2 * Copyright (C) 2007 Oracle. All rights reserved.
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public
6 * License v2 as published by the Free Software Foundation.
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
13 * You should have received a copy of the GNU General Public
14 * License along with this program; if not, write to the
15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
16 * Boston, MA 021110-1307, USA.
19 #include <linux/blkdev.h>
20 #include <linux/module.h>
21 #include <linux/buffer_head.h>
23 #include <linux/pagemap.h>
24 #include <linux/highmem.h>
25 #include <linux/time.h>
26 #include <linux/init.h>
27 #include <linux/seq_file.h>
28 #include <linux/string.h>
29 #include <linux/smp_lock.h>
30 #include <linux/backing-dev.h>
31 #include <linux/mount.h>
32 #include <linux/mpage.h>
33 #include <linux/swap.h>
34 #include <linux/writeback.h>
35 #include <linux/statfs.h>
36 #include <linux/compat.h>
37 #include <linux/parser.h>
38 #include <linux/ctype.h>
39 #include <linux/namei.h>
40 #include <linux/miscdevice.h>
41 #include <linux/magic.h>
45 #include "transaction.h"
46 #include "btrfs_inode.h"
48 #include "print-tree.h"
53 #include "compression.h"
55 static struct super_operations btrfs_super_ops;
57 static void btrfs_put_super(struct super_block *sb)
59 struct btrfs_root *root = btrfs_sb(sb);
62 ret = close_ctree(root);
67 Opt_degraded, Opt_subvol, Opt_device, Opt_nodatasum, Opt_nodatacow,
68 Opt_max_extent, Opt_max_inline, Opt_alloc_start, Opt_nobarrier,
69 Opt_ssd, Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl,
70 Opt_compress, Opt_notreelog, Opt_ratio, Opt_flushoncommit, Opt_err,
73 static match_table_t tokens = {
74 {Opt_degraded, "degraded"},
75 {Opt_subvol, "subvol=%s"},
76 {Opt_device, "device=%s"},
77 {Opt_nodatasum, "nodatasum"},
78 {Opt_nodatacow, "nodatacow"},
79 {Opt_nobarrier, "nobarrier"},
80 {Opt_max_extent, "max_extent=%s"},
81 {Opt_max_inline, "max_inline=%s"},
82 {Opt_alloc_start, "alloc_start=%s"},
83 {Opt_thread_pool, "thread_pool=%d"},
84 {Opt_compress, "compress"},
86 {Opt_ssd_spread, "ssd_spread"},
89 {Opt_notreelog, "notreelog"},
90 {Opt_flushoncommit, "flushoncommit"},
91 {Opt_ratio, "metadata_ratio=%d"},
95 u64 btrfs_parse_size(char *str)
102 res = simple_strtoul(str, &end, 10);
106 last = tolower(last);
121 * Regular mount options parser. Everything that is needed only when
122 * reading in a new superblock is parsed here.
124 int btrfs_parse_options(struct btrfs_root *root, char *options)
126 struct btrfs_fs_info *info = root->fs_info;
127 substring_t args[MAX_OPT_ARGS];
135 * strsep changes the string, duplicate it because parse_options
138 options = kstrdup(options, GFP_NOFS);
143 while ((p = strsep(&options, ",")) != NULL) {
148 token = match_token(p, tokens, args);
151 printk(KERN_INFO "btrfs: allowing degraded mounts\n");
152 btrfs_set_opt(info->mount_opt, DEGRADED);
157 * These are parsed by btrfs_parse_early_options
158 * and can be happily ignored here.
162 printk(KERN_INFO "btrfs: setting nodatasum\n");
163 btrfs_set_opt(info->mount_opt, NODATASUM);
166 printk(KERN_INFO "btrfs: setting nodatacow\n");
167 btrfs_set_opt(info->mount_opt, NODATACOW);
168 btrfs_set_opt(info->mount_opt, NODATASUM);
171 printk(KERN_INFO "btrfs: use compression\n");
172 btrfs_set_opt(info->mount_opt, COMPRESS);
175 printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
176 btrfs_set_opt(info->mount_opt, SSD);
179 printk(KERN_INFO "btrfs: use spread ssd "
180 "allocation scheme\n");
181 btrfs_set_opt(info->mount_opt, SSD);
182 btrfs_set_opt(info->mount_opt, SSD_SPREAD);
185 printk(KERN_INFO "btrfs: not using ssd allocation "
187 btrfs_set_opt(info->mount_opt, NOSSD);
188 btrfs_clear_opt(info->mount_opt, SSD);
189 btrfs_clear_opt(info->mount_opt, SSD_SPREAD);
192 printk(KERN_INFO "btrfs: turning off barriers\n");
193 btrfs_set_opt(info->mount_opt, NOBARRIER);
195 case Opt_thread_pool:
197 match_int(&args[0], &intarg);
199 info->thread_pool_size = intarg;
200 printk(KERN_INFO "btrfs: thread pool %d\n",
201 info->thread_pool_size);
205 num = match_strdup(&args[0]);
207 info->max_extent = btrfs_parse_size(num);
210 info->max_extent = max_t(u64,
211 info->max_extent, root->sectorsize);
212 printk(KERN_INFO "btrfs: max_extent at %llu\n",
213 (unsigned long long)info->max_extent);
217 num = match_strdup(&args[0]);
219 info->max_inline = btrfs_parse_size(num);
222 if (info->max_inline) {
223 info->max_inline = max_t(u64,
227 printk(KERN_INFO "btrfs: max_inline at %llu\n",
228 (unsigned long long)info->max_inline);
231 case Opt_alloc_start:
232 num = match_strdup(&args[0]);
234 info->alloc_start = btrfs_parse_size(num);
237 "btrfs: allocations start at %llu\n",
238 (unsigned long long)info->alloc_start);
242 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
245 printk(KERN_INFO "btrfs: disabling tree log\n");
246 btrfs_set_opt(info->mount_opt, NOTREELOG);
248 case Opt_flushoncommit:
249 printk(KERN_INFO "btrfs: turning on flush-on-commit\n");
250 btrfs_set_opt(info->mount_opt, FLUSHONCOMMIT);
254 match_int(&args[0], &intarg);
256 info->metadata_ratio = intarg;
257 printk(KERN_INFO "btrfs: metadata ratio %d\n",
258 info->metadata_ratio);
270 * Parse mount options that are required early in the mount process.
272 * All other options will be parsed on much later in the mount process and
273 * only when we need to allocate a new super block.
275 static int btrfs_parse_early_options(const char *options, fmode_t flags,
276 void *holder, char **subvol_name,
277 struct btrfs_fs_devices **fs_devices)
279 substring_t args[MAX_OPT_ARGS];
287 * strsep changes the string, duplicate it because parse_options
290 opts = kstrdup(options, GFP_KERNEL);
294 while ((p = strsep(&opts, ",")) != NULL) {
299 token = match_token(p, tokens, args);
302 *subvol_name = match_strdup(&args[0]);
305 error = btrfs_scan_one_device(match_strdup(&args[0]),
306 flags, holder, fs_devices);
319 * If no subvolume name is specified we use the default one. Allocate
320 * a copy of the string "." here so that code later in the
321 * mount path doesn't care if it's the default volume or another one.
324 *subvol_name = kstrdup(".", GFP_KERNEL);
331 static int btrfs_fill_super(struct super_block *sb,
332 struct btrfs_fs_devices *fs_devices,
333 void *data, int silent)
336 struct dentry *root_dentry;
337 struct btrfs_super_block *disk_super;
338 struct btrfs_root *tree_root;
339 struct btrfs_key key;
342 sb->s_maxbytes = MAX_LFS_FILESIZE;
343 sb->s_magic = BTRFS_SUPER_MAGIC;
344 sb->s_op = &btrfs_super_ops;
345 sb->s_export_op = &btrfs_export_ops;
346 sb->s_xattr = btrfs_xattr_handlers;
348 sb->s_flags |= MS_POSIXACL;
350 tree_root = open_ctree(sb, fs_devices, (char *)data);
352 if (IS_ERR(tree_root)) {
353 printk("btrfs: open_ctree failed\n");
354 return PTR_ERR(tree_root);
356 sb->s_fs_info = tree_root;
357 disk_super = &tree_root->fs_info->super_copy;
359 key.objectid = BTRFS_FIRST_FREE_OBJECTID;
360 key.type = BTRFS_INODE_ITEM_KEY;
362 inode = btrfs_iget(sb, &key, tree_root->fs_info->fs_root);
364 err = PTR_ERR(inode);
368 root_dentry = d_alloc_root(inode);
375 /* this does the super kobj at the same time */
376 err = btrfs_sysfs_add_super(tree_root->fs_info);
381 sb->s_root = root_dentry;
383 save_mount_options(sb, data);
387 close_ctree(tree_root);
391 int btrfs_sync_fs(struct super_block *sb, int wait)
393 struct btrfs_trans_handle *trans;
394 struct btrfs_root *root = btrfs_sb(sb);
398 filemap_flush(root->fs_info->btree_inode->i_mapping);
402 btrfs_start_delalloc_inodes(root);
403 btrfs_wait_ordered_extents(root, 0);
405 trans = btrfs_start_transaction(root, 1);
406 ret = btrfs_commit_transaction(trans, root);
410 static int btrfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
412 struct btrfs_root *root = btrfs_sb(vfs->mnt_sb);
413 struct btrfs_fs_info *info = root->fs_info;
415 if (btrfs_test_opt(root, DEGRADED))
416 seq_puts(seq, ",degraded");
417 if (btrfs_test_opt(root, NODATASUM))
418 seq_puts(seq, ",nodatasum");
419 if (btrfs_test_opt(root, NODATACOW))
420 seq_puts(seq, ",nodatacow");
421 if (btrfs_test_opt(root, NOBARRIER))
422 seq_puts(seq, ",nobarrier");
423 if (info->max_extent != (u64)-1)
424 seq_printf(seq, ",max_extent=%llu",
425 (unsigned long long)info->max_extent);
426 if (info->max_inline != 8192 * 1024)
427 seq_printf(seq, ",max_inline=%llu",
428 (unsigned long long)info->max_inline);
429 if (info->alloc_start != 0)
430 seq_printf(seq, ",alloc_start=%llu",
431 (unsigned long long)info->alloc_start);
432 if (info->thread_pool_size != min_t(unsigned long,
433 num_online_cpus() + 2, 8))
434 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
435 if (btrfs_test_opt(root, COMPRESS))
436 seq_puts(seq, ",compress");
437 if (btrfs_test_opt(root, NOSSD))
438 seq_puts(seq, ",nossd");
439 if (btrfs_test_opt(root, SSD_SPREAD))
440 seq_puts(seq, ",ssd_spread");
441 else if (btrfs_test_opt(root, SSD))
442 seq_puts(seq, ",ssd");
443 if (btrfs_test_opt(root, NOTREELOG))
444 seq_puts(seq, ",notreelog");
445 if (btrfs_test_opt(root, FLUSHONCOMMIT))
446 seq_puts(seq, ",flushoncommit");
447 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
448 seq_puts(seq, ",noacl");
452 static int btrfs_test_super(struct super_block *s, void *data)
454 struct btrfs_fs_devices *test_fs_devices = data;
455 struct btrfs_root *root = btrfs_sb(s);
457 return root->fs_info->fs_devices == test_fs_devices;
461 * Find a superblock for the given device / mount point.
463 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
464 * for multiple device setup. Make sure to keep it in sync.
466 static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
467 const char *dev_name, void *data, struct vfsmount *mnt)
469 char *subvol_name = NULL;
470 struct block_device *bdev = NULL;
471 struct super_block *s;
473 struct btrfs_fs_devices *fs_devices = NULL;
474 fmode_t mode = FMODE_READ;
477 if (!(flags & MS_RDONLY))
480 error = btrfs_parse_early_options(data, mode, fs_type,
481 &subvol_name, &fs_devices);
485 error = btrfs_scan_one_device(dev_name, mode, fs_type, &fs_devices);
487 goto error_free_subvol_name;
489 error = btrfs_open_devices(fs_devices, mode, fs_type);
491 goto error_free_subvol_name;
493 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
495 goto error_close_devices;
498 bdev = fs_devices->latest_bdev;
499 s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
504 if ((flags ^ s->s_flags) & MS_RDONLY) {
505 deactivate_locked_super(s);
507 goto error_close_devices;
510 btrfs_close_devices(fs_devices);
512 char b[BDEVNAME_SIZE];
515 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
516 error = btrfs_fill_super(s, fs_devices, data,
517 flags & MS_SILENT ? 1 : 0);
519 deactivate_locked_super(s);
520 goto error_free_subvol_name;
523 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
524 s->s_flags |= MS_ACTIVE;
527 if (!strcmp(subvol_name, "."))
528 root = dget(s->s_root);
530 mutex_lock(&s->s_root->d_inode->i_mutex);
531 root = lookup_one_len(subvol_name, s->s_root,
532 strlen(subvol_name));
533 mutex_unlock(&s->s_root->d_inode->i_mutex);
536 deactivate_locked_super(s);
537 error = PTR_ERR(root);
538 goto error_free_subvol_name;
540 if (!root->d_inode) {
542 deactivate_locked_super(s);
544 goto error_free_subvol_name;
549 mnt->mnt_root = root;
557 btrfs_close_devices(fs_devices);
558 error_free_subvol_name:
563 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
565 struct btrfs_root *root = btrfs_sb(sb);
568 ret = btrfs_parse_options(root, data);
572 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
575 if (*flags & MS_RDONLY) {
576 sb->s_flags |= MS_RDONLY;
578 ret = btrfs_commit_super(root);
581 if (root->fs_info->fs_devices->rw_devices == 0)
584 if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
587 /* recover relocation */
588 ret = btrfs_recover_relocation(root);
591 ret = btrfs_cleanup_fs_roots(root->fs_info);
594 sb->s_flags &= ~MS_RDONLY;
600 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
602 struct btrfs_root *root = btrfs_sb(dentry->d_sb);
603 struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
604 int bits = dentry->d_sb->s_blocksize_bits;
605 __be32 *fsid = (__be32 *)root->fs_info->fsid;
607 buf->f_namelen = BTRFS_NAME_LEN;
608 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
609 buf->f_bfree = buf->f_blocks -
610 (btrfs_super_bytes_used(disk_super) >> bits);
611 buf->f_bavail = buf->f_bfree;
612 buf->f_bsize = dentry->d_sb->s_blocksize;
613 buf->f_type = BTRFS_SUPER_MAGIC;
615 /* We treat it as constant endianness (it doesn't matter _which_)
616 because we want the fsid to come out the same whether mounted
617 on a big-endian or little-endian host */
618 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
619 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
620 /* Mask in the root object ID too, to disambiguate subvols */
621 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
622 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
627 static struct file_system_type btrfs_fs_type = {
628 .owner = THIS_MODULE,
630 .get_sb = btrfs_get_sb,
631 .kill_sb = kill_anon_super,
632 .fs_flags = FS_REQUIRES_DEV,
636 * used by btrfsctl to scan devices when no FS is mounted
638 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
641 struct btrfs_ioctl_vol_args *vol;
642 struct btrfs_fs_devices *fs_devices;
645 if (!capable(CAP_SYS_ADMIN))
648 vol = memdup_user((void __user *)arg, sizeof(*vol));
653 case BTRFS_IOC_SCAN_DEV:
654 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
655 &btrfs_fs_type, &fs_devices);
663 static int btrfs_freeze(struct super_block *sb)
665 struct btrfs_root *root = btrfs_sb(sb);
666 mutex_lock(&root->fs_info->transaction_kthread_mutex);
667 mutex_lock(&root->fs_info->cleaner_mutex);
671 static int btrfs_unfreeze(struct super_block *sb)
673 struct btrfs_root *root = btrfs_sb(sb);
674 mutex_unlock(&root->fs_info->cleaner_mutex);
675 mutex_unlock(&root->fs_info->transaction_kthread_mutex);
679 static struct super_operations btrfs_super_ops = {
680 .delete_inode = btrfs_delete_inode,
681 .put_super = btrfs_put_super,
682 .sync_fs = btrfs_sync_fs,
683 .show_options = btrfs_show_options,
684 .write_inode = btrfs_write_inode,
685 .dirty_inode = btrfs_dirty_inode,
686 .alloc_inode = btrfs_alloc_inode,
687 .destroy_inode = btrfs_destroy_inode,
688 .statfs = btrfs_statfs,
689 .remount_fs = btrfs_remount,
690 .freeze_fs = btrfs_freeze,
691 .unfreeze_fs = btrfs_unfreeze,
694 static const struct file_operations btrfs_ctl_fops = {
695 .unlocked_ioctl = btrfs_control_ioctl,
696 .compat_ioctl = btrfs_control_ioctl,
697 .owner = THIS_MODULE,
700 static struct miscdevice btrfs_misc = {
701 .minor = MISC_DYNAMIC_MINOR,
702 .name = "btrfs-control",
703 .fops = &btrfs_ctl_fops
706 static int btrfs_interface_init(void)
708 return misc_register(&btrfs_misc);
711 static void btrfs_interface_exit(void)
713 if (misc_deregister(&btrfs_misc) < 0)
714 printk(KERN_INFO "misc_deregister failed for control device");
717 static int __init init_btrfs_fs(void)
721 err = btrfs_init_sysfs();
725 err = btrfs_init_cachep();
729 err = extent_io_init();
733 err = extent_map_init();
737 err = btrfs_interface_init();
739 goto free_extent_map;
741 err = register_filesystem(&btrfs_fs_type);
743 goto unregister_ioctl;
745 printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
749 btrfs_interface_exit();
755 btrfs_destroy_cachep();
761 static void __exit exit_btrfs_fs(void)
763 btrfs_destroy_cachep();
766 btrfs_interface_exit();
767 unregister_filesystem(&btrfs_fs_type);
769 btrfs_cleanup_fs_uuids();
773 module_init(init_btrfs_fs)
774 module_exit(exit_btrfs_fs)
776 MODULE_LICENSE("GPL");