Btrfs: corret fmode_t annotations
[linux-2.6/btrfs-unstable.git] / fs / btrfs / super.c
blob09908f25fca9c759407377f51206a5cbe15b6a78
1 /*
2 * Copyright (C) 2007 Oracle. All rights reserved.
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public
6 * License v2 as published by the Free Software Foundation.
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
13 * You should have received a copy of the GNU General Public
14 * License along with this program; if not, write to the
15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
16 * Boston, MA 021110-1307, USA.
19 #include <linux/blkdev.h>
20 #include <linux/module.h>
21 #include <linux/buffer_head.h>
22 #include <linux/fs.h>
23 #include <linux/pagemap.h>
24 #include <linux/highmem.h>
25 #include <linux/time.h>
26 #include <linux/init.h>
27 #include <linux/string.h>
28 #include <linux/smp_lock.h>
29 #include <linux/backing-dev.h>
30 #include <linux/mount.h>
31 #include <linux/mpage.h>
32 #include <linux/swap.h>
33 #include <linux/writeback.h>
34 #include <linux/statfs.h>
35 #include <linux/compat.h>
36 #include <linux/parser.h>
37 #include <linux/ctype.h>
38 #include <linux/namei.h>
39 #include <linux/miscdevice.h>
40 #include <linux/version.h>
41 #include "compat.h"
42 #include "ctree.h"
43 #include "disk-io.h"
44 #include "transaction.h"
45 #include "btrfs_inode.h"
46 #include "ioctl.h"
47 #include "print-tree.h"
48 #include "xattr.h"
49 #include "volumes.h"
50 #include "version.h"
51 #include "export.h"
52 #include "compression.h"
54 #define BTRFS_SUPER_MAGIC 0x9123683E
56 static struct super_operations btrfs_super_ops;
58 static void btrfs_put_super (struct super_block * sb)
60 struct btrfs_root *root = btrfs_sb(sb);
61 struct btrfs_fs_info *fs = root->fs_info;
62 int ret;
64 ret = close_ctree(root);
65 if (ret) {
66 printk("close ctree returns %d\n", ret);
68 btrfs_sysfs_del_super(fs);
69 sb->s_fs_info = NULL;
72 enum {
73 Opt_degraded, Opt_subvol, Opt_device, Opt_nodatasum, Opt_nodatacow,
74 Opt_max_extent, Opt_max_inline, Opt_alloc_start, Opt_nobarrier,
75 Opt_ssd, Opt_thread_pool, Opt_noacl, Opt_compress, Opt_err,
78 static match_table_t tokens = {
79 {Opt_degraded, "degraded"},
80 {Opt_subvol, "subvol=%s"},
81 {Opt_device, "device=%s"},
82 {Opt_nodatasum, "nodatasum"},
83 {Opt_nodatacow, "nodatacow"},
84 {Opt_nobarrier, "nobarrier"},
85 {Opt_max_extent, "max_extent=%s"},
86 {Opt_max_inline, "max_inline=%s"},
87 {Opt_alloc_start, "alloc_start=%s"},
88 {Opt_thread_pool, "thread_pool=%d"},
89 {Opt_compress, "compress"},
90 {Opt_ssd, "ssd"},
91 {Opt_noacl, "noacl"},
92 {Opt_err, NULL},
95 u64 btrfs_parse_size(char *str)
97 u64 res;
98 int mult = 1;
99 char *end;
100 char last;
102 res = simple_strtoul(str, &end, 10);
104 last = end[0];
105 if (isalpha(last)) {
106 last = tolower(last);
107 switch (last) {
108 case 'g':
109 mult *= 1024;
110 case 'm':
111 mult *= 1024;
112 case 'k':
113 mult *= 1024;
115 res = res * mult;
117 return res;
121 * Regular mount options parser. Everything that is needed only when
122 * reading in a new superblock is parsed here.
124 int btrfs_parse_options(struct btrfs_root *root, char *options)
126 struct btrfs_fs_info *info = root->fs_info;
127 substring_t args[MAX_OPT_ARGS];
128 char *p, *num;
129 int intarg;
131 if (!options)
132 return 0;
135 * strsep changes the string, duplicate it because parse_options
136 * gets called twice
138 options = kstrdup(options, GFP_NOFS);
139 if (!options)
140 return -ENOMEM;
143 while ((p = strsep(&options, ",")) != NULL) {
144 int token;
145 if (!*p)
146 continue;
148 token = match_token(p, tokens, args);
149 switch (token) {
150 case Opt_degraded:
151 printk(KERN_INFO "btrfs: allowing degraded mounts\n");
152 btrfs_set_opt(info->mount_opt, DEGRADED);
153 break;
154 case Opt_subvol:
155 case Opt_device:
157 * These are parsed by btrfs_parse_early_options
158 * and can be happily ignored here.
160 break;
161 case Opt_nodatasum:
162 printk(KERN_INFO "btrfs: setting nodatacsum\n");
163 btrfs_set_opt(info->mount_opt, NODATASUM);
164 break;
165 case Opt_nodatacow:
166 printk(KERN_INFO "btrfs: setting nodatacow\n");
167 btrfs_set_opt(info->mount_opt, NODATACOW);
168 btrfs_set_opt(info->mount_opt, NODATASUM);
169 break;
170 case Opt_compress:
171 printk(KERN_INFO "btrfs: use compression\n");
172 btrfs_set_opt(info->mount_opt, COMPRESS);
173 break;
174 case Opt_ssd:
175 printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
176 btrfs_set_opt(info->mount_opt, SSD);
177 break;
178 case Opt_nobarrier:
179 printk(KERN_INFO "btrfs: turning off barriers\n");
180 btrfs_set_opt(info->mount_opt, NOBARRIER);
181 break;
182 case Opt_thread_pool:
183 intarg = 0;
184 match_int(&args[0], &intarg);
185 if (intarg) {
186 info->thread_pool_size = intarg;
187 printk(KERN_INFO "btrfs: thread pool %d\n",
188 info->thread_pool_size);
190 break;
191 case Opt_max_extent:
192 num = match_strdup(&args[0]);
193 if (num) {
194 info->max_extent = btrfs_parse_size(num);
195 kfree(num);
197 info->max_extent = max_t(u64,
198 info->max_extent, root->sectorsize);
199 printk(KERN_INFO "btrfs: max_extent at %llu\n",
200 info->max_extent);
202 break;
203 case Opt_max_inline:
204 num = match_strdup(&args[0]);
205 if (num) {
206 info->max_inline = btrfs_parse_size(num);
207 kfree(num);
209 if (info->max_inline) {
210 info->max_inline = max_t(u64,
211 info->max_inline,
212 root->sectorsize);
214 printk(KERN_INFO "btrfs: max_inline at %llu\n",
215 info->max_inline);
217 break;
218 case Opt_alloc_start:
219 num = match_strdup(&args[0]);
220 if (num) {
221 info->alloc_start = btrfs_parse_size(num);
222 kfree(num);
223 printk(KERN_INFO
224 "btrfs: allocations start at %llu\n",
225 info->alloc_start);
227 break;
228 case Opt_noacl:
229 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
230 break;
231 default:
232 break;
235 kfree(options);
236 return 0;
240 * Parse mount options that are required early in the mount process.
242 * All other options will be parsed on much later in the mount process and
243 * only when we need to allocate a new super block.
245 static int btrfs_parse_early_options(const char *options, fmode_t flags,
246 void *holder, char **subvol_name,
247 struct btrfs_fs_devices **fs_devices)
249 substring_t args[MAX_OPT_ARGS];
250 char *opts, *p;
251 int error = 0;
253 if (!options)
254 goto out;
257 * strsep changes the string, duplicate it because parse_options
258 * gets called twice
260 opts = kstrdup(options, GFP_KERNEL);
261 if (!opts)
262 return -ENOMEM;
264 while ((p = strsep(&opts, ",")) != NULL) {
265 int token;
266 if (!*p)
267 continue;
269 token = match_token(p, tokens, args);
270 switch (token) {
271 case Opt_subvol:
272 *subvol_name = match_strdup(&args[0]);
273 break;
274 case Opt_device:
275 error = btrfs_scan_one_device(match_strdup(&args[0]),
276 flags, holder, fs_devices);
277 if (error)
278 goto out_free_opts;
279 break;
280 default:
281 break;
285 out_free_opts:
286 kfree(opts);
287 out:
289 * If no subvolume name is specified we use the default one. Allocate
290 * a copy of the string "." here so that code later in the
291 * mount path doesn't care if it's the default volume or another one.
293 if (!*subvol_name) {
294 *subvol_name = kstrdup(".", GFP_KERNEL);
295 if (!*subvol_name)
296 return -ENOMEM;
298 return error;
301 static int btrfs_fill_super(struct super_block * sb,
302 struct btrfs_fs_devices *fs_devices,
303 void * data, int silent)
305 struct inode * inode;
306 struct dentry * root_dentry;
307 struct btrfs_super_block *disk_super;
308 struct btrfs_root *tree_root;
309 struct btrfs_inode *bi;
310 int err;
312 sb->s_maxbytes = MAX_LFS_FILESIZE;
313 sb->s_magic = BTRFS_SUPER_MAGIC;
314 sb->s_op = &btrfs_super_ops;
315 sb->s_export_op = &btrfs_export_ops;
316 sb->s_xattr = btrfs_xattr_handlers;
317 sb->s_time_gran = 1;
318 sb->s_flags |= MS_POSIXACL;
320 tree_root = open_ctree(sb, fs_devices, (char *)data);
322 if (IS_ERR(tree_root)) {
323 printk("btrfs: open_ctree failed\n");
324 return PTR_ERR(tree_root);
326 sb->s_fs_info = tree_root;
327 disk_super = &tree_root->fs_info->super_copy;
328 inode = btrfs_iget_locked(sb, BTRFS_FIRST_FREE_OBJECTID,
329 tree_root->fs_info->fs_root);
330 bi = BTRFS_I(inode);
331 bi->location.objectid = inode->i_ino;
332 bi->location.offset = 0;
333 bi->root = tree_root->fs_info->fs_root;
335 btrfs_set_key_type(&bi->location, BTRFS_INODE_ITEM_KEY);
337 if (!inode) {
338 err = -ENOMEM;
339 goto fail_close;
341 if (inode->i_state & I_NEW) {
342 btrfs_read_locked_inode(inode);
343 unlock_new_inode(inode);
346 root_dentry = d_alloc_root(inode);
347 if (!root_dentry) {
348 iput(inode);
349 err = -ENOMEM;
350 goto fail_close;
353 /* this does the super kobj at the same time */
354 err = btrfs_sysfs_add_super(tree_root->fs_info);
355 if (err)
356 goto fail_close;
358 sb->s_root = root_dentry;
360 save_mount_options(sb, data);
361 return 0;
363 fail_close:
364 close_ctree(tree_root);
365 return err;
368 int btrfs_sync_fs(struct super_block *sb, int wait)
370 struct btrfs_trans_handle *trans;
371 struct btrfs_root *root;
372 int ret;
373 root = btrfs_sb(sb);
375 if (sb->s_flags & MS_RDONLY)
376 return 0;
378 sb->s_dirt = 0;
379 if (!wait) {
380 filemap_flush(root->fs_info->btree_inode->i_mapping);
381 return 0;
384 btrfs_start_delalloc_inodes(root);
385 btrfs_wait_ordered_extents(root, 0);
387 btrfs_clean_old_snapshots(root);
388 trans = btrfs_start_transaction(root, 1);
389 ret = btrfs_commit_transaction(trans, root);
390 sb->s_dirt = 0;
391 return ret;
394 static void btrfs_write_super(struct super_block *sb)
396 sb->s_dirt = 0;
399 static int btrfs_test_super(struct super_block *s, void *data)
401 struct btrfs_fs_devices *test_fs_devices = data;
402 struct btrfs_root *root = btrfs_sb(s);
404 return root->fs_info->fs_devices == test_fs_devices;
408 * Find a superblock for the given device / mount point.
410 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
411 * for multiple device setup. Make sure to keep it in sync.
413 static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
414 const char *dev_name, void *data, struct vfsmount *mnt)
416 char *subvol_name = NULL;
417 struct block_device *bdev = NULL;
418 struct super_block *s;
419 struct dentry *root;
420 struct btrfs_fs_devices *fs_devices = NULL;
421 fmode_t mode = FMODE_READ;
422 int error = 0;
424 if (!(flags & MS_RDONLY))
425 mode |= FMODE_WRITE;
427 error = btrfs_parse_early_options(data, mode, fs_type,
428 &subvol_name, &fs_devices);
429 if (error)
430 goto error;
432 error = btrfs_scan_one_device(dev_name, mode, fs_type, &fs_devices);
433 if (error)
434 goto error_free_subvol_name;
436 error = btrfs_open_devices(fs_devices, mode, fs_type);
437 if (error)
438 goto error_free_subvol_name;
440 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
441 error = -EACCES;
442 goto error_close_devices;
445 bdev = fs_devices->latest_bdev;
446 s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
447 if (IS_ERR(s))
448 goto error_s;
450 if (s->s_root) {
451 if ((flags ^ s->s_flags) & MS_RDONLY) {
452 up_write(&s->s_umount);
453 deactivate_super(s);
454 error = -EBUSY;
455 goto error_close_devices;
458 btrfs_close_devices(fs_devices);
459 } else {
460 char b[BDEVNAME_SIZE];
462 s->s_flags = flags;
463 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
464 error = btrfs_fill_super(s, fs_devices, data,
465 flags & MS_SILENT ? 1 : 0);
466 if (error) {
467 up_write(&s->s_umount);
468 deactivate_super(s);
469 goto error;
472 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
473 s->s_flags |= MS_ACTIVE;
476 if (!strcmp(subvol_name, "."))
477 root = dget(s->s_root);
478 else {
479 mutex_lock(&s->s_root->d_inode->i_mutex);
480 root = lookup_one_len(subvol_name, s->s_root, strlen(subvol_name));
481 mutex_unlock(&s->s_root->d_inode->i_mutex);
482 if (IS_ERR(root)) {
483 up_write(&s->s_umount);
484 deactivate_super(s);
485 error = PTR_ERR(root);
486 goto error;
488 if (!root->d_inode) {
489 dput(root);
490 up_write(&s->s_umount);
491 deactivate_super(s);
492 error = -ENXIO;
493 goto error;
497 mnt->mnt_sb = s;
498 mnt->mnt_root = root;
500 kfree(subvol_name);
501 return 0;
503 error_s:
504 error = PTR_ERR(s);
505 error_close_devices:
506 btrfs_close_devices(fs_devices);
507 error_free_subvol_name:
508 kfree(subvol_name);
509 error:
510 return error;
513 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
515 struct btrfs_root *root = btrfs_sb(sb);
516 int ret;
518 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
519 return 0;
521 if (*flags & MS_RDONLY) {
522 sb->s_flags |= MS_RDONLY;
524 ret = btrfs_commit_super(root);
525 WARN_ON(ret);
526 } else {
527 if (root->fs_info->fs_devices->rw_devices == 0)
528 return -EACCES;
530 if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
531 return -EINVAL;
533 ret = btrfs_cleanup_reloc_trees(root);
534 WARN_ON(ret);
536 ret = btrfs_cleanup_fs_roots(root->fs_info);
537 WARN_ON(ret);
539 sb->s_flags &= ~MS_RDONLY;
542 return 0;
545 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
547 struct btrfs_root *root = btrfs_sb(dentry->d_sb);
548 struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
549 int bits = dentry->d_sb->s_blocksize_bits;
550 __be32 *fsid = (__be32 *)root->fs_info->fsid;
552 buf->f_namelen = BTRFS_NAME_LEN;
553 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
554 buf->f_bfree = buf->f_blocks -
555 (btrfs_super_bytes_used(disk_super) >> bits);
556 buf->f_bavail = buf->f_bfree;
557 buf->f_bsize = dentry->d_sb->s_blocksize;
558 buf->f_type = BTRFS_SUPER_MAGIC;
559 /* We treat it as constant endianness (it doesn't matter _which_)
560 because we want the fsid to come out the same whether mounted
561 on a big-endian or little-endian host */
562 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
563 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
564 /* Mask in the root object ID too, to disambiguate subvols */
565 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
566 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
568 return 0;
571 static struct file_system_type btrfs_fs_type = {
572 .owner = THIS_MODULE,
573 .name = "btrfs",
574 .get_sb = btrfs_get_sb,
575 .kill_sb = kill_anon_super,
576 .fs_flags = FS_REQUIRES_DEV,
580 * used by btrfsctl to scan devices when no FS is mounted
582 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
583 unsigned long arg)
585 struct btrfs_ioctl_vol_args *vol;
586 struct btrfs_fs_devices *fs_devices;
587 int ret = 0;
588 int len;
590 vol = kmalloc(sizeof(*vol), GFP_KERNEL);
591 if (copy_from_user(vol, (void __user *)arg, sizeof(*vol))) {
592 ret = -EFAULT;
593 goto out;
595 len = strnlen(vol->name, BTRFS_PATH_NAME_MAX);
596 switch (cmd) {
597 case BTRFS_IOC_SCAN_DEV:
598 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
599 &btrfs_fs_type, &fs_devices);
600 break;
602 out:
603 kfree(vol);
604 return ret;
607 static void btrfs_write_super_lockfs(struct super_block *sb)
609 struct btrfs_root *root = btrfs_sb(sb);
610 mutex_lock(&root->fs_info->transaction_kthread_mutex);
611 mutex_lock(&root->fs_info->cleaner_mutex);
614 static void btrfs_unlockfs(struct super_block *sb)
616 struct btrfs_root *root = btrfs_sb(sb);
617 mutex_unlock(&root->fs_info->cleaner_mutex);
618 mutex_unlock(&root->fs_info->transaction_kthread_mutex);
621 static struct super_operations btrfs_super_ops = {
622 .delete_inode = btrfs_delete_inode,
623 .put_super = btrfs_put_super,
624 .write_super = btrfs_write_super,
625 .sync_fs = btrfs_sync_fs,
626 .show_options = generic_show_options,
627 .write_inode = btrfs_write_inode,
628 .dirty_inode = btrfs_dirty_inode,
629 .alloc_inode = btrfs_alloc_inode,
630 .destroy_inode = btrfs_destroy_inode,
631 .statfs = btrfs_statfs,
632 .remount_fs = btrfs_remount,
633 .write_super_lockfs = btrfs_write_super_lockfs,
634 .unlockfs = btrfs_unlockfs,
637 static const struct file_operations btrfs_ctl_fops = {
638 .unlocked_ioctl = btrfs_control_ioctl,
639 .compat_ioctl = btrfs_control_ioctl,
640 .owner = THIS_MODULE,
643 static struct miscdevice btrfs_misc = {
644 .minor = MISC_DYNAMIC_MINOR,
645 .name = "btrfs-control",
646 .fops = &btrfs_ctl_fops
649 static int btrfs_interface_init(void)
651 return misc_register(&btrfs_misc);
654 static void btrfs_interface_exit(void)
656 if (misc_deregister(&btrfs_misc) < 0)
657 printk("misc_deregister failed for control device");
660 static int __init init_btrfs_fs(void)
662 int err;
664 err = btrfs_init_sysfs();
665 if (err)
666 return err;
668 err = btrfs_init_cachep();
669 if (err)
670 goto free_sysfs;
672 err = extent_io_init();
673 if (err)
674 goto free_cachep;
676 err = extent_map_init();
677 if (err)
678 goto free_extent_io;
680 err = btrfs_interface_init();
681 if (err)
682 goto free_extent_map;
684 err = register_filesystem(&btrfs_fs_type);
685 if (err)
686 goto unregister_ioctl;
688 printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
689 return 0;
691 unregister_ioctl:
692 btrfs_interface_exit();
693 free_extent_map:
694 extent_map_exit();
695 free_extent_io:
696 extent_io_exit();
697 free_cachep:
698 btrfs_destroy_cachep();
699 free_sysfs:
700 btrfs_exit_sysfs();
701 return err;
704 static void __exit exit_btrfs_fs(void)
706 btrfs_destroy_cachep();
707 extent_map_exit();
708 extent_io_exit();
709 btrfs_interface_exit();
710 unregister_filesystem(&btrfs_fs_type);
711 btrfs_exit_sysfs();
712 btrfs_cleanup_fs_uuids();
713 btrfs_zlib_exit();
716 module_init(init_btrfs_fs)
717 module_exit(exit_btrfs_fs)
719 MODULE_LICENSE("GPL");