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/assert.h"
  32 #include "wimlib/endianness.h"
  33 #include "wimlib/error.h"
  34 #include "wimlib/lookup_table.h"
  35 #include "wimlib/metadata.h"
  36 #include "wimlib/ntfs_3g.h"
  37 #include "wimlib/resource.h"
  38 #include "wimlib/util.h"
  39 #include "wimlib/write.h"
  40
  41 #include <stdlib.h>
  42 #include <string.h>
  43 #include <unistd.h> /* for unlink()  */
  44
  45 /* WIM lookup table:
  46  *
  47  * This is a logical mapping from SHA1 message digests to the data streams
  48  * contained in a WIM.
  49  *
  50  * Here it is implemented as a hash table.
  51  *
  52  * Note: Everything will break horribly if there is a SHA1 collision.
  53  */
  54 struct wim_lookup_table {
  55         struct hlist_head *array;
  56         size_t num_entries;
  57         size_t capacity;
  58 };
  59
  60 struct wim_lookup_table *
  61 new_lookup_table(size_t capacity)
  62 {
  63         struct wim_lookup_table *table;
  64         struct hlist_head *array;
  65
  66         table = MALLOC(sizeof(struct wim_lookup_table));
  67         if (table == NULL)
  68                 goto oom;
  69
  70         array = CALLOC(capacity, sizeof(array[0]));
  71         if (array == NULL) {
  72                 FREE(table);
  73                 goto oom;
  74         }
  75
  76         table->num_entries = 0;
  77         table->capacity = capacity;
  78         table->array = array;
  79         return table;
  80
  81 oom:
  82         ERROR("Failed to allocate memory for lookup table "
  83               "with capacity %zu", capacity);
  84         return NULL;
  85 }
  86
  87 static int
  88 do_free_lookup_table_entry(struct wim_lookup_table_entry *entry, void *ignore)
  89 {
  90         free_lookup_table_entry(entry);
  91         return 0;
  92 }
  93
  94 void
  95 free_lookup_table(struct wim_lookup_table *table)
  96 {
  97         DEBUG("Freeing lookup table.");
  98         if (table == NULL)
  99                 return;
 100
 101         if (table->array) {
 102                 for_lookup_table_entry(table,
 103                                        do_free_lookup_table_entry,
 104                                        NULL);
 105                 FREE(table->array);
 106         }
 107         FREE(table);
 108 }
 109
 110 struct wim_lookup_table_entry *
 111 new_lookup_table_entry(void)
 112 {
 113         struct wim_lookup_table_entry *lte;
 114
 115         lte = CALLOC(1, sizeof(struct wim_lookup_table_entry));
 116         if (lte == NULL)
 117                 return NULL;
 118
 119         lte->refcnt = 1;
 120
 121         /* lte->resource_location = RESOURCE_NONEXISTENT  */
 122         BUILD_BUG_ON(RESOURCE_NONEXISTENT != 0);
 123
 124         return lte;
 125 }
 126
 127 struct wim_lookup_table_entry *
 128 clone_lookup_table_entry(const struct wim_lookup_table_entry *old)
 129 {
 130         struct wim_lookup_table_entry *new;
 131
 132         new = memdup(old, sizeof(struct wim_lookup_table_entry));
 133         if (new == NULL)
 134                 return NULL;
 135
 136         new->extracted_file = NULL;
 137         switch (new->resource_location) {
 138         case RESOURCE_IN_WIM:
 139                 list_add(&new->rspec_node, &new->rspec->stream_list);
 140                 break;
 141
 142         case RESOURCE_IN_FILE_ON_DISK:
 143 #ifdef __WIN32__
 144         case RESOURCE_WIN32_ENCRYPTED:
 145 #endif
 146 #ifdef WITH_FUSE
 147         case RESOURCE_IN_STAGING_FILE:
 148                 BUILD_BUG_ON((void*)&old->file_on_disk !=
 149                              (void*)&old->staging_file_name);
 150 #endif
 151                 new->file_on_disk = TSTRDUP(old->file_on_disk);
 152                 if (new->file_on_disk == NULL)
 153                         goto out_free;
 154                 break;
 155         case RESOURCE_IN_ATTACHED_BUFFER:
 156                 new->attached_buffer = memdup(old->attached_buffer, old->size);
 157                 if (new->attached_buffer == NULL)
 158                         goto out_free;
 159                 break;
 160 #ifdef WITH_NTFS_3G
 161         case RESOURCE_IN_NTFS_VOLUME:
 162                 if (old->ntfs_loc) {
 163                         struct ntfs_location *loc;
 164                         loc = memdup(old->ntfs_loc, sizeof(struct ntfs_location));
 165                         if (loc == NULL)
 166                                 goto out_free;
 167                         loc->path = NULL;
 168                         loc->stream_name = NULL;
 169                         new->ntfs_loc = loc;
 170                         loc->path = STRDUP(old->ntfs_loc->path);
 171                         if (loc->path == NULL)
 172                                 goto out_free;
 173                         if (loc->stream_name_nchars != 0) {
 174                                 loc->stream_name = memdup(old->ntfs_loc->stream_name,
 175                                                           loc->stream_name_nchars * 2);
 176                                 if (loc->stream_name == NULL)
 177                                         goto out_free;
 178                         }
 179                 }
 180                 break;
 181 #endif
 182         default:
 183                 break;
 184         }
 185         return new;
 186
 187 out_free:
 188         free_lookup_table_entry(new);
 189         return NULL;
 190 }
 191
 192 void
 193 lte_put_resource(struct wim_lookup_table_entry *lte)
 194 {
 195         switch (lte->resource_location) {
 196         case RESOURCE_IN_WIM:
 197                 list_del(&lte->rspec_node);
 198                 if (list_empty(&lte->rspec->stream_list))
 199                         FREE(lte->rspec);
 200                 break;
 201         case RESOURCE_IN_FILE_ON_DISK:
 202 #ifdef __WIN32__
 203         case RESOURCE_WIN32_ENCRYPTED:
 204 #endif
 205 #ifdef WITH_FUSE
 206         case RESOURCE_IN_STAGING_FILE:
 207                 BUILD_BUG_ON((void*)&lte->file_on_disk !=
 208                              (void*)&lte->staging_file_name);
 209 #endif
 210         case RESOURCE_IN_ATTACHED_BUFFER:
 211                 BUILD_BUG_ON((void*)&lte->file_on_disk !=
 212                              (void*)&lte->attached_buffer);
 213                 FREE(lte->file_on_disk);
 214                 break;
 215 #ifdef WITH_NTFS_3G
 216         case RESOURCE_IN_NTFS_VOLUME:
 217                 if (lte->ntfs_loc) {
 218                         FREE(lte->ntfs_loc->path);
 219                         FREE(lte->ntfs_loc->stream_name);
 220                         FREE(lte->ntfs_loc);
 221                 }
 222                 break;
 223 #endif
 224         default:
 225                 break;
 226         }
 227 }
 228
 229 void
 230 free_lookup_table_entry(struct wim_lookup_table_entry *lte)
 231 {
 232         if (lte) {
 233                 lte_put_resource(lte);
 234                 FREE(lte);
 235         }
 236 }
 237
 238 /* Decrements the reference count for the lookup table entry @lte.  If its
 239  * reference count reaches 0, it is unlinked from the lookup table.  If,
 240  * furthermore, the entry has no opened file descriptors associated with it, the
 241  * entry is freed.  */
 242 void
 243 lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
 244                      struct wim_lookup_table *table)
 245 {
 246         wimlib_assert(lte->refcnt != 0);
 247
 248         if (--lte->refcnt == 0) {
 249                 if (lte->unhashed) {
 250                         list_del(&lte->unhashed_list);
 251                 #ifdef WITH_FUSE
 252                         /* If the stream has been extracted to a staging file
 253                          * for a FUSE mount, unlink the staging file.  (Note
 254                          * that there still may be open file descriptors to it.)
 255                          * */
 256                         if (lte->resource_location == RESOURCE_IN_STAGING_FILE)
 257                                 unlink(lte->staging_file_name);
 258                 #endif
 259                 } else {
 260                         lookup_table_unlink(table, lte);
 261                 }
 262
 263                 /* If FUSE mounts are enabled, we don't actually free the entry
 264                  * until the last file descriptor has been closed by
 265                  * lte_decrement_num_opened_fds().  */
 266 #ifdef WITH_FUSE
 267                 if (lte->num_opened_fds == 0)
 268 #endif
 269                         free_lookup_table_entry(lte);
 270         }
 271 }
 272
 273 #ifdef WITH_FUSE
 274 void
 275 lte_decrement_num_opened_fds(struct wim_lookup_table_entry *lte)
 276 {
 277         wimlib_assert(lte->num_opened_fds != 0);
 278
 279         if (--lte->num_opened_fds == 0 && lte->refcnt == 0)
 280                 free_lookup_table_entry(lte);
 281 }
 282 #endif
 283
 284 static void
 285 lookup_table_insert_raw(struct wim_lookup_table *table,
 286                         struct wim_lookup_table_entry *lte)
 287 {
 288         size_t i = lte->hash_short % table->capacity;
 289
 290         hlist_add_head(&lte->hash_list, &table->array[i]);
 291 }
 292
 293 static void
 294 enlarge_lookup_table(struct wim_lookup_table *table)
 295 {
 296         size_t old_capacity, new_capacity;
 297         struct hlist_head *old_array, *new_array;
 298         struct wim_lookup_table_entry *lte;
 299         struct hlist_node *cur, *tmp;
 300         size_t i;
 301
 302         old_capacity = table->capacity;
 303         new_capacity = old_capacity * 2;
 304         new_array = CALLOC(new_capacity, sizeof(struct hlist_head));
 305         if (new_array == NULL)
 306                 return;
 307         old_array = table->array;
 308         table->array = new_array;
 309         table->capacity = new_capacity;
 310
 311         for (i = 0; i < old_capacity; i++) {
 312                 hlist_for_each_entry_safe(lte, cur, tmp, &old_array[i], hash_list) {
 313                         hlist_del(&lte->hash_list);
 314                         lookup_table_insert_raw(table, lte);
 315                 }
 316         }
 317         FREE(old_array);
 318 }
 319
 320 /* Inserts an entry into the lookup table.  */
 321 void
 322 lookup_table_insert(struct wim_lookup_table *table,
 323                     struct wim_lookup_table_entry *lte)
 324 {
 325         lookup_table_insert_raw(table, lte);
 326         if (++table->num_entries > table->capacity)
 327                 enlarge_lookup_table(table);
 328 }
 329
 330 /* Unlinks a lookup table entry from the table; does not free it.  */
 331 void
 332 lookup_table_unlink(struct wim_lookup_table *table,
 333                     struct wim_lookup_table_entry *lte)
 334 {
 335         wimlib_assert(!lte->unhashed);
 336         wimlib_assert(table->num_entries != 0);
 337
 338         hlist_del(&lte->hash_list);
 339         table->num_entries--;
 340 }
 341
 342 /* Given a SHA1 message digest, return the corresponding entry in the WIM's
 343  * lookup table, or NULL if there is none.  */
 344 struct wim_lookup_table_entry *
 345 lookup_stream(const struct wim_lookup_table *table, const u8 hash[])
 346 {
 347         size_t i;
 348         struct wim_lookup_table_entry *lte;
 349         struct hlist_node *pos;
 350
 351         i = *(size_t*)hash % table->capacity;
 352         hlist_for_each_entry(lte, pos, &table->array[i], hash_list)
 353                 if (hashes_equal(hash, lte->hash))
 354                         return lte;
 355         return NULL;
 356 }
 357
 358 /* Calls a function on all the entries in the WIM lookup table.  Stop early and
 359  * return nonzero if any call to the function returns nonzero. */
 360 int
 361 for_lookup_table_entry(struct wim_lookup_table *table,
 362                        int (*visitor)(struct wim_lookup_table_entry *, void *),
 363                        void *arg)
 364 {
 365         struct wim_lookup_table_entry *lte;
 366         struct hlist_node *pos, *tmp;
 367         int ret;
 368
 369         for (size_t i = 0; i < table->capacity; i++) {
 370                 hlist_for_each_entry_safe(lte, pos, tmp, &table->array[i],
 371                                           hash_list)
 372                 {
 373                         ret = visitor(lte, arg);
 374                         if (ret)
 375                                 return ret;
 376                 }
 377         }
 378         return 0;
 379 }
 380
 381 /* qsort() callback that sorts streams (represented by `struct
 382  * wim_lookup_table_entry's) into an order optimized for reading.
 383  *
 384  * Sorting is done primarily by resource location, then secondarily by a
 385  * per-resource location order.  For example, resources in WIM files are sorted
 386  * primarily by part number, then secondarily by offset, as to implement optimal
 387  * reading of either a standalone or split WIM.  */
 388 static int
 389 cmp_streams_by_sequential_order(const void *p1, const void *p2)
 390 {
 391         const struct wim_lookup_table_entry *lte1, *lte2;
 392         int v;
 393         WIMStruct *wim1, *wim2;
 394
 395         lte1 = *(const struct wim_lookup_table_entry**)p1;
 396         lte2 = *(const struct wim_lookup_table_entry**)p2;
 397
 398         v = (int)lte1->resource_location - (int)lte2->resource_location;
 399
 400         /* Different resource locations?  */
 401         if (v)
 402                 return v;
 403
 404         switch (lte1->resource_location) {
 405         case RESOURCE_IN_WIM:
 406                 wim1 = lte1->rspec->wim;
 407                 wim2 = lte2->rspec->wim;
 408
 409                 /* Different (possibly split) WIMs?  */
 410                 if (wim1 != wim2) {
 411                         v = memcmp(wim1->hdr.guid, wim2->hdr.guid, WIM_GID_LEN);
 412                         if (v)
 413                                 return v;
 414                 }
 415
 416                 /* Different part numbers in the same WIM?  */
 417                 v = (int)wim1->hdr.part_number - (int)wim2->hdr.part_number;
 418                 if (v)
 419                         return v;
 420
 421                 if (lte1->rspec->offset_in_wim != lte2->rspec->offset_in_wim)
 422                         return cmp_u64(lte1->rspec->offset_in_wim,
 423                                        lte2->rspec->offset_in_wim);
 424
 425                 return cmp_u64(lte1->offset_in_res, lte2->offset_in_res);
 426
 427         case RESOURCE_IN_FILE_ON_DISK:
 428 #ifdef WITH_FUSE
 429         case RESOURCE_IN_STAGING_FILE:
 430 #endif
 431 #ifdef __WIN32__
 432         case RESOURCE_WIN32_ENCRYPTED:
 433 #endif
 434                 /* Compare files by path: just a heuristic that will place files
 435                  * in the same directory next to each other.  */
 436                 return tstrcmp(lte1->file_on_disk, lte2->file_on_disk);
 437 #ifdef WITH_NTFS_3G
 438         case RESOURCE_IN_NTFS_VOLUME:
 439                 return tstrcmp(lte1->ntfs_loc->path, lte2->ntfs_loc->path);
 440 #endif
 441         default:
 442                 /* No additional sorting order defined for this resource
 443                  * location (e.g. RESOURCE_IN_ATTACHED_BUFFER); simply compare
 444                  * everything equal to each other.  */
 445                 return 0;
 446         }
 447 }
 448
 449 int
 450 sort_stream_list(struct list_head *stream_list,
 451                  size_t list_head_offset,
 452                  int (*compar)(const void *, const void*))
 453 {
 454         struct list_head *cur;
 455         struct wim_lookup_table_entry **array;
 456         size_t i;
 457         size_t array_size;
 458         size_t num_streams = 0;
 459
 460         list_for_each(cur, stream_list)
 461                 num_streams++;
 462
 463         if (num_streams <= 1)
 464                 return 0;
 465
 466         array_size = num_streams * sizeof(array[0]);
 467         array = MALLOC(array_size);
 468         if (array == NULL)
 469                 return WIMLIB_ERR_NOMEM;
 470
 471         cur = stream_list->next;
 472         for (i = 0; i < num_streams; i++) {
 473                 array[i] = (struct wim_lookup_table_entry*)((u8*)cur -
 474                                                             list_head_offset);
 475                 cur = cur->next;
 476         }
 477
 478         qsort(array, num_streams, sizeof(array[0]), compar);
 479
 480         INIT_LIST_HEAD(stream_list);
 481         for (i = 0; i < num_streams; i++) {
 482                 list_add_tail((struct list_head*)
 483                                ((u8*)array[i] + list_head_offset),
 484                               stream_list);
 485         }
 486         FREE(array);
 487         return 0;
 488 }
 489
 490 /* Sort the specified list of streams in an order optimized for reading.  */
 491 int
 492 sort_stream_list_by_sequential_order(struct list_head *stream_list,
 493                                      size_t list_head_offset)
 494 {
 495         return sort_stream_list(stream_list, list_head_offset,
 496                                 cmp_streams_by_sequential_order);
 497 }
 498
 499
 500 static int
 501 add_lte_to_array(struct wim_lookup_table_entry *lte,
 502                  void *_pp)
 503 {
 504         struct wim_lookup_table_entry ***pp = _pp;
 505         *(*pp)++ = lte;
 506         return 0;
 507 }
 508
 509 /* Iterate through the lookup table entries, but first sort them by stream
 510  * offset in the WIM.  Caution: this is intended to be used when the stream
 511  * offset field has actually been set. */
 512 int
 513 for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
 514                                   int (*visitor)(struct wim_lookup_table_entry *,
 515                                                  void *),
 516                                   void *arg)
 517 {
 518         struct wim_lookup_table_entry **lte_array, **p;
 519         size_t num_streams = table->num_entries;
 520         int ret;
 521
 522         lte_array = MALLOC(num_streams * sizeof(lte_array[0]));
 523         if (!lte_array)
 524                 return WIMLIB_ERR_NOMEM;
 525         p = lte_array;
 526         for_lookup_table_entry(table, add_lte_to_array, &p);
 527
 528         wimlib_assert(p == lte_array + num_streams);
 529
 530         qsort(lte_array, num_streams, sizeof(lte_array[0]),
 531               cmp_streams_by_sequential_order);
 532         ret = 0;
 533         for (size_t i = 0; i < num_streams; i++) {
 534                 ret = visitor(lte_array[i], arg);
 535                 if (ret)
 536                         break;
 537         }
 538         FREE(lte_array);
 539         return ret;
 540 }
 541
 542 /* On-disk format of a WIM lookup table entry (stream entry). */
 543 struct wim_lookup_table_entry_disk {
 544         /* Size, offset, and flags of the stream.  */
 545         struct wim_reshdr_disk reshdr;
 546
 547         /* Which part of the split WIM this stream is in; indexed from 1. */
 548         le16 part_number;
 549
 550         /* Reference count of this stream over all WIM images. */
 551         le32 refcnt;
 552
 553         /* SHA1 message digest of the uncompressed data of this stream, or
 554          * optionally all zeroes if this stream is of zero length. */
 555         u8 hash[SHA1_HASH_SIZE];
 556 } _packed_attribute;
 557
 558 #define WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE 50
 559
 560 /* Validate the size and location of a WIM resource.  */
 561 static int
 562 validate_resource(const struct wim_resource_spec *rspec)
 563 {
 564         struct wim_lookup_table_entry *lte;
 565         u64 cur_offset;
 566
 567         /* Verify that calculating the offset of the end of the resource doesn't
 568          * overflow.  */
 569         if (rspec->offset_in_wim + rspec->size_in_wim < rspec->size_in_wim)
 570                 goto invalid;
 571
 572         /* Verify that each stream in the resource has a valid offset and size,
 573          * and that no streams overlap, and that the streams were added in order
 574          * of increasing offset.  */
 575         cur_offset = 0;
 576         list_for_each_entry(lte, &rspec->stream_list, rspec_node) {
 577                 if (lte->offset_in_res + lte->size < lte->size ||
 578                     lte->offset_in_res + lte->size > rspec->uncompressed_size ||
 579                     lte->offset_in_res < cur_offset)
 580                         goto invalid;
 581
 582                 cur_offset = lte->offset_in_res + lte->size;
 583         }
 584         return 0;
 585
 586 invalid:
 587
 588         ERROR("Invalid resource entry!");
 589         return WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 590 }
 591
 592 /*
 593  * Reads the lookup table from a WIM file.  Each entry specifies a stream that
 594  * the WIM file contains, along with its location and SHA1 message digest.
 595  *
 596  * Saves lookup table entries for non-metadata streams in a hash table, and
 597  * saves the metadata entry for each image in a special per-image location (the
 598  * image_metadata array).
 599  *
 600  * Return values:
 601  *      WIMLIB_ERR_SUCCESS (0)
 602  *      WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY
 603  *      WIMLIB_ERR_RESOURCE_NOT_FOUND
 604  *
 605  *      Or an error code caused by failure to read the lookup table into memory.
 606  */
 607 int
 608 read_wim_lookup_table(WIMStruct *wim)
 609 {
 610         int ret;
 611         size_t i;
 612         size_t num_entries;
 613         struct wim_lookup_table *table;
 614         struct wim_lookup_table_entry *cur_entry, *duplicate_entry;
 615         struct wim_resource_spec *cur_rspec;
 616         void *buf;
 617         bool back_to_back_pack;
 618
 619         DEBUG("Reading lookup table.");
 620
 621         /* Sanity check: lookup table entries are 50 bytes each.  */
 622         BUILD_BUG_ON(sizeof(struct wim_lookup_table_entry_disk) !=
 623                      WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE);
 624
 625         /* Calculate number of entries in the lookup table.  */
 626         num_entries = wim->hdr.lookup_table_reshdr.uncompressed_size /
 627                       sizeof(struct wim_lookup_table_entry_disk);
 628
 629         /* Read the lookup table into a buffer.  */
 630         ret = wim_reshdr_to_data(&wim->hdr.lookup_table_reshdr, wim, &buf);
 631         if (ret)
 632                 goto out;
 633
 634         /* Allocate a hash table to map SHA1 message digests into stream
 635          * specifications.  This is the in-memory "lookup table".  */
 636         table = new_lookup_table(num_entries * 2 + 1);
 637         if (table == NULL) {
 638                 ERROR("Not enough memory to read lookup table.");
 639                 ret = WIMLIB_ERR_NOMEM;
 640                 goto out_free_buf;
 641         }
 642
 643         /* Allocate and initalize stream entries from the raw lookup table
 644          * buffer.  */
 645         wim->current_image = 0;
 646         cur_rspec = NULL;
 647         for (i = 0; i < num_entries; i++) {
 648                 const struct wim_lookup_table_entry_disk *disk_entry =
 649                         &((const struct wim_lookup_table_entry_disk*)buf)[i];
 650                 u16 part_number;
 651                 struct wim_reshdr reshdr;
 652
 653                 get_wim_reshdr(&disk_entry->reshdr, &reshdr);
 654
 655                 DEBUG("reshdr: size_in_wim=%"PRIu64", "
 656                       "uncompressed_size=%"PRIu64", "
 657                       "offset_in_wim=%"PRIu64", "
 658                       "flags=0x%02x",
 659                       reshdr.size_in_wim, reshdr.uncompressed_size,
 660                       reshdr.offset_in_wim, reshdr.flags);
 661
 662                 if (wim->hdr.wim_version == WIM_VERSION_DEFAULT)
 663                         reshdr.flags &= ~WIM_RESHDR_FLAG_PACKED_STREAMS;
 664
 665                 cur_entry = new_lookup_table_entry();
 666                 if (cur_entry == NULL) {
 667                         ERROR("Not enough memory to read lookup table!");
 668                         ret = WIMLIB_ERR_NOMEM;
 669                         goto err;
 670                 }
 671
 672                 part_number = le16_to_cpu(disk_entry->part_number);
 673                 cur_entry->refcnt = le32_to_cpu(disk_entry->refcnt);
 674                 copy_hash(cur_entry->hash, disk_entry->hash);
 675
 676                 if (part_number != wim->hdr.part_number) {
 677                         WARNING("A lookup table entry in part %hu of the WIM "
 678                                 "points to part %hu (ignoring it)",
 679                                 wim->hdr.part_number, part_number);
 680                         free_lookup_table_entry(cur_entry);
 681                         continue;
 682                 }
 683
 684                 if (!(reshdr.flags & (WIM_RESHDR_FLAG_PACKED_STREAMS |
 685                                       WIM_RESHDR_FLAG_COMPRESSED))) {
 686                         if (reshdr.uncompressed_size != reshdr.size_in_wim) {
 687                                 ERROR("Invalid resource entry!");
 688                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 689                                 goto err;
 690                         }
 691                 }
 692
 693                 back_to_back_pack = false;
 694                 if (!(reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) ||
 695                     cur_rspec == NULL ||
 696                     (back_to_back_pack =
 697                      ((reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) &&
 698                       reshdr.uncompressed_size == WIM_PACK_MAGIC_NUMBER &&
 699                       cur_rspec != NULL &&
 700                       cur_rspec->size_in_wim != 0)))
 701                 {
 702                         /* Starting new run of streams that share the same WIM
 703                          * resource.  */
 704                         struct wim_lookup_table_entry *prev_entry = NULL;
 705
 706                         if (back_to_back_pack &&
 707                             !list_empty(&cur_rspec->stream_list))
 708                         {
 709                                 prev_entry = list_entry(cur_rspec->stream_list.prev,
 710                                                         struct wim_lookup_table_entry,
 711                                                         rspec_node);
 712                                 lte_unbind_wim_resource_spec(prev_entry);
 713                         }
 714                         if (cur_rspec != NULL) {
 715                                 ret = validate_resource(cur_rspec);
 716                                 if (ret)
 717                                         goto err;
 718                         }
 719
 720                         /* Allocate the resource specification and initialize it
 721                          * with values from the current stream entry.  */
 722                         cur_rspec = MALLOC(sizeof(*cur_rspec));
 723                         if (cur_rspec == NULL) {
 724                                 ERROR("Not enough memory to read lookup table!");
 725                                 ret = WIMLIB_ERR_NOMEM;
 726                                 goto err;
 727                         }
 728                         wim_res_hdr_to_spec(&reshdr, wim, cur_rspec);
 729
 730                         /* If this is a packed run, the current stream entry may
 731                          * specify a stream within the resource, and not the
 732                          * resource itself.  Zero possibly irrelevant data until
 733                          * it is read for certain.  (Note that the computation
 734                          * of 'back_to_back_pack' tests if 'size_in_wim' is
 735                          * nonzero to see if the resource info has been read;
 736                          * hence we need to set it to 0 here.)  */
 737                         if (reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
 738                                 cur_rspec->size_in_wim = 0;
 739                                 cur_rspec->uncompressed_size = 0;
 740                                 cur_rspec->offset_in_wim = 0;
 741                         }
 742
 743                         if (prev_entry)
 744                                 lte_bind_wim_resource_spec(prev_entry, cur_rspec);
 745                 }
 746
 747                 if ((reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) &&
 748                     reshdr.uncompressed_size == WIM_PACK_MAGIC_NUMBER)
 749                 {
 750                         /* Found the specification for the packed resource.
 751                          * Transfer the values to the `struct
 752                          * wim_resource_spec', and discard the current stream
 753                          * since this lookup table entry did not, in fact,
 754                          * correspond to a "stream".
 755                          */
 756
 757                         /* Uncompressed size of the resource pack is actually
 758                          * stored in the header of the resource itself.  Read
 759                          * it, and also grab the chunk size and compression type
 760                          * (which are not necessarily the defaults from the WIM
 761                          * header).  */
 762                         struct alt_chunk_table_header_disk hdr;
 763
 764                         ret = full_pread(&wim->in_fd, &hdr,
 765                                          sizeof(hdr), reshdr.offset_in_wim);
 766                         if (ret)
 767                                 goto err;
 768
 769                         cur_rspec->uncompressed_size = le64_to_cpu(hdr.res_usize);
 770                         cur_rspec->offset_in_wim = reshdr.offset_in_wim;
 771                         cur_rspec->size_in_wim = reshdr.size_in_wim;
 772                         cur_rspec->flags = reshdr.flags;
 773
 774                         /* Compression format numbers must be the same as in
 775                          * WIMGAPI to be compatible here.  */
 776                         BUILD_BUG_ON(WIMLIB_COMPRESSION_TYPE_NONE != 0);
 777                         BUILD_BUG_ON(WIMLIB_COMPRESSION_TYPE_XPRESS != 1);
 778                         BUILD_BUG_ON(WIMLIB_COMPRESSION_TYPE_LZX != 2);
 779                         BUILD_BUG_ON(WIMLIB_COMPRESSION_TYPE_LZMS != 3);
 780                         cur_rspec->compression_type = le32_to_cpu(hdr.compression_format);
 781
 782                         cur_rspec->chunk_size = le32_to_cpu(hdr.chunk_size);
 783
 784                         DEBUG("Full pack is %"PRIu64" compressed bytes "
 785                               "at file offset %"PRIu64" (flags 0x%02x)",
 786                               cur_rspec->size_in_wim,
 787                               cur_rspec->offset_in_wim,
 788                               cur_rspec->flags);
 789                         free_lookup_table_entry(cur_entry);
 790                         continue;
 791                 }
 792
 793                 if (is_zero_hash(cur_entry->hash)) {
 794                         free_lookup_table_entry(cur_entry);
 795                         continue;
 796                 }
 797
 798                 if (reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
 799                         /* Continuing the pack with another stream.  */
 800                         DEBUG("Continuing pack with stream: "
 801                               "%"PRIu64" uncompressed bytes @ "
 802                               "resource offset %"PRIu64")",
 803                               reshdr.size_in_wim, reshdr.offset_in_wim);
 804                 }
 805
 806                 lte_bind_wim_resource_spec(cur_entry, cur_rspec);
 807                 if (reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
 808                         /* In packed runs, the offset field is used for
 809                          * in-resource offset, not the in-WIM offset, and the
 810                          * size field is used for the uncompressed size, not the
 811                          * compressed size.  */
 812                         cur_entry->offset_in_res = reshdr.offset_in_wim;
 813                         cur_entry->size = reshdr.size_in_wim;
 814                         cur_entry->flags = reshdr.flags;
 815                 } else {
 816                         /* Normal case: The stream corresponds one-to-one with
 817                          * the resource entry.  */
 818                         cur_entry->offset_in_res = 0;
 819                         cur_entry->size = reshdr.uncompressed_size;
 820                         cur_entry->flags = reshdr.flags;
 821                         cur_rspec = NULL;
 822                 }
 823
 824                 if (cur_entry->flags & WIM_RESHDR_FLAG_METADATA) {
 825                         /* Lookup table entry for a metadata resource */
 826
 827                         /* Metadata entries with no references must be ignored;
 828                          * see for example the WinPE WIMs from the WAIK v2.1.
 829                          * */
 830                         if (cur_entry->refcnt == 0) {
 831                                 free_lookup_table_entry(cur_entry);
 832                                 continue;
 833                         }
 834
 835                         if (cur_entry->refcnt != 1) {
 836                                 ERROR("Found metadata resource with refcnt != 1");
 837                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 838                                 goto err;
 839                         }
 840
 841                         if (wim->hdr.part_number != 1) {
 842                                 WARNING("Ignoring metadata resource found in a "
 843                                         "non-first part of the split WIM");
 844                                 free_lookup_table_entry(cur_entry);
 845                                 continue;
 846                         }
 847                         if (wim->current_image == wim->hdr.image_count) {
 848                                 WARNING("The WIM header says there are %u images "
 849                                         "in the WIM, but we found more metadata "
 850                                         "resources than this (ignoring the extra)",
 851                                         wim->hdr.image_count);
 852                                 free_lookup_table_entry(cur_entry);
 853                                 continue;
 854                         }
 855
 856                         /* Notice very carefully:  We are assigning the metadata
 857                          * resources in the exact order mirrored by their lookup
 858                          * table entries on disk, which is the behavior of
 859                          * Microsoft's software.  In particular, this overrides
 860                          * the actual locations of the metadata resources
 861                          * themselves in the WIM file as well as any information
 862                          * written in the XML data. */
 863                         DEBUG("Found metadata resource for image %u at "
 864                               "offset %"PRIu64".",
 865                               wim->current_image + 1,
 866                               cur_entry->rspec->offset_in_wim);
 867                         wim->image_metadata[
 868                                 wim->current_image++]->metadata_lte = cur_entry;
 869                         continue;
 870                 }
 871
 872                 /* Lookup table entry for a stream that is not a metadata
 873                  * resource.  */
 874                 duplicate_entry = lookup_stream(table, cur_entry->hash);
 875                 if (duplicate_entry) {
 876                         WARNING("The WIM lookup table contains two entries "
 877                                 "with the same SHA1 message digest!");
 878                         free_lookup_table_entry(cur_entry);
 879                         continue;
 880                 }
 881
 882                 /* Finally, insert the stream into the lookup table, keyed by
 883                  * its SHA1 message digest.  */
 884                 lookup_table_insert(table, cur_entry);
 885         }
 886         cur_entry = NULL;
 887
 888         /* Validate the last resource.  */
 889         if (cur_rspec != NULL) {
 890                 ret = validate_resource(cur_rspec);
 891                 if (ret)
 892                         goto err;
 893         }
 894
 895         if (wim->hdr.part_number == 1 && wim->current_image != wim->hdr.image_count) {
 896                 WARNING("The header of \"%"TS"\" says there are %u images in\n"
 897                         "          the WIM, but we only found %d metadata resources!  Acting as if\n"
 898                         "          the header specified only %d images instead.",
 899                         wim->filename, wim->hdr.image_count,
 900                         wim->current_image, wim->current_image);
 901                 for (int i = wim->current_image; i < wim->hdr.image_count; i++)
 902                         put_image_metadata(wim->image_metadata[i], NULL);
 903                 wim->hdr.image_count = wim->current_image;
 904         }
 905         DEBUG("Done reading lookup table.");
 906         wim->lookup_table = table;
 907         ret = 0;
 908         goto out_free_buf;
 909
 910 err:
 911         if (cur_rspec && list_empty(&cur_rspec->stream_list))
 912                 FREE(cur_rspec);
 913         free_lookup_table_entry(cur_entry);
 914         free_lookup_table(table);
 915 out_free_buf:
 916         FREE(buf);
 917 out:
 918         wim->current_image = 0;
 919         return ret;
 920 }
 921
 922 static void
 923 put_wim_lookup_table_entry(struct wim_lookup_table_entry_disk *disk_entry,
 924                            const struct wim_reshdr *out_reshdr,
 925                            u16 part_number, u32 refcnt, const u8 *hash)
 926 {
 927         put_wim_reshdr(out_reshdr, &disk_entry->reshdr);
 928         disk_entry->part_number = cpu_to_le16(part_number);
 929         disk_entry->refcnt = cpu_to_le32(refcnt);
 930         copy_hash(disk_entry->hash, hash);
 931 }
 932
 933 int
 934 write_wim_lookup_table_from_stream_list(struct list_head *stream_list,
 935                                         struct filedes *out_fd,
 936                                         u16 part_number,
 937                                         struct wim_reshdr *out_reshdr,
 938                                         int write_resource_flags)
 939 {
 940         size_t table_size;
 941         struct wim_lookup_table_entry *lte;
 942         struct wim_lookup_table_entry_disk *table_buf;
 943         struct wim_lookup_table_entry_disk *table_buf_ptr;
 944         int ret;
 945         u64 prev_res_offset_in_wim = ~0ULL;
 946
 947         table_size = 0;
 948         list_for_each_entry(lte, stream_list, lookup_table_list) {
 949                 table_size += sizeof(struct wim_lookup_table_entry_disk);
 950
 951                 if (lte->out_reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS &&
 952                     lte->out_res_offset_in_wim != prev_res_offset_in_wim)
 953                 {
 954                         table_size += sizeof(struct wim_lookup_table_entry_disk);
 955                         prev_res_offset_in_wim = lte->out_res_offset_in_wim;
 956                 }
 957         }
 958
 959         DEBUG("Writing WIM lookup table (size=%zu, offset=%"PRIu64")",
 960               table_size, out_fd->offset);
 961
 962         table_buf = MALLOC(table_size);
 963         if (table_buf == NULL) {
 964                 ERROR("Failed to allocate %zu bytes for temporary lookup table",
 965                       table_size);
 966                 return WIMLIB_ERR_NOMEM;
 967         }
 968         table_buf_ptr = table_buf;
 969
 970         prev_res_offset_in_wim = ~0ULL;
 971         list_for_each_entry(lte, stream_list, lookup_table_list) {
 972
 973                 put_wim_lookup_table_entry(table_buf_ptr++,
 974                                            &lte->out_reshdr,
 975                                            part_number,
 976                                            lte->out_refcnt,
 977                                            lte->hash);
 978                 if (lte->out_reshdr.flags & WIM_RESHDR_FLAG_PACKED_STREAMS &&
 979                     lte->out_res_offset_in_wim != prev_res_offset_in_wim)
 980                 {
 981                         /* Put the main resource entry for the pack.  */
 982
 983                         struct wim_reshdr reshdr;
 984
 985                         reshdr.offset_in_wim = lte->out_res_offset_in_wim;
 986                         reshdr.size_in_wim = lte->out_res_size_in_wim;
 987                         reshdr.uncompressed_size = WIM_PACK_MAGIC_NUMBER;
 988                         reshdr.flags = WIM_RESHDR_FLAG_PACKED_STREAMS;
 989
 990                         DEBUG("Putting main entry for pack: "
 991                               "size_in_wim=%"PRIu64", "
 992                               "offset_in_wim=%"PRIu64", "
 993                               "uncompressed_size=%"PRIu64,
 994                               reshdr.size_in_wim,
 995                               reshdr.offset_in_wim,
 996                               reshdr.uncompressed_size);
 997
 998                         put_wim_lookup_table_entry(table_buf_ptr++,
 999                                                    &reshdr,
1000                                                    part_number,
1001                                                    1, zero_hash);
1002                         prev_res_offset_in_wim = lte->out_res_offset_in_wim;
1003                 }
1004
1005         }
1006         wimlib_assert((u8*)table_buf_ptr - (u8*)table_buf == table_size);
1007
1008         /* Write the lookup table uncompressed.  Although wimlib can handle a
1009          * compressed lookup table, MS software cannot.  */
1010         ret = write_wim_resource_from_buffer(table_buf,
1011                                              table_size,
1012                                              WIM_RESHDR_FLAG_METADATA,
1013                                              out_fd,
1014                                              WIMLIB_COMPRESSION_TYPE_NONE,
1015                                              0,
1016                                              out_reshdr,
1017                                              NULL,
1018                                              write_resource_flags);
1019         FREE(table_buf);
1020         DEBUG("ret=%d", ret);
1021         return ret;
1022 }
1023
1024 int
1025 lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
1026 {
1027         lte->real_refcnt = 0;
1028         return 0;
1029 }
1030
1031 int
1032 lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
1033 {
1034         lte->out_refcnt = 0;
1035         return 0;
1036 }
1037
1038 int
1039 lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *_ignore)
1040 {
1041         if (lte->extracted_file != NULL) {
1042                 FREE(lte->extracted_file);
1043                 lte->extracted_file = NULL;
1044         }
1045         return 0;
1046 }
1047
1048 /* Allocate a stream entry for the contents of the buffer, or re-use an existing
1049  * entry in @lookup_table for the same stream.  */
1050 struct wim_lookup_table_entry *
1051 new_stream_from_data_buffer(const void *buffer, size_t size,
1052                             struct wim_lookup_table *lookup_table)
1053 {
1054         u8 hash[SHA1_HASH_SIZE];
1055         struct wim_lookup_table_entry *lte, *existing_lte;
1056
1057         sha1_buffer(buffer, size, hash);
1058         existing_lte = lookup_stream(lookup_table, hash);
1059         if (existing_lte) {
1060                 wimlib_assert(existing_lte->size == size);
1061                 lte = existing_lte;
1062                 lte->refcnt++;
1063         } else {
1064                 void *buffer_copy;
1065                 lte = new_lookup_table_entry();
1066                 if (lte == NULL)
1067                         return NULL;
1068                 buffer_copy = memdup(buffer, size);
1069                 if (buffer_copy == NULL) {
1070                         free_lookup_table_entry(lte);
1071                         return NULL;
1072                 }
1073                 lte->resource_location  = RESOURCE_IN_ATTACHED_BUFFER;
1074                 lte->attached_buffer    = buffer_copy;
1075                 lte->size               = size;
1076                 copy_hash(lte->hash, hash);
1077                 lookup_table_insert(lookup_table, lte);
1078         }
1079         return lte;
1080 }
1081
1082 /* Calculate the SHA1 message digest of a stream and move it from the list of
1083  * unhashed streams to the stream lookup table, possibly joining it with an
1084  * existing lookup table entry for an identical stream.
1085  *
1086  * @lte:  An unhashed lookup table entry.
1087  * @lookup_table:  Lookup table for the WIM.
1088  * @lte_ret:  On success, write a pointer to the resulting lookup table
1089  *            entry to this location.  This will be the same as @lte
1090  *            if it was inserted into the lookup table, or different if
1091  *            a duplicate stream was found.
1092  *
1093  * Returns 0 on success; nonzero if there is an error reading the stream.
1094  */
1095 int
1096 hash_unhashed_stream(struct wim_lookup_table_entry *lte,
1097                      struct wim_lookup_table *lookup_table,
1098                      struct wim_lookup_table_entry **lte_ret)
1099 {
1100         int ret;
1101         struct wim_lookup_table_entry *duplicate_lte;
1102         struct wim_lookup_table_entry **back_ptr;
1103
1104         wimlib_assert(lte->unhashed);
1105
1106         /* back_ptr must be saved because @back_inode and @back_stream_id are in
1107          * union with the SHA1 message digest and will no longer be valid once
1108          * the SHA1 has been calculated. */
1109         back_ptr = retrieve_lte_pointer(lte);
1110
1111         ret = sha1_stream(lte);
1112         if (ret)
1113                 return ret;
1114
1115         /* Look for a duplicate stream */
1116         duplicate_lte = lookup_stream(lookup_table, lte->hash);
1117         list_del(&lte->unhashed_list);
1118         if (duplicate_lte) {
1119                 /* We have a duplicate stream.  Transfer the reference counts
1120                  * from this stream to the duplicate and update the reference to
1121                  * this stream (in an inode or ads_entry) to point to the
1122                  * duplicate.  The caller is responsible for freeing @lte if
1123                  * needed.  */
1124                 wimlib_assert(!(duplicate_lte->unhashed));
1125                 wimlib_assert(duplicate_lte->size == lte->size);
1126                 duplicate_lte->refcnt += lte->refcnt;
1127                 lte->refcnt = 0;
1128                 *back_ptr = duplicate_lte;
1129                 lte = duplicate_lte;
1130         } else {
1131                 /* No duplicate stream, so we need to insert this stream into
1132                  * the lookup table and treat it as a hashed stream. */
1133                 lookup_table_insert(lookup_table, lte);
1134                 lte->unhashed = 0;
1135         }
1136         *lte_ret = lte;
1137         return 0;
1138 }
1139
1140 void
1141 lte_to_wimlib_resource_entry(const struct wim_lookup_table_entry *lte,
1142                              struct wimlib_resource_entry *wentry)
1143 {
1144         memset(wentry, 0, sizeof(*wentry));
1145
1146         wentry->uncompressed_size = lte->size;
1147         if (lte->resource_location == RESOURCE_IN_WIM) {
1148                 wentry->part_number = lte->rspec->wim->hdr.part_number;
1149                 if (lte->flags & WIM_RESHDR_FLAG_PACKED_STREAMS) {
1150                         wentry->compressed_size = 0;
1151                         wentry->offset = lte->offset_in_res;
1152                 } else {
1153                         wentry->compressed_size = lte->rspec->size_in_wim;
1154                         wentry->offset = lte->rspec->offset_in_wim;
1155                 }
1156                 wentry->raw_resource_offset_in_wim = lte->rspec->offset_in_wim;
1157                 /*wentry->raw_resource_uncompressed_size = lte->rspec->uncompressed_size;*/
1158                 wentry->raw_resource_compressed_size = lte->rspec->size_in_wim;
1159         }
1160         copy_hash(wentry->sha1_hash, lte->hash);
1161         wentry->reference_count = lte->refcnt;
1162         wentry->is_compressed = (lte->flags & WIM_RESHDR_FLAG_COMPRESSED) != 0;
1163         wentry->is_metadata = (lte->flags & WIM_RESHDR_FLAG_METADATA) != 0;
1164         wentry->is_free = (lte->flags & WIM_RESHDR_FLAG_FREE) != 0;
1165         wentry->is_spanned = (lte->flags & WIM_RESHDR_FLAG_SPANNED) != 0;
1166         wentry->packed = (lte->flags & WIM_RESHDR_FLAG_PACKED_STREAMS) != 0;
1167 }
1168
1169 struct iterate_lte_context {
1170         wimlib_iterate_lookup_table_callback_t cb;
1171         void *user_ctx;
1172 };
1173
1174 static int
1175 do_iterate_lte(struct wim_lookup_table_entry *lte, void *_ctx)
1176 {
1177         struct iterate_lte_context *ctx = _ctx;
1178         struct wimlib_resource_entry entry;
1179
1180         lte_to_wimlib_resource_entry(lte, &entry);
1181         return (*ctx->cb)(&entry, ctx->user_ctx);
1182 }
1183
1184 /* API function documented in wimlib.h  */
1185 WIMLIBAPI int
1186 wimlib_iterate_lookup_table(WIMStruct *wim, int flags,
1187                             wimlib_iterate_lookup_table_callback_t cb,
1188                             void *user_ctx)
1189 {
1190         if (flags != 0)
1191                 return WIMLIB_ERR_INVALID_PARAM;
1192
1193         struct iterate_lte_context ctx = {
1194                 .cb = cb,
1195                 .user_ctx = user_ctx,
1196         };
1197         if (wim->hdr.part_number == 1) {
1198                 int ret;
1199                 for (int i = 0; i < wim->hdr.image_count; i++) {
1200                         ret = do_iterate_lte(wim->image_metadata[i]->metadata_lte,
1201                                              &ctx);
1202                         if (ret)
1203                                 return ret;
1204                 }
1205         }
1206         return for_lookup_table_entry(wim->lookup_table, do_iterate_lte, &ctx);
1207 }