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                 if (entry->d_name[0] == '.' &&
 184                     (entry->d_name[1] == '\0' ||
 185                      (entry->d_name[1] == '.' && entry->d_name[2] == '\0')))
 186                         continue;
 187
 188                 full_path[full_path_len] = '/';
 189                 name_len = strlen(entry->d_name);
 190                 memcpy(&full_path[full_path_len + 1], entry->d_name, name_len + 1);
 191                 ret = unix_build_dentry_tree_recursive(&child,
 192                                                        full_path,
 193                                                        full_path_len + 1 + name_len,
 194                                                        dirfd,
 195                                                        &full_path[full_path_len + 1],
 196                                                        params);
 197                 full_path[full_path_len] = '\0';
 198                 if (ret)
 199                         break;
 200                 if (child)
 201                         dentry_add_child(dir_dentry, child);
 202         }
 203         closedir(dir);
 204         return ret;
 205 }
 206
 207 /* Given an absolute symbolic link target @dest (UNIX-style, beginning
 208  * with '/'), determine whether it points into the directory specified by
 209  * @ino and @dev.  If so, return the target modified to be "absolute"
 210  * relative to this directory.  Otherwise, return NULL.  */
 211 static char *
 212 unix_fixup_abslink(char *dest, u64 ino, u64 dev)
 213 {
 214         char *p = dest;
 215
 216         do {
 217                 char save;
 218                 struct stat stbuf;
 219                 int ret;
 220
 221                 /* Skip non-slashes.  */
 222                 while (*p && *p != '/')
 223                         p++;
 224
 225                 /* Skip slashes.  */
 226                 while (*p && *p == '/')
 227                         p++;
 228
 229                 /* Get inode and device for this prefix.  */
 230                 save = *p;
 231                 *p = '\0';
 232                 ret = stat(dest, &stbuf);
 233                 *p = save;
 234
 235                 if (ret) {
 236                         /* stat() failed.  Assume the link points outside the
 237                          * directory tree being captured.  */
 238                         break;
 239                 }
 240
 241                 if (stbuf.st_ino == ino && stbuf.st_dev == dev) {
 242                         /* Link points inside directory tree being captured.
 243                          * Return abbreviated path.  */
 244                         *--p = '/';
 245                         while (p > dest && *(p - 1) == '/')
 246                                 p--;
 247                         return p;
 248                 }
 249         } while (*p);
 250
 251         /* Link does not point inside directory tree being captured.  */
 252         return NULL;
 253 }
 254
 255 static int
 256 unix_scan_symlink(const char *full_path, int dirfd, const char *relpath,
 257                   struct wim_inode *inode, struct capture_params *params)
 258 {
 259         char deref_name_buf[4096];
 260         ssize_t deref_name_len;
 261         char *dest;
 262         int ret;
 263
 264         inode->i_attributes = FILE_ATTRIBUTE_REPARSE_POINT;
 265         inode->i_reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
 266
 267         /* The idea here is to call readlink() to get the UNIX target of the
 268          * symbolic link, then turn the target into a reparse point data buffer
 269          * that contains a relative or absolute symbolic link. */
 270         deref_name_len = my_readlinkat(full_path, dirfd, relpath,
 271                                        deref_name_buf, sizeof(deref_name_buf) - 1);
 272         if (deref_name_len < 0) {
 273                 ERROR_WITH_ERRNO("\"%s\": Can't read target of symbolic link",
 274                                  full_path);
 275                 return WIMLIB_ERR_READLINK;
 276         }
 277
 278         dest = deref_name_buf;
 279
 280         dest[deref_name_len] = '\0';
 281
 282         if ((params->add_flags & WIMLIB_ADD_FLAG_RPFIX) &&
 283              dest[0] == '/')
 284         {
 285                 char *fixed_dest;
 286
 287                 /* RPFIX (reparse point fixup) mode:  Change target of absolute
 288                  * symbolic link to be "absolute" relative to the tree being
 289                  * captured.  */
 290                 fixed_dest = unix_fixup_abslink(dest,
 291                                                 params->capture_root_ino,
 292                                                 params->capture_root_dev);
 293                 params->progress.scan.cur_path = full_path;
 294                 params->progress.scan.symlink_target = deref_name_buf;
 295                 if (fixed_dest) {
 296                         /* Link points inside the tree being captured, so it was
 297                          * fixed.  */
 298                         inode->i_not_rpfixed = 0;
 299                         dest = fixed_dest;
 300                         ret = do_capture_progress(params,
 301                                                   WIMLIB_SCAN_DENTRY_FIXED_SYMLINK,
 302                                                   NULL);
 303                 } else {
 304                         /* Link points outside the tree being captured, so it
 305                          * was not fixed.  */
 306                         ret = do_capture_progress(params,
 307                                                   WIMLIB_SCAN_DENTRY_NOT_FIXED_SYMLINK,
 308                                                   NULL);
 309                 }
 310                 if (ret)
 311                         return ret;
 312         }
 313         ret = wim_inode_set_symlink(inode, dest, params->blob_table);
 314         if (ret)
 315                 return ret;
 316
 317         /* Unfortunately, Windows seems to have the concept of "file" symbolic
 318          * links as being different from "directory" symbolic links...  so
 319          * FILE_ATTRIBUTE_DIRECTORY needs to be set on the symbolic link if the
 320          * *target* of the symbolic link is a directory.  */
 321         struct stat stbuf;
 322         if (my_fstatat(full_path, dirfd, relpath, &stbuf, 0) == 0 &&
 323             S_ISDIR(stbuf.st_mode))
 324                 inode->i_attributes |= FILE_ATTRIBUTE_DIRECTORY;
 325         return 0;
 326 }
 327
 328 static int
 329 unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 330                                  char *full_path, size_t full_path_len,
 331                                  int dirfd, const char *relpath,
 332                                  struct capture_params *params)
 333 {
 334         struct wim_dentry *tree = NULL;
 335         struct wim_inode *inode = NULL;
 336         int ret;
 337         struct stat stbuf;
 338         int stat_flags;
 339
 340         ret = try_exclude(full_path, full_path_len, params);
 341         if (ret < 0) /* Excluded? */
 342                 goto out_progress;
 343         if (ret > 0) /* Error? */
 344                 goto out;
 345
 346         if (params->add_flags & (WIMLIB_ADD_FLAG_DEREFERENCE |
 347                                  WIMLIB_ADD_FLAG_ROOT))
 348                 stat_flags = 0;
 349         else
 350                 stat_flags = AT_SYMLINK_NOFOLLOW;
 351
 352         ret = my_fstatat(full_path, dirfd, relpath, &stbuf, stat_flags);
 353
 354         if (ret) {
 355                 ERROR_WITH_ERRNO("\"%s\": Can't read metadata", full_path);
 356                 ret = WIMLIB_ERR_STAT;
 357                 goto out;
 358         }
 359
 360         if (!(params->add_flags & WIMLIB_ADD_FLAG_UNIX_DATA)) {
 361                 if (unlikely(!S_ISREG(stbuf.st_mode) &&
 362                              !S_ISDIR(stbuf.st_mode) &&
 363                              !S_ISLNK(stbuf.st_mode)))
 364                 {
 365                         if (params->add_flags &
 366                             WIMLIB_ADD_FLAG_NO_UNSUPPORTED_EXCLUDE)
 367                         {
 368                                 ERROR("\"%s\": File type is unsupported",
 369                                       full_path);
 370                                 ret = WIMLIB_ERR_UNSUPPORTED_FILE;
 371                                 goto out;
 372                         }
 373                         params->progress.scan.cur_path = full_path;
 374                         ret = do_capture_progress(params,
 375                                                   WIMLIB_SCAN_DENTRY_UNSUPPORTED,
 376                                                   NULL);
 377                         goto out;
 378                 }
 379         }
 380
 381         ret = inode_table_new_dentry(params->inode_table, relpath,
 382                                      stbuf.st_ino, stbuf.st_dev,
 383                                      S_ISDIR(stbuf.st_mode), &tree);
 384         if (ret)
 385                 goto out;
 386
 387         inode = tree->d_inode;
 388
 389         /* Already seen this inode?  */
 390         if (inode->i_nlink > 1)
 391                 goto out_progress;
 392
 393 #ifdef HAVE_STAT_NANOSECOND_PRECISION
 394         inode->i_creation_time = timespec_to_wim_timestamp(&stbuf.st_mtim);
 395         inode->i_last_write_time = timespec_to_wim_timestamp(&stbuf.st_mtim);
 396         inode->i_last_access_time = timespec_to_wim_timestamp(&stbuf.st_atim);
 397 #else
 398         inode->i_creation_time = time_t_to_wim_timestamp(stbuf.st_mtime);
 399         inode->i_last_write_time = time_t_to_wim_timestamp(stbuf.st_mtime);
 400         inode->i_last_access_time = time_t_to_wim_timestamp(stbuf.st_atime);
 401 #endif
 402         if (params->add_flags & WIMLIB_ADD_FLAG_UNIX_DATA) {
 403                 struct wimlib_unix_data unix_data;
 404
 405                 unix_data.uid = stbuf.st_uid;
 406                 unix_data.gid = stbuf.st_gid;
 407                 unix_data.mode = stbuf.st_mode;
 408                 unix_data.rdev = stbuf.st_rdev;
 409                 if (!inode_set_unix_data(inode, &unix_data, UNIX_DATA_ALL)) {
 410                         ret = WIMLIB_ERR_NOMEM;
 411                         goto out;
 412                 }
 413         }
 414
 415         if (params->add_flags & WIMLIB_ADD_FLAG_ROOT) {
 416                 params->capture_root_ino = stbuf.st_ino;
 417                 params->capture_root_dev = stbuf.st_dev;
 418                 params->add_flags &= ~WIMLIB_ADD_FLAG_ROOT;
 419         }
 420
 421         if (S_ISREG(stbuf.st_mode)) {
 422                 ret = unix_scan_regular_file(full_path, stbuf.st_size,
 423                                              inode, params->unhashed_blobs);
 424         } else if (S_ISDIR(stbuf.st_mode)) {
 425                 ret = unix_scan_directory(tree, full_path, full_path_len,
 426                                           dirfd, relpath, params);
 427         } else if (S_ISLNK(stbuf.st_mode)) {
 428                 ret = unix_scan_symlink(full_path, dirfd, relpath,
 429                                         inode, params);
 430         }
 431
 432         if (ret)
 433                 goto out;
 434
 435 out_progress:
 436         params->progress.scan.cur_path = full_path;
 437         if (likely(tree))
 438                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_OK, inode);
 439         else
 440                 ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_EXCLUDED, NULL);
 441 out:
 442         if (unlikely(ret)) {
 443                 free_dentry_tree(tree, params->blob_table);
 444                 tree = NULL;
 445                 ret = report_capture_error(params, ret, full_path);
 446         }
 447         *tree_ret = tree;
 448         return ret;
 449 }
 450
 451 /*
 452  * unix_build_dentry_tree():
 453  *      Builds a tree of WIM dentries from an on-disk directory tree (UNIX
 454  *      version; no NTFS-specific data is captured).
 455  *
 456  * @root_ret:   Place to return a pointer to the root of the dentry tree.  Only
 457  *              modified if successful.  Set to NULL if the file or directory was
 458  *              excluded from capture.
 459  *
 460  * @root_disk_path:  The path to the root of the directory tree on disk.
 461  *
 462  * @params:     See doc for `struct capture_params'.
 463  *
 464  * @return:     0 on success, nonzero on failure.  It is a failure if any of
 465  *              the files cannot be `stat'ed, or if any of the needed
 466  *              directories cannot be opened or read.  Failure to add the files
 467  *              to the WIM may still occur later when trying to actually read
 468  *              the on-disk files during a call to wimlib_write() or
 469  *              wimlib_overwrite().
 470  */
 471 int
 472 unix_build_dentry_tree(struct wim_dentry **root_ret,
 473                        const char *root_disk_path,
 474                        struct capture_params *params)
 475 {
 476         size_t path_len;
 477         size_t path_bufsz;
 478         char *path_buf;
 479         int ret;
 480
 481         path_len = strlen(root_disk_path);
 482         path_bufsz = min(32790, PATH_MAX + 1);
 483
 484         if (path_len >= path_bufsz)
 485                 return WIMLIB_ERR_INVALID_PARAM;
 486
 487         path_buf = MALLOC(path_bufsz);
 488         if (!path_buf)
 489                 return WIMLIB_ERR_NOMEM;
 490         memcpy(path_buf, root_disk_path, path_len + 1);
 491
 492         params->capture_root_nchars = path_len;
 493
 494         ret = unix_build_dentry_tree_recursive(root_ret, path_buf, path_len,
 495                                                AT_FDCWD, path_buf, params);
 496         FREE(path_buf);
 497         return ret;
 498 }
 499
 500 #endif /* !__WIN32__ */