Commit | Line | Data |
---|---|---|
f46b5a66 CH |
1 | /* |
2 | * Copyright (C) 2007 Oracle. All rights reserved. | |
3 | * | |
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. | |
7 | * | |
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. | |
12 | * | |
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. | |
17 | */ | |
18 | ||
19 | #include <linux/kernel.h> | |
20 | #include <linux/bio.h> | |
21 | #include <linux/buffer_head.h> | |
22 | #include <linux/file.h> | |
23 | #include <linux/fs.h> | |
cb8e7090 | 24 | #include <linux/fsnotify.h> |
f46b5a66 CH |
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> | |
f46b5a66 | 30 | #include <linux/backing-dev.h> |
cb8e7090 | 31 | #include <linux/mount.h> |
f46b5a66 | 32 | #include <linux/mpage.h> |
cb8e7090 | 33 | #include <linux/namei.h> |
f46b5a66 CH |
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> | |
cb8e7090 | 39 | #include <linux/security.h> |
f46b5a66 | 40 | #include <linux/xattr.h> |
7ea394f1 | 41 | #include <linux/vmalloc.h> |
5a0e3ad6 | 42 | #include <linux/slab.h> |
f7039b1d | 43 | #include <linux/blkdev.h> |
8ea05e3a | 44 | #include <linux/uuid.h> |
55e301fd | 45 | #include <linux/btrfs.h> |
4b4e25f2 | 46 | #include "compat.h" |
f46b5a66 CH |
47 | #include "ctree.h" |
48 | #include "disk-io.h" | |
49 | #include "transaction.h" | |
50 | #include "btrfs_inode.h" | |
f46b5a66 CH |
51 | #include "print-tree.h" |
52 | #include "volumes.h" | |
925baedd | 53 | #include "locking.h" |
581bb050 | 54 | #include "inode-map.h" |
d7728c96 | 55 | #include "backref.h" |
606686ee | 56 | #include "rcu-string.h" |
31db9f7c | 57 | #include "send.h" |
3f6bcfbd | 58 | #include "dev-replace.h" |
f46b5a66 | 59 | |
6cbff00f CH |
60 | /* Mask out flags that are inappropriate for the given type of inode. */ |
61 | static inline __u32 btrfs_mask_flags(umode_t mode, __u32 flags) | |
62 | { | |
63 | if (S_ISDIR(mode)) | |
64 | return flags; | |
65 | else if (S_ISREG(mode)) | |
66 | return flags & ~FS_DIRSYNC_FL; | |
67 | else | |
68 | return flags & (FS_NODUMP_FL | FS_NOATIME_FL); | |
69 | } | |
70 | ||
71 | /* | |
72 | * Export inode flags to the format expected by the FS_IOC_GETFLAGS ioctl. | |
73 | */ | |
74 | static unsigned int btrfs_flags_to_ioctl(unsigned int flags) | |
75 | { | |
76 | unsigned int iflags = 0; | |
77 | ||
78 | if (flags & BTRFS_INODE_SYNC) | |
79 | iflags |= FS_SYNC_FL; | |
80 | if (flags & BTRFS_INODE_IMMUTABLE) | |
81 | iflags |= FS_IMMUTABLE_FL; | |
82 | if (flags & BTRFS_INODE_APPEND) | |
83 | iflags |= FS_APPEND_FL; | |
84 | if (flags & BTRFS_INODE_NODUMP) | |
85 | iflags |= FS_NODUMP_FL; | |
86 | if (flags & BTRFS_INODE_NOATIME) | |
87 | iflags |= FS_NOATIME_FL; | |
88 | if (flags & BTRFS_INODE_DIRSYNC) | |
89 | iflags |= FS_DIRSYNC_FL; | |
d0092bdd LZ |
90 | if (flags & BTRFS_INODE_NODATACOW) |
91 | iflags |= FS_NOCOW_FL; | |
92 | ||
93 | if ((flags & BTRFS_INODE_COMPRESS) && !(flags & BTRFS_INODE_NOCOMPRESS)) | |
94 | iflags |= FS_COMPR_FL; | |
95 | else if (flags & BTRFS_INODE_NOCOMPRESS) | |
96 | iflags |= FS_NOCOMP_FL; | |
6cbff00f CH |
97 | |
98 | return iflags; | |
99 | } | |
100 | ||
101 | /* | |
102 | * Update inode->i_flags based on the btrfs internal flags. | |
103 | */ | |
104 | void btrfs_update_iflags(struct inode *inode) | |
105 | { | |
106 | struct btrfs_inode *ip = BTRFS_I(inode); | |
107 | ||
108 | inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC); | |
109 | ||
110 | if (ip->flags & BTRFS_INODE_SYNC) | |
111 | inode->i_flags |= S_SYNC; | |
112 | if (ip->flags & BTRFS_INODE_IMMUTABLE) | |
113 | inode->i_flags |= S_IMMUTABLE; | |
114 | if (ip->flags & BTRFS_INODE_APPEND) | |
115 | inode->i_flags |= S_APPEND; | |
116 | if (ip->flags & BTRFS_INODE_NOATIME) | |
117 | inode->i_flags |= S_NOATIME; | |
118 | if (ip->flags & BTRFS_INODE_DIRSYNC) | |
119 | inode->i_flags |= S_DIRSYNC; | |
120 | } | |
121 | ||
122 | /* | |
123 | * Inherit flags from the parent inode. | |
124 | * | |
e27425d6 | 125 | * Currently only the compression flags and the cow flags are inherited. |
6cbff00f CH |
126 | */ |
127 | void btrfs_inherit_iflags(struct inode *inode, struct inode *dir) | |
128 | { | |
0b4dcea5 CM |
129 | unsigned int flags; |
130 | ||
131 | if (!dir) | |
132 | return; | |
133 | ||
134 | flags = BTRFS_I(dir)->flags; | |
6cbff00f | 135 | |
e27425d6 JB |
136 | if (flags & BTRFS_INODE_NOCOMPRESS) { |
137 | BTRFS_I(inode)->flags &= ~BTRFS_INODE_COMPRESS; | |
138 | BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS; | |
139 | } else if (flags & BTRFS_INODE_COMPRESS) { | |
140 | BTRFS_I(inode)->flags &= ~BTRFS_INODE_NOCOMPRESS; | |
141 | BTRFS_I(inode)->flags |= BTRFS_INODE_COMPRESS; | |
142 | } | |
143 | ||
213490b3 | 144 | if (flags & BTRFS_INODE_NODATACOW) { |
e27425d6 | 145 | BTRFS_I(inode)->flags |= BTRFS_INODE_NODATACOW; |
213490b3 LB |
146 | if (S_ISREG(inode->i_mode)) |
147 | BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM; | |
148 | } | |
6cbff00f | 149 | |
6cbff00f CH |
150 | btrfs_update_iflags(inode); |
151 | } | |
152 | ||
153 | static int btrfs_ioctl_getflags(struct file *file, void __user *arg) | |
154 | { | |
496ad9aa | 155 | struct btrfs_inode *ip = BTRFS_I(file_inode(file)); |
6cbff00f CH |
156 | unsigned int flags = btrfs_flags_to_ioctl(ip->flags); |
157 | ||
158 | if (copy_to_user(arg, &flags, sizeof(flags))) | |
159 | return -EFAULT; | |
160 | return 0; | |
161 | } | |
162 | ||
75e7cb7f LB |
163 | static int check_flags(unsigned int flags) |
164 | { | |
165 | if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \ | |
166 | FS_NOATIME_FL | FS_NODUMP_FL | \ | |
167 | FS_SYNC_FL | FS_DIRSYNC_FL | \ | |
e1e8fb6a LZ |
168 | FS_NOCOMP_FL | FS_COMPR_FL | |
169 | FS_NOCOW_FL)) | |
75e7cb7f LB |
170 | return -EOPNOTSUPP; |
171 | ||
172 | if ((flags & FS_NOCOMP_FL) && (flags & FS_COMPR_FL)) | |
173 | return -EINVAL; | |
174 | ||
75e7cb7f LB |
175 | return 0; |
176 | } | |
177 | ||
6cbff00f CH |
178 | static int btrfs_ioctl_setflags(struct file *file, void __user *arg) |
179 | { | |
496ad9aa | 180 | struct inode *inode = file_inode(file); |
6cbff00f CH |
181 | struct btrfs_inode *ip = BTRFS_I(inode); |
182 | struct btrfs_root *root = ip->root; | |
183 | struct btrfs_trans_handle *trans; | |
184 | unsigned int flags, oldflags; | |
185 | int ret; | |
f062abf0 LZ |
186 | u64 ip_oldflags; |
187 | unsigned int i_oldflags; | |
7e97b8da | 188 | umode_t mode; |
6cbff00f | 189 | |
b83cc969 LZ |
190 | if (btrfs_root_readonly(root)) |
191 | return -EROFS; | |
192 | ||
6cbff00f CH |
193 | if (copy_from_user(&flags, arg, sizeof(flags))) |
194 | return -EFAULT; | |
195 | ||
75e7cb7f LB |
196 | ret = check_flags(flags); |
197 | if (ret) | |
198 | return ret; | |
f46b5a66 | 199 | |
2e149670 | 200 | if (!inode_owner_or_capable(inode)) |
6cbff00f CH |
201 | return -EACCES; |
202 | ||
e7848683 JK |
203 | ret = mnt_want_write_file(file); |
204 | if (ret) | |
205 | return ret; | |
206 | ||
6cbff00f CH |
207 | mutex_lock(&inode->i_mutex); |
208 | ||
f062abf0 LZ |
209 | ip_oldflags = ip->flags; |
210 | i_oldflags = inode->i_flags; | |
7e97b8da | 211 | mode = inode->i_mode; |
f062abf0 | 212 | |
6cbff00f CH |
213 | flags = btrfs_mask_flags(inode->i_mode, flags); |
214 | oldflags = btrfs_flags_to_ioctl(ip->flags); | |
215 | if ((flags ^ oldflags) & (FS_APPEND_FL | FS_IMMUTABLE_FL)) { | |
216 | if (!capable(CAP_LINUX_IMMUTABLE)) { | |
217 | ret = -EPERM; | |
218 | goto out_unlock; | |
219 | } | |
220 | } | |
221 | ||
6cbff00f CH |
222 | if (flags & FS_SYNC_FL) |
223 | ip->flags |= BTRFS_INODE_SYNC; | |
224 | else | |
225 | ip->flags &= ~BTRFS_INODE_SYNC; | |
226 | if (flags & FS_IMMUTABLE_FL) | |
227 | ip->flags |= BTRFS_INODE_IMMUTABLE; | |
228 | else | |
229 | ip->flags &= ~BTRFS_INODE_IMMUTABLE; | |
230 | if (flags & FS_APPEND_FL) | |
231 | ip->flags |= BTRFS_INODE_APPEND; | |
232 | else | |
233 | ip->flags &= ~BTRFS_INODE_APPEND; | |
234 | if (flags & FS_NODUMP_FL) | |
235 | ip->flags |= BTRFS_INODE_NODUMP; | |
236 | else | |
237 | ip->flags &= ~BTRFS_INODE_NODUMP; | |
238 | if (flags & FS_NOATIME_FL) | |
239 | ip->flags |= BTRFS_INODE_NOATIME; | |
240 | else | |
241 | ip->flags &= ~BTRFS_INODE_NOATIME; | |
242 | if (flags & FS_DIRSYNC_FL) | |
243 | ip->flags |= BTRFS_INODE_DIRSYNC; | |
244 | else | |
245 | ip->flags &= ~BTRFS_INODE_DIRSYNC; | |
7e97b8da DS |
246 | if (flags & FS_NOCOW_FL) { |
247 | if (S_ISREG(mode)) { | |
248 | /* | |
249 | * It's safe to turn csums off here, no extents exist. | |
250 | * Otherwise we want the flag to reflect the real COW | |
251 | * status of the file and will not set it. | |
252 | */ | |
253 | if (inode->i_size == 0) | |
254 | ip->flags |= BTRFS_INODE_NODATACOW | |
255 | | BTRFS_INODE_NODATASUM; | |
256 | } else { | |
257 | ip->flags |= BTRFS_INODE_NODATACOW; | |
258 | } | |
259 | } else { | |
260 | /* | |
261 | * Revert back under same assuptions as above | |
262 | */ | |
263 | if (S_ISREG(mode)) { | |
264 | if (inode->i_size == 0) | |
265 | ip->flags &= ~(BTRFS_INODE_NODATACOW | |
266 | | BTRFS_INODE_NODATASUM); | |
267 | } else { | |
268 | ip->flags &= ~BTRFS_INODE_NODATACOW; | |
269 | } | |
270 | } | |
6cbff00f | 271 | |
75e7cb7f LB |
272 | /* |
273 | * The COMPRESS flag can only be changed by users, while the NOCOMPRESS | |
274 | * flag may be changed automatically if compression code won't make | |
275 | * things smaller. | |
276 | */ | |
277 | if (flags & FS_NOCOMP_FL) { | |
278 | ip->flags &= ~BTRFS_INODE_COMPRESS; | |
279 | ip->flags |= BTRFS_INODE_NOCOMPRESS; | |
280 | } else if (flags & FS_COMPR_FL) { | |
281 | ip->flags |= BTRFS_INODE_COMPRESS; | |
282 | ip->flags &= ~BTRFS_INODE_NOCOMPRESS; | |
ebcb904d LZ |
283 | } else { |
284 | ip->flags &= ~(BTRFS_INODE_COMPRESS | BTRFS_INODE_NOCOMPRESS); | |
75e7cb7f | 285 | } |
6cbff00f | 286 | |
4da6f1a3 | 287 | trans = btrfs_start_transaction(root, 1); |
f062abf0 LZ |
288 | if (IS_ERR(trans)) { |
289 | ret = PTR_ERR(trans); | |
290 | goto out_drop; | |
291 | } | |
6cbff00f | 292 | |
306424cc | 293 | btrfs_update_iflags(inode); |
0c4d2d95 | 294 | inode_inc_iversion(inode); |
306424cc | 295 | inode->i_ctime = CURRENT_TIME; |
6cbff00f | 296 | ret = btrfs_update_inode(trans, root, inode); |
6cbff00f | 297 | |
6cbff00f | 298 | btrfs_end_transaction(trans, root); |
f062abf0 LZ |
299 | out_drop: |
300 | if (ret) { | |
301 | ip->flags = ip_oldflags; | |
302 | inode->i_flags = i_oldflags; | |
303 | } | |
6cbff00f | 304 | |
6cbff00f CH |
305 | out_unlock: |
306 | mutex_unlock(&inode->i_mutex); | |
e7848683 | 307 | mnt_drop_write_file(file); |
2d4e6f6a | 308 | return ret; |
6cbff00f CH |
309 | } |
310 | ||
311 | static int btrfs_ioctl_getversion(struct file *file, int __user *arg) | |
312 | { | |
496ad9aa | 313 | struct inode *inode = file_inode(file); |
6cbff00f CH |
314 | |
315 | return put_user(inode->i_generation, arg); | |
316 | } | |
f46b5a66 | 317 | |
f7039b1d LD |
318 | static noinline int btrfs_ioctl_fitrim(struct file *file, void __user *arg) |
319 | { | |
815745cf | 320 | struct btrfs_fs_info *fs_info = btrfs_sb(fdentry(file)->d_sb); |
f7039b1d LD |
321 | struct btrfs_device *device; |
322 | struct request_queue *q; | |
323 | struct fstrim_range range; | |
324 | u64 minlen = ULLONG_MAX; | |
325 | u64 num_devices = 0; | |
815745cf | 326 | u64 total_bytes = btrfs_super_total_bytes(fs_info->super_copy); |
f7039b1d LD |
327 | int ret; |
328 | ||
329 | if (!capable(CAP_SYS_ADMIN)) | |
330 | return -EPERM; | |
331 | ||
1f78160c XG |
332 | rcu_read_lock(); |
333 | list_for_each_entry_rcu(device, &fs_info->fs_devices->devices, | |
334 | dev_list) { | |
f7039b1d LD |
335 | if (!device->bdev) |
336 | continue; | |
337 | q = bdev_get_queue(device->bdev); | |
338 | if (blk_queue_discard(q)) { | |
339 | num_devices++; | |
340 | minlen = min((u64)q->limits.discard_granularity, | |
341 | minlen); | |
342 | } | |
343 | } | |
1f78160c | 344 | rcu_read_unlock(); |
f4c697e6 | 345 | |
f7039b1d LD |
346 | if (!num_devices) |
347 | return -EOPNOTSUPP; | |
f7039b1d LD |
348 | if (copy_from_user(&range, arg, sizeof(range))) |
349 | return -EFAULT; | |
e515c18b LC |
350 | if (range.start > total_bytes || |
351 | range.len < fs_info->sb->s_blocksize) | |
f4c697e6 | 352 | return -EINVAL; |
f7039b1d | 353 | |
f4c697e6 | 354 | range.len = min(range.len, total_bytes - range.start); |
f7039b1d | 355 | range.minlen = max(range.minlen, minlen); |
815745cf | 356 | ret = btrfs_trim_fs(fs_info->tree_root, &range); |
f7039b1d LD |
357 | if (ret < 0) |
358 | return ret; | |
359 | ||
360 | if (copy_to_user(arg, &range, sizeof(range))) | |
361 | return -EFAULT; | |
362 | ||
363 | return 0; | |
364 | } | |
365 | ||
d5c12070 | 366 | static noinline int create_subvol(struct inode *dir, |
cb8e7090 | 367 | struct dentry *dentry, |
72fd032e | 368 | char *name, int namelen, |
6f72c7e2 | 369 | u64 *async_transid, |
8696c533 | 370 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 CH |
371 | { |
372 | struct btrfs_trans_handle *trans; | |
373 | struct btrfs_key key; | |
374 | struct btrfs_root_item root_item; | |
375 | struct btrfs_inode_item *inode_item; | |
376 | struct extent_buffer *leaf; | |
d5c12070 | 377 | struct btrfs_root *root = BTRFS_I(dir)->root; |
76dda93c | 378 | struct btrfs_root *new_root; |
d5c12070 | 379 | struct btrfs_block_rsv block_rsv; |
8ea05e3a | 380 | struct timespec cur_time = CURRENT_TIME; |
f46b5a66 CH |
381 | int ret; |
382 | int err; | |
383 | u64 objectid; | |
384 | u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID; | |
3de4586c | 385 | u64 index = 0; |
d5c12070 | 386 | u64 qgroup_reserved; |
8ea05e3a | 387 | uuid_le new_uuid; |
f46b5a66 | 388 | |
581bb050 | 389 | ret = btrfs_find_free_objectid(root->fs_info->tree_root, &objectid); |
2fbe8c8a | 390 | if (ret) |
a22285a6 | 391 | return ret; |
6a912213 | 392 | |
d5c12070 | 393 | btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP); |
9ed74f2d | 394 | /* |
d5c12070 MX |
395 | * The same as the snapshot creation, please see the comment |
396 | * of create_snapshot(). | |
9ed74f2d | 397 | */ |
d5c12070 MX |
398 | ret = btrfs_subvolume_reserve_metadata(root, &block_rsv, |
399 | 7, &qgroup_reserved); | |
400 | if (ret) | |
401 | return ret; | |
402 | ||
403 | trans = btrfs_start_transaction(root, 0); | |
404 | if (IS_ERR(trans)) { | |
405 | ret = PTR_ERR(trans); | |
406 | goto out; | |
407 | } | |
408 | trans->block_rsv = &block_rsv; | |
409 | trans->bytes_reserved = block_rsv.size; | |
f46b5a66 | 410 | |
8696c533 | 411 | ret = btrfs_qgroup_inherit(trans, root->fs_info, 0, objectid, inherit); |
6f72c7e2 AJ |
412 | if (ret) |
413 | goto fail; | |
414 | ||
5d4f98a2 | 415 | leaf = btrfs_alloc_free_block(trans, root, root->leafsize, |
5581a51a | 416 | 0, objectid, NULL, 0, 0, 0); |
8e8a1e31 JB |
417 | if (IS_ERR(leaf)) { |
418 | ret = PTR_ERR(leaf); | |
419 | goto fail; | |
420 | } | |
f46b5a66 | 421 | |
5d4f98a2 | 422 | memset_extent_buffer(leaf, 0, 0, sizeof(struct btrfs_header)); |
f46b5a66 CH |
423 | btrfs_set_header_bytenr(leaf, leaf->start); |
424 | btrfs_set_header_generation(leaf, trans->transid); | |
5d4f98a2 | 425 | btrfs_set_header_backref_rev(leaf, BTRFS_MIXED_BACKREF_REV); |
f46b5a66 CH |
426 | btrfs_set_header_owner(leaf, objectid); |
427 | ||
428 | write_extent_buffer(leaf, root->fs_info->fsid, | |
429 | (unsigned long)btrfs_header_fsid(leaf), | |
430 | BTRFS_FSID_SIZE); | |
5d4f98a2 YZ |
431 | write_extent_buffer(leaf, root->fs_info->chunk_tree_uuid, |
432 | (unsigned long)btrfs_header_chunk_tree_uuid(leaf), | |
433 | BTRFS_UUID_SIZE); | |
f46b5a66 CH |
434 | btrfs_mark_buffer_dirty(leaf); |
435 | ||
8ea05e3a AB |
436 | memset(&root_item, 0, sizeof(root_item)); |
437 | ||
f46b5a66 | 438 | inode_item = &root_item.inode; |
f46b5a66 CH |
439 | inode_item->generation = cpu_to_le64(1); |
440 | inode_item->size = cpu_to_le64(3); | |
441 | inode_item->nlink = cpu_to_le32(1); | |
a76a3cd4 | 442 | inode_item->nbytes = cpu_to_le64(root->leafsize); |
f46b5a66 CH |
443 | inode_item->mode = cpu_to_le32(S_IFDIR | 0755); |
444 | ||
08fe4db1 LZ |
445 | root_item.flags = 0; |
446 | root_item.byte_limit = 0; | |
447 | inode_item->flags = cpu_to_le64(BTRFS_INODE_ROOT_ITEM_INIT); | |
448 | ||
f46b5a66 | 449 | btrfs_set_root_bytenr(&root_item, leaf->start); |
84234f3a | 450 | btrfs_set_root_generation(&root_item, trans->transid); |
f46b5a66 CH |
451 | btrfs_set_root_level(&root_item, 0); |
452 | btrfs_set_root_refs(&root_item, 1); | |
86b9f2ec | 453 | btrfs_set_root_used(&root_item, leaf->len); |
80ff3856 | 454 | btrfs_set_root_last_snapshot(&root_item, 0); |
f46b5a66 | 455 | |
8ea05e3a AB |
456 | btrfs_set_root_generation_v2(&root_item, |
457 | btrfs_root_generation(&root_item)); | |
458 | uuid_le_gen(&new_uuid); | |
459 | memcpy(root_item.uuid, new_uuid.b, BTRFS_UUID_SIZE); | |
460 | root_item.otime.sec = cpu_to_le64(cur_time.tv_sec); | |
dadd1105 | 461 | root_item.otime.nsec = cpu_to_le32(cur_time.tv_nsec); |
8ea05e3a AB |
462 | root_item.ctime = root_item.otime; |
463 | btrfs_set_root_ctransid(&root_item, trans->transid); | |
464 | btrfs_set_root_otransid(&root_item, trans->transid); | |
f46b5a66 | 465 | |
925baedd | 466 | btrfs_tree_unlock(leaf); |
f46b5a66 CH |
467 | free_extent_buffer(leaf); |
468 | leaf = NULL; | |
469 | ||
470 | btrfs_set_root_dirid(&root_item, new_dirid); | |
471 | ||
472 | key.objectid = objectid; | |
5d4f98a2 | 473 | key.offset = 0; |
f46b5a66 CH |
474 | btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY); |
475 | ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key, | |
476 | &root_item); | |
477 | if (ret) | |
478 | goto fail; | |
479 | ||
76dda93c YZ |
480 | key.offset = (u64)-1; |
481 | new_root = btrfs_read_fs_root_no_name(root->fs_info, &key); | |
79787eaa JM |
482 | if (IS_ERR(new_root)) { |
483 | btrfs_abort_transaction(trans, root, PTR_ERR(new_root)); | |
484 | ret = PTR_ERR(new_root); | |
485 | goto fail; | |
486 | } | |
76dda93c YZ |
487 | |
488 | btrfs_record_root_in_trans(trans, new_root); | |
489 | ||
d82a6f1d | 490 | ret = btrfs_create_subvol_root(trans, new_root, new_dirid); |
ce598979 MF |
491 | if (ret) { |
492 | /* We potentially lose an unused inode item here */ | |
79787eaa | 493 | btrfs_abort_transaction(trans, root, ret); |
ce598979 MF |
494 | goto fail; |
495 | } | |
496 | ||
f46b5a66 CH |
497 | /* |
498 | * insert the directory item | |
499 | */ | |
3de4586c | 500 | ret = btrfs_set_inode_index(dir, &index); |
79787eaa JM |
501 | if (ret) { |
502 | btrfs_abort_transaction(trans, root, ret); | |
503 | goto fail; | |
504 | } | |
3de4586c CM |
505 | |
506 | ret = btrfs_insert_dir_item(trans, root, | |
16cdcec7 | 507 | name, namelen, dir, &key, |
3de4586c | 508 | BTRFS_FT_DIR, index); |
79787eaa JM |
509 | if (ret) { |
510 | btrfs_abort_transaction(trans, root, ret); | |
f46b5a66 | 511 | goto fail; |
79787eaa | 512 | } |
0660b5af | 513 | |
52c26179 YZ |
514 | btrfs_i_size_write(dir, dir->i_size + namelen * 2); |
515 | ret = btrfs_update_inode(trans, root, dir); | |
516 | BUG_ON(ret); | |
517 | ||
0660b5af | 518 | ret = btrfs_add_root_ref(trans, root->fs_info->tree_root, |
4df27c4d | 519 | objectid, root->root_key.objectid, |
33345d01 | 520 | btrfs_ino(dir), index, name, namelen); |
0660b5af | 521 | |
76dda93c | 522 | BUG_ON(ret); |
f46b5a66 | 523 | |
f46b5a66 | 524 | fail: |
d5c12070 MX |
525 | trans->block_rsv = NULL; |
526 | trans->bytes_reserved = 0; | |
72fd032e SW |
527 | if (async_transid) { |
528 | *async_transid = trans->transid; | |
529 | err = btrfs_commit_transaction_async(trans, root, 1); | |
00d71c9c MX |
530 | if (err) |
531 | err = btrfs_commit_transaction(trans, root); | |
72fd032e SW |
532 | } else { |
533 | err = btrfs_commit_transaction(trans, root); | |
534 | } | |
f46b5a66 CH |
535 | if (err && !ret) |
536 | ret = err; | |
1a65e24b CM |
537 | |
538 | if (!ret) | |
539 | d_instantiate(dentry, btrfs_lookup_dentry(dir, dentry)); | |
d5c12070 MX |
540 | out: |
541 | btrfs_subvolume_release_metadata(root, &block_rsv, qgroup_reserved); | |
f46b5a66 CH |
542 | return ret; |
543 | } | |
544 | ||
e9662f70 MX |
545 | static int create_snapshot(struct btrfs_root *root, struct inode *dir, |
546 | struct dentry *dentry, char *name, int namelen, | |
547 | u64 *async_transid, bool readonly, | |
548 | struct btrfs_qgroup_inherit *inherit) | |
f46b5a66 | 549 | { |
2e4bfab9 | 550 | struct inode *inode; |
f46b5a66 CH |
551 | struct btrfs_pending_snapshot *pending_snapshot; |
552 | struct btrfs_trans_handle *trans; | |
2e4bfab9 | 553 | int ret; |
f46b5a66 CH |
554 | |
555 | if (!root->ref_cows) | |
556 | return -EINVAL; | |
557 | ||
6a03843d MX |
558 | ret = btrfs_start_delalloc_inodes(root, 0); |
559 | if (ret) | |
560 | return ret; | |
561 | ||
562 | btrfs_wait_ordered_extents(root, 0); | |
563 | ||
3de4586c | 564 | pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS); |
a22285a6 YZ |
565 | if (!pending_snapshot) |
566 | return -ENOMEM; | |
567 | ||
66d8f3dd MX |
568 | btrfs_init_block_rsv(&pending_snapshot->block_rsv, |
569 | BTRFS_BLOCK_RSV_TEMP); | |
d5c12070 MX |
570 | /* |
571 | * 1 - parent dir inode | |
572 | * 2 - dir entries | |
573 | * 1 - root item | |
574 | * 2 - root ref/backref | |
575 | * 1 - root of snapshot | |
576 | */ | |
577 | ret = btrfs_subvolume_reserve_metadata(BTRFS_I(dir)->root, | |
578 | &pending_snapshot->block_rsv, 7, | |
579 | &pending_snapshot->qgroup_reserved); | |
580 | if (ret) | |
581 | goto out; | |
582 | ||
3de4586c | 583 | pending_snapshot->dentry = dentry; |
f46b5a66 | 584 | pending_snapshot->root = root; |
b83cc969 | 585 | pending_snapshot->readonly = readonly; |
e9662f70 | 586 | pending_snapshot->dir = dir; |
8696c533 | 587 | pending_snapshot->inherit = inherit; |
a22285a6 | 588 | |
d5c12070 | 589 | trans = btrfs_start_transaction(root, 0); |
a22285a6 YZ |
590 | if (IS_ERR(trans)) { |
591 | ret = PTR_ERR(trans); | |
592 | goto fail; | |
593 | } | |
594 | ||
8351583e | 595 | spin_lock(&root->fs_info->trans_lock); |
f46b5a66 CH |
596 | list_add(&pending_snapshot->list, |
597 | &trans->transaction->pending_snapshots); | |
8351583e | 598 | spin_unlock(&root->fs_info->trans_lock); |
72fd032e SW |
599 | if (async_transid) { |
600 | *async_transid = trans->transid; | |
601 | ret = btrfs_commit_transaction_async(trans, | |
602 | root->fs_info->extent_root, 1); | |
00d71c9c MX |
603 | if (ret) |
604 | ret = btrfs_commit_transaction(trans, root); | |
72fd032e SW |
605 | } else { |
606 | ret = btrfs_commit_transaction(trans, | |
607 | root->fs_info->extent_root); | |
608 | } | |
aec8030a | 609 | if (ret) |
c37b2b62 | 610 | goto fail; |
a22285a6 YZ |
611 | |
612 | ret = pending_snapshot->error; | |
613 | if (ret) | |
614 | goto fail; | |
615 | ||
66b4ffd1 JB |
616 | ret = btrfs_orphan_cleanup(pending_snapshot->snap); |
617 | if (ret) | |
618 | goto fail; | |
f46b5a66 | 619 | |
2fbe8c8a | 620 | inode = btrfs_lookup_dentry(dentry->d_parent->d_inode, dentry); |
2e4bfab9 YZ |
621 | if (IS_ERR(inode)) { |
622 | ret = PTR_ERR(inode); | |
623 | goto fail; | |
624 | } | |
625 | BUG_ON(!inode); | |
626 | d_instantiate(dentry, inode); | |
627 | ret = 0; | |
628 | fail: | |
d5c12070 MX |
629 | btrfs_subvolume_release_metadata(BTRFS_I(dir)->root, |
630 | &pending_snapshot->block_rsv, | |
631 | pending_snapshot->qgroup_reserved); | |
632 | out: | |
a22285a6 | 633 | kfree(pending_snapshot); |
f46b5a66 CH |
634 | return ret; |
635 | } | |
636 | ||
4260f7c7 SW |
637 | /* copy of check_sticky in fs/namei.c() |
638 | * It's inline, so penalty for filesystems that don't use sticky bit is | |
639 | * minimal. | |
640 | */ | |
641 | static inline int btrfs_check_sticky(struct inode *dir, struct inode *inode) | |
642 | { | |
2f2f43d3 | 643 | kuid_t fsuid = current_fsuid(); |
4260f7c7 SW |
644 | |
645 | if (!(dir->i_mode & S_ISVTX)) | |
646 | return 0; | |
2f2f43d3 | 647 | if (uid_eq(inode->i_uid, fsuid)) |
4260f7c7 | 648 | return 0; |
2f2f43d3 | 649 | if (uid_eq(dir->i_uid, fsuid)) |
4260f7c7 SW |
650 | return 0; |
651 | return !capable(CAP_FOWNER); | |
652 | } | |
653 | ||
654 | /* copy of may_delete in fs/namei.c() | |
655 | * Check whether we can remove a link victim from directory dir, check | |
656 | * whether the type of victim is right. | |
657 | * 1. We can't do it if dir is read-only (done in permission()) | |
658 | * 2. We should have write and exec permissions on dir | |
659 | * 3. We can't remove anything from append-only dir | |
660 | * 4. We can't do anything with immutable dir (done in permission()) | |
661 | * 5. If the sticky bit on dir is set we should either | |
662 | * a. be owner of dir, or | |
663 | * b. be owner of victim, or | |
664 | * c. have CAP_FOWNER capability | |
665 | * 6. If the victim is append-only or immutable we can't do antyhing with | |
666 | * links pointing to it. | |
667 | * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR. | |
668 | * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR. | |
669 | * 9. We can't remove a root or mountpoint. | |
670 | * 10. We don't allow removal of NFS sillyrenamed files; it's handled by | |
671 | * nfs_async_unlink(). | |
672 | */ | |
673 | ||
674 | static int btrfs_may_delete(struct inode *dir,struct dentry *victim,int isdir) | |
675 | { | |
676 | int error; | |
677 | ||
678 | if (!victim->d_inode) | |
679 | return -ENOENT; | |
680 | ||
681 | BUG_ON(victim->d_parent->d_inode != dir); | |
4fa6b5ec | 682 | audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE); |
4260f7c7 SW |
683 | |
684 | error = inode_permission(dir, MAY_WRITE | MAY_EXEC); | |
685 | if (error) | |
686 | return error; | |
687 | if (IS_APPEND(dir)) | |
688 | return -EPERM; | |
689 | if (btrfs_check_sticky(dir, victim->d_inode)|| | |
690 | IS_APPEND(victim->d_inode)|| | |
691 | IS_IMMUTABLE(victim->d_inode) || IS_SWAPFILE(victim->d_inode)) | |
692 | return -EPERM; | |
693 | if (isdir) { | |
694 | if (!S_ISDIR(victim->d_inode->i_mode)) | |
695 | return -ENOTDIR; | |
696 | if (IS_ROOT(victim)) | |
697 | return -EBUSY; | |
698 | } else if (S_ISDIR(victim->d_inode->i_mode)) | |
699 | return -EISDIR; | |
700 | if (IS_DEADDIR(dir)) | |
701 | return -ENOENT; | |
702 | if (victim->d_flags & DCACHE_NFSFS_RENAMED) | |
703 | return -EBUSY; | |
704 | return 0; | |
705 | } | |
706 | ||
cb8e7090 CH |
707 | /* copy of may_create in fs/namei.c() */ |
708 | static inline int btrfs_may_create(struct inode *dir, struct dentry *child) | |
709 | { | |
710 | if (child->d_inode) | |
711 | return -EEXIST; | |
712 | if (IS_DEADDIR(dir)) | |
713 | return -ENOENT; | |
714 | return inode_permission(dir, MAY_WRITE | MAY_EXEC); | |
715 | } | |
716 | ||
717 | /* | |
718 | * Create a new subvolume below @parent. This is largely modeled after | |
719 | * sys_mkdirat and vfs_mkdir, but we only do a single component lookup | |
720 | * inside this filesystem so it's quite a bit simpler. | |
721 | */ | |
76dda93c YZ |
722 | static noinline int btrfs_mksubvol(struct path *parent, |
723 | char *name, int namelen, | |
72fd032e | 724 | struct btrfs_root *snap_src, |
6f72c7e2 | 725 | u64 *async_transid, bool readonly, |
8696c533 | 726 | struct btrfs_qgroup_inherit *inherit) |
cb8e7090 | 727 | { |
76dda93c | 728 | struct inode *dir = parent->dentry->d_inode; |
cb8e7090 CH |
729 | struct dentry *dentry; |
730 | int error; | |
731 | ||
5c50c9b8 DS |
732 | error = mutex_lock_killable_nested(&dir->i_mutex, I_MUTEX_PARENT); |
733 | if (error == -EINTR) | |
734 | return error; | |
cb8e7090 CH |
735 | |
736 | dentry = lookup_one_len(name, parent->dentry, namelen); | |
737 | error = PTR_ERR(dentry); | |
738 | if (IS_ERR(dentry)) | |
739 | goto out_unlock; | |
740 | ||
741 | error = -EEXIST; | |
742 | if (dentry->d_inode) | |
743 | goto out_dput; | |
744 | ||
76dda93c | 745 | error = btrfs_may_create(dir, dentry); |
cb8e7090 | 746 | if (error) |
a874a63e | 747 | goto out_dput; |
cb8e7090 | 748 | |
9c52057c CM |
749 | /* |
750 | * even if this name doesn't exist, we may get hash collisions. | |
751 | * check for them now when we can safely fail | |
752 | */ | |
753 | error = btrfs_check_dir_item_collision(BTRFS_I(dir)->root, | |
754 | dir->i_ino, name, | |
755 | namelen); | |
756 | if (error) | |
757 | goto out_dput; | |
758 | ||
76dda93c YZ |
759 | down_read(&BTRFS_I(dir)->root->fs_info->subvol_sem); |
760 | ||
761 | if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0) | |
762 | goto out_up_read; | |
763 | ||
3de4586c | 764 | if (snap_src) { |
e9662f70 | 765 | error = create_snapshot(snap_src, dir, dentry, name, namelen, |
6f72c7e2 | 766 | async_transid, readonly, inherit); |
3de4586c | 767 | } else { |
d5c12070 MX |
768 | error = create_subvol(dir, dentry, name, namelen, |
769 | async_transid, inherit); | |
3de4586c | 770 | } |
76dda93c YZ |
771 | if (!error) |
772 | fsnotify_mkdir(dir, dentry); | |
773 | out_up_read: | |
774 | up_read(&BTRFS_I(dir)->root->fs_info->subvol_sem); | |
cb8e7090 CH |
775 | out_dput: |
776 | dput(dentry); | |
777 | out_unlock: | |
76dda93c | 778 | mutex_unlock(&dir->i_mutex); |
cb8e7090 CH |
779 | return error; |
780 | } | |
781 | ||
4cb5300b CM |
782 | /* |
783 | * When we're defragging a range, we don't want to kick it off again | |
784 | * if it is really just waiting for delalloc to send it down. | |
785 | * If we find a nice big extent or delalloc range for the bytes in the | |
786 | * file you want to defrag, we return 0 to let you know to skip this | |
787 | * part of the file | |
788 | */ | |
789 | static int check_defrag_in_cache(struct inode *inode, u64 offset, int thresh) | |
790 | { | |
791 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; | |
792 | struct extent_map *em = NULL; | |
793 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
794 | u64 end; | |
795 | ||
796 | read_lock(&em_tree->lock); | |
797 | em = lookup_extent_mapping(em_tree, offset, PAGE_CACHE_SIZE); | |
798 | read_unlock(&em_tree->lock); | |
799 | ||
800 | if (em) { | |
801 | end = extent_map_end(em); | |
802 | free_extent_map(em); | |
803 | if (end - offset > thresh) | |
804 | return 0; | |
805 | } | |
806 | /* if we already have a nice delalloc here, just stop */ | |
807 | thresh /= 2; | |
808 | end = count_range_bits(io_tree, &offset, offset + thresh, | |
809 | thresh, EXTENT_DELALLOC, 1); | |
810 | if (end >= thresh) | |
811 | return 0; | |
812 | return 1; | |
813 | } | |
814 | ||
815 | /* | |
816 | * helper function to walk through a file and find extents | |
817 | * newer than a specific transid, and smaller than thresh. | |
818 | * | |
819 | * This is used by the defragging code to find new and small | |
820 | * extents | |
821 | */ | |
822 | static int find_new_extents(struct btrfs_root *root, | |
823 | struct inode *inode, u64 newer_than, | |
824 | u64 *off, int thresh) | |
825 | { | |
826 | struct btrfs_path *path; | |
827 | struct btrfs_key min_key; | |
828 | struct btrfs_key max_key; | |
829 | struct extent_buffer *leaf; | |
830 | struct btrfs_file_extent_item *extent; | |
831 | int type; | |
832 | int ret; | |
a4689d2b | 833 | u64 ino = btrfs_ino(inode); |
4cb5300b CM |
834 | |
835 | path = btrfs_alloc_path(); | |
836 | if (!path) | |
837 | return -ENOMEM; | |
838 | ||
a4689d2b | 839 | min_key.objectid = ino; |
4cb5300b CM |
840 | min_key.type = BTRFS_EXTENT_DATA_KEY; |
841 | min_key.offset = *off; | |
842 | ||
a4689d2b | 843 | max_key.objectid = ino; |
4cb5300b CM |
844 | max_key.type = (u8)-1; |
845 | max_key.offset = (u64)-1; | |
846 | ||
847 | path->keep_locks = 1; | |
848 | ||
849 | while(1) { | |
850 | ret = btrfs_search_forward(root, &min_key, &max_key, | |
de78b51a | 851 | path, newer_than); |
4cb5300b CM |
852 | if (ret != 0) |
853 | goto none; | |
a4689d2b | 854 | if (min_key.objectid != ino) |
4cb5300b CM |
855 | goto none; |
856 | if (min_key.type != BTRFS_EXTENT_DATA_KEY) | |
857 | goto none; | |
858 | ||
859 | leaf = path->nodes[0]; | |
860 | extent = btrfs_item_ptr(leaf, path->slots[0], | |
861 | struct btrfs_file_extent_item); | |
862 | ||
863 | type = btrfs_file_extent_type(leaf, extent); | |
864 | if (type == BTRFS_FILE_EXTENT_REG && | |
865 | btrfs_file_extent_num_bytes(leaf, extent) < thresh && | |
866 | check_defrag_in_cache(inode, min_key.offset, thresh)) { | |
867 | *off = min_key.offset; | |
868 | btrfs_free_path(path); | |
869 | return 0; | |
870 | } | |
871 | ||
872 | if (min_key.offset == (u64)-1) | |
873 | goto none; | |
874 | ||
875 | min_key.offset++; | |
876 | btrfs_release_path(path); | |
877 | } | |
878 | none: | |
879 | btrfs_free_path(path); | |
880 | return -ENOENT; | |
881 | } | |
882 | ||
6c282eb4 | 883 | static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start) |
17ce6ef8 LB |
884 | { |
885 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
6c282eb4 LZ |
886 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; |
887 | struct extent_map *em; | |
888 | u64 len = PAGE_CACHE_SIZE; | |
17ce6ef8 | 889 | |
6c282eb4 LZ |
890 | /* |
891 | * hopefully we have this extent in the tree already, try without | |
892 | * the full extent lock | |
893 | */ | |
17ce6ef8 | 894 | read_lock(&em_tree->lock); |
6c282eb4 | 895 | em = lookup_extent_mapping(em_tree, start, len); |
17ce6ef8 LB |
896 | read_unlock(&em_tree->lock); |
897 | ||
6c282eb4 LZ |
898 | if (!em) { |
899 | /* get the big lock and read metadata off disk */ | |
900 | lock_extent(io_tree, start, start + len - 1); | |
901 | em = btrfs_get_extent(inode, NULL, 0, start, len, 0); | |
902 | unlock_extent(io_tree, start, start + len - 1); | |
903 | ||
904 | if (IS_ERR(em)) | |
905 | return NULL; | |
906 | } | |
907 | ||
908 | return em; | |
909 | } | |
17ce6ef8 | 910 | |
6c282eb4 LZ |
911 | static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em) |
912 | { | |
913 | struct extent_map *next; | |
914 | bool ret = true; | |
915 | ||
916 | /* this is the last extent */ | |
917 | if (em->start + em->len >= i_size_read(inode)) | |
918 | return false; | |
919 | ||
920 | next = defrag_lookup_extent(inode, em->start + em->len); | |
921 | if (!next || next->block_start >= EXTENT_MAP_LAST_BYTE) | |
922 | ret = false; | |
923 | ||
924 | free_extent_map(next); | |
17ce6ef8 LB |
925 | return ret; |
926 | } | |
927 | ||
6c282eb4 | 928 | static int should_defrag_range(struct inode *inode, u64 start, int thresh, |
a43a2111 AM |
929 | u64 *last_len, u64 *skip, u64 *defrag_end, |
930 | int compress) | |
940100a4 | 931 | { |
6c282eb4 | 932 | struct extent_map *em; |
940100a4 | 933 | int ret = 1; |
6c282eb4 | 934 | bool next_mergeable = true; |
940100a4 CM |
935 | |
936 | /* | |
008873ea | 937 | * make sure that once we start defragging an extent, we keep on |
940100a4 CM |
938 | * defragging it |
939 | */ | |
940 | if (start < *defrag_end) | |
941 | return 1; | |
942 | ||
943 | *skip = 0; | |
944 | ||
6c282eb4 LZ |
945 | em = defrag_lookup_extent(inode, start); |
946 | if (!em) | |
947 | return 0; | |
940100a4 CM |
948 | |
949 | /* this will cover holes, and inline extents */ | |
17ce6ef8 | 950 | if (em->block_start >= EXTENT_MAP_LAST_BYTE) { |
940100a4 | 951 | ret = 0; |
17ce6ef8 LB |
952 | goto out; |
953 | } | |
954 | ||
6c282eb4 | 955 | next_mergeable = defrag_check_next_extent(inode, em); |
940100a4 CM |
956 | |
957 | /* | |
6c282eb4 LZ |
958 | * we hit a real extent, if it is big or the next extent is not a |
959 | * real extent, don't bother defragging it | |
940100a4 | 960 | */ |
a43a2111 | 961 | if (!compress && (*last_len == 0 || *last_len >= thresh) && |
6c282eb4 | 962 | (em->len >= thresh || !next_mergeable)) |
940100a4 | 963 | ret = 0; |
17ce6ef8 | 964 | out: |
940100a4 CM |
965 | /* |
966 | * last_len ends up being a counter of how many bytes we've defragged. | |
967 | * every time we choose not to defrag an extent, we reset *last_len | |
968 | * so that the next tiny extent will force a defrag. | |
969 | * | |
970 | * The end result of this is that tiny extents before a single big | |
971 | * extent will force at least part of that big extent to be defragged. | |
972 | */ | |
973 | if (ret) { | |
940100a4 CM |
974 | *defrag_end = extent_map_end(em); |
975 | } else { | |
976 | *last_len = 0; | |
977 | *skip = extent_map_end(em); | |
978 | *defrag_end = 0; | |
979 | } | |
980 | ||
981 | free_extent_map(em); | |
982 | return ret; | |
983 | } | |
984 | ||
4cb5300b CM |
985 | /* |
986 | * it doesn't do much good to defrag one or two pages | |
987 | * at a time. This pulls in a nice chunk of pages | |
988 | * to COW and defrag. | |
989 | * | |
990 | * It also makes sure the delalloc code has enough | |
991 | * dirty data to avoid making new small extents as part | |
992 | * of the defrag | |
993 | * | |
994 | * It's a good idea to start RA on this range | |
995 | * before calling this. | |
996 | */ | |
997 | static int cluster_pages_for_defrag(struct inode *inode, | |
998 | struct page **pages, | |
999 | unsigned long start_index, | |
1000 | int num_pages) | |
f46b5a66 | 1001 | { |
4cb5300b CM |
1002 | unsigned long file_end; |
1003 | u64 isize = i_size_read(inode); | |
1004 | u64 page_start; | |
1005 | u64 page_end; | |
1f12bd06 | 1006 | u64 page_cnt; |
4cb5300b CM |
1007 | int ret; |
1008 | int i; | |
1009 | int i_done; | |
3eaa2885 | 1010 | struct btrfs_ordered_extent *ordered; |
4cb5300b | 1011 | struct extent_state *cached_state = NULL; |
600a45e1 | 1012 | struct extent_io_tree *tree; |
3b16a4e3 | 1013 | gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping); |
4cb5300b | 1014 | |
4cb5300b | 1015 | file_end = (isize - 1) >> PAGE_CACHE_SHIFT; |
1f12bd06 LB |
1016 | if (!isize || start_index > file_end) |
1017 | return 0; | |
1018 | ||
1019 | page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1); | |
4cb5300b CM |
1020 | |
1021 | ret = btrfs_delalloc_reserve_space(inode, | |
1f12bd06 | 1022 | page_cnt << PAGE_CACHE_SHIFT); |
4cb5300b CM |
1023 | if (ret) |
1024 | return ret; | |
4cb5300b | 1025 | i_done = 0; |
600a45e1 | 1026 | tree = &BTRFS_I(inode)->io_tree; |
4cb5300b CM |
1027 | |
1028 | /* step one, lock all the pages */ | |
1f12bd06 | 1029 | for (i = 0; i < page_cnt; i++) { |
4cb5300b | 1030 | struct page *page; |
600a45e1 | 1031 | again: |
a94733d0 | 1032 | page = find_or_create_page(inode->i_mapping, |
600a45e1 | 1033 | start_index + i, mask); |
4cb5300b CM |
1034 | if (!page) |
1035 | break; | |
1036 | ||
600a45e1 MX |
1037 | page_start = page_offset(page); |
1038 | page_end = page_start + PAGE_CACHE_SIZE - 1; | |
1039 | while (1) { | |
d0082371 | 1040 | lock_extent(tree, page_start, page_end); |
600a45e1 MX |
1041 | ordered = btrfs_lookup_ordered_extent(inode, |
1042 | page_start); | |
d0082371 | 1043 | unlock_extent(tree, page_start, page_end); |
600a45e1 MX |
1044 | if (!ordered) |
1045 | break; | |
1046 | ||
1047 | unlock_page(page); | |
1048 | btrfs_start_ordered_extent(inode, ordered, 1); | |
1049 | btrfs_put_ordered_extent(ordered); | |
1050 | lock_page(page); | |
1f12bd06 LB |
1051 | /* |
1052 | * we unlocked the page above, so we need check if | |
1053 | * it was released or not. | |
1054 | */ | |
1055 | if (page->mapping != inode->i_mapping) { | |
1056 | unlock_page(page); | |
1057 | page_cache_release(page); | |
1058 | goto again; | |
1059 | } | |
600a45e1 MX |
1060 | } |
1061 | ||
4cb5300b CM |
1062 | if (!PageUptodate(page)) { |
1063 | btrfs_readpage(NULL, page); | |
1064 | lock_page(page); | |
1065 | if (!PageUptodate(page)) { | |
1066 | unlock_page(page); | |
1067 | page_cache_release(page); | |
1068 | ret = -EIO; | |
1069 | break; | |
1070 | } | |
1071 | } | |
600a45e1 | 1072 | |
600a45e1 MX |
1073 | if (page->mapping != inode->i_mapping) { |
1074 | unlock_page(page); | |
1075 | page_cache_release(page); | |
1076 | goto again; | |
1077 | } | |
1078 | ||
4cb5300b CM |
1079 | pages[i] = page; |
1080 | i_done++; | |
1081 | } | |
1082 | if (!i_done || ret) | |
1083 | goto out; | |
1084 | ||
1085 | if (!(inode->i_sb->s_flags & MS_ACTIVE)) | |
1086 | goto out; | |
1087 | ||
1088 | /* | |
1089 | * so now we have a nice long stream of locked | |
1090 | * and up to date pages, lets wait on them | |
1091 | */ | |
1092 | for (i = 0; i < i_done; i++) | |
1093 | wait_on_page_writeback(pages[i]); | |
1094 | ||
1095 | page_start = page_offset(pages[0]); | |
1096 | page_end = page_offset(pages[i_done - 1]) + PAGE_CACHE_SIZE; | |
1097 | ||
1098 | lock_extent_bits(&BTRFS_I(inode)->io_tree, | |
d0082371 | 1099 | page_start, page_end - 1, 0, &cached_state); |
4cb5300b CM |
1100 | clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, |
1101 | page_end - 1, EXTENT_DIRTY | EXTENT_DELALLOC | | |
9e8a4a8b LB |
1102 | EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 0, 0, |
1103 | &cached_state, GFP_NOFS); | |
4cb5300b | 1104 | |
1f12bd06 | 1105 | if (i_done != page_cnt) { |
9e0baf60 JB |
1106 | spin_lock(&BTRFS_I(inode)->lock); |
1107 | BTRFS_I(inode)->outstanding_extents++; | |
1108 | spin_unlock(&BTRFS_I(inode)->lock); | |
4cb5300b | 1109 | btrfs_delalloc_release_space(inode, |
1f12bd06 | 1110 | (page_cnt - i_done) << PAGE_CACHE_SHIFT); |
4cb5300b CM |
1111 | } |
1112 | ||
1113 | ||
9e8a4a8b LB |
1114 | set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1, |
1115 | &cached_state, GFP_NOFS); | |
4cb5300b CM |
1116 | |
1117 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, | |
1118 | page_start, page_end - 1, &cached_state, | |
1119 | GFP_NOFS); | |
1120 | ||
1121 | for (i = 0; i < i_done; i++) { | |
1122 | clear_page_dirty_for_io(pages[i]); | |
1123 | ClearPageChecked(pages[i]); | |
1124 | set_page_extent_mapped(pages[i]); | |
1125 | set_page_dirty(pages[i]); | |
1126 | unlock_page(pages[i]); | |
1127 | page_cache_release(pages[i]); | |
1128 | } | |
1129 | return i_done; | |
1130 | out: | |
1131 | for (i = 0; i < i_done; i++) { | |
1132 | unlock_page(pages[i]); | |
1133 | page_cache_release(pages[i]); | |
1134 | } | |
1f12bd06 | 1135 | btrfs_delalloc_release_space(inode, page_cnt << PAGE_CACHE_SHIFT); |
4cb5300b CM |
1136 | return ret; |
1137 | ||
1138 | } | |
1139 | ||
1140 | int btrfs_defrag_file(struct inode *inode, struct file *file, | |
1141 | struct btrfs_ioctl_defrag_range_args *range, | |
1142 | u64 newer_than, unsigned long max_to_defrag) | |
1143 | { | |
1144 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4cb5300b | 1145 | struct file_ra_state *ra = NULL; |
f46b5a66 | 1146 | unsigned long last_index; |
151a31b2 | 1147 | u64 isize = i_size_read(inode); |
940100a4 CM |
1148 | u64 last_len = 0; |
1149 | u64 skip = 0; | |
1150 | u64 defrag_end = 0; | |
4cb5300b | 1151 | u64 newer_off = range->start; |
f46b5a66 | 1152 | unsigned long i; |
008873ea | 1153 | unsigned long ra_index = 0; |
f46b5a66 | 1154 | int ret; |
4cb5300b | 1155 | int defrag_count = 0; |
1a419d85 | 1156 | int compress_type = BTRFS_COMPRESS_ZLIB; |
4cb5300b | 1157 | int extent_thresh = range->extent_thresh; |
008873ea LZ |
1158 | int max_cluster = (256 * 1024) >> PAGE_CACHE_SHIFT; |
1159 | int cluster = max_cluster; | |
4cb5300b CM |
1160 | u64 new_align = ~((u64)128 * 1024 - 1); |
1161 | struct page **pages = NULL; | |
1162 | ||
0abd5b17 LB |
1163 | if (isize == 0) |
1164 | return 0; | |
1165 | ||
1166 | if (range->start >= isize) | |
1167 | return -EINVAL; | |
1a419d85 LZ |
1168 | |
1169 | if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS) { | |
1170 | if (range->compress_type > BTRFS_COMPRESS_TYPES) | |
1171 | return -EINVAL; | |
1172 | if (range->compress_type) | |
1173 | compress_type = range->compress_type; | |
1174 | } | |
f46b5a66 | 1175 | |
0abd5b17 LB |
1176 | if (extent_thresh == 0) |
1177 | extent_thresh = 256 * 1024; | |
940100a4 | 1178 | |
4cb5300b CM |
1179 | /* |
1180 | * if we were not given a file, allocate a readahead | |
1181 | * context | |
1182 | */ | |
1183 | if (!file) { | |
1184 | ra = kzalloc(sizeof(*ra), GFP_NOFS); | |
1185 | if (!ra) | |
1186 | return -ENOMEM; | |
1187 | file_ra_state_init(ra, inode->i_mapping); | |
1188 | } else { | |
1189 | ra = &file->f_ra; | |
1190 | } | |
1191 | ||
008873ea | 1192 | pages = kmalloc(sizeof(struct page *) * max_cluster, |
4cb5300b CM |
1193 | GFP_NOFS); |
1194 | if (!pages) { | |
1195 | ret = -ENOMEM; | |
1196 | goto out_ra; | |
1197 | } | |
1198 | ||
1199 | /* find the last page to defrag */ | |
1e701a32 | 1200 | if (range->start + range->len > range->start) { |
151a31b2 | 1201 | last_index = min_t(u64, isize - 1, |
1e701a32 CM |
1202 | range->start + range->len - 1) >> PAGE_CACHE_SHIFT; |
1203 | } else { | |
151a31b2 | 1204 | last_index = (isize - 1) >> PAGE_CACHE_SHIFT; |
1e701a32 CM |
1205 | } |
1206 | ||
4cb5300b CM |
1207 | if (newer_than) { |
1208 | ret = find_new_extents(root, inode, newer_than, | |
1209 | &newer_off, 64 * 1024); | |
1210 | if (!ret) { | |
1211 | range->start = newer_off; | |
1212 | /* | |
1213 | * we always align our defrag to help keep | |
1214 | * the extents in the file evenly spaced | |
1215 | */ | |
1216 | i = (newer_off & new_align) >> PAGE_CACHE_SHIFT; | |
4cb5300b CM |
1217 | } else |
1218 | goto out_ra; | |
1219 | } else { | |
1220 | i = range->start >> PAGE_CACHE_SHIFT; | |
1221 | } | |
1222 | if (!max_to_defrag) | |
7ec31b54 | 1223 | max_to_defrag = last_index + 1; |
4cb5300b | 1224 | |
2a0f7f57 LZ |
1225 | /* |
1226 | * make writeback starts from i, so the defrag range can be | |
1227 | * written sequentially. | |
1228 | */ | |
1229 | if (i < inode->i_mapping->writeback_index) | |
1230 | inode->i_mapping->writeback_index = i; | |
1231 | ||
f7f43cc8 CM |
1232 | while (i <= last_index && defrag_count < max_to_defrag && |
1233 | (i < (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >> | |
1234 | PAGE_CACHE_SHIFT)) { | |
4cb5300b CM |
1235 | /* |
1236 | * make sure we stop running if someone unmounts | |
1237 | * the FS | |
1238 | */ | |
1239 | if (!(inode->i_sb->s_flags & MS_ACTIVE)) | |
1240 | break; | |
1241 | ||
210549eb DS |
1242 | if (btrfs_defrag_cancelled(root->fs_info)) { |
1243 | printk(KERN_DEBUG "btrfs: defrag_file cancelled\n"); | |
1244 | ret = -EAGAIN; | |
1245 | break; | |
1246 | } | |
1247 | ||
66c26892 | 1248 | if (!should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT, |
6c282eb4 | 1249 | extent_thresh, &last_len, &skip, |
a43a2111 AM |
1250 | &defrag_end, range->flags & |
1251 | BTRFS_DEFRAG_RANGE_COMPRESS)) { | |
940100a4 CM |
1252 | unsigned long next; |
1253 | /* | |
1254 | * the should_defrag function tells us how much to skip | |
1255 | * bump our counter by the suggested amount | |
1256 | */ | |
1257 | next = (skip + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; | |
1258 | i = max(i + 1, next); | |
1259 | continue; | |
1260 | } | |
008873ea LZ |
1261 | |
1262 | if (!newer_than) { | |
1263 | cluster = (PAGE_CACHE_ALIGN(defrag_end) >> | |
1264 | PAGE_CACHE_SHIFT) - i; | |
1265 | cluster = min(cluster, max_cluster); | |
1266 | } else { | |
1267 | cluster = max_cluster; | |
1268 | } | |
1269 | ||
1e701a32 | 1270 | if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS) |
1a419d85 | 1271 | BTRFS_I(inode)->force_compress = compress_type; |
940100a4 | 1272 | |
008873ea LZ |
1273 | if (i + cluster > ra_index) { |
1274 | ra_index = max(i, ra_index); | |
1275 | btrfs_force_ra(inode->i_mapping, ra, file, ra_index, | |
1276 | cluster); | |
1277 | ra_index += max_cluster; | |
1278 | } | |
940100a4 | 1279 | |
ecb8bea8 | 1280 | mutex_lock(&inode->i_mutex); |
008873ea | 1281 | ret = cluster_pages_for_defrag(inode, pages, i, cluster); |
ecb8bea8 LB |
1282 | if (ret < 0) { |
1283 | mutex_unlock(&inode->i_mutex); | |
4cb5300b | 1284 | goto out_ra; |
ecb8bea8 | 1285 | } |
4cb5300b CM |
1286 | |
1287 | defrag_count += ret; | |
d0e1d66b | 1288 | balance_dirty_pages_ratelimited(inode->i_mapping); |
ecb8bea8 | 1289 | mutex_unlock(&inode->i_mutex); |
4cb5300b CM |
1290 | |
1291 | if (newer_than) { | |
1292 | if (newer_off == (u64)-1) | |
1293 | break; | |
1294 | ||
e1f041e1 LB |
1295 | if (ret > 0) |
1296 | i += ret; | |
1297 | ||
4cb5300b CM |
1298 | newer_off = max(newer_off + 1, |
1299 | (u64)i << PAGE_CACHE_SHIFT); | |
1300 | ||
1301 | ret = find_new_extents(root, inode, | |
1302 | newer_than, &newer_off, | |
1303 | 64 * 1024); | |
1304 | if (!ret) { | |
1305 | range->start = newer_off; | |
1306 | i = (newer_off & new_align) >> PAGE_CACHE_SHIFT; | |
4cb5300b CM |
1307 | } else { |
1308 | break; | |
f46b5a66 | 1309 | } |
4cb5300b | 1310 | } else { |
008873ea | 1311 | if (ret > 0) { |
cbcc8326 | 1312 | i += ret; |
008873ea LZ |
1313 | last_len += ret << PAGE_CACHE_SHIFT; |
1314 | } else { | |
cbcc8326 | 1315 | i++; |
008873ea LZ |
1316 | last_len = 0; |
1317 | } | |
f46b5a66 | 1318 | } |
f46b5a66 CH |
1319 | } |
1320 | ||
1e701a32 CM |
1321 | if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO)) |
1322 | filemap_flush(inode->i_mapping); | |
1323 | ||
1324 | if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { | |
1325 | /* the filemap_flush will queue IO into the worker threads, but | |
1326 | * we have to make sure the IO is actually started and that | |
1327 | * ordered extents get created before we return | |
1328 | */ | |
1329 | atomic_inc(&root->fs_info->async_submit_draining); | |
1330 | while (atomic_read(&root->fs_info->nr_async_submits) || | |
1331 | atomic_read(&root->fs_info->async_delalloc_pages)) { | |
1332 | wait_event(root->fs_info->async_submit_wait, | |
1333 | (atomic_read(&root->fs_info->nr_async_submits) == 0 && | |
1334 | atomic_read(&root->fs_info->async_delalloc_pages) == 0)); | |
1335 | } | |
1336 | atomic_dec(&root->fs_info->async_submit_draining); | |
1337 | ||
1338 | mutex_lock(&inode->i_mutex); | |
261507a0 | 1339 | BTRFS_I(inode)->force_compress = BTRFS_COMPRESS_NONE; |
1e701a32 CM |
1340 | mutex_unlock(&inode->i_mutex); |
1341 | } | |
1342 | ||
1a419d85 | 1343 | if (range->compress_type == BTRFS_COMPRESS_LZO) { |
2b0ce2c2 | 1344 | btrfs_set_fs_incompat(root->fs_info, COMPRESS_LZO); |
1a419d85 LZ |
1345 | } |
1346 | ||
60ccf82f | 1347 | ret = defrag_count; |
940100a4 | 1348 | |
4cb5300b CM |
1349 | out_ra: |
1350 | if (!file) | |
1351 | kfree(ra); | |
1352 | kfree(pages); | |
940100a4 | 1353 | return ret; |
f46b5a66 CH |
1354 | } |
1355 | ||
198605a8 | 1356 | static noinline int btrfs_ioctl_resize(struct file *file, |
76dda93c | 1357 | void __user *arg) |
f46b5a66 CH |
1358 | { |
1359 | u64 new_size; | |
1360 | u64 old_size; | |
1361 | u64 devid = 1; | |
496ad9aa | 1362 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
f46b5a66 CH |
1363 | struct btrfs_ioctl_vol_args *vol_args; |
1364 | struct btrfs_trans_handle *trans; | |
1365 | struct btrfs_device *device = NULL; | |
1366 | char *sizestr; | |
1367 | char *devstr = NULL; | |
1368 | int ret = 0; | |
f46b5a66 CH |
1369 | int mod = 0; |
1370 | ||
e441d54d CM |
1371 | if (!capable(CAP_SYS_ADMIN)) |
1372 | return -EPERM; | |
1373 | ||
198605a8 MX |
1374 | ret = mnt_want_write_file(file); |
1375 | if (ret) | |
1376 | return ret; | |
1377 | ||
5ac00add SB |
1378 | if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running, |
1379 | 1)) { | |
1380 | pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n"); | |
97547676 | 1381 | mnt_drop_write_file(file); |
2c0c9da0 | 1382 | return -EINVAL; |
c9e9f97b ID |
1383 | } |
1384 | ||
5ac00add | 1385 | mutex_lock(&root->fs_info->volume_mutex); |
dae7b665 | 1386 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
1387 | if (IS_ERR(vol_args)) { |
1388 | ret = PTR_ERR(vol_args); | |
1389 | goto out; | |
1390 | } | |
5516e595 MF |
1391 | |
1392 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
f46b5a66 | 1393 | |
f46b5a66 CH |
1394 | sizestr = vol_args->name; |
1395 | devstr = strchr(sizestr, ':'); | |
1396 | if (devstr) { | |
1397 | char *end; | |
1398 | sizestr = devstr + 1; | |
1399 | *devstr = '\0'; | |
1400 | devstr = vol_args->name; | |
1401 | devid = simple_strtoull(devstr, &end, 10); | |
dfd79829 MX |
1402 | if (!devid) { |
1403 | ret = -EINVAL; | |
1404 | goto out_free; | |
1405 | } | |
5bb14682 | 1406 | printk(KERN_INFO "btrfs: resizing devid %llu\n", |
21380931 | 1407 | (unsigned long long)devid); |
f46b5a66 | 1408 | } |
dba60f3f | 1409 | |
aa1b8cd4 | 1410 | device = btrfs_find_device(root->fs_info, devid, NULL, NULL); |
f46b5a66 | 1411 | if (!device) { |
5bb14682 | 1412 | printk(KERN_INFO "btrfs: resizer unable to find device %llu\n", |
21380931 | 1413 | (unsigned long long)devid); |
dfd79829 | 1414 | ret = -ENODEV; |
c9e9f97b | 1415 | goto out_free; |
f46b5a66 | 1416 | } |
dba60f3f MX |
1417 | |
1418 | if (!device->writeable) { | |
4e42ae1b | 1419 | printk(KERN_INFO "btrfs: resizer unable to apply on " |
dba60f3f | 1420 | "readonly device %llu\n", |
a8c4a33b | 1421 | (unsigned long long)devid); |
dfd79829 | 1422 | ret = -EPERM; |
4e42ae1b LB |
1423 | goto out_free; |
1424 | } | |
1425 | ||
f46b5a66 CH |
1426 | if (!strcmp(sizestr, "max")) |
1427 | new_size = device->bdev->bd_inode->i_size; | |
1428 | else { | |
1429 | if (sizestr[0] == '-') { | |
1430 | mod = -1; | |
1431 | sizestr++; | |
1432 | } else if (sizestr[0] == '+') { | |
1433 | mod = 1; | |
1434 | sizestr++; | |
1435 | } | |
91748467 | 1436 | new_size = memparse(sizestr, NULL); |
f46b5a66 CH |
1437 | if (new_size == 0) { |
1438 | ret = -EINVAL; | |
c9e9f97b | 1439 | goto out_free; |
f46b5a66 CH |
1440 | } |
1441 | } | |
1442 | ||
63a212ab | 1443 | if (device->is_tgtdev_for_dev_replace) { |
dfd79829 | 1444 | ret = -EPERM; |
63a212ab SB |
1445 | goto out_free; |
1446 | } | |
1447 | ||
f46b5a66 CH |
1448 | old_size = device->total_bytes; |
1449 | ||
1450 | if (mod < 0) { | |
1451 | if (new_size > old_size) { | |
1452 | ret = -EINVAL; | |
c9e9f97b | 1453 | goto out_free; |
f46b5a66 CH |
1454 | } |
1455 | new_size = old_size - new_size; | |
1456 | } else if (mod > 0) { | |
1457 | new_size = old_size + new_size; | |
1458 | } | |
1459 | ||
1460 | if (new_size < 256 * 1024 * 1024) { | |
1461 | ret = -EINVAL; | |
c9e9f97b | 1462 | goto out_free; |
f46b5a66 CH |
1463 | } |
1464 | if (new_size > device->bdev->bd_inode->i_size) { | |
1465 | ret = -EFBIG; | |
c9e9f97b | 1466 | goto out_free; |
f46b5a66 CH |
1467 | } |
1468 | ||
1469 | do_div(new_size, root->sectorsize); | |
1470 | new_size *= root->sectorsize; | |
1471 | ||
606686ee JB |
1472 | printk_in_rcu(KERN_INFO "btrfs: new size for %s is %llu\n", |
1473 | rcu_str_deref(device->name), | |
1474 | (unsigned long long)new_size); | |
f46b5a66 CH |
1475 | |
1476 | if (new_size > old_size) { | |
a22285a6 | 1477 | trans = btrfs_start_transaction(root, 0); |
98d5dc13 TI |
1478 | if (IS_ERR(trans)) { |
1479 | ret = PTR_ERR(trans); | |
c9e9f97b | 1480 | goto out_free; |
98d5dc13 | 1481 | } |
f46b5a66 CH |
1482 | ret = btrfs_grow_device(trans, device, new_size); |
1483 | btrfs_commit_transaction(trans, root); | |
ece7d20e | 1484 | } else if (new_size < old_size) { |
f46b5a66 | 1485 | ret = btrfs_shrink_device(device, new_size); |
0253f40e | 1486 | } /* equal, nothing need to do */ |
f46b5a66 | 1487 | |
c9e9f97b | 1488 | out_free: |
f46b5a66 | 1489 | kfree(vol_args); |
c9e9f97b ID |
1490 | out: |
1491 | mutex_unlock(&root->fs_info->volume_mutex); | |
5ac00add | 1492 | atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0); |
18f39c41 | 1493 | mnt_drop_write_file(file); |
f46b5a66 CH |
1494 | return ret; |
1495 | } | |
1496 | ||
72fd032e | 1497 | static noinline int btrfs_ioctl_snap_create_transid(struct file *file, |
6f72c7e2 AJ |
1498 | char *name, unsigned long fd, int subvol, |
1499 | u64 *transid, bool readonly, | |
8696c533 | 1500 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 1501 | { |
f46b5a66 | 1502 | int namelen; |
3de4586c | 1503 | int ret = 0; |
f46b5a66 | 1504 | |
a874a63e LB |
1505 | ret = mnt_want_write_file(file); |
1506 | if (ret) | |
1507 | goto out; | |
1508 | ||
72fd032e SW |
1509 | namelen = strlen(name); |
1510 | if (strchr(name, '/')) { | |
f46b5a66 | 1511 | ret = -EINVAL; |
a874a63e | 1512 | goto out_drop_write; |
f46b5a66 CH |
1513 | } |
1514 | ||
16780cab CM |
1515 | if (name[0] == '.' && |
1516 | (namelen == 1 || (name[1] == '.' && namelen == 2))) { | |
1517 | ret = -EEXIST; | |
a874a63e | 1518 | goto out_drop_write; |
16780cab CM |
1519 | } |
1520 | ||
3de4586c | 1521 | if (subvol) { |
72fd032e | 1522 | ret = btrfs_mksubvol(&file->f_path, name, namelen, |
6f72c7e2 | 1523 | NULL, transid, readonly, inherit); |
cb8e7090 | 1524 | } else { |
2903ff01 | 1525 | struct fd src = fdget(fd); |
3de4586c | 1526 | struct inode *src_inode; |
2903ff01 | 1527 | if (!src.file) { |
3de4586c | 1528 | ret = -EINVAL; |
a874a63e | 1529 | goto out_drop_write; |
3de4586c CM |
1530 | } |
1531 | ||
496ad9aa AV |
1532 | src_inode = file_inode(src.file); |
1533 | if (src_inode->i_sb != file_inode(file)->i_sb) { | |
d397712b CM |
1534 | printk(KERN_INFO "btrfs: Snapshot src from " |
1535 | "another FS\n"); | |
3de4586c | 1536 | ret = -EINVAL; |
ecd18815 AV |
1537 | } else { |
1538 | ret = btrfs_mksubvol(&file->f_path, name, namelen, | |
1539 | BTRFS_I(src_inode)->root, | |
1540 | transid, readonly, inherit); | |
3de4586c | 1541 | } |
2903ff01 | 1542 | fdput(src); |
cb8e7090 | 1543 | } |
a874a63e LB |
1544 | out_drop_write: |
1545 | mnt_drop_write_file(file); | |
f46b5a66 | 1546 | out: |
72fd032e SW |
1547 | return ret; |
1548 | } | |
1549 | ||
1550 | static noinline int btrfs_ioctl_snap_create(struct file *file, | |
fa0d2b9b | 1551 | void __user *arg, int subvol) |
72fd032e | 1552 | { |
fa0d2b9b | 1553 | struct btrfs_ioctl_vol_args *vol_args; |
72fd032e SW |
1554 | int ret; |
1555 | ||
fa0d2b9b LZ |
1556 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1557 | if (IS_ERR(vol_args)) | |
1558 | return PTR_ERR(vol_args); | |
1559 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
72fd032e | 1560 | |
fa0d2b9b | 1561 | ret = btrfs_ioctl_snap_create_transid(file, vol_args->name, |
b83cc969 | 1562 | vol_args->fd, subvol, |
6f72c7e2 | 1563 | NULL, false, NULL); |
fdfb1e4f | 1564 | |
fa0d2b9b LZ |
1565 | kfree(vol_args); |
1566 | return ret; | |
1567 | } | |
fdfb1e4f | 1568 | |
fa0d2b9b LZ |
1569 | static noinline int btrfs_ioctl_snap_create_v2(struct file *file, |
1570 | void __user *arg, int subvol) | |
1571 | { | |
1572 | struct btrfs_ioctl_vol_args_v2 *vol_args; | |
1573 | int ret; | |
1574 | u64 transid = 0; | |
1575 | u64 *ptr = NULL; | |
b83cc969 | 1576 | bool readonly = false; |
6f72c7e2 | 1577 | struct btrfs_qgroup_inherit *inherit = NULL; |
75eaa0e2 | 1578 | |
fa0d2b9b LZ |
1579 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1580 | if (IS_ERR(vol_args)) | |
1581 | return PTR_ERR(vol_args); | |
1582 | vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0'; | |
75eaa0e2 | 1583 | |
b83cc969 | 1584 | if (vol_args->flags & |
6f72c7e2 AJ |
1585 | ~(BTRFS_SUBVOL_CREATE_ASYNC | BTRFS_SUBVOL_RDONLY | |
1586 | BTRFS_SUBVOL_QGROUP_INHERIT)) { | |
b83cc969 | 1587 | ret = -EOPNOTSUPP; |
fa0d2b9b | 1588 | goto out; |
72fd032e | 1589 | } |
fa0d2b9b LZ |
1590 | |
1591 | if (vol_args->flags & BTRFS_SUBVOL_CREATE_ASYNC) | |
1592 | ptr = &transid; | |
b83cc969 LZ |
1593 | if (vol_args->flags & BTRFS_SUBVOL_RDONLY) |
1594 | readonly = true; | |
6f72c7e2 AJ |
1595 | if (vol_args->flags & BTRFS_SUBVOL_QGROUP_INHERIT) { |
1596 | if (vol_args->size > PAGE_CACHE_SIZE) { | |
1597 | ret = -EINVAL; | |
1598 | goto out; | |
1599 | } | |
1600 | inherit = memdup_user(vol_args->qgroup_inherit, vol_args->size); | |
1601 | if (IS_ERR(inherit)) { | |
1602 | ret = PTR_ERR(inherit); | |
1603 | goto out; | |
1604 | } | |
1605 | } | |
fa0d2b9b LZ |
1606 | |
1607 | ret = btrfs_ioctl_snap_create_transid(file, vol_args->name, | |
6f72c7e2 | 1608 | vol_args->fd, subvol, ptr, |
8696c533 | 1609 | readonly, inherit); |
fa0d2b9b LZ |
1610 | |
1611 | if (ret == 0 && ptr && | |
1612 | copy_to_user(arg + | |
1613 | offsetof(struct btrfs_ioctl_vol_args_v2, | |
1614 | transid), ptr, sizeof(*ptr))) | |
1615 | ret = -EFAULT; | |
fdfb1e4f | 1616 | out: |
f46b5a66 | 1617 | kfree(vol_args); |
6f72c7e2 | 1618 | kfree(inherit); |
f46b5a66 CH |
1619 | return ret; |
1620 | } | |
1621 | ||
0caa102d LZ |
1622 | static noinline int btrfs_ioctl_subvol_getflags(struct file *file, |
1623 | void __user *arg) | |
1624 | { | |
496ad9aa | 1625 | struct inode *inode = file_inode(file); |
0caa102d LZ |
1626 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1627 | int ret = 0; | |
1628 | u64 flags = 0; | |
1629 | ||
33345d01 | 1630 | if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) |
0caa102d LZ |
1631 | return -EINVAL; |
1632 | ||
1633 | down_read(&root->fs_info->subvol_sem); | |
1634 | if (btrfs_root_readonly(root)) | |
1635 | flags |= BTRFS_SUBVOL_RDONLY; | |
1636 | up_read(&root->fs_info->subvol_sem); | |
1637 | ||
1638 | if (copy_to_user(arg, &flags, sizeof(flags))) | |
1639 | ret = -EFAULT; | |
1640 | ||
1641 | return ret; | |
1642 | } | |
1643 | ||
1644 | static noinline int btrfs_ioctl_subvol_setflags(struct file *file, | |
1645 | void __user *arg) | |
1646 | { | |
496ad9aa | 1647 | struct inode *inode = file_inode(file); |
0caa102d LZ |
1648 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1649 | struct btrfs_trans_handle *trans; | |
1650 | u64 root_flags; | |
1651 | u64 flags; | |
1652 | int ret = 0; | |
1653 | ||
b9ca0664 LB |
1654 | ret = mnt_want_write_file(file); |
1655 | if (ret) | |
1656 | goto out; | |
0caa102d | 1657 | |
b9ca0664 LB |
1658 | if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) { |
1659 | ret = -EINVAL; | |
1660 | goto out_drop_write; | |
1661 | } | |
0caa102d | 1662 | |
b9ca0664 LB |
1663 | if (copy_from_user(&flags, arg, sizeof(flags))) { |
1664 | ret = -EFAULT; | |
1665 | goto out_drop_write; | |
1666 | } | |
0caa102d | 1667 | |
b9ca0664 LB |
1668 | if (flags & BTRFS_SUBVOL_CREATE_ASYNC) { |
1669 | ret = -EINVAL; | |
1670 | goto out_drop_write; | |
1671 | } | |
0caa102d | 1672 | |
b9ca0664 LB |
1673 | if (flags & ~BTRFS_SUBVOL_RDONLY) { |
1674 | ret = -EOPNOTSUPP; | |
1675 | goto out_drop_write; | |
1676 | } | |
0caa102d | 1677 | |
b9ca0664 LB |
1678 | if (!inode_owner_or_capable(inode)) { |
1679 | ret = -EACCES; | |
1680 | goto out_drop_write; | |
1681 | } | |
b4dc2b8c | 1682 | |
0caa102d LZ |
1683 | down_write(&root->fs_info->subvol_sem); |
1684 | ||
1685 | /* nothing to do */ | |
1686 | if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root)) | |
b9ca0664 | 1687 | goto out_drop_sem; |
0caa102d LZ |
1688 | |
1689 | root_flags = btrfs_root_flags(&root->root_item); | |
1690 | if (flags & BTRFS_SUBVOL_RDONLY) | |
1691 | btrfs_set_root_flags(&root->root_item, | |
1692 | root_flags | BTRFS_ROOT_SUBVOL_RDONLY); | |
1693 | else | |
1694 | btrfs_set_root_flags(&root->root_item, | |
1695 | root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY); | |
1696 | ||
1697 | trans = btrfs_start_transaction(root, 1); | |
1698 | if (IS_ERR(trans)) { | |
1699 | ret = PTR_ERR(trans); | |
1700 | goto out_reset; | |
1701 | } | |
1702 | ||
b4dc2b8c | 1703 | ret = btrfs_update_root(trans, root->fs_info->tree_root, |
0caa102d LZ |
1704 | &root->root_key, &root->root_item); |
1705 | ||
1706 | btrfs_commit_transaction(trans, root); | |
1707 | out_reset: | |
1708 | if (ret) | |
1709 | btrfs_set_root_flags(&root->root_item, root_flags); | |
b9ca0664 | 1710 | out_drop_sem: |
0caa102d | 1711 | up_write(&root->fs_info->subvol_sem); |
b9ca0664 LB |
1712 | out_drop_write: |
1713 | mnt_drop_write_file(file); | |
1714 | out: | |
0caa102d LZ |
1715 | return ret; |
1716 | } | |
1717 | ||
76dda93c YZ |
1718 | /* |
1719 | * helper to check if the subvolume references other subvolumes | |
1720 | */ | |
1721 | static noinline int may_destroy_subvol(struct btrfs_root *root) | |
1722 | { | |
1723 | struct btrfs_path *path; | |
1724 | struct btrfs_key key; | |
1725 | int ret; | |
1726 | ||
1727 | path = btrfs_alloc_path(); | |
1728 | if (!path) | |
1729 | return -ENOMEM; | |
1730 | ||
1731 | key.objectid = root->root_key.objectid; | |
1732 | key.type = BTRFS_ROOT_REF_KEY; | |
1733 | key.offset = (u64)-1; | |
1734 | ||
1735 | ret = btrfs_search_slot(NULL, root->fs_info->tree_root, | |
1736 | &key, path, 0, 0); | |
1737 | if (ret < 0) | |
1738 | goto out; | |
1739 | BUG_ON(ret == 0); | |
1740 | ||
1741 | ret = 0; | |
1742 | if (path->slots[0] > 0) { | |
1743 | path->slots[0]--; | |
1744 | btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]); | |
1745 | if (key.objectid == root->root_key.objectid && | |
1746 | key.type == BTRFS_ROOT_REF_KEY) | |
1747 | ret = -ENOTEMPTY; | |
1748 | } | |
1749 | out: | |
1750 | btrfs_free_path(path); | |
1751 | return ret; | |
1752 | } | |
1753 | ||
ac8e9819 CM |
1754 | static noinline int key_in_sk(struct btrfs_key *key, |
1755 | struct btrfs_ioctl_search_key *sk) | |
1756 | { | |
abc6e134 CM |
1757 | struct btrfs_key test; |
1758 | int ret; | |
1759 | ||
1760 | test.objectid = sk->min_objectid; | |
1761 | test.type = sk->min_type; | |
1762 | test.offset = sk->min_offset; | |
1763 | ||
1764 | ret = btrfs_comp_cpu_keys(key, &test); | |
1765 | if (ret < 0) | |
ac8e9819 | 1766 | return 0; |
abc6e134 CM |
1767 | |
1768 | test.objectid = sk->max_objectid; | |
1769 | test.type = sk->max_type; | |
1770 | test.offset = sk->max_offset; | |
1771 | ||
1772 | ret = btrfs_comp_cpu_keys(key, &test); | |
1773 | if (ret > 0) | |
ac8e9819 CM |
1774 | return 0; |
1775 | return 1; | |
1776 | } | |
1777 | ||
1778 | static noinline int copy_to_sk(struct btrfs_root *root, | |
1779 | struct btrfs_path *path, | |
1780 | struct btrfs_key *key, | |
1781 | struct btrfs_ioctl_search_key *sk, | |
1782 | char *buf, | |
1783 | unsigned long *sk_offset, | |
1784 | int *num_found) | |
1785 | { | |
1786 | u64 found_transid; | |
1787 | struct extent_buffer *leaf; | |
1788 | struct btrfs_ioctl_search_header sh; | |
1789 | unsigned long item_off; | |
1790 | unsigned long item_len; | |
1791 | int nritems; | |
1792 | int i; | |
1793 | int slot; | |
ac8e9819 CM |
1794 | int ret = 0; |
1795 | ||
1796 | leaf = path->nodes[0]; | |
1797 | slot = path->slots[0]; | |
1798 | nritems = btrfs_header_nritems(leaf); | |
1799 | ||
1800 | if (btrfs_header_generation(leaf) > sk->max_transid) { | |
1801 | i = nritems; | |
1802 | goto advance_key; | |
1803 | } | |
1804 | found_transid = btrfs_header_generation(leaf); | |
1805 | ||
1806 | for (i = slot; i < nritems; i++) { | |
1807 | item_off = btrfs_item_ptr_offset(leaf, i); | |
1808 | item_len = btrfs_item_size_nr(leaf, i); | |
1809 | ||
03b71c6c GP |
1810 | btrfs_item_key_to_cpu(leaf, key, i); |
1811 | if (!key_in_sk(key, sk)) | |
1812 | continue; | |
1813 | ||
1814 | if (sizeof(sh) + item_len > BTRFS_SEARCH_ARGS_BUFSIZE) | |
ac8e9819 CM |
1815 | item_len = 0; |
1816 | ||
1817 | if (sizeof(sh) + item_len + *sk_offset > | |
1818 | BTRFS_SEARCH_ARGS_BUFSIZE) { | |
1819 | ret = 1; | |
1820 | goto overflow; | |
1821 | } | |
1822 | ||
ac8e9819 CM |
1823 | sh.objectid = key->objectid; |
1824 | sh.offset = key->offset; | |
1825 | sh.type = key->type; | |
1826 | sh.len = item_len; | |
1827 | sh.transid = found_transid; | |
1828 | ||
1829 | /* copy search result header */ | |
1830 | memcpy(buf + *sk_offset, &sh, sizeof(sh)); | |
1831 | *sk_offset += sizeof(sh); | |
1832 | ||
1833 | if (item_len) { | |
1834 | char *p = buf + *sk_offset; | |
1835 | /* copy the item */ | |
1836 | read_extent_buffer(leaf, p, | |
1837 | item_off, item_len); | |
1838 | *sk_offset += item_len; | |
ac8e9819 | 1839 | } |
e2156867 | 1840 | (*num_found)++; |
ac8e9819 CM |
1841 | |
1842 | if (*num_found >= sk->nr_items) | |
1843 | break; | |
1844 | } | |
1845 | advance_key: | |
abc6e134 CM |
1846 | ret = 0; |
1847 | if (key->offset < (u64)-1 && key->offset < sk->max_offset) | |
ac8e9819 | 1848 | key->offset++; |
abc6e134 CM |
1849 | else if (key->type < (u8)-1 && key->type < sk->max_type) { |
1850 | key->offset = 0; | |
ac8e9819 | 1851 | key->type++; |
abc6e134 CM |
1852 | } else if (key->objectid < (u64)-1 && key->objectid < sk->max_objectid) { |
1853 | key->offset = 0; | |
1854 | key->type = 0; | |
ac8e9819 | 1855 | key->objectid++; |
abc6e134 CM |
1856 | } else |
1857 | ret = 1; | |
ac8e9819 | 1858 | overflow: |
ac8e9819 CM |
1859 | return ret; |
1860 | } | |
1861 | ||
1862 | static noinline int search_ioctl(struct inode *inode, | |
1863 | struct btrfs_ioctl_search_args *args) | |
1864 | { | |
1865 | struct btrfs_root *root; | |
1866 | struct btrfs_key key; | |
1867 | struct btrfs_key max_key; | |
1868 | struct btrfs_path *path; | |
1869 | struct btrfs_ioctl_search_key *sk = &args->key; | |
1870 | struct btrfs_fs_info *info = BTRFS_I(inode)->root->fs_info; | |
1871 | int ret; | |
1872 | int num_found = 0; | |
1873 | unsigned long sk_offset = 0; | |
1874 | ||
1875 | path = btrfs_alloc_path(); | |
1876 | if (!path) | |
1877 | return -ENOMEM; | |
1878 | ||
1879 | if (sk->tree_id == 0) { | |
1880 | /* search the root of the inode that was passed */ | |
1881 | root = BTRFS_I(inode)->root; | |
1882 | } else { | |
1883 | key.objectid = sk->tree_id; | |
1884 | key.type = BTRFS_ROOT_ITEM_KEY; | |
1885 | key.offset = (u64)-1; | |
1886 | root = btrfs_read_fs_root_no_name(info, &key); | |
1887 | if (IS_ERR(root)) { | |
1888 | printk(KERN_ERR "could not find root %llu\n", | |
1889 | sk->tree_id); | |
1890 | btrfs_free_path(path); | |
1891 | return -ENOENT; | |
1892 | } | |
1893 | } | |
1894 | ||
1895 | key.objectid = sk->min_objectid; | |
1896 | key.type = sk->min_type; | |
1897 | key.offset = sk->min_offset; | |
1898 | ||
1899 | max_key.objectid = sk->max_objectid; | |
1900 | max_key.type = sk->max_type; | |
1901 | max_key.offset = sk->max_offset; | |
1902 | ||
1903 | path->keep_locks = 1; | |
1904 | ||
1905 | while(1) { | |
de78b51a | 1906 | ret = btrfs_search_forward(root, &key, &max_key, path, |
ac8e9819 CM |
1907 | sk->min_transid); |
1908 | if (ret != 0) { | |
1909 | if (ret > 0) | |
1910 | ret = 0; | |
1911 | goto err; | |
1912 | } | |
1913 | ret = copy_to_sk(root, path, &key, sk, args->buf, | |
1914 | &sk_offset, &num_found); | |
b3b4aa74 | 1915 | btrfs_release_path(path); |
ac8e9819 CM |
1916 | if (ret || num_found >= sk->nr_items) |
1917 | break; | |
1918 | ||
1919 | } | |
1920 | ret = 0; | |
1921 | err: | |
1922 | sk->nr_items = num_found; | |
1923 | btrfs_free_path(path); | |
1924 | return ret; | |
1925 | } | |
1926 | ||
1927 | static noinline int btrfs_ioctl_tree_search(struct file *file, | |
1928 | void __user *argp) | |
1929 | { | |
1930 | struct btrfs_ioctl_search_args *args; | |
1931 | struct inode *inode; | |
1932 | int ret; | |
1933 | ||
1934 | if (!capable(CAP_SYS_ADMIN)) | |
1935 | return -EPERM; | |
1936 | ||
2354d08f JL |
1937 | args = memdup_user(argp, sizeof(*args)); |
1938 | if (IS_ERR(args)) | |
1939 | return PTR_ERR(args); | |
ac8e9819 | 1940 | |
496ad9aa | 1941 | inode = file_inode(file); |
ac8e9819 CM |
1942 | ret = search_ioctl(inode, args); |
1943 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) | |
1944 | ret = -EFAULT; | |
1945 | kfree(args); | |
1946 | return ret; | |
1947 | } | |
1948 | ||
98d377a0 | 1949 | /* |
ac8e9819 CM |
1950 | * Search INODE_REFs to identify path name of 'dirid' directory |
1951 | * in a 'tree_id' tree. and sets path name to 'name'. | |
1952 | */ | |
98d377a0 TH |
1953 | static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, |
1954 | u64 tree_id, u64 dirid, char *name) | |
1955 | { | |
1956 | struct btrfs_root *root; | |
1957 | struct btrfs_key key; | |
ac8e9819 | 1958 | char *ptr; |
98d377a0 TH |
1959 | int ret = -1; |
1960 | int slot; | |
1961 | int len; | |
1962 | int total_len = 0; | |
1963 | struct btrfs_inode_ref *iref; | |
1964 | struct extent_buffer *l; | |
1965 | struct btrfs_path *path; | |
1966 | ||
1967 | if (dirid == BTRFS_FIRST_FREE_OBJECTID) { | |
1968 | name[0]='\0'; | |
1969 | return 0; | |
1970 | } | |
1971 | ||
1972 | path = btrfs_alloc_path(); | |
1973 | if (!path) | |
1974 | return -ENOMEM; | |
1975 | ||
ac8e9819 | 1976 | ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX]; |
98d377a0 TH |
1977 | |
1978 | key.objectid = tree_id; | |
1979 | key.type = BTRFS_ROOT_ITEM_KEY; | |
1980 | key.offset = (u64)-1; | |
1981 | root = btrfs_read_fs_root_no_name(info, &key); | |
1982 | if (IS_ERR(root)) { | |
1983 | printk(KERN_ERR "could not find root %llu\n", tree_id); | |
8ad6fcab CM |
1984 | ret = -ENOENT; |
1985 | goto out; | |
98d377a0 TH |
1986 | } |
1987 | ||
1988 | key.objectid = dirid; | |
1989 | key.type = BTRFS_INODE_REF_KEY; | |
8ad6fcab | 1990 | key.offset = (u64)-1; |
98d377a0 TH |
1991 | |
1992 | while(1) { | |
1993 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
1994 | if (ret < 0) | |
1995 | goto out; | |
1996 | ||
1997 | l = path->nodes[0]; | |
1998 | slot = path->slots[0]; | |
8ad6fcab CM |
1999 | if (ret > 0 && slot > 0) |
2000 | slot--; | |
98d377a0 TH |
2001 | btrfs_item_key_to_cpu(l, &key, slot); |
2002 | ||
2003 | if (ret > 0 && (key.objectid != dirid || | |
ac8e9819 CM |
2004 | key.type != BTRFS_INODE_REF_KEY)) { |
2005 | ret = -ENOENT; | |
98d377a0 | 2006 | goto out; |
ac8e9819 | 2007 | } |
98d377a0 TH |
2008 | |
2009 | iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref); | |
2010 | len = btrfs_inode_ref_name_len(l, iref); | |
2011 | ptr -= len + 1; | |
2012 | total_len += len + 1; | |
ac8e9819 | 2013 | if (ptr < name) |
98d377a0 TH |
2014 | goto out; |
2015 | ||
2016 | *(ptr + len) = '/'; | |
2017 | read_extent_buffer(l, ptr,(unsigned long)(iref + 1), len); | |
2018 | ||
2019 | if (key.offset == BTRFS_FIRST_FREE_OBJECTID) | |
2020 | break; | |
2021 | ||
b3b4aa74 | 2022 | btrfs_release_path(path); |
98d377a0 | 2023 | key.objectid = key.offset; |
8ad6fcab | 2024 | key.offset = (u64)-1; |
98d377a0 | 2025 | dirid = key.objectid; |
98d377a0 | 2026 | } |
ac8e9819 | 2027 | if (ptr < name) |
98d377a0 | 2028 | goto out; |
77906a50 | 2029 | memmove(name, ptr, total_len); |
98d377a0 TH |
2030 | name[total_len]='\0'; |
2031 | ret = 0; | |
2032 | out: | |
2033 | btrfs_free_path(path); | |
ac8e9819 CM |
2034 | return ret; |
2035 | } | |
2036 | ||
2037 | static noinline int btrfs_ioctl_ino_lookup(struct file *file, | |
2038 | void __user *argp) | |
2039 | { | |
2040 | struct btrfs_ioctl_ino_lookup_args *args; | |
2041 | struct inode *inode; | |
2042 | int ret; | |
2043 | ||
2044 | if (!capable(CAP_SYS_ADMIN)) | |
2045 | return -EPERM; | |
2046 | ||
2354d08f JL |
2047 | args = memdup_user(argp, sizeof(*args)); |
2048 | if (IS_ERR(args)) | |
2049 | return PTR_ERR(args); | |
c2b96929 | 2050 | |
496ad9aa | 2051 | inode = file_inode(file); |
ac8e9819 | 2052 | |
1b53ac4d CM |
2053 | if (args->treeid == 0) |
2054 | args->treeid = BTRFS_I(inode)->root->root_key.objectid; | |
2055 | ||
ac8e9819 CM |
2056 | ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info, |
2057 | args->treeid, args->objectid, | |
2058 | args->name); | |
2059 | ||
2060 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) | |
2061 | ret = -EFAULT; | |
2062 | ||
2063 | kfree(args); | |
98d377a0 TH |
2064 | return ret; |
2065 | } | |
2066 | ||
76dda93c YZ |
2067 | static noinline int btrfs_ioctl_snap_destroy(struct file *file, |
2068 | void __user *arg) | |
2069 | { | |
2070 | struct dentry *parent = fdentry(file); | |
2071 | struct dentry *dentry; | |
2072 | struct inode *dir = parent->d_inode; | |
2073 | struct inode *inode; | |
2074 | struct btrfs_root *root = BTRFS_I(dir)->root; | |
2075 | struct btrfs_root *dest = NULL; | |
2076 | struct btrfs_ioctl_vol_args *vol_args; | |
2077 | struct btrfs_trans_handle *trans; | |
c58aaad2 MX |
2078 | struct btrfs_block_rsv block_rsv; |
2079 | u64 qgroup_reserved; | |
76dda93c YZ |
2080 | int namelen; |
2081 | int ret; | |
2082 | int err = 0; | |
2083 | ||
76dda93c YZ |
2084 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
2085 | if (IS_ERR(vol_args)) | |
2086 | return PTR_ERR(vol_args); | |
2087 | ||
2088 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
2089 | namelen = strlen(vol_args->name); | |
2090 | if (strchr(vol_args->name, '/') || | |
2091 | strncmp(vol_args->name, "..", namelen) == 0) { | |
2092 | err = -EINVAL; | |
2093 | goto out; | |
2094 | } | |
2095 | ||
a561be71 | 2096 | err = mnt_want_write_file(file); |
76dda93c YZ |
2097 | if (err) |
2098 | goto out; | |
2099 | ||
5c50c9b8 DS |
2100 | err = mutex_lock_killable_nested(&dir->i_mutex, I_MUTEX_PARENT); |
2101 | if (err == -EINTR) | |
2102 | goto out; | |
76dda93c YZ |
2103 | dentry = lookup_one_len(vol_args->name, parent, namelen); |
2104 | if (IS_ERR(dentry)) { | |
2105 | err = PTR_ERR(dentry); | |
2106 | goto out_unlock_dir; | |
2107 | } | |
2108 | ||
2109 | if (!dentry->d_inode) { | |
2110 | err = -ENOENT; | |
2111 | goto out_dput; | |
2112 | } | |
2113 | ||
2114 | inode = dentry->d_inode; | |
4260f7c7 SW |
2115 | dest = BTRFS_I(inode)->root; |
2116 | if (!capable(CAP_SYS_ADMIN)){ | |
2117 | /* | |
2118 | * Regular user. Only allow this with a special mount | |
2119 | * option, when the user has write+exec access to the | |
2120 | * subvol root, and when rmdir(2) would have been | |
2121 | * allowed. | |
2122 | * | |
2123 | * Note that this is _not_ check that the subvol is | |
2124 | * empty or doesn't contain data that we wouldn't | |
2125 | * otherwise be able to delete. | |
2126 | * | |
2127 | * Users who want to delete empty subvols should try | |
2128 | * rmdir(2). | |
2129 | */ | |
2130 | err = -EPERM; | |
2131 | if (!btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED)) | |
2132 | goto out_dput; | |
2133 | ||
2134 | /* | |
2135 | * Do not allow deletion if the parent dir is the same | |
2136 | * as the dir to be deleted. That means the ioctl | |
2137 | * must be called on the dentry referencing the root | |
2138 | * of the subvol, not a random directory contained | |
2139 | * within it. | |
2140 | */ | |
2141 | err = -EINVAL; | |
2142 | if (root == dest) | |
2143 | goto out_dput; | |
2144 | ||
2145 | err = inode_permission(inode, MAY_WRITE | MAY_EXEC); | |
2146 | if (err) | |
2147 | goto out_dput; | |
4260f7c7 SW |
2148 | } |
2149 | ||
5c39da5b MX |
2150 | /* check if subvolume may be deleted by a user */ |
2151 | err = btrfs_may_delete(dir, dentry, 1); | |
2152 | if (err) | |
2153 | goto out_dput; | |
2154 | ||
33345d01 | 2155 | if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) { |
76dda93c YZ |
2156 | err = -EINVAL; |
2157 | goto out_dput; | |
2158 | } | |
2159 | ||
76dda93c YZ |
2160 | mutex_lock(&inode->i_mutex); |
2161 | err = d_invalidate(dentry); | |
2162 | if (err) | |
2163 | goto out_unlock; | |
2164 | ||
2165 | down_write(&root->fs_info->subvol_sem); | |
2166 | ||
2167 | err = may_destroy_subvol(dest); | |
2168 | if (err) | |
2169 | goto out_up_write; | |
2170 | ||
c58aaad2 MX |
2171 | btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP); |
2172 | /* | |
2173 | * One for dir inode, two for dir entries, two for root | |
2174 | * ref/backref. | |
2175 | */ | |
2176 | err = btrfs_subvolume_reserve_metadata(root, &block_rsv, | |
2177 | 5, &qgroup_reserved); | |
2178 | if (err) | |
2179 | goto out_up_write; | |
2180 | ||
a22285a6 YZ |
2181 | trans = btrfs_start_transaction(root, 0); |
2182 | if (IS_ERR(trans)) { | |
2183 | err = PTR_ERR(trans); | |
c58aaad2 | 2184 | goto out_release; |
a22285a6 | 2185 | } |
c58aaad2 MX |
2186 | trans->block_rsv = &block_rsv; |
2187 | trans->bytes_reserved = block_rsv.size; | |
a22285a6 | 2188 | |
76dda93c YZ |
2189 | ret = btrfs_unlink_subvol(trans, root, dir, |
2190 | dest->root_key.objectid, | |
2191 | dentry->d_name.name, | |
2192 | dentry->d_name.len); | |
79787eaa JM |
2193 | if (ret) { |
2194 | err = ret; | |
2195 | btrfs_abort_transaction(trans, root, ret); | |
2196 | goto out_end_trans; | |
2197 | } | |
76dda93c YZ |
2198 | |
2199 | btrfs_record_root_in_trans(trans, dest); | |
2200 | ||
2201 | memset(&dest->root_item.drop_progress, 0, | |
2202 | sizeof(dest->root_item.drop_progress)); | |
2203 | dest->root_item.drop_level = 0; | |
2204 | btrfs_set_root_refs(&dest->root_item, 0); | |
2205 | ||
d68fc57b YZ |
2206 | if (!xchg(&dest->orphan_item_inserted, 1)) { |
2207 | ret = btrfs_insert_orphan_item(trans, | |
2208 | root->fs_info->tree_root, | |
2209 | dest->root_key.objectid); | |
79787eaa JM |
2210 | if (ret) { |
2211 | btrfs_abort_transaction(trans, root, ret); | |
2212 | err = ret; | |
2213 | goto out_end_trans; | |
2214 | } | |
d68fc57b | 2215 | } |
79787eaa | 2216 | out_end_trans: |
c58aaad2 MX |
2217 | trans->block_rsv = NULL; |
2218 | trans->bytes_reserved = 0; | |
531cb13f | 2219 | ret = btrfs_end_transaction(trans, root); |
79787eaa JM |
2220 | if (ret && !err) |
2221 | err = ret; | |
76dda93c | 2222 | inode->i_flags |= S_DEAD; |
c58aaad2 MX |
2223 | out_release: |
2224 | btrfs_subvolume_release_metadata(root, &block_rsv, qgroup_reserved); | |
76dda93c YZ |
2225 | out_up_write: |
2226 | up_write(&root->fs_info->subvol_sem); | |
2227 | out_unlock: | |
2228 | mutex_unlock(&inode->i_mutex); | |
2229 | if (!err) { | |
efefb143 | 2230 | shrink_dcache_sb(root->fs_info->sb); |
76dda93c YZ |
2231 | btrfs_invalidate_inodes(dest); |
2232 | d_delete(dentry); | |
fa6ac876 LB |
2233 | |
2234 | /* the last ref */ | |
2235 | if (dest->cache_inode) { | |
2236 | iput(dest->cache_inode); | |
2237 | dest->cache_inode = NULL; | |
2238 | } | |
76dda93c YZ |
2239 | } |
2240 | out_dput: | |
2241 | dput(dentry); | |
2242 | out_unlock_dir: | |
2243 | mutex_unlock(&dir->i_mutex); | |
2a79f17e | 2244 | mnt_drop_write_file(file); |
76dda93c YZ |
2245 | out: |
2246 | kfree(vol_args); | |
2247 | return err; | |
2248 | } | |
2249 | ||
1e701a32 | 2250 | static int btrfs_ioctl_defrag(struct file *file, void __user *argp) |
f46b5a66 | 2251 | { |
496ad9aa | 2252 | struct inode *inode = file_inode(file); |
f46b5a66 | 2253 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1e701a32 | 2254 | struct btrfs_ioctl_defrag_range_args *range; |
c146afad YZ |
2255 | int ret; |
2256 | ||
25122d15 ID |
2257 | ret = mnt_want_write_file(file); |
2258 | if (ret) | |
2259 | return ret; | |
b83cc969 | 2260 | |
25122d15 ID |
2261 | if (btrfs_root_readonly(root)) { |
2262 | ret = -EROFS; | |
2263 | goto out; | |
5ac00add | 2264 | } |
f46b5a66 CH |
2265 | |
2266 | switch (inode->i_mode & S_IFMT) { | |
2267 | case S_IFDIR: | |
e441d54d CM |
2268 | if (!capable(CAP_SYS_ADMIN)) { |
2269 | ret = -EPERM; | |
2270 | goto out; | |
2271 | } | |
de78b51a | 2272 | ret = btrfs_defrag_root(root); |
8929ecfa YZ |
2273 | if (ret) |
2274 | goto out; | |
de78b51a | 2275 | ret = btrfs_defrag_root(root->fs_info->extent_root); |
f46b5a66 CH |
2276 | break; |
2277 | case S_IFREG: | |
e441d54d CM |
2278 | if (!(file->f_mode & FMODE_WRITE)) { |
2279 | ret = -EINVAL; | |
2280 | goto out; | |
2281 | } | |
1e701a32 CM |
2282 | |
2283 | range = kzalloc(sizeof(*range), GFP_KERNEL); | |
2284 | if (!range) { | |
2285 | ret = -ENOMEM; | |
2286 | goto out; | |
2287 | } | |
2288 | ||
2289 | if (argp) { | |
2290 | if (copy_from_user(range, argp, | |
2291 | sizeof(*range))) { | |
2292 | ret = -EFAULT; | |
2293 | kfree(range); | |
683be16e | 2294 | goto out; |
1e701a32 CM |
2295 | } |
2296 | /* compression requires us to start the IO */ | |
2297 | if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { | |
2298 | range->flags |= BTRFS_DEFRAG_RANGE_START_IO; | |
2299 | range->extent_thresh = (u32)-1; | |
2300 | } | |
2301 | } else { | |
2302 | /* the rest are all set to zero by kzalloc */ | |
2303 | range->len = (u64)-1; | |
2304 | } | |
496ad9aa | 2305 | ret = btrfs_defrag_file(file_inode(file), file, |
4cb5300b CM |
2306 | range, 0, 0); |
2307 | if (ret > 0) | |
2308 | ret = 0; | |
1e701a32 | 2309 | kfree(range); |
f46b5a66 | 2310 | break; |
8929ecfa YZ |
2311 | default: |
2312 | ret = -EINVAL; | |
f46b5a66 | 2313 | } |
e441d54d | 2314 | out: |
25122d15 | 2315 | mnt_drop_write_file(file); |
e441d54d | 2316 | return ret; |
f46b5a66 CH |
2317 | } |
2318 | ||
b2950863 | 2319 | static long btrfs_ioctl_add_dev(struct btrfs_root *root, void __user *arg) |
f46b5a66 CH |
2320 | { |
2321 | struct btrfs_ioctl_vol_args *vol_args; | |
2322 | int ret; | |
2323 | ||
e441d54d CM |
2324 | if (!capable(CAP_SYS_ADMIN)) |
2325 | return -EPERM; | |
2326 | ||
5ac00add SB |
2327 | if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running, |
2328 | 1)) { | |
2329 | pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n"); | |
2c0c9da0 | 2330 | return -EINVAL; |
c9e9f97b ID |
2331 | } |
2332 | ||
5ac00add | 2333 | mutex_lock(&root->fs_info->volume_mutex); |
dae7b665 | 2334 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
2335 | if (IS_ERR(vol_args)) { |
2336 | ret = PTR_ERR(vol_args); | |
2337 | goto out; | |
2338 | } | |
f46b5a66 | 2339 | |
5516e595 | 2340 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
f46b5a66 CH |
2341 | ret = btrfs_init_new_device(root, vol_args->name); |
2342 | ||
f46b5a66 | 2343 | kfree(vol_args); |
c9e9f97b ID |
2344 | out: |
2345 | mutex_unlock(&root->fs_info->volume_mutex); | |
5ac00add | 2346 | atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0); |
f46b5a66 CH |
2347 | return ret; |
2348 | } | |
2349 | ||
da24927b | 2350 | static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg) |
f46b5a66 | 2351 | { |
496ad9aa | 2352 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
f46b5a66 CH |
2353 | struct btrfs_ioctl_vol_args *vol_args; |
2354 | int ret; | |
2355 | ||
e441d54d CM |
2356 | if (!capable(CAP_SYS_ADMIN)) |
2357 | return -EPERM; | |
2358 | ||
da24927b MX |
2359 | ret = mnt_want_write_file(file); |
2360 | if (ret) | |
2361 | return ret; | |
c146afad | 2362 | |
dae7b665 | 2363 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
2364 | if (IS_ERR(vol_args)) { |
2365 | ret = PTR_ERR(vol_args); | |
2366 | goto out; | |
2367 | } | |
f46b5a66 | 2368 | |
5516e595 | 2369 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
f46b5a66 | 2370 | |
183860f6 AJ |
2371 | if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running, |
2372 | 1)) { | |
2373 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; | |
2374 | goto out; | |
2375 | } | |
2376 | ||
2377 | mutex_lock(&root->fs_info->volume_mutex); | |
2378 | ret = btrfs_rm_device(root, vol_args->name); | |
c9e9f97b | 2379 | mutex_unlock(&root->fs_info->volume_mutex); |
5ac00add | 2380 | atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0); |
183860f6 AJ |
2381 | |
2382 | out: | |
2383 | kfree(vol_args); | |
4ac20c70 | 2384 | mnt_drop_write_file(file); |
f46b5a66 CH |
2385 | return ret; |
2386 | } | |
2387 | ||
475f6387 JS |
2388 | static long btrfs_ioctl_fs_info(struct btrfs_root *root, void __user *arg) |
2389 | { | |
027ed2f0 | 2390 | struct btrfs_ioctl_fs_info_args *fi_args; |
475f6387 JS |
2391 | struct btrfs_device *device; |
2392 | struct btrfs_device *next; | |
2393 | struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices; | |
027ed2f0 | 2394 | int ret = 0; |
475f6387 JS |
2395 | |
2396 | if (!capable(CAP_SYS_ADMIN)) | |
2397 | return -EPERM; | |
2398 | ||
027ed2f0 LZ |
2399 | fi_args = kzalloc(sizeof(*fi_args), GFP_KERNEL); |
2400 | if (!fi_args) | |
2401 | return -ENOMEM; | |
2402 | ||
2403 | fi_args->num_devices = fs_devices->num_devices; | |
2404 | memcpy(&fi_args->fsid, root->fs_info->fsid, sizeof(fi_args->fsid)); | |
475f6387 JS |
2405 | |
2406 | mutex_lock(&fs_devices->device_list_mutex); | |
2407 | list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) { | |
027ed2f0 LZ |
2408 | if (device->devid > fi_args->max_id) |
2409 | fi_args->max_id = device->devid; | |
475f6387 JS |
2410 | } |
2411 | mutex_unlock(&fs_devices->device_list_mutex); | |
2412 | ||
027ed2f0 LZ |
2413 | if (copy_to_user(arg, fi_args, sizeof(*fi_args))) |
2414 | ret = -EFAULT; | |
475f6387 | 2415 | |
027ed2f0 LZ |
2416 | kfree(fi_args); |
2417 | return ret; | |
475f6387 JS |
2418 | } |
2419 | ||
2420 | static long btrfs_ioctl_dev_info(struct btrfs_root *root, void __user *arg) | |
2421 | { | |
2422 | struct btrfs_ioctl_dev_info_args *di_args; | |
2423 | struct btrfs_device *dev; | |
2424 | struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices; | |
2425 | int ret = 0; | |
2426 | char *s_uuid = NULL; | |
2427 | char empty_uuid[BTRFS_UUID_SIZE] = {0}; | |
2428 | ||
2429 | if (!capable(CAP_SYS_ADMIN)) | |
2430 | return -EPERM; | |
2431 | ||
2432 | di_args = memdup_user(arg, sizeof(*di_args)); | |
2433 | if (IS_ERR(di_args)) | |
2434 | return PTR_ERR(di_args); | |
2435 | ||
2436 | if (memcmp(empty_uuid, di_args->uuid, BTRFS_UUID_SIZE) != 0) | |
2437 | s_uuid = di_args->uuid; | |
2438 | ||
2439 | mutex_lock(&fs_devices->device_list_mutex); | |
aa1b8cd4 | 2440 | dev = btrfs_find_device(root->fs_info, di_args->devid, s_uuid, NULL); |
475f6387 JS |
2441 | |
2442 | if (!dev) { | |
2443 | ret = -ENODEV; | |
2444 | goto out; | |
2445 | } | |
2446 | ||
2447 | di_args->devid = dev->devid; | |
2448 | di_args->bytes_used = dev->bytes_used; | |
2449 | di_args->total_bytes = dev->total_bytes; | |
2450 | memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid)); | |
a27202fb | 2451 | if (dev->name) { |
606686ee JB |
2452 | struct rcu_string *name; |
2453 | ||
2454 | rcu_read_lock(); | |
2455 | name = rcu_dereference(dev->name); | |
2456 | strncpy(di_args->path, name->str, sizeof(di_args->path)); | |
2457 | rcu_read_unlock(); | |
a27202fb JM |
2458 | di_args->path[sizeof(di_args->path) - 1] = 0; |
2459 | } else { | |
99ba55ad | 2460 | di_args->path[0] = '\0'; |
a27202fb | 2461 | } |
475f6387 JS |
2462 | |
2463 | out: | |
55793c0d | 2464 | mutex_unlock(&fs_devices->device_list_mutex); |
475f6387 JS |
2465 | if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args))) |
2466 | ret = -EFAULT; | |
2467 | ||
2468 | kfree(di_args); | |
2469 | return ret; | |
2470 | } | |
2471 | ||
76dda93c YZ |
2472 | static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd, |
2473 | u64 off, u64 olen, u64 destoff) | |
f46b5a66 | 2474 | { |
496ad9aa | 2475 | struct inode *inode = file_inode(file); |
f46b5a66 | 2476 | struct btrfs_root *root = BTRFS_I(inode)->root; |
2903ff01 | 2477 | struct fd src_file; |
f46b5a66 CH |
2478 | struct inode *src; |
2479 | struct btrfs_trans_handle *trans; | |
f46b5a66 | 2480 | struct btrfs_path *path; |
f46b5a66 | 2481 | struct extent_buffer *leaf; |
ae01a0ab YZ |
2482 | char *buf; |
2483 | struct btrfs_key key; | |
f46b5a66 CH |
2484 | u32 nritems; |
2485 | int slot; | |
ae01a0ab | 2486 | int ret; |
c5c9cd4d SW |
2487 | u64 len = olen; |
2488 | u64 bs = root->fs_info->sb->s_blocksize; | |
a96fbc72 | 2489 | int same_inode = 0; |
d20f7043 | 2490 | |
c5c9cd4d SW |
2491 | /* |
2492 | * TODO: | |
2493 | * - split compressed inline extents. annoying: we need to | |
2494 | * decompress into destination's address_space (the file offset | |
2495 | * may change, so source mapping won't do), then recompress (or | |
2496 | * otherwise reinsert) a subrange. | |
2497 | * - allow ranges within the same file to be cloned (provided | |
2498 | * they don't overlap)? | |
2499 | */ | |
2500 | ||
e441d54d | 2501 | /* the destination must be opened for writing */ |
2ebc3464 | 2502 | if (!(file->f_mode & FMODE_WRITE) || (file->f_flags & O_APPEND)) |
e441d54d CM |
2503 | return -EINVAL; |
2504 | ||
b83cc969 LZ |
2505 | if (btrfs_root_readonly(root)) |
2506 | return -EROFS; | |
2507 | ||
a561be71 | 2508 | ret = mnt_want_write_file(file); |
c146afad YZ |
2509 | if (ret) |
2510 | return ret; | |
2511 | ||
2903ff01 AV |
2512 | src_file = fdget(srcfd); |
2513 | if (!src_file.file) { | |
ab67b7c1 YZ |
2514 | ret = -EBADF; |
2515 | goto out_drop_write; | |
2516 | } | |
5dc64164 | 2517 | |
362a20c5 | 2518 | ret = -EXDEV; |
2903ff01 | 2519 | if (src_file.file->f_path.mnt != file->f_path.mnt) |
362a20c5 DS |
2520 | goto out_fput; |
2521 | ||
496ad9aa | 2522 | src = file_inode(src_file.file); |
f46b5a66 | 2523 | |
c5c9cd4d SW |
2524 | ret = -EINVAL; |
2525 | if (src == inode) | |
a96fbc72 | 2526 | same_inode = 1; |
c5c9cd4d | 2527 | |
5dc64164 | 2528 | /* the src must be open for reading */ |
2903ff01 | 2529 | if (!(src_file.file->f_mode & FMODE_READ)) |
5dc64164 DR |
2530 | goto out_fput; |
2531 | ||
0e7b824c LZ |
2532 | /* don't make the dst file partly checksummed */ |
2533 | if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) != | |
2534 | (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) | |
2535 | goto out_fput; | |
2536 | ||
ae01a0ab YZ |
2537 | ret = -EISDIR; |
2538 | if (S_ISDIR(src->i_mode) || S_ISDIR(inode->i_mode)) | |
2539 | goto out_fput; | |
2540 | ||
f46b5a66 | 2541 | ret = -EXDEV; |
362a20c5 | 2542 | if (src->i_sb != inode->i_sb) |
ae01a0ab YZ |
2543 | goto out_fput; |
2544 | ||
2545 | ret = -ENOMEM; | |
2546 | buf = vmalloc(btrfs_level_size(root, 0)); | |
2547 | if (!buf) | |
2548 | goto out_fput; | |
2549 | ||
2550 | path = btrfs_alloc_path(); | |
2551 | if (!path) { | |
2552 | vfree(buf); | |
f46b5a66 | 2553 | goto out_fput; |
ae01a0ab YZ |
2554 | } |
2555 | path->reada = 2; | |
f46b5a66 | 2556 | |
a96fbc72 LB |
2557 | if (!same_inode) { |
2558 | if (inode < src) { | |
2559 | mutex_lock_nested(&inode->i_mutex, I_MUTEX_PARENT); | |
2560 | mutex_lock_nested(&src->i_mutex, I_MUTEX_CHILD); | |
2561 | } else { | |
2562 | mutex_lock_nested(&src->i_mutex, I_MUTEX_PARENT); | |
2563 | mutex_lock_nested(&inode->i_mutex, I_MUTEX_CHILD); | |
2564 | } | |
f46b5a66 | 2565 | } else { |
a96fbc72 | 2566 | mutex_lock(&src->i_mutex); |
f46b5a66 CH |
2567 | } |
2568 | ||
c5c9cd4d SW |
2569 | /* determine range to clone */ |
2570 | ret = -EINVAL; | |
2ebc3464 | 2571 | if (off + len > src->i_size || off + len < off) |
f46b5a66 | 2572 | goto out_unlock; |
c5c9cd4d SW |
2573 | if (len == 0) |
2574 | olen = len = src->i_size - off; | |
2575 | /* if we extend to eof, continue to block boundary */ | |
2576 | if (off + len == src->i_size) | |
2a6b8dae | 2577 | len = ALIGN(src->i_size, bs) - off; |
c5c9cd4d SW |
2578 | |
2579 | /* verify the end result is block aligned */ | |
2a6b8dae LZ |
2580 | if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs) || |
2581 | !IS_ALIGNED(destoff, bs)) | |
c5c9cd4d SW |
2582 | goto out_unlock; |
2583 | ||
a96fbc72 LB |
2584 | /* verify if ranges are overlapped within the same file */ |
2585 | if (same_inode) { | |
2586 | if (destoff + len > off && destoff < off + len) | |
2587 | goto out_unlock; | |
2588 | } | |
2589 | ||
d525e8ab LZ |
2590 | if (destoff > inode->i_size) { |
2591 | ret = btrfs_cont_expand(inode, inode->i_size, destoff); | |
2592 | if (ret) | |
2593 | goto out_unlock; | |
2594 | } | |
2595 | ||
71ef0786 LZ |
2596 | /* truncate page cache pages from target inode range */ |
2597 | truncate_inode_pages_range(&inode->i_data, destoff, | |
2598 | PAGE_CACHE_ALIGN(destoff + len) - 1); | |
2599 | ||
f46b5a66 CH |
2600 | /* do any pending delalloc/csum calc on src, one way or |
2601 | another, and lock file content */ | |
2602 | while (1) { | |
31840ae1 | 2603 | struct btrfs_ordered_extent *ordered; |
aa42ffd9 LB |
2604 | lock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1); |
2605 | ordered = btrfs_lookup_first_ordered_extent(src, off + len - 1); | |
9a019196 | 2606 | if (!ordered && |
aa42ffd9 LB |
2607 | !test_range_bit(&BTRFS_I(src)->io_tree, off, off + len - 1, |
2608 | EXTENT_DELALLOC, 0, NULL)) | |
f46b5a66 | 2609 | break; |
aa42ffd9 | 2610 | unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1); |
ae01a0ab YZ |
2611 | if (ordered) |
2612 | btrfs_put_ordered_extent(ordered); | |
9a019196 | 2613 | btrfs_wait_ordered_range(src, off, len); |
f46b5a66 CH |
2614 | } |
2615 | ||
c5c9cd4d | 2616 | /* clone data */ |
33345d01 | 2617 | key.objectid = btrfs_ino(src); |
ae01a0ab YZ |
2618 | key.type = BTRFS_EXTENT_DATA_KEY; |
2619 | key.offset = 0; | |
f46b5a66 CH |
2620 | |
2621 | while (1) { | |
2622 | /* | |
2623 | * note the key will change type as we walk through the | |
2624 | * tree. | |
2625 | */ | |
362a20c5 DS |
2626 | ret = btrfs_search_slot(NULL, BTRFS_I(src)->root, &key, path, |
2627 | 0, 0); | |
f46b5a66 CH |
2628 | if (ret < 0) |
2629 | goto out; | |
2630 | ||
ae01a0ab YZ |
2631 | nritems = btrfs_header_nritems(path->nodes[0]); |
2632 | if (path->slots[0] >= nritems) { | |
362a20c5 | 2633 | ret = btrfs_next_leaf(BTRFS_I(src)->root, path); |
f46b5a66 CH |
2634 | if (ret < 0) |
2635 | goto out; | |
2636 | if (ret > 0) | |
2637 | break; | |
ae01a0ab | 2638 | nritems = btrfs_header_nritems(path->nodes[0]); |
f46b5a66 CH |
2639 | } |
2640 | leaf = path->nodes[0]; | |
2641 | slot = path->slots[0]; | |
f46b5a66 | 2642 | |
ae01a0ab | 2643 | btrfs_item_key_to_cpu(leaf, &key, slot); |
d20f7043 | 2644 | if (btrfs_key_type(&key) > BTRFS_EXTENT_DATA_KEY || |
33345d01 | 2645 | key.objectid != btrfs_ino(src)) |
f46b5a66 CH |
2646 | break; |
2647 | ||
c5c9cd4d SW |
2648 | if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) { |
2649 | struct btrfs_file_extent_item *extent; | |
2650 | int type; | |
31840ae1 ZY |
2651 | u32 size; |
2652 | struct btrfs_key new_key; | |
c5c9cd4d SW |
2653 | u64 disko = 0, diskl = 0; |
2654 | u64 datao = 0, datal = 0; | |
2655 | u8 comp; | |
b5384d48 | 2656 | u64 endoff; |
31840ae1 ZY |
2657 | |
2658 | size = btrfs_item_size_nr(leaf, slot); | |
2659 | read_extent_buffer(leaf, buf, | |
2660 | btrfs_item_ptr_offset(leaf, slot), | |
2661 | size); | |
c5c9cd4d SW |
2662 | |
2663 | extent = btrfs_item_ptr(leaf, slot, | |
2664 | struct btrfs_file_extent_item); | |
2665 | comp = btrfs_file_extent_compression(leaf, extent); | |
2666 | type = btrfs_file_extent_type(leaf, extent); | |
c8a894d7 CM |
2667 | if (type == BTRFS_FILE_EXTENT_REG || |
2668 | type == BTRFS_FILE_EXTENT_PREALLOC) { | |
d397712b CM |
2669 | disko = btrfs_file_extent_disk_bytenr(leaf, |
2670 | extent); | |
2671 | diskl = btrfs_file_extent_disk_num_bytes(leaf, | |
2672 | extent); | |
c5c9cd4d | 2673 | datao = btrfs_file_extent_offset(leaf, extent); |
d397712b CM |
2674 | datal = btrfs_file_extent_num_bytes(leaf, |
2675 | extent); | |
c5c9cd4d SW |
2676 | } else if (type == BTRFS_FILE_EXTENT_INLINE) { |
2677 | /* take upper bound, may be compressed */ | |
2678 | datal = btrfs_file_extent_ram_bytes(leaf, | |
2679 | extent); | |
2680 | } | |
b3b4aa74 | 2681 | btrfs_release_path(path); |
31840ae1 | 2682 | |
050006a7 | 2683 | if (key.offset + datal <= off || |
aa42ffd9 | 2684 | key.offset >= off + len - 1) |
c5c9cd4d SW |
2685 | goto next; |
2686 | ||
31840ae1 | 2687 | memcpy(&new_key, &key, sizeof(new_key)); |
33345d01 | 2688 | new_key.objectid = btrfs_ino(inode); |
4d728ec7 LZ |
2689 | if (off <= key.offset) |
2690 | new_key.offset = key.offset + destoff - off; | |
2691 | else | |
2692 | new_key.offset = destoff; | |
31840ae1 | 2693 | |
b6f3409b SW |
2694 | /* |
2695 | * 1 - adjusting old extent (we may have to split it) | |
2696 | * 1 - add new extent | |
2697 | * 1 - inode update | |
2698 | */ | |
2699 | trans = btrfs_start_transaction(root, 3); | |
a22285a6 YZ |
2700 | if (IS_ERR(trans)) { |
2701 | ret = PTR_ERR(trans); | |
2702 | goto out; | |
2703 | } | |
2704 | ||
c8a894d7 CM |
2705 | if (type == BTRFS_FILE_EXTENT_REG || |
2706 | type == BTRFS_FILE_EXTENT_PREALLOC) { | |
d72c0842 LZ |
2707 | /* |
2708 | * a | --- range to clone ---| b | |
2709 | * | ------------- extent ------------- | | |
2710 | */ | |
2711 | ||
2712 | /* substract range b */ | |
2713 | if (key.offset + datal > off + len) | |
2714 | datal = off + len - key.offset; | |
2715 | ||
2716 | /* substract range a */ | |
a22285a6 YZ |
2717 | if (off > key.offset) { |
2718 | datao += off - key.offset; | |
2719 | datal -= off - key.offset; | |
2720 | } | |
2721 | ||
5dc562c5 | 2722 | ret = btrfs_drop_extents(trans, root, inode, |
a22285a6 YZ |
2723 | new_key.offset, |
2724 | new_key.offset + datal, | |
2671485d | 2725 | 1); |
79787eaa JM |
2726 | if (ret) { |
2727 | btrfs_abort_transaction(trans, root, | |
2728 | ret); | |
2729 | btrfs_end_transaction(trans, root); | |
2730 | goto out; | |
2731 | } | |
a22285a6 | 2732 | |
c5c9cd4d SW |
2733 | ret = btrfs_insert_empty_item(trans, root, path, |
2734 | &new_key, size); | |
79787eaa JM |
2735 | if (ret) { |
2736 | btrfs_abort_transaction(trans, root, | |
2737 | ret); | |
2738 | btrfs_end_transaction(trans, root); | |
2739 | goto out; | |
2740 | } | |
c5c9cd4d SW |
2741 | |
2742 | leaf = path->nodes[0]; | |
2743 | slot = path->slots[0]; | |
2744 | write_extent_buffer(leaf, buf, | |
31840ae1 ZY |
2745 | btrfs_item_ptr_offset(leaf, slot), |
2746 | size); | |
ae01a0ab | 2747 | |
c5c9cd4d | 2748 | extent = btrfs_item_ptr(leaf, slot, |
f46b5a66 | 2749 | struct btrfs_file_extent_item); |
c5c9cd4d | 2750 | |
c5c9cd4d SW |
2751 | /* disko == 0 means it's a hole */ |
2752 | if (!disko) | |
2753 | datao = 0; | |
c5c9cd4d SW |
2754 | |
2755 | btrfs_set_file_extent_offset(leaf, extent, | |
2756 | datao); | |
2757 | btrfs_set_file_extent_num_bytes(leaf, extent, | |
2758 | datal); | |
2759 | if (disko) { | |
2760 | inode_add_bytes(inode, datal); | |
ae01a0ab | 2761 | ret = btrfs_inc_extent_ref(trans, root, |
5d4f98a2 YZ |
2762 | disko, diskl, 0, |
2763 | root->root_key.objectid, | |
33345d01 | 2764 | btrfs_ino(inode), |
66d7e7f0 AJ |
2765 | new_key.offset - datao, |
2766 | 0); | |
79787eaa JM |
2767 | if (ret) { |
2768 | btrfs_abort_transaction(trans, | |
2769 | root, | |
2770 | ret); | |
2771 | btrfs_end_transaction(trans, | |
2772 | root); | |
2773 | goto out; | |
2774 | ||
2775 | } | |
f46b5a66 | 2776 | } |
c5c9cd4d SW |
2777 | } else if (type == BTRFS_FILE_EXTENT_INLINE) { |
2778 | u64 skip = 0; | |
2779 | u64 trim = 0; | |
2780 | if (off > key.offset) { | |
2781 | skip = off - key.offset; | |
2782 | new_key.offset += skip; | |
2783 | } | |
d397712b | 2784 | |
aa42ffd9 LB |
2785 | if (key.offset + datal > off + len) |
2786 | trim = key.offset + datal - (off + len); | |
d397712b | 2787 | |
c5c9cd4d | 2788 | if (comp && (skip || trim)) { |
c5c9cd4d | 2789 | ret = -EINVAL; |
a22285a6 | 2790 | btrfs_end_transaction(trans, root); |
c5c9cd4d SW |
2791 | goto out; |
2792 | } | |
2793 | size -= skip + trim; | |
2794 | datal -= skip + trim; | |
a22285a6 | 2795 | |
5dc562c5 | 2796 | ret = btrfs_drop_extents(trans, root, inode, |
a22285a6 YZ |
2797 | new_key.offset, |
2798 | new_key.offset + datal, | |
2671485d | 2799 | 1); |
79787eaa JM |
2800 | if (ret) { |
2801 | btrfs_abort_transaction(trans, root, | |
2802 | ret); | |
2803 | btrfs_end_transaction(trans, root); | |
2804 | goto out; | |
2805 | } | |
a22285a6 | 2806 | |
c5c9cd4d SW |
2807 | ret = btrfs_insert_empty_item(trans, root, path, |
2808 | &new_key, size); | |
79787eaa JM |
2809 | if (ret) { |
2810 | btrfs_abort_transaction(trans, root, | |
2811 | ret); | |
2812 | btrfs_end_transaction(trans, root); | |
2813 | goto out; | |
2814 | } | |
c5c9cd4d SW |
2815 | |
2816 | if (skip) { | |
d397712b CM |
2817 | u32 start = |
2818 | btrfs_file_extent_calc_inline_size(0); | |
c5c9cd4d SW |
2819 | memmove(buf+start, buf+start+skip, |
2820 | datal); | |
2821 | } | |
2822 | ||
2823 | leaf = path->nodes[0]; | |
2824 | slot = path->slots[0]; | |
2825 | write_extent_buffer(leaf, buf, | |
2826 | btrfs_item_ptr_offset(leaf, slot), | |
2827 | size); | |
2828 | inode_add_bytes(inode, datal); | |
f46b5a66 | 2829 | } |
c5c9cd4d SW |
2830 | |
2831 | btrfs_mark_buffer_dirty(leaf); | |
b3b4aa74 | 2832 | btrfs_release_path(path); |
c5c9cd4d | 2833 | |
0c4d2d95 | 2834 | inode_inc_iversion(inode); |
a22285a6 | 2835 | inode->i_mtime = inode->i_ctime = CURRENT_TIME; |
b5384d48 SW |
2836 | |
2837 | /* | |
2838 | * we round up to the block size at eof when | |
2839 | * determining which extents to clone above, | |
2840 | * but shouldn't round up the file size | |
2841 | */ | |
2842 | endoff = new_key.offset + datal; | |
5f3888ff LZ |
2843 | if (endoff > destoff+olen) |
2844 | endoff = destoff+olen; | |
b5384d48 SW |
2845 | if (endoff > inode->i_size) |
2846 | btrfs_i_size_write(inode, endoff); | |
2847 | ||
a22285a6 | 2848 | ret = btrfs_update_inode(trans, root, inode); |
79787eaa JM |
2849 | if (ret) { |
2850 | btrfs_abort_transaction(trans, root, ret); | |
2851 | btrfs_end_transaction(trans, root); | |
2852 | goto out; | |
2853 | } | |
2854 | ret = btrfs_end_transaction(trans, root); | |
a22285a6 | 2855 | } |
d397712b | 2856 | next: |
b3b4aa74 | 2857 | btrfs_release_path(path); |
f46b5a66 | 2858 | key.offset++; |
f46b5a66 | 2859 | } |
f46b5a66 CH |
2860 | ret = 0; |
2861 | out: | |
b3b4aa74 | 2862 | btrfs_release_path(path); |
aa42ffd9 | 2863 | unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1); |
f46b5a66 CH |
2864 | out_unlock: |
2865 | mutex_unlock(&src->i_mutex); | |
a96fbc72 LB |
2866 | if (!same_inode) |
2867 | mutex_unlock(&inode->i_mutex); | |
ae01a0ab YZ |
2868 | vfree(buf); |
2869 | btrfs_free_path(path); | |
f46b5a66 | 2870 | out_fput: |
2903ff01 | 2871 | fdput(src_file); |
ab67b7c1 | 2872 | out_drop_write: |
2a79f17e | 2873 | mnt_drop_write_file(file); |
f46b5a66 CH |
2874 | return ret; |
2875 | } | |
2876 | ||
7a865e8a | 2877 | static long btrfs_ioctl_clone_range(struct file *file, void __user *argp) |
c5c9cd4d SW |
2878 | { |
2879 | struct btrfs_ioctl_clone_range_args args; | |
2880 | ||
7a865e8a | 2881 | if (copy_from_user(&args, argp, sizeof(args))) |
c5c9cd4d SW |
2882 | return -EFAULT; |
2883 | return btrfs_ioctl_clone(file, args.src_fd, args.src_offset, | |
2884 | args.src_length, args.dest_offset); | |
2885 | } | |
2886 | ||
f46b5a66 CH |
2887 | /* |
2888 | * there are many ways the trans_start and trans_end ioctls can lead | |
2889 | * to deadlocks. They should only be used by applications that | |
2890 | * basically own the machine, and have a very in depth understanding | |
2891 | * of all the possible deadlocks and enospc problems. | |
2892 | */ | |
b2950863 | 2893 | static long btrfs_ioctl_trans_start(struct file *file) |
f46b5a66 | 2894 | { |
496ad9aa | 2895 | struct inode *inode = file_inode(file); |
f46b5a66 CH |
2896 | struct btrfs_root *root = BTRFS_I(inode)->root; |
2897 | struct btrfs_trans_handle *trans; | |
1ab86aed | 2898 | int ret; |
f46b5a66 | 2899 | |
1ab86aed | 2900 | ret = -EPERM; |
df5b5520 | 2901 | if (!capable(CAP_SYS_ADMIN)) |
1ab86aed | 2902 | goto out; |
df5b5520 | 2903 | |
1ab86aed SW |
2904 | ret = -EINPROGRESS; |
2905 | if (file->private_data) | |
f46b5a66 | 2906 | goto out; |
9ca9ee09 | 2907 | |
b83cc969 LZ |
2908 | ret = -EROFS; |
2909 | if (btrfs_root_readonly(root)) | |
2910 | goto out; | |
2911 | ||
a561be71 | 2912 | ret = mnt_want_write_file(file); |
c146afad YZ |
2913 | if (ret) |
2914 | goto out; | |
2915 | ||
a4abeea4 | 2916 | atomic_inc(&root->fs_info->open_ioctl_trans); |
9ca9ee09 | 2917 | |
1ab86aed | 2918 | ret = -ENOMEM; |
7a7eaa40 | 2919 | trans = btrfs_start_ioctl_transaction(root); |
abd30bb0 | 2920 | if (IS_ERR(trans)) |
1ab86aed SW |
2921 | goto out_drop; |
2922 | ||
2923 | file->private_data = trans; | |
2924 | return 0; | |
2925 | ||
2926 | out_drop: | |
a4abeea4 | 2927 | atomic_dec(&root->fs_info->open_ioctl_trans); |
2a79f17e | 2928 | mnt_drop_write_file(file); |
f46b5a66 | 2929 | out: |
f46b5a66 CH |
2930 | return ret; |
2931 | } | |
2932 | ||
6ef5ed0d JB |
2933 | static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp) |
2934 | { | |
496ad9aa | 2935 | struct inode *inode = file_inode(file); |
6ef5ed0d JB |
2936 | struct btrfs_root *root = BTRFS_I(inode)->root; |
2937 | struct btrfs_root *new_root; | |
2938 | struct btrfs_dir_item *di; | |
2939 | struct btrfs_trans_handle *trans; | |
2940 | struct btrfs_path *path; | |
2941 | struct btrfs_key location; | |
2942 | struct btrfs_disk_key disk_key; | |
6ef5ed0d JB |
2943 | u64 objectid = 0; |
2944 | u64 dir_id; | |
3c04ce01 | 2945 | int ret; |
6ef5ed0d JB |
2946 | |
2947 | if (!capable(CAP_SYS_ADMIN)) | |
2948 | return -EPERM; | |
2949 | ||
3c04ce01 MX |
2950 | ret = mnt_want_write_file(file); |
2951 | if (ret) | |
2952 | return ret; | |
2953 | ||
2954 | if (copy_from_user(&objectid, argp, sizeof(objectid))) { | |
2955 | ret = -EFAULT; | |
2956 | goto out; | |
2957 | } | |
6ef5ed0d JB |
2958 | |
2959 | if (!objectid) | |
2960 | objectid = root->root_key.objectid; | |
2961 | ||
2962 | location.objectid = objectid; | |
2963 | location.type = BTRFS_ROOT_ITEM_KEY; | |
2964 | location.offset = (u64)-1; | |
2965 | ||
2966 | new_root = btrfs_read_fs_root_no_name(root->fs_info, &location); | |
3c04ce01 MX |
2967 | if (IS_ERR(new_root)) { |
2968 | ret = PTR_ERR(new_root); | |
2969 | goto out; | |
2970 | } | |
6ef5ed0d | 2971 | |
6ef5ed0d | 2972 | path = btrfs_alloc_path(); |
3c04ce01 MX |
2973 | if (!path) { |
2974 | ret = -ENOMEM; | |
2975 | goto out; | |
2976 | } | |
6ef5ed0d JB |
2977 | path->leave_spinning = 1; |
2978 | ||
2979 | trans = btrfs_start_transaction(root, 1); | |
98d5dc13 | 2980 | if (IS_ERR(trans)) { |
6ef5ed0d | 2981 | btrfs_free_path(path); |
3c04ce01 MX |
2982 | ret = PTR_ERR(trans); |
2983 | goto out; | |
6ef5ed0d JB |
2984 | } |
2985 | ||
6c41761f | 2986 | dir_id = btrfs_super_root_dir(root->fs_info->super_copy); |
6ef5ed0d JB |
2987 | di = btrfs_lookup_dir_item(trans, root->fs_info->tree_root, path, |
2988 | dir_id, "default", 7, 1); | |
cf1e99a4 | 2989 | if (IS_ERR_OR_NULL(di)) { |
6ef5ed0d JB |
2990 | btrfs_free_path(path); |
2991 | btrfs_end_transaction(trans, root); | |
2992 | printk(KERN_ERR "Umm, you don't have the default dir item, " | |
2993 | "this isn't going to work\n"); | |
3c04ce01 MX |
2994 | ret = -ENOENT; |
2995 | goto out; | |
6ef5ed0d JB |
2996 | } |
2997 | ||
2998 | btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key); | |
2999 | btrfs_set_dir_item_key(path->nodes[0], di, &disk_key); | |
3000 | btrfs_mark_buffer_dirty(path->nodes[0]); | |
3001 | btrfs_free_path(path); | |
3002 | ||
2b0ce2c2 | 3003 | btrfs_set_fs_incompat(root->fs_info, DEFAULT_SUBVOL); |
6ef5ed0d | 3004 | btrfs_end_transaction(trans, root); |
3c04ce01 MX |
3005 | out: |
3006 | mnt_drop_write_file(file); | |
3007 | return ret; | |
6ef5ed0d JB |
3008 | } |
3009 | ||
5af3e8cc SB |
3010 | void btrfs_get_block_group_info(struct list_head *groups_list, |
3011 | struct btrfs_ioctl_space_info *space) | |
bf5fc093 JB |
3012 | { |
3013 | struct btrfs_block_group_cache *block_group; | |
3014 | ||
3015 | space->total_bytes = 0; | |
3016 | space->used_bytes = 0; | |
3017 | space->flags = 0; | |
3018 | list_for_each_entry(block_group, groups_list, list) { | |
3019 | space->flags = block_group->flags; | |
3020 | space->total_bytes += block_group->key.offset; | |
3021 | space->used_bytes += | |
3022 | btrfs_block_group_used(&block_group->item); | |
3023 | } | |
3024 | } | |
3025 | ||
48a3b636 | 3026 | static long btrfs_ioctl_space_info(struct btrfs_root *root, void __user *arg) |
1406e432 JB |
3027 | { |
3028 | struct btrfs_ioctl_space_args space_args; | |
3029 | struct btrfs_ioctl_space_info space; | |
3030 | struct btrfs_ioctl_space_info *dest; | |
7fde62bf | 3031 | struct btrfs_ioctl_space_info *dest_orig; |
13f2696f | 3032 | struct btrfs_ioctl_space_info __user *user_dest; |
1406e432 | 3033 | struct btrfs_space_info *info; |
bf5fc093 JB |
3034 | u64 types[] = {BTRFS_BLOCK_GROUP_DATA, |
3035 | BTRFS_BLOCK_GROUP_SYSTEM, | |
3036 | BTRFS_BLOCK_GROUP_METADATA, | |
3037 | BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA}; | |
3038 | int num_types = 4; | |
7fde62bf | 3039 | int alloc_size; |
1406e432 | 3040 | int ret = 0; |
51788b1b | 3041 | u64 slot_count = 0; |
bf5fc093 | 3042 | int i, c; |
1406e432 JB |
3043 | |
3044 | if (copy_from_user(&space_args, | |
3045 | (struct btrfs_ioctl_space_args __user *)arg, | |
3046 | sizeof(space_args))) | |
3047 | return -EFAULT; | |
3048 | ||
bf5fc093 JB |
3049 | for (i = 0; i < num_types; i++) { |
3050 | struct btrfs_space_info *tmp; | |
3051 | ||
3052 | info = NULL; | |
3053 | rcu_read_lock(); | |
3054 | list_for_each_entry_rcu(tmp, &root->fs_info->space_info, | |
3055 | list) { | |
3056 | if (tmp->flags == types[i]) { | |
3057 | info = tmp; | |
3058 | break; | |
3059 | } | |
3060 | } | |
3061 | rcu_read_unlock(); | |
3062 | ||
3063 | if (!info) | |
3064 | continue; | |
3065 | ||
3066 | down_read(&info->groups_sem); | |
3067 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
3068 | if (!list_empty(&info->block_groups[c])) | |
3069 | slot_count++; | |
3070 | } | |
3071 | up_read(&info->groups_sem); | |
3072 | } | |
7fde62bf CM |
3073 | |
3074 | /* space_slots == 0 means they are asking for a count */ | |
3075 | if (space_args.space_slots == 0) { | |
3076 | space_args.total_spaces = slot_count; | |
3077 | goto out; | |
3078 | } | |
bf5fc093 | 3079 | |
51788b1b | 3080 | slot_count = min_t(u64, space_args.space_slots, slot_count); |
bf5fc093 | 3081 | |
7fde62bf | 3082 | alloc_size = sizeof(*dest) * slot_count; |
bf5fc093 | 3083 | |
7fde62bf CM |
3084 | /* we generally have at most 6 or so space infos, one for each raid |
3085 | * level. So, a whole page should be more than enough for everyone | |
3086 | */ | |
3087 | if (alloc_size > PAGE_CACHE_SIZE) | |
3088 | return -ENOMEM; | |
3089 | ||
1406e432 | 3090 | space_args.total_spaces = 0; |
7fde62bf CM |
3091 | dest = kmalloc(alloc_size, GFP_NOFS); |
3092 | if (!dest) | |
3093 | return -ENOMEM; | |
3094 | dest_orig = dest; | |
1406e432 | 3095 | |
7fde62bf | 3096 | /* now we have a buffer to copy into */ |
bf5fc093 JB |
3097 | for (i = 0; i < num_types; i++) { |
3098 | struct btrfs_space_info *tmp; | |
3099 | ||
51788b1b DR |
3100 | if (!slot_count) |
3101 | break; | |
3102 | ||
bf5fc093 JB |
3103 | info = NULL; |
3104 | rcu_read_lock(); | |
3105 | list_for_each_entry_rcu(tmp, &root->fs_info->space_info, | |
3106 | list) { | |
3107 | if (tmp->flags == types[i]) { | |
3108 | info = tmp; | |
3109 | break; | |
3110 | } | |
3111 | } | |
3112 | rcu_read_unlock(); | |
7fde62bf | 3113 | |
bf5fc093 JB |
3114 | if (!info) |
3115 | continue; | |
3116 | down_read(&info->groups_sem); | |
3117 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
3118 | if (!list_empty(&info->block_groups[c])) { | |
5af3e8cc SB |
3119 | btrfs_get_block_group_info( |
3120 | &info->block_groups[c], &space); | |
bf5fc093 JB |
3121 | memcpy(dest, &space, sizeof(space)); |
3122 | dest++; | |
3123 | space_args.total_spaces++; | |
51788b1b | 3124 | slot_count--; |
bf5fc093 | 3125 | } |
51788b1b DR |
3126 | if (!slot_count) |
3127 | break; | |
bf5fc093 JB |
3128 | } |
3129 | up_read(&info->groups_sem); | |
1406e432 | 3130 | } |
1406e432 | 3131 | |
2eec6c81 | 3132 | user_dest = (struct btrfs_ioctl_space_info __user *) |
7fde62bf CM |
3133 | (arg + sizeof(struct btrfs_ioctl_space_args)); |
3134 | ||
3135 | if (copy_to_user(user_dest, dest_orig, alloc_size)) | |
3136 | ret = -EFAULT; | |
3137 | ||
3138 | kfree(dest_orig); | |
3139 | out: | |
3140 | if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args))) | |
1406e432 JB |
3141 | ret = -EFAULT; |
3142 | ||
3143 | return ret; | |
3144 | } | |
3145 | ||
f46b5a66 CH |
3146 | /* |
3147 | * there are many ways the trans_start and trans_end ioctls can lead | |
3148 | * to deadlocks. They should only be used by applications that | |
3149 | * basically own the machine, and have a very in depth understanding | |
3150 | * of all the possible deadlocks and enospc problems. | |
3151 | */ | |
3152 | long btrfs_ioctl_trans_end(struct file *file) | |
3153 | { | |
496ad9aa | 3154 | struct inode *inode = file_inode(file); |
f46b5a66 CH |
3155 | struct btrfs_root *root = BTRFS_I(inode)->root; |
3156 | struct btrfs_trans_handle *trans; | |
f46b5a66 | 3157 | |
f46b5a66 | 3158 | trans = file->private_data; |
1ab86aed SW |
3159 | if (!trans) |
3160 | return -EINVAL; | |
b214107e | 3161 | file->private_data = NULL; |
9ca9ee09 | 3162 | |
1ab86aed SW |
3163 | btrfs_end_transaction(trans, root); |
3164 | ||
a4abeea4 | 3165 | atomic_dec(&root->fs_info->open_ioctl_trans); |
9ca9ee09 | 3166 | |
2a79f17e | 3167 | mnt_drop_write_file(file); |
1ab86aed | 3168 | return 0; |
f46b5a66 CH |
3169 | } |
3170 | ||
9a8c28be MX |
3171 | static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root, |
3172 | void __user *argp) | |
46204592 | 3173 | { |
46204592 SW |
3174 | struct btrfs_trans_handle *trans; |
3175 | u64 transid; | |
db5b493a | 3176 | int ret; |
46204592 | 3177 | |
d4edf39b | 3178 | trans = btrfs_attach_transaction_barrier(root); |
ff7c1d33 MX |
3179 | if (IS_ERR(trans)) { |
3180 | if (PTR_ERR(trans) != -ENOENT) | |
3181 | return PTR_ERR(trans); | |
3182 | ||
3183 | /* No running transaction, don't bother */ | |
3184 | transid = root->fs_info->last_trans_committed; | |
3185 | goto out; | |
3186 | } | |
46204592 | 3187 | transid = trans->transid; |
db5b493a | 3188 | ret = btrfs_commit_transaction_async(trans, root, 0); |
8b2b2d3c TI |
3189 | if (ret) { |
3190 | btrfs_end_transaction(trans, root); | |
db5b493a | 3191 | return ret; |
8b2b2d3c | 3192 | } |
ff7c1d33 | 3193 | out: |
46204592 SW |
3194 | if (argp) |
3195 | if (copy_to_user(argp, &transid, sizeof(transid))) | |
3196 | return -EFAULT; | |
3197 | return 0; | |
3198 | } | |
3199 | ||
9a8c28be MX |
3200 | static noinline long btrfs_ioctl_wait_sync(struct btrfs_root *root, |
3201 | void __user *argp) | |
46204592 | 3202 | { |
46204592 SW |
3203 | u64 transid; |
3204 | ||
3205 | if (argp) { | |
3206 | if (copy_from_user(&transid, argp, sizeof(transid))) | |
3207 | return -EFAULT; | |
3208 | } else { | |
3209 | transid = 0; /* current trans */ | |
3210 | } | |
3211 | return btrfs_wait_for_commit(root, transid); | |
3212 | } | |
3213 | ||
b8e95489 | 3214 | static long btrfs_ioctl_scrub(struct file *file, void __user *arg) |
475f6387 | 3215 | { |
496ad9aa | 3216 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
475f6387 | 3217 | struct btrfs_ioctl_scrub_args *sa; |
b8e95489 | 3218 | int ret; |
475f6387 JS |
3219 | |
3220 | if (!capable(CAP_SYS_ADMIN)) | |
3221 | return -EPERM; | |
3222 | ||
3223 | sa = memdup_user(arg, sizeof(*sa)); | |
3224 | if (IS_ERR(sa)) | |
3225 | return PTR_ERR(sa); | |
3226 | ||
b8e95489 MX |
3227 | if (!(sa->flags & BTRFS_SCRUB_READONLY)) { |
3228 | ret = mnt_want_write_file(file); | |
3229 | if (ret) | |
3230 | goto out; | |
3231 | } | |
3232 | ||
aa1b8cd4 | 3233 | ret = btrfs_scrub_dev(root->fs_info, sa->devid, sa->start, sa->end, |
63a212ab SB |
3234 | &sa->progress, sa->flags & BTRFS_SCRUB_READONLY, |
3235 | 0); | |
475f6387 JS |
3236 | |
3237 | if (copy_to_user(arg, sa, sizeof(*sa))) | |
3238 | ret = -EFAULT; | |
3239 | ||
b8e95489 MX |
3240 | if (!(sa->flags & BTRFS_SCRUB_READONLY)) |
3241 | mnt_drop_write_file(file); | |
3242 | out: | |
475f6387 JS |
3243 | kfree(sa); |
3244 | return ret; | |
3245 | } | |
3246 | ||
3247 | static long btrfs_ioctl_scrub_cancel(struct btrfs_root *root, void __user *arg) | |
3248 | { | |
3249 | if (!capable(CAP_SYS_ADMIN)) | |
3250 | return -EPERM; | |
3251 | ||
aa1b8cd4 | 3252 | return btrfs_scrub_cancel(root->fs_info); |
475f6387 JS |
3253 | } |
3254 | ||
3255 | static long btrfs_ioctl_scrub_progress(struct btrfs_root *root, | |
3256 | void __user *arg) | |
3257 | { | |
3258 | struct btrfs_ioctl_scrub_args *sa; | |
3259 | int ret; | |
3260 | ||
3261 | if (!capable(CAP_SYS_ADMIN)) | |
3262 | return -EPERM; | |
3263 | ||
3264 | sa = memdup_user(arg, sizeof(*sa)); | |
3265 | if (IS_ERR(sa)) | |
3266 | return PTR_ERR(sa); | |
3267 | ||
3268 | ret = btrfs_scrub_progress(root, sa->devid, &sa->progress); | |
3269 | ||
3270 | if (copy_to_user(arg, sa, sizeof(*sa))) | |
3271 | ret = -EFAULT; | |
3272 | ||
3273 | kfree(sa); | |
3274 | return ret; | |
3275 | } | |
3276 | ||
c11d2c23 | 3277 | static long btrfs_ioctl_get_dev_stats(struct btrfs_root *root, |
b27f7c0c | 3278 | void __user *arg) |
c11d2c23 SB |
3279 | { |
3280 | struct btrfs_ioctl_get_dev_stats *sa; | |
3281 | int ret; | |
3282 | ||
c11d2c23 SB |
3283 | sa = memdup_user(arg, sizeof(*sa)); |
3284 | if (IS_ERR(sa)) | |
3285 | return PTR_ERR(sa); | |
3286 | ||
b27f7c0c DS |
3287 | if ((sa->flags & BTRFS_DEV_STATS_RESET) && !capable(CAP_SYS_ADMIN)) { |
3288 | kfree(sa); | |
3289 | return -EPERM; | |
3290 | } | |
3291 | ||
3292 | ret = btrfs_get_dev_stats(root, sa); | |
c11d2c23 SB |
3293 | |
3294 | if (copy_to_user(arg, sa, sizeof(*sa))) | |
3295 | ret = -EFAULT; | |
3296 | ||
3297 | kfree(sa); | |
3298 | return ret; | |
3299 | } | |
3300 | ||
3f6bcfbd SB |
3301 | static long btrfs_ioctl_dev_replace(struct btrfs_root *root, void __user *arg) |
3302 | { | |
3303 | struct btrfs_ioctl_dev_replace_args *p; | |
3304 | int ret; | |
3305 | ||
3306 | if (!capable(CAP_SYS_ADMIN)) | |
3307 | return -EPERM; | |
3308 | ||
3309 | p = memdup_user(arg, sizeof(*p)); | |
3310 | if (IS_ERR(p)) | |
3311 | return PTR_ERR(p); | |
3312 | ||
3313 | switch (p->cmd) { | |
3314 | case BTRFS_IOCTL_DEV_REPLACE_CMD_START: | |
3315 | if (atomic_xchg( | |
3316 | &root->fs_info->mutually_exclusive_operation_running, | |
3317 | 1)) { | |
3318 | pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n"); | |
3319 | ret = -EINPROGRESS; | |
3320 | } else { | |
3321 | ret = btrfs_dev_replace_start(root, p); | |
3322 | atomic_set( | |
3323 | &root->fs_info->mutually_exclusive_operation_running, | |
3324 | 0); | |
3325 | } | |
3326 | break; | |
3327 | case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS: | |
3328 | btrfs_dev_replace_status(root->fs_info, p); | |
3329 | ret = 0; | |
3330 | break; | |
3331 | case BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL: | |
3332 | ret = btrfs_dev_replace_cancel(root->fs_info, p); | |
3333 | break; | |
3334 | default: | |
3335 | ret = -EINVAL; | |
3336 | break; | |
3337 | } | |
3338 | ||
3339 | if (copy_to_user(arg, p, sizeof(*p))) | |
3340 | ret = -EFAULT; | |
3341 | ||
3342 | kfree(p); | |
3343 | return ret; | |
3344 | } | |
3345 | ||
d7728c96 JS |
3346 | static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg) |
3347 | { | |
3348 | int ret = 0; | |
3349 | int i; | |
740c3d22 | 3350 | u64 rel_ptr; |
d7728c96 | 3351 | int size; |
806468f8 | 3352 | struct btrfs_ioctl_ino_path_args *ipa = NULL; |
d7728c96 JS |
3353 | struct inode_fs_paths *ipath = NULL; |
3354 | struct btrfs_path *path; | |
3355 | ||
82b22ac8 | 3356 | if (!capable(CAP_DAC_READ_SEARCH)) |
d7728c96 JS |
3357 | return -EPERM; |
3358 | ||
3359 | path = btrfs_alloc_path(); | |
3360 | if (!path) { | |
3361 | ret = -ENOMEM; | |
3362 | goto out; | |
3363 | } | |
3364 | ||
3365 | ipa = memdup_user(arg, sizeof(*ipa)); | |
3366 | if (IS_ERR(ipa)) { | |
3367 | ret = PTR_ERR(ipa); | |
3368 | ipa = NULL; | |
3369 | goto out; | |
3370 | } | |
3371 | ||
3372 | size = min_t(u32, ipa->size, 4096); | |
3373 | ipath = init_ipath(size, root, path); | |
3374 | if (IS_ERR(ipath)) { | |
3375 | ret = PTR_ERR(ipath); | |
3376 | ipath = NULL; | |
3377 | goto out; | |
3378 | } | |
3379 | ||
3380 | ret = paths_from_inode(ipa->inum, ipath); | |
3381 | if (ret < 0) | |
3382 | goto out; | |
3383 | ||
3384 | for (i = 0; i < ipath->fspath->elem_cnt; ++i) { | |
745c4d8e JM |
3385 | rel_ptr = ipath->fspath->val[i] - |
3386 | (u64)(unsigned long)ipath->fspath->val; | |
740c3d22 | 3387 | ipath->fspath->val[i] = rel_ptr; |
d7728c96 JS |
3388 | } |
3389 | ||
745c4d8e JM |
3390 | ret = copy_to_user((void *)(unsigned long)ipa->fspath, |
3391 | (void *)(unsigned long)ipath->fspath, size); | |
d7728c96 JS |
3392 | if (ret) { |
3393 | ret = -EFAULT; | |
3394 | goto out; | |
3395 | } | |
3396 | ||
3397 | out: | |
3398 | btrfs_free_path(path); | |
3399 | free_ipath(ipath); | |
3400 | kfree(ipa); | |
3401 | ||
3402 | return ret; | |
3403 | } | |
3404 | ||
3405 | static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx) | |
3406 | { | |
3407 | struct btrfs_data_container *inodes = ctx; | |
3408 | const size_t c = 3 * sizeof(u64); | |
3409 | ||
3410 | if (inodes->bytes_left >= c) { | |
3411 | inodes->bytes_left -= c; | |
3412 | inodes->val[inodes->elem_cnt] = inum; | |
3413 | inodes->val[inodes->elem_cnt + 1] = offset; | |
3414 | inodes->val[inodes->elem_cnt + 2] = root; | |
3415 | inodes->elem_cnt += 3; | |
3416 | } else { | |
3417 | inodes->bytes_missing += c - inodes->bytes_left; | |
3418 | inodes->bytes_left = 0; | |
3419 | inodes->elem_missed += 3; | |
3420 | } | |
3421 | ||
3422 | return 0; | |
3423 | } | |
3424 | ||
3425 | static long btrfs_ioctl_logical_to_ino(struct btrfs_root *root, | |
3426 | void __user *arg) | |
3427 | { | |
3428 | int ret = 0; | |
3429 | int size; | |
d7728c96 JS |
3430 | struct btrfs_ioctl_logical_ino_args *loi; |
3431 | struct btrfs_data_container *inodes = NULL; | |
3432 | struct btrfs_path *path = NULL; | |
d7728c96 JS |
3433 | |
3434 | if (!capable(CAP_SYS_ADMIN)) | |
3435 | return -EPERM; | |
3436 | ||
3437 | loi = memdup_user(arg, sizeof(*loi)); | |
3438 | if (IS_ERR(loi)) { | |
3439 | ret = PTR_ERR(loi); | |
3440 | loi = NULL; | |
3441 | goto out; | |
3442 | } | |
3443 | ||
3444 | path = btrfs_alloc_path(); | |
3445 | if (!path) { | |
3446 | ret = -ENOMEM; | |
3447 | goto out; | |
3448 | } | |
3449 | ||
425d17a2 | 3450 | size = min_t(u32, loi->size, 64 * 1024); |
d7728c96 JS |
3451 | inodes = init_data_container(size); |
3452 | if (IS_ERR(inodes)) { | |
3453 | ret = PTR_ERR(inodes); | |
3454 | inodes = NULL; | |
3455 | goto out; | |
3456 | } | |
3457 | ||
df031f07 LB |
3458 | ret = iterate_inodes_from_logical(loi->logical, root->fs_info, path, |
3459 | build_ino_list, inodes); | |
3460 | if (ret == -EINVAL) | |
d7728c96 JS |
3461 | ret = -ENOENT; |
3462 | if (ret < 0) | |
3463 | goto out; | |
3464 | ||
745c4d8e JM |
3465 | ret = copy_to_user((void *)(unsigned long)loi->inodes, |
3466 | (void *)(unsigned long)inodes, size); | |
d7728c96 JS |
3467 | if (ret) |
3468 | ret = -EFAULT; | |
3469 | ||
3470 | out: | |
3471 | btrfs_free_path(path); | |
425d17a2 | 3472 | vfree(inodes); |
d7728c96 JS |
3473 | kfree(loi); |
3474 | ||
3475 | return ret; | |
3476 | } | |
3477 | ||
19a39dce | 3478 | void update_ioctl_balance_args(struct btrfs_fs_info *fs_info, int lock, |
c9e9f97b ID |
3479 | struct btrfs_ioctl_balance_args *bargs) |
3480 | { | |
3481 | struct btrfs_balance_control *bctl = fs_info->balance_ctl; | |
3482 | ||
3483 | bargs->flags = bctl->flags; | |
3484 | ||
837d5b6e ID |
3485 | if (atomic_read(&fs_info->balance_running)) |
3486 | bargs->state |= BTRFS_BALANCE_STATE_RUNNING; | |
3487 | if (atomic_read(&fs_info->balance_pause_req)) | |
3488 | bargs->state |= BTRFS_BALANCE_STATE_PAUSE_REQ; | |
a7e99c69 ID |
3489 | if (atomic_read(&fs_info->balance_cancel_req)) |
3490 | bargs->state |= BTRFS_BALANCE_STATE_CANCEL_REQ; | |
837d5b6e | 3491 | |
c9e9f97b ID |
3492 | memcpy(&bargs->data, &bctl->data, sizeof(bargs->data)); |
3493 | memcpy(&bargs->meta, &bctl->meta, sizeof(bargs->meta)); | |
3494 | memcpy(&bargs->sys, &bctl->sys, sizeof(bargs->sys)); | |
19a39dce ID |
3495 | |
3496 | if (lock) { | |
3497 | spin_lock(&fs_info->balance_lock); | |
3498 | memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat)); | |
3499 | spin_unlock(&fs_info->balance_lock); | |
3500 | } else { | |
3501 | memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat)); | |
3502 | } | |
c9e9f97b ID |
3503 | } |
3504 | ||
9ba1f6e4 | 3505 | static long btrfs_ioctl_balance(struct file *file, void __user *arg) |
c9e9f97b | 3506 | { |
496ad9aa | 3507 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
c9e9f97b ID |
3508 | struct btrfs_fs_info *fs_info = root->fs_info; |
3509 | struct btrfs_ioctl_balance_args *bargs; | |
3510 | struct btrfs_balance_control *bctl; | |
ed0fb78f | 3511 | bool need_unlock; /* for mut. excl. ops lock */ |
c9e9f97b ID |
3512 | int ret; |
3513 | ||
3514 | if (!capable(CAP_SYS_ADMIN)) | |
3515 | return -EPERM; | |
3516 | ||
e54bfa31 | 3517 | ret = mnt_want_write_file(file); |
9ba1f6e4 LB |
3518 | if (ret) |
3519 | return ret; | |
3520 | ||
ed0fb78f ID |
3521 | again: |
3522 | if (!atomic_xchg(&fs_info->mutually_exclusive_operation_running, 1)) { | |
3523 | mutex_lock(&fs_info->volume_mutex); | |
3524 | mutex_lock(&fs_info->balance_mutex); | |
3525 | need_unlock = true; | |
3526 | goto locked; | |
3527 | } | |
3528 | ||
3529 | /* | |
3530 | * mut. excl. ops lock is locked. Three possibilites: | |
3531 | * (1) some other op is running | |
3532 | * (2) balance is running | |
3533 | * (3) balance is paused -- special case (think resume) | |
3534 | */ | |
c9e9f97b | 3535 | mutex_lock(&fs_info->balance_mutex); |
ed0fb78f ID |
3536 | if (fs_info->balance_ctl) { |
3537 | /* this is either (2) or (3) */ | |
3538 | if (!atomic_read(&fs_info->balance_running)) { | |
3539 | mutex_unlock(&fs_info->balance_mutex); | |
3540 | if (!mutex_trylock(&fs_info->volume_mutex)) | |
3541 | goto again; | |
3542 | mutex_lock(&fs_info->balance_mutex); | |
3543 | ||
3544 | if (fs_info->balance_ctl && | |
3545 | !atomic_read(&fs_info->balance_running)) { | |
3546 | /* this is (3) */ | |
3547 | need_unlock = false; | |
3548 | goto locked; | |
3549 | } | |
3550 | ||
3551 | mutex_unlock(&fs_info->balance_mutex); | |
3552 | mutex_unlock(&fs_info->volume_mutex); | |
3553 | goto again; | |
3554 | } else { | |
3555 | /* this is (2) */ | |
3556 | mutex_unlock(&fs_info->balance_mutex); | |
3557 | ret = -EINPROGRESS; | |
3558 | goto out; | |
3559 | } | |
3560 | } else { | |
3561 | /* this is (1) */ | |
3562 | mutex_unlock(&fs_info->balance_mutex); | |
3563 | pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n"); | |
3564 | ret = -EINVAL; | |
3565 | goto out; | |
3566 | } | |
3567 | ||
3568 | locked: | |
3569 | BUG_ON(!atomic_read(&fs_info->mutually_exclusive_operation_running)); | |
c9e9f97b ID |
3570 | |
3571 | if (arg) { | |
3572 | bargs = memdup_user(arg, sizeof(*bargs)); | |
3573 | if (IS_ERR(bargs)) { | |
3574 | ret = PTR_ERR(bargs); | |
ed0fb78f | 3575 | goto out_unlock; |
c9e9f97b | 3576 | } |
de322263 ID |
3577 | |
3578 | if (bargs->flags & BTRFS_BALANCE_RESUME) { | |
3579 | if (!fs_info->balance_ctl) { | |
3580 | ret = -ENOTCONN; | |
3581 | goto out_bargs; | |
3582 | } | |
3583 | ||
3584 | bctl = fs_info->balance_ctl; | |
3585 | spin_lock(&fs_info->balance_lock); | |
3586 | bctl->flags |= BTRFS_BALANCE_RESUME; | |
3587 | spin_unlock(&fs_info->balance_lock); | |
3588 | ||
3589 | goto do_balance; | |
3590 | } | |
c9e9f97b ID |
3591 | } else { |
3592 | bargs = NULL; | |
3593 | } | |
3594 | ||
ed0fb78f | 3595 | if (fs_info->balance_ctl) { |
837d5b6e ID |
3596 | ret = -EINPROGRESS; |
3597 | goto out_bargs; | |
3598 | } | |
3599 | ||
c9e9f97b ID |
3600 | bctl = kzalloc(sizeof(*bctl), GFP_NOFS); |
3601 | if (!bctl) { | |
3602 | ret = -ENOMEM; | |
3603 | goto out_bargs; | |
3604 | } | |
3605 | ||
3606 | bctl->fs_info = fs_info; | |
3607 | if (arg) { | |
3608 | memcpy(&bctl->data, &bargs->data, sizeof(bctl->data)); | |
3609 | memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta)); | |
3610 | memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys)); | |
3611 | ||
3612 | bctl->flags = bargs->flags; | |
f43ffb60 ID |
3613 | } else { |
3614 | /* balance everything - no filters */ | |
3615 | bctl->flags |= BTRFS_BALANCE_TYPE_MASK; | |
c9e9f97b ID |
3616 | } |
3617 | ||
de322263 | 3618 | do_balance: |
c9e9f97b | 3619 | /* |
ed0fb78f ID |
3620 | * Ownership of bctl and mutually_exclusive_operation_running |
3621 | * goes to to btrfs_balance. bctl is freed in __cancel_balance, | |
3622 | * or, if restriper was paused all the way until unmount, in | |
3623 | * free_fs_info. mutually_exclusive_operation_running is | |
3624 | * cleared in __cancel_balance. | |
c9e9f97b | 3625 | */ |
ed0fb78f ID |
3626 | need_unlock = false; |
3627 | ||
3628 | ret = btrfs_balance(bctl, bargs); | |
3629 | ||
c9e9f97b ID |
3630 | if (arg) { |
3631 | if (copy_to_user(arg, bargs, sizeof(*bargs))) | |
3632 | ret = -EFAULT; | |
3633 | } | |
3634 | ||
3635 | out_bargs: | |
3636 | kfree(bargs); | |
ed0fb78f | 3637 | out_unlock: |
c9e9f97b ID |
3638 | mutex_unlock(&fs_info->balance_mutex); |
3639 | mutex_unlock(&fs_info->volume_mutex); | |
ed0fb78f ID |
3640 | if (need_unlock) |
3641 | atomic_set(&fs_info->mutually_exclusive_operation_running, 0); | |
3642 | out: | |
e54bfa31 | 3643 | mnt_drop_write_file(file); |
c9e9f97b ID |
3644 | return ret; |
3645 | } | |
3646 | ||
837d5b6e ID |
3647 | static long btrfs_ioctl_balance_ctl(struct btrfs_root *root, int cmd) |
3648 | { | |
3649 | if (!capable(CAP_SYS_ADMIN)) | |
3650 | return -EPERM; | |
3651 | ||
3652 | switch (cmd) { | |
3653 | case BTRFS_BALANCE_CTL_PAUSE: | |
3654 | return btrfs_pause_balance(root->fs_info); | |
a7e99c69 ID |
3655 | case BTRFS_BALANCE_CTL_CANCEL: |
3656 | return btrfs_cancel_balance(root->fs_info); | |
837d5b6e ID |
3657 | } |
3658 | ||
3659 | return -EINVAL; | |
3660 | } | |
3661 | ||
19a39dce ID |
3662 | static long btrfs_ioctl_balance_progress(struct btrfs_root *root, |
3663 | void __user *arg) | |
3664 | { | |
3665 | struct btrfs_fs_info *fs_info = root->fs_info; | |
3666 | struct btrfs_ioctl_balance_args *bargs; | |
3667 | int ret = 0; | |
3668 | ||
3669 | if (!capable(CAP_SYS_ADMIN)) | |
3670 | return -EPERM; | |
3671 | ||
3672 | mutex_lock(&fs_info->balance_mutex); | |
3673 | if (!fs_info->balance_ctl) { | |
3674 | ret = -ENOTCONN; | |
3675 | goto out; | |
3676 | } | |
3677 | ||
3678 | bargs = kzalloc(sizeof(*bargs), GFP_NOFS); | |
3679 | if (!bargs) { | |
3680 | ret = -ENOMEM; | |
3681 | goto out; | |
3682 | } | |
3683 | ||
3684 | update_ioctl_balance_args(fs_info, 1, bargs); | |
3685 | ||
3686 | if (copy_to_user(arg, bargs, sizeof(*bargs))) | |
3687 | ret = -EFAULT; | |
3688 | ||
3689 | kfree(bargs); | |
3690 | out: | |
3691 | mutex_unlock(&fs_info->balance_mutex); | |
3692 | return ret; | |
3693 | } | |
3694 | ||
905b0dda | 3695 | static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg) |
5d13a37b | 3696 | { |
496ad9aa | 3697 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
5d13a37b AJ |
3698 | struct btrfs_ioctl_quota_ctl_args *sa; |
3699 | struct btrfs_trans_handle *trans = NULL; | |
3700 | int ret; | |
3701 | int err; | |
3702 | ||
3703 | if (!capable(CAP_SYS_ADMIN)) | |
3704 | return -EPERM; | |
3705 | ||
905b0dda MX |
3706 | ret = mnt_want_write_file(file); |
3707 | if (ret) | |
3708 | return ret; | |
5d13a37b AJ |
3709 | |
3710 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
3711 | if (IS_ERR(sa)) { |
3712 | ret = PTR_ERR(sa); | |
3713 | goto drop_write; | |
3714 | } | |
5d13a37b | 3715 | |
7708f029 | 3716 | down_write(&root->fs_info->subvol_sem); |
2f232036 JS |
3717 | trans = btrfs_start_transaction(root->fs_info->tree_root, 2); |
3718 | if (IS_ERR(trans)) { | |
3719 | ret = PTR_ERR(trans); | |
3720 | goto out; | |
5d13a37b AJ |
3721 | } |
3722 | ||
3723 | switch (sa->cmd) { | |
3724 | case BTRFS_QUOTA_CTL_ENABLE: | |
3725 | ret = btrfs_quota_enable(trans, root->fs_info); | |
3726 | break; | |
3727 | case BTRFS_QUOTA_CTL_DISABLE: | |
3728 | ret = btrfs_quota_disable(trans, root->fs_info); | |
3729 | break; | |
5d13a37b AJ |
3730 | default: |
3731 | ret = -EINVAL; | |
3732 | break; | |
3733 | } | |
3734 | ||
2f232036 JS |
3735 | err = btrfs_commit_transaction(trans, root->fs_info->tree_root); |
3736 | if (err && !ret) | |
3737 | ret = err; | |
5d13a37b AJ |
3738 | out: |
3739 | kfree(sa); | |
7708f029 | 3740 | up_write(&root->fs_info->subvol_sem); |
905b0dda MX |
3741 | drop_write: |
3742 | mnt_drop_write_file(file); | |
5d13a37b AJ |
3743 | return ret; |
3744 | } | |
3745 | ||
905b0dda | 3746 | static long btrfs_ioctl_qgroup_assign(struct file *file, void __user *arg) |
5d13a37b | 3747 | { |
496ad9aa | 3748 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
5d13a37b AJ |
3749 | struct btrfs_ioctl_qgroup_assign_args *sa; |
3750 | struct btrfs_trans_handle *trans; | |
3751 | int ret; | |
3752 | int err; | |
3753 | ||
3754 | if (!capable(CAP_SYS_ADMIN)) | |
3755 | return -EPERM; | |
3756 | ||
905b0dda MX |
3757 | ret = mnt_want_write_file(file); |
3758 | if (ret) | |
3759 | return ret; | |
5d13a37b AJ |
3760 | |
3761 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
3762 | if (IS_ERR(sa)) { |
3763 | ret = PTR_ERR(sa); | |
3764 | goto drop_write; | |
3765 | } | |
5d13a37b AJ |
3766 | |
3767 | trans = btrfs_join_transaction(root); | |
3768 | if (IS_ERR(trans)) { | |
3769 | ret = PTR_ERR(trans); | |
3770 | goto out; | |
3771 | } | |
3772 | ||
3773 | /* FIXME: check if the IDs really exist */ | |
3774 | if (sa->assign) { | |
3775 | ret = btrfs_add_qgroup_relation(trans, root->fs_info, | |
3776 | sa->src, sa->dst); | |
3777 | } else { | |
3778 | ret = btrfs_del_qgroup_relation(trans, root->fs_info, | |
3779 | sa->src, sa->dst); | |
3780 | } | |
3781 | ||
3782 | err = btrfs_end_transaction(trans, root); | |
3783 | if (err && !ret) | |
3784 | ret = err; | |
3785 | ||
3786 | out: | |
3787 | kfree(sa); | |
905b0dda MX |
3788 | drop_write: |
3789 | mnt_drop_write_file(file); | |
5d13a37b AJ |
3790 | return ret; |
3791 | } | |
3792 | ||
905b0dda | 3793 | static long btrfs_ioctl_qgroup_create(struct file *file, void __user *arg) |
5d13a37b | 3794 | { |
496ad9aa | 3795 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
5d13a37b AJ |
3796 | struct btrfs_ioctl_qgroup_create_args *sa; |
3797 | struct btrfs_trans_handle *trans; | |
3798 | int ret; | |
3799 | int err; | |
3800 | ||
3801 | if (!capable(CAP_SYS_ADMIN)) | |
3802 | return -EPERM; | |
3803 | ||
905b0dda MX |
3804 | ret = mnt_want_write_file(file); |
3805 | if (ret) | |
3806 | return ret; | |
5d13a37b AJ |
3807 | |
3808 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
3809 | if (IS_ERR(sa)) { |
3810 | ret = PTR_ERR(sa); | |
3811 | goto drop_write; | |
3812 | } | |
5d13a37b | 3813 | |
d86e56cf MX |
3814 | if (!sa->qgroupid) { |
3815 | ret = -EINVAL; | |
3816 | goto out; | |
3817 | } | |
3818 | ||
5d13a37b AJ |
3819 | trans = btrfs_join_transaction(root); |
3820 | if (IS_ERR(trans)) { | |
3821 | ret = PTR_ERR(trans); | |
3822 | goto out; | |
3823 | } | |
3824 | ||
3825 | /* FIXME: check if the IDs really exist */ | |
3826 | if (sa->create) { | |
3827 | ret = btrfs_create_qgroup(trans, root->fs_info, sa->qgroupid, | |
3828 | NULL); | |
3829 | } else { | |
3830 | ret = btrfs_remove_qgroup(trans, root->fs_info, sa->qgroupid); | |
3831 | } | |
3832 | ||
3833 | err = btrfs_end_transaction(trans, root); | |
3834 | if (err && !ret) | |
3835 | ret = err; | |
3836 | ||
3837 | out: | |
3838 | kfree(sa); | |
905b0dda MX |
3839 | drop_write: |
3840 | mnt_drop_write_file(file); | |
5d13a37b AJ |
3841 | return ret; |
3842 | } | |
3843 | ||
905b0dda | 3844 | static long btrfs_ioctl_qgroup_limit(struct file *file, void __user *arg) |
5d13a37b | 3845 | { |
496ad9aa | 3846 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
5d13a37b AJ |
3847 | struct btrfs_ioctl_qgroup_limit_args *sa; |
3848 | struct btrfs_trans_handle *trans; | |
3849 | int ret; | |
3850 | int err; | |
3851 | u64 qgroupid; | |
3852 | ||
3853 | if (!capable(CAP_SYS_ADMIN)) | |
3854 | return -EPERM; | |
3855 | ||
905b0dda MX |
3856 | ret = mnt_want_write_file(file); |
3857 | if (ret) | |
3858 | return ret; | |
5d13a37b AJ |
3859 | |
3860 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
3861 | if (IS_ERR(sa)) { |
3862 | ret = PTR_ERR(sa); | |
3863 | goto drop_write; | |
3864 | } | |
5d13a37b AJ |
3865 | |
3866 | trans = btrfs_join_transaction(root); | |
3867 | if (IS_ERR(trans)) { | |
3868 | ret = PTR_ERR(trans); | |
3869 | goto out; | |
3870 | } | |
3871 | ||
3872 | qgroupid = sa->qgroupid; | |
3873 | if (!qgroupid) { | |
3874 | /* take the current subvol as qgroup */ | |
3875 | qgroupid = root->root_key.objectid; | |
3876 | } | |
3877 | ||
3878 | /* FIXME: check if the IDs really exist */ | |
3879 | ret = btrfs_limit_qgroup(trans, root->fs_info, qgroupid, &sa->lim); | |
3880 | ||
3881 | err = btrfs_end_transaction(trans, root); | |
3882 | if (err && !ret) | |
3883 | ret = err; | |
3884 | ||
3885 | out: | |
3886 | kfree(sa); | |
905b0dda MX |
3887 | drop_write: |
3888 | mnt_drop_write_file(file); | |
5d13a37b AJ |
3889 | return ret; |
3890 | } | |
3891 | ||
2f232036 JS |
3892 | static long btrfs_ioctl_quota_rescan(struct file *file, void __user *arg) |
3893 | { | |
6d0379ec | 3894 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
2f232036 JS |
3895 | struct btrfs_ioctl_quota_rescan_args *qsa; |
3896 | int ret; | |
3897 | ||
3898 | if (!capable(CAP_SYS_ADMIN)) | |
3899 | return -EPERM; | |
3900 | ||
3901 | ret = mnt_want_write_file(file); | |
3902 | if (ret) | |
3903 | return ret; | |
3904 | ||
3905 | qsa = memdup_user(arg, sizeof(*qsa)); | |
3906 | if (IS_ERR(qsa)) { | |
3907 | ret = PTR_ERR(qsa); | |
3908 | goto drop_write; | |
3909 | } | |
3910 | ||
3911 | if (qsa->flags) { | |
3912 | ret = -EINVAL; | |
3913 | goto out; | |
3914 | } | |
3915 | ||
3916 | ret = btrfs_qgroup_rescan(root->fs_info); | |
3917 | ||
3918 | out: | |
3919 | kfree(qsa); | |
3920 | drop_write: | |
3921 | mnt_drop_write_file(file); | |
3922 | return ret; | |
3923 | } | |
3924 | ||
3925 | static long btrfs_ioctl_quota_rescan_status(struct file *file, void __user *arg) | |
3926 | { | |
6d0379ec | 3927 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
2f232036 JS |
3928 | struct btrfs_ioctl_quota_rescan_args *qsa; |
3929 | int ret = 0; | |
3930 | ||
3931 | if (!capable(CAP_SYS_ADMIN)) | |
3932 | return -EPERM; | |
3933 | ||
3934 | qsa = kzalloc(sizeof(*qsa), GFP_NOFS); | |
3935 | if (!qsa) | |
3936 | return -ENOMEM; | |
3937 | ||
3938 | if (root->fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) { | |
3939 | qsa->flags = 1; | |
3940 | qsa->progress = root->fs_info->qgroup_rescan_progress.objectid; | |
3941 | } | |
3942 | ||
3943 | if (copy_to_user(arg, qsa, sizeof(*qsa))) | |
3944 | ret = -EFAULT; | |
3945 | ||
3946 | kfree(qsa); | |
3947 | return ret; | |
3948 | } | |
3949 | ||
57254b6e JS |
3950 | static long btrfs_ioctl_quota_rescan_wait(struct file *file, void __user *arg) |
3951 | { | |
3952 | struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root; | |
3953 | ||
3954 | if (!capable(CAP_SYS_ADMIN)) | |
3955 | return -EPERM; | |
3956 | ||
3957 | return btrfs_qgroup_wait_for_completion(root->fs_info); | |
3958 | } | |
3959 | ||
8ea05e3a AB |
3960 | static long btrfs_ioctl_set_received_subvol(struct file *file, |
3961 | void __user *arg) | |
3962 | { | |
3963 | struct btrfs_ioctl_received_subvol_args *sa = NULL; | |
496ad9aa | 3964 | struct inode *inode = file_inode(file); |
8ea05e3a AB |
3965 | struct btrfs_root *root = BTRFS_I(inode)->root; |
3966 | struct btrfs_root_item *root_item = &root->root_item; | |
3967 | struct btrfs_trans_handle *trans; | |
3968 | struct timespec ct = CURRENT_TIME; | |
3969 | int ret = 0; | |
3970 | ||
3971 | ret = mnt_want_write_file(file); | |
3972 | if (ret < 0) | |
3973 | return ret; | |
3974 | ||
3975 | down_write(&root->fs_info->subvol_sem); | |
3976 | ||
3977 | if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) { | |
3978 | ret = -EINVAL; | |
3979 | goto out; | |
3980 | } | |
3981 | ||
3982 | if (btrfs_root_readonly(root)) { | |
3983 | ret = -EROFS; | |
3984 | goto out; | |
3985 | } | |
3986 | ||
3987 | if (!inode_owner_or_capable(inode)) { | |
3988 | ret = -EACCES; | |
3989 | goto out; | |
3990 | } | |
3991 | ||
3992 | sa = memdup_user(arg, sizeof(*sa)); | |
3993 | if (IS_ERR(sa)) { | |
3994 | ret = PTR_ERR(sa); | |
3995 | sa = NULL; | |
3996 | goto out; | |
3997 | } | |
3998 | ||
3999 | trans = btrfs_start_transaction(root, 1); | |
4000 | if (IS_ERR(trans)) { | |
4001 | ret = PTR_ERR(trans); | |
4002 | trans = NULL; | |
4003 | goto out; | |
4004 | } | |
4005 | ||
4006 | sa->rtransid = trans->transid; | |
4007 | sa->rtime.sec = ct.tv_sec; | |
4008 | sa->rtime.nsec = ct.tv_nsec; | |
4009 | ||
4010 | memcpy(root_item->received_uuid, sa->uuid, BTRFS_UUID_SIZE); | |
4011 | btrfs_set_root_stransid(root_item, sa->stransid); | |
4012 | btrfs_set_root_rtransid(root_item, sa->rtransid); | |
4013 | root_item->stime.sec = cpu_to_le64(sa->stime.sec); | |
4014 | root_item->stime.nsec = cpu_to_le32(sa->stime.nsec); | |
4015 | root_item->rtime.sec = cpu_to_le64(sa->rtime.sec); | |
4016 | root_item->rtime.nsec = cpu_to_le32(sa->rtime.nsec); | |
4017 | ||
4018 | ret = btrfs_update_root(trans, root->fs_info->tree_root, | |
4019 | &root->root_key, &root->root_item); | |
4020 | if (ret < 0) { | |
4021 | btrfs_end_transaction(trans, root); | |
4022 | trans = NULL; | |
4023 | goto out; | |
4024 | } else { | |
4025 | ret = btrfs_commit_transaction(trans, root); | |
4026 | if (ret < 0) | |
4027 | goto out; | |
4028 | } | |
4029 | ||
4030 | ret = copy_to_user(arg, sa, sizeof(*sa)); | |
4031 | if (ret) | |
4032 | ret = -EFAULT; | |
4033 | ||
4034 | out: | |
4035 | kfree(sa); | |
4036 | up_write(&root->fs_info->subvol_sem); | |
4037 | mnt_drop_write_file(file); | |
4038 | return ret; | |
4039 | } | |
4040 | ||
867ab667 | 4041 | static int btrfs_ioctl_get_fslabel(struct file *file, void __user *arg) |
4042 | { | |
6d0379ec | 4043 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
867ab667 | 4044 | const char *label = root->fs_info->super_copy->label; |
4045 | size_t len = strnlen(label, BTRFS_LABEL_SIZE); | |
4046 | int ret; | |
4047 | ||
4048 | if (len == BTRFS_LABEL_SIZE) { | |
4049 | pr_warn("btrfs: label is too long, return the first %zu bytes\n", | |
4050 | --len); | |
4051 | } | |
4052 | ||
4053 | mutex_lock(&root->fs_info->volume_mutex); | |
4054 | ret = copy_to_user(arg, label, len); | |
4055 | mutex_unlock(&root->fs_info->volume_mutex); | |
4056 | ||
4057 | return ret ? -EFAULT : 0; | |
4058 | } | |
4059 | ||
a8bfd4ab | 4060 | static int btrfs_ioctl_set_fslabel(struct file *file, void __user *arg) |
4061 | { | |
6d0379ec | 4062 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
a8bfd4ab | 4063 | struct btrfs_super_block *super_block = root->fs_info->super_copy; |
4064 | struct btrfs_trans_handle *trans; | |
4065 | char label[BTRFS_LABEL_SIZE]; | |
4066 | int ret; | |
4067 | ||
4068 | if (!capable(CAP_SYS_ADMIN)) | |
4069 | return -EPERM; | |
4070 | ||
4071 | if (copy_from_user(label, arg, sizeof(label))) | |
4072 | return -EFAULT; | |
4073 | ||
4074 | if (strnlen(label, BTRFS_LABEL_SIZE) == BTRFS_LABEL_SIZE) { | |
4075 | pr_err("btrfs: unable to set label with more than %d bytes\n", | |
4076 | BTRFS_LABEL_SIZE - 1); | |
4077 | return -EINVAL; | |
4078 | } | |
4079 | ||
4080 | ret = mnt_want_write_file(file); | |
4081 | if (ret) | |
4082 | return ret; | |
4083 | ||
4084 | mutex_lock(&root->fs_info->volume_mutex); | |
4085 | trans = btrfs_start_transaction(root, 0); | |
4086 | if (IS_ERR(trans)) { | |
4087 | ret = PTR_ERR(trans); | |
4088 | goto out_unlock; | |
4089 | } | |
4090 | ||
4091 | strcpy(super_block->label, label); | |
4092 | ret = btrfs_end_transaction(trans, root); | |
4093 | ||
4094 | out_unlock: | |
4095 | mutex_unlock(&root->fs_info->volume_mutex); | |
4096 | mnt_drop_write_file(file); | |
4097 | return ret; | |
4098 | } | |
4099 | ||
f46b5a66 CH |
4100 | long btrfs_ioctl(struct file *file, unsigned int |
4101 | cmd, unsigned long arg) | |
4102 | { | |
496ad9aa | 4103 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
4bcabaa3 | 4104 | void __user *argp = (void __user *)arg; |
f46b5a66 CH |
4105 | |
4106 | switch (cmd) { | |
6cbff00f CH |
4107 | case FS_IOC_GETFLAGS: |
4108 | return btrfs_ioctl_getflags(file, argp); | |
4109 | case FS_IOC_SETFLAGS: | |
4110 | return btrfs_ioctl_setflags(file, argp); | |
4111 | case FS_IOC_GETVERSION: | |
4112 | return btrfs_ioctl_getversion(file, argp); | |
f7039b1d LD |
4113 | case FITRIM: |
4114 | return btrfs_ioctl_fitrim(file, argp); | |
f46b5a66 | 4115 | case BTRFS_IOC_SNAP_CREATE: |
fa0d2b9b | 4116 | return btrfs_ioctl_snap_create(file, argp, 0); |
fdfb1e4f | 4117 | case BTRFS_IOC_SNAP_CREATE_V2: |
fa0d2b9b | 4118 | return btrfs_ioctl_snap_create_v2(file, argp, 0); |
3de4586c | 4119 | case BTRFS_IOC_SUBVOL_CREATE: |
fa0d2b9b | 4120 | return btrfs_ioctl_snap_create(file, argp, 1); |
6f72c7e2 AJ |
4121 | case BTRFS_IOC_SUBVOL_CREATE_V2: |
4122 | return btrfs_ioctl_snap_create_v2(file, argp, 1); | |
76dda93c YZ |
4123 | case BTRFS_IOC_SNAP_DESTROY: |
4124 | return btrfs_ioctl_snap_destroy(file, argp); | |
0caa102d LZ |
4125 | case BTRFS_IOC_SUBVOL_GETFLAGS: |
4126 | return btrfs_ioctl_subvol_getflags(file, argp); | |
4127 | case BTRFS_IOC_SUBVOL_SETFLAGS: | |
4128 | return btrfs_ioctl_subvol_setflags(file, argp); | |
6ef5ed0d JB |
4129 | case BTRFS_IOC_DEFAULT_SUBVOL: |
4130 | return btrfs_ioctl_default_subvol(file, argp); | |
f46b5a66 | 4131 | case BTRFS_IOC_DEFRAG: |
1e701a32 CM |
4132 | return btrfs_ioctl_defrag(file, NULL); |
4133 | case BTRFS_IOC_DEFRAG_RANGE: | |
4134 | return btrfs_ioctl_defrag(file, argp); | |
f46b5a66 | 4135 | case BTRFS_IOC_RESIZE: |
198605a8 | 4136 | return btrfs_ioctl_resize(file, argp); |
f46b5a66 | 4137 | case BTRFS_IOC_ADD_DEV: |
4bcabaa3 | 4138 | return btrfs_ioctl_add_dev(root, argp); |
f46b5a66 | 4139 | case BTRFS_IOC_RM_DEV: |
da24927b | 4140 | return btrfs_ioctl_rm_dev(file, argp); |
475f6387 JS |
4141 | case BTRFS_IOC_FS_INFO: |
4142 | return btrfs_ioctl_fs_info(root, argp); | |
4143 | case BTRFS_IOC_DEV_INFO: | |
4144 | return btrfs_ioctl_dev_info(root, argp); | |
f46b5a66 | 4145 | case BTRFS_IOC_BALANCE: |
9ba1f6e4 | 4146 | return btrfs_ioctl_balance(file, NULL); |
f46b5a66 | 4147 | case BTRFS_IOC_CLONE: |
c5c9cd4d SW |
4148 | return btrfs_ioctl_clone(file, arg, 0, 0, 0); |
4149 | case BTRFS_IOC_CLONE_RANGE: | |
7a865e8a | 4150 | return btrfs_ioctl_clone_range(file, argp); |
f46b5a66 CH |
4151 | case BTRFS_IOC_TRANS_START: |
4152 | return btrfs_ioctl_trans_start(file); | |
4153 | case BTRFS_IOC_TRANS_END: | |
4154 | return btrfs_ioctl_trans_end(file); | |
ac8e9819 CM |
4155 | case BTRFS_IOC_TREE_SEARCH: |
4156 | return btrfs_ioctl_tree_search(file, argp); | |
4157 | case BTRFS_IOC_INO_LOOKUP: | |
4158 | return btrfs_ioctl_ino_lookup(file, argp); | |
d7728c96 JS |
4159 | case BTRFS_IOC_INO_PATHS: |
4160 | return btrfs_ioctl_ino_to_path(root, argp); | |
4161 | case BTRFS_IOC_LOGICAL_INO: | |
4162 | return btrfs_ioctl_logical_to_ino(root, argp); | |
1406e432 JB |
4163 | case BTRFS_IOC_SPACE_INFO: |
4164 | return btrfs_ioctl_space_info(root, argp); | |
f46b5a66 CH |
4165 | case BTRFS_IOC_SYNC: |
4166 | btrfs_sync_fs(file->f_dentry->d_sb, 1); | |
4167 | return 0; | |
46204592 | 4168 | case BTRFS_IOC_START_SYNC: |
9a8c28be | 4169 | return btrfs_ioctl_start_sync(root, argp); |
46204592 | 4170 | case BTRFS_IOC_WAIT_SYNC: |
9a8c28be | 4171 | return btrfs_ioctl_wait_sync(root, argp); |
475f6387 | 4172 | case BTRFS_IOC_SCRUB: |
b8e95489 | 4173 | return btrfs_ioctl_scrub(file, argp); |
475f6387 JS |
4174 | case BTRFS_IOC_SCRUB_CANCEL: |
4175 | return btrfs_ioctl_scrub_cancel(root, argp); | |
4176 | case BTRFS_IOC_SCRUB_PROGRESS: | |
4177 | return btrfs_ioctl_scrub_progress(root, argp); | |
c9e9f97b | 4178 | case BTRFS_IOC_BALANCE_V2: |
9ba1f6e4 | 4179 | return btrfs_ioctl_balance(file, argp); |
837d5b6e ID |
4180 | case BTRFS_IOC_BALANCE_CTL: |
4181 | return btrfs_ioctl_balance_ctl(root, arg); | |
19a39dce ID |
4182 | case BTRFS_IOC_BALANCE_PROGRESS: |
4183 | return btrfs_ioctl_balance_progress(root, argp); | |
8ea05e3a AB |
4184 | case BTRFS_IOC_SET_RECEIVED_SUBVOL: |
4185 | return btrfs_ioctl_set_received_subvol(file, argp); | |
31db9f7c AB |
4186 | case BTRFS_IOC_SEND: |
4187 | return btrfs_ioctl_send(file, argp); | |
c11d2c23 | 4188 | case BTRFS_IOC_GET_DEV_STATS: |
b27f7c0c | 4189 | return btrfs_ioctl_get_dev_stats(root, argp); |
5d13a37b | 4190 | case BTRFS_IOC_QUOTA_CTL: |
905b0dda | 4191 | return btrfs_ioctl_quota_ctl(file, argp); |
5d13a37b | 4192 | case BTRFS_IOC_QGROUP_ASSIGN: |
905b0dda | 4193 | return btrfs_ioctl_qgroup_assign(file, argp); |
5d13a37b | 4194 | case BTRFS_IOC_QGROUP_CREATE: |
905b0dda | 4195 | return btrfs_ioctl_qgroup_create(file, argp); |
5d13a37b | 4196 | case BTRFS_IOC_QGROUP_LIMIT: |
905b0dda | 4197 | return btrfs_ioctl_qgroup_limit(file, argp); |
2f232036 JS |
4198 | case BTRFS_IOC_QUOTA_RESCAN: |
4199 | return btrfs_ioctl_quota_rescan(file, argp); | |
4200 | case BTRFS_IOC_QUOTA_RESCAN_STATUS: | |
4201 | return btrfs_ioctl_quota_rescan_status(file, argp); | |
57254b6e JS |
4202 | case BTRFS_IOC_QUOTA_RESCAN_WAIT: |
4203 | return btrfs_ioctl_quota_rescan_wait(file, argp); | |
3f6bcfbd SB |
4204 | case BTRFS_IOC_DEV_REPLACE: |
4205 | return btrfs_ioctl_dev_replace(root, argp); | |
867ab667 | 4206 | case BTRFS_IOC_GET_FSLABEL: |
4207 | return btrfs_ioctl_get_fslabel(file, argp); | |
a8bfd4ab | 4208 | case BTRFS_IOC_SET_FSLABEL: |
4209 | return btrfs_ioctl_set_fslabel(file, argp); | |
f46b5a66 CH |
4210 | } |
4211 | ||
4212 | return -ENOTTY; | |
4213 | } |