zstdlib 0.9.0-x86-mingw32 → 0.11.0-x86-mingw32

data/ext/zstdlib_c/{zstd-1.5.2 → zstd-1.5.5}/lib/compress/zstd_lazy.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) Yann Collet, Facebook, Inc.
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the
@@ -25,6 +25,8 @@ extern "C" {
  */
 #define ZSTD_LAZY_DDSS_BUCKET_LOG 2
 
+#define ZSTD_ROW_HASH_TAG_BITS 8        /* nb bits to use for the tag */
+
 U32 ZSTD_insertAndFindFirstIndex(ZSTD_matchState_t* ms, const BYTE* ip);
 void ZSTD_row_update(ZSTD_matchState_t* const ms, const BYTE* ip);
 
@@ -116,7 +118,7 @@ size_t ZSTD_compressBlock_lazy2_extDict_row(
 size_t ZSTD_compressBlock_btlazy2_extDict(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
-        
+
 
 #if defined (__cplusplus)
 }

data/ext/zstdlib_c/{zstd-1.5.2 → zstd-1.5.5}/lib/compress/zstd_ldm.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) Yann Collet, Facebook, Inc.
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the
@@ -242,11 +242,11 @@ static size_t ZSTD_ldm_fillFastTables(ZSTD_matchState_t* ms,
     switch(ms->cParams.strategy)
     {
     case ZSTD_fast:
-        ZSTD_fillHashTable(ms, iend, ZSTD_dtlm_fast);
+        ZSTD_fillHashTable(ms, iend, ZSTD_dtlm_fast, ZSTD_tfp_forCCtx);
         break;
 
     case ZSTD_dfast:
-        ZSTD_fillDoubleHashTable(ms, iend, ZSTD_dtlm_fast);
+        ZSTD_fillDoubleHashTable(ms, iend, ZSTD_dtlm_fast, ZSTD_tfp_forCCtx);
         break;
 
     case ZSTD_greedy:
@@ -549,7 +549,7 @@ size_t ZSTD_ldm_generateSequences(
          * the window through early invalidation.
          * TODO: * Test the chunk size.
          *       * Try invalidation after the sequence generation and test the
-         *         the offset against maxDist directly.
+         *         offset against maxDist directly.
          *
          * NOTE: Because of dictionaries + sequence splitting we MUST make sure
          * that any offset used is valid at the END of the sequence, since it may
@@ -711,7 +711,7 @@ size_t ZSTD_ldm_blockCompress(rawSeqStore_t* rawSeqStore,
             rep[0] = sequence.offset;
             /* Store the sequence */
             ZSTD_storeSeq(seqStore, newLitLength, ip - newLitLength, iend,
-                          STORE_OFFSET(sequence.offset),
+                          OFFSET_TO_OFFBASE(sequence.offset),
                           sequence.matchLength);
             ip += sequence.matchLength;
         }

data/ext/zstdlib_c/{zstd-1.5.2 → zstd-1.5.5}/lib/compress/zstd_ldm.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) Yann Collet, Facebook, Inc.
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the

data/ext/zstdlib_c/{zstd-1.5.2 → zstd-1.5.5}/lib/compress/zstd_ldm_geartab.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) Yann Collet, Facebook, Inc.
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the

data/ext/zstdlib_c/{zstd-1.5.2 → zstd-1.5.5}/lib/compress/zstd_opt.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) Przemyslaw Skibinski, Yann Collet, Facebook, Inc.
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the
@@ -16,7 +16,7 @@
 #define ZSTD_LITFREQ_ADD    2   /* scaling factor for litFreq, so that frequencies adapt faster to new stats */
 #define ZSTD_MAX_PRICE     (1<<30)
 
-#define ZSTD_PREDEF_THRESHOLD 1024   /* if srcSize < ZSTD_PREDEF_THRESHOLD, symbols' cost is assumed static, directly determined by pre-defined distributions */
+#define ZSTD_PREDEF_THRESHOLD 8   /* if srcSize < ZSTD_PREDEF_THRESHOLD, symbols' cost is assumed static, directly determined by pre-defined distributions */
 
 
 /*-*************************************
@@ -26,27 +26,35 @@
 #if 0    /* approximation at bit level (for tests) */
 #  define BITCOST_ACCURACY 0
 #  define BITCOST_MULTIPLIER (1 << BITCOST_ACCURACY)
-#  define WEIGHT(stat, opt) ((void)opt, ZSTD_bitWeight(stat))
+#  define WEIGHT(stat, opt) ((void)(opt), ZSTD_bitWeight(stat))
 #elif 0  /* fractional bit accuracy (for tests) */
 #  define BITCOST_ACCURACY 8
 #  define BITCOST_MULTIPLIER (1 << BITCOST_ACCURACY)
-#  define WEIGHT(stat,opt) ((void)opt, ZSTD_fracWeight(stat))
+#  define WEIGHT(stat,opt) ((void)(opt), ZSTD_fracWeight(stat))
 #else    /* opt==approx, ultra==accurate */
 #  define BITCOST_ACCURACY 8
 #  define BITCOST_MULTIPLIER (1 << BITCOST_ACCURACY)
-#  define WEIGHT(stat,opt) (opt ? ZSTD_fracWeight(stat) : ZSTD_bitWeight(stat))
+#  define WEIGHT(stat,opt) ((opt) ? ZSTD_fracWeight(stat) : ZSTD_bitWeight(stat))
 #endif
 
+/* ZSTD_bitWeight() :
+ * provide estimated "cost" of a stat in full bits only */
 MEM_STATIC U32 ZSTD_bitWeight(U32 stat)
 {
     return (ZSTD_highbit32(stat+1) * BITCOST_MULTIPLIER);
 }
 
+/* ZSTD_fracWeight() :
+ * provide fractional-bit "cost" of a stat,
+ * using linear interpolation approximation */
 MEM_STATIC U32 ZSTD_fracWeight(U32 rawStat)
 {
     U32 const stat = rawStat + 1;
     U32 const hb = ZSTD_highbit32(stat);
     U32 const BWeight = hb * BITCOST_MULTIPLIER;
+    /* Fweight was meant for "Fractional weight"
+     * but it's effectively a value between 1 and 2
+     * using fixed point arithmetic */
     U32 const FWeight = (stat << BITCOST_ACCURACY) >> hb;
     U32 const weight = BWeight + FWeight;
     assert(hb + BITCOST_ACCURACY < 31);
@@ -57,7 +65,7 @@ MEM_STATIC U32 ZSTD_fracWeight(U32 rawStat)
 /* debugging function,
  * @return price in bytes as fractional value
  * for debug messages only */
-MEM_STATIC double ZSTD_fCost(U32 price)
+MEM_STATIC double ZSTD_fCost(int price)
 {
     return (double)price / (BITCOST_MULTIPLIER*8);
 }
@@ -88,20 +96,26 @@ static U32 sum_u32(const unsigned table[], size_t nbElts)
     return total;
 }
 
-static U32 ZSTD_downscaleStats(unsigned* table, U32 lastEltIndex, U32 shift)
+typedef enum { base_0possible=0, base_1guaranteed=1 } base_directive_e;
+
+static U32
+ZSTD_downscaleStats(unsigned* table, U32 lastEltIndex, U32 shift, base_directive_e base1)
 {
     U32 s, sum=0;
-    DEBUGLOG(5, "ZSTD_downscaleStats (nbElts=%u, shift=%u)", (unsigned)lastEltIndex+1, (unsigned)shift);
+    DEBUGLOG(5, "ZSTD_downscaleStats (nbElts=%u, shift=%u)",
+            (unsigned)lastEltIndex+1, (unsigned)shift );
     assert(shift < 30);
     for (s=0; s<lastEltIndex+1; s++) {
-        table[s] = 1 + (table[s] >> shift);
-        sum += table[s];
+        unsigned const base = base1 ? 1 : (table[s]>0);
+        unsigned const newStat = base + (table[s] >> shift);
+        sum += newStat;
+        table[s] = newStat;
     }
     return sum;
 }
 
 /* ZSTD_scaleStats() :
- * reduce all elements in table is sum too large
+ * reduce all elt frequencies in table if sum too large
  * return the resulting sum of elements */
 static U32 ZSTD_scaleStats(unsigned* table, U32 lastEltIndex, U32 logTarget)
 {
@@ -110,7 +124,7 @@ static U32 ZSTD_scaleStats(unsigned* table, U32 lastEltIndex, U32 logTarget)
     DEBUGLOG(5, "ZSTD_scaleStats (nbElts=%u, target=%u)", (unsigned)lastEltIndex+1, (unsigned)logTarget);
     assert(logTarget < 30);
     if (factor <= 1) return prevsum;
-    return ZSTD_downscaleStats(table, lastEltIndex, ZSTD_highbit32(factor));
+    return ZSTD_downscaleStats(table, lastEltIndex, ZSTD_highbit32(factor), base_1guaranteed);
 }
 
 /* ZSTD_rescaleFreqs() :
@@ -129,18 +143,22 @@ ZSTD_rescaleFreqs(optState_t* const optPtr,
     DEBUGLOG(5, "ZSTD_rescaleFreqs (srcSize=%u)", (unsigned)srcSize);
     optPtr->priceType = zop_dynamic;
 
-    if (optPtr->litLengthSum == 0) {  /* first block : init */
-        if (srcSize <= ZSTD_PREDEF_THRESHOLD) {  /* heuristic */
-            DEBUGLOG(5, "(srcSize <= ZSTD_PREDEF_THRESHOLD) => zop_predef");
+    if (optPtr->litLengthSum == 0) {  /* no literals stats collected -> first block assumed -> init */
+
+        /* heuristic: use pre-defined stats for too small inputs */
+        if (srcSize <= ZSTD_PREDEF_THRESHOLD) {
+            DEBUGLOG(5, "srcSize <= %i : use predefined stats", ZSTD_PREDEF_THRESHOLD);
             optPtr->priceType = zop_predef;
         }
 
         assert(optPtr->symbolCosts != NULL);
         if (optPtr->symbolCosts->huf.repeatMode == HUF_repeat_valid) {
-            /* huffman table presumed generated by dictionary */
+
+            /* huffman stats covering the full value set : table presumed generated by dictionary */
             optPtr->priceType = zop_dynamic;
 
             if (compressedLiterals) {
+                /* generate literals statistics from huffman table */
                 unsigned lit;
                 assert(optPtr->litFreq != NULL);
                 optPtr->litSum = 0;
@@ -188,13 +206,14 @@ ZSTD_rescaleFreqs(optState_t* const optPtr,
                     optPtr->offCodeSum += optPtr->offCodeFreq[of];
             }   }
 
-        } else {  /* not a dictionary */
+        } else {  /* first block, no dictionary */
 
             assert(optPtr->litFreq != NULL);
             if (compressedLiterals) {
+                /* base initial cost of literals on direct frequency within src */
                 unsigned lit = MaxLit;
                 HIST_count_simple(optPtr->litFreq, &lit, src, srcSize);   /* use raw first block to init statistics */
-                optPtr->litSum = ZSTD_downscaleStats(optPtr->litFreq, MaxLit, 8);
+                optPtr->litSum = ZSTD_downscaleStats(optPtr->litFreq, MaxLit, 8, base_0possible);
             }
 
             {   unsigned const baseLLfreqs[MaxLL+1] = {
@@ -224,10 +243,9 @@ ZSTD_rescaleFreqs(optState_t* const optPtr,
                 optPtr->offCodeSum = sum_u32(baseOFCfreqs, MaxOff+1);
             }
 
-
         }
 
-    } else {   /* new block : re-use previous statistics, scaled down */
+    } else {   /* new block : scale down accumulated statistics */
 
         if (compressedLiterals)
             optPtr->litSum = ZSTD_scaleStats(optPtr->litFreq, MaxLit, 12);
@@ -255,11 +273,14 @@ static U32 ZSTD_rawLiteralsCost(const BYTE* const literals, U32 const litLength,
         return (litLength*6) * BITCOST_MULTIPLIER;  /* 6 bit per literal - no statistic used */
 
     /* dynamic statistics */
-    {   U32 price = litLength * optPtr->litSumBasePrice;
+    {   U32 price = optPtr->litSumBasePrice * litLength;
+        U32 const litPriceMax = optPtr->litSumBasePrice - BITCOST_MULTIPLIER;
         U32 u;
+        assert(optPtr->litSumBasePrice >= BITCOST_MULTIPLIER);
         for (u=0; u < litLength; u++) {
-            assert(WEIGHT(optPtr->litFreq[literals[u]], optLevel) <= optPtr->litSumBasePrice);   /* literal cost should never be negative */
-            price -= WEIGHT(optPtr->litFreq[literals[u]], optLevel);
+            U32 litPrice = WEIGHT(optPtr->litFreq[literals[u]], optLevel);
+            if (UNLIKELY(litPrice > litPriceMax)) litPrice = litPriceMax;
+            price -= litPrice;
         }
         return price;
     }
@@ -272,10 +293,11 @@ static U32 ZSTD_litLengthPrice(U32 const litLength, const optState_t* const optP
     assert(litLength <= ZSTD_BLOCKSIZE_MAX);
     if (optPtr->priceType == zop_predef)
         return WEIGHT(litLength, optLevel);
-    /* We can't compute the litLength price for sizes >= ZSTD_BLOCKSIZE_MAX
-     * because it isn't representable in the zstd format. So instead just
-     * call it 1 bit more than ZSTD_BLOCKSIZE_MAX - 1. In this case the block
-     * would be all literals.
+
+    /* ZSTD_LLcode() can't compute litLength price for sizes >= ZSTD_BLOCKSIZE_MAX
+     * because it isn't representable in the zstd format.
+     * So instead just pretend it would cost 1 bit more than ZSTD_BLOCKSIZE_MAX - 1.
+     * In such a case, the block would be all literals.
      */
     if (litLength == ZSTD_BLOCKSIZE_MAX)
         return BITCOST_MULTIPLIER + ZSTD_litLengthPrice(ZSTD_BLOCKSIZE_MAX - 1, optPtr, optLevel);
@@ -289,24 +311,25 @@ static U32 ZSTD_litLengthPrice(U32 const litLength, const optState_t* const optP
 }
 
 /* ZSTD_getMatchPrice() :
- * Provides the cost of the match part (offset + matchLength) of a sequence
+ * Provides the cost of the match part (offset + matchLength) of a sequence.
  * Must be combined with ZSTD_fullLiteralsCost() to get the full cost of a sequence.
- * @offcode : expects a scale where 0,1,2 are repcodes 1-3, and 3+ are real_offsets+2
+ * @offBase : sumtype, representing an offset or a repcode, and using numeric representation of ZSTD_storeSeq()
  * @optLevel: when <2, favors small offset for decompression speed (improved cache efficiency)
  */
 FORCE_INLINE_TEMPLATE U32
-ZSTD_getMatchPrice(U32 const offcode,
+ZSTD_getMatchPrice(U32 const offBase,
                    U32 const matchLength,
              const optState_t* const optPtr,
                    int const optLevel)
 {
     U32 price;
-    U32 const offCode = ZSTD_highbit32(STORED_TO_OFFBASE(offcode));
+    U32 const offCode = ZSTD_highbit32(offBase);
     U32 const mlBase = matchLength - MINMATCH;
     assert(matchLength >= MINMATCH);
 
-    if (optPtr->priceType == zop_predef)  /* fixed scheme, do not use statistics */
-        return WEIGHT(mlBase, optLevel) + ((16 + offCode) * BITCOST_MULTIPLIER);
+    if (optPtr->priceType == zop_predef)  /* fixed scheme, does not use statistics */
+        return WEIGHT(mlBase, optLevel)
+             + ((16 + offCode) * BITCOST_MULTIPLIER); /* emulated offset cost */
 
     /* dynamic statistics */
     price = (offCode * BITCOST_MULTIPLIER) + (optPtr->offCodeSumBasePrice - WEIGHT(optPtr->offCodeFreq[offCode], optLevel));
@@ -325,10 +348,10 @@ ZSTD_getMatchPrice(U32 const offcode,
 }
 
 /* ZSTD_updateStats() :
- * assumption : literals + litLengtn <= iend */
+ * assumption : literals + litLength <= iend */
 static void ZSTD_updateStats(optState_t* const optPtr,
                              U32 litLength, const BYTE* literals,
-                             U32 offsetCode, U32 matchLength)
+                             U32 offBase, U32 matchLength)
 {
     /* literals */
     if (ZSTD_compressedLiterals(optPtr)) {
@@ -344,8 +367,8 @@ static void ZSTD_updateStats(optState_t* const optPtr,
         optPtr->litLengthSum++;
     }
 
-    /* offset code : expected to follow storeSeq() numeric representation */
-    {   U32 const offCode = ZSTD_highbit32(STORED_TO_OFFBASE(offsetCode));
+    /* offset code : follows storeSeq() numeric representation */
+    {   U32 const offCode = ZSTD_highbit32(offBase);
         assert(offCode <= MaxOff);
         optPtr->offCodeFreq[offCode]++;
         optPtr->offCodeSum++;
@@ -552,16 +575,17 @@ void ZSTD_updateTree(ZSTD_matchState_t* ms, const BYTE* ip, const BYTE* iend) {
     ZSTD_updateTree_internal(ms, ip, iend, ms->cParams.minMatch, ZSTD_noDict);
 }
 
-FORCE_INLINE_TEMPLATE
-U32 ZSTD_insertBtAndGetAllMatches (
-                    ZSTD_match_t* matches,   /* store result (found matches) in this table (presumed large enough) */
-                    ZSTD_matchState_t* ms,
-                    U32* nextToUpdate3,
-                    const BYTE* const ip, const BYTE* const iLimit, const ZSTD_dictMode_e dictMode,
-                    const U32 rep[ZSTD_REP_NUM],
-                    U32 const ll0,   /* tells if associated literal length is 0 or not. This value must be 0 or 1 */
-                    const U32 lengthToBeat,
-                    U32 const mls /* template */)
+FORCE_INLINE_TEMPLATE U32
+ZSTD_insertBtAndGetAllMatches (
+                ZSTD_match_t* matches,  /* store result (found matches) in this table (presumed large enough) */
+                ZSTD_matchState_t* ms,
+                U32* nextToUpdate3,
+                const BYTE* const ip, const BYTE* const iLimit,
+                const ZSTD_dictMode_e dictMode,
+                const U32 rep[ZSTD_REP_NUM],
+                const U32 ll0,  /* tells if associated literal length is 0 or not. This value must be 0 or 1 */
+                const U32 lengthToBeat,
+                const U32 mls /* template */)
 {
     const ZSTD_compressionParameters* const cParams = &ms->cParams;
     U32 const sufficient_len = MIN(cParams->targetLength, ZSTD_OPT_NUM -1);
@@ -644,7 +668,7 @@ U32 ZSTD_insertBtAndGetAllMatches (
                 DEBUGLOG(8, "found repCode %u (ll0:%u, offset:%u) of length %u",
                             repCode, ll0, repOffset, repLen);
                 bestLength = repLen;
-                matches[mnum].off = STORE_REPCODE(repCode - ll0 + 1);  /* expect value between 1 and 3 */
+                matches[mnum].off = REPCODE_TO_OFFBASE(repCode - ll0 + 1);  /* expect value between 1 and 3 */
                 matches[mnum].len = (U32)repLen;
                 mnum++;
                 if ( (repLen > sufficient_len)
@@ -673,7 +697,7 @@ U32 ZSTD_insertBtAndGetAllMatches (
                 bestLength = mlen;
                 assert(curr > matchIndex3);
                 assert(mnum==0);  /* no prior solution */
-                matches[0].off = STORE_OFFSET(curr - matchIndex3);
+                matches[0].off = OFFSET_TO_OFFBASE(curr - matchIndex3);
                 matches[0].len = (U32)mlen;
                 mnum = 1;
                 if ( (mlen > sufficient_len) |
@@ -706,13 +730,13 @@ U32 ZSTD_insertBtAndGetAllMatches (
         }
 
         if (matchLength > bestLength) {
-            DEBUGLOG(8, "found match of length %u at distance %u (offCode=%u)",
-                    (U32)matchLength, curr - matchIndex, STORE_OFFSET(curr - matchIndex));
+            DEBUGLOG(8, "found match of length %u at distance %u (offBase=%u)",
+                    (U32)matchLength, curr - matchIndex, OFFSET_TO_OFFBASE(curr - matchIndex));
             assert(matchEndIdx > matchIndex);
             if (matchLength > matchEndIdx - matchIndex)
                 matchEndIdx = matchIndex + (U32)matchLength;
             bestLength = matchLength;
-            matches[mnum].off = STORE_OFFSET(curr - matchIndex);
+            matches[mnum].off = OFFSET_TO_OFFBASE(curr - matchIndex);
             matches[mnum].len = (U32)matchLength;
             mnum++;
             if ( (matchLength > ZSTD_OPT_NUM)
@@ -754,12 +778,12 @@ U32 ZSTD_insertBtAndGetAllMatches (
 
             if (matchLength > bestLength) {
                 matchIndex = dictMatchIndex + dmsIndexDelta;
-                DEBUGLOG(8, "found dms match of length %u at distance %u (offCode=%u)",
-                        (U32)matchLength, curr - matchIndex, STORE_OFFSET(curr - matchIndex));
+                DEBUGLOG(8, "found dms match of length %u at distance %u (offBase=%u)",
+                        (U32)matchLength, curr - matchIndex, OFFSET_TO_OFFBASE(curr - matchIndex));
                 if (matchLength > matchEndIdx - matchIndex)
                     matchEndIdx = matchIndex + (U32)matchLength;
                 bestLength = matchLength;
-                matches[mnum].off = STORE_OFFSET(curr - matchIndex);
+                matches[mnum].off = OFFSET_TO_OFFBASE(curr - matchIndex);
                 matches[mnum].len = (U32)matchLength;
                 mnum++;
                 if ( (matchLength > ZSTD_OPT_NUM)
@@ -960,7 +984,7 @@ static void ZSTD_optLdm_maybeAddMatch(ZSTD_match_t* matches, U32* nbMatches,
                                       const ZSTD_optLdm_t* optLdm, U32 currPosInBlock)
 {
     U32 const posDiff = currPosInBlock - optLdm->startPosInBlock;
-    /* Note: ZSTD_match_t actually contains offCode and matchLength (before subtracting MINMATCH) */
+    /* Note: ZSTD_match_t actually contains offBase and matchLength (before subtracting MINMATCH) */
     U32 const candidateMatchLength = optLdm->endPosInBlock - optLdm->startPosInBlock - posDiff;
 
     /* Ensure that current block position is not outside of the match */
@@ -971,11 +995,11 @@ static void ZSTD_optLdm_maybeAddMatch(ZSTD_match_t* matches, U32* nbMatches,
     }
 
     if (*nbMatches == 0 || ((candidateMatchLength > matches[*nbMatches-1].len) && *nbMatches < ZSTD_OPT_NUM)) {
-        U32 const candidateOffCode = STORE_OFFSET(optLdm->offset);
-        DEBUGLOG(6, "ZSTD_optLdm_maybeAddMatch(): Adding ldm candidate match (offCode: %u matchLength %u) at block position=%u",
-                 candidateOffCode, candidateMatchLength, currPosInBlock);
+        U32 const candidateOffBase = OFFSET_TO_OFFBASE(optLdm->offset);
+        DEBUGLOG(6, "ZSTD_optLdm_maybeAddMatch(): Adding ldm candidate match (offBase: %u matchLength %u) at block position=%u",
+                 candidateOffBase, candidateMatchLength, currPosInBlock);
         matches[*nbMatches].len = candidateMatchLength;
-        matches[*nbMatches].off = candidateOffCode;
+        matches[*nbMatches].off = candidateOffBase;
         (*nbMatches)++;
     }
 }
@@ -1062,6 +1086,8 @@ ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
     ZSTD_optimal_t lastSequence;
     ZSTD_optLdm_t optLdm;
 
+    ZSTD_memset(&lastSequence, 0, sizeof(ZSTD_optimal_t));
+
     optLdm.seqStore = ms->ldmSeqStore ? *ms->ldmSeqStore : kNullRawSeqStore;
     optLdm.endPosInBlock = optLdm.startPosInBlock = optLdm.offset = 0;
     ZSTD_opt_getNextMatchAndUpdateSeqStore(&optLdm, (U32)(ip-istart), (U32)(iend-ip));
@@ -1098,14 +1124,14 @@ ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
 
             /* large match -> immediate encoding */
             {   U32 const maxML = matches[nbMatches-1].len;
-                U32 const maxOffcode = matches[nbMatches-1].off;
-                DEBUGLOG(6, "found %u matches of maxLength=%u and maxOffCode=%u at cPos=%u => start new series",
-                            nbMatches, maxML, maxOffcode, (U32)(ip-prefixStart));
+                U32 const maxOffBase = matches[nbMatches-1].off;
+                DEBUGLOG(6, "found %u matches of maxLength=%u and maxOffBase=%u at cPos=%u => start new series",
+                            nbMatches, maxML, maxOffBase, (U32)(ip-prefixStart));
 
                 if (maxML > sufficient_len) {
                     lastSequence.litlen = litlen;
                     lastSequence.mlen = maxML;
-                    lastSequence.off = maxOffcode;
+                    lastSequence.off = maxOffBase;
                     DEBUGLOG(6, "large match (%u>%u), immediate encoding",
                                 maxML, sufficient_len);
                     cur = 0;
@@ -1122,15 +1148,15 @@ ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
                     opt[pos].price = ZSTD_MAX_PRICE;   /* mlen, litlen and price will be fixed during forward scanning */
                 }
                 for (matchNb = 0; matchNb < nbMatches; matchNb++) {
-                    U32 const offcode = matches[matchNb].off;
+                    U32 const offBase = matches[matchNb].off;
                     U32 const end = matches[matchNb].len;
                     for ( ; pos <= end ; pos++ ) {
-                        U32 const matchPrice = ZSTD_getMatchPrice(offcode, pos, optStatePtr, optLevel);
+                        U32 const matchPrice = ZSTD_getMatchPrice(offBase, pos, optStatePtr, optLevel);
                         U32 const sequencePrice = literalsPrice + matchPrice;
                         DEBUGLOG(7, "rPos:%u => set initial price : %.2f",
-                                    pos, ZSTD_fCost(sequencePrice));
+                                    pos, ZSTD_fCost((int)sequencePrice));
                         opt[pos].mlen = pos;
-                        opt[pos].off = offcode;
+                        opt[pos].off = offBase;
                         opt[pos].litlen = litlen;
                         opt[pos].price = (int)sequencePrice;
                 }   }
@@ -1230,7 +1256,7 @@ ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
                     U32 const startML = (matchNb>0) ? matches[matchNb-1].len+1 : minMatch;
                     U32 mlen;
 
-                    DEBUGLOG(7, "testing match %u => offCode=%4u, mlen=%2u, llen=%2u",
+                    DEBUGLOG(7, "testing match %u => offBase=%4u, mlen=%2u, llen=%2u",
                                 matchNb, matches[matchNb].off, lastML, litlen);
 
                     for (mlen = lastML; mlen >= startML; mlen--) {  /* scan downward */
@@ -1296,7 +1322,7 @@ _shortestPath:   /* cur, last_pos, best_mlen, best_off have to be set */
                 for (storePos=storeStart; storePos <= storeEnd; storePos++) {
                     U32 const llen = opt[storePos].litlen;
                     U32 const mlen = opt[storePos].mlen;
-                    U32 const offCode = opt[storePos].off;
+                    U32 const offBase = opt[storePos].off;
                     U32 const advance = llen + mlen;
                     DEBUGLOG(6, "considering seq starting at %zi, llen=%u, mlen=%u",
                                 anchor - istart, (unsigned)llen, (unsigned)mlen);
@@ -1308,8 +1334,8 @@ _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, iend, offCode, mlen);
+                    ZSTD_updateStats(optStatePtr, llen, anchor, offBase, mlen);
+                    ZSTD_storeSeq(seqStore, llen, anchor, iend, offBase, mlen);
                     anchor += advance;
                     ip = anchor;
             }   }
@@ -1349,7 +1375,7 @@ size_t ZSTD_compressBlock_btopt(
 /* ZSTD_initStats_ultra():
  * make a first compression pass, just to seed stats with more accurate starting values.
  * only works on first block, with no dictionary and no ldm.
- * this function cannot error, hence its contract must be respected.
+ * this function cannot error out, its narrow contract must be respected.
  */
 static void
 ZSTD_initStats_ultra(ZSTD_matchState_t* ms,
@@ -1368,7 +1394,7 @@ ZSTD_initStats_ultra(ZSTD_matchState_t* ms,
 
     ZSTD_compressBlock_opt2(ms, seqStore, tmpRep, src, srcSize, ZSTD_noDict);   /* generate stats into ms->opt*/
 
-    /* invalidate first scan from history */
+    /* invalidate first scan from history, only keep entropy stats */
     ZSTD_resetSeqStore(seqStore);
     ms->window.base -= srcSize;
     ms->window.dictLimit += (U32)srcSize;
@@ -1392,20 +1418,20 @@ size_t ZSTD_compressBlock_btultra2(
     U32 const curr = (U32)((const BYTE*)src - ms->window.base);
     DEBUGLOG(5, "ZSTD_compressBlock_btultra2 (srcSize=%zu)", srcSize);
 
-    /* 2-pass strategy:
+    /* 2-passes strategy:
      * this strategy makes a first pass over first block to collect statistics
-     * and seed next round's statistics with it.
-     * After 1st pass, function forgets everything, and starts a new block.
+     * in order to seed next round's statistics with it.
+     * After 1st pass, function forgets history, and starts a new block.
      * Consequently, this can only work if no data has been previously loaded in tables,
      * aka, no dictionary, no prefix, no ldm preprocessing.
      * The compression ratio gain is generally small (~0.5% on first block),
-     * the cost is 2x cpu time on first block. */
+    ** the cost is 2x cpu time on first block. */
     assert(srcSize <= ZSTD_BLOCKSIZE_MAX);
     if ( (ms->opt.litLengthSum==0)   /* first block */
       && (seqStore->sequences == seqStore->sequencesStart)  /* no ldm */
       && (ms->window.dictLimit == ms->window.lowLimit)   /* no dictionary */
-      && (curr == ms->window.dictLimit)   /* start of frame, nothing already loaded nor skipped */
-      && (srcSize > ZSTD_PREDEF_THRESHOLD)
+      && (curr == ms->window.dictLimit)    /* start of frame, nothing already loaded nor skipped */
+      && (srcSize > ZSTD_PREDEF_THRESHOLD) /* input large enough to not employ default stats */
       ) {
         ZSTD_initStats_ultra(ms, seqStore, rep, src, srcSize);
     }

data/ext/zstdlib_c/{zstd-1.5.2 → zstd-1.5.5}/lib/compress/zstd_opt.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) Yann Collet, Facebook, Inc.
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the

data/ext/zstdlib_c/{zstd-1.5.2 → zstd-1.5.5}/lib/compress/zstdmt_compress.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) Yann Collet, Facebook, Inc.
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the
@@ -20,6 +20,7 @@
 
 
 /* ======   Dependencies   ====== */
+#include "../common/allocations.h"  /* ZSTD_customMalloc, ZSTD_customCalloc, ZSTD_customFree */
 #include "../common/zstd_deps.h"   /* ZSTD_memcpy, ZSTD_memset, INT_MAX, UINT_MAX */
 #include "../common/mem.h"         /* MEM_STATIC */
 #include "../common/pool.h"        /* threadpool */
@@ -266,11 +267,11 @@ static void ZSTDMT_releaseBuffer(ZSTDMT_bufferPool* bufPool, buffer_t buf)
  *   1 buffer for input loading
  *   1 buffer for "next input" when submitting current one
  *   1 buffer stuck in queue */
-#define BUF_POOL_MAX_NB_BUFFERS(nbWorkers) 2*nbWorkers + 3
+#define BUF_POOL_MAX_NB_BUFFERS(nbWorkers) (2*(nbWorkers) + 3)
 
 /* After a worker releases its rawSeqStore, it is immediately ready for reuse.
  * So we only need one seq buffer per worker. */
-#define SEQ_POOL_MAX_NB_BUFFERS(nbWorkers) nbWorkers
+#define SEQ_POOL_MAX_NB_BUFFERS(nbWorkers) (nbWorkers)
 
 /* =====   Seq Pool Wrapper   ====== */
 
@@ -719,7 +720,7 @@ static void ZSTDMT_compressionJob(void* jobDescription)
     ZSTDMT_serialState_update(job->serial, cctx, rawSeqStore, job->src, job->jobID);
 
     if (!job->firstJob) {  /* flush and overwrite frame header when it's not first job */
-        size_t const hSize = ZSTD_compressContinue(cctx, dstBuff.start, dstBuff.capacity, job->src.start, 0);
+        size_t const hSize = ZSTD_compressContinue_public(cctx, dstBuff.start, dstBuff.capacity, job->src.start, 0);
         if (ZSTD_isError(hSize)) JOB_ERROR(hSize);
         DEBUGLOG(5, "ZSTDMT_compressionJob: flush and overwrite %u bytes of frame header (not first job)", (U32)hSize);
         ZSTD_invalidateRepCodes(cctx);
@@ -737,7 +738,7 @@ static void ZSTDMT_compressionJob(void* jobDescription)
         DEBUGLOG(5, "ZSTDMT_compressionJob: compress %u bytes in %i blocks", (U32)job->src.size, nbChunks);
         assert(job->cSize == 0);
         for (chunkNb = 1; chunkNb < nbChunks; chunkNb++) {
-            size_t const cSize = ZSTD_compressContinue(cctx, op, oend-op, ip, chunkSize);
+            size_t const cSize = ZSTD_compressContinue_public(cctx, op, oend-op, ip, chunkSize);
             if (ZSTD_isError(cSize)) JOB_ERROR(cSize);
             ip += chunkSize;
             op += cSize; assert(op < oend);
@@ -757,8 +758,8 @@ static void ZSTDMT_compressionJob(void* jobDescription)
             size_t const lastBlockSize1 = job->src.size & (chunkSize-1);
             size_t const lastBlockSize = ((lastBlockSize1==0) & (job->src.size>=chunkSize)) ? chunkSize : lastBlockSize1;
             size_t const cSize = (job->lastJob) ?
-                 ZSTD_compressEnd     (cctx, op, oend-op, ip, lastBlockSize) :
-                 ZSTD_compressContinue(cctx, op, oend-op, ip, lastBlockSize);
+                 ZSTD_compressEnd_public(cctx, op, oend-op, ip, lastBlockSize) :
+                 ZSTD_compressContinue_public(cctx, op, oend-op, ip, lastBlockSize);
             if (ZSTD_isError(cSize)) JOB_ERROR(cSize);
             lastCBlockSize = cSize;
     }   }
@@ -1734,7 +1735,7 @@ findSynchronizationPoint(ZSTDMT_CCtx const* mtctx, ZSTD_inBuffer const input)
         }
     } else {
         /* We have enough bytes buffered to initialize the hash,
-         * and are have processed enough bytes to find a sync point.
+         * and have processed enough bytes to find a sync point.
          * Start scanning at the beginning of the input.
          */
         assert(mtctx->inBuff.filled >= RSYNC_MIN_BLOCK_SIZE);
@@ -1761,17 +1762,24 @@ findSynchronizationPoint(ZSTDMT_CCtx const* mtctx, ZSTD_inBuffer const input)
      * then a block will be emitted anyways, but this is okay, since if we
      * are already synchronized we will remain synchronized.
      */
+    assert(pos < RSYNC_LENGTH || ZSTD_rollingHash_compute(istart + pos - RSYNC_LENGTH, RSYNC_LENGTH) == hash);
     for (; pos < syncPoint.toLoad; ++pos) {
         BYTE const toRemove = pos < RSYNC_LENGTH ? prev[pos] : istart[pos - RSYNC_LENGTH];
-        assert(pos < RSYNC_LENGTH || ZSTD_rollingHash_compute(istart + pos - RSYNC_LENGTH, RSYNC_LENGTH) == hash);
+        /* This assert is very expensive, and Debian compiles with asserts enabled.
+         * So disable it for now. We can get similar coverage by checking it at the
+         * beginning & end of the loop.
+         * assert(pos < RSYNC_LENGTH || ZSTD_rollingHash_compute(istart + pos - RSYNC_LENGTH, RSYNC_LENGTH) == hash);
+         */
         hash = ZSTD_rollingHash_rotate(hash, toRemove, istart[pos], primePower);
         assert(mtctx->inBuff.filled + pos >= RSYNC_MIN_BLOCK_SIZE);
         if ((hash & hitMask) == hitMask) {
             syncPoint.toLoad = pos + 1;
             syncPoint.flush = 1;
+            ++pos; /* for assert */
             break;
         }
     }
+    assert(pos < RSYNC_LENGTH || ZSTD_rollingHash_compute(istart + pos - RSYNC_LENGTH, RSYNC_LENGTH) == hash);
     return syncPoint;
 }