diff options
author | robot-contrib <robot-contrib@yandex-team.com> | 2024-07-24 12:03:09 +0300 |
---|---|---|
committer | robot-contrib <robot-contrib@yandex-team.com> | 2024-07-24 12:15:10 +0300 |
commit | 2cda5506bf4b2fb1c7870fd352b1d4f8dd1c76bf (patch) | |
tree | 7976a154a3d209dd8e55a834b02c4b4c3467b897 /contrib | |
parent | af231441770fd2f0caaa23a8bd4106eaefe80759 (diff) | |
download | ydb-2cda5506bf4b2fb1c7870fd352b1d4f8dd1c76bf.tar.gz |
Update contrib/libs/lz4 to 1.10.0
46c03ca2c4745ff3ce7fdebd47424736854d67e9
Diffstat (limited to 'contrib')
-rw-r--r-- | contrib/libs/lz4/README.md | 26 | ||||
-rw-r--r-- | contrib/libs/lz4/lz4.c | 305 | ||||
-rw-r--r-- | contrib/libs/lz4/lz4.h | 180 | ||||
-rw-r--r-- | contrib/libs/lz4/lz4file.c | 192 | ||||
-rw-r--r-- | contrib/libs/lz4/lz4file.h | 4 | ||||
-rw-r--r-- | contrib/libs/lz4/lz4frame.c | 300 | ||||
-rw-r--r-- | contrib/libs/lz4/lz4frame.h | 265 | ||||
-rw-r--r-- | contrib/libs/lz4/lz4hc.c | 1309 | ||||
-rw-r--r-- | contrib/libs/lz4/lz4hc.h | 85 | ||||
-rw-r--r-- | contrib/libs/lz4/ya.make | 4 |
10 files changed, 1776 insertions, 894 deletions
diff --git a/contrib/libs/lz4/README.md b/contrib/libs/lz4/README.md index 08d1cef2bf..9039da35af 100644 --- a/contrib/libs/lz4/README.md +++ b/contrib/libs/lz4/README.md @@ -100,7 +100,7 @@ The following build macro can be selected to adjust source code behavior at comp passed as argument to become a compression state is suitably aligned. This test can be disabled if it proves flaky, by setting this value to 0. -- `LZ4_USER_MEMORY_FUNCTIONS` : replace calls to `<stdlib,h>`'s `malloc()`, `calloc()` and `free()` +- `LZ4_USER_MEMORY_FUNCTIONS` : replace calls to `<stdlib.h>`'s `malloc()`, `calloc()` and `free()` by user-defined functions, which must be named `LZ4_malloc()`, `LZ4_calloc()` and `LZ4_free()`. User functions must be available at link time. @@ -108,6 +108,12 @@ The following build macro can be selected to adjust source code behavior at comp Remove support of dynamic memory allocation. For more details, see description of this macro in `lib/lz4.c`. +- `LZ4_STATIC_LINKING_ONLY_ENDIANNESS_INDEPENDENT_OUTPUT` : experimental feature aimed at producing the same + compressed output on platforms of different endianness (i.e. little-endian and big-endian). + Output on little-endian platforms shall remain unchanged, while big-endian platforms will start producing + the same output as little-endian ones. This isn't expected to impact backward- and forward-compatibility + in any way. + - `LZ4_FREESTANDING` : by setting this build macro to 1, LZ4/HC removes dependencies on the C standard library, including allocation functions and `memmove()`, `memcpy()`, and `memset()`. @@ -115,6 +121,24 @@ The following build macro can be selected to adjust source code behavior at comp (embedded, bootloader, etc). For more details, see description of this macro in `lib/lz4.h`. +- `LZ4_HEAPMODE` : Select how stateless compression functions like `LZ4_compress_default()` + allocate memory for their hash table, + in memory stack (0:default, fastest), or in memory heap (1:requires malloc()). + +- `LZ4HC_HEAPMODE` : Select how stateless HC compression functions like `LZ4_compress_HC()` + allocate memory for their workspace: + in stack (0), or in heap (1:default). + Since workspace is rather large, stack can be inconvenient, hence heap mode is recommended. + +- `LZ4F_HEAPMODE` : selects how `LZ4F_compressFrame()` allocates the compression state, + either on stack (default, value 0) or using heap memory (value 1). + + +#### Makefile variables + +The following `Makefile` variables can be selected to alter the profile of produced binaries : +- `BUILD_SHARED` : generate `liblz4` dynamic library (enabled by default) +- `BUILD_STATIC` : generate `liblz4` static library (enabled by default) #### Amalgamation diff --git a/contrib/libs/lz4/lz4.c b/contrib/libs/lz4/lz4.c index 654bfdf32f..a2f7abee19 100644 --- a/contrib/libs/lz4/lz4.c +++ b/contrib/libs/lz4/lz4.c @@ -1,6 +1,6 @@ /* LZ4 - Fast LZ compression algorithm - Copyright (C) 2011-2020, Yann Collet. + Copyright (C) 2011-2023, Yann Collet. BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) @@ -37,7 +37,8 @@ **************************************/ /* * LZ4_HEAPMODE : - * Select how default compression functions will allocate memory for their hash table, + * Select how stateless compression functions like `LZ4_compress_default()` + * allocate memory for their hash table, * in memory stack (0:default, fastest), or in memory heap (1:requires malloc()). */ #ifndef LZ4_HEAPMODE @@ -78,7 +79,7 @@ ( defined(__ARM_ARCH_6__) || defined(__ARM_ARCH_6J__) || defined(__ARM_ARCH_6K__) \ || defined(__ARM_ARCH_6Z__) || defined(__ARM_ARCH_6ZK__) || defined(__ARM_ARCH_6T2__) ) # define LZ4_FORCE_MEMORY_ACCESS 2 -# elif (defined(__INTEL_COMPILER) && !defined(_WIN32)) || defined(__GNUC__) +# elif (defined(__INTEL_COMPILER) && !defined(_WIN32)) || defined(__GNUC__) || defined(_MSC_VER) # define LZ4_FORCE_MEMORY_ACCESS 1 # endif #endif @@ -105,15 +106,13 @@ # define LZ4_SRC_INCLUDED 1 #endif -#ifndef LZ4_STATIC_LINKING_ONLY -#define LZ4_STATIC_LINKING_ONLY -#endif - #ifndef LZ4_DISABLE_DEPRECATE_WARNINGS -#define LZ4_DISABLE_DEPRECATE_WARNINGS /* due to LZ4_decompress_safe_withPrefix64k */ +# define LZ4_DISABLE_DEPRECATE_WARNINGS /* due to LZ4_decompress_safe_withPrefix64k */ #endif -#define LZ4_STATIC_LINKING_ONLY /* LZ4_DISTANCE_MAX */ +#ifndef LZ4_STATIC_LINKING_ONLY +# define LZ4_STATIC_LINKING_ONLY +#endif #include "lz4.h" /* see also "memory routines" below */ @@ -125,14 +124,17 @@ # include <intrin.h> /* only present in VS2005+ */ # pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */ # pragma warning(disable : 6237) /* disable: C6237: conditional expression is always 0 */ +# pragma warning(disable : 6239) /* disable: C6239: (<non-zero constant> && <expression>) always evaluates to the result of <expression> */ +# pragma warning(disable : 6240) /* disable: C6240: (<expression> && <non-zero constant>) always evaluates to the result of <expression> */ +# pragma warning(disable : 6326) /* disable: C6326: Potential comparison of a constant with another constant */ #endif /* _MSC_VER */ #ifndef LZ4_FORCE_INLINE -# ifdef _MSC_VER /* Visual Studio */ +# if defined (_MSC_VER) && !defined (__clang__) /* MSVC */ # define LZ4_FORCE_INLINE static __forceinline # else # if defined (__cplusplus) || defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L /* C99 */ -# ifdef __GNUC__ +# if defined (__GNUC__) || defined (__clang__) # define LZ4_FORCE_INLINE static inline __attribute__((always_inline)) # else # define LZ4_FORCE_INLINE static inline @@ -279,7 +281,7 @@ static const int LZ4_minLength = (MFLIMIT+1); static int g_debuglog_enable = 1; # define DEBUGLOG(l, ...) { \ if ((g_debuglog_enable) && (l<=LZ4_DEBUG)) { \ - fprintf(stderr, __FILE__ ": "); \ + fprintf(stderr, __FILE__ " %i: ", __LINE__); \ fprintf(stderr, __VA_ARGS__); \ fprintf(stderr, " \n"); \ } } @@ -364,6 +366,11 @@ static unsigned LZ4_isLittleEndian(void) return one.c[0]; } +#if defined(__GNUC__) || defined(__INTEL_COMPILER) +#define LZ4_PACK( __Declaration__ ) __Declaration__ __attribute__((__packed__)) +#elif defined(_MSC_VER) +#define LZ4_PACK( __Declaration__ ) __pragma( pack(push, 1) ) __Declaration__ __pragma( pack(pop)) +#endif #if defined(LZ4_FORCE_MEMORY_ACCESS) && (LZ4_FORCE_MEMORY_ACCESS==2) /* lie to the compiler about data alignment; use with caution */ @@ -379,14 +386,16 @@ static void LZ4_write32(void* memPtr, U32 value) { *(U32*)memPtr = value; } /* __pack instructions are safer, but compiler specific, hence potentially problematic for some compilers */ /* currently only defined for gcc and icc */ -typedef union { U16 u16; U32 u32; reg_t uArch; } __attribute__((packed)) LZ4_unalign; +LZ4_PACK(typedef struct { U16 u16; }) LZ4_unalign16; +LZ4_PACK(typedef struct { U32 u32; }) LZ4_unalign32; +LZ4_PACK(typedef struct { reg_t uArch; }) LZ4_unalignST; -static U16 LZ4_read16(const void* ptr) { return ((const LZ4_unalign*)ptr)->u16; } -static U32 LZ4_read32(const void* ptr) { return ((const LZ4_unalign*)ptr)->u32; } -static reg_t LZ4_read_ARCH(const void* ptr) { return ((const LZ4_unalign*)ptr)->uArch; } +static U16 LZ4_read16(const void* ptr) { return ((const LZ4_unalign16*)ptr)->u16; } +static U32 LZ4_read32(const void* ptr) { return ((const LZ4_unalign32*)ptr)->u32; } +static reg_t LZ4_read_ARCH(const void* ptr) { return ((const LZ4_unalignST*)ptr)->uArch; } -static void LZ4_write16(void* memPtr, U16 value) { ((LZ4_unalign*)memPtr)->u16 = value; } -static void LZ4_write32(void* memPtr, U32 value) { ((LZ4_unalign*)memPtr)->u32 = value; } +static void LZ4_write16(void* memPtr, U16 value) { ((LZ4_unalign16*)memPtr)->u16 = value; } +static void LZ4_write32(void* memPtr, U32 value) { ((LZ4_unalign32*)memPtr)->u32 = value; } #else /* safe and portable access using memcpy() */ @@ -424,9 +433,21 @@ static U16 LZ4_readLE16(const void* memPtr) return LZ4_read16(memPtr); } else { const BYTE* p = (const BYTE*)memPtr; - return (U16)((U16)p[0] + (p[1]<<8)); + return (U16)((U16)p[0] | (p[1]<<8)); + } +} + +#ifdef LZ4_STATIC_LINKING_ONLY_ENDIANNESS_INDEPENDENT_OUTPUT +static U32 LZ4_readLE32(const void* memPtr) +{ + if (LZ4_isLittleEndian()) { + return LZ4_read32(memPtr); + } else { + const BYTE* p = (const BYTE*)memPtr; + return (U32)p[0] | (p[1]<<8) | (p[2]<<16) | (p[3]<<24); } } +#endif static void LZ4_writeLE16(void* memPtr, U16 value) { @@ -509,7 +530,7 @@ LZ4_wildCopy32(void* dstPtr, const void* srcPtr, void* dstEnd) /* LZ4_memcpy_using_offset() presumes : * - dstEnd >= dstPtr + MINMATCH - * - there is at least 8 bytes available to write after dstEnd */ + * - there is at least 12 bytes available to write after dstEnd */ LZ4_FORCE_INLINE void LZ4_memcpy_using_offset(BYTE* dstPtr, const BYTE* srcPtr, BYTE* dstEnd, const size_t offset) { @@ -524,12 +545,12 @@ LZ4_memcpy_using_offset(BYTE* dstPtr, const BYTE* srcPtr, BYTE* dstEnd, const si case 2: LZ4_memcpy(v, srcPtr, 2); LZ4_memcpy(&v[2], srcPtr, 2); -#if defined(_MSC_VER) && (_MSC_VER <= 1933) /* MSVC 2022 ver 17.3 or earlier */ +#if defined(_MSC_VER) && (_MSC_VER <= 1937) /* MSVC 2022 ver 17.7 or earlier */ # pragma warning(push) # pragma warning(disable : 6385) /* warning C6385: Reading invalid data from 'v'. */ #endif LZ4_memcpy(&v[4], v, 4); -#if defined(_MSC_VER) && (_MSC_VER <= 1933) /* MSVC 2022 ver 17.3 or earlier */ +#if defined(_MSC_VER) && (_MSC_VER <= 1937) /* MSVC 2022 ver 17.7 or earlier */ # pragma warning(pop) #endif break; @@ -776,7 +797,12 @@ LZ4_FORCE_INLINE U32 LZ4_hash5(U64 sequence, tableType_t const tableType) LZ4_FORCE_INLINE U32 LZ4_hashPosition(const void* const p, tableType_t const tableType) { if ((sizeof(reg_t)==8) && (tableType != byU16)) return LZ4_hash5(LZ4_read_ARCH(p), tableType); + +#ifdef LZ4_STATIC_LINKING_ONLY_ENDIANNESS_INDEPENDENT_OUTPUT + return LZ4_hash4(LZ4_readLE32(p), tableType); +#else return LZ4_hash4(LZ4_read32(p), tableType); +#endif } LZ4_FORCE_INLINE void LZ4_clearHash(U32 h, void* tableBase, tableType_t const tableType) @@ -803,23 +829,19 @@ LZ4_FORCE_INLINE void LZ4_putIndexOnHash(U32 idx, U32 h, void* tableBase, tableT } } +/* LZ4_putPosition*() : only used in byPtr mode */ LZ4_FORCE_INLINE void LZ4_putPositionOnHash(const BYTE* p, U32 h, - void* tableBase, tableType_t const tableType, - const BYTE* srcBase) + void* tableBase, tableType_t const tableType) { - switch (tableType) - { - case clearedTable: { /* illegal! */ assert(0); return; } - case byPtr: { const BYTE** hashTable = (const BYTE**)tableBase; hashTable[h] = p; return; } - case byU32: { U32* hashTable = (U32*) tableBase; hashTable[h] = (U32)(p-srcBase); return; } - case byU16: { U16* hashTable = (U16*) tableBase; hashTable[h] = (U16)(p-srcBase); return; } - } + const BYTE** const hashTable = (const BYTE**)tableBase; + assert(tableType == byPtr); (void)tableType; + hashTable[h] = p; } -LZ4_FORCE_INLINE void LZ4_putPosition(const BYTE* p, void* tableBase, tableType_t tableType, const BYTE* srcBase) +LZ4_FORCE_INLINE void LZ4_putPosition(const BYTE* p, void* tableBase, tableType_t tableType) { U32 const h = LZ4_hashPosition(p, tableType); - LZ4_putPositionOnHash(p, h, tableBase, tableType, srcBase); + LZ4_putPositionOnHash(p, h, tableBase, tableType); } /* LZ4_getIndexOnHash() : @@ -844,20 +866,18 @@ LZ4_FORCE_INLINE U32 LZ4_getIndexOnHash(U32 h, const void* tableBase, tableType_ assert(0); return 0; /* forbidden case */ } -static const BYTE* LZ4_getPositionOnHash(U32 h, const void* tableBase, tableType_t tableType, const BYTE* srcBase) +static const BYTE* LZ4_getPositionOnHash(U32 h, const void* tableBase, tableType_t tableType) { - if (tableType == byPtr) { const BYTE* const* hashTable = (const BYTE* const*) tableBase; return hashTable[h]; } - if (tableType == byU32) { const U32* const hashTable = (const U32*) tableBase; return hashTable[h] + srcBase; } - { const U16* const hashTable = (const U16*) tableBase; return hashTable[h] + srcBase; } /* default, to ensure a return */ + assert(tableType == byPtr); (void)tableType; + { const BYTE* const* hashTable = (const BYTE* const*) tableBase; return hashTable[h]; } } LZ4_FORCE_INLINE const BYTE* LZ4_getPosition(const BYTE* p, - const void* tableBase, tableType_t tableType, - const BYTE* srcBase) + const void* tableBase, tableType_t tableType) { U32 const h = LZ4_hashPosition(p, tableType); - return LZ4_getPositionOnHash(h, tableBase, tableType, srcBase); + return LZ4_getPositionOnHash(h, tableBase, tableType); } LZ4_FORCE_INLINE void @@ -901,9 +921,9 @@ LZ4_prepareTable(LZ4_stream_t_internal* const cctx, cctx->dictSize = 0; } -/** LZ4_compress_generic() : +/** LZ4_compress_generic_validated() : * inlined, to ensure branches are decided at compilation time. - * Presumed already validated at this stage: + * The following conditions are presumed already validated: * - source != NULL * - inputSize > 0 */ @@ -921,10 +941,10 @@ LZ4_FORCE_INLINE int LZ4_compress_generic_validated( const int acceleration) { int result; - const BYTE* ip = (const BYTE*) source; + const BYTE* ip = (const BYTE*)source; U32 const startIndex = cctx->currentOffset; - const BYTE* base = (const BYTE*) source - startIndex; + const BYTE* base = (const BYTE*)source - startIndex; const BYTE* lowLimit; const LZ4_stream_t_internal* dictCtx = (const LZ4_stream_t_internal*) cctx->dictCtx; @@ -932,7 +952,8 @@ LZ4_FORCE_INLINE int LZ4_compress_generic_validated( dictDirective == usingDictCtx ? dictCtx->dictionary : cctx->dictionary; const U32 dictSize = dictDirective == usingDictCtx ? dictCtx->dictSize : cctx->dictSize; - const U32 dictDelta = (dictDirective == usingDictCtx) ? startIndex - dictCtx->currentOffset : 0; /* make indexes in dictCtx comparable with index in current context */ + const U32 dictDelta = + (dictDirective == usingDictCtx) ? startIndex - dictCtx->currentOffset : 0; /* make indexes in dictCtx comparable with indexes in current context */ int const maybe_extMem = (dictDirective == usingExtDict) || (dictDirective == usingDictCtx); U32 const prefixIdxLimit = startIndex - dictSize; /* used when dictDirective == dictSmall */ @@ -957,11 +978,11 @@ LZ4_FORCE_INLINE int LZ4_compress_generic_validated( DEBUGLOG(5, "LZ4_compress_generic_validated: srcSize=%i, tableType=%u", inputSize, tableType); assert(ip != NULL); + if (tableType == byU16) assert(inputSize<LZ4_64Klimit); /* Size too large (not within 64K limit) */ + if (tableType == byPtr) assert(dictDirective==noDict); /* only supported use case with byPtr */ /* If init conditions are not met, we don't have to mark stream * as having dirty context, since no action was taken yet */ if (outputDirective == fillOutput && maxOutputSize < 1) { return 0; } /* Impossible to store anything */ - if ((tableType == byU16) && (inputSize>=LZ4_64Klimit)) { return 0; } /* Size too large (not within 64K limit) */ - if (tableType==byPtr) assert(dictDirective==noDict); /* only supported use case with byPtr */ assert(acceleration >= 1); lowLimit = (const BYTE*)source - (dictDirective == withPrefix64k ? dictSize : 0); @@ -981,7 +1002,12 @@ LZ4_FORCE_INLINE int LZ4_compress_generic_validated( if (inputSize<LZ4_minLength) goto _last_literals; /* Input too small, no compression (all literals) */ /* First Byte */ - LZ4_putPosition(ip, cctx->hashTable, tableType, base); + { U32 const h = LZ4_hashPosition(ip, tableType); + if (tableType == byPtr) { + LZ4_putPositionOnHash(ip, h, cctx->hashTable, byPtr); + } else { + LZ4_putIndexOnHash(startIndex, h, cctx->hashTable, tableType); + } } ip++; forwardH = LZ4_hashPosition(ip, tableType); /* Main Loop */ @@ -1004,9 +1030,9 @@ LZ4_FORCE_INLINE int LZ4_compress_generic_validated( if (unlikely(forwardIp > mflimitPlusOne)) goto _last_literals; assert(ip < mflimitPlusOne); - match = LZ4_getPositionOnHash(h, cctx->hashTable, tableType, base); + match = LZ4_getPositionOnHash(h, cctx->hashTable, tableType); forwardH = LZ4_hashPosition(forwardIp, tableType); - LZ4_putPositionOnHash(ip, h, cctx->hashTable, tableType, base); + LZ4_putPositionOnHash(ip, h, cctx->hashTable, tableType); } while ( (match+LZ4_DISTANCE_MAX < ip) || (LZ4_read32(match) != LZ4_read32(ip)) ); @@ -1077,7 +1103,10 @@ LZ4_FORCE_INLINE int LZ4_compress_generic_validated( /* Catch up */ filledIp = ip; - while (((ip>anchor) & (match > lowLimit)) && (unlikely(ip[-1]==match[-1]))) { ip--; match--; } + assert(ip > anchor); /* this is always true as ip has been advanced before entering the main loop */ + if ((match > lowLimit) && unlikely(ip[-1] == match[-1])) { + do { ip--; match--; } while (((ip > anchor) & (match > lowLimit)) && (unlikely(ip[-1] == match[-1]))); + } /* Encode Literals */ { unsigned const litLength = (unsigned)(ip - anchor); @@ -1092,7 +1121,7 @@ LZ4_FORCE_INLINE int LZ4_compress_generic_validated( goto _last_literals; } if (litLength >= RUN_MASK) { - int len = (int)(litLength - RUN_MASK); + unsigned len = litLength - RUN_MASK; *token = (RUN_MASK<<ML_BITS); for(; len >= 255 ; len-=255) *op++ = 255; *op++ = (BYTE)len; @@ -1204,13 +1233,19 @@ _next_match: if (ip >= mflimitPlusOne) break; /* Fill table */ - LZ4_putPosition(ip-2, cctx->hashTable, tableType, base); + { U32 const h = LZ4_hashPosition(ip-2, tableType); + if (tableType == byPtr) { + LZ4_putPositionOnHash(ip-2, h, cctx->hashTable, byPtr); + } else { + U32 const idx = (U32)((ip-2) - base); + LZ4_putIndexOnHash(idx, h, cctx->hashTable, tableType); + } } /* Test next position */ if (tableType == byPtr) { - match = LZ4_getPosition(ip, cctx->hashTable, tableType, base); - LZ4_putPosition(ip, cctx->hashTable, tableType, base); + match = LZ4_getPosition(ip, cctx->hashTable, tableType); + LZ4_putPosition(ip, cctx->hashTable, tableType); if ( (match+LZ4_DISTANCE_MAX >= ip) && (LZ4_read32(match) == LZ4_read32(ip)) ) { token=op++; *token=0; goto _next_match; } @@ -1224,6 +1259,7 @@ _next_match: if (dictDirective == usingDictCtx) { if (matchIndex < startIndex) { /* there was no match, try the dictionary */ + assert(tableType == byU32); matchIndex = LZ4_getIndexOnHash(h, dictCtx->hashTable, byU32); match = dictBase + matchIndex; lowLimit = dictionary; /* required for match length counter */ @@ -1377,9 +1413,10 @@ int LZ4_compress_fast_extState(void* state, const char* source, char* dest, int */ int LZ4_compress_fast_extState_fastReset(void* state, const char* src, char* dst, int srcSize, int dstCapacity, int acceleration) { - LZ4_stream_t_internal* ctx = &((LZ4_stream_t*)state)->internal_donotuse; + LZ4_stream_t_internal* const ctx = &((LZ4_stream_t*)state)->internal_donotuse; if (acceleration < 1) acceleration = LZ4_ACCELERATION_DEFAULT; if (acceleration > LZ4_ACCELERATION_MAX) acceleration = LZ4_ACCELERATION_MAX; + assert(ctx != NULL); if (dstCapacity >= LZ4_compressBound(srcSize)) { if (srcSize < LZ4_64Klimit) { @@ -1413,17 +1450,17 @@ int LZ4_compress_fast_extState_fastReset(void* state, const char* src, char* dst } -int LZ4_compress_fast(const char* source, char* dest, int inputSize, int maxOutputSize, int acceleration) +int LZ4_compress_fast(const char* src, char* dest, int srcSize, int dstCapacity, int acceleration) { int result; #if (LZ4_HEAPMODE) - LZ4_stream_t* ctxPtr = (LZ4_stream_t*)ALLOC(sizeof(LZ4_stream_t)); /* malloc-calloc always properly aligned */ + LZ4_stream_t* const ctxPtr = (LZ4_stream_t*)ALLOC(sizeof(LZ4_stream_t)); /* malloc-calloc always properly aligned */ if (ctxPtr == NULL) return 0; #else LZ4_stream_t ctx; LZ4_stream_t* const ctxPtr = &ctx; #endif - result = LZ4_compress_fast_extState(ctxPtr, source, dest, inputSize, maxOutputSize, acceleration); + result = LZ4_compress_fast_extState(ctxPtr, src, dest, srcSize, dstCapacity, acceleration); #if (LZ4_HEAPMODE) FREEMEM(ctxPtr); @@ -1432,43 +1469,51 @@ int LZ4_compress_fast(const char* source, char* dest, int inputSize, int maxOutp } -int LZ4_compress_default(const char* src, char* dst, int srcSize, int maxOutputSize) +int LZ4_compress_default(const char* src, char* dst, int srcSize, int dstCapacity) { - return LZ4_compress_fast(src, dst, srcSize, maxOutputSize, 1); + return LZ4_compress_fast(src, dst, srcSize, dstCapacity, 1); } /* Note!: This function leaves the stream in an unclean/broken state! * It is not safe to subsequently use the same state with a _fastReset() or * _continue() call without resetting it. */ -static int LZ4_compress_destSize_extState (LZ4_stream_t* state, const char* src, char* dst, int* srcSizePtr, int targetDstSize) +static int LZ4_compress_destSize_extState_internal(LZ4_stream_t* state, const char* src, char* dst, int* srcSizePtr, int targetDstSize, int acceleration) { void* const s = LZ4_initStream(state, sizeof (*state)); assert(s != NULL); (void)s; if (targetDstSize >= LZ4_compressBound(*srcSizePtr)) { /* compression success is guaranteed */ - return LZ4_compress_fast_extState(state, src, dst, *srcSizePtr, targetDstSize, 1); + return LZ4_compress_fast_extState(state, src, dst, *srcSizePtr, targetDstSize, acceleration); } else { if (*srcSizePtr < LZ4_64Klimit) { - return LZ4_compress_generic(&state->internal_donotuse, src, dst, *srcSizePtr, srcSizePtr, targetDstSize, fillOutput, byU16, noDict, noDictIssue, 1); + return LZ4_compress_generic(&state->internal_donotuse, src, dst, *srcSizePtr, srcSizePtr, targetDstSize, fillOutput, byU16, noDict, noDictIssue, acceleration); } else { tableType_t const addrMode = ((sizeof(void*)==4) && ((uptrval)src > LZ4_DISTANCE_MAX)) ? byPtr : byU32; - return LZ4_compress_generic(&state->internal_donotuse, src, dst, *srcSizePtr, srcSizePtr, targetDstSize, fillOutput, addrMode, noDict, noDictIssue, 1); + return LZ4_compress_generic(&state->internal_donotuse, src, dst, *srcSizePtr, srcSizePtr, targetDstSize, fillOutput, addrMode, noDict, noDictIssue, acceleration); } } } +int LZ4_compress_destSize_extState(void* state, const char* src, char* dst, int* srcSizePtr, int targetDstSize, int acceleration) +{ + int const r = LZ4_compress_destSize_extState_internal((LZ4_stream_t*)state, src, dst, srcSizePtr, targetDstSize, acceleration); + /* clean the state on exit */ + LZ4_initStream(state, sizeof (LZ4_stream_t)); + return r; +} + int LZ4_compress_destSize(const char* src, char* dst, int* srcSizePtr, int targetDstSize) { #if (LZ4_HEAPMODE) - LZ4_stream_t* ctx = (LZ4_stream_t*)ALLOC(sizeof(LZ4_stream_t)); /* malloc-calloc always properly aligned */ + LZ4_stream_t* const ctx = (LZ4_stream_t*)ALLOC(sizeof(LZ4_stream_t)); /* malloc-calloc always properly aligned */ if (ctx == NULL) return 0; #else LZ4_stream_t ctxBody; - LZ4_stream_t* ctx = &ctxBody; + LZ4_stream_t* const ctx = &ctxBody; #endif - int result = LZ4_compress_destSize_extState(ctx, src, dst, srcSizePtr, targetDstSize); + int result = LZ4_compress_destSize_extState_internal(ctx, src, dst, srcSizePtr, targetDstSize, 1); #if (LZ4_HEAPMODE) FREEMEM(ctx); @@ -1537,14 +1582,17 @@ int LZ4_freeStream (LZ4_stream_t* LZ4_stream) #endif +typedef enum { _ld_fast, _ld_slow } LoadDict_mode_e; #define HASH_UNIT sizeof(reg_t) -int LZ4_loadDict (LZ4_stream_t* LZ4_dict, const char* dictionary, int dictSize) +int LZ4_loadDict_internal(LZ4_stream_t* LZ4_dict, + const char* dictionary, int dictSize, + LoadDict_mode_e _ld) { - LZ4_stream_t_internal* dict = &LZ4_dict->internal_donotuse; + LZ4_stream_t_internal* const dict = &LZ4_dict->internal_donotuse; const tableType_t tableType = byU32; const BYTE* p = (const BYTE*)dictionary; const BYTE* const dictEnd = p + dictSize; - const BYTE* base; + U32 idx32; DEBUGLOG(4, "LZ4_loadDict (%i bytes from %p into %p)", dictSize, dictionary, LZ4_dict); @@ -1567,19 +1615,46 @@ int LZ4_loadDict (LZ4_stream_t* LZ4_dict, const char* dictionary, int dictSize) } if ((dictEnd - p) > 64 KB) p = dictEnd - 64 KB; - base = dictEnd - dict->currentOffset; dict->dictionary = p; dict->dictSize = (U32)(dictEnd - p); dict->tableType = (U32)tableType; + idx32 = dict->currentOffset - dict->dictSize; while (p <= dictEnd-HASH_UNIT) { - LZ4_putPosition(p, dict->hashTable, tableType, base); - p+=3; + U32 const h = LZ4_hashPosition(p, tableType); + /* Note: overwriting => favors positions end of dictionary */ + LZ4_putIndexOnHash(idx32, h, dict->hashTable, tableType); + p+=3; idx32+=3; + } + + if (_ld == _ld_slow) { + /* Fill hash table with additional references, to improve compression capability */ + p = dict->dictionary; + idx32 = dict->currentOffset - dict->dictSize; + while (p <= dictEnd-HASH_UNIT) { + U32 const h = LZ4_hashPosition(p, tableType); + U32 const limit = dict->currentOffset - 64 KB; + if (LZ4_getIndexOnHash(h, dict->hashTable, tableType) <= limit) { + /* Note: not overwriting => favors positions beginning of dictionary */ + LZ4_putIndexOnHash(idx32, h, dict->hashTable, tableType); + } + p++; idx32++; + } } return (int)dict->dictSize; } +int LZ4_loadDict(LZ4_stream_t* LZ4_dict, const char* dictionary, int dictSize) +{ + return LZ4_loadDict_internal(LZ4_dict, dictionary, dictSize, _ld_fast); +} + +int LZ4_loadDictSlow(LZ4_stream_t* LZ4_dict, const char* dictionary, int dictSize) +{ + return LZ4_loadDict_internal(LZ4_dict, dictionary, dictSize, _ld_slow); +} + void LZ4_attach_dictionary(LZ4_stream_t* workingStream, const LZ4_stream_t* dictionaryStream) { const LZ4_stream_t_internal* dictCtx = (dictionaryStream == NULL) ? NULL : @@ -1711,7 +1786,7 @@ int LZ4_compress_fast_continue (LZ4_stream_t* LZ4_stream, /* Hidden debug function, to force-test external dictionary mode */ int LZ4_compress_forceExtDict (LZ4_stream_t* LZ4_dict, const char* source, char* dest, int srcSize) { - LZ4_stream_t_internal* streamPtr = &LZ4_dict->internal_donotuse; + LZ4_stream_t_internal* const streamPtr = &LZ4_dict->internal_donotuse; int result; LZ4_renormDictT(streamPtr, srcSize); @@ -1774,7 +1849,7 @@ typedef enum { decode_full_block = 0, partial_decode = 1 } earlyEnd_directive; * does not know end of input * presumes input is well formed * note : will consume at least one byte */ -size_t read_long_length_no_check(const BYTE** pp) +static size_t read_long_length_no_check(const BYTE** pp) { size_t b, l = 0; do { b = **pp; (*pp)++; l += b; } while (b==255); @@ -1911,6 +1986,17 @@ read_variable_length(const BYTE** ip, const BYTE* ilimit, if (initial_check && unlikely((*ip) >= ilimit)) { /* read limit reached */ return rvl_error; } + s = **ip; + (*ip)++; + length += s; + if (unlikely((*ip) > ilimit)) { /* read limit reached */ + return rvl_error; + } + /* accumulator overflow detection (32-bit mode only) */ + if ((sizeof(length) < 8) && unlikely(length > ((Rvl_t)(-1)/2)) ) { + return rvl_error; + } + if (likely(s != 255)) return length; do { s = **ip; (*ip)++; @@ -1919,10 +2005,10 @@ read_variable_length(const BYTE** ip, const BYTE* ilimit, return rvl_error; } /* accumulator overflow detection (32-bit mode only) */ - if ((sizeof(length)<8) && unlikely(length > ((Rvl_t)(-1)/2)) ) { + if ((sizeof(length) < 8) && unlikely(length > ((Rvl_t)(-1)/2)) ) { return rvl_error; } - } while (s==255); + } while (s == 255); return length; } @@ -1988,63 +2074,73 @@ LZ4_decompress_generic( * note : fast loop may show a regression for some client arm chips. */ #if LZ4_FAST_DEC_LOOP if ((oend - op) < FASTLOOP_SAFE_DISTANCE) { - DEBUGLOG(6, "skip fast decode loop"); + DEBUGLOG(6, "move to safe decode loop"); goto safe_decode; } /* Fast loop : decode sequences as long as output < oend-FASTLOOP_SAFE_DISTANCE */ + DEBUGLOG(6, "using fast decode loop"); while (1) { /* Main fastloop assertion: We can always wildcopy FASTLOOP_SAFE_DISTANCE */ assert(oend - op >= FASTLOOP_SAFE_DISTANCE); assert(ip < iend); token = *ip++; length = token >> ML_BITS; /* literal length */ + DEBUGLOG(7, "blockPos%6u: litLength token = %u", (unsigned)(op-(BYTE*)dst), (unsigned)length); /* decode literal length */ if (length == RUN_MASK) { size_t const addl = read_variable_length(&ip, iend-RUN_MASK, 1); - if (addl == rvl_error) { goto _output_error; } + if (addl == rvl_error) { + DEBUGLOG(6, "error reading long literal length"); + goto _output_error; + } length += addl; if (unlikely((uptrval)(op)+length<(uptrval)(op))) { goto _output_error; } /* overflow detection */ if (unlikely((uptrval)(ip)+length<(uptrval)(ip))) { goto _output_error; } /* overflow detection */ /* copy literals */ - cpy = op+length; LZ4_STATIC_ASSERT(MFLIMIT >= WILDCOPYLENGTH); - if ((cpy>oend-32) || (ip+length>iend-32)) { goto safe_literal_copy; } - LZ4_wildCopy32(op, ip, cpy); - ip += length; op = cpy; - } else { - cpy = op+length; - DEBUGLOG(7, "copy %u bytes in a 16-bytes stripe", (unsigned)length); + if ((op+length>oend-32) || (ip+length>iend-32)) { goto safe_literal_copy; } + LZ4_wildCopy32(op, ip, op+length); + ip += length; op += length; + } else if (ip <= iend-(16 + 1/*max lit + offset + nextToken*/)) { /* We don't need to check oend, since we check it once for each loop below */ - if (ip > iend-(16 + 1/*max lit + offset + nextToken*/)) { goto safe_literal_copy; } + DEBUGLOG(7, "copy %u bytes in a 16-bytes stripe", (unsigned)length); /* Literals can only be <= 14, but hope compilers optimize better when copy by a register size */ LZ4_memcpy(op, ip, 16); - ip += length; op = cpy; + ip += length; op += length; + } else { + goto safe_literal_copy; } /* get offset */ offset = LZ4_readLE16(ip); ip+=2; + DEBUGLOG(6, "blockPos%6u: offset = %u", (unsigned)(op-(BYTE*)dst), (unsigned)offset); match = op - offset; assert(match <= op); /* overflow check */ /* get matchlength */ length = token & ML_MASK; + DEBUGLOG(7, " match length token = %u (len==%u)", (unsigned)length, (unsigned)length+MINMATCH); if (length == ML_MASK) { size_t const addl = read_variable_length(&ip, iend - LASTLITERALS + 1, 0); - if (addl == rvl_error) { goto _output_error; } + if (addl == rvl_error) { + DEBUGLOG(5, "error reading long match length"); + goto _output_error; + } length += addl; length += MINMATCH; + DEBUGLOG(7, " long match length == %u", (unsigned)length); if (unlikely((uptrval)(op)+length<(uptrval)op)) { goto _output_error; } /* overflow detection */ - if ((checkOffset) && (unlikely(match + dictSize < lowPrefix))) { goto _output_error; } /* Error : offset outside buffers */ if (op + length >= oend - FASTLOOP_SAFE_DISTANCE) { goto safe_match_copy; } } else { length += MINMATCH; if (op + length >= oend - FASTLOOP_SAFE_DISTANCE) { + DEBUGLOG(7, "moving to safe_match_copy (ml==%u)", (unsigned)length); goto safe_match_copy; } @@ -2062,7 +2158,10 @@ LZ4_decompress_generic( continue; } } } - if (checkOffset && (unlikely(match + dictSize < lowPrefix))) { goto _output_error; } /* Error : offset outside buffers */ + if ( checkOffset && (unlikely(match + dictSize < lowPrefix)) ) { + DEBUGLOG(5, "Error : pos=%zi, offset=%zi => outside buffers", op-lowPrefix, op-match); + goto _output_error; + } /* match starting within external dictionary */ if ((dict==usingExtDict) && (match < lowPrefix)) { assert(dictEnd != NULL); @@ -2071,7 +2170,8 @@ LZ4_decompress_generic( DEBUGLOG(7, "partialDecoding: dictionary match, close to dstEnd"); length = MIN(length, (size_t)(oend-op)); } else { - goto _output_error; /* end-of-block condition violated */ + DEBUGLOG(6, "end-of-block condition violated") + goto _output_error; } } if (length <= (size_t)(lowPrefix-match)) { @@ -2111,10 +2211,12 @@ LZ4_decompress_generic( #endif /* Main Loop : decode remaining sequences where output < FASTLOOP_SAFE_DISTANCE */ + DEBUGLOG(6, "using safe decode loop"); while (1) { assert(ip < iend); token = *ip++; length = token >> ML_BITS; /* literal length */ + DEBUGLOG(7, "blockPos%6u: litLength token = %u", (unsigned)(op-(BYTE*)dst), (unsigned)length); /* A two-stage shortcut for the most common case: * 1) If the literal length is 0..14, and there is enough space, @@ -2135,6 +2237,7 @@ LZ4_decompress_generic( /* The second stage: prepare for match copying, decode full info. * If it doesn't work out, the info won't be wasted. */ length = token & ML_MASK; /* match length */ + DEBUGLOG(7, "blockPos%6u: matchLength token = %u (len=%u)", (unsigned)(op-(BYTE*)dst), (unsigned)length, (unsigned)length + 4); offset = LZ4_readLE16(ip); ip += 2; match = op - offset; assert(match <= op); /* check overflow */ @@ -2166,11 +2269,12 @@ LZ4_decompress_generic( if (unlikely((uptrval)(ip)+length<(uptrval)(ip))) { goto _output_error; } /* overflow detection */ } - /* copy literals */ - cpy = op+length; #if LZ4_FAST_DEC_LOOP safe_literal_copy: #endif + /* copy literals */ + cpy = op+length; + LZ4_STATIC_ASSERT(MFLIMIT >= WILDCOPYLENGTH); if ((cpy>oend-MFLIMIT) || (ip+length>iend-(2+1+LASTLITERALS))) { /* We've either hit the input parsing restriction or the output parsing restriction. @@ -2206,9 +2310,10 @@ LZ4_decompress_generic( * so check that we exactly consume the input and don't overrun the output buffer. */ if ((ip+length != iend) || (cpy > oend)) { - DEBUGLOG(6, "should have been last run of literals") - DEBUGLOG(6, "ip(%p) + length(%i) = %p != iend (%p)", ip, (int)length, ip+length, iend); - DEBUGLOG(6, "or cpy(%p) > oend(%p)", cpy, oend); + DEBUGLOG(5, "should have been last run of literals") + DEBUGLOG(5, "ip(%p) + length(%i) = %p != iend (%p)", ip, (int)length, ip+length, iend); + DEBUGLOG(5, "or cpy(%p) > (oend-MFLIMIT)(%p)", cpy, oend-MFLIMIT); + DEBUGLOG(5, "after writing %u bytes / %i bytes available", (unsigned)(op-(BYTE*)dst), outputSize); goto _output_error; } } @@ -2234,6 +2339,7 @@ LZ4_decompress_generic( /* get matchlength */ length = token & ML_MASK; + DEBUGLOG(7, "blockPos%6u: matchLength token = %u", (unsigned)(op-(BYTE*)dst), (unsigned)length); _copy_match: if (length == ML_MASK) { @@ -2323,7 +2429,7 @@ LZ4_decompress_generic( while (op < cpy) { *op++ = *match++; } } else { LZ4_memcpy(op, match, 8); - if (length > 16) { LZ4_wildCopy8(op+8, match+8, cpy); } + if (length > 16) { LZ4_wildCopy8(op+8, match+8, cpy); } } op = cpy; /* wildcopy correction */ } @@ -2418,6 +2524,7 @@ int LZ4_decompress_safe_forceExtDict(const char* source, char* dest, int compressedSize, int maxOutputSize, const void* dictStart, size_t dictSize) { + DEBUGLOG(5, "LZ4_decompress_safe_forceExtDict"); return LZ4_decompress_generic(source, dest, compressedSize, maxOutputSize, decode_full_block, usingExtDict, (BYTE*)dest, (const BYTE*)dictStart, dictSize); diff --git a/contrib/libs/lz4/lz4.h b/contrib/libs/lz4/lz4.h index 46caa7f6f5..427c1e6e94 100644 --- a/contrib/libs/lz4/lz4.h +++ b/contrib/libs/lz4/lz4.h @@ -1,7 +1,7 @@ /* * LZ4 - Fast LZ compression algorithm * Header File - * Copyright (C) 2011-2020, Yann Collet. + * Copyright (C) 2011-2023, Yann Collet. BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) @@ -129,8 +129,8 @@ extern "C" { /*------ Version ------*/ #define LZ4_VERSION_MAJOR 1 /* for breaking interface changes */ -#define LZ4_VERSION_MINOR 9 /* for new (non-breaking) interface capabilities */ -#define LZ4_VERSION_RELEASE 4 /* for tweaks, bug-fixes, or development */ +#define LZ4_VERSION_MINOR 10 /* for new (non-breaking) interface capabilities */ +#define LZ4_VERSION_RELEASE 0 /* for tweaks, bug-fixes, or development */ #define LZ4_VERSION_NUMBER (LZ4_VERSION_MAJOR *100*100 + LZ4_VERSION_MINOR *100 + LZ4_VERSION_RELEASE) @@ -144,23 +144,25 @@ LZ4LIB_API const char* LZ4_versionString (void); /**< library version string; /*-************************************ -* Tuning parameter +* Tuning memory usage **************************************/ -#define LZ4_MEMORY_USAGE_MIN 10 -#define LZ4_MEMORY_USAGE_DEFAULT 14 -#define LZ4_MEMORY_USAGE_MAX 20 - /*! * LZ4_MEMORY_USAGE : - * Memory usage formula : N->2^N Bytes (examples : 10 -> 1KB; 12 -> 4KB ; 16 -> 64KB; 20 -> 1MB; ) - * Increasing memory usage improves compression ratio, at the cost of speed. + * Can be selected at compile time, by setting LZ4_MEMORY_USAGE. + * Memory usage formula : N->2^N Bytes (examples : 10 -> 1KB; 12 -> 4KB ; 16 -> 64KB; 20 -> 1MB) + * Increasing memory usage improves compression ratio, generally at the cost of speed. * Reduced memory usage may improve speed at the cost of ratio, thanks to better cache locality. - * Default value is 14, for 16KB, which nicely fits into Intel x86 L1 cache + * Default value is 14, for 16KB, which nicely fits into most L1 caches. */ #ifndef LZ4_MEMORY_USAGE # define LZ4_MEMORY_USAGE LZ4_MEMORY_USAGE_DEFAULT #endif +/* These are absolute limits, they should not be changed by users */ +#define LZ4_MEMORY_USAGE_MIN 10 +#define LZ4_MEMORY_USAGE_DEFAULT 14 +#define LZ4_MEMORY_USAGE_MAX 20 + #if (LZ4_MEMORY_USAGE < LZ4_MEMORY_USAGE_MIN) # error "LZ4_MEMORY_USAGE is too small !" #endif @@ -189,8 +191,9 @@ LZ4LIB_API const char* LZ4_versionString (void); /**< library version string; LZ4LIB_API int LZ4_compress_default(const char* src, char* dst, int srcSize, int dstCapacity); /*! LZ4_decompress_safe() : - * compressedSize : is the exact complete size of the compressed block. - * dstCapacity : is the size of destination buffer (which must be already allocated), presumed an upper bound of decompressed size. + * @compressedSize : is the exact complete size of the compressed block. + * @dstCapacity : is the size of destination buffer (which must be already allocated), + * presumed an upper bound of decompressed size. * @return : the number of bytes decompressed into destination buffer (necessarily <= dstCapacity) * If destination buffer is not large enough, decoding will stop and output an error code (negative value). * If the source stream is detected malformed, the function will stop decoding and return a negative result. @@ -242,20 +245,20 @@ LZ4LIB_API int LZ4_compress_fast (const char* src, char* dst, int srcSize, int d LZ4LIB_API int LZ4_sizeofState(void); LZ4LIB_API int LZ4_compress_fast_extState (void* state, const char* src, char* dst, int srcSize, int dstCapacity, int acceleration); - /*! LZ4_compress_destSize() : * Reverse the logic : compresses as much data as possible from 'src' buffer - * into already allocated buffer 'dst', of size >= 'targetDestSize'. + * into already allocated buffer 'dst', of size >= 'dstCapacity'. * This function either compresses the entire 'src' content into 'dst' if it's large enough, * or fill 'dst' buffer completely with as much data as possible from 'src'. * note: acceleration parameter is fixed to "default". * - * *srcSizePtr : will be modified to indicate how many bytes where read from 'src' to fill 'dst'. + * *srcSizePtr : in+out parameter. Initially contains size of input. + * Will be modified to indicate how many bytes where read from 'src' to fill 'dst'. * New value is necessarily <= input value. - * @return : Nb bytes written into 'dst' (necessarily <= targetDestSize) + * @return : Nb bytes written into 'dst' (necessarily <= dstCapacity) * or 0 if compression fails. * - * Note : from v1.8.2 to v1.9.1, this function had a bug (fixed un v1.9.2+): + * Note : from v1.8.2 to v1.9.1, this function had a bug (fixed in v1.9.2+): * the produced compressed content could, in specific circumstances, * require to be decompressed into a destination buffer larger * by at least 1 byte than the content to decompress. @@ -266,8 +269,7 @@ LZ4LIB_API int LZ4_compress_fast_extState (void* state, const char* src, char* d * a dstCapacity which is > decompressedSize, by at least 1 byte. * See https://github.com/lz4/lz4/issues/859 for details */ -LZ4LIB_API int LZ4_compress_destSize (const char* src, char* dst, int* srcSizePtr, int targetDstSize); - +LZ4LIB_API int LZ4_compress_destSize(const char* src, char* dst, int* srcSizePtr, int targetDstSize); /*! LZ4_decompress_safe_partial() : * Decompress an LZ4 compressed block, of size 'srcSize' at position 'src', @@ -311,7 +313,7 @@ LZ4LIB_API int LZ4_decompress_safe_partial (const char* src, char* dst, int srcS ***********************************************/ typedef union LZ4_stream_u LZ4_stream_t; /* incomplete type (defined later) */ -/** +/*! Note about RC_INVOKED - RC_INVOKED is predefined symbol of rc.exe (the resource compiler which is part of MSVC/Visual Studio). @@ -361,13 +363,58 @@ LZ4LIB_API void LZ4_resetStream_fast (LZ4_stream_t* streamPtr); * LZ4_loadDict() triggers a reset, so any previous data will be forgotten. * The same dictionary will have to be loaded on decompression side for successful decoding. * Dictionary are useful for better compression of small data (KB range). - * While LZ4 accept any input as dictionary, - * results are generally better when using Zstandard's Dictionary Builder. + * While LZ4 itself accepts any input as dictionary, dictionary efficiency is also a topic. + * When in doubt, employ the Zstandard's Dictionary Builder. * Loading a size of 0 is allowed, and is the same as reset. - * @return : loaded dictionary size, in bytes (necessarily <= 64 KB) + * @return : loaded dictionary size, in bytes (note: only the last 64 KB are loaded) */ LZ4LIB_API int LZ4_loadDict (LZ4_stream_t* streamPtr, const char* dictionary, int dictSize); +/*! LZ4_loadDictSlow() : v1.10.0+ + * Same as LZ4_loadDict(), + * but uses a bit more cpu to reference the dictionary content more thoroughly. + * This is expected to slightly improve compression ratio. + * The extra-cpu cost is likely worth it if the dictionary is re-used across multiple sessions. + * @return : loaded dictionary size, in bytes (note: only the last 64 KB are loaded) + */ +LZ4LIB_API int LZ4_loadDictSlow(LZ4_stream_t* streamPtr, const char* dictionary, int dictSize); + +/*! LZ4_attach_dictionary() : stable since v1.10.0 + * + * This allows efficient re-use of a static dictionary multiple times. + * + * Rather than re-loading the dictionary buffer into a working context before + * each compression, or copying a pre-loaded dictionary's LZ4_stream_t into a + * working LZ4_stream_t, this function introduces a no-copy setup mechanism, + * in which the working stream references @dictionaryStream in-place. + * + * Several assumptions are made about the state of @dictionaryStream. + * Currently, only states which have been prepared by LZ4_loadDict() or + * LZ4_loadDictSlow() should be expected to work. + * + * Alternatively, the provided @dictionaryStream may be NULL, + * in which case any existing dictionary stream is unset. + * + * If a dictionary is provided, it replaces any pre-existing stream history. + * The dictionary contents are the only history that can be referenced and + * logically immediately precede the data compressed in the first subsequent + * compression call. + * + * The dictionary will only remain attached to the working stream through the + * first compression call, at the end of which it is cleared. + * @dictionaryStream stream (and source buffer) must remain in-place / accessible / unchanged + * through the completion of the compression session. + * + * Note: there is no equivalent LZ4_attach_*() method on the decompression side + * because there is no initialization cost, hence no need to share the cost across multiple sessions. + * To decompress LZ4 blocks using dictionary, attached or not, + * just employ the regular LZ4_setStreamDecode() for streaming, + * or the stateless LZ4_decompress_safe_usingDict() for one-shot decompression. + */ +LZ4LIB_API void +LZ4_attach_dictionary(LZ4_stream_t* workingStream, + const LZ4_stream_t* dictionaryStream); + /*! LZ4_compress_fast_continue() : * Compress 'src' content using data from previously compressed blocks, for better compression ratio. * 'dst' buffer must be already allocated. @@ -443,11 +490,24 @@ LZ4LIB_API int LZ4_setStreamDecode (LZ4_streamDecode_t* LZ4_streamDecode, const LZ4LIB_API int LZ4_decoderRingBufferSize(int maxBlockSize); #define LZ4_DECODER_RING_BUFFER_SIZE(maxBlockSize) (65536 + 14 + (maxBlockSize)) /* for static allocation; maxBlockSize presumed valid */ -/*! LZ4_decompress_*_continue() : - * These decoding functions allow decompression of consecutive blocks in "streaming" mode. - * A block is an unsplittable entity, it must be presented entirely to a decompression function. - * Decompression functions only accepts one block at a time. - * The last 64KB of previously decoded data *must* remain available and unmodified at the memory position where they were decoded. +/*! LZ4_decompress_safe_continue() : + * This decoding function allows decompression of consecutive blocks in "streaming" mode. + * The difference with the usual independent blocks is that + * new blocks are allowed to find references into former blocks. + * A block is an unsplittable entity, and must be presented entirely to the decompression function. + * LZ4_decompress_safe_continue() only accepts one block at a time. + * It's modeled after `LZ4_decompress_safe()` and behaves similarly. + * + * @LZ4_streamDecode : decompression state, tracking the position in memory of past data + * @compressedSize : exact complete size of one compressed block. + * @dstCapacity : size of destination buffer (which must be already allocated), + * must be an upper bound of decompressed size. + * @return : number of bytes decompressed into destination buffer (necessarily <= dstCapacity) + * If destination buffer is not large enough, decoding will stop and output an error code (negative value). + * If the source stream is detected malformed, the function will stop decoding and return a negative result. + * + * The last 64KB of previously decoded data *must* remain available and unmodified + * at the memory position where they were previously decoded. * If less than 64KB of data has been decoded, all the data must be present. * * Special : if decompression side sets a ring buffer, it must respect one of the following conditions : @@ -474,10 +534,10 @@ LZ4_decompress_safe_continue (LZ4_streamDecode_t* LZ4_streamDecode, int srcSize, int dstCapacity); -/*! LZ4_decompress_*_usingDict() : - * These decoding functions work the same as - * a combination of LZ4_setStreamDecode() followed by LZ4_decompress_*_continue() - * They are stand-alone, and don't need an LZ4_streamDecode_t structure. +/*! LZ4_decompress_safe_usingDict() : + * Works the same as + * a combination of LZ4_setStreamDecode() followed by LZ4_decompress_safe_continue() + * However, it's stateless: it doesn't need any LZ4_streamDecode_t state. * Dictionary is presumed stable : it must remain accessible and unmodified during decompression. * Performance tip : Decompression speed can be substantially increased * when dst == dictStart + dictSize. @@ -487,6 +547,12 @@ LZ4_decompress_safe_usingDict(const char* src, char* dst, int srcSize, int dstCapacity, const char* dictStart, int dictSize); +/*! LZ4_decompress_safe_partial_usingDict() : + * Behaves the same as LZ4_decompress_safe_partial() + * with the added ability to specify a memory segment for past data. + * Performance tip : Decompression speed can be substantially increased + * when dst == dictStart + dictSize. + */ LZ4LIB_API int LZ4_decompress_safe_partial_usingDict(const char* src, char* dst, int compressedSize, @@ -526,9 +592,9 @@ LZ4_decompress_safe_partial_usingDict(const char* src, char* dst, #define LZ4_STATIC_3504398509 #ifdef LZ4_PUBLISH_STATIC_FUNCTIONS -#define LZ4LIB_STATIC_API LZ4LIB_API +# define LZ4LIB_STATIC_API LZ4LIB_API #else -#define LZ4LIB_STATIC_API +# define LZ4LIB_STATIC_API #endif @@ -544,36 +610,11 @@ LZ4_decompress_safe_partial_usingDict(const char* src, char* dst, */ LZ4LIB_STATIC_API int LZ4_compress_fast_extState_fastReset (void* state, const char* src, char* dst, int srcSize, int dstCapacity, int acceleration); -/*! LZ4_attach_dictionary() : - * This is an experimental API that allows - * efficient use of a static dictionary many times. - * - * Rather than re-loading the dictionary buffer into a working context before - * each compression, or copying a pre-loaded dictionary's LZ4_stream_t into a - * working LZ4_stream_t, this function introduces a no-copy setup mechanism, - * in which the working stream references the dictionary stream in-place. - * - * Several assumptions are made about the state of the dictionary stream. - * Currently, only streams which have been prepared by LZ4_loadDict() should - * be expected to work. - * - * Alternatively, the provided dictionaryStream may be NULL, - * in which case any existing dictionary stream is unset. - * - * If a dictionary is provided, it replaces any pre-existing stream history. - * The dictionary contents are the only history that can be referenced and - * logically immediately precede the data compressed in the first subsequent - * compression call. - * - * The dictionary will only remain attached to the working stream through the - * first compression call, at the end of which it is cleared. The dictionary - * stream (and source buffer) must remain in-place / accessible / unchanged - * through the completion of the first compression call on the stream. +/*! LZ4_compress_destSize_extState() : introduced in v1.10.0 + * Same as LZ4_compress_destSize(), but using an externally allocated state. + * Also: exposes @acceleration */ -LZ4LIB_STATIC_API void -LZ4_attach_dictionary(LZ4_stream_t* workingStream, - const LZ4_stream_t* dictionaryStream); - +int LZ4_compress_destSize_extState(void* state, const char* src, char* dst, int* srcSizePtr, int targetDstSize, int acceleration); /*! In-place compression and decompression * @@ -685,7 +726,7 @@ struct LZ4_stream_t_internal { /* Implicit padding to ensure structure is aligned */ }; -#define LZ4_STREAM_MINSIZE ((1UL << LZ4_MEMORY_USAGE) + 32) /* static size, for inter-version compatibility */ +#define LZ4_STREAM_MINSIZE ((1UL << (LZ4_MEMORY_USAGE)) + 32) /* static size, for inter-version compatibility */ union LZ4_stream_u { char minStateSize[LZ4_STREAM_MINSIZE]; LZ4_stream_t_internal internal_donotuse; @@ -706,7 +747,7 @@ union LZ4_stream_u { * Note2: An LZ4_stream_t structure guarantees correct alignment and size. * Note3: Before v1.9.0, use LZ4_resetStream() instead **/ -LZ4LIB_API LZ4_stream_t* LZ4_initStream (void* buffer, size_t size); +LZ4LIB_API LZ4_stream_t* LZ4_initStream (void* stateBuffer, size_t size); /*! LZ4_streamDecode_t : @@ -823,11 +864,12 @@ LZ4_DEPRECATED("use LZ4_decompress_fast_usingDict() instead") LZ4LIB_API int LZ4 * But they may happen if input data is invalid (error or intentional tampering). * As a consequence, use these functions in trusted environments with trusted data **only**. */ -LZ4_DEPRECATED("This function is deprecated and unsafe. Consider using LZ4_decompress_safe() instead") +LZ4_DEPRECATED("This function is deprecated and unsafe. Consider using LZ4_decompress_safe_partial() instead") LZ4LIB_API int LZ4_decompress_fast (const char* src, char* dst, int originalSize); -LZ4_DEPRECATED("This function is deprecated and unsafe. Consider using LZ4_decompress_safe_continue() instead") +LZ4_DEPRECATED("This function is deprecated and unsafe. Consider migrating towards LZ4_decompress_safe_continue() instead. " + "Note that the contract will change (requires block's compressed size, instead of decompressed size)") LZ4LIB_API int LZ4_decompress_fast_continue (LZ4_streamDecode_t* LZ4_streamDecode, const char* src, char* dst, int originalSize); -LZ4_DEPRECATED("This function is deprecated and unsafe. Consider using LZ4_decompress_safe_usingDict() instead") +LZ4_DEPRECATED("This function is deprecated and unsafe. Consider using LZ4_decompress_safe_partial_usingDict() instead") LZ4LIB_API int LZ4_decompress_fast_usingDict (const char* src, char* dst, int originalSize, const char* dictStart, int dictSize); /*! LZ4_resetStream() : diff --git a/contrib/libs/lz4/lz4file.c b/contrib/libs/lz4/lz4file.c index eaf9b1704d..a4197ea80d 100644 --- a/contrib/libs/lz4/lz4file.c +++ b/contrib/libs/lz4/lz4file.c @@ -31,11 +31,21 @@ * - LZ4 homepage : http://www.lz4.org * - LZ4 source repository : https://github.com/lz4/lz4 */ -#include <stdlib.h> +#include <stdlib.h> /* malloc, free */ #include <string.h> +#include <assert.h> #include "lz4.h" #include "lz4file.h" +static LZ4F_errorCode_t returnErrorCode(LZ4F_errorCodes code) +{ + return (LZ4F_errorCode_t)-(ptrdiff_t)code; +} +#undef RETURN_ERROR +#define RETURN_ERROR(e) return returnErrorCode(LZ4F_ERROR_ ## e) + +/* ===== read API ===== */ + struct LZ4_readFile_s { LZ4F_dctx* dctxPtr; FILE* fp; @@ -45,76 +55,80 @@ struct LZ4_readFile_s { size_t srcBufMaxSize; }; -struct LZ4_writeFile_s { - LZ4F_cctx* cctxPtr; - FILE* fp; - LZ4_byte* dstBuf; - size_t maxWriteSize; - size_t dstBufMaxSize; - LZ4F_errorCode_t errCode; -}; +static void LZ4F_freeReadFile(LZ4_readFile_t* lz4fRead) +{ + if (lz4fRead==NULL) return; + LZ4F_freeDecompressionContext(lz4fRead->dctxPtr); + free(lz4fRead->srcBuf); + free(lz4fRead); +} + +static void LZ4F_freeAndNullReadFile(LZ4_readFile_t** statePtr) +{ + assert(statePtr != NULL); + LZ4F_freeReadFile(*statePtr); + *statePtr = NULL; +} LZ4F_errorCode_t LZ4F_readOpen(LZ4_readFile_t** lz4fRead, FILE* fp) { char buf[LZ4F_HEADER_SIZE_MAX]; size_t consumedSize; LZ4F_errorCode_t ret; - LZ4F_frameInfo_t info; if (fp == NULL || lz4fRead == NULL) { - return -LZ4F_ERROR_GENERIC; + RETURN_ERROR(parameter_null); } *lz4fRead = (LZ4_readFile_t*)calloc(1, sizeof(LZ4_readFile_t)); if (*lz4fRead == NULL) { - return -LZ4F_ERROR_allocation_failed; + RETURN_ERROR(allocation_failed); } - ret = LZ4F_createDecompressionContext(&(*lz4fRead)->dctxPtr, LZ4F_getVersion()); + ret = LZ4F_createDecompressionContext(&(*lz4fRead)->dctxPtr, LZ4F_VERSION); if (LZ4F_isError(ret)) { - free(*lz4fRead); + LZ4F_freeAndNullReadFile(lz4fRead); return ret; } (*lz4fRead)->fp = fp; consumedSize = fread(buf, 1, sizeof(buf), (*lz4fRead)->fp); if (consumedSize != sizeof(buf)) { - free(*lz4fRead); - return -LZ4F_ERROR_GENERIC; + LZ4F_freeAndNullReadFile(lz4fRead); + RETURN_ERROR(io_read); } - ret = LZ4F_getFrameInfo((*lz4fRead)->dctxPtr, &info, buf, &consumedSize); - if (LZ4F_isError(ret)) { - LZ4F_freeDecompressionContext((*lz4fRead)->dctxPtr); - free(*lz4fRead); - return ret; + { LZ4F_frameInfo_t info; + LZ4F_errorCode_t const r = LZ4F_getFrameInfo((*lz4fRead)->dctxPtr, &info, buf, &consumedSize); + if (LZ4F_isError(r)) { + LZ4F_freeAndNullReadFile(lz4fRead); + return r; } - switch (info.blockSizeID) { - case LZ4F_default : - case LZ4F_max64KB : - (*lz4fRead)->srcBufMaxSize = 64 * 1024; - break; - case LZ4F_max256KB: - (*lz4fRead)->srcBufMaxSize = 256 * 1024; - break; - case LZ4F_max1MB: - (*lz4fRead)->srcBufMaxSize = 1 * 1024 * 1024; - break; - case LZ4F_max4MB: - (*lz4fRead)->srcBufMaxSize = 4 * 1024 * 1024; - break; - default: - LZ4F_freeDecompressionContext((*lz4fRead)->dctxPtr); - free(*lz4fRead); - return -LZ4F_ERROR_maxBlockSize_invalid; + switch (info.blockSizeID) { + case LZ4F_default : + case LZ4F_max64KB : + (*lz4fRead)->srcBufMaxSize = 64 * 1024; + break; + case LZ4F_max256KB: + (*lz4fRead)->srcBufMaxSize = 256 * 1024; + break; + case LZ4F_max1MB: + (*lz4fRead)->srcBufMaxSize = 1 * 1024 * 1024; + break; + case LZ4F_max4MB: + (*lz4fRead)->srcBufMaxSize = 4 * 1024 * 1024; + break; + default: + LZ4F_freeAndNullReadFile(lz4fRead); + RETURN_ERROR(maxBlockSize_invalid); + } } (*lz4fRead)->srcBuf = (LZ4_byte*)malloc((*lz4fRead)->srcBufMaxSize); if ((*lz4fRead)->srcBuf == NULL) { - LZ4F_freeDecompressionContext((*lz4fRead)->dctxPtr); - free(lz4fRead); - return -LZ4F_ERROR_allocation_failed; + LZ4F_freeAndNullReadFile(lz4fRead); + RETURN_ERROR(allocation_failed); } (*lz4fRead)->srcBufSize = sizeof(buf) - consumedSize; @@ -129,7 +143,7 @@ size_t LZ4F_read(LZ4_readFile_t* lz4fRead, void* buf, size_t size) size_t next = 0; if (lz4fRead == NULL || buf == NULL) - return -LZ4F_ERROR_GENERIC; + RETURN_ERROR(parameter_null); while (next < size) { size_t srcsize = lz4fRead->srcBufSize - lz4fRead->srcBufNext; @@ -142,12 +156,10 @@ size_t LZ4F_read(LZ4_readFile_t* lz4fRead, void* buf, size_t size) lz4fRead->srcBufSize = ret; srcsize = lz4fRead->srcBufSize; lz4fRead->srcBufNext = 0; - } - else if (ret == 0) { + } else if (ret == 0) { break; - } - else { - return -LZ4F_ERROR_GENERIC; + } else { + RETURN_ERROR(io_read); } } @@ -171,24 +183,48 @@ size_t LZ4F_read(LZ4_readFile_t* lz4fRead, void* buf, size_t size) LZ4F_errorCode_t LZ4F_readClose(LZ4_readFile_t* lz4fRead) { if (lz4fRead == NULL) - return -LZ4F_ERROR_GENERIC; - LZ4F_freeDecompressionContext(lz4fRead->dctxPtr); - free(lz4fRead->srcBuf); - free(lz4fRead); + RETURN_ERROR(parameter_null); + LZ4F_freeReadFile(lz4fRead); return LZ4F_OK_NoError; } +/* ===== write API ===== */ + +struct LZ4_writeFile_s { + LZ4F_cctx* cctxPtr; + FILE* fp; + LZ4_byte* dstBuf; + size_t maxWriteSize; + size_t dstBufMaxSize; + LZ4F_errorCode_t errCode; +}; + +static void LZ4F_freeWriteFile(LZ4_writeFile_t* state) +{ + if (state == NULL) return; + LZ4F_freeCompressionContext(state->cctxPtr); + free(state->dstBuf); + free(state); +} + +static void LZ4F_freeAndNullWriteFile(LZ4_writeFile_t** statePtr) +{ + assert(statePtr != NULL); + LZ4F_freeWriteFile(*statePtr); + *statePtr = NULL; +} + LZ4F_errorCode_t LZ4F_writeOpen(LZ4_writeFile_t** lz4fWrite, FILE* fp, const LZ4F_preferences_t* prefsPtr) { LZ4_byte buf[LZ4F_HEADER_SIZE_MAX]; size_t ret; if (fp == NULL || lz4fWrite == NULL) - return -LZ4F_ERROR_GENERIC; + RETURN_ERROR(parameter_null); - *lz4fWrite = (LZ4_writeFile_t*)malloc(sizeof(LZ4_writeFile_t)); + *lz4fWrite = (LZ4_writeFile_t*)calloc(1, sizeof(LZ4_writeFile_t)); if (*lz4fWrite == NULL) { - return -LZ4F_ERROR_allocation_failed; + RETURN_ERROR(allocation_failed); } if (prefsPtr != NULL) { switch (prefsPtr->frameInfo.blockSizeID) { @@ -206,8 +242,8 @@ LZ4F_errorCode_t LZ4F_writeOpen(LZ4_writeFile_t** lz4fWrite, FILE* fp, const LZ4 (*lz4fWrite)->maxWriteSize = 4 * 1024 * 1024; break; default: - free(lz4fWrite); - return -LZ4F_ERROR_maxBlockSize_invalid; + LZ4F_freeAndNullWriteFile(lz4fWrite); + RETURN_ERROR(maxBlockSize_invalid); } } else { (*lz4fWrite)->maxWriteSize = 64 * 1024; @@ -216,30 +252,25 @@ LZ4F_errorCode_t LZ4F_writeOpen(LZ4_writeFile_t** lz4fWrite, FILE* fp, const LZ4 (*lz4fWrite)->dstBufMaxSize = LZ4F_compressBound((*lz4fWrite)->maxWriteSize, prefsPtr); (*lz4fWrite)->dstBuf = (LZ4_byte*)malloc((*lz4fWrite)->dstBufMaxSize); if ((*lz4fWrite)->dstBuf == NULL) { - free(*lz4fWrite); - return -LZ4F_ERROR_allocation_failed; + LZ4F_freeAndNullWriteFile(lz4fWrite); + RETURN_ERROR(allocation_failed); } - ret = LZ4F_createCompressionContext(&(*lz4fWrite)->cctxPtr, LZ4F_getVersion()); + ret = LZ4F_createCompressionContext(&(*lz4fWrite)->cctxPtr, LZ4F_VERSION); if (LZ4F_isError(ret)) { - free((*lz4fWrite)->dstBuf); - free(*lz4fWrite); + LZ4F_freeAndNullWriteFile(lz4fWrite); return ret; } ret = LZ4F_compressBegin((*lz4fWrite)->cctxPtr, buf, LZ4F_HEADER_SIZE_MAX, prefsPtr); if (LZ4F_isError(ret)) { - LZ4F_freeCompressionContext((*lz4fWrite)->cctxPtr); - free((*lz4fWrite)->dstBuf); - free(*lz4fWrite); + LZ4F_freeAndNullWriteFile(lz4fWrite); return ret; } if (ret != fwrite(buf, 1, ret, fp)) { - LZ4F_freeCompressionContext((*lz4fWrite)->cctxPtr); - free((*lz4fWrite)->dstBuf); - free(*lz4fWrite); - return -LZ4F_ERROR_GENERIC; + LZ4F_freeAndNullWriteFile(lz4fWrite); + RETURN_ERROR(io_write); } (*lz4fWrite)->fp = fp; @@ -247,15 +278,15 @@ LZ4F_errorCode_t LZ4F_writeOpen(LZ4_writeFile_t** lz4fWrite, FILE* fp, const LZ4 return LZ4F_OK_NoError; } -size_t LZ4F_write(LZ4_writeFile_t* lz4fWrite, void* buf, size_t size) +size_t LZ4F_write(LZ4_writeFile_t* lz4fWrite, const void* buf, size_t size) { - LZ4_byte* p = (LZ4_byte*)buf; + const LZ4_byte* p = (const LZ4_byte*)buf; size_t remain = size; size_t chunk; size_t ret; if (lz4fWrite == NULL || buf == NULL) - return -LZ4F_ERROR_GENERIC; + RETURN_ERROR(parameter_null); while (remain) { if (remain > lz4fWrite->maxWriteSize) chunk = lz4fWrite->maxWriteSize; @@ -271,9 +302,9 @@ size_t LZ4F_write(LZ4_writeFile_t* lz4fWrite, void* buf, size_t size) return ret; } - if(ret != fwrite(lz4fWrite->dstBuf, 1, ret, lz4fWrite->fp)) { - lz4fWrite->errCode = -LZ4F_ERROR_GENERIC; - return -LZ4F_ERROR_GENERIC; + if (ret != fwrite(lz4fWrite->dstBuf, 1, ret, lz4fWrite->fp)) { + lz4fWrite->errCode = returnErrorCode(LZ4F_ERROR_io_write); + RETURN_ERROR(io_write); } p += chunk; @@ -287,8 +318,9 @@ LZ4F_errorCode_t LZ4F_writeClose(LZ4_writeFile_t* lz4fWrite) { LZ4F_errorCode_t ret = LZ4F_OK_NoError; - if (lz4fWrite == NULL) - return -LZ4F_ERROR_GENERIC; + if (lz4fWrite == NULL) { + RETURN_ERROR(parameter_null); + } if (lz4fWrite->errCode == LZ4F_OK_NoError) { ret = LZ4F_compressEnd(lz4fWrite->cctxPtr, @@ -299,13 +331,11 @@ LZ4F_errorCode_t LZ4F_writeClose(LZ4_writeFile_t* lz4fWrite) } if (ret != fwrite(lz4fWrite->dstBuf, 1, ret, lz4fWrite->fp)) { - ret = -LZ4F_ERROR_GENERIC; + ret = returnErrorCode(LZ4F_ERROR_io_write); } } out: - LZ4F_freeCompressionContext(lz4fWrite->cctxPtr); - free(lz4fWrite->dstBuf); - free(lz4fWrite); + LZ4F_freeWriteFile(lz4fWrite); return ret; } diff --git a/contrib/libs/lz4/lz4file.h b/contrib/libs/lz4/lz4file.h index 5527130720..598ad705ea 100644 --- a/contrib/libs/lz4/lz4file.h +++ b/contrib/libs/lz4/lz4file.h @@ -38,7 +38,7 @@ extern "C" { #ifndef LZ4FILE_H #define LZ4FILE_H -#include <stdio.h> +#include <stdio.h> /* FILE* */ #include "lz4frame_static.h" typedef struct LZ4_readFile_s LZ4_readFile_t; @@ -78,7 +78,7 @@ LZ4FLIB_STATIC_API LZ4F_errorCode_t LZ4F_writeOpen(LZ4_writeFile_t** lz4fWrite, * `buf` write data buffer. * `size` write data buffer size. */ -LZ4FLIB_STATIC_API size_t LZ4F_write(LZ4_writeFile_t* lz4fWrite, void* buf, size_t size); +LZ4FLIB_STATIC_API size_t LZ4F_write(LZ4_writeFile_t* lz4fWrite, const void* buf, size_t size); /*! LZ4F_writeClose() : * Close lz4file handle. diff --git a/contrib/libs/lz4/lz4frame.c b/contrib/libs/lz4/lz4frame.c index 174f9ae4f2..f89c055799 100644 --- a/contrib/libs/lz4/lz4frame.c +++ b/contrib/libs/lz4/lz4frame.c @@ -44,6 +44,7 @@ /*-************************************ * Compiler Options **************************************/ +#include <limits.h> #ifdef _MSC_VER /* Visual Studio */ # pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */ #endif @@ -54,8 +55,8 @@ **************************************/ /* * LZ4F_HEAPMODE : - * Select how default compression functions will allocate memory for their hash table, - * in memory stack (0:default, fastest), or in memory heap (1:requires malloc()). + * Control how LZ4F_compressFrame allocates the Compression State, + * either on stack (0:default, fastest), or in memory heap (1:requires malloc()). */ #ifndef LZ4F_HEAPMODE # define LZ4F_HEAPMODE 0 @@ -125,8 +126,9 @@ static void* LZ4F_malloc(size_t s, LZ4F_CustomMem cmem) static void LZ4F_free(void* p, LZ4F_CustomMem cmem) { - /* custom malloc defined : use it */ + if (p == NULL) return; if (cmem.customFree != NULL) { + /* custom allocation defined : use it */ cmem.customFree(cmem.opaqueState, p); return; } @@ -153,7 +155,7 @@ static void LZ4F_free(void* p, LZ4F_CustomMem cmem) static int g_debuglog_enable = 1; # define DEBUGLOG(l, ...) { \ if ((g_debuglog_enable) && (l<=LZ4_DEBUG)) { \ - fprintf(stderr, __FILE__ ": "); \ + fprintf(stderr, __FILE__ " (%i): ", __LINE__ ); \ fprintf(stderr, __VA_ARGS__); \ fprintf(stderr, " \n"); \ } } @@ -186,9 +188,9 @@ static U32 LZ4F_readLE32 (const void* src) { const BYTE* const srcPtr = (const BYTE*)src; U32 value32 = srcPtr[0]; - value32 += ((U32)srcPtr[1])<< 8; - value32 += ((U32)srcPtr[2])<<16; - value32 += ((U32)srcPtr[3])<<24; + value32 |= ((U32)srcPtr[1])<< 8; + value32 |= ((U32)srcPtr[2])<<16; + value32 |= ((U32)srcPtr[3])<<24; return value32; } @@ -205,13 +207,13 @@ static U64 LZ4F_readLE64 (const void* src) { const BYTE* const srcPtr = (const BYTE*)src; U64 value64 = srcPtr[0]; - value64 += ((U64)srcPtr[1]<<8); - value64 += ((U64)srcPtr[2]<<16); - value64 += ((U64)srcPtr[3]<<24); - value64 += ((U64)srcPtr[4]<<32); - value64 += ((U64)srcPtr[5]<<40); - value64 += ((U64)srcPtr[6]<<48); - value64 += ((U64)srcPtr[7]<<56); + value64 |= ((U64)srcPtr[1]<<8); + value64 |= ((U64)srcPtr[2]<<16); + value64 |= ((U64)srcPtr[3]<<24); + value64 |= ((U64)srcPtr[4]<<32); + value64 |= ((U64)srcPtr[5]<<40); + value64 |= ((U64)srcPtr[6]<<48); + value64 |= ((U64)srcPtr[7]<<56); return value64; } @@ -257,14 +259,15 @@ static const size_t BFSize = LZ4F_BLOCK_CHECKSUM_SIZE; /* block footer : checks * Structures and local types **************************************/ -typedef enum { LZ4B_COMPRESSED, LZ4B_UNCOMPRESSED} LZ4F_blockCompression_t; +typedef enum { LZ4B_COMPRESSED, LZ4B_UNCOMPRESSED} LZ4F_BlockCompressMode_e; +typedef enum { ctxNone, ctxFast, ctxHC } LZ4F_CtxType_e; typedef struct LZ4F_cctx_s { LZ4F_CustomMem cmem; LZ4F_preferences_t prefs; U32 version; - U32 cStage; + U32 cStage; /* 0 : compression uninitialized ; 1 : initialized, can compress */ const LZ4F_CDict* cdict; size_t maxBlockSize; size_t maxBufferSize; @@ -275,8 +278,8 @@ typedef struct LZ4F_cctx_s XXH32_state_t xxh; void* lz4CtxPtr; U16 lz4CtxAlloc; /* sized for: 0 = none, 1 = lz4 ctx, 2 = lz4hc ctx */ - U16 lz4CtxState; /* in use as: 0 = none, 1 = lz4 ctx, 2 = lz4hc ctx */ - LZ4F_blockCompression_t blockCompression; + U16 lz4CtxType; /* in use as: 0 = none, 1 = lz4 ctx, 2 = lz4hc ctx */ + LZ4F_BlockCompressMode_e blockCompressMode; } LZ4F_cctx_t; @@ -314,9 +317,14 @@ static LZ4F_errorCode_t LZ4F_returnErrorCode(LZ4F_errorCodes code) #define RETURN_ERROR(e) return LZ4F_returnErrorCode(LZ4F_ERROR_ ## e) -#define RETURN_ERROR_IF(c,e) if (c) RETURN_ERROR(e) +#define RETURN_ERROR_IF(c,e) do { \ + if (c) { \ + DEBUGLOG(3, "Error: " #c); \ + RETURN_ERROR(e); \ + } \ + } while (0) -#define FORWARD_IF_ERROR(r) if (LZ4F_isError(r)) return (r) +#define FORWARD_IF_ERROR(r) do { if (LZ4F_isError(r)) return (r); } while (0) unsigned LZ4F_getVersion(void) { return LZ4F_VERSION; } @@ -429,6 +437,7 @@ size_t LZ4F_compressFrame_usingCDict(LZ4F_cctx* cctx, BYTE* dstPtr = dstStart; BYTE* const dstEnd = dstStart + dstCapacity; + DEBUGLOG(4, "LZ4F_compressFrame_usingCDict (srcSize=%u)", (unsigned)srcSize); if (preferencesPtr!=NULL) prefs = *preferencesPtr; else @@ -494,7 +503,7 @@ size_t LZ4F_compressFrame(void* dstBuffer, size_t dstCapacity, LZ4_initStream(&lz4ctx, sizeof(lz4ctx)); cctxPtr->lz4CtxPtr = &lz4ctx; cctxPtr->lz4CtxAlloc = 1; - cctxPtr->lz4CtxState = 1; + cctxPtr->lz4CtxType = ctxFast; } #endif DEBUGLOG(4, "LZ4F_compressFrame"); @@ -539,18 +548,19 @@ LZ4F_createCDict_advanced(LZ4F_CustomMem cmem, const void* dictBuffer, size_t di dictSize = 64 KB; } cdict->dictContent = LZ4F_malloc(dictSize, cmem); + /* note: using @cmem to allocate => can't use default create */ cdict->fastCtx = (LZ4_stream_t*)LZ4F_malloc(sizeof(LZ4_stream_t), cmem); - if (cdict->fastCtx) - LZ4_initStream(cdict->fastCtx, sizeof(LZ4_stream_t)); cdict->HCCtx = (LZ4_streamHC_t*)LZ4F_malloc(sizeof(LZ4_streamHC_t), cmem); - if (cdict->HCCtx) - LZ4_initStream(cdict->HCCtx, sizeof(LZ4_streamHC_t)); if (!cdict->dictContent || !cdict->fastCtx || !cdict->HCCtx) { LZ4F_freeCDict(cdict); return NULL; } memcpy(cdict->dictContent, dictStart, dictSize); - LZ4_loadDict (cdict->fastCtx, (const char*)cdict->dictContent, (int)dictSize); + LZ4_initStream(cdict->fastCtx, sizeof(LZ4_stream_t)); + LZ4_loadDictSlow(cdict->fastCtx, (const char*)cdict->dictContent, (int)dictSize); + LZ4_initStreamHC(cdict->HCCtx, sizeof(LZ4_streamHC_t)); + /* note: we don't know at this point which compression level is going to be used + * as a consequence, HCCtx is created for the more common HC mode */ LZ4_setCompressionLevel(cdict->HCCtx, LZ4HC_CLEVEL_DEFAULT); LZ4_loadDictHC(cdict->HCCtx, (const char*)cdict->dictContent, (int)dictSize); return cdict; @@ -616,7 +626,6 @@ LZ4F_createCompressionContext(LZ4F_cctx** LZ4F_compressionContextPtr, unsigned v return LZ4F_OK_NoError; } - LZ4F_errorCode_t LZ4F_freeCompressionContext(LZ4F_cctx* cctxPtr) { if (cctxPtr != NULL) { /* support free on NULL */ @@ -641,7 +650,7 @@ static void LZ4F_initStream(void* ctx, int level, LZ4F_blockMode_t blockMode) { if (level < LZ4HC_CLEVEL_MIN) { - if (cdict != NULL || blockMode == LZ4F_blockLinked) { + if (cdict || blockMode == LZ4F_blockLinked) { /* In these cases, we will call LZ4_compress_fast_continue(), * which needs an already reset context. Otherwise, we'll call a * one-shot API. The non-continued APIs internally perform their own @@ -649,11 +658,18 @@ static void LZ4F_initStream(void* ctx, * tableType they need the context to be in. So in that case this * would be misguided / wasted work. */ LZ4_resetStream_fast((LZ4_stream_t*)ctx); + if (cdict) + LZ4_attach_dictionary((LZ4_stream_t*)ctx, cdict->fastCtx); } - LZ4_attach_dictionary((LZ4_stream_t *)ctx, cdict ? cdict->fastCtx : NULL); + /* In these cases, we'll call a one-shot API. + * The non-continued APIs internally perform their own resets + * at the beginning of their calls, where they know + * which tableType they need the context to be in. + * Therefore, a reset here would be wasted work. */ } else { LZ4_resetStreamHC_fast((LZ4_streamHC_t*)ctx, level); - LZ4_attach_HC_dictionary((LZ4_streamHC_t *)ctx, cdict ? cdict->HCCtx : NULL); + if (cdict) + LZ4_attach_HC_dictionary((LZ4_streamHC_t*)ctx, cdict->HCCtx); } } @@ -668,14 +684,12 @@ static int ctxTypeID_to_size(int ctxTypeID) { } } -/*! LZ4F_compressBegin_usingCDict() : - * init streaming compression AND writes frame header into @dstBuffer. - * @dstCapacity must be >= LZ4F_HEADER_SIZE_MAX bytes. - * @return : number of bytes written into @dstBuffer for the header - * or an error code (can be tested using LZ4F_isError()) +/* LZ4F_compressBegin_internal() + * Note: only accepts @cdict _or_ @dictBuffer as non NULL. */ -size_t LZ4F_compressBegin_usingCDict(LZ4F_cctx* cctxPtr, +size_t LZ4F_compressBegin_internal(LZ4F_cctx* cctx, void* dstBuffer, size_t dstCapacity, + const void* dictBuffer, size_t dictSize, const LZ4F_CDict* cdict, const LZ4F_preferences_t* preferencesPtr) { @@ -685,71 +699,85 @@ size_t LZ4F_compressBegin_usingCDict(LZ4F_cctx* cctxPtr, RETURN_ERROR_IF(dstCapacity < maxFHSize, dstMaxSize_tooSmall); if (preferencesPtr == NULL) preferencesPtr = &prefNull; - cctxPtr->prefs = *preferencesPtr; + cctx->prefs = *preferencesPtr; /* cctx Management */ - { U16 const ctxTypeID = (cctxPtr->prefs.compressionLevel < LZ4HC_CLEVEL_MIN) ? 1 : 2; + { U16 const ctxTypeID = (cctx->prefs.compressionLevel < LZ4HC_CLEVEL_MIN) ? 1 : 2; int requiredSize = ctxTypeID_to_size(ctxTypeID); - int allocatedSize = ctxTypeID_to_size(cctxPtr->lz4CtxAlloc); + int allocatedSize = ctxTypeID_to_size(cctx->lz4CtxAlloc); if (allocatedSize < requiredSize) { /* not enough space allocated */ - LZ4F_free(cctxPtr->lz4CtxPtr, cctxPtr->cmem); - if (cctxPtr->prefs.compressionLevel < LZ4HC_CLEVEL_MIN) { + LZ4F_free(cctx->lz4CtxPtr, cctx->cmem); + if (cctx->prefs.compressionLevel < LZ4HC_CLEVEL_MIN) { /* must take ownership of memory allocation, * in order to respect custom allocator contract */ - cctxPtr->lz4CtxPtr = LZ4F_malloc(sizeof(LZ4_stream_t), cctxPtr->cmem); - if (cctxPtr->lz4CtxPtr) - LZ4_initStream(cctxPtr->lz4CtxPtr, sizeof(LZ4_stream_t)); + cctx->lz4CtxPtr = LZ4F_malloc(sizeof(LZ4_stream_t), cctx->cmem); + if (cctx->lz4CtxPtr) + LZ4_initStream(cctx->lz4CtxPtr, sizeof(LZ4_stream_t)); } else { - cctxPtr->lz4CtxPtr = LZ4F_malloc(sizeof(LZ4_streamHC_t), cctxPtr->cmem); - if (cctxPtr->lz4CtxPtr) - LZ4_initStreamHC(cctxPtr->lz4CtxPtr, sizeof(LZ4_streamHC_t)); + cctx->lz4CtxPtr = LZ4F_malloc(sizeof(LZ4_streamHC_t), cctx->cmem); + if (cctx->lz4CtxPtr) + LZ4_initStreamHC(cctx->lz4CtxPtr, sizeof(LZ4_streamHC_t)); } - RETURN_ERROR_IF(cctxPtr->lz4CtxPtr == NULL, allocation_failed); - cctxPtr->lz4CtxAlloc = ctxTypeID; - cctxPtr->lz4CtxState = ctxTypeID; - } else if (cctxPtr->lz4CtxState != ctxTypeID) { + RETURN_ERROR_IF(cctx->lz4CtxPtr == NULL, allocation_failed); + cctx->lz4CtxAlloc = ctxTypeID; + cctx->lz4CtxType = ctxTypeID; + } else if (cctx->lz4CtxType != ctxTypeID) { /* otherwise, a sufficient buffer is already allocated, * but we need to reset it to the correct context type */ - if (cctxPtr->prefs.compressionLevel < LZ4HC_CLEVEL_MIN) { - LZ4_initStream((LZ4_stream_t*)cctxPtr->lz4CtxPtr, sizeof(LZ4_stream_t)); + if (cctx->prefs.compressionLevel < LZ4HC_CLEVEL_MIN) { + LZ4_initStream((LZ4_stream_t*)cctx->lz4CtxPtr, sizeof(LZ4_stream_t)); } else { - LZ4_initStreamHC((LZ4_streamHC_t*)cctxPtr->lz4CtxPtr, sizeof(LZ4_streamHC_t)); - LZ4_setCompressionLevel((LZ4_streamHC_t*)cctxPtr->lz4CtxPtr, cctxPtr->prefs.compressionLevel); + LZ4_initStreamHC((LZ4_streamHC_t*)cctx->lz4CtxPtr, sizeof(LZ4_streamHC_t)); + LZ4_setCompressionLevel((LZ4_streamHC_t*)cctx->lz4CtxPtr, cctx->prefs.compressionLevel); } - cctxPtr->lz4CtxState = ctxTypeID; + cctx->lz4CtxType = ctxTypeID; } } /* Buffer Management */ - if (cctxPtr->prefs.frameInfo.blockSizeID == 0) - cctxPtr->prefs.frameInfo.blockSizeID = LZ4F_BLOCKSIZEID_DEFAULT; - cctxPtr->maxBlockSize = LZ4F_getBlockSize(cctxPtr->prefs.frameInfo.blockSizeID); + if (cctx->prefs.frameInfo.blockSizeID == 0) + cctx->prefs.frameInfo.blockSizeID = LZ4F_BLOCKSIZEID_DEFAULT; + cctx->maxBlockSize = LZ4F_getBlockSize(cctx->prefs.frameInfo.blockSizeID); { size_t const requiredBuffSize = preferencesPtr->autoFlush ? - ((cctxPtr->prefs.frameInfo.blockMode == LZ4F_blockLinked) ? 64 KB : 0) : /* only needs past data up to window size */ - cctxPtr->maxBlockSize + ((cctxPtr->prefs.frameInfo.blockMode == LZ4F_blockLinked) ? 128 KB : 0); - - if (cctxPtr->maxBufferSize < requiredBuffSize) { - cctxPtr->maxBufferSize = 0; - LZ4F_free(cctxPtr->tmpBuff, cctxPtr->cmem); - cctxPtr->tmpBuff = (BYTE*)LZ4F_calloc(requiredBuffSize, cctxPtr->cmem); - RETURN_ERROR_IF(cctxPtr->tmpBuff == NULL, allocation_failed); - cctxPtr->maxBufferSize = requiredBuffSize; + ((cctx->prefs.frameInfo.blockMode == LZ4F_blockLinked) ? 64 KB : 0) : /* only needs past data up to window size */ + cctx->maxBlockSize + ((cctx->prefs.frameInfo.blockMode == LZ4F_blockLinked) ? 128 KB : 0); + + if (cctx->maxBufferSize < requiredBuffSize) { + cctx->maxBufferSize = 0; + LZ4F_free(cctx->tmpBuff, cctx->cmem); + cctx->tmpBuff = (BYTE*)LZ4F_malloc(requiredBuffSize, cctx->cmem); + RETURN_ERROR_IF(cctx->tmpBuff == NULL, allocation_failed); + cctx->maxBufferSize = requiredBuffSize; } } - cctxPtr->tmpIn = cctxPtr->tmpBuff; - cctxPtr->tmpInSize = 0; - (void)XXH32_reset(&(cctxPtr->xxh), 0); + cctx->tmpIn = cctx->tmpBuff; + cctx->tmpInSize = 0; + (void)XXH32_reset(&(cctx->xxh), 0); /* context init */ - cctxPtr->cdict = cdict; - if (cctxPtr->prefs.frameInfo.blockMode == LZ4F_blockLinked) { + cctx->cdict = cdict; + if (cctx->prefs.frameInfo.blockMode == LZ4F_blockLinked) { /* frame init only for blockLinked : blockIndependent will be init at each block */ - LZ4F_initStream(cctxPtr->lz4CtxPtr, cdict, cctxPtr->prefs.compressionLevel, LZ4F_blockLinked); + LZ4F_initStream(cctx->lz4CtxPtr, cdict, cctx->prefs.compressionLevel, LZ4F_blockLinked); } if (preferencesPtr->compressionLevel >= LZ4HC_CLEVEL_MIN) { - LZ4_favorDecompressionSpeed((LZ4_streamHC_t*)cctxPtr->lz4CtxPtr, (int)preferencesPtr->favorDecSpeed); + LZ4_favorDecompressionSpeed((LZ4_streamHC_t*)cctx->lz4CtxPtr, (int)preferencesPtr->favorDecSpeed); + } + if (dictBuffer) { + assert(cdict == NULL); + RETURN_ERROR_IF(dictSize > INT_MAX, parameter_invalid); + if (cctx->lz4CtxType == ctxFast) { + /* lz4 fast*/ + LZ4_loadDict((LZ4_stream_t*)cctx->lz4CtxPtr, (const char*)dictBuffer, (int)dictSize); + } else { + /* lz4hc */ + assert(cctx->lz4CtxType == ctxHC); + LZ4_loadDictHC((LZ4_streamHC_t*)cctx->lz4CtxPtr, (const char*)dictBuffer, (int)dictSize); + } } + /* Stage 2 : Write Frame Header */ + /* Magic Number */ LZ4F_writeLE32(dstPtr, LZ4F_MAGICNUMBER); dstPtr += 4; @@ -757,22 +785,22 @@ size_t LZ4F_compressBegin_usingCDict(LZ4F_cctx* cctxPtr, /* FLG Byte */ *dstPtr++ = (BYTE)(((1 & _2BITS) << 6) /* Version('01') */ - + ((cctxPtr->prefs.frameInfo.blockMode & _1BIT ) << 5) - + ((cctxPtr->prefs.frameInfo.blockChecksumFlag & _1BIT ) << 4) - + ((unsigned)(cctxPtr->prefs.frameInfo.contentSize > 0) << 3) - + ((cctxPtr->prefs.frameInfo.contentChecksumFlag & _1BIT ) << 2) - + (cctxPtr->prefs.frameInfo.dictID > 0) ); + + ((cctx->prefs.frameInfo.blockMode & _1BIT ) << 5) + + ((cctx->prefs.frameInfo.blockChecksumFlag & _1BIT ) << 4) + + ((unsigned)(cctx->prefs.frameInfo.contentSize > 0) << 3) + + ((cctx->prefs.frameInfo.contentChecksumFlag & _1BIT ) << 2) + + (cctx->prefs.frameInfo.dictID > 0) ); /* BD Byte */ - *dstPtr++ = (BYTE)((cctxPtr->prefs.frameInfo.blockSizeID & _3BITS) << 4); + *dstPtr++ = (BYTE)((cctx->prefs.frameInfo.blockSizeID & _3BITS) << 4); /* Optional Frame content size field */ - if (cctxPtr->prefs.frameInfo.contentSize) { - LZ4F_writeLE64(dstPtr, cctxPtr->prefs.frameInfo.contentSize); + if (cctx->prefs.frameInfo.contentSize) { + LZ4F_writeLE64(dstPtr, cctx->prefs.frameInfo.contentSize); dstPtr += 8; - cctxPtr->totalInSize = 0; + cctx->totalInSize = 0; } /* Optional dictionary ID field */ - if (cctxPtr->prefs.frameInfo.dictID) { - LZ4F_writeLE32(dstPtr, cctxPtr->prefs.frameInfo.dictID); + if (cctx->prefs.frameInfo.dictID) { + LZ4F_writeLE32(dstPtr, cctx->prefs.frameInfo.dictID); dstPtr += 4; } /* Header CRC Byte */ @@ -780,24 +808,54 @@ size_t LZ4F_compressBegin_usingCDict(LZ4F_cctx* cctxPtr, dstPtr++; } - cctxPtr->cStage = 1; /* header written, now request input data block */ + cctx->cStage = 1; /* header written, now request input data block */ return (size_t)(dstPtr - dstStart); } +size_t LZ4F_compressBegin(LZ4F_cctx* cctx, + void* dstBuffer, size_t dstCapacity, + const LZ4F_preferences_t* preferencesPtr) +{ + return LZ4F_compressBegin_internal(cctx, dstBuffer, dstCapacity, + NULL, 0, + NULL, preferencesPtr); +} -/*! LZ4F_compressBegin() : - * init streaming compression AND writes frame header into @dstBuffer. - * @dstCapacity must be >= LZ4F_HEADER_SIZE_MAX bytes. - * @preferencesPtr can be NULL, in which case default parameters are selected. - * @return : number of bytes written into dstBuffer for the header - * or an error code (can be tested using LZ4F_isError()) - */ -size_t LZ4F_compressBegin(LZ4F_cctx* cctxPtr, +/* LZ4F_compressBegin_usingDictOnce: + * Hidden implementation, + * employed for multi-threaded compression + * when frame defines linked blocks */ +size_t LZ4F_compressBegin_usingDictOnce(LZ4F_cctx* cctx, + void* dstBuffer, size_t dstCapacity, + const void* dict, size_t dictSize, + const LZ4F_preferences_t* preferencesPtr) +{ + return LZ4F_compressBegin_internal(cctx, dstBuffer, dstCapacity, + dict, dictSize, + NULL, preferencesPtr); +} + +size_t LZ4F_compressBegin_usingDict(LZ4F_cctx* cctx, + void* dstBuffer, size_t dstCapacity, + const void* dict, size_t dictSize, + const LZ4F_preferences_t* preferencesPtr) +{ + /* note : incorrect implementation : + * this will only use the dictionary once, + * instead of once *per* block when frames defines independent blocks */ + return LZ4F_compressBegin_usingDictOnce(cctx, dstBuffer, dstCapacity, + dict, dictSize, + preferencesPtr); +} + +size_t LZ4F_compressBegin_usingCDict(LZ4F_cctx* cctx, void* dstBuffer, size_t dstCapacity, + const LZ4F_CDict* cdict, const LZ4F_preferences_t* preferencesPtr) { - return LZ4F_compressBegin_usingCDict(cctxPtr, dstBuffer, dstCapacity, - NULL, preferencesPtr); + return LZ4F_compressBegin_internal(cctx, dstBuffer, dstCapacity, + NULL, 0, + cdict, preferencesPtr); } @@ -891,9 +949,10 @@ static int LZ4F_doNotCompressBlock(void* ctx, const char* src, char* dst, int sr return 0; } -static compressFunc_t LZ4F_selectCompression(LZ4F_blockMode_t blockMode, int level, LZ4F_blockCompression_t compressMode) +static compressFunc_t LZ4F_selectCompression(LZ4F_blockMode_t blockMode, int level, LZ4F_BlockCompressMode_e compressMode) { - if (compressMode == LZ4B_UNCOMPRESSED) return LZ4F_doNotCompressBlock; + if (compressMode == LZ4B_UNCOMPRESSED) + return LZ4F_doNotCompressBlock; if (level < LZ4HC_CLEVEL_MIN) { if (blockMode == LZ4F_blockIndependent) return LZ4F_compressBlock; return LZ4F_compressBlock_continue; @@ -931,7 +990,7 @@ static size_t LZ4F_compressUpdateImpl(LZ4F_cctx* cctxPtr, void* dstBuffer, size_t dstCapacity, const void* srcBuffer, size_t srcSize, const LZ4F_compressOptions_t* compressOptionsPtr, - LZ4F_blockCompression_t blockCompression) + LZ4F_BlockCompressMode_e blockCompression) { size_t const blockSize = cctxPtr->maxBlockSize; const BYTE* srcPtr = (const BYTE*)srcBuffer; @@ -951,10 +1010,10 @@ static size_t LZ4F_compressUpdateImpl(LZ4F_cctx* cctxPtr, RETURN_ERROR(dstMaxSize_tooSmall); /* flush currently written block, to continue with new block compression */ - if (cctxPtr->blockCompression != blockCompression) { + if (cctxPtr->blockCompressMode != blockCompression) { bytesWritten = LZ4F_flush(cctxPtr, dstBuffer, dstCapacity, compressOptionsPtr); dstPtr += bytesWritten; - cctxPtr->blockCompression = blockCompression; + cctxPtr->blockCompressMode = blockCompression; } if (compressOptionsPtr == NULL) compressOptionsPtr = &k_cOptionsNull; @@ -1068,13 +1127,9 @@ size_t LZ4F_compressUpdate(LZ4F_cctx* cctxPtr, compressOptionsPtr, LZ4B_COMPRESSED); } -/*! LZ4F_compressUpdate() : - * LZ4F_compressUpdate() can be called repetitively to compress as much data as necessary. - * When successful, the function always entirely consumes @srcBuffer. - * src data is either buffered or compressed into @dstBuffer. - * If previously an uncompressed block was written, buffered data is flushed - * before appending compressed data is continued. - * This is only supported when LZ4F_blockIndependent is used +/*! LZ4F_uncompressedUpdate() : + * Same as LZ4F_compressUpdate(), but requests blocks to be sent uncompressed. + * This symbol is only supported when LZ4F_blockIndependent is used * @dstCapacity MUST be >= LZ4F_compressBound(srcSize, preferencesPtr). * @compressOptionsPtr is optional : provide NULL to mean "default". * @return : the number of bytes written into dstBuffer. It can be zero, meaning input data was just buffered. @@ -1084,8 +1139,8 @@ size_t LZ4F_compressUpdate(LZ4F_cctx* cctxPtr, size_t LZ4F_uncompressedUpdate(LZ4F_cctx* cctxPtr, void* dstBuffer, size_t dstCapacity, const void* srcBuffer, size_t srcSize, - const LZ4F_compressOptions_t* compressOptionsPtr) { - RETURN_ERROR_IF(cctxPtr->prefs.frameInfo.blockMode != LZ4F_blockIndependent, blockMode_invalid); + const LZ4F_compressOptions_t* compressOptionsPtr) +{ return LZ4F_compressUpdateImpl(cctxPtr, dstBuffer, dstCapacity, srcBuffer, srcSize, @@ -1115,7 +1170,7 @@ size_t LZ4F_flush(LZ4F_cctx* cctxPtr, (void)compressOptionsPtr; /* not useful (yet) */ /* select compression function */ - compress = LZ4F_selectCompression(cctxPtr->prefs.frameInfo.blockMode, cctxPtr->prefs.compressionLevel, cctxPtr->blockCompression); + compress = LZ4F_selectCompression(cctxPtr->prefs.frameInfo.blockMode, cctxPtr->prefs.compressionLevel, cctxPtr->blockCompressMode); /* compress tmp buffer */ dstPtr += LZ4F_makeBlock(dstPtr, @@ -1170,13 +1225,12 @@ size_t LZ4F_compressEnd(LZ4F_cctx* cctxPtr, if (cctxPtr->prefs.frameInfo.contentChecksumFlag == LZ4F_contentChecksumEnabled) { U32 const xxh = XXH32_digest(&(cctxPtr->xxh)); RETURN_ERROR_IF(dstCapacity < 8, dstMaxSize_tooSmall); - DEBUGLOG(5,"Writing 32-bit content checksum"); + DEBUGLOG(5,"Writing 32-bit content checksum (0x%0X)", xxh); LZ4F_writeLE32(dstPtr, xxh); dstPtr+=4; /* content Checksum */ } cctxPtr->cStage = 0; /* state is now re-usable (with identical preferences) */ - cctxPtr->maxBufferSize = 0; /* reuse HC context */ if (cctxPtr->prefs.frameInfo.contentSize) { if (cctxPtr->prefs.frameInfo.contentSize != cctxPtr->totalInSize) @@ -1270,13 +1324,14 @@ LZ4F_errorCode_t LZ4F_freeDecompressionContext(LZ4F_dctx* dctx) /*==--- Streaming Decompression operations ---==*/ - void LZ4F_resetDecompressionContext(LZ4F_dctx* dctx) { + DEBUGLOG(5, "LZ4F_resetDecompressionContext"); dctx->dStage = dstage_getFrameHeader; dctx->dict = NULL; dctx->dictSize = 0; dctx->skipChecksum = 0; + dctx->frameRemainingSize = 0; } @@ -1333,6 +1388,7 @@ static size_t LZ4F_decodeHeader(LZ4F_dctx* dctx, const void* src, size_t srcSize if (((FLG>>1)&_1BIT) != 0) RETURN_ERROR(reservedFlag_set); /* Reserved bit */ if (version != 1) RETURN_ERROR(headerVersion_wrong); /* Version Number, only supported value */ } + DEBUGLOG(6, "contentSizeFlag: %u", contentSizeFlag); /* Frame Header Size */ frameHeaderSize = minFHSize + (contentSizeFlag?8:0) + (dictIDFlag?4:0); @@ -1369,8 +1425,9 @@ static size_t LZ4F_decodeHeader(LZ4F_dctx* dctx, const void* src, size_t srcSize dctx->frameInfo.contentChecksumFlag = (LZ4F_contentChecksum_t)contentChecksumFlag; dctx->frameInfo.blockSizeID = (LZ4F_blockSizeID_t)blockSizeID; dctx->maxBlockSize = LZ4F_getBlockSize((LZ4F_blockSizeID_t)blockSizeID); - if (contentSizeFlag) + if (contentSizeFlag) { dctx->frameRemainingSize = dctx->frameInfo.contentSize = LZ4F_readLE64(srcPtr+6); + } if (dictIDFlag) dctx->frameInfo.dictID = LZ4F_readLE32(srcPtr + frameHeaderSize - 5); @@ -1570,7 +1627,7 @@ size_t LZ4F_decompress(LZ4F_dctx* dctx, size_t nextSrcSizeHint = 1; - DEBUGLOG(5, "LZ4F_decompress : %p,%u => %p,%u", + DEBUGLOG(5, "LZ4F_decompress: src[%p](%u) => dst[%p](%u)", srcBuffer, (unsigned)*srcSizePtr, dstBuffer, (unsigned)*dstSizePtr); if (dstBuffer == NULL) assert(*dstSizePtr == 0); MEM_INIT(&optionsNull, 0, sizeof(optionsNull)); @@ -1722,10 +1779,10 @@ size_t LZ4F_decompress(LZ4F_dctx* dctx, /* history management (linked blocks only)*/ if (dctx->frameInfo.blockMode == LZ4F_blockLinked) { LZ4F_updateDict(dctx, dstPtr, sizeToCopy, dstStart, 0); - } } - - srcPtr += sizeToCopy; - dstPtr += sizeToCopy; + } + srcPtr += sizeToCopy; + dstPtr += sizeToCopy; + } if (sizeToCopy == dctx->tmpInTarget) { /* all done */ if (dctx->frameInfo.blockChecksumFlag) { dctx->tmpInSize = 0; @@ -1959,6 +2016,7 @@ size_t LZ4F_decompress(LZ4F_dctx* dctx, if (!dctx->skipChecksum) { U32 const readCRC = LZ4F_readLE32(selectedIn); U32 const resultCRC = XXH32_digest(&(dctx->xxh)); + DEBUGLOG(4, "frame checksum: stored 0x%0X vs 0x%0X processed", readCRC, resultCRC); #ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION RETURN_ERROR_IF(readCRC != resultCRC, contentChecksum_invalid); #else diff --git a/contrib/libs/lz4/lz4frame.h b/contrib/libs/lz4/lz4frame.h index 1bdf6c4fcb..b8ae322767 100644 --- a/contrib/libs/lz4/lz4frame.h +++ b/contrib/libs/lz4/lz4frame.h @@ -173,16 +173,16 @@ typedef LZ4F_contentChecksum_t contentChecksum_t; * setting all parameters to default. * It's then possible to update selectively some parameters */ typedef struct { - LZ4F_blockSizeID_t blockSizeID; /* max64KB, max256KB, max1MB, max4MB; 0 == default */ - LZ4F_blockMode_t blockMode; /* LZ4F_blockLinked, LZ4F_blockIndependent; 0 == default */ - LZ4F_contentChecksum_t contentChecksumFlag; /* 1: frame terminated with 32-bit checksum of decompressed data; 0: disabled (default) */ + LZ4F_blockSizeID_t blockSizeID; /* max64KB, max256KB, max1MB, max4MB; 0 == default (LZ4F_max64KB) */ + LZ4F_blockMode_t blockMode; /* LZ4F_blockLinked, LZ4F_blockIndependent; 0 == default (LZ4F_blockLinked) */ + LZ4F_contentChecksum_t contentChecksumFlag; /* 1: add a 32-bit checksum of frame's decompressed data; 0 == default (disabled) */ LZ4F_frameType_t frameType; /* read-only field : LZ4F_frame or LZ4F_skippableFrame */ unsigned long long contentSize; /* Size of uncompressed content ; 0 == unknown */ unsigned dictID; /* Dictionary ID, sent by compressor to help decoder select correct dictionary; 0 == no dictID provided */ - LZ4F_blockChecksum_t blockChecksumFlag; /* 1: each block followed by a checksum of block's compressed data; 0: disabled (default) */ + LZ4F_blockChecksum_t blockChecksumFlag; /* 1: each block followed by a checksum of block's compressed data; 0 == default (disabled) */ } LZ4F_frameInfo_t; -#define LZ4F_INIT_FRAMEINFO { LZ4F_default, LZ4F_blockLinked, LZ4F_noContentChecksum, LZ4F_frame, 0ULL, 0U, LZ4F_noBlockChecksum } /* v1.8.3+ */ +#define LZ4F_INIT_FRAMEINFO { LZ4F_max64KB, LZ4F_blockLinked, LZ4F_noContentChecksum, LZ4F_frame, 0ULL, 0U, LZ4F_noBlockChecksum } /* v1.8.3+ */ /*! LZ4F_preferences_t : * makes it possible to supply advanced compression instructions to streaming interface. @@ -204,7 +204,26 @@ typedef struct { * Simple compression function ***********************************/ -LZ4FLIB_API int LZ4F_compressionLevel_max(void); /* v1.8.0+ */ +/*! LZ4F_compressFrame() : + * Compress srcBuffer content into an LZ4-compressed frame. + * It's a one shot operation, all input content is consumed, and all output is generated. + * + * Note : it's a stateless operation (no LZ4F_cctx state needed). + * In order to reduce load on the allocator, LZ4F_compressFrame(), by default, + * uses the stack to allocate space for the compression state and some table. + * If this usage of the stack is too much for your application, + * consider compiling `lz4frame.c` with compile-time macro LZ4F_HEAPMODE set to 1 instead. + * All state allocations will use the Heap. + * It also means each invocation of LZ4F_compressFrame() will trigger several internal alloc/free invocations. + * + * @dstCapacity MUST be >= LZ4F_compressFrameBound(srcSize, preferencesPtr). + * @preferencesPtr is optional : one can provide NULL, in which case all preferences are set to default. + * @return : number of bytes written into dstBuffer. + * or an error code if it fails (can be tested using LZ4F_isError()) + */ +LZ4FLIB_API size_t LZ4F_compressFrame(void* dstBuffer, size_t dstCapacity, + const void* srcBuffer, size_t srcSize, + const LZ4F_preferences_t* preferencesPtr); /*! LZ4F_compressFrameBound() : * Returns the maximum possible compressed size with LZ4F_compressFrame() given srcSize and preferences. @@ -214,16 +233,11 @@ LZ4FLIB_API int LZ4F_compressionLevel_max(void); /* v1.8.0+ */ */ LZ4FLIB_API size_t LZ4F_compressFrameBound(size_t srcSize, const LZ4F_preferences_t* preferencesPtr); -/*! LZ4F_compressFrame() : - * Compress an entire srcBuffer into a valid LZ4 frame. - * dstCapacity MUST be >= LZ4F_compressFrameBound(srcSize, preferencesPtr). - * The LZ4F_preferences_t structure is optional : you can provide NULL as argument. All preferences will be set to default. - * @return : number of bytes written into dstBuffer. - * or an error code if it fails (can be tested using LZ4F_isError()) + +/*! LZ4F_compressionLevel_max() : + * @return maximum allowed compression level (currently: 12) */ -LZ4FLIB_API size_t LZ4F_compressFrame(void* dstBuffer, size_t dstCapacity, - const void* srcBuffer, size_t srcSize, - const LZ4F_preferences_t* preferencesPtr); +LZ4FLIB_API int LZ4F_compressionLevel_max(void); /* v1.8.0+ */ /*-*********************************** @@ -278,7 +292,7 @@ LZ4FLIB_API LZ4F_errorCode_t LZ4F_freeCompressionContext(LZ4F_cctx* cctx); /*! LZ4F_compressBegin() : * will write the frame header into dstBuffer. * dstCapacity must be >= LZ4F_HEADER_SIZE_MAX bytes. - * `prefsPtr` is optional : you can provide NULL as argument, all preferences will then be set to default. + * `prefsPtr` is optional : NULL can be provided to set all preferences to default. * @return : number of bytes written into dstBuffer for the header * or an error code (which can be tested using LZ4F_isError()) */ @@ -355,8 +369,9 @@ typedef struct LZ4F_dctx_s LZ4F_dctx; /* incomplete type */ typedef LZ4F_dctx* LZ4F_decompressionContext_t; /* compatibility with previous API versions */ typedef struct { - unsigned stableDst; /* pledges that last 64KB decompressed data will remain available unmodified between invocations. - * This optimization skips storage operations in tmp buffers. */ + unsigned stableDst; /* pledges that last 64KB decompressed data is present right before @dstBuffer pointer. + * This optimization skips internal storage operations. + * Once set, this pledge must remain valid up to the end of current frame. */ unsigned skipChecksums; /* disable checksum calculation and verification, even when one is present in frame, to save CPU time. * Setting this option to 1 once disables all checksums for the rest of the frame. */ unsigned reserved1; /* must be set to zero for forward compatibility */ @@ -463,6 +478,11 @@ LZ4F_getFrameInfo(LZ4F_dctx* dctx, * `dstBuffer` can freely change between each consecutive function invocation. * `dstBuffer` content will be overwritten. * + * Note: if `LZ4F_getFrameInfo()` is called before `LZ4F_decompress()`, srcBuffer must be updated to reflect + * the number of bytes consumed after reading the frame header. Failure to update srcBuffer before calling + * `LZ4F_decompress()` will cause decompression failure or, even worse, successful but incorrect decompression. + * See the `LZ4F_getFrameInfo()` docs for details. + * * @return : an hint of how many `srcSize` bytes LZ4F_decompress() expects for next call. * Schematically, it's the size of the current (or remaining) compressed block + header of next block. * Respecting the hint provides some small speed benefit, because it skips intermediate buffers. @@ -493,6 +513,109 @@ LZ4F_decompress(LZ4F_dctx* dctx, LZ4FLIB_API void LZ4F_resetDecompressionContext(LZ4F_dctx* dctx); /* always successful */ +/********************************** + * Dictionary compression API + *********************************/ + +/* A Dictionary is useful for the compression of small messages (KB range). + * It dramatically improves compression efficiency. + * + * LZ4 can ingest any input as dictionary, though only the last 64 KB are useful. + * Better results are generally achieved by using Zstandard's Dictionary Builder + * to generate a high-quality dictionary from a set of samples. + * + * The same dictionary will have to be used on the decompression side + * for decoding to be successful. + * To help identify the correct dictionary at decoding stage, + * the frame header allows optional embedding of a dictID field. + */ + +/*! LZ4F_compressBegin_usingDict() : stable since v1.10 + * Inits dictionary compression streaming, and writes the frame header into dstBuffer. + * @dstCapacity must be >= LZ4F_HEADER_SIZE_MAX bytes. + * @prefsPtr is optional : one may provide NULL as argument, + * however, it's the only way to provide dictID in the frame header. + * @dictBuffer must outlive the compression session. + * @return : number of bytes written into dstBuffer for the header, + * or an error code (which can be tested using LZ4F_isError()) + * NOTE: The LZ4Frame spec allows each independent block to be compressed with the dictionary, + * but this entry supports a more limited scenario, where only the first block uses the dictionary. + * This is still useful for small data, which only need one block anyway. + * For larger inputs, one may be more interested in LZ4F_compressFrame_usingCDict() below. + */ +LZ4FLIB_API size_t +LZ4F_compressBegin_usingDict(LZ4F_cctx* cctx, + void* dstBuffer, size_t dstCapacity, + const void* dictBuffer, size_t dictSize, + const LZ4F_preferences_t* prefsPtr); + +/*! LZ4F_decompress_usingDict() : stable since v1.10 + * Same as LZ4F_decompress(), using a predefined dictionary. + * Dictionary is used "in place", without any preprocessing. +** It must remain accessible throughout the entire frame decoding. */ +LZ4FLIB_API size_t +LZ4F_decompress_usingDict(LZ4F_dctx* dctxPtr, + void* dstBuffer, size_t* dstSizePtr, + const void* srcBuffer, size_t* srcSizePtr, + const void* dict, size_t dictSize, + const LZ4F_decompressOptions_t* decompressOptionsPtr); + +/***************************************** + * Bulk processing dictionary compression + *****************************************/ + +/* Loading a dictionary has a cost, since it involves construction of tables. + * The Bulk processing dictionary API makes it possible to share this cost + * over an arbitrary number of compression jobs, even concurrently, + * markedly improving compression latency for these cases. + * + * Note that there is no corresponding bulk API for the decompression side, + * because dictionary does not carry any initialization cost for decompression. + * Use the regular LZ4F_decompress_usingDict() there. + */ +typedef struct LZ4F_CDict_s LZ4F_CDict; + +/*! LZ4_createCDict() : stable since v1.10 + * When compressing multiple messages / blocks using the same dictionary, it's recommended to initialize it just once. + * LZ4_createCDict() will create a digested dictionary, ready to start future compression operations without startup delay. + * LZ4_CDict can be created once and shared by multiple threads concurrently, since its usage is read-only. + * @dictBuffer can be released after LZ4_CDict creation, since its content is copied within CDict. */ +LZ4FLIB_API LZ4F_CDict* LZ4F_createCDict(const void* dictBuffer, size_t dictSize); +LZ4FLIB_API void LZ4F_freeCDict(LZ4F_CDict* CDict); + +/*! LZ4_compressFrame_usingCDict() : stable since v1.10 + * Compress an entire srcBuffer into a valid LZ4 frame using a digested Dictionary. + * @cctx must point to a context created by LZ4F_createCompressionContext(). + * If @cdict==NULL, compress without a dictionary. + * @dstBuffer MUST be >= LZ4F_compressFrameBound(srcSize, preferencesPtr). + * If this condition is not respected, function will fail (@return an errorCode). + * The LZ4F_preferences_t structure is optional : one may provide NULL as argument, + * but it's not recommended, as it's the only way to provide @dictID in the frame header. + * @return : number of bytes written into dstBuffer. + * or an error code if it fails (can be tested using LZ4F_isError()) + * Note: for larger inputs generating multiple independent blocks, + * this entry point uses the dictionary for each block. */ +LZ4FLIB_API size_t +LZ4F_compressFrame_usingCDict(LZ4F_cctx* cctx, + void* dst, size_t dstCapacity, + const void* src, size_t srcSize, + const LZ4F_CDict* cdict, + const LZ4F_preferences_t* preferencesPtr); + +/*! LZ4F_compressBegin_usingCDict() : stable since v1.10 + * Inits streaming dictionary compression, and writes the frame header into dstBuffer. + * @dstCapacity must be >= LZ4F_HEADER_SIZE_MAX bytes. + * @prefsPtr is optional : one may provide NULL as argument, + * note however that it's the only way to insert a @dictID in the frame header. + * @cdict must outlive the compression session. + * @return : number of bytes written into dstBuffer for the header, + * or an error code, which can be tested using LZ4F_isError(). */ +LZ4FLIB_API size_t +LZ4F_compressBegin_usingCDict(LZ4F_cctx* cctx, + void* dstBuffer, size_t dstCapacity, + const LZ4F_CDict* cdict, + const LZ4F_preferences_t* prefsPtr); + #if defined (__cplusplus) } @@ -503,11 +626,8 @@ LZ4FLIB_API void LZ4F_resetDecompressionContext(LZ4F_dctx* dctx); /* always su #if defined(LZ4F_STATIC_LINKING_ONLY) && !defined(LZ4F_H_STATIC_09782039843) #define LZ4F_H_STATIC_09782039843 -#if defined (__cplusplus) -extern "C" { -#endif - -/* These declarations are not stable and may change in the future. +/* Note : + * The below declarations are not stable and may change in the future. * They are therefore only safe to depend on * when the caller is statically linked against the library. * To access their declarations, define LZ4F_STATIC_LINKING_ONLY. @@ -517,6 +637,11 @@ extern "C" { * by defining LZ4F_PUBLISH_STATIC_FUNCTIONS. * Use at your own risk. */ + +#if defined (__cplusplus) +extern "C" { +#endif + #ifdef LZ4F_PUBLISH_STATIC_FUNCTIONS # define LZ4FLIB_STATIC_API LZ4FLIB_API #else @@ -530,7 +655,7 @@ extern "C" { ITEM(ERROR_GENERIC) \ ITEM(ERROR_maxBlockSize_invalid) \ ITEM(ERROR_blockMode_invalid) \ - ITEM(ERROR_contentChecksumFlag_invalid) \ + ITEM(ERROR_parameter_invalid) \ ITEM(ERROR_compressionLevel_invalid) \ ITEM(ERROR_headerVersion_wrong) \ ITEM(ERROR_blockChecksum_invalid) \ @@ -548,6 +673,8 @@ extern "C" { ITEM(ERROR_frameDecoding_alreadyStarted) \ ITEM(ERROR_compressionState_uninitialized) \ ITEM(ERROR_parameter_null) \ + ITEM(ERROR_io_write) \ + ITEM(ERROR_io_read) \ ITEM(ERROR_maxCode) #define LZ4F_GENERATE_ENUM(ENUM) LZ4F_##ENUM, @@ -558,22 +685,26 @@ typedef enum { LZ4F_LIST_ERRORS(LZ4F_GENERATE_ENUM) LZ4FLIB_STATIC_API LZ4F_errorCodes LZ4F_getErrorCode(size_t functionResult); +/********************************** + * Advanced compression operations + *********************************/ /*! LZ4F_getBlockSize() : - * Return, in scalar format (size_t), - * the maximum block size associated with blockSizeID. + * @return, in scalar format (size_t), + * the maximum block size associated with @blockSizeID, + * or an error code (can be tested using LZ4F_isError()) if @blockSizeID is invalid. **/ LZ4FLIB_STATIC_API size_t LZ4F_getBlockSize(LZ4F_blockSizeID_t blockSizeID); /*! LZ4F_uncompressedUpdate() : - * LZ4F_uncompressedUpdate() can be called repetitively to add as much data uncompressed data as necessary. + * LZ4F_uncompressedUpdate() can be called repetitively to add data stored as uncompressed blocks. * Important rule: dstCapacity MUST be large enough to store the entire source buffer as * no compression is done for this operation * If this condition is not respected, LZ4F_uncompressedUpdate() will fail (result is an errorCode). * After an error, the state is left in a UB state, and must be re-initialized or freed. - * If previously a compressed block was written, buffered data is flushed + * If previously a compressed block was written, buffered data is flushed first, * before appending uncompressed data is continued. - * This is only supported when LZ4F_blockIndependent is used + * This operation is only supported when LZ4F_blockIndependent is used. * `cOptPtr` is optional : NULL can be provided, in which case all options are set to default. * @return : number of bytes written into `dstBuffer` (it can be zero, meaning input data was just buffered). * or an error code if it fails (which can be tested using LZ4F_isError()) @@ -585,82 +716,10 @@ LZ4F_uncompressedUpdate(LZ4F_cctx* cctx, const LZ4F_compressOptions_t* cOptPtr); /********************************** - * Bulk processing dictionary API + * Custom memory allocation *********************************/ -/* A Dictionary is useful for the compression of small messages (KB range). - * It dramatically improves compression efficiency. - * - * LZ4 can ingest any input as dictionary, though only the last 64 KB are useful. - * Best results are generally achieved by using Zstandard's Dictionary Builder - * to generate a high-quality dictionary from a set of samples. - * - * Loading a dictionary has a cost, since it involves construction of tables. - * The Bulk processing dictionary API makes it possible to share this cost - * over an arbitrary number of compression jobs, even concurrently, - * markedly improving compression latency for these cases. - * - * The same dictionary will have to be used on the decompression side - * for decoding to be successful. - * To help identify the correct dictionary at decoding stage, - * the frame header allows optional embedding of a dictID field. - */ -typedef struct LZ4F_CDict_s LZ4F_CDict; - -/*! LZ4_createCDict() : - * When compressing multiple messages / blocks using the same dictionary, it's recommended to load it just once. - * LZ4_createCDict() will create a digested dictionary, ready to start future compression operations without startup delay. - * LZ4_CDict can be created once and shared by multiple threads concurrently, since its usage is read-only. - * `dictBuffer` can be released after LZ4_CDict creation, since its content is copied within CDict */ -LZ4FLIB_STATIC_API LZ4F_CDict* LZ4F_createCDict(const void* dictBuffer, size_t dictSize); -LZ4FLIB_STATIC_API void LZ4F_freeCDict(LZ4F_CDict* CDict); - - -/*! LZ4_compressFrame_usingCDict() : - * Compress an entire srcBuffer into a valid LZ4 frame using a digested Dictionary. - * cctx must point to a context created by LZ4F_createCompressionContext(). - * If cdict==NULL, compress without a dictionary. - * dstBuffer MUST be >= LZ4F_compressFrameBound(srcSize, preferencesPtr). - * If this condition is not respected, function will fail (@return an errorCode). - * The LZ4F_preferences_t structure is optional : you may provide NULL as argument, - * but it's not recommended, as it's the only way to provide dictID in the frame header. - * @return : number of bytes written into dstBuffer. - * or an error code if it fails (can be tested using LZ4F_isError()) */ -LZ4FLIB_STATIC_API size_t -LZ4F_compressFrame_usingCDict(LZ4F_cctx* cctx, - void* dst, size_t dstCapacity, - const void* src, size_t srcSize, - const LZ4F_CDict* cdict, - const LZ4F_preferences_t* preferencesPtr); - - -/*! LZ4F_compressBegin_usingCDict() : - * Inits streaming dictionary compression, and writes the frame header into dstBuffer. - * dstCapacity must be >= LZ4F_HEADER_SIZE_MAX bytes. - * `prefsPtr` is optional : you may provide NULL as argument, - * however, it's the only way to provide dictID in the frame header. - * @return : number of bytes written into dstBuffer for the header, - * or an error code (which can be tested using LZ4F_isError()) */ -LZ4FLIB_STATIC_API size_t -LZ4F_compressBegin_usingCDict(LZ4F_cctx* cctx, - void* dstBuffer, size_t dstCapacity, - const LZ4F_CDict* cdict, - const LZ4F_preferences_t* prefsPtr); - - -/*! LZ4F_decompress_usingDict() : - * Same as LZ4F_decompress(), using a predefined dictionary. - * Dictionary is used "in place", without any preprocessing. -** It must remain accessible throughout the entire frame decoding. */ -LZ4FLIB_STATIC_API size_t -LZ4F_decompress_usingDict(LZ4F_dctx* dctxPtr, - void* dstBuffer, size_t* dstSizePtr, - const void* srcBuffer, size_t* srcSizePtr, - const void* dict, size_t dictSize, - const LZ4F_decompressOptions_t* decompressOptionsPtr); - - -/*! Custom memory allocation : +/*! Custom memory allocation : v1.9.4+ * These prototypes make it possible to pass custom allocation/free functions. * LZ4F_customMem is provided at state creation time, using LZ4F_create*_advanced() listed below. * All allocation/free operations will be completed using these custom variants instead of regular <stdlib.h> ones. diff --git a/contrib/libs/lz4/lz4hc.c b/contrib/libs/lz4/lz4hc.c index b21ad6bb59..4d8c36a697 100644 --- a/contrib/libs/lz4/lz4hc.c +++ b/contrib/libs/lz4/lz4hc.c @@ -39,9 +39,10 @@ ***************************************/ /*! HEAPMODE : - * Select how default compression function will allocate workplace memory, - * in stack (0:fastest), or in heap (1:requires malloc()). - * Since workplace is rather large, heap mode is recommended. + * Select how stateless HC compression functions like `LZ4_compress_HC()` + * allocate memory for their workspace: + * in stack (0:fastest), or in heap (1:default, requires malloc()). + * Since workspace is rather large, heap mode is recommended. **/ #ifndef LZ4HC_HEAPMODE # define LZ4HC_HEAPMODE 1 @@ -51,19 +52,19 @@ /*=== Dependency ===*/ #define LZ4_HC_STATIC_LINKING_ONLY #include "lz4hc.h" +#include <limits.h> -/*=== Common definitions ===*/ -#if defined(__GNUC__) +/*=== Shared lz4.c code ===*/ +#ifndef LZ4_SRC_INCLUDED +# if defined(__GNUC__) # pragma GCC diagnostic ignored "-Wunused-function" -#endif -#if defined (__clang__) +# endif +# if defined (__clang__) # pragma clang diagnostic ignored "-Wunused-function" -#endif - -#define LZ4_COMMONDEFS_ONLY -#ifndef LZ4_SRC_INCLUDED -#include "lz4.c" /* LZ4_count, constants, mem */ +# endif +# define LZ4_COMMONDEFS_ONLY +# include "lz4.c" /* LZ4_count, constants, mem */ #endif @@ -79,17 +80,158 @@ typedef enum { noDictCtx, usingDictCtxHc } dictCtx_directive; /*=== Macros ===*/ #define MIN(a,b) ( (a) < (b) ? (a) : (b) ) #define MAX(a,b) ( (a) > (b) ? (a) : (b) ) -#define HASH_FUNCTION(i) (((i) * 2654435761U) >> ((MINMATCH*8)-LZ4HC_HASH_LOG)) -#define DELTANEXTMAXD(p) chainTable[(p) & LZ4HC_MAXD_MASK] /* flexible, LZ4HC_MAXD dependent */ + + +/*=== Levels definition ===*/ +typedef enum { lz4mid, lz4hc, lz4opt } lz4hc_strat_e; +typedef struct { + lz4hc_strat_e strat; + int nbSearches; + U32 targetLength; +} cParams_t; +static const cParams_t k_clTable[LZ4HC_CLEVEL_MAX+1] = { + { lz4mid, 2, 16 }, /* 0, unused */ + { lz4mid, 2, 16 }, /* 1, unused */ + { lz4mid, 2, 16 }, /* 2 */ + { lz4hc, 4, 16 }, /* 3 */ + { lz4hc, 8, 16 }, /* 4 */ + { lz4hc, 16, 16 }, /* 5 */ + { lz4hc, 32, 16 }, /* 6 */ + { lz4hc, 64, 16 }, /* 7 */ + { lz4hc, 128, 16 }, /* 8 */ + { lz4hc, 256, 16 }, /* 9 */ + { lz4opt, 96, 64 }, /*10==LZ4HC_CLEVEL_OPT_MIN*/ + { lz4opt, 512,128 }, /*11 */ + { lz4opt,16384,LZ4_OPT_NUM }, /* 12==LZ4HC_CLEVEL_MAX */ +}; + +static cParams_t LZ4HC_getCLevelParams(int cLevel) +{ + /* note : clevel convention is a bit different from lz4frame, + * possibly something worth revisiting for consistency */ + if (cLevel < 1) + cLevel = LZ4HC_CLEVEL_DEFAULT; + cLevel = MIN(LZ4HC_CLEVEL_MAX, cLevel); + return k_clTable[cLevel]; +} + + +/*=== Hashing ===*/ +#define LZ4HC_HASHSIZE 4 +#define HASH_FUNCTION(i) (((i) * 2654435761U) >> ((MINMATCH*8)-LZ4HC_HASH_LOG)) +static U32 LZ4HC_hashPtr(const void* ptr) { return HASH_FUNCTION(LZ4_read32(ptr)); } + +#if defined(LZ4_FORCE_MEMORY_ACCESS) && (LZ4_FORCE_MEMORY_ACCESS==2) +/* lie to the compiler about data alignment; use with caution */ +static U64 LZ4_read64(const void* memPtr) { return *(const U64*) memPtr; } + +#elif defined(LZ4_FORCE_MEMORY_ACCESS) && (LZ4_FORCE_MEMORY_ACCESS==1) +/* __pack instructions are safer, but compiler specific */ +LZ4_PACK(typedef struct { U64 u64; }) LZ4_unalign64; +static U64 LZ4_read64(const void* ptr) { return ((const LZ4_unalign64*)ptr)->u64; } + +#else /* safe and portable access using memcpy() */ +static U64 LZ4_read64(const void* memPtr) +{ + U64 val; LZ4_memcpy(&val, memPtr, sizeof(val)); return val; +} + +#endif /* LZ4_FORCE_MEMORY_ACCESS */ + +#define LZ4MID_HASHSIZE 8 +#define LZ4MID_HASHLOG (LZ4HC_HASH_LOG-1) +#define LZ4MID_HASHTABLESIZE (1 << LZ4MID_HASHLOG) + +static U32 LZ4MID_hash4(U32 v) { return (v * 2654435761U) >> (32-LZ4MID_HASHLOG); } +static U32 LZ4MID_hash4Ptr(const void* ptr) { return LZ4MID_hash4(LZ4_read32(ptr)); } +/* note: hash7 hashes the lower 56-bits. + * It presumes input was read using little endian.*/ +static U32 LZ4MID_hash7(U64 v) { return (U32)(((v << (64-56)) * 58295818150454627ULL) >> (64-LZ4MID_HASHLOG)) ; } +static U64 LZ4_readLE64(const void* memPtr); +static U32 LZ4MID_hash8Ptr(const void* ptr) { return LZ4MID_hash7(LZ4_readLE64(ptr)); } + +static U64 LZ4_readLE64(const void* memPtr) +{ + if (LZ4_isLittleEndian()) { + return LZ4_read64(memPtr); + } else { + const BYTE* p = (const BYTE*)memPtr; + /* note: relies on the compiler to simplify this expression */ + return (U64)p[0] | ((U64)p[1]<<8) | ((U64)p[2]<<16) | ((U64)p[3]<<24) + | ((U64)p[4]<<32) | ((U64)p[5]<<40) | ((U64)p[6]<<48) | ((U64)p[7]<<56); + } +} + + +/*=== Count match length ===*/ +LZ4_FORCE_INLINE +unsigned LZ4HC_NbCommonBytes32(U32 val) +{ + assert(val != 0); + if (LZ4_isLittleEndian()) { +# if defined(_MSC_VER) && (_MSC_VER >= 1400) && !defined(LZ4_FORCE_SW_BITCOUNT) + unsigned long r; + _BitScanReverse(&r, val); + return (unsigned)((31 - r) >> 3); +# elif (defined(__clang__) || (defined(__GNUC__) && ((__GNUC__ > 3) || \ + ((__GNUC__ == 3) && (__GNUC_MINOR__ >= 4))))) && \ + !defined(LZ4_FORCE_SW_BITCOUNT) + return (unsigned)__builtin_clz(val) >> 3; +# else + val >>= 8; + val = ((((val + 0x00FFFF00) | 0x00FFFFFF) + val) | + (val + 0x00FF0000)) >> 24; + return (unsigned)val ^ 3; +# endif + } else { +# if defined(_MSC_VER) && (_MSC_VER >= 1400) && !defined(LZ4_FORCE_SW_BITCOUNT) + unsigned long r; + _BitScanForward(&r, val); + return (unsigned)(r >> 3); +# elif (defined(__clang__) || (defined(__GNUC__) && ((__GNUC__ > 3) || \ + ((__GNUC__ == 3) && (__GNUC_MINOR__ >= 4))))) && \ + !defined(LZ4_FORCE_SW_BITCOUNT) + return (unsigned)__builtin_ctz(val) >> 3; +# else + const U32 m = 0x01010101; + return (unsigned)((((val - 1) ^ val) & (m - 1)) * m) >> 24; +# endif + } +} + +/** LZ4HC_countBack() : + * @return : negative value, nb of common bytes before ip/match */ +LZ4_FORCE_INLINE +int LZ4HC_countBack(const BYTE* const ip, const BYTE* const match, + const BYTE* const iMin, const BYTE* const mMin) +{ + int back = 0; + int const min = (int)MAX(iMin - ip, mMin - match); + assert(min <= 0); + assert(ip >= iMin); assert((size_t)(ip-iMin) < (1U<<31)); + assert(match >= mMin); assert((size_t)(match - mMin) < (1U<<31)); + + while ((back - min) > 3) { + U32 const v = LZ4_read32(ip + back - 4) ^ LZ4_read32(match + back - 4); + if (v) { + return (back - (int)LZ4HC_NbCommonBytes32(v)); + } else back -= 4; /* 4-byte step */ + } + /* check remainder if any */ + while ( (back > min) + && (ip[back-1] == match[back-1]) ) + back--; + return back; +} + +/*=== Chain table updates ===*/ #define DELTANEXTU16(table, pos) table[(U16)(pos)] /* faster */ /* Make fields passed to, and updated by LZ4HC_encodeSequence explicit */ #define UPDATABLE(ip, op, anchor) &ip, &op, &anchor -static U32 LZ4HC_hashPtr(const void* ptr) { return HASH_FUNCTION(LZ4_read32(ptr)); } - /************************************** -* HC Compression +* Init **************************************/ static void LZ4HC_clearTables (LZ4HC_CCtx_internal* hc4) { @@ -101,6 +243,7 @@ static void LZ4HC_init_internal (LZ4HC_CCtx_internal* hc4, const BYTE* start) { size_t const bufferSize = (size_t)(hc4->end - hc4->prefixStart); size_t newStartingOffset = bufferSize + hc4->dictLimit; + DEBUGLOG(5, "LZ4HC_init_internal"); assert(newStartingOffset >= bufferSize); /* check overflow */ if (newStartingOffset > 1 GB) { LZ4HC_clearTables(hc4); @@ -116,6 +259,524 @@ static void LZ4HC_init_internal (LZ4HC_CCtx_internal* hc4, const BYTE* start) } +/************************************** +* Encode +**************************************/ +/* LZ4HC_encodeSequence() : + * @return : 0 if ok, + * 1 if buffer issue detected */ +LZ4_FORCE_INLINE int LZ4HC_encodeSequence ( + const BYTE** _ip, + BYTE** _op, + const BYTE** _anchor, + int matchLength, + int offset, + limitedOutput_directive limit, + BYTE* oend) +{ +#define ip (*_ip) +#define op (*_op) +#define anchor (*_anchor) + + size_t length; + BYTE* const token = op++; + +#if defined(LZ4_DEBUG) && (LZ4_DEBUG >= 6) + static const BYTE* start = NULL; + static U32 totalCost = 0; + U32 const pos = (start==NULL) ? 0 : (U32)(anchor - start); + U32 const ll = (U32)(ip - anchor); + U32 const llAdd = (ll>=15) ? ((ll-15) / 255) + 1 : 0; + U32 const mlAdd = (matchLength>=19) ? ((matchLength-19) / 255) + 1 : 0; + U32 const cost = 1 + llAdd + ll + 2 + mlAdd; + if (start==NULL) start = anchor; /* only works for single segment */ + /* g_debuglog_enable = (pos >= 2228) & (pos <= 2262); */ + DEBUGLOG(6, "pos:%7u -- literals:%4u, match:%4i, offset:%5i, cost:%4u + %5u", + pos, + (U32)(ip - anchor), matchLength, offset, + cost, totalCost); + totalCost += cost; +#endif + + /* Encode Literal length */ + length = (size_t)(ip - anchor); + LZ4_STATIC_ASSERT(notLimited == 0); + /* Check output limit */ + if (limit && ((op + (length / 255) + length + (2 + 1 + LASTLITERALS)) > oend)) { + DEBUGLOG(6, "Not enough room to write %i literals (%i bytes remaining)", + (int)length, (int)(oend - op)); + return 1; + } + if (length >= RUN_MASK) { + size_t len = length - RUN_MASK; + *token = (RUN_MASK << ML_BITS); + for(; len >= 255 ; len -= 255) *op++ = 255; + *op++ = (BYTE)len; + } else { + *token = (BYTE)(length << ML_BITS); + } + + /* Copy Literals */ + LZ4_wildCopy8(op, anchor, op + length); + op += length; + + /* Encode Offset */ + assert(offset <= LZ4_DISTANCE_MAX ); + assert(offset > 0); + LZ4_writeLE16(op, (U16)(offset)); op += 2; + + /* Encode MatchLength */ + assert(matchLength >= MINMATCH); + length = (size_t)matchLength - MINMATCH; + if (limit && (op + (length / 255) + (1 + LASTLITERALS) > oend)) { + DEBUGLOG(6, "Not enough room to write match length"); + return 1; /* Check output limit */ + } + if (length >= ML_MASK) { + *token += ML_MASK; + length -= ML_MASK; + for(; length >= 510 ; length -= 510) { *op++ = 255; *op++ = 255; } + if (length >= 255) { length -= 255; *op++ = 255; } + *op++ = (BYTE)length; + } else { + *token += (BYTE)(length); + } + + /* Prepare next loop */ + ip += matchLength; + anchor = ip; + + return 0; + +#undef ip +#undef op +#undef anchor +} + + +typedef struct { + int off; + int len; + int back; /* negative value */ +} LZ4HC_match_t; + +LZ4HC_match_t LZ4HC_searchExtDict(const BYTE* ip, U32 ipIndex, + const BYTE* const iLowLimit, const BYTE* const iHighLimit, + const LZ4HC_CCtx_internal* dictCtx, U32 gDictEndIndex, + int currentBestML, int nbAttempts) +{ + size_t const lDictEndIndex = (size_t)(dictCtx->end - dictCtx->prefixStart) + dictCtx->dictLimit; + U32 lDictMatchIndex = dictCtx->hashTable[LZ4HC_hashPtr(ip)]; + U32 matchIndex = lDictMatchIndex + gDictEndIndex - (U32)lDictEndIndex; + int offset = 0, sBack = 0; + assert(lDictEndIndex <= 1 GB); + if (lDictMatchIndex>0) + DEBUGLOG(7, "lDictEndIndex = %zu, lDictMatchIndex = %u", lDictEndIndex, lDictMatchIndex); + while (ipIndex - matchIndex <= LZ4_DISTANCE_MAX && nbAttempts--) { + const BYTE* const matchPtr = dictCtx->prefixStart - dictCtx->dictLimit + lDictMatchIndex; + + if (LZ4_read32(matchPtr) == LZ4_read32(ip)) { + int mlt; + int back = 0; + const BYTE* vLimit = ip + (lDictEndIndex - lDictMatchIndex); + if (vLimit > iHighLimit) vLimit = iHighLimit; + mlt = (int)LZ4_count(ip+MINMATCH, matchPtr+MINMATCH, vLimit) + MINMATCH; + back = (ip > iLowLimit) ? LZ4HC_countBack(ip, matchPtr, iLowLimit, dictCtx->prefixStart) : 0; + mlt -= back; + if (mlt > currentBestML) { + currentBestML = mlt; + offset = (int)(ipIndex - matchIndex); + sBack = back; + DEBUGLOG(7, "found match of length %i within extDictCtx", currentBestML); + } } + + { U32 const nextOffset = DELTANEXTU16(dictCtx->chainTable, lDictMatchIndex); + lDictMatchIndex -= nextOffset; + matchIndex -= nextOffset; + } } + + { LZ4HC_match_t md; + md.len = currentBestML; + md.off = offset; + md.back = sBack; + return md; + } +} + +typedef LZ4HC_match_t (*LZ4MID_searchIntoDict_f)(const BYTE* ip, U32 ipIndex, + const BYTE* const iHighLimit, + const LZ4HC_CCtx_internal* dictCtx, U32 gDictEndIndex); + +static LZ4HC_match_t LZ4MID_searchHCDict(const BYTE* ip, U32 ipIndex, + const BYTE* const iHighLimit, + const LZ4HC_CCtx_internal* dictCtx, U32 gDictEndIndex) +{ + return LZ4HC_searchExtDict(ip,ipIndex, + ip, iHighLimit, + dictCtx, gDictEndIndex, + MINMATCH-1, 2); +} + +static LZ4HC_match_t LZ4MID_searchExtDict(const BYTE* ip, U32 ipIndex, + const BYTE* const iHighLimit, + const LZ4HC_CCtx_internal* dictCtx, U32 gDictEndIndex) +{ + size_t const lDictEndIndex = (size_t)(dictCtx->end - dictCtx->prefixStart) + dictCtx->dictLimit; + const U32* const hash4Table = dictCtx->hashTable; + const U32* const hash8Table = hash4Table + LZ4MID_HASHTABLESIZE; + DEBUGLOG(7, "LZ4MID_searchExtDict (ipIdx=%u)", ipIndex); + + /* search long match first */ + { U32 l8DictMatchIndex = hash8Table[LZ4MID_hash8Ptr(ip)]; + U32 m8Index = l8DictMatchIndex + gDictEndIndex - (U32)lDictEndIndex; + assert(lDictEndIndex <= 1 GB); + if (ipIndex - m8Index <= LZ4_DISTANCE_MAX) { + const BYTE* const matchPtr = dictCtx->prefixStart - dictCtx->dictLimit + l8DictMatchIndex; + const size_t safeLen = MIN(lDictEndIndex - l8DictMatchIndex, (size_t)(iHighLimit - ip)); + int mlt = (int)LZ4_count(ip, matchPtr, ip + safeLen); + if (mlt >= MINMATCH) { + LZ4HC_match_t md; + DEBUGLOG(7, "Found long ExtDict match of len=%u", mlt); + md.len = mlt; + md.off = (int)(ipIndex - m8Index); + md.back = 0; + return md; + } + } + } + + /* search for short match second */ + { U32 l4DictMatchIndex = hash4Table[LZ4MID_hash4Ptr(ip)]; + U32 m4Index = l4DictMatchIndex + gDictEndIndex - (U32)lDictEndIndex; + if (ipIndex - m4Index <= LZ4_DISTANCE_MAX) { + const BYTE* const matchPtr = dictCtx->prefixStart - dictCtx->dictLimit + l4DictMatchIndex; + const size_t safeLen = MIN(lDictEndIndex - l4DictMatchIndex, (size_t)(iHighLimit - ip)); + int mlt = (int)LZ4_count(ip, matchPtr, ip + safeLen); + if (mlt >= MINMATCH) { + LZ4HC_match_t md; + DEBUGLOG(7, "Found short ExtDict match of len=%u", mlt); + md.len = mlt; + md.off = (int)(ipIndex - m4Index); + md.back = 0; + return md; + } + } + } + + /* nothing found */ + { LZ4HC_match_t const md = {0, 0, 0 }; + return md; + } +} + +/************************************** +* Mid Compression (level 2) +**************************************/ + +LZ4_FORCE_INLINE void +LZ4MID_addPosition(U32* hTable, U32 hValue, U32 index) +{ + hTable[hValue] = index; +} + +#define ADDPOS8(_p, _idx) LZ4MID_addPosition(hash8Table, LZ4MID_hash8Ptr(_p), _idx) +#define ADDPOS4(_p, _idx) LZ4MID_addPosition(hash4Table, LZ4MID_hash4Ptr(_p), _idx) + +/* Fill hash tables with references into dictionary. + * The resulting table is only exploitable by LZ4MID (level 2) */ +static void +LZ4MID_fillHTable (LZ4HC_CCtx_internal* cctx, const void* dict, size_t size) +{ + U32* const hash4Table = cctx->hashTable; + U32* const hash8Table = hash4Table + LZ4MID_HASHTABLESIZE; + const BYTE* const prefixPtr = (const BYTE*)dict; + U32 const prefixIdx = cctx->dictLimit; + U32 const target = prefixIdx + (U32)size - LZ4MID_HASHSIZE; + U32 idx = cctx->nextToUpdate; + assert(dict == cctx->prefixStart); + DEBUGLOG(4, "LZ4MID_fillHTable (size:%zu)", size); + if (size <= LZ4MID_HASHSIZE) + return; + + for (; idx < target; idx += 3) { + ADDPOS4(prefixPtr+idx-prefixIdx, idx); + ADDPOS8(prefixPtr+idx+1-prefixIdx, idx+1); + } + + idx = (size > 32 KB + LZ4MID_HASHSIZE) ? target - 32 KB : cctx->nextToUpdate; + for (; idx < target; idx += 1) { + ADDPOS8(prefixPtr+idx-prefixIdx, idx); + } + + cctx->nextToUpdate = target; +} + +static LZ4MID_searchIntoDict_f select_searchDict_function(const LZ4HC_CCtx_internal* dictCtx) +{ + if (dictCtx == NULL) return NULL; + if (LZ4HC_getCLevelParams(dictCtx->compressionLevel).strat == lz4mid) + return LZ4MID_searchExtDict; + return LZ4MID_searchHCDict; +} + +static int LZ4MID_compress ( + LZ4HC_CCtx_internal* const ctx, + const char* const src, + char* const dst, + int* srcSizePtr, + int const maxOutputSize, + const limitedOutput_directive limit, + const dictCtx_directive dict + ) +{ + U32* const hash4Table = ctx->hashTable; + U32* const hash8Table = hash4Table + LZ4MID_HASHTABLESIZE; + const BYTE* ip = (const BYTE*)src; + const BYTE* anchor = ip; + const BYTE* const iend = ip + *srcSizePtr; + const BYTE* const mflimit = iend - MFLIMIT; + const BYTE* const matchlimit = (iend - LASTLITERALS); + const BYTE* const ilimit = (iend - LZ4MID_HASHSIZE); + BYTE* op = (BYTE*)dst; + BYTE* oend = op + maxOutputSize; + + const BYTE* const prefixPtr = ctx->prefixStart; + const U32 prefixIdx = ctx->dictLimit; + const U32 ilimitIdx = (U32)(ilimit - prefixPtr) + prefixIdx; + const BYTE* const dictStart = ctx->dictStart; + const U32 dictIdx = ctx->lowLimit; + const U32 gDictEndIndex = ctx->lowLimit; + const LZ4MID_searchIntoDict_f searchIntoDict = (dict == usingDictCtxHc) ? select_searchDict_function(ctx->dictCtx) : NULL; + unsigned matchLength; + unsigned matchDistance; + + /* input sanitization */ + DEBUGLOG(5, "LZ4MID_compress (%i bytes)", *srcSizePtr); + if (dict == usingDictCtxHc) DEBUGLOG(5, "usingDictCtxHc"); + assert(*srcSizePtr >= 0); + if (*srcSizePtr) assert(src != NULL); + if (maxOutputSize) assert(dst != NULL); + if (*srcSizePtr < 0) return 0; /* invalid */ + if (maxOutputSize < 0) return 0; /* invalid */ + if (*srcSizePtr > LZ4_MAX_INPUT_SIZE) { + /* forbidden: no input is allowed to be that large */ + return 0; + } + if (limit == fillOutput) oend -= LASTLITERALS; /* Hack for support LZ4 format restriction */ + if (*srcSizePtr < LZ4_minLength) + goto _lz4mid_last_literals; /* Input too small, no compression (all literals) */ + + /* main loop */ + while (ip <= mflimit) { + const U32 ipIndex = (U32)(ip - prefixPtr) + prefixIdx; + /* search long match */ + { U32 const h8 = LZ4MID_hash8Ptr(ip); + U32 const pos8 = hash8Table[h8]; + assert(h8 < LZ4MID_HASHTABLESIZE); + assert(pos8 < ipIndex); + LZ4MID_addPosition(hash8Table, h8, ipIndex); + if (ipIndex - pos8 <= LZ4_DISTANCE_MAX) { + /* match candidate found */ + if (pos8 >= prefixIdx) { + const BYTE* const matchPtr = prefixPtr + pos8 - prefixIdx; + assert(matchPtr < ip); + matchLength = LZ4_count(ip, matchPtr, matchlimit); + if (matchLength >= MINMATCH) { + DEBUGLOG(7, "found long match at pos %u (len=%u)", pos8, matchLength); + matchDistance = ipIndex - pos8; + goto _lz4mid_encode_sequence; + } + } else { + if (pos8 >= dictIdx) { + /* extDict match candidate */ + const BYTE* const matchPtr = dictStart + (pos8 - dictIdx); + const size_t safeLen = MIN(prefixIdx - pos8, (size_t)(matchlimit - ip)); + matchLength = LZ4_count(ip, matchPtr, ip + safeLen); + if (matchLength >= MINMATCH) { + DEBUGLOG(7, "found long match at ExtDict pos %u (len=%u)", pos8, matchLength); + matchDistance = ipIndex - pos8; + goto _lz4mid_encode_sequence; + } + } + } + } } + /* search short match */ + { U32 const h4 = LZ4MID_hash4Ptr(ip); + U32 const pos4 = hash4Table[h4]; + assert(h4 < LZ4MID_HASHTABLESIZE); + assert(pos4 < ipIndex); + LZ4MID_addPosition(hash4Table, h4, ipIndex); + if (ipIndex - pos4 <= LZ4_DISTANCE_MAX) { + /* match candidate found */ + if (pos4 >= prefixIdx) { + /* only search within prefix */ + const BYTE* const matchPtr = prefixPtr + (pos4 - prefixIdx); + assert(matchPtr < ip); + assert(matchPtr >= prefixPtr); + matchLength = LZ4_count(ip, matchPtr, matchlimit); + if (matchLength >= MINMATCH) { + /* short match found, let's just check ip+1 for longer */ + U32 const h8 = LZ4MID_hash8Ptr(ip+1); + U32 const pos8 = hash8Table[h8]; + U32 const m2Distance = ipIndex + 1 - pos8; + matchDistance = ipIndex - pos4; + if ( m2Distance <= LZ4_DISTANCE_MAX + && pos8 >= prefixIdx /* only search within prefix */ + && likely(ip < mflimit) + ) { + const BYTE* const m2Ptr = prefixPtr + (pos8 - prefixIdx); + unsigned ml2 = LZ4_count(ip+1, m2Ptr, matchlimit); + if (ml2 > matchLength) { + LZ4MID_addPosition(hash8Table, h8, ipIndex+1); + ip++; + matchLength = ml2; + matchDistance = m2Distance; + } } + goto _lz4mid_encode_sequence; + } + } else { + if (pos4 >= dictIdx) { + /* extDict match candidate */ + const BYTE* const matchPtr = dictStart + (pos4 - dictIdx); + const size_t safeLen = MIN(prefixIdx - pos4, (size_t)(matchlimit - ip)); + matchLength = LZ4_count(ip, matchPtr, ip + safeLen); + if (matchLength >= MINMATCH) { + DEBUGLOG(7, "found match at ExtDict pos %u (len=%u)", pos4, matchLength); + matchDistance = ipIndex - pos4; + goto _lz4mid_encode_sequence; + } + } + } + } } + /* no match found in prefix */ + if ( (dict == usingDictCtxHc) + && (ipIndex - gDictEndIndex < LZ4_DISTANCE_MAX - 8) ) { + /* search a match into external dictionary */ + LZ4HC_match_t dMatch = searchIntoDict(ip, ipIndex, + matchlimit, + ctx->dictCtx, gDictEndIndex); + if (dMatch.len >= MINMATCH) { + DEBUGLOG(7, "found Dictionary match (offset=%i)", dMatch.off); + assert(dMatch.back == 0); + matchLength = (unsigned)dMatch.len; + matchDistance = (unsigned)dMatch.off; + goto _lz4mid_encode_sequence; + } + } + /* no match found */ + ip += 1 + ((ip-anchor) >> 9); /* skip faster over incompressible data */ + continue; + +_lz4mid_encode_sequence: + /* catch back */ + while (((ip > anchor) & ((U32)(ip-prefixPtr) > matchDistance)) && (unlikely(ip[-1] == ip[-(int)matchDistance-1]))) { + ip--; matchLength++; + }; + + /* fill table with beginning of match */ + ADDPOS8(ip+1, ipIndex+1); + ADDPOS8(ip+2, ipIndex+2); + ADDPOS4(ip+1, ipIndex+1); + + /* encode */ + { BYTE* const saved_op = op; + /* LZ4HC_encodeSequence always updates @op; on success, it updates @ip and @anchor */ + if (LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), + (int)matchLength, (int)matchDistance, + limit, oend) ) { + op = saved_op; /* restore @op value before failed LZ4HC_encodeSequence */ + goto _lz4mid_dest_overflow; + } + } + + /* fill table with end of match */ + { U32 endMatchIdx = (U32)(ip-prefixPtr) + prefixIdx; + U32 pos_m2 = endMatchIdx - 2; + if (pos_m2 < ilimitIdx) { + if (likely(ip - prefixPtr > 5)) { + ADDPOS8(ip-5, endMatchIdx - 5); + } + ADDPOS8(ip-3, endMatchIdx - 3); + ADDPOS8(ip-2, endMatchIdx - 2); + ADDPOS4(ip-2, endMatchIdx - 2); + ADDPOS4(ip-1, endMatchIdx - 1); + } + } + } + +_lz4mid_last_literals: + /* Encode Last Literals */ + { size_t lastRunSize = (size_t)(iend - anchor); /* literals */ + size_t llAdd = (lastRunSize + 255 - RUN_MASK) / 255; + size_t const totalSize = 1 + llAdd + lastRunSize; + if (limit == fillOutput) oend += LASTLITERALS; /* restore correct value */ + if (limit && (op + totalSize > oend)) { + if (limit == limitedOutput) return 0; /* not enough space in @dst */ + /* adapt lastRunSize to fill 'dest' */ + lastRunSize = (size_t)(oend - op) - 1 /*token*/; + llAdd = (lastRunSize + 256 - RUN_MASK) / 256; + lastRunSize -= llAdd; + } + DEBUGLOG(6, "Final literal run : %i literals", (int)lastRunSize); + ip = anchor + lastRunSize; /* can be != iend if limit==fillOutput */ + + if (lastRunSize >= RUN_MASK) { + size_t accumulator = lastRunSize - RUN_MASK; + *op++ = (RUN_MASK << ML_BITS); + for(; accumulator >= 255 ; accumulator -= 255) + *op++ = 255; + *op++ = (BYTE) accumulator; + } else { + *op++ = (BYTE)(lastRunSize << ML_BITS); + } + assert(lastRunSize <= (size_t)(oend - op)); + LZ4_memcpy(op, anchor, lastRunSize); + op += lastRunSize; + } + + /* End */ + DEBUGLOG(5, "compressed %i bytes into %i bytes", *srcSizePtr, (int)((char*)op - dst)); + assert(ip >= (const BYTE*)src); + assert(ip <= iend); + *srcSizePtr = (int)(ip - (const BYTE*)src); + assert((char*)op >= dst); + assert(op <= oend); + assert((char*)op - dst < INT_MAX); + return (int)((char*)op - dst); + +_lz4mid_dest_overflow: + if (limit == fillOutput) { + /* Assumption : @ip, @anchor, @optr and @matchLength must be set correctly */ + size_t const ll = (size_t)(ip - anchor); + size_t const ll_addbytes = (ll + 240) / 255; + size_t const ll_totalCost = 1 + ll_addbytes + ll; + BYTE* const maxLitPos = oend - 3; /* 2 for offset, 1 for token */ + DEBUGLOG(6, "Last sequence is overflowing : %u literals, %u remaining space", + (unsigned)ll, (unsigned)(oend-op)); + if (op + ll_totalCost <= maxLitPos) { + /* ll validated; now adjust match length */ + size_t const bytesLeftForMl = (size_t)(maxLitPos - (op+ll_totalCost)); + size_t const maxMlSize = MINMATCH + (ML_MASK-1) + (bytesLeftForMl * 255); + assert(maxMlSize < INT_MAX); + if ((size_t)matchLength > maxMlSize) matchLength= (unsigned)maxMlSize; + if ((oend + LASTLITERALS) - (op + ll_totalCost + 2) - 1 + matchLength >= MFLIMIT) { + DEBUGLOG(6, "Let's encode a last sequence (ll=%u, ml=%u)", (unsigned)ll, matchLength); + LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), + (int)matchLength, (int)matchDistance, + notLimited, oend); + } } + DEBUGLOG(6, "Let's finish with a run of literals (%u bytes left)", (unsigned)(oend-op)); + goto _lz4mid_last_literals; + } + /* compression failed */ + return 0; +} + + +/************************************** +* HC Compression - Search +**************************************/ + /* Update chains up to ip (excluded) */ LZ4_FORCE_INLINE void LZ4HC_Insert (LZ4HC_CCtx_internal* hc4, const BYTE* ip) { @@ -140,23 +801,6 @@ LZ4_FORCE_INLINE void LZ4HC_Insert (LZ4HC_CCtx_internal* hc4, const BYTE* ip) hc4->nextToUpdate = target; } -/** LZ4HC_countBack() : - * @return : negative value, nb of common bytes before ip/match */ -LZ4_FORCE_INLINE -int LZ4HC_countBack(const BYTE* const ip, const BYTE* const match, - const BYTE* const iMin, const BYTE* const mMin) -{ - int back = 0; - int const min = (int)MAX(iMin - ip, mMin - match); - assert(min <= 0); - assert(ip >= iMin); assert((size_t)(ip-iMin) < (1U<<31)); - assert(match >= mMin); assert((size_t)(match - mMin) < (1U<<31)); - while ( (back > min) - && (ip[back-1] == match[back-1]) ) - back--; - return back; -} - #if defined(_MSC_VER) # define LZ4HC_rotl32(x,r) _rotl(x,r) #else @@ -236,22 +880,21 @@ static int LZ4HC_protectDictEnd(U32 const dictLimit, U32 const matchIndex) typedef enum { rep_untested, rep_not, rep_confirmed } repeat_state_e; typedef enum { favorCompressionRatio=0, favorDecompressionSpeed } HCfavor_e; -LZ4_FORCE_INLINE int + +LZ4_FORCE_INLINE LZ4HC_match_t LZ4HC_InsertAndGetWiderMatch ( LZ4HC_CCtx_internal* const hc4, const BYTE* const ip, const BYTE* const iLowLimit, const BYTE* const iHighLimit, int longest, - const BYTE** matchpos, - const BYTE** startpos, const int maxNbAttempts, const int patternAnalysis, const int chainSwap, const dictCtx_directive dict, const HCfavor_e favorDecSpeed) { U16* const chainTable = hc4->chainTable; - U32* const HashTable = hc4->hashTable; - const LZ4HC_CCtx_internal * const dictCtx = hc4->dictCtx; + U32* const hashTable = hc4->hashTable; + const LZ4HC_CCtx_internal* const dictCtx = hc4->dictCtx; const BYTE* const prefixPtr = hc4->prefixStart; const U32 prefixIdx = hc4->dictLimit; const U32 ipIndex = (U32)(ip - prefixPtr) + prefixIdx; @@ -267,22 +910,24 @@ LZ4HC_InsertAndGetWiderMatch ( U32 matchIndex; repeat_state_e repeat = rep_untested; size_t srcPatternLength = 0; + int offset = 0, sBack = 0; DEBUGLOG(7, "LZ4HC_InsertAndGetWiderMatch"); /* First Match */ - LZ4HC_Insert(hc4, ip); - matchIndex = HashTable[LZ4HC_hashPtr(ip)]; - DEBUGLOG(7, "First match at index %u / %u (lowestMatchIndex)", - matchIndex, lowestMatchIndex); + LZ4HC_Insert(hc4, ip); /* insert all prior positions up to ip (excluded) */ + matchIndex = hashTable[LZ4HC_hashPtr(ip)]; + DEBUGLOG(7, "First candidate match for pos %u found at index %u / %u (lowestMatchIndex)", + ipIndex, matchIndex, lowestMatchIndex); while ((matchIndex>=lowestMatchIndex) && (nbAttempts>0)) { int matchLength=0; nbAttempts--; assert(matchIndex < ipIndex); if (favorDecSpeed && (ipIndex - matchIndex < 8)) { - /* do nothing */ + /* do nothing: + * favorDecSpeed intentionally skips matches with offset < 8 */ } else if (matchIndex >= prefixIdx) { /* within current Prefix */ - const BYTE* const matchPtr = prefixPtr + matchIndex - prefixIdx; + const BYTE* const matchPtr = prefixPtr + (matchIndex - prefixIdx); assert(matchPtr < ip); assert(longest >= 1); if (LZ4_read16(iLowLimit + longest - 1) == LZ4_read16(matchPtr - lookBackLength + longest - 1)) { @@ -292,10 +937,11 @@ LZ4HC_InsertAndGetWiderMatch ( matchLength -= back; if (matchLength > longest) { longest = matchLength; - *matchpos = matchPtr + back; - *startpos = ip + back; + offset = (int)(ipIndex - matchIndex); + sBack = back; + DEBUGLOG(7, "Found match of len=%i within prefix, offset=%i, back=%i", longest, offset, -back); } } } - } else { /* lowestMatchIndex <= matchIndex < dictLimit */ + } else { /* lowestMatchIndex <= matchIndex < dictLimit : within Ext Dict */ const BYTE* const matchPtr = dictStart + (matchIndex - dictIdx); assert(matchIndex >= dictIdx); if ( likely(matchIndex <= prefixIdx - 4) @@ -310,8 +956,9 @@ LZ4HC_InsertAndGetWiderMatch ( matchLength -= back; if (matchLength > longest) { longest = matchLength; - *matchpos = prefixPtr - prefixIdx + matchIndex + back; /* virtual pos, relative to ip, to retrieve offset */ - *startpos = ip + back; + offset = (int)(ipIndex - matchIndex); + sBack = back; + DEBUGLOG(7, "Found match of len=%i within dict, offset=%i, back=%i", longest, offset, -back); } } } if (chainSwap && matchLength==longest) { /* better match => select a better chain */ @@ -344,6 +991,7 @@ LZ4HC_InsertAndGetWiderMatch ( if (repeat == rep_untested) { if ( ((pattern & 0xFFFF) == (pattern >> 16)) & ((pattern & 0xFF) == (pattern >> 24)) ) { + DEBUGLOG(7, "Repeat pattern detected, char %02X", pattern >> 24); repeat = rep_confirmed; srcPatternLength = LZ4HC_countPattern(ip+sizeof(pattern), iHighLimit, pattern) + sizeof(pattern); } else { @@ -352,7 +1000,7 @@ LZ4HC_InsertAndGetWiderMatch ( if ( (repeat == rep_confirmed) && (matchCandidateIdx >= lowestMatchIndex) && LZ4HC_protectDictEnd(prefixIdx, matchCandidateIdx) ) { const int extDict = matchCandidateIdx < prefixIdx; - const BYTE* const matchPtr = (extDict ? dictStart - dictIdx : prefixPtr - prefixIdx) + matchCandidateIdx; + const BYTE* const matchPtr = extDict ? dictStart + (matchCandidateIdx - dictIdx) : prefixPtr + (matchCandidateIdx - prefixIdx); if (LZ4_read32(matchPtr) == pattern) { /* good candidate */ const BYTE* const iLimit = extDict ? dictEnd : iHighLimit; size_t forwardPatternLength = LZ4HC_countPattern(matchPtr+sizeof(pattern), iLimit, pattern) + sizeof(pattern); @@ -398,8 +1046,9 @@ LZ4HC_InsertAndGetWiderMatch ( if ((size_t)(ip - prefixPtr) + prefixIdx - matchIndex > LZ4_DISTANCE_MAX) break; assert(maxML < 2 GB); longest = (int)maxML; - *matchpos = prefixPtr - prefixIdx + matchIndex; /* virtual pos, relative to ip, to retrieve offset */ - *startpos = ip; + offset = (int)(ipIndex - matchIndex); + assert(sBack == 0); + DEBUGLOG(7, "Found repeat pattern match of len=%i, offset=%i", longest, offset); } { U32 const distToNextPattern = DELTANEXTU16(chainTable, matchIndex); if (distToNextPattern > matchIndex) break; /* avoid overflow */ @@ -416,11 +1065,12 @@ LZ4HC_InsertAndGetWiderMatch ( if ( dict == usingDictCtxHc && nbAttempts > 0 - && ipIndex - lowestMatchIndex < LZ4_DISTANCE_MAX) { + && withinStartDistance) { size_t const dictEndOffset = (size_t)(dictCtx->end - dictCtx->prefixStart) + dictCtx->dictLimit; U32 dictMatchIndex = dictCtx->hashTable[LZ4HC_hashPtr(ip)]; assert(dictEndOffset <= 1 GB); matchIndex = dictMatchIndex + lowestMatchIndex - (U32)dictEndOffset; + if (dictMatchIndex>0) DEBUGLOG(7, "dictEndOffset = %zu, dictMatchIndex = %u => relative matchIndex = %i", dictEndOffset, dictMatchIndex, (int)dictMatchIndex - (int)dictEndOffset); while (ipIndex - matchIndex <= LZ4_DISTANCE_MAX && nbAttempts--) { const BYTE* const matchPtr = dictCtx->prefixStart - dictCtx->dictLimit + dictMatchIndex; @@ -434,8 +1084,9 @@ LZ4HC_InsertAndGetWiderMatch ( mlt -= back; if (mlt > longest) { longest = mlt; - *matchpos = prefixPtr - prefixIdx + matchIndex + back; - *startpos = ip + back; + offset = (int)(ipIndex - matchIndex); + sBack = back; + DEBUGLOG(7, "found match of length %i within extDictCtx", longest); } } { U32 const nextOffset = DELTANEXTU16(dictCtx->chainTable, dictMatchIndex); @@ -443,112 +1094,29 @@ LZ4HC_InsertAndGetWiderMatch ( matchIndex -= nextOffset; } } } - return longest; + { LZ4HC_match_t md; + assert(longest >= 0); + md.len = longest; + md.off = offset; + md.back = sBack; + return md; + } } -LZ4_FORCE_INLINE int +LZ4_FORCE_INLINE LZ4HC_match_t LZ4HC_InsertAndFindBestMatch(LZ4HC_CCtx_internal* const hc4, /* Index table will be updated */ const BYTE* const ip, const BYTE* const iLimit, - const BYTE** matchpos, const int maxNbAttempts, const int patternAnalysis, const dictCtx_directive dict) { - const BYTE* uselessPtr = ip; + DEBUGLOG(7, "LZ4HC_InsertAndFindBestMatch"); /* note : LZ4HC_InsertAndGetWiderMatch() is able to modify the starting position of a match (*startpos), * but this won't be the case here, as we define iLowLimit==ip, * so LZ4HC_InsertAndGetWiderMatch() won't be allowed to search past ip */ - return LZ4HC_InsertAndGetWiderMatch(hc4, ip, ip, iLimit, MINMATCH-1, matchpos, &uselessPtr, maxNbAttempts, patternAnalysis, 0 /*chainSwap*/, dict, favorCompressionRatio); + return LZ4HC_InsertAndGetWiderMatch(hc4, ip, ip, iLimit, MINMATCH-1, maxNbAttempts, patternAnalysis, 0 /*chainSwap*/, dict, favorCompressionRatio); } -/* LZ4HC_encodeSequence() : - * @return : 0 if ok, - * 1 if buffer issue detected */ -LZ4_FORCE_INLINE int LZ4HC_encodeSequence ( - const BYTE** _ip, - BYTE** _op, - const BYTE** _anchor, - int matchLength, - const BYTE* const match, - limitedOutput_directive limit, - BYTE* oend) -{ -#define ip (*_ip) -#define op (*_op) -#define anchor (*_anchor) - - size_t length; - BYTE* const token = op++; - -#if defined(LZ4_DEBUG) && (LZ4_DEBUG >= 6) - static const BYTE* start = NULL; - static U32 totalCost = 0; - U32 const pos = (start==NULL) ? 0 : (U32)(anchor - start); - U32 const ll = (U32)(ip - anchor); - U32 const llAdd = (ll>=15) ? ((ll-15) / 255) + 1 : 0; - U32 const mlAdd = (matchLength>=19) ? ((matchLength-19) / 255) + 1 : 0; - U32 const cost = 1 + llAdd + ll + 2 + mlAdd; - if (start==NULL) start = anchor; /* only works for single segment */ - /* g_debuglog_enable = (pos >= 2228) & (pos <= 2262); */ - DEBUGLOG(6, "pos:%7u -- literals:%4u, match:%4i, offset:%5u, cost:%4u + %5u", - pos, - (U32)(ip - anchor), matchLength, (U32)(ip-match), - cost, totalCost); - totalCost += cost; -#endif - - /* Encode Literal length */ - length = (size_t)(ip - anchor); - LZ4_STATIC_ASSERT(notLimited == 0); - /* Check output limit */ - if (limit && ((op + (length / 255) + length + (2 + 1 + LASTLITERALS)) > oend)) { - DEBUGLOG(6, "Not enough room to write %i literals (%i bytes remaining)", - (int)length, (int)(oend - op)); - return 1; - } - if (length >= RUN_MASK) { - size_t len = length - RUN_MASK; - *token = (RUN_MASK << ML_BITS); - for(; len >= 255 ; len -= 255) *op++ = 255; - *op++ = (BYTE)len; - } else { - *token = (BYTE)(length << ML_BITS); - } - - /* Copy Literals */ - LZ4_wildCopy8(op, anchor, op + length); - op += length; - - /* Encode Offset */ - assert( (ip - match) <= LZ4_DISTANCE_MAX ); /* note : consider providing offset as a value, rather than as a pointer difference */ - LZ4_writeLE16(op, (U16)(ip - match)); op += 2; - - /* Encode MatchLength */ - assert(matchLength >= MINMATCH); - length = (size_t)matchLength - MINMATCH; - if (limit && (op + (length / 255) + (1 + LASTLITERALS) > oend)) { - DEBUGLOG(6, "Not enough room to write match length"); - return 1; /* Check output limit */ - } - if (length >= ML_MASK) { - *token += ML_MASK; - length -= ML_MASK; - for(; length >= 510 ; length -= 510) { *op++ = 255; *op++ = 255; } - if (length >= 255) { length -= 255; *op++ = 255; } - *op++ = (BYTE)length; - } else { - *token += (BYTE)(length); - } - - /* Prepare next loop */ - ip += matchLength; - anchor = ip; - - return 0; -} -#undef ip -#undef op -#undef anchor LZ4_FORCE_INLINE int LZ4HC_compress_hashChain ( LZ4HC_CCtx_internal* const ctx, @@ -574,127 +1142,130 @@ LZ4_FORCE_INLINE int LZ4HC_compress_hashChain ( BYTE* op = (BYTE*) dest; BYTE* oend = op + maxOutputSize; - int ml0, ml, ml2, ml3; const BYTE* start0; - const BYTE* ref0; - const BYTE* ref = NULL; const BYTE* start2 = NULL; - const BYTE* ref2 = NULL; const BYTE* start3 = NULL; - const BYTE* ref3 = NULL; + LZ4HC_match_t m0, m1, m2, m3; + const LZ4HC_match_t nomatch = {0, 0, 0}; /* init */ + DEBUGLOG(5, "LZ4HC_compress_hashChain (dict?=>%i)", dict); *srcSizePtr = 0; if (limit == fillOutput) oend -= LASTLITERALS; /* Hack for support LZ4 format restriction */ if (inputSize < LZ4_minLength) goto _last_literals; /* Input too small, no compression (all literals) */ /* Main Loop */ while (ip <= mflimit) { - ml = LZ4HC_InsertAndFindBestMatch(ctx, ip, matchlimit, &ref, maxNbAttempts, patternAnalysis, dict); - if (ml<MINMATCH) { ip++; continue; } + m1 = LZ4HC_InsertAndFindBestMatch(ctx, ip, matchlimit, maxNbAttempts, patternAnalysis, dict); + if (m1.len<MINMATCH) { ip++; continue; } /* saved, in case we would skip too much */ - start0 = ip; ref0 = ref; ml0 = ml; + start0 = ip; m0 = m1; _Search2: - if (ip+ml <= mflimit) { - ml2 = LZ4HC_InsertAndGetWiderMatch(ctx, - ip + ml - 2, ip + 0, matchlimit, ml, &ref2, &start2, + DEBUGLOG(7, "_Search2 (currently found match of size %i)", m1.len); + if (ip+m1.len <= mflimit) { + start2 = ip + m1.len - 2; + m2 = LZ4HC_InsertAndGetWiderMatch(ctx, + start2, ip + 0, matchlimit, m1.len, maxNbAttempts, patternAnalysis, 0, dict, favorCompressionRatio); + start2 += m2.back; } else { - ml2 = ml; + m2 = nomatch; /* do not search further */ } - if (ml2 == ml) { /* No better match => encode ML1 */ + if (m2.len <= m1.len) { /* No better match => encode ML1 immediately */ optr = op; - if (LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), ml, ref, limit, oend)) goto _dest_overflow; + if (LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), + m1.len, m1.off, + limit, oend) ) + goto _dest_overflow; continue; } if (start0 < ip) { /* first match was skipped at least once */ - if (start2 < ip + ml0) { /* squeezing ML1 between ML0(original ML1) and ML2 */ - ip = start0; ref = ref0; ml = ml0; /* restore initial ML1 */ + if (start2 < ip + m0.len) { /* squeezing ML1 between ML0(original ML1) and ML2 */ + ip = start0; m1 = m0; /* restore initial Match1 */ } } /* Here, start0==ip */ if ((start2 - ip) < 3) { /* First Match too small : removed */ - ml = ml2; ip = start2; - ref =ref2; + m1 = m2; goto _Search2; } _Search3: - /* At this stage, we have : - * ml2 > ml1, and - * ip1+3 <= ip2 (usually < ip1+ml1) */ if ((start2 - ip) < OPTIMAL_ML) { int correction; - int new_ml = ml; + int new_ml = m1.len; if (new_ml > OPTIMAL_ML) new_ml = OPTIMAL_ML; - if (ip+new_ml > start2 + ml2 - MINMATCH) new_ml = (int)(start2 - ip) + ml2 - MINMATCH; + if (ip+new_ml > start2 + m2.len - MINMATCH) + new_ml = (int)(start2 - ip) + m2.len - MINMATCH; correction = new_ml - (int)(start2 - ip); if (correction > 0) { start2 += correction; - ref2 += correction; - ml2 -= correction; + m2.len -= correction; } } - /* Now, we have start2 = ip+new_ml, with new_ml = min(ml, OPTIMAL_ML=18) */ - if (start2 + ml2 <= mflimit) { - ml3 = LZ4HC_InsertAndGetWiderMatch(ctx, - start2 + ml2 - 3, start2, matchlimit, ml2, &ref3, &start3, + if (start2 + m2.len <= mflimit) { + start3 = start2 + m2.len - 3; + m3 = LZ4HC_InsertAndGetWiderMatch(ctx, + start3, start2, matchlimit, m2.len, maxNbAttempts, patternAnalysis, 0, dict, favorCompressionRatio); + start3 += m3.back; } else { - ml3 = ml2; + m3 = nomatch; /* do not search further */ } - if (ml3 == ml2) { /* No better match => encode ML1 and ML2 */ + if (m3.len <= m2.len) { /* No better match => encode ML1 and ML2 */ /* ip & ref are known; Now for ml */ - if (start2 < ip+ml) ml = (int)(start2 - ip); + if (start2 < ip+m1.len) m1.len = (int)(start2 - ip); /* Now, encode 2 sequences */ optr = op; - if (LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), ml, ref, limit, oend)) goto _dest_overflow; + if (LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), + m1.len, m1.off, + limit, oend) ) + goto _dest_overflow; ip = start2; optr = op; - if (LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), ml2, ref2, limit, oend)) { - ml = ml2; - ref = ref2; + if (LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), + m2.len, m2.off, + limit, oend) ) { + m1 = m2; goto _dest_overflow; } continue; } - if (start3 < ip+ml+3) { /* Not enough space for match 2 : remove it */ - if (start3 >= (ip+ml)) { /* can write Seq1 immediately ==> Seq2 is removed, so Seq3 becomes Seq1 */ - if (start2 < ip+ml) { - int correction = (int)(ip+ml - start2); + if (start3 < ip+m1.len+3) { /* Not enough space for match 2 : remove it */ + if (start3 >= (ip+m1.len)) { /* can write Seq1 immediately ==> Seq2 is removed, so Seq3 becomes Seq1 */ + if (start2 < ip+m1.len) { + int correction = (int)(ip+m1.len - start2); start2 += correction; - ref2 += correction; - ml2 -= correction; - if (ml2 < MINMATCH) { + m2.len -= correction; + if (m2.len < MINMATCH) { start2 = start3; - ref2 = ref3; - ml2 = ml3; + m2 = m3; } } optr = op; - if (LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), ml, ref, limit, oend)) goto _dest_overflow; + if (LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), + m1.len, m1.off, + limit, oend) ) + goto _dest_overflow; ip = start3; - ref = ref3; - ml = ml3; + m1 = m3; start0 = start2; - ref0 = ref2; - ml0 = ml2; + m0 = m2; goto _Search2; } start2 = start3; - ref2 = ref3; - ml2 = ml3; + m2 = m3; goto _Search3; } @@ -703,29 +1274,32 @@ _Search3: * let's write the first one ML1. * ip & ref are known; Now decide ml. */ - if (start2 < ip+ml) { + if (start2 < ip+m1.len) { if ((start2 - ip) < OPTIMAL_ML) { int correction; - if (ml > OPTIMAL_ML) ml = OPTIMAL_ML; - if (ip + ml > start2 + ml2 - MINMATCH) ml = (int)(start2 - ip) + ml2 - MINMATCH; - correction = ml - (int)(start2 - ip); + if (m1.len > OPTIMAL_ML) m1.len = OPTIMAL_ML; + if (ip + m1.len > start2 + m2.len - MINMATCH) + m1.len = (int)(start2 - ip) + m2.len - MINMATCH; + correction = m1.len - (int)(start2 - ip); if (correction > 0) { start2 += correction; - ref2 += correction; - ml2 -= correction; + m2.len -= correction; } } else { - ml = (int)(start2 - ip); + m1.len = (int)(start2 - ip); } } optr = op; - if (LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), ml, ref, limit, oend)) goto _dest_overflow; + if ( LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), + m1.len, m1.off, + limit, oend) ) + goto _dest_overflow; /* ML2 becomes ML1 */ - ip = start2; ref = ref2; ml = ml2; + ip = start2; m1 = m2; /* ML3 becomes ML2 */ - start2 = start3; ref2 = ref3; ml2 = ml3; + start2 = start3; m2 = m3; /* let's find a new ML3 */ goto _Search3; @@ -765,7 +1339,7 @@ _last_literals: _dest_overflow: if (limit == fillOutput) { - /* Assumption : ip, anchor, ml and ref must be set correctly */ + /* Assumption : @ip, @anchor, @optr and @m1 must be set correctly */ size_t const ll = (size_t)(ip - anchor); size_t const ll_addbytes = (ll + 240) / 255; size_t const ll_totalCost = 1 + ll_addbytes + ll; @@ -776,10 +1350,10 @@ _dest_overflow: /* ll validated; now adjust match length */ size_t const bytesLeftForMl = (size_t)(maxLitPos - (op+ll_totalCost)); size_t const maxMlSize = MINMATCH + (ML_MASK-1) + (bytesLeftForMl * 255); - assert(maxMlSize < INT_MAX); assert(ml >= 0); - if ((size_t)ml > maxMlSize) ml = (int)maxMlSize; - if ((oend + LASTLITERALS) - (op + ll_totalCost + 2) - 1 + ml >= MFLIMIT) { - LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), ml, ref, notLimited, oend); + assert(maxMlSize < INT_MAX); assert(m1.len >= 0); + if ((size_t)m1.len > maxMlSize) m1.len = (int)maxMlSize; + if ((oend + LASTLITERALS) - (op + ll_totalCost + 2) - 1 + m1.len >= MFLIMIT) { + LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), m1.len, m1.off, notLimited, oend); } } goto _last_literals; } @@ -796,54 +1370,34 @@ static int LZ4HC_compress_optimal( LZ4HC_CCtx_internal* ctx, const dictCtx_directive dict, const HCfavor_e favorDecSpeed); - -LZ4_FORCE_INLINE int LZ4HC_compress_generic_internal ( - LZ4HC_CCtx_internal* const ctx, - const char* const src, - char* const dst, - int* const srcSizePtr, - int const dstCapacity, - int cLevel, - const limitedOutput_directive limit, - const dictCtx_directive dict - ) +LZ4_FORCE_INLINE int +LZ4HC_compress_generic_internal ( + LZ4HC_CCtx_internal* const ctx, + const char* const src, + char* const dst, + int* const srcSizePtr, + int const dstCapacity, + int cLevel, + const limitedOutput_directive limit, + const dictCtx_directive dict + ) { - typedef enum { lz4hc, lz4opt } lz4hc_strat_e; - typedef struct { - lz4hc_strat_e strat; - int nbSearches; - U32 targetLength; - } cParams_t; - static const cParams_t clTable[LZ4HC_CLEVEL_MAX+1] = { - { lz4hc, 2, 16 }, /* 0, unused */ - { lz4hc, 2, 16 }, /* 1, unused */ - { lz4hc, 2, 16 }, /* 2, unused */ - { lz4hc, 4, 16 }, /* 3 */ - { lz4hc, 8, 16 }, /* 4 */ - { lz4hc, 16, 16 }, /* 5 */ - { lz4hc, 32, 16 }, /* 6 */ - { lz4hc, 64, 16 }, /* 7 */ - { lz4hc, 128, 16 }, /* 8 */ - { lz4hc, 256, 16 }, /* 9 */ - { lz4opt, 96, 64 }, /*10==LZ4HC_CLEVEL_OPT_MIN*/ - { lz4opt, 512,128 }, /*11 */ - { lz4opt,16384,LZ4_OPT_NUM }, /* 12==LZ4HC_CLEVEL_MAX */ - }; - - DEBUGLOG(4, "LZ4HC_compress_generic(ctx=%p, src=%p, srcSize=%d, limit=%d)", - ctx, src, *srcSizePtr, limit); + DEBUGLOG(5, "LZ4HC_compress_generic_internal(src=%p, srcSize=%d)", + src, *srcSizePtr); if (limit == fillOutput && dstCapacity < 1) return 0; /* Impossible to store anything */ - if ((U32)*srcSizePtr > (U32)LZ4_MAX_INPUT_SIZE) return 0; /* Unsupported input size (too large or negative) */ + if ((U32)*srcSizePtr > (U32)LZ4_MAX_INPUT_SIZE) return 0; /* Unsupported input size (too large or negative) */ ctx->end += *srcSizePtr; - if (cLevel < 1) cLevel = LZ4HC_CLEVEL_DEFAULT; /* note : convention is different from lz4frame, maybe something to review */ - cLevel = MIN(LZ4HC_CLEVEL_MAX, cLevel); - { cParams_t const cParam = clTable[cLevel]; + { cParams_t const cParam = LZ4HC_getCLevelParams(cLevel); HCfavor_e const favor = ctx->favorDecSpeed ? favorDecompressionSpeed : favorCompressionRatio; int result; - if (cParam.strat == lz4hc) { + if (cParam.strat == lz4mid) { + result = LZ4MID_compress(ctx, + src, dst, srcSizePtr, dstCapacity, + limit, dict); + } else if (cParam.strat == lz4hc) { result = LZ4HC_compress_hashChain(ctx, src, dst, srcSizePtr, dstCapacity, cParam.nbSearches, limit, dict); @@ -852,7 +1406,7 @@ LZ4_FORCE_INLINE int LZ4HC_compress_generic_internal ( result = LZ4HC_compress_optimal(ctx, src, dst, srcSizePtr, dstCapacity, cParam.nbSearches, cParam.targetLength, limit, - cLevel == LZ4HC_CLEVEL_MAX, /* ultra mode */ + cLevel >= LZ4HC_CLEVEL_MAX, /* ultra mode */ dict, favor); } if (result <= 0) ctx->dirty = 1; @@ -877,6 +1431,13 @@ LZ4HC_compress_generic_noDictCtx ( return LZ4HC_compress_generic_internal(ctx, src, dst, srcSizePtr, dstCapacity, cLevel, limit, noDictCtx); } +static int isStateCompatible(const LZ4HC_CCtx_internal* ctx1, const LZ4HC_CCtx_internal* ctx2) +{ + int const isMid1 = LZ4HC_getCLevelParams(ctx1->compressionLevel).strat == lz4mid; + int const isMid2 = LZ4HC_getCLevelParams(ctx2->compressionLevel).strat == lz4mid; + return !(isMid1 ^ isMid2); +} + static int LZ4HC_compress_generic_dictCtx ( LZ4HC_CCtx_internal* const ctx, @@ -893,7 +1454,7 @@ LZ4HC_compress_generic_dictCtx ( if (position >= 64 KB) { ctx->dictCtx = NULL; return LZ4HC_compress_generic_noDictCtx(ctx, src, dst, srcSizePtr, dstCapacity, cLevel, limit); - } else if (position == 0 && *srcSizePtr > 4 KB) { + } else if (position == 0 && *srcSizePtr > 4 KB && isStateCompatible(ctx, ctx->dictCtx)) { LZ4_memcpy(ctx, ctx->dictCtx, sizeof(LZ4HC_CCtx_internal)); LZ4HC_setExternalDict(ctx, (const BYTE *)src); ctx->compressionLevel = (short)cLevel; @@ -965,6 +1526,7 @@ int LZ4_compress_HC(const char* src, char* dst, int srcSize, int dstCapacity, in LZ4_streamHC_t state; LZ4_streamHC_t* const statePtr = &state; #endif + DEBUGLOG(5, "LZ4_compress_HC") cSize = LZ4_compress_HC_extStateHC(statePtr, src, dst, srcSize, dstCapacity, compressionLevel); #if defined(LZ4HC_HEAPMODE) && LZ4HC_HEAPMODE==1 FREEMEM(statePtr); @@ -1032,18 +1594,16 @@ void LZ4_resetStreamHC (LZ4_streamHC_t* LZ4_streamHCPtr, int compressionLevel) void LZ4_resetStreamHC_fast (LZ4_streamHC_t* LZ4_streamHCPtr, int compressionLevel) { - DEBUGLOG(4, "LZ4_resetStreamHC_fast(%p, %d)", LZ4_streamHCPtr, compressionLevel); - if (LZ4_streamHCPtr->internal_donotuse.dirty) { + LZ4HC_CCtx_internal* const s = &LZ4_streamHCPtr->internal_donotuse; + DEBUGLOG(5, "LZ4_resetStreamHC_fast(%p, %d)", LZ4_streamHCPtr, compressionLevel); + if (s->dirty) { LZ4_initStreamHC(LZ4_streamHCPtr, sizeof(*LZ4_streamHCPtr)); } else { - /* preserve end - prefixStart : can trigger clearTable's threshold */ - if (LZ4_streamHCPtr->internal_donotuse.end != NULL) { - LZ4_streamHCPtr->internal_donotuse.end -= (uptrval)LZ4_streamHCPtr->internal_donotuse.prefixStart; - } else { - assert(LZ4_streamHCPtr->internal_donotuse.prefixStart == NULL); - } - LZ4_streamHCPtr->internal_donotuse.prefixStart = NULL; - LZ4_streamHCPtr->internal_donotuse.dictCtx = NULL; + assert(s->end >= s->prefixStart); + s->dictLimit += (U32)(s->end - s->prefixStart); + s->prefixStart = NULL; + s->end = NULL; + s->dictCtx = NULL; } LZ4_setCompressionLevel(LZ4_streamHCPtr, compressionLevel); } @@ -1067,7 +1627,9 @@ int LZ4_loadDictHC (LZ4_streamHC_t* LZ4_streamHCPtr, const char* dictionary, int dictSize) { LZ4HC_CCtx_internal* const ctxPtr = &LZ4_streamHCPtr->internal_donotuse; - DEBUGLOG(4, "LZ4_loadDictHC(ctx:%p, dict:%p, dictSize:%d)", LZ4_streamHCPtr, dictionary, dictSize); + cParams_t cp; + DEBUGLOG(4, "LZ4_loadDictHC(ctx:%p, dict:%p, dictSize:%d, clevel=%d)", LZ4_streamHCPtr, dictionary, dictSize, ctxPtr->compressionLevel); + assert(dictSize >= 0); assert(LZ4_streamHCPtr != NULL); if (dictSize > 64 KB) { dictionary += (size_t)dictSize - 64 KB; @@ -1077,10 +1639,15 @@ int LZ4_loadDictHC (LZ4_streamHC_t* LZ4_streamHCPtr, { int const cLevel = ctxPtr->compressionLevel; LZ4_initStreamHC(LZ4_streamHCPtr, sizeof(*LZ4_streamHCPtr)); LZ4_setCompressionLevel(LZ4_streamHCPtr, cLevel); + cp = LZ4HC_getCLevelParams(cLevel); } LZ4HC_init_internal (ctxPtr, (const BYTE*)dictionary); ctxPtr->end = (const BYTE*)dictionary + dictSize; - if (dictSize >= 4) LZ4HC_Insert (ctxPtr, ctxPtr->end-3); + if (cp.strat == lz4mid) { + LZ4MID_fillHTable (ctxPtr, dictionary, (size_t)dictSize); + } else { + if (dictSize >= LZ4HC_HASHSIZE) LZ4HC_Insert (ctxPtr, ctxPtr->end-3); + } return dictSize; } @@ -1093,8 +1660,10 @@ void LZ4_attach_HC_dictionary(LZ4_streamHC_t *working_stream, const LZ4_streamHC static void LZ4HC_setExternalDict(LZ4HC_CCtx_internal* ctxPtr, const BYTE* newBlock) { DEBUGLOG(4, "LZ4HC_setExternalDict(%p, %p)", ctxPtr, newBlock); - if (ctxPtr->end >= ctxPtr->prefixStart + 4) - LZ4HC_Insert (ctxPtr, ctxPtr->end-3); /* Referencing remaining dictionary content */ + if ( (ctxPtr->end >= ctxPtr->prefixStart + 4) + && (LZ4HC_getCLevelParams(ctxPtr->compressionLevel).strat != lz4mid) ) { + LZ4HC_Insert (ctxPtr, ctxPtr->end-3); /* Referencing remaining dictionary content */ + } /* Only one memory segment for extDict, so any previous extDict is lost at this stage */ ctxPtr->lowLimit = ctxPtr->dictLimit; @@ -1119,7 +1688,8 @@ LZ4_compressHC_continue_generic (LZ4_streamHC_t* LZ4_streamHCPtr, LZ4_streamHCPtr, src, *srcSizePtr, limit); assert(ctxPtr != NULL); /* auto-init if forgotten */ - if (ctxPtr->prefixStart == NULL) LZ4HC_init_internal (ctxPtr, (const BYTE*) src); + if (ctxPtr->prefixStart == NULL) + LZ4HC_init_internal (ctxPtr, (const BYTE*) src); /* Check overflow */ if ((size_t)(ctxPtr->end - ctxPtr->prefixStart) + ctxPtr->dictLimit > 2 GB) { @@ -1140,7 +1710,8 @@ LZ4_compressHC_continue_generic (LZ4_streamHC_t* LZ4_streamHCPtr, if (sourceEnd > dictEnd) sourceEnd = dictEnd; ctxPtr->lowLimit += (U32)(sourceEnd - ctxPtr->dictStart); ctxPtr->dictStart += (U32)(sourceEnd - ctxPtr->dictStart); - if (ctxPtr->dictLimit - ctxPtr->lowLimit < 4) { + /* invalidate dictionary is it's too small */ + if (ctxPtr->dictLimit - ctxPtr->lowLimit < LZ4HC_HASHSIZE) { ctxPtr->lowLimit = ctxPtr->dictLimit; ctxPtr->dictStart = ctxPtr->prefixStart; } } } @@ -1150,6 +1721,7 @@ LZ4_compressHC_continue_generic (LZ4_streamHC_t* LZ4_streamHCPtr, int LZ4_compress_HC_continue (LZ4_streamHC_t* LZ4_streamHCPtr, const char* src, char* dst, int srcSize, int dstCapacity) { + DEBUGLOG(5, "LZ4_compress_HC_continue"); if (dstCapacity < LZ4_compressBound(srcSize)) return LZ4_compressHC_continue_generic (LZ4_streamHCPtr, src, dst, &srcSize, dstCapacity, limitedOutput); else @@ -1162,7 +1734,6 @@ int LZ4_compress_HC_continue_destSize (LZ4_streamHC_t* LZ4_streamHCPtr, const ch } - /* LZ4_saveDictHC : * save history content * into a user-provided buffer @@ -1179,10 +1750,10 @@ int LZ4_saveDictHC (LZ4_streamHC_t* LZ4_streamHCPtr, char* safeBuffer, int dictS if (dictSize > prefixSize) dictSize = prefixSize; if (safeBuffer == NULL) assert(dictSize == 0); if (dictSize > 0) - LZ4_memmove(safeBuffer, streamPtr->end - dictSize, dictSize); + LZ4_memmove(safeBuffer, streamPtr->end - dictSize, (size_t)dictSize); { U32 const endIndex = (U32)(streamPtr->end - streamPtr->prefixStart) + streamPtr->dictLimit; - streamPtr->end = (const BYTE*)safeBuffer + dictSize; - streamPtr->prefixStart = streamPtr->end - dictSize; + streamPtr->end = (safeBuffer == NULL) ? NULL : (const BYTE*)safeBuffer + dictSize; + streamPtr->prefixStart = (const BYTE*)safeBuffer; streamPtr->dictLimit = endIndex - (U32)dictSize; streamPtr->lowLimit = endIndex - (U32)dictSize; streamPtr->dictStart = streamPtr->prefixStart; @@ -1193,75 +1764,6 @@ int LZ4_saveDictHC (LZ4_streamHC_t* LZ4_streamHCPtr, char* safeBuffer, int dictS } -/*************************************************** -* Deprecated Functions -***************************************************/ - -/* These functions currently generate deprecation warnings */ - -/* Wrappers for deprecated compression functions */ -int LZ4_compressHC(const char* src, char* dst, int srcSize) { return LZ4_compress_HC (src, dst, srcSize, LZ4_compressBound(srcSize), 0); } -int LZ4_compressHC_limitedOutput(const char* src, char* dst, int srcSize, int maxDstSize) { return LZ4_compress_HC(src, dst, srcSize, maxDstSize, 0); } -int LZ4_compressHC2(const char* src, char* dst, int srcSize, int cLevel) { return LZ4_compress_HC (src, dst, srcSize, LZ4_compressBound(srcSize), cLevel); } -int LZ4_compressHC2_limitedOutput(const char* src, char* dst, int srcSize, int maxDstSize, int cLevel) { return LZ4_compress_HC(src, dst, srcSize, maxDstSize, cLevel); } -int LZ4_compressHC_withStateHC (void* state, const char* src, char* dst, int srcSize) { return LZ4_compress_HC_extStateHC (state, src, dst, srcSize, LZ4_compressBound(srcSize), 0); } -int LZ4_compressHC_limitedOutput_withStateHC (void* state, const char* src, char* dst, int srcSize, int maxDstSize) { return LZ4_compress_HC_extStateHC (state, src, dst, srcSize, maxDstSize, 0); } -int LZ4_compressHC2_withStateHC (void* state, const char* src, char* dst, int srcSize, int cLevel) { return LZ4_compress_HC_extStateHC(state, src, dst, srcSize, LZ4_compressBound(srcSize), cLevel); } -int LZ4_compressHC2_limitedOutput_withStateHC (void* state, const char* src, char* dst, int srcSize, int maxDstSize, int cLevel) { return LZ4_compress_HC_extStateHC(state, src, dst, srcSize, maxDstSize, cLevel); } -int LZ4_compressHC_continue (LZ4_streamHC_t* ctx, const char* src, char* dst, int srcSize) { return LZ4_compress_HC_continue (ctx, src, dst, srcSize, LZ4_compressBound(srcSize)); } -int LZ4_compressHC_limitedOutput_continue (LZ4_streamHC_t* ctx, const char* src, char* dst, int srcSize, int maxDstSize) { return LZ4_compress_HC_continue (ctx, src, dst, srcSize, maxDstSize); } - - -/* Deprecated streaming functions */ -int LZ4_sizeofStreamStateHC(void) { return sizeof(LZ4_streamHC_t); } - -/* state is presumed correctly sized, aka >= sizeof(LZ4_streamHC_t) - * @return : 0 on success, !=0 if error */ -int LZ4_resetStreamStateHC(void* state, char* inputBuffer) -{ - LZ4_streamHC_t* const hc4 = LZ4_initStreamHC(state, sizeof(*hc4)); - if (hc4 == NULL) return 1; /* init failed */ - LZ4HC_init_internal (&hc4->internal_donotuse, (const BYTE*)inputBuffer); - return 0; -} - -#if !defined(LZ4_STATIC_LINKING_ONLY_DISABLE_MEMORY_ALLOCATION) -void* LZ4_createHC (const char* inputBuffer) -{ - LZ4_streamHC_t* const hc4 = LZ4_createStreamHC(); - if (hc4 == NULL) return NULL; /* not enough memory */ - LZ4HC_init_internal (&hc4->internal_donotuse, (const BYTE*)inputBuffer); - return hc4; -} - -int LZ4_freeHC (void* LZ4HC_Data) -{ - if (!LZ4HC_Data) return 0; /* support free on NULL */ - FREEMEM(LZ4HC_Data); - return 0; -} -#endif - -int LZ4_compressHC2_continue (void* LZ4HC_Data, const char* src, char* dst, int srcSize, int cLevel) -{ - return LZ4HC_compress_generic (&((LZ4_streamHC_t*)LZ4HC_Data)->internal_donotuse, src, dst, &srcSize, 0, cLevel, notLimited); -} - -int LZ4_compressHC2_limitedOutput_continue (void* LZ4HC_Data, const char* src, char* dst, int srcSize, int dstCapacity, int cLevel) -{ - return LZ4HC_compress_generic (&((LZ4_streamHC_t*)LZ4HC_Data)->internal_donotuse, src, dst, &srcSize, dstCapacity, cLevel, limitedOutput); -} - -char* LZ4_slideInputBufferHC(void* LZ4HC_Data) -{ - LZ4_streamHC_t* const ctx = (LZ4_streamHC_t*)LZ4HC_Data; - const BYTE* bufferStart = ctx->internal_donotuse.prefixStart - ctx->internal_donotuse.dictLimit + ctx->internal_donotuse.lowLimit; - LZ4_resetStreamHC_fast(ctx, ctx->internal_donotuse.compressionLevel); - /* avoid const char * -> char * conversion warning :( */ - return (char*)(uptrval)bufferStart; -} - - /* ================================================ * LZ4 Optimal parser (levels [LZ4HC_CLEVEL_OPT_MIN - LZ4HC_CLEVEL_MAX]) * ===============================================*/ @@ -1282,7 +1784,6 @@ LZ4_FORCE_INLINE int LZ4HC_literalsPrice(int const litlen) return price; } - /* requires mlen >= MINMATCH */ LZ4_FORCE_INLINE int LZ4HC_sequencePrice(int litlen, int mlen) { @@ -1298,12 +1799,6 @@ LZ4_FORCE_INLINE int LZ4HC_sequencePrice(int litlen, int mlen) return price; } - -typedef struct { - int off; - int len; -} LZ4HC_match_t; - LZ4_FORCE_INLINE LZ4HC_match_t LZ4HC_FindLongerMatch(LZ4HC_CCtx_internal* const ctx, const BYTE* ip, const BYTE* const iHighLimit, @@ -1311,19 +1806,17 @@ LZ4HC_FindLongerMatch(LZ4HC_CCtx_internal* const ctx, const dictCtx_directive dict, const HCfavor_e favorDecSpeed) { - LZ4HC_match_t match = { 0 , 0 }; - const BYTE* matchPtr = NULL; + LZ4HC_match_t const match0 = { 0 , 0, 0 }; /* note : LZ4HC_InsertAndGetWiderMatch() is able to modify the starting position of a match (*startpos), * but this won't be the case here, as we define iLowLimit==ip, - * so LZ4HC_InsertAndGetWiderMatch() won't be allowed to search past ip */ - int matchLength = LZ4HC_InsertAndGetWiderMatch(ctx, ip, ip, iHighLimit, minLen, &matchPtr, &ip, nbSearches, 1 /*patternAnalysis*/, 1 /*chainSwap*/, dict, favorDecSpeed); - if (matchLength <= minLen) return match; + ** so LZ4HC_InsertAndGetWiderMatch() won't be allowed to search past ip */ + LZ4HC_match_t md = LZ4HC_InsertAndGetWiderMatch(ctx, ip, ip, iHighLimit, minLen, nbSearches, 1 /*patternAnalysis*/, 1 /*chainSwap*/, dict, favorDecSpeed); + assert(md.back == 0); + if (md.len <= minLen) return match0; if (favorDecSpeed) { - if ((matchLength>18) & (matchLength<=36)) matchLength=18; /* favor shortcut */ + if ((md.len>18) & (md.len<=36)) md.len=18; /* favor dec.speed (shortcut) */ } - match.len = matchLength; - match.off = (int)(ip-matchPtr); - return match; + return md; } @@ -1356,7 +1849,7 @@ static int LZ4HC_compress_optimal ( LZ4HC_CCtx_internal* ctx, BYTE* opSaved = (BYTE*) dst; BYTE* oend = op + dstCapacity; int ovml = MINMATCH; /* overflow - last sequence */ - const BYTE* ovref = NULL; + int ovoff = 0; /* init */ #if defined(LZ4HC_HEAPMODE) && LZ4HC_HEAPMODE==1 @@ -1379,11 +1872,10 @@ static int LZ4HC_compress_optimal ( LZ4HC_CCtx_internal* ctx, if ((size_t)firstMatch.len > sufficient_len) { /* good enough solution : immediate encoding */ int const firstML = firstMatch.len; - const BYTE* const matchPos = ip - firstMatch.off; opSaved = op; - if ( LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), firstML, matchPos, limit, oend) ) { /* updates ip, op and anchor */ + if ( LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), firstML, firstMatch.off, limit, oend) ) { /* updates ip, op and anchor */ ovml = firstML; - ovref = matchPos; + ovoff = firstMatch.off; goto _dest_overflow; } continue; @@ -1401,11 +1893,11 @@ static int LZ4HC_compress_optimal ( LZ4HC_CCtx_internal* ctx, rPos, cost, opt[rPos].litlen); } } /* set prices using initial match */ - { int mlen = MINMATCH; - int const matchML = firstMatch.len; /* necessarily < sufficient_len < LZ4_OPT_NUM */ + { int const matchML = firstMatch.len; /* necessarily < sufficient_len < LZ4_OPT_NUM */ int const offset = firstMatch.off; + int mlen; assert(matchML < LZ4_OPT_NUM); - for ( ; mlen <= matchML ; mlen++) { + for (mlen = MINMATCH ; mlen <= matchML ; mlen++) { int const cost = LZ4HC_sequencePrice(llen, mlen); opt[mlen].mlen = mlen; opt[mlen].off = offset; @@ -1557,9 +2049,9 @@ encode: /* cur, last_match_pos, best_mlen, best_off must be set */ assert(ml >= MINMATCH); assert((offset >= 1) && (offset <= LZ4_DISTANCE_MAX)); opSaved = op; - if ( LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), ml, ip - offset, limit, oend) ) { /* updates ip, op and anchor */ + if ( LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), ml, offset, limit, oend) ) { /* updates ip, op and anchor */ ovml = ml; - ovref = ip - offset; + ovoff = offset; goto _dest_overflow; } } } } /* while (ip <= mflimit) */ @@ -1618,14 +2110,83 @@ if (limit == fillOutput) { if ((oend + LASTLITERALS) - (op + ll_totalCost + 2) - 1 + ovml >= MFLIMIT) { DEBUGLOG(6, "Space to end : %i + ml (%i)", (int)((oend + LASTLITERALS) - (op + ll_totalCost + 2) - 1), ovml); DEBUGLOG(6, "Before : ip = %p, anchor = %p", ip, anchor); - LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), ovml, ovref, notLimited, oend); + LZ4HC_encodeSequence(UPDATABLE(ip, op, anchor), ovml, ovoff, notLimited, oend); DEBUGLOG(6, "After : ip = %p, anchor = %p", ip, anchor); } } goto _last_literals; } _return_label: #if defined(LZ4HC_HEAPMODE) && LZ4HC_HEAPMODE==1 - FREEMEM(opt); + if (opt) FREEMEM(opt); #endif return retval; } + + +/*************************************************** +* Deprecated Functions +***************************************************/ + +/* These functions currently generate deprecation warnings */ + +/* Wrappers for deprecated compression functions */ +int LZ4_compressHC(const char* src, char* dst, int srcSize) { return LZ4_compress_HC (src, dst, srcSize, LZ4_compressBound(srcSize), 0); } +int LZ4_compressHC_limitedOutput(const char* src, char* dst, int srcSize, int maxDstSize) { return LZ4_compress_HC(src, dst, srcSize, maxDstSize, 0); } +int LZ4_compressHC2(const char* src, char* dst, int srcSize, int cLevel) { return LZ4_compress_HC (src, dst, srcSize, LZ4_compressBound(srcSize), cLevel); } +int LZ4_compressHC2_limitedOutput(const char* src, char* dst, int srcSize, int maxDstSize, int cLevel) { return LZ4_compress_HC(src, dst, srcSize, maxDstSize, cLevel); } +int LZ4_compressHC_withStateHC (void* state, const char* src, char* dst, int srcSize) { return LZ4_compress_HC_extStateHC (state, src, dst, srcSize, LZ4_compressBound(srcSize), 0); } +int LZ4_compressHC_limitedOutput_withStateHC (void* state, const char* src, char* dst, int srcSize, int maxDstSize) { return LZ4_compress_HC_extStateHC (state, src, dst, srcSize, maxDstSize, 0); } +int LZ4_compressHC2_withStateHC (void* state, const char* src, char* dst, int srcSize, int cLevel) { return LZ4_compress_HC_extStateHC(state, src, dst, srcSize, LZ4_compressBound(srcSize), cLevel); } +int LZ4_compressHC2_limitedOutput_withStateHC (void* state, const char* src, char* dst, int srcSize, int maxDstSize, int cLevel) { return LZ4_compress_HC_extStateHC(state, src, dst, srcSize, maxDstSize, cLevel); } +int LZ4_compressHC_continue (LZ4_streamHC_t* ctx, const char* src, char* dst, int srcSize) { return LZ4_compress_HC_continue (ctx, src, dst, srcSize, LZ4_compressBound(srcSize)); } +int LZ4_compressHC_limitedOutput_continue (LZ4_streamHC_t* ctx, const char* src, char* dst, int srcSize, int maxDstSize) { return LZ4_compress_HC_continue (ctx, src, dst, srcSize, maxDstSize); } + + +/* Deprecated streaming functions */ +int LZ4_sizeofStreamStateHC(void) { return sizeof(LZ4_streamHC_t); } + +/* state is presumed correctly sized, aka >= sizeof(LZ4_streamHC_t) + * @return : 0 on success, !=0 if error */ +int LZ4_resetStreamStateHC(void* state, char* inputBuffer) +{ + LZ4_streamHC_t* const hc4 = LZ4_initStreamHC(state, sizeof(*hc4)); + if (hc4 == NULL) return 1; /* init failed */ + LZ4HC_init_internal (&hc4->internal_donotuse, (const BYTE*)inputBuffer); + return 0; +} + +#if !defined(LZ4_STATIC_LINKING_ONLY_DISABLE_MEMORY_ALLOCATION) +void* LZ4_createHC (const char* inputBuffer) +{ + LZ4_streamHC_t* const hc4 = LZ4_createStreamHC(); + if (hc4 == NULL) return NULL; /* not enough memory */ + LZ4HC_init_internal (&hc4->internal_donotuse, (const BYTE*)inputBuffer); + return hc4; +} + +int LZ4_freeHC (void* LZ4HC_Data) +{ + if (!LZ4HC_Data) return 0; /* support free on NULL */ + FREEMEM(LZ4HC_Data); + return 0; +} +#endif + +int LZ4_compressHC2_continue (void* LZ4HC_Data, const char* src, char* dst, int srcSize, int cLevel) +{ + return LZ4HC_compress_generic (&((LZ4_streamHC_t*)LZ4HC_Data)->internal_donotuse, src, dst, &srcSize, 0, cLevel, notLimited); +} + +int LZ4_compressHC2_limitedOutput_continue (void* LZ4HC_Data, const char* src, char* dst, int srcSize, int dstCapacity, int cLevel) +{ + return LZ4HC_compress_generic (&((LZ4_streamHC_t*)LZ4HC_Data)->internal_donotuse, src, dst, &srcSize, dstCapacity, cLevel, limitedOutput); +} + +char* LZ4_slideInputBufferHC(void* LZ4HC_Data) +{ + LZ4HC_CCtx_internal* const s = &((LZ4_streamHC_t*)LZ4HC_Data)->internal_donotuse; + const BYTE* const bufferStart = s->prefixStart - s->dictLimit + s->lowLimit; + LZ4_resetStreamHC_fast((LZ4_streamHC_t*)LZ4HC_Data, s->compressionLevel); + /* ugly conversion trick, required to evade (const char*) -> (char*) cast-qual warning :( */ + return (char*)(uptrval)bufferStart; +} diff --git a/contrib/libs/lz4/lz4hc.h b/contrib/libs/lz4/lz4hc.h index e937acfefd..992bc8cdd7 100644 --- a/contrib/libs/lz4/lz4hc.h +++ b/contrib/libs/lz4/lz4hc.h @@ -44,7 +44,7 @@ extern "C" { /* --- Useful constants --- */ -#define LZ4HC_CLEVEL_MIN 3 +#define LZ4HC_CLEVEL_MIN 2 #define LZ4HC_CLEVEL_DEFAULT 9 #define LZ4HC_CLEVEL_OPT_MIN 10 #define LZ4HC_CLEVEL_MAX 12 @@ -126,6 +126,8 @@ LZ4LIB_API int LZ4_freeStreamHC (LZ4_streamHC_t* streamHCPtr); After reset, a first "fictional block" can be designated as initial dictionary, using LZ4_loadDictHC() (Optional). + Note: In order for LZ4_loadDictHC() to create the correct data structure, + it is essential to set the compression level _before_ loading the dictionary. Invoke LZ4_compress_HC_continue() to compress each successive block. The number of blocks is unlimited. @@ -135,12 +137,12 @@ LZ4LIB_API int LZ4_freeStreamHC (LZ4_streamHC_t* streamHCPtr); It's allowed to update compression level anytime between blocks, using LZ4_setCompressionLevel() (experimental). - 'dst' buffer should be sized to handle worst case scenarios + @dst buffer should be sized to handle worst case scenarios (see LZ4_compressBound(), it ensures compression success). In case of failure, the API does not guarantee recovery, so the state _must_ be reset. To ensure compression success - whenever `dst` buffer size cannot be made >= LZ4_compressBound(), + whenever @dst buffer size cannot be made >= LZ4_compressBound(), consider using LZ4_compress_HC_continue_destSize(). Whenever previous input blocks can't be preserved unmodified in-place during compression of next blocks, @@ -176,6 +178,34 @@ LZ4LIB_API int LZ4_compress_HC_continue_destSize(LZ4_streamHC_t* LZ4_streamHCPtr LZ4LIB_API int LZ4_saveDictHC (LZ4_streamHC_t* streamHCPtr, char* safeBuffer, int maxDictSize); +/*! LZ4_attach_HC_dictionary() : stable since v1.10.0 + * This API allows for the efficient re-use of a static dictionary many times. + * + * Rather than re-loading the dictionary buffer into a working context before + * each compression, or copying a pre-loaded dictionary's LZ4_streamHC_t into a + * working LZ4_streamHC_t, this function introduces a no-copy setup mechanism, + * in which the working stream references the dictionary stream in-place. + * + * Several assumptions are made about the state of the dictionary stream. + * Currently, only streams which have been prepared by LZ4_loadDictHC() should + * be expected to work. + * + * Alternatively, the provided dictionary stream pointer may be NULL, in which + * case any existing dictionary stream is unset. + * + * A dictionary should only be attached to a stream without any history (i.e., + * a stream that has just been reset). + * + * The dictionary will remain attached to the working stream only for the + * current stream session. Calls to LZ4_resetStreamHC(_fast) will remove the + * dictionary context association from the working stream. The dictionary + * stream (and source buffer) must remain in-place / accessible / unchanged + * through the lifetime of the stream session. + */ +LZ4LIB_API void +LZ4_attach_HC_dictionary(LZ4_streamHC_t* working_stream, + const LZ4_streamHC_t* dictionary_stream); + /*^********************************************** * !!!!!! STATIC LINKING ONLY !!!!!! @@ -204,18 +234,18 @@ LZ4LIB_API int LZ4_saveDictHC (LZ4_streamHC_t* streamHCPtr, char* safeBuffer, in typedef struct LZ4HC_CCtx_internal LZ4HC_CCtx_internal; struct LZ4HC_CCtx_internal { - LZ4_u32 hashTable[LZ4HC_HASHTABLESIZE]; - LZ4_u16 chainTable[LZ4HC_MAXD]; - const LZ4_byte* end; /* next block here to continue on current prefix */ + LZ4_u32 hashTable[LZ4HC_HASHTABLESIZE]; + LZ4_u16 chainTable[LZ4HC_MAXD]; + const LZ4_byte* end; /* next block here to continue on current prefix */ const LZ4_byte* prefixStart; /* Indexes relative to this position */ const LZ4_byte* dictStart; /* alternate reference for extDict */ - LZ4_u32 dictLimit; /* below that point, need extDict */ - LZ4_u32 lowLimit; /* below that point, no more dict */ - LZ4_u32 nextToUpdate; /* index from which to continue dictionary update */ - short compressionLevel; - LZ4_i8 favorDecSpeed; /* favor decompression speed if this flag set, - otherwise, favor compression ratio */ - LZ4_i8 dirty; /* stream has to be fully reset if this flag is set */ + LZ4_u32 dictLimit; /* below that point, need extDict */ + LZ4_u32 lowLimit; /* below that point, no more history */ + LZ4_u32 nextToUpdate; /* index from which to continue dictionary update */ + short compressionLevel; + LZ4_i8 favorDecSpeed; /* favor decompression speed if this flag set, + otherwise, favor compression ratio */ + LZ4_i8 dirty; /* stream has to be fully reset if this flag is set */ const LZ4HC_CCtx_internal* dictCtx; }; @@ -376,35 +406,6 @@ LZ4LIB_STATIC_API int LZ4_compress_HC_extStateHC_fastReset ( int srcSize, int dstCapacity, int compressionLevel); -/*! LZ4_attach_HC_dictionary() : - * This is an experimental API that allows for the efficient use of a - * static dictionary many times. - * - * Rather than re-loading the dictionary buffer into a working context before - * each compression, or copying a pre-loaded dictionary's LZ4_streamHC_t into a - * working LZ4_streamHC_t, this function introduces a no-copy setup mechanism, - * in which the working stream references the dictionary stream in-place. - * - * Several assumptions are made about the state of the dictionary stream. - * Currently, only streams which have been prepared by LZ4_loadDictHC() should - * be expected to work. - * - * Alternatively, the provided dictionary stream pointer may be NULL, in which - * case any existing dictionary stream is unset. - * - * A dictionary should only be attached to a stream without any history (i.e., - * a stream that has just been reset). - * - * The dictionary will remain attached to the working stream only for the - * current stream session. Calls to LZ4_resetStreamHC(_fast) will remove the - * dictionary context association from the working stream. The dictionary - * stream (and source buffer) must remain in-place / accessible / unchanged - * through the lifetime of the stream session. - */ -LZ4LIB_STATIC_API void LZ4_attach_HC_dictionary( - LZ4_streamHC_t *working_stream, - const LZ4_streamHC_t *dictionary_stream); - #if defined (__cplusplus) } #endif diff --git a/contrib/libs/lz4/ya.make b/contrib/libs/lz4/ya.make index 78477250e2..8c8428bca6 100644 --- a/contrib/libs/lz4/ya.make +++ b/contrib/libs/lz4/ya.make @@ -6,9 +6,9 @@ LICENSE(BSD-2-Clause) LICENSE_TEXTS(.yandex_meta/licenses.list.txt) -VERSION(1.9.4) +VERSION(1.10.0) -ORIGINAL_SOURCE(https://github.com/lz4/lz4/archive/v1.9.4.tar.gz) +ORIGINAL_SOURCE(https://github.com/lz4/lz4/archive/v1.10.0.tar.gz) PEERDIR( contrib/libs/xxhash |