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                                 /* Metadata entries with no references must be
 578                                  * ignored.  See for example the WinPE WIMs from
 579                                  * WAIK v2.1.  */
 580                                 if (cur_entry->refcnt == 0) {
 581                                         free_lookup_table_entry(cur_entry);
 582                                         continue;
 583                                 }
 584                                 if (wimlib_print_errors) {
 585                                         ERROR("Found metadata resource with refcnt != 1:");
 586                                         print_lookup_table_entry(cur_entry, stderr);
 587                                 }
 588                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 589                                 goto out_free_cur_entry;
 590                         }
 591
 592                         if (wim->hdr.part_number != 1) {
 593                                 WARNING("Ignoring metadata resource found in a "
 594                                         "non-first part of the split WIM");
 595                                 free_lookup_table_entry(cur_entry);
 596                                 continue;
 597                         }
 598                         if (wim->current_image == wim->hdr.image_count) {
 599                                 WARNING("The WIM header says there are %u images "
 600                                         "in the WIM, but we found more metadata "
 601                                         "resources than this (ignoring the extra)",
 602                                         wim->hdr.image_count);
 603                                 free_lookup_table_entry(cur_entry);
 604                                 continue;
 605                         }
 606
 607                         /* Notice very carefully:  We are assigning the metadata
 608                          * resources in the exact order mirrored by their lookup
 609                          * table entries on disk, which is the behavior of
 610                          * Microsoft's software.  In particular, this overrides
 611                          * the actual locations of the metadata resources
 612                          * themselves in the WIM file as well as any information
 613                          * written in the XML data. */
 614                         DEBUG("Found metadata resource for image %u at "
 615                               "offset %"PRIu64".",
 616                               wim->current_image + 1,
 617                               cur_entry->resource_entry.offset);
 618                         wim->image_metadata[
 619                                 wim->current_image++]->metadata_lte = cur_entry;
 620                 } else {
 621                         /* Lookup table entry for a stream that is not a
 622                          * metadata resource */
 623                         duplicate_entry = lookup_resource(table, cur_entry->hash);
 624                         if (duplicate_entry) {
 625                                 if (wimlib_print_errors) {
 626                                         WARNING("The WIM lookup table contains two entries with the "
 627                                               "same SHA1 message digest!");
 628                                         WARNING("The first entry is:");
 629                                         print_lookup_table_entry(duplicate_entry, stderr);
 630                                         WARNING("The second entry is:");
 631                                         print_lookup_table_entry(cur_entry, stderr);
 632                                 }
 633                                 free_lookup_table_entry(cur_entry);
 634                                 continue;
 635                         } else {
 636                                 lookup_table_insert(table, cur_entry);
 637                         }
 638                 }
 639         }
 640
 641         if (wim->hdr.part_number == 1 && wim->current_image != wim->hdr.image_count) {
 642                 WARNING("The header of \"%"TS"\" says there are %u images in\n"
 643                         "          the WIM, but we only found %d metadata resources!  Acting as if\n"
 644                         "          the header specified only %d images instead.",
 645                         wim->filename, wim->hdr.image_count,
 646                         wim->current_image, wim->current_image);
 647                 for (int i = wim->current_image; i < wim->hdr.image_count; i++)
 648                         put_image_metadata(wim->image_metadata[i], NULL);
 649                 wim->hdr.image_count = wim->current_image;
 650         }
 651         DEBUG("Done reading lookup table.");
 652         wim->lookup_table = table;
 653         ret = 0;
 654         goto out_free_buf;
 655 out_free_cur_entry:
 656         FREE(cur_entry);
 657 out_free_lookup_table:
 658         free_lookup_table(table);
 659 out_free_buf:
 660         FREE(buf);
 661 out:
 662         wim->current_image = 0;
 663         return ret;
 664 }
 665
 666
 667 static void
 668 write_wim_lookup_table_entry(const struct wim_lookup_table_entry *lte,
 669                              struct wim_lookup_table_entry_disk *disk_entry)
 670 {
 671         put_resource_entry(&lte->output_resource_entry, &disk_entry->resource_entry);
 672         disk_entry->part_number = cpu_to_le16(lte->part_number);
 673         disk_entry->refcnt = cpu_to_le32(lte->out_refcnt);
 674         copy_hash(disk_entry->hash, lte->hash);
 675 }
 676
 677 static int
 678 write_wim_lookup_table_from_stream_list(struct list_head *stream_list,
 679                                         struct filedes *out_fd,
 680                                         struct resource_entry *out_res_entry,
 681                                         int write_resource_flags,
 682                                         struct wimlib_lzx_context **comp_ctx)
 683 {
 684         size_t table_size;
 685         struct wim_lookup_table_entry *lte;
 686         struct wim_lookup_table_entry_disk *table_buf;
 687         struct wim_lookup_table_entry_disk *table_buf_ptr;
 688         int ret;
 689
 690         table_size = 0;
 691         list_for_each_entry(lte, stream_list, lookup_table_list)
 692                 table_size += sizeof(struct wim_lookup_table_entry_disk);
 693
 694         DEBUG("Writing WIM lookup table (size=%zu, offset=%"PRIu64")",
 695               table_size, out_fd->offset);
 696
 697         table_buf = MALLOC(table_size);
 698         if (!table_buf) {
 699                 ERROR("Failed to allocate %zu bytes for temporary lookup table",
 700                       table_size);
 701                 return WIMLIB_ERR_NOMEM;
 702         }
 703         table_buf_ptr = table_buf;
 704         list_for_each_entry(lte, stream_list, lookup_table_list)
 705                 write_wim_lookup_table_entry(lte, table_buf_ptr++);
 706
 707         /* Write the lookup table uncompressed.  Although wimlib can handle a
 708          * compressed lookup table, MS software cannot.  */
 709         ret = write_wim_resource_from_buffer(table_buf,
 710                                              table_size,
 711                                              WIM_RESHDR_FLAG_METADATA,
 712                                              out_fd,
 713                                              WIMLIB_COMPRESSION_TYPE_NONE,
 714                                              out_res_entry,
 715                                              NULL,
 716                                              write_resource_flags,
 717                                              comp_ctx);
 718         FREE(table_buf);
 719         return ret;
 720 }
 721
 722 static int
 723 append_lookup_table_entry(struct wim_lookup_table_entry *lte, void *_list)
 724 {
 725         /* Lookup table entries with 'out_refcnt' == 0 correspond to streams not
 726          * written and not present in the resulting WIM file, and should not be
 727          * included in the lookup table.
 728          *
 729          * Lookup table entries marked as filtered (EXTERNAL_WIM) with
 730          * 'out_refcnt != 0' were referenced as part of the logical write but
 731          * correspond to streams that were not in fact written, and should not
 732          * be included in the lookup table.
 733          *
 734          * Lookup table entries marked as filtered (SAME_WIM) with 'out_refcnt
 735          * != 0' were referenced as part of the logical write but correspond to
 736          * streams that were not in fact written, but nevertheless were already
 737          * present in the WIM being overwritten in-place.  These entries must be
 738          * included in the lookup table, and the resource information to write
 739          * needs to be copied from the resource information read originally.
 740          */
 741         if (lte->out_refcnt != 0 && !(lte->filtered & FILTERED_EXTERNAL_WIM)) {
 742                 if (lte->filtered & FILTERED_SAME_WIM) {
 743                         copy_resource_entry(&lte->output_resource_entry,
 744                                             &lte->resource_entry);
 745                 }
 746                 list_add_tail(&lte->lookup_table_list, (struct list_head*)_list);
 747         }
 748         return 0;
 749 }
 750
 751 int
 752 write_wim_lookup_table(WIMStruct *wim, int image, int write_flags,
 753                        struct resource_entry *out_res_entry,
 754                        struct list_head *stream_list_override)
 755 {
 756         int write_resource_flags;
 757         struct list_head _stream_list;
 758         struct list_head *stream_list;
 759
 760         if (stream_list_override) {
 761                 stream_list = stream_list_override;
 762         } else {
 763                 stream_list = &_stream_list;
 764                 INIT_LIST_HEAD(stream_list);
 765         }
 766
 767         if (!(write_flags & WIMLIB_WRITE_FLAG_NO_METADATA)) {
 768                 int start_image;
 769                 int end_image;
 770
 771                 if (image == WIMLIB_ALL_IMAGES) {
 772                         start_image = 1;
 773                         end_image = wim->hdr.image_count;
 774                 } else {
 775                         start_image = image;
 776                         end_image = image;
 777                 }
 778
 779                 /* Push metadata resource lookup table entries onto the front of
 780                  * the list in reverse order, so that they're written in order.
 781                  */
 782                 for (int i = end_image; i >= start_image; i--) {
 783                         struct wim_lookup_table_entry *metadata_lte;
 784
 785                         metadata_lte = wim->image_metadata[i - 1]->metadata_lte;
 786                         metadata_lte->out_refcnt = 1;
 787                         metadata_lte->part_number = wim->hdr.part_number;
 788                         metadata_lte->output_resource_entry.flags |= WIM_RESHDR_FLAG_METADATA;
 789
 790                         list_add(&metadata_lte->lookup_table_list, stream_list);
 791                 }
 792         }
 793
 794         /* Append additional lookup table entries that need to be written, with
 795          * some special handling for streams that have been marked as filtered.
 796          */
 797         if (!stream_list_override) {
 798                 for_lookup_table_entry(wim->lookup_table,
 799                                        append_lookup_table_entry, stream_list);
 800         }
 801
 802         write_resource_flags = 0;
 803         if (write_flags & WIMLIB_WRITE_FLAG_PIPABLE)
 804                 write_resource_flags |= WIMLIB_WRITE_RESOURCE_FLAG_PIPABLE;
 805         return write_wim_lookup_table_from_stream_list(stream_list,
 806                                                        &wim->out_fd,
 807                                                        out_res_entry,
 808                                                        write_resource_flags,
 809                                                        &wim->lzx_context);
 810 }
 811
 812
 813 int
 814 lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 815 {
 816         lte->real_refcnt = 0;
 817         return 0;
 818 }
 819
 820 int
 821 lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 822 {
 823         lte->out_refcnt = 0;
 824         return 0;
 825 }
 826
 827 int
 828 lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *_ignore)
 829 {
 830         if (lte->extracted_file != NULL) {
 831                 FREE(lte->extracted_file);
 832                 lte->extracted_file = NULL;
 833         }
 834         return 0;
 835 }
 836
 837 void
 838 print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 839 {
 840         if (!lte) {
 841                 tputc(T('\n'), out);
 842                 return;
 843         }
 844         tfprintf(out, T("Offset            = %"PRIu64" bytes\n"),
 845                  lte->resource_entry.offset);
 846
 847         tfprintf(out, T("Size              = %"PRIu64" bytes\n"),
 848                  (u64)lte->resource_entry.size);
 849
 850         tfprintf(out, T("Original size     = %"PRIu64" bytes\n"),
 851                  lte->resource_entry.original_size);
 852
 853         tfprintf(out, T("Part Number       = %hu\n"), lte->part_number);
 854         tfprintf(out, T("Reference Count   = %u\n"), lte->refcnt);
 855
 856         if (lte->unhashed) {
 857                 tfprintf(out, T("(Unhashed: inode %p, stream_id = %u)\n"),
 858                          lte->back_inode, lte->back_stream_id);
 859         } else {
 860                 tfprintf(out, T("Hash              = 0x"));
 861                 print_hash(lte->hash, out);
 862                 tputc(T('\n'), out);
 863         }
 864
 865         tfprintf(out, T("Flags             = "));
 866         u8 flags = lte->resource_entry.flags;
 867         if (flags & WIM_RESHDR_FLAG_COMPRESSED)
 868                 tfputs(T("WIM_RESHDR_FLAG_COMPRESSED, "), out);
 869         if (flags & WIM_RESHDR_FLAG_FREE)
 870                 tfputs(T("WIM_RESHDR_FLAG_FREE, "), out);
 871         if (flags & WIM_RESHDR_FLAG_METADATA)
 872                 tfputs(T("WIM_RESHDR_FLAG_METADATA, "), out);
 873         if (flags & WIM_RESHDR_FLAG_SPANNED)
 874                 tfputs(T("WIM_RESHDR_FLAG_SPANNED, "), out);
 875         tputc(T('\n'), out);
 876         switch (lte->resource_location) {
 877         case RESOURCE_IN_WIM:
 878                 if (lte->wim->filename) {
 879                         tfprintf(out, T("WIM file          = `%"TS"'\n"),
 880                                  lte->wim->filename);
 881                 }
 882                 break;
 883 #ifdef __WIN32__
 884         case RESOURCE_WIN32_ENCRYPTED:
 885 #endif
 886         case RESOURCE_IN_FILE_ON_DISK:
 887                 tfprintf(out, T("File on Disk      = `%"TS"'\n"),
 888                          lte->file_on_disk);
 889                 break;
 890 #ifdef WITH_FUSE
 891         case RESOURCE_IN_STAGING_FILE:
 892                 tfprintf(out, T("Staging File      = `%"TS"'\n"),
 893                                 lte->staging_file_name);
 894                 break;
 895 #endif
 896         default:
 897                 break;
 898         }
 899         tputc(T('\n'), out);
 900 }
 901
 902 void
 903 lte_to_wimlib_resource_entry(const struct wim_lookup_table_entry *lte,
 904                              struct wimlib_resource_entry *wentry)
 905 {
 906         wentry->uncompressed_size = lte->resource_entry.original_size;
 907         wentry->compressed_size = lte->resource_entry.size;
 908         wentry->offset = lte->resource_entry.offset;
 909         copy_hash(wentry->sha1_hash, lte->hash);
 910         wentry->part_number = lte->part_number;
 911         wentry->reference_count = lte->refcnt;
 912         wentry->is_compressed = (lte->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED) != 0;
 913         wentry->is_metadata = (lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA) != 0;
 914         wentry->is_free = (lte->resource_entry.flags & WIM_RESHDR_FLAG_FREE) != 0;
 915         wentry->is_spanned = (lte->resource_entry.flags & WIM_RESHDR_FLAG_SPANNED) != 0;
 916 }
 917
 918 struct iterate_lte_context {
 919         wimlib_iterate_lookup_table_callback_t cb;
 920         void *user_ctx;
 921 };
 922
 923 static int
 924 do_iterate_lte(struct wim_lookup_table_entry *lte, void *_ctx)
 925 {
 926         struct iterate_lte_context *ctx = _ctx;
 927         struct wimlib_resource_entry entry;
 928
 929         lte_to_wimlib_resource_entry(lte, &entry);
 930         return (*ctx->cb)(&entry, ctx->user_ctx);
 931 }
 932
 933 /* API function documented in wimlib.h  */
 934 WIMLIBAPI int
 935 wimlib_iterate_lookup_table(WIMStruct *wim, int flags,
 936                             wimlib_iterate_lookup_table_callback_t cb,
 937                             void *user_ctx)
 938 {
 939         struct iterate_lte_context ctx = {
 940                 .cb = cb,
 941                 .user_ctx = user_ctx,
 942         };
 943         if (wim->hdr.part_number == 1) {
 944                 int ret;
 945                 for (int i = 0; i < wim->hdr.image_count; i++) {
 946                         ret = do_iterate_lte(wim->image_metadata[i]->metadata_lte,
 947                                              &ctx);
 948                         if (ret)
 949                                 return ret;
 950                 }
 951         }
 952         return for_lookup_table_entry(wim->lookup_table, do_iterate_lte, &ctx);
 953 }
 954
 955 /* Given a SHA1 message digest, return the corresponding entry in the WIM's
 956  * lookup table, or NULL if there is none.  */
 957 struct wim_lookup_table_entry *
 958 lookup_resource(const struct wim_lookup_table *table, const u8 hash[])
 959 {
 960         size_t i;
 961         struct wim_lookup_table_entry *lte;
 962         struct hlist_node *pos;
 963
 964         wimlib_assert(table != NULL);
 965         wimlib_assert(hash != NULL);
 966
 967         i = *(size_t*)hash % table->capacity;
 968         hlist_for_each_entry(lte, pos, &table->array[i], hash_list)
 969                 if (hashes_equal(hash, lte->hash))
 970                         return lte;
 971         return NULL;
 972 }
 973
 974 #ifdef WITH_FUSE
 975 /*
 976  * Finds the dentry, lookup table entry, and stream index for a WIM file stream,
 977  * given a path name.
 978  *
 979  * This is only for pre-resolved inodes.
 980  */
 981 int
 982 wim_pathname_to_stream(WIMStruct *wim,
 983                        const tchar *path,
 984                        int lookup_flags,
 985                        struct wim_dentry **dentry_ret,
 986                        struct wim_lookup_table_entry **lte_ret,
 987                        u16 *stream_idx_ret)
 988 {
 989         struct wim_dentry *dentry;
 990         struct wim_lookup_table_entry *lte;
 991         u16 stream_idx;
 992         const tchar *stream_name = NULL;
 993         struct wim_inode *inode;
 994         tchar *p = NULL;
 995
 996         if (lookup_flags & LOOKUP_FLAG_ADS_OK) {
 997                 stream_name = path_stream_name(path);
 998                 if (stream_name) {
 999                         p = (tchar*)stream_name - 1;
1000                         *p = T('\0');
1001                 }
1002         }
1003
1004         dentry = get_dentry(wim, path);
1005         if (p)
1006                 *p = T(':');
1007         if (!dentry)
1008                 return -errno;
1009
1010         inode = dentry->d_inode;
1011
1012         if (!inode->i_resolved)
1013                 if (inode_resolve_ltes(inode, wim->lookup_table, false))
1014                         return -EIO;
1015
1016         if (!(lookup_flags & LOOKUP_FLAG_DIRECTORY_OK)
1017               && inode_is_directory(inode))
1018                 return -EISDIR;
1019
1020         if (stream_name) {
1021                 struct wim_ads_entry *ads_entry;
1022                 u16 ads_idx;
1023                 ads_entry = inode_get_ads_entry(inode, stream_name,
1024                                                 &ads_idx);
1025                 if (ads_entry) {
1026                         stream_idx = ads_idx + 1;
1027                         lte = ads_entry->lte;
1028                         goto out;
1029                 } else {
1030                         return -ENOENT;
1031                 }
1032         } else {
1033                 lte = inode_unnamed_stream_resolved(inode, &stream_idx);
1034         }
1035 out:
1036         if (dentry_ret)
1037                 *dentry_ret = dentry;
1038         if (lte_ret)
1039                 *lte_ret = lte;
1040         if (stream_idx_ret)
1041                 *stream_idx_ret = stream_idx;
1042         return 0;
1043 }
1044 #endif
1045
1046 int
1047 resource_not_found_error(const struct wim_inode *inode, const u8 *hash)
1048 {
1049         if (wimlib_print_errors) {
1050                 ERROR("\"%"TS"\": resource not found", inode_first_full_path(inode));
1051                 tfprintf(stderr, T("        SHA-1 message digest of missing resource:\n        "));
1052                 print_hash(hash, stderr);
1053                 tputc(T('\n'), stderr);
1054         }
1055         return WIMLIB_ERR_RESOURCE_NOT_FOUND;
1056 }
1057
1058 /*
1059  * Resolve an inode's lookup table entries.
1060  *
1061  * This replaces the SHA1 hash fields (which are used to lookup an entry in the
1062  * lookup table) with pointers directly to the lookup table entries.
1063  *
1064  * If @force is %false:
1065  *      If any needed SHA1 message digests are not found in the lookup table,
1066  *      WIMLIB_ERR_RESOURCE_NOT_FOUND is returned and the inode is left
1067  *      unmodified.
1068  * If @force is %true:
1069  *      If any needed SHA1 message digests are not found in the lookup table,
1070  *      new entries are allocated and inserted into the lookup table.
1071  */
1072 int
1073 inode_resolve_ltes(struct wim_inode *inode, struct wim_lookup_table *table,
1074                    bool force)
1075 {
1076         const u8 *hash;
1077
1078         if (!inode->i_resolved) {
1079                 struct wim_lookup_table_entry *lte, *ads_lte;
1080
1081                 /* Resolve the default file stream */
1082                 lte = NULL;
1083                 hash = inode->i_hash;
1084                 if (!is_zero_hash(hash)) {
1085                         lte = lookup_resource(table, hash);
1086                         if (!lte) {
1087                                 if (force) {
1088                                         lte = new_lookup_table_entry();
1089                                         if (!lte)
1090                                                 return WIMLIB_ERR_NOMEM;
1091                                         copy_hash(lte->hash, hash);
1092                                         lookup_table_insert(table, lte);
1093                                 } else {
1094                                         goto resource_not_found;
1095                                 }
1096                         }
1097                 }
1098
1099                 /* Resolve the alternate data streams */
1100                 struct wim_lookup_table_entry *ads_ltes[inode->i_num_ads];
1101                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1102                         struct wim_ads_entry *cur_entry;
1103
1104                         ads_lte = NULL;
1105                         cur_entry = &inode->i_ads_entries[i];
1106                         hash = cur_entry->hash;
1107                         if (!is_zero_hash(hash)) {
1108                                 ads_lte = lookup_resource(table, hash);
1109                                 if (!ads_lte) {
1110                                         if (force) {
1111                                                 ads_lte = new_lookup_table_entry();
1112                                                 if (!ads_lte)
1113                                                         return WIMLIB_ERR_NOMEM;
1114                                                 copy_hash(ads_lte->hash, hash);
1115                                                 lookup_table_insert(table, ads_lte);
1116                                         } else {
1117                                                 goto resource_not_found;
1118                                         }
1119                                 }
1120                         }
1121                         ads_ltes[i] = ads_lte;
1122                 }
1123                 inode->i_lte = lte;
1124                 for (u16 i = 0; i < inode->i_num_ads; i++)
1125                         inode->i_ads_entries[i].lte = ads_ltes[i];
1126                 inode->i_resolved = 1;
1127         }
1128         return 0;
1129
1130 resource_not_found:
1131         return resource_not_found_error(inode, hash);
1132 }
1133
1134 void
1135 inode_unresolve_ltes(struct wim_inode *inode)
1136 {
1137         if (inode->i_resolved) {
1138                 if (inode->i_lte)
1139                         copy_hash(inode->i_hash, inode->i_lte->hash);
1140                 else
1141                         zero_out_hash(inode->i_hash);
1142
1143                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1144                         if (inode->i_ads_entries[i].lte)
1145                                 copy_hash(inode->i_ads_entries[i].hash,
1146                                           inode->i_ads_entries[i].lte->hash);
1147                         else
1148                                 zero_out_hash(inode->i_ads_entries[i].hash);
1149                 }
1150                 inode->i_resolved = 0;
1151         }
1152 }
1153
1154 /*
1155  * Returns the lookup table entry for stream @stream_idx of the inode, where
1156  * stream_idx = 0 means the default un-named file stream, and stream_idx >= 1
1157  * corresponds to an alternate data stream.
1158  *
1159  * This works for both resolved and un-resolved inodes.
1160  */
1161 struct wim_lookup_table_entry *
1162 inode_stream_lte(const struct wim_inode *inode, unsigned stream_idx,
1163                  const struct wim_lookup_table *table)
1164 {
1165         if (inode->i_resolved)
1166                 return inode_stream_lte_resolved(inode, stream_idx);
1167         else
1168                 return inode_stream_lte_unresolved(inode, stream_idx, table);
1169 }
1170
1171 struct wim_lookup_table_entry *
1172 inode_unnamed_stream_resolved(const struct wim_inode *inode, u16 *stream_idx_ret)
1173 {
1174         wimlib_assert(inode->i_resolved);
1175         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
1176                 if (inode_stream_name_nbytes(inode, i) == 0 &&
1177                     !is_zero_hash(inode_stream_hash_resolved(inode, i)))
1178                 {
1179                         *stream_idx_ret = i;
1180                         return inode_stream_lte_resolved(inode, i);
1181                 }
1182         }
1183         *stream_idx_ret = 0;
1184         return NULL;
1185 }
1186
1187 struct wim_lookup_table_entry *
1188 inode_unnamed_lte_resolved(const struct wim_inode *inode)
1189 {
1190         u16 stream_idx;
1191         return inode_unnamed_stream_resolved(inode, &stream_idx);
1192 }
1193
1194 struct wim_lookup_table_entry *
1195 inode_unnamed_lte_unresolved(const struct wim_inode *inode,
1196                              const struct wim_lookup_table *table)
1197 {
1198         wimlib_assert(!inode->i_resolved);
1199         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
1200                 if (inode_stream_name_nbytes(inode, i) == 0 &&
1201                     !is_zero_hash(inode_stream_hash_unresolved(inode, i)))
1202                 {
1203                         return inode_stream_lte_unresolved(inode, i, table);
1204                 }
1205         }
1206         return NULL;
1207 }
1208
1209 /* Return the lookup table entry for the unnamed data stream of an inode, or
1210  * NULL if there is none.
1211  *
1212  * You'd think this would be easier than it actually is, since the unnamed data
1213  * stream should be the one referenced from the inode itself.  Alas, if there
1214  * are named data streams, Microsoft's "imagex.exe" program will put the unnamed
1215  * data stream in one of the alternate data streams instead of inside the WIM
1216  * dentry itself.  So we need to check the alternate data streams too.
1217  *
1218  * Also, note that a dentry may appear to have more than one unnamed stream, but
1219  * if the SHA1 message digest is all 0's then the corresponding stream does not
1220  * really "count" (this is the case for the inode's own file stream when the
1221  * file stream that should be there is actually in one of the alternate stream
1222  * entries.).  This is despite the fact that we may need to extract such a
1223  * missing entry as an empty file or empty named data stream.
1224  */
1225 struct wim_lookup_table_entry *
1226 inode_unnamed_lte(const struct wim_inode *inode,
1227                   const struct wim_lookup_table *table)
1228 {
1229         if (inode->i_resolved)
1230                 return inode_unnamed_lte_resolved(inode);
1231         else
1232                 return inode_unnamed_lte_unresolved(inode, table);
1233 }
1234
1235 /* Returns the SHA1 message digest of the unnamed data stream of a WIM inode, or
1236  * 'zero_hash' if the unnamed data stream is missing has all zeroes in its SHA1
1237  * message digest field.  */
1238 const u8 *
1239 inode_unnamed_stream_hash(const struct wim_inode *inode)
1240 {
1241         const u8 *hash;
1242
1243         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
1244                 if (inode_stream_name_nbytes(inode, i) == 0) {
1245                         hash = inode_stream_hash(inode, i);
1246                         if (!is_zero_hash(hash))
1247                                 return hash;
1248                 }
1249         }
1250         return zero_hash;
1251 }
1252
1253
1254 static int
1255 lte_add_stream_size(struct wim_lookup_table_entry *lte, void *total_bytes_p)
1256 {
1257         *(u64*)total_bytes_p += lte->resource_entry.size;
1258         return 0;
1259 }
1260
1261 u64
1262 lookup_table_total_stream_size(struct wim_lookup_table *table)
1263 {
1264         u64 total_size = 0;
1265         for_lookup_table_entry(table, lte_add_stream_size, &total_size);
1266         return total_size;
1267 }
1268
1269 struct wim_lookup_table_entry **
1270 retrieve_lte_pointer(struct wim_lookup_table_entry *lte)
1271 {
1272         wimlib_assert(lte->unhashed);
1273         struct wim_inode *inode = lte->back_inode;
1274         u32 stream_id = lte->back_stream_id;
1275         if (stream_id == 0)
1276                 return &inode->i_lte;
1277         else
1278                 for (u16 i = 0; i < inode->i_num_ads; i++)
1279                         if (inode->i_ads_entries[i].stream_id == stream_id)
1280                                 return &inode->i_ads_entries[i].lte;
1281         wimlib_assert(0);
1282         return NULL;
1283 }
1284
1285 /* Calculate the SHA1 message digest of a stream and move it from the list of
1286  * unhashed streams to the stream lookup table, possibly joining it with an
1287  * existing lookup table entry for an identical stream.
1288  *
1289  * @lte:  An unhashed lookup table entry.
1290  * @lookup_table:  Lookup table for the WIM.
1291  * @lte_ret:  On success, write a pointer to the resulting lookup table
1292  *            entry to this location.  This will be the same as @lte
1293  *            if it was inserted into the lookup table, or different if
1294  *            a duplicate stream was found.
1295  *
1296  * Returns 0 on success; nonzero if there is an error reading the stream.
1297  */
1298 int
1299 hash_unhashed_stream(struct wim_lookup_table_entry *lte,
1300                      struct wim_lookup_table *lookup_table,
1301                      struct wim_lookup_table_entry **lte_ret)
1302 {
1303         int ret;
1304         struct wim_lookup_table_entry *duplicate_lte;
1305         struct wim_lookup_table_entry **back_ptr;
1306
1307         wimlib_assert(lte->unhashed);
1308
1309         /* back_ptr must be saved because @back_inode and @back_stream_id are in
1310          * union with the SHA1 message digest and will no longer be valid once
1311          * the SHA1 has been calculated. */
1312         back_ptr = retrieve_lte_pointer(lte);
1313
1314         ret = sha1_resource(lte);
1315         if (ret)
1316                 return ret;
1317
1318         /* Look for a duplicate stream */
1319         duplicate_lte = lookup_resource(lookup_table, lte->hash);
1320         list_del(&lte->unhashed_list);
1321         if (duplicate_lte) {
1322                 /* We have a duplicate stream.  Transfer the reference counts
1323                  * from this stream to the duplicate, update the reference to
1324                  * this stream (in an inode or ads_entry) to point to the
1325                  * duplicate, then free this stream. */
1326                 wimlib_assert(!(duplicate_lte->unhashed));
1327                 duplicate_lte->refcnt += lte->refcnt;
1328                 duplicate_lte->out_refcnt += lte->out_refcnt;
1329                 *back_ptr = duplicate_lte;
1330                 free_lookup_table_entry(lte);
1331                 lte = duplicate_lte;
1332         } else {
1333                 /* No duplicate stream, so we need to insert
1334                  * this stream into the lookup table and treat
1335                  * it as a hashed stream. */
1336                 lookup_table_insert(lookup_table, lte);
1337                 lte->unhashed = 0;
1338         }
1339         if (lte_ret)
1340                 *lte_ret = lte;
1341         return 0;
1342 }
1343
1344 static int
1345 lte_clone_if_new(struct wim_lookup_table_entry *lte, void *_lookup_table)
1346 {
1347         struct wim_lookup_table *lookup_table = _lookup_table;
1348
1349         if (lookup_resource(lookup_table, lte->hash))
1350                 return 0;  /*  Resource already present.  */
1351
1352         lte = clone_lookup_table_entry(lte);
1353         if (!lte)
1354                 return WIMLIB_ERR_NOMEM;
1355         lte->out_refcnt = 1;
1356         lookup_table_insert(lookup_table, lte);
1357         return 0;
1358 }
1359
1360 static int
1361 lte_delete_if_new(struct wim_lookup_table_entry *lte, void *_lookup_table)
1362 {
1363         struct wim_lookup_table *lookup_table = _lookup_table;
1364
1365         if (lte->out_refcnt) {
1366                 lookup_table_unlink(lookup_table, lte);
1367                 free_lookup_table_entry(lte);
1368         }
1369         return 0;
1370 }
1371
1372 /* API function documented in wimlib.h  */
1373 WIMLIBAPI int
1374 wimlib_reference_resources(WIMStruct *wim,
1375                            WIMStruct **resource_wims, unsigned num_resource_wims,
1376                            int ref_flags)
1377 {
1378         int ret;
1379         unsigned i;
1380
1381         if (wim == NULL)
1382                 return WIMLIB_ERR_INVALID_PARAM;
1383
1384         if (num_resource_wims != 0 && resource_wims == NULL)
1385                 return WIMLIB_ERR_INVALID_PARAM;
1386
1387         for (i = 0; i < num_resource_wims; i++)
1388                 if (resource_wims[i] == NULL)
1389                         return WIMLIB_ERR_INVALID_PARAM;
1390
1391         for_lookup_table_entry(wim->lookup_table, lte_zero_out_refcnt, NULL);
1392
1393         for (i = 0; i < num_resource_wims; i++) {
1394                 ret = for_lookup_table_entry(resource_wims[i]->lookup_table,
1395                                              lte_clone_if_new,
1396                                              wim->lookup_table);
1397                 if (ret)
1398                         goto out_rollback;
1399         }
1400         return 0;
1401
1402 out_rollback:
1403         for_lookup_table_entry(wim->lookup_table, lte_delete_if_new,
1404                                wim->lookup_table);
1405         return ret;
1406 }
1407
1408 static int
1409 reference_resource_paths(WIMStruct *wim,
1410                          const tchar * const *resource_wimfiles,
1411                          unsigned num_resource_wimfiles,
1412                          int ref_flags,
1413                          int open_flags,
1414                          wimlib_progress_func_t progress_func)
1415 {
1416         WIMStruct **resource_wims;
1417         unsigned i;
1418         int ret;
1419
1420         resource_wims = CALLOC(num_resource_wimfiles, sizeof(resource_wims[0]));
1421         if (!resource_wims)
1422                 return WIMLIB_ERR_NOMEM;
1423
1424         for (i = 0; i < num_resource_wimfiles; i++) {
1425                 DEBUG("Referencing resources from path \"%"TS"\"",
1426                       resource_wimfiles[i]);
1427                 ret = wimlib_open_wim(resource_wimfiles[i], open_flags,
1428                                       &resource_wims[i], progress_func);
1429                 if (ret)
1430                         goto out_free_resource_wims;
1431         }
1432
1433         ret = wimlib_reference_resources(wim, resource_wims,
1434                                          num_resource_wimfiles, ref_flags);
1435         if (ret)
1436                 goto out_free_resource_wims;
1437
1438         for (i = 0; i < num_resource_wimfiles; i++)
1439                 list_add_tail(&resource_wims[i]->subwim_node, &wim->subwims);
1440
1441         ret = 0;
1442         goto out_free_array;
1443
1444 out_free_resource_wims:
1445         for (i = 0; i < num_resource_wimfiles; i++)
1446                 wimlib_free(resource_wims[i]);
1447 out_free_array:
1448         FREE(resource_wims);
1449         return ret;
1450 }
1451
1452 static int
1453 reference_resource_glob(WIMStruct *wim, const tchar *refglob,
1454                         int ref_flags, int open_flags,
1455                         wimlib_progress_func_t progress_func)
1456 {
1457         glob_t globbuf;
1458         int ret;
1459
1460         /* Note: glob() is replaced in Windows native builds.  */
1461         ret = tglob(refglob, GLOB_ERR | GLOB_NOSORT, NULL, &globbuf);
1462         if (ret) {
1463                 if (ret == GLOB_NOMATCH) {
1464                         if (ref_flags & WIMLIB_REF_FLAG_GLOB_ERR_ON_NOMATCH) {
1465                                 ERROR("Found no files for glob \"%"TS"\"", refglob);
1466                                 return WIMLIB_ERR_GLOB_HAD_NO_MATCHES;
1467                         } else {
1468                                 return reference_resource_paths(wim,
1469                                                                 &refglob,
1470                                                                 1,
1471                                                                 ref_flags,
1472                                                                 open_flags,
1473                                                                 progress_func);
1474                         }
1475                 } else {
1476                         ERROR_WITH_ERRNO("Failed to process glob \"%"TS"\"", refglob);
1477                         if (ret == GLOB_NOSPACE)
1478                                 return WIMLIB_ERR_NOMEM;
1479                         else
1480                                 return WIMLIB_ERR_READ;
1481                 }
1482         }
1483
1484         ret = reference_resource_paths(wim,
1485                                        (const tchar * const *)globbuf.gl_pathv,
1486                                        globbuf.gl_pathc,
1487                                        ref_flags,
1488                                        open_flags,
1489                                        progress_func);
1490         globfree(&globbuf);
1491         return ret;
1492 }
1493
1494 /* API function documented in wimlib.h  */
1495 WIMLIBAPI int
1496 wimlib_reference_resource_files(WIMStruct *wim,
1497                                 const tchar * const * resource_wimfiles_or_globs,
1498                                 unsigned count,
1499                                 int ref_flags,
1500                                 int open_flags,
1501                                 wimlib_progress_func_t progress_func)
1502 {
1503         unsigned i;
1504         int ret;
1505
1506         if (ref_flags & WIMLIB_REF_FLAG_GLOB_ENABLE) {
1507                 for (i = 0; i < count; i++) {
1508                         ret = reference_resource_glob(wim,
1509                                                       resource_wimfiles_or_globs[i],
1510                                                       ref_flags,
1511                                                       open_flags,
1512                                                       progress_func);
1513                         if (ret)
1514                                 return ret;
1515                 }
1516                 return 0;
1517         } else {
1518                 return reference_resource_paths(wim, resource_wimfiles_or_globs,
1519                                                 count, ref_flags,
1520                                                 open_flags, progress_func);
1521         }
1522 }