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; plus code to read and write the corresponding on-disk data.
   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 #ifdef HAVE_CONFIG_H
  28 #  include "config.h"
  29 #endif
  30
  31 #include "wimlib/endianness.h"
  32 #include "wimlib/error.h"
  33 #include "wimlib/file_io.h"
  34 #include "wimlib/glob.h"
  35 #include "wimlib/lookup_table.h"
  36 #include "wimlib/metadata.h"
  37 #include "wimlib/paths.h"
  38 #include "wimlib/resource.h"
  39 #include "wimlib/util.h"
  40 #include "wimlib/write.h"
  41
  42 #include <errno.h>
  43 #include <stdlib.h>
  44 #ifdef WITH_FUSE
  45 #  include <unistd.h> /* for unlink() */
  46 #endif
  47
  48 struct wim_lookup_table *
  49 new_lookup_table(size_t capacity)
  50 {
  51         struct wim_lookup_table *table;
  52         struct hlist_head *array;
  53
  54         table = CALLOC(1, sizeof(struct wim_lookup_table));
  55         if (table) {
  56                 array = CALLOC(capacity, sizeof(array[0]));
  57                 if (array) {
  58                         table->num_entries = 0;
  59                         table->capacity = capacity;
  60                         table->array = array;
  61                 } else {
  62                         FREE(table);
  63                         table = NULL;
  64                         ERROR("Failed to allocate memory for lookup table "
  65                               "with capacity %zu", capacity);
  66                 }
  67         }
  68         return table;
  69 }
  70
  71 struct wim_lookup_table_entry *
  72 new_lookup_table_entry(void)
  73 {
  74         struct wim_lookup_table_entry *lte;
  75
  76         lte = CALLOC(1, sizeof(struct wim_lookup_table_entry));
  77         if (lte == NULL) {
  78                 ERROR("Out of memory (tried to allocate %zu bytes for "
  79                       "lookup table entry)",
  80                       sizeof(struct wim_lookup_table_entry));
  81                 return NULL;
  82         }
  83         lte->refcnt = 1;
  84         BUILD_BUG_ON(RESOURCE_NONEXISTENT != 0);
  85         return lte;
  86 }
  87
  88 struct wim_lookup_table_entry *
  89 clone_lookup_table_entry(const struct wim_lookup_table_entry *old)
  90 {
  91         struct wim_lookup_table_entry *new;
  92
  93         new = memdup(old, sizeof(struct wim_lookup_table_entry));
  94         if (new == NULL)
  95                 return NULL;
  96
  97         new->extracted_file = NULL;
  98         switch (new->resource_location) {
  99         case RESOURCE_IN_WIM:
 100                 list_add(&new->rspec_node, &new->rspec->stream_list);
 101                 break;
 102
 103         case RESOURCE_IN_FILE_ON_DISK:
 104 #ifdef __WIN32__
 105         case RESOURCE_WIN32_ENCRYPTED:
 106 #endif
 107 #ifdef WITH_FUSE
 108         case RESOURCE_IN_STAGING_FILE:
 109                 BUILD_BUG_ON((void*)&old->file_on_disk !=
 110                              (void*)&old->staging_file_name);
 111 #endif
 112                 new->file_on_disk = TSTRDUP(old->file_on_disk);
 113                 if (new->file_on_disk == NULL)
 114                         goto out_free;
 115                 break;
 116         case RESOURCE_IN_ATTACHED_BUFFER:
 117                 new->attached_buffer = memdup(old->attached_buffer, old->size);
 118                 if (new->attached_buffer == NULL)
 119                         goto out_free;
 120                 break;
 121 #ifdef WITH_NTFS_3G
 122         case RESOURCE_IN_NTFS_VOLUME:
 123                 if (old->ntfs_loc) {
 124                         struct ntfs_location *loc;
 125                         loc = memdup(old->ntfs_loc, sizeof(struct ntfs_location));
 126                         if (loc == NULL)
 127                                 goto out_free;
 128                         loc->path = NULL;
 129                         loc->stream_name = NULL;
 130                         new->ntfs_loc = loc;
 131                         loc->path = STRDUP(old->ntfs_loc->path);
 132                         if (loc->path == NULL)
 133                                 goto out_free;
 134                         if (loc->stream_name_nchars != 0) {
 135                                 loc->stream_name = memdup(old->ntfs_loc->stream_name,
 136                                                           loc->stream_name_nchars * 2);
 137                                 if (loc->stream_name == NULL)
 138                                         goto out_free;
 139                         }
 140                 }
 141                 break;
 142 #endif
 143         default:
 144                 break;
 145         }
 146         return new;
 147 out_free:
 148         free_lookup_table_entry(new);
 149         return NULL;
 150 }
 151
 152 void
 153 free_lookup_table_entry(struct wim_lookup_table_entry *lte)
 154 {
 155         if (lte) {
 156                 switch (lte->resource_location) {
 157                 case RESOURCE_IN_WIM:
 158                         list_del(&lte->rspec_node);
 159                         if (list_empty(&lte->rspec->stream_list))
 160                                 FREE(lte->rspec);
 161                         break;
 162                 case RESOURCE_IN_FILE_ON_DISK:
 163         #ifdef __WIN32__
 164                 case RESOURCE_WIN32_ENCRYPTED:
 165         #endif
 166         #ifdef WITH_FUSE
 167                 case RESOURCE_IN_STAGING_FILE:
 168                         BUILD_BUG_ON((void*)&lte->file_on_disk !=
 169                                      (void*)&lte->staging_file_name);
 170         #endif
 171                 case RESOURCE_IN_ATTACHED_BUFFER:
 172                         BUILD_BUG_ON((void*)&lte->file_on_disk !=
 173                                      (void*)&lte->attached_buffer);
 174                         FREE(lte->file_on_disk);
 175                         break;
 176 #ifdef WITH_NTFS_3G
 177                 case RESOURCE_IN_NTFS_VOLUME:
 178                         if (lte->ntfs_loc) {
 179                                 FREE(lte->ntfs_loc->path);
 180                                 FREE(lte->ntfs_loc->stream_name);
 181                                 FREE(lte->ntfs_loc);
 182                         }
 183                         break;
 184 #endif
 185                 default:
 186                         break;
 187                 }
 188                 FREE(lte);
 189         }
 190 }
 191
 192 static int
 193 do_free_lookup_table_entry(struct wim_lookup_table_entry *entry, void *ignore)
 194 {
 195         free_lookup_table_entry(entry);
 196         return 0;
 197 }
 198
 199
 200 void
 201 free_lookup_table(struct wim_lookup_table *table)
 202 {
 203         DEBUG("Freeing lookup table.");
 204         if (table) {
 205                 if (table->array) {
 206                         for_lookup_table_entry(table,
 207                                                do_free_lookup_table_entry,
 208                                                NULL);
 209                         FREE(table->array);
 210                 }
 211                 FREE(table);
 212         }
 213 }
 214
 215 static void
 216 lookup_table_insert_raw(struct wim_lookup_table *table,
 217                         struct wim_lookup_table_entry *lte)
 218 {
 219         size_t i = lte->hash_short % table->capacity;
 220
 221         hlist_add_head(&lte->hash_list, &table->array[i]);
 222 }
 223
 224 static void
 225 enlarge_lookup_table(struct wim_lookup_table *table)
 226 {
 227         size_t old_capacity, new_capacity;
 228         struct hlist_head *old_array, *new_array;
 229         struct wim_lookup_table_entry *lte;
 230         struct hlist_node *cur, *tmp;
 231         size_t i;
 232
 233         old_capacity = table->capacity;
 234         new_capacity = old_capacity * 2;
 235         new_array = CALLOC(new_capacity, sizeof(struct hlist_head));
 236         if (new_array == NULL)
 237                 return;
 238         old_array = table->array;
 239         table->array = new_array;
 240         table->capacity = new_capacity;
 241
 242         for (i = 0; i < old_capacity; i++) {
 243                 hlist_for_each_entry_safe(lte, cur, tmp, &old_array[i], hash_list) {
 244                         hlist_del(&lte->hash_list);
 245                         lookup_table_insert_raw(table, lte);
 246                 }
 247         }
 248         FREE(old_array);
 249 }
 250
 251
 252 /*
 253  * Inserts an entry into the lookup table.
 254  *
 255  * @table:      A pointer to the lookup table.
 256  * @lte:        A pointer to the entry to insert.
 257  */
 258 void
 259 lookup_table_insert(struct wim_lookup_table *table,
 260                     struct wim_lookup_table_entry *lte)
 261 {
 262         lookup_table_insert_raw(table, lte);
 263         if (++table->num_entries > table->capacity)
 264                 enlarge_lookup_table(table);
 265 }
 266
 267 static void
 268 finalize_lte(struct wim_lookup_table_entry *lte)
 269 {
 270         #ifdef WITH_FUSE
 271         if (lte->resource_location == RESOURCE_IN_STAGING_FILE) {
 272                 unlink(lte->staging_file_name);
 273                 list_del(&lte->unhashed_list);
 274         }
 275         #endif
 276         free_lookup_table_entry(lte);
 277 }
 278
 279 /* Decrements the reference count for the lookup table entry @lte.  If its
 280  * reference count reaches 0, it is unlinked from the lookup table.  If,
 281  * furthermore, the entry has no opened file descriptors associated with it, the
 282  * entry is freed.  */
 283 void
 284 lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
 285                      struct wim_lookup_table *table)
 286 {
 287         wimlib_assert(lte != NULL);
 288         wimlib_assert(lte->refcnt != 0);
 289         if (--lte->refcnt == 0) {
 290                 if (lte->unhashed)
 291                         list_del(&lte->unhashed_list);
 292                 else
 293                         lookup_table_unlink(table, lte);
 294         #ifdef WITH_FUSE
 295                 if (lte->num_opened_fds == 0)
 296         #endif
 297                         finalize_lte(lte);
 298         }
 299 }
 300
 301 #ifdef WITH_FUSE
 302 void
 303 lte_decrement_num_opened_fds(struct wim_lookup_table_entry *lte)
 304 {
 305         if (lte->num_opened_fds != 0)
 306                 if (--lte->num_opened_fds == 0 && lte->refcnt == 0)
 307                         finalize_lte(lte);
 308 }
 309 #endif
 310
 311 /* Calls a function on all the entries in the WIM lookup table.  Stop early and
 312  * return nonzero if any call to the function returns nonzero. */
 313 int
 314 for_lookup_table_entry(struct wim_lookup_table *table,
 315                        int (*visitor)(struct wim_lookup_table_entry *, void *),
 316                        void *arg)
 317 {
 318         struct wim_lookup_table_entry *lte;
 319         struct hlist_node *pos, *tmp;
 320         int ret;
 321
 322         for (size_t i = 0; i < table->capacity; i++) {
 323                 hlist_for_each_entry_safe(lte, pos, tmp, &table->array[i],
 324                                           hash_list)
 325                 {
 326                         ret = visitor(lte, arg);
 327                         if (ret)
 328                                 return ret;
 329                 }
 330         }
 331         return 0;
 332 }
 333
 334 /* qsort() callback that sorts streams (represented by `struct
 335  * wim_lookup_table_entry's) into an order optimized for reading.
 336  *
 337  * Sorting is done primarily by resource location, then secondarily by a
 338  * per-resource location order.  For example, resources in WIM files are sorted
 339  * primarily by part number, then secondarily by offset, as to implement optimal
 340  * reading of either a standalone or split WIM.  */
 341 static int
 342 cmp_streams_by_sequential_order(const void *p1, const void *p2)
 343 {
 344         const struct wim_lookup_table_entry *lte1, *lte2;
 345         int v;
 346         WIMStruct *wim1, *wim2;
 347
 348         lte1 = *(const struct wim_lookup_table_entry**)p1;
 349         lte2 = *(const struct wim_lookup_table_entry**)p2;
 350
 351         v = (int)lte1->resource_location - (int)lte2->resource_location;
 352
 353         /* Different resource locations?  */
 354         if (v)
 355                 return v;
 356
 357         switch (lte1->resource_location) {
 358         case RESOURCE_IN_WIM:
 359                 wim1 = lte1->rspec->wim;
 360                 wim2 = lte2->rspec->wim;
 361
 362                 /* Different (possibly split) WIMs?  */
 363                 if (wim1 != wim2) {
 364                         v = memcmp(wim1->hdr.guid, wim2->hdr.guid, WIM_GID_LEN);
 365                         if (v)
 366                                 return v;
 367                 }
 368
 369                 /* Different part numbers in the same WIM?  */
 370                 v = (int)wim1->hdr.part_number - (int)wim2->hdr.part_number;
 371                 if (v)
 372                         return v;
 373
 374                 if (lte1->rspec->offset_in_wim != lte2->rspec->offset_in_wim)
 375                         return cmp_u64(lte1->rspec->offset_in_wim,
 376                                        lte2->rspec->offset_in_wim);
 377
 378                 return cmp_u64(lte1->offset_in_res, lte2->offset_in_res);
 379
 380         case RESOURCE_IN_FILE_ON_DISK:
 381 #ifdef WITH_FUSE
 382         case RESOURCE_IN_STAGING_FILE:
 383 #endif
 384 #ifdef __WIN32__
 385         case RESOURCE_WIN32_ENCRYPTED:
 386 #endif
 387                 /* Compare files by path: just a heuristic that will place files
 388                  * in the same directory next to each other.  */
 389                 return tstrcmp(lte1->file_on_disk, lte2->file_on_disk);
 390 #ifdef WITH_NTFS_3G
 391         case RESOURCE_IN_NTFS_VOLUME:
 392                 return tstrcmp(lte1->ntfs_loc->path, lte2->ntfs_loc->path);
 393 #endif
 394         default:
 395                 /* No additional sorting order defined for this resource
 396                  * location (e.g. RESOURCE_IN_ATTACHED_BUFFER); simply compare
 397                  * everything equal to each other.  */
 398                 return 0;
 399         }
 400 }
 401
 402 int
 403 sort_stream_list(struct list_head *stream_list,
 404                  size_t list_head_offset,
 405                  int (*compar)(const void *, const void*))
 406 {
 407         struct list_head *cur;
 408         struct wim_lookup_table_entry **array;
 409         size_t i;
 410         size_t array_size;
 411         size_t num_streams = 0;
 412
 413         list_for_each(cur, stream_list)
 414                 num_streams++;
 415
 416         if (num_streams <= 1)
 417                 return 0;
 418
 419         array_size = num_streams * sizeof(array[0]);
 420         array = MALLOC(array_size);
 421         if (array == NULL)
 422                 return WIMLIB_ERR_NOMEM;
 423
 424         cur = stream_list->next;
 425         for (i = 0; i < num_streams; i++) {
 426                 array[i] = (struct wim_lookup_table_entry*)((u8*)cur -
 427                                                             list_head_offset);
 428                 cur = cur->next;
 429         }
 430
 431         qsort(array, num_streams, sizeof(array[0]), compar);
 432
 433         INIT_LIST_HEAD(stream_list);
 434         for (i = 0; i < num_streams; i++) {
 435                 list_add_tail((struct list_head*)
 436                                ((u8*)array[i] + list_head_offset),
 437                               stream_list);
 438         }
 439         FREE(array);
 440         return 0;
 441 }
 442
 443 /* Sort the specified list of streams in an order optimized for reading.  */
 444 int
 445 sort_stream_list_by_sequential_order(struct list_head *stream_list,
 446                                      size_t list_head_offset)
 447 {
 448         return sort_stream_list(stream_list, list_head_offset,
 449                                 cmp_streams_by_sequential_order);
 450 }
 451
 452
 453 static int
 454 add_lte_to_array(struct wim_lookup_table_entry *lte,
 455                  void *_pp)
 456 {
 457         struct wim_lookup_table_entry ***pp = _pp;
 458         *(*pp)++ = lte;
 459         return 0;
 460 }
 461
 462 /* Iterate through the lookup table entries, but first sort them by stream
 463  * offset in the WIM.  Caution: this is intended to be used when the stream
 464  * offset field has actually been set. */
 465 int
 466 for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
 467                                   int (*visitor)(struct wim_lookup_table_entry *,
 468                                                  void *),
 469                                   void *arg)
 470 {
 471         struct wim_lookup_table_entry **lte_array, **p;
 472         size_t num_streams = table->num_entries;
 473         int ret;
 474
 475         lte_array = MALLOC(num_streams * sizeof(lte_array[0]));
 476         if (!lte_array)
 477                 return WIMLIB_ERR_NOMEM;
 478         p = lte_array;
 479         for_lookup_table_entry(table, add_lte_to_array, &p);
 480
 481         wimlib_assert(p == lte_array + num_streams);
 482
 483         qsort(lte_array, num_streams, sizeof(lte_array[0]),
 484               cmp_streams_by_sequential_order);
 485         ret = 0;
 486         for (size_t i = 0; i < num_streams; i++) {
 487                 ret = visitor(lte_array[i], arg);
 488                 if (ret)
 489                         break;
 490         }
 491         FREE(lte_array);
 492         return ret;
 493 }
 494
 495 /* On-disk format of a WIM lookup table entry (stream entry). */
 496 struct wim_lookup_table_entry_disk {
 497         /* Size, offset, and flags of the stream.  */
 498         struct wim_reshdr_disk reshdr;
 499
 500         /* Which part of the split WIM this stream is in; indexed from 1. */
 501         le16 part_number;
 502
 503         /* Reference count of this stream over all WIM images. */
 504         le32 refcnt;
 505
 506         /* SHA1 message digest of the uncompressed data of this stream, or
 507          * optionally all zeroes if this stream is of zero length. */
 508         u8 hash[SHA1_HASH_SIZE];
 509 } _packed_attribute;
 510
 511 #define WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE 50
 512
 513 /* Validate the size and location of a WIM resource.  */
 514 static int
 515 validate_resource(const struct wim_resource_spec *rspec)
 516 {
 517         struct wim_lookup_table_entry *lte;
 518         u64 cur_offset;
 519
 520         /* Verify that calculating the offset of the end of the resource doesn't
 521          * overflow.  */
 522         if (rspec->offset_in_wim + rspec->size_in_wim < rspec->size_in_wim)
 523                 goto invalid;
 524
 525         /* Verify that each stream in the resource has a valid offset and size,
 526          * and that no streams overlap, and that the streams were added in order
 527          * of increasing offset.  */
 528         cur_offset = 0;
 529         list_for_each_entry(lte, &rspec->stream_list, rspec_node) {
 530                 if (lte->offset_in_res + lte->size < lte->size ||
 531                     lte->offset_in_res + lte->size > rspec->uncompressed_size ||
 532                     lte->offset_in_res < cur_offset)
 533                         goto invalid;
 534
 535                 cur_offset = lte->offset_in_res + lte->size;
 536         }
 537         return 0;
 538
 539 invalid:
 540
 541         ERROR("Invalid resource entry!");
 542         return WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 543 }
 544
 545 /*
 546  * Reads the lookup table from a WIM file.  Each entry specifies a stream that
 547  * the WIM file contains, along with its location and SHA1 message digest.
 548  *
 549  * Saves lookup table entries for non-metadata streams in a hash table, and
 550  * saves the metadata entry for each image in a special per-image location (the
 551  * image_metadata array).
 552  *
 553  * Return values:
 554  *      WIMLIB_ERR_SUCCESS (0)
 555  *      WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY
 556  *      WIMLIB_ERR_RESOURCE_NOT_FOUND
 557  *
 558  *      Or an error code caused by failure to read the lookup table into memory.
 559  */
 560 int
 561 read_wim_lookup_table(WIMStruct *wim)
 562 {
 563         int ret;
 564         size_t i;
 565         size_t num_entries;
 566         struct wim_lookup_table *table;
 567         struct wim_lookup_table_entry *cur_entry, *duplicate_entry;
 568         struct wim_resource_spec *cur_rspec;
 569         void *buf;
 570         bool back_to_back_pack;
 571
 572         DEBUG("Reading lookup table.");
 573
 574         /* Sanity check: lookup table entries are 50 bytes each.  */
 575         BUILD_BUG_ON(sizeof(struct wim_lookup_table_entry_disk) !=
 576                      WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE);
 577
 578         /* Calculate number of entries in the lookup table.  */
 579         num_entries = wim->hdr.lookup_table_reshdr.uncompressed_size /
 580                       sizeof(struct wim_lookup_table_entry_disk);
 581
 582         /* Read the lookup table into a buffer.  */
 583         ret = wim_reshdr_to_data(&wim->hdr.lookup_table_reshdr, wim, &buf);
 584         if (ret)
 585                 goto out;
 586
 587         /* Allocate a hash table to map SHA1 message digests into stream
 588          * specifications.  This is the in-memory "lookup table".  */
 589         table = new_lookup_table(num_entries * 2 + 1);
 590         if (table == NULL) {
 591                 ERROR("Not enough memory to read lookup table.");
 592                 ret = WIMLIB_ERR_NOMEM;
 593                 goto out_free_buf;
 594         }
 595
 596         /* Allocate and initalize stream entries from the raw lookup table
 597          * buffer.  */
 598         wim->current_image = 0;
 599         cur_rspec = NULL;
 600         for (i = 0; i < num_entries; i++) {
 601                 const struct wim_lookup_table_entry_disk *disk_entry =
 602                         &((const struct wim_lookup_table_entry_disk*)buf)[i];
 603                 u16 part_number;
 604                 struct wim_reshdr reshdr;
 605
 606                 get_wim_reshdr(&disk_entry->reshdr, &reshdr);
 607
 608                 DEBUG("reshdr: size_in_wim=%"PRIu64", "
 609                       "uncompressed_size=%"PRIu64", "
 610                       "offset_in_wim=%"PRIu64", "
 611                       "flags=0x%02x",
 612                       reshdr.size_in_wim, reshdr.uncompressed_size,
 613                       reshdr.offset_in_wim, reshdr.flags);
 614
 615                 if (wim->hdr.wim_version == WIM_VERSION_DEFAULT)
 616                         reshdr.flags &= ~WIM_RESHDR_FLAG_PACKED_STREAMS;
 617
 618                 cur_entry = new_lookup_table_entry();
 619                 if (cur_entry == NULL) {
 620                         ERROR("Not enough memory to read lookup table!");
 621                         ret = WIMLIB_ERR_NOMEM;
 622                         goto out_free_lookup_table;
 623                 }
 624
 625                 part_number = le16_to_cpu(disk_entry->part_number);
 626                 cur_entry->refcnt = le32_to_cpu(disk_entry->refcnt);
 627                 copy_hash(cur_entry->hash, disk_entry->hash);
 628
 629                 if (part_number != wim->hdr.part_number) {
 630                         WARNING("A lookup table entry in part %hu of the WIM "
 631                                 "points to part %hu (ignoring it)",
 632                                 wim->hdr.part_number, part_number);
 633                         free_lookup_table_entry(cur_entry);
 634                         continue;
 635                 }
 636
 637                 if (!(reshdr.flags & (WIM_RESHDR_FLAG_PACKED_STREAMS |
 638                                       WIM_RESHDR_FLAG_COMPRESSED))) {
 639                         if (reshdr.uncompressed_size != reshdr.size_in_wim) {
 640                                 ERROR("Invalid resource entry!");
 641                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 642                                 goto out_free_cur_entry;
 643                         }
 644                 }
 645
 646                 back_to_back_pack = false;
 647                 if (!(reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) ||
 648                     cur_rspec == NULL ||
 649                     (back_to_back_pack =
 650                      ((reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) &&
 651                       reshdr.uncompressed_size == WIM_PACK_MAGIC_NUMBER &&
 652                       cur_rspec != NULL &&
 653                       cur_rspec->size_in_wim != 0)))
 654                 {
 655                         /* Starting new run of streams that share the same WIM
 656                          * resource.  */
 657                         struct wim_lookup_table_entry *prev_entry = NULL;
 658
 659                         if (back_to_back_pack) {
 660                                 prev_entry = list_entry(cur_rspec->stream_list.prev,
 661                                                         struct wim_lookup_table_entry,
 662                                                         rspec_node);
 663                                 lte_unbind_wim_resource_spec(prev_entry);
 664                                 cur_rspec->uncompressed_size -= prev_entry->size;
 665                         }
 666                         if (cur_rspec != NULL) {
 667                                 ret = validate_resource(cur_rspec);
 668                                 if (ret)
 669                                         goto out_free_cur_entry;
 670                         }
 671
 672                         /* Allocate the resource specification and initialize it
 673                          * with values from the current stream entry.  */
 674                         cur_rspec = MALLOC(sizeof(*cur_rspec));
 675                         if (cur_rspec == NULL) {
 676                                 ERROR("Not enough memory to read lookup table!");
 677                                 ret = WIMLIB_ERR_NOMEM;
 678                                 goto out_free_cur_entry;
 679                         }
 680                         wim_res_hdr_to_spec(&reshdr, wim, cur_rspec);
 681
 682                         /* If this is a packed run, the current stream entry may
 683                          * specify a stream within the resource, and not the
 684                          * resource itself.  Zero possibly irrelevant data until
 685                          * it is read for certain.  */
 686                         if (reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
 687                                 cur_rspec->size_in_wim = 0;
 688                                 cur_rspec->uncompressed_size = 0;
 689                                 cur_rspec->offset_in_wim = 0;
 690                         }
 691
 692                         if (prev_entry) {
 693                                 lte_bind_wim_resource_spec(prev_entry, cur_rspec);
 694                                 cur_rspec->uncompressed_size = prev_entry->size;
 695                         }
 696                 }
 697
 698                 if ((reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) &&
 699                     reshdr.uncompressed_size == WIM_PACK_MAGIC_NUMBER)
 700                 {
 701                         /* Found the specification for the packed resource.
 702                          * Transfer the values to the `struct
 703                          * wim_resource_spec', and discard the current stream
 704                          * since this lookup table entry did not, in fact,
 705                          * correspond to a "stream".  */
 706
 707                         cur_rspec->offset_in_wim = reshdr.offset_in_wim;
 708                         cur_rspec->size_in_wim = reshdr.size_in_wim;
 709                         cur_rspec->flags = reshdr.flags;
 710                         DEBUG("Full pack is %"PRIu64" compressed bytes "
 711                               "at file offset %"PRIu64" (flags 0x%02x)",
 712                               cur_rspec->size_in_wim,
 713                               cur_rspec->offset_in_wim,
 714                               cur_rspec->flags);
 715                         free_lookup_table_entry(cur_entry);
 716                         continue;
 717                 }
 718
 719                 if (is_zero_hash(cur_entry->hash)) {
 720                         free_lookup_table_entry(cur_entry);
 721                         continue;
 722                 }
 723
 724                 if (reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
 725                         /* Continuing the pack with another stream.  */
 726                         DEBUG("Continuing packed run with stream: "
 727                               "%"PRIu64" uncompressed bytes @ resource offset %"PRIu64")",
 728                               reshdr.size_in_wim, reshdr.offset_in_wim);
 729                         cur_rspec->uncompressed_size += reshdr.size_in_wim;
 730                 }
 731
 732                 lte_bind_wim_resource_spec(cur_entry, cur_rspec);
 733                 if (reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
 734                         /* In packed runs, the offset field is used for
 735                          * in-resource offset, not the in-WIM offset, and the
 736                          * size field is used for the uncompressed size, not the
 737                          * compressed size.  */
 738                         cur_entry->offset_in_res = reshdr.offset_in_wim;
 739                         cur_entry->size = reshdr.size_in_wim;
 740                         cur_entry->flags = reshdr.flags;
 741                 } else {
 742                         /* Normal case: The stream corresponds one-to-one with
 743                          * the resource entry.  */
 744                         cur_entry->offset_in_res = 0;
 745                         cur_entry->size = reshdr.uncompressed_size;
 746                         cur_entry->flags = reshdr.flags;
 747                         cur_rspec = NULL;
 748                 }
 749
 750                 if (cur_entry->flags & WIM_RESHDR_FLAG_METADATA) {
 751                         /* Lookup table entry for a metadata resource */
 752
 753                         /* Metadata entries with no references must be ignored;
 754                          * see for example the WinPE WIMs from the WAIK v2.1.
 755                          * */
 756                         if (cur_entry->refcnt == 0) {
 757                                 free_lookup_table_entry(cur_entry);
 758                                 continue;
 759                         }
 760
 761                         if (cur_entry->refcnt != 1) {
 762                                 if (wimlib_print_errors) {
 763                                         ERROR("Found metadata resource with refcnt != 1:");
 764                                         print_lookup_table_entry(cur_entry, stderr);
 765                                 }
 766                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 767                                 goto out_free_cur_entry;
 768                         }
 769
 770                         if (wim->hdr.part_number != 1) {
 771                                 WARNING("Ignoring metadata resource found in a "
 772                                         "non-first part of the split WIM");
 773                                 free_lookup_table_entry(cur_entry);
 774                                 continue;
 775                         }
 776                         if (wim->current_image == wim->hdr.image_count) {
 777                                 WARNING("The WIM header says there are %u images "
 778                                         "in the WIM, but we found more metadata "
 779                                         "resources than this (ignoring the extra)",
 780                                         wim->hdr.image_count);
 781                                 free_lookup_table_entry(cur_entry);
 782                                 continue;
 783                         }
 784
 785                         /* Notice very carefully:  We are assigning the metadata
 786                          * resources in the exact order mirrored by their lookup
 787                          * table entries on disk, which is the behavior of
 788                          * Microsoft's software.  In particular, this overrides
 789                          * the actual locations of the metadata resources
 790                          * themselves in the WIM file as well as any information
 791                          * written in the XML data. */
 792                         DEBUG("Found metadata resource for image %u at "
 793                               "offset %"PRIu64".",
 794                               wim->current_image + 1,
 795                               cur_entry->rspec->offset_in_wim);
 796                         wim->image_metadata[
 797                                 wim->current_image++]->metadata_lte = cur_entry;
 798                         continue;
 799                 }
 800
 801                 /* Lookup table entry for a stream that is not a metadata
 802                  * resource.  */
 803                 duplicate_entry = lookup_resource(table, cur_entry->hash);
 804                 if (duplicate_entry) {
 805                         if (wimlib_print_errors) {
 806                                 WARNING("The WIM lookup table contains two entries with the "
 807                                       "same SHA1 message digest!");
 808                                 WARNING("The first entry is:");
 809                                 print_lookup_table_entry(duplicate_entry, stderr);
 810                                 WARNING("The second entry is:");
 811                                 print_lookup_table_entry(cur_entry, stderr);
 812                         }
 813                         free_lookup_table_entry(cur_entry);
 814                         continue;
 815                 }
 816
 817                 /* Finally, insert the stream into the lookup table, keyed by
 818                  * its SHA1 message digest.  */
 819                 lookup_table_insert(table, cur_entry);
 820         }
 821
 822         /* Validate the last resource.  */
 823         if (cur_rspec != NULL) {
 824                 ret = validate_resource(cur_rspec);
 825                 if (ret)
 826                         goto out_free_lookup_table;
 827         }
 828
 829         if (wim->hdr.part_number == 1 && wim->current_image != wim->hdr.image_count) {
 830                 WARNING("The header of \"%"TS"\" says there are %u images in\n"
 831                         "          the WIM, but we only found %d metadata resources!  Acting as if\n"
 832                         "          the header specified only %d images instead.",
 833                         wim->filename, wim->hdr.image_count,
 834                         wim->current_image, wim->current_image);
 835                 for (int i = wim->current_image; i < wim->hdr.image_count; i++)
 836                         put_image_metadata(wim->image_metadata[i], NULL);
 837                 wim->hdr.image_count = wim->current_image;
 838         }
 839         DEBUG("Done reading lookup table.");
 840         wim->lookup_table = table;
 841         ret = 0;
 842         goto out_free_buf;
 843
 844 out_free_cur_entry:
 845         FREE(cur_entry);
 846 out_free_lookup_table:
 847         free_lookup_table(table);
 848 out_free_buf:
 849         FREE(buf);
 850 out:
 851         wim->current_image = 0;
 852         return ret;
 853 }
 854
 855 static void
 856 put_wim_lookup_table_entry(struct wim_lookup_table_entry_disk *disk_entry,
 857                            const struct wim_reshdr *out_reshdr,
 858                            u16 part_number, u32 refcnt, const u8 *hash)
 859 {
 860         put_wim_reshdr(out_reshdr, &disk_entry->reshdr);
 861         disk_entry->part_number = cpu_to_le16(part_number);
 862         disk_entry->refcnt = cpu_to_le32(refcnt);
 863         copy_hash(disk_entry->hash, hash);
 864 }
 865
 866 int
 867 write_wim_lookup_table_from_stream_list(struct list_head *stream_list,
 868                                         struct filedes *out_fd,
 869                                         u16 part_number,
 870                                         struct wim_reshdr *out_reshdr,
 871                                         int write_resource_flags)
 872 {
 873         size_t table_size;
 874         struct wim_lookup_table_entry *lte;
 875         struct wim_lookup_table_entry_disk *table_buf;
 876         struct wim_lookup_table_entry_disk *table_buf_ptr;
 877         int ret;
 878         u64 prev_res_offset_in_wim = ~0ULL;
 879
 880         table_size = 0;
 881         list_for_each_entry(lte, stream_list, lookup_table_list) {
 882                 table_size += sizeof(struct wim_lookup_table_entry_disk);
 883
 884                 if (lte->out_reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS &&
 885                     lte->out_res_offset_in_wim != prev_res_offset_in_wim)
 886                 {
 887                         table_size += sizeof(struct wim_lookup_table_entry_disk);
 888                         prev_res_offset_in_wim = lte->out_res_offset_in_wim;
 889                 }
 890         }
 891
 892         DEBUG("Writing WIM lookup table (size=%zu, offset=%"PRIu64")",
 893               table_size, out_fd->offset);
 894
 895         table_buf = MALLOC(table_size);
 896         if (table_buf == NULL) {
 897                 ERROR("Failed to allocate %zu bytes for temporary lookup table",
 898                       table_size);
 899                 return WIMLIB_ERR_NOMEM;
 900         }
 901         table_buf_ptr = table_buf;
 902
 903         prev_res_offset_in_wim = ~0ULL;
 904         list_for_each_entry(lte, stream_list, lookup_table_list) {
 905
 906                 put_wim_lookup_table_entry(table_buf_ptr++,
 907                                            &lte->out_reshdr,
 908                                            part_number,
 909                                            lte->out_refcnt,
 910                                            lte->hash);
 911                 if (lte->out_reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS &&
 912                     lte->out_res_offset_in_wim != prev_res_offset_in_wim)
 913                 {
 914                         /* Put the main resource entry for the pack.  */
 915
 916                         struct wim_reshdr reshdr;
 917
 918                         reshdr.offset_in_wim = lte->out_res_offset_in_wim;
 919                         reshdr.size_in_wim = lte->out_res_size_in_wim;
 920                         reshdr.uncompressed_size = WIM_PACK_MAGIC_NUMBER;
 921                         reshdr.flags = WIM_RESHDR_FLAG_PACKED_STREAMS;
 922
 923                         DEBUG("Putting main entry for pack: "
 924                               "size_in_wim=%"PRIu64", "
 925                               "offset_in_wim=%"PRIu64", "
 926                               "uncompressed_size=%"PRIu64,
 927                               reshdr.size_in_wim,
 928                               reshdr.offset_in_wim,
 929                               reshdr.uncompressed_size);
 930
 931                         put_wim_lookup_table_entry(table_buf_ptr++,
 932                                                    &reshdr,
 933                                                    part_number,
 934                                                    1, zero_hash);
 935                         prev_res_offset_in_wim = lte->out_res_offset_in_wim;
 936                 }
 937
 938         }
 939         wimlib_assert((u8*)table_buf_ptr - (u8*)table_buf == table_size);
 940
 941         /* Write the lookup table uncompressed.  Although wimlib can handle a
 942          * compressed lookup table, MS software cannot.  */
 943         ret = write_wim_resource_from_buffer(table_buf,
 944                                              table_size,
 945                                              WIM_RESHDR_FLAG_METADATA,
 946                                              out_fd,
 947                                              WIMLIB_COMPRESSION_TYPE_NONE,
 948                                              0,
 949                                              out_reshdr,
 950                                              NULL,
 951                                              write_resource_flags);
 952         FREE(table_buf);
 953         DEBUG("ret=%d", ret);
 954         return ret;
 955 }
 956
 957 int
 958 lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 959 {
 960         lte->real_refcnt = 0;
 961         return 0;
 962 }
 963
 964 int
 965 lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 966 {
 967         lte->out_refcnt = 0;
 968         return 0;
 969 }
 970
 971 int
 972 lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *_ignore)
 973 {
 974         if (lte->extracted_file != NULL) {
 975                 FREE(lte->extracted_file);
 976                 lte->extracted_file = NULL;
 977         }
 978         return 0;
 979 }
 980
 981 void
 982 print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 983 {
 984         if (lte == NULL) {
 985                 tputc(T('\n'), out);
 986                 return;
 987         }
 988
 989
 990         tprintf(T("Uncompressed size     = %"PRIu64" bytes\n"),
 991                 lte->size);
 992         if (lte->flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
 993                 tprintf(T("Offset                = %"PRIu64" bytes\n"),
 994                         lte->offset_in_res);
 995
 996                 tprintf(T("Raw uncompressed size = %"PRIu64" bytes\n"),
 997                         lte->rspec->uncompressed_size);
 998
 999                 tprintf(T("Raw compressed size   = %"PRIu64" bytes\n"),
1000                         lte->rspec->size_in_wim);
1001
1002                 tprintf(T("Raw offset            = %"PRIu64" bytes\n"),
1003                         lte->rspec->offset_in_wim);
1004         } else if (lte->resource_location == RESOURCE_IN_WIM) {
1005                 tprintf(T("Compressed size       = %"PRIu64" bytes\n"),
1006                         lte->rspec->size_in_wim);
1007
1008                 tprintf(T("Offset                = %"PRIu64" bytes\n"),
1009                         lte->rspec->offset_in_wim);
1010         }
1011
1012         tfprintf(out, T("Reference Count       = %u\n"), lte->refcnt);
1013
1014         if (lte->unhashed) {
1015                 tfprintf(out, T("(Unhashed: inode %p, stream_id = %u)\n"),
1016                          lte->back_inode, lte->back_stream_id);
1017         } else {
1018                 tfprintf(out, T("Hash                  = 0x"));
1019                 print_hash(lte->hash, out);
1020                 tputc(T('\n'), out);
1021         }
1022
1023         tfprintf(out, T("Flags                 = "));
1024         u8 flags = lte->flags;
1025         if (flags & WIM_RESHDR_FLAG_COMPRESSED)
1026                 tfputs(T("WIM_RESHDR_FLAG_COMPRESSED, "), out);
1027         if (flags & WIM_RESHDR_FLAG_FREE)
1028                 tfputs(T("WIM_RESHDR_FLAG_FREE, "), out);
1029         if (flags & WIM_RESHDR_FLAG_METADATA)
1030                 tfputs(T("WIM_RESHDR_FLAG_METADATA, "), out);
1031         if (flags & WIM_RESHDR_FLAG_SPANNED)
1032                 tfputs(T("WIM_RESHDR_FLAG_SPANNED, "), out);
1033         if (flags & WIM_RESHDR_FLAG_PACKED_STREAMS)
1034                 tfputs(T("WIM_RESHDR_FLAG_PACKED_STREAMS, "), out);
1035         tputc(T('\n'), out);
1036         switch (lte->resource_location) {
1037         case RESOURCE_IN_WIM:
1038                 if (lte->rspec->wim->filename) {
1039                         tfprintf(out, T("WIM file              = `%"TS"'\n"),
1040                                  lte->rspec->wim->filename);
1041                 }
1042                 break;
1043 #ifdef __WIN32__
1044         case RESOURCE_WIN32_ENCRYPTED:
1045 #endif
1046         case RESOURCE_IN_FILE_ON_DISK:
1047                 tfprintf(out, T("File on Disk          = `%"TS"'\n"),
1048                          lte->file_on_disk);
1049                 break;
1050 #ifdef WITH_FUSE
1051         case RESOURCE_IN_STAGING_FILE:
1052                 tfprintf(out, T("Staging File          = `%"TS"'\n"),
1053                                 lte->staging_file_name);
1054                 break;
1055 #endif
1056         default:
1057                 break;
1058         }
1059         tputc(T('\n'), out);
1060 }
1061
1062 void
1063 lte_to_wimlib_resource_entry(const struct wim_lookup_table_entry *lte,
1064                              struct wimlib_resource_entry *wentry)
1065 {
1066         memset(wentry, 0, sizeof(*wentry));
1067
1068         wentry->uncompressed_size = lte->size;
1069         if (lte->resource_location == RESOURCE_IN_WIM) {
1070                 wentry->part_number = lte->rspec->wim->hdr.part_number;
1071                 if (lte->flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
1072                         wentry->compressed_size = 0;
1073                         wentry->offset = lte->offset_in_res;
1074                 } else {
1075                         wentry->compressed_size = lte->rspec->size_in_wim;
1076                         wentry->offset = lte->rspec->offset_in_wim;
1077                 }
1078                 wentry->raw_resource_offset_in_wim = lte->rspec->offset_in_wim;
1079                 /*wentry->raw_resource_uncompressed_size = lte->rspec->uncompressed_size;*/
1080                 wentry->raw_resource_compressed_size = lte->rspec->size_in_wim;
1081         }
1082         copy_hash(wentry->sha1_hash, lte->hash);
1083         wentry->reference_count = lte->refcnt;
1084         wentry->is_compressed = (lte->flags & WIM_RESHDR_FLAG_COMPRESSED) != 0;
1085         wentry->is_metadata = (lte->flags & WIM_RESHDR_FLAG_METADATA) != 0;
1086         wentry->is_free = (lte->flags & WIM_RESHDR_FLAG_FREE) != 0;
1087         wentry->is_spanned = (lte->flags & WIM_RESHDR_FLAG_SPANNED) != 0;
1088         wentry->packed = (lte->flags & WIM_RESHDR_FLAG_PACKED_STREAMS) != 0;
1089 }
1090
1091 struct iterate_lte_context {
1092         wimlib_iterate_lookup_table_callback_t cb;
1093         void *user_ctx;
1094 };
1095
1096 static int
1097 do_iterate_lte(struct wim_lookup_table_entry *lte, void *_ctx)
1098 {
1099         struct iterate_lte_context *ctx = _ctx;
1100         struct wimlib_resource_entry entry;
1101
1102         lte_to_wimlib_resource_entry(lte, &entry);
1103         return (*ctx->cb)(&entry, ctx->user_ctx);
1104 }
1105
1106 /* API function documented in wimlib.h  */
1107 WIMLIBAPI int
1108 wimlib_iterate_lookup_table(WIMStruct *wim, int flags,
1109                             wimlib_iterate_lookup_table_callback_t cb,
1110                             void *user_ctx)
1111 {
1112         struct iterate_lte_context ctx = {
1113                 .cb = cb,
1114                 .user_ctx = user_ctx,
1115         };
1116         if (wim->hdr.part_number == 1) {
1117                 int ret;
1118                 for (int i = 0; i < wim->hdr.image_count; i++) {
1119                         ret = do_iterate_lte(wim->image_metadata[i]->metadata_lte,
1120                                              &ctx);
1121                         if (ret)
1122                                 return ret;
1123                 }
1124         }
1125         return for_lookup_table_entry(wim->lookup_table, do_iterate_lte, &ctx);
1126 }
1127
1128 /* Given a SHA1 message digest, return the corresponding entry in the WIM's
1129  * lookup table, or NULL if there is none.  */
1130 struct wim_lookup_table_entry *
1131 lookup_resource(const struct wim_lookup_table *table, const u8 hash[])
1132 {
1133         size_t i;
1134         struct wim_lookup_table_entry *lte;
1135         struct hlist_node *pos;
1136
1137         wimlib_assert(table != NULL);
1138         wimlib_assert(hash != NULL);
1139
1140         i = *(size_t*)hash % table->capacity;
1141         hlist_for_each_entry(lte, pos, &table->array[i], hash_list)
1142                 if (hashes_equal(hash, lte->hash))
1143                         return lte;
1144         return NULL;
1145 }
1146
1147 #ifdef WITH_FUSE
1148 /*
1149  * Finds the dentry, lookup table entry, and stream index for a WIM file stream,
1150  * given a path name.
1151  *
1152  * This is only for pre-resolved inodes.
1153  */
1154 int
1155 wim_pathname_to_stream(WIMStruct *wim,
1156                        const tchar *path,
1157                        int lookup_flags,
1158                        struct wim_dentry **dentry_ret,
1159                        struct wim_lookup_table_entry **lte_ret,
1160                        u16 *stream_idx_ret)
1161 {
1162         struct wim_dentry *dentry;
1163         struct wim_lookup_table_entry *lte;
1164         u16 stream_idx;
1165         const tchar *stream_name = NULL;
1166         struct wim_inode *inode;
1167         tchar *p = NULL;
1168
1169         if (lookup_flags & LOOKUP_FLAG_ADS_OK) {
1170                 stream_name = path_stream_name(path);
1171                 if (stream_name) {
1172                         p = (tchar*)stream_name - 1;
1173                         *p = T('\0');
1174                 }
1175         }
1176
1177         dentry = get_dentry(wim, path);
1178         if (p)
1179                 *p = T(':');
1180         if (!dentry)
1181                 return -errno;
1182
1183         inode = dentry->d_inode;
1184
1185         if (!inode->i_resolved)
1186                 if (inode_resolve_ltes(inode, wim->lookup_table, false))
1187                         return -EIO;
1188
1189         if (!(lookup_flags & LOOKUP_FLAG_DIRECTORY_OK)
1190               && inode_is_directory(inode))
1191                 return -EISDIR;
1192
1193         if (stream_name) {
1194                 struct wim_ads_entry *ads_entry;
1195                 u16 ads_idx;
1196                 ads_entry = inode_get_ads_entry(inode, stream_name,
1197                                                 &ads_idx);
1198                 if (ads_entry) {
1199                         stream_idx = ads_idx + 1;
1200                         lte = ads_entry->lte;
1201                         goto out;
1202                 } else {
1203                         return -ENOENT;
1204                 }
1205         } else {
1206                 lte = inode_unnamed_stream_resolved(inode, &stream_idx);
1207         }
1208 out:
1209         if (dentry_ret)
1210                 *dentry_ret = dentry;
1211         if (lte_ret)
1212                 *lte_ret = lte;
1213         if (stream_idx_ret)
1214                 *stream_idx_ret = stream_idx;
1215         return 0;
1216 }
1217 #endif
1218
1219 int
1220 resource_not_found_error(const struct wim_inode *inode, const u8 *hash)
1221 {
1222         if (wimlib_print_errors) {
1223                 ERROR("\"%"TS"\": resource not found", inode_first_full_path(inode));
1224                 tfprintf(stderr, T("        SHA-1 message digest of missing resource:\n        "));
1225                 print_hash(hash, stderr);
1226                 tputc(T('\n'), stderr);
1227         }
1228         return WIMLIB_ERR_RESOURCE_NOT_FOUND;
1229 }
1230
1231 /*
1232  * Resolve an inode's lookup table entries.
1233  *
1234  * This replaces the SHA1 hash fields (which are used to lookup an entry in the
1235  * lookup table) with pointers directly to the lookup table entries.
1236  *
1237  * If @force is %false:
1238  *      If any needed SHA1 message digests are not found in the lookup table,
1239  *      WIMLIB_ERR_RESOURCE_NOT_FOUND is returned and the inode is left
1240  *      unmodified.
1241  * If @force is %true:
1242  *      If any needed SHA1 message digests are not found in the lookup table,
1243  *      new entries are allocated and inserted into the lookup table.
1244  */
1245 int
1246 inode_resolve_ltes(struct wim_inode *inode, struct wim_lookup_table *table,
1247                    bool force)
1248 {
1249         const u8 *hash;
1250
1251         if (!inode->i_resolved) {
1252                 struct wim_lookup_table_entry *lte, *ads_lte;
1253
1254                 /* Resolve the default file stream */
1255                 lte = NULL;
1256                 hash = inode->i_hash;
1257                 if (!is_zero_hash(hash)) {
1258                         lte = lookup_resource(table, hash);
1259                         if (!lte) {
1260                                 if (force) {
1261                                         lte = new_lookup_table_entry();
1262                                         if (!lte)
1263                                                 return WIMLIB_ERR_NOMEM;
1264                                         copy_hash(lte->hash, hash);
1265                                         lookup_table_insert(table, lte);
1266                                 } else {
1267                                         goto resource_not_found;
1268                                 }
1269                         }
1270                 }
1271
1272                 /* Resolve the alternate data streams */
1273                 struct wim_lookup_table_entry *ads_ltes[inode->i_num_ads];
1274                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1275                         struct wim_ads_entry *cur_entry;
1276
1277                         ads_lte = NULL;
1278                         cur_entry = &inode->i_ads_entries[i];
1279                         hash = cur_entry->hash;
1280                         if (!is_zero_hash(hash)) {
1281                                 ads_lte = lookup_resource(table, hash);
1282                                 if (!ads_lte) {
1283                                         if (force) {
1284                                                 ads_lte = new_lookup_table_entry();
1285                                                 if (!ads_lte)
1286                                                         return WIMLIB_ERR_NOMEM;
1287                                                 copy_hash(ads_lte->hash, hash);
1288                                                 lookup_table_insert(table, ads_lte);
1289                                         } else {
1290                                                 goto resource_not_found;
1291                                         }
1292                                 }
1293                         }
1294                         ads_ltes[i] = ads_lte;
1295                 }
1296                 inode->i_lte = lte;
1297                 for (u16 i = 0; i < inode->i_num_ads; i++)
1298                         inode->i_ads_entries[i].lte = ads_ltes[i];
1299                 inode->i_resolved = 1;
1300         }
1301         return 0;
1302
1303 resource_not_found:
1304         return resource_not_found_error(inode, hash);
1305 }
1306
1307 void
1308 inode_unresolve_ltes(struct wim_inode *inode)
1309 {
1310         if (inode->i_resolved) {
1311                 if (inode->i_lte)
1312                         copy_hash(inode->i_hash, inode->i_lte->hash);
1313                 else
1314                         zero_out_hash(inode->i_hash);
1315
1316                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1317                         if (inode->i_ads_entries[i].lte)
1318                                 copy_hash(inode->i_ads_entries[i].hash,
1319                                           inode->i_ads_entries[i].lte->hash);
1320                         else
1321                                 zero_out_hash(inode->i_ads_entries[i].hash);
1322                 }
1323                 inode->i_resolved = 0;
1324         }
1325 }
1326
1327 /*
1328  * Returns the lookup table entry for stream @stream_idx of the inode, where
1329  * stream_idx = 0 means the default un-named file stream, and stream_idx >= 1
1330  * corresponds to an alternate data stream.
1331  *
1332  * This works for both resolved and un-resolved inodes.
1333  */
1334 struct wim_lookup_table_entry *
1335 inode_stream_lte(const struct wim_inode *inode, unsigned stream_idx,
1336                  const struct wim_lookup_table *table)
1337 {
1338         if (inode->i_resolved)
1339                 return inode_stream_lte_resolved(inode, stream_idx);
1340         else
1341                 return inode_stream_lte_unresolved(inode, stream_idx, table);
1342 }
1343
1344 struct wim_lookup_table_entry *
1345 inode_unnamed_stream_resolved(const struct wim_inode *inode, u16 *stream_idx_ret)
1346 {
1347         wimlib_assert(inode->i_resolved);
1348         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
1349                 if (inode_stream_name_nbytes(inode, i) == 0 &&
1350                     !is_zero_hash(inode_stream_hash_resolved(inode, i)))
1351                 {
1352                         *stream_idx_ret = i;
1353                         return inode_stream_lte_resolved(inode, i);
1354                 }
1355         }
1356         *stream_idx_ret = 0;
1357         return NULL;
1358 }
1359
1360 struct wim_lookup_table_entry *
1361 inode_unnamed_lte_resolved(const struct wim_inode *inode)
1362 {
1363         u16 stream_idx;
1364         return inode_unnamed_stream_resolved(inode, &stream_idx);
1365 }
1366
1367 struct wim_lookup_table_entry *
1368 inode_unnamed_lte_unresolved(const struct wim_inode *inode,
1369                              const struct wim_lookup_table *table)
1370 {
1371         wimlib_assert(!inode->i_resolved);
1372         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
1373                 if (inode_stream_name_nbytes(inode, i) == 0 &&
1374                     !is_zero_hash(inode_stream_hash_unresolved(inode, i)))
1375                 {
1376                         return inode_stream_lte_unresolved(inode, i, table);
1377                 }
1378         }
1379         return NULL;
1380 }
1381
1382 /* Return the lookup table entry for the unnamed data stream of an inode, or
1383  * NULL if there is none.
1384  *
1385  * You'd think this would be easier than it actually is, since the unnamed data
1386  * stream should be the one referenced from the inode itself.  Alas, if there
1387  * are named data streams, Microsoft's "imagex.exe" program will put the unnamed
1388  * data stream in one of the alternate data streams instead of inside the WIM
1389  * dentry itself.  So we need to check the alternate data streams too.
1390  *
1391  * Also, note that a dentry may appear to have more than one unnamed stream, but
1392  * if the SHA1 message digest is all 0's then the corresponding stream does not
1393  * really "count" (this is the case for the inode's own file stream when the
1394  * file stream that should be there is actually in one of the alternate stream
1395  * entries.).  This is despite the fact that we may need to extract such a
1396  * missing entry as an empty file or empty named data stream.
1397  */
1398 struct wim_lookup_table_entry *
1399 inode_unnamed_lte(const struct wim_inode *inode,
1400                   const struct wim_lookup_table *table)
1401 {
1402         if (inode->i_resolved)
1403                 return inode_unnamed_lte_resolved(inode);
1404         else
1405                 return inode_unnamed_lte_unresolved(inode, table);
1406 }
1407
1408 /* Returns the SHA1 message digest of the unnamed data stream of a WIM inode, or
1409  * 'zero_hash' if the unnamed data stream is missing has all zeroes in its SHA1
1410  * message digest field.  */
1411 const u8 *
1412 inode_unnamed_stream_hash(const struct wim_inode *inode)
1413 {
1414         const u8 *hash;
1415
1416         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
1417                 if (inode_stream_name_nbytes(inode, i) == 0) {
1418                         hash = inode_stream_hash(inode, i);
1419                         if (!is_zero_hash(hash))
1420                                 return hash;
1421                 }
1422         }
1423         return zero_hash;
1424 }
1425
1426 struct wim_lookup_table_entry **
1427 retrieve_lte_pointer(struct wim_lookup_table_entry *lte)
1428 {
1429         wimlib_assert(lte->unhashed);
1430         struct wim_inode *inode = lte->back_inode;
1431         u32 stream_id = lte->back_stream_id;
1432         if (stream_id == 0)
1433                 return &inode->i_lte;
1434         else
1435                 for (u16 i = 0; i < inode->i_num_ads; i++)
1436                         if (inode->i_ads_entries[i].stream_id == stream_id)
1437                                 return &inode->i_ads_entries[i].lte;
1438         wimlib_assert(0);
1439         return NULL;
1440 }
1441
1442 /* Calculate the SHA1 message digest of a stream and move it from the list of
1443  * unhashed streams to the stream lookup table, possibly joining it with an
1444  * existing lookup table entry for an identical stream.
1445  *
1446  * @lte:  An unhashed lookup table entry.
1447  * @lookup_table:  Lookup table for the WIM.
1448  * @lte_ret:  On success, write a pointer to the resulting lookup table
1449  *            entry to this location.  This will be the same as @lte
1450  *            if it was inserted into the lookup table, or different if
1451  *            a duplicate stream was found.
1452  *
1453  * Returns 0 on success; nonzero if there is an error reading the stream.
1454  */
1455 int
1456 hash_unhashed_stream(struct wim_lookup_table_entry *lte,
1457                      struct wim_lookup_table *lookup_table,
1458                      struct wim_lookup_table_entry **lte_ret)
1459 {
1460         int ret;
1461         struct wim_lookup_table_entry *duplicate_lte;
1462         struct wim_lookup_table_entry **back_ptr;
1463
1464         wimlib_assert(lte->unhashed);
1465
1466         /* back_ptr must be saved because @back_inode and @back_stream_id are in
1467          * union with the SHA1 message digest and will no longer be valid once
1468          * the SHA1 has been calculated. */
1469         back_ptr = retrieve_lte_pointer(lte);
1470
1471         ret = sha1_stream(lte);
1472         if (ret)
1473                 return ret;
1474
1475         /* Look for a duplicate stream */
1476         duplicate_lte = lookup_resource(lookup_table, lte->hash);
1477         list_del(&lte->unhashed_list);
1478         if (duplicate_lte) {
1479                 /* We have a duplicate stream.  Transfer the reference counts
1480                  * from this stream to the duplicate and update the reference to
1481                  * this stream (in an inode or ads_entry) to point to the
1482                  * duplicate.  The caller is responsible for freeing @lte if
1483                  * needed.  */
1484                 wimlib_assert(!(duplicate_lte->unhashed));
1485                 wimlib_assert(duplicate_lte->size == lte->size);
1486                 duplicate_lte->refcnt += lte->refcnt;
1487                 lte->refcnt = 0;
1488                 *back_ptr = duplicate_lte;
1489                 lte = duplicate_lte;
1490         } else {
1491                 /* No duplicate stream, so we need to insert this stream into
1492                  * the lookup table and treat it as a hashed stream. */
1493                 lookup_table_insert(lookup_table, lte);
1494                 lte->unhashed = 0;
1495         }
1496         *lte_ret = lte;
1497         return 0;
1498 }
1499
1500 static int
1501 lte_clone_if_new(struct wim_lookup_table_entry *lte, void *_lookup_table)
1502 {
1503         struct wim_lookup_table *lookup_table = _lookup_table;
1504
1505         if (lookup_resource(lookup_table, lte->hash))
1506                 return 0;  /*  Resource already present.  */
1507
1508         lte = clone_lookup_table_entry(lte);
1509         if (lte == NULL)
1510                 return WIMLIB_ERR_NOMEM;
1511         lte->out_refcnt = 1;
1512         lookup_table_insert(lookup_table, lte);
1513         return 0;
1514 }
1515
1516 static int
1517 lte_delete_if_new(struct wim_lookup_table_entry *lte, void *_lookup_table)
1518 {
1519         struct wim_lookup_table *lookup_table = _lookup_table;
1520
1521         if (lte->out_refcnt) {
1522                 lookup_table_unlink(lookup_table, lte);
1523                 free_lookup_table_entry(lte);
1524         }
1525         return 0;
1526 }
1527
1528 /* API function documented in wimlib.h  */
1529 WIMLIBAPI int
1530 wimlib_reference_resources(WIMStruct *wim,
1531                            WIMStruct **resource_wims, unsigned num_resource_wims,
1532                            int ref_flags)
1533 {
1534         int ret;
1535         unsigned i;
1536
1537         if (wim == NULL)
1538                 return WIMLIB_ERR_INVALID_PARAM;
1539
1540         if (num_resource_wims != 0 && resource_wims == NULL)
1541                 return WIMLIB_ERR_INVALID_PARAM;
1542
1543         for (i = 0; i < num_resource_wims; i++)
1544                 if (resource_wims[i] == NULL)
1545                         return WIMLIB_ERR_INVALID_PARAM;
1546
1547         for_lookup_table_entry(wim->lookup_table, lte_zero_out_refcnt, NULL);
1548
1549         for (i = 0; i < num_resource_wims; i++) {
1550                 ret = for_lookup_table_entry(resource_wims[i]->lookup_table,
1551                                              lte_clone_if_new,
1552                                              wim->lookup_table);
1553                 if (ret)
1554                         goto out_rollback;
1555         }
1556         return 0;
1557
1558 out_rollback:
1559         for_lookup_table_entry(wim->lookup_table, lte_delete_if_new,
1560                                wim->lookup_table);
1561         return ret;
1562 }
1563
1564 static int
1565 reference_resource_paths(WIMStruct *wim,
1566                          const tchar * const *resource_wimfiles,
1567                          unsigned num_resource_wimfiles,
1568                          int ref_flags,
1569                          int open_flags,
1570                          wimlib_progress_func_t progress_func)
1571 {
1572         WIMStruct **resource_wims;
1573         unsigned i;
1574         int ret;
1575
1576         resource_wims = CALLOC(num_resource_wimfiles, sizeof(resource_wims[0]));
1577         if (!resource_wims)
1578                 return WIMLIB_ERR_NOMEM;
1579
1580         for (i = 0; i < num_resource_wimfiles; i++) {
1581                 DEBUG("Referencing resources from path \"%"TS"\"",
1582                       resource_wimfiles[i]);
1583                 ret = wimlib_open_wim(resource_wimfiles[i], open_flags,
1584                                       &resource_wims[i], progress_func);
1585                 if (ret)
1586                         goto out_free_resource_wims;
1587         }
1588
1589         ret = wimlib_reference_resources(wim, resource_wims,
1590                                          num_resource_wimfiles, ref_flags);
1591         if (ret)
1592                 goto out_free_resource_wims;
1593
1594         for (i = 0; i < num_resource_wimfiles; i++)
1595                 list_add_tail(&resource_wims[i]->subwim_node, &wim->subwims);
1596
1597         ret = 0;
1598         goto out_free_array;
1599
1600 out_free_resource_wims:
1601         for (i = 0; i < num_resource_wimfiles; i++)
1602                 wimlib_free(resource_wims[i]);
1603 out_free_array:
1604         FREE(resource_wims);
1605         return ret;
1606 }
1607
1608 static int
1609 reference_resource_glob(WIMStruct *wim, const tchar *refglob,
1610                         int ref_flags, int open_flags,
1611                         wimlib_progress_func_t progress_func)
1612 {
1613         glob_t globbuf;
1614         int ret;
1615
1616         /* Note: glob() is replaced in Windows native builds.  */
1617         ret = tglob(refglob, GLOB_ERR | GLOB_NOSORT, NULL, &globbuf);
1618         if (ret) {
1619                 if (ret == GLOB_NOMATCH) {
1620                         if (ref_flags & WIMLIB_REF_FLAG_GLOB_ERR_ON_NOMATCH) {
1621                                 ERROR("Found no files for glob \"%"TS"\"", refglob);
1622                                 return WIMLIB_ERR_GLOB_HAD_NO_MATCHES;
1623                         } else {
1624                                 return reference_resource_paths(wim,
1625                                                                 &refglob,
1626                                                                 1,
1627                                                                 ref_flags,
1628                                                                 open_flags,
1629                                                                 progress_func);
1630                         }
1631                 } else {
1632                         ERROR_WITH_ERRNO("Failed to process glob \"%"TS"\"", refglob);
1633                         if (ret == GLOB_NOSPACE)
1634                                 return WIMLIB_ERR_NOMEM;
1635                         else
1636                                 return WIMLIB_ERR_READ;
1637                 }
1638         }
1639
1640         ret = reference_resource_paths(wim,
1641                                        (const tchar * const *)globbuf.gl_pathv,
1642                                        globbuf.gl_pathc,
1643                                        ref_flags,
1644                                        open_flags,
1645                                        progress_func);
1646         globfree(&globbuf);
1647         return ret;
1648 }
1649
1650 /* API function documented in wimlib.h  */
1651 WIMLIBAPI int
1652 wimlib_reference_resource_files(WIMStruct *wim,
1653                                 const tchar * const * resource_wimfiles_or_globs,
1654                                 unsigned count,
1655                                 int ref_flags,
1656                                 int open_flags,
1657                                 wimlib_progress_func_t progress_func)
1658 {
1659         unsigned i;
1660         int ret;
1661
1662         if (ref_flags & WIMLIB_REF_FLAG_GLOB_ENABLE) {
1663                 for (i = 0; i < count; i++) {
1664                         ret = reference_resource_glob(wim,
1665                                                       resource_wimfiles_or_globs[i],
1666                                                       ref_flags,
1667                                                       open_flags,
1668                                                       progress_func);
1669                         if (ret)
1670                                 return ret;
1671                 }
1672                 return 0;
1673         } else {
1674                 return reference_resource_paths(wim, resource_wimfiles_or_globs,
1675                                                 count, ref_flags,
1676                                                 open_flags, progress_func);
1677         }
1678 }