zstd_lazy.c (63555B)
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#include "zstd_compress_internal.h" 12#include "zstd_lazy.h" 13 14 15/*-************************************* 16* Binary Tree search 17***************************************/ 18 19static void 20ZSTD_updateDUBT(ZSTD_matchState_t* ms, 21 const BYTE* ip, const BYTE* iend, 22 U32 mls) 23{ 24 const ZSTD_compressionParameters* const cParams = &ms->cParams; 25 U32* const hashTable = ms->hashTable; 26 U32 const hashLog = cParams->hashLog; 27 28 U32* const bt = ms->chainTable; 29 U32 const btLog = cParams->chainLog - 1; 30 U32 const btMask = (1 << btLog) - 1; 31 32 const BYTE* const base = ms->window.base; 33 U32 const target = (U32)(ip - base); 34 U32 idx = ms->nextToUpdate; 35 36 if (idx != target) 37 DEBUGLOG(7, "ZSTD_updateDUBT, from %u to %u (dictLimit:%u)", 38 idx, target, ms->window.dictLimit); 39 assert(ip + 8 <= iend); /* condition for ZSTD_hashPtr */ 40 (void)iend; 41 42 assert(idx >= ms->window.dictLimit); /* condition for valid base+idx */ 43 for ( ; idx < target ; idx++) { 44 size_t const h = ZSTD_hashPtr(base + idx, hashLog, mls); /* assumption : ip + 8 <= iend */ 45 U32 const matchIndex = hashTable[h]; 46 47 U32* const nextCandidatePtr = bt + 2*(idx&btMask); 48 U32* const sortMarkPtr = nextCandidatePtr + 1; 49 50 DEBUGLOG(8, "ZSTD_updateDUBT: insert %u", idx); 51 hashTable[h] = idx; /* Update Hash Table */ 52 *nextCandidatePtr = matchIndex; /* update BT like a chain */ 53 *sortMarkPtr = ZSTD_DUBT_UNSORTED_MARK; 54 } 55 ms->nextToUpdate = target; 56} 57 58 59/* ZSTD_insertDUBT1() : 60 * sort one already inserted but unsorted position 61 * assumption : curr >= btlow == (curr - btmask) 62 * doesn't fail */ 63static void 64ZSTD_insertDUBT1(ZSTD_matchState_t* ms, 65 U32 curr, const BYTE* inputEnd, 66 U32 nbCompares, U32 btLow, 67 const ZSTD_dictMode_e dictMode) 68{ 69 const ZSTD_compressionParameters* const cParams = &ms->cParams; 70 U32* const bt = ms->chainTable; 71 U32 const btLog = cParams->chainLog - 1; 72 U32 const btMask = (1 << btLog) - 1; 73 size_t commonLengthSmaller=0, commonLengthLarger=0; 74 const BYTE* const base = ms->window.base; 75 const BYTE* const dictBase = ms->window.dictBase; 76 const U32 dictLimit = ms->window.dictLimit; 77 const BYTE* const ip = (curr>=dictLimit) ? base + curr : dictBase + curr; 78 const BYTE* const iend = (curr>=dictLimit) ? inputEnd : dictBase + dictLimit; 79 const BYTE* const dictEnd = dictBase + dictLimit; 80 const BYTE* const prefixStart = base + dictLimit; 81 const BYTE* match; 82 U32* smallerPtr = bt + 2*(curr&btMask); 83 U32* largerPtr = smallerPtr + 1; 84 U32 matchIndex = *smallerPtr; /* this candidate is unsorted : next sorted candidate is reached through *smallerPtr, while *largerPtr contains previous unsorted candidate (which is already saved and can be overwritten) */ 85 U32 dummy32; /* to be nullified at the end */ 86 U32 const windowValid = ms->window.lowLimit; 87 U32 const maxDistance = 1U << cParams->windowLog; 88 U32 const windowLow = (curr - windowValid > maxDistance) ? curr - maxDistance : windowValid; 89 90 91 DEBUGLOG(8, "ZSTD_insertDUBT1(%u) (dictLimit=%u, lowLimit=%u)", 92 curr, dictLimit, windowLow); 93 assert(curr >= btLow); 94 assert(ip < iend); /* condition for ZSTD_count */ 95 96 for (; nbCompares && (matchIndex > windowLow); --nbCompares) { 97 U32* const nextPtr = bt + 2*(matchIndex & btMask); 98 size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */ 99 assert(matchIndex < curr); 100 /* note : all candidates are now supposed sorted, 101 * but it's still possible to have nextPtr[1] == ZSTD_DUBT_UNSORTED_MARK 102 * when a real index has the same value as ZSTD_DUBT_UNSORTED_MARK */ 103 104 if ( (dictMode != ZSTD_extDict) 105 || (matchIndex+matchLength >= dictLimit) /* both in current segment*/ 106 || (curr < dictLimit) /* both in extDict */) { 107 const BYTE* const mBase = ( (dictMode != ZSTD_extDict) 108 || (matchIndex+matchLength >= dictLimit)) ? 109 base : dictBase; 110 assert( (matchIndex+matchLength >= dictLimit) /* might be wrong if extDict is incorrectly set to 0 */ 111 || (curr < dictLimit) ); 112 match = mBase + matchIndex; 113 matchLength += ZSTD_count(ip+matchLength, match+matchLength, iend); 114 } else { 115 match = dictBase + matchIndex; 116 matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iend, dictEnd, prefixStart); 117 if (matchIndex+matchLength >= dictLimit) 118 match = base + matchIndex; /* preparation for next read of match[matchLength] */ 119 } 120 121 DEBUGLOG(8, "ZSTD_insertDUBT1: comparing %u with %u : found %u common bytes ", 122 curr, matchIndex, (U32)matchLength); 123 124 if (ip+matchLength == iend) { /* equal : no way to know if inf or sup */ 125 break; /* drop , to guarantee consistency ; miss a bit of compression, but other solutions can corrupt tree */ 126 } 127 128 if (match[matchLength] < ip[matchLength]) { /* necessarily within buffer */ 129 /* match is smaller than current */ 130 *smallerPtr = matchIndex; /* update smaller idx */ 131 commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */ 132 if (matchIndex <= btLow) { smallerPtr=&dummy32; break; } /* beyond tree size, stop searching */ 133 DEBUGLOG(8, "ZSTD_insertDUBT1: %u (>btLow=%u) is smaller : next => %u", 134 matchIndex, btLow, nextPtr[1]); 135 smallerPtr = nextPtr+1; /* new "candidate" => larger than match, which was smaller than target */ 136 matchIndex = nextPtr[1]; /* new matchIndex, larger than previous and closer to current */ 137 } else { 138 /* match is larger than current */ 139 *largerPtr = matchIndex; 140 commonLengthLarger = matchLength; 141 if (matchIndex <= btLow) { largerPtr=&dummy32; break; } /* beyond tree size, stop searching */ 142 DEBUGLOG(8, "ZSTD_insertDUBT1: %u (>btLow=%u) is larger => %u", 143 matchIndex, btLow, nextPtr[0]); 144 largerPtr = nextPtr; 145 matchIndex = nextPtr[0]; 146 } } 147 148 *smallerPtr = *largerPtr = 0; 149} 150 151 152static size_t 153ZSTD_DUBT_findBetterDictMatch ( 154 ZSTD_matchState_t* ms, 155 const BYTE* const ip, const BYTE* const iend, 156 size_t* offsetPtr, 157 size_t bestLength, 158 U32 nbCompares, 159 U32 const mls, 160 const ZSTD_dictMode_e dictMode) 161{ 162 const ZSTD_matchState_t * const dms = ms->dictMatchState; 163 const ZSTD_compressionParameters* const dmsCParams = &dms->cParams; 164 const U32 * const dictHashTable = dms->hashTable; 165 U32 const hashLog = dmsCParams->hashLog; 166 size_t const h = ZSTD_hashPtr(ip, hashLog, mls); 167 U32 dictMatchIndex = dictHashTable[h]; 168 169 const BYTE* const base = ms->window.base; 170 const BYTE* const prefixStart = base + ms->window.dictLimit; 171 U32 const curr = (U32)(ip-base); 172 const BYTE* const dictBase = dms->window.base; 173 const BYTE* const dictEnd = dms->window.nextSrc; 174 U32 const dictHighLimit = (U32)(dms->window.nextSrc - dms->window.base); 175 U32 const dictLowLimit = dms->window.lowLimit; 176 U32 const dictIndexDelta = ms->window.lowLimit - dictHighLimit; 177 178 U32* const dictBt = dms->chainTable; 179 U32 const btLog = dmsCParams->chainLog - 1; 180 U32 const btMask = (1 << btLog) - 1; 181 U32 const btLow = (btMask >= dictHighLimit - dictLowLimit) ? dictLowLimit : dictHighLimit - btMask; 182 183 size_t commonLengthSmaller=0, commonLengthLarger=0; 184 185 (void)dictMode; 186 assert(dictMode == ZSTD_dictMatchState); 187 188 for (; nbCompares && (dictMatchIndex > dictLowLimit); --nbCompares) { 189 U32* const nextPtr = dictBt + 2*(dictMatchIndex & btMask); 190 size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */ 191 const BYTE* match = dictBase + dictMatchIndex; 192 matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iend, dictEnd, prefixStart); 193 if (dictMatchIndex+matchLength >= dictHighLimit) 194 match = base + dictMatchIndex + dictIndexDelta; /* to prepare for next usage of match[matchLength] */ 195 196 if (matchLength > bestLength) { 197 U32 matchIndex = dictMatchIndex + dictIndexDelta; 198 if ( (4*(int)(matchLength-bestLength)) > (int)(ZSTD_highbit32(curr-matchIndex+1) - ZSTD_highbit32((U32)offsetPtr[0]+1)) ) { 199 DEBUGLOG(9, "ZSTD_DUBT_findBetterDictMatch(%u) : found better match length %u -> %u and offsetCode %u -> %u (dictMatchIndex %u, matchIndex %u)", 200 curr, (U32)bestLength, (U32)matchLength, (U32)*offsetPtr, ZSTD_REP_MOVE + curr - matchIndex, dictMatchIndex, matchIndex); 201 bestLength = matchLength, *offsetPtr = ZSTD_REP_MOVE + curr - matchIndex; 202 } 203 if (ip+matchLength == iend) { /* reached end of input : ip[matchLength] is not valid, no way to know if it's larger or smaller than match */ 204 break; /* drop, to guarantee consistency (miss a little bit of compression) */ 205 } 206 } 207 208 if (match[matchLength] < ip[matchLength]) { 209 if (dictMatchIndex <= btLow) { break; } /* beyond tree size, stop the search */ 210 commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */ 211 dictMatchIndex = nextPtr[1]; /* new matchIndex larger than previous (closer to current) */ 212 } else { 213 /* match is larger than current */ 214 if (dictMatchIndex <= btLow) { break; } /* beyond tree size, stop the search */ 215 commonLengthLarger = matchLength; 216 dictMatchIndex = nextPtr[0]; 217 } 218 } 219 220 if (bestLength >= MINMATCH) { 221 U32 const mIndex = curr - ((U32)*offsetPtr - ZSTD_REP_MOVE); (void)mIndex; 222 DEBUGLOG(8, "ZSTD_DUBT_findBetterDictMatch(%u) : found match of length %u and offsetCode %u (pos %u)", 223 curr, (U32)bestLength, (U32)*offsetPtr, mIndex); 224 } 225 return bestLength; 226 227} 228 229 230static size_t 231ZSTD_DUBT_findBestMatch(ZSTD_matchState_t* ms, 232 const BYTE* const ip, const BYTE* const iend, 233 size_t* offsetPtr, 234 U32 const mls, 235 const ZSTD_dictMode_e dictMode) 236{ 237 const ZSTD_compressionParameters* const cParams = &ms->cParams; 238 U32* const hashTable = ms->hashTable; 239 U32 const hashLog = cParams->hashLog; 240 size_t const h = ZSTD_hashPtr(ip, hashLog, mls); 241 U32 matchIndex = hashTable[h]; 242 243 const BYTE* const base = ms->window.base; 244 U32 const curr = (U32)(ip-base); 245 U32 const windowLow = ZSTD_getLowestMatchIndex(ms, curr, cParams->windowLog); 246 247 U32* const bt = ms->chainTable; 248 U32 const btLog = cParams->chainLog - 1; 249 U32 const btMask = (1 << btLog) - 1; 250 U32 const btLow = (btMask >= curr) ? 0 : curr - btMask; 251 U32 const unsortLimit = MAX(btLow, windowLow); 252 253 U32* nextCandidate = bt + 2*(matchIndex&btMask); 254 U32* unsortedMark = bt + 2*(matchIndex&btMask) + 1; 255 U32 nbCompares = 1U << cParams->searchLog; 256 U32 nbCandidates = nbCompares; 257 U32 previousCandidate = 0; 258 259 DEBUGLOG(7, "ZSTD_DUBT_findBestMatch (%u) ", curr); 260 assert(ip <= iend-8); /* required for h calculation */ 261 assert(dictMode != ZSTD_dedicatedDictSearch); 262 263 /* reach end of unsorted candidates list */ 264 while ( (matchIndex > unsortLimit) 265 && (*unsortedMark == ZSTD_DUBT_UNSORTED_MARK) 266 && (nbCandidates > 1) ) { 267 DEBUGLOG(8, "ZSTD_DUBT_findBestMatch: candidate %u is unsorted", 268 matchIndex); 269 *unsortedMark = previousCandidate; /* the unsortedMark becomes a reversed chain, to move up back to original position */ 270 previousCandidate = matchIndex; 271 matchIndex = *nextCandidate; 272 nextCandidate = bt + 2*(matchIndex&btMask); 273 unsortedMark = bt + 2*(matchIndex&btMask) + 1; 274 nbCandidates --; 275 } 276 277 /* nullify last candidate if it's still unsorted 278 * simplification, detrimental to compression ratio, beneficial for speed */ 279 if ( (matchIndex > unsortLimit) 280 && (*unsortedMark==ZSTD_DUBT_UNSORTED_MARK) ) { 281 DEBUGLOG(7, "ZSTD_DUBT_findBestMatch: nullify last unsorted candidate %u", 282 matchIndex); 283 *nextCandidate = *unsortedMark = 0; 284 } 285 286 /* batch sort stacked candidates */ 287 matchIndex = previousCandidate; 288 while (matchIndex) { /* will end on matchIndex == 0 */ 289 U32* const nextCandidateIdxPtr = bt + 2*(matchIndex&btMask) + 1; 290 U32 const nextCandidateIdx = *nextCandidateIdxPtr; 291 ZSTD_insertDUBT1(ms, matchIndex, iend, 292 nbCandidates, unsortLimit, dictMode); 293 matchIndex = nextCandidateIdx; 294 nbCandidates++; 295 } 296 297 /* find longest match */ 298 { size_t commonLengthSmaller = 0, commonLengthLarger = 0; 299 const BYTE* const dictBase = ms->window.dictBase; 300 const U32 dictLimit = ms->window.dictLimit; 301 const BYTE* const dictEnd = dictBase + dictLimit; 302 const BYTE* const prefixStart = base + dictLimit; 303 U32* smallerPtr = bt + 2*(curr&btMask); 304 U32* largerPtr = bt + 2*(curr&btMask) + 1; 305 U32 matchEndIdx = curr + 8 + 1; 306 U32 dummy32; /* to be nullified at the end */ 307 size_t bestLength = 0; 308 309 matchIndex = hashTable[h]; 310 hashTable[h] = curr; /* Update Hash Table */ 311 312 for (; nbCompares && (matchIndex > windowLow); --nbCompares) { 313 U32* const nextPtr = bt + 2*(matchIndex & btMask); 314 size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */ 315 const BYTE* match; 316 317 if ((dictMode != ZSTD_extDict) || (matchIndex+matchLength >= dictLimit)) { 318 match = base + matchIndex; 319 matchLength += ZSTD_count(ip+matchLength, match+matchLength, iend); 320 } else { 321 match = dictBase + matchIndex; 322 matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iend, dictEnd, prefixStart); 323 if (matchIndex+matchLength >= dictLimit) 324 match = base + matchIndex; /* to prepare for next usage of match[matchLength] */ 325 } 326 327 if (matchLength > bestLength) { 328 if (matchLength > matchEndIdx - matchIndex) 329 matchEndIdx = matchIndex + (U32)matchLength; 330 if ( (4*(int)(matchLength-bestLength)) > (int)(ZSTD_highbit32(curr-matchIndex+1) - ZSTD_highbit32((U32)offsetPtr[0]+1)) ) 331 bestLength = matchLength, *offsetPtr = ZSTD_REP_MOVE + curr - matchIndex; 332 if (ip+matchLength == iend) { /* equal : no way to know if inf or sup */ 333 if (dictMode == ZSTD_dictMatchState) { 334 nbCompares = 0; /* in addition to avoiding checking any 335 * further in this loop, make sure we 336 * skip checking in the dictionary. */ 337 } 338 break; /* drop, to guarantee consistency (miss a little bit of compression) */ 339 } 340 } 341 342 if (match[matchLength] < ip[matchLength]) { 343 /* match is smaller than current */ 344 *smallerPtr = matchIndex; /* update smaller idx */ 345 commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */ 346 if (matchIndex <= btLow) { smallerPtr=&dummy32; break; } /* beyond tree size, stop the search */ 347 smallerPtr = nextPtr+1; /* new "smaller" => larger of match */ 348 matchIndex = nextPtr[1]; /* new matchIndex larger than previous (closer to current) */ 349 } else { 350 /* match is larger than current */ 351 *largerPtr = matchIndex; 352 commonLengthLarger = matchLength; 353 if (matchIndex <= btLow) { largerPtr=&dummy32; break; } /* beyond tree size, stop the search */ 354 largerPtr = nextPtr; 355 matchIndex = nextPtr[0]; 356 } } 357 358 *smallerPtr = *largerPtr = 0; 359 360 assert(nbCompares <= (1U << ZSTD_SEARCHLOG_MAX)); /* Check we haven't underflowed. */ 361 if (dictMode == ZSTD_dictMatchState && nbCompares) { 362 bestLength = ZSTD_DUBT_findBetterDictMatch( 363 ms, ip, iend, 364 offsetPtr, bestLength, nbCompares, 365 mls, dictMode); 366 } 367 368 assert(matchEndIdx > curr+8); /* ensure nextToUpdate is increased */ 369 ms->nextToUpdate = matchEndIdx - 8; /* skip repetitive patterns */ 370 if (bestLength >= MINMATCH) { 371 U32 const mIndex = curr - ((U32)*offsetPtr - ZSTD_REP_MOVE); (void)mIndex; 372 DEBUGLOG(8, "ZSTD_DUBT_findBestMatch(%u) : found match of length %u and offsetCode %u (pos %u)", 373 curr, (U32)bestLength, (U32)*offsetPtr, mIndex); 374 } 375 return bestLength; 376 } 377} 378 379 380/* ZSTD_BtFindBestMatch() : Tree updater, providing best match */ 381FORCE_INLINE_TEMPLATE size_t 382ZSTD_BtFindBestMatch( ZSTD_matchState_t* ms, 383 const BYTE* const ip, const BYTE* const iLimit, 384 size_t* offsetPtr, 385 const U32 mls /* template */, 386 const ZSTD_dictMode_e dictMode) 387{ 388 DEBUGLOG(7, "ZSTD_BtFindBestMatch"); 389 if (ip < ms->window.base + ms->nextToUpdate) return 0; /* skipped area */ 390 ZSTD_updateDUBT(ms, ip, iLimit, mls); 391 return ZSTD_DUBT_findBestMatch(ms, ip, iLimit, offsetPtr, mls, dictMode); 392} 393 394 395static size_t 396ZSTD_BtFindBestMatch_selectMLS ( ZSTD_matchState_t* ms, 397 const BYTE* ip, const BYTE* const iLimit, 398 size_t* offsetPtr) 399{ 400 switch(ms->cParams.minMatch) 401 { 402 default : /* includes case 3 */ 403 case 4 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 4, ZSTD_noDict); 404 case 5 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 5, ZSTD_noDict); 405 case 7 : 406 case 6 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 6, ZSTD_noDict); 407 } 408} 409 410 411static size_t ZSTD_BtFindBestMatch_dictMatchState_selectMLS ( 412 ZSTD_matchState_t* ms, 413 const BYTE* ip, const BYTE* const iLimit, 414 size_t* offsetPtr) 415{ 416 switch(ms->cParams.minMatch) 417 { 418 default : /* includes case 3 */ 419 case 4 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 4, ZSTD_dictMatchState); 420 case 5 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 5, ZSTD_dictMatchState); 421 case 7 : 422 case 6 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 6, ZSTD_dictMatchState); 423 } 424} 425 426 427static size_t ZSTD_BtFindBestMatch_extDict_selectMLS ( 428 ZSTD_matchState_t* ms, 429 const BYTE* ip, const BYTE* const iLimit, 430 size_t* offsetPtr) 431{ 432 switch(ms->cParams.minMatch) 433 { 434 default : /* includes case 3 */ 435 case 4 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 4, ZSTD_extDict); 436 case 5 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 5, ZSTD_extDict); 437 case 7 : 438 case 6 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 6, ZSTD_extDict); 439 } 440} 441 442 443 444/* ********************************* 445* Hash Chain 446***********************************/ 447#define NEXT_IN_CHAIN(d, mask) chainTable[(d) & (mask)] 448 449/* Update chains up to ip (excluded) 450 Assumption : always within prefix (i.e. not within extDict) */ 451FORCE_INLINE_TEMPLATE U32 ZSTD_insertAndFindFirstIndex_internal( 452 ZSTD_matchState_t* ms, 453 const ZSTD_compressionParameters* const cParams, 454 const BYTE* ip, U32 const mls) 455{ 456 U32* const hashTable = ms->hashTable; 457 const U32 hashLog = cParams->hashLog; 458 U32* const chainTable = ms->chainTable; 459 const U32 chainMask = (1 << cParams->chainLog) - 1; 460 const BYTE* const base = ms->window.base; 461 const U32 target = (U32)(ip - base); 462 U32 idx = ms->nextToUpdate; 463 464 while(idx < target) { /* catch up */ 465 size_t const h = ZSTD_hashPtr(base+idx, hashLog, mls); 466 NEXT_IN_CHAIN(idx, chainMask) = hashTable[h]; 467 hashTable[h] = idx; 468 idx++; 469 } 470 471 ms->nextToUpdate = target; 472 return hashTable[ZSTD_hashPtr(ip, hashLog, mls)]; 473} 474 475U32 ZSTD_insertAndFindFirstIndex(ZSTD_matchState_t* ms, const BYTE* ip) { 476 const ZSTD_compressionParameters* const cParams = &ms->cParams; 477 return ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, ms->cParams.minMatch); 478} 479 480void ZSTD_dedicatedDictSearch_lazy_loadDictionary(ZSTD_matchState_t* ms, const BYTE* const ip) 481{ 482 const BYTE* const base = ms->window.base; 483 U32 const target = (U32)(ip - base); 484 U32* const hashTable = ms->hashTable; 485 U32* const chainTable = ms->chainTable; 486 U32 const chainSize = 1 << ms->cParams.chainLog; 487 U32 idx = ms->nextToUpdate; 488 U32 const minChain = chainSize < target ? target - chainSize : idx; 489 U32 const bucketSize = 1 << ZSTD_LAZY_DDSS_BUCKET_LOG; 490 U32 const cacheSize = bucketSize - 1; 491 U32 const chainAttempts = (1 << ms->cParams.searchLog) - cacheSize; 492 U32 const chainLimit = chainAttempts > 255 ? 255 : chainAttempts; 493 494 /* We know the hashtable is oversized by a factor of `bucketSize`. 495 * We are going to temporarily pretend `bucketSize == 1`, keeping only a 496 * single entry. We will use the rest of the space to construct a temporary 497 * chaintable. 498 */ 499 U32 const hashLog = ms->cParams.hashLog - ZSTD_LAZY_DDSS_BUCKET_LOG; 500 U32* const tmpHashTable = hashTable; 501 U32* const tmpChainTable = hashTable + ((size_t)1 << hashLog); 502 U32 const tmpChainSize = ((1 << ZSTD_LAZY_DDSS_BUCKET_LOG) - 1) << hashLog; 503 U32 const tmpMinChain = tmpChainSize < target ? target - tmpChainSize : idx; 504 505 U32 hashIdx; 506 507 assert(ms->cParams.chainLog <= 24); 508 assert(ms->cParams.hashLog >= ms->cParams.chainLog); 509 assert(idx != 0); 510 assert(tmpMinChain <= minChain); 511 512 /* fill conventional hash table and conventional chain table */ 513 for ( ; idx < target; idx++) { 514 U32 const h = (U32)ZSTD_hashPtr(base + idx, hashLog, ms->cParams.minMatch); 515 if (idx >= tmpMinChain) { 516 tmpChainTable[idx - tmpMinChain] = hashTable[h]; 517 } 518 tmpHashTable[h] = idx; 519 } 520 521 /* sort chains into ddss chain table */ 522 { 523 U32 chainPos = 0; 524 for (hashIdx = 0; hashIdx < (1U << hashLog); hashIdx++) { 525 U32 count; 526 U32 countBeyondMinChain = 0; 527 U32 i = tmpHashTable[hashIdx]; 528 for (count = 0; i >= tmpMinChain && count < cacheSize; count++) { 529 /* skip through the chain to the first position that won't be 530 * in the hash cache bucket */ 531 if (i < minChain) { 532 countBeyondMinChain++; 533 } 534 i = tmpChainTable[i - tmpMinChain]; 535 } 536 if (count == cacheSize) { 537 for (count = 0; count < chainLimit;) { 538 if (i < minChain) { 539 if (!i || countBeyondMinChain++ > cacheSize) { 540 /* only allow pulling `cacheSize` number of entries 541 * into the cache or chainTable beyond `minChain`, 542 * to replace the entries pulled out of the 543 * chainTable into the cache. This lets us reach 544 * back further without increasing the total number 545 * of entries in the chainTable, guaranteeing the 546 * DDSS chain table will fit into the space 547 * allocated for the regular one. */ 548 break; 549 } 550 } 551 chainTable[chainPos++] = i; 552 count++; 553 if (i < tmpMinChain) { 554 break; 555 } 556 i = tmpChainTable[i - tmpMinChain]; 557 } 558 } else { 559 count = 0; 560 } 561 if (count) { 562 tmpHashTable[hashIdx] = ((chainPos - count) << 8) + count; 563 } else { 564 tmpHashTable[hashIdx] = 0; 565 } 566 } 567 assert(chainPos <= chainSize); /* I believe this is guaranteed... */ 568 } 569 570 /* move chain pointers into the last entry of each hash bucket */ 571 for (hashIdx = (1 << hashLog); hashIdx; ) { 572 U32 const bucketIdx = --hashIdx << ZSTD_LAZY_DDSS_BUCKET_LOG; 573 U32 const chainPackedPointer = tmpHashTable[hashIdx]; 574 U32 i; 575 for (i = 0; i < cacheSize; i++) { 576 hashTable[bucketIdx + i] = 0; 577 } 578 hashTable[bucketIdx + bucketSize - 1] = chainPackedPointer; 579 } 580 581 /* fill the buckets of the hash table */ 582 for (idx = ms->nextToUpdate; idx < target; idx++) { 583 U32 const h = (U32)ZSTD_hashPtr(base + idx, hashLog, ms->cParams.minMatch) 584 << ZSTD_LAZY_DDSS_BUCKET_LOG; 585 U32 i; 586 /* Shift hash cache down 1. */ 587 for (i = cacheSize - 1; i; i--) 588 hashTable[h + i] = hashTable[h + i - 1]; 589 hashTable[h] = idx; 590 } 591 592 ms->nextToUpdate = target; 593} 594 595 596/* inlining is important to hardwire a hot branch (template emulation) */ 597FORCE_INLINE_TEMPLATE 598size_t ZSTD_HcFindBestMatch_generic ( 599 ZSTD_matchState_t* ms, 600 const BYTE* const ip, const BYTE* const iLimit, 601 size_t* offsetPtr, 602 const U32 mls, const ZSTD_dictMode_e dictMode) 603{ 604 const ZSTD_compressionParameters* const cParams = &ms->cParams; 605 U32* const chainTable = ms->chainTable; 606 const U32 chainSize = (1 << cParams->chainLog); 607 const U32 chainMask = chainSize-1; 608 const BYTE* const base = ms->window.base; 609 const BYTE* const dictBase = ms->window.dictBase; 610 const U32 dictLimit = ms->window.dictLimit; 611 const BYTE* const prefixStart = base + dictLimit; 612 const BYTE* const dictEnd = dictBase + dictLimit; 613 const U32 curr = (U32)(ip-base); 614 const U32 maxDistance = 1U << cParams->windowLog; 615 const U32 lowestValid = ms->window.lowLimit; 616 const U32 withinMaxDistance = (curr - lowestValid > maxDistance) ? curr - maxDistance : lowestValid; 617 const U32 isDictionary = (ms->loadedDictEnd != 0); 618 const U32 lowLimit = isDictionary ? lowestValid : withinMaxDistance; 619 const U32 minChain = curr > chainSize ? curr - chainSize : 0; 620 U32 nbAttempts = 1U << cParams->searchLog; 621 size_t ml=4-1; 622 623 const ZSTD_matchState_t* const dms = ms->dictMatchState; 624 const U32 ddsHashLog = dictMode == ZSTD_dedicatedDictSearch 625 ? dms->cParams.hashLog - ZSTD_LAZY_DDSS_BUCKET_LOG : 0; 626 const size_t ddsIdx = dictMode == ZSTD_dedicatedDictSearch 627 ? ZSTD_hashPtr(ip, ddsHashLog, mls) << ZSTD_LAZY_DDSS_BUCKET_LOG : 0; 628 629 U32 matchIndex; 630 631 if (dictMode == ZSTD_dedicatedDictSearch) { 632 const U32* entry = &dms->hashTable[ddsIdx]; 633 PREFETCH_L1(entry); 634 } 635 636 /* HC4 match finder */ 637 matchIndex = ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, mls); 638 639 for ( ; (matchIndex>=lowLimit) & (nbAttempts>0) ; nbAttempts--) { 640 size_t currentMl=0; 641 if ((dictMode != ZSTD_extDict) || matchIndex >= dictLimit) { 642 const BYTE* const match = base + matchIndex; 643 assert(matchIndex >= dictLimit); /* ensures this is true if dictMode != ZSTD_extDict */ 644 if (match[ml] == ip[ml]) /* potentially better */ 645 currentMl = ZSTD_count(ip, match, iLimit); 646 } else { 647 const BYTE* const match = dictBase + matchIndex; 648 assert(match+4 <= dictEnd); 649 if (MEM_read32(match) == MEM_read32(ip)) /* assumption : matchIndex <= dictLimit-4 (by table construction) */ 650 currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, dictEnd, prefixStart) + 4; 651 } 652 653 /* save best solution */ 654 if (currentMl > ml) { 655 ml = currentMl; 656 *offsetPtr = curr - matchIndex + ZSTD_REP_MOVE; 657 if (ip+currentMl == iLimit) break; /* best possible, avoids read overflow on next attempt */ 658 } 659 660 if (matchIndex <= minChain) break; 661 matchIndex = NEXT_IN_CHAIN(matchIndex, chainMask); 662 } 663 664 assert(nbAttempts <= (1U << ZSTD_SEARCHLOG_MAX)); /* Check we haven't underflowed. */ 665 if (dictMode == ZSTD_dedicatedDictSearch) { 666 const U32 ddsLowestIndex = dms->window.dictLimit; 667 const BYTE* const ddsBase = dms->window.base; 668 const BYTE* const ddsEnd = dms->window.nextSrc; 669 const U32 ddsSize = (U32)(ddsEnd - ddsBase); 670 const U32 ddsIndexDelta = dictLimit - ddsSize; 671 const U32 bucketSize = (1 << ZSTD_LAZY_DDSS_BUCKET_LOG); 672 const U32 bucketLimit = nbAttempts < bucketSize - 1 ? nbAttempts : bucketSize - 1; 673 U32 ddsAttempt; 674 675 for (ddsAttempt = 0; ddsAttempt < bucketSize - 1; ddsAttempt++) { 676 PREFETCH_L1(ddsBase + dms->hashTable[ddsIdx + ddsAttempt]); 677 } 678 679 { 680 U32 const chainPackedPointer = dms->hashTable[ddsIdx + bucketSize - 1]; 681 U32 const chainIndex = chainPackedPointer >> 8; 682 683 PREFETCH_L1(&dms->chainTable[chainIndex]); 684 } 685 686 for (ddsAttempt = 0; ddsAttempt < bucketLimit; ddsAttempt++) { 687 size_t currentMl=0; 688 const BYTE* match; 689 matchIndex = dms->hashTable[ddsIdx + ddsAttempt]; 690 match = ddsBase + matchIndex; 691 692 if (!matchIndex) { 693 return ml; 694 } 695 696 /* guaranteed by table construction */ 697 (void)ddsLowestIndex; 698 assert(matchIndex >= ddsLowestIndex); 699 assert(match+4 <= ddsEnd); 700 if (MEM_read32(match) == MEM_read32(ip)) { 701 /* assumption : matchIndex <= dictLimit-4 (by table construction) */ 702 currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, ddsEnd, prefixStart) + 4; 703 } 704 705 /* save best solution */ 706 if (currentMl > ml) { 707 ml = currentMl; 708 *offsetPtr = curr - (matchIndex + ddsIndexDelta) + ZSTD_REP_MOVE; 709 if (ip+currentMl == iLimit) { 710 /* best possible, avoids read overflow on next attempt */ 711 return ml; 712 } 713 } 714 } 715 716 { 717 U32 const chainPackedPointer = dms->hashTable[ddsIdx + bucketSize - 1]; 718 U32 chainIndex = chainPackedPointer >> 8; 719 U32 const chainLength = chainPackedPointer & 0xFF; 720 U32 const chainAttempts = nbAttempts - ddsAttempt; 721 U32 const chainLimit = chainAttempts > chainLength ? chainLength : chainAttempts; 722 U32 chainAttempt; 723 724 for (chainAttempt = 0 ; chainAttempt < chainLimit; chainAttempt++) { 725 PREFETCH_L1(ddsBase + dms->chainTable[chainIndex + chainAttempt]); 726 } 727 728 for (chainAttempt = 0 ; chainAttempt < chainLimit; chainAttempt++, chainIndex++) { 729 size_t currentMl=0; 730 const BYTE* match; 731 matchIndex = dms->chainTable[chainIndex]; 732 match = ddsBase + matchIndex; 733 734 /* guaranteed by table construction */ 735 assert(matchIndex >= ddsLowestIndex); 736 assert(match+4 <= ddsEnd); 737 if (MEM_read32(match) == MEM_read32(ip)) { 738 /* assumption : matchIndex <= dictLimit-4 (by table construction) */ 739 currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, ddsEnd, prefixStart) + 4; 740 } 741 742 /* save best solution */ 743 if (currentMl > ml) { 744 ml = currentMl; 745 *offsetPtr = curr - (matchIndex + ddsIndexDelta) + ZSTD_REP_MOVE; 746 if (ip+currentMl == iLimit) break; /* best possible, avoids read overflow on next attempt */ 747 } 748 } 749 } 750 } else if (dictMode == ZSTD_dictMatchState) { 751 const U32* const dmsChainTable = dms->chainTable; 752 const U32 dmsChainSize = (1 << dms->cParams.chainLog); 753 const U32 dmsChainMask = dmsChainSize - 1; 754 const U32 dmsLowestIndex = dms->window.dictLimit; 755 const BYTE* const dmsBase = dms->window.base; 756 const BYTE* const dmsEnd = dms->window.nextSrc; 757 const U32 dmsSize = (U32)(dmsEnd - dmsBase); 758 const U32 dmsIndexDelta = dictLimit - dmsSize; 759 const U32 dmsMinChain = dmsSize > dmsChainSize ? dmsSize - dmsChainSize : 0; 760 761 matchIndex = dms->hashTable[ZSTD_hashPtr(ip, dms->cParams.hashLog, mls)]; 762 763 for ( ; (matchIndex>=dmsLowestIndex) & (nbAttempts>0) ; nbAttempts--) { 764 size_t currentMl=0; 765 const BYTE* const match = dmsBase + matchIndex; 766 assert(match+4 <= dmsEnd); 767 if (MEM_read32(match) == MEM_read32(ip)) /* assumption : matchIndex <= dictLimit-4 (by table construction) */ 768 currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, dmsEnd, prefixStart) + 4; 769 770 /* save best solution */ 771 if (currentMl > ml) { 772 ml = currentMl; 773 *offsetPtr = curr - (matchIndex + dmsIndexDelta) + ZSTD_REP_MOVE; 774 if (ip+currentMl == iLimit) break; /* best possible, avoids read overflow on next attempt */ 775 } 776 777 if (matchIndex <= dmsMinChain) break; 778 779 matchIndex = dmsChainTable[matchIndex & dmsChainMask]; 780 } 781 } 782 783 return ml; 784} 785 786 787FORCE_INLINE_TEMPLATE size_t ZSTD_HcFindBestMatch_selectMLS ( 788 ZSTD_matchState_t* ms, 789 const BYTE* ip, const BYTE* const iLimit, 790 size_t* offsetPtr) 791{ 792 switch(ms->cParams.minMatch) 793 { 794 default : /* includes case 3 */ 795 case 4 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 4, ZSTD_noDict); 796 case 5 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 5, ZSTD_noDict); 797 case 7 : 798 case 6 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 6, ZSTD_noDict); 799 } 800} 801 802 803static size_t ZSTD_HcFindBestMatch_dictMatchState_selectMLS ( 804 ZSTD_matchState_t* ms, 805 const BYTE* ip, const BYTE* const iLimit, 806 size_t* offsetPtr) 807{ 808 switch(ms->cParams.minMatch) 809 { 810 default : /* includes case 3 */ 811 case 4 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 4, ZSTD_dictMatchState); 812 case 5 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 5, ZSTD_dictMatchState); 813 case 7 : 814 case 6 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 6, ZSTD_dictMatchState); 815 } 816} 817 818 819static size_t ZSTD_HcFindBestMatch_dedicatedDictSearch_selectMLS ( 820 ZSTD_matchState_t* ms, 821 const BYTE* ip, const BYTE* const iLimit, 822 size_t* offsetPtr) 823{ 824 switch(ms->cParams.minMatch) 825 { 826 default : /* includes case 3 */ 827 case 4 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 4, ZSTD_dedicatedDictSearch); 828 case 5 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 5, ZSTD_dedicatedDictSearch); 829 case 7 : 830 case 6 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 6, ZSTD_dedicatedDictSearch); 831 } 832} 833 834 835FORCE_INLINE_TEMPLATE size_t ZSTD_HcFindBestMatch_extDict_selectMLS ( 836 ZSTD_matchState_t* ms, 837 const BYTE* ip, const BYTE* const iLimit, 838 size_t* offsetPtr) 839{ 840 switch(ms->cParams.minMatch) 841 { 842 default : /* includes case 3 */ 843 case 4 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 4, ZSTD_extDict); 844 case 5 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 5, ZSTD_extDict); 845 case 7 : 846 case 6 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 6, ZSTD_extDict); 847 } 848} 849 850 851/* ******************************* 852* Common parser - lazy strategy 853*********************************/ 854typedef enum { search_hashChain, search_binaryTree } searchMethod_e; 855 856FORCE_INLINE_TEMPLATE size_t 857ZSTD_compressBlock_lazy_generic( 858 ZSTD_matchState_t* ms, seqStore_t* seqStore, 859 U32 rep[ZSTD_REP_NUM], 860 const void* src, size_t srcSize, 861 const searchMethod_e searchMethod, const U32 depth, 862 ZSTD_dictMode_e const dictMode) 863{ 864 const BYTE* const istart = (const BYTE*)src; 865 const BYTE* ip = istart; 866 const BYTE* anchor = istart; 867 const BYTE* const iend = istart + srcSize; 868 const BYTE* const ilimit = iend - 8; 869 const BYTE* const base = ms->window.base; 870 const U32 prefixLowestIndex = ms->window.dictLimit; 871 const BYTE* const prefixLowest = base + prefixLowestIndex; 872 873 typedef size_t (*searchMax_f)( 874 ZSTD_matchState_t* ms, 875 const BYTE* ip, const BYTE* iLimit, size_t* offsetPtr); 876 877 /* 878 * This table is indexed first by the four ZSTD_dictMode_e values, and then 879 * by the two searchMethod_e values. NULLs are placed for configurations 880 * that should never occur (extDict modes go to the other implementation 881 * below and there is no DDSS for binary tree search yet). 882 */ 883 const searchMax_f searchFuncs[4][2] = { 884 { 885 ZSTD_HcFindBestMatch_selectMLS, 886 ZSTD_BtFindBestMatch_selectMLS 887 }, 888 { 889 NULL, 890 NULL 891 }, 892 { 893 ZSTD_HcFindBestMatch_dictMatchState_selectMLS, 894 ZSTD_BtFindBestMatch_dictMatchState_selectMLS 895 }, 896 { 897 ZSTD_HcFindBestMatch_dedicatedDictSearch_selectMLS, 898 NULL 899 } 900 }; 901 902 searchMax_f const searchMax = searchFuncs[dictMode][searchMethod == search_binaryTree]; 903 U32 offset_1 = rep[0], offset_2 = rep[1], savedOffset=0; 904 905 const int isDMS = dictMode == ZSTD_dictMatchState; 906 const int isDDS = dictMode == ZSTD_dedicatedDictSearch; 907 const int isDxS = isDMS || isDDS; 908 const ZSTD_matchState_t* const dms = ms->dictMatchState; 909 const U32 dictLowestIndex = isDxS ? dms->window.dictLimit : 0; 910 const BYTE* const dictBase = isDxS ? dms->window.base : NULL; 911 const BYTE* const dictLowest = isDxS ? dictBase + dictLowestIndex : NULL; 912 const BYTE* const dictEnd = isDxS ? dms->window.nextSrc : NULL; 913 const U32 dictIndexDelta = isDxS ? 914 prefixLowestIndex - (U32)(dictEnd - dictBase) : 915 0; 916 const U32 dictAndPrefixLength = (U32)((ip - prefixLowest) + (dictEnd - dictLowest)); 917 918 assert(searchMax != NULL); 919 920 DEBUGLOG(5, "ZSTD_compressBlock_lazy_generic (dictMode=%u)", (U32)dictMode); 921 922 /* init */ 923 ip += (dictAndPrefixLength == 0); 924 if (dictMode == ZSTD_noDict) { 925 U32 const curr = (U32)(ip - base); 926 U32 const windowLow = ZSTD_getLowestPrefixIndex(ms, curr, ms->cParams.windowLog); 927 U32 const maxRep = curr - windowLow; 928 if (offset_2 > maxRep) savedOffset = offset_2, offset_2 = 0; 929 if (offset_1 > maxRep) savedOffset = offset_1, offset_1 = 0; 930 } 931 if (isDxS) { 932 /* dictMatchState repCode checks don't currently handle repCode == 0 933 * disabling. */ 934 assert(offset_1 <= dictAndPrefixLength); 935 assert(offset_2 <= dictAndPrefixLength); 936 } 937 938 /* Match Loop */ 939#if defined(__x86_64__) 940 /* I've measured random a 5% speed loss on levels 5 & 6 (greedy) when the 941 * code alignment is perturbed. To fix the instability align the loop on 32-bytes. 942 */ 943 __asm__(".p2align 5"); 944#endif 945 while (ip < ilimit) { 946 size_t matchLength=0; 947 size_t offset=0; 948 const BYTE* start=ip+1; 949 950 /* check repCode */ 951 if (isDxS) { 952 const U32 repIndex = (U32)(ip - base) + 1 - offset_1; 953 const BYTE* repMatch = ((dictMode == ZSTD_dictMatchState || dictMode == ZSTD_dedicatedDictSearch) 954 && repIndex < prefixLowestIndex) ? 955 dictBase + (repIndex - dictIndexDelta) : 956 base + repIndex; 957 if (((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */) 958 && (MEM_read32(repMatch) == MEM_read32(ip+1)) ) { 959 const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend; 960 matchLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4; 961 if (depth==0) goto _storeSequence; 962 } 963 } 964 if ( dictMode == ZSTD_noDict 965 && ((offset_1 > 0) & (MEM_read32(ip+1-offset_1) == MEM_read32(ip+1)))) { 966 matchLength = ZSTD_count(ip+1+4, ip+1+4-offset_1, iend) + 4; 967 if (depth==0) goto _storeSequence; 968 } 969 970 /* first search (depth 0) */ 971 { size_t offsetFound = 999999999; 972 size_t const ml2 = searchMax(ms, ip, iend, &offsetFound); 973 if (ml2 > matchLength) 974 matchLength = ml2, start = ip, offset=offsetFound; 975 } 976 977 if (matchLength < 4) { 978 ip += ((ip-anchor) >> kSearchStrength) + 1; /* jump faster over incompressible sections */ 979 continue; 980 } 981 982 /* let's try to find a better solution */ 983 if (depth>=1) 984 while (ip<ilimit) { 985 ip ++; 986 if ( (dictMode == ZSTD_noDict) 987 && (offset) && ((offset_1>0) & (MEM_read32(ip) == MEM_read32(ip - offset_1)))) { 988 size_t const mlRep = ZSTD_count(ip+4, ip+4-offset_1, iend) + 4; 989 int const gain2 = (int)(mlRep * 3); 990 int const gain1 = (int)(matchLength*3 - ZSTD_highbit32((U32)offset+1) + 1); 991 if ((mlRep >= 4) && (gain2 > gain1)) 992 matchLength = mlRep, offset = 0, start = ip; 993 } 994 if (isDxS) { 995 const U32 repIndex = (U32)(ip - base) - offset_1; 996 const BYTE* repMatch = repIndex < prefixLowestIndex ? 997 dictBase + (repIndex - dictIndexDelta) : 998 base + repIndex; 999 if (((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */) 1000 && (MEM_read32(repMatch) == MEM_read32(ip)) ) { 1001 const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend; 1002 size_t const mlRep = ZSTD_count_2segments(ip+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4; 1003 int const gain2 = (int)(mlRep * 3); 1004 int const gain1 = (int)(matchLength*3 - ZSTD_highbit32((U32)offset+1) + 1); 1005 if ((mlRep >= 4) && (gain2 > gain1)) 1006 matchLength = mlRep, offset = 0, start = ip; 1007 } 1008 } 1009 { size_t offset2=999999999; 1010 size_t const ml2 = searchMax(ms, ip, iend, &offset2); 1011 int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)offset2+1)); /* raw approx */ 1012 int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 4); 1013 if ((ml2 >= 4) && (gain2 > gain1)) { 1014 matchLength = ml2, offset = offset2, start = ip; 1015 continue; /* search a better one */ 1016 } } 1017 1018 /* let's find an even better one */ 1019 if ((depth==2) && (ip<ilimit)) { 1020 ip ++; 1021 if ( (dictMode == ZSTD_noDict) 1022 && (offset) && ((offset_1>0) & (MEM_read32(ip) == MEM_read32(ip - offset_1)))) { 1023 size_t const mlRep = ZSTD_count(ip+4, ip+4-offset_1, iend) + 4; 1024 int const gain2 = (int)(mlRep * 4); 1025 int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 1); 1026 if ((mlRep >= 4) && (gain2 > gain1)) 1027 matchLength = mlRep, offset = 0, start = ip; 1028 } 1029 if (isDxS) { 1030 const U32 repIndex = (U32)(ip - base) - offset_1; 1031 const BYTE* repMatch = repIndex < prefixLowestIndex ? 1032 dictBase + (repIndex - dictIndexDelta) : 1033 base + repIndex; 1034 if (((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */) 1035 && (MEM_read32(repMatch) == MEM_read32(ip)) ) { 1036 const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend; 1037 size_t const mlRep = ZSTD_count_2segments(ip+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4; 1038 int const gain2 = (int)(mlRep * 4); 1039 int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 1); 1040 if ((mlRep >= 4) && (gain2 > gain1)) 1041 matchLength = mlRep, offset = 0, start = ip; 1042 } 1043 } 1044 { size_t offset2=999999999; 1045 size_t const ml2 = searchMax(ms, ip, iend, &offset2); 1046 int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)offset2+1)); /* raw approx */ 1047 int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 7); 1048 if ((ml2 >= 4) && (gain2 > gain1)) { 1049 matchLength = ml2, offset = offset2, start = ip; 1050 continue; 1051 } } } 1052 break; /* nothing found : store previous solution */ 1053 } 1054 1055 /* NOTE: 1056 * start[-offset+ZSTD_REP_MOVE-1] is undefined behavior. 1057 * (-offset+ZSTD_REP_MOVE-1) is unsigned, and is added to start, which 1058 * overflows the pointer, which is undefined behavior. 1059 */ 1060 /* catch up */ 1061 if (offset) { 1062 if (dictMode == ZSTD_noDict) { 1063 while ( ((start > anchor) & (start - (offset-ZSTD_REP_MOVE) > prefixLowest)) 1064 && (start[-1] == (start-(offset-ZSTD_REP_MOVE))[-1]) ) /* only search for offset within prefix */ 1065 { start--; matchLength++; } 1066 } 1067 if (isDxS) { 1068 U32 const matchIndex = (U32)((start-base) - (offset - ZSTD_REP_MOVE)); 1069 const BYTE* match = (matchIndex < prefixLowestIndex) ? dictBase + matchIndex - dictIndexDelta : base + matchIndex; 1070 const BYTE* const mStart = (matchIndex < prefixLowestIndex) ? dictLowest : prefixLowest; 1071 while ((start>anchor) && (match>mStart) && (start[-1] == match[-1])) { start--; match--; matchLength++; } /* catch up */ 1072 } 1073 offset_2 = offset_1; offset_1 = (U32)(offset - ZSTD_REP_MOVE); 1074 } 1075 /* store sequence */ 1076_storeSequence: 1077 { size_t const litLength = start - anchor; 1078 ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offset, matchLength-MINMATCH); 1079 anchor = ip = start + matchLength; 1080 } 1081 1082 /* check immediate repcode */ 1083 if (isDxS) { 1084 while (ip <= ilimit) { 1085 U32 const current2 = (U32)(ip-base); 1086 U32 const repIndex = current2 - offset_2; 1087 const BYTE* repMatch = repIndex < prefixLowestIndex ? 1088 dictBase - dictIndexDelta + repIndex : 1089 base + repIndex; 1090 if ( ((U32)((prefixLowestIndex-1) - (U32)repIndex) >= 3 /* intentional overflow */) 1091 && (MEM_read32(repMatch) == MEM_read32(ip)) ) { 1092 const BYTE* const repEnd2 = repIndex < prefixLowestIndex ? dictEnd : iend; 1093 matchLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd2, prefixLowest) + 4; 1094 offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap offset_2 <=> offset_1 */ 1095 ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH); 1096 ip += matchLength; 1097 anchor = ip; 1098 continue; 1099 } 1100 break; 1101 } 1102 } 1103 1104 if (dictMode == ZSTD_noDict) { 1105 while ( ((ip <= ilimit) & (offset_2>0)) 1106 && (MEM_read32(ip) == MEM_read32(ip - offset_2)) ) { 1107 /* store sequence */ 1108 matchLength = ZSTD_count(ip+4, ip+4-offset_2, iend) + 4; 1109 offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap repcodes */ 1110 ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH); 1111 ip += matchLength; 1112 anchor = ip; 1113 continue; /* faster when present ... (?) */ 1114 } } } 1115 1116 /* Save reps for next block */ 1117 rep[0] = offset_1 ? offset_1 : savedOffset; 1118 rep[1] = offset_2 ? offset_2 : savedOffset; 1119 1120 /* Return the last literals size */ 1121 return (size_t)(iend - anchor); 1122} 1123 1124 1125size_t ZSTD_compressBlock_btlazy2( 1126 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1127 void const* src, size_t srcSize) 1128{ 1129 return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2, ZSTD_noDict); 1130} 1131 1132size_t ZSTD_compressBlock_lazy2( 1133 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1134 void const* src, size_t srcSize) 1135{ 1136 return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_noDict); 1137} 1138 1139size_t ZSTD_compressBlock_lazy( 1140 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1141 void const* src, size_t srcSize) 1142{ 1143 return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_noDict); 1144} 1145 1146size_t ZSTD_compressBlock_greedy( 1147 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1148 void const* src, size_t srcSize) 1149{ 1150 return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0, ZSTD_noDict); 1151} 1152 1153size_t ZSTD_compressBlock_btlazy2_dictMatchState( 1154 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1155 void const* src, size_t srcSize) 1156{ 1157 return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2, ZSTD_dictMatchState); 1158} 1159 1160size_t ZSTD_compressBlock_lazy2_dictMatchState( 1161 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1162 void const* src, size_t srcSize) 1163{ 1164 return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_dictMatchState); 1165} 1166 1167size_t ZSTD_compressBlock_lazy_dictMatchState( 1168 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1169 void const* src, size_t srcSize) 1170{ 1171 return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_dictMatchState); 1172} 1173 1174size_t ZSTD_compressBlock_greedy_dictMatchState( 1175 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1176 void const* src, size_t srcSize) 1177{ 1178 return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0, ZSTD_dictMatchState); 1179} 1180 1181 1182size_t ZSTD_compressBlock_lazy2_dedicatedDictSearch( 1183 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1184 void const* src, size_t srcSize) 1185{ 1186 return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_dedicatedDictSearch); 1187} 1188 1189size_t ZSTD_compressBlock_lazy_dedicatedDictSearch( 1190 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1191 void const* src, size_t srcSize) 1192{ 1193 return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_dedicatedDictSearch); 1194} 1195 1196size_t ZSTD_compressBlock_greedy_dedicatedDictSearch( 1197 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1198 void const* src, size_t srcSize) 1199{ 1200 return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0, ZSTD_dedicatedDictSearch); 1201} 1202 1203 1204FORCE_INLINE_TEMPLATE 1205size_t ZSTD_compressBlock_lazy_extDict_generic( 1206 ZSTD_matchState_t* ms, seqStore_t* seqStore, 1207 U32 rep[ZSTD_REP_NUM], 1208 const void* src, size_t srcSize, 1209 const searchMethod_e searchMethod, const U32 depth) 1210{ 1211 const BYTE* const istart = (const BYTE*)src; 1212 const BYTE* ip = istart; 1213 const BYTE* anchor = istart; 1214 const BYTE* const iend = istart + srcSize; 1215 const BYTE* const ilimit = iend - 8; 1216 const BYTE* const base = ms->window.base; 1217 const U32 dictLimit = ms->window.dictLimit; 1218 const BYTE* const prefixStart = base + dictLimit; 1219 const BYTE* const dictBase = ms->window.dictBase; 1220 const BYTE* const dictEnd = dictBase + dictLimit; 1221 const BYTE* const dictStart = dictBase + ms->window.lowLimit; 1222 const U32 windowLog = ms->cParams.windowLog; 1223 1224 typedef size_t (*searchMax_f)( 1225 ZSTD_matchState_t* ms, 1226 const BYTE* ip, const BYTE* iLimit, size_t* offsetPtr); 1227 searchMax_f searchMax = searchMethod==search_binaryTree ? ZSTD_BtFindBestMatch_extDict_selectMLS : ZSTD_HcFindBestMatch_extDict_selectMLS; 1228 1229 U32 offset_1 = rep[0], offset_2 = rep[1]; 1230 1231 DEBUGLOG(5, "ZSTD_compressBlock_lazy_extDict_generic"); 1232 1233 /* init */ 1234 ip += (ip == prefixStart); 1235 1236 /* Match Loop */ 1237#if defined(__x86_64__) 1238 /* I've measured random a 5% speed loss on levels 5 & 6 (greedy) when the 1239 * code alignment is perturbed. To fix the instability align the loop on 32-bytes. 1240 */ 1241 __asm__(".p2align 5"); 1242#endif 1243 while (ip < ilimit) { 1244 size_t matchLength=0; 1245 size_t offset=0; 1246 const BYTE* start=ip+1; 1247 U32 curr = (U32)(ip-base); 1248 1249 /* check repCode */ 1250 { const U32 windowLow = ZSTD_getLowestMatchIndex(ms, curr+1, windowLog); 1251 const U32 repIndex = (U32)(curr+1 - offset_1); 1252 const BYTE* const repBase = repIndex < dictLimit ? dictBase : base; 1253 const BYTE* const repMatch = repBase + repIndex; 1254 if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > windowLow)) /* intentional overflow */ 1255 if (MEM_read32(ip+1) == MEM_read32(repMatch)) { 1256 /* repcode detected we should take it */ 1257 const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend; 1258 matchLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repEnd, prefixStart) + 4; 1259 if (depth==0) goto _storeSequence; 1260 } } 1261 1262 /* first search (depth 0) */ 1263 { size_t offsetFound = 999999999; 1264 size_t const ml2 = searchMax(ms, ip, iend, &offsetFound); 1265 if (ml2 > matchLength) 1266 matchLength = ml2, start = ip, offset=offsetFound; 1267 } 1268 1269 if (matchLength < 4) { 1270 ip += ((ip-anchor) >> kSearchStrength) + 1; /* jump faster over incompressible sections */ 1271 continue; 1272 } 1273 1274 /* let's try to find a better solution */ 1275 if (depth>=1) 1276 while (ip<ilimit) { 1277 ip ++; 1278 curr++; 1279 /* check repCode */ 1280 if (offset) { 1281 const U32 windowLow = ZSTD_getLowestMatchIndex(ms, curr, windowLog); 1282 const U32 repIndex = (U32)(curr - offset_1); 1283 const BYTE* const repBase = repIndex < dictLimit ? dictBase : base; 1284 const BYTE* const repMatch = repBase + repIndex; 1285 if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > windowLow)) /* intentional overflow */ 1286 if (MEM_read32(ip) == MEM_read32(repMatch)) { 1287 /* repcode detected */ 1288 const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend; 1289 size_t const repLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4; 1290 int const gain2 = (int)(repLength * 3); 1291 int const gain1 = (int)(matchLength*3 - ZSTD_highbit32((U32)offset+1) + 1); 1292 if ((repLength >= 4) && (gain2 > gain1)) 1293 matchLength = repLength, offset = 0, start = ip; 1294 } } 1295 1296 /* search match, depth 1 */ 1297 { size_t offset2=999999999; 1298 size_t const ml2 = searchMax(ms, ip, iend, &offset2); 1299 int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)offset2+1)); /* raw approx */ 1300 int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 4); 1301 if ((ml2 >= 4) && (gain2 > gain1)) { 1302 matchLength = ml2, offset = offset2, start = ip; 1303 continue; /* search a better one */ 1304 } } 1305 1306 /* let's find an even better one */ 1307 if ((depth==2) && (ip<ilimit)) { 1308 ip ++; 1309 curr++; 1310 /* check repCode */ 1311 if (offset) { 1312 const U32 windowLow = ZSTD_getLowestMatchIndex(ms, curr, windowLog); 1313 const U32 repIndex = (U32)(curr - offset_1); 1314 const BYTE* const repBase = repIndex < dictLimit ? dictBase : base; 1315 const BYTE* const repMatch = repBase + repIndex; 1316 if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > windowLow)) /* intentional overflow */ 1317 if (MEM_read32(ip) == MEM_read32(repMatch)) { 1318 /* repcode detected */ 1319 const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend; 1320 size_t const repLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4; 1321 int const gain2 = (int)(repLength * 4); 1322 int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 1); 1323 if ((repLength >= 4) && (gain2 > gain1)) 1324 matchLength = repLength, offset = 0, start = ip; 1325 } } 1326 1327 /* search match, depth 2 */ 1328 { size_t offset2=999999999; 1329 size_t const ml2 = searchMax(ms, ip, iend, &offset2); 1330 int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)offset2+1)); /* raw approx */ 1331 int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 7); 1332 if ((ml2 >= 4) && (gain2 > gain1)) { 1333 matchLength = ml2, offset = offset2, start = ip; 1334 continue; 1335 } } } 1336 break; /* nothing found : store previous solution */ 1337 } 1338 1339 /* catch up */ 1340 if (offset) { 1341 U32 const matchIndex = (U32)((start-base) - (offset - ZSTD_REP_MOVE)); 1342 const BYTE* match = (matchIndex < dictLimit) ? dictBase + matchIndex : base + matchIndex; 1343 const BYTE* const mStart = (matchIndex < dictLimit) ? dictStart : prefixStart; 1344 while ((start>anchor) && (match>mStart) && (start[-1] == match[-1])) { start--; match--; matchLength++; } /* catch up */ 1345 offset_2 = offset_1; offset_1 = (U32)(offset - ZSTD_REP_MOVE); 1346 } 1347 1348 /* store sequence */ 1349_storeSequence: 1350 { size_t const litLength = start - anchor; 1351 ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offset, matchLength-MINMATCH); 1352 anchor = ip = start + matchLength; 1353 } 1354 1355 /* check immediate repcode */ 1356 while (ip <= ilimit) { 1357 const U32 repCurrent = (U32)(ip-base); 1358 const U32 windowLow = ZSTD_getLowestMatchIndex(ms, repCurrent, windowLog); 1359 const U32 repIndex = repCurrent - offset_2; 1360 const BYTE* const repBase = repIndex < dictLimit ? dictBase : base; 1361 const BYTE* const repMatch = repBase + repIndex; 1362 if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > windowLow)) /* intentional overflow */ 1363 if (MEM_read32(ip) == MEM_read32(repMatch)) { 1364 /* repcode detected we should take it */ 1365 const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend; 1366 matchLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4; 1367 offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap offset history */ 1368 ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH); 1369 ip += matchLength; 1370 anchor = ip; 1371 continue; /* faster when present ... (?) */ 1372 } 1373 break; 1374 } } 1375 1376 /* Save reps for next block */ 1377 rep[0] = offset_1; 1378 rep[1] = offset_2; 1379 1380 /* Return the last literals size */ 1381 return (size_t)(iend - anchor); 1382} 1383 1384 1385size_t ZSTD_compressBlock_greedy_extDict( 1386 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1387 void const* src, size_t srcSize) 1388{ 1389 return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0); 1390} 1391 1392size_t ZSTD_compressBlock_lazy_extDict( 1393 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1394 void const* src, size_t srcSize) 1395 1396{ 1397 return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1); 1398} 1399 1400size_t ZSTD_compressBlock_lazy2_extDict( 1401 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1402 void const* src, size_t srcSize) 1403 1404{ 1405 return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2); 1406} 1407 1408size_t ZSTD_compressBlock_btlazy2_extDict( 1409 ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM], 1410 void const* src, size_t srcSize) 1411 1412{ 1413 return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2); 1414}