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_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"},
93 u64
btrfs_parse_size(char *str
)
100 res
= simple_strtoul(str
, &end
, 10);
104 last
= tolower(last
);
119 * Regular mount options parser. Everything that is needed only when
120 * reading in a new superblock is parsed here.
122 int btrfs_parse_options(struct btrfs_root
*root
, char *options
)
124 struct btrfs_fs_info
*info
= root
->fs_info
;
125 substring_t args
[MAX_OPT_ARGS
];
133 * strsep changes the string, duplicate it because parse_options
136 options
= kstrdup(options
, GFP_NOFS
);
141 while ((p
= strsep(&options
, ",")) != NULL
) {
146 token
= match_token(p
, tokens
, args
);
149 printk(KERN_INFO
"btrfs: allowing degraded mounts\n");
150 btrfs_set_opt(info
->mount_opt
, DEGRADED
);
155 * These are parsed by btrfs_parse_early_options
156 * and can be happily ignored here.
160 printk(KERN_INFO
"btrfs: setting nodatacsum\n");
161 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
164 printk(KERN_INFO
"btrfs: setting nodatacow\n");
165 btrfs_set_opt(info
->mount_opt
, NODATACOW
);
166 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
169 printk(KERN_INFO
"btrfs: use compression\n");
170 btrfs_set_opt(info
->mount_opt
, COMPRESS
);
173 printk(KERN_INFO
"btrfs: use ssd allocation scheme\n");
174 btrfs_set_opt(info
->mount_opt
, SSD
);
177 printk(KERN_INFO
"btrfs: turning off barriers\n");
178 btrfs_set_opt(info
->mount_opt
, NOBARRIER
);
180 case Opt_thread_pool
:
182 match_int(&args
[0], &intarg
);
184 info
->thread_pool_size
= intarg
;
185 printk(KERN_INFO
"btrfs: thread pool %d\n",
186 info
->thread_pool_size
);
190 num
= match_strdup(&args
[0]);
192 info
->max_extent
= btrfs_parse_size(num
);
195 info
->max_extent
= max_t(u64
,
196 info
->max_extent
, root
->sectorsize
);
197 printk(KERN_INFO
"btrfs: max_extent at %llu\n",
202 num
= match_strdup(&args
[0]);
204 info
->max_inline
= btrfs_parse_size(num
);
207 if (info
->max_inline
) {
208 info
->max_inline
= max_t(u64
,
212 printk(KERN_INFO
"btrfs: max_inline at %llu\n",
216 case Opt_alloc_start
:
217 num
= match_strdup(&args
[0]);
219 info
->alloc_start
= btrfs_parse_size(num
);
222 "btrfs: allocations start at %llu\n",
227 root
->fs_info
->sb
->s_flags
&= ~MS_POSIXACL
;
230 printk(KERN_INFO
"btrfs: disabling tree log\n");
231 btrfs_set_opt(info
->mount_opt
, NOTREELOG
);
233 case Opt_flushoncommit
:
234 printk(KERN_INFO
"btrfs: turning on flush-on-commit\n");
235 btrfs_set_opt(info
->mount_opt
, FLUSHONCOMMIT
);
246 * Parse mount options that are required early in the mount process.
248 * All other options will be parsed on much later in the mount process and
249 * only when we need to allocate a new super block.
251 static int btrfs_parse_early_options(const char *options
, fmode_t flags
,
252 void *holder
, char **subvol_name
,
253 struct btrfs_fs_devices
**fs_devices
)
255 substring_t args
[MAX_OPT_ARGS
];
263 * strsep changes the string, duplicate it because parse_options
266 opts
= kstrdup(options
, GFP_KERNEL
);
270 while ((p
= strsep(&opts
, ",")) != NULL
) {
275 token
= match_token(p
, tokens
, args
);
278 *subvol_name
= match_strdup(&args
[0]);
281 error
= btrfs_scan_one_device(match_strdup(&args
[0]),
282 flags
, holder
, fs_devices
);
295 * If no subvolume name is specified we use the default one. Allocate
296 * a copy of the string "." here so that code later in the
297 * mount path doesn't care if it's the default volume or another one.
300 *subvol_name
= kstrdup(".", GFP_KERNEL
);
307 static int btrfs_fill_super(struct super_block
*sb
,
308 struct btrfs_fs_devices
*fs_devices
,
309 void *data
, int silent
)
312 struct dentry
*root_dentry
;
313 struct btrfs_super_block
*disk_super
;
314 struct btrfs_root
*tree_root
;
315 struct btrfs_inode
*bi
;
318 sb
->s_maxbytes
= MAX_LFS_FILESIZE
;
319 sb
->s_magic
= BTRFS_SUPER_MAGIC
;
320 sb
->s_op
= &btrfs_super_ops
;
321 sb
->s_export_op
= &btrfs_export_ops
;
322 sb
->s_xattr
= btrfs_xattr_handlers
;
324 sb
->s_flags
|= MS_POSIXACL
;
326 tree_root
= open_ctree(sb
, fs_devices
, (char *)data
);
328 if (IS_ERR(tree_root
)) {
329 printk("btrfs: open_ctree failed\n");
330 return PTR_ERR(tree_root
);
332 sb
->s_fs_info
= tree_root
;
333 disk_super
= &tree_root
->fs_info
->super_copy
;
334 inode
= btrfs_iget_locked(sb
, BTRFS_FIRST_FREE_OBJECTID
,
335 tree_root
->fs_info
->fs_root
);
337 bi
->location
.objectid
= inode
->i_ino
;
338 bi
->location
.offset
= 0;
339 bi
->root
= tree_root
->fs_info
->fs_root
;
341 btrfs_set_key_type(&bi
->location
, BTRFS_INODE_ITEM_KEY
);
347 if (inode
->i_state
& I_NEW
) {
348 btrfs_read_locked_inode(inode
);
349 unlock_new_inode(inode
);
352 root_dentry
= d_alloc_root(inode
);
359 /* this does the super kobj at the same time */
360 err
= btrfs_sysfs_add_super(tree_root
->fs_info
);
365 sb
->s_root
= root_dentry
;
367 save_mount_options(sb
, data
);
371 close_ctree(tree_root
);
375 int btrfs_sync_fs(struct super_block
*sb
, int wait
)
377 struct btrfs_trans_handle
*trans
;
378 struct btrfs_root
*root
= btrfs_sb(sb
);
381 if (sb
->s_flags
& MS_RDONLY
)
386 filemap_flush(root
->fs_info
->btree_inode
->i_mapping
);
390 btrfs_start_delalloc_inodes(root
);
391 btrfs_wait_ordered_extents(root
, 0);
393 trans
= btrfs_start_transaction(root
, 1);
394 ret
= btrfs_commit_transaction(trans
, root
);
399 static int btrfs_show_options(struct seq_file
*seq
, struct vfsmount
*vfs
)
401 struct btrfs_root
*root
= btrfs_sb(vfs
->mnt_sb
);
402 struct btrfs_fs_info
*info
= root
->fs_info
;
404 if (btrfs_test_opt(root
, DEGRADED
))
405 seq_puts(seq
, ",degraded");
406 if (btrfs_test_opt(root
, NODATASUM
))
407 seq_puts(seq
, ",nodatasum");
408 if (btrfs_test_opt(root
, NODATACOW
))
409 seq_puts(seq
, ",nodatacow");
410 if (btrfs_test_opt(root
, NOBARRIER
))
411 seq_puts(seq
, ",nobarrier");
412 if (info
->max_extent
!= (u64
)-1)
413 seq_printf(seq
, ",max_extent=%llu", info
->max_extent
);
414 if (info
->max_inline
!= 8192 * 1024)
415 seq_printf(seq
, ",max_inline=%llu", info
->max_inline
);
416 if (info
->alloc_start
!= 0)
417 seq_printf(seq
, ",alloc_start=%llu", info
->alloc_start
);
418 if (info
->thread_pool_size
!= min_t(unsigned long,
419 num_online_cpus() + 2, 8))
420 seq_printf(seq
, ",thread_pool=%d", info
->thread_pool_size
);
421 if (btrfs_test_opt(root
, COMPRESS
))
422 seq_puts(seq
, ",compress");
423 if (btrfs_test_opt(root
, SSD
))
424 seq_puts(seq
, ",ssd");
425 if (btrfs_test_opt(root
, NOTREELOG
))
426 seq_puts(seq
, ",no-treelog");
427 if (btrfs_test_opt(root
, FLUSHONCOMMIT
))
428 seq_puts(seq
, ",flush-on-commit");
429 if (!(root
->fs_info
->sb
->s_flags
& MS_POSIXACL
))
430 seq_puts(seq
, ",noacl");
434 static void btrfs_write_super(struct super_block
*sb
)
439 static int btrfs_test_super(struct super_block
*s
, void *data
)
441 struct btrfs_fs_devices
*test_fs_devices
= data
;
442 struct btrfs_root
*root
= btrfs_sb(s
);
444 return root
->fs_info
->fs_devices
== test_fs_devices
;
448 * Find a superblock for the given device / mount point.
450 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
451 * for multiple device setup. Make sure to keep it in sync.
453 static int btrfs_get_sb(struct file_system_type
*fs_type
, int flags
,
454 const char *dev_name
, void *data
, struct vfsmount
*mnt
)
456 char *subvol_name
= NULL
;
457 struct block_device
*bdev
= NULL
;
458 struct super_block
*s
;
460 struct btrfs_fs_devices
*fs_devices
= NULL
;
461 fmode_t mode
= FMODE_READ
;
464 if (!(flags
& MS_RDONLY
))
467 error
= btrfs_parse_early_options(data
, mode
, fs_type
,
468 &subvol_name
, &fs_devices
);
472 error
= btrfs_scan_one_device(dev_name
, mode
, fs_type
, &fs_devices
);
474 goto error_free_subvol_name
;
476 error
= btrfs_open_devices(fs_devices
, mode
, fs_type
);
478 goto error_free_subvol_name
;
480 if (!(flags
& MS_RDONLY
) && fs_devices
->rw_devices
== 0) {
482 goto error_close_devices
;
485 bdev
= fs_devices
->latest_bdev
;
486 s
= sget(fs_type
, btrfs_test_super
, set_anon_super
, fs_devices
);
491 if ((flags
^ s
->s_flags
) & MS_RDONLY
) {
492 up_write(&s
->s_umount
);
495 goto error_close_devices
;
498 btrfs_close_devices(fs_devices
);
500 char b
[BDEVNAME_SIZE
];
503 strlcpy(s
->s_id
, bdevname(bdev
, b
), sizeof(s
->s_id
));
504 error
= btrfs_fill_super(s
, fs_devices
, data
,
505 flags
& MS_SILENT
? 1 : 0);
507 up_write(&s
->s_umount
);
509 goto error_free_subvol_name
;
512 btrfs_sb(s
)->fs_info
->bdev_holder
= fs_type
;
513 s
->s_flags
|= MS_ACTIVE
;
516 if (!strcmp(subvol_name
, "."))
517 root
= dget(s
->s_root
);
519 mutex_lock(&s
->s_root
->d_inode
->i_mutex
);
520 root
= lookup_one_len(subvol_name
, s
->s_root
,
521 strlen(subvol_name
));
522 mutex_unlock(&s
->s_root
->d_inode
->i_mutex
);
525 up_write(&s
->s_umount
);
527 error
= PTR_ERR(root
);
528 goto error_free_subvol_name
;
530 if (!root
->d_inode
) {
532 up_write(&s
->s_umount
);
535 goto error_free_subvol_name
;
540 mnt
->mnt_root
= root
;
548 btrfs_close_devices(fs_devices
);
549 error_free_subvol_name
:
554 static int btrfs_remount(struct super_block
*sb
, int *flags
, char *data
)
556 struct btrfs_root
*root
= btrfs_sb(sb
);
559 ret
= btrfs_parse_options(root
, data
);
563 if ((*flags
& MS_RDONLY
) == (sb
->s_flags
& MS_RDONLY
))
566 if (*flags
& MS_RDONLY
) {
567 sb
->s_flags
|= MS_RDONLY
;
569 ret
= btrfs_commit_super(root
);
572 if (root
->fs_info
->fs_devices
->rw_devices
== 0)
575 if (btrfs_super_log_root(&root
->fs_info
->super_copy
) != 0)
578 ret
= btrfs_cleanup_reloc_trees(root
);
581 ret
= btrfs_cleanup_fs_roots(root
->fs_info
);
584 sb
->s_flags
&= ~MS_RDONLY
;
590 static int btrfs_statfs(struct dentry
*dentry
, struct kstatfs
*buf
)
592 struct btrfs_root
*root
= btrfs_sb(dentry
->d_sb
);
593 struct btrfs_super_block
*disk_super
= &root
->fs_info
->super_copy
;
594 int bits
= dentry
->d_sb
->s_blocksize_bits
;
595 __be32
*fsid
= (__be32
*)root
->fs_info
->fsid
;
597 buf
->f_namelen
= BTRFS_NAME_LEN
;
598 buf
->f_blocks
= btrfs_super_total_bytes(disk_super
) >> bits
;
599 buf
->f_bfree
= buf
->f_blocks
-
600 (btrfs_super_bytes_used(disk_super
) >> bits
);
601 buf
->f_bavail
= buf
->f_bfree
;
602 buf
->f_bsize
= dentry
->d_sb
->s_blocksize
;
603 buf
->f_type
= BTRFS_SUPER_MAGIC
;
605 /* We treat it as constant endianness (it doesn't matter _which_)
606 because we want the fsid to come out the same whether mounted
607 on a big-endian or little-endian host */
608 buf
->f_fsid
.val
[0] = be32_to_cpu(fsid
[0]) ^ be32_to_cpu(fsid
[2]);
609 buf
->f_fsid
.val
[1] = be32_to_cpu(fsid
[1]) ^ be32_to_cpu(fsid
[3]);
610 /* Mask in the root object ID too, to disambiguate subvols */
611 buf
->f_fsid
.val
[0] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
>> 32;
612 buf
->f_fsid
.val
[1] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
;
617 static struct file_system_type btrfs_fs_type
= {
618 .owner
= THIS_MODULE
,
620 .get_sb
= btrfs_get_sb
,
621 .kill_sb
= kill_anon_super
,
622 .fs_flags
= FS_REQUIRES_DEV
,
626 * used by btrfsctl to scan devices when no FS is mounted
628 static long btrfs_control_ioctl(struct file
*file
, unsigned int cmd
,
631 struct btrfs_ioctl_vol_args
*vol
;
632 struct btrfs_fs_devices
*fs_devices
;
635 if (!capable(CAP_SYS_ADMIN
))
638 vol
= kmalloc(sizeof(*vol
), GFP_KERNEL
);
642 if (copy_from_user(vol
, (void __user
*)arg
, sizeof(*vol
))) {
648 case BTRFS_IOC_SCAN_DEV
:
649 ret
= btrfs_scan_one_device(vol
->name
, FMODE_READ
,
650 &btrfs_fs_type
, &fs_devices
);
658 static int btrfs_freeze(struct super_block
*sb
)
660 struct btrfs_root
*root
= btrfs_sb(sb
);
661 mutex_lock(&root
->fs_info
->transaction_kthread_mutex
);
662 mutex_lock(&root
->fs_info
->cleaner_mutex
);
666 static int btrfs_unfreeze(struct super_block
*sb
)
668 struct btrfs_root
*root
= btrfs_sb(sb
);
669 mutex_unlock(&root
->fs_info
->cleaner_mutex
);
670 mutex_unlock(&root
->fs_info
->transaction_kthread_mutex
);
674 static struct super_operations btrfs_super_ops
= {
675 .delete_inode
= btrfs_delete_inode
,
676 .put_super
= btrfs_put_super
,
677 .write_super
= btrfs_write_super
,
678 .sync_fs
= btrfs_sync_fs
,
679 .show_options
= btrfs_show_options
,
680 .write_inode
= btrfs_write_inode
,
681 .dirty_inode
= btrfs_dirty_inode
,
682 .alloc_inode
= btrfs_alloc_inode
,
683 .destroy_inode
= btrfs_destroy_inode
,
684 .statfs
= btrfs_statfs
,
685 .remount_fs
= btrfs_remount
,
686 .freeze_fs
= btrfs_freeze
,
687 .unfreeze_fs
= btrfs_unfreeze
,
690 static const struct file_operations btrfs_ctl_fops
= {
691 .unlocked_ioctl
= btrfs_control_ioctl
,
692 .compat_ioctl
= btrfs_control_ioctl
,
693 .owner
= THIS_MODULE
,
696 static struct miscdevice btrfs_misc
= {
697 .minor
= MISC_DYNAMIC_MINOR
,
698 .name
= "btrfs-control",
699 .fops
= &btrfs_ctl_fops
702 static int btrfs_interface_init(void)
704 return misc_register(&btrfs_misc
);
707 static void btrfs_interface_exit(void)
709 if (misc_deregister(&btrfs_misc
) < 0)
710 printk(KERN_INFO
"misc_deregister failed for control device");
713 static int __init
init_btrfs_fs(void)
717 err
= btrfs_init_sysfs();
721 err
= btrfs_init_cachep();
725 err
= extent_io_init();
729 err
= extent_map_init();
733 err
= btrfs_interface_init();
735 goto free_extent_map
;
737 err
= register_filesystem(&btrfs_fs_type
);
739 goto unregister_ioctl
;
741 printk(KERN_INFO
"%s loaded\n", BTRFS_BUILD_VERSION
);
745 btrfs_interface_exit();
751 btrfs_destroy_cachep();
757 static void __exit
exit_btrfs_fs(void)
759 btrfs_destroy_cachep();
762 btrfs_interface_exit();
763 unregister_filesystem(&btrfs_fs_type
);
765 btrfs_cleanup_fs_uuids();
769 module_init(init_btrfs_fs
)
770 module_exit(exit_btrfs_fs
)
772 MODULE_LICENSE("GPL");