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

   1 /*
   2  * unix_capture.c:  Capture a directory tree on UNIX.
   3  */
   4
   5 /*
   6  * Copyright (C) 2012, 2013, 2014 Eric Biggers
   7  *
   8  * This file is free software; you can redistribute it and/or modify it under
   9  * the terms of the GNU Lesser General Public License as published by the Free
  10  * Software Foundation; either version 3 of the License, or (at your option) any
  11  * later version.
  12  *
  13  * This file is distributed in the hope that it will be useful, but WITHOUT
  14  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
  15  * FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
  16  * details.
  17  *
  18  * You should have received a copy of the GNU Lesser General Public License
  19  * along with this file; if not, see http://www.gnu.org/licenses/.
  20  */
  21
  22 #ifndef __WIN32__
  23
  24 #ifdef HAVE_CONFIG_H
  25 #  include "config.h"
  26 #endif
  27
  28 #include <dirent.h>
  29 #include <errno.h>
  30 #include <fcntl.h>
  31 #include <limits.h> /* for PATH_MAX */
  32 #include <sys/stat.h>
  33 #include <unistd.h>
  34
  35 #include "wimlib/capture.h"
  36 #include "wimlib/dentry.h"
  37 #include "wimlib/error.h"
  38 #include "wimlib/lookup_table.h"
  39 #include "wimlib/reparse.h"
  40 #include "wimlib/timestamp.h"
  41 #include "wimlib/unix_data.h"
  42
  43 #ifdef HAVE_FDOPENDIR
  44 #  define my_fdopendir(dirfd_p) fdopendir(*(dirfd_p))
  45 #else
  46 static DIR *
  47 my_fdopendir(int *dirfd_p)
  48 {
  49         DIR *dir = NULL;
  50         int old_pwd;
  51
  52         old_pwd = open(".", O_RDONLY);
  53         if (old_pwd >= 0) {
  54                 if (!fchdir(*dirfd_p)) {
  55                         dir = opendir(".");
  56                         if (dir) {
  57                                 close(*dirfd_p);
  58                                 *dirfd_p = dirfd(dir);
  59                         }
  60                         fchdir(old_pwd);
  61                 }
  62                 close(old_pwd);
  63         }
  64         return dir;
  65 }
  66 #endif
  67
  68 #ifdef HAVE_OPENAT
  69 #  define my_openat(full_path, dirfd, relpath, flags) \
  70                 openat((dirfd), (relpath), (flags))
  71 #else
  72 #  define my_openat(full_path, dirfd, relpath, flags) \
  73                 open((full_path), (flags))
  74 #endif
  75
  76 #ifdef HAVE_READLINKAT
  77 #  define my_readlinkat(full_path, dirfd, relpath, buf, bufsize) \
  78                 readlinkat((dirfd), (relpath), (buf), (bufsize))
  79 #else
  80 #  define my_readlinkat(full_path, dirfd, relpath, buf, bufsize) \
  81                 readlink((full_path), (buf), (bufsize))
  82 #endif
  83
  84 #ifdef HAVE_FSTATAT
  85 #  define my_fstatat(full_path, dirfd, relpath, stbuf, flags)   \
  86         fstatat((dirfd), (relpath), (stbuf), (flags))
  87 #else
  88 #  define my_fstatat(full_path, dirfd, relpath, stbuf, flags)   \
  89         ((flags) & AT_SYMLINK_NOFOLLOW) ? \
  90                 lstat((full_path), (stbuf)) : \
  91                 stat((full_path), (stbuf))
  92 #endif
  93
  94 #ifndef AT_FDCWD
  95 #  define AT_FDCWD      -100
  96 #endif
  97
  98 #ifndef AT_SYMLINK_NOFOLLOW
  99 #  define AT_SYMLINK_NOFOLLOW   0x100
 100 #endif
 101
 102 static int
 103 unix_scan_regular_file(const char *path, u64 size, struct wim_inode *inode,
 104                        struct list_head *unhashed_streams)
 105 {
 106         struct wim_lookup_table_entry *lte;
 107         char *file_on_disk;
 108
 109         inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
 110
 111         /* Empty files do not have to have a lookup table entry. */
 112         if (!size)
 113                 return 0;
 114
 115         file_on_disk = STRDUP(path);
 116         if (!file_on_disk)
 117                 return WIMLIB_ERR_NOMEM;
 118         lte = new_lookup_table_entry();
 119         if (!lte) {
 120                 FREE(file_on_disk);
 121                 return WIMLIB_ERR_NOMEM;
 122         }
 123         lte->file_on_disk = file_on_disk;
 124         lte->resource_location = RESOURCE_IN_FILE_ON_DISK;
 125         lte->size = size;
 126         add_unhashed_stream(lte, inode, 0, unhashed_streams);
 127         inode->i_lte = lte;
 128         return 0;
 129 }
 130
 131 static int
 132 unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 133                                  char *path, size_t path_len,
 134                                  int dirfd, const char *relpath,
 135                                  struct add_image_params *params);
 136
 137 static int
 138 unix_scan_directory(struct wim_dentry *dir_dentry,
 139                     char *full_path, size_t full_path_len,
 140                     int parent_dirfd, const char *dir_relpath,
 141                     struct add_image_params *params)
 142 {
 143
 144         int dirfd;
 145         DIR *dir;
 146         int ret;
 147
 148         dirfd = my_openat(full_path, parent_dirfd, dir_relpath, O_RDONLY);
 149         if (dirfd < 0) {
 150                 ERROR_WITH_ERRNO("\"%s\": Can't open directory", full_path);
 151                 return WIMLIB_ERR_OPENDIR;
 152         }
 153
 154         dir_dentry->d_inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 155         dir = my_fdopendir(&dirfd);
 156         if (!dir) {
 157                 ERROR_WITH_ERRNO("\"%s\": Can't open directory", full_path);
 158                 close(dirfd);
 159                 return WIMLIB_ERR_OPENDIR;
 160         }
 161
 162         ret = 0;
 163         for (;;) {
 164                 struct dirent *entry;
 165                 struct wim_dentry *child;
 166                 size_t name_len;
 167
 168                 errno = 0;
 169                 entry = readdir(dir);
 170                 if (!entry) {
 171                         if (errno) {
 172                                 ret = WIMLIB_ERR_READ;
 173                                 ERROR_WITH_ERRNO("\"%s\": Error reading directory",
 174                                                  full_path);
 175                         }
 176                         break;
 177                 }
 178
 179                 if (entry->d_name[0] == '.' &&
 180                     (entry->d_name[1] == '\0' ||
 181                      (entry->d_name[1] == '.' && entry->d_name[2] == '\0')))
 182                         continue;
 183
 184                 full_path[full_path_len] = '/';
 185                 name_len = strlen(entry->d_name);
 186                 memcpy(&full_path[full_path_len + 1], entry->d_name, name_len + 1);
 187                 ret = unix_build_dentry_tree_recursive(&child,
 188                                                        full_path,
 189                                                        full_path_len + 1 + name_len,
 190                                                        dirfd,
 191                                                        &full_path[full_path_len + 1],
 192                                                        params);
 193                 full_path[full_path_len] = '\0';
 194                 if (ret)
 195                         break;
 196                 if (child)
 197                         dentry_add_child(dir_dentry, child);
 198         }
 199         closedir(dir);
 200         return ret;
 201 }
 202
 203 /* Given an absolute symbolic link target @dest (UNIX-style, beginning
 204  * with '/'), determine whether it points into the directory specified by
 205  * @ino and @dev.  If so, return the target modified to be "absolute"
 206  * relative to this directory.  Otherwise, return NULL.  */
 207 static char *
 208 unix_fixup_abslink(char *dest, u64 ino, u64 dev)
 209 {
 210         char *p = dest;
 211
 212         do {
 213                 char save;
 214                 struct stat stbuf;
 215                 int ret;
 216
 217                 /* Skip non-slashes.  */
 218                 while (*p && *p != '/')
 219                         p++;
 220
 221                 /* Skip slashes.  */
 222                 while (*p && *p == '/')
 223                         p++;
 224
 225                 /* Get inode and device for this prefix.  */
 226                 save = *p;
 227                 *p = '\0';
 228                 ret = stat(dest, &stbuf);
 229                 *p = save;
 230
 231                 if (ret) {
 232                         /* stat() failed.  Assume the link points outside the
 233                          * directory tree being captured.  */
 234                         break;
 235                 }
 236
 237                 if (stbuf.st_ino == ino && stbuf.st_dev == dev) {
 238                         /* Link points inside directory tree being captured.
 239                          * Return abbreviated path.  */
 240                         *--p = '/';
 241                         while (p > dest && *(p - 1) == '/')
 242                                 p--;
 243                         return p;
 244                 }
 245         } while (*p);
 246
 247         /* Link does not point inside directory tree being captured.  */
 248         return NULL;
 249 }
 250
 251 static int
 252 unix_scan_symlink(const char *full_path, int dirfd, const char *relpath,
 253                   struct wim_inode *inode, struct add_image_params *params)
 254 {
 255         char deref_name_buf[4096];
 256         ssize_t deref_name_len;
 257         char *dest;
 258         int ret;
 259
 260         inode->i_attributes = FILE_ATTRIBUTE_REPARSE_POINT;
 261         inode->i_reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
 262
 263         /* The idea here is to call readlink() to get the UNIX target of the
 264          * symbolic link, then turn the target into a reparse point data buffer
 265          * that contains a relative or absolute symbolic link. */
 266         deref_name_len = my_readlinkat(full_path, dirfd, relpath,
 267                                        deref_name_buf, sizeof(deref_name_buf) - 1);
 268         if (deref_name_len < 0) {
 269                 ERROR_WITH_ERRNO("\"%s\": Can't read target of symbolic link",
 270                                  full_path);
 271                 return WIMLIB_ERR_READLINK;
 272         }
 273
 274         dest = deref_name_buf;
 275
 276         dest[deref_name_len] = '\0';
 277
 278         if ((params->add_flags & WIMLIB_ADD_FLAG_RPFIX) &&
 279              dest[0] == '/')
 280         {
 281                 char *fixed_dest;
 282
 283                 /* RPFIX (reparse point fixup) mode:  Change target of absolute
 284                  * symbolic link to be "absolute" relative to the tree being
 285                  * captured.  */
 286                 fixed_dest = unix_fixup_abslink(dest,
 287                                                 params->capture_root_ino,
 288                                                 params->capture_root_dev);
 289                 params->progress.scan.cur_path = full_path;
 290                 params->progress.scan.symlink_target = deref_name_buf;
 291                 if (fixed_dest) {
 292                         /* Link points inside the tree being captured, so it was
 293                          * fixed.  */
 294                         inode->i_not_rpfixed = 0;
 295                         dest = fixed_dest;
 296                         ret = do_capture_progress(params,
 297                                                   WIMLIB_SCAN_DENTRY_FIXED_SYMLINK,
 298                                                   NULL);
 299                 } else {
 300                         /* Link points outside the tree being captured, so it
 301                          * was not fixed.  */
 302                         ret = do_capture_progress(params,
 303                                                   WIMLIB_SCAN_DENTRY_NOT_FIXED_SYMLINK,
 304                                                   NULL);
 305                 }
 306                 if (ret)
 307                         return ret;
 308         }
 309         ret = wim_inode_set_symlink(inode, dest, params->lookup_table);
 310         if (ret)
 311                 return ret;
 312
 313         /* Unfortunately, Windows seems to have the concept of "file" symbolic
 314          * links as being different from "directory" symbolic links...  so
 315          * FILE_ATTRIBUTE_DIRECTORY needs to be set on the symbolic link if the
 316          * *target* of the symbolic link is a directory.  */
 317         struct stat stbuf;
 318         if (my_fstatat(full_path, dirfd, relpath, &stbuf, 0) == 0 &&
 319             S_ISDIR(stbuf.st_mode))
 320                 inode->i_attributes |= FILE_ATTRIBUTE_DIRECTORY;
 321         return 0;
 322 }
 323
 324 static int
 325 unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 326                                  char *full_path, size_t full_path_len,
 327                                  int dirfd, const char *relpath,
 328                                  struct add_image_params *params)
 329 {
 330         struct wim_dentry *tree = NULL;
 331         struct wim_inode *inode = NULL;
 332         int ret;
 333         struct stat stbuf;
 334         int stat_flags;
 335
 336         ret = try_exclude(full_path, full_path_len, params);
 337         if (ret < 0) /* Excluded? */
 338                 goto out_progress;
 339         if (ret > 0) /* Error? */
 340                 goto out;
 341
 342         if (params->add_flags & (WIMLIB_ADD_FLAG_DEREFERENCE |
 343                                  WIMLIB_ADD_FLAG_ROOT))
 344                 stat_flags = 0;
 345         else
 346                 stat_flags = AT_SYMLINK_NOFOLLOW;
 347
 348         ret = my_fstatat(full_path, dirfd, relpath, &stbuf, stat_flags);
 349
 350         if (ret) {
 351                 ERROR_WITH_ERRNO("\"%s\": Can't read metadata", full_path);
 352                 ret = WIMLIB_ERR_STAT;
 353                 goto out;
 354         }
 355
 356         if (!(params->add_flags & WIMLIB_ADD_FLAG_UNIX_DATA)) {
 357                 if (unlikely(!S_ISREG(stbuf.st_mode) &&
 358                              !S_ISDIR(stbuf.st_mode) &&
 359                              !S_ISLNK(stbuf.st_mode)))
 360                 {
 361                         if (params->add_flags &
 362                             WIMLIB_ADD_FLAG_NO_UNSUPPORTED_EXCLUDE)
 363                         {
 364                                 ERROR("\"%s\": File type is unsupported",
 365                                       full_path);
 366                                 ret = WIMLIB_ERR_UNSUPPORTED_FILE;
 367                                 goto out;
 368                         }
 369                         params->progress.scan.cur_path = full_path;
 370                         ret = do_capture_progress(params,
 371                                                   WIMLIB_SCAN_DENTRY_UNSUPPORTED,
 372                                                   NULL);
 373                         goto out;
 374                 }
 375         }
 376
 377         ret = inode_table_new_dentry(params->inode_table, relpath,
 378                                      stbuf.st_ino, stbuf.st_dev,
 379                                      S_ISDIR(stbuf.st_mode), &tree);
 380         if (ret)
 381                 goto out;
 382
 383         inode = tree->d_inode;
 384
 385         /* Already seen this inode?  */
 386         if (inode->i_nlink > 1)
 387                 goto out_progress;
 388
 389 #ifdef HAVE_STAT_NANOSECOND_PRECISION
 390         inode->i_creation_time = timespec_to_wim_timestamp(stbuf.st_mtim);
 391         inode->i_last_write_time = timespec_to_wim_timestamp(stbuf.st_mtim);
 392         inode->i_last_access_time = timespec_to_wim_timestamp(stbuf.st_atim);
 393 #else
 394         inode->i_creation_time = unix_timestamp_to_wim(stbuf.st_mtime);
 395         inode->i_last_write_time = unix_timestamp_to_wim(stbuf.st_mtime);
 396         inode->i_last_access_time = unix_timestamp_to_wim(stbuf.st_atime);
 397 #endif
 398         inode->i_resolved = 1;
 399         if (params->add_flags & WIMLIB_ADD_FLAG_UNIX_DATA) {
 400                 struct wimlib_unix_data unix_data;
 401
 402                 unix_data.uid = stbuf.st_uid;
 403                 unix_data.gid = stbuf.st_gid;
 404                 unix_data.mode = stbuf.st_mode;
 405                 unix_data.rdev = stbuf.st_rdev;
 406                 if (!inode_set_unix_data(inode, &unix_data, UNIX_DATA_ALL)) {
 407                         ret = WIMLIB_ERR_NOMEM;
 408                         goto out;
 409                 }
 410         }
 411
 412         if (params->add_flags & WIMLIB_ADD_FLAG_ROOT) {
 413                 params->capture_root_ino = stbuf.st_ino;
 414                 params->capture_root_dev = stbuf.st_dev;
 415                 params->add_flags &= ~WIMLIB_ADD_FLAG_ROOT;
 416         }
 417
 418         if (S_ISREG(stbuf.st_mode)) {
 419                 ret = unix_scan_regular_file(full_path, stbuf.st_size,
 420                                              inode, params->unhashed_streams);
 421         } else if (S_ISDIR(stbuf.st_mode)) {
 422                 ret = unix_scan_directory(tree, full_path, full_path_len,
 423                                           dirfd, relpath, params);
 424         } else if (S_ISLNK(stbuf.st_mode)) {
 425                 ret = unix_scan_symlink(full_path, dirfd, relpath,
 426                                         inode, params);
 427         }
 428
 429         if (ret)
 430                 goto out;
 431
 432 out_progress:
 433         params->progress.scan.cur_path = full_path;
 434         if (likely(tree))
 435                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_OK, inode);
 436         else
 437                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_EXCLUDED, NULL);
 438 out:
 439         if (likely(ret == 0))
 440                 *tree_ret = tree;
 441         else
 442                 free_dentry_tree(tree, params->lookup_table);
 443         return ret;
 444 }
 445
 446 /*
 447  * unix_build_dentry_tree():
 448  *      Builds a tree of WIM dentries from an on-disk directory tree (UNIX
 449  *      version; no NTFS-specific data is captured).
 450  *
 451  * @root_ret:   Place to return a pointer to the root of the dentry tree.  Only
 452  *              modified if successful.  Set to NULL if the file or directory was
 453  *              excluded from capture.
 454  *
 455  * @root_disk_path:  The path to the root of the directory tree on disk.
 456  *
 457  * @params:     See doc for `struct add_image_params'.
 458  *
 459  * @return:     0 on success, nonzero on failure.  It is a failure if any of
 460  *              the files cannot be `stat'ed, or if any of the needed
 461  *              directories cannot be opened or read.  Failure to add the files
 462  *              to the WIM may still occur later when trying to actually read
 463  *              the on-disk files during a call to wimlib_write() or
 464  *              wimlib_overwrite().
 465  */
 466 int
 467 unix_build_dentry_tree(struct wim_dentry **root_ret,
 468                        const char *root_disk_path,
 469                        struct add_image_params *params)
 470 {
 471         size_t path_len;
 472         size_t path_bufsz;
 473         char *path_buf;
 474         int ret;
 475
 476         path_len = strlen(root_disk_path);
 477         path_bufsz = min(32790, PATH_MAX + 1);
 478
 479         if (path_len >= path_bufsz)
 480                 return WIMLIB_ERR_INVALID_PARAM;
 481
 482         path_buf = MALLOC(path_bufsz);
 483         if (!path_buf)
 484                 return WIMLIB_ERR_NOMEM;
 485         memcpy(path_buf, root_disk_path, path_len + 1);
 486
 487         params->capture_root_nchars = path_len;
 488
 489         ret = unix_build_dentry_tree_recursive(root_ret, path_buf, path_len,
 490                                                AT_FDCWD, path_buf, params);
 491         FREE(path_buf);
 492         return ret;
 493 }
 494
 495 #endif /* !__WIN32__ */