]> bbs.cooldavid.org Git - net-next-2.6.git/blame - fs/btrfs/super.c
xps: Transmit Packet Steering
[net-next-2.6.git] / fs / btrfs / super.c
CommitLineData
6cbd5570
CM
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
4b82d6e4 19#include <linux/blkdev.h>
2e635a27 20#include <linux/module.h>
e20d96d6 21#include <linux/buffer_head.h>
2e635a27
CM
22#include <linux/fs.h>
23#include <linux/pagemap.h>
24#include <linux/highmem.h>
25#include <linux/time.h>
26#include <linux/init.h>
a9572a15 27#include <linux/seq_file.h>
2e635a27 28#include <linux/string.h>
2e635a27 29#include <linux/backing-dev.h>
4b82d6e4 30#include <linux/mount.h>
dee26a9f 31#include <linux/mpage.h>
75dfe396
CM
32#include <linux/swap.h>
33#include <linux/writeback.h>
8fd17795 34#include <linux/statfs.h>
08607c1b 35#include <linux/compat.h>
95e05289 36#include <linux/parser.h>
c59f8951 37#include <linux/ctype.h>
6da6abae 38#include <linux/namei.h>
a9218f6b 39#include <linux/miscdevice.h>
1bcbf313 40#include <linux/magic.h>
5a0e3ad6 41#include <linux/slab.h>
4b4e25f2 42#include "compat.h"
2e635a27 43#include "ctree.h"
e20d96d6 44#include "disk-io.h"
d5719762 45#include "transaction.h"
2c90e5d6 46#include "btrfs_inode.h"
c5739bba 47#include "ioctl.h"
3a686375 48#include "print-tree.h"
5103e947 49#include "xattr.h"
8a4b83cc 50#include "volumes.h"
b3c3da71 51#include "version.h"
be6e8dc0 52#include "export.h"
c8b97818 53#include "compression.h"
2e635a27 54
b87221de 55static const struct super_operations btrfs_super_ops;
75dfe396 56
d397712b 57static void btrfs_put_super(struct super_block *sb)
b18c6685 58{
39279cc3 59 struct btrfs_root *root = btrfs_sb(sb);
b18c6685 60 int ret;
b18c6685 61
39279cc3 62 ret = close_ctree(root);
39279cc3 63 sb->s_fs_info = NULL;
559af821
AK
64
65 (void)ret; /* FIXME: need to fix VFS to return error? */
75dfe396
CM
66}
67
95e05289 68enum {
73f73415 69 Opt_degraded, Opt_subvol, Opt_subvolid, Opt_device, Opt_nodatasum,
287a0ab9
JB
70 Opt_nodatacow, Opt_max_inline, Opt_alloc_start, Opt_nobarrier, Opt_ssd,
71 Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl, Opt_compress,
72 Opt_compress_force, Opt_notreelog, Opt_ratio, Opt_flushoncommit,
88c2ba3b 73 Opt_discard, Opt_space_cache, Opt_clear_cache, Opt_err,
4260f7c7 74 Opt_user_subvol_rm_allowed,
95e05289
CM
75};
76
77static match_table_t tokens = {
dfe25020 78 {Opt_degraded, "degraded"},
95e05289 79 {Opt_subvol, "subvol=%s"},
73f73415 80 {Opt_subvolid, "subvolid=%d"},
43e570b0 81 {Opt_device, "device=%s"},
b6cda9bc 82 {Opt_nodatasum, "nodatasum"},
be20aa9d 83 {Opt_nodatacow, "nodatacow"},
21ad10cf 84 {Opt_nobarrier, "nobarrier"},
6f568d35 85 {Opt_max_inline, "max_inline=%s"},
8f662a76 86 {Opt_alloc_start, "alloc_start=%s"},
4543df7e 87 {Opt_thread_pool, "thread_pool=%d"},
c8b97818 88 {Opt_compress, "compress"},
a555f810 89 {Opt_compress_force, "compress-force"},
e18e4809 90 {Opt_ssd, "ssd"},
451d7585 91 {Opt_ssd_spread, "ssd_spread"},
3b30c22f 92 {Opt_nossd, "nossd"},
33268eaf 93 {Opt_noacl, "noacl"},
3a5e1404 94 {Opt_notreelog, "notreelog"},
dccae999 95 {Opt_flushoncommit, "flushoncommit"},
97e728d4 96 {Opt_ratio, "metadata_ratio=%d"},
e244a0ae 97 {Opt_discard, "discard"},
0af3d00b 98 {Opt_space_cache, "space_cache"},
88c2ba3b 99 {Opt_clear_cache, "clear_cache"},
4260f7c7 100 {Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"},
33268eaf 101 {Opt_err, NULL},
95e05289
CM
102};
103
edf24abe
CH
104/*
105 * Regular mount options parser. Everything that is needed only when
106 * reading in a new superblock is parsed here.
107 */
108int btrfs_parse_options(struct btrfs_root *root, char *options)
95e05289 109{
edf24abe 110 struct btrfs_fs_info *info = root->fs_info;
95e05289 111 substring_t args[MAX_OPT_ARGS];
da495ecc 112 char *p, *num, *orig;
4543df7e 113 int intarg;
a7a3f7ca 114 int ret = 0;
b6cda9bc 115
95e05289 116 if (!options)
edf24abe 117 return 0;
95e05289 118
be20aa9d
CM
119 /*
120 * strsep changes the string, duplicate it because parse_options
121 * gets called twice
122 */
123 options = kstrdup(options, GFP_NOFS);
124 if (!options)
125 return -ENOMEM;
126
da495ecc 127 orig = options;
be20aa9d 128
edf24abe 129 while ((p = strsep(&options, ",")) != NULL) {
95e05289
CM
130 int token;
131 if (!*p)
132 continue;
133
134 token = match_token(p, tokens, args);
135 switch (token) {
dfe25020 136 case Opt_degraded:
edf24abe
CH
137 printk(KERN_INFO "btrfs: allowing degraded mounts\n");
138 btrfs_set_opt(info->mount_opt, DEGRADED);
dfe25020 139 break;
95e05289 140 case Opt_subvol:
73f73415 141 case Opt_subvolid:
43e570b0 142 case Opt_device:
edf24abe 143 /*
43e570b0 144 * These are parsed by btrfs_parse_early_options
edf24abe
CH
145 * and can be happily ignored here.
146 */
b6cda9bc
CM
147 break;
148 case Opt_nodatasum:
067c28ad 149 printk(KERN_INFO "btrfs: setting nodatasum\n");
edf24abe 150 btrfs_set_opt(info->mount_opt, NODATASUM);
be20aa9d
CM
151 break;
152 case Opt_nodatacow:
edf24abe
CH
153 printk(KERN_INFO "btrfs: setting nodatacow\n");
154 btrfs_set_opt(info->mount_opt, NODATACOW);
155 btrfs_set_opt(info->mount_opt, NODATASUM);
95e05289 156 break;
c8b97818
CM
157 case Opt_compress:
158 printk(KERN_INFO "btrfs: use compression\n");
159 btrfs_set_opt(info->mount_opt, COMPRESS);
160 break;
a555f810
CM
161 case Opt_compress_force:
162 printk(KERN_INFO "btrfs: forcing compression\n");
163 btrfs_set_opt(info->mount_opt, FORCE_COMPRESS);
164 btrfs_set_opt(info->mount_opt, COMPRESS);
165 break;
e18e4809 166 case Opt_ssd:
edf24abe
CH
167 printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
168 btrfs_set_opt(info->mount_opt, SSD);
e18e4809 169 break;
451d7585
CM
170 case Opt_ssd_spread:
171 printk(KERN_INFO "btrfs: use spread ssd "
172 "allocation scheme\n");
173 btrfs_set_opt(info->mount_opt, SSD);
174 btrfs_set_opt(info->mount_opt, SSD_SPREAD);
175 break;
3b30c22f 176 case Opt_nossd:
451d7585
CM
177 printk(KERN_INFO "btrfs: not using ssd allocation "
178 "scheme\n");
c289811c 179 btrfs_set_opt(info->mount_opt, NOSSD);
3b30c22f 180 btrfs_clear_opt(info->mount_opt, SSD);
451d7585 181 btrfs_clear_opt(info->mount_opt, SSD_SPREAD);
3b30c22f 182 break;
21ad10cf 183 case Opt_nobarrier:
edf24abe
CH
184 printk(KERN_INFO "btrfs: turning off barriers\n");
185 btrfs_set_opt(info->mount_opt, NOBARRIER);
21ad10cf 186 break;
4543df7e
CM
187 case Opt_thread_pool:
188 intarg = 0;
189 match_int(&args[0], &intarg);
190 if (intarg) {
191 info->thread_pool_size = intarg;
192 printk(KERN_INFO "btrfs: thread pool %d\n",
193 info->thread_pool_size);
194 }
195 break;
6f568d35 196 case Opt_max_inline:
edf24abe
CH
197 num = match_strdup(&args[0]);
198 if (num) {
91748467 199 info->max_inline = memparse(num, NULL);
edf24abe
CH
200 kfree(num);
201
15ada040
CM
202 if (info->max_inline) {
203 info->max_inline = max_t(u64,
204 info->max_inline,
205 root->sectorsize);
206 }
edf24abe 207 printk(KERN_INFO "btrfs: max_inline at %llu\n",
21380931 208 (unsigned long long)info->max_inline);
6f568d35
CM
209 }
210 break;
8f662a76 211 case Opt_alloc_start:
edf24abe
CH
212 num = match_strdup(&args[0]);
213 if (num) {
91748467 214 info->alloc_start = memparse(num, NULL);
edf24abe
CH
215 kfree(num);
216 printk(KERN_INFO
217 "btrfs: allocations start at %llu\n",
21380931 218 (unsigned long long)info->alloc_start);
8f662a76
CM
219 }
220 break;
33268eaf
JB
221 case Opt_noacl:
222 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
223 break;
3a5e1404
SW
224 case Opt_notreelog:
225 printk(KERN_INFO "btrfs: disabling tree log\n");
226 btrfs_set_opt(info->mount_opt, NOTREELOG);
227 break;
dccae999
SW
228 case Opt_flushoncommit:
229 printk(KERN_INFO "btrfs: turning on flush-on-commit\n");
230 btrfs_set_opt(info->mount_opt, FLUSHONCOMMIT);
231 break;
97e728d4
JB
232 case Opt_ratio:
233 intarg = 0;
234 match_int(&args[0], &intarg);
235 if (intarg) {
236 info->metadata_ratio = intarg;
237 printk(KERN_INFO "btrfs: metadata ratio %d\n",
238 info->metadata_ratio);
239 }
240 break;
e244a0ae
CH
241 case Opt_discard:
242 btrfs_set_opt(info->mount_opt, DISCARD);
243 break;
0af3d00b
JB
244 case Opt_space_cache:
245 printk(KERN_INFO "btrfs: enabling disk space caching\n");
246 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
88c2ba3b
JB
247 case Opt_clear_cache:
248 printk(KERN_INFO "btrfs: force clearing of disk cache\n");
249 btrfs_set_opt(info->mount_opt, CLEAR_CACHE);
0af3d00b 250 break;
4260f7c7
SW
251 case Opt_user_subvol_rm_allowed:
252 btrfs_set_opt(info->mount_opt, USER_SUBVOL_RM_ALLOWED);
253 break;
a7a3f7ca
SW
254 case Opt_err:
255 printk(KERN_INFO "btrfs: unrecognized mount option "
256 "'%s'\n", p);
257 ret = -EINVAL;
258 goto out;
95e05289 259 default:
be20aa9d 260 break;
95e05289
CM
261 }
262 }
a7a3f7ca 263out:
da495ecc 264 kfree(orig);
a7a3f7ca 265 return ret;
edf24abe
CH
266}
267
268/*
269 * Parse mount options that are required early in the mount process.
270 *
271 * All other options will be parsed on much later in the mount process and
272 * only when we need to allocate a new super block.
273 */
97288f2c 274static int btrfs_parse_early_options(const char *options, fmode_t flags,
73f73415 275 void *holder, char **subvol_name, u64 *subvol_objectid,
43e570b0 276 struct btrfs_fs_devices **fs_devices)
edf24abe
CH
277{
278 substring_t args[MAX_OPT_ARGS];
279 char *opts, *p;
280 int error = 0;
73f73415 281 int intarg;
edf24abe
CH
282
283 if (!options)
284 goto out;
285
286 /*
287 * strsep changes the string, duplicate it because parse_options
288 * gets called twice
289 */
290 opts = kstrdup(options, GFP_KERNEL);
291 if (!opts)
292 return -ENOMEM;
293
294 while ((p = strsep(&opts, ",")) != NULL) {
295 int token;
296 if (!*p)
297 continue;
298
299 token = match_token(p, tokens, args);
300 switch (token) {
301 case Opt_subvol:
302 *subvol_name = match_strdup(&args[0]);
303 break;
73f73415
JB
304 case Opt_subvolid:
305 intarg = 0;
4849f01d
JB
306 error = match_int(&args[0], &intarg);
307 if (!error) {
308 /* we want the original fs_tree */
309 if (!intarg)
310 *subvol_objectid =
311 BTRFS_FS_TREE_OBJECTID;
312 else
313 *subvol_objectid = intarg;
314 }
73f73415 315 break;
43e570b0
CH
316 case Opt_device:
317 error = btrfs_scan_one_device(match_strdup(&args[0]),
318 flags, holder, fs_devices);
319 if (error)
320 goto out_free_opts;
321 break;
edf24abe
CH
322 default:
323 break;
324 }
325 }
326
43e570b0 327 out_free_opts:
edf24abe
CH
328 kfree(opts);
329 out:
330 /*
331 * If no subvolume name is specified we use the default one. Allocate
3de4586c 332 * a copy of the string "." here so that code later in the
edf24abe
CH
333 * mount path doesn't care if it's the default volume or another one.
334 */
335 if (!*subvol_name) {
3de4586c 336 *subvol_name = kstrdup(".", GFP_KERNEL);
edf24abe
CH
337 if (!*subvol_name)
338 return -ENOMEM;
339 }
340 return error;
95e05289
CM
341}
342
73f73415
JB
343static struct dentry *get_default_root(struct super_block *sb,
344 u64 subvol_objectid)
345{
346 struct btrfs_root *root = sb->s_fs_info;
347 struct btrfs_root *new_root;
348 struct btrfs_dir_item *di;
349 struct btrfs_path *path;
350 struct btrfs_key location;
351 struct inode *inode;
352 struct dentry *dentry;
353 u64 dir_id;
354 int new = 0;
355
356 /*
357 * We have a specific subvol we want to mount, just setup location and
358 * go look up the root.
359 */
360 if (subvol_objectid) {
361 location.objectid = subvol_objectid;
362 location.type = BTRFS_ROOT_ITEM_KEY;
363 location.offset = (u64)-1;
364 goto find_root;
365 }
366
367 path = btrfs_alloc_path();
368 if (!path)
369 return ERR_PTR(-ENOMEM);
370 path->leave_spinning = 1;
371
372 /*
373 * Find the "default" dir item which points to the root item that we
374 * will mount by default if we haven't been given a specific subvolume
375 * to mount.
376 */
377 dir_id = btrfs_super_root_dir(&root->fs_info->super_copy);
378 di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0);
fb4f6f91
DC
379 if (IS_ERR(di))
380 return ERR_CAST(di);
73f73415
JB
381 if (!di) {
382 /*
383 * Ok the default dir item isn't there. This is weird since
384 * it's always been there, but don't freak out, just try and
385 * mount to root most subvolume.
386 */
387 btrfs_free_path(path);
388 dir_id = BTRFS_FIRST_FREE_OBJECTID;
389 new_root = root->fs_info->fs_root;
390 goto setup_root;
391 }
392
393 btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
394 btrfs_free_path(path);
395
396find_root:
397 new_root = btrfs_read_fs_root_no_name(root->fs_info, &location);
398 if (IS_ERR(new_root))
d0b678cb 399 return ERR_CAST(new_root);
73f73415
JB
400
401 if (btrfs_root_refs(&new_root->root_item) == 0)
402 return ERR_PTR(-ENOENT);
403
404 dir_id = btrfs_root_dirid(&new_root->root_item);
405setup_root:
406 location.objectid = dir_id;
407 location.type = BTRFS_INODE_ITEM_KEY;
408 location.offset = 0;
409
410 inode = btrfs_iget(sb, &location, new_root, &new);
4cbd1149
DC
411 if (IS_ERR(inode))
412 return ERR_CAST(inode);
73f73415
JB
413
414 /*
415 * If we're just mounting the root most subvol put the inode and return
416 * a reference to the dentry. We will have already gotten a reference
417 * to the inode in btrfs_fill_super so we're good to go.
418 */
419 if (!new && sb->s_root->d_inode == inode) {
420 iput(inode);
421 return dget(sb->s_root);
422 }
423
424 if (new) {
425 const struct qstr name = { .name = "/", .len = 1 };
426
427 /*
428 * New inode, we need to make the dentry a sibling of s_root so
429 * everything gets cleaned up properly on unmount.
430 */
431 dentry = d_alloc(sb->s_root, &name);
432 if (!dentry) {
433 iput(inode);
434 return ERR_PTR(-ENOMEM);
435 }
436 d_splice_alias(inode, dentry);
437 } else {
438 /*
439 * We found the inode in cache, just find a dentry for it and
440 * put the reference to the inode we just got.
441 */
442 dentry = d_find_alias(inode);
443 iput(inode);
444 }
445
446 return dentry;
447}
448
d397712b 449static int btrfs_fill_super(struct super_block *sb,
8a4b83cc 450 struct btrfs_fs_devices *fs_devices,
d397712b 451 void *data, int silent)
75dfe396 452{
d397712b
CM
453 struct inode *inode;
454 struct dentry *root_dentry;
39279cc3 455 struct btrfs_root *tree_root;
5d4f98a2 456 struct btrfs_key key;
39279cc3 457 int err;
a429e513 458
39279cc3
CM
459 sb->s_maxbytes = MAX_LFS_FILESIZE;
460 sb->s_magic = BTRFS_SUPER_MAGIC;
461 sb->s_op = &btrfs_super_ops;
be6e8dc0 462 sb->s_export_op = &btrfs_export_ops;
5103e947 463 sb->s_xattr = btrfs_xattr_handlers;
39279cc3 464 sb->s_time_gran = 1;
0eda294d 465#ifdef CONFIG_BTRFS_FS_POSIX_ACL
33268eaf 466 sb->s_flags |= MS_POSIXACL;
49cf6f45 467#endif
a429e513 468
dfe25020 469 tree_root = open_ctree(sb, fs_devices, (char *)data);
6567e837 470
e58ca020 471 if (IS_ERR(tree_root)) {
39279cc3 472 printk("btrfs: open_ctree failed\n");
e58ca020 473 return PTR_ERR(tree_root);
a429e513 474 }
39279cc3 475 sb->s_fs_info = tree_root;
a429e513 476
5d4f98a2
YZ
477 key.objectid = BTRFS_FIRST_FREE_OBJECTID;
478 key.type = BTRFS_INODE_ITEM_KEY;
479 key.offset = 0;
73f73415 480 inode = btrfs_iget(sb, &key, tree_root->fs_info->fs_root, NULL);
5d4f98a2
YZ
481 if (IS_ERR(inode)) {
482 err = PTR_ERR(inode);
39279cc3 483 goto fail_close;
f254e52c 484 }
f254e52c 485
39279cc3
CM
486 root_dentry = d_alloc_root(inode);
487 if (!root_dentry) {
488 iput(inode);
489 err = -ENOMEM;
490 goto fail_close;
f254e52c 491 }
58176a96 492
39279cc3 493 sb->s_root = root_dentry;
6885f308 494
6885f308 495 save_mount_options(sb, data);
2619ba1f 496 return 0;
39279cc3
CM
497
498fail_close:
499 close_ctree(tree_root);
500 return err;
2619ba1f
CM
501}
502
6bf13c0c 503int btrfs_sync_fs(struct super_block *sb, int wait)
c5739bba
CM
504{
505 struct btrfs_trans_handle *trans;
dccae999 506 struct btrfs_root *root = btrfs_sb(sb);
c5739bba 507 int ret;
2619ba1f 508
39279cc3
CM
509 if (!wait) {
510 filemap_flush(root->fs_info->btree_inode->i_mapping);
511 return 0;
512 }
771ed689 513
24bbcf04
YZ
514 btrfs_start_delalloc_inodes(root, 0);
515 btrfs_wait_ordered_extents(root, 0, 0);
771ed689 516
a22285a6 517 trans = btrfs_start_transaction(root, 0);
c5739bba 518 ret = btrfs_commit_transaction(trans, root);
54aa1f4d 519 return ret;
2c90e5d6
CM
520}
521
a9572a15
EP
522static int btrfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
523{
524 struct btrfs_root *root = btrfs_sb(vfs->mnt_sb);
525 struct btrfs_fs_info *info = root->fs_info;
526
527 if (btrfs_test_opt(root, DEGRADED))
528 seq_puts(seq, ",degraded");
529 if (btrfs_test_opt(root, NODATASUM))
530 seq_puts(seq, ",nodatasum");
531 if (btrfs_test_opt(root, NODATACOW))
532 seq_puts(seq, ",nodatacow");
533 if (btrfs_test_opt(root, NOBARRIER))
534 seq_puts(seq, ",nobarrier");
a9572a15 535 if (info->max_inline != 8192 * 1024)
21380931
JB
536 seq_printf(seq, ",max_inline=%llu",
537 (unsigned long long)info->max_inline);
a9572a15 538 if (info->alloc_start != 0)
21380931
JB
539 seq_printf(seq, ",alloc_start=%llu",
540 (unsigned long long)info->alloc_start);
a9572a15
EP
541 if (info->thread_pool_size != min_t(unsigned long,
542 num_online_cpus() + 2, 8))
543 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
544 if (btrfs_test_opt(root, COMPRESS))
545 seq_puts(seq, ",compress");
c289811c
CM
546 if (btrfs_test_opt(root, NOSSD))
547 seq_puts(seq, ",nossd");
451d7585
CM
548 if (btrfs_test_opt(root, SSD_SPREAD))
549 seq_puts(seq, ",ssd_spread");
550 else if (btrfs_test_opt(root, SSD))
a9572a15 551 seq_puts(seq, ",ssd");
3a5e1404 552 if (btrfs_test_opt(root, NOTREELOG))
6b65c5c6 553 seq_puts(seq, ",notreelog");
dccae999 554 if (btrfs_test_opt(root, FLUSHONCOMMIT))
6b65c5c6 555 seq_puts(seq, ",flushoncommit");
20a5239a
MW
556 if (btrfs_test_opt(root, DISCARD))
557 seq_puts(seq, ",discard");
a9572a15
EP
558 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
559 seq_puts(seq, ",noacl");
560 return 0;
561}
562
a061fc8d 563static int btrfs_test_super(struct super_block *s, void *data)
4b82d6e4 564{
a061fc8d
CM
565 struct btrfs_fs_devices *test_fs_devices = data;
566 struct btrfs_root *root = btrfs_sb(s);
4b82d6e4 567
a061fc8d 568 return root->fs_info->fs_devices == test_fs_devices;
4b82d6e4
Y
569}
570
edf24abe
CH
571/*
572 * Find a superblock for the given device / mount point.
573 *
574 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
575 * for multiple device setup. Make sure to keep it in sync.
576 */
061dbc6b
AV
577static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
578 const char *dev_name, void *data)
4b82d6e4
Y
579{
580 struct block_device *bdev = NULL;
581 struct super_block *s;
582 struct dentry *root;
8a4b83cc 583 struct btrfs_fs_devices *fs_devices = NULL;
97288f2c 584 fmode_t mode = FMODE_READ;
73f73415
JB
585 char *subvol_name = NULL;
586 u64 subvol_objectid = 0;
4b82d6e4
Y
587 int error = 0;
588
97288f2c
CH
589 if (!(flags & MS_RDONLY))
590 mode |= FMODE_WRITE;
591
592 error = btrfs_parse_early_options(data, mode, fs_type,
73f73415
JB
593 &subvol_name, &subvol_objectid,
594 &fs_devices);
edf24abe 595 if (error)
061dbc6b 596 return ERR_PTR(error);
edf24abe 597
97288f2c 598 error = btrfs_scan_one_device(dev_name, mode, fs_type, &fs_devices);
8a4b83cc 599 if (error)
edf24abe 600 goto error_free_subvol_name;
4b82d6e4 601
97288f2c 602 error = btrfs_open_devices(fs_devices, mode, fs_type);
8a4b83cc 603 if (error)
edf24abe 604 goto error_free_subvol_name;
8a4b83cc 605
2b82032c
YZ
606 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
607 error = -EACCES;
608 goto error_close_devices;
609 }
610
dfe25020 611 bdev = fs_devices->latest_bdev;
a061fc8d 612 s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
4b82d6e4
Y
613 if (IS_ERR(s))
614 goto error_s;
615
616 if (s->s_root) {
617 if ((flags ^ s->s_flags) & MS_RDONLY) {
6f5bbff9 618 deactivate_locked_super(s);
4b82d6e4 619 error = -EBUSY;
c146afad 620 goto error_close_devices;
4b82d6e4
Y
621 }
622
2b82032c 623 btrfs_close_devices(fs_devices);
4b82d6e4
Y
624 } else {
625 char b[BDEVNAME_SIZE];
626
627 s->s_flags = flags;
628 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
8a4b83cc
CM
629 error = btrfs_fill_super(s, fs_devices, data,
630 flags & MS_SILENT ? 1 : 0);
4b82d6e4 631 if (error) {
6f5bbff9 632 deactivate_locked_super(s);
1f483660 633 goto error_free_subvol_name;
4b82d6e4
Y
634 }
635
788f20eb 636 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
4b82d6e4
Y
637 s->s_flags |= MS_ACTIVE;
638 }
639
73f73415
JB
640 root = get_default_root(s, subvol_objectid);
641 if (IS_ERR(root)) {
642 error = PTR_ERR(root);
643 deactivate_locked_super(s);
0e78340f 644 goto error_free_subvol_name;
73f73415
JB
645 }
646 /* if they gave us a subvolume name bind mount into that */
647 if (strcmp(subvol_name, ".")) {
648 struct dentry *new_root;
649 mutex_lock(&root->d_inode->i_mutex);
650 new_root = lookup_one_len(subvol_name, root,
d397712b 651 strlen(subvol_name));
73f73415 652 mutex_unlock(&root->d_inode->i_mutex);
d397712b 653
73f73415 654 if (IS_ERR(new_root)) {
6f5bbff9 655 deactivate_locked_super(s);
73f73415
JB
656 error = PTR_ERR(new_root);
657 dput(root);
0e78340f 658 goto error_free_subvol_name;
76fcef19 659 }
73f73415 660 if (!new_root->d_inode) {
76fcef19 661 dput(root);
73f73415 662 dput(new_root);
6f5bbff9 663 deactivate_locked_super(s);
76fcef19 664 error = -ENXIO;
0e78340f 665 goto error_free_subvol_name;
76fcef19 666 }
73f73415
JB
667 dput(root);
668 root = new_root;
4b82d6e4
Y
669 }
670
edf24abe 671 kfree(subvol_name);
061dbc6b 672 return root;
4b82d6e4
Y
673
674error_s:
675 error = PTR_ERR(s);
c146afad 676error_close_devices:
8a4b83cc 677 btrfs_close_devices(fs_devices);
edf24abe
CH
678error_free_subvol_name:
679 kfree(subvol_name);
061dbc6b 680 return ERR_PTR(error);
4b82d6e4 681}
2e635a27 682
c146afad
YZ
683static int btrfs_remount(struct super_block *sb, int *flags, char *data)
684{
685 struct btrfs_root *root = btrfs_sb(sb);
686 int ret;
687
b288052e
CM
688 ret = btrfs_parse_options(root, data);
689 if (ret)
690 return -EINVAL;
691
c146afad
YZ
692 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
693 return 0;
694
695 if (*flags & MS_RDONLY) {
696 sb->s_flags |= MS_RDONLY;
697
698 ret = btrfs_commit_super(root);
699 WARN_ON(ret);
700 } else {
2b82032c
YZ
701 if (root->fs_info->fs_devices->rw_devices == 0)
702 return -EACCES;
703
c146afad
YZ
704 if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
705 return -EINVAL;
706
d68fc57b 707 ret = btrfs_cleanup_fs_roots(root->fs_info);
c146afad
YZ
708 WARN_ON(ret);
709
d68fc57b
YZ
710 /* recover relocation */
711 ret = btrfs_recover_relocation(root);
c146afad
YZ
712 WARN_ON(ret);
713
714 sb->s_flags &= ~MS_RDONLY;
715 }
716
717 return 0;
718}
719
8fd17795
CM
720static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
721{
722 struct btrfs_root *root = btrfs_sb(dentry->d_sb);
4b52dff6 723 struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
bd4d1088
JB
724 struct list_head *head = &root->fs_info->space_info;
725 struct btrfs_space_info *found;
726 u64 total_used = 0;
89a55897 727 u64 total_used_data = 0;
db94535d 728 int bits = dentry->d_sb->s_blocksize_bits;
9d03632e 729 __be32 *fsid = (__be32 *)root->fs_info->fsid;
8fd17795 730
bd4d1088 731 rcu_read_lock();
89a55897
JB
732 list_for_each_entry_rcu(found, head, list) {
733 if (found->flags & (BTRFS_BLOCK_GROUP_METADATA |
734 BTRFS_BLOCK_GROUP_SYSTEM))
735 total_used_data += found->disk_total;
736 else
737 total_used_data += found->disk_used;
b742bb82 738 total_used += found->disk_used;
89a55897 739 }
bd4d1088
JB
740 rcu_read_unlock();
741
8fd17795 742 buf->f_namelen = BTRFS_NAME_LEN;
db94535d 743 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
bd4d1088 744 buf->f_bfree = buf->f_blocks - (total_used >> bits);
89a55897 745 buf->f_bavail = buf->f_blocks - (total_used_data >> bits);
8fd17795
CM
746 buf->f_bsize = dentry->d_sb->s_blocksize;
747 buf->f_type = BTRFS_SUPER_MAGIC;
d397712b 748
9d03632e 749 /* We treat it as constant endianness (it doesn't matter _which_)
d397712b 750 because we want the fsid to come out the same whether mounted
9d03632e
DW
751 on a big-endian or little-endian host */
752 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
753 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
32d48fa1
DW
754 /* Mask in the root object ID too, to disambiguate subvols */
755 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
756 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
757
8fd17795
CM
758 return 0;
759}
b5133862 760
2e635a27
CM
761static struct file_system_type btrfs_fs_type = {
762 .owner = THIS_MODULE,
763 .name = "btrfs",
061dbc6b 764 .mount = btrfs_mount,
a061fc8d 765 .kill_sb = kill_anon_super,
2e635a27
CM
766 .fs_flags = FS_REQUIRES_DEV,
767};
a9218f6b 768
d352ac68
CM
769/*
770 * used by btrfsctl to scan devices when no FS is mounted
771 */
8a4b83cc
CM
772static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
773 unsigned long arg)
774{
775 struct btrfs_ioctl_vol_args *vol;
776 struct btrfs_fs_devices *fs_devices;
c071fcfd 777 int ret = -ENOTTY;
8a4b83cc 778
e441d54d
CM
779 if (!capable(CAP_SYS_ADMIN))
780 return -EPERM;
781
dae7b665
LZ
782 vol = memdup_user((void __user *)arg, sizeof(*vol));
783 if (IS_ERR(vol))
784 return PTR_ERR(vol);
c071fcfd 785
8a4b83cc
CM
786 switch (cmd) {
787 case BTRFS_IOC_SCAN_DEV:
97288f2c 788 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
8a4b83cc
CM
789 &btrfs_fs_type, &fs_devices);
790 break;
791 }
dae7b665 792
8a4b83cc 793 kfree(vol);
f819d837 794 return ret;
8a4b83cc
CM
795}
796
0176260f 797static int btrfs_freeze(struct super_block *sb)
ed0dab6b
Y
798{
799 struct btrfs_root *root = btrfs_sb(sb);
a74a4b97
CM
800 mutex_lock(&root->fs_info->transaction_kthread_mutex);
801 mutex_lock(&root->fs_info->cleaner_mutex);
0176260f 802 return 0;
ed0dab6b
Y
803}
804
0176260f 805static int btrfs_unfreeze(struct super_block *sb)
ed0dab6b
Y
806{
807 struct btrfs_root *root = btrfs_sb(sb);
a74a4b97
CM
808 mutex_unlock(&root->fs_info->cleaner_mutex);
809 mutex_unlock(&root->fs_info->transaction_kthread_mutex);
0176260f 810 return 0;
ed0dab6b 811}
2e635a27 812
b87221de 813static const struct super_operations btrfs_super_ops = {
76dda93c 814 .drop_inode = btrfs_drop_inode,
bd555975 815 .evict_inode = btrfs_evict_inode,
e20d96d6 816 .put_super = btrfs_put_super,
d5719762 817 .sync_fs = btrfs_sync_fs,
a9572a15 818 .show_options = btrfs_show_options,
4730a4bc 819 .write_inode = btrfs_write_inode,
b5133862 820 .dirty_inode = btrfs_dirty_inode,
2c90e5d6
CM
821 .alloc_inode = btrfs_alloc_inode,
822 .destroy_inode = btrfs_destroy_inode,
8fd17795 823 .statfs = btrfs_statfs,
c146afad 824 .remount_fs = btrfs_remount,
0176260f
LT
825 .freeze_fs = btrfs_freeze,
826 .unfreeze_fs = btrfs_unfreeze,
e20d96d6 827};
a9218f6b
CM
828
829static const struct file_operations btrfs_ctl_fops = {
830 .unlocked_ioctl = btrfs_control_ioctl,
831 .compat_ioctl = btrfs_control_ioctl,
832 .owner = THIS_MODULE,
6038f373 833 .llseek = noop_llseek,
a9218f6b
CM
834};
835
836static struct miscdevice btrfs_misc = {
578454ff 837 .minor = BTRFS_MINOR,
a9218f6b
CM
838 .name = "btrfs-control",
839 .fops = &btrfs_ctl_fops
840};
841
578454ff
KS
842MODULE_ALIAS_MISCDEV(BTRFS_MINOR);
843MODULE_ALIAS("devname:btrfs-control");
844
a9218f6b
CM
845static int btrfs_interface_init(void)
846{
847 return misc_register(&btrfs_misc);
848}
849
b2950863 850static void btrfs_interface_exit(void)
a9218f6b
CM
851{
852 if (misc_deregister(&btrfs_misc) < 0)
d397712b 853 printk(KERN_INFO "misc_deregister failed for control device");
a9218f6b
CM
854}
855
2e635a27
CM
856static int __init init_btrfs_fs(void)
857{
2c90e5d6 858 int err;
58176a96
JB
859
860 err = btrfs_init_sysfs();
861 if (err)
862 return err;
863
39279cc3 864 err = btrfs_init_cachep();
2c90e5d6 865 if (err)
a74a4b97 866 goto free_sysfs;
d1310b2e
CM
867
868 err = extent_io_init();
2f4cbe64
WB
869 if (err)
870 goto free_cachep;
871
d1310b2e
CM
872 err = extent_map_init();
873 if (err)
874 goto free_extent_io;
875
a9218f6b 876 err = btrfs_interface_init();
2f4cbe64
WB
877 if (err)
878 goto free_extent_map;
c8b97818 879
a9218f6b
CM
880 err = register_filesystem(&btrfs_fs_type);
881 if (err)
882 goto unregister_ioctl;
b3c3da71
CM
883
884 printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
2f4cbe64
WB
885 return 0;
886
a9218f6b
CM
887unregister_ioctl:
888 btrfs_interface_exit();
2f4cbe64
WB
889free_extent_map:
890 extent_map_exit();
d1310b2e
CM
891free_extent_io:
892 extent_io_exit();
2f4cbe64
WB
893free_cachep:
894 btrfs_destroy_cachep();
a74a4b97 895free_sysfs:
2f4cbe64
WB
896 btrfs_exit_sysfs();
897 return err;
2e635a27
CM
898}
899
900static void __exit exit_btrfs_fs(void)
901{
39279cc3 902 btrfs_destroy_cachep();
a52d9a80 903 extent_map_exit();
d1310b2e 904 extent_io_exit();
a9218f6b 905 btrfs_interface_exit();
2e635a27 906 unregister_filesystem(&btrfs_fs_type);
58176a96 907 btrfs_exit_sysfs();
8a4b83cc 908 btrfs_cleanup_fs_uuids();
c8b97818 909 btrfs_zlib_exit();
2e635a27
CM
910}
911
912module_init(init_btrfs_fs)
913module_exit(exit_btrfs_fs)
914
915MODULE_LICENSE("GPL");