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

   1 /*
   2  * lookup_table.c
   3  *
   4  * Lookup table, implemented as a hash table, that maps SHA1 message digests to
   5  * data streams; plus code to read and write the corresponding on-disk data.
   6  */
   7
   8 /*
   9  * Copyright (C) 2012, 2013 Eric Biggers
  10  *
  11  * This file is part of wimlib, a library for working with WIM files.
  12  *
  13  * wimlib is free software; you can redistribute it and/or modify it under the
  14  * terms of the GNU General Public License as published by the Free
  15  * Software Foundation; either version 3 of the License, or (at your option)
  16  * any later version.
  17  *
  18  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  19  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  20  * A PARTICULAR PURPOSE. See the GNU General Public License for more
  21  * details.
  22  *
  23  * You should have received a copy of the GNU General Public License
  24  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  25  */
  26
  27 #ifdef HAVE_CONFIG_H
  28 #  include "config.h"
  29 #endif
  30
  31 #include "wimlib/endianness.h"
  32 #include "wimlib/error.h"
  33 #include "wimlib/file_io.h"
  34 #include "wimlib/glob.h"
  35 #include "wimlib/lookup_table.h"
  36 #include "wimlib/metadata.h"
  37 #include "wimlib/paths.h"
  38 #include "wimlib/resource.h"
  39 #include "wimlib/util.h"
  40 #include "wimlib/write.h"
  41
  42 #include <errno.h>
  43 #include <stdlib.h>
  44 #ifdef WITH_FUSE
  45 #  include <unistd.h> /* for unlink() */
  46 #endif
  47
  48 struct wim_lookup_table *
  49 new_lookup_table(size_t capacity)
  50 {
  51         struct wim_lookup_table *table;
  52         struct hlist_head *array;
  53
  54         table = CALLOC(1, sizeof(struct wim_lookup_table));
  55         if (table) {
  56                 array = CALLOC(capacity, sizeof(array[0]));
  57                 if (array) {
  58                         table->num_entries = 0;
  59                         table->capacity = capacity;
  60                         table->array = array;
  61                 } else {
  62                         FREE(table);
  63                         table = NULL;
  64                         ERROR("Failed to allocate memory for lookup table "
  65                               "with capacity %zu", capacity);
  66                 }
  67         }
  68         return table;
  69 }
  70
  71 struct wim_lookup_table_entry *
  72 new_lookup_table_entry(void)
  73 {
  74         struct wim_lookup_table_entry *lte;
  75
  76         lte = CALLOC(1, sizeof(struct wim_lookup_table_entry));
  77         if (lte == NULL) {
  78                 ERROR("Out of memory (tried to allocate %zu bytes for "
  79                       "lookup table entry)",
  80                       sizeof(struct wim_lookup_table_entry));
  81                 return NULL;
  82         }
  83         lte->refcnt = 1;
  84         BUILD_BUG_ON(RESOURCE_NONEXISTENT != 0);
  85         BUILD_BUG_ON(WIMLIB_COMPRESSION_TYPE_NONE != 0);
  86         return lte;
  87 }
  88
  89 struct wim_lookup_table_entry *
  90 clone_lookup_table_entry(const struct wim_lookup_table_entry *old)
  91 {
  92         struct wim_lookup_table_entry *new;
  93
  94         new = memdup(old, sizeof(struct wim_lookup_table_entry));
  95         if (!new)
  96                 return NULL;
  97
  98         new->extracted_file = NULL;
  99         switch (new->resource_location) {
 100         case RESOURCE_IN_WIM:
 101                 list_add(&new->wim_resource_list, &new->rspec->lte_list);
 102                 break;
 103
 104         case RESOURCE_IN_FILE_ON_DISK:
 105 #ifdef __WIN32__
 106         case RESOURCE_WIN32_ENCRYPTED:
 107 #endif
 108 #ifdef WITH_FUSE
 109         case RESOURCE_IN_STAGING_FILE:
 110                 BUILD_BUG_ON((void*)&old->file_on_disk !=
 111                              (void*)&old->staging_file_name);
 112 #endif
 113                 new->file_on_disk = TSTRDUP(old->file_on_disk);
 114                 if (!new->file_on_disk)
 115                         goto out_free;
 116                 break;
 117         case RESOURCE_IN_ATTACHED_BUFFER:
 118                 new->attached_buffer = memdup(old->attached_buffer, old->size);
 119                 if (!new->attached_buffer)
 120                         goto out_free;
 121                 break;
 122 #ifdef WITH_NTFS_3G
 123         case RESOURCE_IN_NTFS_VOLUME:
 124                 if (old->ntfs_loc) {
 125                         struct ntfs_location *loc;
 126                         loc = memdup(old->ntfs_loc, sizeof(struct ntfs_location));
 127                         if (!loc)
 128                                 goto out_free;
 129                         loc->path = NULL;
 130                         loc->stream_name = NULL;
 131                         new->ntfs_loc = loc;
 132                         loc->path = STRDUP(old->ntfs_loc->path);
 133                         if (!loc->path)
 134                                 goto out_free;
 135                         if (loc->stream_name_nchars) {
 136                                 loc->stream_name = memdup(old->ntfs_loc->stream_name,
 137                                                           loc->stream_name_nchars * 2);
 138                                 if (!loc->stream_name)
 139                                         goto out_free;
 140                         }
 141                 }
 142                 break;
 143 #endif
 144         default:
 145                 break;
 146         }
 147         return new;
 148 out_free:
 149         free_lookup_table_entry(new);
 150         return NULL;
 151 }
 152
 153 void
 154 free_lookup_table_entry(struct wim_lookup_table_entry *lte)
 155 {
 156         if (lte) {
 157                 switch (lte->resource_location) {
 158                 case RESOURCE_IN_WIM:
 159                         list_del(&lte->wim_resource_list);
 160                         if (list_empty(&lte->rspec->lte_list))
 161                                 FREE(lte->rspec);
 162                         break;
 163                 case RESOURCE_IN_FILE_ON_DISK:
 164         #ifdef __WIN32__
 165                 case RESOURCE_WIN32_ENCRYPTED:
 166         #endif
 167         #ifdef WITH_FUSE
 168                 case RESOURCE_IN_STAGING_FILE:
 169                         BUILD_BUG_ON((void*)&lte->file_on_disk !=
 170                                      (void*)&lte->staging_file_name);
 171         #endif
 172                 case RESOURCE_IN_ATTACHED_BUFFER:
 173                         BUILD_BUG_ON((void*)&lte->file_on_disk !=
 174                                      (void*)&lte->attached_buffer);
 175                         FREE(lte->file_on_disk);
 176                         break;
 177 #ifdef WITH_NTFS_3G
 178                 case RESOURCE_IN_NTFS_VOLUME:
 179                         if (lte->ntfs_loc) {
 180                                 FREE(lte->ntfs_loc->path);
 181                                 FREE(lte->ntfs_loc->stream_name);
 182                                 FREE(lte->ntfs_loc);
 183                         }
 184                         break;
 185 #endif
 186                 default:
 187                         break;
 188                 }
 189                 FREE(lte);
 190         }
 191 }
 192
 193 static int
 194 do_free_lookup_table_entry(struct wim_lookup_table_entry *entry, void *ignore)
 195 {
 196         free_lookup_table_entry(entry);
 197         return 0;
 198 }
 199
 200
 201 void
 202 free_lookup_table(struct wim_lookup_table *table)
 203 {
 204         DEBUG2("Freeing lookup table");
 205         if (table) {
 206                 if (table->array) {
 207                         for_lookup_table_entry(table,
 208                                                do_free_lookup_table_entry,
 209                                                NULL);
 210                         FREE(table->array);
 211                 }
 212                 FREE(table);
 213         }
 214 }
 215
 216 /*
 217  * Inserts an entry into the lookup table.
 218  *
 219  * @table:      A pointer to the lookup table.
 220  * @lte:        A pointer to the entry to insert.
 221  */
 222 void
 223 lookup_table_insert(struct wim_lookup_table *table,
 224                     struct wim_lookup_table_entry *lte)
 225 {
 226         size_t i = lte->hash_short % table->capacity;
 227         hlist_add_head(&lte->hash_list, &table->array[i]);
 228
 229         /* XXX Make the table grow when too many entries have been inserted. */
 230         table->num_entries++;
 231 }
 232
 233 static void
 234 finalize_lte(struct wim_lookup_table_entry *lte)
 235 {
 236         #ifdef WITH_FUSE
 237         if (lte->resource_location == RESOURCE_IN_STAGING_FILE) {
 238                 unlink(lte->staging_file_name);
 239                 list_del(&lte->unhashed_list);
 240         }
 241         #endif
 242         free_lookup_table_entry(lte);
 243 }
 244
 245 /* Decrements the reference count for the lookup table entry @lte.  If its
 246  * reference count reaches 0, it is unlinked from the lookup table.  If,
 247  * furthermore, the entry has no opened file descriptors associated with it, the
 248  * entry is freed.  */
 249 void
 250 lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
 251                      struct wim_lookup_table *table)
 252 {
 253         wimlib_assert(lte != NULL);
 254         wimlib_assert(lte->refcnt != 0);
 255         if (--lte->refcnt == 0) {
 256                 if (lte->unhashed)
 257                         list_del(&lte->unhashed_list);
 258                 else
 259                         lookup_table_unlink(table, lte);
 260         #ifdef WITH_FUSE
 261                 if (lte->num_opened_fds == 0)
 262         #endif
 263                         finalize_lte(lte);
 264         }
 265 }
 266
 267 #ifdef WITH_FUSE
 268 void
 269 lte_decrement_num_opened_fds(struct wim_lookup_table_entry *lte)
 270 {
 271         if (lte->num_opened_fds != 0)
 272                 if (--lte->num_opened_fds == 0 && lte->refcnt == 0)
 273                         finalize_lte(lte);
 274 }
 275 #endif
 276
 277 /* Calls a function on all the entries in the WIM lookup table.  Stop early and
 278  * return nonzero if any call to the function returns nonzero. */
 279 int
 280 for_lookup_table_entry(struct wim_lookup_table *table,
 281                        int (*visitor)(struct wim_lookup_table_entry *, void *),
 282                        void *arg)
 283 {
 284         struct wim_lookup_table_entry *lte;
 285         struct hlist_node *pos, *tmp;
 286         int ret;
 287
 288         for (size_t i = 0; i < table->capacity; i++) {
 289                 hlist_for_each_entry_safe(lte, pos, tmp, &table->array[i],
 290                                           hash_list)
 291                 {
 292                         wimlib_assert2(!(lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA));
 293                         ret = visitor(lte, arg);
 294                         if (ret)
 295                                 return ret;
 296                 }
 297         }
 298         return 0;
 299 }
 300
 301 /* qsort() callback that sorts streams (represented by `struct
 302  * wim_lookup_table_entry's) into an order optimized for reading and writing.
 303  *
 304  * Sorting is done primarily by resource location, then secondarily by a
 305  * per-resource location order.  For example, resources in WIM files are sorted
 306  * primarily by part number, then secondarily by offset, as to implement optimal
 307  * reading of either a standalone or split WIM.  */
 308 static int
 309 cmp_streams_by_sequential_order(const void *p1, const void *p2)
 310 {
 311         const struct wim_lookup_table_entry *lte1, *lte2;
 312         int v;
 313         WIMStruct *wim1, *wim2;
 314
 315         lte1 = *(const struct wim_lookup_table_entry**)p1;
 316         lte2 = *(const struct wim_lookup_table_entry**)p2;
 317
 318         v = (int)lte1->resource_location - (int)lte2->resource_location;
 319
 320         /* Different resource locations?  */
 321         if (v)
 322                 return v;
 323
 324         switch (lte1->resource_location) {
 325         case RESOURCE_IN_WIM:
 326                 wim1 = lte1->rspec->wim;
 327                 wim2 = lte2->rspec->wim;
 328
 329                 /* Different (possibly split) WIMs?  */
 330                 if (wim1 != wim2) {
 331                         v = memcmp(wim1->hdr.guid, wim2->hdr.guid, WIM_GID_LEN);
 332                         if (v)
 333                                 return v;
 334                 }
 335
 336                 /* Different part numbers in the same WIM?  */
 337                 v = (int)wim1->hdr.part_number - (int)wim2->hdr.part_number;
 338                 if (v)
 339                         return v;
 340
 341                 /* Compare by offset.  */
 342                 if (lte1->rspec->offset_in_wim < lte2->rspec->offset_in_wim)
 343                         return -1;
 344                 if (lte1->rspec->offset_in_wim > lte2->rspec->offset_in_wim)
 345                         return 1;
 346                 return 0;
 347         case RESOURCE_IN_FILE_ON_DISK:
 348 #ifdef WITH_FUSE
 349         case RESOURCE_IN_STAGING_FILE:
 350 #endif
 351 #ifdef __WIN32__
 352         case RESOURCE_WIN32_ENCRYPTED:
 353 #endif
 354                 /* Compare files by path: just a heuristic that will place files
 355                  * in the same directory next to each other.  */
 356                 return tstrcmp(lte1->file_on_disk, lte2->file_on_disk);
 357 #ifdef WITH_NTFS_3G
 358         case RESOURCE_IN_NTFS_VOLUME:
 359                 return tstrcmp(lte1->ntfs_loc->path, lte2->ntfs_loc->path);
 360 #endif
 361         default:
 362                 /* No additional sorting order defined for this resource
 363                  * location (e.g. RESOURCE_IN_ATTACHED_BUFFER); simply compare
 364                  * everything equal to each other.  */
 365                 return 0;
 366         }
 367 }
 368
 369 int
 370 sort_stream_list_by_sequential_order(struct list_head *stream_list,
 371                                      size_t list_head_offset)
 372 {
 373         struct list_head *cur;
 374         struct wim_lookup_table_entry **array;
 375         size_t i;
 376         size_t array_size;
 377         size_t num_streams = 0;
 378
 379         list_for_each(cur, stream_list)
 380                 num_streams++;
 381
 382         array_size = num_streams * sizeof(array[0]);
 383         array = MALLOC(array_size);
 384         if (!array)
 385                 return WIMLIB_ERR_NOMEM;
 386         cur = stream_list->next;
 387         for (i = 0; i < num_streams; i++) {
 388                 array[i] = (struct wim_lookup_table_entry*)((u8*)cur -
 389                                                             list_head_offset);
 390                 cur = cur->next;
 391         }
 392
 393         qsort(array, num_streams, sizeof(array[0]),
 394               cmp_streams_by_sequential_order);
 395
 396         INIT_LIST_HEAD(stream_list);
 397         for (i = 0; i < num_streams; i++) {
 398                 list_add_tail((struct list_head*)
 399                                ((u8*)array[i] + list_head_offset),
 400                               stream_list);
 401         }
 402         FREE(array);
 403         return 0;
 404 }
 405
 406
 407 static int
 408 add_lte_to_array(struct wim_lookup_table_entry *lte,
 409                  void *_pp)
 410 {
 411         struct wim_lookup_table_entry ***pp = _pp;
 412         *(*pp)++ = lte;
 413         return 0;
 414 }
 415
 416 /* Iterate through the lookup table entries, but first sort them by stream
 417  * offset in the WIM.  Caution: this is intended to be used when the stream
 418  * offset field has actually been set. */
 419 int
 420 for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
 421                                   int (*visitor)(struct wim_lookup_table_entry *,
 422                                                  void *),
 423                                   void *arg)
 424 {
 425         struct wim_lookup_table_entry **lte_array, **p;
 426         size_t num_streams = table->num_entries;
 427         int ret;
 428
 429         lte_array = MALLOC(num_streams * sizeof(lte_array[0]));
 430         if (!lte_array)
 431                 return WIMLIB_ERR_NOMEM;
 432         p = lte_array;
 433         for_lookup_table_entry(table, add_lte_to_array, &p);
 434
 435         wimlib_assert(p == lte_array + num_streams);
 436
 437         qsort(lte_array, num_streams, sizeof(lte_array[0]),
 438               cmp_streams_by_sequential_order);
 439         ret = 0;
 440         for (size_t i = 0; i < num_streams; i++) {
 441                 ret = visitor(lte_array[i], arg);
 442                 if (ret)
 443                         break;
 444         }
 445         FREE(lte_array);
 446         return ret;
 447 }
 448
 449 /* On-disk format of a WIM lookup table entry (stream entry). */
 450 struct wim_lookup_table_entry_disk {
 451         /* Size, offset, and flags of the stream.  */
 452         struct wim_reshdr_disk reshdr;
 453
 454         /* Which part of the split WIM this stream is in; indexed from 1. */
 455         le16 part_number;
 456
 457         /* Reference count of this stream over all WIM images. */
 458         le32 refcnt;
 459
 460         /* SHA1 message digest of the uncompressed data of this stream, or
 461          * optionally all zeroes if this stream is of zero length. */
 462         u8 hash[SHA1_HASH_SIZE];
 463 } _packed_attribute;
 464
 465 #define WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE 50
 466
 467 /*
 468  * Reads the lookup table from a WIM file.
 469  *
 470  * Saves lookup table entries for non-metadata streams in a hash table, and
 471  * saves the metadata entry for each image in a special per-image location (the
 472  * image_metadata array).
 473  *
 474  * Return values:
 475  *      WIMLIB_ERR_SUCCESS (0)
 476  *      WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY
 477  *      WIMLIB_ERR_RESOURCE_NOT_FOUND
 478  */
 479 int
 480 read_wim_lookup_table(WIMStruct *wim)
 481 {
 482         int ret;
 483         size_t i;
 484         size_t num_entries;
 485         struct wim_lookup_table *table;
 486         struct wim_lookup_table_entry *cur_entry, *duplicate_entry;
 487         void *buf;
 488         bool in_concat_run;
 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.");
 494
 495         /* Calculate number of entries in the lookup table.  */
 496         num_entries = wim->hdr.lookup_table_reshdr.uncompressed_size /
 497                       sizeof(struct wim_lookup_table_entry_disk);
 498
 499         /* Read the lookup table into a buffer.  */
 500         ret = wim_reshdr_to_data(&wim->hdr.lookup_table_reshdr, wim, &buf);
 501         if (ret)
 502                 goto out;
 503
 504         /* Allocate hash table.  */
 505         table = new_lookup_table(num_entries * 2 + 1);
 506         if (!table) {
 507                 ERROR("Not enough memory to read lookup table.");
 508                 ret = WIMLIB_ERR_NOMEM;
 509                 goto out_free_buf;
 510         }
 511
 512         /* Allocate and initalize `struct wim_lookup_table_entry's from the
 513          * on-disk lookup table.  */
 514         wim->current_image = 0;
 515         in_concat_run = false;
 516         for (i = 0; i < num_entries; i++) {
 517                 const struct wim_lookup_table_entry_disk *disk_entry =
 518                         &((const struct wim_lookup_table_entry_disk*)buf)[i];
 519                 u16 part_number;
 520                 struct wim_reshdr reshdr;
 521                 struct wim_resource_spec *cur_rspec;
 522
 523                 cur_entry = new_lookup_table_entry();
 524                 if (cur_entry == NULL) {
 525                         ERROR("Not enough memory to read lookup table.");
 526                         ret = WIMLIB_ERR_NOMEM;
 527                         goto out_free_lookup_table;
 528                 }
 529
 530                 part_number = le16_to_cpu(disk_entry->part_number);
 531                 cur_entry->refcnt = le32_to_cpu(disk_entry->refcnt);
 532                 copy_hash(cur_entry->hash, disk_entry->hash);
 533
 534                 if (part_number != wim->hdr.part_number) {
 535                         WARNING("A lookup table entry in part %hu of the WIM "
 536                                 "points to part %hu (ignoring it)",
 537                                 wim->hdr.part_number, part_number);
 538                         free_lookup_table_entry(cur_entry);
 539                         continue;
 540                 }
 541                 if (is_zero_hash(cur_entry->hash)) {
 542                         WARNING("The WIM lookup table contains an entry with a "
 543                                 "SHA1 message digest of all 0's (ignoring it)");
 544                         free_lookup_table_entry(cur_entry);
 545                         continue;
 546                 }
 547
 548                 cur_rspec = MALLOC(sizeof(struct wim_resource_spec));
 549                 if (cur_rspec == NULL) {
 550                         ERROR("Not enough memory to read lookup table.");
 551                         ret = WIMLIB_ERR_NOMEM;
 552                         goto out_free_cur_entry;
 553                 }
 554
 555                 get_wim_reshdr(&disk_entry->reshdr, &reshdr);
 556                 wim_res_hdr_to_spec(&reshdr, wim, cur_rspec);
 557                 lte_bind_wim_resource_spec(cur_entry, cur_rspec);
 558
 559                 if (cur_entry->flags & WIM_RESHDR_FLAG_METADATA) {
 560                         /* Lookup table entry for a metadata resource */
 561                         if (cur_entry->refcnt != 1) {
 562                                 /* Metadata entries with no references must be
 563                                  * ignored.  See for example the WinPE WIMs from
 564                                  * WAIK v2.1.  */
 565                                 if (cur_entry->refcnt == 0) {
 566                                         free_lookup_table_entry(cur_entry);
 567                                         continue;
 568                                 }
 569                                 if (wimlib_print_errors) {
 570                                         ERROR("Found metadata resource with refcnt != 1:");
 571                                         print_lookup_table_entry(cur_entry, stderr);
 572                                 }
 573                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 574                                 goto out_free_cur_entry;
 575                         }
 576
 577                         if (wim->hdr.part_number != 1) {
 578                                 WARNING("Ignoring metadata resource found in a "
 579                                         "non-first part of the split WIM");
 580                                 free_lookup_table_entry(cur_entry);
 581                                 continue;
 582                         }
 583                         if (wim->current_image == wim->hdr.image_count) {
 584                                 WARNING("The WIM header says there are %u images "
 585                                         "in the WIM, but we found more metadata "
 586                                         "resources than this (ignoring the extra)",
 587                                         wim->hdr.image_count);
 588                                 free_lookup_table_entry(cur_entry);
 589                                 continue;
 590                         }
 591
 592                         /* Notice very carefully:  We are assigning the metadata
 593                          * resources in the exact order mirrored by their lookup
 594                          * table entries on disk, which is the behavior of
 595                          * Microsoft's software.  In particular, this overrides
 596                          * the actual locations of the metadata resources
 597                          * themselves in the WIM file as well as any information
 598                          * written in the XML data. */
 599                         DEBUG("Found metadata resource for image %u at "
 600                               "offset %"PRIu64".",
 601                               wim->current_image + 1,
 602                               cur_entry->rspec->offset_in_wim);
 603                         wim->image_metadata[
 604                                 wim->current_image++]->metadata_lte = cur_entry;
 605                 } else {
 606                         /* Lookup table entry for a stream that is not a
 607                          * metadata resource */
 608                         duplicate_entry = lookup_resource(table, cur_entry->hash);
 609                         if (duplicate_entry) {
 610                                 if (wimlib_print_errors) {
 611                                         WARNING("The WIM lookup table contains two entries with the "
 612                                               "same SHA1 message digest!");
 613                                         WARNING("The first entry is:");
 614                                         print_lookup_table_entry(duplicate_entry, stderr);
 615                                         WARNING("The second entry is:");
 616                                         print_lookup_table_entry(cur_entry, stderr);
 617                                 }
 618                                 free_lookup_table_entry(cur_entry);
 619                                 continue;
 620                         } else {
 621                                 lookup_table_insert(table, cur_entry);
 622                         }
 623                 }
 624         }
 625
 626         if (wim->hdr.part_number == 1 && wim->current_image != wim->hdr.image_count) {
 627                 WARNING("The header of \"%"TS"\" says there are %u images in\n"
 628                         "          the WIM, but we only found %d metadata resources!  Acting as if\n"
 629                         "          the header specified only %d images instead.",
 630                         wim->filename, wim->hdr.image_count,
 631                         wim->current_image, wim->current_image);
 632                 for (int i = wim->current_image; i < wim->hdr.image_count; i++)
 633                         put_image_metadata(wim->image_metadata[i], NULL);
 634                 wim->hdr.image_count = wim->current_image;
 635         }
 636         DEBUG("Done reading lookup table.");
 637         wim->lookup_table = table;
 638         ret = 0;
 639         goto out_free_buf;
 640
 641 out_free_cur_entry:
 642         FREE(cur_entry);
 643 out_free_lookup_table:
 644         free_lookup_table(table);
 645 out_free_buf:
 646         FREE(buf);
 647 out:
 648         wim->current_image = 0;
 649         return ret;
 650 }
 651
 652
 653 static void
 654 write_wim_lookup_table_entry(const struct wim_lookup_table_entry *lte,
 655                              struct wim_lookup_table_entry_disk *disk_entry,
 656                              u16 part_number)
 657 {
 658         put_wim_reshdr(&lte->out_reshdr, &disk_entry->reshdr);
 659         disk_entry->part_number = cpu_to_le16(part_number);
 660         disk_entry->refcnt = cpu_to_le32(lte->out_refcnt);
 661         copy_hash(disk_entry->hash, lte->hash);
 662 }
 663
 664 static int
 665 write_wim_lookup_table_from_stream_list(struct list_head *stream_list,
 666                                         struct filedes *out_fd,
 667                                         u16 part_number,
 668                                         struct wim_reshdr *out_reshdr,
 669                                         int write_resource_flags,
 670                                         struct wimlib_lzx_context **comp_ctx)
 671 {
 672         size_t table_size;
 673         struct wim_lookup_table_entry *lte;
 674         struct wim_lookup_table_entry_disk *table_buf;
 675         struct wim_lookup_table_entry_disk *table_buf_ptr;
 676         int ret;
 677
 678         table_size = 0;
 679         list_for_each_entry(lte, stream_list, lookup_table_list)
 680                 table_size += sizeof(struct wim_lookup_table_entry_disk);
 681
 682         DEBUG("Writing WIM lookup table (size=%zu, offset=%"PRIu64")",
 683               table_size, out_fd->offset);
 684
 685         table_buf = MALLOC(table_size);
 686         if (!table_buf) {
 687                 ERROR("Failed to allocate %zu bytes for temporary lookup table",
 688                       table_size);
 689                 return WIMLIB_ERR_NOMEM;
 690         }
 691         table_buf_ptr = table_buf;
 692         list_for_each_entry(lte, stream_list, lookup_table_list)
 693                 write_wim_lookup_table_entry(lte, table_buf_ptr++, part_number);
 694
 695         /* Write the lookup table uncompressed.  Although wimlib can handle a
 696          * compressed lookup table, MS software cannot.  */
 697         ret = write_wim_resource_from_buffer(table_buf,
 698                                              table_size,
 699                                              WIM_RESHDR_FLAG_METADATA,
 700                                              out_fd,
 701                                              WIMLIB_COMPRESSION_TYPE_NONE,
 702                                              0,
 703                                              out_reshdr,
 704                                              NULL,
 705                                              write_resource_flags,
 706                                              comp_ctx);
 707         FREE(table_buf);
 708         DEBUG("ret=%d", ret);
 709         return ret;
 710 }
 711
 712 static int
 713 append_lookup_table_entry(struct wim_lookup_table_entry *lte, void *_list)
 714 {
 715         /* Lookup table entries with 'out_refcnt' == 0 correspond to streams not
 716          * written and not present in the resulting WIM file, and should not be
 717          * included in the lookup table.
 718          *
 719          * Lookup table entries marked as filtered (EXTERNAL_WIM) with
 720          * 'out_refcnt != 0' were referenced as part of the logical write but
 721          * correspond to streams that were not in fact written, and should not
 722          * be included in the lookup table.
 723          *
 724          * Lookup table entries marked as filtered (SAME_WIM) with 'out_refcnt
 725          * != 0' were referenced as part of the logical write but correspond to
 726          * streams that were not in fact written, but nevertheless were already
 727          * present in the WIM being overwritten in-place.  These entries must be
 728          * included in the lookup table, and the resource information to write
 729          * needs to be copied from the resource information read originally.
 730          */
 731         if (lte->out_refcnt != 0 && !(lte->filtered & FILTERED_EXTERNAL_WIM)) {
 732                 if (lte->filtered & FILTERED_SAME_WIM)
 733                         wim_res_spec_to_hdr(lte->rspec, &lte->out_reshdr);
 734                 list_add_tail(&lte->lookup_table_list, (struct list_head*)_list);
 735         }
 736         return 0;
 737 }
 738
 739 int
 740 write_wim_lookup_table(WIMStruct *wim, int image, int write_flags,
 741                        struct wim_reshdr *out_reshdr,
 742                        struct list_head *stream_list_override)
 743 {
 744         int write_resource_flags;
 745         struct list_head _stream_list;
 746         struct list_head *stream_list;
 747
 748         if (stream_list_override) {
 749                 stream_list = stream_list_override;
 750         } else {
 751                 stream_list = &_stream_list;
 752                 INIT_LIST_HEAD(stream_list);
 753         }
 754
 755         if (!(write_flags & WIMLIB_WRITE_FLAG_NO_METADATA)) {
 756                 int start_image;
 757                 int end_image;
 758
 759                 if (image == WIMLIB_ALL_IMAGES) {
 760                         start_image = 1;
 761                         end_image = wim->hdr.image_count;
 762                 } else {
 763                         start_image = image;
 764                         end_image = image;
 765                 }
 766
 767                 /* Push metadata resource lookup table entries onto the front of
 768                  * the list in reverse order, so that they're written in order.
 769                  */
 770                 for (int i = end_image; i >= start_image; i--) {
 771                         struct wim_lookup_table_entry *metadata_lte;
 772
 773                         metadata_lte = wim->image_metadata[i - 1]->metadata_lte;
 774                         metadata_lte->out_refcnt = 1;
 775                         metadata_lte->out_reshdr.flags |= WIM_RESHDR_FLAG_METADATA;
 776                         list_add(&metadata_lte->lookup_table_list, stream_list);
 777                 }
 778         }
 779
 780         /* Append additional lookup table entries that need to be written, with
 781          * some special handling for streams that have been marked as filtered.
 782          */
 783         if (!stream_list_override) {
 784                 for_lookup_table_entry(wim->lookup_table,
 785                                        append_lookup_table_entry, stream_list);
 786         }
 787
 788         write_resource_flags = 0;
 789         if (write_flags & WIMLIB_WRITE_FLAG_PIPABLE)
 790                 write_resource_flags |= WIMLIB_WRITE_RESOURCE_FLAG_PIPABLE;
 791         return write_wim_lookup_table_from_stream_list(stream_list,
 792                                                        &wim->out_fd,
 793                                                        wim->hdr.part_number,
 794                                                        out_reshdr,
 795                                                        write_resource_flags,
 796                                                        &wim->lzx_context);
 797 }
 798
 799
 800 int
 801 lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 802 {
 803         lte->real_refcnt = 0;
 804         return 0;
 805 }
 806
 807 int
 808 lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 809 {
 810         lte->out_refcnt = 0;
 811         return 0;
 812 }
 813
 814 int
 815 lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *_ignore)
 816 {
 817         if (lte->extracted_file != NULL) {
 818                 FREE(lte->extracted_file);
 819                 lte->extracted_file = NULL;
 820         }
 821         return 0;
 822 }
 823
 824 void
 825 print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 826 {
 827         if (lte == NULL) {
 828                 tputc(T('\n'), out);
 829                 return;
 830         }
 831
 832         tfprintf(out, T("Reference Count    = %u\n"), lte->refcnt);
 833         tfprintf(out, T("Uncompressed Size  = %"PRIu64" bytes\n"), lte->size);
 834
 835         if (lte->resource_location == RESOURCE_IN_WIM) {
 836                 tfprintf(out, T("Offset in WIM    = %"PRIu64" bytes\n"),
 837                          lte->rspec->offset_in_wim);
 838
 839                 tfprintf(out, T("Size in WIM      = %"PRIu64" bytes\n"),
 840                          lte->rspec->size_in_wim);
 841         }
 842
 843         if (lte->unhashed) {
 844                 tfprintf(out, T("(Unhashed: inode %p, stream_id = %u)\n"),
 845                          lte->back_inode, lte->back_stream_id);
 846         } else {
 847                 tfprintf(out, T("Hash              = 0x"));
 848                 print_hash(lte->hash, out);
 849                 tputc(T('\n'), out);
 850         }
 851
 852         tfprintf(out, T("Flags             = "));
 853         u8 flags = lte->flags;
 854         if (flags & WIM_RESHDR_FLAG_COMPRESSED)
 855                 tfputs(T("WIM_RESHDR_FLAG_COMPRESSED, "), out);
 856         if (flags & WIM_RESHDR_FLAG_FREE)
 857                 tfputs(T("WIM_RESHDR_FLAG_FREE, "), out);
 858         if (flags & WIM_RESHDR_FLAG_METADATA)
 859                 tfputs(T("WIM_RESHDR_FLAG_METADATA, "), out);
 860         if (flags & WIM_RESHDR_FLAG_SPANNED)
 861                 tfputs(T("WIM_RESHDR_FLAG_SPANNED, "), out);
 862         if (flags & WIM_RESHDR_FLAG_CONCAT)
 863                 tfputs(T("WIM_RESHDR_FLAG_CONCAT, "), out);
 864         tputc(T('\n'), out);
 865         switch (lte->resource_location) {
 866         case RESOURCE_IN_WIM:
 867                 if (lte->rspec->wim->filename) {
 868                         tfprintf(out, T("WIM file          = `%"TS"'\n"),
 869                                  lte->rspec->wim->filename);
 870                 }
 871                 break;
 872 #ifdef __WIN32__
 873         case RESOURCE_WIN32_ENCRYPTED:
 874 #endif
 875         case RESOURCE_IN_FILE_ON_DISK:
 876                 tfprintf(out, T("File on Disk      = `%"TS"'\n"),
 877                          lte->file_on_disk);
 878                 break;
 879 #ifdef WITH_FUSE
 880         case RESOURCE_IN_STAGING_FILE:
 881                 tfprintf(out, T("Staging File      = `%"TS"'\n"),
 882                                 lte->staging_file_name);
 883                 break;
 884 #endif
 885         default:
 886                 break;
 887         }
 888         tputc(T('\n'), out);
 889 }
 890
 891 void
 892 lte_to_wimlib_resource_entry(const struct wim_lookup_table_entry *lte,
 893                              struct wimlib_resource_entry *wentry)
 894 {
 895         wentry->uncompressed_size = lte->size;
 896
 897         if (lte->resource_location == RESOURCE_IN_WIM) {
 898                 wentry->compressed_size = lte->rspec->size_in_wim;
 899                 wentry->offset = lte->rspec->offset_in_wim;
 900                 wentry->part_number = lte->rspec->wim->hdr.part_number;
 901         } else {
 902                 wentry->compressed_size = 0;
 903                 wentry->offset = 0;
 904                 wentry->part_number = 0;
 905         }
 906         copy_hash(wentry->sha1_hash, lte->hash);
 907         wentry->reference_count = lte->refcnt;
 908         wentry->is_compressed = (lte->flags & WIM_RESHDR_FLAG_COMPRESSED) != 0;
 909         wentry->is_metadata = (lte->flags & WIM_RESHDR_FLAG_METADATA) != 0;
 910         wentry->is_free = (lte->flags & WIM_RESHDR_FLAG_FREE) != 0;
 911         wentry->is_spanned = (lte->flags & WIM_RESHDR_FLAG_SPANNED) != 0;
 912 }
 913
 914 struct iterate_lte_context {
 915         wimlib_iterate_lookup_table_callback_t cb;
 916         void *user_ctx;
 917 };
 918
 919 static int
 920 do_iterate_lte(struct wim_lookup_table_entry *lte, void *_ctx)
 921 {
 922         struct iterate_lte_context *ctx = _ctx;
 923         struct wimlib_resource_entry entry;
 924
 925         lte_to_wimlib_resource_entry(lte, &entry);
 926         return (*ctx->cb)(&entry, ctx->user_ctx);
 927 }
 928
 929 /* API function documented in wimlib.h  */
 930 WIMLIBAPI int
 931 wimlib_iterate_lookup_table(WIMStruct *wim, int flags,
 932                             wimlib_iterate_lookup_table_callback_t cb,
 933                             void *user_ctx)
 934 {
 935         struct iterate_lte_context ctx = {
 936                 .cb = cb,
 937                 .user_ctx = user_ctx,
 938         };
 939         if (wim->hdr.part_number == 1) {
 940                 int ret;
 941                 for (int i = 0; i < wim->hdr.image_count; i++) {
 942                         ret = do_iterate_lte(wim->image_metadata[i]->metadata_lte,
 943                                              &ctx);
 944                         if (ret)
 945                                 return ret;
 946                 }
 947         }
 948         return for_lookup_table_entry(wim->lookup_table, do_iterate_lte, &ctx);
 949 }
 950
 951 /* Given a SHA1 message digest, return the corresponding entry in the WIM's
 952  * lookup table, or NULL if there is none.  */
 953 struct wim_lookup_table_entry *
 954 lookup_resource(const struct wim_lookup_table *table, const u8 hash[])
 955 {
 956         size_t i;
 957         struct wim_lookup_table_entry *lte;
 958         struct hlist_node *pos;
 959
 960         wimlib_assert(table != NULL);
 961         wimlib_assert(hash != NULL);
 962
 963         i = *(size_t*)hash % table->capacity;
 964         hlist_for_each_entry(lte, pos, &table->array[i], hash_list)
 965                 if (hashes_equal(hash, lte->hash))
 966                         return lte;
 967         return NULL;
 968 }
 969
 970 #ifdef WITH_FUSE
 971 /*
 972  * Finds the dentry, lookup table entry, and stream index for a WIM file stream,
 973  * given a path name.
 974  *
 975  * This is only for pre-resolved inodes.
 976  */
 977 int
 978 wim_pathname_to_stream(WIMStruct *wim,
 979                        const tchar *path,
 980                        int lookup_flags,
 981                        struct wim_dentry **dentry_ret,
 982                        struct wim_lookup_table_entry **lte_ret,
 983                        u16 *stream_idx_ret)
 984 {
 985         struct wim_dentry *dentry;
 986         struct wim_lookup_table_entry *lte;
 987         u16 stream_idx;
 988         const tchar *stream_name = NULL;
 989         struct wim_inode *inode;
 990         tchar *p = NULL;
 991
 992         if (lookup_flags & LOOKUP_FLAG_ADS_OK) {
 993                 stream_name = path_stream_name(path);
 994                 if (stream_name) {
 995                         p = (tchar*)stream_name - 1;
 996                         *p = T('\0');
 997                 }
 998         }
 999
1000         dentry = get_dentry(wim, path);
1001         if (p)
1002                 *p = T(':');
1003         if (!dentry)
1004                 return -errno;
1005
1006         inode = dentry->d_inode;
1007
1008         if (!inode->i_resolved)
1009                 if (inode_resolve_ltes(inode, wim->lookup_table, false))
1010                         return -EIO;
1011
1012         if (!(lookup_flags & LOOKUP_FLAG_DIRECTORY_OK)
1013               && inode_is_directory(inode))
1014                 return -EISDIR;
1015
1016         if (stream_name) {
1017                 struct wim_ads_entry *ads_entry;
1018                 u16 ads_idx;
1019                 ads_entry = inode_get_ads_entry(inode, stream_name,
1020                                                 &ads_idx);
1021                 if (ads_entry) {
1022                         stream_idx = ads_idx + 1;
1023                         lte = ads_entry->lte;
1024                         goto out;
1025                 } else {
1026                         return -ENOENT;
1027                 }
1028         } else {
1029                 lte = inode_unnamed_stream_resolved(inode, &stream_idx);
1030         }
1031 out:
1032         if (dentry_ret)
1033                 *dentry_ret = dentry;
1034         if (lte_ret)
1035                 *lte_ret = lte;
1036         if (stream_idx_ret)
1037                 *stream_idx_ret = stream_idx;
1038         return 0;
1039 }
1040 #endif
1041
1042 int
1043 resource_not_found_error(const struct wim_inode *inode, const u8 *hash)
1044 {
1045         if (wimlib_print_errors) {
1046                 ERROR("\"%"TS"\": resource not found", inode_first_full_path(inode));
1047                 tfprintf(stderr, T("        SHA-1 message digest of missing resource:\n        "));
1048                 print_hash(hash, stderr);
1049                 tputc(T('\n'), stderr);
1050         }
1051         return WIMLIB_ERR_RESOURCE_NOT_FOUND;
1052 }
1053
1054 /*
1055  * Resolve an inode's lookup table entries.
1056  *
1057  * This replaces the SHA1 hash fields (which are used to lookup an entry in the
1058  * lookup table) with pointers directly to the lookup table entries.
1059  *
1060  * If @force is %false:
1061  *      If any needed SHA1 message digests are not found in the lookup table,
1062  *      WIMLIB_ERR_RESOURCE_NOT_FOUND is returned and the inode is left
1063  *      unmodified.
1064  * If @force is %true:
1065  *      If any needed SHA1 message digests are not found in the lookup table,
1066  *      new entries are allocated and inserted into the lookup table.
1067  */
1068 int
1069 inode_resolve_ltes(struct wim_inode *inode, struct wim_lookup_table *table,
1070                    bool force)
1071 {
1072         const u8 *hash;
1073
1074         if (!inode->i_resolved) {
1075                 struct wim_lookup_table_entry *lte, *ads_lte;
1076
1077                 /* Resolve the default file stream */
1078                 lte = NULL;
1079                 hash = inode->i_hash;
1080                 if (!is_zero_hash(hash)) {
1081                         lte = lookup_resource(table, hash);
1082                         if (!lte) {
1083                                 if (force) {
1084                                         lte = new_lookup_table_entry();
1085                                         if (!lte)
1086                                                 return WIMLIB_ERR_NOMEM;
1087                                         copy_hash(lte->hash, hash);
1088                                         lookup_table_insert(table, lte);
1089                                 } else {
1090                                         goto resource_not_found;
1091                                 }
1092                         }
1093                 }
1094
1095                 /* Resolve the alternate data streams */
1096                 struct wim_lookup_table_entry *ads_ltes[inode->i_num_ads];
1097                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1098                         struct wim_ads_entry *cur_entry;
1099
1100                         ads_lte = NULL;
1101                         cur_entry = &inode->i_ads_entries[i];
1102                         hash = cur_entry->hash;
1103                         if (!is_zero_hash(hash)) {
1104                                 ads_lte = lookup_resource(table, hash);
1105                                 if (!ads_lte) {
1106                                         if (force) {
1107                                                 ads_lte = new_lookup_table_entry();
1108                                                 if (!ads_lte)
1109                                                         return WIMLIB_ERR_NOMEM;
1110                                                 copy_hash(ads_lte->hash, hash);
1111                                                 lookup_table_insert(table, ads_lte);
1112                                         } else {
1113                                                 goto resource_not_found;
1114                                         }
1115                                 }
1116                         }
1117                         ads_ltes[i] = ads_lte;
1118                 }
1119                 inode->i_lte = lte;
1120                 for (u16 i = 0; i < inode->i_num_ads; i++)
1121                         inode->i_ads_entries[i].lte = ads_ltes[i];
1122                 inode->i_resolved = 1;
1123         }
1124         return 0;
1125
1126 resource_not_found:
1127         return resource_not_found_error(inode, hash);
1128 }
1129
1130 void
1131 inode_unresolve_ltes(struct wim_inode *inode)
1132 {
1133         if (inode->i_resolved) {
1134                 if (inode->i_lte)
1135                         copy_hash(inode->i_hash, inode->i_lte->hash);
1136                 else
1137                         zero_out_hash(inode->i_hash);
1138
1139                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1140                         if (inode->i_ads_entries[i].lte)
1141                                 copy_hash(inode->i_ads_entries[i].hash,
1142                                           inode->i_ads_entries[i].lte->hash);
1143                         else
1144                                 zero_out_hash(inode->i_ads_entries[i].hash);
1145                 }
1146                 inode->i_resolved = 0;
1147         }
1148 }
1149
1150 /*
1151  * Returns the lookup table entry for stream @stream_idx of the inode, where
1152  * stream_idx = 0 means the default un-named file stream, and stream_idx >= 1
1153  * corresponds to an alternate data stream.
1154  *
1155  * This works for both resolved and un-resolved inodes.
1156  */
1157 struct wim_lookup_table_entry *
1158 inode_stream_lte(const struct wim_inode *inode, unsigned stream_idx,
1159                  const struct wim_lookup_table *table)
1160 {
1161         if (inode->i_resolved)
1162                 return inode_stream_lte_resolved(inode, stream_idx);
1163         else
1164                 return inode_stream_lte_unresolved(inode, stream_idx, table);
1165 }
1166
1167 struct wim_lookup_table_entry *
1168 inode_unnamed_stream_resolved(const struct wim_inode *inode, u16 *stream_idx_ret)
1169 {
1170         wimlib_assert(inode->i_resolved);
1171         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
1172                 if (inode_stream_name_nbytes(inode, i) == 0 &&
1173                     !is_zero_hash(inode_stream_hash_resolved(inode, i)))
1174                 {
1175                         *stream_idx_ret = i;
1176                         return inode_stream_lte_resolved(inode, i);
1177                 }
1178         }
1179         *stream_idx_ret = 0;
1180         return NULL;
1181 }
1182
1183 struct wim_lookup_table_entry *
1184 inode_unnamed_lte_resolved(const struct wim_inode *inode)
1185 {
1186         u16 stream_idx;
1187         return inode_unnamed_stream_resolved(inode, &stream_idx);
1188 }
1189
1190 struct wim_lookup_table_entry *
1191 inode_unnamed_lte_unresolved(const struct wim_inode *inode,
1192                              const struct wim_lookup_table *table)
1193 {
1194         wimlib_assert(!inode->i_resolved);
1195         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
1196                 if (inode_stream_name_nbytes(inode, i) == 0 &&
1197                     !is_zero_hash(inode_stream_hash_unresolved(inode, i)))
1198                 {
1199                         return inode_stream_lte_unresolved(inode, i, table);
1200                 }
1201         }
1202         return NULL;
1203 }
1204
1205 /* Return the lookup table entry for the unnamed data stream of an inode, or
1206  * NULL if there is none.
1207  *
1208  * You'd think this would be easier than it actually is, since the unnamed data
1209  * stream should be the one referenced from the inode itself.  Alas, if there
1210  * are named data streams, Microsoft's "imagex.exe" program will put the unnamed
1211  * data stream in one of the alternate data streams instead of inside the WIM
1212  * dentry itself.  So we need to check the alternate data streams too.
1213  *
1214  * Also, note that a dentry may appear to have more than one unnamed stream, but
1215  * if the SHA1 message digest is all 0's then the corresponding stream does not
1216  * really "count" (this is the case for the inode's own file stream when the
1217  * file stream that should be there is actually in one of the alternate stream
1218  * entries.).  This is despite the fact that we may need to extract such a
1219  * missing entry as an empty file or empty named data stream.
1220  */
1221 struct wim_lookup_table_entry *
1222 inode_unnamed_lte(const struct wim_inode *inode,
1223                   const struct wim_lookup_table *table)
1224 {
1225         if (inode->i_resolved)
1226                 return inode_unnamed_lte_resolved(inode);
1227         else
1228                 return inode_unnamed_lte_unresolved(inode, table);
1229 }
1230
1231 /* Returns the SHA1 message digest of the unnamed data stream of a WIM inode, or
1232  * 'zero_hash' if the unnamed data stream is missing has all zeroes in its SHA1
1233  * message digest field.  */
1234 const u8 *
1235 inode_unnamed_stream_hash(const struct wim_inode *inode)
1236 {
1237         const u8 *hash;
1238
1239         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
1240                 if (inode_stream_name_nbytes(inode, i) == 0) {
1241                         hash = inode_stream_hash(inode, i);
1242                         if (!is_zero_hash(hash))
1243                                 return hash;
1244                 }
1245         }
1246         return zero_hash;
1247 }
1248
1249 struct wim_lookup_table_entry **
1250 retrieve_lte_pointer(struct wim_lookup_table_entry *lte)
1251 {
1252         wimlib_assert(lte->unhashed);
1253         struct wim_inode *inode = lte->back_inode;
1254         u32 stream_id = lte->back_stream_id;
1255         if (stream_id == 0)
1256                 return &inode->i_lte;
1257         else
1258                 for (u16 i = 0; i < inode->i_num_ads; i++)
1259                         if (inode->i_ads_entries[i].stream_id == stream_id)
1260                                 return &inode->i_ads_entries[i].lte;
1261         wimlib_assert(0);
1262         return NULL;
1263 }
1264
1265 /* Calculate the SHA1 message digest of a stream and move it from the list of
1266  * unhashed streams to the stream lookup table, possibly joining it with an
1267  * existing lookup table entry for an identical stream.
1268  *
1269  * @lte:  An unhashed lookup table entry.
1270  * @lookup_table:  Lookup table for the WIM.
1271  * @lte_ret:  On success, write a pointer to the resulting lookup table
1272  *            entry to this location.  This will be the same as @lte
1273  *            if it was inserted into the lookup table, or different if
1274  *            a duplicate stream was found.
1275  *
1276  * Returns 0 on success; nonzero if there is an error reading the stream.
1277  */
1278 int
1279 hash_unhashed_stream(struct wim_lookup_table_entry *lte,
1280                      struct wim_lookup_table *lookup_table,
1281                      struct wim_lookup_table_entry **lte_ret)
1282 {
1283         int ret;
1284         struct wim_lookup_table_entry *duplicate_lte;
1285         struct wim_lookup_table_entry **back_ptr;
1286
1287         wimlib_assert(lte->unhashed);
1288
1289         /* back_ptr must be saved because @back_inode and @back_stream_id are in
1290          * union with the SHA1 message digest and will no longer be valid once
1291          * the SHA1 has been calculated. */
1292         back_ptr = retrieve_lte_pointer(lte);
1293
1294         ret = sha1_resource(lte);
1295         if (ret)
1296                 return ret;
1297
1298         /* Look for a duplicate stream */
1299         duplicate_lte = lookup_resource(lookup_table, lte->hash);
1300         list_del(&lte->unhashed_list);
1301         if (duplicate_lte) {
1302                 /* We have a duplicate stream.  Transfer the reference counts
1303                  * from this stream to the duplicate, update the reference to
1304                  * this stream (in an inode or ads_entry) to point to the
1305                  * duplicate, then free this stream. */
1306                 wimlib_assert(!(duplicate_lte->unhashed));
1307                 duplicate_lte->refcnt += lte->refcnt;
1308                 duplicate_lte->out_refcnt += lte->out_refcnt;
1309                 *back_ptr = duplicate_lte;
1310                 free_lookup_table_entry(lte);
1311                 lte = duplicate_lte;
1312         } else {
1313                 /* No duplicate stream, so we need to insert
1314                  * this stream into the lookup table and treat
1315                  * it as a hashed stream. */
1316                 lookup_table_insert(lookup_table, lte);
1317                 lte->unhashed = 0;
1318         }
1319         if (lte_ret)
1320                 *lte_ret = lte;
1321         return 0;
1322 }
1323
1324 static int
1325 lte_clone_if_new(struct wim_lookup_table_entry *lte, void *_lookup_table)
1326 {
1327         struct wim_lookup_table *lookup_table = _lookup_table;
1328
1329         if (lookup_resource(lookup_table, lte->hash))
1330                 return 0;  /*  Resource already present.  */
1331
1332         lte = clone_lookup_table_entry(lte);
1333         if (!lte)
1334                 return WIMLIB_ERR_NOMEM;
1335         lte->out_refcnt = 1;
1336         lookup_table_insert(lookup_table, lte);
1337         return 0;
1338 }
1339
1340 static int
1341 lte_delete_if_new(struct wim_lookup_table_entry *lte, void *_lookup_table)
1342 {
1343         struct wim_lookup_table *lookup_table = _lookup_table;
1344
1345         if (lte->out_refcnt) {
1346                 lookup_table_unlink(lookup_table, lte);
1347                 free_lookup_table_entry(lte);
1348         }
1349         return 0;
1350 }
1351
1352 /* API function documented in wimlib.h  */
1353 WIMLIBAPI int
1354 wimlib_reference_resources(WIMStruct *wim,
1355                            WIMStruct **resource_wims, unsigned num_resource_wims,
1356                            int ref_flags)
1357 {
1358         int ret;
1359         unsigned i;
1360
1361         if (wim == NULL)
1362                 return WIMLIB_ERR_INVALID_PARAM;
1363
1364         if (num_resource_wims != 0 && resource_wims == NULL)
1365                 return WIMLIB_ERR_INVALID_PARAM;
1366
1367         for (i = 0; i < num_resource_wims; i++)
1368                 if (resource_wims[i] == NULL)
1369                         return WIMLIB_ERR_INVALID_PARAM;
1370
1371         for_lookup_table_entry(wim->lookup_table, lte_zero_out_refcnt, NULL);
1372
1373         for (i = 0; i < num_resource_wims; i++) {
1374                 ret = for_lookup_table_entry(resource_wims[i]->lookup_table,
1375                                              lte_clone_if_new,
1376                                              wim->lookup_table);
1377                 if (ret)
1378                         goto out_rollback;
1379         }
1380         return 0;
1381
1382 out_rollback:
1383         for_lookup_table_entry(wim->lookup_table, lte_delete_if_new,
1384                                wim->lookup_table);
1385         return ret;
1386 }
1387
1388 static int
1389 reference_resource_paths(WIMStruct *wim,
1390                          const tchar * const *resource_wimfiles,
1391                          unsigned num_resource_wimfiles,
1392                          int ref_flags,
1393                          int open_flags,
1394                          wimlib_progress_func_t progress_func)
1395 {
1396         WIMStruct **resource_wims;
1397         unsigned i;
1398         int ret;
1399
1400         resource_wims = CALLOC(num_resource_wimfiles, sizeof(resource_wims[0]));
1401         if (!resource_wims)
1402                 return WIMLIB_ERR_NOMEM;
1403
1404         for (i = 0; i < num_resource_wimfiles; i++) {
1405                 DEBUG("Referencing resources from path \"%"TS"\"",
1406                       resource_wimfiles[i]);
1407                 ret = wimlib_open_wim(resource_wimfiles[i], open_flags,
1408                                       &resource_wims[i], progress_func);
1409                 if (ret)
1410                         goto out_free_resource_wims;
1411         }
1412
1413         ret = wimlib_reference_resources(wim, resource_wims,
1414                                          num_resource_wimfiles, ref_flags);
1415         if (ret)
1416                 goto out_free_resource_wims;
1417
1418         for (i = 0; i < num_resource_wimfiles; i++)
1419                 list_add_tail(&resource_wims[i]->subwim_node, &wim->subwims);
1420
1421         ret = 0;
1422         goto out_free_array;
1423
1424 out_free_resource_wims:
1425         for (i = 0; i < num_resource_wimfiles; i++)
1426                 wimlib_free(resource_wims[i]);
1427 out_free_array:
1428         FREE(resource_wims);
1429         return ret;
1430 }
1431
1432 static int
1433 reference_resource_glob(WIMStruct *wim, const tchar *refglob,
1434                         int ref_flags, int open_flags,
1435                         wimlib_progress_func_t progress_func)
1436 {
1437         glob_t globbuf;
1438         int ret;
1439
1440         /* Note: glob() is replaced in Windows native builds.  */
1441         ret = tglob(refglob, GLOB_ERR | GLOB_NOSORT, NULL, &globbuf);
1442         if (ret) {
1443                 if (ret == GLOB_NOMATCH) {
1444                         if (ref_flags & WIMLIB_REF_FLAG_GLOB_ERR_ON_NOMATCH) {
1445                                 ERROR("Found no files for glob \"%"TS"\"", refglob);
1446                                 return WIMLIB_ERR_GLOB_HAD_NO_MATCHES;
1447                         } else {
1448                                 return reference_resource_paths(wim,
1449                                                                 &refglob,
1450                                                                 1,
1451                                                                 ref_flags,
1452                                                                 open_flags,
1453                                                                 progress_func);
1454                         }
1455                 } else {
1456                         ERROR_WITH_ERRNO("Failed to process glob \"%"TS"\"", refglob);
1457                         if (ret == GLOB_NOSPACE)
1458                                 return WIMLIB_ERR_NOMEM;
1459                         else
1460                                 return WIMLIB_ERR_READ;
1461                 }
1462         }
1463
1464         ret = reference_resource_paths(wim,
1465                                        (const tchar * const *)globbuf.gl_pathv,
1466                                        globbuf.gl_pathc,
1467                                        ref_flags,
1468                                        open_flags,
1469                                        progress_func);
1470         globfree(&globbuf);
1471         return ret;
1472 }
1473
1474 /* API function documented in wimlib.h  */
1475 WIMLIBAPI int
1476 wimlib_reference_resource_files(WIMStruct *wim,
1477                                 const tchar * const * resource_wimfiles_or_globs,
1478                                 unsigned count,
1479                                 int ref_flags,
1480                                 int open_flags,
1481                                 wimlib_progress_func_t progress_func)
1482 {
1483         unsigned i;
1484         int ret;
1485
1486         if (ref_flags & WIMLIB_REF_FLAG_GLOB_ENABLE) {
1487                 for (i = 0; i < count; i++) {
1488                         ret = reference_resource_glob(wim,
1489                                                       resource_wimfiles_or_globs[i],
1490                                                       ref_flags,
1491                                                       open_flags,
1492                                                       progress_func);
1493                         if (ret)
1494                                 return ret;
1495                 }
1496                 return 0;
1497         } else {
1498                 return reference_resource_paths(wim, resource_wimfiles_or_globs,
1499                                                 count, ref_flags,
1500                                                 open_flags, progress_func);
1501         }
1502 }