Update zstd to version 1.3.8

C/7zVersion.h

@@ -1,7 +1,7 @@
 #define MY_VER_MAJOR 18
 #define MY_VER_MINOR 05
 #define MY_VER_BUILD 0
-#define MY_VERSION_NUMBERS "18.05 ZS v1.3.7 R3"
+#define MY_VERSION_NUMBERS "18.05 ZS v1.3.8 R1"
 #define MY_VERSION MY_VERSION_NUMBERS
 
 #ifdef MY_CPU_NAME
@@ -10,7 +10,7 @@
   #define MY_VERSION_CPU MY_VERSION
 #endif
 
-#define MY_DATE "2018-11-27"
+#define MY_DATE "2018-12-28"
 #undef MY_COPYRIGHT
 #undef MY_VERSION_COPYRIGHT_DATE
 #define MY_AUTHOR_NAME "Igor Pavlov, Tino Reichardt"

C/zstd/README.md

@@ -7,13 +7,32 @@ in order to make it easier to select or exclude features.
 
 #### Building
 
-`Makefile` script is provided, supporting all standard [Makefile conventions](https://www.gnu.org/prep/standards/html_node/Makefile-Conventions.html#Makefile-Conventions),
+`Makefile` script is provided, supporting [Makefile conventions](https://www.gnu.org/prep/standards/html_node/Makefile-Conventions.html#Makefile-Conventions),
 including commands variables, staged install, directory variables and standard targets.
 - `make` : generates both static and dynamic libraries
-- `make install` : install libraries in default system directories
+- `make install` : install libraries and headers in target system directories
 
-`libzstd` default scope includes compression, decompression, dictionary building,
+`libzstd` default scope is pretty large, including compression, decompression, dictionary builder,
-and decoding support for legacy formats >= v0.5.0.
+and support for decoding legacy formats >= v0.5.0.
+The scope can be reduced on demand (see paragraph _modular build_).
+
+
+#### Multithreading support
+
+Multithreading is disabled by default when building with `make`.
+Enabling multithreading requires 2 conditions :
+- set build macro `ZSTD_MULTITHREAD` (`-DZSTD_MULTITHREAD` for `gcc`)
+- for POSIX systems : compile with pthread (`-pthread` compilation flag for `gcc`)
+
+Both conditions are automatically applied when invoking `make lib-mt` target.
+
+When linking a POSIX program with a multithreaded version of `libzstd`,
+note that it's necessary to request the `-pthread` flag during link stage.
+
+Multithreading capabilities are exposed
+via the [advanced API defined in `lib/zstd.h`](https://github.com/facebook/zstd/blob/v1.3.8/lib/zstd.h#L592).
+This API is still labelled experimental,
+but is expected to become "stable" in the near future.
 
 
 #### API
@@ -26,63 +45,70 @@ Zstandard's stable API is exposed within [lib/zstd.h](zstd.h).
 Optional advanced features are exposed via :
 
 - `lib/common/zstd_errors.h` : translates `size_t` function results
-                              into an `ZSTD_ErrorCode`, for accurate error handling.
+                               into a `ZSTD_ErrorCode`, for accurate error handling.
+
 - `ZSTD_STATIC_LINKING_ONLY` : if this macro is defined _before_ including `zstd.h`,
-                          it unlocks access to advanced experimental API,
+                          it unlocks access to the experimental API,
-                          exposed in second part of `zstd.h`.
+                          exposed in the second part of `zstd.h`.
-                          These APIs are not "stable", their definition may change in the future.
+                          All definitions in the experimental APIs are unstable,
-                          As a consequence, it shall ___never be used with dynamic library___ !
+                          they may still change in the future, or even be removed.
+                          As a consequence, experimental definitions shall ___never be used with dynamic library___ !
                           Only static linking is allowed.
 
 
 #### Modular build
 
-It's possible to compile only a limited set of features.
+It's possible to compile only a limited set of features within `libzstd`.
+The file structure is designed to make this selection manually achievable for any build system :
 
 - Directory `lib/common` is always required, for all variants.
+
 - Compression source code lies in `lib/compress`
+
 - Decompression source code lies in `lib/decompress`
+
 - It's possible to include only `compress` or only `decompress`, they don't depend on each other.
+
 - `lib/dictBuilder` : makes it possible to generate dictionaries from a set of samples.
         The API is exposed in `lib/dictBuilder/zdict.h`.
         This module depends on both `lib/common` and `lib/compress` .
-- `lib/legacy` : source code to decompress legacy zstd formats, starting from `v0.1.0`.
+
+- `lib/legacy` : makes it possible to decompress legacy zstd formats, starting from `v0.1.0`.
         This module depends on `lib/common` and `lib/decompress`.
         To enable this feature, define `ZSTD_LEGACY_SUPPORT` during compilation.
         Specifying a number limits versions supported to that version onward.
         For example, `ZSTD_LEGACY_SUPPORT=2` means : "support legacy formats >= v0.2.0".
-        `ZSTD_LEGACY_SUPPORT=3` means : "support legacy formats >= v0.3.0", and so on.
+        Conversely, `ZSTD_LEGACY_SUPPORT=0` means "do __not__ support legacy formats".
-        Currently, the default library setting is `ZST_LEGACY_SUPPORT=5`.
+        By default, this build macro is set as `ZSTD_LEGACY_SUPPORT=5`.
-        It can be changed at build by any other value.
+        Decoding supported legacy format is a transparent capability triggered within decompression functions.
-        Note that any number >= 8 translates into "do __not__ support legacy formats",
+        It's also allowed to invoke legacy API directly, exposed in `lib/legacy/zstd_legacy.h`.
-        since all versions of `zstd` >= v0.8 are compatible with v1+ specification.
+        Each version does also provide its own set of advanced API.
-        `ZSTD_LEGACY_SUPPORT=0` also means "do __not__ support legacy formats".
-        Once enabled, this capability is transparently triggered within decompression functions.
-        It's also possible to invoke directly legacy API, as exposed in `lib/legacy/zstd_legacy.h`.
-        Each version also provides an additional dedicated set of advanced API.
         For example, advanced API for version `v0.4` is exposed in `lib/legacy/zstd_v04.h` .
-        Note : `lib/legacy` only supports _decoding_ legacy formats.
-- Similarly, you can define `ZSTD_LIB_COMPRESSION, ZSTD_LIB_DECOMPRESSION`, `ZSTD_LIB_DICTBUILDER`,
-        and `ZSTD_LIB_DEPRECATED` as 0 to forgo compilation of the corresponding features. This will
-        also disable compilation of all dependencies (eg. `ZSTD_LIB_COMPRESSION=0` will also disable
-        dictBuilder).
 
+- While invoking `make libzstd`, it's possible to define build macros
+        `ZSTD_LIB_COMPRESSION, ZSTD_LIB_DECOMPRESSION`, `ZSTD_LIB_DICTBUILDER`,
+        and `ZSTD_LIB_DEPRECATED` as `0` to forgo compilation of the corresponding features.
+        This will also disable compilation of all dependencies
+        (eg. `ZSTD_LIB_COMPRESSION=0` will also disable dictBuilder).
 
-#### Multithreading support
+- There are some additional build macros that can be used to minify the decoder.
 
-Multithreading is disabled by default when building with `make`.
+  Zstandard often has more than one implementation of a piece of functionality,
-Enabling multithreading requires 2 conditions :
+  where each implementation optimizes for different scenarios. For example, the
-- set macro `ZSTD_MULTITHREAD`
+  Huffman decoder has complementary implementations that decode the stream one
-- on POSIX systems : compile with pthread (`-pthread` compilation flag for `gcc`)
+  symbol at a time or two symbols at a time. Zstd normally includes both (and
+  dispatches between them at runtime), but by defining `HUF_FORCE_DECOMPRESS_X1`
+  or `HUF_FORCE_DECOMPRESS_X2`, you can force the use of one or the other, avoiding
+  compilation of the other. Similarly, `ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT`
+  and `ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG` force the compilation and use of
+  only one or the other of two decompression implementations. The smallest
+  binary is achieved by using `HUF_FORCE_DECOMPRESS_X1` and
+  `ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT`.
 
-Both conditions are automatically triggered by invoking `make lib-mt` target.
+  For squeezing the last ounce of size out, you can also define
-Note that, when linking a POSIX program with a multithreaded version of `libzstd`,
+  `ZSTD_NO_INLINE`, which disables inlining, and `ZSTD_STRIP_ERROR_STRINGS`,
-it's necessary to trigger `-pthread` flag during link stage.
+  which removes the error messages that are otherwise returned by
-
+  `ZSTD_getErrorName`.
-Multithreading capabilities are exposed
-via [advanced API `ZSTD_compress_generic()` defined in `lib/zstd.h`](https://github.com/facebook/zstd/blob/dev/lib/zstd.h#L919).
-This API is still considered experimental,
-but is expected to become "stable" at some point in the future.
 
 
 #### Windows : using MinGW+MSYS to create DLL
@@ -113,8 +139,8 @@ Consider migrating code towards supported streaming API exposed in `zstd.h`.
 
 The other files are not source code. There are :
 
- - `LICENSE` : contains the BSD license text
- - `Makefile` : `make` script to build and install zstd library (static and dynamic)
  - `BUCK` : support for `buck` build system (https://buckbuild.com/)
- - `libzstd.pc.in` : for `pkg-config` (used in `make install`)
+ - `Makefile` : `make` script to build and install zstd library (static and dynamic)
  - `README.md` : this file
+ - `dll/` : resources directory for Windows compilation
+ - `libzstd.pc.in` : script for `pkg-config` (used in `make install`)

C/zstd/bitstream.h

@@ -389,7 +389,7 @@ MEM_STATIC void BIT_skipBits(BIT_DStream_t* bitD, U32 nbBits)
  *  Read (consume) next n bits from local register and update.
  *  Pay attention to not read more than nbBits contained into local register.
  * @return : extracted value. */
-MEM_STATIC size_t BIT_readBits(BIT_DStream_t* bitD, U32 nbBits)
+MEM_STATIC size_t BIT_readBits(BIT_DStream_t* bitD, unsigned nbBits)
 {
     size_t const value = BIT_lookBits(bitD, nbBits);
     BIT_skipBits(bitD, nbBits);
@@ -398,7 +398,7 @@ MEM_STATIC size_t BIT_readBits(BIT_DStream_t* bitD, U32 nbBits)
 
 /*! BIT_readBitsFast() :
  *  unsafe version; only works only if nbBits >= 1 */
-MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, U32 nbBits)
+MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, unsigned nbBits)
 {
     size_t const value = BIT_lookBitsFast(bitD, nbBits);
     assert(nbBits >= 1);

C/zstd/compiler.h

@@ -15,6 +15,8 @@
 *  Compiler specifics
 *********************************************************/
 /* force inlining */
+
+#if !defined(ZSTD_NO_INLINE)
 #if defined (__GNUC__) || defined(__cplusplus) || defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L   /* C99 */
 #  define INLINE_KEYWORD inline
 #else
@@ -29,6 +31,13 @@
 #  define FORCE_INLINE_ATTR
 #endif
 
+#else
+
+#define INLINE_KEYWORD
+#define FORCE_INLINE_ATTR
+
+#endif
+
 /**
  * FORCE_INLINE_TEMPLATE is used to define C "templates", which take constant
  * parameters. They must be inlined for the compiler to elimininate the constant
@@ -89,23 +98,21 @@
 #endif
 
 /* prefetch
- * can be disabled, by declaring NO_PREFETCH macro
+ * can be disabled, by declaring NO_PREFETCH build macro */
- * All prefetch invocations use a single default locality 2,
- * generating instruction prefetcht1,
- * which, according to Intel, means "load data into L2 cache".
- * This is a good enough "middle ground" for the time being,
- * though in theory, it would be better to specialize locality depending on data being prefetched.
- * Tests could not determine any sensible difference based on locality value. */
 #if defined(NO_PREFETCH)
-#  define PREFETCH(ptr)     (void)(ptr)  /* disabled */
+#  define PREFETCH_L1(ptr)  (void)(ptr)  /* disabled */
+#  define PREFETCH_L2(ptr)  (void)(ptr)  /* disabled */
 #else
 #  if defined(_MSC_VER) && (defined(_M_X64) || defined(_M_I86))  /* _mm_prefetch() is not defined outside of x86/x64 */
 #    include <mmintrin.h>   /* https://msdn.microsoft.com/fr-fr/library/84szxsww(v=vs.90).aspx */
-#    define PREFETCH(ptr)   _mm_prefetch((const char*)(ptr), _MM_HINT_T1)
+#    define PREFETCH_L1(ptr)  _mm_prefetch((const char*)(ptr), _MM_HINT_T0)
+#    define PREFETCH_L2(ptr)  _mm_prefetch((const char*)(ptr), _MM_HINT_T1)
 #  elif defined(__GNUC__) && ( (__GNUC__ >= 4) || ( (__GNUC__ == 3) && (__GNUC_MINOR__ >= 1) ) )
-#    define PREFETCH(ptr)   __builtin_prefetch((ptr), 0 /* rw==read */, 2 /* locality */)
+#    define PREFETCH_L1(ptr)  __builtin_prefetch((ptr), 0 /* rw==read */, 3 /* locality */)
+#    define PREFETCH_L2(ptr)  __builtin_prefetch((ptr), 0 /* rw==read */, 2 /* locality */)
 #  else
-#    define PREFETCH(ptr)   (void)(ptr)  /* disabled */
+#    define PREFETCH_L1(ptr) (void)(ptr)  /* disabled */
+#    define PREFETCH_L2(ptr) (void)(ptr)  /* disabled */
 #  endif
 #endif  /* NO_PREFETCH */
 
@@ -116,7 +123,7 @@
     size_t const _size = (size_t)(s);     \
     size_t _pos;                          \
     for (_pos=0; _pos<_size; _pos+=CACHELINE_SIZE) {  \
-        PREFETCH(_ptr + _pos);            \
+        PREFETCH_L2(_ptr + _pos);         \
     }                                     \
 }
 

C/zstd/cpu.h

@@ -78,7 +78,7 @@ MEM_STATIC ZSTD_cpuid_t ZSTD_cpuid(void) {
       __asm__(
           "pushl %%ebx\n\t"
           "cpuid\n\t"
-          "movl %%ebx, %%eax\n\r"
+          "movl %%ebx, %%eax\n\t"
           "popl %%ebx"
           : "=a"(f7b), "=c"(f7c)
           : "a"(7), "c"(0)

C/zstd/debug.h

@@ -57,9 +57,9 @@ extern "C" {
 #endif
 
 
-/* static assert is triggered at compile time, leaving no runtime artefact,
+/* static assert is triggered at compile time, leaving no runtime artefact.
- * but can only work with compile-time constants.
+ * static assert only works with compile-time constants.
- * This variant can only be used inside a function. */
+ * Also, this variant can only be used inside a function. */
 #define DEBUG_STATIC_ASSERT(c) (void)sizeof(char[(c) ? 1 : -1])
 
 
@@ -70,9 +70,19 @@ extern "C" {
 #  define DEBUGLEVEL 0
 #endif
 
+
+/* DEBUGFILE can be defined externally,
+ * typically through compiler command line.
+ * note : currently useless.
+ * Value must be stderr or stdout */
+#ifndef DEBUGFILE
+#  define DEBUGFILE stderr
+#endif
+
+
 /* recommended values for DEBUGLEVEL :
- * 0 : no debug, all run-time functions disabled
+ * 0 : release mode, no debug, all run-time checks disabled
- * 1 : no display, enables assert() only
+ * 1 : enables assert() only, no display
  * 2 : reserved, for currently active debug path
  * 3 : events once per object lifetime (CCtx, CDict, etc.)
  * 4 : events once per frame
@@ -81,7 +91,7 @@ extern "C" {
  * 7+: events at every position (*very* verbose)
  *
  * It's generally inconvenient to output traces > 5.
- * In which case, it's possible to selectively enable higher verbosity levels
+ * In which case, it's possible to selectively trigger high verbosity levels
  * by modifying g_debug_level.
  */
 
@@ -95,11 +105,12 @@ extern "C" {
 
 #if (DEBUGLEVEL>=2)
 #  include <stdio.h>
-extern int g_debuglevel; /* here, this variable is only declared,
+extern int g_debuglevel; /* the variable is only declared,
-                           it actually lives in debug.c,
+                            it actually lives in debug.c,
-                           and is shared by the whole process.
+                            and is shared by the whole process.
-                           It's typically used to enable very verbose levels
+                            It's not thread-safe.
-                           on selective conditions (such as position in src) */
+                            It's useful when enabling very verbose levels
+                            on selective conditions (such as position in src) */
 
 #  define RAWLOG(l, ...) {                                      \
                 if (l<=g_debuglevel) {                          \

C/zstd/error_private.c

@@ -14,6 +14,10 @@
 
 const char* ERR_getErrorString(ERR_enum code)
 {
+#ifdef ZSTD_STRIP_ERROR_STRINGS
+    (void)code;
+    return "Error strings stripped";
+#else
     static const char* const notErrorCode = "Unspecified error code";
     switch( code )
     {
@@ -39,10 +43,12 @@ const char* ERR_getErrorString(ERR_enum code)
     case PREFIX(dictionaryCreation_failed): return "Cannot create Dictionary from provided samples";
     case PREFIX(dstSize_tooSmall): return "Destination buffer is too small";
     case PREFIX(srcSize_wrong): return "Src size is incorrect";
+    case PREFIX(dstBuffer_null): return "Operation on NULL destination buffer";
         /* following error codes are not stable and may be removed or changed in a future version */
     case PREFIX(frameIndex_tooLarge): return "Frame index is too large";
     case PREFIX(seekableIO): return "An I/O error occurred when reading/seeking";
     case PREFIX(maxCode):
     default: return notErrorCode;
     }
+#endif
 }

C/zstd/fse.h

@@ -512,7 +512,7 @@ MEM_STATIC void FSE_initCState(FSE_CState_t* statePtr, const FSE_CTable* ct)
     const U32 tableLog = MEM_read16(ptr);
     statePtr->value = (ptrdiff_t)1<<tableLog;
     statePtr->stateTable = u16ptr+2;
-    statePtr->symbolTT = ((const U32*)ct + 1 + (tableLog ? (1<<(tableLog-1)) : 1));
+    statePtr->symbolTT = ct + 1 + (tableLog ? (1<<(tableLog-1)) : 1);
     statePtr->stateLog = tableLog;
 }
 
@@ -531,7 +531,7 @@ MEM_STATIC void FSE_initCState2(FSE_CState_t* statePtr, const FSE_CTable* ct, U3
     }
 }
 
-MEM_STATIC void FSE_encodeSymbol(BIT_CStream_t* bitC, FSE_CState_t* statePtr, U32 symbol)
+MEM_STATIC void FSE_encodeSymbol(BIT_CStream_t* bitC, FSE_CState_t* statePtr, unsigned symbol)
 {
     FSE_symbolCompressionTransform const symbolTT = ((const FSE_symbolCompressionTransform*)(statePtr->symbolTT))[symbol];
     const U16* const stateTable = (const U16*)(statePtr->stateTable);

C/zstd/fse_compress.c

@@ -115,7 +115,7 @@ size_t FSE_buildCTable_wksp(FSE_CTable* ct,
     /* symbol start positions */
     {   U32 u;
         cumul[0] = 0;
-        for (u=1; u<=maxSymbolValue+1; u++) {
+        for (u=1; u <= maxSymbolValue+1; u++) {
             if (normalizedCounter[u-1]==-1) {  /* Low proba symbol */
                 cumul[u] = cumul[u-1] + 1;
                 tableSymbol[highThreshold--] = (FSE_FUNCTION_TYPE)(u-1);
@@ -658,7 +658,7 @@ size_t FSE_compress_wksp (void* dst, size_t dstSize, const void* src, size_t src
     BYTE* op = ostart;
     BYTE* const oend = ostart + dstSize;
 
-    U32   count[FSE_MAX_SYMBOL_VALUE+1];
+    unsigned count[FSE_MAX_SYMBOL_VALUE+1];
     S16   norm[FSE_MAX_SYMBOL_VALUE+1];
     FSE_CTable* CTable = (FSE_CTable*)workSpace;
     size_t const CTableSize = FSE_CTABLE_SIZE_U32(tableLog, maxSymbolValue);
@@ -672,7 +672,7 @@ size_t FSE_compress_wksp (void* dst, size_t dstSize, const void* src, size_t src
     if (!tableLog) tableLog = FSE_DEFAULT_TABLELOG;
 
     /* Scan input and build symbol stats */
-    {   CHECK_V_F(maxCount, HIST_count_wksp(count, &maxSymbolValue, src, srcSize, (unsigned*)scratchBuffer) );
+    {   CHECK_V_F(maxCount, HIST_count_wksp(count, &maxSymbolValue, src, srcSize, scratchBuffer, scratchBufferSize) );
         if (maxCount == srcSize) return 1;   /* only a single symbol in src : rle */
         if (maxCount == 1) return 0;         /* each symbol present maximum once => not compressible */
         if (maxCount < (srcSize >> 7)) return 0;   /* Heuristic : not compressible enough */

C/zstd/hist.c

@@ -73,6 +73,7 @@ unsigned HIST_count_simple(unsigned* count, unsigned* maxSymbolValuePtr,
     return largestCount;
 }
 
+typedef enum { trustInput, checkMaxSymbolValue } HIST_checkInput_e;
 
 /* HIST_count_parallel_wksp() :
  * store histogram into 4 intermediate tables, recombined at the end.
@@ -85,8 +86,8 @@ unsigned HIST_count_simple(unsigned* count, unsigned* maxSymbolValuePtr,
 static size_t HIST_count_parallel_wksp(
                                 unsigned* count, unsigned* maxSymbolValuePtr,
                                 const void* source, size_t sourceSize,
-                                unsigned checkMax,
+                                HIST_checkInput_e check,
-                                unsigned* const workSpace)
+                                U32* const workSpace)
 {
     const BYTE* ip = (const BYTE*)source;
     const BYTE* const iend = ip+sourceSize;
@@ -137,7 +138,7 @@ static size_t HIST_count_parallel_wksp(
     /* finish last symbols */
     while (ip<iend) Counting1[*ip++]++;
 
-    if (checkMax) {   /* verify stats will fit into destination table */
+    if (check) {   /* verify stats will fit into destination table */
         U32 s; for (s=255; s>maxSymbolValue; s--) {
             Counting1[s] += Counting2[s] + Counting3[s] + Counting4[s];
             if (Counting1[s]) return ERROR(maxSymbolValue_tooSmall);
@@ -157,14 +158,18 @@ static size_t HIST_count_parallel_wksp(
 
 /* HIST_countFast_wksp() :
  * Same as HIST_countFast(), but using an externally provided scratch buffer.
- * `workSpace` size must be table of >= HIST_WKSP_SIZE_U32 unsigned */
+ * `workSpace` is a writable buffer which must be 4-bytes aligned,
+ * `workSpaceSize` must be >= HIST_WKSP_SIZE
+ */
 size_t HIST_countFast_wksp(unsigned* count, unsigned* maxSymbolValuePtr,
                           const void* source, size_t sourceSize,
-                          unsigned* workSpace)
+                          void* workSpace, size_t workSpaceSize)
 {
     if (sourceSize < 1500) /* heuristic threshold */
         return HIST_count_simple(count, maxSymbolValuePtr, source, sourceSize);
-    return HIST_count_parallel_wksp(count, maxSymbolValuePtr, source, sourceSize, 0, workSpace);
+    if ((size_t)workSpace & 3) return ERROR(GENERIC);  /* must be aligned on 4-bytes boundaries */
+    if (workSpaceSize < HIST_WKSP_SIZE) return ERROR(workSpace_tooSmall);
+    return HIST_count_parallel_wksp(count, maxSymbolValuePtr, source, sourceSize, trustInput, (U32*)workSpace);
 }
 
 /* fast variant (unsafe : won't check if src contains values beyond count[] limit) */
@@ -172,24 +177,27 @@ size_t HIST_countFast(unsigned* count, unsigned* maxSymbolValuePtr,
                      const void* source, size_t sourceSize)
 {
     unsigned tmpCounters[HIST_WKSP_SIZE_U32];
-    return HIST_countFast_wksp(count, maxSymbolValuePtr, source, sourceSize, tmpCounters);
+    return HIST_countFast_wksp(count, maxSymbolValuePtr, source, sourceSize, tmpCounters, sizeof(tmpCounters));
 }
 
 /* HIST_count_wksp() :
  * Same as HIST_count(), but using an externally provided scratch buffer.
  * `workSpace` size must be table of >= HIST_WKSP_SIZE_U32 unsigned */
 size_t HIST_count_wksp(unsigned* count, unsigned* maxSymbolValuePtr,
-                 const void* source, size_t sourceSize, unsigned* workSpace)
+                       const void* source, size_t sourceSize,
+                       void* workSpace, size_t workSpaceSize)
 {
+    if ((size_t)workSpace & 3) return ERROR(GENERIC);  /* must be aligned on 4-bytes boundaries */
+    if (workSpaceSize < HIST_WKSP_SIZE) return ERROR(workSpace_tooSmall);
     if (*maxSymbolValuePtr < 255)
-        return HIST_count_parallel_wksp(count, maxSymbolValuePtr, source, sourceSize, 1, workSpace);
+        return HIST_count_parallel_wksp(count, maxSymbolValuePtr, source, sourceSize, checkMaxSymbolValue, (U32*)workSpace);
     *maxSymbolValuePtr = 255;
-    return HIST_countFast_wksp(count, maxSymbolValuePtr, source, sourceSize, workSpace);
+    return HIST_countFast_wksp(count, maxSymbolValuePtr, source, sourceSize, workSpace, workSpaceSize);
 }
 
 size_t HIST_count(unsigned* count, unsigned* maxSymbolValuePtr,
                  const void* src, size_t srcSize)
 {
     unsigned tmpCounters[HIST_WKSP_SIZE_U32];
-    return HIST_count_wksp(count, maxSymbolValuePtr, src, srcSize, tmpCounters);
+    return HIST_count_wksp(count, maxSymbolValuePtr, src, srcSize, tmpCounters, sizeof(tmpCounters));
 }

C/zstd/hist.h

@@ -41,11 +41,11 @@
 
 /*! HIST_count():
  *  Provides the precise count of each byte within a table 'count'.
- *  'count' is a table of unsigned int, of minimum size (*maxSymbolValuePtr+1).
+ * 'count' is a table of unsigned int, of minimum size (*maxSymbolValuePtr+1).
  *  Updates *maxSymbolValuePtr with actual largest symbol value detected.
- *  @return : count of the most frequent symbol (which isn't identified).
+ * @return : count of the most frequent symbol (which isn't identified).
- *            or an error code, which can be tested using HIST_isError().
+ *           or an error code, which can be tested using HIST_isError().
- *            note : if return == srcSize, there is only one symbol.
+ *           note : if return == srcSize, there is only one symbol.
  */
 size_t HIST_count(unsigned* count, unsigned* maxSymbolValuePtr,
                   const void* src, size_t srcSize);
@@ -56,14 +56,16 @@ unsigned HIST_isError(size_t code);  /**< tells if a return value is an error co
 /* --- advanced histogram functions --- */
 
 #define HIST_WKSP_SIZE_U32 1024
+#define HIST_WKSP_SIZE    (HIST_WKSP_SIZE_U32 * sizeof(unsigned))
 /** HIST_count_wksp() :
  *  Same as HIST_count(), but using an externally provided scratch buffer.
  *  Benefit is this function will use very little stack space.
- * `workSpace` must be a table of unsigned of size >= HIST_WKSP_SIZE_U32
+ * `workSpace` is a writable buffer which must be 4-bytes aligned,
+ * `workSpaceSize` must be >= HIST_WKSP_SIZE
  */
 size_t HIST_count_wksp(unsigned* count, unsigned* maxSymbolValuePtr,
                        const void* src, size_t srcSize,
-                       unsigned* workSpace);
+                       void* workSpace, size_t workSpaceSize);
 
 /** HIST_countFast() :
  *  same as HIST_count(), but blindly trusts that all byte values within src are <= *maxSymbolValuePtr.
@@ -74,11 +76,12 @@ size_t HIST_countFast(unsigned* count, unsigned* maxSymbolValuePtr,
 
 /** HIST_countFast_wksp() :
  *  Same as HIST_countFast(), but using an externally provided scratch buffer.
- * `workSpace` must be a table of unsigned of size >= HIST_WKSP_SIZE_U32
+ * `workSpace` is a writable buffer which must be 4-bytes aligned,
+ * `workSpaceSize` must be >= HIST_WKSP_SIZE
  */
 size_t HIST_countFast_wksp(unsigned* count, unsigned* maxSymbolValuePtr,
                            const void* src, size_t srcSize,
-                           unsigned* workSpace);
+                           void* workSpace, size_t workSpaceSize);
 
 /*! HIST_count_simple() :
  *  Same as HIST_countFast(), this function is unsafe,

C/zstd/huf.h

@@ -173,15 +173,19 @@ typedef U32 HUF_DTable;
 *  Advanced decompression functions
 ******************************************/
 size_t HUF_decompress4X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /**< single-symbol decoder */
+#ifndef HUF_FORCE_DECOMPRESS_X1
 size_t HUF_decompress4X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /**< double-symbols decoder */
+#endif
 
 size_t HUF_decompress4X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /**< decodes RLE and uncompressed */
 size_t HUF_decompress4X_hufOnly(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< considers RLE and uncompressed as errors */
 size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /**< considers RLE and uncompressed as errors */
 size_t HUF_decompress4X1_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /**< single-symbol decoder */
 size_t HUF_decompress4X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /**< single-symbol decoder */
+#ifndef HUF_FORCE_DECOMPRESS_X1
 size_t HUF_decompress4X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /**< double-symbols decoder */
 size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /**< double-symbols decoder */
+#endif
 
 
 /* ****************************************
@@ -228,7 +232,7 @@ size_t HUF_compress4X_repeat(void* dst, size_t dstSize,
 #define HUF_CTABLE_WORKSPACE_SIZE_U32 (2*HUF_SYMBOLVALUE_MAX +1 +1)
 #define HUF_CTABLE_WORKSPACE_SIZE (HUF_CTABLE_WORKSPACE_SIZE_U32 * sizeof(unsigned))
 size_t HUF_buildCTable_wksp (HUF_CElt* tree,
-                       const U32* count, U32 maxSymbolValue, U32 maxNbBits,
+                       const unsigned* count, U32 maxSymbolValue, U32 maxNbBits,
                              void* workSpace, size_t wkspSize);
 
 /*! HUF_readStats() :
@@ -277,14 +281,22 @@ U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize);
 #define HUF_DECOMPRESS_WORKSPACE_SIZE (2 << 10)
 #define HUF_DECOMPRESS_WORKSPACE_SIZE_U32 (HUF_DECOMPRESS_WORKSPACE_SIZE / sizeof(U32))
 
+#ifndef HUF_FORCE_DECOMPRESS_X2
 size_t HUF_readDTableX1 (HUF_DTable* DTable, const void* src, size_t srcSize);
 size_t HUF_readDTableX1_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
+#endif
+#ifndef HUF_FORCE_DECOMPRESS_X1
 size_t HUF_readDTableX2 (HUF_DTable* DTable, const void* src, size_t srcSize);
 size_t HUF_readDTableX2_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
+#endif
 
 size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#ifndef HUF_FORCE_DECOMPRESS_X2
 size_t HUF_decompress4X1_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#endif
+#ifndef HUF_FORCE_DECOMPRESS_X1
 size_t HUF_decompress4X2_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#endif
 
 
 /* ====================== */
@@ -306,24 +318,36 @@ size_t HUF_compress1X_repeat(void* dst, size_t dstSize,
                        HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2);
 
 size_t HUF_decompress1X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /* single-symbol decoder */
+#ifndef HUF_FORCE_DECOMPRESS_X1
 size_t HUF_decompress1X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /* double-symbol decoder */
+#endif
 
 size_t HUF_decompress1X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);
 size_t HUF_decompress1X_DCtx_wksp (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);
+#ifndef HUF_FORCE_DECOMPRESS_X2
 size_t HUF_decompress1X1_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /**< single-symbol decoder */
 size_t HUF_decompress1X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /**< single-symbol decoder */
+#endif
+#ifndef HUF_FORCE_DECOMPRESS_X1
 size_t HUF_decompress1X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /**< double-symbols decoder */
 size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /**< double-symbols decoder */
+#endif
 
 size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);   /**< automatic selection of sing or double symbol decoder, based on DTable */
+#ifndef HUF_FORCE_DECOMPRESS_X2
 size_t HUF_decompress1X1_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#endif
+#ifndef HUF_FORCE_DECOMPRESS_X1
 size_t HUF_decompress1X2_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+#endif
 
 /* BMI2 variants.
  * If the CPU has BMI2 support, pass bmi2=1, otherwise pass bmi2=0.
  */
 size_t HUF_decompress1X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2);
+#ifndef HUF_FORCE_DECOMPRESS_X2
 size_t HUF_decompress1X1_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2);
+#endif
 size_t HUF_decompress4X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2);
 size_t HUF_decompress4X_hufOnly_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2);
 

C/zstd/huf_compress.c

@@ -88,13 +88,13 @@ static size_t HUF_compressWeights (void* dst, size_t dstSize, const void* weight
     BYTE* op = ostart;
     BYTE* const oend = ostart + dstSize;
 
-    U32 maxSymbolValue = HUF_TABLELOG_MAX;
+    unsigned maxSymbolValue = HUF_TABLELOG_MAX;
     U32 tableLog = MAX_FSE_TABLELOG_FOR_HUFF_HEADER;
 
     FSE_CTable CTable[FSE_CTABLE_SIZE_U32(MAX_FSE_TABLELOG_FOR_HUFF_HEADER, HUF_TABLELOG_MAX)];
     BYTE scratchBuffer[1<<MAX_FSE_TABLELOG_FOR_HUFF_HEADER];
 
-    U32 count[HUF_TABLELOG_MAX+1];
+    unsigned count[HUF_TABLELOG_MAX+1];
     S16 norm[HUF_TABLELOG_MAX+1];
 
     /* init conditions */
@@ -134,7 +134,7 @@ struct HUF_CElt_s {
     `CTable` : Huffman tree to save, using huf representation.
     @return : size of saved CTable */
 size_t HUF_writeCTable (void* dst, size_t maxDstSize,
-                        const HUF_CElt* CTable, U32 maxSymbolValue, U32 huffLog)
+                        const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog)
 {
     BYTE bitsToWeight[HUF_TABLELOG_MAX + 1];   /* precomputed conversion table */
     BYTE huffWeight[HUF_SYMBOLVALUE_MAX];
@@ -169,7 +169,7 @@ size_t HUF_writeCTable (void* dst, size_t maxDstSize,
 }
 
 
-size_t HUF_readCTable (HUF_CElt* CTable, U32* maxSymbolValuePtr, const void* src, size_t srcSize)
+size_t HUF_readCTable (HUF_CElt* CTable, unsigned* maxSymbolValuePtr, const void* src, size_t srcSize)
 {
     BYTE huffWeight[HUF_SYMBOLVALUE_MAX + 1];   /* init not required, even though some static analyzer may complain */
     U32 rankVal[HUF_TABLELOG_ABSOLUTEMAX + 1];   /* large enough for values from 0 to 16 */
@@ -315,7 +315,7 @@ typedef struct {
     U32 current;
 } rankPos;
 
-static void HUF_sort(nodeElt* huffNode, const U32* count, U32 maxSymbolValue)
+static void HUF_sort(nodeElt* huffNode, const unsigned* count, U32 maxSymbolValue)
 {
     rankPos rank[32];
     U32 n;
@@ -347,7 +347,7 @@ static void HUF_sort(nodeElt* huffNode, const U32* count, U32 maxSymbolValue)
  */
 #define STARTNODE (HUF_SYMBOLVALUE_MAX+1)
 typedef nodeElt huffNodeTable[HUF_CTABLE_WORKSPACE_SIZE_U32];
-size_t HUF_buildCTable_wksp (HUF_CElt* tree, const U32* count, U32 maxSymbolValue, U32 maxNbBits, void* workSpace, size_t wkspSize)
+size_t HUF_buildCTable_wksp (HUF_CElt* tree, const unsigned* count, U32 maxSymbolValue, U32 maxNbBits, void* workSpace, size_t wkspSize)
 {
     nodeElt* const huffNode0 = (nodeElt*)workSpace;
     nodeElt* const huffNode = huffNode0+1;
@@ -421,7 +421,7 @@ size_t HUF_buildCTable_wksp (HUF_CElt* tree, const U32* count, U32 maxSymbolValu
  * @return : maxNbBits
  *  Note : count is used before tree is written, so they can safely overlap
  */
-size_t HUF_buildCTable (HUF_CElt* tree, const U32* count, U32 maxSymbolValue, U32 maxNbBits)
+size_t HUF_buildCTable (HUF_CElt* tree, const unsigned* count, unsigned maxSymbolValue, unsigned maxNbBits)
 {
     huffNodeTable nodeTable;
     return HUF_buildCTable_wksp(tree, count, maxSymbolValue, maxNbBits, nodeTable, sizeof(nodeTable));
@@ -610,13 +610,14 @@ size_t HUF_compress4X_usingCTable(void* dst, size_t dstSize, const void* src, si
     return HUF_compress4X_usingCTable_internal(dst, dstSize, src, srcSize, CTable, /* bmi2 */ 0);
 }
 
+typedef enum { HUF_singleStream, HUF_fourStreams } HUF_nbStreams_e;
 
 static size_t HUF_compressCTable_internal(
                 BYTE* const ostart, BYTE* op, BYTE* const oend,
                 const void* src, size_t srcSize,
-                unsigned singleStream, const HUF_CElt* CTable, const int bmi2)
+                HUF_nbStreams_e nbStreams, const HUF_CElt* CTable, const int bmi2)
 {
-    size_t const cSize = singleStream ?
+    size_t const cSize = (nbStreams==HUF_singleStream) ?
                          HUF_compress1X_usingCTable_internal(op, oend - op, src, srcSize, CTable, bmi2) :
                          HUF_compress4X_usingCTable_internal(op, oend - op, src, srcSize, CTable, bmi2);
     if (HUF_isError(cSize)) { return cSize; }
@@ -628,21 +629,21 @@ static size_t HUF_compressCTable_internal(
 }
 
 typedef struct {
-    U32 count[HUF_SYMBOLVALUE_MAX + 1];
+    unsigned count[HUF_SYMBOLVALUE_MAX + 1];
     HUF_CElt CTable[HUF_SYMBOLVALUE_MAX + 1];
     huffNodeTable nodeTable;
 } HUF_compress_tables_t;
 
 /* HUF_compress_internal() :
  * `workSpace` must a table of at least HUF_WORKSPACE_SIZE_U32 unsigned */
-static size_t HUF_compress_internal (
+static size_t
-                void* dst, size_t dstSize,
+HUF_compress_internal (void* dst, size_t dstSize,
-                const void* src, size_t srcSize,
+                 const void* src, size_t srcSize,
-                unsigned maxSymbolValue, unsigned huffLog,
+                       unsigned maxSymbolValue, unsigned huffLog,
-                unsigned singleStream,
+                       HUF_nbStreams_e nbStreams,
-                void* workSpace, size_t wkspSize,
+                       void* workSpace, size_t wkspSize,
-                HUF_CElt* oldHufTable, HUF_repeat* repeat, int preferRepeat,
+                       HUF_CElt* oldHufTable, HUF_repeat* repeat, int preferRepeat,
-                const int bmi2)
+                 const int bmi2)
 {
     HUF_compress_tables_t* const table = (HUF_compress_tables_t*)workSpace;
     BYTE* const ostart = (BYTE*)dst;
@@ -651,7 +652,7 @@ static size_t HUF_compress_internal (
 
     /* checks & inits */
     if (((size_t)workSpace & 3) != 0) return ERROR(GENERIC);  /* must be aligned on 4-bytes boundaries */
-    if (wkspSize < sizeof(*table)) return ERROR(workSpace_tooSmall);
+    if (wkspSize < HUF_WORKSPACE_SIZE) return ERROR(workSpace_tooSmall);
     if (!srcSize) return 0;  /* Uncompressed */
     if (!dstSize) return 0;  /* cannot fit anything within dst budget */
     if (srcSize > HUF_BLOCKSIZE_MAX) return ERROR(srcSize_wrong);   /* current block size limit */
@@ -664,11 +665,11 @@ static size_t HUF_compress_internal (
     if (preferRepeat && repeat && *repeat == HUF_repeat_valid) {
         return HUF_compressCTable_internal(ostart, op, oend,
                                            src, srcSize,
-                                           singleStream, oldHufTable, bmi2);
+                                           nbStreams, oldHufTable, bmi2);
     }
 
     /* Scan input and build symbol stats */
-    {   CHECK_V_F(largest, HIST_count_wksp (table->count, &maxSymbolValue, (const BYTE*)src, srcSize, table->count) );
+    {   CHECK_V_F(largest, HIST_count_wksp (table->count, &maxSymbolValue, (const BYTE*)src, srcSize, workSpace, wkspSize) );
         if (largest == srcSize) { *ostart = ((const BYTE*)src)[0]; return 1; }   /* single symbol, rle */
         if (largest <= (srcSize >> 7)+4) return 0;   /* heuristic : probably not compressible enough */
     }
@@ -683,14 +684,15 @@ static size_t HUF_compress_internal (
     if (preferRepeat && repeat && *repeat != HUF_repeat_none) {
         return HUF_compressCTable_internal(ostart, op, oend,
                                            src, srcSize,
-                                           singleStream, oldHufTable, bmi2);
+                                           nbStreams, oldHufTable, bmi2);
     }
 
     /* Build Huffman Tree */
     huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue);
-    {   CHECK_V_F(maxBits, HUF_buildCTable_wksp(table->CTable, table->count,
+    {   size_t const maxBits = HUF_buildCTable_wksp(table->CTable, table->count,
-                                                maxSymbolValue, huffLog,
+                                            maxSymbolValue, huffLog,
-                                                table->nodeTable, sizeof(table->nodeTable)) );
+                                            table->nodeTable, sizeof(table->nodeTable));
+        CHECK_F(maxBits);
         huffLog = (U32)maxBits;
         /* Zero unused symbols in CTable, so we can check it for validity */
         memset(table->CTable + (maxSymbolValue + 1), 0,
@@ -706,7 +708,7 @@ static size_t HUF_compress_internal (
             if (oldSize <= hSize + newSize || hSize + 12 >= srcSize) {
                 return HUF_compressCTable_internal(ostart, op, oend,
                                                    src, srcSize,
-                                                   singleStream, oldHufTable, bmi2);
+                                                   nbStreams, oldHufTable, bmi2);
         }   }
 
         /* Use the new huffman table */
@@ -718,7 +720,7 @@ static size_t HUF_compress_internal (
     }
     return HUF_compressCTable_internal(ostart, op, oend,
                                        src, srcSize,
-                                       singleStream, table->CTable, bmi2);
+                                       nbStreams, table->CTable, bmi2);
 }
 
 
@@ -728,7 +730,7 @@ size_t HUF_compress1X_wksp (void* dst, size_t dstSize,
                       void* workSpace, size_t wkspSize)
 {
     return HUF_compress_internal(dst, dstSize, src, srcSize,
-                                 maxSymbolValue, huffLog, 1 /*single stream*/,
+                                 maxSymbolValue, huffLog, HUF_singleStream,
                                  workSpace, wkspSize,
                                  NULL, NULL, 0, 0 /*bmi2*/);
 }
@@ -740,7 +742,7 @@ size_t HUF_compress1X_repeat (void* dst, size_t dstSize,
                       HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2)
 {
     return HUF_compress_internal(dst, dstSize, src, srcSize,
-                                 maxSymbolValue, huffLog, 1 /*single stream*/,
+                                 maxSymbolValue, huffLog, HUF_singleStream,
                                  workSpace, wkspSize, hufTable,
                                  repeat, preferRepeat, bmi2);
 }
@@ -762,7 +764,7 @@ size_t HUF_compress4X_wksp (void* dst, size_t dstSize,
                       void* workSpace, size_t wkspSize)
 {
     return HUF_compress_internal(dst, dstSize, src, srcSize,
-                                 maxSymbolValue, huffLog, 0 /*4 streams*/,
+                                 maxSymbolValue, huffLog, HUF_fourStreams,
                                  workSpace, wkspSize,
                                  NULL, NULL, 0, 0 /*bmi2*/);
 }
@@ -777,7 +779,7 @@ size_t HUF_compress4X_repeat (void* dst, size_t dstSize,
                       HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2)
 {
     return HUF_compress_internal(dst, dstSize, src, srcSize,
-                                 maxSymbolValue, huffLog, 0 /* 4 streams */,
+                                 maxSymbolValue, huffLog, HUF_fourStreams,
                                  workSpace, wkspSize,
                                  hufTable, repeat, preferRepeat, bmi2);
 }

C/zstd/huf_decompress.c

@@ -43,6 +43,19 @@
 #include "huf.h"
 #include "error_private.h"
 
+/* **************************************************************
+*  Macros
+****************************************************************/
+
+/* These two optional macros force the use one way or another of the two
+ * Huffman decompression implementations. You can't force in both directions
+ * at the same time.
+ */
+#if defined(HUF_FORCE_DECOMPRESS_X1) && \
+    defined(HUF_FORCE_DECOMPRESS_X2)
+#error "Cannot force the use of the X1 and X2 decoders at the same time!"
+#endif
+
 
 /* **************************************************************
 *  Error Management
@@ -58,6 +71,51 @@
 #define HUF_ALIGN_MASK(x, mask) (((x) + (mask)) & ~(mask))
 
 
+/* **************************************************************
+*  BMI2 Variant Wrappers
+****************************************************************/
+#if DYNAMIC_BMI2
+
+#define HUF_DGEN(fn)                                                        \
+                                                                            \
+    static size_t fn##_default(                                             \
+                  void* dst,  size_t dstSize,                               \
+            const void* cSrc, size_t cSrcSize,                              \
+            const HUF_DTable* DTable)                                       \
+    {                                                                       \
+        return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable);             \
+    }                                                                       \
+                                                                            \
+    static TARGET_ATTRIBUTE("bmi2") size_t fn##_bmi2(                       \
+                  void* dst,  size_t dstSize,                               \
+            const void* cSrc, size_t cSrcSize,                              \
+            const HUF_DTable* DTable)                                       \
+    {                                                                       \
+        return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable);             \
+    }                                                                       \
+                                                                            \
+    static size_t fn(void* dst, size_t dstSize, void const* cSrc,           \
+                     size_t cSrcSize, HUF_DTable const* DTable, int bmi2)   \
+    {                                                                       \
+        if (bmi2) {                                                         \
+            return fn##_bmi2(dst, dstSize, cSrc, cSrcSize, DTable);         \
+        }                                                                   \
+        return fn##_default(dst, dstSize, cSrc, cSrcSize, DTable);          \
+    }
+
+#else
+
+#define HUF_DGEN(fn)                                                        \
+    static size_t fn(void* dst, size_t dstSize, void const* cSrc,           \
+                     size_t cSrcSize, HUF_DTable const* DTable, int bmi2)   \
+    {                                                                       \
+        (void)bmi2;                                                         \
+        return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable);             \
+    }
+
+#endif
+
+
 /*-***************************/
 /*  generic DTableDesc       */
 /*-***************************/
@@ -71,6 +129,8 @@ static DTableDesc HUF_getDTableDesc(const HUF_DTable* table)
 }
 
 
+#ifndef HUF_FORCE_DECOMPRESS_X2
+
 /*-***************************/
 /*  single-symbol decoding   */
 /*-***************************/
@@ -307,46 +367,6 @@ typedef size_t (*HUF_decompress_usingDTable_t)(void *dst, size_t dstSize,
                                                const void *cSrc,
                                                size_t cSrcSize,
                                                const HUF_DTable *DTable);
-#if DYNAMIC_BMI2
-
-#define HUF_DGEN(fn)                                                               \
-                                                                            \
-    static size_t fn##_default(                                             \
-                  void* dst,  size_t dstSize,                               \
-            const void* cSrc, size_t cSrcSize,                              \
-            const HUF_DTable* DTable)                                       \
-    {                                                                       \
-        return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable);             \
-    }                                                                       \
-                                                                            \
-    static TARGET_ATTRIBUTE("bmi2") size_t fn##_bmi2(                       \
-                  void* dst,  size_t dstSize,                               \
-            const void* cSrc, size_t cSrcSize,                              \
-            const HUF_DTable* DTable)                                       \
-    {                                                                       \
-        return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable);             \
-    }                                                                       \
-                                                                            \
-    static size_t fn(void* dst, size_t dstSize, void const* cSrc,           \
-                     size_t cSrcSize, HUF_DTable const* DTable, int bmi2)   \
-    {                                                                       \
-        if (bmi2) {                                                         \
-            return fn##_bmi2(dst, dstSize, cSrc, cSrcSize, DTable);         \
-        }                                                                   \
-        return fn##_default(dst, dstSize, cSrc, cSrcSize, DTable);          \
-    }
-
-#else
-
-#define HUF_DGEN(fn)                                                               \
-    static size_t fn(void* dst, size_t dstSize, void const* cSrc,           \
-                     size_t cSrcSize, HUF_DTable const* DTable, int bmi2)   \
-    {                                                                       \
-        (void)bmi2;                                                         \
-        return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable);             \
-    }
-
-#endif
 
 HUF_DGEN(HUF_decompress1X1_usingDTable_internal)
 HUF_DGEN(HUF_decompress4X1_usingDTable_internal)
@@ -437,6 +457,10 @@ size_t HUF_decompress4X1 (void* dst, size_t dstSize, const void* cSrc, size_t cS
     return HUF_decompress4X1_DCtx(DTable, dst, dstSize, cSrc, cSrcSize);
 }
 
+#endif /* HUF_FORCE_DECOMPRESS_X2 */
+
+
+#ifndef HUF_FORCE_DECOMPRESS_X1
 
 /* *************************/
 /* double-symbols decoding */
@@ -911,6 +935,8 @@ size_t HUF_decompress4X2 (void* dst, size_t dstSize, const void* cSrc, size_t cS
     return HUF_decompress4X2_DCtx(DTable, dst, dstSize, cSrc, cSrcSize);
 }
 
+#endif /* HUF_FORCE_DECOMPRESS_X1 */
+
 
 /* ***********************************/
 /* Universal decompression selectors */
@@ -921,8 +947,18 @@ size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize,
                                     const HUF_DTable* DTable)
 {
     DTableDesc const dtd = HUF_getDTableDesc(DTable);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dtd;
+    assert(dtd.tableType == 0);
+    return HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dtd;
+    assert(dtd.tableType == 1);
+    return HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#else
     return dtd.tableType ? HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) :
                            HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#endif
 }
 
 size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize,
@@ -930,11 +966,22 @@ size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize,
                                     const HUF_DTable* DTable)
 {
     DTableDesc const dtd = HUF_getDTableDesc(DTable);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dtd;
+    assert(dtd.tableType == 0);
+    return HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dtd;
+    assert(dtd.tableType == 1);
+    return HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#else
     return dtd.tableType ? HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) :
                            HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+#endif
 }
 
 
+#if !defined(HUF_FORCE_DECOMPRESS_X1) && !defined(HUF_FORCE_DECOMPRESS_X2)
 typedef struct { U32 tableTime; U32 decode256Time; } algo_time_t;
 static const algo_time_t algoTime[16 /* Quantization */][3 /* single, double, quad */] =
 {
@@ -956,6 +1003,7 @@ static const algo_time_t algoTime[16 /* Quantization */][3 /* single, double, qu
     {{1455,128}, {2422,124}, {4174,124}},   /* Q ==14 : 87-93% */
     {{ 722,128}, {1891,145}, {1936,146}},   /* Q ==15 : 93-99% */
 };
+#endif
 
 /** HUF_selectDecoder() :
  *  Tells which decoder is likely to decode faster,
@@ -966,6 +1014,15 @@ U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize)
 {
     assert(dstSize > 0);
     assert(dstSize <= 128*1024);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dstSize;
+    (void)cSrcSize;
+    return 0;
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dstSize;
+    (void)cSrcSize;
+    return 1;
+#else
     /* decoder timing evaluation */
     {   U32 const Q = (cSrcSize >= dstSize) ? 15 : (U32)(cSrcSize * 16 / dstSize);   /* Q < 16 */
         U32 const D256 = (U32)(dstSize >> 8);
@@ -973,14 +1030,18 @@ U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize)
         U32 DTime1 = algoTime[Q][1].tableTime + (algoTime[Q][1].decode256Time * D256);
         DTime1 += DTime1 >> 3;  /* advantage to algorithm using less memory, to reduce cache eviction */
         return DTime1 < DTime0;
-}   }
+    }
+#endif
+}
 
 
 typedef size_t (*decompressionAlgo)(void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);
 
 size_t HUF_decompress (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
 {
+#if !defined(HUF_FORCE_DECOMPRESS_X1) && !defined(HUF_FORCE_DECOMPRESS_X2)
     static const decompressionAlgo decompress[2] = { HUF_decompress4X1, HUF_decompress4X2 };
+#endif
 
     /* validation checks */
     if (dstSize == 0) return ERROR(dstSize_tooSmall);
@@ -989,7 +1050,17 @@ size_t HUF_decompress (void* dst, size_t dstSize, const void* cSrc, size_t cSrcS
     if (cSrcSize == 1) { memset(dst, *(const BYTE*)cSrc, dstSize); return dstSize; }   /* RLE */
 
     {   U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+        (void)algoNb;
+        assert(algoNb == 0);
+        return HUF_decompress4X1(dst, dstSize, cSrc, cSrcSize);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+        (void)algoNb;
+        assert(algoNb == 1);
+        return HUF_decompress4X2(dst, dstSize, cSrc, cSrcSize);
+#else
         return decompress[algoNb](dst, dstSize, cSrc, cSrcSize);
+#endif
     }
 }
 
@@ -1002,8 +1073,18 @@ size_t HUF_decompress4X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const
     if (cSrcSize == 1) { memset(dst, *(const BYTE*)cSrc, dstSize); return dstSize; }   /* RLE */
 
     {   U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+        (void)algoNb;
+        assert(algoNb == 0);
+        return HUF_decompress4X1_DCtx(dctx, dst, dstSize, cSrc, cSrcSize);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+        (void)algoNb;
+        assert(algoNb == 1);
+        return HUF_decompress4X2_DCtx(dctx, dst, dstSize, cSrc, cSrcSize);
+#else
         return algoNb ? HUF_decompress4X2_DCtx(dctx, dst, dstSize, cSrc, cSrcSize) :
                         HUF_decompress4X1_DCtx(dctx, dst, dstSize, cSrc, cSrcSize) ;
+#endif
     }
 }
 
@@ -1025,8 +1106,19 @@ size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable* dctx, void* dst,
     if (cSrcSize == 0) return ERROR(corruption_detected);
 
     {   U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
-        return algoNb ? HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize):
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+        (void)algoNb;
+        assert(algoNb == 0);
+        return HUF_decompress4X1_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+        (void)algoNb;
+        assert(algoNb == 1);
+        return HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize);
+#else
+        return algoNb ? HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc,
+                            cSrcSize, workSpace, wkspSize):
                         HUF_decompress4X1_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize);
+#endif
     }
 }
 
@@ -1041,10 +1133,22 @@ size_t HUF_decompress1X_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
     if (cSrcSize == 1) { memset(dst, *(const BYTE*)cSrc, dstSize); return dstSize; }   /* RLE */
 
     {   U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+        (void)algoNb;
+        assert(algoNb == 0);
+        return HUF_decompress1X1_DCtx_wksp(dctx, dst, dstSize, cSrc,
+                                cSrcSize, workSpace, wkspSize);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+        (void)algoNb;
+        assert(algoNb == 1);
+        return HUF_decompress1X2_DCtx_wksp(dctx, dst, dstSize, cSrc,
+                                cSrcSize, workSpace, wkspSize);
+#else
         return algoNb ? HUF_decompress1X2_DCtx_wksp(dctx, dst, dstSize, cSrc,
                                 cSrcSize, workSpace, wkspSize):
                         HUF_decompress1X1_DCtx_wksp(dctx, dst, dstSize, cSrc,
                                 cSrcSize, workSpace, wkspSize);
+#endif
     }
 }
 
@@ -1060,10 +1164,21 @@ size_t HUF_decompress1X_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize,
 size_t HUF_decompress1X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2)
 {
     DTableDesc const dtd = HUF_getDTableDesc(DTable);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dtd;
+    assert(dtd.tableType == 0);
+    return HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dtd;
+    assert(dtd.tableType == 1);
+    return HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#else
     return dtd.tableType ? HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2) :
                            HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#endif
 }
 
+#ifndef HUF_FORCE_DECOMPRESS_X2
 size_t HUF_decompress1X1_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2)
 {
     const BYTE* ip = (const BYTE*) cSrc;
@@ -1075,12 +1190,23 @@ size_t HUF_decompress1X1_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstS
 
     return HUF_decompress1X1_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
 }
+#endif
 
 size_t HUF_decompress4X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2)
 {
     DTableDesc const dtd = HUF_getDTableDesc(DTable);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+    (void)dtd;
+    assert(dtd.tableType == 0);
+    return HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+    (void)dtd;
+    assert(dtd.tableType == 1);
+    return HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#else
     return dtd.tableType ? HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2) :
                            HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+#endif
 }
 
 size_t HUF_decompress4X_hufOnly_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2)
@@ -1090,7 +1216,17 @@ size_t HUF_decompress4X_hufOnly_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t ds
     if (cSrcSize == 0) return ERROR(corruption_detected);
 
     {   U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
+#if defined(HUF_FORCE_DECOMPRESS_X1)
+        (void)algoNb;
+        assert(algoNb == 0);
+        return HUF_decompress4X1_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
+#elif defined(HUF_FORCE_DECOMPRESS_X2)
+        (void)algoNb;
+        assert(algoNb == 1);
+        return HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
+#else
         return algoNb ? HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2) :
                         HUF_decompress4X1_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
+#endif
     }
 }

C/zstd/pool.c

@@ -88,8 +88,8 @@ static void* POOL_thread(void* opaque) {
             ctx->numThreadsBusy++;
             ctx->queueEmpty = ctx->queueHead == ctx->queueTail;
             /* Unlock the mutex, signal a pusher, and run the job */
-            ZSTD_pthread_mutex_unlock(&ctx->queueMutex);
             ZSTD_pthread_cond_signal(&ctx->queuePushCond);
+            ZSTD_pthread_mutex_unlock(&ctx->queueMutex);
 
             job.function(job.opaque);
 

C/zstd/zstd_common.c

@@ -30,8 +30,10 @@ const char* ZSTD_versionString(void) { return ZSTD_VERSION_STRING; }
 /*-****************************************
 *  ZSTD Error Management
 ******************************************/
+#undef ZSTD_isError   /* defined within zstd_internal.h */
 /*! ZSTD_isError() :
- *  tells if a return value is an error code */
+ *  tells if a return value is an error code
+ *  symbol is required for external callers */
 unsigned ZSTD_isError(size_t code) { return ERR_isError(code); }
 
 /*! ZSTD_getErrorName() :

C/zstd/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 */
+    unsigned* litFreq;           /* table of literals statistics, of size 256 */
-    U32* litLengthFreq;          /* table of litLength statistics, of size (MaxLL+1) */
+    unsigned* litLengthFreq;     /* table of litLength statistics, of size (MaxLL+1) */
-    U32* matchLengthFreq;        /* table of matchLength statistics, of size (MaxML+1) */
+    unsigned* matchLengthFreq;   /* table of matchLength statistics, of size (MaxML+1) */
-    U32* offCodeFreq;            /* table of offCode statistics, of size (MaxOff+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;
+    int nbWorkers;
-    unsigned jobSize;
+    size_t jobSize;
-    unsigned overlapSizeLog;
+    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,
+MEM_STATIC void
-                                           void const* srcEnd, U32 maxDist,
+ZSTD_window_enforceMaxDist(ZSTD_window_t* window,
-                                           U32* loadedDictEndPtr,
+                           void const* srcEnd,
-                                           const ZSTD_matchState_t** dictMatchStatePtr)
+                           U32 maxDist,
+                           U32* loadedDictEndPtr,
+                     const ZSTD_matchState_t** dictMatchStatePtr)
 {
-    U32 const current = (U32)((BYTE const*)srcEnd - window->base);
+    U32 const blockEndIdx = (U32)((BYTE const*)srcEnd - window->base);
-    U32 loadedDictEnd = loadedDictEndPtr != NULL ? *loadedDictEndPtr : 0;
+    U32 loadedDictEnd = (loadedDictEndPtr != NULL) ? *loadedDictEndPtr : 0;
-    DEBUGLOG(5, "ZSTD_window_enforceMaxDist: current=%u, maxDist=%u", current, maxDist);
+    DEBUGLOG(5, "ZSTD_window_enforceMaxDist: blockEndIdx=%u, maxDist=%u",
-    if (current > maxDist + loadedDictEnd) {
+                (unsigned)blockEndIdx, (unsigned)maxDist);
-        U32 const newLowLimit = current - 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 newStat = rawStat + 1;
-    U32 const hb = ZSTD_highbit32(stat);
+    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

C/zstd/zstd_ddict.c

@@ -0,0 +1,240 @@
+/*
+ * Copyright (c) 2016-present, 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.
+ */
+
+/* zstd_ddict.c :
+ * concentrates all logic that needs to know the internals of ZSTD_DDict object */
+
+/*-*******************************************************
+*  Dependencies
+*********************************************************/
+#include <string.h>      /* memcpy, memmove, memset */
+#include "cpu.h"         /* bmi2 */
+#include "mem.h"         /* low level memory routines */
+#define FSE_STATIC_LINKING_ONLY
+#include "fse.h"
+#define HUF_STATIC_LINKING_ONLY
+#include "huf.h"
+#include "zstd_decompress_internal.h"
+#include "zstd_ddict.h"
+
+#if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT>=1)
+#  include "zstd_legacy.h"
+#endif
+
+
+
+/*-*******************************************************
+*  Types
+*********************************************************/
+struct ZSTD_DDict_s {
+    void* dictBuffer;
+    const void* dictContent;
+    size_t dictSize;
+    ZSTD_entropyDTables_t entropy;
+    U32 dictID;
+    U32 entropyPresent;
+    ZSTD_customMem cMem;
+};  /* typedef'd to ZSTD_DDict within "zstd.h" */
+
+const void* ZSTD_DDict_dictContent(const ZSTD_DDict* ddict)
+{
+    assert(ddict != NULL);
+    return ddict->dictContent;
+}
+
+size_t ZSTD_DDict_dictSize(const ZSTD_DDict* ddict)
+{
+    assert(ddict != NULL);
+    return ddict->dictSize;
+}
+
+void ZSTD_copyDDictParameters(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict)
+{
+    DEBUGLOG(4, "ZSTD_copyDDictParameters");
+    assert(dctx != NULL);
+    assert(ddict != NULL);
+    dctx->dictID = ddict->dictID;
+    dctx->prefixStart = ddict->dictContent;
+    dctx->virtualStart = ddict->dictContent;
+    dctx->dictEnd = (const BYTE*)ddict->dictContent + ddict->dictSize;
+    dctx->previousDstEnd = dctx->dictEnd;
+    if (ddict->entropyPresent) {
+        dctx->litEntropy = 1;
+        dctx->fseEntropy = 1;
+        dctx->LLTptr = ddict->entropy.LLTable;
+        dctx->MLTptr = ddict->entropy.MLTable;
+        dctx->OFTptr = ddict->entropy.OFTable;
+        dctx->HUFptr = ddict->entropy.hufTable;
+        dctx->entropy.rep[0] = ddict->entropy.rep[0];
+        dctx->entropy.rep[1] = ddict->entropy.rep[1];
+        dctx->entropy.rep[2] = ddict->entropy.rep[2];
+    } else {
+        dctx->litEntropy = 0;
+        dctx->fseEntropy = 0;
+    }
+}
+
+
+static size_t
+ZSTD_loadEntropy_intoDDict(ZSTD_DDict* ddict,
+                           ZSTD_dictContentType_e dictContentType)
+{
+    ddict->dictID = 0;
+    ddict->entropyPresent = 0;
+    if (dictContentType == ZSTD_dct_rawContent) return 0;
+
+    if (ddict->dictSize < 8) {
+        if (dictContentType == ZSTD_dct_fullDict)
+            return ERROR(dictionary_corrupted);   /* only accept specified dictionaries */
+        return 0;   /* pure content mode */
+    }
+    {   U32 const magic = MEM_readLE32(ddict->dictContent);
+        if (magic != ZSTD_MAGIC_DICTIONARY) {
+            if (dictContentType == ZSTD_dct_fullDict)
+                return ERROR(dictionary_corrupted);   /* only accept specified dictionaries */
+            return 0;   /* pure content mode */
+        }
+    }
+    ddict->dictID = MEM_readLE32((const char*)ddict->dictContent + ZSTD_FRAMEIDSIZE);
+
+    /* load entropy tables */
+    CHECK_E( ZSTD_loadDEntropy(&ddict->entropy,
+                                ddict->dictContent, ddict->dictSize),
+             dictionary_corrupted );
+    ddict->entropyPresent = 1;
+    return 0;
+}
+
+
+static size_t ZSTD_initDDict_internal(ZSTD_DDict* ddict,
+                                      const void* dict, size_t dictSize,
+                                      ZSTD_dictLoadMethod_e dictLoadMethod,
+                                      ZSTD_dictContentType_e dictContentType)
+{
+    if ((dictLoadMethod == ZSTD_dlm_byRef) || (!dict) || (!dictSize)) {
+        ddict->dictBuffer = NULL;
+        ddict->dictContent = dict;
+        if (!dict) dictSize = 0;
+    } else {
+        void* const internalBuffer = ZSTD_malloc(dictSize, ddict->cMem);
+        ddict->dictBuffer = internalBuffer;
+        ddict->dictContent = internalBuffer;
+        if (!internalBuffer) return ERROR(memory_allocation);
+        memcpy(internalBuffer, dict, dictSize);
+    }
+    ddict->dictSize = dictSize;
+    ddict->entropy.hufTable[0] = (HUF_DTable)((HufLog)*0x1000001);  /* cover both little and big endian */
+
+    /* parse dictionary content */
+    CHECK_F( ZSTD_loadEntropy_intoDDict(ddict, dictContentType) );
+
+    return 0;
+}
+
+ZSTD_DDict* ZSTD_createDDict_advanced(const void* dict, size_t dictSize,
+                                      ZSTD_dictLoadMethod_e dictLoadMethod,
+                                      ZSTD_dictContentType_e dictContentType,
+                                      ZSTD_customMem customMem)
+{
+    if (!customMem.customAlloc ^ !customMem.customFree) return NULL;
+
+    {   ZSTD_DDict* const ddict = (ZSTD_DDict*) ZSTD_malloc(sizeof(ZSTD_DDict), customMem);
+        if (ddict == NULL) return NULL;
+        ddict->cMem = customMem;
+        {   size_t const initResult = ZSTD_initDDict_internal(ddict,
+                                            dict, dictSize,
+                                            dictLoadMethod, dictContentType);
+            if (ZSTD_isError(initResult)) {
+                ZSTD_freeDDict(ddict);
+                return NULL;
+        }   }
+        return ddict;
+    }
+}
+
+/*! ZSTD_createDDict() :
+*   Create a digested dictionary, to start decompression without startup delay.
+*   `dict` content is copied inside DDict.
+*   Consequently, `dict` can be released after `ZSTD_DDict` creation */
+ZSTD_DDict* ZSTD_createDDict(const void* dict, size_t dictSize)
+{
+    ZSTD_customMem const allocator = { NULL, NULL, NULL };
+    return ZSTD_createDDict_advanced(dict, dictSize, ZSTD_dlm_byCopy, ZSTD_dct_auto, allocator);
+}
+
+/*! ZSTD_createDDict_byReference() :
+ *  Create a digested dictionary, to start decompression without startup delay.
+ *  Dictionary content is simply referenced, it will be accessed during decompression.
+ *  Warning : dictBuffer must outlive DDict (DDict must be freed before dictBuffer) */
+ZSTD_DDict* ZSTD_createDDict_byReference(const void* dictBuffer, size_t dictSize)
+{
+    ZSTD_customMem const allocator = { NULL, NULL, NULL };
+    return ZSTD_createDDict_advanced(dictBuffer, dictSize, ZSTD_dlm_byRef, ZSTD_dct_auto, allocator);
+}
+
+
+const ZSTD_DDict* ZSTD_initStaticDDict(
+                                void* sBuffer, size_t sBufferSize,
+                                const void* dict, size_t dictSize,
+                                ZSTD_dictLoadMethod_e dictLoadMethod,
+                                ZSTD_dictContentType_e dictContentType)
+{
+    size_t const neededSpace = sizeof(ZSTD_DDict)
+                             + (dictLoadMethod == ZSTD_dlm_byRef ? 0 : dictSize);
+    ZSTD_DDict* const ddict = (ZSTD_DDict*)sBuffer;
+    assert(sBuffer != NULL);
+    assert(dict != NULL);
+    if ((size_t)sBuffer & 7) return NULL;   /* 8-aligned */
+    if (sBufferSize < neededSpace) return NULL;
+    if (dictLoadMethod == ZSTD_dlm_byCopy) {
+        memcpy(ddict+1, dict, dictSize);  /* local copy */
+        dict = ddict+1;
+    }
+    if (ZSTD_isError( ZSTD_initDDict_internal(ddict,
+                                              dict, dictSize,
+                                              ZSTD_dlm_byRef, dictContentType) ))
+        return NULL;
+    return ddict;
+}
+
+
+size_t ZSTD_freeDDict(ZSTD_DDict* ddict)
+{
+    if (ddict==NULL) return 0;   /* support free on NULL */
+    {   ZSTD_customMem const cMem = ddict->cMem;
+        ZSTD_free(ddict->dictBuffer, cMem);
+        ZSTD_free(ddict, cMem);
+        return 0;
+    }
+}
+
+/*! ZSTD_estimateDDictSize() :
+ *  Estimate amount of memory that will be needed to create a dictionary for decompression.
+ *  Note : dictionary created by reference using ZSTD_dlm_byRef are smaller */
+size_t ZSTD_estimateDDictSize(size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod)
+{
+    return sizeof(ZSTD_DDict) + (dictLoadMethod == ZSTD_dlm_byRef ? 0 : dictSize);
+}
+
+size_t ZSTD_sizeof_DDict(const ZSTD_DDict* ddict)
+{
+    if (ddict==NULL) return 0;   /* support sizeof on NULL */
+    return sizeof(*ddict) + (ddict->dictBuffer ? ddict->dictSize : 0) ;
+}
+
+/*! ZSTD_getDictID_fromDDict() :
+ *  Provides the dictID of the dictionary loaded into `ddict`.
+ *  If @return == 0, the dictionary is not conformant to Zstandard specification, or empty.
+ *  Non-conformant dictionaries can still be loaded, but as content-only dictionaries. */
+unsigned ZSTD_getDictID_fromDDict(const ZSTD_DDict* ddict)
+{
+    if (ddict==NULL) return 0;
+    return ZSTD_getDictID_fromDict(ddict->dictContent, ddict->dictSize);
+}

C/zstd/zstd_ddict.h

@@ -0,0 +1,44 @@
+/*
+ * Copyright (c) 2016-present, 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.
+ */
+
+
+#ifndef ZSTD_DDICT_H
+#define ZSTD_DDICT_H
+
+/*-*******************************************************
+ *  Dependencies
+ *********************************************************/
+#include <stddef.h>   /* size_t */
+#include "zstd.h"     /* ZSTD_DDict, and several public functions */
+
+
+/*-*******************************************************
+ *  Interface
+ *********************************************************/
+
+/* note: several prototypes are already published in `zstd.h` :
+ * ZSTD_createDDict()
+ * ZSTD_createDDict_byReference()
+ * ZSTD_createDDict_advanced()
+ * ZSTD_freeDDict()
+ * ZSTD_initStaticDDict()
+ * ZSTD_sizeof_DDict()
+ * ZSTD_estimateDDictSize()
+ * ZSTD_getDictID_fromDict()
+ */
+
+const void* ZSTD_DDict_dictContent(const ZSTD_DDict* ddict);
+size_t ZSTD_DDict_dictSize(const ZSTD_DDict* ddict);
+
+void ZSTD_copyDDictParameters(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict);
+
+
+
+#endif /* ZSTD_DDICT_H */

C/zstd/zstd_decompress_block.h

@@ -0,0 +1,59 @@
+/*
+ * Copyright (c) 2016-present, 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.
+ */
+
+
+#ifndef ZSTD_DEC_BLOCK_H
+#define ZSTD_DEC_BLOCK_H
+
+/*-*******************************************************
+ *  Dependencies
+ *********************************************************/
+#include <stddef.h>   /* size_t */
+#include "zstd.h"    /* DCtx, and some public functions */
+#include "zstd_internal.h"  /* blockProperties_t, and some public functions */
+#include "zstd_decompress_internal.h"  /* ZSTD_seqSymbol */
+
+
+/* ===   Prototypes   === */
+
+/* note: prototypes already published within `zstd.h` :
+ * ZSTD_decompressBlock()
+ */
+
+/* note: prototypes already published within `zstd_internal.h` :
+ * ZSTD_getcBlockSize()
+ * ZSTD_decodeSeqHeaders()
+ */
+
+
+/* ZSTD_decompressBlock_internal() :
+ * decompress block, starting at `src`,
+ * into destination buffer `dst`.
+ * @return : decompressed block size,
+ *           or an error code (which can be tested using ZSTD_isError())
+ */
+size_t ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx,
+                               void* dst, size_t dstCapacity,
+                         const void* src, size_t srcSize, const int frame);
+
+/* ZSTD_buildFSETable() :
+ * generate FSE decoding table for one symbol (ll, ml or off)
+ * this function must be called with valid parameters only
+ * (dt is large enough, normalizedCounter distribution total is a power of 2, max is within range, etc.)
+ * in which case it cannot fail.
+ * Internal use only.
+ */
+void ZSTD_buildFSETable(ZSTD_seqSymbol* dt,
+             const short* normalizedCounter, unsigned maxSymbolValue,
+             const U32* baseValue, const U32* nbAdditionalBits,
+                   unsigned tableLog);
+
+
+#endif /* ZSTD_DEC_BLOCK_H */

C/zstd/zstd_decompress_internal.h

@@ -0,0 +1,168 @@
+/*
+ * Copyright (c) 2016-present, 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.
+ */
+
+
+/* zstd_decompress_internal:
+ * objects and definitions shared within lib/decompress modules */
+
+ #ifndef ZSTD_DECOMPRESS_INTERNAL_H
+ #define ZSTD_DECOMPRESS_INTERNAL_H
+
+
+/*-*******************************************************
+ *  Dependencies
+ *********************************************************/
+#include "mem.h"             /* BYTE, U16, U32 */
+#include "zstd_internal.h"   /* ZSTD_seqSymbol */
+
+
+
+/*-*******************************************************
+ *  Constants
+ *********************************************************/
+static const U32 LL_base[MaxLL+1] = {
+                 0,    1,    2,     3,     4,     5,     6,      7,
+                 8,    9,   10,    11,    12,    13,    14,     15,
+                16,   18,   20,    22,    24,    28,    32,     40,
+                48,   64, 0x80, 0x100, 0x200, 0x400, 0x800, 0x1000,
+                0x2000, 0x4000, 0x8000, 0x10000 };
+
+static const U32 OF_base[MaxOff+1] = {
+                 0,        1,       1,       5,     0xD,     0x1D,     0x3D,     0x7D,
+                 0xFD,   0x1FD,   0x3FD,   0x7FD,   0xFFD,   0x1FFD,   0x3FFD,   0x7FFD,
+                 0xFFFD, 0x1FFFD, 0x3FFFD, 0x7FFFD, 0xFFFFD, 0x1FFFFD, 0x3FFFFD, 0x7FFFFD,
+                 0xFFFFFD, 0x1FFFFFD, 0x3FFFFFD, 0x7FFFFFD, 0xFFFFFFD, 0x1FFFFFFD, 0x3FFFFFFD, 0x7FFFFFFD };
+
+static const U32 OF_bits[MaxOff+1] = {
+                     0,  1,  2,  3,  4,  5,  6,  7,
+                     8,  9, 10, 11, 12, 13, 14, 15,
+                    16, 17, 18, 19, 20, 21, 22, 23,
+                    24, 25, 26, 27, 28, 29, 30, 31 };
+
+static const U32 ML_base[MaxML+1] = {
+                     3,  4,  5,    6,     7,     8,     9,    10,
+                    11, 12, 13,   14,    15,    16,    17,    18,
+                    19, 20, 21,   22,    23,    24,    25,    26,
+                    27, 28, 29,   30,    31,    32,    33,    34,
+                    35, 37, 39,   41,    43,    47,    51,    59,
+                    67, 83, 99, 0x83, 0x103, 0x203, 0x403, 0x803,
+                    0x1003, 0x2003, 0x4003, 0x8003, 0x10003 };
+
+
+/*-*******************************************************
+ *  Decompression types
+ *********************************************************/
+ typedef struct {
+     U32 fastMode;
+     U32 tableLog;
+ } ZSTD_seqSymbol_header;
+
+ typedef struct {
+     U16  nextState;
+     BYTE nbAdditionalBits;
+     BYTE nbBits;
+     U32  baseValue;
+ } ZSTD_seqSymbol;
+
+ #define SEQSYMBOL_TABLE_SIZE(log)   (1 + (1 << (log)))
+
+typedef struct {
+    ZSTD_seqSymbol LLTable[SEQSYMBOL_TABLE_SIZE(LLFSELog)];    /* Note : Space reserved for FSE Tables */
+    ZSTD_seqSymbol OFTable[SEQSYMBOL_TABLE_SIZE(OffFSELog)];   /* is also used as temporary workspace while building hufTable during DDict creation */
+    ZSTD_seqSymbol MLTable[SEQSYMBOL_TABLE_SIZE(MLFSELog)];    /* and therefore must be at least HUF_DECOMPRESS_WORKSPACE_SIZE large */
+    HUF_DTable hufTable[HUF_DTABLE_SIZE(HufLog)];  /* can accommodate HUF_decompress4X */
+    U32 rep[ZSTD_REP_NUM];
+} ZSTD_entropyDTables_t;
+
+typedef enum { ZSTDds_getFrameHeaderSize, ZSTDds_decodeFrameHeader,
+               ZSTDds_decodeBlockHeader, ZSTDds_decompressBlock,
+               ZSTDds_decompressLastBlock, ZSTDds_checkChecksum,
+               ZSTDds_decodeSkippableHeader, ZSTDds_skipFrame } ZSTD_dStage;
+
+typedef enum { zdss_init=0, zdss_loadHeader,
+               zdss_read, zdss_load, zdss_flush } ZSTD_dStreamStage;
+
+struct ZSTD_DCtx_s
+{
+    const ZSTD_seqSymbol* LLTptr;
+    const ZSTD_seqSymbol* MLTptr;
+    const ZSTD_seqSymbol* OFTptr;
+    const HUF_DTable* HUFptr;
+    ZSTD_entropyDTables_t entropy;
+    U32 workspace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];   /* space needed when building huffman tables */
+    const void* previousDstEnd;   /* detect continuity */
+    const void* prefixStart;      /* start of current segment */
+    const void* virtualStart;     /* virtual start of previous segment if it was just before current one */
+    const void* dictEnd;          /* end of previous segment */
+    size_t expected;
+    ZSTD_frameHeader fParams;
+    U64 decodedSize;
+    blockType_e bType;            /* used in ZSTD_decompressContinue(), store blockType between block header decoding and block decompression stages */
+    ZSTD_dStage stage;
+    U32 litEntropy;
+    U32 fseEntropy;
+    XXH64_state_t xxhState;
+    size_t headerSize;
+    ZSTD_format_e format;
+    const BYTE* litPtr;
+    ZSTD_customMem customMem;
+    size_t litSize;
+    size_t rleSize;
+    size_t staticSize;
+    int bmi2;                     /* == 1 if the CPU supports BMI2 and 0 otherwise. CPU support is determined dynamically once per context lifetime. */
+
+    /* dictionary */
+    ZSTD_DDict* ddictLocal;
+    const ZSTD_DDict* ddict;     /* set by ZSTD_initDStream_usingDDict(), or ZSTD_DCtx_refDDict() */
+    U32 dictID;
+    int ddictIsCold;             /* if == 1 : dictionary is "new" for working context, and presumed "cold" (not in cpu cache) */
+
+    /* streaming */
+    ZSTD_dStreamStage streamStage;
+    char*  inBuff;
+    size_t inBuffSize;
+    size_t inPos;
+    size_t maxWindowSize;
+    char*  outBuff;
+    size_t outBuffSize;
+    size_t outStart;
+    size_t outEnd;
+    size_t lhSize;
+    void* legacyContext;
+    U32 previousLegacyVersion;
+    U32 legacyVersion;
+    U32 hostageByte;
+    int noForwardProgress;
+
+    /* workspace */
+    BYTE litBuffer[ZSTD_BLOCKSIZE_MAX + WILDCOPY_OVERLENGTH];
+    BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX];
+};  /* typedef'd to ZSTD_DCtx within "zstd.h" */
+
+
+/*-*******************************************************
+ *  Shared internal functions
+ *********************************************************/
+
+/*! ZSTD_loadDEntropy() :
+ *  dict : must point at beginning of a valid zstd dictionary.
+ * @return : size of entropy tables read */
+size_t ZSTD_loadDEntropy(ZSTD_entropyDTables_t* entropy,
+                   const void* const dict, size_t const dictSize);
+
+/*! ZSTD_checkContinuity() :
+ *  check if next `dst` follows previous position, where decompression ended.
+ *  If yes, do nothing (continue on current segment).
+ *  If not, classify previous segment as "external dictionary", and start a new segment.
+ *  This function cannot fail. */
+void ZSTD_checkContinuity(ZSTD_DCtx* dctx, const void* dst);
+
+
+#endif /* ZSTD_DECOMPRESS_INTERNAL_H */

C/zstd/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 */

C/zstd/zstd_errors.h

@@ -72,6 +72,7 @@ typedef enum {
   ZSTD_error_workSpace_tooSmall= 66,
   ZSTD_error_dstSize_tooSmall = 70,
   ZSTD_error_srcSize_wrong    = 72,
+  ZSTD_error_dstBuffer_null   = 74,
   /* following error codes are __NOT STABLE__, they can be removed or changed in future versions */
   ZSTD_error_frameIndex_tooLarge = 100,
   ZSTD_error_seekableIO          = 102,

C/zstd/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;
+        size_t const hash0 = ZSTD_hashPtr(ip, hBits, mls);
-        for (i = 0; i < fastHashFillStep; ++i) {
+        hashTable[hash0] = current;
-            size_t const hash = ZSTD_hashPtr(ip + i, hBits, mls);
+        if (dtlm == ZSTD_dtlm_fast) continue;
-            if (i == 0 || hashTable[hash] == 0)
+        /* Only load extra positions for ZSTD_dtlm_full */
-                hashTable[hash] = current + i;
+        {   U32 p;
-            /* Only load extra positions for ZSTD_dtlm_full */
+            for (p = 1; p < fastHashFillStep; ++p) {
-            if (dtlm == ZSTD_dtlm_fast)
+                size_t const hash = ZSTD_hashPtr(ip + p, hBits, mls);
-                break;
+                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 */

C/zstd/zstd_internal.h

@@ -41,6 +41,9 @@ extern "C" {
 
 /* ---- static assert (debug) --- */
 #define ZSTD_STATIC_ASSERT(c) DEBUG_STATIC_ASSERT(c)
+#define ZSTD_isError ERR_isError   /* for inlining */
+#define FSE_isError  ERR_isError
+#define HUF_isError  ERR_isError
 
 
 /*-*************************************
@@ -75,7 +78,6 @@ static const U32 repStartValue[ZSTD_REP_NUM] = { 1, 4, 8 };
 #define BIT0   1
 
 #define ZSTD_WINDOWLOG_ABSOLUTEMIN 10
-#define ZSTD_WINDOWLOG_DEFAULTMAX 27 /* Default maximum allowed window log */
 static const size_t ZSTD_fcs_fieldSize[4] = { 0, 2, 4, 8 };
 static const size_t ZSTD_did_fieldSize[4] = { 0, 1, 2, 4 };
 
@@ -242,7 +244,7 @@ typedef struct {
     blockType_e blockType;
     U32 lastBlock;
     U32 origSize;
-} blockProperties_t;
+} blockProperties_t;   /* declared here for decompress and fullbench */
 
 /*! ZSTD_getcBlockSize() :
  *  Provides the size of compressed block from block header `src` */
@@ -250,6 +252,13 @@ typedef struct {
 size_t ZSTD_getcBlockSize(const void* src, size_t srcSize,
                           blockProperties_t* bpPtr);
 
+/*! ZSTD_decodeSeqHeaders() :
+ *  decode sequence header from src */
+/* Used by: decompress, fullbench (does not get its definition from here) */
+size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr,
+                       const void* src, size_t srcSize);
+
+
 #if defined (__cplusplus)
 }
 #endif

C/zstd/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 bt = ms->chainTable;
-    U32    const btLog  = cParams->chainLog - 1;
+    U32  const btLog  = cParams->chainLog - 1;
-    U32    const btMask = (1 << btLog) - 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);

C/zstd/zstd_ldm.c

@@ -37,8 +37,8 @@ void ZSTD_ldm_adjustParameters(ldmParams_t* params,
         params->hashLog = MAX(ZSTD_HASHLOG_MIN, params->windowLog - LDM_HASH_RLOG);
         assert(params->hashLog <= ZSTD_HASHLOG_MAX);
     }
-    if (params->hashEveryLog == 0) {
+    if (params->hashRateLog == 0) {
-        params->hashEveryLog = params->windowLog < params->hashLog
+        params->hashRateLog = params->windowLog < params->hashLog
                                    ? 0
                                    : params->windowLog - params->hashLog;
     }
@@ -119,20 +119,20 @@ static void ZSTD_ldm_insertEntry(ldmState_t* ldmState,
  *
  *  Gets the small hash, checksum, and tag from the rollingHash.
  *
- *  If the tag matches (1 << ldmParams.hashEveryLog)-1, then
+ *  If the tag matches (1 << ldmParams.hashRateLog)-1, then
  *  creates an ldmEntry from the offset, and inserts it into the hash table.
  *
  *  hBits is the length of the small hash, which is the most significant hBits
  *  of rollingHash. The checksum is the next 32 most significant bits, followed
- *  by ldmParams.hashEveryLog bits that make up the tag. */
+ *  by ldmParams.hashRateLog bits that make up the tag. */
 static void ZSTD_ldm_makeEntryAndInsertByTag(ldmState_t* ldmState,
                                              U64 const rollingHash,
                                              U32 const hBits,
                                              U32 const offset,
                                              ldmParams_t const ldmParams)
 {
-    U32 const tag = ZSTD_ldm_getTag(rollingHash, hBits, ldmParams.hashEveryLog);
+    U32 const tag = ZSTD_ldm_getTag(rollingHash, hBits, ldmParams.hashRateLog);
-    U32 const tagMask = ((U32)1 << ldmParams.hashEveryLog) - 1;
+    U32 const tagMask = ((U32)1 << ldmParams.hashRateLog) - 1;
     if (tag == tagMask) {
         U32 const hash = ZSTD_ldm_getSmallHash(rollingHash, hBits);
         U32 const checksum = ZSTD_ldm_getChecksum(rollingHash, hBits);
@@ -143,56 +143,6 @@ static void ZSTD_ldm_makeEntryAndInsertByTag(ldmState_t* ldmState,
     }
 }
 
-/** ZSTD_ldm_getRollingHash() :
- *  Get a 64-bit hash using the first len bytes from buf.
- *
- *  Giving bytes s = s_1, s_2, ... s_k, the hash is defined to be
- *  H(s) = s_1*(a^(k-1)) + s_2*(a^(k-2)) + ... + s_k*(a^0)
- *
- *  where the constant a is defined to be prime8bytes.
- *
- *  The implementation adds an offset to each byte, so
- *  H(s) = (s_1 + HASH_CHAR_OFFSET)*(a^(k-1)) + ... */
-static U64 ZSTD_ldm_getRollingHash(const BYTE* buf, U32 len)
-{
-    U64 ret = 0;
-    U32 i;
-    for (i = 0; i < len; i++) {
-        ret *= prime8bytes;
-        ret += buf[i] + LDM_HASH_CHAR_OFFSET;
-    }
-    return ret;
-}
-
-/** ZSTD_ldm_ipow() :
- *  Return base^exp. */
-static U64 ZSTD_ldm_ipow(U64 base, U64 exp)
-{
-    U64 ret = 1;
-    while (exp) {
-        if (exp & 1) { ret *= base; }
-        exp >>= 1;
-        base *= base;
-    }
-    return ret;
-}
-
-U64 ZSTD_ldm_getHashPower(U32 minMatchLength) {
-    DEBUGLOG(4, "ZSTD_ldm_getHashPower: mml=%u", minMatchLength);
-    assert(minMatchLength >= ZSTD_LDM_MINMATCH_MIN);
-    return ZSTD_ldm_ipow(prime8bytes, minMatchLength - 1);
-}
-
-/** ZSTD_ldm_updateHash() :
- *  Updates hash by removing toRemove and adding toAdd. */
-static U64 ZSTD_ldm_updateHash(U64 hash, BYTE toRemove, BYTE toAdd, U64 hashPower)
-{
-    hash -= ((toRemove + LDM_HASH_CHAR_OFFSET) * hashPower);
-    hash *= prime8bytes;
-    hash += toAdd + LDM_HASH_CHAR_OFFSET;
-    return hash;
-}
-
 /** ZSTD_ldm_countBackwardsMatch() :
  *  Returns the number of bytes that match backwards before pIn and pMatch.
  *
@@ -238,6 +188,7 @@ static size_t ZSTD_ldm_fillFastTables(ZSTD_matchState_t* ms,
     case ZSTD_btlazy2:
     case ZSTD_btopt:
     case ZSTD_btultra:
+    case ZSTD_btultra2:
         break;
     default:
         assert(0);  /* not possible : not a valid strategy id */
@@ -261,9 +212,9 @@ static U64 ZSTD_ldm_fillLdmHashTable(ldmState_t* state,
     const BYTE* cur = lastHashed + 1;
 
     while (cur < iend) {
-        rollingHash = ZSTD_ldm_updateHash(rollingHash, cur[-1],
+        rollingHash = ZSTD_rollingHash_rotate(rollingHash, cur[-1],
-                                          cur[ldmParams.minMatchLength-1],
+                                              cur[ldmParams.minMatchLength-1],
-                                          state->hashPower);
+                                              state->hashPower);
         ZSTD_ldm_makeEntryAndInsertByTag(state,
                                          rollingHash, hBits,
                                          (U32)(cur - base), ldmParams);
@@ -297,8 +248,8 @@ static size_t ZSTD_ldm_generateSequences_internal(
     U64 const hashPower = ldmState->hashPower;
     U32 const hBits = params->hashLog - params->bucketSizeLog;
     U32 const ldmBucketSize = 1U << params->bucketSizeLog;
-    U32 const hashEveryLog = params->hashEveryLog;
+    U32 const hashRateLog = params->hashRateLog;
-    U32 const ldmTagMask = (1U << params->hashEveryLog) - 1;
+    U32 const ldmTagMask = (1U << params->hashRateLog) - 1;
     /* Prefix and extDict parameters */
     U32 const dictLimit = ldmState->window.dictLimit;
     U32 const lowestIndex = extDict ? ldmState->window.lowLimit : dictLimit;
@@ -324,16 +275,16 @@ static size_t ZSTD_ldm_generateSequences_internal(
         size_t forwardMatchLength = 0, backwardMatchLength = 0;
         ldmEntry_t* bestEntry = NULL;
         if (ip != istart) {
-            rollingHash = ZSTD_ldm_updateHash(rollingHash, lastHashed[0],
+            rollingHash = ZSTD_rollingHash_rotate(rollingHash, lastHashed[0],
-                                              lastHashed[minMatchLength],
+                                                  lastHashed[minMatchLength],
-                                              hashPower);
+                                                  hashPower);
         } else {
-            rollingHash = ZSTD_ldm_getRollingHash(ip, minMatchLength);
+            rollingHash = ZSTD_rollingHash_compute(ip, minMatchLength);
         }
         lastHashed = ip;
 
         /* Do not insert and do not look for a match */
-        if (ZSTD_ldm_getTag(rollingHash, hBits, hashEveryLog) != ldmTagMask) {
+        if (ZSTD_ldm_getTag(rollingHash, hBits, hashRateLog) != ldmTagMask) {
            ip++;
            continue;
         }
@@ -593,7 +544,7 @@ size_t ZSTD_ldm_blockCompress(rawSeqStore_t* rawSeqStore,
     void const* src, size_t srcSize)
 {
     const ZSTD_compressionParameters* const cParams = &ms->cParams;
-    unsigned const minMatch = cParams->searchLength;
+    unsigned const minMatch = cParams->minMatch;
     ZSTD_blockCompressor const blockCompressor =
         ZSTD_selectBlockCompressor(cParams->strategy, ZSTD_matchState_dictMode(ms));
     /* Input bounds */

C/zstd/zstd_ldm.h

@@ -21,7 +21,7 @@ extern "C" {
 *  Long distance matching
 ***************************************/
 
-#define ZSTD_LDM_DEFAULT_WINDOW_LOG ZSTD_WINDOWLOG_DEFAULTMAX
+#define ZSTD_LDM_DEFAULT_WINDOW_LOG ZSTD_WINDOWLOG_LIMIT_DEFAULT
 
 /**
  * ZSTD_ldm_generateSequences():
@@ -86,12 +86,8 @@ size_t ZSTD_ldm_getTableSize(ldmParams_t params);
  */
 size_t ZSTD_ldm_getMaxNbSeq(ldmParams_t params, size_t maxChunkSize);
 
-/** ZSTD_ldm_getTableSize() :
- *  Return prime8bytes^(minMatchLength-1) */
-U64 ZSTD_ldm_getHashPower(U32 minMatchLength);
-
 /** ZSTD_ldm_adjustParameters() :
- *  If the params->hashEveryLog is not set, set it to its default value based on
+ *  If the params->hashRateLog is not set, set it to its default value based on
  *  windowLog and params->hashLog.
  *
  *  Ensures that params->bucketSizeLog is <= params->hashLog (setting it to

C/zstd/zstd_opt.c

@@ -17,6 +17,8 @@
 #define ZSTD_FREQ_DIV       4   /* log factor when using previous stats to init next 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 */
+
 
 /*-*************************************
 *  Price functions for optimal parser
@@ -52,11 +54,15 @@ MEM_STATIC U32 ZSTD_fracWeight(U32 rawStat)
     return weight;
 }
 
-/* debugging function, @return price in bytes */
+#if (DEBUGLEVEL>=2)
+/* debugging function,
+ * @return price in bytes as fractional value
+ * for debug messages only */
 MEM_STATIC double ZSTD_fCost(U32 price)
 {
     return (double)price / (BITCOST_MULTIPLIER*8);
 }
+#endif
 
 static void ZSTD_setBasePrices(optState_t* optPtr, int optLevel)
 {
@@ -67,29 +73,44 @@ static void ZSTD_setBasePrices(optState_t* optPtr, int optLevel)
 }
 
 
-static U32 ZSTD_downscaleStat(U32* table, U32 lastEltIndex, int malus)
+/* ZSTD_downscaleStat() :
+ * reduce all elements in table by a factor 2^(ZSTD_FREQ_DIV+malus)
+ * return the resulting sum of elements */
+static U32 ZSTD_downscaleStat(unsigned* table, U32 lastEltIndex, int malus)
 {
     U32 s, sum=0;
+    DEBUGLOG(5, "ZSTD_downscaleStat (nbElts=%u)", (unsigned)lastEltIndex+1);
     assert(ZSTD_FREQ_DIV+malus > 0 && ZSTD_FREQ_DIV+malus < 31);
-    for (s=0; s<=lastEltIndex; s++) {
+    for (s=0; s<lastEltIndex+1; s++) {
         table[s] = 1 + (table[s] >> (ZSTD_FREQ_DIV+malus));
         sum += table[s];
     }
     return sum;
 }
 
-static void ZSTD_rescaleFreqs(optState_t* const optPtr,
+/* ZSTD_rescaleFreqs() :
-                              const BYTE* const src, size_t const srcSize,
+ * if first block (detected by optPtr->litLengthSum == 0) : init statistics
-                              int optLevel)
+ *    take hints from dictionary if there is one
+ *    or init from zero, using src for literals stats, or flat 1 for match symbols
+ * otherwise downscale existing stats, to be used as seed for next block.
+ */
+static void
+ZSTD_rescaleFreqs(optState_t* const optPtr,
+            const BYTE* const src, size_t const srcSize,
+                  int const optLevel)
 {
+    DEBUGLOG(5, "ZSTD_rescaleFreqs (srcSize=%u)", (unsigned)srcSize);
     optPtr->priceType = zop_dynamic;
 
     if (optPtr->litLengthSum == 0) {  /* first block : init */
-        if (srcSize <= 1024)   /* heuristic */
+        if (srcSize <= ZSTD_PREDEF_THRESHOLD) {  /* heuristic */
+            DEBUGLOG(5, "(srcSize <= ZSTD_PREDEF_THRESHOLD) => zop_predef");
             optPtr->priceType = zop_predef;
+        }
 
         assert(optPtr->symbolCosts != NULL);
-        if (optPtr->symbolCosts->huf.repeatMode == HUF_repeat_valid) { /* huffman table presumed generated by dictionary */
+        if (optPtr->symbolCosts->huf.repeatMode == HUF_repeat_valid) {
+            /* huffman table presumed generated by dictionary */
             optPtr->priceType = zop_dynamic;
 
             assert(optPtr->litFreq != NULL);
@@ -208,7 +229,9 @@ static U32 ZSTD_litLengthPrice(U32 const litLength, const optState_t* const optP
 
     /* dynamic statistics */
     {   U32 const llCode = ZSTD_LLcode(litLength);
-        return (LL_bits[llCode] * BITCOST_MULTIPLIER) + (optPtr->litLengthSumBasePrice - WEIGHT(optPtr->litLengthFreq[llCode], optLevel));
+        return (LL_bits[llCode] * BITCOST_MULTIPLIER)
+             + optPtr->litLengthSumBasePrice
+             - WEIGHT(optPtr->litLengthFreq[llCode], optLevel);
     }
 }
 
@@ -253,7 +276,7 @@ static int ZSTD_literalsContribution(const BYTE* const literals, U32 const litLe
 FORCE_INLINE_TEMPLATE U32
 ZSTD_getMatchPrice(U32 const offset,
                    U32 const matchLength,
-                   const optState_t* const optPtr,
+             const optState_t* const optPtr,
                    int const optLevel)
 {
     U32 price;
@@ -385,7 +408,6 @@ static U32 ZSTD_insertBt1(
     U32* largerPtr  = smallerPtr + 1;
     U32 dummy32;   /* to be nullified at the end */
     U32 const windowLow = ms->window.lowLimit;
-    U32 const matchLow = windowLow ? windowLow : 1;
     U32 matchEndIdx = current+8+1;
     size_t bestLength = 8;
     U32 nbCompares = 1U << cParams->searchLog;
@@ -401,7 +423,8 @@ static U32 ZSTD_insertBt1(
     assert(ip <= iend-8);   /* required for h calculation */
     hashTable[h] = current;   /* Update Hash Table */
 
-    while (nbCompares-- && (matchIndex >= matchLow)) {
+    assert(windowLow > 0);
+    while (nbCompares-- && (matchIndex >= windowLow)) {
         U32* const nextPtr = bt + 2*(matchIndex & btMask);
         size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger);   /* guaranteed minimum nb of common bytes */
         assert(matchIndex < current);
@@ -479,7 +502,7 @@ void ZSTD_updateTree_internal(
     const BYTE* const base = ms->window.base;
     U32 const target = (U32)(ip - base);
     U32 idx = ms->nextToUpdate;
-    DEBUGLOG(5, "ZSTD_updateTree_internal, from %u to %u  (dictMode:%u)",
+    DEBUGLOG(6, "ZSTD_updateTree_internal, from %u to %u  (dictMode:%u)",
                 idx, target, dictMode);
 
     while(idx < target)
@@ -488,15 +511,18 @@ void ZSTD_updateTree_internal(
 }
 
 void ZSTD_updateTree(ZSTD_matchState_t* ms, const BYTE* ip, const BYTE* iend) {
-    ZSTD_updateTree_internal(ms, ip, iend, ms->cParams.searchLength, ZSTD_noDict);
+    ZSTD_updateTree_internal(ms, ip, iend, ms->cParams.minMatch, ZSTD_noDict);
 }
 
 FORCE_INLINE_TEMPLATE
 U32 ZSTD_insertBtAndGetAllMatches (
                     ZSTD_matchState_t* ms,
                     const BYTE* const ip, const BYTE* const iLimit, const ZSTD_dictMode_e dictMode,
-                    U32 rep[ZSTD_REP_NUM], U32 const ll0,
+                    U32 rep[ZSTD_REP_NUM],
-                    ZSTD_match_t* matches, const U32 lengthToBeat, U32 const mls /* template */)
+                    U32 const ll0,   /* tells if associated literal length is 0 or not. This value must be 0 or 1 */
+                    ZSTD_match_t* matches,
+                    const U32 lengthToBeat,
+                    U32 const mls /* template */)
 {
     const ZSTD_compressionParameters* const cParams = &ms->cParams;
     U32 const sufficient_len = MIN(cParams->targetLength, ZSTD_OPT_NUM -1);
@@ -542,6 +568,7 @@ U32 ZSTD_insertBtAndGetAllMatches (
     DEBUGLOG(8, "ZSTD_insertBtAndGetAllMatches: current=%u", current);
 
     /* check repCode */
+    assert(ll0 <= 1);   /* necessarily 1 or 0 */
     {   U32 const lastR = ZSTD_REP_NUM + ll0;
         U32 repCode;
         for (repCode = ll0; repCode < lastR; repCode++) {
@@ -724,7 +751,7 @@ FORCE_INLINE_TEMPLATE U32 ZSTD_BtGetAllMatches (
                         ZSTD_match_t* matches, U32 const lengthToBeat)
 {
     const ZSTD_compressionParameters* const cParams = &ms->cParams;
-    U32 const matchLengthSearch = cParams->searchLength;
+    U32 const matchLengthSearch = cParams->minMatch;
     DEBUGLOG(8, "ZSTD_BtGetAllMatches");
     if (ip < ms->window.base + ms->nextToUpdate) return 0;   /* skipped area */
     ZSTD_updateTree_internal(ms, ip, iHighLimit, matchLengthSearch, dictMode);
@@ -774,12 +801,30 @@ static U32 ZSTD_totalLen(ZSTD_optimal_t sol)
     return sol.litlen + sol.mlen;
 }
 
+#if 0 /* debug */
+
+static void
+listStats(const U32* table, int lastEltID)
+{
+    int const nbElts = lastEltID + 1;
+    int enb;
+    for (enb=0; enb < nbElts; enb++) {
+        (void)table;
+        //RAWLOG(2, "%3i:%3i,  ", enb, table[enb]);
+        RAWLOG(2, "%4i,", table[enb]);
+    }
+    RAWLOG(2, " \n");
+}
+
+#endif
+
 FORCE_INLINE_TEMPLATE size_t
 ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
                                seqStore_t* seqStore,
                                U32 rep[ZSTD_REP_NUM],
-                               const void* src, size_t srcSize,
+                         const void* src, size_t srcSize,
-                               const int optLevel, const ZSTD_dictMode_e dictMode)
+                         const int optLevel,
+                         const ZSTD_dictMode_e dictMode)
 {
     optState_t* const optStatePtr = &ms->opt;
     const BYTE* const istart = (const BYTE*)src;
@@ -792,14 +837,15 @@ ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
     const ZSTD_compressionParameters* const cParams = &ms->cParams;
 
     U32 const sufficient_len = MIN(cParams->targetLength, ZSTD_OPT_NUM -1);
-    U32 const minMatch = (cParams->searchLength == 3) ? 3 : 4;
+    U32 const minMatch = (cParams->minMatch == 3) ? 3 : 4;
 
     ZSTD_optimal_t* const opt = optStatePtr->priceTable;
     ZSTD_match_t* const matches = optStatePtr->matchTable;
     ZSTD_optimal_t lastSequence;
 
     /* init */
-    DEBUGLOG(5, "ZSTD_compressBlock_opt_generic");
+    DEBUGLOG(5, "ZSTD_compressBlock_opt_generic: current=%u, prefix=%u, nextToUpdate=%u",
+                (U32)(ip - base), ms->window.dictLimit, ms->nextToUpdate);
     assert(optLevel <= 2);
     ms->nextToUpdate3 = ms->nextToUpdate;
     ZSTD_rescaleFreqs(optStatePtr, (const BYTE*)src, srcSize, optLevel);
@@ -999,7 +1045,7 @@ _shortestPath:   /* cur, last_pos, best_mlen, best_off have to be set */
                     U32 const offCode = opt[storePos].off;
                     U32 const advance = llen + mlen;
                     DEBUGLOG(6, "considering seq starting at %zi, llen=%u, mlen=%u",
-                                anchor - istart, llen, mlen);
+                                anchor - istart, (unsigned)llen, (unsigned)mlen);
 
                     if (mlen==0) {  /* only literals => must be last "sequence", actually starting a new stream of sequences */
                         assert(storePos == storeEnd);   /* must be last sequence */
@@ -1047,11 +1093,11 @@ size_t ZSTD_compressBlock_btopt(
 
 
 /* used in 2-pass strategy */
-static U32 ZSTD_upscaleStat(U32* table, U32 lastEltIndex, int bonus)
+static U32 ZSTD_upscaleStat(unsigned* table, U32 lastEltIndex, int bonus)
 {
     U32 s, sum=0;
-    assert(ZSTD_FREQ_DIV+bonus > 0);
+    assert(ZSTD_FREQ_DIV+bonus >= 0);
-    for (s=0; s<=lastEltIndex; s++) {
+    for (s=0; s<lastEltIndex+1; s++) {
         table[s] <<= ZSTD_FREQ_DIV+bonus;
         table[s]--;
         sum += table[s];
@@ -1063,9 +1109,43 @@ static U32 ZSTD_upscaleStat(U32* table, U32 lastEltIndex, int bonus)
 MEM_STATIC void ZSTD_upscaleStats(optState_t* optPtr)
 {
     optPtr->litSum = ZSTD_upscaleStat(optPtr->litFreq, MaxLit, 0);
-    optPtr->litLengthSum = ZSTD_upscaleStat(optPtr->litLengthFreq, MaxLL, 1);
+    optPtr->litLengthSum = ZSTD_upscaleStat(optPtr->litLengthFreq, MaxLL, 0);
-    optPtr->matchLengthSum = ZSTD_upscaleStat(optPtr->matchLengthFreq, MaxML, 1);
+    optPtr->matchLengthSum = ZSTD_upscaleStat(optPtr->matchLengthFreq, MaxML, 0);
-    optPtr->offCodeSum = ZSTD_upscaleStat(optPtr->offCodeFreq, MaxOff, 1);
+    optPtr->offCodeSum = ZSTD_upscaleStat(optPtr->offCodeFreq, MaxOff, 0);
+}
+
+/* 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 constract must be respected.
+ */
+static void
+ZSTD_initStats_ultra(ZSTD_matchState_t* ms,
+                     seqStore_t* seqStore,
+                     U32 rep[ZSTD_REP_NUM],
+               const void* src, size_t srcSize)
+{
+    U32 tmpRep[ZSTD_REP_NUM];  /* updated rep codes will sink here */
+    memcpy(tmpRep, rep, sizeof(tmpRep));
+
+    DEBUGLOG(4, "ZSTD_initStats_ultra (srcSize=%zu)", srcSize);
+    assert(ms->opt.litLengthSum == 0);    /* first block */
+    assert(seqStore->sequences == seqStore->sequencesStart);   /* no ldm */
+    assert(ms->window.dictLimit == ms->window.lowLimit);   /* no dictionary */
+    assert(ms->window.dictLimit - ms->nextToUpdate <= 1);  /* no prefix (note: intentional overflow, defined as 2-complement) */
+
+    ZSTD_compressBlock_opt_generic(ms, seqStore, tmpRep, src, srcSize, 2 /*optLevel*/, ZSTD_noDict);   /* generate stats into ms->opt*/
+
+    /* invalidate first scan from history */
+    ZSTD_resetSeqStore(seqStore);
+    ms->window.base -= srcSize;
+    ms->window.dictLimit += (U32)srcSize;
+    ms->window.lowLimit = ms->window.dictLimit;
+    ms->nextToUpdate = ms->window.dictLimit;
+    ms->nextToUpdate3 = ms->window.dictLimit;
+
+    /* re-inforce weight of collected statistics */
+    ZSTD_upscaleStats(&ms->opt);
 }
 
 size_t ZSTD_compressBlock_btultra(
@@ -1073,33 +1153,34 @@ size_t ZSTD_compressBlock_btultra(
         const void* src, size_t srcSize)
 {
     DEBUGLOG(5, "ZSTD_compressBlock_btultra (srcSize=%zu)", srcSize);
-#if 0
+    return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 2 /*optLevel*/, ZSTD_noDict);
-    /* 2-pass strategy (disabled)
+}
+
+size_t ZSTD_compressBlock_btultra2(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        const void* src, size_t srcSize)
+{
+    U32 const current = (U32)((const BYTE*)src - ms->window.base);
+    DEBUGLOG(5, "ZSTD_compressBlock_btultra2 (srcSize=%zu)", srcSize);
+
+    /* 2-pass 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.
+     * 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. */
     assert(srcSize <= ZSTD_BLOCKSIZE_MAX);
     if ( (ms->opt.litLengthSum==0)   /* first block */
-      && (seqStore->sequences == seqStore->sequencesStart)   /* no ldm */
+      && (seqStore->sequences == seqStore->sequencesStart)  /* no ldm */
-      && (ms->window.dictLimit == ms->window.lowLimit) ) {   /* no dictionary */
+      && (ms->window.dictLimit == ms->window.lowLimit)   /* no dictionary */
-        U32 tmpRep[ZSTD_REP_NUM];
+      && (current == ms->window.dictLimit)   /* start of frame, nothing already loaded nor skipped */
-        DEBUGLOG(5, "ZSTD_compressBlock_btultra: first block: collecting statistics");
+      && (srcSize > ZSTD_PREDEF_THRESHOLD)
-        assert(ms->nextToUpdate >= ms->window.dictLimit
+      ) {
-            && ms->nextToUpdate <= ms->window.dictLimit + 1);
+        ZSTD_initStats_ultra(ms, seqStore, rep, src, srcSize);
-        memcpy(tmpRep, rep, sizeof(tmpRep));
-        ZSTD_compressBlock_opt_generic(ms, seqStore, tmpRep, src, srcSize, 2 /*optLevel*/, ZSTD_noDict);   /* generate stats into ms->opt*/
-        ZSTD_resetSeqStore(seqStore);
-        /* invalidate first scan from history */
-        ms->window.base -= srcSize;
-        ms->window.dictLimit += (U32)srcSize;
-        ms->window.lowLimit = ms->window.dictLimit;
-        ms->nextToUpdate = ms->window.dictLimit;
-        ms->nextToUpdate3 = ms->window.dictLimit;
-        /* re-inforce weight of collected statistics */
-        ZSTD_upscaleStats(&ms->opt);
     }
-#endif
+
     return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 2 /*optLevel*/, ZSTD_noDict);
 }
 
@@ -1130,3 +1211,7 @@ size_t ZSTD_compressBlock_btultra_extDict(
 {
     return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 2 /*optLevel*/, ZSTD_extDict);
 }
+
+/* note : no btultra2 variant for extDict nor dictMatchState,
+ * because btultra2 is not meant to work with dictionaries
+ * and is only specific for the first block (no prefix) */

C/zstd/zstd_opt.h

@@ -26,6 +26,10 @@ size_t ZSTD_compressBlock_btopt(
 size_t ZSTD_compressBlock_btultra(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_btultra2(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+
 
 size_t ZSTD_compressBlock_btopt_dictMatchState(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
@@ -41,6 +45,10 @@ size_t ZSTD_compressBlock_btultra_extDict(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
 
+        /* note : no btultra2 variant for extDict nor dictMatchState,
+         * because btultra2 is not meant to work with dictionaries
+         * and is only specific for the first block (no prefix) */
+
 #if defined (__cplusplus)
 }
 #endif

C/zstd/zstd_v04.c

@@ -240,17 +240,7 @@ MEM_STATIC size_t MEM_readLEST(const void* memPtr)
 /* *************************************
 *  Types
 ***************************************/
-#define ZSTD_WINDOWLOG_MAX 26
-#define ZSTD_WINDOWLOG_MIN 18
 #define ZSTD_WINDOWLOG_ABSOLUTEMIN 11
-#define ZSTD_CONTENTLOG_MAX (ZSTD_WINDOWLOG_MAX+1)
-#define ZSTD_CONTENTLOG_MIN 4
-#define ZSTD_HASHLOG_MAX 28
-#define ZSTD_HASHLOG_MIN 4
-#define ZSTD_SEARCHLOG_MAX (ZSTD_CONTENTLOG_MAX-1)
-#define ZSTD_SEARCHLOG_MIN 1
-#define ZSTD_SEARCHLENGTH_MAX 7
-#define ZSTD_SEARCHLENGTH_MIN 4
 
 /** from faster to stronger */
 typedef enum { ZSTD_fast, ZSTD_greedy, ZSTD_lazy, ZSTD_lazy2, ZSTD_btlazy2 } ZSTD_strategy;

C/zstd/zstd_v05.c

@@ -836,7 +836,7 @@ MEM_STATIC void BITv05_skipBits(BITv05_DStream_t* bitD, U32 nbBits)
     bitD->bitsConsumed += nbBits;
 }
 
-MEM_STATIC size_t BITv05_readBits(BITv05_DStream_t* bitD, U32 nbBits)
+MEM_STATIC size_t BITv05_readBits(BITv05_DStream_t* bitD, unsigned nbBits)
 {
     size_t value = BITv05_lookBits(bitD, nbBits);
     BITv05_skipBits(bitD, nbBits);
@@ -845,7 +845,7 @@ MEM_STATIC size_t BITv05_readBits(BITv05_DStream_t* bitD, U32 nbBits)
 
 /*!BITv05_readBitsFast :
 *  unsafe version; only works only if nbBits >= 1 */
-MEM_STATIC size_t BITv05_readBitsFast(BITv05_DStream_t* bitD, U32 nbBits)
+MEM_STATIC size_t BITv05_readBitsFast(BITv05_DStream_t* bitD, unsigned nbBits)
 {
     size_t value = BITv05_lookBitsFast(bitD, nbBits);
     BITv05_skipBits(bitD, nbBits);
@@ -1162,7 +1162,7 @@ MEM_STATIC unsigned FSEv05_endOfDState(const FSEv05_DState_t* DStatePtr)
 /* **************************************************************
 *  Complex types
 ****************************************************************/
-typedef U32 DTable_max_t[FSEv05_DTABLE_SIZE_U32(FSEv05_MAX_TABLELOG)];
+typedef unsigned DTable_max_t[FSEv05_DTABLE_SIZE_U32(FSEv05_MAX_TABLELOG)];
 
 
 /* **************************************************************
@@ -2191,7 +2191,7 @@ static void HUFv05_fillDTableX4(HUFv05_DEltX4* DTable, const U32 targetLog,
     }
 }
 
-size_t HUFv05_readDTableX4 (U32* DTable, const void* src, size_t srcSize)
+size_t HUFv05_readDTableX4 (unsigned* DTable, const void* src, size_t srcSize)
 {
     BYTE weightList[HUFv05_MAX_SYMBOL_VALUE + 1];
     sortedSymbol_t sortedSymbol[HUFv05_MAX_SYMBOL_VALUE + 1];
@@ -2205,7 +2205,7 @@ size_t HUFv05_readDTableX4 (U32* DTable, const void* src, size_t srcSize)
     void* dtPtr = DTable;
     HUFv05_DEltX4* const dt = ((HUFv05_DEltX4*)dtPtr) + 1;
 
-    HUFv05_STATIC_ASSERT(sizeof(HUFv05_DEltX4) == sizeof(U32));   /* if compilation fails here, assertion is false */
+    HUFv05_STATIC_ASSERT(sizeof(HUFv05_DEltX4) == sizeof(unsigned));   /* if compilation fails here, assertion is false */
     if (memLog > HUFv05_ABSOLUTEMAX_TABLELOG) return ERROR(tableLog_tooLarge);
     //memset(weightList, 0, sizeof(weightList));   /* is not necessary, even though some analyzer complain ... */
 
@@ -2332,7 +2332,7 @@ static inline size_t HUFv05_decodeStreamX4(BYTE* p, BITv05_DStream_t* bitDPtr, B
 size_t HUFv05_decompress1X4_usingDTable(
           void* dst,  size_t dstSize,
     const void* cSrc, size_t cSrcSize,
-    const U32* DTable)
+    const unsigned* DTable)
 {
     const BYTE* const istart = (const BYTE*) cSrc;
     BYTE* const ostart = (BYTE*) dst;
@@ -2375,7 +2375,7 @@ size_t HUFv05_decompress1X4 (void* dst, size_t dstSize, const void* cSrc, size_t
 size_t HUFv05_decompress4X4_usingDTable(
           void* dst,  size_t dstSize,
     const void* cSrc, size_t cSrcSize,
-    const U32* DTable)
+    const unsigned* DTable)
 {
     if (cSrcSize < 10) return ERROR(corruption_detected);   /* strict minimum : jump table + 1 byte per stream */
 
@@ -2999,7 +2999,7 @@ static size_t ZSTDv05_decodeSeqHeaders(int* nbSeq, const BYTE** dumpsPtr, size_t
     const BYTE* ip = istart;
     const BYTE* const iend = istart + srcSize;
     U32 LLtype, Offtype, MLtype;
-    U32 LLlog, Offlog, MLlog;
+    unsigned LLlog, Offlog, MLlog;
     size_t dumpsLength;
 
     /* check */
@@ -3057,7 +3057,7 @@ static size_t ZSTDv05_decodeSeqHeaders(int* nbSeq, const BYTE** dumpsPtr, size_t
             break;
         case FSEv05_ENCODING_DYNAMIC :
         default :   /* impossible */
-            {   U32 max = MaxLL;
+            {   unsigned max = MaxLL;
                 headerSize = FSEv05_readNCount(norm, &max, &LLlog, ip, iend-ip);
                 if (FSEv05_isError(headerSize)) return ERROR(GENERIC);
                 if (LLlog > LLFSEv05Log) return ERROR(corruption_detected);
@@ -3081,7 +3081,7 @@ static size_t ZSTDv05_decodeSeqHeaders(int* nbSeq, const BYTE** dumpsPtr, size_t
             break;
         case FSEv05_ENCODING_DYNAMIC :
         default :   /* impossible */
-            {   U32 max = MaxOff;
+            {   unsigned max = MaxOff;
                 headerSize = FSEv05_readNCount(norm, &max, &Offlog, ip, iend-ip);
                 if (FSEv05_isError(headerSize)) return ERROR(GENERIC);
                 if (Offlog > OffFSEv05Log) return ERROR(corruption_detected);
@@ -3105,7 +3105,7 @@ static size_t ZSTDv05_decodeSeqHeaders(int* nbSeq, const BYTE** dumpsPtr, size_t
             break;
         case FSEv05_ENCODING_DYNAMIC :
         default :   /* impossible */
-            {   U32 max = MaxML;
+            {   unsigned max = MaxML;
                 headerSize = FSEv05_readNCount(norm, &max, &MLlog, ip, iend-ip);
                 if (FSEv05_isError(headerSize)) return ERROR(GENERIC);
                 if (MLlog > MLFSEv05Log) return ERROR(corruption_detected);
@@ -3305,9 +3305,9 @@ static size_t ZSTDv05_decompressSequences(
     const BYTE* const litEnd = litPtr + dctx->litSize;
     int nbSeq=0;
     const BYTE* dumps = NULL;
-    U32* DTableLL = dctx->LLTable;
+    unsigned* DTableLL = dctx->LLTable;
-    U32* DTableML = dctx->MLTable;
+    unsigned* DTableML = dctx->MLTable;
-    U32* DTableOffb = dctx->OffTable;
+    unsigned* DTableOffb = dctx->OffTable;
     const BYTE* const base = (const BYTE*) (dctx->base);
     const BYTE* const vBase = (const BYTE*) (dctx->vBase);
     const BYTE* const dictEnd = (const BYTE*) (dctx->dictEnd);
@@ -3633,7 +3633,7 @@ static size_t ZSTDv05_loadEntropy(ZSTDv05_DCtx* dctx, const void* dict, size_t d
 {
     size_t hSize, offcodeHeaderSize, matchlengthHeaderSize, errorCode, litlengthHeaderSize;
     short offcodeNCount[MaxOff+1];
-    U32 offcodeMaxValue=MaxOff, offcodeLog;
+    unsigned offcodeMaxValue=MaxOff, offcodeLog;
     short matchlengthNCount[MaxML+1];
     unsigned matchlengthMaxValue = MaxML, matchlengthLog;
     short litlengthNCount[MaxLL+1];

C/zstd/zstdmt_compress.c

@@ -9,21 +9,19 @@
  */
 
 
-/* ======   Tuning parameters   ====== */
-#define ZSTDMT_NBWORKERS_MAX 200
-#define ZSTDMT_JOBSIZE_MAX  (MEM_32bits() ? (512 MB) : (2 GB))  /* note : limited by `jobSize` type, which is `unsigned` */
-#define ZSTDMT_OVERLAPLOG_DEFAULT 6
-
-
 /* ======   Compiler specifics   ====== */
 #if defined(_MSC_VER)
 #  pragma warning(disable : 4204)   /* disable: C4204: non-constant aggregate initializer */
 #endif
 
 
+/* ======   Constants   ====== */
+#define ZSTDMT_OVERLAPLOG_DEFAULT 0
+
+
 /* ======   Dependencies   ====== */
 #include <string.h>      /* memcpy, memset */
-#include <limits.h>      /* INT_MAX */
+#include <limits.h>      /* INT_MAX, UINT_MAX */
 #include "pool.h"        /* threadpool */
 #include "threading.h"   /* mutex */
 #include "zstd_compress_internal.h"  /* MIN, ERROR, ZSTD_*, ZSTD_highbit32 */
@@ -57,9 +55,9 @@ static unsigned long long GetCurrentClockTimeMicroseconds(void)
    static clock_t _ticksPerSecond = 0;
    if (_ticksPerSecond <= 0) _ticksPerSecond = sysconf(_SC_CLK_TCK);
 
-   { struct tms junk; clock_t newTicks = (clock_t) times(&junk);
+   {   struct tms junk; clock_t newTicks = (clock_t) times(&junk);
-     return ((((unsigned long long)newTicks)*(1000000))/_ticksPerSecond); }
+       return ((((unsigned long long)newTicks)*(1000000))/_ticksPerSecond);
-}
+}  }
 
 #define MUTEX_WAIT_TIME_DLEVEL 6
 #define ZSTD_PTHREAD_MUTEX_LOCK(mutex) {          \
@@ -342,8 +340,8 @@ static ZSTDMT_seqPool* ZSTDMT_expandSeqPool(ZSTDMT_seqPool* pool, U32 nbWorkers)
 
 typedef struct {
     ZSTD_pthread_mutex_t poolMutex;
-    unsigned totalCCtx;
+    int totalCCtx;
-    unsigned availCCtx;
+    int availCCtx;
     ZSTD_customMem cMem;
     ZSTD_CCtx* cctx[1];   /* variable size */
 } ZSTDMT_CCtxPool;
@@ -351,16 +349,16 @@ typedef struct {
 /* note : all CCtx borrowed from the pool should be released back to the pool _before_ freeing the pool */
 static void ZSTDMT_freeCCtxPool(ZSTDMT_CCtxPool* pool)
 {
-    unsigned u;
+    int cid;
-    for (u=0; u<pool->totalCCtx; u++)
+    for (cid=0; cid<pool->totalCCtx; cid++)
-        ZSTD_freeCCtx(pool->cctx[u]);  /* note : compatible with free on NULL */
+        ZSTD_freeCCtx(pool->cctx[cid]);  /* note : compatible with free on NULL */
     ZSTD_pthread_mutex_destroy(&pool->poolMutex);
     ZSTD_free(pool, pool->cMem);
 }
 
 /* ZSTDMT_createCCtxPool() :
  * implies nbWorkers >= 1 , checked by caller ZSTDMT_createCCtx() */
-static ZSTDMT_CCtxPool* ZSTDMT_createCCtxPool(unsigned nbWorkers,
+static ZSTDMT_CCtxPool* ZSTDMT_createCCtxPool(int nbWorkers,
                                               ZSTD_customMem cMem)
 {
     ZSTDMT_CCtxPool* const cctxPool = (ZSTDMT_CCtxPool*) ZSTD_calloc(
@@ -381,7 +379,7 @@ static ZSTDMT_CCtxPool* ZSTDMT_createCCtxPool(unsigned nbWorkers,
 }
 
 static ZSTDMT_CCtxPool* ZSTDMT_expandCCtxPool(ZSTDMT_CCtxPool* srcPool,
-                                              unsigned nbWorkers)
+                                              int nbWorkers)
 {
     if (srcPool==NULL) return NULL;
     if (nbWorkers <= srcPool->totalCCtx) return srcPool;   /* good enough */
@@ -469,9 +467,9 @@ static int ZSTDMT_serialState_reset(serialState_t* serialState, ZSTDMT_seqPool*
         DEBUGLOG(4, "LDM window size = %u KB", (1U << params.cParams.windowLog) >> 10);
         ZSTD_ldm_adjustParameters(&params.ldmParams, &params.cParams);
         assert(params.ldmParams.hashLog >= params.ldmParams.bucketSizeLog);
-        assert(params.ldmParams.hashEveryLog < 32);
+        assert(params.ldmParams.hashRateLog < 32);
         serialState->ldmState.hashPower =
-                ZSTD_ldm_getHashPower(params.ldmParams.minMatchLength);
+                ZSTD_rollingHash_primePower(params.ldmParams.minMatchLength);
     } else {
         memset(&params.ldmParams, 0, sizeof(params.ldmParams));
     }
@@ -674,7 +672,7 @@ static void ZSTDMT_compressionJob(void* jobDescription)
         if (ZSTD_isError(initError)) JOB_ERROR(initError);
     } else {  /* srcStart points at reloaded section */
         U64 const pledgedSrcSize = job->firstJob ? job->fullFrameSize : job->src.size;
-        {   size_t const forceWindowError = ZSTD_CCtxParam_setParameter(&jobParams, ZSTD_p_forceMaxWindow, !job->firstJob);
+        {   size_t const forceWindowError = ZSTD_CCtxParam_setParameter(&jobParams, ZSTD_c_forceMaxWindow, !job->firstJob);
             if (ZSTD_isError(forceWindowError)) JOB_ERROR(forceWindowError);
         }
         {   size_t const initError = ZSTD_compressBegin_advanced_internal(cctx,
@@ -777,6 +775,14 @@ typedef struct {
 
 static const roundBuff_t kNullRoundBuff = {NULL, 0, 0};
 
+#define RSYNC_LENGTH 32
+
+typedef struct {
+  U64 hash;
+  U64 hitMask;
+  U64 primePower;
+} rsyncState_t;
+
 struct ZSTDMT_CCtx_s {
     POOL_ctx* factory;
     ZSTDMT_jobDescription* jobs;
@@ -790,6 +796,7 @@ struct ZSTDMT_CCtx_s {
     inBuff_t inBuff;
     roundBuff_t roundBuff;
     serialState_t serial;
+    rsyncState_t rsync;
     unsigned singleBlockingThread;
     unsigned jobIDMask;
     unsigned doneJobID;
@@ -859,7 +866,7 @@ size_t ZSTDMT_CCtxParam_setNbWorkers(ZSTD_CCtx_params* params, unsigned nbWorker
 {
     if (nbWorkers > ZSTDMT_NBWORKERS_MAX) nbWorkers = ZSTDMT_NBWORKERS_MAX;
     params->nbWorkers = nbWorkers;
-    params->overlapSizeLog = ZSTDMT_OVERLAPLOG_DEFAULT;
+    params->overlapLog = ZSTDMT_OVERLAPLOG_DEFAULT;
     params->jobSize = 0;
     return nbWorkers;
 }
@@ -969,52 +976,59 @@ size_t ZSTDMT_sizeof_CCtx(ZSTDMT_CCtx* mtctx)
 }
 
 /* Internal only */
-size_t ZSTDMT_CCtxParam_setMTCtxParameter(ZSTD_CCtx_params* params,
+size_t
-                                ZSTDMT_parameter parameter, unsigned value) {
+ZSTDMT_CCtxParam_setMTCtxParameter(ZSTD_CCtx_params* params,
+                                   ZSTDMT_parameter parameter,
+                                   int value)
+{
     DEBUGLOG(4, "ZSTDMT_CCtxParam_setMTCtxParameter");
     switch(parameter)
     {
     case ZSTDMT_p_jobSize :
-        DEBUGLOG(4, "ZSTDMT_CCtxParam_setMTCtxParameter : set jobSize to %u", value);
+        DEBUGLOG(4, "ZSTDMT_CCtxParam_setMTCtxParameter : set jobSize to %i", value);
-        if ( (value > 0)  /* value==0 => automatic job size */
+        if ( value != 0  /* default */
-           & (value < ZSTDMT_JOBSIZE_MIN) )
+          && value < ZSTDMT_JOBSIZE_MIN)
             value = ZSTDMT_JOBSIZE_MIN;
-        if (value > ZSTDMT_JOBSIZE_MAX)
+        assert(value >= 0);
-            value = ZSTDMT_JOBSIZE_MAX;
+        if (value > ZSTDMT_JOBSIZE_MAX) value = ZSTDMT_JOBSIZE_MAX;
         params->jobSize = value;
         return value;
-    case ZSTDMT_p_overlapSectionLog :
+
-        if (value > 9) value = 9;
+    case ZSTDMT_p_overlapLog :
-        DEBUGLOG(4, "ZSTDMT_p_overlapSectionLog : %u", value);
+        DEBUGLOG(4, "ZSTDMT_p_overlapLog : %i", value);
-        params->overlapSizeLog = (value >= 9) ? 9 : value;
+        if (value < ZSTD_OVERLAPLOG_MIN) value = ZSTD_OVERLAPLOG_MIN;
+        if (value > ZSTD_OVERLAPLOG_MAX) value = ZSTD_OVERLAPLOG_MAX;
+        params->overlapLog = value;
         return value;
+
+    case ZSTDMT_p_rsyncable :
+        value = (value != 0);
+        params->rsyncable = value;
+        return value;
+
     default :
         return ERROR(parameter_unsupported);
     }
 }
 
-size_t ZSTDMT_setMTCtxParameter(ZSTDMT_CCtx* mtctx, ZSTDMT_parameter parameter, unsigned value)
+size_t ZSTDMT_setMTCtxParameter(ZSTDMT_CCtx* mtctx, ZSTDMT_parameter parameter, int value)
 {
     DEBUGLOG(4, "ZSTDMT_setMTCtxParameter");
-    switch(parameter)
+    return ZSTDMT_CCtxParam_setMTCtxParameter(&mtctx->params, parameter, value);
-    {
-    case ZSTDMT_p_jobSize :
-        return ZSTDMT_CCtxParam_setMTCtxParameter(&mtctx->params, parameter, value);
-    case ZSTDMT_p_overlapSectionLog :
-        return ZSTDMT_CCtxParam_setMTCtxParameter(&mtctx->params, parameter, value);
-    default :
-        return ERROR(parameter_unsupported);
-    }
 }
 
-size_t ZSTDMT_getMTCtxParameter(ZSTDMT_CCtx* mtctx, ZSTDMT_parameter parameter, unsigned* value)
+size_t ZSTDMT_getMTCtxParameter(ZSTDMT_CCtx* mtctx, ZSTDMT_parameter parameter, int* value)
 {
     switch (parameter) {
     case ZSTDMT_p_jobSize:
-        *value = mtctx->params.jobSize;
+        assert(mtctx->params.jobSize <= INT_MAX);
+        *value = (int)(mtctx->params.jobSize);
         break;
-    case ZSTDMT_p_overlapSectionLog:
+    case ZSTDMT_p_overlapLog:
-        *value = mtctx->params.overlapSizeLog;
+        *value = mtctx->params.overlapLog;
+        break;
+    case ZSTDMT_p_rsyncable:
+        *value = mtctx->params.rsyncable;
         break;
     default:
         return ERROR(parameter_unsupported);
@@ -1140,22 +1154,66 @@ size_t ZSTDMT_toFlushNow(ZSTDMT_CCtx* mtctx)
 /* =====   Multi-threaded compression   ===== */
 /* ------------------------------------------ */
 
-static size_t ZSTDMT_computeTargetJobLog(ZSTD_CCtx_params const params)
+static unsigned ZSTDMT_computeTargetJobLog(ZSTD_CCtx_params const params)
 {
     if (params.ldmParams.enableLdm)
+        /* In Long Range Mode, the windowLog is typically oversized.
+         * In which case, it's preferable to determine the jobSize
+         * based on chainLog instead. */
         return MAX(21, params.cParams.chainLog + 4);
     return MAX(20, params.cParams.windowLog + 2);
 }
 
-static size_t ZSTDMT_computeOverlapLog(ZSTD_CCtx_params const params)
+static int ZSTDMT_overlapLog_default(ZSTD_strategy strat)
 {
-    unsigned const overlapRLog = (params.overlapSizeLog>9) ? 0 : 9-params.overlapSizeLog;
+    switch(strat)
-    if (params.ldmParams.enableLdm)
+    {
-        return (MIN(params.cParams.windowLog, ZSTDMT_computeTargetJobLog(params) - 2) - overlapRLog);
+        case ZSTD_btultra2:
-    return overlapRLog >= 9 ? 0 : (params.cParams.windowLog - overlapRLog);
+            return 9;
+        case ZSTD_btultra:
+        case ZSTD_btopt:
+            return 8;
+        case ZSTD_btlazy2:
+        case ZSTD_lazy2:
+            return 7;
+        case ZSTD_lazy:
+        case ZSTD_greedy:
+        case ZSTD_dfast:
+        case ZSTD_fast:
+        default:;
+    }
+    return 6;
 }
 
-static unsigned ZSTDMT_computeNbJobs(ZSTD_CCtx_params params, size_t srcSize, unsigned nbWorkers) {
+static int ZSTDMT_overlapLog(int ovlog, ZSTD_strategy strat)
+{
+    assert(0 <= ovlog && ovlog <= 9);
+    if (ovlog == 0) return ZSTDMT_overlapLog_default(strat);
+    return ovlog;
+}
+
+static size_t ZSTDMT_computeOverlapSize(ZSTD_CCtx_params const params)
+{
+    int const overlapRLog = 9 - ZSTDMT_overlapLog(params.overlapLog, params.cParams.strategy);
+    int ovLog = (overlapRLog >= 8) ? 0 : (params.cParams.windowLog - overlapRLog);
+    assert(0 <= overlapRLog && overlapRLog <= 8);
+    if (params.ldmParams.enableLdm) {
+        /* In Long Range Mode, the windowLog is typically oversized.
+         * In which case, it's preferable to determine the jobSize
+         * based on chainLog instead.
+         * Then, ovLog becomes a fraction of the jobSize, rather than windowSize */
+        ovLog = MIN(params.cParams.windowLog, ZSTDMT_computeTargetJobLog(params) - 2)
+                - overlapRLog;
+    }
+    assert(0 <= ovLog && ovLog <= 30);
+    DEBUGLOG(4, "overlapLog : %i", params.overlapLog);
+    DEBUGLOG(4, "overlap size : %i", 1 << ovLog);
+    return (ovLog==0) ? 0 : (size_t)1 << ovLog;
+}
+
+static unsigned
+ZSTDMT_computeNbJobs(ZSTD_CCtx_params params, size_t srcSize, unsigned nbWorkers)
+{
     assert(nbWorkers>0);
     {   size_t const jobSizeTarget = (size_t)1 << ZSTDMT_computeTargetJobLog(params);
         size_t const jobMaxSize = jobSizeTarget << 2;
@@ -1178,7 +1236,7 @@ static size_t ZSTDMT_compress_advanced_internal(
                 ZSTD_CCtx_params params)
 {
     ZSTD_CCtx_params const jobParams = ZSTDMT_initJobCCtxParams(params);
-    size_t const overlapSize = (size_t)1 << ZSTDMT_computeOverlapLog(params);
+    size_t const overlapSize = ZSTDMT_computeOverlapSize(params);
     unsigned const nbJobs = ZSTDMT_computeNbJobs(params, srcSize, params.nbWorkers);
     size_t const proposedJobSize = (srcSize + (nbJobs-1)) / nbJobs;
     size_t const avgJobSize = (((proposedJobSize-1) & 0x1FFFF) < 0x7FFF) ? proposedJobSize + 0xFFFF : proposedJobSize;   /* avoid too small last block */
@@ -1289,16 +1347,17 @@ static size_t ZSTDMT_compress_advanced_internal(
 }
 
 size_t ZSTDMT_compress_advanced(ZSTDMT_CCtx* mtctx,
-                               void* dst, size_t dstCapacity,
+                                void* dst, size_t dstCapacity,
-                         const void* src, size_t srcSize,
+                          const void* src, size_t srcSize,
-                         const ZSTD_CDict* cdict,
+                          const ZSTD_CDict* cdict,
-                               ZSTD_parameters params,
+                                ZSTD_parameters params,
-                               unsigned overlapLog)
+                                int overlapLog)
 {
     ZSTD_CCtx_params cctxParams = mtctx->params;
     cctxParams.cParams = params.cParams;
     cctxParams.fParams = params.fParams;
-    cctxParams.overlapSizeLog = overlapLog;
+    assert(ZSTD_OVERLAPLOG_MIN <= overlapLog && overlapLog <= ZSTD_OVERLAPLOG_MAX);
+    cctxParams.overlapLog = overlapLog;
     return ZSTDMT_compress_advanced_internal(mtctx,
                                              dst, dstCapacity,
                                              src, srcSize,
@@ -1311,8 +1370,8 @@ size_t ZSTDMT_compressCCtx(ZSTDMT_CCtx* mtctx,
                      const void* src, size_t srcSize,
                            int compressionLevel)
 {
-    U32 const overlapLog = (compressionLevel >= ZSTD_maxCLevel()) ? 9 : ZSTDMT_OVERLAPLOG_DEFAULT;
     ZSTD_parameters params = ZSTD_getParams(compressionLevel, srcSize, 0);
+    int const overlapLog = ZSTDMT_overlapLog_default(params.cParams.strategy);
     params.fParams.contentSizeFlag = 1;
     return ZSTDMT_compress_advanced(mtctx, dst, dstCapacity, src, srcSize, NULL, params, overlapLog);
 }
@@ -1339,8 +1398,8 @@ size_t ZSTDMT_initCStream_internal(
     if (params.nbWorkers != mtctx->params.nbWorkers)
         CHECK_F( ZSTDMT_resize(mtctx, params.nbWorkers) );
 
-    if (params.jobSize > 0 && params.jobSize < ZSTDMT_JOBSIZE_MIN) params.jobSize = ZSTDMT_JOBSIZE_MIN;
+    if (params.jobSize != 0 && params.jobSize < ZSTDMT_JOBSIZE_MIN) params.jobSize = ZSTDMT_JOBSIZE_MIN;
-    if (params.jobSize > ZSTDMT_JOBSIZE_MAX) params.jobSize = ZSTDMT_JOBSIZE_MAX;
+    if (params.jobSize > (size_t)ZSTDMT_JOBSIZE_MAX) params.jobSize = ZSTDMT_JOBSIZE_MAX;
 
     mtctx->singleBlockingThread = (pledgedSrcSize <= ZSTDMT_JOBSIZE_MIN);  /* do not trigger multi-threading when srcSize is too small */
     if (mtctx->singleBlockingThread) {
@@ -1375,14 +1434,24 @@ size_t ZSTDMT_initCStream_internal(
         mtctx->cdict = cdict;
     }
 
-    mtctx->targetPrefixSize = (size_t)1 << ZSTDMT_computeOverlapLog(params);
+    mtctx->targetPrefixSize = ZSTDMT_computeOverlapSize(params);
-    DEBUGLOG(4, "overlapLog=%u => %u KB", params.overlapSizeLog, (U32)(mtctx->targetPrefixSize>>10));
+    DEBUGLOG(4, "overlapLog=%i => %u KB", params.overlapLog, (U32)(mtctx->targetPrefixSize>>10));
     mtctx->targetSectionSize = params.jobSize;
     if (mtctx->targetSectionSize == 0) {
         mtctx->targetSectionSize = 1ULL << ZSTDMT_computeTargetJobLog(params);
     }
+    if (params.rsyncable) {
+        /* Aim for the targetsectionSize as the average job size. */
+        U32 const jobSizeMB = (U32)(mtctx->targetSectionSize >> 20);
+        U32 const rsyncBits = ZSTD_highbit32(jobSizeMB) + 20;
+        assert(jobSizeMB >= 1);
+        DEBUGLOG(4, "rsyncLog = %u", rsyncBits);
+        mtctx->rsync.hash = 0;
+        mtctx->rsync.hitMask = (1ULL << rsyncBits) - 1;
+        mtctx->rsync.primePower = ZSTD_rollingHash_primePower(RSYNC_LENGTH);
+    }
     if (mtctx->targetSectionSize < mtctx->targetPrefixSize) mtctx->targetSectionSize = mtctx->targetPrefixSize;  /* job size must be >= overlap size */
-    DEBUGLOG(4, "Job Size : %u KB (note : set to %u)", (U32)(mtctx->targetSectionSize>>10), params.jobSize);
+    DEBUGLOG(4, "Job Size : %u KB (note : set to %u)", (U32)(mtctx->targetSectionSize>>10), (U32)params.jobSize);
     DEBUGLOG(4, "inBuff Size : %u KB", (U32)(mtctx->targetSectionSize>>10));
     ZSTDMT_setBufferSize(mtctx->bufPool, ZSTD_compressBound(mtctx->targetSectionSize));
     {
@@ -1818,6 +1887,89 @@ static int ZSTDMT_tryGetInputRange(ZSTDMT_CCtx* mtctx)
     return 1;
 }
 
+typedef struct {
+  size_t toLoad;  /* The number of bytes to load from the input. */
+  int flush;      /* Boolean declaring if we must flush because we found a synchronization point. */
+} syncPoint_t;
+
+/**
+ * Searches through the input for a synchronization point. If one is found, we
+ * will instruct the caller to flush, and return the number of bytes to load.
+ * Otherwise, we will load as many bytes as possible and instruct the caller
+ * to continue as normal.
+ */
+static syncPoint_t
+findSynchronizationPoint(ZSTDMT_CCtx const* mtctx, ZSTD_inBuffer const input)
+{
+    BYTE const* const istart = (BYTE const*)input.src + input.pos;
+    U64 const primePower = mtctx->rsync.primePower;
+    U64 const hitMask = mtctx->rsync.hitMask;
+
+    syncPoint_t syncPoint;
+    U64 hash;
+    BYTE const* prev;
+    size_t pos;
+
+    syncPoint.toLoad = MIN(input.size - input.pos, mtctx->targetSectionSize - mtctx->inBuff.filled);
+    syncPoint.flush = 0;
+    if (!mtctx->params.rsyncable)
+        /* Rsync is disabled. */
+        return syncPoint;
+    if (mtctx->inBuff.filled + syncPoint.toLoad < RSYNC_LENGTH)
+        /* Not enough to compute the hash.
+         * We will miss any synchronization points in this RSYNC_LENGTH byte
+         * window. However, since it depends only in the internal buffers, if the
+         * state is already synchronized, we will remain synchronized.
+         * Additionally, the probability that we miss a synchronization point is
+         * low: RSYNC_LENGTH / targetSectionSize.
+         */
+        return syncPoint;
+    /* Initialize the loop variables. */
+    if (mtctx->inBuff.filled >= RSYNC_LENGTH) {
+        /* We have enough bytes buffered to initialize the hash.
+         * Start scanning at the beginning of the input.
+         */
+        pos = 0;
+        prev = (BYTE const*)mtctx->inBuff.buffer.start + mtctx->inBuff.filled - RSYNC_LENGTH;
+        hash = ZSTD_rollingHash_compute(prev, RSYNC_LENGTH);
+    } else {
+        /* We don't have enough bytes buffered to initialize the hash, but
+         * we know we have at least RSYNC_LENGTH bytes total.
+         * Start scanning after the first RSYNC_LENGTH bytes less the bytes
+         * already buffered.
+         */
+        pos = RSYNC_LENGTH - mtctx->inBuff.filled;
+        prev = (BYTE const*)mtctx->inBuff.buffer.start - pos;
+        hash = ZSTD_rollingHash_compute(mtctx->inBuff.buffer.start, mtctx->inBuff.filled);
+        hash = ZSTD_rollingHash_append(hash, istart, pos);
+    }
+    /* Starting with the hash of the previous RSYNC_LENGTH bytes, roll
+     * through the input. If we hit a synchronization point, then cut the
+     * job off, and tell the compressor to flush the job. Otherwise, load
+     * all the bytes and continue as normal.
+     * If we go too long without a synchronization point (targetSectionSize)
+     * then a block will be emitted anyways, but this is okay, since if we
+     * are already synchronized we will remain synchronized.
+     */
+    for (; pos < syncPoint.toLoad; ++pos) {
+        BYTE const toRemove = pos < RSYNC_LENGTH ? prev[pos] : istart[pos - RSYNC_LENGTH];
+        /* if (pos >= RSYNC_LENGTH) assert(ZSTD_rollingHash_compute(istart + pos - RSYNC_LENGTH, RSYNC_LENGTH) == hash); */
+        hash = ZSTD_rollingHash_rotate(hash, toRemove, istart[pos], primePower);
+        if ((hash & hitMask) == hitMask) {
+            syncPoint.toLoad = pos + 1;
+            syncPoint.flush = 1;
+            break;
+        }
+    }
+    return syncPoint;
+}
+
+size_t ZSTDMT_nextInputSizeHint(const ZSTDMT_CCtx* mtctx)
+{
+    size_t hintInSize = mtctx->targetSectionSize - mtctx->inBuff.filled;
+    if (hintInSize==0) hintInSize = mtctx->targetSectionSize;
+    return hintInSize;
+}
 
 /** ZSTDMT_compressStream_generic() :
  *  internal use only - exposed to be invoked from zstd_compress.c
@@ -1844,7 +1996,8 @@ size_t ZSTDMT_compressStream_generic(ZSTDMT_CCtx* mtctx,
     }
 
     /* single-pass shortcut (note : synchronous-mode) */
-    if ( (mtctx->nextJobID == 0)      /* just started */
+    if ( (!mtctx->params.rsyncable)   /* rsyncable mode is disabled */
+      && (mtctx->nextJobID == 0)      /* just started */
       && (mtctx->inBuff.filled == 0)  /* nothing buffered */
       && (!mtctx->jobReady)           /* no job already created */
       && (endOp == ZSTD_e_end)        /* end order */
@@ -1876,14 +2029,17 @@ size_t ZSTDMT_compressStream_generic(ZSTDMT_CCtx* mtctx,
                 DEBUGLOG(5, "ZSTDMT_tryGetInputRange completed successfully : mtctx->inBuff.buffer.start = %p", mtctx->inBuff.buffer.start);
         }
         if (mtctx->inBuff.buffer.start != NULL) {
-            size_t const toLoad = MIN(input->size - input->pos, mtctx->targetSectionSize - mtctx->inBuff.filled);
+            syncPoint_t const syncPoint = findSynchronizationPoint(mtctx, *input);
+            if (syncPoint.flush && endOp == ZSTD_e_continue) {
+                endOp = ZSTD_e_flush;
+            }
             assert(mtctx->inBuff.buffer.capacity >= mtctx->targetSectionSize);
             DEBUGLOG(5, "ZSTDMT_compressStream_generic: adding %u bytes on top of %u to buffer of size %u",
-                        (U32)toLoad, (U32)mtctx->inBuff.filled, (U32)mtctx->targetSectionSize);
+                        (U32)syncPoint.toLoad, (U32)mtctx->inBuff.filled, (U32)mtctx->targetSectionSize);
-            memcpy((char*)mtctx->inBuff.buffer.start + mtctx->inBuff.filled, (const char*)input->src + input->pos, toLoad);
+            memcpy((char*)mtctx->inBuff.buffer.start + mtctx->inBuff.filled, (const char*)input->src + input->pos, syncPoint.toLoad);
-            input->pos += toLoad;
+            input->pos += syncPoint.toLoad;
-            mtctx->inBuff.filled += toLoad;
+            mtctx->inBuff.filled += syncPoint.toLoad;
-            forwardInputProgress = toLoad>0;
+            forwardInputProgress = syncPoint.toLoad>0;
         }
         if ((input->pos < input->size) && (endOp == ZSTD_e_end))
             endOp = ZSTD_e_flush;   /* can't end now : not all input consumed */

C/zstd/zstdmt_compress.h

@@ -28,6 +28,16 @@
 #include "zstd.h"            /* ZSTD_inBuffer, ZSTD_outBuffer, ZSTDLIB_API */
 
 
+/* ===   Constants   === */
+#ifndef ZSTDMT_NBWORKERS_MAX
+#  define ZSTDMT_NBWORKERS_MAX 200
+#endif
+#ifndef ZSTDMT_JOBSIZE_MIN
+#  define ZSTDMT_JOBSIZE_MIN (1 MB)
+#endif
+#define ZSTDMT_JOBSIZE_MAX  (MEM_32bits() ? (512 MB) : (1024 MB))
+
+
 /* ===   Memory management   === */
 typedef struct ZSTDMT_CCtx_s ZSTDMT_CCtx;
 ZSTDLIB_API ZSTDMT_CCtx* ZSTDMT_createCCtx(unsigned nbWorkers);
@@ -52,6 +62,7 @@ ZSTDLIB_API size_t ZSTDMT_compressCCtx(ZSTDMT_CCtx* mtctx,
 ZSTDLIB_API size_t ZSTDMT_initCStream(ZSTDMT_CCtx* mtctx, int compressionLevel);
 ZSTDLIB_API size_t ZSTDMT_resetCStream(ZSTDMT_CCtx* mtctx, unsigned long long pledgedSrcSize);  /**< if srcSize is not known at reset time, use ZSTD_CONTENTSIZE_UNKNOWN. Note: for compatibility with older programs, 0 means the same as ZSTD_CONTENTSIZE_UNKNOWN, but it will change in the future to mean "empty" */
 
+ZSTDLIB_API size_t ZSTDMT_nextInputSizeHint(const ZSTDMT_CCtx* mtctx);
 ZSTDLIB_API size_t ZSTDMT_compressStream(ZSTDMT_CCtx* mtctx, ZSTD_outBuffer* output, ZSTD_inBuffer* input);
 
 ZSTDLIB_API size_t ZSTDMT_flushStream(ZSTDMT_CCtx* mtctx, ZSTD_outBuffer* output);   /**< @return : 0 == all flushed; >0 : still some data to be flushed; or an error code (ZSTD_isError()) */
@@ -60,16 +71,12 @@ ZSTDLIB_API size_t ZSTDMT_endStream(ZSTDMT_CCtx* mtctx, ZSTD_outBuffer* output);
 
 /* ===   Advanced functions and parameters  === */
 
-#ifndef ZSTDMT_JOBSIZE_MIN
-#  define ZSTDMT_JOBSIZE_MIN (1U << 20)   /* 1 MB - Minimum size of each compression job */
-#endif
-
 ZSTDLIB_API size_t ZSTDMT_compress_advanced(ZSTDMT_CCtx* mtctx,
                                            void* dst, size_t dstCapacity,
                                      const void* src, size_t srcSize,
                                      const ZSTD_CDict* cdict,
                                            ZSTD_parameters params,
-                                           unsigned overlapLog);
+                                           int overlapLog);
 
 ZSTDLIB_API size_t ZSTDMT_initCStream_advanced(ZSTDMT_CCtx* mtctx,
                                         const void* dict, size_t dictSize,   /* dict can be released after init, a local copy is preserved within zcs */
@@ -84,8 +91,9 @@ ZSTDLIB_API size_t ZSTDMT_initCStream_usingCDict(ZSTDMT_CCtx* mtctx,
 /* ZSTDMT_parameter :
  * List of parameters that can be set using ZSTDMT_setMTCtxParameter() */
 typedef enum {
-    ZSTDMT_p_jobSize,           /* Each job is compressed in parallel. By default, this value is dynamically determined depending on compression parameters. Can be set explicitly here. */
+    ZSTDMT_p_jobSize,     /* Each job is compressed in parallel. By default, this value is dynamically determined depending on compression parameters. Can be set explicitly here. */
-    ZSTDMT_p_overlapSectionLog  /* Each job may reload a part of previous job to enhance compressionr ratio; 0 == no overlap, 6(default) == use 1/8th of window, >=9 == use full window. This is a "sticky" parameter : its value will be re-used on next compression job */
+    ZSTDMT_p_overlapLog,  /* Each job may reload a part of previous job to enhance compressionr ratio; 0 == no overlap, 6(default) == use 1/8th of window, >=9 == use full window. This is a "sticky" parameter : its value will be re-used on next compression job */
+    ZSTDMT_p_rsyncable    /* Enables rsyncable mode. */
 } ZSTDMT_parameter;
 
 /* ZSTDMT_setMTCtxParameter() :
@@ -93,12 +101,12 @@ typedef enum {
  * The function must be called typically after ZSTD_createCCtx() but __before ZSTDMT_init*() !__
  * Parameters not explicitly reset by ZSTDMT_init*() remain the same in consecutive compression sessions.
  * @return : 0, or an error code (which can be tested using ZSTD_isError()) */
-ZSTDLIB_API size_t ZSTDMT_setMTCtxParameter(ZSTDMT_CCtx* mtctx, ZSTDMT_parameter parameter, unsigned value);
+ZSTDLIB_API size_t ZSTDMT_setMTCtxParameter(ZSTDMT_CCtx* mtctx, ZSTDMT_parameter parameter, int value);
 
 /* ZSTDMT_getMTCtxParameter() :
  * Query the ZSTDMT_CCtx for a parameter value.
  * @return : 0, or an error code (which can be tested using ZSTD_isError()) */
-ZSTDLIB_API size_t ZSTDMT_getMTCtxParameter(ZSTDMT_CCtx* mtctx, ZSTDMT_parameter parameter, unsigned* value);
+ZSTDLIB_API size_t ZSTDMT_getMTCtxParameter(ZSTDMT_CCtx* mtctx, ZSTDMT_parameter parameter, int* value);
 
 
 /*! ZSTDMT_compressStream_generic() :
@@ -129,7 +137,7 @@ size_t ZSTDMT_toFlushNow(ZSTDMT_CCtx* mtctx);
 
 /*! ZSTDMT_CCtxParam_setMTCtxParameter()
  *  like ZSTDMT_setMTCtxParameter(), but into a ZSTD_CCtx_Params */
-size_t ZSTDMT_CCtxParam_setMTCtxParameter(ZSTD_CCtx_params* params, ZSTDMT_parameter parameter, unsigned value);
+size_t ZSTDMT_CCtxParam_setMTCtxParameter(ZSTD_CCtx_params* params, ZSTDMT_parameter parameter, int value);
 
 /*! ZSTDMT_CCtxParam_setNbWorkers()
  *  Set nbWorkers, and clamp it.

CPP/7zip/Bundles/Alone/makefile

@@ -291,6 +291,8 @@ ZSTD_OBJS = \
   $O\xxhash.obj \
   $O\zstd_common.obj \
   $O\zstd_compress.obj \
+  $O\zstd_ddict.obj \
+  $O\zstd_decompress_block.obj \
   $O\zstd_decompress.obj \
   $O\zstd_double_fast.obj \
   $O\zstd_fast.obj \

CPP/7zip/Bundles/Codec_zstd/makefile

@@ -35,6 +35,8 @@ ZSTD_OBJS = \
   $O\xxhash.obj \
   $O\zstd_common.obj \
   $O\zstd_compress.obj \
+  $O\zstd_ddict.obj \
+  $O\zstd_decompress_block.obj \
   $O\zstd_decompress.obj \
   $O\zstd_double_fast.obj \
   $O\zstd_fast.obj \

CPP/7zip/Bundles/Format7z/makefile

@@ -213,6 +213,8 @@ ZSTD_OBJS = \
   $O\xxhash.obj \
   $O\zstd_common.obj \
   $O\zstd_compress.obj \
+  $O\zstd_ddict.obj \
+  $O\zstd_decompress_block.obj \
   $O\zstd_decompress.obj \
   $O\zstd_double_fast.obj \
   $O\zstd_fast.obj \

CPP/7zip/Bundles/Format7zExtract/makefile

@@ -173,6 +173,8 @@ ZSTD_OBJS = \
   $O\xxhash.obj \
   $O\zstd_common.obj \
   $O\zstd_compress.obj \
+  $O\zstd_ddict.obj \
+  $O\zstd_decompress_block.obj \
   $O\zstd_decompress.obj \
   $O\zstd_double_fast.obj \
   $O\zstd_fast.obj \

CPP/7zip/Bundles/Format7zF/makefile

@@ -88,6 +88,8 @@ ZSTD_OBJS = \
   $O\xxhash.obj \
   $O\zstd_common.obj \
   $O\zstd_compress.obj \
+  $O\zstd_ddict.obj \
+  $O\zstd_decompress_block.obj \
   $O\zstd_decompress.obj \
   $O\zstd_double_fast.obj \
   $O\zstd_fast.obj \

CPP/7zip/Bundles/SFXCon/makefile

@@ -151,6 +151,8 @@ ZSTD_OBJS = \
   $O\xxhash.obj \
   $O\zstd_common.obj \
   $O\zstd_compress.obj \
+  $O\zstd_ddict.obj \
+  $O\zstd_decompress_block.obj \
   $O\zstd_decompress.obj \
   $O\zstd_double_fast.obj \
   $O\zstd_fast.obj \

CPP/7zip/Bundles/SFXWin/makefile

@@ -169,6 +169,8 @@ ZSTD_OBJS = \
   $O\xxhash.obj \
   $O\zstd_common.obj \
   $O\zstd_compress.obj \
+  $O\zstd_ddict.obj \
+  $O\zstd_decompress_block.obj \
   $O\zstd_decompress.obj \
   $O\zstd_double_fast.obj \
   $O\zstd_fast.obj \

CPP/7zip/Compress/ZstdEncoder.cpp

@@ -23,20 +23,20 @@ CEncoder::CEncoder():
   _processedIn(0),
   _processedOut(0),
   _numThreads(NWindows::NSystem::GetNumberOfProcessors()),
-  _Strategy(-1),
   _Long(-1),
   _Level(ZSTD_CLEVEL_DEFAULT),
+  _Strategy(-1),
   _WindowLog(-1),
   _HashLog(-1),
   _ChainLog(-1),
   _SearchLog(-1),
-  _SearchLength(-1),
+  _MinMatch(-1),
   _TargetLen(-1),
   _OverlapLog(-1),
   _LdmHashLog(-1),
-  _LdmSearchLength(-1),
+  _LdmMinMatch(-1),
   _LdmBucketSizeLog(-1),
-  _LdmHashEveryLog(-1)
+  _LdmHashRateLog(-1)
 {
   _props.clear();
   _hMutex = CreateMutex(NULL, FALSE, NULL);
@@ -153,11 +153,11 @@ STDMETHODIMP CEncoder::SetCoderProperties(const PROPID * propIDs, const PROPVARI
         _SearchLog = v;
         break;
       }
-    case NCoderPropID::kSearchLength:
+    case NCoderPropID::kMinMatch:
       {
-        if (v < ZSTD_SEARCHLENGTH_MIN) v = ZSTD_SEARCHLENGTH_MIN;
+        if (v < ZSTD_MINMATCH_MIN) v = ZSTD_MINMATCH_MIN;
-        if (v > ZSTD_SEARCHLENGTH_MAX) v = ZSTD_SEARCHLENGTH_MAX;
+        if (v > ZSTD_MINMATCH_MAX) v = ZSTD_MINMATCH_MAX;
-        _SearchLength = v;
+        _MinMatch = v;
         break;
       }
     case NCoderPropID::kTargetLen:
@@ -185,7 +185,7 @@ STDMETHODIMP CEncoder::SetCoderProperties(const PROPID * propIDs, const PROPVARI
       {
         if (v < ZSTD_LDM_MINMATCH_MIN) v = ZSTD_LDM_MINMATCH_MIN;
         if (v > ZSTD_LDM_MINMATCH_MAX) v = ZSTD_LDM_MINMATCH_MAX;
-        _LdmSearchLength = v;
+        _LdmMinMatch = v;
         break;
       }
     case NCoderPropID::kLdmBucketSizeLog:
@@ -195,11 +195,11 @@ STDMETHODIMP CEncoder::SetCoderProperties(const PROPID * propIDs, const PROPVARI
         _LdmBucketSizeLog = v;
         break;
       }
-    case NCoderPropID::kLdmHashEveryLog:
+    case NCoderPropID::kLdmHashRateLog:
       {
         if (v < 0) v = 0; /* 0 => automatic mode */
         if (v > (ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN)) v = (ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN);
-        _LdmHashEveryLog = v;
+        _LdmHashRateLog = v;
         break;
       }
     default:
@@ -240,15 +240,15 @@ STDMETHODIMP CEncoder::Code(ISequentialInStream *inStream,
       return E_OUTOFMEMORY;
 
     /* setup level */
-    err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_compressionLevel, (UInt32)_Level);
+    err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_compressionLevel, (UInt32)_Level);
     if (ZSTD_isError(err)) return E_INVALIDARG;
 
     /* setup thread count */
-    err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_nbWorkers, _numThreads);
+    err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_nbWorkers, _numThreads);
     if (ZSTD_isError(err)) return E_INVALIDARG;
 
     /* set the content size flag */
-    err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_contentSizeFlag, 1);
+    err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_contentSizeFlag, 1);
     if (ZSTD_isError(err)) return E_INVALIDARG;
 
     /* enable ldm for large windowlog values */
@@ -257,67 +257,67 @@ STDMETHODIMP CEncoder::Code(ISequentialInStream *inStream,
 
     /* set ldm */
     if (_Long != -1) {
-      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_enableLongDistanceMatching, _Long);
+      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_enableLongDistanceMatching, _Long);
       if (ZSTD_isError(err)) return E_INVALIDARG;
     }
 
     if (_Strategy != -1) {
-      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_compressionStrategy, _Strategy);
+      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_strategy, _Strategy);
       if (ZSTD_isError(err)) return E_INVALIDARG;
     }
 
     if (_WindowLog != -1) {
-      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_windowLog, _WindowLog);
+      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_windowLog, _WindowLog);
       if (ZSTD_isError(err)) return E_INVALIDARG;
     }
 
     if (_HashLog != -1) {
-      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_hashLog, _HashLog);
+      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_hashLog, _HashLog);
       if (ZSTD_isError(err)) return E_INVALIDARG;
     }
 
     if (_ChainLog != -1) {
-      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_chainLog, _ChainLog);
+      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_chainLog, _ChainLog);
       if (ZSTD_isError(err)) return E_INVALIDARG;
     }
 
     if (_SearchLog != -1) {
-      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_searchLog, _SearchLog);
+      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_searchLog, _SearchLog);
       if (ZSTD_isError(err)) return E_INVALIDARG;
     }
 
-    if (_SearchLength != -1) {
+    if (_MinMatch != -1) {
-      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_minMatch, _SearchLength);
+      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_minMatch, _MinMatch);
       if (ZSTD_isError(err)) return E_INVALIDARG;
     }
 
     if (_TargetLen != -1) {
-      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_targetLength, _TargetLen);
+      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_targetLength, _TargetLen);
       if (ZSTD_isError(err)) return E_INVALIDARG;
     }
 
     if (_OverlapLog != -1) {
-      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_overlapSizeLog, _OverlapLog);
+      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_overlapLog, _OverlapLog);
       if (ZSTD_isError(err)) return E_INVALIDARG;
     }
 
     if (_LdmHashLog != -1) {
-      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_ldmHashLog, _LdmHashLog);
+      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_ldmHashLog, _LdmHashLog);
       if (ZSTD_isError(err)) return E_INVALIDARG;
     }
 
-    if (_LdmSearchLength != -1) {
+    if (_LdmMinMatch != -1) {
-      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_ldmMinMatch, _LdmSearchLength);
+      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_ldmMinMatch, _LdmMinMatch);
       if (ZSTD_isError(err)) return E_INVALIDARG;
     }
 
     if (_LdmBucketSizeLog != -1) {
-      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_ldmBucketSizeLog, _LdmBucketSizeLog);
+      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_ldmBucketSizeLog, _LdmBucketSizeLog);
       if (ZSTD_isError(err)) return E_INVALIDARG;
     }
 
-    if (_LdmHashEveryLog != -1) {
+    if (_LdmHashRateLog != -1) {
-      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_p_ldmHashEveryLog, _LdmHashEveryLog);
+      err = ZSTD_CCtx_setParameter(_ctx, ZSTD_c_ldmHashRateLog, _LdmHashRateLog);
       if (ZSTD_isError(err)) return E_INVALIDARG;
     }
   }
@@ -352,7 +352,7 @@ STDMETHODIMP CEncoder::Code(ISequentialInStream *inStream,
         inBuff.pos = 0;
       }
 
-      err = ZSTD_compress_generic(_ctx, &outBuff, &inBuff, ZSTD_todo);
+      err = ZSTD_compressStream2(_ctx, &outBuff, &inBuff, ZSTD_todo);
       if (ZSTD_isError(err)) return E_FAIL;
 
 #if DEBUG

CPP/7zip/Compress/ZstdEncoder.h

@@ -52,20 +52,20 @@ class CEncoder:
   HANDLE _hMutex;
 
   /* zstd advanced compression options */
-  Int32 _Strategy;
   Int32 _Long;
   Int32 _Level;
+  Int32 _Strategy;
   Int32 _WindowLog;
   Int32 _HashLog;
   Int32 _ChainLog;
   Int32 _SearchLog;
-  Int32 _SearchLength;
+  Int32 _MinMatch;
   Int32 _TargetLen;
   Int32 _OverlapLog;
   Int32 _LdmHashLog;
-  Int32 _LdmSearchLength;
+  Int32 _LdmMinMatch;
   Int32 _LdmBucketSizeLog;
-  Int32 _LdmHashEveryLog;
+  Int32 _LdmHashRateLog;
 
 public:
   MY_QUERYINTERFACE_BEGIN2(ICompressCoder)

CPP/7zip/ICoder.h

@@ -141,13 +141,13 @@ namespace NCoderPropID
     kHashLog,           // VT_UI4 The minimum hlog is 6 (64 B) and the maximum is 26 (128 MiB).
     kChainLog,          // VT_UI4 The minimum clog is 6 (64 B) and the maximum is 28 (256 MiB)
     kSearchLog,         // VT_UI4 The minimum slog is 1 and the maximum is 26
-    kSearchLength,      // VT_UI4 The minimum slen is 3 and the maximum is 7.
+    kMinMatch,          // VT_UI4 The minimum slen is 3 and the maximum is 7.
     kTargetLen,         // VT_UI4 The minimum tlen is 0 and the maximum is 999.
     kOverlapLog,        // VT_UI4 The minimum ovlog is 0 and the maximum is 9.  (default: 6)
     kLdmHashLog,        // VT_UI4 The minimum ldmhlog is 6 and the maximum is 26 (default: 20).
     kLdmSearchLength,   // VT_UI4 The minimum ldmslen is 4 and the maximum is 4096 (default: 64).
     kLdmBucketSizeLog,  // VT_UI4 The minimum ldmblog is 0 and the maximum is 8 (default: 3).
-    kLdmHashEveryLog    // VT_UI4 The default value is wlog - ldmhlog.
+    kLdmHashRateLog    // VT_UI4 The default value is wlog - ldmhlog.
   };
 }