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 /*
  43  * Returns true if @dentry has the UTF-8 file name @name that has length
  44  * @name_len.
  45  */
  46 static bool dentry_has_name(const struct dentry *dentry, const char *name,
  47                             size_t name_len)
  48 {
  49         if (dentry->file_name_utf8_len != name_len)
  50                 return false;
  51         return memcmp(dentry->file_name_utf8, name, name_len) == 0;
  52 }
  53
  54 /* Real length of a dentry, including the alternate data stream entries, which
  55  * are not included in the dentry->length field... */
  56 u64 dentry_total_length(const struct dentry *dentry)
  57 {
  58         u64 length = (dentry->length + 7) & ~7;
  59         for (u16 i = 0 ; i < dentry->num_ads; i++)
  60                 length += ads_entry_length(&dentry->ads_entries[i]);
  61         return length;
  62 }
  63
  64 /* Transfers file attributes from a `stat' buffer to a struct dentry. */
  65 void stbuf_to_dentry(const struct stat *stbuf, struct dentry *dentry)
  66 {
  67         if (S_ISLNK(stbuf->st_mode)) {
  68                 dentry->attributes = FILE_ATTRIBUTE_REPARSE_POINT;
  69                 dentry->reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
  70         } else if (S_ISDIR(stbuf->st_mode)) {
  71                 dentry->attributes = FILE_ATTRIBUTE_DIRECTORY;
  72         } else {
  73                 dentry->attributes = FILE_ATTRIBUTE_NORMAL;
  74         }
  75         if (sizeof(ino_t) >= 8)
  76                 dentry->hard_link = (u64)stbuf->st_ino;
  77         else
  78                 dentry->hard_link = (u64)stbuf->st_ino |
  79                                    ((u64)stbuf->st_dev << (sizeof(ino_t) * 8));
  80 }
  81
  82 /* Transfers file attributes from a struct dentry to a `stat' buffer. */
  83 void dentry_to_stbuf(const struct dentry *dentry, struct stat *stbuf,
  84                      const struct lookup_table *table)
  85 {
  86         struct lookup_table_entry *lte;
  87
  88         if (dentry_is_symlink(dentry))
  89                 stbuf->st_mode = S_IFLNK | 0777;
  90         else if (dentry_is_directory(dentry))
  91                 stbuf->st_mode = S_IFDIR | 0755;
  92         else
  93                 stbuf->st_mode = S_IFREG | 0644;
  94
  95         /* Use the size of the unnamed (default) file stream. */
  96         if (table && (lte = __lookup_resource(table, dentry_hash(dentry))))
  97                 stbuf->st_size = lte->resource_entry.original_size;
  98         else
  99                 stbuf->st_size = 0;
 100
 101         stbuf->st_nlink   = dentry_link_group_size(dentry);
 102         stbuf->st_uid     = getuid();
 103         stbuf->st_gid     = getgid();
 104         stbuf->st_atime   = ms_timestamp_to_unix(dentry->last_access_time);
 105         stbuf->st_mtime   = ms_timestamp_to_unix(dentry->last_write_time);
 106         stbuf->st_ctime   = ms_timestamp_to_unix(dentry->creation_time);
 107         stbuf->st_blocks  = (stbuf->st_size + 511) / 512;
 108 }
 109
 110 /* Makes all timestamp fields for the dentry be the current time. */
 111 void dentry_update_all_timestamps(struct dentry *dentry)
 112 {
 113         u64 now = get_timestamp();
 114         dentry->creation_time    = now;
 115         dentry->last_access_time = now;
 116         dentry->last_write_time  = now;
 117 }
 118
 119 struct ads_entry *dentry_get_ads_entry(struct dentry *dentry,
 120                                        const char *stream_name)
 121 {
 122         size_t stream_name_len = strlen(stream_name);
 123         if (!stream_name)
 124                 return NULL;
 125         for (u16 i = 0; i < dentry->num_ads; i++)
 126                 if (ads_entry_has_name(&dentry->ads_entries[i],
 127                                        stream_name, stream_name_len))
 128                         return &dentry->ads_entries[i];
 129         return NULL;
 130 }
 131
 132 /* Add an alternate stream entry to a dentry and return a pointer to it, or NULL
 133  * on failure. */
 134 struct ads_entry *dentry_add_ads(struct dentry *dentry, const char *stream_name)
 135 {
 136         u16 num_ads = dentry->num_ads + 1;
 137         struct ads_entry *ads_entries;
 138         struct ads_entry *new_entry;
 139         if (num_ads == 0xffff)
 140                 return NULL;
 141         ads_entries = MALLOC(num_ads * sizeof(struct ads_entry));
 142         if (!ads_entries)
 143                 return NULL;
 144
 145         new_entry = &ads_entries[num_ads - 1];
 146         if (change_ads_name(new_entry, stream_name) != 0) {
 147                 FREE(ads_entries);
 148                 return NULL;
 149         }
 150
 151         memcpy(ads_entries, dentry->ads_entries,
 152                (num_ads - 1) * sizeof(struct ads_entry));
 153         FREE(dentry->ads_entries);
 154         dentry->ads_entries = ads_entries;
 155         dentry->num_ads = num_ads;
 156         return memset(new_entry, 0, sizeof(struct ads_entry));
 157 }
 158
 159 void dentry_remove_ads(struct dentry *dentry, struct ads_entry *sentry)
 160 {
 161         destroy_ads_entry(sentry);
 162         memcpy(sentry, sentry + 1,
 163                (dentry->num_ads - (sentry - dentry->ads_entries))
 164                  * sizeof(struct ads_entry));
 165         dentry->num_ads--;
 166 }
 167
 168 /*
 169  * Calls a function on all directory entries in a directory tree.  It is called
 170  * on a parent before its children.
 171  */
 172 int for_dentry_in_tree(struct dentry *root,
 173                        int (*visitor)(struct dentry*, void*), void *arg)
 174 {
 175         int ret;
 176         struct dentry *child;
 177
 178         ret = visitor(root, arg);
 179
 180         if (ret != 0)
 181                 return ret;
 182
 183         child = root->children;
 184
 185         if (!child)
 186                 return 0;
 187
 188         do {
 189                 ret = for_dentry_in_tree(child, visitor, arg);
 190                 if (ret != 0)
 191                         return ret;
 192                 child = child->next;
 193         } while (child != root->children);
 194         return 0;
 195 }
 196
 197 /*
 198  * Like for_dentry_in_tree(), but the visitor function is always called on a
 199  * dentry's children before on itself.
 200  */
 201 int for_dentry_in_tree_depth(struct dentry *root,
 202                              int (*visitor)(struct dentry*, void*), void *arg)
 203 {
 204         int ret;
 205         struct dentry *child;
 206         struct dentry *next;
 207
 208         child = root->children;
 209         if (child) {
 210                 do {
 211                         next = child->next;
 212                         ret = for_dentry_in_tree_depth(child, visitor, arg);
 213                         if (ret != 0)
 214                                 return ret;
 215                         child = next;
 216                 } while (child != root->children);
 217         }
 218         return visitor(root, arg);
 219 }
 220
 221 /*
 222  * Calculate the full path of @dentry, based on its parent's full path and on
 223  * its UTF-8 file name.
 224  */
 225 int calculate_dentry_full_path(struct dentry *dentry, void *ignore)
 226 {
 227         char *full_path;
 228         u32 full_path_len;
 229         if (dentry_is_root(dentry)) {
 230                 full_path = MALLOC(2);
 231                 if (!full_path)
 232                         goto oom;
 233                 full_path[0] = '/';
 234                 full_path[1] = '\0';
 235                 full_path_len = 1;
 236         } else {
 237                 char *parent_full_path;
 238                 u32 parent_full_path_len;
 239                 const struct dentry *parent = dentry->parent;
 240
 241                 if (dentry_is_root(parent)) {
 242                         parent_full_path = "";
 243                         parent_full_path_len = 0;
 244                 } else {
 245                         parent_full_path = parent->full_path_utf8;
 246                         parent_full_path_len = parent->full_path_utf8_len;
 247                 }
 248
 249                 full_path_len = parent_full_path_len + 1 +
 250                                 dentry->file_name_utf8_len;
 251                 full_path = MALLOC(full_path_len + 1);
 252                 if (!full_path)
 253                         goto oom;
 254
 255                 memcpy(full_path, parent_full_path, parent_full_path_len);
 256                 full_path[parent_full_path_len] = '/';
 257                 memcpy(full_path + parent_full_path_len + 1,
 258                        dentry->file_name_utf8,
 259                        dentry->file_name_utf8_len);
 260                 full_path[full_path_len] = '\0';
 261         }
 262         FREE(dentry->full_path_utf8);
 263         dentry->full_path_utf8 = full_path;
 264         dentry->full_path_utf8_len = full_path_len;
 265         return 0;
 266 oom:
 267         ERROR("Out of memory while calculating dentry full path");
 268         return WIMLIB_ERR_NOMEM;
 269 }
 270
 271 /*
 272  * Recursively calculates the subdir offsets for a directory tree.
 273  *
 274  * @dentry:  The root of the directory tree.
 275  * @subdir_offset_p:  The current subdirectory offset; i.e., the subdirectory
 276  *      offset for @dentry.
 277  */
 278 void calculate_subdir_offsets(struct dentry *dentry, u64 *subdir_offset_p)
 279 {
 280         struct dentry *child;
 281
 282         child = dentry->children;
 283         dentry->subdir_offset = *subdir_offset_p;
 284         if (child) {
 285
 286                 /* Advance the subdir offset by the amount of space the children
 287                  * of this dentry take up. */
 288                 do {
 289                         *subdir_offset_p += dentry_total_length(child);
 290                         child = child->next;
 291                 } while (child != dentry->children);
 292
 293                 /* End-of-directory dentry on disk. */
 294                 *subdir_offset_p += 8;
 295
 296                 /* Recursively call calculate_subdir_offsets() on all the
 297                  * children. */
 298                 do {
 299                         calculate_subdir_offsets(child, subdir_offset_p);
 300                         child = child->next;
 301                 } while (child != dentry->children);
 302         } else {
 303                 /* On disk, childless directories have a valid subdir_offset
 304                  * that points to an 8-byte end-of-directory dentry.  Regular
 305                  * files have a subdir_offset of 0. */
 306                 if (dentry_is_directory(dentry))
 307                         *subdir_offset_p += 8;
 308                 else
 309                         dentry->subdir_offset = 0;
 310         }
 311 }
 312
 313
 314 /* Returns the child of @dentry that has the file name @name.
 315  * Returns NULL if no child has the name. */
 316 struct dentry *get_dentry_child_with_name(const struct dentry *dentry,
 317                                                         const char *name)
 318 {
 319         struct dentry *child;
 320         size_t name_len;
 321
 322         child = dentry->children;
 323         if (child) {
 324                 name_len = strlen(name);
 325                 do {
 326                         if (dentry_has_name(child, name, name_len))
 327                                 return child;
 328                         child = child->next;
 329                 } while (child != dentry->children);
 330         }
 331         return NULL;
 332 }
 333
 334 /* Retrieves the dentry that has the UTF-8 @path relative to the dentry
 335  * @cur_dir.  Returns NULL if no dentry having the path is found. */
 336 static struct dentry *get_dentry_relative_path(struct dentry *cur_dir, const char *path)
 337 {
 338         struct dentry *child;
 339         size_t base_len;
 340         const char *new_path;
 341
 342         if (*path == '\0')
 343                 return cur_dir;
 344
 345         child = cur_dir->children;
 346         if (child) {
 347                 new_path = path_next_part(path, &base_len);
 348                 do {
 349                         if (dentry_has_name(child, path, base_len))
 350                                 return get_dentry_relative_path(child, new_path);
 351                         child = child->next;
 352                 } while (child != cur_dir->children);
 353         }
 354         return NULL;
 355 }
 356
 357 /* Returns the dentry corresponding to the UTF-8 @path, or NULL if there is no
 358  * such dentry. */
 359 struct dentry *get_dentry(WIMStruct *w, const char *path)
 360 {
 361         struct dentry *root = wim_root_dentry(w);
 362         while (*path == '/')
 363                 path++;
 364         return get_dentry_relative_path(root, path);
 365 }
 366
 367 /* Returns the parent directory for the @path. */
 368 struct dentry *get_parent_dentry(WIMStruct *w, const char *path)
 369 {
 370         size_t path_len = strlen(path);
 371         char buf[path_len + 1];
 372
 373         memcpy(buf, path, path_len + 1);
 374
 375         to_parent_name(buf, path_len);
 376
 377         return get_dentry(w, buf);
 378 }
 379
 380 /* Prints the full path of a dentry. */
 381 int print_dentry_full_path(struct dentry *dentry, void *ignore)
 382 {
 383         if (dentry->full_path_utf8)
 384                 puts(dentry->full_path_utf8);
 385         return 0;
 386 }
 387
 388 struct file_attr_flag {
 389         u32 flag;
 390         const char *name;
 391 };
 392 struct file_attr_flag file_attr_flags[] = {
 393         {FILE_ATTRIBUTE_READONLY,               "READONLY"},
 394         {FILE_ATTRIBUTE_HIDDEN,         "HIDDEN"},
 395         {FILE_ATTRIBUTE_SYSTEM,         "SYSTEM"},
 396         {FILE_ATTRIBUTE_DIRECTORY,              "DIRECTORY"},
 397         {FILE_ATTRIBUTE_ARCHIVE,                "ARCHIVE"},
 398         {FILE_ATTRIBUTE_DEVICE,         "DEVICE"},
 399         {FILE_ATTRIBUTE_NORMAL,         "NORMAL"},
 400         {FILE_ATTRIBUTE_TEMPORARY,              "TEMPORARY"},
 401         {FILE_ATTRIBUTE_SPARSE_FILE,    "SPARSE_FILE"},
 402         {FILE_ATTRIBUTE_REPARSE_POINT,  "REPARSE_POINT"},
 403         {FILE_ATTRIBUTE_COMPRESSED,             "COMPRESSED"},
 404         {FILE_ATTRIBUTE_OFFLINE,                "OFFLINE"},
 405         {FILE_ATTRIBUTE_NOT_CONTENT_INDEXED,"NOT_CONTENT_INDEXED"},
 406         {FILE_ATTRIBUTE_ENCRYPTED,              "ENCRYPTED"},
 407         {FILE_ATTRIBUTE_VIRTUAL,                "VIRTUAL"},
 408 };
 409
 410 /* Prints a directory entry.  @lookup_table is a pointer to the lookup table, or
 411  * NULL if the resource entry for the dentry is not to be printed. */
 412 int print_dentry(struct dentry *dentry, void *lookup_table)
 413 {
 414         struct lookup_table_entry *lte;
 415         unsigned i;
 416
 417         printf("[DENTRY]\n");
 418         printf("Length            = %"PRIu64"\n", dentry->length);
 419         printf("Attributes        = 0x%x\n", dentry->attributes);
 420         for (i = 0; i < ARRAY_LEN(file_attr_flags); i++)
 421                 if (file_attr_flags[i].flag & dentry->attributes)
 422                         printf("    FILE_ATTRIBUTE_%s is set\n",
 423                                 file_attr_flags[i].name);
 424         printf("Security ID       = %d\n", dentry->security_id);
 425         printf("Subdir offset     = %"PRIu64"\n", dentry->subdir_offset);
 426         /*printf("Unused1           = 0x%"PRIu64"\n", dentry->unused1);*/
 427         /*printf("Unused2           = %"PRIu64"\n", dentry->unused2);*/
 428         printf("Creation Time     = 0x%"PRIx64"\n", dentry->creation_time);
 429         printf("Last Access Time  = 0x%"PRIx64"\n", dentry->last_access_time);
 430         printf("Last Write Time   = 0x%"PRIx64"\n", dentry->last_write_time);
 431         printf("Hash              = 0x");
 432         print_hash(dentry->hash);
 433         putchar('\n');
 434         printf("Reparse Tag       = 0x%"PRIx32"\n", dentry->reparse_tag);
 435         printf("Hard Link Group   = 0x%"PRIx64"\n", dentry->hard_link);
 436         printf("Number of Alternate Data Streams = %hu\n", dentry->num_ads);
 437         printf("Filename          = \"");
 438         print_string(dentry->file_name, dentry->file_name_len);
 439         puts("\"");
 440         printf("Filename Length   = %hu\n", dentry->file_name_len);
 441         printf("Filename (UTF-8)  = \"%s\"\n", dentry->file_name_utf8);
 442         printf("Filename (UTF-8) Length = %hu\n", dentry->file_name_utf8_len);
 443         printf("Short Name        = \"");
 444         print_string(dentry->short_name, dentry->short_name_len);
 445         puts("\"");
 446         printf("Short Name Length = %hu\n", dentry->short_name_len);
 447         printf("Full Path (UTF-8) = \"%s\"\n", dentry->full_path_utf8);
 448         if (lookup_table && (lte = __lookup_resource(lookup_table, dentry->hash)))
 449                 print_lookup_table_entry(lte, NULL);
 450         else
 451                 putchar('\n');
 452         for (u16 i = 0; i < dentry->num_ads; i++) {
 453                 printf("[Alternate Stream Entry %u]\n", i);
 454                 printf("Name = \"%s\"\n", dentry->ads_entries[i].stream_name_utf8);
 455                 printf("Name Length (UTF-16) = %u\n",
 456                                 dentry->ads_entries[i].stream_name_len);
 457                 printf("Hash              = 0x");
 458                 print_hash(dentry->ads_entries[i].hash);
 459                 if (lookup_table &&
 460                      (lte = __lookup_resource(lookup_table,
 461                                               dentry->ads_entries[i].hash)))
 462                 {
 463                         print_lookup_table_entry(lte, NULL);
 464                 } else {
 465                         putchar('\n');
 466                 }
 467         }
 468         return 0;
 469 }
 470
 471 static inline void dentry_common_init(struct dentry *dentry)
 472 {
 473         memset(dentry, 0, sizeof(struct dentry));
 474         dentry->refcnt = 1;
 475         dentry->security_id = -1;
 476 }
 477
 478 /*
 479  * Creates an unlinked directory entry.
 480  *
 481  * @name:    The base name of the new dentry.
 482  * @return:  A pointer to the new dentry, or NULL if out of memory.
 483  */
 484 struct dentry *new_dentry(const char *name)
 485 {
 486         struct dentry *dentry;
 487
 488         dentry = MALLOC(sizeof(struct dentry));
 489         if (!dentry)
 490                 goto err;
 491
 492         dentry_common_init(dentry);
 493         if (change_dentry_name(dentry, name) != 0)
 494                 goto err;
 495
 496         dentry_update_all_timestamps(dentry);
 497         dentry->next   = dentry;
 498         dentry->prev   = dentry;
 499         dentry->parent = dentry;
 500         return dentry;
 501 err:
 502         FREE(dentry);
 503         ERROR("Failed to allocate new dentry");
 504         return NULL;
 505 }
 506
 507 static void dentry_free_ads_entries(struct dentry *dentry)
 508 {
 509         for (u16 i = 0; i < dentry->num_ads; i++)
 510                 destroy_ads_entry(&dentry->ads_entries[i]);
 511         FREE(dentry->ads_entries);
 512         dentry->ads_entries = NULL;
 513         dentry->num_ads = 0;
 514 }
 515
 516
 517 void free_dentry(struct dentry *dentry)
 518 {
 519         FREE(dentry->file_name);
 520         FREE(dentry->file_name_utf8);
 521         FREE(dentry->short_name);
 522         FREE(dentry->full_path_utf8);
 523         dentry_free_ads_entries(dentry);
 524         FREE(dentry);
 525 }
 526
 527 /* Arguments for do_free_dentry(). */
 528 struct free_dentry_args {
 529         struct lookup_table *lookup_table;
 530         bool lt_decrement_refcnt;
 531 };
 532
 533 /*
 534  * This function is passed as an argument to for_dentry_in_tree_depth() in order
 535  * to free a directory tree.  __args is a pointer to a `struct free_dentry_args'.
 536  */
 537 static int do_free_dentry(struct dentry *dentry, void *__args)
 538 {
 539         struct free_dentry_args *args = (struct free_dentry_args*)__args;
 540
 541         if (args->lt_decrement_refcnt && !dentry_is_directory(dentry)) {
 542                 lookup_table_decrement_refcnt(args->lookup_table,
 543                                               dentry->hash);
 544         }
 545
 546         wimlib_assert(dentry->refcnt >= 1);
 547         if (--dentry->refcnt == 0)
 548                 free_dentry(dentry);
 549         return 0;
 550 }
 551
 552 /*
 553  * Unlinks and frees a dentry tree.
 554  *
 555  * @root:               The root of the tree.
 556  * @lookup_table:       The lookup table for dentries.
 557  * @decrement_refcnt:   True if the dentries in the tree are to have their
 558  *                      reference counts in the lookup table decremented.
 559  */
 560 void free_dentry_tree(struct dentry *root, struct lookup_table *lookup_table,
 561                       bool lt_decrement_refcnt)
 562 {
 563         if (!root || !root->parent)
 564                 return;
 565
 566         struct free_dentry_args args;
 567         args.lookup_table        = lookup_table;
 568         args.lt_decrement_refcnt = lt_decrement_refcnt;
 569         for_dentry_in_tree_depth(root, do_free_dentry, &args);
 570 }
 571
 572 int increment_dentry_refcnt(struct dentry *dentry, void *ignore)
 573 {
 574         dentry->refcnt++;
 575         return 0;
 576 }
 577
 578 /*
 579  * Links a dentry into the directory tree.
 580  *
 581  * @dentry: The dentry to link.
 582  * @parent: The dentry that will be the parent of @dentry.
 583  */
 584 void link_dentry(struct dentry *dentry, struct dentry *parent)
 585 {
 586         dentry->parent = parent;
 587         if (parent->children) {
 588                 /* Not an only child; link to siblings. */
 589                 dentry->next = parent->children;
 590                 dentry->prev = parent->children->prev;
 591                 dentry->next->prev = dentry;
 592                 dentry->prev->next = dentry;
 593         } else {
 594                 /* Only child; link to parent. */
 595                 parent->children = dentry;
 596                 dentry->next = dentry;
 597                 dentry->prev = dentry;
 598         }
 599 }
 600
 601 /* Unlink a dentry from the directory tree. */
 602 void unlink_dentry(struct dentry *dentry)
 603 {
 604         if (dentry_is_root(dentry))
 605                 return;
 606         if (dentry_is_only_child(dentry)) {
 607                 dentry->parent->children = NULL;
 608         } else {
 609                 if (dentry_is_first_sibling(dentry))
 610                         dentry->parent->children = dentry->next;
 611                 dentry->next->prev = dentry->prev;
 612                 dentry->prev->next = dentry->next;
 613         }
 614 }
 615
 616
 617 /* Recalculates the length of @dentry based on its file name length and short
 618  * name length.  */
 619 static inline void recalculate_dentry_size(struct dentry *dentry)
 620 {
 621         dentry->length = WIM_DENTRY_DISK_SIZE + dentry->file_name_len +
 622                          2 + dentry->short_name_len;
 623         /* Must be multiple of 8. */
 624         dentry->length = (dentry->length + 7) & ~7;
 625 }
 626
 627 static int do_name_change(char **file_name_ret,
 628                           char **file_name_utf8_ret,
 629                           u16 *file_name_len_ret,
 630                           u16 *file_name_utf8_len_ret,
 631                           const char *new_name)
 632 {
 633         size_t utf8_len;
 634         size_t utf16_len;
 635         char *file_name, *file_name_utf8;
 636
 637         utf8_len = strlen(new_name);
 638
 639         file_name = utf8_to_utf16(new_name, utf8_len, &utf16_len);
 640
 641         if (!file_name)
 642                 return WIMLIB_ERR_NOMEM;
 643
 644         file_name_utf8 = MALLOC(utf8_len + 1);
 645         if (!file_name_utf8) {
 646                 FREE(file_name);
 647                 return WIMLIB_ERR_NOMEM;
 648         }
 649         memcpy(file_name_utf8, new_name, utf8_len + 1);
 650
 651         FREE(*file_name_ret);
 652         FREE(*file_name_utf8_ret);
 653         *file_name_ret          = file_name;
 654         *file_name_utf8_ret     = file_name_utf8;
 655         *file_name_len_ret      = utf16_len;
 656         *file_name_utf8_len_ret = utf8_len;
 657         return 0;
 658 }
 659
 660 /* Changes the name of a dentry to @new_name.  Only changes the file_name and
 661  * file_name_utf8 fields; does not change the short_name, short_name_utf8, or
 662  * full_path_utf8 fields.  Also recalculates its length. */
 663 int change_dentry_name(struct dentry *dentry, const char *new_name)
 664 {
 665         int ret;
 666
 667         ret = do_name_change(&dentry->file_name, &dentry->file_name_utf8,
 668                              &dentry->file_name_len, &dentry->file_name_utf8_len,
 669                              new_name);
 670         if (ret == 0)
 671                 recalculate_dentry_size(dentry);
 672         return ret;
 673 }
 674
 675 int change_ads_name(struct ads_entry *entry, const char *new_name)
 676 {
 677         return do_name_change(&entry->stream_name, &entry->stream_name_utf8,
 678                               &entry->stream_name_len,
 679                               &entry->stream_name_utf8_len,
 680                               new_name);
 681 }
 682
 683 /* Parameters for calculate_dentry_statistics(). */
 684 struct image_statistics {
 685         struct lookup_table *lookup_table;
 686         u64 *dir_count;
 687         u64 *file_count;
 688         u64 *total_bytes;
 689         u64 *hard_link_bytes;
 690 };
 691
 692 static int calculate_dentry_statistics(struct dentry *dentry, void *arg)
 693 {
 694         struct image_statistics *stats;
 695         struct lookup_table_entry *lte;
 696         u16 i;
 697
 698         stats = arg;
 699
 700         if (dentry_is_directory(dentry) && !dentry_is_root(dentry))
 701                 ++*stats->dir_count;
 702         else
 703                 ++*stats->file_count;
 704
 705         lte = __lookup_resource(stats->lookup_table, dentry->hash);
 706         i = 0;
 707         while (1) {
 708                 if (lte) {
 709                         u64 size = lte->resource_entry.original_size;
 710                         *stats->total_bytes += size;
 711                         if (++lte->out_refcnt == 1)
 712                                 *stats->hard_link_bytes += size;
 713                 }
 714                 if (i == dentry->num_ads)
 715                         break;
 716                 lte = __lookup_resource(stats->lookup_table,
 717                                         dentry->ads_entries[i].hash);
 718                 i++;
 719         }
 720
 721         return 0;
 722 }
 723
 724 void calculate_dir_tree_statistics(struct dentry *root, struct lookup_table *table,
 725                                    u64 *dir_count_ret, u64 *file_count_ret,
 726                                    u64 *total_bytes_ret,
 727                                    u64 *hard_link_bytes_ret)
 728 {
 729         struct image_statistics stats;
 730         *dir_count_ret         = 0;
 731         *file_count_ret        = 0;
 732         *total_bytes_ret       = 0;
 733         *hard_link_bytes_ret   = 0;
 734         stats.lookup_table     = table;
 735         stats.dir_count       = dir_count_ret;
 736         stats.file_count      = file_count_ret;
 737         stats.total_bytes     = total_bytes_ret;
 738         stats.hard_link_bytes = hard_link_bytes_ret;
 739         for_lookup_table_entry(table, zero_out_refcnts, NULL);
 740         for_dentry_in_tree(root, calculate_dentry_statistics, &stats);
 741 }
 742
 743 static int read_ads_entries(const u8 *p, struct dentry *dentry,
 744                             u64 remaining_size)
 745 {
 746         u16 num_ads = dentry->num_ads;
 747         struct ads_entry *ads_entries = CALLOC(num_ads, sizeof(struct ads_entry));
 748         int ret;
 749         if (!ads_entries) {
 750                 ERROR("Could not allocate memory for %"PRIu16" "
 751                       "alternate data stream entries", num_ads);
 752                 return WIMLIB_ERR_NOMEM;
 753         }
 754         DEBUG2("Reading %"PRIu16" alternate data streams "
 755                "(remaining size = %"PRIu64")", num_ads, remaining_size);
 756
 757         for (u16 i = 0; i < num_ads; i++) {
 758                 struct ads_entry *cur_entry = &ads_entries[i];
 759                 u64 length;
 760                 size_t utf8_len;
 761                 const char *p_save = p;
 762                 /* Read the base stream entry, excluding the stream name. */
 763                 if (remaining_size < WIM_ADS_ENTRY_DISK_SIZE) {
 764                         ERROR("Stream entries go past end of metadata resource");
 765                         ERROR("(remaining_size = %"PRIu64")", remaining_size);
 766                         ret = WIMLIB_ERR_INVALID_DENTRY;
 767                         goto out_free_ads_entries;
 768                 }
 769                 remaining_size -= WIM_ADS_ENTRY_DISK_SIZE;
 770
 771                 p = get_u64(p, &length); /* ADS entry length */
 772
 773                 DEBUG2("ADS length = %"PRIu64, length);
 774
 775                 p += 8; /* Unused */
 776                 p = get_bytes(p, WIM_HASH_SIZE, (u8*)cur_entry->hash);
 777                 p = get_u16(p, &cur_entry->stream_name_len);
 778
 779                 DEBUG2("Stream name length = %u", cur_entry->stream_name_len);
 780
 781                 cur_entry->stream_name = NULL;
 782                 cur_entry->stream_name_utf8 = NULL;
 783
 784                 if (remaining_size < cur_entry->stream_name_len + 2) {
 785                         ERROR("Stream entries go past end of metadata resource");
 786                         ERROR("(remaining_size = %"PRIu64" bytes, stream_name_len "
 787                               "= %"PRIu16" bytes", remaining_size,
 788                               cur_entry->stream_name_len);
 789                         ret = WIMLIB_ERR_INVALID_DENTRY;
 790                         goto out_free_ads_entries;
 791                 }
 792                 remaining_size -= cur_entry->stream_name_len + 2;
 793
 794                 cur_entry->stream_name = MALLOC(cur_entry->stream_name_len);
 795                 if (!cur_entry->stream_name) {
 796                         ret = WIMLIB_ERR_NOMEM;
 797                         goto out_free_ads_entries;
 798                 }
 799                 get_bytes(p, cur_entry->stream_name_len,
 800                           (u8*)cur_entry->stream_name);
 801                 cur_entry->stream_name_utf8 = utf16_to_utf8(cur_entry->stream_name,
 802                                                             cur_entry->stream_name_len,
 803                                                             &utf8_len);
 804                 cur_entry->stream_name_utf8_len = utf8_len;
 805
 806                 if (!cur_entry->stream_name_utf8) {
 807                         ret = WIMLIB_ERR_NOMEM;
 808                         goto out_free_ads_entries;
 809                 }
 810                 p = p_save + ads_entry_length(cur_entry);
 811         }
 812         dentry->ads_entries = ads_entries;
 813         return 0;
 814 out_free_ads_entries:
 815         for (u16 i = 0; i < num_ads; i++) {
 816                 FREE(ads_entries[i].stream_name);
 817                 FREE(ads_entries[i].stream_name_utf8);
 818         }
 819         FREE(ads_entries);
 820         return ret;
 821 }
 822
 823 /*
 824  * Reads a directory entry from the metadata resource.
 825  */
 826 int read_dentry(const u8 metadata_resource[], u64 metadata_resource_len,
 827                 u64 offset, struct dentry *dentry)
 828 {
 829         const u8 *p;
 830         u64 calculated_size;
 831         char *file_name;
 832         char *file_name_utf8;
 833         char *short_name;
 834         u16 short_name_len;
 835         u16 file_name_len;
 836         size_t file_name_utf8_len;
 837         int ret;
 838
 839         dentry_common_init(dentry);
 840
 841         /*Make sure the dentry really fits into the metadata resource.*/
 842         if (offset + 8 > metadata_resource_len) {
 843                 ERROR("Directory entry starting at %"PRIu64" ends past the "
 844                       "end of the metadata resource (size %"PRIu64")",
 845                       offset, metadata_resource_len);
 846                 return WIMLIB_ERR_INVALID_DENTRY;
 847         }
 848
 849         /* Before reading the whole entry, we need to read just the length.
 850          * This is because an entry of length 8 (that is, just the length field)
 851          * terminates the list of sibling directory entries. */
 852
 853         p = get_u64(&metadata_resource[offset], &dentry->length);
 854
 855         /* A zero length field (really a length of 8, since that's how big the
 856          * directory entry is...) indicates that this is the end of directory
 857          * dentry.  We do not read it into memory as an actual dentry, so just
 858          * return true in that case. */
 859         if (dentry->length == 0)
 860                 return 0;
 861
 862         if (offset + dentry->length >= metadata_resource_len) {
 863                 ERROR("Directory entry at offset %"PRIu64" and with size "
 864                       "%"PRIu64" ends past the end of the metadata resource "
 865                       "(size %"PRIu64")",
 866                       offset, dentry->length, metadata_resource_len);
 867                 return WIMLIB_ERR_INVALID_DENTRY;
 868         }
 869
 870         /* If it is a recognized length, read the rest of the directory entry.
 871          * Note: The root directory entry has no name, and its length does not
 872          * include the short name length field.  */
 873         if (dentry->length < WIM_DENTRY_DISK_SIZE) {
 874                 ERROR("Directory entry has invalid length of %"PRIu64" bytes",
 875                       dentry->length);
 876                 return WIMLIB_ERR_INVALID_DENTRY;
 877         }
 878
 879         p = get_u32(p, &dentry->attributes);
 880         p = get_u32(p, (u32*)&dentry->security_id);
 881         p = get_u64(p, &dentry->subdir_offset);
 882
 883         /* 2 unused fields */
 884         p += 2 * sizeof(u64);
 885         /*p = get_u64(p, &dentry->unused1);*/
 886         /*p = get_u64(p, &dentry->unused2);*/
 887
 888         p = get_u64(p, &dentry->creation_time);
 889         p = get_u64(p, &dentry->last_access_time);
 890         p = get_u64(p, &dentry->last_write_time);
 891
 892         p = get_bytes(p, WIM_HASH_SIZE, dentry->hash);
 893
 894         /*
 895          * I don't know what's going on here.  It seems like M$ screwed up the
 896          * reparse points, then put the fields in the same place and didn't
 897          * document it.  The WIM_HDR_FLAG_RP_FIX flag in the WIM header might
 898          * have something to do with this, but it's not documented.
 899          */
 900         if (dentry->attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
 901                 /* ??? */
 902                 u32 u1, u2;
 903                 p = get_u32(p, &u1);
 904                 /*p += 4;*/
 905                 p = get_u32(p, &dentry->reparse_tag);
 906                 p = get_u32(p, &u2);
 907                 /*p += 4;*/
 908                 dentry->hard_link = (u64)(u1) | ((u64)(u2) << 32);
 909         } else {
 910                 p = get_u32(p, &dentry->reparse_tag);
 911                 p = get_u64(p, &dentry->hard_link);
 912         }
 913
 914         /* By the way, the reparse_reserved field does not actually exist (at
 915          * least when the file is not a reparse point) */
 916
 917
 918         p = get_u16(p, &dentry->num_ads);
 919
 920         p = get_u16(p, &short_name_len);
 921         p = get_u16(p, &file_name_len);
 922
 923         calculated_size = WIM_DENTRY_DISK_SIZE + file_name_len + 2 +
 924                           short_name_len;
 925
 926         if (dentry->length < calculated_size) {
 927                 ERROR("Unexpected end of directory entry! (Expected "
 928                       "%"PRIu64" bytes, got %"PRIu64" bytes. "
 929                       "short_name_len = %hu, file_name_len = %hu)",
 930                       calculated_size, dentry->length,
 931                       short_name_len, file_name_len);
 932                 return WIMLIB_ERR_INVALID_DENTRY;
 933         }
 934
 935         /* Read the filename. */
 936         file_name = MALLOC(file_name_len);
 937         if (!file_name) {
 938                 ERROR("Failed to allocate %hu bytes for dentry file name",
 939                       file_name_len);
 940                 return WIMLIB_ERR_NOMEM;
 941         }
 942         p = get_bytes(p, file_name_len, file_name);
 943
 944         /* Convert filename to UTF-8. */
 945         file_name_utf8 = utf16_to_utf8(file_name, file_name_len,
 946                                        &file_name_utf8_len);
 947
 948         if (!file_name_utf8) {
 949                 ERROR("Failed to allocate memory to convert UTF-16 "
 950                       "filename (%hu bytes) to UTF-8", file_name_len);
 951                 ret = WIMLIB_ERR_NOMEM;
 952                 goto out_free_file_name;
 953         }
 954
 955         /* Undocumented padding between file name and short name.  This probably
 956          * is supposed to be a terminating null character. */
 957         p += 2;
 958
 959         /* Read the short filename. */
 960         short_name = MALLOC(short_name_len);
 961         if (!short_name) {
 962                 ERROR("Failed to allocate %hu bytes for short filename",
 963                       short_name_len);
 964                 ret = WIMLIB_ERR_NOMEM;
 965                 goto out_free_file_name_utf8;
 966         }
 967
 968         p = get_bytes(p, short_name_len, short_name);
 969
 970         /* Some directory entries inexplicibly have a little over 70 bytes of
 971          * extra data.  The exact amount of data seems to be 72 bytes, but it is
 972          * aligned on the next 8-byte boundary.  Here's an example of the
 973          * aligned data:
 974          *
 975          * 01000000400000006c786bbac58ede11b0bb00261870892ab6adb76fe63a3
 976          * e468fca86530d2effa16c786bbac58ede11b0bb00261870892a0000000000
 977          * 0000000000000000000000
 978          *
 979          * Here's one interpretation of how the data is laid out.
 980          *
 981          * struct unknown {
 982          *      u32 field1; (always 0x00000001)
 983          *      u32 field2; (always 0x40000000)
 984          *      u16 field3;
 985          *      u32 field4;
 986          *      u32 field5;
 987          *      u32 field6;
 988          *      u8  data[48]; (???)
 989          *      u64 reserved1; (always 0)
 990          *      u64 reserved2; (always 0)
 991          * };*/
 992 #if 0
 993         if (dentry->length - calculated_size >= WIM_ADS_ENTRY_DISK_SIZE) {
 994                 printf("%s: %lu / %lu (", file_name_utf8,
 995                                 calculated_size, dentry->length);
 996                 print_string(p + WIM_ADS_ENTRY_DISK_SIZE, dentry->length - calculated_size - WIM_ADS_ENTRY_DISK_SIZE);
 997                 puts(")");
 998                 print_byte_field(p, dentry->length - calculated_size);
 999                 putchar('\n');
1000         }
1001 #endif
1002
1003         if (dentry->num_ads != 0) {
1004                 calculated_size = (calculated_size + 7) & ~7;
1005                 if (calculated_size > metadata_resource_len - offset) {
1006                         ERROR("Not enough space in metadata resource for "
1007                               "alternate stream entries");
1008                         ret = WIMLIB_ERR_INVALID_DENTRY;
1009                         goto out_free_short_name;
1010                 }
1011                 ret = read_ads_entries(&metadata_resource[offset + calculated_size],
1012                                        dentry,
1013                                        metadata_resource_len - offset - calculated_size);
1014                 if (ret != 0)
1015                         goto out_free_short_name;
1016         }
1017
1018         dentry->file_name          = file_name;
1019         dentry->file_name_utf8     = file_name_utf8;
1020         dentry->short_name         = short_name;
1021         dentry->file_name_len      = file_name_len;
1022         dentry->file_name_utf8_len = file_name_utf8_len;
1023         dentry->short_name_len     = short_name_len;
1024         return 0;
1025 out_free_short_name:
1026         FREE(short_name);
1027 out_free_file_name_utf8:
1028         FREE(file_name_utf8);
1029 out_free_file_name:
1030         FREE(file_name);
1031         return ret;
1032 }
1033
1034 /*
1035  * Writes a dentry to an output buffer.
1036  *
1037  * @dentry:  The dentry structure.
1038  * @p:       The memory location to write the data to.
1039  * @return:  Pointer to the byte after the last byte we wrote as part of the
1040  *              dentry.
1041  */
1042 static u8 *write_dentry(const struct dentry *dentry, u8 *p)
1043 {
1044         u8 *orig_p = p;
1045         unsigned padding;
1046         memset(p, 0, dentry->length);
1047         p = put_u64(p, dentry->length);
1048         p = put_u32(p, dentry->attributes);
1049         p = put_u32(p, dentry->security_id);
1050         p = put_u64(p, dentry->subdir_offset);
1051         p = put_u64(p, 0); /* unused1 */
1052         p = put_u64(p, 0); /* unused2 */
1053         p = put_u64(p, dentry->creation_time);
1054         p = put_u64(p, dentry->last_access_time);
1055         p = put_u64(p, dentry->last_write_time);
1056         memcpy(p, dentry->hash, WIM_HASH_SIZE);
1057         p += WIM_HASH_SIZE;
1058         p = put_u32(p, dentry->reparse_tag);
1059         p = put_u64(p, dentry->hard_link);
1060         p = put_u16(p, dentry->num_ads); /*streams */
1061         p = put_u16(p, dentry->short_name_len);
1062         p = put_u16(p, dentry->file_name_len);
1063         p = put_bytes(p, dentry->file_name_len, (u8*)dentry->file_name);
1064         p = put_u16(p, 0); /* filename padding, 2 bytes. */
1065         p = put_bytes(p, dentry->short_name_len, (u8*)dentry->short_name);
1066
1067         wimlib_assert(p - orig_p <= dentry->length);
1068         if (p - orig_p < dentry->length)
1069                 p = put_zeroes(p, dentry->length - (p - orig_p));
1070
1071         p = put_zeroes(p, (8 - (p - orig_p) % 8) % 8);
1072
1073         for (u16 i = 0; i < dentry->num_ads; i++) {
1074                 p = put_u64(p, ads_entry_length(&dentry->ads_entries[i]));
1075                 p = put_u64(p, 0); /* Unused */
1076                 p = put_bytes(p, WIM_HASH_SIZE, dentry->ads_entries[i].hash);
1077                 p = put_u16(p, dentry->ads_entries[i].stream_name_len);
1078                 p = put_bytes(p, dentry->ads_entries[i].stream_name_len,
1079                                  (u8*)dentry->ads_entries[i].stream_name);
1080                 p = put_zeroes(p, (8 - (p - orig_p) % 8) % 8);
1081         }
1082         return p;
1083 }
1084
1085 /* Recursive function that writes a dentry tree rooted at @tree, not including
1086  * @tree itself, which has already been written, except in the case of the root
1087  * dentry, which is written right away, along with an end-of-directory entry. */
1088 u8 *write_dentry_tree(const struct dentry *tree, u8 *p)
1089 {
1090         const struct dentry *child;
1091
1092         if (dentry_is_root(tree)) {
1093                 p = write_dentry(tree, p);
1094
1095                 /* write end of directory entry */
1096                 p = put_u64(p, 0);
1097         } else {
1098                 /* Nothing to do for a regular file. */
1099                 if (dentry_is_regular_file(tree))
1100                         return p;
1101         }
1102
1103         /* Write child dentries and end-of-directory entry. */
1104         child = tree->children;
1105         if (child) {
1106                 do {
1107                         p = write_dentry(child, p);
1108                         child = child->next;
1109                 } while (child != tree->children);
1110         }
1111
1112         /* write end of directory entry */
1113         p = put_u64(p, 0);
1114
1115         /* Recurse on children. */
1116         if (child) {
1117                 do {
1118                         p = write_dentry_tree(child, p);
1119                         child = child->next;
1120                 } while (child != tree->children);
1121         }
1122         return p;
1123 }
1124
1125 /* Reads the children of a dentry, and all their children, ..., etc. from the
1126  * metadata resource and into the dentry tree.
1127  *
1128  * @metadata_resource:  An array that contains the uncompressed metadata
1129  *                      resource for the WIM file.
1130  * @metadata_resource_len:      The length of @metadata_resource.
1131  * @dentry:     A pointer to a struct dentry that is the root of the directory
1132  *              tree and has already been read from the metadata resource.  It
1133  *              does not need to be the real root because this procedure is
1134  *              called recursively.
1135  *
1136  * @return:     Zero on success, nonzero on failure.
1137  */
1138 int read_dentry_tree(const u8 metadata_resource[], u64 metadata_resource_len,
1139                      struct dentry *dentry)
1140 {
1141         u64 cur_offset = dentry->subdir_offset;
1142         struct dentry *prev_child = NULL;
1143         struct dentry *first_child = NULL;
1144         struct dentry *child;
1145         struct dentry cur_child;
1146         int ret;
1147
1148         /* If @dentry is a regular file, nothing more needs to be done for this
1149          * branch. */
1150         if (cur_offset == 0)
1151                 return 0;
1152
1153         /* Find and read all the children of @dentry. */
1154         while (1) {
1155
1156                 /* Read next child of @dentry into @cur_child. */
1157                 ret = read_dentry(metadata_resource, metadata_resource_len,
1158                                   cur_offset, &cur_child);
1159                 if (ret != 0)
1160                         break;
1161
1162                 /* Check for end of directory. */
1163                 if (cur_child.length == 0) {
1164                         ret = 0;
1165                         break;
1166                 }
1167
1168                 /* Not end of directory.  Allocate this child permanently and
1169                  * link it to the parent and previous child. */
1170                 child = MALLOC(sizeof(struct dentry));
1171                 if (!child) {
1172                         ERROR("Failed to allocate %zu bytes for new dentry",
1173                               sizeof(struct dentry));
1174                         ret = WIMLIB_ERR_NOMEM;
1175                         break;
1176                 }
1177                 memcpy(child, &cur_child, sizeof(struct dentry));
1178
1179                 if (prev_child) {
1180                         prev_child->next = child;
1181                         child->prev = prev_child;
1182                 } else {
1183                         first_child = child;
1184                 }
1185
1186                 child->parent = dentry;
1187                 prev_child = child;
1188
1189                 /* If there are children of this child, call this procedure
1190                  * recursively. */
1191                 if (child->subdir_offset != 0) {
1192                         ret = read_dentry_tree(metadata_resource,
1193                                                metadata_resource_len, child);
1194                         if (ret != 0)
1195                                 break;
1196                 }
1197
1198                 /* Advance to the offset of the next child. */
1199                 cur_offset += dentry_total_length(child);
1200         }
1201
1202         /* Link last child to first one, and set parent's
1203          * children pointer to the first child.  */
1204         if (prev_child) {
1205                 prev_child->next = first_child;
1206                 first_child->prev = prev_child;
1207         }
1208         dentry->children = first_child;
1209         return ret;
1210 }
1211
1212 int dentry_set_symlink_buf(struct dentry *dentry, const u8 symlink_buf_hash[])
1213 {
1214         struct ads_entry *ads_entries;
1215
1216         ads_entries = CALLOC(2, sizeof(struct ads_entry));
1217         if (!ads_entries)
1218                 return WIMLIB_ERR_NOMEM;
1219         memcpy(ads_entries[1].hash, symlink_buf_hash, WIM_HASH_SIZE);
1220         dentry_free_ads_entries(dentry);
1221         dentry->num_ads = 2;
1222         dentry->ads_entries = ads_entries;
1223         return 0;
1224 }