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 struct sd_tree {
  52         u32 num_sds;
  53         struct wim_security_data *sd;
  54         struct sd_node *root;
  55 };
  56
  57 struct sd_node {
  58         int security_id;
  59         u8 hash[SHA1_HASH_SIZE];
  60         struct sd_node *left;
  61         struct sd_node *right;
  62 };
  63
  64 static void free_sd_tree(struct sd_node *root)
  65 {
  66         if (root) {
  67                 free_sd_tree(root->left);
  68                 free_sd_tree(root->right);
  69                 FREE(root);
  70         }
  71 }
  72
  73 static void insert_sd_node(struct sd_node *new, struct sd_node *root)
  74 {
  75         int cmp = hashes_cmp(root->hash, new->hash);
  76         if (cmp < 0) {
  77                 if (root->left)
  78                         insert_sd_node(new, root->left);
  79                 else
  80                         root->left = new;
  81         } else if (cmp > 0) {
  82                 if (root->right)
  83                         insert_sd_node(new, root->right);
  84                 else
  85                         root->right = new;
  86         } else {
  87                 wimlib_assert(0);
  88         }
  89 }
  90
  91 static int lookup_sd(const u8 hash[SHA1_HASH_SIZE], struct sd_node *node)
  92 {
  93         int cmp;
  94         if (!node)
  95                 return -1;
  96         cmp = hashes_cmp(hash, node->hash);
  97         if (cmp < 0)
  98                 return lookup_sd(hash, node->left);
  99         else if (cmp > 0)
 100                 return lookup_sd(hash, node->right);
 101         else
 102                 return node->security_id;
 103 }
 104
 105 static int tree_add_sd(struct sd_tree *tree, const u8 *descriptor,
 106                        size_t size)
 107 {
 108         u8 hash[SHA1_HASH_SIZE];
 109         int security_id;
 110         struct sd_node *new;
 111         u8 **descriptors;
 112         u64 *sizes;
 113         u8 *descr_copy;
 114         struct wim_security_data *sd = tree->sd;
 115         sha1_buffer(descriptor, size, hash);
 116
 117         security_id = lookup_sd(hash, tree->root);
 118         if (security_id >= 0)
 119                 return security_id;
 120
 121         new = MALLOC(sizeof(struct sd_node));
 122         if (!new)
 123                 return -1;
 124         descr_copy = MALLOC(size);
 125         if (!descr_copy)
 126                 goto out_free_node;
 127         memcpy(descr_copy, descriptor, size);
 128         new->security_id = tree->num_sds++;
 129         new->left = NULL;
 130         new->right = NULL;
 131         copy_hash(new->hash, hash);
 132
 133         descriptors = REALLOC(sd->descriptors,
 134                               (sd->num_entries + 1) * sizeof(sd->descriptors[0]));
 135         if (!descriptors)
 136                 goto out_free_descr;
 137         sd->descriptors = descriptors;
 138         sizes = REALLOC(sd->sizes,
 139                         (sd->num_entries + 1) * sizeof(sd->sizes[0]));
 140         if (!sizes)
 141                 goto out_free_descr;
 142         sd->sizes = sizes;
 143         sd->descriptors[sd->num_entries] = descr_copy;
 144         sd->sizes[sd->num_entries] = size;
 145         sd->num_entries++;
 146         sd->total_length += size + 8;
 147
 148         if (tree->root)
 149                 insert_sd_node(tree->root, new);
 150         else
 151                 tree->root = new;
 152         return new->security_id;
 153 out_free_descr:
 154         FREE(descr_copy);
 155 out_free_node:
 156         FREE(new);
 157         return -1;
 158 }
 159
 160 #if 0
 161 static int build_sd_tree(struct wim_security_data *sd, struct sd_tree *tree)
 162 {
 163         int ret;
 164         u32 orig_num_entries = sd->num_entries;
 165         u32 orig_total_length = sd->total_length;
 166
 167         tree->num_sds = 0;
 168         tree->sd = sd;
 169         tree->root = NULL;
 170
 171         for (u32 i = 0; i < sd->num_entries; i++) {
 172                 ret = tree_add_sd(tree, sd->descriptors[i], sd->sizes[i]);
 173                 if (ret < 0)
 174                         goto out_revert;
 175         }
 176         return 0;
 177 out_revert:
 178         sd->num_entries = orig_num_entries;
 179         sd->total_length = orig_total_length;
 180         free_sd_tree(tree->root);
 181         return ret;
 182 }
 183 #endif
 184
 185 static int ntfs_attr_sha1sum(ntfs_inode *ni, ATTR_RECORD *ar,
 186                              u8 md[SHA1_HASH_SIZE])
 187 {
 188         s64 pos = 0;
 189         s64 bytes_remaining;
 190         char buf[4096];
 191         ntfs_attr *na;
 192         SHA_CTX ctx;
 193
 194         na = ntfs_attr_open(ni, ar->type,
 195                             (ntfschar*)((u8*)ar + le16_to_cpu(ar->name_offset)),
 196                             ar->name_length);
 197         if (!na) {
 198                 ERROR_WITH_ERRNO("Failed to open NTFS attribute");
 199                 return WIMLIB_ERR_NTFS_3G;
 200         }
 201
 202         bytes_remaining = na->data_size;
 203         sha1_init(&ctx);
 204
 205         while (bytes_remaining) {
 206                 s64 to_read = min(bytes_remaining, sizeof(buf));
 207                 if (ntfs_attr_pread(na, pos, to_read, buf) != to_read) {
 208                         ERROR_WITH_ERRNO("Error reading NTFS attribute");
 209                         return WIMLIB_ERR_NTFS_3G;
 210                 }
 211                 sha1_update(&ctx, buf, to_read);
 212                 pos += to_read;
 213                 bytes_remaining -= to_read;
 214         }
 215         sha1_final(md, &ctx);
 216         ntfs_attr_close(na);
 217         return 0;
 218 }
 219
 220 /* Load the streams from a WIM file or reparse point in the NTFS volume into the
 221  * WIM lookup table */
 222 static int capture_ntfs_streams(struct dentry *dentry, ntfs_inode *ni,
 223                                 char path[], size_t path_len,
 224                                 struct lookup_table *lookup_table,
 225                                 ntfs_volume **ntfs_vol_p,
 226                                 ATTR_TYPES type)
 227 {
 228
 229         ntfs_attr_search_ctx *actx;
 230         u8 attr_hash[SHA1_HASH_SIZE];
 231         struct ntfs_location *ntfs_loc;
 232         struct lookup_table_entry *lte;
 233         int ret = 0;
 234
 235         /* Get context to search the streams of the NTFS file. */
 236         actx = ntfs_attr_get_search_ctx(ni, NULL);
 237         if (!actx) {
 238                 ERROR_WITH_ERRNO("Cannot get attribute search "
 239                                  "context");
 240                 return WIMLIB_ERR_NTFS_3G;
 241         }
 242
 243         /* Capture each data stream or reparse data stream. */
 244         while (!ntfs_attr_lookup(type, NULL, 0,
 245                                  CASE_SENSITIVE, 0, NULL, 0, actx))
 246         {
 247                 char *stream_name_utf8;
 248                 size_t stream_name_utf16_len;
 249
 250                 /* Checksum the stream. */
 251                 ret = ntfs_attr_sha1sum(ni, actx->attr, attr_hash);
 252                 if (ret != 0)
 253                         goto out_put_actx;
 254
 255                 /* Make a lookup table entry for the stream, or use an existing
 256                  * one if there's already an identical stream. */
 257                 lte = __lookup_resource(lookup_table, attr_hash);
 258                 ret = WIMLIB_ERR_NOMEM;
 259                 if (lte) {
 260                         lte->refcnt++;
 261                 } else {
 262                         struct ntfs_location *ntfs_loc;
 263
 264
 265                         ntfs_loc = CALLOC(1, sizeof(*ntfs_loc));
 266                         if (!ntfs_loc) {
 267                                 goto out_put_actx;
 268                         }
 269                         ntfs_loc->ntfs_vol_p = ntfs_vol_p;
 270                         ntfs_loc->path_utf8 = MALLOC(path_len + 1);
 271                         if (!ntfs_loc->path_utf8)
 272                                 goto out_put_actx;
 273                         memcpy(ntfs_loc->path_utf8, path, path_len + 1);
 274                         ntfs_loc->stream_name_utf16 = MALLOC(actx->attr->name_length * 2);
 275                         if (!ntfs_loc->stream_name_utf16)
 276                                 goto out_free_ntfs_loc;
 277                         memcpy(ntfs_loc->stream_name_utf16,
 278                                (u8*)actx->attr +
 279                                         le16_to_cpu(actx->attr->name_offset),
 280                                actx->attr->name_length * 2);
 281
 282                         ntfs_loc->stream_name_utf16_num_chars = actx->attr->name_length;
 283                         lte = new_lookup_table_entry();
 284                         if (!lte)
 285                                 goto out_free_ntfs_loc;
 286                         lte->ntfs_loc = ntfs_loc;
 287                         lte->resource_location = RESOURCE_IN_NTFS_VOLUME;
 288                         lte->resource_entry.original_size = actx->attr->data_size;
 289                         lte->resource_entry.size = actx->attr->data_size;
 290                         copy_hash(lte->hash, attr_hash);
 291                         lookup_table_insert(lookup_table, lte);
 292                 }
 293                 if (actx->attr->name_length == 0) {
 294                         wimlib_assert(!dentry->lte);
 295                         dentry->lte = lte;
 296                 } else {
 297                         struct ads_entry *new_ads_entry;
 298                         stream_name_utf8 = utf16_to_utf8((u8*)actx->attr +
 299                                                          le16_to_cpu(actx->attr->name_offset),
 300                                                          actx->attr->name_length,
 301                                                          &stream_name_utf16_len);
 302                         if (!stream_name_utf8)
 303                                 goto out_free_lte;
 304                         FREE(stream_name_utf8);
 305                         new_ads_entry = dentry_add_ads(dentry, stream_name_utf8);
 306                         if (!new_ads_entry)
 307                                 goto out_free_lte;
 308
 309                         new_ads_entry->lte = lte;
 310                 }
 311         }
 312         ret = 0;
 313         goto out_put_actx;
 314 out_free_lte:
 315         free_lookup_table_entry(lte);
 316 out_free_ntfs_loc:
 317         if (ntfs_loc) {
 318                 FREE(ntfs_loc->path_utf8);
 319                 FREE(ntfs_loc->stream_name_utf16);
 320                 FREE(ntfs_loc);
 321         }
 322 out_put_actx:
 323         ntfs_attr_put_search_ctx(actx);
 324         return ret;
 325 }
 326
 327 struct readdir_ctx {
 328         struct dentry       *dentry;
 329         ntfs_inode          *dir_ni;
 330         char                *path;
 331         size_t               path_len;
 332         struct lookup_table *lookup_table;
 333         struct sd_tree      *tree;
 334         ntfs_volume        **ntfs_vol_p;
 335 };
 336
 337 static int __build_dentry_tree_ntfs(struct dentry *dentry, ntfs_inode *ni,
 338                                     char path[], size_t path_len,
 339                                     struct lookup_table *lookup_table,
 340                                     struct sd_tree *tree,
 341                                     ntfs_volume **ntfs_vol_p);
 342
 343
 344 static int filldir(void *dirent, const ntfschar *name,
 345                    const int name_len, const int name_type, const s64 pos,
 346                    const MFT_REF mref, const unsigned dt_type)
 347 {
 348         struct readdir_ctx *ctx;
 349         size_t utf8_name_len;
 350         char *utf8_name;
 351         struct dentry *child;
 352         int ret;
 353         size_t path_len;
 354
 355         ret = -1;
 356
 357         utf8_name = utf16_to_utf8((const u8*)name, name_len * 2,
 358                                   &utf8_name_len);
 359         if (!utf8_name)
 360                 goto out;
 361
 362         ctx = dirent;
 363
 364         ntfs_inode *ni = ntfs_inode_open(ctx->dir_ni->vol, mref);
 365         if (!ni) {
 366                 ERROR_WITH_ERRNO("Failed to open NTFS inode");
 367                 ret = 1;
 368         }
 369         child = new_dentry(utf8_name);
 370         if (!child)
 371                 goto out_close_ni;
 372
 373         memcpy(ctx->path + ctx->path_len, utf8_name, utf8_name_len + 1);
 374         path_len = ctx->path_len + utf8_name_len;
 375         ret = __build_dentry_tree_ntfs(child, ni, ctx->path, path_len,
 376                                        ctx->lookup_table, ctx->tree,
 377                                        ctx->ntfs_vol_p);
 378         link_dentry(child, ctx->dentry);
 379 out_close_ni:
 380         ntfs_inode_close(ni);
 381 out_free_utf8_name:
 382         FREE(utf8_name);
 383 out:
 384         return ret;
 385 }
 386
 387 /* Recursively build a WIM dentry tree corresponding to a NTFS volume.
 388  * At the same time, update the WIM lookup table with lookup table entries for
 389  * the NTFS streams, and build an array of security descriptors.
 390  */
 391 static int __build_dentry_tree_ntfs(struct dentry *dentry, ntfs_inode *ni,
 392                                     char path[], size_t path_len,
 393                                     struct lookup_table *lookup_table,
 394                                     struct sd_tree *tree,
 395                                     ntfs_volume **ntfs_vol_p)
 396 {
 397         u32 attributes = ntfs_inode_get_attributes(ni);
 398         int mrec_flags = ni->mrec->flags;
 399         u32 sd_size;
 400         int ret = 0;
 401
 402         dentry->creation_time    = le64_to_cpu(ni->creation_time);
 403         dentry->last_write_time  = le64_to_cpu(ni->last_data_change_time);
 404         dentry->last_access_time = le64_to_cpu(ni->last_access_time);
 405         dentry->security_id      = le32_to_cpu(ni->security_id);
 406         dentry->attributes       = le32_to_cpu(attributes);
 407         dentry->hard_link        = ni->mft_no;
 408         dentry->resolved = true;
 409
 410         if (attributes & FILE_ATTR_REPARSE_POINT) {
 411                 /* Junction point, symbolic link, or other reparse point */
 412                 ret = capture_ntfs_streams(dentry, ni, path, path_len,
 413                                            lookup_table, ntfs_vol_p,
 414                                            AT_REPARSE_POINT);
 415         } else if (mrec_flags & MFT_RECORD_IS_DIRECTORY) {
 416                 /* Normal directory */
 417                 s64 pos = 0;
 418                 struct readdir_ctx ctx = {
 419                         .dentry       = dentry,
 420                         .dir_ni       = ni,
 421                         .path         = path,
 422                         .path_len     = path_len,
 423                         .lookup_table = lookup_table,
 424                         .tree         = tree,
 425                         .ntfs_vol_p   = ntfs_vol_p,
 426                 };
 427                 ret = ntfs_readdir(ni, &pos, &ctx, filldir);
 428                 if (ret != 0)
 429                         ret = WIMLIB_ERR_NTFS_3G;
 430         } else {
 431                 /* Normal file */
 432                 ret = capture_ntfs_streams(dentry, ni, path, path_len,
 433                                            lookup_table, ntfs_vol_p,
 434                                            AT_DATA);
 435         }
 436         if (ret != 0)
 437                 return ret;
 438         ret = ntfs_inode_get_security(ni,
 439                                       OWNER_SECURITY_INFORMATION |
 440                                       GROUP_SECURITY_INFORMATION |
 441                                       DACL_SECURITY_INFORMATION  |
 442                                       SACL_SECURITY_INFORMATION,
 443                                       NULL, 0, &sd_size);
 444         u8 sd[sd_size];
 445         ret = ntfs_inode_get_security(ni,
 446                                       OWNER_SECURITY_INFORMATION |
 447                                       GROUP_SECURITY_INFORMATION |
 448                                       DACL_SECURITY_INFORMATION  |
 449                                       SACL_SECURITY_INFORMATION,
 450                                       sd, sd_size, &sd_size);
 451         dentry->security_id = tree_add_sd(tree, sd, sd_size);
 452         return 0;
 453 }
 454
 455 static int build_dentry_tree_ntfs(struct dentry *root_dentry,
 456                                   const char *device,
 457                                   struct lookup_table *lookup_table,
 458                                   struct wim_security_data *sd,
 459                                   int flags,
 460                                   void *extra_arg)
 461 {
 462         ntfs_volume *vol;
 463         ntfs_inode *root_ni;
 464         int ret = 0;
 465         struct sd_tree tree;
 466         tree.sd = sd;
 467         tree.root = NULL;
 468         ntfs_volume **ntfs_vol_p = extra_arg;
 469
 470         vol = ntfs_mount(device, MS_RDONLY);
 471         if (!vol) {
 472                 ERROR_WITH_ERRNO("Failed to mount NTFS volume `%s' read-only",
 473                                  device);
 474                 return WIMLIB_ERR_NTFS_3G;
 475         }
 476         root_ni = ntfs_inode_open(vol, FILE_root);
 477         if (!root_ni) {
 478                 ERROR_WITH_ERRNO("Failed to open root inode of NTFS volume "
 479                                  "`%s'", device);
 480                 ret = WIMLIB_ERR_NTFS_3G;
 481                 goto out;
 482         }
 483         char path[4096];
 484         path[0] = '/';
 485         path[1] = '\0';
 486         ret = __build_dentry_tree_ntfs(root_dentry, root_ni, path, 1,
 487                                        lookup_table, &tree, ntfs_vol_p);
 488         ntfs_inode_close(root_ni);
 489
 490 out:
 491         if (ntfs_umount(vol, FALSE) != 0) {
 492                 ERROR_WITH_ERRNO("Failed to unmount NTFS volume `%s'", device);
 493                 if (ret == 0)
 494                         ret = WIMLIB_ERR_NTFS_3G;
 495         }
 496         return ret;
 497 }
 498
 499 WIMLIBAPI int wimlib_add_image_from_ntfs_volume(WIMStruct *w,
 500                                                 const char *device,
 501                                                 const char *name,
 502                                                 const char *description,
 503                                                 const char *flags_element,
 504                                                 int flags)
 505 {
 506         if (flags & (WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE)) {
 507                 ERROR("Cannot dereference files when capturing directly from NTFS");
 508                 return WIMLIB_ERR_INVALID_PARAM;
 509         }
 510         return do_add_image(w, device, name, description, flags_element, flags,
 511                             build_dentry_tree_ntfs,
 512                             &w->ntfs_vol);
 513 }
 514
 515 #else /* WITH_NTFS_3G */
 516 WIMLIBAPI int wimlib_add_image_from_ntfs_volume(WIMStruct *w,
 517                                                 const char *device,
 518                                                 const char *name,
 519                                                 const char *description,
 520                                                 const char *flags_element,
 521                                                 int flags)
 522 {
 523         ERROR("wimlib was compiled without support for NTFS-3g, so");
 524         ERROR("we cannot capture a WIM image directly from a NTFS volume");
 525         return WIMLIB_ERR_UNSUPPORTED;
 526 }
 527 #endif /* WITH_NTFS_3G */