wimlib.net Git - wimlib/blob - src/verify.c

   1 /*
   2  * verify.c
   3  *
   4  * Some functions to verify that stuff in the WIM is valid.  Of course, not
   5  * *all* the verifications of the input data are in this file.
   6  */
   7
   8 /*
   9  * Copyright (C) 2012, 2013 Eric Biggers
  10  *
  11  * wimlib - Library for working with WIM files
  12  *
  13  * This file is part of wimlib, a library for working with WIM files.
  14  *
  15  * wimlib is free software; you can redistribute it and/or modify it under the
  16  * terms of the GNU General Public License as published by the Free
  17  * Software Foundation; either version 3 of the License, or (at your option)
  18  * any later version.
  19  *
  20  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  21  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  22  * A PARTICULAR PURPOSE. See the GNU General Public License for more
  23  * details.
  24  *
  25  * You should have received a copy of the GNU General Public License
  26  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  27  */
  28
  29 #include "wimlib_internal.h"
  30 #include "dentry.h"
  31 #include "lookup_table.h"
  32
  33 static int
  34 verify_inode(struct wim_inode *inode, const WIMStruct *w)
  35 {
  36         const struct wim_lookup_table *table = w->lookup_table;
  37         const struct wim_security_data *sd = wim_const_security_data(w);
  38         const struct wim_dentry *first_dentry = inode_first_dentry(inode);
  39         const struct wim_dentry *dentry;
  40         int ret = WIMLIB_ERR_INVALID_DENTRY;
  41
  42         /* Check the security ID.  -1 is valid and means "no security
  43          * descriptor".  Anything else has to be a valid index into the WIM
  44          * image's security descriptors table. */
  45         if (inode->i_security_id < -1) {
  46                 ERROR("Dentry `%"TS"' has an invalid security ID (%d)",
  47                       first_dentry->full_path, inode->i_security_id);
  48                 goto out;
  49         }
  50
  51         if (inode->i_security_id >= sd->num_entries) {
  52                 ERROR("Dentry `%"TS"' has an invalid security ID (%d) "
  53                       "(there are only %u entries in the security table)",
  54                       first_dentry->full_path, inode->i_security_id,
  55                       sd->num_entries);
  56                 goto out;
  57         }
  58
  59         /* Check that lookup table entries for all the inode's stream exist,
  60          * except if the SHA1 message digest is all 0's, which indicates an
  61          * empty stream.
  62          *
  63          * This check is skipped on split WIMs. */
  64         if (w->hdr.total_parts == 1) {
  65                 for (unsigned i = 0; i <= inode->i_num_ads; i++) {
  66                         struct wim_lookup_table_entry *lte;
  67                         const u8 *hash;
  68                         hash = inode_stream_hash_unresolved(inode, i);
  69                         lte = __lookup_resource(table, hash);
  70                         if (!lte && !is_zero_hash(hash)) {
  71                                 ERROR("Could not find lookup table entry for stream "
  72                                       "%u of dentry `%"TS"'",
  73                                       i, first_dentry->full_path);
  74                                 goto out;
  75                         }
  76                         if (lte)
  77                                 lte->real_refcnt += inode->i_nlink;
  78                 }
  79         }
  80
  81         /* Make sure there is only one unnamed data stream. */
  82         unsigned num_unnamed_streams = 0;
  83         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
  84                 const u8 *hash;
  85                 hash = inode_stream_hash_unresolved(inode, i);
  86                 if (inode_stream_name_nbytes(inode, i) == 0 && !is_zero_hash(hash))
  87                         num_unnamed_streams++;
  88         }
  89         if (num_unnamed_streams > 1) {
  90                 ERROR("Dentry `%"TS"' has multiple (%u) un-named streams",
  91                       first_dentry->full_path, num_unnamed_streams);
  92                 goto out;
  93         }
  94
  95         /* Files cannot have multiple DOS names, even if they have multiple
  96          * names in multiple directories (i.e. hard links).
  97          * Source: NTFS-3g authors. */
  98         const struct wim_dentry *dentry_with_dos_name = NULL;
  99         inode_for_each_dentry(dentry, inode) {
 100                 if (dentry_has_short_name(dentry)) {
 101                         if (dentry_with_dos_name) {
 102                                 ERROR("Hard-linked file has a DOS name at "
 103                                       "both `%"TS"' and `%"TS"'",
 104                                       dentry_with_dos_name->full_path,
 105                                       dentry->full_path);
 106                                 goto out;
 107                         }
 108                         dentry_with_dos_name = dentry;
 109                 }
 110         }
 111
 112         /* Directories with multiple links have not been tested. XXX */
 113         if (inode->i_nlink > 1 && inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) {
 114                 ERROR("Hard-linked directory `%"TS"' is unsupported",
 115                       first_dentry->full_path);
 116                 goto out;
 117         }
 118
 119         inode->i_verified = 1;
 120         ret = 0;
 121 out:
 122         return ret;
 123 }
 124
 125 /* Run some miscellaneous verifications on a WIM dentry */
 126 int
 127 verify_dentry(struct wim_dentry *dentry, void *wim)
 128 {
 129         int ret;
 130         WIMStruct *w = wim;
 131
 132         /* Verify the associated inode, but only one time no matter how many
 133          * dentries it has (unless we are doing a full verification of the WIM,
 134          * in which case we need to force the inode to be verified again.) */
 135         if (!dentry->d_inode->i_verified || w->full_verification_in_progress) {
 136                 ret = verify_inode(dentry->d_inode, w);
 137                 if (ret != 0)
 138                         return ret;
 139         }
 140
 141         /* Make sure root dentry is unnamed, while every other dentry has at
 142          * least a long name.
 143          *
 144          * I am assuming that dentries having only a DOS name is illegal; i.e.,
 145          * Windows will always combine the Win32 name and DOS name for a file
 146          * into a single WIM dentry, even if they are stored separately on NTFS.
 147          * (This seems to be the case...) */
 148         if (dentry_is_root(dentry)) {
 149                 if (dentry_has_long_name(dentry) || dentry_has_short_name(dentry)) {
 150                         ERROR("The root dentry has a nonempty name!");
 151                         return WIMLIB_ERR_INVALID_DENTRY;
 152                 }
 153         } else {
 154                 if (!dentry_has_long_name(dentry)) {
 155                         ERROR("Dentry `%"TS"' has no long name!",
 156                               dentry->full_path);
 157                         return WIMLIB_ERR_INVALID_DENTRY;
 158                 }
 159         }
 160
 161 #if 0
 162         /* Check timestamps */
 163         if (inode->i_last_access_time < inode->i_creation_time ||
 164             inode->i_last_write_time < inode->i_creation_time) {
 165                 WARNING("Dentry `%"TS"' was created after it was last accessed or "
 166                         "written to", dentry->full_path);
 167         }
 168 #endif
 169
 170         return 0;
 171 }
 172
 173 static int
 174 image_run_full_verifications(WIMStruct *w)
 175 {
 176         return for_dentry_in_tree(wim_root_dentry(w), verify_dentry, w);
 177 }
 178
 179 static int
 180 lte_fix_refcnt(struct wim_lookup_table_entry *lte, void *ctr)
 181 {
 182         if (lte->refcnt != lte->real_refcnt) {
 183         #ifdef ENABLE_ERROR_MESSAGES
 184                 WARNING("The following lookup table entry has a reference "
 185                         "count of %u, but", lte->refcnt);
 186                 WARNING("We found %u references to it",
 187                         lte->real_refcnt);
 188                 print_lookup_table_entry(lte, stderr);
 189         #endif
 190                 lte->refcnt = lte->real_refcnt;
 191                 ++*(unsigned long *)ctr;
 192         }
 193         return 0;
 194 }
 195
 196 /* Ideally this would be unnecessary... however, the WIMs for Windows 8 are
 197  * screwed up because some lookup table entries are referenced more times than
 198  * their stated reference counts.  So theoretically, if we delete all the
 199  * references to a stream and then remove it, it might still be referenced
 200  * somewhere else, making a file be missing from the WIM... So, work around this
 201  * problem by looking at ALL the images to re-calculate the reference count of
 202  * EVERY lookup table entry.  This only absolutely has to be done before an image
 203  * is deleted or before an image is mounted read-write. */
 204 int
 205 wim_run_full_verifications(WIMStruct *w)
 206 {
 207         int ret;
 208
 209         for_lookup_table_entry(w->lookup_table, lte_zero_real_refcnt, NULL);
 210         w->all_images_verified = 1;
 211         w->full_verification_in_progress = 1;
 212         ret = for_image(w, WIMLIB_ALL_IMAGES, image_run_full_verifications);
 213         w->full_verification_in_progress = 0;
 214         if (ret == 0) {
 215                 unsigned long num_ltes_with_bogus_refcnt = 0;
 216                 for (int i = 0; i < w->hdr.image_count; i++)
 217                         w->image_metadata[i].metadata_lte->real_refcnt++;
 218                 for_lookup_table_entry(w->lookup_table, lte_fix_refcnt,
 219                                        &num_ltes_with_bogus_refcnt);
 220                 if (num_ltes_with_bogus_refcnt != 0) {
 221                         WARNING("A total of %lu entries in the WIM's stream "
 222                                 "lookup table had to have\n"
 223                                 "          their reference counts fixed.",
 224                                 num_ltes_with_bogus_refcnt);
 225                 }
 226         } else {
 227                 w->all_images_verified = 0;
 228         }
 229         return ret;
 230 }
 231
 232 /*
 233  * verify_swm_set: - Sanity checks to make sure a set of WIMs correctly
 234  *                   correspond to a spanned set.
 235  *
 236  * @w:
 237  *      Part 1 of the set.
 238  *
 239  * @additional_swms:
 240  *      All parts of the set other than part 1.
 241  *
 242  * @num_additional_swms:
 243  *      Number of WIMStructs in @additional_swms.  Or, the total number of parts
 244  *      in the set minus 1.
 245  *
 246  * @return:
 247  *      0 on success; WIMLIB_ERR_SPLIT_INVALID if the set is not valid.
 248  */
 249 int
 250 verify_swm_set(WIMStruct *w, WIMStruct **additional_swms,
 251                unsigned num_additional_swms)
 252 {
 253         unsigned total_parts = w->hdr.total_parts;
 254         int ctype;
 255         const u8 *guid;
 256
 257         if (total_parts != num_additional_swms + 1) {
 258                 ERROR("`%"TS"' says there are %u parts in the spanned set, "
 259                       "but %"TS"%u part%"TS" provided",
 260                       w->filename, total_parts,
 261                       (num_additional_swms + 1 < total_parts) ? T("only ") : T(""),
 262                       num_additional_swms + 1,
 263                       (num_additional_swms) ? T("s were") : T(" was"));
 264                 return WIMLIB_ERR_SPLIT_INVALID;
 265         }
 266         if (w->hdr.part_number != 1) {
 267                 ERROR("WIM `%"TS"' is not the first part of the split WIM.",
 268                       w->filename);
 269                 return WIMLIB_ERR_SPLIT_INVALID;
 270         }
 271         for (unsigned i = 0; i < num_additional_swms; i++) {
 272                 if (additional_swms[i]->hdr.total_parts != total_parts) {
 273                         ERROR("WIM `%"TS"' says there are %u parts in the "
 274                               "spanned set, but %u parts were provided",
 275                               additional_swms[i]->filename,
 276                               additional_swms[i]->hdr.total_parts,
 277                               total_parts);
 278                         return WIMLIB_ERR_SPLIT_INVALID;
 279                 }
 280         }
 281
 282         /* keep track of ctype and guid just to make sure they are the same for
 283          * all the WIMs. */
 284         ctype = wimlib_get_compression_type(w);
 285         guid = w->hdr.guid;
 286
 287         {
 288                 /* parts_to_swms is not allocated at function scope because it
 289                  * should only be allocated after num_additional_swms was
 290                  * checked to be the same as w->hdr.total_parts.  Otherwise, it
 291                  * could be unexpectedly high and cause a stack overflow. */
 292                 WIMStruct *parts_to_swms[num_additional_swms];
 293                 ZERO_ARRAY(parts_to_swms);
 294                 for (unsigned i = 0; i < num_additional_swms; i++) {
 295
 296                         WIMStruct *swm = additional_swms[i];
 297
 298                         if (wimlib_get_compression_type(swm) != ctype) {
 299                                 ERROR("The split WIMs do not all have the same "
 300                                       "compression type");
 301                                 return WIMLIB_ERR_SPLIT_INVALID;
 302                         }
 303                         if (memcmp(guid, swm->hdr.guid, WIM_GID_LEN) != 0) {
 304                                 ERROR("The split WIMs do not all have the same "
 305                                       "GUID");
 306                                 return WIMLIB_ERR_SPLIT_INVALID;
 307                         }
 308                         if (swm->hdr.part_number == 1) {
 309                                 ERROR("WIMs `%"TS"' and `%"TS"' both are marked "
 310                                       "as the first WIM in the spanned set",
 311                                       w->filename, swm->filename);
 312                                 return WIMLIB_ERR_SPLIT_INVALID;
 313                         }
 314                         if (swm->hdr.part_number == 0 ||
 315                             swm->hdr.part_number > total_parts)
 316                         {
 317                                 ERROR("WIM `%"TS"' says it is part %u in the "
 318                                       "spanned set, but the part number must "
 319                                       "be in the range [1, %u]",
 320                                       swm->filename, swm->hdr.part_number, total_parts);
 321                                 return WIMLIB_ERR_SPLIT_INVALID;
 322                         }
 323                         if (parts_to_swms[swm->hdr.part_number - 2])
 324                         {
 325                                 ERROR("`%"TS"' and `%"TS"' are both marked as "
 326                                       "part %u of %u in the spanned set",
 327                                       parts_to_swms[swm->hdr.part_number - 2]->filename,
 328                                       swm->filename,
 329                                       swm->hdr.part_number,
 330                                       total_parts);
 331                                 return WIMLIB_ERR_SPLIT_INVALID;
 332                         } else {
 333                                 parts_to_swms[swm->hdr.part_number - 2] = swm;
 334                         }
 335                 }
 336         }
 337         return 0;
 338 }