2 * Copyright (C) 2007 Oracle. All rights reserved.
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public
6 * License v2 as published by the Free Software Foundation.
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
13 * You should have received a copy of the GNU General Public
14 * License along with this program; if not, write to the
15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
16 * Boston, MA 021110-1307, USA.
19 #include <linux/kernel.h>
20 #include <linux/bio.h>
21 #include <linux/buffer_head.h>
22 #include <linux/file.h>
24 #include <linux/fsnotify.h>
25 #include <linux/pagemap.h>
26 #include <linux/highmem.h>
27 #include <linux/time.h>
28 #include <linux/init.h>
29 #include <linux/string.h>
30 #include <linux/backing-dev.h>
31 #include <linux/mount.h>
32 #include <linux/mpage.h>
33 #include <linux/namei.h>
34 #include <linux/swap.h>
35 #include <linux/writeback.h>
36 #include <linux/statfs.h>
37 #include <linux/compat.h>
38 #include <linux/bit_spinlock.h>
39 #include <linux/security.h>
40 #include <linux/xattr.h>
41 #include <linux/vmalloc.h>
42 #include <linux/slab.h>
43 #include <linux/blkdev.h>
47 #include "transaction.h"
48 #include "btrfs_inode.h"
50 #include "print-tree.h"
53 #include "inode-map.h"
55 /* Mask out flags that are inappropriate for the given type of inode. */
56 static inline __u32 btrfs_mask_flags(umode_t mode, __u32 flags)
60 else if (S_ISREG(mode))
61 return flags & ~FS_DIRSYNC_FL;
63 return flags & (FS_NODUMP_FL | FS_NOATIME_FL);
67 * Export inode flags to the format expected by the FS_IOC_GETFLAGS ioctl.
69 static unsigned int btrfs_flags_to_ioctl(unsigned int flags)
71 unsigned int iflags = 0;
73 if (flags & BTRFS_INODE_SYNC)
75 if (flags & BTRFS_INODE_IMMUTABLE)
76 iflags |= FS_IMMUTABLE_FL;
77 if (flags & BTRFS_INODE_APPEND)
78 iflags |= FS_APPEND_FL;
79 if (flags & BTRFS_INODE_NODUMP)
80 iflags |= FS_NODUMP_FL;
81 if (flags & BTRFS_INODE_NOATIME)
82 iflags |= FS_NOATIME_FL;
83 if (flags & BTRFS_INODE_DIRSYNC)
84 iflags |= FS_DIRSYNC_FL;
85 if (flags & BTRFS_INODE_NODATACOW)
86 iflags |= FS_NOCOW_FL;
88 if ((flags & BTRFS_INODE_COMPRESS) && !(flags & BTRFS_INODE_NOCOMPRESS))
89 iflags |= FS_COMPR_FL;
90 else if (flags & BTRFS_INODE_NOCOMPRESS)
91 iflags |= FS_NOCOMP_FL;
97 * Update inode->i_flags based on the btrfs internal flags.
99 void btrfs_update_iflags(struct inode *inode)
101 struct btrfs_inode *ip = BTRFS_I(inode);
103 inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
105 if (ip->flags & BTRFS_INODE_SYNC)
106 inode->i_flags |= S_SYNC;
107 if (ip->flags & BTRFS_INODE_IMMUTABLE)
108 inode->i_flags |= S_IMMUTABLE;
109 if (ip->flags & BTRFS_INODE_APPEND)
110 inode->i_flags |= S_APPEND;
111 if (ip->flags & BTRFS_INODE_NOATIME)
112 inode->i_flags |= S_NOATIME;
113 if (ip->flags & BTRFS_INODE_DIRSYNC)
114 inode->i_flags |= S_DIRSYNC;
118 * Inherit flags from the parent inode.
120 * Unlike extN we don't have any flags we don't want to inherit currently.
122 void btrfs_inherit_iflags(struct inode *inode, struct inode *dir)
129 flags = BTRFS_I(dir)->flags;
131 if (S_ISREG(inode->i_mode))
132 flags &= ~BTRFS_INODE_DIRSYNC;
133 else if (!S_ISDIR(inode->i_mode))
134 flags &= (BTRFS_INODE_NODUMP | BTRFS_INODE_NOATIME);
136 BTRFS_I(inode)->flags = flags;
137 btrfs_update_iflags(inode);
140 static int btrfs_ioctl_getflags(struct file *file, void __user *arg)
142 struct btrfs_inode *ip = BTRFS_I(file->f_path.dentry->d_inode);
143 unsigned int flags = btrfs_flags_to_ioctl(ip->flags);
145 if (copy_to_user(arg, &flags, sizeof(flags)))
150 static int check_flags(unsigned int flags)
152 if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \
153 FS_NOATIME_FL | FS_NODUMP_FL | \
154 FS_SYNC_FL | FS_DIRSYNC_FL | \
155 FS_NOCOMP_FL | FS_COMPR_FL |
159 if ((flags & FS_NOCOMP_FL) && (flags & FS_COMPR_FL))
165 static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
167 struct inode *inode = file->f_path.dentry->d_inode;
168 struct btrfs_inode *ip = BTRFS_I(inode);
169 struct btrfs_root *root = ip->root;
170 struct btrfs_trans_handle *trans;
171 unsigned int flags, oldflags;
174 if (btrfs_root_readonly(root))
177 if (copy_from_user(&flags, arg, sizeof(flags)))
180 ret = check_flags(flags);
184 if (!inode_owner_or_capable(inode))
187 mutex_lock(&inode->i_mutex);
189 flags = btrfs_mask_flags(inode->i_mode, flags);
190 oldflags = btrfs_flags_to_ioctl(ip->flags);
191 if ((flags ^ oldflags) & (FS_APPEND_FL | FS_IMMUTABLE_FL)) {
192 if (!capable(CAP_LINUX_IMMUTABLE)) {
198 ret = mnt_want_write(file->f_path.mnt);
202 if (flags & FS_SYNC_FL)
203 ip->flags |= BTRFS_INODE_SYNC;
205 ip->flags &= ~BTRFS_INODE_SYNC;
206 if (flags & FS_IMMUTABLE_FL)
207 ip->flags |= BTRFS_INODE_IMMUTABLE;
209 ip->flags &= ~BTRFS_INODE_IMMUTABLE;
210 if (flags & FS_APPEND_FL)
211 ip->flags |= BTRFS_INODE_APPEND;
213 ip->flags &= ~BTRFS_INODE_APPEND;
214 if (flags & FS_NODUMP_FL)
215 ip->flags |= BTRFS_INODE_NODUMP;
217 ip->flags &= ~BTRFS_INODE_NODUMP;
218 if (flags & FS_NOATIME_FL)
219 ip->flags |= BTRFS_INODE_NOATIME;
221 ip->flags &= ~BTRFS_INODE_NOATIME;
222 if (flags & FS_DIRSYNC_FL)
223 ip->flags |= BTRFS_INODE_DIRSYNC;
225 ip->flags &= ~BTRFS_INODE_DIRSYNC;
226 if (flags & FS_NOCOW_FL)
227 ip->flags |= BTRFS_INODE_NODATACOW;
229 ip->flags &= ~BTRFS_INODE_NODATACOW;
232 * The COMPRESS flag can only be changed by users, while the NOCOMPRESS
233 * flag may be changed automatically if compression code won't make
236 if (flags & FS_NOCOMP_FL) {
237 ip->flags &= ~BTRFS_INODE_COMPRESS;
238 ip->flags |= BTRFS_INODE_NOCOMPRESS;
239 } else if (flags & FS_COMPR_FL) {
240 ip->flags |= BTRFS_INODE_COMPRESS;
241 ip->flags &= ~BTRFS_INODE_NOCOMPRESS;
243 ip->flags &= ~(BTRFS_INODE_COMPRESS | BTRFS_INODE_NOCOMPRESS);
246 trans = btrfs_join_transaction(root);
247 BUG_ON(IS_ERR(trans));
249 ret = btrfs_update_inode(trans, root, inode);
252 btrfs_update_iflags(inode);
253 inode->i_ctime = CURRENT_TIME;
254 btrfs_end_transaction(trans, root);
256 mnt_drop_write(file->f_path.mnt);
260 mutex_unlock(&inode->i_mutex);
264 static int btrfs_ioctl_getversion(struct file *file, int __user *arg)
266 struct inode *inode = file->f_path.dentry->d_inode;
268 return put_user(inode->i_generation, arg);
271 static noinline int btrfs_ioctl_fitrim(struct file *file, void __user *arg)
273 struct btrfs_root *root = fdentry(file)->d_sb->s_fs_info;
274 struct btrfs_fs_info *fs_info = root->fs_info;
275 struct btrfs_device *device;
276 struct request_queue *q;
277 struct fstrim_range range;
278 u64 minlen = ULLONG_MAX;
282 if (!capable(CAP_SYS_ADMIN))
286 list_for_each_entry_rcu(device, &fs_info->fs_devices->devices,
290 q = bdev_get_queue(device->bdev);
291 if (blk_queue_discard(q)) {
293 minlen = min((u64)q->limits.discard_granularity,
301 if (copy_from_user(&range, arg, sizeof(range)))
304 range.minlen = max(range.minlen, minlen);
305 ret = btrfs_trim_fs(root, &range);
309 if (copy_to_user(arg, &range, sizeof(range)))
315 static noinline int create_subvol(struct btrfs_root *root,
316 struct dentry *dentry,
317 char *name, int namelen,
320 struct btrfs_trans_handle *trans;
321 struct btrfs_key key;
322 struct btrfs_root_item root_item;
323 struct btrfs_inode_item *inode_item;
324 struct extent_buffer *leaf;
325 struct btrfs_root *new_root;
326 struct dentry *parent = dget_parent(dentry);
331 u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
334 ret = btrfs_find_free_objectid(root->fs_info->tree_root, &objectid);
340 dir = parent->d_inode;
348 trans = btrfs_start_transaction(root, 6);
351 return PTR_ERR(trans);
354 leaf = btrfs_alloc_free_block(trans, root, root->leafsize,
355 0, objectid, NULL, 0, 0, 0);
361 memset_extent_buffer(leaf, 0, 0, sizeof(struct btrfs_header));
362 btrfs_set_header_bytenr(leaf, leaf->start);
363 btrfs_set_header_generation(leaf, trans->transid);
364 btrfs_set_header_backref_rev(leaf, BTRFS_MIXED_BACKREF_REV);
365 btrfs_set_header_owner(leaf, objectid);
367 write_extent_buffer(leaf, root->fs_info->fsid,
368 (unsigned long)btrfs_header_fsid(leaf),
370 write_extent_buffer(leaf, root->fs_info->chunk_tree_uuid,
371 (unsigned long)btrfs_header_chunk_tree_uuid(leaf),
373 btrfs_mark_buffer_dirty(leaf);
375 inode_item = &root_item.inode;
376 memset(inode_item, 0, sizeof(*inode_item));
377 inode_item->generation = cpu_to_le64(1);
378 inode_item->size = cpu_to_le64(3);
379 inode_item->nlink = cpu_to_le32(1);
380 inode_item->nbytes = cpu_to_le64(root->leafsize);
381 inode_item->mode = cpu_to_le32(S_IFDIR | 0755);
384 root_item.byte_limit = 0;
385 inode_item->flags = cpu_to_le64(BTRFS_INODE_ROOT_ITEM_INIT);
387 btrfs_set_root_bytenr(&root_item, leaf->start);
388 btrfs_set_root_generation(&root_item, trans->transid);
389 btrfs_set_root_level(&root_item, 0);
390 btrfs_set_root_refs(&root_item, 1);
391 btrfs_set_root_used(&root_item, leaf->len);
392 btrfs_set_root_last_snapshot(&root_item, 0);
394 memset(&root_item.drop_progress, 0, sizeof(root_item.drop_progress));
395 root_item.drop_level = 0;
397 btrfs_tree_unlock(leaf);
398 free_extent_buffer(leaf);
401 btrfs_set_root_dirid(&root_item, new_dirid);
403 key.objectid = objectid;
405 btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
406 ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key,
411 key.offset = (u64)-1;
412 new_root = btrfs_read_fs_root_no_name(root->fs_info, &key);
413 BUG_ON(IS_ERR(new_root));
415 btrfs_record_root_in_trans(trans, new_root);
417 ret = btrfs_create_subvol_root(trans, new_root, new_dirid);
419 * insert the directory item
421 ret = btrfs_set_inode_index(dir, &index);
424 ret = btrfs_insert_dir_item(trans, root,
425 name, namelen, dir, &key,
426 BTRFS_FT_DIR, index);
430 btrfs_i_size_write(dir, dir->i_size + namelen * 2);
431 ret = btrfs_update_inode(trans, root, dir);
434 ret = btrfs_add_root_ref(trans, root->fs_info->tree_root,
435 objectid, root->root_key.objectid,
436 btrfs_ino(dir), index, name, namelen);
440 d_instantiate(dentry, btrfs_lookup_dentry(dir, dentry));
444 *async_transid = trans->transid;
445 err = btrfs_commit_transaction_async(trans, root, 1);
447 err = btrfs_commit_transaction(trans, root);
454 static int create_snapshot(struct btrfs_root *root, struct dentry *dentry,
455 char *name, int namelen, u64 *async_transid,
459 struct dentry *parent;
460 struct btrfs_pending_snapshot *pending_snapshot;
461 struct btrfs_trans_handle *trans;
467 pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS);
468 if (!pending_snapshot)
471 btrfs_init_block_rsv(&pending_snapshot->block_rsv);
472 pending_snapshot->dentry = dentry;
473 pending_snapshot->root = root;
474 pending_snapshot->readonly = readonly;
476 trans = btrfs_start_transaction(root->fs_info->extent_root, 5);
478 ret = PTR_ERR(trans);
482 ret = btrfs_snap_reserve_metadata(trans, pending_snapshot);
485 list_add(&pending_snapshot->list,
486 &trans->transaction->pending_snapshots);
488 *async_transid = trans->transid;
489 ret = btrfs_commit_transaction_async(trans,
490 root->fs_info->extent_root, 1);
492 ret = btrfs_commit_transaction(trans,
493 root->fs_info->extent_root);
497 ret = pending_snapshot->error;
501 ret = btrfs_orphan_cleanup(pending_snapshot->snap);
505 parent = dget_parent(dentry);
506 inode = btrfs_lookup_dentry(parent->d_inode, dentry);
509 ret = PTR_ERR(inode);
513 d_instantiate(dentry, inode);
516 kfree(pending_snapshot);
520 /* copy of check_sticky in fs/namei.c()
521 * It's inline, so penalty for filesystems that don't use sticky bit is
524 static inline int btrfs_check_sticky(struct inode *dir, struct inode *inode)
526 uid_t fsuid = current_fsuid();
528 if (!(dir->i_mode & S_ISVTX))
530 if (inode->i_uid == fsuid)
532 if (dir->i_uid == fsuid)
534 return !capable(CAP_FOWNER);
537 /* copy of may_delete in fs/namei.c()
538 * Check whether we can remove a link victim from directory dir, check
539 * whether the type of victim is right.
540 * 1. We can't do it if dir is read-only (done in permission())
541 * 2. We should have write and exec permissions on dir
542 * 3. We can't remove anything from append-only dir
543 * 4. We can't do anything with immutable dir (done in permission())
544 * 5. If the sticky bit on dir is set we should either
545 * a. be owner of dir, or
546 * b. be owner of victim, or
547 * c. have CAP_FOWNER capability
548 * 6. If the victim is append-only or immutable we can't do antyhing with
549 * links pointing to it.
550 * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR.
551 * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR.
552 * 9. We can't remove a root or mountpoint.
553 * 10. We don't allow removal of NFS sillyrenamed files; it's handled by
554 * nfs_async_unlink().
557 static int btrfs_may_delete(struct inode *dir,struct dentry *victim,int isdir)
561 if (!victim->d_inode)
564 BUG_ON(victim->d_parent->d_inode != dir);
565 audit_inode_child(victim, dir);
567 error = inode_permission(dir, MAY_WRITE | MAY_EXEC);
572 if (btrfs_check_sticky(dir, victim->d_inode)||
573 IS_APPEND(victim->d_inode)||
574 IS_IMMUTABLE(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
577 if (!S_ISDIR(victim->d_inode->i_mode))
581 } else if (S_ISDIR(victim->d_inode->i_mode))
585 if (victim->d_flags & DCACHE_NFSFS_RENAMED)
590 /* copy of may_create in fs/namei.c() */
591 static inline int btrfs_may_create(struct inode *dir, struct dentry *child)
597 return inode_permission(dir, MAY_WRITE | MAY_EXEC);
601 * Create a new subvolume below @parent. This is largely modeled after
602 * sys_mkdirat and vfs_mkdir, but we only do a single component lookup
603 * inside this filesystem so it's quite a bit simpler.
605 static noinline int btrfs_mksubvol(struct path *parent,
606 char *name, int namelen,
607 struct btrfs_root *snap_src,
608 u64 *async_transid, bool readonly)
610 struct inode *dir = parent->dentry->d_inode;
611 struct dentry *dentry;
614 mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
616 dentry = lookup_one_len(name, parent->dentry, namelen);
617 error = PTR_ERR(dentry);
625 error = mnt_want_write(parent->mnt);
629 error = btrfs_may_create(dir, dentry);
633 down_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
635 if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0)
639 error = create_snapshot(snap_src, dentry,
640 name, namelen, async_transid, readonly);
642 error = create_subvol(BTRFS_I(dir)->root, dentry,
643 name, namelen, async_transid);
646 fsnotify_mkdir(dir, dentry);
648 up_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
650 mnt_drop_write(parent->mnt);
654 mutex_unlock(&dir->i_mutex);
659 * When we're defragging a range, we don't want to kick it off again
660 * if it is really just waiting for delalloc to send it down.
661 * If we find a nice big extent or delalloc range for the bytes in the
662 * file you want to defrag, we return 0 to let you know to skip this
665 static int check_defrag_in_cache(struct inode *inode, u64 offset, int thresh)
667 struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
668 struct extent_map *em = NULL;
669 struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
672 read_lock(&em_tree->lock);
673 em = lookup_extent_mapping(em_tree, offset, PAGE_CACHE_SIZE);
674 read_unlock(&em_tree->lock);
677 end = extent_map_end(em);
679 if (end - offset > thresh)
682 /* if we already have a nice delalloc here, just stop */
684 end = count_range_bits(io_tree, &offset, offset + thresh,
685 thresh, EXTENT_DELALLOC, 1);
692 * helper function to walk through a file and find extents
693 * newer than a specific transid, and smaller than thresh.
695 * This is used by the defragging code to find new and small
698 static int find_new_extents(struct btrfs_root *root,
699 struct inode *inode, u64 newer_than,
700 u64 *off, int thresh)
702 struct btrfs_path *path;
703 struct btrfs_key min_key;
704 struct btrfs_key max_key;
705 struct extent_buffer *leaf;
706 struct btrfs_file_extent_item *extent;
710 path = btrfs_alloc_path();
714 min_key.objectid = inode->i_ino;
715 min_key.type = BTRFS_EXTENT_DATA_KEY;
716 min_key.offset = *off;
718 max_key.objectid = inode->i_ino;
719 max_key.type = (u8)-1;
720 max_key.offset = (u64)-1;
722 path->keep_locks = 1;
725 ret = btrfs_search_forward(root, &min_key, &max_key,
726 path, 0, newer_than);
729 if (min_key.objectid != inode->i_ino)
731 if (min_key.type != BTRFS_EXTENT_DATA_KEY)
734 leaf = path->nodes[0];
735 extent = btrfs_item_ptr(leaf, path->slots[0],
736 struct btrfs_file_extent_item);
738 type = btrfs_file_extent_type(leaf, extent);
739 if (type == BTRFS_FILE_EXTENT_REG &&
740 btrfs_file_extent_num_bytes(leaf, extent) < thresh &&
741 check_defrag_in_cache(inode, min_key.offset, thresh)) {
742 *off = min_key.offset;
743 btrfs_free_path(path);
747 if (min_key.offset == (u64)-1)
751 btrfs_release_path(path);
754 btrfs_free_path(path);
758 static int should_defrag_range(struct inode *inode, u64 start, u64 len,
759 int thresh, u64 *last_len, u64 *skip,
762 struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
763 struct extent_map *em = NULL;
764 struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
768 * make sure that once we start defragging and extent, we keep on
771 if (start < *defrag_end)
777 * hopefully we have this extent in the tree already, try without
778 * the full extent lock
780 read_lock(&em_tree->lock);
781 em = lookup_extent_mapping(em_tree, start, len);
782 read_unlock(&em_tree->lock);
785 /* get the big lock and read metadata off disk */
786 lock_extent(io_tree, start, start + len - 1, GFP_NOFS);
787 em = btrfs_get_extent(inode, NULL, 0, start, len, 0);
788 unlock_extent(io_tree, start, start + len - 1, GFP_NOFS);
794 /* this will cover holes, and inline extents */
795 if (em->block_start >= EXTENT_MAP_LAST_BYTE)
799 * we hit a real extent, if it is big don't bother defragging it again
801 if ((*last_len == 0 || *last_len >= thresh) && em->len >= thresh)
805 * last_len ends up being a counter of how many bytes we've defragged.
806 * every time we choose not to defrag an extent, we reset *last_len
807 * so that the next tiny extent will force a defrag.
809 * The end result of this is that tiny extents before a single big
810 * extent will force at least part of that big extent to be defragged.
814 *defrag_end = extent_map_end(em);
817 *skip = extent_map_end(em);
826 * it doesn't do much good to defrag one or two pages
827 * at a time. This pulls in a nice chunk of pages
830 * It also makes sure the delalloc code has enough
831 * dirty data to avoid making new small extents as part
834 * It's a good idea to start RA on this range
835 * before calling this.
837 static int cluster_pages_for_defrag(struct inode *inode,
839 unsigned long start_index,
842 unsigned long file_end;
843 u64 isize = i_size_read(inode);
849 struct btrfs_ordered_extent *ordered;
850 struct extent_state *cached_state = NULL;
854 file_end = (isize - 1) >> PAGE_CACHE_SHIFT;
856 ret = btrfs_delalloc_reserve_space(inode,
857 num_pages << PAGE_CACHE_SHIFT);
864 /* step one, lock all the pages */
865 for (i = 0; i < num_pages; i++) {
867 page = grab_cache_page(inode->i_mapping,
872 if (!PageUptodate(page)) {
873 btrfs_readpage(NULL, page);
875 if (!PageUptodate(page)) {
877 page_cache_release(page);
882 isize = i_size_read(inode);
883 file_end = (isize - 1) >> PAGE_CACHE_SHIFT;
884 if (!isize || page->index > file_end ||
885 page->mapping != inode->i_mapping) {
886 /* whoops, we blew past eof, skip this page */
888 page_cache_release(page);
897 if (!(inode->i_sb->s_flags & MS_ACTIVE))
901 * so now we have a nice long stream of locked
902 * and up to date pages, lets wait on them
904 for (i = 0; i < i_done; i++)
905 wait_on_page_writeback(pages[i]);
907 page_start = page_offset(pages[0]);
908 page_end = page_offset(pages[i_done - 1]) + PAGE_CACHE_SIZE;
910 lock_extent_bits(&BTRFS_I(inode)->io_tree,
911 page_start, page_end - 1, 0, &cached_state,
913 ordered = btrfs_lookup_first_ordered_extent(inode, page_end - 1);
915 ordered->file_offset + ordered->len > page_start &&
916 ordered->file_offset < page_end) {
917 btrfs_put_ordered_extent(ordered);
918 unlock_extent_cached(&BTRFS_I(inode)->io_tree,
919 page_start, page_end - 1,
920 &cached_state, GFP_NOFS);
921 for (i = 0; i < i_done; i++) {
922 unlock_page(pages[i]);
923 page_cache_release(pages[i]);
925 btrfs_wait_ordered_range(inode, page_start,
926 page_end - page_start);
930 btrfs_put_ordered_extent(ordered);
932 clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start,
933 page_end - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
934 EXTENT_DO_ACCOUNTING, 0, 0, &cached_state,
937 if (i_done != num_pages) {
938 atomic_inc(&BTRFS_I(inode)->outstanding_extents);
939 btrfs_delalloc_release_space(inode,
940 (num_pages - i_done) << PAGE_CACHE_SHIFT);
944 btrfs_set_extent_delalloc(inode, page_start, page_end - 1,
947 unlock_extent_cached(&BTRFS_I(inode)->io_tree,
948 page_start, page_end - 1, &cached_state,
951 for (i = 0; i < i_done; i++) {
952 clear_page_dirty_for_io(pages[i]);
953 ClearPageChecked(pages[i]);
954 set_page_extent_mapped(pages[i]);
955 set_page_dirty(pages[i]);
956 unlock_page(pages[i]);
957 page_cache_release(pages[i]);
961 for (i = 0; i < i_done; i++) {
962 unlock_page(pages[i]);
963 page_cache_release(pages[i]);
965 btrfs_delalloc_release_space(inode, num_pages << PAGE_CACHE_SHIFT);
970 int btrfs_defrag_file(struct inode *inode, struct file *file,
971 struct btrfs_ioctl_defrag_range_args *range,
972 u64 newer_than, unsigned long max_to_defrag)
974 struct btrfs_root *root = BTRFS_I(inode)->root;
975 struct btrfs_super_block *disk_super;
976 struct file_ra_state *ra = NULL;
977 unsigned long last_index;
982 u64 newer_off = range->start;
986 int defrag_count = 0;
987 int compress_type = BTRFS_COMPRESS_ZLIB;
988 int extent_thresh = range->extent_thresh;
989 int newer_cluster = (256 * 1024) >> PAGE_CACHE_SHIFT;
990 u64 new_align = ~((u64)128 * 1024 - 1);
991 struct page **pages = NULL;
993 if (extent_thresh == 0)
994 extent_thresh = 256 * 1024;
996 if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS) {
997 if (range->compress_type > BTRFS_COMPRESS_TYPES)
999 if (range->compress_type)
1000 compress_type = range->compress_type;
1003 if (inode->i_size == 0)
1007 * if we were not given a file, allocate a readahead
1011 ra = kzalloc(sizeof(*ra), GFP_NOFS);
1014 file_ra_state_init(ra, inode->i_mapping);
1019 pages = kmalloc(sizeof(struct page *) * newer_cluster,
1026 /* find the last page to defrag */
1027 if (range->start + range->len > range->start) {
1028 last_index = min_t(u64, inode->i_size - 1,
1029 range->start + range->len - 1) >> PAGE_CACHE_SHIFT;
1031 last_index = (inode->i_size - 1) >> PAGE_CACHE_SHIFT;
1035 ret = find_new_extents(root, inode, newer_than,
1036 &newer_off, 64 * 1024);
1038 range->start = newer_off;
1040 * we always align our defrag to help keep
1041 * the extents in the file evenly spaced
1043 i = (newer_off & new_align) >> PAGE_CACHE_SHIFT;
1044 newer_left = newer_cluster;
1048 i = range->start >> PAGE_CACHE_SHIFT;
1051 max_to_defrag = last_index - 1;
1053 while (i <= last_index && defrag_count < max_to_defrag) {
1055 * make sure we stop running if someone unmounts
1058 if (!(inode->i_sb->s_flags & MS_ACTIVE))
1062 !should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT,
1069 * the should_defrag function tells us how much to skip
1070 * bump our counter by the suggested amount
1072 next = (skip + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
1073 i = max(i + 1, next);
1076 if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)
1077 BTRFS_I(inode)->force_compress = compress_type;
1079 btrfs_force_ra(inode->i_mapping, ra, file, i, newer_cluster);
1081 ret = cluster_pages_for_defrag(inode, pages, i, newer_cluster);
1085 defrag_count += ret;
1086 balance_dirty_pages_ratelimited_nr(inode->i_mapping, ret);
1090 if (newer_off == (u64)-1)
1093 newer_off = max(newer_off + 1,
1094 (u64)i << PAGE_CACHE_SHIFT);
1096 ret = find_new_extents(root, inode,
1097 newer_than, &newer_off,
1100 range->start = newer_off;
1101 i = (newer_off & new_align) >> PAGE_CACHE_SHIFT;
1102 newer_left = newer_cluster;
1111 if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO))
1112 filemap_flush(inode->i_mapping);
1114 if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
1115 /* the filemap_flush will queue IO into the worker threads, but
1116 * we have to make sure the IO is actually started and that
1117 * ordered extents get created before we return
1119 atomic_inc(&root->fs_info->async_submit_draining);
1120 while (atomic_read(&root->fs_info->nr_async_submits) ||
1121 atomic_read(&root->fs_info->async_delalloc_pages)) {
1122 wait_event(root->fs_info->async_submit_wait,
1123 (atomic_read(&root->fs_info->nr_async_submits) == 0 &&
1124 atomic_read(&root->fs_info->async_delalloc_pages) == 0));
1126 atomic_dec(&root->fs_info->async_submit_draining);
1128 mutex_lock(&inode->i_mutex);
1129 BTRFS_I(inode)->force_compress = BTRFS_COMPRESS_NONE;
1130 mutex_unlock(&inode->i_mutex);
1133 disk_super = &root->fs_info->super_copy;
1134 features = btrfs_super_incompat_flags(disk_super);
1135 if (range->compress_type == BTRFS_COMPRESS_LZO) {
1136 features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
1137 btrfs_set_super_incompat_flags(disk_super, features);
1142 return defrag_count;
1151 static noinline int btrfs_ioctl_resize(struct btrfs_root *root,
1157 struct btrfs_ioctl_vol_args *vol_args;
1158 struct btrfs_trans_handle *trans;
1159 struct btrfs_device *device = NULL;
1161 char *devstr = NULL;
1165 if (root->fs_info->sb->s_flags & MS_RDONLY)
1168 if (!capable(CAP_SYS_ADMIN))
1171 vol_args = memdup_user(arg, sizeof(*vol_args));
1172 if (IS_ERR(vol_args))
1173 return PTR_ERR(vol_args);
1175 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
1177 mutex_lock(&root->fs_info->volume_mutex);
1178 sizestr = vol_args->name;
1179 devstr = strchr(sizestr, ':');
1182 sizestr = devstr + 1;
1184 devstr = vol_args->name;
1185 devid = simple_strtoull(devstr, &end, 10);
1186 printk(KERN_INFO "resizing devid %llu\n",
1187 (unsigned long long)devid);
1189 device = btrfs_find_device(root, devid, NULL, NULL);
1191 printk(KERN_INFO "resizer unable to find device %llu\n",
1192 (unsigned long long)devid);
1196 if (!strcmp(sizestr, "max"))
1197 new_size = device->bdev->bd_inode->i_size;
1199 if (sizestr[0] == '-') {
1202 } else if (sizestr[0] == '+') {
1206 new_size = memparse(sizestr, NULL);
1207 if (new_size == 0) {
1213 old_size = device->total_bytes;
1216 if (new_size > old_size) {
1220 new_size = old_size - new_size;
1221 } else if (mod > 0) {
1222 new_size = old_size + new_size;
1225 if (new_size < 256 * 1024 * 1024) {
1229 if (new_size > device->bdev->bd_inode->i_size) {
1234 do_div(new_size, root->sectorsize);
1235 new_size *= root->sectorsize;
1237 printk(KERN_INFO "new size for %s is %llu\n",
1238 device->name, (unsigned long long)new_size);
1240 if (new_size > old_size) {
1241 trans = btrfs_start_transaction(root, 0);
1242 if (IS_ERR(trans)) {
1243 ret = PTR_ERR(trans);
1246 ret = btrfs_grow_device(trans, device, new_size);
1247 btrfs_commit_transaction(trans, root);
1249 ret = btrfs_shrink_device(device, new_size);
1253 mutex_unlock(&root->fs_info->volume_mutex);
1258 static noinline int btrfs_ioctl_snap_create_transid(struct file *file,
1265 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
1266 struct file *src_file;
1270 if (root->fs_info->sb->s_flags & MS_RDONLY)
1273 namelen = strlen(name);
1274 if (strchr(name, '/')) {
1280 ret = btrfs_mksubvol(&file->f_path, name, namelen,
1281 NULL, transid, readonly);
1283 struct inode *src_inode;
1284 src_file = fget(fd);
1290 src_inode = src_file->f_path.dentry->d_inode;
1291 if (src_inode->i_sb != file->f_path.dentry->d_inode->i_sb) {
1292 printk(KERN_INFO "btrfs: Snapshot src from "
1298 ret = btrfs_mksubvol(&file->f_path, name, namelen,
1299 BTRFS_I(src_inode)->root,
1307 static noinline int btrfs_ioctl_snap_create(struct file *file,
1308 void __user *arg, int subvol)
1310 struct btrfs_ioctl_vol_args *vol_args;
1313 vol_args = memdup_user(arg, sizeof(*vol_args));
1314 if (IS_ERR(vol_args))
1315 return PTR_ERR(vol_args);
1316 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
1318 ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
1319 vol_args->fd, subvol,
1326 static noinline int btrfs_ioctl_snap_create_v2(struct file *file,
1327 void __user *arg, int subvol)
1329 struct btrfs_ioctl_vol_args_v2 *vol_args;
1333 bool readonly = false;
1335 vol_args = memdup_user(arg, sizeof(*vol_args));
1336 if (IS_ERR(vol_args))
1337 return PTR_ERR(vol_args);
1338 vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
1340 if (vol_args->flags &
1341 ~(BTRFS_SUBVOL_CREATE_ASYNC | BTRFS_SUBVOL_RDONLY)) {
1346 if (vol_args->flags & BTRFS_SUBVOL_CREATE_ASYNC)
1348 if (vol_args->flags & BTRFS_SUBVOL_RDONLY)
1351 ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
1352 vol_args->fd, subvol,
1355 if (ret == 0 && ptr &&
1357 offsetof(struct btrfs_ioctl_vol_args_v2,
1358 transid), ptr, sizeof(*ptr)))
1365 static noinline int btrfs_ioctl_subvol_getflags(struct file *file,
1368 struct inode *inode = fdentry(file)->d_inode;
1369 struct btrfs_root *root = BTRFS_I(inode)->root;
1373 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID)
1376 down_read(&root->fs_info->subvol_sem);
1377 if (btrfs_root_readonly(root))
1378 flags |= BTRFS_SUBVOL_RDONLY;
1379 up_read(&root->fs_info->subvol_sem);
1381 if (copy_to_user(arg, &flags, sizeof(flags)))
1387 static noinline int btrfs_ioctl_subvol_setflags(struct file *file,
1390 struct inode *inode = fdentry(file)->d_inode;
1391 struct btrfs_root *root = BTRFS_I(inode)->root;
1392 struct btrfs_trans_handle *trans;
1397 if (root->fs_info->sb->s_flags & MS_RDONLY)
1400 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID)
1403 if (copy_from_user(&flags, arg, sizeof(flags)))
1406 if (flags & BTRFS_SUBVOL_CREATE_ASYNC)
1409 if (flags & ~BTRFS_SUBVOL_RDONLY)
1412 if (!inode_owner_or_capable(inode))
1415 down_write(&root->fs_info->subvol_sem);
1418 if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root))
1421 root_flags = btrfs_root_flags(&root->root_item);
1422 if (flags & BTRFS_SUBVOL_RDONLY)
1423 btrfs_set_root_flags(&root->root_item,
1424 root_flags | BTRFS_ROOT_SUBVOL_RDONLY);
1426 btrfs_set_root_flags(&root->root_item,
1427 root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY);
1429 trans = btrfs_start_transaction(root, 1);
1430 if (IS_ERR(trans)) {
1431 ret = PTR_ERR(trans);
1435 ret = btrfs_update_root(trans, root->fs_info->tree_root,
1436 &root->root_key, &root->root_item);
1438 btrfs_commit_transaction(trans, root);
1441 btrfs_set_root_flags(&root->root_item, root_flags);
1443 up_write(&root->fs_info->subvol_sem);
1448 * helper to check if the subvolume references other subvolumes
1450 static noinline int may_destroy_subvol(struct btrfs_root *root)
1452 struct btrfs_path *path;
1453 struct btrfs_key key;
1456 path = btrfs_alloc_path();
1460 key.objectid = root->root_key.objectid;
1461 key.type = BTRFS_ROOT_REF_KEY;
1462 key.offset = (u64)-1;
1464 ret = btrfs_search_slot(NULL, root->fs_info->tree_root,
1471 if (path->slots[0] > 0) {
1473 btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
1474 if (key.objectid == root->root_key.objectid &&
1475 key.type == BTRFS_ROOT_REF_KEY)
1479 btrfs_free_path(path);
1483 static noinline int key_in_sk(struct btrfs_key *key,
1484 struct btrfs_ioctl_search_key *sk)
1486 struct btrfs_key test;
1489 test.objectid = sk->min_objectid;
1490 test.type = sk->min_type;
1491 test.offset = sk->min_offset;
1493 ret = btrfs_comp_cpu_keys(key, &test);
1497 test.objectid = sk->max_objectid;
1498 test.type = sk->max_type;
1499 test.offset = sk->max_offset;
1501 ret = btrfs_comp_cpu_keys(key, &test);
1507 static noinline int copy_to_sk(struct btrfs_root *root,
1508 struct btrfs_path *path,
1509 struct btrfs_key *key,
1510 struct btrfs_ioctl_search_key *sk,
1512 unsigned long *sk_offset,
1516 struct extent_buffer *leaf;
1517 struct btrfs_ioctl_search_header sh;
1518 unsigned long item_off;
1519 unsigned long item_len;
1525 leaf = path->nodes[0];
1526 slot = path->slots[0];
1527 nritems = btrfs_header_nritems(leaf);
1529 if (btrfs_header_generation(leaf) > sk->max_transid) {
1533 found_transid = btrfs_header_generation(leaf);
1535 for (i = slot; i < nritems; i++) {
1536 item_off = btrfs_item_ptr_offset(leaf, i);
1537 item_len = btrfs_item_size_nr(leaf, i);
1539 if (item_len > BTRFS_SEARCH_ARGS_BUFSIZE)
1542 if (sizeof(sh) + item_len + *sk_offset >
1543 BTRFS_SEARCH_ARGS_BUFSIZE) {
1548 btrfs_item_key_to_cpu(leaf, key, i);
1549 if (!key_in_sk(key, sk))
1552 sh.objectid = key->objectid;
1553 sh.offset = key->offset;
1554 sh.type = key->type;
1556 sh.transid = found_transid;
1558 /* copy search result header */
1559 memcpy(buf + *sk_offset, &sh, sizeof(sh));
1560 *sk_offset += sizeof(sh);
1563 char *p = buf + *sk_offset;
1565 read_extent_buffer(leaf, p,
1566 item_off, item_len);
1567 *sk_offset += item_len;
1571 if (*num_found >= sk->nr_items)
1576 if (key->offset < (u64)-1 && key->offset < sk->max_offset)
1578 else if (key->type < (u8)-1 && key->type < sk->max_type) {
1581 } else if (key->objectid < (u64)-1 && key->objectid < sk->max_objectid) {
1591 static noinline int search_ioctl(struct inode *inode,
1592 struct btrfs_ioctl_search_args *args)
1594 struct btrfs_root *root;
1595 struct btrfs_key key;
1596 struct btrfs_key max_key;
1597 struct btrfs_path *path;
1598 struct btrfs_ioctl_search_key *sk = &args->key;
1599 struct btrfs_fs_info *info = BTRFS_I(inode)->root->fs_info;
1602 unsigned long sk_offset = 0;
1604 path = btrfs_alloc_path();
1608 if (sk->tree_id == 0) {
1609 /* search the root of the inode that was passed */
1610 root = BTRFS_I(inode)->root;
1612 key.objectid = sk->tree_id;
1613 key.type = BTRFS_ROOT_ITEM_KEY;
1614 key.offset = (u64)-1;
1615 root = btrfs_read_fs_root_no_name(info, &key);
1617 printk(KERN_ERR "could not find root %llu\n",
1619 btrfs_free_path(path);
1624 key.objectid = sk->min_objectid;
1625 key.type = sk->min_type;
1626 key.offset = sk->min_offset;
1628 max_key.objectid = sk->max_objectid;
1629 max_key.type = sk->max_type;
1630 max_key.offset = sk->max_offset;
1632 path->keep_locks = 1;
1635 ret = btrfs_search_forward(root, &key, &max_key, path, 0,
1642 ret = copy_to_sk(root, path, &key, sk, args->buf,
1643 &sk_offset, &num_found);
1644 btrfs_release_path(path);
1645 if (ret || num_found >= sk->nr_items)
1651 sk->nr_items = num_found;
1652 btrfs_free_path(path);
1656 static noinline int btrfs_ioctl_tree_search(struct file *file,
1659 struct btrfs_ioctl_search_args *args;
1660 struct inode *inode;
1663 if (!capable(CAP_SYS_ADMIN))
1666 args = memdup_user(argp, sizeof(*args));
1668 return PTR_ERR(args);
1670 inode = fdentry(file)->d_inode;
1671 ret = search_ioctl(inode, args);
1672 if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
1679 * Search INODE_REFs to identify path name of 'dirid' directory
1680 * in a 'tree_id' tree. and sets path name to 'name'.
1682 static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info,
1683 u64 tree_id, u64 dirid, char *name)
1685 struct btrfs_root *root;
1686 struct btrfs_key key;
1692 struct btrfs_inode_ref *iref;
1693 struct extent_buffer *l;
1694 struct btrfs_path *path;
1696 if (dirid == BTRFS_FIRST_FREE_OBJECTID) {
1701 path = btrfs_alloc_path();
1705 ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX];
1707 key.objectid = tree_id;
1708 key.type = BTRFS_ROOT_ITEM_KEY;
1709 key.offset = (u64)-1;
1710 root = btrfs_read_fs_root_no_name(info, &key);
1712 printk(KERN_ERR "could not find root %llu\n", tree_id);
1717 key.objectid = dirid;
1718 key.type = BTRFS_INODE_REF_KEY;
1719 key.offset = (u64)-1;
1722 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1727 slot = path->slots[0];
1728 if (ret > 0 && slot > 0)
1730 btrfs_item_key_to_cpu(l, &key, slot);
1732 if (ret > 0 && (key.objectid != dirid ||
1733 key.type != BTRFS_INODE_REF_KEY)) {
1738 iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref);
1739 len = btrfs_inode_ref_name_len(l, iref);
1741 total_len += len + 1;
1746 read_extent_buffer(l, ptr,(unsigned long)(iref + 1), len);
1748 if (key.offset == BTRFS_FIRST_FREE_OBJECTID)
1751 btrfs_release_path(path);
1752 key.objectid = key.offset;
1753 key.offset = (u64)-1;
1754 dirid = key.objectid;
1759 memcpy(name, ptr, total_len);
1760 name[total_len]='\0';
1763 btrfs_free_path(path);
1767 static noinline int btrfs_ioctl_ino_lookup(struct file *file,
1770 struct btrfs_ioctl_ino_lookup_args *args;
1771 struct inode *inode;
1774 if (!capable(CAP_SYS_ADMIN))
1777 args = memdup_user(argp, sizeof(*args));
1779 return PTR_ERR(args);
1781 inode = fdentry(file)->d_inode;
1783 if (args->treeid == 0)
1784 args->treeid = BTRFS_I(inode)->root->root_key.objectid;
1786 ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info,
1787 args->treeid, args->objectid,
1790 if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
1797 static noinline int btrfs_ioctl_snap_destroy(struct file *file,
1800 struct dentry *parent = fdentry(file);
1801 struct dentry *dentry;
1802 struct inode *dir = parent->d_inode;
1803 struct inode *inode;
1804 struct btrfs_root *root = BTRFS_I(dir)->root;
1805 struct btrfs_root *dest = NULL;
1806 struct btrfs_ioctl_vol_args *vol_args;
1807 struct btrfs_trans_handle *trans;
1812 vol_args = memdup_user(arg, sizeof(*vol_args));
1813 if (IS_ERR(vol_args))
1814 return PTR_ERR(vol_args);
1816 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
1817 namelen = strlen(vol_args->name);
1818 if (strchr(vol_args->name, '/') ||
1819 strncmp(vol_args->name, "..", namelen) == 0) {
1824 err = mnt_want_write(file->f_path.mnt);
1828 mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
1829 dentry = lookup_one_len(vol_args->name, parent, namelen);
1830 if (IS_ERR(dentry)) {
1831 err = PTR_ERR(dentry);
1832 goto out_unlock_dir;
1835 if (!dentry->d_inode) {
1840 inode = dentry->d_inode;
1841 dest = BTRFS_I(inode)->root;
1842 if (!capable(CAP_SYS_ADMIN)){
1844 * Regular user. Only allow this with a special mount
1845 * option, when the user has write+exec access to the
1846 * subvol root, and when rmdir(2) would have been
1849 * Note that this is _not_ check that the subvol is
1850 * empty or doesn't contain data that we wouldn't
1851 * otherwise be able to delete.
1853 * Users who want to delete empty subvols should try
1857 if (!btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED))
1861 * Do not allow deletion if the parent dir is the same
1862 * as the dir to be deleted. That means the ioctl
1863 * must be called on the dentry referencing the root
1864 * of the subvol, not a random directory contained
1871 err = inode_permission(inode, MAY_WRITE | MAY_EXEC);
1875 /* check if subvolume may be deleted by a non-root user */
1876 err = btrfs_may_delete(dir, dentry, 1);
1881 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) {
1886 mutex_lock(&inode->i_mutex);
1887 err = d_invalidate(dentry);
1891 down_write(&root->fs_info->subvol_sem);
1893 err = may_destroy_subvol(dest);
1897 trans = btrfs_start_transaction(root, 0);
1898 if (IS_ERR(trans)) {
1899 err = PTR_ERR(trans);
1902 trans->block_rsv = &root->fs_info->global_block_rsv;
1904 ret = btrfs_unlink_subvol(trans, root, dir,
1905 dest->root_key.objectid,
1906 dentry->d_name.name,
1907 dentry->d_name.len);
1910 btrfs_record_root_in_trans(trans, dest);
1912 memset(&dest->root_item.drop_progress, 0,
1913 sizeof(dest->root_item.drop_progress));
1914 dest->root_item.drop_level = 0;
1915 btrfs_set_root_refs(&dest->root_item, 0);
1917 if (!xchg(&dest->orphan_item_inserted, 1)) {
1918 ret = btrfs_insert_orphan_item(trans,
1919 root->fs_info->tree_root,
1920 dest->root_key.objectid);
1924 ret = btrfs_end_transaction(trans, root);
1926 inode->i_flags |= S_DEAD;
1928 up_write(&root->fs_info->subvol_sem);
1930 mutex_unlock(&inode->i_mutex);
1932 shrink_dcache_sb(root->fs_info->sb);
1933 btrfs_invalidate_inodes(dest);
1939 mutex_unlock(&dir->i_mutex);
1940 mnt_drop_write(file->f_path.mnt);
1946 static int btrfs_ioctl_defrag(struct file *file, void __user *argp)
1948 struct inode *inode = fdentry(file)->d_inode;
1949 struct btrfs_root *root = BTRFS_I(inode)->root;
1950 struct btrfs_ioctl_defrag_range_args *range;
1953 if (btrfs_root_readonly(root))
1956 ret = mnt_want_write(file->f_path.mnt);
1960 switch (inode->i_mode & S_IFMT) {
1962 if (!capable(CAP_SYS_ADMIN)) {
1966 ret = btrfs_defrag_root(root, 0);
1969 ret = btrfs_defrag_root(root->fs_info->extent_root, 0);
1972 if (!(file->f_mode & FMODE_WRITE)) {
1977 range = kzalloc(sizeof(*range), GFP_KERNEL);
1984 if (copy_from_user(range, argp,
1990 /* compression requires us to start the IO */
1991 if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
1992 range->flags |= BTRFS_DEFRAG_RANGE_START_IO;
1993 range->extent_thresh = (u32)-1;
1996 /* the rest are all set to zero by kzalloc */
1997 range->len = (u64)-1;
1999 ret = btrfs_defrag_file(fdentry(file)->d_inode, file,
2009 mnt_drop_write(file->f_path.mnt);
2013 static long btrfs_ioctl_add_dev(struct btrfs_root *root, void __user *arg)
2015 struct btrfs_ioctl_vol_args *vol_args;
2018 if (!capable(CAP_SYS_ADMIN))
2021 vol_args = memdup_user(arg, sizeof(*vol_args));
2022 if (IS_ERR(vol_args))
2023 return PTR_ERR(vol_args);
2025 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2026 ret = btrfs_init_new_device(root, vol_args->name);
2032 static long btrfs_ioctl_rm_dev(struct btrfs_root *root, void __user *arg)
2034 struct btrfs_ioctl_vol_args *vol_args;
2037 if (!capable(CAP_SYS_ADMIN))
2040 if (root->fs_info->sb->s_flags & MS_RDONLY)
2043 vol_args = memdup_user(arg, sizeof(*vol_args));
2044 if (IS_ERR(vol_args))
2045 return PTR_ERR(vol_args);
2047 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2048 ret = btrfs_rm_device(root, vol_args->name);
2054 static long btrfs_ioctl_fs_info(struct btrfs_root *root, void __user *arg)
2056 struct btrfs_ioctl_fs_info_args fi_args;
2057 struct btrfs_device *device;
2058 struct btrfs_device *next;
2059 struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
2061 if (!capable(CAP_SYS_ADMIN))
2064 fi_args.num_devices = fs_devices->num_devices;
2066 memcpy(&fi_args.fsid, root->fs_info->fsid, sizeof(fi_args.fsid));
2068 mutex_lock(&fs_devices->device_list_mutex);
2069 list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
2070 if (device->devid > fi_args.max_id)
2071 fi_args.max_id = device->devid;
2073 mutex_unlock(&fs_devices->device_list_mutex);
2075 if (copy_to_user(arg, &fi_args, sizeof(fi_args)))
2081 static long btrfs_ioctl_dev_info(struct btrfs_root *root, void __user *arg)
2083 struct btrfs_ioctl_dev_info_args *di_args;
2084 struct btrfs_device *dev;
2085 struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
2087 char *s_uuid = NULL;
2088 char empty_uuid[BTRFS_UUID_SIZE] = {0};
2090 if (!capable(CAP_SYS_ADMIN))
2093 di_args = memdup_user(arg, sizeof(*di_args));
2094 if (IS_ERR(di_args))
2095 return PTR_ERR(di_args);
2097 if (memcmp(empty_uuid, di_args->uuid, BTRFS_UUID_SIZE) != 0)
2098 s_uuid = di_args->uuid;
2100 mutex_lock(&fs_devices->device_list_mutex);
2101 dev = btrfs_find_device(root, di_args->devid, s_uuid, NULL);
2102 mutex_unlock(&fs_devices->device_list_mutex);
2109 di_args->devid = dev->devid;
2110 di_args->bytes_used = dev->bytes_used;
2111 di_args->total_bytes = dev->total_bytes;
2112 memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid));
2113 strncpy(di_args->path, dev->name, sizeof(di_args->path));
2116 if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args)))
2123 static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
2124 u64 off, u64 olen, u64 destoff)
2126 struct inode *inode = fdentry(file)->d_inode;
2127 struct btrfs_root *root = BTRFS_I(inode)->root;
2128 struct file *src_file;
2130 struct btrfs_trans_handle *trans;
2131 struct btrfs_path *path;
2132 struct extent_buffer *leaf;
2134 struct btrfs_key key;
2139 u64 bs = root->fs_info->sb->s_blocksize;
2144 * - split compressed inline extents. annoying: we need to
2145 * decompress into destination's address_space (the file offset
2146 * may change, so source mapping won't do), then recompress (or
2147 * otherwise reinsert) a subrange.
2148 * - allow ranges within the same file to be cloned (provided
2149 * they don't overlap)?
2152 /* the destination must be opened for writing */
2153 if (!(file->f_mode & FMODE_WRITE) || (file->f_flags & O_APPEND))
2156 if (btrfs_root_readonly(root))
2159 ret = mnt_want_write(file->f_path.mnt);
2163 src_file = fget(srcfd);
2166 goto out_drop_write;
2169 src = src_file->f_dentry->d_inode;
2175 /* the src must be open for reading */
2176 if (!(src_file->f_mode & FMODE_READ))
2180 if (S_ISDIR(src->i_mode) || S_ISDIR(inode->i_mode))
2184 if (src->i_sb != inode->i_sb || BTRFS_I(src)->root != root)
2188 buf = vmalloc(btrfs_level_size(root, 0));
2192 path = btrfs_alloc_path();
2200 mutex_lock_nested(&inode->i_mutex, I_MUTEX_PARENT);
2201 mutex_lock_nested(&src->i_mutex, I_MUTEX_CHILD);
2203 mutex_lock_nested(&src->i_mutex, I_MUTEX_PARENT);
2204 mutex_lock_nested(&inode->i_mutex, I_MUTEX_CHILD);
2207 /* determine range to clone */
2209 if (off + len > src->i_size || off + len < off)
2212 olen = len = src->i_size - off;
2213 /* if we extend to eof, continue to block boundary */
2214 if (off + len == src->i_size)
2215 len = ALIGN(src->i_size, bs) - off;
2217 /* verify the end result is block aligned */
2218 if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs) ||
2219 !IS_ALIGNED(destoff, bs))
2222 /* do any pending delalloc/csum calc on src, one way or
2223 another, and lock file content */
2225 struct btrfs_ordered_extent *ordered;
2226 lock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
2227 ordered = btrfs_lookup_first_ordered_extent(src, off+len);
2229 !test_range_bit(&BTRFS_I(src)->io_tree, off, off+len,
2230 EXTENT_DELALLOC, 0, NULL))
2232 unlock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
2234 btrfs_put_ordered_extent(ordered);
2235 btrfs_wait_ordered_range(src, off, len);
2239 key.objectid = btrfs_ino(src);
2240 key.type = BTRFS_EXTENT_DATA_KEY;
2245 * note the key will change type as we walk through the
2248 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
2252 nritems = btrfs_header_nritems(path->nodes[0]);
2253 if (path->slots[0] >= nritems) {
2254 ret = btrfs_next_leaf(root, path);
2259 nritems = btrfs_header_nritems(path->nodes[0]);
2261 leaf = path->nodes[0];
2262 slot = path->slots[0];
2264 btrfs_item_key_to_cpu(leaf, &key, slot);
2265 if (btrfs_key_type(&key) > BTRFS_EXTENT_DATA_KEY ||
2266 key.objectid != btrfs_ino(src))
2269 if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) {
2270 struct btrfs_file_extent_item *extent;
2273 struct btrfs_key new_key;
2274 u64 disko = 0, diskl = 0;
2275 u64 datao = 0, datal = 0;
2279 size = btrfs_item_size_nr(leaf, slot);
2280 read_extent_buffer(leaf, buf,
2281 btrfs_item_ptr_offset(leaf, slot),
2284 extent = btrfs_item_ptr(leaf, slot,
2285 struct btrfs_file_extent_item);
2286 comp = btrfs_file_extent_compression(leaf, extent);
2287 type = btrfs_file_extent_type(leaf, extent);
2288 if (type == BTRFS_FILE_EXTENT_REG ||
2289 type == BTRFS_FILE_EXTENT_PREALLOC) {
2290 disko = btrfs_file_extent_disk_bytenr(leaf,
2292 diskl = btrfs_file_extent_disk_num_bytes(leaf,
2294 datao = btrfs_file_extent_offset(leaf, extent);
2295 datal = btrfs_file_extent_num_bytes(leaf,
2297 } else if (type == BTRFS_FILE_EXTENT_INLINE) {
2298 /* take upper bound, may be compressed */
2299 datal = btrfs_file_extent_ram_bytes(leaf,
2302 btrfs_release_path(path);
2304 if (key.offset + datal <= off ||
2305 key.offset >= off+len)
2308 memcpy(&new_key, &key, sizeof(new_key));
2309 new_key.objectid = btrfs_ino(inode);
2310 if (off <= key.offset)
2311 new_key.offset = key.offset + destoff - off;
2313 new_key.offset = destoff;
2315 trans = btrfs_start_transaction(root, 1);
2316 if (IS_ERR(trans)) {
2317 ret = PTR_ERR(trans);
2321 if (type == BTRFS_FILE_EXTENT_REG ||
2322 type == BTRFS_FILE_EXTENT_PREALLOC) {
2323 if (off > key.offset) {
2324 datao += off - key.offset;
2325 datal -= off - key.offset;
2328 if (key.offset + datal > off + len)
2329 datal = off + len - key.offset;
2331 ret = btrfs_drop_extents(trans, inode,
2333 new_key.offset + datal,
2337 ret = btrfs_insert_empty_item(trans, root, path,
2341 leaf = path->nodes[0];
2342 slot = path->slots[0];
2343 write_extent_buffer(leaf, buf,
2344 btrfs_item_ptr_offset(leaf, slot),
2347 extent = btrfs_item_ptr(leaf, slot,
2348 struct btrfs_file_extent_item);
2350 /* disko == 0 means it's a hole */
2354 btrfs_set_file_extent_offset(leaf, extent,
2356 btrfs_set_file_extent_num_bytes(leaf, extent,
2359 inode_add_bytes(inode, datal);
2360 ret = btrfs_inc_extent_ref(trans, root,
2362 root->root_key.objectid,
2364 new_key.offset - datao);
2367 } else if (type == BTRFS_FILE_EXTENT_INLINE) {
2370 if (off > key.offset) {
2371 skip = off - key.offset;
2372 new_key.offset += skip;
2375 if (key.offset + datal > off+len)
2376 trim = key.offset + datal - (off+len);
2378 if (comp && (skip || trim)) {
2380 btrfs_end_transaction(trans, root);
2383 size -= skip + trim;
2384 datal -= skip + trim;
2386 ret = btrfs_drop_extents(trans, inode,
2388 new_key.offset + datal,
2392 ret = btrfs_insert_empty_item(trans, root, path,
2398 btrfs_file_extent_calc_inline_size(0);
2399 memmove(buf+start, buf+start+skip,
2403 leaf = path->nodes[0];
2404 slot = path->slots[0];
2405 write_extent_buffer(leaf, buf,
2406 btrfs_item_ptr_offset(leaf, slot),
2408 inode_add_bytes(inode, datal);
2411 btrfs_mark_buffer_dirty(leaf);
2412 btrfs_release_path(path);
2414 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
2417 * we round up to the block size at eof when
2418 * determining which extents to clone above,
2419 * but shouldn't round up the file size
2421 endoff = new_key.offset + datal;
2422 if (endoff > destoff+olen)
2423 endoff = destoff+olen;
2424 if (endoff > inode->i_size)
2425 btrfs_i_size_write(inode, endoff);
2427 BTRFS_I(inode)->flags = BTRFS_I(src)->flags;
2428 ret = btrfs_update_inode(trans, root, inode);
2430 btrfs_end_transaction(trans, root);
2433 btrfs_release_path(path);
2438 btrfs_release_path(path);
2439 unlock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
2441 mutex_unlock(&src->i_mutex);
2442 mutex_unlock(&inode->i_mutex);
2444 btrfs_free_path(path);
2448 mnt_drop_write(file->f_path.mnt);
2452 static long btrfs_ioctl_clone_range(struct file *file, void __user *argp)
2454 struct btrfs_ioctl_clone_range_args args;
2456 if (copy_from_user(&args, argp, sizeof(args)))
2458 return btrfs_ioctl_clone(file, args.src_fd, args.src_offset,
2459 args.src_length, args.dest_offset);
2463 * there are many ways the trans_start and trans_end ioctls can lead
2464 * to deadlocks. They should only be used by applications that
2465 * basically own the machine, and have a very in depth understanding
2466 * of all the possible deadlocks and enospc problems.
2468 static long btrfs_ioctl_trans_start(struct file *file)
2470 struct inode *inode = fdentry(file)->d_inode;
2471 struct btrfs_root *root = BTRFS_I(inode)->root;
2472 struct btrfs_trans_handle *trans;
2476 if (!capable(CAP_SYS_ADMIN))
2480 if (file->private_data)
2484 if (btrfs_root_readonly(root))
2487 ret = mnt_want_write(file->f_path.mnt);
2491 atomic_inc(&root->fs_info->open_ioctl_trans);
2494 trans = btrfs_start_ioctl_transaction(root);
2498 file->private_data = trans;
2502 atomic_dec(&root->fs_info->open_ioctl_trans);
2503 mnt_drop_write(file->f_path.mnt);
2508 static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp)
2510 struct inode *inode = fdentry(file)->d_inode;
2511 struct btrfs_root *root = BTRFS_I(inode)->root;
2512 struct btrfs_root *new_root;
2513 struct btrfs_dir_item *di;
2514 struct btrfs_trans_handle *trans;
2515 struct btrfs_path *path;
2516 struct btrfs_key location;
2517 struct btrfs_disk_key disk_key;
2518 struct btrfs_super_block *disk_super;
2523 if (!capable(CAP_SYS_ADMIN))
2526 if (copy_from_user(&objectid, argp, sizeof(objectid)))
2530 objectid = root->root_key.objectid;
2532 location.objectid = objectid;
2533 location.type = BTRFS_ROOT_ITEM_KEY;
2534 location.offset = (u64)-1;
2536 new_root = btrfs_read_fs_root_no_name(root->fs_info, &location);
2537 if (IS_ERR(new_root))
2538 return PTR_ERR(new_root);
2540 if (btrfs_root_refs(&new_root->root_item) == 0)
2543 path = btrfs_alloc_path();
2546 path->leave_spinning = 1;
2548 trans = btrfs_start_transaction(root, 1);
2549 if (IS_ERR(trans)) {
2550 btrfs_free_path(path);
2551 return PTR_ERR(trans);
2554 dir_id = btrfs_super_root_dir(&root->fs_info->super_copy);
2555 di = btrfs_lookup_dir_item(trans, root->fs_info->tree_root, path,
2556 dir_id, "default", 7, 1);
2557 if (IS_ERR_OR_NULL(di)) {
2558 btrfs_free_path(path);
2559 btrfs_end_transaction(trans, root);
2560 printk(KERN_ERR "Umm, you don't have the default dir item, "
2561 "this isn't going to work\n");
2565 btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key);
2566 btrfs_set_dir_item_key(path->nodes[0], di, &disk_key);
2567 btrfs_mark_buffer_dirty(path->nodes[0]);
2568 btrfs_free_path(path);
2570 disk_super = &root->fs_info->super_copy;
2571 features = btrfs_super_incompat_flags(disk_super);
2572 if (!(features & BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL)) {
2573 features |= BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL;
2574 btrfs_set_super_incompat_flags(disk_super, features);
2576 btrfs_end_transaction(trans, root);
2581 static void get_block_group_info(struct list_head *groups_list,
2582 struct btrfs_ioctl_space_info *space)
2584 struct btrfs_block_group_cache *block_group;
2586 space->total_bytes = 0;
2587 space->used_bytes = 0;
2589 list_for_each_entry(block_group, groups_list, list) {
2590 space->flags = block_group->flags;
2591 space->total_bytes += block_group->key.offset;
2592 space->used_bytes +=
2593 btrfs_block_group_used(&block_group->item);
2597 long btrfs_ioctl_space_info(struct btrfs_root *root, void __user *arg)
2599 struct btrfs_ioctl_space_args space_args;
2600 struct btrfs_ioctl_space_info space;
2601 struct btrfs_ioctl_space_info *dest;
2602 struct btrfs_ioctl_space_info *dest_orig;
2603 struct btrfs_ioctl_space_info __user *user_dest;
2604 struct btrfs_space_info *info;
2605 u64 types[] = {BTRFS_BLOCK_GROUP_DATA,
2606 BTRFS_BLOCK_GROUP_SYSTEM,
2607 BTRFS_BLOCK_GROUP_METADATA,
2608 BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA};
2615 if (copy_from_user(&space_args,
2616 (struct btrfs_ioctl_space_args __user *)arg,
2617 sizeof(space_args)))
2620 for (i = 0; i < num_types; i++) {
2621 struct btrfs_space_info *tmp;
2625 list_for_each_entry_rcu(tmp, &root->fs_info->space_info,
2627 if (tmp->flags == types[i]) {
2637 down_read(&info->groups_sem);
2638 for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
2639 if (!list_empty(&info->block_groups[c]))
2642 up_read(&info->groups_sem);
2645 /* space_slots == 0 means they are asking for a count */
2646 if (space_args.space_slots == 0) {
2647 space_args.total_spaces = slot_count;
2651 slot_count = min_t(u64, space_args.space_slots, slot_count);
2653 alloc_size = sizeof(*dest) * slot_count;
2655 /* we generally have at most 6 or so space infos, one for each raid
2656 * level. So, a whole page should be more than enough for everyone
2658 if (alloc_size > PAGE_CACHE_SIZE)
2661 space_args.total_spaces = 0;
2662 dest = kmalloc(alloc_size, GFP_NOFS);
2667 /* now we have a buffer to copy into */
2668 for (i = 0; i < num_types; i++) {
2669 struct btrfs_space_info *tmp;
2676 list_for_each_entry_rcu(tmp, &root->fs_info->space_info,
2678 if (tmp->flags == types[i]) {
2687 down_read(&info->groups_sem);
2688 for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
2689 if (!list_empty(&info->block_groups[c])) {
2690 get_block_group_info(&info->block_groups[c],
2692 memcpy(dest, &space, sizeof(space));
2694 space_args.total_spaces++;
2700 up_read(&info->groups_sem);
2703 user_dest = (struct btrfs_ioctl_space_info *)
2704 (arg + sizeof(struct btrfs_ioctl_space_args));
2706 if (copy_to_user(user_dest, dest_orig, alloc_size))
2711 if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args)))
2718 * there are many ways the trans_start and trans_end ioctls can lead
2719 * to deadlocks. They should only be used by applications that
2720 * basically own the machine, and have a very in depth understanding
2721 * of all the possible deadlocks and enospc problems.
2723 long btrfs_ioctl_trans_end(struct file *file)
2725 struct inode *inode = fdentry(file)->d_inode;
2726 struct btrfs_root *root = BTRFS_I(inode)->root;
2727 struct btrfs_trans_handle *trans;
2729 trans = file->private_data;
2732 file->private_data = NULL;
2734 btrfs_end_transaction(trans, root);
2736 atomic_dec(&root->fs_info->open_ioctl_trans);
2738 mnt_drop_write(file->f_path.mnt);
2742 static noinline long btrfs_ioctl_start_sync(struct file *file, void __user *argp)
2744 struct btrfs_root *root = BTRFS_I(file->f_dentry->d_inode)->root;
2745 struct btrfs_trans_handle *trans;
2749 trans = btrfs_start_transaction(root, 0);
2751 return PTR_ERR(trans);
2752 transid = trans->transid;
2753 ret = btrfs_commit_transaction_async(trans, root, 0);
2755 btrfs_end_transaction(trans, root);
2760 if (copy_to_user(argp, &transid, sizeof(transid)))
2765 static noinline long btrfs_ioctl_wait_sync(struct file *file, void __user *argp)
2767 struct btrfs_root *root = BTRFS_I(file->f_dentry->d_inode)->root;
2771 if (copy_from_user(&transid, argp, sizeof(transid)))
2774 transid = 0; /* current trans */
2776 return btrfs_wait_for_commit(root, transid);
2779 static long btrfs_ioctl_scrub(struct btrfs_root *root, void __user *arg)
2782 struct btrfs_ioctl_scrub_args *sa;
2784 if (!capable(CAP_SYS_ADMIN))
2787 sa = memdup_user(arg, sizeof(*sa));
2791 ret = btrfs_scrub_dev(root, sa->devid, sa->start, sa->end,
2792 &sa->progress, sa->flags & BTRFS_SCRUB_READONLY);
2794 if (copy_to_user(arg, sa, sizeof(*sa)))
2801 static long btrfs_ioctl_scrub_cancel(struct btrfs_root *root, void __user *arg)
2803 if (!capable(CAP_SYS_ADMIN))
2806 return btrfs_scrub_cancel(root);
2809 static long btrfs_ioctl_scrub_progress(struct btrfs_root *root,
2812 struct btrfs_ioctl_scrub_args *sa;
2815 if (!capable(CAP_SYS_ADMIN))
2818 sa = memdup_user(arg, sizeof(*sa));
2822 ret = btrfs_scrub_progress(root, sa->devid, &sa->progress);
2824 if (copy_to_user(arg, sa, sizeof(*sa)))
2831 long btrfs_ioctl(struct file *file, unsigned int
2832 cmd, unsigned long arg)
2834 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
2835 void __user *argp = (void __user *)arg;
2838 case FS_IOC_GETFLAGS:
2839 return btrfs_ioctl_getflags(file, argp);
2840 case FS_IOC_SETFLAGS:
2841 return btrfs_ioctl_setflags(file, argp);
2842 case FS_IOC_GETVERSION:
2843 return btrfs_ioctl_getversion(file, argp);
2845 return btrfs_ioctl_fitrim(file, argp);
2846 case BTRFS_IOC_SNAP_CREATE:
2847 return btrfs_ioctl_snap_create(file, argp, 0);
2848 case BTRFS_IOC_SNAP_CREATE_V2:
2849 return btrfs_ioctl_snap_create_v2(file, argp, 0);
2850 case BTRFS_IOC_SUBVOL_CREATE:
2851 return btrfs_ioctl_snap_create(file, argp, 1);
2852 case BTRFS_IOC_SNAP_DESTROY:
2853 return btrfs_ioctl_snap_destroy(file, argp);
2854 case BTRFS_IOC_SUBVOL_GETFLAGS:
2855 return btrfs_ioctl_subvol_getflags(file, argp);
2856 case BTRFS_IOC_SUBVOL_SETFLAGS:
2857 return btrfs_ioctl_subvol_setflags(file, argp);
2858 case BTRFS_IOC_DEFAULT_SUBVOL:
2859 return btrfs_ioctl_default_subvol(file, argp);
2860 case BTRFS_IOC_DEFRAG:
2861 return btrfs_ioctl_defrag(file, NULL);
2862 case BTRFS_IOC_DEFRAG_RANGE:
2863 return btrfs_ioctl_defrag(file, argp);
2864 case BTRFS_IOC_RESIZE:
2865 return btrfs_ioctl_resize(root, argp);
2866 case BTRFS_IOC_ADD_DEV:
2867 return btrfs_ioctl_add_dev(root, argp);
2868 case BTRFS_IOC_RM_DEV:
2869 return btrfs_ioctl_rm_dev(root, argp);
2870 case BTRFS_IOC_FS_INFO:
2871 return btrfs_ioctl_fs_info(root, argp);
2872 case BTRFS_IOC_DEV_INFO:
2873 return btrfs_ioctl_dev_info(root, argp);
2874 case BTRFS_IOC_BALANCE:
2875 return btrfs_balance(root->fs_info->dev_root);
2876 case BTRFS_IOC_CLONE:
2877 return btrfs_ioctl_clone(file, arg, 0, 0, 0);
2878 case BTRFS_IOC_CLONE_RANGE:
2879 return btrfs_ioctl_clone_range(file, argp);
2880 case BTRFS_IOC_TRANS_START:
2881 return btrfs_ioctl_trans_start(file);
2882 case BTRFS_IOC_TRANS_END:
2883 return btrfs_ioctl_trans_end(file);
2884 case BTRFS_IOC_TREE_SEARCH:
2885 return btrfs_ioctl_tree_search(file, argp);
2886 case BTRFS_IOC_INO_LOOKUP:
2887 return btrfs_ioctl_ino_lookup(file, argp);
2888 case BTRFS_IOC_SPACE_INFO:
2889 return btrfs_ioctl_space_info(root, argp);
2890 case BTRFS_IOC_SYNC:
2891 btrfs_sync_fs(file->f_dentry->d_sb, 1);
2893 case BTRFS_IOC_START_SYNC:
2894 return btrfs_ioctl_start_sync(file, argp);
2895 case BTRFS_IOC_WAIT_SYNC:
2896 return btrfs_ioctl_wait_sync(file, argp);
2897 case BTRFS_IOC_SCRUB:
2898 return btrfs_ioctl_scrub(root, argp);
2899 case BTRFS_IOC_SCRUB_CANCEL:
2900 return btrfs_ioctl_scrub_cancel(root, argp);
2901 case BTRFS_IOC_SCRUB_PROGRESS:
2902 return btrfs_ioctl_scrub_progress(root, argp);