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,
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"},
92 {Opt_discard, "discard"},
96 u64 btrfs_parse_size(char *str)
103 res = simple_strtoul(str, &end, 10);
107 last = tolower(last);
122 * Regular mount options parser. Everything that is needed only when
123 * reading in a new superblock is parsed here.
125 int btrfs_parse_options(struct btrfs_root *root, char *options)
127 struct btrfs_fs_info *info = root->fs_info;
128 substring_t args[MAX_OPT_ARGS];
137 * strsep changes the string, duplicate it because parse_options
140 options = kstrdup(options, GFP_NOFS);
145 while ((p = strsep(&options, ",")) != NULL) {
150 token = match_token(p, tokens, args);
153 printk(KERN_INFO "btrfs: allowing degraded mounts\n");
154 btrfs_set_opt(info->mount_opt, DEGRADED);
159 * These are parsed by btrfs_parse_early_options
160 * and can be happily ignored here.
164 printk(KERN_INFO "btrfs: setting nodatasum\n");
165 btrfs_set_opt(info->mount_opt, NODATASUM);
168 printk(KERN_INFO "btrfs: setting nodatacow\n");
169 btrfs_set_opt(info->mount_opt, NODATACOW);
170 btrfs_set_opt(info->mount_opt, NODATASUM);
173 printk(KERN_INFO "btrfs: use compression\n");
174 btrfs_set_opt(info->mount_opt, COMPRESS);
177 printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
178 btrfs_set_opt(info->mount_opt, SSD);
181 printk(KERN_INFO "btrfs: use spread ssd "
182 "allocation scheme\n");
183 btrfs_set_opt(info->mount_opt, SSD);
184 btrfs_set_opt(info->mount_opt, SSD_SPREAD);
187 printk(KERN_INFO "btrfs: not using ssd allocation "
189 btrfs_set_opt(info->mount_opt, NOSSD);
190 btrfs_clear_opt(info->mount_opt, SSD);
191 btrfs_clear_opt(info->mount_opt, SSD_SPREAD);
194 printk(KERN_INFO "btrfs: turning off barriers\n");
195 btrfs_set_opt(info->mount_opt, NOBARRIER);
197 case Opt_thread_pool:
199 match_int(&args[0], &intarg);
201 info->thread_pool_size = intarg;
202 printk(KERN_INFO "btrfs: thread pool %d\n",
203 info->thread_pool_size);
207 num = match_strdup(&args[0]);
209 info->max_extent = btrfs_parse_size(num);
212 info->max_extent = max_t(u64,
213 info->max_extent, root->sectorsize);
214 printk(KERN_INFO "btrfs: max_extent at %llu\n",
215 (unsigned long long)info->max_extent);
219 num = match_strdup(&args[0]);
221 info->max_inline = btrfs_parse_size(num);
224 if (info->max_inline) {
225 info->max_inline = max_t(u64,
229 printk(KERN_INFO "btrfs: max_inline at %llu\n",
230 (unsigned long long)info->max_inline);
233 case Opt_alloc_start:
234 num = match_strdup(&args[0]);
236 info->alloc_start = btrfs_parse_size(num);
239 "btrfs: allocations start at %llu\n",
240 (unsigned long long)info->alloc_start);
244 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
247 printk(KERN_INFO "btrfs: disabling tree log\n");
248 btrfs_set_opt(info->mount_opt, NOTREELOG);
250 case Opt_flushoncommit:
251 printk(KERN_INFO "btrfs: turning on flush-on-commit\n");
252 btrfs_set_opt(info->mount_opt, FLUSHONCOMMIT);
256 match_int(&args[0], &intarg);
258 info->metadata_ratio = intarg;
259 printk(KERN_INFO "btrfs: metadata ratio %d\n",
260 info->metadata_ratio);
264 btrfs_set_opt(info->mount_opt, DISCARD);
267 printk(KERN_INFO "btrfs: unrecognized mount option "
281 * Parse mount options that are required early in the mount process.
283 * All other options will be parsed on much later in the mount process and
284 * only when we need to allocate a new super block.
286 static int btrfs_parse_early_options(const char *options, fmode_t flags,
287 void *holder, char **subvol_name,
288 struct btrfs_fs_devices **fs_devices)
290 substring_t args[MAX_OPT_ARGS];
298 * strsep changes the string, duplicate it because parse_options
301 opts = kstrdup(options, GFP_KERNEL);
305 while ((p = strsep(&opts, ",")) != NULL) {
310 token = match_token(p, tokens, args);
313 *subvol_name = match_strdup(&args[0]);
316 error = btrfs_scan_one_device(match_strdup(&args[0]),
317 flags, holder, fs_devices);
330 * If no subvolume name is specified we use the default one. Allocate
331 * a copy of the string "." here so that code later in the
332 * mount path doesn't care if it's the default volume or another one.
335 *subvol_name = kstrdup(".", GFP_KERNEL);
342 static int btrfs_fill_super(struct super_block *sb,
343 struct btrfs_fs_devices *fs_devices,
344 void *data, int silent)
347 struct dentry *root_dentry;
348 struct btrfs_super_block *disk_super;
349 struct btrfs_root *tree_root;
350 struct btrfs_key key;
353 sb->s_maxbytes = MAX_LFS_FILESIZE;
354 sb->s_magic = BTRFS_SUPER_MAGIC;
355 sb->s_op = &btrfs_super_ops;
356 sb->s_export_op = &btrfs_export_ops;
357 sb->s_xattr = btrfs_xattr_handlers;
359 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
360 sb->s_flags |= MS_POSIXACL;
363 tree_root = open_ctree(sb, fs_devices, (char *)data);
365 if (IS_ERR(tree_root)) {
366 printk("btrfs: open_ctree failed\n");
367 return PTR_ERR(tree_root);
369 sb->s_fs_info = tree_root;
370 disk_super = &tree_root->fs_info->super_copy;
372 key.objectid = BTRFS_FIRST_FREE_OBJECTID;
373 key.type = BTRFS_INODE_ITEM_KEY;
375 inode = btrfs_iget(sb, &key, tree_root->fs_info->fs_root);
377 err = PTR_ERR(inode);
381 root_dentry = d_alloc_root(inode);
388 /* this does the super kobj at the same time */
389 err = btrfs_sysfs_add_super(tree_root->fs_info);
394 sb->s_root = root_dentry;
396 save_mount_options(sb, data);
400 close_ctree(tree_root);
404 int btrfs_sync_fs(struct super_block *sb, int wait)
406 struct btrfs_trans_handle *trans;
407 struct btrfs_root *root = btrfs_sb(sb);
411 filemap_flush(root->fs_info->btree_inode->i_mapping);
415 btrfs_start_delalloc_inodes(root, 0);
416 btrfs_wait_ordered_extents(root, 0, 0);
418 trans = btrfs_start_transaction(root, 1);
419 ret = btrfs_commit_transaction(trans, root);
423 static int btrfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
425 struct btrfs_root *root = btrfs_sb(vfs->mnt_sb);
426 struct btrfs_fs_info *info = root->fs_info;
428 if (btrfs_test_opt(root, DEGRADED))
429 seq_puts(seq, ",degraded");
430 if (btrfs_test_opt(root, NODATASUM))
431 seq_puts(seq, ",nodatasum");
432 if (btrfs_test_opt(root, NODATACOW))
433 seq_puts(seq, ",nodatacow");
434 if (btrfs_test_opt(root, NOBARRIER))
435 seq_puts(seq, ",nobarrier");
436 if (info->max_extent != (u64)-1)
437 seq_printf(seq, ",max_extent=%llu",
438 (unsigned long long)info->max_extent);
439 if (info->max_inline != 8192 * 1024)
440 seq_printf(seq, ",max_inline=%llu",
441 (unsigned long long)info->max_inline);
442 if (info->alloc_start != 0)
443 seq_printf(seq, ",alloc_start=%llu",
444 (unsigned long long)info->alloc_start);
445 if (info->thread_pool_size != min_t(unsigned long,
446 num_online_cpus() + 2, 8))
447 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
448 if (btrfs_test_opt(root, COMPRESS))
449 seq_puts(seq, ",compress");
450 if (btrfs_test_opt(root, NOSSD))
451 seq_puts(seq, ",nossd");
452 if (btrfs_test_opt(root, SSD_SPREAD))
453 seq_puts(seq, ",ssd_spread");
454 else if (btrfs_test_opt(root, SSD))
455 seq_puts(seq, ",ssd");
456 if (btrfs_test_opt(root, NOTREELOG))
457 seq_puts(seq, ",notreelog");
458 if (btrfs_test_opt(root, FLUSHONCOMMIT))
459 seq_puts(seq, ",flushoncommit");
460 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
461 seq_puts(seq, ",noacl");
465 static int btrfs_test_super(struct super_block *s, void *data)
467 struct btrfs_fs_devices *test_fs_devices = data;
468 struct btrfs_root *root = btrfs_sb(s);
470 return root->fs_info->fs_devices == test_fs_devices;
474 * Find a superblock for the given device / mount point.
476 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
477 * for multiple device setup. Make sure to keep it in sync.
479 static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
480 const char *dev_name, void *data, struct vfsmount *mnt)
482 char *subvol_name = NULL;
483 struct block_device *bdev = NULL;
484 struct super_block *s;
486 struct btrfs_fs_devices *fs_devices = NULL;
487 fmode_t mode = FMODE_READ;
490 if (!(flags & MS_RDONLY))
493 error = btrfs_parse_early_options(data, mode, fs_type,
494 &subvol_name, &fs_devices);
498 error = btrfs_scan_one_device(dev_name, mode, fs_type, &fs_devices);
500 goto error_free_subvol_name;
502 error = btrfs_open_devices(fs_devices, mode, fs_type);
504 goto error_free_subvol_name;
506 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
508 goto error_close_devices;
511 bdev = fs_devices->latest_bdev;
512 s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
517 if ((flags ^ s->s_flags) & MS_RDONLY) {
518 deactivate_locked_super(s);
520 goto error_close_devices;
523 btrfs_close_devices(fs_devices);
525 char b[BDEVNAME_SIZE];
528 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
529 error = btrfs_fill_super(s, fs_devices, data,
530 flags & MS_SILENT ? 1 : 0);
532 deactivate_locked_super(s);
533 goto error_free_subvol_name;
536 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
537 s->s_flags |= MS_ACTIVE;
540 if (!strcmp(subvol_name, "."))
541 root = dget(s->s_root);
543 mutex_lock(&s->s_root->d_inode->i_mutex);
544 root = lookup_one_len(subvol_name, s->s_root,
545 strlen(subvol_name));
546 mutex_unlock(&s->s_root->d_inode->i_mutex);
549 deactivate_locked_super(s);
550 error = PTR_ERR(root);
551 goto error_free_subvol_name;
553 if (!root->d_inode) {
555 deactivate_locked_super(s);
557 goto error_free_subvol_name;
562 mnt->mnt_root = root;
570 btrfs_close_devices(fs_devices);
571 error_free_subvol_name:
576 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
578 struct btrfs_root *root = btrfs_sb(sb);
581 ret = btrfs_parse_options(root, data);
585 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
588 if (*flags & MS_RDONLY) {
589 sb->s_flags |= MS_RDONLY;
591 ret = btrfs_commit_super(root);
594 if (root->fs_info->fs_devices->rw_devices == 0)
597 if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
600 /* recover relocation */
601 ret = btrfs_recover_relocation(root);
604 ret = btrfs_cleanup_fs_roots(root->fs_info);
607 sb->s_flags &= ~MS_RDONLY;
613 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
615 struct btrfs_root *root = btrfs_sb(dentry->d_sb);
616 struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
617 int bits = dentry->d_sb->s_blocksize_bits;
618 __be32 *fsid = (__be32 *)root->fs_info->fsid;
620 buf->f_namelen = BTRFS_NAME_LEN;
621 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
622 buf->f_bfree = buf->f_blocks -
623 (btrfs_super_bytes_used(disk_super) >> bits);
624 buf->f_bavail = buf->f_bfree;
625 buf->f_bsize = dentry->d_sb->s_blocksize;
626 buf->f_type = BTRFS_SUPER_MAGIC;
628 /* We treat it as constant endianness (it doesn't matter _which_)
629 because we want the fsid to come out the same whether mounted
630 on a big-endian or little-endian host */
631 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
632 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
633 /* Mask in the root object ID too, to disambiguate subvols */
634 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
635 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
640 static struct file_system_type btrfs_fs_type = {
641 .owner = THIS_MODULE,
643 .get_sb = btrfs_get_sb,
644 .kill_sb = kill_anon_super,
645 .fs_flags = FS_REQUIRES_DEV,
649 * used by btrfsctl to scan devices when no FS is mounted
651 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
654 struct btrfs_ioctl_vol_args *vol;
655 struct btrfs_fs_devices *fs_devices;
658 if (!capable(CAP_SYS_ADMIN))
661 vol = memdup_user((void __user *)arg, sizeof(*vol));
666 case BTRFS_IOC_SCAN_DEV:
667 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
668 &btrfs_fs_type, &fs_devices);
676 static int btrfs_freeze(struct super_block *sb)
678 struct btrfs_root *root = btrfs_sb(sb);
679 mutex_lock(&root->fs_info->transaction_kthread_mutex);
680 mutex_lock(&root->fs_info->cleaner_mutex);
684 static int btrfs_unfreeze(struct super_block *sb)
686 struct btrfs_root *root = btrfs_sb(sb);
687 mutex_unlock(&root->fs_info->cleaner_mutex);
688 mutex_unlock(&root->fs_info->transaction_kthread_mutex);
692 static const struct super_operations btrfs_super_ops = {
693 .drop_inode = btrfs_drop_inode,
694 .delete_inode = btrfs_delete_inode,
695 .put_super = btrfs_put_super,
696 .sync_fs = btrfs_sync_fs,
697 .show_options = btrfs_show_options,
698 .write_inode = btrfs_write_inode,
699 .dirty_inode = btrfs_dirty_inode,
700 .alloc_inode = btrfs_alloc_inode,
701 .destroy_inode = btrfs_destroy_inode,
702 .statfs = btrfs_statfs,
703 .remount_fs = btrfs_remount,
704 .freeze_fs = btrfs_freeze,
705 .unfreeze_fs = btrfs_unfreeze,
708 static const struct file_operations btrfs_ctl_fops = {
709 .unlocked_ioctl = btrfs_control_ioctl,
710 .compat_ioctl = btrfs_control_ioctl,
711 .owner = THIS_MODULE,
714 static struct miscdevice btrfs_misc = {
715 .minor = MISC_DYNAMIC_MINOR,
716 .name = "btrfs-control",
717 .fops = &btrfs_ctl_fops
720 static int btrfs_interface_init(void)
722 return misc_register(&btrfs_misc);
725 static void btrfs_interface_exit(void)
727 if (misc_deregister(&btrfs_misc) < 0)
728 printk(KERN_INFO "misc_deregister failed for control device");
731 static int __init init_btrfs_fs(void)
735 err = btrfs_init_sysfs();
739 err = btrfs_init_cachep();
743 err = extent_io_init();
747 err = extent_map_init();
751 err = btrfs_interface_init();
753 goto free_extent_map;
755 err = register_filesystem(&btrfs_fs_type);
757 goto unregister_ioctl;
759 printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
763 btrfs_interface_exit();
769 btrfs_destroy_cachep();
775 static void __exit exit_btrfs_fs(void)
777 btrfs_destroy_cachep();
780 btrfs_interface_exit();
781 unregister_filesystem(&btrfs_fs_type);
783 btrfs_cleanup_fs_uuids();
787 module_init(init_btrfs_fs)
788 module_exit(exit_btrfs_fs)
790 MODULE_LICENSE("GPL");