zstd-ruby 1.3.3.0 → 1.3.4.0

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

@@ -30,15 +30,15 @@
 
 /* ===   Memory management   === */
 typedef struct ZSTDMT_CCtx_s ZSTDMT_CCtx;
-ZSTDLIB_API ZSTDMT_CCtx* ZSTDMT_createCCtx(unsigned nbThreads);
-ZSTDLIB_API ZSTDMT_CCtx* ZSTDMT_createCCtx_advanced(unsigned nbThreads,
+ZSTDLIB_API ZSTDMT_CCtx* ZSTDMT_createCCtx(unsigned nbWorkers);
+ZSTDLIB_API ZSTDMT_CCtx* ZSTDMT_createCCtx_advanced(unsigned nbWorkers,
                                                     ZSTD_customMem cMem);
 ZSTDLIB_API size_t ZSTDMT_freeCCtx(ZSTDMT_CCtx* mtctx);
 
 ZSTDLIB_API size_t ZSTDMT_sizeof_CCtx(ZSTDMT_CCtx* mtctx);
 
 
-/* ===   Simple buffer-to-butter one-pass function   === */
+/* ===   Simple one-pass compression function   === */
 
 ZSTDLIB_API size_t ZSTDMT_compressCCtx(ZSTDMT_CCtx* mtctx,
                                        void* dst, size_t dstCapacity,
@@ -50,7 +50,7 @@ ZSTDLIB_API size_t ZSTDMT_compressCCtx(ZSTDMT_CCtx* mtctx,
 /* ===   Streaming functions   === */
 
 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 may change in the future, to mean "empty" */
+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_compressStream(ZSTDMT_CCtx* mtctx, ZSTD_outBuffer* output, ZSTD_inBuffer* input);
 
@@ -68,7 +68,7 @@ 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 const params,
+                                           ZSTD_parameters params,
                                            unsigned overlapLog);
 
 ZSTDLIB_API size_t ZSTDMT_initCStream_advanced(ZSTDMT_CCtx* mtctx,
@@ -85,7 +85,7 @@ ZSTDLIB_API size_t ZSTDMT_initCStream_usingCDict(ZSTDMT_CCtx* mtctx,
  * 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_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 */
+    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_parameter;
 
 /* ZSTDMT_setMTCtxParameter() :
@@ -97,30 +97,46 @@ ZSTDLIB_API size_t ZSTDMT_setMTCtxParameter(ZSTDMT_CCtx* mtctx, ZSTDMT_parameter
 
 
 /*! ZSTDMT_compressStream_generic() :
- *  Combines ZSTDMT_compressStream() with ZSTDMT_flushStream() or ZSTDMT_endStream()
+ *  Combines ZSTDMT_compressStream() with optional ZSTDMT_flushStream() or ZSTDMT_endStream()
  *  depending on flush directive.
  * @return : minimum amount of data still to be flushed
  *           0 if fully flushed
- *           or an error code */
+ *           or an error code
+ *  note : needs to be init using any ZSTD_initCStream*() variant */
 ZSTDLIB_API size_t ZSTDMT_compressStream_generic(ZSTDMT_CCtx* mtctx,
                                                 ZSTD_outBuffer* output,
                                                 ZSTD_inBuffer* input,
                                                 ZSTD_EndDirective endOp);
 
 
-/* ===   Private definitions; never ever use directly  === */
+/* ========================================================
+ * ===  Private interface, for use by ZSTD_compress.c   ===
+ * ===  Not exposed in libzstd. Never invoke directly   ===
+ * ======================================================== */
 
 size_t ZSTDMT_CCtxParam_setMTCtxParameter(ZSTD_CCtx_params* params, ZSTDMT_parameter parameter, unsigned value);
 
-/* ZSTDMT_CCtxParam_setNbThreads()
- * Set nbThreads, and clamp it correctly,
- * also reset jobSize and overlapLog */
-size_t ZSTDMT_CCtxParam_setNbThreads(ZSTD_CCtx_params* params, unsigned nbThreads);
+/* ZSTDMT_CCtxParam_setNbWorkers()
+ * Set nbWorkers, and clamp it.
+ * Also reset jobSize and overlapLog */
+size_t ZSTDMT_CCtxParam_setNbWorkers(ZSTD_CCtx_params* params, unsigned nbWorkers);
 
-/* ZSTDMT_getNbThreads():
+/*! ZSTDMT_updateCParams_whileCompressing() :
+ *  Updates only a selected set of compression parameters, to remain compatible with current frame.
+ *  New parameters will be applied to next compression job. */
+void ZSTDMT_updateCParams_whileCompressing(ZSTDMT_CCtx* mtctx, const ZSTD_CCtx_params* cctxParams);
+
+/* ZSTDMT_getNbWorkers():
  * @return nb threads currently active in mtctx.
  * mtctx must be valid */
-size_t ZSTDMT_getNbThreads(const ZSTDMT_CCtx* mtctx);
+unsigned ZSTDMT_getNbWorkers(const ZSTDMT_CCtx* mtctx);
+
+/* ZSTDMT_getFrameProgression():
+ * tells how much data has been consumed (input) and produced (output) for current frame.
+ * able to count progression inside worker threads.
+ */
+ZSTD_frameProgression ZSTDMT_getFrameProgression(ZSTDMT_CCtx* mtctx);
+
 
 /*! ZSTDMT_initCStream_internal() :
  *  Private use only. Init streaming operation.
@@ -128,7 +144,7 @@ size_t ZSTDMT_getNbThreads(const ZSTDMT_CCtx* mtctx);
  *  must receive dict, or cdict, or none, but not both.
  *  @return : 0, or an error code */
 size_t ZSTDMT_initCStream_internal(ZSTDMT_CCtx* zcs,
-                    const void* dict, size_t dictSize, ZSTD_dictMode_e dictMode,
+                    const void* dict, size_t dictSize, ZSTD_dictContentType_e dictContentType,
                     const ZSTD_CDict* cdict,
                     ZSTD_CCtx_params params, unsigned long long pledgedSrcSize);
 

data/ext/zstdruby/libzstd/decompress/huf_decompress.c

@@ -49,18 +49,19 @@
 ****************************************************************/
 #define HUF_isError ERR_isError
 #define HUF_STATIC_ASSERT(c) { enum { HUF_static_assert = 1/(int)(!!(c)) }; }   /* use only *after* variable declarations */
+#define CHECK_F(f) { size_t const err_ = (f); if (HUF_isError(err_)) return err_; }
 
 
 /* **************************************************************
 *  Byte alignment for workSpace management
 ****************************************************************/
-#define HUF_ALIGN(x, a) HUF_ALIGN_MASK((x), (a) - 1)
+#define HUF_ALIGN(x, a)         HUF_ALIGN_MASK((x), (a) - 1)
 #define HUF_ALIGN_MASK(x, mask) (((x) + (mask)) & ~(mask))
 
+
 /*-***************************/
 /*  generic DTableDesc       */
 /*-***************************/
-
 typedef struct { BYTE maxTableLog; BYTE tableType; BYTE tableLog; BYTE reserved; } DTableDesc;
 
 static DTableDesc HUF_getDTableDesc(const HUF_DTable* table)
@@ -74,7 +75,6 @@ static DTableDesc HUF_getDTableDesc(const HUF_DTable* table)
 /*-***************************/
 /*  single-symbol decoding   */
 /*-***************************/
-
 typedef struct { BYTE byte; BYTE nbBits; } HUF_DEltX2;   /* single-symbol decoding */
 
 size_t HUF_readDTableX2_wksp(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize)
@@ -94,10 +94,7 @@ size_t HUF_readDTableX2_wksp(HUF_DTable* DTable, const void* src, size_t srcSize
     huffWeight = (BYTE *)((U32 *)workSpace + spaceUsed32);
     spaceUsed32 += HUF_ALIGN(HUF_SYMBOLVALUE_MAX + 1, sizeof(U32)) >> 2;
 
-    if ((spaceUsed32 << 2) > wkspSize)
-        return ERROR(tableLog_tooLarge);
-    workSpace = (U32 *)workSpace + spaceUsed32;
-    wkspSize -= (spaceUsed32 << 2);
+    if ((spaceUsed32 << 2) > wkspSize) return ERROR(tableLog_tooLarge);
 
     HUF_STATIC_ASSERT(sizeof(DTableDesc) == sizeof(HUF_DTable));
     /* memset(huffWeight, 0, sizeof(huffWeight)); */   /* is not necessary, even though some analyzer complain ... */
@@ -144,8 +141,10 @@ size_t HUF_readDTableX2(HUF_DTable* DTable, const void* src, size_t srcSize)
                                  workSpace, sizeof(workSpace));
 }
 
+typedef struct { U16 sequence; BYTE nbBits; BYTE length; } HUF_DEltX4;  /* double-symbols decoding */
 
-static BYTE HUF_decodeSymbolX2(BIT_DStream_t* Dstream, const HUF_DEltX2* dt, const U32 dtLog)
+FORCE_INLINE_TEMPLATE BYTE
+HUF_decodeSymbolX2(BIT_DStream_t* Dstream, const HUF_DEltX2* dt, const U32 dtLog)
 {
     size_t const val = BIT_lookBitsFast(Dstream, dtLog); /* note : dtLog >= 1 */
     BYTE const c = dt[val].byte;
@@ -156,7 +155,7 @@ static BYTE HUF_decodeSymbolX2(BIT_DStream_t* Dstream, const HUF_DEltX2* dt, con
 #define HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) \
     *ptr++ = HUF_decodeSymbolX2(DStreamPtr, dt, dtLog)
 
-#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr) \
+#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr)  \
     if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
         HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr)
 
@@ -164,30 +163,33 @@ static BYTE HUF_decodeSymbolX2(BIT_DStream_t* Dstream, const HUF_DEltX2* dt, con
     if (MEM_64bits()) \
         HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr)
 
-HINT_INLINE size_t HUF_decodeStreamX2(BYTE* p, BIT_DStream_t* const bitDPtr, BYTE* const pEnd, const HUF_DEltX2* const dt, const U32 dtLog)
+HINT_INLINE size_t
+HUF_decodeStreamX2(BYTE* p, BIT_DStream_t* const bitDPtr, BYTE* const pEnd, const HUF_DEltX2* const dt, const U32 dtLog)
 {
     BYTE* const pStart = p;
 
     /* up to 4 symbols at a time */
-    while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p <= pEnd-4)) {
+    while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd-3)) {
         HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
         HUF_DECODE_SYMBOLX2_1(p, bitDPtr);
         HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
         HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
     }
 
-    /* closer to the end */
-    while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p < pEnd))
-        HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
+    /* [0-3] symbols remaining */
+    if (MEM_32bits())
+        while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd))
+            HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
 
-    /* no more data to retrieve from bitstream, hence no need to reload */
+    /* no more data to retrieve from bitstream, no need to reload */
     while (p < pEnd)
         HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
 
     return pEnd-pStart;
 }
 
-static size_t HUF_decompress1X2_usingDTable_internal(
+FORCE_INLINE_TEMPLATE size_t
+HUF_decompress1X2_usingDTable_internal_body(
           void* dst,  size_t dstSize,
     const void* cSrc, size_t cSrcSize,
     const HUF_DTable* DTable)
@@ -200,58 +202,17 @@ static size_t HUF_decompress1X2_usingDTable_internal(
     DTableDesc const dtd = HUF_getDTableDesc(DTable);
     U32 const dtLog = dtd.tableLog;
 
-    { size_t const errorCode = BIT_initDStream(&bitD, cSrc, cSrcSize);
-      if (HUF_isError(errorCode)) return errorCode; }
+    CHECK_F( BIT_initDStream(&bitD, cSrc, cSrcSize) );
 
     HUF_decodeStreamX2(op, &bitD, oend, dt, dtLog);
 
-    /* check */
     if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected);
 
     return dstSize;
 }
 
-size_t HUF_decompress1X2_usingDTable(
-          void* dst,  size_t dstSize,
-    const void* cSrc, size_t cSrcSize,
-    const HUF_DTable* DTable)
-{
-    DTableDesc dtd = HUF_getDTableDesc(DTable);
-    if (dtd.tableType != 0) return ERROR(GENERIC);
-    return HUF_decompress1X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
-}
-
-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)
-{
-    const BYTE* ip = (const BYTE*) cSrc;
-
-    size_t const hSize = HUF_readDTableX2_wksp(DCtx, cSrc, cSrcSize, workSpace, wkspSize);
-    if (HUF_isError(hSize)) return hSize;
-    if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
-    ip += hSize; cSrcSize -= hSize;
-
-    return HUF_decompress1X2_usingDTable_internal (dst, dstSize, ip, cSrcSize, DCtx);
-}
-
-
-size_t HUF_decompress1X2_DCtx(HUF_DTable* DCtx, void* dst, size_t dstSize,
-                              const void* cSrc, size_t cSrcSize)
-{
-    U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];
-    return HUF_decompress1X2_DCtx_wksp(DCtx, dst, dstSize, cSrc, cSrcSize,
-                                       workSpace, sizeof(workSpace));
-}
-
-size_t HUF_decompress1X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
-{
-    HUF_CREATE_STATIC_DTABLEX2(DTable, HUF_TABLELOG_MAX);
-    return HUF_decompress1X2_DCtx (DTable, dst, dstSize, cSrc, cSrcSize);
-}
-
-
-static size_t HUF_decompress4X2_usingDTable_internal(
+FORCE_INLINE_TEMPLATE size_t
+HUF_decompress4X2_usingDTable_internal_body(
           void* dst,  size_t dstSize,
     const void* cSrc, size_t cSrcSize,
     const HUF_DTable* DTable)
@@ -286,23 +247,19 @@ static size_t HUF_decompress4X2_usingDTable_internal(
         BYTE* op2 = opStart2;
         BYTE* op3 = opStart3;
         BYTE* op4 = opStart4;
-        U32 endSignal;
+        U32 endSignal = BIT_DStream_unfinished;
         DTableDesc const dtd = HUF_getDTableDesc(DTable);
         U32 const dtLog = dtd.tableLog;
 
         if (length4 > cSrcSize) return ERROR(corruption_detected);   /* overflow */
-        { size_t const errorCode = BIT_initDStream(&bitD1, istart1, length1);
-          if (HUF_isError(errorCode)) return errorCode; }
-        { size_t const errorCode = BIT_initDStream(&bitD2, istart2, length2);
-          if (HUF_isError(errorCode)) return errorCode; }
-        { size_t const errorCode = BIT_initDStream(&bitD3, istart3, length3);
-          if (HUF_isError(errorCode)) return errorCode; }
-        { size_t const errorCode = BIT_initDStream(&bitD4, istart4, length4);
-          if (HUF_isError(errorCode)) return errorCode; }
+        CHECK_F( BIT_initDStream(&bitD1, istart1, length1) );
+        CHECK_F( BIT_initDStream(&bitD2, istart2, length2) );
+        CHECK_F( BIT_initDStream(&bitD3, istart3, length3) );
+        CHECK_F( BIT_initDStream(&bitD4, istart4, length4) );
 
-        /* 16-32 symbols per loop (4-8 symbols per stream) */
+        /* up to 16 symbols per loop (4 symbols per stream) in 64-bit mode */
         endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
-        for ( ; (endSignal==BIT_DStream_unfinished) && (op4<(oend-7)) ; ) {
+        while ( (endSignal==BIT_DStream_unfinished) && (op4<(oend-3)) ) {
             HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
             HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
             HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
@@ -319,10 +276,15 @@ static size_t HUF_decompress4X2_usingDTable_internal(
             HUF_DECODE_SYMBOLX2_0(op2, &bitD2);
             HUF_DECODE_SYMBOLX2_0(op3, &bitD3);
             HUF_DECODE_SYMBOLX2_0(op4, &bitD4);
-            endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
+            BIT_reloadDStream(&bitD1);
+            BIT_reloadDStream(&bitD2);
+            BIT_reloadDStream(&bitD3);
+            BIT_reloadDStream(&bitD4);
         }
 
         /* check corruption */
+        /* note : should not be necessary : op# advance in lock step, and we control op4.
+         *        but curiously, binary generated by gcc 7.2 & 7.3 with -mbmi2 runs faster when >=1 test is present */
         if (op1 > opStart2) return ERROR(corruption_detected);
         if (op2 > opStart3) return ERROR(corruption_detected);
         if (op3 > opStart4) return ERROR(corruption_detected);
@@ -335,8 +297,8 @@ static size_t HUF_decompress4X2_usingDTable_internal(
         HUF_decodeStreamX2(op4, &bitD4, oend,     dt, dtLog);
 
         /* check */
-        endSignal = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
-        if (!endSignal) return ERROR(corruption_detected);
+        { U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
+          if (!endCheck) return ERROR(corruption_detected); }
 
         /* decoded size */
         return dstSize;
@@ -344,30 +306,309 @@ static size_t HUF_decompress4X2_usingDTable_internal(
 }
 
 
-size_t HUF_decompress4X2_usingDTable(
+FORCE_INLINE_TEMPLATE U32
+HUF_decodeSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const U32 dtLog)
+{
+    size_t const val = BIT_lookBitsFast(DStream, dtLog);   /* note : dtLog >= 1 */
+    memcpy(op, dt+val, 2);
+    BIT_skipBits(DStream, dt[val].nbBits);
+    return dt[val].length;
+}
+
+FORCE_INLINE_TEMPLATE U32
+HUF_decodeLastSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const U32 dtLog)
+{
+    size_t const val = BIT_lookBitsFast(DStream, dtLog);   /* note : dtLog >= 1 */
+    memcpy(op, dt+val, 1);
+    if (dt[val].length==1) BIT_skipBits(DStream, dt[val].nbBits);
+    else {
+        if (DStream->bitsConsumed < (sizeof(DStream->bitContainer)*8)) {
+            BIT_skipBits(DStream, dt[val].nbBits);
+            if (DStream->bitsConsumed > (sizeof(DStream->bitContainer)*8))
+                /* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */
+                DStream->bitsConsumed = (sizeof(DStream->bitContainer)*8);
+    }   }
+    return 1;
+}
+
+#define HUF_DECODE_SYMBOLX4_0(ptr, DStreamPtr) \
+    ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
+
+#define HUF_DECODE_SYMBOLX4_1(ptr, DStreamPtr) \
+    if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
+        ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
+
+#define HUF_DECODE_SYMBOLX4_2(ptr, DStreamPtr) \
+    if (MEM_64bits()) \
+        ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
+
+HINT_INLINE size_t
+HUF_decodeStreamX4(BYTE* p, BIT_DStream_t* bitDPtr, BYTE* const pEnd,
+                const HUF_DEltX4* const dt, const U32 dtLog)
+{
+    BYTE* const pStart = p;
+
+    /* up to 8 symbols at a time */
+    while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd-(sizeof(bitDPtr->bitContainer)-1))) {
+        HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
+        HUF_DECODE_SYMBOLX4_1(p, bitDPtr);
+        HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
+        HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
+    }
+
+    /* closer to end : up to 2 symbols at a time */
+    while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p <= pEnd-2))
+        HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
+
+    while (p <= pEnd-2)
+        HUF_DECODE_SYMBOLX4_0(p, bitDPtr);   /* no need to reload : reached the end of DStream */
+
+    if (p < pEnd)
+        p += HUF_decodeLastSymbolX4(p, bitDPtr, dt, dtLog);
+
+    return p-pStart;
+}
+
+FORCE_INLINE_TEMPLATE size_t
+HUF_decompress1X4_usingDTable_internal_body(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    BIT_DStream_t bitD;
+
+    /* Init */
+    CHECK_F( BIT_initDStream(&bitD, cSrc, cSrcSize) );
+
+    /* decode */
+    {   BYTE* const ostart = (BYTE*) dst;
+        BYTE* const oend = ostart + dstSize;
+        const void* const dtPtr = DTable+1;   /* force compiler to not use strict-aliasing */
+        const HUF_DEltX4* const dt = (const HUF_DEltX4*)dtPtr;
+        DTableDesc const dtd = HUF_getDTableDesc(DTable);
+        HUF_decodeStreamX4(ostart, &bitD, oend, dt, dtd.tableLog);
+    }
+
+    /* check */
+    if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected);
+
+    /* decoded size */
+    return dstSize;
+}
+
+
+FORCE_INLINE_TEMPLATE size_t
+HUF_decompress4X4_usingDTable_internal_body(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    if (cSrcSize < 10) return ERROR(corruption_detected);   /* strict minimum : jump table + 1 byte per stream */
+
+    {   const BYTE* const istart = (const BYTE*) cSrc;
+        BYTE* const ostart = (BYTE*) dst;
+        BYTE* const oend = ostart + dstSize;
+        const void* const dtPtr = DTable+1;
+        const HUF_DEltX4* const dt = (const HUF_DEltX4*)dtPtr;
+
+        /* Init */
+        BIT_DStream_t bitD1;
+        BIT_DStream_t bitD2;
+        BIT_DStream_t bitD3;
+        BIT_DStream_t bitD4;
+        size_t const length1 = MEM_readLE16(istart);
+        size_t const length2 = MEM_readLE16(istart+2);
+        size_t const length3 = MEM_readLE16(istart+4);
+        size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
+        const BYTE* const istart1 = istart + 6;  /* jumpTable */
+        const BYTE* const istart2 = istart1 + length1;
+        const BYTE* const istart3 = istart2 + length2;
+        const BYTE* const istart4 = istart3 + length3;
+        size_t const segmentSize = (dstSize+3) / 4;
+        BYTE* const opStart2 = ostart + segmentSize;
+        BYTE* const opStart3 = opStart2 + segmentSize;
+        BYTE* const opStart4 = opStart3 + segmentSize;
+        BYTE* op1 = ostart;
+        BYTE* op2 = opStart2;
+        BYTE* op3 = opStart3;
+        BYTE* op4 = opStart4;
+        U32 endSignal;
+        DTableDesc const dtd = HUF_getDTableDesc(DTable);
+        U32 const dtLog = dtd.tableLog;
+
+        if (length4 > cSrcSize) return ERROR(corruption_detected);   /* overflow */
+        CHECK_F( BIT_initDStream(&bitD1, istart1, length1) );
+        CHECK_F( BIT_initDStream(&bitD2, istart2, length2) );
+        CHECK_F( BIT_initDStream(&bitD3, istart3, length3) );
+        CHECK_F( BIT_initDStream(&bitD4, istart4, length4) );
+
+        /* 16-32 symbols per loop (4-8 symbols per stream) */
+        endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
+        for ( ; (endSignal==BIT_DStream_unfinished) & (op4<(oend-(sizeof(bitD4.bitContainer)-1))) ; ) {
+            HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
+            HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
+            HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
+            HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
+            HUF_DECODE_SYMBOLX4_1(op1, &bitD1);
+            HUF_DECODE_SYMBOLX4_1(op2, &bitD2);
+            HUF_DECODE_SYMBOLX4_1(op3, &bitD3);
+            HUF_DECODE_SYMBOLX4_1(op4, &bitD4);
+            HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
+            HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
+            HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
+            HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
+            HUF_DECODE_SYMBOLX4_0(op1, &bitD1);
+            HUF_DECODE_SYMBOLX4_0(op2, &bitD2);
+            HUF_DECODE_SYMBOLX4_0(op3, &bitD3);
+            HUF_DECODE_SYMBOLX4_0(op4, &bitD4);
+
+            endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
+        }
+
+        /* check corruption */
+        if (op1 > opStart2) return ERROR(corruption_detected);
+        if (op2 > opStart3) return ERROR(corruption_detected);
+        if (op3 > opStart4) return ERROR(corruption_detected);
+        /* note : op4 already verified within main loop */
+
+        /* finish bitStreams one by one */
+        HUF_decodeStreamX4(op1, &bitD1, opStart2, dt, dtLog);
+        HUF_decodeStreamX4(op2, &bitD2, opStart3, dt, dtLog);
+        HUF_decodeStreamX4(op3, &bitD3, opStart4, dt, dtLog);
+        HUF_decodeStreamX4(op4, &bitD4, oend,     dt, dtLog);
+
+        /* check */
+        { U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
+          if (!endCheck) return ERROR(corruption_detected); }
+
+        /* decoded size */
+        return dstSize;
+    }
+}
+
+
+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 X(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 X(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
+
+X(HUF_decompress1X2_usingDTable_internal)
+X(HUF_decompress4X2_usingDTable_internal)
+X(HUF_decompress1X4_usingDTable_internal)
+X(HUF_decompress4X4_usingDTable_internal)
+
+#undef X
+
+
+size_t HUF_decompress1X2_usingDTable(
           void* dst,  size_t dstSize,
     const void* cSrc, size_t cSrcSize,
     const HUF_DTable* DTable)
 {
     DTableDesc dtd = HUF_getDTableDesc(DTable);
     if (dtd.tableType != 0) return ERROR(GENERIC);
-    return HUF_decompress4X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
+    return HUF_decompress1X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
 }
 
-
-size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
+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)
 {
     const BYTE* ip = (const BYTE*) cSrc;
 
+    size_t const hSize = HUF_readDTableX2_wksp(DCtx, cSrc, cSrcSize, workSpace, wkspSize);
+    if (HUF_isError(hSize)) return hSize;
+    if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
+    ip += hSize; cSrcSize -= hSize;
+
+    return HUF_decompress1X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx, /* bmi2 */ 0);
+}
+
+
+size_t HUF_decompress1X2_DCtx(HUF_DTable* DCtx, void* dst, size_t dstSize,
+                              const void* cSrc, size_t cSrcSize)
+{
+    U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];
+    return HUF_decompress1X2_DCtx_wksp(DCtx, dst, dstSize, cSrc, cSrcSize,
+                                       workSpace, sizeof(workSpace));
+}
+
+size_t HUF_decompress1X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
+{
+    HUF_CREATE_STATIC_DTABLEX2(DTable, HUF_TABLELOG_MAX);
+    return HUF_decompress1X2_DCtx (DTable, dst, dstSize, cSrc, cSrcSize);
+}
+
+size_t HUF_decompress4X2_usingDTable(
+          void* dst,  size_t dstSize,
+    const void* cSrc, size_t cSrcSize,
+    const HUF_DTable* DTable)
+{
+    DTableDesc dtd = HUF_getDTableDesc(DTable);
+    if (dtd.tableType != 0) return ERROR(GENERIC);
+    return HUF_decompress4X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
+}
+
+static size_t HUF_decompress4X2_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;
+
     size_t const hSize = HUF_readDTableX2_wksp (dctx, cSrc, cSrcSize,
                                                 workSpace, wkspSize);
     if (HUF_isError(hSize)) return hSize;
     if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
     ip += hSize; cSrcSize -= hSize;
 
-    return HUF_decompress4X2_usingDTable_internal (dst, dstSize, ip, cSrcSize, dctx);
+    return HUF_decompress4X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
+}
+
+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)
+{
+    return HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, 0);
 }
 
 
@@ -387,8 +628,6 @@ size_t HUF_decompress4X2 (void* dst, size_t dstSize, const void* cSrc, size_t cS
 /* *************************/
 /* double-symbols decoding */
 /* *************************/
-typedef struct { U16 sequence; BYTE nbBits; BYTE length; } HUF_DEltX4;  /* double-symbols decoding */
-
 typedef struct { BYTE symbol; BYTE weight; } sortedSymbol_t;
 
 /* HUF_fillDTableX4Level2() :
@@ -508,10 +747,7 @@ size_t HUF_readDTableX4_wksp(HUF_DTable* DTable, const void* src,
     weightList = (BYTE *)((U32 *)workSpace + spaceUsed32);
     spaceUsed32 += HUF_ALIGN(HUF_SYMBOLVALUE_MAX + 1, sizeof(U32)) >> 2;
 
-    if ((spaceUsed32 << 2) > wkspSize)
-        return ERROR(tableLog_tooLarge);
-    workSpace = (U32 *)workSpace + spaceUsed32;
-    wkspSize -= (spaceUsed32 << 2);
+    if ((spaceUsed32 << 2) > wkspSize) return ERROR(tableLog_tooLarge);
 
     rankStart = rankStart0 + 1;
     memset(rankStats, 0, sizeof(U32) * (2 * HUF_TABLELOG_MAX + 2 + 1));
@@ -588,95 +824,6 @@ size_t HUF_readDTableX4(HUF_DTable* DTable, const void* src, size_t srcSize)
                                workSpace, sizeof(workSpace));
 }
 
-static U32 HUF_decodeSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const U32 dtLog)
-{
-    size_t const val = BIT_lookBitsFast(DStream, dtLog);   /* note : dtLog >= 1 */
-    memcpy(op, dt+val, 2);
-    BIT_skipBits(DStream, dt[val].nbBits);
-    return dt[val].length;
-}
-
-static U32 HUF_decodeLastSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const U32 dtLog)
-{
-    size_t const val = BIT_lookBitsFast(DStream, dtLog);   /* note : dtLog >= 1 */
-    memcpy(op, dt+val, 1);
-    if (dt[val].length==1) BIT_skipBits(DStream, dt[val].nbBits);
-    else {
-        if (DStream->bitsConsumed < (sizeof(DStream->bitContainer)*8)) {
-            BIT_skipBits(DStream, dt[val].nbBits);
-            if (DStream->bitsConsumed > (sizeof(DStream->bitContainer)*8))
-                /* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */
-                DStream->bitsConsumed = (sizeof(DStream->bitContainer)*8);
-    }   }
-    return 1;
-}
-
-
-#define HUF_DECODE_SYMBOLX4_0(ptr, DStreamPtr) \
-    ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
-
-#define HUF_DECODE_SYMBOLX4_1(ptr, DStreamPtr) \
-    if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
-        ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
-
-#define HUF_DECODE_SYMBOLX4_2(ptr, DStreamPtr) \
-    if (MEM_64bits()) \
-        ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
-
-HINT_INLINE size_t HUF_decodeStreamX4(BYTE* p, BIT_DStream_t* bitDPtr, BYTE* const pEnd, const HUF_DEltX4* const dt, const U32 dtLog)
-{
-    BYTE* const pStart = p;
-
-    /* up to 8 symbols at a time */
-    while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd-(sizeof(bitDPtr->bitContainer)-1))) {
-        HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
-        HUF_DECODE_SYMBOLX4_1(p, bitDPtr);
-        HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
-        HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
-    }
-
-    /* closer to end : up to 2 symbols at a time */
-    while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p <= pEnd-2))
-        HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
-
-    while (p <= pEnd-2)
-        HUF_DECODE_SYMBOLX4_0(p, bitDPtr);   /* no need to reload : reached the end of DStream */
-
-    if (p < pEnd)
-        p += HUF_decodeLastSymbolX4(p, bitDPtr, dt, dtLog);
-
-    return p-pStart;
-}
-
-
-static size_t HUF_decompress1X4_usingDTable_internal(
-          void* dst,  size_t dstSize,
-    const void* cSrc, size_t cSrcSize,
-    const HUF_DTable* DTable)
-{
-    BIT_DStream_t bitD;
-
-    /* Init */
-    {   size_t const errorCode = BIT_initDStream(&bitD, cSrc, cSrcSize);
-        if (HUF_isError(errorCode)) return errorCode;
-    }
-
-    /* decode */
-    {   BYTE* const ostart = (BYTE*) dst;
-        BYTE* const oend = ostart + dstSize;
-        const void* const dtPtr = DTable+1;   /* force compiler to not use strict-aliasing */
-        const HUF_DEltX4* const dt = (const HUF_DEltX4*)dtPtr;
-        DTableDesc const dtd = HUF_getDTableDesc(DTable);
-        HUF_decodeStreamX4(ostart, &bitD, oend, dt, dtd.tableLog);
-    }
-
-    /* check */
-    if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected);
-
-    /* decoded size */
-    return dstSize;
-}
-
 size_t HUF_decompress1X4_usingDTable(
           void* dst,  size_t dstSize,
     const void* cSrc, size_t cSrcSize,
@@ -684,7 +831,7 @@ size_t HUF_decompress1X4_usingDTable(
 {
     DTableDesc dtd = HUF_getDTableDesc(DTable);
     if (dtd.tableType != 1) return ERROR(GENERIC);
-    return HUF_decompress1X4_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
+    return HUF_decompress1X4_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
 }
 
 size_t HUF_decompress1X4_DCtx_wksp(HUF_DTable* DCtx, void* dst, size_t dstSize,
@@ -699,7 +846,7 @@ size_t HUF_decompress1X4_DCtx_wksp(HUF_DTable* DCtx, void* dst, size_t dstSize,
     if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
     ip += hSize; cSrcSize -= hSize;
 
-    return HUF_decompress1X4_usingDTable_internal (dst, dstSize, ip, cSrcSize, DCtx);
+    return HUF_decompress1X4_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx, /* bmi2 */ 0);
 }
 
 
@@ -717,99 +864,6 @@ size_t HUF_decompress1X4 (void* dst, size_t dstSize, const void* cSrc, size_t cS
     return HUF_decompress1X4_DCtx(DTable, dst, dstSize, cSrc, cSrcSize);
 }
 
-static size_t HUF_decompress4X4_usingDTable_internal(
-          void* dst,  size_t dstSize,
-    const void* cSrc, size_t cSrcSize,
-    const HUF_DTable* DTable)
-{
-    if (cSrcSize < 10) return ERROR(corruption_detected);   /* strict minimum : jump table + 1 byte per stream */
-
-    {   const BYTE* const istart = (const BYTE*) cSrc;
-        BYTE* const ostart = (BYTE*) dst;
-        BYTE* const oend = ostart + dstSize;
-        const void* const dtPtr = DTable+1;
-        const HUF_DEltX4* const dt = (const HUF_DEltX4*)dtPtr;
-
-        /* Init */
-        BIT_DStream_t bitD1;
-        BIT_DStream_t bitD2;
-        BIT_DStream_t bitD3;
-        BIT_DStream_t bitD4;
-        size_t const length1 = MEM_readLE16(istart);
-        size_t const length2 = MEM_readLE16(istart+2);
-        size_t const length3 = MEM_readLE16(istart+4);
-        size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
-        const BYTE* const istart1 = istart + 6;  /* jumpTable */
-        const BYTE* const istart2 = istart1 + length1;
-        const BYTE* const istart3 = istart2 + length2;
-        const BYTE* const istart4 = istart3 + length3;
-        size_t const segmentSize = (dstSize+3) / 4;
-        BYTE* const opStart2 = ostart + segmentSize;
-        BYTE* const opStart3 = opStart2 + segmentSize;
-        BYTE* const opStart4 = opStart3 + segmentSize;
-        BYTE* op1 = ostart;
-        BYTE* op2 = opStart2;
-        BYTE* op3 = opStart3;
-        BYTE* op4 = opStart4;
-        U32 endSignal;
-        DTableDesc const dtd = HUF_getDTableDesc(DTable);
-        U32 const dtLog = dtd.tableLog;
-
-        if (length4 > cSrcSize) return ERROR(corruption_detected);   /* overflow */
-        { size_t const errorCode = BIT_initDStream(&bitD1, istart1, length1);
-          if (HUF_isError(errorCode)) return errorCode; }
-        { size_t const errorCode = BIT_initDStream(&bitD2, istart2, length2);
-          if (HUF_isError(errorCode)) return errorCode; }
-        { size_t const errorCode = BIT_initDStream(&bitD3, istart3, length3);
-          if (HUF_isError(errorCode)) return errorCode; }
-        { size_t const errorCode = BIT_initDStream(&bitD4, istart4, length4);
-          if (HUF_isError(errorCode)) return errorCode; }
-
-        /* 16-32 symbols per loop (4-8 symbols per stream) */
-        endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
-        for ( ; (endSignal==BIT_DStream_unfinished) & (op4<(oend-(sizeof(bitD4.bitContainer)-1))) ; ) {
-            HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
-            HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
-            HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
-            HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
-            HUF_DECODE_SYMBOLX4_1(op1, &bitD1);
-            HUF_DECODE_SYMBOLX4_1(op2, &bitD2);
-            HUF_DECODE_SYMBOLX4_1(op3, &bitD3);
-            HUF_DECODE_SYMBOLX4_1(op4, &bitD4);
-            HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
-            HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
-            HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
-            HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
-            HUF_DECODE_SYMBOLX4_0(op1, &bitD1);
-            HUF_DECODE_SYMBOLX4_0(op2, &bitD2);
-            HUF_DECODE_SYMBOLX4_0(op3, &bitD3);
-            HUF_DECODE_SYMBOLX4_0(op4, &bitD4);
-
-            endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
-        }
-
-        /* check corruption */
-        if (op1 > opStart2) return ERROR(corruption_detected);
-        if (op2 > opStart3) return ERROR(corruption_detected);
-        if (op3 > opStart4) return ERROR(corruption_detected);
-        /* note : op4 already verified within main loop */
-
-        /* finish bitStreams one by one */
-        HUF_decodeStreamX4(op1, &bitD1, opStart2, dt, dtLog);
-        HUF_decodeStreamX4(op2, &bitD2, opStart3, dt, dtLog);
-        HUF_decodeStreamX4(op3, &bitD3, opStart4, dt, dtLog);
-        HUF_decodeStreamX4(op4, &bitD4, oend,     dt, dtLog);
-
-        /* check */
-        { U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
-          if (!endCheck) return ERROR(corruption_detected); }
-
-        /* decoded size */
-        return dstSize;
-    }
-}
-
-
 size_t HUF_decompress4X4_usingDTable(
           void* dst,  size_t dstSize,
     const void* cSrc, size_t cSrcSize,
@@ -817,13 +871,12 @@ size_t HUF_decompress4X4_usingDTable(
 {
     DTableDesc dtd = HUF_getDTableDesc(DTable);
     if (dtd.tableType != 1) return ERROR(GENERIC);
-    return HUF_decompress4X4_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
+    return HUF_decompress4X4_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
 }
 
-
-size_t HUF_decompress4X4_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
+static size_t HUF_decompress4X4_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize,
                                    const void* cSrc, size_t cSrcSize,
-                                   void* workSpace, size_t wkspSize)
+                                   void* workSpace, size_t wkspSize, int bmi2)
 {
     const BYTE* ip = (const BYTE*) cSrc;
 
@@ -833,7 +886,14 @@ size_t HUF_decompress4X4_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
     if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
     ip += hSize; cSrcSize -= hSize;
 
-    return HUF_decompress4X4_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx);
+    return HUF_decompress4X4_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
+}
+
+size_t HUF_decompress4X4_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
+                                   const void* cSrc, size_t cSrcSize,
+                                   void* workSpace, size_t wkspSize)
+{
+    return HUF_decompress4X4_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, /* bmi2 */ 0);
 }
 
 
@@ -861,8 +921,8 @@ size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize,
                                     const HUF_DTable* DTable)
 {
     DTableDesc const dtd = HUF_getDTableDesc(DTable);
-    return dtd.tableType ? HUF_decompress1X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable) :
-                           HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable);
+    return dtd.tableType ? HUF_decompress1X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) :
+                           HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
 }
 
 size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize,
@@ -870,8 +930,8 @@ size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize,
                                     const HUF_DTable* DTable)
 {
     DTableDesc const dtd = HUF_getDTableDesc(DTable);
-    return dtd.tableType ? HUF_decompress4X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable) :
-                           HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable);
+    return dtd.tableType ? HUF_decompress4X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) :
+                           HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
 }
 
 
@@ -898,21 +958,22 @@ static const algo_time_t algoTime[16 /* Quantization */][3 /* single, double, qu
 };
 
 /** HUF_selectDecoder() :
-*   Tells which decoder is likely to decode faster,
-*   based on a set of pre-determined metrics.
-*   @return : 0==HUF_decompress4X2, 1==HUF_decompress4X4 .
-*   Assumption : 0 < cSrcSize, dstSize <= 128 KB */
+ *  Tells which decoder is likely to decode faster,
+ *  based on a set of pre-computed metrics.
+ * @return : 0==HUF_decompress4X2, 1==HUF_decompress4X4 .
+ *  Assumption : 0 < dstSize <= 128 KB */
 U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize)
 {
+    assert(dstSize > 0);
+    assert(dstSize <= 128 KB);
     /* decoder timing evaluation */
-    U32 const Q = cSrcSize >= dstSize ? 15 : (U32)(cSrcSize * 16 / dstSize);   /* Q < 16 */
-    U32 const D256 = (U32)(dstSize >> 8);
-    U32 const DTime0 = algoTime[Q][0].tableTime + (algoTime[Q][0].decode256Time * D256);
-    U32 DTime1 = algoTime[Q][1].tableTime + (algoTime[Q][1].decode256Time * D256);
-    DTime1 += DTime1 >> 3;  /* advantage to algorithm using less memory, for cache eviction */
-
-    return DTime1 < DTime0;
-}
+    {   U32 const Q = (cSrcSize >= dstSize) ? 15 : (U32)(cSrcSize * 16 / dstSize);   /* Q < 16 */
+        U32 const D256 = (U32)(dstSize >> 8);
+        U32 const DTime0 = algoTime[Q][0].tableTime + (algoTime[Q][0].decode256Time * D256);
+        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;
+}   }
 
 
 typedef size_t (*decompressionAlgo)(void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);
@@ -994,3 +1055,42 @@ size_t HUF_decompress1X_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize,
     return HUF_decompress1X_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize,
                                       workSpace, sizeof(workSpace));
 }
+
+
+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);
+    return dtd.tableType ? HUF_decompress1X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2) :
+                           HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
+}
+
+size_t HUF_decompress1X2_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;
+
+    size_t const hSize = HUF_readDTableX2_wksp(dctx, cSrc, cSrcSize, workSpace, wkspSize);
+    if (HUF_isError(hSize)) return hSize;
+    if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
+    ip += hSize; cSrcSize -= hSize;
+
+    return HUF_decompress1X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
+}
+
+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);
+    return dtd.tableType ? HUF_decompress4X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2) :
+                           HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, 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)
+{
+    /* validation checks */
+    if (dstSize == 0) return ERROR(dstSize_tooSmall);
+    if (cSrcSize == 0) return ERROR(corruption_detected);
+
+    {   U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
+        return algoNb ? HUF_decompress4X4_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2) :
+                        HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
+    }
+}