wimlib.net Git - wimlib/blob - src/dentry.c

   1 /*
   2  * dentry.c
   3  *
   4  * A dentry (directory entry) contains the metadata for a file.  In the WIM file
   5  * format, the dentries are stored in the "metadata resource" section right
   6  * after the security data.  Each image in the WIM file has its own metadata
   7  * resource with its own security data and dentry tree.  Dentries in different
   8  * images may share file resources by referring to the same lookup table
   9  * entries.
  10  */
  11
  12 /*
  13  *
  14  * Copyright (C) 2010 Carl Thijssen
  15  * Copyright (C) 2012 Eric Biggers
  16  *
  17  * This file is part of wimlib, a library for working with WIM files.
  18  *
  19  * wimlib is free software; you can redistribute it and/or modify it under the
  20  * terms of the GNU Lesser General Public License as published by the Free
  21  * Software Foundation; either version 2.1 of the License, or (at your option)
  22  * any later version.
  23  *
  24  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  25  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  26  * A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
  27  * details.
  28  *
  29  * You should have received a copy of the GNU Lesser General Public License
  30  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  31  */
  32
  33 #include "wimlib_internal.h"
  34 #include "dentry.h"
  35 #include "io.h"
  36 #include "timestamp.h"
  37 #include "lookup_table.h"
  38 #include "sha1.h"
  39 #include <unistd.h>
  40 #include <sys/stat.h>
  41
  42 /* Real length of a dentry, including the alternate data stream entries, which
  43  * are not included in the dentry->length field... */
  44 u64 dentry_total_length(const struct dentry *dentry)
  45 {
  46         u64 length = (dentry->length + 7) & ~7;
  47         for (u16 i = 0 ; i < dentry->num_ads; i++)
  48                 length += ads_entry_length(&dentry->ads_entries[i]);
  49         return length;
  50 }
  51
  52 /* Transfers file attributes from a `stat' buffer to a struct dentry. */
  53 void stbuf_to_dentry(const struct stat *stbuf, struct dentry *dentry)
  54 {
  55         if (S_ISLNK(stbuf->st_mode)) {
  56                 dentry->attributes = FILE_ATTRIBUTE_REPARSE_POINT;
  57                 dentry->reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
  58         } else if (S_ISDIR(stbuf->st_mode)) {
  59                 dentry->attributes = FILE_ATTRIBUTE_DIRECTORY;
  60         } else {
  61                 dentry->attributes = FILE_ATTRIBUTE_NORMAL;
  62         }
  63 }
  64
  65 /* Transfers file attributes from a struct dentry to a `stat' buffer. */
  66 void dentry_to_stbuf(const struct dentry *dentry, struct stat *stbuf,
  67                      const struct lookup_table *table)
  68 {
  69         struct lookup_table_entry *lte;
  70
  71         if (dentry_is_symlink(dentry))
  72                 stbuf->st_mode = S_IFLNK | 0777;
  73         else if (dentry_is_directory(dentry))
  74                 stbuf->st_mode = S_IFDIR | 0755;
  75         else
  76                 stbuf->st_mode = S_IFREG | 0644;
  77
  78         if (table)
  79                 lte = lookup_resource(table, dentry->hash);
  80         else
  81                 lte = NULL;
  82
  83         if (lte) {
  84                 stbuf->st_nlink = lte->refcnt;
  85                 stbuf->st_size = lte->resource_entry.original_size;
  86         } else {
  87                 stbuf->st_nlink = 1;
  88                 stbuf->st_size = 0;
  89         }
  90         stbuf->st_uid     = getuid();
  91         stbuf->st_gid     = getgid();
  92         stbuf->st_atime   = ms_timestamp_to_unix(dentry->last_access_time);
  93         stbuf->st_mtime   = ms_timestamp_to_unix(dentry->last_write_time);
  94         stbuf->st_ctime   = ms_timestamp_to_unix(dentry->creation_time);
  95         stbuf->st_blocks  = (stbuf->st_size + 511) / 512;
  96 }
  97
  98 /* Makes all timestamp fields for the dentry be the current time. */
  99 void dentry_update_all_timestamps(struct dentry *dentry)
 100 {
 101         u64 now = get_timestamp();
 102         dentry->creation_time    = now;
 103         dentry->last_access_time = now;
 104         dentry->last_write_time  = now;
 105 }
 106
 107 /*
 108  * Calls a function on all directory entries in a directory tree.  It is called
 109  * on a parent before its children.
 110  */
 111 int for_dentry_in_tree(struct dentry *root,
 112                        int (*visitor)(struct dentry*, void*), void *arg)
 113 {
 114         int ret;
 115         struct dentry *child;
 116
 117         ret = visitor(root, arg);
 118
 119         if (ret != 0)
 120                 return ret;
 121
 122         child = root->children;
 123
 124         if (!child)
 125                 return 0;
 126
 127         do {
 128                 ret = for_dentry_in_tree(child, visitor, arg);
 129                 if (ret != 0)
 130                         return ret;
 131                 child = child->next;
 132         } while (child != root->children);
 133         return 0;
 134 }
 135
 136 /*
 137  * Like for_dentry_in_tree(), but the visitor function is always called on a
 138  * dentry's children before on itself.
 139  */
 140 int for_dentry_in_tree_depth(struct dentry *root,
 141                              int (*visitor)(struct dentry*, void*), void *arg)
 142 {
 143         int ret;
 144         struct dentry *child;
 145         struct dentry *next;
 146
 147         child = root->children;
 148         if (child) {
 149                 do {
 150                         next = child->next;
 151                         ret = for_dentry_in_tree_depth(child, visitor, arg);
 152                         if (ret != 0)
 153                                 return ret;
 154                         child = next;
 155                 } while (child != root->children);
 156         }
 157         return visitor(root, arg);
 158 }
 159
 160 /*
 161  * Calculate the full path of @dentry, based on its parent's full path and on
 162  * its UTF-8 file name.
 163  */
 164 int calculate_dentry_full_path(struct dentry *dentry, void *ignore)
 165 {
 166         char *full_path;
 167         u32 full_path_len;
 168         if (dentry_is_root(dentry)) {
 169                 full_path = MALLOC(2);
 170                 if (!full_path)
 171                         goto oom;
 172                 full_path[0] = '/';
 173                 full_path[1] = '\0';
 174                 full_path_len = 1;
 175         } else {
 176                 char *parent_full_path;
 177                 u32 parent_full_path_len;
 178                 const struct dentry *parent = dentry->parent;
 179
 180                 if (dentry_is_root(parent)) {
 181                         parent_full_path = "";
 182                         parent_full_path_len = 0;
 183                 } else {
 184                         parent_full_path = parent->full_path_utf8;
 185                         parent_full_path_len = parent->full_path_utf8_len;
 186                 }
 187
 188                 full_path_len = parent_full_path_len + 1 +
 189                                 dentry->file_name_utf8_len;
 190                 full_path = MALLOC(full_path_len + 1);
 191                 if (!full_path)
 192                         goto oom;
 193
 194                 memcpy(full_path, parent_full_path, parent_full_path_len);
 195                 full_path[parent_full_path_len] = '/';
 196                 memcpy(full_path + parent_full_path_len + 1,
 197                        dentry->file_name_utf8,
 198                        dentry->file_name_utf8_len);
 199                 full_path[full_path_len] = '\0';
 200         }
 201         FREE(dentry->full_path_utf8);
 202         dentry->full_path_utf8 = full_path;
 203         dentry->full_path_utf8_len = full_path_len;
 204         return 0;
 205 oom:
 206         ERROR("Out of memory while calculating dentry full path");
 207         return WIMLIB_ERR_NOMEM;
 208 }
 209
 210 /*
 211  * Recursively calculates the subdir offsets for a directory tree.
 212  *
 213  * @dentry:  The root of the directory tree.
 214  * @subdir_offset_p:  The current subdirectory offset; i.e., the subdirectory
 215  *      offset for @dentry.
 216  */
 217 void calculate_subdir_offsets(struct dentry *dentry, u64 *subdir_offset_p)
 218 {
 219         struct dentry *child;
 220
 221         child = dentry->children;
 222         dentry->subdir_offset = *subdir_offset_p;
 223         if (child) {
 224
 225                 /* Advance the subdir offset by the amount of space the children
 226                  * of this dentry take up. */
 227                 do {
 228                         *subdir_offset_p += dentry_total_length(child);
 229                         child = child->next;
 230                 } while (child != dentry->children);
 231
 232                 /* End-of-directory dentry on disk. */
 233                 *subdir_offset_p += 8;
 234
 235                 /* Recursively call calculate_subdir_offsets() on all the
 236                  * children. */
 237                 do {
 238                         calculate_subdir_offsets(child, subdir_offset_p);
 239                         child = child->next;
 240                 } while (child != dentry->children);
 241         } else {
 242                 /* On disk, childless directories have a valid subdir_offset
 243                  * that points to an 8-byte end-of-directory dentry.  Regular
 244                  * files have a subdir_offset of 0. */
 245                 if (dentry_is_directory(dentry))
 246                         *subdir_offset_p += 8;
 247                 else
 248                         dentry->subdir_offset = 0;
 249         }
 250 }
 251
 252
 253 /* Returns the child of @dentry that has the file name @name.
 254  * Returns NULL if no child has the name. */
 255 struct dentry *get_dentry_child_with_name(const struct dentry *dentry,
 256                                                         const char *name)
 257 {
 258         struct dentry *child;
 259         size_t name_len;
 260
 261         child = dentry->children;
 262         if (child) {
 263                 name_len = strlen(name);
 264                 do {
 265                         if (dentry_has_name(child, name, name_len))
 266                                 return child;
 267                         child = child->next;
 268                 } while (child != dentry->children);
 269         }
 270         return NULL;
 271 }
 272
 273 /* Retrieves the dentry that has the UTF-8 @path relative to the dentry
 274  * @cur_dir.  Returns NULL if no dentry having the path is found. */
 275 static struct dentry *get_dentry_relative_path(struct dentry *cur_dir, const char *path)
 276 {
 277         struct dentry *child;
 278         size_t base_len;
 279         const char *new_path;
 280
 281         if (*path == '\0')
 282                 return cur_dir;
 283
 284         child = cur_dir->children;
 285         if (child) {
 286                 new_path = path_next_part(path, &base_len);
 287                 do {
 288                         if (dentry_has_name(child, path, base_len))
 289                                 return get_dentry_relative_path(child, new_path);
 290                         child = child->next;
 291                 } while (child != cur_dir->children);
 292         }
 293         return NULL;
 294 }
 295
 296 /* Returns the dentry corresponding to the UTF-8 @path, or NULL if there is no
 297  * such dentry. */
 298 struct dentry *get_dentry(WIMStruct *w, const char *path)
 299 {
 300         struct dentry *root = wim_root_dentry(w);
 301         while (*path == '/')
 302                 path++;
 303         return get_dentry_relative_path(root, path);
 304 }
 305
 306 /* Returns the parent directory for the @path. */
 307 struct dentry *get_parent_dentry(WIMStruct *w, const char *path)
 308 {
 309         size_t path_len = strlen(path);
 310         char buf[path_len + 1];
 311
 312         memcpy(buf, path, path_len + 1);
 313
 314         to_parent_name(buf, path_len);
 315
 316         return get_dentry(w, buf);
 317 }
 318
 319 /* Prints the full path of a dentry. */
 320 int print_dentry_full_path(struct dentry *dentry, void *ignore)
 321 {
 322         if (dentry->full_path_utf8)
 323                 puts(dentry->full_path_utf8);
 324         return 0;
 325 }
 326
 327 struct file_attr_flag {
 328         u32 flag;
 329         const char *name;
 330 };
 331 struct file_attr_flag file_attr_flags[] = {
 332         {FILE_ATTRIBUTE_READONLY,               "READONLY"},
 333         {FILE_ATTRIBUTE_HIDDEN,         "HIDDEN"},
 334         {FILE_ATTRIBUTE_SYSTEM,         "SYSTEM"},
 335         {FILE_ATTRIBUTE_DIRECTORY,              "DIRECTORY"},
 336         {FILE_ATTRIBUTE_ARCHIVE,                "ARCHIVE"},
 337         {FILE_ATTRIBUTE_DEVICE,         "DEVICE"},
 338         {FILE_ATTRIBUTE_NORMAL,         "NORMAL"},
 339         {FILE_ATTRIBUTE_TEMPORARY,              "TEMPORARY"},
 340         {FILE_ATTRIBUTE_SPARSE_FILE,    "SPARSE_FILE"},
 341         {FILE_ATTRIBUTE_REPARSE_POINT,  "REPARSE_POINT"},
 342         {FILE_ATTRIBUTE_COMPRESSED,             "COMPRESSED"},
 343         {FILE_ATTRIBUTE_OFFLINE,                "OFFLINE"},
 344         {FILE_ATTRIBUTE_NOT_CONTENT_INDEXED,"NOT_CONTENT_INDEXED"},
 345         {FILE_ATTRIBUTE_ENCRYPTED,              "ENCRYPTED"},
 346         {FILE_ATTRIBUTE_VIRTUAL,                "VIRTUAL"},
 347 };
 348
 349 /* Prints a directory entry.  @lookup_table is a pointer to the lookup table, or
 350  * NULL if the resource entry for the dentry is not to be printed. */
 351 int print_dentry(struct dentry *dentry, void *lookup_table)
 352 {
 353         struct lookup_table_entry *lte;
 354         unsigned i;
 355
 356         printf("[DENTRY]\n");
 357         printf("Length            = %"PRIu64"\n", dentry->length);
 358         printf("Attributes        = 0x%x\n", dentry->attributes);
 359         for (i = 0; i < ARRAY_LEN(file_attr_flags); i++)
 360                 if (file_attr_flags[i].flag & dentry->attributes)
 361                         printf("    FILE_ATTRIBUTE_%s is set\n",
 362                                 file_attr_flags[i].name);
 363         printf("Security ID       = %d\n", dentry->security_id);
 364         printf("Subdir offset     = %"PRIu64"\n", dentry->subdir_offset);
 365         /*printf("Unused1           = 0x%"PRIu64"\n", dentry->unused1);*/
 366         /*printf("Unused2           = %"PRIu64"\n", dentry->unused2);*/
 367         printf("Creation Time     = 0x%"PRIx64"\n", dentry->creation_time);
 368         printf("Last Access Time  = 0x%"PRIx64"\n", dentry->last_access_time);
 369         printf("Last Write Time   = 0x%"PRIx64"\n", dentry->last_write_time);
 370         printf("Hash              = 0x");
 371         print_hash(dentry->hash);
 372         putchar('\n');
 373         printf("Reparse Tag       = 0x%"PRIx32"\n", dentry->reparse_tag);
 374         printf("Hard Link Group   = 0x%"PRIx64"\n", dentry->hard_link);
 375         printf("Number of Alternate Data Streams = %hu\n", dentry->num_ads);
 376         printf("Filename          = \"");
 377         print_string(dentry->file_name, dentry->file_name_len);
 378         puts("\"");
 379         printf("Filename Length   = %hu\n", dentry->file_name_len);
 380         printf("Filename (UTF-8)  = \"%s\"\n", dentry->file_name_utf8);
 381         printf("Filename (UTF-8) Length = %hu\n", dentry->file_name_utf8_len);
 382         printf("Short Name        = \"");
 383         print_string(dentry->short_name, dentry->short_name_len);
 384         puts("\"");
 385         printf("Short Name Length = %hu\n", dentry->short_name_len);
 386         printf("Full Path (UTF-8) = \"%s\"\n", dentry->full_path_utf8);
 387         if (lookup_table) {
 388                 lte = lookup_resource(lookup_table, dentry->hash);
 389                 if (lte)
 390                         print_lookup_table_entry(lte, NULL);
 391                 else
 392                         putchar('\n');
 393         } else {
 394                 putchar('\n');
 395         }
 396         for (u16 i = 0; i < dentry->num_ads; i++) {
 397                 printf("[Alternate Stream Entry %u]\n", i);
 398                 printf("Name = \"%s\"\n", dentry->ads_entries[i].stream_name_utf8);
 399                 printf("Name Length (UTF-16) = %u\n",
 400                                 dentry->ads_entries[i].stream_name_len);
 401                 printf("Hash              = 0x");
 402                 print_hash(dentry->ads_entries[i].hash);
 403                 putchar('\n');
 404                 lte = lookup_resource(lookup_table, dentry->ads_entries[i].hash);
 405                 if (lte)
 406                         print_lookup_table_entry(lte, NULL);
 407                 else
 408                         putchar('\n');
 409         }
 410         return 0;
 411 }
 412
 413 static inline void dentry_common_init(struct dentry *dentry)
 414 {
 415         memset(dentry, 0, sizeof(struct dentry));
 416         dentry->refcnt = 1;
 417         dentry->security_id = -1;
 418 }
 419
 420 /*
 421  * Creates an unlinked directory entry.
 422  *
 423  * @name:    The base name of the new dentry.
 424  * @return:  A pointer to the new dentry, or NULL if out of memory.
 425  */
 426 struct dentry *new_dentry(const char *name)
 427 {
 428         struct dentry *dentry;
 429
 430         dentry = MALLOC(sizeof(struct dentry));
 431         if (!dentry)
 432                 return NULL;
 433
 434         dentry_common_init(dentry);
 435         if (change_dentry_name(dentry, name) != 0) {
 436                 FREE(dentry);
 437                 return NULL;
 438         }
 439
 440         dentry_update_all_timestamps(dentry);
 441         dentry->next   = dentry;
 442         dentry->prev   = dentry;
 443         dentry->parent = dentry;
 444         return dentry;
 445 }
 446
 447 static void dentry_free_ads_entries(struct dentry *dentry)
 448 {
 449         for (u16 i = 0; i < dentry->num_ads; i++) {
 450                 FREE(dentry->ads_entries[i].stream_name);
 451                 FREE(dentry->ads_entries[i].stream_name_utf8);
 452         }
 453         FREE(dentry->ads_entries);
 454         dentry->ads_entries = NULL;
 455         dentry->num_ads = 0;
 456 }
 457
 458
 459 void free_dentry(struct dentry *dentry)
 460 {
 461         FREE(dentry->file_name);
 462         FREE(dentry->file_name_utf8);
 463         FREE(dentry->short_name);
 464         FREE(dentry->full_path_utf8);
 465         dentry_free_ads_entries(dentry);
 466         FREE(dentry);
 467 }
 468
 469 /* Arguments for do_free_dentry(). */
 470 struct free_dentry_args {
 471         struct lookup_table *lookup_table;
 472         bool lt_decrement_refcnt;
 473 };
 474
 475 /*
 476  * This function is passed as an argument to for_dentry_in_tree_depth() in order
 477  * to free a directory tree.  __args is a pointer to a `struct free_dentry_args'.
 478  */
 479 static int do_free_dentry(struct dentry *dentry, void *__args)
 480 {
 481         struct free_dentry_args *args = (struct free_dentry_args*)__args;
 482
 483         if (args->lt_decrement_refcnt && !dentry_is_directory(dentry)) {
 484                 lookup_table_decrement_refcnt(args->lookup_table,
 485                                               dentry->hash);
 486         }
 487
 488         wimlib_assert(dentry->refcnt >= 1);
 489         if (--dentry->refcnt == 0)
 490                 free_dentry(dentry);
 491         return 0;
 492 }
 493
 494 /*
 495  * Unlinks and frees a dentry tree.
 496  *
 497  * @root:               The root of the tree.
 498  * @lookup_table:       The lookup table for dentries.
 499  * @decrement_refcnt:   True if the dentries in the tree are to have their
 500  *                      reference counts in the lookup table decremented.
 501  */
 502 void free_dentry_tree(struct dentry *root, struct lookup_table *lookup_table,
 503                       bool lt_decrement_refcnt)
 504 {
 505         if (!root || !root->parent)
 506                 return;
 507
 508         struct free_dentry_args args;
 509         args.lookup_table        = lookup_table;
 510         args.lt_decrement_refcnt = lt_decrement_refcnt;
 511         for_dentry_in_tree_depth(root, do_free_dentry, &args);
 512 }
 513
 514 int increment_dentry_refcnt(struct dentry *dentry, void *ignore)
 515 {
 516         dentry->refcnt++;
 517         return 0;
 518 }
 519
 520 /*
 521  * Links a dentry into the directory tree.
 522  *
 523  * @dentry: The dentry to link.
 524  * @parent: The dentry that will be the parent of @dentry.
 525  */
 526 void link_dentry(struct dentry *dentry, struct dentry *parent)
 527 {
 528         dentry->parent = parent;
 529         if (parent->children) {
 530                 /* Not an only child; link to siblings. */
 531                 dentry->next = parent->children;
 532                 dentry->prev = parent->children->prev;
 533                 dentry->next->prev = dentry;
 534                 dentry->prev->next = dentry;
 535         } else {
 536                 /* Only child; link to parent. */
 537                 parent->children = dentry;
 538                 dentry->next = dentry;
 539                 dentry->prev = dentry;
 540         }
 541 }
 542
 543 /* Unlink a dentry from the directory tree. */
 544 void unlink_dentry(struct dentry *dentry)
 545 {
 546         if (dentry_is_root(dentry))
 547                 return;
 548         if (dentry_is_only_child(dentry)) {
 549                 dentry->parent->children = NULL;
 550         } else {
 551                 if (dentry_is_first_sibling(dentry))
 552                         dentry->parent->children = dentry->next;
 553                 dentry->next->prev = dentry->prev;
 554                 dentry->prev->next = dentry->next;
 555         }
 556 }
 557
 558
 559 /* Recalculates the length of @dentry based on its file name length and short
 560  * name length.  */
 561 static inline void recalculate_dentry_size(struct dentry *dentry)
 562 {
 563         dentry->length = WIM_DENTRY_DISK_SIZE + dentry->file_name_len +
 564                          2 + dentry->short_name_len;
 565         /* Must be multiple of 8. */
 566         dentry->length = (dentry->length + 7) & ~7;
 567 }
 568
 569 /* Changes the name of a dentry to @new_name.  Only changes the file_name and
 570  * file_name_utf8 fields; does not change the short_name, short_name_utf8, or
 571  * full_path_utf8 fields.  Also recalculates its length. */
 572 int change_dentry_name(struct dentry *dentry, const char *new_name)
 573 {
 574         size_t utf8_len;
 575         size_t utf16_len;
 576
 577         FREE(dentry->file_name);
 578
 579         utf8_len = strlen(new_name);
 580
 581         dentry->file_name = utf8_to_utf16(new_name, utf8_len, &utf16_len);
 582
 583         if (!dentry->file_name)
 584                 return WIMLIB_ERR_NOMEM;
 585
 586         FREE(dentry->file_name_utf8);
 587         dentry->file_name_utf8 = MALLOC(utf8_len + 1);
 588         if (!dentry->file_name_utf8) {
 589                 FREE(dentry->file_name);
 590                 dentry->file_name = NULL;
 591                 return WIMLIB_ERR_NOMEM;
 592         }
 593
 594         dentry->file_name_len = utf16_len;
 595         dentry->file_name_utf8_len = utf8_len;
 596         memcpy(dentry->file_name_utf8, new_name, utf8_len + 1);
 597         recalculate_dentry_size(dentry);
 598         return 0;
 599 }
 600
 601 /* Parameters for calculate_dentry_statistics(). */
 602 struct image_statistics {
 603         struct lookup_table *lookup_table;
 604         u64 *dir_count;
 605         u64 *file_count;
 606         u64 *total_bytes;
 607         u64 *hard_link_bytes;
 608 };
 609
 610 static int calculate_dentry_statistics(struct dentry *dentry, void *arg)
 611 {
 612         struct image_statistics *stats;
 613         struct lookup_table_entry *lte;
 614
 615         stats = arg;
 616         lte = lookup_resource(stats->lookup_table, dentry->hash);
 617
 618         if (dentry_is_directory(dentry) && !dentry_is_root(dentry))
 619                 ++*stats->dir_count;
 620         else
 621                 ++*stats->file_count;
 622
 623         if (lte) {
 624                 u64 size = lte->resource_entry.original_size;
 625                 *stats->total_bytes += size;
 626                 if (++lte->out_refcnt == 1)
 627                         *stats->hard_link_bytes += size;
 628         }
 629         return 0;
 630 }
 631
 632 void calculate_dir_tree_statistics(struct dentry *root, struct lookup_table *table,
 633                                    u64 *dir_count_ret, u64 *file_count_ret,
 634                                    u64 *total_bytes_ret,
 635                                    u64 *hard_link_bytes_ret)
 636 {
 637         struct image_statistics stats;
 638         *dir_count_ret         = 0;
 639         *file_count_ret        = 0;
 640         *total_bytes_ret       = 0;
 641         *hard_link_bytes_ret   = 0;
 642         stats.lookup_table     = table;
 643         stats.dir_count       = dir_count_ret;
 644         stats.file_count      = file_count_ret;
 645         stats.total_bytes     = total_bytes_ret;
 646         stats.hard_link_bytes = hard_link_bytes_ret;
 647         for_lookup_table_entry(table, zero_out_refcnts, NULL);
 648         for_dentry_in_tree(root, calculate_dentry_statistics, &stats);
 649 }
 650
 651 static int read_ads_entries(const u8 *p, struct dentry *dentry,
 652                             u64 remaining_size)
 653 {
 654         u16 num_ads = dentry->num_ads;
 655         struct ads_entry *ads_entries = CALLOC(num_ads, sizeof(struct ads_entry));
 656         int ret;
 657         if (!ads_entries) {
 658                 ERROR("Could not allocate memory for %"PRIu16" "
 659                       "alternate data stream entries", num_ads);
 660                 return WIMLIB_ERR_NOMEM;
 661         }
 662         DEBUG2("Reading %"PRIu16" alternate data streams "
 663                "(remaining size = %"PRIu64")", num_ads, remaining_size);
 664
 665         for (u16 i = 0; i < num_ads; i++) {
 666                 struct ads_entry *cur_entry = &ads_entries[i];
 667                 u64 length;
 668                 size_t utf8_len;
 669                 const char *p_save = p;
 670                 /* Read the base stream entry, excluding the stream name. */
 671                 if (remaining_size < WIM_ADS_ENTRY_DISK_SIZE) {
 672                         ERROR("Stream entries go past end of metadata resource");
 673                         ERROR("(remaining_size = %"PRIu64")", remaining_size);
 674                         ret = WIMLIB_ERR_INVALID_DENTRY;
 675                         goto out_free_ads_entries;
 676                 }
 677                 remaining_size -= WIM_ADS_ENTRY_DISK_SIZE;
 678
 679                 p = get_u64(p, &length); /* ADS entry length */
 680
 681                 DEBUG2("ADS length = %"PRIu64, length);
 682
 683                 p += 8; /* Unused */
 684                 p = get_bytes(p, WIM_HASH_SIZE, (u8*)cur_entry->hash);
 685                 p = get_u16(p, &cur_entry->stream_name_len);
 686
 687                 DEBUG2("Stream name length = %u", cur_entry->stream_name_len);
 688
 689                 cur_entry->stream_name = NULL;
 690                 cur_entry->stream_name_utf8 = NULL;
 691
 692                 if (remaining_size < cur_entry->stream_name_len + 2) {
 693                         ERROR("Stream entries go past end of metadata resource");
 694                         ERROR("(remaining_size = %"PRIu64" bytes, stream_name_len "
 695                               "= %"PRIu16" bytes", remaining_size,
 696                               cur_entry->stream_name_len);
 697                         ret = WIMLIB_ERR_INVALID_DENTRY;
 698                         goto out_free_ads_entries;
 699                 }
 700                 remaining_size -= cur_entry->stream_name_len + 2;
 701
 702                 cur_entry->stream_name = MALLOC(cur_entry->stream_name_len);
 703                 if (!cur_entry->stream_name) {
 704                         ret = WIMLIB_ERR_NOMEM;
 705                         goto out_free_ads_entries;
 706                 }
 707                 get_bytes(p, cur_entry->stream_name_len,
 708                           (u8*)cur_entry->stream_name);
 709                 cur_entry->stream_name_utf8 = utf16_to_utf8(cur_entry->stream_name,
 710                                                             cur_entry->stream_name_len,
 711                                                             &utf8_len);
 712                 cur_entry->stream_name_len_utf8 = utf8_len;
 713
 714                 if (!cur_entry->stream_name_utf8) {
 715                         ret = WIMLIB_ERR_NOMEM;
 716                         goto out_free_ads_entries;
 717                 }
 718                 p = p_save + ads_entry_length(cur_entry);
 719         }
 720         dentry->ads_entries = ads_entries;
 721         return 0;
 722 out_free_ads_entries:
 723         for (u16 i = 0; i < num_ads; i++) {
 724                 FREE(ads_entries[i].stream_name);
 725                 FREE(ads_entries[i].stream_name_utf8);
 726         }
 727         FREE(ads_entries);
 728         return ret;
 729 }
 730
 731 /*
 732  * Reads a directory entry from the metadata resource.
 733  */
 734 int read_dentry(const u8 metadata_resource[], u64 metadata_resource_len,
 735                 u64 offset, struct dentry *dentry)
 736 {
 737         const u8 *p;
 738         u64 calculated_size;
 739         char *file_name;
 740         char *file_name_utf8;
 741         char *short_name;
 742         u16 short_name_len;
 743         u16 file_name_len;
 744         size_t file_name_utf8_len;
 745         int ret;
 746
 747         dentry_common_init(dentry);
 748
 749         /*Make sure the dentry really fits into the metadata resource.*/
 750         if (offset + 8 > metadata_resource_len) {
 751                 ERROR("Directory entry starting at %"PRIu64" ends past the "
 752                       "end of the metadata resource (size %"PRIu64")",
 753                       offset, metadata_resource_len);
 754                 return WIMLIB_ERR_INVALID_DENTRY;
 755         }
 756
 757         /* Before reading the whole entry, we need to read just the length.
 758          * This is because an entry of length 8 (that is, just the length field)
 759          * terminates the list of sibling directory entries. */
 760
 761         p = get_u64(&metadata_resource[offset], &dentry->length);
 762
 763         /* A zero length field (really a length of 8, since that's how big the
 764          * directory entry is...) indicates that this is the end of directory
 765          * dentry.  We do not read it into memory as an actual dentry, so just
 766          * return true in that case. */
 767         if (dentry->length == 0)
 768                 return 0;
 769
 770         if (offset + dentry->length >= metadata_resource_len) {
 771                 ERROR("Directory entry at offset %"PRIu64" and with size "
 772                       "%"PRIu64" ends past the end of the metadata resource "
 773                       "(size %"PRIu64")",
 774                       offset, dentry->length, metadata_resource_len);
 775                 return WIMLIB_ERR_INVALID_DENTRY;
 776         }
 777
 778         /* If it is a recognized length, read the rest of the directory entry.
 779          * Note: The root directory entry has no name, and its length does not
 780          * include the short name length field.  */
 781         if (dentry->length < WIM_DENTRY_DISK_SIZE) {
 782                 ERROR("Directory entry has invalid length of %"PRIu64" bytes",
 783                       dentry->length);
 784                 return WIMLIB_ERR_INVALID_DENTRY;
 785         }
 786
 787         p = get_u32(p, &dentry->attributes);
 788         p = get_u32(p, (u32*)&dentry->security_id);
 789         p = get_u64(p, &dentry->subdir_offset);
 790
 791         /* 2 unused fields */
 792         p += 2 * sizeof(u64);
 793         /*p = get_u64(p, &dentry->unused1);*/
 794         /*p = get_u64(p, &dentry->unused2);*/
 795
 796         p = get_u64(p, &dentry->creation_time);
 797         p = get_u64(p, &dentry->last_access_time);
 798         p = get_u64(p, &dentry->last_write_time);
 799
 800         p = get_bytes(p, WIM_HASH_SIZE, dentry->hash);
 801
 802         /*
 803          * I don't know what's going on here.  It seems like M$ screwed up the
 804          * reparse points, then put the fields in the same place and didn't
 805          * document it.  The WIM_HDR_FLAG_RP_FIX flag in the WIM header might
 806          * have something to do with this, but it's not documented.
 807          */
 808         if (dentry->attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
 809                 /* ??? */
 810                 u32 u1, u2;
 811                 p = get_u32(p, &u1);
 812                 /*p += 4;*/
 813                 p = get_u32(p, &dentry->reparse_tag);
 814                 p = get_u32(p, &u2);
 815                 /*p += 4;*/
 816                 dentry->hard_link = (u64)(u1) | ((u64)(u2) << 32);
 817         } else {
 818                 p = get_u32(p, &dentry->reparse_tag);
 819                 p = get_u64(p, &dentry->hard_link);
 820         }
 821
 822         /* By the way, the reparse_reserved field does not actually exist (at
 823          * least when the file is not a reparse point) */
 824
 825
 826         p = get_u16(p, &dentry->num_ads);
 827
 828         p = get_u16(p, &short_name_len);
 829         p = get_u16(p, &file_name_len);
 830
 831         calculated_size = WIM_DENTRY_DISK_SIZE + file_name_len + 2 +
 832                           short_name_len;
 833
 834         if (dentry->length < calculated_size) {
 835                 ERROR("Unexpected end of directory entry! (Expected "
 836                       "%"PRIu64" bytes, got %"PRIu64" bytes. "
 837                       "short_name_len = %hu, file_name_len = %hu)",
 838                       calculated_size, dentry->length,
 839                       short_name_len, file_name_len);
 840                 return WIMLIB_ERR_INVALID_DENTRY;
 841         }
 842
 843         /* Read the filename. */
 844         file_name = MALLOC(file_name_len);
 845         if (!file_name) {
 846                 ERROR("Failed to allocate %hu bytes for dentry file name",
 847                       file_name_len);
 848                 return WIMLIB_ERR_NOMEM;
 849         }
 850         p = get_bytes(p, file_name_len, file_name);
 851
 852         /* Convert filename to UTF-8. */
 853         file_name_utf8 = utf16_to_utf8(file_name, file_name_len,
 854                                        &file_name_utf8_len);
 855
 856         if (!file_name_utf8) {
 857                 ERROR("Failed to allocate memory to convert UTF-16 "
 858                       "filename (%hu bytes) to UTF-8", file_name_len);
 859                 ret = WIMLIB_ERR_NOMEM;
 860                 goto out_free_file_name;
 861         }
 862
 863         /* Undocumented padding between file name and short name.  This probably
 864          * is supposed to be a terminating null character. */
 865         p += 2;
 866
 867         /* Read the short filename. */
 868         short_name = MALLOC(short_name_len);
 869         if (!short_name) {
 870                 ERROR("Failed to allocate %hu bytes for short filename",
 871                       short_name_len);
 872                 ret = WIMLIB_ERR_NOMEM;
 873                 goto out_free_file_name_utf8;
 874         }
 875
 876         p = get_bytes(p, short_name_len, short_name);
 877
 878         /* Some directory entries inexplicibly have a little over 70 bytes of
 879          * extra data.  The exact amount of data seems to be 72 bytes, but it is
 880          * aligned on the next 8-byte boundary.  Here's an example of the
 881          * aligned data:
 882          *
 883          * 01000000400000006c786bbac58ede11b0bb00261870892ab6adb76fe63a3
 884          * e468fca86530d2effa16c786bbac58ede11b0bb00261870892a0000000000
 885          * 0000000000000000000000
 886          *
 887          * Here's one interpretation of how the data is laid out.
 888          *
 889          * struct unknown {
 890          *      u32 field1; (always 0x00000001)
 891          *      u32 field2; (always 0x40000000)
 892          *      u16 field3;
 893          *      u32 field4;
 894          *      u32 field5;
 895          *      u32 field6;
 896          *      u8  data[48]; (???)
 897          *      u64 reserved1; (always 0)
 898          *      u64 reserved2; (always 0)
 899          * };*/
 900 #if 0
 901         if (dentry->length - calculated_size >= WIM_ADS_ENTRY_DISK_SIZE) {
 902                 printf("%s: %lu / %lu (", file_name_utf8,
 903                                 calculated_size, dentry->length);
 904                 print_string(p + WIM_ADS_ENTRY_DISK_SIZE, dentry->length - calculated_size - WIM_ADS_ENTRY_DISK_SIZE);
 905                 puts(")");
 906                 print_byte_field(p, dentry->length - calculated_size);
 907                 putchar('\n');
 908         }
 909 #endif
 910
 911         if (dentry->num_ads != 0) {
 912                 calculated_size = (calculated_size + 7) & ~7;
 913                 if (calculated_size > metadata_resource_len - offset) {
 914                         ERROR("Not enough space in metadata resource for "
 915                               "alternate stream entries");
 916                         ret = WIMLIB_ERR_INVALID_DENTRY;
 917                         goto out_free_short_name;
 918                 }
 919                 ret = read_ads_entries(&metadata_resource[offset + calculated_size],
 920                                        dentry,
 921                                        metadata_resource_len - offset - calculated_size);
 922                 if (ret != 0)
 923                         goto out_free_short_name;
 924         }
 925
 926         dentry->file_name          = file_name;
 927         dentry->file_name_utf8     = file_name_utf8;
 928         dentry->short_name         = short_name;
 929         dentry->file_name_len      = file_name_len;
 930         dentry->file_name_utf8_len = file_name_utf8_len;
 931         dentry->short_name_len     = short_name_len;
 932         return 0;
 933 out_free_short_name:
 934         FREE(short_name);
 935 out_free_file_name_utf8:
 936         FREE(file_name_utf8);
 937 out_free_file_name:
 938         FREE(file_name);
 939         return ret;
 940 }
 941
 942 /*
 943  * Writes a dentry to an output buffer.
 944  *
 945  * @dentry:  The dentry structure.
 946  * @p:       The memory location to write the data to.
 947  * @return:  Pointer to the byte after the last byte we wrote as part of the
 948  *              dentry.
 949  */
 950 static u8 *write_dentry(const struct dentry *dentry, u8 *p)
 951 {
 952         u8 *orig_p = p;
 953         unsigned padding;
 954         memset(p, 0, dentry->length);
 955         p = put_u64(p, dentry->length);
 956         p = put_u32(p, dentry->attributes);
 957         p = put_u32(p, dentry->security_id);
 958         p = put_u64(p, dentry->subdir_offset);
 959         p = put_u64(p, 0); /* unused1 */
 960         p = put_u64(p, 0); /* unused2 */
 961         p = put_u64(p, dentry->creation_time);
 962         p = put_u64(p, dentry->last_access_time);
 963         p = put_u64(p, dentry->last_write_time);
 964         memcpy(p, dentry->hash, WIM_HASH_SIZE);
 965         p += WIM_HASH_SIZE;
 966         p = put_u32(p, dentry->reparse_tag);
 967         p = put_u64(p, dentry->hard_link);
 968         p = put_u16(p, dentry->num_ads); /*streams */
 969         p = put_u16(p, dentry->short_name_len);
 970         p = put_u16(p, dentry->file_name_len);
 971         p = put_bytes(p, dentry->file_name_len, (u8*)dentry->file_name);
 972         p = put_u16(p, 0); /* filename padding, 2 bytes. */
 973         p = put_bytes(p, dentry->short_name_len, (u8*)dentry->short_name);
 974
 975         wimlib_assert(p - orig_p <= dentry->length);
 976         if (p - orig_p < dentry->length)
 977                 p = put_zeroes(p, dentry->length - (p - orig_p));
 978
 979         p = put_zeroes(p, (8 - (p - orig_p) % 8) % 8);
 980
 981         for (u16 i = 0; i < dentry->num_ads; i++) {
 982                 p = put_u64(p, ads_entry_length(&dentry->ads_entries[i]));
 983                 p = put_u64(p, 0); /* Unused */
 984                 p = put_bytes(p, WIM_HASH_SIZE, dentry->ads_entries[i].hash);
 985                 p = put_u16(p, dentry->ads_entries[i].stream_name_len);
 986                 p = put_bytes(p, dentry->ads_entries[i].stream_name_len,
 987                                  (u8*)dentry->ads_entries[i].stream_name);
 988                 p = put_zeroes(p, (8 - (p - orig_p) % 8) % 8);
 989         }
 990         return p;
 991 }
 992
 993 /* Recursive function that writes a dentry tree rooted at @tree, not including
 994  * @tree itself, which has already been written, except in the case of the root
 995  * dentry, which is written right away, along with an end-of-directory entry. */
 996 u8 *write_dentry_tree(const struct dentry *tree, u8 *p)
 997 {
 998         const struct dentry *child;
 999
1000         if (dentry_is_root(tree)) {
1001                 p = write_dentry(tree, p);
1002
1003                 /* write end of directory entry */
1004                 p = put_u64(p, 0);
1005         } else {
1006                 /* Nothing to do for a regular file. */
1007                 if (dentry_is_regular_file(tree))
1008                         return p;
1009         }
1010
1011         /* Write child dentries and end-of-directory entry. */
1012         child = tree->children;
1013         if (child) {
1014                 do {
1015                         p = write_dentry(child, p);
1016                         child = child->next;
1017                 } while (child != tree->children);
1018         }
1019
1020         /* write end of directory entry */
1021         p = put_u64(p, 0);
1022
1023         /* Recurse on children. */
1024         if (child) {
1025                 do {
1026                         p = write_dentry_tree(child, p);
1027                         child = child->next;
1028                 } while (child != tree->children);
1029         }
1030         return p;
1031 }
1032
1033 /* Reads the children of a dentry, and all their children, ..., etc. from the
1034  * metadata resource and into the dentry tree.
1035  *
1036  * @metadata_resource:  An array that contains the uncompressed metadata
1037  *                      resource for the WIM file.
1038  * @metadata_resource_len:      The length of @metadata_resource.
1039  * @dentry:     A pointer to a struct dentry that is the root of the directory
1040  *              tree and has already been read from the metadata resource.  It
1041  *              does not need to be the real root because this procedure is
1042  *              called recursively.
1043  *
1044  * @return:     Zero on success, nonzero on failure.
1045  */
1046 int read_dentry_tree(const u8 metadata_resource[], u64 metadata_resource_len,
1047                      struct dentry *dentry)
1048 {
1049         u64 cur_offset = dentry->subdir_offset;
1050         struct dentry *prev_child = NULL;
1051         struct dentry *first_child = NULL;
1052         struct dentry *child;
1053         struct dentry cur_child;
1054         int ret;
1055
1056         /* If @dentry is a regular file, nothing more needs to be done for this
1057          * branch. */
1058         if (cur_offset == 0)
1059                 return 0;
1060
1061         /* Find and read all the children of @dentry. */
1062         while (1) {
1063
1064                 /* Read next child of @dentry into @cur_child. */
1065                 ret = read_dentry(metadata_resource, metadata_resource_len,
1066                                   cur_offset, &cur_child);
1067                 if (ret != 0)
1068                         break;
1069
1070                 /* Check for end of directory. */
1071                 if (cur_child.length == 0) {
1072                         ret = 0;
1073                         break;
1074                 }
1075
1076                 /* Not end of directory.  Allocate this child permanently and
1077                  * link it to the parent and previous child. */
1078                 child = MALLOC(sizeof(struct dentry));
1079                 if (!child) {
1080                         ERROR("Failed to allocate %zu bytes for new dentry",
1081                               sizeof(struct dentry));
1082                         ret = WIMLIB_ERR_NOMEM;
1083                         break;
1084                 }
1085                 memcpy(child, &cur_child, sizeof(struct dentry));
1086
1087                 if (prev_child) {
1088                         prev_child->next = child;
1089                         child->prev = prev_child;
1090                 } else {
1091                         first_child = child;
1092                 }
1093
1094                 child->parent = dentry;
1095                 prev_child = child;
1096
1097                 /* If there are children of this child, call this procedure
1098                  * recursively. */
1099                 if (child->subdir_offset != 0) {
1100                         ret = read_dentry_tree(metadata_resource,
1101                                                metadata_resource_len, child);
1102                         if (ret != 0)
1103                                 break;
1104                 }
1105
1106                 /* Advance to the offset of the next child. */
1107                 cur_offset += dentry_total_length(child);
1108         }
1109
1110         /* Link last child to first one, and set parent's
1111          * children pointer to the first child.  */
1112         if (prev_child) {
1113                 prev_child->next = first_child;
1114                 first_child->prev = prev_child;
1115         }
1116         dentry->children = first_child;
1117         return ret;
1118 }
1119
1120 int dentry_set_symlink_buf(struct dentry *dentry, const u8 symlink_buf_hash[])
1121 {
1122         struct ads_entry *ads_entries;
1123
1124         ads_entries = CALLOC(2, sizeof(struct ads_entry));
1125         if (!ads_entries)
1126                 return WIMLIB_ERR_NOMEM;
1127         memcpy(ads_entries[1].hash, symlink_buf_hash, WIM_HASH_SIZE);
1128         dentry_free_ads_entries(dentry);
1129         dentry->num_ads = 2;
1130         dentry->ads_entries = ads_entries;
1131         return 0;
1132 }