diff --git a/lib/compress/zstd_compress_internal.h b/lib/compress/zstd_compress_internal.h index fefa8aff..579bd5d4 100644 --- a/lib/compress/zstd_compress_internal.h +++ b/lib/compress/zstd_compress_internal.h @@ -344,8 +344,9 @@ MEM_STATIC size_t ZSTD_minGain(size_t srcSize, ZSTD_strategy strat) * Store a sequence (litlen, litPtr, offCode and mlBase) into seqStore_t. * `offCode` : distance to match + ZSTD_REP_MOVE (values <= ZSTD_REP_MOVE are repCodes). * `mlBase` : matchLength - MINMATCH + * Allowed to overread literals up to litLimit. */ -MEM_STATIC void ZSTD_storeSeq(seqStore_t* seqStorePtr, size_t litLength, const void* literals, U32 offCode, size_t mlBase) +MEM_STATIC void ZSTD_storeSeq(seqStore_t* seqStorePtr, size_t litLength, const BYTE* literals, const BYTE* litLimit, U32 offCode, size_t mlBase) { #if defined(DEBUGLEVEL) && (DEBUGLEVEL >= 6) static const BYTE* g_start = NULL; @@ -362,7 +363,11 @@ MEM_STATIC void ZSTD_storeSeq(seqStore_t* seqStorePtr, size_t litLength, const v /* We are guaranteed at least 8 bytes of literals space because of HASH_READ_SIZE and * MINMATCH. */ - ZSTD_wildcopy8(seqStorePtr->lit, literals, (ptrdiff_t)litLength); + assert(litLimit - literals >= HASH_READ_SIZE + MINMATCH); + if (litLimit - literals >= WILDCOPY_OVERLENGTH) + ZSTD_wildcopy(seqStorePtr->lit, literals, (ptrdiff_t)litLength, ZSTD_no_overlap); + else + ZSTD_wildcopy8(seqStorePtr->lit, literals, (ptrdiff_t)litLength); seqStorePtr->lit += litLength; /* literal Length */ diff --git a/lib/compress/zstd_double_fast.c b/lib/compress/zstd_double_fast.c index 54467cc3..a661a485 100644 --- a/lib/compress/zstd_double_fast.c +++ b/lib/compress/zstd_double_fast.c @@ -148,7 +148,7 @@ size_t ZSTD_compressBlock_doubleFast_generic( const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend; mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4; ip++; - ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, 0, mLength-MINMATCH); + ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH); goto _match_stored; } @@ -157,7 +157,7 @@ size_t ZSTD_compressBlock_doubleFast_generic( && ((offset_1 > 0) & (MEM_read32(ip+1-offset_1) == MEM_read32(ip+1)))) { mLength = ZSTD_count(ip+1+4, ip+1+4-offset_1, iend) + 4; ip++; - ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, 0, mLength-MINMATCH); + ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH); goto _match_stored; } @@ -247,7 +247,7 @@ _match_found: offset_2 = offset_1; offset_1 = offset; - ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH); + ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH); _match_stored: /* match found */ @@ -278,7 +278,7 @@ _match_stored: const BYTE* const repEnd2 = repIndex2 < prefixLowestIndex ? dictEnd : iend; size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixLowest) + 4; U32 tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */ - ZSTD_storeSeq(seqStore, 0, anchor, 0, repLength2-MINMATCH); + ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, repLength2-MINMATCH); hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = current2; hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = current2; ip += repLength2; @@ -297,7 +297,7 @@ _match_stored: U32 const tmpOff = offset_2; offset_2 = offset_1; offset_1 = tmpOff; /* swap offset_2 <=> offset_1 */ hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = (U32)(ip-base); hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = (U32)(ip-base); - ZSTD_storeSeq(seqStore, 0, anchor, 0, rLength-MINMATCH); + ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, rLength-MINMATCH); ip += rLength; anchor = ip; continue; /* faster when present ... (?) */ @@ -411,7 +411,7 @@ static size_t ZSTD_compressBlock_doubleFast_extDict_generic( const BYTE* repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend; mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixStart) + 4; ip++; - ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, 0, mLength-MINMATCH); + ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH); } else { if ((matchLongIndex > dictStartIndex) && (MEM_read64(matchLong) == MEM_read64(ip))) { const BYTE* const matchEnd = matchLongIndex < prefixStartIndex ? dictEnd : iend; @@ -422,7 +422,7 @@ static size_t ZSTD_compressBlock_doubleFast_extDict_generic( while (((ip>anchor) & (matchLong>lowMatchPtr)) && (ip[-1] == matchLong[-1])) { ip--; matchLong--; mLength++; } /* catch up */ offset_2 = offset_1; offset_1 = offset; - ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH); + ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH); } else if ((matchIndex > dictStartIndex) && (MEM_read32(match) == MEM_read32(ip))) { size_t const h3 = ZSTD_hashPtr(ip+1, hBitsL, 8); @@ -447,7 +447,7 @@ static size_t ZSTD_compressBlock_doubleFast_extDict_generic( } offset_2 = offset_1; offset_1 = offset; - ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH); + ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH); } else { ip += ((ip-anchor) >> kSearchStrength) + 1; @@ -479,7 +479,7 @@ static size_t ZSTD_compressBlock_doubleFast_extDict_generic( const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend; size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4; U32 const tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */ - ZSTD_storeSeq(seqStore, 0, anchor, 0, repLength2-MINMATCH); + ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, repLength2-MINMATCH); hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = current2; hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = current2; ip += repLength2; diff --git a/lib/compress/zstd_fast.c b/lib/compress/zstd_fast.c index 75aec8a3..6dbefee6 100644 --- a/lib/compress/zstd_fast.c +++ b/lib/compress/zstd_fast.c @@ -136,7 +136,7 @@ _offset: /* Requires: ip0, match0 */ _match: /* Requires: ip0, match0, offcode */ /* Count the forward length */ mLength += ZSTD_count(ip0+mLength+4, match0+mLength+4, iend) + 4; - ZSTD_storeSeq(seqStore, (size_t)(ip0-anchor), anchor, offcode, mLength-MINMATCH); + ZSTD_storeSeq(seqStore, (size_t)(ip0-anchor), anchor, iend, offcode, mLength-MINMATCH); /* match found */ ip0 += mLength; anchor = ip0; @@ -156,7 +156,7 @@ _match: /* Requires: ip0, match0, offcode */ hashTable[ZSTD_hashPtr(ip0, hlog, mls)] = (U32)(ip0-base); ip0 += rLength; ip1 = ip0 + 1; - ZSTD_storeSeq(seqStore, 0 /*litLen*/, anchor, 0 /*offCode*/, rLength-MINMATCH); + ZSTD_storeSeq(seqStore, 0 /*litLen*/, anchor, iend, 0 /*offCode*/, rLength-MINMATCH); anchor = ip0; continue; /* faster when present (confirmed on gcc-8) ... (?) */ } @@ -261,7 +261,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState_generic( const BYTE* const repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend; mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixStart) + 4; ip++; - ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, 0, mLength-MINMATCH); + ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH); } else if ( (matchIndex <= prefixStartIndex) ) { size_t const dictHash = ZSTD_hashPtr(ip, dictHLog, mls); U32 const dictMatchIndex = dictHashTable[dictHash]; @@ -281,7 +281,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState_generic( } /* catch up */ offset_2 = offset_1; offset_1 = offset; - ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH); + ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH); } } else if (MEM_read32(match) != MEM_read32(ip)) { /* it's not a match, and we're not going to check the dictionary */ @@ -296,7 +296,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState_generic( && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */ offset_2 = offset_1; offset_1 = offset; - ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH); + ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH); } /* match found */ @@ -321,7 +321,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState_generic( const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend; size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4; U32 tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */ - ZSTD_storeSeq(seqStore, 0, anchor, 0, repLength2-MINMATCH); + ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, repLength2-MINMATCH); hashTable[ZSTD_hashPtr(ip, hlog, mls)] = current2; ip += repLength2; anchor = ip; @@ -411,7 +411,7 @@ static size_t ZSTD_compressBlock_fast_extDict_generic( const BYTE* const repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend; size_t const rLength = ZSTD_count_2segments(ip+1 +4, repMatch +4, iend, repMatchEnd, prefixStart) + 4; ip++; - ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, 0, rLength-MINMATCH); + ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, rLength-MINMATCH); ip += rLength; anchor = ip; } else { @@ -427,7 +427,7 @@ static size_t ZSTD_compressBlock_fast_extDict_generic( size_t mLength = ZSTD_count_2segments(ip+4, match+4, iend, matchEnd, prefixStart) + 4; while (((ip>anchor) & (match>lowMatchPtr)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */ offset_2 = offset_1; offset_1 = offset; /* update offset history */ - ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH); + ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH); ip += mLength; anchor = ip; } } @@ -446,7 +446,7 @@ static size_t ZSTD_compressBlock_fast_extDict_generic( const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend; size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4; { U32 const tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; } /* swap offset_2 <=> offset_1 */ - ZSTD_storeSeq(seqStore, 0 /*litlen*/, anchor, 0 /*offcode*/, repLength2-MINMATCH); + ZSTD_storeSeq(seqStore, 0 /*litlen*/, anchor, iend, 0 /*offcode*/, repLength2-MINMATCH); hashTable[ZSTD_hashPtr(ip, hlog, mls)] = current2; ip += repLength2; anchor = ip; diff --git a/lib/compress/zstd_lazy.c b/lib/compress/zstd_lazy.c index 0af41724..9ad7e03b 100644 --- a/lib/compress/zstd_lazy.c +++ b/lib/compress/zstd_lazy.c @@ -810,7 +810,7 @@ ZSTD_compressBlock_lazy_generic( /* store sequence */ _storeSequence: { size_t const litLength = start - anchor; - ZSTD_storeSeq(seqStore, litLength, anchor, (U32)offset, matchLength-MINMATCH); + ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offset, matchLength-MINMATCH); anchor = ip = start + matchLength; } @@ -828,7 +828,7 @@ _storeSequence: const BYTE* const repEnd2 = repIndex < prefixLowestIndex ? dictEnd : iend; matchLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd2, prefixLowest) + 4; offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap offset_2 <=> offset_1 */ - ZSTD_storeSeq(seqStore, 0, anchor, 0, matchLength-MINMATCH); + ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH); ip += matchLength; anchor = ip; continue; @@ -843,7 +843,7 @@ _storeSequence: /* store sequence */ matchLength = ZSTD_count(ip+4, ip+4-offset_2, iend) + 4; offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap repcodes */ - ZSTD_storeSeq(seqStore, 0, anchor, 0, matchLength-MINMATCH); + ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH); ip += matchLength; anchor = ip; continue; /* faster when present ... (?) */ @@ -1051,7 +1051,7 @@ size_t ZSTD_compressBlock_lazy_extDict_generic( /* store sequence */ _storeSequence: { size_t const litLength = start - anchor; - ZSTD_storeSeq(seqStore, litLength, anchor, (U32)offset, matchLength-MINMATCH); + ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offset, matchLength-MINMATCH); anchor = ip = start + matchLength; } @@ -1066,7 +1066,7 @@ _storeSequence: const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend; matchLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4; offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap offset history */ - ZSTD_storeSeq(seqStore, 0, anchor, 0, matchLength-MINMATCH); + ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH); ip += matchLength; anchor = ip; continue; /* faster when present ... (?) */ diff --git a/lib/compress/zstd_ldm.c b/lib/compress/zstd_ldm.c index 3dcf86e6..fc3f4694 100644 --- a/lib/compress/zstd_ldm.c +++ b/lib/compress/zstd_ldm.c @@ -583,7 +583,7 @@ size_t ZSTD_ldm_blockCompress(rawSeqStore_t* rawSeqStore, rep[i] = rep[i-1]; rep[0] = sequence.offset; /* Store the sequence */ - ZSTD_storeSeq(seqStore, newLitLength, ip - newLitLength, + ZSTD_storeSeq(seqStore, newLitLength, ip - newLitLength, iend, sequence.offset + ZSTD_REP_MOVE, sequence.matchLength - MINMATCH); ip += sequence.matchLength; diff --git a/lib/compress/zstd_opt.c b/lib/compress/zstd_opt.c index 2da363f9..2e50fca6 100644 --- a/lib/compress/zstd_opt.c +++ b/lib/compress/zstd_opt.c @@ -1098,7 +1098,7 @@ _shortestPath: /* cur, last_pos, best_mlen, best_off have to be set */ assert(anchor + llen <= iend); ZSTD_updateStats(optStatePtr, llen, anchor, offCode, mlen); - ZSTD_storeSeq(seqStore, llen, anchor, offCode, mlen-MINMATCH); + ZSTD_storeSeq(seqStore, llen, anchor, iend, offCode, mlen-MINMATCH); anchor += advance; ip = anchor; } } diff --git a/lib/decompress/zstd_decompress_block.c b/lib/decompress/zstd_decompress_block.c index 27ce137f..cbb66c8d 100644 --- a/lib/decompress/zstd_decompress_block.c +++ b/lib/decompress/zstd_decompress_block.c @@ -761,10 +761,10 @@ size_t ZSTD_execSequence(BYTE* op, * without overlap checking. */ if (sequence.offset >= WILDCOPY_VECLEN) { - /* Split out matchLength <= 32 since it is nearly always true. +1% on gcc-9. - * We copy 32 bytes here since matches are generally longer than literals. - * In silesia, for example ~10% of matches are longer than 16 bytes. - */ + /* We bet on a full wildcopy for matches, since we expect matches to be + * longer than literals (in general). In silesia, ~10% of matches are longer + * than 16 bytes. + */ ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength, ZSTD_no_overlap); return sequenceLength; }