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/blob_table.h"
  36 #include "wimlib/capture.h"
  37 #include "wimlib/dentry.h"
  38 #include "wimlib/error.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_blobs)
 105 {
 106         struct blob_descriptor *blob = NULL;
 107         struct wim_inode_stream *strm;
 108
 109         inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
 110
 111         if (size) {
 112                 blob = new_blob_descriptor();
 113                 if (unlikely(!blob))
 114                         goto err_nomem;
 115                 blob->file_on_disk = STRDUP(path);
 116                 if (unlikely(!blob->file_on_disk))
 117                         goto err_nomem;
 118                 blob->blob_location = BLOB_IN_FILE_ON_DISK;
 119                 blob->size = size;
 120                 blob->file_inode = inode;
 121         }
 122
 123         strm = inode_add_stream(inode, STREAM_TYPE_DATA, NO_STREAM_NAME, blob);
 124         if (unlikely(!strm))
 125                 goto err_nomem;
 126
 127         prepare_unhashed_blob(blob, inode, strm->stream_id, unhashed_blobs);
 128         return 0;
 129
 130 err_nomem:
 131         free_blob_descriptor(blob);
 132         return WIMLIB_ERR_NOMEM;
 133 }
 134
 135 static int
 136 unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 137                                  char *path, size_t path_len,
 138                                  int dirfd, const char *relpath,
 139                                  struct capture_params *params);
 140
 141 static int
 142 unix_scan_directory(struct wim_dentry *dir_dentry,
 143                     char *full_path, size_t full_path_len,
 144                     int parent_dirfd, const char *dir_relpath,
 145                     struct capture_params *params)
 146 {
 147
 148         int dirfd;
 149         DIR *dir;
 150         int ret;
 151
 152         dirfd = my_openat(full_path, parent_dirfd, dir_relpath, O_RDONLY);
 153         if (dirfd < 0) {
 154                 ERROR_WITH_ERRNO("\"%s\": Can't open directory", full_path);
 155                 return WIMLIB_ERR_OPENDIR;
 156         }
 157
 158         dir_dentry->d_inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 159         dir = my_fdopendir(&dirfd);
 160         if (!dir) {
 161                 ERROR_WITH_ERRNO("\"%s\": Can't open directory", full_path);
 162                 close(dirfd);
 163                 return WIMLIB_ERR_OPENDIR;
 164         }
 165
 166         ret = 0;
 167         for (;;) {
 168                 struct dirent *entry;
 169                 struct wim_dentry *child;
 170                 size_t name_len;
 171
 172                 errno = 0;
 173                 entry = readdir(dir);
 174                 if (!entry) {
 175                         if (errno) {
 176                                 ret = WIMLIB_ERR_READ;
 177                                 ERROR_WITH_ERRNO("\"%s\": Error reading directory",
 178                                                  full_path);
 179                         }
 180                         break;
 181                 }
 182
 183                 name_len = strlen(entry->d_name);
 184
 185                 if (should_ignore_filename(entry->d_name, name_len))
 186                         continue;
 187
 188                 full_path[full_path_len] = '/';
 189                 memcpy(&full_path[full_path_len + 1], entry->d_name, name_len + 1);
 190                 ret = unix_build_dentry_tree_recursive(&child,
 191                                                        full_path,
 192                                                        full_path_len + 1 + name_len,
 193                                                        dirfd,
 194                                                        &full_path[full_path_len + 1],
 195                                                        params);
 196                 full_path[full_path_len] = '\0';
 197                 if (ret)
 198                         break;
 199                 if (child)
 200                         dentry_add_child(dir_dentry, child);
 201         }
 202         closedir(dir);
 203         return ret;
 204 }
 205
 206 /*
 207  * Given an absolute symbolic link target (UNIX-style, beginning with '/'),
 208  * determine whether it points into the directory identified by @ino and @dev.
 209  * If yes, return the suffix of @target which is relative to this directory, but
 210  * retaining leading slashes.  If no, return @target.
 211  *
 212  * Here are some examples, assuming that the @ino/@dev directory is "/home/e":
 213  *
 214  *      Original target         New target
 215  *      ---------------         ----------
 216  *      /home/e/test            /test
 217  *      /home/e/test/           /test/
 218  *      //home//e//test//       //test//
 219  *      /home/e                                         (empty string)
 220  *      /home/e/                /
 221  *      /usr/lib                /usr/lib                (external link)
 222  *
 223  * Because of the possibility of other links into the @ino/@dev directory and/or
 224  * multiple path separators, we can't simply do a string comparison; instead we
 225  * need to stat() each ancestor directory.
 226  *
 227  * If the link points directly to the @ino/@dev directory with no trailing
 228  * slashes, then the new target will be an empty string.  This is not a valid
 229  * UNIX symlink target, but we store this in the archive anyway since the target
 230  * is intended to be de-relativized when the link is extracted.
 231  */
 232 static char *
 233 unix_relativize_link_target(char *target, u64 ino, u64 dev)
 234 {
 235         char *p = target;
 236
 237         do {
 238                 char save;
 239                 struct stat stbuf;
 240                 int ret;
 241
 242                 /* Skip slashes (guaranteed to be at least one here)  */
 243                 do {
 244                         p++;
 245                 } while (*p == '/');
 246
 247                 /* End of string?  */
 248                 if (!*p)
 249                         break;
 250
 251                 /* Skip non-slashes (guaranteed to be at least one here)  */
 252                 do {
 253                         p++;
 254                 } while (*p && *p != '/');
 255
 256                 /* Get the inode and device numbers for this prefix.  */
 257                 save = *p;
 258                 *p = '\0';
 259                 ret = stat(target, &stbuf);
 260                 *p = save;
 261
 262                 if (ret) {
 263                         /* stat() failed.  Assume the link points outside the
 264                          * directory tree being captured.  */
 265                         break;
 266                 }
 267
 268                 if (stbuf.st_ino == ino && stbuf.st_dev == dev) {
 269                         /* Link points inside directory tree being captured.
 270                          * Return abbreviated path.  */
 271                         return p;
 272                 }
 273         } while (*p);
 274
 275         /* Link does not point inside directory tree being captured.  */
 276         return target;
 277 }
 278
 279 static noinline_for_stack int
 280 unix_scan_symlink(const char *full_path, int dirfd, const char *relpath,
 281                   struct wim_inode *inode, struct capture_params *params)
 282 {
 283         char orig_target[REPARSE_POINT_MAX_SIZE];
 284         char *target = orig_target;
 285         int ret;
 286
 287         /* Read the UNIX symbolic link target.  */
 288         ret = my_readlinkat(full_path, dirfd, relpath, target,
 289                             sizeof(orig_target));
 290         if (unlikely(ret < 0)) {
 291                 ERROR_WITH_ERRNO("\"%s\": Can't read target of symbolic link",
 292                                  full_path);
 293                 return WIMLIB_ERR_READLINK;
 294         }
 295         if (unlikely(ret >= sizeof(orig_target))) {
 296                 ERROR("\"%s\": target of symbolic link is too long", full_path);
 297                 return WIMLIB_ERR_READLINK;
 298         }
 299         target[ret] = '\0';
 300
 301         /* If the link is absolute and reparse point fixups are enabled, then
 302          * change it to be "absolute" relative to the tree being captured.  */
 303         if (target[0] == '/' && (params->add_flags & WIMLIB_ADD_FLAG_RPFIX)) {
 304                 int status = WIMLIB_SCAN_DENTRY_NOT_FIXED_SYMLINK;
 305
 306                 params->progress.scan.cur_path = full_path;
 307                 params->progress.scan.symlink_target = target;
 308
 309                 target = unix_relativize_link_target(target,
 310                                                      params->capture_root_ino,
 311                                                      params->capture_root_dev);
 312                 if (target != orig_target) {
 313                         /* Link target was fixed.  */
 314                         inode->i_rp_flags &= ~WIM_RP_FLAG_NOT_FIXED;
 315                         status = WIMLIB_SCAN_DENTRY_FIXED_SYMLINK;
 316                 }
 317                 ret = do_capture_progress(params, status, NULL);
 318                 if (ret)
 319                         return ret;
 320         }
 321
 322         /* Translate the UNIX symlink target into a Windows reparse point.  */
 323         ret = wim_inode_set_symlink(inode, target, params->blob_table);
 324         if (ret)
 325                 return ret;
 326
 327         /* On Windows, a reparse point can be set on both directory and
 328          * non-directory files.  Usually, a link that is intended to point to a
 329          * (non-)directory is stored as a reparse point on a (non-)directory
 330          * file.  Replicate this behavior by examining the target file.  */
 331         struct stat stbuf;
 332         if (my_fstatat(full_path, dirfd, relpath, &stbuf, 0) == 0 &&
 333             S_ISDIR(stbuf.st_mode))
 334                 inode->i_attributes |= FILE_ATTRIBUTE_DIRECTORY;
 335         return 0;
 336 }
 337
 338 static int
 339 unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 340                                  char *full_path, size_t full_path_len,
 341                                  int dirfd, const char *relpath,
 342                                  struct capture_params *params)
 343 {
 344         struct wim_dentry *tree = NULL;
 345         struct wim_inode *inode = NULL;
 346         int ret;
 347         struct stat stbuf;
 348         int stat_flags;
 349
 350         ret = try_exclude(full_path, params);
 351         if (unlikely(ret < 0)) /* Excluded? */
 352                 goto out_progress;
 353         if (unlikely(ret > 0)) /* Error? */
 354                 goto out;
 355
 356         if (params->add_flags & (WIMLIB_ADD_FLAG_DEREFERENCE |
 357                                  WIMLIB_ADD_FLAG_ROOT))
 358                 stat_flags = 0;
 359         else
 360                 stat_flags = AT_SYMLINK_NOFOLLOW;
 361
 362         ret = my_fstatat(full_path, dirfd, relpath, &stbuf, stat_flags);
 363
 364         if (ret) {
 365                 ERROR_WITH_ERRNO("\"%s\": Can't read metadata", full_path);
 366                 ret = WIMLIB_ERR_STAT;
 367                 goto out;
 368         }
 369
 370         if (!(params->add_flags & WIMLIB_ADD_FLAG_UNIX_DATA)) {
 371                 if (unlikely(!S_ISREG(stbuf.st_mode) &&
 372                              !S_ISDIR(stbuf.st_mode) &&
 373                              !S_ISLNK(stbuf.st_mode)))
 374                 {
 375                         if (params->add_flags &
 376                             WIMLIB_ADD_FLAG_NO_UNSUPPORTED_EXCLUDE)
 377                         {
 378                                 ERROR("\"%s\": File type is unsupported",
 379                                       full_path);
 380                                 ret = WIMLIB_ERR_UNSUPPORTED_FILE;
 381                                 goto out;
 382                         }
 383                         params->progress.scan.cur_path = full_path;
 384                         ret = do_capture_progress(params,
 385                                                   WIMLIB_SCAN_DENTRY_UNSUPPORTED,
 386                                                   NULL);
 387                         goto out;
 388                 }
 389         }
 390
 391         ret = inode_table_new_dentry(params->inode_table, relpath,
 392                                      stbuf.st_ino, stbuf.st_dev,
 393                                      S_ISDIR(stbuf.st_mode), &tree);
 394         if (ret)
 395                 goto out;
 396
 397         inode = tree->d_inode;
 398
 399         /* Already seen this inode?  */
 400         if (inode->i_nlink > 1)
 401                 goto out_progress;
 402
 403 #ifdef HAVE_STAT_NANOSECOND_PRECISION
 404         inode->i_creation_time = timespec_to_wim_timestamp(&stbuf.st_mtim);
 405         inode->i_last_write_time = timespec_to_wim_timestamp(&stbuf.st_mtim);
 406         inode->i_last_access_time = timespec_to_wim_timestamp(&stbuf.st_atim);
 407 #else
 408         inode->i_creation_time = time_t_to_wim_timestamp(stbuf.st_mtime);
 409         inode->i_last_write_time = time_t_to_wim_timestamp(stbuf.st_mtime);
 410         inode->i_last_access_time = time_t_to_wim_timestamp(stbuf.st_atime);
 411 #endif
 412         if (params->add_flags & WIMLIB_ADD_FLAG_UNIX_DATA) {
 413                 struct wimlib_unix_data unix_data;
 414
 415                 unix_data.uid = stbuf.st_uid;
 416                 unix_data.gid = stbuf.st_gid;
 417                 unix_data.mode = stbuf.st_mode;
 418                 unix_data.rdev = stbuf.st_rdev;
 419                 if (!inode_set_unix_data(inode, &unix_data, UNIX_DATA_ALL)) {
 420                         ret = WIMLIB_ERR_NOMEM;
 421                         goto out;
 422                 }
 423         }
 424
 425         if (params->add_flags & WIMLIB_ADD_FLAG_ROOT) {
 426                 params->capture_root_ino = stbuf.st_ino;
 427                 params->capture_root_dev = stbuf.st_dev;
 428                 params->add_flags &= ~WIMLIB_ADD_FLAG_ROOT;
 429         }
 430
 431         if (S_ISREG(stbuf.st_mode)) {
 432                 ret = unix_scan_regular_file(full_path, stbuf.st_size,
 433                                              inode, params->unhashed_blobs);
 434         } else if (S_ISDIR(stbuf.st_mode)) {
 435                 ret = unix_scan_directory(tree, full_path, full_path_len,
 436                                           dirfd, relpath, params);
 437         } else if (S_ISLNK(stbuf.st_mode)) {
 438                 ret = unix_scan_symlink(full_path, dirfd, relpath,
 439                                         inode, params);
 440         }
 441
 442         if (ret)
 443                 goto out;
 444
 445 out_progress:
 446         params->progress.scan.cur_path = full_path;
 447         if (likely(tree))
 448                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_OK, inode);
 449         else
 450                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_EXCLUDED, NULL);
 451 out:
 452         if (unlikely(ret)) {
 453                 free_dentry_tree(tree, params->blob_table);
 454                 tree = NULL;
 455                 ret = report_capture_error(params, ret, full_path);
 456         }
 457         *tree_ret = tree;
 458         return ret;
 459 }
 460
 461 /*
 462  * unix_build_dentry_tree():
 463  *      Builds a tree of WIM dentries from an on-disk directory tree (UNIX
 464  *      version; no NTFS-specific data is captured).
 465  *
 466  * @root_ret:   Place to return a pointer to the root of the dentry tree.  Only
 467  *              modified if successful.  Set to NULL if the file or directory was
 468  *              excluded from capture.
 469  *
 470  * @root_disk_path:  The path to the root of the directory tree on disk.
 471  *
 472  * @params:     See doc for `struct capture_params'.
 473  *
 474  * @return:     0 on success, nonzero on failure.  It is a failure if any of
 475  *              the files cannot be `stat'ed, or if any of the needed
 476  *              directories cannot be opened or read.  Failure to add the files
 477  *              to the WIM may still occur later when trying to actually read
 478  *              the on-disk files during a call to wimlib_write() or
 479  *              wimlib_overwrite().
 480  */
 481 int
 482 unix_build_dentry_tree(struct wim_dentry **root_ret,
 483                        const char *root_disk_path,
 484                        struct capture_params *params)
 485 {
 486         size_t path_len;
 487         size_t path_bufsz;
 488         char *path_buf;
 489         int ret;
 490
 491         path_len = strlen(root_disk_path);
 492         path_bufsz = min(32790, PATH_MAX + 1);
 493
 494         if (path_len >= path_bufsz)
 495                 return WIMLIB_ERR_INVALID_PARAM;
 496
 497         path_buf = MALLOC(path_bufsz);
 498         if (!path_buf)
 499                 return WIMLIB_ERR_NOMEM;
 500         memcpy(path_buf, root_disk_path, path_len + 1);
 501
 502         params->capture_root_nchars = path_len;
 503
 504         ret = unix_build_dentry_tree_recursive(root_ret, path_buf, path_len,
 505                                                AT_FDCWD, path_buf, params);
 506         FREE(path_buf);
 507         return ret;
 508 }
 509
 510 #endif /* !__WIN32__ */