src/liblzma/common/stream_decoder.c

   1 ///////////////////////////////////////////////////////////////////////////////
   2 //
   3 /// \file       stream_decoder.c
   4 /// \brief      Decodes .xz Streams
   5 //
   6 //  Author:     Lasse Collin
   7 //
   8 //  This file has been put into the public domain.
   9 //  You can do whatever you want with this file.
  10 //
  11 ///////////////////////////////////////////////////////////////////////////////
  12
  13 #include "stream_decoder.h"
  14 #include "block_decoder.h"
  15
  16
  17 struct lzma_coder_s {
  18         enum {
  19                 SEQ_STREAM_HEADER,
  20                 SEQ_BLOCK_HEADER,
  21                 SEQ_BLOCK,
  22                 SEQ_INDEX,
  23                 SEQ_STREAM_FOOTER,
  24                 SEQ_STREAM_PADDING,
  25         } sequence;
  26
  27         /// Block or Metadata decoder. This takes little memory and the same
  28         /// data structure can be used to decode every Block Header, so it's
  29         /// a good idea to have a separate lzma_next_coder structure for it.
  30         lzma_next_coder block_decoder;
  31
  32         /// Block options decoded by the Block Header decoder and used by
  33         /// the Block decoder.
  34         lzma_block block_options;
  35
  36         /// Stream Flags from Stream Header
  37         lzma_stream_flags stream_flags;
  38
  39         /// Index is hashed so that it can be compared to the sizes of Blocks
  40         /// with O(1) memory usage.
  41         lzma_index_hash *index_hash;
  42
  43         /// Memory usage limit
  44         uint64_t memlimit;
  45
  46         /// Amount of memory actually needed (only an estimate)
  47         uint64_t memusage;
  48
  49         /// If true, LZMA_NO_CHECK is returned if the Stream has
  50         /// no integrity check.
  51         bool tell_no_check;
  52
  53         /// If true, LZMA_UNSUPPORTED_CHECK is returned if the Stream has
  54         /// an integrity check that isn't supported by this liblzma build.
  55         bool tell_unsupported_check;
  56
  57         /// If true, LZMA_GET_CHECK is returned after decoding Stream Header.
  58         bool tell_any_check;
  59
  60         /// If true, we will decode concatenated Streams that possibly have
  61         /// Stream Padding between or after them. LZMA_STREAM_END is returned
  62         /// once the application isn't giving us any new input, and we aren't
  63         /// in the middle of a Stream, and possible Stream Padding is a
  64         /// multiple of four bytes.
  65         bool concatenated;
  66
  67         /// When decoding concatenated Streams, this is true as long as we
  68         /// are decoding the first Stream. This is needed to avoid misleading
  69         /// LZMA_FORMAT_ERROR in case the later Streams don't have valid magic
  70         /// bytes.
  71         bool first_stream;
  72
  73         /// Write position in buffer[] and position in Stream Padding
  74         size_t pos;
  75
  76         /// Buffer to hold Stream Header, Block Header, and Stream Footer.
  77         /// Block Header has biggest maximum size.
  78         uint8_t buffer[LZMA_BLOCK_HEADER_SIZE_MAX];
  79 };
  80
  81
  82 static lzma_ret
  83 stream_decoder_reset(lzma_coder *coder, lzma_allocator *allocator)
  84 {
  85         // Initialize the Index hash used to verify the Index.
  86         coder->index_hash = lzma_index_hash_init(coder->index_hash, allocator);
  87         if (coder->index_hash == NULL)
  88                 return LZMA_MEM_ERROR;
  89
  90         // Reset the rest of the variables.
  91         coder->sequence = SEQ_STREAM_HEADER;
  92         coder->pos = 0;
  93
  94         return LZMA_OK;
  95 }
  96
  97
  98 static lzma_ret
  99 stream_decode(lzma_coder *coder, lzma_allocator *allocator,
 100                 const uint8_t *restrict in, size_t *restrict in_pos,
 101                 size_t in_size, uint8_t *restrict out,
 102                 size_t *restrict out_pos, size_t out_size, lzma_action action)
 103 {
 104         // When decoding the actual Block, it may be able to produce more
 105         // output even if we don't give it any new input.
 106         while (true)
 107         switch (coder->sequence) {
 108         case SEQ_STREAM_HEADER: {
 109                 // Copy the Stream Header to the internal buffer.
 110                 lzma_bufcpy(in, in_pos, in_size, coder->buffer, &coder->pos,
 111                                 LZMA_STREAM_HEADER_SIZE);
 112
 113                 // Return if we didn't get the whole Stream Header yet.
 114                 if (coder->pos < LZMA_STREAM_HEADER_SIZE)
 115                         return LZMA_OK;
 116
 117                 coder->pos = 0;
 118
 119                 // Decode the Stream Header.
 120                 const lzma_ret ret = lzma_stream_header_decode(
 121                                 &coder->stream_flags, coder->buffer);
 122                 if (ret != LZMA_OK)
 123                         return ret == LZMA_FORMAT_ERROR && !coder->first_stream
 124                                         ? LZMA_DATA_ERROR : ret;
 125
 126                 // If we are decoding concatenated Streams, and the later
 127                 // Streams have invalid Header Magic Bytes, we give
 128                 // LZMA_DATA_ERROR instead of LZMA_FORMAT_ERROR.
 129                 coder->first_stream = false;
 130
 131                 // Copy the type of the Check so that Block Header and Block
 132                 // decoders see it.
 133                 coder->block_options.check = coder->stream_flags.check;
 134
 135                 // Even if we return LZMA_*_CHECK below, we want
 136                 // to continue from Block Header decoding.
 137                 coder->sequence = SEQ_BLOCK_HEADER;
 138
 139                 // Detect if there's no integrity check or if it is
 140                 // unsupported if those were requested by the application.
 141                 if (coder->tell_no_check && coder->stream_flags.check
 142                                 == LZMA_CHECK_NONE)
 143                         return LZMA_NO_CHECK;
 144
 145                 if (coder->tell_unsupported_check
 146                                 && !lzma_check_is_supported(
 147                                         coder->stream_flags.check))
 148                         return LZMA_UNSUPPORTED_CHECK;
 149
 150                 if (coder->tell_any_check)
 151                         return LZMA_GET_CHECK;
 152         }
 153
 154         // Fall through
 155
 156         case SEQ_BLOCK_HEADER: {
 157                 if (*in_pos >= in_size)
 158                         return LZMA_OK;
 159
 160                 if (coder->pos == 0) {
 161                         // Detect if it's Index.
 162                         if (in[*in_pos] == 0x00) {
 163                                 coder->sequence = SEQ_INDEX;
 164                                 break;
 165                         }
 166
 167                         // Calculate the size of the Block Header. Note that
 168                         // Block Header decoder wants to see this byte too
 169                         // so don't advance *in_pos.
 170                         coder->block_options.header_size
 171                                         = lzma_block_header_size_decode(
 172                                                 in[*in_pos]);
 173                 }
 174
 175                 // Copy the Block Header to the internal buffer.
 176                 lzma_bufcpy(in, in_pos, in_size, coder->buffer, &coder->pos,
 177                                 coder->block_options.header_size);
 178
 179                 // Return if we didn't get the whole Block Header yet.
 180                 if (coder->pos < coder->block_options.header_size)
 181                         return LZMA_OK;
 182
 183                 coder->pos = 0;
 184
 185                 // Version 0 is currently the only possible version.
 186                 coder->block_options.version = 0;
 187
 188                 // Set up a buffer to hold the filter chain. Block Header
 189                 // decoder will initialize all members of this array so
 190                 // we don't need to do it here.
 191                 lzma_filter filters[LZMA_FILTERS_MAX + 1];
 192                 coder->block_options.filters = filters;
 193
 194                 // Decode the Block Header.
 195                 return_if_error(lzma_block_header_decode(&coder->block_options,
 196                                 allocator, coder->buffer));
 197
 198                 // Check the memory usage limit.
 199                 const uint64_t memusage = lzma_raw_decoder_memusage(filters);
 200                 lzma_ret ret;
 201
 202                 if (memusage == UINT64_MAX) {
 203                         // One or more unknown Filter IDs.
 204                         ret = LZMA_OPTIONS_ERROR;
 205                 } else {
 206                         // Now we can set coder->memusage since we know that
 207                         // the filter chain is valid. We don't want
 208                         // lzma_memusage() to return UINT64_MAX in case of
 209                         // invalid filter chain.
 210                         coder->memusage = memusage;
 211
 212                         if (memusage > coder->memlimit) {
 213                                 // The chain would need too much memory.
 214                                 ret = LZMA_MEMLIMIT_ERROR;
 215                         } else {
 216                                 // Memory usage is OK.
 217                                 // Initialize the Block decoder.
 218                                 ret = lzma_block_decoder_init(
 219                                                 &coder->block_decoder,
 220                                                 allocator,
 221                                                 &coder->block_options);
 222                         }
 223                 }
 224
 225                 // Free the allocated filter options since they are needed
 226                 // only to initialize the Block decoder.
 227                 size_t i;
 228                 for (i = 0; i < LZMA_FILTERS_MAX; ++i)
 229                         lzma_free(filters[i].options, allocator);
 230
 231                 coder->block_options.filters = NULL;
 232
 233                 // Check if memory usage calculation and Block enocoder
 234                 // initialization succeeded.
 235                 if (ret != LZMA_OK)
 236                         return ret;
 237
 238                 coder->sequence = SEQ_BLOCK;
 239         }
 240
 241         // Fall through
 242
 243         case SEQ_BLOCK: {
 244                 const lzma_ret ret = coder->block_decoder.code(
 245                                 coder->block_decoder.coder, allocator,
 246                                 in, in_pos, in_size, out, out_pos, out_size,
 247                                 action);
 248
 249                 if (ret != LZMA_STREAM_END)
 250                         return ret;
 251
 252                 // Block decoded successfully. Add the new size pair to
 253                 // the Index hash.
 254                 return_if_error(lzma_index_hash_append(coder->index_hash,
 255                                 lzma_block_unpadded_size(
 256                                         &coder->block_options),
 257                                 coder->block_options.uncompressed_size));
 258
 259                 coder->sequence = SEQ_BLOCK_HEADER;
 260                 break;
 261         }
 262
 263         case SEQ_INDEX: {
 264                 // If we don't have any input, don't call
 265                 // lzma_index_hash_decode() since it would return
 266                 // LZMA_BUF_ERROR, which we must not do here.
 267                 if (*in_pos >= in_size)
 268                         return LZMA_OK;
 269
 270                 // Decode the Index and compare it to the hash calculated
 271                 // from the sizes of the Blocks (if any).
 272                 const lzma_ret ret = lzma_index_hash_decode(coder->index_hash,
 273                                 in, in_pos, in_size);
 274                 if (ret != LZMA_STREAM_END)
 275                         return ret;
 276
 277                 coder->sequence = SEQ_STREAM_FOOTER;
 278         }
 279
 280         // Fall through
 281
 282         case SEQ_STREAM_FOOTER: {
 283                 // Copy the Stream Footer to the internal buffer.
 284                 lzma_bufcpy(in, in_pos, in_size, coder->buffer, &coder->pos,
 285                                 LZMA_STREAM_HEADER_SIZE);
 286
 287                 // Return if we didn't get the whole Stream Footer yet.
 288                 if (coder->pos < LZMA_STREAM_HEADER_SIZE)
 289                         return LZMA_OK;
 290
 291                 coder->pos = 0;
 292
 293                 // Decode the Stream Footer. The decoder gives
 294                 // LZMA_FORMAT_ERROR if the magic bytes don't match,
 295                 // so convert that return code to LZMA_DATA_ERROR.
 296                 lzma_stream_flags footer_flags;
 297                 const lzma_ret ret = lzma_stream_footer_decode(
 298                                 &footer_flags, coder->buffer);
 299                 if (ret != LZMA_OK)
 300                         return ret == LZMA_FORMAT_ERROR
 301                                         ? LZMA_DATA_ERROR : ret;
 302
 303                 // Check that Index Size stored in the Stream Footer matches
 304                 // the real size of the Index field.
 305                 if (lzma_index_hash_size(coder->index_hash)
 306                                 != footer_flags.backward_size)
 307                         return LZMA_DATA_ERROR;
 308
 309                 // Compare that the Stream Flags fields are identical in
 310                 // both Stream Header and Stream Footer.
 311                 return_if_error(lzma_stream_flags_compare(
 312                                 &coder->stream_flags, &footer_flags));
 313
 314                 if (!coder->concatenated)
 315                         return LZMA_STREAM_END;
 316
 317                 coder->sequence = SEQ_STREAM_PADDING;
 318         }
 319
 320         // Fall through
 321
 322         case SEQ_STREAM_PADDING:
 323                 assert(coder->concatenated);
 324
 325                 // Skip over possible Stream Padding.
 326                 while (true) {
 327                         if (*in_pos >= in_size) {
 328                                 // Unless LZMA_FINISH was used, we cannot
 329                                 // know if there's more input coming later.
 330                                 if (action != LZMA_FINISH)
 331                                         return LZMA_OK;
 332
 333                                 // Stream Padding must be a multiple of
 334                                 // four bytes.
 335                                 return coder->pos == 0
 336                                                 ? LZMA_STREAM_END
 337                                                 : LZMA_DATA_ERROR;
 338                         }
 339
 340                         // If the byte is not zero, it probably indicates
 341                         // beginning of a new Stream (or the file is corrupt).
 342                         if (in[*in_pos] != 0x00)
 343                                 break;
 344
 345                         ++*in_pos;
 346                         coder->pos = (coder->pos + 1) & 3;
 347                 }
 348
 349                 // Stream Padding must be a multiple of four bytes (empty
 350                 // Stream Padding is OK).
 351                 if (coder->pos != 0) {
 352                         ++*in_pos;
 353                         return LZMA_DATA_ERROR;
 354                 }
 355
 356                 // Prepare to decode the next Stream.
 357                 return_if_error(stream_decoder_reset(coder, allocator));
 358                 break;
 359
 360         default:
 361                 assert(0);
 362                 return LZMA_PROG_ERROR;
 363         }
 364
 365         // Never reached
 366 }
 367
 368
 369 static void
 370 stream_decoder_end(lzma_coder *coder, lzma_allocator *allocator)
 371 {
 372         lzma_next_end(&coder->block_decoder, allocator);
 373         lzma_index_hash_end(coder->index_hash, allocator);
 374         lzma_free(coder, allocator);
 375         return;
 376 }
 377
 378
 379 static lzma_check
 380 stream_decoder_get_check(const lzma_coder *coder)
 381 {
 382         return coder->stream_flags.check;
 383 }
 384
 385
 386 static lzma_ret
 387 stream_decoder_memconfig(lzma_coder *coder, uint64_t *memusage,
 388                 uint64_t *old_memlimit, uint64_t new_memlimit)
 389 {
 390         *memusage = coder->memusage;
 391         *old_memlimit = coder->memlimit;
 392
 393         if (new_memlimit != 0) {
 394                 if (new_memlimit < coder->memusage)
 395                         return LZMA_MEMLIMIT_ERROR;
 396
 397                 coder->memlimit = new_memlimit;
 398         }
 399
 400         return LZMA_OK;
 401 }
 402
 403
 404 extern lzma_ret
 405 lzma_stream_decoder_init(lzma_next_coder *next, lzma_allocator *allocator,
 406                 uint64_t memlimit, uint32_t flags)
 407 {
 408         lzma_next_coder_init(&lzma_stream_decoder_init, next, allocator);
 409
 410         if (memlimit == 0)
 411                 return LZMA_PROG_ERROR;
 412
 413         if (flags & ~LZMA_SUPPORTED_FLAGS)
 414                 return LZMA_OPTIONS_ERROR;
 415
 416         if (next->coder == NULL) {
 417                 next->coder = lzma_alloc(sizeof(lzma_coder), allocator);
 418                 if (next->coder == NULL)
 419                         return LZMA_MEM_ERROR;
 420
 421                 next->code = &stream_decode;
 422                 next->end = &stream_decoder_end;
 423                 next->get_check = &stream_decoder_get_check;
 424                 next->memconfig = &stream_decoder_memconfig;
 425
 426                 next->coder->block_decoder = LZMA_NEXT_CODER_INIT;
 427                 next->coder->index_hash = NULL;
 428         }
 429
 430         next->coder->memlimit = memlimit;
 431         next->coder->memusage = LZMA_MEMUSAGE_BASE;
 432         next->coder->tell_no_check = (flags & LZMA_TELL_NO_CHECK) != 0;
 433         next->coder->tell_unsupported_check
 434                         = (flags & LZMA_TELL_UNSUPPORTED_CHECK) != 0;
 435         next->coder->tell_any_check = (flags & LZMA_TELL_ANY_CHECK) != 0;
 436         next->coder->concatenated = (flags & LZMA_CONCATENATED) != 0;
 437         next->coder->first_stream = true;
 438
 439         return stream_decoder_reset(next->coder, allocator);
 440 }
 441
 442
 443 extern LZMA_API(lzma_ret)
 444 lzma_stream_decoder(lzma_stream *strm, uint64_t memlimit, uint32_t flags)
 445 {
 446         lzma_next_strm_init(lzma_stream_decoder_init, strm, memlimit, flags);
 447
 448         strm->internal->supported_actions[LZMA_RUN] = true;
 449         strm->internal->supported_actions[LZMA_FINISH] = true;
 450
 451         return LZMA_OK;
 452 }