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/magic.h>
44 #include "transaction.h"
45 #include "btrfs_inode.h"
47 #include "print-tree.h"
52 #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_thread_pool, Opt_noacl, Opt_compress, 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"},
89 u64 btrfs_parse_size(char *str)
96 res = simple_strtoul(str, &end, 10);
100 last = tolower(last);
115 * Regular mount options parser. Everything that is needed only when
116 * reading in a new superblock is parsed here.
118 int btrfs_parse_options(struct btrfs_root *root, char *options)
120 struct btrfs_fs_info *info = root->fs_info;
121 substring_t args[MAX_OPT_ARGS];
129 * strsep changes the string, duplicate it because parse_options
132 options = kstrdup(options, GFP_NOFS);
137 while ((p = strsep(&options, ",")) != NULL) {
142 token = match_token(p, tokens, args);
145 printk(KERN_INFO "btrfs: allowing degraded mounts\n");
146 btrfs_set_opt(info->mount_opt, DEGRADED);
151 * These are parsed by btrfs_parse_early_options
152 * and can be happily ignored here.
156 printk(KERN_INFO "btrfs: setting nodatacsum\n");
157 btrfs_set_opt(info->mount_opt, NODATASUM);
160 printk(KERN_INFO "btrfs: setting nodatacow\n");
161 btrfs_set_opt(info->mount_opt, NODATACOW);
162 btrfs_set_opt(info->mount_opt, NODATASUM);
165 printk(KERN_INFO "btrfs: use compression\n");
166 btrfs_set_opt(info->mount_opt, COMPRESS);
169 printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
170 btrfs_set_opt(info->mount_opt, SSD);
173 printk(KERN_INFO "btrfs: turning off barriers\n");
174 btrfs_set_opt(info->mount_opt, NOBARRIER);
176 case Opt_thread_pool:
178 match_int(&args[0], &intarg);
180 info->thread_pool_size = intarg;
181 printk(KERN_INFO "btrfs: thread pool %d\n",
182 info->thread_pool_size);
186 num = match_strdup(&args[0]);
188 info->max_extent = btrfs_parse_size(num);
191 info->max_extent = max_t(u64,
192 info->max_extent, root->sectorsize);
193 printk(KERN_INFO "btrfs: max_extent at %llu\n",
198 num = match_strdup(&args[0]);
200 info->max_inline = btrfs_parse_size(num);
203 if (info->max_inline) {
204 info->max_inline = max_t(u64,
208 printk(KERN_INFO "btrfs: max_inline at %llu\n",
212 case Opt_alloc_start:
213 num = match_strdup(&args[0]);
215 info->alloc_start = btrfs_parse_size(num);
218 "btrfs: allocations start at %llu\n",
223 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
234 * Parse mount options that are required early in the mount process.
236 * All other options will be parsed on much later in the mount process and
237 * only when we need to allocate a new super block.
239 static int btrfs_parse_early_options(const char *options, fmode_t flags,
240 void *holder, char **subvol_name,
241 struct btrfs_fs_devices **fs_devices)
243 substring_t args[MAX_OPT_ARGS];
251 * strsep changes the string, duplicate it because parse_options
254 opts = kstrdup(options, GFP_KERNEL);
258 while ((p = strsep(&opts, ",")) != NULL) {
263 token = match_token(p, tokens, args);
266 *subvol_name = match_strdup(&args[0]);
269 error = btrfs_scan_one_device(match_strdup(&args[0]),
270 flags, holder, fs_devices);
283 * If no subvolume name is specified we use the default one. Allocate
284 * a copy of the string "." here so that code later in the
285 * mount path doesn't care if it's the default volume or another one.
288 *subvol_name = kstrdup(".", GFP_KERNEL);
295 static int btrfs_fill_super(struct super_block *sb,
296 struct btrfs_fs_devices *fs_devices,
297 void *data, int silent)
300 struct dentry *root_dentry;
301 struct btrfs_super_block *disk_super;
302 struct btrfs_root *tree_root;
303 struct btrfs_inode *bi;
306 sb->s_maxbytes = MAX_LFS_FILESIZE;
307 sb->s_magic = BTRFS_SUPER_MAGIC;
308 sb->s_op = &btrfs_super_ops;
309 sb->s_export_op = &btrfs_export_ops;
310 sb->s_xattr = btrfs_xattr_handlers;
312 sb->s_flags |= MS_POSIXACL;
314 tree_root = open_ctree(sb, fs_devices, (char *)data);
316 if (IS_ERR(tree_root)) {
317 printk("btrfs: open_ctree failed\n");
318 return PTR_ERR(tree_root);
320 sb->s_fs_info = tree_root;
321 disk_super = &tree_root->fs_info->super_copy;
322 inode = btrfs_iget_locked(sb, BTRFS_FIRST_FREE_OBJECTID,
323 tree_root->fs_info->fs_root);
325 bi->location.objectid = inode->i_ino;
326 bi->location.offset = 0;
327 bi->root = tree_root->fs_info->fs_root;
329 btrfs_set_key_type(&bi->location, BTRFS_INODE_ITEM_KEY);
335 if (inode->i_state & I_NEW) {
336 btrfs_read_locked_inode(inode);
337 unlock_new_inode(inode);
340 root_dentry = d_alloc_root(inode);
347 /* this does the super kobj at the same time */
348 err = btrfs_sysfs_add_super(tree_root->fs_info);
353 sb->s_root = root_dentry;
355 save_mount_options(sb, data);
359 close_ctree(tree_root);
363 int btrfs_sync_fs(struct super_block *sb, int wait)
365 struct btrfs_trans_handle *trans;
366 struct btrfs_root *root;
370 if (sb->s_flags & MS_RDONLY)
375 filemap_flush(root->fs_info->btree_inode->i_mapping);
379 btrfs_start_delalloc_inodes(root);
380 btrfs_wait_ordered_extents(root, 0);
382 trans = btrfs_start_transaction(root, 1);
383 ret = btrfs_commit_transaction(trans, root);
388 static void btrfs_write_super(struct super_block *sb)
393 static int btrfs_test_super(struct super_block *s, void *data)
395 struct btrfs_fs_devices *test_fs_devices = data;
396 struct btrfs_root *root = btrfs_sb(s);
398 return root->fs_info->fs_devices == test_fs_devices;
402 * Find a superblock for the given device / mount point.
404 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
405 * for multiple device setup. Make sure to keep it in sync.
407 static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
408 const char *dev_name, void *data, struct vfsmount *mnt)
410 char *subvol_name = NULL;
411 struct block_device *bdev = NULL;
412 struct super_block *s;
414 struct btrfs_fs_devices *fs_devices = NULL;
415 fmode_t mode = FMODE_READ;
418 if (!(flags & MS_RDONLY))
421 error = btrfs_parse_early_options(data, mode, fs_type,
422 &subvol_name, &fs_devices);
426 error = btrfs_scan_one_device(dev_name, mode, fs_type, &fs_devices);
428 goto error_free_subvol_name;
430 error = btrfs_open_devices(fs_devices, mode, fs_type);
432 goto error_free_subvol_name;
434 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
436 goto error_close_devices;
439 bdev = fs_devices->latest_bdev;
440 s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
445 if ((flags ^ s->s_flags) & MS_RDONLY) {
446 up_write(&s->s_umount);
449 goto error_close_devices;
452 btrfs_close_devices(fs_devices);
454 char b[BDEVNAME_SIZE];
457 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
458 error = btrfs_fill_super(s, fs_devices, data,
459 flags & MS_SILENT ? 1 : 0);
461 up_write(&s->s_umount);
463 goto error_free_subvol_name;
466 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
467 s->s_flags |= MS_ACTIVE;
470 if (!strcmp(subvol_name, "."))
471 root = dget(s->s_root);
473 mutex_lock(&s->s_root->d_inode->i_mutex);
474 root = lookup_one_len(subvol_name, s->s_root,
475 strlen(subvol_name));
476 mutex_unlock(&s->s_root->d_inode->i_mutex);
479 up_write(&s->s_umount);
481 error = PTR_ERR(root);
482 goto error_free_subvol_name;
484 if (!root->d_inode) {
486 up_write(&s->s_umount);
489 goto error_free_subvol_name;
494 mnt->mnt_root = root;
502 btrfs_close_devices(fs_devices);
503 error_free_subvol_name:
508 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
510 struct btrfs_root *root = btrfs_sb(sb);
513 ret = btrfs_parse_options(root, data);
517 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
520 if (*flags & MS_RDONLY) {
521 sb->s_flags |= MS_RDONLY;
523 ret = btrfs_commit_super(root);
526 if (root->fs_info->fs_devices->rw_devices == 0)
529 if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
532 ret = btrfs_cleanup_reloc_trees(root);
535 ret = btrfs_cleanup_fs_roots(root->fs_info);
538 sb->s_flags &= ~MS_RDONLY;
544 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
546 struct btrfs_root *root = btrfs_sb(dentry->d_sb);
547 struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
548 int bits = dentry->d_sb->s_blocksize_bits;
549 __be32 *fsid = (__be32 *)root->fs_info->fsid;
551 buf->f_namelen = BTRFS_NAME_LEN;
552 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
553 buf->f_bfree = buf->f_blocks -
554 (btrfs_super_bytes_used(disk_super) >> bits);
555 buf->f_bavail = buf->f_bfree;
556 buf->f_bsize = dentry->d_sb->s_blocksize;
557 buf->f_type = BTRFS_SUPER_MAGIC;
559 /* We treat it as constant endianness (it doesn't matter _which_)
560 because we want the fsid to come out the same whether mounted
561 on a big-endian or little-endian host */
562 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
563 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
564 /* Mask in the root object ID too, to disambiguate subvols */
565 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
566 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
571 static struct file_system_type btrfs_fs_type = {
572 .owner = THIS_MODULE,
574 .get_sb = btrfs_get_sb,
575 .kill_sb = kill_anon_super,
576 .fs_flags = FS_REQUIRES_DEV,
580 * used by btrfsctl to scan devices when no FS is mounted
582 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
585 struct btrfs_ioctl_vol_args *vol;
586 struct btrfs_fs_devices *fs_devices;
589 if (!capable(CAP_SYS_ADMIN))
592 vol = kmalloc(sizeof(*vol), GFP_KERNEL);
596 if (copy_from_user(vol, (void __user *)arg, sizeof(*vol))) {
602 case BTRFS_IOC_SCAN_DEV:
603 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
604 &btrfs_fs_type, &fs_devices);
612 static int btrfs_freeze(struct super_block *sb)
614 struct btrfs_root *root = btrfs_sb(sb);
615 mutex_lock(&root->fs_info->transaction_kthread_mutex);
616 mutex_lock(&root->fs_info->cleaner_mutex);
620 static int btrfs_unfreeze(struct super_block *sb)
622 struct btrfs_root *root = btrfs_sb(sb);
623 mutex_unlock(&root->fs_info->cleaner_mutex);
624 mutex_unlock(&root->fs_info->transaction_kthread_mutex);
628 static struct super_operations btrfs_super_ops = {
629 .delete_inode = btrfs_delete_inode,
630 .put_super = btrfs_put_super,
631 .write_super = btrfs_write_super,
632 .sync_fs = btrfs_sync_fs,
633 .show_options = generic_show_options,
634 .write_inode = btrfs_write_inode,
635 .dirty_inode = btrfs_dirty_inode,
636 .alloc_inode = btrfs_alloc_inode,
637 .destroy_inode = btrfs_destroy_inode,
638 .statfs = btrfs_statfs,
639 .remount_fs = btrfs_remount,
640 .freeze_fs = btrfs_freeze,
641 .unfreeze_fs = btrfs_unfreeze,
644 static const struct file_operations btrfs_ctl_fops = {
645 .unlocked_ioctl = btrfs_control_ioctl,
646 .compat_ioctl = btrfs_control_ioctl,
647 .owner = THIS_MODULE,
650 static struct miscdevice btrfs_misc = {
651 .minor = MISC_DYNAMIC_MINOR,
652 .name = "btrfs-control",
653 .fops = &btrfs_ctl_fops
656 static int btrfs_interface_init(void)
658 return misc_register(&btrfs_misc);
661 static void btrfs_interface_exit(void)
663 if (misc_deregister(&btrfs_misc) < 0)
664 printk(KERN_INFO "misc_deregister failed for control device");
667 static int __init init_btrfs_fs(void)
671 err = btrfs_init_sysfs();
675 err = btrfs_init_cachep();
679 err = extent_io_init();
683 err = extent_map_init();
687 err = btrfs_interface_init();
689 goto free_extent_map;
691 err = register_filesystem(&btrfs_fs_type);
693 goto unregister_ioctl;
695 printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
699 btrfs_interface_exit();
705 btrfs_destroy_cachep();
711 static void __exit exit_btrfs_fs(void)
713 btrfs_destroy_cachep();
716 btrfs_interface_exit();
717 unregister_filesystem(&btrfs_fs_type);
719 btrfs_cleanup_fs_uuids();
723 module_init(init_btrfs_fs)
724 module_exit(exit_btrfs_fs)
726 MODULE_LICENSE("GPL");