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 | ||
3de4586c | 558 | pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS); |
a22285a6 YZ |
559 | if (!pending_snapshot) |
560 | return -ENOMEM; | |
561 | ||
66d8f3dd MX |
562 | btrfs_init_block_rsv(&pending_snapshot->block_rsv, |
563 | BTRFS_BLOCK_RSV_TEMP); | |
d5c12070 MX |
564 | /* |
565 | * 1 - parent dir inode | |
566 | * 2 - dir entries | |
567 | * 1 - root item | |
568 | * 2 - root ref/backref | |
569 | * 1 - root of snapshot | |
570 | */ | |
571 | ret = btrfs_subvolume_reserve_metadata(BTRFS_I(dir)->root, | |
572 | &pending_snapshot->block_rsv, 7, | |
573 | &pending_snapshot->qgroup_reserved); | |
574 | if (ret) | |
575 | goto out; | |
576 | ||
3de4586c | 577 | pending_snapshot->dentry = dentry; |
f46b5a66 | 578 | pending_snapshot->root = root; |
b83cc969 | 579 | pending_snapshot->readonly = readonly; |
e9662f70 | 580 | pending_snapshot->dir = dir; |
8696c533 | 581 | pending_snapshot->inherit = inherit; |
a22285a6 | 582 | |
d5c12070 | 583 | trans = btrfs_start_transaction(root, 0); |
a22285a6 YZ |
584 | if (IS_ERR(trans)) { |
585 | ret = PTR_ERR(trans); | |
586 | goto fail; | |
587 | } | |
588 | ||
8351583e | 589 | spin_lock(&root->fs_info->trans_lock); |
f46b5a66 CH |
590 | list_add(&pending_snapshot->list, |
591 | &trans->transaction->pending_snapshots); | |
8351583e | 592 | spin_unlock(&root->fs_info->trans_lock); |
72fd032e SW |
593 | if (async_transid) { |
594 | *async_transid = trans->transid; | |
595 | ret = btrfs_commit_transaction_async(trans, | |
596 | root->fs_info->extent_root, 1); | |
00d71c9c MX |
597 | if (ret) |
598 | ret = btrfs_commit_transaction(trans, root); | |
72fd032e SW |
599 | } else { |
600 | ret = btrfs_commit_transaction(trans, | |
601 | root->fs_info->extent_root); | |
602 | } | |
aec8030a | 603 | if (ret) |
c37b2b62 | 604 | goto fail; |
a22285a6 YZ |
605 | |
606 | ret = pending_snapshot->error; | |
607 | if (ret) | |
608 | goto fail; | |
609 | ||
66b4ffd1 JB |
610 | ret = btrfs_orphan_cleanup(pending_snapshot->snap); |
611 | if (ret) | |
612 | goto fail; | |
f46b5a66 | 613 | |
2fbe8c8a | 614 | inode = btrfs_lookup_dentry(dentry->d_parent->d_inode, dentry); |
2e4bfab9 YZ |
615 | if (IS_ERR(inode)) { |
616 | ret = PTR_ERR(inode); | |
617 | goto fail; | |
618 | } | |
619 | BUG_ON(!inode); | |
620 | d_instantiate(dentry, inode); | |
621 | ret = 0; | |
622 | fail: | |
d5c12070 MX |
623 | btrfs_subvolume_release_metadata(BTRFS_I(dir)->root, |
624 | &pending_snapshot->block_rsv, | |
625 | pending_snapshot->qgroup_reserved); | |
626 | out: | |
a22285a6 | 627 | kfree(pending_snapshot); |
f46b5a66 CH |
628 | return ret; |
629 | } | |
630 | ||
4260f7c7 SW |
631 | /* copy of check_sticky in fs/namei.c() |
632 | * It's inline, so penalty for filesystems that don't use sticky bit is | |
633 | * minimal. | |
634 | */ | |
635 | static inline int btrfs_check_sticky(struct inode *dir, struct inode *inode) | |
636 | { | |
2f2f43d3 | 637 | kuid_t fsuid = current_fsuid(); |
4260f7c7 SW |
638 | |
639 | if (!(dir->i_mode & S_ISVTX)) | |
640 | return 0; | |
2f2f43d3 | 641 | if (uid_eq(inode->i_uid, fsuid)) |
4260f7c7 | 642 | return 0; |
2f2f43d3 | 643 | if (uid_eq(dir->i_uid, fsuid)) |
4260f7c7 SW |
644 | return 0; |
645 | return !capable(CAP_FOWNER); | |
646 | } | |
647 | ||
648 | /* copy of may_delete in fs/namei.c() | |
649 | * Check whether we can remove a link victim from directory dir, check | |
650 | * whether the type of victim is right. | |
651 | * 1. We can't do it if dir is read-only (done in permission()) | |
652 | * 2. We should have write and exec permissions on dir | |
653 | * 3. We can't remove anything from append-only dir | |
654 | * 4. We can't do anything with immutable dir (done in permission()) | |
655 | * 5. If the sticky bit on dir is set we should either | |
656 | * a. be owner of dir, or | |
657 | * b. be owner of victim, or | |
658 | * c. have CAP_FOWNER capability | |
659 | * 6. If the victim is append-only or immutable we can't do antyhing with | |
660 | * links pointing to it. | |
661 | * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR. | |
662 | * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR. | |
663 | * 9. We can't remove a root or mountpoint. | |
664 | * 10. We don't allow removal of NFS sillyrenamed files; it's handled by | |
665 | * nfs_async_unlink(). | |
666 | */ | |
667 | ||
668 | static int btrfs_may_delete(struct inode *dir,struct dentry *victim,int isdir) | |
669 | { | |
670 | int error; | |
671 | ||
672 | if (!victim->d_inode) | |
673 | return -ENOENT; | |
674 | ||
675 | BUG_ON(victim->d_parent->d_inode != dir); | |
4fa6b5ec | 676 | audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE); |
4260f7c7 SW |
677 | |
678 | error = inode_permission(dir, MAY_WRITE | MAY_EXEC); | |
679 | if (error) | |
680 | return error; | |
681 | if (IS_APPEND(dir)) | |
682 | return -EPERM; | |
683 | if (btrfs_check_sticky(dir, victim->d_inode)|| | |
684 | IS_APPEND(victim->d_inode)|| | |
685 | IS_IMMUTABLE(victim->d_inode) || IS_SWAPFILE(victim->d_inode)) | |
686 | return -EPERM; | |
687 | if (isdir) { | |
688 | if (!S_ISDIR(victim->d_inode->i_mode)) | |
689 | return -ENOTDIR; | |
690 | if (IS_ROOT(victim)) | |
691 | return -EBUSY; | |
692 | } else if (S_ISDIR(victim->d_inode->i_mode)) | |
693 | return -EISDIR; | |
694 | if (IS_DEADDIR(dir)) | |
695 | return -ENOENT; | |
696 | if (victim->d_flags & DCACHE_NFSFS_RENAMED) | |
697 | return -EBUSY; | |
698 | return 0; | |
699 | } | |
700 | ||
cb8e7090 CH |
701 | /* copy of may_create in fs/namei.c() */ |
702 | static inline int btrfs_may_create(struct inode *dir, struct dentry *child) | |
703 | { | |
704 | if (child->d_inode) | |
705 | return -EEXIST; | |
706 | if (IS_DEADDIR(dir)) | |
707 | return -ENOENT; | |
708 | return inode_permission(dir, MAY_WRITE | MAY_EXEC); | |
709 | } | |
710 | ||
711 | /* | |
712 | * Create a new subvolume below @parent. This is largely modeled after | |
713 | * sys_mkdirat and vfs_mkdir, but we only do a single component lookup | |
714 | * inside this filesystem so it's quite a bit simpler. | |
715 | */ | |
76dda93c YZ |
716 | static noinline int btrfs_mksubvol(struct path *parent, |
717 | char *name, int namelen, | |
72fd032e | 718 | struct btrfs_root *snap_src, |
6f72c7e2 | 719 | u64 *async_transid, bool readonly, |
8696c533 | 720 | struct btrfs_qgroup_inherit *inherit) |
cb8e7090 | 721 | { |
76dda93c | 722 | struct inode *dir = parent->dentry->d_inode; |
cb8e7090 CH |
723 | struct dentry *dentry; |
724 | int error; | |
725 | ||
5c50c9b8 DS |
726 | error = mutex_lock_killable_nested(&dir->i_mutex, I_MUTEX_PARENT); |
727 | if (error == -EINTR) | |
728 | return error; | |
cb8e7090 CH |
729 | |
730 | dentry = lookup_one_len(name, parent->dentry, namelen); | |
731 | error = PTR_ERR(dentry); | |
732 | if (IS_ERR(dentry)) | |
733 | goto out_unlock; | |
734 | ||
735 | error = -EEXIST; | |
736 | if (dentry->d_inode) | |
737 | goto out_dput; | |
738 | ||
76dda93c | 739 | error = btrfs_may_create(dir, dentry); |
cb8e7090 | 740 | if (error) |
a874a63e | 741 | goto out_dput; |
cb8e7090 | 742 | |
9c52057c CM |
743 | /* |
744 | * even if this name doesn't exist, we may get hash collisions. | |
745 | * check for them now when we can safely fail | |
746 | */ | |
747 | error = btrfs_check_dir_item_collision(BTRFS_I(dir)->root, | |
748 | dir->i_ino, name, | |
749 | namelen); | |
750 | if (error) | |
751 | goto out_dput; | |
752 | ||
76dda93c YZ |
753 | down_read(&BTRFS_I(dir)->root->fs_info->subvol_sem); |
754 | ||
755 | if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0) | |
756 | goto out_up_read; | |
757 | ||
3de4586c | 758 | if (snap_src) { |
e9662f70 | 759 | error = create_snapshot(snap_src, dir, dentry, name, namelen, |
6f72c7e2 | 760 | async_transid, readonly, inherit); |
3de4586c | 761 | } else { |
d5c12070 MX |
762 | error = create_subvol(dir, dentry, name, namelen, |
763 | async_transid, inherit); | |
3de4586c | 764 | } |
76dda93c YZ |
765 | if (!error) |
766 | fsnotify_mkdir(dir, dentry); | |
767 | out_up_read: | |
768 | up_read(&BTRFS_I(dir)->root->fs_info->subvol_sem); | |
cb8e7090 CH |
769 | out_dput: |
770 | dput(dentry); | |
771 | out_unlock: | |
76dda93c | 772 | mutex_unlock(&dir->i_mutex); |
cb8e7090 CH |
773 | return error; |
774 | } | |
775 | ||
4cb5300b CM |
776 | /* |
777 | * When we're defragging a range, we don't want to kick it off again | |
778 | * if it is really just waiting for delalloc to send it down. | |
779 | * If we find a nice big extent or delalloc range for the bytes in the | |
780 | * file you want to defrag, we return 0 to let you know to skip this | |
781 | * part of the file | |
782 | */ | |
783 | static int check_defrag_in_cache(struct inode *inode, u64 offset, int thresh) | |
784 | { | |
785 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; | |
786 | struct extent_map *em = NULL; | |
787 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
788 | u64 end; | |
789 | ||
790 | read_lock(&em_tree->lock); | |
791 | em = lookup_extent_mapping(em_tree, offset, PAGE_CACHE_SIZE); | |
792 | read_unlock(&em_tree->lock); | |
793 | ||
794 | if (em) { | |
795 | end = extent_map_end(em); | |
796 | free_extent_map(em); | |
797 | if (end - offset > thresh) | |
798 | return 0; | |
799 | } | |
800 | /* if we already have a nice delalloc here, just stop */ | |
801 | thresh /= 2; | |
802 | end = count_range_bits(io_tree, &offset, offset + thresh, | |
803 | thresh, EXTENT_DELALLOC, 1); | |
804 | if (end >= thresh) | |
805 | return 0; | |
806 | return 1; | |
807 | } | |
808 | ||
809 | /* | |
810 | * helper function to walk through a file and find extents | |
811 | * newer than a specific transid, and smaller than thresh. | |
812 | * | |
813 | * This is used by the defragging code to find new and small | |
814 | * extents | |
815 | */ | |
816 | static int find_new_extents(struct btrfs_root *root, | |
817 | struct inode *inode, u64 newer_than, | |
818 | u64 *off, int thresh) | |
819 | { | |
820 | struct btrfs_path *path; | |
821 | struct btrfs_key min_key; | |
822 | struct btrfs_key max_key; | |
823 | struct extent_buffer *leaf; | |
824 | struct btrfs_file_extent_item *extent; | |
825 | int type; | |
826 | int ret; | |
a4689d2b | 827 | u64 ino = btrfs_ino(inode); |
4cb5300b CM |
828 | |
829 | path = btrfs_alloc_path(); | |
830 | if (!path) | |
831 | return -ENOMEM; | |
832 | ||
a4689d2b | 833 | min_key.objectid = ino; |
4cb5300b CM |
834 | min_key.type = BTRFS_EXTENT_DATA_KEY; |
835 | min_key.offset = *off; | |
836 | ||
a4689d2b | 837 | max_key.objectid = ino; |
4cb5300b CM |
838 | max_key.type = (u8)-1; |
839 | max_key.offset = (u64)-1; | |
840 | ||
841 | path->keep_locks = 1; | |
842 | ||
843 | while(1) { | |
844 | ret = btrfs_search_forward(root, &min_key, &max_key, | |
de78b51a | 845 | path, newer_than); |
4cb5300b CM |
846 | if (ret != 0) |
847 | goto none; | |
a4689d2b | 848 | if (min_key.objectid != ino) |
4cb5300b CM |
849 | goto none; |
850 | if (min_key.type != BTRFS_EXTENT_DATA_KEY) | |
851 | goto none; | |
852 | ||
853 | leaf = path->nodes[0]; | |
854 | extent = btrfs_item_ptr(leaf, path->slots[0], | |
855 | struct btrfs_file_extent_item); | |
856 | ||
857 | type = btrfs_file_extent_type(leaf, extent); | |
858 | if (type == BTRFS_FILE_EXTENT_REG && | |
859 | btrfs_file_extent_num_bytes(leaf, extent) < thresh && | |
860 | check_defrag_in_cache(inode, min_key.offset, thresh)) { | |
861 | *off = min_key.offset; | |
862 | btrfs_free_path(path); | |
863 | return 0; | |
864 | } | |
865 | ||
866 | if (min_key.offset == (u64)-1) | |
867 | goto none; | |
868 | ||
869 | min_key.offset++; | |
870 | btrfs_release_path(path); | |
871 | } | |
872 | none: | |
873 | btrfs_free_path(path); | |
874 | return -ENOENT; | |
875 | } | |
876 | ||
6c282eb4 | 877 | static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start) |
17ce6ef8 LB |
878 | { |
879 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
6c282eb4 LZ |
880 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; |
881 | struct extent_map *em; | |
882 | u64 len = PAGE_CACHE_SIZE; | |
17ce6ef8 | 883 | |
6c282eb4 LZ |
884 | /* |
885 | * hopefully we have this extent in the tree already, try without | |
886 | * the full extent lock | |
887 | */ | |
17ce6ef8 | 888 | read_lock(&em_tree->lock); |
6c282eb4 | 889 | em = lookup_extent_mapping(em_tree, start, len); |
17ce6ef8 LB |
890 | read_unlock(&em_tree->lock); |
891 | ||
6c282eb4 LZ |
892 | if (!em) { |
893 | /* get the big lock and read metadata off disk */ | |
894 | lock_extent(io_tree, start, start + len - 1); | |
895 | em = btrfs_get_extent(inode, NULL, 0, start, len, 0); | |
896 | unlock_extent(io_tree, start, start + len - 1); | |
897 | ||
898 | if (IS_ERR(em)) | |
899 | return NULL; | |
900 | } | |
901 | ||
902 | return em; | |
903 | } | |
17ce6ef8 | 904 | |
6c282eb4 LZ |
905 | static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em) |
906 | { | |
907 | struct extent_map *next; | |
908 | bool ret = true; | |
909 | ||
910 | /* this is the last extent */ | |
911 | if (em->start + em->len >= i_size_read(inode)) | |
912 | return false; | |
913 | ||
914 | next = defrag_lookup_extent(inode, em->start + em->len); | |
915 | if (!next || next->block_start >= EXTENT_MAP_LAST_BYTE) | |
916 | ret = false; | |
917 | ||
918 | free_extent_map(next); | |
17ce6ef8 LB |
919 | return ret; |
920 | } | |
921 | ||
6c282eb4 | 922 | static int should_defrag_range(struct inode *inode, u64 start, int thresh, |
a43a2111 AM |
923 | u64 *last_len, u64 *skip, u64 *defrag_end, |
924 | int compress) | |
940100a4 | 925 | { |
6c282eb4 | 926 | struct extent_map *em; |
940100a4 | 927 | int ret = 1; |
6c282eb4 | 928 | bool next_mergeable = true; |
940100a4 CM |
929 | |
930 | /* | |
008873ea | 931 | * make sure that once we start defragging an extent, we keep on |
940100a4 CM |
932 | * defragging it |
933 | */ | |
934 | if (start < *defrag_end) | |
935 | return 1; | |
936 | ||
937 | *skip = 0; | |
938 | ||
6c282eb4 LZ |
939 | em = defrag_lookup_extent(inode, start); |
940 | if (!em) | |
941 | return 0; | |
940100a4 CM |
942 | |
943 | /* this will cover holes, and inline extents */ | |
17ce6ef8 | 944 | if (em->block_start >= EXTENT_MAP_LAST_BYTE) { |
940100a4 | 945 | ret = 0; |
17ce6ef8 LB |
946 | goto out; |
947 | } | |
948 | ||
6c282eb4 | 949 | next_mergeable = defrag_check_next_extent(inode, em); |
940100a4 CM |
950 | |
951 | /* | |
6c282eb4 LZ |
952 | * we hit a real extent, if it is big or the next extent is not a |
953 | * real extent, don't bother defragging it | |
940100a4 | 954 | */ |
a43a2111 | 955 | if (!compress && (*last_len == 0 || *last_len >= thresh) && |
6c282eb4 | 956 | (em->len >= thresh || !next_mergeable)) |
940100a4 | 957 | ret = 0; |
17ce6ef8 | 958 | out: |
940100a4 CM |
959 | /* |
960 | * last_len ends up being a counter of how many bytes we've defragged. | |
961 | * every time we choose not to defrag an extent, we reset *last_len | |
962 | * so that the next tiny extent will force a defrag. | |
963 | * | |
964 | * The end result of this is that tiny extents before a single big | |
965 | * extent will force at least part of that big extent to be defragged. | |
966 | */ | |
967 | if (ret) { | |
940100a4 CM |
968 | *defrag_end = extent_map_end(em); |
969 | } else { | |
970 | *last_len = 0; | |
971 | *skip = extent_map_end(em); | |
972 | *defrag_end = 0; | |
973 | } | |
974 | ||
975 | free_extent_map(em); | |
976 | return ret; | |
977 | } | |
978 | ||
4cb5300b CM |
979 | /* |
980 | * it doesn't do much good to defrag one or two pages | |
981 | * at a time. This pulls in a nice chunk of pages | |
982 | * to COW and defrag. | |
983 | * | |
984 | * It also makes sure the delalloc code has enough | |
985 | * dirty data to avoid making new small extents as part | |
986 | * of the defrag | |
987 | * | |
988 | * It's a good idea to start RA on this range | |
989 | * before calling this. | |
990 | */ | |
991 | static int cluster_pages_for_defrag(struct inode *inode, | |
992 | struct page **pages, | |
993 | unsigned long start_index, | |
994 | int num_pages) | |
f46b5a66 | 995 | { |
4cb5300b CM |
996 | unsigned long file_end; |
997 | u64 isize = i_size_read(inode); | |
998 | u64 page_start; | |
999 | u64 page_end; | |
1f12bd06 | 1000 | u64 page_cnt; |
4cb5300b CM |
1001 | int ret; |
1002 | int i; | |
1003 | int i_done; | |
3eaa2885 | 1004 | struct btrfs_ordered_extent *ordered; |
4cb5300b | 1005 | struct extent_state *cached_state = NULL; |
600a45e1 | 1006 | struct extent_io_tree *tree; |
3b16a4e3 | 1007 | gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping); |
4cb5300b | 1008 | |
4cb5300b | 1009 | file_end = (isize - 1) >> PAGE_CACHE_SHIFT; |
1f12bd06 LB |
1010 | if (!isize || start_index > file_end) |
1011 | return 0; | |
1012 | ||
1013 | page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1); | |
4cb5300b CM |
1014 | |
1015 | ret = btrfs_delalloc_reserve_space(inode, | |
1f12bd06 | 1016 | page_cnt << PAGE_CACHE_SHIFT); |
4cb5300b CM |
1017 | if (ret) |
1018 | return ret; | |
4cb5300b | 1019 | i_done = 0; |
600a45e1 | 1020 | tree = &BTRFS_I(inode)->io_tree; |
4cb5300b CM |
1021 | |
1022 | /* step one, lock all the pages */ | |
1f12bd06 | 1023 | for (i = 0; i < page_cnt; i++) { |
4cb5300b | 1024 | struct page *page; |
600a45e1 | 1025 | again: |
a94733d0 | 1026 | page = find_or_create_page(inode->i_mapping, |
600a45e1 | 1027 | start_index + i, mask); |
4cb5300b CM |
1028 | if (!page) |
1029 | break; | |
1030 | ||
600a45e1 MX |
1031 | page_start = page_offset(page); |
1032 | page_end = page_start + PAGE_CACHE_SIZE - 1; | |
1033 | while (1) { | |
d0082371 | 1034 | lock_extent(tree, page_start, page_end); |
600a45e1 MX |
1035 | ordered = btrfs_lookup_ordered_extent(inode, |
1036 | page_start); | |
d0082371 | 1037 | unlock_extent(tree, page_start, page_end); |
600a45e1 MX |
1038 | if (!ordered) |
1039 | break; | |
1040 | ||
1041 | unlock_page(page); | |
1042 | btrfs_start_ordered_extent(inode, ordered, 1); | |
1043 | btrfs_put_ordered_extent(ordered); | |
1044 | lock_page(page); | |
1f12bd06 LB |
1045 | /* |
1046 | * we unlocked the page above, so we need check if | |
1047 | * it was released or not. | |
1048 | */ | |
1049 | if (page->mapping != inode->i_mapping) { | |
1050 | unlock_page(page); | |
1051 | page_cache_release(page); | |
1052 | goto again; | |
1053 | } | |
600a45e1 MX |
1054 | } |
1055 | ||
4cb5300b CM |
1056 | if (!PageUptodate(page)) { |
1057 | btrfs_readpage(NULL, page); | |
1058 | lock_page(page); | |
1059 | if (!PageUptodate(page)) { | |
1060 | unlock_page(page); | |
1061 | page_cache_release(page); | |
1062 | ret = -EIO; | |
1063 | break; | |
1064 | } | |
1065 | } | |
600a45e1 | 1066 | |
600a45e1 MX |
1067 | if (page->mapping != inode->i_mapping) { |
1068 | unlock_page(page); | |
1069 | page_cache_release(page); | |
1070 | goto again; | |
1071 | } | |
1072 | ||
4cb5300b CM |
1073 | pages[i] = page; |
1074 | i_done++; | |
1075 | } | |
1076 | if (!i_done || ret) | |
1077 | goto out; | |
1078 | ||
1079 | if (!(inode->i_sb->s_flags & MS_ACTIVE)) | |
1080 | goto out; | |
1081 | ||
1082 | /* | |
1083 | * so now we have a nice long stream of locked | |
1084 | * and up to date pages, lets wait on them | |
1085 | */ | |
1086 | for (i = 0; i < i_done; i++) | |
1087 | wait_on_page_writeback(pages[i]); | |
1088 | ||
1089 | page_start = page_offset(pages[0]); | |
1090 | page_end = page_offset(pages[i_done - 1]) + PAGE_CACHE_SIZE; | |
1091 | ||
1092 | lock_extent_bits(&BTRFS_I(inode)->io_tree, | |
d0082371 | 1093 | page_start, page_end - 1, 0, &cached_state); |
4cb5300b CM |
1094 | clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, |
1095 | page_end - 1, EXTENT_DIRTY | EXTENT_DELALLOC | | |
9e8a4a8b LB |
1096 | EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 0, 0, |
1097 | &cached_state, GFP_NOFS); | |
4cb5300b | 1098 | |
1f12bd06 | 1099 | if (i_done != page_cnt) { |
9e0baf60 JB |
1100 | spin_lock(&BTRFS_I(inode)->lock); |
1101 | BTRFS_I(inode)->outstanding_extents++; | |
1102 | spin_unlock(&BTRFS_I(inode)->lock); | |
4cb5300b | 1103 | btrfs_delalloc_release_space(inode, |
1f12bd06 | 1104 | (page_cnt - i_done) << PAGE_CACHE_SHIFT); |
4cb5300b CM |
1105 | } |
1106 | ||
1107 | ||
9e8a4a8b LB |
1108 | set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1, |
1109 | &cached_state, GFP_NOFS); | |
4cb5300b CM |
1110 | |
1111 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, | |
1112 | page_start, page_end - 1, &cached_state, | |
1113 | GFP_NOFS); | |
1114 | ||
1115 | for (i = 0; i < i_done; i++) { | |
1116 | clear_page_dirty_for_io(pages[i]); | |
1117 | ClearPageChecked(pages[i]); | |
1118 | set_page_extent_mapped(pages[i]); | |
1119 | set_page_dirty(pages[i]); | |
1120 | unlock_page(pages[i]); | |
1121 | page_cache_release(pages[i]); | |
1122 | } | |
1123 | return i_done; | |
1124 | out: | |
1125 | for (i = 0; i < i_done; i++) { | |
1126 | unlock_page(pages[i]); | |
1127 | page_cache_release(pages[i]); | |
1128 | } | |
1f12bd06 | 1129 | btrfs_delalloc_release_space(inode, page_cnt << PAGE_CACHE_SHIFT); |
4cb5300b CM |
1130 | return ret; |
1131 | ||
1132 | } | |
1133 | ||
1134 | int btrfs_defrag_file(struct inode *inode, struct file *file, | |
1135 | struct btrfs_ioctl_defrag_range_args *range, | |
1136 | u64 newer_than, unsigned long max_to_defrag) | |
1137 | { | |
1138 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4cb5300b | 1139 | struct file_ra_state *ra = NULL; |
f46b5a66 | 1140 | unsigned long last_index; |
151a31b2 | 1141 | u64 isize = i_size_read(inode); |
940100a4 CM |
1142 | u64 last_len = 0; |
1143 | u64 skip = 0; | |
1144 | u64 defrag_end = 0; | |
4cb5300b | 1145 | u64 newer_off = range->start; |
f46b5a66 | 1146 | unsigned long i; |
008873ea | 1147 | unsigned long ra_index = 0; |
f46b5a66 | 1148 | int ret; |
4cb5300b | 1149 | int defrag_count = 0; |
1a419d85 | 1150 | int compress_type = BTRFS_COMPRESS_ZLIB; |
4cb5300b | 1151 | int extent_thresh = range->extent_thresh; |
008873ea LZ |
1152 | int max_cluster = (256 * 1024) >> PAGE_CACHE_SHIFT; |
1153 | int cluster = max_cluster; | |
4cb5300b CM |
1154 | u64 new_align = ~((u64)128 * 1024 - 1); |
1155 | struct page **pages = NULL; | |
1156 | ||
0abd5b17 LB |
1157 | if (isize == 0) |
1158 | return 0; | |
1159 | ||
1160 | if (range->start >= isize) | |
1161 | return -EINVAL; | |
1a419d85 LZ |
1162 | |
1163 | if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS) { | |
1164 | if (range->compress_type > BTRFS_COMPRESS_TYPES) | |
1165 | return -EINVAL; | |
1166 | if (range->compress_type) | |
1167 | compress_type = range->compress_type; | |
1168 | } | |
f46b5a66 | 1169 | |
0abd5b17 LB |
1170 | if (extent_thresh == 0) |
1171 | extent_thresh = 256 * 1024; | |
940100a4 | 1172 | |
4cb5300b CM |
1173 | /* |
1174 | * if we were not given a file, allocate a readahead | |
1175 | * context | |
1176 | */ | |
1177 | if (!file) { | |
1178 | ra = kzalloc(sizeof(*ra), GFP_NOFS); | |
1179 | if (!ra) | |
1180 | return -ENOMEM; | |
1181 | file_ra_state_init(ra, inode->i_mapping); | |
1182 | } else { | |
1183 | ra = &file->f_ra; | |
1184 | } | |
1185 | ||
008873ea | 1186 | pages = kmalloc(sizeof(struct page *) * max_cluster, |
4cb5300b CM |
1187 | GFP_NOFS); |
1188 | if (!pages) { | |
1189 | ret = -ENOMEM; | |
1190 | goto out_ra; | |
1191 | } | |
1192 | ||
1193 | /* find the last page to defrag */ | |
1e701a32 | 1194 | if (range->start + range->len > range->start) { |
151a31b2 | 1195 | last_index = min_t(u64, isize - 1, |
1e701a32 CM |
1196 | range->start + range->len - 1) >> PAGE_CACHE_SHIFT; |
1197 | } else { | |
151a31b2 | 1198 | last_index = (isize - 1) >> PAGE_CACHE_SHIFT; |
1e701a32 CM |
1199 | } |
1200 | ||
4cb5300b CM |
1201 | if (newer_than) { |
1202 | ret = find_new_extents(root, inode, newer_than, | |
1203 | &newer_off, 64 * 1024); | |
1204 | if (!ret) { | |
1205 | range->start = newer_off; | |
1206 | /* | |
1207 | * we always align our defrag to help keep | |
1208 | * the extents in the file evenly spaced | |
1209 | */ | |
1210 | i = (newer_off & new_align) >> PAGE_CACHE_SHIFT; | |
4cb5300b CM |
1211 | } else |
1212 | goto out_ra; | |
1213 | } else { | |
1214 | i = range->start >> PAGE_CACHE_SHIFT; | |
1215 | } | |
1216 | if (!max_to_defrag) | |
7ec31b54 | 1217 | max_to_defrag = last_index + 1; |
4cb5300b | 1218 | |
2a0f7f57 LZ |
1219 | /* |
1220 | * make writeback starts from i, so the defrag range can be | |
1221 | * written sequentially. | |
1222 | */ | |
1223 | if (i < inode->i_mapping->writeback_index) | |
1224 | inode->i_mapping->writeback_index = i; | |
1225 | ||
f7f43cc8 CM |
1226 | while (i <= last_index && defrag_count < max_to_defrag && |
1227 | (i < (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >> | |
1228 | PAGE_CACHE_SHIFT)) { | |
4cb5300b CM |
1229 | /* |
1230 | * make sure we stop running if someone unmounts | |
1231 | * the FS | |
1232 | */ | |
1233 | if (!(inode->i_sb->s_flags & MS_ACTIVE)) | |
1234 | break; | |
1235 | ||
210549eb DS |
1236 | if (btrfs_defrag_cancelled(root->fs_info)) { |
1237 | printk(KERN_DEBUG "btrfs: defrag_file cancelled\n"); | |
1238 | ret = -EAGAIN; | |
1239 | break; | |
1240 | } | |
1241 | ||
66c26892 | 1242 | if (!should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT, |
6c282eb4 | 1243 | extent_thresh, &last_len, &skip, |
a43a2111 AM |
1244 | &defrag_end, range->flags & |
1245 | BTRFS_DEFRAG_RANGE_COMPRESS)) { | |
940100a4 CM |
1246 | unsigned long next; |
1247 | /* | |
1248 | * the should_defrag function tells us how much to skip | |
1249 | * bump our counter by the suggested amount | |
1250 | */ | |
1251 | next = (skip + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; | |
1252 | i = max(i + 1, next); | |
1253 | continue; | |
1254 | } | |
008873ea LZ |
1255 | |
1256 | if (!newer_than) { | |
1257 | cluster = (PAGE_CACHE_ALIGN(defrag_end) >> | |
1258 | PAGE_CACHE_SHIFT) - i; | |
1259 | cluster = min(cluster, max_cluster); | |
1260 | } else { | |
1261 | cluster = max_cluster; | |
1262 | } | |
1263 | ||
1e701a32 | 1264 | if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS) |
1a419d85 | 1265 | BTRFS_I(inode)->force_compress = compress_type; |
940100a4 | 1266 | |
008873ea LZ |
1267 | if (i + cluster > ra_index) { |
1268 | ra_index = max(i, ra_index); | |
1269 | btrfs_force_ra(inode->i_mapping, ra, file, ra_index, | |
1270 | cluster); | |
1271 | ra_index += max_cluster; | |
1272 | } | |
940100a4 | 1273 | |
ecb8bea8 | 1274 | mutex_lock(&inode->i_mutex); |
008873ea | 1275 | ret = cluster_pages_for_defrag(inode, pages, i, cluster); |
ecb8bea8 LB |
1276 | if (ret < 0) { |
1277 | mutex_unlock(&inode->i_mutex); | |
4cb5300b | 1278 | goto out_ra; |
ecb8bea8 | 1279 | } |
4cb5300b CM |
1280 | |
1281 | defrag_count += ret; | |
d0e1d66b | 1282 | balance_dirty_pages_ratelimited(inode->i_mapping); |
ecb8bea8 | 1283 | mutex_unlock(&inode->i_mutex); |
4cb5300b CM |
1284 | |
1285 | if (newer_than) { | |
1286 | if (newer_off == (u64)-1) | |
1287 | break; | |
1288 | ||
e1f041e1 LB |
1289 | if (ret > 0) |
1290 | i += ret; | |
1291 | ||
4cb5300b CM |
1292 | newer_off = max(newer_off + 1, |
1293 | (u64)i << PAGE_CACHE_SHIFT); | |
1294 | ||
1295 | ret = find_new_extents(root, inode, | |
1296 | newer_than, &newer_off, | |
1297 | 64 * 1024); | |
1298 | if (!ret) { | |
1299 | range->start = newer_off; | |
1300 | i = (newer_off & new_align) >> PAGE_CACHE_SHIFT; | |
4cb5300b CM |
1301 | } else { |
1302 | break; | |
f46b5a66 | 1303 | } |
4cb5300b | 1304 | } else { |
008873ea | 1305 | if (ret > 0) { |
cbcc8326 | 1306 | i += ret; |
008873ea LZ |
1307 | last_len += ret << PAGE_CACHE_SHIFT; |
1308 | } else { | |
cbcc8326 | 1309 | i++; |
008873ea LZ |
1310 | last_len = 0; |
1311 | } | |
f46b5a66 | 1312 | } |
f46b5a66 CH |
1313 | } |
1314 | ||
1e701a32 CM |
1315 | if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO)) |
1316 | filemap_flush(inode->i_mapping); | |
1317 | ||
1318 | if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { | |
1319 | /* the filemap_flush will queue IO into the worker threads, but | |
1320 | * we have to make sure the IO is actually started and that | |
1321 | * ordered extents get created before we return | |
1322 | */ | |
1323 | atomic_inc(&root->fs_info->async_submit_draining); | |
1324 | while (atomic_read(&root->fs_info->nr_async_submits) || | |
1325 | atomic_read(&root->fs_info->async_delalloc_pages)) { | |
1326 | wait_event(root->fs_info->async_submit_wait, | |
1327 | (atomic_read(&root->fs_info->nr_async_submits) == 0 && | |
1328 | atomic_read(&root->fs_info->async_delalloc_pages) == 0)); | |
1329 | } | |
1330 | atomic_dec(&root->fs_info->async_submit_draining); | |
1331 | ||
1332 | mutex_lock(&inode->i_mutex); | |
261507a0 | 1333 | BTRFS_I(inode)->force_compress = BTRFS_COMPRESS_NONE; |
1e701a32 CM |
1334 | mutex_unlock(&inode->i_mutex); |
1335 | } | |
1336 | ||
1a419d85 | 1337 | if (range->compress_type == BTRFS_COMPRESS_LZO) { |
2b0ce2c2 | 1338 | btrfs_set_fs_incompat(root->fs_info, COMPRESS_LZO); |
1a419d85 LZ |
1339 | } |
1340 | ||
60ccf82f | 1341 | ret = defrag_count; |
940100a4 | 1342 | |
4cb5300b CM |
1343 | out_ra: |
1344 | if (!file) | |
1345 | kfree(ra); | |
1346 | kfree(pages); | |
940100a4 | 1347 | return ret; |
f46b5a66 CH |
1348 | } |
1349 | ||
198605a8 | 1350 | static noinline int btrfs_ioctl_resize(struct file *file, |
76dda93c | 1351 | void __user *arg) |
f46b5a66 CH |
1352 | { |
1353 | u64 new_size; | |
1354 | u64 old_size; | |
1355 | u64 devid = 1; | |
496ad9aa | 1356 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
f46b5a66 CH |
1357 | struct btrfs_ioctl_vol_args *vol_args; |
1358 | struct btrfs_trans_handle *trans; | |
1359 | struct btrfs_device *device = NULL; | |
1360 | char *sizestr; | |
1361 | char *devstr = NULL; | |
1362 | int ret = 0; | |
f46b5a66 CH |
1363 | int mod = 0; |
1364 | ||
e441d54d CM |
1365 | if (!capable(CAP_SYS_ADMIN)) |
1366 | return -EPERM; | |
1367 | ||
198605a8 MX |
1368 | ret = mnt_want_write_file(file); |
1369 | if (ret) | |
1370 | return ret; | |
1371 | ||
5ac00add SB |
1372 | if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running, |
1373 | 1)) { | |
1374 | pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n"); | |
97547676 | 1375 | mnt_drop_write_file(file); |
2c0c9da0 | 1376 | return -EINVAL; |
c9e9f97b ID |
1377 | } |
1378 | ||
5ac00add | 1379 | mutex_lock(&root->fs_info->volume_mutex); |
dae7b665 | 1380 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
1381 | if (IS_ERR(vol_args)) { |
1382 | ret = PTR_ERR(vol_args); | |
1383 | goto out; | |
1384 | } | |
5516e595 MF |
1385 | |
1386 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
f46b5a66 | 1387 | |
f46b5a66 CH |
1388 | sizestr = vol_args->name; |
1389 | devstr = strchr(sizestr, ':'); | |
1390 | if (devstr) { | |
1391 | char *end; | |
1392 | sizestr = devstr + 1; | |
1393 | *devstr = '\0'; | |
1394 | devstr = vol_args->name; | |
1395 | devid = simple_strtoull(devstr, &end, 10); | |
dfd79829 MX |
1396 | if (!devid) { |
1397 | ret = -EINVAL; | |
1398 | goto out_free; | |
1399 | } | |
5bb14682 | 1400 | printk(KERN_INFO "btrfs: resizing devid %llu\n", |
21380931 | 1401 | (unsigned long long)devid); |
f46b5a66 | 1402 | } |
dba60f3f | 1403 | |
aa1b8cd4 | 1404 | device = btrfs_find_device(root->fs_info, devid, NULL, NULL); |
f46b5a66 | 1405 | if (!device) { |
5bb14682 | 1406 | printk(KERN_INFO "btrfs: resizer unable to find device %llu\n", |
21380931 | 1407 | (unsigned long long)devid); |
dfd79829 | 1408 | ret = -ENODEV; |
c9e9f97b | 1409 | goto out_free; |
f46b5a66 | 1410 | } |
dba60f3f MX |
1411 | |
1412 | if (!device->writeable) { | |
4e42ae1b | 1413 | printk(KERN_INFO "btrfs: resizer unable to apply on " |
dba60f3f | 1414 | "readonly device %llu\n", |
a8c4a33b | 1415 | (unsigned long long)devid); |
dfd79829 | 1416 | ret = -EPERM; |
4e42ae1b LB |
1417 | goto out_free; |
1418 | } | |
1419 | ||
f46b5a66 CH |
1420 | if (!strcmp(sizestr, "max")) |
1421 | new_size = device->bdev->bd_inode->i_size; | |
1422 | else { | |
1423 | if (sizestr[0] == '-') { | |
1424 | mod = -1; | |
1425 | sizestr++; | |
1426 | } else if (sizestr[0] == '+') { | |
1427 | mod = 1; | |
1428 | sizestr++; | |
1429 | } | |
91748467 | 1430 | new_size = memparse(sizestr, NULL); |
f46b5a66 CH |
1431 | if (new_size == 0) { |
1432 | ret = -EINVAL; | |
c9e9f97b | 1433 | goto out_free; |
f46b5a66 CH |
1434 | } |
1435 | } | |
1436 | ||
63a212ab | 1437 | if (device->is_tgtdev_for_dev_replace) { |
dfd79829 | 1438 | ret = -EPERM; |
63a212ab SB |
1439 | goto out_free; |
1440 | } | |
1441 | ||
f46b5a66 CH |
1442 | old_size = device->total_bytes; |
1443 | ||
1444 | if (mod < 0) { | |
1445 | if (new_size > old_size) { | |
1446 | ret = -EINVAL; | |
c9e9f97b | 1447 | goto out_free; |
f46b5a66 CH |
1448 | } |
1449 | new_size = old_size - new_size; | |
1450 | } else if (mod > 0) { | |
1451 | new_size = old_size + new_size; | |
1452 | } | |
1453 | ||
1454 | if (new_size < 256 * 1024 * 1024) { | |
1455 | ret = -EINVAL; | |
c9e9f97b | 1456 | goto out_free; |
f46b5a66 CH |
1457 | } |
1458 | if (new_size > device->bdev->bd_inode->i_size) { | |
1459 | ret = -EFBIG; | |
c9e9f97b | 1460 | goto out_free; |
f46b5a66 CH |
1461 | } |
1462 | ||
1463 | do_div(new_size, root->sectorsize); | |
1464 | new_size *= root->sectorsize; | |
1465 | ||
606686ee JB |
1466 | printk_in_rcu(KERN_INFO "btrfs: new size for %s is %llu\n", |
1467 | rcu_str_deref(device->name), | |
1468 | (unsigned long long)new_size); | |
f46b5a66 CH |
1469 | |
1470 | if (new_size > old_size) { | |
a22285a6 | 1471 | trans = btrfs_start_transaction(root, 0); |
98d5dc13 TI |
1472 | if (IS_ERR(trans)) { |
1473 | ret = PTR_ERR(trans); | |
c9e9f97b | 1474 | goto out_free; |
98d5dc13 | 1475 | } |
f46b5a66 CH |
1476 | ret = btrfs_grow_device(trans, device, new_size); |
1477 | btrfs_commit_transaction(trans, root); | |
ece7d20e | 1478 | } else if (new_size < old_size) { |
f46b5a66 | 1479 | ret = btrfs_shrink_device(device, new_size); |
0253f40e | 1480 | } /* equal, nothing need to do */ |
f46b5a66 | 1481 | |
c9e9f97b | 1482 | out_free: |
f46b5a66 | 1483 | kfree(vol_args); |
c9e9f97b ID |
1484 | out: |
1485 | mutex_unlock(&root->fs_info->volume_mutex); | |
5ac00add | 1486 | atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0); |
18f39c41 | 1487 | mnt_drop_write_file(file); |
f46b5a66 CH |
1488 | return ret; |
1489 | } | |
1490 | ||
72fd032e | 1491 | static noinline int btrfs_ioctl_snap_create_transid(struct file *file, |
6f72c7e2 AJ |
1492 | char *name, unsigned long fd, int subvol, |
1493 | u64 *transid, bool readonly, | |
8696c533 | 1494 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 1495 | { |
f46b5a66 | 1496 | int namelen; |
3de4586c | 1497 | int ret = 0; |
f46b5a66 | 1498 | |
879149a0 JM |
1499 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1500 | return -ENOTDIR; | |
1501 | ||
a874a63e LB |
1502 | ret = mnt_want_write_file(file); |
1503 | if (ret) | |
1504 | goto out; | |
1505 | ||
72fd032e SW |
1506 | namelen = strlen(name); |
1507 | if (strchr(name, '/')) { | |
f46b5a66 | 1508 | ret = -EINVAL; |
a874a63e | 1509 | goto out_drop_write; |
f46b5a66 CH |
1510 | } |
1511 | ||
16780cab CM |
1512 | if (name[0] == '.' && |
1513 | (namelen == 1 || (name[1] == '.' && namelen == 2))) { | |
1514 | ret = -EEXIST; | |
a874a63e | 1515 | goto out_drop_write; |
16780cab CM |
1516 | } |
1517 | ||
3de4586c | 1518 | if (subvol) { |
72fd032e | 1519 | ret = btrfs_mksubvol(&file->f_path, name, namelen, |
6f72c7e2 | 1520 | NULL, transid, readonly, inherit); |
cb8e7090 | 1521 | } else { |
2903ff01 | 1522 | struct fd src = fdget(fd); |
3de4586c | 1523 | struct inode *src_inode; |
2903ff01 | 1524 | if (!src.file) { |
3de4586c | 1525 | ret = -EINVAL; |
a874a63e | 1526 | goto out_drop_write; |
3de4586c CM |
1527 | } |
1528 | ||
496ad9aa AV |
1529 | src_inode = file_inode(src.file); |
1530 | if (src_inode->i_sb != file_inode(file)->i_sb) { | |
d397712b CM |
1531 | printk(KERN_INFO "btrfs: Snapshot src from " |
1532 | "another FS\n"); | |
3de4586c | 1533 | ret = -EINVAL; |
f0cea52a DS |
1534 | } else if (!inode_owner_or_capable(src_inode)) { |
1535 | /* | |
1536 | * Subvolume creation is not restricted, but snapshots | |
1537 | * are limited to own subvolumes only | |
1538 | */ | |
1539 | ret = -EPERM; | |
ecd18815 AV |
1540 | } else { |
1541 | ret = btrfs_mksubvol(&file->f_path, name, namelen, | |
1542 | BTRFS_I(src_inode)->root, | |
1543 | transid, readonly, inherit); | |
3de4586c | 1544 | } |
2903ff01 | 1545 | fdput(src); |
cb8e7090 | 1546 | } |
a874a63e LB |
1547 | out_drop_write: |
1548 | mnt_drop_write_file(file); | |
f46b5a66 | 1549 | out: |
72fd032e SW |
1550 | return ret; |
1551 | } | |
1552 | ||
1553 | static noinline int btrfs_ioctl_snap_create(struct file *file, | |
fa0d2b9b | 1554 | void __user *arg, int subvol) |
72fd032e | 1555 | { |
fa0d2b9b | 1556 | struct btrfs_ioctl_vol_args *vol_args; |
72fd032e SW |
1557 | int ret; |
1558 | ||
879149a0 JM |
1559 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1560 | return -ENOTDIR; | |
1561 | ||
fa0d2b9b LZ |
1562 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1563 | if (IS_ERR(vol_args)) | |
1564 | return PTR_ERR(vol_args); | |
1565 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
72fd032e | 1566 | |
fa0d2b9b | 1567 | ret = btrfs_ioctl_snap_create_transid(file, vol_args->name, |
b83cc969 | 1568 | vol_args->fd, subvol, |
6f72c7e2 | 1569 | NULL, false, NULL); |
fdfb1e4f | 1570 | |
fa0d2b9b LZ |
1571 | kfree(vol_args); |
1572 | return ret; | |
1573 | } | |
fdfb1e4f | 1574 | |
fa0d2b9b LZ |
1575 | static noinline int btrfs_ioctl_snap_create_v2(struct file *file, |
1576 | void __user *arg, int subvol) | |
1577 | { | |
1578 | struct btrfs_ioctl_vol_args_v2 *vol_args; | |
1579 | int ret; | |
1580 | u64 transid = 0; | |
1581 | u64 *ptr = NULL; | |
b83cc969 | 1582 | bool readonly = false; |
6f72c7e2 | 1583 | struct btrfs_qgroup_inherit *inherit = NULL; |
75eaa0e2 | 1584 | |
879149a0 JM |
1585 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1586 | return -ENOTDIR; | |
1587 | ||
fa0d2b9b LZ |
1588 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1589 | if (IS_ERR(vol_args)) | |
1590 | return PTR_ERR(vol_args); | |
1591 | vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0'; | |
75eaa0e2 | 1592 | |
b83cc969 | 1593 | if (vol_args->flags & |
6f72c7e2 AJ |
1594 | ~(BTRFS_SUBVOL_CREATE_ASYNC | BTRFS_SUBVOL_RDONLY | |
1595 | BTRFS_SUBVOL_QGROUP_INHERIT)) { | |
b83cc969 | 1596 | ret = -EOPNOTSUPP; |
fa0d2b9b | 1597 | goto out; |
72fd032e | 1598 | } |
fa0d2b9b LZ |
1599 | |
1600 | if (vol_args->flags & BTRFS_SUBVOL_CREATE_ASYNC) | |
1601 | ptr = &transid; | |
b83cc969 LZ |
1602 | if (vol_args->flags & BTRFS_SUBVOL_RDONLY) |
1603 | readonly = true; | |
6f72c7e2 AJ |
1604 | if (vol_args->flags & BTRFS_SUBVOL_QGROUP_INHERIT) { |
1605 | if (vol_args->size > PAGE_CACHE_SIZE) { | |
1606 | ret = -EINVAL; | |
1607 | goto out; | |
1608 | } | |
1609 | inherit = memdup_user(vol_args->qgroup_inherit, vol_args->size); | |
1610 | if (IS_ERR(inherit)) { | |
1611 | ret = PTR_ERR(inherit); | |
1612 | goto out; | |
1613 | } | |
1614 | } | |
fa0d2b9b LZ |
1615 | |
1616 | ret = btrfs_ioctl_snap_create_transid(file, vol_args->name, | |
6f72c7e2 | 1617 | vol_args->fd, subvol, ptr, |
8696c533 | 1618 | readonly, inherit); |
fa0d2b9b LZ |
1619 | |
1620 | if (ret == 0 && ptr && | |
1621 | copy_to_user(arg + | |
1622 | offsetof(struct btrfs_ioctl_vol_args_v2, | |
1623 | transid), ptr, sizeof(*ptr))) | |
1624 | ret = -EFAULT; | |
fdfb1e4f | 1625 | out: |
f46b5a66 | 1626 | kfree(vol_args); |
6f72c7e2 | 1627 | kfree(inherit); |
f46b5a66 CH |
1628 | return ret; |
1629 | } | |
1630 | ||
0caa102d LZ |
1631 | static noinline int btrfs_ioctl_subvol_getflags(struct file *file, |
1632 | void __user *arg) | |
1633 | { | |
496ad9aa | 1634 | struct inode *inode = file_inode(file); |
0caa102d LZ |
1635 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1636 | int ret = 0; | |
1637 | u64 flags = 0; | |
1638 | ||
33345d01 | 1639 | if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) |
0caa102d LZ |
1640 | return -EINVAL; |
1641 | ||
1642 | down_read(&root->fs_info->subvol_sem); | |
1643 | if (btrfs_root_readonly(root)) | |
1644 | flags |= BTRFS_SUBVOL_RDONLY; | |
1645 | up_read(&root->fs_info->subvol_sem); | |
1646 | ||
1647 | if (copy_to_user(arg, &flags, sizeof(flags))) | |
1648 | ret = -EFAULT; | |
1649 | ||
1650 | return ret; | |
1651 | } | |
1652 | ||
1653 | static noinline int btrfs_ioctl_subvol_setflags(struct file *file, | |
1654 | void __user *arg) | |
1655 | { | |
496ad9aa | 1656 | struct inode *inode = file_inode(file); |
0caa102d LZ |
1657 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1658 | struct btrfs_trans_handle *trans; | |
1659 | u64 root_flags; | |
1660 | u64 flags; | |
1661 | int ret = 0; | |
1662 | ||
b9ca0664 LB |
1663 | ret = mnt_want_write_file(file); |
1664 | if (ret) | |
1665 | goto out; | |
0caa102d | 1666 | |
b9ca0664 LB |
1667 | if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) { |
1668 | ret = -EINVAL; | |
1669 | goto out_drop_write; | |
1670 | } | |
0caa102d | 1671 | |
b9ca0664 LB |
1672 | if (copy_from_user(&flags, arg, sizeof(flags))) { |
1673 | ret = -EFAULT; | |
1674 | goto out_drop_write; | |
1675 | } | |
0caa102d | 1676 | |
b9ca0664 LB |
1677 | if (flags & BTRFS_SUBVOL_CREATE_ASYNC) { |
1678 | ret = -EINVAL; | |
1679 | goto out_drop_write; | |
1680 | } | |
0caa102d | 1681 | |
b9ca0664 LB |
1682 | if (flags & ~BTRFS_SUBVOL_RDONLY) { |
1683 | ret = -EOPNOTSUPP; | |
1684 | goto out_drop_write; | |
1685 | } | |
0caa102d | 1686 | |
b9ca0664 LB |
1687 | if (!inode_owner_or_capable(inode)) { |
1688 | ret = -EACCES; | |
1689 | goto out_drop_write; | |
1690 | } | |
b4dc2b8c | 1691 | |
0caa102d LZ |
1692 | down_write(&root->fs_info->subvol_sem); |
1693 | ||
1694 | /* nothing to do */ | |
1695 | if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root)) | |
b9ca0664 | 1696 | goto out_drop_sem; |
0caa102d LZ |
1697 | |
1698 | root_flags = btrfs_root_flags(&root->root_item); | |
1699 | if (flags & BTRFS_SUBVOL_RDONLY) | |
1700 | btrfs_set_root_flags(&root->root_item, | |
1701 | root_flags | BTRFS_ROOT_SUBVOL_RDONLY); | |
1702 | else | |
1703 | btrfs_set_root_flags(&root->root_item, | |
1704 | root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY); | |
1705 | ||
1706 | trans = btrfs_start_transaction(root, 1); | |
1707 | if (IS_ERR(trans)) { | |
1708 | ret = PTR_ERR(trans); | |
1709 | goto out_reset; | |
1710 | } | |
1711 | ||
b4dc2b8c | 1712 | ret = btrfs_update_root(trans, root->fs_info->tree_root, |
0caa102d LZ |
1713 | &root->root_key, &root->root_item); |
1714 | ||
1715 | btrfs_commit_transaction(trans, root); | |
1716 | out_reset: | |
1717 | if (ret) | |
1718 | btrfs_set_root_flags(&root->root_item, root_flags); | |
b9ca0664 | 1719 | out_drop_sem: |
0caa102d | 1720 | up_write(&root->fs_info->subvol_sem); |
b9ca0664 LB |
1721 | out_drop_write: |
1722 | mnt_drop_write_file(file); | |
1723 | out: | |
0caa102d LZ |
1724 | return ret; |
1725 | } | |
1726 | ||
76dda93c YZ |
1727 | /* |
1728 | * helper to check if the subvolume references other subvolumes | |
1729 | */ | |
1730 | static noinline int may_destroy_subvol(struct btrfs_root *root) | |
1731 | { | |
1732 | struct btrfs_path *path; | |
1733 | struct btrfs_key key; | |
1734 | int ret; | |
1735 | ||
1736 | path = btrfs_alloc_path(); | |
1737 | if (!path) | |
1738 | return -ENOMEM; | |
1739 | ||
1740 | key.objectid = root->root_key.objectid; | |
1741 | key.type = BTRFS_ROOT_REF_KEY; | |
1742 | key.offset = (u64)-1; | |
1743 | ||
1744 | ret = btrfs_search_slot(NULL, root->fs_info->tree_root, | |
1745 | &key, path, 0, 0); | |
1746 | if (ret < 0) | |
1747 | goto out; | |
1748 | BUG_ON(ret == 0); | |
1749 | ||
1750 | ret = 0; | |
1751 | if (path->slots[0] > 0) { | |
1752 | path->slots[0]--; | |
1753 | btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]); | |
1754 | if (key.objectid == root->root_key.objectid && | |
1755 | key.type == BTRFS_ROOT_REF_KEY) | |
1756 | ret = -ENOTEMPTY; | |
1757 | } | |
1758 | out: | |
1759 | btrfs_free_path(path); | |
1760 | return ret; | |
1761 | } | |
1762 | ||
ac8e9819 CM |
1763 | static noinline int key_in_sk(struct btrfs_key *key, |
1764 | struct btrfs_ioctl_search_key *sk) | |
1765 | { | |
abc6e134 CM |
1766 | struct btrfs_key test; |
1767 | int ret; | |
1768 | ||
1769 | test.objectid = sk->min_objectid; | |
1770 | test.type = sk->min_type; | |
1771 | test.offset = sk->min_offset; | |
1772 | ||
1773 | ret = btrfs_comp_cpu_keys(key, &test); | |
1774 | if (ret < 0) | |
ac8e9819 | 1775 | return 0; |
abc6e134 CM |
1776 | |
1777 | test.objectid = sk->max_objectid; | |
1778 | test.type = sk->max_type; | |
1779 | test.offset = sk->max_offset; | |
1780 | ||
1781 | ret = btrfs_comp_cpu_keys(key, &test); | |
1782 | if (ret > 0) | |
ac8e9819 CM |
1783 | return 0; |
1784 | return 1; | |
1785 | } | |
1786 | ||
1787 | static noinline int copy_to_sk(struct btrfs_root *root, | |
1788 | struct btrfs_path *path, | |
1789 | struct btrfs_key *key, | |
1790 | struct btrfs_ioctl_search_key *sk, | |
1791 | char *buf, | |
1792 | unsigned long *sk_offset, | |
1793 | int *num_found) | |
1794 | { | |
1795 | u64 found_transid; | |
1796 | struct extent_buffer *leaf; | |
1797 | struct btrfs_ioctl_search_header sh; | |
1798 | unsigned long item_off; | |
1799 | unsigned long item_len; | |
1800 | int nritems; | |
1801 | int i; | |
1802 | int slot; | |
ac8e9819 CM |
1803 | int ret = 0; |
1804 | ||
1805 | leaf = path->nodes[0]; | |
1806 | slot = path->slots[0]; | |
1807 | nritems = btrfs_header_nritems(leaf); | |
1808 | ||
1809 | if (btrfs_header_generation(leaf) > sk->max_transid) { | |
1810 | i = nritems; | |
1811 | goto advance_key; | |
1812 | } | |
1813 | found_transid = btrfs_header_generation(leaf); | |
1814 | ||
1815 | for (i = slot; i < nritems; i++) { | |
1816 | item_off = btrfs_item_ptr_offset(leaf, i); | |
1817 | item_len = btrfs_item_size_nr(leaf, i); | |
1818 | ||
03b71c6c GP |
1819 | btrfs_item_key_to_cpu(leaf, key, i); |
1820 | if (!key_in_sk(key, sk)) | |
1821 | continue; | |
1822 | ||
1823 | if (sizeof(sh) + item_len > BTRFS_SEARCH_ARGS_BUFSIZE) | |
ac8e9819 CM |
1824 | item_len = 0; |
1825 | ||
1826 | if (sizeof(sh) + item_len + *sk_offset > | |
1827 | BTRFS_SEARCH_ARGS_BUFSIZE) { | |
1828 | ret = 1; | |
1829 | goto overflow; | |
1830 | } | |
1831 | ||
ac8e9819 CM |
1832 | sh.objectid = key->objectid; |
1833 | sh.offset = key->offset; | |
1834 | sh.type = key->type; | |
1835 | sh.len = item_len; | |
1836 | sh.transid = found_transid; | |
1837 | ||
1838 | /* copy search result header */ | |
1839 | memcpy(buf + *sk_offset, &sh, sizeof(sh)); | |
1840 | *sk_offset += sizeof(sh); | |
1841 | ||
1842 | if (item_len) { | |
1843 | char *p = buf + *sk_offset; | |
1844 | /* copy the item */ | |
1845 | read_extent_buffer(leaf, p, | |
1846 | item_off, item_len); | |
1847 | *sk_offset += item_len; | |
ac8e9819 | 1848 | } |
e2156867 | 1849 | (*num_found)++; |
ac8e9819 CM |
1850 | |
1851 | if (*num_found >= sk->nr_items) | |
1852 | break; | |
1853 | } | |
1854 | advance_key: | |
abc6e134 CM |
1855 | ret = 0; |
1856 | if (key->offset < (u64)-1 && key->offset < sk->max_offset) | |
ac8e9819 | 1857 | key->offset++; |
abc6e134 CM |
1858 | else if (key->type < (u8)-1 && key->type < sk->max_type) { |
1859 | key->offset = 0; | |
ac8e9819 | 1860 | key->type++; |
abc6e134 CM |
1861 | } else if (key->objectid < (u64)-1 && key->objectid < sk->max_objectid) { |
1862 | key->offset = 0; | |
1863 | key->type = 0; | |
ac8e9819 | 1864 | key->objectid++; |
abc6e134 CM |
1865 | } else |
1866 | ret = 1; | |
ac8e9819 | 1867 | overflow: |
ac8e9819 CM |
1868 | return ret; |
1869 | } | |
1870 | ||
1871 | static noinline int search_ioctl(struct inode *inode, | |
1872 | struct btrfs_ioctl_search_args *args) | |
1873 | { | |
1874 | struct btrfs_root *root; | |
1875 | struct btrfs_key key; | |
1876 | struct btrfs_key max_key; | |
1877 | struct btrfs_path *path; | |
1878 | struct btrfs_ioctl_search_key *sk = &args->key; | |
1879 | struct btrfs_fs_info *info = BTRFS_I(inode)->root->fs_info; | |
1880 | int ret; | |
1881 | int num_found = 0; | |
1882 | unsigned long sk_offset = 0; | |
1883 | ||
1884 | path = btrfs_alloc_path(); | |
1885 | if (!path) | |
1886 | return -ENOMEM; | |
1887 | ||
1888 | if (sk->tree_id == 0) { | |
1889 | /* search the root of the inode that was passed */ | |
1890 | root = BTRFS_I(inode)->root; | |
1891 | } else { | |
1892 | key.objectid = sk->tree_id; | |
1893 | key.type = BTRFS_ROOT_ITEM_KEY; | |
1894 | key.offset = (u64)-1; | |
1895 | root = btrfs_read_fs_root_no_name(info, &key); | |
1896 | if (IS_ERR(root)) { | |
1897 | printk(KERN_ERR "could not find root %llu\n", | |
1898 | sk->tree_id); | |
1899 | btrfs_free_path(path); | |
1900 | return -ENOENT; | |
1901 | } | |
1902 | } | |
1903 | ||
1904 | key.objectid = sk->min_objectid; | |
1905 | key.type = sk->min_type; | |
1906 | key.offset = sk->min_offset; | |
1907 | ||
1908 | max_key.objectid = sk->max_objectid; | |
1909 | max_key.type = sk->max_type; | |
1910 | max_key.offset = sk->max_offset; | |
1911 | ||
1912 | path->keep_locks = 1; | |
1913 | ||
1914 | while(1) { | |
de78b51a | 1915 | ret = btrfs_search_forward(root, &key, &max_key, path, |
ac8e9819 CM |
1916 | sk->min_transid); |
1917 | if (ret != 0) { | |
1918 | if (ret > 0) | |
1919 | ret = 0; | |
1920 | goto err; | |
1921 | } | |
1922 | ret = copy_to_sk(root, path, &key, sk, args->buf, | |
1923 | &sk_offset, &num_found); | |
b3b4aa74 | 1924 | btrfs_release_path(path); |
ac8e9819 CM |
1925 | if (ret || num_found >= sk->nr_items) |
1926 | break; | |
1927 | ||
1928 | } | |
1929 | ret = 0; | |
1930 | err: | |
1931 | sk->nr_items = num_found; | |
1932 | btrfs_free_path(path); | |
1933 | return ret; | |
1934 | } | |
1935 | ||
1936 | static noinline int btrfs_ioctl_tree_search(struct file *file, | |
1937 | void __user *argp) | |
1938 | { | |
1939 | struct btrfs_ioctl_search_args *args; | |
1940 | struct inode *inode; | |
1941 | int ret; | |
1942 | ||
1943 | if (!capable(CAP_SYS_ADMIN)) | |
1944 | return -EPERM; | |
1945 | ||
2354d08f JL |
1946 | args = memdup_user(argp, sizeof(*args)); |
1947 | if (IS_ERR(args)) | |
1948 | return PTR_ERR(args); | |
ac8e9819 | 1949 | |
496ad9aa | 1950 | inode = file_inode(file); |
ac8e9819 CM |
1951 | ret = search_ioctl(inode, args); |
1952 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) | |
1953 | ret = -EFAULT; | |
1954 | kfree(args); | |
1955 | return ret; | |
1956 | } | |
1957 | ||
98d377a0 | 1958 | /* |
ac8e9819 CM |
1959 | * Search INODE_REFs to identify path name of 'dirid' directory |
1960 | * in a 'tree_id' tree. and sets path name to 'name'. | |
1961 | */ | |
98d377a0 TH |
1962 | static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, |
1963 | u64 tree_id, u64 dirid, char *name) | |
1964 | { | |
1965 | struct btrfs_root *root; | |
1966 | struct btrfs_key key; | |
ac8e9819 | 1967 | char *ptr; |
98d377a0 TH |
1968 | int ret = -1; |
1969 | int slot; | |
1970 | int len; | |
1971 | int total_len = 0; | |
1972 | struct btrfs_inode_ref *iref; | |
1973 | struct extent_buffer *l; | |
1974 | struct btrfs_path *path; | |
1975 | ||
1976 | if (dirid == BTRFS_FIRST_FREE_OBJECTID) { | |
1977 | name[0]='\0'; | |
1978 | return 0; | |
1979 | } | |
1980 | ||
1981 | path = btrfs_alloc_path(); | |
1982 | if (!path) | |
1983 | return -ENOMEM; | |
1984 | ||
ac8e9819 | 1985 | ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX]; |
98d377a0 TH |
1986 | |
1987 | key.objectid = tree_id; | |
1988 | key.type = BTRFS_ROOT_ITEM_KEY; | |
1989 | key.offset = (u64)-1; | |
1990 | root = btrfs_read_fs_root_no_name(info, &key); | |
1991 | if (IS_ERR(root)) { | |
1992 | printk(KERN_ERR "could not find root %llu\n", tree_id); | |
8ad6fcab CM |
1993 | ret = -ENOENT; |
1994 | goto out; | |
98d377a0 TH |
1995 | } |
1996 | ||
1997 | key.objectid = dirid; | |
1998 | key.type = BTRFS_INODE_REF_KEY; | |
8ad6fcab | 1999 | key.offset = (u64)-1; |
98d377a0 TH |
2000 | |
2001 | while(1) { | |
2002 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
2003 | if (ret < 0) | |
2004 | goto out; | |
2005 | ||
2006 | l = path->nodes[0]; | |
2007 | slot = path->slots[0]; | |
8ad6fcab CM |
2008 | if (ret > 0 && slot > 0) |
2009 | slot--; | |
98d377a0 TH |
2010 | btrfs_item_key_to_cpu(l, &key, slot); |
2011 | ||
2012 | if (ret > 0 && (key.objectid != dirid || | |
ac8e9819 CM |
2013 | key.type != BTRFS_INODE_REF_KEY)) { |
2014 | ret = -ENOENT; | |
98d377a0 | 2015 | goto out; |
ac8e9819 | 2016 | } |
98d377a0 TH |
2017 | |
2018 | iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref); | |
2019 | len = btrfs_inode_ref_name_len(l, iref); | |
2020 | ptr -= len + 1; | |
2021 | total_len += len + 1; | |
ac8e9819 | 2022 | if (ptr < name) |
98d377a0 TH |
2023 | goto out; |
2024 | ||
2025 | *(ptr + len) = '/'; | |
2026 | read_extent_buffer(l, ptr,(unsigned long)(iref + 1), len); | |
2027 | ||
2028 | if (key.offset == BTRFS_FIRST_FREE_OBJECTID) | |
2029 | break; | |
2030 | ||
b3b4aa74 | 2031 | btrfs_release_path(path); |
98d377a0 | 2032 | key.objectid = key.offset; |
8ad6fcab | 2033 | key.offset = (u64)-1; |
98d377a0 | 2034 | dirid = key.objectid; |
98d377a0 | 2035 | } |
ac8e9819 | 2036 | if (ptr < name) |
98d377a0 | 2037 | goto out; |
77906a50 | 2038 | memmove(name, ptr, total_len); |
98d377a0 TH |
2039 | name[total_len]='\0'; |
2040 | ret = 0; | |
2041 | out: | |
2042 | btrfs_free_path(path); | |
ac8e9819 CM |
2043 | return ret; |
2044 | } | |
2045 | ||
2046 | static noinline int btrfs_ioctl_ino_lookup(struct file *file, | |
2047 | void __user *argp) | |
2048 | { | |
2049 | struct btrfs_ioctl_ino_lookup_args *args; | |
2050 | struct inode *inode; | |
2051 | int ret; | |
2052 | ||
2053 | if (!capable(CAP_SYS_ADMIN)) | |
2054 | return -EPERM; | |
2055 | ||
2354d08f JL |
2056 | args = memdup_user(argp, sizeof(*args)); |
2057 | if (IS_ERR(args)) | |
2058 | return PTR_ERR(args); | |
c2b96929 | 2059 | |
496ad9aa | 2060 | inode = file_inode(file); |
ac8e9819 | 2061 | |
1b53ac4d CM |
2062 | if (args->treeid == 0) |
2063 | args->treeid = BTRFS_I(inode)->root->root_key.objectid; | |
2064 | ||
ac8e9819 CM |
2065 | ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info, |
2066 | args->treeid, args->objectid, | |
2067 | args->name); | |
2068 | ||
2069 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) | |
2070 | ret = -EFAULT; | |
2071 | ||
2072 | kfree(args); | |
98d377a0 TH |
2073 | return ret; |
2074 | } | |
2075 | ||
76dda93c YZ |
2076 | static noinline int btrfs_ioctl_snap_destroy(struct file *file, |
2077 | void __user *arg) | |
2078 | { | |
2079 | struct dentry *parent = fdentry(file); | |
2080 | struct dentry *dentry; | |
2081 | struct inode *dir = parent->d_inode; | |
2082 | struct inode *inode; | |
2083 | struct btrfs_root *root = BTRFS_I(dir)->root; | |
2084 | struct btrfs_root *dest = NULL; | |
2085 | struct btrfs_ioctl_vol_args *vol_args; | |
2086 | struct btrfs_trans_handle *trans; | |
c58aaad2 MX |
2087 | struct btrfs_block_rsv block_rsv; |
2088 | u64 qgroup_reserved; | |
76dda93c YZ |
2089 | int namelen; |
2090 | int ret; | |
2091 | int err = 0; | |
2092 | ||
879149a0 JM |
2093 | if (!S_ISDIR(dir->i_mode)) |
2094 | return -ENOTDIR; | |
2095 | ||
76dda93c YZ |
2096 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
2097 | if (IS_ERR(vol_args)) | |
2098 | return PTR_ERR(vol_args); | |
2099 | ||
2100 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
2101 | namelen = strlen(vol_args->name); | |
2102 | if (strchr(vol_args->name, '/') || | |
2103 | strncmp(vol_args->name, "..", namelen) == 0) { | |
2104 | err = -EINVAL; | |
2105 | goto out; | |
2106 | } | |
2107 | ||
a561be71 | 2108 | err = mnt_want_write_file(file); |
76dda93c YZ |
2109 | if (err) |
2110 | goto out; | |
2111 | ||
5c50c9b8 DS |
2112 | err = mutex_lock_killable_nested(&dir->i_mutex, I_MUTEX_PARENT); |
2113 | if (err == -EINTR) | |
89ec7522 | 2114 | goto out_drop_write; |
76dda93c YZ |
2115 | dentry = lookup_one_len(vol_args->name, parent, namelen); |
2116 | if (IS_ERR(dentry)) { | |
2117 | err = PTR_ERR(dentry); | |
2118 | goto out_unlock_dir; | |
2119 | } | |
2120 | ||
2121 | if (!dentry->d_inode) { | |
2122 | err = -ENOENT; | |
2123 | goto out_dput; | |
2124 | } | |
2125 | ||
2126 | inode = dentry->d_inode; | |
4260f7c7 SW |
2127 | dest = BTRFS_I(inode)->root; |
2128 | if (!capable(CAP_SYS_ADMIN)){ | |
2129 | /* | |
2130 | * Regular user. Only allow this with a special mount | |
2131 | * option, when the user has write+exec access to the | |
2132 | * subvol root, and when rmdir(2) would have been | |
2133 | * allowed. | |
2134 | * | |
2135 | * Note that this is _not_ check that the subvol is | |
2136 | * empty or doesn't contain data that we wouldn't | |
2137 | * otherwise be able to delete. | |
2138 | * | |
2139 | * Users who want to delete empty subvols should try | |
2140 | * rmdir(2). | |
2141 | */ | |
2142 | err = -EPERM; | |
2143 | if (!btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED)) | |
2144 | goto out_dput; | |
2145 | ||
2146 | /* | |
2147 | * Do not allow deletion if the parent dir is the same | |
2148 | * as the dir to be deleted. That means the ioctl | |
2149 | * must be called on the dentry referencing the root | |
2150 | * of the subvol, not a random directory contained | |
2151 | * within it. | |
2152 | */ | |
2153 | err = -EINVAL; | |
2154 | if (root == dest) | |
2155 | goto out_dput; | |
2156 | ||
2157 | err = inode_permission(inode, MAY_WRITE | MAY_EXEC); | |
2158 | if (err) | |
2159 | goto out_dput; | |
4260f7c7 SW |
2160 | } |
2161 | ||
5c39da5b MX |
2162 | /* check if subvolume may be deleted by a user */ |
2163 | err = btrfs_may_delete(dir, dentry, 1); | |
2164 | if (err) | |
2165 | goto out_dput; | |
2166 | ||
33345d01 | 2167 | if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) { |
76dda93c YZ |
2168 | err = -EINVAL; |
2169 | goto out_dput; | |
2170 | } | |
2171 | ||
76dda93c YZ |
2172 | mutex_lock(&inode->i_mutex); |
2173 | err = d_invalidate(dentry); | |
2174 | if (err) | |
2175 | goto out_unlock; | |
2176 | ||
2177 | down_write(&root->fs_info->subvol_sem); | |
2178 | ||
2179 | err = may_destroy_subvol(dest); | |
2180 | if (err) | |
2181 | goto out_up_write; | |
2182 | ||
c58aaad2 MX |
2183 | btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP); |
2184 | /* | |
2185 | * One for dir inode, two for dir entries, two for root | |
2186 | * ref/backref. | |
2187 | */ | |
2188 | err = btrfs_subvolume_reserve_metadata(root, &block_rsv, | |
2189 | 5, &qgroup_reserved); | |
2190 | if (err) | |
2191 | goto out_up_write; | |
2192 | ||
a22285a6 YZ |
2193 | trans = btrfs_start_transaction(root, 0); |
2194 | if (IS_ERR(trans)) { | |
2195 | err = PTR_ERR(trans); | |
c58aaad2 | 2196 | goto out_release; |
a22285a6 | 2197 | } |
c58aaad2 MX |
2198 | trans->block_rsv = &block_rsv; |
2199 | trans->bytes_reserved = block_rsv.size; | |
a22285a6 | 2200 | |
76dda93c YZ |
2201 | ret = btrfs_unlink_subvol(trans, root, dir, |
2202 | dest->root_key.objectid, | |
2203 | dentry->d_name.name, | |
2204 | dentry->d_name.len); | |
79787eaa JM |
2205 | if (ret) { |
2206 | err = ret; | |
2207 | btrfs_abort_transaction(trans, root, ret); | |
2208 | goto out_end_trans; | |
2209 | } | |
76dda93c YZ |
2210 | |
2211 | btrfs_record_root_in_trans(trans, dest); | |
2212 | ||
2213 | memset(&dest->root_item.drop_progress, 0, | |
2214 | sizeof(dest->root_item.drop_progress)); | |
2215 | dest->root_item.drop_level = 0; | |
2216 | btrfs_set_root_refs(&dest->root_item, 0); | |
2217 | ||
d68fc57b YZ |
2218 | if (!xchg(&dest->orphan_item_inserted, 1)) { |
2219 | ret = btrfs_insert_orphan_item(trans, | |
2220 | root->fs_info->tree_root, | |
2221 | dest->root_key.objectid); | |
79787eaa JM |
2222 | if (ret) { |
2223 | btrfs_abort_transaction(trans, root, ret); | |
2224 | err = ret; | |
2225 | goto out_end_trans; | |
2226 | } | |
d68fc57b | 2227 | } |
79787eaa | 2228 | out_end_trans: |
c58aaad2 MX |
2229 | trans->block_rsv = NULL; |
2230 | trans->bytes_reserved = 0; | |
531cb13f | 2231 | ret = btrfs_end_transaction(trans, root); |
79787eaa JM |
2232 | if (ret && !err) |
2233 | err = ret; | |
76dda93c | 2234 | inode->i_flags |= S_DEAD; |
c58aaad2 MX |
2235 | out_release: |
2236 | btrfs_subvolume_release_metadata(root, &block_rsv, qgroup_reserved); | |
76dda93c YZ |
2237 | out_up_write: |
2238 | up_write(&root->fs_info->subvol_sem); | |
2239 | out_unlock: | |
2240 | mutex_unlock(&inode->i_mutex); | |
2241 | if (!err) { | |
efefb143 | 2242 | shrink_dcache_sb(root->fs_info->sb); |
76dda93c YZ |
2243 | btrfs_invalidate_inodes(dest); |
2244 | d_delete(dentry); | |
fa6ac876 LB |
2245 | |
2246 | /* the last ref */ | |
2247 | if (dest->cache_inode) { | |
2248 | iput(dest->cache_inode); | |
2249 | dest->cache_inode = NULL; | |
2250 | } | |
76dda93c YZ |
2251 | } |
2252 | out_dput: | |
2253 | dput(dentry); | |
2254 | out_unlock_dir: | |
2255 | mutex_unlock(&dir->i_mutex); | |
89ec7522 | 2256 | out_drop_write: |
2a79f17e | 2257 | mnt_drop_write_file(file); |
76dda93c YZ |
2258 | out: |
2259 | kfree(vol_args); | |
2260 | return err; | |
2261 | } | |
2262 | ||
1e701a32 | 2263 | static int btrfs_ioctl_defrag(struct file *file, void __user *argp) |
f46b5a66 | 2264 | { |
496ad9aa | 2265 | struct inode *inode = file_inode(file); |
f46b5a66 | 2266 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1e701a32 | 2267 | struct btrfs_ioctl_defrag_range_args *range; |
c146afad YZ |
2268 | int ret; |
2269 | ||
25122d15 ID |
2270 | ret = mnt_want_write_file(file); |
2271 | if (ret) | |
2272 | return ret; | |
b83cc969 | 2273 | |
25122d15 ID |
2274 | if (btrfs_root_readonly(root)) { |
2275 | ret = -EROFS; | |
2276 | goto out; | |
5ac00add | 2277 | } |
f46b5a66 CH |
2278 | |
2279 | switch (inode->i_mode & S_IFMT) { | |
2280 | case S_IFDIR: | |
e441d54d CM |
2281 | if (!capable(CAP_SYS_ADMIN)) { |
2282 | ret = -EPERM; | |
2283 | goto out; | |
2284 | } | |
de78b51a | 2285 | ret = btrfs_defrag_root(root); |
8929ecfa YZ |
2286 | if (ret) |
2287 | goto out; | |
de78b51a | 2288 | ret = btrfs_defrag_root(root->fs_info->extent_root); |
f46b5a66 CH |
2289 | break; |
2290 | case S_IFREG: | |
e441d54d CM |
2291 | if (!(file->f_mode & FMODE_WRITE)) { |
2292 | ret = -EINVAL; | |
2293 | goto out; | |
2294 | } | |
1e701a32 CM |
2295 | |
2296 | range = kzalloc(sizeof(*range), GFP_KERNEL); | |
2297 | if (!range) { | |
2298 | ret = -ENOMEM; | |
2299 | goto out; | |
2300 | } | |
2301 | ||
2302 | if (argp) { | |
2303 | if (copy_from_user(range, argp, | |
2304 | sizeof(*range))) { | |
2305 | ret = -EFAULT; | |
2306 | kfree(range); | |
683be16e | 2307 | goto out; |
1e701a32 CM |
2308 | } |
2309 | /* compression requires us to start the IO */ | |
2310 | if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { | |
2311 | range->flags |= BTRFS_DEFRAG_RANGE_START_IO; | |
2312 | range->extent_thresh = (u32)-1; | |
2313 | } | |
2314 | } else { | |
2315 | /* the rest are all set to zero by kzalloc */ | |
2316 | range->len = (u64)-1; | |
2317 | } | |
496ad9aa | 2318 | ret = btrfs_defrag_file(file_inode(file), file, |
4cb5300b CM |
2319 | range, 0, 0); |
2320 | if (ret > 0) | |
2321 | ret = 0; | |
1e701a32 | 2322 | kfree(range); |
f46b5a66 | 2323 | break; |
8929ecfa YZ |
2324 | default: |
2325 | ret = -EINVAL; | |
f46b5a66 | 2326 | } |
e441d54d | 2327 | out: |
25122d15 | 2328 | mnt_drop_write_file(file); |
e441d54d | 2329 | return ret; |
f46b5a66 CH |
2330 | } |
2331 | ||
b2950863 | 2332 | static long btrfs_ioctl_add_dev(struct btrfs_root *root, void __user *arg) |
f46b5a66 CH |
2333 | { |
2334 | struct btrfs_ioctl_vol_args *vol_args; | |
2335 | int ret; | |
2336 | ||
e441d54d CM |
2337 | if (!capable(CAP_SYS_ADMIN)) |
2338 | return -EPERM; | |
2339 | ||
5ac00add SB |
2340 | if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running, |
2341 | 1)) { | |
2342 | pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n"); | |
2c0c9da0 | 2343 | return -EINVAL; |
c9e9f97b ID |
2344 | } |
2345 | ||
5ac00add | 2346 | mutex_lock(&root->fs_info->volume_mutex); |
dae7b665 | 2347 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
2348 | if (IS_ERR(vol_args)) { |
2349 | ret = PTR_ERR(vol_args); | |
2350 | goto out; | |
2351 | } | |
f46b5a66 | 2352 | |
5516e595 | 2353 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
f46b5a66 CH |
2354 | ret = btrfs_init_new_device(root, vol_args->name); |
2355 | ||
f46b5a66 | 2356 | kfree(vol_args); |
c9e9f97b ID |
2357 | out: |
2358 | mutex_unlock(&root->fs_info->volume_mutex); | |
5ac00add | 2359 | atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0); |
f46b5a66 CH |
2360 | return ret; |
2361 | } | |
2362 | ||
da24927b | 2363 | static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg) |
f46b5a66 | 2364 | { |
496ad9aa | 2365 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
f46b5a66 CH |
2366 | struct btrfs_ioctl_vol_args *vol_args; |
2367 | int ret; | |
2368 | ||
e441d54d CM |
2369 | if (!capable(CAP_SYS_ADMIN)) |
2370 | return -EPERM; | |
2371 | ||
da24927b MX |
2372 | ret = mnt_want_write_file(file); |
2373 | if (ret) | |
2374 | return ret; | |
c146afad | 2375 | |
5ac00add SB |
2376 | if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running, |
2377 | 1)) { | |
2378 | pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n"); | |
da24927b | 2379 | mnt_drop_write_file(file); |
2c0c9da0 | 2380 | return -EINVAL; |
c9e9f97b ID |
2381 | } |
2382 | ||
5ac00add | 2383 | mutex_lock(&root->fs_info->volume_mutex); |
dae7b665 | 2384 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
2385 | if (IS_ERR(vol_args)) { |
2386 | ret = PTR_ERR(vol_args); | |
2387 | goto out; | |
2388 | } | |
f46b5a66 | 2389 | |
5516e595 | 2390 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
f46b5a66 CH |
2391 | ret = btrfs_rm_device(root, vol_args->name); |
2392 | ||
f46b5a66 | 2393 | kfree(vol_args); |
c9e9f97b ID |
2394 | out: |
2395 | mutex_unlock(&root->fs_info->volume_mutex); | |
5ac00add | 2396 | atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0); |
4ac20c70 | 2397 | mnt_drop_write_file(file); |
f46b5a66 CH |
2398 | return ret; |
2399 | } | |
2400 | ||
475f6387 JS |
2401 | static long btrfs_ioctl_fs_info(struct btrfs_root *root, void __user *arg) |
2402 | { | |
027ed2f0 | 2403 | struct btrfs_ioctl_fs_info_args *fi_args; |
475f6387 JS |
2404 | struct btrfs_device *device; |
2405 | struct btrfs_device *next; | |
2406 | struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices; | |
027ed2f0 | 2407 | int ret = 0; |
475f6387 JS |
2408 | |
2409 | if (!capable(CAP_SYS_ADMIN)) | |
2410 | return -EPERM; | |
2411 | ||
027ed2f0 LZ |
2412 | fi_args = kzalloc(sizeof(*fi_args), GFP_KERNEL); |
2413 | if (!fi_args) | |
2414 | return -ENOMEM; | |
2415 | ||
2416 | fi_args->num_devices = fs_devices->num_devices; | |
2417 | memcpy(&fi_args->fsid, root->fs_info->fsid, sizeof(fi_args->fsid)); | |
475f6387 JS |
2418 | |
2419 | mutex_lock(&fs_devices->device_list_mutex); | |
2420 | list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) { | |
027ed2f0 LZ |
2421 | if (device->devid > fi_args->max_id) |
2422 | fi_args->max_id = device->devid; | |
475f6387 JS |
2423 | } |
2424 | mutex_unlock(&fs_devices->device_list_mutex); | |
2425 | ||
027ed2f0 LZ |
2426 | if (copy_to_user(arg, fi_args, sizeof(*fi_args))) |
2427 | ret = -EFAULT; | |
475f6387 | 2428 | |
027ed2f0 LZ |
2429 | kfree(fi_args); |
2430 | return ret; | |
475f6387 JS |
2431 | } |
2432 | ||
2433 | static long btrfs_ioctl_dev_info(struct btrfs_root *root, void __user *arg) | |
2434 | { | |
2435 | struct btrfs_ioctl_dev_info_args *di_args; | |
2436 | struct btrfs_device *dev; | |
2437 | struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices; | |
2438 | int ret = 0; | |
2439 | char *s_uuid = NULL; | |
2440 | char empty_uuid[BTRFS_UUID_SIZE] = {0}; | |
2441 | ||
2442 | if (!capable(CAP_SYS_ADMIN)) | |
2443 | return -EPERM; | |
2444 | ||
2445 | di_args = memdup_user(arg, sizeof(*di_args)); | |
2446 | if (IS_ERR(di_args)) | |
2447 | return PTR_ERR(di_args); | |
2448 | ||
2449 | if (memcmp(empty_uuid, di_args->uuid, BTRFS_UUID_SIZE) != 0) | |
2450 | s_uuid = di_args->uuid; | |
2451 | ||
2452 | mutex_lock(&fs_devices->device_list_mutex); | |
aa1b8cd4 | 2453 | dev = btrfs_find_device(root->fs_info, di_args->devid, s_uuid, NULL); |
475f6387 JS |
2454 | |
2455 | if (!dev) { | |
2456 | ret = -ENODEV; | |
2457 | goto out; | |
2458 | } | |
2459 | ||
2460 | di_args->devid = dev->devid; | |
2461 | di_args->bytes_used = dev->bytes_used; | |
2462 | di_args->total_bytes = dev->total_bytes; | |
2463 | memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid)); | |
a27202fb | 2464 | if (dev->name) { |
606686ee JB |
2465 | struct rcu_string *name; |
2466 | ||
2467 | rcu_read_lock(); | |
2468 | name = rcu_dereference(dev->name); | |
2469 | strncpy(di_args->path, name->str, sizeof(di_args->path)); | |
2470 | rcu_read_unlock(); | |
a27202fb JM |
2471 | di_args->path[sizeof(di_args->path) - 1] = 0; |
2472 | } else { | |
99ba55ad | 2473 | di_args->path[0] = '\0'; |
a27202fb | 2474 | } |
475f6387 JS |
2475 | |
2476 | out: | |
55793c0d | 2477 | mutex_unlock(&fs_devices->device_list_mutex); |
475f6387 JS |
2478 | if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args))) |
2479 | ret = -EFAULT; | |
2480 | ||
2481 | kfree(di_args); | |
2482 | return ret; | |
2483 | } | |
2484 | ||
76dda93c YZ |
2485 | static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd, |
2486 | u64 off, u64 olen, u64 destoff) | |
f46b5a66 | 2487 | { |
496ad9aa | 2488 | struct inode *inode = file_inode(file); |
f46b5a66 | 2489 | struct btrfs_root *root = BTRFS_I(inode)->root; |
2903ff01 | 2490 | struct fd src_file; |
f46b5a66 CH |
2491 | struct inode *src; |
2492 | struct btrfs_trans_handle *trans; | |
f46b5a66 | 2493 | struct btrfs_path *path; |
f46b5a66 | 2494 | struct extent_buffer *leaf; |
ae01a0ab YZ |
2495 | char *buf; |
2496 | struct btrfs_key key; | |
f46b5a66 CH |
2497 | u32 nritems; |
2498 | int slot; | |
ae01a0ab | 2499 | int ret; |
c5c9cd4d SW |
2500 | u64 len = olen; |
2501 | u64 bs = root->fs_info->sb->s_blocksize; | |
d20f7043 | 2502 | |
c5c9cd4d SW |
2503 | /* |
2504 | * TODO: | |
2505 | * - split compressed inline extents. annoying: we need to | |
2506 | * decompress into destination's address_space (the file offset | |
2507 | * may change, so source mapping won't do), then recompress (or | |
2508 | * otherwise reinsert) a subrange. | |
2509 | * - allow ranges within the same file to be cloned (provided | |
2510 | * they don't overlap)? | |
2511 | */ | |
2512 | ||
e441d54d | 2513 | /* the destination must be opened for writing */ |
2ebc3464 | 2514 | if (!(file->f_mode & FMODE_WRITE) || (file->f_flags & O_APPEND)) |
e441d54d CM |
2515 | return -EINVAL; |
2516 | ||
b83cc969 LZ |
2517 | if (btrfs_root_readonly(root)) |
2518 | return -EROFS; | |
2519 | ||
a561be71 | 2520 | ret = mnt_want_write_file(file); |
c146afad YZ |
2521 | if (ret) |
2522 | return ret; | |
2523 | ||
2903ff01 AV |
2524 | src_file = fdget(srcfd); |
2525 | if (!src_file.file) { | |
ab67b7c1 YZ |
2526 | ret = -EBADF; |
2527 | goto out_drop_write; | |
2528 | } | |
5dc64164 | 2529 | |
362a20c5 | 2530 | ret = -EXDEV; |
2903ff01 | 2531 | if (src_file.file->f_path.mnt != file->f_path.mnt) |
362a20c5 DS |
2532 | goto out_fput; |
2533 | ||
496ad9aa | 2534 | src = file_inode(src_file.file); |
f46b5a66 | 2535 | |
c5c9cd4d SW |
2536 | ret = -EINVAL; |
2537 | if (src == inode) | |
2538 | goto out_fput; | |
2539 | ||
5dc64164 | 2540 | /* the src must be open for reading */ |
2903ff01 | 2541 | if (!(src_file.file->f_mode & FMODE_READ)) |
5dc64164 DR |
2542 | goto out_fput; |
2543 | ||
0e7b824c LZ |
2544 | /* don't make the dst file partly checksummed */ |
2545 | if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) != | |
2546 | (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) | |
2547 | goto out_fput; | |
2548 | ||
ae01a0ab YZ |
2549 | ret = -EISDIR; |
2550 | if (S_ISDIR(src->i_mode) || S_ISDIR(inode->i_mode)) | |
2551 | goto out_fput; | |
2552 | ||
f46b5a66 | 2553 | ret = -EXDEV; |
362a20c5 | 2554 | if (src->i_sb != inode->i_sb) |
ae01a0ab YZ |
2555 | goto out_fput; |
2556 | ||
2557 | ret = -ENOMEM; | |
2558 | buf = vmalloc(btrfs_level_size(root, 0)); | |
2559 | if (!buf) | |
2560 | goto out_fput; | |
2561 | ||
2562 | path = btrfs_alloc_path(); | |
2563 | if (!path) { | |
2564 | vfree(buf); | |
f46b5a66 | 2565 | goto out_fput; |
ae01a0ab YZ |
2566 | } |
2567 | path->reada = 2; | |
f46b5a66 CH |
2568 | |
2569 | if (inode < src) { | |
fccdae43 SW |
2570 | mutex_lock_nested(&inode->i_mutex, I_MUTEX_PARENT); |
2571 | mutex_lock_nested(&src->i_mutex, I_MUTEX_CHILD); | |
f46b5a66 | 2572 | } else { |
fccdae43 SW |
2573 | mutex_lock_nested(&src->i_mutex, I_MUTEX_PARENT); |
2574 | mutex_lock_nested(&inode->i_mutex, I_MUTEX_CHILD); | |
f46b5a66 CH |
2575 | } |
2576 | ||
c5c9cd4d SW |
2577 | /* determine range to clone */ |
2578 | ret = -EINVAL; | |
2ebc3464 | 2579 | if (off + len > src->i_size || off + len < off) |
f46b5a66 | 2580 | goto out_unlock; |
c5c9cd4d SW |
2581 | if (len == 0) |
2582 | olen = len = src->i_size - off; | |
2583 | /* if we extend to eof, continue to block boundary */ | |
2584 | if (off + len == src->i_size) | |
2a6b8dae | 2585 | len = ALIGN(src->i_size, bs) - off; |
c5c9cd4d | 2586 | |
6073c416 FM |
2587 | if (len == 0) { |
2588 | ret = 0; | |
2589 | goto out_unlock; | |
2590 | } | |
2591 | ||
c5c9cd4d | 2592 | /* verify the end result is block aligned */ |
2a6b8dae LZ |
2593 | if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs) || |
2594 | !IS_ALIGNED(destoff, bs)) | |
c5c9cd4d SW |
2595 | goto out_unlock; |
2596 | ||
d525e8ab LZ |
2597 | if (destoff > inode->i_size) { |
2598 | ret = btrfs_cont_expand(inode, inode->i_size, destoff); | |
2599 | if (ret) | |
2600 | goto out_unlock; | |
2601 | } | |
2602 | ||
71ef0786 LZ |
2603 | /* truncate page cache pages from target inode range */ |
2604 | truncate_inode_pages_range(&inode->i_data, destoff, | |
2605 | PAGE_CACHE_ALIGN(destoff + len) - 1); | |
2606 | ||
f46b5a66 CH |
2607 | /* do any pending delalloc/csum calc on src, one way or |
2608 | another, and lock file content */ | |
2609 | while (1) { | |
31840ae1 | 2610 | struct btrfs_ordered_extent *ordered; |
aa42ffd9 LB |
2611 | lock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1); |
2612 | ordered = btrfs_lookup_first_ordered_extent(src, off + len - 1); | |
9a019196 | 2613 | if (!ordered && |
aa42ffd9 LB |
2614 | !test_range_bit(&BTRFS_I(src)->io_tree, off, off + len - 1, |
2615 | EXTENT_DELALLOC, 0, NULL)) | |
f46b5a66 | 2616 | break; |
aa42ffd9 | 2617 | unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1); |
ae01a0ab YZ |
2618 | if (ordered) |
2619 | btrfs_put_ordered_extent(ordered); | |
9a019196 | 2620 | btrfs_wait_ordered_range(src, off, len); |
f46b5a66 CH |
2621 | } |
2622 | ||
c5c9cd4d | 2623 | /* clone data */ |
33345d01 | 2624 | key.objectid = btrfs_ino(src); |
ae01a0ab YZ |
2625 | key.type = BTRFS_EXTENT_DATA_KEY; |
2626 | key.offset = 0; | |
f46b5a66 CH |
2627 | |
2628 | while (1) { | |
2629 | /* | |
2630 | * note the key will change type as we walk through the | |
2631 | * tree. | |
2632 | */ | |
362a20c5 DS |
2633 | ret = btrfs_search_slot(NULL, BTRFS_I(src)->root, &key, path, |
2634 | 0, 0); | |
f46b5a66 CH |
2635 | if (ret < 0) |
2636 | goto out; | |
2637 | ||
ae01a0ab YZ |
2638 | nritems = btrfs_header_nritems(path->nodes[0]); |
2639 | if (path->slots[0] >= nritems) { | |
362a20c5 | 2640 | ret = btrfs_next_leaf(BTRFS_I(src)->root, path); |
f46b5a66 CH |
2641 | if (ret < 0) |
2642 | goto out; | |
2643 | if (ret > 0) | |
2644 | break; | |
ae01a0ab | 2645 | nritems = btrfs_header_nritems(path->nodes[0]); |
f46b5a66 CH |
2646 | } |
2647 | leaf = path->nodes[0]; | |
2648 | slot = path->slots[0]; | |
f46b5a66 | 2649 | |
ae01a0ab | 2650 | btrfs_item_key_to_cpu(leaf, &key, slot); |
d20f7043 | 2651 | if (btrfs_key_type(&key) > BTRFS_EXTENT_DATA_KEY || |
33345d01 | 2652 | key.objectid != btrfs_ino(src)) |
f46b5a66 CH |
2653 | break; |
2654 | ||
c5c9cd4d SW |
2655 | if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) { |
2656 | struct btrfs_file_extent_item *extent; | |
2657 | int type; | |
31840ae1 ZY |
2658 | u32 size; |
2659 | struct btrfs_key new_key; | |
c5c9cd4d SW |
2660 | u64 disko = 0, diskl = 0; |
2661 | u64 datao = 0, datal = 0; | |
2662 | u8 comp; | |
b5384d48 | 2663 | u64 endoff; |
31840ae1 ZY |
2664 | |
2665 | size = btrfs_item_size_nr(leaf, slot); | |
2666 | read_extent_buffer(leaf, buf, | |
2667 | btrfs_item_ptr_offset(leaf, slot), | |
2668 | size); | |
c5c9cd4d SW |
2669 | |
2670 | extent = btrfs_item_ptr(leaf, slot, | |
2671 | struct btrfs_file_extent_item); | |
2672 | comp = btrfs_file_extent_compression(leaf, extent); | |
2673 | type = btrfs_file_extent_type(leaf, extent); | |
c8a894d7 CM |
2674 | if (type == BTRFS_FILE_EXTENT_REG || |
2675 | type == BTRFS_FILE_EXTENT_PREALLOC) { | |
d397712b CM |
2676 | disko = btrfs_file_extent_disk_bytenr(leaf, |
2677 | extent); | |
2678 | diskl = btrfs_file_extent_disk_num_bytes(leaf, | |
2679 | extent); | |
c5c9cd4d | 2680 | datao = btrfs_file_extent_offset(leaf, extent); |
d397712b CM |
2681 | datal = btrfs_file_extent_num_bytes(leaf, |
2682 | extent); | |
c5c9cd4d SW |
2683 | } else if (type == BTRFS_FILE_EXTENT_INLINE) { |
2684 | /* take upper bound, may be compressed */ | |
2685 | datal = btrfs_file_extent_ram_bytes(leaf, | |
2686 | extent); | |
2687 | } | |
b3b4aa74 | 2688 | btrfs_release_path(path); |
31840ae1 | 2689 | |
050006a7 | 2690 | if (key.offset + datal <= off || |
aa42ffd9 | 2691 | key.offset >= off + len - 1) |
c5c9cd4d SW |
2692 | goto next; |
2693 | ||
31840ae1 | 2694 | memcpy(&new_key, &key, sizeof(new_key)); |
33345d01 | 2695 | new_key.objectid = btrfs_ino(inode); |
4d728ec7 LZ |
2696 | if (off <= key.offset) |
2697 | new_key.offset = key.offset + destoff - off; | |
2698 | else | |
2699 | new_key.offset = destoff; | |
31840ae1 | 2700 | |
b6f3409b SW |
2701 | /* |
2702 | * 1 - adjusting old extent (we may have to split it) | |
2703 | * 1 - add new extent | |
2704 | * 1 - inode update | |
2705 | */ | |
2706 | trans = btrfs_start_transaction(root, 3); | |
a22285a6 YZ |
2707 | if (IS_ERR(trans)) { |
2708 | ret = PTR_ERR(trans); | |
2709 | goto out; | |
2710 | } | |
2711 | ||
c8a894d7 CM |
2712 | if (type == BTRFS_FILE_EXTENT_REG || |
2713 | type == BTRFS_FILE_EXTENT_PREALLOC) { | |
d72c0842 LZ |
2714 | /* |
2715 | * a | --- range to clone ---| b | |
2716 | * | ------------- extent ------------- | | |
2717 | */ | |
2718 | ||
2719 | /* substract range b */ | |
2720 | if (key.offset + datal > off + len) | |
2721 | datal = off + len - key.offset; | |
2722 | ||
2723 | /* substract range a */ | |
a22285a6 YZ |
2724 | if (off > key.offset) { |
2725 | datao += off - key.offset; | |
2726 | datal -= off - key.offset; | |
2727 | } | |
2728 | ||
5dc562c5 | 2729 | ret = btrfs_drop_extents(trans, root, inode, |
a22285a6 YZ |
2730 | new_key.offset, |
2731 | new_key.offset + datal, | |
2671485d | 2732 | 1); |
79787eaa JM |
2733 | if (ret) { |
2734 | btrfs_abort_transaction(trans, root, | |
2735 | ret); | |
2736 | btrfs_end_transaction(trans, root); | |
2737 | goto out; | |
2738 | } | |
a22285a6 | 2739 | |
c5c9cd4d SW |
2740 | ret = btrfs_insert_empty_item(trans, root, path, |
2741 | &new_key, size); | |
79787eaa JM |
2742 | if (ret) { |
2743 | btrfs_abort_transaction(trans, root, | |
2744 | ret); | |
2745 | btrfs_end_transaction(trans, root); | |
2746 | goto out; | |
2747 | } | |
c5c9cd4d SW |
2748 | |
2749 | leaf = path->nodes[0]; | |
2750 | slot = path->slots[0]; | |
2751 | write_extent_buffer(leaf, buf, | |
31840ae1 ZY |
2752 | btrfs_item_ptr_offset(leaf, slot), |
2753 | size); | |
ae01a0ab | 2754 | |
c5c9cd4d | 2755 | extent = btrfs_item_ptr(leaf, slot, |
f46b5a66 | 2756 | struct btrfs_file_extent_item); |
c5c9cd4d | 2757 | |
c5c9cd4d SW |
2758 | /* disko == 0 means it's a hole */ |
2759 | if (!disko) | |
2760 | datao = 0; | |
c5c9cd4d SW |
2761 | |
2762 | btrfs_set_file_extent_offset(leaf, extent, | |
2763 | datao); | |
2764 | btrfs_set_file_extent_num_bytes(leaf, extent, | |
2765 | datal); | |
2766 | if (disko) { | |
2767 | inode_add_bytes(inode, datal); | |
ae01a0ab | 2768 | ret = btrfs_inc_extent_ref(trans, root, |
5d4f98a2 YZ |
2769 | disko, diskl, 0, |
2770 | root->root_key.objectid, | |
33345d01 | 2771 | btrfs_ino(inode), |
66d7e7f0 AJ |
2772 | new_key.offset - datao, |
2773 | 0); | |
79787eaa JM |
2774 | if (ret) { |
2775 | btrfs_abort_transaction(trans, | |
2776 | root, | |
2777 | ret); | |
2778 | btrfs_end_transaction(trans, | |
2779 | root); | |
2780 | goto out; | |
2781 | ||
2782 | } | |
f46b5a66 | 2783 | } |
c5c9cd4d SW |
2784 | } else if (type == BTRFS_FILE_EXTENT_INLINE) { |
2785 | u64 skip = 0; | |
2786 | u64 trim = 0; | |
2787 | if (off > key.offset) { | |
2788 | skip = off - key.offset; | |
2789 | new_key.offset += skip; | |
2790 | } | |
d397712b | 2791 | |
aa42ffd9 LB |
2792 | if (key.offset + datal > off + len) |
2793 | trim = key.offset + datal - (off + len); | |
d397712b | 2794 | |
c5c9cd4d | 2795 | if (comp && (skip || trim)) { |
c5c9cd4d | 2796 | ret = -EINVAL; |
a22285a6 | 2797 | btrfs_end_transaction(trans, root); |
c5c9cd4d SW |
2798 | goto out; |
2799 | } | |
2800 | size -= skip + trim; | |
2801 | datal -= skip + trim; | |
a22285a6 | 2802 | |
5dc562c5 | 2803 | ret = btrfs_drop_extents(trans, root, inode, |
a22285a6 YZ |
2804 | new_key.offset, |
2805 | new_key.offset + datal, | |
2671485d | 2806 | 1); |
79787eaa JM |
2807 | if (ret) { |
2808 | btrfs_abort_transaction(trans, root, | |
2809 | ret); | |
2810 | btrfs_end_transaction(trans, root); | |
2811 | goto out; | |
2812 | } | |
a22285a6 | 2813 | |
c5c9cd4d SW |
2814 | ret = btrfs_insert_empty_item(trans, root, path, |
2815 | &new_key, size); | |
79787eaa JM |
2816 | if (ret) { |
2817 | btrfs_abort_transaction(trans, root, | |
2818 | ret); | |
2819 | btrfs_end_transaction(trans, root); | |
2820 | goto out; | |
2821 | } | |
c5c9cd4d SW |
2822 | |
2823 | if (skip) { | |
d397712b CM |
2824 | u32 start = |
2825 | btrfs_file_extent_calc_inline_size(0); | |
c5c9cd4d SW |
2826 | memmove(buf+start, buf+start+skip, |
2827 | datal); | |
2828 | } | |
2829 | ||
2830 | leaf = path->nodes[0]; | |
2831 | slot = path->slots[0]; | |
2832 | write_extent_buffer(leaf, buf, | |
2833 | btrfs_item_ptr_offset(leaf, slot), | |
2834 | size); | |
2835 | inode_add_bytes(inode, datal); | |
f46b5a66 | 2836 | } |
c5c9cd4d SW |
2837 | |
2838 | btrfs_mark_buffer_dirty(leaf); | |
b3b4aa74 | 2839 | btrfs_release_path(path); |
c5c9cd4d | 2840 | |
0c4d2d95 | 2841 | inode_inc_iversion(inode); |
a22285a6 | 2842 | inode->i_mtime = inode->i_ctime = CURRENT_TIME; |
b5384d48 SW |
2843 | |
2844 | /* | |
2845 | * we round up to the block size at eof when | |
2846 | * determining which extents to clone above, | |
2847 | * but shouldn't round up the file size | |
2848 | */ | |
2849 | endoff = new_key.offset + datal; | |
5f3888ff LZ |
2850 | if (endoff > destoff+olen) |
2851 | endoff = destoff+olen; | |
b5384d48 SW |
2852 | if (endoff > inode->i_size) |
2853 | btrfs_i_size_write(inode, endoff); | |
2854 | ||
a22285a6 | 2855 | ret = btrfs_update_inode(trans, root, inode); |
79787eaa JM |
2856 | if (ret) { |
2857 | btrfs_abort_transaction(trans, root, ret); | |
2858 | btrfs_end_transaction(trans, root); | |
2859 | goto out; | |
2860 | } | |
2861 | ret = btrfs_end_transaction(trans, root); | |
a22285a6 | 2862 | } |
d397712b | 2863 | next: |
b3b4aa74 | 2864 | btrfs_release_path(path); |
f46b5a66 | 2865 | key.offset++; |
f46b5a66 | 2866 | } |
f46b5a66 CH |
2867 | ret = 0; |
2868 | out: | |
b3b4aa74 | 2869 | btrfs_release_path(path); |
aa42ffd9 | 2870 | unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1); |
f46b5a66 CH |
2871 | out_unlock: |
2872 | mutex_unlock(&src->i_mutex); | |
2873 | mutex_unlock(&inode->i_mutex); | |
ae01a0ab YZ |
2874 | vfree(buf); |
2875 | btrfs_free_path(path); | |
f46b5a66 | 2876 | out_fput: |
2903ff01 | 2877 | fdput(src_file); |
ab67b7c1 | 2878 | out_drop_write: |
2a79f17e | 2879 | mnt_drop_write_file(file); |
f46b5a66 CH |
2880 | return ret; |
2881 | } | |
2882 | ||
7a865e8a | 2883 | static long btrfs_ioctl_clone_range(struct file *file, void __user *argp) |
c5c9cd4d SW |
2884 | { |
2885 | struct btrfs_ioctl_clone_range_args args; | |
2886 | ||
7a865e8a | 2887 | if (copy_from_user(&args, argp, sizeof(args))) |
c5c9cd4d SW |
2888 | return -EFAULT; |
2889 | return btrfs_ioctl_clone(file, args.src_fd, args.src_offset, | |
2890 | args.src_length, args.dest_offset); | |
2891 | } | |
2892 | ||
f46b5a66 CH |
2893 | /* |
2894 | * there are many ways the trans_start and trans_end ioctls can lead | |
2895 | * to deadlocks. They should only be used by applications that | |
2896 | * basically own the machine, and have a very in depth understanding | |
2897 | * of all the possible deadlocks and enospc problems. | |
2898 | */ | |
b2950863 | 2899 | static long btrfs_ioctl_trans_start(struct file *file) |
f46b5a66 | 2900 | { |
496ad9aa | 2901 | struct inode *inode = file_inode(file); |
f46b5a66 CH |
2902 | struct btrfs_root *root = BTRFS_I(inode)->root; |
2903 | struct btrfs_trans_handle *trans; | |
1ab86aed | 2904 | int ret; |
f46b5a66 | 2905 | |
1ab86aed | 2906 | ret = -EPERM; |
df5b5520 | 2907 | if (!capable(CAP_SYS_ADMIN)) |
1ab86aed | 2908 | goto out; |
df5b5520 | 2909 | |
1ab86aed SW |
2910 | ret = -EINPROGRESS; |
2911 | if (file->private_data) | |
f46b5a66 | 2912 | goto out; |
9ca9ee09 | 2913 | |
b83cc969 LZ |
2914 | ret = -EROFS; |
2915 | if (btrfs_root_readonly(root)) | |
2916 | goto out; | |
2917 | ||
a561be71 | 2918 | ret = mnt_want_write_file(file); |
c146afad YZ |
2919 | if (ret) |
2920 | goto out; | |
2921 | ||
a4abeea4 | 2922 | atomic_inc(&root->fs_info->open_ioctl_trans); |
9ca9ee09 | 2923 | |
1ab86aed | 2924 | ret = -ENOMEM; |
7a7eaa40 | 2925 | trans = btrfs_start_ioctl_transaction(root); |
abd30bb0 | 2926 | if (IS_ERR(trans)) |
1ab86aed SW |
2927 | goto out_drop; |
2928 | ||
2929 | file->private_data = trans; | |
2930 | return 0; | |
2931 | ||
2932 | out_drop: | |
a4abeea4 | 2933 | atomic_dec(&root->fs_info->open_ioctl_trans); |
2a79f17e | 2934 | mnt_drop_write_file(file); |
f46b5a66 | 2935 | out: |
f46b5a66 CH |
2936 | return ret; |
2937 | } | |
2938 | ||
6ef5ed0d JB |
2939 | static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp) |
2940 | { | |
496ad9aa | 2941 | struct inode *inode = file_inode(file); |
6ef5ed0d JB |
2942 | struct btrfs_root *root = BTRFS_I(inode)->root; |
2943 | struct btrfs_root *new_root; | |
2944 | struct btrfs_dir_item *di; | |
2945 | struct btrfs_trans_handle *trans; | |
2946 | struct btrfs_path *path; | |
2947 | struct btrfs_key location; | |
2948 | struct btrfs_disk_key disk_key; | |
6ef5ed0d JB |
2949 | u64 objectid = 0; |
2950 | u64 dir_id; | |
3c04ce01 | 2951 | int ret; |
6ef5ed0d JB |
2952 | |
2953 | if (!capable(CAP_SYS_ADMIN)) | |
2954 | return -EPERM; | |
2955 | ||
3c04ce01 MX |
2956 | ret = mnt_want_write_file(file); |
2957 | if (ret) | |
2958 | return ret; | |
2959 | ||
2960 | if (copy_from_user(&objectid, argp, sizeof(objectid))) { | |
2961 | ret = -EFAULT; | |
2962 | goto out; | |
2963 | } | |
6ef5ed0d JB |
2964 | |
2965 | if (!objectid) | |
2966 | objectid = root->root_key.objectid; | |
2967 | ||
2968 | location.objectid = objectid; | |
2969 | location.type = BTRFS_ROOT_ITEM_KEY; | |
2970 | location.offset = (u64)-1; | |
2971 | ||
2972 | new_root = btrfs_read_fs_root_no_name(root->fs_info, &location); | |
3c04ce01 MX |
2973 | if (IS_ERR(new_root)) { |
2974 | ret = PTR_ERR(new_root); | |
2975 | goto out; | |
2976 | } | |
827a5cb0 | 2977 | if (!is_fstree(new_root->objectid)) { |
2978 | ret = -ENOENT; | |
2979 | goto out; | |
2980 | } | |
6ef5ed0d | 2981 | |
3c04ce01 MX |
2982 | if (btrfs_root_refs(&new_root->root_item) == 0) { |
2983 | ret = -ENOENT; | |
2984 | goto out; | |
2985 | } | |
6ef5ed0d JB |
2986 | |
2987 | path = btrfs_alloc_path(); | |
3c04ce01 MX |
2988 | if (!path) { |
2989 | ret = -ENOMEM; | |
2990 | goto out; | |
2991 | } | |
6ef5ed0d JB |
2992 | path->leave_spinning = 1; |
2993 | ||
2994 | trans = btrfs_start_transaction(root, 1); | |
98d5dc13 | 2995 | if (IS_ERR(trans)) { |
6ef5ed0d | 2996 | btrfs_free_path(path); |
3c04ce01 MX |
2997 | ret = PTR_ERR(trans); |
2998 | goto out; | |
6ef5ed0d JB |
2999 | } |
3000 | ||
6c41761f | 3001 | dir_id = btrfs_super_root_dir(root->fs_info->super_copy); |
6ef5ed0d JB |
3002 | di = btrfs_lookup_dir_item(trans, root->fs_info->tree_root, path, |
3003 | dir_id, "default", 7, 1); | |
cf1e99a4 | 3004 | if (IS_ERR_OR_NULL(di)) { |
6ef5ed0d JB |
3005 | btrfs_free_path(path); |
3006 | btrfs_end_transaction(trans, root); | |
3007 | printk(KERN_ERR "Umm, you don't have the default dir item, " | |
3008 | "this isn't going to work\n"); | |
3c04ce01 MX |
3009 | ret = -ENOENT; |
3010 | goto out; | |
6ef5ed0d JB |
3011 | } |
3012 | ||
3013 | btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key); | |
3014 | btrfs_set_dir_item_key(path->nodes[0], di, &disk_key); | |
3015 | btrfs_mark_buffer_dirty(path->nodes[0]); | |
3016 | btrfs_free_path(path); | |
3017 | ||
2b0ce2c2 | 3018 | btrfs_set_fs_incompat(root->fs_info, DEFAULT_SUBVOL); |
6ef5ed0d | 3019 | btrfs_end_transaction(trans, root); |
3c04ce01 MX |
3020 | out: |
3021 | mnt_drop_write_file(file); | |
3022 | return ret; | |
6ef5ed0d JB |
3023 | } |
3024 | ||
5af3e8cc SB |
3025 | void btrfs_get_block_group_info(struct list_head *groups_list, |
3026 | struct btrfs_ioctl_space_info *space) | |
bf5fc093 JB |
3027 | { |
3028 | struct btrfs_block_group_cache *block_group; | |
3029 | ||
3030 | space->total_bytes = 0; | |
3031 | space->used_bytes = 0; | |
3032 | space->flags = 0; | |
3033 | list_for_each_entry(block_group, groups_list, list) { | |
3034 | space->flags = block_group->flags; | |
3035 | space->total_bytes += block_group->key.offset; | |
3036 | space->used_bytes += | |
3037 | btrfs_block_group_used(&block_group->item); | |
3038 | } | |
3039 | } | |
3040 | ||
48a3b636 | 3041 | static long btrfs_ioctl_space_info(struct btrfs_root *root, void __user *arg) |
1406e432 JB |
3042 | { |
3043 | struct btrfs_ioctl_space_args space_args; | |
3044 | struct btrfs_ioctl_space_info space; | |
3045 | struct btrfs_ioctl_space_info *dest; | |
7fde62bf | 3046 | struct btrfs_ioctl_space_info *dest_orig; |
13f2696f | 3047 | struct btrfs_ioctl_space_info __user *user_dest; |
1406e432 | 3048 | struct btrfs_space_info *info; |
bf5fc093 JB |
3049 | u64 types[] = {BTRFS_BLOCK_GROUP_DATA, |
3050 | BTRFS_BLOCK_GROUP_SYSTEM, | |
3051 | BTRFS_BLOCK_GROUP_METADATA, | |
3052 | BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA}; | |
3053 | int num_types = 4; | |
7fde62bf | 3054 | int alloc_size; |
1406e432 | 3055 | int ret = 0; |
51788b1b | 3056 | u64 slot_count = 0; |
bf5fc093 | 3057 | int i, c; |
1406e432 JB |
3058 | |
3059 | if (copy_from_user(&space_args, | |
3060 | (struct btrfs_ioctl_space_args __user *)arg, | |
3061 | sizeof(space_args))) | |
3062 | return -EFAULT; | |
3063 | ||
bf5fc093 JB |
3064 | for (i = 0; i < num_types; i++) { |
3065 | struct btrfs_space_info *tmp; | |
3066 | ||
3067 | info = NULL; | |
3068 | rcu_read_lock(); | |
3069 | list_for_each_entry_rcu(tmp, &root->fs_info->space_info, | |
3070 | list) { | |
3071 | if (tmp->flags == types[i]) { | |
3072 | info = tmp; | |
3073 | break; | |
3074 | } | |
3075 | } | |
3076 | rcu_read_unlock(); | |
3077 | ||
3078 | if (!info) | |
3079 | continue; | |
3080 | ||
3081 | down_read(&info->groups_sem); | |
3082 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
3083 | if (!list_empty(&info->block_groups[c])) | |
3084 | slot_count++; | |
3085 | } | |
3086 | up_read(&info->groups_sem); | |
3087 | } | |
7fde62bf CM |
3088 | |
3089 | /* space_slots == 0 means they are asking for a count */ | |
3090 | if (space_args.space_slots == 0) { | |
3091 | space_args.total_spaces = slot_count; | |
3092 | goto out; | |
3093 | } | |
bf5fc093 | 3094 | |
51788b1b | 3095 | slot_count = min_t(u64, space_args.space_slots, slot_count); |
bf5fc093 | 3096 | |
7fde62bf | 3097 | alloc_size = sizeof(*dest) * slot_count; |
bf5fc093 | 3098 | |
7fde62bf CM |
3099 | /* we generally have at most 6 or so space infos, one for each raid |
3100 | * level. So, a whole page should be more than enough for everyone | |
3101 | */ | |
3102 | if (alloc_size > PAGE_CACHE_SIZE) | |
3103 | return -ENOMEM; | |
3104 | ||
1406e432 | 3105 | space_args.total_spaces = 0; |
7fde62bf CM |
3106 | dest = kmalloc(alloc_size, GFP_NOFS); |
3107 | if (!dest) | |
3108 | return -ENOMEM; | |
3109 | dest_orig = dest; | |
1406e432 | 3110 | |
7fde62bf | 3111 | /* now we have a buffer to copy into */ |
bf5fc093 JB |
3112 | for (i = 0; i < num_types; i++) { |
3113 | struct btrfs_space_info *tmp; | |
3114 | ||
51788b1b DR |
3115 | if (!slot_count) |
3116 | break; | |
3117 | ||
bf5fc093 JB |
3118 | info = NULL; |
3119 | rcu_read_lock(); | |
3120 | list_for_each_entry_rcu(tmp, &root->fs_info->space_info, | |
3121 | list) { | |
3122 | if (tmp->flags == types[i]) { | |
3123 | info = tmp; | |
3124 | break; | |
3125 | } | |
3126 | } | |
3127 | rcu_read_unlock(); | |
7fde62bf | 3128 | |
bf5fc093 JB |
3129 | if (!info) |
3130 | continue; | |
3131 | down_read(&info->groups_sem); | |
3132 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
3133 | if (!list_empty(&info->block_groups[c])) { | |
5af3e8cc SB |
3134 | btrfs_get_block_group_info( |
3135 | &info->block_groups[c], &space); | |
bf5fc093 JB |
3136 | memcpy(dest, &space, sizeof(space)); |
3137 | dest++; | |
3138 | space_args.total_spaces++; | |
51788b1b | 3139 | slot_count--; |
bf5fc093 | 3140 | } |
51788b1b DR |
3141 | if (!slot_count) |
3142 | break; | |
bf5fc093 JB |
3143 | } |
3144 | up_read(&info->groups_sem); | |
1406e432 | 3145 | } |
1406e432 | 3146 | |
2eec6c81 | 3147 | user_dest = (struct btrfs_ioctl_space_info __user *) |
7fde62bf CM |
3148 | (arg + sizeof(struct btrfs_ioctl_space_args)); |
3149 | ||
3150 | if (copy_to_user(user_dest, dest_orig, alloc_size)) | |
3151 | ret = -EFAULT; | |
3152 | ||
3153 | kfree(dest_orig); | |
3154 | out: | |
3155 | if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args))) | |
1406e432 JB |
3156 | ret = -EFAULT; |
3157 | ||
3158 | return ret; | |
3159 | } | |
3160 | ||
f46b5a66 CH |
3161 | /* |
3162 | * there are many ways the trans_start and trans_end ioctls can lead | |
3163 | * to deadlocks. They should only be used by applications that | |
3164 | * basically own the machine, and have a very in depth understanding | |
3165 | * of all the possible deadlocks and enospc problems. | |
3166 | */ | |
3167 | long btrfs_ioctl_trans_end(struct file *file) | |
3168 | { | |
496ad9aa | 3169 | struct inode *inode = file_inode(file); |
f46b5a66 CH |
3170 | struct btrfs_root *root = BTRFS_I(inode)->root; |
3171 | struct btrfs_trans_handle *trans; | |
f46b5a66 | 3172 | |
f46b5a66 | 3173 | trans = file->private_data; |
1ab86aed SW |
3174 | if (!trans) |
3175 | return -EINVAL; | |
b214107e | 3176 | file->private_data = NULL; |
9ca9ee09 | 3177 | |
1ab86aed SW |
3178 | btrfs_end_transaction(trans, root); |
3179 | ||
a4abeea4 | 3180 | atomic_dec(&root->fs_info->open_ioctl_trans); |
9ca9ee09 | 3181 | |
2a79f17e | 3182 | mnt_drop_write_file(file); |
1ab86aed | 3183 | return 0; |
f46b5a66 CH |
3184 | } |
3185 | ||
9a8c28be MX |
3186 | static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root, |
3187 | void __user *argp) | |
46204592 | 3188 | { |
46204592 SW |
3189 | struct btrfs_trans_handle *trans; |
3190 | u64 transid; | |
db5b493a | 3191 | int ret; |
46204592 | 3192 | |
d4edf39b | 3193 | trans = btrfs_attach_transaction_barrier(root); |
ff7c1d33 MX |
3194 | if (IS_ERR(trans)) { |
3195 | if (PTR_ERR(trans) != -ENOENT) | |
3196 | return PTR_ERR(trans); | |
3197 | ||
3198 | /* No running transaction, don't bother */ | |
3199 | transid = root->fs_info->last_trans_committed; | |
3200 | goto out; | |
3201 | } | |
46204592 | 3202 | transid = trans->transid; |
db5b493a | 3203 | ret = btrfs_commit_transaction_async(trans, root, 0); |
8b2b2d3c TI |
3204 | if (ret) { |
3205 | btrfs_end_transaction(trans, root); | |
db5b493a | 3206 | return ret; |
8b2b2d3c | 3207 | } |
ff7c1d33 | 3208 | out: |
46204592 SW |
3209 | if (argp) |
3210 | if (copy_to_user(argp, &transid, sizeof(transid))) | |
3211 | return -EFAULT; | |
3212 | return 0; | |
3213 | } | |
3214 | ||
9a8c28be MX |
3215 | static noinline long btrfs_ioctl_wait_sync(struct btrfs_root *root, |
3216 | void __user *argp) | |
46204592 | 3217 | { |
46204592 SW |
3218 | u64 transid; |
3219 | ||
3220 | if (argp) { | |
3221 | if (copy_from_user(&transid, argp, sizeof(transid))) | |
3222 | return -EFAULT; | |
3223 | } else { | |
3224 | transid = 0; /* current trans */ | |
3225 | } | |
3226 | return btrfs_wait_for_commit(root, transid); | |
3227 | } | |
3228 | ||
b8e95489 | 3229 | static long btrfs_ioctl_scrub(struct file *file, void __user *arg) |
475f6387 | 3230 | { |
496ad9aa | 3231 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
475f6387 | 3232 | struct btrfs_ioctl_scrub_args *sa; |
b8e95489 | 3233 | int ret; |
475f6387 JS |
3234 | |
3235 | if (!capable(CAP_SYS_ADMIN)) | |
3236 | return -EPERM; | |
3237 | ||
3238 | sa = memdup_user(arg, sizeof(*sa)); | |
3239 | if (IS_ERR(sa)) | |
3240 | return PTR_ERR(sa); | |
3241 | ||
b8e95489 MX |
3242 | if (!(sa->flags & BTRFS_SCRUB_READONLY)) { |
3243 | ret = mnt_want_write_file(file); | |
3244 | if (ret) | |
3245 | goto out; | |
3246 | } | |
3247 | ||
aa1b8cd4 | 3248 | ret = btrfs_scrub_dev(root->fs_info, sa->devid, sa->start, sa->end, |
63a212ab SB |
3249 | &sa->progress, sa->flags & BTRFS_SCRUB_READONLY, |
3250 | 0); | |
475f6387 JS |
3251 | |
3252 | if (copy_to_user(arg, sa, sizeof(*sa))) | |
3253 | ret = -EFAULT; | |
3254 | ||
b8e95489 MX |
3255 | if (!(sa->flags & BTRFS_SCRUB_READONLY)) |
3256 | mnt_drop_write_file(file); | |
3257 | out: | |
475f6387 JS |
3258 | kfree(sa); |
3259 | return ret; | |
3260 | } | |
3261 | ||
3262 | static long btrfs_ioctl_scrub_cancel(struct btrfs_root *root, void __user *arg) | |
3263 | { | |
3264 | if (!capable(CAP_SYS_ADMIN)) | |
3265 | return -EPERM; | |
3266 | ||
aa1b8cd4 | 3267 | return btrfs_scrub_cancel(root->fs_info); |
475f6387 JS |
3268 | } |
3269 | ||
3270 | static long btrfs_ioctl_scrub_progress(struct btrfs_root *root, | |
3271 | void __user *arg) | |
3272 | { | |
3273 | struct btrfs_ioctl_scrub_args *sa; | |
3274 | int ret; | |
3275 | ||
3276 | if (!capable(CAP_SYS_ADMIN)) | |
3277 | return -EPERM; | |
3278 | ||
3279 | sa = memdup_user(arg, sizeof(*sa)); | |
3280 | if (IS_ERR(sa)) | |
3281 | return PTR_ERR(sa); | |
3282 | ||
3283 | ret = btrfs_scrub_progress(root, sa->devid, &sa->progress); | |
3284 | ||
3285 | if (copy_to_user(arg, sa, sizeof(*sa))) | |
3286 | ret = -EFAULT; | |
3287 | ||
3288 | kfree(sa); | |
3289 | return ret; | |
3290 | } | |
3291 | ||
c11d2c23 | 3292 | static long btrfs_ioctl_get_dev_stats(struct btrfs_root *root, |
b27f7c0c | 3293 | void __user *arg) |
c11d2c23 SB |
3294 | { |
3295 | struct btrfs_ioctl_get_dev_stats *sa; | |
3296 | int ret; | |
3297 | ||
c11d2c23 SB |
3298 | sa = memdup_user(arg, sizeof(*sa)); |
3299 | if (IS_ERR(sa)) | |
3300 | return PTR_ERR(sa); | |
3301 | ||
b27f7c0c DS |
3302 | if ((sa->flags & BTRFS_DEV_STATS_RESET) && !capable(CAP_SYS_ADMIN)) { |
3303 | kfree(sa); | |
3304 | return -EPERM; | |
3305 | } | |
3306 | ||
3307 | ret = btrfs_get_dev_stats(root, sa); | |
c11d2c23 SB |
3308 | |
3309 | if (copy_to_user(arg, sa, sizeof(*sa))) | |
3310 | ret = -EFAULT; | |
3311 | ||
3312 | kfree(sa); | |
3313 | return ret; | |
3314 | } | |
3315 | ||
3f6bcfbd SB |
3316 | static long btrfs_ioctl_dev_replace(struct btrfs_root *root, void __user *arg) |
3317 | { | |
3318 | struct btrfs_ioctl_dev_replace_args *p; | |
3319 | int ret; | |
3320 | ||
3321 | if (!capable(CAP_SYS_ADMIN)) | |
3322 | return -EPERM; | |
3323 | ||
3324 | p = memdup_user(arg, sizeof(*p)); | |
3325 | if (IS_ERR(p)) | |
3326 | return PTR_ERR(p); | |
3327 | ||
3328 | switch (p->cmd) { | |
3329 | case BTRFS_IOCTL_DEV_REPLACE_CMD_START: | |
42cc8e56 SB |
3330 | if (root->fs_info->sb->s_flags & MS_RDONLY) |
3331 | return -EROFS; | |
3332 | ||
3f6bcfbd SB |
3333 | if (atomic_xchg( |
3334 | &root->fs_info->mutually_exclusive_operation_running, | |
3335 | 1)) { | |
3336 | pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n"); | |
3337 | ret = -EINPROGRESS; | |
3338 | } else { | |
3339 | ret = btrfs_dev_replace_start(root, p); | |
3340 | atomic_set( | |
3341 | &root->fs_info->mutually_exclusive_operation_running, | |
3342 | 0); | |
3343 | } | |
3344 | break; | |
3345 | case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS: | |
3346 | btrfs_dev_replace_status(root->fs_info, p); | |
3347 | ret = 0; | |
3348 | break; | |
3349 | case BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL: | |
3350 | ret = btrfs_dev_replace_cancel(root->fs_info, p); | |
3351 | break; | |
3352 | default: | |
3353 | ret = -EINVAL; | |
3354 | break; | |
3355 | } | |
3356 | ||
3357 | if (copy_to_user(arg, p, sizeof(*p))) | |
3358 | ret = -EFAULT; | |
3359 | ||
3360 | kfree(p); | |
3361 | return ret; | |
3362 | } | |
3363 | ||
d7728c96 JS |
3364 | static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg) |
3365 | { | |
3366 | int ret = 0; | |
3367 | int i; | |
740c3d22 | 3368 | u64 rel_ptr; |
d7728c96 | 3369 | int size; |
806468f8 | 3370 | struct btrfs_ioctl_ino_path_args *ipa = NULL; |
d7728c96 JS |
3371 | struct inode_fs_paths *ipath = NULL; |
3372 | struct btrfs_path *path; | |
3373 | ||
82b22ac8 | 3374 | if (!capable(CAP_DAC_READ_SEARCH)) |
d7728c96 JS |
3375 | return -EPERM; |
3376 | ||
3377 | path = btrfs_alloc_path(); | |
3378 | if (!path) { | |
3379 | ret = -ENOMEM; | |
3380 | goto out; | |
3381 | } | |
3382 | ||
3383 | ipa = memdup_user(arg, sizeof(*ipa)); | |
3384 | if (IS_ERR(ipa)) { | |
3385 | ret = PTR_ERR(ipa); | |
3386 | ipa = NULL; | |
3387 | goto out; | |
3388 | } | |
3389 | ||
3390 | size = min_t(u32, ipa->size, 4096); | |
3391 | ipath = init_ipath(size, root, path); | |
3392 | if (IS_ERR(ipath)) { | |
3393 | ret = PTR_ERR(ipath); | |
3394 | ipath = NULL; | |
3395 | goto out; | |
3396 | } | |
3397 | ||
3398 | ret = paths_from_inode(ipa->inum, ipath); | |
3399 | if (ret < 0) | |
3400 | goto out; | |
3401 | ||
3402 | for (i = 0; i < ipath->fspath->elem_cnt; ++i) { | |
745c4d8e JM |
3403 | rel_ptr = ipath->fspath->val[i] - |
3404 | (u64)(unsigned long)ipath->fspath->val; | |
740c3d22 | 3405 | ipath->fspath->val[i] = rel_ptr; |
d7728c96 JS |
3406 | } |
3407 | ||
745c4d8e JM |
3408 | ret = copy_to_user((void *)(unsigned long)ipa->fspath, |
3409 | (void *)(unsigned long)ipath->fspath, size); | |
d7728c96 JS |
3410 | if (ret) { |
3411 | ret = -EFAULT; | |
3412 | goto out; | |
3413 | } | |
3414 | ||
3415 | out: | |
3416 | btrfs_free_path(path); | |
3417 | free_ipath(ipath); | |
3418 | kfree(ipa); | |
3419 | ||
3420 | return ret; | |
3421 | } | |
3422 | ||
3423 | static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx) | |
3424 | { | |
3425 | struct btrfs_data_container *inodes = ctx; | |
3426 | const size_t c = 3 * sizeof(u64); | |
3427 | ||
3428 | if (inodes->bytes_left >= c) { | |
3429 | inodes->bytes_left -= c; | |
3430 | inodes->val[inodes->elem_cnt] = inum; | |
3431 | inodes->val[inodes->elem_cnt + 1] = offset; | |
3432 | inodes->val[inodes->elem_cnt + 2] = root; | |
3433 | inodes->elem_cnt += 3; | |
3434 | } else { | |
3435 | inodes->bytes_missing += c - inodes->bytes_left; | |
3436 | inodes->bytes_left = 0; | |
3437 | inodes->elem_missed += 3; | |
3438 | } | |
3439 | ||
3440 | return 0; | |
3441 | } | |
3442 | ||
3443 | static long btrfs_ioctl_logical_to_ino(struct btrfs_root *root, | |
3444 | void __user *arg) | |
3445 | { | |
3446 | int ret = 0; | |
3447 | int size; | |
d7728c96 JS |
3448 | struct btrfs_ioctl_logical_ino_args *loi; |
3449 | struct btrfs_data_container *inodes = NULL; | |
3450 | struct btrfs_path *path = NULL; | |
d7728c96 JS |
3451 | |
3452 | if (!capable(CAP_SYS_ADMIN)) | |
3453 | return -EPERM; | |
3454 | ||
3455 | loi = memdup_user(arg, sizeof(*loi)); | |
3456 | if (IS_ERR(loi)) { | |
3457 | ret = PTR_ERR(loi); | |
3458 | loi = NULL; | |
3459 | goto out; | |
3460 | } | |
3461 | ||
3462 | path = btrfs_alloc_path(); | |
3463 | if (!path) { | |
3464 | ret = -ENOMEM; | |
3465 | goto out; | |
3466 | } | |
3467 | ||
425d17a2 | 3468 | size = min_t(u32, loi->size, 64 * 1024); |
d7728c96 JS |
3469 | inodes = init_data_container(size); |
3470 | if (IS_ERR(inodes)) { | |
3471 | ret = PTR_ERR(inodes); | |
3472 | inodes = NULL; | |
3473 | goto out; | |
3474 | } | |
3475 | ||
df031f07 LB |
3476 | ret = iterate_inodes_from_logical(loi->logical, root->fs_info, path, |
3477 | build_ino_list, inodes); | |
3478 | if (ret == -EINVAL) | |
d7728c96 JS |
3479 | ret = -ENOENT; |
3480 | if (ret < 0) | |
3481 | goto out; | |
3482 | ||
745c4d8e JM |
3483 | ret = copy_to_user((void *)(unsigned long)loi->inodes, |
3484 | (void *)(unsigned long)inodes, size); | |
d7728c96 JS |
3485 | if (ret) |
3486 | ret = -EFAULT; | |
3487 | ||
3488 | out: | |
3489 | btrfs_free_path(path); | |
425d17a2 | 3490 | vfree(inodes); |
d7728c96 JS |
3491 | kfree(loi); |
3492 | ||
3493 | return ret; | |
3494 | } | |
3495 | ||
19a39dce | 3496 | void update_ioctl_balance_args(struct btrfs_fs_info *fs_info, int lock, |
c9e9f97b ID |
3497 | struct btrfs_ioctl_balance_args *bargs) |
3498 | { | |
3499 | struct btrfs_balance_control *bctl = fs_info->balance_ctl; | |
3500 | ||
3501 | bargs->flags = bctl->flags; | |
3502 | ||
837d5b6e ID |
3503 | if (atomic_read(&fs_info->balance_running)) |
3504 | bargs->state |= BTRFS_BALANCE_STATE_RUNNING; | |
3505 | if (atomic_read(&fs_info->balance_pause_req)) | |
3506 | bargs->state |= BTRFS_BALANCE_STATE_PAUSE_REQ; | |
a7e99c69 ID |
3507 | if (atomic_read(&fs_info->balance_cancel_req)) |
3508 | bargs->state |= BTRFS_BALANCE_STATE_CANCEL_REQ; | |
837d5b6e | 3509 | |
c9e9f97b ID |
3510 | memcpy(&bargs->data, &bctl->data, sizeof(bargs->data)); |
3511 | memcpy(&bargs->meta, &bctl->meta, sizeof(bargs->meta)); | |
3512 | memcpy(&bargs->sys, &bctl->sys, sizeof(bargs->sys)); | |
19a39dce ID |
3513 | |
3514 | if (lock) { | |
3515 | spin_lock(&fs_info->balance_lock); | |
3516 | memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat)); | |
3517 | spin_unlock(&fs_info->balance_lock); | |
3518 | } else { | |
3519 | memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat)); | |
3520 | } | |
c9e9f97b ID |
3521 | } |
3522 | ||
9ba1f6e4 | 3523 | static long btrfs_ioctl_balance(struct file *file, void __user *arg) |
c9e9f97b | 3524 | { |
496ad9aa | 3525 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
c9e9f97b ID |
3526 | struct btrfs_fs_info *fs_info = root->fs_info; |
3527 | struct btrfs_ioctl_balance_args *bargs; | |
3528 | struct btrfs_balance_control *bctl; | |
ed0fb78f | 3529 | bool need_unlock; /* for mut. excl. ops lock */ |
c9e9f97b ID |
3530 | int ret; |
3531 | ||
3532 | if (!capable(CAP_SYS_ADMIN)) | |
3533 | return -EPERM; | |
3534 | ||
e54bfa31 | 3535 | ret = mnt_want_write_file(file); |
9ba1f6e4 LB |
3536 | if (ret) |
3537 | return ret; | |
3538 | ||
ed0fb78f ID |
3539 | again: |
3540 | if (!atomic_xchg(&fs_info->mutually_exclusive_operation_running, 1)) { | |
3541 | mutex_lock(&fs_info->volume_mutex); | |
3542 | mutex_lock(&fs_info->balance_mutex); | |
3543 | need_unlock = true; | |
3544 | goto locked; | |
3545 | } | |
3546 | ||
3547 | /* | |
3548 | * mut. excl. ops lock is locked. Three possibilites: | |
3549 | * (1) some other op is running | |
3550 | * (2) balance is running | |
3551 | * (3) balance is paused -- special case (think resume) | |
3552 | */ | |
c9e9f97b | 3553 | mutex_lock(&fs_info->balance_mutex); |
ed0fb78f ID |
3554 | if (fs_info->balance_ctl) { |
3555 | /* this is either (2) or (3) */ | |
3556 | if (!atomic_read(&fs_info->balance_running)) { | |
3557 | mutex_unlock(&fs_info->balance_mutex); | |
3558 | if (!mutex_trylock(&fs_info->volume_mutex)) | |
3559 | goto again; | |
3560 | mutex_lock(&fs_info->balance_mutex); | |
3561 | ||
3562 | if (fs_info->balance_ctl && | |
3563 | !atomic_read(&fs_info->balance_running)) { | |
3564 | /* this is (3) */ | |
3565 | need_unlock = false; | |
3566 | goto locked; | |
3567 | } | |
3568 | ||
3569 | mutex_unlock(&fs_info->balance_mutex); | |
3570 | mutex_unlock(&fs_info->volume_mutex); | |
3571 | goto again; | |
3572 | } else { | |
3573 | /* this is (2) */ | |
3574 | mutex_unlock(&fs_info->balance_mutex); | |
3575 | ret = -EINPROGRESS; | |
3576 | goto out; | |
3577 | } | |
3578 | } else { | |
3579 | /* this is (1) */ | |
3580 | mutex_unlock(&fs_info->balance_mutex); | |
3581 | pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n"); | |
3582 | ret = -EINVAL; | |
3583 | goto out; | |
3584 | } | |
3585 | ||
3586 | locked: | |
3587 | BUG_ON(!atomic_read(&fs_info->mutually_exclusive_operation_running)); | |
c9e9f97b ID |
3588 | |
3589 | if (arg) { | |
3590 | bargs = memdup_user(arg, sizeof(*bargs)); | |
3591 | if (IS_ERR(bargs)) { | |
3592 | ret = PTR_ERR(bargs); | |
ed0fb78f | 3593 | goto out_unlock; |
c9e9f97b | 3594 | } |
de322263 ID |
3595 | |
3596 | if (bargs->flags & BTRFS_BALANCE_RESUME) { | |
3597 | if (!fs_info->balance_ctl) { | |
3598 | ret = -ENOTCONN; | |
3599 | goto out_bargs; | |
3600 | } | |
3601 | ||
3602 | bctl = fs_info->balance_ctl; | |
3603 | spin_lock(&fs_info->balance_lock); | |
3604 | bctl->flags |= BTRFS_BALANCE_RESUME; | |
3605 | spin_unlock(&fs_info->balance_lock); | |
3606 | ||
3607 | goto do_balance; | |
3608 | } | |
c9e9f97b ID |
3609 | } else { |
3610 | bargs = NULL; | |
3611 | } | |
3612 | ||
ed0fb78f | 3613 | if (fs_info->balance_ctl) { |
837d5b6e ID |
3614 | ret = -EINPROGRESS; |
3615 | goto out_bargs; | |
3616 | } | |
3617 | ||
c9e9f97b ID |
3618 | bctl = kzalloc(sizeof(*bctl), GFP_NOFS); |
3619 | if (!bctl) { | |
3620 | ret = -ENOMEM; | |
3621 | goto out_bargs; | |
3622 | } | |
3623 | ||
3624 | bctl->fs_info = fs_info; | |
3625 | if (arg) { | |
3626 | memcpy(&bctl->data, &bargs->data, sizeof(bctl->data)); | |
3627 | memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta)); | |
3628 | memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys)); | |
3629 | ||
3630 | bctl->flags = bargs->flags; | |
f43ffb60 ID |
3631 | } else { |
3632 | /* balance everything - no filters */ | |
3633 | bctl->flags |= BTRFS_BALANCE_TYPE_MASK; | |
c9e9f97b ID |
3634 | } |
3635 | ||
de322263 | 3636 | do_balance: |
c9e9f97b | 3637 | /* |
ed0fb78f ID |
3638 | * Ownership of bctl and mutually_exclusive_operation_running |
3639 | * goes to to btrfs_balance. bctl is freed in __cancel_balance, | |
3640 | * or, if restriper was paused all the way until unmount, in | |
3641 | * free_fs_info. mutually_exclusive_operation_running is | |
3642 | * cleared in __cancel_balance. | |
c9e9f97b | 3643 | */ |
ed0fb78f ID |
3644 | need_unlock = false; |
3645 | ||
3646 | ret = btrfs_balance(bctl, bargs); | |
3647 | ||
c9e9f97b ID |
3648 | if (arg) { |
3649 | if (copy_to_user(arg, bargs, sizeof(*bargs))) | |
3650 | ret = -EFAULT; | |
3651 | } | |
3652 | ||
3653 | out_bargs: | |
3654 | kfree(bargs); | |
ed0fb78f | 3655 | out_unlock: |
c9e9f97b ID |
3656 | mutex_unlock(&fs_info->balance_mutex); |
3657 | mutex_unlock(&fs_info->volume_mutex); | |
ed0fb78f ID |
3658 | if (need_unlock) |
3659 | atomic_set(&fs_info->mutually_exclusive_operation_running, 0); | |
3660 | out: | |
e54bfa31 | 3661 | mnt_drop_write_file(file); |
c9e9f97b ID |
3662 | return ret; |
3663 | } | |
3664 | ||
837d5b6e ID |
3665 | static long btrfs_ioctl_balance_ctl(struct btrfs_root *root, int cmd) |
3666 | { | |
3667 | if (!capable(CAP_SYS_ADMIN)) | |
3668 | return -EPERM; | |
3669 | ||
3670 | switch (cmd) { | |
3671 | case BTRFS_BALANCE_CTL_PAUSE: | |
3672 | return btrfs_pause_balance(root->fs_info); | |
a7e99c69 ID |
3673 | case BTRFS_BALANCE_CTL_CANCEL: |
3674 | return btrfs_cancel_balance(root->fs_info); | |
837d5b6e ID |
3675 | } |
3676 | ||
3677 | return -EINVAL; | |
3678 | } | |
3679 | ||
19a39dce ID |
3680 | static long btrfs_ioctl_balance_progress(struct btrfs_root *root, |
3681 | void __user *arg) | |
3682 | { | |
3683 | struct btrfs_fs_info *fs_info = root->fs_info; | |
3684 | struct btrfs_ioctl_balance_args *bargs; | |
3685 | int ret = 0; | |
3686 | ||
3687 | if (!capable(CAP_SYS_ADMIN)) | |
3688 | return -EPERM; | |
3689 | ||
3690 | mutex_lock(&fs_info->balance_mutex); | |
3691 | if (!fs_info->balance_ctl) { | |
3692 | ret = -ENOTCONN; | |
3693 | goto out; | |
3694 | } | |
3695 | ||
3696 | bargs = kzalloc(sizeof(*bargs), GFP_NOFS); | |
3697 | if (!bargs) { | |
3698 | ret = -ENOMEM; | |
3699 | goto out; | |
3700 | } | |
3701 | ||
3702 | update_ioctl_balance_args(fs_info, 1, bargs); | |
3703 | ||
3704 | if (copy_to_user(arg, bargs, sizeof(*bargs))) | |
3705 | ret = -EFAULT; | |
3706 | ||
3707 | kfree(bargs); | |
3708 | out: | |
3709 | mutex_unlock(&fs_info->balance_mutex); | |
3710 | return ret; | |
3711 | } | |
3712 | ||
905b0dda | 3713 | static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg) |
5d13a37b | 3714 | { |
496ad9aa | 3715 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
5d13a37b AJ |
3716 | struct btrfs_ioctl_quota_ctl_args *sa; |
3717 | struct btrfs_trans_handle *trans = NULL; | |
3718 | int ret; | |
3719 | int err; | |
3720 | ||
3721 | if (!capable(CAP_SYS_ADMIN)) | |
3722 | return -EPERM; | |
3723 | ||
905b0dda MX |
3724 | ret = mnt_want_write_file(file); |
3725 | if (ret) | |
3726 | return ret; | |
5d13a37b AJ |
3727 | |
3728 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
3729 | if (IS_ERR(sa)) { |
3730 | ret = PTR_ERR(sa); | |
3731 | goto drop_write; | |
3732 | } | |
5d13a37b | 3733 | |
7708f029 | 3734 | down_write(&root->fs_info->subvol_sem); |
2f232036 JS |
3735 | trans = btrfs_start_transaction(root->fs_info->tree_root, 2); |
3736 | if (IS_ERR(trans)) { | |
3737 | ret = PTR_ERR(trans); | |
3738 | goto out; | |
5d13a37b AJ |
3739 | } |
3740 | ||
3741 | switch (sa->cmd) { | |
3742 | case BTRFS_QUOTA_CTL_ENABLE: | |
3743 | ret = btrfs_quota_enable(trans, root->fs_info); | |
3744 | break; | |
3745 | case BTRFS_QUOTA_CTL_DISABLE: | |
3746 | ret = btrfs_quota_disable(trans, root->fs_info); | |
3747 | break; | |
5d13a37b AJ |
3748 | default: |
3749 | ret = -EINVAL; | |
3750 | break; | |
3751 | } | |
3752 | ||
3753 | if (copy_to_user(arg, sa, sizeof(*sa))) | |
3754 | ret = -EFAULT; | |
3755 | ||
2f232036 JS |
3756 | err = btrfs_commit_transaction(trans, root->fs_info->tree_root); |
3757 | if (err && !ret) | |
3758 | ret = err; | |
5d13a37b AJ |
3759 | out: |
3760 | kfree(sa); | |
7708f029 | 3761 | up_write(&root->fs_info->subvol_sem); |
905b0dda MX |
3762 | drop_write: |
3763 | mnt_drop_write_file(file); | |
5d13a37b AJ |
3764 | return ret; |
3765 | } | |
3766 | ||
905b0dda | 3767 | static long btrfs_ioctl_qgroup_assign(struct file *file, void __user *arg) |
5d13a37b | 3768 | { |
496ad9aa | 3769 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
5d13a37b AJ |
3770 | struct btrfs_ioctl_qgroup_assign_args *sa; |
3771 | struct btrfs_trans_handle *trans; | |
3772 | int ret; | |
3773 | int err; | |
3774 | ||
3775 | if (!capable(CAP_SYS_ADMIN)) | |
3776 | return -EPERM; | |
3777 | ||
905b0dda MX |
3778 | ret = mnt_want_write_file(file); |
3779 | if (ret) | |
3780 | return ret; | |
5d13a37b AJ |
3781 | |
3782 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
3783 | if (IS_ERR(sa)) { |
3784 | ret = PTR_ERR(sa); | |
3785 | goto drop_write; | |
3786 | } | |
5d13a37b AJ |
3787 | |
3788 | trans = btrfs_join_transaction(root); | |
3789 | if (IS_ERR(trans)) { | |
3790 | ret = PTR_ERR(trans); | |
3791 | goto out; | |
3792 | } | |
3793 | ||
3794 | /* FIXME: check if the IDs really exist */ | |
3795 | if (sa->assign) { | |
3796 | ret = btrfs_add_qgroup_relation(trans, root->fs_info, | |
3797 | sa->src, sa->dst); | |
3798 | } else { | |
3799 | ret = btrfs_del_qgroup_relation(trans, root->fs_info, | |
3800 | sa->src, sa->dst); | |
3801 | } | |
3802 | ||
3803 | err = btrfs_end_transaction(trans, root); | |
3804 | if (err && !ret) | |
3805 | ret = err; | |
3806 | ||
3807 | out: | |
3808 | kfree(sa); | |
905b0dda MX |
3809 | drop_write: |
3810 | mnt_drop_write_file(file); | |
5d13a37b AJ |
3811 | return ret; |
3812 | } | |
3813 | ||
905b0dda | 3814 | static long btrfs_ioctl_qgroup_create(struct file *file, void __user *arg) |
5d13a37b | 3815 | { |
496ad9aa | 3816 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
5d13a37b AJ |
3817 | struct btrfs_ioctl_qgroup_create_args *sa; |
3818 | struct btrfs_trans_handle *trans; | |
3819 | int ret; | |
3820 | int err; | |
3821 | ||
3822 | if (!capable(CAP_SYS_ADMIN)) | |
3823 | return -EPERM; | |
3824 | ||
905b0dda MX |
3825 | ret = mnt_want_write_file(file); |
3826 | if (ret) | |
3827 | return ret; | |
5d13a37b AJ |
3828 | |
3829 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
3830 | if (IS_ERR(sa)) { |
3831 | ret = PTR_ERR(sa); | |
3832 | goto drop_write; | |
3833 | } | |
5d13a37b | 3834 | |
d86e56cf MX |
3835 | if (!sa->qgroupid) { |
3836 | ret = -EINVAL; | |
3837 | goto out; | |
3838 | } | |
3839 | ||
5d13a37b AJ |
3840 | trans = btrfs_join_transaction(root); |
3841 | if (IS_ERR(trans)) { | |
3842 | ret = PTR_ERR(trans); | |
3843 | goto out; | |
3844 | } | |
3845 | ||
3846 | /* FIXME: check if the IDs really exist */ | |
3847 | if (sa->create) { | |
3848 | ret = btrfs_create_qgroup(trans, root->fs_info, sa->qgroupid, | |
3849 | NULL); | |
3850 | } else { | |
3851 | ret = btrfs_remove_qgroup(trans, root->fs_info, sa->qgroupid); | |
3852 | } | |
3853 | ||
3854 | err = btrfs_end_transaction(trans, root); | |
3855 | if (err && !ret) | |
3856 | ret = err; | |
3857 | ||
3858 | out: | |
3859 | kfree(sa); | |
905b0dda MX |
3860 | drop_write: |
3861 | mnt_drop_write_file(file); | |
5d13a37b AJ |
3862 | return ret; |
3863 | } | |
3864 | ||
905b0dda | 3865 | static long btrfs_ioctl_qgroup_limit(struct file *file, void __user *arg) |
5d13a37b | 3866 | { |
496ad9aa | 3867 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
5d13a37b AJ |
3868 | struct btrfs_ioctl_qgroup_limit_args *sa; |
3869 | struct btrfs_trans_handle *trans; | |
3870 | int ret; | |
3871 | int err; | |
3872 | u64 qgroupid; | |
3873 | ||
3874 | if (!capable(CAP_SYS_ADMIN)) | |
3875 | return -EPERM; | |
3876 | ||
905b0dda MX |
3877 | ret = mnt_want_write_file(file); |
3878 | if (ret) | |
3879 | return ret; | |
5d13a37b AJ |
3880 | |
3881 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
3882 | if (IS_ERR(sa)) { |
3883 | ret = PTR_ERR(sa); | |
3884 | goto drop_write; | |
3885 | } | |
5d13a37b AJ |
3886 | |
3887 | trans = btrfs_join_transaction(root); | |
3888 | if (IS_ERR(trans)) { | |
3889 | ret = PTR_ERR(trans); | |
3890 | goto out; | |
3891 | } | |
3892 | ||
3893 | qgroupid = sa->qgroupid; | |
3894 | if (!qgroupid) { | |
3895 | /* take the current subvol as qgroup */ | |
3896 | qgroupid = root->root_key.objectid; | |
3897 | } | |
3898 | ||
3899 | /* FIXME: check if the IDs really exist */ | |
3900 | ret = btrfs_limit_qgroup(trans, root->fs_info, qgroupid, &sa->lim); | |
3901 | ||
3902 | err = btrfs_end_transaction(trans, root); | |
3903 | if (err && !ret) | |
3904 | ret = err; | |
3905 | ||
3906 | out: | |
3907 | kfree(sa); | |
905b0dda MX |
3908 | drop_write: |
3909 | mnt_drop_write_file(file); | |
5d13a37b AJ |
3910 | return ret; |
3911 | } | |
3912 | ||
2f232036 JS |
3913 | static long btrfs_ioctl_quota_rescan(struct file *file, void __user *arg) |
3914 | { | |
3915 | struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root; | |
3916 | struct btrfs_ioctl_quota_rescan_args *qsa; | |
3917 | int ret; | |
3918 | ||
3919 | if (!capable(CAP_SYS_ADMIN)) | |
3920 | return -EPERM; | |
3921 | ||
3922 | ret = mnt_want_write_file(file); | |
3923 | if (ret) | |
3924 | return ret; | |
3925 | ||
3926 | qsa = memdup_user(arg, sizeof(*qsa)); | |
3927 | if (IS_ERR(qsa)) { | |
3928 | ret = PTR_ERR(qsa); | |
3929 | goto drop_write; | |
3930 | } | |
3931 | ||
3932 | if (qsa->flags) { | |
3933 | ret = -EINVAL; | |
3934 | goto out; | |
3935 | } | |
3936 | ||
3937 | ret = btrfs_qgroup_rescan(root->fs_info); | |
3938 | ||
3939 | out: | |
3940 | kfree(qsa); | |
3941 | drop_write: | |
3942 | mnt_drop_write_file(file); | |
3943 | return ret; | |
3944 | } | |
3945 | ||
3946 | static long btrfs_ioctl_quota_rescan_status(struct file *file, void __user *arg) | |
3947 | { | |
3948 | struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root; | |
3949 | struct btrfs_ioctl_quota_rescan_args *qsa; | |
3950 | int ret = 0; | |
3951 | ||
3952 | if (!capable(CAP_SYS_ADMIN)) | |
3953 | return -EPERM; | |
3954 | ||
3955 | qsa = kzalloc(sizeof(*qsa), GFP_NOFS); | |
3956 | if (!qsa) | |
3957 | return -ENOMEM; | |
3958 | ||
3959 | if (root->fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) { | |
3960 | qsa->flags = 1; | |
3961 | qsa->progress = root->fs_info->qgroup_rescan_progress.objectid; | |
3962 | } | |
3963 | ||
3964 | if (copy_to_user(arg, qsa, sizeof(*qsa))) | |
3965 | ret = -EFAULT; | |
3966 | ||
3967 | kfree(qsa); | |
3968 | return ret; | |
3969 | } | |
3970 | ||
8ea05e3a AB |
3971 | static long btrfs_ioctl_set_received_subvol(struct file *file, |
3972 | void __user *arg) | |
3973 | { | |
3974 | struct btrfs_ioctl_received_subvol_args *sa = NULL; | |
496ad9aa | 3975 | struct inode *inode = file_inode(file); |
8ea05e3a AB |
3976 | struct btrfs_root *root = BTRFS_I(inode)->root; |
3977 | struct btrfs_root_item *root_item = &root->root_item; | |
3978 | struct btrfs_trans_handle *trans; | |
3979 | struct timespec ct = CURRENT_TIME; | |
3980 | int ret = 0; | |
3981 | ||
3982 | ret = mnt_want_write_file(file); | |
3983 | if (ret < 0) | |
3984 | return ret; | |
3985 | ||
3986 | down_write(&root->fs_info->subvol_sem); | |
3987 | ||
3988 | if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) { | |
3989 | ret = -EINVAL; | |
3990 | goto out; | |
3991 | } | |
3992 | ||
3993 | if (btrfs_root_readonly(root)) { | |
3994 | ret = -EROFS; | |
3995 | goto out; | |
3996 | } | |
3997 | ||
3998 | if (!inode_owner_or_capable(inode)) { | |
3999 | ret = -EACCES; | |
4000 | goto out; | |
4001 | } | |
4002 | ||
4003 | sa = memdup_user(arg, sizeof(*sa)); | |
4004 | if (IS_ERR(sa)) { | |
4005 | ret = PTR_ERR(sa); | |
4006 | sa = NULL; | |
4007 | goto out; | |
4008 | } | |
4009 | ||
4010 | trans = btrfs_start_transaction(root, 1); | |
4011 | if (IS_ERR(trans)) { | |
4012 | ret = PTR_ERR(trans); | |
4013 | trans = NULL; | |
4014 | goto out; | |
4015 | } | |
4016 | ||
4017 | sa->rtransid = trans->transid; | |
4018 | sa->rtime.sec = ct.tv_sec; | |
4019 | sa->rtime.nsec = ct.tv_nsec; | |
4020 | ||
4021 | memcpy(root_item->received_uuid, sa->uuid, BTRFS_UUID_SIZE); | |
4022 | btrfs_set_root_stransid(root_item, sa->stransid); | |
4023 | btrfs_set_root_rtransid(root_item, sa->rtransid); | |
4024 | root_item->stime.sec = cpu_to_le64(sa->stime.sec); | |
4025 | root_item->stime.nsec = cpu_to_le32(sa->stime.nsec); | |
4026 | root_item->rtime.sec = cpu_to_le64(sa->rtime.sec); | |
4027 | root_item->rtime.nsec = cpu_to_le32(sa->rtime.nsec); | |
4028 | ||
4029 | ret = btrfs_update_root(trans, root->fs_info->tree_root, | |
4030 | &root->root_key, &root->root_item); | |
4031 | if (ret < 0) { | |
4032 | btrfs_end_transaction(trans, root); | |
4033 | trans = NULL; | |
4034 | goto out; | |
4035 | } else { | |
4036 | ret = btrfs_commit_transaction(trans, root); | |
4037 | if (ret < 0) | |
4038 | goto out; | |
4039 | } | |
4040 | ||
4041 | ret = copy_to_user(arg, sa, sizeof(*sa)); | |
4042 | if (ret) | |
4043 | ret = -EFAULT; | |
4044 | ||
4045 | out: | |
4046 | kfree(sa); | |
4047 | up_write(&root->fs_info->subvol_sem); | |
4048 | mnt_drop_write_file(file); | |
4049 | return ret; | |
4050 | } | |
4051 | ||
867ab667 | 4052 | static int btrfs_ioctl_get_fslabel(struct file *file, void __user *arg) |
4053 | { | |
4054 | struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root; | |
4055 | const char *label = root->fs_info->super_copy->label; | |
4056 | size_t len = strnlen(label, BTRFS_LABEL_SIZE); | |
4057 | int ret; | |
4058 | ||
4059 | if (len == BTRFS_LABEL_SIZE) { | |
4060 | pr_warn("btrfs: label is too long, return the first %zu bytes\n", | |
4061 | --len); | |
4062 | } | |
4063 | ||
4064 | mutex_lock(&root->fs_info->volume_mutex); | |
4065 | ret = copy_to_user(arg, label, len); | |
4066 | mutex_unlock(&root->fs_info->volume_mutex); | |
4067 | ||
4068 | return ret ? -EFAULT : 0; | |
4069 | } | |
4070 | ||
a8bfd4ab | 4071 | static int btrfs_ioctl_set_fslabel(struct file *file, void __user *arg) |
4072 | { | |
4073 | struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root; | |
4074 | struct btrfs_super_block *super_block = root->fs_info->super_copy; | |
4075 | struct btrfs_trans_handle *trans; | |
4076 | char label[BTRFS_LABEL_SIZE]; | |
4077 | int ret; | |
4078 | ||
4079 | if (!capable(CAP_SYS_ADMIN)) | |
4080 | return -EPERM; | |
4081 | ||
4082 | if (copy_from_user(label, arg, sizeof(label))) | |
4083 | return -EFAULT; | |
4084 | ||
4085 | if (strnlen(label, BTRFS_LABEL_SIZE) == BTRFS_LABEL_SIZE) { | |
4086 | pr_err("btrfs: unable to set label with more than %d bytes\n", | |
4087 | BTRFS_LABEL_SIZE - 1); | |
4088 | return -EINVAL; | |
4089 | } | |
4090 | ||
4091 | ret = mnt_want_write_file(file); | |
4092 | if (ret) | |
4093 | return ret; | |
4094 | ||
4095 | mutex_lock(&root->fs_info->volume_mutex); | |
4096 | trans = btrfs_start_transaction(root, 0); | |
4097 | if (IS_ERR(trans)) { | |
4098 | ret = PTR_ERR(trans); | |
4099 | goto out_unlock; | |
4100 | } | |
4101 | ||
4102 | strcpy(super_block->label, label); | |
4103 | ret = btrfs_end_transaction(trans, root); | |
4104 | ||
4105 | out_unlock: | |
4106 | mutex_unlock(&root->fs_info->volume_mutex); | |
4107 | mnt_drop_write_file(file); | |
4108 | return ret; | |
4109 | } | |
4110 | ||
f46b5a66 CH |
4111 | long btrfs_ioctl(struct file *file, unsigned int |
4112 | cmd, unsigned long arg) | |
4113 | { | |
496ad9aa | 4114 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
4bcabaa3 | 4115 | void __user *argp = (void __user *)arg; |
f46b5a66 CH |
4116 | |
4117 | switch (cmd) { | |
6cbff00f CH |
4118 | case FS_IOC_GETFLAGS: |
4119 | return btrfs_ioctl_getflags(file, argp); | |
4120 | case FS_IOC_SETFLAGS: | |
4121 | return btrfs_ioctl_setflags(file, argp); | |
4122 | case FS_IOC_GETVERSION: | |
4123 | return btrfs_ioctl_getversion(file, argp); | |
f7039b1d LD |
4124 | case FITRIM: |
4125 | return btrfs_ioctl_fitrim(file, argp); | |
f46b5a66 | 4126 | case BTRFS_IOC_SNAP_CREATE: |
fa0d2b9b | 4127 | return btrfs_ioctl_snap_create(file, argp, 0); |
fdfb1e4f | 4128 | case BTRFS_IOC_SNAP_CREATE_V2: |
fa0d2b9b | 4129 | return btrfs_ioctl_snap_create_v2(file, argp, 0); |
3de4586c | 4130 | case BTRFS_IOC_SUBVOL_CREATE: |
fa0d2b9b | 4131 | return btrfs_ioctl_snap_create(file, argp, 1); |
6f72c7e2 AJ |
4132 | case BTRFS_IOC_SUBVOL_CREATE_V2: |
4133 | return btrfs_ioctl_snap_create_v2(file, argp, 1); | |
76dda93c YZ |
4134 | case BTRFS_IOC_SNAP_DESTROY: |
4135 | return btrfs_ioctl_snap_destroy(file, argp); | |
0caa102d LZ |
4136 | case BTRFS_IOC_SUBVOL_GETFLAGS: |
4137 | return btrfs_ioctl_subvol_getflags(file, argp); | |
4138 | case BTRFS_IOC_SUBVOL_SETFLAGS: | |
4139 | return btrfs_ioctl_subvol_setflags(file, argp); | |
6ef5ed0d JB |
4140 | case BTRFS_IOC_DEFAULT_SUBVOL: |
4141 | return btrfs_ioctl_default_subvol(file, argp); | |
f46b5a66 | 4142 | case BTRFS_IOC_DEFRAG: |
1e701a32 CM |
4143 | return btrfs_ioctl_defrag(file, NULL); |
4144 | case BTRFS_IOC_DEFRAG_RANGE: | |
4145 | return btrfs_ioctl_defrag(file, argp); | |
f46b5a66 | 4146 | case BTRFS_IOC_RESIZE: |
198605a8 | 4147 | return btrfs_ioctl_resize(file, argp); |
f46b5a66 | 4148 | case BTRFS_IOC_ADD_DEV: |
4bcabaa3 | 4149 | return btrfs_ioctl_add_dev(root, argp); |
f46b5a66 | 4150 | case BTRFS_IOC_RM_DEV: |
da24927b | 4151 | return btrfs_ioctl_rm_dev(file, argp); |
475f6387 JS |
4152 | case BTRFS_IOC_FS_INFO: |
4153 | return btrfs_ioctl_fs_info(root, argp); | |
4154 | case BTRFS_IOC_DEV_INFO: | |
4155 | return btrfs_ioctl_dev_info(root, argp); | |
f46b5a66 | 4156 | case BTRFS_IOC_BALANCE: |
9ba1f6e4 | 4157 | return btrfs_ioctl_balance(file, NULL); |
f46b5a66 | 4158 | case BTRFS_IOC_CLONE: |
c5c9cd4d SW |
4159 | return btrfs_ioctl_clone(file, arg, 0, 0, 0); |
4160 | case BTRFS_IOC_CLONE_RANGE: | |
7a865e8a | 4161 | return btrfs_ioctl_clone_range(file, argp); |
f46b5a66 CH |
4162 | case BTRFS_IOC_TRANS_START: |
4163 | return btrfs_ioctl_trans_start(file); | |
4164 | case BTRFS_IOC_TRANS_END: | |
4165 | return btrfs_ioctl_trans_end(file); | |
ac8e9819 CM |
4166 | case BTRFS_IOC_TREE_SEARCH: |
4167 | return btrfs_ioctl_tree_search(file, argp); | |
4168 | case BTRFS_IOC_INO_LOOKUP: | |
4169 | return btrfs_ioctl_ino_lookup(file, argp); | |
d7728c96 JS |
4170 | case BTRFS_IOC_INO_PATHS: |
4171 | return btrfs_ioctl_ino_to_path(root, argp); | |
4172 | case BTRFS_IOC_LOGICAL_INO: | |
4173 | return btrfs_ioctl_logical_to_ino(root, argp); | |
1406e432 JB |
4174 | case BTRFS_IOC_SPACE_INFO: |
4175 | return btrfs_ioctl_space_info(root, argp); | |
f46b5a66 CH |
4176 | case BTRFS_IOC_SYNC: |
4177 | btrfs_sync_fs(file->f_dentry->d_sb, 1); | |
4178 | return 0; | |
46204592 | 4179 | case BTRFS_IOC_START_SYNC: |
9a8c28be | 4180 | return btrfs_ioctl_start_sync(root, argp); |
46204592 | 4181 | case BTRFS_IOC_WAIT_SYNC: |
9a8c28be | 4182 | return btrfs_ioctl_wait_sync(root, argp); |
475f6387 | 4183 | case BTRFS_IOC_SCRUB: |
b8e95489 | 4184 | return btrfs_ioctl_scrub(file, argp); |
475f6387 JS |
4185 | case BTRFS_IOC_SCRUB_CANCEL: |
4186 | return btrfs_ioctl_scrub_cancel(root, argp); | |
4187 | case BTRFS_IOC_SCRUB_PROGRESS: | |
4188 | return btrfs_ioctl_scrub_progress(root, argp); | |
c9e9f97b | 4189 | case BTRFS_IOC_BALANCE_V2: |
9ba1f6e4 | 4190 | return btrfs_ioctl_balance(file, argp); |
837d5b6e ID |
4191 | case BTRFS_IOC_BALANCE_CTL: |
4192 | return btrfs_ioctl_balance_ctl(root, arg); | |
19a39dce ID |
4193 | case BTRFS_IOC_BALANCE_PROGRESS: |
4194 | return btrfs_ioctl_balance_progress(root, argp); | |
8ea05e3a AB |
4195 | case BTRFS_IOC_SET_RECEIVED_SUBVOL: |
4196 | return btrfs_ioctl_set_received_subvol(file, argp); | |
31db9f7c AB |
4197 | case BTRFS_IOC_SEND: |
4198 | return btrfs_ioctl_send(file, argp); | |
c11d2c23 | 4199 | case BTRFS_IOC_GET_DEV_STATS: |
b27f7c0c | 4200 | return btrfs_ioctl_get_dev_stats(root, argp); |
5d13a37b | 4201 | case BTRFS_IOC_QUOTA_CTL: |
905b0dda | 4202 | return btrfs_ioctl_quota_ctl(file, argp); |
5d13a37b | 4203 | case BTRFS_IOC_QGROUP_ASSIGN: |
905b0dda | 4204 | return btrfs_ioctl_qgroup_assign(file, argp); |
5d13a37b | 4205 | case BTRFS_IOC_QGROUP_CREATE: |
905b0dda | 4206 | return btrfs_ioctl_qgroup_create(file, argp); |
5d13a37b | 4207 | case BTRFS_IOC_QGROUP_LIMIT: |
905b0dda | 4208 | return btrfs_ioctl_qgroup_limit(file, argp); |
2f232036 JS |
4209 | case BTRFS_IOC_QUOTA_RESCAN: |
4210 | return btrfs_ioctl_quota_rescan(file, argp); | |
4211 | case BTRFS_IOC_QUOTA_RESCAN_STATUS: | |
4212 | return btrfs_ioctl_quota_rescan_status(file, argp); | |
3f6bcfbd SB |
4213 | case BTRFS_IOC_DEV_REPLACE: |
4214 | return btrfs_ioctl_dev_replace(root, argp); | |
867ab667 | 4215 | case BTRFS_IOC_GET_FSLABEL: |
4216 | return btrfs_ioctl_get_fslabel(file, argp); | |
a8bfd4ab | 4217 | case BTRFS_IOC_SET_FSLABEL: |
4218 | return btrfs_ioctl_set_fslabel(file, argp); | |
f46b5a66 CH |
4219 | } |
4220 | ||
4221 | return -ENOTTY; | |
4222 | } |