2008-06-12 03:53:53 +02:00
|
|
|
/*
|
|
|
|
* Copyright (C) 2007 Oracle. All rights reserved.
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU General Public
|
|
|
|
* License v2 as published by the Free Software Foundation.
|
|
|
|
*
|
|
|
|
* This program is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU General Public
|
|
|
|
* License along with this program; if not, write to the
|
|
|
|
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
|
|
|
|
* Boston, MA 021110-1307, USA.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/kernel.h>
|
|
|
|
#include <linux/bio.h>
|
|
|
|
#include <linux/buffer_head.h>
|
|
|
|
#include <linux/file.h>
|
|
|
|
#include <linux/fs.h>
|
2008-10-09 19:39:39 +02:00
|
|
|
#include <linux/fsnotify.h>
|
2008-06-12 03:53:53 +02:00
|
|
|
#include <linux/pagemap.h>
|
|
|
|
#include <linux/highmem.h>
|
|
|
|
#include <linux/time.h>
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/string.h>
|
|
|
|
#include <linux/smp_lock.h>
|
|
|
|
#include <linux/backing-dev.h>
|
2008-10-09 19:39:39 +02:00
|
|
|
#include <linux/mount.h>
|
2008-06-12 03:53:53 +02:00
|
|
|
#include <linux/mpage.h>
|
2008-10-09 19:39:39 +02:00
|
|
|
#include <linux/namei.h>
|
2008-06-12 03:53:53 +02:00
|
|
|
#include <linux/swap.h>
|
|
|
|
#include <linux/writeback.h>
|
|
|
|
#include <linux/statfs.h>
|
|
|
|
#include <linux/compat.h>
|
|
|
|
#include <linux/bit_spinlock.h>
|
2008-10-09 19:39:39 +02:00
|
|
|
#include <linux/security.h>
|
2008-06-12 03:53:53 +02:00
|
|
|
#include <linux/xattr.h>
|
2008-08-05 19:05:02 +02:00
|
|
|
#include <linux/vmalloc.h>
|
2008-11-20 16:22:27 +01:00
|
|
|
#include "compat.h"
|
2008-06-12 03:53:53 +02:00
|
|
|
#include "ctree.h"
|
|
|
|
#include "disk-io.h"
|
|
|
|
#include "transaction.h"
|
|
|
|
#include "btrfs_inode.h"
|
|
|
|
#include "ioctl.h"
|
|
|
|
#include "print-tree.h"
|
|
|
|
#include "volumes.h"
|
2008-06-25 22:01:30 +02:00
|
|
|
#include "locking.h"
|
2008-06-12 03:53:53 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
2008-10-09 19:39:39 +02:00
|
|
|
static noinline int create_subvol(struct btrfs_root *root,
|
|
|
|
struct dentry *dentry,
|
|
|
|
char *name, int namelen)
|
2008-06-12 03:53:53 +02:00
|
|
|
{
|
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
struct btrfs_key key;
|
|
|
|
struct btrfs_root_item root_item;
|
|
|
|
struct btrfs_inode_item *inode_item;
|
|
|
|
struct extent_buffer *leaf;
|
|
|
|
struct btrfs_root *new_root = root;
|
|
|
|
struct inode *dir;
|
|
|
|
int ret;
|
|
|
|
int err;
|
|
|
|
u64 objectid;
|
|
|
|
u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
|
2008-11-18 03:02:50 +01:00
|
|
|
u64 index = 0;
|
2008-06-12 03:53:53 +02:00
|
|
|
unsigned long nr = 1;
|
|
|
|
|
|
|
|
ret = btrfs_check_free_space(root, 1, 0);
|
|
|
|
if (ret)
|
|
|
|
goto fail_commit;
|
|
|
|
|
|
|
|
trans = btrfs_start_transaction(root, 1);
|
|
|
|
BUG_ON(!trans);
|
|
|
|
|
|
|
|
ret = btrfs_find_free_objectid(trans, root->fs_info->tree_root,
|
|
|
|
0, &objectid);
|
|
|
|
if (ret)
|
|
|
|
goto fail;
|
|
|
|
|
2008-09-23 19:14:14 +02:00
|
|
|
leaf = btrfs_alloc_free_block(trans, root, root->leafsize, 0,
|
|
|
|
objectid, trans->transid, 0, 0, 0);
|
2008-07-24 18:17:14 +02:00
|
|
|
if (IS_ERR(leaf)) {
|
|
|
|
ret = PTR_ERR(leaf);
|
|
|
|
goto fail;
|
|
|
|
}
|
2008-06-12 03:53:53 +02:00
|
|
|
|
|
|
|
btrfs_set_header_nritems(leaf, 0);
|
|
|
|
btrfs_set_header_level(leaf, 0);
|
|
|
|
btrfs_set_header_bytenr(leaf, leaf->start);
|
|
|
|
btrfs_set_header_generation(leaf, trans->transid);
|
|
|
|
btrfs_set_header_owner(leaf, objectid);
|
|
|
|
|
|
|
|
write_extent_buffer(leaf, root->fs_info->fsid,
|
|
|
|
(unsigned long)btrfs_header_fsid(leaf),
|
|
|
|
BTRFS_FSID_SIZE);
|
|
|
|
btrfs_mark_buffer_dirty(leaf);
|
|
|
|
|
|
|
|
inode_item = &root_item.inode;
|
|
|
|
memset(inode_item, 0, sizeof(*inode_item));
|
|
|
|
inode_item->generation = cpu_to_le64(1);
|
|
|
|
inode_item->size = cpu_to_le64(3);
|
|
|
|
inode_item->nlink = cpu_to_le32(1);
|
2008-10-09 17:46:29 +02:00
|
|
|
inode_item->nbytes = cpu_to_le64(root->leafsize);
|
2008-06-12 03:53:53 +02:00
|
|
|
inode_item->mode = cpu_to_le32(S_IFDIR | 0755);
|
|
|
|
|
|
|
|
btrfs_set_root_bytenr(&root_item, leaf->start);
|
2008-10-29 19:49:05 +01:00
|
|
|
btrfs_set_root_generation(&root_item, trans->transid);
|
2008-06-12 03:53:53 +02:00
|
|
|
btrfs_set_root_level(&root_item, 0);
|
|
|
|
btrfs_set_root_refs(&root_item, 1);
|
|
|
|
btrfs_set_root_used(&root_item, 0);
|
2008-10-30 19:20:02 +01:00
|
|
|
btrfs_set_root_last_snapshot(&root_item, 0);
|
2008-06-12 03:53:53 +02:00
|
|
|
|
|
|
|
memset(&root_item.drop_progress, 0, sizeof(root_item.drop_progress));
|
|
|
|
root_item.drop_level = 0;
|
|
|
|
|
2008-06-25 22:01:30 +02:00
|
|
|
btrfs_tree_unlock(leaf);
|
2008-06-12 03:53:53 +02:00
|
|
|
free_extent_buffer(leaf);
|
|
|
|
leaf = NULL;
|
|
|
|
|
|
|
|
btrfs_set_root_dirid(&root_item, new_dirid);
|
|
|
|
|
|
|
|
key.objectid = objectid;
|
|
|
|
key.offset = 1;
|
|
|
|
btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
|
|
|
|
ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key,
|
|
|
|
&root_item);
|
|
|
|
if (ret)
|
|
|
|
goto fail;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* insert the directory item
|
|
|
|
*/
|
|
|
|
key.offset = (u64)-1;
|
2008-11-18 03:02:50 +01:00
|
|
|
dir = dentry->d_parent->d_inode;
|
|
|
|
ret = btrfs_set_inode_index(dir, &index);
|
|
|
|
BUG_ON(ret);
|
|
|
|
|
|
|
|
ret = btrfs_insert_dir_item(trans, root,
|
2008-06-12 03:53:53 +02:00
|
|
|
name, namelen, dir->i_ino, &key,
|
2008-11-18 03:02:50 +01:00
|
|
|
BTRFS_FT_DIR, index);
|
2008-06-12 03:53:53 +02:00
|
|
|
if (ret)
|
|
|
|
goto fail;
|
2008-11-18 02:37:39 +01:00
|
|
|
|
2009-01-05 21:43:43 +01:00
|
|
|
btrfs_i_size_write(dir, dir->i_size + namelen * 2);
|
|
|
|
ret = btrfs_update_inode(trans, root, dir);
|
|
|
|
BUG_ON(ret);
|
|
|
|
|
2008-11-18 02:37:39 +01:00
|
|
|
/* add the backref first */
|
|
|
|
ret = btrfs_add_root_ref(trans, root->fs_info->tree_root,
|
|
|
|
objectid, BTRFS_ROOT_BACKREF_KEY,
|
|
|
|
root->root_key.objectid,
|
|
|
|
dir->i_ino, index, name, namelen);
|
|
|
|
|
|
|
|
BUG_ON(ret);
|
|
|
|
|
|
|
|
/* now add the forward ref */
|
|
|
|
ret = btrfs_add_root_ref(trans, root->fs_info->tree_root,
|
|
|
|
root->root_key.objectid, BTRFS_ROOT_REF_KEY,
|
|
|
|
objectid,
|
|
|
|
dir->i_ino, index, name, namelen);
|
|
|
|
|
|
|
|
BUG_ON(ret);
|
|
|
|
|
2008-06-12 03:53:53 +02:00
|
|
|
ret = btrfs_commit_transaction(trans, root);
|
|
|
|
if (ret)
|
|
|
|
goto fail_commit;
|
|
|
|
|
2008-11-18 03:02:50 +01:00
|
|
|
new_root = btrfs_read_fs_root_no_name(root->fs_info, &key);
|
2008-06-12 03:53:53 +02:00
|
|
|
BUG_ON(!new_root);
|
|
|
|
|
|
|
|
trans = btrfs_start_transaction(new_root, 1);
|
|
|
|
BUG_ON(!trans);
|
|
|
|
|
2008-12-11 22:30:39 +01:00
|
|
|
ret = btrfs_create_subvol_root(trans, new_root, dentry, new_dirid,
|
2008-06-12 03:53:53 +02:00
|
|
|
BTRFS_I(dir)->block_group);
|
|
|
|
if (ret)
|
|
|
|
goto fail;
|
|
|
|
|
|
|
|
fail:
|
|
|
|
nr = trans->blocks_used;
|
|
|
|
err = btrfs_commit_transaction(trans, new_root);
|
|
|
|
if (err && !ret)
|
|
|
|
ret = err;
|
|
|
|
fail_commit:
|
|
|
|
btrfs_btree_balance_dirty(root, nr);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2008-11-18 03:02:50 +01:00
|
|
|
static int create_snapshot(struct btrfs_root *root, struct dentry *dentry,
|
|
|
|
char *name, int namelen)
|
2008-06-12 03:53:53 +02:00
|
|
|
{
|
|
|
|
struct btrfs_pending_snapshot *pending_snapshot;
|
|
|
|
struct btrfs_trans_handle *trans;
|
2008-11-18 03:02:50 +01:00
|
|
|
int ret = 0;
|
2008-06-12 03:53:53 +02:00
|
|
|
int err;
|
|
|
|
unsigned long nr = 0;
|
|
|
|
|
|
|
|
if (!root->ref_cows)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
ret = btrfs_check_free_space(root, 1, 0);
|
|
|
|
if (ret)
|
|
|
|
goto fail_unlock;
|
|
|
|
|
2008-11-18 03:02:50 +01:00
|
|
|
pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS);
|
2008-06-12 03:53:53 +02:00
|
|
|
if (!pending_snapshot) {
|
|
|
|
ret = -ENOMEM;
|
|
|
|
goto fail_unlock;
|
|
|
|
}
|
|
|
|
pending_snapshot->name = kmalloc(namelen + 1, GFP_NOFS);
|
|
|
|
if (!pending_snapshot->name) {
|
|
|
|
ret = -ENOMEM;
|
|
|
|
kfree(pending_snapshot);
|
|
|
|
goto fail_unlock;
|
|
|
|
}
|
|
|
|
memcpy(pending_snapshot->name, name, namelen);
|
|
|
|
pending_snapshot->name[namelen] = '\0';
|
2008-11-18 03:02:50 +01:00
|
|
|
pending_snapshot->dentry = dentry;
|
2008-06-12 03:53:53 +02:00
|
|
|
trans = btrfs_start_transaction(root, 1);
|
|
|
|
BUG_ON(!trans);
|
|
|
|
pending_snapshot->root = root;
|
|
|
|
list_add(&pending_snapshot->list,
|
|
|
|
&trans->transaction->pending_snapshots);
|
|
|
|
err = btrfs_commit_transaction(trans, root);
|
|
|
|
|
|
|
|
fail_unlock:
|
|
|
|
btrfs_btree_balance_dirty(root, nr);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2008-10-09 19:39:39 +02:00
|
|
|
/* copy of may_create in fs/namei.c() */
|
|
|
|
static inline int btrfs_may_create(struct inode *dir, struct dentry *child)
|
|
|
|
{
|
|
|
|
if (child->d_inode)
|
|
|
|
return -EEXIST;
|
|
|
|
if (IS_DEADDIR(dir))
|
|
|
|
return -ENOENT;
|
|
|
|
return inode_permission(dir, MAY_WRITE | MAY_EXEC);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Create a new subvolume below @parent. This is largely modeled after
|
|
|
|
* sys_mkdirat and vfs_mkdir, but we only do a single component lookup
|
|
|
|
* inside this filesystem so it's quite a bit simpler.
|
|
|
|
*/
|
|
|
|
static noinline int btrfs_mksubvol(struct path *parent, char *name,
|
2008-11-18 03:02:50 +01:00
|
|
|
int mode, int namelen,
|
|
|
|
struct btrfs_root *snap_src)
|
2008-10-09 19:39:39 +02:00
|
|
|
{
|
|
|
|
struct dentry *dentry;
|
|
|
|
int error;
|
|
|
|
|
|
|
|
mutex_lock_nested(&parent->dentry->d_inode->i_mutex, I_MUTEX_PARENT);
|
|
|
|
|
|
|
|
dentry = lookup_one_len(name, parent->dentry, namelen);
|
|
|
|
error = PTR_ERR(dentry);
|
|
|
|
if (IS_ERR(dentry))
|
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
error = -EEXIST;
|
|
|
|
if (dentry->d_inode)
|
|
|
|
goto out_dput;
|
|
|
|
|
|
|
|
if (!IS_POSIXACL(parent->dentry->d_inode))
|
|
|
|
mode &= ~current->fs->umask;
|
2008-11-18 03:02:50 +01:00
|
|
|
|
2008-10-09 19:39:39 +02:00
|
|
|
error = mnt_want_write(parent->mnt);
|
|
|
|
if (error)
|
|
|
|
goto out_dput;
|
|
|
|
|
|
|
|
error = btrfs_may_create(parent->dentry->d_inode, dentry);
|
|
|
|
if (error)
|
|
|
|
goto out_drop_write;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Actually perform the low-level subvolume creation after all
|
|
|
|
* this VFS fuzz.
|
|
|
|
*
|
|
|
|
* Eventually we want to pass in an inode under which we create this
|
|
|
|
* subvolume, but for now all are under the filesystem root.
|
|
|
|
*
|
|
|
|
* Also we should pass on the mode eventually to allow creating new
|
|
|
|
* subvolume with specific mode bits.
|
|
|
|
*/
|
2008-11-18 03:02:50 +01:00
|
|
|
if (snap_src) {
|
2008-11-18 03:14:24 +01:00
|
|
|
struct dentry *dir = dentry->d_parent;
|
|
|
|
struct dentry *test = dir->d_parent;
|
|
|
|
struct btrfs_path *path = btrfs_alloc_path();
|
|
|
|
int ret;
|
|
|
|
u64 test_oid;
|
|
|
|
u64 parent_oid = BTRFS_I(dir->d_inode)->root->root_key.objectid;
|
|
|
|
|
|
|
|
test_oid = snap_src->root_key.objectid;
|
|
|
|
|
|
|
|
ret = btrfs_find_root_ref(snap_src->fs_info->tree_root,
|
|
|
|
path, parent_oid, test_oid);
|
|
|
|
if (ret == 0)
|
|
|
|
goto create;
|
|
|
|
btrfs_release_path(snap_src->fs_info->tree_root, path);
|
|
|
|
|
|
|
|
/* we need to make sure we aren't creating a directory loop
|
|
|
|
* by taking a snapshot of something that has our current
|
|
|
|
* subvol in its directory tree. So, this loops through
|
|
|
|
* the dentries and checks the forward refs for each subvolume
|
|
|
|
* to see if is references the subvolume where we are
|
|
|
|
* placing this new snapshot.
|
|
|
|
*/
|
2009-01-06 03:25:51 +01:00
|
|
|
while (1) {
|
2008-11-18 03:14:24 +01:00
|
|
|
if (!test ||
|
|
|
|
dir == snap_src->fs_info->sb->s_root ||
|
|
|
|
test == snap_src->fs_info->sb->s_root ||
|
|
|
|
test->d_inode->i_sb != snap_src->fs_info->sb) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (S_ISLNK(test->d_inode->i_mode)) {
|
2009-01-06 03:25:51 +01:00
|
|
|
printk(KERN_INFO "Btrfs symlink in snapshot "
|
|
|
|
"path, failed\n");
|
2008-11-18 03:14:24 +01:00
|
|
|
error = -EMLINK;
|
|
|
|
btrfs_free_path(path);
|
|
|
|
goto out_drop_write;
|
|
|
|
}
|
|
|
|
test_oid =
|
|
|
|
BTRFS_I(test->d_inode)->root->root_key.objectid;
|
|
|
|
ret = btrfs_find_root_ref(snap_src->fs_info->tree_root,
|
|
|
|
path, test_oid, parent_oid);
|
|
|
|
if (ret == 0) {
|
2009-01-06 03:25:51 +01:00
|
|
|
printk(KERN_INFO "Btrfs snapshot creation "
|
|
|
|
"failed, looping\n");
|
2008-11-18 03:14:24 +01:00
|
|
|
error = -EMLINK;
|
|
|
|
btrfs_free_path(path);
|
|
|
|
goto out_drop_write;
|
|
|
|
}
|
|
|
|
btrfs_release_path(snap_src->fs_info->tree_root, path);
|
|
|
|
test = test->d_parent;
|
|
|
|
}
|
|
|
|
create:
|
|
|
|
btrfs_free_path(path);
|
2008-11-18 03:02:50 +01:00
|
|
|
error = create_snapshot(snap_src, dentry, name, namelen);
|
|
|
|
} else {
|
|
|
|
error = create_subvol(BTRFS_I(parent->dentry->d_inode)->root,
|
|
|
|
dentry, name, namelen);
|
|
|
|
}
|
2008-10-09 19:39:39 +02:00
|
|
|
if (error)
|
|
|
|
goto out_drop_write;
|
|
|
|
|
|
|
|
fsnotify_mkdir(parent->dentry->d_inode, dentry);
|
|
|
|
out_drop_write:
|
|
|
|
mnt_drop_write(parent->mnt);
|
|
|
|
out_dput:
|
|
|
|
dput(dentry);
|
|
|
|
out_unlock:
|
|
|
|
mutex_unlock(&parent->dentry->d_inode->i_mutex);
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2008-12-02 15:54:17 +01:00
|
|
|
static int btrfs_defrag_file(struct file *file)
|
2008-06-12 03:53:53 +02:00
|
|
|
{
|
|
|
|
struct inode *inode = fdentry(file)->d_inode;
|
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
|
|
|
struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
|
2008-07-24 17:57:52 +02:00
|
|
|
struct btrfs_ordered_extent *ordered;
|
2008-06-12 03:53:53 +02:00
|
|
|
struct page *page;
|
|
|
|
unsigned long last_index;
|
|
|
|
unsigned long ra_pages = root->fs_info->bdi.ra_pages;
|
|
|
|
unsigned long total_read = 0;
|
|
|
|
u64 page_start;
|
|
|
|
u64 page_end;
|
|
|
|
unsigned long i;
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
ret = btrfs_check_free_space(root, inode->i_size, 0);
|
|
|
|
if (ret)
|
|
|
|
return -ENOSPC;
|
|
|
|
|
|
|
|
mutex_lock(&inode->i_mutex);
|
|
|
|
last_index = inode->i_size >> PAGE_CACHE_SHIFT;
|
|
|
|
for (i = 0; i <= last_index; i++) {
|
|
|
|
if (total_read % ra_pages == 0) {
|
|
|
|
btrfs_force_ra(inode->i_mapping, &file->f_ra, file, i,
|
|
|
|
min(last_index, i + ra_pages - 1));
|
|
|
|
}
|
|
|
|
total_read++;
|
2008-07-24 17:57:52 +02:00
|
|
|
again:
|
2008-06-12 03:53:53 +02:00
|
|
|
page = grab_cache_page(inode->i_mapping, i);
|
|
|
|
if (!page)
|
|
|
|
goto out_unlock;
|
|
|
|
if (!PageUptodate(page)) {
|
|
|
|
btrfs_readpage(NULL, page);
|
|
|
|
lock_page(page);
|
|
|
|
if (!PageUptodate(page)) {
|
|
|
|
unlock_page(page);
|
|
|
|
page_cache_release(page);
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
wait_on_page_writeback(page);
|
|
|
|
|
|
|
|
page_start = (u64)page->index << PAGE_CACHE_SHIFT;
|
|
|
|
page_end = page_start + PAGE_CACHE_SIZE - 1;
|
|
|
|
lock_extent(io_tree, page_start, page_end, GFP_NOFS);
|
2008-07-24 17:57:52 +02:00
|
|
|
|
|
|
|
ordered = btrfs_lookup_ordered_extent(inode, page_start);
|
|
|
|
if (ordered) {
|
|
|
|
unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
|
|
|
|
unlock_page(page);
|
|
|
|
page_cache_release(page);
|
|
|
|
btrfs_start_ordered_extent(inode, ordered, 1);
|
|
|
|
btrfs_put_ordered_extent(ordered);
|
|
|
|
goto again;
|
|
|
|
}
|
|
|
|
set_page_extent_mapped(page);
|
|
|
|
|
2008-08-01 17:27:23 +02:00
|
|
|
/*
|
|
|
|
* this makes sure page_mkwrite is called on the
|
|
|
|
* page if it is dirtied again later
|
|
|
|
*/
|
|
|
|
clear_page_dirty_for_io(page);
|
|
|
|
|
2008-08-05 05:17:27 +02:00
|
|
|
btrfs_set_extent_delalloc(inode, page_start, page_end);
|
2008-06-12 03:53:53 +02:00
|
|
|
|
|
|
|
unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
|
|
|
|
set_page_dirty(page);
|
|
|
|
unlock_page(page);
|
|
|
|
page_cache_release(page);
|
|
|
|
balance_dirty_pages_ratelimited_nr(inode->i_mapping, 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
out_unlock:
|
|
|
|
mutex_unlock(&inode->i_mutex);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Called inside transaction, so use GFP_NOFS
|
|
|
|
*/
|
|
|
|
|
|
|
|
static int btrfs_ioctl_resize(struct btrfs_root *root, void __user *arg)
|
|
|
|
{
|
|
|
|
u64 new_size;
|
|
|
|
u64 old_size;
|
|
|
|
u64 devid = 1;
|
|
|
|
struct btrfs_ioctl_vol_args *vol_args;
|
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
struct btrfs_device *device = NULL;
|
|
|
|
char *sizestr;
|
|
|
|
char *devstr = NULL;
|
|
|
|
int ret = 0;
|
|
|
|
int namelen;
|
|
|
|
int mod = 0;
|
|
|
|
|
2008-11-12 20:34:12 +01:00
|
|
|
if (root->fs_info->sb->s_flags & MS_RDONLY)
|
|
|
|
return -EROFS;
|
|
|
|
|
2009-01-05 22:57:23 +01:00
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
return -EPERM;
|
|
|
|
|
2008-06-12 03:53:53 +02:00
|
|
|
vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
|
|
|
|
|
|
|
|
if (!vol_args)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
|
|
|
|
ret = -EFAULT;
|
|
|
|
goto out;
|
|
|
|
}
|
2008-07-24 18:20:14 +02:00
|
|
|
|
|
|
|
vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
|
2008-06-12 03:53:53 +02:00
|
|
|
namelen = strlen(vol_args->name);
|
|
|
|
|
2008-07-08 20:19:17 +02:00
|
|
|
mutex_lock(&root->fs_info->volume_mutex);
|
2008-06-12 03:53:53 +02:00
|
|
|
sizestr = vol_args->name;
|
|
|
|
devstr = strchr(sizestr, ':');
|
|
|
|
if (devstr) {
|
|
|
|
char *end;
|
|
|
|
sizestr = devstr + 1;
|
|
|
|
*devstr = '\0';
|
|
|
|
devstr = vol_args->name;
|
|
|
|
devid = simple_strtoull(devstr, &end, 10);
|
|
|
|
printk(KERN_INFO "resizing devid %llu\n", devid);
|
|
|
|
}
|
2008-11-18 03:11:30 +01:00
|
|
|
device = btrfs_find_device(root, devid, NULL, NULL);
|
2008-06-12 03:53:53 +02:00
|
|
|
if (!device) {
|
|
|
|
printk(KERN_INFO "resizer unable to find device %llu\n", devid);
|
|
|
|
ret = -EINVAL;
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
|
|
|
if (!strcmp(sizestr, "max"))
|
|
|
|
new_size = device->bdev->bd_inode->i_size;
|
|
|
|
else {
|
|
|
|
if (sizestr[0] == '-') {
|
|
|
|
mod = -1;
|
|
|
|
sizestr++;
|
|
|
|
} else if (sizestr[0] == '+') {
|
|
|
|
mod = 1;
|
|
|
|
sizestr++;
|
|
|
|
}
|
|
|
|
new_size = btrfs_parse_size(sizestr);
|
|
|
|
if (new_size == 0) {
|
|
|
|
ret = -EINVAL;
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
old_size = device->total_bytes;
|
|
|
|
|
|
|
|
if (mod < 0) {
|
|
|
|
if (new_size > old_size) {
|
|
|
|
ret = -EINVAL;
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
|
|
|
new_size = old_size - new_size;
|
|
|
|
} else if (mod > 0) {
|
|
|
|
new_size = old_size + new_size;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (new_size < 256 * 1024 * 1024) {
|
|
|
|
ret = -EINVAL;
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
|
|
|
if (new_size > device->bdev->bd_inode->i_size) {
|
|
|
|
ret = -EFBIG;
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
|
|
|
|
|
|
|
do_div(new_size, root->sectorsize);
|
|
|
|
new_size *= root->sectorsize;
|
|
|
|
|
|
|
|
printk(KERN_INFO "new size for %s is %llu\n",
|
|
|
|
device->name, (unsigned long long)new_size);
|
|
|
|
|
|
|
|
if (new_size > old_size) {
|
|
|
|
trans = btrfs_start_transaction(root, 1);
|
|
|
|
ret = btrfs_grow_device(trans, device, new_size);
|
|
|
|
btrfs_commit_transaction(trans, root);
|
|
|
|
} else {
|
|
|
|
ret = btrfs_shrink_device(device, new_size);
|
|
|
|
}
|
|
|
|
|
|
|
|
out_unlock:
|
2008-07-08 20:19:17 +02:00
|
|
|
mutex_unlock(&root->fs_info->volume_mutex);
|
2008-06-12 03:53:53 +02:00
|
|
|
out:
|
|
|
|
kfree(vol_args);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2008-10-09 19:39:39 +02:00
|
|
|
static noinline int btrfs_ioctl_snap_create(struct file *file,
|
2008-11-18 03:02:50 +01:00
|
|
|
void __user *arg, int subvol)
|
2008-06-12 03:53:53 +02:00
|
|
|
{
|
2008-10-09 19:39:39 +02:00
|
|
|
struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
|
2008-06-12 03:53:53 +02:00
|
|
|
struct btrfs_ioctl_vol_args *vol_args;
|
|
|
|
struct btrfs_dir_item *di;
|
|
|
|
struct btrfs_path *path;
|
2008-11-18 03:02:50 +01:00
|
|
|
struct file *src_file;
|
2008-06-12 03:53:53 +02:00
|
|
|
u64 root_dirid;
|
|
|
|
int namelen;
|
2008-11-18 03:02:50 +01:00
|
|
|
int ret = 0;
|
2008-06-12 03:53:53 +02:00
|
|
|
|
2008-11-12 20:34:12 +01:00
|
|
|
if (root->fs_info->sb->s_flags & MS_RDONLY)
|
|
|
|
return -EROFS;
|
|
|
|
|
2008-06-12 03:53:53 +02:00
|
|
|
vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
|
|
|
|
|
|
|
|
if (!vol_args)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
|
|
|
|
ret = -EFAULT;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2008-07-24 18:20:14 +02:00
|
|
|
vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
|
2008-06-12 03:53:53 +02:00
|
|
|
namelen = strlen(vol_args->name);
|
|
|
|
if (strchr(vol_args->name, '/')) {
|
|
|
|
ret = -EINVAL;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
path = btrfs_alloc_path();
|
|
|
|
if (!path) {
|
|
|
|
ret = -ENOMEM;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
root_dirid = root->fs_info->sb->s_root->d_inode->i_ino,
|
|
|
|
di = btrfs_lookup_dir_item(NULL, root->fs_info->tree_root,
|
|
|
|
path, root_dirid,
|
|
|
|
vol_args->name, namelen, 0);
|
|
|
|
btrfs_free_path(path);
|
|
|
|
|
|
|
|
if (di && !IS_ERR(di)) {
|
|
|
|
ret = -EEXIST;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (IS_ERR(di)) {
|
|
|
|
ret = PTR_ERR(di);
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2008-11-18 03:02:50 +01:00
|
|
|
if (subvol) {
|
2008-10-09 19:39:39 +02:00
|
|
|
ret = btrfs_mksubvol(&file->f_path, vol_args->name,
|
|
|
|
file->f_path.dentry->d_inode->i_mode,
|
2008-11-18 03:02:50 +01:00
|
|
|
namelen, NULL);
|
2008-10-09 19:39:39 +02:00
|
|
|
} else {
|
2008-11-18 03:02:50 +01:00
|
|
|
struct inode *src_inode;
|
|
|
|
src_file = fget(vol_args->fd);
|
|
|
|
if (!src_file) {
|
|
|
|
ret = -EINVAL;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
src_inode = src_file->f_path.dentry->d_inode;
|
|
|
|
if (src_inode->i_sb != file->f_path.dentry->d_inode->i_sb) {
|
2009-01-06 03:25:51 +01:00
|
|
|
printk(KERN_INFO "btrfs: Snapshot src from "
|
|
|
|
"another FS\n");
|
2008-11-18 03:02:50 +01:00
|
|
|
ret = -EINVAL;
|
|
|
|
fput(src_file);
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
ret = btrfs_mksubvol(&file->f_path, vol_args->name,
|
|
|
|
file->f_path.dentry->d_inode->i_mode,
|
|
|
|
namelen, BTRFS_I(src_inode)->root);
|
|
|
|
fput(src_file);
|
2008-10-09 19:39:39 +02:00
|
|
|
}
|
|
|
|
|
2008-06-12 03:53:53 +02:00
|
|
|
out:
|
|
|
|
kfree(vol_args);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int btrfs_ioctl_defrag(struct file *file)
|
|
|
|
{
|
|
|
|
struct inode *inode = fdentry(file)->d_inode;
|
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
2008-11-12 20:34:12 +01:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
ret = mnt_want_write(file->f_path.mnt);
|
|
|
|
if (ret)
|
|
|
|
return ret;
|
2008-06-12 03:53:53 +02:00
|
|
|
|
|
|
|
switch (inode->i_mode & S_IFMT) {
|
|
|
|
case S_IFDIR:
|
2009-01-05 22:57:23 +01:00
|
|
|
if (!capable(CAP_SYS_ADMIN)) {
|
|
|
|
ret = -EPERM;
|
|
|
|
goto out;
|
|
|
|
}
|
2008-06-12 03:53:53 +02:00
|
|
|
btrfs_defrag_root(root, 0);
|
|
|
|
btrfs_defrag_root(root->fs_info->extent_root, 0);
|
|
|
|
break;
|
|
|
|
case S_IFREG:
|
2009-01-05 22:57:23 +01:00
|
|
|
if (!(file->f_mode & FMODE_WRITE)) {
|
|
|
|
ret = -EINVAL;
|
|
|
|
goto out;
|
|
|
|
}
|
2008-06-12 03:53:53 +02:00
|
|
|
btrfs_defrag_file(file);
|
|
|
|
break;
|
|
|
|
}
|
2009-01-05 22:57:23 +01:00
|
|
|
out:
|
2008-12-19 16:58:39 +01:00
|
|
|
mnt_drop_write(file->f_path.mnt);
|
2009-01-05 22:57:23 +01:00
|
|
|
return ret;
|
2008-06-12 03:53:53 +02:00
|
|
|
}
|
|
|
|
|
2008-12-02 15:54:17 +01:00
|
|
|
static long btrfs_ioctl_add_dev(struct btrfs_root *root, void __user *arg)
|
2008-06-12 03:53:53 +02:00
|
|
|
{
|
|
|
|
struct btrfs_ioctl_vol_args *vol_args;
|
|
|
|
int ret;
|
|
|
|
|
2009-01-05 22:57:23 +01:00
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
return -EPERM;
|
|
|
|
|
2008-06-12 03:53:53 +02:00
|
|
|
vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
|
|
|
|
|
|
|
|
if (!vol_args)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
|
|
|
|
ret = -EFAULT;
|
|
|
|
goto out;
|
|
|
|
}
|
2008-07-24 18:20:14 +02:00
|
|
|
vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
|
2008-06-12 03:53:53 +02:00
|
|
|
ret = btrfs_init_new_device(root, vol_args->name);
|
|
|
|
|
|
|
|
out:
|
|
|
|
kfree(vol_args);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2008-12-02 15:54:17 +01:00
|
|
|
static long btrfs_ioctl_rm_dev(struct btrfs_root *root, void __user *arg)
|
2008-06-12 03:53:53 +02:00
|
|
|
{
|
|
|
|
struct btrfs_ioctl_vol_args *vol_args;
|
|
|
|
int ret;
|
|
|
|
|
2009-01-05 22:57:23 +01:00
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
return -EPERM;
|
|
|
|
|
2008-11-12 20:34:12 +01:00
|
|
|
if (root->fs_info->sb->s_flags & MS_RDONLY)
|
|
|
|
return -EROFS;
|
|
|
|
|
2008-06-12 03:53:53 +02:00
|
|
|
vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
|
|
|
|
|
|
|
|
if (!vol_args)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
|
|
|
|
ret = -EFAULT;
|
|
|
|
goto out;
|
|
|
|
}
|
2008-07-24 18:20:14 +02:00
|
|
|
vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
|
2008-06-12 03:53:53 +02:00
|
|
|
ret = btrfs_rm_device(root, vol_args->name);
|
|
|
|
|
|
|
|
out:
|
|
|
|
kfree(vol_args);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2008-12-02 15:54:17 +01:00
|
|
|
static long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
|
|
|
|
u64 off, u64 olen, u64 destoff)
|
2008-06-12 03:53:53 +02:00
|
|
|
{
|
|
|
|
struct inode *inode = fdentry(file)->d_inode;
|
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
|
|
|
struct file *src_file;
|
|
|
|
struct inode *src;
|
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
struct btrfs_path *path;
|
|
|
|
struct extent_buffer *leaf;
|
2008-08-05 05:23:47 +02:00
|
|
|
char *buf;
|
|
|
|
struct btrfs_key key;
|
2008-06-12 03:53:53 +02:00
|
|
|
u32 nritems;
|
|
|
|
int slot;
|
2008-08-05 05:23:47 +02:00
|
|
|
int ret;
|
2008-11-12 20:32:25 +01:00
|
|
|
u64 len = olen;
|
|
|
|
u64 bs = root->fs_info->sb->s_blocksize;
|
|
|
|
u64 hint_byte;
|
Btrfs: move data checksumming into a dedicated tree
Btrfs stores checksums for each data block. Until now, they have
been stored in the subvolume trees, indexed by the inode that is
referencing the data block. This means that when we read the inode,
we've probably read in at least some checksums as well.
But, this has a few problems:
* The checksums are indexed by logical offset in the file. When
compression is on, this means we have to do the expensive checksumming
on the uncompressed data. It would be faster if we could checksum
the compressed data instead.
* If we implement encryption, we'll be checksumming the plain text and
storing that on disk. This is significantly less secure.
* For either compression or encryption, we have to get the plain text
back before we can verify the checksum as correct. This makes the raid
layer balancing and extent moving much more expensive.
* It makes the front end caching code more complex, as we have touch
the subvolume and inodes as we cache extents.
* There is potentitally one copy of the checksum in each subvolume
referencing an extent.
The solution used here is to store the extent checksums in a dedicated
tree. This allows us to index the checksums by phyiscal extent
start and length. It means:
* The checksum is against the data stored on disk, after any compression
or encryption is done.
* The checksum is stored in a central location, and can be verified without
following back references, or reading inodes.
This makes compression significantly faster by reducing the amount of
data that needs to be checksummed. It will also allow much faster
raid management code in general.
The checksums are indexed by a key with a fixed objectid (a magic value
in ctree.h) and offset set to the starting byte of the extent. This
allows us to copy the checksum items into the fsync log tree directly (or
any other tree), without having to invent a second format for them.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-12-08 22:58:54 +01:00
|
|
|
|
2008-11-12 20:32:25 +01:00
|
|
|
/*
|
|
|
|
* TODO:
|
|
|
|
* - split compressed inline extents. annoying: we need to
|
|
|
|
* decompress into destination's address_space (the file offset
|
|
|
|
* may change, so source mapping won't do), then recompress (or
|
|
|
|
* otherwise reinsert) a subrange.
|
|
|
|
* - allow ranges within the same file to be cloned (provided
|
|
|
|
* they don't overlap)?
|
|
|
|
*/
|
|
|
|
|
2009-01-05 22:57:23 +01:00
|
|
|
/* the destination must be opened for writing */
|
|
|
|
if (!(file->f_mode & FMODE_WRITE))
|
|
|
|
return -EINVAL;
|
|
|
|
|
2008-11-12 20:34:12 +01:00
|
|
|
ret = mnt_want_write(file->f_path.mnt);
|
|
|
|
if (ret)
|
|
|
|
return ret;
|
|
|
|
|
2008-11-12 20:32:25 +01:00
|
|
|
src_file = fget(srcfd);
|
2008-12-19 16:58:39 +01:00
|
|
|
if (!src_file) {
|
|
|
|
ret = -EBADF;
|
|
|
|
goto out_drop_write;
|
|
|
|
}
|
2008-06-12 03:53:53 +02:00
|
|
|
src = src_file->f_dentry->d_inode;
|
|
|
|
|
2008-11-12 20:32:25 +01:00
|
|
|
ret = -EINVAL;
|
|
|
|
if (src == inode)
|
|
|
|
goto out_fput;
|
|
|
|
|
2008-08-05 05:23:47 +02:00
|
|
|
ret = -EISDIR;
|
|
|
|
if (S_ISDIR(src->i_mode) || S_ISDIR(inode->i_mode))
|
|
|
|
goto out_fput;
|
|
|
|
|
2008-06-12 03:53:53 +02:00
|
|
|
ret = -EXDEV;
|
2008-08-05 05:23:47 +02:00
|
|
|
if (src->i_sb != inode->i_sb || BTRFS_I(src)->root != root)
|
|
|
|
goto out_fput;
|
|
|
|
|
|
|
|
ret = -ENOMEM;
|
|
|
|
buf = vmalloc(btrfs_level_size(root, 0));
|
|
|
|
if (!buf)
|
|
|
|
goto out_fput;
|
|
|
|
|
|
|
|
path = btrfs_alloc_path();
|
|
|
|
if (!path) {
|
|
|
|
vfree(buf);
|
2008-06-12 03:53:53 +02:00
|
|
|
goto out_fput;
|
2008-08-05 05:23:47 +02:00
|
|
|
}
|
|
|
|
path->reada = 2;
|
2008-06-12 03:53:53 +02:00
|
|
|
|
|
|
|
if (inode < src) {
|
|
|
|
mutex_lock(&inode->i_mutex);
|
|
|
|
mutex_lock(&src->i_mutex);
|
|
|
|
} else {
|
|
|
|
mutex_lock(&src->i_mutex);
|
|
|
|
mutex_lock(&inode->i_mutex);
|
|
|
|
}
|
|
|
|
|
2008-11-12 20:32:25 +01:00
|
|
|
/* determine range to clone */
|
|
|
|
ret = -EINVAL;
|
|
|
|
if (off >= src->i_size || off + len > src->i_size)
|
2008-06-12 03:53:53 +02:00
|
|
|
goto out_unlock;
|
2008-11-12 20:32:25 +01:00
|
|
|
if (len == 0)
|
|
|
|
olen = len = src->i_size - off;
|
|
|
|
/* if we extend to eof, continue to block boundary */
|
|
|
|
if (off + len == src->i_size)
|
|
|
|
len = ((src->i_size + bs-1) & ~(bs-1))
|
|
|
|
- off;
|
|
|
|
|
|
|
|
/* verify the end result is block aligned */
|
|
|
|
if ((off & (bs-1)) ||
|
|
|
|
((off + len) & (bs-1)))
|
|
|
|
goto out_unlock;
|
|
|
|
|
2008-06-12 03:53:53 +02:00
|
|
|
/* do any pending delalloc/csum calc on src, one way or
|
|
|
|
another, and lock file content */
|
|
|
|
while (1) {
|
2008-09-23 19:14:14 +02:00
|
|
|
struct btrfs_ordered_extent *ordered;
|
2008-11-12 20:32:25 +01:00
|
|
|
lock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
|
|
|
|
ordered = btrfs_lookup_first_ordered_extent(inode, off+len);
|
2008-08-05 05:23:47 +02:00
|
|
|
if (BTRFS_I(src)->delalloc_bytes == 0 && !ordered)
|
2008-06-12 03:53:53 +02:00
|
|
|
break;
|
2008-11-12 20:32:25 +01:00
|
|
|
unlock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
|
2008-08-05 05:23:47 +02:00
|
|
|
if (ordered)
|
|
|
|
btrfs_put_ordered_extent(ordered);
|
2008-11-12 20:32:25 +01:00
|
|
|
btrfs_wait_ordered_range(src, off, off+len);
|
2008-06-12 03:53:53 +02:00
|
|
|
}
|
|
|
|
|
2008-08-05 05:23:47 +02:00
|
|
|
trans = btrfs_start_transaction(root, 1);
|
|
|
|
BUG_ON(!trans);
|
|
|
|
|
2008-11-12 20:32:25 +01:00
|
|
|
/* punch hole in destination first */
|
|
|
|
btrfs_drop_extents(trans, root, inode, off, off+len, 0, &hint_byte);
|
|
|
|
|
|
|
|
/* clone data */
|
2008-06-12 03:53:53 +02:00
|
|
|
key.objectid = src->i_ino;
|
2008-08-05 05:23:47 +02:00
|
|
|
key.type = BTRFS_EXTENT_DATA_KEY;
|
|
|
|
key.offset = 0;
|
2008-06-12 03:53:53 +02:00
|
|
|
|
|
|
|
while (1) {
|
|
|
|
/*
|
|
|
|
* note the key will change type as we walk through the
|
|
|
|
* tree.
|
|
|
|
*/
|
|
|
|
ret = btrfs_search_slot(trans, root, &key, path, 0, 0);
|
|
|
|
if (ret < 0)
|
|
|
|
goto out;
|
|
|
|
|
2008-08-05 05:23:47 +02:00
|
|
|
nritems = btrfs_header_nritems(path->nodes[0]);
|
|
|
|
if (path->slots[0] >= nritems) {
|
2008-06-12 03:53:53 +02:00
|
|
|
ret = btrfs_next_leaf(root, path);
|
|
|
|
if (ret < 0)
|
|
|
|
goto out;
|
|
|
|
if (ret > 0)
|
|
|
|
break;
|
2008-08-05 05:23:47 +02:00
|
|
|
nritems = btrfs_header_nritems(path->nodes[0]);
|
2008-06-12 03:53:53 +02:00
|
|
|
}
|
|
|
|
leaf = path->nodes[0];
|
|
|
|
slot = path->slots[0];
|
|
|
|
|
2008-08-05 05:23:47 +02:00
|
|
|
btrfs_item_key_to_cpu(leaf, &key, slot);
|
Btrfs: move data checksumming into a dedicated tree
Btrfs stores checksums for each data block. Until now, they have
been stored in the subvolume trees, indexed by the inode that is
referencing the data block. This means that when we read the inode,
we've probably read in at least some checksums as well.
But, this has a few problems:
* The checksums are indexed by logical offset in the file. When
compression is on, this means we have to do the expensive checksumming
on the uncompressed data. It would be faster if we could checksum
the compressed data instead.
* If we implement encryption, we'll be checksumming the plain text and
storing that on disk. This is significantly less secure.
* For either compression or encryption, we have to get the plain text
back before we can verify the checksum as correct. This makes the raid
layer balancing and extent moving much more expensive.
* It makes the front end caching code more complex, as we have touch
the subvolume and inodes as we cache extents.
* There is potentitally one copy of the checksum in each subvolume
referencing an extent.
The solution used here is to store the extent checksums in a dedicated
tree. This allows us to index the checksums by phyiscal extent
start and length. It means:
* The checksum is against the data stored on disk, after any compression
or encryption is done.
* The checksum is stored in a central location, and can be verified without
following back references, or reading inodes.
This makes compression significantly faster by reducing the amount of
data that needs to be checksummed. It will also allow much faster
raid management code in general.
The checksums are indexed by a key with a fixed objectid (a magic value
in ctree.h) and offset set to the starting byte of the extent. This
allows us to copy the checksum items into the fsync log tree directly (or
any other tree), without having to invent a second format for them.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-12-08 22:58:54 +01:00
|
|
|
if (btrfs_key_type(&key) > BTRFS_EXTENT_DATA_KEY ||
|
2008-06-12 03:53:53 +02:00
|
|
|
key.objectid != src->i_ino)
|
|
|
|
break;
|
|
|
|
|
2008-11-12 20:32:25 +01:00
|
|
|
if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) {
|
|
|
|
struct btrfs_file_extent_item *extent;
|
|
|
|
int type;
|
2008-09-23 19:14:14 +02:00
|
|
|
u32 size;
|
|
|
|
struct btrfs_key new_key;
|
2008-11-12 20:32:25 +01:00
|
|
|
u64 disko = 0, diskl = 0;
|
|
|
|
u64 datao = 0, datal = 0;
|
|
|
|
u8 comp;
|
2008-09-23 19:14:14 +02:00
|
|
|
|
|
|
|
size = btrfs_item_size_nr(leaf, slot);
|
|
|
|
read_extent_buffer(leaf, buf,
|
|
|
|
btrfs_item_ptr_offset(leaf, slot),
|
|
|
|
size);
|
2008-11-12 20:32:25 +01:00
|
|
|
|
|
|
|
extent = btrfs_item_ptr(leaf, slot,
|
|
|
|
struct btrfs_file_extent_item);
|
|
|
|
comp = btrfs_file_extent_compression(leaf, extent);
|
|
|
|
type = btrfs_file_extent_type(leaf, extent);
|
|
|
|
if (type == BTRFS_FILE_EXTENT_REG) {
|
2009-01-06 03:25:51 +01:00
|
|
|
disko = btrfs_file_extent_disk_bytenr(leaf,
|
|
|
|
extent);
|
|
|
|
diskl = btrfs_file_extent_disk_num_bytes(leaf,
|
|
|
|
extent);
|
2008-11-12 20:32:25 +01:00
|
|
|
datao = btrfs_file_extent_offset(leaf, extent);
|
2009-01-06 03:25:51 +01:00
|
|
|
datal = btrfs_file_extent_num_bytes(leaf,
|
|
|
|
extent);
|
2008-11-12 20:32:25 +01:00
|
|
|
} else if (type == BTRFS_FILE_EXTENT_INLINE) {
|
|
|
|
/* take upper bound, may be compressed */
|
|
|
|
datal = btrfs_file_extent_ram_bytes(leaf,
|
|
|
|
extent);
|
|
|
|
}
|
2008-09-23 19:14:14 +02:00
|
|
|
btrfs_release_path(root, path);
|
|
|
|
|
2008-11-12 20:32:25 +01:00
|
|
|
if (key.offset + datal < off ||
|
|
|
|
key.offset >= off+len)
|
|
|
|
goto next;
|
|
|
|
|
2008-09-23 19:14:14 +02:00
|
|
|
memcpy(&new_key, &key, sizeof(new_key));
|
|
|
|
new_key.objectid = inode->i_ino;
|
2008-11-12 20:32:25 +01:00
|
|
|
new_key.offset = key.offset + destoff - off;
|
2008-09-23 19:14:14 +02:00
|
|
|
|
2008-11-12 20:32:25 +01:00
|
|
|
if (type == BTRFS_FILE_EXTENT_REG) {
|
|
|
|
ret = btrfs_insert_empty_item(trans, root, path,
|
|
|
|
&new_key, size);
|
|
|
|
if (ret)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
leaf = path->nodes[0];
|
|
|
|
slot = path->slots[0];
|
|
|
|
write_extent_buffer(leaf, buf,
|
2008-09-23 19:14:14 +02:00
|
|
|
btrfs_item_ptr_offset(leaf, slot),
|
|
|
|
size);
|
2008-08-05 05:23:47 +02:00
|
|
|
|
2008-11-12 20:32:25 +01:00
|
|
|
extent = btrfs_item_ptr(leaf, slot,
|
2008-06-12 03:53:53 +02:00
|
|
|
struct btrfs_file_extent_item);
|
2008-11-12 20:32:25 +01:00
|
|
|
|
|
|
|
if (off > key.offset) {
|
|
|
|
datao += off - key.offset;
|
|
|
|
datal -= off - key.offset;
|
|
|
|
}
|
|
|
|
if (key.offset + datao + datal + key.offset >
|
|
|
|
off + len)
|
|
|
|
datal = off + len - key.offset - datao;
|
|
|
|
/* disko == 0 means it's a hole */
|
|
|
|
if (!disko)
|
|
|
|
datao = 0;
|
|
|
|
|
|
|
|
btrfs_set_file_extent_offset(leaf, extent,
|
|
|
|
datao);
|
|
|
|
btrfs_set_file_extent_num_bytes(leaf, extent,
|
|
|
|
datal);
|
|
|
|
if (disko) {
|
|
|
|
inode_add_bytes(inode, datal);
|
2008-08-05 05:23:47 +02:00
|
|
|
ret = btrfs_inc_extent_ref(trans, root,
|
2008-11-12 20:32:25 +01:00
|
|
|
disko, diskl, leaf->start,
|
|
|
|
root->root_key.objectid,
|
|
|
|
trans->transid,
|
|
|
|
inode->i_ino);
|
2008-09-23 19:14:14 +02:00
|
|
|
BUG_ON(ret);
|
2008-06-12 03:53:53 +02:00
|
|
|
}
|
2008-11-12 20:32:25 +01:00
|
|
|
} else if (type == BTRFS_FILE_EXTENT_INLINE) {
|
|
|
|
u64 skip = 0;
|
|
|
|
u64 trim = 0;
|
|
|
|
if (off > key.offset) {
|
|
|
|
skip = off - key.offset;
|
|
|
|
new_key.offset += skip;
|
|
|
|
}
|
2009-01-06 03:25:51 +01:00
|
|
|
|
2008-11-12 20:32:25 +01:00
|
|
|
if (key.offset + datal > off+len)
|
|
|
|
trim = key.offset + datal - (off+len);
|
2009-01-06 03:25:51 +01:00
|
|
|
|
2008-11-12 20:32:25 +01:00
|
|
|
if (comp && (skip || trim)) {
|
|
|
|
ret = -EINVAL;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
size -= skip + trim;
|
|
|
|
datal -= skip + trim;
|
|
|
|
ret = btrfs_insert_empty_item(trans, root, path,
|
|
|
|
&new_key, size);
|
|
|
|
if (ret)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
if (skip) {
|
2009-01-06 03:25:51 +01:00
|
|
|
u32 start =
|
|
|
|
btrfs_file_extent_calc_inline_size(0);
|
2008-11-12 20:32:25 +01:00
|
|
|
memmove(buf+start, buf+start+skip,
|
|
|
|
datal);
|
|
|
|
}
|
|
|
|
|
|
|
|
leaf = path->nodes[0];
|
|
|
|
slot = path->slots[0];
|
|
|
|
write_extent_buffer(leaf, buf,
|
|
|
|
btrfs_item_ptr_offset(leaf, slot),
|
|
|
|
size);
|
|
|
|
inode_add_bytes(inode, datal);
|
2008-06-12 03:53:53 +02:00
|
|
|
}
|
2008-11-12 20:32:25 +01:00
|
|
|
|
|
|
|
btrfs_mark_buffer_dirty(leaf);
|
2008-08-05 05:23:47 +02:00
|
|
|
}
|
2008-11-12 20:32:25 +01:00
|
|
|
|
2009-01-06 03:25:51 +01:00
|
|
|
next:
|
2008-09-23 19:14:14 +02:00
|
|
|
btrfs_release_path(root, path);
|
2008-06-12 03:53:53 +02:00
|
|
|
key.offset++;
|
|
|
|
}
|
|
|
|
ret = 0;
|
|
|
|
out:
|
2008-08-05 05:23:47 +02:00
|
|
|
btrfs_release_path(root, path);
|
|
|
|
if (ret == 0) {
|
|
|
|
inode->i_mtime = inode->i_ctime = CURRENT_TIME;
|
2008-11-12 20:32:25 +01:00
|
|
|
if (destoff + olen > inode->i_size)
|
|
|
|
btrfs_i_size_write(inode, destoff + olen);
|
2008-08-05 05:23:47 +02:00
|
|
|
BTRFS_I(inode)->flags = BTRFS_I(src)->flags;
|
|
|
|
ret = btrfs_update_inode(trans, root, inode);
|
|
|
|
}
|
2008-06-12 03:53:53 +02:00
|
|
|
btrfs_end_transaction(trans, root);
|
2008-11-12 20:32:25 +01:00
|
|
|
unlock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
|
2008-08-05 05:23:47 +02:00
|
|
|
if (ret)
|
|
|
|
vmtruncate(inode, 0);
|
2008-06-12 03:53:53 +02:00
|
|
|
out_unlock:
|
|
|
|
mutex_unlock(&src->i_mutex);
|
|
|
|
mutex_unlock(&inode->i_mutex);
|
2008-08-05 05:23:47 +02:00
|
|
|
vfree(buf);
|
|
|
|
btrfs_free_path(path);
|
2008-06-12 03:53:53 +02:00
|
|
|
out_fput:
|
|
|
|
fput(src_file);
|
2008-12-19 16:58:39 +01:00
|
|
|
out_drop_write:
|
|
|
|
mnt_drop_write(file->f_path.mnt);
|
2008-06-12 03:53:53 +02:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2008-12-02 15:52:24 +01:00
|
|
|
static long btrfs_ioctl_clone_range(struct file *file, void __user *argp)
|
2008-11-12 20:32:25 +01:00
|
|
|
{
|
|
|
|
struct btrfs_ioctl_clone_range_args args;
|
|
|
|
|
2008-12-02 15:52:24 +01:00
|
|
|
if (copy_from_user(&args, argp, sizeof(args)))
|
2008-11-12 20:32:25 +01:00
|
|
|
return -EFAULT;
|
|
|
|
return btrfs_ioctl_clone(file, args.src_fd, args.src_offset,
|
|
|
|
args.src_length, args.dest_offset);
|
|
|
|
}
|
|
|
|
|
2008-06-12 03:53:53 +02:00
|
|
|
/*
|
|
|
|
* there are many ways the trans_start and trans_end ioctls can lead
|
|
|
|
* to deadlocks. They should only be used by applications that
|
|
|
|
* basically own the machine, and have a very in depth understanding
|
|
|
|
* of all the possible deadlocks and enospc problems.
|
|
|
|
*/
|
2008-12-02 15:54:17 +01:00
|
|
|
static long btrfs_ioctl_trans_start(struct file *file)
|
2008-06-12 03:53:53 +02:00
|
|
|
{
|
|
|
|
struct inode *inode = fdentry(file)->d_inode;
|
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
int ret = 0;
|
|
|
|
|
2008-06-12 03:53:58 +02:00
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
return -EPERM;
|
|
|
|
|
2008-06-12 03:53:53 +02:00
|
|
|
if (file->private_data) {
|
|
|
|
ret = -EINPROGRESS;
|
|
|
|
goto out;
|
|
|
|
}
|
2008-08-04 16:41:27 +02:00
|
|
|
|
2008-11-12 20:34:12 +01:00
|
|
|
ret = mnt_want_write(file->f_path.mnt);
|
|
|
|
if (ret)
|
|
|
|
goto out;
|
|
|
|
|
2008-08-04 16:41:27 +02:00
|
|
|
mutex_lock(&root->fs_info->trans_mutex);
|
|
|
|
root->fs_info->open_ioctl_trans++;
|
|
|
|
mutex_unlock(&root->fs_info->trans_mutex);
|
|
|
|
|
|
|
|
trans = btrfs_start_ioctl_transaction(root, 0);
|
2008-06-12 03:53:53 +02:00
|
|
|
if (trans)
|
|
|
|
file->private_data = trans;
|
|
|
|
else
|
|
|
|
ret = -ENOMEM;
|
|
|
|
/*printk(KERN_INFO "btrfs_ioctl_trans_start on %p\n", file);*/
|
|
|
|
out:
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* there are many ways the trans_start and trans_end ioctls can lead
|
|
|
|
* to deadlocks. They should only be used by applications that
|
|
|
|
* basically own the machine, and have a very in depth understanding
|
|
|
|
* of all the possible deadlocks and enospc problems.
|
|
|
|
*/
|
|
|
|
long btrfs_ioctl_trans_end(struct file *file)
|
|
|
|
{
|
|
|
|
struct inode *inode = fdentry(file)->d_inode;
|
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
trans = file->private_data;
|
|
|
|
if (!trans) {
|
|
|
|
ret = -EINVAL;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
btrfs_end_transaction(trans, root);
|
2008-09-05 22:43:31 +02:00
|
|
|
file->private_data = NULL;
|
2008-08-04 16:41:27 +02:00
|
|
|
|
|
|
|
mutex_lock(&root->fs_info->trans_mutex);
|
|
|
|
root->fs_info->open_ioctl_trans--;
|
|
|
|
mutex_unlock(&root->fs_info->trans_mutex);
|
|
|
|
|
2008-12-11 22:30:06 +01:00
|
|
|
mnt_drop_write(file->f_path.mnt);
|
|
|
|
|
2008-06-12 03:53:53 +02:00
|
|
|
out:
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
long btrfs_ioctl(struct file *file, unsigned int
|
|
|
|
cmd, unsigned long arg)
|
|
|
|
{
|
|
|
|
struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
|
2008-12-02 12:36:08 +01:00
|
|
|
void __user *argp = (void __user *)arg;
|
2008-06-12 03:53:53 +02:00
|
|
|
|
|
|
|
switch (cmd) {
|
|
|
|
case BTRFS_IOC_SNAP_CREATE:
|
2008-12-02 12:36:08 +01:00
|
|
|
return btrfs_ioctl_snap_create(file, argp, 0);
|
2008-11-18 03:02:50 +01:00
|
|
|
case BTRFS_IOC_SUBVOL_CREATE:
|
2008-12-02 12:36:08 +01:00
|
|
|
return btrfs_ioctl_snap_create(file, argp, 1);
|
2008-06-12 03:53:53 +02:00
|
|
|
case BTRFS_IOC_DEFRAG:
|
|
|
|
return btrfs_ioctl_defrag(file);
|
|
|
|
case BTRFS_IOC_RESIZE:
|
2008-12-02 12:36:08 +01:00
|
|
|
return btrfs_ioctl_resize(root, argp);
|
2008-06-12 03:53:53 +02:00
|
|
|
case BTRFS_IOC_ADD_DEV:
|
2008-12-02 12:36:08 +01:00
|
|
|
return btrfs_ioctl_add_dev(root, argp);
|
2008-06-12 03:53:53 +02:00
|
|
|
case BTRFS_IOC_RM_DEV:
|
2008-12-02 12:36:08 +01:00
|
|
|
return btrfs_ioctl_rm_dev(root, argp);
|
2008-06-12 03:53:53 +02:00
|
|
|
case BTRFS_IOC_BALANCE:
|
|
|
|
return btrfs_balance(root->fs_info->dev_root);
|
|
|
|
case BTRFS_IOC_CLONE:
|
2008-11-12 20:32:25 +01:00
|
|
|
return btrfs_ioctl_clone(file, arg, 0, 0, 0);
|
|
|
|
case BTRFS_IOC_CLONE_RANGE:
|
2008-12-02 15:52:24 +01:00
|
|
|
return btrfs_ioctl_clone_range(file, argp);
|
2008-06-12 03:53:53 +02:00
|
|
|
case BTRFS_IOC_TRANS_START:
|
|
|
|
return btrfs_ioctl_trans_start(file);
|
|
|
|
case BTRFS_IOC_TRANS_END:
|
|
|
|
return btrfs_ioctl_trans_end(file);
|
|
|
|
case BTRFS_IOC_SYNC:
|
|
|
|
btrfs_sync_fs(file->f_dentry->d_sb, 1);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
return -ENOTTY;
|
|
|
|
}
|