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                 duplicate_entry = __lookup_resource(table, cur_entry->hash);
 225                 if (duplicate_entry) {
 226                         ERROR("The WIM lookup table contains two entries with the "
 227                               "same SHA1 message digest!");
 228                         ERROR("The first entry is:");
 229                         print_lookup_table_entry(duplicate_entry);
 230                         ERROR("The second entry is:");
 231                         print_lookup_table_entry(cur_entry);
 232                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 233                         goto out;
 234                 }
 235                 lookup_table_insert(table, cur_entry);
 236
 237                 if (!(cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
 238                     && (cur_entry->resource_entry.size !=
 239                       cur_entry->resource_entry.original_size))
 240                 {
 241                         ERROR("Found uncompressed resource with original size "
 242                               "not the same as compressed size");
 243                         ERROR("The lookup table entry for the resource is as follows:");
 244                         print_lookup_table_entry(cur_entry);
 245                         ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 246                         goto out;
 247                 }
 248         }
 249         DEBUG("Done reading lookup table.");
 250         w->lookup_table = table;
 251         return 0;
 252 out:
 253         free_lookup_table(table);
 254         return ret;
 255 }
 256
 257
 258 /*
 259  * Writes a lookup table entry to the output file.
 260  */
 261 int write_lookup_table_entry(struct lookup_table_entry *lte, void *__out)
 262 {
 263         FILE *out;
 264         u8 buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
 265         u8 *p;
 266
 267         out = __out;
 268
 269         /* do not write lookup table entries for empty files */
 270         if (lte->output_resource_entry.original_size == 0)
 271                 return 0;
 272
 273         /* Don't write entries that have not had file resources or metadata
 274          * resources written for them. */
 275         if (lte->out_refcnt == 0)
 276                 return 0;
 277
 278         if (lte->output_resource_entry.flags & WIM_RESHDR_FLAG_METADATA)
 279                 DEBUG("Writing metadata entry at %lu (orig size = %zu)",
 280                       ftello(out), lte->output_resource_entry.original_size);
 281
 282         p = put_resource_entry(buf, &lte->output_resource_entry);
 283         p = put_u16(p, lte->part_number);
 284         p = put_u32(p, lte->out_refcnt);
 285         p = put_bytes(p, SHA1_HASH_SIZE, lte->hash);
 286         if (fwrite(buf, 1, sizeof(buf), out) != sizeof(buf)) {
 287                 ERROR_WITH_ERRNO("Failed to write lookup table entry");
 288                 return WIMLIB_ERR_WRITE;
 289         }
 290         return 0;
 291 }
 292
 293
 294
 295 int zero_out_refcnts(struct lookup_table_entry *entry, void *ignore)
 296 {
 297         entry->out_refcnt = 0;
 298         return 0;
 299 }
 300
 301 void print_lookup_table_entry(const struct lookup_table_entry *lte)
 302 {
 303         if (!lte) {
 304                 putchar('\n');
 305                 return;
 306         }
 307         printf("Offset            = %"PRIu64" bytes\n",
 308                lte->resource_entry.offset);
 309         printf("Size              = %"PRIu64" bytes\n",
 310                (u64)lte->resource_entry.size);
 311         printf("Original size     = %"PRIu64" bytes\n",
 312                lte->resource_entry.original_size);
 313         printf("Part Number       = %hu\n", lte->part_number);
 314         printf("Reference Count   = %u\n", lte->refcnt);
 315         printf("Hash              = 0x");
 316         print_hash(lte->hash);
 317         putchar('\n');
 318         printf("Flags             = ");
 319         u8 flags = lte->resource_entry.flags;
 320         if (flags & WIM_RESHDR_FLAG_COMPRESSED)
 321                 fputs("WIM_RESHDR_FLAG_COMPRESSED, ", stdout);
 322         if (flags & WIM_RESHDR_FLAG_FREE)
 323                 fputs("WIM_RESHDR_FLAG_FREE, ", stdout);
 324         if (flags & WIM_RESHDR_FLAG_METADATA)
 325                 fputs("WIM_RESHDR_FLAG_METADATA, ", stdout);
 326         if (flags & WIM_RESHDR_FLAG_SPANNED)
 327                 fputs("WIM_RESHDR_FLAG_SPANNED, ", stdout);
 328         putchar('\n');
 329         switch (lte->resource_location) {
 330         case RESOURCE_IN_WIM:
 331                 if (lte->wim->filename) {
 332                         printf("WIM file          = `%s'\n",
 333                                lte->wim->filename);
 334                 }
 335                 break;
 336         case RESOURCE_IN_FILE_ON_DISK:
 337                 printf("File on Disk      = `%s'\n", lte->file_on_disk);
 338                 break;
 339         case RESOURCE_IN_STAGING_FILE:
 340                 printf("Staging File      = `%s'\n", lte->staging_file_name);
 341                 break;
 342         }
 343         putchar('\n');
 344 }
 345
 346 static int do_print_lookup_table_entry(struct lookup_table_entry *lte,
 347                                        void *ignore)
 348 {
 349         print_lookup_table_entry(lte);
 350         return 0;
 351 }
 352
 353 /*
 354  * Prints the lookup table of a WIM file.
 355  */
 356 WIMLIBAPI void wimlib_print_lookup_table(WIMStruct *w)
 357 {
 358         for_lookup_table_entry(w->lookup_table,
 359                                do_print_lookup_table_entry,
 360                                NULL);
 361 }
 362
 363 /*
 364  * Looks up an entry in the lookup table.
 365  */
 366 struct lookup_table_entry *
 367 __lookup_resource(const struct lookup_table *table, const u8 hash[])
 368 {
 369         size_t i;
 370         struct lookup_table_entry *lte;
 371         struct hlist_node *pos;
 372
 373         i = *(size_t*)hash % table->capacity;
 374         hlist_for_each_entry(lte, pos, &table->array[i], hash_list)
 375                 if (hashes_equal(hash, lte->hash))
 376                         return lte;
 377         return NULL;
 378 }
 379
 380 /*
 381  * Finds the dentry, lookup table entry, and stream index for a WIM file stream,
 382  * given a path name.
 383  *
 384  * This is only for pre-resolved dentries.
 385  */
 386 int lookup_resource(WIMStruct *w, const char *path,
 387                     int lookup_flags,
 388                     struct dentry **dentry_ret,
 389                     struct lookup_table_entry **lte_ret,
 390                     unsigned *stream_idx_ret)
 391 {
 392         struct dentry *dentry;
 393         struct lookup_table_entry *lte;
 394         unsigned stream_idx;
 395         const char *stream_name;
 396         char *p = NULL;
 397
 398         if (lookup_flags & LOOKUP_FLAG_ADS_OK) {
 399                 stream_name = path_stream_name(path);
 400                 if (stream_name) {
 401                         p = (char*)stream_name - 1;
 402                         *p = '\0';
 403                 }
 404         }
 405
 406         dentry = get_dentry(w, path);
 407         if (p)
 408                 *p = ':';
 409         if (!dentry)
 410                 return -ENOENT;
 411
 412         wimlib_assert(dentry->resolved);
 413
 414         lte = dentry->lte;
 415         if (!(lookup_flags & LOOKUP_FLAG_DIRECTORY_OK)
 416               && dentry_is_directory(dentry))
 417                 return -EISDIR;
 418         stream_idx = 0;
 419         if (stream_name) {
 420                 size_t stream_name_len = strlen(stream_name);
 421                 for (u16 i = 0; i < dentry->num_ads; i++) {
 422                         if (ads_entry_has_name(&dentry->ads_entries[i],
 423                                                stream_name,
 424                                                stream_name_len))
 425                         {
 426                                 stream_idx = i + 1;
 427                                 lte = dentry->ads_entries[i].lte;
 428                                 goto out;
 429                         }
 430                 }
 431                 return -ENOENT;
 432         }
 433 out:
 434         if (dentry_ret)
 435                 *dentry_ret = dentry;
 436         if (lte_ret)
 437                 *lte_ret = lte;
 438         if (stream_idx_ret)
 439                 *stream_idx_ret = stream_idx;
 440         return 0;
 441 }
 442
 443 /* Resolve a dentry's lookup table entries
 444  *
 445  * This replaces the SHA1 hash fields (which are used to lookup an entry in the
 446  * lookup table) with pointers directly to the lookup table entries.  A circular
 447  * linked list of streams sharing the same lookup table entry is created.
 448  *
 449  * This function always succeeds; unresolved lookup table entries are given a
 450  * NULL pointer.
 451  */
 452 int dentry_resolve_ltes(struct dentry *dentry, void *__table)
 453 {
 454         struct lookup_table *table = __table;
 455         struct lookup_table_entry *lte;
 456
 457         if (dentry->resolved)
 458                 return 0;
 459
 460         /* Resolve the default file stream */
 461         lte = __lookup_resource(table, dentry->hash);
 462         if (lte)
 463                 list_add(&dentry->lte_group_list.list, &lte->lte_group_list);
 464         else
 465                 INIT_LIST_HEAD(&dentry->lte_group_list.list);
 466         dentry->lte = lte;
 467         dentry->lte_group_list.type = STREAM_TYPE_NORMAL;
 468         dentry->resolved = true;
 469
 470         /* Resolve the alternate data streams */
 471         if (dentry->ads_entries_status != ADS_ENTRIES_USER) {
 472                 for (u16 i = 0; i < dentry->num_ads; i++) {
 473                         struct ads_entry *cur_entry = &dentry->ads_entries[i];
 474
 475                         lte = __lookup_resource(table, cur_entry->hash);
 476                         if (lte)
 477                                 list_add(&cur_entry->lte_group_list.list,
 478                                          &lte->lte_group_list);
 479                         else
 480                                 INIT_LIST_HEAD(&cur_entry->lte_group_list.list);
 481                         cur_entry->lte = lte;
 482                         cur_entry->lte_group_list.type = STREAM_TYPE_ADS;
 483                 }
 484         }
 485         return 0;
 486 }
 487
 488 struct lookup_table_entry *
 489 dentry_first_lte(const struct dentry *dentry, const struct lookup_table *table)
 490 {
 491         if (dentry->resolved)
 492                 return dentry_first_lte_resolved(dentry);
 493         else
 494                 return dentry_first_lte_unresolved(dentry, table);
 495 }
 496