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 part of wimlib, a library for working with WIM files.
   9  *
  10  * wimlib is free software; you can redistribute it and/or modify it under the
  11  * terms of the GNU General Public License as published by the Free
  12  * Software Foundation; either version 3 of the License, or (at your option)
  13  * any later version.
  14  *
  15  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  16  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  17  * A PARTICULAR PURPOSE. See the GNU General Public License for more
  18  * details.
  19  *
  20  * You should have received a copy of the GNU General Public License
  21  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  22  */
  23
  24 #ifndef __WIN32__
  25
  26 #ifdef HAVE_CONFIG_H
  27 #  include "config.h"
  28 #endif
  29
  30 #include <dirent.h>
  31 #include <errno.h>
  32 #include <fcntl.h>
  33 #include <limits.h> /* for PATH_MAX */
  34 #include <sys/stat.h>
  35 #include <unistd.h>
  36
  37 #include "wimlib/capture.h"
  38 #include "wimlib/dentry.h"
  39 #include "wimlib/error.h"
  40 #include "wimlib/lookup_table.h"
  41 #include "wimlib/reparse.h"
  42 #include "wimlib/timestamp.h"
  43 #include "wimlib/unix_data.h"
  44
  45 #ifdef HAVE_FDOPENDIR
  46 #  define my_fdopendir(dirfd_p) fdopendir(*(dirfd_p))
  47 #else
  48 static DIR *
  49 my_fdopendir(int *dirfd_p)
  50 {
  51         DIR *dir = NULL;
  52         int old_pwd;
  53
  54         old_pwd = open(".", O_RDONLY);
  55         if (old_pwd >= 0) {
  56                 if (!fchdir(*dirfd_p)) {
  57                         dir = opendir(".");
  58                         if (dir) {
  59                                 close(*dirfd_p);
  60                                 *dirfd_p = dirfd(dir);
  61                         }
  62                         fchdir(old_pwd);
  63                 }
  64                 close(old_pwd);
  65         }
  66         return dir;
  67 }
  68 #endif
  69
  70 #ifdef HAVE_OPENAT
  71 #  define my_openat(full_path, dirfd, relpath, flags) \
  72                 openat((dirfd), (relpath), (flags))
  73 #else
  74 #  define my_openat(full_path, dirfd, relpath, flags) \
  75                 open((full_path), (flags))
  76 #endif
  77
  78 #ifdef HAVE_READLINKAT
  79 #  define my_readlinkat(full_path, dirfd, relpath, buf, bufsize) \
  80                 readlinkat((dirfd), (relpath), (buf), (bufsize))
  81 #else
  82 #  define my_readlinkat(full_path, dirfd, relpath, buf, bufsize) \
  83                 readlink((full_path), (buf), (bufsize))
  84 #endif
  85
  86 #ifdef HAVE_FSTATAT
  87 #  define my_fstatat(full_path, dirfd, relpath, stbuf, flags)   \
  88         fstatat((dirfd), (relpath), (stbuf), (flags))
  89 #else
  90 #  define my_fstatat(full_path, dirfd, relpath, stbuf, flags)   \
  91         ((flags) & AT_SYMLINK_NOFOLLOW) ? \
  92                 lstat((full_path), (stbuf)) : \
  93                 stat((full_path), (stbuf))
  94 #endif
  95
  96 #ifndef AT_FDCWD
  97 #  define AT_FDCWD      -100
  98 #endif
  99
 100 #ifndef AT_SYMLINK_NOFOLLOW
 101 #  define AT_SYMLINK_NOFOLLOW   0x100
 102 #endif
 103
 104 static int
 105 unix_scan_regular_file(const char *path, u64 size, struct wim_inode *inode,
 106                        struct list_head *unhashed_streams)
 107 {
 108         struct wim_lookup_table_entry *lte;
 109         char *file_on_disk;
 110
 111         inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
 112
 113         /* Empty files do not have to have a lookup table entry. */
 114         if (!size)
 115                 return 0;
 116
 117         file_on_disk = STRDUP(path);
 118         if (!file_on_disk)
 119                 return WIMLIB_ERR_NOMEM;
 120         lte = new_lookup_table_entry();
 121         if (!lte) {
 122                 FREE(file_on_disk);
 123                 return WIMLIB_ERR_NOMEM;
 124         }
 125         lte->file_on_disk = file_on_disk;
 126         lte->resource_location = RESOURCE_IN_FILE_ON_DISK;
 127         lte->size = size;
 128         add_unhashed_stream(lte, inode, 0, unhashed_streams);
 129         inode->i_lte = lte;
 130         return 0;
 131 }
 132
 133 static int
 134 unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 135                                  char *path, size_t path_len,
 136                                  int dirfd, const char *relpath,
 137                                  struct add_image_params *params);
 138
 139 static int
 140 unix_scan_directory(struct wim_dentry *dir_dentry,
 141                     char *full_path, size_t full_path_len,
 142                     int parent_dirfd, const char *dir_relpath,
 143                     struct add_image_params *params)
 144 {
 145
 146         int dirfd;
 147         DIR *dir;
 148         int ret;
 149
 150         dirfd = my_openat(full_path, parent_dirfd, dir_relpath, O_RDONLY);
 151         if (dirfd < 0) {
 152                 ERROR_WITH_ERRNO("\"%s\": Can't open directory", full_path);
 153                 return WIMLIB_ERR_OPENDIR;
 154         }
 155
 156         dir_dentry->d_inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 157         dir = my_fdopendir(&dirfd);
 158         if (!dir) {
 159                 ERROR_WITH_ERRNO("\"%s\": Can't open directory", full_path);
 160                 close(dirfd);
 161                 return WIMLIB_ERR_OPENDIR;
 162         }
 163
 164         ret = 0;
 165         for (;;) {
 166                 struct dirent *entry;
 167                 struct wim_dentry *child;
 168                 size_t name_len;
 169
 170                 errno = 0;
 171                 entry = readdir(dir);
 172                 if (!entry) {
 173                         if (errno) {
 174                                 ret = WIMLIB_ERR_READ;
 175                                 ERROR_WITH_ERRNO("\"%s\": Error reading directory",
 176                                                  full_path);
 177                         }
 178                         break;
 179                 }
 180
 181                 if (entry->d_name[0] == '.' &&
 182                     (entry->d_name[1] == '\0' ||
 183                      (entry->d_name[1] == '.' && entry->d_name[2] == '\0')))
 184                         continue;
 185
 186                 full_path[full_path_len] = '/';
 187                 name_len = strlen(entry->d_name);
 188                 memcpy(&full_path[full_path_len + 1], entry->d_name, name_len + 1);
 189                 ret = unix_build_dentry_tree_recursive(&child,
 190                                                        full_path,
 191                                                        full_path_len + 1 + name_len,
 192                                                        dirfd,
 193                                                        &full_path[full_path_len + 1],
 194                                                        params);
 195                 full_path[full_path_len] = '\0';
 196                 if (ret)
 197                         break;
 198                 if (child)
 199                         dentry_add_child(dir_dentry, child);
 200         }
 201         closedir(dir);
 202         return ret;
 203 }
 204
 205 /* Given an absolute symbolic link target @dest (UNIX-style, beginning
 206  * with '/'), determine whether it points into the directory specified by
 207  * @ino and @dev.  If so, return the target modified to be "absolute"
 208  * relative to this directory.  Otherwise, return NULL.  */
 209 static char *
 210 unix_fixup_abslink(char *dest, u64 ino, u64 dev)
 211 {
 212         char *p = dest;
 213
 214         do {
 215                 char save;
 216                 struct stat stbuf;
 217                 int ret;
 218
 219                 /* Skip non-slashes.  */
 220                 while (*p && *p != '/')
 221                         p++;
 222
 223                 /* Skip slashes.  */
 224                 while (*p && *p == '/')
 225                         p++;
 226
 227                 /* Get inode and device for this prefix.  */
 228                 save = *p;
 229                 *p = '\0';
 230                 ret = stat(dest, &stbuf);
 231                 *p = save;
 232
 233                 if (ret) {
 234                         /* stat() failed.  Assume the link points outside the
 235                          * directory tree being captured.  */
 236                         break;
 237                 }
 238
 239                 if (stbuf.st_ino == ino && stbuf.st_dev == dev) {
 240                         /* Link points inside directory tree being captured.
 241                          * Return abbreviated path.  */
 242                         *--p = '/';
 243                         while (p > dest && *(p - 1) == '/')
 244                                 p--;
 245                         return p;
 246                 }
 247         } while (*p);
 248
 249         /* Link does not point inside directory tree being captured.  */
 250         return NULL;
 251 }
 252
 253 static int
 254 unix_scan_symlink(struct wim_dentry **root_p, const char *full_path,
 255                   int dirfd, const char *relpath,
 256                   struct wim_inode *inode, struct add_image_params *params)
 257 {
 258         char deref_name_buf[4096];
 259         ssize_t deref_name_len;
 260         char *dest;
 261         int ret;
 262
 263         inode->i_attributes = FILE_ATTRIBUTE_REPARSE_POINT;
 264         inode->i_reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
 265
 266         /* The idea here is to call readlink() to get the UNIX target of the
 267          * symbolic link, then turn the target into a reparse point data buffer
 268          * that contains a relative or absolute symbolic link. */
 269         deref_name_len = my_readlinkat(full_path, dirfd, relpath,
 270                                        deref_name_buf, sizeof(deref_name_buf) - 1);
 271         if (deref_name_len < 0) {
 272                 ERROR_WITH_ERRNO("\"%s\": Can't read target of symbolic link",
 273                                  full_path);
 274                 return WIMLIB_ERR_READLINK;
 275         }
 276
 277         dest = deref_name_buf;
 278
 279         dest[deref_name_len] = '\0';
 280
 281         if ((params->add_flags & WIMLIB_ADD_FLAG_RPFIX) &&
 282              dest[0] == '/')
 283         {
 284                 /* RPFIX (reparse point fixup) mode:  Change target of absolute
 285                  * symbolic link to be "absolute" relative to the tree being
 286                  * captured.  */
 287                 dest = unix_fixup_abslink(dest,
 288                                           params->capture_root_ino,
 289                                           params->capture_root_dev);
 290                 params->progress.scan.cur_path = full_path;
 291                 params->progress.scan.symlink_target = deref_name_buf;
 292                 if (dest) {
 293                         /* Successfully fixed the link target.  */
 294                         inode->i_not_rpfixed = 0;
 295                         ret = do_capture_progress(params,
 296                                                   WIMLIB_SCAN_DENTRY_FIXED_SYMLINK,
 297                                                   NULL);
 298                         if (ret)
 299                                 return ret;
 300                 } else {
 301                         /* Link points outside of the tree being captured.
 302                          * Exclude it.  */
 303                         free_dentry(*root_p);
 304                         *root_p = NULL;
 305                         return do_capture_progress(params,
 306                                                    WIMLIB_SCAN_DENTRY_EXCLUDED_SYMLINK,
 307                                                    NULL);
 308                 }
 309         }
 310         ret = wim_inode_set_symlink(inode, dest, params->lookup_table);
 311         if (ret)
 312                 return ret;
 313
 314         /* Unfortunately, Windows seems to have the concept of "file" symbolic
 315          * links as being different from "directory" symbolic links...  so
 316          * FILE_ATTRIBUTE_DIRECTORY needs to be set on the symbolic link if the
 317          * *target* of the symbolic link is a directory.  */
 318         struct stat stbuf;
 319         if (my_fstatat(full_path, dirfd, relpath, &stbuf, 0) == 0 &&
 320             S_ISDIR(stbuf.st_mode))
 321                 inode->i_attributes |= FILE_ATTRIBUTE_DIRECTORY;
 322         return 0;
 323 }
 324
 325 static int
 326 unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 327                                  char *full_path, size_t full_path_len,
 328                                  int dirfd, const char *relpath,
 329                                  struct add_image_params *params)
 330 {
 331         struct wim_dentry *tree = NULL;
 332         struct wim_inode *inode = NULL;
 333         int ret;
 334         struct stat stbuf;
 335         int stat_flags;
 336
 337         if (should_exclude_path(full_path + params->capture_root_nchars,
 338                                 full_path_len - params->capture_root_nchars,
 339                                 params->config))
 340                 goto out_progress;
 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(&tree, full_path, dirfd, relpath,
 426                                         inode, params);
 427                 if (!tree)
 428                         goto out;
 429         }
 430
 431         if (ret)
 432                 goto out;
 433
 434 out_progress:
 435         params->progress.scan.cur_path = full_path;
 436         if (likely(tree))
 437                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_OK, inode);
 438         else
 439                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_EXCLUDED, NULL);
 440 out:
 441         if (likely(ret == 0))
 442                 *tree_ret = tree;
 443         else
 444                 free_dentry_tree(tree, params->lookup_table);
 445         return ret;
 446 }
 447
 448 /*
 449  * unix_build_dentry_tree():
 450  *      Builds a tree of WIM dentries from an on-disk directory tree (UNIX
 451  *      version; no NTFS-specific data is captured).
 452  *
 453  * @root_ret:   Place to return a pointer to the root of the dentry tree.  Only
 454  *              modified if successful.  Set to NULL if the file or directory was
 455  *              excluded from capture.
 456  *
 457  * @root_disk_path:  The path to the root of the directory tree on disk.
 458  *
 459  * @params:     See doc for `struct add_image_params'.
 460  *
 461  * @return:     0 on success, nonzero on failure.  It is a failure if any of
 462  *              the files cannot be `stat'ed, or if any of the needed
 463  *              directories cannot be opened or read.  Failure to add the files
 464  *              to the WIM may still occur later when trying to actually read
 465  *              the on-disk files during a call to wimlib_write() or
 466  *              wimlib_overwrite().
 467  */
 468 int
 469 unix_build_dentry_tree(struct wim_dentry **root_ret,
 470                        const char *root_disk_path,
 471                        struct add_image_params *params)
 472 {
 473         size_t path_len;
 474         size_t path_bufsz;
 475         char *path_buf;
 476         int ret;
 477
 478         path_len = strlen(root_disk_path);
 479         path_bufsz = min(32790, PATH_MAX + 1);
 480
 481         if (path_len >= path_bufsz)
 482                 return WIMLIB_ERR_INVALID_PARAM;
 483
 484         path_buf = MALLOC(path_bufsz);
 485         if (!path_buf)
 486                 return WIMLIB_ERR_NOMEM;
 487         memcpy(path_buf, root_disk_path, path_len + 1);
 488
 489         params->capture_root_nchars = path_len;
 490
 491         ret = unix_build_dentry_tree_recursive(root_ret, path_buf, path_len,
 492                                                AT_FDCWD, path_buf, params);
 493         FREE(path_buf);
 494         return ret;
 495 }
 496
 497 #endif /* !__WIN32__ */