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"
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);
68 Opt_degraded, Opt_subvol, Opt_device, Opt_nodatasum, Opt_nodatacow,
69 Opt_max_extent, Opt_max_inline, Opt_alloc_start, Opt_nobarrier,
70 Opt_ssd, Opt_thread_pool, Opt_noacl, Opt_compress, Opt_notreelog,
74 static match_table_t tokens = {
75 {Opt_degraded, "degraded"},
76 {Opt_subvol, "subvol=%s"},
77 {Opt_device, "device=%s"},
78 {Opt_nodatasum, "nodatasum"},
79 {Opt_nodatacow, "nodatacow"},
80 {Opt_nobarrier, "nobarrier"},
81 {Opt_max_extent, "max_extent=%s"},
82 {Opt_max_inline, "max_inline=%s"},
83 {Opt_alloc_start, "alloc_start=%s"},
84 {Opt_thread_pool, "thread_pool=%d"},
85 {Opt_compress, "compress"},
88 {Opt_notreelog, "notreelog"},
92 u64 btrfs_parse_size(char *str)
99 res = simple_strtoul(str, &end, 10);
103 last = tolower(last);
118 * Regular mount options parser. Everything that is needed only when
119 * reading in a new superblock is parsed here.
121 int btrfs_parse_options(struct btrfs_root *root, char *options)
123 struct btrfs_fs_info *info = root->fs_info;
124 substring_t args[MAX_OPT_ARGS];
132 * strsep changes the string, duplicate it because parse_options
135 options = kstrdup(options, GFP_NOFS);
140 while ((p = strsep(&options, ",")) != NULL) {
145 token = match_token(p, tokens, args);
148 printk(KERN_INFO "btrfs: allowing degraded mounts\n");
149 btrfs_set_opt(info->mount_opt, DEGRADED);
154 * These are parsed by btrfs_parse_early_options
155 * and can be happily ignored here.
159 printk(KERN_INFO "btrfs: setting nodatacsum\n");
160 btrfs_set_opt(info->mount_opt, NODATASUM);
163 printk(KERN_INFO "btrfs: setting nodatacow\n");
164 btrfs_set_opt(info->mount_opt, NODATACOW);
165 btrfs_set_opt(info->mount_opt, NODATASUM);
168 printk(KERN_INFO "btrfs: use compression\n");
169 btrfs_set_opt(info->mount_opt, COMPRESS);
172 printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
173 btrfs_set_opt(info->mount_opt, SSD);
176 printk(KERN_INFO "btrfs: turning off barriers\n");
177 btrfs_set_opt(info->mount_opt, NOBARRIER);
179 case Opt_thread_pool:
181 match_int(&args[0], &intarg);
183 info->thread_pool_size = intarg;
184 printk(KERN_INFO "btrfs: thread pool %d\n",
185 info->thread_pool_size);
189 num = match_strdup(&args[0]);
191 info->max_extent = btrfs_parse_size(num);
194 info->max_extent = max_t(u64,
195 info->max_extent, root->sectorsize);
196 printk(KERN_INFO "btrfs: max_extent at %llu\n",
201 num = match_strdup(&args[0]);
203 info->max_inline = btrfs_parse_size(num);
206 if (info->max_inline) {
207 info->max_inline = max_t(u64,
211 printk(KERN_INFO "btrfs: max_inline at %llu\n",
215 case Opt_alloc_start:
216 num = match_strdup(&args[0]);
218 info->alloc_start = btrfs_parse_size(num);
221 "btrfs: allocations start at %llu\n",
226 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
229 printk(KERN_INFO "btrfs: disabling tree log\n");
230 btrfs_set_opt(info->mount_opt, NOTREELOG);
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)
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 trans = btrfs_start_transaction(root, 1);
390 ret = btrfs_commit_transaction(trans, root);
395 static int btrfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
397 struct btrfs_root *root = btrfs_sb(vfs->mnt_sb);
398 struct btrfs_fs_info *info = root->fs_info;
400 if (btrfs_test_opt(root, DEGRADED))
401 seq_puts(seq, ",degraded");
402 if (btrfs_test_opt(root, NODATASUM))
403 seq_puts(seq, ",nodatasum");
404 if (btrfs_test_opt(root, NODATACOW))
405 seq_puts(seq, ",nodatacow");
406 if (btrfs_test_opt(root, NOBARRIER))
407 seq_puts(seq, ",nobarrier");
408 if (info->max_extent != (u64)-1)
409 seq_printf(seq, ",max_extent=%llu", info->max_extent);
410 if (info->max_inline != 8192 * 1024)
411 seq_printf(seq, ",max_inline=%llu", info->max_inline);
412 if (info->alloc_start != 0)
413 seq_printf(seq, ",alloc_start=%llu", info->alloc_start);
414 if (info->thread_pool_size != min_t(unsigned long,
415 num_online_cpus() + 2, 8))
416 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
417 if (btrfs_test_opt(root, COMPRESS))
418 seq_puts(seq, ",compress");
419 if (btrfs_test_opt(root, SSD))
420 seq_puts(seq, ",ssd");
421 if (btrfs_test_opt(root, NOTREELOG))
422 seq_puts(seq, ",notreelog");
423 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
424 seq_puts(seq, ",noacl");
428 static void btrfs_write_super(struct super_block *sb)
433 static int btrfs_test_super(struct super_block *s, void *data)
435 struct btrfs_fs_devices *test_fs_devices = data;
436 struct btrfs_root *root = btrfs_sb(s);
438 return root->fs_info->fs_devices == test_fs_devices;
442 * Find a superblock for the given device / mount point.
444 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
445 * for multiple device setup. Make sure to keep it in sync.
447 static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
448 const char *dev_name, void *data, struct vfsmount *mnt)
450 char *subvol_name = NULL;
451 struct block_device *bdev = NULL;
452 struct super_block *s;
454 struct btrfs_fs_devices *fs_devices = NULL;
455 fmode_t mode = FMODE_READ;
458 if (!(flags & MS_RDONLY))
461 error = btrfs_parse_early_options(data, mode, fs_type,
462 &subvol_name, &fs_devices);
466 error = btrfs_scan_one_device(dev_name, mode, fs_type, &fs_devices);
468 goto error_free_subvol_name;
470 error = btrfs_open_devices(fs_devices, mode, fs_type);
472 goto error_free_subvol_name;
474 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
476 goto error_close_devices;
479 bdev = fs_devices->latest_bdev;
480 s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
485 if ((flags ^ s->s_flags) & MS_RDONLY) {
486 up_write(&s->s_umount);
489 goto error_close_devices;
492 btrfs_close_devices(fs_devices);
494 char b[BDEVNAME_SIZE];
497 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
498 error = btrfs_fill_super(s, fs_devices, data,
499 flags & MS_SILENT ? 1 : 0);
501 up_write(&s->s_umount);
503 goto error_free_subvol_name;
506 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
507 s->s_flags |= MS_ACTIVE;
510 if (!strcmp(subvol_name, "."))
511 root = dget(s->s_root);
513 mutex_lock(&s->s_root->d_inode->i_mutex);
514 root = lookup_one_len(subvol_name, s->s_root,
515 strlen(subvol_name));
516 mutex_unlock(&s->s_root->d_inode->i_mutex);
519 up_write(&s->s_umount);
521 error = PTR_ERR(root);
522 goto error_free_subvol_name;
524 if (!root->d_inode) {
526 up_write(&s->s_umount);
529 goto error_free_subvol_name;
534 mnt->mnt_root = root;
542 btrfs_close_devices(fs_devices);
543 error_free_subvol_name:
548 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
550 struct btrfs_root *root = btrfs_sb(sb);
553 ret = btrfs_parse_options(root, data);
557 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
560 if (*flags & MS_RDONLY) {
561 sb->s_flags |= MS_RDONLY;
563 ret = btrfs_commit_super(root);
566 if (root->fs_info->fs_devices->rw_devices == 0)
569 if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
572 ret = btrfs_cleanup_reloc_trees(root);
575 ret = btrfs_cleanup_fs_roots(root->fs_info);
578 sb->s_flags &= ~MS_RDONLY;
584 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
586 struct btrfs_root *root = btrfs_sb(dentry->d_sb);
587 struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
588 int bits = dentry->d_sb->s_blocksize_bits;
589 __be32 *fsid = (__be32 *)root->fs_info->fsid;
591 buf->f_namelen = BTRFS_NAME_LEN;
592 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
593 buf->f_bfree = buf->f_blocks -
594 (btrfs_super_bytes_used(disk_super) >> bits);
595 buf->f_bavail = buf->f_bfree;
596 buf->f_bsize = dentry->d_sb->s_blocksize;
597 buf->f_type = BTRFS_SUPER_MAGIC;
599 /* We treat it as constant endianness (it doesn't matter _which_)
600 because we want the fsid to come out the same whether mounted
601 on a big-endian or little-endian host */
602 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
603 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
604 /* Mask in the root object ID too, to disambiguate subvols */
605 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
606 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
611 static struct file_system_type btrfs_fs_type = {
612 .owner = THIS_MODULE,
614 .get_sb = btrfs_get_sb,
615 .kill_sb = kill_anon_super,
616 .fs_flags = FS_REQUIRES_DEV,
620 * used by btrfsctl to scan devices when no FS is mounted
622 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
625 struct btrfs_ioctl_vol_args *vol;
626 struct btrfs_fs_devices *fs_devices;
629 if (!capable(CAP_SYS_ADMIN))
632 vol = kmalloc(sizeof(*vol), GFP_KERNEL);
636 if (copy_from_user(vol, (void __user *)arg, sizeof(*vol))) {
642 case BTRFS_IOC_SCAN_DEV:
643 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
644 &btrfs_fs_type, &fs_devices);
652 static int btrfs_freeze(struct super_block *sb)
654 struct btrfs_root *root = btrfs_sb(sb);
655 mutex_lock(&root->fs_info->transaction_kthread_mutex);
656 mutex_lock(&root->fs_info->cleaner_mutex);
660 static int btrfs_unfreeze(struct super_block *sb)
662 struct btrfs_root *root = btrfs_sb(sb);
663 mutex_unlock(&root->fs_info->cleaner_mutex);
664 mutex_unlock(&root->fs_info->transaction_kthread_mutex);
668 static struct super_operations btrfs_super_ops = {
669 .delete_inode = btrfs_delete_inode,
670 .put_super = btrfs_put_super,
671 .write_super = btrfs_write_super,
672 .sync_fs = btrfs_sync_fs,
673 .show_options = btrfs_show_options,
674 .write_inode = btrfs_write_inode,
675 .dirty_inode = btrfs_dirty_inode,
676 .alloc_inode = btrfs_alloc_inode,
677 .destroy_inode = btrfs_destroy_inode,
678 .statfs = btrfs_statfs,
679 .remount_fs = btrfs_remount,
680 .freeze_fs = btrfs_freeze,
681 .unfreeze_fs = btrfs_unfreeze,
684 static const struct file_operations btrfs_ctl_fops = {
685 .unlocked_ioctl = btrfs_control_ioctl,
686 .compat_ioctl = btrfs_control_ioctl,
687 .owner = THIS_MODULE,
690 static struct miscdevice btrfs_misc = {
691 .minor = MISC_DYNAMIC_MINOR,
692 .name = "btrfs-control",
693 .fops = &btrfs_ctl_fops
696 static int btrfs_interface_init(void)
698 return misc_register(&btrfs_misc);
701 static void btrfs_interface_exit(void)
703 if (misc_deregister(&btrfs_misc) < 0)
704 printk(KERN_INFO "misc_deregister failed for control device");
707 static int __init init_btrfs_fs(void)
711 err = btrfs_init_sysfs();
715 err = btrfs_init_cachep();
719 err = extent_io_init();
723 err = extent_map_init();
727 err = btrfs_interface_init();
729 goto free_extent_map;
731 err = register_filesystem(&btrfs_fs_type);
733 goto unregister_ioctl;
735 printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
739 btrfs_interface_exit();
745 btrfs_destroy_cachep();
751 static void __exit exit_btrfs_fs(void)
753 btrfs_destroy_cachep();
756 btrfs_interface_exit();
757 unregister_filesystem(&btrfs_fs_type);
759 btrfs_cleanup_fs_uuids();
763 module_init(init_btrfs_fs)
764 module_exit(exit_btrfs_fs)
766 MODULE_LICENSE("GPL");