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>
47 #include "transaction.h"
48 #include "btrfs_inode.h"
50 #include "print-tree.h"
53 #include "inode-map.h"
55 /* Mask out flags that are inappropriate for the given type of inode. */
56 static inline __u32
btrfs_mask_flags(umode_t mode
, __u32 flags
)
60 else if (S_ISREG(mode
))
61 return flags
& ~FS_DIRSYNC_FL
;
63 return flags
& (FS_NODUMP_FL
| FS_NOATIME_FL
);
67 * Export inode flags to the format expected by the FS_IOC_GETFLAGS ioctl.
69 static unsigned int btrfs_flags_to_ioctl(unsigned int flags
)
71 unsigned int iflags
= 0;
73 if (flags
& BTRFS_INODE_SYNC
)
75 if (flags
& BTRFS_INODE_IMMUTABLE
)
76 iflags
|= FS_IMMUTABLE_FL
;
77 if (flags
& BTRFS_INODE_APPEND
)
78 iflags
|= FS_APPEND_FL
;
79 if (flags
& BTRFS_INODE_NODUMP
)
80 iflags
|= FS_NODUMP_FL
;
81 if (flags
& BTRFS_INODE_NOATIME
)
82 iflags
|= FS_NOATIME_FL
;
83 if (flags
& BTRFS_INODE_DIRSYNC
)
84 iflags
|= FS_DIRSYNC_FL
;
85 if (flags
& BTRFS_INODE_NODATACOW
)
86 iflags
|= FS_NOCOW_FL
;
88 if ((flags
& BTRFS_INODE_COMPRESS
) && !(flags
& BTRFS_INODE_NOCOMPRESS
))
89 iflags
|= FS_COMPR_FL
;
90 else if (flags
& BTRFS_INODE_NOCOMPRESS
)
91 iflags
|= FS_NOCOMP_FL
;
97 * Update inode->i_flags based on the btrfs internal flags.
99 void btrfs_update_iflags(struct inode
*inode
)
101 struct btrfs_inode
*ip
= BTRFS_I(inode
);
103 inode
->i_flags
&= ~(S_SYNC
|S_APPEND
|S_IMMUTABLE
|S_NOATIME
|S_DIRSYNC
);
105 if (ip
->flags
& BTRFS_INODE_SYNC
)
106 inode
->i_flags
|= S_SYNC
;
107 if (ip
->flags
& BTRFS_INODE_IMMUTABLE
)
108 inode
->i_flags
|= S_IMMUTABLE
;
109 if (ip
->flags
& BTRFS_INODE_APPEND
)
110 inode
->i_flags
|= S_APPEND
;
111 if (ip
->flags
& BTRFS_INODE_NOATIME
)
112 inode
->i_flags
|= S_NOATIME
;
113 if (ip
->flags
& BTRFS_INODE_DIRSYNC
)
114 inode
->i_flags
|= S_DIRSYNC
;
118 * Inherit flags from the parent inode.
120 * Currently only the compression flags and the cow flags are inherited.
122 void btrfs_inherit_iflags(struct inode
*inode
, struct inode
*dir
)
129 flags
= BTRFS_I(dir
)->flags
;
131 if (flags
& BTRFS_INODE_NOCOMPRESS
) {
132 BTRFS_I(inode
)->flags
&= ~BTRFS_INODE_COMPRESS
;
133 BTRFS_I(inode
)->flags
|= BTRFS_INODE_NOCOMPRESS
;
134 } else if (flags
& BTRFS_INODE_COMPRESS
) {
135 BTRFS_I(inode
)->flags
&= ~BTRFS_INODE_NOCOMPRESS
;
136 BTRFS_I(inode
)->flags
|= BTRFS_INODE_COMPRESS
;
139 if (flags
& BTRFS_INODE_NODATACOW
)
140 BTRFS_I(inode
)->flags
|= BTRFS_INODE_NODATACOW
;
142 btrfs_update_iflags(inode
);
145 static int btrfs_ioctl_getflags(struct file
*file
, void __user
*arg
)
147 struct btrfs_inode
*ip
= BTRFS_I(file
->f_path
.dentry
->d_inode
);
148 unsigned int flags
= btrfs_flags_to_ioctl(ip
->flags
);
150 if (copy_to_user(arg
, &flags
, sizeof(flags
)))
155 static int check_flags(unsigned int flags
)
157 if (flags
& ~(FS_IMMUTABLE_FL
| FS_APPEND_FL
| \
158 FS_NOATIME_FL
| FS_NODUMP_FL
| \
159 FS_SYNC_FL
| FS_DIRSYNC_FL
| \
160 FS_NOCOMP_FL
| FS_COMPR_FL
|
164 if ((flags
& FS_NOCOMP_FL
) && (flags
& FS_COMPR_FL
))
170 static int btrfs_ioctl_setflags(struct file
*file
, void __user
*arg
)
172 struct inode
*inode
= file
->f_path
.dentry
->d_inode
;
173 struct btrfs_inode
*ip
= BTRFS_I(inode
);
174 struct btrfs_root
*root
= ip
->root
;
175 struct btrfs_trans_handle
*trans
;
176 unsigned int flags
, oldflags
;
179 if (btrfs_root_readonly(root
))
182 if (copy_from_user(&flags
, arg
, sizeof(flags
)))
185 ret
= check_flags(flags
);
189 if (!inode_owner_or_capable(inode
))
192 mutex_lock(&inode
->i_mutex
);
194 flags
= btrfs_mask_flags(inode
->i_mode
, flags
);
195 oldflags
= btrfs_flags_to_ioctl(ip
->flags
);
196 if ((flags
^ oldflags
) & (FS_APPEND_FL
| FS_IMMUTABLE_FL
)) {
197 if (!capable(CAP_LINUX_IMMUTABLE
)) {
203 ret
= mnt_want_write(file
->f_path
.mnt
);
207 if (flags
& FS_SYNC_FL
)
208 ip
->flags
|= BTRFS_INODE_SYNC
;
210 ip
->flags
&= ~BTRFS_INODE_SYNC
;
211 if (flags
& FS_IMMUTABLE_FL
)
212 ip
->flags
|= BTRFS_INODE_IMMUTABLE
;
214 ip
->flags
&= ~BTRFS_INODE_IMMUTABLE
;
215 if (flags
& FS_APPEND_FL
)
216 ip
->flags
|= BTRFS_INODE_APPEND
;
218 ip
->flags
&= ~BTRFS_INODE_APPEND
;
219 if (flags
& FS_NODUMP_FL
)
220 ip
->flags
|= BTRFS_INODE_NODUMP
;
222 ip
->flags
&= ~BTRFS_INODE_NODUMP
;
223 if (flags
& FS_NOATIME_FL
)
224 ip
->flags
|= BTRFS_INODE_NOATIME
;
226 ip
->flags
&= ~BTRFS_INODE_NOATIME
;
227 if (flags
& FS_DIRSYNC_FL
)
228 ip
->flags
|= BTRFS_INODE_DIRSYNC
;
230 ip
->flags
&= ~BTRFS_INODE_DIRSYNC
;
231 if (flags
& FS_NOCOW_FL
)
232 ip
->flags
|= BTRFS_INODE_NODATACOW
;
234 ip
->flags
&= ~BTRFS_INODE_NODATACOW
;
237 * The COMPRESS flag can only be changed by users, while the NOCOMPRESS
238 * flag may be changed automatically if compression code won't make
241 if (flags
& FS_NOCOMP_FL
) {
242 ip
->flags
&= ~BTRFS_INODE_COMPRESS
;
243 ip
->flags
|= BTRFS_INODE_NOCOMPRESS
;
244 } else if (flags
& FS_COMPR_FL
) {
245 ip
->flags
|= BTRFS_INODE_COMPRESS
;
246 ip
->flags
&= ~BTRFS_INODE_NOCOMPRESS
;
248 ip
->flags
&= ~(BTRFS_INODE_COMPRESS
| BTRFS_INODE_NOCOMPRESS
);
251 trans
= btrfs_join_transaction(root
);
252 BUG_ON(IS_ERR(trans
));
254 ret
= btrfs_update_inode(trans
, root
, inode
);
257 btrfs_update_iflags(inode
);
258 inode
->i_ctime
= CURRENT_TIME
;
259 btrfs_end_transaction(trans
, root
);
261 mnt_drop_write(file
->f_path
.mnt
);
265 mutex_unlock(&inode
->i_mutex
);
269 static int btrfs_ioctl_getversion(struct file
*file
, int __user
*arg
)
271 struct inode
*inode
= file
->f_path
.dentry
->d_inode
;
273 return put_user(inode
->i_generation
, arg
);
276 static noinline
int btrfs_ioctl_fitrim(struct file
*file
, void __user
*arg
)
278 struct btrfs_root
*root
= fdentry(file
)->d_sb
->s_fs_info
;
279 struct btrfs_fs_info
*fs_info
= root
->fs_info
;
280 struct btrfs_device
*device
;
281 struct request_queue
*q
;
282 struct fstrim_range range
;
283 u64 minlen
= ULLONG_MAX
;
287 if (!capable(CAP_SYS_ADMIN
))
291 list_for_each_entry_rcu(device
, &fs_info
->fs_devices
->devices
,
295 q
= bdev_get_queue(device
->bdev
);
296 if (blk_queue_discard(q
)) {
298 minlen
= min((u64
)q
->limits
.discard_granularity
,
306 if (copy_from_user(&range
, arg
, sizeof(range
)))
309 range
.minlen
= max(range
.minlen
, minlen
);
310 ret
= btrfs_trim_fs(root
, &range
);
314 if (copy_to_user(arg
, &range
, sizeof(range
)))
320 static noinline
int create_subvol(struct btrfs_root
*root
,
321 struct dentry
*dentry
,
322 char *name
, int namelen
,
325 struct btrfs_trans_handle
*trans
;
326 struct btrfs_key key
;
327 struct btrfs_root_item root_item
;
328 struct btrfs_inode_item
*inode_item
;
329 struct extent_buffer
*leaf
;
330 struct btrfs_root
*new_root
;
331 struct dentry
*parent
= dentry
->d_parent
;
336 u64 new_dirid
= BTRFS_FIRST_FREE_OBJECTID
;
339 ret
= btrfs_find_free_objectid(root
->fs_info
->tree_root
, &objectid
);
343 dir
= parent
->d_inode
;
351 trans
= btrfs_start_transaction(root
, 6);
353 return PTR_ERR(trans
);
355 leaf
= btrfs_alloc_free_block(trans
, root
, root
->leafsize
,
356 0, objectid
, NULL
, 0, 0, 0);
362 memset_extent_buffer(leaf
, 0, 0, sizeof(struct btrfs_header
));
363 btrfs_set_header_bytenr(leaf
, leaf
->start
);
364 btrfs_set_header_generation(leaf
, trans
->transid
);
365 btrfs_set_header_backref_rev(leaf
, BTRFS_MIXED_BACKREF_REV
);
366 btrfs_set_header_owner(leaf
, objectid
);
368 write_extent_buffer(leaf
, root
->fs_info
->fsid
,
369 (unsigned long)btrfs_header_fsid(leaf
),
371 write_extent_buffer(leaf
, root
->fs_info
->chunk_tree_uuid
,
372 (unsigned long)btrfs_header_chunk_tree_uuid(leaf
),
374 btrfs_mark_buffer_dirty(leaf
);
376 inode_item
= &root_item
.inode
;
377 memset(inode_item
, 0, sizeof(*inode_item
));
378 inode_item
->generation
= cpu_to_le64(1);
379 inode_item
->size
= cpu_to_le64(3);
380 inode_item
->nlink
= cpu_to_le32(1);
381 inode_item
->nbytes
= cpu_to_le64(root
->leafsize
);
382 inode_item
->mode
= cpu_to_le32(S_IFDIR
| 0755);
385 root_item
.byte_limit
= 0;
386 inode_item
->flags
= cpu_to_le64(BTRFS_INODE_ROOT_ITEM_INIT
);
388 btrfs_set_root_bytenr(&root_item
, leaf
->start
);
389 btrfs_set_root_generation(&root_item
, trans
->transid
);
390 btrfs_set_root_level(&root_item
, 0);
391 btrfs_set_root_refs(&root_item
, 1);
392 btrfs_set_root_used(&root_item
, leaf
->len
);
393 btrfs_set_root_last_snapshot(&root_item
, 0);
395 memset(&root_item
.drop_progress
, 0, sizeof(root_item
.drop_progress
));
396 root_item
.drop_level
= 0;
398 btrfs_tree_unlock(leaf
);
399 free_extent_buffer(leaf
);
402 btrfs_set_root_dirid(&root_item
, new_dirid
);
404 key
.objectid
= objectid
;
406 btrfs_set_key_type(&key
, BTRFS_ROOT_ITEM_KEY
);
407 ret
= btrfs_insert_root(trans
, root
->fs_info
->tree_root
, &key
,
412 key
.offset
= (u64
)-1;
413 new_root
= btrfs_read_fs_root_no_name(root
->fs_info
, &key
);
414 BUG_ON(IS_ERR(new_root
));
416 btrfs_record_root_in_trans(trans
, new_root
);
418 ret
= btrfs_create_subvol_root(trans
, new_root
, new_dirid
);
420 * insert the directory item
422 ret
= btrfs_set_inode_index(dir
, &index
);
425 ret
= btrfs_insert_dir_item(trans
, root
,
426 name
, namelen
, dir
, &key
,
427 BTRFS_FT_DIR
, index
);
431 btrfs_i_size_write(dir
, dir
->i_size
+ namelen
* 2);
432 ret
= btrfs_update_inode(trans
, root
, dir
);
435 ret
= btrfs_add_root_ref(trans
, root
->fs_info
->tree_root
,
436 objectid
, root
->root_key
.objectid
,
437 btrfs_ino(dir
), index
, name
, namelen
);
441 d_instantiate(dentry
, btrfs_lookup_dentry(dir
, dentry
));
444 *async_transid
= trans
->transid
;
445 err
= btrfs_commit_transaction_async(trans
, root
, 1);
447 err
= btrfs_commit_transaction(trans
, root
);
454 static int create_snapshot(struct btrfs_root
*root
, struct dentry
*dentry
,
455 char *name
, int namelen
, u64
*async_transid
,
459 struct btrfs_pending_snapshot
*pending_snapshot
;
460 struct btrfs_trans_handle
*trans
;
466 pending_snapshot
= kzalloc(sizeof(*pending_snapshot
), GFP_NOFS
);
467 if (!pending_snapshot
)
470 btrfs_init_block_rsv(&pending_snapshot
->block_rsv
);
471 pending_snapshot
->dentry
= dentry
;
472 pending_snapshot
->root
= root
;
473 pending_snapshot
->readonly
= readonly
;
475 trans
= btrfs_start_transaction(root
->fs_info
->extent_root
, 5);
477 ret
= PTR_ERR(trans
);
481 ret
= btrfs_snap_reserve_metadata(trans
, pending_snapshot
);
484 spin_lock(&root
->fs_info
->trans_lock
);
485 list_add(&pending_snapshot
->list
,
486 &trans
->transaction
->pending_snapshots
);
487 spin_unlock(&root
->fs_info
->trans_lock
);
489 *async_transid
= trans
->transid
;
490 ret
= btrfs_commit_transaction_async(trans
,
491 root
->fs_info
->extent_root
, 1);
493 ret
= btrfs_commit_transaction(trans
,
494 root
->fs_info
->extent_root
);
498 ret
= pending_snapshot
->error
;
502 ret
= btrfs_orphan_cleanup(pending_snapshot
->snap
);
506 inode
= btrfs_lookup_dentry(dentry
->d_parent
->d_inode
, dentry
);
508 ret
= PTR_ERR(inode
);
512 d_instantiate(dentry
, inode
);
515 kfree(pending_snapshot
);
519 /* copy of check_sticky in fs/namei.c()
520 * It's inline, so penalty for filesystems that don't use sticky bit is
523 static inline int btrfs_check_sticky(struct inode
*dir
, struct inode
*inode
)
525 uid_t fsuid
= current_fsuid();
527 if (!(dir
->i_mode
& S_ISVTX
))
529 if (inode
->i_uid
== fsuid
)
531 if (dir
->i_uid
== fsuid
)
533 return !capable(CAP_FOWNER
);
536 /* copy of may_delete in fs/namei.c()
537 * Check whether we can remove a link victim from directory dir, check
538 * whether the type of victim is right.
539 * 1. We can't do it if dir is read-only (done in permission())
540 * 2. We should have write and exec permissions on dir
541 * 3. We can't remove anything from append-only dir
542 * 4. We can't do anything with immutable dir (done in permission())
543 * 5. If the sticky bit on dir is set we should either
544 * a. be owner of dir, or
545 * b. be owner of victim, or
546 * c. have CAP_FOWNER capability
547 * 6. If the victim is append-only or immutable we can't do antyhing with
548 * links pointing to it.
549 * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR.
550 * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR.
551 * 9. We can't remove a root or mountpoint.
552 * 10. We don't allow removal of NFS sillyrenamed files; it's handled by
553 * nfs_async_unlink().
556 static int btrfs_may_delete(struct inode
*dir
,struct dentry
*victim
,int isdir
)
560 if (!victim
->d_inode
)
563 BUG_ON(victim
->d_parent
->d_inode
!= dir
);
564 audit_inode_child(victim
, dir
);
566 error
= inode_permission(dir
, MAY_WRITE
| MAY_EXEC
);
571 if (btrfs_check_sticky(dir
, victim
->d_inode
)||
572 IS_APPEND(victim
->d_inode
)||
573 IS_IMMUTABLE(victim
->d_inode
) || IS_SWAPFILE(victim
->d_inode
))
576 if (!S_ISDIR(victim
->d_inode
->i_mode
))
580 } else if (S_ISDIR(victim
->d_inode
->i_mode
))
584 if (victim
->d_flags
& DCACHE_NFSFS_RENAMED
)
589 /* copy of may_create in fs/namei.c() */
590 static inline int btrfs_may_create(struct inode
*dir
, struct dentry
*child
)
596 return inode_permission(dir
, MAY_WRITE
| MAY_EXEC
);
600 * Create a new subvolume below @parent. This is largely modeled after
601 * sys_mkdirat and vfs_mkdir, but we only do a single component lookup
602 * inside this filesystem so it's quite a bit simpler.
604 static noinline
int btrfs_mksubvol(struct path
*parent
,
605 char *name
, int namelen
,
606 struct btrfs_root
*snap_src
,
607 u64
*async_transid
, bool readonly
)
609 struct inode
*dir
= parent
->dentry
->d_inode
;
610 struct dentry
*dentry
;
613 mutex_lock_nested(&dir
->i_mutex
, I_MUTEX_PARENT
);
615 dentry
= lookup_one_len(name
, parent
->dentry
, namelen
);
616 error
= PTR_ERR(dentry
);
624 error
= mnt_want_write(parent
->mnt
);
628 error
= btrfs_may_create(dir
, dentry
);
632 down_read(&BTRFS_I(dir
)->root
->fs_info
->subvol_sem
);
634 if (btrfs_root_refs(&BTRFS_I(dir
)->root
->root_item
) == 0)
638 error
= create_snapshot(snap_src
, dentry
,
639 name
, namelen
, async_transid
, readonly
);
641 error
= create_subvol(BTRFS_I(dir
)->root
, dentry
,
642 name
, namelen
, async_transid
);
645 fsnotify_mkdir(dir
, dentry
);
647 up_read(&BTRFS_I(dir
)->root
->fs_info
->subvol_sem
);
649 mnt_drop_write(parent
->mnt
);
653 mutex_unlock(&dir
->i_mutex
);
658 * When we're defragging a range, we don't want to kick it off again
659 * if it is really just waiting for delalloc to send it down.
660 * If we find a nice big extent or delalloc range for the bytes in the
661 * file you want to defrag, we return 0 to let you know to skip this
664 static int check_defrag_in_cache(struct inode
*inode
, u64 offset
, int thresh
)
666 struct extent_io_tree
*io_tree
= &BTRFS_I(inode
)->io_tree
;
667 struct extent_map
*em
= NULL
;
668 struct extent_map_tree
*em_tree
= &BTRFS_I(inode
)->extent_tree
;
671 read_lock(&em_tree
->lock
);
672 em
= lookup_extent_mapping(em_tree
, offset
, PAGE_CACHE_SIZE
);
673 read_unlock(&em_tree
->lock
);
676 end
= extent_map_end(em
);
678 if (end
- offset
> thresh
)
681 /* if we already have a nice delalloc here, just stop */
683 end
= count_range_bits(io_tree
, &offset
, offset
+ thresh
,
684 thresh
, EXTENT_DELALLOC
, 1);
691 * helper function to walk through a file and find extents
692 * newer than a specific transid, and smaller than thresh.
694 * This is used by the defragging code to find new and small
697 static int find_new_extents(struct btrfs_root
*root
,
698 struct inode
*inode
, u64 newer_than
,
699 u64
*off
, int thresh
)
701 struct btrfs_path
*path
;
702 struct btrfs_key min_key
;
703 struct btrfs_key max_key
;
704 struct extent_buffer
*leaf
;
705 struct btrfs_file_extent_item
*extent
;
708 u64 ino
= btrfs_ino(inode
);
710 path
= btrfs_alloc_path();
714 min_key
.objectid
= ino
;
715 min_key
.type
= BTRFS_EXTENT_DATA_KEY
;
716 min_key
.offset
= *off
;
718 max_key
.objectid
= ino
;
719 max_key
.type
= (u8
)-1;
720 max_key
.offset
= (u64
)-1;
722 path
->keep_locks
= 1;
725 ret
= btrfs_search_forward(root
, &min_key
, &max_key
,
726 path
, 0, newer_than
);
729 if (min_key
.objectid
!= ino
)
731 if (min_key
.type
!= BTRFS_EXTENT_DATA_KEY
)
734 leaf
= path
->nodes
[0];
735 extent
= btrfs_item_ptr(leaf
, path
->slots
[0],
736 struct btrfs_file_extent_item
);
738 type
= btrfs_file_extent_type(leaf
, extent
);
739 if (type
== BTRFS_FILE_EXTENT_REG
&&
740 btrfs_file_extent_num_bytes(leaf
, extent
) < thresh
&&
741 check_defrag_in_cache(inode
, min_key
.offset
, thresh
)) {
742 *off
= min_key
.offset
;
743 btrfs_free_path(path
);
747 if (min_key
.offset
== (u64
)-1)
751 btrfs_release_path(path
);
754 btrfs_free_path(path
);
758 static int should_defrag_range(struct inode
*inode
, u64 start
, u64 len
,
759 int thresh
, u64
*last_len
, u64
*skip
,
762 struct extent_io_tree
*io_tree
= &BTRFS_I(inode
)->io_tree
;
763 struct extent_map
*em
= NULL
;
764 struct extent_map_tree
*em_tree
= &BTRFS_I(inode
)->extent_tree
;
768 * make sure that once we start defragging and extent, we keep on
771 if (start
< *defrag_end
)
777 * hopefully we have this extent in the tree already, try without
778 * the full extent lock
780 read_lock(&em_tree
->lock
);
781 em
= lookup_extent_mapping(em_tree
, start
, len
);
782 read_unlock(&em_tree
->lock
);
785 /* get the big lock and read metadata off disk */
786 lock_extent(io_tree
, start
, start
+ len
- 1, GFP_NOFS
);
787 em
= btrfs_get_extent(inode
, NULL
, 0, start
, len
, 0);
788 unlock_extent(io_tree
, start
, start
+ len
- 1, GFP_NOFS
);
794 /* this will cover holes, and inline extents */
795 if (em
->block_start
>= EXTENT_MAP_LAST_BYTE
)
799 * we hit a real extent, if it is big don't bother defragging it again
801 if ((*last_len
== 0 || *last_len
>= thresh
) && em
->len
>= thresh
)
805 * last_len ends up being a counter of how many bytes we've defragged.
806 * every time we choose not to defrag an extent, we reset *last_len
807 * so that the next tiny extent will force a defrag.
809 * The end result of this is that tiny extents before a single big
810 * extent will force at least part of that big extent to be defragged.
814 *defrag_end
= extent_map_end(em
);
817 *skip
= extent_map_end(em
);
826 * it doesn't do much good to defrag one or two pages
827 * at a time. This pulls in a nice chunk of pages
830 * It also makes sure the delalloc code has enough
831 * dirty data to avoid making new small extents as part
834 * It's a good idea to start RA on this range
835 * before calling this.
837 static int cluster_pages_for_defrag(struct inode
*inode
,
839 unsigned long start_index
,
842 unsigned long file_end
;
843 u64 isize
= i_size_read(inode
);
849 struct btrfs_ordered_extent
*ordered
;
850 struct extent_state
*cached_state
= NULL
;
851 gfp_t mask
= btrfs_alloc_write_mask(inode
->i_mapping
);
855 file_end
= (isize
- 1) >> PAGE_CACHE_SHIFT
;
857 ret
= btrfs_delalloc_reserve_space(inode
,
858 num_pages
<< PAGE_CACHE_SHIFT
);
865 /* step one, lock all the pages */
866 for (i
= 0; i
< num_pages
; i
++) {
868 page
= find_or_create_page(inode
->i_mapping
,
869 start_index
+ i
, mask
);
873 if (!PageUptodate(page
)) {
874 btrfs_readpage(NULL
, page
);
876 if (!PageUptodate(page
)) {
878 page_cache_release(page
);
883 isize
= i_size_read(inode
);
884 file_end
= (isize
- 1) >> PAGE_CACHE_SHIFT
;
885 if (!isize
|| page
->index
> file_end
||
886 page
->mapping
!= inode
->i_mapping
) {
887 /* whoops, we blew past eof, skip this page */
889 page_cache_release(page
);
898 if (!(inode
->i_sb
->s_flags
& MS_ACTIVE
))
902 * so now we have a nice long stream of locked
903 * and up to date pages, lets wait on them
905 for (i
= 0; i
< i_done
; i
++)
906 wait_on_page_writeback(pages
[i
]);
908 page_start
= page_offset(pages
[0]);
909 page_end
= page_offset(pages
[i_done
- 1]) + PAGE_CACHE_SIZE
;
911 lock_extent_bits(&BTRFS_I(inode
)->io_tree
,
912 page_start
, page_end
- 1, 0, &cached_state
,
914 ordered
= btrfs_lookup_first_ordered_extent(inode
, page_end
- 1);
916 ordered
->file_offset
+ ordered
->len
> page_start
&&
917 ordered
->file_offset
< page_end
) {
918 btrfs_put_ordered_extent(ordered
);
919 unlock_extent_cached(&BTRFS_I(inode
)->io_tree
,
920 page_start
, page_end
- 1,
921 &cached_state
, GFP_NOFS
);
922 for (i
= 0; i
< i_done
; i
++) {
923 unlock_page(pages
[i
]);
924 page_cache_release(pages
[i
]);
926 btrfs_wait_ordered_range(inode
, page_start
,
927 page_end
- page_start
);
931 btrfs_put_ordered_extent(ordered
);
933 clear_extent_bit(&BTRFS_I(inode
)->io_tree
, page_start
,
934 page_end
- 1, EXTENT_DIRTY
| EXTENT_DELALLOC
|
935 EXTENT_DO_ACCOUNTING
, 0, 0, &cached_state
,
938 if (i_done
!= num_pages
) {
939 spin_lock(&BTRFS_I(inode
)->lock
);
940 BTRFS_I(inode
)->outstanding_extents
++;
941 spin_unlock(&BTRFS_I(inode
)->lock
);
942 btrfs_delalloc_release_space(inode
,
943 (num_pages
- i_done
) << PAGE_CACHE_SHIFT
);
947 btrfs_set_extent_delalloc(inode
, page_start
, page_end
- 1,
950 unlock_extent_cached(&BTRFS_I(inode
)->io_tree
,
951 page_start
, page_end
- 1, &cached_state
,
954 for (i
= 0; i
< i_done
; i
++) {
955 clear_page_dirty_for_io(pages
[i
]);
956 ClearPageChecked(pages
[i
]);
957 set_page_extent_mapped(pages
[i
]);
958 set_page_dirty(pages
[i
]);
959 unlock_page(pages
[i
]);
960 page_cache_release(pages
[i
]);
964 for (i
= 0; i
< i_done
; i
++) {
965 unlock_page(pages
[i
]);
966 page_cache_release(pages
[i
]);
968 btrfs_delalloc_release_space(inode
, num_pages
<< PAGE_CACHE_SHIFT
);
973 int btrfs_defrag_file(struct inode
*inode
, struct file
*file
,
974 struct btrfs_ioctl_defrag_range_args
*range
,
975 u64 newer_than
, unsigned long max_to_defrag
)
977 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
978 struct btrfs_super_block
*disk_super
;
979 struct file_ra_state
*ra
= NULL
;
980 unsigned long last_index
;
985 u64 newer_off
= range
->start
;
989 int defrag_count
= 0;
990 int compress_type
= BTRFS_COMPRESS_ZLIB
;
991 int extent_thresh
= range
->extent_thresh
;
992 int newer_cluster
= (256 * 1024) >> PAGE_CACHE_SHIFT
;
993 u64 new_align
= ~((u64
)128 * 1024 - 1);
994 struct page
**pages
= NULL
;
996 if (extent_thresh
== 0)
997 extent_thresh
= 256 * 1024;
999 if (range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
) {
1000 if (range
->compress_type
> BTRFS_COMPRESS_TYPES
)
1002 if (range
->compress_type
)
1003 compress_type
= range
->compress_type
;
1006 if (inode
->i_size
== 0)
1010 * if we were not given a file, allocate a readahead
1014 ra
= kzalloc(sizeof(*ra
), GFP_NOFS
);
1017 file_ra_state_init(ra
, inode
->i_mapping
);
1022 pages
= kmalloc(sizeof(struct page
*) * newer_cluster
,
1029 /* find the last page to defrag */
1030 if (range
->start
+ range
->len
> range
->start
) {
1031 last_index
= min_t(u64
, inode
->i_size
- 1,
1032 range
->start
+ range
->len
- 1) >> PAGE_CACHE_SHIFT
;
1034 last_index
= (inode
->i_size
- 1) >> PAGE_CACHE_SHIFT
;
1038 ret
= find_new_extents(root
, inode
, newer_than
,
1039 &newer_off
, 64 * 1024);
1041 range
->start
= newer_off
;
1043 * we always align our defrag to help keep
1044 * the extents in the file evenly spaced
1046 i
= (newer_off
& new_align
) >> PAGE_CACHE_SHIFT
;
1047 newer_left
= newer_cluster
;
1051 i
= range
->start
>> PAGE_CACHE_SHIFT
;
1054 max_to_defrag
= last_index
- 1;
1056 while (i
<= last_index
&& defrag_count
< max_to_defrag
) {
1058 * make sure we stop running if someone unmounts
1061 if (!(inode
->i_sb
->s_flags
& MS_ACTIVE
))
1065 !should_defrag_range(inode
, (u64
)i
<< PAGE_CACHE_SHIFT
,
1072 * the should_defrag function tells us how much to skip
1073 * bump our counter by the suggested amount
1075 next
= (skip
+ PAGE_CACHE_SIZE
- 1) >> PAGE_CACHE_SHIFT
;
1076 i
= max(i
+ 1, next
);
1079 if (range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
)
1080 BTRFS_I(inode
)->force_compress
= compress_type
;
1082 btrfs_force_ra(inode
->i_mapping
, ra
, file
, i
, newer_cluster
);
1084 ret
= cluster_pages_for_defrag(inode
, pages
, i
, newer_cluster
);
1088 defrag_count
+= ret
;
1089 balance_dirty_pages_ratelimited_nr(inode
->i_mapping
, ret
);
1093 if (newer_off
== (u64
)-1)
1096 newer_off
= max(newer_off
+ 1,
1097 (u64
)i
<< PAGE_CACHE_SHIFT
);
1099 ret
= find_new_extents(root
, inode
,
1100 newer_than
, &newer_off
,
1103 range
->start
= newer_off
;
1104 i
= (newer_off
& new_align
) >> PAGE_CACHE_SHIFT
;
1105 newer_left
= newer_cluster
;
1114 if ((range
->flags
& BTRFS_DEFRAG_RANGE_START_IO
))
1115 filemap_flush(inode
->i_mapping
);
1117 if ((range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
)) {
1118 /* the filemap_flush will queue IO into the worker threads, but
1119 * we have to make sure the IO is actually started and that
1120 * ordered extents get created before we return
1122 atomic_inc(&root
->fs_info
->async_submit_draining
);
1123 while (atomic_read(&root
->fs_info
->nr_async_submits
) ||
1124 atomic_read(&root
->fs_info
->async_delalloc_pages
)) {
1125 wait_event(root
->fs_info
->async_submit_wait
,
1126 (atomic_read(&root
->fs_info
->nr_async_submits
) == 0 &&
1127 atomic_read(&root
->fs_info
->async_delalloc_pages
) == 0));
1129 atomic_dec(&root
->fs_info
->async_submit_draining
);
1131 mutex_lock(&inode
->i_mutex
);
1132 BTRFS_I(inode
)->force_compress
= BTRFS_COMPRESS_NONE
;
1133 mutex_unlock(&inode
->i_mutex
);
1136 disk_super
= &root
->fs_info
->super_copy
;
1137 features
= btrfs_super_incompat_flags(disk_super
);
1138 if (range
->compress_type
== BTRFS_COMPRESS_LZO
) {
1139 features
|= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO
;
1140 btrfs_set_super_incompat_flags(disk_super
, features
);
1145 return defrag_count
;
1154 static noinline
int btrfs_ioctl_resize(struct btrfs_root
*root
,
1160 struct btrfs_ioctl_vol_args
*vol_args
;
1161 struct btrfs_trans_handle
*trans
;
1162 struct btrfs_device
*device
= NULL
;
1164 char *devstr
= NULL
;
1168 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
)
1171 if (!capable(CAP_SYS_ADMIN
))
1174 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1175 if (IS_ERR(vol_args
))
1176 return PTR_ERR(vol_args
);
1178 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
1180 mutex_lock(&root
->fs_info
->volume_mutex
);
1181 sizestr
= vol_args
->name
;
1182 devstr
= strchr(sizestr
, ':');
1185 sizestr
= devstr
+ 1;
1187 devstr
= vol_args
->name
;
1188 devid
= simple_strtoull(devstr
, &end
, 10);
1189 printk(KERN_INFO
"resizing devid %llu\n",
1190 (unsigned long long)devid
);
1192 device
= btrfs_find_device(root
, devid
, NULL
, NULL
);
1194 printk(KERN_INFO
"resizer unable to find device %llu\n",
1195 (unsigned long long)devid
);
1199 if (!strcmp(sizestr
, "max"))
1200 new_size
= device
->bdev
->bd_inode
->i_size
;
1202 if (sizestr
[0] == '-') {
1205 } else if (sizestr
[0] == '+') {
1209 new_size
= memparse(sizestr
, NULL
);
1210 if (new_size
== 0) {
1216 old_size
= device
->total_bytes
;
1219 if (new_size
> old_size
) {
1223 new_size
= old_size
- new_size
;
1224 } else if (mod
> 0) {
1225 new_size
= old_size
+ new_size
;
1228 if (new_size
< 256 * 1024 * 1024) {
1232 if (new_size
> device
->bdev
->bd_inode
->i_size
) {
1237 do_div(new_size
, root
->sectorsize
);
1238 new_size
*= root
->sectorsize
;
1240 printk(KERN_INFO
"new size for %s is %llu\n",
1241 device
->name
, (unsigned long long)new_size
);
1243 if (new_size
> old_size
) {
1244 trans
= btrfs_start_transaction(root
, 0);
1245 if (IS_ERR(trans
)) {
1246 ret
= PTR_ERR(trans
);
1249 ret
= btrfs_grow_device(trans
, device
, new_size
);
1250 btrfs_commit_transaction(trans
, root
);
1252 ret
= btrfs_shrink_device(device
, new_size
);
1256 mutex_unlock(&root
->fs_info
->volume_mutex
);
1261 static noinline
int btrfs_ioctl_snap_create_transid(struct file
*file
,
1268 struct btrfs_root
*root
= BTRFS_I(fdentry(file
)->d_inode
)->root
;
1269 struct file
*src_file
;
1273 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
)
1276 namelen
= strlen(name
);
1277 if (strchr(name
, '/')) {
1283 ret
= btrfs_mksubvol(&file
->f_path
, name
, namelen
,
1284 NULL
, transid
, readonly
);
1286 struct inode
*src_inode
;
1287 src_file
= fget(fd
);
1293 src_inode
= src_file
->f_path
.dentry
->d_inode
;
1294 if (src_inode
->i_sb
!= file
->f_path
.dentry
->d_inode
->i_sb
) {
1295 printk(KERN_INFO
"btrfs: Snapshot src from "
1301 ret
= btrfs_mksubvol(&file
->f_path
, name
, namelen
,
1302 BTRFS_I(src_inode
)->root
,
1310 static noinline
int btrfs_ioctl_snap_create(struct file
*file
,
1311 void __user
*arg
, int subvol
)
1313 struct btrfs_ioctl_vol_args
*vol_args
;
1316 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1317 if (IS_ERR(vol_args
))
1318 return PTR_ERR(vol_args
);
1319 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
1321 ret
= btrfs_ioctl_snap_create_transid(file
, vol_args
->name
,
1322 vol_args
->fd
, subvol
,
1329 static noinline
int btrfs_ioctl_snap_create_v2(struct file
*file
,
1330 void __user
*arg
, int subvol
)
1332 struct btrfs_ioctl_vol_args_v2
*vol_args
;
1336 bool readonly
= false;
1338 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1339 if (IS_ERR(vol_args
))
1340 return PTR_ERR(vol_args
);
1341 vol_args
->name
[BTRFS_SUBVOL_NAME_MAX
] = '\0';
1343 if (vol_args
->flags
&
1344 ~(BTRFS_SUBVOL_CREATE_ASYNC
| BTRFS_SUBVOL_RDONLY
)) {
1349 if (vol_args
->flags
& BTRFS_SUBVOL_CREATE_ASYNC
)
1351 if (vol_args
->flags
& BTRFS_SUBVOL_RDONLY
)
1354 ret
= btrfs_ioctl_snap_create_transid(file
, vol_args
->name
,
1355 vol_args
->fd
, subvol
,
1358 if (ret
== 0 && ptr
&&
1360 offsetof(struct btrfs_ioctl_vol_args_v2
,
1361 transid
), ptr
, sizeof(*ptr
)))
1368 static noinline
int btrfs_ioctl_subvol_getflags(struct file
*file
,
1371 struct inode
*inode
= fdentry(file
)->d_inode
;
1372 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
1376 if (btrfs_ino(inode
) != BTRFS_FIRST_FREE_OBJECTID
)
1379 down_read(&root
->fs_info
->subvol_sem
);
1380 if (btrfs_root_readonly(root
))
1381 flags
|= BTRFS_SUBVOL_RDONLY
;
1382 up_read(&root
->fs_info
->subvol_sem
);
1384 if (copy_to_user(arg
, &flags
, sizeof(flags
)))
1390 static noinline
int btrfs_ioctl_subvol_setflags(struct file
*file
,
1393 struct inode
*inode
= fdentry(file
)->d_inode
;
1394 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
1395 struct btrfs_trans_handle
*trans
;
1400 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
)
1403 if (btrfs_ino(inode
) != BTRFS_FIRST_FREE_OBJECTID
)
1406 if (copy_from_user(&flags
, arg
, sizeof(flags
)))
1409 if (flags
& BTRFS_SUBVOL_CREATE_ASYNC
)
1412 if (flags
& ~BTRFS_SUBVOL_RDONLY
)
1415 if (!inode_owner_or_capable(inode
))
1418 down_write(&root
->fs_info
->subvol_sem
);
1421 if (!!(flags
& BTRFS_SUBVOL_RDONLY
) == btrfs_root_readonly(root
))
1424 root_flags
= btrfs_root_flags(&root
->root_item
);
1425 if (flags
& BTRFS_SUBVOL_RDONLY
)
1426 btrfs_set_root_flags(&root
->root_item
,
1427 root_flags
| BTRFS_ROOT_SUBVOL_RDONLY
);
1429 btrfs_set_root_flags(&root
->root_item
,
1430 root_flags
& ~BTRFS_ROOT_SUBVOL_RDONLY
);
1432 trans
= btrfs_start_transaction(root
, 1);
1433 if (IS_ERR(trans
)) {
1434 ret
= PTR_ERR(trans
);
1438 ret
= btrfs_update_root(trans
, root
->fs_info
->tree_root
,
1439 &root
->root_key
, &root
->root_item
);
1441 btrfs_commit_transaction(trans
, root
);
1444 btrfs_set_root_flags(&root
->root_item
, root_flags
);
1446 up_write(&root
->fs_info
->subvol_sem
);
1451 * helper to check if the subvolume references other subvolumes
1453 static noinline
int may_destroy_subvol(struct btrfs_root
*root
)
1455 struct btrfs_path
*path
;
1456 struct btrfs_key key
;
1459 path
= btrfs_alloc_path();
1463 key
.objectid
= root
->root_key
.objectid
;
1464 key
.type
= BTRFS_ROOT_REF_KEY
;
1465 key
.offset
= (u64
)-1;
1467 ret
= btrfs_search_slot(NULL
, root
->fs_info
->tree_root
,
1474 if (path
->slots
[0] > 0) {
1476 btrfs_item_key_to_cpu(path
->nodes
[0], &key
, path
->slots
[0]);
1477 if (key
.objectid
== root
->root_key
.objectid
&&
1478 key
.type
== BTRFS_ROOT_REF_KEY
)
1482 btrfs_free_path(path
);
1486 static noinline
int key_in_sk(struct btrfs_key
*key
,
1487 struct btrfs_ioctl_search_key
*sk
)
1489 struct btrfs_key test
;
1492 test
.objectid
= sk
->min_objectid
;
1493 test
.type
= sk
->min_type
;
1494 test
.offset
= sk
->min_offset
;
1496 ret
= btrfs_comp_cpu_keys(key
, &test
);
1500 test
.objectid
= sk
->max_objectid
;
1501 test
.type
= sk
->max_type
;
1502 test
.offset
= sk
->max_offset
;
1504 ret
= btrfs_comp_cpu_keys(key
, &test
);
1510 static noinline
int copy_to_sk(struct btrfs_root
*root
,
1511 struct btrfs_path
*path
,
1512 struct btrfs_key
*key
,
1513 struct btrfs_ioctl_search_key
*sk
,
1515 unsigned long *sk_offset
,
1519 struct extent_buffer
*leaf
;
1520 struct btrfs_ioctl_search_header sh
;
1521 unsigned long item_off
;
1522 unsigned long item_len
;
1528 leaf
= path
->nodes
[0];
1529 slot
= path
->slots
[0];
1530 nritems
= btrfs_header_nritems(leaf
);
1532 if (btrfs_header_generation(leaf
) > sk
->max_transid
) {
1536 found_transid
= btrfs_header_generation(leaf
);
1538 for (i
= slot
; i
< nritems
; i
++) {
1539 item_off
= btrfs_item_ptr_offset(leaf
, i
);
1540 item_len
= btrfs_item_size_nr(leaf
, i
);
1542 if (item_len
> BTRFS_SEARCH_ARGS_BUFSIZE
)
1545 if (sizeof(sh
) + item_len
+ *sk_offset
>
1546 BTRFS_SEARCH_ARGS_BUFSIZE
) {
1551 btrfs_item_key_to_cpu(leaf
, key
, i
);
1552 if (!key_in_sk(key
, sk
))
1555 sh
.objectid
= key
->objectid
;
1556 sh
.offset
= key
->offset
;
1557 sh
.type
= key
->type
;
1559 sh
.transid
= found_transid
;
1561 /* copy search result header */
1562 memcpy(buf
+ *sk_offset
, &sh
, sizeof(sh
));
1563 *sk_offset
+= sizeof(sh
);
1566 char *p
= buf
+ *sk_offset
;
1568 read_extent_buffer(leaf
, p
,
1569 item_off
, item_len
);
1570 *sk_offset
+= item_len
;
1574 if (*num_found
>= sk
->nr_items
)
1579 if (key
->offset
< (u64
)-1 && key
->offset
< sk
->max_offset
)
1581 else if (key
->type
< (u8
)-1 && key
->type
< sk
->max_type
) {
1584 } else if (key
->objectid
< (u64
)-1 && key
->objectid
< sk
->max_objectid
) {
1594 static noinline
int search_ioctl(struct inode
*inode
,
1595 struct btrfs_ioctl_search_args
*args
)
1597 struct btrfs_root
*root
;
1598 struct btrfs_key key
;
1599 struct btrfs_key max_key
;
1600 struct btrfs_path
*path
;
1601 struct btrfs_ioctl_search_key
*sk
= &args
->key
;
1602 struct btrfs_fs_info
*info
= BTRFS_I(inode
)->root
->fs_info
;
1605 unsigned long sk_offset
= 0;
1607 path
= btrfs_alloc_path();
1611 if (sk
->tree_id
== 0) {
1612 /* search the root of the inode that was passed */
1613 root
= BTRFS_I(inode
)->root
;
1615 key
.objectid
= sk
->tree_id
;
1616 key
.type
= BTRFS_ROOT_ITEM_KEY
;
1617 key
.offset
= (u64
)-1;
1618 root
= btrfs_read_fs_root_no_name(info
, &key
);
1620 printk(KERN_ERR
"could not find root %llu\n",
1622 btrfs_free_path(path
);
1627 key
.objectid
= sk
->min_objectid
;
1628 key
.type
= sk
->min_type
;
1629 key
.offset
= sk
->min_offset
;
1631 max_key
.objectid
= sk
->max_objectid
;
1632 max_key
.type
= sk
->max_type
;
1633 max_key
.offset
= sk
->max_offset
;
1635 path
->keep_locks
= 1;
1638 ret
= btrfs_search_forward(root
, &key
, &max_key
, path
, 0,
1645 ret
= copy_to_sk(root
, path
, &key
, sk
, args
->buf
,
1646 &sk_offset
, &num_found
);
1647 btrfs_release_path(path
);
1648 if (ret
|| num_found
>= sk
->nr_items
)
1654 sk
->nr_items
= num_found
;
1655 btrfs_free_path(path
);
1659 static noinline
int btrfs_ioctl_tree_search(struct file
*file
,
1662 struct btrfs_ioctl_search_args
*args
;
1663 struct inode
*inode
;
1666 if (!capable(CAP_SYS_ADMIN
))
1669 args
= memdup_user(argp
, sizeof(*args
));
1671 return PTR_ERR(args
);
1673 inode
= fdentry(file
)->d_inode
;
1674 ret
= search_ioctl(inode
, args
);
1675 if (ret
== 0 && copy_to_user(argp
, args
, sizeof(*args
)))
1682 * Search INODE_REFs to identify path name of 'dirid' directory
1683 * in a 'tree_id' tree. and sets path name to 'name'.
1685 static noinline
int btrfs_search_path_in_tree(struct btrfs_fs_info
*info
,
1686 u64 tree_id
, u64 dirid
, char *name
)
1688 struct btrfs_root
*root
;
1689 struct btrfs_key key
;
1695 struct btrfs_inode_ref
*iref
;
1696 struct extent_buffer
*l
;
1697 struct btrfs_path
*path
;
1699 if (dirid
== BTRFS_FIRST_FREE_OBJECTID
) {
1704 path
= btrfs_alloc_path();
1708 ptr
= &name
[BTRFS_INO_LOOKUP_PATH_MAX
];
1710 key
.objectid
= tree_id
;
1711 key
.type
= BTRFS_ROOT_ITEM_KEY
;
1712 key
.offset
= (u64
)-1;
1713 root
= btrfs_read_fs_root_no_name(info
, &key
);
1715 printk(KERN_ERR
"could not find root %llu\n", tree_id
);
1720 key
.objectid
= dirid
;
1721 key
.type
= BTRFS_INODE_REF_KEY
;
1722 key
.offset
= (u64
)-1;
1725 ret
= btrfs_search_slot(NULL
, root
, &key
, path
, 0, 0);
1730 slot
= path
->slots
[0];
1731 if (ret
> 0 && slot
> 0)
1733 btrfs_item_key_to_cpu(l
, &key
, slot
);
1735 if (ret
> 0 && (key
.objectid
!= dirid
||
1736 key
.type
!= BTRFS_INODE_REF_KEY
)) {
1741 iref
= btrfs_item_ptr(l
, slot
, struct btrfs_inode_ref
);
1742 len
= btrfs_inode_ref_name_len(l
, iref
);
1744 total_len
+= len
+ 1;
1749 read_extent_buffer(l
, ptr
,(unsigned long)(iref
+ 1), len
);
1751 if (key
.offset
== BTRFS_FIRST_FREE_OBJECTID
)
1754 btrfs_release_path(path
);
1755 key
.objectid
= key
.offset
;
1756 key
.offset
= (u64
)-1;
1757 dirid
= key
.objectid
;
1761 memmove(name
, ptr
, total_len
);
1762 name
[total_len
]='\0';
1765 btrfs_free_path(path
);
1769 static noinline
int btrfs_ioctl_ino_lookup(struct file
*file
,
1772 struct btrfs_ioctl_ino_lookup_args
*args
;
1773 struct inode
*inode
;
1776 if (!capable(CAP_SYS_ADMIN
))
1779 args
= memdup_user(argp
, sizeof(*args
));
1781 return PTR_ERR(args
);
1783 inode
= fdentry(file
)->d_inode
;
1785 if (args
->treeid
== 0)
1786 args
->treeid
= BTRFS_I(inode
)->root
->root_key
.objectid
;
1788 ret
= btrfs_search_path_in_tree(BTRFS_I(inode
)->root
->fs_info
,
1789 args
->treeid
, args
->objectid
,
1792 if (ret
== 0 && copy_to_user(argp
, args
, sizeof(*args
)))
1799 static noinline
int btrfs_ioctl_snap_destroy(struct file
*file
,
1802 struct dentry
*parent
= fdentry(file
);
1803 struct dentry
*dentry
;
1804 struct inode
*dir
= parent
->d_inode
;
1805 struct inode
*inode
;
1806 struct btrfs_root
*root
= BTRFS_I(dir
)->root
;
1807 struct btrfs_root
*dest
= NULL
;
1808 struct btrfs_ioctl_vol_args
*vol_args
;
1809 struct btrfs_trans_handle
*trans
;
1814 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1815 if (IS_ERR(vol_args
))
1816 return PTR_ERR(vol_args
);
1818 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
1819 namelen
= strlen(vol_args
->name
);
1820 if (strchr(vol_args
->name
, '/') ||
1821 strncmp(vol_args
->name
, "..", namelen
) == 0) {
1826 err
= mnt_want_write(file
->f_path
.mnt
);
1830 mutex_lock_nested(&dir
->i_mutex
, I_MUTEX_PARENT
);
1831 dentry
= lookup_one_len(vol_args
->name
, parent
, namelen
);
1832 if (IS_ERR(dentry
)) {
1833 err
= PTR_ERR(dentry
);
1834 goto out_unlock_dir
;
1837 if (!dentry
->d_inode
) {
1842 inode
= dentry
->d_inode
;
1843 dest
= BTRFS_I(inode
)->root
;
1844 if (!capable(CAP_SYS_ADMIN
)){
1846 * Regular user. Only allow this with a special mount
1847 * option, when the user has write+exec access to the
1848 * subvol root, and when rmdir(2) would have been
1851 * Note that this is _not_ check that the subvol is
1852 * empty or doesn't contain data that we wouldn't
1853 * otherwise be able to delete.
1855 * Users who want to delete empty subvols should try
1859 if (!btrfs_test_opt(root
, USER_SUBVOL_RM_ALLOWED
))
1863 * Do not allow deletion if the parent dir is the same
1864 * as the dir to be deleted. That means the ioctl
1865 * must be called on the dentry referencing the root
1866 * of the subvol, not a random directory contained
1873 err
= inode_permission(inode
, MAY_WRITE
| MAY_EXEC
);
1877 /* check if subvolume may be deleted by a non-root user */
1878 err
= btrfs_may_delete(dir
, dentry
, 1);
1883 if (btrfs_ino(inode
) != BTRFS_FIRST_FREE_OBJECTID
) {
1888 mutex_lock(&inode
->i_mutex
);
1889 err
= d_invalidate(dentry
);
1893 down_write(&root
->fs_info
->subvol_sem
);
1895 err
= may_destroy_subvol(dest
);
1899 trans
= btrfs_start_transaction(root
, 0);
1900 if (IS_ERR(trans
)) {
1901 err
= PTR_ERR(trans
);
1904 trans
->block_rsv
= &root
->fs_info
->global_block_rsv
;
1906 ret
= btrfs_unlink_subvol(trans
, root
, dir
,
1907 dest
->root_key
.objectid
,
1908 dentry
->d_name
.name
,
1909 dentry
->d_name
.len
);
1912 btrfs_record_root_in_trans(trans
, dest
);
1914 memset(&dest
->root_item
.drop_progress
, 0,
1915 sizeof(dest
->root_item
.drop_progress
));
1916 dest
->root_item
.drop_level
= 0;
1917 btrfs_set_root_refs(&dest
->root_item
, 0);
1919 if (!xchg(&dest
->orphan_item_inserted
, 1)) {
1920 ret
= btrfs_insert_orphan_item(trans
,
1921 root
->fs_info
->tree_root
,
1922 dest
->root_key
.objectid
);
1926 ret
= btrfs_end_transaction(trans
, root
);
1928 inode
->i_flags
|= S_DEAD
;
1930 up_write(&root
->fs_info
->subvol_sem
);
1932 mutex_unlock(&inode
->i_mutex
);
1934 shrink_dcache_sb(root
->fs_info
->sb
);
1935 btrfs_invalidate_inodes(dest
);
1941 mutex_unlock(&dir
->i_mutex
);
1942 mnt_drop_write(file
->f_path
.mnt
);
1948 static int btrfs_ioctl_defrag(struct file
*file
, void __user
*argp
)
1950 struct inode
*inode
= fdentry(file
)->d_inode
;
1951 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
1952 struct btrfs_ioctl_defrag_range_args
*range
;
1955 if (btrfs_root_readonly(root
))
1958 ret
= mnt_want_write(file
->f_path
.mnt
);
1962 switch (inode
->i_mode
& S_IFMT
) {
1964 if (!capable(CAP_SYS_ADMIN
)) {
1968 ret
= btrfs_defrag_root(root
, 0);
1971 ret
= btrfs_defrag_root(root
->fs_info
->extent_root
, 0);
1974 if (!(file
->f_mode
& FMODE_WRITE
)) {
1979 range
= kzalloc(sizeof(*range
), GFP_KERNEL
);
1986 if (copy_from_user(range
, argp
,
1992 /* compression requires us to start the IO */
1993 if ((range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
)) {
1994 range
->flags
|= BTRFS_DEFRAG_RANGE_START_IO
;
1995 range
->extent_thresh
= (u32
)-1;
1998 /* the rest are all set to zero by kzalloc */
1999 range
->len
= (u64
)-1;
2001 ret
= btrfs_defrag_file(fdentry(file
)->d_inode
, file
,
2011 mnt_drop_write(file
->f_path
.mnt
);
2015 static long btrfs_ioctl_add_dev(struct btrfs_root
*root
, void __user
*arg
)
2017 struct btrfs_ioctl_vol_args
*vol_args
;
2020 if (!capable(CAP_SYS_ADMIN
))
2023 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
2024 if (IS_ERR(vol_args
))
2025 return PTR_ERR(vol_args
);
2027 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
2028 ret
= btrfs_init_new_device(root
, vol_args
->name
);
2034 static long btrfs_ioctl_rm_dev(struct btrfs_root
*root
, void __user
*arg
)
2036 struct btrfs_ioctl_vol_args
*vol_args
;
2039 if (!capable(CAP_SYS_ADMIN
))
2042 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
)
2045 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
2046 if (IS_ERR(vol_args
))
2047 return PTR_ERR(vol_args
);
2049 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
2050 ret
= btrfs_rm_device(root
, vol_args
->name
);
2056 static long btrfs_ioctl_fs_info(struct btrfs_root
*root
, void __user
*arg
)
2058 struct btrfs_ioctl_fs_info_args
*fi_args
;
2059 struct btrfs_device
*device
;
2060 struct btrfs_device
*next
;
2061 struct btrfs_fs_devices
*fs_devices
= root
->fs_info
->fs_devices
;
2064 if (!capable(CAP_SYS_ADMIN
))
2067 fi_args
= kzalloc(sizeof(*fi_args
), GFP_KERNEL
);
2071 fi_args
->num_devices
= fs_devices
->num_devices
;
2072 memcpy(&fi_args
->fsid
, root
->fs_info
->fsid
, sizeof(fi_args
->fsid
));
2074 mutex_lock(&fs_devices
->device_list_mutex
);
2075 list_for_each_entry_safe(device
, next
, &fs_devices
->devices
, dev_list
) {
2076 if (device
->devid
> fi_args
->max_id
)
2077 fi_args
->max_id
= device
->devid
;
2079 mutex_unlock(&fs_devices
->device_list_mutex
);
2081 if (copy_to_user(arg
, fi_args
, sizeof(*fi_args
)))
2088 static long btrfs_ioctl_dev_info(struct btrfs_root
*root
, void __user
*arg
)
2090 struct btrfs_ioctl_dev_info_args
*di_args
;
2091 struct btrfs_device
*dev
;
2092 struct btrfs_fs_devices
*fs_devices
= root
->fs_info
->fs_devices
;
2094 char *s_uuid
= NULL
;
2095 char empty_uuid
[BTRFS_UUID_SIZE
] = {0};
2097 if (!capable(CAP_SYS_ADMIN
))
2100 di_args
= memdup_user(arg
, sizeof(*di_args
));
2101 if (IS_ERR(di_args
))
2102 return PTR_ERR(di_args
);
2104 if (memcmp(empty_uuid
, di_args
->uuid
, BTRFS_UUID_SIZE
) != 0)
2105 s_uuid
= di_args
->uuid
;
2107 mutex_lock(&fs_devices
->device_list_mutex
);
2108 dev
= btrfs_find_device(root
, di_args
->devid
, s_uuid
, NULL
);
2109 mutex_unlock(&fs_devices
->device_list_mutex
);
2116 di_args
->devid
= dev
->devid
;
2117 di_args
->bytes_used
= dev
->bytes_used
;
2118 di_args
->total_bytes
= dev
->total_bytes
;
2119 memcpy(di_args
->uuid
, dev
->uuid
, sizeof(di_args
->uuid
));
2120 strncpy(di_args
->path
, dev
->name
, sizeof(di_args
->path
));
2123 if (ret
== 0 && copy_to_user(arg
, di_args
, sizeof(*di_args
)))
2130 static noinline
long btrfs_ioctl_clone(struct file
*file
, unsigned long srcfd
,
2131 u64 off
, u64 olen
, u64 destoff
)
2133 struct inode
*inode
= fdentry(file
)->d_inode
;
2134 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
2135 struct file
*src_file
;
2137 struct btrfs_trans_handle
*trans
;
2138 struct btrfs_path
*path
;
2139 struct extent_buffer
*leaf
;
2141 struct btrfs_key key
;
2146 u64 bs
= root
->fs_info
->sb
->s_blocksize
;
2151 * - split compressed inline extents. annoying: we need to
2152 * decompress into destination's address_space (the file offset
2153 * may change, so source mapping won't do), then recompress (or
2154 * otherwise reinsert) a subrange.
2155 * - allow ranges within the same file to be cloned (provided
2156 * they don't overlap)?
2159 /* the destination must be opened for writing */
2160 if (!(file
->f_mode
& FMODE_WRITE
) || (file
->f_flags
& O_APPEND
))
2163 if (btrfs_root_readonly(root
))
2166 ret
= mnt_want_write(file
->f_path
.mnt
);
2170 src_file
= fget(srcfd
);
2173 goto out_drop_write
;
2176 src
= src_file
->f_dentry
->d_inode
;
2182 /* the src must be open for reading */
2183 if (!(src_file
->f_mode
& FMODE_READ
))
2186 /* don't make the dst file partly checksummed */
2187 if ((BTRFS_I(src
)->flags
& BTRFS_INODE_NODATASUM
) !=
2188 (BTRFS_I(inode
)->flags
& BTRFS_INODE_NODATASUM
))
2192 if (S_ISDIR(src
->i_mode
) || S_ISDIR(inode
->i_mode
))
2196 if (src
->i_sb
!= inode
->i_sb
|| BTRFS_I(src
)->root
!= root
)
2200 buf
= vmalloc(btrfs_level_size(root
, 0));
2204 path
= btrfs_alloc_path();
2212 mutex_lock_nested(&inode
->i_mutex
, I_MUTEX_PARENT
);
2213 mutex_lock_nested(&src
->i_mutex
, I_MUTEX_CHILD
);
2215 mutex_lock_nested(&src
->i_mutex
, I_MUTEX_PARENT
);
2216 mutex_lock_nested(&inode
->i_mutex
, I_MUTEX_CHILD
);
2219 /* determine range to clone */
2221 if (off
+ len
> src
->i_size
|| off
+ len
< off
)
2224 olen
= len
= src
->i_size
- off
;
2225 /* if we extend to eof, continue to block boundary */
2226 if (off
+ len
== src
->i_size
)
2227 len
= ALIGN(src
->i_size
, bs
) - off
;
2229 /* verify the end result is block aligned */
2230 if (!IS_ALIGNED(off
, bs
) || !IS_ALIGNED(off
+ len
, bs
) ||
2231 !IS_ALIGNED(destoff
, bs
))
2234 if (destoff
> inode
->i_size
) {
2235 ret
= btrfs_cont_expand(inode
, inode
->i_size
, destoff
);
2240 /* truncate page cache pages from target inode range */
2241 truncate_inode_pages_range(&inode
->i_data
, destoff
,
2242 PAGE_CACHE_ALIGN(destoff
+ len
) - 1);
2244 /* do any pending delalloc/csum calc on src, one way or
2245 another, and lock file content */
2247 struct btrfs_ordered_extent
*ordered
;
2248 lock_extent(&BTRFS_I(src
)->io_tree
, off
, off
+len
, GFP_NOFS
);
2249 ordered
= btrfs_lookup_first_ordered_extent(src
, off
+len
);
2251 !test_range_bit(&BTRFS_I(src
)->io_tree
, off
, off
+len
,
2252 EXTENT_DELALLOC
, 0, NULL
))
2254 unlock_extent(&BTRFS_I(src
)->io_tree
, off
, off
+len
, GFP_NOFS
);
2256 btrfs_put_ordered_extent(ordered
);
2257 btrfs_wait_ordered_range(src
, off
, len
);
2261 key
.objectid
= btrfs_ino(src
);
2262 key
.type
= BTRFS_EXTENT_DATA_KEY
;
2267 * note the key will change type as we walk through the
2270 ret
= btrfs_search_slot(NULL
, root
, &key
, path
, 0, 0);
2274 nritems
= btrfs_header_nritems(path
->nodes
[0]);
2275 if (path
->slots
[0] >= nritems
) {
2276 ret
= btrfs_next_leaf(root
, path
);
2281 nritems
= btrfs_header_nritems(path
->nodes
[0]);
2283 leaf
= path
->nodes
[0];
2284 slot
= path
->slots
[0];
2286 btrfs_item_key_to_cpu(leaf
, &key
, slot
);
2287 if (btrfs_key_type(&key
) > BTRFS_EXTENT_DATA_KEY
||
2288 key
.objectid
!= btrfs_ino(src
))
2291 if (btrfs_key_type(&key
) == BTRFS_EXTENT_DATA_KEY
) {
2292 struct btrfs_file_extent_item
*extent
;
2295 struct btrfs_key new_key
;
2296 u64 disko
= 0, diskl
= 0;
2297 u64 datao
= 0, datal
= 0;
2301 size
= btrfs_item_size_nr(leaf
, slot
);
2302 read_extent_buffer(leaf
, buf
,
2303 btrfs_item_ptr_offset(leaf
, slot
),
2306 extent
= btrfs_item_ptr(leaf
, slot
,
2307 struct btrfs_file_extent_item
);
2308 comp
= btrfs_file_extent_compression(leaf
, extent
);
2309 type
= btrfs_file_extent_type(leaf
, extent
);
2310 if (type
== BTRFS_FILE_EXTENT_REG
||
2311 type
== BTRFS_FILE_EXTENT_PREALLOC
) {
2312 disko
= btrfs_file_extent_disk_bytenr(leaf
,
2314 diskl
= btrfs_file_extent_disk_num_bytes(leaf
,
2316 datao
= btrfs_file_extent_offset(leaf
, extent
);
2317 datal
= btrfs_file_extent_num_bytes(leaf
,
2319 } else if (type
== BTRFS_FILE_EXTENT_INLINE
) {
2320 /* take upper bound, may be compressed */
2321 datal
= btrfs_file_extent_ram_bytes(leaf
,
2324 btrfs_release_path(path
);
2326 if (key
.offset
+ datal
<= off
||
2327 key
.offset
>= off
+len
)
2330 memcpy(&new_key
, &key
, sizeof(new_key
));
2331 new_key
.objectid
= btrfs_ino(inode
);
2332 if (off
<= key
.offset
)
2333 new_key
.offset
= key
.offset
+ destoff
- off
;
2335 new_key
.offset
= destoff
;
2338 * 1 - adjusting old extent (we may have to split it)
2339 * 1 - add new extent
2342 trans
= btrfs_start_transaction(root
, 3);
2343 if (IS_ERR(trans
)) {
2344 ret
= PTR_ERR(trans
);
2348 if (type
== BTRFS_FILE_EXTENT_REG
||
2349 type
== BTRFS_FILE_EXTENT_PREALLOC
) {
2351 * a | --- range to clone ---| b
2352 * | ------------- extent ------------- |
2355 /* substract range b */
2356 if (key
.offset
+ datal
> off
+ len
)
2357 datal
= off
+ len
- key
.offset
;
2359 /* substract range a */
2360 if (off
> key
.offset
) {
2361 datao
+= off
- key
.offset
;
2362 datal
-= off
- key
.offset
;
2365 ret
= btrfs_drop_extents(trans
, inode
,
2367 new_key
.offset
+ datal
,
2371 ret
= btrfs_insert_empty_item(trans
, root
, path
,
2375 leaf
= path
->nodes
[0];
2376 slot
= path
->slots
[0];
2377 write_extent_buffer(leaf
, buf
,
2378 btrfs_item_ptr_offset(leaf
, slot
),
2381 extent
= btrfs_item_ptr(leaf
, slot
,
2382 struct btrfs_file_extent_item
);
2384 /* disko == 0 means it's a hole */
2388 btrfs_set_file_extent_offset(leaf
, extent
,
2390 btrfs_set_file_extent_num_bytes(leaf
, extent
,
2393 inode_add_bytes(inode
, datal
);
2394 ret
= btrfs_inc_extent_ref(trans
, root
,
2396 root
->root_key
.objectid
,
2398 new_key
.offset
- datao
);
2401 } else if (type
== BTRFS_FILE_EXTENT_INLINE
) {
2404 if (off
> key
.offset
) {
2405 skip
= off
- key
.offset
;
2406 new_key
.offset
+= skip
;
2409 if (key
.offset
+ datal
> off
+len
)
2410 trim
= key
.offset
+ datal
- (off
+len
);
2412 if (comp
&& (skip
|| trim
)) {
2414 btrfs_end_transaction(trans
, root
);
2417 size
-= skip
+ trim
;
2418 datal
-= skip
+ trim
;
2420 ret
= btrfs_drop_extents(trans
, inode
,
2422 new_key
.offset
+ datal
,
2426 ret
= btrfs_insert_empty_item(trans
, root
, path
,
2432 btrfs_file_extent_calc_inline_size(0);
2433 memmove(buf
+start
, buf
+start
+skip
,
2437 leaf
= path
->nodes
[0];
2438 slot
= path
->slots
[0];
2439 write_extent_buffer(leaf
, buf
,
2440 btrfs_item_ptr_offset(leaf
, slot
),
2442 inode_add_bytes(inode
, datal
);
2445 btrfs_mark_buffer_dirty(leaf
);
2446 btrfs_release_path(path
);
2448 inode
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
2451 * we round up to the block size at eof when
2452 * determining which extents to clone above,
2453 * but shouldn't round up the file size
2455 endoff
= new_key
.offset
+ datal
;
2456 if (endoff
> destoff
+olen
)
2457 endoff
= destoff
+olen
;
2458 if (endoff
> inode
->i_size
)
2459 btrfs_i_size_write(inode
, endoff
);
2461 ret
= btrfs_update_inode(trans
, root
, inode
);
2463 btrfs_end_transaction(trans
, root
);
2466 btrfs_release_path(path
);
2471 btrfs_release_path(path
);
2472 unlock_extent(&BTRFS_I(src
)->io_tree
, off
, off
+len
, GFP_NOFS
);
2474 mutex_unlock(&src
->i_mutex
);
2475 mutex_unlock(&inode
->i_mutex
);
2477 btrfs_free_path(path
);
2481 mnt_drop_write(file
->f_path
.mnt
);
2485 static long btrfs_ioctl_clone_range(struct file
*file
, void __user
*argp
)
2487 struct btrfs_ioctl_clone_range_args args
;
2489 if (copy_from_user(&args
, argp
, sizeof(args
)))
2491 return btrfs_ioctl_clone(file
, args
.src_fd
, args
.src_offset
,
2492 args
.src_length
, args
.dest_offset
);
2496 * there are many ways the trans_start and trans_end ioctls can lead
2497 * to deadlocks. They should only be used by applications that
2498 * basically own the machine, and have a very in depth understanding
2499 * of all the possible deadlocks and enospc problems.
2501 static long btrfs_ioctl_trans_start(struct file
*file
)
2503 struct inode
*inode
= fdentry(file
)->d_inode
;
2504 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
2505 struct btrfs_trans_handle
*trans
;
2509 if (!capable(CAP_SYS_ADMIN
))
2513 if (file
->private_data
)
2517 if (btrfs_root_readonly(root
))
2520 ret
= mnt_want_write(file
->f_path
.mnt
);
2524 atomic_inc(&root
->fs_info
->open_ioctl_trans
);
2527 trans
= btrfs_start_ioctl_transaction(root
);
2531 file
->private_data
= trans
;
2535 atomic_dec(&root
->fs_info
->open_ioctl_trans
);
2536 mnt_drop_write(file
->f_path
.mnt
);
2541 static long btrfs_ioctl_default_subvol(struct file
*file
, void __user
*argp
)
2543 struct inode
*inode
= fdentry(file
)->d_inode
;
2544 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
2545 struct btrfs_root
*new_root
;
2546 struct btrfs_dir_item
*di
;
2547 struct btrfs_trans_handle
*trans
;
2548 struct btrfs_path
*path
;
2549 struct btrfs_key location
;
2550 struct btrfs_disk_key disk_key
;
2551 struct btrfs_super_block
*disk_super
;
2556 if (!capable(CAP_SYS_ADMIN
))
2559 if (copy_from_user(&objectid
, argp
, sizeof(objectid
)))
2563 objectid
= root
->root_key
.objectid
;
2565 location
.objectid
= objectid
;
2566 location
.type
= BTRFS_ROOT_ITEM_KEY
;
2567 location
.offset
= (u64
)-1;
2569 new_root
= btrfs_read_fs_root_no_name(root
->fs_info
, &location
);
2570 if (IS_ERR(new_root
))
2571 return PTR_ERR(new_root
);
2573 if (btrfs_root_refs(&new_root
->root_item
) == 0)
2576 path
= btrfs_alloc_path();
2579 path
->leave_spinning
= 1;
2581 trans
= btrfs_start_transaction(root
, 1);
2582 if (IS_ERR(trans
)) {
2583 btrfs_free_path(path
);
2584 return PTR_ERR(trans
);
2587 dir_id
= btrfs_super_root_dir(&root
->fs_info
->super_copy
);
2588 di
= btrfs_lookup_dir_item(trans
, root
->fs_info
->tree_root
, path
,
2589 dir_id
, "default", 7, 1);
2590 if (IS_ERR_OR_NULL(di
)) {
2591 btrfs_free_path(path
);
2592 btrfs_end_transaction(trans
, root
);
2593 printk(KERN_ERR
"Umm, you don't have the default dir item, "
2594 "this isn't going to work\n");
2598 btrfs_cpu_key_to_disk(&disk_key
, &new_root
->root_key
);
2599 btrfs_set_dir_item_key(path
->nodes
[0], di
, &disk_key
);
2600 btrfs_mark_buffer_dirty(path
->nodes
[0]);
2601 btrfs_free_path(path
);
2603 disk_super
= &root
->fs_info
->super_copy
;
2604 features
= btrfs_super_incompat_flags(disk_super
);
2605 if (!(features
& BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL
)) {
2606 features
|= BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL
;
2607 btrfs_set_super_incompat_flags(disk_super
, features
);
2609 btrfs_end_transaction(trans
, root
);
2614 static void get_block_group_info(struct list_head
*groups_list
,
2615 struct btrfs_ioctl_space_info
*space
)
2617 struct btrfs_block_group_cache
*block_group
;
2619 space
->total_bytes
= 0;
2620 space
->used_bytes
= 0;
2622 list_for_each_entry(block_group
, groups_list
, list
) {
2623 space
->flags
= block_group
->flags
;
2624 space
->total_bytes
+= block_group
->key
.offset
;
2625 space
->used_bytes
+=
2626 btrfs_block_group_used(&block_group
->item
);
2630 long btrfs_ioctl_space_info(struct btrfs_root
*root
, void __user
*arg
)
2632 struct btrfs_ioctl_space_args space_args
;
2633 struct btrfs_ioctl_space_info space
;
2634 struct btrfs_ioctl_space_info
*dest
;
2635 struct btrfs_ioctl_space_info
*dest_orig
;
2636 struct btrfs_ioctl_space_info __user
*user_dest
;
2637 struct btrfs_space_info
*info
;
2638 u64 types
[] = {BTRFS_BLOCK_GROUP_DATA
,
2639 BTRFS_BLOCK_GROUP_SYSTEM
,
2640 BTRFS_BLOCK_GROUP_METADATA
,
2641 BTRFS_BLOCK_GROUP_DATA
| BTRFS_BLOCK_GROUP_METADATA
};
2648 if (copy_from_user(&space_args
,
2649 (struct btrfs_ioctl_space_args __user
*)arg
,
2650 sizeof(space_args
)))
2653 for (i
= 0; i
< num_types
; i
++) {
2654 struct btrfs_space_info
*tmp
;
2658 list_for_each_entry_rcu(tmp
, &root
->fs_info
->space_info
,
2660 if (tmp
->flags
== types
[i
]) {
2670 down_read(&info
->groups_sem
);
2671 for (c
= 0; c
< BTRFS_NR_RAID_TYPES
; c
++) {
2672 if (!list_empty(&info
->block_groups
[c
]))
2675 up_read(&info
->groups_sem
);
2678 /* space_slots == 0 means they are asking for a count */
2679 if (space_args
.space_slots
== 0) {
2680 space_args
.total_spaces
= slot_count
;
2684 slot_count
= min_t(u64
, space_args
.space_slots
, slot_count
);
2686 alloc_size
= sizeof(*dest
) * slot_count
;
2688 /* we generally have at most 6 or so space infos, one for each raid
2689 * level. So, a whole page should be more than enough for everyone
2691 if (alloc_size
> PAGE_CACHE_SIZE
)
2694 space_args
.total_spaces
= 0;
2695 dest
= kmalloc(alloc_size
, GFP_NOFS
);
2700 /* now we have a buffer to copy into */
2701 for (i
= 0; i
< num_types
; i
++) {
2702 struct btrfs_space_info
*tmp
;
2709 list_for_each_entry_rcu(tmp
, &root
->fs_info
->space_info
,
2711 if (tmp
->flags
== types
[i
]) {
2720 down_read(&info
->groups_sem
);
2721 for (c
= 0; c
< BTRFS_NR_RAID_TYPES
; c
++) {
2722 if (!list_empty(&info
->block_groups
[c
])) {
2723 get_block_group_info(&info
->block_groups
[c
],
2725 memcpy(dest
, &space
, sizeof(space
));
2727 space_args
.total_spaces
++;
2733 up_read(&info
->groups_sem
);
2736 user_dest
= (struct btrfs_ioctl_space_info
*)
2737 (arg
+ sizeof(struct btrfs_ioctl_space_args
));
2739 if (copy_to_user(user_dest
, dest_orig
, alloc_size
))
2744 if (ret
== 0 && copy_to_user(arg
, &space_args
, sizeof(space_args
)))
2751 * there are many ways the trans_start and trans_end ioctls can lead
2752 * to deadlocks. They should only be used by applications that
2753 * basically own the machine, and have a very in depth understanding
2754 * of all the possible deadlocks and enospc problems.
2756 long btrfs_ioctl_trans_end(struct file
*file
)
2758 struct inode
*inode
= fdentry(file
)->d_inode
;
2759 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
2760 struct btrfs_trans_handle
*trans
;
2762 trans
= file
->private_data
;
2765 file
->private_data
= NULL
;
2767 btrfs_end_transaction(trans
, root
);
2769 atomic_dec(&root
->fs_info
->open_ioctl_trans
);
2771 mnt_drop_write(file
->f_path
.mnt
);
2775 static noinline
long btrfs_ioctl_start_sync(struct file
*file
, void __user
*argp
)
2777 struct btrfs_root
*root
= BTRFS_I(file
->f_dentry
->d_inode
)->root
;
2778 struct btrfs_trans_handle
*trans
;
2782 trans
= btrfs_start_transaction(root
, 0);
2784 return PTR_ERR(trans
);
2785 transid
= trans
->transid
;
2786 ret
= btrfs_commit_transaction_async(trans
, root
, 0);
2788 btrfs_end_transaction(trans
, root
);
2793 if (copy_to_user(argp
, &transid
, sizeof(transid
)))
2798 static noinline
long btrfs_ioctl_wait_sync(struct file
*file
, void __user
*argp
)
2800 struct btrfs_root
*root
= BTRFS_I(file
->f_dentry
->d_inode
)->root
;
2804 if (copy_from_user(&transid
, argp
, sizeof(transid
)))
2807 transid
= 0; /* current trans */
2809 return btrfs_wait_for_commit(root
, transid
);
2812 static long btrfs_ioctl_scrub(struct btrfs_root
*root
, void __user
*arg
)
2815 struct btrfs_ioctl_scrub_args
*sa
;
2817 if (!capable(CAP_SYS_ADMIN
))
2820 sa
= memdup_user(arg
, sizeof(*sa
));
2824 ret
= btrfs_scrub_dev(root
, sa
->devid
, sa
->start
, sa
->end
,
2825 &sa
->progress
, sa
->flags
& BTRFS_SCRUB_READONLY
);
2827 if (copy_to_user(arg
, sa
, sizeof(*sa
)))
2834 static long btrfs_ioctl_scrub_cancel(struct btrfs_root
*root
, void __user
*arg
)
2836 if (!capable(CAP_SYS_ADMIN
))
2839 return btrfs_scrub_cancel(root
);
2842 static long btrfs_ioctl_scrub_progress(struct btrfs_root
*root
,
2845 struct btrfs_ioctl_scrub_args
*sa
;
2848 if (!capable(CAP_SYS_ADMIN
))
2851 sa
= memdup_user(arg
, sizeof(*sa
));
2855 ret
= btrfs_scrub_progress(root
, sa
->devid
, &sa
->progress
);
2857 if (copy_to_user(arg
, sa
, sizeof(*sa
)))
2864 long btrfs_ioctl(struct file
*file
, unsigned int
2865 cmd
, unsigned long arg
)
2867 struct btrfs_root
*root
= BTRFS_I(fdentry(file
)->d_inode
)->root
;
2868 void __user
*argp
= (void __user
*)arg
;
2871 case FS_IOC_GETFLAGS
:
2872 return btrfs_ioctl_getflags(file
, argp
);
2873 case FS_IOC_SETFLAGS
:
2874 return btrfs_ioctl_setflags(file
, argp
);
2875 case FS_IOC_GETVERSION
:
2876 return btrfs_ioctl_getversion(file
, argp
);
2878 return btrfs_ioctl_fitrim(file
, argp
);
2879 case BTRFS_IOC_SNAP_CREATE
:
2880 return btrfs_ioctl_snap_create(file
, argp
, 0);
2881 case BTRFS_IOC_SNAP_CREATE_V2
:
2882 return btrfs_ioctl_snap_create_v2(file
, argp
, 0);
2883 case BTRFS_IOC_SUBVOL_CREATE
:
2884 return btrfs_ioctl_snap_create(file
, argp
, 1);
2885 case BTRFS_IOC_SNAP_DESTROY
:
2886 return btrfs_ioctl_snap_destroy(file
, argp
);
2887 case BTRFS_IOC_SUBVOL_GETFLAGS
:
2888 return btrfs_ioctl_subvol_getflags(file
, argp
);
2889 case BTRFS_IOC_SUBVOL_SETFLAGS
:
2890 return btrfs_ioctl_subvol_setflags(file
, argp
);
2891 case BTRFS_IOC_DEFAULT_SUBVOL
:
2892 return btrfs_ioctl_default_subvol(file
, argp
);
2893 case BTRFS_IOC_DEFRAG
:
2894 return btrfs_ioctl_defrag(file
, NULL
);
2895 case BTRFS_IOC_DEFRAG_RANGE
:
2896 return btrfs_ioctl_defrag(file
, argp
);
2897 case BTRFS_IOC_RESIZE
:
2898 return btrfs_ioctl_resize(root
, argp
);
2899 case BTRFS_IOC_ADD_DEV
:
2900 return btrfs_ioctl_add_dev(root
, argp
);
2901 case BTRFS_IOC_RM_DEV
:
2902 return btrfs_ioctl_rm_dev(root
, argp
);
2903 case BTRFS_IOC_FS_INFO
:
2904 return btrfs_ioctl_fs_info(root
, argp
);
2905 case BTRFS_IOC_DEV_INFO
:
2906 return btrfs_ioctl_dev_info(root
, argp
);
2907 case BTRFS_IOC_BALANCE
:
2908 return btrfs_balance(root
->fs_info
->dev_root
);
2909 case BTRFS_IOC_CLONE
:
2910 return btrfs_ioctl_clone(file
, arg
, 0, 0, 0);
2911 case BTRFS_IOC_CLONE_RANGE
:
2912 return btrfs_ioctl_clone_range(file
, argp
);
2913 case BTRFS_IOC_TRANS_START
:
2914 return btrfs_ioctl_trans_start(file
);
2915 case BTRFS_IOC_TRANS_END
:
2916 return btrfs_ioctl_trans_end(file
);
2917 case BTRFS_IOC_TREE_SEARCH
:
2918 return btrfs_ioctl_tree_search(file
, argp
);
2919 case BTRFS_IOC_INO_LOOKUP
:
2920 return btrfs_ioctl_ino_lookup(file
, argp
);
2921 case BTRFS_IOC_SPACE_INFO
:
2922 return btrfs_ioctl_space_info(root
, argp
);
2923 case BTRFS_IOC_SYNC
:
2924 btrfs_sync_fs(file
->f_dentry
->d_sb
, 1);
2926 case BTRFS_IOC_START_SYNC
:
2927 return btrfs_ioctl_start_sync(file
, argp
);
2928 case BTRFS_IOC_WAIT_SYNC
:
2929 return btrfs_ioctl_wait_sync(file
, argp
);
2930 case BTRFS_IOC_SCRUB
:
2931 return btrfs_ioctl_scrub(root
, argp
);
2932 case BTRFS_IOC_SCRUB_CANCEL
:
2933 return btrfs_ioctl_scrub_cancel(root
, argp
);
2934 case BTRFS_IOC_SCRUB_PROGRESS
:
2935 return btrfs_ioctl_scrub_progress(root
, argp
);