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_notreelog
, Opt_ratio
, Opt_flushoncommit
, Opt_err
,
72 static match_table_t tokens
= {
73 {Opt_degraded
, "degraded"},
74 {Opt_subvol
, "subvol=%s"},
75 {Opt_device
, "device=%s"},
76 {Opt_nodatasum
, "nodatasum"},
77 {Opt_nodatacow
, "nodatacow"},
78 {Opt_nobarrier
, "nobarrier"},
79 {Opt_max_extent
, "max_extent=%s"},
80 {Opt_max_inline
, "max_inline=%s"},
81 {Opt_alloc_start
, "alloc_start=%s"},
82 {Opt_thread_pool
, "thread_pool=%d"},
83 {Opt_compress
, "compress"},
85 {Opt_ssd_spread
, "ssd_spread"},
88 {Opt_notreelog
, "notreelog"},
89 {Opt_flushoncommit
, "flushoncommit"},
90 {Opt_ratio
, "metadata_ratio=%d"},
94 u64
btrfs_parse_size(char *str
)
101 res
= simple_strtoul(str
, &end
, 10);
105 last
= tolower(last
);
120 * Regular mount options parser. Everything that is needed only when
121 * reading in a new superblock is parsed here.
123 int btrfs_parse_options(struct btrfs_root
*root
, char *options
)
125 struct btrfs_fs_info
*info
= root
->fs_info
;
126 substring_t args
[MAX_OPT_ARGS
];
134 * strsep changes the string, duplicate it because parse_options
137 options
= kstrdup(options
, GFP_NOFS
);
142 while ((p
= strsep(&options
, ",")) != NULL
) {
147 token
= match_token(p
, tokens
, args
);
150 printk(KERN_INFO
"btrfs: allowing degraded mounts\n");
151 btrfs_set_opt(info
->mount_opt
, DEGRADED
);
156 * These are parsed by btrfs_parse_early_options
157 * and can be happily ignored here.
161 printk(KERN_INFO
"btrfs: setting nodatasum\n");
162 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
165 printk(KERN_INFO
"btrfs: setting nodatacow\n");
166 btrfs_set_opt(info
->mount_opt
, NODATACOW
);
167 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
170 printk(KERN_INFO
"btrfs: use compression\n");
171 btrfs_set_opt(info
->mount_opt
, COMPRESS
);
174 printk(KERN_INFO
"btrfs: use ssd allocation scheme\n");
175 btrfs_set_opt(info
->mount_opt
, SSD
);
178 printk(KERN_INFO
"btrfs: use spread ssd "
179 "allocation scheme\n");
180 btrfs_set_opt(info
->mount_opt
, SSD
);
181 btrfs_set_opt(info
->mount_opt
, SSD_SPREAD
);
184 printk(KERN_INFO
"btrfs: not using ssd allocation "
186 btrfs_set_opt(info
->mount_opt
, NOSSD
);
187 btrfs_clear_opt(info
->mount_opt
, SSD
);
188 btrfs_clear_opt(info
->mount_opt
, SSD_SPREAD
);
191 printk(KERN_INFO
"btrfs: turning off barriers\n");
192 btrfs_set_opt(info
->mount_opt
, NOBARRIER
);
194 case Opt_thread_pool
:
196 match_int(&args
[0], &intarg
);
198 info
->thread_pool_size
= intarg
;
199 printk(KERN_INFO
"btrfs: thread pool %d\n",
200 info
->thread_pool_size
);
204 num
= match_strdup(&args
[0]);
206 info
->max_extent
= btrfs_parse_size(num
);
209 info
->max_extent
= max_t(u64
,
210 info
->max_extent
, root
->sectorsize
);
211 printk(KERN_INFO
"btrfs: max_extent at %llu\n",
212 (unsigned long long)info
->max_extent
);
216 num
= match_strdup(&args
[0]);
218 info
->max_inline
= btrfs_parse_size(num
);
221 if (info
->max_inline
) {
222 info
->max_inline
= max_t(u64
,
226 printk(KERN_INFO
"btrfs: max_inline at %llu\n",
227 (unsigned long long)info
->max_inline
);
230 case Opt_alloc_start
:
231 num
= match_strdup(&args
[0]);
233 info
->alloc_start
= btrfs_parse_size(num
);
236 "btrfs: allocations start at %llu\n",
237 (unsigned long long)info
->alloc_start
);
241 root
->fs_info
->sb
->s_flags
&= ~MS_POSIXACL
;
244 printk(KERN_INFO
"btrfs: disabling tree log\n");
245 btrfs_set_opt(info
->mount_opt
, NOTREELOG
);
247 case Opt_flushoncommit
:
248 printk(KERN_INFO
"btrfs: turning on flush-on-commit\n");
249 btrfs_set_opt(info
->mount_opt
, FLUSHONCOMMIT
);
253 match_int(&args
[0], &intarg
);
255 info
->metadata_ratio
= intarg
;
256 printk(KERN_INFO
"btrfs: metadata ratio %d\n",
257 info
->metadata_ratio
);
269 * Parse mount options that are required early in the mount process.
271 * All other options will be parsed on much later in the mount process and
272 * only when we need to allocate a new super block.
274 static int btrfs_parse_early_options(const char *options
, fmode_t flags
,
275 void *holder
, char **subvol_name
,
276 struct btrfs_fs_devices
**fs_devices
)
278 substring_t args
[MAX_OPT_ARGS
];
286 * strsep changes the string, duplicate it because parse_options
289 opts
= kstrdup(options
, GFP_KERNEL
);
293 while ((p
= strsep(&opts
, ",")) != NULL
) {
298 token
= match_token(p
, tokens
, args
);
301 *subvol_name
= match_strdup(&args
[0]);
304 error
= btrfs_scan_one_device(match_strdup(&args
[0]),
305 flags
, holder
, fs_devices
);
318 * If no subvolume name is specified we use the default one. Allocate
319 * a copy of the string "." here so that code later in the
320 * mount path doesn't care if it's the default volume or another one.
323 *subvol_name
= kstrdup(".", GFP_KERNEL
);
330 static int btrfs_fill_super(struct super_block
*sb
,
331 struct btrfs_fs_devices
*fs_devices
,
332 void *data
, int silent
)
335 struct dentry
*root_dentry
;
336 struct btrfs_super_block
*disk_super
;
337 struct btrfs_root
*tree_root
;
338 struct btrfs_key key
;
341 sb
->s_maxbytes
= MAX_LFS_FILESIZE
;
342 sb
->s_magic
= BTRFS_SUPER_MAGIC
;
343 sb
->s_op
= &btrfs_super_ops
;
344 sb
->s_export_op
= &btrfs_export_ops
;
345 sb
->s_xattr
= btrfs_xattr_handlers
;
347 sb
->s_flags
|= MS_POSIXACL
;
349 tree_root
= open_ctree(sb
, fs_devices
, (char *)data
);
351 if (IS_ERR(tree_root
)) {
352 printk("btrfs: open_ctree failed\n");
353 return PTR_ERR(tree_root
);
355 sb
->s_fs_info
= tree_root
;
356 disk_super
= &tree_root
->fs_info
->super_copy
;
358 key
.objectid
= BTRFS_FIRST_FREE_OBJECTID
;
359 key
.type
= BTRFS_INODE_ITEM_KEY
;
361 inode
= btrfs_iget(sb
, &key
, tree_root
->fs_info
->fs_root
);
363 err
= PTR_ERR(inode
);
367 root_dentry
= d_alloc_root(inode
);
374 /* this does the super kobj at the same time */
375 err
= btrfs_sysfs_add_super(tree_root
->fs_info
);
380 sb
->s_root
= root_dentry
;
382 save_mount_options(sb
, data
);
386 close_ctree(tree_root
);
390 int btrfs_sync_fs(struct super_block
*sb
, int wait
)
392 struct btrfs_trans_handle
*trans
;
393 struct btrfs_root
*root
= btrfs_sb(sb
);
397 filemap_flush(root
->fs_info
->btree_inode
->i_mapping
);
401 btrfs_start_delalloc_inodes(root
);
402 btrfs_wait_ordered_extents(root
, 0);
404 trans
= btrfs_start_transaction(root
, 1);
405 ret
= btrfs_commit_transaction(trans
, root
);
409 static int btrfs_show_options(struct seq_file
*seq
, struct vfsmount
*vfs
)
411 struct btrfs_root
*root
= btrfs_sb(vfs
->mnt_sb
);
412 struct btrfs_fs_info
*info
= root
->fs_info
;
414 if (btrfs_test_opt(root
, DEGRADED
))
415 seq_puts(seq
, ",degraded");
416 if (btrfs_test_opt(root
, NODATASUM
))
417 seq_puts(seq
, ",nodatasum");
418 if (btrfs_test_opt(root
, NODATACOW
))
419 seq_puts(seq
, ",nodatacow");
420 if (btrfs_test_opt(root
, NOBARRIER
))
421 seq_puts(seq
, ",nobarrier");
422 if (info
->max_extent
!= (u64
)-1)
423 seq_printf(seq
, ",max_extent=%llu",
424 (unsigned long long)info
->max_extent
);
425 if (info
->max_inline
!= 8192 * 1024)
426 seq_printf(seq
, ",max_inline=%llu",
427 (unsigned long long)info
->max_inline
);
428 if (info
->alloc_start
!= 0)
429 seq_printf(seq
, ",alloc_start=%llu",
430 (unsigned long long)info
->alloc_start
);
431 if (info
->thread_pool_size
!= min_t(unsigned long,
432 num_online_cpus() + 2, 8))
433 seq_printf(seq
, ",thread_pool=%d", info
->thread_pool_size
);
434 if (btrfs_test_opt(root
, COMPRESS
))
435 seq_puts(seq
, ",compress");
436 if (btrfs_test_opt(root
, NOSSD
))
437 seq_puts(seq
, ",nossd");
438 if (btrfs_test_opt(root
, SSD_SPREAD
))
439 seq_puts(seq
, ",ssd_spread");
440 else if (btrfs_test_opt(root
, SSD
))
441 seq_puts(seq
, ",ssd");
442 if (btrfs_test_opt(root
, NOTREELOG
))
443 seq_puts(seq
, ",notreelog");
444 if (btrfs_test_opt(root
, FLUSHONCOMMIT
))
445 seq_puts(seq
, ",flushoncommit");
446 if (!(root
->fs_info
->sb
->s_flags
& MS_POSIXACL
))
447 seq_puts(seq
, ",noacl");
451 static int btrfs_test_super(struct super_block
*s
, void *data
)
453 struct btrfs_fs_devices
*test_fs_devices
= data
;
454 struct btrfs_root
*root
= btrfs_sb(s
);
456 return root
->fs_info
->fs_devices
== test_fs_devices
;
460 * Find a superblock for the given device / mount point.
462 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
463 * for multiple device setup. Make sure to keep it in sync.
465 static int btrfs_get_sb(struct file_system_type
*fs_type
, int flags
,
466 const char *dev_name
, void *data
, struct vfsmount
*mnt
)
468 char *subvol_name
= NULL
;
469 struct block_device
*bdev
= NULL
;
470 struct super_block
*s
;
472 struct btrfs_fs_devices
*fs_devices
= NULL
;
473 fmode_t mode
= FMODE_READ
;
476 if (!(flags
& MS_RDONLY
))
479 error
= btrfs_parse_early_options(data
, mode
, fs_type
,
480 &subvol_name
, &fs_devices
);
484 error
= btrfs_scan_one_device(dev_name
, mode
, fs_type
, &fs_devices
);
486 goto error_free_subvol_name
;
488 error
= btrfs_open_devices(fs_devices
, mode
, fs_type
);
490 goto error_free_subvol_name
;
492 if (!(flags
& MS_RDONLY
) && fs_devices
->rw_devices
== 0) {
494 goto error_close_devices
;
497 bdev
= fs_devices
->latest_bdev
;
498 s
= sget(fs_type
, btrfs_test_super
, set_anon_super
, fs_devices
);
503 if ((flags
^ s
->s_flags
) & MS_RDONLY
) {
504 deactivate_locked_super(s
);
506 goto error_close_devices
;
509 btrfs_close_devices(fs_devices
);
511 char b
[BDEVNAME_SIZE
];
514 strlcpy(s
->s_id
, bdevname(bdev
, b
), sizeof(s
->s_id
));
515 error
= btrfs_fill_super(s
, fs_devices
, data
,
516 flags
& MS_SILENT
? 1 : 0);
518 deactivate_locked_super(s
);
519 goto error_free_subvol_name
;
522 btrfs_sb(s
)->fs_info
->bdev_holder
= fs_type
;
523 s
->s_flags
|= MS_ACTIVE
;
526 if (!strcmp(subvol_name
, "."))
527 root
= dget(s
->s_root
);
529 mutex_lock(&s
->s_root
->d_inode
->i_mutex
);
530 root
= lookup_one_len(subvol_name
, s
->s_root
,
531 strlen(subvol_name
));
532 mutex_unlock(&s
->s_root
->d_inode
->i_mutex
);
535 deactivate_locked_super(s
);
536 error
= PTR_ERR(root
);
537 goto error_free_subvol_name
;
539 if (!root
->d_inode
) {
541 deactivate_locked_super(s
);
543 goto error_free_subvol_name
;
548 mnt
->mnt_root
= root
;
556 btrfs_close_devices(fs_devices
);
557 error_free_subvol_name
:
562 static int btrfs_remount(struct super_block
*sb
, int *flags
, char *data
)
564 struct btrfs_root
*root
= btrfs_sb(sb
);
567 ret
= btrfs_parse_options(root
, data
);
571 if ((*flags
& MS_RDONLY
) == (sb
->s_flags
& MS_RDONLY
))
574 if (*flags
& MS_RDONLY
) {
575 sb
->s_flags
|= MS_RDONLY
;
577 ret
= btrfs_commit_super(root
);
580 if (root
->fs_info
->fs_devices
->rw_devices
== 0)
583 if (btrfs_super_log_root(&root
->fs_info
->super_copy
) != 0)
586 /* recover relocation */
587 ret
= btrfs_recover_relocation(root
);
590 ret
= btrfs_cleanup_fs_roots(root
->fs_info
);
593 sb
->s_flags
&= ~MS_RDONLY
;
599 static int btrfs_statfs(struct dentry
*dentry
, struct kstatfs
*buf
)
601 struct btrfs_root
*root
= btrfs_sb(dentry
->d_sb
);
602 struct btrfs_super_block
*disk_super
= &root
->fs_info
->super_copy
;
603 int bits
= dentry
->d_sb
->s_blocksize_bits
;
604 __be32
*fsid
= (__be32
*)root
->fs_info
->fsid
;
606 buf
->f_namelen
= BTRFS_NAME_LEN
;
607 buf
->f_blocks
= btrfs_super_total_bytes(disk_super
) >> bits
;
608 buf
->f_bfree
= buf
->f_blocks
-
609 (btrfs_super_bytes_used(disk_super
) >> bits
);
610 buf
->f_bavail
= buf
->f_bfree
;
611 buf
->f_bsize
= dentry
->d_sb
->s_blocksize
;
612 buf
->f_type
= BTRFS_SUPER_MAGIC
;
614 /* We treat it as constant endianness (it doesn't matter _which_)
615 because we want the fsid to come out the same whether mounted
616 on a big-endian or little-endian host */
617 buf
->f_fsid
.val
[0] = be32_to_cpu(fsid
[0]) ^ be32_to_cpu(fsid
[2]);
618 buf
->f_fsid
.val
[1] = be32_to_cpu(fsid
[1]) ^ be32_to_cpu(fsid
[3]);
619 /* Mask in the root object ID too, to disambiguate subvols */
620 buf
->f_fsid
.val
[0] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
>> 32;
621 buf
->f_fsid
.val
[1] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
;
626 static struct file_system_type btrfs_fs_type
= {
627 .owner
= THIS_MODULE
,
629 .get_sb
= btrfs_get_sb
,
630 .kill_sb
= kill_anon_super
,
631 .fs_flags
= FS_REQUIRES_DEV
,
635 * used by btrfsctl to scan devices when no FS is mounted
637 static long btrfs_control_ioctl(struct file
*file
, unsigned int cmd
,
640 struct btrfs_ioctl_vol_args
*vol
;
641 struct btrfs_fs_devices
*fs_devices
;
644 if (!capable(CAP_SYS_ADMIN
))
647 vol
= memdup_user((void __user
*)arg
, sizeof(*vol
));
652 case BTRFS_IOC_SCAN_DEV
:
653 ret
= btrfs_scan_one_device(vol
->name
, FMODE_READ
,
654 &btrfs_fs_type
, &fs_devices
);
662 static int btrfs_freeze(struct super_block
*sb
)
664 struct btrfs_root
*root
= btrfs_sb(sb
);
665 mutex_lock(&root
->fs_info
->transaction_kthread_mutex
);
666 mutex_lock(&root
->fs_info
->cleaner_mutex
);
670 static int btrfs_unfreeze(struct super_block
*sb
)
672 struct btrfs_root
*root
= btrfs_sb(sb
);
673 mutex_unlock(&root
->fs_info
->cleaner_mutex
);
674 mutex_unlock(&root
->fs_info
->transaction_kthread_mutex
);
678 static const struct super_operations btrfs_super_ops
= {
679 .delete_inode
= btrfs_delete_inode
,
680 .put_super
= btrfs_put_super
,
681 .sync_fs
= btrfs_sync_fs
,
682 .show_options
= btrfs_show_options
,
683 .write_inode
= btrfs_write_inode
,
684 .dirty_inode
= btrfs_dirty_inode
,
685 .alloc_inode
= btrfs_alloc_inode
,
686 .destroy_inode
= btrfs_destroy_inode
,
687 .statfs
= btrfs_statfs
,
688 .remount_fs
= btrfs_remount
,
689 .freeze_fs
= btrfs_freeze
,
690 .unfreeze_fs
= btrfs_unfreeze
,
693 static const struct file_operations btrfs_ctl_fops
= {
694 .unlocked_ioctl
= btrfs_control_ioctl
,
695 .compat_ioctl
= btrfs_control_ioctl
,
696 .owner
= THIS_MODULE
,
699 static struct miscdevice btrfs_misc
= {
700 .minor
= MISC_DYNAMIC_MINOR
,
701 .name
= "btrfs-control",
702 .fops
= &btrfs_ctl_fops
705 static int btrfs_interface_init(void)
707 return misc_register(&btrfs_misc
);
710 static void btrfs_interface_exit(void)
712 if (misc_deregister(&btrfs_misc
) < 0)
713 printk(KERN_INFO
"misc_deregister failed for control device");
716 static int __init
init_btrfs_fs(void)
720 err
= btrfs_init_sysfs();
724 err
= btrfs_init_cachep();
728 err
= extent_io_init();
732 err
= extent_map_init();
736 err
= btrfs_interface_init();
738 goto free_extent_map
;
740 err
= register_filesystem(&btrfs_fs_type
);
742 goto unregister_ioctl
;
744 printk(KERN_INFO
"%s loaded\n", BTRFS_BUILD_VERSION
);
748 btrfs_interface_exit();
754 btrfs_destroy_cachep();
760 static void __exit
exit_btrfs_fs(void)
762 btrfs_destroy_cachep();
765 btrfs_interface_exit();
766 unregister_filesystem(&btrfs_fs_type
);
768 btrfs_cleanup_fs_uuids();
772 module_init(init_btrfs_fs
)
773 module_exit(exit_btrfs_fs
)
775 MODULE_LICENSE("GPL");