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/string.h>
28 #include <linux/smp_lock.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/version.h>
44 #include "transaction.h"
45 #include "btrfs_inode.h"
47 #include "print-tree.h"
52 #include "compression.h"
54 #define BTRFS_SUPER_MAGIC 0x9123683E
56 static struct super_operations btrfs_super_ops;
58 static void btrfs_put_super (struct super_block * sb)
60 struct btrfs_root *root = btrfs_sb(sb);
63 ret = close_ctree(root);
65 printk("close ctree returns %d\n", ret);
68 btrfs_sysfs_del_super(root->fs_info);
74 Opt_degraded, Opt_subvol, Opt_device, Opt_nodatasum, Opt_nodatacow,
75 Opt_max_extent, Opt_max_inline, Opt_alloc_start, Opt_nobarrier,
76 Opt_ssd, Opt_thread_pool, Opt_noacl, Opt_compress, Opt_err,
79 static match_table_t tokens = {
80 {Opt_degraded, "degraded"},
81 {Opt_subvol, "subvol=%s"},
82 {Opt_device, "device=%s"},
83 {Opt_nodatasum, "nodatasum"},
84 {Opt_nodatacow, "nodatacow"},
85 {Opt_nobarrier, "nobarrier"},
86 {Opt_max_extent, "max_extent=%s"},
87 {Opt_max_inline, "max_inline=%s"},
88 {Opt_alloc_start, "alloc_start=%s"},
89 {Opt_thread_pool, "thread_pool=%d"},
90 {Opt_compress, "compress"},
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];
136 * strsep changes the string, duplicate it because parse_options
139 options = kstrdup(options, GFP_NOFS);
144 while ((p = strsep(&options, ",")) != NULL) {
149 token = match_token(p, tokens, args);
152 printk(KERN_INFO "btrfs: allowing degraded mounts\n");
153 btrfs_set_opt(info->mount_opt, DEGRADED);
158 * These are parsed by btrfs_parse_early_options
159 * and can be happily ignored here.
163 printk(KERN_INFO "btrfs: setting nodatacsum\n");
164 btrfs_set_opt(info->mount_opt, NODATASUM);
167 printk(KERN_INFO "btrfs: setting nodatacow\n");
168 btrfs_set_opt(info->mount_opt, NODATACOW);
169 btrfs_set_opt(info->mount_opt, NODATASUM);
172 printk(KERN_INFO "btrfs: use compression\n");
173 btrfs_set_opt(info->mount_opt, COMPRESS);
176 printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
177 btrfs_set_opt(info->mount_opt, SSD);
180 printk(KERN_INFO "btrfs: turning off barriers\n");
181 btrfs_set_opt(info->mount_opt, NOBARRIER);
183 case Opt_thread_pool:
185 match_int(&args[0], &intarg);
187 info->thread_pool_size = intarg;
188 printk(KERN_INFO "btrfs: thread pool %d\n",
189 info->thread_pool_size);
193 num = match_strdup(&args[0]);
195 info->max_extent = btrfs_parse_size(num);
198 info->max_extent = max_t(u64,
199 info->max_extent, root->sectorsize);
200 printk(KERN_INFO "btrfs: max_extent at %llu\n",
205 num = match_strdup(&args[0]);
207 info->max_inline = btrfs_parse_size(num);
210 if (info->max_inline) {
211 info->max_inline = max_t(u64,
215 printk(KERN_INFO "btrfs: max_inline at %llu\n",
219 case Opt_alloc_start:
220 num = match_strdup(&args[0]);
222 info->alloc_start = btrfs_parse_size(num);
225 "btrfs: allocations start at %llu\n",
230 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
241 * Parse mount options that are required early in the mount process.
243 * All other options will be parsed on much later in the mount process and
244 * only when we need to allocate a new super block.
246 static int btrfs_parse_early_options(const char *options, fmode_t flags,
247 void *holder, char **subvol_name,
248 struct btrfs_fs_devices **fs_devices)
250 substring_t args[MAX_OPT_ARGS];
258 * strsep changes the string, duplicate it because parse_options
261 opts = kstrdup(options, GFP_KERNEL);
265 while ((p = strsep(&opts, ",")) != NULL) {
270 token = match_token(p, tokens, args);
273 *subvol_name = match_strdup(&args[0]);
276 error = btrfs_scan_one_device(match_strdup(&args[0]),
277 flags, holder, fs_devices);
290 * If no subvolume name is specified we use the default one. Allocate
291 * a copy of the string "." here so that code later in the
292 * mount path doesn't care if it's the default volume or another one.
295 *subvol_name = kstrdup(".", GFP_KERNEL);
302 static int btrfs_fill_super(struct super_block * sb,
303 struct btrfs_fs_devices *fs_devices,
304 void * data, int silent)
306 struct inode * inode;
307 struct dentry * root_dentry;
308 struct btrfs_super_block *disk_super;
309 struct btrfs_root *tree_root;
310 struct btrfs_inode *bi;
313 sb->s_maxbytes = MAX_LFS_FILESIZE;
314 sb->s_magic = BTRFS_SUPER_MAGIC;
315 sb->s_op = &btrfs_super_ops;
316 sb->s_export_op = &btrfs_export_ops;
317 sb->s_xattr = btrfs_xattr_handlers;
319 sb->s_flags |= MS_POSIXACL;
321 tree_root = open_ctree(sb, fs_devices, (char *)data);
323 if (IS_ERR(tree_root)) {
324 printk("btrfs: open_ctree failed\n");
325 return PTR_ERR(tree_root);
327 sb->s_fs_info = tree_root;
328 disk_super = &tree_root->fs_info->super_copy;
329 inode = btrfs_iget_locked(sb, BTRFS_FIRST_FREE_OBJECTID,
330 tree_root->fs_info->fs_root);
332 bi->location.objectid = inode->i_ino;
333 bi->location.offset = 0;
334 bi->root = tree_root->fs_info->fs_root;
336 btrfs_set_key_type(&bi->location, BTRFS_INODE_ITEM_KEY);
342 if (inode->i_state & I_NEW) {
343 btrfs_read_locked_inode(inode);
344 unlock_new_inode(inode);
347 root_dentry = d_alloc_root(inode);
354 /* this does the super kobj at the same time */
355 err = btrfs_sysfs_add_super(tree_root->fs_info);
360 sb->s_root = root_dentry;
362 save_mount_options(sb, data);
366 close_ctree(tree_root);
370 int btrfs_sync_fs(struct super_block *sb, int wait)
372 struct btrfs_trans_handle *trans;
373 struct btrfs_root *root;
377 if (sb->s_flags & MS_RDONLY)
382 filemap_flush(root->fs_info->btree_inode->i_mapping);
386 btrfs_start_delalloc_inodes(root);
387 btrfs_wait_ordered_extents(root, 0);
389 btrfs_clean_old_snapshots(root);
390 trans = btrfs_start_transaction(root, 1);
391 ret = btrfs_commit_transaction(trans, root);
396 static void btrfs_write_super(struct super_block *sb)
401 static int btrfs_test_super(struct super_block *s, void *data)
403 struct btrfs_fs_devices *test_fs_devices = data;
404 struct btrfs_root *root = btrfs_sb(s);
406 return root->fs_info->fs_devices == test_fs_devices;
410 * Find a superblock for the given device / mount point.
412 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
413 * for multiple device setup. Make sure to keep it in sync.
415 static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
416 const char *dev_name, void *data, struct vfsmount *mnt)
418 char *subvol_name = NULL;
419 struct block_device *bdev = NULL;
420 struct super_block *s;
422 struct btrfs_fs_devices *fs_devices = NULL;
423 fmode_t mode = FMODE_READ;
426 if (!(flags & MS_RDONLY))
429 error = btrfs_parse_early_options(data, mode, fs_type,
430 &subvol_name, &fs_devices);
434 error = btrfs_scan_one_device(dev_name, mode, fs_type, &fs_devices);
436 goto error_free_subvol_name;
438 error = btrfs_open_devices(fs_devices, mode, fs_type);
440 goto error_free_subvol_name;
442 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
444 goto error_close_devices;
447 bdev = fs_devices->latest_bdev;
448 s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
453 if ((flags ^ s->s_flags) & MS_RDONLY) {
454 up_write(&s->s_umount);
457 goto error_close_devices;
460 btrfs_close_devices(fs_devices);
462 char b[BDEVNAME_SIZE];
465 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
466 error = btrfs_fill_super(s, fs_devices, data,
467 flags & MS_SILENT ? 1 : 0);
469 up_write(&s->s_umount);
471 goto error_free_subvol_name;
474 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
475 s->s_flags |= MS_ACTIVE;
478 if (!strcmp(subvol_name, "."))
479 root = dget(s->s_root);
481 mutex_lock(&s->s_root->d_inode->i_mutex);
482 root = lookup_one_len(subvol_name, s->s_root, strlen(subvol_name));
483 mutex_unlock(&s->s_root->d_inode->i_mutex);
485 up_write(&s->s_umount);
487 error = PTR_ERR(root);
488 goto error_free_subvol_name;
490 if (!root->d_inode) {
492 up_write(&s->s_umount);
495 goto error_free_subvol_name;
500 mnt->mnt_root = root;
508 btrfs_close_devices(fs_devices);
509 error_free_subvol_name:
514 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
516 struct btrfs_root *root = btrfs_sb(sb);
519 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
522 if (*flags & MS_RDONLY) {
523 sb->s_flags |= MS_RDONLY;
525 ret = btrfs_commit_super(root);
528 if (root->fs_info->fs_devices->rw_devices == 0)
531 if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
534 ret = btrfs_cleanup_reloc_trees(root);
537 ret = btrfs_cleanup_fs_roots(root->fs_info);
540 sb->s_flags &= ~MS_RDONLY;
546 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
548 struct btrfs_root *root = btrfs_sb(dentry->d_sb);
549 struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
550 int bits = dentry->d_sb->s_blocksize_bits;
551 __be32 *fsid = (__be32 *)root->fs_info->fsid;
553 buf->f_namelen = BTRFS_NAME_LEN;
554 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
555 buf->f_bfree = buf->f_blocks -
556 (btrfs_super_bytes_used(disk_super) >> bits);
557 buf->f_bavail = buf->f_bfree;
558 buf->f_bsize = dentry->d_sb->s_blocksize;
559 buf->f_type = BTRFS_SUPER_MAGIC;
560 /* We treat it as constant endianness (it doesn't matter _which_)
561 because we want the fsid to come out the same whether mounted
562 on a big-endian or little-endian host */
563 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
564 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
565 /* Mask in the root object ID too, to disambiguate subvols */
566 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
567 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
572 static struct file_system_type btrfs_fs_type = {
573 .owner = THIS_MODULE,
575 .get_sb = btrfs_get_sb,
576 .kill_sb = kill_anon_super,
577 .fs_flags = FS_REQUIRES_DEV,
581 * used by btrfsctl to scan devices when no FS is mounted
583 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
586 struct btrfs_ioctl_vol_args *vol;
587 struct btrfs_fs_devices *fs_devices;
591 if (!capable(CAP_SYS_ADMIN))
594 vol = kmalloc(sizeof(*vol), GFP_KERNEL);
595 if (copy_from_user(vol, (void __user *)arg, sizeof(*vol))) {
599 len = strnlen(vol->name, BTRFS_PATH_NAME_MAX);
601 case BTRFS_IOC_SCAN_DEV:
602 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
603 &btrfs_fs_type, &fs_devices);
611 static void btrfs_write_super_lockfs(struct super_block *sb)
613 struct btrfs_root *root = btrfs_sb(sb);
614 mutex_lock(&root->fs_info->transaction_kthread_mutex);
615 mutex_lock(&root->fs_info->cleaner_mutex);
618 static void btrfs_unlockfs(struct super_block *sb)
620 struct btrfs_root *root = btrfs_sb(sb);
621 mutex_unlock(&root->fs_info->cleaner_mutex);
622 mutex_unlock(&root->fs_info->transaction_kthread_mutex);
625 static struct super_operations btrfs_super_ops = {
626 .delete_inode = btrfs_delete_inode,
627 .put_super = btrfs_put_super,
628 .write_super = btrfs_write_super,
629 .sync_fs = btrfs_sync_fs,
630 .show_options = generic_show_options,
631 .write_inode = btrfs_write_inode,
632 .dirty_inode = btrfs_dirty_inode,
633 .alloc_inode = btrfs_alloc_inode,
634 .destroy_inode = btrfs_destroy_inode,
635 .statfs = btrfs_statfs,
636 .remount_fs = btrfs_remount,
637 .write_super_lockfs = btrfs_write_super_lockfs,
638 .unlockfs = btrfs_unlockfs,
641 static const struct file_operations btrfs_ctl_fops = {
642 .unlocked_ioctl = btrfs_control_ioctl,
643 .compat_ioctl = btrfs_control_ioctl,
644 .owner = THIS_MODULE,
647 static struct miscdevice btrfs_misc = {
648 .minor = MISC_DYNAMIC_MINOR,
649 .name = "btrfs-control",
650 .fops = &btrfs_ctl_fops
653 static int btrfs_interface_init(void)
655 return misc_register(&btrfs_misc);
658 static void btrfs_interface_exit(void)
660 if (misc_deregister(&btrfs_misc) < 0)
661 printk("misc_deregister failed for control device");
664 static int __init init_btrfs_fs(void)
668 err = btrfs_init_sysfs();
672 err = btrfs_init_cachep();
676 err = extent_io_init();
680 err = extent_map_init();
684 err = btrfs_interface_init();
686 goto free_extent_map;
688 err = register_filesystem(&btrfs_fs_type);
690 goto unregister_ioctl;
692 printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
696 btrfs_interface_exit();
702 btrfs_destroy_cachep();
708 static void __exit exit_btrfs_fs(void)
710 btrfs_destroy_cachep();
713 btrfs_interface_exit();
714 unregister_filesystem(&btrfs_fs_type);
716 btrfs_cleanup_fs_uuids();
720 module_init(init_btrfs_fs)
721 module_exit(exit_btrfs_fs)
723 MODULE_LICENSE("GPL");