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