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/smp_lock.h>
30 #include <linux/backing-dev.h>
31 #include <linux/mount.h>
32 #include <linux/mpage.h>
33 #include <linux/swap.h>
34 #include <linux/writeback.h>
35 #include <linux/statfs.h>
36 #include <linux/compat.h>
37 #include <linux/parser.h>
38 #include <linux/ctype.h>
39 #include <linux/namei.h>
40 #include <linux/miscdevice.h>
41 #include <linux/magic.h>
45 #include "transaction.h"
46 #include "btrfs_inode.h"
48 #include "print-tree.h"
53 #include "compression.h"
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
);
63 ret
= close_ctree(root
);
68 Opt_degraded
, Opt_subvol
, Opt_device
, Opt_nodatasum
, Opt_nodatacow
,
69 Opt_max_extent
, Opt_max_inline
, Opt_alloc_start
, Opt_nobarrier
,
70 Opt_ssd
, Opt_thread_pool
, Opt_noacl
, Opt_compress
, Opt_notreelog
,
71 Opt_ratio
, Opt_flushoncommit
, Opt_err
,
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"},
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 nodatacsum\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: turning off barriers\n");
179 btrfs_set_opt(info
->mount_opt
, NOBARRIER
);
181 case Opt_thread_pool
:
183 match_int(&args
[0], &intarg
);
185 info
->thread_pool_size
= intarg
;
186 printk(KERN_INFO
"btrfs: thread pool %d\n",
187 info
->thread_pool_size
);
191 num
= match_strdup(&args
[0]);
193 info
->max_extent
= btrfs_parse_size(num
);
196 info
->max_extent
= max_t(u64
,
197 info
->max_extent
, root
->sectorsize
);
198 printk(KERN_INFO
"btrfs: max_extent at %llu\n",
199 (unsigned long long)info
->max_extent
);
203 num
= match_strdup(&args
[0]);
205 info
->max_inline
= btrfs_parse_size(num
);
208 if (info
->max_inline
) {
209 info
->max_inline
= max_t(u64
,
213 printk(KERN_INFO
"btrfs: max_inline at %llu\n",
214 (unsigned long long)info
->max_inline
);
217 case Opt_alloc_start
:
218 num
= match_strdup(&args
[0]);
220 info
->alloc_start
= btrfs_parse_size(num
);
223 "btrfs: allocations start at %llu\n",
224 (unsigned long long)info
->alloc_start
);
228 root
->fs_info
->sb
->s_flags
&= ~MS_POSIXACL
;
231 printk(KERN_INFO
"btrfs: disabling tree log\n");
232 btrfs_set_opt(info
->mount_opt
, NOTREELOG
);
234 case Opt_flushoncommit
:
235 printk(KERN_INFO
"btrfs: turning on flush-on-commit\n");
236 btrfs_set_opt(info
->mount_opt
, FLUSHONCOMMIT
);
240 match_int(&args
[0], &intarg
);
242 info
->metadata_ratio
= intarg
;
243 printk(KERN_INFO
"btrfs: metadata ratio %d\n",
244 info
->metadata_ratio
);
256 * Parse mount options that are required early in the mount process.
258 * All other options will be parsed on much later in the mount process and
259 * only when we need to allocate a new super block.
261 static int btrfs_parse_early_options(const char *options
, fmode_t flags
,
262 void *holder
, char **subvol_name
,
263 struct btrfs_fs_devices
**fs_devices
)
265 substring_t args
[MAX_OPT_ARGS
];
273 * strsep changes the string, duplicate it because parse_options
276 opts
= kstrdup(options
, GFP_KERNEL
);
280 while ((p
= strsep(&opts
, ",")) != NULL
) {
285 token
= match_token(p
, tokens
, args
);
288 *subvol_name
= match_strdup(&args
[0]);
291 error
= btrfs_scan_one_device(match_strdup(&args
[0]),
292 flags
, holder
, fs_devices
);
305 * If no subvolume name is specified we use the default one. Allocate
306 * a copy of the string "." here so that code later in the
307 * mount path doesn't care if it's the default volume or another one.
310 *subvol_name
= kstrdup(".", GFP_KERNEL
);
317 static int btrfs_fill_super(struct super_block
*sb
,
318 struct btrfs_fs_devices
*fs_devices
,
319 void *data
, int silent
)
322 struct dentry
*root_dentry
;
323 struct btrfs_super_block
*disk_super
;
324 struct btrfs_root
*tree_root
;
325 struct btrfs_inode
*bi
;
328 sb
->s_maxbytes
= MAX_LFS_FILESIZE
;
329 sb
->s_magic
= BTRFS_SUPER_MAGIC
;
330 sb
->s_op
= &btrfs_super_ops
;
331 sb
->s_export_op
= &btrfs_export_ops
;
332 sb
->s_xattr
= btrfs_xattr_handlers
;
334 sb
->s_flags
|= MS_POSIXACL
;
336 tree_root
= open_ctree(sb
, fs_devices
, (char *)data
);
338 if (IS_ERR(tree_root
)) {
339 printk("btrfs: open_ctree failed\n");
340 return PTR_ERR(tree_root
);
342 sb
->s_fs_info
= tree_root
;
343 disk_super
= &tree_root
->fs_info
->super_copy
;
344 inode
= btrfs_iget_locked(sb
, BTRFS_FIRST_FREE_OBJECTID
,
345 tree_root
->fs_info
->fs_root
);
347 bi
->location
.objectid
= inode
->i_ino
;
348 bi
->location
.offset
= 0;
349 bi
->root
= tree_root
->fs_info
->fs_root
;
351 btrfs_set_key_type(&bi
->location
, BTRFS_INODE_ITEM_KEY
);
357 if (inode
->i_state
& I_NEW
) {
358 btrfs_read_locked_inode(inode
);
359 unlock_new_inode(inode
);
362 root_dentry
= d_alloc_root(inode
);
369 /* this does the super kobj at the same time */
370 err
= btrfs_sysfs_add_super(tree_root
->fs_info
);
375 sb
->s_root
= root_dentry
;
377 save_mount_options(sb
, data
);
381 close_ctree(tree_root
);
385 int btrfs_sync_fs(struct super_block
*sb
, int wait
)
387 struct btrfs_trans_handle
*trans
;
388 struct btrfs_root
*root
= btrfs_sb(sb
);
391 if (sb
->s_flags
& MS_RDONLY
)
396 filemap_flush(root
->fs_info
->btree_inode
->i_mapping
);
400 btrfs_start_delalloc_inodes(root
);
401 btrfs_wait_ordered_extents(root
, 0);
403 trans
= btrfs_start_transaction(root
, 1);
404 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
, SSD
))
437 seq_puts(seq
, ",ssd");
438 if (btrfs_test_opt(root
, NOTREELOG
))
439 seq_puts(seq
, ",no-treelog");
440 if (btrfs_test_opt(root
, FLUSHONCOMMIT
))
441 seq_puts(seq
, ",flush-on-commit");
442 if (!(root
->fs_info
->sb
->s_flags
& MS_POSIXACL
))
443 seq_puts(seq
, ",noacl");
447 static void btrfs_write_super(struct super_block
*sb
)
452 static int btrfs_test_super(struct super_block
*s
, void *data
)
454 struct btrfs_fs_devices
*test_fs_devices
= data
;
455 struct btrfs_root
*root
= btrfs_sb(s
);
457 return root
->fs_info
->fs_devices
== test_fs_devices
;
461 * Find a superblock for the given device / mount point.
463 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
464 * for multiple device setup. Make sure to keep it in sync.
466 static int btrfs_get_sb(struct file_system_type
*fs_type
, int flags
,
467 const char *dev_name
, void *data
, struct vfsmount
*mnt
)
469 char *subvol_name
= NULL
;
470 struct block_device
*bdev
= NULL
;
471 struct super_block
*s
;
473 struct btrfs_fs_devices
*fs_devices
= NULL
;
474 fmode_t mode
= FMODE_READ
;
477 if (!(flags
& MS_RDONLY
))
480 error
= btrfs_parse_early_options(data
, mode
, fs_type
,
481 &subvol_name
, &fs_devices
);
485 error
= btrfs_scan_one_device(dev_name
, mode
, fs_type
, &fs_devices
);
487 goto error_free_subvol_name
;
489 error
= btrfs_open_devices(fs_devices
, mode
, fs_type
);
491 goto error_free_subvol_name
;
493 if (!(flags
& MS_RDONLY
) && fs_devices
->rw_devices
== 0) {
495 goto error_close_devices
;
498 bdev
= fs_devices
->latest_bdev
;
499 s
= sget(fs_type
, btrfs_test_super
, set_anon_super
, fs_devices
);
504 if ((flags
^ s
->s_flags
) & MS_RDONLY
) {
505 up_write(&s
->s_umount
);
508 goto error_close_devices
;
511 btrfs_close_devices(fs_devices
);
513 char b
[BDEVNAME_SIZE
];
516 strlcpy(s
->s_id
, bdevname(bdev
, b
), sizeof(s
->s_id
));
517 error
= btrfs_fill_super(s
, fs_devices
, data
,
518 flags
& MS_SILENT
? 1 : 0);
520 up_write(&s
->s_umount
);
522 goto error_free_subvol_name
;
525 btrfs_sb(s
)->fs_info
->bdev_holder
= fs_type
;
526 s
->s_flags
|= MS_ACTIVE
;
529 if (!strcmp(subvol_name
, "."))
530 root
= dget(s
->s_root
);
532 mutex_lock(&s
->s_root
->d_inode
->i_mutex
);
533 root
= lookup_one_len(subvol_name
, s
->s_root
,
534 strlen(subvol_name
));
535 mutex_unlock(&s
->s_root
->d_inode
->i_mutex
);
538 up_write(&s
->s_umount
);
540 error
= PTR_ERR(root
);
541 goto error_free_subvol_name
;
543 if (!root
->d_inode
) {
545 up_write(&s
->s_umount
);
548 goto error_free_subvol_name
;
553 mnt
->mnt_root
= root
;
561 btrfs_close_devices(fs_devices
);
562 error_free_subvol_name
:
567 static int btrfs_remount(struct super_block
*sb
, int *flags
, char *data
)
569 struct btrfs_root
*root
= btrfs_sb(sb
);
572 ret
= btrfs_parse_options(root
, data
);
576 if ((*flags
& MS_RDONLY
) == (sb
->s_flags
& MS_RDONLY
))
579 if (*flags
& MS_RDONLY
) {
580 sb
->s_flags
|= MS_RDONLY
;
582 ret
= btrfs_commit_super(root
);
585 if (root
->fs_info
->fs_devices
->rw_devices
== 0)
588 if (btrfs_super_log_root(&root
->fs_info
->super_copy
) != 0)
591 ret
= btrfs_cleanup_reloc_trees(root
);
594 ret
= btrfs_cleanup_fs_roots(root
->fs_info
);
597 sb
->s_flags
&= ~MS_RDONLY
;
603 static int btrfs_statfs(struct dentry
*dentry
, struct kstatfs
*buf
)
605 struct btrfs_root
*root
= btrfs_sb(dentry
->d_sb
);
606 struct btrfs_super_block
*disk_super
= &root
->fs_info
->super_copy
;
607 int bits
= dentry
->d_sb
->s_blocksize_bits
;
608 __be32
*fsid
= (__be32
*)root
->fs_info
->fsid
;
610 buf
->f_namelen
= BTRFS_NAME_LEN
;
611 buf
->f_blocks
= btrfs_super_total_bytes(disk_super
) >> bits
;
612 buf
->f_bfree
= buf
->f_blocks
-
613 (btrfs_super_bytes_used(disk_super
) >> bits
);
614 buf
->f_bavail
= buf
->f_bfree
;
615 buf
->f_bsize
= dentry
->d_sb
->s_blocksize
;
616 buf
->f_type
= BTRFS_SUPER_MAGIC
;
618 /* We treat it as constant endianness (it doesn't matter _which_)
619 because we want the fsid to come out the same whether mounted
620 on a big-endian or little-endian host */
621 buf
->f_fsid
.val
[0] = be32_to_cpu(fsid
[0]) ^ be32_to_cpu(fsid
[2]);
622 buf
->f_fsid
.val
[1] = be32_to_cpu(fsid
[1]) ^ be32_to_cpu(fsid
[3]);
623 /* Mask in the root object ID too, to disambiguate subvols */
624 buf
->f_fsid
.val
[0] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
>> 32;
625 buf
->f_fsid
.val
[1] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
;
630 static struct file_system_type btrfs_fs_type
= {
631 .owner
= THIS_MODULE
,
633 .get_sb
= btrfs_get_sb
,
634 .kill_sb
= kill_anon_super
,
635 .fs_flags
= FS_REQUIRES_DEV
,
639 * used by btrfsctl to scan devices when no FS is mounted
641 static long btrfs_control_ioctl(struct file
*file
, unsigned int cmd
,
644 struct btrfs_ioctl_vol_args
*vol
;
645 struct btrfs_fs_devices
*fs_devices
;
648 if (!capable(CAP_SYS_ADMIN
))
651 vol
= memdup_user((void __user
*)arg
, sizeof(*vol
));
656 case BTRFS_IOC_SCAN_DEV
:
657 ret
= btrfs_scan_one_device(vol
->name
, FMODE_READ
,
658 &btrfs_fs_type
, &fs_devices
);
666 static int btrfs_freeze(struct super_block
*sb
)
668 struct btrfs_root
*root
= btrfs_sb(sb
);
669 mutex_lock(&root
->fs_info
->transaction_kthread_mutex
);
670 mutex_lock(&root
->fs_info
->cleaner_mutex
);
674 static int btrfs_unfreeze(struct super_block
*sb
)
676 struct btrfs_root
*root
= btrfs_sb(sb
);
677 mutex_unlock(&root
->fs_info
->cleaner_mutex
);
678 mutex_unlock(&root
->fs_info
->transaction_kthread_mutex
);
682 static struct super_operations btrfs_super_ops
= {
683 .delete_inode
= btrfs_delete_inode
,
684 .put_super
= btrfs_put_super
,
685 .write_super
= btrfs_write_super
,
686 .sync_fs
= btrfs_sync_fs
,
687 .show_options
= btrfs_show_options
,
688 .write_inode
= btrfs_write_inode
,
689 .dirty_inode
= btrfs_dirty_inode
,
690 .alloc_inode
= btrfs_alloc_inode
,
691 .destroy_inode
= btrfs_destroy_inode
,
692 .statfs
= btrfs_statfs
,
693 .remount_fs
= btrfs_remount
,
694 .freeze_fs
= btrfs_freeze
,
695 .unfreeze_fs
= btrfs_unfreeze
,
698 static const struct file_operations btrfs_ctl_fops
= {
699 .unlocked_ioctl
= btrfs_control_ioctl
,
700 .compat_ioctl
= btrfs_control_ioctl
,
701 .owner
= THIS_MODULE
,
704 static struct miscdevice btrfs_misc
= {
705 .minor
= MISC_DYNAMIC_MINOR
,
706 .name
= "btrfs-control",
707 .fops
= &btrfs_ctl_fops
710 static int btrfs_interface_init(void)
712 return misc_register(&btrfs_misc
);
715 static void btrfs_interface_exit(void)
717 if (misc_deregister(&btrfs_misc
) < 0)
718 printk(KERN_INFO
"misc_deregister failed for control device");
721 static int __init
init_btrfs_fs(void)
725 err
= btrfs_init_sysfs();
729 err
= btrfs_init_cachep();
733 err
= extent_io_init();
737 err
= extent_map_init();
741 err
= btrfs_interface_init();
743 goto free_extent_map
;
745 err
= register_filesystem(&btrfs_fs_type
);
747 goto unregister_ioctl
;
749 printk(KERN_INFO
"%s loaded\n", BTRFS_BUILD_VERSION
);
753 btrfs_interface_exit();
759 btrfs_destroy_cachep();
765 static void __exit
exit_btrfs_fs(void)
767 btrfs_destroy_cachep();
770 btrfs_interface_exit();
771 unregister_filesystem(&btrfs_fs_type
);
773 btrfs_cleanup_fs_uuids();
777 module_init(init_btrfs_fs
)
778 module_exit(exit_btrfs_fs
)
780 MODULE_LICENSE("GPL");