wimlib.net Git - wimlib/blob - src/ntfs-capture.c

   1 /*
   2  * ntfs-capture.c
   3  *
   4  * Capture a WIM image from a NTFS volume.  We capture everything we can,
   5  * including security data and alternate data streams.  There should be no loss
   6  * of information.
   7  */
   8
   9 /*
  10  * Copyright (C) 2012 Eric Biggers
  11  *
  12  * This file is part of wimlib, a library for working with WIM files.
  13  *
  14  * wimlib is free software; you can redistribute it and/or modify it under the
  15  * terms of the GNU Lesser General Public License as published by the Free
  16  * Software Foundation; either version 2.1 of the License, or (at your option)
  17  * any later version.
  18  *
  19  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  20  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  21  * A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
  22  * details.
  23  *
  24  * You should have received a copy of the GNU Lesser General Public License
  25  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  26  */
  27
  28 #include "config.h"
  29 #include "wimlib_internal.h"
  30
  31
  32 #ifdef WITH_NTFS_3G
  33 #include "dentry.h"
  34 #include "lookup_table.h"
  35 #include "io.h"
  36 #include <ntfs-3g/layout.h>
  37 #include <ntfs-3g/acls.h>
  38 #include <ntfs-3g/attrib.h>
  39 #include <ntfs-3g/misc.h>
  40 #include <ntfs-3g/reparse.h>
  41 #include <ntfs-3g/security.h>
  42 #include <ntfs-3g/volume.h>
  43 #include <stdlib.h>
  44 #include <unistd.h>
  45
  46 extern int ntfs_inode_get_security(ntfs_inode *ni, u32 selection, char *buf,
  47                                    u32 buflen, u32 *psize);
  48
  49 extern int ntfs_inode_get_attributes(ntfs_inode *ni);
  50
  51 /* Structure that allows searching the security descriptors by SHA1 message
  52  * digest. */
  53 struct sd_set {
  54         struct wim_security_data *sd;
  55         struct sd_node *root;
  56 };
  57
  58 /* Binary tree node of security descriptors, indexed by the @hash field. */
  59 struct sd_node {
  60         int security_id;
  61         u8 hash[SHA1_HASH_SIZE];
  62         struct sd_node *left;
  63         struct sd_node *right;
  64 };
  65
  66 /* Frees a security descriptor index tree. */
  67 static void free_sd_set(struct sd_node *root)
  68 {
  69         if (root) {
  70                 free_sd_set(root->left);
  71                 free_sd_set(root->right);
  72                 FREE(root);
  73         }
  74 }
  75
  76 /* Inserts a a new node into the security descriptor index tree. */
  77 static void insert_sd_node(struct sd_node *new, struct sd_node *root)
  78 {
  79         int cmp = hashes_cmp(new->hash, root->hash);
  80         if (cmp < 0) {
  81                 if (root->left)
  82                         insert_sd_node(new, root->left);
  83                 else
  84                         root->left = new;
  85         } else if (cmp > 0) {
  86                 if (root->right)
  87                         insert_sd_node(new, root->right);
  88                 else
  89                         root->right = new;
  90         } else {
  91                 wimlib_assert(0);
  92         }
  93 }
  94
  95 /* Returns the security ID of the security data having a SHA1 message digest of
  96  * @hash in the security descriptor index tree rooted at @root.
  97  *
  98  * If not found, return -1. */
  99 static int lookup_sd(const u8 hash[SHA1_HASH_SIZE], struct sd_node *root)
 100 {
 101         int cmp;
 102         if (!root)
 103                 return -1;
 104         cmp = hashes_cmp(hash, root->hash);
 105         if (cmp < 0)
 106                 return lookup_sd(hash, root->left);
 107         else if (cmp > 0)
 108                 return lookup_sd(hash, root->right);
 109         else
 110                 return root->security_id;
 111 }
 112
 113 /*
 114  * Adds a security descriptor to the indexed security descriptor set as well as
 115  * the corresponding `struct wim_security_data', and returns the new security
 116  * ID; or, if there is an existing security descriptor that is the same, return
 117  * the security ID for it.  If a new security descriptor cannot be allocated,
 118  * return -1.
 119  */
 120 static int sd_set_add_sd(struct sd_set *sd_set, const u8 *descriptor,
 121                          size_t size)
 122 {
 123         u8 hash[SHA1_HASH_SIZE];
 124         int security_id;
 125         struct sd_node *new;
 126         u8 **descriptors;
 127         u64 *sizes;
 128         u8 *descr_copy;
 129         struct wim_security_data *sd;
 130
 131         sha1_buffer(descriptor, size, hash);
 132         security_id = lookup_sd(hash, sd_set->root);
 133         if (security_id >= 0)
 134                 return security_id;
 135
 136         new = MALLOC(sizeof(*new));
 137         if (!new)
 138                 goto out;
 139         descr_copy = MALLOC(size);
 140         if (!descr_copy)
 141                 goto out_free_node;
 142
 143         sd = sd_set->sd;
 144
 145         memcpy(descr_copy, descriptor, size);
 146         new->security_id = sd->num_entries;
 147         new->left = NULL;
 148         new->right = NULL;
 149         copy_hash(new->hash, hash);
 150
 151
 152         descriptors = REALLOC(sd->descriptors,
 153                               (sd->num_entries + 1) * sizeof(sd->descriptors[0]));
 154         if (!descriptors)
 155                 goto out_free_descr;
 156         sd->descriptors = descriptors;
 157         sizes = REALLOC(sd->sizes,
 158                         (sd->num_entries + 1) * sizeof(sd->sizes[0]));
 159         if (!sizes)
 160                 goto out_free_descr;
 161         sd->sizes = sizes;
 162         sd->descriptors[sd->num_entries] = descr_copy;
 163         sd->sizes[sd->num_entries] = size;
 164         sd->num_entries++;
 165         sd->total_length += size + sizeof(sd->sizes[0]);
 166
 167         if (sd_set->root)
 168                 insert_sd_node(sd_set->root, new);
 169         else
 170                 sd_set->root = new;
 171         return new->security_id;
 172 out_free_descr:
 173         FREE(descr_copy);
 174 out_free_node:
 175         FREE(new);
 176 out:
 177         return -1;
 178 }
 179
 180 static inline ntfschar *attr_record_name(ATTR_RECORD *ar)
 181 {
 182         return (ntfschar*)((u8*)ar + le16_to_cpu(ar->name_offset));
 183 }
 184
 185 /* Calculates the SHA1 message digest of a NTFS attribute.
 186  *
 187  * @ni:  The NTFS inode containing the attribute.
 188  * @ar:  The ATTR_RECORD describing the attribute.
 189  * @md:  If successful, the returned SHA1 message digest.
 190  *
 191  * Return 0 on success or nonzero on error.
 192  */
 193 static int ntfs_attr_sha1sum(ntfs_inode *ni, ATTR_RECORD *ar,
 194                              u8 md[SHA1_HASH_SIZE])
 195 {
 196         s64 pos = 0;
 197         s64 bytes_remaining;
 198         char buf[4096];
 199         ntfs_attr *na;
 200         SHA_CTX ctx;
 201
 202         na = ntfs_attr_open(ni, ar->type, attr_record_name(ar),
 203                             ar->name_length);
 204         if (!na) {
 205                 ERROR_WITH_ERRNO("Failed to open NTFS attribute");
 206                 return WIMLIB_ERR_NTFS_3G;
 207         }
 208
 209         bytes_remaining = na->data_size;
 210         sha1_init(&ctx);
 211
 212         DEBUG("Calculating SHA1 message digest (%"PRIu64" bytes)",
 213                         bytes_remaining);
 214
 215         while (bytes_remaining) {
 216                 s64 to_read = min(bytes_remaining, sizeof(buf));
 217                 if (ntfs_attr_pread(na, pos, to_read, buf) != to_read) {
 218                         ERROR_WITH_ERRNO("Error reading NTFS attribute");
 219                         return WIMLIB_ERR_NTFS_3G;
 220                 }
 221                 sha1_update(&ctx, buf, to_read);
 222                 pos += to_read;
 223                 bytes_remaining -= to_read;
 224         }
 225         sha1_final(md, &ctx);
 226         ntfs_attr_close(na);
 227         return 0;
 228 }
 229
 230 /* Load the streams from a WIM file or reparse point in the NTFS volume into the
 231  * WIM lookup table */
 232 static int capture_ntfs_streams(struct dentry *dentry, ntfs_inode *ni,
 233                                 char path[], size_t path_len,
 234                                 struct lookup_table *lookup_table,
 235                                 ntfs_volume **ntfs_vol_p,
 236                                 ATTR_TYPES type)
 237 {
 238
 239         ntfs_attr_search_ctx *actx;
 240         u8 attr_hash[SHA1_HASH_SIZE];
 241         struct ntfs_location *ntfs_loc;
 242         struct lookup_table_entry *lte;
 243         int ret = 0;
 244
 245         DEBUG("Capturing NTFS data streams from `%s'", path);
 246
 247         /* Get context to search the streams of the NTFS file. */
 248         actx = ntfs_attr_get_search_ctx(ni, NULL);
 249         if (!actx) {
 250                 ERROR_WITH_ERRNO("Cannot get attribute search "
 251                                  "context");
 252                 return WIMLIB_ERR_NTFS_3G;
 253         }
 254
 255         /* Capture each data stream or reparse data stream. */
 256         while (!ntfs_attr_lookup(type, NULL, 0,
 257                                  CASE_SENSITIVE, 0, NULL, 0, actx))
 258         {
 259                 char *stream_name_utf8;
 260                 size_t stream_name_utf16_len;
 261
 262                 /* Checksum the stream. */
 263                 ret = ntfs_attr_sha1sum(ni, actx->attr, attr_hash);
 264                 if (ret != 0)
 265                         goto out_put_actx;
 266
 267                 /* Make a lookup table entry for the stream, or use an existing
 268                  * one if there's already an identical stream. */
 269                 lte = __lookup_resource(lookup_table, attr_hash);
 270                 ret = WIMLIB_ERR_NOMEM;
 271                 if (lte) {
 272                         lte->refcnt++;
 273                 } else {
 274                         struct ntfs_location *ntfs_loc;
 275
 276                         ntfs_loc = CALLOC(1, sizeof(*ntfs_loc));
 277                         if (!ntfs_loc)
 278                                 goto out_put_actx;
 279                         ntfs_loc->ntfs_vol_p = ntfs_vol_p;
 280                         ntfs_loc->path_utf8 = MALLOC(path_len + 1);
 281                         if (!ntfs_loc->path_utf8)
 282                                 goto out_free_ntfs_loc;
 283                         memcpy(ntfs_loc->path_utf8, path, path_len + 1);
 284                         ntfs_loc->stream_name_utf16 = MALLOC(actx->attr->name_length * 2);
 285                         if (!ntfs_loc->stream_name_utf16)
 286                                 goto out_free_ntfs_loc;
 287                         memcpy(ntfs_loc->stream_name_utf16,
 288                                attr_record_name(actx->attr),
 289                                actx->attr->name_length * 2);
 290
 291                         ntfs_loc->stream_name_utf16_num_chars = actx->attr->name_length;
 292                         ntfs_loc->is_reparse_point = (type == AT_REPARSE_POINT);
 293                         lte = new_lookup_table_entry();
 294                         if (!lte)
 295                                 goto out_free_ntfs_loc;
 296                         lte->ntfs_loc = ntfs_loc;
 297                         lte->resource_location = RESOURCE_IN_NTFS_VOLUME;
 298                         lte->resource_entry.original_size = actx->attr->data_size;
 299                         lte->resource_entry.size = actx->attr->data_size;
 300                         DEBUG("Add resource for `%s' (size = %zu)",
 301                                 dentry->file_name_utf8,
 302                                 lte->resource_entry.original_size);
 303                         copy_hash(lte->hash, attr_hash);
 304                         lookup_table_insert(lookup_table, lte);
 305                 }
 306                 if (actx->attr->name_length == 0) {
 307                         if (dentry->lte) {
 308                                 ERROR("Found two un-named data streams for "
 309                                       "`%s'", path);
 310                                 ret = WIMLIB_ERR_NTFS_3G;
 311                                 goto out_free_lte;
 312                         }
 313                         dentry->lte = lte;
 314                 } else {
 315                         struct ads_entry *new_ads_entry;
 316                         stream_name_utf8 = utf16_to_utf8((const u8*)attr_record_name(actx->attr),
 317                                                          actx->attr->name_length,
 318                                                          &stream_name_utf16_len);
 319                         if (!stream_name_utf8)
 320                                 goto out_free_lte;
 321                         new_ads_entry = dentry_add_ads(dentry, stream_name_utf8);
 322                         FREE(stream_name_utf8);
 323                         if (!new_ads_entry)
 324                                 goto out_free_lte;
 325
 326                         new_ads_entry->lte = lte;
 327                 }
 328         }
 329         ret = 0;
 330         goto out_put_actx;
 331 out_free_lte:
 332         free_lookup_table_entry(lte);
 333 out_free_ntfs_loc:
 334         if (ntfs_loc) {
 335                 FREE(ntfs_loc->path_utf8);
 336                 FREE(ntfs_loc->stream_name_utf16);
 337                 FREE(ntfs_loc);
 338         }
 339 out_put_actx:
 340         ntfs_attr_put_search_ctx(actx);
 341         if (ret == 0)
 342                 DEBUG("Successfully captured NTFS streams from `%s'", path);
 343         else
 344                 DEBUG("Failed to capture NTFS streams from `%s", path);
 345         return ret;
 346 }
 347
 348 struct readdir_ctx {
 349         struct dentry       *parent;
 350         ntfs_inode          *dir_ni;
 351         char                *path;
 352         size_t               path_len;
 353         struct lookup_table *lookup_table;
 354         struct sd_set       *sd_set;
 355         const struct capture_config *config;
 356         ntfs_volume        **ntfs_vol_p;
 357 };
 358
 359 static int __build_dentry_tree_ntfs(struct dentry **root_p, ntfs_inode *ni,
 360                                     char path[], size_t path_len,
 361                                     struct lookup_table *lookup_table,
 362                                     struct sd_set *sd_set,
 363                                     const struct capture_config *config,
 364                                     ntfs_volume **ntfs_vol_p);
 365
 366
 367 static int wim_ntfs_capture_filldir(void *dirent, const ntfschar *name,
 368                                     const int name_len, const int name_type,
 369                                     const s64 pos, const MFT_REF mref,
 370                                     const unsigned dt_type)
 371 {
 372         struct readdir_ctx *ctx;
 373         size_t utf8_name_len;
 374         char *utf8_name;
 375         struct dentry *child = NULL;
 376         int ret;
 377         size_t path_len;
 378
 379         if (name_type == FILE_NAME_DOS)
 380                 return 0;
 381
 382         ret = -1;
 383
 384         utf8_name = utf16_to_utf8((const u8*)name, name_len * 2,
 385                                   &utf8_name_len);
 386         if (!utf8_name)
 387                 goto out;
 388
 389         if (utf8_name[0] == '.' &&
 390              (utf8_name[1] == '\0' ||
 391               (utf8_name[1] == '.' && utf8_name[2] == '\0'))) {
 392                 DEBUG("Skipping dentry `%s'", utf8_name);
 393                 ret = 0;
 394                 goto out_free_utf8_name;
 395         }
 396
 397         DEBUG("Opening inode for `%s'", utf8_name);
 398
 399         ctx = dirent;
 400
 401         ntfs_inode *ni = ntfs_inode_open(ctx->dir_ni->vol, mref);
 402         if (!ni) {
 403                 ERROR_WITH_ERRNO("Failed to open NTFS inode");
 404                 ret = 1;
 405         }
 406         path_len = ctx->path_len;
 407         if (path_len != 1)
 408                 ctx->path[path_len++] = '/';
 409         memcpy(ctx->path + path_len, utf8_name, utf8_name_len + 1);
 410         path_len += utf8_name_len;
 411         ret = __build_dentry_tree_ntfs(&child, ni, ctx->path, path_len,
 412                                        ctx->lookup_table, ctx->sd_set,
 413                                        ctx->config, ctx->ntfs_vol_p);
 414
 415         if (child) {
 416                 DEBUG("Linking dentry `%s' with parent `%s'",
 417                       child->file_name_utf8, ctx->parent->file_name_utf8);
 418                 link_dentry(child, ctx->parent);
 419         }
 420 out_close_ni:
 421         ntfs_inode_close(ni);
 422 out_free_utf8_name:
 423         FREE(utf8_name);
 424 out:
 425         return ret;
 426 }
 427
 428 /* Recursively build a WIM dentry tree corresponding to a NTFS volume.
 429  * At the same time, update the WIM lookup table with lookup table entries for
 430  * the NTFS streams, and build an array of security descriptors.
 431  */
 432 static int __build_dentry_tree_ntfs(struct dentry **root_p, ntfs_inode *ni,
 433                                     char path[], size_t path_len,
 434                                     struct lookup_table *lookup_table,
 435                                     struct sd_set *sd_set,
 436                                     const struct capture_config *config,
 437                                     ntfs_volume **ntfs_vol_p)
 438 {
 439         u32 attributes;
 440         int mrec_flags;
 441         u32 sd_size;
 442         int ret = 0;
 443         struct dentry *root;
 444
 445         if (exclude_path(path, config, false)) {
 446                 DEBUG("Excluding `%s' from capture", path);
 447                 return 0;
 448         }
 449
 450         DEBUG("Starting recursive capture at path = `%s'", path);
 451         mrec_flags = ni->mrec->flags;
 452         attributes = ntfs_inode_get_attributes(ni);
 453
 454         root = new_dentry(path_basename(path));
 455         if (!root)
 456                 return WIMLIB_ERR_NOMEM;
 457
 458         root->creation_time    = le64_to_cpu(ni->creation_time);
 459         root->last_write_time  = le64_to_cpu(ni->last_data_change_time);
 460         root->last_access_time = le64_to_cpu(ni->last_access_time);
 461         root->security_id      = le32_to_cpu(ni->security_id);
 462         root->attributes       = le32_to_cpu(attributes);
 463         root->hard_link  = ni->mft_no;
 464         root->resolved = true;
 465
 466         if (attributes & FILE_ATTR_REPARSE_POINT) {
 467                 DEBUG("Reparse point `%s'", path);
 468                 /* Junction point, symbolic link, or other reparse point */
 469                 ret = capture_ntfs_streams(root, ni, path, path_len,
 470                                            lookup_table, ntfs_vol_p,
 471                                            AT_REPARSE_POINT);
 472         } else if (mrec_flags & MFT_RECORD_IS_DIRECTORY) {
 473                 DEBUG("Directory `%s'", path);
 474
 475                 /* Normal directory */
 476                 s64 pos = 0;
 477                 struct readdir_ctx ctx = {
 478                         .parent       = root,
 479                         .dir_ni       = ni,
 480                         .path         = path,
 481                         .path_len     = path_len,
 482                         .lookup_table = lookup_table,
 483                         .sd_set       = sd_set,
 484                         .config       = config,
 485                         .ntfs_vol_p   = ntfs_vol_p,
 486                 };
 487                 ret = ntfs_readdir(ni, &pos, &ctx, wim_ntfs_capture_filldir);
 488                 if (ret != 0) {
 489                         ERROR_WITH_ERRNO("ntfs_readdir()");
 490                         ret = WIMLIB_ERR_NTFS_3G;
 491                 }
 492         } else {
 493                 DEBUG("Normal file `%s'", path);
 494                 /* Normal file */
 495                 ret = capture_ntfs_streams(root, ni, path, path_len,
 496                                            lookup_table, ntfs_vol_p,
 497                                            AT_DATA);
 498         }
 499         if (ret != 0)
 500                 return ret;
 501
 502         ret = ntfs_inode_get_security(ni,
 503                                       OWNER_SECURITY_INFORMATION |
 504                                       GROUP_SECURITY_INFORMATION |
 505                                       DACL_SECURITY_INFORMATION  |
 506                                       SACL_SECURITY_INFORMATION,
 507                                       NULL, 0, &sd_size);
 508         u8 sd[sd_size];
 509         ret = ntfs_inode_get_security(ni,
 510                                       OWNER_SECURITY_INFORMATION |
 511                                       GROUP_SECURITY_INFORMATION |
 512                                       DACL_SECURITY_INFORMATION  |
 513                                       SACL_SECURITY_INFORMATION,
 514                                       sd, sd_size, &sd_size);
 515         if (ret == 0) {
 516                 ERROR_WITH_ERRNO("Failed to get security information from "
 517                                  "`%s'", path);
 518                 ret = WIMLIB_ERR_NTFS_3G;
 519         } else {
 520                 if (ret > 0) {
 521                         /*print_security_descriptor(sd, sd_size);*/
 522                         root->security_id = sd_set_add_sd(sd_set, sd, sd_size);
 523                         DEBUG("Added security ID = %u for `%s'",
 524                               root->security_id, path);
 525                 } else {
 526                         root->security_id = -1;
 527                         DEBUG("No security ID for `%s'", path);
 528                 }
 529                 ret = 0;
 530         }
 531         *root_p = root;
 532         return ret;
 533 }
 534
 535 static int build_dentry_tree_ntfs(struct dentry **root_p,
 536                                   const char *device,
 537                                   struct lookup_table *lookup_table,
 538                                   struct wim_security_data *sd,
 539                                   const struct capture_config *config,
 540                                   int flags,
 541                                   void *extra_arg)
 542 {
 543         ntfs_volume *vol;
 544         ntfs_inode *root_ni;
 545         int ret = 0;
 546         struct sd_set tree;
 547         tree.sd = sd;
 548         tree.root = NULL;
 549         ntfs_volume **ntfs_vol_p = extra_arg;
 550
 551         DEBUG("Mounting NTFS volume `%s' read-only", device);
 552
 553         vol = ntfs_mount(device, MS_RDONLY);
 554         if (!vol) {
 555                 ERROR_WITH_ERRNO("Failed to mount NTFS volume `%s' read-only",
 556                                  device);
 557                 return WIMLIB_ERR_NTFS_3G;
 558         }
 559
 560         NVolClearShowSysFiles(vol);
 561
 562         DEBUG("Opening root NTFS dentry");
 563         root_ni = ntfs_inode_open(vol, FILE_root);
 564         if (!root_ni) {
 565                 ERROR_WITH_ERRNO("Failed to open root inode of NTFS volume "
 566                                  "`%s'", device);
 567                 ret = WIMLIB_ERR_NTFS_3G;
 568                 goto out;
 569         }
 570         char path[4096];
 571         path[0] = '/';
 572         path[1] = '\0';
 573         ret = __build_dentry_tree_ntfs(root_p, root_ni, path, 1,
 574                                        lookup_table, &tree, config,
 575                                        ntfs_vol_p);
 576         ntfs_inode_close(root_ni);
 577
 578 out:
 579         if (ret) {
 580                 if (ntfs_umount(vol, FALSE) != 0) {
 581                         ERROR_WITH_ERRNO("Failed to unmount NTFS volume `%s'",
 582                                          device);
 583                         if (ret == 0)
 584                                 ret = WIMLIB_ERR_NTFS_3G;
 585                 }
 586         } else {
 587                 *ntfs_vol_p = vol;
 588         }
 589         return ret;
 590 }
 591
 592
 593
 594 WIMLIBAPI int wimlib_add_image_from_ntfs_volume(WIMStruct *w,
 595                                                 const char *device,
 596                                                 const char *name,
 597                                                 const char *config_str,
 598                                                 size_t config_len,
 599                                                 int flags)
 600 {
 601         if (flags & (WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE)) {
 602                 ERROR("Cannot dereference files when capturing directly from NTFS");
 603                 return WIMLIB_ERR_INVALID_PARAM;
 604         }
 605         return do_add_image(w, device, name, config_str, config_len, flags,
 606                             build_dentry_tree_ntfs, &w->ntfs_vol);
 607 }
 608
 609 #else /* WITH_NTFS_3G */
 610 WIMLIBAPI int wimlib_add_image_from_ntfs_volume(WIMStruct *w,
 611                                                 const char *device,
 612                                                 const char *name,
 613                                                 const char *description,
 614                                                 const char *flags_element,
 615                                                 int flags,
 616                                                 const char *config_str,
 617                                                 size_t config_len)
 618 {
 619         ERROR("wimlib was compiled without support for NTFS-3g, so");
 620         ERROR("we cannot capture a WIM image directly from a NTFS volume");
 621         return WIMLIB_ERR_UNSUPPORTED;
 622 }
 623 #endif /* WITH_NTFS_3G */