1 | /* |
2 | * Copyright (c) 2016-present, 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 | /* This header contains definitions |
12 | * that shall **only** be used by modules within lib/compress. |
13 | */ |
14 | |
15 | #ifndef ZSTD_COMPRESS_H |
16 | #define ZSTD_COMPRESS_H |
17 | |
18 | /*-************************************* |
19 | * Dependencies |
20 | ***************************************/ |
21 | #include "zstd_internal.h" |
22 | #ifdef ZSTD_MULTITHREAD |
23 | # include "zstdmt_compress.h" |
24 | #endif |
25 | |
26 | #if defined (__cplusplus) |
27 | extern "C" { |
28 | #endif |
29 | |
30 | /*-************************************* |
31 | * Constants |
32 | ***************************************/ |
33 | #define kSearchStrength 8 |
34 | #define HASH_READ_SIZE 8 |
35 | #define ZSTD_DUBT_UNSORTED_MARK 1 /* For btlazy2 strategy, index 1 now means "unsorted". |
36 | It could be confused for a real successor at index "1", if sorted as larger than its predecessor. |
37 | It's not a big deal though : candidate will just be sorted again. |
38 | Additionnally, candidate position 1 will be lost. |
39 | But candidate 1 cannot hide a large tree of candidates, so it's a minimal loss. |
40 | The benefit is that ZSTD_DUBT_UNSORTED_MARK cannot be misdhandled after table re-use with a different strategy */ |
41 | |
42 | |
43 | /*-************************************* |
44 | * Context memory management |
45 | ***************************************/ |
46 | typedef enum { ZSTDcs_created=0, ZSTDcs_init, ZSTDcs_ongoing, ZSTDcs_ending } ZSTD_compressionStage_e; |
47 | typedef enum { zcss_init=0, zcss_load, zcss_flush } ZSTD_cStreamStage; |
48 | |
49 | typedef struct ZSTD_prefixDict_s { |
50 | const void* dict; |
51 | size_t dictSize; |
52 | ZSTD_dictContentType_e dictContentType; |
53 | } ZSTD_prefixDict; |
54 | |
55 | typedef struct { |
56 | U32 hufCTable[HUF_CTABLE_SIZE_U32(255)]; |
57 | FSE_CTable offcodeCTable[FSE_CTABLE_SIZE_U32(OffFSELog, MaxOff)]; |
58 | FSE_CTable matchlengthCTable[FSE_CTABLE_SIZE_U32(MLFSELog, MaxML)]; |
59 | FSE_CTable litlengthCTable[FSE_CTABLE_SIZE_U32(LLFSELog, MaxLL)]; |
60 | HUF_repeat hufCTable_repeatMode; |
61 | FSE_repeat offcode_repeatMode; |
62 | FSE_repeat matchlength_repeatMode; |
63 | FSE_repeat litlength_repeatMode; |
64 | } ZSTD_entropyCTables_t; |
65 | |
66 | typedef struct { |
67 | U32 off; |
68 | U32 len; |
69 | } ZSTD_match_t; |
70 | |
71 | typedef struct { |
72 | int price; |
73 | U32 off; |
74 | U32 mlen; |
75 | U32 litlen; |
76 | U32 rep[ZSTD_REP_NUM]; |
77 | } ZSTD_optimal_t; |
78 | |
79 | typedef struct { |
80 | /* All tables are allocated inside cctx->workspace by ZSTD_resetCCtx_internal() */ |
81 | U32* litFreq; /* table of literals statistics, of size 256 */ |
82 | U32* litLengthFreq; /* table of litLength statistics, of size (MaxLL+1) */ |
83 | U32* matchLengthFreq; /* table of matchLength statistics, of size (MaxML+1) */ |
84 | U32* offCodeFreq; /* table of offCode statistics, of size (MaxOff+1) */ |
85 | ZSTD_match_t* matchTable; /* list of found matches, of size ZSTD_OPT_NUM+1 */ |
86 | ZSTD_optimal_t* priceTable; /* All positions tracked by optimal parser, of size ZSTD_OPT_NUM+1 */ |
87 | |
88 | U32 litSum; /* nb of literals */ |
89 | U32 litLengthSum; /* nb of litLength codes */ |
90 | U32 matchLengthSum; /* nb of matchLength codes */ |
91 | U32 offCodeSum; /* nb of offset codes */ |
92 | /* begin updated by ZSTD_setLog2Prices */ |
93 | U32 log2litSum; /* pow2 to compare log2(litfreq) to */ |
94 | U32 log2litLengthSum; /* pow2 to compare log2(llfreq) to */ |
95 | U32 log2matchLengthSum; /* pow2 to compare log2(mlfreq) to */ |
96 | U32 log2offCodeSum; /* pow2 to compare log2(offreq) to */ |
97 | /* end : updated by ZSTD_setLog2Prices */ |
98 | U32 staticPrices; /* prices follow a pre-defined cost structure, statistics are irrelevant */ |
99 | } optState_t; |
100 | |
101 | typedef struct { |
102 | ZSTD_entropyCTables_t entropy; |
103 | U32 rep[ZSTD_REP_NUM]; |
104 | } ZSTD_compressedBlockState_t; |
105 | |
106 | typedef struct { |
107 | BYTE const* nextSrc; /* next block here to continue on current prefix */ |
108 | BYTE const* base; /* All regular indexes relative to this position */ |
109 | BYTE const* dictBase; /* extDict indexes relative to this position */ |
110 | U32 dictLimit; /* below that point, need extDict */ |
111 | U32 lowLimit; /* below that point, no more data */ |
112 | } ZSTD_window_t; |
113 | |
114 | typedef struct { |
115 | ZSTD_window_t window; /* State for window round buffer management */ |
116 | U32 loadedDictEnd; /* index of end of dictionary */ |
117 | U32 nextToUpdate; /* index from which to continue table update */ |
118 | U32 nextToUpdate3; /* index from which to continue table update */ |
119 | U32 hashLog3; /* dispatch table : larger == faster, more memory */ |
120 | U32* hashTable; |
121 | U32* hashTable3; |
122 | U32* chainTable; |
123 | optState_t opt; /* optimal parser state */ |
124 | } ZSTD_matchState_t; |
125 | |
126 | typedef struct { |
127 | ZSTD_compressedBlockState_t* prevCBlock; |
128 | ZSTD_compressedBlockState_t* nextCBlock; |
129 | ZSTD_matchState_t matchState; |
130 | } ZSTD_blockState_t; |
131 | |
132 | typedef struct { |
133 | U32 offset; |
134 | U32 checksum; |
135 | } ldmEntry_t; |
136 | |
137 | typedef struct { |
138 | ZSTD_window_t window; /* State for the window round buffer management */ |
139 | ldmEntry_t* hashTable; |
140 | BYTE* bucketOffsets; /* Next position in bucket to insert entry */ |
141 | U64 hashPower; /* Used to compute the rolling hash. |
142 | * Depends on ldmParams.minMatchLength */ |
143 | } ldmState_t; |
144 | |
145 | typedef struct { |
146 | U32 enableLdm; /* 1 if enable long distance matching */ |
147 | U32 hashLog; /* Log size of hashTable */ |
148 | U32 bucketSizeLog; /* Log bucket size for collision resolution, at most 8 */ |
149 | U32 minMatchLength; /* Minimum match length */ |
150 | U32 hashEveryLog; /* Log number of entries to skip */ |
151 | U32 windowLog; /* Window log for the LDM */ |
152 | } ldmParams_t; |
153 | |
154 | typedef struct { |
155 | U32 offset; |
156 | U32 litLength; |
157 | U32 matchLength; |
158 | } rawSeq; |
159 | |
160 | typedef struct { |
161 | rawSeq* seq; /* The start of the sequences */ |
162 | size_t pos; /* The position where reading stopped. <= size. */ |
163 | size_t size; /* The number of sequences. <= capacity. */ |
164 | size_t capacity; /* The capacity of the `seq` pointer */ |
165 | } rawSeqStore_t; |
166 | |
167 | struct ZSTD_CCtx_params_s { |
168 | ZSTD_format_e format; |
169 | ZSTD_compressionParameters cParams; |
170 | ZSTD_frameParameters fParams; |
171 | |
172 | int compressionLevel; |
173 | int disableLiteralCompression; |
174 | int forceWindow; /* force back-references to respect limit of |
175 | * 1<<wLog, even for dictionary */ |
176 | |
177 | /* Multithreading: used to pass parameters to mtctx */ |
178 | unsigned nbWorkers; |
179 | unsigned jobSize; |
180 | unsigned overlapSizeLog; |
181 | |
182 | /* Long distance matching parameters */ |
183 | ldmParams_t ldmParams; |
184 | |
185 | /* Internal use, for createCCtxParams() and freeCCtxParams() only */ |
186 | ZSTD_customMem customMem; |
187 | }; /* typedef'd to ZSTD_CCtx_params within "zstd.h" */ |
188 | |
189 | struct ZSTD_CCtx_s { |
190 | ZSTD_compressionStage_e stage; |
191 | int cParamsChanged; /* == 1 if cParams(except wlog) or compression level are changed in requestedParams. Triggers transmission of new params to ZSTDMT (if available) then reset to 0. */ |
192 | int bmi2; /* == 1 if the CPU supports BMI2 and 0 otherwise. CPU support is determined dynamically once per context lifetime. */ |
193 | ZSTD_CCtx_params requestedParams; |
194 | ZSTD_CCtx_params appliedParams; |
195 | U32 dictID; |
196 | void* workSpace; |
197 | size_t workSpaceSize; |
198 | size_t blockSize; |
199 | unsigned long long pledgedSrcSizePlusOne; /* this way, 0 (default) == unknown */ |
200 | unsigned long long consumedSrcSize; |
201 | unsigned long long producedCSize; |
202 | XXH64_state_t xxhState; |
203 | ZSTD_customMem customMem; |
204 | size_t staticSize; |
205 | |
206 | seqStore_t seqStore; /* sequences storage ptrs */ |
207 | ldmState_t ldmState; /* long distance matching state */ |
208 | rawSeq* ldmSequences; /* Storage for the ldm output sequences */ |
209 | size_t maxNbLdmSequences; |
210 | rawSeqStore_t externSeqStore; /* Mutable reference to external sequences */ |
211 | ZSTD_blockState_t blockState; |
212 | U32* entropyWorkspace; /* entropy workspace of HUF_WORKSPACE_SIZE bytes */ |
213 | |
214 | /* streaming */ |
215 | char* inBuff; |
216 | size_t inBuffSize; |
217 | size_t inToCompress; |
218 | size_t inBuffPos; |
219 | size_t inBuffTarget; |
220 | char* outBuff; |
221 | size_t outBuffSize; |
222 | size_t outBuffContentSize; |
223 | size_t outBuffFlushedSize; |
224 | ZSTD_cStreamStage streamStage; |
225 | U32 frameEnded; |
226 | |
227 | /* Dictionary */ |
228 | ZSTD_CDict* cdictLocal; |
229 | const ZSTD_CDict* cdict; |
230 | ZSTD_prefixDict prefixDict; /* single-usage dictionary */ |
231 | |
232 | /* Multi-threading */ |
233 | #ifdef ZSTD_MULTITHREAD |
234 | ZSTDMT_CCtx* mtctx; |
235 | #endif |
236 | }; |
237 | |
238 | |
239 | typedef size_t (*ZSTD_blockCompressor) ( |
240 | ZSTD_matchState_t* bs, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], |
241 | ZSTD_compressionParameters const* cParams, void const* src, size_t srcSize); |
242 | ZSTD_blockCompressor ZSTD_selectBlockCompressor(ZSTD_strategy strat, int extDict); |
243 | |
244 | |
245 | MEM_STATIC U32 ZSTD_LLcode(U32 litLength) |
246 | { |
247 | static const BYTE LL_Code[64] = { 0, 1, 2, 3, 4, 5, 6, 7, |
248 | 8, 9, 10, 11, 12, 13, 14, 15, |
249 | 16, 16, 17, 17, 18, 18, 19, 19, |
250 | 20, 20, 20, 20, 21, 21, 21, 21, |
251 | 22, 22, 22, 22, 22, 22, 22, 22, |
252 | 23, 23, 23, 23, 23, 23, 23, 23, |
253 | 24, 24, 24, 24, 24, 24, 24, 24, |
254 | 24, 24, 24, 24, 24, 24, 24, 24 }; |
255 | static const U32 LL_deltaCode = 19; |
256 | return (litLength > 63) ? ZSTD_highbit32(litLength) + LL_deltaCode : LL_Code[litLength]; |
257 | } |
258 | |
259 | /* ZSTD_MLcode() : |
260 | * note : mlBase = matchLength - MINMATCH; |
261 | * because it's the format it's stored in seqStore->sequences */ |
262 | MEM_STATIC U32 ZSTD_MLcode(U32 mlBase) |
263 | { |
264 | static const BYTE ML_Code[128] = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, |
265 | 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, |
266 | 32, 32, 33, 33, 34, 34, 35, 35, 36, 36, 36, 36, 37, 37, 37, 37, |
267 | 38, 38, 38, 38, 38, 38, 38, 38, 39, 39, 39, 39, 39, 39, 39, 39, |
268 | 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, |
269 | 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, |
270 | 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, |
271 | 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42 }; |
272 | static const U32 ML_deltaCode = 36; |
273 | return (mlBase > 127) ? ZSTD_highbit32(mlBase) + ML_deltaCode : ML_Code[mlBase]; |
274 | } |
275 | |
276 | /*! ZSTD_storeSeq() : |
277 | * Store a sequence (literal length, literals, offset code and match length code) into seqStore_t. |
278 | * `offsetCode` : distance to match + 3 (values 1-3 are repCodes). |
279 | * `mlBase` : matchLength - MINMATCH |
280 | */ |
281 | MEM_STATIC void ZSTD_storeSeq(seqStore_t* seqStorePtr, size_t litLength, const void* literals, U32 offsetCode, size_t mlBase) |
282 | { |
283 | #if defined(ZSTD_DEBUG) && (ZSTD_DEBUG >= 6) |
284 | static const BYTE* g_start = NULL; |
285 | if (g_start==NULL) g_start = (const BYTE*)literals; /* note : index only works for compression within a single segment */ |
286 | { U32 const pos = (U32)((const BYTE*)literals - g_start); |
287 | DEBUGLOG(6, "Cpos%7u :%3u literals, match%3u bytes at dist.code%7u" , |
288 | pos, (U32)litLength, (U32)mlBase+MINMATCH, (U32)offsetCode); |
289 | } |
290 | #endif |
291 | /* copy Literals */ |
292 | assert(seqStorePtr->lit + litLength <= seqStorePtr->litStart + 128 KB); |
293 | ZSTD_wildcopy(seqStorePtr->lit, literals, litLength); |
294 | seqStorePtr->lit += litLength; |
295 | |
296 | /* literal Length */ |
297 | if (litLength>0xFFFF) { |
298 | assert(seqStorePtr->longLengthID == 0); /* there can only be a single long length */ |
299 | seqStorePtr->longLengthID = 1; |
300 | seqStorePtr->longLengthPos = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart); |
301 | } |
302 | seqStorePtr->sequences[0].litLength = (U16)litLength; |
303 | |
304 | /* match offset */ |
305 | seqStorePtr->sequences[0].offset = offsetCode + 1; |
306 | |
307 | /* match Length */ |
308 | if (mlBase>0xFFFF) { |
309 | assert(seqStorePtr->longLengthID == 0); /* there can only be a single long length */ |
310 | seqStorePtr->longLengthID = 2; |
311 | seqStorePtr->longLengthPos = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart); |
312 | } |
313 | seqStorePtr->sequences[0].matchLength = (U16)mlBase; |
314 | |
315 | seqStorePtr->sequences++; |
316 | } |
317 | |
318 | |
319 | /*-************************************* |
320 | * Match length counter |
321 | ***************************************/ |
322 | static unsigned ZSTD_NbCommonBytes (size_t val) |
323 | { |
324 | if (MEM_isLittleEndian()) { |
325 | if (MEM_64bits()) { |
326 | # if defined(_MSC_VER) && defined(_WIN64) |
327 | unsigned long r = 0; |
328 | _BitScanForward64( &r, (U64)val ); |
329 | return (unsigned)(r>>3); |
330 | # elif defined(__GNUC__) && (__GNUC__ >= 4) |
331 | return (__builtin_ctzll((U64)val) >> 3); |
332 | # else |
333 | static const int DeBruijnBytePos[64] = { 0, 0, 0, 0, 0, 1, 1, 2, |
334 | 0, 3, 1, 3, 1, 4, 2, 7, |
335 | 0, 2, 3, 6, 1, 5, 3, 5, |
336 | 1, 3, 4, 4, 2, 5, 6, 7, |
337 | 7, 0, 1, 2, 3, 3, 4, 6, |
338 | 2, 6, 5, 5, 3, 4, 5, 6, |
339 | 7, 1, 2, 4, 6, 4, 4, 5, |
340 | 7, 2, 6, 5, 7, 6, 7, 7 }; |
341 | return DeBruijnBytePos[((U64)((val & -(long long)val) * 0x0218A392CDABBD3FULL)) >> 58]; |
342 | # endif |
343 | } else { /* 32 bits */ |
344 | # if defined(_MSC_VER) |
345 | unsigned long r=0; |
346 | _BitScanForward( &r, (U32)val ); |
347 | return (unsigned)(r>>3); |
348 | # elif defined(__GNUC__) && (__GNUC__ >= 3) |
349 | return (__builtin_ctz((U32)val) >> 3); |
350 | # else |
351 | static const int DeBruijnBytePos[32] = { 0, 0, 3, 0, 3, 1, 3, 0, |
352 | 3, 2, 2, 1, 3, 2, 0, 1, |
353 | 3, 3, 1, 2, 2, 2, 2, 0, |
354 | 3, 1, 2, 0, 1, 0, 1, 1 }; |
355 | return DeBruijnBytePos[((U32)((val & -(S32)val) * 0x077CB531U)) >> 27]; |
356 | # endif |
357 | } |
358 | } else { /* Big Endian CPU */ |
359 | if (MEM_64bits()) { |
360 | # if defined(_MSC_VER) && defined(_WIN64) |
361 | unsigned long r = 0; |
362 | _BitScanReverse64( &r, val ); |
363 | return (unsigned)(r>>3); |
364 | # elif defined(__GNUC__) && (__GNUC__ >= 4) |
365 | return (__builtin_clzll(val) >> 3); |
366 | # else |
367 | unsigned r; |
368 | const unsigned n32 = sizeof(size_t)*4; /* calculate this way due to compiler complaining in 32-bits mode */ |
369 | if (!(val>>n32)) { r=4; } else { r=0; val>>=n32; } |
370 | if (!(val>>16)) { r+=2; val>>=8; } else { val>>=24; } |
371 | r += (!val); |
372 | return r; |
373 | # endif |
374 | } else { /* 32 bits */ |
375 | # if defined(_MSC_VER) |
376 | unsigned long r = 0; |
377 | _BitScanReverse( &r, (unsigned long)val ); |
378 | return (unsigned)(r>>3); |
379 | # elif defined(__GNUC__) && (__GNUC__ >= 3) |
380 | return (__builtin_clz((U32)val) >> 3); |
381 | # else |
382 | unsigned r; |
383 | if (!(val>>16)) { r=2; val>>=8; } else { r=0; val>>=24; } |
384 | r += (!val); |
385 | return r; |
386 | # endif |
387 | } } |
388 | } |
389 | |
390 | |
391 | MEM_STATIC size_t ZSTD_count(const BYTE* pIn, const BYTE* pMatch, const BYTE* const pInLimit) |
392 | { |
393 | const BYTE* const pStart = pIn; |
394 | const BYTE* const pInLoopLimit = pInLimit - (sizeof(size_t)-1); |
395 | |
396 | if (pIn < pInLoopLimit) { |
397 | { size_t const diff = MEM_readST(pMatch) ^ MEM_readST(pIn); |
398 | if (diff) return ZSTD_NbCommonBytes(diff); } |
399 | pIn+=sizeof(size_t); pMatch+=sizeof(size_t); |
400 | while (pIn < pInLoopLimit) { |
401 | size_t const diff = MEM_readST(pMatch) ^ MEM_readST(pIn); |
402 | if (!diff) { pIn+=sizeof(size_t); pMatch+=sizeof(size_t); continue; } |
403 | pIn += ZSTD_NbCommonBytes(diff); |
404 | return (size_t)(pIn - pStart); |
405 | } } |
406 | if (MEM_64bits() && (pIn<(pInLimit-3)) && (MEM_read32(pMatch) == MEM_read32(pIn))) { pIn+=4; pMatch+=4; } |
407 | if ((pIn<(pInLimit-1)) && (MEM_read16(pMatch) == MEM_read16(pIn))) { pIn+=2; pMatch+=2; } |
408 | if ((pIn<pInLimit) && (*pMatch == *pIn)) pIn++; |
409 | return (size_t)(pIn - pStart); |
410 | } |
411 | |
412 | /** ZSTD_count_2segments() : |
413 | * can count match length with `ip` & `match` in 2 different segments. |
414 | * convention : on reaching mEnd, match count continue starting from iStart |
415 | */ |
416 | MEM_STATIC size_t |
417 | ZSTD_count_2segments(const BYTE* ip, const BYTE* match, |
418 | const BYTE* iEnd, const BYTE* mEnd, const BYTE* iStart) |
419 | { |
420 | const BYTE* const vEnd = MIN( ip + (mEnd - match), iEnd); |
421 | size_t const matchLength = ZSTD_count(ip, match, vEnd); |
422 | if (match + matchLength != mEnd) return matchLength; |
423 | return matchLength + ZSTD_count(ip+matchLength, iStart, iEnd); |
424 | } |
425 | |
426 | |
427 | /*-************************************* |
428 | * Hashes |
429 | ***************************************/ |
430 | static const U32 prime3bytes = 506832829U; |
431 | static U32 ZSTD_hash3(U32 u, U32 h) { return ((u << (32-24)) * prime3bytes) >> (32-h) ; } |
432 | MEM_STATIC size_t ZSTD_hash3Ptr(const void* ptr, U32 h) { return ZSTD_hash3(MEM_readLE32(ptr), h); } /* only in zstd_opt.h */ |
433 | |
434 | static const U32 prime4bytes = 2654435761U; |
435 | static U32 ZSTD_hash4(U32 u, U32 h) { return (u * prime4bytes) >> (32-h) ; } |
436 | static size_t ZSTD_hash4Ptr(const void* ptr, U32 h) { return ZSTD_hash4(MEM_read32(ptr), h); } |
437 | |
438 | static const U64 prime5bytes = 889523592379ULL; |
439 | static size_t ZSTD_hash5(U64 u, U32 h) { return (size_t)(((u << (64-40)) * prime5bytes) >> (64-h)) ; } |
440 | static size_t ZSTD_hash5Ptr(const void* p, U32 h) { return ZSTD_hash5(MEM_readLE64(p), h); } |
441 | |
442 | static const U64 prime6bytes = 227718039650203ULL; |
443 | static size_t ZSTD_hash6(U64 u, U32 h) { return (size_t)(((u << (64-48)) * prime6bytes) >> (64-h)) ; } |
444 | static size_t ZSTD_hash6Ptr(const void* p, U32 h) { return ZSTD_hash6(MEM_readLE64(p), h); } |
445 | |
446 | static const U64 prime7bytes = 58295818150454627ULL; |
447 | static size_t ZSTD_hash7(U64 u, U32 h) { return (size_t)(((u << (64-56)) * prime7bytes) >> (64-h)) ; } |
448 | static size_t ZSTD_hash7Ptr(const void* p, U32 h) { return ZSTD_hash7(MEM_readLE64(p), h); } |
449 | |
450 | static const U64 prime8bytes = 0xCF1BBCDCB7A56463ULL; |
451 | static size_t ZSTD_hash8(U64 u, U32 h) { return (size_t)(((u) * prime8bytes) >> (64-h)) ; } |
452 | static size_t ZSTD_hash8Ptr(const void* p, U32 h) { return ZSTD_hash8(MEM_readLE64(p), h); } |
453 | |
454 | MEM_STATIC size_t ZSTD_hashPtr(const void* p, U32 hBits, U32 mls) |
455 | { |
456 | switch(mls) |
457 | { |
458 | default: |
459 | case 4: return ZSTD_hash4Ptr(p, hBits); |
460 | case 5: return ZSTD_hash5Ptr(p, hBits); |
461 | case 6: return ZSTD_hash6Ptr(p, hBits); |
462 | case 7: return ZSTD_hash7Ptr(p, hBits); |
463 | case 8: return ZSTD_hash8Ptr(p, hBits); |
464 | } |
465 | } |
466 | |
467 | /*-************************************* |
468 | * Round buffer management |
469 | ***************************************/ |
470 | /* Max current allowed */ |
471 | #define ZSTD_CURRENT_MAX ((3U << 29) + (1U << ZSTD_WINDOWLOG_MAX)) |
472 | /* Maximum chunk size before overflow correction needs to be called again */ |
473 | #define ZSTD_CHUNKSIZE_MAX \ |
474 | ( ((U32)-1) /* Maximum ending current index */ \ |
475 | - ZSTD_CURRENT_MAX) /* Maximum beginning lowLimit */ |
476 | |
477 | /** |
478 | * ZSTD_window_clear(): |
479 | * Clears the window containing the history by simply setting it to empty. |
480 | */ |
481 | MEM_STATIC void ZSTD_window_clear(ZSTD_window_t* window) |
482 | { |
483 | size_t const endT = (size_t)(window->nextSrc - window->base); |
484 | U32 const end = (U32)endT; |
485 | |
486 | window->lowLimit = end; |
487 | window->dictLimit = end; |
488 | } |
489 | |
490 | /** |
491 | * ZSTD_window_hasExtDict(): |
492 | * Returns non-zero if the window has a non-empty extDict. |
493 | */ |
494 | MEM_STATIC U32 ZSTD_window_hasExtDict(ZSTD_window_t const window) |
495 | { |
496 | return window.lowLimit < window.dictLimit; |
497 | } |
498 | |
499 | /** |
500 | * ZSTD_window_needOverflowCorrection(): |
501 | * Returns non-zero if the indices are getting too large and need overflow |
502 | * protection. |
503 | */ |
504 | MEM_STATIC U32 ZSTD_window_needOverflowCorrection(ZSTD_window_t const window, |
505 | void const* srcEnd) |
506 | { |
507 | U32 const current = (U32)((BYTE const*)srcEnd - window.base); |
508 | return current > ZSTD_CURRENT_MAX; |
509 | } |
510 | |
511 | /** |
512 | * ZSTD_window_correctOverflow(): |
513 | * Reduces the indices to protect from index overflow. |
514 | * Returns the correction made to the indices, which must be applied to every |
515 | * stored index. |
516 | * |
517 | * The least significant cycleLog bits of the indices must remain the same, |
518 | * which may be 0. Every index up to maxDist in the past must be valid. |
519 | * NOTE: (maxDist & cycleMask) must be zero. |
520 | */ |
521 | MEM_STATIC U32 ZSTD_window_correctOverflow(ZSTD_window_t* window, U32 cycleLog, |
522 | U32 maxDist, void const* src) |
523 | { |
524 | /* preemptive overflow correction: |
525 | * 1. correction is large enough: |
526 | * lowLimit > (3<<29) ==> current > 3<<29 + 1<<windowLog |
527 | * 1<<windowLog <= newCurrent < 1<<chainLog + 1<<windowLog |
528 | * |
529 | * current - newCurrent |
530 | * > (3<<29 + 1<<windowLog) - (1<<windowLog + 1<<chainLog) |
531 | * > (3<<29) - (1<<chainLog) |
532 | * > (3<<29) - (1<<30) (NOTE: chainLog <= 30) |
533 | * > 1<<29 |
534 | * |
535 | * 2. (ip+ZSTD_CHUNKSIZE_MAX - cctx->base) doesn't overflow: |
536 | * After correction, current is less than (1<<chainLog + 1<<windowLog). |
537 | * In 64-bit mode we are safe, because we have 64-bit ptrdiff_t. |
538 | * In 32-bit mode we are safe, because (chainLog <= 29), so |
539 | * ip+ZSTD_CHUNKSIZE_MAX - cctx->base < 1<<32. |
540 | * 3. (cctx->lowLimit + 1<<windowLog) < 1<<32: |
541 | * windowLog <= 31 ==> 3<<29 + 1<<windowLog < 7<<29 < 1<<32. |
542 | */ |
543 | U32 const cycleMask = (1U << cycleLog) - 1; |
544 | U32 const current = (U32)((BYTE const*)src - window->base); |
545 | U32 const newCurrent = (current & cycleMask) + maxDist; |
546 | U32 const correction = current - newCurrent; |
547 | assert((maxDist & cycleMask) == 0); |
548 | assert(current > newCurrent); |
549 | /* Loose bound, should be around 1<<29 (see above) */ |
550 | assert(correction > 1<<28); |
551 | |
552 | window->base += correction; |
553 | window->dictBase += correction; |
554 | window->lowLimit -= correction; |
555 | window->dictLimit -= correction; |
556 | |
557 | DEBUGLOG(4, "Correction of 0x%x bytes to lowLimit=0x%x" , correction, |
558 | window->lowLimit); |
559 | return correction; |
560 | } |
561 | |
562 | /** |
563 | * ZSTD_window_enforceMaxDist(): |
564 | * Updates lowLimit so that: |
565 | * (srcEnd - base) - lowLimit == maxDist + loadedDictEnd |
566 | * This allows a simple check that index >= lowLimit to see if index is valid. |
567 | * This must be called before a block compression call, with srcEnd as the block |
568 | * source end. |
569 | * If loadedDictEndPtr is not NULL, we set it to zero once we update lowLimit. |
570 | * This is because dictionaries are allowed to be referenced as long as the last |
571 | * byte of the dictionary is in the window, but once they are out of range, |
572 | * they cannot be referenced. If loadedDictEndPtr is NULL, we use |
573 | * loadedDictEnd == 0. |
574 | */ |
575 | MEM_STATIC void ZSTD_window_enforceMaxDist(ZSTD_window_t* window, |
576 | void const* srcEnd, U32 maxDist, |
577 | U32* loadedDictEndPtr) |
578 | { |
579 | U32 const current = (U32)((BYTE const*)srcEnd - window->base); |
580 | U32 loadedDictEnd = loadedDictEndPtr != NULL ? *loadedDictEndPtr : 0; |
581 | if (current > maxDist + loadedDictEnd) { |
582 | U32 const newLowLimit = current - maxDist; |
583 | if (window->lowLimit < newLowLimit) window->lowLimit = newLowLimit; |
584 | if (window->dictLimit < window->lowLimit) { |
585 | DEBUGLOG(5, "Update dictLimit from %u to %u" , window->dictLimit, |
586 | window->lowLimit); |
587 | window->dictLimit = window->lowLimit; |
588 | } |
589 | if (loadedDictEndPtr) |
590 | *loadedDictEndPtr = 0; |
591 | } |
592 | } |
593 | |
594 | /** |
595 | * ZSTD_window_update(): |
596 | * Updates the window by appending [src, src + srcSize) to the window. |
597 | * If it is not contiguous, the current prefix becomes the extDict, and we |
598 | * forget about the extDict. Handles overlap of the prefix and extDict. |
599 | * Returns non-zero if the segment is contiguous. |
600 | */ |
601 | MEM_STATIC U32 ZSTD_window_update(ZSTD_window_t* window, |
602 | void const* src, size_t srcSize) |
603 | { |
604 | BYTE const* const ip = (BYTE const*)src; |
605 | U32 contiguous = 1; |
606 | /* Check if blocks follow each other */ |
607 | if (src != window->nextSrc) { |
608 | /* not contiguous */ |
609 | size_t const distanceFromBase = (size_t)(window->nextSrc - window->base); |
610 | DEBUGLOG(5, "Non contiguous blocks, new segment starts at %u" , |
611 | window->dictLimit); |
612 | window->lowLimit = window->dictLimit; |
613 | assert(distanceFromBase == (size_t)(U32)distanceFromBase); /* should never overflow */ |
614 | window->dictLimit = (U32)distanceFromBase; |
615 | window->dictBase = window->base; |
616 | window->base = ip - distanceFromBase; |
617 | // ms->nextToUpdate = window->dictLimit; |
618 | if (window->dictLimit - window->lowLimit < HASH_READ_SIZE) window->lowLimit = window->dictLimit; /* too small extDict */ |
619 | contiguous = 0; |
620 | } |
621 | window->nextSrc = ip + srcSize; |
622 | /* if input and dictionary overlap : reduce dictionary (area presumed modified by input) */ |
623 | if ( (ip+srcSize > window->dictBase + window->lowLimit) |
624 | & (ip < window->dictBase + window->dictLimit)) { |
625 | ptrdiff_t const highInputIdx = (ip + srcSize) - window->dictBase; |
626 | U32 const lowLimitMax = (highInputIdx > (ptrdiff_t)window->dictLimit) ? window->dictLimit : (U32)highInputIdx; |
627 | window->lowLimit = lowLimitMax; |
628 | } |
629 | return contiguous; |
630 | } |
631 | |
632 | #if defined (__cplusplus) |
633 | } |
634 | #endif |
635 | |
636 | |
637 | /* ============================================================== |
638 | * Private declarations |
639 | * These prototypes shall only be called from within lib/compress |
640 | * ============================================================== */ |
641 | |
642 | /* ZSTD_getCParamsFromCCtxParams() : |
643 | * cParams are built depending on compressionLevel, src size hints, |
644 | * LDM and manually set compression parameters. |
645 | */ |
646 | ZSTD_compressionParameters ZSTD_getCParamsFromCCtxParams( |
647 | const ZSTD_CCtx_params* CCtxParams, U64 srcSizeHint, size_t dictSize); |
648 | |
649 | /*! ZSTD_initCStream_internal() : |
650 | * Private use only. Init streaming operation. |
651 | * expects params to be valid. |
652 | * must receive dict, or cdict, or none, but not both. |
653 | * @return : 0, or an error code */ |
654 | size_t ZSTD_initCStream_internal(ZSTD_CStream* zcs, |
655 | const void* dict, size_t dictSize, |
656 | const ZSTD_CDict* cdict, |
657 | ZSTD_CCtx_params params, unsigned long long pledgedSrcSize); |
658 | |
659 | /*! ZSTD_compressStream_generic() : |
660 | * Private use only. To be called from zstdmt_compress.c in single-thread mode. */ |
661 | size_t ZSTD_compressStream_generic(ZSTD_CStream* zcs, |
662 | ZSTD_outBuffer* output, |
663 | ZSTD_inBuffer* input, |
664 | ZSTD_EndDirective const flushMode); |
665 | |
666 | /*! ZSTD_getCParamsFromCDict() : |
667 | * as the name implies */ |
668 | ZSTD_compressionParameters ZSTD_getCParamsFromCDict(const ZSTD_CDict* cdict); |
669 | |
670 | /* ZSTD_compressBegin_advanced_internal() : |
671 | * Private use only. To be called from zstdmt_compress.c. */ |
672 | size_t ZSTD_compressBegin_advanced_internal(ZSTD_CCtx* cctx, |
673 | const void* dict, size_t dictSize, |
674 | ZSTD_dictContentType_e dictContentType, |
675 | const ZSTD_CDict* cdict, |
676 | ZSTD_CCtx_params params, |
677 | unsigned long long pledgedSrcSize); |
678 | |
679 | /* ZSTD_compress_advanced_internal() : |
680 | * Private use only. To be called from zstdmt_compress.c. */ |
681 | size_t ZSTD_compress_advanced_internal(ZSTD_CCtx* cctx, |
682 | void* dst, size_t dstCapacity, |
683 | const void* src, size_t srcSize, |
684 | const void* dict,size_t dictSize, |
685 | ZSTD_CCtx_params params); |
686 | |
687 | |
688 | /* ZSTD_writeLastEmptyBlock() : |
689 | * output an empty Block with end-of-frame mark to complete a frame |
690 | * @return : size of data written into `dst` (== ZSTD_blockHeaderSize (defined in zstd_internal.h)) |
691 | * or an error code if `dstCapcity` is too small (<ZSTD_blockHeaderSize) |
692 | */ |
693 | size_t ZSTD_writeLastEmptyBlock(void* dst, size_t dstCapacity); |
694 | |
695 | |
696 | /* ZSTD_referenceExternalSequences() : |
697 | * Must be called before starting a compression operation. |
698 | * seqs must parse a prefix of the source. |
699 | * This cannot be used when long range matching is enabled. |
700 | * Zstd will use these sequences, and pass the literals to a secondary block |
701 | * compressor. |
702 | * @return : An error code on failure. |
703 | * NOTE: seqs are not verified! Invalid sequences can cause out-of-bounds memory |
704 | * access and data corruption. |
705 | */ |
706 | size_t ZSTD_referenceExternalSequences(ZSTD_CCtx* cctx, rawSeq* seq, size_t nbSeq); |
707 | |
708 | |
709 | #endif /* ZSTD_COMPRESS_H */ |
710 | |