contrib/xz/src/liblzma/lzma/lzma_decoder.c

   1 ///////////////////////////////////////////////////////////////////////////////
   2 //
   3 /// \file       lzma_decoder.c
   4 /// \brief      LZMA decoder
   5 ///
   6 //  Authors:    Igor Pavlov
   7 //              Lasse Collin
   8 //
   9 //  This file has been put into the public domain.
  10 //  You can do whatever you want with this file.
  11 //
  12 ///////////////////////////////////////////////////////////////////////////////
  13
  14 #include "lz_decoder.h"
  15 #include "lzma_common.h"
  16 #include "lzma_decoder.h"
  17 #include "range_decoder.h"
  18
  19
  20 #ifdef HAVE_SMALL
  21
  22 // Macros for (somewhat) size-optimized code.
  23 #define seq_4(seq) seq
  24
  25 #define seq_6(seq) seq
  26
  27 #define seq_8(seq) seq
  28
  29 #define seq_len(seq) \
  30         seq ## _CHOICE, \
  31         seq ## _CHOICE2, \
  32         seq ## _BITTREE
  33
  34 #define len_decode(target, ld, pos_state, seq) \
  35 do { \
  36 case seq ## _CHOICE: \
  37         rc_if_0(ld.choice, seq ## _CHOICE) { \
  38                 rc_update_0(ld.choice); \
  39                 probs = ld.low[pos_state];\
  40                 limit = LEN_LOW_SYMBOLS; \
  41                 target = MATCH_LEN_MIN; \
  42         } else { \
  43                 rc_update_1(ld.choice); \
  44 case seq ## _CHOICE2: \
  45                 rc_if_0(ld.choice2, seq ## _CHOICE2) { \
  46                         rc_update_0(ld.choice2); \
  47                         probs = ld.mid[pos_state]; \
  48                         limit = LEN_MID_SYMBOLS; \
  49                         target = MATCH_LEN_MIN + LEN_LOW_SYMBOLS; \
  50                 } else { \
  51                         rc_update_1(ld.choice2); \
  52                         probs = ld.high; \
  53                         limit = LEN_HIGH_SYMBOLS; \
  54                         target = MATCH_LEN_MIN + LEN_LOW_SYMBOLS \
  55                                         + LEN_MID_SYMBOLS; \
  56                 } \
  57         } \
  58         symbol = 1; \
  59 case seq ## _BITTREE: \
  60         do { \
  61                 rc_bit(probs[symbol], , , seq ## _BITTREE); \
  62         } while (symbol < limit); \
  63         target += symbol - limit; \
  64 } while (0)
  65
  66 #else // HAVE_SMALL
  67
  68 // Unrolled versions
  69 #define seq_4(seq) \
  70         seq ## 0, \
  71         seq ## 1, \
  72         seq ## 2, \
  73         seq ## 3
  74
  75 #define seq_6(seq) \
  76         seq ## 0, \
  77         seq ## 1, \
  78         seq ## 2, \
  79         seq ## 3, \
  80         seq ## 4, \
  81         seq ## 5
  82
  83 #define seq_8(seq) \
  84         seq ## 0, \
  85         seq ## 1, \
  86         seq ## 2, \
  87         seq ## 3, \
  88         seq ## 4, \
  89         seq ## 5, \
  90         seq ## 6, \
  91         seq ## 7
  92
  93 #define seq_len(seq) \
  94         seq ## _CHOICE, \
  95         seq ## _LOW0, \
  96         seq ## _LOW1, \
  97         seq ## _LOW2, \
  98         seq ## _CHOICE2, \
  99         seq ## _MID0, \
 100         seq ## _MID1, \
 101         seq ## _MID2, \
 102         seq ## _HIGH0, \
 103         seq ## _HIGH1, \
 104         seq ## _HIGH2, \
 105         seq ## _HIGH3, \
 106         seq ## _HIGH4, \
 107         seq ## _HIGH5, \
 108         seq ## _HIGH6, \
 109         seq ## _HIGH7
 110
 111 #define len_decode(target, ld, pos_state, seq) \
 112 do { \
 113         symbol = 1; \
 114 case seq ## _CHOICE: \
 115         rc_if_0(ld.choice, seq ## _CHOICE) { \
 116                 rc_update_0(ld.choice); \
 117                 rc_bit_case(ld.low[pos_state][symbol], , , seq ## _LOW0); \
 118                 rc_bit_case(ld.low[pos_state][symbol], , , seq ## _LOW1); \
 119                 rc_bit_case(ld.low[pos_state][symbol], , , seq ## _LOW2); \
 120                 target = symbol - LEN_LOW_SYMBOLS + MATCH_LEN_MIN; \
 121         } else { \
 122                 rc_update_1(ld.choice); \
 123 case seq ## _CHOICE2: \
 124                 rc_if_0(ld.choice2, seq ## _CHOICE2) { \
 125                         rc_update_0(ld.choice2); \
 126                         rc_bit_case(ld.mid[pos_state][symbol], , , \
 127                                         seq ## _MID0); \
 128                         rc_bit_case(ld.mid[pos_state][symbol], , , \
 129                                         seq ## _MID1); \
 130                         rc_bit_case(ld.mid[pos_state][symbol], , , \
 131                                         seq ## _MID2); \
 132                         target = symbol - LEN_MID_SYMBOLS \
 133                                         + MATCH_LEN_MIN + LEN_LOW_SYMBOLS; \
 134                 } else { \
 135                         rc_update_1(ld.choice2); \
 136                         rc_bit_case(ld.high[symbol], , , seq ## _HIGH0); \
 137                         rc_bit_case(ld.high[symbol], , , seq ## _HIGH1); \
 138                         rc_bit_case(ld.high[symbol], , , seq ## _HIGH2); \
 139                         rc_bit_case(ld.high[symbol], , , seq ## _HIGH3); \
 140                         rc_bit_case(ld.high[symbol], , , seq ## _HIGH4); \
 141                         rc_bit_case(ld.high[symbol], , , seq ## _HIGH5); \
 142                         rc_bit_case(ld.high[symbol], , , seq ## _HIGH6); \
 143                         rc_bit_case(ld.high[symbol], , , seq ## _HIGH7); \
 144                         target = symbol - LEN_HIGH_SYMBOLS \
 145                                         + MATCH_LEN_MIN \
 146                                         + LEN_LOW_SYMBOLS + LEN_MID_SYMBOLS; \
 147                 } \
 148         } \
 149 } while (0)
 150
 151 #endif // HAVE_SMALL
 152
 153
 154 /// Length decoder probabilities; see comments in lzma_common.h.
 155 typedef struct {
 156         probability choice;
 157         probability choice2;
 158         probability low[POS_STATES_MAX][LEN_LOW_SYMBOLS];
 159         probability mid[POS_STATES_MAX][LEN_MID_SYMBOLS];
 160         probability high[LEN_HIGH_SYMBOLS];
 161 } lzma_length_decoder;
 162
 163
 164 struct lzma_coder_s {
 165         ///////////////////
 166         // Probabilities //
 167         ///////////////////
 168
 169         /// Literals; see comments in lzma_common.h.
 170         probability literal[LITERAL_CODERS_MAX][LITERAL_CODER_SIZE];
 171
 172         /// If 1, it's a match. Otherwise it's a single 8-bit literal.
 173         probability is_match[STATES][POS_STATES_MAX];
 174
 175         /// If 1, it's a repeated match. The distance is one of rep0 .. rep3.
 176         probability is_rep[STATES];
 177
 178         /// If 0, distance of a repeated match is rep0.
 179         /// Otherwise check is_rep1.
 180         probability is_rep0[STATES];
 181
 182         /// If 0, distance of a repeated match is rep1.
 183         /// Otherwise check is_rep2.
 184         probability is_rep1[STATES];
 185
 186         /// If 0, distance of a repeated match is rep2. Otherwise it is rep3.
 187         probability is_rep2[STATES];
 188
 189         /// If 1, the repeated match has length of one byte. Otherwise
 190         /// the length is decoded from rep_len_decoder.
 191         probability is_rep0_long[STATES][POS_STATES_MAX];
 192
 193         /// Probability tree for the highest two bits of the match distance.
 194         /// There is a separate probability tree for match lengths of
 195         /// 2 (i.e. MATCH_LEN_MIN), 3, 4, and [5, 273].
 196         probability dist_slot[DIST_STATES][DIST_SLOTS];
 197
 198         /// Probability trees for additional bits for match distance when the
 199         /// distance is in the range [4, 127].
 200         probability pos_special[FULL_DISTANCES - DIST_MODEL_END];
 201
 202         /// Probability tree for the lowest four bits of a match distance
 203         /// that is equal to or greater than 128.
 204         probability pos_align[ALIGN_SIZE];
 205
 206         /// Length of a normal match
 207         lzma_length_decoder match_len_decoder;
 208
 209         /// Length of a repeated match
 210         lzma_length_decoder rep_len_decoder;
 211
 212         ///////////////////
 213         // Decoder state //
 214         ///////////////////
 215
 216         // Range coder
 217         lzma_range_decoder rc;
 218
 219         // Types of the most recently seen LZMA symbols
 220         lzma_lzma_state state;
 221
 222         uint32_t rep0;      ///< Distance of the latest match
 223         uint32_t rep1;      ///< Distance of second latest match
 224         uint32_t rep2;      ///< Distance of third latest match
 225         uint32_t rep3;      ///< Distance of fourth latest match
 226
 227         uint32_t pos_mask; // (1U << pb) - 1
 228         uint32_t literal_context_bits;
 229         uint32_t literal_pos_mask;
 230
 231         /// Uncompressed size as bytes, or LZMA_VLI_UNKNOWN if end of
 232         /// payload marker is expected.
 233         lzma_vli uncompressed_size;
 234
 235         ////////////////////////////////
 236         // State of incomplete symbol //
 237         ////////////////////////////////
 238
 239         /// Position where to continue the decoder loop
 240         enum {
 241                 SEQ_NORMALIZE,
 242                 SEQ_IS_MATCH,
 243                 seq_8(SEQ_LITERAL),
 244                 seq_8(SEQ_LITERAL_MATCHED),
 245                 SEQ_LITERAL_WRITE,
 246                 SEQ_IS_REP,
 247                 seq_len(SEQ_MATCH_LEN),
 248                 seq_6(SEQ_DIST_SLOT),
 249                 SEQ_DIST_MODEL,
 250                 SEQ_DIRECT,
 251                 seq_4(SEQ_ALIGN),
 252                 SEQ_EOPM,
 253                 SEQ_IS_REP0,
 254                 SEQ_SHORTREP,
 255                 SEQ_IS_REP0_LONG,
 256                 SEQ_IS_REP1,
 257                 SEQ_IS_REP2,
 258                 seq_len(SEQ_REP_LEN),
 259                 SEQ_COPY,
 260         } sequence;
 261
 262         /// Base of the current probability tree
 263         probability *probs;
 264
 265         /// Symbol being decoded. This is also used as an index variable in
 266         /// bittree decoders: probs[symbol]
 267         uint32_t symbol;
 268
 269         /// Used as a loop termination condition on bittree decoders and
 270         /// direct bits decoder.
 271         uint32_t limit;
 272
 273         /// Matched literal decoder: 0x100 or 0 to help avoiding branches.
 274         /// Bittree reverse decoders: Offset of the next bit: 1 << offset
 275         uint32_t offset;
 276
 277         /// If decoding a literal: match byte.
 278         /// If decoding a match: length of the match.
 279         uint32_t len;
 280 };
 281
 282
 283 static lzma_ret
 284 lzma_decode(lzma_coder *restrict coder, lzma_dict *restrict dictptr,
 285                 const uint8_t *restrict in,
 286                 size_t *restrict in_pos, size_t in_size)
 287 {
 288         ////////////////////
 289         // Initialization //
 290         ////////////////////
 291
 292         {
 293                 const lzma_ret ret = rc_read_init(
 294                                 &coder->rc, in, in_pos, in_size);
 295                 if (ret != LZMA_STREAM_END)
 296                         return ret;
 297         }
 298
 299         ///////////////
 300         // Variables //
 301         ///////////////
 302
 303         // Making local copies of often-used variables improves both
 304         // speed and readability.
 305
 306         lzma_dict dict = *dictptr;
 307
 308         const size_t dict_start = dict.pos;
 309
 310         // Range decoder
 311         rc_to_local(coder->rc, *in_pos);
 312
 313         // State
 314         uint32_t state = coder->state;
 315         uint32_t rep0 = coder->rep0;
 316         uint32_t rep1 = coder->rep1;
 317         uint32_t rep2 = coder->rep2;
 318         uint32_t rep3 = coder->rep3;
 319
 320         const uint32_t pos_mask = coder->pos_mask;
 321
 322         // These variables are actually needed only if we last time ran
 323         // out of input in the middle of the decoder loop.
 324         probability *probs = coder->probs;
 325         uint32_t symbol = coder->symbol;
 326         uint32_t limit = coder->limit;
 327         uint32_t offset = coder->offset;
 328         uint32_t len = coder->len;
 329
 330         const uint32_t literal_pos_mask = coder->literal_pos_mask;
 331         const uint32_t literal_context_bits = coder->literal_context_bits;
 332
 333         // Temporary variables
 334         uint32_t pos_state = dict.pos & pos_mask;
 335
 336         lzma_ret ret = LZMA_OK;
 337
 338         // If uncompressed size is known, there must be no end of payload
 339         // marker.
 340         const bool no_eopm = coder->uncompressed_size
 341                         != LZMA_VLI_UNKNOWN;
 342         if (no_eopm && coder->uncompressed_size < dict.limit - dict.pos)
 343                 dict.limit = dict.pos + (size_t)(coder->uncompressed_size);
 344
 345         // The main decoder loop. The "switch" is used to restart the decoder at
 346         // correct location. Once restarted, the "switch" is no longer used.
 347         switch (coder->sequence)
 348         while (true) {
 349                 // Calculate new pos_state. This is skipped on the first loop
 350                 // since we already calculated it when setting up the local
 351                 // variables.
 352                 pos_state = dict.pos & pos_mask;
 353
 354         case SEQ_NORMALIZE:
 355         case SEQ_IS_MATCH:
 356                 if (unlikely(no_eopm && dict.pos == dict.limit))
 357                         break;
 358
 359                 rc_if_0(coder->is_match[state][pos_state], SEQ_IS_MATCH) {
 360                         rc_update_0(coder->is_match[state][pos_state]);
 361
 362                         // It's a literal i.e. a single 8-bit byte.
 363
 364                         probs = literal_subcoder(coder->literal,
 365                                         literal_context_bits, literal_pos_mask,
 366                                         dict.pos, dict_get(&dict, 0));
 367                         symbol = 1;
 368
 369                         if (is_literal_state(state)) {
 370                                 // Decode literal without match byte.
 371 #ifdef HAVE_SMALL
 372         case SEQ_LITERAL:
 373                                 do {
 374                                         rc_bit(probs[symbol], , , SEQ_LITERAL);
 375                                 } while (symbol < (1 << 8));
 376 #else
 377                                 rc_bit_case(probs[symbol], , , SEQ_LITERAL0);
 378                                 rc_bit_case(probs[symbol], , , SEQ_LITERAL1);
 379                                 rc_bit_case(probs[symbol], , , SEQ_LITERAL2);
 380                                 rc_bit_case(probs[symbol], , , SEQ_LITERAL3);
 381                                 rc_bit_case(probs[symbol], , , SEQ_LITERAL4);
 382                                 rc_bit_case(probs[symbol], , , SEQ_LITERAL5);
 383                                 rc_bit_case(probs[symbol], , , SEQ_LITERAL6);
 384                                 rc_bit_case(probs[symbol], , , SEQ_LITERAL7);
 385 #endif
 386                         } else {
 387                                 // Decode literal with match byte.
 388                                 //
 389                                 // We store the byte we compare against
 390                                 // ("match byte") to "len" to minimize the
 391                                 // number of variables we need to store
 392                                 // between decoder calls.
 393                                 len = dict_get(&dict, rep0) << 1;
 394
 395                                 // The usage of "offset" allows omitting some
 396                                 // branches, which should give tiny speed
 397                                 // improvement on some CPUs. "offset" gets
 398                                 // set to zero if match_bit didn't match.
 399                                 offset = 0x100;
 400
 401 #ifdef HAVE_SMALL
 402         case SEQ_LITERAL_MATCHED:
 403                                 do {
 404                                         const uint32_t match_bit
 405                                                         = len & offset;
 406                                         const uint32_t subcoder_index
 407                                                         = offset + match_bit
 408                                                         + symbol;
 409
 410                                         rc_bit(probs[subcoder_index],
 411                                                         offset &= ~match_bit,
 412                                                         offset &= match_bit,
 413                                                         SEQ_LITERAL_MATCHED);
 414
 415                                         // It seems to be faster to do this
 416                                         // here instead of putting it to the
 417                                         // beginning of the loop and then
 418                                         // putting the "case" in the middle
 419                                         // of the loop.
 420                                         len <<= 1;
 421
 422                                 } while (symbol < (1 << 8));
 423 #else
 424                                 // Unroll the loop.
 425                                 uint32_t match_bit;
 426                                 uint32_t subcoder_index;
 427
 428 #       define d(seq) \
 429                 case seq: \
 430                         match_bit = len & offset; \
 431                         subcoder_index = offset + match_bit + symbol; \
 432                         rc_bit(probs[subcoder_index], \
 433                                         offset &= ~match_bit, \
 434                                         offset &= match_bit, \
 435                                         seq)
 436
 437                                 d(SEQ_LITERAL_MATCHED0);
 438                                 len <<= 1;
 439                                 d(SEQ_LITERAL_MATCHED1);
 440                                 len <<= 1;
 441                                 d(SEQ_LITERAL_MATCHED2);
 442                                 len <<= 1;
 443                                 d(SEQ_LITERAL_MATCHED3);
 444                                 len <<= 1;
 445                                 d(SEQ_LITERAL_MATCHED4);
 446                                 len <<= 1;
 447                                 d(SEQ_LITERAL_MATCHED5);
 448                                 len <<= 1;
 449                                 d(SEQ_LITERAL_MATCHED6);
 450                                 len <<= 1;
 451                                 d(SEQ_LITERAL_MATCHED7);
 452 #       undef d
 453 #endif
 454                         }
 455
 456                         //update_literal(state);
 457                         // Use a lookup table to update to literal state,
 458                         // since compared to other state updates, this would
 459                         // need two branches.
 460                         static const lzma_lzma_state next_state[] = {
 461                                 STATE_LIT_LIT,
 462                                 STATE_LIT_LIT,
 463                                 STATE_LIT_LIT,
 464                                 STATE_LIT_LIT,
 465                                 STATE_MATCH_LIT_LIT,
 466                                 STATE_REP_LIT_LIT,
 467                                 STATE_SHORTREP_LIT_LIT,
 468                                 STATE_MATCH_LIT,
 469                                 STATE_REP_LIT,
 470                                 STATE_SHORTREP_LIT,
 471                                 STATE_MATCH_LIT,
 472                                 STATE_REP_LIT
 473                         };
 474                         state = next_state[state];
 475
 476         case SEQ_LITERAL_WRITE:
 477                         if (unlikely(dict_put(&dict, symbol))) {
 478                                 coder->sequence = SEQ_LITERAL_WRITE;
 479                                 goto out;
 480                         }
 481
 482                         continue;
 483                 }
 484
 485                 // Instead of a new byte we are going to get a byte range
 486                 // (distance and length) which will be repeated from our
 487                 // output history.
 488
 489                 rc_update_1(coder->is_match[state][pos_state]);
 490
 491         case SEQ_IS_REP:
 492                 rc_if_0(coder->is_rep[state], SEQ_IS_REP) {
 493                         // Not a repeated match
 494                         rc_update_0(coder->is_rep[state]);
 495                         update_match(state);
 496
 497                         // The latest three match distances are kept in
 498                         // memory in case there are repeated matches.
 499                         rep3 = rep2;
 500                         rep2 = rep1;
 501                         rep1 = rep0;
 502
 503                         // Decode the length of the match.
 504                         len_decode(len, coder->match_len_decoder,
 505                                         pos_state, SEQ_MATCH_LEN);
 506
 507                         // Prepare to decode the highest two bits of the
 508                         // match distance.
 509                         probs = coder->dist_slot[get_dist_state(len)];
 510                         symbol = 1;
 511
 512 #ifdef HAVE_SMALL
 513         case SEQ_DIST_SLOT:
 514                         do {
 515                                 rc_bit(probs[symbol], , , SEQ_DIST_SLOT);
 516                         } while (symbol < DIST_SLOTS);
 517 #else
 518                         rc_bit_case(probs[symbol], , , SEQ_DIST_SLOT0);
 519                         rc_bit_case(probs[symbol], , , SEQ_DIST_SLOT1);
 520                         rc_bit_case(probs[symbol], , , SEQ_DIST_SLOT2);
 521                         rc_bit_case(probs[symbol], , , SEQ_DIST_SLOT3);
 522                         rc_bit_case(probs[symbol], , , SEQ_DIST_SLOT4);
 523                         rc_bit_case(probs[symbol], , , SEQ_DIST_SLOT5);
 524 #endif
 525                         // Get rid of the highest bit that was needed for
 526                         // indexing of the probability array.
 527                         symbol -= DIST_SLOTS;
 528                         assert(symbol <= 63);
 529
 530                         if (symbol < DIST_MODEL_START) {
 531                                 // Match distances [0, 3] have only two bits.
 532                                 rep0 = symbol;
 533                         } else {
 534                                 // Decode the lowest [1, 29] bits of
 535                                 // the match distance.
 536                                 limit = (symbol >> 1) - 1;
 537                                 assert(limit >= 1 && limit <= 30);
 538                                 rep0 = 2 + (symbol & 1);
 539
 540                                 if (symbol < DIST_MODEL_END) {
 541                                         // Prepare to decode the low bits for
 542                                         // a distance of [4, 127].
 543                                         assert(limit <= 5);
 544                                         rep0 <<= limit;
 545                                         assert(rep0 <= 96);
 546                                         // -1 is fine, because we start
 547                                         // decoding at probs[1], not probs[0].
 548                                         // NOTE: This violates the C standard,
 549                                         // since we are doing pointer
 550                                         // arithmetic past the beginning of
 551                                         // the array.
 552                                         assert((int32_t)(rep0 - symbol - 1)
 553                                                         >= -1);
 554                                         assert((int32_t)(rep0 - symbol - 1)
 555                                                         <= 82);
 556                                         probs = coder->pos_special + rep0
 557                                                         - symbol - 1;
 558                                         symbol = 1;
 559                                         offset = 0;
 560         case SEQ_DIST_MODEL:
 561 #ifdef HAVE_SMALL
 562                                         do {
 563                                                 rc_bit(probs[symbol], ,
 564                                                         rep0 += 1 << offset,
 565                                                         SEQ_DIST_MODEL);
 566                                         } while (++offset < limit);
 567 #else
 568                                         switch (limit) {
 569                                         case 5:
 570                                                 assert(offset == 0);
 571                                                 rc_bit(probs[symbol], ,
 572                                                         rep0 += 1,
 573                                                         SEQ_DIST_MODEL);
 574                                                 ++offset;
 575                                                 --limit;
 576                                         case 4:
 577                                                 rc_bit(probs[symbol], ,
 578                                                         rep0 += 1 << offset,
 579                                                         SEQ_DIST_MODEL);
 580                                                 ++offset;
 581                                                 --limit;
 582                                         case 3:
 583                                                 rc_bit(probs[symbol], ,
 584                                                         rep0 += 1 << offset,
 585                                                         SEQ_DIST_MODEL);
 586                                                 ++offset;
 587                                                 --limit;
 588                                         case 2:
 589                                                 rc_bit(probs[symbol], ,
 590                                                         rep0 += 1 << offset,
 591                                                         SEQ_DIST_MODEL);
 592                                                 ++offset;
 593                                                 --limit;
 594                                         case 1:
 595                                                 // We need "symbol" only for
 596                                                 // indexing the probability
 597                                                 // array, thus we can use
 598                                                 // rc_bit_last() here to omit
 599                                                 // the unneeded updating of
 600                                                 // "symbol".
 601                                                 rc_bit_last(probs[symbol], ,
 602                                                         rep0 += 1 << offset,
 603                                                         SEQ_DIST_MODEL);
 604                                         }
 605 #endif
 606                                 } else {
 607                                         // The distance is >= 128. Decode the
 608                                         // lower bits without probabilities
 609                                         // except the lowest four bits.
 610                                         assert(symbol >= 14);
 611                                         assert(limit >= 6);
 612                                         limit -= ALIGN_BITS;
 613                                         assert(limit >= 2);
 614         case SEQ_DIRECT:
 615                                         // Not worth manual unrolling
 616                                         do {
 617                                                 rc_direct(rep0, SEQ_DIRECT);
 618                                         } while (--limit > 0);
 619
 620                                         // Decode the lowest four bits using
 621                                         // probabilities.
 622                                         rep0 <<= ALIGN_BITS;
 623                                         symbol = 1;
 624 #ifdef HAVE_SMALL
 625                                         offset = 0;
 626         case SEQ_ALIGN:
 627                                         do {
 628                                                 rc_bit(coder->pos_align[
 629                                                                 symbol], ,
 630                                                         rep0 += 1 << offset,
 631                                                         SEQ_ALIGN);
 632                                         } while (++offset < ALIGN_BITS);
 633 #else
 634         case SEQ_ALIGN0:
 635                                         rc_bit(coder->pos_align[symbol], ,
 636                                                         rep0 += 1, SEQ_ALIGN0);
 637         case SEQ_ALIGN1:
 638                                         rc_bit(coder->pos_align[symbol], ,
 639                                                         rep0 += 2, SEQ_ALIGN1);
 640         case SEQ_ALIGN2:
 641                                         rc_bit(coder->pos_align[symbol], ,
 642                                                         rep0 += 4, SEQ_ALIGN2);
 643         case SEQ_ALIGN3:
 644                                         // Like in SEQ_DIST_MODEL, we don't
 645                                         // need "symbol" for anything else
 646                                         // than indexing the probability array.
 647                                         rc_bit_last(coder->pos_align[symbol], ,
 648                                                         rep0 += 8, SEQ_ALIGN3);
 649 #endif
 650
 651                                         if (rep0 == UINT32_MAX) {
 652                                                 // End of payload marker was
 653                                                 // found. It must not be
 654                                                 // present if uncompressed
 655                                                 // size is known.
 656                                                 if (coder->uncompressed_size
 657                                                 != LZMA_VLI_UNKNOWN) {
 658                                                         ret = LZMA_DATA_ERROR;
 659                                                         goto out;
 660                                                 }
 661
 662         case SEQ_EOPM:
 663                                                 // LZMA1 stream with
 664                                                 // end-of-payload marker.
 665                                                 rc_normalize(SEQ_EOPM);
 666                                                 ret = LZMA_STREAM_END;
 667                                                 goto out;
 668                                         }
 669                                 }
 670                         }
 671
 672                         // Validate the distance we just decoded.
 673                         if (unlikely(!dict_is_distance_valid(&dict, rep0))) {
 674                                 ret = LZMA_DATA_ERROR;
 675                                 goto out;
 676                         }
 677
 678                 } else {
 679                         rc_update_1(coder->is_rep[state]);
 680
 681                         // Repeated match
 682                         //
 683                         // The match distance is a value that we have had
 684                         // earlier. The latest four match distances are
 685                         // available as rep0, rep1, rep2 and rep3. We will
 686                         // now decode which of them is the new distance.
 687                         //
 688                         // There cannot be a match if we haven't produced
 689                         // any output, so check that first.
 690                         if (unlikely(!dict_is_distance_valid(&dict, 0))) {
 691                                 ret = LZMA_DATA_ERROR;
 692                                 goto out;
 693                         }
 694
 695         case SEQ_IS_REP0:
 696                         rc_if_0(coder->is_rep0[state], SEQ_IS_REP0) {
 697                                 rc_update_0(coder->is_rep0[state]);
 698                                 // The distance is rep0.
 699
 700         case SEQ_IS_REP0_LONG:
 701                                 rc_if_0(coder->is_rep0_long[state][pos_state],
 702                                                 SEQ_IS_REP0_LONG) {
 703                                         rc_update_0(coder->is_rep0_long[
 704                                                         state][pos_state]);
 705
 706                                         update_short_rep(state);
 707
 708         case SEQ_SHORTREP:
 709                                         if (unlikely(dict_put(&dict, dict_get(
 710                                                         &dict, rep0)))) {
 711                                                 coder->sequence = SEQ_SHORTREP;
 712                                                 goto out;
 713                                         }
 714
 715                                         continue;
 716                                 }
 717
 718                                 // Repeating more than one byte at
 719                                 // distance of rep0.
 720                                 rc_update_1(coder->is_rep0_long[
 721                                                 state][pos_state]);
 722
 723                         } else {
 724                                 rc_update_1(coder->is_rep0[state]);
 725
 726         case SEQ_IS_REP1:
 727                                 // The distance is rep1, rep2 or rep3. Once
 728                                 // we find out which one of these three, it
 729                                 // is stored to rep0 and rep1, rep2 and rep3
 730                                 // are updated accordingly.
 731                                 rc_if_0(coder->is_rep1[state], SEQ_IS_REP1) {
 732                                         rc_update_0(coder->is_rep1[state]);
 733
 734                                         const uint32_t distance = rep1;
 735                                         rep1 = rep0;
 736                                         rep0 = distance;
 737
 738                                 } else {
 739                                         rc_update_1(coder->is_rep1[state]);
 740         case SEQ_IS_REP2:
 741                                         rc_if_0(coder->is_rep2[state],
 742                                                         SEQ_IS_REP2) {
 743                                                 rc_update_0(coder->is_rep2[
 744                                                                 state]);
 745
 746                                                 const uint32_t distance = rep2;
 747                                                 rep2 = rep1;
 748                                                 rep1 = rep0;
 749                                                 rep0 = distance;
 750
 751                                         } else {
 752                                                 rc_update_1(coder->is_rep2[
 753                                                                 state]);
 754
 755                                                 const uint32_t distance = rep3;
 756                                                 rep3 = rep2;
 757                                                 rep2 = rep1;
 758                                                 rep1 = rep0;
 759                                                 rep0 = distance;
 760                                         }
 761                                 }
 762                         }
 763
 764                         update_long_rep(state);
 765
 766                         // Decode the length of the repeated match.
 767                         len_decode(len, coder->rep_len_decoder,
 768                                         pos_state, SEQ_REP_LEN);
 769                 }
 770
 771                 /////////////////////////////////
 772                 // Repeat from history buffer. //
 773                 /////////////////////////////////
 774
 775                 // The length is always between these limits. There is no way
 776                 // to trigger the algorithm to set len outside this range.
 777                 assert(len >= MATCH_LEN_MIN);
 778                 assert(len <= MATCH_LEN_MAX);
 779
 780         case SEQ_COPY:
 781                 // Repeat len bytes from distance of rep0.
 782                 if (unlikely(dict_repeat(&dict, rep0, &len))) {
 783                         coder->sequence = SEQ_COPY;
 784                         goto out;
 785                 }
 786         }
 787
 788         rc_normalize(SEQ_NORMALIZE);
 789         coder->sequence = SEQ_IS_MATCH;
 790
 791 out:
 792         // Save state
 793
 794         // NOTE: Must not copy dict.limit.
 795         dictptr->pos = dict.pos;
 796         dictptr->full = dict.full;
 797
 798         rc_from_local(coder->rc, *in_pos);
 799
 800         coder->state = state;
 801         coder->rep0 = rep0;
 802         coder->rep1 = rep1;
 803         coder->rep2 = rep2;
 804         coder->rep3 = rep3;
 805
 806         coder->probs = probs;
 807         coder->symbol = symbol;
 808         coder->limit = limit;
 809         coder->offset = offset;
 810         coder->len = len;
 811
 812         // Update the remaining amount of uncompressed data if uncompressed
 813         // size was known.
 814         if (coder->uncompressed_size != LZMA_VLI_UNKNOWN) {
 815                 coder->uncompressed_size -= dict.pos - dict_start;
 816
 817                 // Since there cannot be end of payload marker if the
 818                 // uncompressed size was known, we check here if we
 819                 // finished decoding.
 820                 if (coder->uncompressed_size == 0 && ret == LZMA_OK
 821                                 && coder->sequence != SEQ_NORMALIZE)
 822                         ret = coder->sequence == SEQ_IS_MATCH
 823                                         ? LZMA_STREAM_END : LZMA_DATA_ERROR;
 824         }
 825
 826         // We can do an additional check in the range decoder to catch some
 827         // corrupted files.
 828         if (ret == LZMA_STREAM_END) {
 829                 if (!rc_is_finished(coder->rc))
 830                         ret = LZMA_DATA_ERROR;
 831
 832                 // Reset the range decoder so that it is ready to reinitialize
 833                 // for a new LZMA2 chunk.
 834                 rc_reset(coder->rc);
 835         }
 836
 837         return ret;
 838 }
 839
 840
 841
 842 static void
 843 lzma_decoder_uncompressed(lzma_coder *coder, lzma_vli uncompressed_size)
 844 {
 845         coder->uncompressed_size = uncompressed_size;
 846 }
 847
 848 /*
 849 extern void
 850 lzma_lzma_decoder_uncompressed(void *coder_ptr, lzma_vli uncompressed_size)
 851 {
 852         // This is hack.
 853         (*(lzma_coder **)(coder))->uncompressed_size = uncompressed_size;
 854 }
 855 */
 856
 857 static void
 858 lzma_decoder_reset(lzma_coder *coder, const void *opt)
 859 {
 860         const lzma_options_lzma *options = opt;
 861
 862         // NOTE: We assume that lc/lp/pb are valid since they were
 863         // successfully decoded with lzma_lzma_decode_properties().
 864
 865         // Calculate pos_mask. We don't need pos_bits as is for anything.
 866         coder->pos_mask = (1U << options->pb) - 1;
 867
 868         // Initialize the literal decoder.
 869         literal_init(coder->literal, options->lc, options->lp);
 870
 871         coder->literal_context_bits = options->lc;
 872         coder->literal_pos_mask = (1U << options->lp) - 1;
 873
 874         // State
 875         coder->state = STATE_LIT_LIT;
 876         coder->rep0 = 0;
 877         coder->rep1 = 0;
 878         coder->rep2 = 0;
 879         coder->rep3 = 0;
 880         coder->pos_mask = (1U << options->pb) - 1;
 881
 882         // Range decoder
 883         rc_reset(coder->rc);
 884
 885         // Bit and bittree decoders
 886         for (uint32_t i = 0; i < STATES; ++i) {
 887                 for (uint32_t j = 0; j <= coder->pos_mask; ++j) {
 888                         bit_reset(coder->is_match[i][j]);
 889                         bit_reset(coder->is_rep0_long[i][j]);
 890                 }
 891
 892                 bit_reset(coder->is_rep[i]);
 893                 bit_reset(coder->is_rep0[i]);
 894                 bit_reset(coder->is_rep1[i]);
 895                 bit_reset(coder->is_rep2[i]);
 896         }
 897
 898         for (uint32_t i = 0; i < DIST_STATES; ++i)
 899                 bittree_reset(coder->dist_slot[i], DIST_SLOT_BITS);
 900
 901         for (uint32_t i = 0; i < FULL_DISTANCES - DIST_MODEL_END; ++i)
 902                 bit_reset(coder->pos_special[i]);
 903
 904         bittree_reset(coder->pos_align, ALIGN_BITS);
 905
 906         // Len decoders (also bit/bittree)
 907         const uint32_t num_pos_states = 1U << options->pb;
 908         bit_reset(coder->match_len_decoder.choice);
 909         bit_reset(coder->match_len_decoder.choice2);
 910         bit_reset(coder->rep_len_decoder.choice);
 911         bit_reset(coder->rep_len_decoder.choice2);
 912
 913         for (uint32_t pos_state = 0; pos_state < num_pos_states; ++pos_state) {
 914                 bittree_reset(coder->match_len_decoder.low[pos_state],
 915                                 LEN_LOW_BITS);
 916                 bittree_reset(coder->match_len_decoder.mid[pos_state],
 917                                 LEN_MID_BITS);
 918
 919                 bittree_reset(coder->rep_len_decoder.low[pos_state],
 920                                 LEN_LOW_BITS);
 921                 bittree_reset(coder->rep_len_decoder.mid[pos_state],
 922                                 LEN_MID_BITS);
 923         }
 924
 925         bittree_reset(coder->match_len_decoder.high, LEN_HIGH_BITS);
 926         bittree_reset(coder->rep_len_decoder.high, LEN_HIGH_BITS);
 927
 928         coder->sequence = SEQ_IS_MATCH;
 929         coder->probs = NULL;
 930         coder->symbol = 0;
 931         coder->limit = 0;
 932         coder->offset = 0;
 933         coder->len = 0;
 934
 935         return;
 936 }
 937
 938
 939 extern lzma_ret
 940 lzma_lzma_decoder_create(lzma_lz_decoder *lz, const lzma_allocator *allocator,
 941                 const void *opt, lzma_lz_options *lz_options)
 942 {
 943         if (lz->coder == NULL) {
 944                 lz->coder = lzma_alloc(sizeof(lzma_coder), allocator);
 945                 if (lz->coder == NULL)
 946                         return LZMA_MEM_ERROR;
 947
 948                 lz->code = &lzma_decode;
 949                 lz->reset = &lzma_decoder_reset;
 950                 lz->set_uncompressed = &lzma_decoder_uncompressed;
 951         }
 952
 953         // All dictionary sizes are OK here. LZ decoder will take care of
 954         // the special cases.
 955         const lzma_options_lzma *options = opt;
 956         lz_options->dict_size = options->dict_size;
 957         lz_options->preset_dict = options->preset_dict;
 958         lz_options->preset_dict_size = options->preset_dict_size;
 959
 960         return LZMA_OK;
 961 }
 962
 963
 964 /// Allocate and initialize LZMA decoder. This is used only via LZ
 965 /// initialization (lzma_lzma_decoder_init() passes function pointer to
 966 /// the LZ initialization).
 967 static lzma_ret
 968 lzma_decoder_init(lzma_lz_decoder *lz, const lzma_allocator *allocator,
 969                 const void *options, lzma_lz_options *lz_options)
 970 {
 971         if (!is_lclppb_valid(options))
 972                 return LZMA_PROG_ERROR;
 973
 974         return_if_error(lzma_lzma_decoder_create(
 975                         lz, allocator, options, lz_options));
 976
 977         lzma_decoder_reset(lz->coder, options);
 978         lzma_decoder_uncompressed(lz->coder, LZMA_VLI_UNKNOWN);
 979
 980         return LZMA_OK;
 981 }
 982
 983
 984 extern lzma_ret
 985 lzma_lzma_decoder_init(lzma_next_coder *next, const lzma_allocator *allocator,
 986                 const lzma_filter_info *filters)
 987 {
 988         // LZMA can only be the last filter in the chain. This is enforced
 989         // by the raw_decoder initialization.
 990         assert(filters[1].init == NULL);
 991
 992         return lzma_lz_decoder_init(next, allocator, filters,
 993                         &lzma_decoder_init);
 994 }
 995
 996
 997 extern bool
 998 lzma_lzma_lclppb_decode(lzma_options_lzma *options, uint8_t byte)
 999 {
1000         if (byte > (4 * 5 + 4) * 9 + 8)
1001                 return true;
1002
1003         // See the file format specification to understand this.
1004         options->pb = byte / (9 * 5);
1005         byte -= options->pb * 9 * 5;
1006         options->lp = byte / 9;
1007         options->lc = byte - options->lp * 9;
1008
1009         return options->lc + options->lp > LZMA_LCLP_MAX;
1010 }
1011
1012
1013 extern uint64_t
1014 lzma_lzma_decoder_memusage_nocheck(const void *options)
1015 {
1016         const lzma_options_lzma *const opt = options;
1017         return sizeof(lzma_coder) + lzma_lz_decoder_memusage(opt->dict_size);
1018 }
1019
1020
1021 extern uint64_t
1022 lzma_lzma_decoder_memusage(const void *options)
1023 {
1024         if (!is_lclppb_valid(options))
1025                 return UINT64_MAX;
1026
1027         return lzma_lzma_decoder_memusage_nocheck(options);
1028 }
1029
1030
1031 extern lzma_ret
1032 lzma_lzma_props_decode(void **options, const lzma_allocator *allocator,
1033                 const uint8_t *props, size_t props_size)
1034 {
1035         if (props_size != 5)
1036                 return LZMA_OPTIONS_ERROR;
1037
1038         lzma_options_lzma *opt
1039                         = lzma_alloc(sizeof(lzma_options_lzma), allocator);
1040         if (opt == NULL)
1041                 return LZMA_MEM_ERROR;
1042
1043         if (lzma_lzma_lclppb_decode(opt, props[0]))
1044                 goto error;
1045
1046         // All dictionary sizes are accepted, including zero. LZ decoder
1047         // will automatically use a dictionary at least a few KiB even if
1048         // a smaller dictionary is requested.
1049         opt->dict_size = unaligned_read32le(props + 1);
1050
1051         opt->preset_dict = NULL;
1052         opt->preset_dict_size = 0;
1053
1054         *options = opt;
1055
1056         return LZMA_OK;
1057
1058 error:
1059         lzma_free(opt, allocator);
1060         return LZMA_OPTIONS_ERROR;
1061 }