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