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
@@ -1,5 +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
@@ -27,9 +27,11 @@ const char* ERR_getErrorString(ERR_enum code)
27
27
  case PREFIX(version_unsupported): return "Version not supported";
28
28
  case PREFIX(frameParameter_unsupported): return "Unsupported frame parameter";
29
29
  case PREFIX(frameParameter_windowTooLarge): return "Frame requires too much memory for decoding";
30
- case PREFIX(corruption_detected): return "Corrupted block detected";
30
+ case PREFIX(corruption_detected): return "Data corruption detected";
31
31
  case PREFIX(checksum_wrong): return "Restored data doesn't match checksum";
32
+ case PREFIX(literals_headerWrong): return "Header of Literals' block doesn't respect format specification";
32
33
  case PREFIX(parameter_unsupported): return "Unsupported parameter";
34
+ case PREFIX(parameter_combination_unsupported): return "Unsupported combination of parameters";
33
35
  case PREFIX(parameter_outOfBound): return "Parameter is out of bound";
34
36
  case PREFIX(init_missing): return "Context should be init first";
35
37
  case PREFIX(memory_allocation): return "Allocation error : not enough memory";
@@ -38,17 +40,22 @@ const char* ERR_getErrorString(ERR_enum code)
38
40
  case PREFIX(tableLog_tooLarge): return "tableLog requires too much memory : unsupported";
39
41
  case PREFIX(maxSymbolValue_tooLarge): return "Unsupported max Symbol Value : too large";
40
42
  case PREFIX(maxSymbolValue_tooSmall): return "Specified maxSymbolValue is too small";
43
+ case PREFIX(stabilityCondition_notRespected): return "pledged buffer stability condition is not respected";
41
44
  case PREFIX(dictionary_corrupted): return "Dictionary is corrupted";
42
45
  case PREFIX(dictionary_wrong): return "Dictionary mismatch";
43
46
  case PREFIX(dictionaryCreation_failed): return "Cannot create Dictionary from provided samples";
44
47
  case PREFIX(dstSize_tooSmall): return "Destination buffer is too small";
45
48
  case PREFIX(srcSize_wrong): return "Src size is incorrect";
46
49
  case PREFIX(dstBuffer_null): return "Operation on NULL destination buffer";
50
+ case PREFIX(noForwardProgress_destFull): return "Operation made no progress over multiple calls, due to output buffer being full";
51
+ case PREFIX(noForwardProgress_inputEmpty): return "Operation made no progress over multiple calls, due to input being empty";
47
52
  /* following error codes are not stable and may be removed or changed in a future version */
48
53
  case PREFIX(frameIndex_tooLarge): return "Frame index is too large";
49
54
  case PREFIX(seekableIO): return "An I/O error occurred when reading/seeking";
50
55
  case PREFIX(dstBuffer_wrong): return "Destination buffer is wrong";
51
56
  case PREFIX(srcBuffer_wrong): return "Source buffer is wrong";
57
+ case PREFIX(sequenceProducer_failed): return "Block-level external sequence producer returned an error code";
58
+ case PREFIX(externalSequences_invalid): return "External sequences are not valid";
52
59
  case PREFIX(maxCode):
53
60
  default: return notErrorCode;
54
61
  }
@@ -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
@@ -21,8 +21,10 @@ extern "C" {
21
21
  /* ****************************************
22
22
  * Dependencies
23
23
  ******************************************/
24
- #include "zstd_deps.h" /* size_t */
25
- #include "zstd_errors.h" /* enum list */
24
+ #include "../zstd_errors.h" /* enum list */
25
+ #include "compiler.h"
26
+ #include "debug.h"
27
+ #include "zstd_deps.h" /* size_t */
26
28
 
27
29
 
28
30
  /* ****************************************
@@ -58,8 +60,13 @@ ERR_STATIC unsigned ERR_isError(size_t code) { return (code > ERROR(maxCode)); }
58
60
  ERR_STATIC ERR_enum ERR_getErrorCode(size_t code) { if (!ERR_isError(code)) return (ERR_enum)0; return (ERR_enum) (0-code); }
59
61
 
60
62
  /* check and forward error code */
61
- #define CHECK_V_F(e, f) size_t const e = f; if (ERR_isError(e)) return e
62
- #define CHECK_F(f) { CHECK_V_F(_var_err__, f); }
63
+ #define CHECK_V_F(e, f) \
64
+ size_t const e = f; \
65
+ do { \
66
+ if (ERR_isError(e)) \
67
+ return e; \
68
+ } while (0)
69
+ #define CHECK_F(f) do { CHECK_V_F(_var_err__, f); } while (0)
63
70
 
64
71
 
65
72
  /*-****************************************
@@ -73,6 +80,87 @@ ERR_STATIC const char* ERR_getErrorName(size_t code)
73
80
  return ERR_getErrorString(ERR_getErrorCode(code));
74
81
  }
75
82
 
83
+ /**
84
+ * Ignore: this is an internal helper.
85
+ *
86
+ * This is a helper function to help force C99-correctness during compilation.
87
+ * Under strict compilation modes, variadic macro arguments can't be empty.
88
+ * However, variadic function arguments can be. Using a function therefore lets
89
+ * us statically check that at least one (string) argument was passed,
90
+ * independent of the compilation flags.
91
+ */
92
+ static INLINE_KEYWORD UNUSED_ATTR
93
+ void _force_has_format_string(const char *format, ...) {
94
+ (void)format;
95
+ }
96
+
97
+ /**
98
+ * Ignore: this is an internal helper.
99
+ *
100
+ * We want to force this function invocation to be syntactically correct, but
101
+ * we don't want to force runtime evaluation of its arguments.
102
+ */
103
+ #define _FORCE_HAS_FORMAT_STRING(...) \
104
+ do { \
105
+ if (0) { \
106
+ _force_has_format_string(__VA_ARGS__); \
107
+ } \
108
+ } while (0)
109
+
110
+ #define ERR_QUOTE(str) #str
111
+
112
+ /**
113
+ * Return the specified error if the condition evaluates to true.
114
+ *
115
+ * In debug modes, prints additional information.
116
+ * In order to do that (particularly, printing the conditional that failed),
117
+ * this can't just wrap RETURN_ERROR().
118
+ */
119
+ #define RETURN_ERROR_IF(cond, err, ...) \
120
+ do { \
121
+ if (cond) { \
122
+ RAWLOG(3, "%s:%d: ERROR!: check %s failed, returning %s", \
123
+ __FILE__, __LINE__, ERR_QUOTE(cond), ERR_QUOTE(ERROR(err))); \
124
+ _FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
125
+ RAWLOG(3, ": " __VA_ARGS__); \
126
+ RAWLOG(3, "\n"); \
127
+ return ERROR(err); \
128
+ } \
129
+ } while (0)
130
+
131
+ /**
132
+ * Unconditionally return the specified error.
133
+ *
134
+ * In debug modes, prints additional information.
135
+ */
136
+ #define RETURN_ERROR(err, ...) \
137
+ do { \
138
+ RAWLOG(3, "%s:%d: ERROR!: unconditional check failed, returning %s", \
139
+ __FILE__, __LINE__, ERR_QUOTE(ERROR(err))); \
140
+ _FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
141
+ RAWLOG(3, ": " __VA_ARGS__); \
142
+ RAWLOG(3, "\n"); \
143
+ return ERROR(err); \
144
+ } while(0)
145
+
146
+ /**
147
+ * If the provided expression evaluates to an error code, returns that error code.
148
+ *
149
+ * In debug modes, prints additional information.
150
+ */
151
+ #define FORWARD_IF_ERROR(err, ...) \
152
+ do { \
153
+ size_t const err_code = (err); \
154
+ if (ERR_isError(err_code)) { \
155
+ RAWLOG(3, "%s:%d: ERROR!: forwarding error in %s: %s", \
156
+ __FILE__, __LINE__, ERR_QUOTE(err), ERR_getErrorName(err_code)); \
157
+ _FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
158
+ RAWLOG(3, ": " __VA_ARGS__); \
159
+ RAWLOG(3, "\n"); \
160
+ return err_code; \
161
+ } \
162
+ } while(0)
163
+
76
164
  #if defined (__cplusplus)
77
165
  }
78
166
  #endif
@@ -1,7 +1,7 @@
1
1
  /* ******************************************************************
2
2
  * FSE : Finite State Entropy codec
3
3
  * Public Prototypes declaration
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
@@ -53,34 +53,6 @@ extern "C" {
53
53
  FSE_PUBLIC_API unsigned FSE_versionNumber(void); /**< library version number; to be used when checking dll version */
54
54
 
55
55
 
56
- /*-****************************************
57
- * FSE simple functions
58
- ******************************************/
59
- /*! FSE_compress() :
60
- Compress content of buffer 'src', of size 'srcSize', into destination buffer 'dst'.
61
- 'dst' buffer must be already allocated. Compression runs faster is dstCapacity >= FSE_compressBound(srcSize).
62
- @return : size of compressed data (<= dstCapacity).
63
- Special values : if return == 0, srcData is not compressible => Nothing is stored within dst !!!
64
- if return == 1, srcData is a single byte symbol * srcSize times. Use RLE compression instead.
65
- if FSE_isError(return), compression failed (more details using FSE_getErrorName())
66
- */
67
- FSE_PUBLIC_API size_t FSE_compress(void* dst, size_t dstCapacity,
68
- const void* src, size_t srcSize);
69
-
70
- /*! FSE_decompress():
71
- Decompress FSE data from buffer 'cSrc', of size 'cSrcSize',
72
- into already allocated destination buffer 'dst', of size 'dstCapacity'.
73
- @return : size of regenerated data (<= maxDstSize),
74
- or an error code, which can be tested using FSE_isError() .
75
-
76
- ** Important ** : FSE_decompress() does not decompress non-compressible nor RLE data !!!
77
- Why ? : making this distinction requires a header.
78
- Header management is intentionally delegated to the user layer, which can better manage special cases.
79
- */
80
- FSE_PUBLIC_API size_t FSE_decompress(void* dst, size_t dstCapacity,
81
- const void* cSrc, size_t cSrcSize);
82
-
83
-
84
56
  /*-*****************************************
85
57
  * Tool functions
86
58
  ******************************************/
@@ -91,20 +63,6 @@ FSE_PUBLIC_API unsigned FSE_isError(size_t code); /* tells if a return
91
63
  FSE_PUBLIC_API const char* FSE_getErrorName(size_t code); /* provides error code string (useful for debugging) */
92
64
 
93
65
 
94
- /*-*****************************************
95
- * FSE advanced functions
96
- ******************************************/
97
- /*! FSE_compress2() :
98
- Same as FSE_compress(), but allows the selection of 'maxSymbolValue' and 'tableLog'
99
- Both parameters can be defined as '0' to mean : use default value
100
- @return : size of compressed data
101
- Special values : if return == 0, srcData is not compressible => Nothing is stored within cSrc !!!
102
- if return == 1, srcData is a single byte symbol * srcSize times. Use RLE compression.
103
- if FSE_isError(return), it's an error code.
104
- */
105
- FSE_PUBLIC_API size_t FSE_compress2 (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog);
106
-
107
-
108
66
  /*-*****************************************
109
67
  * FSE detailed API
110
68
  ******************************************/
@@ -164,8 +122,6 @@ FSE_PUBLIC_API size_t FSE_writeNCount (void* buffer, size_t bufferSize,
164
122
  /*! Constructor and Destructor of FSE_CTable.
165
123
  Note that FSE_CTable size depends on 'tableLog' and 'maxSymbolValue' */
166
124
  typedef unsigned FSE_CTable; /* don't allocate that. It's only meant to be more restrictive than void* */
167
- FSE_PUBLIC_API FSE_CTable* FSE_createCTable (unsigned maxSymbolValue, unsigned tableLog);
168
- FSE_PUBLIC_API void FSE_freeCTable (FSE_CTable* ct);
169
125
 
170
126
  /*! FSE_buildCTable():
171
127
  Builds `ct`, which must be already allocated, using FSE_createCTable().
@@ -241,23 +197,7 @@ FSE_PUBLIC_API size_t FSE_readNCount_bmi2(short* normalizedCounter,
241
197
  unsigned* maxSymbolValuePtr, unsigned* tableLogPtr,
242
198
  const void* rBuffer, size_t rBuffSize, int bmi2);
243
199
 
244
- /*! Constructor and Destructor of FSE_DTable.
245
- Note that its size depends on 'tableLog' */
246
200
  typedef unsigned FSE_DTable; /* don't allocate that. It's just a way to be more restrictive than void* */
247
- FSE_PUBLIC_API FSE_DTable* FSE_createDTable(unsigned tableLog);
248
- FSE_PUBLIC_API void FSE_freeDTable(FSE_DTable* dt);
249
-
250
- /*! FSE_buildDTable():
251
- Builds 'dt', which must be already allocated, using FSE_createDTable().
252
- return : 0, or an errorCode, which can be tested using FSE_isError() */
253
- FSE_PUBLIC_API size_t FSE_buildDTable (FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog);
254
-
255
- /*! FSE_decompress_usingDTable():
256
- Decompress compressed source `cSrc` of size `cSrcSize` using `dt`
257
- into `dst` which must be already allocated.
258
- @return : size of regenerated data (necessarily <= `dstCapacity`),
259
- or an errorCode, which can be tested using FSE_isError() */
260
- FSE_PUBLIC_API size_t FSE_decompress_usingDTable(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, const FSE_DTable* dt);
261
201
 
262
202
  /*!
263
203
  Tutorial :
@@ -289,6 +229,7 @@ If there is an error, the function will return an error code, which can be teste
289
229
 
290
230
  #endif /* FSE_H */
291
231
 
232
+
292
233
  #if defined(FSE_STATIC_LINKING_ONLY) && !defined(FSE_H_FSE_STATIC_LINKING_ONLY)
293
234
  #define FSE_H_FSE_STATIC_LINKING_ONLY
294
235
 
@@ -320,24 +261,16 @@ If there is an error, the function will return an error code, which can be teste
320
261
  unsigned FSE_optimalTableLog_internal(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, unsigned minus);
321
262
  /**< same as FSE_optimalTableLog(), which used `minus==2` */
322
263
 
323
- /* FSE_compress_wksp() :
324
- * Same as FSE_compress2(), but using an externally allocated scratch buffer (`workSpace`).
325
- * FSE_COMPRESS_WKSP_SIZE_U32() provides the minimum size required for `workSpace` as a table of FSE_CTable.
326
- */
327
- #define FSE_COMPRESS_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) ( FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue) + ((maxTableLog > 12) ? (1 << (maxTableLog - 2)) : 1024) )
328
- size_t FSE_compress_wksp (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);
329
-
330
- size_t FSE_buildCTable_raw (FSE_CTable* ct, unsigned nbBits);
331
- /**< build a fake FSE_CTable, designed for a flat distribution, where each symbol uses nbBits */
332
-
333
264
  size_t FSE_buildCTable_rle (FSE_CTable* ct, unsigned char symbolValue);
334
265
  /**< build a fake FSE_CTable, designed to compress always the same symbolValue */
335
266
 
336
267
  /* FSE_buildCTable_wksp() :
337
268
  * Same as FSE_buildCTable(), but using an externally allocated scratch buffer (`workSpace`).
338
- * `wkspSize` must be >= `FSE_BUILD_CTABLE_WORKSPACE_SIZE(maxSymbolValue, tableLog)`.
269
+ * `wkspSize` must be >= `FSE_BUILD_CTABLE_WORKSPACE_SIZE_U32(maxSymbolValue, tableLog)` of `unsigned`.
270
+ * See FSE_buildCTable_wksp() for breakdown of workspace usage.
339
271
  */
340
- #define FSE_BUILD_CTABLE_WORKSPACE_SIZE(maxSymbolValue, tableLog) (sizeof(unsigned) * (maxSymbolValue + 2) + (1ull << tableLog))
272
+ #define FSE_BUILD_CTABLE_WORKSPACE_SIZE_U32(maxSymbolValue, tableLog) (((maxSymbolValue + 2) + (1ull << (tableLog)))/2 + sizeof(U64)/sizeof(U32) /* additional 8 bytes for potential table overwrite */)
273
+ #define FSE_BUILD_CTABLE_WORKSPACE_SIZE(maxSymbolValue, tableLog) (sizeof(unsigned) * FSE_BUILD_CTABLE_WORKSPACE_SIZE_U32(maxSymbolValue, tableLog))
341
274
  size_t FSE_buildCTable_wksp(FSE_CTable* ct, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);
342
275
 
343
276
  #define FSE_BUILD_DTABLE_WKSP_SIZE(maxTableLog, maxSymbolValue) (sizeof(short) * (maxSymbolValue + 1) + (1ULL << maxTableLog) + 8)
@@ -345,19 +278,11 @@ size_t FSE_buildCTable_wksp(FSE_CTable* ct, const short* normalizedCounter, unsi
345
278
  FSE_PUBLIC_API size_t FSE_buildDTable_wksp(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);
346
279
  /**< Same as FSE_buildDTable(), using an externally allocated `workspace` produced with `FSE_BUILD_DTABLE_WKSP_SIZE_U32(maxSymbolValue)` */
347
280
 
348
- size_t FSE_buildDTable_raw (FSE_DTable* dt, unsigned nbBits);
349
- /**< build a fake FSE_DTable, designed to read a flat distribution where each symbol uses nbBits */
350
-
351
- size_t FSE_buildDTable_rle (FSE_DTable* dt, unsigned char symbolValue);
352
- /**< build a fake FSE_DTable, designed to always generate the same symbolValue */
353
-
354
- #define FSE_DECOMPRESS_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) (FSE_DTABLE_SIZE_U32(maxTableLog) + FSE_BUILD_DTABLE_WKSP_SIZE_U32(maxTableLog, maxSymbolValue))
281
+ #define FSE_DECOMPRESS_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) (FSE_DTABLE_SIZE_U32(maxTableLog) + 1 + FSE_BUILD_DTABLE_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) + (FSE_MAX_SYMBOL_VALUE + 1) / 2 + 1)
355
282
  #define FSE_DECOMPRESS_WKSP_SIZE(maxTableLog, maxSymbolValue) (FSE_DECOMPRESS_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) * sizeof(unsigned))
356
- size_t FSE_decompress_wksp(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize);
357
- /**< same as FSE_decompress(), using an externally allocated `workSpace` produced with `FSE_DECOMPRESS_WKSP_SIZE_U32(maxLog, maxSymbolValue)` */
358
-
359
283
  size_t FSE_decompress_wksp_bmi2(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize, int bmi2);
360
- /**< Same as FSE_decompress_wksp() but with dynamic BMI2 support. Pass 1 if your CPU supports BMI2 or 0 if it doesn't. */
284
+ /**< same as FSE_decompress(), using an externally allocated `workSpace` produced with `FSE_DECOMPRESS_WKSP_SIZE_U32(maxLog, maxSymbolValue)`.
285
+ * Set bmi2 to 1 if your CPU supports BMI2 or 0 if it doesn't */
361
286
 
362
287
  typedef enum {
363
288
  FSE_repeat_none, /**< Cannot use the previous table */
@@ -540,20 +465,20 @@ MEM_STATIC void FSE_encodeSymbol(BIT_CStream_t* bitC, FSE_CState_t* statePtr, un
540
465
  FSE_symbolCompressionTransform const symbolTT = ((const FSE_symbolCompressionTransform*)(statePtr->symbolTT))[symbol];
541
466
  const U16* const stateTable = (const U16*)(statePtr->stateTable);
542
467
  U32 const nbBitsOut = (U32)((statePtr->value + symbolTT.deltaNbBits) >> 16);
543
- BIT_addBits(bitC, statePtr->value, nbBitsOut);
468
+ BIT_addBits(bitC, (size_t)statePtr->value, nbBitsOut);
544
469
  statePtr->value = stateTable[ (statePtr->value >> nbBitsOut) + symbolTT.deltaFindState];
545
470
  }
546
471
 
547
472
  MEM_STATIC void FSE_flushCState(BIT_CStream_t* bitC, const FSE_CState_t* statePtr)
548
473
  {
549
- BIT_addBits(bitC, statePtr->value, statePtr->stateLog);
474
+ BIT_addBits(bitC, (size_t)statePtr->value, statePtr->stateLog);
550
475
  BIT_flushBits(bitC);
551
476
  }
552
477
 
553
478
 
554
479
  /* FSE_getMaxNbBits() :
555
480
  * Approximate maximum cost of a symbol, in bits.
556
- * Fractional get rounded up (i.e : a symbol with a normalized frequency of 3 gives the same result as a frequency of 2)
481
+ * Fractional get rounded up (i.e. a symbol with a normalized frequency of 3 gives the same result as a frequency of 2)
557
482
  * note 1 : assume symbolValue is valid (<= maxSymbolValue)
558
483
  * note 2 : if freq[symbolValue]==0, @return a fake cost of tableLog+1 bits */
559
484
  MEM_STATIC U32 FSE_getMaxNbBits(const void* symbolTTPtr, U32 symbolValue)
@@ -1,6 +1,6 @@
1
1
  /* ******************************************************************
2
2
  * FSE : Finite State Entropy decoder
3
- * Copyright (c) 2013-2020, Yann Collet, Facebook, Inc.
3
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
4
4
  *
5
5
  * You can contact the author at :
6
6
  * - FSE source repository : https://github.com/Cyan4973/FiniteStateEntropy
@@ -22,8 +22,8 @@
22
22
  #define FSE_STATIC_LINKING_ONLY
23
23
  #include "fse.h"
24
24
  #include "error_private.h"
25
- #define ZSTD_DEPS_NEED_MALLOC
26
- #include "zstd_deps.h"
25
+ #include "zstd_deps.h" /* ZSTD_memcpy */
26
+ #include "bits.h" /* ZSTD_highbit32 */
27
27
 
28
28
 
29
29
  /* **************************************************************
@@ -55,19 +55,6 @@
55
55
  #define FSE_FUNCTION_NAME(X,Y) FSE_CAT(X,Y)
56
56
  #define FSE_TYPE_NAME(X,Y) FSE_CAT(X,Y)
57
57
 
58
-
59
- /* Function templates */
60
- FSE_DTable* FSE_createDTable (unsigned tableLog)
61
- {
62
- if (tableLog > FSE_TABLELOG_ABSOLUTE_MAX) tableLog = FSE_TABLELOG_ABSOLUTE_MAX;
63
- return (FSE_DTable*)ZSTD_malloc( FSE_DTABLE_SIZE_U32(tableLog) * sizeof (U32) );
64
- }
65
-
66
- void FSE_freeDTable (FSE_DTable* dt)
67
- {
68
- ZSTD_free(dt);
69
- }
70
-
71
58
  static size_t FSE_buildDTable_internal(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize)
72
59
  {
73
60
  void* const tdPtr = dt+1; /* because *dt is unsigned, 32-bits aligned on 32-bits */
@@ -96,7 +83,7 @@ static size_t FSE_buildDTable_internal(FSE_DTable* dt, const short* normalizedCo
96
83
  symbolNext[s] = 1;
97
84
  } else {
98
85
  if (normalizedCounter[s] >= largeLimit) DTableH.fastMode=0;
99
- symbolNext[s] = normalizedCounter[s];
86
+ symbolNext[s] = (U16)normalizedCounter[s];
100
87
  } } }
101
88
  ZSTD_memcpy(dt, &DTableH, sizeof(DTableH));
102
89
  }
@@ -111,8 +98,7 @@ static size_t FSE_buildDTable_internal(FSE_DTable* dt, const short* normalizedCo
111
98
  * all symbols have counts <= 8. We ensure we have 8 bytes at the end of
112
99
  * our buffer to handle the over-write.
113
100
  */
114
- {
115
- U64 const add = 0x0101010101010101ull;
101
+ { U64 const add = 0x0101010101010101ull;
116
102
  size_t pos = 0;
117
103
  U64 sv = 0;
118
104
  U32 s;
@@ -123,14 +109,13 @@ static size_t FSE_buildDTable_internal(FSE_DTable* dt, const short* normalizedCo
123
109
  for (i = 8; i < n; i += 8) {
124
110
  MEM_write64(spread + pos + i, sv);
125
111
  }
126
- pos += n;
127
- }
128
- }
112
+ pos += (size_t)n;
113
+ } }
129
114
  /* Now we spread those positions across the table.
130
- * The benefit of doing it in two stages is that we avoid the the
115
+ * The benefit of doing it in two stages is that we avoid the
131
116
  * variable size inner loop, which caused lots of branch misses.
132
117
  * Now we can run through all the positions without any branch misses.
133
- * We unroll the loop twice, since that is what emperically worked best.
118
+ * We unroll the loop twice, since that is what empirically worked best.
134
119
  */
135
120
  {
136
121
  size_t position = 0;
@@ -166,7 +151,7 @@ static size_t FSE_buildDTable_internal(FSE_DTable* dt, const short* normalizedCo
166
151
  for (u=0; u<tableSize; u++) {
167
152
  FSE_FUNCTION_TYPE const symbol = (FSE_FUNCTION_TYPE)(tableDecode[u].symbol);
168
153
  U32 const nextState = symbolNext[symbol]++;
169
- tableDecode[u].nbBits = (BYTE) (tableLog - BIT_highbit32(nextState) );
154
+ tableDecode[u].nbBits = (BYTE) (tableLog - ZSTD_highbit32(nextState) );
170
155
  tableDecode[u].newState = (U16) ( (nextState << tableDecode[u].nbBits) - tableSize);
171
156
  } }
172
157
 
@@ -184,49 +169,6 @@ size_t FSE_buildDTable_wksp(FSE_DTable* dt, const short* normalizedCounter, unsi
184
169
  /*-*******************************************************
185
170
  * Decompression (Byte symbols)
186
171
  *********************************************************/
187
- size_t FSE_buildDTable_rle (FSE_DTable* dt, BYTE symbolValue)
188
- {
189
- void* ptr = dt;
190
- FSE_DTableHeader* const DTableH = (FSE_DTableHeader*)ptr;
191
- void* dPtr = dt + 1;
192
- FSE_decode_t* const cell = (FSE_decode_t*)dPtr;
193
-
194
- DTableH->tableLog = 0;
195
- DTableH->fastMode = 0;
196
-
197
- cell->newState = 0;
198
- cell->symbol = symbolValue;
199
- cell->nbBits = 0;
200
-
201
- return 0;
202
- }
203
-
204
-
205
- size_t FSE_buildDTable_raw (FSE_DTable* dt, unsigned nbBits)
206
- {
207
- void* ptr = dt;
208
- FSE_DTableHeader* const DTableH = (FSE_DTableHeader*)ptr;
209
- void* dPtr = dt + 1;
210
- FSE_decode_t* const dinfo = (FSE_decode_t*)dPtr;
211
- const unsigned tableSize = 1 << nbBits;
212
- const unsigned tableMask = tableSize - 1;
213
- const unsigned maxSV1 = tableMask+1;
214
- unsigned s;
215
-
216
- /* Sanity checks */
217
- if (nbBits < 1) return ERROR(GENERIC); /* min size */
218
-
219
- /* Build Decoding Table */
220
- DTableH->tableLog = (U16)nbBits;
221
- DTableH->fastMode = 1;
222
- for (s=0; s<maxSV1; s++) {
223
- dinfo[s].newState = 0;
224
- dinfo[s].symbol = (BYTE)s;
225
- dinfo[s].nbBits = (BYTE)nbBits;
226
- }
227
-
228
- return 0;
229
- }
230
172
 
231
173
  FORCE_INLINE_TEMPLATE size_t FSE_decompress_usingDTable_generic(
232
174
  void* dst, size_t maxDstSize,
@@ -287,29 +229,15 @@ FORCE_INLINE_TEMPLATE size_t FSE_decompress_usingDTable_generic(
287
229
  break;
288
230
  } }
289
231
 
290
- return op-ostart;
232
+ assert(op >= ostart);
233
+ return (size_t)(op-ostart);
291
234
  }
292
235
 
293
-
294
- size_t FSE_decompress_usingDTable(void* dst, size_t originalSize,
295
- const void* cSrc, size_t cSrcSize,
296
- const FSE_DTable* dt)
297
- {
298
- const void* ptr = dt;
299
- const FSE_DTableHeader* DTableH = (const FSE_DTableHeader*)ptr;
300
- const U32 fastMode = DTableH->fastMode;
301
-
302
- /* select fast mode (static) */
303
- if (fastMode) return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 1);
304
- return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 0);
305
- }
236
+ typedef struct {
237
+ short ncount[FSE_MAX_SYMBOL_VALUE + 1];
238
+ } FSE_DecompressWksp;
306
239
 
307
240
 
308
- size_t FSE_decompress_wksp(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize)
309
- {
310
- return FSE_decompress_wksp_bmi2(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize, /* bmi2 */ 0);
311
- }
312
-
313
241
  FORCE_INLINE_TEMPLATE size_t FSE_decompress_wksp_body(
314
242
  void* dst, size_t dstCapacity,
315
243
  const void* cSrc, size_t cSrcSize,
@@ -318,24 +246,34 @@ FORCE_INLINE_TEMPLATE size_t FSE_decompress_wksp_body(
318
246
  {
319
247
  const BYTE* const istart = (const BYTE*)cSrc;
320
248
  const BYTE* ip = istart;
321
- short counting[FSE_MAX_SYMBOL_VALUE+1];
322
249
  unsigned tableLog;
323
250
  unsigned maxSymbolValue = FSE_MAX_SYMBOL_VALUE;
324
- FSE_DTable* const dtable = (FSE_DTable*)workSpace;
251
+ FSE_DecompressWksp* const wksp = (FSE_DecompressWksp*)workSpace;
252
+ size_t const dtablePos = sizeof(FSE_DecompressWksp) / sizeof(FSE_DTable);
253
+ FSE_DTable* const dtable = (FSE_DTable*)workSpace + dtablePos;
254
+
255
+ FSE_STATIC_ASSERT((FSE_MAX_SYMBOL_VALUE + 1) % 2 == 0);
256
+ if (wkspSize < sizeof(*wksp)) return ERROR(GENERIC);
257
+
258
+ /* correct offset to dtable depends on this property */
259
+ FSE_STATIC_ASSERT(sizeof(FSE_DecompressWksp) % sizeof(FSE_DTable) == 0);
325
260
 
326
261
  /* normal FSE decoding mode */
327
- size_t const NCountLength = FSE_readNCount_bmi2(counting, &maxSymbolValue, &tableLog, istart, cSrcSize, bmi2);
328
- if (FSE_isError(NCountLength)) return NCountLength;
329
- if (tableLog > maxLog) return ERROR(tableLog_tooLarge);
330
- assert(NCountLength <= cSrcSize);
331
- ip += NCountLength;
332
- cSrcSize -= NCountLength;
262
+ { size_t const NCountLength =
263
+ FSE_readNCount_bmi2(wksp->ncount, &maxSymbolValue, &tableLog, istart, cSrcSize, bmi2);
264
+ if (FSE_isError(NCountLength)) return NCountLength;
265
+ if (tableLog > maxLog) return ERROR(tableLog_tooLarge);
266
+ assert(NCountLength <= cSrcSize);
267
+ ip += NCountLength;
268
+ cSrcSize -= NCountLength;
269
+ }
333
270
 
334
271
  if (FSE_DECOMPRESS_WKSP_SIZE(tableLog, maxSymbolValue) > wkspSize) return ERROR(tableLog_tooLarge);
335
- workSpace = dtable + FSE_DTABLE_SIZE_U32(tableLog);
336
- wkspSize -= FSE_DTABLE_SIZE(tableLog);
272
+ assert(sizeof(*wksp) + FSE_DTABLE_SIZE(tableLog) <= wkspSize);
273
+ workSpace = (BYTE*)workSpace + sizeof(*wksp) + FSE_DTABLE_SIZE(tableLog);
274
+ wkspSize -= sizeof(*wksp) + FSE_DTABLE_SIZE(tableLog);
337
275
 
338
- CHECK_F( FSE_buildDTable_internal(dtable, counting, maxSymbolValue, tableLog, workSpace, wkspSize) );
276
+ CHECK_F( FSE_buildDTable_internal(dtable, wksp->ncount, maxSymbolValue, tableLog, workSpace, wkspSize) );
339
277
 
340
278
  {
341
279
  const void* ptr = dtable;
@@ -355,7 +293,7 @@ static size_t FSE_decompress_wksp_body_default(void* dst, size_t dstCapacity, co
355
293
  }
356
294
 
357
295
  #if DYNAMIC_BMI2
358
- TARGET_ATTRIBUTE("bmi2") static size_t FSE_decompress_wksp_body_bmi2(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize)
296
+ BMI2_TARGET_ATTRIBUTE static size_t FSE_decompress_wksp_body_bmi2(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize)
359
297
  {
360
298
  return FSE_decompress_wksp_body(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize, 1);
361
299
  }
@@ -372,22 +310,4 @@ size_t FSE_decompress_wksp_bmi2(void* dst, size_t dstCapacity, const void* cSrc,
372
310
  return FSE_decompress_wksp_body_default(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize);
373
311
  }
374
312
 
375
-
376
- typedef FSE_DTable DTable_max_t[FSE_DTABLE_SIZE_U32(FSE_MAX_TABLELOG)];
377
-
378
- #ifndef ZSTD_NO_UNUSED_FUNCTIONS
379
- size_t FSE_buildDTable(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog) {
380
- U32 wksp[FSE_BUILD_DTABLE_WKSP_SIZE_U32(FSE_TABLELOG_ABSOLUTE_MAX, FSE_MAX_SYMBOL_VALUE)];
381
- return FSE_buildDTable_wksp(dt, normalizedCounter, maxSymbolValue, tableLog, wksp, sizeof(wksp));
382
- }
383
-
384
- size_t FSE_decompress(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize)
385
- {
386
- /* Static analyzer seems unable to understand this table will be properly initialized later */
387
- U32 wksp[FSE_DECOMPRESS_WKSP_SIZE_U32(FSE_MAX_TABLELOG, FSE_MAX_SYMBOL_VALUE)];
388
- return FSE_decompress_wksp(dst, dstCapacity, cSrc, cSrcSize, FSE_MAX_TABLELOG, wksp, sizeof(wksp));
389
- }
390
- #endif
391
-
392
-
393
313
  #endif /* FSE_COMMONDEFS_ONLY */