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

   1 /*
   2  * lookup_table.c
   3  *
   4  * Lookup table, implemented as a hash table, that maps SHA1 message digests to
   5  * data streams; plus code to read and write the corresponding on-disk data.
   6  */
   7
   8 /*
   9  * Copyright (C) 2012, 2013 Eric Biggers
  10  *
  11  * This file is part of wimlib, a library for working with WIM files.
  12  *
  13  * wimlib is free software; you can redistribute it and/or modify it under the
  14  * terms of the GNU General Public License as published by the Free
  15  * Software Foundation; either version 3 of the License, or (at your option)
  16  * any later version.
  17  *
  18  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  19  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  20  * A PARTICULAR PURPOSE. See the GNU General Public License for more
  21  * details.
  22  *
  23  * You should have received a copy of the GNU General Public License
  24  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  25  */
  26
  27 #ifdef HAVE_CONFIG_H
  28 #  include "config.h"
  29 #endif
  30
  31 #include "wimlib/endianness.h"
  32 #include "wimlib/error.h"
  33 #include "wimlib/file_io.h"
  34 #include "wimlib/glob.h"
  35 #include "wimlib/lookup_table.h"
  36 #include "wimlib/metadata.h"
  37 #include "wimlib/paths.h"
  38 #include "wimlib/resource.h"
  39 #include "wimlib/util.h"
  40 #include "wimlib/write.h"
  41
  42 #include <errno.h>
  43 #include <stdlib.h>
  44 #ifdef WITH_FUSE
  45 #  include <unistd.h> /* for unlink() */
  46 #endif
  47
  48 struct wim_lookup_table *
  49 new_lookup_table(size_t capacity)
  50 {
  51         struct wim_lookup_table *table;
  52         struct hlist_head *array;
  53
  54         table = CALLOC(1, sizeof(struct wim_lookup_table));
  55         if (table) {
  56                 array = CALLOC(capacity, sizeof(array[0]));
  57                 if (array) {
  58                         table->num_entries = 0;
  59                         table->capacity = capacity;
  60                         table->array = array;
  61                 } else {
  62                         FREE(table);
  63                         table = NULL;
  64                         ERROR("Failed to allocate memory for lookup table "
  65                               "with capacity %zu", capacity);
  66                 }
  67         }
  68         return table;
  69 }
  70
  71 struct wim_lookup_table_entry *
  72 new_lookup_table_entry(void)
  73 {
  74         struct wim_lookup_table_entry *lte;
  75
  76         lte = CALLOC(1, sizeof(struct wim_lookup_table_entry));
  77         if (lte == NULL) {
  78                 ERROR("Out of memory (tried to allocate %zu bytes for "
  79                       "lookup table entry)",
  80                       sizeof(struct wim_lookup_table_entry));
  81                 return NULL;
  82         }
  83         lte->refcnt = 1;
  84         BUILD_BUG_ON(RESOURCE_NONEXISTENT != 0);
  85         return lte;
  86 }
  87
  88 struct wim_lookup_table_entry *
  89 clone_lookup_table_entry(const struct wim_lookup_table_entry *old)
  90 {
  91         struct wim_lookup_table_entry *new;
  92
  93         new = memdup(old, sizeof(struct wim_lookup_table_entry));
  94         if (!new)
  95                 return NULL;
  96
  97         new->extracted_file = NULL;
  98         switch (new->resource_location) {
  99         case RESOURCE_IN_WIM:
 100                 list_add(&new->wim_resource_list, &new->rspec->stream_list);
 101                 break;
 102
 103         case RESOURCE_IN_FILE_ON_DISK:
 104 #ifdef __WIN32__
 105         case RESOURCE_WIN32_ENCRYPTED:
 106 #endif
 107 #ifdef WITH_FUSE
 108         case RESOURCE_IN_STAGING_FILE:
 109                 BUILD_BUG_ON((void*)&old->file_on_disk !=
 110                              (void*)&old->staging_file_name);
 111 #endif
 112                 new->file_on_disk = TSTRDUP(old->file_on_disk);
 113                 if (!new->file_on_disk)
 114                         goto out_free;
 115                 break;
 116         case RESOURCE_IN_ATTACHED_BUFFER:
 117                 new->attached_buffer = memdup(old->attached_buffer, old->size);
 118                 if (!new->attached_buffer)
 119                         goto out_free;
 120                 break;
 121 #ifdef WITH_NTFS_3G
 122         case RESOURCE_IN_NTFS_VOLUME:
 123                 if (old->ntfs_loc) {
 124                         struct ntfs_location *loc;
 125                         loc = memdup(old->ntfs_loc, sizeof(struct ntfs_location));
 126                         if (!loc)
 127                                 goto out_free;
 128                         loc->path = NULL;
 129                         loc->stream_name = NULL;
 130                         new->ntfs_loc = loc;
 131                         loc->path = STRDUP(old->ntfs_loc->path);
 132                         if (!loc->path)
 133                                 goto out_free;
 134                         if (loc->stream_name_nchars) {
 135                                 loc->stream_name = memdup(old->ntfs_loc->stream_name,
 136                                                           loc->stream_name_nchars * 2);
 137                                 if (!loc->stream_name)
 138                                         goto out_free;
 139                         }
 140                 }
 141                 break;
 142 #endif
 143         default:
 144                 break;
 145         }
 146         return new;
 147 out_free:
 148         free_lookup_table_entry(new);
 149         return NULL;
 150 }
 151
 152 void
 153 free_lookup_table_entry(struct wim_lookup_table_entry *lte)
 154 {
 155         if (lte) {
 156                 switch (lte->resource_location) {
 157                 case RESOURCE_IN_WIM:
 158                         list_del(&lte->wim_resource_list);
 159                         if (list_empty(&lte->rspec->stream_list))
 160                                 FREE(lte->rspec);
 161                         break;
 162                 case RESOURCE_IN_FILE_ON_DISK:
 163         #ifdef __WIN32__
 164                 case RESOURCE_WIN32_ENCRYPTED:
 165         #endif
 166         #ifdef WITH_FUSE
 167                 case RESOURCE_IN_STAGING_FILE:
 168                         BUILD_BUG_ON((void*)&lte->file_on_disk !=
 169                                      (void*)&lte->staging_file_name);
 170         #endif
 171                 case RESOURCE_IN_ATTACHED_BUFFER:
 172                         BUILD_BUG_ON((void*)&lte->file_on_disk !=
 173                                      (void*)&lte->attached_buffer);
 174                         FREE(lte->file_on_disk);
 175                         break;
 176 #ifdef WITH_NTFS_3G
 177                 case RESOURCE_IN_NTFS_VOLUME:
 178                         if (lte->ntfs_loc) {
 179                                 FREE(lte->ntfs_loc->path);
 180                                 FREE(lte->ntfs_loc->stream_name);
 181                                 FREE(lte->ntfs_loc);
 182                         }
 183                         break;
 184 #endif
 185                 default:
 186                         break;
 187                 }
 188                 FREE(lte);
 189         }
 190 }
 191
 192 static int
 193 do_free_lookup_table_entry(struct wim_lookup_table_entry *entry, void *ignore)
 194 {
 195         free_lookup_table_entry(entry);
 196         return 0;
 197 }
 198
 199
 200 void
 201 free_lookup_table(struct wim_lookup_table *table)
 202 {
 203         DEBUG("Freeing lookup table.");
 204         if (table) {
 205                 if (table->array) {
 206                         for_lookup_table_entry(table,
 207                                                do_free_lookup_table_entry,
 208                                                NULL);
 209                         FREE(table->array);
 210                 }
 211                 FREE(table);
 212         }
 213 }
 214
 215 /*
 216  * Inserts an entry into the lookup table.
 217  *
 218  * @table:      A pointer to the lookup table.
 219  * @lte:        A pointer to the entry to insert.
 220  */
 221 void
 222 lookup_table_insert(struct wim_lookup_table *table,
 223                     struct wim_lookup_table_entry *lte)
 224 {
 225         size_t i = lte->hash_short % table->capacity;
 226         hlist_add_head(&lte->hash_list, &table->array[i]);
 227
 228         /* XXX Make the table grow when too many entries have been inserted. */
 229         table->num_entries++;
 230 }
 231
 232 static void
 233 finalize_lte(struct wim_lookup_table_entry *lte)
 234 {
 235         #ifdef WITH_FUSE
 236         if (lte->resource_location == RESOURCE_IN_STAGING_FILE) {
 237                 unlink(lte->staging_file_name);
 238                 list_del(&lte->unhashed_list);
 239         }
 240         #endif
 241         free_lookup_table_entry(lte);
 242 }
 243
 244 /* Decrements the reference count for the lookup table entry @lte.  If its
 245  * reference count reaches 0, it is unlinked from the lookup table.  If,
 246  * furthermore, the entry has no opened file descriptors associated with it, the
 247  * entry is freed.  */
 248 void
 249 lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
 250                      struct wim_lookup_table *table)
 251 {
 252         wimlib_assert(lte != NULL);
 253         wimlib_assert(lte->refcnt != 0);
 254         if (--lte->refcnt == 0) {
 255                 if (lte->unhashed)
 256                         list_del(&lte->unhashed_list);
 257                 else
 258                         lookup_table_unlink(table, lte);
 259         #ifdef WITH_FUSE
 260                 if (lte->num_opened_fds == 0)
 261         #endif
 262                         finalize_lte(lte);
 263         }
 264 }
 265
 266 #ifdef WITH_FUSE
 267 void
 268 lte_decrement_num_opened_fds(struct wim_lookup_table_entry *lte)
 269 {
 270         if (lte->num_opened_fds != 0)
 271                 if (--lte->num_opened_fds == 0 && lte->refcnt == 0)
 272                         finalize_lte(lte);
 273 }
 274 #endif
 275
 276 /* Calls a function on all the entries in the WIM lookup table.  Stop early and
 277  * return nonzero if any call to the function returns nonzero. */
 278 int
 279 for_lookup_table_entry(struct wim_lookup_table *table,
 280                        int (*visitor)(struct wim_lookup_table_entry *, void *),
 281                        void *arg)
 282 {
 283         struct wim_lookup_table_entry *lte;
 284         struct hlist_node *pos, *tmp;
 285         int ret;
 286
 287         for (size_t i = 0; i < table->capacity; i++) {
 288                 hlist_for_each_entry_safe(lte, pos, tmp, &table->array[i],
 289                                           hash_list)
 290                 {
 291                         wimlib_assert2(!(lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA));
 292                         ret = visitor(lte, arg);
 293                         if (ret)
 294                                 return ret;
 295                 }
 296         }
 297         return 0;
 298 }
 299
 300 /* qsort() callback that sorts streams (represented by `struct
 301  * wim_lookup_table_entry's) into an order optimized for reading and writing.
 302  *
 303  * Sorting is done primarily by resource location, then secondarily by a
 304  * per-resource location order.  For example, resources in WIM files are sorted
 305  * primarily by part number, then secondarily by offset, as to implement optimal
 306  * reading of either a standalone or split WIM.  */
 307 static int
 308 cmp_streams_by_sequential_order(const void *p1, const void *p2)
 309 {
 310         const struct wim_lookup_table_entry *lte1, *lte2;
 311         int v;
 312         WIMStruct *wim1, *wim2;
 313         u64 offset1, offset2;
 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                 offset1 = lte1->rspec->offset_in_wim + lte1->offset_in_res;
 343                 offset2 = lte2->rspec->offset_in_wim + lte2->offset_in_res;
 344
 345                 if (offset1 < offset2)
 346                         return -1;
 347                 if (offset1 > offset2)
 348                         return 1;
 349                 return 0;
 350         case RESOURCE_IN_FILE_ON_DISK:
 351 #ifdef WITH_FUSE
 352         case RESOURCE_IN_STAGING_FILE:
 353 #endif
 354 #ifdef __WIN32__
 355         case RESOURCE_WIN32_ENCRYPTED:
 356 #endif
 357                 /* Compare files by path: just a heuristic that will place files
 358                  * in the same directory next to each other.  */
 359                 return tstrcmp(lte1->file_on_disk, lte2->file_on_disk);
 360 #ifdef WITH_NTFS_3G
 361         case RESOURCE_IN_NTFS_VOLUME:
 362                 return tstrcmp(lte1->ntfs_loc->path, lte2->ntfs_loc->path);
 363 #endif
 364         default:
 365                 /* No additional sorting order defined for this resource
 366                  * location (e.g. RESOURCE_IN_ATTACHED_BUFFER); simply compare
 367                  * everything equal to each other.  */
 368                 return 0;
 369         }
 370 }
 371
 372 int
 373 sort_stream_list_by_sequential_order(struct list_head *stream_list,
 374                                      size_t list_head_offset)
 375 {
 376         struct list_head *cur;
 377         struct wim_lookup_table_entry **array;
 378         size_t i;
 379         size_t array_size;
 380         size_t num_streams = 0;
 381
 382         list_for_each(cur, stream_list)
 383                 num_streams++;
 384
 385         array_size = num_streams * sizeof(array[0]);
 386         array = MALLOC(array_size);
 387         if (array == NULL)
 388                 return WIMLIB_ERR_NOMEM;
 389         cur = stream_list->next;
 390         for (i = 0; i < num_streams; i++) {
 391                 array[i] = (struct wim_lookup_table_entry*)((u8*)cur -
 392                                                             list_head_offset);
 393                 cur = cur->next;
 394         }
 395
 396         qsort(array, num_streams, sizeof(array[0]),
 397               cmp_streams_by_sequential_order);
 398
 399         INIT_LIST_HEAD(stream_list);
 400         for (i = 0; i < num_streams; i++) {
 401                 list_add_tail((struct list_head*)
 402                                ((u8*)array[i] + list_head_offset),
 403                               stream_list);
 404         }
 405         FREE(array);
 406         return 0;
 407 }
 408
 409
 410 static int
 411 add_lte_to_array(struct wim_lookup_table_entry *lte,
 412                  void *_pp)
 413 {
 414         struct wim_lookup_table_entry ***pp = _pp;
 415         *(*pp)++ = lte;
 416         return 0;
 417 }
 418
 419 /* Iterate through the lookup table entries, but first sort them by stream
 420  * offset in the WIM.  Caution: this is intended to be used when the stream
 421  * offset field has actually been set. */
 422 int
 423 for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
 424                                   int (*visitor)(struct wim_lookup_table_entry *,
 425                                                  void *),
 426                                   void *arg)
 427 {
 428         struct wim_lookup_table_entry **lte_array, **p;
 429         size_t num_streams = table->num_entries;
 430         int ret;
 431
 432         lte_array = MALLOC(num_streams * sizeof(lte_array[0]));
 433         if (!lte_array)
 434                 return WIMLIB_ERR_NOMEM;
 435         p = lte_array;
 436         for_lookup_table_entry(table, add_lte_to_array, &p);
 437
 438         wimlib_assert(p == lte_array + num_streams);
 439
 440         qsort(lte_array, num_streams, sizeof(lte_array[0]),
 441               cmp_streams_by_sequential_order);
 442         ret = 0;
 443         for (size_t i = 0; i < num_streams; i++) {
 444                 ret = visitor(lte_array[i], arg);
 445                 if (ret)
 446                         break;
 447         }
 448         FREE(lte_array);
 449         return ret;
 450 }
 451
 452 /* On-disk format of a WIM lookup table entry (stream entry). */
 453 struct wim_lookup_table_entry_disk {
 454         /* Size, offset, and flags of the stream.  */
 455         struct wim_reshdr_disk reshdr;
 456
 457         /* Which part of the split WIM this stream is in; indexed from 1. */
 458         le16 part_number;
 459
 460         /* Reference count of this stream over all WIM images. */
 461         le32 refcnt;
 462
 463         /* SHA1 message digest of the uncompressed data of this stream, or
 464          * optionally all zeroes if this stream is of zero length. */
 465         u8 hash[SHA1_HASH_SIZE];
 466 } _packed_attribute;
 467
 468 #define WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE 50
 469
 470 /* Validate the size and location of a WIM resource.  */
 471 static int
 472 validate_resource(const struct wim_resource_spec *rspec)
 473 {
 474         struct wim_lookup_table_entry *lte;
 475         u64 cur_offset;
 476
 477         /* Verify that calculating the offset of the end of the resource doesn't
 478          * overflow.  */
 479         if (rspec->offset_in_wim + rspec->size_in_wim < rspec->size_in_wim)
 480                 goto invalid;
 481
 482         /* Verify that each stream in the resource has a valid offset and size,
 483          * and that no streams overlap.  */
 484         cur_offset = 0;
 485         list_for_each_entry(lte, &rspec->stream_list, wim_resource_list) {
 486                 if (lte->offset_in_res + lte->size < lte->size ||
 487                     lte->offset_in_res + lte->size > rspec->uncompressed_size ||
 488                     lte->offset_in_res < cur_offset)
 489                         goto invalid;
 490
 491                 cur_offset = lte->offset_in_res;
 492         }
 493         return 0;
 494
 495 invalid:
 496
 497         ERROR("Invalid resource entry!");
 498         return WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 499 }
 500
 501 /*
 502  * Reads the lookup table from a WIM file.  Each entry specifies a stream that
 503  * the WIM file contains, along with its location and SHA1 message digest.
 504  *
 505  * Saves lookup table entries for non-metadata streams in a hash table, and
 506  * saves the metadata entry for each image in a special per-image location (the
 507  * image_metadata array).
 508  *
 509  * Return values:
 510  *      WIMLIB_ERR_SUCCESS (0)
 511  *      WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY
 512  *      WIMLIB_ERR_RESOURCE_NOT_FOUND
 513  *
 514  *      Or an error code caused by failure to read the lookup table into memory.
 515  */
 516 int
 517 read_wim_lookup_table(WIMStruct *wim)
 518 {
 519         int ret;
 520         size_t i;
 521         size_t num_entries;
 522         struct wim_lookup_table *table;
 523         struct wim_lookup_table_entry *cur_entry, *duplicate_entry;
 524         struct wim_resource_spec *cur_rspec;
 525         void *buf;
 526
 527         DEBUG("Reading lookup table.");
 528
 529         /* Sanity check: lookup table entries are 50 bytes each.  */
 530         BUILD_BUG_ON(sizeof(struct wim_lookup_table_entry_disk) !=
 531                      WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE);
 532
 533         /* Calculate number of entries in the lookup table.  */
 534         num_entries = wim->hdr.lookup_table_reshdr.uncompressed_size /
 535                       sizeof(struct wim_lookup_table_entry_disk);
 536
 537         /* Read the lookup table into a buffer.  */
 538         ret = wim_reshdr_to_data(&wim->hdr.lookup_table_reshdr, wim, &buf);
 539         if (ret)
 540                 goto out;
 541
 542         /* Allocate a hash table to map SHA1 message digests into stream
 543          * specifications.  This is the in-memory "lookup table".  */
 544         table = new_lookup_table(num_entries * 2 + 1);
 545         if (table == NULL) {
 546                 ERROR("Not enough memory to read lookup table.");
 547                 ret = WIMLIB_ERR_NOMEM;
 548                 goto out_free_buf;
 549         }
 550
 551         /* Allocate and initalize stream entries from the raw lookup table
 552          * buffer.  */
 553         wim->current_image = 0;
 554         cur_rspec = NULL;
 555         for (i = 0; i < num_entries; i++) {
 556                 const struct wim_lookup_table_entry_disk *disk_entry =
 557                         &((const struct wim_lookup_table_entry_disk*)buf)[i];
 558                 u16 part_number;
 559                 struct wim_reshdr reshdr;
 560
 561                 get_wim_reshdr(&disk_entry->reshdr, &reshdr);
 562
 563                 DEBUG("reshdr: size_in_wim=%"PRIu64", "
 564                       "uncompressed_size=%"PRIu64", "
 565                       "offset_in_wim=%"PRIu64", "
 566                       "flags=0x%02x",
 567                       reshdr.size_in_wim, reshdr.uncompressed_size,
 568                       reshdr.offset_in_wim, reshdr.flags);
 569
 570                 cur_entry = new_lookup_table_entry();
 571                 if (cur_entry == NULL) {
 572                         ERROR("Not enough memory to read lookup table!");
 573                         ret = WIMLIB_ERR_NOMEM;
 574                         goto out_free_lookup_table;
 575                 }
 576
 577                 part_number = le16_to_cpu(disk_entry->part_number);
 578                 cur_entry->refcnt = le32_to_cpu(disk_entry->refcnt);
 579                 copy_hash(cur_entry->hash, disk_entry->hash);
 580
 581                 if (part_number != wim->hdr.part_number) {
 582                         WARNING("A lookup table entry in part %hu of the WIM "
 583                                 "points to part %hu (ignoring it)",
 584                                 wim->hdr.part_number, part_number);
 585                         free_lookup_table_entry(cur_entry);
 586                         continue;
 587                 }
 588
 589                 if (!(reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) ||
 590                     cur_rspec == NULL)
 591                 {
 592                         /* Starting new run of streams that share the same WIM
 593                          * resource.  */
 594                         if (cur_rspec != NULL) {
 595                                 ret = validate_resource(cur_rspec);
 596                                 if (ret)
 597                                         goto out_free_cur_entry;
 598                         }
 599
 600                         /* Allocate the resource specification and initialize it
 601                          * with values from the current stream entry.  */
 602                         cur_rspec = MALLOC(sizeof(*cur_rspec));
 603                         if (cur_rspec == NULL) {
 604                                 ERROR("Not enough memory to read lookup table!");
 605                                 ret = WIMLIB_ERR_NOMEM;
 606                                 goto out_free_cur_entry;
 607                         }
 608                         wim_res_hdr_to_spec(&reshdr, wim, cur_rspec);
 609
 610                         /* If this is a packed run, the current stream entry may
 611                          * specify a stream within the resource, and not the
 612                          * resource itself.  Zero possibly irrelevant data until
 613                          * it is read for certain.  */
 614                         if (reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
 615                                 cur_rspec->size_in_wim = 0;
 616                                 cur_rspec->uncompressed_size = 0;
 617                                 cur_rspec->flags = 0;
 618                         }
 619                 }
 620
 621                 if (is_zero_hash(cur_entry->hash)) {
 622                         if (reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
 623                                 /* Found the specification for the packed resource.
 624                                  * Transfer the values to the `struct
 625                                  * wim_resource_spec', and discard the current stream
 626                                  * since this lookup table entry did not, in fact,
 627                                  * correspond to a "stream".  */
 628                                 cur_rspec->offset_in_wim = reshdr.offset_in_wim;
 629                                 cur_rspec->size_in_wim = reshdr.size_in_wim;
 630                                 cur_rspec->flags = reshdr.flags;
 631                                 DEBUG("Full run is %"PRIu64" compressed bytes "
 632                                       "at file offset %"PRIu64" (flags 0x%02x)",
 633                                       cur_rspec->size_in_wim,
 634                                       cur_rspec->offset_in_wim,
 635                                       cur_rspec->flags);
 636                         }
 637                         free_lookup_table_entry(cur_entry);
 638                         continue;
 639                 }
 640
 641                 if (reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
 642                         /* Continuing the run with another stream.  */
 643                         DEBUG("Continuing packed run with stream: "
 644                               "%"PRIu64" uncompressed bytes @ resource offset %"PRIu64")",
 645                               reshdr.size_in_wim, reshdr.offset_in_wim);
 646                         cur_rspec->uncompressed_size += reshdr.size_in_wim;
 647                 }
 648
 649                 lte_bind_wim_resource_spec(cur_entry, cur_rspec);
 650                 if (reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
 651                         /* In packed runs, the offset field is used for
 652                          * in-resource offset, not the in-WIM offset, and the
 653                          * size field is used for the uncompressed size, not the
 654                          * compressed size.  */
 655                         cur_entry->offset_in_res = reshdr.offset_in_wim;
 656                         cur_entry->size = reshdr.size_in_wim;
 657                         cur_entry->flags = reshdr.flags;
 658                 } else {
 659                         /* Normal case: The stream corresponds one-to-one with
 660                          * the resource entry.  */
 661                         cur_entry->offset_in_res = 0;
 662                         cur_entry->size = reshdr.uncompressed_size;
 663                         cur_entry->flags = reshdr.flags;
 664                         cur_rspec = NULL;
 665                 }
 666
 667                 if (cur_entry->flags & WIM_RESHDR_FLAG_METADATA) {
 668                         /* Lookup table entry for a metadata resource */
 669                         if (cur_entry->refcnt != 1) {
 670                                 /* Metadata entries with no references must be
 671                                  * ignored.  See for example the WinPE WIMs from
 672                                  * WAIK v2.1.  */
 673                                 if (cur_entry->refcnt == 0) {
 674                                         free_lookup_table_entry(cur_entry);
 675                                         continue;
 676                                 }
 677                                 if (wimlib_print_errors) {
 678                                         ERROR("Found metadata resource with refcnt != 1:");
 679                                         print_lookup_table_entry(cur_entry, stderr);
 680                                 }
 681                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 682                                 goto out_free_cur_entry;
 683                         }
 684
 685                         if (wim->hdr.part_number != 1) {
 686                                 WARNING("Ignoring metadata resource found in a "
 687                                         "non-first part of the split WIM");
 688                                 free_lookup_table_entry(cur_entry);
 689                                 continue;
 690                         }
 691                         if (wim->current_image == wim->hdr.image_count) {
 692                                 WARNING("The WIM header says there are %u images "
 693                                         "in the WIM, but we found more metadata "
 694                                         "resources than this (ignoring the extra)",
 695                                         wim->hdr.image_count);
 696                                 free_lookup_table_entry(cur_entry);
 697                                 continue;
 698                         }
 699
 700                         /* Notice very carefully:  We are assigning the metadata
 701                          * resources in the exact order mirrored by their lookup
 702                          * table entries on disk, which is the behavior of
 703                          * Microsoft's software.  In particular, this overrides
 704                          * the actual locations of the metadata resources
 705                          * themselves in the WIM file as well as any information
 706                          * written in the XML data. */
 707                         DEBUG("Found metadata resource for image %u at "
 708                               "offset %"PRIu64".",
 709                               wim->current_image + 1,
 710                               cur_entry->rspec->offset_in_wim);
 711                         wim->image_metadata[
 712                                 wim->current_image++]->metadata_lte = cur_entry;
 713                         continue;
 714                 }
 715
 716                 /* Lookup table entry for a stream that is not a metadata
 717                  * resource.  */
 718                 duplicate_entry = lookup_resource(table, cur_entry->hash);
 719                 if (duplicate_entry) {
 720                         if (wimlib_print_errors) {
 721                                 WARNING("The WIM lookup table contains two entries with the "
 722                                       "same SHA1 message digest!");
 723                                 WARNING("The first entry is:");
 724                                 print_lookup_table_entry(duplicate_entry, stderr);
 725                                 WARNING("The second entry is:");
 726                                 print_lookup_table_entry(cur_entry, stderr);
 727                         }
 728                         free_lookup_table_entry(cur_entry);
 729                         continue;
 730                 }
 731
 732                 /* Finally, insert the stream into the lookup table, keyed by
 733                  * its SHA1 message digest.  */
 734                 lookup_table_insert(table, cur_entry);
 735         }
 736
 737         /* Validate the last resource.  */
 738         if (cur_rspec != NULL) {
 739                 ret = validate_resource(cur_rspec);
 740                 if (ret)
 741                         goto out_free_cur_entry;
 742         }
 743
 744         if (wim->hdr.part_number == 1 && wim->current_image != wim->hdr.image_count) {
 745                 WARNING("The header of \"%"TS"\" says there are %u images in\n"
 746                         "          the WIM, but we only found %d metadata resources!  Acting as if\n"
 747                         "          the header specified only %d images instead.",
 748                         wim->filename, wim->hdr.image_count,
 749                         wim->current_image, wim->current_image);
 750                 for (int i = wim->current_image; i < wim->hdr.image_count; i++)
 751                         put_image_metadata(wim->image_metadata[i], NULL);
 752                 wim->hdr.image_count = wim->current_image;
 753         }
 754         DEBUG("Done reading lookup table.");
 755         wim->lookup_table = table;
 756         ret = 0;
 757         goto out_free_buf;
 758
 759 out_free_cur_entry:
 760         FREE(cur_entry);
 761 out_free_lookup_table:
 762         free_lookup_table(table);
 763 out_free_buf:
 764         FREE(buf);
 765 out:
 766         wim->current_image = 0;
 767         return ret;
 768 }
 769
 770
 771 static void
 772 write_wim_lookup_table_entry(const struct wim_lookup_table_entry *lte,
 773                              struct wim_lookup_table_entry_disk *disk_entry,
 774                              u16 part_number)
 775 {
 776         put_wim_reshdr(&lte->out_reshdr, &disk_entry->reshdr);
 777         disk_entry->part_number = cpu_to_le16(part_number);
 778         disk_entry->refcnt = cpu_to_le32(lte->out_refcnt);
 779         copy_hash(disk_entry->hash, lte->hash);
 780 }
 781
 782 static int
 783 write_wim_lookup_table_from_stream_list(struct list_head *stream_list,
 784                                         struct filedes *out_fd,
 785                                         u16 part_number,
 786                                         struct wim_reshdr *out_reshdr,
 787                                         int write_resource_flags,
 788                                         struct wimlib_lzx_context **comp_ctx)
 789 {
 790         size_t table_size;
 791         struct wim_lookup_table_entry *lte;
 792         struct wim_lookup_table_entry_disk *table_buf;
 793         struct wim_lookup_table_entry_disk *table_buf_ptr;
 794         int ret;
 795
 796         table_size = 0;
 797         list_for_each_entry(lte, stream_list, lookup_table_list)
 798                 table_size += sizeof(struct wim_lookup_table_entry_disk);
 799
 800         DEBUG("Writing WIM lookup table (size=%zu, offset=%"PRIu64")",
 801               table_size, out_fd->offset);
 802
 803         table_buf = MALLOC(table_size);
 804         if (!table_buf) {
 805                 ERROR("Failed to allocate %zu bytes for temporary lookup table",
 806                       table_size);
 807                 return WIMLIB_ERR_NOMEM;
 808         }
 809         table_buf_ptr = table_buf;
 810         list_for_each_entry(lte, stream_list, lookup_table_list)
 811                 write_wim_lookup_table_entry(lte, table_buf_ptr++, part_number);
 812
 813         /* Write the lookup table uncompressed.  Although wimlib can handle a
 814          * compressed lookup table, MS software cannot.  */
 815         ret = write_wim_resource_from_buffer(table_buf,
 816                                              table_size,
 817                                              WIM_RESHDR_FLAG_METADATA,
 818                                              out_fd,
 819                                              WIMLIB_COMPRESSION_TYPE_NONE,
 820                                              0,
 821                                              out_reshdr,
 822                                              NULL,
 823                                              write_resource_flags,
 824                                              comp_ctx);
 825         FREE(table_buf);
 826         DEBUG("ret=%d", ret);
 827         return ret;
 828 }
 829
 830 static int
 831 append_lookup_table_entry(struct wim_lookup_table_entry *lte, void *_list)
 832 {
 833         /* Lookup table entries with 'out_refcnt' == 0 correspond to streams not
 834          * written and not present in the resulting WIM file, and should not be
 835          * included in the lookup table.
 836          *
 837          * Lookup table entries marked as filtered (EXTERNAL_WIM) with
 838          * 'out_refcnt != 0' were referenced as part of the logical write but
 839          * correspond to streams that were not in fact written, and should not
 840          * be included in the lookup table.
 841          *
 842          * Lookup table entries marked as filtered (SAME_WIM) with 'out_refcnt
 843          * != 0' were referenced as part of the logical write but correspond to
 844          * streams that were not in fact written, but nevertheless were already
 845          * present in the WIM being overwritten in-place.  These entries must be
 846          * included in the lookup table, and the resource information to write
 847          * needs to be copied from the resource information read originally.
 848          */
 849         if (lte->out_refcnt != 0 && !(lte->filtered & FILTERED_EXTERNAL_WIM)) {
 850                 if (lte->filtered & FILTERED_SAME_WIM)
 851                         wim_res_spec_to_hdr(lte->rspec, &lte->out_reshdr);
 852                 list_add_tail(&lte->lookup_table_list, (struct list_head*)_list);
 853         }
 854         return 0;
 855 }
 856
 857 int
 858 write_wim_lookup_table(WIMStruct *wim, int image, int write_flags,
 859                        struct wim_reshdr *out_reshdr,
 860                        struct list_head *stream_list_override)
 861 {
 862         int write_resource_flags;
 863         struct list_head _stream_list;
 864         struct list_head *stream_list;
 865
 866         if (stream_list_override) {
 867                 stream_list = stream_list_override;
 868         } else {
 869                 stream_list = &_stream_list;
 870                 INIT_LIST_HEAD(stream_list);
 871         }
 872
 873         if (!(write_flags & WIMLIB_WRITE_FLAG_NO_METADATA)) {
 874                 int start_image;
 875                 int end_image;
 876
 877                 if (image == WIMLIB_ALL_IMAGES) {
 878                         start_image = 1;
 879                         end_image = wim->hdr.image_count;
 880                 } else {
 881                         start_image = image;
 882                         end_image = image;
 883                 }
 884
 885                 /* Push metadata resource lookup table entries onto the front of
 886                  * the list in reverse order, so that they're written in order.
 887                  */
 888                 for (int i = end_image; i >= start_image; i--) {
 889                         struct wim_lookup_table_entry *metadata_lte;
 890
 891                         metadata_lte = wim->image_metadata[i - 1]->metadata_lte;
 892                         metadata_lte->out_refcnt = 1;
 893                         metadata_lte->out_reshdr.flags |= WIM_RESHDR_FLAG_METADATA;
 894                         list_add(&metadata_lte->lookup_table_list, stream_list);
 895                 }
 896         }
 897
 898         /* Append additional lookup table entries that need to be written, with
 899          * some special handling for streams that have been marked as filtered.
 900          */
 901         if (!stream_list_override) {
 902                 for_lookup_table_entry(wim->lookup_table,
 903                                        append_lookup_table_entry, stream_list);
 904         }
 905
 906         write_resource_flags = 0;
 907         if (write_flags & WIMLIB_WRITE_FLAG_PIPABLE)
 908                 write_resource_flags |= WIMLIB_WRITE_RESOURCE_FLAG_PIPABLE;
 909         return write_wim_lookup_table_from_stream_list(stream_list,
 910                                                        &wim->out_fd,
 911                                                        wim->hdr.part_number,
 912                                                        out_reshdr,
 913                                                        write_resource_flags,
 914                                                        &wim->lzx_context);
 915 }
 916
 917
 918 int
 919 lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 920 {
 921         lte->real_refcnt = 0;
 922         return 0;
 923 }
 924
 925 int
 926 lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 927 {
 928         lte->out_refcnt = 0;
 929         return 0;
 930 }
 931
 932 int
 933 lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *_ignore)
 934 {
 935         if (lte->extracted_file != NULL) {
 936                 FREE(lte->extracted_file);
 937                 lte->extracted_file = NULL;
 938         }
 939         return 0;
 940 }
 941
 942 void
 943 print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 944 {
 945         if (lte == NULL) {
 946                 tputc(T('\n'), out);
 947                 return;
 948         }
 949
 950
 951         tprintf(T("Uncompressed size     = %"PRIu64" bytes\n"),
 952                 lte->size);
 953         if (lte_is_partial(lte)) {
 954                 tprintf(T("Offset                = %"PRIu64" bytes\n"),
 955                         lte->offset_in_res);
 956
 957                 tprintf(T("Raw uncompressed size = %"PRIu64" bytes\n"),
 958                         lte->rspec->uncompressed_size);
 959
 960                 tprintf(T("Raw compressed size   = %"PRIu64" bytes\n"),
 961                         lte->rspec->size_in_wim);
 962
 963                 tprintf(T("Raw offset            = %"PRIu64" bytes\n"),
 964                         lte->rspec->offset_in_wim);
 965         } else if (lte->resource_location == RESOURCE_IN_WIM) {
 966                 tprintf(T("Compressed size       = %"PRIu64" bytes\n"),
 967                         lte->rspec->size_in_wim);
 968
 969                 tprintf(T("Offset                = %"PRIu64" bytes\n"),
 970                         lte->rspec->offset_in_wim);
 971         }
 972
 973         tfprintf(out, T("Reference Count       = %u\n"), lte->refcnt);
 974
 975         if (lte->unhashed) {
 976                 tfprintf(out, T("(Unhashed: inode %p, stream_id = %u)\n"),
 977                          lte->back_inode, lte->back_stream_id);
 978         } else {
 979                 tfprintf(out, T("Hash                  = 0x"));
 980                 print_hash(lte->hash, out);
 981                 tputc(T('\n'), out);
 982         }
 983
 984         tfprintf(out, T("Flags                 = "));
 985         u8 flags = lte->flags;
 986         if (flags & WIM_RESHDR_FLAG_COMPRESSED)
 987                 tfputs(T("WIM_RESHDR_FLAG_COMPRESSED, "), out);
 988         if (flags & WIM_RESHDR_FLAG_FREE)
 989                 tfputs(T("WIM_RESHDR_FLAG_FREE, "), out);
 990         if (flags & WIM_RESHDR_FLAG_METADATA)
 991                 tfputs(T("WIM_RESHDR_FLAG_METADATA, "), out);
 992         if (flags & WIM_RESHDR_FLAG_SPANNED)
 993                 tfputs(T("WIM_RESHDR_FLAG_SPANNED, "), out);
 994         if (flags & WIM_RESHDR_FLAG_PACKED_STREAMS)
 995                 tfputs(T("WIM_RESHDR_FLAG_PACKED_STREAMS, "), out);
 996         tputc(T('\n'), out);
 997         switch (lte->resource_location) {
 998         case RESOURCE_IN_WIM:
 999                 if (lte->rspec->wim->filename) {
1000                         tfprintf(out, T("WIM file              = `%"TS"'\n"),
1001                                  lte->rspec->wim->filename);
1002                 }
1003                 break;
1004 #ifdef __WIN32__
1005         case RESOURCE_WIN32_ENCRYPTED:
1006 #endif
1007         case RESOURCE_IN_FILE_ON_DISK:
1008                 tfprintf(out, T("File on Disk          = `%"TS"'\n"),
1009                          lte->file_on_disk);
1010                 break;
1011 #ifdef WITH_FUSE
1012         case RESOURCE_IN_STAGING_FILE:
1013                 tfprintf(out, T("Staging File          = `%"TS"'\n"),
1014                                 lte->staging_file_name);
1015                 break;
1016 #endif
1017         default:
1018                 break;
1019         }
1020         tputc(T('\n'), out);
1021 }
1022
1023 void
1024 lte_to_wimlib_resource_entry(const struct wim_lookup_table_entry *lte,
1025                              struct wimlib_resource_entry *wentry)
1026 {
1027         memset(wentry, 0, sizeof(*wentry));
1028
1029         wentry->uncompressed_size = lte->size;
1030         if (lte->resource_location == RESOURCE_IN_WIM) {
1031                 wentry->part_number = lte->rspec->wim->hdr.part_number;
1032                 if (lte_is_partial(lte)) {
1033                         wentry->compressed_size = 0;
1034                         wentry->offset = lte->offset_in_res;
1035                 } else {
1036                         wentry->compressed_size = lte->rspec->size_in_wim;
1037                         wentry->offset = lte->rspec->offset_in_wim;
1038                 }
1039                 wentry->raw_resource_offset_in_wim = lte->rspec->offset_in_wim;
1040                 wentry->raw_resource_uncompressed_size = lte->rspec->uncompressed_size;
1041                 wentry->raw_resource_compressed_size = lte->rspec->size_in_wim;
1042         }
1043         copy_hash(wentry->sha1_hash, lte->hash);
1044         wentry->reference_count = lte->refcnt;
1045         wentry->is_compressed = (lte->flags & WIM_RESHDR_FLAG_COMPRESSED) != 0;
1046         wentry->is_metadata = (lte->flags & WIM_RESHDR_FLAG_METADATA) != 0;
1047         wentry->is_free = (lte->flags & WIM_RESHDR_FLAG_FREE) != 0;
1048         wentry->is_spanned = (lte->flags & WIM_RESHDR_FLAG_SPANNED) != 0;
1049         wentry->is_partial = lte_is_partial(lte);
1050 }
1051
1052 struct iterate_lte_context {
1053         wimlib_iterate_lookup_table_callback_t cb;
1054         void *user_ctx;
1055 };
1056
1057 static int
1058 do_iterate_lte(struct wim_lookup_table_entry *lte, void *_ctx)
1059 {
1060         struct iterate_lte_context *ctx = _ctx;
1061         struct wimlib_resource_entry entry;
1062
1063         lte_to_wimlib_resource_entry(lte, &entry);
1064         return (*ctx->cb)(&entry, ctx->user_ctx);
1065 }
1066
1067 /* API function documented in wimlib.h  */
1068 WIMLIBAPI int
1069 wimlib_iterate_lookup_table(WIMStruct *wim, int flags,
1070                             wimlib_iterate_lookup_table_callback_t cb,
1071                             void *user_ctx)
1072 {
1073         struct iterate_lte_context ctx = {
1074                 .cb = cb,
1075                 .user_ctx = user_ctx,
1076         };
1077         if (wim->hdr.part_number == 1) {
1078                 int ret;
1079                 for (int i = 0; i < wim->hdr.image_count; i++) {
1080                         ret = do_iterate_lte(wim->image_metadata[i]->metadata_lte,
1081                                              &ctx);
1082                         if (ret)
1083                                 return ret;
1084                 }
1085         }
1086         return for_lookup_table_entry(wim->lookup_table, do_iterate_lte, &ctx);
1087 }
1088
1089 /* Given a SHA1 message digest, return the corresponding entry in the WIM's
1090  * lookup table, or NULL if there is none.  */
1091 struct wim_lookup_table_entry *
1092 lookup_resource(const struct wim_lookup_table *table, const u8 hash[])
1093 {
1094         size_t i;
1095         struct wim_lookup_table_entry *lte;
1096         struct hlist_node *pos;
1097
1098         wimlib_assert(table != NULL);
1099         wimlib_assert(hash != NULL);
1100
1101         i = *(size_t*)hash % table->capacity;
1102         hlist_for_each_entry(lte, pos, &table->array[i], hash_list)
1103                 if (hashes_equal(hash, lte->hash))
1104                         return lte;
1105         return NULL;
1106 }
1107
1108 #ifdef WITH_FUSE
1109 /*
1110  * Finds the dentry, lookup table entry, and stream index for a WIM file stream,
1111  * given a path name.
1112  *
1113  * This is only for pre-resolved inodes.
1114  */
1115 int
1116 wim_pathname_to_stream(WIMStruct *wim,
1117                        const tchar *path,
1118                        int lookup_flags,
1119                        struct wim_dentry **dentry_ret,
1120                        struct wim_lookup_table_entry **lte_ret,
1121                        u16 *stream_idx_ret)
1122 {
1123         struct wim_dentry *dentry;
1124         struct wim_lookup_table_entry *lte;
1125         u16 stream_idx;
1126         const tchar *stream_name = NULL;
1127         struct wim_inode *inode;
1128         tchar *p = NULL;
1129
1130         if (lookup_flags & LOOKUP_FLAG_ADS_OK) {
1131                 stream_name = path_stream_name(path);
1132                 if (stream_name) {
1133                         p = (tchar*)stream_name - 1;
1134                         *p = T('\0');
1135                 }
1136         }
1137
1138         dentry = get_dentry(wim, path);
1139         if (p)
1140                 *p = T(':');
1141         if (!dentry)
1142                 return -errno;
1143
1144         inode = dentry->d_inode;
1145
1146         if (!inode->i_resolved)
1147                 if (inode_resolve_ltes(inode, wim->lookup_table, false))
1148                         return -EIO;
1149
1150         if (!(lookup_flags & LOOKUP_FLAG_DIRECTORY_OK)
1151               && inode_is_directory(inode))
1152                 return -EISDIR;
1153
1154         if (stream_name) {
1155                 struct wim_ads_entry *ads_entry;
1156                 u16 ads_idx;
1157                 ads_entry = inode_get_ads_entry(inode, stream_name,
1158                                                 &ads_idx);
1159                 if (ads_entry) {
1160                         stream_idx = ads_idx + 1;
1161                         lte = ads_entry->lte;
1162                         goto out;
1163                 } else {
1164                         return -ENOENT;
1165                 }
1166         } else {
1167                 lte = inode_unnamed_stream_resolved(inode, &stream_idx);
1168         }
1169 out:
1170         if (dentry_ret)
1171                 *dentry_ret = dentry;
1172         if (lte_ret)
1173                 *lte_ret = lte;
1174         if (stream_idx_ret)
1175                 *stream_idx_ret = stream_idx;
1176         return 0;
1177 }
1178 #endif
1179
1180 int
1181 resource_not_found_error(const struct wim_inode *inode, const u8 *hash)
1182 {
1183         if (wimlib_print_errors) {
1184                 ERROR("\"%"TS"\": resource not found", inode_first_full_path(inode));
1185                 tfprintf(stderr, T("        SHA-1 message digest of missing resource:\n        "));
1186                 print_hash(hash, stderr);
1187                 tputc(T('\n'), stderr);
1188         }
1189         return WIMLIB_ERR_RESOURCE_NOT_FOUND;
1190 }
1191
1192 /*
1193  * Resolve an inode's lookup table entries.
1194  *
1195  * This replaces the SHA1 hash fields (which are used to lookup an entry in the
1196  * lookup table) with pointers directly to the lookup table entries.
1197  *
1198  * If @force is %false:
1199  *      If any needed SHA1 message digests are not found in the lookup table,
1200  *      WIMLIB_ERR_RESOURCE_NOT_FOUND is returned and the inode is left
1201  *      unmodified.
1202  * If @force is %true:
1203  *      If any needed SHA1 message digests are not found in the lookup table,
1204  *      new entries are allocated and inserted into the lookup table.
1205  */
1206 int
1207 inode_resolve_ltes(struct wim_inode *inode, struct wim_lookup_table *table,
1208                    bool force)
1209 {
1210         const u8 *hash;
1211
1212         if (!inode->i_resolved) {
1213                 struct wim_lookup_table_entry *lte, *ads_lte;
1214
1215                 /* Resolve the default file stream */
1216                 lte = NULL;
1217                 hash = inode->i_hash;
1218                 if (!is_zero_hash(hash)) {
1219                         lte = lookup_resource(table, hash);
1220                         if (!lte) {
1221                                 if (force) {
1222                                         lte = new_lookup_table_entry();
1223                                         if (!lte)
1224                                                 return WIMLIB_ERR_NOMEM;
1225                                         copy_hash(lte->hash, hash);
1226                                         lookup_table_insert(table, lte);
1227                                 } else {
1228                                         goto resource_not_found;
1229                                 }
1230                         }
1231                 }
1232
1233                 /* Resolve the alternate data streams */
1234                 struct wim_lookup_table_entry *ads_ltes[inode->i_num_ads];
1235                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1236                         struct wim_ads_entry *cur_entry;
1237
1238                         ads_lte = NULL;
1239                         cur_entry = &inode->i_ads_entries[i];
1240                         hash = cur_entry->hash;
1241                         if (!is_zero_hash(hash)) {
1242                                 ads_lte = lookup_resource(table, hash);
1243                                 if (!ads_lte) {
1244                                         if (force) {
1245                                                 ads_lte = new_lookup_table_entry();
1246                                                 if (!ads_lte)
1247                                                         return WIMLIB_ERR_NOMEM;
1248                                                 copy_hash(ads_lte->hash, hash);
1249                                                 lookup_table_insert(table, ads_lte);
1250                                         } else {
1251                                                 goto resource_not_found;
1252                                         }
1253                                 }
1254                         }
1255                         ads_ltes[i] = ads_lte;
1256                 }
1257                 inode->i_lte = lte;
1258                 for (u16 i = 0; i < inode->i_num_ads; i++)
1259                         inode->i_ads_entries[i].lte = ads_ltes[i];
1260                 inode->i_resolved = 1;
1261         }
1262         return 0;
1263
1264 resource_not_found:
1265         return resource_not_found_error(inode, hash);
1266 }
1267
1268 void
1269 inode_unresolve_ltes(struct wim_inode *inode)
1270 {
1271         if (inode->i_resolved) {
1272                 if (inode->i_lte)
1273                         copy_hash(inode->i_hash, inode->i_lte->hash);
1274                 else
1275                         zero_out_hash(inode->i_hash);
1276
1277                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1278                         if (inode->i_ads_entries[i].lte)
1279                                 copy_hash(inode->i_ads_entries[i].hash,
1280                                           inode->i_ads_entries[i].lte->hash);
1281                         else
1282                                 zero_out_hash(inode->i_ads_entries[i].hash);
1283                 }
1284                 inode->i_resolved = 0;
1285         }
1286 }
1287
1288 /*
1289  * Returns the lookup table entry for stream @stream_idx of the inode, where
1290  * stream_idx = 0 means the default un-named file stream, and stream_idx >= 1
1291  * corresponds to an alternate data stream.
1292  *
1293  * This works for both resolved and un-resolved inodes.
1294  */
1295 struct wim_lookup_table_entry *
1296 inode_stream_lte(const struct wim_inode *inode, unsigned stream_idx,
1297                  const struct wim_lookup_table *table)
1298 {
1299         if (inode->i_resolved)
1300                 return inode_stream_lte_resolved(inode, stream_idx);
1301         else
1302                 return inode_stream_lte_unresolved(inode, stream_idx, table);
1303 }
1304
1305 struct wim_lookup_table_entry *
1306 inode_unnamed_stream_resolved(const struct wim_inode *inode, u16 *stream_idx_ret)
1307 {
1308         wimlib_assert(inode->i_resolved);
1309         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
1310                 if (inode_stream_name_nbytes(inode, i) == 0 &&
1311                     !is_zero_hash(inode_stream_hash_resolved(inode, i)))
1312                 {
1313                         *stream_idx_ret = i;
1314                         return inode_stream_lte_resolved(inode, i);
1315                 }
1316         }
1317         *stream_idx_ret = 0;
1318         return NULL;
1319 }
1320
1321 struct wim_lookup_table_entry *
1322 inode_unnamed_lte_resolved(const struct wim_inode *inode)
1323 {
1324         u16 stream_idx;
1325         return inode_unnamed_stream_resolved(inode, &stream_idx);
1326 }
1327
1328 struct wim_lookup_table_entry *
1329 inode_unnamed_lte_unresolved(const struct wim_inode *inode,
1330                              const struct wim_lookup_table *table)
1331 {
1332         wimlib_assert(!inode->i_resolved);
1333         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
1334                 if (inode_stream_name_nbytes(inode, i) == 0 &&
1335                     !is_zero_hash(inode_stream_hash_unresolved(inode, i)))
1336                 {
1337                         return inode_stream_lte_unresolved(inode, i, table);
1338                 }
1339         }
1340         return NULL;
1341 }
1342
1343 /* Return the lookup table entry for the unnamed data stream of an inode, or
1344  * NULL if there is none.
1345  *
1346  * You'd think this would be easier than it actually is, since the unnamed data
1347  * stream should be the one referenced from the inode itself.  Alas, if there
1348  * are named data streams, Microsoft's "imagex.exe" program will put the unnamed
1349  * data stream in one of the alternate data streams instead of inside the WIM
1350  * dentry itself.  So we need to check the alternate data streams too.
1351  *
1352  * Also, note that a dentry may appear to have more than one unnamed stream, but
1353  * if the SHA1 message digest is all 0's then the corresponding stream does not
1354  * really "count" (this is the case for the inode's own file stream when the
1355  * file stream that should be there is actually in one of the alternate stream
1356  * entries.).  This is despite the fact that we may need to extract such a
1357  * missing entry as an empty file or empty named data stream.
1358  */
1359 struct wim_lookup_table_entry *
1360 inode_unnamed_lte(const struct wim_inode *inode,
1361                   const struct wim_lookup_table *table)
1362 {
1363         if (inode->i_resolved)
1364                 return inode_unnamed_lte_resolved(inode);
1365         else
1366                 return inode_unnamed_lte_unresolved(inode, table);
1367 }
1368
1369 /* Returns the SHA1 message digest of the unnamed data stream of a WIM inode, or
1370  * 'zero_hash' if the unnamed data stream is missing has all zeroes in its SHA1
1371  * message digest field.  */
1372 const u8 *
1373 inode_unnamed_stream_hash(const struct wim_inode *inode)
1374 {
1375         const u8 *hash;
1376
1377         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
1378                 if (inode_stream_name_nbytes(inode, i) == 0) {
1379                         hash = inode_stream_hash(inode, i);
1380                         if (!is_zero_hash(hash))
1381                                 return hash;
1382                 }
1383         }
1384         return zero_hash;
1385 }
1386
1387 struct wim_lookup_table_entry **
1388 retrieve_lte_pointer(struct wim_lookup_table_entry *lte)
1389 {
1390         wimlib_assert(lte->unhashed);
1391         struct wim_inode *inode = lte->back_inode;
1392         u32 stream_id = lte->back_stream_id;
1393         if (stream_id == 0)
1394                 return &inode->i_lte;
1395         else
1396                 for (u16 i = 0; i < inode->i_num_ads; i++)
1397                         if (inode->i_ads_entries[i].stream_id == stream_id)
1398                                 return &inode->i_ads_entries[i].lte;
1399         wimlib_assert(0);
1400         return NULL;
1401 }
1402
1403 /* Calculate the SHA1 message digest of a stream and move it from the list of
1404  * unhashed streams to the stream lookup table, possibly joining it with an
1405  * existing lookup table entry for an identical stream.
1406  *
1407  * @lte:  An unhashed lookup table entry.
1408  * @lookup_table:  Lookup table for the WIM.
1409  * @lte_ret:  On success, write a pointer to the resulting lookup table
1410  *            entry to this location.  This will be the same as @lte
1411  *            if it was inserted into the lookup table, or different if
1412  *            a duplicate stream was found.
1413  *
1414  * Returns 0 on success; nonzero if there is an error reading the stream.
1415  */
1416 int
1417 hash_unhashed_stream(struct wim_lookup_table_entry *lte,
1418                      struct wim_lookup_table *lookup_table,
1419                      struct wim_lookup_table_entry **lte_ret)
1420 {
1421         int ret;
1422         struct wim_lookup_table_entry *duplicate_lte;
1423         struct wim_lookup_table_entry **back_ptr;
1424
1425         wimlib_assert(lte->unhashed);
1426
1427         /* back_ptr must be saved because @back_inode and @back_stream_id are in
1428          * union with the SHA1 message digest and will no longer be valid once
1429          * the SHA1 has been calculated. */
1430         back_ptr = retrieve_lte_pointer(lte);
1431
1432         ret = sha1_stream(lte);
1433         if (ret)
1434                 return ret;
1435
1436         /* Look for a duplicate stream */
1437         duplicate_lte = lookup_resource(lookup_table, lte->hash);
1438         list_del(&lte->unhashed_list);
1439         if (duplicate_lte) {
1440                 /* We have a duplicate stream.  Transfer the reference counts
1441                  * from this stream to the duplicate, update the reference to
1442                  * this stream (in an inode or ads_entry) to point to the
1443                  * duplicate, then free this stream. */
1444                 wimlib_assert(!(duplicate_lte->unhashed));
1445                 duplicate_lte->refcnt += lte->refcnt;
1446                 duplicate_lte->out_refcnt += lte->out_refcnt;
1447                 *back_ptr = duplicate_lte;
1448                 free_lookup_table_entry(lte);
1449                 lte = duplicate_lte;
1450         } else {
1451                 /* No duplicate stream, so we need to insert
1452                  * this stream into the lookup table and treat
1453                  * it as a hashed stream. */
1454                 lookup_table_insert(lookup_table, lte);
1455                 lte->unhashed = 0;
1456         }
1457         if (lte_ret)
1458                 *lte_ret = lte;
1459         return 0;
1460 }
1461
1462 static int
1463 lte_clone_if_new(struct wim_lookup_table_entry *lte, void *_lookup_table)
1464 {
1465         struct wim_lookup_table *lookup_table = _lookup_table;
1466
1467         if (lookup_resource(lookup_table, lte->hash))
1468                 return 0;  /*  Resource already present.  */
1469
1470         lte = clone_lookup_table_entry(lte);
1471         if (!lte)
1472                 return WIMLIB_ERR_NOMEM;
1473         lte->out_refcnt = 1;
1474         lookup_table_insert(lookup_table, lte);
1475         return 0;
1476 }
1477
1478 static int
1479 lte_delete_if_new(struct wim_lookup_table_entry *lte, void *_lookup_table)
1480 {
1481         struct wim_lookup_table *lookup_table = _lookup_table;
1482
1483         if (lte->out_refcnt) {
1484                 lookup_table_unlink(lookup_table, lte);
1485                 free_lookup_table_entry(lte);
1486         }
1487         return 0;
1488 }
1489
1490 /* API function documented in wimlib.h  */
1491 WIMLIBAPI int
1492 wimlib_reference_resources(WIMStruct *wim,
1493                            WIMStruct **resource_wims, unsigned num_resource_wims,
1494                            int ref_flags)
1495 {
1496         int ret;
1497         unsigned i;
1498
1499         if (wim == NULL)
1500                 return WIMLIB_ERR_INVALID_PARAM;
1501
1502         if (num_resource_wims != 0 && resource_wims == NULL)
1503                 return WIMLIB_ERR_INVALID_PARAM;
1504
1505         for (i = 0; i < num_resource_wims; i++)
1506                 if (resource_wims[i] == NULL)
1507                         return WIMLIB_ERR_INVALID_PARAM;
1508
1509         for_lookup_table_entry(wim->lookup_table, lte_zero_out_refcnt, NULL);
1510
1511         for (i = 0; i < num_resource_wims; i++) {
1512                 ret = for_lookup_table_entry(resource_wims[i]->lookup_table,
1513                                              lte_clone_if_new,
1514                                              wim->lookup_table);
1515                 if (ret)
1516                         goto out_rollback;
1517         }
1518         return 0;
1519
1520 out_rollback:
1521         for_lookup_table_entry(wim->lookup_table, lte_delete_if_new,
1522                                wim->lookup_table);
1523         return ret;
1524 }
1525
1526 static int
1527 reference_resource_paths(WIMStruct *wim,
1528                          const tchar * const *resource_wimfiles,
1529                          unsigned num_resource_wimfiles,
1530                          int ref_flags,
1531                          int open_flags,
1532                          wimlib_progress_func_t progress_func)
1533 {
1534         WIMStruct **resource_wims;
1535         unsigned i;
1536         int ret;
1537
1538         resource_wims = CALLOC(num_resource_wimfiles, sizeof(resource_wims[0]));
1539         if (!resource_wims)
1540                 return WIMLIB_ERR_NOMEM;
1541
1542         for (i = 0; i < num_resource_wimfiles; i++) {
1543                 DEBUG("Referencing resources from path \"%"TS"\"",
1544                       resource_wimfiles[i]);
1545                 ret = wimlib_open_wim(resource_wimfiles[i], open_flags,
1546                                       &resource_wims[i], progress_func);
1547                 if (ret)
1548                         goto out_free_resource_wims;
1549         }
1550
1551         ret = wimlib_reference_resources(wim, resource_wims,
1552                                          num_resource_wimfiles, ref_flags);
1553         if (ret)
1554                 goto out_free_resource_wims;
1555
1556         for (i = 0; i < num_resource_wimfiles; i++)
1557                 list_add_tail(&resource_wims[i]->subwim_node, &wim->subwims);
1558
1559         ret = 0;
1560         goto out_free_array;
1561
1562 out_free_resource_wims:
1563         for (i = 0; i < num_resource_wimfiles; i++)
1564                 wimlib_free(resource_wims[i]);
1565 out_free_array:
1566         FREE(resource_wims);
1567         return ret;
1568 }
1569
1570 static int
1571 reference_resource_glob(WIMStruct *wim, const tchar *refglob,
1572                         int ref_flags, int open_flags,
1573                         wimlib_progress_func_t progress_func)
1574 {
1575         glob_t globbuf;
1576         int ret;
1577
1578         /* Note: glob() is replaced in Windows native builds.  */
1579         ret = tglob(refglob, GLOB_ERR | GLOB_NOSORT, NULL, &globbuf);
1580         if (ret) {
1581                 if (ret == GLOB_NOMATCH) {
1582                         if (ref_flags & WIMLIB_REF_FLAG_GLOB_ERR_ON_NOMATCH) {
1583                                 ERROR("Found no files for glob \"%"TS"\"", refglob);
1584                                 return WIMLIB_ERR_GLOB_HAD_NO_MATCHES;
1585                         } else {
1586                                 return reference_resource_paths(wim,
1587                                                                 &refglob,
1588                                                                 1,
1589                                                                 ref_flags,
1590                                                                 open_flags,
1591                                                                 progress_func);
1592                         }
1593                 } else {
1594                         ERROR_WITH_ERRNO("Failed to process glob \"%"TS"\"", refglob);
1595                         if (ret == GLOB_NOSPACE)
1596                                 return WIMLIB_ERR_NOMEM;
1597                         else
1598                                 return WIMLIB_ERR_READ;
1599                 }
1600         }
1601
1602         ret = reference_resource_paths(wim,
1603                                        (const tchar * const *)globbuf.gl_pathv,
1604                                        globbuf.gl_pathc,
1605                                        ref_flags,
1606                                        open_flags,
1607                                        progress_func);
1608         globfree(&globbuf);
1609         return ret;
1610 }
1611
1612 /* API function documented in wimlib.h  */
1613 WIMLIBAPI int
1614 wimlib_reference_resource_files(WIMStruct *wim,
1615                                 const tchar * const * resource_wimfiles_or_globs,
1616                                 unsigned count,
1617                                 int ref_flags,
1618                                 int open_flags,
1619                                 wimlib_progress_func_t progress_func)
1620 {
1621         unsigned i;
1622         int ret;
1623
1624         if (ref_flags & WIMLIB_REF_FLAG_GLOB_ENABLE) {
1625                 for (i = 0; i < count; i++) {
1626                         ret = reference_resource_glob(wim,
1627                                                       resource_wimfiles_or_globs[i],
1628                                                       ref_flags,
1629                                                       open_flags,
1630                                                       progress_func);
1631                         if (ret)
1632                                 return ret;
1633                 }
1634                 return 0;
1635         } else {
1636                 return reference_resource_paths(wim, resource_wimfiles_or_globs,
1637                                                 count, ref_flags,
1638                                                 open_flags, progress_func);
1639         }
1640 }