wimlib.net Git - wimlib/blob - src/lookup_table.c

   1 /*
   2  * lookup_table.c
   3  *
   4  * Lookup table, implemented as a hash table, that maps SHA1 message digests to
   5  * data streams.
   6  */
   7
   8 /*
   9  * Copyright (C) 2012, 2013 Eric Biggers
  10  *
  11  * This file is part of wimlib, a library for working with WIM files.
  12  *
  13  * wimlib is free software; you can redistribute it and/or modify it under the
  14  * terms of the GNU General Public License as published by the Free
  15  * Software Foundation; either version 3 of the License, or (at your option)
  16  * any later version.
  17  *
  18  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  19  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  20  * A PARTICULAR PURPOSE. See the GNU General Public License for more
  21  * details.
  22  *
  23  * You should have received a copy of the GNU General Public License
  24  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  25  */
  26
  27 #include "wimlib_internal.h"
  28 #include "lookup_table.h"
  29 #include "buffer_io.h"
  30 #include <errno.h>
  31 #include <stdlib.h>
  32
  33 #ifdef WITH_FUSE
  34 #include <unistd.h>
  35 #endif
  36
  37 struct wim_lookup_table *
  38 new_lookup_table(size_t capacity)
  39 {
  40         struct wim_lookup_table *table;
  41         struct hlist_head *array;
  42
  43         table = CALLOC(1, sizeof(struct wim_lookup_table));
  44         if (table) {
  45                 array = CALLOC(capacity, sizeof(array[0]));
  46                 if (array) {
  47                         table->num_entries = 0;
  48                         table->capacity = capacity;
  49                         table->array = array;
  50                 } else {
  51                         FREE(table);
  52                         table = NULL;
  53                         ERROR("Failed to allocate memory for lookup table with capacity %zu",
  54                               capacity);
  55                 }
  56         }
  57         return table;
  58 }
  59
  60 struct wim_lookup_table_entry *
  61 new_lookup_table_entry()
  62 {
  63         struct wim_lookup_table_entry *lte;
  64
  65         lte = CALLOC(1, sizeof(struct wim_lookup_table_entry));
  66         if (lte) {
  67                 lte->part_number  = 1;
  68                 lte->refcnt       = 1;
  69         } else {
  70                 ERROR("Out of memory (tried to allocate %zu bytes for "
  71                       "lookup table entry)",
  72                       sizeof(struct wim_lookup_table_entry));
  73         }
  74         return lte;
  75 }
  76
  77 struct wim_lookup_table_entry *
  78 clone_lookup_table_entry(const struct wim_lookup_table_entry *old)
  79 {
  80         struct wim_lookup_table_entry *new;
  81
  82         new = MALLOC(sizeof(*new));
  83         if (!new)
  84                 return NULL;
  85
  86         memcpy(new, old, sizeof(*old));
  87         new->extracted_file = NULL;
  88         switch (new->resource_location) {
  89 #ifdef __WIN32__
  90         case RESOURCE_WIN32:
  91         case RESOURCE_WIN32_ENCRYPTED:
  92 #else
  93         case RESOURCE_IN_FILE_ON_DISK:
  94 #endif
  95 #ifdef WITH_FUSE
  96         case RESOURCE_IN_STAGING_FILE:
  97                 BUILD_BUG_ON((void*)&old->file_on_disk !=
  98                              (void*)&old->staging_file_name);
  99 #endif
 100                 new->file_on_disk = TSTRDUP(old->file_on_disk);
 101                 if (!new->file_on_disk)
 102                         goto out_free;
 103                 break;
 104         case RESOURCE_IN_ATTACHED_BUFFER:
 105                 new->attached_buffer = MALLOC(wim_resource_size(old));
 106                 if (!new->attached_buffer)
 107                         goto out_free;
 108                 memcpy(new->attached_buffer, old->attached_buffer,
 109                        wim_resource_size(old));
 110                 break;
 111 #ifdef WITH_NTFS_3G
 112         case RESOURCE_IN_NTFS_VOLUME:
 113                 if (old->ntfs_loc) {
 114                         struct ntfs_location *loc;
 115                         loc = MALLOC(sizeof(*loc));
 116                         if (!loc)
 117                                 goto out_free;
 118                         memcpy(loc, old->ntfs_loc, sizeof(*loc));
 119                         loc->path = NULL;
 120                         loc->stream_name = NULL;
 121                         new->ntfs_loc = loc;
 122                         loc->path = STRDUP(old->ntfs_loc->path);
 123                         if (!loc->path)
 124                                 goto out_free;
 125                         loc->stream_name = MALLOC((loc->stream_name_nchars + 1) * 2);
 126                         if (!loc->stream_name)
 127                                 goto out_free;
 128                         memcpy(loc->stream_name,
 129                                old->ntfs_loc->stream_name,
 130                                (loc->stream_name_nchars + 1) * 2);
 131                 }
 132                 break;
 133 #endif
 134         default:
 135                 break;
 136         }
 137         return new;
 138 out_free:
 139         free_lookup_table_entry(new);
 140         return NULL;
 141 }
 142
 143 void
 144 free_lookup_table_entry(struct wim_lookup_table_entry *lte)
 145 {
 146         if (lte) {
 147                 switch (lte->resource_location) {
 148         #ifdef __WIN32__
 149                 case RESOURCE_WIN32:
 150                 case RESOURCE_WIN32_ENCRYPTED:
 151         #else
 152                 case RESOURCE_IN_FILE_ON_DISK:
 153         #endif
 154         #ifdef WITH_FUSE
 155                 case RESOURCE_IN_STAGING_FILE:
 156                         BUILD_BUG_ON((void*)&lte->file_on_disk !=
 157                                      (void*)&lte->staging_file_name);
 158         #endif
 159                 case RESOURCE_IN_ATTACHED_BUFFER:
 160                         BUILD_BUG_ON((void*)&lte->file_on_disk !=
 161                                      (void*)&lte->attached_buffer);
 162                         FREE(lte->file_on_disk);
 163                         break;
 164 #ifdef WITH_NTFS_3G
 165                 case RESOURCE_IN_NTFS_VOLUME:
 166                         if (lte->ntfs_loc) {
 167                                 FREE(lte->ntfs_loc->path);
 168                                 FREE(lte->ntfs_loc->stream_name);
 169                                 FREE(lte->ntfs_loc);
 170                         }
 171                         break;
 172 #endif
 173                 default:
 174                         break;
 175                 }
 176                 FREE(lte);
 177         }
 178 }
 179
 180 static int
 181 do_free_lookup_table_entry(struct wim_lookup_table_entry *entry, void *ignore)
 182 {
 183         free_lookup_table_entry(entry);
 184         return 0;
 185 }
 186
 187
 188 void
 189 free_lookup_table(struct wim_lookup_table *table)
 190 {
 191         DEBUG2("Freeing lookup table");
 192         if (table) {
 193                 if (table->array) {
 194                         for_lookup_table_entry(table,
 195                                                do_free_lookup_table_entry,
 196                                                NULL);
 197                         FREE(table->array);
 198                 }
 199                 FREE(table);
 200         }
 201 }
 202
 203 /*
 204  * Inserts an entry into the lookup table.
 205  *
 206  * @table:      A pointer to the lookup table.
 207  * @lte:        A pointer to the entry to insert.
 208  */
 209 void
 210 lookup_table_insert(struct wim_lookup_table *table,
 211                     struct wim_lookup_table_entry *lte)
 212 {
 213         size_t i = lte->hash_short % table->capacity;
 214         hlist_add_head(&lte->hash_list, &table->array[i]);
 215
 216         /* XXX Make the table grow when too many entries have been inserted. */
 217         table->num_entries++;
 218 }
 219
 220 static void
 221 finalize_lte(struct wim_lookup_table_entry *lte)
 222 {
 223         #ifdef WITH_FUSE
 224         if (lte->resource_location == RESOURCE_IN_STAGING_FILE) {
 225                 unlink(lte->staging_file_name);
 226                 list_del(&lte->unhashed_list);
 227         }
 228         #endif
 229         free_lookup_table_entry(lte);
 230 }
 231
 232 /* Decrements the reference count for the lookup table entry @lte.  If its
 233  * reference count reaches 0, it is unlinked from the lookup table.  If,
 234  * furthermore, the entry has no opened file descriptors associated with it, the
 235  * entry is freed.  */
 236 void
 237 lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
 238                      struct wim_lookup_table *table)
 239 {
 240         wimlib_assert(lte != NULL);
 241         wimlib_assert(lte->refcnt != 0);
 242         if (--lte->refcnt == 0) {
 243                 if (!lte->unhashed)
 244                         lookup_table_unlink(table, lte);
 245         #ifdef WITH_FUSE
 246                 if (lte->num_opened_fds == 0)
 247         #endif
 248                         finalize_lte(lte);
 249         }
 250 }
 251
 252 #ifdef WITH_FUSE
 253 void
 254 lte_decrement_num_opened_fds(struct wim_lookup_table_entry *lte)
 255 {
 256         if (lte->num_opened_fds != 0)
 257                 if (--lte->num_opened_fds == 0 && lte->refcnt == 0)
 258                         finalize_lte(lte);
 259 }
 260 #endif
 261
 262 /* Calls a function on all the entries in the WIM lookup table.  Stop early and
 263  * return nonzero if any call to the function returns nonzero. */
 264 int
 265 for_lookup_table_entry(struct wim_lookup_table *table,
 266                        int (*visitor)(struct wim_lookup_table_entry *, void *),
 267                        void *arg)
 268 {
 269         struct wim_lookup_table_entry *lte;
 270         struct hlist_node *pos, *tmp;
 271         int ret;
 272
 273         for (size_t i = 0; i < table->capacity; i++) {
 274                 hlist_for_each_entry_safe(lte, pos, tmp, &table->array[i],
 275                                           hash_list)
 276                 {
 277                         wimlib_assert2(!(lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA));
 278                         ret = visitor(lte, arg);
 279                         if (ret)
 280                                 return ret;
 281                 }
 282         }
 283         return 0;
 284 }
 285
 286 int
 287 cmp_streams_by_wim_position(const void *p1, const void *p2)
 288 {
 289         const struct wim_lookup_table_entry *lte1, *lte2;
 290         lte1 = *(const struct wim_lookup_table_entry**)p1;
 291         lte2 = *(const struct wim_lookup_table_entry**)p2;
 292         if (lte1->resource_entry.offset < lte2->resource_entry.offset)
 293                 return -1;
 294         else if (lte1->resource_entry.offset > lte2->resource_entry.offset)
 295                 return 1;
 296         else
 297                 return 0;
 298 }
 299
 300
 301 static int
 302 add_lte_to_array(struct wim_lookup_table_entry *lte,
 303                  void *_pp)
 304 {
 305         struct wim_lookup_table_entry ***pp = _pp;
 306         *(*pp)++ = lte;
 307         return 0;
 308 }
 309
 310 /* Iterate through the lookup table entries, but first sort them by stream
 311  * offset in the WIM.  Caution: this is intended to be used when the stream
 312  * offset field has actually been set. */
 313 int
 314 for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
 315                                   int (*visitor)(struct wim_lookup_table_entry *,
 316                                                  void *),
 317                                   void *arg)
 318 {
 319         struct wim_lookup_table_entry **lte_array, **p;
 320         size_t num_streams = table->num_entries;
 321         int ret;
 322
 323         lte_array = MALLOC(num_streams * sizeof(lte_array[0]));
 324         if (!lte_array)
 325                 return WIMLIB_ERR_NOMEM;
 326         p = lte_array;
 327         for_lookup_table_entry(table, add_lte_to_array, &p);
 328
 329         wimlib_assert(p == lte_array + num_streams);
 330
 331         qsort(lte_array, num_streams, sizeof(lte_array[0]),
 332               cmp_streams_by_wim_position);
 333         ret = 0;
 334         for (size_t i = 0; i < num_streams; i++) {
 335                 ret = visitor(lte_array[i], arg);
 336                 if (ret)
 337                         break;
 338         }
 339         FREE(lte_array);
 340         return ret;
 341 }
 342
 343 /*
 344  * Reads the lookup table from a WIM file.
 345  *
 346  * Saves lookup table entries for non-metadata streams in a hash table, and
 347  * saves the metadata entry for each image in a special per-image location (the
 348  * image_metadata array).
 349  */
 350 int
 351 read_lookup_table(WIMStruct *w)
 352 {
 353         u64 num_entries;
 354         u8 buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
 355         int ret;
 356         struct wim_lookup_table *table;
 357         struct wim_lookup_table_entry *cur_entry, *duplicate_entry;
 358
 359         if (resource_is_compressed(&w->hdr.lookup_table_res_entry)) {
 360                 ERROR("Didn't expect a compressed lookup table!");
 361                 ERROR("Ask the author to implement support for this.");
 362                 return WIMLIB_ERR_COMPRESSED_LOOKUP_TABLE;
 363         }
 364
 365         DEBUG("Reading lookup table: offset %"PRIu64", size %"PRIu64"",
 366               w->hdr.lookup_table_res_entry.offset,
 367               w->hdr.lookup_table_res_entry.original_size);
 368
 369         if (fseeko(w->fp, w->hdr.lookup_table_res_entry.offset, SEEK_SET) != 0)
 370         {
 371                 ERROR_WITH_ERRNO("Failed to seek to byte %"PRIu64" to read "
 372                                  "lookup table",
 373                                  w->hdr.lookup_table_res_entry.offset);
 374                 return WIMLIB_ERR_READ;
 375         }
 376
 377         num_entries = w->hdr.lookup_table_res_entry.original_size /
 378                       WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE;
 379         table = new_lookup_table(num_entries * 2 + 1);
 380         if (!table)
 381                 return WIMLIB_ERR_NOMEM;
 382
 383         w->current_image = 0;
 384         while (num_entries--) {
 385                 const u8 *p;
 386
 387                 if (fread(buf, 1, sizeof(buf), w->fp) != sizeof(buf)) {
 388                         if (feof(w->fp)) {
 389                                 ERROR("Unexpected EOF in WIM lookup table!");
 390                         } else {
 391                                 ERROR_WITH_ERRNO("Error reading WIM lookup "
 392                                                  "table");
 393                         }
 394                         ret = WIMLIB_ERR_READ;
 395                         goto out_free_lookup_table;
 396                 }
 397                 cur_entry = new_lookup_table_entry();
 398                 if (!cur_entry) {
 399                         ret = WIMLIB_ERR_NOMEM;
 400                         goto out_free_lookup_table;
 401                 }
 402
 403                 cur_entry->wim = w;
 404                 cur_entry->resource_location = RESOURCE_IN_WIM;
 405                 p = get_resource_entry(buf, &cur_entry->resource_entry);
 406                 p = get_u16(p, &cur_entry->part_number);
 407                 p = get_u32(p, &cur_entry->refcnt);
 408                 p = get_bytes(p, SHA1_HASH_SIZE, cur_entry->hash);
 409
 410                 if (cur_entry->part_number != w->hdr.part_number) {
 411                         ERROR("A lookup table entry in part %hu of the WIM "
 412                               "points to part %hu",
 413                               w->hdr.part_number, cur_entry->part_number);
 414                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 415                         goto out_free_cur_entry;
 416                 }
 417
 418                 if (is_zero_hash(cur_entry->hash)) {
 419                         ERROR("The WIM lookup table contains an entry with a "
 420                               "SHA1 message digest of all 0's");
 421                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 422                         goto out_free_cur_entry;
 423                 }
 424
 425                 if (!(cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
 426                     && (cur_entry->resource_entry.size !=
 427                         cur_entry->resource_entry.original_size))
 428                 {
 429                 #ifdef ENABLE_ERROR_MESSAGES
 430                         ERROR("Found uncompressed resource with original size "
 431                               "not the same as compressed size");
 432                         ERROR("The lookup table entry for the resource is as follows:");
 433                         print_lookup_table_entry(cur_entry, stderr);
 434                 #endif
 435                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 436                         goto out_free_cur_entry;
 437                 }
 438
 439                 if (cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_METADATA) {
 440                         /* Lookup table entry for a metadata resource */
 441                         if (cur_entry->refcnt != 1) {
 442                         #ifdef ENABLE_ERROR_MESSAGES
 443                                 ERROR("Found metadata resource with refcnt != 1:");
 444                                 print_lookup_table_entry(cur_entry, stderr);
 445                         #endif
 446                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 447                                 goto out_free_cur_entry;
 448                         }
 449
 450                         if (w->hdr.part_number != 1) {
 451                                 ERROR("Found a metadata resource in a "
 452                                       "non-first part of the split WIM!");
 453                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 454                                 goto out_free_cur_entry;
 455                         }
 456                         if (w->current_image == w->hdr.image_count) {
 457                                 ERROR("The WIM header says there are %u images "
 458                                       "in the WIM, but we found more metadata "
 459                                       "resources than this", w->hdr.image_count);
 460                                 ret = WIMLIB_ERR_IMAGE_COUNT;
 461                                 goto out_free_cur_entry;
 462                         }
 463
 464                         /* Notice very carefully:  We are assigning the metadata
 465                          * resources in the exact order mirrored by their lookup
 466                          * table entries on disk, which is the behavior of
 467                          * Microsoft's software.  In particular, this overrides
 468                          * the actual locations of the metadata resources
 469                          * themselves in the WIM file as well as any information
 470                          * written in the XML data. */
 471                         DEBUG("Found metadata resource for image %u at "
 472                               "offset %"PRIu64".",
 473                               w->current_image + 1,
 474                               cur_entry->resource_entry.offset);
 475                         w->image_metadata[
 476                                 w->current_image++]->metadata_lte = cur_entry;
 477                 } else {
 478                         /* Lookup table entry for a stream that is not a
 479                          * metadata resource */
 480                         duplicate_entry = __lookup_resource(table, cur_entry->hash);
 481                         if (duplicate_entry) {
 482                         #ifdef ENABLE_ERROR_MESSAGES
 483                                 ERROR("The WIM lookup table contains two entries with the "
 484                                       "same SHA1 message digest!");
 485                                 ERROR("The first entry is:");
 486                                 print_lookup_table_entry(duplicate_entry, stderr);
 487                                 ERROR("The second entry is:");
 488                                 print_lookup_table_entry(cur_entry, stderr);
 489                         #endif
 490                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 491                                 goto out_free_cur_entry;
 492                         }
 493                         lookup_table_insert(table, cur_entry);
 494                 }
 495         }
 496
 497         if (w->hdr.part_number == 1 &&
 498             w->current_image != w->hdr.image_count)
 499         {
 500                 ERROR("The WIM header says there are %u images "
 501                       "in the WIM, but we only found %d metadata "
 502                       "resources!", w->hdr.image_count, w->current_image);
 503                 ret = WIMLIB_ERR_IMAGE_COUNT;
 504                 goto out_free_lookup_table;
 505         }
 506         DEBUG("Done reading lookup table.");
 507         w->lookup_table = table;
 508         ret = 0;
 509         goto out;
 510 out_free_cur_entry:
 511         FREE(cur_entry);
 512 out_free_lookup_table:
 513         free_lookup_table(table);
 514 out:
 515         w->current_image = 0;
 516         return ret;
 517 }
 518
 519
 520 /*
 521  * Writes a lookup table entry to the output file.
 522  */
 523 int
 524 write_lookup_table_entry(struct wim_lookup_table_entry *lte, void *_out)
 525 {
 526         FILE *out;
 527         u8 buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
 528         u8 *p;
 529
 530         out = _out;
 531
 532         /* Don't write entries that have not had file resources or metadata
 533          * resources written for them. */
 534         if (lte->out_refcnt == 0)
 535                 return 0;
 536
 537         if (lte->output_resource_entry.flags & WIM_RESHDR_FLAG_METADATA) {
 538                 DEBUG("Writing metadata entry at %"PRIu64" "
 539                       "(orig size = %"PRIu64")",
 540                       ftello(out), lte->output_resource_entry.original_size);
 541         }
 542
 543         p = put_resource_entry(buf, &lte->output_resource_entry);
 544         p = put_u16(p, lte->part_number);
 545         p = put_u32(p, lte->out_refcnt);
 546         p = put_bytes(p, SHA1_HASH_SIZE, lte->hash);
 547         if (fwrite(buf, 1, sizeof(buf), out) != sizeof(buf)) {
 548                 ERROR_WITH_ERRNO("Failed to write lookup table entry");
 549                 return WIMLIB_ERR_WRITE;
 550         }
 551         return 0;
 552 }
 553
 554 /* Writes the WIM lookup table to the output file. */
 555 int
 556 write_lookup_table(WIMStruct *w, int image, struct resource_entry *out_res_entry)
 557 {
 558         FILE *out = w->out_fp;
 559         off_t start_offset, end_offset;
 560         int ret;
 561         int start_image, end_image;
 562
 563         start_offset = ftello(out);
 564         if (start_offset == -1)
 565                 return WIMLIB_ERR_WRITE;
 566
 567         /* Write lookup table entries for metadata resources */
 568         if (image == WIMLIB_ALL_IMAGES) {
 569                 start_image = 1;
 570                 end_image = w->hdr.image_count;
 571         } else {
 572                 start_image = image;
 573                 end_image = image;
 574         }
 575         for (int i = start_image; i <= end_image; i++) {
 576                 struct wim_lookup_table_entry *metadata_lte;
 577
 578                 metadata_lte = w->image_metadata[i - 1]->metadata_lte;
 579                 metadata_lte->out_refcnt = 1;
 580                 metadata_lte->output_resource_entry.flags |= WIM_RESHDR_FLAG_METADATA;
 581                 ret = write_lookup_table_entry(metadata_lte, out);
 582                 if (ret)
 583                         return ret;
 584         }
 585
 586         /* Write lookup table entries for other resources */
 587         ret = for_lookup_table_entry(w->lookup_table, write_lookup_table_entry, out);
 588         if (ret)
 589                 return ret;
 590
 591         /* Fill in the resource entry for the lookup table itself */
 592         end_offset = ftello(out);
 593         if (end_offset == -1)
 594                 return WIMLIB_ERR_WRITE;
 595
 596         out_res_entry->offset        = start_offset;
 597         out_res_entry->size          = end_offset - start_offset;
 598         out_res_entry->original_size = end_offset - start_offset;
 599         out_res_entry->flags         = WIM_RESHDR_FLAG_METADATA;
 600         return 0;
 601 }
 602
 603 int
 604 lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 605 {
 606         lte->real_refcnt = 0;
 607         return 0;
 608 }
 609
 610 int
 611 lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 612 {
 613         lte->out_refcnt = 0;
 614         return 0;
 615 }
 616
 617 int
 618 lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *_ignore)
 619 {
 620         if (lte->extracted_file != NULL) {
 621                 FREE(lte->extracted_file);
 622                 lte->extracted_file = NULL;
 623         }
 624         return 0;
 625 }
 626
 627 void
 628 print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 629 {
 630         if (!lte) {
 631                 tputc(T('\n'), out);
 632                 return;
 633         }
 634         tfprintf(out, T("Offset            = %"PRIu64" bytes\n"),
 635                  lte->resource_entry.offset);
 636
 637         tfprintf(out, T("Size              = %"PRIu64" bytes\n"),
 638                  (u64)lte->resource_entry.size);
 639
 640         tfprintf(out, T("Original size     = %"PRIu64" bytes\n"),
 641                  lte->resource_entry.original_size);
 642
 643         tfprintf(out, T("Part Number       = %hu\n"), lte->part_number);
 644         tfprintf(out, T("Reference Count   = %u\n"), lte->refcnt);
 645
 646         if (lte->unhashed) {
 647                 tfprintf(out, T("(Unhashed: inode %p, stream_id = %u)\n"),
 648                          lte->back_inode, lte->back_stream_id);
 649         } else {
 650                 tfprintf(out, T("Hash              = 0x"));
 651                 print_hash(lte->hash, out);
 652                 tputc(T('\n'), out);
 653         }
 654
 655         tfprintf(out, T("Flags             = "));
 656         u8 flags = lte->resource_entry.flags;
 657         if (flags & WIM_RESHDR_FLAG_COMPRESSED)
 658                 tfputs(T("WIM_RESHDR_FLAG_COMPRESSED, "), out);
 659         if (flags & WIM_RESHDR_FLAG_FREE)
 660                 tfputs(T("WIM_RESHDR_FLAG_FREE, "), out);
 661         if (flags & WIM_RESHDR_FLAG_METADATA)
 662                 tfputs(T("WIM_RESHDR_FLAG_METADATA, "), out);
 663         if (flags & WIM_RESHDR_FLAG_SPANNED)
 664                 tfputs(T("WIM_RESHDR_FLAG_SPANNED, "), out);
 665         tputc(T('\n'), out);
 666         switch (lte->resource_location) {
 667         case RESOURCE_IN_WIM:
 668                 if (lte->wim->filename) {
 669                         tfprintf(out, T("WIM file          = `%"TS"'\n"),
 670                                  lte->wim->filename);
 671                 }
 672                 break;
 673 #ifdef __WIN32__
 674         case RESOURCE_WIN32:
 675         case RESOURCE_WIN32_ENCRYPTED:
 676 #else
 677         case RESOURCE_IN_FILE_ON_DISK:
 678 #endif
 679                 tfprintf(out, T("File on Disk      = `%"TS"'\n"),
 680                          lte->file_on_disk);
 681                 break;
 682 #ifdef WITH_FUSE
 683         case RESOURCE_IN_STAGING_FILE:
 684                 tfprintf(out, T("Staging File      = `%"TS"'\n"),
 685                                 lte->staging_file_name);
 686                 break;
 687 #endif
 688         default:
 689                 break;
 690         }
 691         tputc(T('\n'), out);
 692 }
 693
 694 static int
 695 do_print_lookup_table_entry(struct wim_lookup_table_entry *lte, void *fp)
 696 {
 697         print_lookup_table_entry(lte, (FILE*)fp);
 698         return 0;
 699 }
 700
 701 /*
 702  * Prints the lookup table of a WIM file.
 703  */
 704 WIMLIBAPI void
 705 wimlib_print_lookup_table(WIMStruct *w)
 706 {
 707         for_lookup_table_entry(w->lookup_table,
 708                                do_print_lookup_table_entry,
 709                                stdout);
 710 }
 711
 712 /* Given a SHA1 message digest, return the corresponding entry in the WIM's
 713  * lookup table, or NULL if there is none.  */
 714 struct wim_lookup_table_entry *
 715 __lookup_resource(const struct wim_lookup_table *table, const u8 hash[])
 716 {
 717         size_t i;
 718         struct wim_lookup_table_entry *lte;
 719         struct hlist_node *pos;
 720
 721         wimlib_assert(table != NULL);
 722         wimlib_assert(hash != NULL);
 723
 724         i = *(size_t*)hash % table->capacity;
 725         hlist_for_each_entry(lte, pos, &table->array[i], hash_list)
 726                 if (hashes_equal(hash, lte->hash))
 727                         return lte;
 728         return NULL;
 729 }
 730
 731 #ifdef WITH_FUSE
 732 /*
 733  * Finds the dentry, lookup table entry, and stream index for a WIM file stream,
 734  * given a path name.
 735  *
 736  * This is only for pre-resolved inodes.
 737  */
 738 int
 739 lookup_resource(WIMStruct *w,
 740                 const tchar *path,
 741                 int lookup_flags,
 742                 struct wim_dentry **dentry_ret,
 743                 struct wim_lookup_table_entry **lte_ret,
 744                 u16 *stream_idx_ret)
 745 {
 746         struct wim_dentry *dentry;
 747         struct wim_lookup_table_entry *lte;
 748         u16 stream_idx;
 749         const tchar *stream_name = NULL;
 750         struct wim_inode *inode;
 751         tchar *p = NULL;
 752
 753         if (lookup_flags & LOOKUP_FLAG_ADS_OK) {
 754                 stream_name = path_stream_name(path);
 755                 if (stream_name) {
 756                         p = (tchar*)stream_name - 1;
 757                         *p = T('\0');
 758                 }
 759         }
 760
 761         dentry = get_dentry(w, path);
 762         if (p)
 763                 *p = T(':');
 764         if (!dentry)
 765                 return -errno;
 766
 767         inode = dentry->d_inode;
 768
 769         wimlib_assert(inode->i_resolved);
 770
 771         if (!(lookup_flags & LOOKUP_FLAG_DIRECTORY_OK)
 772               && inode_is_directory(inode))
 773                 return -EISDIR;
 774
 775         if (stream_name) {
 776                 struct wim_ads_entry *ads_entry;
 777                 u16 ads_idx;
 778                 ads_entry = inode_get_ads_entry(inode, stream_name,
 779                                                 &ads_idx);
 780                 if (ads_entry) {
 781                         stream_idx = ads_idx + 1;
 782                         lte = ads_entry->lte;
 783                         goto out;
 784                 } else {
 785                         return -ENOENT;
 786                 }
 787         } else {
 788                 lte = inode->i_lte;
 789                 stream_idx = 0;
 790         }
 791 out:
 792         if (dentry_ret)
 793                 *dentry_ret = dentry;
 794         if (lte_ret)
 795                 *lte_ret = lte;
 796         if (stream_idx_ret)
 797                 *stream_idx_ret = stream_idx;
 798         return 0;
 799 }
 800 #endif
 801
 802 /* Resolve an inode's lookup table entries
 803  *
 804  * This replaces the SHA1 hash fields (which are used to lookup an entry in the
 805  * lookup table) with pointers directly to the lookup table entries.  A circular
 806  * linked list of streams sharing the same lookup table entry is created.
 807  *
 808  * This function always succeeds; unresolved lookup table entries are given a
 809  * NULL pointer.
 810  */
 811 void
 812 inode_resolve_ltes(struct wim_inode *inode, struct wim_lookup_table *table)
 813 {
 814
 815         if (!inode->i_resolved) {
 816                 struct wim_lookup_table_entry *lte;
 817                 /* Resolve the default file stream */
 818                 lte = __lookup_resource(table, inode->i_hash);
 819                 inode->i_lte = lte;
 820                 inode->i_resolved = 1;
 821
 822                 /* Resolve the alternate data streams */
 823                 for (u16 i = 0; i < inode->i_num_ads; i++) {
 824                         struct wim_ads_entry *cur_entry = &inode->i_ads_entries[i];
 825                         lte = __lookup_resource(table, cur_entry->hash);
 826                         cur_entry->lte = lte;
 827                 }
 828         }
 829 }
 830
 831 void
 832 inode_unresolve_ltes(struct wim_inode *inode)
 833 {
 834         if (inode->i_resolved) {
 835                 if (inode->i_lte)
 836                         copy_hash(inode->i_hash, inode->i_lte->hash);
 837                 else
 838                         zero_out_hash(inode->i_hash);
 839
 840                 for (u16 i = 0; i < inode->i_num_ads; i++) {
 841                         if (inode->i_ads_entries[i].lte)
 842                                 copy_hash(inode->i_ads_entries[i].hash,
 843                                           inode->i_ads_entries[i].lte->hash);
 844                         else
 845                                 zero_out_hash(inode->i_ads_entries[i].hash);
 846                 }
 847                 inode->i_resolved = 0;
 848         }
 849 }
 850
 851 /*
 852  * Returns the lookup table entry for stream @stream_idx of the inode, where
 853  * stream_idx = 0 means the default un-named file stream, and stream_idx >= 1
 854  * corresponds to an alternate data stream.
 855  *
 856  * This works for both resolved and un-resolved inodes.
 857  */
 858 struct wim_lookup_table_entry *
 859 inode_stream_lte(const struct wim_inode *inode, unsigned stream_idx,
 860                  const struct wim_lookup_table *table)
 861 {
 862         if (inode->i_resolved)
 863                 return inode_stream_lte_resolved(inode, stream_idx);
 864         else
 865                 return inode_stream_lte_unresolved(inode, stream_idx, table);
 866 }
 867
 868 struct wim_lookup_table_entry *
 869 inode_unnamed_lte_resolved(const struct wim_inode *inode)
 870 {
 871         wimlib_assert(inode->i_resolved);
 872         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
 873                 if (inode_stream_name_nbytes(inode, i) == 0 &&
 874                     !is_zero_hash(inode_stream_hash_resolved(inode, i)))
 875                 {
 876                         return inode_stream_lte_resolved(inode, i);
 877                 }
 878         }
 879         return NULL;
 880 }
 881
 882 struct wim_lookup_table_entry *
 883 inode_unnamed_lte_unresolved(const struct wim_inode *inode,
 884                              const struct wim_lookup_table *table)
 885 {
 886         wimlib_assert(!inode->i_resolved);
 887         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
 888                 if (inode_stream_name_nbytes(inode, i) == 0 &&
 889                     !is_zero_hash(inode_stream_hash_unresolved(inode, i)))
 890                 {
 891                         return inode_stream_lte_unresolved(inode, i, table);
 892                 }
 893         }
 894         return NULL;
 895 }
 896
 897 /* Return the lookup table entry for the unnamed data stream of an inode, or
 898  * NULL if there is none.
 899  *
 900  * You'd think this would be easier than it actually is, since the unnamed data
 901  * stream should be the one referenced from the inode itself.  Alas, if there
 902  * are named data streams, Microsoft's "imagex.exe" program will put the unnamed
 903  * data stream in one of the alternate data streams instead of inside the WIM
 904  * dentry itself.  So we need to check the alternate data streams too.
 905  *
 906  * Also, note that a dentry may appear to have more than one unnamed stream, but
 907  * if the SHA1 message digest is all 0's then the corresponding stream does not
 908  * really "count" (this is the case for the inode's own file stream when the
 909  * file stream that should be there is actually in one of the alternate stream
 910  * entries.).  This is despite the fact that we may need to extract such a
 911  * missing entry as an empty file or empty named data stream.
 912  */
 913 struct wim_lookup_table_entry *
 914 inode_unnamed_lte(const struct wim_inode *inode,
 915                   const struct wim_lookup_table *table)
 916 {
 917         if (inode->i_resolved)
 918                 return inode_unnamed_lte_resolved(inode);
 919         else
 920                 return inode_unnamed_lte_unresolved(inode, table);
 921 }
 922
 923 static int
 924 lte_add_stream_size(struct wim_lookup_table_entry *lte, void *total_bytes_p)
 925 {
 926         *(u64*)total_bytes_p += lte->resource_entry.size;
 927         return 0;
 928 }
 929
 930 u64
 931 lookup_table_total_stream_size(struct wim_lookup_table *table)
 932 {
 933         u64 total_size = 0;
 934         for_lookup_table_entry(table, lte_add_stream_size, &total_size);
 935         return total_size;
 936 }
 937
 938 struct wim_lookup_table_entry **
 939 retrieve_lte_pointer(struct wim_lookup_table_entry *lte)
 940 {
 941         wimlib_assert(lte->unhashed);
 942         struct wim_inode *inode = lte->back_inode;
 943         u32 stream_id = lte->back_stream_id;
 944         if (stream_id == 0)
 945                 return &inode->i_lte;
 946         else
 947                 for (u16 i = 0; i < inode->i_num_ads; i++)
 948                         if (inode->i_ads_entries[i].stream_id == stream_id)
 949                                 return &inode->i_ads_entries[i].lte;
 950         wimlib_assert(0);
 951         return NULL;
 952 }
 953
 954 /* Calculate the SHA1 message digest of a stream and move it from the list of
 955  * unhashed streams to the stream lookup table, possibly joining it with an
 956  * existing lookup table entry for an identical stream.
 957  *
 958  * @lte:  An unhashed lookup table entry.
 959  * @lookup_table:  Lookup table for the WIM.
 960  * @lte_ret:  On success, write a pointer to the resulting lookup table
 961  *            entry to this location.  This will be the same as @lte
 962  *            if it was inserted into the lookup table, or different if
 963  *            a duplicate stream was found.
 964  *
 965  * Returns 0 on success; nonzero if there is an error reading the stream.
 966  */
 967 int
 968 hash_unhashed_stream(struct wim_lookup_table_entry *lte,
 969                      struct wim_lookup_table *lookup_table,
 970                      struct wim_lookup_table_entry **lte_ret)
 971 {
 972         int ret;
 973         struct wim_lookup_table_entry *duplicate_lte;
 974         struct wim_lookup_table_entry **back_ptr;
 975
 976         wimlib_assert(lte->unhashed);
 977
 978         /* back_ptr must be saved because @back_inode and @back_stream_id are in
 979          * union with the SHA1 message digest and will no longer be valid once
 980          * the SHA1 has been calculated. */
 981         back_ptr = retrieve_lte_pointer(lte);
 982
 983         ret = sha1_resource(lte);
 984         if (ret)
 985                 return ret;
 986
 987         /* Look for a duplicate stream */
 988         duplicate_lte = __lookup_resource(lookup_table, lte->hash);
 989         list_del(&lte->unhashed_list);
 990         if (duplicate_lte) {
 991                 /* We have a duplicate stream.  Transfer the reference counts
 992                  * from this stream to the duplicate, update the reference to
 993                  * this stream (in an inode or ads_entry) to point to the
 994                  * duplicate, then free this stream. */
 995                 wimlib_assert(!(duplicate_lte->unhashed));
 996                 duplicate_lte->refcnt += lte->refcnt;
 997                 duplicate_lte->out_refcnt += lte->refcnt;
 998                 *back_ptr = duplicate_lte;
 999                 free_lookup_table_entry(lte);
1000                 lte = duplicate_lte;
1001         } else {
1002                 /* No duplicate stream, so we need to insert
1003                  * this stream into the lookup table and treat
1004                  * it as a hashed stream. */
1005                 lookup_table_insert(lookup_table, lte);
1006                 lte->unhashed = 0;
1007         }
1008         if (lte_ret)
1009                 *lte_ret = lte;
1010         return 0;
1011 }
1012