2 * Copyright (C) 2007 Oracle. All rights reserved.
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public
6 * License v2 as published by the Free Software Foundation.
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
13 * You should have received a copy of the GNU General Public
14 * License along with this program; if not, write to the
15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
16 * Boston, MA 021110-1307, USA.
19 #include <linux/kernel.h>
20 #include <linux/bio.h>
21 #include <linux/buffer_head.h>
22 #include <linux/file.h>
24 #include <linux/fsnotify.h>
25 #include <linux/pagemap.h>
26 #include <linux/highmem.h>
27 #include <linux/time.h>
28 #include <linux/init.h>
29 #include <linux/string.h>
30 #include <linux/backing-dev.h>
31 #include <linux/mount.h>
32 #include <linux/mpage.h>
33 #include <linux/namei.h>
34 #include <linux/swap.h>
35 #include <linux/writeback.h>
36 #include <linux/statfs.h>
37 #include <linux/compat.h>
38 #include <linux/bit_spinlock.h>
39 #include <linux/security.h>
40 #include <linux/xattr.h>
41 #include <linux/vmalloc.h>
42 #include <linux/slab.h>
43 #include <linux/blkdev.h>
44 #include <linux/uuid.h>
45 #include <linux/btrfs.h>
49 #include "transaction.h"
50 #include "btrfs_inode.h"
51 #include "print-tree.h"
54 #include "inode-map.h"
56 #include "rcu-string.h"
58 #include "dev-replace.h"
60 /* Mask out flags that are inappropriate for the given type of inode. */
61 static inline __u32 btrfs_mask_flags(umode_t mode, __u32 flags)
65 else if (S_ISREG(mode))
66 return flags & ~FS_DIRSYNC_FL;
68 return flags & (FS_NODUMP_FL | FS_NOATIME_FL);
72 * Export inode flags to the format expected by the FS_IOC_GETFLAGS ioctl.
74 static unsigned int btrfs_flags_to_ioctl(unsigned int flags)
76 unsigned int iflags = 0;
78 if (flags & BTRFS_INODE_SYNC)
80 if (flags & BTRFS_INODE_IMMUTABLE)
81 iflags |= FS_IMMUTABLE_FL;
82 if (flags & BTRFS_INODE_APPEND)
83 iflags |= FS_APPEND_FL;
84 if (flags & BTRFS_INODE_NODUMP)
85 iflags |= FS_NODUMP_FL;
86 if (flags & BTRFS_INODE_NOATIME)
87 iflags |= FS_NOATIME_FL;
88 if (flags & BTRFS_INODE_DIRSYNC)
89 iflags |= FS_DIRSYNC_FL;
90 if (flags & BTRFS_INODE_NODATACOW)
91 iflags |= FS_NOCOW_FL;
93 if ((flags & BTRFS_INODE_COMPRESS) && !(flags & BTRFS_INODE_NOCOMPRESS))
94 iflags |= FS_COMPR_FL;
95 else if (flags & BTRFS_INODE_NOCOMPRESS)
96 iflags |= FS_NOCOMP_FL;
102 * Update inode->i_flags based on the btrfs internal flags.
104 void btrfs_update_iflags(struct inode *inode)
106 struct btrfs_inode *ip = BTRFS_I(inode);
108 inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
110 if (ip->flags & BTRFS_INODE_SYNC)
111 inode->i_flags |= S_SYNC;
112 if (ip->flags & BTRFS_INODE_IMMUTABLE)
113 inode->i_flags |= S_IMMUTABLE;
114 if (ip->flags & BTRFS_INODE_APPEND)
115 inode->i_flags |= S_APPEND;
116 if (ip->flags & BTRFS_INODE_NOATIME)
117 inode->i_flags |= S_NOATIME;
118 if (ip->flags & BTRFS_INODE_DIRSYNC)
119 inode->i_flags |= S_DIRSYNC;
123 * Inherit flags from the parent inode.
125 * Currently only the compression flags and the cow flags are inherited.
127 void btrfs_inherit_iflags(struct inode *inode, struct inode *dir)
134 flags = BTRFS_I(dir)->flags;
136 if (flags & BTRFS_INODE_NOCOMPRESS) {
137 BTRFS_I(inode)->flags &= ~BTRFS_INODE_COMPRESS;
138 BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS;
139 } else if (flags & BTRFS_INODE_COMPRESS) {
140 BTRFS_I(inode)->flags &= ~BTRFS_INODE_NOCOMPRESS;
141 BTRFS_I(inode)->flags |= BTRFS_INODE_COMPRESS;
144 if (flags & BTRFS_INODE_NODATACOW) {
145 BTRFS_I(inode)->flags |= BTRFS_INODE_NODATACOW;
146 if (S_ISREG(inode->i_mode))
147 BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM;
150 btrfs_update_iflags(inode);
153 static int btrfs_ioctl_getflags(struct file *file, void __user *arg)
155 struct btrfs_inode *ip = BTRFS_I(file_inode(file));
156 unsigned int flags = btrfs_flags_to_ioctl(ip->flags);
158 if (copy_to_user(arg, &flags, sizeof(flags)))
163 static int check_flags(unsigned int flags)
165 if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \
166 FS_NOATIME_FL | FS_NODUMP_FL | \
167 FS_SYNC_FL | FS_DIRSYNC_FL | \
168 FS_NOCOMP_FL | FS_COMPR_FL |
172 if ((flags & FS_NOCOMP_FL) && (flags & FS_COMPR_FL))
178 static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
180 struct inode *inode = file_inode(file);
181 struct btrfs_inode *ip = BTRFS_I(inode);
182 struct btrfs_root *root = ip->root;
183 struct btrfs_trans_handle *trans;
184 unsigned int flags, oldflags;
187 unsigned int i_oldflags;
190 if (btrfs_root_readonly(root))
193 if (copy_from_user(&flags, arg, sizeof(flags)))
196 ret = check_flags(flags);
200 if (!inode_owner_or_capable(inode))
203 ret = mnt_want_write_file(file);
207 mutex_lock(&inode->i_mutex);
209 ip_oldflags = ip->flags;
210 i_oldflags = inode->i_flags;
211 mode = inode->i_mode;
213 flags = btrfs_mask_flags(inode->i_mode, flags);
214 oldflags = btrfs_flags_to_ioctl(ip->flags);
215 if ((flags ^ oldflags) & (FS_APPEND_FL | FS_IMMUTABLE_FL)) {
216 if (!capable(CAP_LINUX_IMMUTABLE)) {
222 if (flags & FS_SYNC_FL)
223 ip->flags |= BTRFS_INODE_SYNC;
225 ip->flags &= ~BTRFS_INODE_SYNC;
226 if (flags & FS_IMMUTABLE_FL)
227 ip->flags |= BTRFS_INODE_IMMUTABLE;
229 ip->flags &= ~BTRFS_INODE_IMMUTABLE;
230 if (flags & FS_APPEND_FL)
231 ip->flags |= BTRFS_INODE_APPEND;
233 ip->flags &= ~BTRFS_INODE_APPEND;
234 if (flags & FS_NODUMP_FL)
235 ip->flags |= BTRFS_INODE_NODUMP;
237 ip->flags &= ~BTRFS_INODE_NODUMP;
238 if (flags & FS_NOATIME_FL)
239 ip->flags |= BTRFS_INODE_NOATIME;
241 ip->flags &= ~BTRFS_INODE_NOATIME;
242 if (flags & FS_DIRSYNC_FL)
243 ip->flags |= BTRFS_INODE_DIRSYNC;
245 ip->flags &= ~BTRFS_INODE_DIRSYNC;
246 if (flags & FS_NOCOW_FL) {
249 * It's safe to turn csums off here, no extents exist.
250 * Otherwise we want the flag to reflect the real COW
251 * status of the file and will not set it.
253 if (inode->i_size == 0)
254 ip->flags |= BTRFS_INODE_NODATACOW
255 | BTRFS_INODE_NODATASUM;
257 ip->flags |= BTRFS_INODE_NODATACOW;
261 * Revert back under same assuptions as above
264 if (inode->i_size == 0)
265 ip->flags &= ~(BTRFS_INODE_NODATACOW
266 | BTRFS_INODE_NODATASUM);
268 ip->flags &= ~BTRFS_INODE_NODATACOW;
273 * The COMPRESS flag can only be changed by users, while the NOCOMPRESS
274 * flag may be changed automatically if compression code won't make
277 if (flags & FS_NOCOMP_FL) {
278 ip->flags &= ~BTRFS_INODE_COMPRESS;
279 ip->flags |= BTRFS_INODE_NOCOMPRESS;
280 } else if (flags & FS_COMPR_FL) {
281 ip->flags |= BTRFS_INODE_COMPRESS;
282 ip->flags &= ~BTRFS_INODE_NOCOMPRESS;
284 ip->flags &= ~(BTRFS_INODE_COMPRESS | BTRFS_INODE_NOCOMPRESS);
287 trans = btrfs_start_transaction(root, 1);
289 ret = PTR_ERR(trans);
293 btrfs_update_iflags(inode);
294 inode_inc_iversion(inode);
295 inode->i_ctime = CURRENT_TIME;
296 ret = btrfs_update_inode(trans, root, inode);
298 btrfs_end_transaction(trans, root);
301 ip->flags = ip_oldflags;
302 inode->i_flags = i_oldflags;
306 mutex_unlock(&inode->i_mutex);
307 mnt_drop_write_file(file);
311 static int btrfs_ioctl_getversion(struct file *file, int __user *arg)
313 struct inode *inode = file_inode(file);
315 return put_user(inode->i_generation, arg);
318 static noinline int btrfs_ioctl_fitrim(struct file *file, void __user *arg)
320 struct btrfs_fs_info *fs_info = btrfs_sb(fdentry(file)->d_sb);
321 struct btrfs_device *device;
322 struct request_queue *q;
323 struct fstrim_range range;
324 u64 minlen = ULLONG_MAX;
326 u64 total_bytes = btrfs_super_total_bytes(fs_info->super_copy);
329 if (!capable(CAP_SYS_ADMIN))
333 list_for_each_entry_rcu(device, &fs_info->fs_devices->devices,
337 q = bdev_get_queue(device->bdev);
338 if (blk_queue_discard(q)) {
340 minlen = min((u64)q->limits.discard_granularity,
348 if (copy_from_user(&range, arg, sizeof(range)))
350 if (range.start > total_bytes ||
351 range.len < fs_info->sb->s_blocksize)
354 range.len = min(range.len, total_bytes - range.start);
355 range.minlen = max(range.minlen, minlen);
356 ret = btrfs_trim_fs(fs_info->tree_root, &range);
360 if (copy_to_user(arg, &range, sizeof(range)))
366 static noinline int create_subvol(struct inode *dir,
367 struct dentry *dentry,
368 char *name, int namelen,
370 struct btrfs_qgroup_inherit *inherit)
372 struct btrfs_trans_handle *trans;
373 struct btrfs_key key;
374 struct btrfs_root_item root_item;
375 struct btrfs_inode_item *inode_item;
376 struct extent_buffer *leaf;
377 struct btrfs_root *root = BTRFS_I(dir)->root;
378 struct btrfs_root *new_root;
379 struct btrfs_block_rsv block_rsv;
380 struct timespec cur_time = CURRENT_TIME;
384 u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
389 ret = btrfs_find_free_objectid(root->fs_info->tree_root, &objectid);
393 btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP);
395 * The same as the snapshot creation, please see the comment
396 * of create_snapshot().
398 ret = btrfs_subvolume_reserve_metadata(root, &block_rsv,
399 7, &qgroup_reserved);
403 trans = btrfs_start_transaction(root, 0);
405 ret = PTR_ERR(trans);
408 trans->block_rsv = &block_rsv;
409 trans->bytes_reserved = block_rsv.size;
411 ret = btrfs_qgroup_inherit(trans, root->fs_info, 0, objectid, inherit);
415 leaf = btrfs_alloc_free_block(trans, root, root->leafsize,
416 0, objectid, NULL, 0, 0, 0);
422 memset_extent_buffer(leaf, 0, 0, sizeof(struct btrfs_header));
423 btrfs_set_header_bytenr(leaf, leaf->start);
424 btrfs_set_header_generation(leaf, trans->transid);
425 btrfs_set_header_backref_rev(leaf, BTRFS_MIXED_BACKREF_REV);
426 btrfs_set_header_owner(leaf, objectid);
428 write_extent_buffer(leaf, root->fs_info->fsid,
429 (unsigned long)btrfs_header_fsid(leaf),
431 write_extent_buffer(leaf, root->fs_info->chunk_tree_uuid,
432 (unsigned long)btrfs_header_chunk_tree_uuid(leaf),
434 btrfs_mark_buffer_dirty(leaf);
436 memset(&root_item, 0, sizeof(root_item));
438 inode_item = &root_item.inode;
439 inode_item->generation = cpu_to_le64(1);
440 inode_item->size = cpu_to_le64(3);
441 inode_item->nlink = cpu_to_le32(1);
442 inode_item->nbytes = cpu_to_le64(root->leafsize);
443 inode_item->mode = cpu_to_le32(S_IFDIR | 0755);
446 root_item.byte_limit = 0;
447 inode_item->flags = cpu_to_le64(BTRFS_INODE_ROOT_ITEM_INIT);
449 btrfs_set_root_bytenr(&root_item, leaf->start);
450 btrfs_set_root_generation(&root_item, trans->transid);
451 btrfs_set_root_level(&root_item, 0);
452 btrfs_set_root_refs(&root_item, 1);
453 btrfs_set_root_used(&root_item, leaf->len);
454 btrfs_set_root_last_snapshot(&root_item, 0);
456 btrfs_set_root_generation_v2(&root_item,
457 btrfs_root_generation(&root_item));
458 uuid_le_gen(&new_uuid);
459 memcpy(root_item.uuid, new_uuid.b, BTRFS_UUID_SIZE);
460 root_item.otime.sec = cpu_to_le64(cur_time.tv_sec);
461 root_item.otime.nsec = cpu_to_le32(cur_time.tv_nsec);
462 root_item.ctime = root_item.otime;
463 btrfs_set_root_ctransid(&root_item, trans->transid);
464 btrfs_set_root_otransid(&root_item, trans->transid);
466 btrfs_tree_unlock(leaf);
467 free_extent_buffer(leaf);
470 btrfs_set_root_dirid(&root_item, new_dirid);
472 key.objectid = objectid;
474 btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
475 ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key,
480 key.offset = (u64)-1;
481 new_root = btrfs_read_fs_root_no_name(root->fs_info, &key);
482 if (IS_ERR(new_root)) {
483 btrfs_abort_transaction(trans, root, PTR_ERR(new_root));
484 ret = PTR_ERR(new_root);
488 btrfs_record_root_in_trans(trans, new_root);
490 ret = btrfs_create_subvol_root(trans, new_root, new_dirid);
492 /* We potentially lose an unused inode item here */
493 btrfs_abort_transaction(trans, root, ret);
498 * insert the directory item
500 ret = btrfs_set_inode_index(dir, &index);
502 btrfs_abort_transaction(trans, root, ret);
506 ret = btrfs_insert_dir_item(trans, root,
507 name, namelen, dir, &key,
508 BTRFS_FT_DIR, index);
510 btrfs_abort_transaction(trans, root, ret);
514 btrfs_i_size_write(dir, dir->i_size + namelen * 2);
515 ret = btrfs_update_inode(trans, root, dir);
518 ret = btrfs_add_root_ref(trans, root->fs_info->tree_root,
519 objectid, root->root_key.objectid,
520 btrfs_ino(dir), index, name, namelen);
525 trans->block_rsv = NULL;
526 trans->bytes_reserved = 0;
528 *async_transid = trans->transid;
529 err = btrfs_commit_transaction_async(trans, root, 1);
531 err = btrfs_commit_transaction(trans, root);
533 err = btrfs_commit_transaction(trans, root);
539 d_instantiate(dentry, btrfs_lookup_dentry(dir, dentry));
541 btrfs_subvolume_release_metadata(root, &block_rsv, qgroup_reserved);
545 static int create_snapshot(struct btrfs_root *root, struct inode *dir,
546 struct dentry *dentry, char *name, int namelen,
547 u64 *async_transid, bool readonly,
548 struct btrfs_qgroup_inherit *inherit)
551 struct btrfs_pending_snapshot *pending_snapshot;
552 struct btrfs_trans_handle *trans;
558 ret = btrfs_start_delalloc_inodes(root, 0);
562 btrfs_wait_ordered_extents(root, 0);
564 pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS);
565 if (!pending_snapshot)
568 btrfs_init_block_rsv(&pending_snapshot->block_rsv,
569 BTRFS_BLOCK_RSV_TEMP);
571 * 1 - parent dir inode
574 * 2 - root ref/backref
575 * 1 - root of snapshot
577 ret = btrfs_subvolume_reserve_metadata(BTRFS_I(dir)->root,
578 &pending_snapshot->block_rsv, 7,
579 &pending_snapshot->qgroup_reserved);
583 pending_snapshot->dentry = dentry;
584 pending_snapshot->root = root;
585 pending_snapshot->readonly = readonly;
586 pending_snapshot->dir = dir;
587 pending_snapshot->inherit = inherit;
589 trans = btrfs_start_transaction(root, 0);
591 ret = PTR_ERR(trans);
595 spin_lock(&root->fs_info->trans_lock);
596 list_add(&pending_snapshot->list,
597 &trans->transaction->pending_snapshots);
598 spin_unlock(&root->fs_info->trans_lock);
600 *async_transid = trans->transid;
601 ret = btrfs_commit_transaction_async(trans,
602 root->fs_info->extent_root, 1);
604 ret = btrfs_commit_transaction(trans, root);
606 ret = btrfs_commit_transaction(trans,
607 root->fs_info->extent_root);
612 ret = pending_snapshot->error;
616 ret = btrfs_orphan_cleanup(pending_snapshot->snap);
620 inode = btrfs_lookup_dentry(dentry->d_parent->d_inode, dentry);
622 ret = PTR_ERR(inode);
626 d_instantiate(dentry, inode);
629 btrfs_subvolume_release_metadata(BTRFS_I(dir)->root,
630 &pending_snapshot->block_rsv,
631 pending_snapshot->qgroup_reserved);
633 kfree(pending_snapshot);
637 /* copy of check_sticky in fs/namei.c()
638 * It's inline, so penalty for filesystems that don't use sticky bit is
641 static inline int btrfs_check_sticky(struct inode *dir, struct inode *inode)
643 kuid_t fsuid = current_fsuid();
645 if (!(dir->i_mode & S_ISVTX))
647 if (uid_eq(inode->i_uid, fsuid))
649 if (uid_eq(dir->i_uid, fsuid))
651 return !capable(CAP_FOWNER);
654 /* copy of may_delete in fs/namei.c()
655 * Check whether we can remove a link victim from directory dir, check
656 * whether the type of victim is right.
657 * 1. We can't do it if dir is read-only (done in permission())
658 * 2. We should have write and exec permissions on dir
659 * 3. We can't remove anything from append-only dir
660 * 4. We can't do anything with immutable dir (done in permission())
661 * 5. If the sticky bit on dir is set we should either
662 * a. be owner of dir, or
663 * b. be owner of victim, or
664 * c. have CAP_FOWNER capability
665 * 6. If the victim is append-only or immutable we can't do antyhing with
666 * links pointing to it.
667 * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR.
668 * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR.
669 * 9. We can't remove a root or mountpoint.
670 * 10. We don't allow removal of NFS sillyrenamed files; it's handled by
671 * nfs_async_unlink().
674 static int btrfs_may_delete(struct inode *dir,struct dentry *victim,int isdir)
678 if (!victim->d_inode)
681 BUG_ON(victim->d_parent->d_inode != dir);
682 audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE);
684 error = inode_permission(dir, MAY_WRITE | MAY_EXEC);
689 if (btrfs_check_sticky(dir, victim->d_inode)||
690 IS_APPEND(victim->d_inode)||
691 IS_IMMUTABLE(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
694 if (!S_ISDIR(victim->d_inode->i_mode))
698 } else if (S_ISDIR(victim->d_inode->i_mode))
702 if (victim->d_flags & DCACHE_NFSFS_RENAMED)
707 /* copy of may_create in fs/namei.c() */
708 static inline int btrfs_may_create(struct inode *dir, struct dentry *child)
714 return inode_permission(dir, MAY_WRITE | MAY_EXEC);
718 * Create a new subvolume below @parent. This is largely modeled after
719 * sys_mkdirat and vfs_mkdir, but we only do a single component lookup
720 * inside this filesystem so it's quite a bit simpler.
722 static noinline int btrfs_mksubvol(struct path *parent,
723 char *name, int namelen,
724 struct btrfs_root *snap_src,
725 u64 *async_transid, bool readonly,
726 struct btrfs_qgroup_inherit *inherit)
728 struct inode *dir = parent->dentry->d_inode;
729 struct dentry *dentry;
732 error = mutex_lock_killable_nested(&dir->i_mutex, I_MUTEX_PARENT);
736 dentry = lookup_one_len(name, parent->dentry, namelen);
737 error = PTR_ERR(dentry);
745 error = btrfs_may_create(dir, dentry);
750 * even if this name doesn't exist, we may get hash collisions.
751 * check for them now when we can safely fail
753 error = btrfs_check_dir_item_collision(BTRFS_I(dir)->root,
759 down_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
761 if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0)
765 error = create_snapshot(snap_src, dir, dentry, name, namelen,
766 async_transid, readonly, inherit);
768 error = create_subvol(dir, dentry, name, namelen,
769 async_transid, inherit);
772 fsnotify_mkdir(dir, dentry);
774 up_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
778 mutex_unlock(&dir->i_mutex);
783 * When we're defragging a range, we don't want to kick it off again
784 * if it is really just waiting for delalloc to send it down.
785 * If we find a nice big extent or delalloc range for the bytes in the
786 * file you want to defrag, we return 0 to let you know to skip this
789 static int check_defrag_in_cache(struct inode *inode, u64 offset, int thresh)
791 struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
792 struct extent_map *em = NULL;
793 struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
796 read_lock(&em_tree->lock);
797 em = lookup_extent_mapping(em_tree, offset, PAGE_CACHE_SIZE);
798 read_unlock(&em_tree->lock);
801 end = extent_map_end(em);
803 if (end - offset > thresh)
806 /* if we already have a nice delalloc here, just stop */
808 end = count_range_bits(io_tree, &offset, offset + thresh,
809 thresh, EXTENT_DELALLOC, 1);
816 * helper function to walk through a file and find extents
817 * newer than a specific transid, and smaller than thresh.
819 * This is used by the defragging code to find new and small
822 static int find_new_extents(struct btrfs_root *root,
823 struct inode *inode, u64 newer_than,
824 u64 *off, int thresh)
826 struct btrfs_path *path;
827 struct btrfs_key min_key;
828 struct btrfs_key max_key;
829 struct extent_buffer *leaf;
830 struct btrfs_file_extent_item *extent;
833 u64 ino = btrfs_ino(inode);
835 path = btrfs_alloc_path();
839 min_key.objectid = ino;
840 min_key.type = BTRFS_EXTENT_DATA_KEY;
841 min_key.offset = *off;
843 max_key.objectid = ino;
844 max_key.type = (u8)-1;
845 max_key.offset = (u64)-1;
847 path->keep_locks = 1;
850 ret = btrfs_search_forward(root, &min_key, &max_key,
854 if (min_key.objectid != ino)
856 if (min_key.type != BTRFS_EXTENT_DATA_KEY)
859 leaf = path->nodes[0];
860 extent = btrfs_item_ptr(leaf, path->slots[0],
861 struct btrfs_file_extent_item);
863 type = btrfs_file_extent_type(leaf, extent);
864 if (type == BTRFS_FILE_EXTENT_REG &&
865 btrfs_file_extent_num_bytes(leaf, extent) < thresh &&
866 check_defrag_in_cache(inode, min_key.offset, thresh)) {
867 *off = min_key.offset;
868 btrfs_free_path(path);
872 if (min_key.offset == (u64)-1)
876 btrfs_release_path(path);
879 btrfs_free_path(path);
883 static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start)
885 struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
886 struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
887 struct extent_map *em;
888 u64 len = PAGE_CACHE_SIZE;
891 * hopefully we have this extent in the tree already, try without
892 * the full extent lock
894 read_lock(&em_tree->lock);
895 em = lookup_extent_mapping(em_tree, start, len);
896 read_unlock(&em_tree->lock);
899 /* get the big lock and read metadata off disk */
900 lock_extent(io_tree, start, start + len - 1);
901 em = btrfs_get_extent(inode, NULL, 0, start, len, 0);
902 unlock_extent(io_tree, start, start + len - 1);
911 static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em)
913 struct extent_map *next;
916 /* this is the last extent */
917 if (em->start + em->len >= i_size_read(inode))
920 next = defrag_lookup_extent(inode, em->start + em->len);
921 if (!next || next->block_start >= EXTENT_MAP_LAST_BYTE)
924 free_extent_map(next);
928 static int should_defrag_range(struct inode *inode, u64 start, int thresh,
929 u64 *last_len, u64 *skip, u64 *defrag_end,
932 struct extent_map *em;
934 bool next_mergeable = true;
937 * make sure that once we start defragging an extent, we keep on
940 if (start < *defrag_end)
945 em = defrag_lookup_extent(inode, start);
949 /* this will cover holes, and inline extents */
950 if (em->block_start >= EXTENT_MAP_LAST_BYTE) {
955 next_mergeable = defrag_check_next_extent(inode, em);
958 * we hit a real extent, if it is big or the next extent is not a
959 * real extent, don't bother defragging it
961 if (!compress && (*last_len == 0 || *last_len >= thresh) &&
962 (em->len >= thresh || !next_mergeable))
966 * last_len ends up being a counter of how many bytes we've defragged.
967 * every time we choose not to defrag an extent, we reset *last_len
968 * so that the next tiny extent will force a defrag.
970 * The end result of this is that tiny extents before a single big
971 * extent will force at least part of that big extent to be defragged.
974 *defrag_end = extent_map_end(em);
977 *skip = extent_map_end(em);
986 * it doesn't do much good to defrag one or two pages
987 * at a time. This pulls in a nice chunk of pages
990 * It also makes sure the delalloc code has enough
991 * dirty data to avoid making new small extents as part
994 * It's a good idea to start RA on this range
995 * before calling this.
997 static int cluster_pages_for_defrag(struct inode *inode,
999 unsigned long start_index,
1002 unsigned long file_end;
1003 u64 isize = i_size_read(inode);
1010 struct btrfs_ordered_extent *ordered;
1011 struct extent_state *cached_state = NULL;
1012 struct extent_io_tree *tree;
1013 gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
1015 file_end = (isize - 1) >> PAGE_CACHE_SHIFT;
1016 if (!isize || start_index > file_end)
1019 page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1);
1021 ret = btrfs_delalloc_reserve_space(inode,
1022 page_cnt << PAGE_CACHE_SHIFT);
1026 tree = &BTRFS_I(inode)->io_tree;
1028 /* step one, lock all the pages */
1029 for (i = 0; i < page_cnt; i++) {
1032 page = find_or_create_page(inode->i_mapping,
1033 start_index + i, mask);
1037 page_start = page_offset(page);
1038 page_end = page_start + PAGE_CACHE_SIZE - 1;
1040 lock_extent(tree, page_start, page_end);
1041 ordered = btrfs_lookup_ordered_extent(inode,
1043 unlock_extent(tree, page_start, page_end);
1048 btrfs_start_ordered_extent(inode, ordered, 1);
1049 btrfs_put_ordered_extent(ordered);
1052 * we unlocked the page above, so we need check if
1053 * it was released or not.
1055 if (page->mapping != inode->i_mapping) {
1057 page_cache_release(page);
1062 if (!PageUptodate(page)) {
1063 btrfs_readpage(NULL, page);
1065 if (!PageUptodate(page)) {
1067 page_cache_release(page);
1073 if (page->mapping != inode->i_mapping) {
1075 page_cache_release(page);
1085 if (!(inode->i_sb->s_flags & MS_ACTIVE))
1089 * so now we have a nice long stream of locked
1090 * and up to date pages, lets wait on them
1092 for (i = 0; i < i_done; i++)
1093 wait_on_page_writeback(pages[i]);
1095 page_start = page_offset(pages[0]);
1096 page_end = page_offset(pages[i_done - 1]) + PAGE_CACHE_SIZE;
1098 lock_extent_bits(&BTRFS_I(inode)->io_tree,
1099 page_start, page_end - 1, 0, &cached_state);
1100 clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start,
1101 page_end - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
1102 EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 0, 0,
1103 &cached_state, GFP_NOFS);
1105 if (i_done != page_cnt) {
1106 spin_lock(&BTRFS_I(inode)->lock);
1107 BTRFS_I(inode)->outstanding_extents++;
1108 spin_unlock(&BTRFS_I(inode)->lock);
1109 btrfs_delalloc_release_space(inode,
1110 (page_cnt - i_done) << PAGE_CACHE_SHIFT);
1114 set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1,
1115 &cached_state, GFP_NOFS);
1117 unlock_extent_cached(&BTRFS_I(inode)->io_tree,
1118 page_start, page_end - 1, &cached_state,
1121 for (i = 0; i < i_done; i++) {
1122 clear_page_dirty_for_io(pages[i]);
1123 ClearPageChecked(pages[i]);
1124 set_page_extent_mapped(pages[i]);
1125 set_page_dirty(pages[i]);
1126 unlock_page(pages[i]);
1127 page_cache_release(pages[i]);
1131 for (i = 0; i < i_done; i++) {
1132 unlock_page(pages[i]);
1133 page_cache_release(pages[i]);
1135 btrfs_delalloc_release_space(inode, page_cnt << PAGE_CACHE_SHIFT);
1140 int btrfs_defrag_file(struct inode *inode, struct file *file,
1141 struct btrfs_ioctl_defrag_range_args *range,
1142 u64 newer_than, unsigned long max_to_defrag)
1144 struct btrfs_root *root = BTRFS_I(inode)->root;
1145 struct file_ra_state *ra = NULL;
1146 unsigned long last_index;
1147 u64 isize = i_size_read(inode);
1151 u64 newer_off = range->start;
1153 unsigned long ra_index = 0;
1155 int defrag_count = 0;
1156 int compress_type = BTRFS_COMPRESS_ZLIB;
1157 int extent_thresh = range->extent_thresh;
1158 int max_cluster = (256 * 1024) >> PAGE_CACHE_SHIFT;
1159 int cluster = max_cluster;
1160 u64 new_align = ~((u64)128 * 1024 - 1);
1161 struct page **pages = NULL;
1166 if (range->start >= isize)
1169 if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS) {
1170 if (range->compress_type > BTRFS_COMPRESS_TYPES)
1172 if (range->compress_type)
1173 compress_type = range->compress_type;
1176 if (extent_thresh == 0)
1177 extent_thresh = 256 * 1024;
1180 * if we were not given a file, allocate a readahead
1184 ra = kzalloc(sizeof(*ra), GFP_NOFS);
1187 file_ra_state_init(ra, inode->i_mapping);
1192 pages = kmalloc(sizeof(struct page *) * max_cluster,
1199 /* find the last page to defrag */
1200 if (range->start + range->len > range->start) {
1201 last_index = min_t(u64, isize - 1,
1202 range->start + range->len - 1) >> PAGE_CACHE_SHIFT;
1204 last_index = (isize - 1) >> PAGE_CACHE_SHIFT;
1208 ret = find_new_extents(root, inode, newer_than,
1209 &newer_off, 64 * 1024);
1211 range->start = newer_off;
1213 * we always align our defrag to help keep
1214 * the extents in the file evenly spaced
1216 i = (newer_off & new_align) >> PAGE_CACHE_SHIFT;
1220 i = range->start >> PAGE_CACHE_SHIFT;
1223 max_to_defrag = last_index + 1;
1226 * make writeback starts from i, so the defrag range can be
1227 * written sequentially.
1229 if (i < inode->i_mapping->writeback_index)
1230 inode->i_mapping->writeback_index = i;
1232 while (i <= last_index && defrag_count < max_to_defrag &&
1233 (i < (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >>
1234 PAGE_CACHE_SHIFT)) {
1236 * make sure we stop running if someone unmounts
1239 if (!(inode->i_sb->s_flags & MS_ACTIVE))
1242 if (btrfs_defrag_cancelled(root->fs_info)) {
1243 printk(KERN_DEBUG "btrfs: defrag_file cancelled\n");
1248 if (!should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT,
1249 extent_thresh, &last_len, &skip,
1250 &defrag_end, range->flags &
1251 BTRFS_DEFRAG_RANGE_COMPRESS)) {
1254 * the should_defrag function tells us how much to skip
1255 * bump our counter by the suggested amount
1257 next = (skip + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
1258 i = max(i + 1, next);
1263 cluster = (PAGE_CACHE_ALIGN(defrag_end) >>
1264 PAGE_CACHE_SHIFT) - i;
1265 cluster = min(cluster, max_cluster);
1267 cluster = max_cluster;
1270 if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)
1271 BTRFS_I(inode)->force_compress = compress_type;
1273 if (i + cluster > ra_index) {
1274 ra_index = max(i, ra_index);
1275 btrfs_force_ra(inode->i_mapping, ra, file, ra_index,
1277 ra_index += max_cluster;
1280 mutex_lock(&inode->i_mutex);
1281 ret = cluster_pages_for_defrag(inode, pages, i, cluster);
1283 mutex_unlock(&inode->i_mutex);
1287 defrag_count += ret;
1288 balance_dirty_pages_ratelimited(inode->i_mapping);
1289 mutex_unlock(&inode->i_mutex);
1292 if (newer_off == (u64)-1)
1298 newer_off = max(newer_off + 1,
1299 (u64)i << PAGE_CACHE_SHIFT);
1301 ret = find_new_extents(root, inode,
1302 newer_than, &newer_off,
1305 range->start = newer_off;
1306 i = (newer_off & new_align) >> PAGE_CACHE_SHIFT;
1313 last_len += ret << PAGE_CACHE_SHIFT;
1321 if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO))
1322 filemap_flush(inode->i_mapping);
1324 if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
1325 /* the filemap_flush will queue IO into the worker threads, but
1326 * we have to make sure the IO is actually started and that
1327 * ordered extents get created before we return
1329 atomic_inc(&root->fs_info->async_submit_draining);
1330 while (atomic_read(&root->fs_info->nr_async_submits) ||
1331 atomic_read(&root->fs_info->async_delalloc_pages)) {
1332 wait_event(root->fs_info->async_submit_wait,
1333 (atomic_read(&root->fs_info->nr_async_submits) == 0 &&
1334 atomic_read(&root->fs_info->async_delalloc_pages) == 0));
1336 atomic_dec(&root->fs_info->async_submit_draining);
1338 mutex_lock(&inode->i_mutex);
1339 BTRFS_I(inode)->force_compress = BTRFS_COMPRESS_NONE;
1340 mutex_unlock(&inode->i_mutex);
1343 if (range->compress_type == BTRFS_COMPRESS_LZO) {
1344 btrfs_set_fs_incompat(root->fs_info, COMPRESS_LZO);
1356 static noinline int btrfs_ioctl_resize(struct file *file,
1362 struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
1363 struct btrfs_ioctl_vol_args *vol_args;
1364 struct btrfs_trans_handle *trans;
1365 struct btrfs_device *device = NULL;
1367 char *devstr = NULL;
1371 if (!capable(CAP_SYS_ADMIN))
1374 ret = mnt_want_write_file(file);
1378 if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running,
1380 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
1381 mnt_drop_write_file(file);
1385 mutex_lock(&root->fs_info->volume_mutex);
1386 vol_args = memdup_user(arg, sizeof(*vol_args));
1387 if (IS_ERR(vol_args)) {
1388 ret = PTR_ERR(vol_args);
1392 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
1394 sizestr = vol_args->name;
1395 devstr = strchr(sizestr, ':');
1398 sizestr = devstr + 1;
1400 devstr = vol_args->name;
1401 devid = simple_strtoull(devstr, &end, 10);
1406 printk(KERN_INFO "btrfs: resizing devid %llu\n",
1407 (unsigned long long)devid);
1410 device = btrfs_find_device(root->fs_info, devid, NULL, NULL);
1412 printk(KERN_INFO "btrfs: resizer unable to find device %llu\n",
1413 (unsigned long long)devid);
1418 if (!device->writeable) {
1419 printk(KERN_INFO "btrfs: resizer unable to apply on "
1420 "readonly device %llu\n",
1421 (unsigned long long)devid);
1426 if (!strcmp(sizestr, "max"))
1427 new_size = device->bdev->bd_inode->i_size;
1429 if (sizestr[0] == '-') {
1432 } else if (sizestr[0] == '+') {
1436 new_size = memparse(sizestr, NULL);
1437 if (new_size == 0) {
1443 if (device->is_tgtdev_for_dev_replace) {
1448 old_size = device->total_bytes;
1451 if (new_size > old_size) {
1455 new_size = old_size - new_size;
1456 } else if (mod > 0) {
1457 new_size = old_size + new_size;
1460 if (new_size < 256 * 1024 * 1024) {
1464 if (new_size > device->bdev->bd_inode->i_size) {
1469 do_div(new_size, root->sectorsize);
1470 new_size *= root->sectorsize;
1472 printk_in_rcu(KERN_INFO "btrfs: new size for %s is %llu\n",
1473 rcu_str_deref(device->name),
1474 (unsigned long long)new_size);
1476 if (new_size > old_size) {
1477 trans = btrfs_start_transaction(root, 0);
1478 if (IS_ERR(trans)) {
1479 ret = PTR_ERR(trans);
1482 ret = btrfs_grow_device(trans, device, new_size);
1483 btrfs_commit_transaction(trans, root);
1484 } else if (new_size < old_size) {
1485 ret = btrfs_shrink_device(device, new_size);
1486 } /* equal, nothing need to do */
1491 mutex_unlock(&root->fs_info->volume_mutex);
1492 atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0);
1493 mnt_drop_write_file(file);
1497 static noinline int btrfs_ioctl_snap_create_transid(struct file *file,
1498 char *name, unsigned long fd, int subvol,
1499 u64 *transid, bool readonly,
1500 struct btrfs_qgroup_inherit *inherit)
1505 ret = mnt_want_write_file(file);
1509 namelen = strlen(name);
1510 if (strchr(name, '/')) {
1512 goto out_drop_write;
1515 if (name[0] == '.' &&
1516 (namelen == 1 || (name[1] == '.' && namelen == 2))) {
1518 goto out_drop_write;
1522 ret = btrfs_mksubvol(&file->f_path, name, namelen,
1523 NULL, transid, readonly, inherit);
1525 struct fd src = fdget(fd);
1526 struct inode *src_inode;
1529 goto out_drop_write;
1532 src_inode = file_inode(src.file);
1533 if (src_inode->i_sb != file_inode(file)->i_sb) {
1534 printk(KERN_INFO "btrfs: Snapshot src from "
1538 ret = btrfs_mksubvol(&file->f_path, name, namelen,
1539 BTRFS_I(src_inode)->root,
1540 transid, readonly, inherit);
1545 mnt_drop_write_file(file);
1550 static noinline int btrfs_ioctl_snap_create(struct file *file,
1551 void __user *arg, int subvol)
1553 struct btrfs_ioctl_vol_args *vol_args;
1556 vol_args = memdup_user(arg, sizeof(*vol_args));
1557 if (IS_ERR(vol_args))
1558 return PTR_ERR(vol_args);
1559 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
1561 ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
1562 vol_args->fd, subvol,
1569 static noinline int btrfs_ioctl_snap_create_v2(struct file *file,
1570 void __user *arg, int subvol)
1572 struct btrfs_ioctl_vol_args_v2 *vol_args;
1576 bool readonly = false;
1577 struct btrfs_qgroup_inherit *inherit = NULL;
1579 vol_args = memdup_user(arg, sizeof(*vol_args));
1580 if (IS_ERR(vol_args))
1581 return PTR_ERR(vol_args);
1582 vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
1584 if (vol_args->flags &
1585 ~(BTRFS_SUBVOL_CREATE_ASYNC | BTRFS_SUBVOL_RDONLY |
1586 BTRFS_SUBVOL_QGROUP_INHERIT)) {
1591 if (vol_args->flags & BTRFS_SUBVOL_CREATE_ASYNC)
1593 if (vol_args->flags & BTRFS_SUBVOL_RDONLY)
1595 if (vol_args->flags & BTRFS_SUBVOL_QGROUP_INHERIT) {
1596 if (vol_args->size > PAGE_CACHE_SIZE) {
1600 inherit = memdup_user(vol_args->qgroup_inherit, vol_args->size);
1601 if (IS_ERR(inherit)) {
1602 ret = PTR_ERR(inherit);
1607 ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
1608 vol_args->fd, subvol, ptr,
1611 if (ret == 0 && ptr &&
1613 offsetof(struct btrfs_ioctl_vol_args_v2,
1614 transid), ptr, sizeof(*ptr)))
1622 static noinline int btrfs_ioctl_subvol_getflags(struct file *file,
1625 struct inode *inode = file_inode(file);
1626 struct btrfs_root *root = BTRFS_I(inode)->root;
1630 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID)
1633 down_read(&root->fs_info->subvol_sem);
1634 if (btrfs_root_readonly(root))
1635 flags |= BTRFS_SUBVOL_RDONLY;
1636 up_read(&root->fs_info->subvol_sem);
1638 if (copy_to_user(arg, &flags, sizeof(flags)))
1644 static noinline int btrfs_ioctl_subvol_setflags(struct file *file,
1647 struct inode *inode = file_inode(file);
1648 struct btrfs_root *root = BTRFS_I(inode)->root;
1649 struct btrfs_trans_handle *trans;
1654 ret = mnt_want_write_file(file);
1658 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) {
1660 goto out_drop_write;
1663 if (copy_from_user(&flags, arg, sizeof(flags))) {
1665 goto out_drop_write;
1668 if (flags & BTRFS_SUBVOL_CREATE_ASYNC) {
1670 goto out_drop_write;
1673 if (flags & ~BTRFS_SUBVOL_RDONLY) {
1675 goto out_drop_write;
1678 if (!inode_owner_or_capable(inode)) {
1680 goto out_drop_write;
1683 down_write(&root->fs_info->subvol_sem);
1686 if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root))
1689 root_flags = btrfs_root_flags(&root->root_item);
1690 if (flags & BTRFS_SUBVOL_RDONLY)
1691 btrfs_set_root_flags(&root->root_item,
1692 root_flags | BTRFS_ROOT_SUBVOL_RDONLY);
1694 btrfs_set_root_flags(&root->root_item,
1695 root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY);
1697 trans = btrfs_start_transaction(root, 1);
1698 if (IS_ERR(trans)) {
1699 ret = PTR_ERR(trans);
1703 ret = btrfs_update_root(trans, root->fs_info->tree_root,
1704 &root->root_key, &root->root_item);
1706 btrfs_commit_transaction(trans, root);
1709 btrfs_set_root_flags(&root->root_item, root_flags);
1711 up_write(&root->fs_info->subvol_sem);
1713 mnt_drop_write_file(file);
1719 * helper to check if the subvolume references other subvolumes
1721 static noinline int may_destroy_subvol(struct btrfs_root *root)
1723 struct btrfs_path *path;
1724 struct btrfs_key key;
1727 path = btrfs_alloc_path();
1731 key.objectid = root->root_key.objectid;
1732 key.type = BTRFS_ROOT_REF_KEY;
1733 key.offset = (u64)-1;
1735 ret = btrfs_search_slot(NULL, root->fs_info->tree_root,
1742 if (path->slots[0] > 0) {
1744 btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
1745 if (key.objectid == root->root_key.objectid &&
1746 key.type == BTRFS_ROOT_REF_KEY)
1750 btrfs_free_path(path);
1754 static noinline int key_in_sk(struct btrfs_key *key,
1755 struct btrfs_ioctl_search_key *sk)
1757 struct btrfs_key test;
1760 test.objectid = sk->min_objectid;
1761 test.type = sk->min_type;
1762 test.offset = sk->min_offset;
1764 ret = btrfs_comp_cpu_keys(key, &test);
1768 test.objectid = sk->max_objectid;
1769 test.type = sk->max_type;
1770 test.offset = sk->max_offset;
1772 ret = btrfs_comp_cpu_keys(key, &test);
1778 static noinline int copy_to_sk(struct btrfs_root *root,
1779 struct btrfs_path *path,
1780 struct btrfs_key *key,
1781 struct btrfs_ioctl_search_key *sk,
1783 unsigned long *sk_offset,
1787 struct extent_buffer *leaf;
1788 struct btrfs_ioctl_search_header sh;
1789 unsigned long item_off;
1790 unsigned long item_len;
1796 leaf = path->nodes[0];
1797 slot = path->slots[0];
1798 nritems = btrfs_header_nritems(leaf);
1800 if (btrfs_header_generation(leaf) > sk->max_transid) {
1804 found_transid = btrfs_header_generation(leaf);
1806 for (i = slot; i < nritems; i++) {
1807 item_off = btrfs_item_ptr_offset(leaf, i);
1808 item_len = btrfs_item_size_nr(leaf, i);
1810 btrfs_item_key_to_cpu(leaf, key, i);
1811 if (!key_in_sk(key, sk))
1814 if (sizeof(sh) + item_len > BTRFS_SEARCH_ARGS_BUFSIZE)
1817 if (sizeof(sh) + item_len + *sk_offset >
1818 BTRFS_SEARCH_ARGS_BUFSIZE) {
1823 sh.objectid = key->objectid;
1824 sh.offset = key->offset;
1825 sh.type = key->type;
1827 sh.transid = found_transid;
1829 /* copy search result header */
1830 memcpy(buf + *sk_offset, &sh, sizeof(sh));
1831 *sk_offset += sizeof(sh);
1834 char *p = buf + *sk_offset;
1836 read_extent_buffer(leaf, p,
1837 item_off, item_len);
1838 *sk_offset += item_len;
1842 if (*num_found >= sk->nr_items)
1847 if (key->offset < (u64)-1 && key->offset < sk->max_offset)
1849 else if (key->type < (u8)-1 && key->type < sk->max_type) {
1852 } else if (key->objectid < (u64)-1 && key->objectid < sk->max_objectid) {
1862 static noinline int search_ioctl(struct inode *inode,
1863 struct btrfs_ioctl_search_args *args)
1865 struct btrfs_root *root;
1866 struct btrfs_key key;
1867 struct btrfs_key max_key;
1868 struct btrfs_path *path;
1869 struct btrfs_ioctl_search_key *sk = &args->key;
1870 struct btrfs_fs_info *info = BTRFS_I(inode)->root->fs_info;
1873 unsigned long sk_offset = 0;
1875 path = btrfs_alloc_path();
1879 if (sk->tree_id == 0) {
1880 /* search the root of the inode that was passed */
1881 root = BTRFS_I(inode)->root;
1883 key.objectid = sk->tree_id;
1884 key.type = BTRFS_ROOT_ITEM_KEY;
1885 key.offset = (u64)-1;
1886 root = btrfs_read_fs_root_no_name(info, &key);
1888 printk(KERN_ERR "could not find root %llu\n",
1890 btrfs_free_path(path);
1895 key.objectid = sk->min_objectid;
1896 key.type = sk->min_type;
1897 key.offset = sk->min_offset;
1899 max_key.objectid = sk->max_objectid;
1900 max_key.type = sk->max_type;
1901 max_key.offset = sk->max_offset;
1903 path->keep_locks = 1;
1906 ret = btrfs_search_forward(root, &key, &max_key, path,
1913 ret = copy_to_sk(root, path, &key, sk, args->buf,
1914 &sk_offset, &num_found);
1915 btrfs_release_path(path);
1916 if (ret || num_found >= sk->nr_items)
1922 sk->nr_items = num_found;
1923 btrfs_free_path(path);
1927 static noinline int btrfs_ioctl_tree_search(struct file *file,
1930 struct btrfs_ioctl_search_args *args;
1931 struct inode *inode;
1934 if (!capable(CAP_SYS_ADMIN))
1937 args = memdup_user(argp, sizeof(*args));
1939 return PTR_ERR(args);
1941 inode = file_inode(file);
1942 ret = search_ioctl(inode, args);
1943 if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
1950 * Search INODE_REFs to identify path name of 'dirid' directory
1951 * in a 'tree_id' tree. and sets path name to 'name'.
1953 static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info,
1954 u64 tree_id, u64 dirid, char *name)
1956 struct btrfs_root *root;
1957 struct btrfs_key key;
1963 struct btrfs_inode_ref *iref;
1964 struct extent_buffer *l;
1965 struct btrfs_path *path;
1967 if (dirid == BTRFS_FIRST_FREE_OBJECTID) {
1972 path = btrfs_alloc_path();
1976 ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX];
1978 key.objectid = tree_id;
1979 key.type = BTRFS_ROOT_ITEM_KEY;
1980 key.offset = (u64)-1;
1981 root = btrfs_read_fs_root_no_name(info, &key);
1983 printk(KERN_ERR "could not find root %llu\n", tree_id);
1988 key.objectid = dirid;
1989 key.type = BTRFS_INODE_REF_KEY;
1990 key.offset = (u64)-1;
1993 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1998 slot = path->slots[0];
1999 if (ret > 0 && slot > 0)
2001 btrfs_item_key_to_cpu(l, &key, slot);
2003 if (ret > 0 && (key.objectid != dirid ||
2004 key.type != BTRFS_INODE_REF_KEY)) {
2009 iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref);
2010 len = btrfs_inode_ref_name_len(l, iref);
2012 total_len += len + 1;
2017 read_extent_buffer(l, ptr,(unsigned long)(iref + 1), len);
2019 if (key.offset == BTRFS_FIRST_FREE_OBJECTID)
2022 btrfs_release_path(path);
2023 key.objectid = key.offset;
2024 key.offset = (u64)-1;
2025 dirid = key.objectid;
2029 memmove(name, ptr, total_len);
2030 name[total_len]='\0';
2033 btrfs_free_path(path);
2037 static noinline int btrfs_ioctl_ino_lookup(struct file *file,
2040 struct btrfs_ioctl_ino_lookup_args *args;
2041 struct inode *inode;
2044 if (!capable(CAP_SYS_ADMIN))
2047 args = memdup_user(argp, sizeof(*args));
2049 return PTR_ERR(args);
2051 inode = file_inode(file);
2053 if (args->treeid == 0)
2054 args->treeid = BTRFS_I(inode)->root->root_key.objectid;
2056 ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info,
2057 args->treeid, args->objectid,
2060 if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
2067 static noinline int btrfs_ioctl_snap_destroy(struct file *file,
2070 struct dentry *parent = fdentry(file);
2071 struct dentry *dentry;
2072 struct inode *dir = parent->d_inode;
2073 struct inode *inode;
2074 struct btrfs_root *root = BTRFS_I(dir)->root;
2075 struct btrfs_root *dest = NULL;
2076 struct btrfs_ioctl_vol_args *vol_args;
2077 struct btrfs_trans_handle *trans;
2078 struct btrfs_block_rsv block_rsv;
2079 u64 qgroup_reserved;
2084 vol_args = memdup_user(arg, sizeof(*vol_args));
2085 if (IS_ERR(vol_args))
2086 return PTR_ERR(vol_args);
2088 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2089 namelen = strlen(vol_args->name);
2090 if (strchr(vol_args->name, '/') ||
2091 strncmp(vol_args->name, "..", namelen) == 0) {
2096 err = mnt_want_write_file(file);
2100 err = mutex_lock_killable_nested(&dir->i_mutex, I_MUTEX_PARENT);
2103 dentry = lookup_one_len(vol_args->name, parent, namelen);
2104 if (IS_ERR(dentry)) {
2105 err = PTR_ERR(dentry);
2106 goto out_unlock_dir;
2109 if (!dentry->d_inode) {
2114 inode = dentry->d_inode;
2115 dest = BTRFS_I(inode)->root;
2116 if (!capable(CAP_SYS_ADMIN)){
2118 * Regular user. Only allow this with a special mount
2119 * option, when the user has write+exec access to the
2120 * subvol root, and when rmdir(2) would have been
2123 * Note that this is _not_ check that the subvol is
2124 * empty or doesn't contain data that we wouldn't
2125 * otherwise be able to delete.
2127 * Users who want to delete empty subvols should try
2131 if (!btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED))
2135 * Do not allow deletion if the parent dir is the same
2136 * as the dir to be deleted. That means the ioctl
2137 * must be called on the dentry referencing the root
2138 * of the subvol, not a random directory contained
2145 err = inode_permission(inode, MAY_WRITE | MAY_EXEC);
2150 /* check if subvolume may be deleted by a user */
2151 err = btrfs_may_delete(dir, dentry, 1);
2155 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) {
2160 mutex_lock(&inode->i_mutex);
2161 err = d_invalidate(dentry);
2165 down_write(&root->fs_info->subvol_sem);
2167 err = may_destroy_subvol(dest);
2171 btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP);
2173 * One for dir inode, two for dir entries, two for root
2176 err = btrfs_subvolume_reserve_metadata(root, &block_rsv,
2177 5, &qgroup_reserved);
2181 trans = btrfs_start_transaction(root, 0);
2182 if (IS_ERR(trans)) {
2183 err = PTR_ERR(trans);
2186 trans->block_rsv = &block_rsv;
2187 trans->bytes_reserved = block_rsv.size;
2189 ret = btrfs_unlink_subvol(trans, root, dir,
2190 dest->root_key.objectid,
2191 dentry->d_name.name,
2192 dentry->d_name.len);
2195 btrfs_abort_transaction(trans, root, ret);
2199 btrfs_record_root_in_trans(trans, dest);
2201 memset(&dest->root_item.drop_progress, 0,
2202 sizeof(dest->root_item.drop_progress));
2203 dest->root_item.drop_level = 0;
2204 btrfs_set_root_refs(&dest->root_item, 0);
2206 if (!xchg(&dest->orphan_item_inserted, 1)) {
2207 ret = btrfs_insert_orphan_item(trans,
2208 root->fs_info->tree_root,
2209 dest->root_key.objectid);
2211 btrfs_abort_transaction(trans, root, ret);
2217 trans->block_rsv = NULL;
2218 trans->bytes_reserved = 0;
2219 ret = btrfs_end_transaction(trans, root);
2222 inode->i_flags |= S_DEAD;
2224 btrfs_subvolume_release_metadata(root, &block_rsv, qgroup_reserved);
2226 up_write(&root->fs_info->subvol_sem);
2228 mutex_unlock(&inode->i_mutex);
2230 shrink_dcache_sb(root->fs_info->sb);
2231 btrfs_invalidate_inodes(dest);
2235 if (dest->cache_inode) {
2236 iput(dest->cache_inode);
2237 dest->cache_inode = NULL;
2243 mutex_unlock(&dir->i_mutex);
2244 mnt_drop_write_file(file);
2250 static int btrfs_ioctl_defrag(struct file *file, void __user *argp)
2252 struct inode *inode = file_inode(file);
2253 struct btrfs_root *root = BTRFS_I(inode)->root;
2254 struct btrfs_ioctl_defrag_range_args *range;
2257 ret = mnt_want_write_file(file);
2261 if (btrfs_root_readonly(root)) {
2266 switch (inode->i_mode & S_IFMT) {
2268 if (!capable(CAP_SYS_ADMIN)) {
2272 ret = btrfs_defrag_root(root);
2275 ret = btrfs_defrag_root(root->fs_info->extent_root);
2278 if (!(file->f_mode & FMODE_WRITE)) {
2283 range = kzalloc(sizeof(*range), GFP_KERNEL);
2290 if (copy_from_user(range, argp,
2296 /* compression requires us to start the IO */
2297 if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
2298 range->flags |= BTRFS_DEFRAG_RANGE_START_IO;
2299 range->extent_thresh = (u32)-1;
2302 /* the rest are all set to zero by kzalloc */
2303 range->len = (u64)-1;
2305 ret = btrfs_defrag_file(file_inode(file), file,
2315 mnt_drop_write_file(file);
2319 static long btrfs_ioctl_add_dev(struct btrfs_root *root, void __user *arg)
2321 struct btrfs_ioctl_vol_args *vol_args;
2324 if (!capable(CAP_SYS_ADMIN))
2327 if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running,
2329 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
2333 mutex_lock(&root->fs_info->volume_mutex);
2334 vol_args = memdup_user(arg, sizeof(*vol_args));
2335 if (IS_ERR(vol_args)) {
2336 ret = PTR_ERR(vol_args);
2340 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2341 ret = btrfs_init_new_device(root, vol_args->name);
2345 mutex_unlock(&root->fs_info->volume_mutex);
2346 atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0);
2350 static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg)
2352 struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
2353 struct btrfs_ioctl_vol_args *vol_args;
2356 if (!capable(CAP_SYS_ADMIN))
2359 ret = mnt_want_write_file(file);
2363 vol_args = memdup_user(arg, sizeof(*vol_args));
2364 if (IS_ERR(vol_args)) {
2365 ret = PTR_ERR(vol_args);
2369 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2371 if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running,
2373 ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
2377 mutex_lock(&root->fs_info->volume_mutex);
2378 ret = btrfs_rm_device(root, vol_args->name);
2379 mutex_unlock(&root->fs_info->volume_mutex);
2380 atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0);
2384 mnt_drop_write_file(file);
2388 static long btrfs_ioctl_fs_info(struct btrfs_root *root, void __user *arg)
2390 struct btrfs_ioctl_fs_info_args *fi_args;
2391 struct btrfs_device *device;
2392 struct btrfs_device *next;
2393 struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
2396 if (!capable(CAP_SYS_ADMIN))
2399 fi_args = kzalloc(sizeof(*fi_args), GFP_KERNEL);
2403 fi_args->num_devices = fs_devices->num_devices;
2404 memcpy(&fi_args->fsid, root->fs_info->fsid, sizeof(fi_args->fsid));
2406 mutex_lock(&fs_devices->device_list_mutex);
2407 list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
2408 if (device->devid > fi_args->max_id)
2409 fi_args->max_id = device->devid;
2411 mutex_unlock(&fs_devices->device_list_mutex);
2413 if (copy_to_user(arg, fi_args, sizeof(*fi_args)))
2420 static long btrfs_ioctl_dev_info(struct btrfs_root *root, void __user *arg)
2422 struct btrfs_ioctl_dev_info_args *di_args;
2423 struct btrfs_device *dev;
2424 struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
2426 char *s_uuid = NULL;
2427 char empty_uuid[BTRFS_UUID_SIZE] = {0};
2429 if (!capable(CAP_SYS_ADMIN))
2432 di_args = memdup_user(arg, sizeof(*di_args));
2433 if (IS_ERR(di_args))
2434 return PTR_ERR(di_args);
2436 if (memcmp(empty_uuid, di_args->uuid, BTRFS_UUID_SIZE) != 0)
2437 s_uuid = di_args->uuid;
2439 mutex_lock(&fs_devices->device_list_mutex);
2440 dev = btrfs_find_device(root->fs_info, di_args->devid, s_uuid, NULL);
2447 di_args->devid = dev->devid;
2448 di_args->bytes_used = dev->bytes_used;
2449 di_args->total_bytes = dev->total_bytes;
2450 memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid));
2452 struct rcu_string *name;
2455 name = rcu_dereference(dev->name);
2456 strncpy(di_args->path, name->str, sizeof(di_args->path));
2458 di_args->path[sizeof(di_args->path) - 1] = 0;
2460 di_args->path[0] = '\0';
2464 mutex_unlock(&fs_devices->device_list_mutex);
2465 if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args)))
2472 static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
2473 u64 off, u64 olen, u64 destoff)
2475 struct inode *inode = file_inode(file);
2476 struct btrfs_root *root = BTRFS_I(inode)->root;
2479 struct btrfs_trans_handle *trans;
2480 struct btrfs_path *path;
2481 struct extent_buffer *leaf;
2483 struct btrfs_key key;
2488 u64 bs = root->fs_info->sb->s_blocksize;
2493 * - split compressed inline extents. annoying: we need to
2494 * decompress into destination's address_space (the file offset
2495 * may change, so source mapping won't do), then recompress (or
2496 * otherwise reinsert) a subrange.
2497 * - allow ranges within the same file to be cloned (provided
2498 * they don't overlap)?
2501 /* the destination must be opened for writing */
2502 if (!(file->f_mode & FMODE_WRITE) || (file->f_flags & O_APPEND))
2505 if (btrfs_root_readonly(root))
2508 ret = mnt_want_write_file(file);
2512 src_file = fdget(srcfd);
2513 if (!src_file.file) {
2515 goto out_drop_write;
2519 if (src_file.file->f_path.mnt != file->f_path.mnt)
2522 src = file_inode(src_file.file);
2528 /* the src must be open for reading */
2529 if (!(src_file.file->f_mode & FMODE_READ))
2532 /* don't make the dst file partly checksummed */
2533 if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) !=
2534 (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM))
2538 if (S_ISDIR(src->i_mode) || S_ISDIR(inode->i_mode))
2542 if (src->i_sb != inode->i_sb)
2546 buf = vmalloc(btrfs_level_size(root, 0));
2550 path = btrfs_alloc_path();
2559 mutex_lock_nested(&inode->i_mutex, I_MUTEX_PARENT);
2560 mutex_lock_nested(&src->i_mutex, I_MUTEX_CHILD);
2562 mutex_lock_nested(&src->i_mutex, I_MUTEX_PARENT);
2563 mutex_lock_nested(&inode->i_mutex, I_MUTEX_CHILD);
2566 mutex_lock(&src->i_mutex);
2569 /* determine range to clone */
2571 if (off + len > src->i_size || off + len < off)
2574 olen = len = src->i_size - off;
2575 /* if we extend to eof, continue to block boundary */
2576 if (off + len == src->i_size)
2577 len = ALIGN(src->i_size, bs) - off;
2579 /* verify the end result is block aligned */
2580 if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs) ||
2581 !IS_ALIGNED(destoff, bs))
2584 /* verify if ranges are overlapped within the same file */
2586 if (destoff + len > off && destoff < off + len)
2590 if (destoff > inode->i_size) {
2591 ret = btrfs_cont_expand(inode, inode->i_size, destoff);
2596 /* truncate page cache pages from target inode range */
2597 truncate_inode_pages_range(&inode->i_data, destoff,
2598 PAGE_CACHE_ALIGN(destoff + len) - 1);
2600 /* do any pending delalloc/csum calc on src, one way or
2601 another, and lock file content */
2603 struct btrfs_ordered_extent *ordered;
2604 lock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1);
2605 ordered = btrfs_lookup_first_ordered_extent(src, off + len - 1);
2607 !test_range_bit(&BTRFS_I(src)->io_tree, off, off + len - 1,
2608 EXTENT_DELALLOC, 0, NULL))
2610 unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1);
2612 btrfs_put_ordered_extent(ordered);
2613 btrfs_wait_ordered_range(src, off, len);
2617 key.objectid = btrfs_ino(src);
2618 key.type = BTRFS_EXTENT_DATA_KEY;
2623 * note the key will change type as we walk through the
2626 ret = btrfs_search_slot(NULL, BTRFS_I(src)->root, &key, path,
2631 nritems = btrfs_header_nritems(path->nodes[0]);
2632 if (path->slots[0] >= nritems) {
2633 ret = btrfs_next_leaf(BTRFS_I(src)->root, path);
2638 nritems = btrfs_header_nritems(path->nodes[0]);
2640 leaf = path->nodes[0];
2641 slot = path->slots[0];
2643 btrfs_item_key_to_cpu(leaf, &key, slot);
2644 if (btrfs_key_type(&key) > BTRFS_EXTENT_DATA_KEY ||
2645 key.objectid != btrfs_ino(src))
2648 if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) {
2649 struct btrfs_file_extent_item *extent;
2652 struct btrfs_key new_key;
2653 u64 disko = 0, diskl = 0;
2654 u64 datao = 0, datal = 0;
2658 size = btrfs_item_size_nr(leaf, slot);
2659 read_extent_buffer(leaf, buf,
2660 btrfs_item_ptr_offset(leaf, slot),
2663 extent = btrfs_item_ptr(leaf, slot,
2664 struct btrfs_file_extent_item);
2665 comp = btrfs_file_extent_compression(leaf, extent);
2666 type = btrfs_file_extent_type(leaf, extent);
2667 if (type == BTRFS_FILE_EXTENT_REG ||
2668 type == BTRFS_FILE_EXTENT_PREALLOC) {
2669 disko = btrfs_file_extent_disk_bytenr(leaf,
2671 diskl = btrfs_file_extent_disk_num_bytes(leaf,
2673 datao = btrfs_file_extent_offset(leaf, extent);
2674 datal = btrfs_file_extent_num_bytes(leaf,
2676 } else if (type == BTRFS_FILE_EXTENT_INLINE) {
2677 /* take upper bound, may be compressed */
2678 datal = btrfs_file_extent_ram_bytes(leaf,
2681 btrfs_release_path(path);
2683 if (key.offset + datal <= off ||
2684 key.offset >= off + len - 1)
2687 memcpy(&new_key, &key, sizeof(new_key));
2688 new_key.objectid = btrfs_ino(inode);
2689 if (off <= key.offset)
2690 new_key.offset = key.offset + destoff - off;
2692 new_key.offset = destoff;
2695 * 1 - adjusting old extent (we may have to split it)
2696 * 1 - add new extent
2699 trans = btrfs_start_transaction(root, 3);
2700 if (IS_ERR(trans)) {
2701 ret = PTR_ERR(trans);
2705 if (type == BTRFS_FILE_EXTENT_REG ||
2706 type == BTRFS_FILE_EXTENT_PREALLOC) {
2708 * a | --- range to clone ---| b
2709 * | ------------- extent ------------- |
2712 /* substract range b */
2713 if (key.offset + datal > off + len)
2714 datal = off + len - key.offset;
2716 /* substract range a */
2717 if (off > key.offset) {
2718 datao += off - key.offset;
2719 datal -= off - key.offset;
2722 ret = btrfs_drop_extents(trans, root, inode,
2724 new_key.offset + datal,
2727 btrfs_abort_transaction(trans, root,
2729 btrfs_end_transaction(trans, root);
2733 ret = btrfs_insert_empty_item(trans, root, path,
2736 btrfs_abort_transaction(trans, root,
2738 btrfs_end_transaction(trans, root);
2742 leaf = path->nodes[0];
2743 slot = path->slots[0];
2744 write_extent_buffer(leaf, buf,
2745 btrfs_item_ptr_offset(leaf, slot),
2748 extent = btrfs_item_ptr(leaf, slot,
2749 struct btrfs_file_extent_item);
2751 /* disko == 0 means it's a hole */
2755 btrfs_set_file_extent_offset(leaf, extent,
2757 btrfs_set_file_extent_num_bytes(leaf, extent,
2760 inode_add_bytes(inode, datal);
2761 ret = btrfs_inc_extent_ref(trans, root,
2763 root->root_key.objectid,
2765 new_key.offset - datao,
2768 btrfs_abort_transaction(trans,
2771 btrfs_end_transaction(trans,
2777 } else if (type == BTRFS_FILE_EXTENT_INLINE) {
2780 if (off > key.offset) {
2781 skip = off - key.offset;
2782 new_key.offset += skip;
2785 if (key.offset + datal > off + len)
2786 trim = key.offset + datal - (off + len);
2788 if (comp && (skip || trim)) {
2790 btrfs_end_transaction(trans, root);
2793 size -= skip + trim;
2794 datal -= skip + trim;
2796 ret = btrfs_drop_extents(trans, root, inode,
2798 new_key.offset + datal,
2801 btrfs_abort_transaction(trans, root,
2803 btrfs_end_transaction(trans, root);
2807 ret = btrfs_insert_empty_item(trans, root, path,
2810 btrfs_abort_transaction(trans, root,
2812 btrfs_end_transaction(trans, root);
2818 btrfs_file_extent_calc_inline_size(0);
2819 memmove(buf+start, buf+start+skip,
2823 leaf = path->nodes[0];
2824 slot = path->slots[0];
2825 write_extent_buffer(leaf, buf,
2826 btrfs_item_ptr_offset(leaf, slot),
2828 inode_add_bytes(inode, datal);
2831 btrfs_mark_buffer_dirty(leaf);
2832 btrfs_release_path(path);
2834 inode_inc_iversion(inode);
2835 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
2838 * we round up to the block size at eof when
2839 * determining which extents to clone above,
2840 * but shouldn't round up the file size
2842 endoff = new_key.offset + datal;
2843 if (endoff > destoff+olen)
2844 endoff = destoff+olen;
2845 if (endoff > inode->i_size)
2846 btrfs_i_size_write(inode, endoff);
2848 ret = btrfs_update_inode(trans, root, inode);
2850 btrfs_abort_transaction(trans, root, ret);
2851 btrfs_end_transaction(trans, root);
2854 ret = btrfs_end_transaction(trans, root);
2857 btrfs_release_path(path);
2862 btrfs_release_path(path);
2863 unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1);
2865 mutex_unlock(&src->i_mutex);
2867 mutex_unlock(&inode->i_mutex);
2869 btrfs_free_path(path);
2873 mnt_drop_write_file(file);
2877 static long btrfs_ioctl_clone_range(struct file *file, void __user *argp)
2879 struct btrfs_ioctl_clone_range_args args;
2881 if (copy_from_user(&args, argp, sizeof(args)))
2883 return btrfs_ioctl_clone(file, args.src_fd, args.src_offset,
2884 args.src_length, args.dest_offset);
2888 * there are many ways the trans_start and trans_end ioctls can lead
2889 * to deadlocks. They should only be used by applications that
2890 * basically own the machine, and have a very in depth understanding
2891 * of all the possible deadlocks and enospc problems.
2893 static long btrfs_ioctl_trans_start(struct file *file)
2895 struct inode *inode = file_inode(file);
2896 struct btrfs_root *root = BTRFS_I(inode)->root;
2897 struct btrfs_trans_handle *trans;
2901 if (!capable(CAP_SYS_ADMIN))
2905 if (file->private_data)
2909 if (btrfs_root_readonly(root))
2912 ret = mnt_want_write_file(file);
2916 atomic_inc(&root->fs_info->open_ioctl_trans);
2919 trans = btrfs_start_ioctl_transaction(root);
2923 file->private_data = trans;
2927 atomic_dec(&root->fs_info->open_ioctl_trans);
2928 mnt_drop_write_file(file);
2933 static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp)
2935 struct inode *inode = file_inode(file);
2936 struct btrfs_root *root = BTRFS_I(inode)->root;
2937 struct btrfs_root *new_root;
2938 struct btrfs_dir_item *di;
2939 struct btrfs_trans_handle *trans;
2940 struct btrfs_path *path;
2941 struct btrfs_key location;
2942 struct btrfs_disk_key disk_key;
2947 if (!capable(CAP_SYS_ADMIN))
2950 ret = mnt_want_write_file(file);
2954 if (copy_from_user(&objectid, argp, sizeof(objectid))) {
2960 objectid = root->root_key.objectid;
2962 location.objectid = objectid;
2963 location.type = BTRFS_ROOT_ITEM_KEY;
2964 location.offset = (u64)-1;
2966 new_root = btrfs_read_fs_root_no_name(root->fs_info, &location);
2967 if (IS_ERR(new_root)) {
2968 ret = PTR_ERR(new_root);
2972 path = btrfs_alloc_path();
2977 path->leave_spinning = 1;
2979 trans = btrfs_start_transaction(root, 1);
2980 if (IS_ERR(trans)) {
2981 btrfs_free_path(path);
2982 ret = PTR_ERR(trans);
2986 dir_id = btrfs_super_root_dir(root->fs_info->super_copy);
2987 di = btrfs_lookup_dir_item(trans, root->fs_info->tree_root, path,
2988 dir_id, "default", 7, 1);
2989 if (IS_ERR_OR_NULL(di)) {
2990 btrfs_free_path(path);
2991 btrfs_end_transaction(trans, root);
2992 printk(KERN_ERR "Umm, you don't have the default dir item, "
2993 "this isn't going to work\n");
2998 btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key);
2999 btrfs_set_dir_item_key(path->nodes[0], di, &disk_key);
3000 btrfs_mark_buffer_dirty(path->nodes[0]);
3001 btrfs_free_path(path);
3003 btrfs_set_fs_incompat(root->fs_info, DEFAULT_SUBVOL);
3004 btrfs_end_transaction(trans, root);
3006 mnt_drop_write_file(file);
3010 void btrfs_get_block_group_info(struct list_head *groups_list,
3011 struct btrfs_ioctl_space_info *space)
3013 struct btrfs_block_group_cache *block_group;
3015 space->total_bytes = 0;
3016 space->used_bytes = 0;
3018 list_for_each_entry(block_group, groups_list, list) {
3019 space->flags = block_group->flags;
3020 space->total_bytes += block_group->key.offset;
3021 space->used_bytes +=
3022 btrfs_block_group_used(&block_group->item);
3026 static long btrfs_ioctl_space_info(struct btrfs_root *root, void __user *arg)
3028 struct btrfs_ioctl_space_args space_args;
3029 struct btrfs_ioctl_space_info space;
3030 struct btrfs_ioctl_space_info *dest;
3031 struct btrfs_ioctl_space_info *dest_orig;
3032 struct btrfs_ioctl_space_info __user *user_dest;
3033 struct btrfs_space_info *info;
3034 u64 types[] = {BTRFS_BLOCK_GROUP_DATA,
3035 BTRFS_BLOCK_GROUP_SYSTEM,
3036 BTRFS_BLOCK_GROUP_METADATA,
3037 BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA};
3044 if (copy_from_user(&space_args,
3045 (struct btrfs_ioctl_space_args __user *)arg,
3046 sizeof(space_args)))
3049 for (i = 0; i < num_types; i++) {
3050 struct btrfs_space_info *tmp;
3054 list_for_each_entry_rcu(tmp, &root->fs_info->space_info,
3056 if (tmp->flags == types[i]) {
3066 down_read(&info->groups_sem);
3067 for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
3068 if (!list_empty(&info->block_groups[c]))
3071 up_read(&info->groups_sem);
3074 /* space_slots == 0 means they are asking for a count */
3075 if (space_args.space_slots == 0) {
3076 space_args.total_spaces = slot_count;
3080 slot_count = min_t(u64, space_args.space_slots, slot_count);
3082 alloc_size = sizeof(*dest) * slot_count;
3084 /* we generally have at most 6 or so space infos, one for each raid
3085 * level. So, a whole page should be more than enough for everyone
3087 if (alloc_size > PAGE_CACHE_SIZE)
3090 space_args.total_spaces = 0;
3091 dest = kmalloc(alloc_size, GFP_NOFS);
3096 /* now we have a buffer to copy into */
3097 for (i = 0; i < num_types; i++) {
3098 struct btrfs_space_info *tmp;
3105 list_for_each_entry_rcu(tmp, &root->fs_info->space_info,
3107 if (tmp->flags == types[i]) {
3116 down_read(&info->groups_sem);
3117 for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
3118 if (!list_empty(&info->block_groups[c])) {
3119 btrfs_get_block_group_info(
3120 &info->block_groups[c], &space);
3121 memcpy(dest, &space, sizeof(space));
3123 space_args.total_spaces++;
3129 up_read(&info->groups_sem);
3132 user_dest = (struct btrfs_ioctl_space_info __user *)
3133 (arg + sizeof(struct btrfs_ioctl_space_args));
3135 if (copy_to_user(user_dest, dest_orig, alloc_size))
3140 if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args)))
3147 * there are many ways the trans_start and trans_end ioctls can lead
3148 * to deadlocks. They should only be used by applications that
3149 * basically own the machine, and have a very in depth understanding
3150 * of all the possible deadlocks and enospc problems.
3152 long btrfs_ioctl_trans_end(struct file *file)
3154 struct inode *inode = file_inode(file);
3155 struct btrfs_root *root = BTRFS_I(inode)->root;
3156 struct btrfs_trans_handle *trans;
3158 trans = file->private_data;
3161 file->private_data = NULL;
3163 btrfs_end_transaction(trans, root);
3165 atomic_dec(&root->fs_info->open_ioctl_trans);
3167 mnt_drop_write_file(file);
3171 static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root,
3174 struct btrfs_trans_handle *trans;
3178 trans = btrfs_attach_transaction_barrier(root);
3179 if (IS_ERR(trans)) {
3180 if (PTR_ERR(trans) != -ENOENT)
3181 return PTR_ERR(trans);
3183 /* No running transaction, don't bother */
3184 transid = root->fs_info->last_trans_committed;
3187 transid = trans->transid;
3188 ret = btrfs_commit_transaction_async(trans, root, 0);
3190 btrfs_end_transaction(trans, root);
3195 if (copy_to_user(argp, &transid, sizeof(transid)))
3200 static noinline long btrfs_ioctl_wait_sync(struct btrfs_root *root,
3206 if (copy_from_user(&transid, argp, sizeof(transid)))
3209 transid = 0; /* current trans */
3211 return btrfs_wait_for_commit(root, transid);
3214 static long btrfs_ioctl_scrub(struct file *file, void __user *arg)
3216 struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
3217 struct btrfs_ioctl_scrub_args *sa;
3220 if (!capable(CAP_SYS_ADMIN))
3223 sa = memdup_user(arg, sizeof(*sa));
3227 if (!(sa->flags & BTRFS_SCRUB_READONLY)) {
3228 ret = mnt_want_write_file(file);
3233 ret = btrfs_scrub_dev(root->fs_info, sa->devid, sa->start, sa->end,
3234 &sa->progress, sa->flags & BTRFS_SCRUB_READONLY,
3237 if (copy_to_user(arg, sa, sizeof(*sa)))
3240 if (!(sa->flags & BTRFS_SCRUB_READONLY))
3241 mnt_drop_write_file(file);
3247 static long btrfs_ioctl_scrub_cancel(struct btrfs_root *root, void __user *arg)
3249 if (!capable(CAP_SYS_ADMIN))
3252 return btrfs_scrub_cancel(root->fs_info);
3255 static long btrfs_ioctl_scrub_progress(struct btrfs_root *root,
3258 struct btrfs_ioctl_scrub_args *sa;
3261 if (!capable(CAP_SYS_ADMIN))
3264 sa = memdup_user(arg, sizeof(*sa));
3268 ret = btrfs_scrub_progress(root, sa->devid, &sa->progress);
3270 if (copy_to_user(arg, sa, sizeof(*sa)))
3277 static long btrfs_ioctl_get_dev_stats(struct btrfs_root *root,
3280 struct btrfs_ioctl_get_dev_stats *sa;
3283 sa = memdup_user(arg, sizeof(*sa));
3287 if ((sa->flags & BTRFS_DEV_STATS_RESET) && !capable(CAP_SYS_ADMIN)) {
3292 ret = btrfs_get_dev_stats(root, sa);
3294 if (copy_to_user(arg, sa, sizeof(*sa)))
3301 static long btrfs_ioctl_dev_replace(struct btrfs_root *root, void __user *arg)
3303 struct btrfs_ioctl_dev_replace_args *p;
3306 if (!capable(CAP_SYS_ADMIN))
3309 p = memdup_user(arg, sizeof(*p));
3314 case BTRFS_IOCTL_DEV_REPLACE_CMD_START:
3316 &root->fs_info->mutually_exclusive_operation_running,
3318 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
3321 ret = btrfs_dev_replace_start(root, p);
3323 &root->fs_info->mutually_exclusive_operation_running,
3327 case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS:
3328 btrfs_dev_replace_status(root->fs_info, p);
3331 case BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL:
3332 ret = btrfs_dev_replace_cancel(root->fs_info, p);
3339 if (copy_to_user(arg, p, sizeof(*p)))
3346 static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg)
3352 struct btrfs_ioctl_ino_path_args *ipa = NULL;
3353 struct inode_fs_paths *ipath = NULL;
3354 struct btrfs_path *path;
3356 if (!capable(CAP_DAC_READ_SEARCH))
3359 path = btrfs_alloc_path();
3365 ipa = memdup_user(arg, sizeof(*ipa));
3372 size = min_t(u32, ipa->size, 4096);
3373 ipath = init_ipath(size, root, path);
3374 if (IS_ERR(ipath)) {
3375 ret = PTR_ERR(ipath);
3380 ret = paths_from_inode(ipa->inum, ipath);
3384 for (i = 0; i < ipath->fspath->elem_cnt; ++i) {
3385 rel_ptr = ipath->fspath->val[i] -
3386 (u64)(unsigned long)ipath->fspath->val;
3387 ipath->fspath->val[i] = rel_ptr;
3390 ret = copy_to_user((void *)(unsigned long)ipa->fspath,
3391 (void *)(unsigned long)ipath->fspath, size);
3398 btrfs_free_path(path);
3405 static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx)
3407 struct btrfs_data_container *inodes = ctx;
3408 const size_t c = 3 * sizeof(u64);
3410 if (inodes->bytes_left >= c) {
3411 inodes->bytes_left -= c;
3412 inodes->val[inodes->elem_cnt] = inum;
3413 inodes->val[inodes->elem_cnt + 1] = offset;
3414 inodes->val[inodes->elem_cnt + 2] = root;
3415 inodes->elem_cnt += 3;
3417 inodes->bytes_missing += c - inodes->bytes_left;
3418 inodes->bytes_left = 0;
3419 inodes->elem_missed += 3;
3425 static long btrfs_ioctl_logical_to_ino(struct btrfs_root *root,
3430 struct btrfs_ioctl_logical_ino_args *loi;
3431 struct btrfs_data_container *inodes = NULL;
3432 struct btrfs_path *path = NULL;
3434 if (!capable(CAP_SYS_ADMIN))
3437 loi = memdup_user(arg, sizeof(*loi));
3444 path = btrfs_alloc_path();
3450 size = min_t(u32, loi->size, 64 * 1024);
3451 inodes = init_data_container(size);
3452 if (IS_ERR(inodes)) {
3453 ret = PTR_ERR(inodes);
3458 ret = iterate_inodes_from_logical(loi->logical, root->fs_info, path,
3459 build_ino_list, inodes);
3465 ret = copy_to_user((void *)(unsigned long)loi->inodes,
3466 (void *)(unsigned long)inodes, size);
3471 btrfs_free_path(path);
3478 void update_ioctl_balance_args(struct btrfs_fs_info *fs_info, int lock,
3479 struct btrfs_ioctl_balance_args *bargs)
3481 struct btrfs_balance_control *bctl = fs_info->balance_ctl;
3483 bargs->flags = bctl->flags;
3485 if (atomic_read(&fs_info->balance_running))
3486 bargs->state |= BTRFS_BALANCE_STATE_RUNNING;
3487 if (atomic_read(&fs_info->balance_pause_req))
3488 bargs->state |= BTRFS_BALANCE_STATE_PAUSE_REQ;
3489 if (atomic_read(&fs_info->balance_cancel_req))
3490 bargs->state |= BTRFS_BALANCE_STATE_CANCEL_REQ;
3492 memcpy(&bargs->data, &bctl->data, sizeof(bargs->data));
3493 memcpy(&bargs->meta, &bctl->meta, sizeof(bargs->meta));
3494 memcpy(&bargs->sys, &bctl->sys, sizeof(bargs->sys));
3497 spin_lock(&fs_info->balance_lock);
3498 memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat));
3499 spin_unlock(&fs_info->balance_lock);
3501 memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat));
3505 static long btrfs_ioctl_balance(struct file *file, void __user *arg)
3507 struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
3508 struct btrfs_fs_info *fs_info = root->fs_info;
3509 struct btrfs_ioctl_balance_args *bargs;
3510 struct btrfs_balance_control *bctl;
3511 bool need_unlock; /* for mut. excl. ops lock */
3514 if (!capable(CAP_SYS_ADMIN))
3517 ret = mnt_want_write_file(file);
3522 if (!atomic_xchg(&fs_info->mutually_exclusive_operation_running, 1)) {
3523 mutex_lock(&fs_info->volume_mutex);
3524 mutex_lock(&fs_info->balance_mutex);
3530 * mut. excl. ops lock is locked. Three possibilites:
3531 * (1) some other op is running
3532 * (2) balance is running
3533 * (3) balance is paused -- special case (think resume)
3535 mutex_lock(&fs_info->balance_mutex);
3536 if (fs_info->balance_ctl) {
3537 /* this is either (2) or (3) */
3538 if (!atomic_read(&fs_info->balance_running)) {
3539 mutex_unlock(&fs_info->balance_mutex);
3540 if (!mutex_trylock(&fs_info->volume_mutex))
3542 mutex_lock(&fs_info->balance_mutex);
3544 if (fs_info->balance_ctl &&
3545 !atomic_read(&fs_info->balance_running)) {
3547 need_unlock = false;
3551 mutex_unlock(&fs_info->balance_mutex);
3552 mutex_unlock(&fs_info->volume_mutex);
3556 mutex_unlock(&fs_info->balance_mutex);
3562 mutex_unlock(&fs_info->balance_mutex);
3563 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
3569 BUG_ON(!atomic_read(&fs_info->mutually_exclusive_operation_running));
3572 bargs = memdup_user(arg, sizeof(*bargs));
3573 if (IS_ERR(bargs)) {
3574 ret = PTR_ERR(bargs);
3578 if (bargs->flags & BTRFS_BALANCE_RESUME) {
3579 if (!fs_info->balance_ctl) {
3584 bctl = fs_info->balance_ctl;
3585 spin_lock(&fs_info->balance_lock);
3586 bctl->flags |= BTRFS_BALANCE_RESUME;
3587 spin_unlock(&fs_info->balance_lock);
3595 if (fs_info->balance_ctl) {
3600 bctl = kzalloc(sizeof(*bctl), GFP_NOFS);
3606 bctl->fs_info = fs_info;
3608 memcpy(&bctl->data, &bargs->data, sizeof(bctl->data));
3609 memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta));
3610 memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys));
3612 bctl->flags = bargs->flags;
3614 /* balance everything - no filters */
3615 bctl->flags |= BTRFS_BALANCE_TYPE_MASK;
3620 * Ownership of bctl and mutually_exclusive_operation_running
3621 * goes to to btrfs_balance. bctl is freed in __cancel_balance,
3622 * or, if restriper was paused all the way until unmount, in
3623 * free_fs_info. mutually_exclusive_operation_running is
3624 * cleared in __cancel_balance.
3626 need_unlock = false;
3628 ret = btrfs_balance(bctl, bargs);
3631 if (copy_to_user(arg, bargs, sizeof(*bargs)))
3638 mutex_unlock(&fs_info->balance_mutex);
3639 mutex_unlock(&fs_info->volume_mutex);
3641 atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
3643 mnt_drop_write_file(file);
3647 static long btrfs_ioctl_balance_ctl(struct btrfs_root *root, int cmd)
3649 if (!capable(CAP_SYS_ADMIN))
3653 case BTRFS_BALANCE_CTL_PAUSE:
3654 return btrfs_pause_balance(root->fs_info);
3655 case BTRFS_BALANCE_CTL_CANCEL:
3656 return btrfs_cancel_balance(root->fs_info);
3662 static long btrfs_ioctl_balance_progress(struct btrfs_root *root,
3665 struct btrfs_fs_info *fs_info = root->fs_info;
3666 struct btrfs_ioctl_balance_args *bargs;
3669 if (!capable(CAP_SYS_ADMIN))
3672 mutex_lock(&fs_info->balance_mutex);
3673 if (!fs_info->balance_ctl) {
3678 bargs = kzalloc(sizeof(*bargs), GFP_NOFS);
3684 update_ioctl_balance_args(fs_info, 1, bargs);
3686 if (copy_to_user(arg, bargs, sizeof(*bargs)))
3691 mutex_unlock(&fs_info->balance_mutex);
3695 static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg)
3697 struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
3698 struct btrfs_ioctl_quota_ctl_args *sa;
3699 struct btrfs_trans_handle *trans = NULL;
3703 if (!capable(CAP_SYS_ADMIN))
3706 ret = mnt_want_write_file(file);
3710 sa = memdup_user(arg, sizeof(*sa));
3716 down_write(&root->fs_info->subvol_sem);
3717 trans = btrfs_start_transaction(root->fs_info->tree_root, 2);
3718 if (IS_ERR(trans)) {
3719 ret = PTR_ERR(trans);
3724 case BTRFS_QUOTA_CTL_ENABLE:
3725 ret = btrfs_quota_enable(trans, root->fs_info);
3727 case BTRFS_QUOTA_CTL_DISABLE:
3728 ret = btrfs_quota_disable(trans, root->fs_info);
3735 err = btrfs_commit_transaction(trans, root->fs_info->tree_root);
3740 up_write(&root->fs_info->subvol_sem);
3742 mnt_drop_write_file(file);
3746 static long btrfs_ioctl_qgroup_assign(struct file *file, void __user *arg)
3748 struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
3749 struct btrfs_ioctl_qgroup_assign_args *sa;
3750 struct btrfs_trans_handle *trans;
3754 if (!capable(CAP_SYS_ADMIN))
3757 ret = mnt_want_write_file(file);
3761 sa = memdup_user(arg, sizeof(*sa));
3767 trans = btrfs_join_transaction(root);
3768 if (IS_ERR(trans)) {
3769 ret = PTR_ERR(trans);
3773 /* FIXME: check if the IDs really exist */
3775 ret = btrfs_add_qgroup_relation(trans, root->fs_info,
3778 ret = btrfs_del_qgroup_relation(trans, root->fs_info,
3782 err = btrfs_end_transaction(trans, root);
3789 mnt_drop_write_file(file);
3793 static long btrfs_ioctl_qgroup_create(struct file *file, void __user *arg)
3795 struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
3796 struct btrfs_ioctl_qgroup_create_args *sa;
3797 struct btrfs_trans_handle *trans;
3801 if (!capable(CAP_SYS_ADMIN))
3804 ret = mnt_want_write_file(file);
3808 sa = memdup_user(arg, sizeof(*sa));
3814 if (!sa->qgroupid) {
3819 trans = btrfs_join_transaction(root);
3820 if (IS_ERR(trans)) {
3821 ret = PTR_ERR(trans);
3825 /* FIXME: check if the IDs really exist */
3827 ret = btrfs_create_qgroup(trans, root->fs_info, sa->qgroupid,
3830 ret = btrfs_remove_qgroup(trans, root->fs_info, sa->qgroupid);
3833 err = btrfs_end_transaction(trans, root);
3840 mnt_drop_write_file(file);
3844 static long btrfs_ioctl_qgroup_limit(struct file *file, void __user *arg)
3846 struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
3847 struct btrfs_ioctl_qgroup_limit_args *sa;
3848 struct btrfs_trans_handle *trans;
3853 if (!capable(CAP_SYS_ADMIN))
3856 ret = mnt_want_write_file(file);
3860 sa = memdup_user(arg, sizeof(*sa));
3866 trans = btrfs_join_transaction(root);
3867 if (IS_ERR(trans)) {
3868 ret = PTR_ERR(trans);
3872 qgroupid = sa->qgroupid;
3874 /* take the current subvol as qgroup */
3875 qgroupid = root->root_key.objectid;
3878 /* FIXME: check if the IDs really exist */
3879 ret = btrfs_limit_qgroup(trans, root->fs_info, qgroupid, &sa->lim);
3881 err = btrfs_end_transaction(trans, root);
3888 mnt_drop_write_file(file);
3892 static long btrfs_ioctl_quota_rescan(struct file *file, void __user *arg)
3894 struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
3895 struct btrfs_ioctl_quota_rescan_args *qsa;
3898 if (!capable(CAP_SYS_ADMIN))
3901 ret = mnt_want_write_file(file);
3905 qsa = memdup_user(arg, sizeof(*qsa));
3916 ret = btrfs_qgroup_rescan(root->fs_info);
3921 mnt_drop_write_file(file);
3925 static long btrfs_ioctl_quota_rescan_status(struct file *file, void __user *arg)
3927 struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
3928 struct btrfs_ioctl_quota_rescan_args *qsa;
3931 if (!capable(CAP_SYS_ADMIN))
3934 qsa = kzalloc(sizeof(*qsa), GFP_NOFS);
3938 if (root->fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) {
3940 qsa->progress = root->fs_info->qgroup_rescan_progress.objectid;
3943 if (copy_to_user(arg, qsa, sizeof(*qsa)))
3950 static long btrfs_ioctl_quota_rescan_wait(struct file *file, void __user *arg)
3952 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
3954 if (!capable(CAP_SYS_ADMIN))
3957 return btrfs_qgroup_wait_for_completion(root->fs_info);
3960 static long btrfs_ioctl_set_received_subvol(struct file *file,
3963 struct btrfs_ioctl_received_subvol_args *sa = NULL;
3964 struct inode *inode = file_inode(file);
3965 struct btrfs_root *root = BTRFS_I(inode)->root;
3966 struct btrfs_root_item *root_item = &root->root_item;
3967 struct btrfs_trans_handle *trans;
3968 struct timespec ct = CURRENT_TIME;
3971 ret = mnt_want_write_file(file);
3975 down_write(&root->fs_info->subvol_sem);
3977 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) {
3982 if (btrfs_root_readonly(root)) {
3987 if (!inode_owner_or_capable(inode)) {
3992 sa = memdup_user(arg, sizeof(*sa));
3999 trans = btrfs_start_transaction(root, 1);
4000 if (IS_ERR(trans)) {
4001 ret = PTR_ERR(trans);
4006 sa->rtransid = trans->transid;
4007 sa->rtime.sec = ct.tv_sec;
4008 sa->rtime.nsec = ct.tv_nsec;
4010 memcpy(root_item->received_uuid, sa->uuid, BTRFS_UUID_SIZE);
4011 btrfs_set_root_stransid(root_item, sa->stransid);
4012 btrfs_set_root_rtransid(root_item, sa->rtransid);
4013 root_item->stime.sec = cpu_to_le64(sa->stime.sec);
4014 root_item->stime.nsec = cpu_to_le32(sa->stime.nsec);
4015 root_item->rtime.sec = cpu_to_le64(sa->rtime.sec);
4016 root_item->rtime.nsec = cpu_to_le32(sa->rtime.nsec);
4018 ret = btrfs_update_root(trans, root->fs_info->tree_root,
4019 &root->root_key, &root->root_item);
4021 btrfs_end_transaction(trans, root);
4025 ret = btrfs_commit_transaction(trans, root);
4030 ret = copy_to_user(arg, sa, sizeof(*sa));
4036 up_write(&root->fs_info->subvol_sem);
4037 mnt_drop_write_file(file);
4041 static int btrfs_ioctl_get_fslabel(struct file *file, void __user *arg)
4043 struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
4044 const char *label = root->fs_info->super_copy->label;
4045 size_t len = strnlen(label, BTRFS_LABEL_SIZE);
4048 if (len == BTRFS_LABEL_SIZE) {
4049 pr_warn("btrfs: label is too long, return the first %zu bytes\n",
4053 mutex_lock(&root->fs_info->volume_mutex);
4054 ret = copy_to_user(arg, label, len);
4055 mutex_unlock(&root->fs_info->volume_mutex);
4057 return ret ? -EFAULT : 0;
4060 static int btrfs_ioctl_set_fslabel(struct file *file, void __user *arg)
4062 struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
4063 struct btrfs_super_block *super_block = root->fs_info->super_copy;
4064 struct btrfs_trans_handle *trans;
4065 char label[BTRFS_LABEL_SIZE];
4068 if (!capable(CAP_SYS_ADMIN))
4071 if (copy_from_user(label, arg, sizeof(label)))
4074 if (strnlen(label, BTRFS_LABEL_SIZE) == BTRFS_LABEL_SIZE) {
4075 pr_err("btrfs: unable to set label with more than %d bytes\n",
4076 BTRFS_LABEL_SIZE - 1);
4080 ret = mnt_want_write_file(file);
4084 mutex_lock(&root->fs_info->volume_mutex);
4085 trans = btrfs_start_transaction(root, 0);
4086 if (IS_ERR(trans)) {
4087 ret = PTR_ERR(trans);
4091 strcpy(super_block->label, label);
4092 ret = btrfs_end_transaction(trans, root);
4095 mutex_unlock(&root->fs_info->volume_mutex);
4096 mnt_drop_write_file(file);
4100 long btrfs_ioctl(struct file *file, unsigned int
4101 cmd, unsigned long arg)
4103 struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
4104 void __user *argp = (void __user *)arg;
4107 case FS_IOC_GETFLAGS:
4108 return btrfs_ioctl_getflags(file, argp);
4109 case FS_IOC_SETFLAGS:
4110 return btrfs_ioctl_setflags(file, argp);
4111 case FS_IOC_GETVERSION:
4112 return btrfs_ioctl_getversion(file, argp);
4114 return btrfs_ioctl_fitrim(file, argp);
4115 case BTRFS_IOC_SNAP_CREATE:
4116 return btrfs_ioctl_snap_create(file, argp, 0);
4117 case BTRFS_IOC_SNAP_CREATE_V2:
4118 return btrfs_ioctl_snap_create_v2(file, argp, 0);
4119 case BTRFS_IOC_SUBVOL_CREATE:
4120 return btrfs_ioctl_snap_create(file, argp, 1);
4121 case BTRFS_IOC_SUBVOL_CREATE_V2:
4122 return btrfs_ioctl_snap_create_v2(file, argp, 1);
4123 case BTRFS_IOC_SNAP_DESTROY:
4124 return btrfs_ioctl_snap_destroy(file, argp);
4125 case BTRFS_IOC_SUBVOL_GETFLAGS:
4126 return btrfs_ioctl_subvol_getflags(file, argp);
4127 case BTRFS_IOC_SUBVOL_SETFLAGS:
4128 return btrfs_ioctl_subvol_setflags(file, argp);
4129 case BTRFS_IOC_DEFAULT_SUBVOL:
4130 return btrfs_ioctl_default_subvol(file, argp);
4131 case BTRFS_IOC_DEFRAG:
4132 return btrfs_ioctl_defrag(file, NULL);
4133 case BTRFS_IOC_DEFRAG_RANGE:
4134 return btrfs_ioctl_defrag(file, argp);
4135 case BTRFS_IOC_RESIZE:
4136 return btrfs_ioctl_resize(file, argp);
4137 case BTRFS_IOC_ADD_DEV:
4138 return btrfs_ioctl_add_dev(root, argp);
4139 case BTRFS_IOC_RM_DEV:
4140 return btrfs_ioctl_rm_dev(file, argp);
4141 case BTRFS_IOC_FS_INFO:
4142 return btrfs_ioctl_fs_info(root, argp);
4143 case BTRFS_IOC_DEV_INFO:
4144 return btrfs_ioctl_dev_info(root, argp);
4145 case BTRFS_IOC_BALANCE:
4146 return btrfs_ioctl_balance(file, NULL);
4147 case BTRFS_IOC_CLONE:
4148 return btrfs_ioctl_clone(file, arg, 0, 0, 0);
4149 case BTRFS_IOC_CLONE_RANGE:
4150 return btrfs_ioctl_clone_range(file, argp);
4151 case BTRFS_IOC_TRANS_START:
4152 return btrfs_ioctl_trans_start(file);
4153 case BTRFS_IOC_TRANS_END:
4154 return btrfs_ioctl_trans_end(file);
4155 case BTRFS_IOC_TREE_SEARCH:
4156 return btrfs_ioctl_tree_search(file, argp);
4157 case BTRFS_IOC_INO_LOOKUP:
4158 return btrfs_ioctl_ino_lookup(file, argp);
4159 case BTRFS_IOC_INO_PATHS:
4160 return btrfs_ioctl_ino_to_path(root, argp);
4161 case BTRFS_IOC_LOGICAL_INO:
4162 return btrfs_ioctl_logical_to_ino(root, argp);
4163 case BTRFS_IOC_SPACE_INFO:
4164 return btrfs_ioctl_space_info(root, argp);
4165 case BTRFS_IOC_SYNC:
4166 btrfs_sync_fs(file->f_dentry->d_sb, 1);
4168 case BTRFS_IOC_START_SYNC:
4169 return btrfs_ioctl_start_sync(root, argp);
4170 case BTRFS_IOC_WAIT_SYNC:
4171 return btrfs_ioctl_wait_sync(root, argp);
4172 case BTRFS_IOC_SCRUB:
4173 return btrfs_ioctl_scrub(file, argp);
4174 case BTRFS_IOC_SCRUB_CANCEL:
4175 return btrfs_ioctl_scrub_cancel(root, argp);
4176 case BTRFS_IOC_SCRUB_PROGRESS:
4177 return btrfs_ioctl_scrub_progress(root, argp);
4178 case BTRFS_IOC_BALANCE_V2:
4179 return btrfs_ioctl_balance(file, argp);
4180 case BTRFS_IOC_BALANCE_CTL:
4181 return btrfs_ioctl_balance_ctl(root, arg);
4182 case BTRFS_IOC_BALANCE_PROGRESS:
4183 return btrfs_ioctl_balance_progress(root, argp);
4184 case BTRFS_IOC_SET_RECEIVED_SUBVOL:
4185 return btrfs_ioctl_set_received_subvol(file, argp);
4186 case BTRFS_IOC_SEND:
4187 return btrfs_ioctl_send(file, argp);
4188 case BTRFS_IOC_GET_DEV_STATS:
4189 return btrfs_ioctl_get_dev_stats(root, argp);
4190 case BTRFS_IOC_QUOTA_CTL:
4191 return btrfs_ioctl_quota_ctl(file, argp);
4192 case BTRFS_IOC_QGROUP_ASSIGN:
4193 return btrfs_ioctl_qgroup_assign(file, argp);
4194 case BTRFS_IOC_QGROUP_CREATE:
4195 return btrfs_ioctl_qgroup_create(file, argp);
4196 case BTRFS_IOC_QGROUP_LIMIT:
4197 return btrfs_ioctl_qgroup_limit(file, argp);
4198 case BTRFS_IOC_QUOTA_RESCAN:
4199 return btrfs_ioctl_quota_rescan(file, argp);
4200 case BTRFS_IOC_QUOTA_RESCAN_STATUS:
4201 return btrfs_ioctl_quota_rescan_status(file, argp);
4202 case BTRFS_IOC_QUOTA_RESCAN_WAIT:
4203 return btrfs_ioctl_quota_rescan_wait(file, argp);
4204 case BTRFS_IOC_DEV_REPLACE:
4205 return btrfs_ioctl_dev_replace(root, argp);
4206 case BTRFS_IOC_GET_FSLABEL:
4207 return btrfs_ioctl_get_fslabel(file, argp);
4208 case BTRFS_IOC_SET_FSLABEL:
4209 return btrfs_ioctl_set_fslabel(file, argp);