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 dentries to file
   5  * resources.
   6  */
   7
   8 /*
   9  * Copyright (C) 2012 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 "io.h"
  30 #include <errno.h>
  31
  32 struct lookup_table *new_lookup_table(size_t capacity)
  33 {
  34         struct lookup_table *table;
  35         struct hlist_head *array;
  36
  37         table = MALLOC(sizeof(struct lookup_table));
  38         if (!table)
  39                 goto err;
  40         array = CALLOC(capacity, sizeof(array[0]));
  41         if (!array) {
  42                 FREE(table);
  43                 goto err;
  44         }
  45         table->num_entries = 0;
  46         table->capacity = capacity;
  47         table->array = array;
  48         return table;
  49 err:
  50         ERROR("Failed to allocate memory for lookup table with capacity %zu",
  51               capacity);
  52         return NULL;
  53 }
  54
  55 struct lookup_table_entry *new_lookup_table_entry()
  56 {
  57         struct lookup_table_entry *lte;
  58
  59         lte = CALLOC(1, sizeof(struct lookup_table_entry));
  60         if (!lte) {
  61                 ERROR("Out of memory (tried to allocate %zu bytes for "
  62                       "lookup table entry)",
  63                       sizeof(struct lookup_table_entry));
  64                 return NULL;
  65         }
  66
  67         lte->part_number  = 1;
  68         lte->refcnt       = 1;
  69         INIT_LIST_HEAD(&lte->lte_group_list);
  70         return lte;
  71 }
  72
  73 void free_lookup_table_entry(struct lookup_table_entry *lte)
  74 {
  75         if (lte) {
  76 #ifdef WITH_FUSE
  77                 if (lte->staging_list.next)
  78                         list_del(&lte->staging_list);
  79 #endif
  80                 switch (lte->resource_location) {
  81                 case RESOURCE_IN_STAGING_FILE:
  82                 case RESOURCE_IN_ATTACHED_BUFFER:
  83                 case RESOURCE_IN_FILE_ON_DISK:
  84                         wimlib_assert(((void*)&lte->file_on_disk ==
  85                                       (void*)&lte->staging_file_name)
  86                                       && ((void*)&lte->file_on_disk ==
  87                                       (void*)&lte->attached_buffer));
  88                         FREE(lte->file_on_disk);
  89                         break;
  90 #ifdef WITH_NTFS_3G
  91                 case RESOURCE_IN_NTFS_VOLUME:
  92                         if (lte->ntfs_loc) {
  93                                 FREE(lte->ntfs_loc->path_utf8);
  94                                 FREE(lte->ntfs_loc->stream_name_utf16);
  95                                 FREE(lte->ntfs_loc);
  96                         }
  97                         break;
  98 #endif
  99                 default:
 100                         break;
 101                 }
 102                 FREE(lte);
 103         }
 104 }
 105
 106 static int do_free_lookup_table_entry(struct lookup_table_entry *entry,
 107                                       void *ignore)
 108 {
 109         free_lookup_table_entry(entry);
 110         return 0;
 111 }
 112
 113
 114 void free_lookup_table(struct lookup_table *table)
 115 {
 116         DEBUG("Freeing lookup table");
 117         if (table) {
 118                 if (table->array) {
 119                         for_lookup_table_entry(table,
 120                                                do_free_lookup_table_entry,
 121                                                NULL);
 122                         FREE(table->array);
 123                 }
 124                 FREE(table);
 125         }
 126 }
 127
 128 /*
 129  * Inserts an entry into the lookup table.
 130  *
 131  * @table:      A pointer to the lookup table.
 132  * @entry:      A pointer to the entry to insert.
 133  */
 134 void lookup_table_insert(struct lookup_table *table,
 135                          struct lookup_table_entry *lte)
 136 {
 137         size_t i = lte->hash_short % table->capacity;
 138         hlist_add_head(&lte->hash_list, &table->array[i]);
 139
 140         /* XXX Make the table grow when too many entries have been inserted. */
 141         table->num_entries++;
 142 }
 143
 144
 145
 146 /* Decrements the reference count for the lookup table entry @lte.  If its
 147  * reference count reaches 0, it is unlinked from the lookup table.  If,
 148  * furthermore, the entry has no opened file descriptors associated with it, the
 149  * entry is freed.  */
 150 struct lookup_table_entry *
 151 lte_decrement_refcnt(struct lookup_table_entry *lte, struct lookup_table *table)
 152 {
 153         if (lte) {
 154                 wimlib_assert(lte->refcnt);
 155                 if (--lte->refcnt == 0) {
 156                         lookup_table_unlink(table, lte);
 157                 #ifdef WITH_FUSE
 158                         if (lte->num_opened_fds == 0)
 159                 #endif
 160                         {
 161                                 free_lookup_table_entry(lte);
 162                                 lte = NULL;
 163                         }
 164                 }
 165         }
 166         return lte;
 167 }
 168
 169 /*
 170  * Calls a function on all the entries in the lookup table.  Stop early and
 171  * return nonzero if any call to the function returns nonzero.
 172  */
 173 int for_lookup_table_entry(struct lookup_table *table,
 174                            int (*visitor)(struct lookup_table_entry *, void *),
 175                            void *arg)
 176 {
 177         struct lookup_table_entry *lte;
 178         struct hlist_node *pos, *tmp;
 179         int ret;
 180
 181         for (size_t i = 0; i < table->capacity; i++) {
 182                 hlist_for_each_entry_safe(lte, pos, tmp, &table->array[i],
 183                                           hash_list)
 184                 {
 185                         ret = visitor(lte, arg);
 186                         if (ret != 0)
 187                                 return ret;
 188                 }
 189         }
 190         return 0;
 191 }
 192
 193
 194 /*
 195  * Reads the lookup table from a WIM file.
 196  */
 197 int read_lookup_table(WIMStruct *w)
 198 {
 199         u64    num_entries;
 200         u8     buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
 201         int    ret;
 202         struct lookup_table *table;
 203
 204         DEBUG("Reading lookup table: offset %"PRIu64", size %"PRIu64"",
 205               w->hdr.lookup_table_res_entry.offset,
 206               w->hdr.lookup_table_res_entry.original_size);
 207
 208         if (fseeko(w->fp, w->hdr.lookup_table_res_entry.offset, SEEK_SET) != 0) {
 209                 ERROR_WITH_ERRNO("Failed to seek to byte %"PRIu64" to read "
 210                                  "lookup table",
 211                                  w->hdr.lookup_table_res_entry.offset);
 212                 return WIMLIB_ERR_READ;
 213         }
 214
 215         num_entries = w->hdr.lookup_table_res_entry.original_size /
 216                       WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE;
 217         table = new_lookup_table(num_entries * 2 + 1);
 218         if (!table)
 219                 return WIMLIB_ERR_NOMEM;
 220
 221         while (num_entries--) {
 222                 const u8 *p;
 223                 struct lookup_table_entry *cur_entry, *duplicate_entry;
 224
 225                 if (fread(buf, 1, sizeof(buf), w->fp) != sizeof(buf)) {
 226                         if (feof(w->fp)) {
 227                                 ERROR("Unexpected EOF in WIM lookup table!");
 228                         } else {
 229                                 ERROR_WITH_ERRNO("Error reading WIM lookup "
 230                                                  "table");
 231                         }
 232                         ret = WIMLIB_ERR_READ;
 233                         goto out;
 234                 }
 235                 cur_entry = new_lookup_table_entry();
 236                 if (!cur_entry) {
 237                         ret = WIMLIB_ERR_NOMEM;
 238                         goto out;
 239                 }
 240                 cur_entry->wim = w;
 241                 cur_entry->resource_location = RESOURCE_IN_WIM;
 242
 243                 p = get_resource_entry(buf, &cur_entry->resource_entry);
 244                 p = get_u16(p, &cur_entry->part_number);
 245                 p = get_u32(p, &cur_entry->refcnt);
 246                 p = get_bytes(p, SHA1_HASH_SIZE, cur_entry->hash);
 247
 248                 if (is_zero_hash(cur_entry->hash)) {
 249                         ERROR("The WIM lookup table contains an entry with a "
 250                               "SHA1 message digest of all 0's");
 251                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 252                         FREE(cur_entry);
 253                         goto out;
 254                 }
 255
 256                 duplicate_entry = __lookup_resource(table, cur_entry->hash);
 257                 if (duplicate_entry) {
 258                         ERROR("The WIM lookup table contains two entries with the "
 259                               "same SHA1 message digest!");
 260                         ERROR("The first entry is:");
 261                         print_lookup_table_entry(duplicate_entry);
 262                         ERROR("The second entry is:");
 263                         print_lookup_table_entry(cur_entry);
 264                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 265                         FREE(cur_entry);
 266                         goto out;
 267                 }
 268                 lookup_table_insert(table, cur_entry);
 269
 270                 if (!(cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
 271                     && (cur_entry->resource_entry.size !=
 272                       cur_entry->resource_entry.original_size))
 273                 {
 274                         ERROR("Found uncompressed resource with original size "
 275                               "not the same as compressed size");
 276                         ERROR("The lookup table entry for the resource is as follows:");
 277                         print_lookup_table_entry(cur_entry);
 278                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 279                         goto out;
 280                 }
 281         }
 282         DEBUG("Done reading lookup table.");
 283         w->lookup_table = table;
 284         return 0;
 285 out:
 286         free_lookup_table(table);
 287         return ret;
 288 }
 289
 290
 291 /*
 292  * Writes a lookup table entry to the output file.
 293  */
 294 int write_lookup_table_entry(struct lookup_table_entry *lte, void *__out)
 295 {
 296         FILE *out;
 297         u8 buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
 298         u8 *p;
 299
 300         out = __out;
 301
 302         /* Don't write entries that have not had file resources or metadata
 303          * resources written for them. */
 304         if (lte->out_refcnt == 0)
 305                 return 0;
 306
 307         if (lte->output_resource_entry.flags & WIM_RESHDR_FLAG_METADATA)
 308                 DEBUG("Writing metadata entry at %lu (orig size = %zu)",
 309                       ftello(out), lte->output_resource_entry.original_size);
 310
 311         p = put_resource_entry(buf, &lte->output_resource_entry);
 312         p = put_u16(p, lte->part_number);
 313         p = put_u32(p, lte->out_refcnt);
 314         p = put_bytes(p, SHA1_HASH_SIZE, lte->hash);
 315         if (fwrite(buf, 1, sizeof(buf), out) != sizeof(buf)) {
 316                 ERROR_WITH_ERRNO("Failed to write lookup table entry");
 317                 return WIMLIB_ERR_WRITE;
 318         }
 319         return 0;
 320 }
 321
 322
 323
 324 int zero_out_refcnts(struct lookup_table_entry *entry, void *ignore)
 325 {
 326         entry->out_refcnt = 0;
 327         return 0;
 328 }
 329
 330 void print_lookup_table_entry(const struct lookup_table_entry *lte)
 331 {
 332         if (!lte) {
 333                 putchar('\n');
 334                 return;
 335         }
 336         printf("Offset            = %"PRIu64" bytes\n",
 337                lte->resource_entry.offset);
 338         printf("Size              = %"PRIu64" bytes\n",
 339                (u64)lte->resource_entry.size);
 340         printf("Original size     = %"PRIu64" bytes\n",
 341                lte->resource_entry.original_size);
 342         printf("Part Number       = %hu\n", lte->part_number);
 343         printf("Reference Count   = %u\n", lte->refcnt);
 344         printf("Hash              = 0x");
 345         print_hash(lte->hash);
 346         putchar('\n');
 347         printf("Flags             = ");
 348         u8 flags = lte->resource_entry.flags;
 349         if (flags & WIM_RESHDR_FLAG_COMPRESSED)
 350                 fputs("WIM_RESHDR_FLAG_COMPRESSED, ", stdout);
 351         if (flags & WIM_RESHDR_FLAG_FREE)
 352                 fputs("WIM_RESHDR_FLAG_FREE, ", stdout);
 353         if (flags & WIM_RESHDR_FLAG_METADATA)
 354                 fputs("WIM_RESHDR_FLAG_METADATA, ", stdout);
 355         if (flags & WIM_RESHDR_FLAG_SPANNED)
 356                 fputs("WIM_RESHDR_FLAG_SPANNED, ", stdout);
 357         putchar('\n');
 358         switch (lte->resource_location) {
 359         case RESOURCE_IN_WIM:
 360                 if (lte->wim->filename) {
 361                         printf("WIM file          = `%s'\n",
 362                                lte->wim->filename);
 363                 }
 364                 break;
 365         case RESOURCE_IN_FILE_ON_DISK:
 366                 printf("File on Disk      = `%s'\n", lte->file_on_disk);
 367                 break;
 368         case RESOURCE_IN_STAGING_FILE:
 369                 printf("Staging File      = `%s'\n", lte->staging_file_name);
 370                 break;
 371         default:
 372                 break;
 373         }
 374         putchar('\n');
 375 }
 376
 377 static int do_print_lookup_table_entry(struct lookup_table_entry *lte,
 378                                        void *ignore)
 379 {
 380         print_lookup_table_entry(lte);
 381         return 0;
 382 }
 383
 384 /*
 385  * Prints the lookup table of a WIM file.
 386  */
 387 WIMLIBAPI void wimlib_print_lookup_table(WIMStruct *w)
 388 {
 389         for_lookup_table_entry(w->lookup_table,
 390                                do_print_lookup_table_entry,
 391                                NULL);
 392 }
 393
 394 /*
 395  * Looks up an entry in the lookup table.
 396  */
 397 struct lookup_table_entry *
 398 __lookup_resource(const struct lookup_table *table, const u8 hash[])
 399 {
 400         size_t i;
 401         struct lookup_table_entry *lte;
 402         struct hlist_node *pos;
 403
 404         i = *(size_t*)hash % table->capacity;
 405         hlist_for_each_entry(lte, pos, &table->array[i], hash_list)
 406                 if (hashes_equal(hash, lte->hash))
 407                         return lte;
 408         return NULL;
 409 }
 410
 411 /*
 412  * Finds the dentry, lookup table entry, and stream index for a WIM file stream,
 413  * given a path name.
 414  *
 415  * This is only for pre-resolved dentries.
 416  */
 417 int lookup_resource(WIMStruct *w, const char *path,
 418                     int lookup_flags,
 419                     struct dentry **dentry_ret,
 420                     struct lookup_table_entry **lte_ret,
 421                     unsigned *stream_idx_ret)
 422 {
 423         struct dentry *dentry;
 424         struct lookup_table_entry *lte;
 425         unsigned stream_idx;
 426         const char *stream_name = NULL;
 427         char *p = NULL;
 428
 429         if (lookup_flags & LOOKUP_FLAG_ADS_OK) {
 430                 stream_name = path_stream_name(path);
 431                 if (stream_name) {
 432                         p = (char*)stream_name - 1;
 433                         *p = '\0';
 434                 }
 435         }
 436
 437         dentry = get_dentry(w, path);
 438         if (p)
 439                 *p = ':';
 440         if (!dentry)
 441                 return -ENOENT;
 442
 443         wimlib_assert(dentry->resolved);
 444
 445         lte = dentry->lte;
 446         if (!(lookup_flags & LOOKUP_FLAG_DIRECTORY_OK)
 447               && dentry_is_directory(dentry))
 448                 return -EISDIR;
 449         stream_idx = 0;
 450         if (stream_name) {
 451                 size_t stream_name_len = strlen(stream_name);
 452                 for (u16 i = 0; i < dentry->num_ads; i++) {
 453                         if (ads_entry_has_name(&dentry->ads_entries[i],
 454                                                stream_name,
 455                                                stream_name_len))
 456                         {
 457                                 stream_idx = i + 1;
 458                                 lte = dentry->ads_entries[i].lte;
 459                                 goto out;
 460                         }
 461                 }
 462                 return -ENOENT;
 463         }
 464 out:
 465         if (dentry_ret)
 466                 *dentry_ret = dentry;
 467         if (lte_ret)
 468                 *lte_ret = lte;
 469         if (stream_idx_ret)
 470                 *stream_idx_ret = stream_idx;
 471         return 0;
 472 }
 473
 474 /* Resolve a dentry's lookup table entries
 475  *
 476  * This replaces the SHA1 hash fields (which are used to lookup an entry in the
 477  * lookup table) with pointers directly to the lookup table entries.  A circular
 478  * linked list of streams sharing the same lookup table entry is created.
 479  *
 480  * This function always succeeds; unresolved lookup table entries are given a
 481  * NULL pointer.
 482  */
 483 int dentry_resolve_ltes(struct dentry *dentry, void *__table)
 484 {
 485         struct lookup_table *table = __table;
 486         struct lookup_table_entry *lte;
 487
 488         if (dentry->resolved)
 489                 return 0;
 490
 491         /* Resolve the default file stream */
 492         lte = __lookup_resource(table, dentry->hash);
 493         if (lte)
 494                 list_add(&dentry->lte_group_list.list, &lte->lte_group_list);
 495         else
 496                 INIT_LIST_HEAD(&dentry->lte_group_list.list);
 497         dentry->lte = lte;
 498         dentry->lte_group_list.type = STREAM_TYPE_NORMAL;
 499         dentry->resolved = true;
 500
 501         /* Resolve the alternate data streams */
 502         if (dentry->ads_entries_status != ADS_ENTRIES_USER) {
 503                 for (u16 i = 0; i < dentry->num_ads; i++) {
 504                         struct ads_entry *cur_entry = &dentry->ads_entries[i];
 505
 506                         lte = __lookup_resource(table, cur_entry->hash);
 507                         if (lte)
 508                                 list_add(&cur_entry->lte_group_list.list,
 509                                          &lte->lte_group_list);
 510                         else
 511                                 INIT_LIST_HEAD(&cur_entry->lte_group_list.list);
 512                         cur_entry->lte = lte;
 513                         cur_entry->lte_group_list.type = STREAM_TYPE_ADS;
 514                 }
 515         }
 516         return 0;
 517 }
 518
 519 /* Return the lookup table entry for the unnamed data stream of a dentry, or
 520  * NULL if there is none.
 521  *
 522  * You'd think this would be easier than it actually is, since the unnamed data
 523  * stream should be the one referenced from the dentry itself.  Alas, if there
 524  * are named data streams, Microsoft's "imagex.exe" program will put the unnamed
 525  * data stream in one of the alternate data streams instead of inside the
 526  * dentry.  So we need to check the alternate data streams too.
 527  *
 528  * Also, note that a dentry may appear to have than one unnamed stream, but if
 529  * the SHA1 message digest is all 0's then the corresponding stream does not
 530  * really "count" (this is the case for the dentry's own file stream when the
 531  * file stream that should be there is actually in one of the alternate stream
 532  * entries.).  This is despite the fact that we may need to extract such a
 533  * missing entry as an empty file or empty named data stream.
 534  */
 535 struct lookup_table_entry *
 536 dentry_unnamed_lte(const struct dentry *dentry,
 537                    const struct lookup_table *table)
 538 {
 539         if (dentry->resolved)
 540                 return dentry_unnamed_lte_resolved(dentry);
 541         else
 542                 return dentry_unnamed_lte_unresolved(dentry, table);
 543 }
 544