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.
   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 #include "wimlib_internal.h"
  28 #include "lookup_table.h"
  29 #include "buffer_io.h"
  30 #include <errno.h>
  31 #include <stdlib.h>
  32
  33 #ifdef WITH_FUSE
  34 #include <unistd.h>
  35 #endif
  36
  37 struct wim_lookup_table *
  38 new_lookup_table(size_t capacity)
  39 {
  40         struct wim_lookup_table *table;
  41         struct hlist_head *array;
  42
  43         table = CALLOC(1, sizeof(struct wim_lookup_table));
  44         if (table) {
  45                 array = CALLOC(capacity, sizeof(array[0]));
  46                 if (array) {
  47                         table->num_entries = 0;
  48                         table->capacity = capacity;
  49                         table->array = array;
  50                 } else {
  51                         FREE(table);
  52                         table = NULL;
  53                         ERROR("Failed to allocate memory for lookup table with capacity %zu",
  54                               capacity);
  55                 }
  56         }
  57         return table;
  58 }
  59
  60 struct wim_lookup_table_entry *
  61 new_lookup_table_entry()
  62 {
  63         struct wim_lookup_table_entry *lte;
  64
  65         lte = CALLOC(1, sizeof(struct wim_lookup_table_entry));
  66         if (lte) {
  67                 lte->part_number  = 1;
  68                 lte->refcnt       = 1;
  69         } else {
  70                 ERROR("Out of memory (tried to allocate %zu bytes for "
  71                       "lookup table entry)",
  72                       sizeof(struct wim_lookup_table_entry));
  73         }
  74         return lte;
  75 }
  76
  77 struct wim_lookup_table_entry *
  78 clone_lookup_table_entry(const struct wim_lookup_table_entry *old)
  79 {
  80         struct wim_lookup_table_entry *new;
  81
  82         new = MALLOC(sizeof(*new));
  83         if (!new)
  84                 return NULL;
  85
  86         memcpy(new, old, sizeof(*old));
  87         new->extracted_file = NULL;
  88         switch (new->resource_location) {
  89 #ifdef __WIN32__
  90         case RESOURCE_WIN32:
  91         case RESOURCE_WIN32_ENCRYPTED:
  92 #else
  93         case RESOURCE_IN_FILE_ON_DISK:
  94 #endif
  95 #ifdef WITH_FUSE
  96         case RESOURCE_IN_STAGING_FILE:
  97                 BUILD_BUG_ON((void*)&old->file_on_disk !=
  98                              (void*)&old->staging_file_name);
  99 #endif
 100                 new->file_on_disk = TSTRDUP(old->file_on_disk);
 101                 if (!new->file_on_disk)
 102                         goto out_free;
 103                 break;
 104         case RESOURCE_IN_ATTACHED_BUFFER:
 105                 new->attached_buffer = MALLOC(wim_resource_size(old));
 106                 if (!new->attached_buffer)
 107                         goto out_free;
 108                 memcpy(new->attached_buffer, old->attached_buffer,
 109                        wim_resource_size(old));
 110                 break;
 111 #ifdef WITH_NTFS_3G
 112         case RESOURCE_IN_NTFS_VOLUME:
 113                 if (old->ntfs_loc) {
 114                         struct ntfs_location *loc;
 115                         loc = MALLOC(sizeof(*loc));
 116                         if (!loc)
 117                                 goto out_free;
 118                         memcpy(loc, old->ntfs_loc, sizeof(*loc));
 119                         loc->path = NULL;
 120                         loc->stream_name = NULL;
 121                         new->ntfs_loc = loc;
 122                         loc->path = STRDUP(old->ntfs_loc->path);
 123                         if (!loc->path)
 124                                 goto out_free;
 125                         loc->stream_name = MALLOC((loc->stream_name_nchars + 1) * 2);
 126                         if (!loc->stream_name)
 127                                 goto out_free;
 128                         memcpy(loc->stream_name,
 129                                old->ntfs_loc->stream_name,
 130                                (loc->stream_name_nchars + 1) * 2);
 131                 }
 132                 break;
 133 #endif
 134         default:
 135                 break;
 136         }
 137         return new;
 138 out_free:
 139         free_lookup_table_entry(new);
 140         return NULL;
 141 }
 142
 143 void
 144 free_lookup_table_entry(struct wim_lookup_table_entry *lte)
 145 {
 146         if (lte) {
 147                 switch (lte->resource_location) {
 148         #ifdef __WIN32__
 149                 case RESOURCE_WIN32:
 150                 case RESOURCE_WIN32_ENCRYPTED:
 151         #else
 152                 case RESOURCE_IN_FILE_ON_DISK:
 153         #endif
 154         #ifdef WITH_FUSE
 155                 case RESOURCE_IN_STAGING_FILE:
 156                         BUILD_BUG_ON((void*)&lte->file_on_disk !=
 157                                      (void*)&lte->staging_file_name);
 158         #endif
 159                 case RESOURCE_IN_ATTACHED_BUFFER:
 160                         BUILD_BUG_ON((void*)&lte->file_on_disk !=
 161                                      (void*)&lte->attached_buffer);
 162                         FREE(lte->file_on_disk);
 163                         break;
 164 #ifdef WITH_NTFS_3G
 165                 case RESOURCE_IN_NTFS_VOLUME:
 166                         if (lte->ntfs_loc) {
 167                                 FREE(lte->ntfs_loc->path);
 168                                 FREE(lte->ntfs_loc->stream_name);
 169                                 FREE(lte->ntfs_loc);
 170                         }
 171                         break;
 172 #endif
 173                 default:
 174                         break;
 175                 }
 176                 FREE(lte);
 177         }
 178 }
 179
 180 static int
 181 do_free_lookup_table_entry(struct wim_lookup_table_entry *entry, void *ignore)
 182 {
 183         free_lookup_table_entry(entry);
 184         return 0;
 185 }
 186
 187
 188 void
 189 free_lookup_table(struct wim_lookup_table *table)
 190 {
 191         DEBUG2("Freeing lookup table");
 192         if (table) {
 193                 if (table->array) {
 194                         for_lookup_table_entry(table,
 195                                                do_free_lookup_table_entry,
 196                                                NULL);
 197                         FREE(table->array);
 198                 }
 199                 FREE(table);
 200         }
 201 }
 202
 203 /*
 204  * Inserts an entry into the lookup table.
 205  *
 206  * @table:      A pointer to the lookup table.
 207  * @lte:        A pointer to the entry to insert.
 208  */
 209 void
 210 lookup_table_insert(struct wim_lookup_table *table,
 211                     struct wim_lookup_table_entry *lte)
 212 {
 213         size_t i = lte->hash_short % table->capacity;
 214         hlist_add_head(&lte->hash_list, &table->array[i]);
 215
 216         /* XXX Make the table grow when too many entries have been inserted. */
 217         table->num_entries++;
 218 }
 219
 220 static void
 221 finalize_lte(struct wim_lookup_table_entry *lte)
 222 {
 223         #ifdef WITH_FUSE
 224         if (lte->resource_location == RESOURCE_IN_STAGING_FILE) {
 225                 unlink(lte->staging_file_name);
 226                 list_del(&lte->unhashed_list);
 227         }
 228         #endif
 229         free_lookup_table_entry(lte);
 230 }
 231
 232 /* Decrements the reference count for the lookup table entry @lte.  If its
 233  * reference count reaches 0, it is unlinked from the lookup table.  If,
 234  * furthermore, the entry has no opened file descriptors associated with it, the
 235  * entry is freed.  */
 236 void
 237 lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
 238                      struct wim_lookup_table *table)
 239 {
 240         wimlib_assert(lte != NULL);
 241         wimlib_assert(lte->refcnt != 0);
 242         if (--lte->refcnt == 0) {
 243                 if (!lte->unhashed)
 244                         lookup_table_unlink(table, lte);
 245         #ifdef WITH_FUSE
 246                 if (lte->num_opened_fds == 0)
 247         #endif
 248                         finalize_lte(lte);
 249         }
 250 }
 251
 252 #ifdef WITH_FUSE
 253 void
 254 lte_decrement_num_opened_fds(struct wim_lookup_table_entry *lte)
 255 {
 256         if (lte->num_opened_fds != 0)
 257                 if (--lte->num_opened_fds == 0 && lte->refcnt == 0)
 258                         finalize_lte(lte);
 259 }
 260 #endif
 261
 262 /* Calls a function on all the entries in the WIM lookup table.  Stop early and
 263  * return nonzero if any call to the function returns nonzero. */
 264 int
 265 for_lookup_table_entry(struct wim_lookup_table *table,
 266                        int (*visitor)(struct wim_lookup_table_entry *, void *),
 267                        void *arg)
 268 {
 269         struct wim_lookup_table_entry *lte;
 270         struct hlist_node *pos, *tmp;
 271         int ret;
 272
 273         for (size_t i = 0; i < table->capacity; i++) {
 274                 hlist_for_each_entry_safe(lte, pos, tmp, &table->array[i],
 275                                           hash_list)
 276                 {
 277                         wimlib_assert2(!(lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA));
 278                         ret = visitor(lte, arg);
 279                         if (ret)
 280                                 return ret;
 281                 }
 282         }
 283         return 0;
 284 }
 285
 286 int
 287 cmp_streams_by_wim_position(const void *p1, const void *p2)
 288 {
 289         const struct wim_lookup_table_entry *lte1, *lte2;
 290         lte1 = *(const struct wim_lookup_table_entry**)p1;
 291         lte2 = *(const struct wim_lookup_table_entry**)p2;
 292         if (lte1->resource_entry.offset < lte2->resource_entry.offset)
 293                 return -1;
 294         else if (lte1->resource_entry.offset > lte2->resource_entry.offset)
 295                 return 1;
 296         else
 297                 return 0;
 298 }
 299
 300
 301 static int
 302 add_lte_to_array(struct wim_lookup_table_entry *lte,
 303                  void *_pp)
 304 {
 305         struct wim_lookup_table_entry ***pp = _pp;
 306         *(*pp)++ = lte;
 307         return 0;
 308 }
 309
 310 /* Iterate through the lookup table entries, but first sort them by stream
 311  * offset in the WIM.  Caution: this is intended to be used when the stream
 312  * offset field has actually been set. */
 313 int
 314 for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
 315                                   int (*visitor)(struct wim_lookup_table_entry *,
 316                                                  void *),
 317                                   void *arg)
 318 {
 319         struct wim_lookup_table_entry **lte_array, **p;
 320         size_t num_streams = table->num_entries;
 321         int ret;
 322
 323         lte_array = MALLOC(num_streams * sizeof(lte_array[0]));
 324         if (!lte_array)
 325                 return WIMLIB_ERR_NOMEM;
 326         p = lte_array;
 327         for_lookup_table_entry(table, add_lte_to_array, &p);
 328
 329         wimlib_assert(p == lte_array + num_streams);
 330
 331         qsort(lte_array, num_streams, sizeof(lte_array[0]),
 332               cmp_streams_by_wim_position);
 333         ret = 0;
 334         for (size_t i = 0; i < num_streams; i++) {
 335                 ret = visitor(lte_array[i], arg);
 336                 if (ret)
 337                         break;
 338         }
 339         FREE(lte_array);
 340         return ret;
 341 }
 342
 343 /*
 344  * Reads the lookup table from a WIM file.
 345  *
 346  * Saves lookup table entries for non-metadata streams in a hash table, and
 347  * saves the metadata entry for each image in a special per-image location (the
 348  * image_metadata array).
 349  */
 350 int
 351 read_lookup_table(WIMStruct *w)
 352 {
 353         u64 num_entries;
 354         int ret;
 355         struct wim_lookup_table *table;
 356         struct wim_lookup_table_entry *cur_entry, *duplicate_entry;
 357         void *table_buf;
 358         size_t table_size;
 359         const void *p;
 360
 361         DEBUG("Reading lookup table: offset %"PRIu64", size %"PRIu64"",
 362               w->hdr.lookup_table_res_entry.offset,
 363               w->hdr.lookup_table_res_entry.original_size);
 364
 365         if (resource_is_compressed(&w->hdr.lookup_table_res_entry)) {
 366                 ERROR("Didn't expect a compressed lookup table!");
 367                 ERROR("Ask the author to implement support for this.");
 368                 return WIMLIB_ERR_COMPRESSED_LOOKUP_TABLE;
 369         }
 370
 371         table_size = w->hdr.lookup_table_res_entry.size;
 372         if (table_size != w->hdr.lookup_table_res_entry.size) {
 373                 ERROR("Lookup table is invalid");
 374                 return WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 375         }
 376
 377         table_buf = MALLOC(table_size);
 378         if (!table_buf)
 379                 return WIMLIB_ERR_NOMEM;
 380
 381         if (full_pread(w->in_fd, table_buf, table_size,
 382                        w->hdr.lookup_table_res_entry.offset) != table_size)
 383         {
 384                 ret = WIMLIB_ERR_READ;
 385                 goto out_free_table_buf;
 386         }
 387
 388         num_entries = w->hdr.lookup_table_res_entry.original_size /
 389                       WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE;
 390         table = new_lookup_table(num_entries * 2 + 1);
 391         if (!table) {
 392                 ret = WIMLIB_ERR_NOMEM;
 393                 goto out_free_table_buf;
 394         }
 395
 396         w->current_image = 0;
 397         p = table_buf;
 398         while (num_entries--) {
 399                 cur_entry = new_lookup_table_entry();
 400                 if (!cur_entry) {
 401                         ret = WIMLIB_ERR_NOMEM;
 402                         goto out_free_lookup_table;
 403                 }
 404
 405                 cur_entry->wim = w;
 406                 cur_entry->resource_location = RESOURCE_IN_WIM;
 407                 p = get_resource_entry(p, &cur_entry->resource_entry);
 408                 p = get_u16(p, &cur_entry->part_number);
 409                 p = get_u32(p, &cur_entry->refcnt);
 410                 p = get_bytes(p, SHA1_HASH_SIZE, cur_entry->hash);
 411
 412                 if (cur_entry->part_number != w->hdr.part_number) {
 413                         ERROR("A lookup table entry in part %hu of the WIM "
 414                               "points to part %hu",
 415                               w->hdr.part_number, cur_entry->part_number);
 416                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 417                         goto out_free_cur_entry;
 418                 }
 419
 420                 if (is_zero_hash(cur_entry->hash)) {
 421                         ERROR("The WIM lookup table contains an entry with a "
 422                               "SHA1 message digest of all 0's");
 423                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 424                         goto out_free_cur_entry;
 425                 }
 426
 427                 if (!(cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
 428                     && (cur_entry->resource_entry.size !=
 429                         cur_entry->resource_entry.original_size))
 430                 {
 431                 #ifdef ENABLE_ERROR_MESSAGES
 432                         ERROR("Found uncompressed resource with original size "
 433                               "not the same as compressed size");
 434                         ERROR("The lookup table entry for the resource is as follows:");
 435                         print_lookup_table_entry(cur_entry, stderr);
 436                 #endif
 437                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 438                         goto out_free_cur_entry;
 439                 }
 440
 441                 if (cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_METADATA) {
 442                         /* Lookup table entry for a metadata resource */
 443                         if (cur_entry->refcnt != 1) {
 444                         #ifdef ENABLE_ERROR_MESSAGES
 445                                 ERROR("Found metadata resource with refcnt != 1:");
 446                                 print_lookup_table_entry(cur_entry, stderr);
 447                         #endif
 448                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 449                                 goto out_free_cur_entry;
 450                         }
 451
 452                         if (w->hdr.part_number != 1) {
 453                                 ERROR("Found a metadata resource in a "
 454                                       "non-first part of the split WIM!");
 455                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 456                                 goto out_free_cur_entry;
 457                         }
 458                         if (w->current_image == w->hdr.image_count) {
 459                                 ERROR("The WIM header says there are %u images "
 460                                       "in the WIM, but we found more metadata "
 461                                       "resources than this", w->hdr.image_count);
 462                                 ret = WIMLIB_ERR_IMAGE_COUNT;
 463                                 goto out_free_cur_entry;
 464                         }
 465
 466                         /* Notice very carefully:  We are assigning the metadata
 467                          * resources in the exact order mirrored by their lookup
 468                          * table entries on disk, which is the behavior of
 469                          * Microsoft's software.  In particular, this overrides
 470                          * the actual locations of the metadata resources
 471                          * themselves in the WIM file as well as any information
 472                          * written in the XML data. */
 473                         DEBUG("Found metadata resource for image %u at "
 474                               "offset %"PRIu64".",
 475                               w->current_image + 1,
 476                               cur_entry->resource_entry.offset);
 477                         w->image_metadata[
 478                                 w->current_image++]->metadata_lte = cur_entry;
 479                 } else {
 480                         /* Lookup table entry for a stream that is not a
 481                          * metadata resource */
 482                         duplicate_entry = __lookup_resource(table, cur_entry->hash);
 483                         if (duplicate_entry) {
 484                         #ifdef ENABLE_ERROR_MESSAGES
 485                                 ERROR("The WIM lookup table contains two entries with the "
 486                                       "same SHA1 message digest!");
 487                                 ERROR("The first entry is:");
 488                                 print_lookup_table_entry(duplicate_entry, stderr);
 489                                 ERROR("The second entry is:");
 490                                 print_lookup_table_entry(cur_entry, stderr);
 491                         #endif
 492                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 493                                 goto out_free_cur_entry;
 494                         }
 495                         lookup_table_insert(table, cur_entry);
 496                 }
 497         }
 498
 499         if (w->hdr.part_number == 1 &&
 500             w->current_image != w->hdr.image_count)
 501         {
 502                 ERROR("The WIM header says there are %u images "
 503                       "in the WIM, but we only found %d metadata "
 504                       "resources!", w->hdr.image_count, w->current_image);
 505                 ret = WIMLIB_ERR_IMAGE_COUNT;
 506                 goto out_free_lookup_table;
 507         }
 508         DEBUG("Done reading lookup table.");
 509         w->lookup_table = table;
 510         ret = 0;
 511         goto out_free_table_buf;
 512 out_free_cur_entry:
 513         FREE(cur_entry);
 514 out_free_lookup_table:
 515         free_lookup_table(table);
 516 out_free_table_buf:
 517         FREE(table_buf);
 518         w->current_image = 0;
 519         return ret;
 520 }
 521
 522
 523 /*
 524  * Writes a lookup table entry to the output file.
 525  */
 526 static void *
 527 write_lookup_table_entry(struct wim_lookup_table_entry *lte, void *buf_p)
 528 {
 529         buf_p = put_resource_entry(buf_p, &lte->output_resource_entry);
 530         buf_p = put_u16(buf_p, lte->part_number);
 531         buf_p = put_u32(buf_p, lte->out_refcnt);
 532         buf_p = put_bytes(buf_p, SHA1_HASH_SIZE, lte->hash);
 533         return buf_p;
 534 }
 535
 536 int
 537 write_lookup_table_from_stream_list(struct list_head *stream_list,
 538                                     filedes_t out_fd,
 539                                     struct resource_entry *out_res_entry)
 540 {
 541         size_t num_entries;
 542         struct list_head *pos;
 543         size_t table_size;
 544         void *table_buf;
 545         void *buf_p;
 546         struct wim_lookup_table_entry *lte;
 547         off_t start_offset;
 548         int ret;
 549
 550         start_offset = filedes_offset(out_fd);
 551         if (start_offset == -1)
 552                 return WIMLIB_ERR_WRITE;
 553
 554         num_entries = 0;
 555         list_for_each(pos, stream_list)
 556                 num_entries++;
 557         table_size = num_entries * WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE;
 558
 559         table_buf = MALLOC(table_size);
 560         if (!table_buf) {
 561                 ERROR("Failed to allocate lookup table buffer of %zu bytes", table_size);
 562                 return WIMLIB_ERR_NOMEM;
 563         }
 564
 565         buf_p = table_buf;
 566         list_for_each_entry(lte, stream_list, lookup_table_list)
 567                 buf_p =  write_lookup_table_entry(lte, buf_p);
 568
 569         wimlib_assert(buf_p - table_buf == table_size);
 570
 571         if (full_write(out_fd, table_buf, table_size) != table_size) {
 572                 ERROR_WITH_ERRNO("Failed to write lookup table");
 573                 ret = WIMLIB_ERR_WRITE;
 574                 goto out_free_table_buf;
 575         }
 576
 577         out_res_entry->offset        = start_offset;
 578         out_res_entry->size          = table_size;
 579         out_res_entry->original_size = table_size;
 580         out_res_entry->flags         = WIM_RESHDR_FLAG_METADATA;
 581         ret = 0;
 582 out_free_table_buf:
 583         FREE(table_buf);
 584         return ret;
 585 }
 586
 587 static int
 588 append_lookup_table_entry(struct wim_lookup_table_entry *lte, void *_list)
 589 {
 590         struct list_head *list = _list;
 591
 592         if (lte->out_refcnt != 0)
 593                 list_add_tail(&lte->lookup_table_list, list);
 594         return 0;
 595 }
 596
 597
 598 /* Writes the WIM lookup table to the output file. */
 599 int
 600 write_lookup_table(WIMStruct *w, int image, struct resource_entry *out_res_entry)
 601 {
 602         LIST_HEAD(stream_list);
 603         int start_image;
 604         int end_image;
 605
 606         if (image == WIMLIB_ALL_IMAGES) {
 607                 start_image = 1;
 608                 end_image = w->hdr.image_count;
 609         } else {
 610                 start_image = image;
 611                 end_image = image;
 612         }
 613
 614         for (int i = start_image; i <= end_image; i++) {
 615                 struct wim_lookup_table_entry *metadata_lte;
 616
 617                 metadata_lte = w->image_metadata[i - 1]->metadata_lte;
 618                 metadata_lte->out_refcnt = 1;
 619                 metadata_lte->output_resource_entry.flags |= WIM_RESHDR_FLAG_METADATA;
 620                 append_lookup_table_entry(metadata_lte, &stream_list);
 621         }
 622
 623         for_lookup_table_entry(w->lookup_table,
 624                                append_lookup_table_entry,
 625                                &stream_list);
 626         return write_lookup_table_from_stream_list(&stream_list,
 627                                                    w->out_fd,
 628                                                    out_res_entry);
 629 }
 630
 631 int
 632 lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 633 {
 634         lte->real_refcnt = 0;
 635         return 0;
 636 }
 637
 638 int
 639 lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 640 {
 641         lte->out_refcnt = 0;
 642         return 0;
 643 }
 644
 645 int
 646 lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *_ignore)
 647 {
 648         if (lte->extracted_file != NULL) {
 649                 FREE(lte->extracted_file);
 650                 lte->extracted_file = NULL;
 651         }
 652         return 0;
 653 }
 654
 655 void
 656 print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 657 {
 658         if (!lte) {
 659                 tputc(T('\n'), out);
 660                 return;
 661         }
 662         tfprintf(out, T("Offset            = %"PRIu64" bytes\n"),
 663                  lte->resource_entry.offset);
 664
 665         tfprintf(out, T("Size              = %"PRIu64" bytes\n"),
 666                  (u64)lte->resource_entry.size);
 667
 668         tfprintf(out, T("Original size     = %"PRIu64" bytes\n"),
 669                  lte->resource_entry.original_size);
 670
 671         tfprintf(out, T("Part Number       = %hu\n"), lte->part_number);
 672         tfprintf(out, T("Reference Count   = %u\n"), lte->refcnt);
 673
 674         if (lte->unhashed) {
 675                 tfprintf(out, T("(Unhashed: inode %p, stream_id = %u)\n"),
 676                          lte->back_inode, lte->back_stream_id);
 677         } else {
 678                 tfprintf(out, T("Hash              = 0x"));
 679                 print_hash(lte->hash, out);
 680                 tputc(T('\n'), out);
 681         }
 682
 683         tfprintf(out, T("Flags             = "));
 684         u8 flags = lte->resource_entry.flags;
 685         if (flags & WIM_RESHDR_FLAG_COMPRESSED)
 686                 tfputs(T("WIM_RESHDR_FLAG_COMPRESSED, "), out);
 687         if (flags & WIM_RESHDR_FLAG_FREE)
 688                 tfputs(T("WIM_RESHDR_FLAG_FREE, "), out);
 689         if (flags & WIM_RESHDR_FLAG_METADATA)
 690                 tfputs(T("WIM_RESHDR_FLAG_METADATA, "), out);
 691         if (flags & WIM_RESHDR_FLAG_SPANNED)
 692                 tfputs(T("WIM_RESHDR_FLAG_SPANNED, "), out);
 693         tputc(T('\n'), out);
 694         switch (lte->resource_location) {
 695         case RESOURCE_IN_WIM:
 696                 if (lte->wim->filename) {
 697                         tfprintf(out, T("WIM file          = `%"TS"'\n"),
 698                                  lte->wim->filename);
 699                 }
 700                 break;
 701 #ifdef __WIN32__
 702         case RESOURCE_WIN32:
 703         case RESOURCE_WIN32_ENCRYPTED:
 704 #else
 705         case RESOURCE_IN_FILE_ON_DISK:
 706 #endif
 707                 tfprintf(out, T("File on Disk      = `%"TS"'\n"),
 708                          lte->file_on_disk);
 709                 break;
 710 #ifdef WITH_FUSE
 711         case RESOURCE_IN_STAGING_FILE:
 712                 tfprintf(out, T("Staging File      = `%"TS"'\n"),
 713                                 lte->staging_file_name);
 714                 break;
 715 #endif
 716         default:
 717                 break;
 718         }
 719         tputc(T('\n'), out);
 720 }
 721
 722 static int
 723 do_print_lookup_table_entry(struct wim_lookup_table_entry *lte, void *fp)
 724 {
 725         print_lookup_table_entry(lte, (FILE*)fp);
 726         return 0;
 727 }
 728
 729 /*
 730  * Prints the lookup table of a WIM file.
 731  */
 732 WIMLIBAPI void
 733 wimlib_print_lookup_table(WIMStruct *w)
 734 {
 735         for_lookup_table_entry(w->lookup_table,
 736                                do_print_lookup_table_entry,
 737                                stdout);
 738 }
 739
 740 /* Given a SHA1 message digest, return the corresponding entry in the WIM's
 741  * lookup table, or NULL if there is none.  */
 742 struct wim_lookup_table_entry *
 743 __lookup_resource(const struct wim_lookup_table *table, const u8 hash[])
 744 {
 745         size_t i;
 746         struct wim_lookup_table_entry *lte;
 747         struct hlist_node *pos;
 748
 749         wimlib_assert(table != NULL);
 750         wimlib_assert(hash != NULL);
 751
 752         i = *(size_t*)hash % table->capacity;
 753         hlist_for_each_entry(lte, pos, &table->array[i], hash_list)
 754                 if (hashes_equal(hash, lte->hash))
 755                         return lte;
 756         return NULL;
 757 }
 758
 759 #ifdef WITH_FUSE
 760 /*
 761  * Finds the dentry, lookup table entry, and stream index for a WIM file stream,
 762  * given a path name.
 763  *
 764  * This is only for pre-resolved inodes.
 765  */
 766 int
 767 lookup_resource(WIMStruct *w,
 768                 const tchar *path,
 769                 int lookup_flags,
 770                 struct wim_dentry **dentry_ret,
 771                 struct wim_lookup_table_entry **lte_ret,
 772                 u16 *stream_idx_ret)
 773 {
 774         struct wim_dentry *dentry;
 775         struct wim_lookup_table_entry *lte;
 776         u16 stream_idx;
 777         const tchar *stream_name = NULL;
 778         struct wim_inode *inode;
 779         tchar *p = NULL;
 780
 781         if (lookup_flags & LOOKUP_FLAG_ADS_OK) {
 782                 stream_name = path_stream_name(path);
 783                 if (stream_name) {
 784                         p = (tchar*)stream_name - 1;
 785                         *p = T('\0');
 786                 }
 787         }
 788
 789         dentry = get_dentry(w, path);
 790         if (p)
 791                 *p = T(':');
 792         if (!dentry)
 793                 return -errno;
 794
 795         inode = dentry->d_inode;
 796
 797         wimlib_assert(inode->i_resolved);
 798
 799         if (!(lookup_flags & LOOKUP_FLAG_DIRECTORY_OK)
 800               && inode_is_directory(inode))
 801                 return -EISDIR;
 802
 803         if (stream_name) {
 804                 struct wim_ads_entry *ads_entry;
 805                 u16 ads_idx;
 806                 ads_entry = inode_get_ads_entry(inode, stream_name,
 807                                                 &ads_idx);
 808                 if (ads_entry) {
 809                         stream_idx = ads_idx + 1;
 810                         lte = ads_entry->lte;
 811                         goto out;
 812                 } else {
 813                         return -ENOENT;
 814                 }
 815         } else {
 816                 lte = inode->i_lte;
 817                 stream_idx = 0;
 818         }
 819 out:
 820         if (dentry_ret)
 821                 *dentry_ret = dentry;
 822         if (lte_ret)
 823                 *lte_ret = lte;
 824         if (stream_idx_ret)
 825                 *stream_idx_ret = stream_idx;
 826         return 0;
 827 }
 828 #endif
 829
 830 /* Resolve an inode's lookup table entries
 831  *
 832  * This replaces the SHA1 hash fields (which are used to lookup an entry in the
 833  * lookup table) with pointers directly to the lookup table entries.  A circular
 834  * linked list of streams sharing the same lookup table entry is created.
 835  *
 836  * This function always succeeds; unresolved lookup table entries are given a
 837  * NULL pointer.
 838  */
 839 void
 840 inode_resolve_ltes(struct wim_inode *inode, struct wim_lookup_table *table)
 841 {
 842
 843         if (!inode->i_resolved) {
 844                 struct wim_lookup_table_entry *lte;
 845                 /* Resolve the default file stream */
 846                 lte = __lookup_resource(table, inode->i_hash);
 847                 inode->i_lte = lte;
 848                 inode->i_resolved = 1;
 849
 850                 /* Resolve the alternate data streams */
 851                 for (u16 i = 0; i < inode->i_num_ads; i++) {
 852                         struct wim_ads_entry *cur_entry = &inode->i_ads_entries[i];
 853                         lte = __lookup_resource(table, cur_entry->hash);
 854                         cur_entry->lte = lte;
 855                 }
 856         }
 857 }
 858
 859 void
 860 inode_unresolve_ltes(struct wim_inode *inode)
 861 {
 862         if (inode->i_resolved) {
 863                 if (inode->i_lte)
 864                         copy_hash(inode->i_hash, inode->i_lte->hash);
 865                 else
 866                         zero_out_hash(inode->i_hash);
 867
 868                 for (u16 i = 0; i < inode->i_num_ads; i++) {
 869                         if (inode->i_ads_entries[i].lte)
 870                                 copy_hash(inode->i_ads_entries[i].hash,
 871                                           inode->i_ads_entries[i].lte->hash);
 872                         else
 873                                 zero_out_hash(inode->i_ads_entries[i].hash);
 874                 }
 875                 inode->i_resolved = 0;
 876         }
 877 }
 878
 879 /*
 880  * Returns the lookup table entry for stream @stream_idx of the inode, where
 881  * stream_idx = 0 means the default un-named file stream, and stream_idx >= 1
 882  * corresponds to an alternate data stream.
 883  *
 884  * This works for both resolved and un-resolved inodes.
 885  */
 886 struct wim_lookup_table_entry *
 887 inode_stream_lte(const struct wim_inode *inode, unsigned stream_idx,
 888                  const struct wim_lookup_table *table)
 889 {
 890         if (inode->i_resolved)
 891                 return inode_stream_lte_resolved(inode, stream_idx);
 892         else
 893                 return inode_stream_lte_unresolved(inode, stream_idx, table);
 894 }
 895
 896 struct wim_lookup_table_entry *
 897 inode_unnamed_lte_resolved(const struct wim_inode *inode)
 898 {
 899         wimlib_assert(inode->i_resolved);
 900         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
 901                 if (inode_stream_name_nbytes(inode, i) == 0 &&
 902                     !is_zero_hash(inode_stream_hash_resolved(inode, i)))
 903                 {
 904                         return inode_stream_lte_resolved(inode, i);
 905                 }
 906         }
 907         return NULL;
 908 }
 909
 910 struct wim_lookup_table_entry *
 911 inode_unnamed_lte_unresolved(const struct wim_inode *inode,
 912                              const struct wim_lookup_table *table)
 913 {
 914         wimlib_assert(!inode->i_resolved);
 915         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
 916                 if (inode_stream_name_nbytes(inode, i) == 0 &&
 917                     !is_zero_hash(inode_stream_hash_unresolved(inode, i)))
 918                 {
 919                         return inode_stream_lte_unresolved(inode, i, table);
 920                 }
 921         }
 922         return NULL;
 923 }
 924
 925 /* Return the lookup table entry for the unnamed data stream of an inode, or
 926  * NULL if there is none.
 927  *
 928  * You'd think this would be easier than it actually is, since the unnamed data
 929  * stream should be the one referenced from the inode itself.  Alas, if there
 930  * are named data streams, Microsoft's "imagex.exe" program will put the unnamed
 931  * data stream in one of the alternate data streams instead of inside the WIM
 932  * dentry itself.  So we need to check the alternate data streams too.
 933  *
 934  * Also, note that a dentry may appear to have more than one unnamed stream, but
 935  * if the SHA1 message digest is all 0's then the corresponding stream does not
 936  * really "count" (this is the case for the inode's own file stream when the
 937  * file stream that should be there is actually in one of the alternate stream
 938  * entries.).  This is despite the fact that we may need to extract such a
 939  * missing entry as an empty file or empty named data stream.
 940  */
 941 struct wim_lookup_table_entry *
 942 inode_unnamed_lte(const struct wim_inode *inode,
 943                   const struct wim_lookup_table *table)
 944 {
 945         if (inode->i_resolved)
 946                 return inode_unnamed_lte_resolved(inode);
 947         else
 948                 return inode_unnamed_lte_unresolved(inode, table);
 949 }
 950
 951 static int
 952 lte_add_stream_size(struct wim_lookup_table_entry *lte, void *total_bytes_p)
 953 {
 954         *(u64*)total_bytes_p += lte->resource_entry.size;
 955         return 0;
 956 }
 957
 958 u64
 959 lookup_table_total_stream_size(struct wim_lookup_table *table)
 960 {
 961         u64 total_size = 0;
 962         for_lookup_table_entry(table, lte_add_stream_size, &total_size);
 963         return total_size;
 964 }
 965
 966 struct wim_lookup_table_entry **
 967 retrieve_lte_pointer(struct wim_lookup_table_entry *lte)
 968 {
 969         wimlib_assert(lte->unhashed);
 970         struct wim_inode *inode = lte->back_inode;
 971         u32 stream_id = lte->back_stream_id;
 972         if (stream_id == 0)
 973                 return &inode->i_lte;
 974         else
 975                 for (u16 i = 0; i < inode->i_num_ads; i++)
 976                         if (inode->i_ads_entries[i].stream_id == stream_id)
 977                                 return &inode->i_ads_entries[i].lte;
 978         wimlib_assert(0);
 979         return NULL;
 980 }
 981
 982 /* Calculate the SHA1 message digest of a stream and move it from the list of
 983  * unhashed streams to the stream lookup table, possibly joining it with an
 984  * existing lookup table entry for an identical stream.
 985  *
 986  * @lte:  An unhashed lookup table entry.
 987  * @lookup_table:  Lookup table for the WIM.
 988  * @lte_ret:  On success, write a pointer to the resulting lookup table
 989  *            entry to this location.  This will be the same as @lte
 990  *            if it was inserted into the lookup table, or different if
 991  *            a duplicate stream was found.
 992  *
 993  * Returns 0 on success; nonzero if there is an error reading the stream.
 994  */
 995 int
 996 hash_unhashed_stream(struct wim_lookup_table_entry *lte,
 997                      struct wim_lookup_table *lookup_table,
 998                      struct wim_lookup_table_entry **lte_ret)
 999 {
1000         int ret;
1001         struct wim_lookup_table_entry *duplicate_lte;
1002         struct wim_lookup_table_entry **back_ptr;
1003
1004         wimlib_assert(lte->unhashed);
1005
1006         /* back_ptr must be saved because @back_inode and @back_stream_id are in
1007          * union with the SHA1 message digest and will no longer be valid once
1008          * the SHA1 has been calculated. */
1009         back_ptr = retrieve_lte_pointer(lte);
1010
1011         ret = sha1_resource(lte);
1012         if (ret)
1013                 return ret;
1014
1015         /* Look for a duplicate stream */
1016         duplicate_lte = __lookup_resource(lookup_table, lte->hash);
1017         list_del(&lte->unhashed_list);
1018         if (duplicate_lte) {
1019                 /* We have a duplicate stream.  Transfer the reference counts
1020                  * from this stream to the duplicate, update the reference to
1021                  * this stream (in an inode or ads_entry) to point to the
1022                  * duplicate, then free this stream. */
1023                 wimlib_assert(!(duplicate_lte->unhashed));
1024                 duplicate_lte->refcnt += lte->refcnt;
1025                 duplicate_lte->out_refcnt += lte->refcnt;
1026                 *back_ptr = duplicate_lte;
1027                 free_lookup_table_entry(lte);
1028                 lte = duplicate_lte;
1029         } else {
1030                 /* No duplicate stream, so we need to insert
1031                  * this stream into the lookup table and treat
1032                  * it as a hashed stream. */
1033                 lookup_table_insert(lookup_table, lte);
1034                 lte->unhashed = 0;
1035         }
1036         if (lte_ret)
1037                 *lte_ret = lte;
1038         return 0;
1039 }
1040