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 spin_lock(&root->fs_info->trans_lock);
486 list_add(&pending_snapshot->list,
487 &trans->transaction->pending_snapshots);
488 spin_unlock(&root->fs_info->trans_lock);
490 *async_transid = trans->transid;
491 ret = btrfs_commit_transaction_async(trans,
492 root->fs_info->extent_root, 1);
494 ret = btrfs_commit_transaction(trans,
495 root->fs_info->extent_root);
499 ret = pending_snapshot->error;
503 ret = btrfs_orphan_cleanup(pending_snapshot->snap);
507 parent = dget_parent(dentry);
508 inode = btrfs_lookup_dentry(parent->d_inode, dentry);
511 ret = PTR_ERR(inode);
515 d_instantiate(dentry, inode);
518 kfree(pending_snapshot);
522 /* copy of check_sticky in fs/namei.c()
523 * It's inline, so penalty for filesystems that don't use sticky bit is
526 static inline int btrfs_check_sticky(struct inode *dir, struct inode *inode)
528 uid_t fsuid = current_fsuid();
530 if (!(dir->i_mode & S_ISVTX))
532 if (inode->i_uid == fsuid)
534 if (dir->i_uid == fsuid)
536 return !capable(CAP_FOWNER);
539 /* copy of may_delete in fs/namei.c()
540 * Check whether we can remove a link victim from directory dir, check
541 * whether the type of victim is right.
542 * 1. We can't do it if dir is read-only (done in permission())
543 * 2. We should have write and exec permissions on dir
544 * 3. We can't remove anything from append-only dir
545 * 4. We can't do anything with immutable dir (done in permission())
546 * 5. If the sticky bit on dir is set we should either
547 * a. be owner of dir, or
548 * b. be owner of victim, or
549 * c. have CAP_FOWNER capability
550 * 6. If the victim is append-only or immutable we can't do antyhing with
551 * links pointing to it.
552 * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR.
553 * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR.
554 * 9. We can't remove a root or mountpoint.
555 * 10. We don't allow removal of NFS sillyrenamed files; it's handled by
556 * nfs_async_unlink().
559 static int btrfs_may_delete(struct inode *dir,struct dentry *victim,int isdir)
563 if (!victim->d_inode)
566 BUG_ON(victim->d_parent->d_inode != dir);
567 audit_inode_child(victim, dir);
569 error = inode_permission(dir, MAY_WRITE | MAY_EXEC);
574 if (btrfs_check_sticky(dir, victim->d_inode)||
575 IS_APPEND(victim->d_inode)||
576 IS_IMMUTABLE(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
579 if (!S_ISDIR(victim->d_inode->i_mode))
583 } else if (S_ISDIR(victim->d_inode->i_mode))
587 if (victim->d_flags & DCACHE_NFSFS_RENAMED)
592 /* copy of may_create in fs/namei.c() */
593 static inline int btrfs_may_create(struct inode *dir, struct dentry *child)
599 return inode_permission(dir, MAY_WRITE | MAY_EXEC);
603 * Create a new subvolume below @parent. This is largely modeled after
604 * sys_mkdirat and vfs_mkdir, but we only do a single component lookup
605 * inside this filesystem so it's quite a bit simpler.
607 static noinline int btrfs_mksubvol(struct path *parent,
608 char *name, int namelen,
609 struct btrfs_root *snap_src,
610 u64 *async_transid, bool readonly)
612 struct inode *dir = parent->dentry->d_inode;
613 struct dentry *dentry;
616 mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
618 dentry = lookup_one_len(name, parent->dentry, namelen);
619 error = PTR_ERR(dentry);
627 error = mnt_want_write(parent->mnt);
631 error = btrfs_may_create(dir, dentry);
635 down_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
637 if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0)
641 error = create_snapshot(snap_src, dentry,
642 name, namelen, async_transid, readonly);
644 error = create_subvol(BTRFS_I(dir)->root, dentry,
645 name, namelen, async_transid);
648 fsnotify_mkdir(dir, dentry);
650 up_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
652 mnt_drop_write(parent->mnt);
656 mutex_unlock(&dir->i_mutex);
661 * When we're defragging a range, we don't want to kick it off again
662 * if it is really just waiting for delalloc to send it down.
663 * If we find a nice big extent or delalloc range for the bytes in the
664 * file you want to defrag, we return 0 to let you know to skip this
667 static int check_defrag_in_cache(struct inode *inode, u64 offset, int thresh)
669 struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
670 struct extent_map *em = NULL;
671 struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
674 read_lock(&em_tree->lock);
675 em = lookup_extent_mapping(em_tree, offset, PAGE_CACHE_SIZE);
676 read_unlock(&em_tree->lock);
679 end = extent_map_end(em);
681 if (end - offset > thresh)
684 /* if we already have a nice delalloc here, just stop */
686 end = count_range_bits(io_tree, &offset, offset + thresh,
687 thresh, EXTENT_DELALLOC, 1);
694 * helper function to walk through a file and find extents
695 * newer than a specific transid, and smaller than thresh.
697 * This is used by the defragging code to find new and small
700 static int find_new_extents(struct btrfs_root *root,
701 struct inode *inode, u64 newer_than,
702 u64 *off, int thresh)
704 struct btrfs_path *path;
705 struct btrfs_key min_key;
706 struct btrfs_key max_key;
707 struct extent_buffer *leaf;
708 struct btrfs_file_extent_item *extent;
711 u64 ino = btrfs_ino(inode);
713 path = btrfs_alloc_path();
717 min_key.objectid = ino;
718 min_key.type = BTRFS_EXTENT_DATA_KEY;
719 min_key.offset = *off;
721 max_key.objectid = ino;
722 max_key.type = (u8)-1;
723 max_key.offset = (u64)-1;
725 path->keep_locks = 1;
728 ret = btrfs_search_forward(root, &min_key, &max_key,
729 path, 0, newer_than);
732 if (min_key.objectid != ino)
734 if (min_key.type != BTRFS_EXTENT_DATA_KEY)
737 leaf = path->nodes[0];
738 extent = btrfs_item_ptr(leaf, path->slots[0],
739 struct btrfs_file_extent_item);
741 type = btrfs_file_extent_type(leaf, extent);
742 if (type == BTRFS_FILE_EXTENT_REG &&
743 btrfs_file_extent_num_bytes(leaf, extent) < thresh &&
744 check_defrag_in_cache(inode, min_key.offset, thresh)) {
745 *off = min_key.offset;
746 btrfs_free_path(path);
750 if (min_key.offset == (u64)-1)
754 btrfs_release_path(path);
757 btrfs_free_path(path);
761 static int should_defrag_range(struct inode *inode, u64 start, u64 len,
762 int thresh, u64 *last_len, u64 *skip,
765 struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
766 struct extent_map *em = NULL;
767 struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
771 * make sure that once we start defragging and extent, we keep on
774 if (start < *defrag_end)
780 * hopefully we have this extent in the tree already, try without
781 * the full extent lock
783 read_lock(&em_tree->lock);
784 em = lookup_extent_mapping(em_tree, start, len);
785 read_unlock(&em_tree->lock);
788 /* get the big lock and read metadata off disk */
789 lock_extent(io_tree, start, start + len - 1, GFP_NOFS);
790 em = btrfs_get_extent(inode, NULL, 0, start, len, 0);
791 unlock_extent(io_tree, start, start + len - 1, GFP_NOFS);
797 /* this will cover holes, and inline extents */
798 if (em->block_start >= EXTENT_MAP_LAST_BYTE)
802 * we hit a real extent, if it is big don't bother defragging it again
804 if ((*last_len == 0 || *last_len >= thresh) && em->len >= thresh)
808 * last_len ends up being a counter of how many bytes we've defragged.
809 * every time we choose not to defrag an extent, we reset *last_len
810 * so that the next tiny extent will force a defrag.
812 * The end result of this is that tiny extents before a single big
813 * extent will force at least part of that big extent to be defragged.
817 *defrag_end = extent_map_end(em);
820 *skip = extent_map_end(em);
829 * it doesn't do much good to defrag one or two pages
830 * at a time. This pulls in a nice chunk of pages
833 * It also makes sure the delalloc code has enough
834 * dirty data to avoid making new small extents as part
837 * It's a good idea to start RA on this range
838 * before calling this.
840 static int cluster_pages_for_defrag(struct inode *inode,
842 unsigned long start_index,
845 unsigned long file_end;
846 u64 isize = i_size_read(inode);
852 struct btrfs_ordered_extent *ordered;
853 struct extent_state *cached_state = NULL;
857 file_end = (isize - 1) >> PAGE_CACHE_SHIFT;
859 ret = btrfs_delalloc_reserve_space(inode,
860 num_pages << PAGE_CACHE_SHIFT);
867 /* step one, lock all the pages */
868 for (i = 0; i < num_pages; i++) {
870 page = find_or_create_page(inode->i_mapping,
871 start_index + i, GFP_NOFS);
875 if (!PageUptodate(page)) {
876 btrfs_readpage(NULL, page);
878 if (!PageUptodate(page)) {
880 page_cache_release(page);
885 isize = i_size_read(inode);
886 file_end = (isize - 1) >> PAGE_CACHE_SHIFT;
887 if (!isize || page->index > file_end ||
888 page->mapping != inode->i_mapping) {
889 /* whoops, we blew past eof, skip this page */
891 page_cache_release(page);
900 if (!(inode->i_sb->s_flags & MS_ACTIVE))
904 * so now we have a nice long stream of locked
905 * and up to date pages, lets wait on them
907 for (i = 0; i < i_done; i++)
908 wait_on_page_writeback(pages[i]);
910 page_start = page_offset(pages[0]);
911 page_end = page_offset(pages[i_done - 1]) + PAGE_CACHE_SIZE;
913 lock_extent_bits(&BTRFS_I(inode)->io_tree,
914 page_start, page_end - 1, 0, &cached_state,
916 ordered = btrfs_lookup_first_ordered_extent(inode, page_end - 1);
918 ordered->file_offset + ordered->len > page_start &&
919 ordered->file_offset < page_end) {
920 btrfs_put_ordered_extent(ordered);
921 unlock_extent_cached(&BTRFS_I(inode)->io_tree,
922 page_start, page_end - 1,
923 &cached_state, GFP_NOFS);
924 for (i = 0; i < i_done; i++) {
925 unlock_page(pages[i]);
926 page_cache_release(pages[i]);
928 btrfs_wait_ordered_range(inode, page_start,
929 page_end - page_start);
933 btrfs_put_ordered_extent(ordered);
935 clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start,
936 page_end - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
937 EXTENT_DO_ACCOUNTING, 0, 0, &cached_state,
940 if (i_done != num_pages) {
941 spin_lock(&BTRFS_I(inode)->lock);
942 BTRFS_I(inode)->outstanding_extents++;
943 spin_unlock(&BTRFS_I(inode)->lock);
944 btrfs_delalloc_release_space(inode,
945 (num_pages - i_done) << PAGE_CACHE_SHIFT);
949 btrfs_set_extent_delalloc(inode, page_start, page_end - 1,
952 unlock_extent_cached(&BTRFS_I(inode)->io_tree,
953 page_start, page_end - 1, &cached_state,
956 for (i = 0; i < i_done; i++) {
957 clear_page_dirty_for_io(pages[i]);
958 ClearPageChecked(pages[i]);
959 set_page_extent_mapped(pages[i]);
960 set_page_dirty(pages[i]);
961 unlock_page(pages[i]);
962 page_cache_release(pages[i]);
966 for (i = 0; i < i_done; i++) {
967 unlock_page(pages[i]);
968 page_cache_release(pages[i]);
970 btrfs_delalloc_release_space(inode, num_pages << PAGE_CACHE_SHIFT);
975 int btrfs_defrag_file(struct inode *inode, struct file *file,
976 struct btrfs_ioctl_defrag_range_args *range,
977 u64 newer_than, unsigned long max_to_defrag)
979 struct btrfs_root *root = BTRFS_I(inode)->root;
980 struct btrfs_super_block *disk_super;
981 struct file_ra_state *ra = NULL;
982 unsigned long last_index;
987 u64 newer_off = range->start;
991 int defrag_count = 0;
992 int compress_type = BTRFS_COMPRESS_ZLIB;
993 int extent_thresh = range->extent_thresh;
994 int newer_cluster = (256 * 1024) >> PAGE_CACHE_SHIFT;
995 u64 new_align = ~((u64)128 * 1024 - 1);
996 struct page **pages = NULL;
998 if (extent_thresh == 0)
999 extent_thresh = 256 * 1024;
1001 if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS) {
1002 if (range->compress_type > BTRFS_COMPRESS_TYPES)
1004 if (range->compress_type)
1005 compress_type = range->compress_type;
1008 if (inode->i_size == 0)
1012 * if we were not given a file, allocate a readahead
1016 ra = kzalloc(sizeof(*ra), GFP_NOFS);
1019 file_ra_state_init(ra, inode->i_mapping);
1024 pages = kmalloc(sizeof(struct page *) * newer_cluster,
1031 /* find the last page to defrag */
1032 if (range->start + range->len > range->start) {
1033 last_index = min_t(u64, inode->i_size - 1,
1034 range->start + range->len - 1) >> PAGE_CACHE_SHIFT;
1036 last_index = (inode->i_size - 1) >> PAGE_CACHE_SHIFT;
1040 ret = find_new_extents(root, inode, newer_than,
1041 &newer_off, 64 * 1024);
1043 range->start = newer_off;
1045 * we always align our defrag to help keep
1046 * the extents in the file evenly spaced
1048 i = (newer_off & new_align) >> PAGE_CACHE_SHIFT;
1049 newer_left = newer_cluster;
1053 i = range->start >> PAGE_CACHE_SHIFT;
1056 max_to_defrag = last_index - 1;
1059 * make writeback starts from i, so the defrag range can be
1060 * written sequentially.
1062 if (i < inode->i_mapping->writeback_index)
1063 inode->i_mapping->writeback_index = i;
1065 while (i <= last_index && defrag_count < max_to_defrag) {
1067 * make sure we stop running if someone unmounts
1070 if (!(inode->i_sb->s_flags & MS_ACTIVE))
1074 !should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT,
1081 * the should_defrag function tells us how much to skip
1082 * bump our counter by the suggested amount
1084 next = (skip + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
1085 i = max(i + 1, next);
1088 if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)
1089 BTRFS_I(inode)->force_compress = compress_type;
1091 btrfs_force_ra(inode->i_mapping, ra, file, i, newer_cluster);
1093 ret = cluster_pages_for_defrag(inode, pages, i, newer_cluster);
1097 defrag_count += ret;
1098 balance_dirty_pages_ratelimited_nr(inode->i_mapping, ret);
1102 if (newer_off == (u64)-1)
1105 newer_off = max(newer_off + 1,
1106 (u64)i << PAGE_CACHE_SHIFT);
1108 ret = find_new_extents(root, inode,
1109 newer_than, &newer_off,
1112 range->start = newer_off;
1113 i = (newer_off & new_align) >> PAGE_CACHE_SHIFT;
1114 newer_left = newer_cluster;
1123 if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO))
1124 filemap_flush(inode->i_mapping);
1126 if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
1127 /* the filemap_flush will queue IO into the worker threads, but
1128 * we have to make sure the IO is actually started and that
1129 * ordered extents get created before we return
1131 atomic_inc(&root->fs_info->async_submit_draining);
1132 while (atomic_read(&root->fs_info->nr_async_submits) ||
1133 atomic_read(&root->fs_info->async_delalloc_pages)) {
1134 wait_event(root->fs_info->async_submit_wait,
1135 (atomic_read(&root->fs_info->nr_async_submits) == 0 &&
1136 atomic_read(&root->fs_info->async_delalloc_pages) == 0));
1138 atomic_dec(&root->fs_info->async_submit_draining);
1140 mutex_lock(&inode->i_mutex);
1141 BTRFS_I(inode)->force_compress = BTRFS_COMPRESS_NONE;
1142 mutex_unlock(&inode->i_mutex);
1145 disk_super = &root->fs_info->super_copy;
1146 features = btrfs_super_incompat_flags(disk_super);
1147 if (range->compress_type == BTRFS_COMPRESS_LZO) {
1148 features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
1149 btrfs_set_super_incompat_flags(disk_super, features);
1154 return defrag_count;
1163 static noinline int btrfs_ioctl_resize(struct btrfs_root *root,
1169 struct btrfs_ioctl_vol_args *vol_args;
1170 struct btrfs_trans_handle *trans;
1171 struct btrfs_device *device = NULL;
1173 char *devstr = NULL;
1177 if (root->fs_info->sb->s_flags & MS_RDONLY)
1180 if (!capable(CAP_SYS_ADMIN))
1183 vol_args = memdup_user(arg, sizeof(*vol_args));
1184 if (IS_ERR(vol_args))
1185 return PTR_ERR(vol_args);
1187 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
1189 mutex_lock(&root->fs_info->volume_mutex);
1190 sizestr = vol_args->name;
1191 devstr = strchr(sizestr, ':');
1194 sizestr = devstr + 1;
1196 devstr = vol_args->name;
1197 devid = simple_strtoull(devstr, &end, 10);
1198 printk(KERN_INFO "resizing devid %llu\n",
1199 (unsigned long long)devid);
1201 device = btrfs_find_device(root, devid, NULL, NULL);
1203 printk(KERN_INFO "resizer unable to find device %llu\n",
1204 (unsigned long long)devid);
1208 if (!strcmp(sizestr, "max"))
1209 new_size = device->bdev->bd_inode->i_size;
1211 if (sizestr[0] == '-') {
1214 } else if (sizestr[0] == '+') {
1218 new_size = memparse(sizestr, NULL);
1219 if (new_size == 0) {
1225 old_size = device->total_bytes;
1228 if (new_size > old_size) {
1232 new_size = old_size - new_size;
1233 } else if (mod > 0) {
1234 new_size = old_size + new_size;
1237 if (new_size < 256 * 1024 * 1024) {
1241 if (new_size > device->bdev->bd_inode->i_size) {
1246 do_div(new_size, root->sectorsize);
1247 new_size *= root->sectorsize;
1249 printk(KERN_INFO "new size for %s is %llu\n",
1250 device->name, (unsigned long long)new_size);
1252 if (new_size > old_size) {
1253 trans = btrfs_start_transaction(root, 0);
1254 if (IS_ERR(trans)) {
1255 ret = PTR_ERR(trans);
1258 ret = btrfs_grow_device(trans, device, new_size);
1259 btrfs_commit_transaction(trans, root);
1261 ret = btrfs_shrink_device(device, new_size);
1265 mutex_unlock(&root->fs_info->volume_mutex);
1270 static noinline int btrfs_ioctl_snap_create_transid(struct file *file,
1277 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
1278 struct file *src_file;
1282 if (root->fs_info->sb->s_flags & MS_RDONLY)
1285 namelen = strlen(name);
1286 if (strchr(name, '/')) {
1292 ret = btrfs_mksubvol(&file->f_path, name, namelen,
1293 NULL, transid, readonly);
1295 struct inode *src_inode;
1296 src_file = fget(fd);
1302 src_inode = src_file->f_path.dentry->d_inode;
1303 if (src_inode->i_sb != file->f_path.dentry->d_inode->i_sb) {
1304 printk(KERN_INFO "btrfs: Snapshot src from "
1310 ret = btrfs_mksubvol(&file->f_path, name, namelen,
1311 BTRFS_I(src_inode)->root,
1319 static noinline int btrfs_ioctl_snap_create(struct file *file,
1320 void __user *arg, int subvol)
1322 struct btrfs_ioctl_vol_args *vol_args;
1325 vol_args = memdup_user(arg, sizeof(*vol_args));
1326 if (IS_ERR(vol_args))
1327 return PTR_ERR(vol_args);
1328 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
1330 ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
1331 vol_args->fd, subvol,
1338 static noinline int btrfs_ioctl_snap_create_v2(struct file *file,
1339 void __user *arg, int subvol)
1341 struct btrfs_ioctl_vol_args_v2 *vol_args;
1345 bool readonly = false;
1347 vol_args = memdup_user(arg, sizeof(*vol_args));
1348 if (IS_ERR(vol_args))
1349 return PTR_ERR(vol_args);
1350 vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
1352 if (vol_args->flags &
1353 ~(BTRFS_SUBVOL_CREATE_ASYNC | BTRFS_SUBVOL_RDONLY)) {
1358 if (vol_args->flags & BTRFS_SUBVOL_CREATE_ASYNC)
1360 if (vol_args->flags & BTRFS_SUBVOL_RDONLY)
1363 ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
1364 vol_args->fd, subvol,
1367 if (ret == 0 && ptr &&
1369 offsetof(struct btrfs_ioctl_vol_args_v2,
1370 transid), ptr, sizeof(*ptr)))
1377 static noinline int btrfs_ioctl_subvol_getflags(struct file *file,
1380 struct inode *inode = fdentry(file)->d_inode;
1381 struct btrfs_root *root = BTRFS_I(inode)->root;
1385 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID)
1388 down_read(&root->fs_info->subvol_sem);
1389 if (btrfs_root_readonly(root))
1390 flags |= BTRFS_SUBVOL_RDONLY;
1391 up_read(&root->fs_info->subvol_sem);
1393 if (copy_to_user(arg, &flags, sizeof(flags)))
1399 static noinline int btrfs_ioctl_subvol_setflags(struct file *file,
1402 struct inode *inode = fdentry(file)->d_inode;
1403 struct btrfs_root *root = BTRFS_I(inode)->root;
1404 struct btrfs_trans_handle *trans;
1409 if (root->fs_info->sb->s_flags & MS_RDONLY)
1412 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID)
1415 if (copy_from_user(&flags, arg, sizeof(flags)))
1418 if (flags & BTRFS_SUBVOL_CREATE_ASYNC)
1421 if (flags & ~BTRFS_SUBVOL_RDONLY)
1424 if (!inode_owner_or_capable(inode))
1427 down_write(&root->fs_info->subvol_sem);
1430 if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root))
1433 root_flags = btrfs_root_flags(&root->root_item);
1434 if (flags & BTRFS_SUBVOL_RDONLY)
1435 btrfs_set_root_flags(&root->root_item,
1436 root_flags | BTRFS_ROOT_SUBVOL_RDONLY);
1438 btrfs_set_root_flags(&root->root_item,
1439 root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY);
1441 trans = btrfs_start_transaction(root, 1);
1442 if (IS_ERR(trans)) {
1443 ret = PTR_ERR(trans);
1447 ret = btrfs_update_root(trans, root->fs_info->tree_root,
1448 &root->root_key, &root->root_item);
1450 btrfs_commit_transaction(trans, root);
1453 btrfs_set_root_flags(&root->root_item, root_flags);
1455 up_write(&root->fs_info->subvol_sem);
1460 * helper to check if the subvolume references other subvolumes
1462 static noinline int may_destroy_subvol(struct btrfs_root *root)
1464 struct btrfs_path *path;
1465 struct btrfs_key key;
1468 path = btrfs_alloc_path();
1472 key.objectid = root->root_key.objectid;
1473 key.type = BTRFS_ROOT_REF_KEY;
1474 key.offset = (u64)-1;
1476 ret = btrfs_search_slot(NULL, root->fs_info->tree_root,
1483 if (path->slots[0] > 0) {
1485 btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
1486 if (key.objectid == root->root_key.objectid &&
1487 key.type == BTRFS_ROOT_REF_KEY)
1491 btrfs_free_path(path);
1495 static noinline int key_in_sk(struct btrfs_key *key,
1496 struct btrfs_ioctl_search_key *sk)
1498 struct btrfs_key test;
1501 test.objectid = sk->min_objectid;
1502 test.type = sk->min_type;
1503 test.offset = sk->min_offset;
1505 ret = btrfs_comp_cpu_keys(key, &test);
1509 test.objectid = sk->max_objectid;
1510 test.type = sk->max_type;
1511 test.offset = sk->max_offset;
1513 ret = btrfs_comp_cpu_keys(key, &test);
1519 static noinline int copy_to_sk(struct btrfs_root *root,
1520 struct btrfs_path *path,
1521 struct btrfs_key *key,
1522 struct btrfs_ioctl_search_key *sk,
1524 unsigned long *sk_offset,
1528 struct extent_buffer *leaf;
1529 struct btrfs_ioctl_search_header sh;
1530 unsigned long item_off;
1531 unsigned long item_len;
1537 leaf = path->nodes[0];
1538 slot = path->slots[0];
1539 nritems = btrfs_header_nritems(leaf);
1541 if (btrfs_header_generation(leaf) > sk->max_transid) {
1545 found_transid = btrfs_header_generation(leaf);
1547 for (i = slot; i < nritems; i++) {
1548 item_off = btrfs_item_ptr_offset(leaf, i);
1549 item_len = btrfs_item_size_nr(leaf, i);
1551 if (item_len > BTRFS_SEARCH_ARGS_BUFSIZE)
1554 if (sizeof(sh) + item_len + *sk_offset >
1555 BTRFS_SEARCH_ARGS_BUFSIZE) {
1560 btrfs_item_key_to_cpu(leaf, key, i);
1561 if (!key_in_sk(key, sk))
1564 sh.objectid = key->objectid;
1565 sh.offset = key->offset;
1566 sh.type = key->type;
1568 sh.transid = found_transid;
1570 /* copy search result header */
1571 memcpy(buf + *sk_offset, &sh, sizeof(sh));
1572 *sk_offset += sizeof(sh);
1575 char *p = buf + *sk_offset;
1577 read_extent_buffer(leaf, p,
1578 item_off, item_len);
1579 *sk_offset += item_len;
1583 if (*num_found >= sk->nr_items)
1588 if (key->offset < (u64)-1 && key->offset < sk->max_offset)
1590 else if (key->type < (u8)-1 && key->type < sk->max_type) {
1593 } else if (key->objectid < (u64)-1 && key->objectid < sk->max_objectid) {
1603 static noinline int search_ioctl(struct inode *inode,
1604 struct btrfs_ioctl_search_args *args)
1606 struct btrfs_root *root;
1607 struct btrfs_key key;
1608 struct btrfs_key max_key;
1609 struct btrfs_path *path;
1610 struct btrfs_ioctl_search_key *sk = &args->key;
1611 struct btrfs_fs_info *info = BTRFS_I(inode)->root->fs_info;
1614 unsigned long sk_offset = 0;
1616 path = btrfs_alloc_path();
1620 if (sk->tree_id == 0) {
1621 /* search the root of the inode that was passed */
1622 root = BTRFS_I(inode)->root;
1624 key.objectid = sk->tree_id;
1625 key.type = BTRFS_ROOT_ITEM_KEY;
1626 key.offset = (u64)-1;
1627 root = btrfs_read_fs_root_no_name(info, &key);
1629 printk(KERN_ERR "could not find root %llu\n",
1631 btrfs_free_path(path);
1636 key.objectid = sk->min_objectid;
1637 key.type = sk->min_type;
1638 key.offset = sk->min_offset;
1640 max_key.objectid = sk->max_objectid;
1641 max_key.type = sk->max_type;
1642 max_key.offset = sk->max_offset;
1644 path->keep_locks = 1;
1647 ret = btrfs_search_forward(root, &key, &max_key, path, 0,
1654 ret = copy_to_sk(root, path, &key, sk, args->buf,
1655 &sk_offset, &num_found);
1656 btrfs_release_path(path);
1657 if (ret || num_found >= sk->nr_items)
1663 sk->nr_items = num_found;
1664 btrfs_free_path(path);
1668 static noinline int btrfs_ioctl_tree_search(struct file *file,
1671 struct btrfs_ioctl_search_args *args;
1672 struct inode *inode;
1675 if (!capable(CAP_SYS_ADMIN))
1678 args = memdup_user(argp, sizeof(*args));
1680 return PTR_ERR(args);
1682 inode = fdentry(file)->d_inode;
1683 ret = search_ioctl(inode, args);
1684 if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
1691 * Search INODE_REFs to identify path name of 'dirid' directory
1692 * in a 'tree_id' tree. and sets path name to 'name'.
1694 static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info,
1695 u64 tree_id, u64 dirid, char *name)
1697 struct btrfs_root *root;
1698 struct btrfs_key key;
1704 struct btrfs_inode_ref *iref;
1705 struct extent_buffer *l;
1706 struct btrfs_path *path;
1708 if (dirid == BTRFS_FIRST_FREE_OBJECTID) {
1713 path = btrfs_alloc_path();
1717 ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX];
1719 key.objectid = tree_id;
1720 key.type = BTRFS_ROOT_ITEM_KEY;
1721 key.offset = (u64)-1;
1722 root = btrfs_read_fs_root_no_name(info, &key);
1724 printk(KERN_ERR "could not find root %llu\n", tree_id);
1729 key.objectid = dirid;
1730 key.type = BTRFS_INODE_REF_KEY;
1731 key.offset = (u64)-1;
1734 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1739 slot = path->slots[0];
1740 if (ret > 0 && slot > 0)
1742 btrfs_item_key_to_cpu(l, &key, slot);
1744 if (ret > 0 && (key.objectid != dirid ||
1745 key.type != BTRFS_INODE_REF_KEY)) {
1750 iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref);
1751 len = btrfs_inode_ref_name_len(l, iref);
1753 total_len += len + 1;
1758 read_extent_buffer(l, ptr,(unsigned long)(iref + 1), len);
1760 if (key.offset == BTRFS_FIRST_FREE_OBJECTID)
1763 btrfs_release_path(path);
1764 key.objectid = key.offset;
1765 key.offset = (u64)-1;
1766 dirid = key.objectid;
1770 memmove(name, ptr, total_len);
1771 name[total_len]='\0';
1774 btrfs_free_path(path);
1778 static noinline int btrfs_ioctl_ino_lookup(struct file *file,
1781 struct btrfs_ioctl_ino_lookup_args *args;
1782 struct inode *inode;
1785 if (!capable(CAP_SYS_ADMIN))
1788 args = memdup_user(argp, sizeof(*args));
1790 return PTR_ERR(args);
1792 inode = fdentry(file)->d_inode;
1794 if (args->treeid == 0)
1795 args->treeid = BTRFS_I(inode)->root->root_key.objectid;
1797 ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info,
1798 args->treeid, args->objectid,
1801 if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
1808 static noinline int btrfs_ioctl_snap_destroy(struct file *file,
1811 struct dentry *parent = fdentry(file);
1812 struct dentry *dentry;
1813 struct inode *dir = parent->d_inode;
1814 struct inode *inode;
1815 struct btrfs_root *root = BTRFS_I(dir)->root;
1816 struct btrfs_root *dest = NULL;
1817 struct btrfs_ioctl_vol_args *vol_args;
1818 struct btrfs_trans_handle *trans;
1823 vol_args = memdup_user(arg, sizeof(*vol_args));
1824 if (IS_ERR(vol_args))
1825 return PTR_ERR(vol_args);
1827 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
1828 namelen = strlen(vol_args->name);
1829 if (strchr(vol_args->name, '/') ||
1830 strncmp(vol_args->name, "..", namelen) == 0) {
1835 err = mnt_want_write(file->f_path.mnt);
1839 mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
1840 dentry = lookup_one_len(vol_args->name, parent, namelen);
1841 if (IS_ERR(dentry)) {
1842 err = PTR_ERR(dentry);
1843 goto out_unlock_dir;
1846 if (!dentry->d_inode) {
1851 inode = dentry->d_inode;
1852 dest = BTRFS_I(inode)->root;
1853 if (!capable(CAP_SYS_ADMIN)){
1855 * Regular user. Only allow this with a special mount
1856 * option, when the user has write+exec access to the
1857 * subvol root, and when rmdir(2) would have been
1860 * Note that this is _not_ check that the subvol is
1861 * empty or doesn't contain data that we wouldn't
1862 * otherwise be able to delete.
1864 * Users who want to delete empty subvols should try
1868 if (!btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED))
1872 * Do not allow deletion if the parent dir is the same
1873 * as the dir to be deleted. That means the ioctl
1874 * must be called on the dentry referencing the root
1875 * of the subvol, not a random directory contained
1882 err = inode_permission(inode, MAY_WRITE | MAY_EXEC);
1886 /* check if subvolume may be deleted by a non-root user */
1887 err = btrfs_may_delete(dir, dentry, 1);
1892 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) {
1897 mutex_lock(&inode->i_mutex);
1898 err = d_invalidate(dentry);
1902 down_write(&root->fs_info->subvol_sem);
1904 err = may_destroy_subvol(dest);
1908 trans = btrfs_start_transaction(root, 0);
1909 if (IS_ERR(trans)) {
1910 err = PTR_ERR(trans);
1913 trans->block_rsv = &root->fs_info->global_block_rsv;
1915 ret = btrfs_unlink_subvol(trans, root, dir,
1916 dest->root_key.objectid,
1917 dentry->d_name.name,
1918 dentry->d_name.len);
1921 btrfs_record_root_in_trans(trans, dest);
1923 memset(&dest->root_item.drop_progress, 0,
1924 sizeof(dest->root_item.drop_progress));
1925 dest->root_item.drop_level = 0;
1926 btrfs_set_root_refs(&dest->root_item, 0);
1928 if (!xchg(&dest->orphan_item_inserted, 1)) {
1929 ret = btrfs_insert_orphan_item(trans,
1930 root->fs_info->tree_root,
1931 dest->root_key.objectid);
1935 ret = btrfs_end_transaction(trans, root);
1937 inode->i_flags |= S_DEAD;
1939 up_write(&root->fs_info->subvol_sem);
1941 mutex_unlock(&inode->i_mutex);
1943 shrink_dcache_sb(root->fs_info->sb);
1944 btrfs_invalidate_inodes(dest);
1950 mutex_unlock(&dir->i_mutex);
1951 mnt_drop_write(file->f_path.mnt);
1957 static int btrfs_ioctl_defrag(struct file *file, void __user *argp)
1959 struct inode *inode = fdentry(file)->d_inode;
1960 struct btrfs_root *root = BTRFS_I(inode)->root;
1961 struct btrfs_ioctl_defrag_range_args *range;
1964 if (btrfs_root_readonly(root))
1967 ret = mnt_want_write(file->f_path.mnt);
1971 switch (inode->i_mode & S_IFMT) {
1973 if (!capable(CAP_SYS_ADMIN)) {
1977 ret = btrfs_defrag_root(root, 0);
1980 ret = btrfs_defrag_root(root->fs_info->extent_root, 0);
1983 if (!(file->f_mode & FMODE_WRITE)) {
1988 range = kzalloc(sizeof(*range), GFP_KERNEL);
1995 if (copy_from_user(range, argp,
2001 /* compression requires us to start the IO */
2002 if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
2003 range->flags |= BTRFS_DEFRAG_RANGE_START_IO;
2004 range->extent_thresh = (u32)-1;
2007 /* the rest are all set to zero by kzalloc */
2008 range->len = (u64)-1;
2010 ret = btrfs_defrag_file(fdentry(file)->d_inode, file,
2020 mnt_drop_write(file->f_path.mnt);
2024 static long btrfs_ioctl_add_dev(struct btrfs_root *root, void __user *arg)
2026 struct btrfs_ioctl_vol_args *vol_args;
2029 if (!capable(CAP_SYS_ADMIN))
2032 vol_args = memdup_user(arg, sizeof(*vol_args));
2033 if (IS_ERR(vol_args))
2034 return PTR_ERR(vol_args);
2036 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2037 ret = btrfs_init_new_device(root, vol_args->name);
2043 static long btrfs_ioctl_rm_dev(struct btrfs_root *root, void __user *arg)
2045 struct btrfs_ioctl_vol_args *vol_args;
2048 if (!capable(CAP_SYS_ADMIN))
2051 if (root->fs_info->sb->s_flags & MS_RDONLY)
2054 vol_args = memdup_user(arg, sizeof(*vol_args));
2055 if (IS_ERR(vol_args))
2056 return PTR_ERR(vol_args);
2058 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2059 ret = btrfs_rm_device(root, vol_args->name);
2065 static long btrfs_ioctl_fs_info(struct btrfs_root *root, void __user *arg)
2067 struct btrfs_ioctl_fs_info_args *fi_args;
2068 struct btrfs_device *device;
2069 struct btrfs_device *next;
2070 struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
2073 if (!capable(CAP_SYS_ADMIN))
2076 fi_args = kzalloc(sizeof(*fi_args), GFP_KERNEL);
2080 fi_args->num_devices = fs_devices->num_devices;
2081 memcpy(&fi_args->fsid, root->fs_info->fsid, sizeof(fi_args->fsid));
2083 mutex_lock(&fs_devices->device_list_mutex);
2084 list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
2085 if (device->devid > fi_args->max_id)
2086 fi_args->max_id = device->devid;
2088 mutex_unlock(&fs_devices->device_list_mutex);
2090 if (copy_to_user(arg, fi_args, sizeof(*fi_args)))
2097 static long btrfs_ioctl_dev_info(struct btrfs_root *root, void __user *arg)
2099 struct btrfs_ioctl_dev_info_args *di_args;
2100 struct btrfs_device *dev;
2101 struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
2103 char *s_uuid = NULL;
2104 char empty_uuid[BTRFS_UUID_SIZE] = {0};
2106 if (!capable(CAP_SYS_ADMIN))
2109 di_args = memdup_user(arg, sizeof(*di_args));
2110 if (IS_ERR(di_args))
2111 return PTR_ERR(di_args);
2113 if (memcmp(empty_uuid, di_args->uuid, BTRFS_UUID_SIZE) != 0)
2114 s_uuid = di_args->uuid;
2116 mutex_lock(&fs_devices->device_list_mutex);
2117 dev = btrfs_find_device(root, di_args->devid, s_uuid, NULL);
2118 mutex_unlock(&fs_devices->device_list_mutex);
2125 di_args->devid = dev->devid;
2126 di_args->bytes_used = dev->bytes_used;
2127 di_args->total_bytes = dev->total_bytes;
2128 memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid));
2129 strncpy(di_args->path, dev->name, sizeof(di_args->path));
2132 if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args)))
2139 static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
2140 u64 off, u64 olen, u64 destoff)
2142 struct inode *inode = fdentry(file)->d_inode;
2143 struct btrfs_root *root = BTRFS_I(inode)->root;
2144 struct file *src_file;
2146 struct btrfs_trans_handle *trans;
2147 struct btrfs_path *path;
2148 struct extent_buffer *leaf;
2150 struct btrfs_key key;
2155 u64 bs = root->fs_info->sb->s_blocksize;
2160 * - split compressed inline extents. annoying: we need to
2161 * decompress into destination's address_space (the file offset
2162 * may change, so source mapping won't do), then recompress (or
2163 * otherwise reinsert) a subrange.
2164 * - allow ranges within the same file to be cloned (provided
2165 * they don't overlap)?
2168 /* the destination must be opened for writing */
2169 if (!(file->f_mode & FMODE_WRITE) || (file->f_flags & O_APPEND))
2172 if (btrfs_root_readonly(root))
2175 ret = mnt_want_write(file->f_path.mnt);
2179 src_file = fget(srcfd);
2182 goto out_drop_write;
2185 src = src_file->f_dentry->d_inode;
2191 /* the src must be open for reading */
2192 if (!(src_file->f_mode & FMODE_READ))
2195 /* don't make the dst file partly checksummed */
2196 if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) !=
2197 (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM))
2201 if (S_ISDIR(src->i_mode) || S_ISDIR(inode->i_mode))
2205 if (src->i_sb != inode->i_sb || BTRFS_I(src)->root != root)
2209 buf = vmalloc(btrfs_level_size(root, 0));
2213 path = btrfs_alloc_path();
2221 mutex_lock_nested(&inode->i_mutex, I_MUTEX_PARENT);
2222 mutex_lock_nested(&src->i_mutex, I_MUTEX_CHILD);
2224 mutex_lock_nested(&src->i_mutex, I_MUTEX_PARENT);
2225 mutex_lock_nested(&inode->i_mutex, I_MUTEX_CHILD);
2228 /* determine range to clone */
2230 if (off + len > src->i_size || off + len < off)
2233 olen = len = src->i_size - off;
2234 /* if we extend to eof, continue to block boundary */
2235 if (off + len == src->i_size)
2236 len = ALIGN(src->i_size, bs) - off;
2238 /* verify the end result is block aligned */
2239 if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs) ||
2240 !IS_ALIGNED(destoff, bs))
2243 if (destoff > inode->i_size) {
2244 ret = btrfs_cont_expand(inode, inode->i_size, destoff);
2249 /* truncate page cache pages from target inode range */
2250 truncate_inode_pages_range(&inode->i_data, destoff,
2251 PAGE_CACHE_ALIGN(destoff + len) - 1);
2253 /* do any pending delalloc/csum calc on src, one way or
2254 another, and lock file content */
2256 struct btrfs_ordered_extent *ordered;
2257 lock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
2258 ordered = btrfs_lookup_first_ordered_extent(src, off+len);
2260 !test_range_bit(&BTRFS_I(src)->io_tree, off, off+len,
2261 EXTENT_DELALLOC, 0, NULL))
2263 unlock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
2265 btrfs_put_ordered_extent(ordered);
2266 btrfs_wait_ordered_range(src, off, len);
2270 key.objectid = btrfs_ino(src);
2271 key.type = BTRFS_EXTENT_DATA_KEY;
2276 * note the key will change type as we walk through the
2279 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
2283 nritems = btrfs_header_nritems(path->nodes[0]);
2284 if (path->slots[0] >= nritems) {
2285 ret = btrfs_next_leaf(root, path);
2290 nritems = btrfs_header_nritems(path->nodes[0]);
2292 leaf = path->nodes[0];
2293 slot = path->slots[0];
2295 btrfs_item_key_to_cpu(leaf, &key, slot);
2296 if (btrfs_key_type(&key) > BTRFS_EXTENT_DATA_KEY ||
2297 key.objectid != btrfs_ino(src))
2300 if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) {
2301 struct btrfs_file_extent_item *extent;
2304 struct btrfs_key new_key;
2305 u64 disko = 0, diskl = 0;
2306 u64 datao = 0, datal = 0;
2310 size = btrfs_item_size_nr(leaf, slot);
2311 read_extent_buffer(leaf, buf,
2312 btrfs_item_ptr_offset(leaf, slot),
2315 extent = btrfs_item_ptr(leaf, slot,
2316 struct btrfs_file_extent_item);
2317 comp = btrfs_file_extent_compression(leaf, extent);
2318 type = btrfs_file_extent_type(leaf, extent);
2319 if (type == BTRFS_FILE_EXTENT_REG ||
2320 type == BTRFS_FILE_EXTENT_PREALLOC) {
2321 disko = btrfs_file_extent_disk_bytenr(leaf,
2323 diskl = btrfs_file_extent_disk_num_bytes(leaf,
2325 datao = btrfs_file_extent_offset(leaf, extent);
2326 datal = btrfs_file_extent_num_bytes(leaf,
2328 } else if (type == BTRFS_FILE_EXTENT_INLINE) {
2329 /* take upper bound, may be compressed */
2330 datal = btrfs_file_extent_ram_bytes(leaf,
2333 btrfs_release_path(path);
2335 if (key.offset + datal <= off ||
2336 key.offset >= off+len)
2339 memcpy(&new_key, &key, sizeof(new_key));
2340 new_key.objectid = btrfs_ino(inode);
2341 if (off <= key.offset)
2342 new_key.offset = key.offset + destoff - off;
2344 new_key.offset = destoff;
2347 * 1 - adjusting old extent (we may have to split it)
2348 * 1 - add new extent
2351 trans = btrfs_start_transaction(root, 3);
2352 if (IS_ERR(trans)) {
2353 ret = PTR_ERR(trans);
2357 if (type == BTRFS_FILE_EXTENT_REG ||
2358 type == BTRFS_FILE_EXTENT_PREALLOC) {
2360 * a | --- range to clone ---| b
2361 * | ------------- extent ------------- |
2364 /* substract range b */
2365 if (key.offset + datal > off + len)
2366 datal = off + len - key.offset;
2368 /* substract range a */
2369 if (off > key.offset) {
2370 datao += off - key.offset;
2371 datal -= off - key.offset;
2374 ret = btrfs_drop_extents(trans, inode,
2376 new_key.offset + datal,
2380 ret = btrfs_insert_empty_item(trans, root, path,
2384 leaf = path->nodes[0];
2385 slot = path->slots[0];
2386 write_extent_buffer(leaf, buf,
2387 btrfs_item_ptr_offset(leaf, slot),
2390 extent = btrfs_item_ptr(leaf, slot,
2391 struct btrfs_file_extent_item);
2393 /* disko == 0 means it's a hole */
2397 btrfs_set_file_extent_offset(leaf, extent,
2399 btrfs_set_file_extent_num_bytes(leaf, extent,
2402 inode_add_bytes(inode, datal);
2403 ret = btrfs_inc_extent_ref(trans, root,
2405 root->root_key.objectid,
2407 new_key.offset - datao);
2410 } else if (type == BTRFS_FILE_EXTENT_INLINE) {
2413 if (off > key.offset) {
2414 skip = off - key.offset;
2415 new_key.offset += skip;
2418 if (key.offset + datal > off+len)
2419 trim = key.offset + datal - (off+len);
2421 if (comp && (skip || trim)) {
2423 btrfs_end_transaction(trans, root);
2426 size -= skip + trim;
2427 datal -= skip + trim;
2429 ret = btrfs_drop_extents(trans, inode,
2431 new_key.offset + datal,
2435 ret = btrfs_insert_empty_item(trans, root, path,
2441 btrfs_file_extent_calc_inline_size(0);
2442 memmove(buf+start, buf+start+skip,
2446 leaf = path->nodes[0];
2447 slot = path->slots[0];
2448 write_extent_buffer(leaf, buf,
2449 btrfs_item_ptr_offset(leaf, slot),
2451 inode_add_bytes(inode, datal);
2454 btrfs_mark_buffer_dirty(leaf);
2455 btrfs_release_path(path);
2457 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
2460 * we round up to the block size at eof when
2461 * determining which extents to clone above,
2462 * but shouldn't round up the file size
2464 endoff = new_key.offset + datal;
2465 if (endoff > destoff+olen)
2466 endoff = destoff+olen;
2467 if (endoff > inode->i_size)
2468 btrfs_i_size_write(inode, endoff);
2470 ret = btrfs_update_inode(trans, root, inode);
2472 btrfs_end_transaction(trans, root);
2475 btrfs_release_path(path);
2480 btrfs_release_path(path);
2481 unlock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
2483 mutex_unlock(&src->i_mutex);
2484 mutex_unlock(&inode->i_mutex);
2486 btrfs_free_path(path);
2490 mnt_drop_write(file->f_path.mnt);
2494 static long btrfs_ioctl_clone_range(struct file *file, void __user *argp)
2496 struct btrfs_ioctl_clone_range_args args;
2498 if (copy_from_user(&args, argp, sizeof(args)))
2500 return btrfs_ioctl_clone(file, args.src_fd, args.src_offset,
2501 args.src_length, args.dest_offset);
2505 * there are many ways the trans_start and trans_end ioctls can lead
2506 * to deadlocks. They should only be used by applications that
2507 * basically own the machine, and have a very in depth understanding
2508 * of all the possible deadlocks and enospc problems.
2510 static long btrfs_ioctl_trans_start(struct file *file)
2512 struct inode *inode = fdentry(file)->d_inode;
2513 struct btrfs_root *root = BTRFS_I(inode)->root;
2514 struct btrfs_trans_handle *trans;
2518 if (!capable(CAP_SYS_ADMIN))
2522 if (file->private_data)
2526 if (btrfs_root_readonly(root))
2529 ret = mnt_want_write(file->f_path.mnt);
2533 atomic_inc(&root->fs_info->open_ioctl_trans);
2536 trans = btrfs_start_ioctl_transaction(root);
2540 file->private_data = trans;
2544 atomic_dec(&root->fs_info->open_ioctl_trans);
2545 mnt_drop_write(file->f_path.mnt);
2550 static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp)
2552 struct inode *inode = fdentry(file)->d_inode;
2553 struct btrfs_root *root = BTRFS_I(inode)->root;
2554 struct btrfs_root *new_root;
2555 struct btrfs_dir_item *di;
2556 struct btrfs_trans_handle *trans;
2557 struct btrfs_path *path;
2558 struct btrfs_key location;
2559 struct btrfs_disk_key disk_key;
2560 struct btrfs_super_block *disk_super;
2565 if (!capable(CAP_SYS_ADMIN))
2568 if (copy_from_user(&objectid, argp, sizeof(objectid)))
2572 objectid = root->root_key.objectid;
2574 location.objectid = objectid;
2575 location.type = BTRFS_ROOT_ITEM_KEY;
2576 location.offset = (u64)-1;
2578 new_root = btrfs_read_fs_root_no_name(root->fs_info, &location);
2579 if (IS_ERR(new_root))
2580 return PTR_ERR(new_root);
2582 if (btrfs_root_refs(&new_root->root_item) == 0)
2585 path = btrfs_alloc_path();
2588 path->leave_spinning = 1;
2590 trans = btrfs_start_transaction(root, 1);
2591 if (IS_ERR(trans)) {
2592 btrfs_free_path(path);
2593 return PTR_ERR(trans);
2596 dir_id = btrfs_super_root_dir(&root->fs_info->super_copy);
2597 di = btrfs_lookup_dir_item(trans, root->fs_info->tree_root, path,
2598 dir_id, "default", 7, 1);
2599 if (IS_ERR_OR_NULL(di)) {
2600 btrfs_free_path(path);
2601 btrfs_end_transaction(trans, root);
2602 printk(KERN_ERR "Umm, you don't have the default dir item, "
2603 "this isn't going to work\n");
2607 btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key);
2608 btrfs_set_dir_item_key(path->nodes[0], di, &disk_key);
2609 btrfs_mark_buffer_dirty(path->nodes[0]);
2610 btrfs_free_path(path);
2612 disk_super = &root->fs_info->super_copy;
2613 features = btrfs_super_incompat_flags(disk_super);
2614 if (!(features & BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL)) {
2615 features |= BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL;
2616 btrfs_set_super_incompat_flags(disk_super, features);
2618 btrfs_end_transaction(trans, root);
2623 static void get_block_group_info(struct list_head *groups_list,
2624 struct btrfs_ioctl_space_info *space)
2626 struct btrfs_block_group_cache *block_group;
2628 space->total_bytes = 0;
2629 space->used_bytes = 0;
2631 list_for_each_entry(block_group, groups_list, list) {
2632 space->flags = block_group->flags;
2633 space->total_bytes += block_group->key.offset;
2634 space->used_bytes +=
2635 btrfs_block_group_used(&block_group->item);
2639 long btrfs_ioctl_space_info(struct btrfs_root *root, void __user *arg)
2641 struct btrfs_ioctl_space_args space_args;
2642 struct btrfs_ioctl_space_info space;
2643 struct btrfs_ioctl_space_info *dest;
2644 struct btrfs_ioctl_space_info *dest_orig;
2645 struct btrfs_ioctl_space_info __user *user_dest;
2646 struct btrfs_space_info *info;
2647 u64 types[] = {BTRFS_BLOCK_GROUP_DATA,
2648 BTRFS_BLOCK_GROUP_SYSTEM,
2649 BTRFS_BLOCK_GROUP_METADATA,
2650 BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA};
2657 if (copy_from_user(&space_args,
2658 (struct btrfs_ioctl_space_args __user *)arg,
2659 sizeof(space_args)))
2662 for (i = 0; i < num_types; i++) {
2663 struct btrfs_space_info *tmp;
2667 list_for_each_entry_rcu(tmp, &root->fs_info->space_info,
2669 if (tmp->flags == types[i]) {
2679 down_read(&info->groups_sem);
2680 for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
2681 if (!list_empty(&info->block_groups[c]))
2684 up_read(&info->groups_sem);
2687 /* space_slots == 0 means they are asking for a count */
2688 if (space_args.space_slots == 0) {
2689 space_args.total_spaces = slot_count;
2693 slot_count = min_t(u64, space_args.space_slots, slot_count);
2695 alloc_size = sizeof(*dest) * slot_count;
2697 /* we generally have at most 6 or so space infos, one for each raid
2698 * level. So, a whole page should be more than enough for everyone
2700 if (alloc_size > PAGE_CACHE_SIZE)
2703 space_args.total_spaces = 0;
2704 dest = kmalloc(alloc_size, GFP_NOFS);
2709 /* now we have a buffer to copy into */
2710 for (i = 0; i < num_types; i++) {
2711 struct btrfs_space_info *tmp;
2718 list_for_each_entry_rcu(tmp, &root->fs_info->space_info,
2720 if (tmp->flags == types[i]) {
2729 down_read(&info->groups_sem);
2730 for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
2731 if (!list_empty(&info->block_groups[c])) {
2732 get_block_group_info(&info->block_groups[c],
2734 memcpy(dest, &space, sizeof(space));
2736 space_args.total_spaces++;
2742 up_read(&info->groups_sem);
2745 user_dest = (struct btrfs_ioctl_space_info *)
2746 (arg + sizeof(struct btrfs_ioctl_space_args));
2748 if (copy_to_user(user_dest, dest_orig, alloc_size))
2753 if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args)))
2760 * there are many ways the trans_start and trans_end ioctls can lead
2761 * to deadlocks. They should only be used by applications that
2762 * basically own the machine, and have a very in depth understanding
2763 * of all the possible deadlocks and enospc problems.
2765 long btrfs_ioctl_trans_end(struct file *file)
2767 struct inode *inode = fdentry(file)->d_inode;
2768 struct btrfs_root *root = BTRFS_I(inode)->root;
2769 struct btrfs_trans_handle *trans;
2771 trans = file->private_data;
2774 file->private_data = NULL;
2776 btrfs_end_transaction(trans, root);
2778 atomic_dec(&root->fs_info->open_ioctl_trans);
2780 mnt_drop_write(file->f_path.mnt);
2784 static noinline long btrfs_ioctl_start_sync(struct file *file, void __user *argp)
2786 struct btrfs_root *root = BTRFS_I(file->f_dentry->d_inode)->root;
2787 struct btrfs_trans_handle *trans;
2791 trans = btrfs_start_transaction(root, 0);
2793 return PTR_ERR(trans);
2794 transid = trans->transid;
2795 ret = btrfs_commit_transaction_async(trans, root, 0);
2797 btrfs_end_transaction(trans, root);
2802 if (copy_to_user(argp, &transid, sizeof(transid)))
2807 static noinline long btrfs_ioctl_wait_sync(struct file *file, void __user *argp)
2809 struct btrfs_root *root = BTRFS_I(file->f_dentry->d_inode)->root;
2813 if (copy_from_user(&transid, argp, sizeof(transid)))
2816 transid = 0; /* current trans */
2818 return btrfs_wait_for_commit(root, transid);
2821 static long btrfs_ioctl_scrub(struct btrfs_root *root, void __user *arg)
2824 struct btrfs_ioctl_scrub_args *sa;
2826 if (!capable(CAP_SYS_ADMIN))
2829 sa = memdup_user(arg, sizeof(*sa));
2833 ret = btrfs_scrub_dev(root, sa->devid, sa->start, sa->end,
2834 &sa->progress, sa->flags & BTRFS_SCRUB_READONLY);
2836 if (copy_to_user(arg, sa, sizeof(*sa)))
2843 static long btrfs_ioctl_scrub_cancel(struct btrfs_root *root, void __user *arg)
2845 if (!capable(CAP_SYS_ADMIN))
2848 return btrfs_scrub_cancel(root);
2851 static long btrfs_ioctl_scrub_progress(struct btrfs_root *root,
2854 struct btrfs_ioctl_scrub_args *sa;
2857 if (!capable(CAP_SYS_ADMIN))
2860 sa = memdup_user(arg, sizeof(*sa));
2864 ret = btrfs_scrub_progress(root, sa->devid, &sa->progress);
2866 if (copy_to_user(arg, sa, sizeof(*sa)))
2873 long btrfs_ioctl(struct file *file, unsigned int
2874 cmd, unsigned long arg)
2876 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
2877 void __user *argp = (void __user *)arg;
2880 case FS_IOC_GETFLAGS:
2881 return btrfs_ioctl_getflags(file, argp);
2882 case FS_IOC_SETFLAGS:
2883 return btrfs_ioctl_setflags(file, argp);
2884 case FS_IOC_GETVERSION:
2885 return btrfs_ioctl_getversion(file, argp);
2887 return btrfs_ioctl_fitrim(file, argp);
2888 case BTRFS_IOC_SNAP_CREATE:
2889 return btrfs_ioctl_snap_create(file, argp, 0);
2890 case BTRFS_IOC_SNAP_CREATE_V2:
2891 return btrfs_ioctl_snap_create_v2(file, argp, 0);
2892 case BTRFS_IOC_SUBVOL_CREATE:
2893 return btrfs_ioctl_snap_create(file, argp, 1);
2894 case BTRFS_IOC_SNAP_DESTROY:
2895 return btrfs_ioctl_snap_destroy(file, argp);
2896 case BTRFS_IOC_SUBVOL_GETFLAGS:
2897 return btrfs_ioctl_subvol_getflags(file, argp);
2898 case BTRFS_IOC_SUBVOL_SETFLAGS:
2899 return btrfs_ioctl_subvol_setflags(file, argp);
2900 case BTRFS_IOC_DEFAULT_SUBVOL:
2901 return btrfs_ioctl_default_subvol(file, argp);
2902 case BTRFS_IOC_DEFRAG:
2903 return btrfs_ioctl_defrag(file, NULL);
2904 case BTRFS_IOC_DEFRAG_RANGE:
2905 return btrfs_ioctl_defrag(file, argp);
2906 case BTRFS_IOC_RESIZE:
2907 return btrfs_ioctl_resize(root, argp);
2908 case BTRFS_IOC_ADD_DEV:
2909 return btrfs_ioctl_add_dev(root, argp);
2910 case BTRFS_IOC_RM_DEV:
2911 return btrfs_ioctl_rm_dev(root, argp);
2912 case BTRFS_IOC_FS_INFO:
2913 return btrfs_ioctl_fs_info(root, argp);
2914 case BTRFS_IOC_DEV_INFO:
2915 return btrfs_ioctl_dev_info(root, argp);
2916 case BTRFS_IOC_BALANCE:
2917 return btrfs_balance(root->fs_info->dev_root);
2918 case BTRFS_IOC_CLONE:
2919 return btrfs_ioctl_clone(file, arg, 0, 0, 0);
2920 case BTRFS_IOC_CLONE_RANGE:
2921 return btrfs_ioctl_clone_range(file, argp);
2922 case BTRFS_IOC_TRANS_START:
2923 return btrfs_ioctl_trans_start(file);
2924 case BTRFS_IOC_TRANS_END:
2925 return btrfs_ioctl_trans_end(file);
2926 case BTRFS_IOC_TREE_SEARCH:
2927 return btrfs_ioctl_tree_search(file, argp);
2928 case BTRFS_IOC_INO_LOOKUP:
2929 return btrfs_ioctl_ino_lookup(file, argp);
2930 case BTRFS_IOC_SPACE_INFO:
2931 return btrfs_ioctl_space_info(root, argp);
2932 case BTRFS_IOC_SYNC:
2933 btrfs_sync_fs(file->f_dentry->d_sb, 1);
2935 case BTRFS_IOC_START_SYNC:
2936 return btrfs_ioctl_start_sync(file, argp);
2937 case BTRFS_IOC_WAIT_SYNC:
2938 return btrfs_ioctl_wait_sync(file, argp);
2939 case BTRFS_IOC_SCRUB:
2940 return btrfs_ioctl_scrub(root, argp);
2941 case BTRFS_IOC_SCRUB_CANCEL:
2942 return btrfs_ioctl_scrub_cancel(root, argp);
2943 case BTRFS_IOC_SCRUB_PROGRESS:
2944 return btrfs_ioctl_scrub_progress(root, argp);