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/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>
44 #include "transaction.h"
45 #include "btrfs_inode.h"
47 #include "print-tree.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
);
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_err
,
73 static match_table_t tokens
= {
74 {Opt_degraded
, "degraded"},
75 {Opt_subvol
, "subvol=%s"},
76 {Opt_device
, "device=%s"},
77 {Opt_nodatasum
, "nodatasum"},
78 {Opt_nodatacow
, "nodatacow"},
79 {Opt_nobarrier
, "nobarrier"},
80 {Opt_max_extent
, "max_extent=%s"},
81 {Opt_max_inline
, "max_inline=%s"},
82 {Opt_alloc_start
, "alloc_start=%s"},
83 {Opt_thread_pool
, "thread_pool=%d"},
84 {Opt_compress
, "compress"},
90 u64
btrfs_parse_size(char *str
)
97 res
= simple_strtoul(str
, &end
, 10);
101 last
= tolower(last
);
116 * Regular mount options parser. Everything that is needed only when
117 * reading in a new superblock is parsed here.
119 int btrfs_parse_options(struct btrfs_root
*root
, char *options
)
121 struct btrfs_fs_info
*info
= root
->fs_info
;
122 substring_t args
[MAX_OPT_ARGS
];
130 * strsep changes the string, duplicate it because parse_options
133 options
= kstrdup(options
, GFP_NOFS
);
138 while ((p
= strsep(&options
, ",")) != NULL
) {
143 token
= match_token(p
, tokens
, args
);
146 printk(KERN_INFO
"btrfs: allowing degraded mounts\n");
147 btrfs_set_opt(info
->mount_opt
, DEGRADED
);
152 * These are parsed by btrfs_parse_early_options
153 * and can be happily ignored here.
157 printk(KERN_INFO
"btrfs: setting nodatacsum\n");
158 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
161 printk(KERN_INFO
"btrfs: setting nodatacow\n");
162 btrfs_set_opt(info
->mount_opt
, NODATACOW
);
163 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
166 printk(KERN_INFO
"btrfs: use compression\n");
167 btrfs_set_opt(info
->mount_opt
, COMPRESS
);
170 printk(KERN_INFO
"btrfs: use ssd allocation scheme\n");
171 btrfs_set_opt(info
->mount_opt
, SSD
);
174 printk(KERN_INFO
"btrfs: turning off barriers\n");
175 btrfs_set_opt(info
->mount_opt
, NOBARRIER
);
177 case Opt_thread_pool
:
179 match_int(&args
[0], &intarg
);
181 info
->thread_pool_size
= intarg
;
182 printk(KERN_INFO
"btrfs: thread pool %d\n",
183 info
->thread_pool_size
);
187 num
= match_strdup(&args
[0]);
189 info
->max_extent
= btrfs_parse_size(num
);
192 info
->max_extent
= max_t(u64
,
193 info
->max_extent
, root
->sectorsize
);
194 printk(KERN_INFO
"btrfs: max_extent at %llu\n",
199 num
= match_strdup(&args
[0]);
201 info
->max_inline
= btrfs_parse_size(num
);
204 if (info
->max_inline
) {
205 info
->max_inline
= max_t(u64
,
209 printk(KERN_INFO
"btrfs: max_inline at %llu\n",
213 case Opt_alloc_start
:
214 num
= match_strdup(&args
[0]);
216 info
->alloc_start
= btrfs_parse_size(num
);
219 "btrfs: allocations start at %llu\n",
224 root
->fs_info
->sb
->s_flags
&= ~MS_POSIXACL
;
235 * Parse mount options that are required early in the mount process.
237 * All other options will be parsed on much later in the mount process and
238 * only when we need to allocate a new super block.
240 static int btrfs_parse_early_options(const char *options
, fmode_t flags
,
241 void *holder
, char **subvol_name
,
242 struct btrfs_fs_devices
**fs_devices
)
244 substring_t args
[MAX_OPT_ARGS
];
252 * strsep changes the string, duplicate it because parse_options
255 opts
= kstrdup(options
, GFP_KERNEL
);
259 while ((p
= strsep(&opts
, ",")) != NULL
) {
264 token
= match_token(p
, tokens
, args
);
267 *subvol_name
= match_strdup(&args
[0]);
270 error
= btrfs_scan_one_device(match_strdup(&args
[0]),
271 flags
, holder
, fs_devices
);
284 * If no subvolume name is specified we use the default one. Allocate
285 * a copy of the string "." here so that code later in the
286 * mount path doesn't care if it's the default volume or another one.
289 *subvol_name
= kstrdup(".", GFP_KERNEL
);
296 static int btrfs_fill_super(struct super_block
*sb
,
297 struct btrfs_fs_devices
*fs_devices
,
298 void *data
, int silent
)
301 struct dentry
*root_dentry
;
302 struct btrfs_super_block
*disk_super
;
303 struct btrfs_root
*tree_root
;
304 struct btrfs_inode
*bi
;
307 sb
->s_maxbytes
= MAX_LFS_FILESIZE
;
308 sb
->s_magic
= BTRFS_SUPER_MAGIC
;
309 sb
->s_op
= &btrfs_super_ops
;
310 sb
->s_export_op
= &btrfs_export_ops
;
311 sb
->s_xattr
= btrfs_xattr_handlers
;
313 sb
->s_flags
|= MS_POSIXACL
;
315 tree_root
= open_ctree(sb
, fs_devices
, (char *)data
);
317 if (IS_ERR(tree_root
)) {
318 printk("btrfs: open_ctree failed\n");
319 return PTR_ERR(tree_root
);
321 sb
->s_fs_info
= tree_root
;
322 disk_super
= &tree_root
->fs_info
->super_copy
;
323 inode
= btrfs_iget_locked(sb
, BTRFS_FIRST_FREE_OBJECTID
,
324 tree_root
->fs_info
->fs_root
);
326 bi
->location
.objectid
= inode
->i_ino
;
327 bi
->location
.offset
= 0;
328 bi
->root
= tree_root
->fs_info
->fs_root
;
330 btrfs_set_key_type(&bi
->location
, BTRFS_INODE_ITEM_KEY
);
336 if (inode
->i_state
& I_NEW
) {
337 btrfs_read_locked_inode(inode
);
338 unlock_new_inode(inode
);
341 root_dentry
= d_alloc_root(inode
);
348 /* this does the super kobj at the same time */
349 err
= btrfs_sysfs_add_super(tree_root
->fs_info
);
354 sb
->s_root
= root_dentry
;
356 save_mount_options(sb
, data
);
360 close_ctree(tree_root
);
364 int btrfs_sync_fs(struct super_block
*sb
, int wait
)
366 struct btrfs_trans_handle
*trans
;
367 struct btrfs_root
*root
;
371 if (sb
->s_flags
& MS_RDONLY
)
376 filemap_flush(root
->fs_info
->btree_inode
->i_mapping
);
380 btrfs_start_delalloc_inodes(root
);
381 btrfs_wait_ordered_extents(root
, 0);
383 btrfs_clean_old_snapshots(root
);
384 trans
= btrfs_start_transaction(root
, 1);
385 ret
= btrfs_commit_transaction(trans
, root
);
390 static void btrfs_write_super(struct super_block
*sb
)
395 static int btrfs_test_super(struct super_block
*s
, void *data
)
397 struct btrfs_fs_devices
*test_fs_devices
= data
;
398 struct btrfs_root
*root
= btrfs_sb(s
);
400 return root
->fs_info
->fs_devices
== test_fs_devices
;
404 * Find a superblock for the given device / mount point.
406 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
407 * for multiple device setup. Make sure to keep it in sync.
409 static int btrfs_get_sb(struct file_system_type
*fs_type
, int flags
,
410 const char *dev_name
, void *data
, struct vfsmount
*mnt
)
412 char *subvol_name
= NULL
;
413 struct block_device
*bdev
= NULL
;
414 struct super_block
*s
;
416 struct btrfs_fs_devices
*fs_devices
= NULL
;
417 fmode_t mode
= FMODE_READ
;
420 if (!(flags
& MS_RDONLY
))
423 error
= btrfs_parse_early_options(data
, mode
, fs_type
,
424 &subvol_name
, &fs_devices
);
428 error
= btrfs_scan_one_device(dev_name
, mode
, fs_type
, &fs_devices
);
430 goto error_free_subvol_name
;
432 error
= btrfs_open_devices(fs_devices
, mode
, fs_type
);
434 goto error_free_subvol_name
;
436 if (!(flags
& MS_RDONLY
) && fs_devices
->rw_devices
== 0) {
438 goto error_close_devices
;
441 bdev
= fs_devices
->latest_bdev
;
442 s
= sget(fs_type
, btrfs_test_super
, set_anon_super
, fs_devices
);
447 if ((flags
^ s
->s_flags
) & MS_RDONLY
) {
448 up_write(&s
->s_umount
);
451 goto error_close_devices
;
454 btrfs_close_devices(fs_devices
);
456 char b
[BDEVNAME_SIZE
];
459 strlcpy(s
->s_id
, bdevname(bdev
, b
), sizeof(s
->s_id
));
460 error
= btrfs_fill_super(s
, fs_devices
, data
,
461 flags
& MS_SILENT
? 1 : 0);
463 up_write(&s
->s_umount
);
465 goto error_free_subvol_name
;
468 btrfs_sb(s
)->fs_info
->bdev_holder
= fs_type
;
469 s
->s_flags
|= MS_ACTIVE
;
472 if (!strcmp(subvol_name
, "."))
473 root
= dget(s
->s_root
);
475 mutex_lock(&s
->s_root
->d_inode
->i_mutex
);
476 root
= lookup_one_len(subvol_name
, s
->s_root
,
477 strlen(subvol_name
));
478 mutex_unlock(&s
->s_root
->d_inode
->i_mutex
);
481 up_write(&s
->s_umount
);
483 error
= PTR_ERR(root
);
484 goto error_free_subvol_name
;
486 if (!root
->d_inode
) {
488 up_write(&s
->s_umount
);
491 goto error_free_subvol_name
;
496 mnt
->mnt_root
= root
;
504 btrfs_close_devices(fs_devices
);
505 error_free_subvol_name
:
510 static int btrfs_remount(struct super_block
*sb
, int *flags
, char *data
)
512 struct btrfs_root
*root
= btrfs_sb(sb
);
515 if ((*flags
& MS_RDONLY
) == (sb
->s_flags
& MS_RDONLY
))
518 if (*flags
& MS_RDONLY
) {
519 sb
->s_flags
|= MS_RDONLY
;
521 ret
= btrfs_commit_super(root
);
524 if (root
->fs_info
->fs_devices
->rw_devices
== 0)
527 if (btrfs_super_log_root(&root
->fs_info
->super_copy
) != 0)
530 ret
= btrfs_cleanup_reloc_trees(root
);
533 ret
= btrfs_cleanup_fs_roots(root
->fs_info
);
536 sb
->s_flags
&= ~MS_RDONLY
;
542 static int btrfs_statfs(struct dentry
*dentry
, struct kstatfs
*buf
)
544 struct btrfs_root
*root
= btrfs_sb(dentry
->d_sb
);
545 struct btrfs_super_block
*disk_super
= &root
->fs_info
->super_copy
;
546 int bits
= dentry
->d_sb
->s_blocksize_bits
;
547 __be32
*fsid
= (__be32
*)root
->fs_info
->fsid
;
549 buf
->f_namelen
= BTRFS_NAME_LEN
;
550 buf
->f_blocks
= btrfs_super_total_bytes(disk_super
) >> bits
;
551 buf
->f_bfree
= buf
->f_blocks
-
552 (btrfs_super_bytes_used(disk_super
) >> bits
);
553 buf
->f_bavail
= buf
->f_bfree
;
554 buf
->f_bsize
= dentry
->d_sb
->s_blocksize
;
555 buf
->f_type
= BTRFS_SUPER_MAGIC
;
557 /* We treat it as constant endianness (it doesn't matter _which_)
558 because we want the fsid to come out the same whether mounted
559 on a big-endian or little-endian host */
560 buf
->f_fsid
.val
[0] = be32_to_cpu(fsid
[0]) ^ be32_to_cpu(fsid
[2]);
561 buf
->f_fsid
.val
[1] = be32_to_cpu(fsid
[1]) ^ be32_to_cpu(fsid
[3]);
562 /* Mask in the root object ID too, to disambiguate subvols */
563 buf
->f_fsid
.val
[0] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
>> 32;
564 buf
->f_fsid
.val
[1] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
;
569 static struct file_system_type btrfs_fs_type
= {
570 .owner
= THIS_MODULE
,
572 .get_sb
= btrfs_get_sb
,
573 .kill_sb
= kill_anon_super
,
574 .fs_flags
= FS_REQUIRES_DEV
,
578 * used by btrfsctl to scan devices when no FS is mounted
580 static long btrfs_control_ioctl(struct file
*file
, unsigned int cmd
,
583 struct btrfs_ioctl_vol_args
*vol
;
584 struct btrfs_fs_devices
*fs_devices
;
588 if (!capable(CAP_SYS_ADMIN
))
591 vol
= kmalloc(sizeof(*vol
), GFP_KERNEL
);
592 if (copy_from_user(vol
, (void __user
*)arg
, sizeof(*vol
))) {
596 len
= strnlen(vol
->name
, BTRFS_PATH_NAME_MAX
);
598 case BTRFS_IOC_SCAN_DEV
:
599 ret
= btrfs_scan_one_device(vol
->name
, FMODE_READ
,
600 &btrfs_fs_type
, &fs_devices
);
608 static int btrfs_freeze(struct super_block
*sb
)
610 struct btrfs_root
*root
= btrfs_sb(sb
);
611 mutex_lock(&root
->fs_info
->transaction_kthread_mutex
);
612 mutex_lock(&root
->fs_info
->cleaner_mutex
);
616 static int btrfs_unfreeze(struct super_block
*sb
)
618 struct btrfs_root
*root
= btrfs_sb(sb
);
619 mutex_unlock(&root
->fs_info
->cleaner_mutex
);
620 mutex_unlock(&root
->fs_info
->transaction_kthread_mutex
);
624 static struct super_operations btrfs_super_ops
= {
625 .delete_inode
= btrfs_delete_inode
,
626 .put_super
= btrfs_put_super
,
627 .write_super
= btrfs_write_super
,
628 .sync_fs
= btrfs_sync_fs
,
629 .show_options
= generic_show_options
,
630 .write_inode
= btrfs_write_inode
,
631 .dirty_inode
= btrfs_dirty_inode
,
632 .alloc_inode
= btrfs_alloc_inode
,
633 .destroy_inode
= btrfs_destroy_inode
,
634 .statfs
= btrfs_statfs
,
635 .remount_fs
= btrfs_remount
,
636 .freeze_fs
= btrfs_freeze
,
637 .unfreeze_fs
= btrfs_unfreeze
,
640 static const struct file_operations btrfs_ctl_fops
= {
641 .unlocked_ioctl
= btrfs_control_ioctl
,
642 .compat_ioctl
= btrfs_control_ioctl
,
643 .owner
= THIS_MODULE
,
646 static struct miscdevice btrfs_misc
= {
647 .minor
= MISC_DYNAMIC_MINOR
,
648 .name
= "btrfs-control",
649 .fops
= &btrfs_ctl_fops
652 static int btrfs_interface_init(void)
654 return misc_register(&btrfs_misc
);
657 static void btrfs_interface_exit(void)
659 if (misc_deregister(&btrfs_misc
) < 0)
660 printk(KERN_INFO
"misc_deregister failed for control device");
663 static int __init
init_btrfs_fs(void)
667 err
= btrfs_init_sysfs();
671 err
= btrfs_init_cachep();
675 err
= extent_io_init();
679 err
= extent_map_init();
683 err
= btrfs_interface_init();
685 goto free_extent_map
;
687 err
= register_filesystem(&btrfs_fs_type
);
689 goto unregister_ioctl
;
691 printk(KERN_INFO
"%s loaded\n", BTRFS_BUILD_VERSION
);
695 btrfs_interface_exit();
701 btrfs_destroy_cachep();
707 static void __exit
exit_btrfs_fs(void)
709 btrfs_destroy_cachep();
712 btrfs_interface_exit();
713 unregister_filesystem(&btrfs_fs_type
);
715 btrfs_cleanup_fs_uuids();
719 module_init(init_btrfs_fs
)
720 module_exit(exit_btrfs_fs
)
722 MODULE_LICENSE("GPL");