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) 2012-2016 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/dentry.h"
  37 #include "wimlib/error.h"
  38 #include "wimlib/reparse.h"
  39 #include "wimlib/scan.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 blocks, u64 size,
 104                        struct wim_inode *inode,
 105                        struct list_head *unhashed_blobs)
 106 {
 107         struct blob_descriptor *blob = NULL;
 108         struct wim_inode_stream *strm;
 109
 110         /*
 111          * Set FILE_ATTRIBUTE_SPARSE_FILE if the file uses less disk space than
 112          * expected given its size.
 113          */
 114         if (blocks < DIV_ROUND_UP(size, 512))
 115                 inode->i_attributes = FILE_ATTRIBUTE_SPARSE_FILE;
 116         else
 117                 inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
 118
 119         if (size) {
 120                 blob = new_blob_descriptor();
 121                 if (unlikely(!blob))
 122                         goto err_nomem;
 123                 blob->file_on_disk = STRDUP(path);
 124                 if (unlikely(!blob->file_on_disk))
 125                         goto err_nomem;
 126                 blob->blob_location = BLOB_IN_FILE_ON_DISK;
 127                 blob->size = size;
 128                 blob->file_inode = inode;
 129         }
 130
 131         strm = inode_add_stream(inode, STREAM_TYPE_DATA, NO_STREAM_NAME, blob);
 132         if (unlikely(!strm))
 133                 goto err_nomem;
 134
 135         prepare_unhashed_blob(blob, inode, strm->stream_id, unhashed_blobs);
 136         return 0;
 137
 138 err_nomem:
 139         free_blob_descriptor(blob);
 140         return WIMLIB_ERR_NOMEM;
 141 }
 142
 143 static int
 144 unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 145                                  char *path, size_t path_len,
 146                                  int dirfd, const char *relpath,
 147                                  struct scan_params *params);
 148
 149 static int
 150 unix_scan_directory(struct wim_dentry *dir_dentry,
 151                     char *full_path, size_t full_path_len,
 152                     int parent_dirfd, const char *dir_relpath,
 153                     struct scan_params *params)
 154 {
 155
 156         int dirfd;
 157         DIR *dir;
 158         int ret;
 159
 160         dirfd = my_openat(full_path, parent_dirfd, dir_relpath, O_RDONLY);
 161         if (dirfd < 0) {
 162                 ERROR_WITH_ERRNO("\"%s\": Can't open directory", full_path);
 163                 return WIMLIB_ERR_OPENDIR;
 164         }
 165
 166         dir_dentry->d_inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 167         dir = my_fdopendir(&dirfd);
 168         if (!dir) {
 169                 ERROR_WITH_ERRNO("\"%s\": Can't open directory", full_path);
 170                 close(dirfd);
 171                 return WIMLIB_ERR_OPENDIR;
 172         }
 173
 174         ret = 0;
 175         for (;;) {
 176                 struct dirent *entry;
 177                 struct wim_dentry *child;
 178                 size_t name_len;
 179
 180                 errno = 0;
 181                 entry = readdir(dir);
 182                 if (!entry) {
 183                         if (errno) {
 184                                 ret = WIMLIB_ERR_READ;
 185                                 ERROR_WITH_ERRNO("\"%s\": Error reading directory",
 186                                                  full_path);
 187                         }
 188                         break;
 189                 }
 190
 191                 name_len = strlen(entry->d_name);
 192
 193                 if (should_ignore_filename(entry->d_name, name_len))
 194                         continue;
 195
 196                 full_path[full_path_len] = '/';
 197                 memcpy(&full_path[full_path_len + 1], entry->d_name, name_len + 1);
 198                 ret = unix_build_dentry_tree_recursive(&child,
 199                                                        full_path,
 200                                                        full_path_len + 1 + name_len,
 201                                                        dirfd,
 202                                                        &full_path[full_path_len + 1],
 203                                                        params);
 204                 full_path[full_path_len] = '\0';
 205                 if (ret)
 206                         break;
 207                 attach_scanned_tree(dir_dentry, child, params->blob_table);
 208         }
 209         closedir(dir);
 210         return ret;
 211 }
 212
 213 /*
 214  * Given an absolute symbolic link target (UNIX-style, beginning with '/'),
 215  * determine whether it points into the directory identified by @ino and @dev.
 216  * If yes, return the suffix of @target which is relative to this directory, but
 217  * retaining leading slashes.  If no, return @target.
 218  *
 219  * Here are some examples, assuming that the @ino/@dev directory is "/home/e":
 220  *
 221  *      Original target         New target
 222  *      ---------------         ----------
 223  *      /home/e/test            /test
 224  *      /home/e/test/           /test/
 225  *      //home//e//test//       //test//
 226  *      /home/e                                         (empty string)
 227  *      /home/e/                /
 228  *      /usr/lib                /usr/lib                (external link)
 229  *
 230  * Because of the possibility of other links into the @ino/@dev directory and/or
 231  * multiple path separators, we can't simply do a string comparison; instead we
 232  * need to stat() each ancestor directory.
 233  *
 234  * If the link points directly to the @ino/@dev directory with no trailing
 235  * slashes, then the new target will be an empty string.  This is not a valid
 236  * UNIX symlink target, but we store this in the archive anyway since the target
 237  * is intended to be de-relativized when the link is extracted.
 238  */
 239 static char *
 240 unix_relativize_link_target(char *target, u64 ino, u64 dev)
 241 {
 242         char *p = target;
 243
 244         do {
 245                 char save;
 246                 struct stat stbuf;
 247                 int ret;
 248
 249                 /* Skip slashes (guaranteed to be at least one here)  */
 250                 do {
 251                         p++;
 252                 } while (*p == '/');
 253
 254                 /* End of string?  */
 255                 if (!*p)
 256                         break;
 257
 258                 /* Skip non-slashes (guaranteed to be at least one here)  */
 259                 do {
 260                         p++;
 261                 } while (*p && *p != '/');
 262
 263                 /* Get the inode and device numbers for this prefix.  */
 264                 save = *p;
 265                 *p = '\0';
 266                 ret = stat(target, &stbuf);
 267                 *p = save;
 268
 269                 if (ret) {
 270                         /* stat() failed.  Assume the link points outside the
 271                          * directory tree being captured.  */
 272                         break;
 273                 }
 274
 275                 if (stbuf.st_ino == ino && stbuf.st_dev == dev) {
 276                         /* Link points inside directory tree being captured.
 277                          * Return abbreviated path.  */
 278                         return p;
 279                 }
 280         } while (*p);
 281
 282         /* Link does not point inside directory tree being captured.  */
 283         return target;
 284 }
 285
 286 static noinline_for_stack int
 287 unix_scan_symlink(const char *full_path, int dirfd, const char *relpath,
 288                   struct wim_inode *inode, struct scan_params *params)
 289 {
 290         char orig_target[REPARSE_POINT_MAX_SIZE];
 291         char *target = orig_target;
 292         int ret;
 293
 294         /* Read the UNIX symbolic link target.  */
 295         ret = my_readlinkat(full_path, dirfd, relpath, target,
 296                             sizeof(orig_target));
 297         if (unlikely(ret < 0)) {
 298                 ERROR_WITH_ERRNO("\"%s\": Can't read target of symbolic link",
 299                                  full_path);
 300                 return WIMLIB_ERR_READLINK;
 301         }
 302         if (unlikely(ret >= sizeof(orig_target))) {
 303                 ERROR("\"%s\": target of symbolic link is too long", full_path);
 304                 return WIMLIB_ERR_READLINK;
 305         }
 306         target[ret] = '\0';
 307
 308         /* If the link is absolute and reparse point fixups are enabled, then
 309          * change it to be "absolute" relative to the tree being captured.  */
 310         if (target[0] == '/' && (params->add_flags & WIMLIB_ADD_FLAG_RPFIX)) {
 311                 int status = WIMLIB_SCAN_DENTRY_NOT_FIXED_SYMLINK;
 312
 313                 params->progress.scan.cur_path = full_path;
 314                 params->progress.scan.symlink_target = target;
 315
 316                 target = unix_relativize_link_target(target,
 317                                                      params->capture_root_ino,
 318                                                      params->capture_root_dev);
 319                 if (target != orig_target) {
 320                         /* Link target was fixed.  */
 321                         inode->i_rp_flags &= ~WIM_RP_FLAG_NOT_FIXED;
 322                         status = WIMLIB_SCAN_DENTRY_FIXED_SYMLINK;
 323                 }
 324                 ret = do_scan_progress(params, status, NULL);
 325                 if (ret)
 326                         return ret;
 327         }
 328
 329         /* Translate the UNIX symlink target into a Windows reparse point.  */
 330         ret = wim_inode_set_symlink(inode, target, params->blob_table);
 331         if (unlikely(ret)) {
 332                 if (ret == WIMLIB_ERR_INVALID_UTF8_STRING) {
 333                         ERROR("\"%s\": target of symbolic link is not valid "
 334                               "UTF-8.  This is not supported.", full_path);
 335                 }
 336                 return ret;
 337         }
 338
 339         /* On Windows, a reparse point can be set on both directory and
 340          * non-directory files.  Usually, a link that is intended to point to a
 341          * (non-)directory is stored as a reparse point on a (non-)directory
 342          * file.  Replicate this behavior by examining the target file.  */
 343         struct stat stbuf;
 344         if (my_fstatat(full_path, dirfd, relpath, &stbuf, 0) == 0 &&
 345             S_ISDIR(stbuf.st_mode))
 346                 inode->i_attributes |= FILE_ATTRIBUTE_DIRECTORY;
 347         return 0;
 348 }
 349
 350 static int
 351 unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 352                                  char *full_path, size_t full_path_len,
 353                                  int dirfd, const char *relpath,
 354                                  struct scan_params *params)
 355 {
 356         struct wim_dentry *tree = NULL;
 357         struct wim_inode *inode = NULL;
 358         int ret;
 359         struct stat stbuf;
 360         int stat_flags;
 361
 362         ret = try_exclude(full_path, params);
 363         if (unlikely(ret < 0)) /* Excluded? */
 364                 goto out_progress;
 365         if (unlikely(ret > 0)) /* Error? */
 366                 goto out;
 367
 368         if (params->add_flags & (WIMLIB_ADD_FLAG_DEREFERENCE |
 369                                  WIMLIB_ADD_FLAG_ROOT))
 370                 stat_flags = 0;
 371         else
 372                 stat_flags = AT_SYMLINK_NOFOLLOW;
 373
 374         ret = my_fstatat(full_path, dirfd, relpath, &stbuf, stat_flags);
 375
 376         if (ret) {
 377                 ERROR_WITH_ERRNO("\"%s\": Can't read metadata", full_path);
 378                 ret = WIMLIB_ERR_STAT;
 379                 goto out;
 380         }
 381
 382         if (!(params->add_flags & WIMLIB_ADD_FLAG_UNIX_DATA)) {
 383                 if (unlikely(!S_ISREG(stbuf.st_mode) &&
 384                              !S_ISDIR(stbuf.st_mode) &&
 385                              !S_ISLNK(stbuf.st_mode)))
 386                 {
 387                         if (params->add_flags &
 388                             WIMLIB_ADD_FLAG_NO_UNSUPPORTED_EXCLUDE)
 389                         {
 390                                 ERROR("\"%s\": File type is unsupported",
 391                                       full_path);
 392                                 ret = WIMLIB_ERR_UNSUPPORTED_FILE;
 393                                 goto out;
 394                         }
 395                         params->progress.scan.cur_path = full_path;
 396                         ret = do_scan_progress(params,
 397                                                WIMLIB_SCAN_DENTRY_UNSUPPORTED,
 398                                                NULL);
 399                         goto out;
 400                 }
 401         }
 402
 403         ret = inode_table_new_dentry(params->inode_table, relpath,
 404                                      stbuf.st_ino, stbuf.st_dev, false, &tree);
 405         if (unlikely(ret)) {
 406                 if (ret == WIMLIB_ERR_INVALID_UTF8_STRING) {
 407                         ERROR("\"%s\": filename is not valid UTF-8.  "
 408                               "This is not supported.", full_path);
 409                 }
 410                 goto out;
 411         }
 412
 413         inode = tree->d_inode;
 414
 415         /* Already seen this inode?  */
 416         if (inode->i_nlink > 1)
 417                 goto out_progress;
 418
 419 #ifdef HAVE_STAT_NANOSECOND_PRECISION
 420         inode->i_creation_time = timespec_to_wim_timestamp(&stbuf.st_mtim);
 421         inode->i_last_write_time = timespec_to_wim_timestamp(&stbuf.st_mtim);
 422         inode->i_last_access_time = timespec_to_wim_timestamp(&stbuf.st_atim);
 423 #else
 424         inode->i_creation_time = time_t_to_wim_timestamp(stbuf.st_mtime);
 425         inode->i_last_write_time = time_t_to_wim_timestamp(stbuf.st_mtime);
 426         inode->i_last_access_time = time_t_to_wim_timestamp(stbuf.st_atime);
 427 #endif
 428         if (params->add_flags & WIMLIB_ADD_FLAG_UNIX_DATA) {
 429                 struct wimlib_unix_data unix_data;
 430
 431                 unix_data.uid = stbuf.st_uid;
 432                 unix_data.gid = stbuf.st_gid;
 433                 unix_data.mode = stbuf.st_mode;
 434                 unix_data.rdev = stbuf.st_rdev;
 435                 if (!inode_set_unix_data(inode, &unix_data, UNIX_DATA_ALL)) {
 436                         ret = WIMLIB_ERR_NOMEM;
 437                         goto out;
 438                 }
 439         }
 440
 441         if (params->add_flags & WIMLIB_ADD_FLAG_ROOT) {
 442                 params->capture_root_ino = stbuf.st_ino;
 443                 params->capture_root_dev = stbuf.st_dev;
 444                 params->add_flags &= ~WIMLIB_ADD_FLAG_ROOT;
 445         }
 446
 447         if (S_ISREG(stbuf.st_mode)) {
 448                 ret = unix_scan_regular_file(full_path, stbuf.st_blocks,
 449                                              stbuf.st_size, inode,
 450                                              params->unhashed_blobs);
 451         } else if (S_ISDIR(stbuf.st_mode)) {
 452                 ret = unix_scan_directory(tree, full_path, full_path_len,
 453                                           dirfd, relpath, params);
 454         } else if (S_ISLNK(stbuf.st_mode)) {
 455                 ret = unix_scan_symlink(full_path, dirfd, relpath,
 456                                         inode, params);
 457         }
 458
 459         if (ret)
 460                 goto out;
 461
 462 out_progress:
 463         params->progress.scan.cur_path = full_path;
 464         if (likely(tree))
 465                 ret = do_scan_progress(params, WIMLIB_SCAN_DENTRY_OK, inode);
 466         else
 467                 ret = do_scan_progress(params, WIMLIB_SCAN_DENTRY_EXCLUDED, NULL);
 468 out:
 469         if (unlikely(ret)) {
 470                 free_dentry_tree(tree, params->blob_table);
 471                 tree = NULL;
 472                 ret = report_scan_error(params, ret, full_path);
 473         }
 474         *tree_ret = tree;
 475         return ret;
 476 }
 477
 478 /*
 479  * unix_build_dentry_tree():
 480  *      Builds a tree of WIM dentries from an on-disk directory tree (UNIX
 481  *      version; no NTFS-specific data is captured).
 482  *
 483  * @root_ret:   Place to return a pointer to the root of the dentry tree.  Set
 484  *              to NULL if the file or directory was excluded from capture.
 485  *
 486  * @root_disk_path:  The path to the root of the directory tree on disk.
 487  *
 488  * @params:     See doc for `struct scan_params'.
 489  *
 490  * @return:     0 on success, nonzero on failure.  It is a failure if any of
 491  *              the files cannot be `stat'ed, or if any of the needed
 492  *              directories cannot be opened or read.  Failure to add the files
 493  *              to the WIM may still occur later when trying to actually read
 494  *              the on-disk files during a call to wimlib_write() or
 495  *              wimlib_overwrite().
 496  */
 497 int
 498 unix_build_dentry_tree(struct wim_dentry **root_ret,
 499                        const char *root_disk_path, struct scan_params *params)
 500 {
 501         size_t path_len;
 502         size_t path_bufsz;
 503         char *path_buf;
 504         int ret;
 505
 506         path_len = strlen(root_disk_path);
 507         path_bufsz = min(32790, PATH_MAX + 1);
 508
 509         if (path_len >= path_bufsz)
 510                 return WIMLIB_ERR_INVALID_PARAM;
 511
 512         path_buf = MALLOC(path_bufsz);
 513         if (!path_buf)
 514                 return WIMLIB_ERR_NOMEM;
 515         memcpy(path_buf, root_disk_path, path_len + 1);
 516
 517         params->capture_root_nchars = path_len;
 518
 519         ret = unix_build_dentry_tree_recursive(root_ret, path_buf, path_len,
 520                                                AT_FDCWD, path_buf, params);
 521         FREE(path_buf);
 522         return ret;
 523 }
 524
 525 #endif /* !__WIN32__ */