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->f_path.dentry->d_inode);
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->f_path.dentry->d_inode;
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->f_path.dentry->d_inode;
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 btrfs_root *root,
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 *new_root;
378 struct dentry *parent = dentry->d_parent;
380 struct timespec cur_time = CURRENT_TIME;
384 u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
388 ret = btrfs_find_free_objectid(root->fs_info->tree_root, &objectid);
392 dir = parent->d_inode;
400 trans = btrfs_start_transaction(root, 6);
402 return PTR_ERR(trans);
404 ret = btrfs_qgroup_inherit(trans, root->fs_info, 0, objectid, inherit);
408 leaf = btrfs_alloc_free_block(trans, root, root->leafsize,
409 0, objectid, NULL, 0, 0, 0);
415 memset_extent_buffer(leaf, 0, 0, sizeof(struct btrfs_header));
416 btrfs_set_header_bytenr(leaf, leaf->start);
417 btrfs_set_header_generation(leaf, trans->transid);
418 btrfs_set_header_backref_rev(leaf, BTRFS_MIXED_BACKREF_REV);
419 btrfs_set_header_owner(leaf, objectid);
421 write_extent_buffer(leaf, root->fs_info->fsid,
422 (unsigned long)btrfs_header_fsid(leaf),
424 write_extent_buffer(leaf, root->fs_info->chunk_tree_uuid,
425 (unsigned long)btrfs_header_chunk_tree_uuid(leaf),
427 btrfs_mark_buffer_dirty(leaf);
429 memset(&root_item, 0, sizeof(root_item));
431 inode_item = &root_item.inode;
432 inode_item->generation = cpu_to_le64(1);
433 inode_item->size = cpu_to_le64(3);
434 inode_item->nlink = cpu_to_le32(1);
435 inode_item->nbytes = cpu_to_le64(root->leafsize);
436 inode_item->mode = cpu_to_le32(S_IFDIR | 0755);
439 root_item.byte_limit = 0;
440 inode_item->flags = cpu_to_le64(BTRFS_INODE_ROOT_ITEM_INIT);
442 btrfs_set_root_bytenr(&root_item, leaf->start);
443 btrfs_set_root_generation(&root_item, trans->transid);
444 btrfs_set_root_level(&root_item, 0);
445 btrfs_set_root_refs(&root_item, 1);
446 btrfs_set_root_used(&root_item, leaf->len);
447 btrfs_set_root_last_snapshot(&root_item, 0);
449 btrfs_set_root_generation_v2(&root_item,
450 btrfs_root_generation(&root_item));
451 uuid_le_gen(&new_uuid);
452 memcpy(root_item.uuid, new_uuid.b, BTRFS_UUID_SIZE);
453 root_item.otime.sec = cpu_to_le64(cur_time.tv_sec);
454 root_item.otime.nsec = cpu_to_le32(cur_time.tv_nsec);
455 root_item.ctime = root_item.otime;
456 btrfs_set_root_ctransid(&root_item, trans->transid);
457 btrfs_set_root_otransid(&root_item, trans->transid);
459 btrfs_tree_unlock(leaf);
460 free_extent_buffer(leaf);
463 btrfs_set_root_dirid(&root_item, new_dirid);
465 key.objectid = objectid;
467 btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
468 ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key,
473 key.offset = (u64)-1;
474 new_root = btrfs_read_fs_root_no_name(root->fs_info, &key);
475 if (IS_ERR(new_root)) {
476 btrfs_abort_transaction(trans, root, PTR_ERR(new_root));
477 ret = PTR_ERR(new_root);
481 btrfs_record_root_in_trans(trans, new_root);
483 ret = btrfs_create_subvol_root(trans, new_root, new_dirid);
485 /* We potentially lose an unused inode item here */
486 btrfs_abort_transaction(trans, root, ret);
491 * insert the directory item
493 ret = btrfs_set_inode_index(dir, &index);
495 btrfs_abort_transaction(trans, root, ret);
499 ret = btrfs_insert_dir_item(trans, root,
500 name, namelen, dir, &key,
501 BTRFS_FT_DIR, index);
503 btrfs_abort_transaction(trans, root, ret);
507 btrfs_i_size_write(dir, dir->i_size + namelen * 2);
508 ret = btrfs_update_inode(trans, root, dir);
511 ret = btrfs_add_root_ref(trans, root->fs_info->tree_root,
512 objectid, root->root_key.objectid,
513 btrfs_ino(dir), index, name, namelen);
519 *async_transid = trans->transid;
520 err = btrfs_commit_transaction_async(trans, root, 1);
522 err = btrfs_commit_transaction(trans, root);
528 d_instantiate(dentry, btrfs_lookup_dentry(dir, dentry));
533 static int create_snapshot(struct btrfs_root *root, struct inode *dir,
534 struct dentry *dentry, char *name, int namelen,
535 u64 *async_transid, bool readonly,
536 struct btrfs_qgroup_inherit *inherit)
539 struct btrfs_pending_snapshot *pending_snapshot;
540 struct btrfs_trans_handle *trans;
546 pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS);
547 if (!pending_snapshot)
550 btrfs_init_block_rsv(&pending_snapshot->block_rsv,
551 BTRFS_BLOCK_RSV_TEMP);
552 pending_snapshot->dentry = dentry;
553 pending_snapshot->root = root;
554 pending_snapshot->readonly = readonly;
555 pending_snapshot->dir = dir;
556 pending_snapshot->inherit = inherit;
558 trans = btrfs_start_transaction(root->fs_info->extent_root, 6);
560 ret = PTR_ERR(trans);
564 ret = btrfs_snap_reserve_metadata(trans, pending_snapshot);
567 spin_lock(&root->fs_info->trans_lock);
568 list_add(&pending_snapshot->list,
569 &trans->transaction->pending_snapshots);
570 spin_unlock(&root->fs_info->trans_lock);
572 *async_transid = trans->transid;
573 ret = btrfs_commit_transaction_async(trans,
574 root->fs_info->extent_root, 1);
576 ret = btrfs_commit_transaction(trans,
577 root->fs_info->extent_root);
580 /* cleanup_transaction has freed this for us */
582 pending_snapshot = NULL;
586 ret = pending_snapshot->error;
590 ret = btrfs_orphan_cleanup(pending_snapshot->snap);
594 inode = btrfs_lookup_dentry(dentry->d_parent->d_inode, dentry);
596 ret = PTR_ERR(inode);
600 d_instantiate(dentry, inode);
603 kfree(pending_snapshot);
607 /* copy of check_sticky in fs/namei.c()
608 * It's inline, so penalty for filesystems that don't use sticky bit is
611 static inline int btrfs_check_sticky(struct inode *dir, struct inode *inode)
613 kuid_t fsuid = current_fsuid();
615 if (!(dir->i_mode & S_ISVTX))
617 if (uid_eq(inode->i_uid, fsuid))
619 if (uid_eq(dir->i_uid, fsuid))
621 return !capable(CAP_FOWNER);
624 /* copy of may_delete in fs/namei.c()
625 * Check whether we can remove a link victim from directory dir, check
626 * whether the type of victim is right.
627 * 1. We can't do it if dir is read-only (done in permission())
628 * 2. We should have write and exec permissions on dir
629 * 3. We can't remove anything from append-only dir
630 * 4. We can't do anything with immutable dir (done in permission())
631 * 5. If the sticky bit on dir is set we should either
632 * a. be owner of dir, or
633 * b. be owner of victim, or
634 * c. have CAP_FOWNER capability
635 * 6. If the victim is append-only or immutable we can't do antyhing with
636 * links pointing to it.
637 * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR.
638 * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR.
639 * 9. We can't remove a root or mountpoint.
640 * 10. We don't allow removal of NFS sillyrenamed files; it's handled by
641 * nfs_async_unlink().
644 static int btrfs_may_delete(struct inode *dir,struct dentry *victim,int isdir)
648 if (!victim->d_inode)
651 BUG_ON(victim->d_parent->d_inode != dir);
652 audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE);
654 error = inode_permission(dir, MAY_WRITE | MAY_EXEC);
659 if (btrfs_check_sticky(dir, victim->d_inode)||
660 IS_APPEND(victim->d_inode)||
661 IS_IMMUTABLE(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
664 if (!S_ISDIR(victim->d_inode->i_mode))
668 } else if (S_ISDIR(victim->d_inode->i_mode))
672 if (victim->d_flags & DCACHE_NFSFS_RENAMED)
677 /* copy of may_create in fs/namei.c() */
678 static inline int btrfs_may_create(struct inode *dir, struct dentry *child)
684 return inode_permission(dir, MAY_WRITE | MAY_EXEC);
688 * Create a new subvolume below @parent. This is largely modeled after
689 * sys_mkdirat and vfs_mkdir, but we only do a single component lookup
690 * inside this filesystem so it's quite a bit simpler.
692 static noinline int btrfs_mksubvol(struct path *parent,
693 char *name, int namelen,
694 struct btrfs_root *snap_src,
695 u64 *async_transid, bool readonly,
696 struct btrfs_qgroup_inherit *inherit)
698 struct inode *dir = parent->dentry->d_inode;
699 struct dentry *dentry;
702 mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
704 dentry = lookup_one_len(name, parent->dentry, namelen);
705 error = PTR_ERR(dentry);
713 error = btrfs_may_create(dir, dentry);
718 * even if this name doesn't exist, we may get hash collisions.
719 * check for them now when we can safely fail
721 error = btrfs_check_dir_item_collision(BTRFS_I(dir)->root,
727 down_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
729 if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0)
733 error = create_snapshot(snap_src, dir, dentry, name, namelen,
734 async_transid, readonly, inherit);
736 error = create_subvol(BTRFS_I(dir)->root, dentry,
737 name, namelen, async_transid, inherit);
740 fsnotify_mkdir(dir, dentry);
742 up_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
746 mutex_unlock(&dir->i_mutex);
751 * When we're defragging a range, we don't want to kick it off again
752 * if it is really just waiting for delalloc to send it down.
753 * If we find a nice big extent or delalloc range for the bytes in the
754 * file you want to defrag, we return 0 to let you know to skip this
757 static int check_defrag_in_cache(struct inode *inode, u64 offset, int thresh)
759 struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
760 struct extent_map *em = NULL;
761 struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
764 read_lock(&em_tree->lock);
765 em = lookup_extent_mapping(em_tree, offset, PAGE_CACHE_SIZE);
766 read_unlock(&em_tree->lock);
769 end = extent_map_end(em);
771 if (end - offset > thresh)
774 /* if we already have a nice delalloc here, just stop */
776 end = count_range_bits(io_tree, &offset, offset + thresh,
777 thresh, EXTENT_DELALLOC, 1);
784 * helper function to walk through a file and find extents
785 * newer than a specific transid, and smaller than thresh.
787 * This is used by the defragging code to find new and small
790 static int find_new_extents(struct btrfs_root *root,
791 struct inode *inode, u64 newer_than,
792 u64 *off, int thresh)
794 struct btrfs_path *path;
795 struct btrfs_key min_key;
796 struct btrfs_key max_key;
797 struct extent_buffer *leaf;
798 struct btrfs_file_extent_item *extent;
801 u64 ino = btrfs_ino(inode);
803 path = btrfs_alloc_path();
807 min_key.objectid = ino;
808 min_key.type = BTRFS_EXTENT_DATA_KEY;
809 min_key.offset = *off;
811 max_key.objectid = ino;
812 max_key.type = (u8)-1;
813 max_key.offset = (u64)-1;
815 path->keep_locks = 1;
818 ret = btrfs_search_forward(root, &min_key, &max_key,
822 if (min_key.objectid != ino)
824 if (min_key.type != BTRFS_EXTENT_DATA_KEY)
827 leaf = path->nodes[0];
828 extent = btrfs_item_ptr(leaf, path->slots[0],
829 struct btrfs_file_extent_item);
831 type = btrfs_file_extent_type(leaf, extent);
832 if (type == BTRFS_FILE_EXTENT_REG &&
833 btrfs_file_extent_num_bytes(leaf, extent) < thresh &&
834 check_defrag_in_cache(inode, min_key.offset, thresh)) {
835 *off = min_key.offset;
836 btrfs_free_path(path);
840 if (min_key.offset == (u64)-1)
844 btrfs_release_path(path);
847 btrfs_free_path(path);
851 static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start)
853 struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
854 struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
855 struct extent_map *em;
856 u64 len = PAGE_CACHE_SIZE;
859 * hopefully we have this extent in the tree already, try without
860 * the full extent lock
862 read_lock(&em_tree->lock);
863 em = lookup_extent_mapping(em_tree, start, len);
864 read_unlock(&em_tree->lock);
867 /* get the big lock and read metadata off disk */
868 lock_extent(io_tree, start, start + len - 1);
869 em = btrfs_get_extent(inode, NULL, 0, start, len, 0);
870 unlock_extent(io_tree, start, start + len - 1);
879 static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em)
881 struct extent_map *next;
884 /* this is the last extent */
885 if (em->start + em->len >= i_size_read(inode))
888 next = defrag_lookup_extent(inode, em->start + em->len);
889 if (!next || next->block_start >= EXTENT_MAP_LAST_BYTE)
892 free_extent_map(next);
896 static int should_defrag_range(struct inode *inode, u64 start, int thresh,
897 u64 *last_len, u64 *skip, u64 *defrag_end,
900 struct extent_map *em;
902 bool next_mergeable = true;
905 * make sure that once we start defragging an extent, we keep on
908 if (start < *defrag_end)
913 em = defrag_lookup_extent(inode, start);
917 /* this will cover holes, and inline extents */
918 if (em->block_start >= EXTENT_MAP_LAST_BYTE) {
923 next_mergeable = defrag_check_next_extent(inode, em);
926 * we hit a real extent, if it is big or the next extent is not a
927 * real extent, don't bother defragging it
929 if (!compress && (*last_len == 0 || *last_len >= thresh) &&
930 (em->len >= thresh || !next_mergeable))
934 * last_len ends up being a counter of how many bytes we've defragged.
935 * every time we choose not to defrag an extent, we reset *last_len
936 * so that the next tiny extent will force a defrag.
938 * The end result of this is that tiny extents before a single big
939 * extent will force at least part of that big extent to be defragged.
942 *defrag_end = extent_map_end(em);
945 *skip = extent_map_end(em);
954 * it doesn't do much good to defrag one or two pages
955 * at a time. This pulls in a nice chunk of pages
958 * It also makes sure the delalloc code has enough
959 * dirty data to avoid making new small extents as part
962 * It's a good idea to start RA on this range
963 * before calling this.
965 static int cluster_pages_for_defrag(struct inode *inode,
967 unsigned long start_index,
970 unsigned long file_end;
971 u64 isize = i_size_read(inode);
978 struct btrfs_ordered_extent *ordered;
979 struct extent_state *cached_state = NULL;
980 struct extent_io_tree *tree;
981 gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
983 file_end = (isize - 1) >> PAGE_CACHE_SHIFT;
984 if (!isize || start_index > file_end)
987 page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1);
989 ret = btrfs_delalloc_reserve_space(inode,
990 page_cnt << PAGE_CACHE_SHIFT);
994 tree = &BTRFS_I(inode)->io_tree;
996 /* step one, lock all the pages */
997 for (i = 0; i < page_cnt; i++) {
1000 page = find_or_create_page(inode->i_mapping,
1001 start_index + i, mask);
1005 page_start = page_offset(page);
1006 page_end = page_start + PAGE_CACHE_SIZE - 1;
1008 lock_extent(tree, page_start, page_end);
1009 ordered = btrfs_lookup_ordered_extent(inode,
1011 unlock_extent(tree, page_start, page_end);
1016 btrfs_start_ordered_extent(inode, ordered, 1);
1017 btrfs_put_ordered_extent(ordered);
1020 * we unlocked the page above, so we need check if
1021 * it was released or not.
1023 if (page->mapping != inode->i_mapping) {
1025 page_cache_release(page);
1030 if (!PageUptodate(page)) {
1031 btrfs_readpage(NULL, page);
1033 if (!PageUptodate(page)) {
1035 page_cache_release(page);
1041 if (page->mapping != inode->i_mapping) {
1043 page_cache_release(page);
1053 if (!(inode->i_sb->s_flags & MS_ACTIVE))
1057 * so now we have a nice long stream of locked
1058 * and up to date pages, lets wait on them
1060 for (i = 0; i < i_done; i++)
1061 wait_on_page_writeback(pages[i]);
1063 page_start = page_offset(pages[0]);
1064 page_end = page_offset(pages[i_done - 1]) + PAGE_CACHE_SIZE;
1066 lock_extent_bits(&BTRFS_I(inode)->io_tree,
1067 page_start, page_end - 1, 0, &cached_state);
1068 clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start,
1069 page_end - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
1070 EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 0, 0,
1071 &cached_state, GFP_NOFS);
1073 if (i_done != page_cnt) {
1074 spin_lock(&BTRFS_I(inode)->lock);
1075 BTRFS_I(inode)->outstanding_extents++;
1076 spin_unlock(&BTRFS_I(inode)->lock);
1077 btrfs_delalloc_release_space(inode,
1078 (page_cnt - i_done) << PAGE_CACHE_SHIFT);
1082 set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1,
1083 &cached_state, GFP_NOFS);
1085 unlock_extent_cached(&BTRFS_I(inode)->io_tree,
1086 page_start, page_end - 1, &cached_state,
1089 for (i = 0; i < i_done; i++) {
1090 clear_page_dirty_for_io(pages[i]);
1091 ClearPageChecked(pages[i]);
1092 set_page_extent_mapped(pages[i]);
1093 set_page_dirty(pages[i]);
1094 unlock_page(pages[i]);
1095 page_cache_release(pages[i]);
1099 for (i = 0; i < i_done; i++) {
1100 unlock_page(pages[i]);
1101 page_cache_release(pages[i]);
1103 btrfs_delalloc_release_space(inode, page_cnt << PAGE_CACHE_SHIFT);
1108 int btrfs_defrag_file(struct inode *inode, struct file *file,
1109 struct btrfs_ioctl_defrag_range_args *range,
1110 u64 newer_than, unsigned long max_to_defrag)
1112 struct btrfs_root *root = BTRFS_I(inode)->root;
1113 struct file_ra_state *ra = NULL;
1114 unsigned long last_index;
1115 u64 isize = i_size_read(inode);
1119 u64 newer_off = range->start;
1121 unsigned long ra_index = 0;
1123 int defrag_count = 0;
1124 int compress_type = BTRFS_COMPRESS_ZLIB;
1125 int extent_thresh = range->extent_thresh;
1126 int max_cluster = (256 * 1024) >> PAGE_CACHE_SHIFT;
1127 int cluster = max_cluster;
1128 u64 new_align = ~((u64)128 * 1024 - 1);
1129 struct page **pages = NULL;
1131 if (extent_thresh == 0)
1132 extent_thresh = 256 * 1024;
1134 if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS) {
1135 if (range->compress_type > BTRFS_COMPRESS_TYPES)
1137 if (range->compress_type)
1138 compress_type = range->compress_type;
1145 * if we were not given a file, allocate a readahead
1149 ra = kzalloc(sizeof(*ra), GFP_NOFS);
1152 file_ra_state_init(ra, inode->i_mapping);
1157 pages = kmalloc(sizeof(struct page *) * max_cluster,
1164 /* find the last page to defrag */
1165 if (range->start + range->len > range->start) {
1166 last_index = min_t(u64, isize - 1,
1167 range->start + range->len - 1) >> PAGE_CACHE_SHIFT;
1169 last_index = (isize - 1) >> PAGE_CACHE_SHIFT;
1173 ret = find_new_extents(root, inode, newer_than,
1174 &newer_off, 64 * 1024);
1176 range->start = newer_off;
1178 * we always align our defrag to help keep
1179 * the extents in the file evenly spaced
1181 i = (newer_off & new_align) >> PAGE_CACHE_SHIFT;
1185 i = range->start >> PAGE_CACHE_SHIFT;
1188 max_to_defrag = last_index + 1;
1191 * make writeback starts from i, so the defrag range can be
1192 * written sequentially.
1194 if (i < inode->i_mapping->writeback_index)
1195 inode->i_mapping->writeback_index = i;
1197 while (i <= last_index && defrag_count < max_to_defrag &&
1198 (i < (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >>
1199 PAGE_CACHE_SHIFT)) {
1201 * make sure we stop running if someone unmounts
1204 if (!(inode->i_sb->s_flags & MS_ACTIVE))
1207 if (btrfs_defrag_cancelled(root->fs_info)) {
1208 printk(KERN_DEBUG "btrfs: defrag_file cancelled\n");
1213 if (!should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT,
1214 extent_thresh, &last_len, &skip,
1215 &defrag_end, range->flags &
1216 BTRFS_DEFRAG_RANGE_COMPRESS)) {
1219 * the should_defrag function tells us how much to skip
1220 * bump our counter by the suggested amount
1222 next = (skip + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
1223 i = max(i + 1, next);
1228 cluster = (PAGE_CACHE_ALIGN(defrag_end) >>
1229 PAGE_CACHE_SHIFT) - i;
1230 cluster = min(cluster, max_cluster);
1232 cluster = max_cluster;
1235 if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)
1236 BTRFS_I(inode)->force_compress = compress_type;
1238 if (i + cluster > ra_index) {
1239 ra_index = max(i, ra_index);
1240 btrfs_force_ra(inode->i_mapping, ra, file, ra_index,
1242 ra_index += max_cluster;
1245 mutex_lock(&inode->i_mutex);
1246 ret = cluster_pages_for_defrag(inode, pages, i, cluster);
1248 mutex_unlock(&inode->i_mutex);
1252 defrag_count += ret;
1253 balance_dirty_pages_ratelimited(inode->i_mapping);
1254 mutex_unlock(&inode->i_mutex);
1257 if (newer_off == (u64)-1)
1263 newer_off = max(newer_off + 1,
1264 (u64)i << PAGE_CACHE_SHIFT);
1266 ret = find_new_extents(root, inode,
1267 newer_than, &newer_off,
1270 range->start = newer_off;
1271 i = (newer_off & new_align) >> PAGE_CACHE_SHIFT;
1278 last_len += ret << PAGE_CACHE_SHIFT;
1286 if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO))
1287 filemap_flush(inode->i_mapping);
1289 if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
1290 /* the filemap_flush will queue IO into the worker threads, but
1291 * we have to make sure the IO is actually started and that
1292 * ordered extents get created before we return
1294 atomic_inc(&root->fs_info->async_submit_draining);
1295 while (atomic_read(&root->fs_info->nr_async_submits) ||
1296 atomic_read(&root->fs_info->async_delalloc_pages)) {
1297 wait_event(root->fs_info->async_submit_wait,
1298 (atomic_read(&root->fs_info->nr_async_submits) == 0 &&
1299 atomic_read(&root->fs_info->async_delalloc_pages) == 0));
1301 atomic_dec(&root->fs_info->async_submit_draining);
1303 mutex_lock(&inode->i_mutex);
1304 BTRFS_I(inode)->force_compress = BTRFS_COMPRESS_NONE;
1305 mutex_unlock(&inode->i_mutex);
1308 if (range->compress_type == BTRFS_COMPRESS_LZO) {
1309 btrfs_set_fs_incompat(root->fs_info, COMPRESS_LZO);
1321 static noinline int btrfs_ioctl_resize(struct file *file,
1327 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
1328 struct btrfs_ioctl_vol_args *vol_args;
1329 struct btrfs_trans_handle *trans;
1330 struct btrfs_device *device = NULL;
1332 char *devstr = NULL;
1336 if (!capable(CAP_SYS_ADMIN))
1339 ret = mnt_want_write_file(file);
1343 if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running,
1345 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
1346 mnt_drop_write_file(file);
1350 mutex_lock(&root->fs_info->volume_mutex);
1351 vol_args = memdup_user(arg, sizeof(*vol_args));
1352 if (IS_ERR(vol_args)) {
1353 ret = PTR_ERR(vol_args);
1357 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
1359 sizestr = vol_args->name;
1360 devstr = strchr(sizestr, ':');
1363 sizestr = devstr + 1;
1365 devstr = vol_args->name;
1366 devid = simple_strtoull(devstr, &end, 10);
1371 printk(KERN_INFO "btrfs: resizing devid %llu\n",
1372 (unsigned long long)devid);
1375 device = btrfs_find_device(root->fs_info, devid, NULL, NULL);
1377 printk(KERN_INFO "btrfs: resizer unable to find device %llu\n",
1378 (unsigned long long)devid);
1383 if (!device->writeable) {
1384 printk(KERN_INFO "btrfs: resizer unable to apply on "
1385 "readonly device %llu\n",
1386 (unsigned long long)devid);
1391 if (!strcmp(sizestr, "max"))
1392 new_size = device->bdev->bd_inode->i_size;
1394 if (sizestr[0] == '-') {
1397 } else if (sizestr[0] == '+') {
1401 new_size = memparse(sizestr, NULL);
1402 if (new_size == 0) {
1408 if (device->is_tgtdev_for_dev_replace) {
1413 old_size = device->total_bytes;
1416 if (new_size > old_size) {
1420 new_size = old_size - new_size;
1421 } else if (mod > 0) {
1422 new_size = old_size + new_size;
1425 if (new_size < 256 * 1024 * 1024) {
1429 if (new_size > device->bdev->bd_inode->i_size) {
1434 do_div(new_size, root->sectorsize);
1435 new_size *= root->sectorsize;
1437 printk_in_rcu(KERN_INFO "btrfs: new size for %s is %llu\n",
1438 rcu_str_deref(device->name),
1439 (unsigned long long)new_size);
1441 if (new_size > old_size) {
1442 trans = btrfs_start_transaction(root, 0);
1443 if (IS_ERR(trans)) {
1444 ret = PTR_ERR(trans);
1447 ret = btrfs_grow_device(trans, device, new_size);
1448 btrfs_commit_transaction(trans, root);
1449 } else if (new_size < old_size) {
1450 ret = btrfs_shrink_device(device, new_size);
1451 } /* equal, nothing need to do */
1456 mutex_unlock(&root->fs_info->volume_mutex);
1457 atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0);
1458 mnt_drop_write_file(file);
1462 static noinline int btrfs_ioctl_snap_create_transid(struct file *file,
1463 char *name, unsigned long fd, int subvol,
1464 u64 *transid, bool readonly,
1465 struct btrfs_qgroup_inherit *inherit)
1470 ret = mnt_want_write_file(file);
1474 namelen = strlen(name);
1475 if (strchr(name, '/')) {
1477 goto out_drop_write;
1480 if (name[0] == '.' &&
1481 (namelen == 1 || (name[1] == '.' && namelen == 2))) {
1483 goto out_drop_write;
1487 ret = btrfs_mksubvol(&file->f_path, name, namelen,
1488 NULL, transid, readonly, inherit);
1490 struct fd src = fdget(fd);
1491 struct inode *src_inode;
1494 goto out_drop_write;
1497 src_inode = src.file->f_path.dentry->d_inode;
1498 if (src_inode->i_sb != file->f_path.dentry->d_inode->i_sb) {
1499 printk(KERN_INFO "btrfs: Snapshot src from "
1503 ret = btrfs_mksubvol(&file->f_path, name, namelen,
1504 BTRFS_I(src_inode)->root,
1505 transid, readonly, inherit);
1510 mnt_drop_write_file(file);
1515 static noinline int btrfs_ioctl_snap_create(struct file *file,
1516 void __user *arg, int subvol)
1518 struct btrfs_ioctl_vol_args *vol_args;
1521 vol_args = memdup_user(arg, sizeof(*vol_args));
1522 if (IS_ERR(vol_args))
1523 return PTR_ERR(vol_args);
1524 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
1526 ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
1527 vol_args->fd, subvol,
1534 static noinline int btrfs_ioctl_snap_create_v2(struct file *file,
1535 void __user *arg, int subvol)
1537 struct btrfs_ioctl_vol_args_v2 *vol_args;
1541 bool readonly = false;
1542 struct btrfs_qgroup_inherit *inherit = NULL;
1544 vol_args = memdup_user(arg, sizeof(*vol_args));
1545 if (IS_ERR(vol_args))
1546 return PTR_ERR(vol_args);
1547 vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
1549 if (vol_args->flags &
1550 ~(BTRFS_SUBVOL_CREATE_ASYNC | BTRFS_SUBVOL_RDONLY |
1551 BTRFS_SUBVOL_QGROUP_INHERIT)) {
1556 if (vol_args->flags & BTRFS_SUBVOL_CREATE_ASYNC)
1558 if (vol_args->flags & BTRFS_SUBVOL_RDONLY)
1560 if (vol_args->flags & BTRFS_SUBVOL_QGROUP_INHERIT) {
1561 if (vol_args->size > PAGE_CACHE_SIZE) {
1565 inherit = memdup_user(vol_args->qgroup_inherit, vol_args->size);
1566 if (IS_ERR(inherit)) {
1567 ret = PTR_ERR(inherit);
1572 ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
1573 vol_args->fd, subvol, ptr,
1576 if (ret == 0 && ptr &&
1578 offsetof(struct btrfs_ioctl_vol_args_v2,
1579 transid), ptr, sizeof(*ptr)))
1587 static noinline int btrfs_ioctl_subvol_getflags(struct file *file,
1590 struct inode *inode = fdentry(file)->d_inode;
1591 struct btrfs_root *root = BTRFS_I(inode)->root;
1595 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID)
1598 down_read(&root->fs_info->subvol_sem);
1599 if (btrfs_root_readonly(root))
1600 flags |= BTRFS_SUBVOL_RDONLY;
1601 up_read(&root->fs_info->subvol_sem);
1603 if (copy_to_user(arg, &flags, sizeof(flags)))
1609 static noinline int btrfs_ioctl_subvol_setflags(struct file *file,
1612 struct inode *inode = fdentry(file)->d_inode;
1613 struct btrfs_root *root = BTRFS_I(inode)->root;
1614 struct btrfs_trans_handle *trans;
1619 ret = mnt_want_write_file(file);
1623 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) {
1625 goto out_drop_write;
1628 if (copy_from_user(&flags, arg, sizeof(flags))) {
1630 goto out_drop_write;
1633 if (flags & BTRFS_SUBVOL_CREATE_ASYNC) {
1635 goto out_drop_write;
1638 if (flags & ~BTRFS_SUBVOL_RDONLY) {
1640 goto out_drop_write;
1643 if (!inode_owner_or_capable(inode)) {
1645 goto out_drop_write;
1648 down_write(&root->fs_info->subvol_sem);
1651 if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root))
1654 root_flags = btrfs_root_flags(&root->root_item);
1655 if (flags & BTRFS_SUBVOL_RDONLY)
1656 btrfs_set_root_flags(&root->root_item,
1657 root_flags | BTRFS_ROOT_SUBVOL_RDONLY);
1659 btrfs_set_root_flags(&root->root_item,
1660 root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY);
1662 trans = btrfs_start_transaction(root, 1);
1663 if (IS_ERR(trans)) {
1664 ret = PTR_ERR(trans);
1668 ret = btrfs_update_root(trans, root->fs_info->tree_root,
1669 &root->root_key, &root->root_item);
1671 btrfs_commit_transaction(trans, root);
1674 btrfs_set_root_flags(&root->root_item, root_flags);
1676 up_write(&root->fs_info->subvol_sem);
1678 mnt_drop_write_file(file);
1684 * helper to check if the subvolume references other subvolumes
1686 static noinline int may_destroy_subvol(struct btrfs_root *root)
1688 struct btrfs_path *path;
1689 struct btrfs_key key;
1692 path = btrfs_alloc_path();
1696 key.objectid = root->root_key.objectid;
1697 key.type = BTRFS_ROOT_REF_KEY;
1698 key.offset = (u64)-1;
1700 ret = btrfs_search_slot(NULL, root->fs_info->tree_root,
1707 if (path->slots[0] > 0) {
1709 btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
1710 if (key.objectid == root->root_key.objectid &&
1711 key.type == BTRFS_ROOT_REF_KEY)
1715 btrfs_free_path(path);
1719 static noinline int key_in_sk(struct btrfs_key *key,
1720 struct btrfs_ioctl_search_key *sk)
1722 struct btrfs_key test;
1725 test.objectid = sk->min_objectid;
1726 test.type = sk->min_type;
1727 test.offset = sk->min_offset;
1729 ret = btrfs_comp_cpu_keys(key, &test);
1733 test.objectid = sk->max_objectid;
1734 test.type = sk->max_type;
1735 test.offset = sk->max_offset;
1737 ret = btrfs_comp_cpu_keys(key, &test);
1743 static noinline int copy_to_sk(struct btrfs_root *root,
1744 struct btrfs_path *path,
1745 struct btrfs_key *key,
1746 struct btrfs_ioctl_search_key *sk,
1748 unsigned long *sk_offset,
1752 struct extent_buffer *leaf;
1753 struct btrfs_ioctl_search_header sh;
1754 unsigned long item_off;
1755 unsigned long item_len;
1761 leaf = path->nodes[0];
1762 slot = path->slots[0];
1763 nritems = btrfs_header_nritems(leaf);
1765 if (btrfs_header_generation(leaf) > sk->max_transid) {
1769 found_transid = btrfs_header_generation(leaf);
1771 for (i = slot; i < nritems; i++) {
1772 item_off = btrfs_item_ptr_offset(leaf, i);
1773 item_len = btrfs_item_size_nr(leaf, i);
1775 if (item_len > BTRFS_SEARCH_ARGS_BUFSIZE)
1778 if (sizeof(sh) + item_len + *sk_offset >
1779 BTRFS_SEARCH_ARGS_BUFSIZE) {
1784 btrfs_item_key_to_cpu(leaf, key, i);
1785 if (!key_in_sk(key, sk))
1788 sh.objectid = key->objectid;
1789 sh.offset = key->offset;
1790 sh.type = key->type;
1792 sh.transid = found_transid;
1794 /* copy search result header */
1795 memcpy(buf + *sk_offset, &sh, sizeof(sh));
1796 *sk_offset += sizeof(sh);
1799 char *p = buf + *sk_offset;
1801 read_extent_buffer(leaf, p,
1802 item_off, item_len);
1803 *sk_offset += item_len;
1807 if (*num_found >= sk->nr_items)
1812 if (key->offset < (u64)-1 && key->offset < sk->max_offset)
1814 else if (key->type < (u8)-1 && key->type < sk->max_type) {
1817 } else if (key->objectid < (u64)-1 && key->objectid < sk->max_objectid) {
1827 static noinline int search_ioctl(struct inode *inode,
1828 struct btrfs_ioctl_search_args *args)
1830 struct btrfs_root *root;
1831 struct btrfs_key key;
1832 struct btrfs_key max_key;
1833 struct btrfs_path *path;
1834 struct btrfs_ioctl_search_key *sk = &args->key;
1835 struct btrfs_fs_info *info = BTRFS_I(inode)->root->fs_info;
1838 unsigned long sk_offset = 0;
1840 path = btrfs_alloc_path();
1844 if (sk->tree_id == 0) {
1845 /* search the root of the inode that was passed */
1846 root = BTRFS_I(inode)->root;
1848 key.objectid = sk->tree_id;
1849 key.type = BTRFS_ROOT_ITEM_KEY;
1850 key.offset = (u64)-1;
1851 root = btrfs_read_fs_root_no_name(info, &key);
1853 printk(KERN_ERR "could not find root %llu\n",
1855 btrfs_free_path(path);
1860 key.objectid = sk->min_objectid;
1861 key.type = sk->min_type;
1862 key.offset = sk->min_offset;
1864 max_key.objectid = sk->max_objectid;
1865 max_key.type = sk->max_type;
1866 max_key.offset = sk->max_offset;
1868 path->keep_locks = 1;
1871 ret = btrfs_search_forward(root, &key, &max_key, path,
1878 ret = copy_to_sk(root, path, &key, sk, args->buf,
1879 &sk_offset, &num_found);
1880 btrfs_release_path(path);
1881 if (ret || num_found >= sk->nr_items)
1887 sk->nr_items = num_found;
1888 btrfs_free_path(path);
1892 static noinline int btrfs_ioctl_tree_search(struct file *file,
1895 struct btrfs_ioctl_search_args *args;
1896 struct inode *inode;
1899 if (!capable(CAP_SYS_ADMIN))
1902 args = memdup_user(argp, sizeof(*args));
1904 return PTR_ERR(args);
1906 inode = fdentry(file)->d_inode;
1907 ret = search_ioctl(inode, args);
1908 if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
1915 * Search INODE_REFs to identify path name of 'dirid' directory
1916 * in a 'tree_id' tree. and sets path name to 'name'.
1918 static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info,
1919 u64 tree_id, u64 dirid, char *name)
1921 struct btrfs_root *root;
1922 struct btrfs_key key;
1928 struct btrfs_inode_ref *iref;
1929 struct extent_buffer *l;
1930 struct btrfs_path *path;
1932 if (dirid == BTRFS_FIRST_FREE_OBJECTID) {
1937 path = btrfs_alloc_path();
1941 ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX];
1943 key.objectid = tree_id;
1944 key.type = BTRFS_ROOT_ITEM_KEY;
1945 key.offset = (u64)-1;
1946 root = btrfs_read_fs_root_no_name(info, &key);
1948 printk(KERN_ERR "could not find root %llu\n", tree_id);
1953 key.objectid = dirid;
1954 key.type = BTRFS_INODE_REF_KEY;
1955 key.offset = (u64)-1;
1958 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1963 slot = path->slots[0];
1964 if (ret > 0 && slot > 0)
1966 btrfs_item_key_to_cpu(l, &key, slot);
1968 if (ret > 0 && (key.objectid != dirid ||
1969 key.type != BTRFS_INODE_REF_KEY)) {
1974 iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref);
1975 len = btrfs_inode_ref_name_len(l, iref);
1977 total_len += len + 1;
1982 read_extent_buffer(l, ptr,(unsigned long)(iref + 1), len);
1984 if (key.offset == BTRFS_FIRST_FREE_OBJECTID)
1987 btrfs_release_path(path);
1988 key.objectid = key.offset;
1989 key.offset = (u64)-1;
1990 dirid = key.objectid;
1994 memmove(name, ptr, total_len);
1995 name[total_len]='\0';
1998 btrfs_free_path(path);
2002 static noinline int btrfs_ioctl_ino_lookup(struct file *file,
2005 struct btrfs_ioctl_ino_lookup_args *args;
2006 struct inode *inode;
2009 if (!capable(CAP_SYS_ADMIN))
2012 args = memdup_user(argp, sizeof(*args));
2014 return PTR_ERR(args);
2016 inode = fdentry(file)->d_inode;
2018 if (args->treeid == 0)
2019 args->treeid = BTRFS_I(inode)->root->root_key.objectid;
2021 ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info,
2022 args->treeid, args->objectid,
2025 if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
2032 static noinline int btrfs_ioctl_snap_destroy(struct file *file,
2035 struct dentry *parent = fdentry(file);
2036 struct dentry *dentry;
2037 struct inode *dir = parent->d_inode;
2038 struct inode *inode;
2039 struct btrfs_root *root = BTRFS_I(dir)->root;
2040 struct btrfs_root *dest = NULL;
2041 struct btrfs_ioctl_vol_args *vol_args;
2042 struct btrfs_trans_handle *trans;
2047 vol_args = memdup_user(arg, sizeof(*vol_args));
2048 if (IS_ERR(vol_args))
2049 return PTR_ERR(vol_args);
2051 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2052 namelen = strlen(vol_args->name);
2053 if (strchr(vol_args->name, '/') ||
2054 strncmp(vol_args->name, "..", namelen) == 0) {
2059 err = mnt_want_write_file(file);
2063 mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
2064 dentry = lookup_one_len(vol_args->name, parent, namelen);
2065 if (IS_ERR(dentry)) {
2066 err = PTR_ERR(dentry);
2067 goto out_unlock_dir;
2070 if (!dentry->d_inode) {
2075 inode = dentry->d_inode;
2076 dest = BTRFS_I(inode)->root;
2077 if (!capable(CAP_SYS_ADMIN)){
2079 * Regular user. Only allow this with a special mount
2080 * option, when the user has write+exec access to the
2081 * subvol root, and when rmdir(2) would have been
2084 * Note that this is _not_ check that the subvol is
2085 * empty or doesn't contain data that we wouldn't
2086 * otherwise be able to delete.
2088 * Users who want to delete empty subvols should try
2092 if (!btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED))
2096 * Do not allow deletion if the parent dir is the same
2097 * as the dir to be deleted. That means the ioctl
2098 * must be called on the dentry referencing the root
2099 * of the subvol, not a random directory contained
2106 err = inode_permission(inode, MAY_WRITE | MAY_EXEC);
2111 /* check if subvolume may be deleted by a user */
2112 err = btrfs_may_delete(dir, dentry, 1);
2116 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) {
2121 mutex_lock(&inode->i_mutex);
2122 err = d_invalidate(dentry);
2126 down_write(&root->fs_info->subvol_sem);
2128 err = may_destroy_subvol(dest);
2132 trans = btrfs_start_transaction(root, 0);
2133 if (IS_ERR(trans)) {
2134 err = PTR_ERR(trans);
2137 trans->block_rsv = &root->fs_info->global_block_rsv;
2139 ret = btrfs_unlink_subvol(trans, root, dir,
2140 dest->root_key.objectid,
2141 dentry->d_name.name,
2142 dentry->d_name.len);
2145 btrfs_abort_transaction(trans, root, ret);
2149 btrfs_record_root_in_trans(trans, dest);
2151 memset(&dest->root_item.drop_progress, 0,
2152 sizeof(dest->root_item.drop_progress));
2153 dest->root_item.drop_level = 0;
2154 btrfs_set_root_refs(&dest->root_item, 0);
2156 if (!xchg(&dest->orphan_item_inserted, 1)) {
2157 ret = btrfs_insert_orphan_item(trans,
2158 root->fs_info->tree_root,
2159 dest->root_key.objectid);
2161 btrfs_abort_transaction(trans, root, ret);
2167 ret = btrfs_end_transaction(trans, root);
2170 inode->i_flags |= S_DEAD;
2172 up_write(&root->fs_info->subvol_sem);
2174 mutex_unlock(&inode->i_mutex);
2176 shrink_dcache_sb(root->fs_info->sb);
2177 btrfs_invalidate_inodes(dest);
2181 if (dest->cache_inode) {
2182 iput(dest->cache_inode);
2183 dest->cache_inode = NULL;
2189 mutex_unlock(&dir->i_mutex);
2190 mnt_drop_write_file(file);
2196 static int btrfs_ioctl_defrag(struct file *file, void __user *argp)
2198 struct inode *inode = fdentry(file)->d_inode;
2199 struct btrfs_root *root = BTRFS_I(inode)->root;
2200 struct btrfs_ioctl_defrag_range_args *range;
2203 ret = mnt_want_write_file(file);
2207 if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running,
2209 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
2210 mnt_drop_write_file(file);
2214 if (btrfs_root_readonly(root)) {
2219 switch (inode->i_mode & S_IFMT) {
2221 if (!capable(CAP_SYS_ADMIN)) {
2225 ret = btrfs_defrag_root(root);
2228 ret = btrfs_defrag_root(root->fs_info->extent_root);
2231 if (!(file->f_mode & FMODE_WRITE)) {
2236 range = kzalloc(sizeof(*range), GFP_KERNEL);
2243 if (copy_from_user(range, argp,
2249 /* compression requires us to start the IO */
2250 if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
2251 range->flags |= BTRFS_DEFRAG_RANGE_START_IO;
2252 range->extent_thresh = (u32)-1;
2255 /* the rest are all set to zero by kzalloc */
2256 range->len = (u64)-1;
2258 ret = btrfs_defrag_file(fdentry(file)->d_inode, file,
2268 atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0);
2269 mnt_drop_write_file(file);
2273 static long btrfs_ioctl_add_dev(struct btrfs_root *root, void __user *arg)
2275 struct btrfs_ioctl_vol_args *vol_args;
2278 if (!capable(CAP_SYS_ADMIN))
2281 if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running,
2283 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
2287 mutex_lock(&root->fs_info->volume_mutex);
2288 vol_args = memdup_user(arg, sizeof(*vol_args));
2289 if (IS_ERR(vol_args)) {
2290 ret = PTR_ERR(vol_args);
2294 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2295 ret = btrfs_init_new_device(root, vol_args->name);
2299 mutex_unlock(&root->fs_info->volume_mutex);
2300 atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0);
2304 static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg)
2306 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
2307 struct btrfs_ioctl_vol_args *vol_args;
2310 if (!capable(CAP_SYS_ADMIN))
2313 ret = mnt_want_write_file(file);
2317 if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running,
2319 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
2320 mnt_drop_write_file(file);
2324 mutex_lock(&root->fs_info->volume_mutex);
2325 vol_args = memdup_user(arg, sizeof(*vol_args));
2326 if (IS_ERR(vol_args)) {
2327 ret = PTR_ERR(vol_args);
2331 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2332 ret = btrfs_rm_device(root, vol_args->name);
2336 mutex_unlock(&root->fs_info->volume_mutex);
2337 atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0);
2338 mnt_drop_write_file(file);
2342 static long btrfs_ioctl_fs_info(struct btrfs_root *root, void __user *arg)
2344 struct btrfs_ioctl_fs_info_args *fi_args;
2345 struct btrfs_device *device;
2346 struct btrfs_device *next;
2347 struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
2350 if (!capable(CAP_SYS_ADMIN))
2353 fi_args = kzalloc(sizeof(*fi_args), GFP_KERNEL);
2357 fi_args->num_devices = fs_devices->num_devices;
2358 memcpy(&fi_args->fsid, root->fs_info->fsid, sizeof(fi_args->fsid));
2360 mutex_lock(&fs_devices->device_list_mutex);
2361 list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
2362 if (device->devid > fi_args->max_id)
2363 fi_args->max_id = device->devid;
2365 mutex_unlock(&fs_devices->device_list_mutex);
2367 if (copy_to_user(arg, fi_args, sizeof(*fi_args)))
2374 static long btrfs_ioctl_dev_info(struct btrfs_root *root, void __user *arg)
2376 struct btrfs_ioctl_dev_info_args *di_args;
2377 struct btrfs_device *dev;
2378 struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
2380 char *s_uuid = NULL;
2381 char empty_uuid[BTRFS_UUID_SIZE] = {0};
2383 if (!capable(CAP_SYS_ADMIN))
2386 di_args = memdup_user(arg, sizeof(*di_args));
2387 if (IS_ERR(di_args))
2388 return PTR_ERR(di_args);
2390 if (memcmp(empty_uuid, di_args->uuid, BTRFS_UUID_SIZE) != 0)
2391 s_uuid = di_args->uuid;
2393 mutex_lock(&fs_devices->device_list_mutex);
2394 dev = btrfs_find_device(root->fs_info, di_args->devid, s_uuid, NULL);
2395 mutex_unlock(&fs_devices->device_list_mutex);
2402 di_args->devid = dev->devid;
2403 di_args->bytes_used = dev->bytes_used;
2404 di_args->total_bytes = dev->total_bytes;
2405 memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid));
2407 struct rcu_string *name;
2410 name = rcu_dereference(dev->name);
2411 strncpy(di_args->path, name->str, sizeof(di_args->path));
2413 di_args->path[sizeof(di_args->path) - 1] = 0;
2415 di_args->path[0] = '\0';
2419 if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args)))
2426 static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
2427 u64 off, u64 olen, u64 destoff)
2429 struct inode *inode = fdentry(file)->d_inode;
2430 struct btrfs_root *root = BTRFS_I(inode)->root;
2433 struct btrfs_trans_handle *trans;
2434 struct btrfs_path *path;
2435 struct extent_buffer *leaf;
2437 struct btrfs_key key;
2442 u64 bs = root->fs_info->sb->s_blocksize;
2446 * - split compressed inline extents. annoying: we need to
2447 * decompress into destination's address_space (the file offset
2448 * may change, so source mapping won't do), then recompress (or
2449 * otherwise reinsert) a subrange.
2450 * - allow ranges within the same file to be cloned (provided
2451 * they don't overlap)?
2454 /* the destination must be opened for writing */
2455 if (!(file->f_mode & FMODE_WRITE) || (file->f_flags & O_APPEND))
2458 if (btrfs_root_readonly(root))
2461 ret = mnt_want_write_file(file);
2465 src_file = fdget(srcfd);
2466 if (!src_file.file) {
2468 goto out_drop_write;
2472 if (src_file.file->f_path.mnt != file->f_path.mnt)
2475 src = src_file.file->f_dentry->d_inode;
2481 /* the src must be open for reading */
2482 if (!(src_file.file->f_mode & FMODE_READ))
2485 /* don't make the dst file partly checksummed */
2486 if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) !=
2487 (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM))
2491 if (S_ISDIR(src->i_mode) || S_ISDIR(inode->i_mode))
2495 if (src->i_sb != inode->i_sb)
2499 buf = vmalloc(btrfs_level_size(root, 0));
2503 path = btrfs_alloc_path();
2511 mutex_lock_nested(&inode->i_mutex, I_MUTEX_PARENT);
2512 mutex_lock_nested(&src->i_mutex, I_MUTEX_CHILD);
2514 mutex_lock_nested(&src->i_mutex, I_MUTEX_PARENT);
2515 mutex_lock_nested(&inode->i_mutex, I_MUTEX_CHILD);
2518 /* determine range to clone */
2520 if (off + len > src->i_size || off + len < off)
2523 olen = len = src->i_size - off;
2524 /* if we extend to eof, continue to block boundary */
2525 if (off + len == src->i_size)
2526 len = ALIGN(src->i_size, bs) - off;
2528 /* verify the end result is block aligned */
2529 if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs) ||
2530 !IS_ALIGNED(destoff, bs))
2533 if (destoff > inode->i_size) {
2534 ret = btrfs_cont_expand(inode, inode->i_size, destoff);
2539 /* truncate page cache pages from target inode range */
2540 truncate_inode_pages_range(&inode->i_data, destoff,
2541 PAGE_CACHE_ALIGN(destoff + len) - 1);
2543 /* do any pending delalloc/csum calc on src, one way or
2544 another, and lock file content */
2546 struct btrfs_ordered_extent *ordered;
2547 lock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1);
2548 ordered = btrfs_lookup_first_ordered_extent(src, off + len - 1);
2550 !test_range_bit(&BTRFS_I(src)->io_tree, off, off + len - 1,
2551 EXTENT_DELALLOC, 0, NULL))
2553 unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1);
2555 btrfs_put_ordered_extent(ordered);
2556 btrfs_wait_ordered_range(src, off, len);
2560 key.objectid = btrfs_ino(src);
2561 key.type = BTRFS_EXTENT_DATA_KEY;
2566 * note the key will change type as we walk through the
2569 ret = btrfs_search_slot(NULL, BTRFS_I(src)->root, &key, path,
2574 nritems = btrfs_header_nritems(path->nodes[0]);
2575 if (path->slots[0] >= nritems) {
2576 ret = btrfs_next_leaf(BTRFS_I(src)->root, path);
2581 nritems = btrfs_header_nritems(path->nodes[0]);
2583 leaf = path->nodes[0];
2584 slot = path->slots[0];
2586 btrfs_item_key_to_cpu(leaf, &key, slot);
2587 if (btrfs_key_type(&key) > BTRFS_EXTENT_DATA_KEY ||
2588 key.objectid != btrfs_ino(src))
2591 if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) {
2592 struct btrfs_file_extent_item *extent;
2595 struct btrfs_key new_key;
2596 u64 disko = 0, diskl = 0;
2597 u64 datao = 0, datal = 0;
2601 size = btrfs_item_size_nr(leaf, slot);
2602 read_extent_buffer(leaf, buf,
2603 btrfs_item_ptr_offset(leaf, slot),
2606 extent = btrfs_item_ptr(leaf, slot,
2607 struct btrfs_file_extent_item);
2608 comp = btrfs_file_extent_compression(leaf, extent);
2609 type = btrfs_file_extent_type(leaf, extent);
2610 if (type == BTRFS_FILE_EXTENT_REG ||
2611 type == BTRFS_FILE_EXTENT_PREALLOC) {
2612 disko = btrfs_file_extent_disk_bytenr(leaf,
2614 diskl = btrfs_file_extent_disk_num_bytes(leaf,
2616 datao = btrfs_file_extent_offset(leaf, extent);
2617 datal = btrfs_file_extent_num_bytes(leaf,
2619 } else if (type == BTRFS_FILE_EXTENT_INLINE) {
2620 /* take upper bound, may be compressed */
2621 datal = btrfs_file_extent_ram_bytes(leaf,
2624 btrfs_release_path(path);
2626 if (key.offset + datal <= off ||
2627 key.offset >= off + len - 1)
2630 memcpy(&new_key, &key, sizeof(new_key));
2631 new_key.objectid = btrfs_ino(inode);
2632 if (off <= key.offset)
2633 new_key.offset = key.offset + destoff - off;
2635 new_key.offset = destoff;
2638 * 1 - adjusting old extent (we may have to split it)
2639 * 1 - add new extent
2642 trans = btrfs_start_transaction(root, 3);
2643 if (IS_ERR(trans)) {
2644 ret = PTR_ERR(trans);
2648 if (type == BTRFS_FILE_EXTENT_REG ||
2649 type == BTRFS_FILE_EXTENT_PREALLOC) {
2651 * a | --- range to clone ---| b
2652 * | ------------- extent ------------- |
2655 /* substract range b */
2656 if (key.offset + datal > off + len)
2657 datal = off + len - key.offset;
2659 /* substract range a */
2660 if (off > key.offset) {
2661 datao += off - key.offset;
2662 datal -= off - key.offset;
2665 ret = btrfs_drop_extents(trans, root, inode,
2667 new_key.offset + datal,
2670 btrfs_abort_transaction(trans, root,
2672 btrfs_end_transaction(trans, root);
2676 ret = btrfs_insert_empty_item(trans, root, path,
2679 btrfs_abort_transaction(trans, root,
2681 btrfs_end_transaction(trans, root);
2685 leaf = path->nodes[0];
2686 slot = path->slots[0];
2687 write_extent_buffer(leaf, buf,
2688 btrfs_item_ptr_offset(leaf, slot),
2691 extent = btrfs_item_ptr(leaf, slot,
2692 struct btrfs_file_extent_item);
2694 /* disko == 0 means it's a hole */
2698 btrfs_set_file_extent_offset(leaf, extent,
2700 btrfs_set_file_extent_num_bytes(leaf, extent,
2703 inode_add_bytes(inode, datal);
2704 ret = btrfs_inc_extent_ref(trans, root,
2706 root->root_key.objectid,
2708 new_key.offset - datao,
2711 btrfs_abort_transaction(trans,
2714 btrfs_end_transaction(trans,
2720 } else if (type == BTRFS_FILE_EXTENT_INLINE) {
2723 if (off > key.offset) {
2724 skip = off - key.offset;
2725 new_key.offset += skip;
2728 if (key.offset + datal > off + len)
2729 trim = key.offset + datal - (off + len);
2731 if (comp && (skip || trim)) {
2733 btrfs_end_transaction(trans, root);
2736 size -= skip + trim;
2737 datal -= skip + trim;
2739 ret = btrfs_drop_extents(trans, root, inode,
2741 new_key.offset + datal,
2744 btrfs_abort_transaction(trans, root,
2746 btrfs_end_transaction(trans, root);
2750 ret = btrfs_insert_empty_item(trans, root, path,
2753 btrfs_abort_transaction(trans, root,
2755 btrfs_end_transaction(trans, root);
2761 btrfs_file_extent_calc_inline_size(0);
2762 memmove(buf+start, buf+start+skip,
2766 leaf = path->nodes[0];
2767 slot = path->slots[0];
2768 write_extent_buffer(leaf, buf,
2769 btrfs_item_ptr_offset(leaf, slot),
2771 inode_add_bytes(inode, datal);
2774 btrfs_mark_buffer_dirty(leaf);
2775 btrfs_release_path(path);
2777 inode_inc_iversion(inode);
2778 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
2781 * we round up to the block size at eof when
2782 * determining which extents to clone above,
2783 * but shouldn't round up the file size
2785 endoff = new_key.offset + datal;
2786 if (endoff > destoff+olen)
2787 endoff = destoff+olen;
2788 if (endoff > inode->i_size)
2789 btrfs_i_size_write(inode, endoff);
2791 ret = btrfs_update_inode(trans, root, inode);
2793 btrfs_abort_transaction(trans, root, ret);
2794 btrfs_end_transaction(trans, root);
2797 ret = btrfs_end_transaction(trans, root);
2800 btrfs_release_path(path);
2805 btrfs_release_path(path);
2806 unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1);
2808 mutex_unlock(&src->i_mutex);
2809 mutex_unlock(&inode->i_mutex);
2811 btrfs_free_path(path);
2815 mnt_drop_write_file(file);
2819 static long btrfs_ioctl_clone_range(struct file *file, void __user *argp)
2821 struct btrfs_ioctl_clone_range_args args;
2823 if (copy_from_user(&args, argp, sizeof(args)))
2825 return btrfs_ioctl_clone(file, args.src_fd, args.src_offset,
2826 args.src_length, args.dest_offset);
2830 * there are many ways the trans_start and trans_end ioctls can lead
2831 * to deadlocks. They should only be used by applications that
2832 * basically own the machine, and have a very in depth understanding
2833 * of all the possible deadlocks and enospc problems.
2835 static long btrfs_ioctl_trans_start(struct file *file)
2837 struct inode *inode = fdentry(file)->d_inode;
2838 struct btrfs_root *root = BTRFS_I(inode)->root;
2839 struct btrfs_trans_handle *trans;
2843 if (!capable(CAP_SYS_ADMIN))
2847 if (file->private_data)
2851 if (btrfs_root_readonly(root))
2854 ret = mnt_want_write_file(file);
2858 atomic_inc(&root->fs_info->open_ioctl_trans);
2861 trans = btrfs_start_ioctl_transaction(root);
2865 file->private_data = trans;
2869 atomic_dec(&root->fs_info->open_ioctl_trans);
2870 mnt_drop_write_file(file);
2875 static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp)
2877 struct inode *inode = fdentry(file)->d_inode;
2878 struct btrfs_root *root = BTRFS_I(inode)->root;
2879 struct btrfs_root *new_root;
2880 struct btrfs_dir_item *di;
2881 struct btrfs_trans_handle *trans;
2882 struct btrfs_path *path;
2883 struct btrfs_key location;
2884 struct btrfs_disk_key disk_key;
2889 if (!capable(CAP_SYS_ADMIN))
2892 ret = mnt_want_write_file(file);
2896 if (copy_from_user(&objectid, argp, sizeof(objectid))) {
2902 objectid = root->root_key.objectid;
2904 location.objectid = objectid;
2905 location.type = BTRFS_ROOT_ITEM_KEY;
2906 location.offset = (u64)-1;
2908 new_root = btrfs_read_fs_root_no_name(root->fs_info, &location);
2909 if (IS_ERR(new_root)) {
2910 ret = PTR_ERR(new_root);
2914 if (btrfs_root_refs(&new_root->root_item) == 0) {
2919 path = btrfs_alloc_path();
2924 path->leave_spinning = 1;
2926 trans = btrfs_start_transaction(root, 1);
2927 if (IS_ERR(trans)) {
2928 btrfs_free_path(path);
2929 ret = PTR_ERR(trans);
2933 dir_id = btrfs_super_root_dir(root->fs_info->super_copy);
2934 di = btrfs_lookup_dir_item(trans, root->fs_info->tree_root, path,
2935 dir_id, "default", 7, 1);
2936 if (IS_ERR_OR_NULL(di)) {
2937 btrfs_free_path(path);
2938 btrfs_end_transaction(trans, root);
2939 printk(KERN_ERR "Umm, you don't have the default dir item, "
2940 "this isn't going to work\n");
2945 btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key);
2946 btrfs_set_dir_item_key(path->nodes[0], di, &disk_key);
2947 btrfs_mark_buffer_dirty(path->nodes[0]);
2948 btrfs_free_path(path);
2950 btrfs_set_fs_incompat(root->fs_info, DEFAULT_SUBVOL);
2951 btrfs_end_transaction(trans, root);
2953 mnt_drop_write_file(file);
2957 void btrfs_get_block_group_info(struct list_head *groups_list,
2958 struct btrfs_ioctl_space_info *space)
2960 struct btrfs_block_group_cache *block_group;
2962 space->total_bytes = 0;
2963 space->used_bytes = 0;
2965 list_for_each_entry(block_group, groups_list, list) {
2966 space->flags = block_group->flags;
2967 space->total_bytes += block_group->key.offset;
2968 space->used_bytes +=
2969 btrfs_block_group_used(&block_group->item);
2973 long btrfs_ioctl_space_info(struct btrfs_root *root, void __user *arg)
2975 struct btrfs_ioctl_space_args space_args;
2976 struct btrfs_ioctl_space_info space;
2977 struct btrfs_ioctl_space_info *dest;
2978 struct btrfs_ioctl_space_info *dest_orig;
2979 struct btrfs_ioctl_space_info __user *user_dest;
2980 struct btrfs_space_info *info;
2981 u64 types[] = {BTRFS_BLOCK_GROUP_DATA,
2982 BTRFS_BLOCK_GROUP_SYSTEM,
2983 BTRFS_BLOCK_GROUP_METADATA,
2984 BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA};
2991 if (copy_from_user(&space_args,
2992 (struct btrfs_ioctl_space_args __user *)arg,
2993 sizeof(space_args)))
2996 for (i = 0; i < num_types; i++) {
2997 struct btrfs_space_info *tmp;
3001 list_for_each_entry_rcu(tmp, &root->fs_info->space_info,
3003 if (tmp->flags == types[i]) {
3013 down_read(&info->groups_sem);
3014 for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
3015 if (!list_empty(&info->block_groups[c]))
3018 up_read(&info->groups_sem);
3021 /* space_slots == 0 means they are asking for a count */
3022 if (space_args.space_slots == 0) {
3023 space_args.total_spaces = slot_count;
3027 slot_count = min_t(u64, space_args.space_slots, slot_count);
3029 alloc_size = sizeof(*dest) * slot_count;
3031 /* we generally have at most 6 or so space infos, one for each raid
3032 * level. So, a whole page should be more than enough for everyone
3034 if (alloc_size > PAGE_CACHE_SIZE)
3037 space_args.total_spaces = 0;
3038 dest = kmalloc(alloc_size, GFP_NOFS);
3043 /* now we have a buffer to copy into */
3044 for (i = 0; i < num_types; i++) {
3045 struct btrfs_space_info *tmp;
3052 list_for_each_entry_rcu(tmp, &root->fs_info->space_info,
3054 if (tmp->flags == types[i]) {
3063 down_read(&info->groups_sem);
3064 for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
3065 if (!list_empty(&info->block_groups[c])) {
3066 btrfs_get_block_group_info(
3067 &info->block_groups[c], &space);
3068 memcpy(dest, &space, sizeof(space));
3070 space_args.total_spaces++;
3076 up_read(&info->groups_sem);
3079 user_dest = (struct btrfs_ioctl_space_info __user *)
3080 (arg + sizeof(struct btrfs_ioctl_space_args));
3082 if (copy_to_user(user_dest, dest_orig, alloc_size))
3087 if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args)))
3094 * there are many ways the trans_start and trans_end ioctls can lead
3095 * to deadlocks. They should only be used by applications that
3096 * basically own the machine, and have a very in depth understanding
3097 * of all the possible deadlocks and enospc problems.
3099 long btrfs_ioctl_trans_end(struct file *file)
3101 struct inode *inode = fdentry(file)->d_inode;
3102 struct btrfs_root *root = BTRFS_I(inode)->root;
3103 struct btrfs_trans_handle *trans;
3105 trans = file->private_data;
3108 file->private_data = NULL;
3110 btrfs_end_transaction(trans, root);
3112 atomic_dec(&root->fs_info->open_ioctl_trans);
3114 mnt_drop_write_file(file);
3118 static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root,
3121 struct btrfs_trans_handle *trans;
3125 trans = btrfs_attach_transaction_barrier(root);
3126 if (IS_ERR(trans)) {
3127 if (PTR_ERR(trans) != -ENOENT)
3128 return PTR_ERR(trans);
3130 /* No running transaction, don't bother */
3131 transid = root->fs_info->last_trans_committed;
3134 transid = trans->transid;
3135 ret = btrfs_commit_transaction_async(trans, root, 0);
3137 btrfs_end_transaction(trans, root);
3142 if (copy_to_user(argp, &transid, sizeof(transid)))
3147 static noinline long btrfs_ioctl_wait_sync(struct btrfs_root *root,
3153 if (copy_from_user(&transid, argp, sizeof(transid)))
3156 transid = 0; /* current trans */
3158 return btrfs_wait_for_commit(root, transid);
3161 static long btrfs_ioctl_scrub(struct file *file, void __user *arg)
3163 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
3164 struct btrfs_ioctl_scrub_args *sa;
3167 if (!capable(CAP_SYS_ADMIN))
3170 sa = memdup_user(arg, sizeof(*sa));
3174 if (!(sa->flags & BTRFS_SCRUB_READONLY)) {
3175 ret = mnt_want_write_file(file);
3180 ret = btrfs_scrub_dev(root->fs_info, sa->devid, sa->start, sa->end,
3181 &sa->progress, sa->flags & BTRFS_SCRUB_READONLY,
3184 if (copy_to_user(arg, sa, sizeof(*sa)))
3187 if (!(sa->flags & BTRFS_SCRUB_READONLY))
3188 mnt_drop_write_file(file);
3194 static long btrfs_ioctl_scrub_cancel(struct btrfs_root *root, void __user *arg)
3196 if (!capable(CAP_SYS_ADMIN))
3199 return btrfs_scrub_cancel(root->fs_info);
3202 static long btrfs_ioctl_scrub_progress(struct btrfs_root *root,
3205 struct btrfs_ioctl_scrub_args *sa;
3208 if (!capable(CAP_SYS_ADMIN))
3211 sa = memdup_user(arg, sizeof(*sa));
3215 ret = btrfs_scrub_progress(root, sa->devid, &sa->progress);
3217 if (copy_to_user(arg, sa, sizeof(*sa)))
3224 static long btrfs_ioctl_get_dev_stats(struct btrfs_root *root,
3227 struct btrfs_ioctl_get_dev_stats *sa;
3230 sa = memdup_user(arg, sizeof(*sa));
3234 if ((sa->flags & BTRFS_DEV_STATS_RESET) && !capable(CAP_SYS_ADMIN)) {
3239 ret = btrfs_get_dev_stats(root, sa);
3241 if (copy_to_user(arg, sa, sizeof(*sa)))
3248 static long btrfs_ioctl_dev_replace(struct btrfs_root *root, void __user *arg)
3250 struct btrfs_ioctl_dev_replace_args *p;
3253 if (!capable(CAP_SYS_ADMIN))
3256 p = memdup_user(arg, sizeof(*p));
3261 case BTRFS_IOCTL_DEV_REPLACE_CMD_START:
3263 &root->fs_info->mutually_exclusive_operation_running,
3265 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
3268 ret = btrfs_dev_replace_start(root, p);
3270 &root->fs_info->mutually_exclusive_operation_running,
3274 case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS:
3275 btrfs_dev_replace_status(root->fs_info, p);
3278 case BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL:
3279 ret = btrfs_dev_replace_cancel(root->fs_info, p);
3286 if (copy_to_user(arg, p, sizeof(*p)))
3293 static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg)
3299 struct btrfs_ioctl_ino_path_args *ipa = NULL;
3300 struct inode_fs_paths *ipath = NULL;
3301 struct btrfs_path *path;
3303 if (!capable(CAP_DAC_READ_SEARCH))
3306 path = btrfs_alloc_path();
3312 ipa = memdup_user(arg, sizeof(*ipa));
3319 size = min_t(u32, ipa->size, 4096);
3320 ipath = init_ipath(size, root, path);
3321 if (IS_ERR(ipath)) {
3322 ret = PTR_ERR(ipath);
3327 ret = paths_from_inode(ipa->inum, ipath);
3331 for (i = 0; i < ipath->fspath->elem_cnt; ++i) {
3332 rel_ptr = ipath->fspath->val[i] -
3333 (u64)(unsigned long)ipath->fspath->val;
3334 ipath->fspath->val[i] = rel_ptr;
3337 ret = copy_to_user((void *)(unsigned long)ipa->fspath,
3338 (void *)(unsigned long)ipath->fspath, size);
3345 btrfs_free_path(path);
3352 static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx)
3354 struct btrfs_data_container *inodes = ctx;
3355 const size_t c = 3 * sizeof(u64);
3357 if (inodes->bytes_left >= c) {
3358 inodes->bytes_left -= c;
3359 inodes->val[inodes->elem_cnt] = inum;
3360 inodes->val[inodes->elem_cnt + 1] = offset;
3361 inodes->val[inodes->elem_cnt + 2] = root;
3362 inodes->elem_cnt += 3;
3364 inodes->bytes_missing += c - inodes->bytes_left;
3365 inodes->bytes_left = 0;
3366 inodes->elem_missed += 3;
3372 static long btrfs_ioctl_logical_to_ino(struct btrfs_root *root,
3377 struct btrfs_ioctl_logical_ino_args *loi;
3378 struct btrfs_data_container *inodes = NULL;
3379 struct btrfs_path *path = NULL;
3381 if (!capable(CAP_SYS_ADMIN))
3384 loi = memdup_user(arg, sizeof(*loi));
3391 path = btrfs_alloc_path();
3397 size = min_t(u32, loi->size, 64 * 1024);
3398 inodes = init_data_container(size);
3399 if (IS_ERR(inodes)) {
3400 ret = PTR_ERR(inodes);
3405 ret = iterate_inodes_from_logical(loi->logical, root->fs_info, path,
3406 build_ino_list, inodes);
3412 ret = copy_to_user((void *)(unsigned long)loi->inodes,
3413 (void *)(unsigned long)inodes, size);
3418 btrfs_free_path(path);
3425 void update_ioctl_balance_args(struct btrfs_fs_info *fs_info, int lock,
3426 struct btrfs_ioctl_balance_args *bargs)
3428 struct btrfs_balance_control *bctl = fs_info->balance_ctl;
3430 bargs->flags = bctl->flags;
3432 if (atomic_read(&fs_info->balance_running))
3433 bargs->state |= BTRFS_BALANCE_STATE_RUNNING;
3434 if (atomic_read(&fs_info->balance_pause_req))
3435 bargs->state |= BTRFS_BALANCE_STATE_PAUSE_REQ;
3436 if (atomic_read(&fs_info->balance_cancel_req))
3437 bargs->state |= BTRFS_BALANCE_STATE_CANCEL_REQ;
3439 memcpy(&bargs->data, &bctl->data, sizeof(bargs->data));
3440 memcpy(&bargs->meta, &bctl->meta, sizeof(bargs->meta));
3441 memcpy(&bargs->sys, &bctl->sys, sizeof(bargs->sys));
3444 spin_lock(&fs_info->balance_lock);
3445 memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat));
3446 spin_unlock(&fs_info->balance_lock);
3448 memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat));
3452 static long btrfs_ioctl_balance(struct file *file, void __user *arg)
3454 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
3455 struct btrfs_fs_info *fs_info = root->fs_info;
3456 struct btrfs_ioctl_balance_args *bargs;
3457 struct btrfs_balance_control *bctl;
3458 bool need_unlock; /* for mut. excl. ops lock */
3461 if (!capable(CAP_SYS_ADMIN))
3464 ret = mnt_want_write_file(file);
3469 if (!atomic_xchg(&fs_info->mutually_exclusive_operation_running, 1)) {
3470 mutex_lock(&fs_info->volume_mutex);
3471 mutex_lock(&fs_info->balance_mutex);
3477 * mut. excl. ops lock is locked. Three possibilites:
3478 * (1) some other op is running
3479 * (2) balance is running
3480 * (3) balance is paused -- special case (think resume)
3482 mutex_lock(&fs_info->balance_mutex);
3483 if (fs_info->balance_ctl) {
3484 /* this is either (2) or (3) */
3485 if (!atomic_read(&fs_info->balance_running)) {
3486 mutex_unlock(&fs_info->balance_mutex);
3487 if (!mutex_trylock(&fs_info->volume_mutex))
3489 mutex_lock(&fs_info->balance_mutex);
3491 if (fs_info->balance_ctl &&
3492 !atomic_read(&fs_info->balance_running)) {
3494 need_unlock = false;
3498 mutex_unlock(&fs_info->balance_mutex);
3499 mutex_unlock(&fs_info->volume_mutex);
3503 mutex_unlock(&fs_info->balance_mutex);
3509 mutex_unlock(&fs_info->balance_mutex);
3510 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
3516 BUG_ON(!atomic_read(&fs_info->mutually_exclusive_operation_running));
3519 bargs = memdup_user(arg, sizeof(*bargs));
3520 if (IS_ERR(bargs)) {
3521 ret = PTR_ERR(bargs);
3525 if (bargs->flags & BTRFS_BALANCE_RESUME) {
3526 if (!fs_info->balance_ctl) {
3531 bctl = fs_info->balance_ctl;
3532 spin_lock(&fs_info->balance_lock);
3533 bctl->flags |= BTRFS_BALANCE_RESUME;
3534 spin_unlock(&fs_info->balance_lock);
3542 if (fs_info->balance_ctl) {
3547 bctl = kzalloc(sizeof(*bctl), GFP_NOFS);
3553 bctl->fs_info = fs_info;
3555 memcpy(&bctl->data, &bargs->data, sizeof(bctl->data));
3556 memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta));
3557 memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys));
3559 bctl->flags = bargs->flags;
3561 /* balance everything - no filters */
3562 bctl->flags |= BTRFS_BALANCE_TYPE_MASK;
3567 * Ownership of bctl and mutually_exclusive_operation_running
3568 * goes to to btrfs_balance. bctl is freed in __cancel_balance,
3569 * or, if restriper was paused all the way until unmount, in
3570 * free_fs_info. mutually_exclusive_operation_running is
3571 * cleared in __cancel_balance.
3573 need_unlock = false;
3575 ret = btrfs_balance(bctl, bargs);
3578 if (copy_to_user(arg, bargs, sizeof(*bargs)))
3585 mutex_unlock(&fs_info->balance_mutex);
3586 mutex_unlock(&fs_info->volume_mutex);
3588 atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
3590 mnt_drop_write_file(file);
3594 static long btrfs_ioctl_balance_ctl(struct btrfs_root *root, int cmd)
3596 if (!capable(CAP_SYS_ADMIN))
3600 case BTRFS_BALANCE_CTL_PAUSE:
3601 return btrfs_pause_balance(root->fs_info);
3602 case BTRFS_BALANCE_CTL_CANCEL:
3603 return btrfs_cancel_balance(root->fs_info);
3609 static long btrfs_ioctl_balance_progress(struct btrfs_root *root,
3612 struct btrfs_fs_info *fs_info = root->fs_info;
3613 struct btrfs_ioctl_balance_args *bargs;
3616 if (!capable(CAP_SYS_ADMIN))
3619 mutex_lock(&fs_info->balance_mutex);
3620 if (!fs_info->balance_ctl) {
3625 bargs = kzalloc(sizeof(*bargs), GFP_NOFS);
3631 update_ioctl_balance_args(fs_info, 1, bargs);
3633 if (copy_to_user(arg, bargs, sizeof(*bargs)))
3638 mutex_unlock(&fs_info->balance_mutex);
3642 static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg)
3644 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
3645 struct btrfs_ioctl_quota_ctl_args *sa;
3646 struct btrfs_trans_handle *trans = NULL;
3650 if (!capable(CAP_SYS_ADMIN))
3653 ret = mnt_want_write_file(file);
3657 sa = memdup_user(arg, sizeof(*sa));
3663 if (sa->cmd != BTRFS_QUOTA_CTL_RESCAN) {
3664 trans = btrfs_start_transaction(root, 2);
3665 if (IS_ERR(trans)) {
3666 ret = PTR_ERR(trans);
3672 case BTRFS_QUOTA_CTL_ENABLE:
3673 ret = btrfs_quota_enable(trans, root->fs_info);
3675 case BTRFS_QUOTA_CTL_DISABLE:
3676 ret = btrfs_quota_disable(trans, root->fs_info);
3678 case BTRFS_QUOTA_CTL_RESCAN:
3679 ret = btrfs_quota_rescan(root->fs_info);
3686 if (copy_to_user(arg, sa, sizeof(*sa)))
3690 err = btrfs_commit_transaction(trans, root);
3697 mnt_drop_write_file(file);
3701 static long btrfs_ioctl_qgroup_assign(struct file *file, void __user *arg)
3703 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
3704 struct btrfs_ioctl_qgroup_assign_args *sa;
3705 struct btrfs_trans_handle *trans;
3709 if (!capable(CAP_SYS_ADMIN))
3712 ret = mnt_want_write_file(file);
3716 sa = memdup_user(arg, sizeof(*sa));
3722 trans = btrfs_join_transaction(root);
3723 if (IS_ERR(trans)) {
3724 ret = PTR_ERR(trans);
3728 /* FIXME: check if the IDs really exist */
3730 ret = btrfs_add_qgroup_relation(trans, root->fs_info,
3733 ret = btrfs_del_qgroup_relation(trans, root->fs_info,
3737 err = btrfs_end_transaction(trans, root);
3744 mnt_drop_write_file(file);
3748 static long btrfs_ioctl_qgroup_create(struct file *file, void __user *arg)
3750 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
3751 struct btrfs_ioctl_qgroup_create_args *sa;
3752 struct btrfs_trans_handle *trans;
3756 if (!capable(CAP_SYS_ADMIN))
3759 ret = mnt_want_write_file(file);
3763 sa = memdup_user(arg, sizeof(*sa));
3769 if (!sa->qgroupid) {
3774 trans = btrfs_join_transaction(root);
3775 if (IS_ERR(trans)) {
3776 ret = PTR_ERR(trans);
3780 /* FIXME: check if the IDs really exist */
3782 ret = btrfs_create_qgroup(trans, root->fs_info, sa->qgroupid,
3785 ret = btrfs_remove_qgroup(trans, root->fs_info, sa->qgroupid);
3788 err = btrfs_end_transaction(trans, root);
3795 mnt_drop_write_file(file);
3799 static long btrfs_ioctl_qgroup_limit(struct file *file, void __user *arg)
3801 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
3802 struct btrfs_ioctl_qgroup_limit_args *sa;
3803 struct btrfs_trans_handle *trans;
3808 if (!capable(CAP_SYS_ADMIN))
3811 ret = mnt_want_write_file(file);
3815 sa = memdup_user(arg, sizeof(*sa));
3821 trans = btrfs_join_transaction(root);
3822 if (IS_ERR(trans)) {
3823 ret = PTR_ERR(trans);
3827 qgroupid = sa->qgroupid;
3829 /* take the current subvol as qgroup */
3830 qgroupid = root->root_key.objectid;
3833 /* FIXME: check if the IDs really exist */
3834 ret = btrfs_limit_qgroup(trans, root->fs_info, qgroupid, &sa->lim);
3836 err = btrfs_end_transaction(trans, root);
3843 mnt_drop_write_file(file);
3847 static long btrfs_ioctl_set_received_subvol(struct file *file,
3850 struct btrfs_ioctl_received_subvol_args *sa = NULL;
3851 struct inode *inode = fdentry(file)->d_inode;
3852 struct btrfs_root *root = BTRFS_I(inode)->root;
3853 struct btrfs_root_item *root_item = &root->root_item;
3854 struct btrfs_trans_handle *trans;
3855 struct timespec ct = CURRENT_TIME;
3858 ret = mnt_want_write_file(file);
3862 down_write(&root->fs_info->subvol_sem);
3864 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) {
3869 if (btrfs_root_readonly(root)) {
3874 if (!inode_owner_or_capable(inode)) {
3879 sa = memdup_user(arg, sizeof(*sa));
3886 trans = btrfs_start_transaction(root, 1);
3887 if (IS_ERR(trans)) {
3888 ret = PTR_ERR(trans);
3893 sa->rtransid = trans->transid;
3894 sa->rtime.sec = ct.tv_sec;
3895 sa->rtime.nsec = ct.tv_nsec;
3897 memcpy(root_item->received_uuid, sa->uuid, BTRFS_UUID_SIZE);
3898 btrfs_set_root_stransid(root_item, sa->stransid);
3899 btrfs_set_root_rtransid(root_item, sa->rtransid);
3900 root_item->stime.sec = cpu_to_le64(sa->stime.sec);
3901 root_item->stime.nsec = cpu_to_le32(sa->stime.nsec);
3902 root_item->rtime.sec = cpu_to_le64(sa->rtime.sec);
3903 root_item->rtime.nsec = cpu_to_le32(sa->rtime.nsec);
3905 ret = btrfs_update_root(trans, root->fs_info->tree_root,
3906 &root->root_key, &root->root_item);
3908 btrfs_end_transaction(trans, root);
3912 ret = btrfs_commit_transaction(trans, root);
3917 ret = copy_to_user(arg, sa, sizeof(*sa));
3923 up_write(&root->fs_info->subvol_sem);
3924 mnt_drop_write_file(file);
3928 static int btrfs_ioctl_get_fslabel(struct file *file, void __user *arg)
3930 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
3931 const char *label = root->fs_info->super_copy->label;
3932 size_t len = strnlen(label, BTRFS_LABEL_SIZE);
3935 if (len == BTRFS_LABEL_SIZE) {
3936 pr_warn("btrfs: label is too long, return the first %zu bytes\n",
3940 mutex_lock(&root->fs_info->volume_mutex);
3941 ret = copy_to_user(arg, label, len);
3942 mutex_unlock(&root->fs_info->volume_mutex);
3944 return ret ? -EFAULT : 0;
3947 static int btrfs_ioctl_set_fslabel(struct file *file, void __user *arg)
3949 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
3950 struct btrfs_super_block *super_block = root->fs_info->super_copy;
3951 struct btrfs_trans_handle *trans;
3952 char label[BTRFS_LABEL_SIZE];
3955 if (!capable(CAP_SYS_ADMIN))
3958 if (copy_from_user(label, arg, sizeof(label)))
3961 if (strnlen(label, BTRFS_LABEL_SIZE) == BTRFS_LABEL_SIZE) {
3962 pr_err("btrfs: unable to set label with more than %d bytes\n",
3963 BTRFS_LABEL_SIZE - 1);
3967 ret = mnt_want_write_file(file);
3971 mutex_lock(&root->fs_info->volume_mutex);
3972 trans = btrfs_start_transaction(root, 0);
3973 if (IS_ERR(trans)) {
3974 ret = PTR_ERR(trans);
3978 strcpy(super_block->label, label);
3979 ret = btrfs_end_transaction(trans, root);
3982 mutex_unlock(&root->fs_info->volume_mutex);
3983 mnt_drop_write_file(file);
3987 long btrfs_ioctl(struct file *file, unsigned int
3988 cmd, unsigned long arg)
3990 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
3991 void __user *argp = (void __user *)arg;
3994 case FS_IOC_GETFLAGS:
3995 return btrfs_ioctl_getflags(file, argp);
3996 case FS_IOC_SETFLAGS:
3997 return btrfs_ioctl_setflags(file, argp);
3998 case FS_IOC_GETVERSION:
3999 return btrfs_ioctl_getversion(file, argp);
4001 return btrfs_ioctl_fitrim(file, argp);
4002 case BTRFS_IOC_SNAP_CREATE:
4003 return btrfs_ioctl_snap_create(file, argp, 0);
4004 case BTRFS_IOC_SNAP_CREATE_V2:
4005 return btrfs_ioctl_snap_create_v2(file, argp, 0);
4006 case BTRFS_IOC_SUBVOL_CREATE:
4007 return btrfs_ioctl_snap_create(file, argp, 1);
4008 case BTRFS_IOC_SUBVOL_CREATE_V2:
4009 return btrfs_ioctl_snap_create_v2(file, argp, 1);
4010 case BTRFS_IOC_SNAP_DESTROY:
4011 return btrfs_ioctl_snap_destroy(file, argp);
4012 case BTRFS_IOC_SUBVOL_GETFLAGS:
4013 return btrfs_ioctl_subvol_getflags(file, argp);
4014 case BTRFS_IOC_SUBVOL_SETFLAGS:
4015 return btrfs_ioctl_subvol_setflags(file, argp);
4016 case BTRFS_IOC_DEFAULT_SUBVOL:
4017 return btrfs_ioctl_default_subvol(file, argp);
4018 case BTRFS_IOC_DEFRAG:
4019 return btrfs_ioctl_defrag(file, NULL);
4020 case BTRFS_IOC_DEFRAG_RANGE:
4021 return btrfs_ioctl_defrag(file, argp);
4022 case BTRFS_IOC_RESIZE:
4023 return btrfs_ioctl_resize(file, argp);
4024 case BTRFS_IOC_ADD_DEV:
4025 return btrfs_ioctl_add_dev(root, argp);
4026 case BTRFS_IOC_RM_DEV:
4027 return btrfs_ioctl_rm_dev(file, argp);
4028 case BTRFS_IOC_FS_INFO:
4029 return btrfs_ioctl_fs_info(root, argp);
4030 case BTRFS_IOC_DEV_INFO:
4031 return btrfs_ioctl_dev_info(root, argp);
4032 case BTRFS_IOC_BALANCE:
4033 return btrfs_ioctl_balance(file, NULL);
4034 case BTRFS_IOC_CLONE:
4035 return btrfs_ioctl_clone(file, arg, 0, 0, 0);
4036 case BTRFS_IOC_CLONE_RANGE:
4037 return btrfs_ioctl_clone_range(file, argp);
4038 case BTRFS_IOC_TRANS_START:
4039 return btrfs_ioctl_trans_start(file);
4040 case BTRFS_IOC_TRANS_END:
4041 return btrfs_ioctl_trans_end(file);
4042 case BTRFS_IOC_TREE_SEARCH:
4043 return btrfs_ioctl_tree_search(file, argp);
4044 case BTRFS_IOC_INO_LOOKUP:
4045 return btrfs_ioctl_ino_lookup(file, argp);
4046 case BTRFS_IOC_INO_PATHS:
4047 return btrfs_ioctl_ino_to_path(root, argp);
4048 case BTRFS_IOC_LOGICAL_INO:
4049 return btrfs_ioctl_logical_to_ino(root, argp);
4050 case BTRFS_IOC_SPACE_INFO:
4051 return btrfs_ioctl_space_info(root, argp);
4052 case BTRFS_IOC_SYNC:
4053 btrfs_sync_fs(file->f_dentry->d_sb, 1);
4055 case BTRFS_IOC_START_SYNC:
4056 return btrfs_ioctl_start_sync(root, argp);
4057 case BTRFS_IOC_WAIT_SYNC:
4058 return btrfs_ioctl_wait_sync(root, argp);
4059 case BTRFS_IOC_SCRUB:
4060 return btrfs_ioctl_scrub(file, argp);
4061 case BTRFS_IOC_SCRUB_CANCEL:
4062 return btrfs_ioctl_scrub_cancel(root, argp);
4063 case BTRFS_IOC_SCRUB_PROGRESS:
4064 return btrfs_ioctl_scrub_progress(root, argp);
4065 case BTRFS_IOC_BALANCE_V2:
4066 return btrfs_ioctl_balance(file, argp);
4067 case BTRFS_IOC_BALANCE_CTL:
4068 return btrfs_ioctl_balance_ctl(root, arg);
4069 case BTRFS_IOC_BALANCE_PROGRESS:
4070 return btrfs_ioctl_balance_progress(root, argp);
4071 case BTRFS_IOC_SET_RECEIVED_SUBVOL:
4072 return btrfs_ioctl_set_received_subvol(file, argp);
4073 case BTRFS_IOC_SEND:
4074 return btrfs_ioctl_send(file, argp);
4075 case BTRFS_IOC_GET_DEV_STATS:
4076 return btrfs_ioctl_get_dev_stats(root, argp);
4077 case BTRFS_IOC_QUOTA_CTL:
4078 return btrfs_ioctl_quota_ctl(file, argp);
4079 case BTRFS_IOC_QGROUP_ASSIGN:
4080 return btrfs_ioctl_qgroup_assign(file, argp);
4081 case BTRFS_IOC_QGROUP_CREATE:
4082 return btrfs_ioctl_qgroup_create(file, argp);
4083 case BTRFS_IOC_QGROUP_LIMIT:
4084 return btrfs_ioctl_qgroup_limit(file, argp);
4085 case BTRFS_IOC_DEV_REPLACE:
4086 return btrfs_ioctl_dev_replace(root, argp);
4087 case BTRFS_IOC_GET_FSLABEL:
4088 return btrfs_ioctl_get_fslabel(file, argp);
4089 case BTRFS_IOC_SET_FSLABEL:
4090 return btrfs_ioctl_set_fslabel(file, argp);