zstd-ruby 1.3.8.0 → 1.4.5.0

data/ext/zstdruby/libzstd/dictBuilder/zdict.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the
@@ -46,7 +46,12 @@ extern "C" {
  *  The resulting dictionary will be saved into `dictBuffer`.
  * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
  *          or an error code, which can be tested with ZDICT_isError().
- *  Note: ZDICT_trainFromBuffer() requires about 9 bytes of memory for each input byte.
+ *  Note:  Dictionary training will fail if there are not enough samples to construct a
+ *         dictionary, or if most of the samples are too small (< 8 bytes being the lower limit).
+ *         If dictionary training fails, you should use zstd without a dictionary, as the dictionary
+ *         would've been ineffective anyways. If you believe your samples would benefit from a dictionary
+ *         please open an issue with details, and we can look into it.
+ *  Note: ZDICT_trainFromBuffer()'s memory usage is about 6 MB.
  *  Tips: In general, a reasonable dictionary has a size of ~ 100 KB.
  *        It's possible to select smaller or larger size, just by specifying `dictBufferCapacity`.
  *        In general, it's recommended to provide a few thousands samples, though this can vary a lot.
@@ -56,9 +61,57 @@ ZDICTLIB_API size_t ZDICT_trainFromBuffer(void* dictBuffer, size_t dictBufferCap
                                     const void* samplesBuffer,
                                     const size_t* samplesSizes, unsigned nbSamples);
 
+typedef struct {
+    int      compressionLevel;   /*< optimize for a specific zstd compression level; 0 means default */
+    unsigned notificationLevel;  /*< Write log to stderr; 0 = none (default); 1 = errors; 2 = progression; 3 = details; 4 = debug; */
+    unsigned dictID;             /*< force dictID value; 0 means auto mode (32-bits random value) */
+} ZDICT_params_t;
+
+/*! ZDICT_finalizeDictionary():
+ * Given a custom content as a basis for dictionary, and a set of samples,
+ * finalize dictionary by adding headers and statistics according to the zstd
+ * dictionary format.
+ *
+ * Samples must be stored concatenated in a flat buffer `samplesBuffer`,
+ * supplied with an array of sizes `samplesSizes`, providing the size of each
+ * sample in order. The samples are used to construct the statistics, so they
+ * should be representative of what you will compress with this dictionary.
+ *
+ * The compression level can be set in `parameters`. You should pass the
+ * compression level you expect to use in production. The statistics for each
+ * compression level differ, so tuning the dictionary for the compression level
+ * can help quite a bit.
+ *
+ * You can set an explicit dictionary ID in `parameters`, or allow us to pick
+ * a random dictionary ID for you, but we can't guarantee no collisions.
+ *
+ * The dstDictBuffer and the dictContent may overlap, and the content will be
+ * appended to the end of the header. If the header + the content doesn't fit in
+ * maxDictSize the beginning of the content is truncated to make room, since it
+ * is presumed that the most profitable content is at the end of the dictionary,
+ * since that is the cheapest to reference.
+ *
+ * `dictContentSize` must be >= ZDICT_CONTENTSIZE_MIN bytes.
+ * `maxDictSize` must be >= max(dictContentSize, ZSTD_DICTSIZE_MIN).
+ *
+ * @return: size of dictionary stored into `dstDictBuffer` (<= `maxDictSize`),
+ *          or an error code, which can be tested by ZDICT_isError().
+ * Note: ZDICT_finalizeDictionary() will push notifications into stderr if
+ *       instructed to, using notificationLevel>0.
+ * NOTE: This function currently may fail in several edge cases including:
+ *         * Not enough samples
+ *         * Samples are uncompressible
+ *         * Samples are all exactly the same
+ */
+ZDICTLIB_API size_t ZDICT_finalizeDictionary(void* dstDictBuffer, size_t maxDictSize,
+                                const void* dictContent, size_t dictContentSize,
+                                const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
+                                ZDICT_params_t parameters);
+
 
 /*======   Helper functions   ======*/
 ZDICTLIB_API unsigned ZDICT_getDictID(const void* dictBuffer, size_t dictSize);  /**< extracts dictID; @return zero if error (not a valid dictionary) */
+ZDICTLIB_API size_t ZDICT_getDictHeaderSize(const void* dictBuffer, size_t dictSize);  /* returns dict header size; returns a ZSTD error code on failure */
 ZDICTLIB_API unsigned ZDICT_isError(size_t errorCode);
 ZDICTLIB_API const char* ZDICT_getErrorName(size_t errorCode);
 
@@ -73,11 +126,8 @@ ZDICTLIB_API const char* ZDICT_getErrorName(size_t errorCode);
  * Use them only in association with static linking.
  * ==================================================================================== */
 
-typedef struct {
-    int      compressionLevel;   /* optimize for a specific zstd compression level; 0 means default */
-    unsigned notificationLevel;  /* Write log to stderr; 0 = none (default); 1 = errors; 2 = progression; 3 = details; 4 = debug; */
-    unsigned dictID;             /* force dictID value; 0 means auto mode (32-bits random value) */
-} ZDICT_params_t;
+#define ZDICT_CONTENTSIZE_MIN 128
+#define ZDICT_DICTSIZE_MIN    256
 
 /*! ZDICT_cover_params_t:
  *  k and d are the only required parameters.
@@ -89,6 +139,8 @@ typedef struct {
     unsigned steps;              /* Number of steps : Only used for optimization : 0 means default (40) : Higher means more parameters checked */
     unsigned nbThreads;          /* Number of threads : constraint: 0 < nbThreads : 1 means single-threaded : Only used for optimization : Ignored if ZSTD_MULTITHREAD is not defined */
     double splitPoint;           /* Percentage of samples used for training: Only used for optimization : the first nbSamples * splitPoint samples will be used to training, the last nbSamples * (1 - splitPoint) samples will be used for testing, 0 means default (1.0), 1.0 when all samples are used for both training and testing */
+    unsigned shrinkDict;         /* Train dictionaries to shrink in size starting from the minimum size and selects the smallest dictionary that is shrinkDictMaxRegression% worse than the largest dictionary. 0 means no shrinking and 1 means shrinking  */
+    unsigned shrinkDictMaxRegression; /* Sets shrinkDictMaxRegression so that a smaller dictionary can be at worse shrinkDictMaxRegression% worse than the max dict size dictionary. */
     ZDICT_params_t zParams;
 } ZDICT_cover_params_t;
 
@@ -100,6 +152,9 @@ typedef struct {
     unsigned nbThreads;          /* Number of threads : constraint: 0 < nbThreads : 1 means single-threaded : Only used for optimization : Ignored if ZSTD_MULTITHREAD is not defined */
     double splitPoint;           /* Percentage of samples used for training: Only used for optimization : the first nbSamples * splitPoint samples will be used to training, the last nbSamples * (1 - splitPoint) samples will be used for testing, 0 means default (0.75), 1.0 when all samples are used for both training and testing */
     unsigned accel;              /* Acceleration level: constraint: 0 < accel <= 10, higher means faster and less accurate, 0 means default(1) */
+    unsigned shrinkDict;         /* Train dictionaries to shrink in size starting from the minimum size and selects the smallest dictionary that is shrinkDictMaxRegression% worse than the largest dictionary. 0 means no shrinking and 1 means shrinking  */
+    unsigned shrinkDictMaxRegression; /* Sets shrinkDictMaxRegression so that a smaller dictionary can be at worse shrinkDictMaxRegression% worse than the max dict size dictionary. */
+
     ZDICT_params_t zParams;
 } ZDICT_fastCover_params_t;
 
@@ -110,6 +165,7 @@ typedef struct {
  *  The resulting dictionary will be saved into `dictBuffer`.
  * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
  *          or an error code, which can be tested with ZDICT_isError().
+ *          See ZDICT_trainFromBuffer() for details on failure modes.
  *  Note: ZDICT_trainFromBuffer_cover() requires about 9 bytes of memory for each input byte.
  *  Tips: In general, a reasonable dictionary has a size of ~ 100 KB.
  *        It's possible to select smaller or larger size, just by specifying `dictBufferCapacity`.
@@ -133,8 +189,9 @@ ZDICTLIB_API size_t ZDICT_trainFromBuffer_cover(
  * If k is non-zero then we don't check multiple values of k, otherwise we check steps values in [50, 2000].
  *
  * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
- *           or an error code, which can be tested with ZDICT_isError().
- *           On success `*parameters` contains the parameters selected.
+ *          or an error code, which can be tested with ZDICT_isError().
+ *          On success `*parameters` contains the parameters selected.
+ *          See ZDICT_trainFromBuffer() for details on failure modes.
  * Note: ZDICT_optimizeTrainFromBuffer_cover() requires about 8 bytes of memory for each input byte and additionally another 5 bytes of memory for each byte of memory for each thread.
  */
 ZDICTLIB_API size_t ZDICT_optimizeTrainFromBuffer_cover(
@@ -151,7 +208,8 @@ ZDICTLIB_API size_t ZDICT_optimizeTrainFromBuffer_cover(
  *  The resulting dictionary will be saved into `dictBuffer`.
  * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
  *          or an error code, which can be tested with ZDICT_isError().
- *  Note: ZDICT_trainFromBuffer_fastCover() requires about 1 bytes of memory for each input byte and additionally another 6 * 2^f bytes of memory .
+ *          See ZDICT_trainFromBuffer() for details on failure modes.
+ *  Note: ZDICT_trainFromBuffer_fastCover() requires 6 * 2^f bytes of memory.
  *  Tips: In general, a reasonable dictionary has a size of ~ 100 KB.
  *        It's possible to select smaller or larger size, just by specifying `dictBufferCapacity`.
  *        In general, it's recommended to provide a few thousands samples, though this can vary a lot.
@@ -175,37 +233,16 @@ ZDICTLIB_API size_t ZDICT_trainFromBuffer_fastCover(void *dictBuffer,
  * If accel is zero, default value of 1 is used.
  *
  * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
- *           or an error code, which can be tested with ZDICT_isError().
- *           On success `*parameters` contains the parameters selected.
- * Note: ZDICT_optimizeTrainFromBuffer_fastCover() requires about 1 byte of memory for each input byte and additionally another 6 * 2^f bytes of memory for each thread.
+ *          or an error code, which can be tested with ZDICT_isError().
+ *          On success `*parameters` contains the parameters selected.
+ *          See ZDICT_trainFromBuffer() for details on failure modes.
+ * Note: ZDICT_optimizeTrainFromBuffer_fastCover() requires about 6 * 2^f bytes of memory for each thread.
  */
 ZDICTLIB_API size_t ZDICT_optimizeTrainFromBuffer_fastCover(void* dictBuffer,
                     size_t dictBufferCapacity, const void* samplesBuffer,
                     const size_t* samplesSizes, unsigned nbSamples,
                     ZDICT_fastCover_params_t* parameters);
 
-/*! ZDICT_finalizeDictionary():
- * Given a custom content as a basis for dictionary, and a set of samples,
- * finalize dictionary by adding headers and statistics.
- *
- * Samples must be stored concatenated in a flat buffer `samplesBuffer`,
- * supplied with an array of sizes `samplesSizes`, providing the size of each sample in order.
- *
- * dictContentSize must be >= ZDICT_CONTENTSIZE_MIN bytes.
- * maxDictSize must be >= dictContentSize, and must be >= ZDICT_DICTSIZE_MIN bytes.
- *
- * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`),
- *           or an error code, which can be tested by ZDICT_isError().
- * Note: ZDICT_finalizeDictionary() will push notifications into stderr if instructed to, using notificationLevel>0.
- * Note 2: dictBuffer and dictContent can overlap
- */
-#define ZDICT_CONTENTSIZE_MIN 128
-#define ZDICT_DICTSIZE_MIN    256
-ZDICTLIB_API size_t ZDICT_finalizeDictionary(void* dictBuffer, size_t dictBufferCapacity,
-                                const void* dictContent, size_t dictContentSize,
-                                const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
-                                ZDICT_params_t parameters);
-
 typedef struct {
     unsigned selectivityLevel;   /* 0 means default; larger => select more => larger dictionary */
     ZDICT_params_t zParams;
@@ -219,6 +256,7 @@ typedef struct {
  * `parameters` is optional and can be provided with values set to 0 to mean "default".
  * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
  *          or an error code, which can be tested with ZDICT_isError().
+ *          See ZDICT_trainFromBuffer() for details on failure modes.
  *  Tips: In general, a reasonable dictionary has a size of ~ 100 KB.
  *        It's possible to select smaller or larger size, just by specifying `dictBufferCapacity`.
  *        In general, it's recommended to provide a few thousands samples, though this can vary a lot.

data/ext/zstdruby/libzstd/dll/example/Makefile

@@ -1,10 +1,11 @@
 # ################################################################
-# Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+# Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
 # All rights reserved.
 #
 # This source code is licensed under both the BSD-style license (found in the
 # LICENSE file in the root directory of this source tree) and the GPLv2 (found
 # in the COPYING file in the root directory of this source tree).
+# You may select, at your option, one of the above-listed licenses.
 # ################################################################
 
 VOID    := /dev/null

data/ext/zstdruby/libzstd/dll/example/build_package.bat

@@ -6,14 +6,15 @@ COPY programs\datagen.h bin\example\
 COPY programs\util.h bin\example\
 COPY programs\platform.h bin\example\
 COPY lib\common\mem.h bin\example\
-COPY lib\common\zstd_errors.h bin\example\
 COPY lib\common\zstd_internal.h bin\example\
 COPY lib\common\error_private.h bin\example\
 COPY lib\common\xxhash.h bin\example\
-COPY lib\zstd.h bin\include\
 COPY lib\libzstd.a bin\static\libzstd_static.lib
 COPY lib\dll\libzstd.* bin\dll\
 COPY lib\dll\example\Makefile bin\example\
 COPY lib\dll\example\fullbench-dll.* bin\example\
 COPY lib\dll\example\README.md bin\
+COPY lib\zstd.h bin\include\
+COPY lib\common\zstd_errors.h bin\include\
+COPY lib\dictBuilder\zdict.h bin\include\
 COPY programs\zstd.exe bin\zstd.exe

data/ext/zstdruby/libzstd/legacy/zstd_legacy.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the
@@ -18,9 +18,9 @@ extern "C" {
 /* *************************************
 *  Includes
 ***************************************/
-#include "mem.h"            /* MEM_STATIC */
-#include "error_private.h"  /* ERROR */
-#include "zstd.h"           /* ZSTD_inBuffer, ZSTD_outBuffer */
+#include "../common/mem.h"            /* MEM_STATIC */
+#include "../common/error_private.h"  /* ERROR */
+#include "../common/zstd_internal.h"  /* ZSTD_inBuffer, ZSTD_outBuffer, ZSTD_frameSizeInfo */
 
 #if !defined (ZSTD_LEGACY_SUPPORT) || (ZSTD_LEGACY_SUPPORT == 0)
 #  undef ZSTD_LEGACY_SUPPORT
@@ -178,43 +178,77 @@ MEM_STATIC size_t ZSTD_decompressLegacy(
     }
 }
 
-MEM_STATIC size_t ZSTD_findFrameCompressedSizeLegacy(const void *src,
-                                             size_t compressedSize)
+MEM_STATIC ZSTD_frameSizeInfo ZSTD_findFrameSizeInfoLegacy(const void *src, size_t srcSize)
 {
-    U32 const version = ZSTD_isLegacy(src, compressedSize);
+    ZSTD_frameSizeInfo frameSizeInfo;
+    U32 const version = ZSTD_isLegacy(src, srcSize);
     switch(version)
     {
 #if (ZSTD_LEGACY_SUPPORT <= 1)
         case 1 :
-            return ZSTDv01_findFrameCompressedSize(src, compressedSize);
+            ZSTDv01_findFrameSizeInfoLegacy(src, srcSize,
+                &frameSizeInfo.compressedSize,
+                &frameSizeInfo.decompressedBound);
+            break;
 #endif
 #if (ZSTD_LEGACY_SUPPORT <= 2)
         case 2 :
-            return ZSTDv02_findFrameCompressedSize(src, compressedSize);
+            ZSTDv02_findFrameSizeInfoLegacy(src, srcSize,
+                &frameSizeInfo.compressedSize,
+                &frameSizeInfo.decompressedBound);
+            break;
 #endif
 #if (ZSTD_LEGACY_SUPPORT <= 3)
         case 3 :
-            return ZSTDv03_findFrameCompressedSize(src, compressedSize);
+            ZSTDv03_findFrameSizeInfoLegacy(src, srcSize,
+                &frameSizeInfo.compressedSize,
+                &frameSizeInfo.decompressedBound);
+            break;
 #endif
 #if (ZSTD_LEGACY_SUPPORT <= 4)
         case 4 :
-            return ZSTDv04_findFrameCompressedSize(src, compressedSize);
+            ZSTDv04_findFrameSizeInfoLegacy(src, srcSize,
+                &frameSizeInfo.compressedSize,
+                &frameSizeInfo.decompressedBound);
+            break;
 #endif
 #if (ZSTD_LEGACY_SUPPORT <= 5)
         case 5 :
-            return ZSTDv05_findFrameCompressedSize(src, compressedSize);
+            ZSTDv05_findFrameSizeInfoLegacy(src, srcSize,
+                &frameSizeInfo.compressedSize,
+                &frameSizeInfo.decompressedBound);
+            break;
 #endif
 #if (ZSTD_LEGACY_SUPPORT <= 6)
         case 6 :
-            return ZSTDv06_findFrameCompressedSize(src, compressedSize);
+            ZSTDv06_findFrameSizeInfoLegacy(src, srcSize,
+                &frameSizeInfo.compressedSize,
+                &frameSizeInfo.decompressedBound);
+            break;
 #endif
 #if (ZSTD_LEGACY_SUPPORT <= 7)
         case 7 :
-            return ZSTDv07_findFrameCompressedSize(src, compressedSize);
+            ZSTDv07_findFrameSizeInfoLegacy(src, srcSize,
+                &frameSizeInfo.compressedSize,
+                &frameSizeInfo.decompressedBound);
+            break;
 #endif
         default :
-            return ERROR(prefix_unknown);
+            frameSizeInfo.compressedSize = ERROR(prefix_unknown);
+            frameSizeInfo.decompressedBound = ZSTD_CONTENTSIZE_ERROR;
+            break;
+    }
+    if (!ZSTD_isError(frameSizeInfo.compressedSize) && frameSizeInfo.compressedSize > srcSize) {
+        frameSizeInfo.compressedSize = ERROR(srcSize_wrong);
+        frameSizeInfo.decompressedBound = ZSTD_CONTENTSIZE_ERROR;
     }
+    return frameSizeInfo;
+}
+
+MEM_STATIC size_t ZSTD_findFrameCompressedSizeLegacy(const void *src, size_t srcSize)
+{
+    ZSTD_frameSizeInfo frameSizeInfo = ZSTD_findFrameSizeInfoLegacy(src, srcSize);
+    return frameSizeInfo.compressedSize;
 }
 
 MEM_STATIC size_t ZSTD_freeLegacyStreamContext(void* legacyContext, U32 version)

data/ext/zstdruby/libzstd/legacy/zstd_v01.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the
@@ -14,7 +14,7 @@
 ******************************************/
 #include <stddef.h>    /* size_t, ptrdiff_t */
 #include "zstd_v01.h"
-#include "error_private.h"
+#include "../common/error_private.h"
 
 
 /******************************************
@@ -257,7 +257,7 @@ static U64 FSE_read64(const void* memPtr)
     U64 val; memcpy(&val, memPtr, sizeof(val)); return val;
 }
 
-#endif // FSE_FORCE_MEMORY_ACCESS
+#endif /* FSE_FORCE_MEMORY_ACCESS */
 
 static U16 FSE_readLE16(const void* memPtr)
 {
@@ -346,7 +346,7 @@ FORCE_INLINE unsigned FSE_highbit32 (U32 val)
     _BitScanReverse ( &r, val );
     return (unsigned) r;
 #   elif defined(__GNUC__) && (GCC_VERSION >= 304)   /* GCC Intrinsic */
-    return 31 - __builtin_clz (val);
+    return __builtin_clz (val) ^ 31;
 #   else   /* Software version */
     static const unsigned DeBruijnClz[32] = { 0, 9, 1, 10, 13, 21, 2, 29, 11, 14, 16, 18, 22, 25, 3, 30, 8, 12, 20, 28, 15, 17, 24, 7, 19, 27, 23, 6, 26, 5, 4, 31 };
     U32 v = val;
@@ -1073,99 +1073,102 @@ static size_t HUF_decompress_usingDTable(   /* -3% slower when non static */
     const void* cSrc, size_t cSrcSize,
     const U16* DTable)
 {
-    BYTE* const ostart = (BYTE*) dst;
-    BYTE* op = ostart;
-    BYTE* const omax = op + maxDstSize;
-    BYTE* const olimit = omax-15;
-
-    const void* ptr = DTable;
-    const HUF_DElt* const dt = (const HUF_DElt*)(ptr)+1;
-    const U32 dtLog = DTable[0];
-    size_t errorCode;
-    U32 reloadStatus;
-
-    /* Init */
-
-    const U16* jumpTable = (const U16*)cSrc;
-    const size_t length1 = FSE_readLE16(jumpTable);
-    const size_t length2 = FSE_readLE16(jumpTable+1);
-    const size_t length3 = FSE_readLE16(jumpTable+2);
-    const size_t length4 = cSrcSize - 6 - length1 - length2 - length3;   // check coherency !!
-    const char* const start1 = (const char*)(cSrc) + 6;
-    const char* const start2 = start1 + length1;
-    const char* const start3 = start2 + length2;
-    const char* const start4 = start3 + length3;
-    FSE_DStream_t bitD1, bitD2, bitD3, bitD4;
-
-    if (length1+length2+length3+6 >= cSrcSize) return (size_t)-FSE_ERROR_srcSize_wrong;
-
-    errorCode = FSE_initDStream(&bitD1, start1, length1);
-    if (FSE_isError(errorCode)) return errorCode;
-    errorCode = FSE_initDStream(&bitD2, start2, length2);
-    if (FSE_isError(errorCode)) return errorCode;
-    errorCode = FSE_initDStream(&bitD3, start3, length3);
-    if (FSE_isError(errorCode)) return errorCode;
-    errorCode = FSE_initDStream(&bitD4, start4, length4);
-    if (FSE_isError(errorCode)) return errorCode;
-
-    reloadStatus=FSE_reloadDStream(&bitD2);
-
-    /* 16 symbols per loop */
-    for ( ; (reloadStatus<FSE_DStream_completed) && (op<olimit);  /* D2-3-4 are supposed to be synchronized and finish together */
-        op+=16, reloadStatus = FSE_reloadDStream(&bitD2) | FSE_reloadDStream(&bitD3) | FSE_reloadDStream(&bitD4), FSE_reloadDStream(&bitD1))
+    if (cSrcSize < 6) return (size_t)-FSE_ERROR_srcSize_wrong;
     {
-#define HUF_DECODE_SYMBOL_0(n, Dstream) \
-        op[n] = HUF_decodeSymbol(&Dstream, dt, dtLog);
-
-#define HUF_DECODE_SYMBOL_1(n, Dstream) \
-        op[n] = HUF_decodeSymbol(&Dstream, dt, dtLog); \
-        if (FSE_32bits() && (HUF_MAX_TABLELOG>12)) FSE_reloadDStream(&Dstream)
-
-#define HUF_DECODE_SYMBOL_2(n, Dstream) \
-        op[n] = HUF_decodeSymbol(&Dstream, dt, dtLog); \
-        if (FSE_32bits()) FSE_reloadDStream(&Dstream)
-
-        HUF_DECODE_SYMBOL_1( 0, bitD1);
-        HUF_DECODE_SYMBOL_1( 1, bitD2);
-        HUF_DECODE_SYMBOL_1( 2, bitD3);
-        HUF_DECODE_SYMBOL_1( 3, bitD4);
-        HUF_DECODE_SYMBOL_2( 4, bitD1);
-        HUF_DECODE_SYMBOL_2( 5, bitD2);
-        HUF_DECODE_SYMBOL_2( 6, bitD3);
-        HUF_DECODE_SYMBOL_2( 7, bitD4);
-        HUF_DECODE_SYMBOL_1( 8, bitD1);
-        HUF_DECODE_SYMBOL_1( 9, bitD2);
-        HUF_DECODE_SYMBOL_1(10, bitD3);
-        HUF_DECODE_SYMBOL_1(11, bitD4);
-        HUF_DECODE_SYMBOL_0(12, bitD1);
-        HUF_DECODE_SYMBOL_0(13, bitD2);
-        HUF_DECODE_SYMBOL_0(14, bitD3);
-        HUF_DECODE_SYMBOL_0(15, bitD4);
-    }
+        BYTE* const ostart = (BYTE*) dst;
+        BYTE* op = ostart;
+        BYTE* const omax = op + maxDstSize;
+        BYTE* const olimit = maxDstSize < 15 ? op : omax-15;
+
+        const void* ptr = DTable;
+        const HUF_DElt* const dt = (const HUF_DElt*)(ptr)+1;
+        const U32 dtLog = DTable[0];
+        size_t errorCode;
+        U32 reloadStatus;
+
+        /* Init */
+
+        const U16* jumpTable = (const U16*)cSrc;
+        const size_t length1 = FSE_readLE16(jumpTable);
+        const size_t length2 = FSE_readLE16(jumpTable+1);
+        const size_t length3 = FSE_readLE16(jumpTable+2);
+        const size_t length4 = cSrcSize - 6 - length1 - length2 - length3;   /* check coherency !! */
+        const char* const start1 = (const char*)(cSrc) + 6;
+        const char* const start2 = start1 + length1;
+        const char* const start3 = start2 + length2;
+        const char* const start4 = start3 + length3;
+        FSE_DStream_t bitD1, bitD2, bitD3, bitD4;
+
+        if (length1+length2+length3+6 >= cSrcSize) return (size_t)-FSE_ERROR_srcSize_wrong;
+
+        errorCode = FSE_initDStream(&bitD1, start1, length1);
+        if (FSE_isError(errorCode)) return errorCode;
+        errorCode = FSE_initDStream(&bitD2, start2, length2);
+        if (FSE_isError(errorCode)) return errorCode;
+        errorCode = FSE_initDStream(&bitD3, start3, length3);
+        if (FSE_isError(errorCode)) return errorCode;
+        errorCode = FSE_initDStream(&bitD4, start4, length4);
+        if (FSE_isError(errorCode)) return errorCode;
+
+        reloadStatus=FSE_reloadDStream(&bitD2);
+
+        /* 16 symbols per loop */
+        for ( ; (reloadStatus<FSE_DStream_completed) && (op<olimit);  /* D2-3-4 are supposed to be synchronized and finish together */
+            op+=16, reloadStatus = FSE_reloadDStream(&bitD2) | FSE_reloadDStream(&bitD3) | FSE_reloadDStream(&bitD4), FSE_reloadDStream(&bitD1))
+        {
+    #define HUF_DECODE_SYMBOL_0(n, Dstream) \
+            op[n] = HUF_decodeSymbol(&Dstream, dt, dtLog);
+
+    #define HUF_DECODE_SYMBOL_1(n, Dstream) \
+            op[n] = HUF_decodeSymbol(&Dstream, dt, dtLog); \
+            if (FSE_32bits() && (HUF_MAX_TABLELOG>12)) FSE_reloadDStream(&Dstream)
+
+    #define HUF_DECODE_SYMBOL_2(n, Dstream) \
+            op[n] = HUF_decodeSymbol(&Dstream, dt, dtLog); \
+            if (FSE_32bits()) FSE_reloadDStream(&Dstream)
+
+            HUF_DECODE_SYMBOL_1( 0, bitD1);
+            HUF_DECODE_SYMBOL_1( 1, bitD2);
+            HUF_DECODE_SYMBOL_1( 2, bitD3);
+            HUF_DECODE_SYMBOL_1( 3, bitD4);
+            HUF_DECODE_SYMBOL_2( 4, bitD1);
+            HUF_DECODE_SYMBOL_2( 5, bitD2);
+            HUF_DECODE_SYMBOL_2( 6, bitD3);
+            HUF_DECODE_SYMBOL_2( 7, bitD4);
+            HUF_DECODE_SYMBOL_1( 8, bitD1);
+            HUF_DECODE_SYMBOL_1( 9, bitD2);
+            HUF_DECODE_SYMBOL_1(10, bitD3);
+            HUF_DECODE_SYMBOL_1(11, bitD4);
+            HUF_DECODE_SYMBOL_0(12, bitD1);
+            HUF_DECODE_SYMBOL_0(13, bitD2);
+            HUF_DECODE_SYMBOL_0(14, bitD3);
+            HUF_DECODE_SYMBOL_0(15, bitD4);
+        }
 
-    if (reloadStatus!=FSE_DStream_completed)   /* not complete : some bitStream might be FSE_DStream_unfinished */
-        return (size_t)-FSE_ERROR_corruptionDetected;
+        if (reloadStatus!=FSE_DStream_completed)   /* not complete : some bitStream might be FSE_DStream_unfinished */
+            return (size_t)-FSE_ERROR_corruptionDetected;
 
-    /* tail */
-    {
-        // bitTail = bitD1;   // *much* slower : -20% !??!
-        FSE_DStream_t bitTail;
-        bitTail.ptr = bitD1.ptr;
-        bitTail.bitsConsumed = bitD1.bitsConsumed;
-        bitTail.bitContainer = bitD1.bitContainer;   // required in case of FSE_DStream_endOfBuffer
-        bitTail.start = start1;
-        for ( ; (FSE_reloadDStream(&bitTail) < FSE_DStream_completed) && (op<omax) ; op++)
+        /* tail */
         {
-            HUF_DECODE_SYMBOL_0(0, bitTail);
+            /* bitTail = bitD1; */   /* *much* slower : -20% !??! */
+            FSE_DStream_t bitTail;
+            bitTail.ptr = bitD1.ptr;
+            bitTail.bitsConsumed = bitD1.bitsConsumed;
+            bitTail.bitContainer = bitD1.bitContainer;   /* required in case of FSE_DStream_endOfBuffer */
+            bitTail.start = start1;
+            for ( ; (FSE_reloadDStream(&bitTail) < FSE_DStream_completed) && (op<omax) ; op++)
+            {
+                HUF_DECODE_SYMBOL_0(0, bitTail);
+            }
+
+            if (FSE_endOfDStream(&bitTail))
+                return op-ostart;
         }
 
-        if (FSE_endOfDStream(&bitTail))
-            return op-ostart;
-    }
+        if (op==omax) return (size_t)-FSE_ERROR_dstSize_tooSmall;   /* dst buffer is full, but cSrc unfinished */
 
-    if (op==omax) return (size_t)-FSE_ERROR_dstSize_tooSmall;   /* dst buffer is full, but cSrc unfinished */
-
-    return (size_t)-FSE_ERROR_corruptionDetected;
+        return (size_t)-FSE_ERROR_corruptionDetected;
+    }
 }
 
 
@@ -1336,6 +1339,8 @@ static const U32 ZSTD_magicNumber = 0xFD2FB51E;   /* 3rd version : seqNb header
 #define LITERAL_NOENTROPY 63
 #define COMMAND_NOENTROPY 7   /* to remove */
 
+#define ZSTD_CONTENTSIZE_ERROR   (0ULL - 2)
+
 static const size_t ZSTD_blockHeaderSize = 3;
 static const size_t ZSTD_frameHeaderSize = 4;
 
@@ -1353,8 +1358,6 @@ static unsigned ZSTD_isLittleEndian(void)
 
 static U16    ZSTD_read16(const void* p) { U16 r; memcpy(&r, p, sizeof(r)); return r; }
 
-static U32    ZSTD_read32(const void* p) { U32 r; memcpy(&r, p, sizeof(r)); return r; }
-
 static void   ZSTD_copy4(void* dst, const void* src) { memcpy(dst, src, 4); }
 
 static void   ZSTD_copy8(void* dst, const void* src) { memcpy(dst, src, 8); }
@@ -1379,16 +1382,9 @@ static U16 ZSTD_readLE16(const void* memPtr)
     }
 }
 
-
-static U32 ZSTD_readLE32(const void* memPtr)
+static U32 ZSTD_readLE24(const void* memPtr)
 {
-    if (ZSTD_isLittleEndian())
-        return ZSTD_read32(memPtr);
-    else
-    {
-        const BYTE* p = (const BYTE*)memPtr;
-        return (U32)((U32)p[0] + ((U32)p[1]<<8) + ((U32)p[2]<<16) + ((U32)p[3]<<24));
-    }
+    return ZSTD_readLE16(memPtr) + (((const BYTE*)memPtr)[2] << 16);
 }
 
 static U32 ZSTD_readBE32(const void* memPtr)
@@ -1487,7 +1483,9 @@ static size_t ZSTDv01_getcBlockSize(const void* src, size_t srcSize, blockProper
 static size_t ZSTD_copyUncompressedBlock(void* dst, size_t maxDstSize, const void* src, size_t srcSize)
 {
     if (srcSize > maxDstSize) return ERROR(dstSize_tooSmall);
-    memcpy(dst, src, srcSize);
+    if (srcSize > 0) {
+        memcpy(dst, src, srcSize);
+    }
     return srcSize;
 }
 
@@ -1506,7 +1504,7 @@ static size_t ZSTD_decompressLiterals(void* ctx,
     if (srcSize <= 3) return ERROR(corruption_detected);
 
     litSize = ip[1] + (ip[0]<<8);
-    litSize += ((ip[-3] >> 3) & 7) << 16;   // mmmmh....
+    litSize += ((ip[-3] >> 3) & 7) << 16;   /* mmmmh.... */
     op = oend - litSize;
 
     (void)ctx;
@@ -1545,7 +1543,9 @@ static size_t ZSTDv01_decodeLiteralsBlock(void* ctx,
             size_t rleSize = litbp.origSize;
             if (rleSize>maxDstSize) return ERROR(dstSize_tooSmall);
             if (!srcSize) return ERROR(srcSize_wrong);
-            memset(oend - rleSize, *ip, rleSize);
+            if (rleSize > 0) {
+                memset(oend - rleSize, *ip, rleSize);
+            }
             *litStart = oend - rleSize;
             *litSize = rleSize;
             ip++;
@@ -1702,13 +1702,13 @@ static void ZSTD_decodeSequence(seq_t* seq, seqState_t* seqState)
     seqState->prevOffset = seq->offset;
     if (litLength == MaxLL)
     {
-        U32 add = dumps<de ? *dumps++ : 0;
+        const U32 add = dumps<de ? *dumps++ : 0;
         if (add < 255) litLength += add;
         else
         {
             if (dumps<=(de-3))
             {
-                litLength = ZSTD_readLE32(dumps) & 0xFFFFFF;  /* no pb : dumps is always followed by seq tables > 1 byte */
+                litLength = ZSTD_readLE24(dumps);
                 dumps += 3;
             }
         }
@@ -1730,13 +1730,13 @@ static void ZSTD_decodeSequence(seq_t* seq, seqState_t* seqState)
     matchLength = FSE_decodeSymbol(&(seqState->stateML), &(seqState->DStream));
     if (matchLength == MaxML)
     {
-        U32 add = dumps<de ? *dumps++ : 0;
+        const U32 add = dumps<de ? *dumps++ : 0;
         if (add < 255) matchLength += add;
         else
         {
             if (dumps<=(de-3))
             {
-                matchLength = ZSTD_readLE32(dumps) & 0xFFFFFF;  /* no pb : dumps is always followed by seq tables > 1 byte */
+                matchLength = ZSTD_readLE24(dumps);
                 dumps += 3;
             }
         }
@@ -1757,7 +1757,7 @@ static size_t ZSTD_execSequence(BYTE* op,
                                 BYTE* const base, BYTE* const oend)
 {
     static const int dec32table[] = {0, 1, 2, 1, 4, 4, 4, 4};   /* added */
-    static const int dec64table[] = {8, 8, 8, 7, 8, 9,10,11};   /* substracted */
+    static const int dec64table[] = {8, 8, 8, 7, 8, 9,10,11};   /* subtracted */
     const BYTE* const ostart = op;
     const size_t litLength = sequence.litLength;
     BYTE* const endMatch = op + litLength + sequence.matchLength;    /* risk : address space overflow (32-bits) */
@@ -1905,8 +1905,10 @@ static size_t ZSTD_decompressSequences(
         {
             size_t lastLLSize = litEnd - litPtr;
             if (op+lastLLSize > oend) return ERROR(dstSize_tooSmall);
-            if (op != litPtr) memmove(op, litPtr, lastLLSize);
-            op += lastLLSize;
+            if (lastLLSize > 0) {
+                if (op != litPtr) memmove(op, litPtr, lastLLSize);
+                op += lastLLSize;
+            }
         }
     }
 
@@ -1999,36 +2001,59 @@ size_t ZSTDv01_decompress(void* dst, size_t maxDstSize, const void* src, size_t
     return ZSTDv01_decompressDCtx(&ctx, dst, maxDstSize, src, srcSize);
 }
 
-size_t ZSTDv01_findFrameCompressedSize(const void* src, size_t srcSize)
+/* ZSTD_errorFrameSizeInfoLegacy() :
+   assumes `cSize` and `dBound` are _not_ NULL */
+static void ZSTD_errorFrameSizeInfoLegacy(size_t* cSize, unsigned long long* dBound, size_t ret)
+{
+    *cSize = ret;
+    *dBound = ZSTD_CONTENTSIZE_ERROR;
+}
+
+void ZSTDv01_findFrameSizeInfoLegacy(const void *src, size_t srcSize, size_t* cSize, unsigned long long* dBound)
 {
     const BYTE* ip = (const BYTE*)src;
     size_t remainingSize = srcSize;
+    size_t nbBlocks = 0;
     U32 magicNumber;
     blockProperties_t blockProperties;
 
     /* Frame Header */
-    if (srcSize < ZSTD_frameHeaderSize+ZSTD_blockHeaderSize) return ERROR(srcSize_wrong);
+    if (srcSize < ZSTD_frameHeaderSize+ZSTD_blockHeaderSize) {
+        ZSTD_errorFrameSizeInfoLegacy(cSize, dBound, ERROR(srcSize_wrong));
+        return;
+    }
     magicNumber = ZSTD_readBE32(src);
-    if (magicNumber != ZSTD_magicNumber) return ERROR(prefix_unknown);
+    if (magicNumber != ZSTD_magicNumber) {
+        ZSTD_errorFrameSizeInfoLegacy(cSize, dBound, ERROR(prefix_unknown));
+        return;
+    }
     ip += ZSTD_frameHeaderSize; remainingSize -= ZSTD_frameHeaderSize;
 
     /* Loop on each block */
     while (1)
     {
         size_t blockSize = ZSTDv01_getcBlockSize(ip, remainingSize, &blockProperties);
-        if (ZSTDv01_isError(blockSize)) return blockSize;
+        if (ZSTDv01_isError(blockSize)) {
+            ZSTD_errorFrameSizeInfoLegacy(cSize, dBound, blockSize);
+            return;
+        }
 
         ip += ZSTD_blockHeaderSize;
         remainingSize -= ZSTD_blockHeaderSize;
-        if (blockSize > remainingSize) return ERROR(srcSize_wrong);
+        if (blockSize > remainingSize) {
+            ZSTD_errorFrameSizeInfoLegacy(cSize, dBound, ERROR(srcSize_wrong));
+            return;
+        }
 
         if (blockSize == 0) break;   /* bt_end */
 
         ip += blockSize;
         remainingSize -= blockSize;
+        nbBlocks++;
     }
 
-    return ip - (const BYTE*)src;
+    *cSize = ip - (const BYTE*)src;
+    *dBound = nbBlocks * BLOCKSIZE;
 }
 
 /*******************************