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
  44 #ifdef HAVE_FDOPENDIR
  45 #  define my_fdopendir(dirfd_p) fdopendir(*(dirfd_p))
  46 #else
  47 static DIR *
  48 my_fdopendir(int *dirfd_p)
  49 {
  50         DIR *dir = NULL;
  51         int old_pwd;
  52
  53         old_pwd = open(".", O_RDONLY);
  54         if (old_pwd >= 0) {
  55                 if (!fchdir(*dirfd_p)) {
  56                         dir = opendir(".");
  57                         if (dir) {
  58                                 close(*dirfd_p);
  59                                 *dirfd_p = dirfd(dir);
  60                         }
  61                         fchdir(old_pwd);
  62                 }
  63                 close(old_pwd);
  64         }
  65         return dir;
  66 }
  67 #endif
  68
  69 #ifdef HAVE_OPENAT
  70 #  define my_openat(full_path, dirfd, relpath, flags) \
  71                 openat((dirfd), (relpath), (flags))
  72 #else
  73 #  define my_openat(full_path, dirfd, relpath, flags) \
  74                 open((full_path), (flags))
  75 #endif
  76
  77 #ifdef HAVE_READLINKAT
  78 #  define my_readlinkat(full_path, dirfd, relpath, buf, bufsize) \
  79                 readlinkat((dirfd), (relpath), (buf), (bufsize))
  80 #else
  81 #  define my_readlinkat(full_path, dirfd, relpath, buf, bufsize) \
  82                 readlink((full_path), (buf), (bufsize))
  83 #endif
  84
  85 #ifdef HAVE_FSTATAT
  86 #  define my_fstatat(full_path, dirfd, relpath, stbuf, flags)   \
  87         fstatat((dirfd), (relpath), (stbuf), (flags))
  88 #else
  89 #  define my_fstatat(full_path, dirfd, relpath, stbuf, flags)   \
  90         ((flags) & AT_SYMLINK_NOFOLLOW) ? \
  91                 lstat((full_path), (stbuf)) : \
  92                 stat((full_path), (stbuf))
  93 #endif
  94
  95 #ifndef AT_FDCWD
  96 #  define AT_FDCWD      -100
  97 #endif
  98
  99 #ifndef AT_SYMLINK_NOFOLLOW
 100 #  define AT_SYMLINK_NOFOLLOW   0x100
 101 #endif
 102
 103 static int
 104 unix_scan_regular_file(const char *path, u64 size, struct wim_inode *inode,
 105                        struct list_head *unhashed_streams)
 106 {
 107         struct wim_lookup_table_entry *lte;
 108         char *file_on_disk;
 109
 110         inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
 111
 112         /* Empty files do not have to have a lookup table entry. */
 113         if (!size)
 114                 return 0;
 115
 116         file_on_disk = STRDUP(path);
 117         if (!file_on_disk)
 118                 return WIMLIB_ERR_NOMEM;
 119         lte = new_lookup_table_entry();
 120         if (!lte) {
 121                 FREE(file_on_disk);
 122                 return WIMLIB_ERR_NOMEM;
 123         }
 124         lte->file_on_disk = file_on_disk;
 125         lte->resource_location = RESOURCE_IN_FILE_ON_DISK;
 126         lte->size = size;
 127         add_unhashed_stream(lte, inode, 0, unhashed_streams);
 128         inode->i_lte = lte;
 129         return 0;
 130 }
 131
 132 static int
 133 unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 134                                  char *path, size_t path_len,
 135                                  int dirfd, const char *relpath,
 136                                  struct add_image_params *params);
 137
 138 static int
 139 unix_scan_directory(struct wim_dentry *dir_dentry,
 140                     char *full_path, size_t full_path_len,
 141                     int parent_dirfd, const char *dir_relpath,
 142                     struct add_image_params *params)
 143 {
 144
 145         int dirfd;
 146         DIR *dir;
 147         int ret;
 148
 149         dirfd = my_openat(full_path, parent_dirfd, dir_relpath, O_RDONLY);
 150         if (dirfd < 0) {
 151                 ERROR_WITH_ERRNO("\"%s\": Can't open directory", full_path);
 152                 return WIMLIB_ERR_OPENDIR;
 153         }
 154
 155         dir_dentry->d_inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 156         dir = my_fdopendir(&dirfd);
 157         if (!dir) {
 158                 ERROR_WITH_ERRNO("\"%s\": Can't open directory", full_path);
 159                 close(dirfd);
 160                 return WIMLIB_ERR_OPENDIR;
 161         }
 162
 163         ret = 0;
 164         for (;;) {
 165                 struct dirent *entry;
 166                 struct wim_dentry *child;
 167                 size_t name_len;
 168
 169                 errno = 0;
 170                 entry = readdir(dir);
 171                 if (!entry) {
 172                         if (errno) {
 173                                 ret = WIMLIB_ERR_READ;
 174                                 ERROR_WITH_ERRNO("\"%s\": Error reading directory",
 175                                                  full_path);
 176                         }
 177                         break;
 178                 }
 179
 180                 if (entry->d_name[0] == '.' &&
 181                     (entry->d_name[1] == '\0' ||
 182                      (entry->d_name[1] == '.' && entry->d_name[2] == '\0')))
 183                         continue;
 184
 185                 full_path[full_path_len] = '/';
 186                 name_len = strlen(entry->d_name);
 187                 memcpy(&full_path[full_path_len + 1], entry->d_name, name_len + 1);
 188                 ret = unix_build_dentry_tree_recursive(&child,
 189                                                        full_path,
 190                                                        full_path_len + 1 + name_len,
 191                                                        dirfd,
 192                                                        &full_path[full_path_len + 1],
 193                                                        params);
 194                 full_path[full_path_len] = '\0';
 195                 if (ret)
 196                         break;
 197                 if (child)
 198                         dentry_add_child(dir_dentry, child);
 199         }
 200         closedir(dir);
 201         return ret;
 202 }
 203
 204 static int
 205 unix_scan_symlink(struct wim_dentry **root_p, const char *full_path,
 206                   int dirfd, const char *relpath,
 207                   struct wim_inode *inode, struct add_image_params *params)
 208 {
 209         char deref_name_buf[4096];
 210         ssize_t deref_name_len;
 211         char *dest;
 212         int ret;
 213
 214         inode->i_attributes = FILE_ATTRIBUTE_REPARSE_POINT;
 215         inode->i_reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
 216
 217         /* The idea here is to call readlink() to get the UNIX target of the
 218          * symbolic link, then turn the target into a reparse point data buffer
 219          * that contains a relative or absolute symbolic link. */
 220         deref_name_len = my_readlinkat(full_path, dirfd, relpath,
 221                                        deref_name_buf, sizeof(deref_name_buf) - 1);
 222         if (deref_name_len < 0) {
 223                 ERROR_WITH_ERRNO("\"%s\": Can't read target of symbolic link",
 224                                  full_path);
 225                 return WIMLIB_ERR_READLINK;
 226         }
 227
 228         dest = deref_name_buf;
 229
 230         dest[deref_name_len] = '\0';
 231
 232         if ((params->add_flags & WIMLIB_ADD_FLAG_RPFIX) &&
 233              dest[0] == '/')
 234         {
 235                 dest = capture_fixup_absolute_symlink(dest,
 236                                                       params->capture_root_ino,
 237                                                       params->capture_root_dev);
 238                 if (!dest) {
 239                         /* RPFIX (reparse point fixup) mode:  Ignore
 240                          * absolute symbolic link that points out of the
 241                          * tree to be captured.  */
 242                         free_dentry(*root_p);
 243                         *root_p = NULL;
 244                         params->progress.scan.cur_path = full_path;
 245                         params->progress.scan.symlink_target = deref_name_buf;
 246                         do_capture_progress(params,
 247                                             WIMLIB_SCAN_DENTRY_EXCLUDED_SYMLINK,
 248                                             NULL);
 249                         return 0;
 250                 }
 251                 inode->i_not_rpfixed = 0;
 252         }
 253         ret = wim_inode_set_symlink(inode, dest, params->lookup_table);
 254         if (ret)
 255                 return ret;
 256
 257         /* Unfortunately, Windows seems to have the concept of "file" symbolic
 258          * links as being different from "directory" symbolic links...  so
 259          * FILE_ATTRIBUTE_DIRECTORY needs to be set on the symbolic link if the
 260          * *target* of the symbolic link is a directory.  */
 261         struct stat stbuf;
 262         if (my_fstatat(full_path, dirfd, relpath, &stbuf, 0) == 0 &&
 263             S_ISDIR(stbuf.st_mode))
 264                 inode->i_attributes |= FILE_ATTRIBUTE_DIRECTORY;
 265         return 0;
 266 }
 267
 268 static int
 269 unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 270                                  char *full_path, size_t full_path_len,
 271                                  int dirfd, const char *relpath,
 272                                  struct add_image_params *params)
 273 {
 274         struct wim_dentry *tree = NULL;
 275         struct wim_inode *inode = NULL;
 276         int ret;
 277         struct stat stbuf;
 278         int stat_flags;
 279
 280         if (should_exclude_path(full_path + params->capture_root_nchars,
 281                                 full_path_len - params->capture_root_nchars,
 282                                 params->config))
 283         {
 284                 ret = 0;
 285                 goto out_progress;
 286         }
 287
 288         if (params->add_flags & (WIMLIB_ADD_FLAG_DEREFERENCE |
 289                                  WIMLIB_ADD_FLAG_ROOT))
 290                 stat_flags = 0;
 291         else
 292                 stat_flags = AT_SYMLINK_NOFOLLOW;
 293
 294         ret = my_fstatat(full_path, dirfd, relpath, &stbuf, stat_flags);
 295
 296         if (ret) {
 297                 ERROR_WITH_ERRNO("\"%s\": Can't read metadata", full_path);
 298                 ret = WIMLIB_ERR_STAT;
 299                 goto out;
 300         }
 301
 302         if (unlikely(!S_ISREG(stbuf.st_mode) &&
 303                      !S_ISDIR(stbuf.st_mode) &&
 304                      !S_ISLNK(stbuf.st_mode)))
 305         {
 306                 if (params->add_flags & WIMLIB_ADD_FLAG_NO_UNSUPPORTED_EXCLUDE)
 307                 {
 308                         ERROR("\"%s\": File type is unsupported", full_path);
 309                         ret = WIMLIB_ERR_UNSUPPORTED_FILE;
 310                         goto out;
 311                 }
 312                 params->progress.scan.cur_path = full_path;
 313                 do_capture_progress(params, WIMLIB_SCAN_DENTRY_UNSUPPORTED, NULL);
 314                 ret = 0;
 315                 goto out;
 316         }
 317
 318         ret = inode_table_new_dentry(params->inode_table, relpath,
 319                                      stbuf.st_ino, stbuf.st_dev,
 320                                      S_ISDIR(stbuf.st_mode), &tree);
 321         if (ret)
 322                 goto out;
 323
 324         inode = tree->d_inode;
 325
 326         if (inode->i_nlink > 1) {
 327                 /* Already seen this inode?  */
 328                 ret = 0;
 329                 goto out_progress;
 330         }
 331
 332 #ifdef HAVE_STAT_NANOSECOND_PRECISION
 333         inode->i_creation_time = timespec_to_wim_timestamp(stbuf.st_mtim);
 334         inode->i_last_write_time = timespec_to_wim_timestamp(stbuf.st_mtim);
 335         inode->i_last_access_time = timespec_to_wim_timestamp(stbuf.st_atim);
 336 #else
 337         inode->i_creation_time = unix_timestamp_to_wim(stbuf.st_mtime);
 338         inode->i_last_write_time = unix_timestamp_to_wim(stbuf.st_mtime);
 339         inode->i_last_access_time = unix_timestamp_to_wim(stbuf.st_atime);
 340 #endif
 341         inode->i_resolved = 1;
 342         if (params->add_flags & WIMLIB_ADD_FLAG_UNIX_DATA) {
 343                 ret = inode_set_unix_data(inode, stbuf.st_uid,
 344                                           stbuf.st_gid,
 345                                           stbuf.st_mode,
 346                                           params->lookup_table,
 347                                           UNIX_DATA_ALL | UNIX_DATA_CREATE);
 348                 if (ret)
 349                         goto out;
 350         }
 351
 352         if (params->add_flags & WIMLIB_ADD_FLAG_ROOT) {
 353                 params->capture_root_ino = stbuf.st_ino;
 354                 params->capture_root_dev = stbuf.st_dev;
 355                 params->add_flags &= ~WIMLIB_ADD_FLAG_ROOT;
 356         }
 357
 358         if (S_ISREG(stbuf.st_mode)) {
 359                 ret = unix_scan_regular_file(full_path, stbuf.st_size,
 360                                              inode, params->unhashed_streams);
 361         } else if (S_ISDIR(stbuf.st_mode)) {
 362                 ret = unix_scan_directory(tree, full_path, full_path_len,
 363                                           dirfd, relpath, params);
 364         } else {
 365                 ret = unix_scan_symlink(&tree, full_path, dirfd, relpath,
 366                                         inode, params);
 367                 if (!tree)
 368                         goto out;
 369         }
 370
 371         if (ret)
 372                 goto out;
 373
 374 out_progress:
 375         params->progress.scan.cur_path = full_path;
 376         if (likely(tree))
 377                 do_capture_progress(params, WIMLIB_SCAN_DENTRY_OK, inode);
 378         else
 379                 do_capture_progress(params, WIMLIB_SCAN_DENTRY_EXCLUDED, NULL);
 380 out:
 381         if (likely(ret == 0))
 382                 *tree_ret = tree;
 383         else
 384                 free_dentry_tree(tree, params->lookup_table);
 385         return ret;
 386 }
 387
 388 /*
 389  * unix_build_dentry_tree():
 390  *      Builds a tree of WIM dentries from an on-disk directory tree (UNIX
 391  *      version; no NTFS-specific data is captured).
 392  *
 393  * @root_ret:   Place to return a pointer to the root of the dentry tree.  Only
 394  *              modified if successful.  Set to NULL if the file or directory was
 395  *              excluded from capture.
 396  *
 397  * @root_disk_path:  The path to the root of the directory tree on disk.
 398  *
 399  * @params:     See doc for `struct add_image_params'.
 400  *
 401  * @return:     0 on success, nonzero on failure.  It is a failure if any of
 402  *              the files cannot be `stat'ed, or if any of the needed
 403  *              directories cannot be opened or read.  Failure to add the files
 404  *              to the WIM may still occur later when trying to actually read
 405  *              the on-disk files during a call to wimlib_write() or
 406  *              wimlib_overwrite().
 407  */
 408 int
 409 unix_build_dentry_tree(struct wim_dentry **root_ret,
 410                        const char *root_disk_path,
 411                        struct add_image_params *params)
 412 {
 413         size_t path_len;
 414         size_t path_bufsz;
 415         char *path_buf;
 416         int ret;
 417
 418         path_len = strlen(root_disk_path);
 419         path_bufsz = min(32790, PATH_MAX + 1);
 420
 421         if (path_len >= path_bufsz)
 422                 return WIMLIB_ERR_INVALID_PARAM;
 423
 424         path_buf = MALLOC(path_bufsz);
 425         if (!path_buf)
 426                 return WIMLIB_ERR_NOMEM;
 427         memcpy(path_buf, root_disk_path, path_len + 1);
 428
 429         params->capture_root_nchars = path_len;
 430
 431         ret = unix_build_dentry_tree_recursive(root_ret, path_buf, path_len,
 432                                                AT_FDCWD, path_buf, params);
 433         FREE(path_buf);
 434         return ret;
 435 }
 436
 437 #endif /* !__WIN32__ */