extzstd 0.3.2 → 0.4

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (112) hide show
  1. checksums.yaml +4 -4
  2. data/README.md +4 -3
  3. data/contrib/zstd/CHANGELOG +225 -1
  4. data/contrib/zstd/CONTRIBUTING.md +158 -75
  5. data/contrib/zstd/LICENSE +4 -4
  6. data/contrib/zstd/Makefile +106 -69
  7. data/contrib/zstd/Package.swift +36 -0
  8. data/contrib/zstd/README.md +64 -36
  9. data/contrib/zstd/SECURITY.md +15 -0
  10. data/contrib/zstd/TESTING.md +2 -3
  11. data/contrib/zstd/lib/BUCK +5 -7
  12. data/contrib/zstd/lib/Makefile +117 -199
  13. data/contrib/zstd/lib/README.md +37 -7
  14. data/contrib/zstd/lib/common/allocations.h +55 -0
  15. data/contrib/zstd/lib/common/bits.h +200 -0
  16. data/contrib/zstd/lib/common/bitstream.h +80 -86
  17. data/contrib/zstd/lib/common/compiler.h +225 -63
  18. data/contrib/zstd/lib/common/cpu.h +37 -1
  19. data/contrib/zstd/lib/common/debug.c +7 -1
  20. data/contrib/zstd/lib/common/debug.h +21 -12
  21. data/contrib/zstd/lib/common/entropy_common.c +15 -37
  22. data/contrib/zstd/lib/common/error_private.c +9 -2
  23. data/contrib/zstd/lib/common/error_private.h +93 -5
  24. data/contrib/zstd/lib/common/fse.h +12 -87
  25. data/contrib/zstd/lib/common/fse_decompress.c +37 -117
  26. data/contrib/zstd/lib/common/huf.h +97 -172
  27. data/contrib/zstd/lib/common/mem.h +58 -58
  28. data/contrib/zstd/lib/common/pool.c +38 -17
  29. data/contrib/zstd/lib/common/pool.h +10 -4
  30. data/contrib/zstd/lib/common/portability_macros.h +158 -0
  31. data/contrib/zstd/lib/common/threading.c +74 -14
  32. data/contrib/zstd/lib/common/threading.h +5 -10
  33. data/contrib/zstd/lib/common/xxhash.c +6 -814
  34. data/contrib/zstd/lib/common/xxhash.h +6930 -195
  35. data/contrib/zstd/lib/common/zstd_common.c +1 -36
  36. data/contrib/zstd/lib/common/zstd_deps.h +1 -1
  37. data/contrib/zstd/lib/common/zstd_internal.h +68 -154
  38. data/contrib/zstd/lib/common/zstd_trace.h +163 -0
  39. data/contrib/zstd/lib/compress/clevels.h +134 -0
  40. data/contrib/zstd/lib/compress/fse_compress.c +75 -155
  41. data/contrib/zstd/lib/compress/hist.c +1 -1
  42. data/contrib/zstd/lib/compress/hist.h +1 -1
  43. data/contrib/zstd/lib/compress/huf_compress.c +810 -259
  44. data/contrib/zstd/lib/compress/zstd_compress.c +2864 -919
  45. data/contrib/zstd/lib/compress/zstd_compress_internal.h +523 -192
  46. data/contrib/zstd/lib/compress/zstd_compress_literals.c +117 -40
  47. data/contrib/zstd/lib/compress/zstd_compress_literals.h +16 -6
  48. data/contrib/zstd/lib/compress/zstd_compress_sequences.c +28 -19
  49. data/contrib/zstd/lib/compress/zstd_compress_sequences.h +1 -1
  50. data/contrib/zstd/lib/compress/zstd_compress_superblock.c +251 -412
  51. data/contrib/zstd/lib/compress/zstd_compress_superblock.h +1 -1
  52. data/contrib/zstd/lib/compress/zstd_cwksp.h +284 -97
  53. data/contrib/zstd/lib/compress/zstd_double_fast.c +382 -133
  54. data/contrib/zstd/lib/compress/zstd_double_fast.h +14 -2
  55. data/contrib/zstd/lib/compress/zstd_fast.c +732 -260
  56. data/contrib/zstd/lib/compress/zstd_fast.h +3 -2
  57. data/contrib/zstd/lib/compress/zstd_lazy.c +1177 -390
  58. data/contrib/zstd/lib/compress/zstd_lazy.h +129 -14
  59. data/contrib/zstd/lib/compress/zstd_ldm.c +280 -210
  60. data/contrib/zstd/lib/compress/zstd_ldm.h +3 -2
  61. data/contrib/zstd/lib/compress/zstd_ldm_geartab.h +106 -0
  62. data/contrib/zstd/lib/compress/zstd_opt.c +516 -285
  63. data/contrib/zstd/lib/compress/zstd_opt.h +32 -8
  64. data/contrib/zstd/lib/compress/zstdmt_compress.c +202 -131
  65. data/contrib/zstd/lib/compress/zstdmt_compress.h +9 -6
  66. data/contrib/zstd/lib/decompress/huf_decompress.c +1149 -555
  67. data/contrib/zstd/lib/decompress/huf_decompress_amd64.S +595 -0
  68. data/contrib/zstd/lib/decompress/zstd_ddict.c +4 -4
  69. data/contrib/zstd/lib/decompress/zstd_ddict.h +1 -1
  70. data/contrib/zstd/lib/decompress/zstd_decompress.c +583 -106
  71. data/contrib/zstd/lib/decompress/zstd_decompress_block.c +1054 -379
  72. data/contrib/zstd/lib/decompress/zstd_decompress_block.h +14 -3
  73. data/contrib/zstd/lib/decompress/zstd_decompress_internal.h +56 -6
  74. data/contrib/zstd/lib/deprecated/zbuff.h +1 -1
  75. data/contrib/zstd/lib/deprecated/zbuff_common.c +1 -1
  76. data/contrib/zstd/lib/deprecated/zbuff_compress.c +24 -4
  77. data/contrib/zstd/lib/deprecated/zbuff_decompress.c +3 -1
  78. data/contrib/zstd/lib/dictBuilder/cover.c +60 -44
  79. data/contrib/zstd/lib/dictBuilder/cover.h +6 -11
  80. data/contrib/zstd/lib/dictBuilder/divsufsort.c +1 -1
  81. data/contrib/zstd/lib/dictBuilder/fastcover.c +26 -18
  82. data/contrib/zstd/lib/dictBuilder/zdict.c +100 -101
  83. data/contrib/zstd/lib/legacy/zstd_legacy.h +38 -1
  84. data/contrib/zstd/lib/legacy/zstd_v01.c +18 -53
  85. data/contrib/zstd/lib/legacy/zstd_v01.h +1 -1
  86. data/contrib/zstd/lib/legacy/zstd_v02.c +28 -85
  87. data/contrib/zstd/lib/legacy/zstd_v02.h +1 -1
  88. data/contrib/zstd/lib/legacy/zstd_v03.c +29 -88
  89. data/contrib/zstd/lib/legacy/zstd_v03.h +1 -1
  90. data/contrib/zstd/lib/legacy/zstd_v04.c +27 -80
  91. data/contrib/zstd/lib/legacy/zstd_v04.h +1 -1
  92. data/contrib/zstd/lib/legacy/zstd_v05.c +36 -85
  93. data/contrib/zstd/lib/legacy/zstd_v05.h +1 -1
  94. data/contrib/zstd/lib/legacy/zstd_v06.c +44 -96
  95. data/contrib/zstd/lib/legacy/zstd_v06.h +1 -1
  96. data/contrib/zstd/lib/legacy/zstd_v07.c +37 -92
  97. data/contrib/zstd/lib/legacy/zstd_v07.h +1 -1
  98. data/contrib/zstd/lib/libzstd.mk +237 -0
  99. data/contrib/zstd/lib/libzstd.pc.in +4 -3
  100. data/contrib/zstd/lib/module.modulemap +35 -0
  101. data/contrib/zstd/lib/{dictBuilder/zdict.h → zdict.h} +202 -33
  102. data/contrib/zstd/lib/zstd.h +1030 -332
  103. data/contrib/zstd/lib/{common/zstd_errors.h → zstd_errors.h} +27 -8
  104. data/ext/extconf.rb +26 -7
  105. data/ext/extzstd.c +51 -24
  106. data/ext/extzstd.h +33 -6
  107. data/ext/extzstd_stream.c +74 -31
  108. data/ext/libzstd_conf.h +0 -1
  109. data/ext/zstd_decompress_asm.S +1 -0
  110. metadata +17 -7
  111. data/contrib/zstd/appveyor.yml +0 -292
  112. data/ext/depend +0 -2
@@ -1,7 +1,7 @@
1
1
  /* ******************************************************************
2
2
  * huff0 huffman codec,
3
3
  * part of Finite State Entropy library
4
- * Copyright (c) 2013-2020, Yann Collet, Facebook, Inc.
4
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
5
5
  *
6
6
  * You can contact the author at :
7
7
  * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
@@ -21,106 +21,29 @@ extern "C" {
21
21
 
22
22
  /* *** Dependencies *** */
23
23
  #include "zstd_deps.h" /* size_t */
24
-
25
-
26
- /* *** library symbols visibility *** */
27
- /* Note : when linking with -fvisibility=hidden on gcc, or by default on Visual,
28
- * HUF symbols remain "private" (internal symbols for library only).
29
- * Set macro FSE_DLL_EXPORT to 1 if you want HUF symbols visible on DLL interface */
30
- #if defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1) && defined(__GNUC__) && (__GNUC__ >= 4)
31
- # define HUF_PUBLIC_API __attribute__ ((visibility ("default")))
32
- #elif defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1) /* Visual expected */
33
- # define HUF_PUBLIC_API __declspec(dllexport)
34
- #elif defined(FSE_DLL_IMPORT) && (FSE_DLL_IMPORT==1)
35
- # define HUF_PUBLIC_API __declspec(dllimport) /* not required, just to generate faster code (saves a function pointer load from IAT and an indirect jump) */
36
- #else
37
- # define HUF_PUBLIC_API
38
- #endif
39
-
40
-
41
- /* ========================== */
42
- /* *** simple functions *** */
43
- /* ========================== */
44
-
45
- /** HUF_compress() :
46
- * Compress content from buffer 'src', of size 'srcSize', into buffer 'dst'.
47
- * 'dst' buffer must be already allocated.
48
- * Compression runs faster if `dstCapacity` >= HUF_compressBound(srcSize).
49
- * `srcSize` must be <= `HUF_BLOCKSIZE_MAX` == 128 KB.
50
- * @return : size of compressed data (<= `dstCapacity`).
51
- * Special values : if return == 0, srcData is not compressible => Nothing is stored within dst !!!
52
- * if HUF_isError(return), compression failed (more details using HUF_getErrorName())
53
- */
54
- HUF_PUBLIC_API size_t HUF_compress(void* dst, size_t dstCapacity,
55
- const void* src, size_t srcSize);
56
-
57
- /** HUF_decompress() :
58
- * Decompress HUF data from buffer 'cSrc', of size 'cSrcSize',
59
- * into already allocated buffer 'dst', of minimum size 'dstSize'.
60
- * `originalSize` : **must** be the ***exact*** size of original (uncompressed) data.
61
- * Note : in contrast with FSE, HUF_decompress can regenerate
62
- * RLE (cSrcSize==1) and uncompressed (cSrcSize==dstSize) data,
63
- * because it knows size to regenerate (originalSize).
64
- * @return : size of regenerated data (== originalSize),
65
- * or an error code, which can be tested using HUF_isError()
66
- */
67
- HUF_PUBLIC_API size_t HUF_decompress(void* dst, size_t originalSize,
68
- const void* cSrc, size_t cSrcSize);
24
+ #include "mem.h" /* U32 */
25
+ #define FSE_STATIC_LINKING_ONLY
26
+ #include "fse.h"
69
27
 
70
28
 
71
29
  /* *** Tool functions *** */
72
- #define HUF_BLOCKSIZE_MAX (128 * 1024) /**< maximum input size for a single block compressed with HUF_compress */
73
- HUF_PUBLIC_API size_t HUF_compressBound(size_t size); /**< maximum compressed size (worst case) */
30
+ #define HUF_BLOCKSIZE_MAX (128 * 1024) /**< maximum input size for a single block compressed with HUF_compress */
31
+ size_t HUF_compressBound(size_t size); /**< maximum compressed size (worst case) */
74
32
 
75
33
  /* Error Management */
76
- HUF_PUBLIC_API unsigned HUF_isError(size_t code); /**< tells if a return value is an error code */
77
- HUF_PUBLIC_API const char* HUF_getErrorName(size_t code); /**< provides error code string (useful for debugging) */
78
-
79
-
80
- /* *** Advanced function *** */
81
-
82
- /** HUF_compress2() :
83
- * Same as HUF_compress(), but offers control over `maxSymbolValue` and `tableLog`.
84
- * `maxSymbolValue` must be <= HUF_SYMBOLVALUE_MAX .
85
- * `tableLog` must be `<= HUF_TABLELOG_MAX` . */
86
- HUF_PUBLIC_API size_t HUF_compress2 (void* dst, size_t dstCapacity,
87
- const void* src, size_t srcSize,
88
- unsigned maxSymbolValue, unsigned tableLog);
89
-
90
- /** HUF_compress4X_wksp() :
91
- * Same as HUF_compress2(), but uses externally allocated `workSpace`.
92
- * `workspace` must have minimum alignment of 4, and be at least as large as HUF_WORKSPACE_SIZE */
93
- #define HUF_WORKSPACE_SIZE ((6 << 10) + 256)
94
- #define HUF_WORKSPACE_SIZE_U32 (HUF_WORKSPACE_SIZE / sizeof(U32))
95
- HUF_PUBLIC_API size_t HUF_compress4X_wksp (void* dst, size_t dstCapacity,
96
- const void* src, size_t srcSize,
97
- unsigned maxSymbolValue, unsigned tableLog,
98
- void* workSpace, size_t wkspSize);
99
-
100
- #endif /* HUF_H_298734234 */
101
-
102
- /* ******************************************************************
103
- * WARNING !!
104
- * The following section contains advanced and experimental definitions
105
- * which shall never be used in the context of a dynamic library,
106
- * because they are not guaranteed to remain stable in the future.
107
- * Only consider them in association with static linking.
108
- * *****************************************************************/
109
- #if defined(HUF_STATIC_LINKING_ONLY) && !defined(HUF_H_HUF_STATIC_LINKING_ONLY)
110
- #define HUF_H_HUF_STATIC_LINKING_ONLY
34
+ unsigned HUF_isError(size_t code); /**< tells if a return value is an error code */
35
+ const char* HUF_getErrorName(size_t code); /**< provides error code string (useful for debugging) */
111
36
 
112
- /* *** Dependencies *** */
113
- #include "mem.h" /* U32 */
114
- #define FSE_STATIC_LINKING_ONLY
115
- #include "fse.h"
116
37
 
38
+ #define HUF_WORKSPACE_SIZE ((8 << 10) + 512 /* sorting scratch space */)
39
+ #define HUF_WORKSPACE_SIZE_U64 (HUF_WORKSPACE_SIZE / sizeof(U64))
117
40
 
118
41
  /* *** Constants *** */
119
- #define HUF_TABLELOG_MAX 12 /* max runtime value of tableLog (due to static allocation); can be modified up to HUF_ABSOLUTEMAX_TABLELOG */
42
+ #define HUF_TABLELOG_MAX 12 /* max runtime value of tableLog (due to static allocation); can be modified up to HUF_TABLELOG_ABSOLUTEMAX */
120
43
  #define HUF_TABLELOG_DEFAULT 11 /* default tableLog value when none specified */
121
44
  #define HUF_SYMBOLVALUE_MAX 255
122
45
 
123
- #define HUF_TABLELOG_ABSOLUTEMAX 15 /* absolute limit of HUF_MAX_TABLELOG. Beyond that value, code does not work */
46
+ #define HUF_TABLELOG_ABSOLUTEMAX 12 /* absolute limit of HUF_MAX_TABLELOG. Beyond that value, code does not work */
124
47
  #if (HUF_TABLELOG_MAX > HUF_TABLELOG_ABSOLUTEMAX)
125
48
  # error "HUF_TABLELOG_MAX is too large !"
126
49
  #endif
@@ -136,15 +59,11 @@ HUF_PUBLIC_API size_t HUF_compress4X_wksp (void* dst, size_t dstCapacity,
136
59
 
137
60
  /* static allocation of HUF's Compression Table */
138
61
  /* this is a private definition, just exposed for allocation and strict aliasing purpose. never EVER access its members directly */
139
- struct HUF_CElt_s {
140
- U16 val;
141
- BYTE nbBits;
142
- }; /* typedef'd to HUF_CElt */
143
- typedef struct HUF_CElt_s HUF_CElt; /* consider it an incomplete type */
144
- #define HUF_CTABLE_SIZE_U32(maxSymbolValue) ((maxSymbolValue)+1) /* Use tables of U32, for proper alignment */
145
- #define HUF_CTABLE_SIZE(maxSymbolValue) (HUF_CTABLE_SIZE_U32(maxSymbolValue) * sizeof(U32))
62
+ typedef size_t HUF_CElt; /* consider it an incomplete type */
63
+ #define HUF_CTABLE_SIZE_ST(maxSymbolValue) ((maxSymbolValue)+2) /* Use tables of size_t, for proper alignment */
64
+ #define HUF_CTABLE_SIZE(maxSymbolValue) (HUF_CTABLE_SIZE_ST(maxSymbolValue) * sizeof(size_t))
146
65
  #define HUF_CREATE_STATIC_CTABLE(name, maxSymbolValue) \
147
- HUF_CElt name[HUF_CTABLE_SIZE_U32(maxSymbolValue)] /* no final ; */
66
+ HUF_CElt name[HUF_CTABLE_SIZE_ST(maxSymbolValue)] /* no final ; */
148
67
 
149
68
  /* static allocation of HUF's DTable */
150
69
  typedef U32 HUF_DTable;
@@ -158,25 +77,49 @@ typedef U32 HUF_DTable;
158
77
  /* ****************************************
159
78
  * Advanced decompression functions
160
79
  ******************************************/
161
- size_t HUF_decompress4X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< single-symbol decoder */
162
- #ifndef HUF_FORCE_DECOMPRESS_X1
163
- size_t HUF_decompress4X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< double-symbols decoder */
164
- #endif
165
80
 
166
- size_t HUF_decompress4X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< decodes RLE and uncompressed */
167
- 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 */
168
- 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 */
169
- size_t HUF_decompress4X1_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< single-symbol decoder */
170
- 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 */
171
- #ifndef HUF_FORCE_DECOMPRESS_X1
172
- size_t HUF_decompress4X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< double-symbols decoder */
173
- 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 */
174
- #endif
81
+ /**
82
+ * Huffman flags bitset.
83
+ * For all flags, 0 is the default value.
84
+ */
85
+ typedef enum {
86
+ /**
87
+ * If compiled with DYNAMIC_BMI2: Set flag only if the CPU supports BMI2 at runtime.
88
+ * Otherwise: Ignored.
89
+ */
90
+ HUF_flags_bmi2 = (1 << 0),
91
+ /**
92
+ * If set: Test possible table depths to find the one that produces the smallest header + encoded size.
93
+ * If unset: Use heuristic to find the table depth.
94
+ */
95
+ HUF_flags_optimalDepth = (1 << 1),
96
+ /**
97
+ * If set: If the previous table can encode the input, always reuse the previous table.
98
+ * If unset: If the previous table can encode the input, reuse the previous table if it results in a smaller output.
99
+ */
100
+ HUF_flags_preferRepeat = (1 << 2),
101
+ /**
102
+ * If set: Sample the input and check if the sample is uncompressible, if it is then don't attempt to compress.
103
+ * If unset: Always histogram the entire input.
104
+ */
105
+ HUF_flags_suspectUncompressible = (1 << 3),
106
+ /**
107
+ * If set: Don't use assembly implementations
108
+ * If unset: Allow using assembly implementations
109
+ */
110
+ HUF_flags_disableAsm = (1 << 4),
111
+ /**
112
+ * If set: Don't use the fast decoding loop, always use the fallback decoding loop.
113
+ * If unset: Use the fast decoding loop when possible.
114
+ */
115
+ HUF_flags_disableFast = (1 << 5)
116
+ } HUF_flags_e;
175
117
 
176
118
 
177
119
  /* ****************************************
178
120
  * HUF detailed API
179
121
  * ****************************************/
122
+ #define HUF_OPTIMAL_DEPTH_THRESHOLD ZSTD_btultra
180
123
 
181
124
  /*! HUF_compress() does the following:
182
125
  * 1. count symbol occurrence from source[] into table count[] using FSE_count() (exposed within "fse.h")
@@ -189,10 +132,12 @@ size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
189
132
  * For example, it's possible to compress several blocks using the same 'CTable',
190
133
  * or to save and regenerate 'CTable' using external methods.
191
134
  */
192
- unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue);
193
- size_t HUF_buildCTable (HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue, unsigned maxNbBits); /* @return : maxNbBits; CTable and count can overlap. In which case, CTable will overwrite count content */
194
- size_t HUF_writeCTable (void* dst, size_t maxDstSize, const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog);
195
- size_t HUF_compress4X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable);
135
+ unsigned HUF_minTableLog(unsigned symbolCardinality);
136
+ unsigned HUF_cardinality(const unsigned* count, unsigned maxSymbolValue);
137
+ unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, void* workSpace,
138
+ size_t wkspSize, HUF_CElt* table, const unsigned* count, int flags); /* table is used as scratch space for building and testing tables, not a return value */
139
+ size_t HUF_writeCTable_wksp(void* dst, size_t maxDstSize, const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog, void* workspace, size_t workspaceSize);
140
+ size_t HUF_compress4X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable, int flags);
196
141
  size_t HUF_estimateCompressedSize(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue);
197
142
  int HUF_validateCTable(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue);
198
143
 
@@ -201,22 +146,24 @@ typedef enum {
201
146
  HUF_repeat_check, /**< Can use the previous table but it must be checked. Note : The previous table must have been constructed by HUF_compress{1, 4}X_repeat */
202
147
  HUF_repeat_valid /**< Can use the previous table and it is assumed to be valid */
203
148
  } HUF_repeat;
149
+
204
150
  /** HUF_compress4X_repeat() :
205
151
  * Same as HUF_compress4X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
206
152
  * If it uses hufTable it does not modify hufTable or repeat.
207
153
  * If it doesn't, it sets *repeat = HUF_repeat_none, and it sets hufTable to the table used.
208
- * If preferRepeat then the old table will always be used if valid. */
154
+ * If preferRepeat then the old table will always be used if valid.
155
+ * If suspectUncompressible then some sampling checks will be run to potentially skip huffman coding */
209
156
  size_t HUF_compress4X_repeat(void* dst, size_t dstSize,
210
157
  const void* src, size_t srcSize,
211
158
  unsigned maxSymbolValue, unsigned tableLog,
212
159
  void* workSpace, size_t wkspSize, /**< `workSpace` must be aligned on 4-bytes boundaries, `wkspSize` must be >= HUF_WORKSPACE_SIZE */
213
- HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2);
160
+ HUF_CElt* hufTable, HUF_repeat* repeat, int flags);
214
161
 
215
162
  /** HUF_buildCTable_wksp() :
216
163
  * Same as HUF_buildCTable(), but using externally allocated scratch buffer.
217
164
  * `workSpace` must be aligned on 4-bytes boundaries, and its size must be >= HUF_CTABLE_WORKSPACE_SIZE.
218
165
  */
219
- #define HUF_CTABLE_WORKSPACE_SIZE_U32 (2*HUF_SYMBOLVALUE_MAX +1 +1)
166
+ #define HUF_CTABLE_WORKSPACE_SIZE_U32 ((4 * (HUF_SYMBOLVALUE_MAX + 1)) + 192)
220
167
  #define HUF_CTABLE_WORKSPACE_SIZE (HUF_CTABLE_WORKSPACE_SIZE_U32 * sizeof(unsigned))
221
168
  size_t HUF_buildCTable_wksp (HUF_CElt* tree,
222
169
  const unsigned* count, U32 maxSymbolValue, U32 maxNbBits,
@@ -242,17 +189,29 @@ size_t HUF_readStats_wksp(BYTE* huffWeight, size_t hwSize,
242
189
  U32* rankStats, U32* nbSymbolsPtr, U32* tableLogPtr,
243
190
  const void* src, size_t srcSize,
244
191
  void* workspace, size_t wkspSize,
245
- int bmi2);
192
+ int flags);
246
193
 
247
194
  /** HUF_readCTable() :
248
195
  * Loading a CTable saved with HUF_writeCTable() */
249
196
  size_t HUF_readCTable (HUF_CElt* CTable, unsigned* maxSymbolValuePtr, const void* src, size_t srcSize, unsigned *hasZeroWeights);
250
197
 
251
- /** HUF_getNbBits() :
198
+ /** HUF_getNbBitsFromCTable() :
252
199
  * Read nbBits from CTable symbolTable, for symbol `symbolValue` presumed <= HUF_SYMBOLVALUE_MAX
253
- * Note 1 : is not inlined, as HUF_CElt definition is private
254
- * Note 2 : const void* used, so that it can provide a statically allocated table as argument (which uses type U32) */
255
- U32 HUF_getNbBits(const void* symbolTable, U32 symbolValue);
200
+ * Note 1 : If symbolValue > HUF_readCTableHeader(symbolTable).maxSymbolValue, returns 0
201
+ * Note 2 : is not inlined, as HUF_CElt definition is private
202
+ */
203
+ U32 HUF_getNbBitsFromCTable(const HUF_CElt* symbolTable, U32 symbolValue);
204
+
205
+ typedef struct {
206
+ BYTE tableLog;
207
+ BYTE maxSymbolValue;
208
+ BYTE unused[sizeof(size_t) - 2];
209
+ } HUF_CTableHeader;
210
+
211
+ /** HUF_readCTableHeader() :
212
+ * @returns The header from the CTable specifying the tableLog and the maxSymbolValue.
213
+ */
214
+ HUF_CTableHeader HUF_readCTableHeader(HUF_CElt const* ctable);
256
215
 
257
216
  /*
258
217
  * HUF_decompress() does the following:
@@ -278,83 +237,49 @@ U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize);
278
237
  * a required workspace size greater than that specified in the following
279
238
  * macro.
280
239
  */
281
- #define HUF_DECOMPRESS_WORKSPACE_SIZE (2 << 10)
240
+ #define HUF_DECOMPRESS_WORKSPACE_SIZE ((2 << 10) + (1 << 9))
282
241
  #define HUF_DECOMPRESS_WORKSPACE_SIZE_U32 (HUF_DECOMPRESS_WORKSPACE_SIZE / sizeof(U32))
283
242
 
284
- #ifndef HUF_FORCE_DECOMPRESS_X2
285
- size_t HUF_readDTableX1 (HUF_DTable* DTable, const void* src, size_t srcSize);
286
- size_t HUF_readDTableX1_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
287
- #endif
288
- #ifndef HUF_FORCE_DECOMPRESS_X1
289
- size_t HUF_readDTableX2 (HUF_DTable* DTable, const void* src, size_t srcSize);
290
- size_t HUF_readDTableX2_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
291
- #endif
292
-
293
- size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
294
- #ifndef HUF_FORCE_DECOMPRESS_X2
295
- size_t HUF_decompress4X1_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
296
- #endif
297
- #ifndef HUF_FORCE_DECOMPRESS_X1
298
- size_t HUF_decompress4X2_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
299
- #endif
300
-
301
243
 
302
244
  /* ====================== */
303
245
  /* single stream variants */
304
246
  /* ====================== */
305
247
 
306
- size_t HUF_compress1X (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog);
307
- size_t HUF_compress1X_wksp (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize); /**< `workSpace` must be a table of at least HUF_WORKSPACE_SIZE_U32 unsigned */
308
- size_t HUF_compress1X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable);
248
+ size_t HUF_compress1X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable, int flags);
309
249
  /** HUF_compress1X_repeat() :
310
250
  * Same as HUF_compress1X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
311
251
  * If it uses hufTable it does not modify hufTable or repeat.
312
252
  * If it doesn't, it sets *repeat = HUF_repeat_none, and it sets hufTable to the table used.
313
- * If preferRepeat then the old table will always be used if valid. */
253
+ * If preferRepeat then the old table will always be used if valid.
254
+ * If suspectUncompressible then some sampling checks will be run to potentially skip huffman coding */
314
255
  size_t HUF_compress1X_repeat(void* dst, size_t dstSize,
315
256
  const void* src, size_t srcSize,
316
257
  unsigned maxSymbolValue, unsigned tableLog,
317
258
  void* workSpace, size_t wkspSize, /**< `workSpace` must be aligned on 4-bytes boundaries, `wkspSize` must be >= HUF_WORKSPACE_SIZE */
318
- HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2);
259
+ HUF_CElt* hufTable, HUF_repeat* repeat, int flags);
319
260
 
320
- size_t HUF_decompress1X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /* single-symbol decoder */
261
+ 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, int flags);
321
262
  #ifndef HUF_FORCE_DECOMPRESS_X1
322
- size_t HUF_decompress1X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /* double-symbol decoder */
323
- #endif
324
-
325
- size_t HUF_decompress1X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);
326
- 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);
327
- #ifndef HUF_FORCE_DECOMPRESS_X2
328
- size_t HUF_decompress1X1_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< single-symbol decoder */
329
- 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 */
330
- #endif
331
- #ifndef HUF_FORCE_DECOMPRESS_X1
332
- size_t HUF_decompress1X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< double-symbols decoder */
333
- 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 */
334
- #endif
335
-
336
- 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 */
337
- #ifndef HUF_FORCE_DECOMPRESS_X2
338
- size_t HUF_decompress1X1_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
339
- #endif
340
- #ifndef HUF_FORCE_DECOMPRESS_X1
341
- size_t HUF_decompress1X2_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
263
+ 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, int flags); /**< double-symbols decoder */
342
264
  #endif
343
265
 
344
266
  /* BMI2 variants.
345
267
  * If the CPU has BMI2 support, pass bmi2=1, otherwise pass bmi2=0.
346
268
  */
347
- size_t HUF_decompress1X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2);
269
+ size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int flags);
348
270
  #ifndef HUF_FORCE_DECOMPRESS_X2
349
- 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);
271
+ 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, int flags);
350
272
  #endif
351
- size_t HUF_decompress4X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2);
352
- 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);
273
+ size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int flags);
274
+ 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, int flags);
353
275
  #ifndef HUF_FORCE_DECOMPRESS_X2
354
- size_t HUF_readDTableX1_wksp_bmi2(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize, int bmi2);
276
+ size_t HUF_readDTableX1_wksp(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize, int flags);
277
+ #endif
278
+ #ifndef HUF_FORCE_DECOMPRESS_X1
279
+ size_t HUF_readDTableX2_wksp(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize, int flags);
355
280
  #endif
356
281
 
357
- #endif /* HUF_STATIC_LINKING_ONLY */
282
+ #endif /* HUF_H_298734234 */
358
283
 
359
284
  #if defined (__cplusplus)
360
285
  }
@@ -1,5 +1,5 @@
1
1
  /*
2
- * Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
2
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
3
3
  * All rights reserved.
4
4
  *
5
5
  * This source code is licensed under both the BSD-style license (found in the
@@ -31,15 +31,6 @@ extern "C" {
31
31
  # include <stdlib.h> /* _byteswap_ulong */
32
32
  # include <intrin.h> /* _byteswap_* */
33
33
  #endif
34
- #if defined(__GNUC__)
35
- # define MEM_STATIC static __inline __attribute__((unused))
36
- #elif defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */)
37
- # define MEM_STATIC static inline
38
- #elif defined(_MSC_VER)
39
- # define MEM_STATIC static __inline
40
- #else
41
- # define MEM_STATIC static /* this version may generate warnings for unused static functions; disable the relevant warning */
42
- #endif
43
34
 
44
35
  /*-**************************************************************
45
36
  * Basic Types
@@ -51,6 +42,8 @@ extern "C" {
51
42
  # include <stdint.h> /* intptr_t */
52
43
  # endif
53
44
  typedef uint8_t BYTE;
45
+ typedef uint8_t U8;
46
+ typedef int8_t S8;
54
47
  typedef uint16_t U16;
55
48
  typedef int16_t S16;
56
49
  typedef uint32_t U32;
@@ -63,6 +56,8 @@ extern "C" {
63
56
  # error "this implementation requires char to be exactly 8-bit type"
64
57
  #endif
65
58
  typedef unsigned char BYTE;
59
+ typedef unsigned char U8;
60
+ typedef signed char S8;
66
61
  #if USHRT_MAX != 65535
67
62
  # error "this implementation requires short to be exactly 16-bit type"
68
63
  #endif
@@ -129,23 +124,15 @@ MEM_STATIC size_t MEM_swapST(size_t in);
129
124
  /*-**************************************************************
130
125
  * Memory I/O Implementation
131
126
  *****************************************************************/
132
- /* MEM_FORCE_MEMORY_ACCESS :
133
- * By default, access to unaligned memory is controlled by `memcpy()`, which is safe and portable.
134
- * Unfortunately, on some target/compiler combinations, the generated assembly is sub-optimal.
135
- * The below switch allow to select different access method for improved performance.
136
- * Method 0 (default) : use `memcpy()`. Safe and portable.
137
- * Method 1 : `__packed` statement. It depends on compiler extension (i.e., not portable).
138
- * This method is safe if your compiler supports it, and *generally* as fast or faster than `memcpy`.
127
+ /* MEM_FORCE_MEMORY_ACCESS : For accessing unaligned memory:
128
+ * Method 0 : always use `memcpy()`. Safe and portable.
129
+ * Method 1 : Use compiler extension to set unaligned access.
139
130
  * Method 2 : direct access. This method is portable but violate C standard.
140
131
  * It can generate buggy code on targets depending on alignment.
141
- * In some circumstances, it's the only known way to get the most performance (i.e. GCC + ARMv6)
142
- * See http://fastcompression.blogspot.fr/2015/08/accessing-unaligned-memory.html for details.
143
- * Prefer these methods in priority order (0 > 1 > 2)
132
+ * Default : method 1 if supported, else method 0
144
133
  */
145
134
  #ifndef MEM_FORCE_MEMORY_ACCESS /* can be defined externally, on command line for example */
146
- # if defined(__GNUC__) && ( defined(__ARM_ARCH_6__) || defined(__ARM_ARCH_6J__) || defined(__ARM_ARCH_6K__) || defined(__ARM_ARCH_6Z__) || defined(__ARM_ARCH_6ZK__) || defined(__ARM_ARCH_6T2__) )
147
- # define MEM_FORCE_MEMORY_ACCESS 2
148
- # elif defined(__INTEL_COMPILER) || defined(__GNUC__) || defined(__ICCARM__)
135
+ # ifdef __GNUC__
149
136
  # define MEM_FORCE_MEMORY_ACCESS 1
150
137
  # endif
151
138
  #endif
@@ -155,8 +142,22 @@ MEM_STATIC unsigned MEM_64bits(void) { return sizeof(size_t)==8; }
155
142
 
156
143
  MEM_STATIC unsigned MEM_isLittleEndian(void)
157
144
  {
145
+ #if defined(__BYTE_ORDER__) && defined(__ORDER_LITTLE_ENDIAN__) && (__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__)
146
+ return 1;
147
+ #elif defined(__BYTE_ORDER__) && defined(__ORDER_BIG_ENDIAN__) && (__BYTE_ORDER__ == __ORDER_BIG_ENDIAN__)
148
+ return 0;
149
+ #elif defined(__clang__) && __LITTLE_ENDIAN__
150
+ return 1;
151
+ #elif defined(__clang__) && __BIG_ENDIAN__
152
+ return 0;
153
+ #elif defined(_MSC_VER) && (_M_AMD64 || _M_IX86)
154
+ return 1;
155
+ #elif defined(__DMC__) && defined(_M_IX86)
156
+ return 1;
157
+ #else
158
158
  const union { U32 u; BYTE c[4]; } one = { 1 }; /* don't use static : performance detrimental */
159
159
  return one.c[0];
160
+ #endif
160
161
  }
161
162
 
162
163
  #if defined(MEM_FORCE_MEMORY_ACCESS) && (MEM_FORCE_MEMORY_ACCESS==2)
@@ -174,30 +175,19 @@ MEM_STATIC void MEM_write64(void* memPtr, U64 value) { *(U64*)memPtr = value; }
174
175
 
175
176
  #elif defined(MEM_FORCE_MEMORY_ACCESS) && (MEM_FORCE_MEMORY_ACCESS==1)
176
177
 
177
- /* __pack instructions are safer, but compiler specific, hence potentially problematic for some compilers */
178
- /* currently only defined for gcc and icc */
179
- #if defined(_MSC_VER) || (defined(__INTEL_COMPILER) && defined(WIN32))
180
- __pragma( pack(push, 1) )
181
- typedef struct { U16 v; } unalign16;
182
- typedef struct { U32 v; } unalign32;
183
- typedef struct { U64 v; } unalign64;
184
- typedef struct { size_t v; } unalignArch;
185
- __pragma( pack(pop) )
186
- #else
187
- typedef struct { U16 v; } __attribute__((packed)) unalign16;
188
- typedef struct { U32 v; } __attribute__((packed)) unalign32;
189
- typedef struct { U64 v; } __attribute__((packed)) unalign64;
190
- typedef struct { size_t v; } __attribute__((packed)) unalignArch;
191
- #endif
178
+ typedef __attribute__((aligned(1))) U16 unalign16;
179
+ typedef __attribute__((aligned(1))) U32 unalign32;
180
+ typedef __attribute__((aligned(1))) U64 unalign64;
181
+ typedef __attribute__((aligned(1))) size_t unalignArch;
192
182
 
193
- MEM_STATIC U16 MEM_read16(const void* ptr) { return ((const unalign16*)ptr)->v; }
194
- MEM_STATIC U32 MEM_read32(const void* ptr) { return ((const unalign32*)ptr)->v; }
195
- MEM_STATIC U64 MEM_read64(const void* ptr) { return ((const unalign64*)ptr)->v; }
196
- MEM_STATIC size_t MEM_readST(const void* ptr) { return ((const unalignArch*)ptr)->v; }
183
+ MEM_STATIC U16 MEM_read16(const void* ptr) { return *(const unalign16*)ptr; }
184
+ MEM_STATIC U32 MEM_read32(const void* ptr) { return *(const unalign32*)ptr; }
185
+ MEM_STATIC U64 MEM_read64(const void* ptr) { return *(const unalign64*)ptr; }
186
+ MEM_STATIC size_t MEM_readST(const void* ptr) { return *(const unalignArch*)ptr; }
197
187
 
198
- MEM_STATIC void MEM_write16(void* memPtr, U16 value) { ((unalign16*)memPtr)->v = value; }
199
- MEM_STATIC void MEM_write32(void* memPtr, U32 value) { ((unalign32*)memPtr)->v = value; }
200
- MEM_STATIC void MEM_write64(void* memPtr, U64 value) { ((unalign64*)memPtr)->v = value; }
188
+ MEM_STATIC void MEM_write16(void* memPtr, U16 value) { *(unalign16*)memPtr = value; }
189
+ MEM_STATIC void MEM_write32(void* memPtr, U32 value) { *(unalign32*)memPtr = value; }
190
+ MEM_STATIC void MEM_write64(void* memPtr, U64 value) { *(unalign64*)memPtr = value; }
201
191
 
202
192
  #else
203
193
 
@@ -241,6 +231,14 @@ MEM_STATIC void MEM_write64(void* memPtr, U64 value)
241
231
 
242
232
  #endif /* MEM_FORCE_MEMORY_ACCESS */
243
233
 
234
+ MEM_STATIC U32 MEM_swap32_fallback(U32 in)
235
+ {
236
+ return ((in << 24) & 0xff000000 ) |
237
+ ((in << 8) & 0x00ff0000 ) |
238
+ ((in >> 8) & 0x0000ff00 ) |
239
+ ((in >> 24) & 0x000000ff );
240
+ }
241
+
244
242
  MEM_STATIC U32 MEM_swap32(U32 in)
245
243
  {
246
244
  #if defined(_MSC_VER) /* Visual Studio */
@@ -249,22 +247,13 @@ MEM_STATIC U32 MEM_swap32(U32 in)
249
247
  || (defined(__clang__) && __has_builtin(__builtin_bswap32))
250
248
  return __builtin_bswap32(in);
251
249
  #else
252
- return ((in << 24) & 0xff000000 ) |
253
- ((in << 8) & 0x00ff0000 ) |
254
- ((in >> 8) & 0x0000ff00 ) |
255
- ((in >> 24) & 0x000000ff );
250
+ return MEM_swap32_fallback(in);
256
251
  #endif
257
252
  }
258
253
 
259
- MEM_STATIC U64 MEM_swap64(U64 in)
254
+ MEM_STATIC U64 MEM_swap64_fallback(U64 in)
260
255
  {
261
- #if defined(_MSC_VER) /* Visual Studio */
262
- return _byteswap_uint64(in);
263
- #elif (defined (__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__ >= 403)) \
264
- || (defined(__clang__) && __has_builtin(__builtin_bswap64))
265
- return __builtin_bswap64(in);
266
- #else
267
- return ((in << 56) & 0xff00000000000000ULL) |
256
+ return ((in << 56) & 0xff00000000000000ULL) |
268
257
  ((in << 40) & 0x00ff000000000000ULL) |
269
258
  ((in << 24) & 0x0000ff0000000000ULL) |
270
259
  ((in << 8) & 0x000000ff00000000ULL) |
@@ -272,6 +261,17 @@ MEM_STATIC U64 MEM_swap64(U64 in)
272
261
  ((in >> 24) & 0x0000000000ff0000ULL) |
273
262
  ((in >> 40) & 0x000000000000ff00ULL) |
274
263
  ((in >> 56) & 0x00000000000000ffULL);
264
+ }
265
+
266
+ MEM_STATIC U64 MEM_swap64(U64 in)
267
+ {
268
+ #if defined(_MSC_VER) /* Visual Studio */
269
+ return _byteswap_uint64(in);
270
+ #elif (defined (__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__ >= 403)) \
271
+ || (defined(__clang__) && __has_builtin(__builtin_bswap64))
272
+ return __builtin_bswap64(in);
273
+ #else
274
+ return MEM_swap64_fallback(in);
275
275
  #endif
276
276
  }
277
277
 
@@ -308,7 +308,7 @@ MEM_STATIC void MEM_writeLE16(void* memPtr, U16 val)
308
308
 
309
309
  MEM_STATIC U32 MEM_readLE24(const void* memPtr)
310
310
  {
311
- return MEM_readLE16(memPtr) + (((const BYTE*)memPtr)[2] << 16);
311
+ return (U32)MEM_readLE16(memPtr) + ((U32)(((const BYTE*)memPtr)[2]) << 16);
312
312
  }
313
313
 
314
314
  MEM_STATIC void MEM_writeLE24(void* memPtr, U32 val)