src/liblzma/common/metadata_decoder.c

   1 ///////////////////////////////////////////////////////////////////////////////
   2 //
   3 /// \file       metadata_decoder.c
   4 /// \brief      Decodes metadata stored in Metadata Blocks
   5 //
   6 //  Copyright (C) 2007 Lasse Collin
   7 //
   8 //  This library is free software; you can redistribute it and/or
   9 //  modify it under the terms of the GNU Lesser General Public
  10 //  License as published by the Free Software Foundation; either
  11 //  version 2.1 of the License, or (at your option) any later version.
  12 //
  13 //  This library is distributed in the hope that it will be useful,
  14 //  but WITHOUT ANY WARRANTY; without even the implied warranty of
  15 //  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  16 //  Lesser General Public License for more details.
  17 //
  18 ///////////////////////////////////////////////////////////////////////////////
  19
  20 #include "metadata_decoder.h"
  21 #include "block_decoder.h"
  22
  23
  24 /// Maximum size of a single Extra Record. Again, this is mostly to make
  25 /// sure that the parsed lzma_vli fits into size_t. Still, maybe this should
  26 /// be smaller.
  27 #define EXTRA_SIZE_MAX (SIZE_MAX / 4)
  28
  29
  30 struct lzma_coder_s {
  31         enum {
  32                 SEQ_FLAGS,
  33                 SEQ_HEADER_METADATA_SIZE,
  34                 SEQ_TOTAL_SIZE,
  35                 SEQ_UNCOMPRESSED_SIZE,
  36                 SEQ_INDEX_COUNT,
  37                 SEQ_INDEX_ALLOC,
  38                 SEQ_INDEX_TOTAL_SIZE,
  39                 SEQ_INDEX_UNCOMPRESSED_SIZE,
  40                 SEQ_EXTRA_PREPARE,
  41                 SEQ_EXTRA_ALLOC,
  42                 SEQ_EXTRA_ID,
  43                 SEQ_EXTRA_SIZE,
  44                 SEQ_EXTRA_DATA_ALLOC,
  45                 SEQ_EXTRA_DATA_COPY,
  46                 SEQ_EXTRA_DUMMY_ALLOC,
  47                 SEQ_EXTRA_DUMMY_ID,
  48                 SEQ_EXTRA_DUMMY_SIZE,
  49                 SEQ_EXTRA_DUMMY_COPY,
  50         } sequence;
  51
  52         /// Number of "things" left to be parsed. If we hit end of input
  53         /// when this isn't zero, we have corrupt Metadata Block.
  54         size_t todo_count;
  55
  56         /// Position in variable-length integers
  57         size_t pos;
  58
  59         /// Temporary variable needed to decode variables whose type
  60         /// is size_t instead of lzma_vli.
  61         lzma_vli tmp;
  62
  63         /// Pointer to target structure to hold the parsed results.
  64         lzma_metadata *metadata;
  65
  66         /// The Index Record we currently are parsing
  67         lzma_index *index_current;
  68
  69         /// Number of Records in Index
  70         size_t index_count;
  71
  72         /// Sum of Total Size fields in the Index
  73         lzma_vli index_total_size;
  74
  75         /// Sum of Uncompressed Size fields in the Index
  76         lzma_vli index_uncompressed_size;
  77
  78         /// True if Extra is present.
  79         bool has_extra;
  80
  81         /// True if we have been requested to store the Extra to *metadata.
  82         bool want_extra;
  83
  84         /// Pointer to the end of the Extra Record list.
  85         lzma_extra *extra_tail;
  86
  87         /// Dummy Extra Record used when only verifying integrity of Extra
  88         /// (not storing it to RAM).
  89         lzma_extra extra_dummy;
  90
  91         /// Block decoder
  92         lzma_next_coder block_decoder;
  93
  94         /// buffer[buffer_pos] is the next byte to process.
  95         size_t buffer_pos;
  96
  97         /// buffer[buffer_size] is the first byte to not process.
  98         size_t buffer_size;
  99
 100         /// Temporary buffer to which encoded Metadata is read before
 101         /// it is parsed.
 102         uint8_t buffer[LZMA_BUFFER_SIZE];
 103 };
 104
 105
 106 /// Reads a variable-length integer to coder->num.
 107 #define read_vli(num) \
 108 do { \
 109         const lzma_ret ret = lzma_vli_decode( \
 110                         &num, &coder->pos, \
 111                         coder->buffer, &coder->buffer_pos, \
 112                         coder->buffer_size); \
 113         if (ret != LZMA_STREAM_END) \
 114                 return ret; \
 115         \
 116         coder->pos = 0; \
 117 } while (0)
 118
 119
 120 static lzma_ret
 121 process(lzma_coder *coder, lzma_allocator *allocator)
 122 {
 123         while (coder->buffer_pos < coder->buffer_size)
 124         switch (coder->sequence) {
 125         case SEQ_FLAGS:
 126                 // Reserved bits must be unset.
 127                 if (coder->buffer[coder->buffer_pos] & 0x70)
 128                         return LZMA_HEADER_ERROR;
 129
 130                 // If Size of Header Metadata is present, prepare the
 131                 // variable for variable-length integer decoding. Otherwise
 132                 // set it to LZMA_VLI_VALUE_UNKNOWN to indicate that the
 133                 // field isn't present.
 134                 if (coder->buffer[coder->buffer_pos] & 0x01) {
 135                         coder->metadata->header_metadata_size = 0;
 136                         ++coder->todo_count;
 137                 }
 138
 139                 if (coder->buffer[coder->buffer_pos] & 0x02) {
 140                         coder->metadata->total_size = 0;
 141                         ++coder->todo_count;
 142                 }
 143
 144                 if (coder->buffer[coder->buffer_pos] & 0x04) {
 145                         coder->metadata->uncompressed_size = 0;
 146                         ++coder->todo_count;
 147                 }
 148
 149                 if (coder->buffer[coder->buffer_pos] & 0x08) {
 150                         // Setting index_count to 1 is just to indicate that
 151                         // Index is present. The real size is parsed later.
 152                         coder->index_count = 1;
 153                         ++coder->todo_count;
 154                 }
 155
 156                 coder->has_extra = (coder->buffer[coder->buffer_pos] & 0x80)
 157                                 != 0;
 158
 159                 ++coder->buffer_pos;
 160                 coder->sequence = SEQ_HEADER_METADATA_SIZE;
 161                 break;
 162
 163         case SEQ_HEADER_METADATA_SIZE:
 164                 if (coder->metadata->header_metadata_size
 165                                 != LZMA_VLI_VALUE_UNKNOWN) {
 166                         read_vli(coder->metadata->header_metadata_size);
 167
 168                         if (coder->metadata->header_metadata_size == 0)
 169                                 return LZMA_DATA_ERROR;
 170
 171                         --coder->todo_count;
 172                 } else {
 173                         // Zero indicates that Size of Header Metadata Block
 174                         // is not present. That is, after successful Metadata
 175                         // decoding, metadata->header_metadata_size is
 176                         // never LZMA_VLI_VALUE_UNKNOWN.
 177                         coder->metadata->header_metadata_size = 0;
 178                 }
 179
 180                 coder->sequence = SEQ_TOTAL_SIZE;
 181                 break;
 182
 183         case SEQ_TOTAL_SIZE:
 184                 if (coder->metadata->total_size != LZMA_VLI_VALUE_UNKNOWN) {
 185                         read_vli(coder->metadata->total_size);
 186
 187                         if (coder->metadata->total_size == 0)
 188                                 return LZMA_DATA_ERROR;
 189
 190                         --coder->todo_count;
 191                 }
 192
 193                 coder->sequence = SEQ_UNCOMPRESSED_SIZE;
 194                 break;
 195
 196         case SEQ_UNCOMPRESSED_SIZE:
 197                 if (coder->metadata->uncompressed_size
 198                                 != LZMA_VLI_VALUE_UNKNOWN) {
 199                         read_vli(coder->metadata->uncompressed_size);
 200                         --coder->todo_count;
 201                 }
 202
 203                 coder->sequence = SEQ_INDEX_COUNT;
 204                 break;
 205
 206         case SEQ_INDEX_COUNT:
 207                 if (coder->index_count == 0) {
 208                         coder->sequence = SEQ_EXTRA_PREPARE;
 209                         break;
 210                 }
 211
 212                 read_vli(coder->tmp);
 213
 214                 // Index must not be empty nor far too big (wouldn't fit
 215                 // in RAM).
 216                 if (coder->tmp == 0 || coder->tmp
 217                                 >= SIZE_MAX / sizeof(lzma_index))
 218                         return LZMA_DATA_ERROR;
 219
 220                 coder->index_count = (size_t)(coder->tmp);
 221                 coder->tmp = 0;
 222
 223                 coder->sequence = SEQ_INDEX_ALLOC;
 224                 break;
 225
 226         case SEQ_INDEX_ALLOC: {
 227                 lzma_index *i = lzma_alloc(sizeof(lzma_index), allocator);
 228                 if (i == NULL)
 229                         return LZMA_MEM_ERROR;
 230
 231                 i->total_size = 0;
 232                 i->uncompressed_size = 0;
 233                 i->next = NULL;
 234
 235                 if (coder->metadata->index == NULL)
 236                         coder->metadata->index = i;
 237                 else
 238                         coder->index_current->next = i;
 239
 240                 coder->index_current = i;
 241
 242                 coder->sequence = SEQ_INDEX_TOTAL_SIZE;
 243         }
 244
 245         // Fall through
 246
 247         case SEQ_INDEX_TOTAL_SIZE: {
 248                 read_vli(coder->index_current->total_size);
 249
 250                 coder->index_total_size += coder->index_current->total_size;
 251                 if (coder->index_total_size > LZMA_VLI_VALUE_MAX)
 252                         return LZMA_DATA_ERROR;
 253
 254                 // No Block can have Total Size of zero bytes.
 255                 if (coder->index_current->total_size == 0)
 256                         return LZMA_DATA_ERROR;
 257
 258                 if (--coder->index_count == 0) {
 259                         // If Total Size is present, it must match the sum
 260                         // of Total Sizes in Index.
 261                         if (coder->metadata->total_size
 262                                                 != LZMA_VLI_VALUE_UNKNOWN
 263                                         && coder->metadata->total_size
 264                                                 != coder->index_total_size)
 265                                 return LZMA_DATA_ERROR;
 266
 267                         coder->index_current = coder->metadata->index;
 268                         coder->sequence = SEQ_INDEX_UNCOMPRESSED_SIZE;
 269                 } else {
 270                         coder->sequence = SEQ_INDEX_ALLOC;
 271                 }
 272
 273                 break;
 274         }
 275
 276         case SEQ_INDEX_UNCOMPRESSED_SIZE: {
 277                 read_vli(coder->index_current->uncompressed_size);
 278
 279                 coder->index_uncompressed_size
 280                                 += coder->index_current->uncompressed_size;
 281                 if (coder->index_uncompressed_size > LZMA_VLI_VALUE_MAX)
 282                         return LZMA_DATA_ERROR;
 283
 284                 coder->index_current = coder->index_current->next;
 285                 if (coder->index_current == NULL) {
 286                         if (coder->metadata->uncompressed_size
 287                                                 != LZMA_VLI_VALUE_UNKNOWN
 288                                         && coder->metadata->uncompressed_size
 289                                         != coder->index_uncompressed_size)
 290                                 return LZMA_DATA_ERROR;
 291
 292                         --coder->todo_count;
 293                         coder->sequence = SEQ_EXTRA_PREPARE;
 294                 }
 295
 296                 break;
 297         }
 298
 299         case SEQ_EXTRA_PREPARE:
 300                 assert(coder->todo_count == 0);
 301
 302                 // If we get here, we have at least one byte of input left.
 303                 // If "Extra is present" flag is unset in Metadata Flags,
 304                 // it means that there is some garbage and we return an error.
 305                 if (!coder->has_extra)
 306                         return LZMA_DATA_ERROR;
 307
 308                 if (!coder->want_extra) {
 309                         coder->extra_tail = &coder->extra_dummy;
 310                         coder->sequence = SEQ_EXTRA_DUMMY_ALLOC;
 311                         break;
 312                 }
 313
 314                 coder->sequence = SEQ_EXTRA_ALLOC;
 315
 316         // Fall through
 317
 318         case SEQ_EXTRA_ALLOC: {
 319                 lzma_extra *e = lzma_alloc(sizeof(lzma_extra), allocator);
 320                 if (e == NULL)
 321                         return LZMA_MEM_ERROR;
 322
 323                 e->next = NULL;
 324                 e->id = 0;
 325                 e->size = 0;
 326                 e->data = NULL;
 327
 328                 if (coder->metadata->extra == NULL)
 329                         coder->metadata->extra = e;
 330                 else
 331                         coder->extra_tail->next = e;
 332
 333                 coder->extra_tail = e;
 334
 335                 coder->todo_count = 1;
 336                 coder->sequence = SEQ_EXTRA_ID;
 337         }
 338
 339         // Fall through
 340
 341         case SEQ_EXTRA_ID:
 342         case SEQ_EXTRA_DUMMY_ID:
 343                 read_vli(coder->extra_tail->id);
 344
 345                 if (coder->extra_tail->id == 0) {
 346                         coder->extra_tail->size = 0;
 347                         coder->extra_tail->data = NULL;
 348                         coder->todo_count = 0;
 349                         --coder->sequence;
 350                 } else {
 351                         ++coder->sequence;
 352                 }
 353
 354                 break;
 355
 356         case SEQ_EXTRA_SIZE:
 357         case SEQ_EXTRA_DUMMY_SIZE:
 358                 read_vli(coder->tmp);
 359                 ++coder->sequence;
 360                 break;
 361
 362         case SEQ_EXTRA_DATA_ALLOC: {
 363                 if (coder->tmp > EXTRA_SIZE_MAX)
 364                         return LZMA_DATA_ERROR;
 365
 366                 coder->extra_tail->size = (size_t)(coder->tmp);
 367                 coder->tmp = 0;
 368
 369                 uint8_t *d = lzma_alloc((size_t)(coder->extra_tail->size),
 370                                 allocator);
 371                 if (d == NULL)
 372                         return LZMA_MEM_ERROR;
 373
 374                 coder->extra_tail->data = d;
 375                 coder->sequence = SEQ_EXTRA_DATA_COPY;
 376         }
 377
 378         // Fall through
 379
 380         case SEQ_EXTRA_DATA_COPY:
 381                 bufcpy(coder->buffer, &coder->buffer_pos, coder->buffer_size,
 382                                 coder->extra_tail->data, &coder->pos,
 383                                 (size_t)(coder->extra_tail->size));
 384
 385                 if ((size_t)(coder->extra_tail->size) == coder->pos) {
 386                         coder->pos = 0;
 387                         coder->todo_count = 0;
 388                         coder->sequence = SEQ_EXTRA_ALLOC;
 389                 }
 390
 391                 break;
 392
 393         case SEQ_EXTRA_DUMMY_ALLOC:
 394                 // Not really alloc, just initialize the dummy entry.
 395                 coder->extra_dummy = (lzma_extra){
 396                         .next = NULL,
 397                         .id = 0,
 398                         .size = 0,
 399                         .data = NULL,
 400                 };
 401
 402                 coder->todo_count = 1;
 403                 coder->sequence = SEQ_EXTRA_DUMMY_ID;
 404                 break;
 405
 406         case SEQ_EXTRA_DUMMY_COPY: {
 407                 // Simply skip as many bytes as indicated by Extra Record Size.
 408                 // We don't check lzma_extra_size_max because we don't
 409                 // allocate any memory to hold the data.
 410                 const size_t in_avail = coder->buffer_size - coder->buffer_pos;
 411                 const size_t skip = MIN((lzma_vli)(in_avail), coder->tmp);
 412                 coder->buffer_pos += skip;
 413                 coder->tmp -= skip;
 414
 415                 if (coder->tmp == 0) {
 416                         coder->todo_count = 0;
 417                         coder->sequence = SEQ_EXTRA_DUMMY_ALLOC;
 418                 }
 419
 420                 break;
 421         }
 422
 423         default:
 424                 return LZMA_PROG_ERROR;
 425         }
 426
 427         return LZMA_OK;
 428 }
 429
 430
 431 static lzma_ret
 432 metadata_decode(lzma_coder *coder, lzma_allocator *allocator,
 433                 const uint8_t *restrict in, size_t *restrict in_pos,
 434                 size_t in_size, uint8_t *restrict out lzma_attribute((unused)),
 435                 size_t *restrict out_pos lzma_attribute((unused)),
 436                 size_t out_size lzma_attribute((unused)),
 437                 lzma_action action lzma_attribute((unused)))
 438 {
 439         bool end_was_reached = false;
 440
 441         while (true) {
 442                 // Fill the buffer if it is empty.
 443                 if (coder->buffer_pos == coder->buffer_size) {
 444                         coder->buffer_pos = 0;
 445                         coder->buffer_size = 0;
 446
 447                         const lzma_ret ret = coder->block_decoder.code(
 448                                         coder->block_decoder.coder, allocator,
 449                                         in, in_pos, in_size, coder->buffer,
 450                                         &coder->buffer_size, LZMA_BUFFER_SIZE,
 451                                         LZMA_RUN);
 452
 453                         switch (ret) {
 454                         case LZMA_OK:
 455                                 // Return immediatelly if we got no new data.
 456                                 if (coder->buffer_size == 0)
 457                                         return LZMA_OK;
 458
 459                                 break;
 460
 461                         case LZMA_STREAM_END:
 462                                 end_was_reached = true;
 463                                 break;
 464
 465                         default:
 466                                 return ret;
 467                         }
 468                 }
 469
 470                 // Process coder->buffer.
 471                 const lzma_ret ret = process(coder, allocator);
 472                 if (ret != LZMA_OK)
 473                         return ret;
 474
 475                 // On success, process() eats all the input.
 476                 assert(coder->buffer_pos == coder->buffer_size);
 477
 478                 if (end_was_reached) {
 479                         // Check that the sequence is not in the
 480                         // middle of anything.
 481                         if (coder->todo_count != 0)
 482                                 return LZMA_DATA_ERROR;
 483
 484                         return LZMA_STREAM_END;
 485                 }
 486         }
 487 }
 488
 489
 490 static void
 491 metadata_decoder_end(lzma_coder *coder, lzma_allocator *allocator)
 492 {
 493         lzma_next_coder_end(&coder->block_decoder, allocator);
 494         lzma_free(coder, allocator);
 495         return;
 496 }
 497
 498
 499 static lzma_ret
 500 metadata_decoder_init(lzma_next_coder *next, lzma_allocator *allocator,
 501                 lzma_options_block *options, lzma_metadata *metadata,
 502                 bool want_extra)
 503 {
 504         if (options == NULL || metadata == NULL)
 505                 return LZMA_PROG_ERROR;
 506
 507         if (next->coder == NULL) {
 508                 next->coder = lzma_alloc(sizeof(lzma_coder), allocator);
 509                 if (next->coder == NULL)
 510                         return LZMA_MEM_ERROR;
 511
 512                 next->code = &metadata_decode;
 513                 next->end = &metadata_decoder_end;
 514                 next->coder->block_decoder = LZMA_NEXT_CODER_INIT;
 515         }
 516
 517         metadata->header_metadata_size = LZMA_VLI_VALUE_UNKNOWN;
 518         metadata->total_size = LZMA_VLI_VALUE_UNKNOWN;
 519         metadata->uncompressed_size = LZMA_VLI_VALUE_UNKNOWN;
 520         metadata->index = NULL;
 521         metadata->extra = NULL;
 522
 523         next->coder->sequence = SEQ_FLAGS;
 524         next->coder->todo_count = 0;
 525         next->coder->pos = 0;
 526         next->coder->tmp = 0;
 527         next->coder->metadata = metadata;
 528         next->coder->index_current = NULL;
 529         next->coder->index_count = 0;
 530         next->coder->index_total_size = 0;
 531         next->coder->index_uncompressed_size = 0;
 532         next->coder->want_extra = want_extra;
 533         next->coder->extra_tail = NULL;
 534         next->coder->buffer_pos = 0;
 535         next->coder->buffer_size = 0;
 536
 537         return lzma_block_decoder_init(
 538                         &next->coder->block_decoder, allocator, options);
 539 }
 540
 541
 542 extern lzma_ret
 543 lzma_metadata_decoder_init(lzma_next_coder *next, lzma_allocator *allocator,
 544                 lzma_options_block *options, lzma_metadata *metadata,
 545                 bool want_extra)
 546 {
 547         lzma_next_coder_init(metadata_decoder_init, next, allocator,
 548                         options, metadata, want_extra);
 549 }
 550
 551
 552 extern LZMA_API lzma_ret
 553 lzma_metadata_decoder(lzma_stream *strm, lzma_options_block *options,
 554                 lzma_metadata *metadata, lzma_bool want_extra)
 555 {
 556         lzma_next_strm_init(strm, lzma_metadata_decoder_init,
 557                         options, metadata, want_extra);
 558
 559         strm->internal->supported_actions[LZMA_RUN] = true;
 560
 561         return LZMA_OK;
 562 }