zstdlib 0.3.0-x86-mingw32 → 0.8.0-x86-mingw32
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.
- checksums.yaml +4 -4
- data/CHANGES.md +30 -1
- data/README.md +2 -2
- data/Rakefile +1 -1
- data/ext/zstdlib/extconf.rb +3 -3
- data/ext/zstdlib/ruby/zlib-2.7/zstdlib.c +4895 -0
- data/ext/zstdlib/ruby/zlib-3.0/zstdlib.c +4994 -0
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/bitstream.h +59 -51
- data/ext/zstdlib/zstd-1.5.0/lib/common/compiler.h +289 -0
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/cpu.h +1 -3
- data/ext/zstdlib/zstd-1.5.0/lib/common/debug.c +24 -0
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/debug.h +22 -49
- data/ext/zstdlib/zstd-1.5.0/lib/common/entropy_common.c +362 -0
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/error_private.c +3 -1
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/error_private.h +8 -4
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/fse.h +50 -42
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/fse_decompress.c +149 -55
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/huf.h +43 -39
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/mem.h +69 -25
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/pool.c +30 -20
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/pool.h +3 -3
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/threading.c +51 -4
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/threading.h +36 -4
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/xxhash.c +40 -92
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/xxhash.h +12 -32
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/common/zstd_common.c +10 -10
- data/ext/zstdlib/zstd-1.5.0/lib/common/zstd_deps.h +111 -0
- data/ext/zstdlib/zstd-1.5.0/lib/common/zstd_internal.h +490 -0
- data/ext/zstdlib/zstd-1.5.0/lib/common/zstd_trace.h +154 -0
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/fse_compress.c +47 -63
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/hist.c +41 -63
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/hist.h +13 -33
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/huf_compress.c +332 -193
- data/ext/zstdlib/zstd-1.5.0/lib/compress/zstd_compress.c +6393 -0
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/zstd_compress_internal.h +522 -86
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/zstd_compress_literals.c +25 -16
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/zstd_compress_literals.h +2 -2
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/zstd_compress_sequences.c +50 -24
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/zstd_compress_sequences.h +11 -4
- data/ext/zstdlib/zstd-1.5.0/lib/compress/zstd_compress_superblock.c +572 -0
- data/ext/zstdlib/zstd-1.5.0/lib/compress/zstd_compress_superblock.h +32 -0
- data/ext/zstdlib/zstd-1.5.0/lib/compress/zstd_cwksp.h +662 -0
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/zstd_double_fast.c +43 -41
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/zstd_double_fast.h +2 -2
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/zstd_fast.c +85 -80
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/zstd_fast.h +2 -2
- data/ext/zstdlib/zstd-1.5.0/lib/compress/zstd_lazy.c +2184 -0
- data/ext/zstdlib/zstd-1.5.0/lib/compress/zstd_lazy.h +125 -0
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/zstd_ldm.c +333 -208
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/zstd_ldm.h +15 -3
- data/ext/zstdlib/zstd-1.5.0/lib/compress/zstd_ldm_geartab.h +103 -0
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/zstd_opt.c +228 -129
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/zstd_opt.h +1 -1
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/compress/zstdmt_compress.c +151 -440
- data/ext/zstdlib/zstd-1.5.0/lib/compress/zstdmt_compress.h +110 -0
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/decompress/huf_decompress.c +395 -276
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/decompress/zstd_ddict.c +20 -16
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/decompress/zstd_ddict.h +3 -3
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/decompress/zstd_decompress.c +628 -231
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/decompress/zstd_decompress_block.c +606 -380
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/decompress/zstd_decompress_block.h +8 -5
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/decompress/zstd_decompress_internal.h +39 -9
- data/ext/zstdlib/zstd-1.5.0/lib/zdict.h +452 -0
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/lib/zstd.h +740 -153
- data/ext/zstdlib/{zstd-1.4.2/lib/common → zstd-1.5.0/lib}/zstd_errors.h +3 -1
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/zlibWrapper/gzclose.c +1 -1
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/zlibWrapper/gzcompatibility.h +1 -1
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/zlibWrapper/gzguts.h +0 -0
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/zlibWrapper/gzlib.c +9 -9
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/zlibWrapper/gzread.c +16 -8
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/zlibWrapper/gzwrite.c +8 -8
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/zlibWrapper/zstd_zlibwrapper.c +131 -45
- data/ext/zstdlib/{zstd-1.4.2 → zstd-1.5.0}/zlibWrapper/zstd_zlibwrapper.h +1 -1
- data/lib/2.2/zstdlib.so +0 -0
- data/lib/2.3/zstdlib.so +0 -0
- data/lib/2.4/zstdlib.so +0 -0
- data/lib/2.5/zstdlib.so +0 -0
- data/lib/2.6/zstdlib.so +0 -0
- data/lib/2.7/zstdlib.so +0 -0
- metadata +76 -67
- data/ext/zstdlib/zstd-1.4.2/lib/common/compiler.h +0 -147
- data/ext/zstdlib/zstd-1.4.2/lib/common/debug.c +0 -44
- data/ext/zstdlib/zstd-1.4.2/lib/common/entropy_common.c +0 -236
- data/ext/zstdlib/zstd-1.4.2/lib/common/zstd_internal.h +0 -371
- data/ext/zstdlib/zstd-1.4.2/lib/compress/zstd_compress.c +0 -3904
- data/ext/zstdlib/zstd-1.4.2/lib/compress/zstd_lazy.c +0 -1111
- data/ext/zstdlib/zstd-1.4.2/lib/compress/zstd_lazy.h +0 -67
- data/ext/zstdlib/zstd-1.4.2/lib/compress/zstdmt_compress.h +0 -192
@@ -1,35 +1,15 @@
|
|
1
1
|
/* ******************************************************************
|
2
|
-
|
3
|
-
|
4
|
-
|
5
|
-
|
6
|
-
|
7
|
-
|
8
|
-
|
9
|
-
|
10
|
-
|
11
|
-
|
12
|
-
|
13
|
-
notice, this list of conditions and the following disclaimer.
|
14
|
-
* Redistributions in binary form must reproduce the above
|
15
|
-
copyright notice, this list of conditions and the following disclaimer
|
16
|
-
in the documentation and/or other materials provided with the
|
17
|
-
distribution.
|
18
|
-
|
19
|
-
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
|
20
|
-
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
|
21
|
-
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
|
22
|
-
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
|
23
|
-
OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
|
24
|
-
SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
|
25
|
-
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
|
26
|
-
DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
|
27
|
-
THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
28
|
-
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
29
|
-
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
30
|
-
|
31
|
-
You can contact the author at :
|
32
|
-
- Source repository : https://github.com/Cyan4973/FiniteStateEntropy
|
2
|
+
* bitstream
|
3
|
+
* Part of FSE library
|
4
|
+
* Copyright (c) Yann Collet, Facebook, Inc.
|
5
|
+
*
|
6
|
+
* You can contact the author at :
|
7
|
+
* - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
|
8
|
+
*
|
9
|
+
* This source code is licensed under both the BSD-style license (found in the
|
10
|
+
* LICENSE file in the root directory of this source tree) and the GPLv2 (found
|
11
|
+
* in the COPYING file in the root directory of this source tree).
|
12
|
+
* You may select, at your option, one of the above-listed licenses.
|
33
13
|
****************************************************************** */
|
34
14
|
#ifndef BITSTREAM_H_MODULE
|
35
15
|
#define BITSTREAM_H_MODULE
|
@@ -37,7 +17,6 @@
|
|
37
17
|
#if defined (__cplusplus)
|
38
18
|
extern "C" {
|
39
19
|
#endif
|
40
|
-
|
41
20
|
/*
|
42
21
|
* This API consists of small unitary functions, which must be inlined for best performance.
|
43
22
|
* Since link-time-optimization is not available for all compilers,
|
@@ -48,6 +27,7 @@ extern "C" {
|
|
48
27
|
* Dependencies
|
49
28
|
******************************************/
|
50
29
|
#include "mem.h" /* unaligned access routines */
|
30
|
+
#include "compiler.h" /* UNLIKELY() */
|
51
31
|
#include "debug.h" /* assert(), DEBUGLOG(), RAWLOG() */
|
52
32
|
#include "error_private.h" /* error codes and messages */
|
53
33
|
|
@@ -55,8 +35,12 @@ extern "C" {
|
|
55
35
|
/*=========================================
|
56
36
|
* Target specific
|
57
37
|
=========================================*/
|
58
|
-
#
|
59
|
-
#
|
38
|
+
#ifndef ZSTD_NO_INTRINSICS
|
39
|
+
# if defined(__BMI__) && defined(__GNUC__)
|
40
|
+
# include <immintrin.h> /* support for bextr (experimental) */
|
41
|
+
# elif defined(__ICCARM__)
|
42
|
+
# include <intrinsics.h>
|
43
|
+
# endif
|
60
44
|
#endif
|
61
45
|
|
62
46
|
#define STREAM_ACCUMULATOR_MIN_32 25
|
@@ -158,11 +142,16 @@ MEM_STATIC unsigned BIT_highbit32 (U32 val)
|
|
158
142
|
assert(val != 0);
|
159
143
|
{
|
160
144
|
# if defined(_MSC_VER) /* Visual */
|
161
|
-
|
162
|
-
|
163
|
-
|
145
|
+
# if STATIC_BMI2 == 1
|
146
|
+
return _lzcnt_u32(val) ^ 31;
|
147
|
+
# else
|
148
|
+
unsigned long r = 0;
|
149
|
+
return _BitScanReverse(&r, val) ? (unsigned)r : 0;
|
150
|
+
# endif
|
164
151
|
# elif defined(__GNUC__) && (__GNUC__ >= 3) /* Use GCC Intrinsic */
|
165
|
-
return
|
152
|
+
return __builtin_clz (val) ^ 31;
|
153
|
+
# elif defined(__ICCARM__) /* IAR Intrinsic */
|
154
|
+
return 31 - __CLZ(val);
|
166
155
|
# else /* Software version */
|
167
156
|
static const unsigned DeBruijnClz[32] = { 0, 9, 1, 10, 13, 21, 2, 29,
|
168
157
|
11, 14, 16, 18, 22, 25, 3, 30,
|
@@ -214,7 +203,7 @@ MEM_STATIC size_t BIT_initCStream(BIT_CStream_t* bitC,
|
|
214
203
|
MEM_STATIC void BIT_addBits(BIT_CStream_t* bitC,
|
215
204
|
size_t value, unsigned nbBits)
|
216
205
|
{
|
217
|
-
|
206
|
+
DEBUG_STATIC_ASSERT(BIT_MASK_SIZE == 32);
|
218
207
|
assert(nbBits < BIT_MASK_SIZE);
|
219
208
|
assert(nbBits + bitC->bitPos < sizeof(bitC->bitContainer) * 8);
|
220
209
|
bitC->bitContainer |= (value & BIT_mask[nbBits]) << bitC->bitPos;
|
@@ -240,9 +229,9 @@ MEM_STATIC void BIT_flushBitsFast(BIT_CStream_t* bitC)
|
|
240
229
|
{
|
241
230
|
size_t const nbBytes = bitC->bitPos >> 3;
|
242
231
|
assert(bitC->bitPos < sizeof(bitC->bitContainer) * 8);
|
232
|
+
assert(bitC->ptr <= bitC->endPtr);
|
243
233
|
MEM_writeLEST(bitC->ptr, bitC->bitContainer);
|
244
234
|
bitC->ptr += nbBytes;
|
245
|
-
assert(bitC->ptr <= bitC->endPtr);
|
246
235
|
bitC->bitPos &= 7;
|
247
236
|
bitC->bitContainer >>= nbBytes*8;
|
248
237
|
}
|
@@ -256,6 +245,7 @@ MEM_STATIC void BIT_flushBits(BIT_CStream_t* bitC)
|
|
256
245
|
{
|
257
246
|
size_t const nbBytes = bitC->bitPos >> 3;
|
258
247
|
assert(bitC->bitPos < sizeof(bitC->bitContainer) * 8);
|
248
|
+
assert(bitC->ptr <= bitC->endPtr);
|
259
249
|
MEM_writeLEST(bitC->ptr, bitC->bitContainer);
|
260
250
|
bitC->ptr += nbBytes;
|
261
251
|
if (bitC->ptr > bitC->endPtr) bitC->ptr = bitC->endPtr;
|
@@ -286,7 +276,7 @@ MEM_STATIC size_t BIT_closeCStream(BIT_CStream_t* bitC)
|
|
286
276
|
*/
|
287
277
|
MEM_STATIC size_t BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, size_t srcSize)
|
288
278
|
{
|
289
|
-
if (srcSize < 1) {
|
279
|
+
if (srcSize < 1) { ZSTD_memset(bitD, 0, sizeof(*bitD)); return ERROR(srcSize_wrong); }
|
290
280
|
|
291
281
|
bitD->start = (const char*)srcBuffer;
|
292
282
|
bitD->limitPtr = bitD->start + sizeof(bitD->bitContainer);
|
@@ -332,12 +322,12 @@ MEM_STATIC size_t BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, si
|
|
332
322
|
return srcSize;
|
333
323
|
}
|
334
324
|
|
335
|
-
MEM_STATIC size_t BIT_getUpperBits(size_t bitContainer, U32 const start)
|
325
|
+
MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getUpperBits(size_t bitContainer, U32 const start)
|
336
326
|
{
|
337
327
|
return bitContainer >> start;
|
338
328
|
}
|
339
329
|
|
340
|
-
MEM_STATIC size_t BIT_getMiddleBits(size_t bitContainer, U32 const start, U32 const nbBits)
|
330
|
+
MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getMiddleBits(size_t bitContainer, U32 const start, U32 const nbBits)
|
341
331
|
{
|
342
332
|
U32 const regMask = sizeof(bitContainer)*8 - 1;
|
343
333
|
/* if start > regMask, bitstream is corrupted, and result is undefined */
|
@@ -345,10 +335,14 @@ MEM_STATIC size_t BIT_getMiddleBits(size_t bitContainer, U32 const start, U32 co
|
|
345
335
|
return (bitContainer >> (start & regMask)) & BIT_mask[nbBits];
|
346
336
|
}
|
347
337
|
|
348
|
-
MEM_STATIC size_t BIT_getLowerBits(size_t bitContainer, U32 const nbBits)
|
338
|
+
MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getLowerBits(size_t bitContainer, U32 const nbBits)
|
349
339
|
{
|
340
|
+
#if defined(STATIC_BMI2) && STATIC_BMI2 == 1
|
341
|
+
return _bzhi_u64(bitContainer, nbBits);
|
342
|
+
#else
|
350
343
|
assert(nbBits < BIT_MASK_SIZE);
|
351
344
|
return bitContainer & BIT_mask[nbBits];
|
345
|
+
#endif
|
352
346
|
}
|
353
347
|
|
354
348
|
/*! BIT_lookBits() :
|
@@ -357,7 +351,7 @@ MEM_STATIC size_t BIT_getLowerBits(size_t bitContainer, U32 const nbBits)
|
|
357
351
|
* On 32-bits, maxNbBits==24.
|
358
352
|
* On 64-bits, maxNbBits==56.
|
359
353
|
* @return : value extracted */
|
360
|
-
MEM_STATIC size_t BIT_lookBits(const BIT_DStream_t*
|
354
|
+
MEM_STATIC FORCE_INLINE_ATTR size_t BIT_lookBits(const BIT_DStream_t* bitD, U32 nbBits)
|
361
355
|
{
|
362
356
|
/* arbitrate between double-shift and shift+mask */
|
363
357
|
#if 1
|
@@ -380,7 +374,7 @@ MEM_STATIC size_t BIT_lookBitsFast(const BIT_DStream_t* bitD, U32 nbBits)
|
|
380
374
|
return (bitD->bitContainer << (bitD->bitsConsumed & regMask)) >> (((regMask+1)-nbBits) & regMask);
|
381
375
|
}
|
382
376
|
|
383
|
-
MEM_STATIC void BIT_skipBits(BIT_DStream_t* bitD, U32 nbBits)
|
377
|
+
MEM_STATIC FORCE_INLINE_ATTR void BIT_skipBits(BIT_DStream_t* bitD, U32 nbBits)
|
384
378
|
{
|
385
379
|
bitD->bitsConsumed += nbBits;
|
386
380
|
}
|
@@ -389,7 +383,7 @@ MEM_STATIC void BIT_skipBits(BIT_DStream_t* bitD, U32 nbBits)
|
|
389
383
|
* Read (consume) next n bits from local register and update.
|
390
384
|
* Pay attention to not read more than nbBits contained into local register.
|
391
385
|
* @return : extracted value. */
|
392
|
-
MEM_STATIC size_t BIT_readBits(BIT_DStream_t* bitD, unsigned nbBits)
|
386
|
+
MEM_STATIC FORCE_INLINE_ATTR size_t BIT_readBits(BIT_DStream_t* bitD, unsigned nbBits)
|
393
387
|
{
|
394
388
|
size_t const value = BIT_lookBits(bitD, nbBits);
|
395
389
|
BIT_skipBits(bitD, nbBits);
|
@@ -406,6 +400,23 @@ MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, unsigned nbBits)
|
|
406
400
|
return value;
|
407
401
|
}
|
408
402
|
|
403
|
+
/*! BIT_reloadDStreamFast() :
|
404
|
+
* Similar to BIT_reloadDStream(), but with two differences:
|
405
|
+
* 1. bitsConsumed <= sizeof(bitD->bitContainer)*8 must hold!
|
406
|
+
* 2. Returns BIT_DStream_overflow when bitD->ptr < bitD->limitPtr, at this
|
407
|
+
* point you must use BIT_reloadDStream() to reload.
|
408
|
+
*/
|
409
|
+
MEM_STATIC BIT_DStream_status BIT_reloadDStreamFast(BIT_DStream_t* bitD)
|
410
|
+
{
|
411
|
+
if (UNLIKELY(bitD->ptr < bitD->limitPtr))
|
412
|
+
return BIT_DStream_overflow;
|
413
|
+
assert(bitD->bitsConsumed <= sizeof(bitD->bitContainer)*8);
|
414
|
+
bitD->ptr -= bitD->bitsConsumed >> 3;
|
415
|
+
bitD->bitsConsumed &= 7;
|
416
|
+
bitD->bitContainer = MEM_readLEST(bitD->ptr);
|
417
|
+
return BIT_DStream_unfinished;
|
418
|
+
}
|
419
|
+
|
409
420
|
/*! BIT_reloadDStream() :
|
410
421
|
* Refill `bitD` from buffer previously set in BIT_initDStream() .
|
411
422
|
* This function is safe, it guarantees it will not read beyond src buffer.
|
@@ -417,10 +428,7 @@ MEM_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD)
|
|
417
428
|
return BIT_DStream_overflow;
|
418
429
|
|
419
430
|
if (bitD->ptr >= bitD->limitPtr) {
|
420
|
-
|
421
|
-
bitD->bitsConsumed &= 7;
|
422
|
-
bitD->bitContainer = MEM_readLEST(bitD->ptr);
|
423
|
-
return BIT_DStream_unfinished;
|
431
|
+
return BIT_reloadDStreamFast(bitD);
|
424
432
|
}
|
425
433
|
if (bitD->ptr == bitD->start) {
|
426
434
|
if (bitD->bitsConsumed < sizeof(bitD->bitContainer)*8) return BIT_DStream_endOfBuffer;
|
@@ -0,0 +1,289 @@
|
|
1
|
+
/*
|
2
|
+
* Copyright (c) Yann Collet, Facebook, Inc.
|
3
|
+
* All rights reserved.
|
4
|
+
*
|
5
|
+
* This source code is licensed under both the BSD-style license (found in the
|
6
|
+
* LICENSE file in the root directory of this source tree) and the GPLv2 (found
|
7
|
+
* in the COPYING file in the root directory of this source tree).
|
8
|
+
* You may select, at your option, one of the above-listed licenses.
|
9
|
+
*/
|
10
|
+
|
11
|
+
#ifndef ZSTD_COMPILER_H
|
12
|
+
#define ZSTD_COMPILER_H
|
13
|
+
|
14
|
+
/*-*******************************************************
|
15
|
+
* Compiler specifics
|
16
|
+
*********************************************************/
|
17
|
+
/* force inlining */
|
18
|
+
|
19
|
+
#if !defined(ZSTD_NO_INLINE)
|
20
|
+
#if (defined(__GNUC__) && !defined(__STRICT_ANSI__)) || defined(__cplusplus) || defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L /* C99 */
|
21
|
+
# define INLINE_KEYWORD inline
|
22
|
+
#else
|
23
|
+
# define INLINE_KEYWORD
|
24
|
+
#endif
|
25
|
+
|
26
|
+
#if defined(__GNUC__) || defined(__ICCARM__)
|
27
|
+
# define FORCE_INLINE_ATTR __attribute__((always_inline))
|
28
|
+
#elif defined(_MSC_VER)
|
29
|
+
# define FORCE_INLINE_ATTR __forceinline
|
30
|
+
#else
|
31
|
+
# define FORCE_INLINE_ATTR
|
32
|
+
#endif
|
33
|
+
|
34
|
+
#else
|
35
|
+
|
36
|
+
#define INLINE_KEYWORD
|
37
|
+
#define FORCE_INLINE_ATTR
|
38
|
+
|
39
|
+
#endif
|
40
|
+
|
41
|
+
/**
|
42
|
+
On MSVC qsort requires that functions passed into it use the __cdecl calling conversion(CC).
|
43
|
+
This explictly marks such functions as __cdecl so that the code will still compile
|
44
|
+
if a CC other than __cdecl has been made the default.
|
45
|
+
*/
|
46
|
+
#if defined(_MSC_VER)
|
47
|
+
# define WIN_CDECL __cdecl
|
48
|
+
#else
|
49
|
+
# define WIN_CDECL
|
50
|
+
#endif
|
51
|
+
|
52
|
+
/**
|
53
|
+
* FORCE_INLINE_TEMPLATE is used to define C "templates", which take constant
|
54
|
+
* parameters. They must be inlined for the compiler to eliminate the constant
|
55
|
+
* branches.
|
56
|
+
*/
|
57
|
+
#define FORCE_INLINE_TEMPLATE static INLINE_KEYWORD FORCE_INLINE_ATTR
|
58
|
+
/**
|
59
|
+
* HINT_INLINE is used to help the compiler generate better code. It is *not*
|
60
|
+
* used for "templates", so it can be tweaked based on the compilers
|
61
|
+
* performance.
|
62
|
+
*
|
63
|
+
* gcc-4.8 and gcc-4.9 have been shown to benefit from leaving off the
|
64
|
+
* always_inline attribute.
|
65
|
+
*
|
66
|
+
* clang up to 5.0.0 (trunk) benefit tremendously from the always_inline
|
67
|
+
* attribute.
|
68
|
+
*/
|
69
|
+
#if !defined(__clang__) && defined(__GNUC__) && __GNUC__ >= 4 && __GNUC_MINOR__ >= 8 && __GNUC__ < 5
|
70
|
+
# define HINT_INLINE static INLINE_KEYWORD
|
71
|
+
#else
|
72
|
+
# define HINT_INLINE static INLINE_KEYWORD FORCE_INLINE_ATTR
|
73
|
+
#endif
|
74
|
+
|
75
|
+
/* UNUSED_ATTR tells the compiler it is okay if the function is unused. */
|
76
|
+
#if defined(__GNUC__)
|
77
|
+
# define UNUSED_ATTR __attribute__((unused))
|
78
|
+
#else
|
79
|
+
# define UNUSED_ATTR
|
80
|
+
#endif
|
81
|
+
|
82
|
+
/* force no inlining */
|
83
|
+
#ifdef _MSC_VER
|
84
|
+
# define FORCE_NOINLINE static __declspec(noinline)
|
85
|
+
#else
|
86
|
+
# if defined(__GNUC__) || defined(__ICCARM__)
|
87
|
+
# define FORCE_NOINLINE static __attribute__((__noinline__))
|
88
|
+
# else
|
89
|
+
# define FORCE_NOINLINE static
|
90
|
+
# endif
|
91
|
+
#endif
|
92
|
+
|
93
|
+
|
94
|
+
/* target attribute */
|
95
|
+
#ifndef __has_attribute
|
96
|
+
#define __has_attribute(x) 0 /* Compatibility with non-clang compilers. */
|
97
|
+
#endif
|
98
|
+
#if defined(__GNUC__) || defined(__ICCARM__)
|
99
|
+
# define TARGET_ATTRIBUTE(target) __attribute__((__target__(target)))
|
100
|
+
#else
|
101
|
+
# define TARGET_ATTRIBUTE(target)
|
102
|
+
#endif
|
103
|
+
|
104
|
+
/* Enable runtime BMI2 dispatch based on the CPU.
|
105
|
+
* Enabled for clang & gcc >=4.8 on x86 when BMI2 isn't enabled by default.
|
106
|
+
*/
|
107
|
+
#ifndef DYNAMIC_BMI2
|
108
|
+
#if ((defined(__clang__) && __has_attribute(__target__)) \
|
109
|
+
|| (defined(__GNUC__) \
|
110
|
+
&& (__GNUC__ >= 5 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 8)))) \
|
111
|
+
&& (defined(__x86_64__) || defined(_M_X86)) \
|
112
|
+
&& !defined(__BMI2__)
|
113
|
+
# define DYNAMIC_BMI2 1
|
114
|
+
#else
|
115
|
+
# define DYNAMIC_BMI2 0
|
116
|
+
#endif
|
117
|
+
#endif
|
118
|
+
|
119
|
+
/* prefetch
|
120
|
+
* can be disabled, by declaring NO_PREFETCH build macro */
|
121
|
+
#if defined(NO_PREFETCH)
|
122
|
+
# define PREFETCH_L1(ptr) (void)(ptr) /* disabled */
|
123
|
+
# define PREFETCH_L2(ptr) (void)(ptr) /* disabled */
|
124
|
+
#else
|
125
|
+
# if defined(_MSC_VER) && (defined(_M_X64) || defined(_M_I86)) /* _mm_prefetch() is not defined outside of x86/x64 */
|
126
|
+
# include <mmintrin.h> /* https://msdn.microsoft.com/fr-fr/library/84szxsww(v=vs.90).aspx */
|
127
|
+
# define PREFETCH_L1(ptr) _mm_prefetch((const char*)(ptr), _MM_HINT_T0)
|
128
|
+
# define PREFETCH_L2(ptr) _mm_prefetch((const char*)(ptr), _MM_HINT_T1)
|
129
|
+
# elif defined(__GNUC__) && ( (__GNUC__ >= 4) || ( (__GNUC__ == 3) && (__GNUC_MINOR__ >= 1) ) )
|
130
|
+
# define PREFETCH_L1(ptr) __builtin_prefetch((ptr), 0 /* rw==read */, 3 /* locality */)
|
131
|
+
# define PREFETCH_L2(ptr) __builtin_prefetch((ptr), 0 /* rw==read */, 2 /* locality */)
|
132
|
+
# elif defined(__aarch64__)
|
133
|
+
# define PREFETCH_L1(ptr) __asm__ __volatile__("prfm pldl1keep, %0" ::"Q"(*(ptr)))
|
134
|
+
# define PREFETCH_L2(ptr) __asm__ __volatile__("prfm pldl2keep, %0" ::"Q"(*(ptr)))
|
135
|
+
# else
|
136
|
+
# define PREFETCH_L1(ptr) (void)(ptr) /* disabled */
|
137
|
+
# define PREFETCH_L2(ptr) (void)(ptr) /* disabled */
|
138
|
+
# endif
|
139
|
+
#endif /* NO_PREFETCH */
|
140
|
+
|
141
|
+
#define CACHELINE_SIZE 64
|
142
|
+
|
143
|
+
#define PREFETCH_AREA(p, s) { \
|
144
|
+
const char* const _ptr = (const char*)(p); \
|
145
|
+
size_t const _size = (size_t)(s); \
|
146
|
+
size_t _pos; \
|
147
|
+
for (_pos=0; _pos<_size; _pos+=CACHELINE_SIZE) { \
|
148
|
+
PREFETCH_L2(_ptr + _pos); \
|
149
|
+
} \
|
150
|
+
}
|
151
|
+
|
152
|
+
/* vectorization
|
153
|
+
* older GCC (pre gcc-4.3 picked as the cutoff) uses a different syntax */
|
154
|
+
#if !defined(__INTEL_COMPILER) && !defined(__clang__) && defined(__GNUC__)
|
155
|
+
# if (__GNUC__ == 4 && __GNUC_MINOR__ > 3) || (__GNUC__ >= 5)
|
156
|
+
# define DONT_VECTORIZE __attribute__((optimize("no-tree-vectorize")))
|
157
|
+
# else
|
158
|
+
# define DONT_VECTORIZE _Pragma("GCC optimize(\"no-tree-vectorize\")")
|
159
|
+
# endif
|
160
|
+
#else
|
161
|
+
# define DONT_VECTORIZE
|
162
|
+
#endif
|
163
|
+
|
164
|
+
/* Tell the compiler that a branch is likely or unlikely.
|
165
|
+
* Only use these macros if it causes the compiler to generate better code.
|
166
|
+
* If you can remove a LIKELY/UNLIKELY annotation without speed changes in gcc
|
167
|
+
* and clang, please do.
|
168
|
+
*/
|
169
|
+
#if defined(__GNUC__)
|
170
|
+
#define LIKELY(x) (__builtin_expect((x), 1))
|
171
|
+
#define UNLIKELY(x) (__builtin_expect((x), 0))
|
172
|
+
#else
|
173
|
+
#define LIKELY(x) (x)
|
174
|
+
#define UNLIKELY(x) (x)
|
175
|
+
#endif
|
176
|
+
|
177
|
+
/* disable warnings */
|
178
|
+
#ifdef _MSC_VER /* Visual Studio */
|
179
|
+
# include <intrin.h> /* For Visual 2005 */
|
180
|
+
# pragma warning(disable : 4100) /* disable: C4100: unreferenced formal parameter */
|
181
|
+
# pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */
|
182
|
+
# pragma warning(disable : 4204) /* disable: C4204: non-constant aggregate initializer */
|
183
|
+
# pragma warning(disable : 4214) /* disable: C4214: non-int bitfields */
|
184
|
+
# pragma warning(disable : 4324) /* disable: C4324: padded structure */
|
185
|
+
#endif
|
186
|
+
|
187
|
+
/*Like DYNAMIC_BMI2 but for compile time determination of BMI2 support*/
|
188
|
+
#ifndef STATIC_BMI2
|
189
|
+
# if defined(_MSC_VER) && (defined(_M_X64) || defined(_M_I86))
|
190
|
+
# ifdef __AVX2__ //MSVC does not have a BMI2 specific flag, but every CPU that supports AVX2 also supports BMI2
|
191
|
+
# define STATIC_BMI2 1
|
192
|
+
# endif
|
193
|
+
# endif
|
194
|
+
#endif
|
195
|
+
|
196
|
+
#ifndef STATIC_BMI2
|
197
|
+
#define STATIC_BMI2 0
|
198
|
+
#endif
|
199
|
+
|
200
|
+
/* compat. with non-clang compilers */
|
201
|
+
#ifndef __has_builtin
|
202
|
+
# define __has_builtin(x) 0
|
203
|
+
#endif
|
204
|
+
|
205
|
+
/* compat. with non-clang compilers */
|
206
|
+
#ifndef __has_feature
|
207
|
+
# define __has_feature(x) 0
|
208
|
+
#endif
|
209
|
+
|
210
|
+
/* detects whether we are being compiled under msan */
|
211
|
+
#ifndef ZSTD_MEMORY_SANITIZER
|
212
|
+
# if __has_feature(memory_sanitizer)
|
213
|
+
# define ZSTD_MEMORY_SANITIZER 1
|
214
|
+
# else
|
215
|
+
# define ZSTD_MEMORY_SANITIZER 0
|
216
|
+
# endif
|
217
|
+
#endif
|
218
|
+
|
219
|
+
#if ZSTD_MEMORY_SANITIZER
|
220
|
+
/* Not all platforms that support msan provide sanitizers/msan_interface.h.
|
221
|
+
* We therefore declare the functions we need ourselves, rather than trying to
|
222
|
+
* include the header file... */
|
223
|
+
#include <stddef.h> /* size_t */
|
224
|
+
#define ZSTD_DEPS_NEED_STDINT
|
225
|
+
#include "zstd_deps.h" /* intptr_t */
|
226
|
+
|
227
|
+
/* Make memory region fully initialized (without changing its contents). */
|
228
|
+
void __msan_unpoison(const volatile void *a, size_t size);
|
229
|
+
|
230
|
+
/* Make memory region fully uninitialized (without changing its contents).
|
231
|
+
This is a legacy interface that does not update origin information. Use
|
232
|
+
__msan_allocated_memory() instead. */
|
233
|
+
void __msan_poison(const volatile void *a, size_t size);
|
234
|
+
|
235
|
+
/* Returns the offset of the first (at least partially) poisoned byte in the
|
236
|
+
memory range, or -1 if the whole range is good. */
|
237
|
+
intptr_t __msan_test_shadow(const volatile void *x, size_t size);
|
238
|
+
#endif
|
239
|
+
|
240
|
+
/* detects whether we are being compiled under asan */
|
241
|
+
#ifndef ZSTD_ADDRESS_SANITIZER
|
242
|
+
# if __has_feature(address_sanitizer)
|
243
|
+
# define ZSTD_ADDRESS_SANITIZER 1
|
244
|
+
# elif defined(__SANITIZE_ADDRESS__)
|
245
|
+
# define ZSTD_ADDRESS_SANITIZER 1
|
246
|
+
# else
|
247
|
+
# define ZSTD_ADDRESS_SANITIZER 0
|
248
|
+
# endif
|
249
|
+
#endif
|
250
|
+
|
251
|
+
#if ZSTD_ADDRESS_SANITIZER
|
252
|
+
/* Not all platforms that support asan provide sanitizers/asan_interface.h.
|
253
|
+
* We therefore declare the functions we need ourselves, rather than trying to
|
254
|
+
* include the header file... */
|
255
|
+
#include <stddef.h> /* size_t */
|
256
|
+
|
257
|
+
/**
|
258
|
+
* Marks a memory region (<c>[addr, addr+size)</c>) as unaddressable.
|
259
|
+
*
|
260
|
+
* This memory must be previously allocated by your program. Instrumented
|
261
|
+
* code is forbidden from accessing addresses in this region until it is
|
262
|
+
* unpoisoned. This function is not guaranteed to poison the entire region -
|
263
|
+
* it could poison only a subregion of <c>[addr, addr+size)</c> due to ASan
|
264
|
+
* alignment restrictions.
|
265
|
+
*
|
266
|
+
* \note This function is not thread-safe because no two threads can poison or
|
267
|
+
* unpoison memory in the same memory region simultaneously.
|
268
|
+
*
|
269
|
+
* \param addr Start of memory region.
|
270
|
+
* \param size Size of memory region. */
|
271
|
+
void __asan_poison_memory_region(void const volatile *addr, size_t size);
|
272
|
+
|
273
|
+
/**
|
274
|
+
* Marks a memory region (<c>[addr, addr+size)</c>) as addressable.
|
275
|
+
*
|
276
|
+
* This memory must be previously allocated by your program. Accessing
|
277
|
+
* addresses in this region is allowed until this region is poisoned again.
|
278
|
+
* This function could unpoison a super-region of <c>[addr, addr+size)</c> due
|
279
|
+
* to ASan alignment restrictions.
|
280
|
+
*
|
281
|
+
* \note This function is not thread-safe because no two threads can
|
282
|
+
* poison or unpoison memory in the same memory region simultaneously.
|
283
|
+
*
|
284
|
+
* \param addr Start of memory region.
|
285
|
+
* \param size Size of memory region. */
|
286
|
+
void __asan_unpoison_memory_region(void const volatile *addr, size_t size);
|
287
|
+
#endif
|
288
|
+
|
289
|
+
#endif /* ZSTD_COMPILER_H */
|