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         if (table)
  96                 lte = __lookup_resource(table, dentry_hash(dentry));
  97         else
  98                 lte = NULL;
  99
 100         if (lte) {
 101                 stbuf->st_nlink = lte->refcnt;
 102                 stbuf->st_size = lte->resource_entry.original_size;
 103         } else {
 104                 stbuf->st_nlink = 1;
 105                 stbuf->st_size = 0;
 106         }
 107         stbuf->st_uid     = getuid();
 108         stbuf->st_gid     = getgid();
 109         stbuf->st_atime   = ms_timestamp_to_unix(dentry->last_access_time);
 110         stbuf->st_mtime   = ms_timestamp_to_unix(dentry->last_write_time);
 111         stbuf->st_ctime   = ms_timestamp_to_unix(dentry->creation_time);
 112         stbuf->st_blocks  = (stbuf->st_size + 511) / 512;
 113 }
 114
 115 /* Makes all timestamp fields for the dentry be the current time. */
 116 void dentry_update_all_timestamps(struct dentry *dentry)
 117 {
 118         u64 now = get_timestamp();
 119         dentry->creation_time    = now;
 120         dentry->last_access_time = now;
 121         dentry->last_write_time  = now;
 122 }
 123
 124 struct ads_entry *dentry_get_ads_entry(struct dentry *dentry,
 125                                        const char *stream_name)
 126 {
 127         size_t stream_name_len = strlen(stream_name);
 128         if (!stream_name)
 129                 return NULL;
 130         for (u16 i = 0; i < dentry->num_ads; i++)
 131                 if (ads_entry_has_name(&dentry->ads_entries[i],
 132                                        stream_name, stream_name_len))
 133                         return &dentry->ads_entries[i];
 134         return NULL;
 135 }
 136
 137 /* Add an alternate stream entry to a dentry and return a pointer to it, or NULL
 138  * on failure. */
 139 struct ads_entry *dentry_add_ads(struct dentry *dentry, const char *stream_name)
 140 {
 141         u16 num_ads = dentry->num_ads + 1;
 142         struct ads_entry *ads_entries;
 143         struct ads_entry *new_entry;
 144         if (num_ads == 0xffff)
 145                 return NULL;
 146         ads_entries = MALLOC(num_ads * sizeof(struct ads_entry));
 147         if (!ads_entries)
 148                 return NULL;
 149
 150         new_entry = &ads_entries[num_ads - 1];
 151         if (change_ads_name(new_entry, stream_name) != 0) {
 152                 FREE(ads_entries);
 153                 return NULL;
 154         }
 155
 156         memcpy(ads_entries, dentry->ads_entries,
 157                (num_ads - 1) * sizeof(struct ads_entry));
 158         FREE(dentry->ads_entries);
 159         dentry->ads_entries = ads_entries;
 160         dentry->num_ads = num_ads;
 161         return memset(new_entry, 0, sizeof(struct ads_entry));
 162 }
 163
 164 void dentry_remove_ads(struct dentry *dentry, struct ads_entry *sentry)
 165 {
 166         destroy_ads_entry(sentry);
 167         memcpy(sentry, sentry + 1,
 168                (dentry->num_ads - (sentry - dentry->ads_entries))
 169                  * sizeof(struct ads_entry));
 170         dentry->num_ads--;
 171 }
 172
 173 /*
 174  * Calls a function on all directory entries in a directory tree.  It is called
 175  * on a parent before its children.
 176  */
 177 int for_dentry_in_tree(struct dentry *root,
 178                        int (*visitor)(struct dentry*, void*), void *arg)
 179 {
 180         int ret;
 181         struct dentry *child;
 182
 183         ret = visitor(root, arg);
 184
 185         if (ret != 0)
 186                 return ret;
 187
 188         child = root->children;
 189
 190         if (!child)
 191                 return 0;
 192
 193         do {
 194                 ret = for_dentry_in_tree(child, visitor, arg);
 195                 if (ret != 0)
 196                         return ret;
 197                 child = child->next;
 198         } while (child != root->children);
 199         return 0;
 200 }
 201
 202 /*
 203  * Like for_dentry_in_tree(), but the visitor function is always called on a
 204  * dentry's children before on itself.
 205  */
 206 int for_dentry_in_tree_depth(struct dentry *root,
 207                              int (*visitor)(struct dentry*, void*), void *arg)
 208 {
 209         int ret;
 210         struct dentry *child;
 211         struct dentry *next;
 212
 213         child = root->children;
 214         if (child) {
 215                 do {
 216                         next = child->next;
 217                         ret = for_dentry_in_tree_depth(child, visitor, arg);
 218                         if (ret != 0)
 219                                 return ret;
 220                         child = next;
 221                 } while (child != root->children);
 222         }
 223         return visitor(root, arg);
 224 }
 225
 226 /*
 227  * Calculate the full path of @dentry, based on its parent's full path and on
 228  * its UTF-8 file name.
 229  */
 230 int calculate_dentry_full_path(struct dentry *dentry, void *ignore)
 231 {
 232         char *full_path;
 233         u32 full_path_len;
 234         if (dentry_is_root(dentry)) {
 235                 full_path = MALLOC(2);
 236                 if (!full_path)
 237                         goto oom;
 238                 full_path[0] = '/';
 239                 full_path[1] = '\0';
 240                 full_path_len = 1;
 241         } else {
 242                 char *parent_full_path;
 243                 u32 parent_full_path_len;
 244                 const struct dentry *parent = dentry->parent;
 245
 246                 if (dentry_is_root(parent)) {
 247                         parent_full_path = "";
 248                         parent_full_path_len = 0;
 249                 } else {
 250                         parent_full_path = parent->full_path_utf8;
 251                         parent_full_path_len = parent->full_path_utf8_len;
 252                 }
 253
 254                 full_path_len = parent_full_path_len + 1 +
 255                                 dentry->file_name_utf8_len;
 256                 full_path = MALLOC(full_path_len + 1);
 257                 if (!full_path)
 258                         goto oom;
 259
 260                 memcpy(full_path, parent_full_path, parent_full_path_len);
 261                 full_path[parent_full_path_len] = '/';
 262                 memcpy(full_path + parent_full_path_len + 1,
 263                        dentry->file_name_utf8,
 264                        dentry->file_name_utf8_len);
 265                 full_path[full_path_len] = '\0';
 266         }
 267         FREE(dentry->full_path_utf8);
 268         dentry->full_path_utf8 = full_path;
 269         dentry->full_path_utf8_len = full_path_len;
 270         return 0;
 271 oom:
 272         ERROR("Out of memory while calculating dentry full path");
 273         return WIMLIB_ERR_NOMEM;
 274 }
 275
 276 /*
 277  * Recursively calculates the subdir offsets for a directory tree.
 278  *
 279  * @dentry:  The root of the directory tree.
 280  * @subdir_offset_p:  The current subdirectory offset; i.e., the subdirectory
 281  *      offset for @dentry.
 282  */
 283 void calculate_subdir_offsets(struct dentry *dentry, u64 *subdir_offset_p)
 284 {
 285         struct dentry *child;
 286
 287         child = dentry->children;
 288         dentry->subdir_offset = *subdir_offset_p;
 289         if (child) {
 290
 291                 /* Advance the subdir offset by the amount of space the children
 292                  * of this dentry take up. */
 293                 do {
 294                         *subdir_offset_p += dentry_total_length(child);
 295                         child = child->next;
 296                 } while (child != dentry->children);
 297
 298                 /* End-of-directory dentry on disk. */
 299                 *subdir_offset_p += 8;
 300
 301                 /* Recursively call calculate_subdir_offsets() on all the
 302                  * children. */
 303                 do {
 304                         calculate_subdir_offsets(child, subdir_offset_p);
 305                         child = child->next;
 306                 } while (child != dentry->children);
 307         } else {
 308                 /* On disk, childless directories have a valid subdir_offset
 309                  * that points to an 8-byte end-of-directory dentry.  Regular
 310                  * files have a subdir_offset of 0. */
 311                 if (dentry_is_directory(dentry))
 312                         *subdir_offset_p += 8;
 313                 else
 314                         dentry->subdir_offset = 0;
 315         }
 316 }
 317
 318
 319 /* Returns the child of @dentry that has the file name @name.
 320  * Returns NULL if no child has the name. */
 321 struct dentry *get_dentry_child_with_name(const struct dentry *dentry,
 322                                                         const char *name)
 323 {
 324         struct dentry *child;
 325         size_t name_len;
 326
 327         child = dentry->children;
 328         if (child) {
 329                 name_len = strlen(name);
 330                 do {
 331                         if (dentry_has_name(child, name, name_len))
 332                                 return child;
 333                         child = child->next;
 334                 } while (child != dentry->children);
 335         }
 336         return NULL;
 337 }
 338
 339 /* Retrieves the dentry that has the UTF-8 @path relative to the dentry
 340  * @cur_dir.  Returns NULL if no dentry having the path is found. */
 341 static struct dentry *get_dentry_relative_path(struct dentry *cur_dir, const char *path)
 342 {
 343         struct dentry *child;
 344         size_t base_len;
 345         const char *new_path;
 346
 347         if (*path == '\0')
 348                 return cur_dir;
 349
 350         child = cur_dir->children;
 351         if (child) {
 352                 new_path = path_next_part(path, &base_len);
 353                 do {
 354                         if (dentry_has_name(child, path, base_len))
 355                                 return get_dentry_relative_path(child, new_path);
 356                         child = child->next;
 357                 } while (child != cur_dir->children);
 358         }
 359         return NULL;
 360 }
 361
 362 /* Returns the dentry corresponding to the UTF-8 @path, or NULL if there is no
 363  * such dentry. */
 364 struct dentry *get_dentry(WIMStruct *w, const char *path)
 365 {
 366         struct dentry *root = wim_root_dentry(w);
 367         while (*path == '/')
 368                 path++;
 369         return get_dentry_relative_path(root, path);
 370 }
 371
 372 /* Returns the parent directory for the @path. */
 373 struct dentry *get_parent_dentry(WIMStruct *w, const char *path)
 374 {
 375         size_t path_len = strlen(path);
 376         char buf[path_len + 1];
 377
 378         memcpy(buf, path, path_len + 1);
 379
 380         to_parent_name(buf, path_len);
 381
 382         return get_dentry(w, buf);
 383 }
 384
 385 /* Prints the full path of a dentry. */
 386 int print_dentry_full_path(struct dentry *dentry, void *ignore)
 387 {
 388         if (dentry->full_path_utf8)
 389                 puts(dentry->full_path_utf8);
 390         return 0;
 391 }
 392
 393 struct file_attr_flag {
 394         u32 flag;
 395         const char *name;
 396 };
 397 struct file_attr_flag file_attr_flags[] = {
 398         {FILE_ATTRIBUTE_READONLY,               "READONLY"},
 399         {FILE_ATTRIBUTE_HIDDEN,         "HIDDEN"},
 400         {FILE_ATTRIBUTE_SYSTEM,         "SYSTEM"},
 401         {FILE_ATTRIBUTE_DIRECTORY,              "DIRECTORY"},
 402         {FILE_ATTRIBUTE_ARCHIVE,                "ARCHIVE"},
 403         {FILE_ATTRIBUTE_DEVICE,         "DEVICE"},
 404         {FILE_ATTRIBUTE_NORMAL,         "NORMAL"},
 405         {FILE_ATTRIBUTE_TEMPORARY,              "TEMPORARY"},
 406         {FILE_ATTRIBUTE_SPARSE_FILE,    "SPARSE_FILE"},
 407         {FILE_ATTRIBUTE_REPARSE_POINT,  "REPARSE_POINT"},
 408         {FILE_ATTRIBUTE_COMPRESSED,             "COMPRESSED"},
 409         {FILE_ATTRIBUTE_OFFLINE,                "OFFLINE"},
 410         {FILE_ATTRIBUTE_NOT_CONTENT_INDEXED,"NOT_CONTENT_INDEXED"},
 411         {FILE_ATTRIBUTE_ENCRYPTED,              "ENCRYPTED"},
 412         {FILE_ATTRIBUTE_VIRTUAL,                "VIRTUAL"},
 413 };
 414
 415 /* Prints a directory entry.  @lookup_table is a pointer to the lookup table, or
 416  * NULL if the resource entry for the dentry is not to be printed. */
 417 int print_dentry(struct dentry *dentry, void *lookup_table)
 418 {
 419         struct lookup_table_entry *lte;
 420         unsigned i;
 421
 422         printf("[DENTRY]\n");
 423         printf("Length            = %"PRIu64"\n", dentry->length);
 424         printf("Attributes        = 0x%x\n", dentry->attributes);
 425         for (i = 0; i < ARRAY_LEN(file_attr_flags); i++)
 426                 if (file_attr_flags[i].flag & dentry->attributes)
 427                         printf("    FILE_ATTRIBUTE_%s is set\n",
 428                                 file_attr_flags[i].name);
 429         printf("Security ID       = %d\n", dentry->security_id);
 430         printf("Subdir offset     = %"PRIu64"\n", dentry->subdir_offset);
 431         /*printf("Unused1           = 0x%"PRIu64"\n", dentry->unused1);*/
 432         /*printf("Unused2           = %"PRIu64"\n", dentry->unused2);*/
 433         printf("Creation Time     = 0x%"PRIx64"\n", dentry->creation_time);
 434         printf("Last Access Time  = 0x%"PRIx64"\n", dentry->last_access_time);
 435         printf("Last Write Time   = 0x%"PRIx64"\n", dentry->last_write_time);
 436         printf("Hash              = 0x");
 437         print_hash(dentry->hash);
 438         putchar('\n');
 439         printf("Reparse Tag       = 0x%"PRIx32"\n", dentry->reparse_tag);
 440         printf("Hard Link Group   = 0x%"PRIx64"\n", dentry->hard_link);
 441         printf("Number of Alternate Data Streams = %hu\n", dentry->num_ads);
 442         printf("Filename          = \"");
 443         print_string(dentry->file_name, dentry->file_name_len);
 444         puts("\"");
 445         printf("Filename Length   = %hu\n", dentry->file_name_len);
 446         printf("Filename (UTF-8)  = \"%s\"\n", dentry->file_name_utf8);
 447         printf("Filename (UTF-8) Length = %hu\n", dentry->file_name_utf8_len);
 448         printf("Short Name        = \"");
 449         print_string(dentry->short_name, dentry->short_name_len);
 450         puts("\"");
 451         printf("Short Name Length = %hu\n", dentry->short_name_len);
 452         printf("Full Path (UTF-8) = \"%s\"\n", dentry->full_path_utf8);
 453         if (lookup_table && (lte = __lookup_resource(lookup_table, dentry->hash)))
 454                 print_lookup_table_entry(lte, NULL);
 455         else
 456                 putchar('\n');
 457         for (u16 i = 0; i < dentry->num_ads; i++) {
 458                 printf("[Alternate Stream Entry %u]\n", i);
 459                 printf("Name = \"%s\"\n", dentry->ads_entries[i].stream_name_utf8);
 460                 printf("Name Length (UTF-16) = %u\n",
 461                                 dentry->ads_entries[i].stream_name_len);
 462                 printf("Hash              = 0x");
 463                 print_hash(dentry->ads_entries[i].hash);
 464                 if (lookup_table &&
 465                      (lte = __lookup_resource(lookup_table,
 466                                               dentry->ads_entries[i].hash)))
 467                 {
 468                         print_lookup_table_entry(lte, NULL);
 469                 } else {
 470                         putchar('\n');
 471                 }
 472         }
 473         return 0;
 474 }
 475
 476 static inline void dentry_common_init(struct dentry *dentry)
 477 {
 478         memset(dentry, 0, sizeof(struct dentry));
 479         dentry->refcnt = 1;
 480         dentry->security_id = -1;
 481 }
 482
 483 /*
 484  * Creates an unlinked directory entry.
 485  *
 486  * @name:    The base name of the new dentry.
 487  * @return:  A pointer to the new dentry, or NULL if out of memory.
 488  */
 489 struct dentry *new_dentry(const char *name)
 490 {
 491         struct dentry *dentry;
 492
 493         dentry = MALLOC(sizeof(struct dentry));
 494         if (!dentry)
 495                 goto err;
 496
 497         dentry_common_init(dentry);
 498         if (change_dentry_name(dentry, name) != 0)
 499                 goto err;
 500
 501         dentry_update_all_timestamps(dentry);
 502         dentry->next   = dentry;
 503         dentry->prev   = dentry;
 504         dentry->parent = dentry;
 505         return dentry;
 506 err:
 507         FREE(dentry);
 508         ERROR("Failed to allocate new dentry");
 509         return NULL;
 510 }
 511
 512 static void dentry_free_ads_entries(struct dentry *dentry)
 513 {
 514         for (u16 i = 0; i < dentry->num_ads; i++)
 515                 destroy_ads_entry(&dentry->ads_entries[i]);
 516         FREE(dentry->ads_entries);
 517         dentry->ads_entries = NULL;
 518         dentry->num_ads = 0;
 519 }
 520
 521
 522 void free_dentry(struct dentry *dentry)
 523 {
 524         FREE(dentry->file_name);
 525         FREE(dentry->file_name_utf8);
 526         FREE(dentry->short_name);
 527         FREE(dentry->full_path_utf8);
 528         dentry_free_ads_entries(dentry);
 529         FREE(dentry);
 530 }
 531
 532 /* Arguments for do_free_dentry(). */
 533 struct free_dentry_args {
 534         struct lookup_table *lookup_table;
 535         bool lt_decrement_refcnt;
 536 };
 537
 538 /*
 539  * This function is passed as an argument to for_dentry_in_tree_depth() in order
 540  * to free a directory tree.  __args is a pointer to a `struct free_dentry_args'.
 541  */
 542 static int do_free_dentry(struct dentry *dentry, void *__args)
 543 {
 544         struct free_dentry_args *args = (struct free_dentry_args*)__args;
 545
 546         if (args->lt_decrement_refcnt && !dentry_is_directory(dentry)) {
 547                 lookup_table_decrement_refcnt(args->lookup_table,
 548                                               dentry->hash);
 549         }
 550
 551         wimlib_assert(dentry->refcnt >= 1);
 552         if (--dentry->refcnt == 0)
 553                 free_dentry(dentry);
 554         return 0;
 555 }
 556
 557 /*
 558  * Unlinks and frees a dentry tree.
 559  *
 560  * @root:               The root of the tree.
 561  * @lookup_table:       The lookup table for dentries.
 562  * @decrement_refcnt:   True if the dentries in the tree are to have their
 563  *                      reference counts in the lookup table decremented.
 564  */
 565 void free_dentry_tree(struct dentry *root, struct lookup_table *lookup_table,
 566                       bool lt_decrement_refcnt)
 567 {
 568         if (!root || !root->parent)
 569                 return;
 570
 571         struct free_dentry_args args;
 572         args.lookup_table        = lookup_table;
 573         args.lt_decrement_refcnt = lt_decrement_refcnt;
 574         for_dentry_in_tree_depth(root, do_free_dentry, &args);
 575 }
 576
 577 int increment_dentry_refcnt(struct dentry *dentry, void *ignore)
 578 {
 579         dentry->refcnt++;
 580         return 0;
 581 }
 582
 583 /*
 584  * Links a dentry into the directory tree.
 585  *
 586  * @dentry: The dentry to link.
 587  * @parent: The dentry that will be the parent of @dentry.
 588  */
 589 void link_dentry(struct dentry *dentry, struct dentry *parent)
 590 {
 591         dentry->parent = parent;
 592         if (parent->children) {
 593                 /* Not an only child; link to siblings. */
 594                 dentry->next = parent->children;
 595                 dentry->prev = parent->children->prev;
 596                 dentry->next->prev = dentry;
 597                 dentry->prev->next = dentry;
 598         } else {
 599                 /* Only child; link to parent. */
 600                 parent->children = dentry;
 601                 dentry->next = dentry;
 602                 dentry->prev = dentry;
 603         }
 604 }
 605
 606 /* Unlink a dentry from the directory tree. */
 607 void unlink_dentry(struct dentry *dentry)
 608 {
 609         if (dentry_is_root(dentry))
 610                 return;
 611         if (dentry_is_only_child(dentry)) {
 612                 dentry->parent->children = NULL;
 613         } else {
 614                 if (dentry_is_first_sibling(dentry))
 615                         dentry->parent->children = dentry->next;
 616                 dentry->next->prev = dentry->prev;
 617                 dentry->prev->next = dentry->next;
 618         }
 619 }
 620
 621
 622 /* Recalculates the length of @dentry based on its file name length and short
 623  * name length.  */
 624 static inline void recalculate_dentry_size(struct dentry *dentry)
 625 {
 626         dentry->length = WIM_DENTRY_DISK_SIZE + dentry->file_name_len +
 627                          2 + dentry->short_name_len;
 628         /* Must be multiple of 8. */
 629         dentry->length = (dentry->length + 7) & ~7;
 630 }
 631
 632 static int do_name_change(char **file_name_ret,
 633                           char **file_name_utf8_ret,
 634                           u16 *file_name_len_ret,
 635                           u16 *file_name_utf8_len_ret,
 636                           const char *new_name)
 637 {
 638         size_t utf8_len;
 639         size_t utf16_len;
 640         char *file_name, *file_name_utf8;
 641
 642         utf8_len = strlen(new_name);
 643
 644         file_name = utf8_to_utf16(new_name, utf8_len, &utf16_len);
 645
 646         if (!file_name)
 647                 return WIMLIB_ERR_NOMEM;
 648
 649         file_name_utf8 = MALLOC(utf8_len + 1);
 650         if (!file_name_utf8) {
 651                 FREE(file_name);
 652                 return WIMLIB_ERR_NOMEM;
 653         }
 654         memcpy(file_name_utf8, new_name, utf8_len + 1);
 655
 656         FREE(*file_name_ret);
 657         FREE(*file_name_utf8_ret);
 658         *file_name_ret          = file_name;
 659         *file_name_utf8_ret     = file_name_utf8;
 660         *file_name_len_ret      = utf16_len;
 661         *file_name_utf8_len_ret = utf8_len;
 662         return 0;
 663 }
 664
 665 /* Changes the name of a dentry to @new_name.  Only changes the file_name and
 666  * file_name_utf8 fields; does not change the short_name, short_name_utf8, or
 667  * full_path_utf8 fields.  Also recalculates its length. */
 668 int change_dentry_name(struct dentry *dentry, const char *new_name)
 669 {
 670         int ret;
 671
 672         ret = do_name_change(&dentry->file_name, &dentry->file_name_utf8,
 673                              &dentry->file_name_len, &dentry->file_name_utf8_len,
 674                              new_name);
 675         if (ret == 0)
 676                 recalculate_dentry_size(dentry);
 677         return ret;
 678 }
 679
 680 int change_ads_name(struct ads_entry *entry, const char *new_name)
 681 {
 682         return do_name_change(&entry->stream_name, &entry->stream_name_utf8,
 683                               &entry->stream_name_len,
 684                               &entry->stream_name_utf8_len,
 685                               new_name);
 686 }
 687
 688 /* Parameters for calculate_dentry_statistics(). */
 689 struct image_statistics {
 690         struct lookup_table *lookup_table;
 691         u64 *dir_count;
 692         u64 *file_count;
 693         u64 *total_bytes;
 694         u64 *hard_link_bytes;
 695 };
 696
 697 static int calculate_dentry_statistics(struct dentry *dentry, void *arg)
 698 {
 699         struct image_statistics *stats;
 700         struct lookup_table_entry *lte;
 701         u16 i;
 702
 703         stats = arg;
 704
 705         if (dentry_is_directory(dentry) && !dentry_is_root(dentry))
 706                 ++*stats->dir_count;
 707         else
 708                 ++*stats->file_count;
 709
 710         lte = __lookup_resource(stats->lookup_table, dentry->hash);
 711         i = 0;
 712         while (1) {
 713                 if (lte) {
 714                         u64 size = lte->resource_entry.original_size;
 715                         *stats->total_bytes += size;
 716                         if (++lte->out_refcnt == 1)
 717                                 *stats->hard_link_bytes += size;
 718                 }
 719                 if (i == dentry->num_ads)
 720                         break;
 721                 lte = __lookup_resource(stats->lookup_table,
 722                                         dentry->ads_entries[i].hash);
 723                 i++;
 724         }
 725
 726         return 0;
 727 }
 728
 729 void calculate_dir_tree_statistics(struct dentry *root, struct lookup_table *table,
 730                                    u64 *dir_count_ret, u64 *file_count_ret,
 731                                    u64 *total_bytes_ret,
 732                                    u64 *hard_link_bytes_ret)
 733 {
 734         struct image_statistics stats;
 735         *dir_count_ret         = 0;
 736         *file_count_ret        = 0;
 737         *total_bytes_ret       = 0;
 738         *hard_link_bytes_ret   = 0;
 739         stats.lookup_table     = table;
 740         stats.dir_count       = dir_count_ret;
 741         stats.file_count      = file_count_ret;
 742         stats.total_bytes     = total_bytes_ret;
 743         stats.hard_link_bytes = hard_link_bytes_ret;
 744         for_lookup_table_entry(table, zero_out_refcnts, NULL);
 745         for_dentry_in_tree(root, calculate_dentry_statistics, &stats);
 746 }
 747
 748 static int read_ads_entries(const u8 *p, struct dentry *dentry,
 749                             u64 remaining_size)
 750 {
 751         u16 num_ads = dentry->num_ads;
 752         struct ads_entry *ads_entries = CALLOC(num_ads, sizeof(struct ads_entry));
 753         int ret;
 754         if (!ads_entries) {
 755                 ERROR("Could not allocate memory for %"PRIu16" "
 756                       "alternate data stream entries", num_ads);
 757                 return WIMLIB_ERR_NOMEM;
 758         }
 759         DEBUG2("Reading %"PRIu16" alternate data streams "
 760                "(remaining size = %"PRIu64")", num_ads, remaining_size);
 761
 762         for (u16 i = 0; i < num_ads; i++) {
 763                 struct ads_entry *cur_entry = &ads_entries[i];
 764                 u64 length;
 765                 size_t utf8_len;
 766                 const char *p_save = p;
 767                 /* Read the base stream entry, excluding the stream name. */
 768                 if (remaining_size < WIM_ADS_ENTRY_DISK_SIZE) {
 769                         ERROR("Stream entries go past end of metadata resource");
 770                         ERROR("(remaining_size = %"PRIu64")", remaining_size);
 771                         ret = WIMLIB_ERR_INVALID_DENTRY;
 772                         goto out_free_ads_entries;
 773                 }
 774                 remaining_size -= WIM_ADS_ENTRY_DISK_SIZE;
 775
 776                 p = get_u64(p, &length); /* ADS entry length */
 777
 778                 DEBUG2("ADS length = %"PRIu64, length);
 779
 780                 p += 8; /* Unused */
 781                 p = get_bytes(p, WIM_HASH_SIZE, (u8*)cur_entry->hash);
 782                 p = get_u16(p, &cur_entry->stream_name_len);
 783
 784                 DEBUG2("Stream name length = %u", cur_entry->stream_name_len);
 785
 786                 cur_entry->stream_name = NULL;
 787                 cur_entry->stream_name_utf8 = NULL;
 788
 789                 if (remaining_size < cur_entry->stream_name_len + 2) {
 790                         ERROR("Stream entries go past end of metadata resource");
 791                         ERROR("(remaining_size = %"PRIu64" bytes, stream_name_len "
 792                               "= %"PRIu16" bytes", remaining_size,
 793                               cur_entry->stream_name_len);
 794                         ret = WIMLIB_ERR_INVALID_DENTRY;
 795                         goto out_free_ads_entries;
 796                 }
 797                 remaining_size -= cur_entry->stream_name_len + 2;
 798
 799                 cur_entry->stream_name = MALLOC(cur_entry->stream_name_len);
 800                 if (!cur_entry->stream_name) {
 801                         ret = WIMLIB_ERR_NOMEM;
 802                         goto out_free_ads_entries;
 803                 }
 804                 get_bytes(p, cur_entry->stream_name_len,
 805                           (u8*)cur_entry->stream_name);
 806                 cur_entry->stream_name_utf8 = utf16_to_utf8(cur_entry->stream_name,
 807                                                             cur_entry->stream_name_len,
 808                                                             &utf8_len);
 809                 cur_entry->stream_name_utf8_len = utf8_len;
 810
 811                 if (!cur_entry->stream_name_utf8) {
 812                         ret = WIMLIB_ERR_NOMEM;
 813                         goto out_free_ads_entries;
 814                 }
 815                 p = p_save + ads_entry_length(cur_entry);
 816         }
 817         dentry->ads_entries = ads_entries;
 818         return 0;
 819 out_free_ads_entries:
 820         for (u16 i = 0; i < num_ads; i++) {
 821                 FREE(ads_entries[i].stream_name);
 822                 FREE(ads_entries[i].stream_name_utf8);
 823         }
 824         FREE(ads_entries);
 825         return ret;
 826 }
 827
 828 /*
 829  * Reads a directory entry from the metadata resource.
 830  */
 831 int read_dentry(const u8 metadata_resource[], u64 metadata_resource_len,
 832                 u64 offset, struct dentry *dentry)
 833 {
 834         const u8 *p;
 835         u64 calculated_size;
 836         char *file_name;
 837         char *file_name_utf8;
 838         char *short_name;
 839         u16 short_name_len;
 840         u16 file_name_len;
 841         size_t file_name_utf8_len;
 842         int ret;
 843
 844         dentry_common_init(dentry);
 845
 846         /*Make sure the dentry really fits into the metadata resource.*/
 847         if (offset + 8 > metadata_resource_len) {
 848                 ERROR("Directory entry starting at %"PRIu64" ends past the "
 849                       "end of the metadata resource (size %"PRIu64")",
 850                       offset, metadata_resource_len);
 851                 return WIMLIB_ERR_INVALID_DENTRY;
 852         }
 853
 854         /* Before reading the whole entry, we need to read just the length.
 855          * This is because an entry of length 8 (that is, just the length field)
 856          * terminates the list of sibling directory entries. */
 857
 858         p = get_u64(&metadata_resource[offset], &dentry->length);
 859
 860         /* A zero length field (really a length of 8, since that's how big the
 861          * directory entry is...) indicates that this is the end of directory
 862          * dentry.  We do not read it into memory as an actual dentry, so just
 863          * return true in that case. */
 864         if (dentry->length == 0)
 865                 return 0;
 866
 867         if (offset + dentry->length >= metadata_resource_len) {
 868                 ERROR("Directory entry at offset %"PRIu64" and with size "
 869                       "%"PRIu64" ends past the end of the metadata resource "
 870                       "(size %"PRIu64")",
 871                       offset, dentry->length, metadata_resource_len);
 872                 return WIMLIB_ERR_INVALID_DENTRY;
 873         }
 874
 875         /* If it is a recognized length, read the rest of the directory entry.
 876          * Note: The root directory entry has no name, and its length does not
 877          * include the short name length field.  */
 878         if (dentry->length < WIM_DENTRY_DISK_SIZE) {
 879                 ERROR("Directory entry has invalid length of %"PRIu64" bytes",
 880                       dentry->length);
 881                 return WIMLIB_ERR_INVALID_DENTRY;
 882         }
 883
 884         p = get_u32(p, &dentry->attributes);
 885         p = get_u32(p, (u32*)&dentry->security_id);
 886         p = get_u64(p, &dentry->subdir_offset);
 887
 888         /* 2 unused fields */
 889         p += 2 * sizeof(u64);
 890         /*p = get_u64(p, &dentry->unused1);*/
 891         /*p = get_u64(p, &dentry->unused2);*/
 892
 893         p = get_u64(p, &dentry->creation_time);
 894         p = get_u64(p, &dentry->last_access_time);
 895         p = get_u64(p, &dentry->last_write_time);
 896
 897         p = get_bytes(p, WIM_HASH_SIZE, dentry->hash);
 898
 899         /*
 900          * I don't know what's going on here.  It seems like M$ screwed up the
 901          * reparse points, then put the fields in the same place and didn't
 902          * document it.  The WIM_HDR_FLAG_RP_FIX flag in the WIM header might
 903          * have something to do with this, but it's not documented.
 904          */
 905         if (dentry->attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
 906                 /* ??? */
 907                 u32 u1, u2;
 908                 p = get_u32(p, &u1);
 909                 /*p += 4;*/
 910                 p = get_u32(p, &dentry->reparse_tag);
 911                 p = get_u32(p, &u2);
 912                 /*p += 4;*/
 913                 dentry->hard_link = (u64)(u1) | ((u64)(u2) << 32);
 914         } else {
 915                 p = get_u32(p, &dentry->reparse_tag);
 916                 p = get_u64(p, &dentry->hard_link);
 917         }
 918
 919         /* By the way, the reparse_reserved field does not actually exist (at
 920          * least when the file is not a reparse point) */
 921
 922
 923         p = get_u16(p, &dentry->num_ads);
 924
 925         p = get_u16(p, &short_name_len);
 926         p = get_u16(p, &file_name_len);
 927
 928         calculated_size = WIM_DENTRY_DISK_SIZE + file_name_len + 2 +
 929                           short_name_len;
 930
 931         if (dentry->length < calculated_size) {
 932                 ERROR("Unexpected end of directory entry! (Expected "
 933                       "%"PRIu64" bytes, got %"PRIu64" bytes. "
 934                       "short_name_len = %hu, file_name_len = %hu)",
 935                       calculated_size, dentry->length,
 936                       short_name_len, file_name_len);
 937                 return WIMLIB_ERR_INVALID_DENTRY;
 938         }
 939
 940         /* Read the filename. */
 941         file_name = MALLOC(file_name_len);
 942         if (!file_name) {
 943                 ERROR("Failed to allocate %hu bytes for dentry file name",
 944                       file_name_len);
 945                 return WIMLIB_ERR_NOMEM;
 946         }
 947         p = get_bytes(p, file_name_len, file_name);
 948
 949         /* Convert filename to UTF-8. */
 950         file_name_utf8 = utf16_to_utf8(file_name, file_name_len,
 951                                        &file_name_utf8_len);
 952
 953         if (!file_name_utf8) {
 954                 ERROR("Failed to allocate memory to convert UTF-16 "
 955                       "filename (%hu bytes) to UTF-8", file_name_len);
 956                 ret = WIMLIB_ERR_NOMEM;
 957                 goto out_free_file_name;
 958         }
 959
 960         /* Undocumented padding between file name and short name.  This probably
 961          * is supposed to be a terminating null character. */
 962         p += 2;
 963
 964         /* Read the short filename. */
 965         short_name = MALLOC(short_name_len);
 966         if (!short_name) {
 967                 ERROR("Failed to allocate %hu bytes for short filename",
 968                       short_name_len);
 969                 ret = WIMLIB_ERR_NOMEM;
 970                 goto out_free_file_name_utf8;
 971         }
 972
 973         p = get_bytes(p, short_name_len, short_name);
 974
 975         /* Some directory entries inexplicibly have a little over 70 bytes of
 976          * extra data.  The exact amount of data seems to be 72 bytes, but it is
 977          * aligned on the next 8-byte boundary.  Here's an example of the
 978          * aligned data:
 979          *
 980          * 01000000400000006c786bbac58ede11b0bb00261870892ab6adb76fe63a3
 981          * e468fca86530d2effa16c786bbac58ede11b0bb00261870892a0000000000
 982          * 0000000000000000000000
 983          *
 984          * Here's one interpretation of how the data is laid out.
 985          *
 986          * struct unknown {
 987          *      u32 field1; (always 0x00000001)
 988          *      u32 field2; (always 0x40000000)
 989          *      u16 field3;
 990          *      u32 field4;
 991          *      u32 field5;
 992          *      u32 field6;
 993          *      u8  data[48]; (???)
 994          *      u64 reserved1; (always 0)
 995          *      u64 reserved2; (always 0)
 996          * };*/
 997 #if 0
 998         if (dentry->length - calculated_size >= WIM_ADS_ENTRY_DISK_SIZE) {
 999                 printf("%s: %lu / %lu (", file_name_utf8,
1000                                 calculated_size, dentry->length);
1001                 print_string(p + WIM_ADS_ENTRY_DISK_SIZE, dentry->length - calculated_size - WIM_ADS_ENTRY_DISK_SIZE);
1002                 puts(")");
1003                 print_byte_field(p, dentry->length - calculated_size);
1004                 putchar('\n');
1005         }
1006 #endif
1007
1008         if (dentry->num_ads != 0) {
1009                 calculated_size = (calculated_size + 7) & ~7;
1010                 if (calculated_size > metadata_resource_len - offset) {
1011                         ERROR("Not enough space in metadata resource for "
1012                               "alternate stream entries");
1013                         ret = WIMLIB_ERR_INVALID_DENTRY;
1014                         goto out_free_short_name;
1015                 }
1016                 ret = read_ads_entries(&metadata_resource[offset + calculated_size],
1017                                        dentry,
1018                                        metadata_resource_len - offset - calculated_size);
1019                 if (ret != 0)
1020                         goto out_free_short_name;
1021         }
1022
1023         dentry->file_name          = file_name;
1024         dentry->file_name_utf8     = file_name_utf8;
1025         dentry->short_name         = short_name;
1026         dentry->file_name_len      = file_name_len;
1027         dentry->file_name_utf8_len = file_name_utf8_len;
1028         dentry->short_name_len     = short_name_len;
1029         return 0;
1030 out_free_short_name:
1031         FREE(short_name);
1032 out_free_file_name_utf8:
1033         FREE(file_name_utf8);
1034 out_free_file_name:
1035         FREE(file_name);
1036         return ret;
1037 }
1038
1039 /*
1040  * Writes a dentry to an output buffer.
1041  *
1042  * @dentry:  The dentry structure.
1043  * @p:       The memory location to write the data to.
1044  * @return:  Pointer to the byte after the last byte we wrote as part of the
1045  *              dentry.
1046  */
1047 static u8 *write_dentry(const struct dentry *dentry, u8 *p)
1048 {
1049         u8 *orig_p = p;
1050         unsigned padding;
1051         memset(p, 0, dentry->length);
1052         p = put_u64(p, dentry->length);
1053         p = put_u32(p, dentry->attributes);
1054         p = put_u32(p, dentry->security_id);
1055         p = put_u64(p, dentry->subdir_offset);
1056         p = put_u64(p, 0); /* unused1 */
1057         p = put_u64(p, 0); /* unused2 */
1058         p = put_u64(p, dentry->creation_time);
1059         p = put_u64(p, dentry->last_access_time);
1060         p = put_u64(p, dentry->last_write_time);
1061         memcpy(p, dentry->hash, WIM_HASH_SIZE);
1062         p += WIM_HASH_SIZE;
1063         p = put_u32(p, dentry->reparse_tag);
1064         p = put_u64(p, dentry->hard_link);
1065         p = put_u16(p, dentry->num_ads); /*streams */
1066         p = put_u16(p, dentry->short_name_len);
1067         p = put_u16(p, dentry->file_name_len);
1068         p = put_bytes(p, dentry->file_name_len, (u8*)dentry->file_name);
1069         p = put_u16(p, 0); /* filename padding, 2 bytes. */
1070         p = put_bytes(p, dentry->short_name_len, (u8*)dentry->short_name);
1071
1072         wimlib_assert(p - orig_p <= dentry->length);
1073         if (p - orig_p < dentry->length)
1074                 p = put_zeroes(p, dentry->length - (p - orig_p));
1075
1076         p = put_zeroes(p, (8 - (p - orig_p) % 8) % 8);
1077
1078         for (u16 i = 0; i < dentry->num_ads; i++) {
1079                 p = put_u64(p, ads_entry_length(&dentry->ads_entries[i]));
1080                 p = put_u64(p, 0); /* Unused */
1081                 p = put_bytes(p, WIM_HASH_SIZE, dentry->ads_entries[i].hash);
1082                 p = put_u16(p, dentry->ads_entries[i].stream_name_len);
1083                 p = put_bytes(p, dentry->ads_entries[i].stream_name_len,
1084                                  (u8*)dentry->ads_entries[i].stream_name);
1085                 p = put_zeroes(p, (8 - (p - orig_p) % 8) % 8);
1086         }
1087         return p;
1088 }
1089
1090 /* Recursive function that writes a dentry tree rooted at @tree, not including
1091  * @tree itself, which has already been written, except in the case of the root
1092  * dentry, which is written right away, along with an end-of-directory entry. */
1093 u8 *write_dentry_tree(const struct dentry *tree, u8 *p)
1094 {
1095         const struct dentry *child;
1096
1097         if (dentry_is_root(tree)) {
1098                 p = write_dentry(tree, p);
1099
1100                 /* write end of directory entry */
1101                 p = put_u64(p, 0);
1102         } else {
1103                 /* Nothing to do for a regular file. */
1104                 if (dentry_is_regular_file(tree))
1105                         return p;
1106         }
1107
1108         /* Write child dentries and end-of-directory entry. */
1109         child = tree->children;
1110         if (child) {
1111                 do {
1112                         p = write_dentry(child, p);
1113                         child = child->next;
1114                 } while (child != tree->children);
1115         }
1116
1117         /* write end of directory entry */
1118         p = put_u64(p, 0);
1119
1120         /* Recurse on children. */
1121         if (child) {
1122                 do {
1123                         p = write_dentry_tree(child, p);
1124                         child = child->next;
1125                 } while (child != tree->children);
1126         }
1127         return p;
1128 }
1129
1130 /* Reads the children of a dentry, and all their children, ..., etc. from the
1131  * metadata resource and into the dentry tree.
1132  *
1133  * @metadata_resource:  An array that contains the uncompressed metadata
1134  *                      resource for the WIM file.
1135  * @metadata_resource_len:      The length of @metadata_resource.
1136  * @dentry:     A pointer to a struct dentry that is the root of the directory
1137  *              tree and has already been read from the metadata resource.  It
1138  *              does not need to be the real root because this procedure is
1139  *              called recursively.
1140  *
1141  * @return:     Zero on success, nonzero on failure.
1142  */
1143 int read_dentry_tree(const u8 metadata_resource[], u64 metadata_resource_len,
1144                      struct dentry *dentry)
1145 {
1146         u64 cur_offset = dentry->subdir_offset;
1147         struct dentry *prev_child = NULL;
1148         struct dentry *first_child = NULL;
1149         struct dentry *child;
1150         struct dentry cur_child;
1151         int ret;
1152
1153         /* If @dentry is a regular file, nothing more needs to be done for this
1154          * branch. */
1155         if (cur_offset == 0)
1156                 return 0;
1157
1158         /* Find and read all the children of @dentry. */
1159         while (1) {
1160
1161                 /* Read next child of @dentry into @cur_child. */
1162                 ret = read_dentry(metadata_resource, metadata_resource_len,
1163                                   cur_offset, &cur_child);
1164                 if (ret != 0)
1165                         break;
1166
1167                 /* Check for end of directory. */
1168                 if (cur_child.length == 0) {
1169                         ret = 0;
1170                         break;
1171                 }
1172
1173                 /* Not end of directory.  Allocate this child permanently and
1174                  * link it to the parent and previous child. */
1175                 child = MALLOC(sizeof(struct dentry));
1176                 if (!child) {
1177                         ERROR("Failed to allocate %zu bytes for new dentry",
1178                               sizeof(struct dentry));
1179                         ret = WIMLIB_ERR_NOMEM;
1180                         break;
1181                 }
1182                 memcpy(child, &cur_child, sizeof(struct dentry));
1183
1184                 if (prev_child) {
1185                         prev_child->next = child;
1186                         child->prev = prev_child;
1187                 } else {
1188                         first_child = child;
1189                 }
1190
1191                 child->parent = dentry;
1192                 prev_child = child;
1193
1194                 /* If there are children of this child, call this procedure
1195                  * recursively. */
1196                 if (child->subdir_offset != 0) {
1197                         ret = read_dentry_tree(metadata_resource,
1198                                                metadata_resource_len, child);
1199                         if (ret != 0)
1200                                 break;
1201                 }
1202
1203                 /* Advance to the offset of the next child. */
1204                 cur_offset += dentry_total_length(child);
1205         }
1206
1207         /* Link last child to first one, and set parent's
1208          * children pointer to the first child.  */
1209         if (prev_child) {
1210                 prev_child->next = first_child;
1211                 first_child->prev = prev_child;
1212         }
1213         dentry->children = first_child;
1214         return ret;
1215 }
1216
1217 int dentry_set_symlink_buf(struct dentry *dentry, const u8 symlink_buf_hash[])
1218 {
1219         struct ads_entry *ads_entries;
1220
1221         ads_entries = CALLOC(2, sizeof(struct ads_entry));
1222         if (!ads_entries)
1223                 return WIMLIB_ERR_NOMEM;
1224         memcpy(ads_entries[1].hash, symlink_buf_hash, WIM_HASH_SIZE);
1225         dentry_free_ads_entries(dentry);
1226         dentry->num_ads = 2;
1227         dentry->ads_entries = ads_entries;
1228         return 0;
1229 }