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_err,
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"},
90 u64 btrfs_parse_size(char *str)
97 res = simple_strtoul(str, &end, 10);
101 last = tolower(last);
116 * Regular mount options parser. Everything that is needed only when
117 * reading in a new superblock is parsed here.
119 int btrfs_parse_options(struct btrfs_root *root, char *options)
121 struct btrfs_fs_info *info = root->fs_info;
122 substring_t args[MAX_OPT_ARGS];
130 * strsep changes the string, duplicate it because parse_options
133 options = kstrdup(options, GFP_NOFS);
138 while ((p = strsep(&options, ",")) != NULL) {
143 token = match_token(p, tokens, args);
146 printk(KERN_INFO "btrfs: allowing degraded mounts\n");
147 btrfs_set_opt(info->mount_opt, DEGRADED);
152 * These are parsed by btrfs_parse_early_options
153 * and can be happily ignored here.
157 printk(KERN_INFO "btrfs: setting nodatacsum\n");
158 btrfs_set_opt(info->mount_opt, NODATASUM);
161 printk(KERN_INFO "btrfs: setting nodatacow\n");
162 btrfs_set_opt(info->mount_opt, NODATACOW);
163 btrfs_set_opt(info->mount_opt, NODATASUM);
166 printk(KERN_INFO "btrfs: use compression\n");
167 btrfs_set_opt(info->mount_opt, COMPRESS);
170 printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
171 btrfs_set_opt(info->mount_opt, SSD);
174 printk(KERN_INFO "btrfs: turning off barriers\n");
175 btrfs_set_opt(info->mount_opt, NOBARRIER);
177 case Opt_thread_pool:
179 match_int(&args[0], &intarg);
181 info->thread_pool_size = intarg;
182 printk(KERN_INFO "btrfs: thread pool %d\n",
183 info->thread_pool_size);
187 num = match_strdup(&args[0]);
189 info->max_extent = btrfs_parse_size(num);
192 info->max_extent = max_t(u64,
193 info->max_extent, root->sectorsize);
194 printk(KERN_INFO "btrfs: max_extent at %llu\n",
199 num = match_strdup(&args[0]);
201 info->max_inline = btrfs_parse_size(num);
204 if (info->max_inline) {
205 info->max_inline = max_t(u64,
209 printk(KERN_INFO "btrfs: max_inline at %llu\n",
213 case Opt_alloc_start:
214 num = match_strdup(&args[0]);
216 info->alloc_start = btrfs_parse_size(num);
219 "btrfs: allocations start at %llu\n",
224 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
235 * Parse mount options that are required early in the mount process.
237 * All other options will be parsed on much later in the mount process and
238 * only when we need to allocate a new super block.
240 static int btrfs_parse_early_options(const char *options, fmode_t flags,
241 void *holder, char **subvol_name,
242 struct btrfs_fs_devices **fs_devices)
244 substring_t args[MAX_OPT_ARGS];
252 * strsep changes the string, duplicate it because parse_options
255 opts = kstrdup(options, GFP_KERNEL);
259 while ((p = strsep(&opts, ",")) != NULL) {
264 token = match_token(p, tokens, args);
267 *subvol_name = match_strdup(&args[0]);
270 error = btrfs_scan_one_device(match_strdup(&args[0]),
271 flags, holder, fs_devices);
284 * If no subvolume name is specified we use the default one. Allocate
285 * a copy of the string "." here so that code later in the
286 * mount path doesn't care if it's the default volume or another one.
289 *subvol_name = kstrdup(".", GFP_KERNEL);
296 static int btrfs_fill_super(struct super_block *sb,
297 struct btrfs_fs_devices *fs_devices,
298 void *data, int silent)
301 struct dentry *root_dentry;
302 struct btrfs_super_block *disk_super;
303 struct btrfs_root *tree_root;
304 struct btrfs_inode *bi;
307 sb->s_maxbytes = MAX_LFS_FILESIZE;
308 sb->s_magic = BTRFS_SUPER_MAGIC;
309 sb->s_op = &btrfs_super_ops;
310 sb->s_export_op = &btrfs_export_ops;
311 sb->s_xattr = btrfs_xattr_handlers;
313 sb->s_flags |= MS_POSIXACL;
315 tree_root = open_ctree(sb, fs_devices, (char *)data);
317 if (IS_ERR(tree_root)) {
318 printk("btrfs: open_ctree failed\n");
319 return PTR_ERR(tree_root);
321 sb->s_fs_info = tree_root;
322 disk_super = &tree_root->fs_info->super_copy;
323 inode = btrfs_iget_locked(sb, BTRFS_FIRST_FREE_OBJECTID,
324 tree_root->fs_info->fs_root);
326 bi->location.objectid = inode->i_ino;
327 bi->location.offset = 0;
328 bi->root = tree_root->fs_info->fs_root;
330 btrfs_set_key_type(&bi->location, BTRFS_INODE_ITEM_KEY);
336 if (inode->i_state & I_NEW) {
337 btrfs_read_locked_inode(inode);
338 unlock_new_inode(inode);
341 root_dentry = d_alloc_root(inode);
348 /* this does the super kobj at the same time */
349 err = btrfs_sysfs_add_super(tree_root->fs_info);
354 sb->s_root = root_dentry;
356 save_mount_options(sb, data);
360 close_ctree(tree_root);
364 int btrfs_sync_fs(struct super_block *sb, int wait)
366 struct btrfs_trans_handle *trans;
367 struct btrfs_root *root;
371 if (sb->s_flags & MS_RDONLY)
376 filemap_flush(root->fs_info->btree_inode->i_mapping);
380 btrfs_start_delalloc_inodes(root);
381 btrfs_wait_ordered_extents(root, 0);
383 trans = btrfs_start_transaction(root, 1);
384 ret = btrfs_commit_transaction(trans, root);
389 static int btrfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
391 struct btrfs_root *root = btrfs_sb(vfs->mnt_sb);
392 struct btrfs_fs_info *info = root->fs_info;
394 if (btrfs_test_opt(root, DEGRADED))
395 seq_puts(seq, ",degraded");
396 if (btrfs_test_opt(root, NODATASUM))
397 seq_puts(seq, ",nodatasum");
398 if (btrfs_test_opt(root, NODATACOW))
399 seq_puts(seq, ",nodatacow");
400 if (btrfs_test_opt(root, NOBARRIER))
401 seq_puts(seq, ",nobarrier");
402 if (info->max_extent != (u64)-1)
403 seq_printf(seq, ",max_extent=%llu", info->max_extent);
404 if (info->max_inline != 8192 * 1024)
405 seq_printf(seq, ",max_inline=%llu", info->max_inline);
406 if (info->alloc_start != 0)
407 seq_printf(seq, ",alloc_start=%llu", info->alloc_start);
408 if (info->thread_pool_size != min_t(unsigned long,
409 num_online_cpus() + 2, 8))
410 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
411 if (btrfs_test_opt(root, COMPRESS))
412 seq_puts(seq, ",compress");
413 if (btrfs_test_opt(root, SSD))
414 seq_puts(seq, ",ssd");
415 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
416 seq_puts(seq, ",noacl");
420 static void btrfs_write_super(struct super_block *sb)
425 static int btrfs_test_super(struct super_block *s, void *data)
427 struct btrfs_fs_devices *test_fs_devices = data;
428 struct btrfs_root *root = btrfs_sb(s);
430 return root->fs_info->fs_devices == test_fs_devices;
434 * Find a superblock for the given device / mount point.
436 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
437 * for multiple device setup. Make sure to keep it in sync.
439 static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
440 const char *dev_name, void *data, struct vfsmount *mnt)
442 char *subvol_name = NULL;
443 struct block_device *bdev = NULL;
444 struct super_block *s;
446 struct btrfs_fs_devices *fs_devices = NULL;
447 fmode_t mode = FMODE_READ;
450 if (!(flags & MS_RDONLY))
453 error = btrfs_parse_early_options(data, mode, fs_type,
454 &subvol_name, &fs_devices);
458 error = btrfs_scan_one_device(dev_name, mode, fs_type, &fs_devices);
460 goto error_free_subvol_name;
462 error = btrfs_open_devices(fs_devices, mode, fs_type);
464 goto error_free_subvol_name;
466 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
468 goto error_close_devices;
471 bdev = fs_devices->latest_bdev;
472 s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
477 if ((flags ^ s->s_flags) & MS_RDONLY) {
478 up_write(&s->s_umount);
481 goto error_close_devices;
484 btrfs_close_devices(fs_devices);
486 char b[BDEVNAME_SIZE];
489 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
490 error = btrfs_fill_super(s, fs_devices, data,
491 flags & MS_SILENT ? 1 : 0);
493 up_write(&s->s_umount);
495 goto error_free_subvol_name;
498 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
499 s->s_flags |= MS_ACTIVE;
502 if (!strcmp(subvol_name, "."))
503 root = dget(s->s_root);
505 mutex_lock(&s->s_root->d_inode->i_mutex);
506 root = lookup_one_len(subvol_name, s->s_root,
507 strlen(subvol_name));
508 mutex_unlock(&s->s_root->d_inode->i_mutex);
511 up_write(&s->s_umount);
513 error = PTR_ERR(root);
514 goto error_free_subvol_name;
516 if (!root->d_inode) {
518 up_write(&s->s_umount);
521 goto error_free_subvol_name;
526 mnt->mnt_root = root;
534 btrfs_close_devices(fs_devices);
535 error_free_subvol_name:
540 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
542 struct btrfs_root *root = btrfs_sb(sb);
545 ret = btrfs_parse_options(root, data);
549 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
552 if (*flags & MS_RDONLY) {
553 sb->s_flags |= MS_RDONLY;
555 ret = btrfs_commit_super(root);
558 if (root->fs_info->fs_devices->rw_devices == 0)
561 if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
564 ret = btrfs_cleanup_reloc_trees(root);
567 ret = btrfs_cleanup_fs_roots(root->fs_info);
570 sb->s_flags &= ~MS_RDONLY;
576 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
578 struct btrfs_root *root = btrfs_sb(dentry->d_sb);
579 struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
580 int bits = dentry->d_sb->s_blocksize_bits;
581 __be32 *fsid = (__be32 *)root->fs_info->fsid;
583 buf->f_namelen = BTRFS_NAME_LEN;
584 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
585 buf->f_bfree = buf->f_blocks -
586 (btrfs_super_bytes_used(disk_super) >> bits);
587 buf->f_bavail = buf->f_bfree;
588 buf->f_bsize = dentry->d_sb->s_blocksize;
589 buf->f_type = BTRFS_SUPER_MAGIC;
591 /* We treat it as constant endianness (it doesn't matter _which_)
592 because we want the fsid to come out the same whether mounted
593 on a big-endian or little-endian host */
594 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
595 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
596 /* Mask in the root object ID too, to disambiguate subvols */
597 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
598 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
603 static struct file_system_type btrfs_fs_type = {
604 .owner = THIS_MODULE,
606 .get_sb = btrfs_get_sb,
607 .kill_sb = kill_anon_super,
608 .fs_flags = FS_REQUIRES_DEV,
612 * used by btrfsctl to scan devices when no FS is mounted
614 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
617 struct btrfs_ioctl_vol_args *vol;
618 struct btrfs_fs_devices *fs_devices;
621 if (!capable(CAP_SYS_ADMIN))
624 vol = kmalloc(sizeof(*vol), GFP_KERNEL);
628 if (copy_from_user(vol, (void __user *)arg, sizeof(*vol))) {
634 case BTRFS_IOC_SCAN_DEV:
635 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
636 &btrfs_fs_type, &fs_devices);
644 static int btrfs_freeze(struct super_block *sb)
646 struct btrfs_root *root = btrfs_sb(sb);
647 mutex_lock(&root->fs_info->transaction_kthread_mutex);
648 mutex_lock(&root->fs_info->cleaner_mutex);
652 static int btrfs_unfreeze(struct super_block *sb)
654 struct btrfs_root *root = btrfs_sb(sb);
655 mutex_unlock(&root->fs_info->cleaner_mutex);
656 mutex_unlock(&root->fs_info->transaction_kthread_mutex);
660 static struct super_operations btrfs_super_ops = {
661 .delete_inode = btrfs_delete_inode,
662 .put_super = btrfs_put_super,
663 .write_super = btrfs_write_super,
664 .sync_fs = btrfs_sync_fs,
665 .show_options = btrfs_show_options,
666 .write_inode = btrfs_write_inode,
667 .dirty_inode = btrfs_dirty_inode,
668 .alloc_inode = btrfs_alloc_inode,
669 .destroy_inode = btrfs_destroy_inode,
670 .statfs = btrfs_statfs,
671 .remount_fs = btrfs_remount,
672 .freeze_fs = btrfs_freeze,
673 .unfreeze_fs = btrfs_unfreeze,
676 static const struct file_operations btrfs_ctl_fops = {
677 .unlocked_ioctl = btrfs_control_ioctl,
678 .compat_ioctl = btrfs_control_ioctl,
679 .owner = THIS_MODULE,
682 static struct miscdevice btrfs_misc = {
683 .minor = MISC_DYNAMIC_MINOR,
684 .name = "btrfs-control",
685 .fops = &btrfs_ctl_fops
688 static int btrfs_interface_init(void)
690 return misc_register(&btrfs_misc);
693 static void btrfs_interface_exit(void)
695 if (misc_deregister(&btrfs_misc) < 0)
696 printk(KERN_INFO "misc_deregister failed for control device");
699 static int __init init_btrfs_fs(void)
703 err = btrfs_init_sysfs();
707 err = btrfs_init_cachep();
711 err = extent_io_init();
715 err = extent_map_init();
719 err = btrfs_interface_init();
721 goto free_extent_map;
723 err = register_filesystem(&btrfs_fs_type);
725 goto unregister_ioctl;
727 printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
731 btrfs_interface_exit();
737 btrfs_destroy_cachep();
743 static void __exit exit_btrfs_fs(void)
745 btrfs_destroy_cachep();
748 btrfs_interface_exit();
749 unregister_filesystem(&btrfs_fs_type);
751 btrfs_cleanup_fs_uuids();
755 module_init(init_btrfs_fs)
756 module_exit(exit_btrfs_fs)
758 MODULE_LICENSE("GPL");