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         while (bytes_remaining) {
 213                 s64 to_read = min(bytes_remaining, sizeof(buf));
 214                 if (ntfs_attr_pread(na, pos, to_read, buf) != to_read) {
 215                         ERROR_WITH_ERRNO("Error reading NTFS attribute");
 216                         return WIMLIB_ERR_NTFS_3G;
 217                 }
 218                 sha1_update(&ctx, buf, to_read);
 219                 pos += to_read;
 220                 bytes_remaining -= to_read;
 221         }
 222         sha1_final(md, &ctx);
 223         ntfs_attr_close(na);
 224         return 0;
 225 }
 226
 227 /* Load the streams from a WIM file or reparse point in the NTFS volume into the
 228  * WIM lookup table */
 229 static int capture_ntfs_streams(struct dentry *dentry, ntfs_inode *ni,
 230                                 char path[], size_t path_len,
 231                                 struct lookup_table *lookup_table,
 232                                 ntfs_volume **ntfs_vol_p,
 233                                 ATTR_TYPES type)
 234 {
 235
 236         ntfs_attr_search_ctx *actx;
 237         u8 attr_hash[SHA1_HASH_SIZE];
 238         struct ntfs_location *ntfs_loc;
 239         struct lookup_table_entry *lte;
 240         int ret = 0;
 241
 242         /* Get context to search the streams of the NTFS file. */
 243         actx = ntfs_attr_get_search_ctx(ni, NULL);
 244         if (!actx) {
 245                 ERROR_WITH_ERRNO("Cannot get attribute search "
 246                                  "context");
 247                 return WIMLIB_ERR_NTFS_3G;
 248         }
 249
 250         /* Capture each data stream or reparse data stream. */
 251         while (!ntfs_attr_lookup(type, NULL, 0,
 252                                  CASE_SENSITIVE, 0, NULL, 0, actx))
 253         {
 254                 char *stream_name_utf8;
 255                 size_t stream_name_utf16_len;
 256
 257                 /* Checksum the stream. */
 258                 ret = ntfs_attr_sha1sum(ni, actx->attr, attr_hash);
 259                 if (ret != 0)
 260                         goto out_put_actx;
 261
 262                 /* Make a lookup table entry for the stream, or use an existing
 263                  * one if there's already an identical stream. */
 264                 lte = __lookup_resource(lookup_table, attr_hash);
 265                 ret = WIMLIB_ERR_NOMEM;
 266                 if (lte) {
 267                         lte->refcnt++;
 268                 } else {
 269                         struct ntfs_location *ntfs_loc;
 270
 271                         ntfs_loc = CALLOC(1, sizeof(*ntfs_loc));
 272                         if (!ntfs_loc)
 273                                 goto out_put_actx;
 274                         ntfs_loc->ntfs_vol_p = ntfs_vol_p;
 275                         ntfs_loc->path_utf8 = MALLOC(path_len + 1);
 276                         if (!ntfs_loc->path_utf8)
 277                                 goto out_free_ntfs_loc;
 278                         memcpy(ntfs_loc->path_utf8, path, path_len + 1);
 279                         ntfs_loc->stream_name_utf16 = MALLOC(actx->attr->name_length * 2);
 280                         if (!ntfs_loc->stream_name_utf16)
 281                                 goto out_free_ntfs_loc;
 282                         memcpy(ntfs_loc->stream_name_utf16,
 283                                attr_record_name(actx->attr),
 284                                actx->attr->name_length * 2);
 285
 286                         ntfs_loc->stream_name_utf16_num_chars = actx->attr->name_length;
 287                         ntfs_loc->is_reparse_point = (type == AT_REPARSE_POINT);
 288                         lte = new_lookup_table_entry();
 289                         if (!lte)
 290                                 goto out_free_ntfs_loc;
 291                         lte->ntfs_loc = ntfs_loc;
 292                         lte->resource_location = RESOURCE_IN_NTFS_VOLUME;
 293                         lte->resource_entry.original_size = actx->attr->data_size;
 294                         lte->resource_entry.size = actx->attr->data_size;
 295                         copy_hash(lte->hash, attr_hash);
 296                         lookup_table_insert(lookup_table, lte);
 297                 }
 298                 if (actx->attr->name_length == 0) {
 299                         if (dentry->lte) {
 300                                 ERROR("Found two un-named data streams for "
 301                                       "`%s'", path);
 302                                 ret = WIMLIB_ERR_NTFS_3G;
 303                                 goto out_free_lte;
 304                         }
 305                         dentry->lte = lte;
 306                 } else {
 307                         struct ads_entry *new_ads_entry;
 308                         stream_name_utf8 = utf16_to_utf8((const u8*)attr_record_name(actx->attr),
 309                                                          actx->attr->name_length,
 310                                                          &stream_name_utf16_len);
 311                         if (!stream_name_utf8)
 312                                 goto out_free_lte;
 313                         new_ads_entry = dentry_add_ads(dentry, stream_name_utf8);
 314                         FREE(stream_name_utf8);
 315                         if (!new_ads_entry)
 316                                 goto out_free_lte;
 317
 318                         new_ads_entry->lte = lte;
 319                 }
 320         }
 321         ret = 0;
 322         goto out_put_actx;
 323 out_free_lte:
 324         free_lookup_table_entry(lte);
 325 out_free_ntfs_loc:
 326         if (ntfs_loc) {
 327                 FREE(ntfs_loc->path_utf8);
 328                 FREE(ntfs_loc->stream_name_utf16);
 329                 FREE(ntfs_loc);
 330         }
 331 out_put_actx:
 332         ntfs_attr_put_search_ctx(actx);
 333         return ret;
 334 }
 335
 336 struct readdir_ctx {
 337         struct dentry       *dentry;
 338         ntfs_inode          *dir_ni;
 339         char                *path;
 340         size_t               path_len;
 341         struct lookup_table *lookup_table;
 342         struct sd_set       *sd_set;
 343         ntfs_volume        **ntfs_vol_p;
 344 };
 345
 346 static int __build_dentry_tree_ntfs(struct dentry *dentry, ntfs_inode *ni,
 347                                     char path[], size_t path_len,
 348                                     struct lookup_table *lookup_table,
 349                                     struct sd_set *sd_set,
 350                                     ntfs_volume **ntfs_vol_p);
 351
 352
 353 static int filldir(void *dirent, const ntfschar *name,
 354                    const int name_len, const int name_type, const s64 pos,
 355                    const MFT_REF mref, const unsigned dt_type)
 356 {
 357         struct readdir_ctx *ctx;
 358         size_t utf8_name_len;
 359         char *utf8_name;
 360         struct dentry *child;
 361         int ret;
 362         size_t path_len;
 363
 364         ret = -1;
 365
 366         utf8_name = utf16_to_utf8((const u8*)name, name_len * 2,
 367                                   &utf8_name_len);
 368         if (!utf8_name)
 369                 goto out;
 370
 371         ctx = dirent;
 372
 373         ntfs_inode *ni = ntfs_inode_open(ctx->dir_ni->vol, mref);
 374         if (!ni) {
 375                 ERROR_WITH_ERRNO("Failed to open NTFS inode");
 376                 ret = 1;
 377         }
 378         child = new_dentry(utf8_name);
 379         if (!child)
 380                 goto out_close_ni;
 381
 382         memcpy(ctx->path + ctx->path_len, utf8_name, utf8_name_len + 1);
 383         path_len = ctx->path_len + utf8_name_len;
 384         ret = __build_dentry_tree_ntfs(child, ni, ctx->path, path_len,
 385                                        ctx->lookup_table, ctx->sd_set,
 386                                        ctx->ntfs_vol_p);
 387         link_dentry(child, ctx->dentry);
 388 out_close_ni:
 389         ntfs_inode_close(ni);
 390 out_free_utf8_name:
 391         FREE(utf8_name);
 392 out:
 393         return ret;
 394 }
 395
 396 /* Recursively build a WIM dentry tree corresponding to a NTFS volume.
 397  * At the same time, update the WIM lookup table with lookup table entries for
 398  * the NTFS streams, and build an array of security descriptors.
 399  */
 400 static int __build_dentry_tree_ntfs(struct dentry *dentry, ntfs_inode *ni,
 401                                     char path[], size_t path_len,
 402                                     struct lookup_table *lookup_table,
 403                                     struct sd_set *sd_set,
 404                                     ntfs_volume **ntfs_vol_p)
 405 {
 406         u32 attributes = ntfs_inode_get_attributes(ni);
 407         int mrec_flags = ni->mrec->flags;
 408         u32 sd_size;
 409         int ret = 0;
 410
 411         dentry->creation_time    = le64_to_cpu(ni->creation_time);
 412         dentry->last_write_time  = le64_to_cpu(ni->last_data_change_time);
 413         dentry->last_access_time = le64_to_cpu(ni->last_access_time);
 414         dentry->security_id      = le32_to_cpu(ni->security_id);
 415         dentry->attributes       = le32_to_cpu(attributes);
 416         dentry->hard_link        = ni->mft_no;
 417         dentry->resolved = true;
 418
 419         if (attributes & FILE_ATTR_REPARSE_POINT) {
 420                 /* Junction point, symbolic link, or other reparse point */
 421                 ret = capture_ntfs_streams(dentry, ni, path, path_len,
 422                                            lookup_table, ntfs_vol_p,
 423                                            AT_REPARSE_POINT);
 424         } else if (mrec_flags & MFT_RECORD_IS_DIRECTORY) {
 425                 /* Normal directory */
 426                 s64 pos = 0;
 427                 struct readdir_ctx ctx = {
 428                         .dentry       = dentry,
 429                         .dir_ni       = ni,
 430                         .path         = path,
 431                         .path_len     = path_len,
 432                         .lookup_table = lookup_table,
 433                         .sd_set       = sd_set,
 434                         .ntfs_vol_p   = ntfs_vol_p,
 435                 };
 436                 ret = ntfs_readdir(ni, &pos, &ctx, filldir);
 437                 if (ret != 0)
 438                         ret = WIMLIB_ERR_NTFS_3G;
 439         } else {
 440                 /* Normal file */
 441                 ret = capture_ntfs_streams(dentry, ni, path, path_len,
 442                                            lookup_table, ntfs_vol_p,
 443                                            AT_DATA);
 444         }
 445         if (ret != 0)
 446                 return ret;
 447         ret = ntfs_inode_get_security(ni,
 448                                       OWNER_SECURITY_INFORMATION |
 449                                       GROUP_SECURITY_INFORMATION |
 450                                       DACL_SECURITY_INFORMATION  |
 451                                       SACL_SECURITY_INFORMATION,
 452                                       NULL, 0, &sd_size);
 453         u8 sd[sd_size];
 454         ret = ntfs_inode_get_security(ni,
 455                                       OWNER_SECURITY_INFORMATION |
 456                                       GROUP_SECURITY_INFORMATION |
 457                                       DACL_SECURITY_INFORMATION  |
 458                                       SACL_SECURITY_INFORMATION,
 459                                       sd, sd_size, &sd_size);
 460         dentry->security_id = sd_set_add_sd(sd_set, sd, sd_size);
 461         if (dentry->security_id == -1) {
 462                 ERROR("Could not allocate security ID");
 463                 ret = WIMLIB_ERR_NOMEM;
 464         }
 465         return ret;
 466 }
 467
 468 static int build_dentry_tree_ntfs(struct dentry *root_dentry,
 469                                   const char *device,
 470                                   struct lookup_table *lookup_table,
 471                                   struct wim_security_data *sd,
 472                                   int flags,
 473                                   void *extra_arg)
 474 {
 475         ntfs_volume *vol;
 476         ntfs_inode *root_ni;
 477         int ret = 0;
 478         struct sd_set tree;
 479         tree.sd = sd;
 480         tree.root = NULL;
 481         ntfs_volume **ntfs_vol_p = extra_arg;
 482
 483         vol = ntfs_mount(device, MS_RDONLY);
 484         if (!vol) {
 485                 ERROR_WITH_ERRNO("Failed to mount NTFS volume `%s' read-only",
 486                                  device);
 487                 return WIMLIB_ERR_NTFS_3G;
 488         }
 489         root_ni = ntfs_inode_open(vol, FILE_root);
 490         if (!root_ni) {
 491                 ERROR_WITH_ERRNO("Failed to open root inode of NTFS volume "
 492                                  "`%s'", device);
 493                 ret = WIMLIB_ERR_NTFS_3G;
 494                 goto out;
 495         }
 496         char path[4096];
 497         path[0] = '/';
 498         path[1] = '\0';
 499         ret = __build_dentry_tree_ntfs(root_dentry, root_ni, path, 1,
 500                                        lookup_table, &tree, ntfs_vol_p);
 501         ntfs_inode_close(root_ni);
 502
 503 out:
 504         if (ntfs_umount(vol, FALSE) != 0) {
 505                 ERROR_WITH_ERRNO("Failed to unmount NTFS volume `%s'", device);
 506                 if (ret == 0)
 507                         ret = WIMLIB_ERR_NTFS_3G;
 508         }
 509         return ret;
 510 }
 511
 512 WIMLIBAPI int wimlib_add_image_from_ntfs_volume(WIMStruct *w,
 513                                                 const char *device,
 514                                                 const char *name,
 515                                                 const char *description,
 516                                                 const char *flags_element,
 517                                                 int flags)
 518 {
 519         if (flags & (WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE)) {
 520                 ERROR("Cannot dereference files when capturing directly from NTFS");
 521                 return WIMLIB_ERR_INVALID_PARAM;
 522         }
 523         return do_add_image(w, device, name, description, flags_element, flags,
 524                             build_dentry_tree_ntfs,
 525                             &w->ntfs_vol);
 526 }
 527
 528 #else /* WITH_NTFS_3G */
 529 WIMLIBAPI int wimlib_add_image_from_ntfs_volume(WIMStruct *w,
 530                                                 const char *device,
 531                                                 const char *name,
 532                                                 const char *description,
 533                                                 const char *flags_element,
 534                                                 int flags)
 535 {
 536         ERROR("wimlib was compiled without support for NTFS-3g, so");
 537         ERROR("we cannot capture a WIM image directly from a NTFS volume");
 538         return WIMLIB_ERR_UNSUPPORTED;
 539 }
 540 #endif /* WITH_NTFS_3G */