?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Avoid ARC buffer transfrom operations in prefetch
[zfs.git] / module / zstd / lib / decompress / zstd_decompress_block.c
blob6f09e61b70ccd24262610ec4e2e72d0e486ab761
1 /*
2 * Copyright (c) 2016-2020, 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 /* zstd_decompress_block :
12 * this module takes care of decompressing _compressed_ block */
13
14 /*-*******************************************************
15 * Dependencies
16 *********************************************************/
17 #include <string.h> /* memcpy, memmove, memset */
18 #include "../common/compiler.h" /* prefetch */
19 #include "../common/cpu.h" /* bmi2 */
20 #include "../common/mem.h" /* low level memory routines */
21 #define FSE_STATIC_LINKING_ONLY
22 #include "../common/fse.h"
23 #define HUF_STATIC_LINKING_ONLY
24 #include "../common/huf.h"
25 #include "../common/zstd_internal.h"
26 #include "zstd_decompress_internal.h" /* ZSTD_DCtx */
27 #include "zstd_ddict.h" /* ZSTD_DDictDictContent */
28 #include "zstd_decompress_block.h"
29
30 /*_*******************************************************
31 * Macros
32 **********************************************************/
33
34 /* These two optional macros force the use one way or another of the two
35 * ZSTD_decompressSequences implementations. You can't force in both directions
36 * at the same time.
37 */
38 #if defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
39 defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
40 #error "Cannot force the use of the short and the long ZSTD_decompressSequences variants!"
41 #endif
42
43
44 /*_*******************************************************
45 * Memory operations
46 **********************************************************/
47 static void ZSTD_copy4(void* dst, const void* src) { memcpy(dst, src, 4); }
48
49
50 /*-*************************************************************
51 * Block decoding
52 ***************************************************************/
53
54 /*! ZSTD_getcBlockSize() :
55 * Provides the size of compressed block from block header `src` */
56 size_t ZSTD_getcBlockSize(const void* src, size_t srcSize,
57 blockProperties_t* bpPtr)
58 {
59 RETURN_ERROR_IF(srcSize < ZSTD_blockHeaderSize, srcSize_wrong, "");
60
61 { U32 const cBlockHeader = MEM_readLE24(src);
62 U32 const cSize = cBlockHeader >> 3;
63 bpPtr->lastBlock = cBlockHeader & 1;
64 bpPtr->blockType = (blockType_e)((cBlockHeader >> 1) & 3);
65 bpPtr->origSize = cSize; /* only useful for RLE */
66 if (bpPtr->blockType == bt_rle) return 1;
67 RETURN_ERROR_IF(bpPtr->blockType == bt_reserved, corruption_detected, "");
68 return cSize;
69 }
70 }
71
72
73 /* Hidden declaration for fullbench */
74 size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx,
75 const void* src, size_t srcSize);
76 /*! ZSTD_decodeLiteralsBlock() :
77 * @return : nb of bytes read from src (< srcSize )
78 * note : symbol not declared but exposed for fullbench */
79 size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx,
80 const void* src, size_t srcSize) /* note : srcSize < BLOCKSIZE */
81 {
82 DEBUGLOG(5, "ZSTD_decodeLiteralsBlock");
83 RETURN_ERROR_IF(srcSize < MIN_CBLOCK_SIZE, corruption_detected, "");
84
85 { const BYTE* const istart = (const BYTE*) src;
86 symbolEncodingType_e const litEncType = (symbolEncodingType_e)(istart[0] & 3);
87
88 switch(litEncType)
89 {
90 case set_repeat:
91 DEBUGLOG(5, "set_repeat flag : re-using stats from previous compressed literals block");
92 RETURN_ERROR_IF(dctx->litEntropy==0, dictionary_corrupted, "");
93 /* fall-through */
94
95 case set_compressed:
96 RETURN_ERROR_IF(srcSize < 5, corruption_detected, "srcSize >= MIN_CBLOCK_SIZE == 3; here we need up to 5 for case 3");
97 { size_t lhSize, litSize, litCSize;
98 U32 singleStream=0;
99 U32 const lhlCode = (istart[0] >> 2) & 3;
100 U32 const lhc = MEM_readLE32(istart);
101 size_t hufSuccess;
102 switch(lhlCode)
103 {
104 case 0: case 1: default: /* note : default is impossible, since lhlCode into [0..3] */
105 /* 2 - 2 - 10 - 10 */
106 singleStream = !lhlCode;
107 lhSize = 3;
108 litSize = (lhc >> 4) & 0x3FF;
109 litCSize = (lhc >> 14) & 0x3FF;
110 break;
111 case 2:
112 /* 2 - 2 - 14 - 14 */
113 lhSize = 4;
114 litSize = (lhc >> 4) & 0x3FFF;
115 litCSize = lhc >> 18;
116 break;
117 case 3:
118 /* 2 - 2 - 18 - 18 */
119 lhSize = 5;
120 litSize = (lhc >> 4) & 0x3FFFF;
121 litCSize = (lhc >> 22) + ((size_t)istart[4] << 10);
122 break;
123 }
124 RETURN_ERROR_IF(litSize > ZSTD_BLOCKSIZE_MAX, corruption_detected, "");
125 RETURN_ERROR_IF(litCSize + lhSize > srcSize, corruption_detected, "");
126
127 /* prefetch huffman table if cold */
128 if (dctx->ddictIsCold && (litSize > 768 /* heuristic */)) {
129 PREFETCH_AREA(dctx->HUFptr, sizeof(dctx->entropy.hufTable));
130 }
131
132 if (litEncType==set_repeat) {
133 if (singleStream) {
134 hufSuccess = HUF_decompress1X_usingDTable_bmi2(
135 dctx->litBuffer, litSize, istart+lhSize, litCSize,
136 dctx->HUFptr, dctx->bmi2);
137 } else {
138 hufSuccess = HUF_decompress4X_usingDTable_bmi2(
139 dctx->litBuffer, litSize, istart+lhSize, litCSize,
140 dctx->HUFptr, dctx->bmi2);
141 }
142 } else {
143 if (singleStream) {
144 #if defined(HUF_FORCE_DECOMPRESS_X2)
145 hufSuccess = HUF_decompress1X_DCtx_wksp(
146 dctx->entropy.hufTable, dctx->litBuffer, litSize,
147 istart+lhSize, litCSize, dctx->workspace,
148 sizeof(dctx->workspace));
149 #else
150 hufSuccess = HUF_decompress1X1_DCtx_wksp_bmi2(
151 dctx->entropy.hufTable, dctx->litBuffer, litSize,
152 istart+lhSize, litCSize, dctx->workspace,
153 sizeof(dctx->workspace), dctx->bmi2);
154 #endif
155 } else {
156 hufSuccess = HUF_decompress4X_hufOnly_wksp_bmi2(
157 dctx->entropy.hufTable, dctx->litBuffer, litSize,
158 istart+lhSize, litCSize, dctx->workspace,
159 sizeof(dctx->workspace), dctx->bmi2);
160 }
161 }
162
163 RETURN_ERROR_IF(HUF_isError(hufSuccess), corruption_detected, "");
164
165 dctx->litPtr = dctx->litBuffer;
166 dctx->litSize = litSize;
167 dctx->litEntropy = 1;
168 if (litEncType==set_compressed) dctx->HUFptr = dctx->entropy.hufTable;
169 memset(dctx->litBuffer + dctx->litSize, 0, WILDCOPY_OVERLENGTH);
170 return litCSize + lhSize;
171 }
172
173 case set_basic:
174 { size_t litSize, lhSize;
175 U32 const lhlCode = ((istart[0]) >> 2) & 3;
176 switch(lhlCode)
177 {
178 case 0: case 2: default: /* note : default is impossible, since lhlCode into [0..3] */
179 lhSize = 1;
180 litSize = istart[0] >> 3;
181 break;
182 case 1:
183 lhSize = 2;
184 litSize = MEM_readLE16(istart) >> 4;
185 break;
186 case 3:
187 lhSize = 3;
188 litSize = MEM_readLE24(istart) >> 4;
189 break;
190 }
191
192 if (lhSize+litSize+WILDCOPY_OVERLENGTH > srcSize) { /* risk reading beyond src buffer with wildcopy */
193 RETURN_ERROR_IF(litSize+lhSize > srcSize, corruption_detected, "");
194 memcpy(dctx->litBuffer, istart+lhSize, litSize);
195 dctx->litPtr = dctx->litBuffer;
196 dctx->litSize = litSize;
197 memset(dctx->litBuffer + dctx->litSize, 0, WILDCOPY_OVERLENGTH);
198 return lhSize+litSize;
199 }
200 /* direct reference into compressed stream */
201 dctx->litPtr = istart+lhSize;
202 dctx->litSize = litSize;
203 return lhSize+litSize;
204 }
205
206 case set_rle:
207 { U32 const lhlCode = ((istart[0]) >> 2) & 3;
208 size_t litSize, lhSize;
209 switch(lhlCode)
210 {
211 case 0: case 2: default: /* note : default is impossible, since lhlCode into [0..3] */
212 lhSize = 1;
213 litSize = istart[0] >> 3;
214 break;
215 case 1:
216 lhSize = 2;
217 litSize = MEM_readLE16(istart) >> 4;
218 break;
219 case 3:
220 lhSize = 3;
221 litSize = MEM_readLE24(istart) >> 4;
222 RETURN_ERROR_IF(srcSize<4, corruption_detected, "srcSize >= MIN_CBLOCK_SIZE == 3; here we need lhSize+1 = 4");
223 break;
224 }
225 RETURN_ERROR_IF(litSize > ZSTD_BLOCKSIZE_MAX, corruption_detected, "");
226 memset(dctx->litBuffer, istart[lhSize], litSize + WILDCOPY_OVERLENGTH);
227 dctx->litPtr = dctx->litBuffer;
228 dctx->litSize = litSize;
229 return lhSize+1;
230 }
231 default:
232 RETURN_ERROR(corruption_detected, "impossible");
233 }
234 }
235 }
236
237 /* Default FSE distribution tables.
238 * These are pre-calculated FSE decoding tables using default distributions as defined in specification :
239 * https://github.com/facebook/zstd/blob/master/doc/zstd_compression_format.md#default-distributions
240 * They were generated programmatically with following method :
241 * - start from default distributions, present in /lib/common/zstd_internal.h
242 * - generate tables normally, using ZSTD_buildFSETable()
243 * - printout the content of tables
244 * - pretify output, report below, test with fuzzer to ensure it's correct */
245
246 /* Default FSE distribution table for Literal Lengths */
247 static const ZSTD_seqSymbol LL_defaultDTable[(1<<LL_DEFAULTNORMLOG)+1] = {
248 { 1, 1, 1, LL_DEFAULTNORMLOG}, /* header : fastMode, tableLog */
249 /* nextState, nbAddBits, nbBits, baseVal */
250 { 0, 0, 4, 0}, { 16, 0, 4, 0},
251 { 32, 0, 5, 1}, { 0, 0, 5, 3},
252 { 0, 0, 5, 4}, { 0, 0, 5, 6},
253 { 0, 0, 5, 7}, { 0, 0, 5, 9},
254 { 0, 0, 5, 10}, { 0, 0, 5, 12},
255 { 0, 0, 6, 14}, { 0, 1, 5, 16},
256 { 0, 1, 5, 20}, { 0, 1, 5, 22},
257 { 0, 2, 5, 28}, { 0, 3, 5, 32},
258 { 0, 4, 5, 48}, { 32, 6, 5, 64},
259 { 0, 7, 5, 128}, { 0, 8, 6, 256},
260 { 0, 10, 6, 1024}, { 0, 12, 6, 4096},
261 { 32, 0, 4, 0}, { 0, 0, 4, 1},
262 { 0, 0, 5, 2}, { 32, 0, 5, 4},
263 { 0, 0, 5, 5}, { 32, 0, 5, 7},
264 { 0, 0, 5, 8}, { 32, 0, 5, 10},
265 { 0, 0, 5, 11}, { 0, 0, 6, 13},
266 { 32, 1, 5, 16}, { 0, 1, 5, 18},
267 { 32, 1, 5, 22}, { 0, 2, 5, 24},
268 { 32, 3, 5, 32}, { 0, 3, 5, 40},
269 { 0, 6, 4, 64}, { 16, 6, 4, 64},
270 { 32, 7, 5, 128}, { 0, 9, 6, 512},
271 { 0, 11, 6, 2048}, { 48, 0, 4, 0},
272 { 16, 0, 4, 1}, { 32, 0, 5, 2},
273 { 32, 0, 5, 3}, { 32, 0, 5, 5},
274 { 32, 0, 5, 6}, { 32, 0, 5, 8},
275 { 32, 0, 5, 9}, { 32, 0, 5, 11},
276 { 32, 0, 5, 12}, { 0, 0, 6, 15},
277 { 32, 1, 5, 18}, { 32, 1, 5, 20},
278 { 32, 2, 5, 24}, { 32, 2, 5, 28},
279 { 32, 3, 5, 40}, { 32, 4, 5, 48},
280 { 0, 16, 6,65536}, { 0, 15, 6,32768},
281 { 0, 14, 6,16384}, { 0, 13, 6, 8192},
282 }; /* LL_defaultDTable */
283
284 /* Default FSE distribution table for Offset Codes */
285 static const ZSTD_seqSymbol OF_defaultDTable[(1<<OF_DEFAULTNORMLOG)+1] = {
286 { 1, 1, 1, OF_DEFAULTNORMLOG}, /* header : fastMode, tableLog */
287 /* nextState, nbAddBits, nbBits, baseVal */
288 { 0, 0, 5, 0}, { 0, 6, 4, 61},
289 { 0, 9, 5, 509}, { 0, 15, 5,32765},
290 { 0, 21, 5,2097149}, { 0, 3, 5, 5},
291 { 0, 7, 4, 125}, { 0, 12, 5, 4093},
292 { 0, 18, 5,262141}, { 0, 23, 5,8388605},
293 { 0, 5, 5, 29}, { 0, 8, 4, 253},
294 { 0, 14, 5,16381}, { 0, 20, 5,1048573},
295 { 0, 2, 5, 1}, { 16, 7, 4, 125},
296 { 0, 11, 5, 2045}, { 0, 17, 5,131069},
297 { 0, 22, 5,4194301}, { 0, 4, 5, 13},
298 { 16, 8, 4, 253}, { 0, 13, 5, 8189},
299 { 0, 19, 5,524285}, { 0, 1, 5, 1},
300 { 16, 6, 4, 61}, { 0, 10, 5, 1021},
301 { 0, 16, 5,65533}, { 0, 28, 5,268435453},
302 { 0, 27, 5,134217725}, { 0, 26, 5,67108861},
303 { 0, 25, 5,33554429}, { 0, 24, 5,16777213},
304 }; /* OF_defaultDTable */
305
306
307 /* Default FSE distribution table for Match Lengths */
308 static const ZSTD_seqSymbol ML_defaultDTable[(1<<ML_DEFAULTNORMLOG)+1] = {
309 { 1, 1, 1, ML_DEFAULTNORMLOG}, /* header : fastMode, tableLog */
310 /* nextState, nbAddBits, nbBits, baseVal */
311 { 0, 0, 6, 3}, { 0, 0, 4, 4},
312 { 32, 0, 5, 5}, { 0, 0, 5, 6},
313 { 0, 0, 5, 8}, { 0, 0, 5, 9},
314 { 0, 0, 5, 11}, { 0, 0, 6, 13},
315 { 0, 0, 6, 16}, { 0, 0, 6, 19},
316 { 0, 0, 6, 22}, { 0, 0, 6, 25},
317 { 0, 0, 6, 28}, { 0, 0, 6, 31},
318 { 0, 0, 6, 34}, { 0, 1, 6, 37},
319 { 0, 1, 6, 41}, { 0, 2, 6, 47},
320 { 0, 3, 6, 59}, { 0, 4, 6, 83},
321 { 0, 7, 6, 131}, { 0, 9, 6, 515},
322 { 16, 0, 4, 4}, { 0, 0, 4, 5},
323 { 32, 0, 5, 6}, { 0, 0, 5, 7},
324 { 32, 0, 5, 9}, { 0, 0, 5, 10},
325 { 0, 0, 6, 12}, { 0, 0, 6, 15},
326 { 0, 0, 6, 18}, { 0, 0, 6, 21},
327 { 0, 0, 6, 24}, { 0, 0, 6, 27},
328 { 0, 0, 6, 30}, { 0, 0, 6, 33},
329 { 0, 1, 6, 35}, { 0, 1, 6, 39},
330 { 0, 2, 6, 43}, { 0, 3, 6, 51},
331 { 0, 4, 6, 67}, { 0, 5, 6, 99},
332 { 0, 8, 6, 259}, { 32, 0, 4, 4},
333 { 48, 0, 4, 4}, { 16, 0, 4, 5},
334 { 32, 0, 5, 7}, { 32, 0, 5, 8},
335 { 32, 0, 5, 10}, { 32, 0, 5, 11},
336 { 0, 0, 6, 14}, { 0, 0, 6, 17},
337 { 0, 0, 6, 20}, { 0, 0, 6, 23},
338 { 0, 0, 6, 26}, { 0, 0, 6, 29},
339 { 0, 0, 6, 32}, { 0, 16, 6,65539},
340 { 0, 15, 6,32771}, { 0, 14, 6,16387},
341 { 0, 13, 6, 8195}, { 0, 12, 6, 4099},
342 { 0, 11, 6, 2051}, { 0, 10, 6, 1027},
343 }; /* ML_defaultDTable */
344
345
346 static void ZSTD_buildSeqTable_rle(ZSTD_seqSymbol* dt, U32 baseValue, U32 nbAddBits)
347 {
348 void* ptr = dt;
349 ZSTD_seqSymbol_header* const DTableH = (ZSTD_seqSymbol_header*)ptr;
350 ZSTD_seqSymbol* const cell = dt + 1;
351
352 DTableH->tableLog = 0;
353 DTableH->fastMode = 0;
354
355 cell->nbBits = 0;
356 cell->nextState = 0;
357 assert(nbAddBits < 255);
358 cell->nbAdditionalBits = (BYTE)nbAddBits;
359 cell->baseValue = baseValue;
360 }
361
362
363 /* ZSTD_buildFSETable() :
364 * generate FSE decoding table for one symbol (ll, ml or off)
365 * cannot fail if input is valid =>
366 * all inputs are presumed validated at this stage */
367 void
368 ZSTD_buildFSETable(ZSTD_seqSymbol* dt,
369 const short* normalizedCounter, unsigned maxSymbolValue,
370 const U32* baseValue, const U32* nbAdditionalBits,
371 unsigned tableLog)
372 {
373 ZSTD_seqSymbol* const tableDecode = dt+1;
374 U16 symbolNext[MaxSeq+1];
375
376 U32 const maxSV1 = maxSymbolValue + 1;
377 U32 const tableSize = 1 << tableLog;
378 U32 highThreshold = tableSize-1;
379
380 /* Sanity Checks */
381 assert(maxSymbolValue <= MaxSeq);
382 assert(tableLog <= MaxFSELog);
383
384 /* Init, lay down lowprob symbols */
385 { ZSTD_seqSymbol_header DTableH;
386 DTableH.tableLog = tableLog;
387 DTableH.fastMode = 1;
388 { S16 const largeLimit= (S16)(1 << (tableLog-1));
389 U32 s;
390 for (s=0; s<maxSV1; s++) {
391 if (normalizedCounter[s]==-1) {
392 tableDecode[highThreshold--].baseValue = s;
393 symbolNext[s] = 1;
394 } else {
395 if (normalizedCounter[s] >= largeLimit) DTableH.fastMode=0;
396 assert(normalizedCounter[s]>=0);
397 symbolNext[s] = (U16)normalizedCounter[s];
398 } } }
399 memcpy(dt, &DTableH, sizeof(DTableH));
400 }
401
402 /* Spread symbols */
403 { U32 const tableMask = tableSize-1;
404 U32 const step = FSE_TABLESTEP(tableSize);
405 U32 s, position = 0;
406 for (s=0; s<maxSV1; s++) {
407 int i;
408 for (i=0; i<normalizedCounter[s]; i++) {
409 tableDecode[position].baseValue = s;
410 position = (position + step) & tableMask;
411 while (position > highThreshold) position = (position + step) & tableMask; /* lowprob area */
412 } }
413 assert(position == 0); /* position must reach all cells once, otherwise normalizedCounter is incorrect */
414 }
415
416 /* Build Decoding table */
417 { U32 u;
418 for (u=0; u<tableSize; u++) {
419 U32 const symbol = tableDecode[u].baseValue;
420 U32 const nextState = symbolNext[symbol]++;
421 tableDecode[u].nbBits = (BYTE) (tableLog - BIT_highbit32(nextState) );
422 tableDecode[u].nextState = (U16) ( (nextState << tableDecode[u].nbBits) - tableSize);
423 assert(nbAdditionalBits[symbol] < 255);
424 tableDecode[u].nbAdditionalBits = (BYTE)nbAdditionalBits[symbol];
425 tableDecode[u].baseValue = baseValue[symbol];
426 } }
427 }
428
429
430 /*! ZSTD_buildSeqTable() :
431 * @return : nb bytes read from src,
432 * or an error code if it fails */
433 static size_t ZSTD_buildSeqTable(ZSTD_seqSymbol* DTableSpace, const ZSTD_seqSymbol** DTablePtr,
434 symbolEncodingType_e type, unsigned max, U32 maxLog,
435 const void* src, size_t srcSize,
436 const U32* baseValue, const U32* nbAdditionalBits,
437 const ZSTD_seqSymbol* defaultTable, U32 flagRepeatTable,
438 int ddictIsCold, int nbSeq)
439 {
440 switch(type)
441 {
442 case set_rle :
443 RETURN_ERROR_IF(!srcSize, srcSize_wrong, "");
444 RETURN_ERROR_IF((*(const BYTE*)src) > max, corruption_detected, "");
445 { U32 const symbol = *(const BYTE*)src;
446 U32 const baseline = baseValue[symbol];
447 U32 const nbBits = nbAdditionalBits[symbol];
448 ZSTD_buildSeqTable_rle(DTableSpace, baseline, nbBits);
449 }
450 *DTablePtr = DTableSpace;
451 return 1;
452 case set_basic :
453 *DTablePtr = defaultTable;
454 return 0;
455 case set_repeat:
456 RETURN_ERROR_IF(!flagRepeatTable, corruption_detected, "");
457 /* prefetch FSE table if used */
458 if (ddictIsCold && (nbSeq > 24 /* heuristic */)) {
459 const void* const pStart = *DTablePtr;
460 size_t const pSize = sizeof(ZSTD_seqSymbol) * (SEQSYMBOL_TABLE_SIZE(maxLog));
461 PREFETCH_AREA(pStart, pSize);
462 }
463 return 0;
464 case set_compressed :
465 { unsigned tableLog;
466 S16 norm[MaxSeq+1];
467 size_t const headerSize = FSE_readNCount(norm, &max, &tableLog, src, srcSize);
468 RETURN_ERROR_IF(FSE_isError(headerSize), corruption_detected, "");
469 RETURN_ERROR_IF(tableLog > maxLog, corruption_detected, "");
470 ZSTD_buildFSETable(DTableSpace, norm, max, baseValue, nbAdditionalBits, tableLog);
471 *DTablePtr = DTableSpace;
472 return headerSize;
473 }
474 default :
475 assert(0);
476 RETURN_ERROR(GENERIC, "impossible");
477 }
478 }
479
480 size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr,
481 const void* src, size_t srcSize)
482 {
483 const BYTE* const istart = (const BYTE* const)src;
484 const BYTE* const iend = istart + srcSize;
485 const BYTE* ip = istart;
486 int nbSeq;
487 DEBUGLOG(5, "ZSTD_decodeSeqHeaders");
488
489 /* check */
490 RETURN_ERROR_IF(srcSize < MIN_SEQUENCES_SIZE, srcSize_wrong, "");
491
492 /* SeqHead */
493 nbSeq = *ip++;
494 if (!nbSeq) {
495 *nbSeqPtr=0;
496 RETURN_ERROR_IF(srcSize != 1, srcSize_wrong, "");
497 return 1;
498 }
499 if (nbSeq > 0x7F) {
500 if (nbSeq == 0xFF) {
501 RETURN_ERROR_IF(ip+2 > iend, srcSize_wrong, "");
502 nbSeq = MEM_readLE16(ip) + LONGNBSEQ, ip+=2;
503 } else {
504 RETURN_ERROR_IF(ip >= iend, srcSize_wrong, "");
505 nbSeq = ((nbSeq-0x80)<<8) + *ip++;
506 }
507 }
508 *nbSeqPtr = nbSeq;
509
510 /* FSE table descriptors */
511 RETURN_ERROR_IF(ip+1 > iend, srcSize_wrong, ""); /* minimum possible size: 1 byte for symbol encoding types */
512 { symbolEncodingType_e const LLtype = (symbolEncodingType_e)(*ip >> 6);
513 symbolEncodingType_e const OFtype = (symbolEncodingType_e)((*ip >> 4) & 3);
514 symbolEncodingType_e const MLtype = (symbolEncodingType_e)((*ip >> 2) & 3);
515 ip++;
516
517 /* Build DTables */
518 { size_t const llhSize = ZSTD_buildSeqTable(dctx->entropy.LLTable, &dctx->LLTptr,
519 LLtype, MaxLL, LLFSELog,
520 ip, iend-ip,
521 LL_base, LL_bits,
522 LL_defaultDTable, dctx->fseEntropy,
523 dctx->ddictIsCold, nbSeq);
524 RETURN_ERROR_IF(ZSTD_isError(llhSize), corruption_detected, "ZSTD_buildSeqTable failed");
525 ip += llhSize;
526 }
527
528 { size_t const ofhSize = ZSTD_buildSeqTable(dctx->entropy.OFTable, &dctx->OFTptr,
529 OFtype, MaxOff, OffFSELog,
530 ip, iend-ip,
531 OF_base, OF_bits,
532 OF_defaultDTable, dctx->fseEntropy,
533 dctx->ddictIsCold, nbSeq);
534 RETURN_ERROR_IF(ZSTD_isError(ofhSize), corruption_detected, "ZSTD_buildSeqTable failed");
535 ip += ofhSize;
536 }
537
538 { size_t const mlhSize = ZSTD_buildSeqTable(dctx->entropy.MLTable, &dctx->MLTptr,
539 MLtype, MaxML, MLFSELog,
540 ip, iend-ip,
541 ML_base, ML_bits,
542 ML_defaultDTable, dctx->fseEntropy,
543 dctx->ddictIsCold, nbSeq);
544 RETURN_ERROR_IF(ZSTD_isError(mlhSize), corruption_detected, "ZSTD_buildSeqTable failed");
545 ip += mlhSize;
546 }
547 }
548
549 return ip-istart;
550 }
551
552
553 typedef struct {
554 size_t litLength;
555 size_t matchLength;
556 size_t offset;
557 } seq_t;
558
559 typedef struct {
560 size_t state;
561 const ZSTD_seqSymbol* table;
562 } ZSTD_fseState;
563
564 typedef struct {
565 BIT_DStream_t DStream;
566 ZSTD_fseState stateLL;
567 ZSTD_fseState stateOffb;
568 ZSTD_fseState stateML;
569 size_t prevOffset[ZSTD_REP_NUM];
570 } seqState_t;
571
572 /*! ZSTD_overlapCopy8() :
573 * Copies 8 bytes from ip to op and updates op and ip where ip <= op.
574 * If the offset is < 8 then the offset is spread to at least 8 bytes.
575 *
576 * Precondition: *ip <= *op
577 * Postcondition: *op - *op >= 8
578 */
579 HINT_INLINE void ZSTD_overlapCopy8(BYTE** op, BYTE const** ip, size_t offset) {
580 assert(*ip <= *op);
581 if (offset < 8) {
582 /* close range match, overlap */
583 static const U32 dec32table[] = { 0, 1, 2, 1, 4, 4, 4, 4 }; /* added */
584 static const int dec64table[] = { 8, 8, 8, 7, 8, 9,10,11 }; /* subtracted */
585 int const sub2 = dec64table[offset];
586 (*op)[0] = (*ip)[0];
587 (*op)[1] = (*ip)[1];
588 (*op)[2] = (*ip)[2];
589 (*op)[3] = (*ip)[3];
590 *ip += dec32table[offset];
591 ZSTD_copy4(*op+4, *ip);
592 *ip -= sub2;
593 } else {
594 ZSTD_copy8(*op, *ip);
595 }
596 *ip += 8;
597 *op += 8;
598 assert(*op - *ip >= 8);
599 }
600
601 /*! ZSTD_safecopy() :
602 * Specialized version of memcpy() that is allowed to READ up to WILDCOPY_OVERLENGTH past the input buffer
603 * and write up to 16 bytes past oend_w (op >= oend_w is allowed).
604 * This function is only called in the uncommon case where the sequence is near the end of the block. It
605 * should be fast for a single long sequence, but can be slow for several short sequences.
606 *
607 * @param ovtype controls the overlap detection
608 * - ZSTD_no_overlap: The source and destination are guaranteed to be at least WILDCOPY_VECLEN bytes apart.
609 * - ZSTD_overlap_src_before_dst: The src and dst may overlap and may be any distance apart.
610 * The src buffer must be before the dst buffer.
611 */
612 static void ZSTD_safecopy(BYTE* op, BYTE* const oend_w, BYTE const* ip, ptrdiff_t length, ZSTD_overlap_e ovtype) {
613 ptrdiff_t const diff = op - ip;
614 BYTE* const oend = op + length;
615
616 assert((ovtype == ZSTD_no_overlap && (diff <= -8 || diff >= 8 || op >= oend_w)) ||
617 (ovtype == ZSTD_overlap_src_before_dst && diff >= 0));
618
619 if (length < 8) {
620 /* Handle short lengths. */
621 while (op < oend) *op++ = *ip++;
622 return;
623 }
624 if (ovtype == ZSTD_overlap_src_before_dst) {
625 /* Copy 8 bytes and ensure the offset >= 8 when there can be overlap. */
626 assert(length >= 8);
627 ZSTD_overlapCopy8(&op, &ip, diff);
628 assert(op - ip >= 8);
629 assert(op <= oend);
630 }
631
632 if (oend <= oend_w) {
633 /* No risk of overwrite. */
634 ZSTD_wildcopy(op, ip, length, ovtype);
635 return;
636 }
637 if (op <= oend_w) {
638 /* Wildcopy until we get close to the end. */
639 assert(oend > oend_w);
640 ZSTD_wildcopy(op, ip, oend_w - op, ovtype);
641 ip += oend_w - op;
642 op = oend_w;
643 }
644 /* Handle the leftovers. */
645 while (op < oend) *op++ = *ip++;
646 }
647
648 /* ZSTD_execSequenceEnd():
649 * This version handles cases that are near the end of the output buffer. It requires
650 * more careful checks to make sure there is no overflow. By separating out these hard
651 * and unlikely cases, we can speed up the common cases.
652 *
653 * NOTE: This function needs to be fast for a single long sequence, but doesn't need
654 * to be optimized for many small sequences, since those fall into ZSTD_execSequence().
655 */
656 FORCE_NOINLINE
657 size_t ZSTD_execSequenceEnd(BYTE* op,
658 BYTE* const oend, seq_t sequence,
659 const BYTE** litPtr, const BYTE* const litLimit,
660 const BYTE* const prefixStart, const BYTE* const virtualStart, const BYTE* const dictEnd)
661 {
662 BYTE* const oLitEnd = op + sequence.litLength;
663 size_t const sequenceLength = sequence.litLength + sequence.matchLength;
664 const BYTE* const iLitEnd = *litPtr + sequence.litLength;
665 const BYTE* match = oLitEnd - sequence.offset;
666 BYTE* const oend_w = oend - WILDCOPY_OVERLENGTH;
667
668 /* bounds checks : careful of address space overflow in 32-bit mode */
669 RETURN_ERROR_IF(sequenceLength > (size_t)(oend - op), dstSize_tooSmall, "last match must fit within dstBuffer");
670 RETURN_ERROR_IF(sequence.litLength > (size_t)(litLimit - *litPtr), corruption_detected, "try to read beyond literal buffer");
671 assert(op < op + sequenceLength);
672 assert(oLitEnd < op + sequenceLength);
673
674 /* copy literals */
675 ZSTD_safecopy(op, oend_w, *litPtr, sequence.litLength, ZSTD_no_overlap);
676 op = oLitEnd;
677 *litPtr = iLitEnd;
678
679 /* copy Match */
680 if (sequence.offset > (size_t)(oLitEnd - prefixStart)) {
681 /* offset beyond prefix */
682 RETURN_ERROR_IF(sequence.offset > (size_t)(oLitEnd - virtualStart), corruption_detected, "");
683 match = dictEnd - (prefixStart-match);
684 if (match + sequence.matchLength <= dictEnd) {
685 memmove(oLitEnd, match, sequence.matchLength);
686 return sequenceLength;
687 }
688 /* span extDict & currentPrefixSegment */
689 { size_t const length1 = dictEnd - match;
690 memmove(oLitEnd, match, length1);
691 op = oLitEnd + length1;
692 sequence.matchLength -= length1;
693 match = prefixStart;
694 } }
695 ZSTD_safecopy(op, oend_w, match, sequence.matchLength, ZSTD_overlap_src_before_dst);
696 return sequenceLength;
697 }
698
699 HINT_INLINE
700 size_t ZSTD_execSequence(BYTE* op,
701 BYTE* const oend, seq_t sequence,
702 const BYTE** litPtr, const BYTE* const litLimit,
703 const BYTE* const prefixStart, const BYTE* const virtualStart, const BYTE* const dictEnd)
704 {
705 BYTE* const oLitEnd = op + sequence.litLength;
706 size_t const sequenceLength = sequence.litLength + sequence.matchLength;
707 BYTE* const oMatchEnd = op + sequenceLength; /* risk : address space overflow (32-bits) */
708 BYTE* const oend_w = oend - WILDCOPY_OVERLENGTH; /* risk : address space underflow on oend=NULL */
709 const BYTE* const iLitEnd = *litPtr + sequence.litLength;
710 const BYTE* match = oLitEnd - sequence.offset;
711
712 assert(op != NULL /* Precondition */);
713 assert(oend_w < oend /* No underflow */);
714 /* Handle edge cases in a slow path:
715 * - Read beyond end of literals
716 * - Match end is within WILDCOPY_OVERLIMIT of oend
717 * - 32-bit mode and the match length overflows
718 */
719 if (UNLIKELY(
720 iLitEnd > litLimit ||
721 oMatchEnd > oend_w ||
722 (MEM_32bits() && (size_t)(oend - op) < sequenceLength + WILDCOPY_OVERLENGTH)))
723 return ZSTD_execSequenceEnd(op, oend, sequence, litPtr, litLimit, prefixStart, virtualStart, dictEnd);
724
725 /* Assumptions (everything else goes into ZSTD_execSequenceEnd()) */
726 assert(op <= oLitEnd /* No overflow */);
727 assert(oLitEnd < oMatchEnd /* Non-zero match & no overflow */);
728 assert(oMatchEnd <= oend /* No underflow */);
729 assert(iLitEnd <= litLimit /* Literal length is in bounds */);
730 assert(oLitEnd <= oend_w /* Can wildcopy literals */);
731 assert(oMatchEnd <= oend_w /* Can wildcopy matches */);
732
733 /* Copy Literals:
734 * Split out litLength <= 16 since it is nearly always true. +1.6% on gcc-9.
735 * We likely don't need the full 32-byte wildcopy.
736 */
737 assert(WILDCOPY_OVERLENGTH >= 16);
738 ZSTD_copy16(op, (*litPtr));
739 if (UNLIKELY(sequence.litLength > 16)) {
740 ZSTD_wildcopy(op+16, (*litPtr)+16, sequence.litLength-16, ZSTD_no_overlap);
741 }
742 op = oLitEnd;
743 *litPtr = iLitEnd; /* update for next sequence */
744
745 /* Copy Match */
746 if (sequence.offset > (size_t)(oLitEnd - prefixStart)) {
747 /* offset beyond prefix -> go into extDict */
748 RETURN_ERROR_IF(UNLIKELY(sequence.offset > (size_t)(oLitEnd - virtualStart)), corruption_detected, "");
749 match = dictEnd + (match - prefixStart);
750 if (match + sequence.matchLength <= dictEnd) {
751 memmove(oLitEnd, match, sequence.matchLength);
752 return sequenceLength;
753 }
754 /* span extDict & currentPrefixSegment */
755 { size_t const length1 = dictEnd - match;
756 memmove(oLitEnd, match, length1);
757 op = oLitEnd + length1;
758 sequence.matchLength -= length1;
759 match = prefixStart;
760 } }
761 /* Match within prefix of 1 or more bytes */
762 assert(op <= oMatchEnd);
763 assert(oMatchEnd <= oend_w);
764 assert(match >= prefixStart);
765 assert(sequence.matchLength >= 1);
766
767 /* Nearly all offsets are >= WILDCOPY_VECLEN bytes, which means we can use wildcopy
768 * without overlap checking.
769 */
770 if (LIKELY(sequence.offset >= WILDCOPY_VECLEN)) {
771 /* We bet on a full wildcopy for matches, since we expect matches to be
772 * longer than literals (in general). In silesia, ~10% of matches are longer
773 * than 16 bytes.
774 */
775 ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength, ZSTD_no_overlap);
776 return sequenceLength;
777 }
778 assert(sequence.offset < WILDCOPY_VECLEN);
779
780 /* Copy 8 bytes and spread the offset to be >= 8. */
781 ZSTD_overlapCopy8(&op, &match, sequence.offset);
782
783 /* If the match length is > 8 bytes, then continue with the wildcopy. */
784 if (sequence.matchLength > 8) {
785 assert(op < oMatchEnd);
786 ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength-8, ZSTD_overlap_src_before_dst);
787 }
788 return sequenceLength;
789 }
790
791 static void
792 ZSTD_initFseState(ZSTD_fseState* DStatePtr, BIT_DStream_t* bitD, const ZSTD_seqSymbol* dt)
793 {
794 const void* ptr = dt;
795 const ZSTD_seqSymbol_header* const DTableH = (const ZSTD_seqSymbol_header*)ptr;
796 DStatePtr->state = BIT_readBits(bitD, DTableH->tableLog);
797 DEBUGLOG(6, "ZSTD_initFseState : val=%u using %u bits",
798 (U32)DStatePtr->state, DTableH->tableLog);
799 BIT_reloadDStream(bitD);
800 DStatePtr->table = dt + 1;
801 }
802
803 FORCE_INLINE_TEMPLATE void
804 ZSTD_updateFseState(ZSTD_fseState* DStatePtr, BIT_DStream_t* bitD)
805 {
806 ZSTD_seqSymbol const DInfo = DStatePtr->table[DStatePtr->state];
807 U32 const nbBits = DInfo.nbBits;
808 size_t const lowBits = BIT_readBits(bitD, nbBits);
809 DStatePtr->state = DInfo.nextState + lowBits;
810 }
811
812 FORCE_INLINE_TEMPLATE void
813 ZSTD_updateFseStateWithDInfo(ZSTD_fseState* DStatePtr, BIT_DStream_t* bitD, ZSTD_seqSymbol const DInfo)
814 {
815 U32 const nbBits = DInfo.nbBits;
816 size_t const lowBits = BIT_readBits(bitD, nbBits);
817 DStatePtr->state = DInfo.nextState + lowBits;
818 }
819
820 /* We need to add at most (ZSTD_WINDOWLOG_MAX_32 - 1) bits to read the maximum
821 * offset bits. But we can only read at most (STREAM_ACCUMULATOR_MIN_32 - 1)
822 * bits before reloading. This value is the maximum number of bytes we read
823 * after reloading when we are decoding long offsets.
824 */
825 #define LONG_OFFSETS_MAX_EXTRA_BITS_32 \
826 (ZSTD_WINDOWLOG_MAX_32 > STREAM_ACCUMULATOR_MIN_32 \
827 ? ZSTD_WINDOWLOG_MAX_32 - STREAM_ACCUMULATOR_MIN_32 \
828 : 0)
829
830 typedef enum { ZSTD_lo_isRegularOffset, ZSTD_lo_isLongOffset=1 } ZSTD_longOffset_e;
831
832 FORCE_INLINE_TEMPLATE seq_t
833 ZSTD_decodeSequence(seqState_t* seqState, const ZSTD_longOffset_e longOffsets)
834 {
835 seq_t seq;
836 ZSTD_seqSymbol const llDInfo = seqState->stateLL.table[seqState->stateLL.state];
837 ZSTD_seqSymbol const mlDInfo = seqState->stateML.table[seqState->stateML.state];
838 ZSTD_seqSymbol const ofDInfo = seqState->stateOffb.table[seqState->stateOffb.state];
839 U32 const llBase = llDInfo.baseValue;
840 U32 const mlBase = mlDInfo.baseValue;
841 U32 const ofBase = ofDInfo.baseValue;
842 BYTE const llBits = llDInfo.nbAdditionalBits;
843 BYTE const mlBits = mlDInfo.nbAdditionalBits;
844 BYTE const ofBits = ofDInfo.nbAdditionalBits;
845 BYTE const totalBits = llBits+mlBits+ofBits;
846
847 /* sequence */
848 { size_t offset;
849 if (ofBits > 1) {
850 ZSTD_STATIC_ASSERT(ZSTD_lo_isLongOffset == 1);
851 ZSTD_STATIC_ASSERT(LONG_OFFSETS_MAX_EXTRA_BITS_32 == 5);
852 assert(ofBits <= MaxOff);
853 if (MEM_32bits() && longOffsets && (ofBits >= STREAM_ACCUMULATOR_MIN_32)) {
854 U32 const extraBits = ofBits - MIN(ofBits, 32 - seqState->DStream.bitsConsumed);
855 offset = ofBase + (BIT_readBitsFast(&seqState->DStream, ofBits - extraBits) << extraBits);
856 BIT_reloadDStream(&seqState->DStream);
857 if (extraBits) offset += BIT_readBitsFast(&seqState->DStream, extraBits);
858 assert(extraBits <= LONG_OFFSETS_MAX_EXTRA_BITS_32); /* to avoid another reload */
859 } else {
860 offset = ofBase + BIT_readBitsFast(&seqState->DStream, ofBits/*>0*/); /* <= (ZSTD_WINDOWLOG_MAX-1) bits */
861 if (MEM_32bits()) BIT_reloadDStream(&seqState->DStream);
862 }
863 seqState->prevOffset[2] = seqState->prevOffset[1];
864 seqState->prevOffset[1] = seqState->prevOffset[0];
865 seqState->prevOffset[0] = offset;
866 } else {
867 U32 const ll0 = (llBase == 0);
868 if (LIKELY((ofBits == 0))) {
869 if (LIKELY(!ll0))
870 offset = seqState->prevOffset[0];
871 else {
872 offset = seqState->prevOffset[1];
873 seqState->prevOffset[1] = seqState->prevOffset[0];
874 seqState->prevOffset[0] = offset;
875 }
876 } else {
877 offset = ofBase + ll0 + BIT_readBitsFast(&seqState->DStream, 1);
878 { size_t temp = (offset==3) ? seqState->prevOffset[0] - 1 : seqState->prevOffset[offset];
879 temp += !temp; /* 0 is not valid; input is corrupted; force offset to 1 */
880 if (offset != 1) seqState->prevOffset[2] = seqState->prevOffset[1];
881 seqState->prevOffset[1] = seqState->prevOffset[0];
882 seqState->prevOffset[0] = offset = temp;
883 } } }
884 seq.offset = offset;
885 }
886
887 seq.matchLength = mlBase;
888 if (mlBits > 0)
889 seq.matchLength += BIT_readBitsFast(&seqState->DStream, mlBits/*>0*/);
890
891 if (MEM_32bits() && (mlBits+llBits >= STREAM_ACCUMULATOR_MIN_32-LONG_OFFSETS_MAX_EXTRA_BITS_32))
892 BIT_reloadDStream(&seqState->DStream);
893 if (MEM_64bits() && UNLIKELY(totalBits >= STREAM_ACCUMULATOR_MIN_64-(LLFSELog+MLFSELog+OffFSELog)))
894 BIT_reloadDStream(&seqState->DStream);
895 /* Ensure there are enough bits to read the rest of data in 64-bit mode. */
896 ZSTD_STATIC_ASSERT(16+LLFSELog+MLFSELog+OffFSELog < STREAM_ACCUMULATOR_MIN_64);
897
898 seq.litLength = llBase;
899 if (llBits > 0)
900 seq.litLength += BIT_readBitsFast(&seqState->DStream, llBits/*>0*/);
901
902 if (MEM_32bits())
903 BIT_reloadDStream(&seqState->DStream);
904
905 DEBUGLOG(6, "seq: litL=%u, matchL=%u, offset=%u",
906 (U32)seq.litLength, (U32)seq.matchLength, (U32)seq.offset);
907
908 /* ANS state update
909 * gcc-9.0.0 does 2.5% worse with ZSTD_updateFseStateWithDInfo().
910 * clang-9.2.0 does 7% worse with ZSTD_updateFseState().
911 * Naturally it seems like ZSTD_updateFseStateWithDInfo() should be the
912 * better option, so it is the default for other compilers. But, if you
913 * measure that it is worse, please put up a pull request.
914 */
915 {
916 #if defined(__GNUC__) && !defined(__clang__)
917 const int kUseUpdateFseState = 1;
918 #else
919 const int kUseUpdateFseState = 0;
920 #endif
921 if (kUseUpdateFseState) {
922 ZSTD_updateFseState(&seqState->stateLL, &seqState->DStream); /* <= 9 bits */
923 ZSTD_updateFseState(&seqState->stateML, &seqState->DStream); /* <= 9 bits */
924 if (MEM_32bits()) BIT_reloadDStream(&seqState->DStream); /* <= 18 bits */
925 ZSTD_updateFseState(&seqState->stateOffb, &seqState->DStream); /* <= 8 bits */
926 } else {
927 ZSTD_updateFseStateWithDInfo(&seqState->stateLL, &seqState->DStream, llDInfo); /* <= 9 bits */
928 ZSTD_updateFseStateWithDInfo(&seqState->stateML, &seqState->DStream, mlDInfo); /* <= 9 bits */
929 if (MEM_32bits()) BIT_reloadDStream(&seqState->DStream); /* <= 18 bits */
930 ZSTD_updateFseStateWithDInfo(&seqState->stateOffb, &seqState->DStream, ofDInfo); /* <= 8 bits */
931 }
932 }
933
934 return seq;
935 }
936
937 #ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
938 static int ZSTD_dictionaryIsActive(ZSTD_DCtx const* dctx, BYTE const* prefixStart, BYTE const* oLitEnd)
939 {
940 size_t const windowSize = dctx->fParams.windowSize;
941 /* No dictionary used. */
942 if (dctx->dictContentEndForFuzzing == NULL) return 0;
943 /* Dictionary is our prefix. */
944 if (prefixStart == dctx->dictContentBeginForFuzzing) return 1;
945 /* Dictionary is not our ext-dict. */
946 if (dctx->dictEnd != dctx->dictContentEndForFuzzing) return 0;
947 /* Dictionary is not within our window size. */
948 if ((size_t)(oLitEnd - prefixStart) >= windowSize) return 0;
949 /* Dictionary is active. */
950 return 1;
951 }
952
953 MEM_STATIC void ZSTD_assertValidSequence(
954 ZSTD_DCtx const* dctx,
955 BYTE const* op, BYTE const* oend,
956 seq_t const seq,
957 BYTE const* prefixStart, BYTE const* virtualStart)
958 {
959 size_t const windowSize = dctx->fParams.windowSize;
960 size_t const sequenceSize = seq.litLength + seq.matchLength;
961 BYTE const* const oLitEnd = op + seq.litLength;
962 DEBUGLOG(6, "Checking sequence: litL=%u matchL=%u offset=%u",
963 (U32)seq.litLength, (U32)seq.matchLength, (U32)seq.offset);
964 assert(op <= oend);
965 assert((size_t)(oend - op) >= sequenceSize);
966 assert(sequenceSize <= ZSTD_BLOCKSIZE_MAX);
967 if (ZSTD_dictionaryIsActive(dctx, prefixStart, oLitEnd)) {
968 size_t const dictSize = (size_t)((char const*)dctx->dictContentEndForFuzzing - (char const*)dctx->dictContentBeginForFuzzing);
969 /* Offset must be within the dictionary. */
970 assert(seq.offset <= (size_t)(oLitEnd - virtualStart));
971 assert(seq.offset <= windowSize + dictSize);
972 } else {
973 /* Offset must be within our window. */
974 assert(seq.offset <= windowSize);
975 }
976 }
977 #endif
978
979 #ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
980 FORCE_INLINE_TEMPLATE size_t
981 DONT_VECTORIZE
982 ZSTD_decompressSequences_body( ZSTD_DCtx* dctx,
983 void* dst, size_t maxDstSize,
984 const void* seqStart, size_t seqSize, int nbSeq,
985 const ZSTD_longOffset_e isLongOffset,
986 const int frame)
987 {
988 const BYTE* ip = (const BYTE*)seqStart;
989 const BYTE* const iend = ip + seqSize;
990 BYTE* const ostart = (BYTE* const)dst;
991 BYTE* const oend = ostart + maxDstSize;
992 BYTE* op = ostart;
993 const BYTE* litPtr = dctx->litPtr;
994 const BYTE* const litEnd = litPtr + dctx->litSize;
995 const BYTE* const prefixStart = (const BYTE*) (dctx->prefixStart);
996 const BYTE* const vBase = (const BYTE*) (dctx->virtualStart);
997 const BYTE* const dictEnd = (const BYTE*) (dctx->dictEnd);
998 DEBUGLOG(5, "ZSTD_decompressSequences_body");
999 (void)frame;
1000
1001 /* Regen sequences */
1002 if (nbSeq) {
1003 seqState_t seqState;
1004 size_t error = 0;
1005 dctx->fseEntropy = 1;
1006 { U32 i; for (i=0; i<ZSTD_REP_NUM; i++) seqState.prevOffset[i] = dctx->entropy.rep[i]; }
1007 RETURN_ERROR_IF(
1008 ERR_isError(BIT_initDStream(&seqState.DStream, ip, iend-ip)),
1009 corruption_detected, "");
1010 ZSTD_initFseState(&seqState.stateLL, &seqState.DStream, dctx->LLTptr);
1011 ZSTD_initFseState(&seqState.stateOffb, &seqState.DStream, dctx->OFTptr);
1012 ZSTD_initFseState(&seqState.stateML, &seqState.DStream, dctx->MLTptr);
1013 assert(dst != NULL);
1014
1015 ZSTD_STATIC_ASSERT(
1016 BIT_DStream_unfinished < BIT_DStream_completed &&
1017 BIT_DStream_endOfBuffer < BIT_DStream_completed &&
1018 BIT_DStream_completed < BIT_DStream_overflow);
1019
1020 #if defined(__GNUC__) && defined(__x86_64__)
1021 /* Align the decompression loop to 32 + 16 bytes.
1022 *
1023 * zstd compiled with gcc-9 on an Intel i9-9900k shows 10% decompression
1024 * speed swings based on the alignment of the decompression loop. This
1025 * performance swing is caused by parts of the decompression loop falling
1026 * out of the DSB. The entire decompression loop should fit in the DSB,
1027 * when it can't we get much worse performance. You can measure if you've
1028 * hit the good case or the bad case with this perf command for some
1029 * compressed file test.zst:
1030 *
1031 * perf stat -e cycles -e instructions -e idq.all_dsb_cycles_any_uops \
1032 * -e idq.all_mite_cycles_any_uops -- ./zstd -tq test.zst
1033 *
1034 * If you see most cycles served out of the MITE you've hit the bad case.
1035 * If you see most cycles served out of the DSB you've hit the good case.
1036 * If it is pretty even then you may be in an okay case.
1037 *
1038 * I've been able to reproduce this issue on the following CPUs:
1039 * - Kabylake: Macbook Pro (15-inch, 2019) 2.4 GHz Intel Core i9
1040 * Use Instruments->Counters to get DSB/MITE cycles.
1041 * I never got performance swings, but I was able to
1042 * go from the good case of mostly DSB to half of the
1043 * cycles served from MITE.
1044 * - Coffeelake: Intel i9-9900k
1045 *
1046 * I haven't been able to reproduce the instability or DSB misses on any
1047 * of the following CPUS:
1048 * - Haswell
1049 * - Broadwell: Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GH
1050 * - Skylake
1051 *
1052 * If you are seeing performance stability this script can help test.
1053 * It tests on 4 commits in zstd where I saw performance change.
1054 *
1055 * https://gist.github.com/terrelln/9889fc06a423fd5ca6e99351564473f4
1056 */
1057 __asm__(".p2align 5");
1058 __asm__("nop");
1059 __asm__(".p2align 4");
1060 #endif
1061 for ( ; ; ) {
1062 seq_t const sequence = ZSTD_decodeSequence(&seqState, isLongOffset);
1063 size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequence, &litPtr, litEnd, prefixStart, vBase, dictEnd);
1064 #if defined(FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION) && defined(FUZZING_ASSERT_VALID_SEQUENCE)
1065 assert(!ZSTD_isError(oneSeqSize));
1066 if (frame) ZSTD_assertValidSequence(dctx, op, oend, sequence, prefixStart, vBase);
1067 #endif
1068 DEBUGLOG(6, "regenerated sequence size : %u", (U32)oneSeqSize);
1069 BIT_reloadDStream(&(seqState.DStream));
1070 /* gcc and clang both don't like early returns in this loop.
1071 * gcc doesn't like early breaks either.
1072 * Instead save an error and report it at the end.
1073 * When there is an error, don't increment op, so we don't
1074 * overwrite.
1075 */
1076 if (UNLIKELY(ZSTD_isError(oneSeqSize))) error = oneSeqSize;
1077 else op += oneSeqSize;
1078 if (UNLIKELY(!--nbSeq)) break;
1079 }
1080
1081 /* check if reached exact end */
1082 DEBUGLOG(5, "ZSTD_decompressSequences_body: after decode loop, remaining nbSeq : %i", nbSeq);
1083 if (ZSTD_isError(error)) return error;
1084 RETURN_ERROR_IF(nbSeq, corruption_detected, "");
1085 RETURN_ERROR_IF(BIT_reloadDStream(&seqState.DStream) < BIT_DStream_completed, corruption_detected, "");
1086 /* save reps for next block */
1087 { U32 i; for (i=0; i<ZSTD_REP_NUM; i++) dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]); }
1088 }
1089
1090 /* last literal segment */
1091 { size_t const lastLLSize = litEnd - litPtr;
1092 RETURN_ERROR_IF(lastLLSize > (size_t)(oend-op), dstSize_tooSmall, "");
1093 if (op != NULL) {
1094 memcpy(op, litPtr, lastLLSize);
1095 op += lastLLSize;
1096 }
1097 }
1098
1099 return op-ostart;
1100 }
1101
1102 static size_t
1103 ZSTD_decompressSequences_default(ZSTD_DCtx* dctx,
1104 void* dst, size_t maxDstSize,
1105 const void* seqStart, size_t seqSize, int nbSeq,
1106 const ZSTD_longOffset_e isLongOffset,
1107 const int frame)
1108 {
1109 return ZSTD_decompressSequences_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
1110 }
1111 #endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG */
1112
1113 #ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
1114
1115 FORCE_INLINE_TEMPLATE size_t
1116 ZSTD_prefetchMatch(size_t prefixPos, seq_t const sequence,
1117 const BYTE* const prefixStart, const BYTE* const dictEnd)
1118 {
1119 prefixPos += sequence.litLength;
1120 { const BYTE* const matchBase = (sequence.offset > prefixPos) ? dictEnd : prefixStart;
1121 const BYTE* const match = matchBase + prefixPos - sequence.offset; /* note : this operation can overflow when seq.offset is really too large, which can only happen when input is corrupted.
1122 * No consequence though : no memory access will occur, offset is only used for prefetching */
1123 PREFETCH_L1(match); PREFETCH_L1(match + sequence.matchLength - 1); /* note : it's safe to invoke PREFETCH() on any memory address, including invalid ones */
1124 }
1125 return prefixPos + sequence.matchLength;
1126 }
1127
1128 /* This decoding function employs prefetching
1129 * to reduce latency impact of cache misses.
1130 * It's generally employed when block contains a significant portion of long-distance matches
1131 * or when coupled with a "cold" dictionary */
1132 FORCE_INLINE_TEMPLATE size_t
1133 ZSTD_decompressSequencesLong_body(
1134 ZSTD_DCtx* dctx,
1135 void* dst, size_t maxDstSize,
1136 const void* seqStart, size_t seqSize, int nbSeq,
1137 const ZSTD_longOffset_e isLongOffset,
1138 const int frame)
1139 {
1140 const BYTE* ip = (const BYTE*)seqStart;
1141 const BYTE* const iend = ip + seqSize;
1142 BYTE* const ostart = (BYTE* const)dst;
1143 BYTE* const oend = ostart + maxDstSize;
1144 BYTE* op = ostart;
1145 const BYTE* litPtr = dctx->litPtr;
1146 const BYTE* const litEnd = litPtr + dctx->litSize;
1147 const BYTE* const prefixStart = (const BYTE*) (dctx->prefixStart);
1148 const BYTE* const dictStart = (const BYTE*) (dctx->virtualStart);
1149 const BYTE* const dictEnd = (const BYTE*) (dctx->dictEnd);
1150 (void)frame;
1151
1152 /* Regen sequences */
1153 if (nbSeq) {
1154 #define STORED_SEQS 4
1155 #define STORED_SEQS_MASK (STORED_SEQS-1)
1156 #define ADVANCED_SEQS 4
1157 seq_t sequences[STORED_SEQS];
1158 int const seqAdvance = MIN(nbSeq, ADVANCED_SEQS);
1159 seqState_t seqState;
1160 int seqNb;
1161 size_t prefixPos = (size_t)(op-prefixStart); /* track position relative to prefixStart */
1162
1163 dctx->fseEntropy = 1;
1164 { int i; for (i=0; i<ZSTD_REP_NUM; i++) seqState.prevOffset[i] = dctx->entropy.rep[i]; }
1165 assert(dst != NULL);
1166 assert(iend >= ip);
1167 RETURN_ERROR_IF(
1168 ERR_isError(BIT_initDStream(&seqState.DStream, ip, iend-ip)),
1169 corruption_detected, "");
1170 ZSTD_initFseState(&seqState.stateLL, &seqState.DStream, dctx->LLTptr);
1171 ZSTD_initFseState(&seqState.stateOffb, &seqState.DStream, dctx->OFTptr);
1172 ZSTD_initFseState(&seqState.stateML, &seqState.DStream, dctx->MLTptr);
1173
1174 /* prepare in advance */
1175 for (seqNb=0; (BIT_reloadDStream(&seqState.DStream) <= BIT_DStream_completed) && (seqNb<seqAdvance); seqNb++) {
1176 seq_t const sequence = ZSTD_decodeSequence(&seqState, isLongOffset);
1177 prefixPos = ZSTD_prefetchMatch(prefixPos, sequence, prefixStart, dictEnd);
1178 sequences[seqNb] = sequence;
1179 }
1180 RETURN_ERROR_IF(seqNb<seqAdvance, corruption_detected, "");
1181
1182 /* decode and decompress */
1183 for ( ; (BIT_reloadDStream(&(seqState.DStream)) <= BIT_DStream_completed) && (seqNb<nbSeq) ; seqNb++) {
1184 seq_t const sequence = ZSTD_decodeSequence(&seqState, isLongOffset);
1185 size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequences[(seqNb-ADVANCED_SEQS) & STORED_SEQS_MASK], &litPtr, litEnd, prefixStart, dictStart, dictEnd);
1186 #if defined(FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION) && defined(FUZZING_ASSERT_VALID_SEQUENCE)
1187 assert(!ZSTD_isError(oneSeqSize));
1188 if (frame) ZSTD_assertValidSequence(dctx, op, oend, sequences[(seqNb-ADVANCED_SEQS) & STORED_SEQS_MASK], prefixStart, dictStart);
1189 #endif
1190 if (ZSTD_isError(oneSeqSize)) return oneSeqSize;
1191
1192 prefixPos = ZSTD_prefetchMatch(prefixPos, sequence, prefixStart, dictEnd);
1193 sequences[seqNb & STORED_SEQS_MASK] = sequence;
1194 op += oneSeqSize;
1195 }
1196 RETURN_ERROR_IF(seqNb<nbSeq, corruption_detected, "");
1197
1198 /* finish queue */
1199 seqNb -= seqAdvance;
1200 for ( ; seqNb<nbSeq ; seqNb++) {
1201 size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequences[seqNb&STORED_SEQS_MASK], &litPtr, litEnd, prefixStart, dictStart, dictEnd);
1202 #if defined(FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION) && defined(FUZZING_ASSERT_VALID_SEQUENCE)
1203 assert(!ZSTD_isError(oneSeqSize));
1204 if (frame) ZSTD_assertValidSequence(dctx, op, oend, sequences[seqNb&STORED_SEQS_MASK], prefixStart, dictStart);
1205 #endif
1206 if (ZSTD_isError(oneSeqSize)) return oneSeqSize;
1207 op += oneSeqSize;
1208 }
1209
1210 /* save reps for next block */
1211 { U32 i; for (i=0; i<ZSTD_REP_NUM; i++) dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]); }
1212 }
1213
1214 /* last literal segment */
1215 { size_t const lastLLSize = litEnd - litPtr;
1216 RETURN_ERROR_IF(lastLLSize > (size_t)(oend-op), dstSize_tooSmall, "");
1217 if (op != NULL) {
1218 memcpy(op, litPtr, lastLLSize);
1219 op += lastLLSize;
1220 }
1221 }
1222
1223 return op-ostart;
1224 }
1225
1226 static size_t
1227 ZSTD_decompressSequencesLong_default(ZSTD_DCtx* dctx,
1228 void* dst, size_t maxDstSize,
1229 const void* seqStart, size_t seqSize, int nbSeq,
1230 const ZSTD_longOffset_e isLongOffset,
1231 const int frame)
1232 {
1233 return ZSTD_decompressSequencesLong_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
1234 }
1235 #endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT */
1236
1237
1238
1239 #if DYNAMIC_BMI2
1240
1241 #ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
1242 static TARGET_ATTRIBUTE("bmi2") size_t
1243 DONT_VECTORIZE
1244 ZSTD_decompressSequences_bmi2(ZSTD_DCtx* dctx,
1245 void* dst, size_t maxDstSize,
1246 const void* seqStart, size_t seqSize, int nbSeq,
1247 const ZSTD_longOffset_e isLongOffset,
1248 const int frame)
1249 {
1250 return ZSTD_decompressSequences_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
1251 }
1252 #endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG */
1253
1254 #ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
1255 static TARGET_ATTRIBUTE("bmi2") size_t
1256 ZSTD_decompressSequencesLong_bmi2(ZSTD_DCtx* dctx,
1257 void* dst, size_t maxDstSize,
1258 const void* seqStart, size_t seqSize, int nbSeq,
1259 const ZSTD_longOffset_e isLongOffset,
1260 const int frame)
1261 {
1262 return ZSTD_decompressSequencesLong_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
1263 }
1264 #endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT */
1265
1266 #endif /* DYNAMIC_BMI2 */
1267
1268 typedef size_t (*ZSTD_decompressSequences_t)(
1269 ZSTD_DCtx* dctx,
1270 void* dst, size_t maxDstSize,
1271 const void* seqStart, size_t seqSize, int nbSeq,
1272 const ZSTD_longOffset_e isLongOffset,
1273 const int frame);
1274
1275 #ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
1276 static size_t
1277 ZSTD_decompressSequences(ZSTD_DCtx* dctx, void* dst, size_t maxDstSize,
1278 const void* seqStart, size_t seqSize, int nbSeq,
1279 const ZSTD_longOffset_e isLongOffset,
1280 const int frame)
1281 {
1282 DEBUGLOG(5, "ZSTD_decompressSequences");
1283 #if DYNAMIC_BMI2
1284 if (dctx->bmi2) {
1285 return ZSTD_decompressSequences_bmi2(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
1286 }
1287 #endif
1288 return ZSTD_decompressSequences_default(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
1289 }
1290 #endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG */
1291
1292
1293 #ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
1294 /* ZSTD_decompressSequencesLong() :
1295 * decompression function triggered when a minimum share of offsets is considered "long",
1296 * aka out of cache.
1297 * note : "long" definition seems overloaded here, sometimes meaning "wider than bitstream register", and sometimes meaning "farther than memory cache distance".
1298 * This function will try to mitigate main memory latency through the use of prefetching */
1299 static size_t
1300 ZSTD_decompressSequencesLong(ZSTD_DCtx* dctx,
1301 void* dst, size_t maxDstSize,
1302 const void* seqStart, size_t seqSize, int nbSeq,
1303 const ZSTD_longOffset_e isLongOffset,
1304 const int frame)
1305 {
1306 DEBUGLOG(5, "ZSTD_decompressSequencesLong");
1307 #if DYNAMIC_BMI2
1308 if (dctx->bmi2) {
1309 return ZSTD_decompressSequencesLong_bmi2(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
1310 }
1311 #endif
1312 return ZSTD_decompressSequencesLong_default(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
1313 }
1314 #endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT */
1315
1316
1317
1318 #if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
1319 !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
1320 /* ZSTD_getLongOffsetsShare() :
1321 * condition : offTable must be valid
1322 * @return : "share" of long offsets (arbitrarily defined as > (1<<23))
1323 * compared to maximum possible of (1<<OffFSELog) */
1324 static unsigned
1325 ZSTD_getLongOffsetsShare(const ZSTD_seqSymbol* offTable)
1326 {
1327 const void* ptr = offTable;
1328 U32 const tableLog = ((const ZSTD_seqSymbol_header*)ptr)[0].tableLog;
1329 const ZSTD_seqSymbol* table = offTable + 1;
1330 U32 const max = 1 << tableLog;
1331 U32 u, total = 0;
1332 DEBUGLOG(5, "ZSTD_getLongOffsetsShare: (tableLog=%u)", tableLog);
1333
1334 assert(max <= (1 << OffFSELog)); /* max not too large */
1335 for (u=0; u<max; u++) {
1336 if (table[u].nbAdditionalBits > 22) total += 1;
1337 }
1338
1339 assert(tableLog <= OffFSELog);
1340 total <<= (OffFSELog - tableLog); /* scale to OffFSELog */
1341
1342 return total;
1343 }
1344 #endif
1345
1346 size_t
1347 ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx,
1348 void* dst, size_t dstCapacity,
1349 const void* src, size_t srcSize, const int frame)
1350 { /* blockType == blockCompressed */
1351 const BYTE* ip = (const BYTE*)src;
1352 /* isLongOffset must be true if there are long offsets.
1353 * Offsets are long if they are larger than 2^STREAM_ACCUMULATOR_MIN.
1354 * We don't expect that to be the case in 64-bit mode.
1355 * In block mode, window size is not known, so we have to be conservative.
1356 * (note: but it could be evaluated from current-lowLimit)
1357 */
1358 ZSTD_longOffset_e const isLongOffset = (ZSTD_longOffset_e)(MEM_32bits() && (!frame || (dctx->fParams.windowSize > (1ULL << STREAM_ACCUMULATOR_MIN))));
1359 DEBUGLOG(5, "ZSTD_decompressBlock_internal (size : %u)", (U32)srcSize);
1360
1361 RETURN_ERROR_IF(srcSize >= ZSTD_BLOCKSIZE_MAX, srcSize_wrong, "");
1362
1363 /* Decode literals section */
1364 { size_t const litCSize = ZSTD_decodeLiteralsBlock(dctx, src, srcSize);
1365 DEBUGLOG(5, "ZSTD_decodeLiteralsBlock : %u", (U32)litCSize);
1366 if (ZSTD_isError(litCSize)) return litCSize;
1367 ip += litCSize;
1368 srcSize -= litCSize;
1369 }
1370
1371 /* Build Decoding Tables */
1372 {
1373 /* These macros control at build-time which decompressor implementation
1374 * we use. If neither is defined, we do some inspection and dispatch at
1375 * runtime.
1376 */
1377 #if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
1378 !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
1379 int usePrefetchDecoder = dctx->ddictIsCold;
1380 #endif
1381 int nbSeq;
1382 size_t const seqHSize = ZSTD_decodeSeqHeaders(dctx, &nbSeq, ip, srcSize);
1383 if (ZSTD_isError(seqHSize)) return seqHSize;
1384 ip += seqHSize;
1385 srcSize -= seqHSize;
1386
1387 RETURN_ERROR_IF(dst == NULL && nbSeq > 0, dstSize_tooSmall, "NULL not handled");
1388
1389 #if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
1390 !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
1391 if ( !usePrefetchDecoder
1392 && (!frame || (dctx->fParams.windowSize > (1<<24)))
1393 && (nbSeq>ADVANCED_SEQS) ) { /* could probably use a larger nbSeq limit */
1394 U32 const shareLongOffsets = ZSTD_getLongOffsetsShare(dctx->OFTptr);
1395 U32 const minShare = MEM_64bits() ? 7 : 20; /* heuristic values, correspond to 2.73% and 7.81% */
1396 usePrefetchDecoder = (shareLongOffsets >= minShare);
1397 }
1398 #endif
1399
1400 dctx->ddictIsCold = 0;
1401
1402 #if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
1403 !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
1404 if (usePrefetchDecoder)
1405 #endif
1406 #ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
1407 return ZSTD_decompressSequencesLong(dctx, dst, dstCapacity, ip, srcSize, nbSeq, isLongOffset, frame);
1408 #endif
1409
1410 #ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
1411 /* else */
1412 return ZSTD_decompressSequences(dctx, dst, dstCapacity, ip, srcSize, nbSeq, isLongOffset, frame);
1413 #endif
1414 }
1415 }
1416
1417
1418 void ZSTD_checkContinuity(ZSTD_DCtx* dctx, const void* dst)
1419 {
1420 if (dst != dctx->previousDstEnd) { /* not contiguous */
1421 dctx->dictEnd = dctx->previousDstEnd;
1422 dctx->virtualStart = (const char*)dst - ((const char*)(dctx->previousDstEnd) - (const char*)(dctx->prefixStart));
1423 dctx->prefixStart = dst;
1424 dctx->previousDstEnd = dst;
1425 }
1426 }
1427
1428
1429 size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx,
1430 void* dst, size_t dstCapacity,
1431 const void* src, size_t srcSize)
1432 {
1433 size_t dSize;
1434 ZSTD_checkContinuity(dctx, dst);
1435 dSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize, /* frame */ 0);
1436 dctx->previousDstEnd = (char*)dst + dSize;
1437 return dSize;
1438 }
OpenZFS on Linux and FreeBSD