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 Lesser General Public License as published by the Free
  15  * Software Foundation; either version 2.1 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 Lesser General Public License for more
  21  * details.
  22  *
  23  * You should have received a copy of the GNU Lesser 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                 if (lte->staging_list.next)
  77                         list_del(&lte->staging_list);
  78                 if (lte->resource_location != RESOURCE_IN_WIM &&
  79                     lte->resource_location != RESOURCE_NONEXISTENT)
  80                         FREE(lte->file_on_disk);
  81                 FREE(lte);
  82         }
  83 }
  84
  85 static int do_free_lookup_table_entry(struct lookup_table_entry *entry,
  86                                       void *ignore)
  87 {
  88         free_lookup_table_entry(entry);
  89         return 0;
  90 }
  91
  92
  93 void free_lookup_table(struct lookup_table *table)
  94 {
  95         DEBUG("Freeing lookup table");
  96         if (table) {
  97                 if (table->array) {
  98                         for_lookup_table_entry(table,
  99                                                do_free_lookup_table_entry,
 100                                                NULL);
 101                         FREE(table->array);
 102                 }
 103                 FREE(table);
 104         }
 105 }
 106
 107 /*
 108  * Inserts an entry into the lookup table.
 109  *
 110  * @table:      A pointer to the lookup table.
 111  * @entry:      A pointer to the entry to insert.
 112  */
 113 void lookup_table_insert(struct lookup_table *table,
 114                          struct lookup_table_entry *lte)
 115 {
 116         size_t i = lte->hash_short % table->capacity;
 117         hlist_add_head(&lte->hash_list, &table->array[i]);
 118
 119         /* XXX Make the table grow when too many entries have been inserted. */
 120         table->num_entries++;
 121 }
 122
 123
 124
 125 /* Decrements the reference count for the lookup table entry @lte.  If its
 126  * reference count reaches 0, it is unlinked from the lookup table.  If,
 127  * furthermore, the entry has no opened file descriptors associated with it, the
 128  * entry is freed.  */
 129 struct lookup_table_entry *
 130 lte_decrement_refcnt(struct lookup_table_entry *lte, struct lookup_table *table)
 131 {
 132         if (lte) {
 133                 wimlib_assert(lte->refcnt);
 134                 if (--lte->refcnt == 0) {
 135                         lookup_table_unlink(table, lte);
 136                         if (lte->num_opened_fds == 0) {
 137                                 free_lookup_table_entry(lte);
 138                                 lte = NULL;
 139                         }
 140                 }
 141         }
 142         return lte;
 143 }
 144
 145 /*
 146  * Calls a function on all the entries in the lookup table.  Stop early and
 147  * return nonzero if any call to the function returns nonzero.
 148  */
 149 int for_lookup_table_entry(struct lookup_table *table,
 150                            int (*visitor)(struct lookup_table_entry *, void *),
 151                            void *arg)
 152 {
 153         struct lookup_table_entry *lte;
 154         struct hlist_node *pos, *tmp;
 155         int ret;
 156
 157         for (size_t i = 0; i < table->capacity; i++) {
 158                 hlist_for_each_entry_safe(lte, pos, tmp, &table->array[i],
 159                                           hash_list)
 160                 {
 161                         ret = visitor(lte, arg);
 162                         if (ret != 0)
 163                                 return ret;
 164                 }
 165         }
 166         return 0;
 167 }
 168
 169
 170 /*
 171  * Reads the lookup table from a WIM file.
 172  */
 173 int read_lookup_table(WIMStruct *w)
 174 {
 175         u64    num_entries;
 176         u8     buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
 177         int    ret;
 178         struct lookup_table *table;
 179
 180         DEBUG("Reading lookup table: offset %"PRIu64", size %"PRIu64"",
 181               w->hdr.lookup_table_res_entry.offset,
 182               w->hdr.lookup_table_res_entry.original_size);
 183
 184         if (fseeko(w->fp, w->hdr.lookup_table_res_entry.offset, SEEK_SET) != 0) {
 185                 ERROR_WITH_ERRNO("Failed to seek to byte %"PRIu64" to read "
 186                                  "lookup table",
 187                                  w->hdr.lookup_table_res_entry.offset);
 188                 return WIMLIB_ERR_READ;
 189         }
 190
 191         num_entries = w->hdr.lookup_table_res_entry.original_size /
 192                       WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE;
 193         table = new_lookup_table(num_entries * 2 + 1);
 194         if (!table)
 195                 return WIMLIB_ERR_NOMEM;
 196
 197         while (num_entries--) {
 198                 const u8 *p;
 199                 struct lookup_table_entry *cur_entry, *duplicate_entry;
 200
 201                 if (fread(buf, 1, sizeof(buf), w->fp) != sizeof(buf)) {
 202                         if (feof(w->fp)) {
 203                                 ERROR("Unexpected EOF in WIM lookup table!");
 204                         } else {
 205                                 ERROR_WITH_ERRNO("Error reading WIM lookup "
 206                                                  "table");
 207                         }
 208                         ret = WIMLIB_ERR_READ;
 209                         goto out;
 210                 }
 211                 cur_entry = new_lookup_table_entry();
 212                 if (!cur_entry) {
 213                         ret = WIMLIB_ERR_NOMEM;
 214                         goto out;
 215                 }
 216                 cur_entry->wim = w;
 217                 cur_entry->resource_location = RESOURCE_IN_WIM;
 218
 219                 p = get_resource_entry(buf, &cur_entry->resource_entry);
 220                 p = get_u16(p, &cur_entry->part_number);
 221                 p = get_u32(p, &cur_entry->refcnt);
 222                 p = get_bytes(p, SHA1_HASH_SIZE, cur_entry->hash);
 223
 224                 if (is_zero_hash(cur_entry->hash)) {
 225                         ERROR("The WIM lookup table contains an entry with a "
 226                               "SHA1 message digest of all 0's");
 227                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 228                         FREE(cur_entry);
 229                         goto out;
 230                 }
 231
 232                 duplicate_entry = __lookup_resource(table, cur_entry->hash);
 233                 if (duplicate_entry) {
 234                         ERROR("The WIM lookup table contains two entries with the "
 235                               "same SHA1 message digest!");
 236                         ERROR("The first entry is:");
 237                         print_lookup_table_entry(duplicate_entry);
 238                         ERROR("The second entry is:");
 239                         print_lookup_table_entry(cur_entry);
 240                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 241                         FREE(cur_entry);
 242                         goto out;
 243                 }
 244                 lookup_table_insert(table, cur_entry);
 245
 246                 if (!(cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
 247                     && (cur_entry->resource_entry.size !=
 248                       cur_entry->resource_entry.original_size))
 249                 {
 250                         ERROR("Found uncompressed resource with original size "
 251                               "not the same as compressed size");
 252                         ERROR("The lookup table entry for the resource is as follows:");
 253                         print_lookup_table_entry(cur_entry);
 254                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 255                         goto out;
 256                 }
 257         }
 258         DEBUG("Done reading lookup table.");
 259         w->lookup_table = table;
 260         return 0;
 261 out:
 262         free_lookup_table(table);
 263         return ret;
 264 }
 265
 266
 267 /*
 268  * Writes a lookup table entry to the output file.
 269  */
 270 int write_lookup_table_entry(struct lookup_table_entry *lte, void *__out)
 271 {
 272         FILE *out;
 273         u8 buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
 274         u8 *p;
 275
 276         out = __out;
 277
 278         /* Don't write entries that have not had file resources or metadata
 279          * resources written for them. */
 280         if (lte->out_refcnt == 0)
 281                 return 0;
 282
 283         if (lte->output_resource_entry.flags & WIM_RESHDR_FLAG_METADATA)
 284                 DEBUG("Writing metadata entry at %lu (orig size = %zu)",
 285                       ftello(out), lte->output_resource_entry.original_size);
 286
 287         p = put_resource_entry(buf, &lte->output_resource_entry);
 288         p = put_u16(p, lte->part_number);
 289         p = put_u32(p, lte->out_refcnt);
 290         p = put_bytes(p, SHA1_HASH_SIZE, lte->hash);
 291         if (fwrite(buf, 1, sizeof(buf), out) != sizeof(buf)) {
 292                 ERROR_WITH_ERRNO("Failed to write lookup table entry");
 293                 return WIMLIB_ERR_WRITE;
 294         }
 295         return 0;
 296 }
 297
 298
 299
 300 int zero_out_refcnts(struct lookup_table_entry *entry, void *ignore)
 301 {
 302         entry->out_refcnt = 0;
 303         return 0;
 304 }
 305
 306 void print_lookup_table_entry(const struct lookup_table_entry *lte)
 307 {
 308         if (!lte) {
 309                 putchar('\n');
 310                 return;
 311         }
 312         printf("Offset            = %"PRIu64" bytes\n",
 313                lte->resource_entry.offset);
 314         printf("Size              = %"PRIu64" bytes\n",
 315                (u64)lte->resource_entry.size);
 316         printf("Original size     = %"PRIu64" bytes\n",
 317                lte->resource_entry.original_size);
 318         printf("Part Number       = %hu\n", lte->part_number);
 319         printf("Reference Count   = %u\n", lte->refcnt);
 320         printf("Hash              = 0x");
 321         print_hash(lte->hash);
 322         putchar('\n');
 323         printf("Flags             = ");
 324         u8 flags = lte->resource_entry.flags;
 325         if (flags & WIM_RESHDR_FLAG_COMPRESSED)
 326                 fputs("WIM_RESHDR_FLAG_COMPRESSED, ", stdout);
 327         if (flags & WIM_RESHDR_FLAG_FREE)
 328                 fputs("WIM_RESHDR_FLAG_FREE, ", stdout);
 329         if (flags & WIM_RESHDR_FLAG_METADATA)
 330                 fputs("WIM_RESHDR_FLAG_METADATA, ", stdout);
 331         if (flags & WIM_RESHDR_FLAG_SPANNED)
 332                 fputs("WIM_RESHDR_FLAG_SPANNED, ", stdout);
 333         putchar('\n');
 334         switch (lte->resource_location) {
 335         case RESOURCE_IN_WIM:
 336                 if (lte->wim->filename) {
 337                         printf("WIM file          = `%s'\n",
 338                                lte->wim->filename);
 339                 }
 340                 break;
 341         case RESOURCE_IN_FILE_ON_DISK:
 342                 printf("File on Disk      = `%s'\n", lte->file_on_disk);
 343                 break;
 344         case RESOURCE_IN_STAGING_FILE:
 345                 printf("Staging File      = `%s'\n", lte->staging_file_name);
 346                 break;
 347         default:
 348                 break;
 349         }
 350         putchar('\n');
 351 }
 352
 353 static int do_print_lookup_table_entry(struct lookup_table_entry *lte,
 354                                        void *ignore)
 355 {
 356         print_lookup_table_entry(lte);
 357         return 0;
 358 }
 359
 360 /*
 361  * Prints the lookup table of a WIM file.
 362  */
 363 WIMLIBAPI void wimlib_print_lookup_table(WIMStruct *w)
 364 {
 365         for_lookup_table_entry(w->lookup_table,
 366                                do_print_lookup_table_entry,
 367                                NULL);
 368 }
 369
 370 /*
 371  * Looks up an entry in the lookup table.
 372  */
 373 struct lookup_table_entry *
 374 __lookup_resource(const struct lookup_table *table, const u8 hash[])
 375 {
 376         size_t i;
 377         struct lookup_table_entry *lte;
 378         struct hlist_node *pos;
 379
 380         i = *(size_t*)hash % table->capacity;
 381         hlist_for_each_entry(lte, pos, &table->array[i], hash_list)
 382                 if (hashes_equal(hash, lte->hash))
 383                         return lte;
 384         return NULL;
 385 }
 386
 387 /*
 388  * Finds the dentry, lookup table entry, and stream index for a WIM file stream,
 389  * given a path name.
 390  *
 391  * This is only for pre-resolved dentries.
 392  */
 393 int lookup_resource(WIMStruct *w, const char *path,
 394                     int lookup_flags,
 395                     struct dentry **dentry_ret,
 396                     struct lookup_table_entry **lte_ret,
 397                     unsigned *stream_idx_ret)
 398 {
 399         struct dentry *dentry;
 400         struct lookup_table_entry *lte;
 401         unsigned stream_idx;
 402         const char *stream_name = NULL;
 403         char *p = NULL;
 404
 405         if (lookup_flags & LOOKUP_FLAG_ADS_OK) {
 406                 stream_name = path_stream_name(path);
 407                 if (stream_name) {
 408                         p = (char*)stream_name - 1;
 409                         *p = '\0';
 410                 }
 411         }
 412
 413         dentry = get_dentry(w, path);
 414         if (p)
 415                 *p = ':';
 416         if (!dentry)
 417                 return -ENOENT;
 418
 419         wimlib_assert(dentry->resolved);
 420
 421         lte = dentry->lte;
 422         if (!(lookup_flags & LOOKUP_FLAG_DIRECTORY_OK)
 423               && dentry_is_directory(dentry))
 424                 return -EISDIR;
 425         stream_idx = 0;
 426         if (stream_name) {
 427                 size_t stream_name_len = strlen(stream_name);
 428                 for (u16 i = 0; i < dentry->num_ads; i++) {
 429                         if (ads_entry_has_name(&dentry->ads_entries[i],
 430                                                stream_name,
 431                                                stream_name_len))
 432                         {
 433                                 stream_idx = i + 1;
 434                                 lte = dentry->ads_entries[i].lte;
 435                                 goto out;
 436                         }
 437                 }
 438                 return -ENOENT;
 439         }
 440 out:
 441         if (dentry_ret)
 442                 *dentry_ret = dentry;
 443         if (lte_ret)
 444                 *lte_ret = lte;
 445         if (stream_idx_ret)
 446                 *stream_idx_ret = stream_idx;
 447         return 0;
 448 }
 449
 450 /* Resolve a dentry's lookup table entries
 451  *
 452  * This replaces the SHA1 hash fields (which are used to lookup an entry in the
 453  * lookup table) with pointers directly to the lookup table entries.  A circular
 454  * linked list of streams sharing the same lookup table entry is created.
 455  *
 456  * This function always succeeds; unresolved lookup table entries are given a
 457  * NULL pointer.
 458  */
 459 int dentry_resolve_ltes(struct dentry *dentry, void *__table)
 460 {
 461         struct lookup_table *table = __table;
 462         struct lookup_table_entry *lte;
 463
 464         if (dentry->resolved)
 465                 return 0;
 466
 467         /* Resolve the default file stream */
 468         lte = __lookup_resource(table, dentry->hash);
 469         if (lte)
 470                 list_add(&dentry->lte_group_list.list, &lte->lte_group_list);
 471         else
 472                 INIT_LIST_HEAD(&dentry->lte_group_list.list);
 473         dentry->lte = lte;
 474         dentry->lte_group_list.type = STREAM_TYPE_NORMAL;
 475         dentry->resolved = true;
 476
 477         /* Resolve the alternate data streams */
 478         if (dentry->ads_entries_status != ADS_ENTRIES_USER) {
 479                 for (u16 i = 0; i < dentry->num_ads; i++) {
 480                         struct ads_entry *cur_entry = &dentry->ads_entries[i];
 481
 482                         lte = __lookup_resource(table, cur_entry->hash);
 483                         if (lte)
 484                                 list_add(&cur_entry->lte_group_list.list,
 485                                          &lte->lte_group_list);
 486                         else
 487                                 INIT_LIST_HEAD(&cur_entry->lte_group_list.list);
 488                         cur_entry->lte = lte;
 489                         cur_entry->lte_group_list.type = STREAM_TYPE_ADS;
 490                 }
 491         }
 492         return 0;
 493 }
 494
 495 /* Return the lookup table entry for the unnamed data stream of a dentry, or
 496  * NULL if there is none.
 497  *
 498  * You'd think this would be easier than it actually is, since the unnamed data
 499  * stream should be the one referenced from the dentry itself.  Alas, if there
 500  * are named data streams, Microsoft's "imagex.exe" program will put the unnamed
 501  * data stream in one of the alternate data streams instead of inside the
 502  * dentry.  So we need to check the alternate data streams too.
 503  *
 504  * Also, note that a dentry may appear to have than one unnamed stream, but if
 505  * the SHA1 message digest is all 0's then the corresponding stream does not
 506  * really "count" (this is the case for the dentry's own file stream when the
 507  * file stream that should be there is actually in one of the alternate stream
 508  * entries.).  This is despite the fact that we may need to extract such a
 509  * missing entry as an empty file or empty named data stream.
 510  */
 511 struct lookup_table_entry *
 512 dentry_unnamed_lte(const struct dentry *dentry,
 513                    const struct lookup_table *table)
 514 {
 515         if (dentry->resolved)
 516                 return dentry_unnamed_lte_resolved(dentry);
 517         else
 518                 return dentry_unnamed_lte_unresolved(dentry, table);
 519 }
 520