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,
71 Opt_ratio, Opt_flushoncommit, Opt_err,
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"},
89 {Opt_flushoncommit, "flushoncommit"},
90 {Opt_ratio, "metadata_ratio=%d"},
94 u64 btrfs_parse_size(char *str)
101 res = simple_strtoul(str, &end, 10);
105 last = tolower(last);
120 * Regular mount options parser. Everything that is needed only when
121 * reading in a new superblock is parsed here.
123 int btrfs_parse_options(struct btrfs_root *root, char *options)
125 struct btrfs_fs_info *info = root->fs_info;
126 substring_t args[MAX_OPT_ARGS];
134 * strsep changes the string, duplicate it because parse_options
137 options = kstrdup(options, GFP_NOFS);
142 while ((p = strsep(&options, ",")) != NULL) {
147 token = match_token(p, tokens, args);
150 printk(KERN_INFO "btrfs: allowing degraded mounts\n");
151 btrfs_set_opt(info->mount_opt, DEGRADED);
156 * These are parsed by btrfs_parse_early_options
157 * and can be happily ignored here.
161 printk(KERN_INFO "btrfs: setting nodatacsum\n");
162 btrfs_set_opt(info->mount_opt, NODATASUM);
165 printk(KERN_INFO "btrfs: setting nodatacow\n");
166 btrfs_set_opt(info->mount_opt, NODATACOW);
167 btrfs_set_opt(info->mount_opt, NODATASUM);
170 printk(KERN_INFO "btrfs: use compression\n");
171 btrfs_set_opt(info->mount_opt, COMPRESS);
174 printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
175 btrfs_set_opt(info->mount_opt, SSD);
178 printk(KERN_INFO "btrfs: turning off barriers\n");
179 btrfs_set_opt(info->mount_opt, NOBARRIER);
181 case Opt_thread_pool:
183 match_int(&args[0], &intarg);
185 info->thread_pool_size = intarg;
186 printk(KERN_INFO "btrfs: thread pool %d\n",
187 info->thread_pool_size);
191 num = match_strdup(&args[0]);
193 info->max_extent = btrfs_parse_size(num);
196 info->max_extent = max_t(u64,
197 info->max_extent, root->sectorsize);
198 printk(KERN_INFO "btrfs: max_extent at %llu\n",
199 (unsigned long long)info->max_extent);
203 num = match_strdup(&args[0]);
205 info->max_inline = btrfs_parse_size(num);
208 if (info->max_inline) {
209 info->max_inline = max_t(u64,
213 printk(KERN_INFO "btrfs: max_inline at %llu\n",
214 (unsigned long long)info->max_inline);
217 case Opt_alloc_start:
218 num = match_strdup(&args[0]);
220 info->alloc_start = btrfs_parse_size(num);
223 "btrfs: allocations start at %llu\n",
224 (unsigned long long)info->alloc_start);
228 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
231 printk(KERN_INFO "btrfs: disabling tree log\n");
232 btrfs_set_opt(info->mount_opt, NOTREELOG);
234 case Opt_flushoncommit:
235 printk(KERN_INFO "btrfs: turning on flush-on-commit\n");
236 btrfs_set_opt(info->mount_opt, FLUSHONCOMMIT);
240 match_int(&args[0], &intarg);
242 info->metadata_ratio = intarg;
243 printk(KERN_INFO "btrfs: metadata ratio %d\n",
244 info->metadata_ratio);
256 * Parse mount options that are required early in the mount process.
258 * All other options will be parsed on much later in the mount process and
259 * only when we need to allocate a new super block.
261 static int btrfs_parse_early_options(const char *options, fmode_t flags,
262 void *holder, char **subvol_name,
263 struct btrfs_fs_devices **fs_devices)
265 substring_t args[MAX_OPT_ARGS];
273 * strsep changes the string, duplicate it because parse_options
276 opts = kstrdup(options, GFP_KERNEL);
280 while ((p = strsep(&opts, ",")) != NULL) {
285 token = match_token(p, tokens, args);
288 *subvol_name = match_strdup(&args[0]);
291 error = btrfs_scan_one_device(match_strdup(&args[0]),
292 flags, holder, fs_devices);
305 * If no subvolume name is specified we use the default one. Allocate
306 * a copy of the string "." here so that code later in the
307 * mount path doesn't care if it's the default volume or another one.
310 *subvol_name = kstrdup(".", GFP_KERNEL);
317 static int btrfs_fill_super(struct super_block *sb,
318 struct btrfs_fs_devices *fs_devices,
319 void *data, int silent)
322 struct dentry *root_dentry;
323 struct btrfs_super_block *disk_super;
324 struct btrfs_root *tree_root;
325 struct btrfs_inode *bi;
328 sb->s_maxbytes = MAX_LFS_FILESIZE;
329 sb->s_magic = BTRFS_SUPER_MAGIC;
330 sb->s_op = &btrfs_super_ops;
331 sb->s_export_op = &btrfs_export_ops;
332 sb->s_xattr = btrfs_xattr_handlers;
334 sb->s_flags |= MS_POSIXACL;
336 tree_root = open_ctree(sb, fs_devices, (char *)data);
338 if (IS_ERR(tree_root)) {
339 printk("btrfs: open_ctree failed\n");
340 return PTR_ERR(tree_root);
342 sb->s_fs_info = tree_root;
343 disk_super = &tree_root->fs_info->super_copy;
344 inode = btrfs_iget_locked(sb, BTRFS_FIRST_FREE_OBJECTID,
345 tree_root->fs_info->fs_root);
347 bi->location.objectid = inode->i_ino;
348 bi->location.offset = 0;
349 bi->root = tree_root->fs_info->fs_root;
351 btrfs_set_key_type(&bi->location, BTRFS_INODE_ITEM_KEY);
357 if (inode->i_state & I_NEW) {
358 btrfs_read_locked_inode(inode);
359 unlock_new_inode(inode);
362 root_dentry = d_alloc_root(inode);
369 /* this does the super kobj at the same time */
370 err = btrfs_sysfs_add_super(tree_root->fs_info);
375 sb->s_root = root_dentry;
377 save_mount_options(sb, data);
381 close_ctree(tree_root);
385 int btrfs_sync_fs(struct super_block *sb, int wait)
387 struct btrfs_trans_handle *trans;
388 struct btrfs_root *root = btrfs_sb(sb);
391 if (sb->s_flags & MS_RDONLY)
396 filemap_flush(root->fs_info->btree_inode->i_mapping);
400 btrfs_start_delalloc_inodes(root);
401 btrfs_wait_ordered_extents(root, 0);
403 trans = btrfs_start_transaction(root, 1);
404 ret = btrfs_commit_transaction(trans, root);
409 static int btrfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
411 struct btrfs_root *root = btrfs_sb(vfs->mnt_sb);
412 struct btrfs_fs_info *info = root->fs_info;
414 if (btrfs_test_opt(root, DEGRADED))
415 seq_puts(seq, ",degraded");
416 if (btrfs_test_opt(root, NODATASUM))
417 seq_puts(seq, ",nodatasum");
418 if (btrfs_test_opt(root, NODATACOW))
419 seq_puts(seq, ",nodatacow");
420 if (btrfs_test_opt(root, NOBARRIER))
421 seq_puts(seq, ",nobarrier");
422 if (info->max_extent != (u64)-1)
423 seq_printf(seq, ",max_extent=%llu",
424 (unsigned long long)info->max_extent);
425 if (info->max_inline != 8192 * 1024)
426 seq_printf(seq, ",max_inline=%llu",
427 (unsigned long long)info->max_inline);
428 if (info->alloc_start != 0)
429 seq_printf(seq, ",alloc_start=%llu",
430 (unsigned long long)info->alloc_start);
431 if (info->thread_pool_size != min_t(unsigned long,
432 num_online_cpus() + 2, 8))
433 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
434 if (btrfs_test_opt(root, COMPRESS))
435 seq_puts(seq, ",compress");
436 if (btrfs_test_opt(root, SSD))
437 seq_puts(seq, ",ssd");
438 if (btrfs_test_opt(root, NOTREELOG))
439 seq_puts(seq, ",notreelog");
440 if (btrfs_test_opt(root, FLUSHONCOMMIT))
441 seq_puts(seq, ",flushoncommit");
442 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
443 seq_puts(seq, ",noacl");
447 static void btrfs_write_super(struct super_block *sb)
452 static int btrfs_test_super(struct super_block *s, void *data)
454 struct btrfs_fs_devices *test_fs_devices = data;
455 struct btrfs_root *root = btrfs_sb(s);
457 return root->fs_info->fs_devices == test_fs_devices;
461 * Find a superblock for the given device / mount point.
463 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
464 * for multiple device setup. Make sure to keep it in sync.
466 static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
467 const char *dev_name, void *data, struct vfsmount *mnt)
469 char *subvol_name = NULL;
470 struct block_device *bdev = NULL;
471 struct super_block *s;
473 struct btrfs_fs_devices *fs_devices = NULL;
474 fmode_t mode = FMODE_READ;
477 if (!(flags & MS_RDONLY))
480 error = btrfs_parse_early_options(data, mode, fs_type,
481 &subvol_name, &fs_devices);
485 error = btrfs_scan_one_device(dev_name, mode, fs_type, &fs_devices);
487 goto error_free_subvol_name;
489 error = btrfs_open_devices(fs_devices, mode, fs_type);
491 goto error_free_subvol_name;
493 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
495 goto error_close_devices;
498 bdev = fs_devices->latest_bdev;
499 s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
504 if ((flags ^ s->s_flags) & MS_RDONLY) {
505 deactivate_locked_super(s);
507 goto error_close_devices;
510 btrfs_close_devices(fs_devices);
512 char b[BDEVNAME_SIZE];
515 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
516 error = btrfs_fill_super(s, fs_devices, data,
517 flags & MS_SILENT ? 1 : 0);
519 deactivate_locked_super(s);
520 goto error_free_subvol_name;
523 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
524 s->s_flags |= MS_ACTIVE;
527 if (!strcmp(subvol_name, "."))
528 root = dget(s->s_root);
530 mutex_lock(&s->s_root->d_inode->i_mutex);
531 root = lookup_one_len(subvol_name, s->s_root,
532 strlen(subvol_name));
533 mutex_unlock(&s->s_root->d_inode->i_mutex);
536 deactivate_locked_super(s);
537 error = PTR_ERR(root);
538 goto error_free_subvol_name;
540 if (!root->d_inode) {
542 deactivate_locked_super(s);
544 goto error_free_subvol_name;
549 mnt->mnt_root = root;
557 btrfs_close_devices(fs_devices);
558 error_free_subvol_name:
563 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
565 struct btrfs_root *root = btrfs_sb(sb);
568 ret = btrfs_parse_options(root, data);
572 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
575 if (*flags & MS_RDONLY) {
576 sb->s_flags |= MS_RDONLY;
578 ret = btrfs_commit_super(root);
581 if (root->fs_info->fs_devices->rw_devices == 0)
584 if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
587 ret = btrfs_cleanup_reloc_trees(root);
590 ret = btrfs_cleanup_fs_roots(root->fs_info);
593 sb->s_flags &= ~MS_RDONLY;
599 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
601 struct btrfs_root *root = btrfs_sb(dentry->d_sb);
602 struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
603 int bits = dentry->d_sb->s_blocksize_bits;
604 __be32 *fsid = (__be32 *)root->fs_info->fsid;
606 buf->f_namelen = BTRFS_NAME_LEN;
607 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
608 buf->f_bfree = buf->f_blocks -
609 (btrfs_super_bytes_used(disk_super) >> bits);
610 buf->f_bavail = buf->f_bfree;
611 buf->f_bsize = dentry->d_sb->s_blocksize;
612 buf->f_type = BTRFS_SUPER_MAGIC;
614 /* We treat it as constant endianness (it doesn't matter _which_)
615 because we want the fsid to come out the same whether mounted
616 on a big-endian or little-endian host */
617 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
618 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
619 /* Mask in the root object ID too, to disambiguate subvols */
620 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
621 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
626 static struct file_system_type btrfs_fs_type = {
627 .owner = THIS_MODULE,
629 .get_sb = btrfs_get_sb,
630 .kill_sb = kill_anon_super,
631 .fs_flags = FS_REQUIRES_DEV,
635 * used by btrfsctl to scan devices when no FS is mounted
637 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
640 struct btrfs_ioctl_vol_args *vol;
641 struct btrfs_fs_devices *fs_devices;
644 if (!capable(CAP_SYS_ADMIN))
647 vol = memdup_user((void __user *)arg, sizeof(*vol));
652 case BTRFS_IOC_SCAN_DEV:
653 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
654 &btrfs_fs_type, &fs_devices);
662 static int btrfs_freeze(struct super_block *sb)
664 struct btrfs_root *root = btrfs_sb(sb);
665 mutex_lock(&root->fs_info->transaction_kthread_mutex);
666 mutex_lock(&root->fs_info->cleaner_mutex);
670 static int btrfs_unfreeze(struct super_block *sb)
672 struct btrfs_root *root = btrfs_sb(sb);
673 mutex_unlock(&root->fs_info->cleaner_mutex);
674 mutex_unlock(&root->fs_info->transaction_kthread_mutex);
678 static struct super_operations btrfs_super_ops = {
679 .delete_inode = btrfs_delete_inode,
680 .put_super = btrfs_put_super,
681 .write_super = btrfs_write_super,
682 .sync_fs = btrfs_sync_fs,
683 .show_options = btrfs_show_options,
684 .write_inode = btrfs_write_inode,
685 .dirty_inode = btrfs_dirty_inode,
686 .alloc_inode = btrfs_alloc_inode,
687 .destroy_inode = btrfs_destroy_inode,
688 .statfs = btrfs_statfs,
689 .remount_fs = btrfs_remount,
690 .freeze_fs = btrfs_freeze,
691 .unfreeze_fs = btrfs_unfreeze,
694 static const struct file_operations btrfs_ctl_fops = {
695 .unlocked_ioctl = btrfs_control_ioctl,
696 .compat_ioctl = btrfs_control_ioctl,
697 .owner = THIS_MODULE,
700 static struct miscdevice btrfs_misc = {
701 .minor = MISC_DYNAMIC_MINOR,
702 .name = "btrfs-control",
703 .fops = &btrfs_ctl_fops
706 static int btrfs_interface_init(void)
708 return misc_register(&btrfs_misc);
711 static void btrfs_interface_exit(void)
713 if (misc_deregister(&btrfs_misc) < 0)
714 printk(KERN_INFO "misc_deregister failed for control device");
717 static int __init init_btrfs_fs(void)
721 err = btrfs_init_sysfs();
725 err = btrfs_init_cachep();
729 err = extent_io_init();
733 err = extent_map_init();
737 err = btrfs_interface_init();
739 goto free_extent_map;
741 err = register_filesystem(&btrfs_fs_type);
743 goto unregister_ioctl;
745 printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
749 btrfs_interface_exit();
755 btrfs_destroy_cachep();
761 static void __exit exit_btrfs_fs(void)
763 btrfs_destroy_cachep();
766 btrfs_interface_exit();
767 unregister_filesystem(&btrfs_fs_type);
769 btrfs_cleanup_fs_uuids();
773 module_init(init_btrfs_fs)
774 module_exit(exit_btrfs_fs)
776 MODULE_LICENSE("GPL");