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 * Currently only the compression flags and the cow flags are inherited.
122 void btrfs_inherit_iflags(struct inode *inode, struct inode *dir)
129 flags = BTRFS_I(dir)->flags;
131 if (flags & BTRFS_INODE_NOCOMPRESS) {
132 BTRFS_I(inode)->flags &= ~BTRFS_INODE_COMPRESS;
133 BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS;
134 } else if (flags & BTRFS_INODE_COMPRESS) {
135 BTRFS_I(inode)->flags &= ~BTRFS_INODE_NOCOMPRESS;
136 BTRFS_I(inode)->flags |= BTRFS_INODE_COMPRESS;
139 if (flags & BTRFS_INODE_NODATACOW)
140 BTRFS_I(inode)->flags |= BTRFS_INODE_NODATACOW;
142 btrfs_update_iflags(inode);
145 static int btrfs_ioctl_getflags(struct file *file, void __user *arg)
147 struct btrfs_inode *ip = BTRFS_I(file->f_path.dentry->d_inode);
148 unsigned int flags = btrfs_flags_to_ioctl(ip->flags);
150 if (copy_to_user(arg, &flags, sizeof(flags)))
155 static int check_flags(unsigned int flags)
157 if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \
158 FS_NOATIME_FL | FS_NODUMP_FL | \
159 FS_SYNC_FL | FS_DIRSYNC_FL | \
160 FS_NOCOMP_FL | FS_COMPR_FL |
164 if ((flags & FS_NOCOMP_FL) && (flags & FS_COMPR_FL))
170 static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
172 struct inode *inode = file->f_path.dentry->d_inode;
173 struct btrfs_inode *ip = BTRFS_I(inode);
174 struct btrfs_root *root = ip->root;
175 struct btrfs_trans_handle *trans;
176 unsigned int flags, oldflags;
179 if (btrfs_root_readonly(root))
182 if (copy_from_user(&flags, arg, sizeof(flags)))
185 ret = check_flags(flags);
189 if (!inode_owner_or_capable(inode))
192 mutex_lock(&inode->i_mutex);
194 flags = btrfs_mask_flags(inode->i_mode, flags);
195 oldflags = btrfs_flags_to_ioctl(ip->flags);
196 if ((flags ^ oldflags) & (FS_APPEND_FL | FS_IMMUTABLE_FL)) {
197 if (!capable(CAP_LINUX_IMMUTABLE)) {
203 ret = mnt_want_write(file->f_path.mnt);
207 if (flags & FS_SYNC_FL)
208 ip->flags |= BTRFS_INODE_SYNC;
210 ip->flags &= ~BTRFS_INODE_SYNC;
211 if (flags & FS_IMMUTABLE_FL)
212 ip->flags |= BTRFS_INODE_IMMUTABLE;
214 ip->flags &= ~BTRFS_INODE_IMMUTABLE;
215 if (flags & FS_APPEND_FL)
216 ip->flags |= BTRFS_INODE_APPEND;
218 ip->flags &= ~BTRFS_INODE_APPEND;
219 if (flags & FS_NODUMP_FL)
220 ip->flags |= BTRFS_INODE_NODUMP;
222 ip->flags &= ~BTRFS_INODE_NODUMP;
223 if (flags & FS_NOATIME_FL)
224 ip->flags |= BTRFS_INODE_NOATIME;
226 ip->flags &= ~BTRFS_INODE_NOATIME;
227 if (flags & FS_DIRSYNC_FL)
228 ip->flags |= BTRFS_INODE_DIRSYNC;
230 ip->flags &= ~BTRFS_INODE_DIRSYNC;
231 if (flags & FS_NOCOW_FL)
232 ip->flags |= BTRFS_INODE_NODATACOW;
234 ip->flags &= ~BTRFS_INODE_NODATACOW;
237 * The COMPRESS flag can only be changed by users, while the NOCOMPRESS
238 * flag may be changed automatically if compression code won't make
241 if (flags & FS_NOCOMP_FL) {
242 ip->flags &= ~BTRFS_INODE_COMPRESS;
243 ip->flags |= BTRFS_INODE_NOCOMPRESS;
244 } else if (flags & FS_COMPR_FL) {
245 ip->flags |= BTRFS_INODE_COMPRESS;
246 ip->flags &= ~BTRFS_INODE_NOCOMPRESS;
248 ip->flags &= ~(BTRFS_INODE_COMPRESS | BTRFS_INODE_NOCOMPRESS);
251 trans = btrfs_join_transaction(root);
252 BUG_ON(IS_ERR(trans));
254 ret = btrfs_update_inode(trans, root, inode);
257 btrfs_update_iflags(inode);
258 inode->i_ctime = CURRENT_TIME;
259 btrfs_end_transaction(trans, root);
261 mnt_drop_write(file->f_path.mnt);
265 mutex_unlock(&inode->i_mutex);
269 static int btrfs_ioctl_getversion(struct file *file, int __user *arg)
271 struct inode *inode = file->f_path.dentry->d_inode;
273 return put_user(inode->i_generation, arg);
276 static noinline int btrfs_ioctl_fitrim(struct file *file, void __user *arg)
278 struct btrfs_root *root = fdentry(file)->d_sb->s_fs_info;
279 struct btrfs_fs_info *fs_info = root->fs_info;
280 struct btrfs_device *device;
281 struct request_queue *q;
282 struct fstrim_range range;
283 u64 minlen = ULLONG_MAX;
287 if (!capable(CAP_SYS_ADMIN))
291 list_for_each_entry_rcu(device, &fs_info->fs_devices->devices,
295 q = bdev_get_queue(device->bdev);
296 if (blk_queue_discard(q)) {
298 minlen = min((u64)q->limits.discard_granularity,
306 if (copy_from_user(&range, arg, sizeof(range)))
309 range.minlen = max(range.minlen, minlen);
310 ret = btrfs_trim_fs(root, &range);
314 if (copy_to_user(arg, &range, sizeof(range)))
320 static noinline int create_subvol(struct btrfs_root *root,
321 struct dentry *dentry,
322 char *name, int namelen,
325 struct btrfs_trans_handle *trans;
326 struct btrfs_key key;
327 struct btrfs_root_item root_item;
328 struct btrfs_inode_item *inode_item;
329 struct extent_buffer *leaf;
330 struct btrfs_root *new_root;
331 struct dentry *parent = dentry->d_parent;
336 u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
339 ret = btrfs_find_free_objectid(root->fs_info->tree_root, &objectid);
343 dir = parent->d_inode;
351 trans = btrfs_start_transaction(root, 6);
353 return PTR_ERR(trans);
355 leaf = btrfs_alloc_free_block(trans, root, root->leafsize,
356 0, objectid, NULL, 0, 0, 0);
362 memset_extent_buffer(leaf, 0, 0, sizeof(struct btrfs_header));
363 btrfs_set_header_bytenr(leaf, leaf->start);
364 btrfs_set_header_generation(leaf, trans->transid);
365 btrfs_set_header_backref_rev(leaf, BTRFS_MIXED_BACKREF_REV);
366 btrfs_set_header_owner(leaf, objectid);
368 write_extent_buffer(leaf, root->fs_info->fsid,
369 (unsigned long)btrfs_header_fsid(leaf),
371 write_extent_buffer(leaf, root->fs_info->chunk_tree_uuid,
372 (unsigned long)btrfs_header_chunk_tree_uuid(leaf),
374 btrfs_mark_buffer_dirty(leaf);
376 inode_item = &root_item.inode;
377 memset(inode_item, 0, sizeof(*inode_item));
378 inode_item->generation = cpu_to_le64(1);
379 inode_item->size = cpu_to_le64(3);
380 inode_item->nlink = cpu_to_le32(1);
381 inode_item->nbytes = cpu_to_le64(root->leafsize);
382 inode_item->mode = cpu_to_le32(S_IFDIR | 0755);
385 root_item.byte_limit = 0;
386 inode_item->flags = cpu_to_le64(BTRFS_INODE_ROOT_ITEM_INIT);
388 btrfs_set_root_bytenr(&root_item, leaf->start);
389 btrfs_set_root_generation(&root_item, trans->transid);
390 btrfs_set_root_level(&root_item, 0);
391 btrfs_set_root_refs(&root_item, 1);
392 btrfs_set_root_used(&root_item, leaf->len);
393 btrfs_set_root_last_snapshot(&root_item, 0);
395 memset(&root_item.drop_progress, 0, sizeof(root_item.drop_progress));
396 root_item.drop_level = 0;
398 btrfs_tree_unlock(leaf);
399 free_extent_buffer(leaf);
402 btrfs_set_root_dirid(&root_item, new_dirid);
404 key.objectid = objectid;
406 btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
407 ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key,
412 key.offset = (u64)-1;
413 new_root = btrfs_read_fs_root_no_name(root->fs_info, &key);
414 BUG_ON(IS_ERR(new_root));
416 btrfs_record_root_in_trans(trans, new_root);
418 ret = btrfs_create_subvol_root(trans, new_root, new_dirid);
420 * insert the directory item
422 ret = btrfs_set_inode_index(dir, &index);
425 ret = btrfs_insert_dir_item(trans, root,
426 name, namelen, dir, &key,
427 BTRFS_FT_DIR, index);
431 btrfs_i_size_write(dir, dir->i_size + namelen * 2);
432 ret = btrfs_update_inode(trans, root, dir);
435 ret = btrfs_add_root_ref(trans, root->fs_info->tree_root,
436 objectid, root->root_key.objectid,
437 btrfs_ino(dir), index, name, namelen);
441 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 btrfs_pending_snapshot *pending_snapshot;
460 struct btrfs_trans_handle *trans;
466 pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS);
467 if (!pending_snapshot)
470 btrfs_init_block_rsv(&pending_snapshot->block_rsv);
471 pending_snapshot->dentry = dentry;
472 pending_snapshot->root = root;
473 pending_snapshot->readonly = readonly;
475 trans = btrfs_start_transaction(root->fs_info->extent_root, 5);
477 ret = PTR_ERR(trans);
481 ret = btrfs_snap_reserve_metadata(trans, pending_snapshot);
484 spin_lock(&root->fs_info->trans_lock);
485 list_add(&pending_snapshot->list,
486 &trans->transaction->pending_snapshots);
487 spin_unlock(&root->fs_info->trans_lock);
489 *async_transid = trans->transid;
490 ret = btrfs_commit_transaction_async(trans,
491 root->fs_info->extent_root, 1);
493 ret = btrfs_commit_transaction(trans,
494 root->fs_info->extent_root);
498 ret = pending_snapshot->error;
502 ret = btrfs_orphan_cleanup(pending_snapshot->snap);
506 inode = btrfs_lookup_dentry(dentry->d_parent->d_inode, dentry);
508 ret = PTR_ERR(inode);
512 d_instantiate(dentry, inode);
515 kfree(pending_snapshot);
519 /* copy of check_sticky in fs/namei.c()
520 * It's inline, so penalty for filesystems that don't use sticky bit is
523 static inline int btrfs_check_sticky(struct inode *dir, struct inode *inode)
525 uid_t fsuid = current_fsuid();
527 if (!(dir->i_mode & S_ISVTX))
529 if (inode->i_uid == fsuid)
531 if (dir->i_uid == fsuid)
533 return !capable(CAP_FOWNER);
536 /* copy of may_delete in fs/namei.c()
537 * Check whether we can remove a link victim from directory dir, check
538 * whether the type of victim is right.
539 * 1. We can't do it if dir is read-only (done in permission())
540 * 2. We should have write and exec permissions on dir
541 * 3. We can't remove anything from append-only dir
542 * 4. We can't do anything with immutable dir (done in permission())
543 * 5. If the sticky bit on dir is set we should either
544 * a. be owner of dir, or
545 * b. be owner of victim, or
546 * c. have CAP_FOWNER capability
547 * 6. If the victim is append-only or immutable we can't do antyhing with
548 * links pointing to it.
549 * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR.
550 * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR.
551 * 9. We can't remove a root or mountpoint.
552 * 10. We don't allow removal of NFS sillyrenamed files; it's handled by
553 * nfs_async_unlink().
556 static int btrfs_may_delete(struct inode *dir,struct dentry *victim,int isdir)
560 if (!victim->d_inode)
563 BUG_ON(victim->d_parent->d_inode != dir);
564 audit_inode_child(victim, dir);
566 error = inode_permission(dir, MAY_WRITE | MAY_EXEC);
571 if (btrfs_check_sticky(dir, victim->d_inode)||
572 IS_APPEND(victim->d_inode)||
573 IS_IMMUTABLE(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
576 if (!S_ISDIR(victim->d_inode->i_mode))
580 } else if (S_ISDIR(victim->d_inode->i_mode))
584 if (victim->d_flags & DCACHE_NFSFS_RENAMED)
589 /* copy of may_create in fs/namei.c() */
590 static inline int btrfs_may_create(struct inode *dir, struct dentry *child)
596 return inode_permission(dir, MAY_WRITE | MAY_EXEC);
600 * Create a new subvolume below @parent. This is largely modeled after
601 * sys_mkdirat and vfs_mkdir, but we only do a single component lookup
602 * inside this filesystem so it's quite a bit simpler.
604 static noinline int btrfs_mksubvol(struct path *parent,
605 char *name, int namelen,
606 struct btrfs_root *snap_src,
607 u64 *async_transid, bool readonly)
609 struct inode *dir = parent->dentry->d_inode;
610 struct dentry *dentry;
613 mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
615 dentry = lookup_one_len(name, parent->dentry, namelen);
616 error = PTR_ERR(dentry);
624 error = mnt_want_write(parent->mnt);
628 error = btrfs_may_create(dir, dentry);
632 down_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
634 if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0)
638 error = create_snapshot(snap_src, dentry,
639 name, namelen, async_transid, readonly);
641 error = create_subvol(BTRFS_I(dir)->root, dentry,
642 name, namelen, async_transid);
645 fsnotify_mkdir(dir, dentry);
647 up_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
649 mnt_drop_write(parent->mnt);
653 mutex_unlock(&dir->i_mutex);
658 * When we're defragging a range, we don't want to kick it off again
659 * if it is really just waiting for delalloc to send it down.
660 * If we find a nice big extent or delalloc range for the bytes in the
661 * file you want to defrag, we return 0 to let you know to skip this
664 static int check_defrag_in_cache(struct inode *inode, u64 offset, int thresh)
666 struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
667 struct extent_map *em = NULL;
668 struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
671 read_lock(&em_tree->lock);
672 em = lookup_extent_mapping(em_tree, offset, PAGE_CACHE_SIZE);
673 read_unlock(&em_tree->lock);
676 end = extent_map_end(em);
678 if (end - offset > thresh)
681 /* if we already have a nice delalloc here, just stop */
683 end = count_range_bits(io_tree, &offset, offset + thresh,
684 thresh, EXTENT_DELALLOC, 1);
691 * helper function to walk through a file and find extents
692 * newer than a specific transid, and smaller than thresh.
694 * This is used by the defragging code to find new and small
697 static int find_new_extents(struct btrfs_root *root,
698 struct inode *inode, u64 newer_than,
699 u64 *off, int thresh)
701 struct btrfs_path *path;
702 struct btrfs_key min_key;
703 struct btrfs_key max_key;
704 struct extent_buffer *leaf;
705 struct btrfs_file_extent_item *extent;
708 u64 ino = btrfs_ino(inode);
710 path = btrfs_alloc_path();
714 min_key.objectid = ino;
715 min_key.type = BTRFS_EXTENT_DATA_KEY;
716 min_key.offset = *off;
718 max_key.objectid = 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 != 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;
851 gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
855 file_end = (isize - 1) >> PAGE_CACHE_SHIFT;
857 ret = btrfs_delalloc_reserve_space(inode,
858 num_pages << PAGE_CACHE_SHIFT);
865 /* step one, lock all the pages */
866 for (i = 0; i < num_pages; i++) {
868 page = find_or_create_page(inode->i_mapping,
869 start_index + i, mask);
873 if (!PageUptodate(page)) {
874 btrfs_readpage(NULL, page);
876 if (!PageUptodate(page)) {
878 page_cache_release(page);
883 isize = i_size_read(inode);
884 file_end = (isize - 1) >> PAGE_CACHE_SHIFT;
885 if (!isize || page->index > file_end ||
886 page->mapping != inode->i_mapping) {
887 /* whoops, we blew past eof, skip this page */
889 page_cache_release(page);
898 if (!(inode->i_sb->s_flags & MS_ACTIVE))
902 * so now we have a nice long stream of locked
903 * and up to date pages, lets wait on them
905 for (i = 0; i < i_done; i++)
906 wait_on_page_writeback(pages[i]);
908 page_start = page_offset(pages[0]);
909 page_end = page_offset(pages[i_done - 1]) + PAGE_CACHE_SIZE;
911 lock_extent_bits(&BTRFS_I(inode)->io_tree,
912 page_start, page_end - 1, 0, &cached_state,
914 ordered = btrfs_lookup_first_ordered_extent(inode, page_end - 1);
916 ordered->file_offset + ordered->len > page_start &&
917 ordered->file_offset < page_end) {
918 btrfs_put_ordered_extent(ordered);
919 unlock_extent_cached(&BTRFS_I(inode)->io_tree,
920 page_start, page_end - 1,
921 &cached_state, GFP_NOFS);
922 for (i = 0; i < i_done; i++) {
923 unlock_page(pages[i]);
924 page_cache_release(pages[i]);
926 btrfs_wait_ordered_range(inode, page_start,
927 page_end - page_start);
931 btrfs_put_ordered_extent(ordered);
933 clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start,
934 page_end - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
935 EXTENT_DO_ACCOUNTING, 0, 0, &cached_state,
938 if (i_done != num_pages) {
939 spin_lock(&BTRFS_I(inode)->lock);
940 BTRFS_I(inode)->outstanding_extents++;
941 spin_unlock(&BTRFS_I(inode)->lock);
942 btrfs_delalloc_release_space(inode,
943 (num_pages - i_done) << PAGE_CACHE_SHIFT);
947 btrfs_set_extent_delalloc(inode, page_start, page_end - 1,
950 unlock_extent_cached(&BTRFS_I(inode)->io_tree,
951 page_start, page_end - 1, &cached_state,
954 for (i = 0; i < i_done; i++) {
955 clear_page_dirty_for_io(pages[i]);
956 ClearPageChecked(pages[i]);
957 set_page_extent_mapped(pages[i]);
958 set_page_dirty(pages[i]);
959 unlock_page(pages[i]);
960 page_cache_release(pages[i]);
964 for (i = 0; i < i_done; i++) {
965 unlock_page(pages[i]);
966 page_cache_release(pages[i]);
968 btrfs_delalloc_release_space(inode, num_pages << PAGE_CACHE_SHIFT);
973 int btrfs_defrag_file(struct inode *inode, struct file *file,
974 struct btrfs_ioctl_defrag_range_args *range,
975 u64 newer_than, unsigned long max_to_defrag)
977 struct btrfs_root *root = BTRFS_I(inode)->root;
978 struct btrfs_super_block *disk_super;
979 struct file_ra_state *ra = NULL;
980 unsigned long last_index;
985 u64 newer_off = range->start;
989 int defrag_count = 0;
990 int compress_type = BTRFS_COMPRESS_ZLIB;
991 int extent_thresh = range->extent_thresh;
992 int newer_cluster = (256 * 1024) >> PAGE_CACHE_SHIFT;
993 u64 new_align = ~((u64)128 * 1024 - 1);
994 struct page **pages = NULL;
996 if (extent_thresh == 0)
997 extent_thresh = 256 * 1024;
999 if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS) {
1000 if (range->compress_type > BTRFS_COMPRESS_TYPES)
1002 if (range->compress_type)
1003 compress_type = range->compress_type;
1006 if (inode->i_size == 0)
1010 * if we were not given a file, allocate a readahead
1014 ra = kzalloc(sizeof(*ra), GFP_NOFS);
1017 file_ra_state_init(ra, inode->i_mapping);
1022 pages = kmalloc(sizeof(struct page *) * newer_cluster,
1029 /* find the last page to defrag */
1030 if (range->start + range->len > range->start) {
1031 last_index = min_t(u64, inode->i_size - 1,
1032 range->start + range->len - 1) >> PAGE_CACHE_SHIFT;
1034 last_index = (inode->i_size - 1) >> PAGE_CACHE_SHIFT;
1038 ret = find_new_extents(root, inode, newer_than,
1039 &newer_off, 64 * 1024);
1041 range->start = newer_off;
1043 * we always align our defrag to help keep
1044 * the extents in the file evenly spaced
1046 i = (newer_off & new_align) >> PAGE_CACHE_SHIFT;
1047 newer_left = newer_cluster;
1051 i = range->start >> PAGE_CACHE_SHIFT;
1054 max_to_defrag = last_index - 1;
1056 while (i <= last_index && defrag_count < max_to_defrag) {
1058 * make sure we stop running if someone unmounts
1061 if (!(inode->i_sb->s_flags & MS_ACTIVE))
1065 !should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT,
1072 * the should_defrag function tells us how much to skip
1073 * bump our counter by the suggested amount
1075 next = (skip + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
1076 i = max(i + 1, next);
1079 if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)
1080 BTRFS_I(inode)->force_compress = compress_type;
1082 btrfs_force_ra(inode->i_mapping, ra, file, i, newer_cluster);
1084 ret = cluster_pages_for_defrag(inode, pages, i, newer_cluster);
1088 defrag_count += ret;
1089 balance_dirty_pages_ratelimited_nr(inode->i_mapping, ret);
1093 if (newer_off == (u64)-1)
1096 newer_off = max(newer_off + 1,
1097 (u64)i << PAGE_CACHE_SHIFT);
1099 ret = find_new_extents(root, inode,
1100 newer_than, &newer_off,
1103 range->start = newer_off;
1104 i = (newer_off & new_align) >> PAGE_CACHE_SHIFT;
1105 newer_left = newer_cluster;
1114 if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO))
1115 filemap_flush(inode->i_mapping);
1117 if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
1118 /* the filemap_flush will queue IO into the worker threads, but
1119 * we have to make sure the IO is actually started and that
1120 * ordered extents get created before we return
1122 atomic_inc(&root->fs_info->async_submit_draining);
1123 while (atomic_read(&root->fs_info->nr_async_submits) ||
1124 atomic_read(&root->fs_info->async_delalloc_pages)) {
1125 wait_event(root->fs_info->async_submit_wait,
1126 (atomic_read(&root->fs_info->nr_async_submits) == 0 &&
1127 atomic_read(&root->fs_info->async_delalloc_pages) == 0));
1129 atomic_dec(&root->fs_info->async_submit_draining);
1131 mutex_lock(&inode->i_mutex);
1132 BTRFS_I(inode)->force_compress = BTRFS_COMPRESS_NONE;
1133 mutex_unlock(&inode->i_mutex);
1136 disk_super = &root->fs_info->super_copy;
1137 features = btrfs_super_incompat_flags(disk_super);
1138 if (range->compress_type == BTRFS_COMPRESS_LZO) {
1139 features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
1140 btrfs_set_super_incompat_flags(disk_super, features);
1152 static noinline int btrfs_ioctl_resize(struct btrfs_root *root,
1158 struct btrfs_ioctl_vol_args *vol_args;
1159 struct btrfs_trans_handle *trans;
1160 struct btrfs_device *device = NULL;
1162 char *devstr = NULL;
1166 if (root->fs_info->sb->s_flags & MS_RDONLY)
1169 if (!capable(CAP_SYS_ADMIN))
1172 vol_args = memdup_user(arg, sizeof(*vol_args));
1173 if (IS_ERR(vol_args))
1174 return PTR_ERR(vol_args);
1176 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
1178 mutex_lock(&root->fs_info->volume_mutex);
1179 sizestr = vol_args->name;
1180 devstr = strchr(sizestr, ':');
1183 sizestr = devstr + 1;
1185 devstr = vol_args->name;
1186 devid = simple_strtoull(devstr, &end, 10);
1187 printk(KERN_INFO "resizing devid %llu\n",
1188 (unsigned long long)devid);
1190 device = btrfs_find_device(root, devid, NULL, NULL);
1192 printk(KERN_INFO "resizer unable to find device %llu\n",
1193 (unsigned long long)devid);
1197 if (!strcmp(sizestr, "max"))
1198 new_size = device->bdev->bd_inode->i_size;
1200 if (sizestr[0] == '-') {
1203 } else if (sizestr[0] == '+') {
1207 new_size = memparse(sizestr, NULL);
1208 if (new_size == 0) {
1214 old_size = device->total_bytes;
1217 if (new_size > old_size) {
1221 new_size = old_size - new_size;
1222 } else if (mod > 0) {
1223 new_size = old_size + new_size;
1226 if (new_size < 256 * 1024 * 1024) {
1230 if (new_size > device->bdev->bd_inode->i_size) {
1235 do_div(new_size, root->sectorsize);
1236 new_size *= root->sectorsize;
1238 printk(KERN_INFO "new size for %s is %llu\n",
1239 device->name, (unsigned long long)new_size);
1241 if (new_size > old_size) {
1242 trans = btrfs_start_transaction(root, 0);
1243 if (IS_ERR(trans)) {
1244 ret = PTR_ERR(trans);
1247 ret = btrfs_grow_device(trans, device, new_size);
1248 btrfs_commit_transaction(trans, root);
1250 ret = btrfs_shrink_device(device, new_size);
1254 mutex_unlock(&root->fs_info->volume_mutex);
1259 static noinline int btrfs_ioctl_snap_create_transid(struct file *file,
1266 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
1267 struct file *src_file;
1271 if (root->fs_info->sb->s_flags & MS_RDONLY)
1274 namelen = strlen(name);
1275 if (strchr(name, '/')) {
1281 ret = btrfs_mksubvol(&file->f_path, name, namelen,
1282 NULL, transid, readonly);
1284 struct inode *src_inode;
1285 src_file = fget(fd);
1291 src_inode = src_file->f_path.dentry->d_inode;
1292 if (src_inode->i_sb != file->f_path.dentry->d_inode->i_sb) {
1293 printk(KERN_INFO "btrfs: Snapshot src from "
1299 ret = btrfs_mksubvol(&file->f_path, name, namelen,
1300 BTRFS_I(src_inode)->root,
1308 static noinline int btrfs_ioctl_snap_create(struct file *file,
1309 void __user *arg, int subvol)
1311 struct btrfs_ioctl_vol_args *vol_args;
1314 vol_args = memdup_user(arg, sizeof(*vol_args));
1315 if (IS_ERR(vol_args))
1316 return PTR_ERR(vol_args);
1317 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
1319 ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
1320 vol_args->fd, subvol,
1327 static noinline int btrfs_ioctl_snap_create_v2(struct file *file,
1328 void __user *arg, int subvol)
1330 struct btrfs_ioctl_vol_args_v2 *vol_args;
1334 bool readonly = false;
1336 vol_args = memdup_user(arg, sizeof(*vol_args));
1337 if (IS_ERR(vol_args))
1338 return PTR_ERR(vol_args);
1339 vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
1341 if (vol_args->flags &
1342 ~(BTRFS_SUBVOL_CREATE_ASYNC | BTRFS_SUBVOL_RDONLY)) {
1347 if (vol_args->flags & BTRFS_SUBVOL_CREATE_ASYNC)
1349 if (vol_args->flags & BTRFS_SUBVOL_RDONLY)
1352 ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
1353 vol_args->fd, subvol,
1356 if (ret == 0 && ptr &&
1358 offsetof(struct btrfs_ioctl_vol_args_v2,
1359 transid), ptr, sizeof(*ptr)))
1366 static noinline int btrfs_ioctl_subvol_getflags(struct file *file,
1369 struct inode *inode = fdentry(file)->d_inode;
1370 struct btrfs_root *root = BTRFS_I(inode)->root;
1374 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID)
1377 down_read(&root->fs_info->subvol_sem);
1378 if (btrfs_root_readonly(root))
1379 flags |= BTRFS_SUBVOL_RDONLY;
1380 up_read(&root->fs_info->subvol_sem);
1382 if (copy_to_user(arg, &flags, sizeof(flags)))
1388 static noinline int btrfs_ioctl_subvol_setflags(struct file *file,
1391 struct inode *inode = fdentry(file)->d_inode;
1392 struct btrfs_root *root = BTRFS_I(inode)->root;
1393 struct btrfs_trans_handle *trans;
1398 if (root->fs_info->sb->s_flags & MS_RDONLY)
1401 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID)
1404 if (copy_from_user(&flags, arg, sizeof(flags)))
1407 if (flags & BTRFS_SUBVOL_CREATE_ASYNC)
1410 if (flags & ~BTRFS_SUBVOL_RDONLY)
1413 if (!inode_owner_or_capable(inode))
1416 down_write(&root->fs_info->subvol_sem);
1419 if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root))
1422 root_flags = btrfs_root_flags(&root->root_item);
1423 if (flags & BTRFS_SUBVOL_RDONLY)
1424 btrfs_set_root_flags(&root->root_item,
1425 root_flags | BTRFS_ROOT_SUBVOL_RDONLY);
1427 btrfs_set_root_flags(&root->root_item,
1428 root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY);
1430 trans = btrfs_start_transaction(root, 1);
1431 if (IS_ERR(trans)) {
1432 ret = PTR_ERR(trans);
1436 ret = btrfs_update_root(trans, root->fs_info->tree_root,
1437 &root->root_key, &root->root_item);
1439 btrfs_commit_transaction(trans, root);
1442 btrfs_set_root_flags(&root->root_item, root_flags);
1444 up_write(&root->fs_info->subvol_sem);
1449 * helper to check if the subvolume references other subvolumes
1451 static noinline int may_destroy_subvol(struct btrfs_root *root)
1453 struct btrfs_path *path;
1454 struct btrfs_key key;
1457 path = btrfs_alloc_path();
1461 key.objectid = root->root_key.objectid;
1462 key.type = BTRFS_ROOT_REF_KEY;
1463 key.offset = (u64)-1;
1465 ret = btrfs_search_slot(NULL, root->fs_info->tree_root,
1472 if (path->slots[0] > 0) {
1474 btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
1475 if (key.objectid == root->root_key.objectid &&
1476 key.type == BTRFS_ROOT_REF_KEY)
1480 btrfs_free_path(path);
1484 static noinline int key_in_sk(struct btrfs_key *key,
1485 struct btrfs_ioctl_search_key *sk)
1487 struct btrfs_key test;
1490 test.objectid = sk->min_objectid;
1491 test.type = sk->min_type;
1492 test.offset = sk->min_offset;
1494 ret = btrfs_comp_cpu_keys(key, &test);
1498 test.objectid = sk->max_objectid;
1499 test.type = sk->max_type;
1500 test.offset = sk->max_offset;
1502 ret = btrfs_comp_cpu_keys(key, &test);
1508 static noinline int copy_to_sk(struct btrfs_root *root,
1509 struct btrfs_path *path,
1510 struct btrfs_key *key,
1511 struct btrfs_ioctl_search_key *sk,
1513 unsigned long *sk_offset,
1517 struct extent_buffer *leaf;
1518 struct btrfs_ioctl_search_header sh;
1519 unsigned long item_off;
1520 unsigned long item_len;
1526 leaf = path->nodes[0];
1527 slot = path->slots[0];
1528 nritems = btrfs_header_nritems(leaf);
1530 if (btrfs_header_generation(leaf) > sk->max_transid) {
1534 found_transid = btrfs_header_generation(leaf);
1536 for (i = slot; i < nritems; i++) {
1537 item_off = btrfs_item_ptr_offset(leaf, i);
1538 item_len = btrfs_item_size_nr(leaf, i);
1540 if (item_len > BTRFS_SEARCH_ARGS_BUFSIZE)
1543 if (sizeof(sh) + item_len + *sk_offset >
1544 BTRFS_SEARCH_ARGS_BUFSIZE) {
1549 btrfs_item_key_to_cpu(leaf, key, i);
1550 if (!key_in_sk(key, sk))
1553 sh.objectid = key->objectid;
1554 sh.offset = key->offset;
1555 sh.type = key->type;
1557 sh.transid = found_transid;
1559 /* copy search result header */
1560 memcpy(buf + *sk_offset, &sh, sizeof(sh));
1561 *sk_offset += sizeof(sh);
1564 char *p = buf + *sk_offset;
1566 read_extent_buffer(leaf, p,
1567 item_off, item_len);
1568 *sk_offset += item_len;
1572 if (*num_found >= sk->nr_items)
1577 if (key->offset < (u64)-1 && key->offset < sk->max_offset)
1579 else if (key->type < (u8)-1 && key->type < sk->max_type) {
1582 } else if (key->objectid < (u64)-1 && key->objectid < sk->max_objectid) {
1592 static noinline int search_ioctl(struct inode *inode,
1593 struct btrfs_ioctl_search_args *args)
1595 struct btrfs_root *root;
1596 struct btrfs_key key;
1597 struct btrfs_key max_key;
1598 struct btrfs_path *path;
1599 struct btrfs_ioctl_search_key *sk = &args->key;
1600 struct btrfs_fs_info *info = BTRFS_I(inode)->root->fs_info;
1603 unsigned long sk_offset = 0;
1605 path = btrfs_alloc_path();
1609 if (sk->tree_id == 0) {
1610 /* search the root of the inode that was passed */
1611 root = BTRFS_I(inode)->root;
1613 key.objectid = sk->tree_id;
1614 key.type = BTRFS_ROOT_ITEM_KEY;
1615 key.offset = (u64)-1;
1616 root = btrfs_read_fs_root_no_name(info, &key);
1618 printk(KERN_ERR "could not find root %llu\n",
1620 btrfs_free_path(path);
1625 key.objectid = sk->min_objectid;
1626 key.type = sk->min_type;
1627 key.offset = sk->min_offset;
1629 max_key.objectid = sk->max_objectid;
1630 max_key.type = sk->max_type;
1631 max_key.offset = sk->max_offset;
1633 path->keep_locks = 1;
1636 ret = btrfs_search_forward(root, &key, &max_key, path, 0,
1643 ret = copy_to_sk(root, path, &key, sk, args->buf,
1644 &sk_offset, &num_found);
1645 btrfs_release_path(path);
1646 if (ret || num_found >= sk->nr_items)
1652 sk->nr_items = num_found;
1653 btrfs_free_path(path);
1657 static noinline int btrfs_ioctl_tree_search(struct file *file,
1660 struct btrfs_ioctl_search_args *args;
1661 struct inode *inode;
1664 if (!capable(CAP_SYS_ADMIN))
1667 args = memdup_user(argp, sizeof(*args));
1669 return PTR_ERR(args);
1671 inode = fdentry(file)->d_inode;
1672 ret = search_ioctl(inode, args);
1673 if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
1680 * Search INODE_REFs to identify path name of 'dirid' directory
1681 * in a 'tree_id' tree. and sets path name to 'name'.
1683 static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info,
1684 u64 tree_id, u64 dirid, char *name)
1686 struct btrfs_root *root;
1687 struct btrfs_key key;
1693 struct btrfs_inode_ref *iref;
1694 struct extent_buffer *l;
1695 struct btrfs_path *path;
1697 if (dirid == BTRFS_FIRST_FREE_OBJECTID) {
1702 path = btrfs_alloc_path();
1706 ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX];
1708 key.objectid = tree_id;
1709 key.type = BTRFS_ROOT_ITEM_KEY;
1710 key.offset = (u64)-1;
1711 root = btrfs_read_fs_root_no_name(info, &key);
1713 printk(KERN_ERR "could not find root %llu\n", tree_id);
1718 key.objectid = dirid;
1719 key.type = BTRFS_INODE_REF_KEY;
1720 key.offset = (u64)-1;
1723 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1728 slot = path->slots[0];
1729 if (ret > 0 && slot > 0)
1731 btrfs_item_key_to_cpu(l, &key, slot);
1733 if (ret > 0 && (key.objectid != dirid ||
1734 key.type != BTRFS_INODE_REF_KEY)) {
1739 iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref);
1740 len = btrfs_inode_ref_name_len(l, iref);
1742 total_len += len + 1;
1747 read_extent_buffer(l, ptr,(unsigned long)(iref + 1), len);
1749 if (key.offset == BTRFS_FIRST_FREE_OBJECTID)
1752 btrfs_release_path(path);
1753 key.objectid = key.offset;
1754 key.offset = (u64)-1;
1755 dirid = key.objectid;
1759 memmove(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;
2062 if (!capable(CAP_SYS_ADMIN))
2065 fi_args = kzalloc(sizeof(*fi_args), GFP_KERNEL);
2069 fi_args->num_devices = fs_devices->num_devices;
2070 memcpy(&fi_args->fsid, root->fs_info->fsid, sizeof(fi_args->fsid));
2072 mutex_lock(&fs_devices->device_list_mutex);
2073 list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
2074 if (device->devid > fi_args->max_id)
2075 fi_args->max_id = device->devid;
2077 mutex_unlock(&fs_devices->device_list_mutex);
2079 if (copy_to_user(arg, fi_args, sizeof(*fi_args)))
2086 static long btrfs_ioctl_dev_info(struct btrfs_root *root, void __user *arg)
2088 struct btrfs_ioctl_dev_info_args *di_args;
2089 struct btrfs_device *dev;
2090 struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
2092 char *s_uuid = NULL;
2093 char empty_uuid[BTRFS_UUID_SIZE] = {0};
2095 if (!capable(CAP_SYS_ADMIN))
2098 di_args = memdup_user(arg, sizeof(*di_args));
2099 if (IS_ERR(di_args))
2100 return PTR_ERR(di_args);
2102 if (memcmp(empty_uuid, di_args->uuid, BTRFS_UUID_SIZE) != 0)
2103 s_uuid = di_args->uuid;
2105 mutex_lock(&fs_devices->device_list_mutex);
2106 dev = btrfs_find_device(root, di_args->devid, s_uuid, NULL);
2107 mutex_unlock(&fs_devices->device_list_mutex);
2114 di_args->devid = dev->devid;
2115 di_args->bytes_used = dev->bytes_used;
2116 di_args->total_bytes = dev->total_bytes;
2117 memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid));
2118 strncpy(di_args->path, dev->name, sizeof(di_args->path));
2121 if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args)))
2128 static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
2129 u64 off, u64 olen, u64 destoff)
2131 struct inode *inode = fdentry(file)->d_inode;
2132 struct btrfs_root *root = BTRFS_I(inode)->root;
2133 struct file *src_file;
2135 struct btrfs_trans_handle *trans;
2136 struct btrfs_path *path;
2137 struct extent_buffer *leaf;
2139 struct btrfs_key key;
2144 u64 bs = root->fs_info->sb->s_blocksize;
2149 * - split compressed inline extents. annoying: we need to
2150 * decompress into destination's address_space (the file offset
2151 * may change, so source mapping won't do), then recompress (or
2152 * otherwise reinsert) a subrange.
2153 * - allow ranges within the same file to be cloned (provided
2154 * they don't overlap)?
2157 /* the destination must be opened for writing */
2158 if (!(file->f_mode & FMODE_WRITE) || (file->f_flags & O_APPEND))
2161 if (btrfs_root_readonly(root))
2164 ret = mnt_want_write(file->f_path.mnt);
2168 src_file = fget(srcfd);
2171 goto out_drop_write;
2174 src = src_file->f_dentry->d_inode;
2180 /* the src must be open for reading */
2181 if (!(src_file->f_mode & FMODE_READ))
2184 /* don't make the dst file partly checksummed */
2185 if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) !=
2186 (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM))
2190 if (S_ISDIR(src->i_mode) || S_ISDIR(inode->i_mode))
2194 if (src->i_sb != inode->i_sb || BTRFS_I(src)->root != root)
2198 buf = vmalloc(btrfs_level_size(root, 0));
2202 path = btrfs_alloc_path();
2210 mutex_lock_nested(&inode->i_mutex, I_MUTEX_PARENT);
2211 mutex_lock_nested(&src->i_mutex, I_MUTEX_CHILD);
2213 mutex_lock_nested(&src->i_mutex, I_MUTEX_PARENT);
2214 mutex_lock_nested(&inode->i_mutex, I_MUTEX_CHILD);
2217 /* determine range to clone */
2219 if (off + len > src->i_size || off + len < off)
2222 olen = len = src->i_size - off;
2223 /* if we extend to eof, continue to block boundary */
2224 if (off + len == src->i_size)
2225 len = ALIGN(src->i_size, bs) - off;
2227 /* verify the end result is block aligned */
2228 if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs) ||
2229 !IS_ALIGNED(destoff, bs))
2232 if (destoff > inode->i_size) {
2233 ret = btrfs_cont_expand(inode, inode->i_size, destoff);
2238 /* truncate page cache pages from target inode range */
2239 truncate_inode_pages_range(&inode->i_data, destoff,
2240 PAGE_CACHE_ALIGN(destoff + len) - 1);
2242 /* do any pending delalloc/csum calc on src, one way or
2243 another, and lock file content */
2245 struct btrfs_ordered_extent *ordered;
2246 lock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
2247 ordered = btrfs_lookup_first_ordered_extent(src, off+len);
2249 !test_range_bit(&BTRFS_I(src)->io_tree, off, off+len,
2250 EXTENT_DELALLOC, 0, NULL))
2252 unlock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
2254 btrfs_put_ordered_extent(ordered);
2255 btrfs_wait_ordered_range(src, off, len);
2259 key.objectid = btrfs_ino(src);
2260 key.type = BTRFS_EXTENT_DATA_KEY;
2265 * note the key will change type as we walk through the
2268 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
2272 nritems = btrfs_header_nritems(path->nodes[0]);
2273 if (path->slots[0] >= nritems) {
2274 ret = btrfs_next_leaf(root, path);
2279 nritems = btrfs_header_nritems(path->nodes[0]);
2281 leaf = path->nodes[0];
2282 slot = path->slots[0];
2284 btrfs_item_key_to_cpu(leaf, &key, slot);
2285 if (btrfs_key_type(&key) > BTRFS_EXTENT_DATA_KEY ||
2286 key.objectid != btrfs_ino(src))
2289 if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) {
2290 struct btrfs_file_extent_item *extent;
2293 struct btrfs_key new_key;
2294 u64 disko = 0, diskl = 0;
2295 u64 datao = 0, datal = 0;
2299 size = btrfs_item_size_nr(leaf, slot);
2300 read_extent_buffer(leaf, buf,
2301 btrfs_item_ptr_offset(leaf, slot),
2304 extent = btrfs_item_ptr(leaf, slot,
2305 struct btrfs_file_extent_item);
2306 comp = btrfs_file_extent_compression(leaf, extent);
2307 type = btrfs_file_extent_type(leaf, extent);
2308 if (type == BTRFS_FILE_EXTENT_REG ||
2309 type == BTRFS_FILE_EXTENT_PREALLOC) {
2310 disko = btrfs_file_extent_disk_bytenr(leaf,
2312 diskl = btrfs_file_extent_disk_num_bytes(leaf,
2314 datao = btrfs_file_extent_offset(leaf, extent);
2315 datal = btrfs_file_extent_num_bytes(leaf,
2317 } else if (type == BTRFS_FILE_EXTENT_INLINE) {
2318 /* take upper bound, may be compressed */
2319 datal = btrfs_file_extent_ram_bytes(leaf,
2322 btrfs_release_path(path);
2324 if (key.offset + datal <= off ||
2325 key.offset >= off+len)
2328 memcpy(&new_key, &key, sizeof(new_key));
2329 new_key.objectid = btrfs_ino(inode);
2330 if (off <= key.offset)
2331 new_key.offset = key.offset + destoff - off;
2333 new_key.offset = destoff;
2336 * 1 - adjusting old extent (we may have to split it)
2337 * 1 - add new extent
2340 trans = btrfs_start_transaction(root, 3);
2341 if (IS_ERR(trans)) {
2342 ret = PTR_ERR(trans);
2346 if (type == BTRFS_FILE_EXTENT_REG ||
2347 type == BTRFS_FILE_EXTENT_PREALLOC) {
2349 * a | --- range to clone ---| b
2350 * | ------------- extent ------------- |
2353 /* substract range b */
2354 if (key.offset + datal > off + len)
2355 datal = off + len - key.offset;
2357 /* substract range a */
2358 if (off > key.offset) {
2359 datao += off - key.offset;
2360 datal -= off - key.offset;
2363 ret = btrfs_drop_extents(trans, inode,
2365 new_key.offset + datal,
2369 ret = btrfs_insert_empty_item(trans, root, path,
2373 leaf = path->nodes[0];
2374 slot = path->slots[0];
2375 write_extent_buffer(leaf, buf,
2376 btrfs_item_ptr_offset(leaf, slot),
2379 extent = btrfs_item_ptr(leaf, slot,
2380 struct btrfs_file_extent_item);
2382 /* disko == 0 means it's a hole */
2386 btrfs_set_file_extent_offset(leaf, extent,
2388 btrfs_set_file_extent_num_bytes(leaf, extent,
2391 inode_add_bytes(inode, datal);
2392 ret = btrfs_inc_extent_ref(trans, root,
2394 root->root_key.objectid,
2396 new_key.offset - datao);
2399 } else if (type == BTRFS_FILE_EXTENT_INLINE) {
2402 if (off > key.offset) {
2403 skip = off - key.offset;
2404 new_key.offset += skip;
2407 if (key.offset + datal > off+len)
2408 trim = key.offset + datal - (off+len);
2410 if (comp && (skip || trim)) {
2412 btrfs_end_transaction(trans, root);
2415 size -= skip + trim;
2416 datal -= skip + trim;
2418 ret = btrfs_drop_extents(trans, inode,
2420 new_key.offset + datal,
2424 ret = btrfs_insert_empty_item(trans, root, path,
2430 btrfs_file_extent_calc_inline_size(0);
2431 memmove(buf+start, buf+start+skip,
2435 leaf = path->nodes[0];
2436 slot = path->slots[0];
2437 write_extent_buffer(leaf, buf,
2438 btrfs_item_ptr_offset(leaf, slot),
2440 inode_add_bytes(inode, datal);
2443 btrfs_mark_buffer_dirty(leaf);
2444 btrfs_release_path(path);
2446 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
2449 * we round up to the block size at eof when
2450 * determining which extents to clone above,
2451 * but shouldn't round up the file size
2453 endoff = new_key.offset + datal;
2454 if (endoff > destoff+olen)
2455 endoff = destoff+olen;
2456 if (endoff > inode->i_size)
2457 btrfs_i_size_write(inode, endoff);
2459 ret = btrfs_update_inode(trans, root, inode);
2461 btrfs_end_transaction(trans, root);
2464 btrfs_release_path(path);
2469 btrfs_release_path(path);
2470 unlock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
2472 mutex_unlock(&src->i_mutex);
2473 mutex_unlock(&inode->i_mutex);
2475 btrfs_free_path(path);
2479 mnt_drop_write(file->f_path.mnt);
2483 static long btrfs_ioctl_clone_range(struct file *file, void __user *argp)
2485 struct btrfs_ioctl_clone_range_args args;
2487 if (copy_from_user(&args, argp, sizeof(args)))
2489 return btrfs_ioctl_clone(file, args.src_fd, args.src_offset,
2490 args.src_length, args.dest_offset);
2494 * there are many ways the trans_start and trans_end ioctls can lead
2495 * to deadlocks. They should only be used by applications that
2496 * basically own the machine, and have a very in depth understanding
2497 * of all the possible deadlocks and enospc problems.
2499 static long btrfs_ioctl_trans_start(struct file *file)
2501 struct inode *inode = fdentry(file)->d_inode;
2502 struct btrfs_root *root = BTRFS_I(inode)->root;
2503 struct btrfs_trans_handle *trans;
2507 if (!capable(CAP_SYS_ADMIN))
2511 if (file->private_data)
2515 if (btrfs_root_readonly(root))
2518 ret = mnt_want_write(file->f_path.mnt);
2522 atomic_inc(&root->fs_info->open_ioctl_trans);
2525 trans = btrfs_start_ioctl_transaction(root);
2529 file->private_data = trans;
2533 atomic_dec(&root->fs_info->open_ioctl_trans);
2534 mnt_drop_write(file->f_path.mnt);
2539 static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp)
2541 struct inode *inode = fdentry(file)->d_inode;
2542 struct btrfs_root *root = BTRFS_I(inode)->root;
2543 struct btrfs_root *new_root;
2544 struct btrfs_dir_item *di;
2545 struct btrfs_trans_handle *trans;
2546 struct btrfs_path *path;
2547 struct btrfs_key location;
2548 struct btrfs_disk_key disk_key;
2549 struct btrfs_super_block *disk_super;
2554 if (!capable(CAP_SYS_ADMIN))
2557 if (copy_from_user(&objectid, argp, sizeof(objectid)))
2561 objectid = root->root_key.objectid;
2563 location.objectid = objectid;
2564 location.type = BTRFS_ROOT_ITEM_KEY;
2565 location.offset = (u64)-1;
2567 new_root = btrfs_read_fs_root_no_name(root->fs_info, &location);
2568 if (IS_ERR(new_root))
2569 return PTR_ERR(new_root);
2571 if (btrfs_root_refs(&new_root->root_item) == 0)
2574 path = btrfs_alloc_path();
2577 path->leave_spinning = 1;
2579 trans = btrfs_start_transaction(root, 1);
2580 if (IS_ERR(trans)) {
2581 btrfs_free_path(path);
2582 return PTR_ERR(trans);
2585 dir_id = btrfs_super_root_dir(&root->fs_info->super_copy);
2586 di = btrfs_lookup_dir_item(trans, root->fs_info->tree_root, path,
2587 dir_id, "default", 7, 1);
2588 if (IS_ERR_OR_NULL(di)) {
2589 btrfs_free_path(path);
2590 btrfs_end_transaction(trans, root);
2591 printk(KERN_ERR "Umm, you don't have the default dir item, "
2592 "this isn't going to work\n");
2596 btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key);
2597 btrfs_set_dir_item_key(path->nodes[0], di, &disk_key);
2598 btrfs_mark_buffer_dirty(path->nodes[0]);
2599 btrfs_free_path(path);
2601 disk_super = &root->fs_info->super_copy;
2602 features = btrfs_super_incompat_flags(disk_super);
2603 if (!(features & BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL)) {
2604 features |= BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL;
2605 btrfs_set_super_incompat_flags(disk_super, features);
2607 btrfs_end_transaction(trans, root);
2612 static void get_block_group_info(struct list_head *groups_list,
2613 struct btrfs_ioctl_space_info *space)
2615 struct btrfs_block_group_cache *block_group;
2617 space->total_bytes = 0;
2618 space->used_bytes = 0;
2620 list_for_each_entry(block_group, groups_list, list) {
2621 space->flags = block_group->flags;
2622 space->total_bytes += block_group->key.offset;
2623 space->used_bytes +=
2624 btrfs_block_group_used(&block_group->item);
2628 long btrfs_ioctl_space_info(struct btrfs_root *root, void __user *arg)
2630 struct btrfs_ioctl_space_args space_args;
2631 struct btrfs_ioctl_space_info space;
2632 struct btrfs_ioctl_space_info *dest;
2633 struct btrfs_ioctl_space_info *dest_orig;
2634 struct btrfs_ioctl_space_info __user *user_dest;
2635 struct btrfs_space_info *info;
2636 u64 types[] = {BTRFS_BLOCK_GROUP_DATA,
2637 BTRFS_BLOCK_GROUP_SYSTEM,
2638 BTRFS_BLOCK_GROUP_METADATA,
2639 BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA};
2646 if (copy_from_user(&space_args,
2647 (struct btrfs_ioctl_space_args __user *)arg,
2648 sizeof(space_args)))
2651 for (i = 0; i < num_types; i++) {
2652 struct btrfs_space_info *tmp;
2656 list_for_each_entry_rcu(tmp, &root->fs_info->space_info,
2658 if (tmp->flags == types[i]) {
2668 down_read(&info->groups_sem);
2669 for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
2670 if (!list_empty(&info->block_groups[c]))
2673 up_read(&info->groups_sem);
2676 /* space_slots == 0 means they are asking for a count */
2677 if (space_args.space_slots == 0) {
2678 space_args.total_spaces = slot_count;
2682 slot_count = min_t(u64, space_args.space_slots, slot_count);
2684 alloc_size = sizeof(*dest) * slot_count;
2686 /* we generally have at most 6 or so space infos, one for each raid
2687 * level. So, a whole page should be more than enough for everyone
2689 if (alloc_size > PAGE_CACHE_SIZE)
2692 space_args.total_spaces = 0;
2693 dest = kmalloc(alloc_size, GFP_NOFS);
2698 /* now we have a buffer to copy into */
2699 for (i = 0; i < num_types; i++) {
2700 struct btrfs_space_info *tmp;
2707 list_for_each_entry_rcu(tmp, &root->fs_info->space_info,
2709 if (tmp->flags == types[i]) {
2718 down_read(&info->groups_sem);
2719 for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
2720 if (!list_empty(&info->block_groups[c])) {
2721 get_block_group_info(&info->block_groups[c],
2723 memcpy(dest, &space, sizeof(space));
2725 space_args.total_spaces++;
2731 up_read(&info->groups_sem);
2734 user_dest = (struct btrfs_ioctl_space_info *)
2735 (arg + sizeof(struct btrfs_ioctl_space_args));
2737 if (copy_to_user(user_dest, dest_orig, alloc_size))
2742 if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args)))
2749 * there are many ways the trans_start and trans_end ioctls can lead
2750 * to deadlocks. They should only be used by applications that
2751 * basically own the machine, and have a very in depth understanding
2752 * of all the possible deadlocks and enospc problems.
2754 long btrfs_ioctl_trans_end(struct file *file)
2756 struct inode *inode = fdentry(file)->d_inode;
2757 struct btrfs_root *root = BTRFS_I(inode)->root;
2758 struct btrfs_trans_handle *trans;
2760 trans = file->private_data;
2763 file->private_data = NULL;
2765 btrfs_end_transaction(trans, root);
2767 atomic_dec(&root->fs_info->open_ioctl_trans);
2769 mnt_drop_write(file->f_path.mnt);
2773 static noinline long btrfs_ioctl_start_sync(struct file *file, void __user *argp)
2775 struct btrfs_root *root = BTRFS_I(file->f_dentry->d_inode)->root;
2776 struct btrfs_trans_handle *trans;
2780 trans = btrfs_start_transaction(root, 0);
2782 return PTR_ERR(trans);
2783 transid = trans->transid;
2784 ret = btrfs_commit_transaction_async(trans, root, 0);
2786 btrfs_end_transaction(trans, root);
2791 if (copy_to_user(argp, &transid, sizeof(transid)))
2796 static noinline long btrfs_ioctl_wait_sync(struct file *file, void __user *argp)
2798 struct btrfs_root *root = BTRFS_I(file->f_dentry->d_inode)->root;
2802 if (copy_from_user(&transid, argp, sizeof(transid)))
2805 transid = 0; /* current trans */
2807 return btrfs_wait_for_commit(root, transid);
2810 static long btrfs_ioctl_scrub(struct btrfs_root *root, void __user *arg)
2813 struct btrfs_ioctl_scrub_args *sa;
2815 if (!capable(CAP_SYS_ADMIN))
2818 sa = memdup_user(arg, sizeof(*sa));
2822 ret = btrfs_scrub_dev(root, sa->devid, sa->start, sa->end,
2823 &sa->progress, sa->flags & BTRFS_SCRUB_READONLY);
2825 if (copy_to_user(arg, sa, sizeof(*sa)))
2832 static long btrfs_ioctl_scrub_cancel(struct btrfs_root *root, void __user *arg)
2834 if (!capable(CAP_SYS_ADMIN))
2837 return btrfs_scrub_cancel(root);
2840 static long btrfs_ioctl_scrub_progress(struct btrfs_root *root,
2843 struct btrfs_ioctl_scrub_args *sa;
2846 if (!capable(CAP_SYS_ADMIN))
2849 sa = memdup_user(arg, sizeof(*sa));
2853 ret = btrfs_scrub_progress(root, sa->devid, &sa->progress);
2855 if (copy_to_user(arg, sa, sizeof(*sa)))
2862 long btrfs_ioctl(struct file *file, unsigned int
2863 cmd, unsigned long arg)
2865 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
2866 void __user *argp = (void __user *)arg;
2869 case FS_IOC_GETFLAGS:
2870 return btrfs_ioctl_getflags(file, argp);
2871 case FS_IOC_SETFLAGS:
2872 return btrfs_ioctl_setflags(file, argp);
2873 case FS_IOC_GETVERSION:
2874 return btrfs_ioctl_getversion(file, argp);
2876 return btrfs_ioctl_fitrim(file, argp);
2877 case BTRFS_IOC_SNAP_CREATE:
2878 return btrfs_ioctl_snap_create(file, argp, 0);
2879 case BTRFS_IOC_SNAP_CREATE_V2:
2880 return btrfs_ioctl_snap_create_v2(file, argp, 0);
2881 case BTRFS_IOC_SUBVOL_CREATE:
2882 return btrfs_ioctl_snap_create(file, argp, 1);
2883 case BTRFS_IOC_SNAP_DESTROY:
2884 return btrfs_ioctl_snap_destroy(file, argp);
2885 case BTRFS_IOC_SUBVOL_GETFLAGS:
2886 return btrfs_ioctl_subvol_getflags(file, argp);
2887 case BTRFS_IOC_SUBVOL_SETFLAGS:
2888 return btrfs_ioctl_subvol_setflags(file, argp);
2889 case BTRFS_IOC_DEFAULT_SUBVOL:
2890 return btrfs_ioctl_default_subvol(file, argp);
2891 case BTRFS_IOC_DEFRAG:
2892 return btrfs_ioctl_defrag(file, NULL);
2893 case BTRFS_IOC_DEFRAG_RANGE:
2894 return btrfs_ioctl_defrag(file, argp);
2895 case BTRFS_IOC_RESIZE:
2896 return btrfs_ioctl_resize(root, argp);
2897 case BTRFS_IOC_ADD_DEV:
2898 return btrfs_ioctl_add_dev(root, argp);
2899 case BTRFS_IOC_RM_DEV:
2900 return btrfs_ioctl_rm_dev(root, argp);
2901 case BTRFS_IOC_FS_INFO:
2902 return btrfs_ioctl_fs_info(root, argp);
2903 case BTRFS_IOC_DEV_INFO:
2904 return btrfs_ioctl_dev_info(root, argp);
2905 case BTRFS_IOC_BALANCE:
2906 return btrfs_balance(root->fs_info->dev_root);
2907 case BTRFS_IOC_CLONE:
2908 return btrfs_ioctl_clone(file, arg, 0, 0, 0);
2909 case BTRFS_IOC_CLONE_RANGE:
2910 return btrfs_ioctl_clone_range(file, argp);
2911 case BTRFS_IOC_TRANS_START:
2912 return btrfs_ioctl_trans_start(file);
2913 case BTRFS_IOC_TRANS_END:
2914 return btrfs_ioctl_trans_end(file);
2915 case BTRFS_IOC_TREE_SEARCH:
2916 return btrfs_ioctl_tree_search(file, argp);
2917 case BTRFS_IOC_INO_LOOKUP:
2918 return btrfs_ioctl_ino_lookup(file, argp);
2919 case BTRFS_IOC_SPACE_INFO:
2920 return btrfs_ioctl_space_info(root, argp);
2921 case BTRFS_IOC_SYNC:
2922 btrfs_sync_fs(file->f_dentry->d_sb, 1);
2924 case BTRFS_IOC_START_SYNC:
2925 return btrfs_ioctl_start_sync(file, argp);
2926 case BTRFS_IOC_WAIT_SYNC:
2927 return btrfs_ioctl_wait_sync(file, argp);
2928 case BTRFS_IOC_SCRUB:
2929 return btrfs_ioctl_scrub(root, argp);
2930 case BTRFS_IOC_SCRUB_CANCEL:
2931 return btrfs_ioctl_scrub_cancel(root, argp);
2932 case BTRFS_IOC_SCRUB_PROGRESS:
2933 return btrfs_ioctl_scrub_progress(root, argp);