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>
48 #include "transaction.h"
49 #include "btrfs_inode.h"
51 #include "print-tree.h"
54 #include "inode-map.h"
56 #include "rcu-string.h"
59 /* Mask out flags that are inappropriate for the given type of inode. */
60 static inline __u32
btrfs_mask_flags(umode_t mode
, __u32 flags
)
64 else if (S_ISREG(mode
))
65 return flags
& ~FS_DIRSYNC_FL
;
67 return flags
& (FS_NODUMP_FL
| FS_NOATIME_FL
);
71 * Export inode flags to the format expected by the FS_IOC_GETFLAGS ioctl.
73 static unsigned int btrfs_flags_to_ioctl(unsigned int flags
)
75 unsigned int iflags
= 0;
77 if (flags
& BTRFS_INODE_SYNC
)
79 if (flags
& BTRFS_INODE_IMMUTABLE
)
80 iflags
|= FS_IMMUTABLE_FL
;
81 if (flags
& BTRFS_INODE_APPEND
)
82 iflags
|= FS_APPEND_FL
;
83 if (flags
& BTRFS_INODE_NODUMP
)
84 iflags
|= FS_NODUMP_FL
;
85 if (flags
& BTRFS_INODE_NOATIME
)
86 iflags
|= FS_NOATIME_FL
;
87 if (flags
& BTRFS_INODE_DIRSYNC
)
88 iflags
|= FS_DIRSYNC_FL
;
89 if (flags
& BTRFS_INODE_NODATACOW
)
90 iflags
|= FS_NOCOW_FL
;
92 if ((flags
& BTRFS_INODE_COMPRESS
) && !(flags
& BTRFS_INODE_NOCOMPRESS
))
93 iflags
|= FS_COMPR_FL
;
94 else if (flags
& BTRFS_INODE_NOCOMPRESS
)
95 iflags
|= FS_NOCOMP_FL
;
101 * Update inode->i_flags based on the btrfs internal flags.
103 void btrfs_update_iflags(struct inode
*inode
)
105 struct btrfs_inode
*ip
= BTRFS_I(inode
);
107 inode
->i_flags
&= ~(S_SYNC
|S_APPEND
|S_IMMUTABLE
|S_NOATIME
|S_DIRSYNC
);
109 if (ip
->flags
& BTRFS_INODE_SYNC
)
110 inode
->i_flags
|= S_SYNC
;
111 if (ip
->flags
& BTRFS_INODE_IMMUTABLE
)
112 inode
->i_flags
|= S_IMMUTABLE
;
113 if (ip
->flags
& BTRFS_INODE_APPEND
)
114 inode
->i_flags
|= S_APPEND
;
115 if (ip
->flags
& BTRFS_INODE_NOATIME
)
116 inode
->i_flags
|= S_NOATIME
;
117 if (ip
->flags
& BTRFS_INODE_DIRSYNC
)
118 inode
->i_flags
|= S_DIRSYNC
;
122 * Inherit flags from the parent inode.
124 * Currently only the compression flags and the cow flags are inherited.
126 void btrfs_inherit_iflags(struct inode
*inode
, struct inode
*dir
)
133 flags
= BTRFS_I(dir
)->flags
;
135 if (flags
& BTRFS_INODE_NOCOMPRESS
) {
136 BTRFS_I(inode
)->flags
&= ~BTRFS_INODE_COMPRESS
;
137 BTRFS_I(inode
)->flags
|= BTRFS_INODE_NOCOMPRESS
;
138 } else if (flags
& BTRFS_INODE_COMPRESS
) {
139 BTRFS_I(inode
)->flags
&= ~BTRFS_INODE_NOCOMPRESS
;
140 BTRFS_I(inode
)->flags
|= BTRFS_INODE_COMPRESS
;
143 if (flags
& BTRFS_INODE_NODATACOW
)
144 BTRFS_I(inode
)->flags
|= BTRFS_INODE_NODATACOW
;
146 btrfs_update_iflags(inode
);
149 static int btrfs_ioctl_getflags(struct file
*file
, void __user
*arg
)
151 struct btrfs_inode
*ip
= BTRFS_I(file
->f_path
.dentry
->d_inode
);
152 unsigned int flags
= btrfs_flags_to_ioctl(ip
->flags
);
154 if (copy_to_user(arg
, &flags
, sizeof(flags
)))
159 static int check_flags(unsigned int flags
)
161 if (flags
& ~(FS_IMMUTABLE_FL
| FS_APPEND_FL
| \
162 FS_NOATIME_FL
| FS_NODUMP_FL
| \
163 FS_SYNC_FL
| FS_DIRSYNC_FL
| \
164 FS_NOCOMP_FL
| FS_COMPR_FL
|
168 if ((flags
& FS_NOCOMP_FL
) && (flags
& FS_COMPR_FL
))
174 static int btrfs_ioctl_setflags(struct file
*file
, void __user
*arg
)
176 struct inode
*inode
= file
->f_path
.dentry
->d_inode
;
177 struct btrfs_inode
*ip
= BTRFS_I(inode
);
178 struct btrfs_root
*root
= ip
->root
;
179 struct btrfs_trans_handle
*trans
;
180 unsigned int flags
, oldflags
;
183 unsigned int i_oldflags
;
185 if (btrfs_root_readonly(root
))
188 if (copy_from_user(&flags
, arg
, sizeof(flags
)))
191 ret
= check_flags(flags
);
195 if (!inode_owner_or_capable(inode
))
198 ret
= mnt_want_write_file(file
);
202 mutex_lock(&inode
->i_mutex
);
204 ip_oldflags
= ip
->flags
;
205 i_oldflags
= inode
->i_flags
;
207 flags
= btrfs_mask_flags(inode
->i_mode
, flags
);
208 oldflags
= btrfs_flags_to_ioctl(ip
->flags
);
209 if ((flags
^ oldflags
) & (FS_APPEND_FL
| FS_IMMUTABLE_FL
)) {
210 if (!capable(CAP_LINUX_IMMUTABLE
)) {
216 if (flags
& FS_SYNC_FL
)
217 ip
->flags
|= BTRFS_INODE_SYNC
;
219 ip
->flags
&= ~BTRFS_INODE_SYNC
;
220 if (flags
& FS_IMMUTABLE_FL
)
221 ip
->flags
|= BTRFS_INODE_IMMUTABLE
;
223 ip
->flags
&= ~BTRFS_INODE_IMMUTABLE
;
224 if (flags
& FS_APPEND_FL
)
225 ip
->flags
|= BTRFS_INODE_APPEND
;
227 ip
->flags
&= ~BTRFS_INODE_APPEND
;
228 if (flags
& FS_NODUMP_FL
)
229 ip
->flags
|= BTRFS_INODE_NODUMP
;
231 ip
->flags
&= ~BTRFS_INODE_NODUMP
;
232 if (flags
& FS_NOATIME_FL
)
233 ip
->flags
|= BTRFS_INODE_NOATIME
;
235 ip
->flags
&= ~BTRFS_INODE_NOATIME
;
236 if (flags
& FS_DIRSYNC_FL
)
237 ip
->flags
|= BTRFS_INODE_DIRSYNC
;
239 ip
->flags
&= ~BTRFS_INODE_DIRSYNC
;
240 if (flags
& FS_NOCOW_FL
)
241 ip
->flags
|= BTRFS_INODE_NODATACOW
;
243 ip
->flags
&= ~BTRFS_INODE_NODATACOW
;
246 * The COMPRESS flag can only be changed by users, while the NOCOMPRESS
247 * flag may be changed automatically if compression code won't make
250 if (flags
& FS_NOCOMP_FL
) {
251 ip
->flags
&= ~BTRFS_INODE_COMPRESS
;
252 ip
->flags
|= BTRFS_INODE_NOCOMPRESS
;
253 } else if (flags
& FS_COMPR_FL
) {
254 ip
->flags
|= BTRFS_INODE_COMPRESS
;
255 ip
->flags
&= ~BTRFS_INODE_NOCOMPRESS
;
257 ip
->flags
&= ~(BTRFS_INODE_COMPRESS
| BTRFS_INODE_NOCOMPRESS
);
260 trans
= btrfs_start_transaction(root
, 1);
262 ret
= PTR_ERR(trans
);
266 btrfs_update_iflags(inode
);
267 inode_inc_iversion(inode
);
268 inode
->i_ctime
= CURRENT_TIME
;
269 ret
= btrfs_update_inode(trans
, root
, inode
);
271 btrfs_end_transaction(trans
, root
);
274 ip
->flags
= ip_oldflags
;
275 inode
->i_flags
= i_oldflags
;
279 mutex_unlock(&inode
->i_mutex
);
280 mnt_drop_write_file(file
);
284 static int btrfs_ioctl_getversion(struct file
*file
, int __user
*arg
)
286 struct inode
*inode
= file
->f_path
.dentry
->d_inode
;
288 return put_user(inode
->i_generation
, arg
);
291 static noinline
int btrfs_ioctl_fitrim(struct file
*file
, void __user
*arg
)
293 struct btrfs_fs_info
*fs_info
= btrfs_sb(fdentry(file
)->d_sb
);
294 struct btrfs_device
*device
;
295 struct request_queue
*q
;
296 struct fstrim_range range
;
297 u64 minlen
= ULLONG_MAX
;
299 u64 total_bytes
= btrfs_super_total_bytes(fs_info
->super_copy
);
302 if (!capable(CAP_SYS_ADMIN
))
306 list_for_each_entry_rcu(device
, &fs_info
->fs_devices
->devices
,
310 q
= bdev_get_queue(device
->bdev
);
311 if (blk_queue_discard(q
)) {
313 minlen
= min((u64
)q
->limits
.discard_granularity
,
321 if (copy_from_user(&range
, arg
, sizeof(range
)))
323 if (range
.start
> total_bytes
)
326 range
.len
= min(range
.len
, total_bytes
- range
.start
);
327 range
.minlen
= max(range
.minlen
, minlen
);
328 ret
= btrfs_trim_fs(fs_info
->tree_root
, &range
);
332 if (copy_to_user(arg
, &range
, sizeof(range
)))
338 static noinline
int create_subvol(struct btrfs_root
*root
,
339 struct dentry
*dentry
,
340 char *name
, int namelen
,
342 struct btrfs_qgroup_inherit
**inherit
)
344 struct btrfs_trans_handle
*trans
;
345 struct btrfs_key key
;
346 struct btrfs_root_item root_item
;
347 struct btrfs_inode_item
*inode_item
;
348 struct extent_buffer
*leaf
;
349 struct btrfs_root
*new_root
;
350 struct dentry
*parent
= dentry
->d_parent
;
352 struct timespec cur_time
= CURRENT_TIME
;
356 u64 new_dirid
= BTRFS_FIRST_FREE_OBJECTID
;
360 ret
= btrfs_find_free_objectid(root
->fs_info
->tree_root
, &objectid
);
364 dir
= parent
->d_inode
;
372 trans
= btrfs_start_transaction(root
, 6);
374 return PTR_ERR(trans
);
376 ret
= btrfs_qgroup_inherit(trans
, root
->fs_info
, 0, objectid
,
377 inherit
? *inherit
: NULL
);
381 leaf
= btrfs_alloc_free_block(trans
, root
, root
->leafsize
,
382 0, objectid
, NULL
, 0, 0, 0);
388 memset_extent_buffer(leaf
, 0, 0, sizeof(struct btrfs_header
));
389 btrfs_set_header_bytenr(leaf
, leaf
->start
);
390 btrfs_set_header_generation(leaf
, trans
->transid
);
391 btrfs_set_header_backref_rev(leaf
, BTRFS_MIXED_BACKREF_REV
);
392 btrfs_set_header_owner(leaf
, objectid
);
394 write_extent_buffer(leaf
, root
->fs_info
->fsid
,
395 (unsigned long)btrfs_header_fsid(leaf
),
397 write_extent_buffer(leaf
, root
->fs_info
->chunk_tree_uuid
,
398 (unsigned long)btrfs_header_chunk_tree_uuid(leaf
),
400 btrfs_mark_buffer_dirty(leaf
);
402 memset(&root_item
, 0, sizeof(root_item
));
404 inode_item
= &root_item
.inode
;
405 inode_item
->generation
= cpu_to_le64(1);
406 inode_item
->size
= cpu_to_le64(3);
407 inode_item
->nlink
= cpu_to_le32(1);
408 inode_item
->nbytes
= cpu_to_le64(root
->leafsize
);
409 inode_item
->mode
= cpu_to_le32(S_IFDIR
| 0755);
412 root_item
.byte_limit
= 0;
413 inode_item
->flags
= cpu_to_le64(BTRFS_INODE_ROOT_ITEM_INIT
);
415 btrfs_set_root_bytenr(&root_item
, leaf
->start
);
416 btrfs_set_root_generation(&root_item
, trans
->transid
);
417 btrfs_set_root_level(&root_item
, 0);
418 btrfs_set_root_refs(&root_item
, 1);
419 btrfs_set_root_used(&root_item
, leaf
->len
);
420 btrfs_set_root_last_snapshot(&root_item
, 0);
422 btrfs_set_root_generation_v2(&root_item
,
423 btrfs_root_generation(&root_item
));
424 uuid_le_gen(&new_uuid
);
425 memcpy(root_item
.uuid
, new_uuid
.b
, BTRFS_UUID_SIZE
);
426 root_item
.otime
.sec
= cpu_to_le64(cur_time
.tv_sec
);
427 root_item
.otime
.nsec
= cpu_to_le64(cur_time
.tv_nsec
);
428 root_item
.ctime
= root_item
.otime
;
429 btrfs_set_root_ctransid(&root_item
, trans
->transid
);
430 btrfs_set_root_otransid(&root_item
, trans
->transid
);
432 btrfs_tree_unlock(leaf
);
433 free_extent_buffer(leaf
);
436 btrfs_set_root_dirid(&root_item
, new_dirid
);
438 key
.objectid
= objectid
;
440 btrfs_set_key_type(&key
, BTRFS_ROOT_ITEM_KEY
);
441 ret
= btrfs_insert_root(trans
, root
->fs_info
->tree_root
, &key
,
446 key
.offset
= (u64
)-1;
447 new_root
= btrfs_read_fs_root_no_name(root
->fs_info
, &key
);
448 if (IS_ERR(new_root
)) {
449 btrfs_abort_transaction(trans
, root
, PTR_ERR(new_root
));
450 ret
= PTR_ERR(new_root
);
454 btrfs_record_root_in_trans(trans
, new_root
);
456 ret
= btrfs_create_subvol_root(trans
, new_root
, new_dirid
);
458 /* We potentially lose an unused inode item here */
459 btrfs_abort_transaction(trans
, root
, ret
);
464 * insert the directory item
466 ret
= btrfs_set_inode_index(dir
, &index
);
468 btrfs_abort_transaction(trans
, root
, ret
);
472 ret
= btrfs_insert_dir_item(trans
, root
,
473 name
, namelen
, dir
, &key
,
474 BTRFS_FT_DIR
, index
);
476 btrfs_abort_transaction(trans
, root
, ret
);
480 btrfs_i_size_write(dir
, dir
->i_size
+ namelen
* 2);
481 ret
= btrfs_update_inode(trans
, root
, dir
);
484 ret
= btrfs_add_root_ref(trans
, root
->fs_info
->tree_root
,
485 objectid
, root
->root_key
.objectid
,
486 btrfs_ino(dir
), index
, name
, namelen
);
490 d_instantiate(dentry
, btrfs_lookup_dentry(dir
, dentry
));
493 *async_transid
= trans
->transid
;
494 err
= btrfs_commit_transaction_async(trans
, root
, 1);
496 err
= btrfs_commit_transaction(trans
, root
);
503 static int create_snapshot(struct btrfs_root
*root
, struct dentry
*dentry
,
504 char *name
, int namelen
, u64
*async_transid
,
505 bool readonly
, struct btrfs_qgroup_inherit
**inherit
)
508 struct btrfs_pending_snapshot
*pending_snapshot
;
509 struct btrfs_trans_handle
*trans
;
515 pending_snapshot
= kzalloc(sizeof(*pending_snapshot
), GFP_NOFS
);
516 if (!pending_snapshot
)
519 btrfs_init_block_rsv(&pending_snapshot
->block_rsv
);
520 pending_snapshot
->dentry
= dentry
;
521 pending_snapshot
->root
= root
;
522 pending_snapshot
->readonly
= readonly
;
524 pending_snapshot
->inherit
= *inherit
;
525 *inherit
= NULL
; /* take responsibility to free it */
528 trans
= btrfs_start_transaction(root
->fs_info
->extent_root
, 5);
530 ret
= PTR_ERR(trans
);
534 ret
= btrfs_snap_reserve_metadata(trans
, pending_snapshot
);
537 spin_lock(&root
->fs_info
->trans_lock
);
538 list_add(&pending_snapshot
->list
,
539 &trans
->transaction
->pending_snapshots
);
540 spin_unlock(&root
->fs_info
->trans_lock
);
542 *async_transid
= trans
->transid
;
543 ret
= btrfs_commit_transaction_async(trans
,
544 root
->fs_info
->extent_root
, 1);
546 ret
= btrfs_commit_transaction(trans
,
547 root
->fs_info
->extent_root
);
551 ret
= pending_snapshot
->error
;
555 ret
= btrfs_orphan_cleanup(pending_snapshot
->snap
);
559 inode
= btrfs_lookup_dentry(dentry
->d_parent
->d_inode
, dentry
);
561 ret
= PTR_ERR(inode
);
565 d_instantiate(dentry
, inode
);
568 kfree(pending_snapshot
);
572 /* copy of check_sticky in fs/namei.c()
573 * It's inline, so penalty for filesystems that don't use sticky bit is
576 static inline int btrfs_check_sticky(struct inode
*dir
, struct inode
*inode
)
578 uid_t fsuid
= current_fsuid();
580 if (!(dir
->i_mode
& S_ISVTX
))
582 if (inode
->i_uid
== fsuid
)
584 if (dir
->i_uid
== fsuid
)
586 return !capable(CAP_FOWNER
);
589 /* copy of may_delete in fs/namei.c()
590 * Check whether we can remove a link victim from directory dir, check
591 * whether the type of victim is right.
592 * 1. We can't do it if dir is read-only (done in permission())
593 * 2. We should have write and exec permissions on dir
594 * 3. We can't remove anything from append-only dir
595 * 4. We can't do anything with immutable dir (done in permission())
596 * 5. If the sticky bit on dir is set we should either
597 * a. be owner of dir, or
598 * b. be owner of victim, or
599 * c. have CAP_FOWNER capability
600 * 6. If the victim is append-only or immutable we can't do antyhing with
601 * links pointing to it.
602 * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR.
603 * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR.
604 * 9. We can't remove a root or mountpoint.
605 * 10. We don't allow removal of NFS sillyrenamed files; it's handled by
606 * nfs_async_unlink().
609 static int btrfs_may_delete(struct inode
*dir
,struct dentry
*victim
,int isdir
)
613 if (!victim
->d_inode
)
616 BUG_ON(victim
->d_parent
->d_inode
!= dir
);
617 audit_inode_child(victim
, dir
);
619 error
= inode_permission(dir
, MAY_WRITE
| MAY_EXEC
);
624 if (btrfs_check_sticky(dir
, victim
->d_inode
)||
625 IS_APPEND(victim
->d_inode
)||
626 IS_IMMUTABLE(victim
->d_inode
) || IS_SWAPFILE(victim
->d_inode
))
629 if (!S_ISDIR(victim
->d_inode
->i_mode
))
633 } else if (S_ISDIR(victim
->d_inode
->i_mode
))
637 if (victim
->d_flags
& DCACHE_NFSFS_RENAMED
)
642 /* copy of may_create in fs/namei.c() */
643 static inline int btrfs_may_create(struct inode
*dir
, struct dentry
*child
)
649 return inode_permission(dir
, MAY_WRITE
| MAY_EXEC
);
653 * Create a new subvolume below @parent. This is largely modeled after
654 * sys_mkdirat and vfs_mkdir, but we only do a single component lookup
655 * inside this filesystem so it's quite a bit simpler.
657 static noinline
int btrfs_mksubvol(struct path
*parent
,
658 char *name
, int namelen
,
659 struct btrfs_root
*snap_src
,
660 u64
*async_transid
, bool readonly
,
661 struct btrfs_qgroup_inherit
**inherit
)
663 struct inode
*dir
= parent
->dentry
->d_inode
;
664 struct dentry
*dentry
;
667 mutex_lock_nested(&dir
->i_mutex
, I_MUTEX_PARENT
);
669 dentry
= lookup_one_len(name
, parent
->dentry
, namelen
);
670 error
= PTR_ERR(dentry
);
678 error
= btrfs_may_create(dir
, dentry
);
682 down_read(&BTRFS_I(dir
)->root
->fs_info
->subvol_sem
);
684 if (btrfs_root_refs(&BTRFS_I(dir
)->root
->root_item
) == 0)
688 error
= create_snapshot(snap_src
, dentry
, name
, namelen
,
689 async_transid
, readonly
, inherit
);
691 error
= create_subvol(BTRFS_I(dir
)->root
, dentry
,
692 name
, namelen
, async_transid
, inherit
);
695 fsnotify_mkdir(dir
, dentry
);
697 up_read(&BTRFS_I(dir
)->root
->fs_info
->subvol_sem
);
701 mutex_unlock(&dir
->i_mutex
);
706 * When we're defragging a range, we don't want to kick it off again
707 * if it is really just waiting for delalloc to send it down.
708 * If we find a nice big extent or delalloc range for the bytes in the
709 * file you want to defrag, we return 0 to let you know to skip this
712 static int check_defrag_in_cache(struct inode
*inode
, u64 offset
, int thresh
)
714 struct extent_io_tree
*io_tree
= &BTRFS_I(inode
)->io_tree
;
715 struct extent_map
*em
= NULL
;
716 struct extent_map_tree
*em_tree
= &BTRFS_I(inode
)->extent_tree
;
719 read_lock(&em_tree
->lock
);
720 em
= lookup_extent_mapping(em_tree
, offset
, PAGE_CACHE_SIZE
);
721 read_unlock(&em_tree
->lock
);
724 end
= extent_map_end(em
);
726 if (end
- offset
> thresh
)
729 /* if we already have a nice delalloc here, just stop */
731 end
= count_range_bits(io_tree
, &offset
, offset
+ thresh
,
732 thresh
, EXTENT_DELALLOC
, 1);
739 * helper function to walk through a file and find extents
740 * newer than a specific transid, and smaller than thresh.
742 * This is used by the defragging code to find new and small
745 static int find_new_extents(struct btrfs_root
*root
,
746 struct inode
*inode
, u64 newer_than
,
747 u64
*off
, int thresh
)
749 struct btrfs_path
*path
;
750 struct btrfs_key min_key
;
751 struct btrfs_key max_key
;
752 struct extent_buffer
*leaf
;
753 struct btrfs_file_extent_item
*extent
;
756 u64 ino
= btrfs_ino(inode
);
758 path
= btrfs_alloc_path();
762 min_key
.objectid
= ino
;
763 min_key
.type
= BTRFS_EXTENT_DATA_KEY
;
764 min_key
.offset
= *off
;
766 max_key
.objectid
= ino
;
767 max_key
.type
= (u8
)-1;
768 max_key
.offset
= (u64
)-1;
770 path
->keep_locks
= 1;
773 ret
= btrfs_search_forward(root
, &min_key
, &max_key
,
774 path
, 0, newer_than
);
777 if (min_key
.objectid
!= ino
)
779 if (min_key
.type
!= BTRFS_EXTENT_DATA_KEY
)
782 leaf
= path
->nodes
[0];
783 extent
= btrfs_item_ptr(leaf
, path
->slots
[0],
784 struct btrfs_file_extent_item
);
786 type
= btrfs_file_extent_type(leaf
, extent
);
787 if (type
== BTRFS_FILE_EXTENT_REG
&&
788 btrfs_file_extent_num_bytes(leaf
, extent
) < thresh
&&
789 check_defrag_in_cache(inode
, min_key
.offset
, thresh
)) {
790 *off
= min_key
.offset
;
791 btrfs_free_path(path
);
795 if (min_key
.offset
== (u64
)-1)
799 btrfs_release_path(path
);
802 btrfs_free_path(path
);
806 static struct extent_map
*defrag_lookup_extent(struct inode
*inode
, u64 start
)
808 struct extent_map_tree
*em_tree
= &BTRFS_I(inode
)->extent_tree
;
809 struct extent_io_tree
*io_tree
= &BTRFS_I(inode
)->io_tree
;
810 struct extent_map
*em
;
811 u64 len
= PAGE_CACHE_SIZE
;
814 * hopefully we have this extent in the tree already, try without
815 * the full extent lock
817 read_lock(&em_tree
->lock
);
818 em
= lookup_extent_mapping(em_tree
, start
, len
);
819 read_unlock(&em_tree
->lock
);
822 /* get the big lock and read metadata off disk */
823 lock_extent(io_tree
, start
, start
+ len
- 1);
824 em
= btrfs_get_extent(inode
, NULL
, 0, start
, len
, 0);
825 unlock_extent(io_tree
, start
, start
+ len
- 1);
834 static bool defrag_check_next_extent(struct inode
*inode
, struct extent_map
*em
)
836 struct extent_map
*next
;
839 /* this is the last extent */
840 if (em
->start
+ em
->len
>= i_size_read(inode
))
843 next
= defrag_lookup_extent(inode
, em
->start
+ em
->len
);
844 if (!next
|| next
->block_start
>= EXTENT_MAP_LAST_BYTE
)
847 free_extent_map(next
);
851 static int should_defrag_range(struct inode
*inode
, u64 start
, int thresh
,
852 u64
*last_len
, u64
*skip
, u64
*defrag_end
,
855 struct extent_map
*em
;
857 bool next_mergeable
= true;
860 * make sure that once we start defragging an extent, we keep on
863 if (start
< *defrag_end
)
868 em
= defrag_lookup_extent(inode
, start
);
872 /* this will cover holes, and inline extents */
873 if (em
->block_start
>= EXTENT_MAP_LAST_BYTE
) {
878 next_mergeable
= defrag_check_next_extent(inode
, em
);
881 * we hit a real extent, if it is big or the next extent is not a
882 * real extent, don't bother defragging it
884 if (!compress
&& (*last_len
== 0 || *last_len
>= thresh
) &&
885 (em
->len
>= thresh
|| !next_mergeable
))
889 * last_len ends up being a counter of how many bytes we've defragged.
890 * every time we choose not to defrag an extent, we reset *last_len
891 * so that the next tiny extent will force a defrag.
893 * The end result of this is that tiny extents before a single big
894 * extent will force at least part of that big extent to be defragged.
897 *defrag_end
= extent_map_end(em
);
900 *skip
= extent_map_end(em
);
909 * it doesn't do much good to defrag one or two pages
910 * at a time. This pulls in a nice chunk of pages
913 * It also makes sure the delalloc code has enough
914 * dirty data to avoid making new small extents as part
917 * It's a good idea to start RA on this range
918 * before calling this.
920 static int cluster_pages_for_defrag(struct inode
*inode
,
922 unsigned long start_index
,
925 unsigned long file_end
;
926 u64 isize
= i_size_read(inode
);
933 struct btrfs_ordered_extent
*ordered
;
934 struct extent_state
*cached_state
= NULL
;
935 struct extent_io_tree
*tree
;
936 gfp_t mask
= btrfs_alloc_write_mask(inode
->i_mapping
);
938 file_end
= (isize
- 1) >> PAGE_CACHE_SHIFT
;
939 if (!isize
|| start_index
> file_end
)
942 page_cnt
= min_t(u64
, (u64
)num_pages
, (u64
)file_end
- start_index
+ 1);
944 ret
= btrfs_delalloc_reserve_space(inode
,
945 page_cnt
<< PAGE_CACHE_SHIFT
);
949 tree
= &BTRFS_I(inode
)->io_tree
;
951 /* step one, lock all the pages */
952 for (i
= 0; i
< page_cnt
; i
++) {
955 page
= find_or_create_page(inode
->i_mapping
,
956 start_index
+ i
, mask
);
960 page_start
= page_offset(page
);
961 page_end
= page_start
+ PAGE_CACHE_SIZE
- 1;
963 lock_extent(tree
, page_start
, page_end
);
964 ordered
= btrfs_lookup_ordered_extent(inode
,
966 unlock_extent(tree
, page_start
, page_end
);
971 btrfs_start_ordered_extent(inode
, ordered
, 1);
972 btrfs_put_ordered_extent(ordered
);
975 * we unlocked the page above, so we need check if
976 * it was released or not.
978 if (page
->mapping
!= inode
->i_mapping
) {
980 page_cache_release(page
);
985 if (!PageUptodate(page
)) {
986 btrfs_readpage(NULL
, page
);
988 if (!PageUptodate(page
)) {
990 page_cache_release(page
);
996 if (page
->mapping
!= inode
->i_mapping
) {
998 page_cache_release(page
);
1008 if (!(inode
->i_sb
->s_flags
& MS_ACTIVE
))
1012 * so now we have a nice long stream of locked
1013 * and up to date pages, lets wait on them
1015 for (i
= 0; i
< i_done
; i
++)
1016 wait_on_page_writeback(pages
[i
]);
1018 page_start
= page_offset(pages
[0]);
1019 page_end
= page_offset(pages
[i_done
- 1]) + PAGE_CACHE_SIZE
;
1021 lock_extent_bits(&BTRFS_I(inode
)->io_tree
,
1022 page_start
, page_end
- 1, 0, &cached_state
);
1023 clear_extent_bit(&BTRFS_I(inode
)->io_tree
, page_start
,
1024 page_end
- 1, EXTENT_DIRTY
| EXTENT_DELALLOC
|
1025 EXTENT_DO_ACCOUNTING
, 0, 0, &cached_state
,
1028 if (i_done
!= page_cnt
) {
1029 spin_lock(&BTRFS_I(inode
)->lock
);
1030 BTRFS_I(inode
)->outstanding_extents
++;
1031 spin_unlock(&BTRFS_I(inode
)->lock
);
1032 btrfs_delalloc_release_space(inode
,
1033 (page_cnt
- i_done
) << PAGE_CACHE_SHIFT
);
1037 btrfs_set_extent_delalloc(inode
, page_start
, page_end
- 1,
1040 unlock_extent_cached(&BTRFS_I(inode
)->io_tree
,
1041 page_start
, page_end
- 1, &cached_state
,
1044 for (i
= 0; i
< i_done
; i
++) {
1045 clear_page_dirty_for_io(pages
[i
]);
1046 ClearPageChecked(pages
[i
]);
1047 set_page_extent_mapped(pages
[i
]);
1048 set_page_dirty(pages
[i
]);
1049 unlock_page(pages
[i
]);
1050 page_cache_release(pages
[i
]);
1054 for (i
= 0; i
< i_done
; i
++) {
1055 unlock_page(pages
[i
]);
1056 page_cache_release(pages
[i
]);
1058 btrfs_delalloc_release_space(inode
, page_cnt
<< PAGE_CACHE_SHIFT
);
1063 int btrfs_defrag_file(struct inode
*inode
, struct file
*file
,
1064 struct btrfs_ioctl_defrag_range_args
*range
,
1065 u64 newer_than
, unsigned long max_to_defrag
)
1067 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
1068 struct file_ra_state
*ra
= NULL
;
1069 unsigned long last_index
;
1070 u64 isize
= i_size_read(inode
);
1074 u64 newer_off
= range
->start
;
1076 unsigned long ra_index
= 0;
1078 int defrag_count
= 0;
1079 int compress_type
= BTRFS_COMPRESS_ZLIB
;
1080 int extent_thresh
= range
->extent_thresh
;
1081 int max_cluster
= (256 * 1024) >> PAGE_CACHE_SHIFT
;
1082 int cluster
= max_cluster
;
1083 u64 new_align
= ~((u64
)128 * 1024 - 1);
1084 struct page
**pages
= NULL
;
1086 if (extent_thresh
== 0)
1087 extent_thresh
= 256 * 1024;
1089 if (range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
) {
1090 if (range
->compress_type
> BTRFS_COMPRESS_TYPES
)
1092 if (range
->compress_type
)
1093 compress_type
= range
->compress_type
;
1100 * if we were not given a file, allocate a readahead
1104 ra
= kzalloc(sizeof(*ra
), GFP_NOFS
);
1107 file_ra_state_init(ra
, inode
->i_mapping
);
1112 pages
= kmalloc(sizeof(struct page
*) * max_cluster
,
1119 /* find the last page to defrag */
1120 if (range
->start
+ range
->len
> range
->start
) {
1121 last_index
= min_t(u64
, isize
- 1,
1122 range
->start
+ range
->len
- 1) >> PAGE_CACHE_SHIFT
;
1124 last_index
= (isize
- 1) >> PAGE_CACHE_SHIFT
;
1128 ret
= find_new_extents(root
, inode
, newer_than
,
1129 &newer_off
, 64 * 1024);
1131 range
->start
= newer_off
;
1133 * we always align our defrag to help keep
1134 * the extents in the file evenly spaced
1136 i
= (newer_off
& new_align
) >> PAGE_CACHE_SHIFT
;
1140 i
= range
->start
>> PAGE_CACHE_SHIFT
;
1143 max_to_defrag
= last_index
+ 1;
1146 * make writeback starts from i, so the defrag range can be
1147 * written sequentially.
1149 if (i
< inode
->i_mapping
->writeback_index
)
1150 inode
->i_mapping
->writeback_index
= i
;
1152 while (i
<= last_index
&& defrag_count
< max_to_defrag
&&
1153 (i
< (i_size_read(inode
) + PAGE_CACHE_SIZE
- 1) >>
1154 PAGE_CACHE_SHIFT
)) {
1156 * make sure we stop running if someone unmounts
1159 if (!(inode
->i_sb
->s_flags
& MS_ACTIVE
))
1162 if (!should_defrag_range(inode
, (u64
)i
<< PAGE_CACHE_SHIFT
,
1163 extent_thresh
, &last_len
, &skip
,
1164 &defrag_end
, range
->flags
&
1165 BTRFS_DEFRAG_RANGE_COMPRESS
)) {
1168 * the should_defrag function tells us how much to skip
1169 * bump our counter by the suggested amount
1171 next
= (skip
+ PAGE_CACHE_SIZE
- 1) >> PAGE_CACHE_SHIFT
;
1172 i
= max(i
+ 1, next
);
1177 cluster
= (PAGE_CACHE_ALIGN(defrag_end
) >>
1178 PAGE_CACHE_SHIFT
) - i
;
1179 cluster
= min(cluster
, max_cluster
);
1181 cluster
= max_cluster
;
1184 if (range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
)
1185 BTRFS_I(inode
)->force_compress
= compress_type
;
1187 if (i
+ cluster
> ra_index
) {
1188 ra_index
= max(i
, ra_index
);
1189 btrfs_force_ra(inode
->i_mapping
, ra
, file
, ra_index
,
1191 ra_index
+= max_cluster
;
1194 mutex_lock(&inode
->i_mutex
);
1195 ret
= cluster_pages_for_defrag(inode
, pages
, i
, cluster
);
1197 mutex_unlock(&inode
->i_mutex
);
1201 defrag_count
+= ret
;
1202 balance_dirty_pages_ratelimited_nr(inode
->i_mapping
, ret
);
1203 mutex_unlock(&inode
->i_mutex
);
1206 if (newer_off
== (u64
)-1)
1212 newer_off
= max(newer_off
+ 1,
1213 (u64
)i
<< PAGE_CACHE_SHIFT
);
1215 ret
= find_new_extents(root
, inode
,
1216 newer_than
, &newer_off
,
1219 range
->start
= newer_off
;
1220 i
= (newer_off
& new_align
) >> PAGE_CACHE_SHIFT
;
1227 last_len
+= ret
<< PAGE_CACHE_SHIFT
;
1235 if ((range
->flags
& BTRFS_DEFRAG_RANGE_START_IO
))
1236 filemap_flush(inode
->i_mapping
);
1238 if ((range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
)) {
1239 /* the filemap_flush will queue IO into the worker threads, but
1240 * we have to make sure the IO is actually started and that
1241 * ordered extents get created before we return
1243 atomic_inc(&root
->fs_info
->async_submit_draining
);
1244 while (atomic_read(&root
->fs_info
->nr_async_submits
) ||
1245 atomic_read(&root
->fs_info
->async_delalloc_pages
)) {
1246 wait_event(root
->fs_info
->async_submit_wait
,
1247 (atomic_read(&root
->fs_info
->nr_async_submits
) == 0 &&
1248 atomic_read(&root
->fs_info
->async_delalloc_pages
) == 0));
1250 atomic_dec(&root
->fs_info
->async_submit_draining
);
1252 mutex_lock(&inode
->i_mutex
);
1253 BTRFS_I(inode
)->force_compress
= BTRFS_COMPRESS_NONE
;
1254 mutex_unlock(&inode
->i_mutex
);
1257 if (range
->compress_type
== BTRFS_COMPRESS_LZO
) {
1258 btrfs_set_fs_incompat(root
->fs_info
, COMPRESS_LZO
);
1270 static noinline
int btrfs_ioctl_resize(struct btrfs_root
*root
,
1276 struct btrfs_ioctl_vol_args
*vol_args
;
1277 struct btrfs_trans_handle
*trans
;
1278 struct btrfs_device
*device
= NULL
;
1280 char *devstr
= NULL
;
1284 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
)
1287 if (!capable(CAP_SYS_ADMIN
))
1290 mutex_lock(&root
->fs_info
->volume_mutex
);
1291 if (root
->fs_info
->balance_ctl
) {
1292 printk(KERN_INFO
"btrfs: balance in progress\n");
1297 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1298 if (IS_ERR(vol_args
)) {
1299 ret
= PTR_ERR(vol_args
);
1303 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
1305 sizestr
= vol_args
->name
;
1306 devstr
= strchr(sizestr
, ':');
1309 sizestr
= devstr
+ 1;
1311 devstr
= vol_args
->name
;
1312 devid
= simple_strtoull(devstr
, &end
, 10);
1313 printk(KERN_INFO
"btrfs: resizing devid %llu\n",
1314 (unsigned long long)devid
);
1316 device
= btrfs_find_device(root
, devid
, NULL
, NULL
);
1318 printk(KERN_INFO
"btrfs: resizer unable to find device %llu\n",
1319 (unsigned long long)devid
);
1323 if (device
->fs_devices
&& device
->fs_devices
->seeding
) {
1324 printk(KERN_INFO
"btrfs: resizer unable to apply on "
1325 "seeding device %llu\n",
1326 (unsigned long long)devid
);
1331 if (!strcmp(sizestr
, "max"))
1332 new_size
= device
->bdev
->bd_inode
->i_size
;
1334 if (sizestr
[0] == '-') {
1337 } else if (sizestr
[0] == '+') {
1341 new_size
= memparse(sizestr
, NULL
);
1342 if (new_size
== 0) {
1348 old_size
= device
->total_bytes
;
1351 if (new_size
> old_size
) {
1355 new_size
= old_size
- new_size
;
1356 } else if (mod
> 0) {
1357 new_size
= old_size
+ new_size
;
1360 if (new_size
< 256 * 1024 * 1024) {
1364 if (new_size
> device
->bdev
->bd_inode
->i_size
) {
1369 do_div(new_size
, root
->sectorsize
);
1370 new_size
*= root
->sectorsize
;
1372 printk_in_rcu(KERN_INFO
"btrfs: new size for %s is %llu\n",
1373 rcu_str_deref(device
->name
),
1374 (unsigned long long)new_size
);
1376 if (new_size
> old_size
) {
1377 trans
= btrfs_start_transaction(root
, 0);
1378 if (IS_ERR(trans
)) {
1379 ret
= PTR_ERR(trans
);
1382 ret
= btrfs_grow_device(trans
, device
, new_size
);
1383 btrfs_commit_transaction(trans
, root
);
1384 } else if (new_size
< old_size
) {
1385 ret
= btrfs_shrink_device(device
, new_size
);
1391 mutex_unlock(&root
->fs_info
->volume_mutex
);
1395 static noinline
int btrfs_ioctl_snap_create_transid(struct file
*file
,
1396 char *name
, unsigned long fd
, int subvol
,
1397 u64
*transid
, bool readonly
,
1398 struct btrfs_qgroup_inherit
**inherit
)
1400 struct file
*src_file
;
1404 ret
= mnt_want_write_file(file
);
1408 namelen
= strlen(name
);
1409 if (strchr(name
, '/')) {
1411 goto out_drop_write
;
1414 if (name
[0] == '.' &&
1415 (namelen
== 1 || (name
[1] == '.' && namelen
== 2))) {
1417 goto out_drop_write
;
1421 ret
= btrfs_mksubvol(&file
->f_path
, name
, namelen
,
1422 NULL
, transid
, readonly
, inherit
);
1424 struct inode
*src_inode
;
1425 src_file
= fget(fd
);
1428 goto out_drop_write
;
1431 src_inode
= src_file
->f_path
.dentry
->d_inode
;
1432 if (src_inode
->i_sb
!= file
->f_path
.dentry
->d_inode
->i_sb
) {
1433 printk(KERN_INFO
"btrfs: Snapshot src from "
1437 goto out_drop_write
;
1439 ret
= btrfs_mksubvol(&file
->f_path
, name
, namelen
,
1440 BTRFS_I(src_inode
)->root
,
1441 transid
, readonly
, inherit
);
1445 mnt_drop_write_file(file
);
1450 static noinline
int btrfs_ioctl_snap_create(struct file
*file
,
1451 void __user
*arg
, int subvol
)
1453 struct btrfs_ioctl_vol_args
*vol_args
;
1456 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1457 if (IS_ERR(vol_args
))
1458 return PTR_ERR(vol_args
);
1459 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
1461 ret
= btrfs_ioctl_snap_create_transid(file
, vol_args
->name
,
1462 vol_args
->fd
, subvol
,
1469 static noinline
int btrfs_ioctl_snap_create_v2(struct file
*file
,
1470 void __user
*arg
, int subvol
)
1472 struct btrfs_ioctl_vol_args_v2
*vol_args
;
1476 bool readonly
= false;
1477 struct btrfs_qgroup_inherit
*inherit
= NULL
;
1479 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1480 if (IS_ERR(vol_args
))
1481 return PTR_ERR(vol_args
);
1482 vol_args
->name
[BTRFS_SUBVOL_NAME_MAX
] = '\0';
1484 if (vol_args
->flags
&
1485 ~(BTRFS_SUBVOL_CREATE_ASYNC
| BTRFS_SUBVOL_RDONLY
|
1486 BTRFS_SUBVOL_QGROUP_INHERIT
)) {
1491 if (vol_args
->flags
& BTRFS_SUBVOL_CREATE_ASYNC
)
1493 if (vol_args
->flags
& BTRFS_SUBVOL_RDONLY
)
1495 if (vol_args
->flags
& BTRFS_SUBVOL_QGROUP_INHERIT
) {
1496 if (vol_args
->size
> PAGE_CACHE_SIZE
) {
1500 inherit
= memdup_user(vol_args
->qgroup_inherit
, vol_args
->size
);
1501 if (IS_ERR(inherit
)) {
1502 ret
= PTR_ERR(inherit
);
1507 ret
= btrfs_ioctl_snap_create_transid(file
, vol_args
->name
,
1508 vol_args
->fd
, subvol
, ptr
,
1509 readonly
, &inherit
);
1511 if (ret
== 0 && ptr
&&
1513 offsetof(struct btrfs_ioctl_vol_args_v2
,
1514 transid
), ptr
, sizeof(*ptr
)))
1522 static noinline
int btrfs_ioctl_subvol_getflags(struct file
*file
,
1525 struct inode
*inode
= fdentry(file
)->d_inode
;
1526 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
1530 if (btrfs_ino(inode
) != BTRFS_FIRST_FREE_OBJECTID
)
1533 down_read(&root
->fs_info
->subvol_sem
);
1534 if (btrfs_root_readonly(root
))
1535 flags
|= BTRFS_SUBVOL_RDONLY
;
1536 up_read(&root
->fs_info
->subvol_sem
);
1538 if (copy_to_user(arg
, &flags
, sizeof(flags
)))
1544 static noinline
int btrfs_ioctl_subvol_setflags(struct file
*file
,
1547 struct inode
*inode
= fdentry(file
)->d_inode
;
1548 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
1549 struct btrfs_trans_handle
*trans
;
1554 ret
= mnt_want_write_file(file
);
1558 if (btrfs_ino(inode
) != BTRFS_FIRST_FREE_OBJECTID
) {
1560 goto out_drop_write
;
1563 if (copy_from_user(&flags
, arg
, sizeof(flags
))) {
1565 goto out_drop_write
;
1568 if (flags
& BTRFS_SUBVOL_CREATE_ASYNC
) {
1570 goto out_drop_write
;
1573 if (flags
& ~BTRFS_SUBVOL_RDONLY
) {
1575 goto out_drop_write
;
1578 if (!inode_owner_or_capable(inode
)) {
1580 goto out_drop_write
;
1583 down_write(&root
->fs_info
->subvol_sem
);
1586 if (!!(flags
& BTRFS_SUBVOL_RDONLY
) == btrfs_root_readonly(root
))
1589 root_flags
= btrfs_root_flags(&root
->root_item
);
1590 if (flags
& BTRFS_SUBVOL_RDONLY
)
1591 btrfs_set_root_flags(&root
->root_item
,
1592 root_flags
| BTRFS_ROOT_SUBVOL_RDONLY
);
1594 btrfs_set_root_flags(&root
->root_item
,
1595 root_flags
& ~BTRFS_ROOT_SUBVOL_RDONLY
);
1597 trans
= btrfs_start_transaction(root
, 1);
1598 if (IS_ERR(trans
)) {
1599 ret
= PTR_ERR(trans
);
1603 ret
= btrfs_update_root(trans
, root
->fs_info
->tree_root
,
1604 &root
->root_key
, &root
->root_item
);
1606 btrfs_commit_transaction(trans
, root
);
1609 btrfs_set_root_flags(&root
->root_item
, root_flags
);
1611 up_write(&root
->fs_info
->subvol_sem
);
1613 mnt_drop_write_file(file
);
1619 * helper to check if the subvolume references other subvolumes
1621 static noinline
int may_destroy_subvol(struct btrfs_root
*root
)
1623 struct btrfs_path
*path
;
1624 struct btrfs_key key
;
1627 path
= btrfs_alloc_path();
1631 key
.objectid
= root
->root_key
.objectid
;
1632 key
.type
= BTRFS_ROOT_REF_KEY
;
1633 key
.offset
= (u64
)-1;
1635 ret
= btrfs_search_slot(NULL
, root
->fs_info
->tree_root
,
1642 if (path
->slots
[0] > 0) {
1644 btrfs_item_key_to_cpu(path
->nodes
[0], &key
, path
->slots
[0]);
1645 if (key
.objectid
== root
->root_key
.objectid
&&
1646 key
.type
== BTRFS_ROOT_REF_KEY
)
1650 btrfs_free_path(path
);
1654 static noinline
int key_in_sk(struct btrfs_key
*key
,
1655 struct btrfs_ioctl_search_key
*sk
)
1657 struct btrfs_key test
;
1660 test
.objectid
= sk
->min_objectid
;
1661 test
.type
= sk
->min_type
;
1662 test
.offset
= sk
->min_offset
;
1664 ret
= btrfs_comp_cpu_keys(key
, &test
);
1668 test
.objectid
= sk
->max_objectid
;
1669 test
.type
= sk
->max_type
;
1670 test
.offset
= sk
->max_offset
;
1672 ret
= btrfs_comp_cpu_keys(key
, &test
);
1678 static noinline
int copy_to_sk(struct btrfs_root
*root
,
1679 struct btrfs_path
*path
,
1680 struct btrfs_key
*key
,
1681 struct btrfs_ioctl_search_key
*sk
,
1683 unsigned long *sk_offset
,
1687 struct extent_buffer
*leaf
;
1688 struct btrfs_ioctl_search_header sh
;
1689 unsigned long item_off
;
1690 unsigned long item_len
;
1696 leaf
= path
->nodes
[0];
1697 slot
= path
->slots
[0];
1698 nritems
= btrfs_header_nritems(leaf
);
1700 if (btrfs_header_generation(leaf
) > sk
->max_transid
) {
1704 found_transid
= btrfs_header_generation(leaf
);
1706 for (i
= slot
; i
< nritems
; i
++) {
1707 item_off
= btrfs_item_ptr_offset(leaf
, i
);
1708 item_len
= btrfs_item_size_nr(leaf
, i
);
1710 if (item_len
> BTRFS_SEARCH_ARGS_BUFSIZE
)
1713 if (sizeof(sh
) + item_len
+ *sk_offset
>
1714 BTRFS_SEARCH_ARGS_BUFSIZE
) {
1719 btrfs_item_key_to_cpu(leaf
, key
, i
);
1720 if (!key_in_sk(key
, sk
))
1723 sh
.objectid
= key
->objectid
;
1724 sh
.offset
= key
->offset
;
1725 sh
.type
= key
->type
;
1727 sh
.transid
= found_transid
;
1729 /* copy search result header */
1730 memcpy(buf
+ *sk_offset
, &sh
, sizeof(sh
));
1731 *sk_offset
+= sizeof(sh
);
1734 char *p
= buf
+ *sk_offset
;
1736 read_extent_buffer(leaf
, p
,
1737 item_off
, item_len
);
1738 *sk_offset
+= item_len
;
1742 if (*num_found
>= sk
->nr_items
)
1747 if (key
->offset
< (u64
)-1 && key
->offset
< sk
->max_offset
)
1749 else if (key
->type
< (u8
)-1 && key
->type
< sk
->max_type
) {
1752 } else if (key
->objectid
< (u64
)-1 && key
->objectid
< sk
->max_objectid
) {
1762 static noinline
int search_ioctl(struct inode
*inode
,
1763 struct btrfs_ioctl_search_args
*args
)
1765 struct btrfs_root
*root
;
1766 struct btrfs_key key
;
1767 struct btrfs_key max_key
;
1768 struct btrfs_path
*path
;
1769 struct btrfs_ioctl_search_key
*sk
= &args
->key
;
1770 struct btrfs_fs_info
*info
= BTRFS_I(inode
)->root
->fs_info
;
1773 unsigned long sk_offset
= 0;
1775 path
= btrfs_alloc_path();
1779 if (sk
->tree_id
== 0) {
1780 /* search the root of the inode that was passed */
1781 root
= BTRFS_I(inode
)->root
;
1783 key
.objectid
= sk
->tree_id
;
1784 key
.type
= BTRFS_ROOT_ITEM_KEY
;
1785 key
.offset
= (u64
)-1;
1786 root
= btrfs_read_fs_root_no_name(info
, &key
);
1788 printk(KERN_ERR
"could not find root %llu\n",
1790 btrfs_free_path(path
);
1795 key
.objectid
= sk
->min_objectid
;
1796 key
.type
= sk
->min_type
;
1797 key
.offset
= sk
->min_offset
;
1799 max_key
.objectid
= sk
->max_objectid
;
1800 max_key
.type
= sk
->max_type
;
1801 max_key
.offset
= sk
->max_offset
;
1803 path
->keep_locks
= 1;
1806 ret
= btrfs_search_forward(root
, &key
, &max_key
, path
, 0,
1813 ret
= copy_to_sk(root
, path
, &key
, sk
, args
->buf
,
1814 &sk_offset
, &num_found
);
1815 btrfs_release_path(path
);
1816 if (ret
|| num_found
>= sk
->nr_items
)
1822 sk
->nr_items
= num_found
;
1823 btrfs_free_path(path
);
1827 static noinline
int btrfs_ioctl_tree_search(struct file
*file
,
1830 struct btrfs_ioctl_search_args
*args
;
1831 struct inode
*inode
;
1834 if (!capable(CAP_SYS_ADMIN
))
1837 args
= memdup_user(argp
, sizeof(*args
));
1839 return PTR_ERR(args
);
1841 inode
= fdentry(file
)->d_inode
;
1842 ret
= search_ioctl(inode
, args
);
1843 if (ret
== 0 && copy_to_user(argp
, args
, sizeof(*args
)))
1850 * Search INODE_REFs to identify path name of 'dirid' directory
1851 * in a 'tree_id' tree. and sets path name to 'name'.
1853 static noinline
int btrfs_search_path_in_tree(struct btrfs_fs_info
*info
,
1854 u64 tree_id
, u64 dirid
, char *name
)
1856 struct btrfs_root
*root
;
1857 struct btrfs_key key
;
1863 struct btrfs_inode_ref
*iref
;
1864 struct extent_buffer
*l
;
1865 struct btrfs_path
*path
;
1867 if (dirid
== BTRFS_FIRST_FREE_OBJECTID
) {
1872 path
= btrfs_alloc_path();
1876 ptr
= &name
[BTRFS_INO_LOOKUP_PATH_MAX
];
1878 key
.objectid
= tree_id
;
1879 key
.type
= BTRFS_ROOT_ITEM_KEY
;
1880 key
.offset
= (u64
)-1;
1881 root
= btrfs_read_fs_root_no_name(info
, &key
);
1883 printk(KERN_ERR
"could not find root %llu\n", tree_id
);
1888 key
.objectid
= dirid
;
1889 key
.type
= BTRFS_INODE_REF_KEY
;
1890 key
.offset
= (u64
)-1;
1893 ret
= btrfs_search_slot(NULL
, root
, &key
, path
, 0, 0);
1898 slot
= path
->slots
[0];
1899 if (ret
> 0 && slot
> 0)
1901 btrfs_item_key_to_cpu(l
, &key
, slot
);
1903 if (ret
> 0 && (key
.objectid
!= dirid
||
1904 key
.type
!= BTRFS_INODE_REF_KEY
)) {
1909 iref
= btrfs_item_ptr(l
, slot
, struct btrfs_inode_ref
);
1910 len
= btrfs_inode_ref_name_len(l
, iref
);
1912 total_len
+= len
+ 1;
1917 read_extent_buffer(l
, ptr
,(unsigned long)(iref
+ 1), len
);
1919 if (key
.offset
== BTRFS_FIRST_FREE_OBJECTID
)
1922 btrfs_release_path(path
);
1923 key
.objectid
= key
.offset
;
1924 key
.offset
= (u64
)-1;
1925 dirid
= key
.objectid
;
1929 memmove(name
, ptr
, total_len
);
1930 name
[total_len
]='\0';
1933 btrfs_free_path(path
);
1937 static noinline
int btrfs_ioctl_ino_lookup(struct file
*file
,
1940 struct btrfs_ioctl_ino_lookup_args
*args
;
1941 struct inode
*inode
;
1944 if (!capable(CAP_SYS_ADMIN
))
1947 args
= memdup_user(argp
, sizeof(*args
));
1949 return PTR_ERR(args
);
1951 inode
= fdentry(file
)->d_inode
;
1953 if (args
->treeid
== 0)
1954 args
->treeid
= BTRFS_I(inode
)->root
->root_key
.objectid
;
1956 ret
= btrfs_search_path_in_tree(BTRFS_I(inode
)->root
->fs_info
,
1957 args
->treeid
, args
->objectid
,
1960 if (ret
== 0 && copy_to_user(argp
, args
, sizeof(*args
)))
1967 static noinline
int btrfs_ioctl_snap_destroy(struct file
*file
,
1970 struct dentry
*parent
= fdentry(file
);
1971 struct dentry
*dentry
;
1972 struct inode
*dir
= parent
->d_inode
;
1973 struct inode
*inode
;
1974 struct btrfs_root
*root
= BTRFS_I(dir
)->root
;
1975 struct btrfs_root
*dest
= NULL
;
1976 struct btrfs_ioctl_vol_args
*vol_args
;
1977 struct btrfs_trans_handle
*trans
;
1982 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1983 if (IS_ERR(vol_args
))
1984 return PTR_ERR(vol_args
);
1986 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
1987 namelen
= strlen(vol_args
->name
);
1988 if (strchr(vol_args
->name
, '/') ||
1989 strncmp(vol_args
->name
, "..", namelen
) == 0) {
1994 err
= mnt_want_write_file(file
);
1998 mutex_lock_nested(&dir
->i_mutex
, I_MUTEX_PARENT
);
1999 dentry
= lookup_one_len(vol_args
->name
, parent
, namelen
);
2000 if (IS_ERR(dentry
)) {
2001 err
= PTR_ERR(dentry
);
2002 goto out_unlock_dir
;
2005 if (!dentry
->d_inode
) {
2010 inode
= dentry
->d_inode
;
2011 dest
= BTRFS_I(inode
)->root
;
2012 if (!capable(CAP_SYS_ADMIN
)){
2014 * Regular user. Only allow this with a special mount
2015 * option, when the user has write+exec access to the
2016 * subvol root, and when rmdir(2) would have been
2019 * Note that this is _not_ check that the subvol is
2020 * empty or doesn't contain data that we wouldn't
2021 * otherwise be able to delete.
2023 * Users who want to delete empty subvols should try
2027 if (!btrfs_test_opt(root
, USER_SUBVOL_RM_ALLOWED
))
2031 * Do not allow deletion if the parent dir is the same
2032 * as the dir to be deleted. That means the ioctl
2033 * must be called on the dentry referencing the root
2034 * of the subvol, not a random directory contained
2041 err
= inode_permission(inode
, MAY_WRITE
| MAY_EXEC
);
2045 /* check if subvolume may be deleted by a non-root user */
2046 err
= btrfs_may_delete(dir
, dentry
, 1);
2051 if (btrfs_ino(inode
) != BTRFS_FIRST_FREE_OBJECTID
) {
2056 mutex_lock(&inode
->i_mutex
);
2057 err
= d_invalidate(dentry
);
2061 down_write(&root
->fs_info
->subvol_sem
);
2063 err
= may_destroy_subvol(dest
);
2067 trans
= btrfs_start_transaction(root
, 0);
2068 if (IS_ERR(trans
)) {
2069 err
= PTR_ERR(trans
);
2072 trans
->block_rsv
= &root
->fs_info
->global_block_rsv
;
2074 ret
= btrfs_unlink_subvol(trans
, root
, dir
,
2075 dest
->root_key
.objectid
,
2076 dentry
->d_name
.name
,
2077 dentry
->d_name
.len
);
2080 btrfs_abort_transaction(trans
, root
, ret
);
2084 btrfs_record_root_in_trans(trans
, dest
);
2086 memset(&dest
->root_item
.drop_progress
, 0,
2087 sizeof(dest
->root_item
.drop_progress
));
2088 dest
->root_item
.drop_level
= 0;
2089 btrfs_set_root_refs(&dest
->root_item
, 0);
2091 if (!xchg(&dest
->orphan_item_inserted
, 1)) {
2092 ret
= btrfs_insert_orphan_item(trans
,
2093 root
->fs_info
->tree_root
,
2094 dest
->root_key
.objectid
);
2096 btrfs_abort_transaction(trans
, root
, ret
);
2102 ret
= btrfs_end_transaction(trans
, root
);
2105 inode
->i_flags
|= S_DEAD
;
2107 up_write(&root
->fs_info
->subvol_sem
);
2109 mutex_unlock(&inode
->i_mutex
);
2111 shrink_dcache_sb(root
->fs_info
->sb
);
2112 btrfs_invalidate_inodes(dest
);
2118 mutex_unlock(&dir
->i_mutex
);
2119 mnt_drop_write_file(file
);
2125 static int btrfs_ioctl_defrag(struct file
*file
, void __user
*argp
)
2127 struct inode
*inode
= fdentry(file
)->d_inode
;
2128 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
2129 struct btrfs_ioctl_defrag_range_args
*range
;
2132 if (btrfs_root_readonly(root
))
2135 ret
= mnt_want_write_file(file
);
2139 switch (inode
->i_mode
& S_IFMT
) {
2141 if (!capable(CAP_SYS_ADMIN
)) {
2145 ret
= btrfs_defrag_root(root
, 0);
2148 ret
= btrfs_defrag_root(root
->fs_info
->extent_root
, 0);
2151 if (!(file
->f_mode
& FMODE_WRITE
)) {
2156 range
= kzalloc(sizeof(*range
), GFP_KERNEL
);
2163 if (copy_from_user(range
, argp
,
2169 /* compression requires us to start the IO */
2170 if ((range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
)) {
2171 range
->flags
|= BTRFS_DEFRAG_RANGE_START_IO
;
2172 range
->extent_thresh
= (u32
)-1;
2175 /* the rest are all set to zero by kzalloc */
2176 range
->len
= (u64
)-1;
2178 ret
= btrfs_defrag_file(fdentry(file
)->d_inode
, file
,
2188 mnt_drop_write_file(file
);
2192 static long btrfs_ioctl_add_dev(struct btrfs_root
*root
, void __user
*arg
)
2194 struct btrfs_ioctl_vol_args
*vol_args
;
2197 if (!capable(CAP_SYS_ADMIN
))
2200 mutex_lock(&root
->fs_info
->volume_mutex
);
2201 if (root
->fs_info
->balance_ctl
) {
2202 printk(KERN_INFO
"btrfs: balance in progress\n");
2207 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
2208 if (IS_ERR(vol_args
)) {
2209 ret
= PTR_ERR(vol_args
);
2213 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
2214 ret
= btrfs_init_new_device(root
, vol_args
->name
);
2218 mutex_unlock(&root
->fs_info
->volume_mutex
);
2222 static long btrfs_ioctl_rm_dev(struct btrfs_root
*root
, void __user
*arg
)
2224 struct btrfs_ioctl_vol_args
*vol_args
;
2227 if (!capable(CAP_SYS_ADMIN
))
2230 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
)
2233 mutex_lock(&root
->fs_info
->volume_mutex
);
2234 if (root
->fs_info
->balance_ctl
) {
2235 printk(KERN_INFO
"btrfs: balance in progress\n");
2240 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
2241 if (IS_ERR(vol_args
)) {
2242 ret
= PTR_ERR(vol_args
);
2246 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
2247 ret
= btrfs_rm_device(root
, vol_args
->name
);
2251 mutex_unlock(&root
->fs_info
->volume_mutex
);
2255 static long btrfs_ioctl_fs_info(struct btrfs_root
*root
, void __user
*arg
)
2257 struct btrfs_ioctl_fs_info_args
*fi_args
;
2258 struct btrfs_device
*device
;
2259 struct btrfs_device
*next
;
2260 struct btrfs_fs_devices
*fs_devices
= root
->fs_info
->fs_devices
;
2263 if (!capable(CAP_SYS_ADMIN
))
2266 fi_args
= kzalloc(sizeof(*fi_args
), GFP_KERNEL
);
2270 fi_args
->num_devices
= fs_devices
->num_devices
;
2271 memcpy(&fi_args
->fsid
, root
->fs_info
->fsid
, sizeof(fi_args
->fsid
));
2273 mutex_lock(&fs_devices
->device_list_mutex
);
2274 list_for_each_entry_safe(device
, next
, &fs_devices
->devices
, dev_list
) {
2275 if (device
->devid
> fi_args
->max_id
)
2276 fi_args
->max_id
= device
->devid
;
2278 mutex_unlock(&fs_devices
->device_list_mutex
);
2280 if (copy_to_user(arg
, fi_args
, sizeof(*fi_args
)))
2287 static long btrfs_ioctl_dev_info(struct btrfs_root
*root
, void __user
*arg
)
2289 struct btrfs_ioctl_dev_info_args
*di_args
;
2290 struct btrfs_device
*dev
;
2291 struct btrfs_fs_devices
*fs_devices
= root
->fs_info
->fs_devices
;
2293 char *s_uuid
= NULL
;
2294 char empty_uuid
[BTRFS_UUID_SIZE
] = {0};
2296 if (!capable(CAP_SYS_ADMIN
))
2299 di_args
= memdup_user(arg
, sizeof(*di_args
));
2300 if (IS_ERR(di_args
))
2301 return PTR_ERR(di_args
);
2303 if (memcmp(empty_uuid
, di_args
->uuid
, BTRFS_UUID_SIZE
) != 0)
2304 s_uuid
= di_args
->uuid
;
2306 mutex_lock(&fs_devices
->device_list_mutex
);
2307 dev
= btrfs_find_device(root
, di_args
->devid
, s_uuid
, NULL
);
2308 mutex_unlock(&fs_devices
->device_list_mutex
);
2315 di_args
->devid
= dev
->devid
;
2316 di_args
->bytes_used
= dev
->bytes_used
;
2317 di_args
->total_bytes
= dev
->total_bytes
;
2318 memcpy(di_args
->uuid
, dev
->uuid
, sizeof(di_args
->uuid
));
2320 struct rcu_string
*name
;
2323 name
= rcu_dereference(dev
->name
);
2324 strncpy(di_args
->path
, name
->str
, sizeof(di_args
->path
));
2326 di_args
->path
[sizeof(di_args
->path
) - 1] = 0;
2328 di_args
->path
[0] = '\0';
2332 if (ret
== 0 && copy_to_user(arg
, di_args
, sizeof(*di_args
)))
2339 static noinline
long btrfs_ioctl_clone(struct file
*file
, unsigned long srcfd
,
2340 u64 off
, u64 olen
, u64 destoff
)
2342 struct inode
*inode
= fdentry(file
)->d_inode
;
2343 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
2344 struct file
*src_file
;
2346 struct btrfs_trans_handle
*trans
;
2347 struct btrfs_path
*path
;
2348 struct extent_buffer
*leaf
;
2350 struct btrfs_key key
;
2355 u64 bs
= root
->fs_info
->sb
->s_blocksize
;
2360 * - split compressed inline extents. annoying: we need to
2361 * decompress into destination's address_space (the file offset
2362 * may change, so source mapping won't do), then recompress (or
2363 * otherwise reinsert) a subrange.
2364 * - allow ranges within the same file to be cloned (provided
2365 * they don't overlap)?
2368 /* the destination must be opened for writing */
2369 if (!(file
->f_mode
& FMODE_WRITE
) || (file
->f_flags
& O_APPEND
))
2372 if (btrfs_root_readonly(root
))
2375 ret
= mnt_want_write_file(file
);
2379 src_file
= fget(srcfd
);
2382 goto out_drop_write
;
2386 if (src_file
->f_path
.mnt
!= file
->f_path
.mnt
)
2389 src
= src_file
->f_dentry
->d_inode
;
2395 /* the src must be open for reading */
2396 if (!(src_file
->f_mode
& FMODE_READ
))
2399 /* don't make the dst file partly checksummed */
2400 if ((BTRFS_I(src
)->flags
& BTRFS_INODE_NODATASUM
) !=
2401 (BTRFS_I(inode
)->flags
& BTRFS_INODE_NODATASUM
))
2405 if (S_ISDIR(src
->i_mode
) || S_ISDIR(inode
->i_mode
))
2409 if (src
->i_sb
!= inode
->i_sb
)
2413 buf
= vmalloc(btrfs_level_size(root
, 0));
2417 path
= btrfs_alloc_path();
2425 mutex_lock_nested(&inode
->i_mutex
, I_MUTEX_PARENT
);
2426 mutex_lock_nested(&src
->i_mutex
, I_MUTEX_CHILD
);
2428 mutex_lock_nested(&src
->i_mutex
, I_MUTEX_PARENT
);
2429 mutex_lock_nested(&inode
->i_mutex
, I_MUTEX_CHILD
);
2432 /* determine range to clone */
2434 if (off
+ len
> src
->i_size
|| off
+ len
< off
)
2437 olen
= len
= src
->i_size
- off
;
2438 /* if we extend to eof, continue to block boundary */
2439 if (off
+ len
== src
->i_size
)
2440 len
= ALIGN(src
->i_size
, bs
) - off
;
2442 /* verify the end result is block aligned */
2443 if (!IS_ALIGNED(off
, bs
) || !IS_ALIGNED(off
+ len
, bs
) ||
2444 !IS_ALIGNED(destoff
, bs
))
2447 if (destoff
> inode
->i_size
) {
2448 ret
= btrfs_cont_expand(inode
, inode
->i_size
, destoff
);
2453 /* truncate page cache pages from target inode range */
2454 truncate_inode_pages_range(&inode
->i_data
, destoff
,
2455 PAGE_CACHE_ALIGN(destoff
+ len
) - 1);
2457 /* do any pending delalloc/csum calc on src, one way or
2458 another, and lock file content */
2460 struct btrfs_ordered_extent
*ordered
;
2461 lock_extent(&BTRFS_I(src
)->io_tree
, off
, off
+len
);
2462 ordered
= btrfs_lookup_first_ordered_extent(src
, off
+len
);
2464 !test_range_bit(&BTRFS_I(src
)->io_tree
, off
, off
+len
,
2465 EXTENT_DELALLOC
, 0, NULL
))
2467 unlock_extent(&BTRFS_I(src
)->io_tree
, off
, off
+len
);
2469 btrfs_put_ordered_extent(ordered
);
2470 btrfs_wait_ordered_range(src
, off
, len
);
2474 key
.objectid
= btrfs_ino(src
);
2475 key
.type
= BTRFS_EXTENT_DATA_KEY
;
2480 * note the key will change type as we walk through the
2483 ret
= btrfs_search_slot(NULL
, BTRFS_I(src
)->root
, &key
, path
,
2488 nritems
= btrfs_header_nritems(path
->nodes
[0]);
2489 if (path
->slots
[0] >= nritems
) {
2490 ret
= btrfs_next_leaf(BTRFS_I(src
)->root
, path
);
2495 nritems
= btrfs_header_nritems(path
->nodes
[0]);
2497 leaf
= path
->nodes
[0];
2498 slot
= path
->slots
[0];
2500 btrfs_item_key_to_cpu(leaf
, &key
, slot
);
2501 if (btrfs_key_type(&key
) > BTRFS_EXTENT_DATA_KEY
||
2502 key
.objectid
!= btrfs_ino(src
))
2505 if (btrfs_key_type(&key
) == BTRFS_EXTENT_DATA_KEY
) {
2506 struct btrfs_file_extent_item
*extent
;
2509 struct btrfs_key new_key
;
2510 u64 disko
= 0, diskl
= 0;
2511 u64 datao
= 0, datal
= 0;
2515 size
= btrfs_item_size_nr(leaf
, slot
);
2516 read_extent_buffer(leaf
, buf
,
2517 btrfs_item_ptr_offset(leaf
, slot
),
2520 extent
= btrfs_item_ptr(leaf
, slot
,
2521 struct btrfs_file_extent_item
);
2522 comp
= btrfs_file_extent_compression(leaf
, extent
);
2523 type
= btrfs_file_extent_type(leaf
, extent
);
2524 if (type
== BTRFS_FILE_EXTENT_REG
||
2525 type
== BTRFS_FILE_EXTENT_PREALLOC
) {
2526 disko
= btrfs_file_extent_disk_bytenr(leaf
,
2528 diskl
= btrfs_file_extent_disk_num_bytes(leaf
,
2530 datao
= btrfs_file_extent_offset(leaf
, extent
);
2531 datal
= btrfs_file_extent_num_bytes(leaf
,
2533 } else if (type
== BTRFS_FILE_EXTENT_INLINE
) {
2534 /* take upper bound, may be compressed */
2535 datal
= btrfs_file_extent_ram_bytes(leaf
,
2538 btrfs_release_path(path
);
2540 if (key
.offset
+ datal
<= off
||
2541 key
.offset
>= off
+len
)
2544 memcpy(&new_key
, &key
, sizeof(new_key
));
2545 new_key
.objectid
= btrfs_ino(inode
);
2546 if (off
<= key
.offset
)
2547 new_key
.offset
= key
.offset
+ destoff
- off
;
2549 new_key
.offset
= destoff
;
2552 * 1 - adjusting old extent (we may have to split it)
2553 * 1 - add new extent
2556 trans
= btrfs_start_transaction(root
, 3);
2557 if (IS_ERR(trans
)) {
2558 ret
= PTR_ERR(trans
);
2562 if (type
== BTRFS_FILE_EXTENT_REG
||
2563 type
== BTRFS_FILE_EXTENT_PREALLOC
) {
2565 * a | --- range to clone ---| b
2566 * | ------------- extent ------------- |
2569 /* substract range b */
2570 if (key
.offset
+ datal
> off
+ len
)
2571 datal
= off
+ len
- key
.offset
;
2573 /* substract range a */
2574 if (off
> key
.offset
) {
2575 datao
+= off
- key
.offset
;
2576 datal
-= off
- key
.offset
;
2579 ret
= btrfs_drop_extents(trans
, inode
,
2581 new_key
.offset
+ datal
,
2584 btrfs_abort_transaction(trans
, root
,
2586 btrfs_end_transaction(trans
, root
);
2590 ret
= btrfs_insert_empty_item(trans
, root
, path
,
2593 btrfs_abort_transaction(trans
, root
,
2595 btrfs_end_transaction(trans
, root
);
2599 leaf
= path
->nodes
[0];
2600 slot
= path
->slots
[0];
2601 write_extent_buffer(leaf
, buf
,
2602 btrfs_item_ptr_offset(leaf
, slot
),
2605 extent
= btrfs_item_ptr(leaf
, slot
,
2606 struct btrfs_file_extent_item
);
2608 /* disko == 0 means it's a hole */
2612 btrfs_set_file_extent_offset(leaf
, extent
,
2614 btrfs_set_file_extent_num_bytes(leaf
, extent
,
2617 inode_add_bytes(inode
, datal
);
2618 ret
= btrfs_inc_extent_ref(trans
, root
,
2620 root
->root_key
.objectid
,
2622 new_key
.offset
- datao
,
2625 btrfs_abort_transaction(trans
,
2628 btrfs_end_transaction(trans
,
2634 } else if (type
== BTRFS_FILE_EXTENT_INLINE
) {
2637 if (off
> key
.offset
) {
2638 skip
= off
- key
.offset
;
2639 new_key
.offset
+= skip
;
2642 if (key
.offset
+ datal
> off
+len
)
2643 trim
= key
.offset
+ datal
- (off
+len
);
2645 if (comp
&& (skip
|| trim
)) {
2647 btrfs_end_transaction(trans
, root
);
2650 size
-= skip
+ trim
;
2651 datal
-= skip
+ trim
;
2653 ret
= btrfs_drop_extents(trans
, inode
,
2655 new_key
.offset
+ datal
,
2658 btrfs_abort_transaction(trans
, root
,
2660 btrfs_end_transaction(trans
, root
);
2664 ret
= btrfs_insert_empty_item(trans
, root
, path
,
2667 btrfs_abort_transaction(trans
, root
,
2669 btrfs_end_transaction(trans
, root
);
2675 btrfs_file_extent_calc_inline_size(0);
2676 memmove(buf
+start
, buf
+start
+skip
,
2680 leaf
= path
->nodes
[0];
2681 slot
= path
->slots
[0];
2682 write_extent_buffer(leaf
, buf
,
2683 btrfs_item_ptr_offset(leaf
, slot
),
2685 inode_add_bytes(inode
, datal
);
2688 btrfs_mark_buffer_dirty(leaf
);
2689 btrfs_release_path(path
);
2691 inode_inc_iversion(inode
);
2692 inode
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
2695 * we round up to the block size at eof when
2696 * determining which extents to clone above,
2697 * but shouldn't round up the file size
2699 endoff
= new_key
.offset
+ datal
;
2700 if (endoff
> destoff
+olen
)
2701 endoff
= destoff
+olen
;
2702 if (endoff
> inode
->i_size
)
2703 btrfs_i_size_write(inode
, endoff
);
2705 ret
= btrfs_update_inode(trans
, root
, inode
);
2707 btrfs_abort_transaction(trans
, root
, ret
);
2708 btrfs_end_transaction(trans
, root
);
2711 ret
= btrfs_end_transaction(trans
, root
);
2714 btrfs_release_path(path
);
2719 btrfs_release_path(path
);
2720 unlock_extent(&BTRFS_I(src
)->io_tree
, off
, off
+len
);
2722 mutex_unlock(&src
->i_mutex
);
2723 mutex_unlock(&inode
->i_mutex
);
2725 btrfs_free_path(path
);
2729 mnt_drop_write_file(file
);
2733 static long btrfs_ioctl_clone_range(struct file
*file
, void __user
*argp
)
2735 struct btrfs_ioctl_clone_range_args args
;
2737 if (copy_from_user(&args
, argp
, sizeof(args
)))
2739 return btrfs_ioctl_clone(file
, args
.src_fd
, args
.src_offset
,
2740 args
.src_length
, args
.dest_offset
);
2744 * there are many ways the trans_start and trans_end ioctls can lead
2745 * to deadlocks. They should only be used by applications that
2746 * basically own the machine, and have a very in depth understanding
2747 * of all the possible deadlocks and enospc problems.
2749 static long btrfs_ioctl_trans_start(struct file
*file
)
2751 struct inode
*inode
= fdentry(file
)->d_inode
;
2752 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
2753 struct btrfs_trans_handle
*trans
;
2757 if (!capable(CAP_SYS_ADMIN
))
2761 if (file
->private_data
)
2765 if (btrfs_root_readonly(root
))
2768 ret
= mnt_want_write_file(file
);
2772 atomic_inc(&root
->fs_info
->open_ioctl_trans
);
2775 trans
= btrfs_start_ioctl_transaction(root
);
2779 file
->private_data
= trans
;
2783 atomic_dec(&root
->fs_info
->open_ioctl_trans
);
2784 mnt_drop_write_file(file
);
2789 static long btrfs_ioctl_default_subvol(struct file
*file
, void __user
*argp
)
2791 struct inode
*inode
= fdentry(file
)->d_inode
;
2792 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
2793 struct btrfs_root
*new_root
;
2794 struct btrfs_dir_item
*di
;
2795 struct btrfs_trans_handle
*trans
;
2796 struct btrfs_path
*path
;
2797 struct btrfs_key location
;
2798 struct btrfs_disk_key disk_key
;
2802 if (!capable(CAP_SYS_ADMIN
))
2805 if (copy_from_user(&objectid
, argp
, sizeof(objectid
)))
2809 objectid
= root
->root_key
.objectid
;
2811 location
.objectid
= objectid
;
2812 location
.type
= BTRFS_ROOT_ITEM_KEY
;
2813 location
.offset
= (u64
)-1;
2815 new_root
= btrfs_read_fs_root_no_name(root
->fs_info
, &location
);
2816 if (IS_ERR(new_root
))
2817 return PTR_ERR(new_root
);
2819 if (btrfs_root_refs(&new_root
->root_item
) == 0)
2822 path
= btrfs_alloc_path();
2825 path
->leave_spinning
= 1;
2827 trans
= btrfs_start_transaction(root
, 1);
2828 if (IS_ERR(trans
)) {
2829 btrfs_free_path(path
);
2830 return PTR_ERR(trans
);
2833 dir_id
= btrfs_super_root_dir(root
->fs_info
->super_copy
);
2834 di
= btrfs_lookup_dir_item(trans
, root
->fs_info
->tree_root
, path
,
2835 dir_id
, "default", 7, 1);
2836 if (IS_ERR_OR_NULL(di
)) {
2837 btrfs_free_path(path
);
2838 btrfs_end_transaction(trans
, root
);
2839 printk(KERN_ERR
"Umm, you don't have the default dir item, "
2840 "this isn't going to work\n");
2844 btrfs_cpu_key_to_disk(&disk_key
, &new_root
->root_key
);
2845 btrfs_set_dir_item_key(path
->nodes
[0], di
, &disk_key
);
2846 btrfs_mark_buffer_dirty(path
->nodes
[0]);
2847 btrfs_free_path(path
);
2849 btrfs_set_fs_incompat(root
->fs_info
, DEFAULT_SUBVOL
);
2850 btrfs_end_transaction(trans
, root
);
2855 static void get_block_group_info(struct list_head
*groups_list
,
2856 struct btrfs_ioctl_space_info
*space
)
2858 struct btrfs_block_group_cache
*block_group
;
2860 space
->total_bytes
= 0;
2861 space
->used_bytes
= 0;
2863 list_for_each_entry(block_group
, groups_list
, list
) {
2864 space
->flags
= block_group
->flags
;
2865 space
->total_bytes
+= block_group
->key
.offset
;
2866 space
->used_bytes
+=
2867 btrfs_block_group_used(&block_group
->item
);
2871 long btrfs_ioctl_space_info(struct btrfs_root
*root
, void __user
*arg
)
2873 struct btrfs_ioctl_space_args space_args
;
2874 struct btrfs_ioctl_space_info space
;
2875 struct btrfs_ioctl_space_info
*dest
;
2876 struct btrfs_ioctl_space_info
*dest_orig
;
2877 struct btrfs_ioctl_space_info __user
*user_dest
;
2878 struct btrfs_space_info
*info
;
2879 u64 types
[] = {BTRFS_BLOCK_GROUP_DATA
,
2880 BTRFS_BLOCK_GROUP_SYSTEM
,
2881 BTRFS_BLOCK_GROUP_METADATA
,
2882 BTRFS_BLOCK_GROUP_DATA
| BTRFS_BLOCK_GROUP_METADATA
};
2889 if (copy_from_user(&space_args
,
2890 (struct btrfs_ioctl_space_args __user
*)arg
,
2891 sizeof(space_args
)))
2894 for (i
= 0; i
< num_types
; i
++) {
2895 struct btrfs_space_info
*tmp
;
2899 list_for_each_entry_rcu(tmp
, &root
->fs_info
->space_info
,
2901 if (tmp
->flags
== types
[i
]) {
2911 down_read(&info
->groups_sem
);
2912 for (c
= 0; c
< BTRFS_NR_RAID_TYPES
; c
++) {
2913 if (!list_empty(&info
->block_groups
[c
]))
2916 up_read(&info
->groups_sem
);
2919 /* space_slots == 0 means they are asking for a count */
2920 if (space_args
.space_slots
== 0) {
2921 space_args
.total_spaces
= slot_count
;
2925 slot_count
= min_t(u64
, space_args
.space_slots
, slot_count
);
2927 alloc_size
= sizeof(*dest
) * slot_count
;
2929 /* we generally have at most 6 or so space infos, one for each raid
2930 * level. So, a whole page should be more than enough for everyone
2932 if (alloc_size
> PAGE_CACHE_SIZE
)
2935 space_args
.total_spaces
= 0;
2936 dest
= kmalloc(alloc_size
, GFP_NOFS
);
2941 /* now we have a buffer to copy into */
2942 for (i
= 0; i
< num_types
; i
++) {
2943 struct btrfs_space_info
*tmp
;
2950 list_for_each_entry_rcu(tmp
, &root
->fs_info
->space_info
,
2952 if (tmp
->flags
== types
[i
]) {
2961 down_read(&info
->groups_sem
);
2962 for (c
= 0; c
< BTRFS_NR_RAID_TYPES
; c
++) {
2963 if (!list_empty(&info
->block_groups
[c
])) {
2964 get_block_group_info(&info
->block_groups
[c
],
2966 memcpy(dest
, &space
, sizeof(space
));
2968 space_args
.total_spaces
++;
2974 up_read(&info
->groups_sem
);
2977 user_dest
= (struct btrfs_ioctl_space_info __user
*)
2978 (arg
+ sizeof(struct btrfs_ioctl_space_args
));
2980 if (copy_to_user(user_dest
, dest_orig
, alloc_size
))
2985 if (ret
== 0 && copy_to_user(arg
, &space_args
, sizeof(space_args
)))
2992 * there are many ways the trans_start and trans_end ioctls can lead
2993 * to deadlocks. They should only be used by applications that
2994 * basically own the machine, and have a very in depth understanding
2995 * of all the possible deadlocks and enospc problems.
2997 long btrfs_ioctl_trans_end(struct file
*file
)
2999 struct inode
*inode
= fdentry(file
)->d_inode
;
3000 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
3001 struct btrfs_trans_handle
*trans
;
3003 trans
= file
->private_data
;
3006 file
->private_data
= NULL
;
3008 btrfs_end_transaction(trans
, root
);
3010 atomic_dec(&root
->fs_info
->open_ioctl_trans
);
3012 mnt_drop_write_file(file
);
3016 static noinline
long btrfs_ioctl_start_sync(struct file
*file
, void __user
*argp
)
3018 struct btrfs_root
*root
= BTRFS_I(file
->f_dentry
->d_inode
)->root
;
3019 struct btrfs_trans_handle
*trans
;
3023 trans
= btrfs_start_transaction(root
, 0);
3025 return PTR_ERR(trans
);
3026 transid
= trans
->transid
;
3027 ret
= btrfs_commit_transaction_async(trans
, root
, 0);
3029 btrfs_end_transaction(trans
, root
);
3034 if (copy_to_user(argp
, &transid
, sizeof(transid
)))
3039 static noinline
long btrfs_ioctl_wait_sync(struct file
*file
, void __user
*argp
)
3041 struct btrfs_root
*root
= BTRFS_I(file
->f_dentry
->d_inode
)->root
;
3045 if (copy_from_user(&transid
, argp
, sizeof(transid
)))
3048 transid
= 0; /* current trans */
3050 return btrfs_wait_for_commit(root
, transid
);
3053 static long btrfs_ioctl_scrub(struct btrfs_root
*root
, void __user
*arg
)
3056 struct btrfs_ioctl_scrub_args
*sa
;
3058 if (!capable(CAP_SYS_ADMIN
))
3061 sa
= memdup_user(arg
, sizeof(*sa
));
3065 ret
= btrfs_scrub_dev(root
, sa
->devid
, sa
->start
, sa
->end
,
3066 &sa
->progress
, sa
->flags
& BTRFS_SCRUB_READONLY
);
3068 if (copy_to_user(arg
, sa
, sizeof(*sa
)))
3075 static long btrfs_ioctl_scrub_cancel(struct btrfs_root
*root
, void __user
*arg
)
3077 if (!capable(CAP_SYS_ADMIN
))
3080 return btrfs_scrub_cancel(root
);
3083 static long btrfs_ioctl_scrub_progress(struct btrfs_root
*root
,
3086 struct btrfs_ioctl_scrub_args
*sa
;
3089 if (!capable(CAP_SYS_ADMIN
))
3092 sa
= memdup_user(arg
, sizeof(*sa
));
3096 ret
= btrfs_scrub_progress(root
, sa
->devid
, &sa
->progress
);
3098 if (copy_to_user(arg
, sa
, sizeof(*sa
)))
3105 static long btrfs_ioctl_get_dev_stats(struct btrfs_root
*root
,
3108 struct btrfs_ioctl_get_dev_stats
*sa
;
3111 sa
= memdup_user(arg
, sizeof(*sa
));
3115 if ((sa
->flags
& BTRFS_DEV_STATS_RESET
) && !capable(CAP_SYS_ADMIN
)) {
3120 ret
= btrfs_get_dev_stats(root
, sa
);
3122 if (copy_to_user(arg
, sa
, sizeof(*sa
)))
3129 static long btrfs_ioctl_ino_to_path(struct btrfs_root
*root
, void __user
*arg
)
3135 struct btrfs_ioctl_ino_path_args
*ipa
= NULL
;
3136 struct inode_fs_paths
*ipath
= NULL
;
3137 struct btrfs_path
*path
;
3139 if (!capable(CAP_SYS_ADMIN
))
3142 path
= btrfs_alloc_path();
3148 ipa
= memdup_user(arg
, sizeof(*ipa
));
3155 size
= min_t(u32
, ipa
->size
, 4096);
3156 ipath
= init_ipath(size
, root
, path
);
3157 if (IS_ERR(ipath
)) {
3158 ret
= PTR_ERR(ipath
);
3163 ret
= paths_from_inode(ipa
->inum
, ipath
);
3167 for (i
= 0; i
< ipath
->fspath
->elem_cnt
; ++i
) {
3168 rel_ptr
= ipath
->fspath
->val
[i
] -
3169 (u64
)(unsigned long)ipath
->fspath
->val
;
3170 ipath
->fspath
->val
[i
] = rel_ptr
;
3173 ret
= copy_to_user((void *)(unsigned long)ipa
->fspath
,
3174 (void *)(unsigned long)ipath
->fspath
, size
);
3181 btrfs_free_path(path
);
3188 static int build_ino_list(u64 inum
, u64 offset
, u64 root
, void *ctx
)
3190 struct btrfs_data_container
*inodes
= ctx
;
3191 const size_t c
= 3 * sizeof(u64
);
3193 if (inodes
->bytes_left
>= c
) {
3194 inodes
->bytes_left
-= c
;
3195 inodes
->val
[inodes
->elem_cnt
] = inum
;
3196 inodes
->val
[inodes
->elem_cnt
+ 1] = offset
;
3197 inodes
->val
[inodes
->elem_cnt
+ 2] = root
;
3198 inodes
->elem_cnt
+= 3;
3200 inodes
->bytes_missing
+= c
- inodes
->bytes_left
;
3201 inodes
->bytes_left
= 0;
3202 inodes
->elem_missed
+= 3;
3208 static long btrfs_ioctl_logical_to_ino(struct btrfs_root
*root
,
3213 u64 extent_item_pos
;
3214 struct btrfs_ioctl_logical_ino_args
*loi
;
3215 struct btrfs_data_container
*inodes
= NULL
;
3216 struct btrfs_path
*path
= NULL
;
3217 struct btrfs_key key
;
3219 if (!capable(CAP_SYS_ADMIN
))
3222 loi
= memdup_user(arg
, sizeof(*loi
));
3229 path
= btrfs_alloc_path();
3235 size
= min_t(u32
, loi
->size
, 4096);
3236 inodes
= init_data_container(size
);
3237 if (IS_ERR(inodes
)) {
3238 ret
= PTR_ERR(inodes
);
3243 ret
= extent_from_logical(root
->fs_info
, loi
->logical
, path
, &key
);
3244 btrfs_release_path(path
);
3246 if (ret
& BTRFS_EXTENT_FLAG_TREE_BLOCK
)
3251 extent_item_pos
= loi
->logical
- key
.objectid
;
3252 ret
= iterate_extent_inodes(root
->fs_info
, key
.objectid
,
3253 extent_item_pos
, 0, build_ino_list
,
3259 ret
= copy_to_user((void *)(unsigned long)loi
->inodes
,
3260 (void *)(unsigned long)inodes
, size
);
3265 btrfs_free_path(path
);
3272 void update_ioctl_balance_args(struct btrfs_fs_info
*fs_info
, int lock
,
3273 struct btrfs_ioctl_balance_args
*bargs
)
3275 struct btrfs_balance_control
*bctl
= fs_info
->balance_ctl
;
3277 bargs
->flags
= bctl
->flags
;
3279 if (atomic_read(&fs_info
->balance_running
))
3280 bargs
->state
|= BTRFS_BALANCE_STATE_RUNNING
;
3281 if (atomic_read(&fs_info
->balance_pause_req
))
3282 bargs
->state
|= BTRFS_BALANCE_STATE_PAUSE_REQ
;
3283 if (atomic_read(&fs_info
->balance_cancel_req
))
3284 bargs
->state
|= BTRFS_BALANCE_STATE_CANCEL_REQ
;
3286 memcpy(&bargs
->data
, &bctl
->data
, sizeof(bargs
->data
));
3287 memcpy(&bargs
->meta
, &bctl
->meta
, sizeof(bargs
->meta
));
3288 memcpy(&bargs
->sys
, &bctl
->sys
, sizeof(bargs
->sys
));
3291 spin_lock(&fs_info
->balance_lock
);
3292 memcpy(&bargs
->stat
, &bctl
->stat
, sizeof(bargs
->stat
));
3293 spin_unlock(&fs_info
->balance_lock
);
3295 memcpy(&bargs
->stat
, &bctl
->stat
, sizeof(bargs
->stat
));
3299 static long btrfs_ioctl_balance(struct file
*file
, void __user
*arg
)
3301 struct btrfs_root
*root
= BTRFS_I(fdentry(file
)->d_inode
)->root
;
3302 struct btrfs_fs_info
*fs_info
= root
->fs_info
;
3303 struct btrfs_ioctl_balance_args
*bargs
;
3304 struct btrfs_balance_control
*bctl
;
3307 if (!capable(CAP_SYS_ADMIN
))
3310 ret
= mnt_want_write_file(file
);
3314 mutex_lock(&fs_info
->volume_mutex
);
3315 mutex_lock(&fs_info
->balance_mutex
);
3318 bargs
= memdup_user(arg
, sizeof(*bargs
));
3319 if (IS_ERR(bargs
)) {
3320 ret
= PTR_ERR(bargs
);
3324 if (bargs
->flags
& BTRFS_BALANCE_RESUME
) {
3325 if (!fs_info
->balance_ctl
) {
3330 bctl
= fs_info
->balance_ctl
;
3331 spin_lock(&fs_info
->balance_lock
);
3332 bctl
->flags
|= BTRFS_BALANCE_RESUME
;
3333 spin_unlock(&fs_info
->balance_lock
);
3341 if (fs_info
->balance_ctl
) {
3346 bctl
= kzalloc(sizeof(*bctl
), GFP_NOFS
);
3352 bctl
->fs_info
= fs_info
;
3354 memcpy(&bctl
->data
, &bargs
->data
, sizeof(bctl
->data
));
3355 memcpy(&bctl
->meta
, &bargs
->meta
, sizeof(bctl
->meta
));
3356 memcpy(&bctl
->sys
, &bargs
->sys
, sizeof(bctl
->sys
));
3358 bctl
->flags
= bargs
->flags
;
3360 /* balance everything - no filters */
3361 bctl
->flags
|= BTRFS_BALANCE_TYPE_MASK
;
3365 ret
= btrfs_balance(bctl
, bargs
);
3367 * bctl is freed in __cancel_balance or in free_fs_info if
3368 * restriper was paused all the way until unmount
3371 if (copy_to_user(arg
, bargs
, sizeof(*bargs
)))
3378 mutex_unlock(&fs_info
->balance_mutex
);
3379 mutex_unlock(&fs_info
->volume_mutex
);
3380 mnt_drop_write_file(file
);
3384 static long btrfs_ioctl_balance_ctl(struct btrfs_root
*root
, int cmd
)
3386 if (!capable(CAP_SYS_ADMIN
))
3390 case BTRFS_BALANCE_CTL_PAUSE
:
3391 return btrfs_pause_balance(root
->fs_info
);
3392 case BTRFS_BALANCE_CTL_CANCEL
:
3393 return btrfs_cancel_balance(root
->fs_info
);
3399 static long btrfs_ioctl_balance_progress(struct btrfs_root
*root
,
3402 struct btrfs_fs_info
*fs_info
= root
->fs_info
;
3403 struct btrfs_ioctl_balance_args
*bargs
;
3406 if (!capable(CAP_SYS_ADMIN
))
3409 mutex_lock(&fs_info
->balance_mutex
);
3410 if (!fs_info
->balance_ctl
) {
3415 bargs
= kzalloc(sizeof(*bargs
), GFP_NOFS
);
3421 update_ioctl_balance_args(fs_info
, 1, bargs
);
3423 if (copy_to_user(arg
, bargs
, sizeof(*bargs
)))
3428 mutex_unlock(&fs_info
->balance_mutex
);
3432 static long btrfs_ioctl_quota_ctl(struct btrfs_root
*root
, void __user
*arg
)
3434 struct btrfs_ioctl_quota_ctl_args
*sa
;
3435 struct btrfs_trans_handle
*trans
= NULL
;
3439 if (!capable(CAP_SYS_ADMIN
))
3442 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
)
3445 sa
= memdup_user(arg
, sizeof(*sa
));
3449 if (sa
->cmd
!= BTRFS_QUOTA_CTL_RESCAN
) {
3450 trans
= btrfs_start_transaction(root
, 2);
3451 if (IS_ERR(trans
)) {
3452 ret
= PTR_ERR(trans
);
3458 case BTRFS_QUOTA_CTL_ENABLE
:
3459 ret
= btrfs_quota_enable(trans
, root
->fs_info
);
3461 case BTRFS_QUOTA_CTL_DISABLE
:
3462 ret
= btrfs_quota_disable(trans
, root
->fs_info
);
3464 case BTRFS_QUOTA_CTL_RESCAN
:
3465 ret
= btrfs_quota_rescan(root
->fs_info
);
3472 if (copy_to_user(arg
, sa
, sizeof(*sa
)))
3476 err
= btrfs_commit_transaction(trans
, root
);
3486 static long btrfs_ioctl_qgroup_assign(struct btrfs_root
*root
, void __user
*arg
)
3488 struct btrfs_ioctl_qgroup_assign_args
*sa
;
3489 struct btrfs_trans_handle
*trans
;
3493 if (!capable(CAP_SYS_ADMIN
))
3496 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
)
3499 sa
= memdup_user(arg
, sizeof(*sa
));
3503 trans
= btrfs_join_transaction(root
);
3504 if (IS_ERR(trans
)) {
3505 ret
= PTR_ERR(trans
);
3509 /* FIXME: check if the IDs really exist */
3511 ret
= btrfs_add_qgroup_relation(trans
, root
->fs_info
,
3514 ret
= btrfs_del_qgroup_relation(trans
, root
->fs_info
,
3518 err
= btrfs_end_transaction(trans
, root
);
3527 static long btrfs_ioctl_qgroup_create(struct btrfs_root
*root
, void __user
*arg
)
3529 struct btrfs_ioctl_qgroup_create_args
*sa
;
3530 struct btrfs_trans_handle
*trans
;
3534 if (!capable(CAP_SYS_ADMIN
))
3537 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
)
3540 sa
= memdup_user(arg
, sizeof(*sa
));
3544 trans
= btrfs_join_transaction(root
);
3545 if (IS_ERR(trans
)) {
3546 ret
= PTR_ERR(trans
);
3550 /* FIXME: check if the IDs really exist */
3552 ret
= btrfs_create_qgroup(trans
, root
->fs_info
, sa
->qgroupid
,
3555 ret
= btrfs_remove_qgroup(trans
, root
->fs_info
, sa
->qgroupid
);
3558 err
= btrfs_end_transaction(trans
, root
);
3567 static long btrfs_ioctl_qgroup_limit(struct btrfs_root
*root
, void __user
*arg
)
3569 struct btrfs_ioctl_qgroup_limit_args
*sa
;
3570 struct btrfs_trans_handle
*trans
;
3575 if (!capable(CAP_SYS_ADMIN
))
3578 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
)
3581 sa
= memdup_user(arg
, sizeof(*sa
));
3585 trans
= btrfs_join_transaction(root
);
3586 if (IS_ERR(trans
)) {
3587 ret
= PTR_ERR(trans
);
3591 qgroupid
= sa
->qgroupid
;
3593 /* take the current subvol as qgroup */
3594 qgroupid
= root
->root_key
.objectid
;
3597 /* FIXME: check if the IDs really exist */
3598 ret
= btrfs_limit_qgroup(trans
, root
->fs_info
, qgroupid
, &sa
->lim
);
3600 err
= btrfs_end_transaction(trans
, root
);
3609 static long btrfs_ioctl_set_received_subvol(struct file
*file
,
3612 struct btrfs_ioctl_received_subvol_args
*sa
= NULL
;
3613 struct inode
*inode
= fdentry(file
)->d_inode
;
3614 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
3615 struct btrfs_root_item
*root_item
= &root
->root_item
;
3616 struct btrfs_trans_handle
*trans
;
3617 struct timespec ct
= CURRENT_TIME
;
3620 ret
= mnt_want_write_file(file
);
3624 down_write(&root
->fs_info
->subvol_sem
);
3626 if (btrfs_ino(inode
) != BTRFS_FIRST_FREE_OBJECTID
) {
3631 if (btrfs_root_readonly(root
)) {
3636 if (!inode_owner_or_capable(inode
)) {
3641 sa
= memdup_user(arg
, sizeof(*sa
));
3648 trans
= btrfs_start_transaction(root
, 1);
3649 if (IS_ERR(trans
)) {
3650 ret
= PTR_ERR(trans
);
3655 sa
->rtransid
= trans
->transid
;
3656 sa
->rtime
.sec
= ct
.tv_sec
;
3657 sa
->rtime
.nsec
= ct
.tv_nsec
;
3659 memcpy(root_item
->received_uuid
, sa
->uuid
, BTRFS_UUID_SIZE
);
3660 btrfs_set_root_stransid(root_item
, sa
->stransid
);
3661 btrfs_set_root_rtransid(root_item
, sa
->rtransid
);
3662 root_item
->stime
.sec
= cpu_to_le64(sa
->stime
.sec
);
3663 root_item
->stime
.nsec
= cpu_to_le32(sa
->stime
.nsec
);
3664 root_item
->rtime
.sec
= cpu_to_le64(sa
->rtime
.sec
);
3665 root_item
->rtime
.nsec
= cpu_to_le32(sa
->rtime
.nsec
);
3667 ret
= btrfs_update_root(trans
, root
->fs_info
->tree_root
,
3668 &root
->root_key
, &root
->root_item
);
3670 btrfs_end_transaction(trans
, root
);
3674 ret
= btrfs_commit_transaction(trans
, root
);
3679 ret
= copy_to_user(arg
, sa
, sizeof(*sa
));
3685 up_write(&root
->fs_info
->subvol_sem
);
3686 mnt_drop_write_file(file
);
3690 long btrfs_ioctl(struct file
*file
, unsigned int
3691 cmd
, unsigned long arg
)
3693 struct btrfs_root
*root
= BTRFS_I(fdentry(file
)->d_inode
)->root
;
3694 void __user
*argp
= (void __user
*)arg
;
3697 case FS_IOC_GETFLAGS
:
3698 return btrfs_ioctl_getflags(file
, argp
);
3699 case FS_IOC_SETFLAGS
:
3700 return btrfs_ioctl_setflags(file
, argp
);
3701 case FS_IOC_GETVERSION
:
3702 return btrfs_ioctl_getversion(file
, argp
);
3704 return btrfs_ioctl_fitrim(file
, argp
);
3705 case BTRFS_IOC_SNAP_CREATE
:
3706 return btrfs_ioctl_snap_create(file
, argp
, 0);
3707 case BTRFS_IOC_SNAP_CREATE_V2
:
3708 return btrfs_ioctl_snap_create_v2(file
, argp
, 0);
3709 case BTRFS_IOC_SUBVOL_CREATE
:
3710 return btrfs_ioctl_snap_create(file
, argp
, 1);
3711 case BTRFS_IOC_SUBVOL_CREATE_V2
:
3712 return btrfs_ioctl_snap_create_v2(file
, argp
, 1);
3713 case BTRFS_IOC_SNAP_DESTROY
:
3714 return btrfs_ioctl_snap_destroy(file
, argp
);
3715 case BTRFS_IOC_SUBVOL_GETFLAGS
:
3716 return btrfs_ioctl_subvol_getflags(file
, argp
);
3717 case BTRFS_IOC_SUBVOL_SETFLAGS
:
3718 return btrfs_ioctl_subvol_setflags(file
, argp
);
3719 case BTRFS_IOC_DEFAULT_SUBVOL
:
3720 return btrfs_ioctl_default_subvol(file
, argp
);
3721 case BTRFS_IOC_DEFRAG
:
3722 return btrfs_ioctl_defrag(file
, NULL
);
3723 case BTRFS_IOC_DEFRAG_RANGE
:
3724 return btrfs_ioctl_defrag(file
, argp
);
3725 case BTRFS_IOC_RESIZE
:
3726 return btrfs_ioctl_resize(root
, argp
);
3727 case BTRFS_IOC_ADD_DEV
:
3728 return btrfs_ioctl_add_dev(root
, argp
);
3729 case BTRFS_IOC_RM_DEV
:
3730 return btrfs_ioctl_rm_dev(root
, argp
);
3731 case BTRFS_IOC_FS_INFO
:
3732 return btrfs_ioctl_fs_info(root
, argp
);
3733 case BTRFS_IOC_DEV_INFO
:
3734 return btrfs_ioctl_dev_info(root
, argp
);
3735 case BTRFS_IOC_BALANCE
:
3736 return btrfs_ioctl_balance(file
, NULL
);
3737 case BTRFS_IOC_CLONE
:
3738 return btrfs_ioctl_clone(file
, arg
, 0, 0, 0);
3739 case BTRFS_IOC_CLONE_RANGE
:
3740 return btrfs_ioctl_clone_range(file
, argp
);
3741 case BTRFS_IOC_TRANS_START
:
3742 return btrfs_ioctl_trans_start(file
);
3743 case BTRFS_IOC_TRANS_END
:
3744 return btrfs_ioctl_trans_end(file
);
3745 case BTRFS_IOC_TREE_SEARCH
:
3746 return btrfs_ioctl_tree_search(file
, argp
);
3747 case BTRFS_IOC_INO_LOOKUP
:
3748 return btrfs_ioctl_ino_lookup(file
, argp
);
3749 case BTRFS_IOC_INO_PATHS
:
3750 return btrfs_ioctl_ino_to_path(root
, argp
);
3751 case BTRFS_IOC_LOGICAL_INO
:
3752 return btrfs_ioctl_logical_to_ino(root
, argp
);
3753 case BTRFS_IOC_SPACE_INFO
:
3754 return btrfs_ioctl_space_info(root
, argp
);
3755 case BTRFS_IOC_SYNC
:
3756 btrfs_sync_fs(file
->f_dentry
->d_sb
, 1);
3758 case BTRFS_IOC_START_SYNC
:
3759 return btrfs_ioctl_start_sync(file
, argp
);
3760 case BTRFS_IOC_WAIT_SYNC
:
3761 return btrfs_ioctl_wait_sync(file
, argp
);
3762 case BTRFS_IOC_SCRUB
:
3763 return btrfs_ioctl_scrub(root
, argp
);
3764 case BTRFS_IOC_SCRUB_CANCEL
:
3765 return btrfs_ioctl_scrub_cancel(root
, argp
);
3766 case BTRFS_IOC_SCRUB_PROGRESS
:
3767 return btrfs_ioctl_scrub_progress(root
, argp
);
3768 case BTRFS_IOC_BALANCE_V2
:
3769 return btrfs_ioctl_balance(file
, argp
);
3770 case BTRFS_IOC_BALANCE_CTL
:
3771 return btrfs_ioctl_balance_ctl(root
, arg
);
3772 case BTRFS_IOC_BALANCE_PROGRESS
:
3773 return btrfs_ioctl_balance_progress(root
, argp
);
3774 case BTRFS_IOC_SET_RECEIVED_SUBVOL
:
3775 return btrfs_ioctl_set_received_subvol(file
, argp
);
3776 case BTRFS_IOC_SEND
:
3777 return btrfs_ioctl_send(file
, argp
);
3778 case BTRFS_IOC_GET_DEV_STATS
:
3779 return btrfs_ioctl_get_dev_stats(root
, argp
);
3780 case BTRFS_IOC_QUOTA_CTL
:
3781 return btrfs_ioctl_quota_ctl(root
, argp
);
3782 case BTRFS_IOC_QGROUP_ASSIGN
:
3783 return btrfs_ioctl_qgroup_assign(root
, argp
);
3784 case BTRFS_IOC_QGROUP_CREATE
:
3785 return btrfs_ioctl_qgroup_create(root
, argp
);
3786 case BTRFS_IOC_QGROUP_LIMIT
:
3787 return btrfs_ioctl_qgroup_limit(root
, argp
);