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 * Unlike extN we don't have any flags we don't want to inherit currently.
122 void btrfs_inherit_iflags(struct inode
*inode
, struct inode
*dir
)
129 flags
= BTRFS_I(dir
)->flags
;
131 if (S_ISREG(inode
->i_mode
))
132 flags
&= ~BTRFS_INODE_DIRSYNC
;
133 else if (!S_ISDIR(inode
->i_mode
))
134 flags
&= (BTRFS_INODE_NODUMP
| BTRFS_INODE_NOATIME
);
136 BTRFS_I(inode
)->flags
= flags
;
137 btrfs_update_iflags(inode
);
140 static int btrfs_ioctl_getflags(struct file
*file
, void __user
*arg
)
142 struct btrfs_inode
*ip
= BTRFS_I(file
->f_path
.dentry
->d_inode
);
143 unsigned int flags
= btrfs_flags_to_ioctl(ip
->flags
);
145 if (copy_to_user(arg
, &flags
, sizeof(flags
)))
150 static int check_flags(unsigned int flags
)
152 if (flags
& ~(FS_IMMUTABLE_FL
| FS_APPEND_FL
| \
153 FS_NOATIME_FL
| FS_NODUMP_FL
| \
154 FS_SYNC_FL
| FS_DIRSYNC_FL
| \
155 FS_NOCOMP_FL
| FS_COMPR_FL
|
159 if ((flags
& FS_NOCOMP_FL
) && (flags
& FS_COMPR_FL
))
165 static int btrfs_ioctl_setflags(struct file
*file
, void __user
*arg
)
167 struct inode
*inode
= file
->f_path
.dentry
->d_inode
;
168 struct btrfs_inode
*ip
= BTRFS_I(inode
);
169 struct btrfs_root
*root
= ip
->root
;
170 struct btrfs_trans_handle
*trans
;
171 unsigned int flags
, oldflags
;
174 if (btrfs_root_readonly(root
))
177 if (copy_from_user(&flags
, arg
, sizeof(flags
)))
180 ret
= check_flags(flags
);
184 if (!inode_owner_or_capable(inode
))
187 mutex_lock(&inode
->i_mutex
);
189 flags
= btrfs_mask_flags(inode
->i_mode
, flags
);
190 oldflags
= btrfs_flags_to_ioctl(ip
->flags
);
191 if ((flags
^ oldflags
) & (FS_APPEND_FL
| FS_IMMUTABLE_FL
)) {
192 if (!capable(CAP_LINUX_IMMUTABLE
)) {
198 ret
= mnt_want_write(file
->f_path
.mnt
);
202 if (flags
& FS_SYNC_FL
)
203 ip
->flags
|= BTRFS_INODE_SYNC
;
205 ip
->flags
&= ~BTRFS_INODE_SYNC
;
206 if (flags
& FS_IMMUTABLE_FL
)
207 ip
->flags
|= BTRFS_INODE_IMMUTABLE
;
209 ip
->flags
&= ~BTRFS_INODE_IMMUTABLE
;
210 if (flags
& FS_APPEND_FL
)
211 ip
->flags
|= BTRFS_INODE_APPEND
;
213 ip
->flags
&= ~BTRFS_INODE_APPEND
;
214 if (flags
& FS_NODUMP_FL
)
215 ip
->flags
|= BTRFS_INODE_NODUMP
;
217 ip
->flags
&= ~BTRFS_INODE_NODUMP
;
218 if (flags
& FS_NOATIME_FL
)
219 ip
->flags
|= BTRFS_INODE_NOATIME
;
221 ip
->flags
&= ~BTRFS_INODE_NOATIME
;
222 if (flags
& FS_DIRSYNC_FL
)
223 ip
->flags
|= BTRFS_INODE_DIRSYNC
;
225 ip
->flags
&= ~BTRFS_INODE_DIRSYNC
;
226 if (flags
& FS_NOCOW_FL
)
227 ip
->flags
|= BTRFS_INODE_NODATACOW
;
229 ip
->flags
&= ~BTRFS_INODE_NODATACOW
;
232 * The COMPRESS flag can only be changed by users, while the NOCOMPRESS
233 * flag may be changed automatically if compression code won't make
236 if (flags
& FS_NOCOMP_FL
) {
237 ip
->flags
&= ~BTRFS_INODE_COMPRESS
;
238 ip
->flags
|= BTRFS_INODE_NOCOMPRESS
;
239 } else if (flags
& FS_COMPR_FL
) {
240 ip
->flags
|= BTRFS_INODE_COMPRESS
;
241 ip
->flags
&= ~BTRFS_INODE_NOCOMPRESS
;
243 ip
->flags
&= ~(BTRFS_INODE_COMPRESS
| BTRFS_INODE_NOCOMPRESS
);
246 trans
= btrfs_join_transaction(root
);
247 BUG_ON(IS_ERR(trans
));
249 ret
= btrfs_update_inode(trans
, root
, inode
);
252 btrfs_update_iflags(inode
);
253 inode
->i_ctime
= CURRENT_TIME
;
254 btrfs_end_transaction(trans
, root
);
256 mnt_drop_write(file
->f_path
.mnt
);
260 mutex_unlock(&inode
->i_mutex
);
264 static int btrfs_ioctl_getversion(struct file
*file
, int __user
*arg
)
266 struct inode
*inode
= file
->f_path
.dentry
->d_inode
;
268 return put_user(inode
->i_generation
, arg
);
271 static noinline
int btrfs_ioctl_fitrim(struct file
*file
, void __user
*arg
)
273 struct btrfs_root
*root
= fdentry(file
)->d_sb
->s_fs_info
;
274 struct btrfs_fs_info
*fs_info
= root
->fs_info
;
275 struct btrfs_device
*device
;
276 struct request_queue
*q
;
277 struct fstrim_range range
;
278 u64 minlen
= ULLONG_MAX
;
282 if (!capable(CAP_SYS_ADMIN
))
286 list_for_each_entry_rcu(device
, &fs_info
->fs_devices
->devices
,
290 q
= bdev_get_queue(device
->bdev
);
291 if (blk_queue_discard(q
)) {
293 minlen
= min((u64
)q
->limits
.discard_granularity
,
301 if (copy_from_user(&range
, arg
, sizeof(range
)))
304 range
.minlen
= max(range
.minlen
, minlen
);
305 ret
= btrfs_trim_fs(root
, &range
);
309 if (copy_to_user(arg
, &range
, sizeof(range
)))
315 static noinline
int create_subvol(struct btrfs_root
*root
,
316 struct dentry
*dentry
,
317 char *name
, int namelen
,
320 struct btrfs_trans_handle
*trans
;
321 struct btrfs_key key
;
322 struct btrfs_root_item root_item
;
323 struct btrfs_inode_item
*inode_item
;
324 struct extent_buffer
*leaf
;
325 struct btrfs_root
*new_root
;
326 struct dentry
*parent
= dget_parent(dentry
);
331 u64 new_dirid
= BTRFS_FIRST_FREE_OBJECTID
;
334 ret
= btrfs_find_free_objectid(root
->fs_info
->tree_root
, &objectid
);
340 dir
= parent
->d_inode
;
348 trans
= btrfs_start_transaction(root
, 6);
351 return PTR_ERR(trans
);
354 leaf
= btrfs_alloc_free_block(trans
, root
, root
->leafsize
,
355 0, objectid
, NULL
, 0, 0, 0);
361 memset_extent_buffer(leaf
, 0, 0, sizeof(struct btrfs_header
));
362 btrfs_set_header_bytenr(leaf
, leaf
->start
);
363 btrfs_set_header_generation(leaf
, trans
->transid
);
364 btrfs_set_header_backref_rev(leaf
, BTRFS_MIXED_BACKREF_REV
);
365 btrfs_set_header_owner(leaf
, objectid
);
367 write_extent_buffer(leaf
, root
->fs_info
->fsid
,
368 (unsigned long)btrfs_header_fsid(leaf
),
370 write_extent_buffer(leaf
, root
->fs_info
->chunk_tree_uuid
,
371 (unsigned long)btrfs_header_chunk_tree_uuid(leaf
),
373 btrfs_mark_buffer_dirty(leaf
);
375 inode_item
= &root_item
.inode
;
376 memset(inode_item
, 0, sizeof(*inode_item
));
377 inode_item
->generation
= cpu_to_le64(1);
378 inode_item
->size
= cpu_to_le64(3);
379 inode_item
->nlink
= cpu_to_le32(1);
380 inode_item
->nbytes
= cpu_to_le64(root
->leafsize
);
381 inode_item
->mode
= cpu_to_le32(S_IFDIR
| 0755);
384 root_item
.byte_limit
= 0;
385 inode_item
->flags
= cpu_to_le64(BTRFS_INODE_ROOT_ITEM_INIT
);
387 btrfs_set_root_bytenr(&root_item
, leaf
->start
);
388 btrfs_set_root_generation(&root_item
, trans
->transid
);
389 btrfs_set_root_level(&root_item
, 0);
390 btrfs_set_root_refs(&root_item
, 1);
391 btrfs_set_root_used(&root_item
, leaf
->len
);
392 btrfs_set_root_last_snapshot(&root_item
, 0);
394 memset(&root_item
.drop_progress
, 0, sizeof(root_item
.drop_progress
));
395 root_item
.drop_level
= 0;
397 btrfs_tree_unlock(leaf
);
398 free_extent_buffer(leaf
);
401 btrfs_set_root_dirid(&root_item
, new_dirid
);
403 key
.objectid
= objectid
;
405 btrfs_set_key_type(&key
, BTRFS_ROOT_ITEM_KEY
);
406 ret
= btrfs_insert_root(trans
, root
->fs_info
->tree_root
, &key
,
411 key
.offset
= (u64
)-1;
412 new_root
= btrfs_read_fs_root_no_name(root
->fs_info
, &key
);
413 BUG_ON(IS_ERR(new_root
));
415 btrfs_record_root_in_trans(trans
, new_root
);
417 ret
= btrfs_create_subvol_root(trans
, new_root
, new_dirid
);
419 * insert the directory item
421 ret
= btrfs_set_inode_index(dir
, &index
);
424 ret
= btrfs_insert_dir_item(trans
, root
,
425 name
, namelen
, dir
, &key
,
426 BTRFS_FT_DIR
, index
);
430 btrfs_i_size_write(dir
, dir
->i_size
+ namelen
* 2);
431 ret
= btrfs_update_inode(trans
, root
, dir
);
434 ret
= btrfs_add_root_ref(trans
, root
->fs_info
->tree_root
,
435 objectid
, root
->root_key
.objectid
,
436 btrfs_ino(dir
), index
, name
, namelen
);
440 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 dentry
*parent
;
460 struct btrfs_pending_snapshot
*pending_snapshot
;
461 struct btrfs_trans_handle
*trans
;
467 pending_snapshot
= kzalloc(sizeof(*pending_snapshot
), GFP_NOFS
);
468 if (!pending_snapshot
)
471 btrfs_init_block_rsv(&pending_snapshot
->block_rsv
);
472 pending_snapshot
->dentry
= dentry
;
473 pending_snapshot
->root
= root
;
474 pending_snapshot
->readonly
= readonly
;
476 trans
= btrfs_start_transaction(root
->fs_info
->extent_root
, 5);
478 ret
= PTR_ERR(trans
);
482 ret
= btrfs_snap_reserve_metadata(trans
, pending_snapshot
);
485 list_add(&pending_snapshot
->list
,
486 &trans
->transaction
->pending_snapshots
);
488 *async_transid
= trans
->transid
;
489 ret
= btrfs_commit_transaction_async(trans
,
490 root
->fs_info
->extent_root
, 1);
492 ret
= btrfs_commit_transaction(trans
,
493 root
->fs_info
->extent_root
);
497 ret
= pending_snapshot
->error
;
501 ret
= btrfs_orphan_cleanup(pending_snapshot
->snap
);
505 parent
= dget_parent(dentry
);
506 inode
= btrfs_lookup_dentry(parent
->d_inode
, dentry
);
509 ret
= PTR_ERR(inode
);
513 d_instantiate(dentry
, inode
);
516 kfree(pending_snapshot
);
520 /* copy of check_sticky in fs/namei.c()
521 * It's inline, so penalty for filesystems that don't use sticky bit is
524 static inline int btrfs_check_sticky(struct inode
*dir
, struct inode
*inode
)
526 uid_t fsuid
= current_fsuid();
528 if (!(dir
->i_mode
& S_ISVTX
))
530 if (inode
->i_uid
== fsuid
)
532 if (dir
->i_uid
== fsuid
)
534 return !capable(CAP_FOWNER
);
537 /* copy of may_delete in fs/namei.c()
538 * Check whether we can remove a link victim from directory dir, check
539 * whether the type of victim is right.
540 * 1. We can't do it if dir is read-only (done in permission())
541 * 2. We should have write and exec permissions on dir
542 * 3. We can't remove anything from append-only dir
543 * 4. We can't do anything with immutable dir (done in permission())
544 * 5. If the sticky bit on dir is set we should either
545 * a. be owner of dir, or
546 * b. be owner of victim, or
547 * c. have CAP_FOWNER capability
548 * 6. If the victim is append-only or immutable we can't do antyhing with
549 * links pointing to it.
550 * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR.
551 * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR.
552 * 9. We can't remove a root or mountpoint.
553 * 10. We don't allow removal of NFS sillyrenamed files; it's handled by
554 * nfs_async_unlink().
557 static int btrfs_may_delete(struct inode
*dir
,struct dentry
*victim
,int isdir
)
561 if (!victim
->d_inode
)
564 BUG_ON(victim
->d_parent
->d_inode
!= dir
);
565 audit_inode_child(victim
, dir
);
567 error
= inode_permission(dir
, MAY_WRITE
| MAY_EXEC
);
572 if (btrfs_check_sticky(dir
, victim
->d_inode
)||
573 IS_APPEND(victim
->d_inode
)||
574 IS_IMMUTABLE(victim
->d_inode
) || IS_SWAPFILE(victim
->d_inode
))
577 if (!S_ISDIR(victim
->d_inode
->i_mode
))
581 } else if (S_ISDIR(victim
->d_inode
->i_mode
))
585 if (victim
->d_flags
& DCACHE_NFSFS_RENAMED
)
590 /* copy of may_create in fs/namei.c() */
591 static inline int btrfs_may_create(struct inode
*dir
, struct dentry
*child
)
597 return inode_permission(dir
, MAY_WRITE
| MAY_EXEC
);
601 * Create a new subvolume below @parent. This is largely modeled after
602 * sys_mkdirat and vfs_mkdir, but we only do a single component lookup
603 * inside this filesystem so it's quite a bit simpler.
605 static noinline
int btrfs_mksubvol(struct path
*parent
,
606 char *name
, int namelen
,
607 struct btrfs_root
*snap_src
,
608 u64
*async_transid
, bool readonly
)
610 struct inode
*dir
= parent
->dentry
->d_inode
;
611 struct dentry
*dentry
;
614 mutex_lock_nested(&dir
->i_mutex
, I_MUTEX_PARENT
);
616 dentry
= lookup_one_len(name
, parent
->dentry
, namelen
);
617 error
= PTR_ERR(dentry
);
625 error
= mnt_want_write(parent
->mnt
);
629 error
= btrfs_may_create(dir
, dentry
);
633 down_read(&BTRFS_I(dir
)->root
->fs_info
->subvol_sem
);
635 if (btrfs_root_refs(&BTRFS_I(dir
)->root
->root_item
) == 0)
639 error
= create_snapshot(snap_src
, dentry
,
640 name
, namelen
, async_transid
, readonly
);
642 error
= create_subvol(BTRFS_I(dir
)->root
, dentry
,
643 name
, namelen
, async_transid
);
646 fsnotify_mkdir(dir
, dentry
);
648 up_read(&BTRFS_I(dir
)->root
->fs_info
->subvol_sem
);
650 mnt_drop_write(parent
->mnt
);
654 mutex_unlock(&dir
->i_mutex
);
659 * When we're defragging a range, we don't want to kick it off again
660 * if it is really just waiting for delalloc to send it down.
661 * If we find a nice big extent or delalloc range for the bytes in the
662 * file you want to defrag, we return 0 to let you know to skip this
665 static int check_defrag_in_cache(struct inode
*inode
, u64 offset
, int thresh
)
667 struct extent_io_tree
*io_tree
= &BTRFS_I(inode
)->io_tree
;
668 struct extent_map
*em
= NULL
;
669 struct extent_map_tree
*em_tree
= &BTRFS_I(inode
)->extent_tree
;
672 read_lock(&em_tree
->lock
);
673 em
= lookup_extent_mapping(em_tree
, offset
, PAGE_CACHE_SIZE
);
674 read_unlock(&em_tree
->lock
);
677 end
= extent_map_end(em
);
679 if (end
- offset
> thresh
)
682 /* if we already have a nice delalloc here, just stop */
684 end
= count_range_bits(io_tree
, &offset
, offset
+ thresh
,
685 thresh
, EXTENT_DELALLOC
, 1);
692 * helper function to walk through a file and find extents
693 * newer than a specific transid, and smaller than thresh.
695 * This is used by the defragging code to find new and small
698 static int find_new_extents(struct btrfs_root
*root
,
699 struct inode
*inode
, u64 newer_than
,
700 u64
*off
, int thresh
)
702 struct btrfs_path
*path
;
703 struct btrfs_key min_key
;
704 struct btrfs_key max_key
;
705 struct extent_buffer
*leaf
;
706 struct btrfs_file_extent_item
*extent
;
709 u64 ino
= btrfs_ino(inode
);
711 path
= btrfs_alloc_path();
715 min_key
.objectid
= ino
;
716 min_key
.type
= BTRFS_EXTENT_DATA_KEY
;
717 min_key
.offset
= *off
;
719 max_key
.objectid
= ino
;
720 max_key
.type
= (u8
)-1;
721 max_key
.offset
= (u64
)-1;
723 path
->keep_locks
= 1;
726 ret
= btrfs_search_forward(root
, &min_key
, &max_key
,
727 path
, 0, newer_than
);
730 if (min_key
.objectid
!= ino
)
732 if (min_key
.type
!= BTRFS_EXTENT_DATA_KEY
)
735 leaf
= path
->nodes
[0];
736 extent
= btrfs_item_ptr(leaf
, path
->slots
[0],
737 struct btrfs_file_extent_item
);
739 type
= btrfs_file_extent_type(leaf
, extent
);
740 if (type
== BTRFS_FILE_EXTENT_REG
&&
741 btrfs_file_extent_num_bytes(leaf
, extent
) < thresh
&&
742 check_defrag_in_cache(inode
, min_key
.offset
, thresh
)) {
743 *off
= min_key
.offset
;
744 btrfs_free_path(path
);
748 if (min_key
.offset
== (u64
)-1)
752 btrfs_release_path(path
);
755 btrfs_free_path(path
);
759 static int should_defrag_range(struct inode
*inode
, u64 start
, u64 len
,
760 int thresh
, u64
*last_len
, u64
*skip
,
763 struct extent_io_tree
*io_tree
= &BTRFS_I(inode
)->io_tree
;
764 struct extent_map
*em
= NULL
;
765 struct extent_map_tree
*em_tree
= &BTRFS_I(inode
)->extent_tree
;
769 * make sure that once we start defragging and extent, we keep on
772 if (start
< *defrag_end
)
778 * hopefully we have this extent in the tree already, try without
779 * the full extent lock
781 read_lock(&em_tree
->lock
);
782 em
= lookup_extent_mapping(em_tree
, start
, len
);
783 read_unlock(&em_tree
->lock
);
786 /* get the big lock and read metadata off disk */
787 lock_extent(io_tree
, start
, start
+ len
- 1, GFP_NOFS
);
788 em
= btrfs_get_extent(inode
, NULL
, 0, start
, len
, 0);
789 unlock_extent(io_tree
, start
, start
+ len
- 1, GFP_NOFS
);
795 /* this will cover holes, and inline extents */
796 if (em
->block_start
>= EXTENT_MAP_LAST_BYTE
)
800 * we hit a real extent, if it is big don't bother defragging it again
802 if ((*last_len
== 0 || *last_len
>= thresh
) && em
->len
>= thresh
)
806 * last_len ends up being a counter of how many bytes we've defragged.
807 * every time we choose not to defrag an extent, we reset *last_len
808 * so that the next tiny extent will force a defrag.
810 * The end result of this is that tiny extents before a single big
811 * extent will force at least part of that big extent to be defragged.
815 *defrag_end
= extent_map_end(em
);
818 *skip
= extent_map_end(em
);
827 * it doesn't do much good to defrag one or two pages
828 * at a time. This pulls in a nice chunk of pages
831 * It also makes sure the delalloc code has enough
832 * dirty data to avoid making new small extents as part
835 * It's a good idea to start RA on this range
836 * before calling this.
838 static int cluster_pages_for_defrag(struct inode
*inode
,
840 unsigned long start_index
,
843 unsigned long file_end
;
844 u64 isize
= i_size_read(inode
);
850 struct btrfs_ordered_extent
*ordered
;
851 struct extent_state
*cached_state
= NULL
;
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
= grab_cache_page(inode
->i_mapping
,
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 atomic_inc(&BTRFS_I(inode
)->outstanding_extents
);
940 btrfs_delalloc_release_space(inode
,
941 (num_pages
- i_done
) << PAGE_CACHE_SHIFT
);
945 btrfs_set_extent_delalloc(inode
, page_start
, page_end
- 1,
948 unlock_extent_cached(&BTRFS_I(inode
)->io_tree
,
949 page_start
, page_end
- 1, &cached_state
,
952 for (i
= 0; i
< i_done
; i
++) {
953 clear_page_dirty_for_io(pages
[i
]);
954 ClearPageChecked(pages
[i
]);
955 set_page_extent_mapped(pages
[i
]);
956 set_page_dirty(pages
[i
]);
957 unlock_page(pages
[i
]);
958 page_cache_release(pages
[i
]);
962 for (i
= 0; i
< i_done
; i
++) {
963 unlock_page(pages
[i
]);
964 page_cache_release(pages
[i
]);
966 btrfs_delalloc_release_space(inode
, num_pages
<< PAGE_CACHE_SHIFT
);
971 int btrfs_defrag_file(struct inode
*inode
, struct file
*file
,
972 struct btrfs_ioctl_defrag_range_args
*range
,
973 u64 newer_than
, unsigned long max_to_defrag
)
975 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
976 struct btrfs_super_block
*disk_super
;
977 struct file_ra_state
*ra
= NULL
;
978 unsigned long last_index
;
983 u64 newer_off
= range
->start
;
987 int defrag_count
= 0;
988 int compress_type
= BTRFS_COMPRESS_ZLIB
;
989 int extent_thresh
= range
->extent_thresh
;
990 int newer_cluster
= (256 * 1024) >> PAGE_CACHE_SHIFT
;
991 u64 new_align
= ~((u64
)128 * 1024 - 1);
992 struct page
**pages
= NULL
;
994 if (extent_thresh
== 0)
995 extent_thresh
= 256 * 1024;
997 if (range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
) {
998 if (range
->compress_type
> BTRFS_COMPRESS_TYPES
)
1000 if (range
->compress_type
)
1001 compress_type
= range
->compress_type
;
1004 if (inode
->i_size
== 0)
1008 * if we were not given a file, allocate a readahead
1012 ra
= kzalloc(sizeof(*ra
), GFP_NOFS
);
1015 file_ra_state_init(ra
, inode
->i_mapping
);
1020 pages
= kmalloc(sizeof(struct page
*) * newer_cluster
,
1027 /* find the last page to defrag */
1028 if (range
->start
+ range
->len
> range
->start
) {
1029 last_index
= min_t(u64
, inode
->i_size
- 1,
1030 range
->start
+ range
->len
- 1) >> PAGE_CACHE_SHIFT
;
1032 last_index
= (inode
->i_size
- 1) >> PAGE_CACHE_SHIFT
;
1036 ret
= find_new_extents(root
, inode
, newer_than
,
1037 &newer_off
, 64 * 1024);
1039 range
->start
= newer_off
;
1041 * we always align our defrag to help keep
1042 * the extents in the file evenly spaced
1044 i
= (newer_off
& new_align
) >> PAGE_CACHE_SHIFT
;
1045 newer_left
= newer_cluster
;
1049 i
= range
->start
>> PAGE_CACHE_SHIFT
;
1052 max_to_defrag
= last_index
- 1;
1054 while (i
<= last_index
&& defrag_count
< max_to_defrag
) {
1056 * make sure we stop running if someone unmounts
1059 if (!(inode
->i_sb
->s_flags
& MS_ACTIVE
))
1063 !should_defrag_range(inode
, (u64
)i
<< PAGE_CACHE_SHIFT
,
1070 * the should_defrag function tells us how much to skip
1071 * bump our counter by the suggested amount
1073 next
= (skip
+ PAGE_CACHE_SIZE
- 1) >> PAGE_CACHE_SHIFT
;
1074 i
= max(i
+ 1, next
);
1077 if (range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
)
1078 BTRFS_I(inode
)->force_compress
= compress_type
;
1080 btrfs_force_ra(inode
->i_mapping
, ra
, file
, i
, newer_cluster
);
1082 ret
= cluster_pages_for_defrag(inode
, pages
, i
, newer_cluster
);
1086 defrag_count
+= ret
;
1087 balance_dirty_pages_ratelimited_nr(inode
->i_mapping
, ret
);
1091 if (newer_off
== (u64
)-1)
1094 newer_off
= max(newer_off
+ 1,
1095 (u64
)i
<< PAGE_CACHE_SHIFT
);
1097 ret
= find_new_extents(root
, inode
,
1098 newer_than
, &newer_off
,
1101 range
->start
= newer_off
;
1102 i
= (newer_off
& new_align
) >> PAGE_CACHE_SHIFT
;
1103 newer_left
= newer_cluster
;
1112 if ((range
->flags
& BTRFS_DEFRAG_RANGE_START_IO
))
1113 filemap_flush(inode
->i_mapping
);
1115 if ((range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
)) {
1116 /* the filemap_flush will queue IO into the worker threads, but
1117 * we have to make sure the IO is actually started and that
1118 * ordered extents get created before we return
1120 atomic_inc(&root
->fs_info
->async_submit_draining
);
1121 while (atomic_read(&root
->fs_info
->nr_async_submits
) ||
1122 atomic_read(&root
->fs_info
->async_delalloc_pages
)) {
1123 wait_event(root
->fs_info
->async_submit_wait
,
1124 (atomic_read(&root
->fs_info
->nr_async_submits
) == 0 &&
1125 atomic_read(&root
->fs_info
->async_delalloc_pages
) == 0));
1127 atomic_dec(&root
->fs_info
->async_submit_draining
);
1129 mutex_lock(&inode
->i_mutex
);
1130 BTRFS_I(inode
)->force_compress
= BTRFS_COMPRESS_NONE
;
1131 mutex_unlock(&inode
->i_mutex
);
1134 disk_super
= &root
->fs_info
->super_copy
;
1135 features
= btrfs_super_incompat_flags(disk_super
);
1136 if (range
->compress_type
== BTRFS_COMPRESS_LZO
) {
1137 features
|= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO
;
1138 btrfs_set_super_incompat_flags(disk_super
, features
);
1143 return defrag_count
;
1152 static noinline
int btrfs_ioctl_resize(struct btrfs_root
*root
,
1158 struct btrfs_ioctl_vol_args
*vol_args
;
1159 struct btrfs_trans_handle
*trans
;
1160 struct btrfs_device
*device
= NULL
;
1162 char *devstr
= NULL
;
1166 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
)
1169 if (!capable(CAP_SYS_ADMIN
))
1172 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1173 if (IS_ERR(vol_args
))
1174 return PTR_ERR(vol_args
);
1176 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
1178 mutex_lock(&root
->fs_info
->volume_mutex
);
1179 sizestr
= vol_args
->name
;
1180 devstr
= strchr(sizestr
, ':');
1183 sizestr
= devstr
+ 1;
1185 devstr
= vol_args
->name
;
1186 devid
= simple_strtoull(devstr
, &end
, 10);
1187 printk(KERN_INFO
"resizing devid %llu\n",
1188 (unsigned long long)devid
);
1190 device
= btrfs_find_device(root
, devid
, NULL
, NULL
);
1192 printk(KERN_INFO
"resizer unable to find device %llu\n",
1193 (unsigned long long)devid
);
1197 if (!strcmp(sizestr
, "max"))
1198 new_size
= device
->bdev
->bd_inode
->i_size
;
1200 if (sizestr
[0] == '-') {
1203 } else if (sizestr
[0] == '+') {
1207 new_size
= memparse(sizestr
, NULL
);
1208 if (new_size
== 0) {
1214 old_size
= device
->total_bytes
;
1217 if (new_size
> old_size
) {
1221 new_size
= old_size
- new_size
;
1222 } else if (mod
> 0) {
1223 new_size
= old_size
+ new_size
;
1226 if (new_size
< 256 * 1024 * 1024) {
1230 if (new_size
> device
->bdev
->bd_inode
->i_size
) {
1235 do_div(new_size
, root
->sectorsize
);
1236 new_size
*= root
->sectorsize
;
1238 printk(KERN_INFO
"new size for %s is %llu\n",
1239 device
->name
, (unsigned long long)new_size
);
1241 if (new_size
> old_size
) {
1242 trans
= btrfs_start_transaction(root
, 0);
1243 if (IS_ERR(trans
)) {
1244 ret
= PTR_ERR(trans
);
1247 ret
= btrfs_grow_device(trans
, device
, new_size
);
1248 btrfs_commit_transaction(trans
, root
);
1250 ret
= btrfs_shrink_device(device
, new_size
);
1254 mutex_unlock(&root
->fs_info
->volume_mutex
);
1259 static noinline
int btrfs_ioctl_snap_create_transid(struct file
*file
,
1266 struct btrfs_root
*root
= BTRFS_I(fdentry(file
)->d_inode
)->root
;
1267 struct file
*src_file
;
1271 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
)
1274 namelen
= strlen(name
);
1275 if (strchr(name
, '/')) {
1281 ret
= btrfs_mksubvol(&file
->f_path
, name
, namelen
,
1282 NULL
, transid
, readonly
);
1284 struct inode
*src_inode
;
1285 src_file
= fget(fd
);
1291 src_inode
= src_file
->f_path
.dentry
->d_inode
;
1292 if (src_inode
->i_sb
!= file
->f_path
.dentry
->d_inode
->i_sb
) {
1293 printk(KERN_INFO
"btrfs: Snapshot src from "
1299 ret
= btrfs_mksubvol(&file
->f_path
, name
, namelen
,
1300 BTRFS_I(src_inode
)->root
,
1308 static noinline
int btrfs_ioctl_snap_create(struct file
*file
,
1309 void __user
*arg
, int subvol
)
1311 struct btrfs_ioctl_vol_args
*vol_args
;
1314 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1315 if (IS_ERR(vol_args
))
1316 return PTR_ERR(vol_args
);
1317 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
1319 ret
= btrfs_ioctl_snap_create_transid(file
, vol_args
->name
,
1320 vol_args
->fd
, subvol
,
1327 static noinline
int btrfs_ioctl_snap_create_v2(struct file
*file
,
1328 void __user
*arg
, int subvol
)
1330 struct btrfs_ioctl_vol_args_v2
*vol_args
;
1334 bool readonly
= false;
1336 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1337 if (IS_ERR(vol_args
))
1338 return PTR_ERR(vol_args
);
1339 vol_args
->name
[BTRFS_SUBVOL_NAME_MAX
] = '\0';
1341 if (vol_args
->flags
&
1342 ~(BTRFS_SUBVOL_CREATE_ASYNC
| BTRFS_SUBVOL_RDONLY
)) {
1347 if (vol_args
->flags
& BTRFS_SUBVOL_CREATE_ASYNC
)
1349 if (vol_args
->flags
& BTRFS_SUBVOL_RDONLY
)
1352 ret
= btrfs_ioctl_snap_create_transid(file
, vol_args
->name
,
1353 vol_args
->fd
, subvol
,
1356 if (ret
== 0 && ptr
&&
1358 offsetof(struct btrfs_ioctl_vol_args_v2
,
1359 transid
), ptr
, sizeof(*ptr
)))
1366 static noinline
int btrfs_ioctl_subvol_getflags(struct file
*file
,
1369 struct inode
*inode
= fdentry(file
)->d_inode
;
1370 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
1374 if (btrfs_ino(inode
) != BTRFS_FIRST_FREE_OBJECTID
)
1377 down_read(&root
->fs_info
->subvol_sem
);
1378 if (btrfs_root_readonly(root
))
1379 flags
|= BTRFS_SUBVOL_RDONLY
;
1380 up_read(&root
->fs_info
->subvol_sem
);
1382 if (copy_to_user(arg
, &flags
, sizeof(flags
)))
1388 static noinline
int btrfs_ioctl_subvol_setflags(struct file
*file
,
1391 struct inode
*inode
= fdentry(file
)->d_inode
;
1392 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
1393 struct btrfs_trans_handle
*trans
;
1398 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
)
1401 if (btrfs_ino(inode
) != BTRFS_FIRST_FREE_OBJECTID
)
1404 if (copy_from_user(&flags
, arg
, sizeof(flags
)))
1407 if (flags
& BTRFS_SUBVOL_CREATE_ASYNC
)
1410 if (flags
& ~BTRFS_SUBVOL_RDONLY
)
1413 if (!inode_owner_or_capable(inode
))
1416 down_write(&root
->fs_info
->subvol_sem
);
1419 if (!!(flags
& BTRFS_SUBVOL_RDONLY
) == btrfs_root_readonly(root
))
1422 root_flags
= btrfs_root_flags(&root
->root_item
);
1423 if (flags
& BTRFS_SUBVOL_RDONLY
)
1424 btrfs_set_root_flags(&root
->root_item
,
1425 root_flags
| BTRFS_ROOT_SUBVOL_RDONLY
);
1427 btrfs_set_root_flags(&root
->root_item
,
1428 root_flags
& ~BTRFS_ROOT_SUBVOL_RDONLY
);
1430 trans
= btrfs_start_transaction(root
, 1);
1431 if (IS_ERR(trans
)) {
1432 ret
= PTR_ERR(trans
);
1436 ret
= btrfs_update_root(trans
, root
->fs_info
->tree_root
,
1437 &root
->root_key
, &root
->root_item
);
1439 btrfs_commit_transaction(trans
, root
);
1442 btrfs_set_root_flags(&root
->root_item
, root_flags
);
1444 up_write(&root
->fs_info
->subvol_sem
);
1449 * helper to check if the subvolume references other subvolumes
1451 static noinline
int may_destroy_subvol(struct btrfs_root
*root
)
1453 struct btrfs_path
*path
;
1454 struct btrfs_key key
;
1457 path
= btrfs_alloc_path();
1461 key
.objectid
= root
->root_key
.objectid
;
1462 key
.type
= BTRFS_ROOT_REF_KEY
;
1463 key
.offset
= (u64
)-1;
1465 ret
= btrfs_search_slot(NULL
, root
->fs_info
->tree_root
,
1472 if (path
->slots
[0] > 0) {
1474 btrfs_item_key_to_cpu(path
->nodes
[0], &key
, path
->slots
[0]);
1475 if (key
.objectid
== root
->root_key
.objectid
&&
1476 key
.type
== BTRFS_ROOT_REF_KEY
)
1480 btrfs_free_path(path
);
1484 static noinline
int key_in_sk(struct btrfs_key
*key
,
1485 struct btrfs_ioctl_search_key
*sk
)
1487 struct btrfs_key test
;
1490 test
.objectid
= sk
->min_objectid
;
1491 test
.type
= sk
->min_type
;
1492 test
.offset
= sk
->min_offset
;
1494 ret
= btrfs_comp_cpu_keys(key
, &test
);
1498 test
.objectid
= sk
->max_objectid
;
1499 test
.type
= sk
->max_type
;
1500 test
.offset
= sk
->max_offset
;
1502 ret
= btrfs_comp_cpu_keys(key
, &test
);
1508 static noinline
int copy_to_sk(struct btrfs_root
*root
,
1509 struct btrfs_path
*path
,
1510 struct btrfs_key
*key
,
1511 struct btrfs_ioctl_search_key
*sk
,
1513 unsigned long *sk_offset
,
1517 struct extent_buffer
*leaf
;
1518 struct btrfs_ioctl_search_header sh
;
1519 unsigned long item_off
;
1520 unsigned long item_len
;
1526 leaf
= path
->nodes
[0];
1527 slot
= path
->slots
[0];
1528 nritems
= btrfs_header_nritems(leaf
);
1530 if (btrfs_header_generation(leaf
) > sk
->max_transid
) {
1534 found_transid
= btrfs_header_generation(leaf
);
1536 for (i
= slot
; i
< nritems
; i
++) {
1537 item_off
= btrfs_item_ptr_offset(leaf
, i
);
1538 item_len
= btrfs_item_size_nr(leaf
, i
);
1540 if (item_len
> BTRFS_SEARCH_ARGS_BUFSIZE
)
1543 if (sizeof(sh
) + item_len
+ *sk_offset
>
1544 BTRFS_SEARCH_ARGS_BUFSIZE
) {
1549 btrfs_item_key_to_cpu(leaf
, key
, i
);
1550 if (!key_in_sk(key
, sk
))
1553 sh
.objectid
= key
->objectid
;
1554 sh
.offset
= key
->offset
;
1555 sh
.type
= key
->type
;
1557 sh
.transid
= found_transid
;
1559 /* copy search result header */
1560 memcpy(buf
+ *sk_offset
, &sh
, sizeof(sh
));
1561 *sk_offset
+= sizeof(sh
);
1564 char *p
= buf
+ *sk_offset
;
1566 read_extent_buffer(leaf
, p
,
1567 item_off
, item_len
);
1568 *sk_offset
+= item_len
;
1572 if (*num_found
>= sk
->nr_items
)
1577 if (key
->offset
< (u64
)-1 && key
->offset
< sk
->max_offset
)
1579 else if (key
->type
< (u8
)-1 && key
->type
< sk
->max_type
) {
1582 } else if (key
->objectid
< (u64
)-1 && key
->objectid
< sk
->max_objectid
) {
1592 static noinline
int search_ioctl(struct inode
*inode
,
1593 struct btrfs_ioctl_search_args
*args
)
1595 struct btrfs_root
*root
;
1596 struct btrfs_key key
;
1597 struct btrfs_key max_key
;
1598 struct btrfs_path
*path
;
1599 struct btrfs_ioctl_search_key
*sk
= &args
->key
;
1600 struct btrfs_fs_info
*info
= BTRFS_I(inode
)->root
->fs_info
;
1603 unsigned long sk_offset
= 0;
1605 path
= btrfs_alloc_path();
1609 if (sk
->tree_id
== 0) {
1610 /* search the root of the inode that was passed */
1611 root
= BTRFS_I(inode
)->root
;
1613 key
.objectid
= sk
->tree_id
;
1614 key
.type
= BTRFS_ROOT_ITEM_KEY
;
1615 key
.offset
= (u64
)-1;
1616 root
= btrfs_read_fs_root_no_name(info
, &key
);
1618 printk(KERN_ERR
"could not find root %llu\n",
1620 btrfs_free_path(path
);
1625 key
.objectid
= sk
->min_objectid
;
1626 key
.type
= sk
->min_type
;
1627 key
.offset
= sk
->min_offset
;
1629 max_key
.objectid
= sk
->max_objectid
;
1630 max_key
.type
= sk
->max_type
;
1631 max_key
.offset
= sk
->max_offset
;
1633 path
->keep_locks
= 1;
1636 ret
= btrfs_search_forward(root
, &key
, &max_key
, path
, 0,
1643 ret
= copy_to_sk(root
, path
, &key
, sk
, args
->buf
,
1644 &sk_offset
, &num_found
);
1645 btrfs_release_path(path
);
1646 if (ret
|| num_found
>= sk
->nr_items
)
1652 sk
->nr_items
= num_found
;
1653 btrfs_free_path(path
);
1657 static noinline
int btrfs_ioctl_tree_search(struct file
*file
,
1660 struct btrfs_ioctl_search_args
*args
;
1661 struct inode
*inode
;
1664 if (!capable(CAP_SYS_ADMIN
))
1667 args
= memdup_user(argp
, sizeof(*args
));
1669 return PTR_ERR(args
);
1671 inode
= fdentry(file
)->d_inode
;
1672 ret
= search_ioctl(inode
, args
);
1673 if (ret
== 0 && copy_to_user(argp
, args
, sizeof(*args
)))
1680 * Search INODE_REFs to identify path name of 'dirid' directory
1681 * in a 'tree_id' tree. and sets path name to 'name'.
1683 static noinline
int btrfs_search_path_in_tree(struct btrfs_fs_info
*info
,
1684 u64 tree_id
, u64 dirid
, char *name
)
1686 struct btrfs_root
*root
;
1687 struct btrfs_key key
;
1693 struct btrfs_inode_ref
*iref
;
1694 struct extent_buffer
*l
;
1695 struct btrfs_path
*path
;
1697 if (dirid
== BTRFS_FIRST_FREE_OBJECTID
) {
1702 path
= btrfs_alloc_path();
1706 ptr
= &name
[BTRFS_INO_LOOKUP_PATH_MAX
];
1708 key
.objectid
= tree_id
;
1709 key
.type
= BTRFS_ROOT_ITEM_KEY
;
1710 key
.offset
= (u64
)-1;
1711 root
= btrfs_read_fs_root_no_name(info
, &key
);
1713 printk(KERN_ERR
"could not find root %llu\n", tree_id
);
1718 key
.objectid
= dirid
;
1719 key
.type
= BTRFS_INODE_REF_KEY
;
1720 key
.offset
= (u64
)-1;
1723 ret
= btrfs_search_slot(NULL
, root
, &key
, path
, 0, 0);
1728 slot
= path
->slots
[0];
1729 if (ret
> 0 && slot
> 0)
1731 btrfs_item_key_to_cpu(l
, &key
, slot
);
1733 if (ret
> 0 && (key
.objectid
!= dirid
||
1734 key
.type
!= BTRFS_INODE_REF_KEY
)) {
1739 iref
= btrfs_item_ptr(l
, slot
, struct btrfs_inode_ref
);
1740 len
= btrfs_inode_ref_name_len(l
, iref
);
1742 total_len
+= len
+ 1;
1747 read_extent_buffer(l
, ptr
,(unsigned long)(iref
+ 1), len
);
1749 if (key
.offset
== BTRFS_FIRST_FREE_OBJECTID
)
1752 btrfs_release_path(path
);
1753 key
.objectid
= key
.offset
;
1754 key
.offset
= (u64
)-1;
1755 dirid
= key
.objectid
;
1760 memcpy(name
, ptr
, total_len
);
1761 name
[total_len
]='\0';
1764 btrfs_free_path(path
);
1768 static noinline
int btrfs_ioctl_ino_lookup(struct file
*file
,
1771 struct btrfs_ioctl_ino_lookup_args
*args
;
1772 struct inode
*inode
;
1775 if (!capable(CAP_SYS_ADMIN
))
1778 args
= memdup_user(argp
, sizeof(*args
));
1780 return PTR_ERR(args
);
1782 inode
= fdentry(file
)->d_inode
;
1784 if (args
->treeid
== 0)
1785 args
->treeid
= BTRFS_I(inode
)->root
->root_key
.objectid
;
1787 ret
= btrfs_search_path_in_tree(BTRFS_I(inode
)->root
->fs_info
,
1788 args
->treeid
, args
->objectid
,
1791 if (ret
== 0 && copy_to_user(argp
, args
, sizeof(*args
)))
1798 static noinline
int btrfs_ioctl_snap_destroy(struct file
*file
,
1801 struct dentry
*parent
= fdentry(file
);
1802 struct dentry
*dentry
;
1803 struct inode
*dir
= parent
->d_inode
;
1804 struct inode
*inode
;
1805 struct btrfs_root
*root
= BTRFS_I(dir
)->root
;
1806 struct btrfs_root
*dest
= NULL
;
1807 struct btrfs_ioctl_vol_args
*vol_args
;
1808 struct btrfs_trans_handle
*trans
;
1813 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
1814 if (IS_ERR(vol_args
))
1815 return PTR_ERR(vol_args
);
1817 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
1818 namelen
= strlen(vol_args
->name
);
1819 if (strchr(vol_args
->name
, '/') ||
1820 strncmp(vol_args
->name
, "..", namelen
) == 0) {
1825 err
= mnt_want_write(file
->f_path
.mnt
);
1829 mutex_lock_nested(&dir
->i_mutex
, I_MUTEX_PARENT
);
1830 dentry
= lookup_one_len(vol_args
->name
, parent
, namelen
);
1831 if (IS_ERR(dentry
)) {
1832 err
= PTR_ERR(dentry
);
1833 goto out_unlock_dir
;
1836 if (!dentry
->d_inode
) {
1841 inode
= dentry
->d_inode
;
1842 dest
= BTRFS_I(inode
)->root
;
1843 if (!capable(CAP_SYS_ADMIN
)){
1845 * Regular user. Only allow this with a special mount
1846 * option, when the user has write+exec access to the
1847 * subvol root, and when rmdir(2) would have been
1850 * Note that this is _not_ check that the subvol is
1851 * empty or doesn't contain data that we wouldn't
1852 * otherwise be able to delete.
1854 * Users who want to delete empty subvols should try
1858 if (!btrfs_test_opt(root
, USER_SUBVOL_RM_ALLOWED
))
1862 * Do not allow deletion if the parent dir is the same
1863 * as the dir to be deleted. That means the ioctl
1864 * must be called on the dentry referencing the root
1865 * of the subvol, not a random directory contained
1872 err
= inode_permission(inode
, MAY_WRITE
| MAY_EXEC
);
1876 /* check if subvolume may be deleted by a non-root user */
1877 err
= btrfs_may_delete(dir
, dentry
, 1);
1882 if (btrfs_ino(inode
) != BTRFS_FIRST_FREE_OBJECTID
) {
1887 mutex_lock(&inode
->i_mutex
);
1888 err
= d_invalidate(dentry
);
1892 down_write(&root
->fs_info
->subvol_sem
);
1894 err
= may_destroy_subvol(dest
);
1898 trans
= btrfs_start_transaction(root
, 0);
1899 if (IS_ERR(trans
)) {
1900 err
= PTR_ERR(trans
);
1903 trans
->block_rsv
= &root
->fs_info
->global_block_rsv
;
1905 ret
= btrfs_unlink_subvol(trans
, root
, dir
,
1906 dest
->root_key
.objectid
,
1907 dentry
->d_name
.name
,
1908 dentry
->d_name
.len
);
1911 btrfs_record_root_in_trans(trans
, dest
);
1913 memset(&dest
->root_item
.drop_progress
, 0,
1914 sizeof(dest
->root_item
.drop_progress
));
1915 dest
->root_item
.drop_level
= 0;
1916 btrfs_set_root_refs(&dest
->root_item
, 0);
1918 if (!xchg(&dest
->orphan_item_inserted
, 1)) {
1919 ret
= btrfs_insert_orphan_item(trans
,
1920 root
->fs_info
->tree_root
,
1921 dest
->root_key
.objectid
);
1925 ret
= btrfs_end_transaction(trans
, root
);
1927 inode
->i_flags
|= S_DEAD
;
1929 up_write(&root
->fs_info
->subvol_sem
);
1931 mutex_unlock(&inode
->i_mutex
);
1933 shrink_dcache_sb(root
->fs_info
->sb
);
1934 btrfs_invalidate_inodes(dest
);
1940 mutex_unlock(&dir
->i_mutex
);
1941 mnt_drop_write(file
->f_path
.mnt
);
1947 static int btrfs_ioctl_defrag(struct file
*file
, void __user
*argp
)
1949 struct inode
*inode
= fdentry(file
)->d_inode
;
1950 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
1951 struct btrfs_ioctl_defrag_range_args
*range
;
1954 if (btrfs_root_readonly(root
))
1957 ret
= mnt_want_write(file
->f_path
.mnt
);
1961 switch (inode
->i_mode
& S_IFMT
) {
1963 if (!capable(CAP_SYS_ADMIN
)) {
1967 ret
= btrfs_defrag_root(root
, 0);
1970 ret
= btrfs_defrag_root(root
->fs_info
->extent_root
, 0);
1973 if (!(file
->f_mode
& FMODE_WRITE
)) {
1978 range
= kzalloc(sizeof(*range
), GFP_KERNEL
);
1985 if (copy_from_user(range
, argp
,
1991 /* compression requires us to start the IO */
1992 if ((range
->flags
& BTRFS_DEFRAG_RANGE_COMPRESS
)) {
1993 range
->flags
|= BTRFS_DEFRAG_RANGE_START_IO
;
1994 range
->extent_thresh
= (u32
)-1;
1997 /* the rest are all set to zero by kzalloc */
1998 range
->len
= (u64
)-1;
2000 ret
= btrfs_defrag_file(fdentry(file
)->d_inode
, file
,
2010 mnt_drop_write(file
->f_path
.mnt
);
2014 static long btrfs_ioctl_add_dev(struct btrfs_root
*root
, void __user
*arg
)
2016 struct btrfs_ioctl_vol_args
*vol_args
;
2019 if (!capable(CAP_SYS_ADMIN
))
2022 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
2023 if (IS_ERR(vol_args
))
2024 return PTR_ERR(vol_args
);
2026 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
2027 ret
= btrfs_init_new_device(root
, vol_args
->name
);
2033 static long btrfs_ioctl_rm_dev(struct btrfs_root
*root
, void __user
*arg
)
2035 struct btrfs_ioctl_vol_args
*vol_args
;
2038 if (!capable(CAP_SYS_ADMIN
))
2041 if (root
->fs_info
->sb
->s_flags
& MS_RDONLY
)
2044 vol_args
= memdup_user(arg
, sizeof(*vol_args
));
2045 if (IS_ERR(vol_args
))
2046 return PTR_ERR(vol_args
);
2048 vol_args
->name
[BTRFS_PATH_NAME_MAX
] = '\0';
2049 ret
= btrfs_rm_device(root
, vol_args
->name
);
2055 static long btrfs_ioctl_fs_info(struct btrfs_root
*root
, void __user
*arg
)
2057 struct btrfs_ioctl_fs_info_args fi_args
;
2058 struct btrfs_device
*device
;
2059 struct btrfs_device
*next
;
2060 struct btrfs_fs_devices
*fs_devices
= root
->fs_info
->fs_devices
;
2062 if (!capable(CAP_SYS_ADMIN
))
2065 fi_args
.num_devices
= fs_devices
->num_devices
;
2067 memcpy(&fi_args
.fsid
, root
->fs_info
->fsid
, sizeof(fi_args
.fsid
));
2069 mutex_lock(&fs_devices
->device_list_mutex
);
2070 list_for_each_entry_safe(device
, next
, &fs_devices
->devices
, dev_list
) {
2071 if (device
->devid
> fi_args
.max_id
)
2072 fi_args
.max_id
= device
->devid
;
2074 mutex_unlock(&fs_devices
->device_list_mutex
);
2076 if (copy_to_user(arg
, &fi_args
, sizeof(fi_args
)))
2082 static long btrfs_ioctl_dev_info(struct btrfs_root
*root
, void __user
*arg
)
2084 struct btrfs_ioctl_dev_info_args
*di_args
;
2085 struct btrfs_device
*dev
;
2086 struct btrfs_fs_devices
*fs_devices
= root
->fs_info
->fs_devices
;
2088 char *s_uuid
= NULL
;
2089 char empty_uuid
[BTRFS_UUID_SIZE
] = {0};
2091 if (!capable(CAP_SYS_ADMIN
))
2094 di_args
= memdup_user(arg
, sizeof(*di_args
));
2095 if (IS_ERR(di_args
))
2096 return PTR_ERR(di_args
);
2098 if (memcmp(empty_uuid
, di_args
->uuid
, BTRFS_UUID_SIZE
) != 0)
2099 s_uuid
= di_args
->uuid
;
2101 mutex_lock(&fs_devices
->device_list_mutex
);
2102 dev
= btrfs_find_device(root
, di_args
->devid
, s_uuid
, NULL
);
2103 mutex_unlock(&fs_devices
->device_list_mutex
);
2110 di_args
->devid
= dev
->devid
;
2111 di_args
->bytes_used
= dev
->bytes_used
;
2112 di_args
->total_bytes
= dev
->total_bytes
;
2113 memcpy(di_args
->uuid
, dev
->uuid
, sizeof(di_args
->uuid
));
2114 strncpy(di_args
->path
, dev
->name
, sizeof(di_args
->path
));
2117 if (ret
== 0 && copy_to_user(arg
, di_args
, sizeof(*di_args
)))
2124 static noinline
long btrfs_ioctl_clone(struct file
*file
, unsigned long srcfd
,
2125 u64 off
, u64 olen
, u64 destoff
)
2127 struct inode
*inode
= fdentry(file
)->d_inode
;
2128 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
2129 struct file
*src_file
;
2131 struct btrfs_trans_handle
*trans
;
2132 struct btrfs_path
*path
;
2133 struct extent_buffer
*leaf
;
2135 struct btrfs_key key
;
2140 u64 bs
= root
->fs_info
->sb
->s_blocksize
;
2145 * - split compressed inline extents. annoying: we need to
2146 * decompress into destination's address_space (the file offset
2147 * may change, so source mapping won't do), then recompress (or
2148 * otherwise reinsert) a subrange.
2149 * - allow ranges within the same file to be cloned (provided
2150 * they don't overlap)?
2153 /* the destination must be opened for writing */
2154 if (!(file
->f_mode
& FMODE_WRITE
) || (file
->f_flags
& O_APPEND
))
2157 if (btrfs_root_readonly(root
))
2160 ret
= mnt_want_write(file
->f_path
.mnt
);
2164 src_file
= fget(srcfd
);
2167 goto out_drop_write
;
2170 src
= src_file
->f_dentry
->d_inode
;
2176 /* the src must be open for reading */
2177 if (!(src_file
->f_mode
& FMODE_READ
))
2181 if (S_ISDIR(src
->i_mode
) || S_ISDIR(inode
->i_mode
))
2185 if (src
->i_sb
!= inode
->i_sb
|| BTRFS_I(src
)->root
!= root
)
2189 buf
= vmalloc(btrfs_level_size(root
, 0));
2193 path
= btrfs_alloc_path();
2201 mutex_lock_nested(&inode
->i_mutex
, I_MUTEX_PARENT
);
2202 mutex_lock_nested(&src
->i_mutex
, I_MUTEX_CHILD
);
2204 mutex_lock_nested(&src
->i_mutex
, I_MUTEX_PARENT
);
2205 mutex_lock_nested(&inode
->i_mutex
, I_MUTEX_CHILD
);
2208 /* determine range to clone */
2210 if (off
+ len
> src
->i_size
|| off
+ len
< off
)
2213 olen
= len
= src
->i_size
- off
;
2214 /* if we extend to eof, continue to block boundary */
2215 if (off
+ len
== src
->i_size
)
2216 len
= ALIGN(src
->i_size
, bs
) - off
;
2218 /* verify the end result is block aligned */
2219 if (!IS_ALIGNED(off
, bs
) || !IS_ALIGNED(off
+ len
, bs
) ||
2220 !IS_ALIGNED(destoff
, bs
))
2223 /* do any pending delalloc/csum calc on src, one way or
2224 another, and lock file content */
2226 struct btrfs_ordered_extent
*ordered
;
2227 lock_extent(&BTRFS_I(src
)->io_tree
, off
, off
+len
, GFP_NOFS
);
2228 ordered
= btrfs_lookup_first_ordered_extent(src
, off
+len
);
2230 !test_range_bit(&BTRFS_I(src
)->io_tree
, off
, off
+len
,
2231 EXTENT_DELALLOC
, 0, NULL
))
2233 unlock_extent(&BTRFS_I(src
)->io_tree
, off
, off
+len
, GFP_NOFS
);
2235 btrfs_put_ordered_extent(ordered
);
2236 btrfs_wait_ordered_range(src
, off
, len
);
2240 key
.objectid
= btrfs_ino(src
);
2241 key
.type
= BTRFS_EXTENT_DATA_KEY
;
2246 * note the key will change type as we walk through the
2249 ret
= btrfs_search_slot(NULL
, root
, &key
, path
, 0, 0);
2253 nritems
= btrfs_header_nritems(path
->nodes
[0]);
2254 if (path
->slots
[0] >= nritems
) {
2255 ret
= btrfs_next_leaf(root
, path
);
2260 nritems
= btrfs_header_nritems(path
->nodes
[0]);
2262 leaf
= path
->nodes
[0];
2263 slot
= path
->slots
[0];
2265 btrfs_item_key_to_cpu(leaf
, &key
, slot
);
2266 if (btrfs_key_type(&key
) > BTRFS_EXTENT_DATA_KEY
||
2267 key
.objectid
!= btrfs_ino(src
))
2270 if (btrfs_key_type(&key
) == BTRFS_EXTENT_DATA_KEY
) {
2271 struct btrfs_file_extent_item
*extent
;
2274 struct btrfs_key new_key
;
2275 u64 disko
= 0, diskl
= 0;
2276 u64 datao
= 0, datal
= 0;
2280 size
= btrfs_item_size_nr(leaf
, slot
);
2281 read_extent_buffer(leaf
, buf
,
2282 btrfs_item_ptr_offset(leaf
, slot
),
2285 extent
= btrfs_item_ptr(leaf
, slot
,
2286 struct btrfs_file_extent_item
);
2287 comp
= btrfs_file_extent_compression(leaf
, extent
);
2288 type
= btrfs_file_extent_type(leaf
, extent
);
2289 if (type
== BTRFS_FILE_EXTENT_REG
||
2290 type
== BTRFS_FILE_EXTENT_PREALLOC
) {
2291 disko
= btrfs_file_extent_disk_bytenr(leaf
,
2293 diskl
= btrfs_file_extent_disk_num_bytes(leaf
,
2295 datao
= btrfs_file_extent_offset(leaf
, extent
);
2296 datal
= btrfs_file_extent_num_bytes(leaf
,
2298 } else if (type
== BTRFS_FILE_EXTENT_INLINE
) {
2299 /* take upper bound, may be compressed */
2300 datal
= btrfs_file_extent_ram_bytes(leaf
,
2303 btrfs_release_path(path
);
2305 if (key
.offset
+ datal
<= off
||
2306 key
.offset
>= off
+len
)
2309 memcpy(&new_key
, &key
, sizeof(new_key
));
2310 new_key
.objectid
= btrfs_ino(inode
);
2311 if (off
<= key
.offset
)
2312 new_key
.offset
= key
.offset
+ destoff
- off
;
2314 new_key
.offset
= destoff
;
2316 trans
= btrfs_start_transaction(root
, 1);
2317 if (IS_ERR(trans
)) {
2318 ret
= PTR_ERR(trans
);
2322 if (type
== BTRFS_FILE_EXTENT_REG
||
2323 type
== BTRFS_FILE_EXTENT_PREALLOC
) {
2324 if (off
> key
.offset
) {
2325 datao
+= off
- key
.offset
;
2326 datal
-= off
- key
.offset
;
2329 if (key
.offset
+ datal
> off
+ len
)
2330 datal
= off
+ len
- key
.offset
;
2332 ret
= btrfs_drop_extents(trans
, inode
,
2334 new_key
.offset
+ datal
,
2338 ret
= btrfs_insert_empty_item(trans
, root
, path
,
2342 leaf
= path
->nodes
[0];
2343 slot
= path
->slots
[0];
2344 write_extent_buffer(leaf
, buf
,
2345 btrfs_item_ptr_offset(leaf
, slot
),
2348 extent
= btrfs_item_ptr(leaf
, slot
,
2349 struct btrfs_file_extent_item
);
2351 /* disko == 0 means it's a hole */
2355 btrfs_set_file_extent_offset(leaf
, extent
,
2357 btrfs_set_file_extent_num_bytes(leaf
, extent
,
2360 inode_add_bytes(inode
, datal
);
2361 ret
= btrfs_inc_extent_ref(trans
, root
,
2363 root
->root_key
.objectid
,
2365 new_key
.offset
- datao
);
2368 } else if (type
== BTRFS_FILE_EXTENT_INLINE
) {
2371 if (off
> key
.offset
) {
2372 skip
= off
- key
.offset
;
2373 new_key
.offset
+= skip
;
2376 if (key
.offset
+ datal
> off
+len
)
2377 trim
= key
.offset
+ datal
- (off
+len
);
2379 if (comp
&& (skip
|| trim
)) {
2381 btrfs_end_transaction(trans
, root
);
2384 size
-= skip
+ trim
;
2385 datal
-= skip
+ trim
;
2387 ret
= btrfs_drop_extents(trans
, inode
,
2389 new_key
.offset
+ datal
,
2393 ret
= btrfs_insert_empty_item(trans
, root
, path
,
2399 btrfs_file_extent_calc_inline_size(0);
2400 memmove(buf
+start
, buf
+start
+skip
,
2404 leaf
= path
->nodes
[0];
2405 slot
= path
->slots
[0];
2406 write_extent_buffer(leaf
, buf
,
2407 btrfs_item_ptr_offset(leaf
, slot
),
2409 inode_add_bytes(inode
, datal
);
2412 btrfs_mark_buffer_dirty(leaf
);
2413 btrfs_release_path(path
);
2415 inode
->i_mtime
= inode
->i_ctime
= CURRENT_TIME
;
2418 * we round up to the block size at eof when
2419 * determining which extents to clone above,
2420 * but shouldn't round up the file size
2422 endoff
= new_key
.offset
+ datal
;
2423 if (endoff
> destoff
+olen
)
2424 endoff
= destoff
+olen
;
2425 if (endoff
> inode
->i_size
)
2426 btrfs_i_size_write(inode
, endoff
);
2428 BTRFS_I(inode
)->flags
= BTRFS_I(src
)->flags
;
2429 ret
= btrfs_update_inode(trans
, root
, inode
);
2431 btrfs_end_transaction(trans
, root
);
2434 btrfs_release_path(path
);
2439 btrfs_release_path(path
);
2440 unlock_extent(&BTRFS_I(src
)->io_tree
, off
, off
+len
, GFP_NOFS
);
2442 mutex_unlock(&src
->i_mutex
);
2443 mutex_unlock(&inode
->i_mutex
);
2445 btrfs_free_path(path
);
2449 mnt_drop_write(file
->f_path
.mnt
);
2453 static long btrfs_ioctl_clone_range(struct file
*file
, void __user
*argp
)
2455 struct btrfs_ioctl_clone_range_args args
;
2457 if (copy_from_user(&args
, argp
, sizeof(args
)))
2459 return btrfs_ioctl_clone(file
, args
.src_fd
, args
.src_offset
,
2460 args
.src_length
, args
.dest_offset
);
2464 * there are many ways the trans_start and trans_end ioctls can lead
2465 * to deadlocks. They should only be used by applications that
2466 * basically own the machine, and have a very in depth understanding
2467 * of all the possible deadlocks and enospc problems.
2469 static long btrfs_ioctl_trans_start(struct file
*file
)
2471 struct inode
*inode
= fdentry(file
)->d_inode
;
2472 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
2473 struct btrfs_trans_handle
*trans
;
2477 if (!capable(CAP_SYS_ADMIN
))
2481 if (file
->private_data
)
2485 if (btrfs_root_readonly(root
))
2488 ret
= mnt_want_write(file
->f_path
.mnt
);
2492 atomic_inc(&root
->fs_info
->open_ioctl_trans
);
2495 trans
= btrfs_start_ioctl_transaction(root
);
2499 file
->private_data
= trans
;
2503 atomic_dec(&root
->fs_info
->open_ioctl_trans
);
2504 mnt_drop_write(file
->f_path
.mnt
);
2509 static long btrfs_ioctl_default_subvol(struct file
*file
, void __user
*argp
)
2511 struct inode
*inode
= fdentry(file
)->d_inode
;
2512 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
2513 struct btrfs_root
*new_root
;
2514 struct btrfs_dir_item
*di
;
2515 struct btrfs_trans_handle
*trans
;
2516 struct btrfs_path
*path
;
2517 struct btrfs_key location
;
2518 struct btrfs_disk_key disk_key
;
2519 struct btrfs_super_block
*disk_super
;
2524 if (!capable(CAP_SYS_ADMIN
))
2527 if (copy_from_user(&objectid
, argp
, sizeof(objectid
)))
2531 objectid
= root
->root_key
.objectid
;
2533 location
.objectid
= objectid
;
2534 location
.type
= BTRFS_ROOT_ITEM_KEY
;
2535 location
.offset
= (u64
)-1;
2537 new_root
= btrfs_read_fs_root_no_name(root
->fs_info
, &location
);
2538 if (IS_ERR(new_root
))
2539 return PTR_ERR(new_root
);
2541 if (btrfs_root_refs(&new_root
->root_item
) == 0)
2544 path
= btrfs_alloc_path();
2547 path
->leave_spinning
= 1;
2549 trans
= btrfs_start_transaction(root
, 1);
2550 if (IS_ERR(trans
)) {
2551 btrfs_free_path(path
);
2552 return PTR_ERR(trans
);
2555 dir_id
= btrfs_super_root_dir(&root
->fs_info
->super_copy
);
2556 di
= btrfs_lookup_dir_item(trans
, root
->fs_info
->tree_root
, path
,
2557 dir_id
, "default", 7, 1);
2558 if (IS_ERR_OR_NULL(di
)) {
2559 btrfs_free_path(path
);
2560 btrfs_end_transaction(trans
, root
);
2561 printk(KERN_ERR
"Umm, you don't have the default dir item, "
2562 "this isn't going to work\n");
2566 btrfs_cpu_key_to_disk(&disk_key
, &new_root
->root_key
);
2567 btrfs_set_dir_item_key(path
->nodes
[0], di
, &disk_key
);
2568 btrfs_mark_buffer_dirty(path
->nodes
[0]);
2569 btrfs_free_path(path
);
2571 disk_super
= &root
->fs_info
->super_copy
;
2572 features
= btrfs_super_incompat_flags(disk_super
);
2573 if (!(features
& BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL
)) {
2574 features
|= BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL
;
2575 btrfs_set_super_incompat_flags(disk_super
, features
);
2577 btrfs_end_transaction(trans
, root
);
2582 static void get_block_group_info(struct list_head
*groups_list
,
2583 struct btrfs_ioctl_space_info
*space
)
2585 struct btrfs_block_group_cache
*block_group
;
2587 space
->total_bytes
= 0;
2588 space
->used_bytes
= 0;
2590 list_for_each_entry(block_group
, groups_list
, list
) {
2591 space
->flags
= block_group
->flags
;
2592 space
->total_bytes
+= block_group
->key
.offset
;
2593 space
->used_bytes
+=
2594 btrfs_block_group_used(&block_group
->item
);
2598 long btrfs_ioctl_space_info(struct btrfs_root
*root
, void __user
*arg
)
2600 struct btrfs_ioctl_space_args space_args
;
2601 struct btrfs_ioctl_space_info space
;
2602 struct btrfs_ioctl_space_info
*dest
;
2603 struct btrfs_ioctl_space_info
*dest_orig
;
2604 struct btrfs_ioctl_space_info __user
*user_dest
;
2605 struct btrfs_space_info
*info
;
2606 u64 types
[] = {BTRFS_BLOCK_GROUP_DATA
,
2607 BTRFS_BLOCK_GROUP_SYSTEM
,
2608 BTRFS_BLOCK_GROUP_METADATA
,
2609 BTRFS_BLOCK_GROUP_DATA
| BTRFS_BLOCK_GROUP_METADATA
};
2616 if (copy_from_user(&space_args
,
2617 (struct btrfs_ioctl_space_args __user
*)arg
,
2618 sizeof(space_args
)))
2621 for (i
= 0; i
< num_types
; i
++) {
2622 struct btrfs_space_info
*tmp
;
2626 list_for_each_entry_rcu(tmp
, &root
->fs_info
->space_info
,
2628 if (tmp
->flags
== types
[i
]) {
2638 down_read(&info
->groups_sem
);
2639 for (c
= 0; c
< BTRFS_NR_RAID_TYPES
; c
++) {
2640 if (!list_empty(&info
->block_groups
[c
]))
2643 up_read(&info
->groups_sem
);
2646 /* space_slots == 0 means they are asking for a count */
2647 if (space_args
.space_slots
== 0) {
2648 space_args
.total_spaces
= slot_count
;
2652 slot_count
= min_t(u64
, space_args
.space_slots
, slot_count
);
2654 alloc_size
= sizeof(*dest
) * slot_count
;
2656 /* we generally have at most 6 or so space infos, one for each raid
2657 * level. So, a whole page should be more than enough for everyone
2659 if (alloc_size
> PAGE_CACHE_SIZE
)
2662 space_args
.total_spaces
= 0;
2663 dest
= kmalloc(alloc_size
, GFP_NOFS
);
2668 /* now we have a buffer to copy into */
2669 for (i
= 0; i
< num_types
; i
++) {
2670 struct btrfs_space_info
*tmp
;
2677 list_for_each_entry_rcu(tmp
, &root
->fs_info
->space_info
,
2679 if (tmp
->flags
== types
[i
]) {
2688 down_read(&info
->groups_sem
);
2689 for (c
= 0; c
< BTRFS_NR_RAID_TYPES
; c
++) {
2690 if (!list_empty(&info
->block_groups
[c
])) {
2691 get_block_group_info(&info
->block_groups
[c
],
2693 memcpy(dest
, &space
, sizeof(space
));
2695 space_args
.total_spaces
++;
2701 up_read(&info
->groups_sem
);
2704 user_dest
= (struct btrfs_ioctl_space_info
*)
2705 (arg
+ sizeof(struct btrfs_ioctl_space_args
));
2707 if (copy_to_user(user_dest
, dest_orig
, alloc_size
))
2712 if (ret
== 0 && copy_to_user(arg
, &space_args
, sizeof(space_args
)))
2719 * there are many ways the trans_start and trans_end ioctls can lead
2720 * to deadlocks. They should only be used by applications that
2721 * basically own the machine, and have a very in depth understanding
2722 * of all the possible deadlocks and enospc problems.
2724 long btrfs_ioctl_trans_end(struct file
*file
)
2726 struct inode
*inode
= fdentry(file
)->d_inode
;
2727 struct btrfs_root
*root
= BTRFS_I(inode
)->root
;
2728 struct btrfs_trans_handle
*trans
;
2730 trans
= file
->private_data
;
2733 file
->private_data
= NULL
;
2735 btrfs_end_transaction(trans
, root
);
2737 atomic_dec(&root
->fs_info
->open_ioctl_trans
);
2739 mnt_drop_write(file
->f_path
.mnt
);
2743 static noinline
long btrfs_ioctl_start_sync(struct file
*file
, void __user
*argp
)
2745 struct btrfs_root
*root
= BTRFS_I(file
->f_dentry
->d_inode
)->root
;
2746 struct btrfs_trans_handle
*trans
;
2750 trans
= btrfs_start_transaction(root
, 0);
2752 return PTR_ERR(trans
);
2753 transid
= trans
->transid
;
2754 ret
= btrfs_commit_transaction_async(trans
, root
, 0);
2756 btrfs_end_transaction(trans
, root
);
2761 if (copy_to_user(argp
, &transid
, sizeof(transid
)))
2766 static noinline
long btrfs_ioctl_wait_sync(struct file
*file
, void __user
*argp
)
2768 struct btrfs_root
*root
= BTRFS_I(file
->f_dentry
->d_inode
)->root
;
2772 if (copy_from_user(&transid
, argp
, sizeof(transid
)))
2775 transid
= 0; /* current trans */
2777 return btrfs_wait_for_commit(root
, transid
);
2780 static long btrfs_ioctl_scrub(struct btrfs_root
*root
, void __user
*arg
)
2783 struct btrfs_ioctl_scrub_args
*sa
;
2785 if (!capable(CAP_SYS_ADMIN
))
2788 sa
= memdup_user(arg
, sizeof(*sa
));
2792 ret
= btrfs_scrub_dev(root
, sa
->devid
, sa
->start
, sa
->end
,
2793 &sa
->progress
, sa
->flags
& BTRFS_SCRUB_READONLY
);
2795 if (copy_to_user(arg
, sa
, sizeof(*sa
)))
2802 static long btrfs_ioctl_scrub_cancel(struct btrfs_root
*root
, void __user
*arg
)
2804 if (!capable(CAP_SYS_ADMIN
))
2807 return btrfs_scrub_cancel(root
);
2810 static long btrfs_ioctl_scrub_progress(struct btrfs_root
*root
,
2813 struct btrfs_ioctl_scrub_args
*sa
;
2816 if (!capable(CAP_SYS_ADMIN
))
2819 sa
= memdup_user(arg
, sizeof(*sa
));
2823 ret
= btrfs_scrub_progress(root
, sa
->devid
, &sa
->progress
);
2825 if (copy_to_user(arg
, sa
, sizeof(*sa
)))
2832 long btrfs_ioctl(struct file
*file
, unsigned int
2833 cmd
, unsigned long arg
)
2835 struct btrfs_root
*root
= BTRFS_I(fdentry(file
)->d_inode
)->root
;
2836 void __user
*argp
= (void __user
*)arg
;
2839 case FS_IOC_GETFLAGS
:
2840 return btrfs_ioctl_getflags(file
, argp
);
2841 case FS_IOC_SETFLAGS
:
2842 return btrfs_ioctl_setflags(file
, argp
);
2843 case FS_IOC_GETVERSION
:
2844 return btrfs_ioctl_getversion(file
, argp
);
2846 return btrfs_ioctl_fitrim(file
, argp
);
2847 case BTRFS_IOC_SNAP_CREATE
:
2848 return btrfs_ioctl_snap_create(file
, argp
, 0);
2849 case BTRFS_IOC_SNAP_CREATE_V2
:
2850 return btrfs_ioctl_snap_create_v2(file
, argp
, 0);
2851 case BTRFS_IOC_SUBVOL_CREATE
:
2852 return btrfs_ioctl_snap_create(file
, argp
, 1);
2853 case BTRFS_IOC_SNAP_DESTROY
:
2854 return btrfs_ioctl_snap_destroy(file
, argp
);
2855 case BTRFS_IOC_SUBVOL_GETFLAGS
:
2856 return btrfs_ioctl_subvol_getflags(file
, argp
);
2857 case BTRFS_IOC_SUBVOL_SETFLAGS
:
2858 return btrfs_ioctl_subvol_setflags(file
, argp
);
2859 case BTRFS_IOC_DEFAULT_SUBVOL
:
2860 return btrfs_ioctl_default_subvol(file
, argp
);
2861 case BTRFS_IOC_DEFRAG
:
2862 return btrfs_ioctl_defrag(file
, NULL
);
2863 case BTRFS_IOC_DEFRAG_RANGE
:
2864 return btrfs_ioctl_defrag(file
, argp
);
2865 case BTRFS_IOC_RESIZE
:
2866 return btrfs_ioctl_resize(root
, argp
);
2867 case BTRFS_IOC_ADD_DEV
:
2868 return btrfs_ioctl_add_dev(root
, argp
);
2869 case BTRFS_IOC_RM_DEV
:
2870 return btrfs_ioctl_rm_dev(root
, argp
);
2871 case BTRFS_IOC_FS_INFO
:
2872 return btrfs_ioctl_fs_info(root
, argp
);
2873 case BTRFS_IOC_DEV_INFO
:
2874 return btrfs_ioctl_dev_info(root
, argp
);
2875 case BTRFS_IOC_BALANCE
:
2876 return btrfs_balance(root
->fs_info
->dev_root
);
2877 case BTRFS_IOC_CLONE
:
2878 return btrfs_ioctl_clone(file
, arg
, 0, 0, 0);
2879 case BTRFS_IOC_CLONE_RANGE
:
2880 return btrfs_ioctl_clone_range(file
, argp
);
2881 case BTRFS_IOC_TRANS_START
:
2882 return btrfs_ioctl_trans_start(file
);
2883 case BTRFS_IOC_TRANS_END
:
2884 return btrfs_ioctl_trans_end(file
);
2885 case BTRFS_IOC_TREE_SEARCH
:
2886 return btrfs_ioctl_tree_search(file
, argp
);
2887 case BTRFS_IOC_INO_LOOKUP
:
2888 return btrfs_ioctl_ino_lookup(file
, argp
);
2889 case BTRFS_IOC_SPACE_INFO
:
2890 return btrfs_ioctl_space_info(root
, argp
);
2891 case BTRFS_IOC_SYNC
:
2892 btrfs_sync_fs(file
->f_dentry
->d_sb
, 1);
2894 case BTRFS_IOC_START_SYNC
:
2895 return btrfs_ioctl_start_sync(file
, argp
);
2896 case BTRFS_IOC_WAIT_SYNC
:
2897 return btrfs_ioctl_wait_sync(file
, argp
);
2898 case BTRFS_IOC_SCRUB
:
2899 return btrfs_ioctl_scrub(root
, argp
);
2900 case BTRFS_IOC_SCRUB_CANCEL
:
2901 return btrfs_ioctl_scrub_cancel(root
, argp
);
2902 case BTRFS_IOC_SCRUB_PROGRESS
:
2903 return btrfs_ioctl_scrub_progress(root
, argp
);