xref: /freebsd/sys/contrib/zstd/lib/compress/zstd_lazy.c (revision 5ca8e32633c4ffbbcd6762e5888b6a4ba0708c6c)
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 
19 static void
20 ZSTD_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 */
63 static void
64 ZSTD_insertDUBT1(const 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 
152 static size_t
153 ZSTD_DUBT_findBetterDictMatch (
154         const 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, STORE_OFFSET(curr - matchIndex), dictMatchIndex, matchIndex);
201                 bestLength = matchLength, *offsetPtr = STORE_OFFSET(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)STORED_OFFSET(*offsetPtr); (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 
230 static size_t
231 ZSTD_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 = STORE_OFFSET(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)STORED_OFFSET(*offsetPtr); (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 */
381 FORCE_INLINE_TEMPLATE size_t
382 ZSTD_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 /***********************************
395 * Dedicated dict search
396 ***********************************/
397 
398 void ZSTD_dedicatedDictSearch_lazy_loadDictionary(ZSTD_matchState_t* ms, const BYTE* const ip)
399 {
400     const BYTE* const base = ms->window.base;
401     U32 const target = (U32)(ip - base);
402     U32* const hashTable = ms->hashTable;
403     U32* const chainTable = ms->chainTable;
404     U32 const chainSize = 1 << ms->cParams.chainLog;
405     U32 idx = ms->nextToUpdate;
406     U32 const minChain = chainSize < target - idx ? target - chainSize : idx;
407     U32 const bucketSize = 1 << ZSTD_LAZY_DDSS_BUCKET_LOG;
408     U32 const cacheSize = bucketSize - 1;
409     U32 const chainAttempts = (1 << ms->cParams.searchLog) - cacheSize;
410     U32 const chainLimit = chainAttempts > 255 ? 255 : chainAttempts;
411 
412     /* We know the hashtable is oversized by a factor of `bucketSize`.
413      * We are going to temporarily pretend `bucketSize == 1`, keeping only a
414      * single entry. We will use the rest of the space to construct a temporary
415      * chaintable.
416      */
417     U32 const hashLog = ms->cParams.hashLog - ZSTD_LAZY_DDSS_BUCKET_LOG;
418     U32* const tmpHashTable = hashTable;
419     U32* const tmpChainTable = hashTable + ((size_t)1 << hashLog);
420     U32 const tmpChainSize = (U32)((1 << ZSTD_LAZY_DDSS_BUCKET_LOG) - 1) << hashLog;
421     U32 const tmpMinChain = tmpChainSize < target ? target - tmpChainSize : idx;
422     U32 hashIdx;
423 
424     assert(ms->cParams.chainLog <= 24);
425     assert(ms->cParams.hashLog > ms->cParams.chainLog);
426     assert(idx != 0);
427     assert(tmpMinChain <= minChain);
428 
429     /* fill conventional hash table and conventional chain table */
430     for ( ; idx < target; idx++) {
431         U32 const h = (U32)ZSTD_hashPtr(base + idx, hashLog, ms->cParams.minMatch);
432         if (idx >= tmpMinChain) {
433             tmpChainTable[idx - tmpMinChain] = hashTable[h];
434         }
435         tmpHashTable[h] = idx;
436     }
437 
438     /* sort chains into ddss chain table */
439     {
440         U32 chainPos = 0;
441         for (hashIdx = 0; hashIdx < (1U << hashLog); hashIdx++) {
442             U32 count;
443             U32 countBeyondMinChain = 0;
444             U32 i = tmpHashTable[hashIdx];
445             for (count = 0; i >= tmpMinChain && count < cacheSize; count++) {
446                 /* skip through the chain to the first position that won't be
447                  * in the hash cache bucket */
448                 if (i < minChain) {
449                     countBeyondMinChain++;
450                 }
451                 i = tmpChainTable[i - tmpMinChain];
452             }
453             if (count == cacheSize) {
454                 for (count = 0; count < chainLimit;) {
455                     if (i < minChain) {
456                         if (!i || ++countBeyondMinChain > cacheSize) {
457                             /* only allow pulling `cacheSize` number of entries
458                              * into the cache or chainTable beyond `minChain`,
459                              * to replace the entries pulled out of the
460                              * chainTable into the cache. This lets us reach
461                              * back further without increasing the total number
462                              * of entries in the chainTable, guaranteeing the
463                              * DDSS chain table will fit into the space
464                              * allocated for the regular one. */
465                             break;
466                         }
467                     }
468                     chainTable[chainPos++] = i;
469                     count++;
470                     if (i < tmpMinChain) {
471                         break;
472                     }
473                     i = tmpChainTable[i - tmpMinChain];
474                 }
475             } else {
476                 count = 0;
477             }
478             if (count) {
479                 tmpHashTable[hashIdx] = ((chainPos - count) << 8) + count;
480             } else {
481                 tmpHashTable[hashIdx] = 0;
482             }
483         }
484         assert(chainPos <= chainSize); /* I believe this is guaranteed... */
485     }
486 
487     /* move chain pointers into the last entry of each hash bucket */
488     for (hashIdx = (1 << hashLog); hashIdx; ) {
489         U32 const bucketIdx = --hashIdx << ZSTD_LAZY_DDSS_BUCKET_LOG;
490         U32 const chainPackedPointer = tmpHashTable[hashIdx];
491         U32 i;
492         for (i = 0; i < cacheSize; i++) {
493             hashTable[bucketIdx + i] = 0;
494         }
495         hashTable[bucketIdx + bucketSize - 1] = chainPackedPointer;
496     }
497 
498     /* fill the buckets of the hash table */
499     for (idx = ms->nextToUpdate; idx < target; idx++) {
500         U32 const h = (U32)ZSTD_hashPtr(base + idx, hashLog, ms->cParams.minMatch)
501                    << ZSTD_LAZY_DDSS_BUCKET_LOG;
502         U32 i;
503         /* Shift hash cache down 1. */
504         for (i = cacheSize - 1; i; i--)
505             hashTable[h + i] = hashTable[h + i - 1];
506         hashTable[h] = idx;
507     }
508 
509     ms->nextToUpdate = target;
510 }
511 
512 /* Returns the longest match length found in the dedicated dict search structure.
513  * If none are longer than the argument ml, then ml will be returned.
514  */
515 FORCE_INLINE_TEMPLATE
516 size_t ZSTD_dedicatedDictSearch_lazy_search(size_t* offsetPtr, size_t ml, U32 nbAttempts,
517                                             const ZSTD_matchState_t* const dms,
518                                             const BYTE* const ip, const BYTE* const iLimit,
519                                             const BYTE* const prefixStart, const U32 curr,
520                                             const U32 dictLimit, const size_t ddsIdx) {
521     const U32 ddsLowestIndex  = dms->window.dictLimit;
522     const BYTE* const ddsBase = dms->window.base;
523     const BYTE* const ddsEnd  = dms->window.nextSrc;
524     const U32 ddsSize         = (U32)(ddsEnd - ddsBase);
525     const U32 ddsIndexDelta   = dictLimit - ddsSize;
526     const U32 bucketSize      = (1 << ZSTD_LAZY_DDSS_BUCKET_LOG);
527     const U32 bucketLimit     = nbAttempts < bucketSize - 1 ? nbAttempts : bucketSize - 1;
528     U32 ddsAttempt;
529     U32 matchIndex;
530 
531     for (ddsAttempt = 0; ddsAttempt < bucketSize - 1; ddsAttempt++) {
532         PREFETCH_L1(ddsBase + dms->hashTable[ddsIdx + ddsAttempt]);
533     }
534 
535     {
536         U32 const chainPackedPointer = dms->hashTable[ddsIdx + bucketSize - 1];
537         U32 const chainIndex = chainPackedPointer >> 8;
538 
539         PREFETCH_L1(&dms->chainTable[chainIndex]);
540     }
541 
542     for (ddsAttempt = 0; ddsAttempt < bucketLimit; ddsAttempt++) {
543         size_t currentMl=0;
544         const BYTE* match;
545         matchIndex = dms->hashTable[ddsIdx + ddsAttempt];
546         match = ddsBase + matchIndex;
547 
548         if (!matchIndex) {
549             return ml;
550         }
551 
552         /* guaranteed by table construction */
553         (void)ddsLowestIndex;
554         assert(matchIndex >= ddsLowestIndex);
555         assert(match+4 <= ddsEnd);
556         if (MEM_read32(match) == MEM_read32(ip)) {
557             /* assumption : matchIndex <= dictLimit-4 (by table construction) */
558             currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, ddsEnd, prefixStart) + 4;
559         }
560 
561         /* save best solution */
562         if (currentMl > ml) {
563             ml = currentMl;
564             *offsetPtr = STORE_OFFSET(curr - (matchIndex + ddsIndexDelta));
565             if (ip+currentMl == iLimit) {
566                 /* best possible, avoids read overflow on next attempt */
567                 return ml;
568             }
569         }
570     }
571 
572     {
573         U32 const chainPackedPointer = dms->hashTable[ddsIdx + bucketSize - 1];
574         U32 chainIndex = chainPackedPointer >> 8;
575         U32 const chainLength = chainPackedPointer & 0xFF;
576         U32 const chainAttempts = nbAttempts - ddsAttempt;
577         U32 const chainLimit = chainAttempts > chainLength ? chainLength : chainAttempts;
578         U32 chainAttempt;
579 
580         for (chainAttempt = 0 ; chainAttempt < chainLimit; chainAttempt++) {
581             PREFETCH_L1(ddsBase + dms->chainTable[chainIndex + chainAttempt]);
582         }
583 
584         for (chainAttempt = 0 ; chainAttempt < chainLimit; chainAttempt++, chainIndex++) {
585             size_t currentMl=0;
586             const BYTE* match;
587             matchIndex = dms->chainTable[chainIndex];
588             match = ddsBase + matchIndex;
589 
590             /* guaranteed by table construction */
591             assert(matchIndex >= ddsLowestIndex);
592             assert(match+4 <= ddsEnd);
593             if (MEM_read32(match) == MEM_read32(ip)) {
594                 /* assumption : matchIndex <= dictLimit-4 (by table construction) */
595                 currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, ddsEnd, prefixStart) + 4;
596             }
597 
598             /* save best solution */
599             if (currentMl > ml) {
600                 ml = currentMl;
601                 *offsetPtr = STORE_OFFSET(curr - (matchIndex + ddsIndexDelta));
602                 if (ip+currentMl == iLimit) break; /* best possible, avoids read overflow on next attempt */
603             }
604         }
605     }
606     return ml;
607 }
608 
609 
610 /* *********************************
611 *  Hash Chain
612 ***********************************/
613 #define NEXT_IN_CHAIN(d, mask)   chainTable[(d) & (mask)]
614 
615 /* Update chains up to ip (excluded)
616    Assumption : always within prefix (i.e. not within extDict) */
617 FORCE_INLINE_TEMPLATE U32 ZSTD_insertAndFindFirstIndex_internal(
618                         ZSTD_matchState_t* ms,
619                         const ZSTD_compressionParameters* const cParams,
620                         const BYTE* ip, U32 const mls)
621 {
622     U32* const hashTable  = ms->hashTable;
623     const U32 hashLog = cParams->hashLog;
624     U32* const chainTable = ms->chainTable;
625     const U32 chainMask = (1 << cParams->chainLog) - 1;
626     const BYTE* const base = ms->window.base;
627     const U32 target = (U32)(ip - base);
628     U32 idx = ms->nextToUpdate;
629 
630     while(idx < target) { /* catch up */
631         size_t const h = ZSTD_hashPtr(base+idx, hashLog, mls);
632         NEXT_IN_CHAIN(idx, chainMask) = hashTable[h];
633         hashTable[h] = idx;
634         idx++;
635     }
636 
637     ms->nextToUpdate = target;
638     return hashTable[ZSTD_hashPtr(ip, hashLog, mls)];
639 }
640 
641 U32 ZSTD_insertAndFindFirstIndex(ZSTD_matchState_t* ms, const BYTE* ip) {
642     const ZSTD_compressionParameters* const cParams = &ms->cParams;
643     return ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, ms->cParams.minMatch);
644 }
645 
646 /* inlining is important to hardwire a hot branch (template emulation) */
647 FORCE_INLINE_TEMPLATE
648 size_t ZSTD_HcFindBestMatch(
649                         ZSTD_matchState_t* ms,
650                         const BYTE* const ip, const BYTE* const iLimit,
651                         size_t* offsetPtr,
652                         const U32 mls, const ZSTD_dictMode_e dictMode)
653 {
654     const ZSTD_compressionParameters* const cParams = &ms->cParams;
655     U32* const chainTable = ms->chainTable;
656     const U32 chainSize = (1 << cParams->chainLog);
657     const U32 chainMask = chainSize-1;
658     const BYTE* const base = ms->window.base;
659     const BYTE* const dictBase = ms->window.dictBase;
660     const U32 dictLimit = ms->window.dictLimit;
661     const BYTE* const prefixStart = base + dictLimit;
662     const BYTE* const dictEnd = dictBase + dictLimit;
663     const U32 curr = (U32)(ip-base);
664     const U32 maxDistance = 1U << cParams->windowLog;
665     const U32 lowestValid = ms->window.lowLimit;
666     const U32 withinMaxDistance = (curr - lowestValid > maxDistance) ? curr - maxDistance : lowestValid;
667     const U32 isDictionary = (ms->loadedDictEnd != 0);
668     const U32 lowLimit = isDictionary ? lowestValid : withinMaxDistance;
669     const U32 minChain = curr > chainSize ? curr - chainSize : 0;
670     U32 nbAttempts = 1U << cParams->searchLog;
671     size_t ml=4-1;
672 
673     const ZSTD_matchState_t* const dms = ms->dictMatchState;
674     const U32 ddsHashLog = dictMode == ZSTD_dedicatedDictSearch
675                          ? dms->cParams.hashLog - ZSTD_LAZY_DDSS_BUCKET_LOG : 0;
676     const size_t ddsIdx = dictMode == ZSTD_dedicatedDictSearch
677                         ? ZSTD_hashPtr(ip, ddsHashLog, mls) << ZSTD_LAZY_DDSS_BUCKET_LOG : 0;
678 
679     U32 matchIndex;
680 
681     if (dictMode == ZSTD_dedicatedDictSearch) {
682         const U32* entry = &dms->hashTable[ddsIdx];
683         PREFETCH_L1(entry);
684     }
685 
686     /* HC4 match finder */
687     matchIndex = ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, mls);
688 
689     for ( ; (matchIndex>=lowLimit) & (nbAttempts>0) ; nbAttempts--) {
690         size_t currentMl=0;
691         if ((dictMode != ZSTD_extDict) || matchIndex >= dictLimit) {
692             const BYTE* const match = base + matchIndex;
693             assert(matchIndex >= dictLimit);   /* ensures this is true if dictMode != ZSTD_extDict */
694             if (match[ml] == ip[ml])   /* potentially better */
695                 currentMl = ZSTD_count(ip, match, iLimit);
696         } else {
697             const BYTE* const match = dictBase + matchIndex;
698             assert(match+4 <= dictEnd);
699             if (MEM_read32(match) == MEM_read32(ip))   /* assumption : matchIndex <= dictLimit-4 (by table construction) */
700                 currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, dictEnd, prefixStart) + 4;
701         }
702 
703         /* save best solution */
704         if (currentMl > ml) {
705             ml = currentMl;
706             *offsetPtr = STORE_OFFSET(curr - matchIndex);
707             if (ip+currentMl == iLimit) break; /* best possible, avoids read overflow on next attempt */
708         }
709 
710         if (matchIndex <= minChain) break;
711         matchIndex = NEXT_IN_CHAIN(matchIndex, chainMask);
712     }
713 
714     assert(nbAttempts <= (1U << ZSTD_SEARCHLOG_MAX)); /* Check we haven't underflowed. */
715     if (dictMode == ZSTD_dedicatedDictSearch) {
716         ml = ZSTD_dedicatedDictSearch_lazy_search(offsetPtr, ml, nbAttempts, dms,
717                                                   ip, iLimit, prefixStart, curr, dictLimit, ddsIdx);
718     } else if (dictMode == ZSTD_dictMatchState) {
719         const U32* const dmsChainTable = dms->chainTable;
720         const U32 dmsChainSize         = (1 << dms->cParams.chainLog);
721         const U32 dmsChainMask         = dmsChainSize - 1;
722         const U32 dmsLowestIndex       = dms->window.dictLimit;
723         const BYTE* const dmsBase      = dms->window.base;
724         const BYTE* const dmsEnd       = dms->window.nextSrc;
725         const U32 dmsSize              = (U32)(dmsEnd - dmsBase);
726         const U32 dmsIndexDelta        = dictLimit - dmsSize;
727         const U32 dmsMinChain = dmsSize > dmsChainSize ? dmsSize - dmsChainSize : 0;
728 
729         matchIndex = dms->hashTable[ZSTD_hashPtr(ip, dms->cParams.hashLog, mls)];
730 
731         for ( ; (matchIndex>=dmsLowestIndex) & (nbAttempts>0) ; nbAttempts--) {
732             size_t currentMl=0;
733             const BYTE* const match = dmsBase + matchIndex;
734             assert(match+4 <= dmsEnd);
735             if (MEM_read32(match) == MEM_read32(ip))   /* assumption : matchIndex <= dictLimit-4 (by table construction) */
736                 currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, dmsEnd, prefixStart) + 4;
737 
738             /* save best solution */
739             if (currentMl > ml) {
740                 ml = currentMl;
741                 assert(curr > matchIndex + dmsIndexDelta);
742                 *offsetPtr = STORE_OFFSET(curr - (matchIndex + dmsIndexDelta));
743                 if (ip+currentMl == iLimit) break; /* best possible, avoids read overflow on next attempt */
744             }
745 
746             if (matchIndex <= dmsMinChain) break;
747 
748             matchIndex = dmsChainTable[matchIndex & dmsChainMask];
749         }
750     }
751 
752     return ml;
753 }
754 
755 /* *********************************
756 * (SIMD) Row-based matchfinder
757 ***********************************/
758 /* Constants for row-based hash */
759 #define ZSTD_ROW_HASH_TAG_OFFSET 16     /* byte offset of hashes in the match state's tagTable from the beginning of a row */
760 #define ZSTD_ROW_HASH_TAG_BITS 8        /* nb bits to use for the tag */
761 #define ZSTD_ROW_HASH_TAG_MASK ((1u << ZSTD_ROW_HASH_TAG_BITS) - 1)
762 #define ZSTD_ROW_HASH_MAX_ENTRIES 64    /* absolute maximum number of entries per row, for all configurations */
763 
764 #define ZSTD_ROW_HASH_CACHE_MASK (ZSTD_ROW_HASH_CACHE_SIZE - 1)
765 
766 typedef U64 ZSTD_VecMask;   /* Clarifies when we are interacting with a U64 representing a mask of matches */
767 
768 /* ZSTD_VecMask_next():
769  * Starting from the LSB, returns the idx of the next non-zero bit.
770  * Basically counting the nb of trailing zeroes.
771  */
772 static U32 ZSTD_VecMask_next(ZSTD_VecMask val) {
773     assert(val != 0);
774 #   if defined(_MSC_VER) && defined(_WIN64)
775         if (val != 0) {
776             unsigned long r;
777             _BitScanForward64(&r, val);
778             return (U32)(r);
779         } else {
780             /* Should not reach this code path */
781             __assume(0);
782         }
783 #   elif (defined(__GNUC__) && ((__GNUC__ > 3) || ((__GNUC__ == 3) && (__GNUC_MINOR__ >= 4))))
784     if (sizeof(size_t) == 4) {
785         U32 mostSignificantWord = (U32)(val >> 32);
786         U32 leastSignificantWord = (U32)val;
787         if (leastSignificantWord == 0) {
788             return 32 + (U32)__builtin_ctz(mostSignificantWord);
789         } else {
790             return (U32)__builtin_ctz(leastSignificantWord);
791         }
792     } else {
793         return (U32)__builtin_ctzll(val);
794     }
795 #   else
796     /* Software ctz version: http://aggregate.org/MAGIC/#Trailing%20Zero%20Count
797      * and: https://stackoverflow.com/questions/2709430/count-number-of-bits-in-a-64-bit-long-big-integer
798      */
799     val = ~val & (val - 1ULL); /* Lowest set bit mask */
800     val = val - ((val >> 1) & 0x5555555555555555);
801     val = (val & 0x3333333333333333ULL) + ((val >> 2) & 0x3333333333333333ULL);
802     return (U32)((((val + (val >> 4)) & 0xF0F0F0F0F0F0F0FULL) * 0x101010101010101ULL) >> 56);
803 #   endif
804 }
805 
806 /* ZSTD_rotateRight_*():
807  * Rotates a bitfield to the right by "count" bits.
808  * https://en.wikipedia.org/w/index.php?title=Circular_shift&oldid=991635599#Implementing_circular_shifts
809  */
810 FORCE_INLINE_TEMPLATE
811 U64 ZSTD_rotateRight_U64(U64 const value, U32 count) {
812     assert(count < 64);
813     count &= 0x3F; /* for fickle pattern recognition */
814     return (value >> count) | (U64)(value << ((0U - count) & 0x3F));
815 }
816 
817 FORCE_INLINE_TEMPLATE
818 U32 ZSTD_rotateRight_U32(U32 const value, U32 count) {
819     assert(count < 32);
820     count &= 0x1F; /* for fickle pattern recognition */
821     return (value >> count) | (U32)(value << ((0U - count) & 0x1F));
822 }
823 
824 FORCE_INLINE_TEMPLATE
825 U16 ZSTD_rotateRight_U16(U16 const value, U32 count) {
826     assert(count < 16);
827     count &= 0x0F; /* for fickle pattern recognition */
828     return (value >> count) | (U16)(value << ((0U - count) & 0x0F));
829 }
830 
831 /* ZSTD_row_nextIndex():
832  * Returns the next index to insert at within a tagTable row, and updates the "head"
833  * value to reflect the update. Essentially cycles backwards from [0, {entries per row})
834  */
835 FORCE_INLINE_TEMPLATE U32 ZSTD_row_nextIndex(BYTE* const tagRow, U32 const rowMask) {
836   U32 const next = (*tagRow - 1) & rowMask;
837   *tagRow = (BYTE)next;
838   return next;
839 }
840 
841 /* ZSTD_isAligned():
842  * Checks that a pointer is aligned to "align" bytes which must be a power of 2.
843  */
844 MEM_STATIC int ZSTD_isAligned(void const* ptr, size_t align) {
845     assert((align & (align - 1)) == 0);
846     return (((size_t)ptr) & (align - 1)) == 0;
847 }
848 
849 /* ZSTD_row_prefetch():
850  * Performs prefetching for the hashTable and tagTable at a given row.
851  */
852 FORCE_INLINE_TEMPLATE void ZSTD_row_prefetch(U32 const* hashTable, U16 const* tagTable, U32 const relRow, U32 const rowLog) {
853     PREFETCH_L1(hashTable + relRow);
854     if (rowLog >= 5) {
855         PREFETCH_L1(hashTable + relRow + 16);
856         /* Note: prefetching more of the hash table does not appear to be beneficial for 128-entry rows */
857     }
858     PREFETCH_L1(tagTable + relRow);
859     if (rowLog == 6) {
860         PREFETCH_L1(tagTable + relRow + 32);
861     }
862     assert(rowLog == 4 || rowLog == 5 || rowLog == 6);
863     assert(ZSTD_isAligned(hashTable + relRow, 64));                 /* prefetched hash row always 64-byte aligned */
864     assert(ZSTD_isAligned(tagTable + relRow, (size_t)1 << rowLog)); /* prefetched tagRow sits on correct multiple of bytes (32,64,128) */
865 }
866 
867 /* ZSTD_row_fillHashCache():
868  * Fill up the hash cache starting at idx, prefetching up to ZSTD_ROW_HASH_CACHE_SIZE entries,
869  * but not beyond iLimit.
870  */
871 FORCE_INLINE_TEMPLATE void ZSTD_row_fillHashCache(ZSTD_matchState_t* ms, const BYTE* base,
872                                    U32 const rowLog, U32 const mls,
873                                    U32 idx, const BYTE* const iLimit)
874 {
875     U32 const* const hashTable = ms->hashTable;
876     U16 const* const tagTable = ms->tagTable;
877     U32 const hashLog = ms->rowHashLog;
878     U32 const maxElemsToPrefetch = (base + idx) > iLimit ? 0 : (U32)(iLimit - (base + idx) + 1);
879     U32 const lim = idx + MIN(ZSTD_ROW_HASH_CACHE_SIZE, maxElemsToPrefetch);
880 
881     for (; idx < lim; ++idx) {
882         U32 const hash = (U32)ZSTD_hashPtr(base + idx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls);
883         U32 const row = (hash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
884         ZSTD_row_prefetch(hashTable, tagTable, row, rowLog);
885         ms->hashCache[idx & ZSTD_ROW_HASH_CACHE_MASK] = hash;
886     }
887 
888     DEBUGLOG(6, "ZSTD_row_fillHashCache(): [%u %u %u %u %u %u %u %u]", ms->hashCache[0], ms->hashCache[1],
889                                                      ms->hashCache[2], ms->hashCache[3], ms->hashCache[4],
890                                                      ms->hashCache[5], ms->hashCache[6], ms->hashCache[7]);
891 }
892 
893 /* ZSTD_row_nextCachedHash():
894  * Returns the hash of base + idx, and replaces the hash in the hash cache with the byte at
895  * base + idx + ZSTD_ROW_HASH_CACHE_SIZE. Also prefetches the appropriate rows from hashTable and tagTable.
896  */
897 FORCE_INLINE_TEMPLATE U32 ZSTD_row_nextCachedHash(U32* cache, U32 const* hashTable,
898                                                   U16 const* tagTable, BYTE const* base,
899                                                   U32 idx, U32 const hashLog,
900                                                   U32 const rowLog, U32 const mls)
901 {
902     U32 const newHash = (U32)ZSTD_hashPtr(base+idx+ZSTD_ROW_HASH_CACHE_SIZE, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls);
903     U32 const row = (newHash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
904     ZSTD_row_prefetch(hashTable, tagTable, row, rowLog);
905     {   U32 const hash = cache[idx & ZSTD_ROW_HASH_CACHE_MASK];
906         cache[idx & ZSTD_ROW_HASH_CACHE_MASK] = newHash;
907         return hash;
908     }
909 }
910 
911 /* ZSTD_row_update_internalImpl():
912  * Updates the hash table with positions starting from updateStartIdx until updateEndIdx.
913  */
914 FORCE_INLINE_TEMPLATE void ZSTD_row_update_internalImpl(ZSTD_matchState_t* ms,
915                                                         U32 updateStartIdx, U32 const updateEndIdx,
916                                                         U32 const mls, U32 const rowLog,
917                                                         U32 const rowMask, U32 const useCache)
918 {
919     U32* const hashTable = ms->hashTable;
920     U16* const tagTable = ms->tagTable;
921     U32 const hashLog = ms->rowHashLog;
922     const BYTE* const base = ms->window.base;
923 
924     DEBUGLOG(6, "ZSTD_row_update_internalImpl(): updateStartIdx=%u, updateEndIdx=%u", updateStartIdx, updateEndIdx);
925     for (; updateStartIdx < updateEndIdx; ++updateStartIdx) {
926         U32 const hash = useCache ? ZSTD_row_nextCachedHash(ms->hashCache, hashTable, tagTable, base, updateStartIdx, hashLog, rowLog, mls)
927                                   : (U32)ZSTD_hashPtr(base + updateStartIdx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls);
928         U32 const relRow = (hash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
929         U32* const row = hashTable + relRow;
930         BYTE* tagRow = (BYTE*)(tagTable + relRow);  /* Though tagTable is laid out as a table of U16, each tag is only 1 byte.
931                                                        Explicit cast allows us to get exact desired position within each row */
932         U32 const pos = ZSTD_row_nextIndex(tagRow, rowMask);
933 
934         assert(hash == ZSTD_hashPtr(base + updateStartIdx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls));
935         ((BYTE*)tagRow)[pos + ZSTD_ROW_HASH_TAG_OFFSET] = hash & ZSTD_ROW_HASH_TAG_MASK;
936         row[pos] = updateStartIdx;
937     }
938 }
939 
940 /* ZSTD_row_update_internal():
941  * Inserts the byte at ip into the appropriate position in the hash table, and updates ms->nextToUpdate.
942  * Skips sections of long matches as is necessary.
943  */
944 FORCE_INLINE_TEMPLATE void ZSTD_row_update_internal(ZSTD_matchState_t* ms, const BYTE* ip,
945                                                     U32 const mls, U32 const rowLog,
946                                                     U32 const rowMask, U32 const useCache)
947 {
948     U32 idx = ms->nextToUpdate;
949     const BYTE* const base = ms->window.base;
950     const U32 target = (U32)(ip - base);
951     const U32 kSkipThreshold = 384;
952     const U32 kMaxMatchStartPositionsToUpdate = 96;
953     const U32 kMaxMatchEndPositionsToUpdate = 32;
954 
955     if (useCache) {
956         /* Only skip positions when using hash cache, i.e.
957          * if we are loading a dict, don't skip anything.
958          * If we decide to skip, then we only update a set number
959          * of positions at the beginning and end of the match.
960          */
961         if (UNLIKELY(target - idx > kSkipThreshold)) {
962             U32 const bound = idx + kMaxMatchStartPositionsToUpdate;
963             ZSTD_row_update_internalImpl(ms, idx, bound, mls, rowLog, rowMask, useCache);
964             idx = target - kMaxMatchEndPositionsToUpdate;
965             ZSTD_row_fillHashCache(ms, base, rowLog, mls, idx, ip+1);
966         }
967     }
968     assert(target >= idx);
969     ZSTD_row_update_internalImpl(ms, idx, target, mls, rowLog, rowMask, useCache);
970     ms->nextToUpdate = target;
971 }
972 
973 /* ZSTD_row_update():
974  * External wrapper for ZSTD_row_update_internal(). Used for filling the hashtable during dictionary
975  * processing.
976  */
977 void ZSTD_row_update(ZSTD_matchState_t* const ms, const BYTE* ip) {
978     const U32 rowLog = BOUNDED(4, ms->cParams.searchLog, 6);
979     const U32 rowMask = (1u << rowLog) - 1;
980     const U32 mls = MIN(ms->cParams.minMatch, 6 /* mls caps out at 6 */);
981 
982     DEBUGLOG(5, "ZSTD_row_update(), rowLog=%u", rowLog);
983     ZSTD_row_update_internal(ms, ip, mls, rowLog, rowMask, 0 /* dont use cache */);
984 }
985 
986 #if defined(ZSTD_ARCH_X86_SSE2)
987 FORCE_INLINE_TEMPLATE ZSTD_VecMask
988 ZSTD_row_getSSEMask(int nbChunks, const BYTE* const src, const BYTE tag, const U32 head)
989 {
990     const __m128i comparisonMask = _mm_set1_epi8((char)tag);
991     int matches[4] = {0};
992     int i;
993     assert(nbChunks == 1 || nbChunks == 2 || nbChunks == 4);
994     for (i=0; i<nbChunks; i++) {
995         const __m128i chunk = _mm_loadu_si128((const __m128i*)(const void*)(src + 16*i));
996         const __m128i equalMask = _mm_cmpeq_epi8(chunk, comparisonMask);
997         matches[i] = _mm_movemask_epi8(equalMask);
998     }
999     if (nbChunks == 1) return ZSTD_rotateRight_U16((U16)matches[0], head);
1000     if (nbChunks == 2) return ZSTD_rotateRight_U32((U32)matches[1] << 16 | (U32)matches[0], head);
1001     assert(nbChunks == 4);
1002     return ZSTD_rotateRight_U64((U64)matches[3] << 48 | (U64)matches[2] << 32 | (U64)matches[1] << 16 | (U64)matches[0], head);
1003 }
1004 #endif
1005 
1006 /* Returns a ZSTD_VecMask (U32) that has the nth bit set to 1 if the newly-computed "tag" matches
1007  * the hash at the nth position in a row of the tagTable.
1008  * Each row is a circular buffer beginning at the value of "head". So we must rotate the "matches" bitfield
1009  * to match up with the actual layout of the entries within the hashTable */
1010 FORCE_INLINE_TEMPLATE ZSTD_VecMask
1011 ZSTD_row_getMatchMask(const BYTE* const tagRow, const BYTE tag, const U32 head, const U32 rowEntries)
1012 {
1013     const BYTE* const src = tagRow + ZSTD_ROW_HASH_TAG_OFFSET;
1014     assert((rowEntries == 16) || (rowEntries == 32) || rowEntries == 64);
1015     assert(rowEntries <= ZSTD_ROW_HASH_MAX_ENTRIES);
1016 
1017 #if defined(ZSTD_ARCH_X86_SSE2)
1018 
1019     return ZSTD_row_getSSEMask(rowEntries / 16, src, tag, head);
1020 
1021 #else /* SW or NEON-LE */
1022 
1023 # if defined(ZSTD_ARCH_ARM_NEON)
1024   /* This NEON path only works for little endian - otherwise use SWAR below */
1025     if (MEM_isLittleEndian()) {
1026         if (rowEntries == 16) {
1027             const uint8x16_t chunk = vld1q_u8(src);
1028             const uint16x8_t equalMask = vreinterpretq_u16_u8(vceqq_u8(chunk, vdupq_n_u8(tag)));
1029             const uint16x8_t t0 = vshlq_n_u16(equalMask, 7);
1030             const uint32x4_t t1 = vreinterpretq_u32_u16(vsriq_n_u16(t0, t0, 14));
1031             const uint64x2_t t2 = vreinterpretq_u64_u32(vshrq_n_u32(t1, 14));
1032             const uint8x16_t t3 = vreinterpretq_u8_u64(vsraq_n_u64(t2, t2, 28));
1033             const U16 hi = (U16)vgetq_lane_u8(t3, 8);
1034             const U16 lo = (U16)vgetq_lane_u8(t3, 0);
1035             return ZSTD_rotateRight_U16((hi << 8) | lo, head);
1036         } else if (rowEntries == 32) {
1037             const uint16x8x2_t chunk = vld2q_u16((const U16*)(const void*)src);
1038             const uint8x16_t chunk0 = vreinterpretq_u8_u16(chunk.val[0]);
1039             const uint8x16_t chunk1 = vreinterpretq_u8_u16(chunk.val[1]);
1040             const uint8x16_t equalMask0 = vceqq_u8(chunk0, vdupq_n_u8(tag));
1041             const uint8x16_t equalMask1 = vceqq_u8(chunk1, vdupq_n_u8(tag));
1042             const int8x8_t pack0 = vqmovn_s16(vreinterpretq_s16_u8(equalMask0));
1043             const int8x8_t pack1 = vqmovn_s16(vreinterpretq_s16_u8(equalMask1));
1044             const uint8x8_t t0 = vreinterpret_u8_s8(pack0);
1045             const uint8x8_t t1 = vreinterpret_u8_s8(pack1);
1046             const uint8x8_t t2 = vsri_n_u8(t1, t0, 2);
1047             const uint8x8x2_t t3 = vuzp_u8(t2, t0);
1048             const uint8x8_t t4 = vsri_n_u8(t3.val[1], t3.val[0], 4);
1049             const U32 matches = vget_lane_u32(vreinterpret_u32_u8(t4), 0);
1050             return ZSTD_rotateRight_U32(matches, head);
1051         } else { /* rowEntries == 64 */
1052             const uint8x16x4_t chunk = vld4q_u8(src);
1053             const uint8x16_t dup = vdupq_n_u8(tag);
1054             const uint8x16_t cmp0 = vceqq_u8(chunk.val[0], dup);
1055             const uint8x16_t cmp1 = vceqq_u8(chunk.val[1], dup);
1056             const uint8x16_t cmp2 = vceqq_u8(chunk.val[2], dup);
1057             const uint8x16_t cmp3 = vceqq_u8(chunk.val[3], dup);
1058 
1059             const uint8x16_t t0 = vsriq_n_u8(cmp1, cmp0, 1);
1060             const uint8x16_t t1 = vsriq_n_u8(cmp3, cmp2, 1);
1061             const uint8x16_t t2 = vsriq_n_u8(t1, t0, 2);
1062             const uint8x16_t t3 = vsriq_n_u8(t2, t2, 4);
1063             const uint8x8_t t4 = vshrn_n_u16(vreinterpretq_u16_u8(t3), 4);
1064             const U64 matches = vget_lane_u64(vreinterpret_u64_u8(t4), 0);
1065             return ZSTD_rotateRight_U64(matches, head);
1066         }
1067     }
1068 # endif /* ZSTD_ARCH_ARM_NEON */
1069     /* SWAR */
1070     {   const size_t chunkSize = sizeof(size_t);
1071         const size_t shiftAmount = ((chunkSize * 8) - chunkSize);
1072         const size_t xFF = ~((size_t)0);
1073         const size_t x01 = xFF / 0xFF;
1074         const size_t x80 = x01 << 7;
1075         const size_t splatChar = tag * x01;
1076         ZSTD_VecMask matches = 0;
1077         int i = rowEntries - chunkSize;
1078         assert((sizeof(size_t) == 4) || (sizeof(size_t) == 8));
1079         if (MEM_isLittleEndian()) { /* runtime check so have two loops */
1080             const size_t extractMagic = (xFF / 0x7F) >> chunkSize;
1081             do {
1082                 size_t chunk = MEM_readST(&src[i]);
1083                 chunk ^= splatChar;
1084                 chunk = (((chunk | x80) - x01) | chunk) & x80;
1085                 matches <<= chunkSize;
1086                 matches |= (chunk * extractMagic) >> shiftAmount;
1087                 i -= chunkSize;
1088             } while (i >= 0);
1089         } else { /* big endian: reverse bits during extraction */
1090             const size_t msb = xFF ^ (xFF >> 1);
1091             const size_t extractMagic = (msb / 0x1FF) | msb;
1092             do {
1093                 size_t chunk = MEM_readST(&src[i]);
1094                 chunk ^= splatChar;
1095                 chunk = (((chunk | x80) - x01) | chunk) & x80;
1096                 matches <<= chunkSize;
1097                 matches |= ((chunk >> 7) * extractMagic) >> shiftAmount;
1098                 i -= chunkSize;
1099             } while (i >= 0);
1100         }
1101         matches = ~matches;
1102         if (rowEntries == 16) {
1103             return ZSTD_rotateRight_U16((U16)matches, head);
1104         } else if (rowEntries == 32) {
1105             return ZSTD_rotateRight_U32((U32)matches, head);
1106         } else {
1107             return ZSTD_rotateRight_U64((U64)matches, head);
1108         }
1109     }
1110 #endif
1111 }
1112 
1113 /* The high-level approach of the SIMD row based match finder is as follows:
1114  * - Figure out where to insert the new entry:
1115  *      - Generate a hash from a byte along with an additional 1-byte "short hash". The additional byte is our "tag"
1116  *      - The hashTable is effectively split into groups or "rows" of 16 or 32 entries of U32, and the hash determines
1117  *        which row to insert into.
1118  *      - Determine the correct position within the row to insert the entry into. Each row of 16 or 32 can
1119  *        be considered as a circular buffer with a "head" index that resides in the tagTable.
1120  *      - Also insert the "tag" into the equivalent row and position in the tagTable.
1121  *          - Note: The tagTable has 17 or 33 1-byte entries per row, due to 16 or 32 tags, and 1 "head" entry.
1122  *                  The 17 or 33 entry rows are spaced out to occur every 32 or 64 bytes, respectively,
1123  *                  for alignment/performance reasons, leaving some bytes unused.
1124  * - Use SIMD to efficiently compare the tags in the tagTable to the 1-byte "short hash" and
1125  *   generate a bitfield that we can cycle through to check the collisions in the hash table.
1126  * - Pick the longest match.
1127  */
1128 FORCE_INLINE_TEMPLATE
1129 size_t ZSTD_RowFindBestMatch(
1130                         ZSTD_matchState_t* ms,
1131                         const BYTE* const ip, const BYTE* const iLimit,
1132                         size_t* offsetPtr,
1133                         const U32 mls, const ZSTD_dictMode_e dictMode,
1134                         const U32 rowLog)
1135 {
1136     U32* const hashTable = ms->hashTable;
1137     U16* const tagTable = ms->tagTable;
1138     U32* const hashCache = ms->hashCache;
1139     const U32 hashLog = ms->rowHashLog;
1140     const ZSTD_compressionParameters* const cParams = &ms->cParams;
1141     const BYTE* const base = ms->window.base;
1142     const BYTE* const dictBase = ms->window.dictBase;
1143     const U32 dictLimit = ms->window.dictLimit;
1144     const BYTE* const prefixStart = base + dictLimit;
1145     const BYTE* const dictEnd = dictBase + dictLimit;
1146     const U32 curr = (U32)(ip-base);
1147     const U32 maxDistance = 1U << cParams->windowLog;
1148     const U32 lowestValid = ms->window.lowLimit;
1149     const U32 withinMaxDistance = (curr - lowestValid > maxDistance) ? curr - maxDistance : lowestValid;
1150     const U32 isDictionary = (ms->loadedDictEnd != 0);
1151     const U32 lowLimit = isDictionary ? lowestValid : withinMaxDistance;
1152     const U32 rowEntries = (1U << rowLog);
1153     const U32 rowMask = rowEntries - 1;
1154     const U32 cappedSearchLog = MIN(cParams->searchLog, rowLog); /* nb of searches is capped at nb entries per row */
1155     U32 nbAttempts = 1U << cappedSearchLog;
1156     size_t ml=4-1;
1157 
1158     /* DMS/DDS variables that may be referenced laster */
1159     const ZSTD_matchState_t* const dms = ms->dictMatchState;
1160 
1161     /* Initialize the following variables to satisfy static analyzer */
1162     size_t ddsIdx = 0;
1163     U32 ddsExtraAttempts = 0; /* cctx hash tables are limited in searches, but allow extra searches into DDS */
1164     U32 dmsTag = 0;
1165     U32* dmsRow = NULL;
1166     BYTE* dmsTagRow = NULL;
1167 
1168     if (dictMode == ZSTD_dedicatedDictSearch) {
1169         const U32 ddsHashLog = dms->cParams.hashLog - ZSTD_LAZY_DDSS_BUCKET_LOG;
1170         {   /* Prefetch DDS hashtable entry */
1171             ddsIdx = ZSTD_hashPtr(ip, ddsHashLog, mls) << ZSTD_LAZY_DDSS_BUCKET_LOG;
1172             PREFETCH_L1(&dms->hashTable[ddsIdx]);
1173         }
1174         ddsExtraAttempts = cParams->searchLog > rowLog ? 1U << (cParams->searchLog - rowLog) : 0;
1175     }
1176 
1177     if (dictMode == ZSTD_dictMatchState) {
1178         /* Prefetch DMS rows */
1179         U32* const dmsHashTable = dms->hashTable;
1180         U16* const dmsTagTable = dms->tagTable;
1181         U32 const dmsHash = (U32)ZSTD_hashPtr(ip, dms->rowHashLog + ZSTD_ROW_HASH_TAG_BITS, mls);
1182         U32 const dmsRelRow = (dmsHash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
1183         dmsTag = dmsHash & ZSTD_ROW_HASH_TAG_MASK;
1184         dmsTagRow = (BYTE*)(dmsTagTable + dmsRelRow);
1185         dmsRow = dmsHashTable + dmsRelRow;
1186         ZSTD_row_prefetch(dmsHashTable, dmsTagTable, dmsRelRow, rowLog);
1187     }
1188 
1189     /* Update the hashTable and tagTable up to (but not including) ip */
1190     ZSTD_row_update_internal(ms, ip, mls, rowLog, rowMask, 1 /* useCache */);
1191     {   /* Get the hash for ip, compute the appropriate row */
1192         U32 const hash = ZSTD_row_nextCachedHash(hashCache, hashTable, tagTable, base, curr, hashLog, rowLog, mls);
1193         U32 const relRow = (hash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
1194         U32 const tag = hash & ZSTD_ROW_HASH_TAG_MASK;
1195         U32* const row = hashTable + relRow;
1196         BYTE* tagRow = (BYTE*)(tagTable + relRow);
1197         U32 const head = *tagRow & rowMask;
1198         U32 matchBuffer[ZSTD_ROW_HASH_MAX_ENTRIES];
1199         size_t numMatches = 0;
1200         size_t currMatch = 0;
1201         ZSTD_VecMask matches = ZSTD_row_getMatchMask(tagRow, (BYTE)tag, head, rowEntries);
1202 
1203         /* Cycle through the matches and prefetch */
1204         for (; (matches > 0) && (nbAttempts > 0); --nbAttempts, matches &= (matches - 1)) {
1205             U32 const matchPos = (head + ZSTD_VecMask_next(matches)) & rowMask;
1206             U32 const matchIndex = row[matchPos];
1207             assert(numMatches < rowEntries);
1208             if (matchIndex < lowLimit)
1209                 break;
1210             if ((dictMode != ZSTD_extDict) || matchIndex >= dictLimit) {
1211                 PREFETCH_L1(base + matchIndex);
1212             } else {
1213                 PREFETCH_L1(dictBase + matchIndex);
1214             }
1215             matchBuffer[numMatches++] = matchIndex;
1216         }
1217 
1218         /* Speed opt: insert current byte into hashtable too. This allows us to avoid one iteration of the loop
1219            in ZSTD_row_update_internal() at the next search. */
1220         {
1221             U32 const pos = ZSTD_row_nextIndex(tagRow, rowMask);
1222             tagRow[pos + ZSTD_ROW_HASH_TAG_OFFSET] = (BYTE)tag;
1223             row[pos] = ms->nextToUpdate++;
1224         }
1225 
1226         /* Return the longest match */
1227         for (; currMatch < numMatches; ++currMatch) {
1228             U32 const matchIndex = matchBuffer[currMatch];
1229             size_t currentMl=0;
1230             assert(matchIndex < curr);
1231             assert(matchIndex >= lowLimit);
1232 
1233             if ((dictMode != ZSTD_extDict) || matchIndex >= dictLimit) {
1234                 const BYTE* const match = base + matchIndex;
1235                 assert(matchIndex >= dictLimit);   /* ensures this is true if dictMode != ZSTD_extDict */
1236                 if (match[ml] == ip[ml])   /* potentially better */
1237                     currentMl = ZSTD_count(ip, match, iLimit);
1238             } else {
1239                 const BYTE* const match = dictBase + matchIndex;
1240                 assert(match+4 <= dictEnd);
1241                 if (MEM_read32(match) == MEM_read32(ip))   /* assumption : matchIndex <= dictLimit-4 (by table construction) */
1242                     currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, dictEnd, prefixStart) + 4;
1243             }
1244 
1245             /* Save best solution */
1246             if (currentMl > ml) {
1247                 ml = currentMl;
1248                 *offsetPtr = STORE_OFFSET(curr - matchIndex);
1249                 if (ip+currentMl == iLimit) break; /* best possible, avoids read overflow on next attempt */
1250             }
1251         }
1252     }
1253 
1254     assert(nbAttempts <= (1U << ZSTD_SEARCHLOG_MAX)); /* Check we haven't underflowed. */
1255     if (dictMode == ZSTD_dedicatedDictSearch) {
1256         ml = ZSTD_dedicatedDictSearch_lazy_search(offsetPtr, ml, nbAttempts + ddsExtraAttempts, dms,
1257                                                   ip, iLimit, prefixStart, curr, dictLimit, ddsIdx);
1258     } else if (dictMode == ZSTD_dictMatchState) {
1259         /* TODO: Measure and potentially add prefetching to DMS */
1260         const U32 dmsLowestIndex       = dms->window.dictLimit;
1261         const BYTE* const dmsBase      = dms->window.base;
1262         const BYTE* const dmsEnd       = dms->window.nextSrc;
1263         const U32 dmsSize              = (U32)(dmsEnd - dmsBase);
1264         const U32 dmsIndexDelta        = dictLimit - dmsSize;
1265 
1266         {   U32 const head = *dmsTagRow & rowMask;
1267             U32 matchBuffer[ZSTD_ROW_HASH_MAX_ENTRIES];
1268             size_t numMatches = 0;
1269             size_t currMatch = 0;
1270             ZSTD_VecMask matches = ZSTD_row_getMatchMask(dmsTagRow, (BYTE)dmsTag, head, rowEntries);
1271 
1272             for (; (matches > 0) && (nbAttempts > 0); --nbAttempts, matches &= (matches - 1)) {
1273                 U32 const matchPos = (head + ZSTD_VecMask_next(matches)) & rowMask;
1274                 U32 const matchIndex = dmsRow[matchPos];
1275                 if (matchIndex < dmsLowestIndex)
1276                     break;
1277                 PREFETCH_L1(dmsBase + matchIndex);
1278                 matchBuffer[numMatches++] = matchIndex;
1279             }
1280 
1281             /* Return the longest match */
1282             for (; currMatch < numMatches; ++currMatch) {
1283                 U32 const matchIndex = matchBuffer[currMatch];
1284                 size_t currentMl=0;
1285                 assert(matchIndex >= dmsLowestIndex);
1286                 assert(matchIndex < curr);
1287 
1288                 {   const BYTE* const match = dmsBase + matchIndex;
1289                     assert(match+4 <= dmsEnd);
1290                     if (MEM_read32(match) == MEM_read32(ip))
1291                         currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, dmsEnd, prefixStart) + 4;
1292                 }
1293 
1294                 if (currentMl > ml) {
1295                     ml = currentMl;
1296                     assert(curr > matchIndex + dmsIndexDelta);
1297                     *offsetPtr = STORE_OFFSET(curr - (matchIndex + dmsIndexDelta));
1298                     if (ip+currentMl == iLimit) break;
1299                 }
1300             }
1301         }
1302     }
1303     return ml;
1304 }
1305 
1306 
1307 typedef size_t (*searchMax_f)(
1308                     ZSTD_matchState_t* ms,
1309                     const BYTE* ip, const BYTE* iLimit, size_t* offsetPtr);
1310 
1311 /**
1312  * This struct contains the functions necessary for lazy to search.
1313  * Currently, that is only searchMax. However, it is still valuable to have the
1314  * VTable because this makes it easier to add more functions to the VTable later.
1315  *
1316  * TODO: The start of the search function involves loading and calculating a
1317  * bunch of constants from the ZSTD_matchState_t. These computations could be
1318  * done in an initialization function, and saved somewhere in the match state.
1319  * Then we could pass a pointer to the saved state instead of the match state,
1320  * and avoid duplicate computations.
1321  *
1322  * TODO: Move the match re-winding into searchMax. This improves compression
1323  * ratio, and unlocks further simplifications with the next TODO.
1324  *
1325  * TODO: Try moving the repcode search into searchMax. After the re-winding
1326  * and repcode search are in searchMax, there is no more logic in the match
1327  * finder loop that requires knowledge about the dictMode. So we should be
1328  * able to avoid force inlining it, and we can join the extDict loop with
1329  * the single segment loop. It should go in searchMax instead of its own
1330  * function to avoid having multiple virtual function calls per search.
1331  */
1332 typedef struct {
1333     searchMax_f searchMax;
1334 } ZSTD_LazyVTable;
1335 
1336 #define GEN_ZSTD_BT_VTABLE(dictMode, mls)                                             \
1337     static size_t ZSTD_BtFindBestMatch_##dictMode##_##mls(                            \
1338             ZSTD_matchState_t* ms,                                                    \
1339             const BYTE* ip, const BYTE* const iLimit,                                 \
1340             size_t* offsetPtr)                                                        \
1341     {                                                                                 \
1342         assert(MAX(4, MIN(6, ms->cParams.minMatch)) == mls);                          \
1343         return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, mls, ZSTD_##dictMode); \
1344     }                                                                                 \
1345     static const ZSTD_LazyVTable ZSTD_BtVTable_##dictMode##_##mls = {                 \
1346         ZSTD_BtFindBestMatch_##dictMode##_##mls                                       \
1347     };
1348 
1349 #define GEN_ZSTD_HC_VTABLE(dictMode, mls)                                             \
1350     static size_t ZSTD_HcFindBestMatch_##dictMode##_##mls(                            \
1351             ZSTD_matchState_t* ms,                                                    \
1352             const BYTE* ip, const BYTE* const iLimit,                                 \
1353             size_t* offsetPtr)                                                        \
1354     {                                                                                 \
1355         assert(MAX(4, MIN(6, ms->cParams.minMatch)) == mls);                          \
1356         return ZSTD_HcFindBestMatch(ms, ip, iLimit, offsetPtr, mls, ZSTD_##dictMode); \
1357     }                                                                                 \
1358     static const ZSTD_LazyVTable ZSTD_HcVTable_##dictMode##_##mls = {                 \
1359         ZSTD_HcFindBestMatch_##dictMode##_##mls                                       \
1360     };
1361 
1362 #define GEN_ZSTD_ROW_VTABLE(dictMode, mls, rowLog)                                             \
1363     static size_t ZSTD_RowFindBestMatch_##dictMode##_##mls##_##rowLog(                         \
1364             ZSTD_matchState_t* ms,                                                             \
1365             const BYTE* ip, const BYTE* const iLimit,                                          \
1366             size_t* offsetPtr)                                                                 \
1367     {                                                                                          \
1368         assert(MAX(4, MIN(6, ms->cParams.minMatch)) == mls);                                   \
1369         assert(MAX(4, MIN(6, ms->cParams.searchLog)) == rowLog);                               \
1370         return ZSTD_RowFindBestMatch(ms, ip, iLimit, offsetPtr, mls, ZSTD_##dictMode, rowLog); \
1371     }                                                                                          \
1372     static const ZSTD_LazyVTable ZSTD_RowVTable_##dictMode##_##mls##_##rowLog = {              \
1373         ZSTD_RowFindBestMatch_##dictMode##_##mls##_##rowLog                                    \
1374     };
1375 
1376 #define ZSTD_FOR_EACH_ROWLOG(X, dictMode, mls) \
1377     X(dictMode, mls, 4)                        \
1378     X(dictMode, mls, 5)                        \
1379     X(dictMode, mls, 6)
1380 
1381 #define ZSTD_FOR_EACH_MLS_ROWLOG(X, dictMode) \
1382     ZSTD_FOR_EACH_ROWLOG(X, dictMode, 4)      \
1383     ZSTD_FOR_EACH_ROWLOG(X, dictMode, 5)      \
1384     ZSTD_FOR_EACH_ROWLOG(X, dictMode, 6)
1385 
1386 #define ZSTD_FOR_EACH_MLS(X, dictMode) \
1387     X(dictMode, 4)                     \
1388     X(dictMode, 5)                     \
1389     X(dictMode, 6)
1390 
1391 #define ZSTD_FOR_EACH_DICT_MODE(X, ...) \
1392     X(__VA_ARGS__, noDict)              \
1393     X(__VA_ARGS__, extDict)             \
1394     X(__VA_ARGS__, dictMatchState)      \
1395     X(__VA_ARGS__, dedicatedDictSearch)
1396 
1397 /* Generate Row VTables for each combination of (dictMode, mls, rowLog) */
1398 ZSTD_FOR_EACH_DICT_MODE(ZSTD_FOR_EACH_MLS_ROWLOG, GEN_ZSTD_ROW_VTABLE)
1399 /* Generate Binary Tree VTables for each combination of (dictMode, mls) */
1400 ZSTD_FOR_EACH_DICT_MODE(ZSTD_FOR_EACH_MLS, GEN_ZSTD_BT_VTABLE)
1401 /* Generate Hash Chain VTables for each combination of (dictMode, mls) */
1402 ZSTD_FOR_EACH_DICT_MODE(ZSTD_FOR_EACH_MLS, GEN_ZSTD_HC_VTABLE)
1403 
1404 #define GEN_ZSTD_BT_VTABLE_ARRAY(dictMode) \
1405     {                                      \
1406         &ZSTD_BtVTable_##dictMode##_4,     \
1407         &ZSTD_BtVTable_##dictMode##_5,     \
1408         &ZSTD_BtVTable_##dictMode##_6      \
1409     }
1410 
1411 #define GEN_ZSTD_HC_VTABLE_ARRAY(dictMode) \
1412     {                                      \
1413         &ZSTD_HcVTable_##dictMode##_4,     \
1414         &ZSTD_HcVTable_##dictMode##_5,     \
1415         &ZSTD_HcVTable_##dictMode##_6      \
1416     }
1417 
1418 #define GEN_ZSTD_ROW_VTABLE_ARRAY_(dictMode, mls) \
1419     {                                             \
1420         &ZSTD_RowVTable_##dictMode##_##mls##_4,   \
1421         &ZSTD_RowVTable_##dictMode##_##mls##_5,   \
1422         &ZSTD_RowVTable_##dictMode##_##mls##_6    \
1423     }
1424 
1425 #define GEN_ZSTD_ROW_VTABLE_ARRAY(dictMode)      \
1426     {                                            \
1427         GEN_ZSTD_ROW_VTABLE_ARRAY_(dictMode, 4), \
1428         GEN_ZSTD_ROW_VTABLE_ARRAY_(dictMode, 5), \
1429         GEN_ZSTD_ROW_VTABLE_ARRAY_(dictMode, 6)  \
1430     }
1431 
1432 #define GEN_ZSTD_VTABLE_ARRAY(X) \
1433     {                            \
1434         X(noDict),               \
1435         X(extDict),              \
1436         X(dictMatchState),       \
1437         X(dedicatedDictSearch)   \
1438     }
1439 
1440 /* *******************************
1441 *  Common parser - lazy strategy
1442 *********************************/
1443 typedef enum { search_hashChain=0, search_binaryTree=1, search_rowHash=2 } searchMethod_e;
1444 
1445 /**
1446  * This table is indexed first by the four ZSTD_dictMode_e values, and then
1447  * by the two searchMethod_e values. NULLs are placed for configurations
1448  * that should never occur (extDict modes go to the other implementation
1449  * below and there is no DDSS for binary tree search yet).
1450  */
1451 
1452 static ZSTD_LazyVTable const*
1453 ZSTD_selectLazyVTable(ZSTD_matchState_t const* ms, searchMethod_e searchMethod, ZSTD_dictMode_e dictMode)
1454 {
1455     /* Fill the Hc/Bt VTable arrays with the right functions for the (dictMode, mls) combination. */
1456     ZSTD_LazyVTable const* const hcVTables[4][3] = GEN_ZSTD_VTABLE_ARRAY(GEN_ZSTD_HC_VTABLE_ARRAY);
1457     ZSTD_LazyVTable const* const btVTables[4][3] = GEN_ZSTD_VTABLE_ARRAY(GEN_ZSTD_BT_VTABLE_ARRAY);
1458     /* Fill the Row VTable array with the right functions for the (dictMode, mls, rowLog) combination. */
1459     ZSTD_LazyVTable const* const rowVTables[4][3][3] = GEN_ZSTD_VTABLE_ARRAY(GEN_ZSTD_ROW_VTABLE_ARRAY);
1460 
1461     U32 const mls = MAX(4, MIN(6, ms->cParams.minMatch));
1462     U32 const rowLog = MAX(4, MIN(6, ms->cParams.searchLog));
1463     switch (searchMethod) {
1464         case search_hashChain:
1465             return hcVTables[dictMode][mls - 4];
1466         case search_binaryTree:
1467             return btVTables[dictMode][mls - 4];
1468         case search_rowHash:
1469             return rowVTables[dictMode][mls - 4][rowLog - 4];
1470         default:
1471             return NULL;
1472     }
1473 }
1474 
1475 FORCE_INLINE_TEMPLATE size_t
1476 ZSTD_compressBlock_lazy_generic(
1477                         ZSTD_matchState_t* ms, seqStore_t* seqStore,
1478                         U32 rep[ZSTD_REP_NUM],
1479                         const void* src, size_t srcSize,
1480                         const searchMethod_e searchMethod, const U32 depth,
1481                         ZSTD_dictMode_e const dictMode)
1482 {
1483     const BYTE* const istart = (const BYTE*)src;
1484     const BYTE* ip = istart;
1485     const BYTE* anchor = istart;
1486     const BYTE* const iend = istart + srcSize;
1487     const BYTE* const ilimit = (searchMethod == search_rowHash) ? iend - 8 - ZSTD_ROW_HASH_CACHE_SIZE : iend - 8;
1488     const BYTE* const base = ms->window.base;
1489     const U32 prefixLowestIndex = ms->window.dictLimit;
1490     const BYTE* const prefixLowest = base + prefixLowestIndex;
1491 
1492     searchMax_f const searchMax = ZSTD_selectLazyVTable(ms, searchMethod, dictMode)->searchMax;
1493     U32 offset_1 = rep[0], offset_2 = rep[1], savedOffset=0;
1494 
1495     const int isDMS = dictMode == ZSTD_dictMatchState;
1496     const int isDDS = dictMode == ZSTD_dedicatedDictSearch;
1497     const int isDxS = isDMS || isDDS;
1498     const ZSTD_matchState_t* const dms = ms->dictMatchState;
1499     const U32 dictLowestIndex      = isDxS ? dms->window.dictLimit : 0;
1500     const BYTE* const dictBase     = isDxS ? dms->window.base : NULL;
1501     const BYTE* const dictLowest   = isDxS ? dictBase + dictLowestIndex : NULL;
1502     const BYTE* const dictEnd      = isDxS ? dms->window.nextSrc : NULL;
1503     const U32 dictIndexDelta       = isDxS ?
1504                                      prefixLowestIndex - (U32)(dictEnd - dictBase) :
1505                                      0;
1506     const U32 dictAndPrefixLength = (U32)((ip - prefixLowest) + (dictEnd - dictLowest));
1507 
1508     assert(searchMax != NULL);
1509 
1510     DEBUGLOG(5, "ZSTD_compressBlock_lazy_generic (dictMode=%u) (searchFunc=%u)", (U32)dictMode, (U32)searchMethod);
1511     ip += (dictAndPrefixLength == 0);
1512     if (dictMode == ZSTD_noDict) {
1513         U32 const curr = (U32)(ip - base);
1514         U32 const windowLow = ZSTD_getLowestPrefixIndex(ms, curr, ms->cParams.windowLog);
1515         U32 const maxRep = curr - windowLow;
1516         if (offset_2 > maxRep) savedOffset = offset_2, offset_2 = 0;
1517         if (offset_1 > maxRep) savedOffset = offset_1, offset_1 = 0;
1518     }
1519     if (isDxS) {
1520         /* dictMatchState repCode checks don't currently handle repCode == 0
1521          * disabling. */
1522         assert(offset_1 <= dictAndPrefixLength);
1523         assert(offset_2 <= dictAndPrefixLength);
1524     }
1525 
1526     if (searchMethod == search_rowHash) {
1527         const U32 rowLog = MAX(4, MIN(6, ms->cParams.searchLog));
1528         ZSTD_row_fillHashCache(ms, base, rowLog,
1529                             MIN(ms->cParams.minMatch, 6 /* mls caps out at 6 */),
1530                             ms->nextToUpdate, ilimit);
1531     }
1532 
1533     /* Match Loop */
1534 #if defined(__GNUC__) && defined(__x86_64__)
1535     /* I've measured random a 5% speed loss on levels 5 & 6 (greedy) when the
1536      * code alignment is perturbed. To fix the instability align the loop on 32-bytes.
1537      */
1538     __asm__(".p2align 5");
1539 #endif
1540     while (ip < ilimit) {
1541         size_t matchLength=0;
1542         size_t offcode=STORE_REPCODE_1;
1543         const BYTE* start=ip+1;
1544         DEBUGLOG(7, "search baseline (depth 0)");
1545 
1546         /* check repCode */
1547         if (isDxS) {
1548             const U32 repIndex = (U32)(ip - base) + 1 - offset_1;
1549             const BYTE* repMatch = ((dictMode == ZSTD_dictMatchState || dictMode == ZSTD_dedicatedDictSearch)
1550                                 && repIndex < prefixLowestIndex) ?
1551                                    dictBase + (repIndex - dictIndexDelta) :
1552                                    base + repIndex;
1553             if (((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */)
1554                 && (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
1555                 const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend;
1556                 matchLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4;
1557                 if (depth==0) goto _storeSequence;
1558             }
1559         }
1560         if ( dictMode == ZSTD_noDict
1561           && ((offset_1 > 0) & (MEM_read32(ip+1-offset_1) == MEM_read32(ip+1)))) {
1562             matchLength = ZSTD_count(ip+1+4, ip+1+4-offset_1, iend) + 4;
1563             if (depth==0) goto _storeSequence;
1564         }
1565 
1566         /* first search (depth 0) */
1567         {   size_t offsetFound = 999999999;
1568             size_t const ml2 = searchMax(ms, ip, iend, &offsetFound);
1569             if (ml2 > matchLength)
1570                 matchLength = ml2, start = ip, offcode=offsetFound;
1571         }
1572 
1573         if (matchLength < 4) {
1574             ip += ((ip-anchor) >> kSearchStrength) + 1;   /* jump faster over incompressible sections */
1575             continue;
1576         }
1577 
1578         /* let's try to find a better solution */
1579         if (depth>=1)
1580         while (ip<ilimit) {
1581             DEBUGLOG(7, "search depth 1");
1582             ip ++;
1583             if ( (dictMode == ZSTD_noDict)
1584               && (offcode) && ((offset_1>0) & (MEM_read32(ip) == MEM_read32(ip - offset_1)))) {
1585                 size_t const mlRep = ZSTD_count(ip+4, ip+4-offset_1, iend) + 4;
1586                 int const gain2 = (int)(mlRep * 3);
1587                 int const gain1 = (int)(matchLength*3 - ZSTD_highbit32((U32)STORED_TO_OFFBASE(offcode)) + 1);
1588                 if ((mlRep >= 4) && (gain2 > gain1))
1589                     matchLength = mlRep, offcode = STORE_REPCODE_1, start = ip;
1590             }
1591             if (isDxS) {
1592                 const U32 repIndex = (U32)(ip - base) - offset_1;
1593                 const BYTE* repMatch = repIndex < prefixLowestIndex ?
1594                                dictBase + (repIndex - dictIndexDelta) :
1595                                base + repIndex;
1596                 if (((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */)
1597                     && (MEM_read32(repMatch) == MEM_read32(ip)) ) {
1598                     const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend;
1599                     size_t const mlRep = ZSTD_count_2segments(ip+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4;
1600                     int const gain2 = (int)(mlRep * 3);
1601                     int const gain1 = (int)(matchLength*3 - ZSTD_highbit32((U32)STORED_TO_OFFBASE(offcode)) + 1);
1602                     if ((mlRep >= 4) && (gain2 > gain1))
1603                         matchLength = mlRep, offcode = STORE_REPCODE_1, start = ip;
1604                 }
1605             }
1606             {   size_t offset2=999999999;
1607                 size_t const ml2 = searchMax(ms, ip, iend, &offset2);
1608                 int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)STORED_TO_OFFBASE(offset2)));   /* raw approx */
1609                 int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)STORED_TO_OFFBASE(offcode)) + 4);
1610                 if ((ml2 >= 4) && (gain2 > gain1)) {
1611                     matchLength = ml2, offcode = offset2, start = ip;
1612                     continue;   /* search a better one */
1613             }   }
1614 
1615             /* let's find an even better one */
1616             if ((depth==2) && (ip<ilimit)) {
1617                 DEBUGLOG(7, "search depth 2");
1618                 ip ++;
1619                 if ( (dictMode == ZSTD_noDict)
1620                   && (offcode) && ((offset_1>0) & (MEM_read32(ip) == MEM_read32(ip - offset_1)))) {
1621                     size_t const mlRep = ZSTD_count(ip+4, ip+4-offset_1, iend) + 4;
1622                     int const gain2 = (int)(mlRep * 4);
1623                     int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)STORED_TO_OFFBASE(offcode)) + 1);
1624                     if ((mlRep >= 4) && (gain2 > gain1))
1625                         matchLength = mlRep, offcode = STORE_REPCODE_1, start = ip;
1626                 }
1627                 if (isDxS) {
1628                     const U32 repIndex = (U32)(ip - base) - offset_1;
1629                     const BYTE* repMatch = repIndex < prefixLowestIndex ?
1630                                    dictBase + (repIndex - dictIndexDelta) :
1631                                    base + repIndex;
1632                     if (((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */)
1633                         && (MEM_read32(repMatch) == MEM_read32(ip)) ) {
1634                         const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend;
1635                         size_t const mlRep = ZSTD_count_2segments(ip+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4;
1636                         int const gain2 = (int)(mlRep * 4);
1637                         int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)STORED_TO_OFFBASE(offcode)) + 1);
1638                         if ((mlRep >= 4) && (gain2 > gain1))
1639                             matchLength = mlRep, offcode = STORE_REPCODE_1, start = ip;
1640                     }
1641                 }
1642                 {   size_t offset2=999999999;
1643                     size_t const ml2 = searchMax(ms, ip, iend, &offset2);
1644                     int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)STORED_TO_OFFBASE(offset2)));   /* raw approx */
1645                     int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)STORED_TO_OFFBASE(offcode)) + 7);
1646                     if ((ml2 >= 4) && (gain2 > gain1)) {
1647                         matchLength = ml2, offcode = offset2, start = ip;
1648                         continue;
1649             }   }   }
1650             break;  /* nothing found : store previous solution */
1651         }
1652 
1653         /* NOTE:
1654          * Pay attention that `start[-value]` can lead to strange undefined behavior
1655          * notably if `value` is unsigned, resulting in a large positive `-value`.
1656          */
1657         /* catch up */
1658         if (STORED_IS_OFFSET(offcode)) {
1659             if (dictMode == ZSTD_noDict) {
1660                 while ( ((start > anchor) & (start - STORED_OFFSET(offcode) > prefixLowest))
1661                      && (start[-1] == (start-STORED_OFFSET(offcode))[-1]) )  /* only search for offset within prefix */
1662                     { start--; matchLength++; }
1663             }
1664             if (isDxS) {
1665                 U32 const matchIndex = (U32)((size_t)(start-base) - STORED_OFFSET(offcode));
1666                 const BYTE* match = (matchIndex < prefixLowestIndex) ? dictBase + matchIndex - dictIndexDelta : base + matchIndex;
1667                 const BYTE* const mStart = (matchIndex < prefixLowestIndex) ? dictLowest : prefixLowest;
1668                 while ((start>anchor) && (match>mStart) && (start[-1] == match[-1])) { start--; match--; matchLength++; }  /* catch up */
1669             }
1670             offset_2 = offset_1; offset_1 = (U32)STORED_OFFSET(offcode);
1671         }
1672         /* store sequence */
1673 _storeSequence:
1674         {   size_t const litLength = (size_t)(start - anchor);
1675             ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offcode, matchLength);
1676             anchor = ip = start + matchLength;
1677         }
1678 
1679         /* check immediate repcode */
1680         if (isDxS) {
1681             while (ip <= ilimit) {
1682                 U32 const current2 = (U32)(ip-base);
1683                 U32 const repIndex = current2 - offset_2;
1684                 const BYTE* repMatch = repIndex < prefixLowestIndex ?
1685                         dictBase - dictIndexDelta + repIndex :
1686                         base + repIndex;
1687                 if ( ((U32)((prefixLowestIndex-1) - (U32)repIndex) >= 3 /* intentional overflow */)
1688                    && (MEM_read32(repMatch) == MEM_read32(ip)) ) {
1689                     const BYTE* const repEnd2 = repIndex < prefixLowestIndex ? dictEnd : iend;
1690                     matchLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd2, prefixLowest) + 4;
1691                     offcode = offset_2; offset_2 = offset_1; offset_1 = (U32)offcode;   /* swap offset_2 <=> offset_1 */
1692                     ZSTD_storeSeq(seqStore, 0, anchor, iend, STORE_REPCODE_1, matchLength);
1693                     ip += matchLength;
1694                     anchor = ip;
1695                     continue;
1696                 }
1697                 break;
1698             }
1699         }
1700 
1701         if (dictMode == ZSTD_noDict) {
1702             while ( ((ip <= ilimit) & (offset_2>0))
1703                  && (MEM_read32(ip) == MEM_read32(ip - offset_2)) ) {
1704                 /* store sequence */
1705                 matchLength = ZSTD_count(ip+4, ip+4-offset_2, iend) + 4;
1706                 offcode = offset_2; offset_2 = offset_1; offset_1 = (U32)offcode; /* swap repcodes */
1707                 ZSTD_storeSeq(seqStore, 0, anchor, iend, STORE_REPCODE_1, matchLength);
1708                 ip += matchLength;
1709                 anchor = ip;
1710                 continue;   /* faster when present ... (?) */
1711     }   }   }
1712 
1713     /* Save reps for next block */
1714     rep[0] = offset_1 ? offset_1 : savedOffset;
1715     rep[1] = offset_2 ? offset_2 : savedOffset;
1716 
1717     /* Return the last literals size */
1718     return (size_t)(iend - anchor);
1719 }
1720 
1721 
1722 size_t ZSTD_compressBlock_btlazy2(
1723         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1724         void const* src, size_t srcSize)
1725 {
1726     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2, ZSTD_noDict);
1727 }
1728 
1729 size_t ZSTD_compressBlock_lazy2(
1730         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1731         void const* src, size_t srcSize)
1732 {
1733     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_noDict);
1734 }
1735 
1736 size_t ZSTD_compressBlock_lazy(
1737         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1738         void const* src, size_t srcSize)
1739 {
1740     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_noDict);
1741 }
1742 
1743 size_t ZSTD_compressBlock_greedy(
1744         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1745         void const* src, size_t srcSize)
1746 {
1747     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0, ZSTD_noDict);
1748 }
1749 
1750 size_t ZSTD_compressBlock_btlazy2_dictMatchState(
1751         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1752         void const* src, size_t srcSize)
1753 {
1754     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2, ZSTD_dictMatchState);
1755 }
1756 
1757 size_t ZSTD_compressBlock_lazy2_dictMatchState(
1758         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1759         void const* src, size_t srcSize)
1760 {
1761     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_dictMatchState);
1762 }
1763 
1764 size_t ZSTD_compressBlock_lazy_dictMatchState(
1765         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1766         void const* src, size_t srcSize)
1767 {
1768     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_dictMatchState);
1769 }
1770 
1771 size_t ZSTD_compressBlock_greedy_dictMatchState(
1772         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1773         void const* src, size_t srcSize)
1774 {
1775     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0, ZSTD_dictMatchState);
1776 }
1777 
1778 
1779 size_t ZSTD_compressBlock_lazy2_dedicatedDictSearch(
1780         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1781         void const* src, size_t srcSize)
1782 {
1783     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_dedicatedDictSearch);
1784 }
1785 
1786 size_t ZSTD_compressBlock_lazy_dedicatedDictSearch(
1787         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1788         void const* src, size_t srcSize)
1789 {
1790     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_dedicatedDictSearch);
1791 }
1792 
1793 size_t ZSTD_compressBlock_greedy_dedicatedDictSearch(
1794         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1795         void const* src, size_t srcSize)
1796 {
1797     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0, ZSTD_dedicatedDictSearch);
1798 }
1799 
1800 /* Row-based matchfinder */
1801 size_t ZSTD_compressBlock_lazy2_row(
1802         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1803         void const* src, size_t srcSize)
1804 {
1805     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 2, ZSTD_noDict);
1806 }
1807 
1808 size_t ZSTD_compressBlock_lazy_row(
1809         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1810         void const* src, size_t srcSize)
1811 {
1812     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 1, ZSTD_noDict);
1813 }
1814 
1815 size_t ZSTD_compressBlock_greedy_row(
1816         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1817         void const* src, size_t srcSize)
1818 {
1819     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 0, ZSTD_noDict);
1820 }
1821 
1822 size_t ZSTD_compressBlock_lazy2_dictMatchState_row(
1823         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1824         void const* src, size_t srcSize)
1825 {
1826     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 2, ZSTD_dictMatchState);
1827 }
1828 
1829 size_t ZSTD_compressBlock_lazy_dictMatchState_row(
1830         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1831         void const* src, size_t srcSize)
1832 {
1833     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 1, ZSTD_dictMatchState);
1834 }
1835 
1836 size_t ZSTD_compressBlock_greedy_dictMatchState_row(
1837         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1838         void const* src, size_t srcSize)
1839 {
1840     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 0, ZSTD_dictMatchState);
1841 }
1842 
1843 
1844 size_t ZSTD_compressBlock_lazy2_dedicatedDictSearch_row(
1845         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1846         void const* src, size_t srcSize)
1847 {
1848     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 2, ZSTD_dedicatedDictSearch);
1849 }
1850 
1851 size_t ZSTD_compressBlock_lazy_dedicatedDictSearch_row(
1852         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1853         void const* src, size_t srcSize)
1854 {
1855     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 1, ZSTD_dedicatedDictSearch);
1856 }
1857 
1858 size_t ZSTD_compressBlock_greedy_dedicatedDictSearch_row(
1859         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
1860         void const* src, size_t srcSize)
1861 {
1862     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 0, ZSTD_dedicatedDictSearch);
1863 }
1864 
1865 FORCE_INLINE_TEMPLATE
1866 size_t ZSTD_compressBlock_lazy_extDict_generic(
1867                         ZSTD_matchState_t* ms, seqStore_t* seqStore,
1868                         U32 rep[ZSTD_REP_NUM],
1869                         const void* src, size_t srcSize,
1870                         const searchMethod_e searchMethod, const U32 depth)
1871 {
1872     const BYTE* const istart = (const BYTE*)src;
1873     const BYTE* ip = istart;
1874     const BYTE* anchor = istart;
1875     const BYTE* const iend = istart + srcSize;
1876     const BYTE* const ilimit = searchMethod == search_rowHash ? iend - 8 - ZSTD_ROW_HASH_CACHE_SIZE : iend - 8;
1877     const BYTE* const base = ms->window.base;
1878     const U32 dictLimit = ms->window.dictLimit;
1879     const BYTE* const prefixStart = base + dictLimit;
1880     const BYTE* const dictBase = ms->window.dictBase;
1881     const BYTE* const dictEnd  = dictBase + dictLimit;
1882     const BYTE* const dictStart  = dictBase + ms->window.lowLimit;
1883     const U32 windowLog = ms->cParams.windowLog;
1884     const U32 rowLog = ms->cParams.searchLog < 5 ? 4 : 5;
1885 
1886     searchMax_f const searchMax = ZSTD_selectLazyVTable(ms, searchMethod, ZSTD_extDict)->searchMax;
1887     U32 offset_1 = rep[0], offset_2 = rep[1];
1888 
1889     DEBUGLOG(5, "ZSTD_compressBlock_lazy_extDict_generic (searchFunc=%u)", (U32)searchMethod);
1890 
1891     /* init */
1892     ip += (ip == prefixStart);
1893     if (searchMethod == search_rowHash) {
1894         ZSTD_row_fillHashCache(ms, base, rowLog,
1895                                MIN(ms->cParams.minMatch, 6 /* mls caps out at 6 */),
1896                                ms->nextToUpdate, ilimit);
1897     }
1898 
1899     /* Match Loop */
1900 #if defined(__GNUC__) && defined(__x86_64__)
1901     /* I've measured random a 5% speed loss on levels 5 & 6 (greedy) when the
1902      * code alignment is perturbed. To fix the instability align the loop on 32-bytes.
1903      */
1904     __asm__(".p2align 5");
1905 #endif
1906     while (ip < ilimit) {
1907         size_t matchLength=0;
1908         size_t offcode=STORE_REPCODE_1;
1909         const BYTE* start=ip+1;
1910         U32 curr = (U32)(ip-base);
1911 
1912         /* check repCode */
1913         {   const U32 windowLow = ZSTD_getLowestMatchIndex(ms, curr+1, windowLog);
1914             const U32 repIndex = (U32)(curr+1 - offset_1);
1915             const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
1916             const BYTE* const repMatch = repBase + repIndex;
1917             if ( ((U32)((dictLimit-1) - repIndex) >= 3) /* intentional overflow */
1918                & (offset_1 <= curr+1 - windowLow) ) /* note: we are searching at curr+1 */
1919             if (MEM_read32(ip+1) == MEM_read32(repMatch)) {
1920                 /* repcode detected we should take it */
1921                 const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
1922                 matchLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repEnd, prefixStart) + 4;
1923                 if (depth==0) goto _storeSequence;
1924         }   }
1925 
1926         /* first search (depth 0) */
1927         {   size_t offsetFound = 999999999;
1928             size_t const ml2 = searchMax(ms, ip, iend, &offsetFound);
1929             if (ml2 > matchLength)
1930                 matchLength = ml2, start = ip, offcode=offsetFound;
1931         }
1932 
1933         if (matchLength < 4) {
1934             ip += ((ip-anchor) >> kSearchStrength) + 1;   /* jump faster over incompressible sections */
1935             continue;
1936         }
1937 
1938         /* let's try to find a better solution */
1939         if (depth>=1)
1940         while (ip<ilimit) {
1941             ip ++;
1942             curr++;
1943             /* check repCode */
1944             if (offcode) {
1945                 const U32 windowLow = ZSTD_getLowestMatchIndex(ms, curr, windowLog);
1946                 const U32 repIndex = (U32)(curr - offset_1);
1947                 const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
1948                 const BYTE* const repMatch = repBase + repIndex;
1949                 if ( ((U32)((dictLimit-1) - repIndex) >= 3) /* intentional overflow : do not test positions overlapping 2 memory segments  */
1950                    & (offset_1 <= curr - windowLow) ) /* equivalent to `curr > repIndex >= windowLow` */
1951                 if (MEM_read32(ip) == MEM_read32(repMatch)) {
1952                     /* repcode detected */
1953                     const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
1954                     size_t const repLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4;
1955                     int const gain2 = (int)(repLength * 3);
1956                     int const gain1 = (int)(matchLength*3 - ZSTD_highbit32((U32)STORED_TO_OFFBASE(offcode)) + 1);
1957                     if ((repLength >= 4) && (gain2 > gain1))
1958                         matchLength = repLength, offcode = STORE_REPCODE_1, start = ip;
1959             }   }
1960 
1961             /* search match, depth 1 */
1962             {   size_t offset2=999999999;
1963                 size_t const ml2 = searchMax(ms, ip, iend, &offset2);
1964                 int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)STORED_TO_OFFBASE(offset2)));   /* raw approx */
1965                 int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)STORED_TO_OFFBASE(offcode)) + 4);
1966                 if ((ml2 >= 4) && (gain2 > gain1)) {
1967                     matchLength = ml2, offcode = offset2, start = ip;
1968                     continue;   /* search a better one */
1969             }   }
1970 
1971             /* let's find an even better one */
1972             if ((depth==2) && (ip<ilimit)) {
1973                 ip ++;
1974                 curr++;
1975                 /* check repCode */
1976                 if (offcode) {
1977                     const U32 windowLow = ZSTD_getLowestMatchIndex(ms, curr, windowLog);
1978                     const U32 repIndex = (U32)(curr - offset_1);
1979                     const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
1980                     const BYTE* const repMatch = repBase + repIndex;
1981                     if ( ((U32)((dictLimit-1) - repIndex) >= 3) /* intentional overflow : do not test positions overlapping 2 memory segments  */
1982                        & (offset_1 <= curr - windowLow) ) /* equivalent to `curr > repIndex >= windowLow` */
1983                     if (MEM_read32(ip) == MEM_read32(repMatch)) {
1984                         /* repcode detected */
1985                         const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
1986                         size_t const repLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4;
1987                         int const gain2 = (int)(repLength * 4);
1988                         int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)STORED_TO_OFFBASE(offcode)) + 1);
1989                         if ((repLength >= 4) && (gain2 > gain1))
1990                             matchLength = repLength, offcode = STORE_REPCODE_1, start = ip;
1991                 }   }
1992 
1993                 /* search match, depth 2 */
1994                 {   size_t offset2=999999999;
1995                     size_t const ml2 = searchMax(ms, ip, iend, &offset2);
1996                     int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)STORED_TO_OFFBASE(offset2)));   /* raw approx */
1997                     int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)STORED_TO_OFFBASE(offcode)) + 7);
1998                     if ((ml2 >= 4) && (gain2 > gain1)) {
1999                         matchLength = ml2, offcode = offset2, start = ip;
2000                         continue;
2001             }   }   }
2002             break;  /* nothing found : store previous solution */
2003         }
2004 
2005         /* catch up */
2006         if (STORED_IS_OFFSET(offcode)) {
2007             U32 const matchIndex = (U32)((size_t)(start-base) - STORED_OFFSET(offcode));
2008             const BYTE* match = (matchIndex < dictLimit) ? dictBase + matchIndex : base + matchIndex;
2009             const BYTE* const mStart = (matchIndex < dictLimit) ? dictStart : prefixStart;
2010             while ((start>anchor) && (match>mStart) && (start[-1] == match[-1])) { start--; match--; matchLength++; }  /* catch up */
2011             offset_2 = offset_1; offset_1 = (U32)STORED_OFFSET(offcode);
2012         }
2013 
2014         /* store sequence */
2015 _storeSequence:
2016         {   size_t const litLength = (size_t)(start - anchor);
2017             ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offcode, matchLength);
2018             anchor = ip = start + matchLength;
2019         }
2020 
2021         /* check immediate repcode */
2022         while (ip <= ilimit) {
2023             const U32 repCurrent = (U32)(ip-base);
2024             const U32 windowLow = ZSTD_getLowestMatchIndex(ms, repCurrent, windowLog);
2025             const U32 repIndex = repCurrent - offset_2;
2026             const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
2027             const BYTE* const repMatch = repBase + repIndex;
2028             if ( ((U32)((dictLimit-1) - repIndex) >= 3) /* intentional overflow : do not test positions overlapping 2 memory segments  */
2029                & (offset_2 <= repCurrent - windowLow) ) /* equivalent to `curr > repIndex >= windowLow` */
2030             if (MEM_read32(ip) == MEM_read32(repMatch)) {
2031                 /* repcode detected we should take it */
2032                 const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
2033                 matchLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4;
2034                 offcode = offset_2; offset_2 = offset_1; offset_1 = (U32)offcode;   /* swap offset history */
2035                 ZSTD_storeSeq(seqStore, 0, anchor, iend, STORE_REPCODE_1, matchLength);
2036                 ip += matchLength;
2037                 anchor = ip;
2038                 continue;   /* faster when present ... (?) */
2039             }
2040             break;
2041     }   }
2042 
2043     /* Save reps for next block */
2044     rep[0] = offset_1;
2045     rep[1] = offset_2;
2046 
2047     /* Return the last literals size */
2048     return (size_t)(iend - anchor);
2049 }
2050 
2051 
2052 size_t ZSTD_compressBlock_greedy_extDict(
2053         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
2054         void const* src, size_t srcSize)
2055 {
2056     return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0);
2057 }
2058 
2059 size_t ZSTD_compressBlock_lazy_extDict(
2060         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
2061         void const* src, size_t srcSize)
2062 
2063 {
2064     return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1);
2065 }
2066 
2067 size_t ZSTD_compressBlock_lazy2_extDict(
2068         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
2069         void const* src, size_t srcSize)
2070 
2071 {
2072     return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2);
2073 }
2074 
2075 size_t ZSTD_compressBlock_btlazy2_extDict(
2076         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
2077         void const* src, size_t srcSize)
2078 
2079 {
2080     return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2);
2081 }
2082 
2083 size_t ZSTD_compressBlock_greedy_extDict_row(
2084         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
2085         void const* src, size_t srcSize)
2086 {
2087     return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 0);
2088 }
2089 
2090 size_t ZSTD_compressBlock_lazy_extDict_row(
2091         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
2092         void const* src, size_t srcSize)
2093 
2094 {
2095     return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 1);
2096 }
2097 
2098 size_t ZSTD_compressBlock_lazy2_extDict_row(
2099         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
2100         void const* src, size_t srcSize)
2101 
2102 {
2103     return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 2);
2104 }
2105