1 /* 2 * LZMA2 decoder 3 * 4 * Authors: Lasse Collin <lasse.collin@tukaani.org> 5 * Igor Pavlov <https://7-zip.org/> 6 * 7 * This file has been put into the public domain. 8 * You can do whatever you want with this file. 9 */ 10 11 #include "xz_private.h" 12 #include "xz_lzma2.h" 13 14 /* 15 * Range decoder initialization eats the first five bytes of each LZMA chunk. 16 */ 17 #define RC_INIT_BYTES 5 18 19 /* 20 * Minimum number of usable input buffer to safely decode one LZMA symbol. 21 * The worst case is that we decode 22 bits using probabilities and 26 22 * direct bits. This may decode at maximum of 20 bytes of input. However, 23 * lzma_main() does an extra normalization before returning, thus we 24 * need to put 21 here. 25 */ 26 #define LZMA_IN_REQUIRED 21 27 28 /* 29 * Dictionary (history buffer) 30 * 31 * These are always true: 32 * start <= pos <= full <= end 33 * pos <= limit <= end 34 * 35 * In multi-call mode, also these are true: 36 * end == size 37 * size <= size_max 38 * allocated <= size 39 * 40 * Most of these variables are size_t to support single-call mode, 41 * in which the dictionary variables address the actual output 42 * buffer directly. 43 */ 44 struct dictionary { 45 /* Beginning of the history buffer */ 46 uint8_t *buf; 47 48 /* Old position in buf (before decoding more data) */ 49 size_t start; 50 51 /* Position in buf */ 52 size_t pos; 53 54 /* 55 * How full dictionary is. This is used to detect corrupt input that 56 * would read beyond the beginning of the uncompressed stream. 57 */ 58 size_t full; 59 60 /* Write limit; we don't write to buf[limit] or later bytes. */ 61 size_t limit; 62 63 /* 64 * End of the dictionary buffer. In multi-call mode, this is 65 * the same as the dictionary size. In single-call mode, this 66 * indicates the size of the output buffer. 67 */ 68 size_t end; 69 70 /* 71 * Size of the dictionary as specified in Block Header. This is used 72 * together with "full" to detect corrupt input that would make us 73 * read beyond the beginning of the uncompressed stream. 74 */ 75 uint32_t size; 76 77 /* 78 * Maximum allowed dictionary size in multi-call mode. 79 * This is ignored in single-call mode. 80 */ 81 uint32_t size_max; 82 83 /* 84 * Amount of memory currently allocated for the dictionary. 85 * This is used only with XZ_DYNALLOC. (With XZ_PREALLOC, 86 * size_max is always the same as the allocated size.) 87 */ 88 uint32_t allocated; 89 90 /* Operation mode */ 91 enum xz_mode mode; 92 }; 93 94 /* Range decoder */ 95 struct rc_dec { 96 uint32_t range; 97 uint32_t code; 98 99 /* 100 * Number of initializing bytes remaining to be read 101 * by rc_read_init(). 102 */ 103 uint32_t init_bytes_left; 104 105 /* 106 * Buffer from which we read our input. It can be either 107 * temp.buf or the caller-provided input buffer. 108 */ 109 const uint8_t *in; 110 size_t in_pos; 111 size_t in_limit; 112 }; 113 114 /* Probabilities for a length decoder. */ 115 struct lzma_len_dec { 116 /* Probability of match length being at least 10 */ 117 uint16_t choice; 118 119 /* Probability of match length being at least 18 */ 120 uint16_t choice2; 121 122 /* Probabilities for match lengths 2-9 */ 123 uint16_t low[POS_STATES_MAX][LEN_LOW_SYMBOLS]; 124 125 /* Probabilities for match lengths 10-17 */ 126 uint16_t mid[POS_STATES_MAX][LEN_MID_SYMBOLS]; 127 128 /* Probabilities for match lengths 18-273 */ 129 uint16_t high[LEN_HIGH_SYMBOLS]; 130 }; 131 132 struct lzma_dec { 133 /* Distances of latest four matches */ 134 uint32_t rep0; 135 uint32_t rep1; 136 uint32_t rep2; 137 uint32_t rep3; 138 139 /* Types of the most recently seen LZMA symbols */ 140 enum lzma_state state; 141 142 /* 143 * Length of a match. This is updated so that dict_repeat can 144 * be called again to finish repeating the whole match. 145 */ 146 uint32_t len; 147 148 /* 149 * LZMA properties or related bit masks (number of literal 150 * context bits, a mask derived from the number of literal 151 * position bits, and a mask derived from the number 152 * position bits) 153 */ 154 uint32_t lc; 155 uint32_t literal_pos_mask; /* (1 << lp) - 1 */ 156 uint32_t pos_mask; /* (1 << pb) - 1 */ 157 158 /* If 1, it's a match. Otherwise it's a single 8-bit literal. */ 159 uint16_t is_match[STATES][POS_STATES_MAX]; 160 161 /* If 1, it's a repeated match. The distance is one of rep0 .. rep3. */ 162 uint16_t is_rep[STATES]; 163 164 /* 165 * If 0, distance of a repeated match is rep0. 166 * Otherwise check is_rep1. 167 */ 168 uint16_t is_rep0[STATES]; 169 170 /* 171 * If 0, distance of a repeated match is rep1. 172 * Otherwise check is_rep2. 173 */ 174 uint16_t is_rep1[STATES]; 175 176 /* If 0, distance of a repeated match is rep2. Otherwise it is rep3. */ 177 uint16_t is_rep2[STATES]; 178 179 /* 180 * If 1, the repeated match has length of one byte. Otherwise 181 * the length is decoded from rep_len_decoder. 182 */ 183 uint16_t is_rep0_long[STATES][POS_STATES_MAX]; 184 185 /* 186 * Probability tree for the highest two bits of the match 187 * distance. There is a separate probability tree for match 188 * lengths of 2 (i.e. MATCH_LEN_MIN), 3, 4, and [5, 273]. 189 */ 190 uint16_t dist_slot[DIST_STATES][DIST_SLOTS]; 191 192 /* 193 * Probility trees for additional bits for match distance 194 * when the distance is in the range [4, 127]. 195 */ 196 uint16_t dist_special[FULL_DISTANCES - DIST_MODEL_END]; 197 198 /* 199 * Probability tree for the lowest four bits of a match 200 * distance that is equal to or greater than 128. 201 */ 202 uint16_t dist_align[ALIGN_SIZE]; 203 204 /* Length of a normal match */ 205 struct lzma_len_dec match_len_dec; 206 207 /* Length of a repeated match */ 208 struct lzma_len_dec rep_len_dec; 209 210 /* Probabilities of literals */ 211 uint16_t literal[LITERAL_CODERS_MAX][LITERAL_CODER_SIZE]; 212 }; 213 214 struct lzma2_dec { 215 /* Position in xz_dec_lzma2_run(). */ 216 enum lzma2_seq { 217 SEQ_CONTROL, 218 SEQ_UNCOMPRESSED_1, 219 SEQ_UNCOMPRESSED_2, 220 SEQ_COMPRESSED_0, 221 SEQ_COMPRESSED_1, 222 SEQ_PROPERTIES, 223 SEQ_LZMA_PREPARE, 224 SEQ_LZMA_RUN, 225 SEQ_COPY 226 } sequence; 227 228 /* Next position after decoding the compressed size of the chunk. */ 229 enum lzma2_seq next_sequence; 230 231 /* Uncompressed size of LZMA chunk (2 MiB at maximum) */ 232 uint32_t uncompressed; 233 234 /* 235 * Compressed size of LZMA chunk or compressed/uncompressed 236 * size of uncompressed chunk (64 KiB at maximum) 237 */ 238 uint32_t compressed; 239 240 /* 241 * True if dictionary reset is needed. This is false before 242 * the first chunk (LZMA or uncompressed). 243 */ 244 bool need_dict_reset; 245 246 /* 247 * True if new LZMA properties are needed. This is false 248 * before the first LZMA chunk. 249 */ 250 bool need_props; 251 252 #ifdef XZ_DEC_MICROLZMA 253 bool pedantic_microlzma; 254 #endif 255 }; 256 257 struct xz_dec_lzma2 { 258 /* 259 * The order below is important on x86 to reduce code size and 260 * it shouldn't hurt on other platforms. Everything up to and 261 * including lzma.pos_mask are in the first 128 bytes on x86-32, 262 * which allows using smaller instructions to access those 263 * variables. On x86-64, fewer variables fit into the first 128 264 * bytes, but this is still the best order without sacrificing 265 * the readability by splitting the structures. 266 */ 267 struct rc_dec rc; 268 struct dictionary dict; 269 struct lzma2_dec lzma2; 270 struct lzma_dec lzma; 271 272 /* 273 * Temporary buffer which holds small number of input bytes between 274 * decoder calls. See lzma2_lzma() for details. 275 */ 276 struct { 277 uint32_t size; 278 uint8_t buf[3 * LZMA_IN_REQUIRED]; 279 } temp; 280 }; 281 282 /************** 283 * Dictionary * 284 **************/ 285 286 /* 287 * Reset the dictionary state. When in single-call mode, set up the beginning 288 * of the dictionary to point to the actual output buffer. 289 */ 290 static void dict_reset(struct dictionary *dict, struct xz_buf *b) 291 { 292 if (DEC_IS_SINGLE(dict->mode)) { 293 dict->buf = b->out + b->out_pos; 294 dict->end = b->out_size - b->out_pos; 295 } 296 297 dict->start = 0; 298 dict->pos = 0; 299 dict->limit = 0; 300 dict->full = 0; 301 } 302 303 /* Set dictionary write limit */ 304 static void dict_limit(struct dictionary *dict, size_t out_max) 305 { 306 if (dict->end - dict->pos <= out_max) 307 dict->limit = dict->end; 308 else 309 dict->limit = dict->pos + out_max; 310 } 311 312 /* Return true if at least one byte can be written into the dictionary. */ 313 static inline bool dict_has_space(const struct dictionary *dict) 314 { 315 return dict->pos < dict->limit; 316 } 317 318 /* 319 * Get a byte from the dictionary at the given distance. The distance is 320 * assumed to valid, or as a special case, zero when the dictionary is 321 * still empty. This special case is needed for single-call decoding to 322 * avoid writing a '\0' to the end of the destination buffer. 323 */ 324 static inline uint32_t dict_get(const struct dictionary *dict, uint32_t dist) 325 { 326 size_t offset = dict->pos - dist - 1; 327 328 if (dist >= dict->pos) 329 offset += dict->end; 330 331 return dict->full > 0 ? dict->buf[offset] : 0; 332 } 333 334 /* 335 * Put one byte into the dictionary. It is assumed that there is space for it. 336 */ 337 static inline void dict_put(struct dictionary *dict, uint8_t byte) 338 { 339 dict->buf[dict->pos++] = byte; 340 341 if (dict->full < dict->pos) 342 dict->full = dict->pos; 343 } 344 345 /* 346 * Repeat given number of bytes from the given distance. If the distance is 347 * invalid, false is returned. On success, true is returned and *len is 348 * updated to indicate how many bytes were left to be repeated. 349 */ 350 static bool dict_repeat(struct dictionary *dict, uint32_t *len, uint32_t dist) 351 { 352 size_t back; 353 uint32_t left; 354 355 if (dist >= dict->full || dist >= dict->size) 356 return false; 357 358 left = min_t(size_t, dict->limit - dict->pos, *len); 359 *len -= left; 360 361 back = dict->pos - dist - 1; 362 if (dist >= dict->pos) 363 back += dict->end; 364 365 do { 366 dict->buf[dict->pos++] = dict->buf[back++]; 367 if (back == dict->end) 368 back = 0; 369 } while (--left > 0); 370 371 if (dict->full < dict->pos) 372 dict->full = dict->pos; 373 374 return true; 375 } 376 377 /* Copy uncompressed data as is from input to dictionary and output buffers. */ 378 static void dict_uncompressed(struct dictionary *dict, struct xz_buf *b, 379 uint32_t *left) 380 { 381 size_t copy_size; 382 383 while (*left > 0 && b->in_pos < b->in_size 384 && b->out_pos < b->out_size) { 385 copy_size = min(b->in_size - b->in_pos, 386 b->out_size - b->out_pos); 387 if (copy_size > dict->end - dict->pos) 388 copy_size = dict->end - dict->pos; 389 if (copy_size > *left) 390 copy_size = *left; 391 392 *left -= copy_size; 393 394 /* 395 * If doing in-place decompression in single-call mode and the 396 * uncompressed size of the file is larger than the caller 397 * thought (i.e. it is invalid input!), the buffers below may 398 * overlap and cause undefined behavior with memcpy(). 399 * With valid inputs memcpy() would be fine here. 400 */ 401 memmove(dict->buf + dict->pos, b->in + b->in_pos, copy_size); 402 dict->pos += copy_size; 403 404 if (dict->full < dict->pos) 405 dict->full = dict->pos; 406 407 if (DEC_IS_MULTI(dict->mode)) { 408 if (dict->pos == dict->end) 409 dict->pos = 0; 410 411 /* 412 * Like above but for multi-call mode: use memmove() 413 * to avoid undefined behavior with invalid input. 414 */ 415 memmove(b->out + b->out_pos, b->in + b->in_pos, 416 copy_size); 417 } 418 419 dict->start = dict->pos; 420 421 b->out_pos += copy_size; 422 b->in_pos += copy_size; 423 } 424 } 425 426 #ifdef XZ_DEC_MICROLZMA 427 # define DICT_FLUSH_SUPPORTS_SKIPPING true 428 #else 429 # define DICT_FLUSH_SUPPORTS_SKIPPING false 430 #endif 431 432 /* 433 * Flush pending data from dictionary to b->out. It is assumed that there is 434 * enough space in b->out. This is guaranteed because caller uses dict_limit() 435 * before decoding data into the dictionary. 436 */ 437 static uint32_t dict_flush(struct dictionary *dict, struct xz_buf *b) 438 { 439 size_t copy_size = dict->pos - dict->start; 440 441 if (DEC_IS_MULTI(dict->mode)) { 442 if (dict->pos == dict->end) 443 dict->pos = 0; 444 445 /* 446 * These buffers cannot overlap even if doing in-place 447 * decompression because in multi-call mode dict->buf 448 * has been allocated by us in this file; it's not 449 * provided by the caller like in single-call mode. 450 * 451 * With MicroLZMA, b->out can be NULL to skip bytes that 452 * the caller doesn't need. This cannot be done with XZ 453 * because it would break BCJ filters. 454 */ 455 if (!DICT_FLUSH_SUPPORTS_SKIPPING || b->out != NULL) 456 memcpy(b->out + b->out_pos, dict->buf + dict->start, 457 copy_size); 458 } 459 460 dict->start = dict->pos; 461 b->out_pos += copy_size; 462 return copy_size; 463 } 464 465 /***************** 466 * Range decoder * 467 *****************/ 468 469 /* Reset the range decoder. */ 470 static void rc_reset(struct rc_dec *rc) 471 { 472 rc->range = (uint32_t)-1; 473 rc->code = 0; 474 rc->init_bytes_left = RC_INIT_BYTES; 475 } 476 477 /* 478 * Read the first five initial bytes into rc->code if they haven't been 479 * read already. (Yes, the first byte gets completely ignored.) 480 */ 481 static bool rc_read_init(struct rc_dec *rc, struct xz_buf *b) 482 { 483 while (rc->init_bytes_left > 0) { 484 if (b->in_pos == b->in_size) 485 return false; 486 487 rc->code = (rc->code << 8) + b->in[b->in_pos++]; 488 --rc->init_bytes_left; 489 } 490 491 return true; 492 } 493 494 /* Return true if there may not be enough input for the next decoding loop. */ 495 static inline bool rc_limit_exceeded(const struct rc_dec *rc) 496 { 497 return rc->in_pos > rc->in_limit; 498 } 499 500 /* 501 * Return true if it is possible (from point of view of range decoder) that 502 * we have reached the end of the LZMA chunk. 503 */ 504 static inline bool rc_is_finished(const struct rc_dec *rc) 505 { 506 return rc->code == 0; 507 } 508 509 /* Read the next input byte if needed. */ 510 static __always_inline void rc_normalize(struct rc_dec *rc) 511 { 512 if (rc->range < RC_TOP_VALUE) { 513 rc->range <<= RC_SHIFT_BITS; 514 rc->code = (rc->code << RC_SHIFT_BITS) + rc->in[rc->in_pos++]; 515 } 516 } 517 518 /* 519 * Decode one bit. In some versions, this function has been split in three 520 * functions so that the compiler is supposed to be able to more easily avoid 521 * an extra branch. In this particular version of the LZMA decoder, this 522 * doesn't seem to be a good idea (tested with GCC 3.3.6, 3.4.6, and 4.3.3 523 * on x86). Using a non-split version results in nicer looking code too. 524 * 525 * NOTE: This must return an int. Do not make it return a bool or the speed 526 * of the code generated by GCC 3.x decreases 10-15 %. (GCC 4.3 doesn't care, 527 * and it generates 10-20 % faster code than GCC 3.x from this file anyway.) 528 */ 529 static __always_inline int rc_bit(struct rc_dec *rc, uint16_t *prob) 530 { 531 uint32_t bound; 532 int bit; 533 534 rc_normalize(rc); 535 bound = (rc->range >> RC_BIT_MODEL_TOTAL_BITS) * *prob; 536 if (rc->code < bound) { 537 rc->range = bound; 538 *prob += (RC_BIT_MODEL_TOTAL - *prob) >> RC_MOVE_BITS; 539 bit = 0; 540 } else { 541 rc->range -= bound; 542 rc->code -= bound; 543 *prob -= *prob >> RC_MOVE_BITS; 544 bit = 1; 545 } 546 547 return bit; 548 } 549 550 /* Decode a bittree starting from the most significant bit. */ 551 static __always_inline uint32_t rc_bittree(struct rc_dec *rc, 552 uint16_t *probs, uint32_t limit) 553 { 554 uint32_t symbol = 1; 555 556 do { 557 if (rc_bit(rc, &probs[symbol])) 558 symbol = (symbol << 1) + 1; 559 else 560 symbol <<= 1; 561 } while (symbol < limit); 562 563 return symbol; 564 } 565 566 /* Decode a bittree starting from the least significant bit. */ 567 static __always_inline void rc_bittree_reverse(struct rc_dec *rc, 568 uint16_t *probs, 569 uint32_t *dest, uint32_t limit) 570 { 571 uint32_t symbol = 1; 572 uint32_t i = 0; 573 574 do { 575 if (rc_bit(rc, &probs[symbol])) { 576 symbol = (symbol << 1) + 1; 577 *dest += 1 << i; 578 } else { 579 symbol <<= 1; 580 } 581 } while (++i < limit); 582 } 583 584 /* Decode direct bits (fixed fifty-fifty probability) */ 585 static inline void rc_direct(struct rc_dec *rc, uint32_t *dest, uint32_t limit) 586 { 587 uint32_t mask; 588 589 do { 590 rc_normalize(rc); 591 rc->range >>= 1; 592 rc->code -= rc->range; 593 mask = (uint32_t)0 - (rc->code >> 31); 594 rc->code += rc->range & mask; 595 *dest = (*dest << 1) + (mask + 1); 596 } while (--limit > 0); 597 } 598 599 /******** 600 * LZMA * 601 ********/ 602 603 /* Get pointer to literal coder probability array. */ 604 static uint16_t *lzma_literal_probs(struct xz_dec_lzma2 *s) 605 { 606 uint32_t prev_byte = dict_get(&s->dict, 0); 607 uint32_t low = prev_byte >> (8 - s->lzma.lc); 608 uint32_t high = (s->dict.pos & s->lzma.literal_pos_mask) << s->lzma.lc; 609 return s->lzma.literal[low + high]; 610 } 611 612 /* Decode a literal (one 8-bit byte) */ 613 static void lzma_literal(struct xz_dec_lzma2 *s) 614 { 615 uint16_t *probs; 616 uint32_t symbol; 617 uint32_t match_byte; 618 uint32_t match_bit; 619 uint32_t offset; 620 uint32_t i; 621 622 probs = lzma_literal_probs(s); 623 624 if (lzma_state_is_literal(s->lzma.state)) { 625 symbol = rc_bittree(&s->rc, probs, 0x100); 626 } else { 627 symbol = 1; 628 match_byte = dict_get(&s->dict, s->lzma.rep0) << 1; 629 offset = 0x100; 630 631 do { 632 match_bit = match_byte & offset; 633 match_byte <<= 1; 634 i = offset + match_bit + symbol; 635 636 if (rc_bit(&s->rc, &probs[i])) { 637 symbol = (symbol << 1) + 1; 638 offset &= match_bit; 639 } else { 640 symbol <<= 1; 641 offset &= ~match_bit; 642 } 643 } while (symbol < 0x100); 644 } 645 646 dict_put(&s->dict, (uint8_t)symbol); 647 lzma_state_literal(&s->lzma.state); 648 } 649 650 /* Decode the length of the match into s->lzma.len. */ 651 static void lzma_len(struct xz_dec_lzma2 *s, struct lzma_len_dec *l, 652 uint32_t pos_state) 653 { 654 uint16_t *probs; 655 uint32_t limit; 656 657 if (!rc_bit(&s->rc, &l->choice)) { 658 probs = l->low[pos_state]; 659 limit = LEN_LOW_SYMBOLS; 660 s->lzma.len = MATCH_LEN_MIN; 661 } else { 662 if (!rc_bit(&s->rc, &l->choice2)) { 663 probs = l->mid[pos_state]; 664 limit = LEN_MID_SYMBOLS; 665 s->lzma.len = MATCH_LEN_MIN + LEN_LOW_SYMBOLS; 666 } else { 667 probs = l->high; 668 limit = LEN_HIGH_SYMBOLS; 669 s->lzma.len = MATCH_LEN_MIN + LEN_LOW_SYMBOLS 670 + LEN_MID_SYMBOLS; 671 } 672 } 673 674 s->lzma.len += rc_bittree(&s->rc, probs, limit) - limit; 675 } 676 677 /* Decode a match. The distance will be stored in s->lzma.rep0. */ 678 static void lzma_match(struct xz_dec_lzma2 *s, uint32_t pos_state) 679 { 680 uint16_t *probs; 681 uint32_t dist_slot; 682 uint32_t limit; 683 684 lzma_state_match(&s->lzma.state); 685 686 s->lzma.rep3 = s->lzma.rep2; 687 s->lzma.rep2 = s->lzma.rep1; 688 s->lzma.rep1 = s->lzma.rep0; 689 690 lzma_len(s, &s->lzma.match_len_dec, pos_state); 691 692 probs = s->lzma.dist_slot[lzma_get_dist_state(s->lzma.len)]; 693 dist_slot = rc_bittree(&s->rc, probs, DIST_SLOTS) - DIST_SLOTS; 694 695 if (dist_slot < DIST_MODEL_START) { 696 s->lzma.rep0 = dist_slot; 697 } else { 698 limit = (dist_slot >> 1) - 1; 699 s->lzma.rep0 = 2 + (dist_slot & 1); 700 701 if (dist_slot < DIST_MODEL_END) { 702 s->lzma.rep0 <<= limit; 703 probs = s->lzma.dist_special + s->lzma.rep0 704 - dist_slot - 1; 705 rc_bittree_reverse(&s->rc, probs, 706 &s->lzma.rep0, limit); 707 } else { 708 rc_direct(&s->rc, &s->lzma.rep0, limit - ALIGN_BITS); 709 s->lzma.rep0 <<= ALIGN_BITS; 710 rc_bittree_reverse(&s->rc, s->lzma.dist_align, 711 &s->lzma.rep0, ALIGN_BITS); 712 } 713 } 714 } 715 716 /* 717 * Decode a repeated match. The distance is one of the four most recently 718 * seen matches. The distance will be stored in s->lzma.rep0. 719 */ 720 static void lzma_rep_match(struct xz_dec_lzma2 *s, uint32_t pos_state) 721 { 722 uint32_t tmp; 723 724 if (!rc_bit(&s->rc, &s->lzma.is_rep0[s->lzma.state])) { 725 if (!rc_bit(&s->rc, &s->lzma.is_rep0_long[ 726 s->lzma.state][pos_state])) { 727 lzma_state_short_rep(&s->lzma.state); 728 s->lzma.len = 1; 729 return; 730 } 731 } else { 732 if (!rc_bit(&s->rc, &s->lzma.is_rep1[s->lzma.state])) { 733 tmp = s->lzma.rep1; 734 } else { 735 if (!rc_bit(&s->rc, &s->lzma.is_rep2[s->lzma.state])) { 736 tmp = s->lzma.rep2; 737 } else { 738 tmp = s->lzma.rep3; 739 s->lzma.rep3 = s->lzma.rep2; 740 } 741 742 s->lzma.rep2 = s->lzma.rep1; 743 } 744 745 s->lzma.rep1 = s->lzma.rep0; 746 s->lzma.rep0 = tmp; 747 } 748 749 lzma_state_long_rep(&s->lzma.state); 750 lzma_len(s, &s->lzma.rep_len_dec, pos_state); 751 } 752 753 /* LZMA decoder core */ 754 static bool lzma_main(struct xz_dec_lzma2 *s) 755 { 756 uint32_t pos_state; 757 758 /* 759 * If the dictionary was reached during the previous call, try to 760 * finish the possibly pending repeat in the dictionary. 761 */ 762 if (dict_has_space(&s->dict) && s->lzma.len > 0) 763 dict_repeat(&s->dict, &s->lzma.len, s->lzma.rep0); 764 765 /* 766 * Decode more LZMA symbols. One iteration may consume up to 767 * LZMA_IN_REQUIRED - 1 bytes. 768 */ 769 while (dict_has_space(&s->dict) && !rc_limit_exceeded(&s->rc)) { 770 pos_state = s->dict.pos & s->lzma.pos_mask; 771 772 if (!rc_bit(&s->rc, &s->lzma.is_match[ 773 s->lzma.state][pos_state])) { 774 lzma_literal(s); 775 } else { 776 if (rc_bit(&s->rc, &s->lzma.is_rep[s->lzma.state])) 777 lzma_rep_match(s, pos_state); 778 else 779 lzma_match(s, pos_state); 780 781 if (!dict_repeat(&s->dict, &s->lzma.len, s->lzma.rep0)) 782 return false; 783 } 784 } 785 786 /* 787 * Having the range decoder always normalized when we are outside 788 * this function makes it easier to correctly handle end of the chunk. 789 */ 790 rc_normalize(&s->rc); 791 792 return true; 793 } 794 795 /* 796 * Reset the LZMA decoder and range decoder state. Dictionary is not reset 797 * here, because LZMA state may be reset without resetting the dictionary. 798 */ 799 static void lzma_reset(struct xz_dec_lzma2 *s) 800 { 801 uint16_t *probs; 802 size_t i; 803 804 s->lzma.state = STATE_LIT_LIT; 805 s->lzma.rep0 = 0; 806 s->lzma.rep1 = 0; 807 s->lzma.rep2 = 0; 808 s->lzma.rep3 = 0; 809 s->lzma.len = 0; 810 811 /* 812 * All probabilities are initialized to the same value. This hack 813 * makes the code smaller by avoiding a separate loop for each 814 * probability array. 815 * 816 * This could be optimized so that only that part of literal 817 * probabilities that are actually required. In the common case 818 * we would write 12 KiB less. 819 */ 820 probs = s->lzma.is_match[0]; 821 for (i = 0; i < PROBS_TOTAL; ++i) 822 probs[i] = RC_BIT_MODEL_TOTAL / 2; 823 824 rc_reset(&s->rc); 825 } 826 827 /* 828 * Decode and validate LZMA properties (lc/lp/pb) and calculate the bit masks 829 * from the decoded lp and pb values. On success, the LZMA decoder state is 830 * reset and true is returned. 831 */ 832 static bool lzma_props(struct xz_dec_lzma2 *s, uint8_t props) 833 { 834 if (props > (4 * 5 + 4) * 9 + 8) 835 return false; 836 837 s->lzma.pos_mask = 0; 838 while (props >= 9 * 5) { 839 props -= 9 * 5; 840 ++s->lzma.pos_mask; 841 } 842 843 s->lzma.pos_mask = (1 << s->lzma.pos_mask) - 1; 844 845 s->lzma.literal_pos_mask = 0; 846 while (props >= 9) { 847 props -= 9; 848 ++s->lzma.literal_pos_mask; 849 } 850 851 s->lzma.lc = props; 852 853 if (s->lzma.lc + s->lzma.literal_pos_mask > 4) 854 return false; 855 856 s->lzma.literal_pos_mask = (1 << s->lzma.literal_pos_mask) - 1; 857 858 lzma_reset(s); 859 860 return true; 861 } 862 863 /********* 864 * LZMA2 * 865 *********/ 866 867 /* 868 * The LZMA decoder assumes that if the input limit (s->rc.in_limit) hasn't 869 * been exceeded, it is safe to read up to LZMA_IN_REQUIRED bytes. This 870 * wrapper function takes care of making the LZMA decoder's assumption safe. 871 * 872 * As long as there is plenty of input left to be decoded in the current LZMA 873 * chunk, we decode directly from the caller-supplied input buffer until 874 * there's LZMA_IN_REQUIRED bytes left. Those remaining bytes are copied into 875 * s->temp.buf, which (hopefully) gets filled on the next call to this 876 * function. We decode a few bytes from the temporary buffer so that we can 877 * continue decoding from the caller-supplied input buffer again. 878 */ 879 static bool lzma2_lzma(struct xz_dec_lzma2 *s, struct xz_buf *b) 880 { 881 size_t in_avail; 882 uint32_t tmp; 883 884 in_avail = b->in_size - b->in_pos; 885 if (s->temp.size > 0 || s->lzma2.compressed == 0) { 886 tmp = 2 * LZMA_IN_REQUIRED - s->temp.size; 887 if (tmp > s->lzma2.compressed - s->temp.size) 888 tmp = s->lzma2.compressed - s->temp.size; 889 if (tmp > in_avail) 890 tmp = in_avail; 891 892 memcpy(s->temp.buf + s->temp.size, b->in + b->in_pos, tmp); 893 894 if (s->temp.size + tmp == s->lzma2.compressed) { 895 memzero(s->temp.buf + s->temp.size + tmp, 896 sizeof(s->temp.buf) 897 - s->temp.size - tmp); 898 s->rc.in_limit = s->temp.size + tmp; 899 } else if (s->temp.size + tmp < LZMA_IN_REQUIRED) { 900 s->temp.size += tmp; 901 b->in_pos += tmp; 902 return true; 903 } else { 904 s->rc.in_limit = s->temp.size + tmp - LZMA_IN_REQUIRED; 905 } 906 907 s->rc.in = s->temp.buf; 908 s->rc.in_pos = 0; 909 910 if (!lzma_main(s) || s->rc.in_pos > s->temp.size + tmp) 911 return false; 912 913 s->lzma2.compressed -= s->rc.in_pos; 914 915 if (s->rc.in_pos < s->temp.size) { 916 s->temp.size -= s->rc.in_pos; 917 memmove(s->temp.buf, s->temp.buf + s->rc.in_pos, 918 s->temp.size); 919 return true; 920 } 921 922 b->in_pos += s->rc.in_pos - s->temp.size; 923 s->temp.size = 0; 924 } 925 926 in_avail = b->in_size - b->in_pos; 927 if (in_avail >= LZMA_IN_REQUIRED) { 928 s->rc.in = b->in; 929 s->rc.in_pos = b->in_pos; 930 931 if (in_avail >= s->lzma2.compressed + LZMA_IN_REQUIRED) 932 s->rc.in_limit = b->in_pos + s->lzma2.compressed; 933 else 934 s->rc.in_limit = b->in_size - LZMA_IN_REQUIRED; 935 936 if (!lzma_main(s)) 937 return false; 938 939 in_avail = s->rc.in_pos - b->in_pos; 940 if (in_avail > s->lzma2.compressed) 941 return false; 942 943 s->lzma2.compressed -= in_avail; 944 b->in_pos = s->rc.in_pos; 945 } 946 947 in_avail = b->in_size - b->in_pos; 948 if (in_avail < LZMA_IN_REQUIRED) { 949 if (in_avail > s->lzma2.compressed) 950 in_avail = s->lzma2.compressed; 951 952 memcpy(s->temp.buf, b->in + b->in_pos, in_avail); 953 s->temp.size = in_avail; 954 b->in_pos += in_avail; 955 } 956 957 return true; 958 } 959 960 /* 961 * Take care of the LZMA2 control layer, and forward the job of actual LZMA 962 * decoding or copying of uncompressed chunks to other functions. 963 */ 964 XZ_EXTERN enum xz_ret xz_dec_lzma2_run(struct xz_dec_lzma2 *s, 965 struct xz_buf *b) 966 { 967 uint32_t tmp; 968 969 while (b->in_pos < b->in_size || s->lzma2.sequence == SEQ_LZMA_RUN) { 970 switch (s->lzma2.sequence) { 971 case SEQ_CONTROL: 972 /* 973 * LZMA2 control byte 974 * 975 * Exact values: 976 * 0x00 End marker 977 * 0x01 Dictionary reset followed by 978 * an uncompressed chunk 979 * 0x02 Uncompressed chunk (no dictionary reset) 980 * 981 * Highest three bits (s->control & 0xE0): 982 * 0xE0 Dictionary reset, new properties and state 983 * reset, followed by LZMA compressed chunk 984 * 0xC0 New properties and state reset, followed 985 * by LZMA compressed chunk (no dictionary 986 * reset) 987 * 0xA0 State reset using old properties, 988 * followed by LZMA compressed chunk (no 989 * dictionary reset) 990 * 0x80 LZMA chunk (no dictionary or state reset) 991 * 992 * For LZMA compressed chunks, the lowest five bits 993 * (s->control & 1F) are the highest bits of the 994 * uncompressed size (bits 16-20). 995 * 996 * A new LZMA2 stream must begin with a dictionary 997 * reset. The first LZMA chunk must set new 998 * properties and reset the LZMA state. 999 * 1000 * Values that don't match anything described above 1001 * are invalid and we return XZ_DATA_ERROR. 1002 */ 1003 tmp = b->in[b->in_pos++]; 1004 1005 if (tmp == 0x00) 1006 return XZ_STREAM_END; 1007 1008 if (tmp >= 0xE0 || tmp == 0x01) { 1009 s->lzma2.need_props = true; 1010 s->lzma2.need_dict_reset = false; 1011 dict_reset(&s->dict, b); 1012 } else if (s->lzma2.need_dict_reset) { 1013 return XZ_DATA_ERROR; 1014 } 1015 1016 if (tmp >= 0x80) { 1017 s->lzma2.uncompressed = (tmp & 0x1F) << 16; 1018 s->lzma2.sequence = SEQ_UNCOMPRESSED_1; 1019 1020 if (tmp >= 0xC0) { 1021 /* 1022 * When there are new properties, 1023 * state reset is done at 1024 * SEQ_PROPERTIES. 1025 */ 1026 s->lzma2.need_props = false; 1027 s->lzma2.next_sequence 1028 = SEQ_PROPERTIES; 1029 1030 } else if (s->lzma2.need_props) { 1031 return XZ_DATA_ERROR; 1032 1033 } else { 1034 s->lzma2.next_sequence 1035 = SEQ_LZMA_PREPARE; 1036 if (tmp >= 0xA0) 1037 lzma_reset(s); 1038 } 1039 } else { 1040 if (tmp > 0x02) 1041 return XZ_DATA_ERROR; 1042 1043 s->lzma2.sequence = SEQ_COMPRESSED_0; 1044 s->lzma2.next_sequence = SEQ_COPY; 1045 } 1046 1047 break; 1048 1049 case SEQ_UNCOMPRESSED_1: 1050 s->lzma2.uncompressed 1051 += (uint32_t)b->in[b->in_pos++] << 8; 1052 s->lzma2.sequence = SEQ_UNCOMPRESSED_2; 1053 break; 1054 1055 case SEQ_UNCOMPRESSED_2: 1056 s->lzma2.uncompressed 1057 += (uint32_t)b->in[b->in_pos++] + 1; 1058 s->lzma2.sequence = SEQ_COMPRESSED_0; 1059 break; 1060 1061 case SEQ_COMPRESSED_0: 1062 s->lzma2.compressed 1063 = (uint32_t)b->in[b->in_pos++] << 8; 1064 s->lzma2.sequence = SEQ_COMPRESSED_1; 1065 break; 1066 1067 case SEQ_COMPRESSED_1: 1068 s->lzma2.compressed 1069 += (uint32_t)b->in[b->in_pos++] + 1; 1070 s->lzma2.sequence = s->lzma2.next_sequence; 1071 break; 1072 1073 case SEQ_PROPERTIES: 1074 if (!lzma_props(s, b->in[b->in_pos++])) 1075 return XZ_DATA_ERROR; 1076 1077 s->lzma2.sequence = SEQ_LZMA_PREPARE; 1078 1079 /* Fall through */ 1080 1081 case SEQ_LZMA_PREPARE: 1082 if (s->lzma2.compressed < RC_INIT_BYTES) 1083 return XZ_DATA_ERROR; 1084 1085 if (!rc_read_init(&s->rc, b)) 1086 return XZ_OK; 1087 1088 s->lzma2.compressed -= RC_INIT_BYTES; 1089 s->lzma2.sequence = SEQ_LZMA_RUN; 1090 1091 /* Fall through */ 1092 1093 case SEQ_LZMA_RUN: 1094 /* 1095 * Set dictionary limit to indicate how much we want 1096 * to be encoded at maximum. Decode new data into the 1097 * dictionary. Flush the new data from dictionary to 1098 * b->out. Check if we finished decoding this chunk. 1099 * In case the dictionary got full but we didn't fill 1100 * the output buffer yet, we may run this loop 1101 * multiple times without changing s->lzma2.sequence. 1102 */ 1103 dict_limit(&s->dict, min_t(size_t, 1104 b->out_size - b->out_pos, 1105 s->lzma2.uncompressed)); 1106 if (!lzma2_lzma(s, b)) 1107 return XZ_DATA_ERROR; 1108 1109 s->lzma2.uncompressed -= dict_flush(&s->dict, b); 1110 1111 if (s->lzma2.uncompressed == 0) { 1112 if (s->lzma2.compressed > 0 || s->lzma.len > 0 1113 || !rc_is_finished(&s->rc)) 1114 return XZ_DATA_ERROR; 1115 1116 rc_reset(&s->rc); 1117 s->lzma2.sequence = SEQ_CONTROL; 1118 1119 } else if (b->out_pos == b->out_size 1120 || (b->in_pos == b->in_size 1121 && s->temp.size 1122 < s->lzma2.compressed)) { 1123 return XZ_OK; 1124 } 1125 1126 break; 1127 1128 case SEQ_COPY: 1129 dict_uncompressed(&s->dict, b, &s->lzma2.compressed); 1130 if (s->lzma2.compressed > 0) 1131 return XZ_OK; 1132 1133 s->lzma2.sequence = SEQ_CONTROL; 1134 break; 1135 } 1136 } 1137 1138 return XZ_OK; 1139 } 1140 1141 XZ_EXTERN struct xz_dec_lzma2 *xz_dec_lzma2_create(enum xz_mode mode, 1142 uint32_t dict_max) 1143 { 1144 struct xz_dec_lzma2 *s = kmalloc(sizeof(*s), GFP_KERNEL); 1145 if (s == NULL) 1146 return NULL; 1147 1148 s->dict.mode = mode; 1149 s->dict.size_max = dict_max; 1150 1151 if (DEC_IS_PREALLOC(mode)) { 1152 s->dict.buf = vmalloc(dict_max); 1153 if (s->dict.buf == NULL) { 1154 kfree(s); 1155 return NULL; 1156 } 1157 } else if (DEC_IS_DYNALLOC(mode)) { 1158 s->dict.buf = NULL; 1159 s->dict.allocated = 0; 1160 } 1161 1162 return s; 1163 } 1164 1165 XZ_EXTERN enum xz_ret xz_dec_lzma2_reset(struct xz_dec_lzma2 *s, uint8_t props) 1166 { 1167 /* This limits dictionary size to 3 GiB to keep parsing simpler. */ 1168 if (props > 39) 1169 return XZ_OPTIONS_ERROR; 1170 1171 s->dict.size = 2 + (props & 1); 1172 s->dict.size <<= (props >> 1) + 11; 1173 1174 if (DEC_IS_MULTI(s->dict.mode)) { 1175 if (s->dict.size > s->dict.size_max) 1176 return XZ_MEMLIMIT_ERROR; 1177 1178 s->dict.end = s->dict.size; 1179 1180 if (DEC_IS_DYNALLOC(s->dict.mode)) { 1181 if (s->dict.allocated < s->dict.size) { 1182 s->dict.allocated = s->dict.size; 1183 vfree(s->dict.buf); 1184 s->dict.buf = vmalloc(s->dict.size); 1185 if (s->dict.buf == NULL) { 1186 s->dict.allocated = 0; 1187 return XZ_MEM_ERROR; 1188 } 1189 } 1190 } 1191 } 1192 1193 s->lzma2.sequence = SEQ_CONTROL; 1194 s->lzma2.need_dict_reset = true; 1195 1196 s->temp.size = 0; 1197 1198 return XZ_OK; 1199 } 1200 1201 XZ_EXTERN void xz_dec_lzma2_end(struct xz_dec_lzma2 *s) 1202 { 1203 if (DEC_IS_MULTI(s->dict.mode)) 1204 vfree(s->dict.buf); 1205 1206 kfree(s); 1207 } 1208 1209 #ifdef XZ_DEC_MICROLZMA 1210 /* This is a wrapper struct to have a nice struct name in the public API. */ 1211 struct xz_dec_microlzma { 1212 struct xz_dec_lzma2 s; 1213 }; 1214 1215 enum xz_ret xz_dec_microlzma_run(struct xz_dec_microlzma *s_ptr, 1216 struct xz_buf *b) 1217 { 1218 struct xz_dec_lzma2 *s = &s_ptr->s; 1219 1220 /* 1221 * sequence is SEQ_PROPERTIES before the first input byte, 1222 * SEQ_LZMA_PREPARE until a total of five bytes have been read, 1223 * and SEQ_LZMA_RUN for the rest of the input stream. 1224 */ 1225 if (s->lzma2.sequence != SEQ_LZMA_RUN) { 1226 if (s->lzma2.sequence == SEQ_PROPERTIES) { 1227 /* One byte is needed for the props. */ 1228 if (b->in_pos >= b->in_size) 1229 return XZ_OK; 1230 1231 /* 1232 * Don't increment b->in_pos here. The same byte is 1233 * also passed to rc_read_init() which will ignore it. 1234 */ 1235 if (!lzma_props(s, ~b->in[b->in_pos])) 1236 return XZ_DATA_ERROR; 1237 1238 s->lzma2.sequence = SEQ_LZMA_PREPARE; 1239 } 1240 1241 /* 1242 * xz_dec_microlzma_reset() doesn't validate the compressed 1243 * size so we do it here. We have to limit the maximum size 1244 * to avoid integer overflows in lzma2_lzma(). 3 GiB is a nice 1245 * round number and much more than users of this code should 1246 * ever need. 1247 */ 1248 if (s->lzma2.compressed < RC_INIT_BYTES 1249 || s->lzma2.compressed > (3U << 30)) 1250 return XZ_DATA_ERROR; 1251 1252 if (!rc_read_init(&s->rc, b)) 1253 return XZ_OK; 1254 1255 s->lzma2.compressed -= RC_INIT_BYTES; 1256 s->lzma2.sequence = SEQ_LZMA_RUN; 1257 1258 dict_reset(&s->dict, b); 1259 } 1260 1261 /* This is to allow increasing b->out_size between calls. */ 1262 if (DEC_IS_SINGLE(s->dict.mode)) 1263 s->dict.end = b->out_size - b->out_pos; 1264 1265 while (true) { 1266 dict_limit(&s->dict, min_t(size_t, b->out_size - b->out_pos, 1267 s->lzma2.uncompressed)); 1268 1269 if (!lzma2_lzma(s, b)) 1270 return XZ_DATA_ERROR; 1271 1272 s->lzma2.uncompressed -= dict_flush(&s->dict, b); 1273 1274 if (s->lzma2.uncompressed == 0) { 1275 if (s->lzma2.pedantic_microlzma) { 1276 if (s->lzma2.compressed > 0 || s->lzma.len > 0 1277 || !rc_is_finished(&s->rc)) 1278 return XZ_DATA_ERROR; 1279 } 1280 1281 return XZ_STREAM_END; 1282 } 1283 1284 if (b->out_pos == b->out_size) 1285 return XZ_OK; 1286 1287 if (b->in_pos == b->in_size 1288 && s->temp.size < s->lzma2.compressed) 1289 return XZ_OK; 1290 } 1291 } 1292 1293 struct xz_dec_microlzma *xz_dec_microlzma_alloc(enum xz_mode mode, 1294 uint32_t dict_size) 1295 { 1296 struct xz_dec_microlzma *s; 1297 1298 /* Restrict dict_size to the same range as in the LZMA2 code. */ 1299 if (dict_size < 4096 || dict_size > (3U << 30)) 1300 return NULL; 1301 1302 s = kmalloc(sizeof(*s), GFP_KERNEL); 1303 if (s == NULL) 1304 return NULL; 1305 1306 s->s.dict.mode = mode; 1307 s->s.dict.size = dict_size; 1308 1309 if (DEC_IS_MULTI(mode)) { 1310 s->s.dict.end = dict_size; 1311 1312 s->s.dict.buf = vmalloc(dict_size); 1313 if (s->s.dict.buf == NULL) { 1314 kfree(s); 1315 return NULL; 1316 } 1317 } 1318 1319 return s; 1320 } 1321 1322 void xz_dec_microlzma_reset(struct xz_dec_microlzma *s, uint32_t comp_size, 1323 uint32_t uncomp_size, int uncomp_size_is_exact) 1324 { 1325 /* 1326 * comp_size is validated in xz_dec_microlzma_run(). 1327 * uncomp_size can safely be anything. 1328 */ 1329 s->s.lzma2.compressed = comp_size; 1330 s->s.lzma2.uncompressed = uncomp_size; 1331 s->s.lzma2.pedantic_microlzma = uncomp_size_is_exact; 1332 1333 s->s.lzma2.sequence = SEQ_PROPERTIES; 1334 s->s.temp.size = 0; 1335 } 1336 1337 void xz_dec_microlzma_end(struct xz_dec_microlzma *s) 1338 { 1339 if (DEC_IS_MULTI(s->s.dict.mode)) 1340 vfree(s->s.dict.buf); 1341 1342 kfree(s); 1343 } 1344 #endif 1345