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) {
  78                 lte->part_number  = 1;
  79                 lte->refcnt       = 1;
  80         } else {
  81                 ERROR("Out of memory (tried to allocate %zu bytes for "
  82                       "lookup table entry)",
  83                       sizeof(struct wim_lookup_table_entry));
  84         }
  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)
  95                 return NULL;
  96
  97         new->extracted_file = NULL;
  98         switch (new->resource_location) {
  99         case RESOURCE_IN_FILE_ON_DISK:
 100 #ifdef __WIN32__
 101         case RESOURCE_WIN32_ENCRYPTED:
 102 #endif
 103 #ifdef WITH_FUSE
 104         case RESOURCE_IN_STAGING_FILE:
 105                 BUILD_BUG_ON((void*)&old->file_on_disk !=
 106                              (void*)&old->staging_file_name);
 107 #endif
 108                 new->file_on_disk = TSTRDUP(old->file_on_disk);
 109                 if (!new->file_on_disk)
 110                         goto out_free;
 111                 break;
 112         case RESOURCE_IN_ATTACHED_BUFFER:
 113                 new->attached_buffer = memdup(old->attached_buffer,
 114                                               wim_resource_size(old));
 115                 if (!new->attached_buffer)
 116                         goto out_free;
 117                 break;
 118 #ifdef WITH_NTFS_3G
 119         case RESOURCE_IN_NTFS_VOLUME:
 120                 if (old->ntfs_loc) {
 121                         struct ntfs_location *loc;
 122                         loc = memdup(old->ntfs_loc, sizeof(struct ntfs_location));
 123                         if (!loc)
 124                                 goto out_free;
 125                         loc->path = NULL;
 126                         loc->stream_name = NULL;
 127                         new->ntfs_loc = loc;
 128                         loc->path = STRDUP(old->ntfs_loc->path);
 129                         if (!loc->path)
 130                                 goto out_free;
 131                         if (loc->stream_name_nchars) {
 132                                 loc->stream_name = memdup(old->ntfs_loc->stream_name,
 133                                                           loc->stream_name_nchars * 2);
 134                                 if (!loc->stream_name)
 135                                         goto out_free;
 136                         }
 137                 }
 138                 break;
 139 #endif
 140         default:
 141                 break;
 142         }
 143         return new;
 144 out_free:
 145         free_lookup_table_entry(new);
 146         return NULL;
 147 }
 148
 149 void
 150 free_lookup_table_entry(struct wim_lookup_table_entry *lte)
 151 {
 152         if (lte) {
 153                 switch (lte->resource_location) {
 154                 case RESOURCE_IN_FILE_ON_DISK:
 155         #ifdef __WIN32__
 156                 case RESOURCE_WIN32_ENCRYPTED:
 157         #endif
 158         #ifdef WITH_FUSE
 159                 case RESOURCE_IN_STAGING_FILE:
 160                         BUILD_BUG_ON((void*)&lte->file_on_disk !=
 161                                      (void*)&lte->staging_file_name);
 162         #endif
 163                 case RESOURCE_IN_ATTACHED_BUFFER:
 164                         BUILD_BUG_ON((void*)&lte->file_on_disk !=
 165                                      (void*)&lte->attached_buffer);
 166                         FREE(lte->file_on_disk);
 167                         break;
 168 #ifdef WITH_NTFS_3G
 169                 case RESOURCE_IN_NTFS_VOLUME:
 170                         if (lte->ntfs_loc) {
 171                                 FREE(lte->ntfs_loc->path);
 172                                 FREE(lte->ntfs_loc->stream_name);
 173                                 FREE(lte->ntfs_loc);
 174                         }
 175                         break;
 176 #endif
 177                 default:
 178                         break;
 179                 }
 180                 FREE(lte);
 181         }
 182 }
 183
 184 static int
 185 do_free_lookup_table_entry(struct wim_lookup_table_entry *entry, void *ignore)
 186 {
 187         free_lookup_table_entry(entry);
 188         return 0;
 189 }
 190
 191
 192 void
 193 free_lookup_table(struct wim_lookup_table *table)
 194 {
 195         DEBUG2("Freeing lookup table");
 196         if (table) {
 197                 if (table->array) {
 198                         for_lookup_table_entry(table,
 199                                                do_free_lookup_table_entry,
 200                                                NULL);
 201                         FREE(table->array);
 202                 }
 203                 FREE(table);
 204         }
 205 }
 206
 207 /*
 208  * Inserts an entry into the lookup table.
 209  *
 210  * @table:      A pointer to the lookup table.
 211  * @lte:        A pointer to the entry to insert.
 212  */
 213 void
 214 lookup_table_insert(struct wim_lookup_table *table,
 215                     struct wim_lookup_table_entry *lte)
 216 {
 217         size_t i = lte->hash_short % table->capacity;
 218         hlist_add_head(&lte->hash_list, &table->array[i]);
 219
 220         /* XXX Make the table grow when too many entries have been inserted. */
 221         table->num_entries++;
 222 }
 223
 224 static void
 225 finalize_lte(struct wim_lookup_table_entry *lte)
 226 {
 227         #ifdef WITH_FUSE
 228         if (lte->resource_location == RESOURCE_IN_STAGING_FILE) {
 229                 unlink(lte->staging_file_name);
 230                 list_del(&lte->unhashed_list);
 231         }
 232         #endif
 233         free_lookup_table_entry(lte);
 234 }
 235
 236 /* Decrements the reference count for the lookup table entry @lte.  If its
 237  * reference count reaches 0, it is unlinked from the lookup table.  If,
 238  * furthermore, the entry has no opened file descriptors associated with it, the
 239  * entry is freed.  */
 240 void
 241 lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
 242                      struct wim_lookup_table *table)
 243 {
 244         wimlib_assert(lte != NULL);
 245         wimlib_assert(lte->refcnt != 0);
 246         if (--lte->refcnt == 0) {
 247                 if (lte->unhashed)
 248                         list_del(&lte->unhashed_list);
 249                 else
 250                         lookup_table_unlink(table, lte);
 251         #ifdef WITH_FUSE
 252                 if (lte->num_opened_fds == 0)
 253         #endif
 254                         finalize_lte(lte);
 255         }
 256 }
 257
 258 #ifdef WITH_FUSE
 259 void
 260 lte_decrement_num_opened_fds(struct wim_lookup_table_entry *lte)
 261 {
 262         if (lte->num_opened_fds != 0)
 263                 if (--lte->num_opened_fds == 0 && lte->refcnt == 0)
 264                         finalize_lte(lte);
 265 }
 266 #endif
 267
 268 /* Calls a function on all the entries in the WIM lookup table.  Stop early and
 269  * return nonzero if any call to the function returns nonzero. */
 270 int
 271 for_lookup_table_entry(struct wim_lookup_table *table,
 272                        int (*visitor)(struct wim_lookup_table_entry *, void *),
 273                        void *arg)
 274 {
 275         struct wim_lookup_table_entry *lte;
 276         struct hlist_node *pos, *tmp;
 277         int ret;
 278
 279         for (size_t i = 0; i < table->capacity; i++) {
 280                 hlist_for_each_entry_safe(lte, pos, tmp, &table->array[i],
 281                                           hash_list)
 282                 {
 283                         wimlib_assert2(!(lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA));
 284                         ret = visitor(lte, arg);
 285                         if (ret)
 286                                 return ret;
 287                 }
 288         }
 289         return 0;
 290 }
 291
 292 /* qsort() callback that sorts streams (represented by `struct
 293  * wim_lookup_table_entry's) into an order optimized for reading and writing.
 294  *
 295  * Sorting is done primarily by resource location, then secondarily by a
 296  * per-resource location order.  For example, resources in WIM files are sorted
 297  * primarily by part number, then secondarily by offset, as to implement optimal
 298  * reading of either a standalone or split WIM.  */
 299 static int
 300 cmp_streams_by_sequential_order(const void *p1, const void *p2)
 301 {
 302         const struct wim_lookup_table_entry *lte1, *lte2;
 303         int v;
 304
 305         lte1 = *(const struct wim_lookup_table_entry**)p1;
 306         lte2 = *(const struct wim_lookup_table_entry**)p2;
 307
 308         v = (int)lte1->resource_location - (int)lte2->resource_location;
 309
 310         /* Different resource locations?  */
 311         if (v)
 312                 return v;
 313
 314         switch (lte1->resource_location) {
 315         case RESOURCE_IN_WIM:
 316
 317                 /* Different (possibly split) WIMs?  */
 318                 if (lte1->wim != lte2->wim) {
 319                         v = memcmp(lte1->wim->hdr.guid, lte2->wim->hdr.guid,
 320                                    WIM_GID_LEN);
 321                         if (v)
 322                                 return v;
 323                 }
 324
 325                 /* Different part numbers in the same WIM?  */
 326                 v = (int)lte1->wim->hdr.part_number - (int)lte2->wim->hdr.part_number;
 327                 if (v)
 328                         return v;
 329
 330                 /* Compare by offset.  */
 331                 if (lte1->resource_entry.offset < lte2->resource_entry.offset)
 332                         return -1;
 333                 else if (lte1->resource_entry.offset > lte2->resource_entry.offset)
 334                         return 1;
 335                 return 0;
 336         case RESOURCE_IN_FILE_ON_DISK:
 337 #ifdef __WIN32__
 338         case RESOURCE_WIN32_ENCRYPTED:
 339 #endif
 340                 /* Compare files by path: just a heuristic that will place files
 341                  * in the same directory next to each other.  */
 342                 return tstrcmp(lte1->file_on_disk, lte2->file_on_disk);
 343 #ifdef WITH_NTFS_3G
 344         case RESOURCE_IN_NTFS_VOLUME:
 345                 return tstrcmp(lte1->ntfs_loc->path, lte2->ntfs_loc->path);
 346 #endif
 347         default:
 348                 /* No additional sorting order defined for this resource
 349                  * location (e.g. RESOURCE_IN_ATTACHED_BUFFER); simply compare
 350                  * everything equal to each other.  */
 351                 return 0;
 352         }
 353 }
 354
 355 int
 356 sort_stream_list_by_sequential_order(struct list_head *stream_list,
 357                                      size_t list_head_offset)
 358 {
 359         struct list_head *cur;
 360         struct wim_lookup_table_entry **array;
 361         size_t i;
 362         size_t array_size;
 363         size_t num_streams = 0;
 364
 365         list_for_each(cur, stream_list)
 366                 num_streams++;
 367
 368         array_size = num_streams * sizeof(array[0]);
 369         array = MALLOC(array_size);
 370         if (!array)
 371                 return WIMLIB_ERR_NOMEM;
 372         cur = stream_list->next;
 373         for (i = 0; i < num_streams; i++) {
 374                 array[i] = (struct wim_lookup_table_entry*)((u8*)cur -
 375                                                             list_head_offset);
 376                 cur = cur->next;
 377         }
 378
 379         qsort(array, num_streams, sizeof(array[0]),
 380               cmp_streams_by_sequential_order);
 381
 382         INIT_LIST_HEAD(stream_list);
 383         for (i = 0; i < num_streams; i++) {
 384                 list_add_tail((struct list_head*)
 385                                ((u8*)array[i] + list_head_offset),
 386                               stream_list);
 387         }
 388         FREE(array);
 389         return 0;
 390 }
 391
 392
 393 static int
 394 add_lte_to_array(struct wim_lookup_table_entry *lte,
 395                  void *_pp)
 396 {
 397         struct wim_lookup_table_entry ***pp = _pp;
 398         *(*pp)++ = lte;
 399         return 0;
 400 }
 401
 402 /* Iterate through the lookup table entries, but first sort them by stream
 403  * offset in the WIM.  Caution: this is intended to be used when the stream
 404  * offset field has actually been set. */
 405 int
 406 for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
 407                                   int (*visitor)(struct wim_lookup_table_entry *,
 408                                                  void *),
 409                                   void *arg)
 410 {
 411         struct wim_lookup_table_entry **lte_array, **p;
 412         size_t num_streams = table->num_entries;
 413         int ret;
 414
 415         lte_array = MALLOC(num_streams * sizeof(lte_array[0]));
 416         if (!lte_array)
 417                 return WIMLIB_ERR_NOMEM;
 418         p = lte_array;
 419         for_lookup_table_entry(table, add_lte_to_array, &p);
 420
 421         wimlib_assert(p == lte_array + num_streams);
 422
 423         qsort(lte_array, num_streams, sizeof(lte_array[0]),
 424               cmp_streams_by_sequential_order);
 425         ret = 0;
 426         for (size_t i = 0; i < num_streams; i++) {
 427                 ret = visitor(lte_array[i], arg);
 428                 if (ret)
 429                         break;
 430         }
 431         FREE(lte_array);
 432         return ret;
 433 }
 434
 435 /* On-disk format of a WIM lookup table entry (stream entry). */
 436 struct wim_lookup_table_entry_disk {
 437         /* Location, offset, compression status, and metadata status of the
 438          * stream. */
 439         struct resource_entry_disk resource_entry;
 440
 441         /* Which part of the split WIM this stream is in; indexed from 1. */
 442         le16 part_number;
 443
 444         /* Reference count of this stream over all WIM images. */
 445         le32 refcnt;
 446
 447         /* SHA1 message digest of the uncompressed data of this stream, or
 448          * optionally all zeroes if this stream is of zero length. */
 449         u8 hash[SHA1_HASH_SIZE];
 450 } _packed_attribute;
 451
 452 #define WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE 50
 453
 454 void
 455 lte_init_wim(struct wim_lookup_table_entry *lte, WIMStruct *wim)
 456 {
 457         lte->resource_location = RESOURCE_IN_WIM;
 458         lte->wim = wim;
 459         if (lte->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
 460                 lte->compression_type = wim->compression_type;
 461         else
 462                 lte->compression_type = WIMLIB_COMPRESSION_TYPE_NONE;
 463
 464         if (wim_is_pipable(wim))
 465                 lte->is_pipable = 1;
 466 }
 467
 468 /*
 469  * Reads the lookup table from a WIM file.
 470  *
 471  * Saves lookup table entries for non-metadata streams in a hash table, and
 472  * saves the metadata entry for each image in a special per-image location (the
 473  * image_metadata array).
 474  *
 475  * Return values:
 476  *      WIMLIB_ERR_SUCCESS (0)
 477  *      WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY
 478  *      WIMLIB_ERR_RESOURCE_NOT_FOUND
 479  */
 480 int
 481 read_wim_lookup_table(WIMStruct *wim)
 482 {
 483         int ret;
 484         size_t i;
 485         size_t num_entries;
 486         struct wim_lookup_table *table;
 487         struct wim_lookup_table_entry *cur_entry, *duplicate_entry;
 488         void *buf;
 489
 490         BUILD_BUG_ON(sizeof(struct wim_lookup_table_entry_disk) !=
 491                      WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE);
 492
 493         DEBUG("Reading lookup table: offset %"PRIu64", size %"PRIu64"",
 494               wim->hdr.lookup_table_res_entry.offset,
 495               wim->hdr.lookup_table_res_entry.size);
 496
 497         /* Calculate number of entries in the lookup table.  */
 498         num_entries = wim->hdr.lookup_table_res_entry.size /
 499                       sizeof(struct wim_lookup_table_entry_disk);
 500
 501
 502         /* Read the lookup table into a buffer.  */
 503         ret = res_entry_to_data(&wim->hdr.lookup_table_res_entry, wim, &buf);
 504         if (ret)
 505                 goto out;
 506
 507         /* Allocate hash table.  */
 508         table = new_lookup_table(num_entries * 2 + 1);
 509         if (!table) {
 510                 ERROR("Not enough memory to read lookup table.");
 511                 ret = WIMLIB_ERR_NOMEM;
 512                 goto out_free_buf;
 513         }
 514
 515         /* Allocate and initalize `struct wim_lookup_table_entry's from the
 516          * on-disk lookup table.  */
 517         wim->current_image = 0;
 518         for (i = 0; i < num_entries; i++) {
 519                 const struct wim_lookup_table_entry_disk *disk_entry =
 520                         &((const struct wim_lookup_table_entry_disk*)buf)[i];
 521
 522                 cur_entry = new_lookup_table_entry();
 523                 if (!cur_entry) {
 524                         ERROR("Not enough memory to read lookup table.");
 525                         ret = WIMLIB_ERR_NOMEM;
 526                         goto out_free_lookup_table;
 527                 }
 528
 529                 cur_entry->wim = wim;
 530                 cur_entry->resource_location = RESOURCE_IN_WIM;
 531                 get_resource_entry(&disk_entry->resource_entry, &cur_entry->resource_entry);
 532                 cur_entry->part_number = le16_to_cpu(disk_entry->part_number);
 533                 cur_entry->refcnt = le32_to_cpu(disk_entry->refcnt);
 534                 copy_hash(cur_entry->hash, disk_entry->hash);
 535                 lte_init_wim(cur_entry, wim);
 536
 537                 if (cur_entry->part_number != wim->hdr.part_number) {
 538                         WARNING("A lookup table entry in part %hu of the WIM "
 539                                 "points to part %hu (ignoring it)",
 540                                 wim->hdr.part_number, cur_entry->part_number);
 541                         free_lookup_table_entry(cur_entry);
 542                         continue;
 543                 }
 544
 545                 if (is_zero_hash(cur_entry->hash)) {
 546                         WARNING("The WIM lookup table contains an entry with a "
 547                                 "SHA1 message digest of all 0's (ignoring it)");
 548                         free_lookup_table_entry(cur_entry);
 549                         continue;
 550                 }
 551
 552                 if (!(cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
 553                     && (cur_entry->resource_entry.size !=
 554                         cur_entry->resource_entry.original_size))
 555                 {
 556                         if (wimlib_print_errors) {
 557                                 WARNING("Found uncompressed resource with "
 558                                         "original size (%"PRIu64") not the same "
 559                                         "as compressed size (%"PRIu64")",
 560                                         cur_entry->resource_entry.original_size,
 561                                         cur_entry->resource_entry.size);
 562                                 if (cur_entry->resource_entry.original_size) {
 563                                         WARNING("Overriding compressed size with original size.");
 564                                         cur_entry->resource_entry.size =
 565                                                 cur_entry->resource_entry.original_size;
 566                                 } else {
 567                                         WARNING("Overriding original size with compressed size");
 568                                         cur_entry->resource_entry.original_size =
 569                                                 cur_entry->resource_entry.size;
 570                                 }
 571                         }
 572                 }
 573
 574                 if (cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_METADATA) {
 575                         /* Lookup table entry for a metadata resource */
 576                         if (cur_entry->refcnt != 1) {
 577                                 if (wimlib_print_errors) {
 578                                         ERROR("Found metadata resource with refcnt != 1:");
 579                                         print_lookup_table_entry(cur_entry, stderr);
 580                                 }
 581                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 582                                 goto out_free_cur_entry;
 583                         }
 584
 585                         if (wim->hdr.part_number != 1) {
 586                                 WARNING("Ignoring metadata resource found in a "
 587                                         "non-first part of the split WIM");
 588                                 free_lookup_table_entry(cur_entry);
 589                                 continue;
 590                         }
 591                         if (wim->current_image == wim->hdr.image_count) {
 592                                 WARNING("The WIM header says there are %u images "
 593                                         "in the WIM, but we found more metadata "
 594                                         "resources than this (ignoring the extra)",
 595                                         wim->hdr.image_count);
 596                                 free_lookup_table_entry(cur_entry);
 597                                 continue;
 598                         }
 599
 600                         /* Notice very carefully:  We are assigning the metadata
 601                          * resources in the exact order mirrored by their lookup
 602                          * table entries on disk, which is the behavior of
 603                          * Microsoft's software.  In particular, this overrides
 604                          * the actual locations of the metadata resources
 605                          * themselves in the WIM file as well as any information
 606                          * written in the XML data. */
 607                         DEBUG("Found metadata resource for image %u at "
 608                               "offset %"PRIu64".",
 609                               wim->current_image + 1,
 610                               cur_entry->resource_entry.offset);
 611                         wim->image_metadata[
 612                                 wim->current_image++]->metadata_lte = cur_entry;
 613                 } else {
 614                         /* Lookup table entry for a stream that is not a
 615                          * metadata resource */
 616                         duplicate_entry = lookup_resource(table, cur_entry->hash);
 617                         if (duplicate_entry) {
 618                                 if (wimlib_print_errors) {
 619                                         WARNING("The WIM lookup table contains two entries with the "
 620                                               "same SHA1 message digest!");
 621                                         WARNING("The first entry is:");
 622                                         print_lookup_table_entry(duplicate_entry, stderr);
 623                                         WARNING("The second entry is:");
 624                                         print_lookup_table_entry(cur_entry, stderr);
 625                                 }
 626                                 free_lookup_table_entry(cur_entry);
 627                                 continue;
 628                         } else {
 629                                 lookup_table_insert(table, cur_entry);
 630                         }
 631                 }
 632         }
 633
 634         if (wim->hdr.part_number == 1 && wim->current_image != wim->hdr.image_count) {
 635                 WARNING("The header of \"%"TS"\" says there are %u images in\n"
 636                         "          the WIM, but we only found %d metadata resources!  Acting as if\n"
 637                         "          the header specified only %d images instead.",
 638                         wim->filename, wim->hdr.image_count,
 639                         wim->current_image, wim->current_image);
 640                 for (int i = wim->current_image; i < wim->hdr.image_count; i++)
 641                         put_image_metadata(wim->image_metadata[i], NULL);
 642                 wim->hdr.image_count = wim->current_image;
 643         }
 644         DEBUG("Done reading lookup table.");
 645         wim->lookup_table = table;
 646         ret = 0;
 647         goto out_free_buf;
 648 out_free_cur_entry:
 649         FREE(cur_entry);
 650 out_free_lookup_table:
 651         free_lookup_table(table);
 652 out_free_buf:
 653         FREE(buf);
 654 out:
 655         wim->current_image = 0;
 656         return ret;
 657 }
 658
 659
 660 static void
 661 write_wim_lookup_table_entry(const struct wim_lookup_table_entry *lte,
 662                              struct wim_lookup_table_entry_disk *disk_entry)
 663 {
 664         put_resource_entry(&lte->output_resource_entry, &disk_entry->resource_entry);
 665         disk_entry->part_number = cpu_to_le16(lte->part_number);
 666         disk_entry->refcnt = cpu_to_le32(lte->out_refcnt);
 667         copy_hash(disk_entry->hash, lte->hash);
 668 }
 669
 670 static int
 671 write_wim_lookup_table_from_stream_list(struct list_head *stream_list,
 672                                         struct filedes *out_fd,
 673                                         struct resource_entry *out_res_entry,
 674                                         int write_resource_flags)
 675 {
 676         size_t table_size;
 677         struct wim_lookup_table_entry *lte;
 678         struct wim_lookup_table_entry_disk *table_buf;
 679         struct wim_lookup_table_entry_disk *table_buf_ptr;
 680         int ret;
 681
 682         table_size = 0;
 683         list_for_each_entry(lte, stream_list, lookup_table_list)
 684                 table_size += sizeof(struct wim_lookup_table_entry_disk);
 685
 686         DEBUG("Writing WIM lookup table (size=%zu, offset=%"PRIu64")",
 687               table_size, out_fd->offset);
 688
 689         table_buf = MALLOC(table_size);
 690         if (!table_buf) {
 691                 ERROR("Failed to allocate %zu bytes for temporary lookup table",
 692                       table_size);
 693                 return WIMLIB_ERR_NOMEM;
 694         }
 695         table_buf_ptr = table_buf;
 696         list_for_each_entry(lte, stream_list, lookup_table_list)
 697                 write_wim_lookup_table_entry(lte, table_buf_ptr++);
 698
 699         /* Write the lookup table uncompressed.  Although wimlib can handle a
 700          * compressed lookup table, MS software cannot.  */
 701         ret = write_wim_resource_from_buffer(table_buf,
 702                                              table_size,
 703                                              WIM_RESHDR_FLAG_METADATA,
 704                                              out_fd,
 705                                              WIMLIB_COMPRESSION_TYPE_NONE,
 706                                              out_res_entry,
 707                                              NULL,
 708                                              write_resource_flags);
 709         FREE(table_buf);
 710         return ret;
 711 }
 712
 713 static int
 714 append_lookup_table_entry(struct wim_lookup_table_entry *lte, void *_list)
 715 {
 716         /* Lookup table entries with 'out_refcnt' == 0 correspond to streams not
 717          * written and not present in the resulting WIM file, and should not be
 718          * included in the lookup table.
 719          *
 720          * Lookup table entries marked as filtered (EXTERNAL_WIM) with
 721          * 'out_refcnt != 0' were referenced as part of the logical write but
 722          * correspond to streams that were not in fact written, and should not
 723          * be included in the lookup table.
 724          *
 725          * Lookup table entries marked as filtered (SAME_WIM) with 'out_refcnt
 726          * != 0' were referenced as part of the logical write but correspond to
 727          * streams that were not in fact written, but nevertheless were already
 728          * present in the WIM being overwritten in-place.  These entries must be
 729          * included in the lookup table, and the resource information to write
 730          * needs to be copied from the resource information read originally.
 731          */
 732         if (lte->out_refcnt != 0 && !(lte->filtered & FILTERED_EXTERNAL_WIM)) {
 733                 if (lte->filtered & FILTERED_SAME_WIM) {
 734                         copy_resource_entry(&lte->output_resource_entry,
 735                                             &lte->resource_entry);
 736                 }
 737                 list_add_tail(&lte->lookup_table_list, (struct list_head*)_list);
 738         }
 739         return 0;
 740 }
 741
 742 int
 743 write_wim_lookup_table(WIMStruct *wim, int image, int write_flags,
 744                        struct resource_entry *out_res_entry,
 745                        struct list_head *stream_list_override)
 746 {
 747         int write_resource_flags;
 748         struct list_head _stream_list;
 749         struct list_head *stream_list;
 750
 751         if (stream_list_override) {
 752                 stream_list = stream_list_override;
 753         } else {
 754                 stream_list = &_stream_list;
 755                 INIT_LIST_HEAD(stream_list);
 756         }
 757
 758         if (!(write_flags & WIMLIB_WRITE_FLAG_NO_METADATA)) {
 759                 int start_image;
 760                 int end_image;
 761
 762                 if (image == WIMLIB_ALL_IMAGES) {
 763                         start_image = 1;
 764                         end_image = wim->hdr.image_count;
 765                 } else {
 766                         start_image = image;
 767                         end_image = image;
 768                 }
 769
 770                 /* Push metadata resource lookup table entries onto the front of
 771                  * the list in reverse order, so that they're written in order.
 772                  */
 773                 for (int i = end_image; i >= start_image; i--) {
 774                         struct wim_lookup_table_entry *metadata_lte;
 775
 776                         metadata_lte = wim->image_metadata[i - 1]->metadata_lte;
 777                         metadata_lte->out_refcnt = 1;
 778                         metadata_lte->part_number = wim->hdr.part_number;
 779                         metadata_lte->output_resource_entry.flags |= WIM_RESHDR_FLAG_METADATA;
 780
 781                         list_add(&metadata_lte->lookup_table_list, stream_list);
 782                 }
 783         }
 784
 785         /* Append additional lookup table entries that need to be written, with
 786          * some special handling for streams that have been marked as filtered.
 787          */
 788         if (!stream_list_override) {
 789                 for_lookup_table_entry(wim->lookup_table,
 790                                        append_lookup_table_entry, stream_list);
 791         }
 792
 793         write_resource_flags = 0;
 794         if (write_flags & WIMLIB_WRITE_FLAG_PIPABLE)
 795                 write_resource_flags |= WIMLIB_WRITE_RESOURCE_FLAG_PIPABLE;
 796         return write_wim_lookup_table_from_stream_list(stream_list,
 797                                                        &wim->out_fd,
 798                                                        out_res_entry,
 799                                                        write_resource_flags);
 800 }
 801
 802
 803 int
 804 lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 805 {
 806         lte->real_refcnt = 0;
 807         return 0;
 808 }
 809
 810 int
 811 lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 812 {
 813         lte->out_refcnt = 0;
 814         return 0;
 815 }
 816
 817 int
 818 lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *_ignore)
 819 {
 820         if (lte->extracted_file != NULL) {
 821                 FREE(lte->extracted_file);
 822                 lte->extracted_file = NULL;
 823         }
 824         return 0;
 825 }
 826
 827 void
 828 print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 829 {
 830         if (!lte) {
 831                 tputc(T('\n'), out);
 832                 return;
 833         }
 834         tfprintf(out, T("Offset            = %"PRIu64" bytes\n"),
 835                  lte->resource_entry.offset);
 836
 837         tfprintf(out, T("Size              = %"PRIu64" bytes\n"),
 838                  (u64)lte->resource_entry.size);
 839
 840         tfprintf(out, T("Original size     = %"PRIu64" bytes\n"),
 841                  lte->resource_entry.original_size);
 842
 843         tfprintf(out, T("Part Number       = %hu\n"), lte->part_number);
 844         tfprintf(out, T("Reference Count   = %u\n"), lte->refcnt);
 845
 846         if (lte->unhashed) {
 847                 tfprintf(out, T("(Unhashed: inode %p, stream_id = %u)\n"),
 848                          lte->back_inode, lte->back_stream_id);
 849         } else {
 850                 tfprintf(out, T("Hash              = 0x"));
 851                 print_hash(lte->hash, out);
 852                 tputc(T('\n'), out);
 853         }
 854
 855         tfprintf(out, T("Flags             = "));
 856         u8 flags = lte->resource_entry.flags;
 857         if (flags & WIM_RESHDR_FLAG_COMPRESSED)
 858                 tfputs(T("WIM_RESHDR_FLAG_COMPRESSED, "), out);
 859         if (flags & WIM_RESHDR_FLAG_FREE)
 860                 tfputs(T("WIM_RESHDR_FLAG_FREE, "), out);
 861         if (flags & WIM_RESHDR_FLAG_METADATA)
 862                 tfputs(T("WIM_RESHDR_FLAG_METADATA, "), out);
 863         if (flags & WIM_RESHDR_FLAG_SPANNED)
 864                 tfputs(T("WIM_RESHDR_FLAG_SPANNED, "), out);
 865         tputc(T('\n'), out);
 866         switch (lte->resource_location) {
 867         case RESOURCE_IN_WIM:
 868                 if (lte->wim->filename) {
 869                         tfprintf(out, T("WIM file          = `%"TS"'\n"),
 870                                  lte->wim->filename);
 871                 }
 872                 break;
 873 #ifdef __WIN32__
 874         case RESOURCE_WIN32_ENCRYPTED:
 875 #endif
 876         case RESOURCE_IN_FILE_ON_DISK:
 877                 tfprintf(out, T("File on Disk      = `%"TS"'\n"),
 878                          lte->file_on_disk);
 879                 break;
 880 #ifdef WITH_FUSE
 881         case RESOURCE_IN_STAGING_FILE:
 882                 tfprintf(out, T("Staging File      = `%"TS"'\n"),
 883                                 lte->staging_file_name);
 884                 break;
 885 #endif
 886         default:
 887                 break;
 888         }
 889         tputc(T('\n'), out);
 890 }
 891
 892 void
 893 lte_to_wimlib_resource_entry(const struct wim_lookup_table_entry *lte,
 894                              struct wimlib_resource_entry *wentry)
 895 {
 896         wentry->uncompressed_size = lte->resource_entry.original_size;
 897         wentry->compressed_size = lte->resource_entry.size;
 898         wentry->offset = lte->resource_entry.offset;
 899         copy_hash(wentry->sha1_hash, lte->hash);
 900         wentry->part_number = lte->part_number;
 901         wentry->reference_count = lte->refcnt;
 902         wentry->is_compressed = (lte->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED) != 0;
 903         wentry->is_metadata = (lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA) != 0;
 904         wentry->is_free = (lte->resource_entry.flags & WIM_RESHDR_FLAG_FREE) != 0;
 905         wentry->is_spanned = (lte->resource_entry.flags & WIM_RESHDR_FLAG_SPANNED) != 0;
 906 }
 907
 908 struct iterate_lte_context {
 909         wimlib_iterate_lookup_table_callback_t cb;
 910         void *user_ctx;
 911 };
 912
 913 static int
 914 do_iterate_lte(struct wim_lookup_table_entry *lte, void *_ctx)
 915 {
 916         struct iterate_lte_context *ctx = _ctx;
 917         struct wimlib_resource_entry entry;
 918
 919         lte_to_wimlib_resource_entry(lte, &entry);
 920         return (*ctx->cb)(&entry, ctx->user_ctx);
 921 }
 922
 923 /* API function documented in wimlib.h  */
 924 WIMLIBAPI int
 925 wimlib_iterate_lookup_table(WIMStruct *wim, int flags,
 926                             wimlib_iterate_lookup_table_callback_t cb,
 927                             void *user_ctx)
 928 {
 929         struct iterate_lte_context ctx = {
 930                 .cb = cb,
 931                 .user_ctx = user_ctx,
 932         };
 933         if (wim->hdr.part_number == 1) {
 934                 int ret;
 935                 for (int i = 0; i < wim->hdr.image_count; i++) {
 936                         ret = do_iterate_lte(wim->image_metadata[i]->metadata_lte,
 937                                              &ctx);
 938                         if (ret)
 939                                 return ret;
 940                 }
 941         }
 942         return for_lookup_table_entry(wim->lookup_table, do_iterate_lte, &ctx);
 943 }
 944
 945 /* Given a SHA1 message digest, return the corresponding entry in the WIM's
 946  * lookup table, or NULL if there is none.  */
 947 struct wim_lookup_table_entry *
 948 lookup_resource(const struct wim_lookup_table *table, const u8 hash[])
 949 {
 950         size_t i;
 951         struct wim_lookup_table_entry *lte;
 952         struct hlist_node *pos;
 953
 954         wimlib_assert(table != NULL);
 955         wimlib_assert(hash != NULL);
 956
 957         i = *(size_t*)hash % table->capacity;
 958         hlist_for_each_entry(lte, pos, &table->array[i], hash_list)
 959                 if (hashes_equal(hash, lte->hash))
 960                         return lte;
 961         return NULL;
 962 }
 963
 964 #ifdef WITH_FUSE
 965 /*
 966  * Finds the dentry, lookup table entry, and stream index for a WIM file stream,
 967  * given a path name.
 968  *
 969  * This is only for pre-resolved inodes.
 970  */
 971 int
 972 wim_pathname_to_stream(WIMStruct *wim,
 973                        const tchar *path,
 974                        int lookup_flags,
 975                        struct wim_dentry **dentry_ret,
 976                        struct wim_lookup_table_entry **lte_ret,
 977                        u16 *stream_idx_ret)
 978 {
 979         struct wim_dentry *dentry;
 980         struct wim_lookup_table_entry *lte;
 981         u16 stream_idx;
 982         const tchar *stream_name = NULL;
 983         struct wim_inode *inode;
 984         tchar *p = NULL;
 985
 986         if (lookup_flags & LOOKUP_FLAG_ADS_OK) {
 987                 stream_name = path_stream_name(path);
 988                 if (stream_name) {
 989                         p = (tchar*)stream_name - 1;
 990                         *p = T('\0');
 991                 }
 992         }
 993
 994         dentry = get_dentry(wim, path);
 995         if (p)
 996                 *p = T(':');
 997         if (!dentry)
 998                 return -errno;
 999
1000         inode = dentry->d_inode;
1001
1002         if (!inode->i_resolved)
1003                 if (inode_resolve_ltes(inode, wim->lookup_table, false))
1004                         return -EIO;
1005
1006         if (!(lookup_flags & LOOKUP_FLAG_DIRECTORY_OK)
1007               && inode_is_directory(inode))
1008                 return -EISDIR;
1009
1010         if (stream_name) {
1011                 struct wim_ads_entry *ads_entry;
1012                 u16 ads_idx;
1013                 ads_entry = inode_get_ads_entry(inode, stream_name,
1014                                                 &ads_idx);
1015                 if (ads_entry) {
1016                         stream_idx = ads_idx + 1;
1017                         lte = ads_entry->lte;
1018                         goto out;
1019                 } else {
1020                         return -ENOENT;
1021                 }
1022         } else {
1023                 lte = inode->i_lte;
1024                 stream_idx = 0;
1025         }
1026 out:
1027         if (dentry_ret)
1028                 *dentry_ret = dentry;
1029         if (lte_ret)
1030                 *lte_ret = lte;
1031         if (stream_idx_ret)
1032                 *stream_idx_ret = stream_idx;
1033         return 0;
1034 }
1035 #endif
1036
1037 int
1038 resource_not_found_error(struct wim_inode *inode, const u8 *hash)
1039 {
1040         if (wimlib_print_errors) {
1041                 ERROR("\"%"TS"\": resource not found", inode_first_full_path(inode));
1042                 tfprintf(stderr, T("        SHA-1 message digest of missing resource:\n        "));
1043                 print_hash(hash, stderr);
1044                 tputc(T('\n'), stderr);
1045         }
1046         return WIMLIB_ERR_RESOURCE_NOT_FOUND;
1047 }
1048
1049 /*
1050  * Resolve an inode's lookup table entries.
1051  *
1052  * This replaces the SHA1 hash fields (which are used to lookup an entry in the
1053  * lookup table) with pointers directly to the lookup table entries.
1054  *
1055  * If @force is %false:
1056  *      If any needed SHA1 message digests are not found in the lookup table,
1057  *      WIMLIB_ERR_RESOURCE_NOT_FOUND is returned and the inode is left
1058  *      unmodified.
1059  * If @force is %true:
1060  *      If any needed SHA1 message digests are not found in the lookup table,
1061  *      new entries are allocated and inserted into the lookup table.
1062  */
1063 int
1064 inode_resolve_ltes(struct wim_inode *inode, struct wim_lookup_table *table,
1065                    bool force)
1066 {
1067         const u8 *hash;
1068
1069         if (!inode->i_resolved) {
1070                 struct wim_lookup_table_entry *lte, *ads_lte;
1071
1072                 /* Resolve the default file stream */
1073                 lte = NULL;
1074                 hash = inode->i_hash;
1075                 if (!is_zero_hash(hash)) {
1076                         lte = lookup_resource(table, hash);
1077                         if (!lte) {
1078                                 if (force) {
1079                                         lte = new_lookup_table_entry();
1080                                         if (!lte)
1081                                                 return WIMLIB_ERR_NOMEM;
1082                                         copy_hash(lte->hash, hash);
1083                                         lookup_table_insert(table, lte);
1084                                 } else {
1085                                         goto resource_not_found;
1086                                 }
1087                         }
1088                 }
1089
1090                 /* Resolve the alternate data streams */
1091                 struct wim_lookup_table_entry *ads_ltes[inode->i_num_ads];
1092                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1093                         struct wim_ads_entry *cur_entry;
1094
1095                         ads_lte = NULL;
1096                         cur_entry = &inode->i_ads_entries[i];
1097                         hash = cur_entry->hash;
1098                         if (!is_zero_hash(hash)) {
1099                                 ads_lte = lookup_resource(table, hash);
1100                                 if (!ads_lte) {
1101                                         if (force) {
1102                                                 ads_lte = new_lookup_table_entry();
1103                                                 if (!ads_lte)
1104                                                         return WIMLIB_ERR_NOMEM;
1105                                                 copy_hash(ads_lte->hash, hash);
1106                                                 lookup_table_insert(table, ads_lte);
1107                                         } else {
1108                                                 goto resource_not_found;
1109                                         }
1110                                 }
1111                         }
1112                         ads_ltes[i] = ads_lte;
1113                 }
1114                 inode->i_lte = lte;
1115                 for (u16 i = 0; i < inode->i_num_ads; i++)
1116                         inode->i_ads_entries[i].lte = ads_ltes[i];
1117                 inode->i_resolved = 1;
1118         }
1119         return 0;
1120
1121 resource_not_found:
1122         return resource_not_found_error(inode, hash);
1123 }
1124
1125 void
1126 inode_unresolve_ltes(struct wim_inode *inode)
1127 {
1128         if (inode->i_resolved) {
1129                 if (inode->i_lte)
1130                         copy_hash(inode->i_hash, inode->i_lte->hash);
1131                 else
1132                         zero_out_hash(inode->i_hash);
1133
1134                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1135                         if (inode->i_ads_entries[i].lte)
1136                                 copy_hash(inode->i_ads_entries[i].hash,
1137                                           inode->i_ads_entries[i].lte->hash);
1138                         else
1139                                 zero_out_hash(inode->i_ads_entries[i].hash);
1140                 }
1141                 inode->i_resolved = 0;
1142         }
1143 }
1144
1145 /*
1146  * Returns the lookup table entry for stream @stream_idx of the inode, where
1147  * stream_idx = 0 means the default un-named file stream, and stream_idx >= 1
1148  * corresponds to an alternate data stream.
1149  *
1150  * This works for both resolved and un-resolved inodes.
1151  */
1152 struct wim_lookup_table_entry *
1153 inode_stream_lte(const struct wim_inode *inode, unsigned stream_idx,
1154                  const struct wim_lookup_table *table)
1155 {
1156         if (inode->i_resolved)
1157                 return inode_stream_lte_resolved(inode, stream_idx);
1158         else
1159                 return inode_stream_lte_unresolved(inode, stream_idx, table);
1160 }
1161
1162 struct wim_lookup_table_entry *
1163 inode_unnamed_lte_resolved(const struct wim_inode *inode)
1164 {
1165         wimlib_assert(inode->i_resolved);
1166         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
1167                 if (inode_stream_name_nbytes(inode, i) == 0 &&
1168                     !is_zero_hash(inode_stream_hash_resolved(inode, i)))
1169                 {
1170                         return inode_stream_lte_resolved(inode, i);
1171                 }
1172         }
1173         return NULL;
1174 }
1175
1176 struct wim_lookup_table_entry *
1177 inode_unnamed_lte_unresolved(const struct wim_inode *inode,
1178                              const struct wim_lookup_table *table)
1179 {
1180         wimlib_assert(!inode->i_resolved);
1181         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
1182                 if (inode_stream_name_nbytes(inode, i) == 0 &&
1183                     !is_zero_hash(inode_stream_hash_unresolved(inode, i)))
1184                 {
1185                         return inode_stream_lte_unresolved(inode, i, table);
1186                 }
1187         }
1188         return NULL;
1189 }
1190
1191 /* Return the lookup table entry for the unnamed data stream of an inode, or
1192  * NULL if there is none.
1193  *
1194  * You'd think this would be easier than it actually is, since the unnamed data
1195  * stream should be the one referenced from the inode itself.  Alas, if there
1196  * are named data streams, Microsoft's "imagex.exe" program will put the unnamed
1197  * data stream in one of the alternate data streams instead of inside the WIM
1198  * dentry itself.  So we need to check the alternate data streams too.
1199  *
1200  * Also, note that a dentry may appear to have more than one unnamed stream, but
1201  * if the SHA1 message digest is all 0's then the corresponding stream does not
1202  * really "count" (this is the case for the inode's own file stream when the
1203  * file stream that should be there is actually in one of the alternate stream
1204  * entries.).  This is despite the fact that we may need to extract such a
1205  * missing entry as an empty file or empty named data stream.
1206  */
1207 struct wim_lookup_table_entry *
1208 inode_unnamed_lte(const struct wim_inode *inode,
1209                   const struct wim_lookup_table *table)
1210 {
1211         if (inode->i_resolved)
1212                 return inode_unnamed_lte_resolved(inode);
1213         else
1214                 return inode_unnamed_lte_unresolved(inode, table);
1215 }
1216
1217 static int
1218 lte_add_stream_size(struct wim_lookup_table_entry *lte, void *total_bytes_p)
1219 {
1220         *(u64*)total_bytes_p += lte->resource_entry.size;
1221         return 0;
1222 }
1223
1224 u64
1225 lookup_table_total_stream_size(struct wim_lookup_table *table)
1226 {
1227         u64 total_size = 0;
1228         for_lookup_table_entry(table, lte_add_stream_size, &total_size);
1229         return total_size;
1230 }
1231
1232 struct wim_lookup_table_entry **
1233 retrieve_lte_pointer(struct wim_lookup_table_entry *lte)
1234 {
1235         wimlib_assert(lte->unhashed);
1236         struct wim_inode *inode = lte->back_inode;
1237         u32 stream_id = lte->back_stream_id;
1238         if (stream_id == 0)
1239                 return &inode->i_lte;
1240         else
1241                 for (u16 i = 0; i < inode->i_num_ads; i++)
1242                         if (inode->i_ads_entries[i].stream_id == stream_id)
1243                                 return &inode->i_ads_entries[i].lte;
1244         wimlib_assert(0);
1245         return NULL;
1246 }
1247
1248 /* Calculate the SHA1 message digest of a stream and move it from the list of
1249  * unhashed streams to the stream lookup table, possibly joining it with an
1250  * existing lookup table entry for an identical stream.
1251  *
1252  * @lte:  An unhashed lookup table entry.
1253  * @lookup_table:  Lookup table for the WIM.
1254  * @lte_ret:  On success, write a pointer to the resulting lookup table
1255  *            entry to this location.  This will be the same as @lte
1256  *            if it was inserted into the lookup table, or different if
1257  *            a duplicate stream was found.
1258  *
1259  * Returns 0 on success; nonzero if there is an error reading the stream.
1260  */
1261 int
1262 hash_unhashed_stream(struct wim_lookup_table_entry *lte,
1263                      struct wim_lookup_table *lookup_table,
1264                      struct wim_lookup_table_entry **lte_ret)
1265 {
1266         int ret;
1267         struct wim_lookup_table_entry *duplicate_lte;
1268         struct wim_lookup_table_entry **back_ptr;
1269
1270         wimlib_assert(lte->unhashed);
1271
1272         /* back_ptr must be saved because @back_inode and @back_stream_id are in
1273          * union with the SHA1 message digest and will no longer be valid once
1274          * the SHA1 has been calculated. */
1275         back_ptr = retrieve_lte_pointer(lte);
1276
1277         ret = sha1_resource(lte);
1278         if (ret)
1279                 return ret;
1280
1281         /* Look for a duplicate stream */
1282         duplicate_lte = lookup_resource(lookup_table, lte->hash);
1283         list_del(&lte->unhashed_list);
1284         if (duplicate_lte) {
1285                 /* We have a duplicate stream.  Transfer the reference counts
1286                  * from this stream to the duplicate, update the reference to
1287                  * this stream (in an inode or ads_entry) to point to the
1288                  * duplicate, then free this stream. */
1289                 wimlib_assert(!(duplicate_lte->unhashed));
1290                 duplicate_lte->refcnt += lte->refcnt;
1291                 duplicate_lte->out_refcnt += lte->out_refcnt;
1292                 *back_ptr = duplicate_lte;
1293                 free_lookup_table_entry(lte);
1294                 lte = duplicate_lte;
1295         } else {
1296                 /* No duplicate stream, so we need to insert
1297                  * this stream into the lookup table and treat
1298                  * it as a hashed stream. */
1299                 lookup_table_insert(lookup_table, lte);
1300                 lte->unhashed = 0;
1301         }
1302         if (lte_ret)
1303                 *lte_ret = lte;
1304         return 0;
1305 }
1306
1307 static int
1308 lte_clone_if_new(struct wim_lookup_table_entry *lte, void *_lookup_table)
1309 {
1310         struct wim_lookup_table *lookup_table = _lookup_table;
1311
1312         if (lookup_resource(lookup_table, lte->hash))
1313                 return 0;  /*  Resource already present.  */
1314
1315         lte = clone_lookup_table_entry(lte);
1316         if (!lte)
1317                 return WIMLIB_ERR_NOMEM;
1318         lte->out_refcnt = 1;
1319         lookup_table_insert(lookup_table, lte);
1320         return 0;
1321 }
1322
1323 static int
1324 lte_delete_if_new(struct wim_lookup_table_entry *lte, void *_lookup_table)
1325 {
1326         struct wim_lookup_table *lookup_table = _lookup_table;
1327
1328         if (lte->out_refcnt) {
1329                 lookup_table_unlink(lookup_table, lte);
1330                 free_lookup_table_entry(lte);
1331         }
1332         return 0;
1333 }
1334
1335 /* API function documented in wimlib.h  */
1336 WIMLIBAPI int
1337 wimlib_reference_resources(WIMStruct *wim,
1338                            WIMStruct **resource_wims, unsigned num_resource_wims,
1339                            int ref_flags)
1340 {
1341         int ret;
1342         unsigned i;
1343
1344         if (wim == NULL)
1345                 return WIMLIB_ERR_INVALID_PARAM;
1346
1347         if (num_resource_wims != 0 && resource_wims == NULL)
1348                 return WIMLIB_ERR_INVALID_PARAM;
1349
1350         for (i = 0; i < num_resource_wims; i++)
1351                 if (resource_wims[i] == NULL)
1352                         return WIMLIB_ERR_INVALID_PARAM;
1353
1354         for_lookup_table_entry(wim->lookup_table, lte_zero_out_refcnt, NULL);
1355
1356         for (i = 0; i < num_resource_wims; i++) {
1357                 ret = for_lookup_table_entry(resource_wims[i]->lookup_table,
1358                                              lte_clone_if_new,
1359                                              wim->lookup_table);
1360                 if (ret)
1361                         goto out_rollback;
1362         }
1363         return 0;
1364
1365 out_rollback:
1366         for_lookup_table_entry(wim->lookup_table, lte_delete_if_new,
1367                                wim->lookup_table);
1368         return ret;
1369 }
1370
1371 static int
1372 reference_resource_paths(WIMStruct *wim,
1373                          const tchar * const *resource_wimfiles,
1374                          unsigned num_resource_wimfiles,
1375                          int ref_flags,
1376                          int open_flags,
1377                          wimlib_progress_func_t progress_func)
1378 {
1379         WIMStruct **resource_wims;
1380         unsigned i;
1381         int ret;
1382
1383         resource_wims = CALLOC(num_resource_wimfiles, sizeof(resource_wims[0]));
1384         if (!resource_wims)
1385                 return WIMLIB_ERR_NOMEM;
1386
1387         for (i = 0; i < num_resource_wimfiles; i++) {
1388                 DEBUG("Referencing resources from path \"%"TS"\"",
1389                       resource_wimfiles[i]);
1390                 ret = wimlib_open_wim(resource_wimfiles[i], open_flags,
1391                                       &resource_wims[i], progress_func);
1392                 if (ret)
1393                         goto out_free_resource_wims;
1394         }
1395
1396         ret = wimlib_reference_resources(wim, resource_wims,
1397                                          num_resource_wimfiles, ref_flags);
1398         if (ret)
1399                 goto out_free_resource_wims;
1400
1401         for (i = 0; i < num_resource_wimfiles; i++)
1402                 list_add_tail(&resource_wims[i]->subwim_node, &wim->subwims);
1403
1404         ret = 0;
1405         goto out_free_array;
1406
1407 out_free_resource_wims:
1408         for (i = 0; i < num_resource_wimfiles; i++)
1409                 wimlib_free(resource_wims[i]);
1410 out_free_array:
1411         FREE(resource_wims);
1412         return ret;
1413 }
1414
1415 static int
1416 reference_resource_glob(WIMStruct *wim, const tchar *refglob,
1417                         int ref_flags, int open_flags,
1418                         wimlib_progress_func_t progress_func)
1419 {
1420         glob_t globbuf;
1421         int ret;
1422
1423         /* Note: glob() is replaced in Windows native builds.  */
1424         ret = tglob(refglob, GLOB_ERR | GLOB_NOSORT, NULL, &globbuf);
1425         if (ret) {
1426                 if (ret == GLOB_NOMATCH) {
1427                         if (ref_flags & WIMLIB_REF_FLAG_GLOB_ERR_ON_NOMATCH) {
1428                                 ERROR("Found no files for glob \"%"TS"\"", refglob);
1429                                 return WIMLIB_ERR_GLOB_HAD_NO_MATCHES;
1430                         } else {
1431                                 return reference_resource_paths(wim,
1432                                                                 &refglob,
1433                                                                 1,
1434                                                                 ref_flags,
1435                                                                 open_flags,
1436                                                                 progress_func);
1437                         }
1438                 } else {
1439                         ERROR_WITH_ERRNO("Failed to process glob \"%"TS"\"", refglob);
1440                         if (ret == GLOB_NOSPACE)
1441                                 return WIMLIB_ERR_NOMEM;
1442                         else
1443                                 return WIMLIB_ERR_READ;
1444                 }
1445         }
1446
1447         ret = reference_resource_paths(wim,
1448                                        (const tchar * const *)globbuf.gl_pathv,
1449                                        globbuf.gl_pathc,
1450                                        ref_flags,
1451                                        open_flags,
1452                                        progress_func);
1453         globfree(&globbuf);
1454         return ret;
1455 }
1456
1457 /* API function documented in wimlib.h  */
1458 WIMLIBAPI int
1459 wimlib_reference_resource_files(WIMStruct *wim,
1460                                 const tchar * const * resource_wimfiles_or_globs,
1461                                 unsigned count,
1462                                 int ref_flags,
1463                                 int open_flags,
1464                                 wimlib_progress_func_t progress_func)
1465 {
1466         unsigned i;
1467         int ret;
1468
1469         if (ref_flags & WIMLIB_REF_FLAG_GLOB_ENABLE) {
1470                 for (i = 0; i < count; i++) {
1471                         ret = reference_resource_glob(wim,
1472                                                       resource_wimfiles_or_globs[i],
1473                                                       ref_flags,
1474                                                       open_flags,
1475                                                       progress_func);
1476                         if (ret)
1477                                 return ret;
1478                 }
1479                 return 0;
1480         } else {
1481                 return reference_resource_paths(wim, resource_wimfiles_or_globs,
1482                                                 count, ref_flags,
1483                                                 open_flags, progress_func);
1484         }
1485 }