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_thread_pool, Opt_noacl, Opt_compress,
70 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"},
88 {Opt_notreelog, "notreelog"},
89 {Opt_flushoncommit, "flushoncommit"},
90 {Opt_ratio, "metadata_ratio=%d"},
94 u64 btrfs_parse_size(char *str)
101 res = simple_strtoul(str, &end, 10);
105 last = tolower(last);
120 * Regular mount options parser. Everything that is needed only when
121 * reading in a new superblock is parsed here.
123 int btrfs_parse_options(struct btrfs_root *root, char *options)
125 struct btrfs_fs_info *info = root->fs_info;
126 substring_t args[MAX_OPT_ARGS];
134 * strsep changes the string, duplicate it because parse_options
137 options = kstrdup(options, GFP_NOFS);
142 while ((p = strsep(&options, ",")) != NULL) {
147 token = match_token(p, tokens, args);
150 printk(KERN_INFO "btrfs: allowing degraded mounts\n");
151 btrfs_set_opt(info->mount_opt, DEGRADED);
156 * These are parsed by btrfs_parse_early_options
157 * and can be happily ignored here.
161 printk(KERN_INFO "btrfs: setting nodatacsum\n");
162 btrfs_set_opt(info->mount_opt, NODATASUM);
165 printk(KERN_INFO "btrfs: setting nodatacow\n");
166 btrfs_set_opt(info->mount_opt, NODATACOW);
167 btrfs_set_opt(info->mount_opt, NODATASUM);
170 printk(KERN_INFO "btrfs: use compression\n");
171 btrfs_set_opt(info->mount_opt, COMPRESS);
174 printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
175 btrfs_set_opt(info->mount_opt, SSD);
178 printk(KERN_INFO "btrfs: not using ssd allocation scheme\n");
179 btrfs_clear_opt(info->mount_opt, SSD);
182 printk(KERN_INFO "btrfs: turning off barriers\n");
183 btrfs_set_opt(info->mount_opt, NOBARRIER);
185 case Opt_thread_pool:
187 match_int(&args[0], &intarg);
189 info->thread_pool_size = intarg;
190 printk(KERN_INFO "btrfs: thread pool %d\n",
191 info->thread_pool_size);
195 num = match_strdup(&args[0]);
197 info->max_extent = btrfs_parse_size(num);
200 info->max_extent = max_t(u64,
201 info->max_extent, root->sectorsize);
202 printk(KERN_INFO "btrfs: max_extent at %llu\n",
203 (unsigned long long)info->max_extent);
207 num = match_strdup(&args[0]);
209 info->max_inline = btrfs_parse_size(num);
212 if (info->max_inline) {
213 info->max_inline = max_t(u64,
217 printk(KERN_INFO "btrfs: max_inline at %llu\n",
218 (unsigned long long)info->max_inline);
221 case Opt_alloc_start:
222 num = match_strdup(&args[0]);
224 info->alloc_start = btrfs_parse_size(num);
227 "btrfs: allocations start at %llu\n",
228 (unsigned long long)info->alloc_start);
232 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
235 printk(KERN_INFO "btrfs: disabling tree log\n");
236 btrfs_set_opt(info->mount_opt, NOTREELOG);
238 case Opt_flushoncommit:
239 printk(KERN_INFO "btrfs: turning on flush-on-commit\n");
240 btrfs_set_opt(info->mount_opt, FLUSHONCOMMIT);
244 match_int(&args[0], &intarg);
246 info->metadata_ratio = intarg;
247 printk(KERN_INFO "btrfs: metadata ratio %d\n",
248 info->metadata_ratio);
260 * Parse mount options that are required early in the mount process.
262 * All other options will be parsed on much later in the mount process and
263 * only when we need to allocate a new super block.
265 static int btrfs_parse_early_options(const char *options, fmode_t flags,
266 void *holder, char **subvol_name,
267 struct btrfs_fs_devices **fs_devices)
269 substring_t args[MAX_OPT_ARGS];
277 * strsep changes the string, duplicate it because parse_options
280 opts = kstrdup(options, GFP_KERNEL);
284 while ((p = strsep(&opts, ",")) != NULL) {
289 token = match_token(p, tokens, args);
292 *subvol_name = match_strdup(&args[0]);
295 error = btrfs_scan_one_device(match_strdup(&args[0]),
296 flags, holder, fs_devices);
309 * If no subvolume name is specified we use the default one. Allocate
310 * a copy of the string "." here so that code later in the
311 * mount path doesn't care if it's the default volume or another one.
314 *subvol_name = kstrdup(".", GFP_KERNEL);
321 static int btrfs_fill_super(struct super_block *sb,
322 struct btrfs_fs_devices *fs_devices,
323 void *data, int silent)
326 struct dentry *root_dentry;
327 struct btrfs_super_block *disk_super;
328 struct btrfs_root *tree_root;
329 struct btrfs_key key;
332 sb->s_maxbytes = MAX_LFS_FILESIZE;
333 sb->s_magic = BTRFS_SUPER_MAGIC;
334 sb->s_op = &btrfs_super_ops;
335 sb->s_export_op = &btrfs_export_ops;
336 sb->s_xattr = btrfs_xattr_handlers;
338 sb->s_flags |= MS_POSIXACL;
340 tree_root = open_ctree(sb, fs_devices, (char *)data);
342 if (IS_ERR(tree_root)) {
343 printk("btrfs: open_ctree failed\n");
344 return PTR_ERR(tree_root);
346 sb->s_fs_info = tree_root;
347 disk_super = &tree_root->fs_info->super_copy;
349 key.objectid = BTRFS_FIRST_FREE_OBJECTID;
350 key.type = BTRFS_INODE_ITEM_KEY;
352 inode = btrfs_iget(sb, &key, tree_root->fs_info->fs_root);
354 err = PTR_ERR(inode);
358 root_dentry = d_alloc_root(inode);
365 /* this does the super kobj at the same time */
366 err = btrfs_sysfs_add_super(tree_root->fs_info);
371 sb->s_root = root_dentry;
373 save_mount_options(sb, data);
377 close_ctree(tree_root);
381 int btrfs_sync_fs(struct super_block *sb, int wait)
383 struct btrfs_trans_handle *trans;
384 struct btrfs_root *root = btrfs_sb(sb);
387 if (sb->s_flags & MS_RDONLY)
392 filemap_flush(root->fs_info->btree_inode->i_mapping);
396 btrfs_start_delalloc_inodes(root);
397 btrfs_wait_ordered_extents(root, 0);
399 trans = btrfs_start_transaction(root, 1);
400 ret = btrfs_commit_transaction(trans, root);
405 static int btrfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
407 struct btrfs_root *root = btrfs_sb(vfs->mnt_sb);
408 struct btrfs_fs_info *info = root->fs_info;
410 if (btrfs_test_opt(root, DEGRADED))
411 seq_puts(seq, ",degraded");
412 if (btrfs_test_opt(root, NODATASUM))
413 seq_puts(seq, ",nodatasum");
414 if (btrfs_test_opt(root, NODATACOW))
415 seq_puts(seq, ",nodatacow");
416 if (btrfs_test_opt(root, NOBARRIER))
417 seq_puts(seq, ",nobarrier");
418 if (info->max_extent != (u64)-1)
419 seq_printf(seq, ",max_extent=%llu",
420 (unsigned long long)info->max_extent);
421 if (info->max_inline != 8192 * 1024)
422 seq_printf(seq, ",max_inline=%llu",
423 (unsigned long long)info->max_inline);
424 if (info->alloc_start != 0)
425 seq_printf(seq, ",alloc_start=%llu",
426 (unsigned long long)info->alloc_start);
427 if (info->thread_pool_size != min_t(unsigned long,
428 num_online_cpus() + 2, 8))
429 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
430 if (btrfs_test_opt(root, COMPRESS))
431 seq_puts(seq, ",compress");
432 if (btrfs_test_opt(root, SSD))
433 seq_puts(seq, ",ssd");
434 if (btrfs_test_opt(root, NOTREELOG))
435 seq_puts(seq, ",notreelog");
436 if (btrfs_test_opt(root, FLUSHONCOMMIT))
437 seq_puts(seq, ",flushoncommit");
438 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
439 seq_puts(seq, ",noacl");
443 static void btrfs_write_super(struct super_block *sb)
448 static int btrfs_test_super(struct super_block *s, void *data)
450 struct btrfs_fs_devices *test_fs_devices = data;
451 struct btrfs_root *root = btrfs_sb(s);
453 return root->fs_info->fs_devices == test_fs_devices;
457 * Find a superblock for the given device / mount point.
459 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
460 * for multiple device setup. Make sure to keep it in sync.
462 static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
463 const char *dev_name, void *data, struct vfsmount *mnt)
465 char *subvol_name = NULL;
466 struct block_device *bdev = NULL;
467 struct super_block *s;
469 struct btrfs_fs_devices *fs_devices = NULL;
470 fmode_t mode = FMODE_READ;
473 if (!(flags & MS_RDONLY))
476 error = btrfs_parse_early_options(data, mode, fs_type,
477 &subvol_name, &fs_devices);
481 error = btrfs_scan_one_device(dev_name, mode, fs_type, &fs_devices);
483 goto error_free_subvol_name;
485 error = btrfs_open_devices(fs_devices, mode, fs_type);
487 goto error_free_subvol_name;
489 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
491 goto error_close_devices;
494 bdev = fs_devices->latest_bdev;
495 s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
500 if ((flags ^ s->s_flags) & MS_RDONLY) {
501 deactivate_locked_super(s);
503 goto error_close_devices;
506 btrfs_close_devices(fs_devices);
508 char b[BDEVNAME_SIZE];
511 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
512 error = btrfs_fill_super(s, fs_devices, data,
513 flags & MS_SILENT ? 1 : 0);
515 deactivate_locked_super(s);
516 goto error_free_subvol_name;
519 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
520 s->s_flags |= MS_ACTIVE;
523 if (!strcmp(subvol_name, "."))
524 root = dget(s->s_root);
526 mutex_lock(&s->s_root->d_inode->i_mutex);
527 root = lookup_one_len(subvol_name, s->s_root,
528 strlen(subvol_name));
529 mutex_unlock(&s->s_root->d_inode->i_mutex);
532 deactivate_locked_super(s);
533 error = PTR_ERR(root);
534 goto error_free_subvol_name;
536 if (!root->d_inode) {
538 deactivate_locked_super(s);
540 goto error_free_subvol_name;
545 mnt->mnt_root = root;
553 btrfs_close_devices(fs_devices);
554 error_free_subvol_name:
559 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
561 struct btrfs_root *root = btrfs_sb(sb);
564 ret = btrfs_parse_options(root, data);
568 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
571 if (*flags & MS_RDONLY) {
572 sb->s_flags |= MS_RDONLY;
574 ret = btrfs_commit_super(root);
577 if (root->fs_info->fs_devices->rw_devices == 0)
580 if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
583 /* recover relocation */
584 ret = btrfs_recover_relocation(root);
587 ret = btrfs_cleanup_fs_roots(root->fs_info);
590 sb->s_flags &= ~MS_RDONLY;
596 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
598 struct btrfs_root *root = btrfs_sb(dentry->d_sb);
599 struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
600 int bits = dentry->d_sb->s_blocksize_bits;
601 __be32 *fsid = (__be32 *)root->fs_info->fsid;
603 buf->f_namelen = BTRFS_NAME_LEN;
604 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
605 buf->f_bfree = buf->f_blocks -
606 (btrfs_super_bytes_used(disk_super) >> bits);
607 buf->f_bavail = buf->f_bfree;
608 buf->f_bsize = dentry->d_sb->s_blocksize;
609 buf->f_type = BTRFS_SUPER_MAGIC;
611 /* We treat it as constant endianness (it doesn't matter _which_)
612 because we want the fsid to come out the same whether mounted
613 on a big-endian or little-endian host */
614 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
615 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
616 /* Mask in the root object ID too, to disambiguate subvols */
617 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
618 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
623 static struct file_system_type btrfs_fs_type = {
624 .owner = THIS_MODULE,
626 .get_sb = btrfs_get_sb,
627 .kill_sb = kill_anon_super,
628 .fs_flags = FS_REQUIRES_DEV,
632 * used by btrfsctl to scan devices when no FS is mounted
634 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
637 struct btrfs_ioctl_vol_args *vol;
638 struct btrfs_fs_devices *fs_devices;
641 if (!capable(CAP_SYS_ADMIN))
644 vol = memdup_user((void __user *)arg, sizeof(*vol));
649 case BTRFS_IOC_SCAN_DEV:
650 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
651 &btrfs_fs_type, &fs_devices);
659 static int btrfs_freeze(struct super_block *sb)
661 struct btrfs_root *root = btrfs_sb(sb);
662 mutex_lock(&root->fs_info->transaction_kthread_mutex);
663 mutex_lock(&root->fs_info->cleaner_mutex);
667 static int btrfs_unfreeze(struct super_block *sb)
669 struct btrfs_root *root = btrfs_sb(sb);
670 mutex_unlock(&root->fs_info->cleaner_mutex);
671 mutex_unlock(&root->fs_info->transaction_kthread_mutex);
675 static struct super_operations btrfs_super_ops = {
676 .delete_inode = btrfs_delete_inode,
677 .put_super = btrfs_put_super,
678 .write_super = btrfs_write_super,
679 .sync_fs = btrfs_sync_fs,
680 .show_options = btrfs_show_options,
681 .write_inode = btrfs_write_inode,
682 .dirty_inode = btrfs_dirty_inode,
683 .alloc_inode = btrfs_alloc_inode,
684 .destroy_inode = btrfs_destroy_inode,
685 .statfs = btrfs_statfs,
686 .remount_fs = btrfs_remount,
687 .freeze_fs = btrfs_freeze,
688 .unfreeze_fs = btrfs_unfreeze,
691 static const struct file_operations btrfs_ctl_fops = {
692 .unlocked_ioctl = btrfs_control_ioctl,
693 .compat_ioctl = btrfs_control_ioctl,
694 .owner = THIS_MODULE,
697 static struct miscdevice btrfs_misc = {
698 .minor = MISC_DYNAMIC_MINOR,
699 .name = "btrfs-control",
700 .fops = &btrfs_ctl_fops
703 static int btrfs_interface_init(void)
705 return misc_register(&btrfs_misc);
708 static void btrfs_interface_exit(void)
710 if (misc_deregister(&btrfs_misc) < 0)
711 printk(KERN_INFO "misc_deregister failed for control device");
714 static int __init init_btrfs_fs(void)
718 err = btrfs_init_sysfs();
722 err = btrfs_init_cachep();
726 err = extent_io_init();
730 err = extent_map_init();
734 err = btrfs_interface_init();
736 goto free_extent_map;
738 err = register_filesystem(&btrfs_fs_type);
740 goto unregister_ioctl;
742 printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
746 btrfs_interface_exit();
752 btrfs_destroy_cachep();
758 static void __exit exit_btrfs_fs(void)
760 btrfs_destroy_cachep();
763 btrfs_interface_exit();
764 unregister_filesystem(&btrfs_fs_type);
766 btrfs_cleanup_fs_uuids();
770 module_init(init_btrfs_fs)
771 module_exit(exit_btrfs_fs)
773 MODULE_LICENSE("GPL");