zstd-ruby 1.3.7.0 → 1.3.8.0

data/ext/zstdruby/libzstd/compress/zstd_compress_internal.h

@@ -48,12 +48,6 @@ extern "C" {
 typedef enum { ZSTDcs_created=0, ZSTDcs_init, ZSTDcs_ongoing, ZSTDcs_ending } ZSTD_compressionStage_e;
 typedef enum { zcss_init=0, zcss_load, zcss_flush } ZSTD_cStreamStage;
 
-typedef enum {
-    ZSTD_dictDefaultAttach = 0,
-    ZSTD_dictForceAttach = 1,
-    ZSTD_dictForceCopy = -1,
-} ZSTD_dictAttachPref_e;
-
 typedef struct ZSTD_prefixDict_s {
     const void* dict;
     size_t dictSize;
@@ -96,10 +90,10 @@ typedef enum { zop_dynamic=0, zop_predef } ZSTD_OptPrice_e;
 
 typedef struct {
     /* All tables are allocated inside cctx->workspace by ZSTD_resetCCtx_internal() */
-    U32* litFreq;                /* table of literals statistics, of size 256 */
-    U32* litLengthFreq;          /* table of litLength statistics, of size (MaxLL+1) */
-    U32* matchLengthFreq;        /* table of matchLength statistics, of size (MaxML+1) */
-    U32* offCodeFreq;            /* table of offCode statistics, of size (MaxOff+1) */
+    unsigned* litFreq;           /* table of literals statistics, of size 256 */
+    unsigned* litLengthFreq;     /* table of litLength statistics, of size (MaxLL+1) */
+    unsigned* matchLengthFreq;   /* table of matchLength statistics, of size (MaxML+1) */
+    unsigned* offCodeFreq;       /* table of offCode statistics, of size (MaxOff+1) */
     ZSTD_match_t* matchTable;    /* list of found matches, of size ZSTD_OPT_NUM+1 */
     ZSTD_optimal_t* priceTable;  /* All positions tracked by optimal parser, of size ZSTD_OPT_NUM+1 */
 
@@ -139,7 +133,7 @@ struct ZSTD_matchState_t {
     U32* hashTable3;
     U32* chainTable;
     optState_t opt;         /* optimal parser state */
-    const ZSTD_matchState_t *dictMatchState;
+    const ZSTD_matchState_t * dictMatchState;
     ZSTD_compressionParameters cParams;
 };
 
@@ -167,7 +161,7 @@ typedef struct {
     U32 hashLog;            /* Log size of hashTable */
     U32 bucketSizeLog;      /* Log bucket size for collision resolution, at most 8 */
     U32 minMatchLength;     /* Minimum match length */
-    U32 hashEveryLog;       /* Log number of entries to skip */
+    U32 hashRateLog;       /* Log number of entries to skip */
     U32 windowLog;          /* Window log for the LDM */
 } ldmParams_t;
 
@@ -196,9 +190,10 @@ struct ZSTD_CCtx_params_s {
     ZSTD_dictAttachPref_e attachDictPref;
 
     /* Multithreading: used to pass parameters to mtctx */
-    unsigned nbWorkers;
-    unsigned jobSize;
-    unsigned overlapSizeLog;
+    int nbWorkers;
+    size_t jobSize;
+    int overlapLog;
+    int rsyncable;
 
     /* Long distance matching parameters */
     ldmParams_t ldmParams;
@@ -498,6 +493,64 @@ MEM_STATIC size_t ZSTD_hashPtr(const void* p, U32 hBits, U32 mls)
     }
 }
 
+/** ZSTD_ipow() :
+ * Return base^exponent.
+ */
+static U64 ZSTD_ipow(U64 base, U64 exponent)
+{
+    U64 power = 1;
+    while (exponent) {
+      if (exponent & 1) power *= base;
+      exponent >>= 1;
+      base *= base;
+    }
+    return power;
+}
+
+#define ZSTD_ROLL_HASH_CHAR_OFFSET 10
+
+/** ZSTD_rollingHash_append() :
+ * Add the buffer to the hash value.
+ */
+static U64 ZSTD_rollingHash_append(U64 hash, void const* buf, size_t size)
+{
+    BYTE const* istart = (BYTE const*)buf;
+    size_t pos;
+    for (pos = 0; pos < size; ++pos) {
+        hash *= prime8bytes;
+        hash += istart[pos] + ZSTD_ROLL_HASH_CHAR_OFFSET;
+    }
+    return hash;
+}
+
+/** ZSTD_rollingHash_compute() :
+ * Compute the rolling hash value of the buffer.
+ */
+MEM_STATIC U64 ZSTD_rollingHash_compute(void const* buf, size_t size)
+{
+    return ZSTD_rollingHash_append(0, buf, size);
+}
+
+/** ZSTD_rollingHash_primePower() :
+ * Compute the primePower to be passed to ZSTD_rollingHash_rotate() for a hash
+ * over a window of length bytes.
+ */
+MEM_STATIC U64 ZSTD_rollingHash_primePower(U32 length)
+{
+    return ZSTD_ipow(prime8bytes, length - 1);
+}
+
+/** ZSTD_rollingHash_rotate() :
+ * Rotate the rolling hash by one byte.
+ */
+MEM_STATIC U64 ZSTD_rollingHash_rotate(U64 hash, BYTE toRemove, BYTE toAdd, U64 primePower)
+{
+    hash -= (toRemove + ZSTD_ROLL_HASH_CHAR_OFFSET) * primePower;
+    hash *= prime8bytes;
+    hash += toAdd + ZSTD_ROLL_HASH_CHAR_OFFSET;
+    return hash;
+}
+
 /*-*************************************
 *  Round buffer management
 ***************************************/
@@ -626,20 +679,23 @@ MEM_STATIC U32 ZSTD_window_correctOverflow(ZSTD_window_t* window, U32 cycleLog,
  * dictMatchState mode, lowLimit and dictLimit are the same, and the dictionary
  * is below them. forceWindow and dictMatchState are therefore incompatible.
  */
-MEM_STATIC void ZSTD_window_enforceMaxDist(ZSTD_window_t* window,
-                                           void const* srcEnd, U32 maxDist,
-                                           U32* loadedDictEndPtr,
-                                           const ZSTD_matchState_t** dictMatchStatePtr)
+MEM_STATIC void
+ZSTD_window_enforceMaxDist(ZSTD_window_t* window,
+                           void const* srcEnd,
+                           U32 maxDist,
+                           U32* loadedDictEndPtr,
+                     const ZSTD_matchState_t** dictMatchStatePtr)
 {
-    U32 const current = (U32)((BYTE const*)srcEnd - window->base);
-    U32 loadedDictEnd = loadedDictEndPtr != NULL ? *loadedDictEndPtr : 0;
-    DEBUGLOG(5, "ZSTD_window_enforceMaxDist: current=%u, maxDist=%u", current, maxDist);
-    if (current > maxDist + loadedDictEnd) {
-        U32 const newLowLimit = current - maxDist;
+    U32 const blockEndIdx = (U32)((BYTE const*)srcEnd - window->base);
+    U32 loadedDictEnd = (loadedDictEndPtr != NULL) ? *loadedDictEndPtr : 0;
+    DEBUGLOG(5, "ZSTD_window_enforceMaxDist: blockEndIdx=%u, maxDist=%u",
+                (unsigned)blockEndIdx, (unsigned)maxDist);
+    if (blockEndIdx > maxDist + loadedDictEnd) {
+        U32 const newLowLimit = blockEndIdx - maxDist;
         if (window->lowLimit < newLowLimit) window->lowLimit = newLowLimit;
         if (window->dictLimit < window->lowLimit) {
             DEBUGLOG(5, "Update dictLimit to match lowLimit, from %u to %u",
-                        window->dictLimit, window->lowLimit);
+                        (unsigned)window->dictLimit, (unsigned)window->lowLimit);
             window->dictLimit = window->lowLimit;
         }
         if (loadedDictEndPtr)
@@ -690,20 +746,23 @@ MEM_STATIC U32 ZSTD_window_update(ZSTD_window_t* window,
 
 
 /* debug functions */
+#if (DEBUGLEVEL>=2)
 
 MEM_STATIC double ZSTD_fWeight(U32 rawStat)
 {
     U32 const fp_accuracy = 8;
     U32 const fp_multiplier = (1 << fp_accuracy);
-    U32 const stat = rawStat + 1;
-    U32 const hb = ZSTD_highbit32(stat);
+    U32 const newStat = rawStat + 1;
+    U32 const hb = ZSTD_highbit32(newStat);
     U32 const BWeight = hb * fp_multiplier;
-    U32 const FWeight = (stat << fp_accuracy) >> hb;
+    U32 const FWeight = (newStat << fp_accuracy) >> hb;
     U32 const weight = BWeight + FWeight;
     assert(hb + fp_accuracy < 31);
     return (double)weight / fp_multiplier;
 }
 
+/* display a table content,
+ * listing each element, its frequency, and its predicted bit cost */
 MEM_STATIC void ZSTD_debugTable(const U32* table, U32 max)
 {
     unsigned u, sum;
@@ -715,6 +774,9 @@ MEM_STATIC void ZSTD_debugTable(const U32* table, U32 max)
     }
 }
 
+#endif
+
+
 #if defined (__cplusplus)
 }
 #endif

data/ext/zstdruby/libzstd/compress/zstd_double_fast.c

@@ -18,7 +18,7 @@ void ZSTD_fillDoubleHashTable(ZSTD_matchState_t* ms,
     const ZSTD_compressionParameters* const cParams = &ms->cParams;
     U32* const hashLarge = ms->hashTable;
     U32  const hBitsL = cParams->hashLog;
-    U32  const mls = cParams->searchLength;
+    U32  const mls = cParams->minMatch;
     U32* const hashSmall = ms->chainTable;
     U32  const hBitsS = cParams->chainLog;
     const BYTE* const base = ms->window.base;
@@ -309,7 +309,7 @@ size_t ZSTD_compressBlock_doubleFast(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize)
 {
-    const U32 mls = ms->cParams.searchLength;
+    const U32 mls = ms->cParams.minMatch;
     switch(mls)
     {
     default: /* includes case 3 */
@@ -329,7 +329,7 @@ size_t ZSTD_compressBlock_doubleFast_dictMatchState(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize)
 {
-    const U32 mls = ms->cParams.searchLength;
+    const U32 mls = ms->cParams.minMatch;
     switch(mls)
     {
     default: /* includes case 3 */
@@ -483,7 +483,7 @@ size_t ZSTD_compressBlock_doubleFast_extDict(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize)
 {
-    U32 const mls = ms->cParams.searchLength;
+    U32 const mls = ms->cParams.minMatch;
     switch(mls)
     {
     default: /* includes case 3 */

data/ext/zstdruby/libzstd/compress/zstd_fast.c

@@ -18,7 +18,7 @@ void ZSTD_fillHashTable(ZSTD_matchState_t* ms,
     const ZSTD_compressionParameters* const cParams = &ms->cParams;
     U32* const hashTable = ms->hashTable;
     U32  const hBits = cParams->hashLog;
-    U32  const mls = cParams->searchLength;
+    U32  const mls = cParams->minMatch;
     const BYTE* const base = ms->window.base;
     const BYTE* ip = base + ms->nextToUpdate;
     const BYTE* const iend = ((const BYTE*)end) - HASH_READ_SIZE;
@@ -27,18 +27,18 @@ void ZSTD_fillHashTable(ZSTD_matchState_t* ms,
     /* Always insert every fastHashFillStep position into the hash table.
      * Insert the other positions if their hash entry is empty.
      */
-    for (; ip + fastHashFillStep - 1 <= iend; ip += fastHashFillStep) {
+    for ( ; ip + fastHashFillStep < iend + 2; ip += fastHashFillStep) {
         U32 const current = (U32)(ip - base);
-        U32 i;
-        for (i = 0; i < fastHashFillStep; ++i) {
-            size_t const hash = ZSTD_hashPtr(ip + i, hBits, mls);
-            if (i == 0 || hashTable[hash] == 0)
-                hashTable[hash] = current + i;
-            /* Only load extra positions for ZSTD_dtlm_full */
-            if (dtlm == ZSTD_dtlm_fast)
-                break;
-        }
-    }
+        size_t const hash0 = ZSTD_hashPtr(ip, hBits, mls);
+        hashTable[hash0] = current;
+        if (dtlm == ZSTD_dtlm_fast) continue;
+        /* Only load extra positions for ZSTD_dtlm_full */
+        {   U32 p;
+            for (p = 1; p < fastHashFillStep; ++p) {
+                size_t const hash = ZSTD_hashPtr(ip + p, hBits, mls);
+                if (hashTable[hash] == 0) {  /* not yet filled */
+                    hashTable[hash] = current + p;
+    }   }   }   }
 }
 
 FORCE_INLINE_TEMPLATE
@@ -235,7 +235,7 @@ size_t ZSTD_compressBlock_fast(
         void const* src, size_t srcSize)
 {
     ZSTD_compressionParameters const* cParams = &ms->cParams;
-    U32 const mls = cParams->searchLength;
+    U32 const mls = cParams->minMatch;
     assert(ms->dictMatchState == NULL);
     switch(mls)
     {
@@ -256,7 +256,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState(
         void const* src, size_t srcSize)
 {
     ZSTD_compressionParameters const* cParams = &ms->cParams;
-    U32 const mls = cParams->searchLength;
+    U32 const mls = cParams->minMatch;
     assert(ms->dictMatchState != NULL);
     switch(mls)
     {
@@ -375,7 +375,7 @@ size_t ZSTD_compressBlock_fast_extDict(
         void const* src, size_t srcSize)
 {
     ZSTD_compressionParameters const* cParams = &ms->cParams;
-    U32 const mls = cParams->searchLength;
+    U32 const mls = cParams->minMatch;
     switch(mls)
     {
     default: /* includes case 3 */

data/ext/zstdruby/libzstd/compress/zstd_lazy.c

@@ -63,12 +63,13 @@ ZSTD_updateDUBT(ZSTD_matchState_t* ms,
 static void
 ZSTD_insertDUBT1(ZSTD_matchState_t* ms,
                  U32 current, const BYTE* inputEnd,
-                 U32 nbCompares, U32 btLow, const ZSTD_dictMode_e dictMode)
+                 U32 nbCompares, U32 btLow,
+                 const ZSTD_dictMode_e dictMode)
 {
     const ZSTD_compressionParameters* const cParams = &ms->cParams;
-    U32*   const bt = ms->chainTable;
-    U32    const btLog  = cParams->chainLog - 1;
-    U32    const btMask = (1 << btLog) - 1;
+    U32* const bt = ms->chainTable;
+    U32  const btLog  = cParams->chainLog - 1;
+    U32  const btMask = (1 << btLog) - 1;
     size_t commonLengthSmaller=0, commonLengthLarger=0;
     const BYTE* const base = ms->window.base;
     const BYTE* const dictBase = ms->window.dictBase;
@@ -80,7 +81,7 @@ ZSTD_insertDUBT1(ZSTD_matchState_t* ms,
     const BYTE* match;
     U32* smallerPtr = bt + 2*(current&btMask);
     U32* largerPtr  = smallerPtr + 1;
-    U32 matchIndex = *smallerPtr;
+    U32 matchIndex = *smallerPtr;   /* this candidate is unsorted : next sorted candidate is reached through *smallerPtr, while *largerPtr contains previous unsorted candidate (which is already saved and can be overwritten) */
     U32 dummy32;   /* to be nullified at the end */
     U32 const windowLow = ms->window.lowLimit;
 
@@ -93,6 +94,9 @@ ZSTD_insertDUBT1(ZSTD_matchState_t* ms,
         U32* const nextPtr = bt + 2*(matchIndex & btMask);
         size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger);   /* guaranteed minimum nb of common bytes */
         assert(matchIndex < current);
+        /* note : all candidates are now supposed sorted,
+         * but it's still possible to have nextPtr[1] == ZSTD_DUBT_UNSORTED_MARK
+         * when a real index has the same value as ZSTD_DUBT_UNSORTED_MARK */
 
         if ( (dictMode != ZSTD_extDict)
           || (matchIndex+matchLength >= dictLimit)  /* both in current segment*/
@@ -108,7 +112,7 @@ ZSTD_insertDUBT1(ZSTD_matchState_t* ms,
             match = dictBase + matchIndex;
             matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iend, dictEnd, prefixStart);
             if (matchIndex+matchLength >= dictLimit)
-                match = base + matchIndex;   /* to prepare for next usage of match[matchLength] */
+                match = base + matchIndex;   /* preparation for next read of match[matchLength] */
         }
 
         DEBUGLOG(8, "ZSTD_insertDUBT1: comparing %u with %u : found %u common bytes ",
@@ -258,7 +262,7 @@ ZSTD_DUBT_findBestMatch(ZSTD_matchState_t* ms,
          && (nbCandidates > 1) ) {
         DEBUGLOG(8, "ZSTD_DUBT_findBestMatch: candidate %u is unsorted",
                     matchIndex);
-        *unsortedMark = previousCandidate;
+        *unsortedMark = previousCandidate;  /* the unsortedMark becomes a reversed chain, to move up back to original position */
         previousCandidate = matchIndex;
         matchIndex = *nextCandidate;
         nextCandidate = bt + 2*(matchIndex&btMask);
@@ -266,11 +270,13 @@ ZSTD_DUBT_findBestMatch(ZSTD_matchState_t* ms,
         nbCandidates --;
     }
 
+    /* nullify last candidate if it's still unsorted
+     * simplification, detrimental to compression ratio, beneficial for speed */
     if ( (matchIndex > unsortLimit)
       && (*unsortedMark==ZSTD_DUBT_UNSORTED_MARK) ) {
         DEBUGLOG(7, "ZSTD_DUBT_findBestMatch: nullify last unsorted candidate %u",
                     matchIndex);
-        *nextCandidate = *unsortedMark = 0;   /* nullify next candidate if it's still unsorted (note : simplification, detrimental to compression ratio, beneficial for speed) */
+        *nextCandidate = *unsortedMark = 0;
     }
 
     /* batch sort stacked candidates */
@@ -285,14 +291,14 @@ ZSTD_DUBT_findBestMatch(ZSTD_matchState_t* ms,
     }
 
     /* find longest match */
-    {   size_t commonLengthSmaller=0, commonLengthLarger=0;
+    {   size_t commonLengthSmaller = 0, commonLengthLarger = 0;
         const BYTE* const dictBase = ms->window.dictBase;
         const U32 dictLimit = ms->window.dictLimit;
         const BYTE* const dictEnd = dictBase + dictLimit;
         const BYTE* const prefixStart = base + dictLimit;
         U32* smallerPtr = bt + 2*(current&btMask);
         U32* largerPtr  = bt + 2*(current&btMask) + 1;
-        U32 matchEndIdx = current+8+1;
+        U32 matchEndIdx = current + 8 + 1;
         U32 dummy32;   /* to be nullified at the end */
         size_t bestLength = 0;
 
@@ -386,7 +392,7 @@ ZSTD_BtFindBestMatch_selectMLS (  ZSTD_matchState_t* ms,
                             const BYTE* ip, const BYTE* const iLimit,
                                   size_t* offsetPtr)
 {
-    switch(ms->cParams.searchLength)
+    switch(ms->cParams.minMatch)
     {
     default : /* includes case 3 */
     case 4 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 4, ZSTD_noDict);
@@ -402,7 +408,7 @@ static size_t ZSTD_BtFindBestMatch_dictMatchState_selectMLS (
                         const BYTE* ip, const BYTE* const iLimit,
                         size_t* offsetPtr)
 {
-    switch(ms->cParams.searchLength)
+    switch(ms->cParams.minMatch)
     {
     default : /* includes case 3 */
     case 4 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 4, ZSTD_dictMatchState);
@@ -418,7 +424,7 @@ static size_t ZSTD_BtFindBestMatch_extDict_selectMLS (
                         const BYTE* ip, const BYTE* const iLimit,
                         size_t* offsetPtr)
 {
-    switch(ms->cParams.searchLength)
+    switch(ms->cParams.minMatch)
     {
     default : /* includes case 3 */
     case 4 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 4, ZSTD_extDict);
@@ -433,7 +439,7 @@ static size_t ZSTD_BtFindBestMatch_extDict_selectMLS (
 /* *********************************
 *  Hash Chain
 ***********************************/
-#define NEXT_IN_CHAIN(d, mask)   chainTable[(d) & mask]
+#define NEXT_IN_CHAIN(d, mask)   chainTable[(d) & (mask)]
 
 /* Update chains up to ip (excluded)
    Assumption : always within prefix (i.e. not within extDict) */
@@ -463,7 +469,7 @@ static U32 ZSTD_insertAndFindFirstIndex_internal(
 
 U32 ZSTD_insertAndFindFirstIndex(ZSTD_matchState_t* ms, const BYTE* ip) {
     const ZSTD_compressionParameters* const cParams = &ms->cParams;
-    return ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, ms->cParams.searchLength);
+    return ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, ms->cParams.minMatch);
 }
 
 
@@ -497,6 +503,7 @@ size_t ZSTD_HcFindBestMatch_generic (
         size_t currentMl=0;
         if ((dictMode != ZSTD_extDict) || matchIndex >= dictLimit) {
             const BYTE* const match = base + matchIndex;
+            assert(matchIndex >= dictLimit);   /* ensures this is true if dictMode != ZSTD_extDict */
             if (match[ml] == ip[ml])   /* potentially better */
                 currentMl = ZSTD_count(ip, match, iLimit);
         } else {
@@ -559,7 +566,7 @@ FORCE_INLINE_TEMPLATE size_t ZSTD_HcFindBestMatch_selectMLS (
                         const BYTE* ip, const BYTE* const iLimit,
                         size_t* offsetPtr)
 {
-    switch(ms->cParams.searchLength)
+    switch(ms->cParams.minMatch)
     {
     default : /* includes case 3 */
     case 4 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 4, ZSTD_noDict);
@@ -575,7 +582,7 @@ static size_t ZSTD_HcFindBestMatch_dictMatchState_selectMLS (
                         const BYTE* ip, const BYTE* const iLimit,
                         size_t* offsetPtr)
 {
-    switch(ms->cParams.searchLength)
+    switch(ms->cParams.minMatch)
     {
     default : /* includes case 3 */
     case 4 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 4, ZSTD_dictMatchState);
@@ -591,7 +598,7 @@ FORCE_INLINE_TEMPLATE size_t ZSTD_HcFindBestMatch_extDict_selectMLS (
                         const BYTE* ip, const BYTE* const iLimit,
                         size_t* offsetPtr)
 {
-    switch(ms->cParams.searchLength)
+    switch(ms->cParams.minMatch)
     {
     default : /* includes case 3 */
     case 4 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 4, ZSTD_extDict);