projects / GitHub / mt8127 / android_kernel_alcatel_ttab.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
Btrfs: fix wrong return value of btrfs_wait_for_commit()
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / fs / btrfs / ioctl.c
CommitLineData
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>
4b4e25f2 45#include "compat.h"
f46b5a66
CH		 46#include "ctree.h"
47#include "disk-io.h"
48#include "transaction.h"
49#include "btrfs_inode.h"
50#include "ioctl.h"
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. */
61static 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 */
74static 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 */
104void 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 */
127void 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
144 if (flags & BTRFS_INODE_NODATACOW)
145 BTRFS_I(inode)->flags |= BTRFS_INODE_NODATACOW;
6cbff00f 146
6cbff00f
CH		 147 btrfs_update_iflags(inode);
148}
149
150static int btrfs_ioctl_getflags(struct file *file, void __user *arg)
151{
152 struct btrfs_inode *ip = BTRFS_I(file->f_path.dentry->d_inode);
153 unsigned int flags = btrfs_flags_to_ioctl(ip->flags);
154
155 if (copy_to_user(arg, &flags, sizeof(flags)))
156 return -EFAULT;
157 return 0;
158}
159
75e7cb7f
LB		 160static int check_flags(unsigned int flags)
161{
162 if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \
163 FS_NOATIME_FL | FS_NODUMP_FL | \
164 FS_SYNC_FL | FS_DIRSYNC_FL | \
e1e8fb6a
LZ		 165 FS_NOCOMP_FL | FS_COMPR_FL |
166 FS_NOCOW_FL))
75e7cb7f
LB		 167 return -EOPNOTSUPP;
168
169 if ((flags & FS_NOCOMP_FL) && (flags & FS_COMPR_FL))
170 return -EINVAL;
171
75e7cb7f
LB		 172 return 0;
173}
174
6cbff00f
CH		 175static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
176{
177 struct inode *inode = file->f_path.dentry->d_inode;
178 struct btrfs_inode *ip = BTRFS_I(inode);
179 struct btrfs_root *root = ip->root;
180 struct btrfs_trans_handle *trans;
181 unsigned int flags, oldflags;
182 int ret;
f062abf0
LZ		 183 u64 ip_oldflags;
184 unsigned int i_oldflags;
7e97b8da 185 umode_t mode;
6cbff00f 186
b83cc969
LZ		 187 if (btrfs_root_readonly(root))
188 return -EROFS;
189
6cbff00f
CH		 190 if (copy_from_user(&flags, arg, sizeof(flags)))
191 return -EFAULT;
192
75e7cb7f
LB		 193 ret = check_flags(flags);
194 if (ret)
195 return ret;
f46b5a66 196
2e149670 197 if (!inode_owner_or_capable(inode))
6cbff00f
CH		 198 return -EACCES;
199
e7848683
JK		 200 ret = mnt_want_write_file(file);
201 if (ret)
202 return ret;
203
6cbff00f
CH		 204 mutex_lock(&inode->i_mutex);
205
f062abf0
LZ		 206 ip_oldflags = ip->flags;
207 i_oldflags = inode->i_flags;
7e97b8da 208 mode = inode->i_mode;
f062abf0 209
6cbff00f
CH		 210 flags = btrfs_mask_flags(inode->i_mode, flags);
211 oldflags = btrfs_flags_to_ioctl(ip->flags);
212 if ((flags ^ oldflags) & (FS_APPEND_FL | FS_IMMUTABLE_FL)) {
213 if (!capable(CAP_LINUX_IMMUTABLE)) {
214 ret = -EPERM;
215 goto out_unlock;
216 }
217 }
218
6cbff00f
CH		 219 if (flags & FS_SYNC_FL)
220 ip->flags |= BTRFS_INODE_SYNC;
221 else
222 ip->flags &= ~BTRFS_INODE_SYNC;
223 if (flags & FS_IMMUTABLE_FL)
224 ip->flags |= BTRFS_INODE_IMMUTABLE;
225 else
226 ip->flags &= ~BTRFS_INODE_IMMUTABLE;
227 if (flags & FS_APPEND_FL)
228 ip->flags |= BTRFS_INODE_APPEND;
229 else
230 ip->flags &= ~BTRFS_INODE_APPEND;
231 if (flags & FS_NODUMP_FL)
232 ip->flags |= BTRFS_INODE_NODUMP;
233 else
234 ip->flags &= ~BTRFS_INODE_NODUMP;
235 if (flags & FS_NOATIME_FL)
236 ip->flags |= BTRFS_INODE_NOATIME;
237 else
238 ip->flags &= ~BTRFS_INODE_NOATIME;
239 if (flags & FS_DIRSYNC_FL)
240 ip->flags |= BTRFS_INODE_DIRSYNC;
241 else
242 ip->flags &= ~BTRFS_INODE_DIRSYNC;
7e97b8da
DS		 243 if (flags & FS_NOCOW_FL) {
244 if (S_ISREG(mode)) {
245 /*
246 * It's safe to turn csums off here, no extents exist.
247 * Otherwise we want the flag to reflect the real COW
248 * status of the file and will not set it.
249 */
250 if (inode->i_size == 0)
251 ip->flags |= BTRFS_INODE_NODATACOW
252 | BTRFS_INODE_NODATASUM;
253 } else {
254 ip->flags |= BTRFS_INODE_NODATACOW;
255 }
256 } else {
257 /*
258 * Revert back under same assuptions as above
259 */
260 if (S_ISREG(mode)) {
261 if (inode->i_size == 0)
262 ip->flags &= ~(BTRFS_INODE_NODATACOW
263 | BTRFS_INODE_NODATASUM);
264 } else {
265 ip->flags &= ~BTRFS_INODE_NODATACOW;
266 }
267 }
6cbff00f 268
75e7cb7f
LB		 269 /*
270 * The COMPRESS flag can only be changed by users, while the NOCOMPRESS
271 * flag may be changed automatically if compression code won't make
272 * things smaller.
273 */
274 if (flags & FS_NOCOMP_FL) {
275 ip->flags &= ~BTRFS_INODE_COMPRESS;
276 ip->flags |= BTRFS_INODE_NOCOMPRESS;
277 } else if (flags & FS_COMPR_FL) {
278 ip->flags |= BTRFS_INODE_COMPRESS;
279 ip->flags &= ~BTRFS_INODE_NOCOMPRESS;
ebcb904d
LZ		 280 } else {
281 ip->flags &= ~(BTRFS_INODE_COMPRESS | BTRFS_INODE_NOCOMPRESS);
75e7cb7f 282 }
6cbff00f 283
4da6f1a3 284 trans = btrfs_start_transaction(root, 1);
f062abf0
LZ		 285 if (IS_ERR(trans)) {
286 ret = PTR_ERR(trans);
287 goto out_drop;
288 }
6cbff00f 289
306424cc 290 btrfs_update_iflags(inode);
0c4d2d95 291 inode_inc_iversion(inode);
306424cc 292 inode->i_ctime = CURRENT_TIME;
6cbff00f 293 ret = btrfs_update_inode(trans, root, inode);
6cbff00f 294
6cbff00f 295 btrfs_end_transaction(trans, root);
f062abf0
LZ		 296 out_drop:
297 if (ret) {
298 ip->flags = ip_oldflags;
299 inode->i_flags = i_oldflags;
300 }
6cbff00f 301
6cbff00f
CH		 302 out_unlock:
303 mutex_unlock(&inode->i_mutex);
e7848683 304 mnt_drop_write_file(file);
2d4e6f6a 305 return ret;
6cbff00f
CH		 306}
307
308static int btrfs_ioctl_getversion(struct file *file, int __user *arg)
309{
310 struct inode *inode = file->f_path.dentry->d_inode;
311
312 return put_user(inode->i_generation, arg);
313}
f46b5a66 314
f7039b1d
LD		 315static noinline int btrfs_ioctl_fitrim(struct file *file, void __user *arg)
316{
815745cf 317 struct btrfs_fs_info *fs_info = btrfs_sb(fdentry(file)->d_sb);
f7039b1d
LD		 318 struct btrfs_device *device;
319 struct request_queue *q;
320 struct fstrim_range range;
321 u64 minlen = ULLONG_MAX;
322 u64 num_devices = 0;
815745cf 323 u64 total_bytes = btrfs_super_total_bytes(fs_info->super_copy);
f7039b1d
LD		 324 int ret;
325
326 if (!capable(CAP_SYS_ADMIN))
327 return -EPERM;
328
1f78160c
XG		 329 rcu_read_lock();
330 list_for_each_entry_rcu(device, &fs_info->fs_devices->devices,
331 dev_list) {
f7039b1d
LD		 332 if (!device->bdev)
333 continue;
334 q = bdev_get_queue(device->bdev);
335 if (blk_queue_discard(q)) {
336 num_devices++;
337 minlen = min((u64)q->limits.discard_granularity,
338 minlen);
339 }
340 }
1f78160c 341 rcu_read_unlock();
f4c697e6 342
f7039b1d
LD		 343 if (!num_devices)
344 return -EOPNOTSUPP;
f7039b1d
LD		 345 if (copy_from_user(&range, arg, sizeof(range)))
346 return -EFAULT;
e515c18b
LC		 347 if (range.start > total_bytes ||
348 range.len < fs_info->sb->s_blocksize)
f4c697e6 349 return -EINVAL;
f7039b1d 350
f4c697e6 351 range.len = min(range.len, total_bytes - range.start);
f7039b1d 352 range.minlen = max(range.minlen, minlen);
815745cf 353 ret = btrfs_trim_fs(fs_info->tree_root, &range);
f7039b1d
LD		 354 if (ret < 0)
355 return ret;
356
357 if (copy_to_user(arg, &range, sizeof(range)))
358 return -EFAULT;
359
360 return 0;
361}
362
cb8e7090
CH		 363static noinline int create_subvol(struct btrfs_root *root,
364 struct dentry *dentry,
72fd032e 365 char *name, int namelen,
6f72c7e2
AJ		 366 u64 *async_transid,
367 struct btrfs_qgroup_inherit **inherit)
f46b5a66
CH		 368{
369 struct btrfs_trans_handle *trans;
370 struct btrfs_key key;
371 struct btrfs_root_item root_item;
372 struct btrfs_inode_item *inode_item;
373 struct extent_buffer *leaf;
76dda93c 374 struct btrfs_root *new_root;
2fbe8c8a 375 struct dentry *parent = dentry->d_parent;
6a912213 376 struct inode *dir;
8ea05e3a 377 struct timespec cur_time = CURRENT_TIME;
f46b5a66
CH		 378 int ret;
379 int err;
380 u64 objectid;
381 u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
3de4586c 382 u64 index = 0;
8ea05e3a 383 uuid_le new_uuid;
f46b5a66 384
581bb050 385 ret = btrfs_find_free_objectid(root->fs_info->tree_root, &objectid);
2fbe8c8a 386 if (ret)
a22285a6 387 return ret;
6a912213
JB		 388
389 dir = parent->d_inode;
390
9ed74f2d
JB		 391 /*
392 * 1 - inode item
393 * 2 - refs
394 * 1 - root item
395 * 2 - dir items
396 */
a22285a6 397 trans = btrfs_start_transaction(root, 6);
2fbe8c8a 398 if (IS_ERR(trans))
a22285a6 399 return PTR_ERR(trans);
f46b5a66 400
6f72c7e2
AJ		 401 ret = btrfs_qgroup_inherit(trans, root->fs_info, 0, objectid,
402 inherit ? *inherit : NULL);
403 if (ret)
404 goto fail;
405
5d4f98a2 406 leaf = btrfs_alloc_free_block(trans, root, root->leafsize,
5581a51a 407 0, objectid, NULL, 0, 0, 0);
8e8a1e31
JB		 408 if (IS_ERR(leaf)) {
409 ret = PTR_ERR(leaf);
410 goto fail;
411 }
f46b5a66 412
5d4f98a2 413 memset_extent_buffer(leaf, 0, 0, sizeof(struct btrfs_header));
f46b5a66
CH		 414 btrfs_set_header_bytenr(leaf, leaf->start);
415 btrfs_set_header_generation(leaf, trans->transid);
5d4f98a2 416 btrfs_set_header_backref_rev(leaf, BTRFS_MIXED_BACKREF_REV);
f46b5a66
CH		 417 btrfs_set_header_owner(leaf, objectid);
418
419 write_extent_buffer(leaf, root->fs_info->fsid,
420 (unsigned long)btrfs_header_fsid(leaf),
421 BTRFS_FSID_SIZE);
5d4f98a2
YZ		 422 write_extent_buffer(leaf, root->fs_info->chunk_tree_uuid,
423 (unsigned long)btrfs_header_chunk_tree_uuid(leaf),
424 BTRFS_UUID_SIZE);
f46b5a66
CH		 425 btrfs_mark_buffer_dirty(leaf);
426
8ea05e3a
AB		 427 memset(&root_item, 0, sizeof(root_item));
428
f46b5a66 429 inode_item = &root_item.inode;
f46b5a66
CH		 430 inode_item->generation = cpu_to_le64(1);
431 inode_item->size = cpu_to_le64(3);
432 inode_item->nlink = cpu_to_le32(1);
a76a3cd4 433 inode_item->nbytes = cpu_to_le64(root->leafsize);
f46b5a66
CH		 434 inode_item->mode = cpu_to_le32(S_IFDIR | 0755);
435
08fe4db1
LZ		 436 root_item.flags = 0;
437 root_item.byte_limit = 0;
438 inode_item->flags = cpu_to_le64(BTRFS_INODE_ROOT_ITEM_INIT);
439
f46b5a66 440 btrfs_set_root_bytenr(&root_item, leaf->start);
84234f3a 441 btrfs_set_root_generation(&root_item, trans->transid);
f46b5a66
CH		 442 btrfs_set_root_level(&root_item, 0);
443 btrfs_set_root_refs(&root_item, 1);
86b9f2ec 444 btrfs_set_root_used(&root_item, leaf->len);
80ff3856 445 btrfs_set_root_last_snapshot(&root_item, 0);
f46b5a66 446
8ea05e3a
AB		 447 btrfs_set_root_generation_v2(&root_item,
448 btrfs_root_generation(&root_item));
449 uuid_le_gen(&new_uuid);
450 memcpy(root_item.uuid, new_uuid.b, BTRFS_UUID_SIZE);
451 root_item.otime.sec = cpu_to_le64(cur_time.tv_sec);
dadd1105 452 root_item.otime.nsec = cpu_to_le32(cur_time.tv_nsec);
8ea05e3a
AB		 453 root_item.ctime = root_item.otime;
454 btrfs_set_root_ctransid(&root_item, trans->transid);
455 btrfs_set_root_otransid(&root_item, trans->transid);
f46b5a66 456
925baedd 457 btrfs_tree_unlock(leaf);
f46b5a66
CH		 458 free_extent_buffer(leaf);
459 leaf = NULL;
460
461 btrfs_set_root_dirid(&root_item, new_dirid);
462
463 key.objectid = objectid;
5d4f98a2 464 key.offset = 0;
f46b5a66
CH		 465 btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
466 ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key,
467 &root_item);
468 if (ret)
469 goto fail;
470
76dda93c
YZ		 471 key.offset = (u64)-1;
472 new_root = btrfs_read_fs_root_no_name(root->fs_info, &key);
79787eaa
JM		 473 if (IS_ERR(new_root)) {
474 btrfs_abort_transaction(trans, root, PTR_ERR(new_root));
475 ret = PTR_ERR(new_root);
476 goto fail;
477 }
76dda93c
YZ		 478
479 btrfs_record_root_in_trans(trans, new_root);
480
d82a6f1d 481 ret = btrfs_create_subvol_root(trans, new_root, new_dirid);
ce598979
MF		 482 if (ret) {
483 /* We potentially lose an unused inode item here */
79787eaa 484 btrfs_abort_transaction(trans, root, ret);
ce598979
MF		 485 goto fail;
486 }
487
f46b5a66
CH		 488 /*
489 * insert the directory item
490 */
3de4586c 491 ret = btrfs_set_inode_index(dir, &index);
79787eaa
JM		 492 if (ret) {
493 btrfs_abort_transaction(trans, root, ret);
494 goto fail;
495 }
3de4586c
CM		 496
497 ret = btrfs_insert_dir_item(trans, root,
16cdcec7 498 name, namelen, dir, &key,
3de4586c 499 BTRFS_FT_DIR, index);
79787eaa
JM		 500 if (ret) {
501 btrfs_abort_transaction(trans, root, ret);
f46b5a66 502 goto fail;
79787eaa 503 }
0660b5af 504
52c26179
YZ		 505 btrfs_i_size_write(dir, dir->i_size + namelen * 2);
506 ret = btrfs_update_inode(trans, root, dir);
507 BUG_ON(ret);
508
0660b5af 509 ret = btrfs_add_root_ref(trans, root->fs_info->tree_root,
4df27c4d 510 objectid, root->root_key.objectid,
33345d01 511 btrfs_ino(dir), index, name, namelen);
0660b5af 512
76dda93c 513 BUG_ON(ret);
f46b5a66 514
76dda93c 515 d_instantiate(dentry, btrfs_lookup_dentry(dir, dentry));
f46b5a66 516fail:
72fd032e
SW		 517 if (async_transid) {
518 *async_transid = trans->transid;
519 err = btrfs_commit_transaction_async(trans, root, 1);
520 } else {
521 err = btrfs_commit_transaction(trans, root);
522 }
f46b5a66
CH		 523 if (err && !ret)
524 ret = err;
f46b5a66
CH		 525 return ret;
526}
527
72fd032e 528static int create_snapshot(struct btrfs_root *root, struct dentry *dentry,
b83cc969 529 char *name, int namelen, u64 *async_transid,
6f72c7e2 530 bool readonly, struct btrfs_qgroup_inherit **inherit)
f46b5a66 531{
2e4bfab9 532 struct inode *inode;
f46b5a66
CH		 533 struct btrfs_pending_snapshot *pending_snapshot;
534 struct btrfs_trans_handle *trans;
2e4bfab9 535 int ret;
f46b5a66
CH		 536
537 if (!root->ref_cows)
538 return -EINVAL;
539
3de4586c 540 pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS);
a22285a6
YZ		 541 if (!pending_snapshot)
542 return -ENOMEM;
543
66d8f3dd
MX		 544 btrfs_init_block_rsv(&pending_snapshot->block_rsv,
545 BTRFS_BLOCK_RSV_TEMP);
3de4586c 546 pending_snapshot->dentry = dentry;
f46b5a66 547 pending_snapshot->root = root;
b83cc969 548 pending_snapshot->readonly = readonly;
6f72c7e2
AJ		 549 if (inherit) {
550 pending_snapshot->inherit = *inherit;
551 *inherit = NULL; /* take responsibility to free it */
552 }
a22285a6 553
48c03c4b 554 trans = btrfs_start_transaction(root->fs_info->extent_root, 6);
a22285a6
YZ		 555 if (IS_ERR(trans)) {
556 ret = PTR_ERR(trans);
557 goto fail;
558 }
559
560 ret = btrfs_snap_reserve_metadata(trans, pending_snapshot);
561 BUG_ON(ret);
562
8351583e 563 spin_lock(&root->fs_info->trans_lock);
f46b5a66
CH		 564 list_add(&pending_snapshot->list,
565 &trans->transaction->pending_snapshots);
8351583e 566 spin_unlock(&root->fs_info->trans_lock);
72fd032e
SW		 567 if (async_transid) {
568 *async_transid = trans->transid;
569 ret = btrfs_commit_transaction_async(trans,
570 root->fs_info->extent_root, 1);
571 } else {
572 ret = btrfs_commit_transaction(trans,
573 root->fs_info->extent_root);
574 }
109f2365
LB		 575 if (ret) {
576 /* cleanup_transaction has freed this for us */
577 if (trans->aborted)
578 pending_snapshot = NULL;
c37b2b62 579 goto fail;
109f2365 580 }
a22285a6
YZ		 581
582 ret = pending_snapshot->error;
583 if (ret)
584 goto fail;
585
66b4ffd1
JB		 586 ret = btrfs_orphan_cleanup(pending_snapshot->snap);
587 if (ret)
588 goto fail;
f46b5a66 589
2fbe8c8a 590 inode = btrfs_lookup_dentry(dentry->d_parent->d_inode, dentry);
2e4bfab9
YZ		 591 if (IS_ERR(inode)) {
592 ret = PTR_ERR(inode);
593 goto fail;
594 }
595 BUG_ON(!inode);
596 d_instantiate(dentry, inode);
597 ret = 0;
598fail:
a22285a6 599 kfree(pending_snapshot);
f46b5a66
CH		 600 return ret;
601}
602
4260f7c7
SW		 603/* copy of check_sticky in fs/namei.c()
604* It's inline, so penalty for filesystems that don't use sticky bit is
605* minimal.
606*/
607static inline int btrfs_check_sticky(struct inode *dir, struct inode *inode)
608{
2f2f43d3 609 kuid_t fsuid = current_fsuid();
4260f7c7
SW		 610
611 if (!(dir->i_mode & S_ISVTX))
612 return 0;
2f2f43d3 613 if (uid_eq(inode->i_uid, fsuid))
4260f7c7 614 return 0;
2f2f43d3 615 if (uid_eq(dir->i_uid, fsuid))
4260f7c7
SW		 616 return 0;
617 return !capable(CAP_FOWNER);
618}
619
620/* copy of may_delete in fs/namei.c()
621 * Check whether we can remove a link victim from directory dir, check
622 * whether the type of victim is right.
623 * 1. We can't do it if dir is read-only (done in permission())
624 * 2. We should have write and exec permissions on dir
625 * 3. We can't remove anything from append-only dir
626 * 4. We can't do anything with immutable dir (done in permission())
627 * 5. If the sticky bit on dir is set we should either
628 * a. be owner of dir, or
629 * b. be owner of victim, or
630 * c. have CAP_FOWNER capability
631 * 6. If the victim is append-only or immutable we can't do antyhing with
632 * links pointing to it.
633 * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR.
634 * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR.
635 * 9. We can't remove a root or mountpoint.
636 * 10. We don't allow removal of NFS sillyrenamed files; it's handled by
637 * nfs_async_unlink().
638 */
639
640static int btrfs_may_delete(struct inode *dir,struct dentry *victim,int isdir)
641{
642 int error;
643
644 if (!victim->d_inode)
645 return -ENOENT;
646
647 BUG_ON(victim->d_parent->d_inode != dir);
4fa6b5ec 648 audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE);
4260f7c7
SW		 649
650 error = inode_permission(dir, MAY_WRITE | MAY_EXEC);
651 if (error)
652 return error;
653 if (IS_APPEND(dir))
654 return -EPERM;
655 if (btrfs_check_sticky(dir, victim->d_inode)||
656 IS_APPEND(victim->d_inode)||
657 IS_IMMUTABLE(victim->d_inode) || IS_SWAPFILE(victim->d_inode))
658 return -EPERM;
659 if (isdir) {
660 if (!S_ISDIR(victim->d_inode->i_mode))
661 return -ENOTDIR;
662 if (IS_ROOT(victim))
663 return -EBUSY;
664 } else if (S_ISDIR(victim->d_inode->i_mode))
665 return -EISDIR;
666 if (IS_DEADDIR(dir))
667 return -ENOENT;
668 if (victim->d_flags & DCACHE_NFSFS_RENAMED)
669 return -EBUSY;
670 return 0;
671}
672
cb8e7090
CH		 673/* copy of may_create in fs/namei.c() */
674static inline int btrfs_may_create(struct inode *dir, struct dentry *child)
675{
676 if (child->d_inode)
677 return -EEXIST;
678 if (IS_DEADDIR(dir))
679 return -ENOENT;
680 return inode_permission(dir, MAY_WRITE | MAY_EXEC);
681}
682
683/*
684 * Create a new subvolume below @parent. This is largely modeled after
685 * sys_mkdirat and vfs_mkdir, but we only do a single component lookup
686 * inside this filesystem so it's quite a bit simpler.
687 */
76dda93c
YZ		 688static noinline int btrfs_mksubvol(struct path *parent,
689 char *name, int namelen,
72fd032e 690 struct btrfs_root *snap_src,
6f72c7e2
AJ		 691 u64 *async_transid, bool readonly,
692 struct btrfs_qgroup_inherit **inherit)
cb8e7090 693{
76dda93c 694 struct inode *dir = parent->dentry->d_inode;
cb8e7090
CH		 695 struct dentry *dentry;
696 int error;
697
76dda93c 698 mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
cb8e7090
CH		 699
700 dentry = lookup_one_len(name, parent->dentry, namelen);
701 error = PTR_ERR(dentry);
702 if (IS_ERR(dentry))
703 goto out_unlock;
704
705 error = -EEXIST;
706 if (dentry->d_inode)
707 goto out_dput;
708
76dda93c 709 error = btrfs_may_create(dir, dentry);
cb8e7090 710 if (error)
a874a63e 711 goto out_dput;
cb8e7090 712
76dda93c
YZ		 713 down_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
714
715 if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0)
716 goto out_up_read;
717
3de4586c 718 if (snap_src) {
6f72c7e2
AJ		 719 error = create_snapshot(snap_src, dentry, name, namelen,
720 async_transid, readonly, inherit);
3de4586c 721 } else {
76dda93c 722 error = create_subvol(BTRFS_I(dir)->root, dentry,
6f72c7e2 723 name, namelen, async_transid, inherit);
3de4586c 724 }
76dda93c
YZ		 725 if (!error)
726 fsnotify_mkdir(dir, dentry);
727out_up_read:
728 up_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
cb8e7090
CH		 729out_dput:
730 dput(dentry);
731out_unlock:
76dda93c 732 mutex_unlock(&dir->i_mutex);
cb8e7090
CH		 733 return error;
734}
735
4cb5300b
CM		 736/*
737 * When we're defragging a range, we don't want to kick it off again
738 * if it is really just waiting for delalloc to send it down.
739 * If we find a nice big extent or delalloc range for the bytes in the
740 * file you want to defrag, we return 0 to let you know to skip this
741 * part of the file
742 */
743static int check_defrag_in_cache(struct inode *inode, u64 offset, int thresh)
744{
745 struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
746 struct extent_map *em = NULL;
747 struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
748 u64 end;
749
750 read_lock(&em_tree->lock);
751 em = lookup_extent_mapping(em_tree, offset, PAGE_CACHE_SIZE);
752 read_unlock(&em_tree->lock);
753
754 if (em) {
755 end = extent_map_end(em);
756 free_extent_map(em);
757 if (end - offset > thresh)
758 return 0;
759 }
760 /* if we already have a nice delalloc here, just stop */
761 thresh /= 2;
762 end = count_range_bits(io_tree, &offset, offset + thresh,
763 thresh, EXTENT_DELALLOC, 1);
764 if (end >= thresh)
765 return 0;
766 return 1;
767}
768
769/*
770 * helper function to walk through a file and find extents
771 * newer than a specific transid, and smaller than thresh.
772 *
773 * This is used by the defragging code to find new and small
774 * extents
775 */
776static int find_new_extents(struct btrfs_root *root,
777 struct inode *inode, u64 newer_than,
778 u64 *off, int thresh)
779{
780 struct btrfs_path *path;
781 struct btrfs_key min_key;
782 struct btrfs_key max_key;
783 struct extent_buffer *leaf;
784 struct btrfs_file_extent_item *extent;
785 int type;
786 int ret;
a4689d2b 787 u64 ino = btrfs_ino(inode);
4cb5300b
CM		 788
789 path = btrfs_alloc_path();
790 if (!path)
791 return -ENOMEM;
792
a4689d2b 793 min_key.objectid = ino;
4cb5300b
CM		 794 min_key.type = BTRFS_EXTENT_DATA_KEY;
795 min_key.offset = *off;
796
a4689d2b 797 max_key.objectid = ino;
4cb5300b
CM		 798 max_key.type = (u8)-1;
799 max_key.offset = (u64)-1;
800
801 path->keep_locks = 1;
802
803 while(1) {
804 ret = btrfs_search_forward(root, &min_key, &max_key,
805 path, 0, newer_than);
806 if (ret != 0)
807 goto none;
a4689d2b 808 if (min_key.objectid != ino)
4cb5300b
CM		 809 goto none;
810 if (min_key.type != BTRFS_EXTENT_DATA_KEY)
811 goto none;
812
813 leaf = path->nodes[0];
814 extent = btrfs_item_ptr(leaf, path->slots[0],
815 struct btrfs_file_extent_item);
816
817 type = btrfs_file_extent_type(leaf, extent);
818 if (type == BTRFS_FILE_EXTENT_REG &&
819 btrfs_file_extent_num_bytes(leaf, extent) < thresh &&
820 check_defrag_in_cache(inode, min_key.offset, thresh)) {
821 *off = min_key.offset;
822 btrfs_free_path(path);
823 return 0;
824 }
825
826 if (min_key.offset == (u64)-1)
827 goto none;
828
829 min_key.offset++;
830 btrfs_release_path(path);
831 }
832none:
833 btrfs_free_path(path);
834 return -ENOENT;
835}
836
6c282eb4 837static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start)
17ce6ef8
LB		 838{
839 struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
6c282eb4
LZ		 840 struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
841 struct extent_map *em;
842 u64 len = PAGE_CACHE_SIZE;
17ce6ef8 843
6c282eb4
LZ		 844 /*
845 * hopefully we have this extent in the tree already, try without
846 * the full extent lock
847 */
17ce6ef8 848 read_lock(&em_tree->lock);
6c282eb4 849 em = lookup_extent_mapping(em_tree, start, len);
17ce6ef8
LB		 850 read_unlock(&em_tree->lock);
851
6c282eb4
LZ		 852 if (!em) {
853 /* get the big lock and read metadata off disk */
854 lock_extent(io_tree, start, start + len - 1);
855 em = btrfs_get_extent(inode, NULL, 0, start, len, 0);
856 unlock_extent(io_tree, start, start + len - 1);
857
858 if (IS_ERR(em))
859 return NULL;
860 }
861
862 return em;
863}
17ce6ef8 864
6c282eb4
LZ		 865static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em)
866{
867 struct extent_map *next;
868 bool ret = true;
869
870 /* this is the last extent */
871 if (em->start + em->len >= i_size_read(inode))
872 return false;
873
874 next = defrag_lookup_extent(inode, em->start + em->len);
875 if (!next || next->block_start >= EXTENT_MAP_LAST_BYTE)
876 ret = false;
877
878 free_extent_map(next);
17ce6ef8
LB		 879 return ret;
880}
881
6c282eb4 882static int should_defrag_range(struct inode *inode, u64 start, int thresh,
a43a2111
AM		 883 u64 *last_len, u64 *skip, u64 *defrag_end,
884 int compress)
940100a4 885{
6c282eb4 886 struct extent_map *em;
940100a4 887 int ret = 1;
6c282eb4 888 bool next_mergeable = true;
940100a4
CM		 889
890 /*
008873ea 891 * make sure that once we start defragging an extent, we keep on
940100a4
CM		 892 * defragging it
893 */
894 if (start < *defrag_end)
895 return 1;
896
897 *skip = 0;
898
6c282eb4
LZ		 899 em = defrag_lookup_extent(inode, start);
900 if (!em)
901 return 0;
940100a4
CM		 902
903 /* this will cover holes, and inline extents */
17ce6ef8 904 if (em->block_start >= EXTENT_MAP_LAST_BYTE) {
940100a4 905 ret = 0;
17ce6ef8
LB		 906 goto out;
907 }
908
6c282eb4 909 next_mergeable = defrag_check_next_extent(inode, em);
940100a4
CM		 910
911 /*
6c282eb4
LZ		 912 * we hit a real extent, if it is big or the next extent is not a
913 * real extent, don't bother defragging it
940100a4 914 */
a43a2111 915 if (!compress && (*last_len == 0 || *last_len >= thresh) &&
6c282eb4 916 (em->len >= thresh || !next_mergeable))
940100a4 917 ret = 0;
17ce6ef8 918out:
940100a4
CM		 919 /*
920 * last_len ends up being a counter of how many bytes we've defragged.
921 * every time we choose not to defrag an extent, we reset *last_len
922 * so that the next tiny extent will force a defrag.
923 *
924 * The end result of this is that tiny extents before a single big
925 * extent will force at least part of that big extent to be defragged.
926 */
927 if (ret) {
940100a4
CM		 928 *defrag_end = extent_map_end(em);
929 } else {
930 *last_len = 0;
931 *skip = extent_map_end(em);
932 *defrag_end = 0;
933 }
934
935 free_extent_map(em);
936 return ret;
937}
938
4cb5300b
CM		 939/*
940 * it doesn't do much good to defrag one or two pages
941 * at a time. This pulls in a nice chunk of pages
942 * to COW and defrag.
943 *
944 * It also makes sure the delalloc code has enough
945 * dirty data to avoid making new small extents as part
946 * of the defrag
947 *
948 * It's a good idea to start RA on this range
949 * before calling this.
950 */
951static int cluster_pages_for_defrag(struct inode *inode,
952 struct page **pages,
953 unsigned long start_index,
954 int num_pages)
f46b5a66 955{
4cb5300b
CM		 956 unsigned long file_end;
957 u64 isize = i_size_read(inode);
958 u64 page_start;
959 u64 page_end;
1f12bd06 960 u64 page_cnt;
4cb5300b
CM		 961 int ret;
962 int i;
963 int i_done;
3eaa2885 964 struct btrfs_ordered_extent *ordered;
4cb5300b 965 struct extent_state *cached_state = NULL;
600a45e1 966 struct extent_io_tree *tree;
3b16a4e3 967 gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
4cb5300b 968
4cb5300b 969 file_end = (isize - 1) >> PAGE_CACHE_SHIFT;
1f12bd06
LB		 970 if (!isize || start_index > file_end)
971 return 0;
972
973 page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1);
4cb5300b
CM		 974
975 ret = btrfs_delalloc_reserve_space(inode,
1f12bd06 976 page_cnt << PAGE_CACHE_SHIFT);
4cb5300b
CM		 977 if (ret)
978 return ret;
4cb5300b 979 i_done = 0;
600a45e1 980 tree = &BTRFS_I(inode)->io_tree;
4cb5300b
CM		 981
982 /* step one, lock all the pages */
1f12bd06 983 for (i = 0; i < page_cnt; i++) {
4cb5300b 984 struct page *page;
600a45e1 985again:
a94733d0 986 page = find_or_create_page(inode->i_mapping,
600a45e1 987 start_index + i, mask);
4cb5300b
CM		 988 if (!page)
989 break;
990
600a45e1
MX		 991 page_start = page_offset(page);
992 page_end = page_start + PAGE_CACHE_SIZE - 1;
993 while (1) {
d0082371 994 lock_extent(tree, page_start, page_end);
600a45e1
MX		 995 ordered = btrfs_lookup_ordered_extent(inode,
996 page_start);
d0082371 997 unlock_extent(tree, page_start, page_end);
600a45e1
MX		 998 if (!ordered)
999 break;
1000
1001 unlock_page(page);
1002 btrfs_start_ordered_extent(inode, ordered, 1);
1003 btrfs_put_ordered_extent(ordered);
1004 lock_page(page);
1f12bd06
LB		 1005 /*
1006 * we unlocked the page above, so we need check if
1007 * it was released or not.
1008 */
1009 if (page->mapping != inode->i_mapping) {
1010 unlock_page(page);
1011 page_cache_release(page);
1012 goto again;
1013 }
600a45e1
MX		 1014 }
1015
4cb5300b
CM		 1016 if (!PageUptodate(page)) {
1017 btrfs_readpage(NULL, page);
1018 lock_page(page);
1019 if (!PageUptodate(page)) {
1020 unlock_page(page);
1021 page_cache_release(page);
1022 ret = -EIO;
1023 break;
1024 }
1025 }
600a45e1 1026
600a45e1
MX		 1027 if (page->mapping != inode->i_mapping) {
1028 unlock_page(page);
1029 page_cache_release(page);
1030 goto again;
1031 }
1032
4cb5300b
CM		 1033 pages[i] = page;
1034 i_done++;
1035 }
1036 if (!i_done || ret)
1037 goto out;
1038
1039 if (!(inode->i_sb->s_flags & MS_ACTIVE))
1040 goto out;
1041
1042 /*
1043 * so now we have a nice long stream of locked
1044 * and up to date pages, lets wait on them
1045 */
1046 for (i = 0; i < i_done; i++)
1047 wait_on_page_writeback(pages[i]);
1048
1049 page_start = page_offset(pages[0]);
1050 page_end = page_offset(pages[i_done - 1]) + PAGE_CACHE_SIZE;
1051
1052 lock_extent_bits(&BTRFS_I(inode)->io_tree,
d0082371 1053 page_start, page_end - 1, 0, &cached_state);
4cb5300b
CM		 1054 clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start,
1055 page_end - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
9e8a4a8b
LB		 1056 EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 0, 0,
1057 &cached_state, GFP_NOFS);
4cb5300b 1058
1f12bd06 1059 if (i_done != page_cnt) {
9e0baf60
JB		 1060 spin_lock(&BTRFS_I(inode)->lock);
1061 BTRFS_I(inode)->outstanding_extents++;
1062 spin_unlock(&BTRFS_I(inode)->lock);
4cb5300b 1063 btrfs_delalloc_release_space(inode,
1f12bd06 1064 (page_cnt - i_done) << PAGE_CACHE_SHIFT);
4cb5300b
CM		 1065 }
1066
1067
9e8a4a8b
LB		 1068 set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1,
1069 &cached_state, GFP_NOFS);
4cb5300b
CM		 1070
1071 unlock_extent_cached(&BTRFS_I(inode)->io_tree,
1072 page_start, page_end - 1, &cached_state,
1073 GFP_NOFS);
1074
1075 for (i = 0; i < i_done; i++) {
1076 clear_page_dirty_for_io(pages[i]);
1077 ClearPageChecked(pages[i]);
1078 set_page_extent_mapped(pages[i]);
1079 set_page_dirty(pages[i]);
1080 unlock_page(pages[i]);
1081 page_cache_release(pages[i]);
1082 }
1083 return i_done;
1084out:
1085 for (i = 0; i < i_done; i++) {
1086 unlock_page(pages[i]);
1087 page_cache_release(pages[i]);
1088 }
1f12bd06 1089 btrfs_delalloc_release_space(inode, page_cnt << PAGE_CACHE_SHIFT);
4cb5300b
CM		 1090 return ret;
1091
1092}
1093
1094int btrfs_defrag_file(struct inode *inode, struct file *file,
1095 struct btrfs_ioctl_defrag_range_args *range,
1096 u64 newer_than, unsigned long max_to_defrag)
1097{
1098 struct btrfs_root *root = BTRFS_I(inode)->root;
4cb5300b 1099 struct file_ra_state *ra = NULL;
f46b5a66 1100 unsigned long last_index;
151a31b2 1101 u64 isize = i_size_read(inode);
940100a4
CM		 1102 u64 last_len = 0;
1103 u64 skip = 0;
1104 u64 defrag_end = 0;
4cb5300b 1105 u64 newer_off = range->start;
f46b5a66 1106 unsigned long i;
008873ea 1107 unsigned long ra_index = 0;
f46b5a66 1108 int ret;
4cb5300b 1109 int defrag_count = 0;
1a419d85 1110 int compress_type = BTRFS_COMPRESS_ZLIB;
4cb5300b 1111 int extent_thresh = range->extent_thresh;
008873ea
LZ		 1112 int max_cluster = (256 * 1024) >> PAGE_CACHE_SHIFT;
1113 int cluster = max_cluster;
4cb5300b
CM		 1114 u64 new_align = ~((u64)128 * 1024 - 1);
1115 struct page **pages = NULL;
1116
1117 if (extent_thresh == 0)
1118 extent_thresh = 256 * 1024;
1a419d85
LZ		 1119
1120 if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS) {
1121 if (range->compress_type > BTRFS_COMPRESS_TYPES)
1122 return -EINVAL;
1123 if (range->compress_type)
1124 compress_type = range->compress_type;
1125 }
f46b5a66 1126
151a31b2 1127 if (isize == 0)
940100a4
CM		 1128 return 0;
1129
4cb5300b
CM		 1130 /*
1131 * if we were not given a file, allocate a readahead
1132 * context
1133 */
1134 if (!file) {
1135 ra = kzalloc(sizeof(*ra), GFP_NOFS);
1136 if (!ra)
1137 return -ENOMEM;
1138 file_ra_state_init(ra, inode->i_mapping);
1139 } else {
1140 ra = &file->f_ra;
1141 }
1142
008873ea 1143 pages = kmalloc(sizeof(struct page *) * max_cluster,
4cb5300b
CM		 1144 GFP_NOFS);
1145 if (!pages) {
1146 ret = -ENOMEM;
1147 goto out_ra;
1148 }
1149
1150 /* find the last page to defrag */
1e701a32 1151 if (range->start + range->len > range->start) {
151a31b2 1152 last_index = min_t(u64, isize - 1,
1e701a32
CM		 1153 range->start + range->len - 1) >> PAGE_CACHE_SHIFT;
1154 } else {
151a31b2 1155 last_index = (isize - 1) >> PAGE_CACHE_SHIFT;
1e701a32
CM		 1156 }
1157
4cb5300b
CM		 1158 if (newer_than) {
1159 ret = find_new_extents(root, inode, newer_than,
1160 &newer_off, 64 * 1024);
1161 if (!ret) {
1162 range->start = newer_off;
1163 /*
1164 * we always align our defrag to help keep
1165 * the extents in the file evenly spaced
1166 */
1167 i = (newer_off & new_align) >> PAGE_CACHE_SHIFT;
4cb5300b
CM		 1168 } else
1169 goto out_ra;
1170 } else {
1171 i = range->start >> PAGE_CACHE_SHIFT;
1172 }
1173 if (!max_to_defrag)
7ec31b54 1174 max_to_defrag = last_index + 1;
4cb5300b 1175
2a0f7f57
LZ		 1176 /*
1177 * make writeback starts from i, so the defrag range can be
1178 * written sequentially.
1179 */
1180 if (i < inode->i_mapping->writeback_index)
1181 inode->i_mapping->writeback_index = i;
1182
f7f43cc8
CM		 1183 while (i <= last_index && defrag_count < max_to_defrag &&
1184 (i < (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >>
1185 PAGE_CACHE_SHIFT)) {
4cb5300b
CM		 1186 /*
1187 * make sure we stop running if someone unmounts
1188 * the FS
1189 */
1190 if (!(inode->i_sb->s_flags & MS_ACTIVE))
1191 break;
1192
66c26892 1193 if (!should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT,
6c282eb4 1194 extent_thresh, &last_len, &skip,
a43a2111
AM		 1195 &defrag_end, range->flags &
1196 BTRFS_DEFRAG_RANGE_COMPRESS)) {
940100a4
CM		 1197 unsigned long next;
1198 /*
1199 * the should_defrag function tells us how much to skip
1200 * bump our counter by the suggested amount
1201 */
1202 next = (skip + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
1203 i = max(i + 1, next);
1204 continue;
1205 }
008873ea
LZ		 1206
1207 if (!newer_than) {
1208 cluster = (PAGE_CACHE_ALIGN(defrag_end) >>
1209 PAGE_CACHE_SHIFT) - i;
1210 cluster = min(cluster, max_cluster);
1211 } else {
1212 cluster = max_cluster;
1213 }
1214
1e701a32 1215 if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)
1a419d85 1216 BTRFS_I(inode)->force_compress = compress_type;
940100a4 1217
008873ea
LZ		 1218 if (i + cluster > ra_index) {
1219 ra_index = max(i, ra_index);
1220 btrfs_force_ra(inode->i_mapping, ra, file, ra_index,
1221 cluster);
1222 ra_index += max_cluster;
1223 }
940100a4 1224
ecb8bea8 1225 mutex_lock(&inode->i_mutex);
008873ea 1226 ret = cluster_pages_for_defrag(inode, pages, i, cluster);
ecb8bea8
LB		 1227 if (ret < 0) {
1228 mutex_unlock(&inode->i_mutex);
4cb5300b 1229 goto out_ra;
ecb8bea8 1230 }
4cb5300b
CM		 1231
1232 defrag_count += ret;
1233 balance_dirty_pages_ratelimited_nr(inode->i_mapping, ret);
ecb8bea8 1234 mutex_unlock(&inode->i_mutex);
4cb5300b
CM		 1235
1236 if (newer_than) {
1237 if (newer_off == (u64)-1)
1238 break;
1239
e1f041e1
LB		 1240 if (ret > 0)
1241 i += ret;
1242
4cb5300b
CM		 1243 newer_off = max(newer_off + 1,
1244 (u64)i << PAGE_CACHE_SHIFT);
1245
1246 ret = find_new_extents(root, inode,
1247 newer_than, &newer_off,
1248 64 * 1024);
1249 if (!ret) {
1250 range->start = newer_off;
1251 i = (newer_off & new_align) >> PAGE_CACHE_SHIFT;
4cb5300b
CM		 1252 } else {
1253 break;
f46b5a66 1254 }
4cb5300b 1255 } else {
008873ea 1256 if (ret > 0) {
cbcc8326 1257 i += ret;
008873ea
LZ		 1258 last_len += ret << PAGE_CACHE_SHIFT;
1259 } else {
cbcc8326 1260 i++;
008873ea
LZ		 1261 last_len = 0;
1262 }
f46b5a66 1263 }
f46b5a66
CH		 1264 }
1265
1e701a32
CM		 1266 if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO))
1267 filemap_flush(inode->i_mapping);
1268
1269 if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
1270 /* the filemap_flush will queue IO into the worker threads, but
1271 * we have to make sure the IO is actually started and that
1272 * ordered extents get created before we return
1273 */
1274 atomic_inc(&root->fs_info->async_submit_draining);
1275 while (atomic_read(&root->fs_info->nr_async_submits) ||
1276 atomic_read(&root->fs_info->async_delalloc_pages)) {
1277 wait_event(root->fs_info->async_submit_wait,
1278 (atomic_read(&root->fs_info->nr_async_submits) == 0 &&
1279 atomic_read(&root->fs_info->async_delalloc_pages) == 0));
1280 }
1281 atomic_dec(&root->fs_info->async_submit_draining);
1282
1283 mutex_lock(&inode->i_mutex);
261507a0 1284 BTRFS_I(inode)->force_compress = BTRFS_COMPRESS_NONE;
1e701a32
CM		 1285 mutex_unlock(&inode->i_mutex);
1286 }
1287
1a419d85 1288 if (range->compress_type == BTRFS_COMPRESS_LZO) {
2b0ce2c2 1289 btrfs_set_fs_incompat(root->fs_info, COMPRESS_LZO);
1a419d85
LZ		 1290 }
1291
60ccf82f 1292 ret = defrag_count;
940100a4 1293
4cb5300b
CM		 1294out_ra:
1295 if (!file)
1296 kfree(ra);
1297 kfree(pages);
940100a4 1298 return ret;
f46b5a66
CH		 1299}
1300
76dda93c
YZ		 1301static noinline int btrfs_ioctl_resize(struct btrfs_root *root,
1302 void __user *arg)
f46b5a66
CH		 1303{
1304 u64 new_size;
1305 u64 old_size;
1306 u64 devid = 1;
1307 struct btrfs_ioctl_vol_args *vol_args;
1308 struct btrfs_trans_handle *trans;
1309 struct btrfs_device *device = NULL;
1310 char *sizestr;
1311 char *devstr = NULL;
1312 int ret = 0;
f46b5a66
CH		 1313 int mod = 0;
1314
c146afad
YZ		 1315 if (root->fs_info->sb->s_flags & MS_RDONLY)
1316 return -EROFS;
1317
e441d54d
CM		 1318 if (!capable(CAP_SYS_ADMIN))
1319 return -EPERM;
1320
5ac00add
SB		 1321 if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running,
1322 1)) {
1323 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
1324 return -EINPROGRESS;
c9e9f97b
ID		 1325 }
1326
5ac00add 1327 mutex_lock(&root->fs_info->volume_mutex);
dae7b665 1328 vol_args = memdup_user(arg, sizeof(*vol_args));
c9e9f97b
ID		 1329 if (IS_ERR(vol_args)) {
1330 ret = PTR_ERR(vol_args);
1331 goto out;
1332 }
5516e595
MF		 1333
1334 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
f46b5a66 1335
f46b5a66
CH		 1336 sizestr = vol_args->name;
1337 devstr = strchr(sizestr, ':');
1338 if (devstr) {
1339 char *end;
1340 sizestr = devstr + 1;
1341 *devstr = '\0';
1342 devstr = vol_args->name;
1343 devid = simple_strtoull(devstr, &end, 10);
5bb14682 1344 printk(KERN_INFO "btrfs: resizing devid %llu\n",
21380931 1345 (unsigned long long)devid);
f46b5a66 1346 }
aa1b8cd4 1347 device = btrfs_find_device(root->fs_info, devid, NULL, NULL);
f46b5a66 1348 if (!device) {
5bb14682 1349 printk(KERN_INFO "btrfs: resizer unable to find device %llu\n",
21380931 1350 (unsigned long long)devid);
f46b5a66 1351 ret = -EINVAL;
c9e9f97b 1352 goto out_free;
f46b5a66 1353 }
4e42ae1b
LB		 1354 if (device->fs_devices && device->fs_devices->seeding) {
1355 printk(KERN_INFO "btrfs: resizer unable to apply on "
a8c4a33b
CM		 1356 "seeding device %llu\n",
1357 (unsigned long long)devid);
4e42ae1b
LB		 1358 ret = -EINVAL;
1359 goto out_free;
1360 }
1361
f46b5a66
CH		 1362 if (!strcmp(sizestr, "max"))
1363 new_size = device->bdev->bd_inode->i_size;
1364 else {
1365 if (sizestr[0] == '-') {
1366 mod = -1;
1367 sizestr++;
1368 } else if (sizestr[0] == '+') {
1369 mod = 1;
1370 sizestr++;
1371 }
91748467 1372 new_size = memparse(sizestr, NULL);
f46b5a66
CH		 1373 if (new_size == 0) {
1374 ret = -EINVAL;
c9e9f97b 1375 goto out_free;
f46b5a66
CH		 1376 }
1377 }
1378
63a212ab
SB		 1379 if (device->is_tgtdev_for_dev_replace) {
1380 ret = -EINVAL;
1381 goto out_free;
1382 }
1383
f46b5a66
CH		 1384 old_size = device->total_bytes;
1385
1386 if (mod < 0) {
1387 if (new_size > old_size) {
1388 ret = -EINVAL;
c9e9f97b 1389 goto out_free;
f46b5a66
CH		 1390 }
1391 new_size = old_size - new_size;
1392 } else if (mod > 0) {
1393 new_size = old_size + new_size;
1394 }
1395
1396 if (new_size < 256 * 1024 * 1024) {
1397 ret = -EINVAL;
c9e9f97b 1398 goto out_free;
f46b5a66
CH		 1399 }
1400 if (new_size > device->bdev->bd_inode->i_size) {
1401 ret = -EFBIG;
c9e9f97b 1402 goto out_free;
f46b5a66
CH		 1403 }
1404
1405 do_div(new_size, root->sectorsize);
1406 new_size *= root->sectorsize;
1407
606686ee
JB		 1408 printk_in_rcu(KERN_INFO "btrfs: new size for %s is %llu\n",
1409 rcu_str_deref(device->name),
1410 (unsigned long long)new_size);
f46b5a66
CH		 1411
1412 if (new_size > old_size) {
a22285a6 1413 trans = btrfs_start_transaction(root, 0);
98d5dc13
TI		 1414 if (IS_ERR(trans)) {
1415 ret = PTR_ERR(trans);
c9e9f97b 1416 goto out_free;
98d5dc13 1417 }
f46b5a66
CH		 1418 ret = btrfs_grow_device(trans, device, new_size);
1419 btrfs_commit_transaction(trans, root);
ece7d20e 1420 } else if (new_size < old_size) {
f46b5a66 1421 ret = btrfs_shrink_device(device, new_size);
0253f40e 1422 } /* equal, nothing need to do */
f46b5a66 1423
c9e9f97b 1424out_free:
f46b5a66 1425 kfree(vol_args);
c9e9f97b
ID		 1426out:
1427 mutex_unlock(&root->fs_info->volume_mutex);
5ac00add 1428 atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0);
f46b5a66
CH		 1429 return ret;
1430}
1431
72fd032e 1432static noinline int btrfs_ioctl_snap_create_transid(struct file *file,
6f72c7e2
AJ		 1433 char *name, unsigned long fd, int subvol,
1434 u64 *transid, bool readonly,
1435 struct btrfs_qgroup_inherit **inherit)
f46b5a66 1436{
f46b5a66 1437 int namelen;
3de4586c 1438 int ret = 0;
f46b5a66 1439
a874a63e
LB		 1440 ret = mnt_want_write_file(file);
1441 if (ret)
1442 goto out;
1443
72fd032e
SW		 1444 namelen = strlen(name);
1445 if (strchr(name, '/')) {
f46b5a66 1446 ret = -EINVAL;
a874a63e 1447 goto out_drop_write;
f46b5a66
CH		 1448 }
1449
16780cab
CM		 1450 if (name[0] == '.' &&
1451 (namelen == 1 || (name[1] == '.' && namelen == 2))) {
1452 ret = -EEXIST;
a874a63e 1453 goto out_drop_write;
16780cab
CM		 1454 }
1455
3de4586c 1456 if (subvol) {
72fd032e 1457 ret = btrfs_mksubvol(&file->f_path, name, namelen,
6f72c7e2 1458 NULL, transid, readonly, inherit);
cb8e7090 1459 } else {
2903ff01 1460 struct fd src = fdget(fd);
3de4586c 1461 struct inode *src_inode;
2903ff01 1462 if (!src.file) {
3de4586c 1463 ret = -EINVAL;
a874a63e 1464 goto out_drop_write;
3de4586c
CM		 1465 }
1466
2903ff01 1467 src_inode = src.file->f_path.dentry->d_inode;
3de4586c 1468 if (src_inode->i_sb != file->f_path.dentry->d_inode->i_sb) {
d397712b
CM		 1469 printk(KERN_INFO "btrfs: Snapshot src from "
1470 "another FS\n");
3de4586c 1471 ret = -EINVAL;
ecd18815
AV		 1472 } else {
1473 ret = btrfs_mksubvol(&file->f_path, name, namelen,
1474 BTRFS_I(src_inode)->root,
1475 transid, readonly, inherit);
3de4586c 1476 }
2903ff01 1477 fdput(src);
cb8e7090 1478 }
a874a63e
LB		 1479out_drop_write:
1480 mnt_drop_write_file(file);
f46b5a66 1481out:
72fd032e
SW		 1482 return ret;
1483}
1484
1485static noinline int btrfs_ioctl_snap_create(struct file *file,
fa0d2b9b 1486 void __user *arg, int subvol)
72fd032e 1487{
fa0d2b9b 1488 struct btrfs_ioctl_vol_args *vol_args;
72fd032e
SW		 1489 int ret;
1490
fa0d2b9b
LZ		 1491 vol_args = memdup_user(arg, sizeof(*vol_args));
1492 if (IS_ERR(vol_args))
1493 return PTR_ERR(vol_args);
1494 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
72fd032e 1495
fa0d2b9b 1496 ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
b83cc969 1497 vol_args->fd, subvol,
6f72c7e2 1498 NULL, false, NULL);
fdfb1e4f 1499
fa0d2b9b
LZ		 1500 kfree(vol_args);
1501 return ret;
1502}
fdfb1e4f 1503
fa0d2b9b
LZ		 1504static noinline int btrfs_ioctl_snap_create_v2(struct file *file,
1505 void __user *arg, int subvol)
1506{
1507 struct btrfs_ioctl_vol_args_v2 *vol_args;
1508 int ret;
1509 u64 transid = 0;
1510 u64 *ptr = NULL;
b83cc969 1511 bool readonly = false;
6f72c7e2 1512 struct btrfs_qgroup_inherit *inherit = NULL;
75eaa0e2 1513
fa0d2b9b
LZ		 1514 vol_args = memdup_user(arg, sizeof(*vol_args));
1515 if (IS_ERR(vol_args))
1516 return PTR_ERR(vol_args);
1517 vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
75eaa0e2 1518
b83cc969 1519 if (vol_args->flags &
6f72c7e2
AJ		 1520 ~(BTRFS_SUBVOL_CREATE_ASYNC | BTRFS_SUBVOL_RDONLY |
1521 BTRFS_SUBVOL_QGROUP_INHERIT)) {
b83cc969 1522 ret = -EOPNOTSUPP;
fa0d2b9b 1523 goto out;
72fd032e 1524 }
fa0d2b9b
LZ		 1525
1526 if (vol_args->flags & BTRFS_SUBVOL_CREATE_ASYNC)
1527 ptr = &transid;
b83cc969
LZ		 1528 if (vol_args->flags & BTRFS_SUBVOL_RDONLY)
1529 readonly = true;
6f72c7e2
AJ		 1530 if (vol_args->flags & BTRFS_SUBVOL_QGROUP_INHERIT) {
1531 if (vol_args->size > PAGE_CACHE_SIZE) {
1532 ret = -EINVAL;
1533 goto out;
1534 }
1535 inherit = memdup_user(vol_args->qgroup_inherit, vol_args->size);
1536 if (IS_ERR(inherit)) {
1537 ret = PTR_ERR(inherit);
1538 goto out;
1539 }
1540 }
fa0d2b9b
LZ		 1541
1542 ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
6f72c7e2
AJ		 1543 vol_args->fd, subvol, ptr,
1544 readonly, &inherit);
fa0d2b9b
LZ		 1545
1546 if (ret == 0 && ptr &&
1547 copy_to_user(arg +
1548 offsetof(struct btrfs_ioctl_vol_args_v2,
1549 transid), ptr, sizeof(*ptr)))
1550 ret = -EFAULT;
fdfb1e4f 1551out:
f46b5a66 1552 kfree(vol_args);
6f72c7e2 1553 kfree(inherit);
f46b5a66
CH		 1554 return ret;
1555}
1556
0caa102d
LZ		 1557static noinline int btrfs_ioctl_subvol_getflags(struct file *file,
1558 void __user *arg)
1559{
1560 struct inode *inode = fdentry(file)->d_inode;
1561 struct btrfs_root *root = BTRFS_I(inode)->root;
1562 int ret = 0;
1563 u64 flags = 0;
1564
33345d01 1565 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID)
0caa102d
LZ		 1566 return -EINVAL;
1567
1568 down_read(&root->fs_info->subvol_sem);
1569 if (btrfs_root_readonly(root))
1570 flags |= BTRFS_SUBVOL_RDONLY;
1571 up_read(&root->fs_info->subvol_sem);
1572
1573 if (copy_to_user(arg, &flags, sizeof(flags)))
1574 ret = -EFAULT;
1575
1576 return ret;
1577}
1578
1579static noinline int btrfs_ioctl_subvol_setflags(struct file *file,
1580 void __user *arg)
1581{
1582 struct inode *inode = fdentry(file)->d_inode;
1583 struct btrfs_root *root = BTRFS_I(inode)->root;
1584 struct btrfs_trans_handle *trans;
1585 u64 root_flags;
1586 u64 flags;
1587 int ret = 0;
1588
b9ca0664
LB		 1589 ret = mnt_want_write_file(file);
1590 if (ret)
1591 goto out;
0caa102d 1592
b9ca0664
LB		 1593 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) {
1594 ret = -EINVAL;
1595 goto out_drop_write;
1596 }
0caa102d 1597
b9ca0664
LB		 1598 if (copy_from_user(&flags, arg, sizeof(flags))) {
1599 ret = -EFAULT;
1600 goto out_drop_write;
1601 }
0caa102d 1602
b9ca0664
LB		 1603 if (flags & BTRFS_SUBVOL_CREATE_ASYNC) {
1604 ret = -EINVAL;
1605 goto out_drop_write;
1606 }
0caa102d 1607
b9ca0664
LB		 1608 if (flags & ~BTRFS_SUBVOL_RDONLY) {
1609 ret = -EOPNOTSUPP;
1610 goto out_drop_write;
1611 }
0caa102d 1612
b9ca0664
LB		 1613 if (!inode_owner_or_capable(inode)) {
1614 ret = -EACCES;
1615 goto out_drop_write;
1616 }
b4dc2b8c 1617
0caa102d
LZ		 1618 down_write(&root->fs_info->subvol_sem);
1619
1620 /* nothing to do */
1621 if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root))
b9ca0664 1622 goto out_drop_sem;
0caa102d
LZ		 1623
1624 root_flags = btrfs_root_flags(&root->root_item);
1625 if (flags & BTRFS_SUBVOL_RDONLY)
1626 btrfs_set_root_flags(&root->root_item,
1627 root_flags | BTRFS_ROOT_SUBVOL_RDONLY);
1628 else
1629 btrfs_set_root_flags(&root->root_item,
1630 root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY);
1631
1632 trans = btrfs_start_transaction(root, 1);
1633 if (IS_ERR(trans)) {
1634 ret = PTR_ERR(trans);
1635 goto out_reset;
1636 }
1637
b4dc2b8c 1638 ret = btrfs_update_root(trans, root->fs_info->tree_root,
0caa102d
LZ		 1639 &root->root_key, &root->root_item);
1640
1641 btrfs_commit_transaction(trans, root);
1642out_reset:
1643 if (ret)
1644 btrfs_set_root_flags(&root->root_item, root_flags);
b9ca0664 1645out_drop_sem:
0caa102d 1646 up_write(&root->fs_info->subvol_sem);
b9ca0664
LB		 1647out_drop_write:
1648 mnt_drop_write_file(file);
1649out:
0caa102d
LZ		 1650 return ret;
1651}
1652
76dda93c
YZ		 1653/*
1654 * helper to check if the subvolume references other subvolumes
1655 */
1656static noinline int may_destroy_subvol(struct btrfs_root *root)
1657{
1658 struct btrfs_path *path;
1659 struct btrfs_key key;
1660 int ret;
1661
1662 path = btrfs_alloc_path();
1663 if (!path)
1664 return -ENOMEM;
1665
1666 key.objectid = root->root_key.objectid;
1667 key.type = BTRFS_ROOT_REF_KEY;
1668 key.offset = (u64)-1;
1669
1670 ret = btrfs_search_slot(NULL, root->fs_info->tree_root,
1671 &key, path, 0, 0);
1672 if (ret < 0)
1673 goto out;
1674 BUG_ON(ret == 0);
1675
1676 ret = 0;
1677 if (path->slots[0] > 0) {
1678 path->slots[0]--;
1679 btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
1680 if (key.objectid == root->root_key.objectid &&
1681 key.type == BTRFS_ROOT_REF_KEY)
1682 ret = -ENOTEMPTY;
1683 }
1684out:
1685 btrfs_free_path(path);
1686 return ret;
1687}
1688
ac8e9819
CM		 1689static noinline int key_in_sk(struct btrfs_key *key,
1690 struct btrfs_ioctl_search_key *sk)
1691{
abc6e134
CM		 1692 struct btrfs_key test;
1693 int ret;
1694
1695 test.objectid = sk->min_objectid;
1696 test.type = sk->min_type;
1697 test.offset = sk->min_offset;
1698
1699 ret = btrfs_comp_cpu_keys(key, &test);
1700 if (ret < 0)
ac8e9819 1701 return 0;
abc6e134
CM		 1702
1703 test.objectid = sk->max_objectid;
1704 test.type = sk->max_type;
1705 test.offset = sk->max_offset;
1706
1707 ret = btrfs_comp_cpu_keys(key, &test);
1708 if (ret > 0)
ac8e9819
CM		 1709 return 0;
1710 return 1;
1711}
1712
1713static noinline int copy_to_sk(struct btrfs_root *root,
1714 struct btrfs_path *path,
1715 struct btrfs_key *key,
1716 struct btrfs_ioctl_search_key *sk,
1717 char *buf,
1718 unsigned long *sk_offset,
1719 int *num_found)
1720{
1721 u64 found_transid;
1722 struct extent_buffer *leaf;
1723 struct btrfs_ioctl_search_header sh;
1724 unsigned long item_off;
1725 unsigned long item_len;
1726 int nritems;
1727 int i;
1728 int slot;
ac8e9819
CM		 1729 int ret = 0;
1730
1731 leaf = path->nodes[0];
1732 slot = path->slots[0];
1733 nritems = btrfs_header_nritems(leaf);
1734
1735 if (btrfs_header_generation(leaf) > sk->max_transid) {
1736 i = nritems;
1737 goto advance_key;
1738 }
1739 found_transid = btrfs_header_generation(leaf);
1740
1741 for (i = slot; i < nritems; i++) {
1742 item_off = btrfs_item_ptr_offset(leaf, i);
1743 item_len = btrfs_item_size_nr(leaf, i);
1744
1745 if (item_len > BTRFS_SEARCH_ARGS_BUFSIZE)
1746 item_len = 0;
1747
1748 if (sizeof(sh) + item_len + *sk_offset >
1749 BTRFS_SEARCH_ARGS_BUFSIZE) {
1750 ret = 1;
1751 goto overflow;
1752 }
1753
1754 btrfs_item_key_to_cpu(leaf, key, i);
1755 if (!key_in_sk(key, sk))
1756 continue;
1757
1758 sh.objectid = key->objectid;
1759 sh.offset = key->offset;
1760 sh.type = key->type;
1761 sh.len = item_len;
1762 sh.transid = found_transid;
1763
1764 /* copy search result header */
1765 memcpy(buf + *sk_offset, &sh, sizeof(sh));
1766 *sk_offset += sizeof(sh);
1767
1768 if (item_len) {
1769 char *p = buf + *sk_offset;
1770 /* copy the item */
1771 read_extent_buffer(leaf, p,
1772 item_off, item_len);
1773 *sk_offset += item_len;
ac8e9819 1774 }
e2156867 1775 (*num_found)++;
ac8e9819
CM		 1776
1777 if (*num_found >= sk->nr_items)
1778 break;
1779 }
1780advance_key:
abc6e134
CM		 1781 ret = 0;
1782 if (key->offset < (u64)-1 && key->offset < sk->max_offset)
ac8e9819 1783 key->offset++;
abc6e134
CM		 1784 else if (key->type < (u8)-1 && key->type < sk->max_type) {
1785 key->offset = 0;
ac8e9819 1786 key->type++;
abc6e134
CM		 1787 } else if (key->objectid < (u64)-1 && key->objectid < sk->max_objectid) {
1788 key->offset = 0;
1789 key->type = 0;
ac8e9819 1790 key->objectid++;
abc6e134
CM		 1791 } else
1792 ret = 1;
ac8e9819 1793overflow:
ac8e9819
CM		 1794 return ret;
1795}
1796
1797static noinline int search_ioctl(struct inode *inode,
1798 struct btrfs_ioctl_search_args *args)
1799{
1800 struct btrfs_root *root;
1801 struct btrfs_key key;
1802 struct btrfs_key max_key;
1803 struct btrfs_path *path;
1804 struct btrfs_ioctl_search_key *sk = &args->key;
1805 struct btrfs_fs_info *info = BTRFS_I(inode)->root->fs_info;
1806 int ret;
1807 int num_found = 0;
1808 unsigned long sk_offset = 0;
1809
1810 path = btrfs_alloc_path();
1811 if (!path)
1812 return -ENOMEM;
1813
1814 if (sk->tree_id == 0) {
1815 /* search the root of the inode that was passed */
1816 root = BTRFS_I(inode)->root;
1817 } else {
1818 key.objectid = sk->tree_id;
1819 key.type = BTRFS_ROOT_ITEM_KEY;
1820 key.offset = (u64)-1;
1821 root = btrfs_read_fs_root_no_name(info, &key);
1822 if (IS_ERR(root)) {
1823 printk(KERN_ERR "could not find root %llu\n",
1824 sk->tree_id);
1825 btrfs_free_path(path);
1826 return -ENOENT;
1827 }
1828 }
1829
1830 key.objectid = sk->min_objectid;
1831 key.type = sk->min_type;
1832 key.offset = sk->min_offset;
1833
1834 max_key.objectid = sk->max_objectid;
1835 max_key.type = sk->max_type;
1836 max_key.offset = sk->max_offset;
1837
1838 path->keep_locks = 1;
1839
1840 while(1) {
1841 ret = btrfs_search_forward(root, &key, &max_key, path, 0,
1842 sk->min_transid);
1843 if (ret != 0) {
1844 if (ret > 0)
1845 ret = 0;
1846 goto err;
1847 }
1848 ret = copy_to_sk(root, path, &key, sk, args->buf,
1849 &sk_offset, &num_found);
b3b4aa74 1850 btrfs_release_path(path);
ac8e9819
CM		 1851 if (ret || num_found >= sk->nr_items)
1852 break;
1853
1854 }
1855 ret = 0;
1856err:
1857 sk->nr_items = num_found;
1858 btrfs_free_path(path);
1859 return ret;
1860}
1861
1862static noinline int btrfs_ioctl_tree_search(struct file *file,
1863 void __user *argp)
1864{
1865 struct btrfs_ioctl_search_args *args;
1866 struct inode *inode;
1867 int ret;
1868
1869 if (!capable(CAP_SYS_ADMIN))
1870 return -EPERM;
1871
2354d08f
JL		 1872 args = memdup_user(argp, sizeof(*args));
1873 if (IS_ERR(args))
1874 return PTR_ERR(args);
ac8e9819 1875
ac8e9819
CM		 1876 inode = fdentry(file)->d_inode;
1877 ret = search_ioctl(inode, args);
1878 if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
1879 ret = -EFAULT;
1880 kfree(args);
1881 return ret;
1882}
1883
98d377a0 1884/*
ac8e9819
CM		 1885 * Search INODE_REFs to identify path name of 'dirid' directory
1886 * in a 'tree_id' tree. and sets path name to 'name'.
1887 */
98d377a0
TH		 1888static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info,
1889 u64 tree_id, u64 dirid, char *name)
1890{
1891 struct btrfs_root *root;
1892 struct btrfs_key key;
ac8e9819 1893 char *ptr;
98d377a0
TH		 1894 int ret = -1;
1895 int slot;
1896 int len;
1897 int total_len = 0;
1898 struct btrfs_inode_ref *iref;
1899 struct extent_buffer *l;
1900 struct btrfs_path *path;
1901
1902 if (dirid == BTRFS_FIRST_FREE_OBJECTID) {
1903 name[0]='\0';
1904 return 0;
1905 }
1906
1907 path = btrfs_alloc_path();
1908 if (!path)
1909 return -ENOMEM;
1910
ac8e9819 1911 ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX];
98d377a0
TH		 1912
1913 key.objectid = tree_id;
1914 key.type = BTRFS_ROOT_ITEM_KEY;
1915 key.offset = (u64)-1;
1916 root = btrfs_read_fs_root_no_name(info, &key);
1917 if (IS_ERR(root)) {
1918 printk(KERN_ERR "could not find root %llu\n", tree_id);
8ad6fcab
CM		 1919 ret = -ENOENT;
1920 goto out;
98d377a0
TH		 1921 }
1922
1923 key.objectid = dirid;
1924 key.type = BTRFS_INODE_REF_KEY;
8ad6fcab 1925 key.offset = (u64)-1;
98d377a0
TH		 1926
1927 while(1) {
1928 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1929 if (ret < 0)
1930 goto out;
1931
1932 l = path->nodes[0];
1933 slot = path->slots[0];
8ad6fcab
CM		 1934 if (ret > 0 && slot > 0)
1935 slot--;
98d377a0
TH		 1936 btrfs_item_key_to_cpu(l, &key, slot);
1937
1938 if (ret > 0 && (key.objectid != dirid ||
ac8e9819
CM		 1939 key.type != BTRFS_INODE_REF_KEY)) {
1940 ret = -ENOENT;
98d377a0 1941 goto out;
ac8e9819 1942 }
98d377a0
TH		 1943
1944 iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref);
1945 len = btrfs_inode_ref_name_len(l, iref);
1946 ptr -= len + 1;
1947 total_len += len + 1;
ac8e9819 1948 if (ptr < name)
98d377a0
TH		 1949 goto out;
1950
1951 *(ptr + len) = '/';
1952 read_extent_buffer(l, ptr,(unsigned long)(iref + 1), len);
1953
1954 if (key.offset == BTRFS_FIRST_FREE_OBJECTID)
1955 break;
1956
b3b4aa74 1957 btrfs_release_path(path);
98d377a0 1958 key.objectid = key.offset;
8ad6fcab 1959 key.offset = (u64)-1;
98d377a0 1960 dirid = key.objectid;
98d377a0 1961 }
ac8e9819 1962 if (ptr < name)
98d377a0 1963 goto out;
77906a50 1964 memmove(name, ptr, total_len);
98d377a0
TH		 1965 name[total_len]='\0';
1966 ret = 0;
1967out:
1968 btrfs_free_path(path);
ac8e9819
CM		 1969 return ret;
1970}
1971
1972static noinline int btrfs_ioctl_ino_lookup(struct file *file,
1973 void __user *argp)
1974{
1975 struct btrfs_ioctl_ino_lookup_args *args;
1976 struct inode *inode;
1977 int ret;
1978
1979 if (!capable(CAP_SYS_ADMIN))
1980 return -EPERM;
1981
2354d08f
JL		 1982 args = memdup_user(argp, sizeof(*args));
1983 if (IS_ERR(args))
1984 return PTR_ERR(args);
c2b96929 1985
ac8e9819
CM		 1986 inode = fdentry(file)->d_inode;
1987
1b53ac4d
CM		 1988 if (args->treeid == 0)
1989 args->treeid = BTRFS_I(inode)->root->root_key.objectid;
1990
ac8e9819
CM		 1991 ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info,
1992 args->treeid, args->objectid,
1993 args->name);
1994
1995 if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
1996 ret = -EFAULT;
1997
1998 kfree(args);
98d377a0
TH		 1999 return ret;
2000}
2001
76dda93c
YZ		 2002static noinline int btrfs_ioctl_snap_destroy(struct file *file,
2003 void __user *arg)
2004{
2005 struct dentry *parent = fdentry(file);
2006 struct dentry *dentry;
2007 struct inode *dir = parent->d_inode;
2008 struct inode *inode;
2009 struct btrfs_root *root = BTRFS_I(dir)->root;
2010 struct btrfs_root *dest = NULL;
2011 struct btrfs_ioctl_vol_args *vol_args;
2012 struct btrfs_trans_handle *trans;
2013 int namelen;
2014 int ret;
2015 int err = 0;
2016
76dda93c
YZ		 2017 vol_args = memdup_user(arg, sizeof(*vol_args));
2018 if (IS_ERR(vol_args))
2019 return PTR_ERR(vol_args);
2020
2021 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2022 namelen = strlen(vol_args->name);
2023 if (strchr(vol_args->name, '/') ||
2024 strncmp(vol_args->name, "..", namelen) == 0) {
2025 err = -EINVAL;
2026 goto out;
2027 }
2028
a561be71 2029 err = mnt_want_write_file(file);
76dda93c
YZ		 2030 if (err)
2031 goto out;
2032
2033 mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
2034 dentry = lookup_one_len(vol_args->name, parent, namelen);
2035 if (IS_ERR(dentry)) {
2036 err = PTR_ERR(dentry);
2037 goto out_unlock_dir;
2038 }
2039
2040 if (!dentry->d_inode) {
2041 err = -ENOENT;
2042 goto out_dput;
2043 }
2044
2045 inode = dentry->d_inode;
4260f7c7
SW		 2046 dest = BTRFS_I(inode)->root;
2047 if (!capable(CAP_SYS_ADMIN)){
2048 /*
2049 * Regular user. Only allow this with a special mount
2050 * option, when the user has write+exec access to the
2051 * subvol root, and when rmdir(2) would have been
2052 * allowed.
2053 *
2054 * Note that this is _not_ check that the subvol is
2055 * empty or doesn't contain data that we wouldn't
2056 * otherwise be able to delete.
2057 *
2058 * Users who want to delete empty subvols should try
2059 * rmdir(2).
2060 */
2061 err = -EPERM;
2062 if (!btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED))
2063 goto out_dput;
2064
2065 /*
2066 * Do not allow deletion if the parent dir is the same
2067 * as the dir to be deleted. That means the ioctl
2068 * must be called on the dentry referencing the root
2069 * of the subvol, not a random directory contained
2070 * within it.
2071 */
2072 err = -EINVAL;
2073 if (root == dest)
2074 goto out_dput;
2075
2076 err = inode_permission(inode, MAY_WRITE | MAY_EXEC);
2077 if (err)
2078 goto out_dput;
2079
2080 /* check if subvolume may be deleted by a non-root user */
2081 err = btrfs_may_delete(dir, dentry, 1);
2082 if (err)
2083 goto out_dput;
2084 }
2085
33345d01 2086 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) {
76dda93c
YZ		 2087 err = -EINVAL;
2088 goto out_dput;
2089 }
2090
76dda93c
YZ		 2091 mutex_lock(&inode->i_mutex);
2092 err = d_invalidate(dentry);
2093 if (err)
2094 goto out_unlock;
2095
2096 down_write(&root->fs_info->subvol_sem);
2097
2098 err = may_destroy_subvol(dest);
2099 if (err)
2100 goto out_up_write;
2101
a22285a6
YZ		 2102 trans = btrfs_start_transaction(root, 0);
2103 if (IS_ERR(trans)) {
2104 err = PTR_ERR(trans);
d327099a 2105 goto out_up_write;
a22285a6
YZ		 2106 }
2107 trans->block_rsv = &root->fs_info->global_block_rsv;
2108
76dda93c
YZ		 2109 ret = btrfs_unlink_subvol(trans, root, dir,
2110 dest->root_key.objectid,
2111 dentry->d_name.name,
2112 dentry->d_name.len);
79787eaa
JM		 2113 if (ret) {
2114 err = ret;
2115 btrfs_abort_transaction(trans, root, ret);
2116 goto out_end_trans;
2117 }
76dda93c
YZ		 2118
2119 btrfs_record_root_in_trans(trans, dest);
2120
2121 memset(&dest->root_item.drop_progress, 0,
2122 sizeof(dest->root_item.drop_progress));
2123 dest->root_item.drop_level = 0;
2124 btrfs_set_root_refs(&dest->root_item, 0);
2125
d68fc57b
YZ		 2126 if (!xchg(&dest->orphan_item_inserted, 1)) {
2127 ret = btrfs_insert_orphan_item(trans,
2128 root->fs_info->tree_root,
2129 dest->root_key.objectid);
79787eaa
JM		 2130 if (ret) {
2131 btrfs_abort_transaction(trans, root, ret);
2132 err = ret;
2133 goto out_end_trans;
2134 }
d68fc57b 2135 }
79787eaa 2136out_end_trans:
531cb13f 2137 ret = btrfs_end_transaction(trans, root);
79787eaa
JM		 2138 if (ret && !err)
2139 err = ret;
76dda93c
YZ		 2140 inode->i_flags |= S_DEAD;
2141out_up_write:
2142 up_write(&root->fs_info->subvol_sem);
2143out_unlock:
2144 mutex_unlock(&inode->i_mutex);
2145 if (!err) {
efefb143 2146 shrink_dcache_sb(root->fs_info->sb);
76dda93c
YZ		 2147 btrfs_invalidate_inodes(dest);
2148 d_delete(dentry);
2149 }
2150out_dput:
2151 dput(dentry);
2152out_unlock_dir:
2153 mutex_unlock(&dir->i_mutex);
2a79f17e 2154 mnt_drop_write_file(file);
76dda93c
YZ		 2155out:
2156 kfree(vol_args);
2157 return err;
2158}
2159
1e701a32 2160static int btrfs_ioctl_defrag(struct file *file, void __user *argp)
f46b5a66
CH		 2161{
2162 struct inode *inode = fdentry(file)->d_inode;
2163 struct btrfs_root *root = BTRFS_I(inode)->root;
1e701a32 2164 struct btrfs_ioctl_defrag_range_args *range;
c146afad
YZ		 2165 int ret;
2166
b83cc969
LZ		 2167 if (btrfs_root_readonly(root))
2168 return -EROFS;
2169
5ac00add
SB		 2170 if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running,
2171 1)) {
2172 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
2173 return -EINPROGRESS;
2174 }
a561be71 2175 ret = mnt_want_write_file(file);
5ac00add
SB		 2176 if (ret) {
2177 atomic_set(&root->fs_info->mutually_exclusive_operation_running,
2178 0);
c146afad 2179 return ret;
5ac00add 2180 }
f46b5a66
CH		 2181
2182 switch (inode->i_mode & S_IFMT) {
2183 case S_IFDIR:
e441d54d
CM		 2184 if (!capable(CAP_SYS_ADMIN)) {
2185 ret = -EPERM;
2186 goto out;
2187 }
8929ecfa
YZ		 2188 ret = btrfs_defrag_root(root, 0);
2189 if (ret)
2190 goto out;
2191 ret = btrfs_defrag_root(root->fs_info->extent_root, 0);
f46b5a66
CH		 2192 break;
2193 case S_IFREG:
e441d54d
CM		 2194 if (!(file->f_mode & FMODE_WRITE)) {
2195 ret = -EINVAL;
2196 goto out;
2197 }
1e701a32
CM		 2198
2199 range = kzalloc(sizeof(*range), GFP_KERNEL);
2200 if (!range) {
2201 ret = -ENOMEM;
2202 goto out;
2203 }
2204
2205 if (argp) {
2206 if (copy_from_user(range, argp,
2207 sizeof(*range))) {
2208 ret = -EFAULT;
2209 kfree(range);
683be16e 2210 goto out;
1e701a32
CM		 2211 }
2212 /* compression requires us to start the IO */
2213 if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
2214 range->flags |= BTRFS_DEFRAG_RANGE_START_IO;
2215 range->extent_thresh = (u32)-1;
2216 }
2217 } else {
2218 /* the rest are all set to zero by kzalloc */
2219 range->len = (u64)-1;
2220 }
4cb5300b
CM		 2221 ret = btrfs_defrag_file(fdentry(file)->d_inode, file,
2222 range, 0, 0);
2223 if (ret > 0)
2224 ret = 0;
1e701a32 2225 kfree(range);
f46b5a66 2226 break;
8929ecfa
YZ		 2227 default:
2228 ret = -EINVAL;
f46b5a66 2229 }
e441d54d 2230out:
2a79f17e 2231 mnt_drop_write_file(file);
5ac00add 2232 atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0);
e441d54d 2233 return ret;
f46b5a66
CH		 2234}
2235
b2950863 2236static long btrfs_ioctl_add_dev(struct btrfs_root *root, void __user *arg)
f46b5a66
CH		 2237{
2238 struct btrfs_ioctl_vol_args *vol_args;
2239 int ret;
2240
e441d54d
CM		 2241 if (!capable(CAP_SYS_ADMIN))
2242 return -EPERM;
2243
5ac00add
SB		 2244 if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running,
2245 1)) {
2246 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
2247 return -EINPROGRESS;
c9e9f97b
ID		 2248 }
2249
5ac00add 2250 mutex_lock(&root->fs_info->volume_mutex);
dae7b665 2251 vol_args = memdup_user(arg, sizeof(*vol_args));
c9e9f97b
ID		 2252 if (IS_ERR(vol_args)) {
2253 ret = PTR_ERR(vol_args);
2254 goto out;
2255 }
f46b5a66 2256
5516e595 2257 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
f46b5a66
CH		 2258 ret = btrfs_init_new_device(root, vol_args->name);
2259
f46b5a66 2260 kfree(vol_args);
c9e9f97b
ID		 2261out:
2262 mutex_unlock(&root->fs_info->volume_mutex);
5ac00add 2263 atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0);
f46b5a66
CH		 2264 return ret;
2265}
2266
b2950863 2267static long btrfs_ioctl_rm_dev(struct btrfs_root *root, void __user *arg)
f46b5a66
CH		 2268{
2269 struct btrfs_ioctl_vol_args *vol_args;
2270 int ret;
2271
e441d54d
CM		 2272 if (!capable(CAP_SYS_ADMIN))
2273 return -EPERM;
2274
c146afad
YZ		 2275 if (root->fs_info->sb->s_flags & MS_RDONLY)
2276 return -EROFS;
2277
5ac00add
SB		 2278 if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running,
2279 1)) {
2280 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
2281 return -EINPROGRESS;
c9e9f97b
ID		 2282 }
2283
5ac00add 2284 mutex_lock(&root->fs_info->volume_mutex);
dae7b665 2285 vol_args = memdup_user(arg, sizeof(*vol_args));
c9e9f97b
ID		 2286 if (IS_ERR(vol_args)) {
2287 ret = PTR_ERR(vol_args);
2288 goto out;
2289 }
f46b5a66 2290
5516e595 2291 vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
f46b5a66
CH		 2292 ret = btrfs_rm_device(root, vol_args->name);
2293
f46b5a66 2294 kfree(vol_args);
c9e9f97b
ID		 2295out:
2296 mutex_unlock(&root->fs_info->volume_mutex);
5ac00add 2297 atomic_set(&root->fs_info->mutually_exclusive_operation_running, 0);
f46b5a66
CH		 2298 return ret;
2299}
2300
475f6387
JS		 2301static long btrfs_ioctl_fs_info(struct btrfs_root *root, void __user *arg)
2302{
027ed2f0 2303 struct btrfs_ioctl_fs_info_args *fi_args;
475f6387
JS		 2304 struct btrfs_device *device;
2305 struct btrfs_device *next;
2306 struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
027ed2f0 2307 int ret = 0;
475f6387
JS		 2308
2309 if (!capable(CAP_SYS_ADMIN))
2310 return -EPERM;
2311
027ed2f0
LZ		 2312 fi_args = kzalloc(sizeof(*fi_args), GFP_KERNEL);
2313 if (!fi_args)
2314 return -ENOMEM;
2315
2316 fi_args->num_devices = fs_devices->num_devices;
2317 memcpy(&fi_args->fsid, root->fs_info->fsid, sizeof(fi_args->fsid));
475f6387
JS		 2318
2319 mutex_lock(&fs_devices->device_list_mutex);
2320 list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
027ed2f0
LZ		 2321 if (device->devid > fi_args->max_id)
2322 fi_args->max_id = device->devid;
475f6387
JS		 2323 }
2324 mutex_unlock(&fs_devices->device_list_mutex);
2325
027ed2f0
LZ		 2326 if (copy_to_user(arg, fi_args, sizeof(*fi_args)))
2327 ret = -EFAULT;
475f6387 2328
027ed2f0
LZ		 2329 kfree(fi_args);
2330 return ret;
475f6387
JS		 2331}
2332
2333static long btrfs_ioctl_dev_info(struct btrfs_root *root, void __user *arg)
2334{
2335 struct btrfs_ioctl_dev_info_args *di_args;
2336 struct btrfs_device *dev;
2337 struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
2338 int ret = 0;
2339 char *s_uuid = NULL;
2340 char empty_uuid[BTRFS_UUID_SIZE] = {0};
2341
2342 if (!capable(CAP_SYS_ADMIN))
2343 return -EPERM;
2344
2345 di_args = memdup_user(arg, sizeof(*di_args));
2346 if (IS_ERR(di_args))
2347 return PTR_ERR(di_args);
2348
2349 if (memcmp(empty_uuid, di_args->uuid, BTRFS_UUID_SIZE) != 0)
2350 s_uuid = di_args->uuid;
2351
2352 mutex_lock(&fs_devices->device_list_mutex);
aa1b8cd4 2353 dev = btrfs_find_device(root->fs_info, di_args->devid, s_uuid, NULL);
475f6387
JS		 2354 mutex_unlock(&fs_devices->device_list_mutex);
2355
2356 if (!dev) {
2357 ret = -ENODEV;
2358 goto out;
2359 }
2360
2361 di_args->devid = dev->devid;
2362 di_args->bytes_used = dev->bytes_used;
2363 di_args->total_bytes = dev->total_bytes;
2364 memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid));
a27202fb 2365 if (dev->name) {
606686ee
JB		 2366 struct rcu_string *name;
2367
2368 rcu_read_lock();
2369 name = rcu_dereference(dev->name);
2370 strncpy(di_args->path, name->str, sizeof(di_args->path));
2371 rcu_read_unlock();
a27202fb
JM		 2372 di_args->path[sizeof(di_args->path) - 1] = 0;
2373 } else {
99ba55ad 2374 di_args->path[0] = '\0';
a27202fb 2375 }
475f6387
JS		 2376
2377out:
2378 if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args)))
2379 ret = -EFAULT;
2380
2381 kfree(di_args);
2382 return ret;
2383}
2384
76dda93c
YZ		 2385static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
2386 u64 off, u64 olen, u64 destoff)
f46b5a66
CH		 2387{
2388 struct inode *inode = fdentry(file)->d_inode;
2389 struct btrfs_root *root = BTRFS_I(inode)->root;
2903ff01 2390 struct fd src_file;
f46b5a66
CH		 2391 struct inode *src;
2392 struct btrfs_trans_handle *trans;
f46b5a66 2393 struct btrfs_path *path;
f46b5a66 2394 struct extent_buffer *leaf;
ae01a0ab
YZ		 2395 char *buf;
2396 struct btrfs_key key;
f46b5a66
CH		 2397 u32 nritems;
2398 int slot;
ae01a0ab 2399 int ret;
c5c9cd4d
SW		 2400 u64 len = olen;
2401 u64 bs = root->fs_info->sb->s_blocksize;
d20f7043 2402
c5c9cd4d
SW		 2403 /*
2404 * TODO:
2405 * - split compressed inline extents. annoying: we need to
2406 * decompress into destination's address_space (the file offset
2407 * may change, so source mapping won't do), then recompress (or
2408 * otherwise reinsert) a subrange.
2409 * - allow ranges within the same file to be cloned (provided
2410 * they don't overlap)?
2411 */
2412
e441d54d 2413 /* the destination must be opened for writing */
2ebc3464 2414 if (!(file->f_mode & FMODE_WRITE) || (file->f_flags & O_APPEND))
e441d54d
CM		 2415 return -EINVAL;
2416
b83cc969
LZ		 2417 if (btrfs_root_readonly(root))
2418 return -EROFS;
2419
a561be71 2420 ret = mnt_want_write_file(file);
c146afad
YZ		 2421 if (ret)
2422 return ret;
2423
2903ff01
AV		 2424 src_file = fdget(srcfd);
2425 if (!src_file.file) {
ab67b7c1
YZ		 2426 ret = -EBADF;
2427 goto out_drop_write;
2428 }
5dc64164 2429
362a20c5 2430 ret = -EXDEV;
2903ff01 2431 if (src_file.file->f_path.mnt != file->f_path.mnt)
362a20c5
DS		 2432 goto out_fput;
2433
2903ff01 2434 src = src_file.file->f_dentry->d_inode;
f46b5a66 2435
c5c9cd4d
SW		 2436 ret = -EINVAL;
2437 if (src == inode)
2438 goto out_fput;
2439
5dc64164 2440 /* the src must be open for reading */
2903ff01 2441 if (!(src_file.file->f_mode & FMODE_READ))
5dc64164
DR		 2442 goto out_fput;
2443
0e7b824c
LZ		 2444 /* don't make the dst file partly checksummed */
2445 if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) !=
2446 (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM))
2447 goto out_fput;
2448
ae01a0ab
YZ		 2449 ret = -EISDIR;
2450 if (S_ISDIR(src->i_mode) || S_ISDIR(inode->i_mode))
2451 goto out_fput;
2452
f46b5a66 2453 ret = -EXDEV;
362a20c5 2454 if (src->i_sb != inode->i_sb)
ae01a0ab
YZ		 2455 goto out_fput;
2456
2457 ret = -ENOMEM;
2458 buf = vmalloc(btrfs_level_size(root, 0));
2459 if (!buf)
2460 goto out_fput;
2461
2462 path = btrfs_alloc_path();
2463 if (!path) {
2464 vfree(buf);
f46b5a66 2465 goto out_fput;
ae01a0ab
YZ		 2466 }
2467 path->reada = 2;
f46b5a66
CH		 2468
2469 if (inode < src) {
fccdae43
SW		 2470 mutex_lock_nested(&inode->i_mutex, I_MUTEX_PARENT);
2471 mutex_lock_nested(&src->i_mutex, I_MUTEX_CHILD);
f46b5a66 2472 } else {
fccdae43
SW		 2473 mutex_lock_nested(&src->i_mutex, I_MUTEX_PARENT);
2474 mutex_lock_nested(&inode->i_mutex, I_MUTEX_CHILD);
f46b5a66
CH		 2475 }
2476
c5c9cd4d
SW		 2477 /* determine range to clone */
2478 ret = -EINVAL;
2ebc3464 2479 if (off + len > src->i_size || off + len < off)
f46b5a66 2480 goto out_unlock;
c5c9cd4d
SW		 2481 if (len == 0)
2482 olen = len = src->i_size - off;
2483 /* if we extend to eof, continue to block boundary */
2484 if (off + len == src->i_size)
2a6b8dae 2485 len = ALIGN(src->i_size, bs) - off;
c5c9cd4d
SW		 2486
2487 /* verify the end result is block aligned */
2a6b8dae
LZ		 2488 if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs) ||
2489 !IS_ALIGNED(destoff, bs))
c5c9cd4d
SW		 2490 goto out_unlock;
2491
d525e8ab
LZ		 2492 if (destoff > inode->i_size) {
2493 ret = btrfs_cont_expand(inode, inode->i_size, destoff);
2494 if (ret)
2495 goto out_unlock;
2496 }
2497
71ef0786
LZ		 2498 /* truncate page cache pages from target inode range */
2499 truncate_inode_pages_range(&inode->i_data, destoff,
2500 PAGE_CACHE_ALIGN(destoff + len) - 1);
2501
f46b5a66
CH		 2502 /* do any pending delalloc/csum calc on src, one way or
2503 another, and lock file content */
2504 while (1) {
31840ae1 2505 struct btrfs_ordered_extent *ordered;
aa42ffd9
LB		 2506 lock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1);
2507 ordered = btrfs_lookup_first_ordered_extent(src, off + len - 1);
9a019196 2508 if (!ordered &&
aa42ffd9
LB		 2509 !test_range_bit(&BTRFS_I(src)->io_tree, off, off + len - 1,
2510 EXTENT_DELALLOC, 0, NULL))
f46b5a66 2511 break;
aa42ffd9 2512 unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1);
ae01a0ab
YZ		 2513 if (ordered)
2514 btrfs_put_ordered_extent(ordered);
9a019196 2515 btrfs_wait_ordered_range(src, off, len);
f46b5a66
CH		 2516 }
2517
c5c9cd4d 2518 /* clone data */
33345d01 2519 key.objectid = btrfs_ino(src);
ae01a0ab
YZ		 2520 key.type = BTRFS_EXTENT_DATA_KEY;
2521 key.offset = 0;
f46b5a66
CH		 2522
2523 while (1) {
2524 /*
2525 * note the key will change type as we walk through the
2526 * tree.
2527 */
362a20c5
DS		 2528 ret = btrfs_search_slot(NULL, BTRFS_I(src)->root, &key, path,
2529 0, 0);
f46b5a66
CH		 2530 if (ret < 0)
2531 goto out;
2532
ae01a0ab
YZ		 2533 nritems = btrfs_header_nritems(path->nodes[0]);
2534 if (path->slots[0] >= nritems) {
362a20c5 2535 ret = btrfs_next_leaf(BTRFS_I(src)->root, path);
f46b5a66
CH		 2536 if (ret < 0)
2537 goto out;
2538 if (ret > 0)
2539 break;
ae01a0ab 2540 nritems = btrfs_header_nritems(path->nodes[0]);
f46b5a66
CH		 2541 }
2542 leaf = path->nodes[0];
2543 slot = path->slots[0];
f46b5a66 2544
ae01a0ab 2545 btrfs_item_key_to_cpu(leaf, &key, slot);
d20f7043 2546 if (btrfs_key_type(&key) > BTRFS_EXTENT_DATA_KEY ||
33345d01 2547 key.objectid != btrfs_ino(src))
f46b5a66
CH		 2548 break;
2549
c5c9cd4d
SW		 2550 if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) {
2551 struct btrfs_file_extent_item *extent;
2552 int type;
31840ae1
ZY		 2553 u32 size;
2554 struct btrfs_key new_key;
c5c9cd4d
SW		 2555 u64 disko = 0, diskl = 0;
2556 u64 datao = 0, datal = 0;
2557 u8 comp;
b5384d48 2558 u64 endoff;
31840ae1
ZY		 2559
2560 size = btrfs_item_size_nr(leaf, slot);
2561 read_extent_buffer(leaf, buf,
2562 btrfs_item_ptr_offset(leaf, slot),
2563 size);
c5c9cd4d
SW		 2564
2565 extent = btrfs_item_ptr(leaf, slot,
2566 struct btrfs_file_extent_item);
2567 comp = btrfs_file_extent_compression(leaf, extent);
2568 type = btrfs_file_extent_type(leaf, extent);
c8a894d7
CM		 2569 if (type == BTRFS_FILE_EXTENT_REG ||
2570 type == BTRFS_FILE_EXTENT_PREALLOC) {
d397712b
CM		 2571 disko = btrfs_file_extent_disk_bytenr(leaf,
2572 extent);
2573 diskl = btrfs_file_extent_disk_num_bytes(leaf,
2574 extent);
c5c9cd4d 2575 datao = btrfs_file_extent_offset(leaf, extent);
d397712b
CM		 2576 datal = btrfs_file_extent_num_bytes(leaf,
2577 extent);
c5c9cd4d
SW		 2578 } else if (type == BTRFS_FILE_EXTENT_INLINE) {
2579 /* take upper bound, may be compressed */
2580 datal = btrfs_file_extent_ram_bytes(leaf,
2581 extent);
2582 }
b3b4aa74 2583 btrfs_release_path(path);
31840ae1 2584
050006a7 2585 if (key.offset + datal <= off ||
aa42ffd9 2586 key.offset >= off + len - 1)
c5c9cd4d
SW		 2587 goto next;
2588
31840ae1 2589 memcpy(&new_key, &key, sizeof(new_key));
33345d01 2590 new_key.objectid = btrfs_ino(inode);
4d728ec7
LZ		 2591 if (off <= key.offset)
2592 new_key.offset = key.offset + destoff - off;
2593 else
2594 new_key.offset = destoff;
31840ae1 2595
b6f3409b
SW		 2596 /*
2597 * 1 - adjusting old extent (we may have to split it)
2598 * 1 - add new extent
2599 * 1 - inode update
2600 */
2601 trans = btrfs_start_transaction(root, 3);
a22285a6
YZ		 2602 if (IS_ERR(trans)) {
2603 ret = PTR_ERR(trans);
2604 goto out;
2605 }
2606
c8a894d7
CM		 2607 if (type == BTRFS_FILE_EXTENT_REG ||
2608 type == BTRFS_FILE_EXTENT_PREALLOC) {
d72c0842
LZ		 2609 /*
2610 * a | --- range to clone ---| b
2611 * | ------------- extent ------------- |
2612 */
2613
2614 /* substract range b */
2615 if (key.offset + datal > off + len)
2616 datal = off + len - key.offset;
2617
2618 /* substract range a */
a22285a6
YZ		 2619 if (off > key.offset) {
2620 datao += off - key.offset;
2621 datal -= off - key.offset;
2622 }
2623
5dc562c5 2624 ret = btrfs_drop_extents(trans, root, inode,
a22285a6
YZ		 2625 new_key.offset,
2626 new_key.offset + datal,
2671485d 2627 1);
79787eaa
JM		 2628 if (ret) {
2629 btrfs_abort_transaction(trans, root,
2630 ret);
2631 btrfs_end_transaction(trans, root);
2632 goto out;
2633 }
a22285a6 2634
c5c9cd4d
SW		 2635 ret = btrfs_insert_empty_item(trans, root, path,
2636 &new_key, size);
79787eaa
JM		 2637 if (ret) {
2638 btrfs_abort_transaction(trans, root,
2639 ret);
2640 btrfs_end_transaction(trans, root);
2641 goto out;
2642 }
c5c9cd4d
SW		 2643
2644 leaf = path->nodes[0];
2645 slot = path->slots[0];
2646 write_extent_buffer(leaf, buf,
31840ae1
ZY		 2647 btrfs_item_ptr_offset(leaf, slot),
2648 size);
ae01a0ab 2649
c5c9cd4d 2650 extent = btrfs_item_ptr(leaf, slot,
f46b5a66 2651 struct btrfs_file_extent_item);
c5c9cd4d 2652
c5c9cd4d
SW		 2653 /* disko == 0 means it's a hole */
2654 if (!disko)
2655 datao = 0;
c5c9cd4d
SW		 2656
2657 btrfs_set_file_extent_offset(leaf, extent,
2658 datao);
2659 btrfs_set_file_extent_num_bytes(leaf, extent,
2660 datal);
2661 if (disko) {
2662 inode_add_bytes(inode, datal);
ae01a0ab 2663 ret = btrfs_inc_extent_ref(trans, root,
5d4f98a2
YZ		 2664 disko, diskl, 0,
2665 root->root_key.objectid,
33345d01 2666 btrfs_ino(inode),
66d7e7f0
AJ		 2667 new_key.offset - datao,
2668 0);
79787eaa
JM		 2669 if (ret) {
2670 btrfs_abort_transaction(trans,
2671 root,
2672 ret);
2673 btrfs_end_transaction(trans,
2674 root);
2675 goto out;
2676
2677 }
f46b5a66 2678 }
c5c9cd4d
SW		 2679 } else if (type == BTRFS_FILE_EXTENT_INLINE) {
2680 u64 skip = 0;
2681 u64 trim = 0;
2682 if (off > key.offset) {
2683 skip = off - key.offset;
2684 new_key.offset += skip;
2685 }
d397712b 2686
aa42ffd9
LB		 2687 if (key.offset + datal > off + len)
2688 trim = key.offset + datal - (off + len);
d397712b 2689
c5c9cd4d 2690 if (comp && (skip || trim)) {
c5c9cd4d 2691 ret = -EINVAL;
a22285a6 2692 btrfs_end_transaction(trans, root);
c5c9cd4d
SW		 2693 goto out;
2694 }
2695 size -= skip + trim;
2696 datal -= skip + trim;
a22285a6 2697
5dc562c5 2698 ret = btrfs_drop_extents(trans, root, inode,
a22285a6
YZ		 2699 new_key.offset,
2700 new_key.offset + datal,
2671485d 2701 1);
79787eaa
JM		 2702 if (ret) {
2703 btrfs_abort_transaction(trans, root,
2704 ret);
2705 btrfs_end_transaction(trans, root);
2706 goto out;
2707 }
a22285a6 2708
c5c9cd4d
SW		 2709 ret = btrfs_insert_empty_item(trans, root, path,
2710 &new_key, size);
79787eaa
JM		 2711 if (ret) {
2712 btrfs_abort_transaction(trans, root,
2713 ret);
2714 btrfs_end_transaction(trans, root);
2715 goto out;
2716 }
c5c9cd4d
SW		 2717
2718 if (skip) {
d397712b
CM		 2719 u32 start =
2720 btrfs_file_extent_calc_inline_size(0);
c5c9cd4d
SW		 2721 memmove(buf+start, buf+start+skip,
2722 datal);
2723 }
2724
2725 leaf = path->nodes[0];
2726 slot = path->slots[0];
2727 write_extent_buffer(leaf, buf,
2728 btrfs_item_ptr_offset(leaf, slot),
2729 size);
2730 inode_add_bytes(inode, datal);
f46b5a66 2731 }
c5c9cd4d
SW		 2732
2733 btrfs_mark_buffer_dirty(leaf);
b3b4aa74 2734 btrfs_release_path(path);
c5c9cd4d 2735
0c4d2d95 2736 inode_inc_iversion(inode);
a22285a6 2737 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
b5384d48
SW		 2738
2739 /*
2740 * we round up to the block size at eof when
2741 * determining which extents to clone above,
2742 * but shouldn't round up the file size
2743 */
2744 endoff = new_key.offset + datal;
5f3888ff
LZ		 2745 if (endoff > destoff+olen)
2746 endoff = destoff+olen;
b5384d48
SW		 2747 if (endoff > inode->i_size)
2748 btrfs_i_size_write(inode, endoff);
2749
a22285a6 2750 ret = btrfs_update_inode(trans, root, inode);
79787eaa
JM		 2751 if (ret) {
2752 btrfs_abort_transaction(trans, root, ret);
2753 btrfs_end_transaction(trans, root);
2754 goto out;
2755 }
2756 ret = btrfs_end_transaction(trans, root);
a22285a6 2757 }
d397712b 2758next:
b3b4aa74 2759 btrfs_release_path(path);
f46b5a66 2760 key.offset++;
f46b5a66 2761 }
f46b5a66
CH		 2762 ret = 0;
2763out:
b3b4aa74 2764 btrfs_release_path(path);
aa42ffd9 2765 unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1);
f46b5a66
CH		 2766out_unlock:
2767 mutex_unlock(&src->i_mutex);
2768 mutex_unlock(&inode->i_mutex);
ae01a0ab
YZ		 2769 vfree(buf);
2770 btrfs_free_path(path);
f46b5a66 2771out_fput:
2903ff01 2772 fdput(src_file);
ab67b7c1 2773out_drop_write:
2a79f17e 2774 mnt_drop_write_file(file);
f46b5a66
CH		 2775 return ret;
2776}
2777
7a865e8a 2778static long btrfs_ioctl_clone_range(struct file *file, void __user *argp)
c5c9cd4d
SW		 2779{
2780 struct btrfs_ioctl_clone_range_args args;
2781
7a865e8a 2782 if (copy_from_user(&args, argp, sizeof(args)))
c5c9cd4d
SW		 2783 return -EFAULT;
2784 return btrfs_ioctl_clone(file, args.src_fd, args.src_offset,
2785 args.src_length, args.dest_offset);
2786}
2787
f46b5a66
CH		 2788/*
2789 * there are many ways the trans_start and trans_end ioctls can lead
2790 * to deadlocks. They should only be used by applications that
2791 * basically own the machine, and have a very in depth understanding
2792 * of all the possible deadlocks and enospc problems.
2793 */
b2950863 2794static long btrfs_ioctl_trans_start(struct file *file)
f46b5a66
CH		 2795{
2796 struct inode *inode = fdentry(file)->d_inode;
2797 struct btrfs_root *root = BTRFS_I(inode)->root;
2798 struct btrfs_trans_handle *trans;
1ab86aed 2799 int ret;
f46b5a66 2800
1ab86aed 2801 ret = -EPERM;
df5b5520 2802 if (!capable(CAP_SYS_ADMIN))
1ab86aed 2803 goto out;
df5b5520 2804
1ab86aed
SW		 2805 ret = -EINPROGRESS;
2806 if (file->private_data)
f46b5a66 2807 goto out;
9ca9ee09 2808
b83cc969
LZ		 2809 ret = -EROFS;
2810 if (btrfs_root_readonly(root))
2811 goto out;
2812
a561be71 2813 ret = mnt_want_write_file(file);
c146afad
YZ		 2814 if (ret)
2815 goto out;
2816
a4abeea4 2817 atomic_inc(&root->fs_info->open_ioctl_trans);
9ca9ee09 2818
1ab86aed 2819 ret = -ENOMEM;
7a7eaa40 2820 trans = btrfs_start_ioctl_transaction(root);
abd30bb0 2821 if (IS_ERR(trans))
1ab86aed
SW		 2822 goto out_drop;
2823
2824 file->private_data = trans;
2825 return 0;
2826
2827out_drop:
a4abeea4 2828 atomic_dec(&root->fs_info->open_ioctl_trans);
2a79f17e 2829 mnt_drop_write_file(file);
f46b5a66 2830out:
f46b5a66
CH		 2831 return ret;
2832}
2833
6ef5ed0d
JB		 2834static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp)
2835{
2836 struct inode *inode = fdentry(file)->d_inode;
2837 struct btrfs_root *root = BTRFS_I(inode)->root;
2838 struct btrfs_root *new_root;
2839 struct btrfs_dir_item *di;
2840 struct btrfs_trans_handle *trans;
2841 struct btrfs_path *path;
2842 struct btrfs_key location;
2843 struct btrfs_disk_key disk_key;
6ef5ed0d
JB		 2844 u64 objectid = 0;
2845 u64 dir_id;
2846
2847 if (!capable(CAP_SYS_ADMIN))
2848 return -EPERM;
2849
2850 if (copy_from_user(&objectid, argp, sizeof(objectid)))
2851 return -EFAULT;
2852
2853 if (!objectid)
2854 objectid = root->root_key.objectid;
2855
2856 location.objectid = objectid;
2857 location.type = BTRFS_ROOT_ITEM_KEY;
2858 location.offset = (u64)-1;
2859
2860 new_root = btrfs_read_fs_root_no_name(root->fs_info, &location);
2861 if (IS_ERR(new_root))
2862 return PTR_ERR(new_root);
2863
2864 if (btrfs_root_refs(&new_root->root_item) == 0)
2865 return -ENOENT;
2866
2867 path = btrfs_alloc_path();
2868 if (!path)
2869 return -ENOMEM;
2870 path->leave_spinning = 1;
2871
2872 trans = btrfs_start_transaction(root, 1);
98d5dc13 2873 if (IS_ERR(trans)) {
6ef5ed0d 2874 btrfs_free_path(path);
98d5dc13 2875 return PTR_ERR(trans);
6ef5ed0d
JB		 2876 }
2877
6c41761f 2878 dir_id = btrfs_super_root_dir(root->fs_info->super_copy);
6ef5ed0d
JB		 2879 di = btrfs_lookup_dir_item(trans, root->fs_info->tree_root, path,
2880 dir_id, "default", 7, 1);
cf1e99a4 2881 if (IS_ERR_OR_NULL(di)) {
6ef5ed0d
JB		 2882 btrfs_free_path(path);
2883 btrfs_end_transaction(trans, root);
2884 printk(KERN_ERR "Umm, you don't have the default dir item, "
2885 "this isn't going to work\n");
2886 return -ENOENT;
2887 }
2888
2889 btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key);
2890 btrfs_set_dir_item_key(path->nodes[0], di, &disk_key);
2891 btrfs_mark_buffer_dirty(path->nodes[0]);
2892 btrfs_free_path(path);
2893
2b0ce2c2 2894 btrfs_set_fs_incompat(root->fs_info, DEFAULT_SUBVOL);
6ef5ed0d
JB		 2895 btrfs_end_transaction(trans, root);
2896
2897 return 0;
2898}
2899
5af3e8cc
SB		 2900void btrfs_get_block_group_info(struct list_head *groups_list,
2901 struct btrfs_ioctl_space_info *space)
bf5fc093
JB		 2902{
2903 struct btrfs_block_group_cache *block_group;
2904
2905 space->total_bytes = 0;
2906 space->used_bytes = 0;
2907 space->flags = 0;
2908 list_for_each_entry(block_group, groups_list, list) {
2909 space->flags = block_group->flags;
2910 space->total_bytes += block_group->key.offset;
2911 space->used_bytes +=
2912 btrfs_block_group_used(&block_group->item);
2913 }
2914}
2915
1406e432
JB		 2916long btrfs_ioctl_space_info(struct btrfs_root *root, void __user *arg)
2917{
2918 struct btrfs_ioctl_space_args space_args;
2919 struct btrfs_ioctl_space_info space;
2920 struct btrfs_ioctl_space_info *dest;
7fde62bf 2921 struct btrfs_ioctl_space_info *dest_orig;
13f2696f 2922 struct btrfs_ioctl_space_info __user *user_dest;
1406e432 2923 struct btrfs_space_info *info;
bf5fc093
JB		 2924 u64 types[] = {BTRFS_BLOCK_GROUP_DATA,
2925 BTRFS_BLOCK_GROUP_SYSTEM,
2926 BTRFS_BLOCK_GROUP_METADATA,
2927 BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA};
2928 int num_types = 4;
7fde62bf 2929 int alloc_size;
1406e432 2930 int ret = 0;
51788b1b 2931 u64 slot_count = 0;
bf5fc093 2932 int i, c;
1406e432
JB		 2933
2934 if (copy_from_user(&space_args,
2935 (struct btrfs_ioctl_space_args __user *)arg,
2936 sizeof(space_args)))
2937 return -EFAULT;
2938
bf5fc093
JB		 2939 for (i = 0; i < num_types; i++) {
2940 struct btrfs_space_info *tmp;
2941
2942 info = NULL;
2943 rcu_read_lock();
2944 list_for_each_entry_rcu(tmp, &root->fs_info->space_info,
2945 list) {
2946 if (tmp->flags == types[i]) {
2947 info = tmp;
2948 break;
2949 }
2950 }
2951 rcu_read_unlock();
2952
2953 if (!info)
2954 continue;
2955
2956 down_read(&info->groups_sem);
2957 for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
2958 if (!list_empty(&info->block_groups[c]))
2959 slot_count++;
2960 }
2961 up_read(&info->groups_sem);
2962 }
7fde62bf
CM		 2963
2964 /* space_slots == 0 means they are asking for a count */
2965 if (space_args.space_slots == 0) {
2966 space_args.total_spaces = slot_count;
2967 goto out;
2968 }
bf5fc093 2969
51788b1b 2970 slot_count = min_t(u64, space_args.space_slots, slot_count);
bf5fc093 2971
7fde62bf 2972 alloc_size = sizeof(*dest) * slot_count;
bf5fc093 2973
7fde62bf
CM		 2974 /* we generally have at most 6 or so space infos, one for each raid
2975 * level. So, a whole page should be more than enough for everyone
2976 */
2977 if (alloc_size > PAGE_CACHE_SIZE)
2978 return -ENOMEM;
2979
1406e432 2980 space_args.total_spaces = 0;
7fde62bf
CM		 2981 dest = kmalloc(alloc_size, GFP_NOFS);
2982 if (!dest)
2983 return -ENOMEM;
2984 dest_orig = dest;
1406e432 2985
7fde62bf 2986 /* now we have a buffer to copy into */
bf5fc093
JB		 2987 for (i = 0; i < num_types; i++) {
2988 struct btrfs_space_info *tmp;
2989
51788b1b
DR		 2990 if (!slot_count)
2991 break;
2992
bf5fc093
JB		 2993 info = NULL;
2994 rcu_read_lock();
2995 list_for_each_entry_rcu(tmp, &root->fs_info->space_info,
2996 list) {
2997 if (tmp->flags == types[i]) {
2998 info = tmp;
2999 break;
3000 }
3001 }
3002 rcu_read_unlock();
7fde62bf 3003
bf5fc093
JB		 3004 if (!info)
3005 continue;
3006 down_read(&info->groups_sem);
3007 for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
3008 if (!list_empty(&info->block_groups[c])) {
5af3e8cc
SB		 3009 btrfs_get_block_group_info(
3010 &info->block_groups[c], &space);
bf5fc093
JB		 3011 memcpy(dest, &space, sizeof(space));
3012 dest++;
3013 space_args.total_spaces++;
51788b1b 3014 slot_count--;
bf5fc093 3015 }
51788b1b
DR		 3016 if (!slot_count)
3017 break;
bf5fc093
JB		 3018 }
3019 up_read(&info->groups_sem);
1406e432 3020 }
1406e432 3021
2eec6c81 3022 user_dest = (struct btrfs_ioctl_space_info __user *)
7fde62bf
CM		 3023 (arg + sizeof(struct btrfs_ioctl_space_args));
3024
3025 if (copy_to_user(user_dest, dest_orig, alloc_size))
3026 ret = -EFAULT;
3027
3028 kfree(dest_orig);
3029out:
3030 if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args)))
1406e432
JB		 3031 ret = -EFAULT;
3032
3033 return ret;
3034}
3035
f46b5a66
CH		 3036/*
3037 * there are many ways the trans_start and trans_end ioctls can lead
3038 * to deadlocks. They should only be used by applications that
3039 * basically own the machine, and have a very in depth understanding
3040 * of all the possible deadlocks and enospc problems.
3041 */
3042long btrfs_ioctl_trans_end(struct file *file)
3043{
3044 struct inode *inode = fdentry(file)->d_inode;
3045 struct btrfs_root *root = BTRFS_I(inode)->root;
3046 struct btrfs_trans_handle *trans;
f46b5a66 3047
f46b5a66 3048 trans = file->private_data;
1ab86aed
SW		 3049 if (!trans)
3050 return -EINVAL;
b214107e 3051 file->private_data = NULL;
9ca9ee09 3052
1ab86aed
SW		 3053 btrfs_end_transaction(trans, root);
3054
a4abeea4 3055 atomic_dec(&root->fs_info->open_ioctl_trans);
9ca9ee09 3056
2a79f17e 3057 mnt_drop_write_file(file);
1ab86aed 3058 return 0;
f46b5a66
CH		 3059}
3060
9a8c28be
MX		 3061static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root,
3062 void __user *argp)
46204592 3063{
46204592
SW		 3064 struct btrfs_trans_handle *trans;
3065 u64 transid;
db5b493a 3066 int ret;
46204592 3067
ff7c1d33
MX		 3068 trans = btrfs_attach_transaction(root);
3069 if (IS_ERR(trans)) {
3070 if (PTR_ERR(trans) != -ENOENT)
3071 return PTR_ERR(trans);
3072
3073 /* No running transaction, don't bother */
3074 transid = root->fs_info->last_trans_committed;
3075 goto out;
3076 }
46204592 3077 transid = trans->transid;
db5b493a 3078 ret = btrfs_commit_transaction_async(trans, root, 0);
8b2b2d3c
TI		 3079 if (ret) {
3080 btrfs_end_transaction(trans, root);
db5b493a 3081 return ret;
8b2b2d3c 3082 }
ff7c1d33 3083out:
46204592
SW		 3084 if (argp)
3085 if (copy_to_user(argp, &transid, sizeof(transid)))
3086 return -EFAULT;
3087 return 0;
3088}
3089
9a8c28be
MX		 3090static noinline long btrfs_ioctl_wait_sync(struct btrfs_root *root,
3091 void __user *argp)
46204592 3092{
46204592
SW		 3093 u64 transid;
3094
3095 if (argp) {
3096 if (copy_from_user(&transid, argp, sizeof(transid)))
3097 return -EFAULT;
3098 } else {
3099 transid = 0; /* current trans */
3100 }
3101 return btrfs_wait_for_commit(root, transid);
3102}
3103
475f6387
JS		 3104static long btrfs_ioctl_scrub(struct btrfs_root *root, void __user *arg)
3105{
3106 int ret;
3107 struct btrfs_ioctl_scrub_args *sa;
3108
3109 if (!capable(CAP_SYS_ADMIN))
3110 return -EPERM;
3111
3112 sa = memdup_user(arg, sizeof(*sa));
3113 if (IS_ERR(sa))
3114 return PTR_ERR(sa);
3115
aa1b8cd4 3116 ret = btrfs_scrub_dev(root->fs_info, sa->devid, sa->start, sa->end,
63a212ab
SB		 3117 &sa->progress, sa->flags & BTRFS_SCRUB_READONLY,
3118 0);
475f6387
JS		 3119
3120 if (copy_to_user(arg, sa, sizeof(*sa)))
3121 ret = -EFAULT;
3122
3123 kfree(sa);
3124 return ret;
3125}
3126
3127static long btrfs_ioctl_scrub_cancel(struct btrfs_root *root, void __user *arg)
3128{
3129 if (!capable(CAP_SYS_ADMIN))
3130 return -EPERM;
3131
aa1b8cd4 3132 return btrfs_scrub_cancel(root->fs_info);
475f6387
JS		 3133}
3134
3135static long btrfs_ioctl_scrub_progress(struct btrfs_root *root,
3136 void __user *arg)
3137{
3138 struct btrfs_ioctl_scrub_args *sa;
3139 int ret;
3140
3141 if (!capable(CAP_SYS_ADMIN))
3142 return -EPERM;
3143
3144 sa = memdup_user(arg, sizeof(*sa));
3145 if (IS_ERR(sa))
3146 return PTR_ERR(sa);
3147
3148 ret = btrfs_scrub_progress(root, sa->devid, &sa->progress);
3149
3150 if (copy_to_user(arg, sa, sizeof(*sa)))
3151 ret = -EFAULT;
3152
3153 kfree(sa);
3154 return ret;
3155}
3156
c11d2c23 3157static long btrfs_ioctl_get_dev_stats(struct btrfs_root *root,
b27f7c0c 3158 void __user *arg)
c11d2c23
SB		 3159{
3160 struct btrfs_ioctl_get_dev_stats *sa;
3161 int ret;
3162
c11d2c23
SB		 3163 sa = memdup_user(arg, sizeof(*sa));
3164 if (IS_ERR(sa))
3165 return PTR_ERR(sa);
3166
b27f7c0c
DS		 3167 if ((sa->flags & BTRFS_DEV_STATS_RESET) && !capable(CAP_SYS_ADMIN)) {
3168 kfree(sa);
3169 return -EPERM;
3170 }
3171
3172 ret = btrfs_get_dev_stats(root, sa);
c11d2c23
SB		 3173
3174 if (copy_to_user(arg, sa, sizeof(*sa)))
3175 ret = -EFAULT;
3176
3177 kfree(sa);
3178 return ret;
3179}
3180
3f6bcfbd
SB		 3181static long btrfs_ioctl_dev_replace(struct btrfs_root *root, void __user *arg)
3182{
3183 struct btrfs_ioctl_dev_replace_args *p;
3184 int ret;
3185
3186 if (!capable(CAP_SYS_ADMIN))
3187 return -EPERM;
3188
3189 p = memdup_user(arg, sizeof(*p));
3190 if (IS_ERR(p))
3191 return PTR_ERR(p);
3192
3193 switch (p->cmd) {
3194 case BTRFS_IOCTL_DEV_REPLACE_CMD_START:
3195 if (atomic_xchg(
3196 &root->fs_info->mutually_exclusive_operation_running,
3197 1)) {
3198 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
3199 ret = -EINPROGRESS;
3200 } else {
3201 ret = btrfs_dev_replace_start(root, p);
3202 atomic_set(
3203 &root->fs_info->mutually_exclusive_operation_running,
3204 0);
3205 }
3206 break;
3207 case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS:
3208 btrfs_dev_replace_status(root->fs_info, p);
3209 ret = 0;
3210 break;
3211 case BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL:
3212 ret = btrfs_dev_replace_cancel(root->fs_info, p);
3213 break;
3214 default:
3215 ret = -EINVAL;
3216 break;
3217 }
3218
3219 if (copy_to_user(arg, p, sizeof(*p)))
3220 ret = -EFAULT;
3221
3222 kfree(p);
3223 return ret;
3224}
3225
d7728c96
JS		 3226static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg)
3227{
3228 int ret = 0;
3229 int i;
740c3d22 3230 u64 rel_ptr;
d7728c96 3231 int size;
806468f8 3232 struct btrfs_ioctl_ino_path_args *ipa = NULL;
d7728c96
JS		 3233 struct inode_fs_paths *ipath = NULL;
3234 struct btrfs_path *path;
3235
3236 if (!capable(CAP_SYS_ADMIN))
3237 return -EPERM;
3238
3239 path = btrfs_alloc_path();
3240 if (!path) {
3241 ret = -ENOMEM;
3242 goto out;
3243 }
3244
3245 ipa = memdup_user(arg, sizeof(*ipa));
3246 if (IS_ERR(ipa)) {
3247 ret = PTR_ERR(ipa);
3248 ipa = NULL;
3249 goto out;
3250 }
3251
3252 size = min_t(u32, ipa->size, 4096);
3253 ipath = init_ipath(size, root, path);
3254 if (IS_ERR(ipath)) {
3255 ret = PTR_ERR(ipath);
3256 ipath = NULL;
3257 goto out;
3258 }
3259
3260 ret = paths_from_inode(ipa->inum, ipath);
3261 if (ret < 0)
3262 goto out;
3263
3264 for (i = 0; i < ipath->fspath->elem_cnt; ++i) {
745c4d8e
JM		 3265 rel_ptr = ipath->fspath->val[i] -
3266 (u64)(unsigned long)ipath->fspath->val;
740c3d22 3267 ipath->fspath->val[i] = rel_ptr;
d7728c96
JS		 3268 }
3269
745c4d8e
JM		 3270 ret = copy_to_user((void *)(unsigned long)ipa->fspath,
3271 (void *)(unsigned long)ipath->fspath, size);
d7728c96
JS		 3272 if (ret) {
3273 ret = -EFAULT;
3274 goto out;
3275 }
3276
3277out:
3278 btrfs_free_path(path);
3279 free_ipath(ipath);
3280 kfree(ipa);
3281
3282 return ret;
3283}
3284
3285static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx)
3286{
3287 struct btrfs_data_container *inodes = ctx;
3288 const size_t c = 3 * sizeof(u64);
3289
3290 if (inodes->bytes_left >= c) {
3291 inodes->bytes_left -= c;
3292 inodes->val[inodes->elem_cnt] = inum;
3293 inodes->val[inodes->elem_cnt + 1] = offset;
3294 inodes->val[inodes->elem_cnt + 2] = root;
3295 inodes->elem_cnt += 3;
3296 } else {
3297 inodes->bytes_missing += c - inodes->bytes_left;
3298 inodes->bytes_left = 0;
3299 inodes->elem_missed += 3;
3300 }
3301
3302 return 0;
3303}
3304
3305static long btrfs_ioctl_logical_to_ino(struct btrfs_root *root,
3306 void __user *arg)
3307{
3308 int ret = 0;
3309 int size;
d7728c96
JS		 3310 struct btrfs_ioctl_logical_ino_args *loi;
3311 struct btrfs_data_container *inodes = NULL;
3312 struct btrfs_path *path = NULL;
d7728c96
JS		 3313
3314 if (!capable(CAP_SYS_ADMIN))
3315 return -EPERM;
3316
3317 loi = memdup_user(arg, sizeof(*loi));
3318 if (IS_ERR(loi)) {
3319 ret = PTR_ERR(loi);
3320 loi = NULL;
3321 goto out;
3322 }
3323
3324 path = btrfs_alloc_path();
3325 if (!path) {
3326 ret = -ENOMEM;
3327 goto out;
3328 }
3329
425d17a2 3330 size = min_t(u32, loi->size, 64 * 1024);
d7728c96
JS		 3331 inodes = init_data_container(size);
3332 if (IS_ERR(inodes)) {
3333 ret = PTR_ERR(inodes);
3334 inodes = NULL;
3335 goto out;
3336 }
3337
df031f07
LB		 3338 ret = iterate_inodes_from_logical(loi->logical, root->fs_info, path,
3339 build_ino_list, inodes);
3340 if (ret == -EINVAL)
d7728c96
JS		 3341 ret = -ENOENT;
3342 if (ret < 0)
3343 goto out;
3344
745c4d8e
JM		 3345 ret = copy_to_user((void *)(unsigned long)loi->inodes,
3346 (void *)(unsigned long)inodes, size);
d7728c96
JS		 3347 if (ret)
3348 ret = -EFAULT;
3349
3350out:
3351 btrfs_free_path(path);
425d17a2 3352 vfree(inodes);
d7728c96
JS		 3353 kfree(loi);
3354
3355 return ret;
3356}
3357
19a39dce 3358void update_ioctl_balance_args(struct btrfs_fs_info *fs_info, int lock,
c9e9f97b
ID		 3359 struct btrfs_ioctl_balance_args *bargs)
3360{
3361 struct btrfs_balance_control *bctl = fs_info->balance_ctl;
3362
3363 bargs->flags = bctl->flags;
3364
837d5b6e
ID		 3365 if (atomic_read(&fs_info->balance_running))
3366 bargs->state |= BTRFS_BALANCE_STATE_RUNNING;
3367 if (atomic_read(&fs_info->balance_pause_req))
3368 bargs->state |= BTRFS_BALANCE_STATE_PAUSE_REQ;
a7e99c69
ID		 3369 if (atomic_read(&fs_info->balance_cancel_req))
3370 bargs->state |= BTRFS_BALANCE_STATE_CANCEL_REQ;
837d5b6e 3371
c9e9f97b
ID		 3372 memcpy(&bargs->data, &bctl->data, sizeof(bargs->data));
3373 memcpy(&bargs->meta, &bctl->meta, sizeof(bargs->meta));
3374 memcpy(&bargs->sys, &bctl->sys, sizeof(bargs->sys));
19a39dce
ID		 3375
3376 if (lock) {
3377 spin_lock(&fs_info->balance_lock);
3378 memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat));
3379 spin_unlock(&fs_info->balance_lock);
3380 } else {
3381 memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat));
3382 }
c9e9f97b
ID		 3383}
3384
9ba1f6e4 3385static long btrfs_ioctl_balance(struct file *file, void __user *arg)
c9e9f97b 3386{
9ba1f6e4 3387 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
c9e9f97b
ID		 3388 struct btrfs_fs_info *fs_info = root->fs_info;
3389 struct btrfs_ioctl_balance_args *bargs;
3390 struct btrfs_balance_control *bctl;
3391 int ret;
5ac00add 3392 int need_to_clear_lock = 0;
c9e9f97b
ID		 3393
3394 if (!capable(CAP_SYS_ADMIN))
3395 return -EPERM;
3396
e54bfa31 3397 ret = mnt_want_write_file(file);
9ba1f6e4
LB		 3398 if (ret)
3399 return ret;
3400
c9e9f97b
ID		 3401 mutex_lock(&fs_info->volume_mutex);
3402 mutex_lock(&fs_info->balance_mutex);
3403
3404 if (arg) {
3405 bargs = memdup_user(arg, sizeof(*bargs));
3406 if (IS_ERR(bargs)) {
3407 ret = PTR_ERR(bargs);
3408 goto out;
3409 }
de322263
ID		 3410
3411 if (bargs->flags & BTRFS_BALANCE_RESUME) {
3412 if (!fs_info->balance_ctl) {
3413 ret = -ENOTCONN;
3414 goto out_bargs;
3415 }
3416
3417 bctl = fs_info->balance_ctl;
3418 spin_lock(&fs_info->balance_lock);
3419 bctl->flags |= BTRFS_BALANCE_RESUME;
3420 spin_unlock(&fs_info->balance_lock);
3421
3422 goto do_balance;
3423 }
c9e9f97b
ID		 3424 } else {
3425 bargs = NULL;
3426 }
3427
5ac00add
SB		 3428 if (atomic_xchg(&root->fs_info->mutually_exclusive_operation_running,
3429 1)) {
3430 pr_info("btrfs: dev add/delete/balance/replace/resize operation in progress\n");
837d5b6e
ID		 3431 ret = -EINPROGRESS;
3432 goto out_bargs;
3433 }
5ac00add 3434 need_to_clear_lock = 1;
837d5b6e 3435
c9e9f97b
ID		 3436 bctl = kzalloc(sizeof(*bctl), GFP_NOFS);
3437 if (!bctl) {
3438 ret = -ENOMEM;
3439 goto out_bargs;
3440 }
3441
3442 bctl->fs_info = fs_info;
3443 if (arg) {
3444 memcpy(&bctl->data, &bargs->data, sizeof(bctl->data));
3445 memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta));
3446 memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys));
3447
3448 bctl->flags = bargs->flags;
f43ffb60
ID		 3449 } else {
3450 /* balance everything - no filters */
3451 bctl->flags |= BTRFS_BALANCE_TYPE_MASK;
c9e9f97b
ID		 3452 }
3453
de322263 3454do_balance:
c9e9f97b
ID		 3455 ret = btrfs_balance(bctl, bargs);
3456 /*
837d5b6e
ID		 3457 * bctl is freed in __cancel_balance or in free_fs_info if
3458 * restriper was paused all the way until unmount
c9e9f97b
ID		 3459 */
3460 if (arg) {
3461 if (copy_to_user(arg, bargs, sizeof(*bargs)))
3462 ret = -EFAULT;
3463 }
3464
3465out_bargs:
3466 kfree(bargs);
3467out:
5ac00add
SB		 3468 if (need_to_clear_lock)
3469 atomic_set(&root->fs_info->mutually_exclusive_operation_running,
3470 0);
c9e9f97b
ID		 3471 mutex_unlock(&fs_info->balance_mutex);
3472 mutex_unlock(&fs_info->volume_mutex);
e54bfa31 3473 mnt_drop_write_file(file);
c9e9f97b
ID		 3474 return ret;
3475}
3476
837d5b6e
ID		 3477static long btrfs_ioctl_balance_ctl(struct btrfs_root *root, int cmd)
3478{
3479 if (!capable(CAP_SYS_ADMIN))
3480 return -EPERM;
3481
3482 switch (cmd) {
3483 case BTRFS_BALANCE_CTL_PAUSE:
3484 return btrfs_pause_balance(root->fs_info);
a7e99c69
ID		 3485 case BTRFS_BALANCE_CTL_CANCEL:
3486 return btrfs_cancel_balance(root->fs_info);
837d5b6e
ID		 3487 }
3488
3489 return -EINVAL;
3490}
3491
19a39dce
ID		 3492static long btrfs_ioctl_balance_progress(struct btrfs_root *root,
3493 void __user *arg)
3494{
3495 struct btrfs_fs_info *fs_info = root->fs_info;
3496 struct btrfs_ioctl_balance_args *bargs;
3497 int ret = 0;
3498
3499 if (!capable(CAP_SYS_ADMIN))
3500 return -EPERM;
3501
3502 mutex_lock(&fs_info->balance_mutex);
3503 if (!fs_info->balance_ctl) {
3504 ret = -ENOTCONN;
3505 goto out;
3506 }
3507
3508 bargs = kzalloc(sizeof(*bargs), GFP_NOFS);
3509 if (!bargs) {
3510 ret = -ENOMEM;
3511 goto out;
3512 }
3513
3514 update_ioctl_balance_args(fs_info, 1, bargs);
3515
3516 if (copy_to_user(arg, bargs, sizeof(*bargs)))
3517 ret = -EFAULT;
3518
3519 kfree(bargs);
3520out:
3521 mutex_unlock(&fs_info->balance_mutex);
3522 return ret;
3523}
3524
5d13a37b
AJ		 3525static long btrfs_ioctl_quota_ctl(struct btrfs_root *root, void __user *arg)
3526{
3527 struct btrfs_ioctl_quota_ctl_args *sa;
3528 struct btrfs_trans_handle *trans = NULL;
3529 int ret;
3530 int err;
3531
3532 if (!capable(CAP_SYS_ADMIN))
3533 return -EPERM;
3534
3535 if (root->fs_info->sb->s_flags & MS_RDONLY)
3536 return -EROFS;
3537
3538 sa = memdup_user(arg, sizeof(*sa));
3539 if (IS_ERR(sa))
3540 return PTR_ERR(sa);
3541
3542 if (sa->cmd != BTRFS_QUOTA_CTL_RESCAN) {
3543 trans = btrfs_start_transaction(root, 2);
3544 if (IS_ERR(trans)) {
3545 ret = PTR_ERR(trans);
3546 goto out;
3547 }
3548 }
3549
3550 switch (sa->cmd) {
3551 case BTRFS_QUOTA_CTL_ENABLE:
3552 ret = btrfs_quota_enable(trans, root->fs_info);
3553 break;
3554 case BTRFS_QUOTA_CTL_DISABLE:
3555 ret = btrfs_quota_disable(trans, root->fs_info);
3556 break;
3557 case BTRFS_QUOTA_CTL_RESCAN:
3558 ret = btrfs_quota_rescan(root->fs_info);
3559 break;
3560 default:
3561 ret = -EINVAL;
3562 break;
3563 }
3564
3565 if (copy_to_user(arg, sa, sizeof(*sa)))
3566 ret = -EFAULT;
3567
3568 if (trans) {
3569 err = btrfs_commit_transaction(trans, root);
3570 if (err && !ret)
3571 ret = err;
3572 }
3573
3574out:
3575 kfree(sa);
3576 return ret;
3577}
3578
3579static long btrfs_ioctl_qgroup_assign(struct btrfs_root *root, void __user *arg)
3580{
3581 struct btrfs_ioctl_qgroup_assign_args *sa;
3582 struct btrfs_trans_handle *trans;
3583 int ret;
3584 int err;
3585
3586 if (!capable(CAP_SYS_ADMIN))
3587 return -EPERM;
3588
3589 if (root->fs_info->sb->s_flags & MS_RDONLY)
3590 return -EROFS;
3591
3592 sa = memdup_user(arg, sizeof(*sa));
3593 if (IS_ERR(sa))
3594 return PTR_ERR(sa);
3595
3596 trans = btrfs_join_transaction(root);
3597 if (IS_ERR(trans)) {
3598 ret = PTR_ERR(trans);
3599 goto out;
3600 }
3601
3602 /* FIXME: check if the IDs really exist */
3603 if (sa->assign) {
3604 ret = btrfs_add_qgroup_relation(trans, root->fs_info,
3605 sa->src, sa->dst);
3606 } else {
3607 ret = btrfs_del_qgroup_relation(trans, root->fs_info,
3608 sa->src, sa->dst);
3609 }
3610
3611 err = btrfs_end_transaction(trans, root);
3612 if (err && !ret)
3613 ret = err;
3614
3615out:
3616 kfree(sa);
3617 return ret;
3618}
3619
3620static long btrfs_ioctl_qgroup_create(struct btrfs_root *root, void __user *arg)
3621{
3622 struct btrfs_ioctl_qgroup_create_args *sa;
3623 struct btrfs_trans_handle *trans;
3624 int ret;
3625 int err;
3626
3627 if (!capable(CAP_SYS_ADMIN))
3628 return -EPERM;
3629
3630 if (root->fs_info->sb->s_flags & MS_RDONLY)
3631 return -EROFS;
3632
3633 sa = memdup_user(arg, sizeof(*sa));
3634 if (IS_ERR(sa))
3635 return PTR_ERR(sa);
3636
3637 trans = btrfs_join_transaction(root);
3638 if (IS_ERR(trans)) {
3639 ret = PTR_ERR(trans);
3640 goto out;
3641 }
3642
3643 /* FIXME: check if the IDs really exist */
3644 if (sa->create) {
3645 ret = btrfs_create_qgroup(trans, root->fs_info, sa->qgroupid,
3646 NULL);
3647 } else {
3648 ret = btrfs_remove_qgroup(trans, root->fs_info, sa->qgroupid);
3649 }
3650
3651 err = btrfs_end_transaction(trans, root);
3652 if (err && !ret)
3653 ret = err;
3654
3655out:
3656 kfree(sa);
3657 return ret;
3658}
3659
3660static long btrfs_ioctl_qgroup_limit(struct btrfs_root *root, void __user *arg)
3661{
3662 struct btrfs_ioctl_qgroup_limit_args *sa;
3663 struct btrfs_trans_handle *trans;
3664 int ret;
3665 int err;
3666 u64 qgroupid;
3667
3668 if (!capable(CAP_SYS_ADMIN))
3669 return -EPERM;
3670
3671 if (root->fs_info->sb->s_flags & MS_RDONLY)
3672 return -EROFS;
3673
3674 sa = memdup_user(arg, sizeof(*sa));
3675 if (IS_ERR(sa))
3676 return PTR_ERR(sa);
3677
3678 trans = btrfs_join_transaction(root);
3679 if (IS_ERR(trans)) {
3680 ret = PTR_ERR(trans);
3681 goto out;
3682 }
3683
3684 qgroupid = sa->qgroupid;
3685 if (!qgroupid) {
3686 /* take the current subvol as qgroup */
3687 qgroupid = root->root_key.objectid;
3688 }
3689
3690 /* FIXME: check if the IDs really exist */
3691 ret = btrfs_limit_qgroup(trans, root->fs_info, qgroupid, &sa->lim);
3692
3693 err = btrfs_end_transaction(trans, root);
3694 if (err && !ret)
3695 ret = err;
3696
3697out:
3698 kfree(sa);
3699 return ret;
3700}
3701
8ea05e3a
AB		 3702static long btrfs_ioctl_set_received_subvol(struct file *file,
3703 void __user *arg)
3704{
3705 struct btrfs_ioctl_received_subvol_args *sa = NULL;
3706 struct inode *inode = fdentry(file)->d_inode;
3707 struct btrfs_root *root = BTRFS_I(inode)->root;
3708 struct btrfs_root_item *root_item = &root->root_item;
3709 struct btrfs_trans_handle *trans;
3710 struct timespec ct = CURRENT_TIME;
3711 int ret = 0;
3712
3713 ret = mnt_want_write_file(file);
3714 if (ret < 0)
3715 return ret;
3716
3717 down_write(&root->fs_info->subvol_sem);
3718
3719 if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) {
3720 ret = -EINVAL;
3721 goto out;
3722 }
3723
3724 if (btrfs_root_readonly(root)) {
3725 ret = -EROFS;
3726 goto out;
3727 }
3728
3729 if (!inode_owner_or_capable(inode)) {
3730 ret = -EACCES;
3731 goto out;
3732 }
3733
3734 sa = memdup_user(arg, sizeof(*sa));
3735 if (IS_ERR(sa)) {
3736 ret = PTR_ERR(sa);
3737 sa = NULL;
3738 goto out;
3739 }
3740
3741 trans = btrfs_start_transaction(root, 1);
3742 if (IS_ERR(trans)) {
3743 ret = PTR_ERR(trans);
3744 trans = NULL;
3745 goto out;
3746 }
3747
3748 sa->rtransid = trans->transid;
3749 sa->rtime.sec = ct.tv_sec;
3750 sa->rtime.nsec = ct.tv_nsec;
3751
3752 memcpy(root_item->received_uuid, sa->uuid, BTRFS_UUID_SIZE);
3753 btrfs_set_root_stransid(root_item, sa->stransid);
3754 btrfs_set_root_rtransid(root_item, sa->rtransid);
3755 root_item->stime.sec = cpu_to_le64(sa->stime.sec);
3756 root_item->stime.nsec = cpu_to_le32(sa->stime.nsec);
3757 root_item->rtime.sec = cpu_to_le64(sa->rtime.sec);
3758 root_item->rtime.nsec = cpu_to_le32(sa->rtime.nsec);
3759
3760 ret = btrfs_update_root(trans, root->fs_info->tree_root,
3761 &root->root_key, &root->root_item);
3762 if (ret < 0) {
3763 btrfs_end_transaction(trans, root);
3764 trans = NULL;
3765 goto out;
3766 } else {
3767 ret = btrfs_commit_transaction(trans, root);
3768 if (ret < 0)
3769 goto out;
3770 }
3771
3772 ret = copy_to_user(arg, sa, sizeof(*sa));
3773 if (ret)
3774 ret = -EFAULT;
3775
3776out:
3777 kfree(sa);
3778 up_write(&root->fs_info->subvol_sem);
3779 mnt_drop_write_file(file);
3780 return ret;
3781}
3782
f46b5a66
CH		 3783long btrfs_ioctl(struct file *file, unsigned int
3784 cmd, unsigned long arg)
3785{
3786 struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
4bcabaa3 3787 void __user *argp = (void __user *)arg;
f46b5a66
CH		 3788
3789 switch (cmd) {
6cbff00f
CH		 3790 case FS_IOC_GETFLAGS:
3791 return btrfs_ioctl_getflags(file, argp);
3792 case FS_IOC_SETFLAGS:
3793 return btrfs_ioctl_setflags(file, argp);
3794 case FS_IOC_GETVERSION:
3795 return btrfs_ioctl_getversion(file, argp);
f7039b1d
LD		 3796 case FITRIM:
3797 return btrfs_ioctl_fitrim(file, argp);
f46b5a66 3798 case BTRFS_IOC_SNAP_CREATE:
fa0d2b9b 3799 return btrfs_ioctl_snap_create(file, argp, 0);
fdfb1e4f 3800 case BTRFS_IOC_SNAP_CREATE_V2:
fa0d2b9b 3801 return btrfs_ioctl_snap_create_v2(file, argp, 0);
3de4586c 3802 case BTRFS_IOC_SUBVOL_CREATE:
fa0d2b9b 3803 return btrfs_ioctl_snap_create(file, argp, 1);
6f72c7e2
AJ		 3804 case BTRFS_IOC_SUBVOL_CREATE_V2:
3805 return btrfs_ioctl_snap_create_v2(file, argp, 1);
76dda93c
YZ		 3806 case BTRFS_IOC_SNAP_DESTROY:
3807 return btrfs_ioctl_snap_destroy(file, argp);
0caa102d
LZ		 3808 case BTRFS_IOC_SUBVOL_GETFLAGS:
3809 return btrfs_ioctl_subvol_getflags(file, argp);
3810 case BTRFS_IOC_SUBVOL_SETFLAGS:
3811 return btrfs_ioctl_subvol_setflags(file, argp);
6ef5ed0d
JB		 3812 case BTRFS_IOC_DEFAULT_SUBVOL:
3813 return btrfs_ioctl_default_subvol(file, argp);
f46b5a66 3814 case BTRFS_IOC_DEFRAG:
1e701a32
CM		 3815 return btrfs_ioctl_defrag(file, NULL);
3816 case BTRFS_IOC_DEFRAG_RANGE:
3817 return btrfs_ioctl_defrag(file, argp);
f46b5a66 3818 case BTRFS_IOC_RESIZE:
4bcabaa3 3819 return btrfs_ioctl_resize(root, argp);
f46b5a66 3820 case BTRFS_IOC_ADD_DEV:
4bcabaa3 3821 return btrfs_ioctl_add_dev(root, argp);
f46b5a66 3822 case BTRFS_IOC_RM_DEV:
4bcabaa3 3823 return btrfs_ioctl_rm_dev(root, argp);
475f6387
JS		 3824 case BTRFS_IOC_FS_INFO:
3825 return btrfs_ioctl_fs_info(root, argp);
3826 case BTRFS_IOC_DEV_INFO:
3827 return btrfs_ioctl_dev_info(root, argp);
f46b5a66 3828 case BTRFS_IOC_BALANCE:
9ba1f6e4 3829 return btrfs_ioctl_balance(file, NULL);
f46b5a66 3830 case BTRFS_IOC_CLONE:
c5c9cd4d
SW		 3831 return btrfs_ioctl_clone(file, arg, 0, 0, 0);
3832 case BTRFS_IOC_CLONE_RANGE:
7a865e8a 3833 return btrfs_ioctl_clone_range(file, argp);
f46b5a66
CH		 3834 case BTRFS_IOC_TRANS_START:
3835 return btrfs_ioctl_trans_start(file);
3836 case BTRFS_IOC_TRANS_END:
3837 return btrfs_ioctl_trans_end(file);
ac8e9819
CM		 3838 case BTRFS_IOC_TREE_SEARCH:
3839 return btrfs_ioctl_tree_search(file, argp);
3840 case BTRFS_IOC_INO_LOOKUP:
3841 return btrfs_ioctl_ino_lookup(file, argp);
d7728c96
JS		 3842 case BTRFS_IOC_INO_PATHS:
3843 return btrfs_ioctl_ino_to_path(root, argp);
3844 case BTRFS_IOC_LOGICAL_INO:
3845 return btrfs_ioctl_logical_to_ino(root, argp);
1406e432
JB		 3846 case BTRFS_IOC_SPACE_INFO:
3847 return btrfs_ioctl_space_info(root, argp);
f46b5a66
CH		 3848 case BTRFS_IOC_SYNC:
3849 btrfs_sync_fs(file->f_dentry->d_sb, 1);
3850 return 0;
46204592 3851 case BTRFS_IOC_START_SYNC:
9a8c28be 3852 return btrfs_ioctl_start_sync(root, argp);
46204592 3853 case BTRFS_IOC_WAIT_SYNC:
9a8c28be 3854 return btrfs_ioctl_wait_sync(root, argp);
475f6387
JS		 3855 case BTRFS_IOC_SCRUB:
3856 return btrfs_ioctl_scrub(root, argp);
3857 case BTRFS_IOC_SCRUB_CANCEL:
3858 return btrfs_ioctl_scrub_cancel(root, argp);
3859 case BTRFS_IOC_SCRUB_PROGRESS:
3860 return btrfs_ioctl_scrub_progress(root, argp);
c9e9f97b 3861 case BTRFS_IOC_BALANCE_V2:
9ba1f6e4 3862 return btrfs_ioctl_balance(file, argp);
837d5b6e
ID		 3863 case BTRFS_IOC_BALANCE_CTL:
3864 return btrfs_ioctl_balance_ctl(root, arg);
19a39dce
ID		 3865 case BTRFS_IOC_BALANCE_PROGRESS:
3866 return btrfs_ioctl_balance_progress(root, argp);
8ea05e3a
AB		 3867 case BTRFS_IOC_SET_RECEIVED_SUBVOL:
3868 return btrfs_ioctl_set_received_subvol(file, argp);
31db9f7c
AB		 3869 case BTRFS_IOC_SEND:
3870 return btrfs_ioctl_send(file, argp);
c11d2c23 3871 case BTRFS_IOC_GET_DEV_STATS:
b27f7c0c 3872 return btrfs_ioctl_get_dev_stats(root, argp);
5d13a37b
AJ		 3873 case BTRFS_IOC_QUOTA_CTL:
3874 return btrfs_ioctl_quota_ctl(root, argp);
3875 case BTRFS_IOC_QGROUP_ASSIGN:
3876 return btrfs_ioctl_qgroup_assign(root, argp);
3877 case BTRFS_IOC_QGROUP_CREATE:
3878 return btrfs_ioctl_qgroup_create(root, argp);
3879 case BTRFS_IOC_QGROUP_LIMIT:
3880 return btrfs_ioctl_qgroup_limit(root, argp);
3f6bcfbd
SB		 3881 case BTRFS_IOC_DEV_REPLACE:
3882 return btrfs_ioctl_dev_replace(root, argp);
f46b5a66
CH		 3883 }
3884
3885 return -ENOTTY;
3886}
kernel source for telekom puls (alcatel/mediatek)