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/kernel.h>
20 #include <linux/bio.h>
21 #include <linux/buffer_head.h>
22 #include <linux/file.h>
24 #include <linux/fsnotify.h>
25 #include <linux/pagemap.h>
26 #include <linux/highmem.h>
27 #include <linux/time.h>
28 #include <linux/init.h>
29 #include <linux/string.h>
30 #include <linux/backing-dev.h>
31 #include <linux/mount.h>
32 #include <linux/mpage.h>
33 #include <linux/namei.h>
34 #include <linux/swap.h>
35 #include <linux/writeback.h>
36 #include <linux/statfs.h>
37 #include <linux/compat.h>
38 #include <linux/bit_spinlock.h>
39 #include <linux/security.h>
40 #include <linux/xattr.h>
41 #include <linux/vmalloc.h>
42 #include <linux/slab.h>
43 #include <linux/blkdev.h>
44 #include <linux/uuid.h>
45 #include <linux/btrfs.h>
46 #include <linux/uaccess.h>
49 #include "transaction.h"
50 #include "btrfs_inode.h"
51 #include "print-tree.h"
54 #include "inode-map.h"
56 #include "rcu-string.h"
58 #include "dev-replace.h"
64 /* If we have a 32-bit userspace and 64-bit kernel, then the UAPI
65 * structures are incorrect, as the timespec structure from userspace
66 * is 4 bytes too small. We define these alternatives here to teach
67 * the kernel about the 32-bit struct packing.
69 struct btrfs_ioctl_timespec_32
{
72 } __attribute__ ((__packed__
));
74 struct btrfs_ioctl_received_subvol_args_32
{
75 char uuid
[BTRFS_UUID_SIZE
]; /* in */
76 __u64 stransid
; /* in */
77 __u64 rtransid
; /* out */
78 struct btrfs_ioctl_timespec_32 stime
; /* in */
79 struct btrfs_ioctl_timespec_32 rtime
; /* out */
81 __u64 reserved
[16]; /* in */
82 } __attribute__ ((__packed__
));
84 #define BTRFS_IOC_SET_RECEIVED_SUBVOL_32 _IOWR(BTRFS_IOCTL_MAGIC, 37, \
85 struct btrfs_ioctl_received_subvol_args_32)
89 static int btrfs_clone(struct inode
*src
, struct inode
*inode
,
90 u64 off
, u64 olen
, u64 olen_aligned
, u64 destoff
);
92 /* Mask out flags that are inappropriate for the given type of inode. */
93 static inline __u32
btrfs_mask_flags(umode_t mode
, __u32 flags
)
97 else if (S_ISREG(mode
))
98 return flags
& ~FS_DIRSYNC_FL
;
100 return flags
& (FS_NODUMP_FL
| FS_NOATIME_FL
);
104 * Export inode flags to the format expected by the FS_IOC_GETFLAGS ioctl.
106 static unsigned int btrfs_flags_to_ioctl(unsigned int flags
)
108 unsigned int iflags
= 0;
110 if (flags
& BTRFS_INODE_SYNC
)
111 iflags
|= FS_SYNC_FL
;
112 if (flags
& BTRFS_INODE_IMMUTABLE
)
113 iflags
|= FS_IMMUTABLE_FL
;
114 if (flags
& BTRFS_INODE_APPEND
)
115 iflags
|= FS_APPEND_FL
;
116 if (flags
& BTRFS_INODE_NODUMP
)
117 iflags
|= FS_NODUMP_FL
;
118 if (flags
& BTRFS_INODE_NOATIME
)
119 iflags
|= FS_NOATIME_FL
;
120 if (flags
& BTRFS_INODE_DIRSYNC
)
121 iflags
|= FS_DIRSYNC_FL
;
122 if (flags
& BTRFS_INODE_NODATACOW
)
123 iflags
|= FS_NOCOW_FL
;
125 if ((flags
& BTRFS_INODE_COMPRESS
) && !(flags
& BTRFS_INODE_NOCOMPRESS
))
126 iflags
|= FS_COMPR_FL
;
127 else if (flags
& BTRFS_INODE_NOCOMPRESS
)
128 iflags
|= FS_NOCOMP_FL
;
134 * Update inode->i_flags based on the btrfs internal flags.
136 void btrfs_update_iflags(struct inode
*inode
)
138 struct btrfs_inode
*ip
= BTRFS_I(inode
);
139 unsigned int new_fl
= 0;
141 if (ip
->flags
& BTRFS_INODE_SYNC
)
143 if (ip
->flags
& BTRFS_INODE_IMMUTABLE
)
144 new_fl
|= S_IMMUTABLE
;
145 if (ip
->flags
& BTRFS_INODE_APPEND
)
147 if (ip
->flags
& BTRFS_INODE_NOATIME
)
149 if (ip
->flags
& BTRFS_INODE_DIRSYNC
)
152 set_mask_bits(&inode
->i_flags
,
153 S_SYNC
| S_APPEND
| S_IMMUTABLE
| S_NOATIME
| S_DIRSYNC
,
158 * Inherit flags from the parent inode.
160 * Currently only the compression flags and the cow flags are inherited.
162 void btrfs_inherit_iflags(struct inode
*inode
, struct inode
*dir
)
169 flags
= BTRFS_I(dir
)->flags
;
171 if (flags
& BTRFS_INODE_NOCOMPRESS
) {
172 BTRFS_I(inode
)->flags
&= ~BTRFS_INODE_COMPRESS
;
173 BTRFS_I(inode
)->flags
|= BTRFS_INODE_NOCOMPRESS
;
174 } else if (flags
& BTRFS_INODE_COMPRESS
) {
175 BTRFS_I(inode
)->flags
&= ~BTRFS_INODE_NOCOMPRESS
;
176 BTRFS_I(inode
)->flags
|= BTRFS_INODE_COMPRESS
;
179 if (flags
& BTRFS_INODE_NODATACOW
) {
180 BTRFS_I(inode
)->flags
|= BTRFS_INODE_NODATACOW
;
181 if (S_ISREG(inode
->i_mode
))
182 BTRFS_I(inode
)->flags
|= BTRFS_INODE_NODATASUM
;
185 btrfs_update_iflags(inode
);
188 static int btrfs_ioctl_getflags(struct file
*file
, void __user
*arg
)
190 struct btrfs_inode
*ip
= BTRFS_I(file_inode(file
));
191 unsigned int flags
= btrfs_flags_to_ioctl(ip
->flags
);
193 if (copy_to_user(arg
, &flags
, sizeof(flags
)))
198 static int check_flags(unsigned int flags
)
200 if (flags
& ~(FS_IMMUTABLE_FL
| FS_APPEND_FL
| \
201 FS_NOATIME_FL
| FS_NODUMP_FL
| \
202 FS_SYNC_FL
| FS_DIRSYNC_FL
| \
203 FS_NOCOMP_FL
| FS_COMPR_FL
|
207 if ((flags
& FS_NOCOMP_FL
) && (flags
& FS_COMPR_FL
))
213 static int btrfs_ioctl_setflags(struct file
*file
, void __user
*arg
)
215 struct inode
*inode
= file_inode(file
);
216 struct btrfs_inode
*ip
= BTRFS_I(inode
);
217 struct btrfs_root
*root
= ip
->root
;
218 struct btrfs_trans_handle
*trans
;
219 unsigned int flags
, oldflags
;
222 unsigned int i_oldflags
;
225 if (!inode_owner_or_capable(inode
))
228 if (btrfs_root_readonly(root
))
231 if (copy_from_user(&flags
, arg
, sizeof(flags
)))
234 ret
= check_flags(flags
);
238 ret
= mnt_want_write_file(file
);
242 mutex_lock(&inode
->i_mutex
);
244 ip_oldflags
= ip
->flags
;
245 i_oldflags
= inode
->i_flags
;
246 mode
= inode
->i_mode
;
248 flags
= btrfs_mask_flags(inode
->i_mode
, flags
);
249 oldflags
= btrfs_flags_to_ioctl(ip
->flags
);
250 if ((flags
^ oldflags
) & (FS_APPEND_FL
| FS_IMMUTABLE_FL
)) {
251 if (!capable(CAP_LINUX_IMMUTABLE
)) {
257 if (flags
& FS_SYNC_FL
)
258 ip
->flags
|= BTRFS_INODE_SYNC
;
260 ip
->flags
&= ~BTRFS_INODE_SYNC
;
261 if (flags
& FS_IMMUTABLE_FL
)
262 ip
->flags
|= BTRFS_INODE_IMMUTABLE
;
264 ip
->flags
&= ~BTRFS_INODE_IMMUTABLE
;
265 if (flags
& FS_APPEND_FL
)
266 ip
->flags
|= BTRFS_INODE_APPEND
;
268 ip
->flags
&= ~BTRFS_INODE_APPEND
;
269 if (flags
& FS_NODUMP_FL
)
270 ip
->flags
|= BTRFS_INODE_NODUMP
;
272 ip
->flags
&= ~BTRFS_INODE_NODUMP
;
273 if (flags
& FS_NOATIME_FL
)
274 ip
->flags
|= BTRFS_INODE_NOATIME
;
276 ip
->flags
&= ~BTRFS_INODE_NOATIME
;
277 if (flags
& FS_DIRSYNC_FL
)
278 ip
->flags
|= BTRFS_INODE_DIRSYNC
;
280 ip
->flags
&= ~BTRFS_INODE_DIRSYNC
;
281 if (flags
& FS_NOCOW_FL
) {
284 * It's safe to turn csums off here, no extents exist.
285 * Otherwise we want the flag to reflect the real COW
286 * status of the file and will not set it.
288 if (inode
->i_size
== 0)
289 ip
->flags
|= BTRFS_INODE_NODATACOW
290 | BTRFS_INODE_NODATASUM
;
292 ip
->flags
|= BTRFS_INODE_NODATACOW
;
296 * Revert back under same assuptions as above
299 if (inode
->i_size
== 0)
300 ip
->flags
&= ~(BTRFS_INODE_NODATACOW
301 | BTRFS_INODE_NODATASUM
);
303 ip
->flags
&= ~BTRFS_INODE_NODATACOW
;
308 * The COMPRESS flag can only be changed by users, while the NOCOMPRESS
309 * flag may be changed automatically if compression code won't make
312 if (flags
& FS_NOCOMP_FL
) {
313 ip
->flags
&= ~BTRFS_INODE_COMPRESS
;
314 ip
->flags
|= BTRFS_INODE_NOCOMPRESS
;
316 ret
= btrfs_set_prop(inode
, "btrfs.compression", NULL
, 0, 0);
317 if (ret
&& ret
!= -ENODATA
)
319 } else if (flags
& FS_COMPR_FL
) {
322 ip
->flags
|= BTRFS_INODE_COMPRESS
;
323 ip
->flags
&= ~BTRFS_INODE_NOCOMPRESS
;
325 if (root
->fs_info
->compress_type
== BTRFS_COMPRESS_LZO
)
329 ret
= btrfs_set_prop(inode
, "btrfs.compression",
330 comp
, strlen(comp
), 0);
335 ret
= btrfs_set_prop(inode
, "btrfs.compression", NULL
, 0, 0);
336 if (ret
&& ret
!= -ENODATA
)
338 ip
->flags
&= ~(BTRFS_INODE_COMPRESS
| BTRFS_INODE_NOCOMPRESS
);
341 trans
= btrfs_start_transaction(root
, 1);
343 ret
= PTR_ERR(trans
);
347 btrfs_update_iflags(inode
);
348 inode_inc_iversion(inode
);
349 inode
->i_ctime
= CURRENT_TIME
;
350 ret
= btrfs_update_inode(trans
, root
, inode
);
352 btrfs_end_transaction(trans
, root
);
355 ip
->flags
= ip_oldflags
;
356 inode
->i_flags
= i_oldflags
;
360 mutex_unlock(&inode
->i_mutex
);
361 mnt_drop_write_file(file
);
365 static int btrfs_ioctl_getversion(struct file
*file
, int __user
*arg
)
367 struct inode
*inode
= file_inode(file
);
369 return put_user(inode
->i_generation
, arg
);
372 static noinline
int btrfs_ioctl_fitrim(struct file
*file
, void __user
*arg
)
374 struct btrfs_fs_info
*fs_info
= btrfs_sb(file_inode(file
)->i_sb
);
375 struct btrfs_device
*device
;
376 struct request_queue
*q
;
377 struct fstrim_range range
;
378 u64 minlen
= ULLONG_MAX
;
380 u64 total_bytes
= btrfs_super_total_bytes(fs_info
->super_copy
);
383 if (!capable(CAP_SYS_ADMIN
))
387 list_for_each_entry_rcu(device
, &fs_info
->fs_devices
->devices
,
391 q
= bdev_get_queue(device
->bdev
);
392 if (blk_queue_discard(q
)) {
394 minlen
= min((u64
)q
->limits
.discard_granularity
,
402 if (copy_from_user(&range
, arg
, sizeof(range
)))
404 if (range
.start
> total_bytes
||
405 range
.len
< fs_info
->sb
->s_blocksize
)
408 range
.len
= min(range
.len
, total_bytes
- range
.start
);
409 range
.minlen
= max(range
.minlen
, minlen
);
410 ret
= btrfs_trim_fs(fs_info
->tree_root
, &range
);
414 if (copy_to_user(arg
, &range
, sizeof(range
)))
420 int btrfs_is_empty_uuid(u8
*uuid
)
424 for (i
= 0; i
< BTRFS_UUID_SIZE
; i
++) {
431 static noinline
int create_subvol(struct inode
*dir
,
432 struct dentry
*dentry
,
433 char *name
, int namelen
,
435 struct btrfs_qgroup_inherit
*inherit
)
437 struct btrfs_trans_handle
*trans
;
438 struct btrfs_key key
;
439 struct btrfs_root_item root_item
;
440 struct btrfs_inode_item
*inode_item
;
441 struct extent_buffer
*leaf
;
442 struct btrfs_root
*root
= BTRFS_I(dir
)->root
;
443 struct btrfs_root
*new_root
;
444 struct btrfs_block_rsv block_rsv
;
445 struct timespec cur_time
= CURRENT_TIME
;
450 u64 new_dirid
= BTRFS_FIRST_FREE_OBJECTID
;
455 ret
= btrfs_find_free_objectid(root
->fs_info
->tree_root
, &objectid
);
459 btrfs_init_block_rsv(&block_rsv
, BTRFS_BLOCK_RSV_TEMP
);
461 * The same as the snapshot creation, please see the comment
462 * of create_snapshot().
464 ret
= btrfs_subvolume_reserve_metadata(root
, &block_rsv
,
465 8, &qgroup_reserved
, false);
469 trans
= btrfs_start_transaction(root
, 0);
471 ret
= PTR_ERR(trans
);
472 btrfs_subvolume_release_metadata(root
, &block_rsv
,
476 trans
->block_rsv
= &block_rsv
;
477 trans
->bytes_reserved
= block_rsv
.size
;
479 ret
= btrfs_qgroup_inherit(trans
, root
->fs_info
, 0, objectid
, inherit
);
483 leaf
= btrfs_alloc_tree_block(trans
, root
, 0, objectid
, NULL
, 0, 0, 0);
489 memset_extent_buffer(leaf
, 0, 0, sizeof(struct btrfs_header
));
490 btrfs_set_header_bytenr(leaf
, leaf
->start
);
491 btrfs_set_header_generation(leaf
, trans
->transid
);
492 btrfs_set_header_backref_rev(leaf
, BTRFS_MIXED_BACKREF_REV
);
493 btrfs_set_header_owner(leaf
, objectid
);
495 write_extent_buffer(leaf
, root
->fs_info
->fsid
, btrfs_header_fsid(),
497 write_extent_buffer(leaf
, root
->fs_info
->chunk_tree_uuid
,
498 btrfs_header_chunk_tree_uuid(leaf
),
500 btrfs_mark_buffer_dirty(leaf
);
502 memset(&root_item
, 0, sizeof(root_item
));
504 inode_item
= &root_item
.inode
;
505 btrfs_set_stack_inode_generation(inode_item
, 1);
506 btrfs_set_stack_inode_size(inode_item
, 3);
507 btrfs_set_stack_inode_nlink(inode_item
, 1);
508 btrfs_set_stack_inode_nbytes(inode_item
, root
->nodesize
);
509 btrfs_set_stack_inode_mode(inode_item
, S_IFDIR
| 0755);
511 btrfs_set_root_flags(&root_item
, 0);
512 btrfs_set_root_limit(&root_item
, 0);
513 btrfs_set_stack_inode_flags(inode_item
, BTRFS_INODE_ROOT_ITEM_INIT
);
515 btrfs_set_root_bytenr(&root_item
, leaf
->start
);
516 btrfs_set_root_generation(&root_item
, trans
->transid
);
517 btrfs_set_root_level(&root_item
, 0);
518 btrfs_set_root_refs(&root_item
, 1);
519 btrfs_set_root_used(&root_item
, leaf
->len
);
520 btrfs_set_root_last_snapshot(&root_item
, 0);
522 btrfs_set_root_generation_v2(&root_item
,
523 btrfs_root_generation(&root_item
));
524 uuid_le_gen(&new_uuid
);
525 memcpy(root_item
.uuid
, new_uuid
.b
, BTRFS_UUID_SIZE
);
526 btrfs_set_stack_timespec_sec(&root_item
.otime
, cur_time
.tv_sec
);
527 btrfs_set_stack_timespec_nsec(&root_item
.otime
, cur_time
.tv_nsec
);
528 root_item
.ctime
= root_item
.otime
;
529 btrfs_set_root_ctransid(&root_item
, trans
->transid
);
530 btrfs_set_root_otransid(&root_item
, trans
->transid
);
532 btrfs_tree_unlock(leaf
);
533 free_extent_buffer(leaf
);
536 btrfs_set_root_dirid(&root_item
, new_dirid
);
538 key
.objectid
= objectid
;
540 key
.type
= BTRFS_ROOT_ITEM_KEY
;
541 ret
= btrfs_insert_root(trans
, root
->fs_info
->tree_root
, &key
,
546 key
.offset
= (u64
)-1;
547 new_root
= btrfs_read_fs_root_no_name(root
->fs_info
, &key
);
548 if (IS_ERR(new_root
)) {
549 btrfs_abort_transaction(trans
, root
, PTR_ERR(new_root
));
550 ret
= PTR_ERR(new_root
);
554 btrfs_record_root_in_trans(trans
, new_root
);
556 ret
= btrfs_create_subvol_root(trans
, new_root
, root
, new_dirid
);
558 /* We potentially lose an unused inode item here */
559 btrfs_abort_transaction(trans
, root
, ret
);
564 * insert the directory item
566 ret
= btrfs_set_inode_index(dir
, &index
);
568 btrfs_abort_transaction(trans
, root
, ret
);
572 ret
= btrfs_insert_dir_item(trans
, root
,
573 name
, namelen
, dir
, &key
,
574 BTRFS_FT_DIR
, index
);
576 btrfs_abort_transaction(trans
, root
, ret
);
580 btrfs_i_size_write(dir
, dir
->i_size
+ namelen
* 2);
581 ret
= btrfs_update_inode(trans
, root
, dir
);
584 ret
= btrfs_add_root_ref(trans
, root
->fs_info
->tree_root
,
585 objectid
, root
->root_key
.objectid
,
586 btrfs_ino(dir
), index
, name
, namelen
);
589 ret
= btrfs_uuid_tree_add(trans
, root
->fs_info
->uuid_root
,
590 root_item
.uuid
, BTRFS_UUID_KEY_SUBVOL
,
593 btrfs_abort_transaction(trans
, root
, ret
);
596 trans
->block_rsv
= NULL
;
597 trans
->bytes_reserved
= 0;
598 btrfs_subvolume_release_metadata(root
, &block_rsv
, qgroup_reserved
);
601 *async_transid
= trans
->transid
;
602 err
= btrfs_commit_transaction_async(trans
, root
, 1);
604 err
= btrfs_commit_transaction(trans
, root
);
606 err
= btrfs_commit_transaction(trans
, root
);
612 inode
= btrfs_lookup_dentry(dir
, dentry
);
614 return PTR_ERR(inode
);
615 d_instantiate(dentry
, inode
);
620 static void btrfs_wait_for_no_snapshoting_writes(struct btrfs_root
*root
)
626 prepare_to_wait(&root
->subv_writers
->wait
, &wait
,
627 TASK_UNINTERRUPTIBLE
);
629 writers
= percpu_counter_sum(&root
->subv_writers
->counter
);
633 finish_wait(&root
->subv_writers
->wait
, &wait
);
637 static int create_snapshot(struct btrfs_root
*root
, struct inode
*dir
,
638 struct dentry
*dentry
, char *name
, int namelen
,
639 u64
*async_transid
, bool readonly
,
640 struct btrfs_qgroup_inherit
*inherit
)
643 struct btrfs_pending_snapshot
*pending_snapshot
;
644 struct btrfs_trans_handle
*trans
;
647 if (!test_bit(BTRFS_ROOT_REF_COWS
, &root
->state
))
650 atomic_inc(&root
->will_be_snapshoted
);
651 smp_mb__after_atomic();
652 btrfs_wait_for_no_snapshoting_writes(root
);
654 ret
= btrfs_start_delalloc_inodes(root
, 0);
658 btrfs_wait_ordered_extents(root
, -1);
660 pending_snapshot
= kzalloc(sizeof(*pending_snapshot
), GFP_NOFS
);
661 if (!pending_snapshot
) {
666 btrfs_init_block_rsv(&pending_snapshot
->block_rsv
,
667 BTRFS_BLOCK_RSV_TEMP
);
669 * 1 - parent dir inode
672 * 2 - root ref/backref
673 * 1 - root of snapshot
676 ret
= btrfs_subvolume_reserve_metadata(BTRFS_I(dir
)->root
,
677 &pending_snapshot
->block_rsv
, 8,
678 &pending_snapshot
->qgroup_reserved
,
683 pending_snapshot
->dentry
= dentry
;
684 pending_snapshot
->root
= root
;
685 pending_snapshot
->readonly
= readonly
;
686 pending_snapshot
->dir
= dir
;
687 pending_snapshot
->inherit
= inherit
;
689 trans
= btrfs_start_transaction(root
, 0);
691 ret
= PTR_ERR(trans
);
695 spin_lock(&root
->fs_info
->trans_lock
);
696 list_add(&pending_snapshot
->list
,
697 &trans
->transaction
->pending_snapshots
);
698 spin_unlock(&root
->fs_info
->trans_lock
);
700 *async_transid
= trans
->transid
;
701 ret
= btrfs_commit_transaction_async(trans
,
702 root
->fs_info
->extent_root
, 1);
704 ret
= btrfs_commit_transaction(trans
, root
);
706 ret
= btrfs_commit_transaction(trans
,
707 root
->fs_info
->extent_root
);
712 ret
= pending_snapshot
->error
;
716 ret
= btrfs_orphan_cleanup(pending_snapshot
->snap
);
720 inode
= btrfs_lookup_dentry(dentry
->d_parent
->d_inode
, dentry
);
722 ret
= PTR_ERR(inode
);
726 d_instantiate(dentry
, inode
);
729 btrfs_subvolume_release_metadata(BTRFS_I(dir
)->root
,
730 &pending_snapshot
->block_rsv
,
731 pending_snapshot
->qgroup_reserved
);
733 kfree(pending_snapshot
);
735 if (atomic_dec_and_test(&root
->will_be_snapshoted
))
736 wake_up_atomic_t(&root
->will_be_snapshoted
);
740 /* copy of may_delete in fs/namei.c()
741 * Check whether we can remove a link victim from directory dir, check
742 * whether the type of victim is right.
743 * 1. We can't do it if dir is read-only (done in permission())
744 * 2. We should have write and exec permissions on dir
745 * 3. We can't remove anything from append-only dir
746 * 4. We can't do anything with immutable dir (done in permission())
747 * 5. If the sticky bit on dir is set we should either
748 * a. be owner of dir, or
749 * b. be owner of victim, or
750 * c. have CAP_FOWNER capability
751 * 6. If the victim is append-only or immutable we can't do antyhing with
752 * links pointing to it.
753 * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR.
754 * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR.
755 * 9. We can't remove a root or mountpoint.
756 * 10. We don't allow removal of NFS sillyrenamed files; it's handled by
757 * nfs_async_unlink().
760 static int btrfs_may_delete(struct inode
*dir
, struct dentry
*victim
, int isdir
)
764 if (!victim
->d_inode
)
767 BUG_ON(victim
->d_parent
->d_inode
!= dir
);
768 audit_inode_child(dir
, victim
, AUDIT_TYPE_CHILD_DELETE
);
770 error
= inode_permission(dir
, MAY_WRITE
| MAY_EXEC
);
775 if (check_sticky(dir
, victim
->d_inode
) || IS_APPEND(victim
->d_inode
) ||
776 IS_IMMUTABLE(victim
->d_inode
) || IS_SWAPFILE(victim
->d_inode
))
779 if (!d_is_dir(victim
))
783 } else if (d_is_dir(victim
))
787 if (victim
->d_flags
& DCACHE_NFSFS_RENAMED
)
792 /* copy of may_create in fs/namei.c() */
793 static inline int btrfs_may_create(struct inode
*dir
, struct dentry
*child
)
799 return inode_permission(dir
, MAY_WRITE
| MAY_EXEC
);
803 * Create a new subvolume below @parent. This is largely modeled after
804 * sys_mkdirat and vfs_mkdir, but we only do a single component lookup
805 * inside this filesystem so it's quite a bit simpler.
807 static noinline
int btrfs_mksubvol(struct path
*parent
,
808 char *name
, int namelen
,
809 struct btrfs_root
*snap_src
,
810 u64
*async_transid
, bool readonly
,
811 struct btrfs_qgroup_inherit
*inherit
)
813 struct inode
*dir
= parent
->dentry
->d_inode
;
814 struct dentry
*dentry
;
817 error
= mutex_lock_killable_nested(&dir
->i_mutex
, I_MUTEX_PARENT
);
821 dentry
= lookup_one_len(name
, parent
->dentry
, namelen
);
822 error
= PTR_ERR(dentry
);
830 error
= btrfs_may_create(dir
, dentry
);
835 * even if this name doesn't exist, we may get hash collisions.
836 * check for them now when we can safely fail
838 error
= btrfs_check_dir_item_collision(BTRFS_I(dir
)->root
,
844 down_read(&BTRFS_I(dir
)->root
->fs_info
->subvol_sem
);
846 if (btrfs_root_refs(&BTRFS_I(dir
)->root
->root_item
) == 0)
850 error
= create_snapshot(snap_src
, dir
, dentry
, name
, namelen
,
851 async_transid
, readonly
, inherit
);
853 error
= create_subvol(dir
, dentry
, name
, namelen
,
854 async_transid
, inherit
);
857 fsnotify_mkdir(dir
, dentry
);
859 up_read(&BTRFS_I(dir
)->root
->fs_info
->subvol_sem
);
863 mutex_unlock(&dir
->i_mutex
);
868 * When we're defragging a range, we don't want to kick it off again
869 * if it is really just waiting for delalloc to send it down.
870 * If we find a nice big extent or delalloc range for the bytes in the
871 * file you want to defrag, we return 0 to let you know to skip this
874 static int check_defrag_in_cache(struct inode
*inode
, u64 offset
, u32 thresh
)
876 struct extent_io_tree
*io_tree
= &BTRFS_I(inode
)->io_tree
;
877 struct extent_map
*em
= NULL
;
878 struct extent_map_tree
*em_tree
= &BTRFS_I(inode
)->extent_tree
;
881 read_lock(&em_tree
->lock
);
882 em
= lookup_extent_mapping(em_tree
, offset
, PAGE_CACHE_SIZE
);
883 read_unlock(&em_tree
->lock
);
886 end
= extent_map_end(em
);
888 if (end
- offset
> thresh
)
891 /* if we already have a nice delalloc here, just stop */
893 end
= count_range_bits(io_tree
, &offset
, offset
+ thresh
,
894 thresh
, EXTENT_DELALLOC
, 1);
901 * helper function to walk through a file and find extents
902 * newer than a specific transid, and smaller than thresh.
904 * This is used by the defragging code to find new and small
907 static int find_new_extents(struct btrfs_root
*root
,
908 struct inode
*inode
, u64 newer_than
,
909 u64
*off
, u32 thresh
)
911 struct btrfs_path
*path
;
912 struct btrfs_key min_key
;
913 struct extent_buffer
*leaf
;
914 struct btrfs_file_extent_item
*extent
;
917 u64 ino
= btrfs_ino(inode
);
919 path
= btrfs_alloc_path();
923 min_key
.objectid
= ino
;
924 min_key
.type
= BTRFS_EXTENT_DATA_KEY
;
925 min_key
.offset
= *off
;
928 ret
= btrfs_search_forward(root
, &min_key
, path
, newer_than
);
932 if (min_key
.objectid
!= ino
)
934 if (min_key
.type
!= BTRFS_EXTENT_DATA_KEY
)
937 leaf
= path
->nodes
[0];
938 extent
= btrfs_item_ptr(leaf
, path
->slots
[0],
939 struct btrfs_file_extent_item
);
941 type
= btrfs_file_extent_type(leaf
, extent
);
942 if (type
== BTRFS_FILE_EXTENT_REG
&&
943 btrfs_file_extent_num_bytes(leaf
, extent
) < thresh
&&
944 check_defrag_in_cache(inode
, min_key
.offset
, thresh
)) {
945 *off
= min_key
.offset
;
946 btrfs_free_path(path
);
951 if (path
->slots
[0] < btrfs_header_nritems(leaf
)) {
952 btrfs_item_key_to_cpu(leaf
, &min_key
, path
->slots
[0]);
956 if (min_key
.offset
== (u64
)-1)
960 btrfs_release_path(path
);
963 btrfs_free_path(path
);
967 static struct extent_map
*defrag_lookup_extent(struct inode
*inode
, u64 start
)
969 struct extent_map_tree
*em_tree
= &BTRFS_I(inode
)->extent_tree
;
970 struct extent_io_tree
*io_tree
= &BTRFS_I(inode
)->io_tree
;
971 struct extent_map
*em
;
972 u64 len
= PAGE_CACHE_SIZE
;
975 * hopefully we have this extent in the tree already, try without
976 * the full extent lock
978 read_lock(&em_tree
->lock
);
979 em
= lookup_extent_mapping(em_tree
, start
, len
);
980 read_unlock(&em_tree
->lock
);
983 struct extent_state
*cached
= NULL
;
984 u64 end
= start
+ len
- 1;
986 /* get the big lock and read metadata off disk */
987 lock_extent_bits(io_tree
, start
, end
, 0, &cached
);
988 em
= btrfs_get_extent(inode
, NULL
, 0, start
, len
, 0);
989 unlock_extent_cached(io_tree
, start
, end
, &cached
, GFP_NOFS
);
998 static bool defrag_check_next_extent(struct inode
*inode
, struct extent_map
*em
)
1000 struct extent_map
*next
;
1003 /* this is the last extent */
1004 if (em
->start
+ em
->len
>= i_size_read(inode
))
1007 next
= defrag_lookup_extent(inode
, em
->start
+ em
->len
);
1008 if (!next
|| next
->block_start
>= EXTENT_MAP_LAST_BYTE
)
1010 else if ((em
->block_start
+ em
->block_len
== next
->block_start
) &&
1011 (em
->block_len
> 128 * 1024 && next
->block_len
> 128 * 1024))
1014 free_extent_map(next
);
1018 static int should_defrag_range(struct inode
*inode
, u64 start
, u32 thresh
,
1019 u64
*last_len
, u64
*skip
, u64
*defrag_end
,
1022 struct extent_map
*em
;
1024 bool next_mergeable
= true;
1027 * make sure that once we start defragging an extent, we keep on
1030 if (start
< *defrag_end
)
1035 em
= defrag_lookup_extent(inode
, start
);
1039 /* this will cover holes, and inline extents */
1040 if (em
->block_start
>= EXTENT_MAP_LAST_BYTE
) {
1045 next_mergeable
= defrag_check_next_extent(inode
, em
);
1047 * we hit a real extent, if it is big or the next extent is not a
1048 * real extent, don't bother defragging it
1050 if (!compress
&& (*last_len
== 0 || *last_len
>= thresh
) &&
1051 (em
->len
>= thresh
|| !next_mergeable
))
1055 * last_len ends up being a counter of how many bytes we've defragged.
1056 * every time we choose not to defrag an extent, we reset *last_len
1057 * so that the next tiny extent will force a defrag.
1059 * The end result of this is that tiny extents before a single big
1060 * extent will force at least part of that big extent to be defragged.
1063 *defrag_end
= extent_map_end(em
);
1066 *skip
= extent_map_end(em
);
1070 free_extent_map(em
);
1075 * it doesn't do much good to defrag one or two pages
1076 * at a time. This pulls in a nice chunk of pages
1077 * to COW and defrag.
1079 * It also makes sure the delalloc code has enough
1080 * dirty data to avoid making new small extents as part
1083 * It's a good idea to start RA on this range
1084 * before calling this.
1086 static int cluster_pages_for_defrag(struct inode
*inode
,
1087 struct page
**pages
,
1088 unsigned long start_index
,
1089 unsigned long num_pages
)
1091 unsigned long file_end
;
1092 u64 isize
= i_size_read(inode
);
1099 struct btrfs_ordered_extent
*ordered
;
1100 struct extent_state
*cached_state
= NULL
;
1101 struct extent_io_tree
*tree
;
1102 gfp_t mask
= btrfs_alloc_write_mask(inode
->i_mapping
);
1104 file_end
= (isize
- 1) >> PAGE_CACHE_SHIFT
;
1105 if (!isize
|| start_index
> file_end
)
1108 page_cnt
= min_t(u64
, (u64
)num_pages
, (u64
)file_end
- start_index
+ 1);
1110 ret
= btrfs_delalloc_reserve_space(inode
,
1111 page_cnt
<< PAGE_CACHE_SHIFT
);
1115 tree
= &BTRFS_I(inode
)->io_tree
;
1117 /* step one, lock all the pages */
1118 for (i
= 0; i
< page_cnt
; i
++) {
1121 page
= find_or_create_page(inode
->i_mapping
,
1122 start_index
+ i
, mask
);
1126 page_start
= page_offset(page
);
1127 page_end
= page_start
+ PAGE_CACHE_SIZE
- 1;
1129 lock_extent_bits(tree
, page_start
, page_end
,
1131 ordered
= btrfs_lookup_ordered_extent(inode
,
1133 unlock_extent_cached(tree
, page_start
, page_end
,
1134 &cached_state
, GFP_NOFS
);
1139 btrfs_start_ordered_extent(inode
, ordered
, 1);
1140 btrfs_put_ordered_extent(ordered
);
1143 * we unlocked the page above, so we need check if
1144 * it was released or not.
1146 if (page
->mapping
!= inode
->i_mapping
) {
1148 page_cache_release(page
);
1153 if (!PageUptodate(page
)) {
1154 btrfs_readpage(NULL
, page
);
1156 if (!PageUptodate(page
)) {
1158 page_cache_release(page
);
1164 if (page
->mapping
!= inode
->i_mapping
) {
1166 page_cache_release(page
);
1176 if (!(inode
->i_sb
->s_flags
& MS_ACTIVE
))
1180 * so now we have a nice long stream of locked
1181 * and up to date pages, lets wait on them
1183 for (i
= 0; i
< i_done
; i
++)
1184 wait_on_page_writeback(pages
[i
]);
1186 page_start
= page_offset(pages
[0]);
1187 page_end
= page_offset(pages
[i_done
- 1]) + PAGE_CACHE_SIZE
;
1189 lock_extent_bits(&BTRFS_I(inode
)->io_tree
,
1190 page_start
, page_end
- 1, 0, &cached_state
);
1191 clear_extent_bit(&BTRFS_I(inode
)->io_tree
, page_start
,
1192 page_end
- 1, EXTENT_DIRTY
| EXTENT_DELALLOC
|
1193 EXTENT_DO_ACCOUNTING
| EXTENT_DEFRAG
, 0, 0,
1194 &cached_state
, GFP_NOFS
);
1196 if (i_done
!= page_cnt
) {
1197 spin_lock(&BTRFS_I(inode
)->lock
);
1198 BTRFS_I(inode
)->outstanding_extents
++;
1199 spin_unlock(&BTRFS_I(inode
)->lock
);
1200 btrfs_delalloc_release_space(inode
,
1201 (page_cnt
- i_done
) << PAGE_CACHE_SHIFT
);
1205 set_extent_defrag(&BTRFS_I(inode
)->io_tree
, page_start
, page_end
- 1,
1206 &cached_state
, GFP_NOFS
);
1208 unlock_extent_cached(&BTRFS_I(inode
)->io_tree
,
1209 page_start
, page_end
- 1, &cached_state
,
1212 for (i
= 0; i
< i_done
; i
++) {
1213 clear_page_dirty_for_io(pages
[i
]);
1214 ClearPageChecked(pages
[i
]);
1215 set_page_extent_mapped(pages
[i
]);
1216 set_page_dirty(pages
[i
]);
1217 unlock_page(pages
[i
]);
1218 page_cache_release(pages
[i
]);
1222 for (i
= 0; i
< i_done
; i
++) {
1223 unlock_page(pages
[i
]);
1224 page_cache_release(pages
[i
]);
1226 btrfs_delalloc_release_space(inode
, page_cnt
<< PAGE_CACHE_SHIFT
);
1231 int btrfs_defrag_file(struct inode
*inode
, struct file
*file
,
1232 struct btrfs_ioctl_defrag_range_args
*range
,
1233 u64 newer_than
, unsigned long max_to_defrag
)
1235 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
1236 struct file_ra_state
*ra
= NULL
;
1237 unsigned long last_index
;
1238 u64 isize
= i_size_read(inode
);
1242 u64 newer_off
= range
->start
;
1244 unsigned long ra_index
= 0;
1246 int defrag_count
= 0;
1247 int compress_type
= BTRFS_COMPRESS_ZLIB
;
1248 u32 extent_thresh
= range
->extent_thresh
;
1249 unsigned long max_cluster
= (256 * 1024) >> PAGE_CACHE_SHIFT
;
1250 unsigned long cluster
= max_cluster
;
1251 u64 new_align
= ~((u64
)128 * 1024 - 1);
1252 struct page
**pages
= NULL
;
1257 if (range
->start
>= isize
)
1260 if (range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
) {
1261 if (range
->compress_type
> BTRFS_COMPRESS_TYPES
)
1263 if (range
->compress_type
)
1264 compress_type
= range
->compress_type
;
1267 if (extent_thresh
== 0)
1268 extent_thresh
= 256 * 1024;
1271 * if we were not given a file, allocate a readahead
1275 ra
= kzalloc(sizeof(*ra
), GFP_NOFS
);
1278 file_ra_state_init(ra
, inode
->i_mapping
);
1283 pages
= kmalloc_array(max_cluster
, sizeof(struct page
*),
1290 /* find the last page to defrag */
1291 if (range
->start
+ range
->len
> range
->start
) {
1292 last_index
= min_t(u64
, isize
- 1,
1293 range
->start
+ range
->len
- 1) >> PAGE_CACHE_SHIFT
;
1295 last_index
= (isize
- 1) >> PAGE_CACHE_SHIFT
;
1299 ret
= find_new_extents(root
, inode
, newer_than
,
1300 &newer_off
, 64 * 1024);
1302 range
->start
= newer_off
;
1304 * we always align our defrag to help keep
1305 * the extents in the file evenly spaced
1307 i
= (newer_off
& new_align
) >> PAGE_CACHE_SHIFT
;
1311 i
= range
->start
>> PAGE_CACHE_SHIFT
;
1314 max_to_defrag
= last_index
+ 1;
1317 * make writeback starts from i, so the defrag range can be
1318 * written sequentially.
1320 if (i
< inode
->i_mapping
->writeback_index
)
1321 inode
->i_mapping
->writeback_index
= i
;
1323 while (i
<= last_index
&& defrag_count
< max_to_defrag
&&
1324 (i
< DIV_ROUND_UP(i_size_read(inode
), PAGE_CACHE_SIZE
))) {
1326 * make sure we stop running if someone unmounts
1329 if (!(inode
->i_sb
->s_flags
& MS_ACTIVE
))
1332 if (btrfs_defrag_cancelled(root
->fs_info
)) {
1333 printk(KERN_DEBUG
"BTRFS: defrag_file cancelled\n");
1338 if (!should_defrag_range(inode
, (u64
)i
<< PAGE_CACHE_SHIFT
,
1339 extent_thresh
, &last_len
, &skip
,
1340 &defrag_end
, range
->flags
&
1341 BTRFS_DEFRAG_RANGE_COMPRESS
)) {
1344 * the should_defrag function tells us how much to skip
1345 * bump our counter by the suggested amount
1347 next
= DIV_ROUND_UP(skip
, PAGE_CACHE_SIZE
);
1348 i
= max(i
+ 1, next
);
1353 cluster
= (PAGE_CACHE_ALIGN(defrag_end
) >>
1354 PAGE_CACHE_SHIFT
) - i
;
1355 cluster
= min(cluster
, max_cluster
);
1357 cluster
= max_cluster
;
1360 if (i
+ cluster
> ra_index
) {
1361 ra_index
= max(i
, ra_index
);
1362 btrfs_force_ra(inode
->i_mapping
, ra
, file
, ra_index
,
1364 ra_index
+= max_cluster
;
1367 mutex_lock(&inode
->i_mutex
);
1368 if (range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
)
1369 BTRFS_I(inode
)->force_compress
= compress_type
;
1370 ret
= cluster_pages_for_defrag(inode
, pages
, i
, cluster
);
1372 mutex_unlock(&inode
->i_mutex
);
1376 defrag_count
+= ret
;
1377 balance_dirty_pages_ratelimited(inode
->i_mapping
);
1378 mutex_unlock(&inode
->i_mutex
);
1381 if (newer_off
== (u64
)-1)
1387 newer_off
= max(newer_off
+ 1,
1388 (u64
)i
<< PAGE_CACHE_SHIFT
);
1390 ret
= find_new_extents(root
, inode
,
1391 newer_than
, &newer_off
,
1394 range
->start
= newer_off
;
1395 i
= (newer_off
& new_align
) >> PAGE_CACHE_SHIFT
;
1402 last_len
+= ret
<< PAGE_CACHE_SHIFT
;
1410 if ((range
->flags
& BTRFS_DEFRAG_RANGE_START_IO
)) {
1411 filemap_flush(inode
->i_mapping
);
1412 if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT
,
1413 &BTRFS_I(inode
)->runtime_flags
))
1414 filemap_flush(inode
->i_mapping
);
1417 if ((range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
)) {
1418 /* the filemap_flush will queue IO into the worker threads, but
1419 * we have to make sure the IO is actually started and that
1420 * ordered extents get created before we return
1422 atomic_inc(&root
->fs_info
->async_submit_draining
);
1423 while (atomic_read(&root
->fs_info
->nr_async_submits
) ||
1424 atomic_read(&root
->fs_info
->async_delalloc_pages
)) {
1425 wait_event(root
->fs_info
->async_submit_wait
,
1426 (atomic_read(&root
->fs_info
->nr_async_submits
) == 0 &&
1427 atomic_read(&root
->fs_info
->async_delalloc_pages
) == 0));
1429 atomic_dec(&root
->fs_info
->async_submit_draining
);
1432 if (range
->compress_type
== BTRFS_COMPRESS_LZO
) {
1433 btrfs_set_fs_incompat(root
->fs_info
, COMPRESS_LZO
);
1439 if (range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
) {
1440 mutex_lock(&inode
->i_mutex
);
1441 BTRFS_I(inode
)->force_compress
= BTRFS_COMPRESS_NONE
;
1442 mutex_unlock(&inode
->i_mutex
);
1450 static noinline
int btrfs_ioctl_resize(struct file
*file
,
1456 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
1457 struct btrfs_ioctl_vol_args
*vol_args
;
1458 struct btrfs_trans_handle
*trans
;
1459 struct btrfs_device
*device
= NULL
;
1462 char *devstr
= NULL
;
1466 if (!capable(CAP_SYS_ADMIN
))
1469 ret
= mnt_want_write_file(file
);
1473 if (atomic_xchg(&root
->fs_info
->mutually_exclusive_operation_running
,
1475 mnt_drop_write_file(file
);
1476 return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS
;
1479 mutex_lock(&root
->fs_info
->volume_mutex
);
1480 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1481 if (IS_ERR(vol_args
)) {
1482 ret
= PTR_ERR(vol_args
);
1486 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
1488 sizestr
= vol_args
->name
;
1489 devstr
= strchr(sizestr
, ':');
1491 sizestr
= devstr
+ 1;
1493 devstr
= vol_args
->name
;
1494 ret
= kstrtoull(devstr
, 10, &devid
);
1501 btrfs_info(root
->fs_info
, "resizing devid %llu", devid
);
1504 device
= btrfs_find_device(root
->fs_info
, devid
, NULL
, NULL
);
1506 btrfs_info(root
->fs_info
, "resizer unable to find device %llu",
1512 if (!device
->writeable
) {
1513 btrfs_info(root
->fs_info
,
1514 "resizer unable to apply on readonly device %llu",
1520 if (!strcmp(sizestr
, "max"))
1521 new_size
= device
->bdev
->bd_inode
->i_size
;
1523 if (sizestr
[0] == '-') {
1526 } else if (sizestr
[0] == '+') {
1530 new_size
= memparse(sizestr
, &retptr
);
1531 if (*retptr
!= '\0' || new_size
== 0) {
1537 if (device
->is_tgtdev_for_dev_replace
) {
1542 old_size
= btrfs_device_get_total_bytes(device
);
1545 if (new_size
> old_size
) {
1549 new_size
= old_size
- new_size
;
1550 } else if (mod
> 0) {
1551 if (new_size
> ULLONG_MAX
- old_size
) {
1555 new_size
= old_size
+ new_size
;
1558 if (new_size
< 256 * 1024 * 1024) {
1562 if (new_size
> device
->bdev
->bd_inode
->i_size
) {
1567 do_div(new_size
, root
->sectorsize
);
1568 new_size
*= root
->sectorsize
;
1570 printk_in_rcu(KERN_INFO
"BTRFS: new size for %s is %llu\n",
1571 rcu_str_deref(device
->name
), new_size
);
1573 if (new_size
> old_size
) {
1574 trans
= btrfs_start_transaction(root
, 0);
1575 if (IS_ERR(trans
)) {
1576 ret
= PTR_ERR(trans
);
1579 ret
= btrfs_grow_device(trans
, device
, new_size
);
1580 btrfs_commit_transaction(trans
, root
);
1581 } else if (new_size
< old_size
) {
1582 ret
= btrfs_shrink_device(device
, new_size
);
1583 } /* equal, nothing need to do */
1588 mutex_unlock(&root
->fs_info
->volume_mutex
);
1589 atomic_set(&root
->fs_info
->mutually_exclusive_operation_running
, 0);
1590 mnt_drop_write_file(file
);
1594 static noinline
int btrfs_ioctl_snap_create_transid(struct file
*file
,
1595 char *name
, unsigned long fd
, int subvol
,
1596 u64
*transid
, bool readonly
,
1597 struct btrfs_qgroup_inherit
*inherit
)
1602 ret
= mnt_want_write_file(file
);
1606 namelen
= strlen(name
);
1607 if (strchr(name
, '/')) {
1609 goto out_drop_write
;
1612 if (name
[0] == '.' &&
1613 (namelen
== 1 || (name
[1] == '.' && namelen
== 2))) {
1615 goto out_drop_write
;
1619 ret
= btrfs_mksubvol(&file
->f_path
, name
, namelen
,
1620 NULL
, transid
, readonly
, inherit
);
1622 struct fd src
= fdget(fd
);
1623 struct inode
*src_inode
;
1626 goto out_drop_write
;
1629 src_inode
= file_inode(src
.file
);
1630 if (src_inode
->i_sb
!= file_inode(file
)->i_sb
) {
1631 btrfs_info(BTRFS_I(src_inode
)->root
->fs_info
,
1632 "Snapshot src from another FS");
1634 } else if (!inode_owner_or_capable(src_inode
)) {
1636 * Subvolume creation is not restricted, but snapshots
1637 * are limited to own subvolumes only
1641 ret
= btrfs_mksubvol(&file
->f_path
, name
, namelen
,
1642 BTRFS_I(src_inode
)->root
,
1643 transid
, readonly
, inherit
);
1648 mnt_drop_write_file(file
);
1653 static noinline
int btrfs_ioctl_snap_create(struct file
*file
,
1654 void __user
*arg
, int subvol
)
1656 struct btrfs_ioctl_vol_args
*vol_args
;
1659 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1660 if (IS_ERR(vol_args
))
1661 return PTR_ERR(vol_args
);
1662 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
1664 ret
= btrfs_ioctl_snap_create_transid(file
, vol_args
->name
,
1665 vol_args
->fd
, subvol
,
1672 static noinline
int btrfs_ioctl_snap_create_v2(struct file
*file
,
1673 void __user
*arg
, int subvol
)
1675 struct btrfs_ioctl_vol_args_v2
*vol_args
;
1679 bool readonly
= false;
1680 struct btrfs_qgroup_inherit
*inherit
= NULL
;
1682 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1683 if (IS_ERR(vol_args
))
1684 return PTR_ERR(vol_args
);
1685 vol_args
->name
[BTRFS_SUBVOL_NAME_MAX
] = '\0';
1687 if (vol_args
->flags
&
1688 ~(BTRFS_SUBVOL_CREATE_ASYNC
| BTRFS_SUBVOL_RDONLY
|
1689 BTRFS_SUBVOL_QGROUP_INHERIT
)) {
1694 if (vol_args
->flags
& BTRFS_SUBVOL_CREATE_ASYNC
)
1696 if (vol_args
->flags
& BTRFS_SUBVOL_RDONLY
)
1698 if (vol_args
->flags
& BTRFS_SUBVOL_QGROUP_INHERIT
) {
1699 if (vol_args
->size
> PAGE_CACHE_SIZE
) {
1703 inherit
= memdup_user(vol_args
->qgroup_inherit
, vol_args
->size
);
1704 if (IS_ERR(inherit
)) {
1705 ret
= PTR_ERR(inherit
);
1710 ret
= btrfs_ioctl_snap_create_transid(file
, vol_args
->name
,
1711 vol_args
->fd
, subvol
, ptr
,
1716 if (ptr
&& copy_to_user(arg
+
1717 offsetof(struct btrfs_ioctl_vol_args_v2
,
1729 static noinline
int btrfs_ioctl_subvol_getflags(struct file
*file
,
1732 struct inode
*inode
= file_inode(file
);
1733 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
1737 if (btrfs_ino(inode
) != BTRFS_FIRST_FREE_OBJECTID
)
1740 down_read(&root
->fs_info
->subvol_sem
);
1741 if (btrfs_root_readonly(root
))
1742 flags
|= BTRFS_SUBVOL_RDONLY
;
1743 up_read(&root
->fs_info
->subvol_sem
);
1745 if (copy_to_user(arg
, &flags
, sizeof(flags
)))
1751 static noinline
int btrfs_ioctl_subvol_setflags(struct file
*file
,
1754 struct inode
*inode
= file_inode(file
);
1755 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
1756 struct btrfs_trans_handle
*trans
;
1761 if (!inode_owner_or_capable(inode
))
1764 ret
= mnt_want_write_file(file
);
1768 if (btrfs_ino(inode
) != BTRFS_FIRST_FREE_OBJECTID
) {
1770 goto out_drop_write
;
1773 if (copy_from_user(&flags
, arg
, sizeof(flags
))) {
1775 goto out_drop_write
;
1778 if (flags
& BTRFS_SUBVOL_CREATE_ASYNC
) {
1780 goto out_drop_write
;
1783 if (flags
& ~BTRFS_SUBVOL_RDONLY
) {
1785 goto out_drop_write
;
1788 down_write(&root
->fs_info
->subvol_sem
);
1791 if (!!(flags
& BTRFS_SUBVOL_RDONLY
) == btrfs_root_readonly(root
))
1794 root_flags
= btrfs_root_flags(&root
->root_item
);
1795 if (flags
& BTRFS_SUBVOL_RDONLY
) {
1796 btrfs_set_root_flags(&root
->root_item
,
1797 root_flags
| BTRFS_ROOT_SUBVOL_RDONLY
);
1800 * Block RO -> RW transition if this subvolume is involved in
1803 spin_lock(&root
->root_item_lock
);
1804 if (root
->send_in_progress
== 0) {
1805 btrfs_set_root_flags(&root
->root_item
,
1806 root_flags
& ~BTRFS_ROOT_SUBVOL_RDONLY
);
1807 spin_unlock(&root
->root_item_lock
);
1809 spin_unlock(&root
->root_item_lock
);
1810 btrfs_warn(root
->fs_info
,
1811 "Attempt to set subvolume %llu read-write during send",
1812 root
->root_key
.objectid
);
1818 trans
= btrfs_start_transaction(root
, 1);
1819 if (IS_ERR(trans
)) {
1820 ret
= PTR_ERR(trans
);
1824 ret
= btrfs_update_root(trans
, root
->fs_info
->tree_root
,
1825 &root
->root_key
, &root
->root_item
);
1827 btrfs_commit_transaction(trans
, root
);
1830 btrfs_set_root_flags(&root
->root_item
, root_flags
);
1832 up_write(&root
->fs_info
->subvol_sem
);
1834 mnt_drop_write_file(file
);
1840 * helper to check if the subvolume references other subvolumes
1842 static noinline
int may_destroy_subvol(struct btrfs_root
*root
)
1844 struct btrfs_path
*path
;
1845 struct btrfs_dir_item
*di
;
1846 struct btrfs_key key
;
1850 path
= btrfs_alloc_path();
1854 /* Make sure this root isn't set as the default subvol */
1855 dir_id
= btrfs_super_root_dir(root
->fs_info
->super_copy
);
1856 di
= btrfs_lookup_dir_item(NULL
, root
->fs_info
->tree_root
, path
,
1857 dir_id
, "default", 7, 0);
1858 if (di
&& !IS_ERR(di
)) {
1859 btrfs_dir_item_key_to_cpu(path
->nodes
[0], di
, &key
);
1860 if (key
.objectid
== root
->root_key
.objectid
) {
1862 btrfs_err(root
->fs_info
, "deleting default subvolume "
1863 "%llu is not allowed", key
.objectid
);
1866 btrfs_release_path(path
);
1869 key
.objectid
= root
->root_key
.objectid
;
1870 key
.type
= BTRFS_ROOT_REF_KEY
;
1871 key
.offset
= (u64
)-1;
1873 ret
= btrfs_search_slot(NULL
, root
->fs_info
->tree_root
,
1880 if (path
->slots
[0] > 0) {
1882 btrfs_item_key_to_cpu(path
->nodes
[0], &key
, path
->slots
[0]);
1883 if (key
.objectid
== root
->root_key
.objectid
&&
1884 key
.type
== BTRFS_ROOT_REF_KEY
)
1888 btrfs_free_path(path
);
1892 static noinline
int key_in_sk(struct btrfs_key
*key
,
1893 struct btrfs_ioctl_search_key
*sk
)
1895 struct btrfs_key test
;
1898 test
.objectid
= sk
->min_objectid
;
1899 test
.type
= sk
->min_type
;
1900 test
.offset
= sk
->min_offset
;
1902 ret
= btrfs_comp_cpu_keys(key
, &test
);
1906 test
.objectid
= sk
->max_objectid
;
1907 test
.type
= sk
->max_type
;
1908 test
.offset
= sk
->max_offset
;
1910 ret
= btrfs_comp_cpu_keys(key
, &test
);
1916 static noinline
int copy_to_sk(struct btrfs_root
*root
,
1917 struct btrfs_path
*path
,
1918 struct btrfs_key
*key
,
1919 struct btrfs_ioctl_search_key
*sk
,
1922 unsigned long *sk_offset
,
1926 struct extent_buffer
*leaf
;
1927 struct btrfs_ioctl_search_header sh
;
1928 unsigned long item_off
;
1929 unsigned long item_len
;
1935 leaf
= path
->nodes
[0];
1936 slot
= path
->slots
[0];
1937 nritems
= btrfs_header_nritems(leaf
);
1939 if (btrfs_header_generation(leaf
) > sk
->max_transid
) {
1943 found_transid
= btrfs_header_generation(leaf
);
1945 for (i
= slot
; i
< nritems
; i
++) {
1946 item_off
= btrfs_item_ptr_offset(leaf
, i
);
1947 item_len
= btrfs_item_size_nr(leaf
, i
);
1949 btrfs_item_key_to_cpu(leaf
, key
, i
);
1950 if (!key_in_sk(key
, sk
))
1953 if (sizeof(sh
) + item_len
> *buf_size
) {
1960 * return one empty item back for v1, which does not
1964 *buf_size
= sizeof(sh
) + item_len
;
1969 if (sizeof(sh
) + item_len
+ *sk_offset
> *buf_size
) {
1974 sh
.objectid
= key
->objectid
;
1975 sh
.offset
= key
->offset
;
1976 sh
.type
= key
->type
;
1978 sh
.transid
= found_transid
;
1980 /* copy search result header */
1981 if (copy_to_user(ubuf
+ *sk_offset
, &sh
, sizeof(sh
))) {
1986 *sk_offset
+= sizeof(sh
);
1989 char __user
*up
= ubuf
+ *sk_offset
;
1991 if (read_extent_buffer_to_user(leaf
, up
,
1992 item_off
, item_len
)) {
1997 *sk_offset
+= item_len
;
2001 if (ret
) /* -EOVERFLOW from above */
2004 if (*num_found
>= sk
->nr_items
) {
2011 if (key
->offset
< (u64
)-1 && key
->offset
< sk
->max_offset
)
2013 else if (key
->type
< (u8
)-1 && key
->type
< sk
->max_type
) {
2016 } else if (key
->objectid
< (u64
)-1 && key
->objectid
< sk
->max_objectid
) {
2024 * 0: all items from this leaf copied, continue with next
2025 * 1: * more items can be copied, but unused buffer is too small
2026 * * all items were found
2027 * Either way, it will stops the loop which iterates to the next
2029 * -EOVERFLOW: item was to large for buffer
2030 * -EFAULT: could not copy extent buffer back to userspace
2035 static noinline
int search_ioctl(struct inode
*inode
,
2036 struct btrfs_ioctl_search_key
*sk
,
2040 struct btrfs_root
*root
;
2041 struct btrfs_key key
;
2042 struct btrfs_path
*path
;
2043 struct btrfs_fs_info
*info
= BTRFS_I(inode
)->root
->fs_info
;
2046 unsigned long sk_offset
= 0;
2048 if (*buf_size
< sizeof(struct btrfs_ioctl_search_header
)) {
2049 *buf_size
= sizeof(struct btrfs_ioctl_search_header
);
2053 path
= btrfs_alloc_path();
2057 if (sk
->tree_id
== 0) {
2058 /* search the root of the inode that was passed */
2059 root
= BTRFS_I(inode
)->root
;
2061 key
.objectid
= sk
->tree_id
;
2062 key
.type
= BTRFS_ROOT_ITEM_KEY
;
2063 key
.offset
= (u64
)-1;
2064 root
= btrfs_read_fs_root_no_name(info
, &key
);
2066 printk(KERN_ERR
"BTRFS: could not find root %llu\n",
2068 btrfs_free_path(path
);
2073 key
.objectid
= sk
->min_objectid
;
2074 key
.type
= sk
->min_type
;
2075 key
.offset
= sk
->min_offset
;
2078 ret
= btrfs_search_forward(root
, &key
, path
, sk
->min_transid
);
2084 ret
= copy_to_sk(root
, path
, &key
, sk
, buf_size
, ubuf
,
2085 &sk_offset
, &num_found
);
2086 btrfs_release_path(path
);
2094 sk
->nr_items
= num_found
;
2095 btrfs_free_path(path
);
2099 static noinline
int btrfs_ioctl_tree_search(struct file
*file
,
2102 struct btrfs_ioctl_search_args __user
*uargs
;
2103 struct btrfs_ioctl_search_key sk
;
2104 struct inode
*inode
;
2108 if (!capable(CAP_SYS_ADMIN
))
2111 uargs
= (struct btrfs_ioctl_search_args __user
*)argp
;
2113 if (copy_from_user(&sk
, &uargs
->key
, sizeof(sk
)))
2116 buf_size
= sizeof(uargs
->buf
);
2118 inode
= file_inode(file
);
2119 ret
= search_ioctl(inode
, &sk
, &buf_size
, uargs
->buf
);
2122 * In the origin implementation an overflow is handled by returning a
2123 * search header with a len of zero, so reset ret.
2125 if (ret
== -EOVERFLOW
)
2128 if (ret
== 0 && copy_to_user(&uargs
->key
, &sk
, sizeof(sk
)))
2133 static noinline
int btrfs_ioctl_tree_search_v2(struct file
*file
,
2136 struct btrfs_ioctl_search_args_v2 __user
*uarg
;
2137 struct btrfs_ioctl_search_args_v2 args
;
2138 struct inode
*inode
;
2141 const size_t buf_limit
= 16 * 1024 * 1024;
2143 if (!capable(CAP_SYS_ADMIN
))
2146 /* copy search header and buffer size */
2147 uarg
= (struct btrfs_ioctl_search_args_v2 __user
*)argp
;
2148 if (copy_from_user(&args
, uarg
, sizeof(args
)))
2151 buf_size
= args
.buf_size
;
2153 if (buf_size
< sizeof(struct btrfs_ioctl_search_header
))
2156 /* limit result size to 16MB */
2157 if (buf_size
> buf_limit
)
2158 buf_size
= buf_limit
;
2160 inode
= file_inode(file
);
2161 ret
= search_ioctl(inode
, &args
.key
, &buf_size
,
2162 (char *)(&uarg
->buf
[0]));
2163 if (ret
== 0 && copy_to_user(&uarg
->key
, &args
.key
, sizeof(args
.key
)))
2165 else if (ret
== -EOVERFLOW
&&
2166 copy_to_user(&uarg
->buf_size
, &buf_size
, sizeof(buf_size
)))
2173 * Search INODE_REFs to identify path name of 'dirid' directory
2174 * in a 'tree_id' tree. and sets path name to 'name'.
2176 static noinline
int btrfs_search_path_in_tree(struct btrfs_fs_info
*info
,
2177 u64 tree_id
, u64 dirid
, char *name
)
2179 struct btrfs_root
*root
;
2180 struct btrfs_key key
;
2186 struct btrfs_inode_ref
*iref
;
2187 struct extent_buffer
*l
;
2188 struct btrfs_path
*path
;
2190 if (dirid
== BTRFS_FIRST_FREE_OBJECTID
) {
2195 path
= btrfs_alloc_path();
2199 ptr
= &name
[BTRFS_INO_LOOKUP_PATH_MAX
];
2201 key
.objectid
= tree_id
;
2202 key
.type
= BTRFS_ROOT_ITEM_KEY
;
2203 key
.offset
= (u64
)-1;
2204 root
= btrfs_read_fs_root_no_name(info
, &key
);
2206 printk(KERN_ERR
"BTRFS: could not find root %llu\n", tree_id
);
2211 key
.objectid
= dirid
;
2212 key
.type
= BTRFS_INODE_REF_KEY
;
2213 key
.offset
= (u64
)-1;
2216 ret
= btrfs_search_slot(NULL
, root
, &key
, path
, 0, 0);
2220 ret
= btrfs_previous_item(root
, path
, dirid
,
2221 BTRFS_INODE_REF_KEY
);
2231 slot
= path
->slots
[0];
2232 btrfs_item_key_to_cpu(l
, &key
, slot
);
2234 iref
= btrfs_item_ptr(l
, slot
, struct btrfs_inode_ref
);
2235 len
= btrfs_inode_ref_name_len(l
, iref
);
2237 total_len
+= len
+ 1;
2239 ret
= -ENAMETOOLONG
;
2244 read_extent_buffer(l
, ptr
, (unsigned long)(iref
+ 1), len
);
2246 if (key
.offset
== BTRFS_FIRST_FREE_OBJECTID
)
2249 btrfs_release_path(path
);
2250 key
.objectid
= key
.offset
;
2251 key
.offset
= (u64
)-1;
2252 dirid
= key
.objectid
;
2254 memmove(name
, ptr
, total_len
);
2255 name
[total_len
] = '\0';
2258 btrfs_free_path(path
);
2262 static noinline
int btrfs_ioctl_ino_lookup(struct file
*file
,
2265 struct btrfs_ioctl_ino_lookup_args
*args
;
2266 struct inode
*inode
;
2269 if (!capable(CAP_SYS_ADMIN
))
2272 args
= memdup_user(argp
, sizeof(*args
));
2274 return PTR_ERR(args
);
2276 inode
= file_inode(file
);
2278 if (args
->treeid
== 0)
2279 args
->treeid
= BTRFS_I(inode
)->root
->root_key
.objectid
;
2281 ret
= btrfs_search_path_in_tree(BTRFS_I(inode
)->root
->fs_info
,
2282 args
->treeid
, args
->objectid
,
2285 if (ret
== 0 && copy_to_user(argp
, args
, sizeof(*args
)))
2292 static noinline
int btrfs_ioctl_snap_destroy(struct file
*file
,
2295 struct dentry
*parent
= file
->f_path
.dentry
;
2296 struct dentry
*dentry
;
2297 struct inode
*dir
= parent
->d_inode
;
2298 struct inode
*inode
;
2299 struct btrfs_root
*root
= BTRFS_I(dir
)->root
;
2300 struct btrfs_root
*dest
= NULL
;
2301 struct btrfs_ioctl_vol_args
*vol_args
;
2302 struct btrfs_trans_handle
*trans
;
2303 struct btrfs_block_rsv block_rsv
;
2305 u64 qgroup_reserved
;
2310 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
2311 if (IS_ERR(vol_args
))
2312 return PTR_ERR(vol_args
);
2314 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
2315 namelen
= strlen(vol_args
->name
);
2316 if (strchr(vol_args
->name
, '/') ||
2317 strncmp(vol_args
->name
, "..", namelen
) == 0) {
2322 err
= mnt_want_write_file(file
);
2327 err
= mutex_lock_killable_nested(&dir
->i_mutex
, I_MUTEX_PARENT
);
2329 goto out_drop_write
;
2330 dentry
= lookup_one_len(vol_args
->name
, parent
, namelen
);
2331 if (IS_ERR(dentry
)) {
2332 err
= PTR_ERR(dentry
);
2333 goto out_unlock_dir
;
2336 if (!dentry
->d_inode
) {
2341 inode
= dentry
->d_inode
;
2342 dest
= BTRFS_I(inode
)->root
;
2343 if (!capable(CAP_SYS_ADMIN
)) {
2345 * Regular user. Only allow this with a special mount
2346 * option, when the user has write+exec access to the
2347 * subvol root, and when rmdir(2) would have been
2350 * Note that this is _not_ check that the subvol is
2351 * empty or doesn't contain data that we wouldn't
2352 * otherwise be able to delete.
2354 * Users who want to delete empty subvols should try
2358 if (!btrfs_test_opt(root
, USER_SUBVOL_RM_ALLOWED
))
2362 * Do not allow deletion if the parent dir is the same
2363 * as the dir to be deleted. That means the ioctl
2364 * must be called on the dentry referencing the root
2365 * of the subvol, not a random directory contained
2372 err
= inode_permission(inode
, MAY_WRITE
| MAY_EXEC
);
2377 /* check if subvolume may be deleted by a user */
2378 err
= btrfs_may_delete(dir
, dentry
, 1);
2382 if (btrfs_ino(inode
) != BTRFS_FIRST_FREE_OBJECTID
) {
2387 mutex_lock(&inode
->i_mutex
);
2390 * Don't allow to delete a subvolume with send in progress. This is
2391 * inside the i_mutex so the error handling that has to drop the bit
2392 * again is not run concurrently.
2394 spin_lock(&dest
->root_item_lock
);
2395 root_flags
= btrfs_root_flags(&dest
->root_item
);
2396 if (dest
->send_in_progress
== 0) {
2397 btrfs_set_root_flags(&dest
->root_item
,
2398 root_flags
| BTRFS_ROOT_SUBVOL_DEAD
);
2399 spin_unlock(&dest
->root_item_lock
);
2401 spin_unlock(&dest
->root_item_lock
);
2402 btrfs_warn(root
->fs_info
,
2403 "Attempt to delete subvolume %llu during send",
2404 dest
->root_key
.objectid
);
2409 d_invalidate(dentry
);
2411 down_write(&root
->fs_info
->subvol_sem
);
2413 err
= may_destroy_subvol(dest
);
2417 btrfs_init_block_rsv(&block_rsv
, BTRFS_BLOCK_RSV_TEMP
);
2419 * One for dir inode, two for dir entries, two for root
2422 err
= btrfs_subvolume_reserve_metadata(root
, &block_rsv
,
2423 5, &qgroup_reserved
, true);
2427 trans
= btrfs_start_transaction(root
, 0);
2428 if (IS_ERR(trans
)) {
2429 err
= PTR_ERR(trans
);
2432 trans
->block_rsv
= &block_rsv
;
2433 trans
->bytes_reserved
= block_rsv
.size
;
2435 ret
= btrfs_unlink_subvol(trans
, root
, dir
,
2436 dest
->root_key
.objectid
,
2437 dentry
->d_name
.name
,
2438 dentry
->d_name
.len
);
2441 btrfs_abort_transaction(trans
, root
, ret
);
2445 btrfs_record_root_in_trans(trans
, dest
);
2447 memset(&dest
->root_item
.drop_progress
, 0,
2448 sizeof(dest
->root_item
.drop_progress
));
2449 dest
->root_item
.drop_level
= 0;
2450 btrfs_set_root_refs(&dest
->root_item
, 0);
2452 if (!test_and_set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED
, &dest
->state
)) {
2453 ret
= btrfs_insert_orphan_item(trans
,
2454 root
->fs_info
->tree_root
,
2455 dest
->root_key
.objectid
);
2457 btrfs_abort_transaction(trans
, root
, ret
);
2463 ret
= btrfs_uuid_tree_rem(trans
, root
->fs_info
->uuid_root
,
2464 dest
->root_item
.uuid
, BTRFS_UUID_KEY_SUBVOL
,
2465 dest
->root_key
.objectid
);
2466 if (ret
&& ret
!= -ENOENT
) {
2467 btrfs_abort_transaction(trans
, root
, ret
);
2471 if (!btrfs_is_empty_uuid(dest
->root_item
.received_uuid
)) {
2472 ret
= btrfs_uuid_tree_rem(trans
, root
->fs_info
->uuid_root
,
2473 dest
->root_item
.received_uuid
,
2474 BTRFS_UUID_KEY_RECEIVED_SUBVOL
,
2475 dest
->root_key
.objectid
);
2476 if (ret
&& ret
!= -ENOENT
) {
2477 btrfs_abort_transaction(trans
, root
, ret
);
2484 trans
->block_rsv
= NULL
;
2485 trans
->bytes_reserved
= 0;
2486 ret
= btrfs_end_transaction(trans
, root
);
2489 inode
->i_flags
|= S_DEAD
;
2491 btrfs_subvolume_release_metadata(root
, &block_rsv
, qgroup_reserved
);
2493 up_write(&root
->fs_info
->subvol_sem
);
2495 spin_lock(&dest
->root_item_lock
);
2496 root_flags
= btrfs_root_flags(&dest
->root_item
);
2497 btrfs_set_root_flags(&dest
->root_item
,
2498 root_flags
& ~BTRFS_ROOT_SUBVOL_DEAD
);
2499 spin_unlock(&dest
->root_item_lock
);
2501 mutex_unlock(&inode
->i_mutex
);
2503 shrink_dcache_sb(root
->fs_info
->sb
);
2504 btrfs_invalidate_inodes(dest
);
2506 ASSERT(dest
->send_in_progress
== 0);
2509 if (dest
->ino_cache_inode
) {
2510 iput(dest
->ino_cache_inode
);
2511 dest
->ino_cache_inode
= NULL
;
2517 mutex_unlock(&dir
->i_mutex
);
2519 mnt_drop_write_file(file
);
2525 static int btrfs_ioctl_defrag(struct file
*file
, void __user
*argp
)
2527 struct inode
*inode
= file_inode(file
);
2528 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
2529 struct btrfs_ioctl_defrag_range_args
*range
;
2532 ret
= mnt_want_write_file(file
);
2536 if (btrfs_root_readonly(root
)) {
2541 switch (inode
->i_mode
& S_IFMT
) {
2543 if (!capable(CAP_SYS_ADMIN
)) {
2547 ret
= btrfs_defrag_root(root
);
2550 ret
= btrfs_defrag_root(root
->fs_info
->extent_root
);
2553 if (!(file
->f_mode
& FMODE_WRITE
)) {
2558 range
= kzalloc(sizeof(*range
), GFP_KERNEL
);
2565 if (copy_from_user(range
, argp
,
2571 /* compression requires us to start the IO */
2572 if ((range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
)) {
2573 range
->flags
|= BTRFS_DEFRAG_RANGE_START_IO
;
2574 range
->extent_thresh
= (u32
)-1;
2577 /* the rest are all set to zero by kzalloc */
2578 range
->len
= (u64
)-1;
2580 ret
= btrfs_defrag_file(file_inode(file
), file
,
2590 mnt_drop_write_file(file
);
2594 static long btrfs_ioctl_add_dev(struct btrfs_root
*root
, void __user
*arg
)
2596 struct btrfs_ioctl_vol_args
*vol_args
;
2599 if (!capable(CAP_SYS_ADMIN
))
2602 if (atomic_xchg(&root
->fs_info
->mutually_exclusive_operation_running
,
2604 return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS
;
2607 mutex_lock(&root
->fs_info
->volume_mutex
);
2608 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
2609 if (IS_ERR(vol_args
)) {
2610 ret
= PTR_ERR(vol_args
);
2614 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
2615 ret
= btrfs_init_new_device(root
, vol_args
->name
);
2618 btrfs_info(root
->fs_info
, "disk added %s",vol_args
->name
);
2622 mutex_unlock(&root
->fs_info
->volume_mutex
);
2623 atomic_set(&root
->fs_info
->mutually_exclusive_operation_running
, 0);
2627 static long btrfs_ioctl_rm_dev(struct file
*file
, void __user
*arg
)
2629 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
2630 struct btrfs_ioctl_vol_args
*vol_args
;
2633 if (!capable(CAP_SYS_ADMIN
))
2636 ret
= mnt_want_write_file(file
);
2640 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
2641 if (IS_ERR(vol_args
)) {
2642 ret
= PTR_ERR(vol_args
);
2646 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
2648 if (atomic_xchg(&root
->fs_info
->mutually_exclusive_operation_running
,
2650 ret
= BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS
;
2654 mutex_lock(&root
->fs_info
->volume_mutex
);
2655 ret
= btrfs_rm_device(root
, vol_args
->name
);
2656 mutex_unlock(&root
->fs_info
->volume_mutex
);
2657 atomic_set(&root
->fs_info
->mutually_exclusive_operation_running
, 0);
2660 btrfs_info(root
->fs_info
, "disk deleted %s",vol_args
->name
);
2665 mnt_drop_write_file(file
);
2669 static long btrfs_ioctl_fs_info(struct btrfs_root
*root
, void __user
*arg
)
2671 struct btrfs_ioctl_fs_info_args
*fi_args
;
2672 struct btrfs_device
*device
;
2673 struct btrfs_device
*next
;
2674 struct btrfs_fs_devices
*fs_devices
= root
->fs_info
->fs_devices
;
2677 fi_args
= kzalloc(sizeof(*fi_args
), GFP_KERNEL
);
2681 mutex_lock(&fs_devices
->device_list_mutex
);
2682 fi_args
->num_devices
= fs_devices
->num_devices
;
2683 memcpy(&fi_args
->fsid
, root
->fs_info
->fsid
, sizeof(fi_args
->fsid
));
2685 list_for_each_entry_safe(device
, next
, &fs_devices
->devices
, dev_list
) {
2686 if (device
->devid
> fi_args
->max_id
)
2687 fi_args
->max_id
= device
->devid
;
2689 mutex_unlock(&fs_devices
->device_list_mutex
);
2691 fi_args
->nodesize
= root
->fs_info
->super_copy
->nodesize
;
2692 fi_args
->sectorsize
= root
->fs_info
->super_copy
->sectorsize
;
2693 fi_args
->clone_alignment
= root
->fs_info
->super_copy
->sectorsize
;
2695 if (copy_to_user(arg
, fi_args
, sizeof(*fi_args
)))
2702 static long btrfs_ioctl_dev_info(struct btrfs_root
*root
, void __user
*arg
)
2704 struct btrfs_ioctl_dev_info_args
*di_args
;
2705 struct btrfs_device
*dev
;
2706 struct btrfs_fs_devices
*fs_devices
= root
->fs_info
->fs_devices
;
2708 char *s_uuid
= NULL
;
2710 di_args
= memdup_user(arg
, sizeof(*di_args
));
2711 if (IS_ERR(di_args
))
2712 return PTR_ERR(di_args
);
2714 if (!btrfs_is_empty_uuid(di_args
->uuid
))
2715 s_uuid
= di_args
->uuid
;
2717 mutex_lock(&fs_devices
->device_list_mutex
);
2718 dev
= btrfs_find_device(root
->fs_info
, di_args
->devid
, s_uuid
, NULL
);
2725 di_args
->devid
= dev
->devid
;
2726 di_args
->bytes_used
= btrfs_device_get_bytes_used(dev
);
2727 di_args
->total_bytes
= btrfs_device_get_total_bytes(dev
);
2728 memcpy(di_args
->uuid
, dev
->uuid
, sizeof(di_args
->uuid
));
2730 struct rcu_string
*name
;
2733 name
= rcu_dereference(dev
->name
);
2734 strncpy(di_args
->path
, name
->str
, sizeof(di_args
->path
));
2736 di_args
->path
[sizeof(di_args
->path
) - 1] = 0;
2738 di_args
->path
[0] = '\0';
2742 mutex_unlock(&fs_devices
->device_list_mutex
);
2743 if (ret
== 0 && copy_to_user(arg
, di_args
, sizeof(*di_args
)))
2750 static struct page
*extent_same_get_page(struct inode
*inode
, u64 off
)
2754 struct extent_io_tree
*tree
= &BTRFS_I(inode
)->io_tree
;
2756 index
= off
>> PAGE_CACHE_SHIFT
;
2758 page
= grab_cache_page(inode
->i_mapping
, index
);
2762 if (!PageUptodate(page
)) {
2763 if (extent_read_full_page_nolock(tree
, page
, btrfs_get_extent
,
2767 if (!PageUptodate(page
)) {
2769 page_cache_release(page
);
2778 static inline void lock_extent_range(struct inode
*inode
, u64 off
, u64 len
)
2780 /* do any pending delalloc/csum calc on src, one way or
2781 another, and lock file content */
2783 struct btrfs_ordered_extent
*ordered
;
2784 lock_extent(&BTRFS_I(inode
)->io_tree
, off
, off
+ len
- 1);
2785 ordered
= btrfs_lookup_first_ordered_extent(inode
,
2788 ordered
->file_offset
+ ordered
->len
<= off
||
2789 ordered
->file_offset
>= off
+ len
) &&
2790 !test_range_bit(&BTRFS_I(inode
)->io_tree
, off
,
2791 off
+ len
- 1, EXTENT_DELALLOC
, 0, NULL
)) {
2793 btrfs_put_ordered_extent(ordered
);
2796 unlock_extent(&BTRFS_I(inode
)->io_tree
, off
, off
+ len
- 1);
2798 btrfs_put_ordered_extent(ordered
);
2799 btrfs_wait_ordered_range(inode
, off
, len
);
2803 static void btrfs_double_unlock(struct inode
*inode1
, u64 loff1
,
2804 struct inode
*inode2
, u64 loff2
, u64 len
)
2806 unlock_extent(&BTRFS_I(inode1
)->io_tree
, loff1
, loff1
+ len
- 1);
2807 unlock_extent(&BTRFS_I(inode2
)->io_tree
, loff2
, loff2
+ len
- 1);
2809 mutex_unlock(&inode1
->i_mutex
);
2810 mutex_unlock(&inode2
->i_mutex
);
2813 static void btrfs_double_lock(struct inode
*inode1
, u64 loff1
,
2814 struct inode
*inode2
, u64 loff2
, u64 len
)
2816 if (inode1
< inode2
) {
2817 swap(inode1
, inode2
);
2821 mutex_lock_nested(&inode1
->i_mutex
, I_MUTEX_PARENT
);
2822 lock_extent_range(inode1
, loff1
, len
);
2823 if (inode1
!= inode2
) {
2824 mutex_lock_nested(&inode2
->i_mutex
, I_MUTEX_CHILD
);
2825 lock_extent_range(inode2
, loff2
, len
);
2829 static int btrfs_cmp_data(struct inode
*src
, u64 loff
, struct inode
*dst
,
2830 u64 dst_loff
, u64 len
)
2833 struct page
*src_page
, *dst_page
;
2834 unsigned int cmp_len
= PAGE_CACHE_SIZE
;
2835 void *addr
, *dst_addr
;
2838 if (len
< PAGE_CACHE_SIZE
)
2841 src_page
= extent_same_get_page(src
, loff
);
2844 dst_page
= extent_same_get_page(dst
, dst_loff
);
2846 page_cache_release(src_page
);
2849 addr
= kmap_atomic(src_page
);
2850 dst_addr
= kmap_atomic(dst_page
);
2852 flush_dcache_page(src_page
);
2853 flush_dcache_page(dst_page
);
2855 if (memcmp(addr
, dst_addr
, cmp_len
))
2856 ret
= BTRFS_SAME_DATA_DIFFERS
;
2858 kunmap_atomic(addr
);
2859 kunmap_atomic(dst_addr
);
2860 page_cache_release(src_page
);
2861 page_cache_release(dst_page
);
2867 dst_loff
+= cmp_len
;
2874 static int extent_same_check_offsets(struct inode
*inode
, u64 off
, u64 len
)
2876 u64 bs
= BTRFS_I(inode
)->root
->fs_info
->sb
->s_blocksize
;
2878 if (off
+ len
> inode
->i_size
|| off
+ len
< off
)
2880 /* Check that we are block aligned - btrfs_clone() requires this */
2881 if (!IS_ALIGNED(off
, bs
) || !IS_ALIGNED(off
+ len
, bs
))
2887 static int btrfs_extent_same(struct inode
*src
, u64 loff
, u64 len
,
2888 struct inode
*dst
, u64 dst_loff
)
2893 * btrfs_clone() can't handle extents in the same file
2894 * yet. Once that works, we can drop this check and replace it
2895 * with a check for the same inode, but overlapping extents.
2900 btrfs_double_lock(src
, loff
, dst
, dst_loff
, len
);
2902 ret
= extent_same_check_offsets(src
, loff
, len
);
2906 ret
= extent_same_check_offsets(dst
, dst_loff
, len
);
2910 /* don't make the dst file partly checksummed */
2911 if ((BTRFS_I(src
)->flags
& BTRFS_INODE_NODATASUM
) !=
2912 (BTRFS_I(dst
)->flags
& BTRFS_INODE_NODATASUM
)) {
2917 ret
= btrfs_cmp_data(src
, loff
, dst
, dst_loff
, len
);
2919 ret
= btrfs_clone(src
, dst
, loff
, len
, len
, dst_loff
);
2922 btrfs_double_unlock(src
, loff
, dst
, dst_loff
, len
);
2927 #define BTRFS_MAX_DEDUPE_LEN (16 * 1024 * 1024)
2929 static long btrfs_ioctl_file_extent_same(struct file
*file
,
2930 struct btrfs_ioctl_same_args __user
*argp
)
2932 struct btrfs_ioctl_same_args
*same
;
2933 struct btrfs_ioctl_same_extent_info
*info
;
2934 struct inode
*src
= file_inode(file
);
2940 u64 bs
= BTRFS_I(src
)->root
->fs_info
->sb
->s_blocksize
;
2941 bool is_admin
= capable(CAP_SYS_ADMIN
);
2944 if (!(file
->f_mode
& FMODE_READ
))
2947 ret
= mnt_want_write_file(file
);
2951 if (get_user(count
, &argp
->dest_count
)) {
2956 size
= offsetof(struct btrfs_ioctl_same_args __user
, info
[count
]);
2958 same
= memdup_user(argp
, size
);
2961 ret
= PTR_ERR(same
);
2965 off
= same
->logical_offset
;
2969 * Limit the total length we will dedupe for each operation.
2970 * This is intended to bound the total time spent in this
2971 * ioctl to something sane.
2973 if (len
> BTRFS_MAX_DEDUPE_LEN
)
2974 len
= BTRFS_MAX_DEDUPE_LEN
;
2976 if (WARN_ON_ONCE(bs
< PAGE_CACHE_SIZE
)) {
2978 * Btrfs does not support blocksize < page_size. As a
2979 * result, btrfs_cmp_data() won't correctly handle
2980 * this situation without an update.
2987 if (S_ISDIR(src
->i_mode
))
2991 if (!S_ISREG(src
->i_mode
))
2994 /* pre-format output fields to sane values */
2995 for (i
= 0; i
< count
; i
++) {
2996 same
->info
[i
].bytes_deduped
= 0ULL;
2997 same
->info
[i
].status
= 0;
3000 for (i
= 0, info
= same
->info
; i
< count
; i
++, info
++) {
3002 struct fd dst_file
= fdget(info
->fd
);
3003 if (!dst_file
.file
) {
3004 info
->status
= -EBADF
;
3007 dst
= file_inode(dst_file
.file
);
3009 if (!(is_admin
|| (dst_file
.file
->f_mode
& FMODE_WRITE
))) {
3010 info
->status
= -EINVAL
;
3011 } else if (file
->f_path
.mnt
!= dst_file
.file
->f_path
.mnt
) {
3012 info
->status
= -EXDEV
;
3013 } else if (S_ISDIR(dst
->i_mode
)) {
3014 info
->status
= -EISDIR
;
3015 } else if (!S_ISREG(dst
->i_mode
)) {
3016 info
->status
= -EACCES
;
3018 info
->status
= btrfs_extent_same(src
, off
, len
, dst
,
3019 info
->logical_offset
);
3020 if (info
->status
== 0)
3021 info
->bytes_deduped
+= len
;
3026 ret
= copy_to_user(argp
, same
, size
);
3031 mnt_drop_write_file(file
);
3035 /* Helper to check and see if this root currently has a ref on the given disk
3036 * bytenr. If it does then we need to update the quota for this root. This
3037 * doesn't do anything if quotas aren't enabled.
3039 static int check_ref(struct btrfs_trans_handle
*trans
, struct btrfs_root
*root
,
3042 struct seq_list tree_mod_seq_elem
= {};
3043 struct ulist
*roots
;
3044 struct ulist_iterator uiter
;
3045 struct ulist_node
*root_node
= NULL
;
3048 if (!root
->fs_info
->quota_enabled
)
3051 btrfs_get_tree_mod_seq(root
->fs_info
, &tree_mod_seq_elem
);
3052 ret
= btrfs_find_all_roots(trans
, root
->fs_info
, disko
,
3053 tree_mod_seq_elem
.seq
, &roots
);
3057 ULIST_ITER_INIT(&uiter
);
3058 while ((root_node
= ulist_next(roots
, &uiter
))) {
3059 if (root_node
->val
== root
->objectid
) {
3066 btrfs_put_tree_mod_seq(root
->fs_info
, &tree_mod_seq_elem
);
3070 static int clone_finish_inode_update(struct btrfs_trans_handle
*trans
,
3071 struct inode
*inode
,
3076 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
3079 inode_inc_iversion(inode
);
3080 inode
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
3082 * We round up to the block size at eof when determining which
3083 * extents to clone above, but shouldn't round up the file size.
3085 if (endoff
> destoff
+ olen
)
3086 endoff
= destoff
+ olen
;
3087 if (endoff
> inode
->i_size
)
3088 btrfs_i_size_write(inode
, endoff
);
3090 ret
= btrfs_update_inode(trans
, root
, inode
);
3092 btrfs_abort_transaction(trans
, root
, ret
);
3093 btrfs_end_transaction(trans
, root
);
3096 ret
= btrfs_end_transaction(trans
, root
);
3101 static void clone_update_extent_map(struct inode
*inode
,
3102 const struct btrfs_trans_handle
*trans
,
3103 const struct btrfs_path
*path
,
3104 const u64 hole_offset
,
3107 struct extent_map_tree
*em_tree
= &BTRFS_I(inode
)->extent_tree
;
3108 struct extent_map
*em
;
3111 em
= alloc_extent_map();
3113 set_bit(BTRFS_INODE_NEEDS_FULL_SYNC
,
3114 &BTRFS_I(inode
)->runtime_flags
);
3119 struct btrfs_file_extent_item
*fi
;
3121 fi
= btrfs_item_ptr(path
->nodes
[0], path
->slots
[0],
3122 struct btrfs_file_extent_item
);
3123 btrfs_extent_item_to_extent_map(inode
, path
, fi
, false, em
);
3124 em
->generation
= -1;
3125 if (btrfs_file_extent_type(path
->nodes
[0], fi
) ==
3126 BTRFS_FILE_EXTENT_INLINE
)
3127 set_bit(BTRFS_INODE_NEEDS_FULL_SYNC
,
3128 &BTRFS_I(inode
)->runtime_flags
);
3130 em
->start
= hole_offset
;
3132 em
->ram_bytes
= em
->len
;
3133 em
->orig_start
= hole_offset
;
3134 em
->block_start
= EXTENT_MAP_HOLE
;
3136 em
->orig_block_len
= 0;
3137 em
->compress_type
= BTRFS_COMPRESS_NONE
;
3138 em
->generation
= trans
->transid
;
3142 write_lock(&em_tree
->lock
);
3143 ret
= add_extent_mapping(em_tree
, em
, 1);
3144 write_unlock(&em_tree
->lock
);
3145 if (ret
!= -EEXIST
) {
3146 free_extent_map(em
);
3149 btrfs_drop_extent_cache(inode
, em
->start
,
3150 em
->start
+ em
->len
- 1, 0);
3154 set_bit(BTRFS_INODE_NEEDS_FULL_SYNC
,
3155 &BTRFS_I(inode
)->runtime_flags
);
3159 * btrfs_clone() - clone a range from inode file to another
3161 * @src: Inode to clone from
3162 * @inode: Inode to clone to
3163 * @off: Offset within source to start clone from
3164 * @olen: Original length, passed by user, of range to clone
3165 * @olen_aligned: Block-aligned value of olen, extent_same uses
3166 * identical values here
3167 * @destoff: Offset within @inode to start clone
3169 static int btrfs_clone(struct inode
*src
, struct inode
*inode
,
3170 const u64 off
, const u64 olen
, const u64 olen_aligned
,
3173 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
3174 struct btrfs_path
*path
= NULL
;
3175 struct extent_buffer
*leaf
;
3176 struct btrfs_trans_handle
*trans
;
3178 struct btrfs_key key
;
3183 const u64 len
= olen_aligned
;
3185 u64 last_dest_end
= destoff
;
3188 buf
= vmalloc(root
->nodesize
);
3192 path
= btrfs_alloc_path();
3200 key
.objectid
= btrfs_ino(src
);
3201 key
.type
= BTRFS_EXTENT_DATA_KEY
;
3206 * note the key will change type as we walk through the
3209 path
->leave_spinning
= 1;
3210 ret
= btrfs_search_slot(NULL
, BTRFS_I(src
)->root
, &key
, path
,
3215 * First search, if no extent item that starts at offset off was
3216 * found but the previous item is an extent item, it's possible
3217 * it might overlap our target range, therefore process it.
3219 if (key
.offset
== off
&& ret
> 0 && path
->slots
[0] > 0) {
3220 btrfs_item_key_to_cpu(path
->nodes
[0], &key
,
3221 path
->slots
[0] - 1);
3222 if (key
.type
== BTRFS_EXTENT_DATA_KEY
)
3226 nritems
= btrfs_header_nritems(path
->nodes
[0]);
3229 if (path
->slots
[0] >= nritems
) {
3230 ret
= btrfs_next_leaf(BTRFS_I(src
)->root
, path
);
3235 nritems
= btrfs_header_nritems(path
->nodes
[0]);
3237 leaf
= path
->nodes
[0];
3238 slot
= path
->slots
[0];
3240 btrfs_item_key_to_cpu(leaf
, &key
, slot
);
3241 if (key
.type
> BTRFS_EXTENT_DATA_KEY
||
3242 key
.objectid
!= btrfs_ino(src
))
3245 if (key
.type
== BTRFS_EXTENT_DATA_KEY
) {
3246 struct btrfs_file_extent_item
*extent
;
3249 struct btrfs_key new_key
;
3250 u64 disko
= 0, diskl
= 0;
3251 u64 datao
= 0, datal
= 0;
3255 extent
= btrfs_item_ptr(leaf
, slot
,
3256 struct btrfs_file_extent_item
);
3257 comp
= btrfs_file_extent_compression(leaf
, extent
);
3258 type
= btrfs_file_extent_type(leaf
, extent
);
3259 if (type
== BTRFS_FILE_EXTENT_REG
||
3260 type
== BTRFS_FILE_EXTENT_PREALLOC
) {
3261 disko
= btrfs_file_extent_disk_bytenr(leaf
,
3263 diskl
= btrfs_file_extent_disk_num_bytes(leaf
,
3265 datao
= btrfs_file_extent_offset(leaf
, extent
);
3266 datal
= btrfs_file_extent_num_bytes(leaf
,
3268 } else if (type
== BTRFS_FILE_EXTENT_INLINE
) {
3269 /* take upper bound, may be compressed */
3270 datal
= btrfs_file_extent_ram_bytes(leaf
,
3275 * The first search might have left us at an extent
3276 * item that ends before our target range's start, can
3277 * happen if we have holes and NO_HOLES feature enabled.
3279 if (key
.offset
+ datal
<= off
) {
3282 } else if (key
.offset
>= off
+ len
) {
3286 size
= btrfs_item_size_nr(leaf
, slot
);
3287 read_extent_buffer(leaf
, buf
,
3288 btrfs_item_ptr_offset(leaf
, slot
),
3291 btrfs_release_path(path
);
3292 path
->leave_spinning
= 0;
3294 memcpy(&new_key
, &key
, sizeof(new_key
));
3295 new_key
.objectid
= btrfs_ino(inode
);
3296 if (off
<= key
.offset
)
3297 new_key
.offset
= key
.offset
+ destoff
- off
;
3299 new_key
.offset
= destoff
;
3302 * Deal with a hole that doesn't have an extent item
3303 * that represents it (NO_HOLES feature enabled).
3304 * This hole is either in the middle of the cloning
3305 * range or at the beginning (fully overlaps it or
3306 * partially overlaps it).
3308 if (new_key
.offset
!= last_dest_end
)
3309 drop_start
= last_dest_end
;
3311 drop_start
= new_key
.offset
;
3314 * 1 - adjusting old extent (we may have to split it)
3315 * 1 - add new extent
3318 trans
= btrfs_start_transaction(root
, 3);
3319 if (IS_ERR(trans
)) {
3320 ret
= PTR_ERR(trans
);
3324 if (type
== BTRFS_FILE_EXTENT_REG
||
3325 type
== BTRFS_FILE_EXTENT_PREALLOC
) {
3327 * a | --- range to clone ---| b
3328 * | ------------- extent ------------- |
3331 /* subtract range b */
3332 if (key
.offset
+ datal
> off
+ len
)
3333 datal
= off
+ len
- key
.offset
;
3335 /* subtract range a */
3336 if (off
> key
.offset
) {
3337 datao
+= off
- key
.offset
;
3338 datal
-= off
- key
.offset
;
3341 ret
= btrfs_drop_extents(trans
, root
, inode
,
3343 new_key
.offset
+ datal
,
3346 if (ret
!= -EOPNOTSUPP
)
3347 btrfs_abort_transaction(trans
,
3349 btrfs_end_transaction(trans
, root
);
3353 ret
= btrfs_insert_empty_item(trans
, root
, path
,
3356 btrfs_abort_transaction(trans
, root
,
3358 btrfs_end_transaction(trans
, root
);
3362 leaf
= path
->nodes
[0];
3363 slot
= path
->slots
[0];
3364 write_extent_buffer(leaf
, buf
,
3365 btrfs_item_ptr_offset(leaf
, slot
),
3368 extent
= btrfs_item_ptr(leaf
, slot
,
3369 struct btrfs_file_extent_item
);
3371 /* disko == 0 means it's a hole */
3375 btrfs_set_file_extent_offset(leaf
, extent
,
3377 btrfs_set_file_extent_num_bytes(leaf
, extent
,
3381 * We need to look up the roots that point at
3382 * this bytenr and see if the new root does. If
3383 * it does not we need to make sure we update
3384 * quotas appropriately.
3386 if (disko
&& root
!= BTRFS_I(src
)->root
&&
3387 disko
!= last_disko
) {
3388 no_quota
= check_ref(trans
, root
,
3391 btrfs_abort_transaction(trans
,
3394 btrfs_end_transaction(trans
,
3402 inode_add_bytes(inode
, datal
);
3403 ret
= btrfs_inc_extent_ref(trans
, root
,
3405 root
->root_key
.objectid
,
3407 new_key
.offset
- datao
,
3410 btrfs_abort_transaction(trans
,
3413 btrfs_end_transaction(trans
,
3419 } else if (type
== BTRFS_FILE_EXTENT_INLINE
) {
3422 u64 aligned_end
= 0;
3424 if (off
> key
.offset
) {
3425 skip
= off
- key
.offset
;
3426 new_key
.offset
+= skip
;
3429 if (key
.offset
+ datal
> off
+ len
)
3430 trim
= key
.offset
+ datal
- (off
+ len
);
3432 if (comp
&& (skip
|| trim
)) {
3434 btrfs_end_transaction(trans
, root
);
3437 size
-= skip
+ trim
;
3438 datal
-= skip
+ trim
;
3440 aligned_end
= ALIGN(new_key
.offset
+ datal
,
3442 ret
= btrfs_drop_extents(trans
, root
, inode
,
3447 if (ret
!= -EOPNOTSUPP
)
3448 btrfs_abort_transaction(trans
,
3450 btrfs_end_transaction(trans
, root
);
3454 ret
= btrfs_insert_empty_item(trans
, root
, path
,
3457 btrfs_abort_transaction(trans
, root
,
3459 btrfs_end_transaction(trans
, root
);
3465 btrfs_file_extent_calc_inline_size(0);
3466 memmove(buf
+start
, buf
+start
+skip
,
3470 leaf
= path
->nodes
[0];
3471 slot
= path
->slots
[0];
3472 write_extent_buffer(leaf
, buf
,
3473 btrfs_item_ptr_offset(leaf
, slot
),
3475 inode_add_bytes(inode
, datal
);
3478 /* If we have an implicit hole (NO_HOLES feature). */
3479 if (drop_start
< new_key
.offset
)
3480 clone_update_extent_map(inode
, trans
,
3482 new_key
.offset
- drop_start
);
3484 clone_update_extent_map(inode
, trans
, path
, 0, 0);
3486 btrfs_mark_buffer_dirty(leaf
);
3487 btrfs_release_path(path
);
3489 last_dest_end
= ALIGN(new_key
.offset
+ datal
,
3491 ret
= clone_finish_inode_update(trans
, inode
,
3496 if (new_key
.offset
+ datal
>= destoff
+ len
)
3499 btrfs_release_path(path
);
3504 if (last_dest_end
< destoff
+ len
) {
3506 * We have an implicit hole (NO_HOLES feature is enabled) that
3507 * fully or partially overlaps our cloning range at its end.
3509 btrfs_release_path(path
);
3512 * 1 - remove extent(s)
3515 trans
= btrfs_start_transaction(root
, 2);
3516 if (IS_ERR(trans
)) {
3517 ret
= PTR_ERR(trans
);
3520 ret
= btrfs_drop_extents(trans
, root
, inode
,
3521 last_dest_end
, destoff
+ len
, 1);
3523 if (ret
!= -EOPNOTSUPP
)
3524 btrfs_abort_transaction(trans
, root
, ret
);
3525 btrfs_end_transaction(trans
, root
);
3528 clone_update_extent_map(inode
, trans
, NULL
, last_dest_end
,
3529 destoff
+ len
- last_dest_end
);
3530 ret
= clone_finish_inode_update(trans
, inode
, destoff
+ len
,
3535 btrfs_free_path(path
);
3540 static noinline
long btrfs_ioctl_clone(struct file
*file
, unsigned long srcfd
,
3541 u64 off
, u64 olen
, u64 destoff
)
3543 struct inode
*inode
= file_inode(file
);
3544 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
3549 u64 bs
= root
->fs_info
->sb
->s_blocksize
;
3554 * - split compressed inline extents. annoying: we need to
3555 * decompress into destination's address_space (the file offset
3556 * may change, so source mapping won't do), then recompress (or
3557 * otherwise reinsert) a subrange.
3559 * - split destination inode's inline extents. The inline extents can
3560 * be either compressed or non-compressed.
3563 /* the destination must be opened for writing */
3564 if (!(file
->f_mode
& FMODE_WRITE
) || (file
->f_flags
& O_APPEND
))
3567 if (btrfs_root_readonly(root
))
3570 ret
= mnt_want_write_file(file
);
3574 src_file
= fdget(srcfd
);
3575 if (!src_file
.file
) {
3577 goto out_drop_write
;
3581 if (src_file
.file
->f_path
.mnt
!= file
->f_path
.mnt
)
3584 src
= file_inode(src_file
.file
);
3590 /* the src must be open for reading */
3591 if (!(src_file
.file
->f_mode
& FMODE_READ
))
3594 /* don't make the dst file partly checksummed */
3595 if ((BTRFS_I(src
)->flags
& BTRFS_INODE_NODATASUM
) !=
3596 (BTRFS_I(inode
)->flags
& BTRFS_INODE_NODATASUM
))
3600 if (S_ISDIR(src
->i_mode
) || S_ISDIR(inode
->i_mode
))
3604 if (src
->i_sb
!= inode
->i_sb
)
3609 mutex_lock_nested(&inode
->i_mutex
, I_MUTEX_PARENT
);
3610 mutex_lock_nested(&src
->i_mutex
, I_MUTEX_CHILD
);
3612 mutex_lock_nested(&src
->i_mutex
, I_MUTEX_PARENT
);
3613 mutex_lock_nested(&inode
->i_mutex
, I_MUTEX_CHILD
);
3616 mutex_lock(&src
->i_mutex
);
3619 /* determine range to clone */
3621 if (off
+ len
> src
->i_size
|| off
+ len
< off
)
3624 olen
= len
= src
->i_size
- off
;
3625 /* if we extend to eof, continue to block boundary */
3626 if (off
+ len
== src
->i_size
)
3627 len
= ALIGN(src
->i_size
, bs
) - off
;
3629 /* verify the end result is block aligned */
3630 if (!IS_ALIGNED(off
, bs
) || !IS_ALIGNED(off
+ len
, bs
) ||
3631 !IS_ALIGNED(destoff
, bs
))
3634 /* verify if ranges are overlapped within the same file */
3636 if (destoff
+ len
> off
&& destoff
< off
+ len
)
3640 if (destoff
> inode
->i_size
) {
3641 ret
= btrfs_cont_expand(inode
, inode
->i_size
, destoff
);
3647 * Lock the target range too. Right after we replace the file extent
3648 * items in the fs tree (which now point to the cloned data), we might
3649 * have a worker replace them with extent items relative to a write
3650 * operation that was issued before this clone operation (i.e. confront
3651 * with inode.c:btrfs_finish_ordered_io).
3654 u64 lock_start
= min_t(u64
, off
, destoff
);
3655 u64 lock_len
= max_t(u64
, off
, destoff
) + len
- lock_start
;
3657 lock_extent_range(src
, lock_start
, lock_len
);
3659 lock_extent_range(src
, off
, len
);
3660 lock_extent_range(inode
, destoff
, len
);
3663 ret
= btrfs_clone(src
, inode
, off
, olen
, len
, destoff
);
3666 u64 lock_start
= min_t(u64
, off
, destoff
);
3667 u64 lock_end
= max_t(u64
, off
, destoff
) + len
- 1;
3669 unlock_extent(&BTRFS_I(src
)->io_tree
, lock_start
, lock_end
);
3671 unlock_extent(&BTRFS_I(src
)->io_tree
, off
, off
+ len
- 1);
3672 unlock_extent(&BTRFS_I(inode
)->io_tree
, destoff
,
3676 * Truncate page cache pages so that future reads will see the cloned
3677 * data immediately and not the previous data.
3679 truncate_inode_pages_range(&inode
->i_data
, destoff
,
3680 PAGE_CACHE_ALIGN(destoff
+ len
) - 1);
3684 mutex_unlock(&src
->i_mutex
);
3685 mutex_unlock(&inode
->i_mutex
);
3687 mutex_unlock(&inode
->i_mutex
);
3688 mutex_unlock(&src
->i_mutex
);
3691 mutex_unlock(&src
->i_mutex
);
3696 mnt_drop_write_file(file
);
3700 static long btrfs_ioctl_clone_range(struct file
*file
, void __user
*argp
)
3702 struct btrfs_ioctl_clone_range_args args
;
3704 if (copy_from_user(&args
, argp
, sizeof(args
)))
3706 return btrfs_ioctl_clone(file
, args
.src_fd
, args
.src_offset
,
3707 args
.src_length
, args
.dest_offset
);
3711 * there are many ways the trans_start and trans_end ioctls can lead
3712 * to deadlocks. They should only be used by applications that
3713 * basically own the machine, and have a very in depth understanding
3714 * of all the possible deadlocks and enospc problems.
3716 static long btrfs_ioctl_trans_start(struct file
*file
)
3718 struct inode
*inode
= file_inode(file
);
3719 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
3720 struct btrfs_trans_handle
*trans
;
3724 if (!capable(CAP_SYS_ADMIN
))
3728 if (file
->private_data
)
3732 if (btrfs_root_readonly(root
))
3735 ret
= mnt_want_write_file(file
);
3739 atomic_inc(&root
->fs_info
->open_ioctl_trans
);
3742 trans
= btrfs_start_ioctl_transaction(root
);
3746 file
->private_data
= trans
;
3750 atomic_dec(&root
->fs_info
->open_ioctl_trans
);
3751 mnt_drop_write_file(file
);
3756 static long btrfs_ioctl_default_subvol(struct file
*file
, void __user
*argp
)
3758 struct inode
*inode
= file_inode(file
);
3759 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
3760 struct btrfs_root
*new_root
;
3761 struct btrfs_dir_item
*di
;
3762 struct btrfs_trans_handle
*trans
;
3763 struct btrfs_path
*path
;
3764 struct btrfs_key location
;
3765 struct btrfs_disk_key disk_key
;
3770 if (!capable(CAP_SYS_ADMIN
))
3773 ret
= mnt_want_write_file(file
);
3777 if (copy_from_user(&objectid
, argp
, sizeof(objectid
))) {
3783 objectid
= BTRFS_FS_TREE_OBJECTID
;
3785 location
.objectid
= objectid
;
3786 location
.type
= BTRFS_ROOT_ITEM_KEY
;
3787 location
.offset
= (u64
)-1;
3789 new_root
= btrfs_read_fs_root_no_name(root
->fs_info
, &location
);
3790 if (IS_ERR(new_root
)) {
3791 ret
= PTR_ERR(new_root
);
3795 path
= btrfs_alloc_path();
3800 path
->leave_spinning
= 1;
3802 trans
= btrfs_start_transaction(root
, 1);
3803 if (IS_ERR(trans
)) {
3804 btrfs_free_path(path
);
3805 ret
= PTR_ERR(trans
);
3809 dir_id
= btrfs_super_root_dir(root
->fs_info
->super_copy
);
3810 di
= btrfs_lookup_dir_item(trans
, root
->fs_info
->tree_root
, path
,
3811 dir_id
, "default", 7, 1);
3812 if (IS_ERR_OR_NULL(di
)) {
3813 btrfs_free_path(path
);
3814 btrfs_end_transaction(trans
, root
);
3815 btrfs_err(new_root
->fs_info
, "Umm, you don't have the default dir"
3816 "item, this isn't going to work");
3821 btrfs_cpu_key_to_disk(&disk_key
, &new_root
->root_key
);
3822 btrfs_set_dir_item_key(path
->nodes
[0], di
, &disk_key
);
3823 btrfs_mark_buffer_dirty(path
->nodes
[0]);
3824 btrfs_free_path(path
);
3826 btrfs_set_fs_incompat(root
->fs_info
, DEFAULT_SUBVOL
);
3827 btrfs_end_transaction(trans
, root
);
3829 mnt_drop_write_file(file
);
3833 void btrfs_get_block_group_info(struct list_head
*groups_list
,
3834 struct btrfs_ioctl_space_info
*space
)
3836 struct btrfs_block_group_cache
*block_group
;
3838 space
->total_bytes
= 0;
3839 space
->used_bytes
= 0;
3841 list_for_each_entry(block_group
, groups_list
, list
) {
3842 space
->flags
= block_group
->flags
;
3843 space
->total_bytes
+= block_group
->key
.offset
;
3844 space
->used_bytes
+=
3845 btrfs_block_group_used(&block_group
->item
);
3849 static long btrfs_ioctl_space_info(struct btrfs_root
*root
, void __user
*arg
)
3851 struct btrfs_ioctl_space_args space_args
;
3852 struct btrfs_ioctl_space_info space
;
3853 struct btrfs_ioctl_space_info
*dest
;
3854 struct btrfs_ioctl_space_info
*dest_orig
;
3855 struct btrfs_ioctl_space_info __user
*user_dest
;
3856 struct btrfs_space_info
*info
;
3857 u64 types
[] = {BTRFS_BLOCK_GROUP_DATA
,
3858 BTRFS_BLOCK_GROUP_SYSTEM
,
3859 BTRFS_BLOCK_GROUP_METADATA
,
3860 BTRFS_BLOCK_GROUP_DATA
| BTRFS_BLOCK_GROUP_METADATA
};
3867 if (copy_from_user(&space_args
,
3868 (struct btrfs_ioctl_space_args __user
*)arg
,
3869 sizeof(space_args
)))
3872 for (i
= 0; i
< num_types
; i
++) {
3873 struct btrfs_space_info
*tmp
;
3877 list_for_each_entry_rcu(tmp
, &root
->fs_info
->space_info
,
3879 if (tmp
->flags
== types
[i
]) {
3889 down_read(&info
->groups_sem
);
3890 for (c
= 0; c
< BTRFS_NR_RAID_TYPES
; c
++) {
3891 if (!list_empty(&info
->block_groups
[c
]))
3894 up_read(&info
->groups_sem
);
3898 * Global block reserve, exported as a space_info
3902 /* space_slots == 0 means they are asking for a count */
3903 if (space_args
.space_slots
== 0) {
3904 space_args
.total_spaces
= slot_count
;
3908 slot_count
= min_t(u64
, space_args
.space_slots
, slot_count
);
3910 alloc_size
= sizeof(*dest
) * slot_count
;
3912 /* we generally have at most 6 or so space infos, one for each raid
3913 * level. So, a whole page should be more than enough for everyone
3915 if (alloc_size
> PAGE_CACHE_SIZE
)
3918 space_args
.total_spaces
= 0;
3919 dest
= kmalloc(alloc_size
, GFP_NOFS
);
3924 /* now we have a buffer to copy into */
3925 for (i
= 0; i
< num_types
; i
++) {
3926 struct btrfs_space_info
*tmp
;
3933 list_for_each_entry_rcu(tmp
, &root
->fs_info
->space_info
,
3935 if (tmp
->flags
== types
[i
]) {
3944 down_read(&info
->groups_sem
);
3945 for (c
= 0; c
< BTRFS_NR_RAID_TYPES
; c
++) {
3946 if (!list_empty(&info
->block_groups
[c
])) {
3947 btrfs_get_block_group_info(
3948 &info
->block_groups
[c
], &space
);
3949 memcpy(dest
, &space
, sizeof(space
));
3951 space_args
.total_spaces
++;
3957 up_read(&info
->groups_sem
);
3961 * Add global block reserve
3964 struct btrfs_block_rsv
*block_rsv
= &root
->fs_info
->global_block_rsv
;
3966 spin_lock(&block_rsv
->lock
);
3967 space
.total_bytes
= block_rsv
->size
;
3968 space
.used_bytes
= block_rsv
->size
- block_rsv
->reserved
;
3969 spin_unlock(&block_rsv
->lock
);
3970 space
.flags
= BTRFS_SPACE_INFO_GLOBAL_RSV
;
3971 memcpy(dest
, &space
, sizeof(space
));
3972 space_args
.total_spaces
++;
3975 user_dest
= (struct btrfs_ioctl_space_info __user
*)
3976 (arg
+ sizeof(struct btrfs_ioctl_space_args
));
3978 if (copy_to_user(user_dest
, dest_orig
, alloc_size
))
3983 if (ret
== 0 && copy_to_user(arg
, &space_args
, sizeof(space_args
)))
3990 * there are many ways the trans_start and trans_end ioctls can lead
3991 * to deadlocks. They should only be used by applications that
3992 * basically own the machine, and have a very in depth understanding
3993 * of all the possible deadlocks and enospc problems.
3995 long btrfs_ioctl_trans_end(struct file
*file
)
3997 struct inode
*inode
= file_inode(file
);
3998 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
3999 struct btrfs_trans_handle
*trans
;
4001 trans
= file
->private_data
;
4004 file
->private_data
= NULL
;
4006 btrfs_end_transaction(trans
, root
);
4008 atomic_dec(&root
->fs_info
->open_ioctl_trans
);
4010 mnt_drop_write_file(file
);
4014 static noinline
long btrfs_ioctl_start_sync(struct btrfs_root
*root
,
4017 struct btrfs_trans_handle
*trans
;
4021 trans
= btrfs_attach_transaction_barrier(root
);
4022 if (IS_ERR(trans
)) {
4023 if (PTR_ERR(trans
) != -ENOENT
)
4024 return PTR_ERR(trans
);
4026 /* No running transaction, don't bother */
4027 transid
= root
->fs_info
->last_trans_committed
;
4030 transid
= trans
->transid
;
4031 ret
= btrfs_commit_transaction_async(trans
, root
, 0);
4033 btrfs_end_transaction(trans
, root
);
4038 if (copy_to_user(argp
, &transid
, sizeof(transid
)))
4043 static noinline
long btrfs_ioctl_wait_sync(struct btrfs_root
*root
,
4049 if (copy_from_user(&transid
, argp
, sizeof(transid
)))
4052 transid
= 0; /* current trans */
4054 return btrfs_wait_for_commit(root
, transid
);
4057 static long btrfs_ioctl_scrub(struct file
*file
, void __user
*arg
)
4059 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
4060 struct btrfs_ioctl_scrub_args
*sa
;
4063 if (!capable(CAP_SYS_ADMIN
))
4066 sa
= memdup_user(arg
, sizeof(*sa
));
4070 if (!(sa
->flags
& BTRFS_SCRUB_READONLY
)) {
4071 ret
= mnt_want_write_file(file
);
4076 ret
= btrfs_scrub_dev(root
->fs_info
, sa
->devid
, sa
->start
, sa
->end
,
4077 &sa
->progress
, sa
->flags
& BTRFS_SCRUB_READONLY
,
4080 if (copy_to_user(arg
, sa
, sizeof(*sa
)))
4083 if (!(sa
->flags
& BTRFS_SCRUB_READONLY
))
4084 mnt_drop_write_file(file
);
4090 static long btrfs_ioctl_scrub_cancel(struct btrfs_root
*root
, void __user
*arg
)
4092 if (!capable(CAP_SYS_ADMIN
))
4095 return btrfs_scrub_cancel(root
->fs_info
);
4098 static long btrfs_ioctl_scrub_progress(struct btrfs_root
*root
,
4101 struct btrfs_ioctl_scrub_args
*sa
;
4104 if (!capable(CAP_SYS_ADMIN
))
4107 sa
= memdup_user(arg
, sizeof(*sa
));
4111 ret
= btrfs_scrub_progress(root
, sa
->devid
, &sa
->progress
);
4113 if (copy_to_user(arg
, sa
, sizeof(*sa
)))
4120 static long btrfs_ioctl_get_dev_stats(struct btrfs_root
*root
,
4123 struct btrfs_ioctl_get_dev_stats
*sa
;
4126 sa
= memdup_user(arg
, sizeof(*sa
));
4130 if ((sa
->flags
& BTRFS_DEV_STATS_RESET
) && !capable(CAP_SYS_ADMIN
)) {
4135 ret
= btrfs_get_dev_stats(root
, sa
);
4137 if (copy_to_user(arg
, sa
, sizeof(*sa
)))
4144 static long btrfs_ioctl_dev_replace(struct btrfs_root
*root
, void __user
*arg
)
4146 struct btrfs_ioctl_dev_replace_args
*p
;
4149 if (!capable(CAP_SYS_ADMIN
))
4152 p
= memdup_user(arg
, sizeof(*p
));
4157 case BTRFS_IOCTL_DEV_REPLACE_CMD_START
:
4158 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
) {
4163 &root
->fs_info
->mutually_exclusive_operation_running
,
4165 ret
= BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS
;
4167 ret
= btrfs_dev_replace_start(root
, p
);
4169 &root
->fs_info
->mutually_exclusive_operation_running
,
4173 case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS
:
4174 btrfs_dev_replace_status(root
->fs_info
, p
);
4177 case BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL
:
4178 ret
= btrfs_dev_replace_cancel(root
->fs_info
, p
);
4185 if (copy_to_user(arg
, p
, sizeof(*p
)))
4192 static long btrfs_ioctl_ino_to_path(struct btrfs_root
*root
, void __user
*arg
)
4198 struct btrfs_ioctl_ino_path_args
*ipa
= NULL
;
4199 struct inode_fs_paths
*ipath
= NULL
;
4200 struct btrfs_path
*path
;
4202 if (!capable(CAP_DAC_READ_SEARCH
))
4205 path
= btrfs_alloc_path();
4211 ipa
= memdup_user(arg
, sizeof(*ipa
));
4218 size
= min_t(u32
, ipa
->size
, 4096);
4219 ipath
= init_ipath(size
, root
, path
);
4220 if (IS_ERR(ipath
)) {
4221 ret
= PTR_ERR(ipath
);
4226 ret
= paths_from_inode(ipa
->inum
, ipath
);
4230 for (i
= 0; i
< ipath
->fspath
->elem_cnt
; ++i
) {
4231 rel_ptr
= ipath
->fspath
->val
[i
] -
4232 (u64
)(unsigned long)ipath
->fspath
->val
;
4233 ipath
->fspath
->val
[i
] = rel_ptr
;
4236 ret
= copy_to_user((void *)(unsigned long)ipa
->fspath
,
4237 (void *)(unsigned long)ipath
->fspath
, size
);
4244 btrfs_free_path(path
);
4251 static int build_ino_list(u64 inum
, u64 offset
, u64 root
, void *ctx
)
4253 struct btrfs_data_container
*inodes
= ctx
;
4254 const size_t c
= 3 * sizeof(u64
);
4256 if (inodes
->bytes_left
>= c
) {
4257 inodes
->bytes_left
-= c
;
4258 inodes
->val
[inodes
->elem_cnt
] = inum
;
4259 inodes
->val
[inodes
->elem_cnt
+ 1] = offset
;
4260 inodes
->val
[inodes
->elem_cnt
+ 2] = root
;
4261 inodes
->elem_cnt
+= 3;
4263 inodes
->bytes_missing
+= c
- inodes
->bytes_left
;
4264 inodes
->bytes_left
= 0;
4265 inodes
->elem_missed
+= 3;
4271 static long btrfs_ioctl_logical_to_ino(struct btrfs_root
*root
,
4276 struct btrfs_ioctl_logical_ino_args
*loi
;
4277 struct btrfs_data_container
*inodes
= NULL
;
4278 struct btrfs_path
*path
= NULL
;
4280 if (!capable(CAP_SYS_ADMIN
))
4283 loi
= memdup_user(arg
, sizeof(*loi
));
4290 path
= btrfs_alloc_path();
4296 size
= min_t(u32
, loi
->size
, 64 * 1024);
4297 inodes
= init_data_container(size
);
4298 if (IS_ERR(inodes
)) {
4299 ret
= PTR_ERR(inodes
);
4304 ret
= iterate_inodes_from_logical(loi
->logical
, root
->fs_info
, path
,
4305 build_ino_list
, inodes
);
4311 ret
= copy_to_user((void *)(unsigned long)loi
->inodes
,
4312 (void *)(unsigned long)inodes
, size
);
4317 btrfs_free_path(path
);
4324 void update_ioctl_balance_args(struct btrfs_fs_info
*fs_info
, int lock
,
4325 struct btrfs_ioctl_balance_args
*bargs
)
4327 struct btrfs_balance_control
*bctl
= fs_info
->balance_ctl
;
4329 bargs
->flags
= bctl
->flags
;
4331 if (atomic_read(&fs_info
->balance_running
))
4332 bargs
->state
|= BTRFS_BALANCE_STATE_RUNNING
;
4333 if (atomic_read(&fs_info
->balance_pause_req
))
4334 bargs
->state
|= BTRFS_BALANCE_STATE_PAUSE_REQ
;
4335 if (atomic_read(&fs_info
->balance_cancel_req
))
4336 bargs
->state
|= BTRFS_BALANCE_STATE_CANCEL_REQ
;
4338 memcpy(&bargs
->data
, &bctl
->data
, sizeof(bargs
->data
));
4339 memcpy(&bargs
->meta
, &bctl
->meta
, sizeof(bargs
->meta
));
4340 memcpy(&bargs
->sys
, &bctl
->sys
, sizeof(bargs
->sys
));
4343 spin_lock(&fs_info
->balance_lock
);
4344 memcpy(&bargs
->stat
, &bctl
->stat
, sizeof(bargs
->stat
));
4345 spin_unlock(&fs_info
->balance_lock
);
4347 memcpy(&bargs
->stat
, &bctl
->stat
, sizeof(bargs
->stat
));
4351 static long btrfs_ioctl_balance(struct file
*file
, void __user
*arg
)
4353 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
4354 struct btrfs_fs_info
*fs_info
= root
->fs_info
;
4355 struct btrfs_ioctl_balance_args
*bargs
;
4356 struct btrfs_balance_control
*bctl
;
4357 bool need_unlock
; /* for mut. excl. ops lock */
4360 if (!capable(CAP_SYS_ADMIN
))
4363 ret
= mnt_want_write_file(file
);
4368 if (!atomic_xchg(&fs_info
->mutually_exclusive_operation_running
, 1)) {
4369 mutex_lock(&fs_info
->volume_mutex
);
4370 mutex_lock(&fs_info
->balance_mutex
);
4376 * mut. excl. ops lock is locked. Three possibilites:
4377 * (1) some other op is running
4378 * (2) balance is running
4379 * (3) balance is paused -- special case (think resume)
4381 mutex_lock(&fs_info
->balance_mutex
);
4382 if (fs_info
->balance_ctl
) {
4383 /* this is either (2) or (3) */
4384 if (!atomic_read(&fs_info
->balance_running
)) {
4385 mutex_unlock(&fs_info
->balance_mutex
);
4386 if (!mutex_trylock(&fs_info
->volume_mutex
))
4388 mutex_lock(&fs_info
->balance_mutex
);
4390 if (fs_info
->balance_ctl
&&
4391 !atomic_read(&fs_info
->balance_running
)) {
4393 need_unlock
= false;
4397 mutex_unlock(&fs_info
->balance_mutex
);
4398 mutex_unlock(&fs_info
->volume_mutex
);
4402 mutex_unlock(&fs_info
->balance_mutex
);
4408 mutex_unlock(&fs_info
->balance_mutex
);
4409 ret
= BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS
;
4414 BUG_ON(!atomic_read(&fs_info
->mutually_exclusive_operation_running
));
4417 bargs
= memdup_user(arg
, sizeof(*bargs
));
4418 if (IS_ERR(bargs
)) {
4419 ret
= PTR_ERR(bargs
);
4423 if (bargs
->flags
& BTRFS_BALANCE_RESUME
) {
4424 if (!fs_info
->balance_ctl
) {
4429 bctl
= fs_info
->balance_ctl
;
4430 spin_lock(&fs_info
->balance_lock
);
4431 bctl
->flags
|= BTRFS_BALANCE_RESUME
;
4432 spin_unlock(&fs_info
->balance_lock
);
4440 if (fs_info
->balance_ctl
) {
4445 bctl
= kzalloc(sizeof(*bctl
), GFP_NOFS
);
4451 bctl
->fs_info
= fs_info
;
4453 memcpy(&bctl
->data
, &bargs
->data
, sizeof(bctl
->data
));
4454 memcpy(&bctl
->meta
, &bargs
->meta
, sizeof(bctl
->meta
));
4455 memcpy(&bctl
->sys
, &bargs
->sys
, sizeof(bctl
->sys
));
4457 bctl
->flags
= bargs
->flags
;
4459 /* balance everything - no filters */
4460 bctl
->flags
|= BTRFS_BALANCE_TYPE_MASK
;
4465 * Ownership of bctl and mutually_exclusive_operation_running
4466 * goes to to btrfs_balance. bctl is freed in __cancel_balance,
4467 * or, if restriper was paused all the way until unmount, in
4468 * free_fs_info. mutually_exclusive_operation_running is
4469 * cleared in __cancel_balance.
4471 need_unlock
= false;
4473 ret
= btrfs_balance(bctl
, bargs
);
4476 if (copy_to_user(arg
, bargs
, sizeof(*bargs
)))
4483 mutex_unlock(&fs_info
->balance_mutex
);
4484 mutex_unlock(&fs_info
->volume_mutex
);
4486 atomic_set(&fs_info
->mutually_exclusive_operation_running
, 0);
4488 mnt_drop_write_file(file
);
4492 static long btrfs_ioctl_balance_ctl(struct btrfs_root
*root
, int cmd
)
4494 if (!capable(CAP_SYS_ADMIN
))
4498 case BTRFS_BALANCE_CTL_PAUSE
:
4499 return btrfs_pause_balance(root
->fs_info
);
4500 case BTRFS_BALANCE_CTL_CANCEL
:
4501 return btrfs_cancel_balance(root
->fs_info
);
4507 static long btrfs_ioctl_balance_progress(struct btrfs_root
*root
,
4510 struct btrfs_fs_info
*fs_info
= root
->fs_info
;
4511 struct btrfs_ioctl_balance_args
*bargs
;
4514 if (!capable(CAP_SYS_ADMIN
))
4517 mutex_lock(&fs_info
->balance_mutex
);
4518 if (!fs_info
->balance_ctl
) {
4523 bargs
= kzalloc(sizeof(*bargs
), GFP_NOFS
);
4529 update_ioctl_balance_args(fs_info
, 1, bargs
);
4531 if (copy_to_user(arg
, bargs
, sizeof(*bargs
)))
4536 mutex_unlock(&fs_info
->balance_mutex
);
4540 static long btrfs_ioctl_quota_ctl(struct file
*file
, void __user
*arg
)
4542 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
4543 struct btrfs_ioctl_quota_ctl_args
*sa
;
4544 struct btrfs_trans_handle
*trans
= NULL
;
4548 if (!capable(CAP_SYS_ADMIN
))
4551 ret
= mnt_want_write_file(file
);
4555 sa
= memdup_user(arg
, sizeof(*sa
));
4561 down_write(&root
->fs_info
->subvol_sem
);
4562 trans
= btrfs_start_transaction(root
->fs_info
->tree_root
, 2);
4563 if (IS_ERR(trans
)) {
4564 ret
= PTR_ERR(trans
);
4569 case BTRFS_QUOTA_CTL_ENABLE
:
4570 ret
= btrfs_quota_enable(trans
, root
->fs_info
);
4572 case BTRFS_QUOTA_CTL_DISABLE
:
4573 ret
= btrfs_quota_disable(trans
, root
->fs_info
);
4580 err
= btrfs_commit_transaction(trans
, root
->fs_info
->tree_root
);
4585 up_write(&root
->fs_info
->subvol_sem
);
4587 mnt_drop_write_file(file
);
4591 static long btrfs_ioctl_qgroup_assign(struct file
*file
, void __user
*arg
)
4593 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
4594 struct btrfs_ioctl_qgroup_assign_args
*sa
;
4595 struct btrfs_trans_handle
*trans
;
4599 if (!capable(CAP_SYS_ADMIN
))
4602 ret
= mnt_want_write_file(file
);
4606 sa
= memdup_user(arg
, sizeof(*sa
));
4612 trans
= btrfs_join_transaction(root
);
4613 if (IS_ERR(trans
)) {
4614 ret
= PTR_ERR(trans
);
4618 /* FIXME: check if the IDs really exist */
4620 ret
= btrfs_add_qgroup_relation(trans
, root
->fs_info
,
4623 ret
= btrfs_del_qgroup_relation(trans
, root
->fs_info
,
4627 err
= btrfs_end_transaction(trans
, root
);
4634 mnt_drop_write_file(file
);
4638 static long btrfs_ioctl_qgroup_create(struct file
*file
, void __user
*arg
)
4640 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
4641 struct btrfs_ioctl_qgroup_create_args
*sa
;
4642 struct btrfs_trans_handle
*trans
;
4646 if (!capable(CAP_SYS_ADMIN
))
4649 ret
= mnt_want_write_file(file
);
4653 sa
= memdup_user(arg
, sizeof(*sa
));
4659 if (!sa
->qgroupid
) {
4664 trans
= btrfs_join_transaction(root
);
4665 if (IS_ERR(trans
)) {
4666 ret
= PTR_ERR(trans
);
4670 /* FIXME: check if the IDs really exist */
4672 ret
= btrfs_create_qgroup(trans
, root
->fs_info
, sa
->qgroupid
,
4675 ret
= btrfs_remove_qgroup(trans
, root
->fs_info
, sa
->qgroupid
);
4678 err
= btrfs_end_transaction(trans
, root
);
4685 mnt_drop_write_file(file
);
4689 static long btrfs_ioctl_qgroup_limit(struct file
*file
, void __user
*arg
)
4691 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
4692 struct btrfs_ioctl_qgroup_limit_args
*sa
;
4693 struct btrfs_trans_handle
*trans
;
4698 if (!capable(CAP_SYS_ADMIN
))
4701 ret
= mnt_want_write_file(file
);
4705 sa
= memdup_user(arg
, sizeof(*sa
));
4711 trans
= btrfs_join_transaction(root
);
4712 if (IS_ERR(trans
)) {
4713 ret
= PTR_ERR(trans
);
4717 qgroupid
= sa
->qgroupid
;
4719 /* take the current subvol as qgroup */
4720 qgroupid
= root
->root_key
.objectid
;
4723 /* FIXME: check if the IDs really exist */
4724 ret
= btrfs_limit_qgroup(trans
, root
->fs_info
, qgroupid
, &sa
->lim
);
4726 err
= btrfs_end_transaction(trans
, root
);
4733 mnt_drop_write_file(file
);
4737 static long btrfs_ioctl_quota_rescan(struct file
*file
, void __user
*arg
)
4739 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
4740 struct btrfs_ioctl_quota_rescan_args
*qsa
;
4743 if (!capable(CAP_SYS_ADMIN
))
4746 ret
= mnt_want_write_file(file
);
4750 qsa
= memdup_user(arg
, sizeof(*qsa
));
4761 ret
= btrfs_qgroup_rescan(root
->fs_info
);
4766 mnt_drop_write_file(file
);
4770 static long btrfs_ioctl_quota_rescan_status(struct file
*file
, void __user
*arg
)
4772 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
4773 struct btrfs_ioctl_quota_rescan_args
*qsa
;
4776 if (!capable(CAP_SYS_ADMIN
))
4779 qsa
= kzalloc(sizeof(*qsa
), GFP_NOFS
);
4783 if (root
->fs_info
->qgroup_flags
& BTRFS_QGROUP_STATUS_FLAG_RESCAN
) {
4785 qsa
->progress
= root
->fs_info
->qgroup_rescan_progress
.objectid
;
4788 if (copy_to_user(arg
, qsa
, sizeof(*qsa
)))
4795 static long btrfs_ioctl_quota_rescan_wait(struct file
*file
, void __user
*arg
)
4797 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
4799 if (!capable(CAP_SYS_ADMIN
))
4802 return btrfs_qgroup_wait_for_completion(root
->fs_info
);
4805 static long _btrfs_ioctl_set_received_subvol(struct file
*file
,
4806 struct btrfs_ioctl_received_subvol_args
*sa
)
4808 struct inode
*inode
= file_inode(file
);
4809 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
4810 struct btrfs_root_item
*root_item
= &root
->root_item
;
4811 struct btrfs_trans_handle
*trans
;
4812 struct timespec ct
= CURRENT_TIME
;
4814 int received_uuid_changed
;
4816 if (!inode_owner_or_capable(inode
))
4819 ret
= mnt_want_write_file(file
);
4823 down_write(&root
->fs_info
->subvol_sem
);
4825 if (btrfs_ino(inode
) != BTRFS_FIRST_FREE_OBJECTID
) {
4830 if (btrfs_root_readonly(root
)) {
4837 * 2 - uuid items (received uuid + subvol uuid)
4839 trans
= btrfs_start_transaction(root
, 3);
4840 if (IS_ERR(trans
)) {
4841 ret
= PTR_ERR(trans
);
4846 sa
->rtransid
= trans
->transid
;
4847 sa
->rtime
.sec
= ct
.tv_sec
;
4848 sa
->rtime
.nsec
= ct
.tv_nsec
;
4850 received_uuid_changed
= memcmp(root_item
->received_uuid
, sa
->uuid
,
4852 if (received_uuid_changed
&&
4853 !btrfs_is_empty_uuid(root_item
->received_uuid
))
4854 btrfs_uuid_tree_rem(trans
, root
->fs_info
->uuid_root
,
4855 root_item
->received_uuid
,
4856 BTRFS_UUID_KEY_RECEIVED_SUBVOL
,
4857 root
->root_key
.objectid
);
4858 memcpy(root_item
->received_uuid
, sa
->uuid
, BTRFS_UUID_SIZE
);
4859 btrfs_set_root_stransid(root_item
, sa
->stransid
);
4860 btrfs_set_root_rtransid(root_item
, sa
->rtransid
);
4861 btrfs_set_stack_timespec_sec(&root_item
->stime
, sa
->stime
.sec
);
4862 btrfs_set_stack_timespec_nsec(&root_item
->stime
, sa
->stime
.nsec
);
4863 btrfs_set_stack_timespec_sec(&root_item
->rtime
, sa
->rtime
.sec
);
4864 btrfs_set_stack_timespec_nsec(&root_item
->rtime
, sa
->rtime
.nsec
);
4866 ret
= btrfs_update_root(trans
, root
->fs_info
->tree_root
,
4867 &root
->root_key
, &root
->root_item
);
4869 btrfs_end_transaction(trans
, root
);
4872 if (received_uuid_changed
&& !btrfs_is_empty_uuid(sa
->uuid
)) {
4873 ret
= btrfs_uuid_tree_add(trans
, root
->fs_info
->uuid_root
,
4875 BTRFS_UUID_KEY_RECEIVED_SUBVOL
,
4876 root
->root_key
.objectid
);
4877 if (ret
< 0 && ret
!= -EEXIST
) {
4878 btrfs_abort_transaction(trans
, root
, ret
);
4882 ret
= btrfs_commit_transaction(trans
, root
);
4884 btrfs_abort_transaction(trans
, root
, ret
);
4889 up_write(&root
->fs_info
->subvol_sem
);
4890 mnt_drop_write_file(file
);
4895 static long btrfs_ioctl_set_received_subvol_32(struct file
*file
,
4898 struct btrfs_ioctl_received_subvol_args_32
*args32
= NULL
;
4899 struct btrfs_ioctl_received_subvol_args
*args64
= NULL
;
4902 args32
= memdup_user(arg
, sizeof(*args32
));
4903 if (IS_ERR(args32
)) {
4904 ret
= PTR_ERR(args32
);
4909 args64
= kmalloc(sizeof(*args64
), GFP_NOFS
);
4915 memcpy(args64
->uuid
, args32
->uuid
, BTRFS_UUID_SIZE
);
4916 args64
->stransid
= args32
->stransid
;
4917 args64
->rtransid
= args32
->rtransid
;
4918 args64
->stime
.sec
= args32
->stime
.sec
;
4919 args64
->stime
.nsec
= args32
->stime
.nsec
;
4920 args64
->rtime
.sec
= args32
->rtime
.sec
;
4921 args64
->rtime
.nsec
= args32
->rtime
.nsec
;
4922 args64
->flags
= args32
->flags
;
4924 ret
= _btrfs_ioctl_set_received_subvol(file
, args64
);
4928 memcpy(args32
->uuid
, args64
->uuid
, BTRFS_UUID_SIZE
);
4929 args32
->stransid
= args64
->stransid
;
4930 args32
->rtransid
= args64
->rtransid
;
4931 args32
->stime
.sec
= args64
->stime
.sec
;
4932 args32
->stime
.nsec
= args64
->stime
.nsec
;
4933 args32
->rtime
.sec
= args64
->rtime
.sec
;
4934 args32
->rtime
.nsec
= args64
->rtime
.nsec
;
4935 args32
->flags
= args64
->flags
;
4937 ret
= copy_to_user(arg
, args32
, sizeof(*args32
));
4948 static long btrfs_ioctl_set_received_subvol(struct file
*file
,
4951 struct btrfs_ioctl_received_subvol_args
*sa
= NULL
;
4954 sa
= memdup_user(arg
, sizeof(*sa
));
4961 ret
= _btrfs_ioctl_set_received_subvol(file
, sa
);
4966 ret
= copy_to_user(arg
, sa
, sizeof(*sa
));
4975 static int btrfs_ioctl_get_fslabel(struct file
*file
, void __user
*arg
)
4977 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
4980 char label
[BTRFS_LABEL_SIZE
];
4982 spin_lock(&root
->fs_info
->super_lock
);
4983 memcpy(label
, root
->fs_info
->super_copy
->label
, BTRFS_LABEL_SIZE
);
4984 spin_unlock(&root
->fs_info
->super_lock
);
4986 len
= strnlen(label
, BTRFS_LABEL_SIZE
);
4988 if (len
== BTRFS_LABEL_SIZE
) {
4989 btrfs_warn(root
->fs_info
,
4990 "label is too long, return the first %zu bytes", --len
);
4993 ret
= copy_to_user(arg
, label
, len
);
4995 return ret
? -EFAULT
: 0;
4998 static int btrfs_ioctl_set_fslabel(struct file
*file
, void __user
*arg
)
5000 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
5001 struct btrfs_super_block
*super_block
= root
->fs_info
->super_copy
;
5002 struct btrfs_trans_handle
*trans
;
5003 char label
[BTRFS_LABEL_SIZE
];
5006 if (!capable(CAP_SYS_ADMIN
))
5009 if (copy_from_user(label
, arg
, sizeof(label
)))
5012 if (strnlen(label
, BTRFS_LABEL_SIZE
) == BTRFS_LABEL_SIZE
) {
5013 btrfs_err(root
->fs_info
, "unable to set label with more than %d bytes",
5014 BTRFS_LABEL_SIZE
- 1);
5018 ret
= mnt_want_write_file(file
);
5022 trans
= btrfs_start_transaction(root
, 0);
5023 if (IS_ERR(trans
)) {
5024 ret
= PTR_ERR(trans
);
5028 spin_lock(&root
->fs_info
->super_lock
);
5029 strcpy(super_block
->label
, label
);
5030 spin_unlock(&root
->fs_info
->super_lock
);
5031 ret
= btrfs_commit_transaction(trans
, root
);
5034 mnt_drop_write_file(file
);
5038 #define INIT_FEATURE_FLAGS(suffix) \
5039 { .compat_flags = BTRFS_FEATURE_COMPAT_##suffix, \
5040 .compat_ro_flags = BTRFS_FEATURE_COMPAT_RO_##suffix, \
5041 .incompat_flags = BTRFS_FEATURE_INCOMPAT_##suffix }
5043 static int btrfs_ioctl_get_supported_features(struct file
*file
,
5046 static struct btrfs_ioctl_feature_flags features
[3] = {
5047 INIT_FEATURE_FLAGS(SUPP
),
5048 INIT_FEATURE_FLAGS(SAFE_SET
),
5049 INIT_FEATURE_FLAGS(SAFE_CLEAR
)
5052 if (copy_to_user(arg
, &features
, sizeof(features
)))
5058 static int btrfs_ioctl_get_features(struct file
*file
, void __user
*arg
)
5060 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
5061 struct btrfs_super_block
*super_block
= root
->fs_info
->super_copy
;
5062 struct btrfs_ioctl_feature_flags features
;
5064 features
.compat_flags
= btrfs_super_compat_flags(super_block
);
5065 features
.compat_ro_flags
= btrfs_super_compat_ro_flags(super_block
);
5066 features
.incompat_flags
= btrfs_super_incompat_flags(super_block
);
5068 if (copy_to_user(arg
, &features
, sizeof(features
)))
5074 static int check_feature_bits(struct btrfs_root
*root
,
5075 enum btrfs_feature_set set
,
5076 u64 change_mask
, u64 flags
, u64 supported_flags
,
5077 u64 safe_set
, u64 safe_clear
)
5079 const char *type
= btrfs_feature_set_names
[set
];
5081 u64 disallowed
, unsupported
;
5082 u64 set_mask
= flags
& change_mask
;
5083 u64 clear_mask
= ~flags
& change_mask
;
5085 unsupported
= set_mask
& ~supported_flags
;
5087 names
= btrfs_printable_features(set
, unsupported
);
5089 btrfs_warn(root
->fs_info
,
5090 "this kernel does not support the %s feature bit%s",
5091 names
, strchr(names
, ',') ? "s" : "");
5094 btrfs_warn(root
->fs_info
,
5095 "this kernel does not support %s bits 0x%llx",
5100 disallowed
= set_mask
& ~safe_set
;
5102 names
= btrfs_printable_features(set
, disallowed
);
5104 btrfs_warn(root
->fs_info
,
5105 "can't set the %s feature bit%s while mounted",
5106 names
, strchr(names
, ',') ? "s" : "");
5109 btrfs_warn(root
->fs_info
,
5110 "can't set %s bits 0x%llx while mounted",
5115 disallowed
= clear_mask
& ~safe_clear
;
5117 names
= btrfs_printable_features(set
, disallowed
);
5119 btrfs_warn(root
->fs_info
,
5120 "can't clear the %s feature bit%s while mounted",
5121 names
, strchr(names
, ',') ? "s" : "");
5124 btrfs_warn(root
->fs_info
,
5125 "can't clear %s bits 0x%llx while mounted",
5133 #define check_feature(root, change_mask, flags, mask_base) \
5134 check_feature_bits(root, FEAT_##mask_base, change_mask, flags, \
5135 BTRFS_FEATURE_ ## mask_base ## _SUPP, \
5136 BTRFS_FEATURE_ ## mask_base ## _SAFE_SET, \
5137 BTRFS_FEATURE_ ## mask_base ## _SAFE_CLEAR)
5139 static int btrfs_ioctl_set_features(struct file
*file
, void __user
*arg
)
5141 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
5142 struct btrfs_super_block
*super_block
= root
->fs_info
->super_copy
;
5143 struct btrfs_ioctl_feature_flags flags
[2];
5144 struct btrfs_trans_handle
*trans
;
5148 if (!capable(CAP_SYS_ADMIN
))
5151 if (copy_from_user(flags
, arg
, sizeof(flags
)))
5155 if (!flags
[0].compat_flags
&& !flags
[0].compat_ro_flags
&&
5156 !flags
[0].incompat_flags
)
5159 ret
= check_feature(root
, flags
[0].compat_flags
,
5160 flags
[1].compat_flags
, COMPAT
);
5164 ret
= check_feature(root
, flags
[0].compat_ro_flags
,
5165 flags
[1].compat_ro_flags
, COMPAT_RO
);
5169 ret
= check_feature(root
, flags
[0].incompat_flags
,
5170 flags
[1].incompat_flags
, INCOMPAT
);
5174 trans
= btrfs_start_transaction(root
, 0);
5176 return PTR_ERR(trans
);
5178 spin_lock(&root
->fs_info
->super_lock
);
5179 newflags
= btrfs_super_compat_flags(super_block
);
5180 newflags
|= flags
[0].compat_flags
& flags
[1].compat_flags
;
5181 newflags
&= ~(flags
[0].compat_flags
& ~flags
[1].compat_flags
);
5182 btrfs_set_super_compat_flags(super_block
, newflags
);
5184 newflags
= btrfs_super_compat_ro_flags(super_block
);
5185 newflags
|= flags
[0].compat_ro_flags
& flags
[1].compat_ro_flags
;
5186 newflags
&= ~(flags
[0].compat_ro_flags
& ~flags
[1].compat_ro_flags
);
5187 btrfs_set_super_compat_ro_flags(super_block
, newflags
);
5189 newflags
= btrfs_super_incompat_flags(super_block
);
5190 newflags
|= flags
[0].incompat_flags
& flags
[1].incompat_flags
;
5191 newflags
&= ~(flags
[0].incompat_flags
& ~flags
[1].incompat_flags
);
5192 btrfs_set_super_incompat_flags(super_block
, newflags
);
5193 spin_unlock(&root
->fs_info
->super_lock
);
5195 return btrfs_commit_transaction(trans
, root
);
5198 long btrfs_ioctl(struct file
*file
, unsigned int
5199 cmd
, unsigned long arg
)
5201 struct btrfs_root
*root
= BTRFS_I(file_inode(file
))->root
;
5202 void __user
*argp
= (void __user
*)arg
;
5205 case FS_IOC_GETFLAGS
:
5206 return btrfs_ioctl_getflags(file
, argp
);
5207 case FS_IOC_SETFLAGS
:
5208 return btrfs_ioctl_setflags(file
, argp
);
5209 case FS_IOC_GETVERSION
:
5210 return btrfs_ioctl_getversion(file
, argp
);
5212 return btrfs_ioctl_fitrim(file
, argp
);
5213 case BTRFS_IOC_SNAP_CREATE
:
5214 return btrfs_ioctl_snap_create(file
, argp
, 0);
5215 case BTRFS_IOC_SNAP_CREATE_V2
:
5216 return btrfs_ioctl_snap_create_v2(file
, argp
, 0);
5217 case BTRFS_IOC_SUBVOL_CREATE
:
5218 return btrfs_ioctl_snap_create(file
, argp
, 1);
5219 case BTRFS_IOC_SUBVOL_CREATE_V2
:
5220 return btrfs_ioctl_snap_create_v2(file
, argp
, 1);
5221 case BTRFS_IOC_SNAP_DESTROY
:
5222 return btrfs_ioctl_snap_destroy(file
, argp
);
5223 case BTRFS_IOC_SUBVOL_GETFLAGS
:
5224 return btrfs_ioctl_subvol_getflags(file
, argp
);
5225 case BTRFS_IOC_SUBVOL_SETFLAGS
:
5226 return btrfs_ioctl_subvol_setflags(file
, argp
);
5227 case BTRFS_IOC_DEFAULT_SUBVOL
:
5228 return btrfs_ioctl_default_subvol(file
, argp
);
5229 case BTRFS_IOC_DEFRAG
:
5230 return btrfs_ioctl_defrag(file
, NULL
);
5231 case BTRFS_IOC_DEFRAG_RANGE
:
5232 return btrfs_ioctl_defrag(file
, argp
);
5233 case BTRFS_IOC_RESIZE
:
5234 return btrfs_ioctl_resize(file
, argp
);
5235 case BTRFS_IOC_ADD_DEV
:
5236 return btrfs_ioctl_add_dev(root
, argp
);
5237 case BTRFS_IOC_RM_DEV
:
5238 return btrfs_ioctl_rm_dev(file
, argp
);
5239 case BTRFS_IOC_FS_INFO
:
5240 return btrfs_ioctl_fs_info(root
, argp
);
5241 case BTRFS_IOC_DEV_INFO
:
5242 return btrfs_ioctl_dev_info(root
, argp
);
5243 case BTRFS_IOC_BALANCE
:
5244 return btrfs_ioctl_balance(file
, NULL
);
5245 case BTRFS_IOC_CLONE
:
5246 return btrfs_ioctl_clone(file
, arg
, 0, 0, 0);
5247 case BTRFS_IOC_CLONE_RANGE
:
5248 return btrfs_ioctl_clone_range(file
, argp
);
5249 case BTRFS_IOC_TRANS_START
:
5250 return btrfs_ioctl_trans_start(file
);
5251 case BTRFS_IOC_TRANS_END
:
5252 return btrfs_ioctl_trans_end(file
);
5253 case BTRFS_IOC_TREE_SEARCH
:
5254 return btrfs_ioctl_tree_search(file
, argp
);
5255 case BTRFS_IOC_TREE_SEARCH_V2
:
5256 return btrfs_ioctl_tree_search_v2(file
, argp
);
5257 case BTRFS_IOC_INO_LOOKUP
:
5258 return btrfs_ioctl_ino_lookup(file
, argp
);
5259 case BTRFS_IOC_INO_PATHS
:
5260 return btrfs_ioctl_ino_to_path(root
, argp
);
5261 case BTRFS_IOC_LOGICAL_INO
:
5262 return btrfs_ioctl_logical_to_ino(root
, argp
);
5263 case BTRFS_IOC_SPACE_INFO
:
5264 return btrfs_ioctl_space_info(root
, argp
);
5265 case BTRFS_IOC_SYNC
: {
5268 ret
= btrfs_start_delalloc_roots(root
->fs_info
, 0, -1);
5271 ret
= btrfs_sync_fs(file_inode(file
)->i_sb
, 1);
5273 * The transaction thread may want to do more work,
5274 * namely it pokes the cleaner ktread that will start
5275 * processing uncleaned subvols.
5277 wake_up_process(root
->fs_info
->transaction_kthread
);
5280 case BTRFS_IOC_START_SYNC
:
5281 return btrfs_ioctl_start_sync(root
, argp
);
5282 case BTRFS_IOC_WAIT_SYNC
:
5283 return btrfs_ioctl_wait_sync(root
, argp
);
5284 case BTRFS_IOC_SCRUB
:
5285 return btrfs_ioctl_scrub(file
, argp
);
5286 case BTRFS_IOC_SCRUB_CANCEL
:
5287 return btrfs_ioctl_scrub_cancel(root
, argp
);
5288 case BTRFS_IOC_SCRUB_PROGRESS
:
5289 return btrfs_ioctl_scrub_progress(root
, argp
);
5290 case BTRFS_IOC_BALANCE_V2
:
5291 return btrfs_ioctl_balance(file
, argp
);
5292 case BTRFS_IOC_BALANCE_CTL
:
5293 return btrfs_ioctl_balance_ctl(root
, arg
);
5294 case BTRFS_IOC_BALANCE_PROGRESS
:
5295 return btrfs_ioctl_balance_progress(root
, argp
);
5296 case BTRFS_IOC_SET_RECEIVED_SUBVOL
:
5297 return btrfs_ioctl_set_received_subvol(file
, argp
);
5299 case BTRFS_IOC_SET_RECEIVED_SUBVOL_32
:
5300 return btrfs_ioctl_set_received_subvol_32(file
, argp
);
5302 case BTRFS_IOC_SEND
:
5303 return btrfs_ioctl_send(file
, argp
);
5304 case BTRFS_IOC_GET_DEV_STATS
:
5305 return btrfs_ioctl_get_dev_stats(root
, argp
);
5306 case BTRFS_IOC_QUOTA_CTL
:
5307 return btrfs_ioctl_quota_ctl(file
, argp
);
5308 case BTRFS_IOC_QGROUP_ASSIGN
:
5309 return btrfs_ioctl_qgroup_assign(file
, argp
);
5310 case BTRFS_IOC_QGROUP_CREATE
:
5311 return btrfs_ioctl_qgroup_create(file
, argp
);
5312 case BTRFS_IOC_QGROUP_LIMIT
:
5313 return btrfs_ioctl_qgroup_limit(file
, argp
);
5314 case BTRFS_IOC_QUOTA_RESCAN
:
5315 return btrfs_ioctl_quota_rescan(file
, argp
);
5316 case BTRFS_IOC_QUOTA_RESCAN_STATUS
:
5317 return btrfs_ioctl_quota_rescan_status(file
, argp
);
5318 case BTRFS_IOC_QUOTA_RESCAN_WAIT
:
5319 return btrfs_ioctl_quota_rescan_wait(file
, argp
);
5320 case BTRFS_IOC_DEV_REPLACE
:
5321 return btrfs_ioctl_dev_replace(root
, argp
);
5322 case BTRFS_IOC_GET_FSLABEL
:
5323 return btrfs_ioctl_get_fslabel(file
, argp
);
5324 case BTRFS_IOC_SET_FSLABEL
:
5325 return btrfs_ioctl_set_fslabel(file
, argp
);
5326 case BTRFS_IOC_FILE_EXTENT_SAME
:
5327 return btrfs_ioctl_file_extent_same(file
, argp
);
5328 case BTRFS_IOC_GET_SUPPORTED_FEATURES
:
5329 return btrfs_ioctl_get_supported_features(file
, argp
);
5330 case BTRFS_IOC_GET_FEATURES
:
5331 return btrfs_ioctl_get_features(file
, argp
);
5332 case BTRFS_IOC_SET_FEATURES
:
5333 return btrfs_ioctl_set_features(file
, argp
);