zstdlib 0.10.0-arm64-darwin → 0.12.0-arm64-darwin

data/ext/zstdlib_c/{zlib-1.2.12 → zlib-1.3.1}/zutil.c

@@ -24,13 +24,11 @@ z_const char * const z_errmsg[10] = {
 };
 
 
-const char * ZEXPORT zlibVersion()
-{
+const char * ZEXPORT zlibVersion(void) {
     return ZLIB_VERSION;
 }
 
-uLong ZEXPORT zlibCompileFlags()
-{
+uLong ZEXPORT zlibCompileFlags(void) {
     uLong flags;
 
     flags = 0;
@@ -61,9 +59,11 @@ uLong ZEXPORT zlibCompileFlags()
 #ifdef ZLIB_DEBUG
     flags += 1 << 8;
 #endif
+    /*
 #if defined(ASMV) || defined(ASMINF)
     flags += 1 << 9;
 #endif
+     */
 #ifdef ZLIB_WINAPI
     flags += 1 << 10;
 #endif
@@ -119,9 +119,7 @@ uLong ZEXPORT zlibCompileFlags()
 #  endif
 int ZLIB_INTERNAL z_verbose = verbose;
 
-void ZLIB_INTERNAL z_error (m)
-    char *m;
-{
+void ZLIB_INTERNAL z_error(char *m) {
     fprintf(stderr, "%s\n", m);
     exit(1);
 }
@@ -130,9 +128,7 @@ void ZLIB_INTERNAL z_error (m)
 /* exported to allow conversion of error code to string for compress() and
  * uncompress()
  */
-const char * ZEXPORT zError(err)
-    int err;
-{
+const char * ZEXPORT zError(int err) {
     return ERR_MSG(err);
 }
 
@@ -146,22 +142,14 @@ const char * ZEXPORT zError(err)
 
 #ifndef HAVE_MEMCPY
 
-void ZLIB_INTERNAL zmemcpy(dest, source, len)
-    Bytef* dest;
-    const Bytef* source;
-    uInt  len;
-{
+void ZLIB_INTERNAL zmemcpy(Bytef* dest, const Bytef* source, uInt len) {
     if (len == 0) return;
     do {
         *dest++ = *source++; /* ??? to be unrolled */
     } while (--len != 0);
 }
 
-int ZLIB_INTERNAL zmemcmp(s1, s2, len)
-    const Bytef* s1;
-    const Bytef* s2;
-    uInt  len;
-{
+int ZLIB_INTERNAL zmemcmp(const Bytef* s1, const Bytef* s2, uInt len) {
     uInt j;
 
     for (j = 0; j < len; j++) {
@@ -170,10 +158,7 @@ int ZLIB_INTERNAL zmemcmp(s1, s2, len)
     return 0;
 }
 
-void ZLIB_INTERNAL zmemzero(dest, len)
-    Bytef* dest;
-    uInt  len;
-{
+void ZLIB_INTERNAL zmemzero(Bytef* dest, uInt len) {
     if (len == 0) return;
     do {
         *dest++ = 0;  /* ??? to be unrolled */
@@ -214,8 +199,7 @@ local ptr_table table[MAX_PTR];
  * a protected system like OS/2. Use Microsoft C instead.
  */
 
-voidpf ZLIB_INTERNAL zcalloc (voidpf opaque, unsigned items, unsigned size)
-{
+voidpf ZLIB_INTERNAL zcalloc(voidpf opaque, unsigned items, unsigned size) {
     voidpf buf;
     ulg bsize = (ulg)items*size;
 
@@ -240,8 +224,7 @@ voidpf ZLIB_INTERNAL zcalloc (voidpf opaque, unsigned items, unsigned size)
     return buf;
 }
 
-void ZLIB_INTERNAL zcfree (voidpf opaque, voidpf ptr)
-{
+void ZLIB_INTERNAL zcfree(voidpf opaque, voidpf ptr) {
     int n;
 
     (void)opaque;
@@ -277,14 +260,12 @@ void ZLIB_INTERNAL zcfree (voidpf opaque, voidpf ptr)
 #  define _hfree   hfree
 #endif
 
-voidpf ZLIB_INTERNAL zcalloc (voidpf opaque, uInt items, uInt size)
-{
+voidpf ZLIB_INTERNAL zcalloc(voidpf opaque, uInt items, uInt size) {
     (void)opaque;
     return _halloc((long)items, size);
 }
 
-void ZLIB_INTERNAL zcfree (voidpf opaque, voidpf ptr)
-{
+void ZLIB_INTERNAL zcfree(voidpf opaque, voidpf ptr) {
     (void)opaque;
     _hfree(ptr);
 }
@@ -297,25 +278,18 @@ void ZLIB_INTERNAL zcfree (voidpf opaque, voidpf ptr)
 #ifndef MY_ZCALLOC /* Any system without a special alloc function */
 
 #ifndef STDC
-extern voidp  malloc OF((uInt size));
-extern voidp  calloc OF((uInt items, uInt size));
-extern void   free   OF((voidpf ptr));
+extern voidp malloc(uInt size);
+extern voidp calloc(uInt items, uInt size);
+extern void free(voidpf ptr);
 #endif
 
-voidpf ZLIB_INTERNAL zcalloc (opaque, items, size)
-    voidpf opaque;
-    unsigned items;
-    unsigned size;
-{
+voidpf ZLIB_INTERNAL zcalloc(voidpf opaque, unsigned items, unsigned size) {
     (void)opaque;
     return sizeof(uInt) > 2 ? (voidpf)malloc(items * size) :
                               (voidpf)calloc(items, size);
 }
 
-void ZLIB_INTERNAL zcfree (opaque, ptr)
-    voidpf opaque;
-    voidpf ptr;
-{
+void ZLIB_INTERNAL zcfree(voidpf opaque, voidpf ptr) {
     (void)opaque;
     free(ptr);
 }

data/ext/zstdlib_c/{zlib-1.2.12 → zlib-1.3.1}/zutil.h

@@ -1,5 +1,5 @@
 /* zutil.h -- internal interface and configuration of the compression library
- * Copyright (C) 1995-2022 Jean-loup Gailly, Mark Adler
+ * Copyright (C) 1995-2024 Jean-loup Gailly, Mark Adler
  * For conditions of distribution and use, see copyright notice in zlib.h
  */
 
@@ -56,7 +56,7 @@ typedef unsigned long  ulg;
 extern z_const char * const z_errmsg[10]; /* indexed by 2-zlib_error */
 /* (size given to avoid silly warnings with Visual C++) */
 
-#define ERR_MSG(err) z_errmsg[Z_NEED_DICT-(err)]
+#define ERR_MSG(err) z_errmsg[(err) < -6 || (err) > 2 ? 9 : 2 - (err)]
 
 #define ERR_RETURN(strm,err) \
   return (strm->msg = ERR_MSG(err), (err))
@@ -137,17 +137,8 @@ extern z_const char * const z_errmsg[10]; /* indexed by 2-zlib_error */
 #  endif
 #endif
 
-#if defined(MACOS) || defined(TARGET_OS_MAC)
+#if defined(MACOS)
 #  define OS_CODE  7
-#  ifndef Z_SOLO
-#    if defined(__MWERKS__) && __dest_os != __be_os && __dest_os != __win32_os
-#      include <unix.h> /* for fdopen */
-#    else
-#      ifndef fdopen
-#        define fdopen(fd,mode) NULL /* No fdopen() */
-#      endif
-#    endif
-#  endif
 #endif
 
 #ifdef __acorn
@@ -170,18 +161,6 @@ extern z_const char * const z_errmsg[10]; /* indexed by 2-zlib_error */
 #  define OS_CODE 19
 #endif
 
-#if defined(_BEOS_) || defined(RISCOS)
-#  define fdopen(fd,mode) NULL /* No fdopen() */
-#endif
-
-#if (defined(_MSC_VER) && (_MSC_VER > 600)) && !defined __INTERIX
-#  if defined(_WIN32_WCE)
-#    define fdopen(fd,mode) NULL /* No fdopen() */
-#  else
-#    define fdopen(fd,type)  _fdopen(fd,type)
-#  endif
-#endif
-
 #if defined(__BORLANDC__) && !defined(MSDOS)
   #pragma warn -8004
   #pragma warn -8008
@@ -191,8 +170,9 @@ extern z_const char * const z_errmsg[10]; /* indexed by 2-zlib_error */
 /* provide prototypes for these when building zlib without LFS */
 #if !defined(_WIN32) && \
     (!defined(_LARGEFILE64_SOURCE) || _LFS64_LARGEFILE-0 == 0)
-    ZEXTERN uLong ZEXPORT adler32_combine64 OF((uLong, uLong, z_off_t));
-    ZEXTERN uLong ZEXPORT crc32_combine64 OF((uLong, uLong, z_off_t));
+    ZEXTERN uLong ZEXPORT adler32_combine64(uLong, uLong, z_off_t);
+    ZEXTERN uLong ZEXPORT crc32_combine64(uLong, uLong, z_off_t);
+    ZEXTERN uLong ZEXPORT crc32_combine_gen64(z_off_t);
 #endif
 
         /* common defaults */
@@ -231,16 +211,16 @@ extern z_const char * const z_errmsg[10]; /* indexed by 2-zlib_error */
 #    define zmemzero(dest, len) memset(dest, 0, len)
 #  endif
 #else
-   void ZLIB_INTERNAL zmemcpy OF((Bytef* dest, const Bytef* source, uInt len));
-   int ZLIB_INTERNAL zmemcmp OF((const Bytef* s1, const Bytef* s2, uInt len));
-   void ZLIB_INTERNAL zmemzero OF((Bytef* dest, uInt len));
+   void ZLIB_INTERNAL zmemcpy(Bytef* dest, const Bytef* source, uInt len);
+   int ZLIB_INTERNAL zmemcmp(const Bytef* s1, const Bytef* s2, uInt len);
+   void ZLIB_INTERNAL zmemzero(Bytef* dest, uInt len);
 #endif
 
 /* Diagnostic functions */
 #ifdef ZLIB_DEBUG
 #  include <stdio.h>
    extern int ZLIB_INTERNAL z_verbose;
-   extern void ZLIB_INTERNAL z_error OF((char *m));
+   extern void ZLIB_INTERNAL z_error(char *m);
 #  define Assert(cond,msg) {if(!(cond)) z_error(msg);}
 #  define Trace(x) {if (z_verbose>=0) fprintf x ;}
 #  define Tracev(x) {if (z_verbose>0) fprintf x ;}
@@ -257,9 +237,9 @@ extern z_const char * const z_errmsg[10]; /* indexed by 2-zlib_error */
 #endif
 
 #ifndef Z_SOLO
-   voidpf ZLIB_INTERNAL zcalloc OF((voidpf opaque, unsigned items,
-                                    unsigned size));
-   void ZLIB_INTERNAL zcfree  OF((voidpf opaque, voidpf ptr));
+   voidpf ZLIB_INTERNAL zcalloc(voidpf opaque, unsigned items,
+                                unsigned size);
+   void ZLIB_INTERNAL zcfree(voidpf opaque, voidpf ptr);
 #endif
 
 #define ZALLOC(strm, items, size) \

data/ext/zstdlib_c/zstd-1.5.5/lib/common/allocations.h

@@ -0,0 +1,55 @@
+/*
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
+ * 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.
+ */
+
+/* This file provides custom allocation primitives
+ */
+
+#define ZSTD_DEPS_NEED_MALLOC
+#include "zstd_deps.h"   /* ZSTD_malloc, ZSTD_calloc, ZSTD_free, ZSTD_memset */
+
+#include "mem.h" /* MEM_STATIC */
+#define ZSTD_STATIC_LINKING_ONLY
+#include "../zstd.h" /* ZSTD_customMem */
+
+#ifndef ZSTD_ALLOCATIONS_H
+#define ZSTD_ALLOCATIONS_H
+
+/* custom memory allocation functions */
+
+MEM_STATIC void* ZSTD_customMalloc(size_t size, ZSTD_customMem customMem)
+{
+    if (customMem.customAlloc)
+        return customMem.customAlloc(customMem.opaque, size);
+    return ZSTD_malloc(size);
+}
+
+MEM_STATIC void* ZSTD_customCalloc(size_t size, ZSTD_customMem customMem)
+{
+    if (customMem.customAlloc) {
+        /* calloc implemented as malloc+memset;
+         * not as efficient as calloc, but next best guess for custom malloc */
+        void* const ptr = customMem.customAlloc(customMem.opaque, size);
+        ZSTD_memset(ptr, 0, size);
+        return ptr;
+    }
+    return ZSTD_calloc(1, size);
+}
+
+MEM_STATIC void ZSTD_customFree(void* ptr, ZSTD_customMem customMem)
+{
+    if (ptr!=NULL) {
+        if (customMem.customFree)
+            customMem.customFree(customMem.opaque, ptr);
+        else
+            ZSTD_free(ptr);
+    }
+}
+
+#endif /* ZSTD_ALLOCATIONS_H */

data/ext/zstdlib_c/zstd-1.5.5/lib/common/bits.h

@@ -0,0 +1,200 @@
+/*
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
+ * 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_BITS_H
+#define ZSTD_BITS_H
+
+#include "mem.h"
+
+MEM_STATIC unsigned ZSTD_countTrailingZeros32_fallback(U32 val)
+{
+    assert(val != 0);
+    {
+        static const U32 DeBruijnBytePos[32] = {0, 1, 28, 2, 29, 14, 24, 3,
+                                                30, 22, 20, 15, 25, 17, 4, 8,
+                                                31, 27, 13, 23, 21, 19, 16, 7,
+                                                26, 12, 18, 6, 11, 5, 10, 9};
+        return DeBruijnBytePos[((U32) ((val & -(S32) val) * 0x077CB531U)) >> 27];
+    }
+}
+
+MEM_STATIC unsigned ZSTD_countTrailingZeros32(U32 val)
+{
+    assert(val != 0);
+#   if defined(_MSC_VER)
+#       if STATIC_BMI2 == 1
+            return (unsigned)_tzcnt_u32(val);
+#       else
+            if (val != 0) {
+                unsigned long r;
+                _BitScanForward(&r, val);
+                return (unsigned)r;
+            } else {
+                /* Should not reach this code path */
+                __assume(0);
+            }
+#       endif
+#   elif defined(__GNUC__) && (__GNUC__ >= 4)
+        return (unsigned)__builtin_ctz(val);
+#   else
+        return ZSTD_countTrailingZeros32_fallback(val);
+#   endif
+}
+
+MEM_STATIC unsigned ZSTD_countLeadingZeros32_fallback(U32 val) {
+    assert(val != 0);
+    {
+        static const U32 DeBruijnClz[32] = {0, 9, 1, 10, 13, 21, 2, 29,
+                                            11, 14, 16, 18, 22, 25, 3, 30,
+                                            8, 12, 20, 28, 15, 17, 24, 7,
+                                            19, 27, 23, 6, 26, 5, 4, 31};
+        val |= val >> 1;
+        val |= val >> 2;
+        val |= val >> 4;
+        val |= val >> 8;
+        val |= val >> 16;
+        return 31 - DeBruijnClz[(val * 0x07C4ACDDU) >> 27];
+    }
+}
+
+MEM_STATIC unsigned ZSTD_countLeadingZeros32(U32 val)
+{
+    assert(val != 0);
+#   if defined(_MSC_VER)
+#       if STATIC_BMI2 == 1
+            return (unsigned)_lzcnt_u32(val);
+#       else
+            if (val != 0) {
+                unsigned long r;
+                _BitScanReverse(&r, val);
+                return (unsigned)(31 - r);
+            } else {
+                /* Should not reach this code path */
+                __assume(0);
+            }
+#       endif
+#   elif defined(__GNUC__) && (__GNUC__ >= 4)
+        return (unsigned)__builtin_clz(val);
+#   else
+        return ZSTD_countLeadingZeros32_fallback(val);
+#   endif
+}
+
+MEM_STATIC unsigned ZSTD_countTrailingZeros64(U64 val)
+{
+    assert(val != 0);
+#   if defined(_MSC_VER) && defined(_WIN64)
+#       if STATIC_BMI2 == 1
+            return (unsigned)_tzcnt_u64(val);
+#       else
+            if (val != 0) {
+                unsigned long r;
+                _BitScanForward64(&r, val);
+                return (unsigned)r;
+            } else {
+                /* Should not reach this code path */
+                __assume(0);
+            }
+#       endif
+#   elif defined(__GNUC__) && (__GNUC__ >= 4) && defined(__LP64__)
+        return (unsigned)__builtin_ctzll(val);
+#   else
+        {
+            U32 mostSignificantWord = (U32)(val >> 32);
+            U32 leastSignificantWord = (U32)val;
+            if (leastSignificantWord == 0) {
+                return 32 + ZSTD_countTrailingZeros32(mostSignificantWord);
+            } else {
+                return ZSTD_countTrailingZeros32(leastSignificantWord);
+            }
+        }
+#   endif
+}
+
+MEM_STATIC unsigned ZSTD_countLeadingZeros64(U64 val)
+{
+    assert(val != 0);
+#   if defined(_MSC_VER) && defined(_WIN64)
+#       if STATIC_BMI2 == 1
+            return (unsigned)_lzcnt_u64(val);
+#       else
+            if (val != 0) {
+                unsigned long r;
+                _BitScanReverse64(&r, val);
+                return (unsigned)(63 - r);
+            } else {
+                /* Should not reach this code path */
+                __assume(0);
+            }
+#       endif
+#   elif defined(__GNUC__) && (__GNUC__ >= 4)
+        return (unsigned)(__builtin_clzll(val));
+#   else
+        {
+            U32 mostSignificantWord = (U32)(val >> 32);
+            U32 leastSignificantWord = (U32)val;
+            if (mostSignificantWord == 0) {
+                return 32 + ZSTD_countLeadingZeros32(leastSignificantWord);
+            } else {
+                return ZSTD_countLeadingZeros32(mostSignificantWord);
+            }
+        }
+#   endif
+}
+
+MEM_STATIC unsigned ZSTD_NbCommonBytes(size_t val)
+{
+    if (MEM_isLittleEndian()) {
+        if (MEM_64bits()) {
+            return ZSTD_countTrailingZeros64((U64)val) >> 3;
+        } else {
+            return ZSTD_countTrailingZeros32((U32)val) >> 3;
+        }
+    } else {  /* Big Endian CPU */
+        if (MEM_64bits()) {
+            return ZSTD_countLeadingZeros64((U64)val) >> 3;
+        } else {
+            return ZSTD_countLeadingZeros32((U32)val) >> 3;
+        }
+    }
+}
+
+MEM_STATIC unsigned ZSTD_highbit32(U32 val)   /* compress, dictBuilder, decodeCorpus */
+{
+    assert(val != 0);
+    return 31 - ZSTD_countLeadingZeros32(val);
+}
+
+/* ZSTD_rotateRight_*():
+ * Rotates a bitfield to the right by "count" bits.
+ * https://en.wikipedia.org/w/index.php?title=Circular_shift&oldid=991635599#Implementing_circular_shifts
+ */
+MEM_STATIC
+U64 ZSTD_rotateRight_U64(U64 const value, U32 count) {
+    assert(count < 64);
+    count &= 0x3F; /* for fickle pattern recognition */
+    return (value >> count) | (U64)(value << ((0U - count) & 0x3F));
+}
+
+MEM_STATIC
+U32 ZSTD_rotateRight_U32(U32 const value, U32 count) {
+    assert(count < 32);
+    count &= 0x1F; /* for fickle pattern recognition */
+    return (value >> count) | (U32)(value << ((0U - count) & 0x1F));
+}
+
+MEM_STATIC
+U16 ZSTD_rotateRight_U16(U16 const value, U32 count) {
+    assert(count < 16);
+    count &= 0x0F; /* for fickle pattern recognition */
+    return (value >> count) | (U16)(value << ((0U - count) & 0x0F));
+}
+
+#endif /* ZSTD_BITS_H */

data/ext/zstdlib_c/{zstd-1.5.2 → zstd-1.5.5}/lib/common/bitstream.h

@@ -1,7 +1,7 @@
 /* ******************************************************************
  * bitstream
  * Part of FSE library
- * Copyright (c) Yann Collet, Facebook, Inc.
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
  *
  * You can contact the author at :
  * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
@@ -30,14 +30,15 @@ extern "C" {
 #include "compiler.h"       /* UNLIKELY() */
 #include "debug.h"          /* assert(), DEBUGLOG(), RAWLOG() */
 #include "error_private.h"  /* error codes and messages */
+#include "bits.h"           /* ZSTD_highbit32 */
 
 
 /*=========================================
 *  Target specific
 =========================================*/
 #ifndef ZSTD_NO_INTRINSICS
-#  if defined(__BMI__) && defined(__GNUC__)
-#    include <immintrin.h>   /* support for bextr (experimental) */
+#  if (defined(__BMI__) || defined(__BMI2__)) && defined(__GNUC__)
+#    include <immintrin.h>   /* support for bextr (experimental)/bzhi */
 #  elif defined(__ICCARM__)
 #    include <intrinsics.h>
 #  endif
@@ -132,48 +133,6 @@ MEM_STATIC void BIT_flushBitsFast(BIT_CStream_t* bitC);
 MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, unsigned nbBits);
 /* faster, but works only if nbBits >= 1 */
 
-
-
-/*-**************************************************************
-*  Internal functions
-****************************************************************/
-MEM_STATIC unsigned BIT_highbit32 (U32 val)
-{
-    assert(val != 0);
-    {
-#   if defined(_MSC_VER)   /* Visual */
-#       if STATIC_BMI2 == 1
-            return _lzcnt_u32(val) ^ 31;
-#       else
-            if (val != 0) {
-                unsigned long r;
-                _BitScanReverse(&r, val);
-                return (unsigned)r;
-            } else {
-                /* Should not reach this code path */
-                __assume(0);
-            }
-#       endif
-#   elif defined(__GNUC__) && (__GNUC__ >= 3)   /* Use GCC Intrinsic */
-        return __builtin_clz (val) ^ 31;
-#   elif defined(__ICCARM__)    /* IAR Intrinsic */
-        return 31 - __CLZ(val);
-#   else   /* Software version */
-        static const unsigned DeBruijnClz[32] = { 0,  9,  1, 10, 13, 21,  2, 29,
-                                                 11, 14, 16, 18, 22, 25,  3, 30,
-                                                  8, 12, 20, 28, 15, 17, 24,  7,
-                                                 19, 27, 23,  6, 26,  5,  4, 31 };
-        U32 v = val;
-        v |= v >> 1;
-        v |= v >> 2;
-        v |= v >> 4;
-        v |= v >> 8;
-        v |= v >> 16;
-        return DeBruijnClz[ (U32) (v * 0x07C4ACDDU) >> 27];
-#   endif
-    }
-}
-
 /*=====    Local Constants   =====*/
 static const unsigned BIT_mask[] = {
     0,          1,         3,         7,         0xF,       0x1F,
@@ -203,6 +162,16 @@ MEM_STATIC size_t BIT_initCStream(BIT_CStream_t* bitC,
     return 0;
 }
 
+MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getLowerBits(size_t bitContainer, U32 const nbBits)
+{
+#if defined(STATIC_BMI2) && STATIC_BMI2 == 1 && !defined(ZSTD_NO_INTRINSICS)
+    return  _bzhi_u64(bitContainer, nbBits);
+#else
+    assert(nbBits < BIT_MASK_SIZE);
+    return bitContainer & BIT_mask[nbBits];
+#endif
+}
+
 /*! BIT_addBits() :
  *  can add up to 31 bits into `bitC`.
  *  Note : does not check for register overflow ! */
@@ -212,7 +181,7 @@ MEM_STATIC void BIT_addBits(BIT_CStream_t* bitC,
     DEBUG_STATIC_ASSERT(BIT_MASK_SIZE == 32);
     assert(nbBits < BIT_MASK_SIZE);
     assert(nbBits + bitC->bitPos < sizeof(bitC->bitContainer) * 8);
-    bitC->bitContainer |= (value & BIT_mask[nbBits]) << bitC->bitPos;
+    bitC->bitContainer |= BIT_getLowerBits(value, nbBits) << bitC->bitPos;
     bitC->bitPos += nbBits;
 }
 
@@ -291,7 +260,7 @@ MEM_STATIC size_t BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, si
         bitD->ptr   = (const char*)srcBuffer + srcSize - sizeof(bitD->bitContainer);
         bitD->bitContainer = MEM_readLEST(bitD->ptr);
         { BYTE const lastByte = ((const BYTE*)srcBuffer)[srcSize-1];
-          bitD->bitsConsumed = lastByte ? 8 - BIT_highbit32(lastByte) : 0;  /* ensures bitsConsumed is always set */
+          bitD->bitsConsumed = lastByte ? 8 - ZSTD_highbit32(lastByte) : 0;  /* ensures bitsConsumed is always set */
           if (lastByte == 0) return ERROR(GENERIC); /* endMark not present */ }
     } else {
         bitD->ptr   = bitD->start;
@@ -319,7 +288,7 @@ MEM_STATIC size_t BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, si
         default: break;
         }
         {   BYTE const lastByte = ((const BYTE*)srcBuffer)[srcSize-1];
-            bitD->bitsConsumed = lastByte ? 8 - BIT_highbit32(lastByte) : 0;
+            bitD->bitsConsumed = lastByte ? 8 - ZSTD_highbit32(lastByte) : 0;
             if (lastByte == 0) return ERROR(corruption_detected);  /* endMark not present */
         }
         bitD->bitsConsumed += (U32)(sizeof(bitD->bitContainer) - srcSize)*8;
@@ -350,16 +319,6 @@ MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getMiddleBits(size_t bitContainer, U32 c
 #endif
 }
 
-MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getLowerBits(size_t bitContainer, U32 const nbBits)
-{
-#if defined(STATIC_BMI2) && STATIC_BMI2 == 1
-	return  _bzhi_u64(bitContainer, nbBits);
-#else
-    assert(nbBits < BIT_MASK_SIZE);
-    return bitContainer & BIT_mask[nbBits];
-#endif
-}
-
 /*! BIT_lookBits() :
  *  Provides next n bits from local register.
  *  local register is not modified.
@@ -406,7 +365,7 @@ MEM_STATIC FORCE_INLINE_ATTR size_t BIT_readBits(BIT_DStream_t* bitD, unsigned n
 }
 
 /*! BIT_readBitsFast() :
- *  unsafe version; only works only if nbBits >= 1 */
+ *  unsafe version; only works if nbBits >= 1 */
 MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, unsigned nbBits)
 {
     size_t const value = BIT_lookBitsFast(bitD, nbBits);
@@ -437,7 +396,7 @@ MEM_STATIC BIT_DStream_status BIT_reloadDStreamFast(BIT_DStream_t* bitD)
  *  This function is safe, it guarantees it will not read beyond src buffer.
  * @return : status of `BIT_DStream_t` internal register.
  *           when status == BIT_DStream_unfinished, internal register is filled with at least 25 or 57 bits */
-MEM_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD)
+MEM_STATIC FORCE_INLINE_ATTR BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD)
 {
     if (bitD->bitsConsumed > (sizeof(bitD->bitContainer)*8))  /* overflow detected, like end of stream */
         return BIT_DStream_overflow;