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 #ifdef HAVE_CONFIG_H
  28 #  include "config.h"
  29 #endif
  30
  31 #include "wimlib/buffer_io.h"
  32 #include "wimlib/error.h"
  33 #include "wimlib/file_io.h"
  34 #include "wimlib/lookup_table.h"
  35 #include "wimlib/metadata.h"
  36 #include "wimlib/paths.h"
  37 #include "wimlib/resource.h"
  38 #include "wimlib/util.h"
  39
  40 #include <errno.h>
  41 #include <stdlib.h>
  42 #ifdef WITH_FUSE
  43 #  include <unistd.h> /* for unlink() */
  44 #endif
  45
  46 /* Size of each lookup table entry in the WIM file. */
  47 #define WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE 50
  48
  49 struct wim_lookup_table *
  50 new_lookup_table(size_t capacity)
  51 {
  52         struct wim_lookup_table *table;
  53         struct hlist_head *array;
  54
  55         table = CALLOC(1, sizeof(struct wim_lookup_table));
  56         if (table) {
  57                 array = CALLOC(capacity, sizeof(array[0]));
  58                 if (array) {
  59                         table->num_entries = 0;
  60                         table->capacity = capacity;
  61                         table->array = array;
  62                 } else {
  63                         FREE(table);
  64                         table = NULL;
  65                         ERROR("Failed to allocate memory for lookup table "
  66                               "with capacity %zu", capacity);
  67                 }
  68         }
  69         return table;
  70 }
  71
  72 struct wim_lookup_table_entry *
  73 new_lookup_table_entry(void)
  74 {
  75         struct wim_lookup_table_entry *lte;
  76
  77         lte = CALLOC(1, sizeof(struct wim_lookup_table_entry));
  78         if (lte) {
  79                 lte->part_number  = 1;
  80                 lte->refcnt       = 1;
  81         } else {
  82                 ERROR("Out of memory (tried to allocate %zu bytes for "
  83                       "lookup table entry)",
  84                       sizeof(struct wim_lookup_table_entry));
  85         }
  86         return lte;
  87 }
  88
  89 struct wim_lookup_table_entry *
  90 clone_lookup_table_entry(const struct wim_lookup_table_entry *old)
  91 {
  92         struct wim_lookup_table_entry *new;
  93
  94         new = MALLOC(sizeof(*new));
  95         if (!new)
  96                 return NULL;
  97
  98         memcpy(new, old, sizeof(*old));
  99         new->extracted_file = NULL;
 100         switch (new->resource_location) {
 101 #ifdef __WIN32__
 102         case RESOURCE_WIN32:
 103         case RESOURCE_WIN32_ENCRYPTED:
 104 #else
 105         case RESOURCE_IN_FILE_ON_DISK:
 106 #endif
 107 #ifdef WITH_FUSE
 108         case RESOURCE_IN_STAGING_FILE:
 109                 BUILD_BUG_ON((void*)&old->file_on_disk !=
 110                              (void*)&old->staging_file_name);
 111 #endif
 112                 new->file_on_disk = TSTRDUP(old->file_on_disk);
 113                 if (!new->file_on_disk)
 114                         goto out_free;
 115                 break;
 116         case RESOURCE_IN_ATTACHED_BUFFER:
 117                 new->attached_buffer = MALLOC(wim_resource_size(old));
 118                 if (!new->attached_buffer)
 119                         goto out_free;
 120                 memcpy(new->attached_buffer, old->attached_buffer,
 121                        wim_resource_size(old));
 122                 break;
 123 #ifdef WITH_NTFS_3G
 124         case RESOURCE_IN_NTFS_VOLUME:
 125                 if (old->ntfs_loc) {
 126                         struct ntfs_location *loc;
 127                         loc = MALLOC(sizeof(*loc));
 128                         if (!loc)
 129                                 goto out_free;
 130                         memcpy(loc, old->ntfs_loc, sizeof(*loc));
 131                         loc->path = NULL;
 132                         loc->stream_name = NULL;
 133                         new->ntfs_loc = loc;
 134                         loc->path = STRDUP(old->ntfs_loc->path);
 135                         if (!loc->path)
 136                                 goto out_free;
 137                         loc->stream_name = MALLOC((loc->stream_name_nchars + 1) * 2);
 138                         if (!loc->stream_name)
 139                                 goto out_free;
 140                         memcpy(loc->stream_name,
 141                                old->ntfs_loc->stream_name,
 142                                (loc->stream_name_nchars + 1) * 2);
 143                 }
 144                 break;
 145 #endif
 146         default:
 147                 break;
 148         }
 149         return new;
 150 out_free:
 151         free_lookup_table_entry(new);
 152         return NULL;
 153 }
 154
 155 void
 156 free_lookup_table_entry(struct wim_lookup_table_entry *lte)
 157 {
 158         if (lte) {
 159                 switch (lte->resource_location) {
 160         #ifdef __WIN32__
 161                 case RESOURCE_WIN32:
 162                 case RESOURCE_WIN32_ENCRYPTED:
 163         #else
 164                 case RESOURCE_IN_FILE_ON_DISK:
 165         #endif
 166         #ifdef WITH_FUSE
 167                 case RESOURCE_IN_STAGING_FILE:
 168                         BUILD_BUG_ON((void*)&lte->file_on_disk !=
 169                                      (void*)&lte->staging_file_name);
 170         #endif
 171                 case RESOURCE_IN_ATTACHED_BUFFER:
 172                         BUILD_BUG_ON((void*)&lte->file_on_disk !=
 173                                      (void*)&lte->attached_buffer);
 174                         FREE(lte->file_on_disk);
 175                         break;
 176 #ifdef WITH_NTFS_3G
 177                 case RESOURCE_IN_NTFS_VOLUME:
 178                         if (lte->ntfs_loc) {
 179                                 FREE(lte->ntfs_loc->path);
 180                                 FREE(lte->ntfs_loc->stream_name);
 181                                 FREE(lte->ntfs_loc);
 182                         }
 183                         break;
 184 #endif
 185                 default:
 186                         break;
 187                 }
 188                 FREE(lte);
 189         }
 190 }
 191
 192 static int
 193 do_free_lookup_table_entry(struct wim_lookup_table_entry *entry, void *ignore)
 194 {
 195         free_lookup_table_entry(entry);
 196         return 0;
 197 }
 198
 199
 200 void
 201 free_lookup_table(struct wim_lookup_table *table)
 202 {
 203         DEBUG2("Freeing lookup table");
 204         if (table) {
 205                 if (table->array) {
 206                         for_lookup_table_entry(table,
 207                                                do_free_lookup_table_entry,
 208                                                NULL);
 209                         FREE(table->array);
 210                 }
 211                 FREE(table);
 212         }
 213 }
 214
 215 /*
 216  * Inserts an entry into the lookup table.
 217  *
 218  * @table:      A pointer to the lookup table.
 219  * @lte:        A pointer to the entry to insert.
 220  */
 221 void
 222 lookup_table_insert(struct wim_lookup_table *table,
 223                     struct wim_lookup_table_entry *lte)
 224 {
 225         size_t i = lte->hash_short % table->capacity;
 226         hlist_add_head(&lte->hash_list, &table->array[i]);
 227
 228         /* XXX Make the table grow when too many entries have been inserted. */
 229         table->num_entries++;
 230 }
 231
 232 static void
 233 finalize_lte(struct wim_lookup_table_entry *lte)
 234 {
 235         #ifdef WITH_FUSE
 236         if (lte->resource_location == RESOURCE_IN_STAGING_FILE) {
 237                 unlink(lte->staging_file_name);
 238                 list_del(&lte->unhashed_list);
 239         }
 240         #endif
 241         free_lookup_table_entry(lte);
 242 }
 243
 244 /* Decrements the reference count for the lookup table entry @lte.  If its
 245  * reference count reaches 0, it is unlinked from the lookup table.  If,
 246  * furthermore, the entry has no opened file descriptors associated with it, the
 247  * entry is freed.  */
 248 void
 249 lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
 250                      struct wim_lookup_table *table)
 251 {
 252         wimlib_assert(lte != NULL);
 253         wimlib_assert(lte->refcnt != 0);
 254         if (--lte->refcnt == 0) {
 255                 if (lte->unhashed)
 256                         list_del(&lte->unhashed_list);
 257                 else
 258                         lookup_table_unlink(table, lte);
 259         #ifdef WITH_FUSE
 260                 if (lte->num_opened_fds == 0)
 261         #endif
 262                         finalize_lte(lte);
 263         }
 264 }
 265
 266 #ifdef WITH_FUSE
 267 void
 268 lte_decrement_num_opened_fds(struct wim_lookup_table_entry *lte)
 269 {
 270         if (lte->num_opened_fds != 0)
 271                 if (--lte->num_opened_fds == 0 && lte->refcnt == 0)
 272                         finalize_lte(lte);
 273 }
 274 #endif
 275
 276 /* Calls a function on all the entries in the WIM lookup table.  Stop early and
 277  * return nonzero if any call to the function returns nonzero. */
 278 int
 279 for_lookup_table_entry(struct wim_lookup_table *table,
 280                        int (*visitor)(struct wim_lookup_table_entry *, void *),
 281                        void *arg)
 282 {
 283         struct wim_lookup_table_entry *lte;
 284         struct hlist_node *pos, *tmp;
 285         int ret;
 286
 287         for (size_t i = 0; i < table->capacity; i++) {
 288                 hlist_for_each_entry_safe(lte, pos, tmp, &table->array[i],
 289                                           hash_list)
 290                 {
 291                         wimlib_assert2(!(lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA));
 292                         ret = visitor(lte, arg);
 293                         if (ret)
 294                                 return ret;
 295                 }
 296         }
 297         return 0;
 298 }
 299
 300 int
 301 cmp_streams_by_wim_position(const void *p1, const void *p2)
 302 {
 303         const struct wim_lookup_table_entry *lte1, *lte2;
 304         lte1 = *(const struct wim_lookup_table_entry**)p1;
 305         lte2 = *(const struct wim_lookup_table_entry**)p2;
 306         if (lte1->resource_entry.offset < lte2->resource_entry.offset)
 307                 return -1;
 308         else if (lte1->resource_entry.offset > lte2->resource_entry.offset)
 309                 return 1;
 310         else
 311                 return 0;
 312 }
 313
 314
 315 static int
 316 add_lte_to_array(struct wim_lookup_table_entry *lte,
 317                  void *_pp)
 318 {
 319         struct wim_lookup_table_entry ***pp = _pp;
 320         *(*pp)++ = lte;
 321         return 0;
 322 }
 323
 324 /* Iterate through the lookup table entries, but first sort them by stream
 325  * offset in the WIM.  Caution: this is intended to be used when the stream
 326  * offset field has actually been set. */
 327 int
 328 for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
 329                                   int (*visitor)(struct wim_lookup_table_entry *,
 330                                                  void *),
 331                                   void *arg)
 332 {
 333         struct wim_lookup_table_entry **lte_array, **p;
 334         size_t num_streams = table->num_entries;
 335         int ret;
 336
 337         lte_array = MALLOC(num_streams * sizeof(lte_array[0]));
 338         if (!lte_array)
 339                 return WIMLIB_ERR_NOMEM;
 340         p = lte_array;
 341         for_lookup_table_entry(table, add_lte_to_array, &p);
 342
 343         wimlib_assert(p == lte_array + num_streams);
 344
 345         qsort(lte_array, num_streams, sizeof(lte_array[0]),
 346               cmp_streams_by_wim_position);
 347         ret = 0;
 348         for (size_t i = 0; i < num_streams; i++) {
 349                 ret = visitor(lte_array[i], arg);
 350                 if (ret)
 351                         break;
 352         }
 353         FREE(lte_array);
 354         return ret;
 355 }
 356
 357 /*
 358  * Reads the lookup table from a WIM file.
 359  *
 360  * Saves lookup table entries for non-metadata streams in a hash table, and
 361  * saves the metadata entry for each image in a special per-image location (the
 362  * image_metadata array).
 363  */
 364 int
 365 read_lookup_table(WIMStruct *w)
 366 {
 367         int ret;
 368         size_t num_entries;
 369         struct wim_lookup_table *table;
 370         struct wim_lookup_table_entry *cur_entry, *duplicate_entry;
 371         u8 table_buf[(BUFFER_SIZE / WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE) *
 372                         WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
 373         const u8 *p;
 374         off_t offset;
 375         size_t buf_entries_remaining;
 376
 377         DEBUG("Reading lookup table: offset %"PRIu64", size %"PRIu64"",
 378               w->hdr.lookup_table_res_entry.offset,
 379               w->hdr.lookup_table_res_entry.original_size);
 380
 381         if (resource_is_compressed(&w->hdr.lookup_table_res_entry)) {
 382                 ERROR("Didn't expect a compressed lookup table!");
 383                 ERROR("Ask the author to implement support for this.");
 384                 return WIMLIB_ERR_COMPRESSED_LOOKUP_TABLE;
 385         }
 386
 387         num_entries = w->hdr.lookup_table_res_entry.size /
 388                       WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE;
 389         table = new_lookup_table(num_entries * 2 + 1);
 390         if (!table)
 391                 return WIMLIB_ERR_NOMEM;
 392
 393         w->current_image = 0;
 394         offset = w->hdr.lookup_table_res_entry.offset;
 395         buf_entries_remaining = 0;
 396         for (; num_entries != 0; num_entries--, buf_entries_remaining--) {
 397                 if (buf_entries_remaining == 0) {
 398                         size_t entries_to_read, bytes_to_read;
 399
 400                         entries_to_read = min(sizeof(table_buf) /
 401                                                 WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE,
 402                                               num_entries);
 403                         bytes_to_read = entries_to_read *
 404                                                 WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE;
 405                         if (full_pread(w->in_fd, table_buf,
 406                                        bytes_to_read, offset) != bytes_to_read)
 407                         {
 408                                 ERROR_WITH_ERRNO("Error reading lookup table "
 409                                                  "(offset=%"PRIu64")", offset);
 410                                 ret = WIMLIB_ERR_READ;
 411                                 goto out_free_lookup_table;
 412                         }
 413                         offset += bytes_to_read;
 414                         p = table_buf;
 415                         buf_entries_remaining = entries_to_read;
 416                 }
 417                 cur_entry = new_lookup_table_entry();
 418                 if (!cur_entry) {
 419                         ret = WIMLIB_ERR_NOMEM;
 420                         goto out_free_lookup_table;
 421                 }
 422
 423                 cur_entry->wim = w;
 424                 cur_entry->resource_location = RESOURCE_IN_WIM;
 425                 p = get_resource_entry(p, &cur_entry->resource_entry);
 426                 p = get_u16(p, &cur_entry->part_number);
 427                 p = get_u32(p, &cur_entry->refcnt);
 428                 p = get_bytes(p, SHA1_HASH_SIZE, cur_entry->hash);
 429
 430                 if (cur_entry->part_number != w->hdr.part_number) {
 431                         ERROR("A lookup table entry in part %hu of the WIM "
 432                               "points to part %hu",
 433                               w->hdr.part_number, cur_entry->part_number);
 434                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 435                         goto out_free_cur_entry;
 436                 }
 437
 438                 if (is_zero_hash(cur_entry->hash)) {
 439                         ERROR("The WIM lookup table contains an entry with a "
 440                               "SHA1 message digest of all 0's");
 441                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 442                         goto out_free_cur_entry;
 443                 }
 444
 445                 if (!(cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
 446                     && (cur_entry->resource_entry.size !=
 447                         cur_entry->resource_entry.original_size))
 448                 {
 449                 #ifdef ENABLE_ERROR_MESSAGES
 450                         ERROR("Found uncompressed resource with original size "
 451                               "not the same as compressed size");
 452                         ERROR("The lookup table entry for the resource is as follows:");
 453                         print_lookup_table_entry(cur_entry, stderr);
 454                 #endif
 455                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 456                         goto out_free_cur_entry;
 457                 }
 458
 459                 if (cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_METADATA) {
 460                         /* Lookup table entry for a metadata resource */
 461                         if (cur_entry->refcnt != 1) {
 462                         #ifdef ENABLE_ERROR_MESSAGES
 463                                 ERROR("Found metadata resource with refcnt != 1:");
 464                                 print_lookup_table_entry(cur_entry, stderr);
 465                         #endif
 466                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 467                                 goto out_free_cur_entry;
 468                         }
 469
 470                         if (w->hdr.part_number != 1) {
 471                                 ERROR("Found a metadata resource in a "
 472                                       "non-first part of the split WIM!");
 473                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 474                                 goto out_free_cur_entry;
 475                         }
 476                         if (w->current_image == w->hdr.image_count) {
 477                                 ERROR("The WIM header says there are %u images "
 478                                       "in the WIM, but we found more metadata "
 479                                       "resources than this", w->hdr.image_count);
 480                                 ret = WIMLIB_ERR_IMAGE_COUNT;
 481                                 goto out_free_cur_entry;
 482                         }
 483
 484                         /* Notice very carefully:  We are assigning the metadata
 485                          * resources in the exact order mirrored by their lookup
 486                          * table entries on disk, which is the behavior of
 487                          * Microsoft's software.  In particular, this overrides
 488                          * the actual locations of the metadata resources
 489                          * themselves in the WIM file as well as any information
 490                          * written in the XML data. */
 491                         DEBUG("Found metadata resource for image %u at "
 492                               "offset %"PRIu64".",
 493                               w->current_image + 1,
 494                               cur_entry->resource_entry.offset);
 495                         w->image_metadata[
 496                                 w->current_image++]->metadata_lte = cur_entry;
 497                 } else {
 498                         /* Lookup table entry for a stream that is not a
 499                          * metadata resource */
 500                         duplicate_entry = __lookup_resource(table, cur_entry->hash);
 501                         if (duplicate_entry) {
 502                         #ifdef ENABLE_ERROR_MESSAGES
 503                                 ERROR("The WIM lookup table contains two entries with the "
 504                                       "same SHA1 message digest!");
 505                                 ERROR("The first entry is:");
 506                                 print_lookup_table_entry(duplicate_entry, stderr);
 507                                 ERROR("The second entry is:");
 508                                 print_lookup_table_entry(cur_entry, stderr);
 509                         #endif
 510                                 ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 511                                 goto out_free_cur_entry;
 512                         }
 513                         lookup_table_insert(table, cur_entry);
 514                 }
 515         }
 516
 517         if (w->hdr.part_number == 1 && w->current_image != w->hdr.image_count)
 518         {
 519                 ERROR("The WIM header says there are %u images "
 520                       "in the WIM, but we only found %d metadata "
 521                       "resources!", w->hdr.image_count, w->current_image);
 522                 ret = WIMLIB_ERR_IMAGE_COUNT;
 523                 goto out_free_lookup_table;
 524         }
 525         DEBUG("Done reading lookup table.");
 526         w->lookup_table = table;
 527         ret = 0;
 528         goto out;
 529 out_free_cur_entry:
 530         FREE(cur_entry);
 531 out_free_lookup_table:
 532         free_lookup_table(table);
 533 out:
 534         w->current_image = 0;
 535         return ret;
 536 }
 537
 538
 539 static u8 *
 540 write_lookup_table_entry(const struct wim_lookup_table_entry *lte, u8 *buf_p)
 541 {
 542         buf_p = put_resource_entry(buf_p, &lte->output_resource_entry);
 543         buf_p = put_u16(buf_p, lte->part_number);
 544         buf_p = put_u32(buf_p, lte->out_refcnt);
 545         buf_p = put_bytes(buf_p, SHA1_HASH_SIZE, lte->hash);
 546         return buf_p;
 547 }
 548
 549 int
 550 write_lookup_table_from_stream_list(struct list_head *stream_list,
 551                                     int out_fd,
 552                                     struct resource_entry *out_res_entry)
 553 {
 554         int ret;
 555         off_t start_offset;
 556         u8 table_buf[(BUFFER_SIZE / WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE) *
 557                         WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
 558         u8 *buf_p;
 559         size_t table_size;
 560         size_t bytes_to_write;
 561         struct wim_lookup_table_entry *lte;
 562
 563         start_offset = filedes_offset(out_fd);
 564         if (start_offset == -1)
 565                 goto write_error;
 566
 567         buf_p = table_buf;
 568         table_size = 0;
 569         list_for_each_entry(lte, stream_list, lookup_table_list) {
 570                 if (buf_p == table_buf + sizeof(table_buf)) {
 571                         bytes_to_write = sizeof(table_buf);
 572                         if (full_write(out_fd, table_buf,
 573                                        bytes_to_write) != bytes_to_write)
 574                                 goto write_error;
 575                         table_size += bytes_to_write;
 576                         buf_p = table_buf;
 577                 }
 578                 buf_p = write_lookup_table_entry(lte, buf_p);
 579         }
 580         bytes_to_write = buf_p - table_buf;
 581         if (bytes_to_write != 0) {
 582                 if (full_write(out_fd, table_buf,
 583                                bytes_to_write) != bytes_to_write)
 584                         goto write_error;
 585                 table_size += bytes_to_write;
 586         }
 587         out_res_entry->offset        = start_offset;
 588         out_res_entry->size          = table_size;
 589         out_res_entry->original_size = table_size;
 590         out_res_entry->flags         = WIM_RESHDR_FLAG_METADATA;
 591         ret = 0;
 592 out:
 593         return ret;
 594 write_error:
 595         ERROR_WITH_ERRNO("Failed to write lookup table");
 596         ret = WIMLIB_ERR_WRITE;
 597         goto out;
 598 }
 599
 600 static int
 601 append_lookup_table_entry(struct wim_lookup_table_entry *lte, void *_list)
 602 {
 603         if (lte->out_refcnt != 0)
 604                 list_add_tail(&lte->lookup_table_list, (struct list_head*)_list);
 605         return 0;
 606 }
 607
 608 /* Writes the WIM lookup table to the output file. */
 609 int
 610 write_lookup_table(WIMStruct *w, int image, struct resource_entry *out_res_entry)
 611 {
 612         LIST_HEAD(stream_list);
 613         int start_image;
 614         int end_image;
 615
 616         if (image == WIMLIB_ALL_IMAGES) {
 617                 start_image = 1;
 618                 end_image = w->hdr.image_count;
 619         } else {
 620                 start_image = image;
 621                 end_image = image;
 622         }
 623
 624         for (int i = start_image; i <= end_image; i++) {
 625                 struct wim_lookup_table_entry *metadata_lte;
 626
 627                 metadata_lte = w->image_metadata[i - 1]->metadata_lte;
 628                 metadata_lte->out_refcnt = 1;
 629                 metadata_lte->output_resource_entry.flags |= WIM_RESHDR_FLAG_METADATA;
 630                 append_lookup_table_entry(metadata_lte, &stream_list);
 631         }
 632         for_lookup_table_entry(w->lookup_table,
 633                                append_lookup_table_entry,
 634                                &stream_list);
 635         return write_lookup_table_from_stream_list(&stream_list,
 636                                                    w->out_fd,
 637                                                    out_res_entry);
 638 }
 639
 640 int
 641 lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 642 {
 643         lte->real_refcnt = 0;
 644         return 0;
 645 }
 646
 647 int
 648 lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 649 {
 650         lte->out_refcnt = 0;
 651         return 0;
 652 }
 653
 654 int
 655 lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *_ignore)
 656 {
 657         if (lte->extracted_file != NULL) {
 658                 FREE(lte->extracted_file);
 659                 lte->extracted_file = NULL;
 660         }
 661         return 0;
 662 }
 663
 664 void
 665 print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 666 {
 667         if (!lte) {
 668                 tputc(T('\n'), out);
 669                 return;
 670         }
 671         tfprintf(out, T("Offset            = %"PRIu64" bytes\n"),
 672                  lte->resource_entry.offset);
 673
 674         tfprintf(out, T("Size              = %"PRIu64" bytes\n"),
 675                  (u64)lte->resource_entry.size);
 676
 677         tfprintf(out, T("Original size     = %"PRIu64" bytes\n"),
 678                  lte->resource_entry.original_size);
 679
 680         tfprintf(out, T("Part Number       = %hu\n"), lte->part_number);
 681         tfprintf(out, T("Reference Count   = %u\n"), lte->refcnt);
 682
 683         if (lte->unhashed) {
 684                 tfprintf(out, T("(Unhashed: inode %p, stream_id = %u)\n"),
 685                          lte->back_inode, lte->back_stream_id);
 686         } else {
 687                 tfprintf(out, T("Hash              = 0x"));
 688                 print_hash(lte->hash, out);
 689                 tputc(T('\n'), out);
 690         }
 691
 692         tfprintf(out, T("Flags             = "));
 693         u8 flags = lte->resource_entry.flags;
 694         if (flags & WIM_RESHDR_FLAG_COMPRESSED)
 695                 tfputs(T("WIM_RESHDR_FLAG_COMPRESSED, "), out);
 696         if (flags & WIM_RESHDR_FLAG_FREE)
 697                 tfputs(T("WIM_RESHDR_FLAG_FREE, "), out);
 698         if (flags & WIM_RESHDR_FLAG_METADATA)
 699                 tfputs(T("WIM_RESHDR_FLAG_METADATA, "), out);
 700         if (flags & WIM_RESHDR_FLAG_SPANNED)
 701                 tfputs(T("WIM_RESHDR_FLAG_SPANNED, "), out);
 702         tputc(T('\n'), out);
 703         switch (lte->resource_location) {
 704         case RESOURCE_IN_WIM:
 705                 if (lte->wim->filename) {
 706                         tfprintf(out, T("WIM file          = `%"TS"'\n"),
 707                                  lte->wim->filename);
 708                 }
 709                 break;
 710 #ifdef __WIN32__
 711         case RESOURCE_WIN32:
 712         case RESOURCE_WIN32_ENCRYPTED:
 713 #else
 714         case RESOURCE_IN_FILE_ON_DISK:
 715 #endif
 716                 tfprintf(out, T("File on Disk      = `%"TS"'\n"),
 717                          lte->file_on_disk);
 718                 break;
 719 #ifdef WITH_FUSE
 720         case RESOURCE_IN_STAGING_FILE:
 721                 tfprintf(out, T("Staging File      = `%"TS"'\n"),
 722                                 lte->staging_file_name);
 723                 break;
 724 #endif
 725         default:
 726                 break;
 727         }
 728         tputc(T('\n'), out);
 729 }
 730
 731 static int
 732 do_print_lookup_table_entry(struct wim_lookup_table_entry *lte, void *fp)
 733 {
 734         print_lookup_table_entry(lte, (FILE*)fp);
 735         return 0;
 736 }
 737
 738 /*
 739  * Prints the lookup table of a WIM file.
 740  */
 741 WIMLIBAPI void
 742 wimlib_print_lookup_table(WIMStruct *w)
 743 {
 744         for_lookup_table_entry(w->lookup_table,
 745                                do_print_lookup_table_entry,
 746                                stdout);
 747 }
 748
 749 /* Given a SHA1 message digest, return the corresponding entry in the WIM's
 750  * lookup table, or NULL if there is none.  */
 751 struct wim_lookup_table_entry *
 752 __lookup_resource(const struct wim_lookup_table *table, const u8 hash[])
 753 {
 754         size_t i;
 755         struct wim_lookup_table_entry *lte;
 756         struct hlist_node *pos;
 757
 758         wimlib_assert(table != NULL);
 759         wimlib_assert(hash != NULL);
 760
 761         i = *(size_t*)hash % table->capacity;
 762         hlist_for_each_entry(lte, pos, &table->array[i], hash_list)
 763                 if (hashes_equal(hash, lte->hash))
 764                         return lte;
 765         return NULL;
 766 }
 767
 768 #ifdef WITH_FUSE
 769 /*
 770  * Finds the dentry, lookup table entry, and stream index for a WIM file stream,
 771  * given a path name.
 772  *
 773  * This is only for pre-resolved inodes.
 774  */
 775 int
 776 lookup_resource(WIMStruct *w,
 777                 const tchar *path,
 778                 int lookup_flags,
 779                 struct wim_dentry **dentry_ret,
 780                 struct wim_lookup_table_entry **lte_ret,
 781                 u16 *stream_idx_ret)
 782 {
 783         struct wim_dentry *dentry;
 784         struct wim_lookup_table_entry *lte;
 785         u16 stream_idx;
 786         const tchar *stream_name = NULL;
 787         struct wim_inode *inode;
 788         tchar *p = NULL;
 789
 790         if (lookup_flags & LOOKUP_FLAG_ADS_OK) {
 791                 stream_name = path_stream_name(path);
 792                 if (stream_name) {
 793                         p = (tchar*)stream_name - 1;
 794                         *p = T('\0');
 795                 }
 796         }
 797
 798         dentry = get_dentry(w, path);
 799         if (p)
 800                 *p = T(':');
 801         if (!dentry)
 802                 return -errno;
 803
 804         inode = dentry->d_inode;
 805
 806         wimlib_assert(inode->i_resolved);
 807
 808         if (!(lookup_flags & LOOKUP_FLAG_DIRECTORY_OK)
 809               && inode_is_directory(inode))
 810                 return -EISDIR;
 811
 812         if (stream_name) {
 813                 struct wim_ads_entry *ads_entry;
 814                 u16 ads_idx;
 815                 ads_entry = inode_get_ads_entry(inode, stream_name,
 816                                                 &ads_idx);
 817                 if (ads_entry) {
 818                         stream_idx = ads_idx + 1;
 819                         lte = ads_entry->lte;
 820                         goto out;
 821                 } else {
 822                         return -ENOENT;
 823                 }
 824         } else {
 825                 lte = inode->i_lte;
 826                 stream_idx = 0;
 827         }
 828 out:
 829         if (dentry_ret)
 830                 *dentry_ret = dentry;
 831         if (lte_ret)
 832                 *lte_ret = lte;
 833         if (stream_idx_ret)
 834                 *stream_idx_ret = stream_idx;
 835         return 0;
 836 }
 837 #endif
 838
 839 /*
 840  * XXX Probably should store the compression type directly in the lookup table
 841  * entry
 842  */
 843 int
 844 wim_resource_compression_type(const struct wim_lookup_table_entry *lte)
 845 {
 846         if (!(lte->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
 847             || lte->resource_location != RESOURCE_IN_WIM)
 848                 return WIMLIB_COMPRESSION_TYPE_NONE;
 849         return wimlib_get_compression_type(lte->wim);
 850 }
 851
 852 /* Resolve an inode's lookup table entries
 853  *
 854  * This replaces the SHA1 hash fields (which are used to lookup an entry in the
 855  * lookup table) with pointers directly to the lookup table entries.  A circular
 856  * linked list of streams sharing the same lookup table entry is created.
 857  *
 858  * This function always succeeds; unresolved lookup table entries are given a
 859  * NULL pointer.
 860  */
 861 void
 862 inode_resolve_ltes(struct wim_inode *inode, struct wim_lookup_table *table)
 863 {
 864
 865         if (!inode->i_resolved) {
 866                 struct wim_lookup_table_entry *lte;
 867                 /* Resolve the default file stream */
 868                 lte = __lookup_resource(table, inode->i_hash);
 869                 inode->i_lte = lte;
 870                 inode->i_resolved = 1;
 871
 872                 /* Resolve the alternate data streams */
 873                 for (u16 i = 0; i < inode->i_num_ads; i++) {
 874                         struct wim_ads_entry *cur_entry = &inode->i_ads_entries[i];
 875                         lte = __lookup_resource(table, cur_entry->hash);
 876                         cur_entry->lte = lte;
 877                 }
 878         }
 879 }
 880
 881 void
 882 inode_unresolve_ltes(struct wim_inode *inode)
 883 {
 884         if (inode->i_resolved) {
 885                 if (inode->i_lte)
 886                         copy_hash(inode->i_hash, inode->i_lte->hash);
 887                 else
 888                         zero_out_hash(inode->i_hash);
 889
 890                 for (u16 i = 0; i < inode->i_num_ads; i++) {
 891                         if (inode->i_ads_entries[i].lte)
 892                                 copy_hash(inode->i_ads_entries[i].hash,
 893                                           inode->i_ads_entries[i].lte->hash);
 894                         else
 895                                 zero_out_hash(inode->i_ads_entries[i].hash);
 896                 }
 897                 inode->i_resolved = 0;
 898         }
 899 }
 900
 901 /*
 902  * Returns the lookup table entry for stream @stream_idx of the inode, where
 903  * stream_idx = 0 means the default un-named file stream, and stream_idx >= 1
 904  * corresponds to an alternate data stream.
 905  *
 906  * This works for both resolved and un-resolved inodes.
 907  */
 908 struct wim_lookup_table_entry *
 909 inode_stream_lte(const struct wim_inode *inode, unsigned stream_idx,
 910                  const struct wim_lookup_table *table)
 911 {
 912         if (inode->i_resolved)
 913                 return inode_stream_lte_resolved(inode, stream_idx);
 914         else
 915                 return inode_stream_lte_unresolved(inode, stream_idx, table);
 916 }
 917
 918 struct wim_lookup_table_entry *
 919 inode_unnamed_lte_resolved(const struct wim_inode *inode)
 920 {
 921         wimlib_assert(inode->i_resolved);
 922         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
 923                 if (inode_stream_name_nbytes(inode, i) == 0 &&
 924                     !is_zero_hash(inode_stream_hash_resolved(inode, i)))
 925                 {
 926                         return inode_stream_lte_resolved(inode, i);
 927                 }
 928         }
 929         return NULL;
 930 }
 931
 932 struct wim_lookup_table_entry *
 933 inode_unnamed_lte_unresolved(const struct wim_inode *inode,
 934                              const struct wim_lookup_table *table)
 935 {
 936         wimlib_assert(!inode->i_resolved);
 937         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
 938                 if (inode_stream_name_nbytes(inode, i) == 0 &&
 939                     !is_zero_hash(inode_stream_hash_unresolved(inode, i)))
 940                 {
 941                         return inode_stream_lte_unresolved(inode, i, table);
 942                 }
 943         }
 944         return NULL;
 945 }
 946
 947 /* Return the lookup table entry for the unnamed data stream of an inode, or
 948  * NULL if there is none.
 949  *
 950  * You'd think this would be easier than it actually is, since the unnamed data
 951  * stream should be the one referenced from the inode itself.  Alas, if there
 952  * are named data streams, Microsoft's "imagex.exe" program will put the unnamed
 953  * data stream in one of the alternate data streams instead of inside the WIM
 954  * dentry itself.  So we need to check the alternate data streams too.
 955  *
 956  * Also, note that a dentry may appear to have more than one unnamed stream, but
 957  * if the SHA1 message digest is all 0's then the corresponding stream does not
 958  * really "count" (this is the case for the inode's own file stream when the
 959  * file stream that should be there is actually in one of the alternate stream
 960  * entries.).  This is despite the fact that we may need to extract such a
 961  * missing entry as an empty file or empty named data stream.
 962  */
 963 struct wim_lookup_table_entry *
 964 inode_unnamed_lte(const struct wim_inode *inode,
 965                   const struct wim_lookup_table *table)
 966 {
 967         if (inode->i_resolved)
 968                 return inode_unnamed_lte_resolved(inode);
 969         else
 970                 return inode_unnamed_lte_unresolved(inode, table);
 971 }
 972
 973 static int
 974 lte_add_stream_size(struct wim_lookup_table_entry *lte, void *total_bytes_p)
 975 {
 976         *(u64*)total_bytes_p += lte->resource_entry.size;
 977         return 0;
 978 }
 979
 980 u64
 981 lookup_table_total_stream_size(struct wim_lookup_table *table)
 982 {
 983         u64 total_size = 0;
 984         for_lookup_table_entry(table, lte_add_stream_size, &total_size);
 985         return total_size;
 986 }
 987
 988 struct wim_lookup_table_entry **
 989 retrieve_lte_pointer(struct wim_lookup_table_entry *lte)
 990 {
 991         wimlib_assert(lte->unhashed);
 992         struct wim_inode *inode = lte->back_inode;
 993         u32 stream_id = lte->back_stream_id;
 994         if (stream_id == 0)
 995                 return &inode->i_lte;
 996         else
 997                 for (u16 i = 0; i < inode->i_num_ads; i++)
 998                         if (inode->i_ads_entries[i].stream_id == stream_id)
 999                                 return &inode->i_ads_entries[i].lte;
1000         wimlib_assert(0);
1001         return NULL;
1002 }
1003
1004 /* Calculate the SHA1 message digest of a stream and move it from the list of
1005  * unhashed streams to the stream lookup table, possibly joining it with an
1006  * existing lookup table entry for an identical stream.
1007  *
1008  * @lte:  An unhashed lookup table entry.
1009  * @lookup_table:  Lookup table for the WIM.
1010  * @lte_ret:  On success, write a pointer to the resulting lookup table
1011  *            entry to this location.  This will be the same as @lte
1012  *            if it was inserted into the lookup table, or different if
1013  *            a duplicate stream was found.
1014  *
1015  * Returns 0 on success; nonzero if there is an error reading the stream.
1016  */
1017 int
1018 hash_unhashed_stream(struct wim_lookup_table_entry *lte,
1019                      struct wim_lookup_table *lookup_table,
1020                      struct wim_lookup_table_entry **lte_ret)
1021 {
1022         int ret;
1023         struct wim_lookup_table_entry *duplicate_lte;
1024         struct wim_lookup_table_entry **back_ptr;
1025
1026         wimlib_assert(lte->unhashed);
1027
1028         /* back_ptr must be saved because @back_inode and @back_stream_id are in
1029          * union with the SHA1 message digest and will no longer be valid once
1030          * the SHA1 has been calculated. */
1031         back_ptr = retrieve_lte_pointer(lte);
1032
1033         ret = sha1_resource(lte);
1034         if (ret)
1035                 return ret;
1036
1037         /* Look for a duplicate stream */
1038         duplicate_lte = __lookup_resource(lookup_table, lte->hash);
1039         list_del(&lte->unhashed_list);
1040         if (duplicate_lte) {
1041                 /* We have a duplicate stream.  Transfer the reference counts
1042                  * from this stream to the duplicate, update the reference to
1043                  * this stream (in an inode or ads_entry) to point to the
1044                  * duplicate, then free this stream. */
1045                 wimlib_assert(!(duplicate_lte->unhashed));
1046                 duplicate_lte->refcnt += lte->refcnt;
1047                 duplicate_lte->out_refcnt += lte->refcnt;
1048                 *back_ptr = duplicate_lte;
1049                 free_lookup_table_entry(lte);
1050                 lte = duplicate_lte;
1051         } else {
1052                 /* No duplicate stream, so we need to insert
1053                  * this stream into the lookup table and treat
1054                  * it as a hashed stream. */
1055                 lookup_table_insert(lookup_table, lte);
1056                 lte->unhashed = 0;
1057         }
1058         if (lte_ret)
1059                 *lte_ret = lte;
1060         return 0;
1061 }
1062