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>
23 #include <linux/pagemap.h>
24 #include <linux/highmem.h>
25 #include <linux/time.h>
26 #include <linux/init.h>
27 #include <linux/seq_file.h>
28 #include <linux/string.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/magic.h>
44 #include "transaction.h"
45 #include "btrfs_inode.h"
47 #include "print-tree.h"
52 #include "compression.h"
54 static const struct super_operations btrfs_super_ops
;
56 static void btrfs_put_super(struct super_block
*sb
)
58 struct btrfs_root
*root
= btrfs_sb(sb
);
61 ret
= close_ctree(root
);
66 Opt_degraded
, Opt_subvol
, Opt_device
, Opt_nodatasum
, Opt_nodatacow
,
67 Opt_max_extent
, Opt_max_inline
, Opt_alloc_start
, Opt_nobarrier
,
68 Opt_ssd
, Opt_nossd
, Opt_ssd_spread
, Opt_thread_pool
, Opt_noacl
,
69 Opt_compress
, Opt_compress_force
, Opt_notreelog
, Opt_ratio
,
74 static match_table_t tokens
= {
75 {Opt_degraded
, "degraded"},
76 {Opt_subvol
, "subvol=%s"},
77 {Opt_device
, "device=%s"},
78 {Opt_nodatasum
, "nodatasum"},
79 {Opt_nodatacow
, "nodatacow"},
80 {Opt_nobarrier
, "nobarrier"},
81 {Opt_max_extent
, "max_extent=%s"},
82 {Opt_max_inline
, "max_inline=%s"},
83 {Opt_alloc_start
, "alloc_start=%s"},
84 {Opt_thread_pool
, "thread_pool=%d"},
85 {Opt_compress
, "compress"},
86 {Opt_compress_force
, "compress-force"},
88 {Opt_ssd_spread
, "ssd_spread"},
91 {Opt_notreelog
, "notreelog"},
92 {Opt_flushoncommit
, "flushoncommit"},
93 {Opt_ratio
, "metadata_ratio=%d"},
94 {Opt_discard
, "discard"},
98 u64
btrfs_parse_size(char *str
)
105 res
= simple_strtoul(str
, &end
, 10);
109 last
= tolower(last
);
124 * Regular mount options parser. Everything that is needed only when
125 * reading in a new superblock is parsed here.
127 int btrfs_parse_options(struct btrfs_root
*root
, char *options
)
129 struct btrfs_fs_info
*info
= root
->fs_info
;
130 substring_t args
[MAX_OPT_ARGS
];
131 char *p
, *num
, *orig
;
139 * strsep changes the string, duplicate it because parse_options
142 options
= kstrdup(options
, GFP_NOFS
);
148 while ((p
= strsep(&options
, ",")) != NULL
) {
153 token
= match_token(p
, tokens
, args
);
156 printk(KERN_INFO
"btrfs: allowing degraded mounts\n");
157 btrfs_set_opt(info
->mount_opt
, DEGRADED
);
162 * These are parsed by btrfs_parse_early_options
163 * and can be happily ignored here.
167 printk(KERN_INFO
"btrfs: setting nodatasum\n");
168 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
171 printk(KERN_INFO
"btrfs: setting nodatacow\n");
172 btrfs_set_opt(info
->mount_opt
, NODATACOW
);
173 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
176 printk(KERN_INFO
"btrfs: use compression\n");
177 btrfs_set_opt(info
->mount_opt
, COMPRESS
);
179 case Opt_compress_force
:
180 printk(KERN_INFO
"btrfs: forcing compression\n");
181 btrfs_set_opt(info
->mount_opt
, FORCE_COMPRESS
);
182 btrfs_set_opt(info
->mount_opt
, COMPRESS
);
185 printk(KERN_INFO
"btrfs: use ssd allocation scheme\n");
186 btrfs_set_opt(info
->mount_opt
, SSD
);
189 printk(KERN_INFO
"btrfs: use spread ssd "
190 "allocation scheme\n");
191 btrfs_set_opt(info
->mount_opt
, SSD
);
192 btrfs_set_opt(info
->mount_opt
, SSD_SPREAD
);
195 printk(KERN_INFO
"btrfs: not using ssd allocation "
197 btrfs_set_opt(info
->mount_opt
, NOSSD
);
198 btrfs_clear_opt(info
->mount_opt
, SSD
);
199 btrfs_clear_opt(info
->mount_opt
, SSD_SPREAD
);
202 printk(KERN_INFO
"btrfs: turning off barriers\n");
203 btrfs_set_opt(info
->mount_opt
, NOBARRIER
);
205 case Opt_thread_pool
:
207 match_int(&args
[0], &intarg
);
209 info
->thread_pool_size
= intarg
;
210 printk(KERN_INFO
"btrfs: thread pool %d\n",
211 info
->thread_pool_size
);
215 num
= match_strdup(&args
[0]);
217 info
->max_extent
= btrfs_parse_size(num
);
220 info
->max_extent
= max_t(u64
,
221 info
->max_extent
, root
->sectorsize
);
222 printk(KERN_INFO
"btrfs: max_extent at %llu\n",
223 (unsigned long long)info
->max_extent
);
227 num
= match_strdup(&args
[0]);
229 info
->max_inline
= btrfs_parse_size(num
);
232 if (info
->max_inline
) {
233 info
->max_inline
= max_t(u64
,
237 printk(KERN_INFO
"btrfs: max_inline at %llu\n",
238 (unsigned long long)info
->max_inline
);
241 case Opt_alloc_start
:
242 num
= match_strdup(&args
[0]);
244 info
->alloc_start
= btrfs_parse_size(num
);
247 "btrfs: allocations start at %llu\n",
248 (unsigned long long)info
->alloc_start
);
252 root
->fs_info
->sb
->s_flags
&= ~MS_POSIXACL
;
255 printk(KERN_INFO
"btrfs: disabling tree log\n");
256 btrfs_set_opt(info
->mount_opt
, NOTREELOG
);
258 case Opt_flushoncommit
:
259 printk(KERN_INFO
"btrfs: turning on flush-on-commit\n");
260 btrfs_set_opt(info
->mount_opt
, FLUSHONCOMMIT
);
264 match_int(&args
[0], &intarg
);
266 info
->metadata_ratio
= intarg
;
267 printk(KERN_INFO
"btrfs: metadata ratio %d\n",
268 info
->metadata_ratio
);
272 btrfs_set_opt(info
->mount_opt
, DISCARD
);
275 printk(KERN_INFO
"btrfs: unrecognized mount option "
289 * Parse mount options that are required early in the mount process.
291 * All other options will be parsed on much later in the mount process and
292 * only when we need to allocate a new super block.
294 static int btrfs_parse_early_options(const char *options
, fmode_t flags
,
295 void *holder
, char **subvol_name
,
296 struct btrfs_fs_devices
**fs_devices
)
298 substring_t args
[MAX_OPT_ARGS
];
306 * strsep changes the string, duplicate it because parse_options
309 opts
= kstrdup(options
, GFP_KERNEL
);
313 while ((p
= strsep(&opts
, ",")) != NULL
) {
318 token
= match_token(p
, tokens
, args
);
321 *subvol_name
= match_strdup(&args
[0]);
324 error
= btrfs_scan_one_device(match_strdup(&args
[0]),
325 flags
, holder
, fs_devices
);
338 * If no subvolume name is specified we use the default one. Allocate
339 * a copy of the string "." here so that code later in the
340 * mount path doesn't care if it's the default volume or another one.
343 *subvol_name
= kstrdup(".", GFP_KERNEL
);
350 static int btrfs_fill_super(struct super_block
*sb
,
351 struct btrfs_fs_devices
*fs_devices
,
352 void *data
, int silent
)
355 struct dentry
*root_dentry
;
356 struct btrfs_super_block
*disk_super
;
357 struct btrfs_root
*tree_root
;
358 struct btrfs_key key
;
361 sb
->s_maxbytes
= MAX_LFS_FILESIZE
;
362 sb
->s_magic
= BTRFS_SUPER_MAGIC
;
363 sb
->s_op
= &btrfs_super_ops
;
364 sb
->s_export_op
= &btrfs_export_ops
;
365 sb
->s_xattr
= btrfs_xattr_handlers
;
367 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
368 sb
->s_flags
|= MS_POSIXACL
;
371 tree_root
= open_ctree(sb
, fs_devices
, (char *)data
);
373 if (IS_ERR(tree_root
)) {
374 printk("btrfs: open_ctree failed\n");
375 return PTR_ERR(tree_root
);
377 sb
->s_fs_info
= tree_root
;
378 disk_super
= &tree_root
->fs_info
->super_copy
;
380 key
.objectid
= BTRFS_FIRST_FREE_OBJECTID
;
381 key
.type
= BTRFS_INODE_ITEM_KEY
;
383 inode
= btrfs_iget(sb
, &key
, tree_root
->fs_info
->fs_root
);
385 err
= PTR_ERR(inode
);
389 root_dentry
= d_alloc_root(inode
);
396 /* this does the super kobj at the same time */
397 err
= btrfs_sysfs_add_super(tree_root
->fs_info
);
402 sb
->s_root
= root_dentry
;
404 save_mount_options(sb
, data
);
408 close_ctree(tree_root
);
412 int btrfs_sync_fs(struct super_block
*sb
, int wait
)
414 struct btrfs_trans_handle
*trans
;
415 struct btrfs_root
*root
= btrfs_sb(sb
);
419 filemap_flush(root
->fs_info
->btree_inode
->i_mapping
);
423 btrfs_start_delalloc_inodes(root
, 0);
424 btrfs_wait_ordered_extents(root
, 0, 0);
426 trans
= btrfs_start_transaction(root
, 1);
427 ret
= btrfs_commit_transaction(trans
, root
);
431 static int btrfs_show_options(struct seq_file
*seq
, struct vfsmount
*vfs
)
433 struct btrfs_root
*root
= btrfs_sb(vfs
->mnt_sb
);
434 struct btrfs_fs_info
*info
= root
->fs_info
;
436 if (btrfs_test_opt(root
, DEGRADED
))
437 seq_puts(seq
, ",degraded");
438 if (btrfs_test_opt(root
, NODATASUM
))
439 seq_puts(seq
, ",nodatasum");
440 if (btrfs_test_opt(root
, NODATACOW
))
441 seq_puts(seq
, ",nodatacow");
442 if (btrfs_test_opt(root
, NOBARRIER
))
443 seq_puts(seq
, ",nobarrier");
444 if (info
->max_extent
!= (u64
)-1)
445 seq_printf(seq
, ",max_extent=%llu",
446 (unsigned long long)info
->max_extent
);
447 if (info
->max_inline
!= 8192 * 1024)
448 seq_printf(seq
, ",max_inline=%llu",
449 (unsigned long long)info
->max_inline
);
450 if (info
->alloc_start
!= 0)
451 seq_printf(seq
, ",alloc_start=%llu",
452 (unsigned long long)info
->alloc_start
);
453 if (info
->thread_pool_size
!= min_t(unsigned long,
454 num_online_cpus() + 2, 8))
455 seq_printf(seq
, ",thread_pool=%d", info
->thread_pool_size
);
456 if (btrfs_test_opt(root
, COMPRESS
))
457 seq_puts(seq
, ",compress");
458 if (btrfs_test_opt(root
, NOSSD
))
459 seq_puts(seq
, ",nossd");
460 if (btrfs_test_opt(root
, SSD_SPREAD
))
461 seq_puts(seq
, ",ssd_spread");
462 else if (btrfs_test_opt(root
, SSD
))
463 seq_puts(seq
, ",ssd");
464 if (btrfs_test_opt(root
, NOTREELOG
))
465 seq_puts(seq
, ",notreelog");
466 if (btrfs_test_opt(root
, FLUSHONCOMMIT
))
467 seq_puts(seq
, ",flushoncommit");
468 if (btrfs_test_opt(root
, DISCARD
))
469 seq_puts(seq
, ",discard");
470 if (!(root
->fs_info
->sb
->s_flags
& MS_POSIXACL
))
471 seq_puts(seq
, ",noacl");
475 static int btrfs_test_super(struct super_block
*s
, void *data
)
477 struct btrfs_fs_devices
*test_fs_devices
= data
;
478 struct btrfs_root
*root
= btrfs_sb(s
);
480 return root
->fs_info
->fs_devices
== test_fs_devices
;
484 * Find a superblock for the given device / mount point.
486 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
487 * for multiple device setup. Make sure to keep it in sync.
489 static int btrfs_get_sb(struct file_system_type
*fs_type
, int flags
,
490 const char *dev_name
, void *data
, struct vfsmount
*mnt
)
492 char *subvol_name
= NULL
;
493 struct block_device
*bdev
= NULL
;
494 struct super_block
*s
;
496 struct btrfs_fs_devices
*fs_devices
= NULL
;
497 fmode_t mode
= FMODE_READ
;
500 if (!(flags
& MS_RDONLY
))
503 error
= btrfs_parse_early_options(data
, mode
, fs_type
,
504 &subvol_name
, &fs_devices
);
508 error
= btrfs_scan_one_device(dev_name
, mode
, fs_type
, &fs_devices
);
510 goto error_free_subvol_name
;
512 error
= btrfs_open_devices(fs_devices
, mode
, fs_type
);
514 goto error_free_subvol_name
;
516 if (!(flags
& MS_RDONLY
) && fs_devices
->rw_devices
== 0) {
518 goto error_close_devices
;
521 bdev
= fs_devices
->latest_bdev
;
522 s
= sget(fs_type
, btrfs_test_super
, set_anon_super
, fs_devices
);
527 if ((flags
^ s
->s_flags
) & MS_RDONLY
) {
528 deactivate_locked_super(s
);
530 goto error_close_devices
;
533 btrfs_close_devices(fs_devices
);
535 char b
[BDEVNAME_SIZE
];
538 strlcpy(s
->s_id
, bdevname(bdev
, b
), sizeof(s
->s_id
));
539 error
= btrfs_fill_super(s
, fs_devices
, data
,
540 flags
& MS_SILENT
? 1 : 0);
542 deactivate_locked_super(s
);
543 goto error_free_subvol_name
;
546 btrfs_sb(s
)->fs_info
->bdev_holder
= fs_type
;
547 s
->s_flags
|= MS_ACTIVE
;
550 if (!strcmp(subvol_name
, "."))
551 root
= dget(s
->s_root
);
553 mutex_lock(&s
->s_root
->d_inode
->i_mutex
);
554 root
= lookup_one_len(subvol_name
, s
->s_root
,
555 strlen(subvol_name
));
556 mutex_unlock(&s
->s_root
->d_inode
->i_mutex
);
559 deactivate_locked_super(s
);
560 error
= PTR_ERR(root
);
561 goto error_free_subvol_name
;
563 if (!root
->d_inode
) {
565 deactivate_locked_super(s
);
567 goto error_free_subvol_name
;
572 mnt
->mnt_root
= root
;
580 btrfs_close_devices(fs_devices
);
581 error_free_subvol_name
:
586 static int btrfs_remount(struct super_block
*sb
, int *flags
, char *data
)
588 struct btrfs_root
*root
= btrfs_sb(sb
);
591 ret
= btrfs_parse_options(root
, data
);
595 if ((*flags
& MS_RDONLY
) == (sb
->s_flags
& MS_RDONLY
))
598 if (*flags
& MS_RDONLY
) {
599 sb
->s_flags
|= MS_RDONLY
;
601 ret
= btrfs_commit_super(root
);
604 if (root
->fs_info
->fs_devices
->rw_devices
== 0)
607 if (btrfs_super_log_root(&root
->fs_info
->super_copy
) != 0)
610 /* recover relocation */
611 ret
= btrfs_recover_relocation(root
);
614 ret
= btrfs_cleanup_fs_roots(root
->fs_info
);
617 sb
->s_flags
&= ~MS_RDONLY
;
623 static int btrfs_statfs(struct dentry
*dentry
, struct kstatfs
*buf
)
625 struct btrfs_root
*root
= btrfs_sb(dentry
->d_sb
);
626 struct btrfs_super_block
*disk_super
= &root
->fs_info
->super_copy
;
627 int bits
= dentry
->d_sb
->s_blocksize_bits
;
628 __be32
*fsid
= (__be32
*)root
->fs_info
->fsid
;
630 buf
->f_namelen
= BTRFS_NAME_LEN
;
631 buf
->f_blocks
= btrfs_super_total_bytes(disk_super
) >> bits
;
632 buf
->f_bfree
= buf
->f_blocks
-
633 (btrfs_super_bytes_used(disk_super
) >> bits
);
634 buf
->f_bavail
= buf
->f_bfree
;
635 buf
->f_bsize
= dentry
->d_sb
->s_blocksize
;
636 buf
->f_type
= BTRFS_SUPER_MAGIC
;
638 /* We treat it as constant endianness (it doesn't matter _which_)
639 because we want the fsid to come out the same whether mounted
640 on a big-endian or little-endian host */
641 buf
->f_fsid
.val
[0] = be32_to_cpu(fsid
[0]) ^ be32_to_cpu(fsid
[2]);
642 buf
->f_fsid
.val
[1] = be32_to_cpu(fsid
[1]) ^ be32_to_cpu(fsid
[3]);
643 /* Mask in the root object ID too, to disambiguate subvols */
644 buf
->f_fsid
.val
[0] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
>> 32;
645 buf
->f_fsid
.val
[1] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
;
650 static struct file_system_type btrfs_fs_type
= {
651 .owner
= THIS_MODULE
,
653 .get_sb
= btrfs_get_sb
,
654 .kill_sb
= kill_anon_super
,
655 .fs_flags
= FS_REQUIRES_DEV
,
659 * used by btrfsctl to scan devices when no FS is mounted
661 static long btrfs_control_ioctl(struct file
*file
, unsigned int cmd
,
664 struct btrfs_ioctl_vol_args
*vol
;
665 struct btrfs_fs_devices
*fs_devices
;
668 if (!capable(CAP_SYS_ADMIN
))
671 vol
= memdup_user((void __user
*)arg
, sizeof(*vol
));
676 case BTRFS_IOC_SCAN_DEV
:
677 ret
= btrfs_scan_one_device(vol
->name
, FMODE_READ
,
678 &btrfs_fs_type
, &fs_devices
);
686 static int btrfs_freeze(struct super_block
*sb
)
688 struct btrfs_root
*root
= btrfs_sb(sb
);
689 mutex_lock(&root
->fs_info
->transaction_kthread_mutex
);
690 mutex_lock(&root
->fs_info
->cleaner_mutex
);
694 static int btrfs_unfreeze(struct super_block
*sb
)
696 struct btrfs_root
*root
= btrfs_sb(sb
);
697 mutex_unlock(&root
->fs_info
->cleaner_mutex
);
698 mutex_unlock(&root
->fs_info
->transaction_kthread_mutex
);
702 static const struct super_operations btrfs_super_ops
= {
703 .drop_inode
= btrfs_drop_inode
,
704 .delete_inode
= btrfs_delete_inode
,
705 .put_super
= btrfs_put_super
,
706 .sync_fs
= btrfs_sync_fs
,
707 .show_options
= btrfs_show_options
,
708 .write_inode
= btrfs_write_inode
,
709 .dirty_inode
= btrfs_dirty_inode
,
710 .alloc_inode
= btrfs_alloc_inode
,
711 .destroy_inode
= btrfs_destroy_inode
,
712 .statfs
= btrfs_statfs
,
713 .remount_fs
= btrfs_remount
,
714 .freeze_fs
= btrfs_freeze
,
715 .unfreeze_fs
= btrfs_unfreeze
,
718 static const struct file_operations btrfs_ctl_fops
= {
719 .unlocked_ioctl
= btrfs_control_ioctl
,
720 .compat_ioctl
= btrfs_control_ioctl
,
721 .owner
= THIS_MODULE
,
724 static struct miscdevice btrfs_misc
= {
725 .minor
= MISC_DYNAMIC_MINOR
,
726 .name
= "btrfs-control",
727 .fops
= &btrfs_ctl_fops
730 static int btrfs_interface_init(void)
732 return misc_register(&btrfs_misc
);
735 static void btrfs_interface_exit(void)
737 if (misc_deregister(&btrfs_misc
) < 0)
738 printk(KERN_INFO
"misc_deregister failed for control device");
741 static int __init
init_btrfs_fs(void)
745 err
= btrfs_init_sysfs();
749 err
= btrfs_init_cachep();
753 err
= extent_io_init();
757 err
= extent_map_init();
761 err
= btrfs_interface_init();
763 goto free_extent_map
;
765 err
= register_filesystem(&btrfs_fs_type
);
767 goto unregister_ioctl
;
769 printk(KERN_INFO
"%s loaded\n", BTRFS_BUILD_VERSION
);
773 btrfs_interface_exit();
779 btrfs_destroy_cachep();
785 static void __exit
exit_btrfs_fs(void)
787 btrfs_destroy_cachep();
790 btrfs_interface_exit();
791 unregister_filesystem(&btrfs_fs_type
);
793 btrfs_cleanup_fs_uuids();
797 module_init(init_btrfs_fs
)
798 module_exit(exit_btrfs_fs
)
800 MODULE_LICENSE("GPL");