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