERROR_WRITE; + } + else + */ + if (p->readRes != SZ_OK && p->inProcessed == p->readProcessed) + { + res = p->readRes; + stat->ReadRes = res; + stat->CombinedRes_Type = SZ_ERROR_READ; + } + #ifndef _7ZIP_ST + else if (p->mainErrorCode != SZ_OK) + res = p->mainErrorCode; + #endif + } + + stat->CombinedRes = res; + if (stat->CombinedRes_Type == SZ_OK) + stat->CombinedRes_Type = res; + return res; + } +} diff --git a/src/third-party/lzma/XzEnc.c b/src/third-party/lzma/XzEnc.c index 2e5a003ad..309eca949 100644 --- a/src/third-party/lzma/XzEnc.c +++ b/src/third-party/lzma/XzEnc.c @@ -1,5 +1,5 @@ /* XzEnc.c -- Xz Encode -2015-09-16 : Igor Pavlov : Public domain */ +2019-02-02 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -7,7 +7,6 @@ #include #include "7zCrc.h" -#include "Alloc.h" #include "Bra.h" #include "CpuArch.h" @@ -19,22 +18,41 @@ #include "XzEnc.h" +// #define _7ZIP_ST + +#ifndef _7ZIP_ST +#include "MtCoder.h" +#else +#define MTCODER__THREADS_MAX 1 +#define MTCODER__BLOCKS_MAX 1 +#endif + +#define XZ_GET_PAD_SIZE(dataSize) ((4 - ((unsigned)(dataSize) & 3)) & 3) + +/* max pack size for LZMA2 block + check-64bytrs: */ +#define XZ_GET_MAX_BLOCK_PACK_SIZE(unpackSize) ((unpackSize) + ((unpackSize) >> 10) + 16 + 64) + +#define XZ_GET_ESTIMATED_BLOCK_TOTAL_PACK_SIZE(unpackSize) (XZ_BLOCK_HEADER_SIZE_MAX + XZ_GET_MAX_BLOCK_PACK_SIZE(unpackSize)) + + #define XzBlock_ClearFlags(p) (p)->flags = 0; #define XzBlock_SetNumFilters(p, n) (p)->flags |= ((n) - 1); #define XzBlock_SetHasPackSize(p) (p)->flags |= XZ_BF_PACK_SIZE; #define XzBlock_SetHasUnpackSize(p) (p)->flags |= XZ_BF_UNPACK_SIZE; -static SRes WriteBytes(ISeqOutStream *s, const void *buf, UInt32 size) + +static SRes WriteBytes(ISeqOutStream *s, const void *buf, size_t size) { - return (s->Write(s, buf, size) == size) ? SZ_OK : SZ_ERROR_WRITE; + return (ISeqOutStream_Write(s, buf, size) == size) ? SZ_OK : SZ_ERROR_WRITE; } -static SRes WriteBytesAndCrc(ISeqOutStream *s, const void *buf, UInt32 size, UInt32 *crc) +static SRes WriteBytesUpdateCrc(ISeqOutStream *s, const void *buf, size_t size, UInt32 *crc) { *crc = CrcUpdate(*crc, buf, size); return WriteBytes(s, buf, size); } + static SRes Xz_WriteHeader(CXzStreamFlags f, ISeqOutStream *s) { UInt32 crc; @@ -78,98 +96,154 @@ static SRes XzBlock_WriteHeader(const CXzBlock *p, ISeqOutStream *s) } -static SRes Xz_WriteFooter(CXzStream *p, ISeqOutStream *s) + + +typedef struct +{ + size_t numBlocks; + size_t size; + size_t allocated; + Byte *blocks; +} CXzEncIndex; + + +static void XzEncIndex_Construct(CXzEncIndex *p) +{ + p->numBlocks = 0; + p->size = 0; + p->allocated = 0; + p->blocks = NULL; +} + +static void XzEncIndex_Init(CXzEncIndex *p) +{ + p->numBlocks = 0; + p->size = 0; +} + +static void XzEncIndex_Free(CXzEncIndex *p, ISzAllocPtr alloc) +{ + if (p->blocks) + { + ISzAlloc_Free(alloc, p->blocks); + p->blocks = NULL; + } + p->numBlocks = 0; + p->size = 0; + p->allocated = 0; +} + + +static SRes XzEncIndex_ReAlloc(CXzEncIndex *p, size_t newSize, ISzAllocPtr alloc) +{ + Byte *blocks = (Byte *)ISzAlloc_Alloc(alloc, newSize); + if (!blocks) + return SZ_ERROR_MEM; + if (p->size != 0) + memcpy(blocks, p->blocks, p->size); + if (p->blocks) + ISzAlloc_Free(alloc, p->blocks); + p->blocks = blocks; + p->allocated = newSize; + return SZ_OK; +} + + +static SRes XzEncIndex_PreAlloc(CXzEncIndex *p, UInt64 numBlocks, UInt64 unpackSize, UInt64 totalSize, ISzAllocPtr alloc) +{ + UInt64 pos; + { + Byte buf[32]; + unsigned pos2 = Xz_WriteVarInt(buf, totalSize); + pos2 += Xz_WriteVarInt(buf + pos2, unpackSize); + pos = numBlocks * pos2; + } + + if (pos <= p->allocated - p->size) + return SZ_OK; + { + UInt64 newSize64 = p->size + pos; + size_t newSize = (size_t)newSize64; + if (newSize != newSize64) + return SZ_ERROR_MEM; + return XzEncIndex_ReAlloc(p, newSize, alloc); + } +} + + +static SRes XzEncIndex_AddIndexRecord(CXzEncIndex *p, UInt64 unpackSize, UInt64 totalSize, ISzAllocPtr alloc) +{ + Byte buf[32]; + unsigned pos = Xz_WriteVarInt(buf, totalSize); + pos += Xz_WriteVarInt(buf + pos, unpackSize); + + if (pos > p->allocated - p->size) + { + size_t newSize = p->allocated * 2 + 16 * 2; + if (newSize < p->size + pos) + return SZ_ERROR_MEM; + RINOK(XzEncIndex_ReAlloc(p, newSize, alloc)); + } + memcpy(p->blocks + p->size, buf, pos); + p->size += pos; + p->numBlocks++; + return SZ_OK; +} + + +static SRes XzEncIndex_WriteFooter(const CXzEncIndex *p, CXzStreamFlags flags, ISeqOutStream *s) { Byte buf[32]; UInt64 globalPos; - { - UInt32 crc = CRC_INIT_VAL; - unsigned pos = 1 + Xz_WriteVarInt(buf + 1, p->numBlocks); - size_t i; - - globalPos = pos; - buf[0] = 0; - RINOK(WriteBytesAndCrc(s, buf, pos, &crc)); - - for (i = 0; i < p->numBlocks; i++) - { - const CXzBlockSizes *block = &p->blocks[i]; - pos = Xz_WriteVarInt(buf, block->totalSize); - pos += Xz_WriteVarInt(buf + pos, block->unpackSize); - globalPos += pos; - RINOK(WriteBytesAndCrc(s, buf, pos, &crc)); - } - - pos = ((unsigned)globalPos & 3); - - if (pos != 0) - { - buf[0] = buf[1] = buf[2] = 0; - RINOK(WriteBytesAndCrc(s, buf, 4 - pos, &crc)); - globalPos += 4 - pos; - } - { - SetUi32(buf, CRC_GET_DIGEST(crc)); - RINOK(WriteBytes(s, buf, 4)); - globalPos += 4; - } - } - - { - UInt32 indexSize = (UInt32)((globalPos >> 2) - 1); - SetUi32(buf + 4, indexSize); - buf[8] = (Byte)(p->flags >> 8); - buf[9] = (Byte)(p->flags & 0xFF); - SetUi32(buf, CrcCalc(buf + 4, 6)); - memcpy(buf + 10, XZ_FOOTER_SIG, XZ_FOOTER_SIG_SIZE); - return WriteBytes(s, buf, 12); - } + UInt32 crc = CRC_INIT_VAL; + unsigned pos = 1 + Xz_WriteVarInt(buf + 1, p->numBlocks); + + globalPos = pos; + buf[0] = 0; + RINOK(WriteBytesUpdateCrc(s, buf, pos, &crc)); + RINOK(WriteBytesUpdateCrc(s, p->blocks, p->size, &crc)); + globalPos += p->size; + + pos = XZ_GET_PAD_SIZE(globalPos); + buf[1] = 0; + buf[2] = 0; + buf[3] = 0; + globalPos += pos; + + crc = CrcUpdate(crc, buf + 4 - pos, pos); + SetUi32(buf + 4, CRC_GET_DIGEST(crc)); + + SetUi32(buf + 8 + 4, (UInt32)(globalPos >> 2)); + buf[8 + 8] = (Byte)(flags >> 8); + buf[8 + 9] = (Byte)(flags & 0xFF); + SetUi32(buf + 8, CrcCalc(buf + 8 + 4, 6)); + buf[8 + 10] = XZ_FOOTER_SIG_0; + buf[8 + 11] = XZ_FOOTER_SIG_1; + + return WriteBytes(s, buf + 4 - pos, pos + 4 + 12); } -static SRes Xz_AddIndexRecord(CXzStream *p, UInt64 unpackSize, UInt64 totalSize, ISzAlloc *alloc) -{ - if (!p->blocks || p->numBlocksAllocated == p->numBlocks) - { - size_t num = p->numBlocks * 2 + 1; - size_t newSize = sizeof(CXzBlockSizes) * num; - CXzBlockSizes *blocks; - if (newSize / sizeof(CXzBlockSizes) != num) - return SZ_ERROR_MEM; - blocks = (CXzBlockSizes *)alloc->Alloc(alloc, newSize); - if (!blocks) - return SZ_ERROR_MEM; - if (p->numBlocks != 0) - { - memcpy(blocks, p->blocks, p->numBlocks * sizeof(CXzBlockSizes)); - alloc->Free(alloc, p->blocks); - } - p->blocks = blocks; - p->numBlocksAllocated = num; - } - { - CXzBlockSizes *block = &p->blocks[p->numBlocks++]; - block->unpackSize = unpackSize; - block->totalSize = totalSize; - } - return SZ_OK; -} - /* ---------- CSeqCheckInStream ---------- */ typedef struct { - ISeqInStream p; + ISeqInStream vt; ISeqInStream *realStream; + const Byte *data; + UInt64 limit; UInt64 processed; + int realStreamFinished; CXzCheck check; } CSeqCheckInStream; -static void SeqCheckInStream_Init(CSeqCheckInStream *p, unsigned mode) +static void SeqCheckInStream_Init(CSeqCheckInStream *p, unsigned checkMode) { + p->limit = (UInt64)(Int64)-1; p->processed = 0; - XzCheck_Init(&p->check, mode); + p->realStreamFinished = 0; + XzCheck_Init(&p->check, checkMode); } static void SeqCheckInStream_GetDigest(CSeqCheckInStream *p, Byte *digest) @@ -177,12 +251,31 @@ static void SeqCheckInStream_GetDigest(CSeqCheckInStream *p, Byte *digest) XzCheck_Final(&p->check, digest); } -static SRes SeqCheckInStream_Read(void *pp, void *data, size_t *size) +static SRes SeqCheckInStream_Read(const ISeqInStream *pp, void *data, size_t *size) { - CSeqCheckInStream *p = (CSeqCheckInStream *)pp; - SRes res = p->realStream->Read(p->realStream, data, size); - XzCheck_Update(&p->check, data, *size); - p->processed += *size; + CSeqCheckInStream *p = CONTAINER_FROM_VTBL(pp, CSeqCheckInStream, vt); + size_t size2 = *size; + SRes res = SZ_OK; + + if (p->limit != (UInt64)(Int64)-1) + { + UInt64 rem = p->limit - p->processed; + if (size2 > rem) + size2 = (size_t)rem; + } + if (size2 != 0) + { + if (p->realStream) + { + res = ISeqInStream_Read(p->realStream, data, &size2); + p->realStreamFinished = (size2 == 0) ? 1 : 0; + } + else + memcpy(data, p->data + (size_t)p->processed, size2); + XzCheck_Update(&p->check, data, size2); + p->processed += size2; + } + *size = size2; return res; } @@ -191,15 +284,24 @@ static SRes SeqCheckInStream_Read(void *pp, void *data, size_t *size) typedef struct { - ISeqOutStream p; + ISeqOutStream vt; ISeqOutStream *realStream; + Byte *outBuf; + size_t outBufLimit; UInt64 processed; } CSeqSizeOutStream; -static size_t MyWrite(void *pp, const void *data, size_t size) +static size_t SeqSizeOutStream_Write(const ISeqOutStream *pp, const void *data, size_t size) { - CSeqSizeOutStream *p = (CSeqSizeOutStream *)pp; - size = p->realStream->Write(p->realStream, data, size); + CSeqSizeOutStream *p = CONTAINER_FROM_VTBL(pp, CSeqSizeOutStream, vt); + if (p->realStream) + size = ISeqOutStream_Write(p->realStream, data, size); + else + { + if (size > p->outBufLimit - (size_t)p->processed) + return 0; + memcpy(p->outBuf + (size_t)p->processed, data, size); + } p->processed += size; return size; } @@ -220,9 +322,29 @@ typedef struct int srcWasFinished; } CSeqInFilter; -static SRes SeqInFilter_Read(void *pp, void *data, size_t *size) + +SRes BraState_SetFromMethod(IStateCoder *p, UInt64 id, int encodeMode, ISzAllocPtr alloc); + +static SRes SeqInFilter_Init(CSeqInFilter *p, const CXzFilter *props, ISzAllocPtr alloc) { - CSeqInFilter *p = (CSeqInFilter *)pp; + if (!p->buf) + { + p->buf = (Byte *)ISzAlloc_Alloc(alloc, FILTER_BUF_SIZE); + if (!p->buf) + return SZ_ERROR_MEM; + } + p->curPos = p->endPos = 0; + p->srcWasFinished = 0; + RINOK(BraState_SetFromMethod(&p->StateCoder, props->id, 1, alloc)); + RINOK(p->StateCoder.SetProps(p->StateCoder.p, props->props, props->propsSize, alloc)); + p->StateCoder.Init(p->StateCoder.p); + return SZ_OK; +} + + +static SRes SeqInFilter_Read(const ISeqInStream *pp, void *data, size_t *size) +{ + CSeqInFilter *p = CONTAINER_FROM_VTBL(pp, CSeqInFilter, p); size_t sizeOriginal = *size; if (sizeOriginal == 0) return SZ_OK; @@ -234,19 +356,22 @@ static SRes SeqInFilter_Read(void *pp, void *data, size_t *size) { p->curPos = 0; p->endPos = FILTER_BUF_SIZE; - RINOK(p->realStream->Read(p->realStream, p->buf, &p->endPos)); + RINOK(ISeqInStream_Read(p->realStream, p->buf, &p->endPos)); if (p->endPos == 0) p->srcWasFinished = 1; } { SizeT srcLen = p->endPos - p->curPos; - int wasFinished; + ECoderStatus status; SRes res; *size = sizeOriginal; - res = p->StateCoder.Code(p->StateCoder.p, data, size, p->buf + p->curPos, &srcLen, - p->srcWasFinished, CODER_FINISH_ANY, &wasFinished); + res = p->StateCoder.Code2(p->StateCoder.p, + (Byte *)data, size, + p->buf + p->curPos, &srcLen, + p->srcWasFinished, CODER_FINISH_ANY, + &status); p->curPos += srcLen; - if (*size != 0 || srcLen == 0 || res != 0) + if (*size != 0 || srcLen == 0 || res != SZ_OK) return res; } } @@ -255,36 +380,24 @@ static SRes SeqInFilter_Read(void *pp, void *data, size_t *size) static void SeqInFilter_Construct(CSeqInFilter *p) { p->buf = NULL; + p->StateCoder.p = NULL; p->p.Read = SeqInFilter_Read; } -static void SeqInFilter_Free(CSeqInFilter *p) +static void SeqInFilter_Free(CSeqInFilter *p, ISzAllocPtr alloc) { + if (p->StateCoder.p) + { + p->StateCoder.Free(p->StateCoder.p, alloc); + p->StateCoder.p = NULL; + } if (p->buf) { - g_Alloc.Free(&g_Alloc, p->buf); + ISzAlloc_Free(alloc, p->buf); p->buf = NULL; } } -SRes BraState_SetFromMethod(IStateCoder *p, UInt64 id, int encodeMode, ISzAlloc *alloc); - -static SRes SeqInFilter_Init(CSeqInFilter *p, const CXzFilter *props) -{ - if (!p->buf) - { - p->buf = g_Alloc.Alloc(&g_Alloc, FILTER_BUF_SIZE); - if (!p->buf) - return SZ_ERROR_MEM; - } - p->curPos = p->endPos = 0; - p->srcWasFinished = 0; - RINOK(BraState_SetFromMethod(&p->StateCoder, props->id, 1, &g_Alloc)); - RINOK(p->StateCoder.SetProps(p->StateCoder.p, props->props, props->propsSize, &g_Alloc)); - p->StateCoder.Init(p->StateCoder.p); - return SZ_OK; -} - /* ---------- CSbEncInStream ---------- */ @@ -292,17 +405,17 @@ static SRes SeqInFilter_Init(CSeqInFilter *p, const CXzFilter *props) typedef struct { - ISeqInStream p; + ISeqInStream vt; ISeqInStream *inStream; CSbEnc enc; } CSbEncInStream; -static SRes SbEncInStream_Read(void *pp, void *data, size_t *size) +static SRes SbEncInStream_Read(const ISeqInStream *pp, void *data, size_t *size) { - CSbEncInStream *p = (CSbEncInStream *)pp; + CSbEncInStream *p = CONTAINER_FROM_VTBL(pp, CSbEncInStream, vt); size_t sizeOriginal = *size; if (sizeOriginal == 0) - return S_OK; + return SZ_OK; for (;;) { @@ -322,14 +435,14 @@ static SRes SbEncInStream_Read(void *pp, void *data, size_t *size) *size = sizeOriginal; RINOK(SbEnc_Read(&p->enc, data, size)); if (*size != 0 || !p->enc.needRead) - return S_OK; + return SZ_OK; } } -void SbEncInStream_Construct(CSbEncInStream *p, ISzAlloc *alloc) +void SbEncInStream_Construct(CSbEncInStream *p, ISzAllocPtr alloc) { SbEnc_Construct(&p->enc, alloc); - p->p.Read = SbEncInStream_Read; + p->vt.Read = SbEncInStream_Read; } SRes SbEncInStream_Init(CSbEncInStream *p) @@ -345,57 +458,9 @@ void SbEncInStream_Free(CSbEncInStream *p) #endif -typedef struct -{ - CLzma2EncHandle lzma2; - #ifdef USE_SUBBLOCK - CSbEncInStream sb; - #endif - CSeqInFilter filter; - ISzAlloc *alloc; - ISzAlloc *bigAlloc; -} CLzma2WithFilters; +/* ---------- CXzProps ---------- */ -static void Lzma2WithFilters_Construct(CLzma2WithFilters *p, ISzAlloc *alloc, ISzAlloc *bigAlloc) -{ - p->alloc = alloc; - p->bigAlloc = bigAlloc; - p->lzma2 = NULL; - #ifdef USE_SUBBLOCK - SbEncInStream_Construct(&p->sb, alloc); - #endif - SeqInFilter_Construct(&p->filter); -} - -static SRes Lzma2WithFilters_Create(CLzma2WithFilters *p) -{ - p->lzma2 = Lzma2Enc_Create(p->alloc, p->bigAlloc); - if (!p->lzma2) - return SZ_ERROR_MEM; - return SZ_OK; -} - -static void Lzma2WithFilters_Free(CLzma2WithFilters *p) -{ - SeqInFilter_Free(&p->filter); - #ifdef USE_SUBBLOCK - SbEncInStream_Free(&p->sb); - #endif - if (p->lzma2) - { - Lzma2Enc_Destroy(p->lzma2); - p->lzma2 = NULL; - } -} - - -void XzProps_Init(CXzProps *p) -{ - p->lzma2Props = NULL; - p->filterProps = NULL; - p->checkId = XZ_CHECK_CRC32; -} void XzFilterProps_Init(CXzFilterProps *p) { @@ -405,122 +470,848 @@ void XzFilterProps_Init(CXzFilterProps *p) p->ipDefined = False; } - -static SRes Xz_Compress(CXzStream *xz, CLzma2WithFilters *lzmaf, - ISeqOutStream *outStream, ISeqInStream *inStream, - const CXzProps *props, ICompressProgress *progress) +void XzProps_Init(CXzProps *p) { - xz->flags = (Byte)props->checkId; + p->checkId = XZ_CHECK_CRC32; + p->blockSize = XZ_PROPS__BLOCK_SIZE__AUTO; + p->numBlockThreads_Reduced = -1; + p->numBlockThreads_Max = -1; + p->numTotalThreads = -1; + p->reduceSize = (UInt64)(Int64)-1; + p->forceWriteSizesInHeader = 0; + // p->forceWriteSizesInHeader = 1; - RINOK(Lzma2Enc_SetProps(lzmaf->lzma2, props->lzma2Props)); - RINOK(Xz_WriteHeader(xz->flags, outStream)); - - { - CSeqCheckInStream checkInStream; - CSeqSizeOutStream seqSizeOutStream; - CXzBlock block; - unsigned filterIndex = 0; - CXzFilter *filter = NULL; - const CXzFilterProps *fp = props->filterProps; - - XzBlock_ClearFlags(&block); - XzBlock_SetNumFilters(&block, 1 + (fp ? 1 : 0)); - - if (fp) - { - filter = &block.filters[filterIndex++]; - filter->id = fp->id; - filter->propsSize = 0; - - if (fp->id == XZ_ID_Delta) - { - filter->props[0] = (Byte)(fp->delta - 1); - filter->propsSize = 1; - } - else if (fp->ipDefined) - { - SetUi32(filter->props, fp->ip); - filter->propsSize = 4; - } - } - - { - CXzFilter *f = &block.filters[filterIndex++]; - f->id = XZ_ID_LZMA2; - f->propsSize = 1; - f->props[0] = Lzma2Enc_WriteProperties(lzmaf->lzma2); - } - - seqSizeOutStream.p.Write = MyWrite; - seqSizeOutStream.realStream = outStream; - seqSizeOutStream.processed = 0; - - RINOK(XzBlock_WriteHeader(&block, &seqSizeOutStream.p)); - - checkInStream.p.Read = SeqCheckInStream_Read; - checkInStream.realStream = inStream; - SeqCheckInStream_Init(&checkInStream, XzFlags_GetCheckType(xz->flags)); - - if (fp) - { - #ifdef USE_SUBBLOCK - if (fp->id == XZ_ID_Subblock) - { - lzmaf->sb.inStream = &checkInStream.p; - RINOK(SbEncInStream_Init(&lzmaf->sb)); - } - else - #endif - { - lzmaf->filter.realStream = &checkInStream.p; - RINOK(SeqInFilter_Init(&lzmaf->filter, filter)); - } - } - - { - UInt64 packPos = seqSizeOutStream.processed; - - SRes res = Lzma2Enc_Encode(lzmaf->lzma2, &seqSizeOutStream.p, - fp ? - #ifdef USE_SUBBLOCK - (fp->id == XZ_ID_Subblock) ? &lzmaf->sb.p: - #endif - &lzmaf->filter.p: - &checkInStream.p, - progress); - - RINOK(res); - block.unpackSize = checkInStream.processed; - block.packSize = seqSizeOutStream.processed - packPos; - } - - { - unsigned padSize = 0; - Byte buf[128]; - while ((((unsigned)block.packSize + padSize) & 3) != 0) - buf[padSize++] = 0; - SeqCheckInStream_GetDigest(&checkInStream, buf + padSize); - RINOK(WriteBytes(&seqSizeOutStream.p, buf, padSize + XzFlags_GetCheckSize(xz->flags))); - RINOK(Xz_AddIndexRecord(xz, block.unpackSize, seqSizeOutStream.processed - padSize, &g_Alloc)); - } - } - return Xz_WriteFooter(xz, outStream); + XzFilterProps_Init(&p->filterProps); + Lzma2EncProps_Init(&p->lzma2Props); } +static void XzEncProps_Normalize_Fixed(CXzProps *p) +{ + UInt64 fileSize; + int t1, t1n, t2, t2r, t3; + { + CLzma2EncProps tp = p->lzma2Props; + if (tp.numTotalThreads <= 0) + tp.numTotalThreads = p->numTotalThreads; + Lzma2EncProps_Normalize(&tp); + t1n = tp.numTotalThreads; + } + + t1 = p->lzma2Props.numTotalThreads; + t2 = p->numBlockThreads_Max; + t3 = p->numTotalThreads; + + if (t2 > MTCODER__THREADS_MAX) + t2 = MTCODER__THREADS_MAX; + + if (t3 <= 0) + { + if (t2 <= 0) + t2 = 1; + t3 = t1n * t2; + } + else if (t2 <= 0) + { + t2 = t3 / t1n; + if (t2 == 0) + { + t1 = 1; + t2 = t3; + } + if (t2 > MTCODER__THREADS_MAX) + t2 = MTCODER__THREADS_MAX; + } + else if (t1 <= 0) + { + t1 = t3 / t2; + if (t1 == 0) + t1 = 1; + } + else + t3 = t1n * t2; + + p->lzma2Props.numTotalThreads = t1; + + t2r = t2; + + fileSize = p->reduceSize; + + if ((p->blockSize < fileSize || fileSize == (UInt64)(Int64)-1)) + p->lzma2Props.lzmaProps.reduceSize = p->blockSize; + + Lzma2EncProps_Normalize(&p->lzma2Props); + + t1 = p->lzma2Props.numTotalThreads; + + { + if (t2 > 1 && fileSize != (UInt64)(Int64)-1) + { + UInt64 numBlocks = fileSize / p->blockSize; + if (numBlocks * p->blockSize != fileSize) + numBlocks++; + if (numBlocks < (unsigned)t2) + { + t2r = (unsigned)numBlocks; + if (t2r == 0) + t2r = 1; + t3 = t1 * t2r; + } + } + } + + p->numBlockThreads_Max = t2; + p->numBlockThreads_Reduced = t2r; + p->numTotalThreads = t3; +} + + +static void XzProps_Normalize(CXzProps *p) +{ + /* we normalize xzProps properties, but we normalize only some of CXzProps::lzma2Props properties. + Lzma2Enc_SetProps() will normalize lzma2Props later. */ + + if (p->blockSize == XZ_PROPS__BLOCK_SIZE__SOLID) + { + p->lzma2Props.lzmaProps.reduceSize = p->reduceSize; + p->numBlockThreads_Reduced = 1; + p->numBlockThreads_Max = 1; + if (p->lzma2Props.numTotalThreads <= 0) + p->lzma2Props.numTotalThreads = p->numTotalThreads; + return; + } + else + { + CLzma2EncProps *lzma2 = &p->lzma2Props; + if (p->blockSize == LZMA2_ENC_PROPS__BLOCK_SIZE__AUTO) + { + // xz-auto + p->lzma2Props.lzmaProps.reduceSize = p->reduceSize; + + if (lzma2->blockSize == LZMA2_ENC_PROPS__BLOCK_SIZE__SOLID) + { + // if (xz-auto && lzma2-solid) - we use solid for both + p->blockSize = XZ_PROPS__BLOCK_SIZE__SOLID; + p->numBlockThreads_Reduced = 1; + p->numBlockThreads_Max = 1; + if (p->lzma2Props.numTotalThreads <= 0) + p->lzma2Props.numTotalThreads = p->numTotalThreads; + } + else + { + // if (xz-auto && (lzma2-auto || lzma2-fixed_) + // we calculate block size for lzma2 and use that block size for xz, lzma2 uses single-chunk per block + CLzma2EncProps tp = p->lzma2Props; + if (tp.numTotalThreads <= 0) + tp.numTotalThreads = p->numTotalThreads; + + Lzma2EncProps_Normalize(&tp); + + p->blockSize = tp.blockSize; // fixed or solid + p->numBlockThreads_Reduced = tp.numBlockThreads_Reduced; + p->numBlockThreads_Max = tp.numBlockThreads_Max; + if (lzma2->blockSize == LZMA2_ENC_PROPS__BLOCK_SIZE__AUTO) + lzma2->blockSize = tp.blockSize; // fixed or solid, LZMA2_ENC_PROPS__BLOCK_SIZE__SOLID + if (lzma2->lzmaProps.reduceSize > tp.blockSize && tp.blockSize != LZMA2_ENC_PROPS__BLOCK_SIZE__SOLID) + lzma2->lzmaProps.reduceSize = tp.blockSize; + lzma2->numBlockThreads_Reduced = 1; + lzma2->numBlockThreads_Max = 1; + return; + } + } + else + { + // xz-fixed + // we can use xz::reduceSize or xz::blockSize as base for lzmaProps::reduceSize + + p->lzma2Props.lzmaProps.reduceSize = p->reduceSize; + { + UInt64 r = p->reduceSize; + if (r > p->blockSize || r == (UInt64)(Int64)-1) + r = p->blockSize; + lzma2->lzmaProps.reduceSize = r; + } + if (lzma2->blockSize == LZMA2_ENC_PROPS__BLOCK_SIZE__AUTO) + lzma2->blockSize = LZMA2_ENC_PROPS__BLOCK_SIZE__SOLID; + else if (lzma2->blockSize > p->blockSize && lzma2->blockSize != LZMA2_ENC_PROPS__BLOCK_SIZE__SOLID) + lzma2->blockSize = p->blockSize; + + XzEncProps_Normalize_Fixed(p); + } + } +} + + +/* ---------- CLzma2WithFilters ---------- */ + +typedef struct +{ + CLzma2EncHandle lzma2; + CSeqInFilter filter; + + #ifdef USE_SUBBLOCK + CSbEncInStream sb; + #endif +} CLzma2WithFilters; + + +static void Lzma2WithFilters_Construct(CLzma2WithFilters *p) +{ + p->lzma2 = NULL; + SeqInFilter_Construct(&p->filter); + + #ifdef USE_SUBBLOCK + SbEncInStream_Construct(&p->sb, alloc); + #endif +} + + +static SRes Lzma2WithFilters_Create(CLzma2WithFilters *p, ISzAllocPtr alloc, ISzAllocPtr bigAlloc) +{ + if (!p->lzma2) + { + p->lzma2 = Lzma2Enc_Create(alloc, bigAlloc); + if (!p->lzma2) + return SZ_ERROR_MEM; + } + return SZ_OK; +} + + +static void Lzma2WithFilters_Free(CLzma2WithFilters *p, ISzAllocPtr alloc) +{ + #ifdef USE_SUBBLOCK + SbEncInStream_Free(&p->sb); + #endif + + SeqInFilter_Free(&p->filter, alloc); + if (p->lzma2) + { + Lzma2Enc_Destroy(p->lzma2); + p->lzma2 = NULL; + } +} + + +typedef struct +{ + UInt64 unpackSize; + UInt64 totalSize; + size_t headerSize; +} CXzEncBlockInfo; + + +static SRes Xz_CompressBlock( + CLzma2WithFilters *lzmaf, + + ISeqOutStream *outStream, + Byte *outBufHeader, + Byte *outBufData, size_t outBufDataLimit, + + ISeqInStream *inStream, + // UInt64 expectedSize, + const Byte *inBuf, // used if (!inStream) + size_t inBufSize, // used if (!inStream), it's block size, props->blockSize is ignored + + const CXzProps *props, + ICompressProgress *progress, + int *inStreamFinished, /* only for inStream version */ + CXzEncBlockInfo *blockSizes, + ISzAllocPtr alloc, + ISzAllocPtr allocBig) +{ + CSeqCheckInStream checkInStream; + CSeqSizeOutStream seqSizeOutStream; + CXzBlock block; + unsigned filterIndex = 0; + CXzFilter *filter = NULL; + const CXzFilterProps *fp = &props->filterProps; + if (fp->id == 0) + fp = NULL; + + *inStreamFinished = False; + + RINOK(Lzma2WithFilters_Create(lzmaf, alloc, allocBig)); + + RINOK(Lzma2Enc_SetProps(lzmaf->lzma2, &props->lzma2Props)); + + XzBlock_ClearFlags(&block); + XzBlock_SetNumFilters(&block, 1 + (fp ? 1 : 0)); + + if (fp) + { + filter = &block.filters[filterIndex++]; + filter->id = fp->id; + filter->propsSize = 0; + + if (fp->id == XZ_ID_Delta) + { + filter->props[0] = (Byte)(fp->delta - 1); + filter->propsSize = 1; + } + else if (fp->ipDefined) + { + SetUi32(filter->props, fp->ip); + filter->propsSize = 4; + } + } + + { + CXzFilter *f = &block.filters[filterIndex++]; + f->id = XZ_ID_LZMA2; + f->propsSize = 1; + f->props[0] = Lzma2Enc_WriteProperties(lzmaf->lzma2); + } + + seqSizeOutStream.vt.Write = SeqSizeOutStream_Write; + seqSizeOutStream.realStream = outStream; + seqSizeOutStream.outBuf = outBufData; + seqSizeOutStream.outBufLimit = outBufDataLimit; + seqSizeOutStream.processed = 0; + + /* + if (expectedSize != (UInt64)(Int64)-1) + { + block.unpackSize = expectedSize; + if (props->blockSize != (UInt64)(Int64)-1) + if (expectedSize > props->blockSize) + block.unpackSize = props->blockSize; + XzBlock_SetHasUnpackSize(&block); + } + */ + + if (outStream) + { + RINOK(XzBlock_WriteHeader(&block, &seqSizeOutStream.vt)); + } + + checkInStream.vt.Read = SeqCheckInStream_Read; + SeqCheckInStream_Init(&checkInStream, props->checkId); + + checkInStream.realStream = inStream; + checkInStream.data = inBuf; + checkInStream.limit = props->blockSize; + if (!inStream) + checkInStream.limit = inBufSize; + + if (fp) + { + #ifdef USE_SUBBLOCK + if (fp->id == XZ_ID_Subblock) + { + lzmaf->sb.inStream = &checkInStream.vt; + RINOK(SbEncInStream_Init(&lzmaf->sb)); + } + else + #endif + { + lzmaf->filter.realStream = &checkInStream.vt; + RINOK(SeqInFilter_Init(&lzmaf->filter, filter, alloc)); + } + } + + { + SRes res; + Byte *outBuf = NULL; + size_t outSize = 0; + BoolInt useStream = (fp || inStream); + // useStream = True; + + if (!useStream) + { + XzCheck_Update(&checkInStream.check, inBuf, inBufSize); + checkInStream.processed = inBufSize; + } + + if (!outStream) + { + outBuf = seqSizeOutStream.outBuf; // + (size_t)seqSizeOutStream.processed; + outSize = seqSizeOutStream.outBufLimit; // - (size_t)seqSizeOutStream.processed; + } + + res = Lzma2Enc_Encode2(lzmaf->lzma2, + outBuf ? NULL : &seqSizeOutStream.vt, + outBuf, + outBuf ? &outSize : NULL, + + useStream ? + (fp ? + ( + #ifdef USE_SUBBLOCK + (fp->id == XZ_ID_Subblock) ? &lzmaf->sb.vt: + #endif + &lzmaf->filter.p) : + &checkInStream.vt) : NULL, + + useStream ? NULL : inBuf, + useStream ? 0 : inBufSize, + + progress); + + if (outBuf) + seqSizeOutStream.processed += outSize; + + RINOK(res); + blockSizes->unpackSize = checkInStream.processed; + } + { + Byte buf[4 + 64]; + unsigned padSize = XZ_GET_PAD_SIZE(seqSizeOutStream.processed); + UInt64 packSize = seqSizeOutStream.processed; + + buf[0] = 0; + buf[1] = 0; + buf[2] = 0; + buf[3] = 0; + + SeqCheckInStream_GetDigest(&checkInStream, buf + 4); + RINOK(WriteBytes(&seqSizeOutStream.vt, buf + (4 - padSize), padSize + XzFlags_GetCheckSize((CXzStreamFlags)props->checkId))); + + blockSizes->totalSize = seqSizeOutStream.processed - padSize; + + if (!outStream) + { + seqSizeOutStream.outBuf = outBufHeader; + seqSizeOutStream.outBufLimit = XZ_BLOCK_HEADER_SIZE_MAX; + seqSizeOutStream.processed = 0; + + block.unpackSize = blockSizes->unpackSize; + XzBlock_SetHasUnpackSize(&block); + + block.packSize = packSize; + XzBlock_SetHasPackSize(&block); + + RINOK(XzBlock_WriteHeader(&block, &seqSizeOutStream.vt)); + + blockSizes->headerSize = (size_t)seqSizeOutStream.processed; + blockSizes->totalSize += seqSizeOutStream.processed; + } + } + + if (inStream) + *inStreamFinished = checkInStream.realStreamFinished; + else + { + *inStreamFinished = False; + if (checkInStream.processed != inBufSize) + return SZ_ERROR_FAIL; + } + + return SZ_OK; +} + + + +typedef struct +{ + ICompressProgress vt; + ICompressProgress *progress; + UInt64 inOffset; + UInt64 outOffset; +} CCompressProgress_XzEncOffset; + + +static SRes CompressProgress_XzEncOffset_Progress(const ICompressProgress *pp, UInt64 inSize, UInt64 outSize) +{ + const CCompressProgress_XzEncOffset *p = CONTAINER_FROM_VTBL(pp, CCompressProgress_XzEncOffset, vt); + inSize += p->inOffset; + outSize += p->outOffset; + return ICompressProgress_Progress(p->progress, inSize, outSize); +} + + + + +typedef struct +{ + ISzAllocPtr alloc; + ISzAllocPtr allocBig; + + CXzProps xzProps; + UInt64 expectedDataSize; + + CXzEncIndex xzIndex; + + CLzma2WithFilters lzmaf_Items[MTCODER__THREADS_MAX]; + + size_t outBufSize; /* size of allocated outBufs[i] */ + Byte *outBufs[MTCODER__BLOCKS_MAX]; + + #ifndef _7ZIP_ST + unsigned checkType; + ISeqOutStream *outStream; + BoolInt mtCoder_WasConstructed; + CMtCoder mtCoder; + CXzEncBlockInfo EncBlocks[MTCODER__BLOCKS_MAX]; + #endif + +} CXzEnc; + + +static void XzEnc_Construct(CXzEnc *p) +{ + unsigned i; + + XzEncIndex_Construct(&p->xzIndex); + + for (i = 0; i < MTCODER__THREADS_MAX; i++) + Lzma2WithFilters_Construct(&p->lzmaf_Items[i]); + + #ifndef _7ZIP_ST + p->mtCoder_WasConstructed = False; + { + for (i = 0; i < MTCODER__BLOCKS_MAX; i++) + p->outBufs[i] = NULL; + p->outBufSize = 0; + } + #endif +} + + +static void XzEnc_FreeOutBufs(CXzEnc *p) +{ + unsigned i; + for (i = 0; i < MTCODER__BLOCKS_MAX; i++) + if (p->outBufs[i]) + { + ISzAlloc_Free(p->alloc, p->outBufs[i]); + p->outBufs[i] = NULL; + } + p->outBufSize = 0; +} + + +static void XzEnc_Free(CXzEnc *p, ISzAllocPtr alloc) +{ + unsigned i; + + XzEncIndex_Free(&p->xzIndex, alloc); + + for (i = 0; i < MTCODER__THREADS_MAX; i++) + Lzma2WithFilters_Free(&p->lzmaf_Items[i], alloc); + + #ifndef _7ZIP_ST + if (p->mtCoder_WasConstructed) + { + MtCoder_Destruct(&p->mtCoder); + p->mtCoder_WasConstructed = False; + } + XzEnc_FreeOutBufs(p); + #endif +} + + +CXzEncHandle XzEnc_Create(ISzAllocPtr alloc, ISzAllocPtr allocBig) +{ + CXzEnc *p = (CXzEnc *)ISzAlloc_Alloc(alloc, sizeof(CXzEnc)); + if (!p) + return NULL; + XzEnc_Construct(p); + XzProps_Init(&p->xzProps); + XzProps_Normalize(&p->xzProps); + p->expectedDataSize = (UInt64)(Int64)-1; + p->alloc = alloc; + p->allocBig = allocBig; + return p; +} + + +void XzEnc_Destroy(CXzEncHandle pp) +{ + CXzEnc *p = (CXzEnc *)pp; + XzEnc_Free(p, p->alloc); + ISzAlloc_Free(p->alloc, p); +} + + +SRes XzEnc_SetProps(CXzEncHandle pp, const CXzProps *props) +{ + CXzEnc *p = (CXzEnc *)pp; + p->xzProps = *props; + XzProps_Normalize(&p->xzProps); + return SZ_OK; +} + + +void XzEnc_SetDataSize(CXzEncHandle pp, UInt64 expectedDataSiize) +{ + CXzEnc *p = (CXzEnc *)pp; + p->expectedDataSize = expectedDataSiize; +} + + + + +#ifndef _7ZIP_ST + +static SRes XzEnc_MtCallback_Code(void *pp, unsigned coderIndex, unsigned outBufIndex, + const Byte *src, size_t srcSize, int finished) +{ + CXzEnc *me = (CXzEnc *)pp; + SRes res; + CMtProgressThunk progressThunk; + + Byte *dest = me->outBufs[outBufIndex]; + + UNUSED_VAR(finished) + + { + CXzEncBlockInfo *bInfo = &me->EncBlocks[outBufIndex]; + bInfo->totalSize = 0; + bInfo->unpackSize = 0; + bInfo->headerSize = 0; + } + + if (!dest) + { + dest = (Byte *)ISzAlloc_Alloc(me->alloc, me->outBufSize); + if (!dest) + return SZ_ERROR_MEM; + me->outBufs[outBufIndex] = dest; + } + + MtProgressThunk_CreateVTable(&progressThunk); + progressThunk.mtProgress = &me->mtCoder.mtProgress; + MtProgressThunk_Init(&progressThunk); + + { + CXzEncBlockInfo blockSizes; + int inStreamFinished; + + res = Xz_CompressBlock( + &me->lzmaf_Items[coderIndex], + + NULL, + dest, + dest + XZ_BLOCK_HEADER_SIZE_MAX, me->outBufSize - XZ_BLOCK_HEADER_SIZE_MAX, + + NULL, + // srcSize, // expectedSize + src, srcSize, + + &me->xzProps, + &progressThunk.vt, + &inStreamFinished, + &blockSizes, + me->alloc, + me->allocBig); + + if (res == SZ_OK) + me->EncBlocks[outBufIndex] = blockSizes; + + return res; + } +} + + +static SRes XzEnc_MtCallback_Write(void *pp, unsigned outBufIndex) +{ + CXzEnc *me = (CXzEnc *)pp; + + const CXzEncBlockInfo *bInfo = &me->EncBlocks[outBufIndex]; + const Byte *data = me->outBufs[outBufIndex]; + + RINOK(WriteBytes(me->outStream, data, bInfo->headerSize)); + + { + UInt64 totalPackFull = bInfo->totalSize + XZ_GET_PAD_SIZE(bInfo->totalSize); + RINOK(WriteBytes(me->outStream, data + XZ_BLOCK_HEADER_SIZE_MAX, (size_t)totalPackFull - bInfo->headerSize)); + } + + return XzEncIndex_AddIndexRecord(&me->xzIndex, bInfo->unpackSize, bInfo->totalSize, me->alloc); +} + +#endif + + + +SRes XzEnc_Encode(CXzEncHandle pp, ISeqOutStream *outStream, ISeqInStream *inStream, ICompressProgress *progress) +{ + CXzEnc *p = (CXzEnc *)pp; + + const CXzProps *props = &p->xzProps; + + XzEncIndex_Init(&p->xzIndex); + { + UInt64 numBlocks = 1; + UInt64 blockSize = props->blockSize; + + if (blockSize != XZ_PROPS__BLOCK_SIZE__SOLID + && props->reduceSize != (UInt64)(Int64)-1) + { + numBlocks = props->reduceSize / blockSize; + if (numBlocks * blockSize != props->reduceSize) + numBlocks++; + } + else + blockSize = (UInt64)1 << 62; + + RINOK(XzEncIndex_PreAlloc(&p->xzIndex, numBlocks, blockSize, XZ_GET_ESTIMATED_BLOCK_TOTAL_PACK_SIZE(blockSize), p->alloc)); + } + + RINOK(Xz_WriteHeader((CXzStreamFlags)props->checkId, outStream)); + + + #ifndef _7ZIP_ST + if (props->numBlockThreads_Reduced > 1) + { + IMtCoderCallback2 vt; + + if (!p->mtCoder_WasConstructed) + { + p->mtCoder_WasConstructed = True; + MtCoder_Construct(&p->mtCoder); + } + + vt.Code = XzEnc_MtCallback_Code; + vt.Write = XzEnc_MtCallback_Write; + + p->checkType = props->checkId; + p->xzProps = *props; + + p->outStream = outStream; + + p->mtCoder.allocBig = p->allocBig; + p->mtCoder.progress = progress; + p->mtCoder.inStream = inStream; + p->mtCoder.inData = NULL; + p->mtCoder.inDataSize = 0; + p->mtCoder.mtCallback = &vt; + p->mtCoder.mtCallbackObject = p; + + if ( props->blockSize == XZ_PROPS__BLOCK_SIZE__SOLID + || props->blockSize == XZ_PROPS__BLOCK_SIZE__AUTO) + return SZ_ERROR_FAIL; + + p->mtCoder.blockSize = (size_t)props->blockSize; + if (p->mtCoder.blockSize != props->blockSize) + return SZ_ERROR_PARAM; /* SZ_ERROR_MEM */ + + { + size_t destBlockSize = XZ_BLOCK_HEADER_SIZE_MAX + XZ_GET_MAX_BLOCK_PACK_SIZE(p->mtCoder.blockSize); + if (destBlockSize < p->mtCoder.blockSize) + return SZ_ERROR_PARAM; + if (p->outBufSize != destBlockSize) + XzEnc_FreeOutBufs(p); + p->outBufSize = destBlockSize; + } + + p->mtCoder.numThreadsMax = props->numBlockThreads_Max; + p->mtCoder.expectedDataSize = p->expectedDataSize; + + RINOK(MtCoder_Code(&p->mtCoder)); + } + else + #endif + { + int writeStartSizes; + CCompressProgress_XzEncOffset progress2; + Byte *bufData = NULL; + size_t bufSize = 0; + + progress2.vt.Progress = CompressProgress_XzEncOffset_Progress; + progress2.inOffset = 0; + progress2.outOffset = 0; + progress2.progress = progress; + + writeStartSizes = 0; + + if (props->blockSize != XZ_PROPS__BLOCK_SIZE__SOLID) + { + writeStartSizes = (props->forceWriteSizesInHeader > 0); + + if (writeStartSizes) + { + size_t t2; + size_t t = (size_t)props->blockSize; + if (t != props->blockSize) + return SZ_ERROR_PARAM; + t = XZ_GET_MAX_BLOCK_PACK_SIZE(t); + if (t < props->blockSize) + return SZ_ERROR_PARAM; + t2 = XZ_BLOCK_HEADER_SIZE_MAX + t; + if (!p->outBufs[0] || t2 != p->outBufSize) + { + XzEnc_FreeOutBufs(p); + p->outBufs[0] = (Byte *)ISzAlloc_Alloc(p->alloc, t2); + if (!p->outBufs[0]) + return SZ_ERROR_MEM; + p->outBufSize = t2; + } + bufData = p->outBufs[0] + XZ_BLOCK_HEADER_SIZE_MAX; + bufSize = t; + } + } + + for (;;) + { + CXzEncBlockInfo blockSizes; + int inStreamFinished; + + /* + UInt64 rem = (UInt64)(Int64)-1; + if (props->reduceSize != (UInt64)(Int64)-1 + && props->reduceSize >= progress2.inOffset) + rem = props->reduceSize - progress2.inOffset; + */ + + blockSizes.headerSize = 0; // for GCC + + RINOK(Xz_CompressBlock( + &p->lzmaf_Items[0], + + writeStartSizes ? NULL : outStream, + writeStartSizes ? p->outBufs[0] : NULL, + bufData, bufSize, + + inStream, + // rem, + NULL, 0, + + props, + progress ? &progress2.vt : NULL, + &inStreamFinished, + &blockSizes, + p->alloc, + p->allocBig)); + + { + UInt64 totalPackFull = blockSizes.totalSize + XZ_GET_PAD_SIZE(blockSizes.totalSize); + + if (writeStartSizes) + { + RINOK(WriteBytes(outStream, p->outBufs[0], blockSizes.headerSize)); + RINOK(WriteBytes(outStream, bufData, (size_t)totalPackFull - blockSizes.headerSize)); + } + + RINOK(XzEncIndex_AddIndexRecord(&p->xzIndex, blockSizes.unpackSize, blockSizes.totalSize, p->alloc)); + + progress2.inOffset += blockSizes.unpackSize; + progress2.outOffset += totalPackFull; + } + + if (inStreamFinished) + break; + } + } + + return XzEncIndex_WriteFooter(&p->xzIndex, (CXzStreamFlags)props->checkId, outStream); +} + + +#include "Alloc.h" + SRes Xz_Encode(ISeqOutStream *outStream, ISeqInStream *inStream, const CXzProps *props, ICompressProgress *progress) { SRes res; - CXzStream xz; - CLzma2WithFilters lzmaf; - Xz_Construct(&xz); - Lzma2WithFilters_Construct(&lzmaf, &g_Alloc, &g_BigAlloc); - res = Lzma2WithFilters_Create(&lzmaf); + CXzEncHandle xz = XzEnc_Create(&g_Alloc, &g_BigAlloc); + if (!xz) + return SZ_ERROR_MEM; + res = XzEnc_SetProps(xz, props); if (res == SZ_OK) - res = Xz_Compress(&xz, &lzmaf, outStream, inStream, props, progress); - Lzma2WithFilters_Free(&lzmaf); - Xz_Free(&xz, &g_Alloc); + res = XzEnc_Encode(xz, outStream, inStream, progress); + XzEnc_Destroy(xz); return res; } @@ -528,11 +1319,11 @@ SRes Xz_Encode(ISeqOutStream *outStream, ISeqInStream *inStream, SRes Xz_EncodeEmpty(ISeqOutStream *outStream) { SRes res; - CXzStream xz; - Xz_Construct(&xz); - res = Xz_WriteHeader(xz.flags, outStream); + CXzEncIndex xzIndex; + XzEncIndex_Construct(&xzIndex); + res = Xz_WriteHeader((CXzStreamFlags)0, outStream); if (res == SZ_OK) - res = Xz_WriteFooter(&xz, outStream); - Xz_Free(&xz, &g_Alloc); + res = XzEncIndex_WriteFooter(&xzIndex, (CXzStreamFlags)0, outStream); + XzEncIndex_Free(&xzIndex, NULL); // g_Alloc return res; } diff --git a/src/third-party/lzma/XzEnc.h b/src/third-party/lzma/XzEnc.h index e9cea34f7..529ac3fd8 100644 --- a/src/third-party/lzma/XzEnc.h +++ b/src/third-party/lzma/XzEnc.h @@ -1,5 +1,5 @@ /* XzEnc.h -- Xz Encode -2011-02-07 : Igor Pavlov : Public domain */ +2017-06-27 : Igor Pavlov : Public domain */ #ifndef __XZ_ENC_H #define __XZ_ENC_H @@ -10,6 +10,11 @@ EXTERN_C_BEGIN + +#define XZ_PROPS__BLOCK_SIZE__AUTO LZMA2_ENC_PROPS__BLOCK_SIZE__AUTO +#define XZ_PROPS__BLOCK_SIZE__SOLID LZMA2_ENC_PROPS__BLOCK_SIZE__SOLID + + typedef struct { UInt32 id; @@ -20,15 +25,31 @@ typedef struct void XzFilterProps_Init(CXzFilterProps *p); + typedef struct { - const CLzma2EncProps *lzma2Props; - const CXzFilterProps *filterProps; + CLzma2EncProps lzma2Props; + CXzFilterProps filterProps; unsigned checkId; + UInt64 blockSize; + int numBlockThreads_Reduced; + int numBlockThreads_Max; + int numTotalThreads; + int forceWriteSizesInHeader; + UInt64 reduceSize; } CXzProps; void XzProps_Init(CXzProps *p); + +typedef void * CXzEncHandle; + +CXzEncHandle XzEnc_Create(ISzAllocPtr alloc, ISzAllocPtr allocBig); +void XzEnc_Destroy(CXzEncHandle p); +SRes XzEnc_SetProps(CXzEncHandle p, const CXzProps *props); +void XzEnc_SetDataSize(CXzEncHandle p, UInt64 expectedDataSiize); +SRes XzEnc_Encode(CXzEncHandle p, ISeqOutStream *outStream, ISeqInStream *inStream, ICompressProgress *progress); + SRes Xz_Encode(ISeqOutStream *outStream, ISeqInStream *inStream, const CXzProps *props, ICompressProgress *progress); diff --git a/src/third-party/lzma/XzIn.c b/src/third-party/lzma/XzIn.c index aaa059730..792a61786 100644 --- a/src/third-party/lzma/XzIn.c +++ b/src/third-party/lzma/XzIn.c @@ -1,5 +1,5 @@ /* XzIn.c - Xz input -2015-11-08 : Igor Pavlov : Public domain */ +2018-07-04 : Igor Pavlov : Public domain */ #include "Precomp.h" @@ -9,6 +9,12 @@ #include "CpuArch.h" #include "Xz.h" +/* +#define XZ_FOOTER_SIG_CHECK(p) (memcmp((p), XZ_FOOTER_SIG, XZ_FOOTER_SIG_SIZE) == 0) +*/ +#define XZ_FOOTER_SIG_CHECK(p) ((p)[0] == XZ_FOOTER_SIG_0 && (p)[1] == XZ_FOOTER_SIG_1) + + SRes Xz_ReadHeader(CXzStreamFlags *p, ISeqInStream *inStream) { Byte sig[XZ_STREAM_HEADER_SIZE]; @@ -22,13 +28,13 @@ SRes Xz_ReadHeader(CXzStreamFlags *p, ISeqInStream *inStream) { unsigned s = Xz_ReadVarInt(buf + pos, size - pos, res); \ if (s == 0) return SZ_ERROR_ARCHIVE; pos += s; } -SRes XzBlock_ReadHeader(CXzBlock *p, ISeqInStream *inStream, Bool *isIndex, UInt32 *headerSizeRes) +SRes XzBlock_ReadHeader(CXzBlock *p, ISeqInStream *inStream, BoolInt *isIndex, UInt32 *headerSizeRes) { Byte header[XZ_BLOCK_HEADER_SIZE_MAX]; unsigned headerSize; *headerSizeRes = 0; RINOK(SeqInStream_ReadByte(inStream, &header[0])); - headerSize = ((unsigned)header[0] << 2) + 4; + headerSize = (unsigned)header[0]; if (headerSize == 0) { *headerSizeRes = 1; @@ -37,12 +43,13 @@ SRes XzBlock_ReadHeader(CXzBlock *p, ISeqInStream *inStream, Bool *isIndex, UInt } *isIndex = False; + headerSize = (headerSize << 2) + 4; *headerSizeRes = headerSize; RINOK(SeqInStream_Read(inStream, header + 1, headerSize - 1)); return XzBlock_Parse(p, header); } -#define ADD_SIZE_CHECH(size, val) \ +#define ADD_SIZE_CHECK(size, val) \ { UInt64 newSize = size + (val); if (newSize < size) return XZ_SIZE_OVERFLOW; size = newSize; } UInt64 Xz_GetUnpackSize(const CXzStream *p) @@ -50,7 +57,7 @@ UInt64 Xz_GetUnpackSize(const CXzStream *p) UInt64 size = 0; size_t i; for (i = 0; i < p->numBlocks; i++) - ADD_SIZE_CHECH(size, p->blocks[i].unpackSize); + ADD_SIZE_CHECK(size, p->blocks[i].unpackSize); return size; } @@ -59,7 +66,7 @@ UInt64 Xz_GetPackSize(const CXzStream *p) UInt64 size = 0; size_t i; for (i = 0; i < p->numBlocks; i++) - ADD_SIZE_CHECH(size, (p->blocks[i].totalSize + 3) & ~(UInt64)3); + ADD_SIZE_CHECK(size, (p->blocks[i].totalSize + 3) & ~(UInt64)3); return size; } @@ -70,7 +77,7 @@ SRes XzBlock_ReadFooter(CXzBlock *p, CXzStreamFlags f, ISeqInStream *inStream) } */ -static SRes Xz_ReadIndex2(CXzStream *p, const Byte *buf, size_t size, ISzAlloc *alloc) +static SRes Xz_ReadIndex2(CXzStream *p, const Byte *buf, size_t size, ISzAllocPtr alloc) { size_t numBlocks, pos = 1; UInt32 crc; @@ -96,9 +103,8 @@ static SRes Xz_ReadIndex2(CXzStream *p, const Byte *buf, size_t size, ISzAlloc * { size_t i; p->numBlocks = numBlocks; - p->numBlocksAllocated = numBlocks; - p->blocks = alloc->Alloc(alloc, sizeof(CXzBlockSizes) * numBlocks); - if (p->blocks == 0) + p->blocks = (CXzBlockSizes *)ISzAlloc_Alloc(alloc, sizeof(CXzBlockSizes) * numBlocks); + if (!p->blocks) return SZ_ERROR_MEM; for (i = 0; i < numBlocks; i++) { @@ -115,7 +121,7 @@ static SRes Xz_ReadIndex2(CXzStream *p, const Byte *buf, size_t size, ISzAlloc * return (pos == size) ? SZ_OK : SZ_ERROR_ARCHIVE; } -static SRes Xz_ReadIndex(CXzStream *p, ILookInStream *stream, UInt64 indexSize, ISzAlloc *alloc) +static SRes Xz_ReadIndex(CXzStream *p, ILookInStream *stream, UInt64 indexSize, ISzAllocPtr alloc) { SRes res; size_t size; @@ -125,13 +131,13 @@ static SRes Xz_ReadIndex(CXzStream *p, ILookInStream *stream, UInt64 indexSize, size = (size_t)indexSize; if (size != indexSize) return SZ_ERROR_UNSUPPORTED; - buf = alloc->Alloc(alloc, size); - if (buf == 0) + buf = (Byte *)ISzAlloc_Alloc(alloc, size); + if (!buf) return SZ_ERROR_MEM; res = LookInStream_Read2(stream, buf, size, SZ_ERROR_UNSUPPORTED); if (res == SZ_OK) res = Xz_ReadIndex2(p, buf, size, alloc); - alloc->Free(alloc, buf); + ISzAlloc_Free(alloc, buf); return res; } @@ -142,7 +148,7 @@ static SRes LookInStream_SeekRead_ForArc(ILookInStream *stream, UInt64 offset, v /* return LookInStream_Read2(stream, buf, size, SZ_ERROR_NO_ARCHIVE); */ } -static SRes Xz_ReadBackward(CXzStream *p, ILookInStream *stream, Int64 *startOffset, ISzAlloc *alloc) +static SRes Xz_ReadBackward(CXzStream *p, ILookInStream *stream, Int64 *startOffset, ISzAllocPtr alloc) { UInt64 indexSize; Byte buf[XZ_STREAM_FOOTER_SIZE]; @@ -154,7 +160,7 @@ static SRes Xz_ReadBackward(CXzStream *p, ILookInStream *stream, Int64 *startOff pos -= XZ_STREAM_FOOTER_SIZE; RINOK(LookInStream_SeekRead_ForArc(stream, pos, buf, XZ_STREAM_FOOTER_SIZE)); - if (memcmp(buf + 10, XZ_FOOTER_SIG, XZ_FOOTER_SIG_SIZE) != 0) + if (!XZ_FOOTER_SIG_CHECK(buf + 10)) { UInt32 total = 0; pos += XZ_STREAM_FOOTER_SIZE; @@ -187,7 +193,7 @@ static SRes Xz_ReadBackward(CXzStream *p, ILookInStream *stream, Int64 *startOff return SZ_ERROR_NO_ARCHIVE; pos -= XZ_STREAM_FOOTER_SIZE; RINOK(LookInStream_SeekRead_ForArc(stream, pos, buf, XZ_STREAM_FOOTER_SIZE)); - if (memcmp(buf + 10, XZ_FOOTER_SIG, XZ_FOOTER_SIG_SIZE) != 0) + if (!XZ_FOOTER_SIG_CHECK(buf + 10)) return SZ_ERROR_NO_ARCHIVE; } @@ -224,7 +230,7 @@ static SRes Xz_ReadBackward(CXzStream *p, ILookInStream *stream, Int64 *startOff SecToRead_CreateVTable(&secToRead); secToRead.realStream = stream; - RINOK(Xz_ReadHeader(&headerFlags, &secToRead.s)); + RINOK(Xz_ReadHeader(&headerFlags, &secToRead.vt)); return (p->flags == headerFlags) ? SZ_OK : SZ_ERROR_ARCHIVE; } } @@ -238,12 +244,12 @@ void Xzs_Construct(CXzs *p) p->streams = 0; } -void Xzs_Free(CXzs *p, ISzAlloc *alloc) +void Xzs_Free(CXzs *p, ISzAllocPtr alloc) { size_t i; for (i = 0; i < p->num; i++) Xz_Free(&p->streams[i], alloc); - alloc->Free(alloc, p->streams); + ISzAlloc_Free(alloc, p->streams); p->num = p->numAllocated = 0; p->streams = 0; } @@ -262,7 +268,7 @@ UInt64 Xzs_GetUnpackSize(const CXzs *p) UInt64 size = 0; size_t i; for (i = 0; i < p->num; i++) - ADD_SIZE_CHECH(size, Xz_GetUnpackSize(&p->streams[i])); + ADD_SIZE_CHECK(size, Xz_GetUnpackSize(&p->streams[i])); return size; } @@ -272,15 +278,15 @@ UInt64 Xzs_GetPackSize(const CXzs *p) UInt64 size = 0; size_t i; for (i = 0; i < p->num; i++) - ADD_SIZE_CHECH(size, Xz_GetTotalSize(&p->streams[i])); + ADD_SIZE_CHECK(size, Xz_GetTotalSize(&p->streams[i])); return size; } */ -SRes Xzs_ReadBackward(CXzs *p, ILookInStream *stream, Int64 *startOffset, ICompressProgress *progress, ISzAlloc *alloc) +SRes Xzs_ReadBackward(CXzs *p, ILookInStream *stream, Int64 *startOffset, ICompressProgress *progress, ISzAllocPtr alloc) { Int64 endOffset = 0; - RINOK(stream->Seek(stream, &endOffset, SZ_SEEK_END)); + RINOK(ILookInStream_Seek(stream, &endOffset, SZ_SEEK_END)); *startOffset = endOffset; for (;;) { @@ -293,20 +299,20 @@ SRes Xzs_ReadBackward(CXzs *p, ILookInStream *stream, Int64 *startOffset, ICompr if (p->num == p->numAllocated) { size_t newNum = p->num + p->num / 4 + 1; - Byte *data = (Byte *)alloc->Alloc(alloc, newNum * sizeof(CXzStream)); - if (data == 0) + Byte *data = (Byte *)ISzAlloc_Alloc(alloc, newNum * sizeof(CXzStream)); + if (!data) return SZ_ERROR_MEM; p->numAllocated = newNum; if (p->num != 0) memcpy(data, p->streams, p->num * sizeof(CXzStream)); - alloc->Free(alloc, p->streams); + ISzAlloc_Free(alloc, p->streams); p->streams = (CXzStream *)data; } p->streams[p->num++] = st; if (*startOffset == 0) break; RINOK(LookInStream_SeekTo(stream, *startOffset)); - if (progress && progress->Progress(progress, endOffset - *startOffset, (UInt64)(Int64)-1) != SZ_OK) + if (progress && ICompressProgress_Progress(progress, endOffset - *startOffset, (UInt64)(Int64)-1) != SZ_OK) return SZ_ERROR_PROGRESS; } return SZ_OK; diff --git a/src/util/vfs/vfs-lzma.c b/src/util/vfs/vfs-lzma.c index e72f924a0..da2a6fe8e 100644 --- a/src/util/vfs/vfs-lzma.c +++ b/src/util/vfs/vfs-lzma.c @@ -16,6 +16,8 @@ #include "third-party/lzma/7zFile.h" #include "third-party/lzma/7zVersion.h" +#define BUFFER_SIZE 0x2000 + struct VDirEntry7z { struct VDirEntry d; @@ -30,7 +32,7 @@ struct VDir7z { // What is all this garbage? CFileInStream archiveStream; - CLookToRead lookStream; + CLookToRead2 lookStream; CSzArEx db; ISzAlloc allocImp; ISzAlloc allocTempImp; @@ -88,15 +90,18 @@ struct VDir* VDirOpen7z(const char* path, int flags) { vd->allocTempImp.Free = SzFreeTemp; FileInStream_CreateVTable(&vd->archiveStream); - LookToRead_CreateVTable(&vd->lookStream, False); + LookToRead2_CreateVTable(&vd->lookStream, False); - vd->lookStream.realStream = &vd->archiveStream.s; - LookToRead_Init(&vd->lookStream); + vd->lookStream.realStream = &vd->archiveStream.vt; + vd->lookStream.buf = malloc(BUFFER_SIZE); + vd->lookStream.bufSize = BUFFER_SIZE; + + LookToRead2_Init(&vd->lookStream); CrcGenerateTable(); SzArEx_Init(&vd->db); - SRes res = SzArEx_Open(&vd->db, &vd->lookStream.s, &vd->allocImp, &vd->allocTempImp); + SRes res = SzArEx_Open(&vd->db, &vd->lookStream.vt, &vd->allocImp, &vd->allocTempImp); if (res != SZ_OK) { SzArEx_Free(&vd->db, &vd->allocImp); File_Close(&vd->archiveStream.file); @@ -211,6 +216,7 @@ bool _vd7zClose(struct VDir* vd) { SzArEx_Free(&vd7z->db, &vd7z->allocImp); File_Close(&vd7z->archiveStream.file); + free(vd7z->lookStream.buf); free(vd7z->dirent.utf8); vd7z->dirent.utf8 = 0; @@ -281,7 +287,7 @@ struct VFile* _vd7zOpenFile(struct VDir* vd, const char* path, int mode) { UInt32 blockIndex; vf->outBuffer = 0; - SRes res = SzArEx_Extract(&vd7z->db, &vd7z->lookStream.s, i, &blockIndex, + SRes res = SzArEx_Extract(&vd7z->db, &vd7z->lookStream.vt, i, &blockIndex, &vf->outBuffer, &outBufferSize, &vf->bufferOffset, &vf->size, &vd7z->allocImp, &vd7z->allocTempImp);