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
12 /* ***************************************************************
13 * Tuning parameters
14 *****************************************************************/
15 /*!
16 * HEAPMODE :
17 * Select how default decompression function ZSTD_decompress() allocates its context,
18 * on stack (0), or into heap (1, default; requires malloc()).
19 * Note that functions with explicit context such as ZSTD_decompressDCtx() are unaffected.
20 */
21 #ifndef ZSTD_HEAPMODE
22 # define ZSTD_HEAPMODE 1
23 #endif
24
25 /*!
26 * LEGACY_SUPPORT :
27 * if set to 1+, ZSTD_decompress() can decode older formats (v0.1+)
28 */
29 #ifndef ZSTD_LEGACY_SUPPORT
30 # define ZSTD_LEGACY_SUPPORT 0
31 #endif
32
33 /*!
34 * MAXWINDOWSIZE_DEFAULT :
35 * maximum window size accepted by DStream __by default__.
36 * Frames requiring more memory will be rejected.
37 * It's possible to set a different limit using ZSTD_DCtx_setMaxWindowSize().
38 */
39 #ifndef ZSTD_MAXWINDOWSIZE_DEFAULT
40 # define ZSTD_MAXWINDOWSIZE_DEFAULT (((U32)1 << ZSTD_WINDOWLOG_LIMIT_DEFAULT) + 1)
41 #endif
42
43 /*!
44 * NO_FORWARD_PROGRESS_MAX :
45 * maximum allowed nb of calls to ZSTD_decompressStream()
46 * without any forward progress
47 * (defined as: no byte read from input, and no byte flushed to output)
48 * before triggering an error.
49 */
50 #ifndef ZSTD_NO_FORWARD_PROGRESS_MAX
51 # define ZSTD_NO_FORWARD_PROGRESS_MAX 16
52 #endif
53
54
55 /*-*******************************************************
56 * Dependencies
57 *********************************************************/
58 #include "../common/zstd_deps.h" /* ZSTD_memcpy, ZSTD_memmove, ZSTD_memset */
59 #include "../common/cpu.h" /* bmi2 */
60 #include "../common/mem.h" /* low level memory routines */
61 #define FSE_STATIC_LINKING_ONLY
62 #include "../common/fse.h"
63 #define HUF_STATIC_LINKING_ONLY
64 #include "../common/huf.h"
65 #include "../common/xxhash.h" /* XXH64_reset, XXH64_update, XXH64_digest, XXH64 */
66 #include "../common/zstd_internal.h" /* blockProperties_t */
67 #include "zstd_decompress_internal.h" /* ZSTD_DCtx */
68 #include "zstd_ddict.h" /* ZSTD_DDictDictContent */
69 #include "zstd_decompress_block.h" /* ZSTD_decompressBlock_internal */
70
71 #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT>=1)
72 # include "../legacy/zstd_legacy.h"
73 #endif
74
75
76
77 /*************************************
78 * Multiple DDicts Hashset internals *
79 *************************************/
80
81 #define DDICT_HASHSET_MAX_LOAD_FACTOR_COUNT_MULT 4
82 #define DDICT_HASHSET_MAX_LOAD_FACTOR_SIZE_MULT 3 /* These two constants represent SIZE_MULT/COUNT_MULT load factor without using a float.
83 * Currently, that means a 0.75 load factor.
84 * So, if count * COUNT_MULT / size * SIZE_MULT != 0, then we've exceeded
85 * the load factor of the ddict hash set.
86 */
87
88 #define DDICT_HASHSET_TABLE_BASE_SIZE 64
89 #define DDICT_HASHSET_RESIZE_FACTOR 2
90
91 /* Hash function to determine starting position of dict insertion within the table
92 * Returns an index between [0, hashSet->ddictPtrTableSize]
93 */
ZSTD_DDictHashSet_getIndex(const ZSTD_DDictHashSet * hashSet,U32 dictID)94 static size_t ZSTD_DDictHashSet_getIndex(const ZSTD_DDictHashSet* hashSet, U32 dictID) {
95 const U64 hash = XXH64(&dictID, sizeof(U32), 0);
96 /* DDict ptr table size is a multiple of 2, use size - 1 as mask to get index within [0, hashSet->ddictPtrTableSize) */
97 return hash & (hashSet->ddictPtrTableSize - 1);
98 }
99
100 /* Adds DDict to a hashset without resizing it.
101 * If inserting a DDict with a dictID that already exists in the set, replaces the one in the set.
102 * Returns 0 if successful, or a zstd error code if something went wrong.
103 */
ZSTD_DDictHashSet_emplaceDDict(ZSTD_DDictHashSet * hashSet,const ZSTD_DDict * ddict)104 static size_t ZSTD_DDictHashSet_emplaceDDict(ZSTD_DDictHashSet* hashSet, const ZSTD_DDict* ddict) {
105 const U32 dictID = ZSTD_getDictID_fromDDict(ddict);
106 size_t idx = ZSTD_DDictHashSet_getIndex(hashSet, dictID);
107 const size_t idxRangeMask = hashSet->ddictPtrTableSize - 1;
108 RETURN_ERROR_IF(hashSet->ddictPtrCount == hashSet->ddictPtrTableSize, GENERIC, "Hash set is full!");
109 DEBUGLOG(4, "Hashed index: for dictID: %u is %zu", dictID, idx);
110 while (hashSet->ddictPtrTable[idx] != NULL) {
111 /* Replace existing ddict if inserting ddict with same dictID */
112 if (ZSTD_getDictID_fromDDict(hashSet->ddictPtrTable[idx]) == dictID) {
113 DEBUGLOG(4, "DictID already exists, replacing rather than adding");
114 hashSet->ddictPtrTable[idx] = ddict;
115 return 0;
116 }
117 idx &= idxRangeMask;
118 idx++;
119 }
120 DEBUGLOG(4, "Final idx after probing for dictID %u is: %zu", dictID, idx);
121 hashSet->ddictPtrTable[idx] = ddict;
122 hashSet->ddictPtrCount++;
123 return 0;
124 }
125
126 /* Expands hash table by factor of DDICT_HASHSET_RESIZE_FACTOR and
127 * rehashes all values, allocates new table, frees old table.
128 * Returns 0 on success, otherwise a zstd error code.
129 */
ZSTD_DDictHashSet_expand(ZSTD_DDictHashSet * hashSet,ZSTD_customMem customMem)130 static size_t ZSTD_DDictHashSet_expand(ZSTD_DDictHashSet* hashSet, ZSTD_customMem customMem) {
131 size_t newTableSize = hashSet->ddictPtrTableSize * DDICT_HASHSET_RESIZE_FACTOR;
132 const ZSTD_DDict** newTable = (const ZSTD_DDict**)ZSTD_customCalloc(sizeof(ZSTD_DDict*) * newTableSize, customMem);
133 const ZSTD_DDict** oldTable = hashSet->ddictPtrTable;
134 size_t oldTableSize = hashSet->ddictPtrTableSize;
135 size_t i;
136
137 DEBUGLOG(4, "Expanding DDict hash table! Old size: %zu new size: %zu", oldTableSize, newTableSize);
138 RETURN_ERROR_IF(!newTable, memory_allocation, "Expanded hashset allocation failed!");
139 hashSet->ddictPtrTable = newTable;
140 hashSet->ddictPtrTableSize = newTableSize;
141 hashSet->ddictPtrCount = 0;
142 for (i = 0; i < oldTableSize; ++i) {
143 if (oldTable[i] != NULL) {
144 FORWARD_IF_ERROR(ZSTD_DDictHashSet_emplaceDDict(hashSet, oldTable[i]), "");
145 }
146 }
147 ZSTD_customFree((void*)oldTable, customMem);
148 DEBUGLOG(4, "Finished re-hash");
149 return 0;
150 }
151
152 /* Fetches a DDict with the given dictID
153 * Returns the ZSTD_DDict* with the requested dictID. If it doesn't exist, then returns NULL.
154 */
ZSTD_DDictHashSet_getDDict(ZSTD_DDictHashSet * hashSet,U32 dictID)155 static const ZSTD_DDict* ZSTD_DDictHashSet_getDDict(ZSTD_DDictHashSet* hashSet, U32 dictID) {
156 size_t idx = ZSTD_DDictHashSet_getIndex(hashSet, dictID);
157 const size_t idxRangeMask = hashSet->ddictPtrTableSize - 1;
158 DEBUGLOG(4, "Hashed index: for dictID: %u is %zu", dictID, idx);
159 for (;;) {
160 size_t currDictID = ZSTD_getDictID_fromDDict(hashSet->ddictPtrTable[idx]);
161 if (currDictID == dictID || currDictID == 0) {
162 /* currDictID == 0 implies a NULL ddict entry */
163 break;
164 } else {
165 idx &= idxRangeMask; /* Goes to start of table when we reach the end */
166 idx++;
167 }
168 }
169 DEBUGLOG(4, "Final idx after probing for dictID %u is: %zu", dictID, idx);
170 return hashSet->ddictPtrTable[idx];
171 }
172
173 /* Allocates space for and returns a ddict hash set
174 * The hash set's ZSTD_DDict* table has all values automatically set to NULL to begin with.
175 * Returns NULL if allocation failed.
176 */
ZSTD_createDDictHashSet(ZSTD_customMem customMem)177 static ZSTD_DDictHashSet* ZSTD_createDDictHashSet(ZSTD_customMem customMem) {
178 ZSTD_DDictHashSet* ret = (ZSTD_DDictHashSet*)ZSTD_customMalloc(sizeof(ZSTD_DDictHashSet), customMem);
179 DEBUGLOG(4, "Allocating new hash set");
180 ret->ddictPtrTable = (const ZSTD_DDict**)ZSTD_customCalloc(DDICT_HASHSET_TABLE_BASE_SIZE * sizeof(ZSTD_DDict*), customMem);
181 ret->ddictPtrTableSize = DDICT_HASHSET_TABLE_BASE_SIZE;
182 ret->ddictPtrCount = 0;
183 if (!ret || !ret->ddictPtrTable) {
184 return NULL;
185 }
186 return ret;
187 }
188
189 /* Frees the table of ZSTD_DDict* within a hashset, then frees the hashset itself.
190 * Note: The ZSTD_DDict* within the table are NOT freed.
191 */
ZSTD_freeDDictHashSet(ZSTD_DDictHashSet * hashSet,ZSTD_customMem customMem)192 static void ZSTD_freeDDictHashSet(ZSTD_DDictHashSet* hashSet, ZSTD_customMem customMem) {
193 DEBUGLOG(4, "Freeing ddict hash set");
194 if (hashSet && hashSet->ddictPtrTable) {
195 ZSTD_customFree((void*)hashSet->ddictPtrTable, customMem);
196 }
197 if (hashSet) {
198 ZSTD_customFree(hashSet, customMem);
199 }
200 }
201
202 /* Public function: Adds a DDict into the ZSTD_DDictHashSet, possibly triggering a resize of the hash set.
203 * Returns 0 on success, or a ZSTD error.
204 */
ZSTD_DDictHashSet_addDDict(ZSTD_DDictHashSet * hashSet,const ZSTD_DDict * ddict,ZSTD_customMem customMem)205 static size_t ZSTD_DDictHashSet_addDDict(ZSTD_DDictHashSet* hashSet, const ZSTD_DDict* ddict, ZSTD_customMem customMem) {
206 DEBUGLOG(4, "Adding dict ID: %u to hashset with - Count: %zu Tablesize: %zu", ZSTD_getDictID_fromDDict(ddict), hashSet->ddictPtrCount, hashSet->ddictPtrTableSize);
207 if (hashSet->ddictPtrCount * DDICT_HASHSET_MAX_LOAD_FACTOR_COUNT_MULT / hashSet->ddictPtrTableSize * DDICT_HASHSET_MAX_LOAD_FACTOR_SIZE_MULT != 0) {
208 FORWARD_IF_ERROR(ZSTD_DDictHashSet_expand(hashSet, customMem), "");
209 }
210 FORWARD_IF_ERROR(ZSTD_DDictHashSet_emplaceDDict(hashSet, ddict), "");
211 return 0;
212 }
213
214 /*-*************************************************************
215 * Context management
216 ***************************************************************/
ZSTD_sizeof_DCtx(const ZSTD_DCtx * dctx)217 size_t ZSTD_sizeof_DCtx (const ZSTD_DCtx* dctx)
218 {
219 if (dctx==NULL) return 0; /* support sizeof NULL */
220 return sizeof(*dctx)
221 + ZSTD_sizeof_DDict(dctx->ddictLocal)
222 + dctx->inBuffSize + dctx->outBuffSize;
223 }
224
ZSTD_estimateDCtxSize(void)225 size_t ZSTD_estimateDCtxSize(void) { return sizeof(ZSTD_DCtx); }
226
227
ZSTD_startingInputLength(ZSTD_format_e format)228 static size_t ZSTD_startingInputLength(ZSTD_format_e format)
229 {
230 size_t const startingInputLength = ZSTD_FRAMEHEADERSIZE_PREFIX(format);
231 /* only supports formats ZSTD_f_zstd1 and ZSTD_f_zstd1_magicless */
232 assert( (format == ZSTD_f_zstd1) || (format == ZSTD_f_zstd1_magicless) );
233 return startingInputLength;
234 }
235
ZSTD_DCtx_resetParameters(ZSTD_DCtx * dctx)236 static void ZSTD_DCtx_resetParameters(ZSTD_DCtx* dctx)
237 {
238 assert(dctx->streamStage == zdss_init);
239 dctx->format = ZSTD_f_zstd1;
240 dctx->maxWindowSize = ZSTD_MAXWINDOWSIZE_DEFAULT;
241 dctx->outBufferMode = ZSTD_bm_buffered;
242 dctx->forceIgnoreChecksum = ZSTD_d_validateChecksum;
243 dctx->refMultipleDDicts = ZSTD_rmd_refSingleDDict;
244 }
245
ZSTD_initDCtx_internal(ZSTD_DCtx * dctx)246 static void ZSTD_initDCtx_internal(ZSTD_DCtx* dctx)
247 {
248 dctx->staticSize = 0;
249 dctx->ddict = NULL;
250 dctx->ddictLocal = NULL;
251 dctx->dictEnd = NULL;
252 dctx->ddictIsCold = 0;
253 dctx->dictUses = ZSTD_dont_use;
254 dctx->inBuff = NULL;
255 dctx->inBuffSize = 0;
256 dctx->outBuffSize = 0;
257 dctx->streamStage = zdss_init;
258 dctx->legacyContext = NULL;
259 dctx->previousLegacyVersion = 0;
260 dctx->noForwardProgress = 0;
261 dctx->oversizedDuration = 0;
262 dctx->bmi2 = ZSTD_cpuid_bmi2(ZSTD_cpuid());
263 dctx->ddictSet = NULL;
264 ZSTD_DCtx_resetParameters(dctx);
265 #ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
266 dctx->dictContentEndForFuzzing = NULL;
267 #endif
268 }
269
ZSTD_initStaticDCtx(void * workspace,size_t workspaceSize)270 ZSTD_DCtx* ZSTD_initStaticDCtx(void *workspace, size_t workspaceSize)
271 {
272 ZSTD_DCtx* const dctx = (ZSTD_DCtx*) workspace;
273
274 if ((size_t)workspace & 7) return NULL; /* 8-aligned */
275 if (workspaceSize < sizeof(ZSTD_DCtx)) return NULL; /* minimum size */
276
277 ZSTD_initDCtx_internal(dctx);
278 dctx->staticSize = workspaceSize;
279 dctx->inBuff = (char*)(dctx+1);
280 return dctx;
281 }
282
ZSTD_createDCtx_advanced(ZSTD_customMem customMem)283 ZSTD_DCtx* ZSTD_createDCtx_advanced(ZSTD_customMem customMem)
284 {
285 if ((!customMem.customAlloc) ^ (!customMem.customFree)) return NULL;
286
287 { ZSTD_DCtx* const dctx = (ZSTD_DCtx*)ZSTD_customMalloc(sizeof(*dctx), customMem);
288 if (!dctx) return NULL;
289 dctx->customMem = customMem;
290 ZSTD_initDCtx_internal(dctx);
291 return dctx;
292 }
293 }
294
ZSTD_createDCtx(void)295 ZSTD_DCtx* ZSTD_createDCtx(void)
296 {
297 DEBUGLOG(3, "ZSTD_createDCtx");
298 return ZSTD_createDCtx_advanced(ZSTD_defaultCMem);
299 }
300
ZSTD_clearDict(ZSTD_DCtx * dctx)301 static void ZSTD_clearDict(ZSTD_DCtx* dctx)
302 {
303 ZSTD_freeDDict(dctx->ddictLocal);
304 dctx->ddictLocal = NULL;
305 dctx->ddict = NULL;
306 dctx->dictUses = ZSTD_dont_use;
307 }
308
ZSTD_freeDCtx(ZSTD_DCtx * dctx)309 size_t ZSTD_freeDCtx(ZSTD_DCtx* dctx)
310 {
311 if (dctx==NULL) return 0; /* support free on NULL */
312 RETURN_ERROR_IF(dctx->staticSize, memory_allocation, "not compatible with static DCtx");
313 { ZSTD_customMem const cMem = dctx->customMem;
314 ZSTD_clearDict(dctx);
315 ZSTD_customFree(dctx->inBuff, cMem);
316 dctx->inBuff = NULL;
317 #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT >= 1)
318 if (dctx->legacyContext)
319 ZSTD_freeLegacyStreamContext(dctx->legacyContext, dctx->previousLegacyVersion);
320 #endif
321 if (dctx->ddictSet) {
322 ZSTD_freeDDictHashSet(dctx->ddictSet, cMem);
323 dctx->ddictSet = NULL;
324 }
325 ZSTD_customFree(dctx, cMem);
326 return 0;
327 }
328 }
329
330 /* no longer useful */
ZSTD_copyDCtx(ZSTD_DCtx * dstDCtx,const ZSTD_DCtx * srcDCtx)331 void ZSTD_copyDCtx(ZSTD_DCtx* dstDCtx, const ZSTD_DCtx* srcDCtx)
332 {
333 size_t const toCopy = (size_t)((char*)(&dstDCtx->inBuff) - (char*)dstDCtx);
334 ZSTD_memcpy(dstDCtx, srcDCtx, toCopy); /* no need to copy workspace */
335 }
336
337 /* Given a dctx with a digested frame params, re-selects the correct ZSTD_DDict based on
338 * the requested dict ID from the frame. If there exists a reference to the correct ZSTD_DDict, then
339 * accordingly sets the ddict to be used to decompress the frame.
340 *
341 * If no DDict is found, then no action is taken, and the ZSTD_DCtx::ddict remains as-is.
342 *
343 * ZSTD_d_refMultipleDDicts must be enabled for this function to be called.
344 */
ZSTD_DCtx_selectFrameDDict(ZSTD_DCtx * dctx)345 static void ZSTD_DCtx_selectFrameDDict(ZSTD_DCtx* dctx) {
346 assert(dctx->refMultipleDDicts && dctx->ddictSet);
347 DEBUGLOG(4, "Adjusting DDict based on requested dict ID from frame");
348 if (dctx->ddict) {
349 const ZSTD_DDict* frameDDict = ZSTD_DDictHashSet_getDDict(dctx->ddictSet, dctx->fParams.dictID);
350 if (frameDDict) {
351 DEBUGLOG(4, "DDict found!");
352 ZSTD_clearDict(dctx);
353 dctx->dictID = dctx->fParams.dictID;
354 dctx->ddict = frameDDict;
355 dctx->dictUses = ZSTD_use_indefinitely;
356 }
357 }
358 }
359
360
361 /*-*************************************************************
362 * Frame header decoding
363 ***************************************************************/
364
365 /*! ZSTD_isFrame() :
366 * Tells if the content of `buffer` starts with a valid Frame Identifier.
367 * Note : Frame Identifier is 4 bytes. If `size < 4`, @return will always be 0.
368 * Note 2 : Legacy Frame Identifiers are considered valid only if Legacy Support is enabled.
369 * Note 3 : Skippable Frame Identifiers are considered valid. */
ZSTD_isFrame(const void * buffer,size_t size)370 unsigned ZSTD_isFrame(const void* buffer, size_t size)
371 {
372 if (size < ZSTD_FRAMEIDSIZE) return 0;
373 { U32 const magic = MEM_readLE32(buffer);
374 if (magic == ZSTD_MAGICNUMBER) return 1;
375 if ((magic & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) return 1;
376 }
377 #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT >= 1)
378 if (ZSTD_isLegacy(buffer, size)) return 1;
379 #endif
380 return 0;
381 }
382
383 /** ZSTD_frameHeaderSize_internal() :
384 * srcSize must be large enough to reach header size fields.
385 * note : only works for formats ZSTD_f_zstd1 and ZSTD_f_zstd1_magicless.
386 * @return : size of the Frame Header
387 * or an error code, which can be tested with ZSTD_isError() */
ZSTD_frameHeaderSize_internal(const void * src,size_t srcSize,ZSTD_format_e format)388 static size_t ZSTD_frameHeaderSize_internal(const void* src, size_t srcSize, ZSTD_format_e format)
389 {
390 size_t const minInputSize = ZSTD_startingInputLength(format);
391 RETURN_ERROR_IF(srcSize < minInputSize, srcSize_wrong, "");
392
393 { BYTE const fhd = ((const BYTE*)src)[minInputSize-1];
394 U32 const dictID= fhd & 3;
395 U32 const singleSegment = (fhd >> 5) & 1;
396 U32 const fcsId = fhd >> 6;
397 return minInputSize + !singleSegment
398 + ZSTD_did_fieldSize[dictID] + ZSTD_fcs_fieldSize[fcsId]
399 + (singleSegment && !fcsId);
400 }
401 }
402
403 /** ZSTD_frameHeaderSize() :
404 * srcSize must be >= ZSTD_frameHeaderSize_prefix.
405 * @return : size of the Frame Header,
406 * or an error code (if srcSize is too small) */
ZSTD_frameHeaderSize(const void * src,size_t srcSize)407 size_t ZSTD_frameHeaderSize(const void* src, size_t srcSize)
408 {
409 return ZSTD_frameHeaderSize_internal(src, srcSize, ZSTD_f_zstd1);
410 }
411
412
413 /** ZSTD_getFrameHeader_advanced() :
414 * decode Frame Header, or require larger `srcSize`.
415 * note : only works for formats ZSTD_f_zstd1 and ZSTD_f_zstd1_magicless
416 * @return : 0, `zfhPtr` is correctly filled,
417 * >0, `srcSize` is too small, value is wanted `srcSize` amount,
418 * or an error code, which can be tested using ZSTD_isError() */
ZSTD_getFrameHeader_advanced(ZSTD_frameHeader * zfhPtr,const void * src,size_t srcSize,ZSTD_format_e format)419 size_t ZSTD_getFrameHeader_advanced(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format)
420 {
421 const BYTE* ip = (const BYTE*)src;
422 size_t const minInputSize = ZSTD_startingInputLength(format);
423
424 ZSTD_memset(zfhPtr, 0, sizeof(*zfhPtr)); /* not strictly necessary, but static analyzer do not understand that zfhPtr is only going to be read only if return value is zero, since they are 2 different signals */
425 if (srcSize < minInputSize) return minInputSize;
426 RETURN_ERROR_IF(src==NULL, GENERIC, "invalid parameter");
427
428 if ( (format != ZSTD_f_zstd1_magicless)
429 && (MEM_readLE32(src) != ZSTD_MAGICNUMBER) ) {
430 if ((MEM_readLE32(src) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
431 /* skippable frame */
432 if (srcSize < ZSTD_SKIPPABLEHEADERSIZE)
433 return ZSTD_SKIPPABLEHEADERSIZE; /* magic number + frame length */
434 ZSTD_memset(zfhPtr, 0, sizeof(*zfhPtr));
435 zfhPtr->frameContentSize = MEM_readLE32((const char *)src + ZSTD_FRAMEIDSIZE);
436 zfhPtr->frameType = ZSTD_skippableFrame;
437 return 0;
438 }
439 RETURN_ERROR(prefix_unknown, "");
440 }
441
442 /* ensure there is enough `srcSize` to fully read/decode frame header */
443 { size_t const fhsize = ZSTD_frameHeaderSize_internal(src, srcSize, format);
444 if (srcSize < fhsize) return fhsize;
445 zfhPtr->headerSize = (U32)fhsize;
446 }
447
448 { BYTE const fhdByte = ip[minInputSize-1];
449 size_t pos = minInputSize;
450 U32 const dictIDSizeCode = fhdByte&3;
451 U32 const checksumFlag = (fhdByte>>2)&1;
452 U32 const singleSegment = (fhdByte>>5)&1;
453 U32 const fcsID = fhdByte>>6;
454 U64 windowSize = 0;
455 U32 dictID = 0;
456 U64 frameContentSize = ZSTD_CONTENTSIZE_UNKNOWN;
457 RETURN_ERROR_IF((fhdByte & 0x08) != 0, frameParameter_unsupported,
458 "reserved bits, must be zero");
459
460 if (!singleSegment) {
461 BYTE const wlByte = ip[pos++];
462 U32 const windowLog = (wlByte >> 3) + ZSTD_WINDOWLOG_ABSOLUTEMIN;
463 RETURN_ERROR_IF(windowLog > ZSTD_WINDOWLOG_MAX, frameParameter_windowTooLarge, "");
464 windowSize = (1ULL << windowLog);
465 windowSize += (windowSize >> 3) * (wlByte&7);
466 }
467 switch(dictIDSizeCode)
468 {
469 default: assert(0); /* impossible */
470 case 0 : break;
471 case 1 : dictID = ip[pos]; pos++; break;
472 case 2 : dictID = MEM_readLE16(ip+pos); pos+=2; break;
473 case 3 : dictID = MEM_readLE32(ip+pos); pos+=4; break;
474 }
475 switch(fcsID)
476 {
477 default: assert(0); /* impossible */
478 case 0 : if (singleSegment) frameContentSize = ip[pos]; break;
479 case 1 : frameContentSize = MEM_readLE16(ip+pos)+256; break;
480 case 2 : frameContentSize = MEM_readLE32(ip+pos); break;
481 case 3 : frameContentSize = MEM_readLE64(ip+pos); break;
482 }
483 if (singleSegment) windowSize = frameContentSize;
484
485 zfhPtr->frameType = ZSTD_frame;
486 zfhPtr->frameContentSize = frameContentSize;
487 zfhPtr->windowSize = windowSize;
488 zfhPtr->blockSizeMax = (unsigned) MIN(windowSize, ZSTD_BLOCKSIZE_MAX);
489 zfhPtr->dictID = dictID;
490 zfhPtr->checksumFlag = checksumFlag;
491 }
492 return 0;
493 }
494
495 /** ZSTD_getFrameHeader() :
496 * decode Frame Header, or require larger `srcSize`.
497 * note : this function does not consume input, it only reads it.
498 * @return : 0, `zfhPtr` is correctly filled,
499 * >0, `srcSize` is too small, value is wanted `srcSize` amount,
500 * or an error code, which can be tested using ZSTD_isError() */
ZSTD_getFrameHeader(ZSTD_frameHeader * zfhPtr,const void * src,size_t srcSize)501 size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize)
502 {
503 return ZSTD_getFrameHeader_advanced(zfhPtr, src, srcSize, ZSTD_f_zstd1);
504 }
505
506
507 /** ZSTD_getFrameContentSize() :
508 * compatible with legacy mode
509 * @return : decompressed size of the single frame pointed to be `src` if known, otherwise
510 * - ZSTD_CONTENTSIZE_UNKNOWN if the size cannot be determined
511 * - ZSTD_CONTENTSIZE_ERROR if an error occurred (e.g. invalid magic number, srcSize too small) */
ZSTD_getFrameContentSize(const void * src,size_t srcSize)512 unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize)
513 {
514 #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT >= 1)
515 if (ZSTD_isLegacy(src, srcSize)) {
516 unsigned long long const ret = ZSTD_getDecompressedSize_legacy(src, srcSize);
517 return ret == 0 ? ZSTD_CONTENTSIZE_UNKNOWN : ret;
518 }
519 #endif
520 { ZSTD_frameHeader zfh;
521 if (ZSTD_getFrameHeader(&zfh, src, srcSize) != 0)
522 return ZSTD_CONTENTSIZE_ERROR;
523 if (zfh.frameType == ZSTD_skippableFrame) {
524 return 0;
525 } else {
526 return zfh.frameContentSize;
527 } }
528 }
529
readSkippableFrameSize(void const * src,size_t srcSize)530 static size_t readSkippableFrameSize(void const* src, size_t srcSize)
531 {
532 size_t const skippableHeaderSize = ZSTD_SKIPPABLEHEADERSIZE;
533 U32 sizeU32;
534
535 RETURN_ERROR_IF(srcSize < ZSTD_SKIPPABLEHEADERSIZE, srcSize_wrong, "");
536
537 sizeU32 = MEM_readLE32((BYTE const*)src + ZSTD_FRAMEIDSIZE);
538 RETURN_ERROR_IF((U32)(sizeU32 + ZSTD_SKIPPABLEHEADERSIZE) < sizeU32,
539 frameParameter_unsupported, "");
540 {
541 size_t const skippableSize = skippableHeaderSize + sizeU32;
542 RETURN_ERROR_IF(skippableSize > srcSize, srcSize_wrong, "");
543 return skippableSize;
544 }
545 }
546
547 /** ZSTD_findDecompressedSize() :
548 * compatible with legacy mode
549 * `srcSize` must be the exact length of some number of ZSTD compressed and/or
550 * skippable frames
551 * @return : decompressed size of the frames contained */
ZSTD_findDecompressedSize(const void * src,size_t srcSize)552 unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize)
553 {
554 unsigned long long totalDstSize = 0;
555
556 while (srcSize >= ZSTD_startingInputLength(ZSTD_f_zstd1)) {
557 U32 const magicNumber = MEM_readLE32(src);
558
559 if ((magicNumber & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
560 size_t const skippableSize = readSkippableFrameSize(src, srcSize);
561 if (ZSTD_isError(skippableSize)) {
562 return ZSTD_CONTENTSIZE_ERROR;
563 }
564 assert(skippableSize <= srcSize);
565
566 src = (const BYTE *)src + skippableSize;
567 srcSize -= skippableSize;
568 continue;
569 }
570
571 { unsigned long long const ret = ZSTD_getFrameContentSize(src, srcSize);
572 if (ret >= ZSTD_CONTENTSIZE_ERROR) return ret;
573
574 /* check for overflow */
575 if (totalDstSize + ret < totalDstSize) return ZSTD_CONTENTSIZE_ERROR;
576 totalDstSize += ret;
577 }
578 { size_t const frameSrcSize = ZSTD_findFrameCompressedSize(src, srcSize);
579 if (ZSTD_isError(frameSrcSize)) {
580 return ZSTD_CONTENTSIZE_ERROR;
581 }
582
583 src = (const BYTE *)src + frameSrcSize;
584 srcSize -= frameSrcSize;
585 }
586 } /* while (srcSize >= ZSTD_frameHeaderSize_prefix) */
587
588 if (srcSize) return ZSTD_CONTENTSIZE_ERROR;
589
590 return totalDstSize;
591 }
592
593 /** ZSTD_getDecompressedSize() :
594 * compatible with legacy mode
595 * @return : decompressed size if known, 0 otherwise
596 note : 0 can mean any of the following :
597 - frame content is empty
598 - decompressed size field is not present in frame header
599 - frame header unknown / not supported
600 - frame header not complete (`srcSize` too small) */
ZSTD_getDecompressedSize(const void * src,size_t srcSize)601 unsigned long long ZSTD_getDecompressedSize(const void* src, size_t srcSize)
602 {
603 unsigned long long const ret = ZSTD_getFrameContentSize(src, srcSize);
604 ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_ERROR < ZSTD_CONTENTSIZE_UNKNOWN);
605 return (ret >= ZSTD_CONTENTSIZE_ERROR) ? 0 : ret;
606 }
607
608
609 /** ZSTD_decodeFrameHeader() :
610 * `headerSize` must be the size provided by ZSTD_frameHeaderSize().
611 * If multiple DDict references are enabled, also will choose the correct DDict to use.
612 * @return : 0 if success, or an error code, which can be tested using ZSTD_isError() */
ZSTD_decodeFrameHeader(ZSTD_DCtx * dctx,const void * src,size_t headerSize)613 static size_t ZSTD_decodeFrameHeader(ZSTD_DCtx* dctx, const void* src, size_t headerSize)
614 {
615 size_t const result = ZSTD_getFrameHeader_advanced(&(dctx->fParams), src, headerSize, dctx->format);
616 if (ZSTD_isError(result)) return result; /* invalid header */
617 RETURN_ERROR_IF(result>0, srcSize_wrong, "headerSize too small");
618
619 /* Reference DDict requested by frame if dctx references multiple ddicts */
620 if (dctx->refMultipleDDicts == ZSTD_rmd_refMultipleDDicts && dctx->ddictSet) {
621 ZSTD_DCtx_selectFrameDDict(dctx);
622 }
623
624 #ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
625 /* Skip the dictID check in fuzzing mode, because it makes the search
626 * harder.
627 */
628 RETURN_ERROR_IF(dctx->fParams.dictID && (dctx->dictID != dctx->fParams.dictID),
629 dictionary_wrong, "");
630 #endif
631 dctx->validateChecksum = (dctx->fParams.checksumFlag && !dctx->forceIgnoreChecksum) ? 1 : 0;
632 if (dctx->validateChecksum) XXH64_reset(&dctx->xxhState, 0);
633 dctx->processedCSize += headerSize;
634 return 0;
635 }
636
ZSTD_errorFrameSizeInfo(size_t ret)637 static ZSTD_frameSizeInfo ZSTD_errorFrameSizeInfo(size_t ret)
638 {
639 ZSTD_frameSizeInfo frameSizeInfo;
640 frameSizeInfo.compressedSize = ret;
641 frameSizeInfo.decompressedBound = ZSTD_CONTENTSIZE_ERROR;
642 return frameSizeInfo;
643 }
644
ZSTD_findFrameSizeInfo(const void * src,size_t srcSize)645 static ZSTD_frameSizeInfo ZSTD_findFrameSizeInfo(const void* src, size_t srcSize)
646 {
647 ZSTD_frameSizeInfo frameSizeInfo;
648 ZSTD_memset(&frameSizeInfo, 0, sizeof(ZSTD_frameSizeInfo));
649
650 #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT >= 1)
651 if (ZSTD_isLegacy(src, srcSize))
652 return ZSTD_findFrameSizeInfoLegacy(src, srcSize);
653 #endif
654
655 if ((srcSize >= ZSTD_SKIPPABLEHEADERSIZE)
656 && (MEM_readLE32(src) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
657 frameSizeInfo.compressedSize = readSkippableFrameSize(src, srcSize);
658 assert(ZSTD_isError(frameSizeInfo.compressedSize) ||
659 frameSizeInfo.compressedSize <= srcSize);
660 return frameSizeInfo;
661 } else {
662 const BYTE* ip = (const BYTE*)src;
663 const BYTE* const ipstart = ip;
664 size_t remainingSize = srcSize;
665 size_t nbBlocks = 0;
666 ZSTD_frameHeader zfh;
667
668 /* Extract Frame Header */
669 { size_t const ret = ZSTD_getFrameHeader(&zfh, src, srcSize);
670 if (ZSTD_isError(ret))
671 return ZSTD_errorFrameSizeInfo(ret);
672 if (ret > 0)
673 return ZSTD_errorFrameSizeInfo(ERROR(srcSize_wrong));
674 }
675
676 ip += zfh.headerSize;
677 remainingSize -= zfh.headerSize;
678
679 /* Iterate over each block */
680 while (1) {
681 blockProperties_t blockProperties;
682 size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSize, &blockProperties);
683 if (ZSTD_isError(cBlockSize))
684 return ZSTD_errorFrameSizeInfo(cBlockSize);
685
686 if (ZSTD_blockHeaderSize + cBlockSize > remainingSize)
687 return ZSTD_errorFrameSizeInfo(ERROR(srcSize_wrong));
688
689 ip += ZSTD_blockHeaderSize + cBlockSize;
690 remainingSize -= ZSTD_blockHeaderSize + cBlockSize;
691 nbBlocks++;
692
693 if (blockProperties.lastBlock) break;
694 }
695
696 /* Final frame content checksum */
697 if (zfh.checksumFlag) {
698 if (remainingSize < 4)
699 return ZSTD_errorFrameSizeInfo(ERROR(srcSize_wrong));
700 ip += 4;
701 }
702
703 frameSizeInfo.compressedSize = (size_t)(ip - ipstart);
704 frameSizeInfo.decompressedBound = (zfh.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN)
705 ? zfh.frameContentSize
706 : nbBlocks * zfh.blockSizeMax;
707 return frameSizeInfo;
708 }
709 }
710
711 /** ZSTD_findFrameCompressedSize() :
712 * compatible with legacy mode
713 * `src` must point to the start of a ZSTD frame, ZSTD legacy frame, or skippable frame
714 * `srcSize` must be at least as large as the frame contained
715 * @return : the compressed size of the frame starting at `src` */
ZSTD_findFrameCompressedSize(const void * src,size_t srcSize)716 size_t ZSTD_findFrameCompressedSize(const void *src, size_t srcSize)
717 {
718 ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize);
719 return frameSizeInfo.compressedSize;
720 }
721
722 /** ZSTD_decompressBound() :
723 * compatible with legacy mode
724 * `src` must point to the start of a ZSTD frame or a skippeable frame
725 * `srcSize` must be at least as large as the frame contained
726 * @return : the maximum decompressed size of the compressed source
727 */
ZSTD_decompressBound(const void * src,size_t srcSize)728 unsigned long long ZSTD_decompressBound(const void* src, size_t srcSize)
729 {
730 unsigned long long bound = 0;
731 /* Iterate over each frame */
732 while (srcSize > 0) {
733 ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize);
734 size_t const compressedSize = frameSizeInfo.compressedSize;
735 unsigned long long const decompressedBound = frameSizeInfo.decompressedBound;
736 if (ZSTD_isError(compressedSize) || decompressedBound == ZSTD_CONTENTSIZE_ERROR)
737 return ZSTD_CONTENTSIZE_ERROR;
738 assert(srcSize >= compressedSize);
739 src = (const BYTE*)src + compressedSize;
740 srcSize -= compressedSize;
741 bound += decompressedBound;
742 }
743 return bound;
744 }
745
746
747 /*-*************************************************************
748 * Frame decoding
749 ***************************************************************/
750
751 /** ZSTD_insertBlock() :
752 * insert `src` block into `dctx` history. Useful to track uncompressed blocks. */
ZSTD_insertBlock(ZSTD_DCtx * dctx,const void * blockStart,size_t blockSize)753 size_t ZSTD_insertBlock(ZSTD_DCtx* dctx, const void* blockStart, size_t blockSize)
754 {
755 DEBUGLOG(5, "ZSTD_insertBlock: %u bytes", (unsigned)blockSize);
756 ZSTD_checkContinuity(dctx, blockStart, blockSize);
757 dctx->previousDstEnd = (const char*)blockStart + blockSize;
758 return blockSize;
759 }
760
761
ZSTD_copyRawBlock(void * dst,size_t dstCapacity,const void * src,size_t srcSize)762 static size_t ZSTD_copyRawBlock(void* dst, size_t dstCapacity,
763 const void* src, size_t srcSize)
764 {
765 DEBUGLOG(5, "ZSTD_copyRawBlock");
766 RETURN_ERROR_IF(srcSize > dstCapacity, dstSize_tooSmall, "");
767 if (dst == NULL) {
768 if (srcSize == 0) return 0;
769 RETURN_ERROR(dstBuffer_null, "");
770 }
771 ZSTD_memcpy(dst, src, srcSize);
772 return srcSize;
773 }
774
ZSTD_setRleBlock(void * dst,size_t dstCapacity,BYTE b,size_t regenSize)775 static size_t ZSTD_setRleBlock(void* dst, size_t dstCapacity,
776 BYTE b,
777 size_t regenSize)
778 {
779 RETURN_ERROR_IF(regenSize > dstCapacity, dstSize_tooSmall, "");
780 if (dst == NULL) {
781 if (regenSize == 0) return 0;
782 RETURN_ERROR(dstBuffer_null, "");
783 }
784 ZSTD_memset(dst, b, regenSize);
785 return regenSize;
786 }
787
ZSTD_DCtx_trace_end(ZSTD_DCtx const * dctx,U64 uncompressedSize,U64 compressedSize,unsigned streaming)788 static void ZSTD_DCtx_trace_end(ZSTD_DCtx const* dctx, U64 uncompressedSize, U64 compressedSize, unsigned streaming)
789 {
790 #if ZSTD_TRACE
791 if (dctx->traceCtx && ZSTD_trace_decompress_end != NULL) {
792 ZSTD_Trace trace;
793 ZSTD_memset(&trace, 0, sizeof(trace));
794 trace.version = ZSTD_VERSION_NUMBER;
795 trace.streaming = streaming;
796 if (dctx->ddict) {
797 trace.dictionaryID = ZSTD_getDictID_fromDDict(dctx->ddict);
798 trace.dictionarySize = ZSTD_DDict_dictSize(dctx->ddict);
799 trace.dictionaryIsCold = dctx->ddictIsCold;
800 }
801 trace.uncompressedSize = (size_t)uncompressedSize;
802 trace.compressedSize = (size_t)compressedSize;
803 trace.dctx = dctx;
804 ZSTD_trace_decompress_end(dctx->traceCtx, &trace);
805 }
806 #else
807 (void)dctx;
808 (void)uncompressedSize;
809 (void)compressedSize;
810 (void)streaming;
811 #endif
812 }
813
814
815 /*! ZSTD_decompressFrame() :
816 * @dctx must be properly initialized
817 * will update *srcPtr and *srcSizePtr,
818 * to make *srcPtr progress by one frame. */
ZSTD_decompressFrame(ZSTD_DCtx * dctx,void * dst,size_t dstCapacity,const void ** srcPtr,size_t * srcSizePtr)819 static size_t ZSTD_decompressFrame(ZSTD_DCtx* dctx,
820 void* dst, size_t dstCapacity,
821 const void** srcPtr, size_t *srcSizePtr)
822 {
823 const BYTE* const istart = (const BYTE*)(*srcPtr);
824 const BYTE* ip = istart;
825 BYTE* const ostart = (BYTE*)dst;
826 BYTE* const oend = dstCapacity != 0 ? ostart + dstCapacity : ostart;
827 BYTE* op = ostart;
828 size_t remainingSrcSize = *srcSizePtr;
829
830 DEBUGLOG(4, "ZSTD_decompressFrame (srcSize:%i)", (int)*srcSizePtr);
831
832 /* check */
833 RETURN_ERROR_IF(
834 remainingSrcSize < ZSTD_FRAMEHEADERSIZE_MIN(dctx->format)+ZSTD_blockHeaderSize,
835 srcSize_wrong, "");
836
837 /* Frame Header */
838 { size_t const frameHeaderSize = ZSTD_frameHeaderSize_internal(
839 ip, ZSTD_FRAMEHEADERSIZE_PREFIX(dctx->format), dctx->format);
840 if (ZSTD_isError(frameHeaderSize)) return frameHeaderSize;
841 RETURN_ERROR_IF(remainingSrcSize < frameHeaderSize+ZSTD_blockHeaderSize,
842 srcSize_wrong, "");
843 FORWARD_IF_ERROR( ZSTD_decodeFrameHeader(dctx, ip, frameHeaderSize) , "");
844 ip += frameHeaderSize; remainingSrcSize -= frameHeaderSize;
845 }
846
847 /* Loop on each block */
848 while (1) {
849 size_t decodedSize;
850 blockProperties_t blockProperties;
851 size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSrcSize, &blockProperties);
852 if (ZSTD_isError(cBlockSize)) return cBlockSize;
853
854 ip += ZSTD_blockHeaderSize;
855 remainingSrcSize -= ZSTD_blockHeaderSize;
856 RETURN_ERROR_IF(cBlockSize > remainingSrcSize, srcSize_wrong, "");
857
858 switch(blockProperties.blockType)
859 {
860 case bt_compressed:
861 decodedSize = ZSTD_decompressBlock_internal(dctx, op, (size_t)(oend-op), ip, cBlockSize, /* frame */ 1);
862 break;
863 case bt_raw :
864 decodedSize = ZSTD_copyRawBlock(op, (size_t)(oend-op), ip, cBlockSize);
865 break;
866 case bt_rle :
867 decodedSize = ZSTD_setRleBlock(op, (size_t)(oend-op), *ip, blockProperties.origSize);
868 break;
869 case bt_reserved :
870 default:
871 RETURN_ERROR(corruption_detected, "invalid block type");
872 }
873
874 if (ZSTD_isError(decodedSize)) return decodedSize;
875 if (dctx->validateChecksum)
876 XXH64_update(&dctx->xxhState, op, decodedSize);
877 if (decodedSize != 0)
878 op += decodedSize;
879 assert(ip != NULL);
880 ip += cBlockSize;
881 remainingSrcSize -= cBlockSize;
882 if (blockProperties.lastBlock) break;
883 }
884
885 if (dctx->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN) {
886 RETURN_ERROR_IF((U64)(op-ostart) != dctx->fParams.frameContentSize,
887 corruption_detected, "");
888 }
889 if (dctx->fParams.checksumFlag) { /* Frame content checksum verification */
890 RETURN_ERROR_IF(remainingSrcSize<4, checksum_wrong, "");
891 if (!dctx->forceIgnoreChecksum) {
892 U32 const checkCalc = (U32)XXH64_digest(&dctx->xxhState);
893 U32 checkRead;
894 checkRead = MEM_readLE32(ip);
895 RETURN_ERROR_IF(checkRead != checkCalc, checksum_wrong, "");
896 }
897 ip += 4;
898 remainingSrcSize -= 4;
899 }
900 ZSTD_DCtx_trace_end(dctx, (U64)(op-ostart), (U64)(ip-istart), /* streaming */ 0);
901 /* Allow caller to get size read */
902 *srcPtr = ip;
903 *srcSizePtr = remainingSrcSize;
904 return (size_t)(op-ostart);
905 }
906
ZSTD_decompressMultiFrame(ZSTD_DCtx * dctx,void * dst,size_t dstCapacity,const void * src,size_t srcSize,const void * dict,size_t dictSize,const ZSTD_DDict * ddict)907 static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx,
908 void* dst, size_t dstCapacity,
909 const void* src, size_t srcSize,
910 const void* dict, size_t dictSize,
911 const ZSTD_DDict* ddict)
912 {
913 void* const dststart = dst;
914 int moreThan1Frame = 0;
915
916 DEBUGLOG(5, "ZSTD_decompressMultiFrame");
917 assert(dict==NULL || ddict==NULL); /* either dict or ddict set, not both */
918
919 if (ddict) {
920 dict = ZSTD_DDict_dictContent(ddict);
921 dictSize = ZSTD_DDict_dictSize(ddict);
922 }
923
924 while (srcSize >= ZSTD_startingInputLength(dctx->format)) {
925
926 #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT >= 1)
927 if (ZSTD_isLegacy(src, srcSize)) {
928 size_t decodedSize;
929 size_t const frameSize = ZSTD_findFrameCompressedSizeLegacy(src, srcSize);
930 if (ZSTD_isError(frameSize)) return frameSize;
931 RETURN_ERROR_IF(dctx->staticSize, memory_allocation,
932 "legacy support is not compatible with static dctx");
933
934 decodedSize = ZSTD_decompressLegacy(dst, dstCapacity, src, frameSize, dict, dictSize);
935 if (ZSTD_isError(decodedSize)) return decodedSize;
936
937 assert(decodedSize <= dstCapacity);
938 dst = (BYTE*)dst + decodedSize;
939 dstCapacity -= decodedSize;
940
941 src = (const BYTE*)src + frameSize;
942 srcSize -= frameSize;
943
944 continue;
945 }
946 #endif
947
948 { U32 const magicNumber = MEM_readLE32(src);
949 DEBUGLOG(4, "reading magic number %08X (expecting %08X)",
950 (unsigned)magicNumber, ZSTD_MAGICNUMBER);
951 if ((magicNumber & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
952 size_t const skippableSize = readSkippableFrameSize(src, srcSize);
953 FORWARD_IF_ERROR(skippableSize, "readSkippableFrameSize failed");
954 assert(skippableSize <= srcSize);
955
956 src = (const BYTE *)src + skippableSize;
957 srcSize -= skippableSize;
958 continue;
959 } }
960
961 if (ddict) {
962 /* we were called from ZSTD_decompress_usingDDict */
963 FORWARD_IF_ERROR(ZSTD_decompressBegin_usingDDict(dctx, ddict), "");
964 } else {
965 /* this will initialize correctly with no dict if dict == NULL, so
966 * use this in all cases but ddict */
967 FORWARD_IF_ERROR(ZSTD_decompressBegin_usingDict(dctx, dict, dictSize), "");
968 }
969 ZSTD_checkContinuity(dctx, dst, dstCapacity);
970
971 { const size_t res = ZSTD_decompressFrame(dctx, dst, dstCapacity,
972 &src, &srcSize);
973 RETURN_ERROR_IF(
974 (ZSTD_getErrorCode(res) == ZSTD_error_prefix_unknown)
975 && (moreThan1Frame==1),
976 srcSize_wrong,
977 "At least one frame successfully completed, "
978 "but following bytes are garbage: "
979 "it's more likely to be a srcSize error, "
980 "specifying more input bytes than size of frame(s). "
981 "Note: one could be unlucky, it might be a corruption error instead, "
982 "happening right at the place where we expect zstd magic bytes. "
983 "But this is _much_ less likely than a srcSize field error.");
984 if (ZSTD_isError(res)) return res;
985 assert(res <= dstCapacity);
986 if (res != 0)
987 dst = (BYTE*)dst + res;
988 dstCapacity -= res;
989 }
990 moreThan1Frame = 1;
991 } /* while (srcSize >= ZSTD_frameHeaderSize_prefix) */
992
993 RETURN_ERROR_IF(srcSize, srcSize_wrong, "input not entirely consumed");
994
995 return (size_t)((BYTE*)dst - (BYTE*)dststart);
996 }
997
ZSTD_decompress_usingDict(ZSTD_DCtx * dctx,void * dst,size_t dstCapacity,const void * src,size_t srcSize,const void * dict,size_t dictSize)998 size_t ZSTD_decompress_usingDict(ZSTD_DCtx* dctx,
999 void* dst, size_t dstCapacity,
1000 const void* src, size_t srcSize,
1001 const void* dict, size_t dictSize)
1002 {
1003 return ZSTD_decompressMultiFrame(dctx, dst, dstCapacity, src, srcSize, dict, dictSize, NULL);
1004 }
1005
1006
ZSTD_getDDict(ZSTD_DCtx * dctx)1007 static ZSTD_DDict const* ZSTD_getDDict(ZSTD_DCtx* dctx)
1008 {
1009 switch (dctx->dictUses) {
1010 default:
1011 assert(0 /* Impossible */);
1012 /* fall-through */
1013 case ZSTD_dont_use:
1014 ZSTD_clearDict(dctx);
1015 return NULL;
1016 case ZSTD_use_indefinitely:
1017 return dctx->ddict;
1018 case ZSTD_use_once:
1019 dctx->dictUses = ZSTD_dont_use;
1020 return dctx->ddict;
1021 }
1022 }
1023
ZSTD_decompressDCtx(ZSTD_DCtx * dctx,void * dst,size_t dstCapacity,const void * src,size_t srcSize)1024 size_t ZSTD_decompressDCtx(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize)
1025 {
1026 return ZSTD_decompress_usingDDict(dctx, dst, dstCapacity, src, srcSize, ZSTD_getDDict(dctx));
1027 }
1028
1029
ZSTD_decompress(void * dst,size_t dstCapacity,const void * src,size_t srcSize)1030 size_t ZSTD_decompress(void* dst, size_t dstCapacity, const void* src, size_t srcSize)
1031 {
1032 #if defined(ZSTD_HEAPMODE) && (ZSTD_HEAPMODE>=1)
1033 size_t regenSize;
1034 ZSTD_DCtx* const dctx = ZSTD_createDCtx();
1035 RETURN_ERROR_IF(dctx==NULL, memory_allocation, "NULL pointer!");
1036 regenSize = ZSTD_decompressDCtx(dctx, dst, dstCapacity, src, srcSize);
1037 ZSTD_freeDCtx(dctx);
1038 return regenSize;
1039 #else /* stack mode */
1040 ZSTD_DCtx dctx;
1041 ZSTD_initDCtx_internal(&dctx);
1042 return ZSTD_decompressDCtx(&dctx, dst, dstCapacity, src, srcSize);
1043 #endif
1044 }
1045
1046
1047 /*-**************************************
1048 * Advanced Streaming Decompression API
1049 * Bufferless and synchronous
1050 ****************************************/
ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx * dctx)1051 size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx* dctx) { return dctx->expected; }
1052
1053 /**
1054 * Similar to ZSTD_nextSrcSizeToDecompress(), but when when a block input can be streamed,
1055 * we allow taking a partial block as the input. Currently only raw uncompressed blocks can
1056 * be streamed.
1057 *
1058 * For blocks that can be streamed, this allows us to reduce the latency until we produce
1059 * output, and avoid copying the input.
1060 *
1061 * @param inputSize - The total amount of input that the caller currently has.
1062 */
ZSTD_nextSrcSizeToDecompressWithInputSize(ZSTD_DCtx * dctx,size_t inputSize)1063 static size_t ZSTD_nextSrcSizeToDecompressWithInputSize(ZSTD_DCtx* dctx, size_t inputSize) {
1064 if (!(dctx->stage == ZSTDds_decompressBlock || dctx->stage == ZSTDds_decompressLastBlock))
1065 return dctx->expected;
1066 if (dctx->bType != bt_raw)
1067 return dctx->expected;
1068 return MIN(MAX(inputSize, 1), dctx->expected);
1069 }
1070
ZSTD_nextInputType(ZSTD_DCtx * dctx)1071 ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx* dctx) {
1072 switch(dctx->stage)
1073 {
1074 default: /* should not happen */
1075 assert(0);
1076 case ZSTDds_getFrameHeaderSize:
1077 case ZSTDds_decodeFrameHeader:
1078 return ZSTDnit_frameHeader;
1079 case ZSTDds_decodeBlockHeader:
1080 return ZSTDnit_blockHeader;
1081 case ZSTDds_decompressBlock:
1082 return ZSTDnit_block;
1083 case ZSTDds_decompressLastBlock:
1084 return ZSTDnit_lastBlock;
1085 case ZSTDds_checkChecksum:
1086 return ZSTDnit_checksum;
1087 case ZSTDds_decodeSkippableHeader:
1088 case ZSTDds_skipFrame:
1089 return ZSTDnit_skippableFrame;
1090 }
1091 }
1092
ZSTD_isSkipFrame(ZSTD_DCtx * dctx)1093 static int ZSTD_isSkipFrame(ZSTD_DCtx* dctx) { return dctx->stage == ZSTDds_skipFrame; }
1094
1095 /** ZSTD_decompressContinue() :
1096 * srcSize : must be the exact nb of bytes expected (see ZSTD_nextSrcSizeToDecompress())
1097 * @return : nb of bytes generated into `dst` (necessarily <= `dstCapacity)
1098 * or an error code, which can be tested using ZSTD_isError() */
ZSTD_decompressContinue(ZSTD_DCtx * dctx,void * dst,size_t dstCapacity,const void * src,size_t srcSize)1099 size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize)
1100 {
1101 DEBUGLOG(5, "ZSTD_decompressContinue (srcSize:%u)", (unsigned)srcSize);
1102 /* Sanity check */
1103 RETURN_ERROR_IF(srcSize != ZSTD_nextSrcSizeToDecompressWithInputSize(dctx, srcSize), srcSize_wrong, "not allowed");
1104 ZSTD_checkContinuity(dctx, dst, dstCapacity);
1105
1106 dctx->processedCSize += srcSize;
1107
1108 switch (dctx->stage)
1109 {
1110 case ZSTDds_getFrameHeaderSize :
1111 assert(src != NULL);
1112 if (dctx->format == ZSTD_f_zstd1) { /* allows header */
1113 assert(srcSize >= ZSTD_FRAMEIDSIZE); /* to read skippable magic number */
1114 if ((MEM_readLE32(src) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) { /* skippable frame */
1115 ZSTD_memcpy(dctx->headerBuffer, src, srcSize);
1116 dctx->expected = ZSTD_SKIPPABLEHEADERSIZE - srcSize; /* remaining to load to get full skippable frame header */
1117 dctx->stage = ZSTDds_decodeSkippableHeader;
1118 return 0;
1119 } }
1120 dctx->headerSize = ZSTD_frameHeaderSize_internal(src, srcSize, dctx->format);
1121 if (ZSTD_isError(dctx->headerSize)) return dctx->headerSize;
1122 ZSTD_memcpy(dctx->headerBuffer, src, srcSize);
1123 dctx->expected = dctx->headerSize - srcSize;
1124 dctx->stage = ZSTDds_decodeFrameHeader;
1125 return 0;
1126
1127 case ZSTDds_decodeFrameHeader:
1128 assert(src != NULL);
1129 ZSTD_memcpy(dctx->headerBuffer + (dctx->headerSize - srcSize), src, srcSize);
1130 FORWARD_IF_ERROR(ZSTD_decodeFrameHeader(dctx, dctx->headerBuffer, dctx->headerSize), "");
1131 dctx->expected = ZSTD_blockHeaderSize;
1132 dctx->stage = ZSTDds_decodeBlockHeader;
1133 return 0;
1134
1135 case ZSTDds_decodeBlockHeader:
1136 { blockProperties_t bp;
1137 size_t const cBlockSize = ZSTD_getcBlockSize(src, ZSTD_blockHeaderSize, &bp);
1138 if (ZSTD_isError(cBlockSize)) return cBlockSize;
1139 RETURN_ERROR_IF(cBlockSize > dctx->fParams.blockSizeMax, corruption_detected, "Block Size Exceeds Maximum");
1140 dctx->expected = cBlockSize;
1141 dctx->bType = bp.blockType;
1142 dctx->rleSize = bp.origSize;
1143 if (cBlockSize) {
1144 dctx->stage = bp.lastBlock ? ZSTDds_decompressLastBlock : ZSTDds_decompressBlock;
1145 return 0;
1146 }
1147 /* empty block */
1148 if (bp.lastBlock) {
1149 if (dctx->fParams.checksumFlag) {
1150 dctx->expected = 4;
1151 dctx->stage = ZSTDds_checkChecksum;
1152 } else {
1153 dctx->expected = 0; /* end of frame */
1154 dctx->stage = ZSTDds_getFrameHeaderSize;
1155 }
1156 } else {
1157 dctx->expected = ZSTD_blockHeaderSize; /* jump to next header */
1158 dctx->stage = ZSTDds_decodeBlockHeader;
1159 }
1160 return 0;
1161 }
1162
1163 case ZSTDds_decompressLastBlock:
1164 case ZSTDds_decompressBlock:
1165 DEBUGLOG(5, "ZSTD_decompressContinue: case ZSTDds_decompressBlock");
1166 { size_t rSize;
1167 switch(dctx->bType)
1168 {
1169 case bt_compressed:
1170 DEBUGLOG(5, "ZSTD_decompressContinue: case bt_compressed");
1171 rSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize, /* frame */ 1);
1172 dctx->expected = 0; /* Streaming not supported */
1173 break;
1174 case bt_raw :
1175 assert(srcSize <= dctx->expected);
1176 rSize = ZSTD_copyRawBlock(dst, dstCapacity, src, srcSize);
1177 FORWARD_IF_ERROR(rSize, "ZSTD_copyRawBlock failed");
1178 assert(rSize == srcSize);
1179 dctx->expected -= rSize;
1180 break;
1181 case bt_rle :
1182 rSize = ZSTD_setRleBlock(dst, dstCapacity, *(const BYTE*)src, dctx->rleSize);
1183 dctx->expected = 0; /* Streaming not supported */
1184 break;
1185 case bt_reserved : /* should never happen */
1186 default:
1187 RETURN_ERROR(corruption_detected, "invalid block type");
1188 }
1189 FORWARD_IF_ERROR(rSize, "");
1190 RETURN_ERROR_IF(rSize > dctx->fParams.blockSizeMax, corruption_detected, "Decompressed Block Size Exceeds Maximum");
1191 DEBUGLOG(5, "ZSTD_decompressContinue: decoded size from block : %u", (unsigned)rSize);
1192 dctx->decodedSize += rSize;
1193 if (dctx->validateChecksum) XXH64_update(&dctx->xxhState, dst, rSize);
1194 dctx->previousDstEnd = (char*)dst + rSize;
1195
1196 /* Stay on the same stage until we are finished streaming the block. */
1197 if (dctx->expected > 0) {
1198 return rSize;
1199 }
1200
1201 if (dctx->stage == ZSTDds_decompressLastBlock) { /* end of frame */
1202 DEBUGLOG(4, "ZSTD_decompressContinue: decoded size from frame : %u", (unsigned)dctx->decodedSize);
1203 RETURN_ERROR_IF(
1204 dctx->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN
1205 && dctx->decodedSize != dctx->fParams.frameContentSize,
1206 corruption_detected, "");
1207 if (dctx->fParams.checksumFlag) { /* another round for frame checksum */
1208 dctx->expected = 4;
1209 dctx->stage = ZSTDds_checkChecksum;
1210 } else {
1211 ZSTD_DCtx_trace_end(dctx, dctx->decodedSize, dctx->processedCSize, /* streaming */ 1);
1212 dctx->expected = 0; /* ends here */
1213 dctx->stage = ZSTDds_getFrameHeaderSize;
1214 }
1215 } else {
1216 dctx->stage = ZSTDds_decodeBlockHeader;
1217 dctx->expected = ZSTD_blockHeaderSize;
1218 }
1219 return rSize;
1220 }
1221
1222 case ZSTDds_checkChecksum:
1223 assert(srcSize == 4); /* guaranteed by dctx->expected */
1224 {
1225 if (dctx->validateChecksum) {
1226 U32 const h32 = (U32)XXH64_digest(&dctx->xxhState);
1227 U32 const check32 = MEM_readLE32(src);
1228 DEBUGLOG(4, "ZSTD_decompressContinue: checksum : calculated %08X :: %08X read", (unsigned)h32, (unsigned)check32);
1229 RETURN_ERROR_IF(check32 != h32, checksum_wrong, "");
1230 }
1231 ZSTD_DCtx_trace_end(dctx, dctx->decodedSize, dctx->processedCSize, /* streaming */ 1);
1232 dctx->expected = 0;
1233 dctx->stage = ZSTDds_getFrameHeaderSize;
1234 return 0;
1235 }
1236
1237 case ZSTDds_decodeSkippableHeader:
1238 assert(src != NULL);
1239 assert(srcSize <= ZSTD_SKIPPABLEHEADERSIZE);
1240 ZSTD_memcpy(dctx->headerBuffer + (ZSTD_SKIPPABLEHEADERSIZE - srcSize), src, srcSize); /* complete skippable header */
1241 dctx->expected = MEM_readLE32(dctx->headerBuffer + ZSTD_FRAMEIDSIZE); /* note : dctx->expected can grow seriously large, beyond local buffer size */
1242 dctx->stage = ZSTDds_skipFrame;
1243 return 0;
1244
1245 case ZSTDds_skipFrame:
1246 dctx->expected = 0;
1247 dctx->stage = ZSTDds_getFrameHeaderSize;
1248 return 0;
1249
1250 default:
1251 assert(0); /* impossible */
1252 RETURN_ERROR(GENERIC, "impossible to reach"); /* some compiler require default to do something */
1253 }
1254 }
1255
1256
ZSTD_refDictContent(ZSTD_DCtx * dctx,const void * dict,size_t dictSize)1257 static size_t ZSTD_refDictContent(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
1258 {
1259 dctx->dictEnd = dctx->previousDstEnd;
1260 dctx->virtualStart = (const char*)dict - ((const char*)(dctx->previousDstEnd) - (const char*)(dctx->prefixStart));
1261 dctx->prefixStart = dict;
1262 dctx->previousDstEnd = (const char*)dict + dictSize;
1263 #ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
1264 dctx->dictContentBeginForFuzzing = dctx->prefixStart;
1265 dctx->dictContentEndForFuzzing = dctx->previousDstEnd;
1266 #endif
1267 return 0;
1268 }
1269
1270 /*! ZSTD_loadDEntropy() :
1271 * dict : must point at beginning of a valid zstd dictionary.
1272 * @return : size of entropy tables read */
1273 size_t
ZSTD_loadDEntropy(ZSTD_entropyDTables_t * entropy,const void * const dict,size_t const dictSize)1274 ZSTD_loadDEntropy(ZSTD_entropyDTables_t* entropy,
1275 const void* const dict, size_t const dictSize)
1276 {
1277 const BYTE* dictPtr = (const BYTE*)dict;
1278 const BYTE* const dictEnd = dictPtr + dictSize;
1279
1280 RETURN_ERROR_IF(dictSize <= 8, dictionary_corrupted, "dict is too small");
1281 assert(MEM_readLE32(dict) == ZSTD_MAGIC_DICTIONARY); /* dict must be valid */
1282 dictPtr += 8; /* skip header = magic + dictID */
1283
1284 ZSTD_STATIC_ASSERT(offsetof(ZSTD_entropyDTables_t, OFTable) == offsetof(ZSTD_entropyDTables_t, LLTable) + sizeof(entropy->LLTable));
1285 ZSTD_STATIC_ASSERT(offsetof(ZSTD_entropyDTables_t, MLTable) == offsetof(ZSTD_entropyDTables_t, OFTable) + sizeof(entropy->OFTable));
1286 ZSTD_STATIC_ASSERT(sizeof(entropy->LLTable) + sizeof(entropy->OFTable) + sizeof(entropy->MLTable) >= HUF_DECOMPRESS_WORKSPACE_SIZE);
1287 { void* const workspace = &entropy->LLTable; /* use fse tables as temporary workspace; implies fse tables are grouped together */
1288 size_t const workspaceSize = sizeof(entropy->LLTable) + sizeof(entropy->OFTable) + sizeof(entropy->MLTable);
1289 #ifdef HUF_FORCE_DECOMPRESS_X1
1290 /* in minimal huffman, we always use X1 variants */
1291 size_t const hSize = HUF_readDTableX1_wksp(entropy->hufTable,
1292 dictPtr, dictEnd - dictPtr,
1293 workspace, workspaceSize);
1294 #else
1295 size_t const hSize = HUF_readDTableX2_wksp(entropy->hufTable,
1296 dictPtr, (size_t)(dictEnd - dictPtr),
1297 workspace, workspaceSize);
1298 #endif
1299 RETURN_ERROR_IF(HUF_isError(hSize), dictionary_corrupted, "");
1300 dictPtr += hSize;
1301 }
1302
1303 { short offcodeNCount[MaxOff+1];
1304 unsigned offcodeMaxValue = MaxOff, offcodeLog;
1305 size_t const offcodeHeaderSize = FSE_readNCount(offcodeNCount, &offcodeMaxValue, &offcodeLog, dictPtr, (size_t)(dictEnd-dictPtr));
1306 RETURN_ERROR_IF(FSE_isError(offcodeHeaderSize), dictionary_corrupted, "");
1307 RETURN_ERROR_IF(offcodeMaxValue > MaxOff, dictionary_corrupted, "");
1308 RETURN_ERROR_IF(offcodeLog > OffFSELog, dictionary_corrupted, "");
1309 ZSTD_buildFSETable( entropy->OFTable,
1310 offcodeNCount, offcodeMaxValue,
1311 OF_base, OF_bits,
1312 offcodeLog,
1313 entropy->workspace, sizeof(entropy->workspace),
1314 /* bmi2 */0);
1315 dictPtr += offcodeHeaderSize;
1316 }
1317
1318 { short matchlengthNCount[MaxML+1];
1319 unsigned matchlengthMaxValue = MaxML, matchlengthLog;
1320 size_t const matchlengthHeaderSize = FSE_readNCount(matchlengthNCount, &matchlengthMaxValue, &matchlengthLog, dictPtr, (size_t)(dictEnd-dictPtr));
1321 RETURN_ERROR_IF(FSE_isError(matchlengthHeaderSize), dictionary_corrupted, "");
1322 RETURN_ERROR_IF(matchlengthMaxValue > MaxML, dictionary_corrupted, "");
1323 RETURN_ERROR_IF(matchlengthLog > MLFSELog, dictionary_corrupted, "");
1324 ZSTD_buildFSETable( entropy->MLTable,
1325 matchlengthNCount, matchlengthMaxValue,
1326 ML_base, ML_bits,
1327 matchlengthLog,
1328 entropy->workspace, sizeof(entropy->workspace),
1329 /* bmi2 */ 0);
1330 dictPtr += matchlengthHeaderSize;
1331 }
1332
1333 { short litlengthNCount[MaxLL+1];
1334 unsigned litlengthMaxValue = MaxLL, litlengthLog;
1335 size_t const litlengthHeaderSize = FSE_readNCount(litlengthNCount, &litlengthMaxValue, &litlengthLog, dictPtr, (size_t)(dictEnd-dictPtr));
1336 RETURN_ERROR_IF(FSE_isError(litlengthHeaderSize), dictionary_corrupted, "");
1337 RETURN_ERROR_IF(litlengthMaxValue > MaxLL, dictionary_corrupted, "");
1338 RETURN_ERROR_IF(litlengthLog > LLFSELog, dictionary_corrupted, "");
1339 ZSTD_buildFSETable( entropy->LLTable,
1340 litlengthNCount, litlengthMaxValue,
1341 LL_base, LL_bits,
1342 litlengthLog,
1343 entropy->workspace, sizeof(entropy->workspace),
1344 /* bmi2 */ 0);
1345 dictPtr += litlengthHeaderSize;
1346 }
1347
1348 RETURN_ERROR_IF(dictPtr+12 > dictEnd, dictionary_corrupted, "");
1349 { int i;
1350 size_t const dictContentSize = (size_t)(dictEnd - (dictPtr+12));
1351 for (i=0; i<3; i++) {
1352 U32 const rep = MEM_readLE32(dictPtr); dictPtr += 4;
1353 RETURN_ERROR_IF(rep==0 || rep > dictContentSize,
1354 dictionary_corrupted, "");
1355 entropy->rep[i] = rep;
1356 } }
1357
1358 return (size_t)(dictPtr - (const BYTE*)dict);
1359 }
1360
ZSTD_decompress_insertDictionary(ZSTD_DCtx * dctx,const void * dict,size_t dictSize)1361 static size_t ZSTD_decompress_insertDictionary(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
1362 {
1363 if (dictSize < 8) return ZSTD_refDictContent(dctx, dict, dictSize);
1364 { U32 const magic = MEM_readLE32(dict);
1365 if (magic != ZSTD_MAGIC_DICTIONARY) {
1366 return ZSTD_refDictContent(dctx, dict, dictSize); /* pure content mode */
1367 } }
1368 dctx->dictID = MEM_readLE32((const char*)dict + ZSTD_FRAMEIDSIZE);
1369
1370 /* load entropy tables */
1371 { size_t const eSize = ZSTD_loadDEntropy(&dctx->entropy, dict, dictSize);
1372 RETURN_ERROR_IF(ZSTD_isError(eSize), dictionary_corrupted, "");
1373 dict = (const char*)dict + eSize;
1374 dictSize -= eSize;
1375 }
1376 dctx->litEntropy = dctx->fseEntropy = 1;
1377
1378 /* reference dictionary content */
1379 return ZSTD_refDictContent(dctx, dict, dictSize);
1380 }
1381
ZSTD_decompressBegin(ZSTD_DCtx * dctx)1382 size_t ZSTD_decompressBegin(ZSTD_DCtx* dctx)
1383 {
1384 assert(dctx != NULL);
1385 #if ZSTD_TRACE
1386 dctx->traceCtx = (ZSTD_trace_decompress_begin != NULL) ? ZSTD_trace_decompress_begin(dctx) : 0;
1387 #endif
1388 dctx->expected = ZSTD_startingInputLength(dctx->format); /* dctx->format must be properly set */
1389 dctx->stage = ZSTDds_getFrameHeaderSize;
1390 dctx->processedCSize = 0;
1391 dctx->decodedSize = 0;
1392 dctx->previousDstEnd = NULL;
1393 dctx->prefixStart = NULL;
1394 dctx->virtualStart = NULL;
1395 dctx->dictEnd = NULL;
1396 dctx->entropy.hufTable[0] = (HUF_DTable)((HufLog)*0x1000001); /* cover both little and big endian */
1397 dctx->litEntropy = dctx->fseEntropy = 0;
1398 dctx->dictID = 0;
1399 dctx->bType = bt_reserved;
1400 ZSTD_STATIC_ASSERT(sizeof(dctx->entropy.rep) == sizeof(repStartValue));
1401 ZSTD_memcpy(dctx->entropy.rep, repStartValue, sizeof(repStartValue)); /* initial repcodes */
1402 dctx->LLTptr = dctx->entropy.LLTable;
1403 dctx->MLTptr = dctx->entropy.MLTable;
1404 dctx->OFTptr = dctx->entropy.OFTable;
1405 dctx->HUFptr = dctx->entropy.hufTable;
1406 return 0;
1407 }
1408
ZSTD_decompressBegin_usingDict(ZSTD_DCtx * dctx,const void * dict,size_t dictSize)1409 size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
1410 {
1411 FORWARD_IF_ERROR( ZSTD_decompressBegin(dctx) , "");
1412 if (dict && dictSize)
1413 RETURN_ERROR_IF(
1414 ZSTD_isError(ZSTD_decompress_insertDictionary(dctx, dict, dictSize)),
1415 dictionary_corrupted, "");
1416 return 0;
1417 }
1418
1419
1420 /* ====== ZSTD_DDict ====== */
1421
ZSTD_decompressBegin_usingDDict(ZSTD_DCtx * dctx,const ZSTD_DDict * ddict)1422 size_t ZSTD_decompressBegin_usingDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict)
1423 {
1424 DEBUGLOG(4, "ZSTD_decompressBegin_usingDDict");
1425 assert(dctx != NULL);
1426 if (ddict) {
1427 const char* const dictStart = (const char*)ZSTD_DDict_dictContent(ddict);
1428 size_t const dictSize = ZSTD_DDict_dictSize(ddict);
1429 const void* const dictEnd = dictStart + dictSize;
1430 dctx->ddictIsCold = (dctx->dictEnd != dictEnd);
1431 DEBUGLOG(4, "DDict is %s",
1432 dctx->ddictIsCold ? "~cold~" : "hot!");
1433 }
1434 FORWARD_IF_ERROR( ZSTD_decompressBegin(dctx) , "");
1435 if (ddict) { /* NULL ddict is equivalent to no dictionary */
1436 ZSTD_copyDDictParameters(dctx, ddict);
1437 }
1438 return 0;
1439 }
1440
1441 /*! ZSTD_getDictID_fromDict() :
1442 * Provides the dictID stored within dictionary.
1443 * if @return == 0, the dictionary is not conformant with Zstandard specification.
1444 * It can still be loaded, but as a content-only dictionary. */
ZSTD_getDictID_fromDict(const void * dict,size_t dictSize)1445 unsigned ZSTD_getDictID_fromDict(const void* dict, size_t dictSize)
1446 {
1447 if (dictSize < 8) return 0;
1448 if (MEM_readLE32(dict) != ZSTD_MAGIC_DICTIONARY) return 0;
1449 return MEM_readLE32((const char*)dict + ZSTD_FRAMEIDSIZE);
1450 }
1451
1452 /*! ZSTD_getDictID_fromFrame() :
1453 * Provides the dictID required to decompress frame stored within `src`.
1454 * If @return == 0, the dictID could not be decoded.
1455 * This could for one of the following reasons :
1456 * - The frame does not require a dictionary (most common case).
1457 * - The frame was built with dictID intentionally removed.
1458 * Needed dictionary is a hidden information.
1459 * Note : this use case also happens when using a non-conformant dictionary.
1460 * - `srcSize` is too small, and as a result, frame header could not be decoded.
1461 * Note : possible if `srcSize < ZSTD_FRAMEHEADERSIZE_MAX`.
1462 * - This is not a Zstandard frame.
1463 * When identifying the exact failure cause, it's possible to use
1464 * ZSTD_getFrameHeader(), which will provide a more precise error code. */
ZSTD_getDictID_fromFrame(const void * src,size_t srcSize)1465 unsigned ZSTD_getDictID_fromFrame(const void* src, size_t srcSize)
1466 {
1467 ZSTD_frameHeader zfp = { 0, 0, 0, ZSTD_frame, 0, 0, 0 };
1468 size_t const hError = ZSTD_getFrameHeader(&zfp, src, srcSize);
1469 if (ZSTD_isError(hError)) return 0;
1470 return zfp.dictID;
1471 }
1472
1473
1474 /*! ZSTD_decompress_usingDDict() :
1475 * Decompression using a pre-digested Dictionary
1476 * Use dictionary without significant overhead. */
ZSTD_decompress_usingDDict(ZSTD_DCtx * dctx,void * dst,size_t dstCapacity,const void * src,size_t srcSize,const ZSTD_DDict * ddict)1477 size_t ZSTD_decompress_usingDDict(ZSTD_DCtx* dctx,
1478 void* dst, size_t dstCapacity,
1479 const void* src, size_t srcSize,
1480 const ZSTD_DDict* ddict)
1481 {
1482 /* pass content and size in case legacy frames are encountered */
1483 return ZSTD_decompressMultiFrame(dctx, dst, dstCapacity, src, srcSize,
1484 NULL, 0,
1485 ddict);
1486 }
1487
1488
1489 /*=====================================
1490 * Streaming decompression
1491 *====================================*/
1492
ZSTD_createDStream(void)1493 ZSTD_DStream* ZSTD_createDStream(void)
1494 {
1495 DEBUGLOG(3, "ZSTD_createDStream");
1496 return ZSTD_createDStream_advanced(ZSTD_defaultCMem);
1497 }
1498
ZSTD_initStaticDStream(void * workspace,size_t workspaceSize)1499 ZSTD_DStream* ZSTD_initStaticDStream(void *workspace, size_t workspaceSize)
1500 {
1501 return ZSTD_initStaticDCtx(workspace, workspaceSize);
1502 }
1503
ZSTD_createDStream_advanced(ZSTD_customMem customMem)1504 ZSTD_DStream* ZSTD_createDStream_advanced(ZSTD_customMem customMem)
1505 {
1506 return ZSTD_createDCtx_advanced(customMem);
1507 }
1508
ZSTD_freeDStream(ZSTD_DStream * zds)1509 size_t ZSTD_freeDStream(ZSTD_DStream* zds)
1510 {
1511 return ZSTD_freeDCtx(zds);
1512 }
1513
1514
1515 /* *** Initialization *** */
1516
ZSTD_DStreamInSize(void)1517 size_t ZSTD_DStreamInSize(void) { return ZSTD_BLOCKSIZE_MAX + ZSTD_blockHeaderSize; }
ZSTD_DStreamOutSize(void)1518 size_t ZSTD_DStreamOutSize(void) { return ZSTD_BLOCKSIZE_MAX; }
1519
ZSTD_DCtx_loadDictionary_advanced(ZSTD_DCtx * dctx,const void * dict,size_t dictSize,ZSTD_dictLoadMethod_e dictLoadMethod,ZSTD_dictContentType_e dictContentType)1520 size_t ZSTD_DCtx_loadDictionary_advanced(ZSTD_DCtx* dctx,
1521 const void* dict, size_t dictSize,
1522 ZSTD_dictLoadMethod_e dictLoadMethod,
1523 ZSTD_dictContentType_e dictContentType)
1524 {
1525 RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
1526 ZSTD_clearDict(dctx);
1527 if (dict && dictSize != 0) {
1528 dctx->ddictLocal = ZSTD_createDDict_advanced(dict, dictSize, dictLoadMethod, dictContentType, dctx->customMem);
1529 RETURN_ERROR_IF(dctx->ddictLocal == NULL, memory_allocation, "NULL pointer!");
1530 dctx->ddict = dctx->ddictLocal;
1531 dctx->dictUses = ZSTD_use_indefinitely;
1532 }
1533 return 0;
1534 }
1535
ZSTD_DCtx_loadDictionary_byReference(ZSTD_DCtx * dctx,const void * dict,size_t dictSize)1536 size_t ZSTD_DCtx_loadDictionary_byReference(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
1537 {
1538 return ZSTD_DCtx_loadDictionary_advanced(dctx, dict, dictSize, ZSTD_dlm_byRef, ZSTD_dct_auto);
1539 }
1540
ZSTD_DCtx_loadDictionary(ZSTD_DCtx * dctx,const void * dict,size_t dictSize)1541 size_t ZSTD_DCtx_loadDictionary(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
1542 {
1543 return ZSTD_DCtx_loadDictionary_advanced(dctx, dict, dictSize, ZSTD_dlm_byCopy, ZSTD_dct_auto);
1544 }
1545
ZSTD_DCtx_refPrefix_advanced(ZSTD_DCtx * dctx,const void * prefix,size_t prefixSize,ZSTD_dictContentType_e dictContentType)1546 size_t ZSTD_DCtx_refPrefix_advanced(ZSTD_DCtx* dctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType)
1547 {
1548 FORWARD_IF_ERROR(ZSTD_DCtx_loadDictionary_advanced(dctx, prefix, prefixSize, ZSTD_dlm_byRef, dictContentType), "");
1549 dctx->dictUses = ZSTD_use_once;
1550 return 0;
1551 }
1552
ZSTD_DCtx_refPrefix(ZSTD_DCtx * dctx,const void * prefix,size_t prefixSize)1553 size_t ZSTD_DCtx_refPrefix(ZSTD_DCtx* dctx, const void* prefix, size_t prefixSize)
1554 {
1555 return ZSTD_DCtx_refPrefix_advanced(dctx, prefix, prefixSize, ZSTD_dct_rawContent);
1556 }
1557
1558
1559 /* ZSTD_initDStream_usingDict() :
1560 * return : expected size, aka ZSTD_startingInputLength().
1561 * this function cannot fail */
ZSTD_initDStream_usingDict(ZSTD_DStream * zds,const void * dict,size_t dictSize)1562 size_t ZSTD_initDStream_usingDict(ZSTD_DStream* zds, const void* dict, size_t dictSize)
1563 {
1564 DEBUGLOG(4, "ZSTD_initDStream_usingDict");
1565 FORWARD_IF_ERROR( ZSTD_DCtx_reset(zds, ZSTD_reset_session_only) , "");
1566 FORWARD_IF_ERROR( ZSTD_DCtx_loadDictionary(zds, dict, dictSize) , "");
1567 return ZSTD_startingInputLength(zds->format);
1568 }
1569
1570 /* note : this variant can't fail */
ZSTD_initDStream(ZSTD_DStream * zds)1571 size_t ZSTD_initDStream(ZSTD_DStream* zds)
1572 {
1573 DEBUGLOG(4, "ZSTD_initDStream");
1574 return ZSTD_initDStream_usingDDict(zds, NULL);
1575 }
1576
1577 /* ZSTD_initDStream_usingDDict() :
1578 * ddict will just be referenced, and must outlive decompression session
1579 * this function cannot fail */
ZSTD_initDStream_usingDDict(ZSTD_DStream * dctx,const ZSTD_DDict * ddict)1580 size_t ZSTD_initDStream_usingDDict(ZSTD_DStream* dctx, const ZSTD_DDict* ddict)
1581 {
1582 FORWARD_IF_ERROR( ZSTD_DCtx_reset(dctx, ZSTD_reset_session_only) , "");
1583 FORWARD_IF_ERROR( ZSTD_DCtx_refDDict(dctx, ddict) , "");
1584 return ZSTD_startingInputLength(dctx->format);
1585 }
1586
1587 /* ZSTD_resetDStream() :
1588 * return : expected size, aka ZSTD_startingInputLength().
1589 * this function cannot fail */
ZSTD_resetDStream(ZSTD_DStream * dctx)1590 size_t ZSTD_resetDStream(ZSTD_DStream* dctx)
1591 {
1592 FORWARD_IF_ERROR(ZSTD_DCtx_reset(dctx, ZSTD_reset_session_only), "");
1593 return ZSTD_startingInputLength(dctx->format);
1594 }
1595
1596
ZSTD_DCtx_refDDict(ZSTD_DCtx * dctx,const ZSTD_DDict * ddict)1597 size_t ZSTD_DCtx_refDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict)
1598 {
1599 RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
1600 ZSTD_clearDict(dctx);
1601 if (ddict) {
1602 dctx->ddict = ddict;
1603 dctx->dictUses = ZSTD_use_indefinitely;
1604 if (dctx->refMultipleDDicts == ZSTD_rmd_refMultipleDDicts) {
1605 if (dctx->ddictSet == NULL) {
1606 dctx->ddictSet = ZSTD_createDDictHashSet(dctx->customMem);
1607 if (!dctx->ddictSet) {
1608 RETURN_ERROR(memory_allocation, "Failed to allocate memory for hash set!");
1609 }
1610 }
1611 assert(!dctx->staticSize); /* Impossible: ddictSet cannot have been allocated if static dctx */
1612 FORWARD_IF_ERROR(ZSTD_DDictHashSet_addDDict(dctx->ddictSet, ddict, dctx->customMem), "");
1613 }
1614 }
1615 return 0;
1616 }
1617
1618 /* ZSTD_DCtx_setMaxWindowSize() :
1619 * note : no direct equivalence in ZSTD_DCtx_setParameter,
1620 * since this version sets windowSize, and the other sets windowLog */
ZSTD_DCtx_setMaxWindowSize(ZSTD_DCtx * dctx,size_t maxWindowSize)1621 size_t ZSTD_DCtx_setMaxWindowSize(ZSTD_DCtx* dctx, size_t maxWindowSize)
1622 {
1623 ZSTD_bounds const bounds = ZSTD_dParam_getBounds(ZSTD_d_windowLogMax);
1624 size_t const min = (size_t)1 << bounds.lowerBound;
1625 size_t const max = (size_t)1 << bounds.upperBound;
1626 RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
1627 RETURN_ERROR_IF(maxWindowSize < min, parameter_outOfBound, "");
1628 RETURN_ERROR_IF(maxWindowSize > max, parameter_outOfBound, "");
1629 dctx->maxWindowSize = maxWindowSize;
1630 return 0;
1631 }
1632
ZSTD_DCtx_setFormat(ZSTD_DCtx * dctx,ZSTD_format_e format)1633 size_t ZSTD_DCtx_setFormat(ZSTD_DCtx* dctx, ZSTD_format_e format)
1634 {
1635 return ZSTD_DCtx_setParameter(dctx, ZSTD_d_format, (int)format);
1636 }
1637
ZSTD_dParam_getBounds(ZSTD_dParameter dParam)1638 ZSTD_bounds ZSTD_dParam_getBounds(ZSTD_dParameter dParam)
1639 {
1640 ZSTD_bounds bounds = { 0, 0, 0 };
1641 switch(dParam) {
1642 case ZSTD_d_windowLogMax:
1643 bounds.lowerBound = ZSTD_WINDOWLOG_ABSOLUTEMIN;
1644 bounds.upperBound = ZSTD_WINDOWLOG_MAX;
1645 return bounds;
1646 case ZSTD_d_format:
1647 bounds.lowerBound = (int)ZSTD_f_zstd1;
1648 bounds.upperBound = (int)ZSTD_f_zstd1_magicless;
1649 ZSTD_STATIC_ASSERT(ZSTD_f_zstd1 < ZSTD_f_zstd1_magicless);
1650 return bounds;
1651 case ZSTD_d_stableOutBuffer:
1652 bounds.lowerBound = (int)ZSTD_bm_buffered;
1653 bounds.upperBound = (int)ZSTD_bm_stable;
1654 return bounds;
1655 case ZSTD_d_forceIgnoreChecksum:
1656 bounds.lowerBound = (int)ZSTD_d_validateChecksum;
1657 bounds.upperBound = (int)ZSTD_d_ignoreChecksum;
1658 return bounds;
1659 case ZSTD_d_refMultipleDDicts:
1660 bounds.lowerBound = (int)ZSTD_rmd_refSingleDDict;
1661 bounds.upperBound = (int)ZSTD_rmd_refMultipleDDicts;
1662 return bounds;
1663 default:;
1664 }
1665 bounds.error = ERROR(parameter_unsupported);
1666 return bounds;
1667 }
1668
1669 /* ZSTD_dParam_withinBounds:
1670 * @return 1 if value is within dParam bounds,
1671 * 0 otherwise */
ZSTD_dParam_withinBounds(ZSTD_dParameter dParam,int value)1672 static int ZSTD_dParam_withinBounds(ZSTD_dParameter dParam, int value)
1673 {
1674 ZSTD_bounds const bounds = ZSTD_dParam_getBounds(dParam);
1675 if (ZSTD_isError(bounds.error)) return 0;
1676 if (value < bounds.lowerBound) return 0;
1677 if (value > bounds.upperBound) return 0;
1678 return 1;
1679 }
1680
1681 #define CHECK_DBOUNDS(p,v) { \
1682 RETURN_ERROR_IF(!ZSTD_dParam_withinBounds(p, v), parameter_outOfBound, ""); \
1683 }
1684
ZSTD_DCtx_getParameter(ZSTD_DCtx * dctx,ZSTD_dParameter param,int * value)1685 size_t ZSTD_DCtx_getParameter(ZSTD_DCtx* dctx, ZSTD_dParameter param, int* value)
1686 {
1687 switch (param) {
1688 case ZSTD_d_windowLogMax:
1689 *value = (int)ZSTD_highbit32((U32)dctx->maxWindowSize);
1690 return 0;
1691 case ZSTD_d_format:
1692 *value = (int)dctx->format;
1693 return 0;
1694 case ZSTD_d_stableOutBuffer:
1695 *value = (int)dctx->outBufferMode;
1696 return 0;
1697 case ZSTD_d_forceIgnoreChecksum:
1698 *value = (int)dctx->forceIgnoreChecksum;
1699 return 0;
1700 case ZSTD_d_refMultipleDDicts:
1701 *value = (int)dctx->refMultipleDDicts;
1702 return 0;
1703 default:;
1704 }
1705 RETURN_ERROR(parameter_unsupported, "");
1706 }
1707
ZSTD_DCtx_setParameter(ZSTD_DCtx * dctx,ZSTD_dParameter dParam,int value)1708 size_t ZSTD_DCtx_setParameter(ZSTD_DCtx* dctx, ZSTD_dParameter dParam, int value)
1709 {
1710 RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
1711 switch(dParam) {
1712 case ZSTD_d_windowLogMax:
1713 if (value == 0) value = ZSTD_WINDOWLOG_LIMIT_DEFAULT;
1714 CHECK_DBOUNDS(ZSTD_d_windowLogMax, value);
1715 dctx->maxWindowSize = ((size_t)1) << value;
1716 return 0;
1717 case ZSTD_d_format:
1718 CHECK_DBOUNDS(ZSTD_d_format, value);
1719 dctx->format = (ZSTD_format_e)value;
1720 return 0;
1721 case ZSTD_d_stableOutBuffer:
1722 CHECK_DBOUNDS(ZSTD_d_stableOutBuffer, value);
1723 dctx->outBufferMode = (ZSTD_bufferMode_e)value;
1724 return 0;
1725 case ZSTD_d_forceIgnoreChecksum:
1726 CHECK_DBOUNDS(ZSTD_d_forceIgnoreChecksum, value);
1727 dctx->forceIgnoreChecksum = (ZSTD_forceIgnoreChecksum_e)value;
1728 return 0;
1729 case ZSTD_d_refMultipleDDicts:
1730 CHECK_DBOUNDS(ZSTD_d_refMultipleDDicts, value);
1731 if (dctx->staticSize != 0) {
1732 RETURN_ERROR(parameter_unsupported, "Static dctx does not support multiple DDicts!");
1733 }
1734 dctx->refMultipleDDicts = (ZSTD_refMultipleDDicts_e)value;
1735 return 0;
1736 default:;
1737 }
1738 RETURN_ERROR(parameter_unsupported, "");
1739 }
1740
ZSTD_DCtx_reset(ZSTD_DCtx * dctx,ZSTD_ResetDirective reset)1741 size_t ZSTD_DCtx_reset(ZSTD_DCtx* dctx, ZSTD_ResetDirective reset)
1742 {
1743 if ( (reset == ZSTD_reset_session_only)
1744 || (reset == ZSTD_reset_session_and_parameters) ) {
1745 dctx->streamStage = zdss_init;
1746 dctx->noForwardProgress = 0;
1747 }
1748 if ( (reset == ZSTD_reset_parameters)
1749 || (reset == ZSTD_reset_session_and_parameters) ) {
1750 RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
1751 ZSTD_clearDict(dctx);
1752 ZSTD_DCtx_resetParameters(dctx);
1753 }
1754 return 0;
1755 }
1756
1757
ZSTD_sizeof_DStream(const ZSTD_DStream * dctx)1758 size_t ZSTD_sizeof_DStream(const ZSTD_DStream* dctx)
1759 {
1760 return ZSTD_sizeof_DCtx(dctx);
1761 }
1762
ZSTD_decodingBufferSize_min(unsigned long long windowSize,unsigned long long frameContentSize)1763 size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long long frameContentSize)
1764 {
1765 size_t const blockSize = (size_t) MIN(windowSize, ZSTD_BLOCKSIZE_MAX);
1766 unsigned long long const neededRBSize = windowSize + blockSize + (WILDCOPY_OVERLENGTH * 2);
1767 unsigned long long const neededSize = MIN(frameContentSize, neededRBSize);
1768 size_t const minRBSize = (size_t) neededSize;
1769 RETURN_ERROR_IF((unsigned long long)minRBSize != neededSize,
1770 frameParameter_windowTooLarge, "");
1771 return minRBSize;
1772 }
1773
ZSTD_estimateDStreamSize(size_t windowSize)1774 size_t ZSTD_estimateDStreamSize(size_t windowSize)
1775 {
1776 size_t const blockSize = MIN(windowSize, ZSTD_BLOCKSIZE_MAX);
1777 size_t const inBuffSize = blockSize; /* no block can be larger */
1778 size_t const outBuffSize = ZSTD_decodingBufferSize_min(windowSize, ZSTD_CONTENTSIZE_UNKNOWN);
1779 return ZSTD_estimateDCtxSize() + inBuffSize + outBuffSize;
1780 }
1781
ZSTD_estimateDStreamSize_fromFrame(const void * src,size_t srcSize)1782 size_t ZSTD_estimateDStreamSize_fromFrame(const void* src, size_t srcSize)
1783 {
1784 U32 const windowSizeMax = 1U << ZSTD_WINDOWLOG_MAX; /* note : should be user-selectable, but requires an additional parameter (or a dctx) */
1785 ZSTD_frameHeader zfh;
1786 size_t const err = ZSTD_getFrameHeader(&zfh, src, srcSize);
1787 if (ZSTD_isError(err)) return err;
1788 RETURN_ERROR_IF(err>0, srcSize_wrong, "");
1789 RETURN_ERROR_IF(zfh.windowSize > windowSizeMax,
1790 frameParameter_windowTooLarge, "");
1791 return ZSTD_estimateDStreamSize((size_t)zfh.windowSize);
1792 }
1793
1794
1795 /* ***** Decompression ***** */
1796
ZSTD_DCtx_isOverflow(ZSTD_DStream * zds,size_t const neededInBuffSize,size_t const neededOutBuffSize)1797 static int ZSTD_DCtx_isOverflow(ZSTD_DStream* zds, size_t const neededInBuffSize, size_t const neededOutBuffSize)
1798 {
1799 return (zds->inBuffSize + zds->outBuffSize) >= (neededInBuffSize + neededOutBuffSize) * ZSTD_WORKSPACETOOLARGE_FACTOR;
1800 }
1801
ZSTD_DCtx_updateOversizedDuration(ZSTD_DStream * zds,size_t const neededInBuffSize,size_t const neededOutBuffSize)1802 static void ZSTD_DCtx_updateOversizedDuration(ZSTD_DStream* zds, size_t const neededInBuffSize, size_t const neededOutBuffSize)
1803 {
1804 if (ZSTD_DCtx_isOverflow(zds, neededInBuffSize, neededOutBuffSize))
1805 zds->oversizedDuration++;
1806 else
1807 zds->oversizedDuration = 0;
1808 }
1809
ZSTD_DCtx_isOversizedTooLong(ZSTD_DStream * zds)1810 static int ZSTD_DCtx_isOversizedTooLong(ZSTD_DStream* zds)
1811 {
1812 return zds->oversizedDuration >= ZSTD_WORKSPACETOOLARGE_MAXDURATION;
1813 }
1814
1815 /* Checks that the output buffer hasn't changed if ZSTD_obm_stable is used. */
ZSTD_checkOutBuffer(ZSTD_DStream const * zds,ZSTD_outBuffer const * output)1816 static size_t ZSTD_checkOutBuffer(ZSTD_DStream const* zds, ZSTD_outBuffer const* output)
1817 {
1818 ZSTD_outBuffer const expect = zds->expectedOutBuffer;
1819 /* No requirement when ZSTD_obm_stable is not enabled. */
1820 if (zds->outBufferMode != ZSTD_bm_stable)
1821 return 0;
1822 /* Any buffer is allowed in zdss_init, this must be the same for every other call until
1823 * the context is reset.
1824 */
1825 if (zds->streamStage == zdss_init)
1826 return 0;
1827 /* The buffer must match our expectation exactly. */
1828 if (expect.dst == output->dst && expect.pos == output->pos && expect.size == output->size)
1829 return 0;
1830 RETURN_ERROR(dstBuffer_wrong, "ZSTD_d_stableOutBuffer enabled but output differs!");
1831 }
1832
1833 /* Calls ZSTD_decompressContinue() with the right parameters for ZSTD_decompressStream()
1834 * and updates the stage and the output buffer state. This call is extracted so it can be
1835 * used both when reading directly from the ZSTD_inBuffer, and in buffered input mode.
1836 * NOTE: You must break after calling this function since the streamStage is modified.
1837 */
ZSTD_decompressContinueStream(ZSTD_DStream * zds,char ** op,char * oend,void const * src,size_t srcSize)1838 static size_t ZSTD_decompressContinueStream(
1839 ZSTD_DStream* zds, char** op, char* oend,
1840 void const* src, size_t srcSize) {
1841 int const isSkipFrame = ZSTD_isSkipFrame(zds);
1842 if (zds->outBufferMode == ZSTD_bm_buffered) {
1843 size_t const dstSize = isSkipFrame ? 0 : zds->outBuffSize - zds->outStart;
1844 size_t const decodedSize = ZSTD_decompressContinue(zds,
1845 zds->outBuff + zds->outStart, dstSize, src, srcSize);
1846 FORWARD_IF_ERROR(decodedSize, "");
1847 if (!decodedSize && !isSkipFrame) {
1848 zds->streamStage = zdss_read;
1849 } else {
1850 zds->outEnd = zds->outStart + decodedSize;
1851 zds->streamStage = zdss_flush;
1852 }
1853 } else {
1854 /* Write directly into the output buffer */
1855 size_t const dstSize = isSkipFrame ? 0 : (size_t)(oend - *op);
1856 size_t const decodedSize = ZSTD_decompressContinue(zds, *op, dstSize, src, srcSize);
1857 FORWARD_IF_ERROR(decodedSize, "");
1858 *op += decodedSize;
1859 /* Flushing is not needed. */
1860 zds->streamStage = zdss_read;
1861 assert(*op <= oend);
1862 assert(zds->outBufferMode == ZSTD_bm_stable);
1863 }
1864 return 0;
1865 }
1866
ZSTD_decompressStream(ZSTD_DStream * zds,ZSTD_outBuffer * output,ZSTD_inBuffer * input)1867 size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inBuffer* input)
1868 {
1869 const char* const src = (const char*)input->src;
1870 const char* const istart = input->pos != 0 ? src + input->pos : src;
1871 const char* const iend = input->size != 0 ? src + input->size : src;
1872 const char* ip = istart;
1873 char* const dst = (char*)output->dst;
1874 char* const ostart = output->pos != 0 ? dst + output->pos : dst;
1875 char* const oend = output->size != 0 ? dst + output->size : dst;
1876 char* op = ostart;
1877 U32 someMoreWork = 1;
1878
1879 DEBUGLOG(5, "ZSTD_decompressStream");
1880 RETURN_ERROR_IF(
1881 input->pos > input->size,
1882 srcSize_wrong,
1883 "forbidden. in: pos: %u vs size: %u",
1884 (U32)input->pos, (U32)input->size);
1885 RETURN_ERROR_IF(
1886 output->pos > output->size,
1887 dstSize_tooSmall,
1888 "forbidden. out: pos: %u vs size: %u",
1889 (U32)output->pos, (U32)output->size);
1890 DEBUGLOG(5, "input size : %u", (U32)(input->size - input->pos));
1891 FORWARD_IF_ERROR(ZSTD_checkOutBuffer(zds, output), "");
1892
1893 while (someMoreWork) {
1894 switch(zds->streamStage)
1895 {
1896 case zdss_init :
1897 DEBUGLOG(5, "stage zdss_init => transparent reset ");
1898 zds->streamStage = zdss_loadHeader;
1899 zds->lhSize = zds->inPos = zds->outStart = zds->outEnd = 0;
1900 zds->legacyVersion = 0;
1901 zds->hostageByte = 0;
1902 zds->expectedOutBuffer = *output;
1903 /* fall-through */
1904
1905 case zdss_loadHeader :
1906 DEBUGLOG(5, "stage zdss_loadHeader (srcSize : %u)", (U32)(iend - ip));
1907 #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT>=1)
1908 if (zds->legacyVersion) {
1909 RETURN_ERROR_IF(zds->staticSize, memory_allocation,
1910 "legacy support is incompatible with static dctx");
1911 { size_t const hint = ZSTD_decompressLegacyStream(zds->legacyContext, zds->legacyVersion, output, input);
1912 if (hint==0) zds->streamStage = zdss_init;
1913 return hint;
1914 } }
1915 #endif
1916 { size_t const hSize = ZSTD_getFrameHeader_advanced(&zds->fParams, zds->headerBuffer, zds->lhSize, zds->format);
1917 if (zds->refMultipleDDicts && zds->ddictSet) {
1918 ZSTD_DCtx_selectFrameDDict(zds);
1919 }
1920 DEBUGLOG(5, "header size : %u", (U32)hSize);
1921 if (ZSTD_isError(hSize)) {
1922 #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT>=1)
1923 U32 const legacyVersion = ZSTD_isLegacy(istart, iend-istart);
1924 if (legacyVersion) {
1925 ZSTD_DDict const* const ddict = ZSTD_getDDict(zds);
1926 const void* const dict = ddict ? ZSTD_DDict_dictContent(ddict) : NULL;
1927 size_t const dictSize = ddict ? ZSTD_DDict_dictSize(ddict) : 0;
1928 DEBUGLOG(5, "ZSTD_decompressStream: detected legacy version v0.%u", legacyVersion);
1929 RETURN_ERROR_IF(zds->staticSize, memory_allocation,
1930 "legacy support is incompatible with static dctx");
1931 FORWARD_IF_ERROR(ZSTD_initLegacyStream(&zds->legacyContext,
1932 zds->previousLegacyVersion, legacyVersion,
1933 dict, dictSize), "");
1934 zds->legacyVersion = zds->previousLegacyVersion = legacyVersion;
1935 { size_t const hint = ZSTD_decompressLegacyStream(zds->legacyContext, legacyVersion, output, input);
1936 if (hint==0) zds->streamStage = zdss_init; /* or stay in stage zdss_loadHeader */
1937 return hint;
1938 } }
1939 #endif
1940 return hSize; /* error */
1941 }
1942 if (hSize != 0) { /* need more input */
1943 size_t const toLoad = hSize - zds->lhSize; /* if hSize!=0, hSize > zds->lhSize */
1944 size_t const remainingInput = (size_t)(iend-ip);
1945 assert(iend >= ip);
1946 if (toLoad > remainingInput) { /* not enough input to load full header */
1947 if (remainingInput > 0) {
1948 ZSTD_memcpy(zds->headerBuffer + zds->lhSize, ip, remainingInput);
1949 zds->lhSize += remainingInput;
1950 }
1951 input->pos = input->size;
1952 return (MAX((size_t)ZSTD_FRAMEHEADERSIZE_MIN(zds->format), hSize) - zds->lhSize) + ZSTD_blockHeaderSize; /* remaining header bytes + next block header */
1953 }
1954 assert(ip != NULL);
1955 ZSTD_memcpy(zds->headerBuffer + zds->lhSize, ip, toLoad); zds->lhSize = hSize; ip += toLoad;
1956 break;
1957 } }
1958
1959 /* check for single-pass mode opportunity */
1960 if (zds->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN
1961 && zds->fParams.frameType != ZSTD_skippableFrame
1962 && (U64)(size_t)(oend-op) >= zds->fParams.frameContentSize) {
1963 size_t const cSize = ZSTD_findFrameCompressedSize(istart, (size_t)(iend-istart));
1964 if (cSize <= (size_t)(iend-istart)) {
1965 /* shortcut : using single-pass mode */
1966 size_t const decompressedSize = ZSTD_decompress_usingDDict(zds, op, (size_t)(oend-op), istart, cSize, ZSTD_getDDict(zds));
1967 if (ZSTD_isError(decompressedSize)) return decompressedSize;
1968 DEBUGLOG(4, "shortcut to single-pass ZSTD_decompress_usingDDict()")
1969 ip = istart + cSize;
1970 op += decompressedSize;
1971 zds->expected = 0;
1972 zds->streamStage = zdss_init;
1973 someMoreWork = 0;
1974 break;
1975 } }
1976
1977 /* Check output buffer is large enough for ZSTD_odm_stable. */
1978 if (zds->outBufferMode == ZSTD_bm_stable
1979 && zds->fParams.frameType != ZSTD_skippableFrame
1980 && zds->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN
1981 && (U64)(size_t)(oend-op) < zds->fParams.frameContentSize) {
1982 RETURN_ERROR(dstSize_tooSmall, "ZSTD_obm_stable passed but ZSTD_outBuffer is too small");
1983 }
1984
1985 /* Consume header (see ZSTDds_decodeFrameHeader) */
1986 DEBUGLOG(4, "Consume header");
1987 FORWARD_IF_ERROR(ZSTD_decompressBegin_usingDDict(zds, ZSTD_getDDict(zds)), "");
1988
1989 if ((MEM_readLE32(zds->headerBuffer) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) { /* skippable frame */
1990 zds->expected = MEM_readLE32(zds->headerBuffer + ZSTD_FRAMEIDSIZE);
1991 zds->stage = ZSTDds_skipFrame;
1992 } else {
1993 FORWARD_IF_ERROR(ZSTD_decodeFrameHeader(zds, zds->headerBuffer, zds->lhSize), "");
1994 zds->expected = ZSTD_blockHeaderSize;
1995 zds->stage = ZSTDds_decodeBlockHeader;
1996 }
1997
1998 /* control buffer memory usage */
1999 DEBUGLOG(4, "Control max memory usage (%u KB <= max %u KB)",
2000 (U32)(zds->fParams.windowSize >>10),
2001 (U32)(zds->maxWindowSize >> 10) );
2002 zds->fParams.windowSize = MAX(zds->fParams.windowSize, 1U << ZSTD_WINDOWLOG_ABSOLUTEMIN);
2003 RETURN_ERROR_IF(zds->fParams.windowSize > zds->maxWindowSize,
2004 frameParameter_windowTooLarge, "");
2005
2006 /* Adapt buffer sizes to frame header instructions */
2007 { size_t const neededInBuffSize = MAX(zds->fParams.blockSizeMax, 4 /* frame checksum */);
2008 size_t const neededOutBuffSize = zds->outBufferMode == ZSTD_bm_buffered
2009 ? ZSTD_decodingBufferSize_min(zds->fParams.windowSize, zds->fParams.frameContentSize)
2010 : 0;
2011
2012 ZSTD_DCtx_updateOversizedDuration(zds, neededInBuffSize, neededOutBuffSize);
2013
2014 { int const tooSmall = (zds->inBuffSize < neededInBuffSize) || (zds->outBuffSize < neededOutBuffSize);
2015 int const tooLarge = ZSTD_DCtx_isOversizedTooLong(zds);
2016
2017 if (tooSmall || tooLarge) {
2018 size_t const bufferSize = neededInBuffSize + neededOutBuffSize;
2019 DEBUGLOG(4, "inBuff : from %u to %u",
2020 (U32)zds->inBuffSize, (U32)neededInBuffSize);
2021 DEBUGLOG(4, "outBuff : from %u to %u",
2022 (U32)zds->outBuffSize, (U32)neededOutBuffSize);
2023 if (zds->staticSize) { /* static DCtx */
2024 DEBUGLOG(4, "staticSize : %u", (U32)zds->staticSize);
2025 assert(zds->staticSize >= sizeof(ZSTD_DCtx)); /* controlled at init */
2026 RETURN_ERROR_IF(
2027 bufferSize > zds->staticSize - sizeof(ZSTD_DCtx),
2028 memory_allocation, "");
2029 } else {
2030 ZSTD_customFree(zds->inBuff, zds->customMem);
2031 zds->inBuffSize = 0;
2032 zds->outBuffSize = 0;
2033 zds->inBuff = (char*)ZSTD_customMalloc(bufferSize, zds->customMem);
2034 RETURN_ERROR_IF(zds->inBuff == NULL, memory_allocation, "");
2035 }
2036 zds->inBuffSize = neededInBuffSize;
2037 zds->outBuff = zds->inBuff + zds->inBuffSize;
2038 zds->outBuffSize = neededOutBuffSize;
2039 } } }
2040 zds->streamStage = zdss_read;
2041 /* fall-through */
2042
2043 case zdss_read:
2044 DEBUGLOG(5, "stage zdss_read");
2045 { size_t const neededInSize = ZSTD_nextSrcSizeToDecompressWithInputSize(zds, (size_t)(iend - ip));
2046 DEBUGLOG(5, "neededInSize = %u", (U32)neededInSize);
2047 if (neededInSize==0) { /* end of frame */
2048 zds->streamStage = zdss_init;
2049 someMoreWork = 0;
2050 break;
2051 }
2052 if ((size_t)(iend-ip) >= neededInSize) { /* decode directly from src */
2053 FORWARD_IF_ERROR(ZSTD_decompressContinueStream(zds, &op, oend, ip, neededInSize), "");
2054 ip += neededInSize;
2055 /* Function modifies the stage so we must break */
2056 break;
2057 } }
2058 if (ip==iend) { someMoreWork = 0; break; } /* no more input */
2059 zds->streamStage = zdss_load;
2060 /* fall-through */
2061
2062 case zdss_load:
2063 { size_t const neededInSize = ZSTD_nextSrcSizeToDecompress(zds);
2064 size_t const toLoad = neededInSize - zds->inPos;
2065 int const isSkipFrame = ZSTD_isSkipFrame(zds);
2066 size_t loadedSize;
2067 /* At this point we shouldn't be decompressing a block that we can stream. */
2068 assert(neededInSize == ZSTD_nextSrcSizeToDecompressWithInputSize(zds, iend - ip));
2069 if (isSkipFrame) {
2070 loadedSize = MIN(toLoad, (size_t)(iend-ip));
2071 } else {
2072 RETURN_ERROR_IF(toLoad > zds->inBuffSize - zds->inPos,
2073 corruption_detected,
2074 "should never happen");
2075 loadedSize = ZSTD_limitCopy(zds->inBuff + zds->inPos, toLoad, ip, (size_t)(iend-ip));
2076 }
2077 ip += loadedSize;
2078 zds->inPos += loadedSize;
2079 if (loadedSize < toLoad) { someMoreWork = 0; break; } /* not enough input, wait for more */
2080
2081 /* decode loaded input */
2082 zds->inPos = 0; /* input is consumed */
2083 FORWARD_IF_ERROR(ZSTD_decompressContinueStream(zds, &op, oend, zds->inBuff, neededInSize), "");
2084 /* Function modifies the stage so we must break */
2085 break;
2086 }
2087 case zdss_flush:
2088 { size_t const toFlushSize = zds->outEnd - zds->outStart;
2089 size_t const flushedSize = ZSTD_limitCopy(op, (size_t)(oend-op), zds->outBuff + zds->outStart, toFlushSize);
2090 op += flushedSize;
2091 zds->outStart += flushedSize;
2092 if (flushedSize == toFlushSize) { /* flush completed */
2093 zds->streamStage = zdss_read;
2094 if ( (zds->outBuffSize < zds->fParams.frameContentSize)
2095 && (zds->outStart + zds->fParams.blockSizeMax > zds->outBuffSize) ) {
2096 DEBUGLOG(5, "restart filling outBuff from beginning (left:%i, needed:%u)",
2097 (int)(zds->outBuffSize - zds->outStart),
2098 (U32)zds->fParams.blockSizeMax);
2099 zds->outStart = zds->outEnd = 0;
2100 }
2101 break;
2102 } }
2103 /* cannot complete flush */
2104 someMoreWork = 0;
2105 break;
2106
2107 default:
2108 assert(0); /* impossible */
2109 RETURN_ERROR(GENERIC, "impossible to reach"); /* some compiler require default to do something */
2110 } }
2111
2112 /* result */
2113 input->pos = (size_t)(ip - (const char*)(input->src));
2114 output->pos = (size_t)(op - (char*)(output->dst));
2115
2116 /* Update the expected output buffer for ZSTD_obm_stable. */
2117 zds->expectedOutBuffer = *output;
2118
2119 if ((ip==istart) && (op==ostart)) { /* no forward progress */
2120 zds->noForwardProgress ++;
2121 if (zds->noForwardProgress >= ZSTD_NO_FORWARD_PROGRESS_MAX) {
2122 RETURN_ERROR_IF(op==oend, dstSize_tooSmall, "");
2123 RETURN_ERROR_IF(ip==iend, srcSize_wrong, "");
2124 assert(0);
2125 }
2126 } else {
2127 zds->noForwardProgress = 0;
2128 }
2129 { size_t nextSrcSizeHint = ZSTD_nextSrcSizeToDecompress(zds);
2130 if (!nextSrcSizeHint) { /* frame fully decoded */
2131 if (zds->outEnd == zds->outStart) { /* output fully flushed */
2132 if (zds->hostageByte) {
2133 if (input->pos >= input->size) {
2134 /* can't release hostage (not present) */
2135 zds->streamStage = zdss_read;
2136 return 1;
2137 }
2138 input->pos++; /* release hostage */
2139 } /* zds->hostageByte */
2140 return 0;
2141 } /* zds->outEnd == zds->outStart */
2142 if (!zds->hostageByte) { /* output not fully flushed; keep last byte as hostage; will be released when all output is flushed */
2143 input->pos--; /* note : pos > 0, otherwise, impossible to finish reading last block */
2144 zds->hostageByte=1;
2145 }
2146 return 1;
2147 } /* nextSrcSizeHint==0 */
2148 nextSrcSizeHint += ZSTD_blockHeaderSize * (ZSTD_nextInputType(zds) == ZSTDnit_block); /* preload header of next block */
2149 assert(zds->inPos <= nextSrcSizeHint);
2150 nextSrcSizeHint -= zds->inPos; /* part already loaded*/
2151 return nextSrcSizeHint;
2152 }
2153 }
2154
ZSTD_decompressStream_simpleArgs(ZSTD_DCtx * dctx,void * dst,size_t dstCapacity,size_t * dstPos,const void * src,size_t srcSize,size_t * srcPos)2155 size_t ZSTD_decompressStream_simpleArgs (
2156 ZSTD_DCtx* dctx,
2157 void* dst, size_t dstCapacity, size_t* dstPos,
2158 const void* src, size_t srcSize, size_t* srcPos)
2159 {
2160 ZSTD_outBuffer output = { dst, dstCapacity, *dstPos };
2161 ZSTD_inBuffer input = { src, srcSize, *srcPos };
2162 /* ZSTD_compress_generic() will check validity of dstPos and srcPos */
2163 size_t const cErr = ZSTD_decompressStream(dctx, &output, &input);
2164 *dstPos = output.pos;
2165 *srcPos = input.pos;
2166 return cErr;
2167 }
2168