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/backing-dev.h>
30 #include <linux/mount.h>
31 #include <linux/mpage.h>
32 #include <linux/swap.h>
33 #include <linux/writeback.h>
34 #include <linux/statfs.h>
35 #include <linux/compat.h>
36 #include <linux/parser.h>
37 #include <linux/ctype.h>
38 #include <linux/namei.h>
39 #include <linux/miscdevice.h>
40 #include <linux/magic.h>
44 #include "transaction.h"
45 #include "btrfs_inode.h"
47 #include "print-tree.h"
52 #include "compression.h"
54 static const struct super_operations btrfs_super_ops;
56 static void btrfs_put_super(struct super_block *sb)
58 struct btrfs_root *root = btrfs_sb(sb);
61 ret = close_ctree(root);
66 Opt_degraded, Opt_subvol, Opt_device, Opt_nodatasum, Opt_nodatacow,
67 Opt_max_extent, Opt_max_inline, Opt_alloc_start, Opt_nobarrier,
68 Opt_ssd, Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl,
69 Opt_compress, Opt_notreelog, Opt_ratio, Opt_flushoncommit, Opt_err,
72 static match_table_t tokens = {
73 {Opt_degraded, "degraded"},
74 {Opt_subvol, "subvol=%s"},
75 {Opt_device, "device=%s"},
76 {Opt_nodatasum, "nodatasum"},
77 {Opt_nodatacow, "nodatacow"},
78 {Opt_nobarrier, "nobarrier"},
79 {Opt_max_extent, "max_extent=%s"},
80 {Opt_max_inline, "max_inline=%s"},
81 {Opt_alloc_start, "alloc_start=%s"},
82 {Opt_thread_pool, "thread_pool=%d"},
83 {Opt_compress, "compress"},
85 {Opt_ssd_spread, "ssd_spread"},
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 nodatasum\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: use spread ssd "
179 "allocation scheme\n");
180 btrfs_set_opt(info->mount_opt, SSD);
181 btrfs_set_opt(info->mount_opt, SSD_SPREAD);
184 printk(KERN_INFO "btrfs: not using ssd allocation "
186 btrfs_set_opt(info->mount_opt, NOSSD);
187 btrfs_clear_opt(info->mount_opt, SSD);
188 btrfs_clear_opt(info->mount_opt, SSD_SPREAD);
191 printk(KERN_INFO "btrfs: turning off barriers\n");
192 btrfs_set_opt(info->mount_opt, NOBARRIER);
194 case Opt_thread_pool:
196 match_int(&args[0], &intarg);
198 info->thread_pool_size = intarg;
199 printk(KERN_INFO "btrfs: thread pool %d\n",
200 info->thread_pool_size);
204 num = match_strdup(&args[0]);
206 info->max_extent = btrfs_parse_size(num);
209 info->max_extent = max_t(u64,
210 info->max_extent, root->sectorsize);
211 printk(KERN_INFO "btrfs: max_extent at %llu\n",
212 (unsigned long long)info->max_extent);
216 num = match_strdup(&args[0]);
218 info->max_inline = btrfs_parse_size(num);
221 if (info->max_inline) {
222 info->max_inline = max_t(u64,
226 printk(KERN_INFO "btrfs: max_inline at %llu\n",
227 (unsigned long long)info->max_inline);
230 case Opt_alloc_start:
231 num = match_strdup(&args[0]);
233 info->alloc_start = btrfs_parse_size(num);
236 "btrfs: allocations start at %llu\n",
237 (unsigned long long)info->alloc_start);
241 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
244 printk(KERN_INFO "btrfs: disabling tree log\n");
245 btrfs_set_opt(info->mount_opt, NOTREELOG);
247 case Opt_flushoncommit:
248 printk(KERN_INFO "btrfs: turning on flush-on-commit\n");
249 btrfs_set_opt(info->mount_opt, FLUSHONCOMMIT);
253 match_int(&args[0], &intarg);
255 info->metadata_ratio = intarg;
256 printk(KERN_INFO "btrfs: metadata ratio %d\n",
257 info->metadata_ratio);
269 * Parse mount options that are required early in the mount process.
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.
274 static int btrfs_parse_early_options(const char *options, fmode_t flags,
275 void *holder, char **subvol_name,
276 struct btrfs_fs_devices **fs_devices)
278 substring_t args[MAX_OPT_ARGS];
286 * strsep changes the string, duplicate it because parse_options
289 opts = kstrdup(options, GFP_KERNEL);
293 while ((p = strsep(&opts, ",")) != NULL) {
298 token = match_token(p, tokens, args);
301 *subvol_name = match_strdup(&args[0]);
304 error = btrfs_scan_one_device(match_strdup(&args[0]),
305 flags, holder, fs_devices);
318 * If no subvolume name is specified we use the default one. Allocate
319 * a copy of the string "." here so that code later in the
320 * mount path doesn't care if it's the default volume or another one.
323 *subvol_name = kstrdup(".", GFP_KERNEL);
330 static int btrfs_fill_super(struct super_block *sb,
331 struct btrfs_fs_devices *fs_devices,
332 void *data, int silent)
335 struct dentry *root_dentry;
336 struct btrfs_super_block *disk_super;
337 struct btrfs_root *tree_root;
338 struct btrfs_key key;
341 sb->s_maxbytes = MAX_LFS_FILESIZE;
342 sb->s_magic = BTRFS_SUPER_MAGIC;
343 sb->s_op = &btrfs_super_ops;
344 sb->s_export_op = &btrfs_export_ops;
345 sb->s_xattr = btrfs_xattr_handlers;
347 #ifdef CONFIG_BTRFS_POSIX_ACL
348 sb->s_flags |= MS_POSIXACL;
351 tree_root = open_ctree(sb, fs_devices, (char *)data);
353 if (IS_ERR(tree_root)) {
354 printk("btrfs: open_ctree failed\n");
355 return PTR_ERR(tree_root);
357 sb->s_fs_info = tree_root;
358 disk_super = &tree_root->fs_info->super_copy;
360 key.objectid = BTRFS_FIRST_FREE_OBJECTID;
361 key.type = BTRFS_INODE_ITEM_KEY;
363 inode = btrfs_iget(sb, &key, tree_root->fs_info->fs_root);
365 err = PTR_ERR(inode);
369 root_dentry = d_alloc_root(inode);
376 /* this does the super kobj at the same time */
377 err = btrfs_sysfs_add_super(tree_root->fs_info);
382 sb->s_root = root_dentry;
384 save_mount_options(sb, data);
388 close_ctree(tree_root);
392 int btrfs_sync_fs(struct super_block *sb, int wait)
394 struct btrfs_trans_handle *trans;
395 struct btrfs_root *root = btrfs_sb(sb);
399 filemap_flush(root->fs_info->btree_inode->i_mapping);
403 btrfs_start_delalloc_inodes(root);
404 btrfs_wait_ordered_extents(root, 0);
406 trans = btrfs_start_transaction(root, 1);
407 ret = btrfs_commit_transaction(trans, root);
411 static int btrfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
413 struct btrfs_root *root = btrfs_sb(vfs->mnt_sb);
414 struct btrfs_fs_info *info = root->fs_info;
416 if (btrfs_test_opt(root, DEGRADED))
417 seq_puts(seq, ",degraded");
418 if (btrfs_test_opt(root, NODATASUM))
419 seq_puts(seq, ",nodatasum");
420 if (btrfs_test_opt(root, NODATACOW))
421 seq_puts(seq, ",nodatacow");
422 if (btrfs_test_opt(root, NOBARRIER))
423 seq_puts(seq, ",nobarrier");
424 if (info->max_extent != (u64)-1)
425 seq_printf(seq, ",max_extent=%llu",
426 (unsigned long long)info->max_extent);
427 if (info->max_inline != 8192 * 1024)
428 seq_printf(seq, ",max_inline=%llu",
429 (unsigned long long)info->max_inline);
430 if (info->alloc_start != 0)
431 seq_printf(seq, ",alloc_start=%llu",
432 (unsigned long long)info->alloc_start);
433 if (info->thread_pool_size != min_t(unsigned long,
434 num_online_cpus() + 2, 8))
435 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
436 if (btrfs_test_opt(root, COMPRESS))
437 seq_puts(seq, ",compress");
438 if (btrfs_test_opt(root, NOSSD))
439 seq_puts(seq, ",nossd");
440 if (btrfs_test_opt(root, SSD_SPREAD))
441 seq_puts(seq, ",ssd_spread");
442 else if (btrfs_test_opt(root, SSD))
443 seq_puts(seq, ",ssd");
444 if (btrfs_test_opt(root, NOTREELOG))
445 seq_puts(seq, ",notreelog");
446 if (btrfs_test_opt(root, FLUSHONCOMMIT))
447 seq_puts(seq, ",flushoncommit");
448 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
449 seq_puts(seq, ",noacl");
453 static int btrfs_test_super(struct super_block *s, void *data)
455 struct btrfs_fs_devices *test_fs_devices = data;
456 struct btrfs_root *root = btrfs_sb(s);
458 return root->fs_info->fs_devices == test_fs_devices;
462 * Find a superblock for the given device / mount point.
464 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
465 * for multiple device setup. Make sure to keep it in sync.
467 static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
468 const char *dev_name, void *data, struct vfsmount *mnt)
470 char *subvol_name = NULL;
471 struct block_device *bdev = NULL;
472 struct super_block *s;
474 struct btrfs_fs_devices *fs_devices = NULL;
475 fmode_t mode = FMODE_READ;
478 if (!(flags & MS_RDONLY))
481 error = btrfs_parse_early_options(data, mode, fs_type,
482 &subvol_name, &fs_devices);
486 error = btrfs_scan_one_device(dev_name, mode, fs_type, &fs_devices);
488 goto error_free_subvol_name;
490 error = btrfs_open_devices(fs_devices, mode, fs_type);
492 goto error_free_subvol_name;
494 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
496 goto error_close_devices;
499 bdev = fs_devices->latest_bdev;
500 s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
505 if ((flags ^ s->s_flags) & MS_RDONLY) {
506 deactivate_locked_super(s);
508 goto error_close_devices;
511 btrfs_close_devices(fs_devices);
513 char b[BDEVNAME_SIZE];
516 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
517 error = btrfs_fill_super(s, fs_devices, data,
518 flags & MS_SILENT ? 1 : 0);
520 deactivate_locked_super(s);
521 goto error_free_subvol_name;
524 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
525 s->s_flags |= MS_ACTIVE;
528 if (!strcmp(subvol_name, "."))
529 root = dget(s->s_root);
531 mutex_lock(&s->s_root->d_inode->i_mutex);
532 root = lookup_one_len(subvol_name, s->s_root,
533 strlen(subvol_name));
534 mutex_unlock(&s->s_root->d_inode->i_mutex);
537 deactivate_locked_super(s);
538 error = PTR_ERR(root);
539 goto error_free_subvol_name;
541 if (!root->d_inode) {
543 deactivate_locked_super(s);
545 goto error_free_subvol_name;
550 mnt->mnt_root = root;
558 btrfs_close_devices(fs_devices);
559 error_free_subvol_name:
564 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
566 struct btrfs_root *root = btrfs_sb(sb);
569 ret = btrfs_parse_options(root, data);
573 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
576 if (*flags & MS_RDONLY) {
577 sb->s_flags |= MS_RDONLY;
579 ret = btrfs_commit_super(root);
582 if (root->fs_info->fs_devices->rw_devices == 0)
585 if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
588 /* recover relocation */
589 ret = btrfs_recover_relocation(root);
592 ret = btrfs_cleanup_fs_roots(root->fs_info);
595 sb->s_flags &= ~MS_RDONLY;
601 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
603 struct btrfs_root *root = btrfs_sb(dentry->d_sb);
604 struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
605 int bits = dentry->d_sb->s_blocksize_bits;
606 __be32 *fsid = (__be32 *)root->fs_info->fsid;
608 buf->f_namelen = BTRFS_NAME_LEN;
609 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
610 buf->f_bfree = buf->f_blocks -
611 (btrfs_super_bytes_used(disk_super) >> bits);
612 buf->f_bavail = buf->f_bfree;
613 buf->f_bsize = dentry->d_sb->s_blocksize;
614 buf->f_type = BTRFS_SUPER_MAGIC;
616 /* We treat it as constant endianness (it doesn't matter _which_)
617 because we want the fsid to come out the same whether mounted
618 on a big-endian or little-endian host */
619 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
620 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
621 /* Mask in the root object ID too, to disambiguate subvols */
622 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
623 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
628 static struct file_system_type btrfs_fs_type = {
629 .owner = THIS_MODULE,
631 .get_sb = btrfs_get_sb,
632 .kill_sb = kill_anon_super,
633 .fs_flags = FS_REQUIRES_DEV,
637 * used by btrfsctl to scan devices when no FS is mounted
639 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
642 struct btrfs_ioctl_vol_args *vol;
643 struct btrfs_fs_devices *fs_devices;
646 if (!capable(CAP_SYS_ADMIN))
649 vol = memdup_user((void __user *)arg, sizeof(*vol));
654 case BTRFS_IOC_SCAN_DEV:
655 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
656 &btrfs_fs_type, &fs_devices);
664 static int btrfs_freeze(struct super_block *sb)
666 struct btrfs_root *root = btrfs_sb(sb);
667 mutex_lock(&root->fs_info->transaction_kthread_mutex);
668 mutex_lock(&root->fs_info->cleaner_mutex);
672 static int btrfs_unfreeze(struct super_block *sb)
674 struct btrfs_root *root = btrfs_sb(sb);
675 mutex_unlock(&root->fs_info->cleaner_mutex);
676 mutex_unlock(&root->fs_info->transaction_kthread_mutex);
680 static const struct super_operations btrfs_super_ops = {
681 .drop_inode = btrfs_drop_inode,
682 .delete_inode = btrfs_delete_inode,
683 .put_super = btrfs_put_super,
684 .sync_fs = btrfs_sync_fs,
685 .show_options = btrfs_show_options,
686 .write_inode = btrfs_write_inode,
687 .dirty_inode = btrfs_dirty_inode,
688 .alloc_inode = btrfs_alloc_inode,
689 .destroy_inode = btrfs_destroy_inode,
690 .statfs = btrfs_statfs,
691 .remount_fs = btrfs_remount,
692 .freeze_fs = btrfs_freeze,
693 .unfreeze_fs = btrfs_unfreeze,
696 static const struct file_operations btrfs_ctl_fops = {
697 .unlocked_ioctl = btrfs_control_ioctl,
698 .compat_ioctl = btrfs_control_ioctl,
699 .owner = THIS_MODULE,
702 static struct miscdevice btrfs_misc = {
703 .minor = MISC_DYNAMIC_MINOR,
704 .name = "btrfs-control",
705 .fops = &btrfs_ctl_fops
708 static int btrfs_interface_init(void)
710 return misc_register(&btrfs_misc);
713 static void btrfs_interface_exit(void)
715 if (misc_deregister(&btrfs_misc) < 0)
716 printk(KERN_INFO "misc_deregister failed for control device");
719 static int __init init_btrfs_fs(void)
723 err = btrfs_init_sysfs();
727 err = btrfs_init_cachep();
731 err = extent_io_init();
735 err = extent_map_init();
739 err = btrfs_interface_init();
741 goto free_extent_map;
743 err = register_filesystem(&btrfs_fs_type);
745 goto unregister_ioctl;
747 printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
751 btrfs_interface_exit();
757 btrfs_destroy_cachep();
763 static void __exit exit_btrfs_fs(void)
765 btrfs_destroy_cachep();
768 btrfs_interface_exit();
769 unregister_filesystem(&btrfs_fs_type);
771 btrfs_cleanup_fs_uuids();
775 module_init(init_btrfs_fs)
776 module_exit(exit_btrfs_fs)
778 MODULE_LICENSE("GPL");