[net-next-2.6.git] / fs / btrfs / ioctl.c

/*
 * 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>
#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>
#include <linux/mpage.h>
#include <linux/swap.h>
#include <linux/writeback.h>
#include <linux/statfs.h>
#include <linux/compat.h>
#include <linux/bit_spinlock.h>
#include <linux/version.h>
#include <linux/xattr.h>
#include "ctree.h"
#include "disk-io.h"
#include "transaction.h"
#include "btrfs_inode.h"
#include "ioctl.h"
#include "print-tree.h"
#include "volumes.h"
#include "locking.h"


static noinline int create_subvol(struct btrfs_root *root, char *name,
				  int namelen)
{
	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;
	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;

	leaf = btrfs_alloc_free_block(trans, root, root->leafsize,
				      objectid, trans->transid, 0, 0,
				      0, 0);
	if (IS_ERR(leaf))
		return PTR_ERR(leaf);

	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);
	inode_item->nblocks = cpu_to_le64(1);
	inode_item->mode = cpu_to_le32(S_IFDIR | 0755);

	btrfs_set_root_bytenr(&root_item, leaf->start);
	btrfs_set_root_level(&root_item, 0);
	btrfs_set_root_refs(&root_item, 1);
	btrfs_set_root_used(&root_item, 0);

	memset(&root_item.drop_progress, 0, sizeof(root_item.drop_progress));
	root_item.drop_level = 0;

	btrfs_tree_unlock(leaf);
	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;
	dir = root->fs_info->sb->s_root->d_inode;
	ret = btrfs_insert_dir_item(trans, root->fs_info->tree_root,
				    name, namelen, dir->i_ino, &key,
				    BTRFS_FT_DIR);
	if (ret)
		goto fail;

	ret = btrfs_insert_inode_ref(trans, root->fs_info->tree_root,
			     name, namelen, objectid,
			     root->fs_info->sb->s_root->d_inode->i_ino);
	if (ret)
		goto fail;

	ret = btrfs_commit_transaction(trans, root);
	if (ret)
		goto fail_commit;

	new_root = btrfs_read_fs_root(root->fs_info, &key, name, namelen);
	BUG_ON(!new_root);

	trans = btrfs_start_transaction(new_root, 1);
	BUG_ON(!trans);

	ret = btrfs_create_subvol_root(new_root, trans, new_dirid,
				       BTRFS_I(dir)->block_group);
	if (ret)
		goto fail;

	/* Invalidate existing dcache entry for new subvolume. */
	btrfs_invalidate_dcache_root(root, name, namelen);

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;
}

static int create_snapshot(struct btrfs_root *root, char *name, int namelen)
{
	struct btrfs_pending_snapshot *pending_snapshot;
	struct btrfs_trans_handle *trans;
	int ret;
	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;

	pending_snapshot = kmalloc(sizeof(*pending_snapshot), GFP_NOFS);
	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';
	trans = btrfs_start_transaction(root, 1);
	BUG_ON(!trans);
	pending_snapshot->root = root;
	list_add(&pending_snapshot->list,
		 &trans->transaction->pending_snapshots);
	ret = btrfs_update_inode(trans, root, root->inode);
	err = btrfs_commit_transaction(trans, root);

fail_unlock:
	btrfs_btree_balance_dirty(root, nr);
	return ret;
}

int btrfs_defrag_file(struct file *file)
{
	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;
	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++;
		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;
			}
		}

#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,18)
		ClearPageDirty(page);
#else
		cancel_dirty_page(page, PAGE_CACHE_SIZE);
#endif
		wait_on_page_writeback(page);
		set_page_extent_mapped(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);
		set_extent_delalloc(io_tree, page_start,
				    page_end, GFP_NOFS);

		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;

	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;
	}
	namelen = strlen(vol_args->name);
	if (namelen > BTRFS_VOL_NAME_MAX) {
		ret = -EINVAL;
		goto out;
	}

	mutex_lock(&root->fs_info->volume_mutex);
	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);
	}
	device = btrfs_find_device(root, devid, NULL);
	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:
	mutex_unlock(&root->fs_info->volume_mutex);
out:
	kfree(vol_args);
	return ret;
}

static noinline int btrfs_ioctl_snap_create(struct btrfs_root *root,
					    void __user *arg)
{
	struct btrfs_ioctl_vol_args *vol_args;
	struct btrfs_dir_item *di;
	struct btrfs_path *path;
	u64 root_dirid;
	int namelen;
	int ret;

	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;
	}

	namelen = strlen(vol_args->name);
	if (namelen > BTRFS_VOL_NAME_MAX) {
		ret = -EINVAL;
		goto out;
	}
	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;
	}

	mutex_lock(&root->fs_info->drop_mutex);
	if (root == root->fs_info->tree_root)
		ret = create_subvol(root, vol_args->name, namelen);
	else
		ret = create_snapshot(root, vol_args->name, namelen);
	mutex_unlock(&root->fs_info->drop_mutex);
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;

	switch (inode->i_mode & S_IFMT) {
	case S_IFDIR:
		btrfs_defrag_root(root, 0);
		btrfs_defrag_root(root->fs_info->extent_root, 0);
		break;
	case S_IFREG:
		btrfs_defrag_file(file);
		break;
	}

	return 0;
}

long btrfs_ioctl_add_dev(struct btrfs_root *root, void __user *arg)
{
	struct btrfs_ioctl_vol_args *vol_args;
	int ret;

	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;
	}
	ret = btrfs_init_new_device(root, vol_args->name);

out:
	kfree(vol_args);
	return ret;
}

long btrfs_ioctl_rm_dev(struct btrfs_root *root, void __user *arg)
{
	struct btrfs_ioctl_vol_args *vol_args;
	int ret;

	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;
	}
	ret = btrfs_rm_device(root, vol_args->name);

out:
	kfree(vol_args);
	return ret;
}

int dup_item_to_inode(struct btrfs_trans_handle *trans,
		       struct btrfs_root *root,
		       struct btrfs_path *path,
		       struct extent_buffer *leaf,
		       int slot,
		       struct btrfs_key *key,
		       u64 destino)
{
	char *dup;
	int len = btrfs_item_size_nr(leaf, slot);
	struct btrfs_key ckey = *key;
	int ret = 0;

	dup = kmalloc(len, GFP_NOFS);
	if (!dup)
		return -ENOMEM;

	read_extent_buffer(leaf, dup, btrfs_item_ptr_offset(leaf, slot), len);
	btrfs_release_path(root, path);

	ckey.objectid = destino;
	ret = btrfs_insert_item(trans, root, &ckey, dup, len);
	kfree(dup);
	return ret;
}

long btrfs_ioctl_clone(struct file *file, unsigned long src_fd)
{
	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;
	int ret;
	u64 pos;
	struct btrfs_path *path;
	struct btrfs_key key;
	struct extent_buffer *leaf;
	u32 nritems;
	int slot;

	src_file = fget(src_fd);
	if (!src_file)
		return -EBADF;
	src = src_file->f_dentry->d_inode;

	ret = -EXDEV;
	if (src->i_sb != inode->i_sb)
		goto out_fput;

	if (inode < src) {
		mutex_lock(&inode->i_mutex);
		mutex_lock(&src->i_mutex);
	} else {
		mutex_lock(&src->i_mutex);
		mutex_lock(&inode->i_mutex);
	}

	ret = -ENOTEMPTY;
	if (inode->i_size)
		goto out_unlock;

	/* do any pending delalloc/csum calc on src, one way or
	   another, and lock file content */
	while (1) {
		filemap_write_and_wait(src->i_mapping);
		lock_extent(&BTRFS_I(src)->io_tree, 0, (u64)-1, GFP_NOFS);
		if (BTRFS_I(src)->delalloc_bytes == 0)
			break;
		unlock_extent(&BTRFS_I(src)->io_tree, 0, (u64)-1, GFP_NOFS);
	}

	trans = btrfs_start_transaction(root, 0);
	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto out;
	}
	key.offset = 0;
	key.type = BTRFS_EXTENT_DATA_KEY;
	key.objectid = src->i_ino;
	pos = 0;
	path->reada = 2;

	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;

		if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
			ret = btrfs_next_leaf(root, path);
			if (ret < 0)
				goto out;
			if (ret > 0)
				break;
		}
		leaf = path->nodes[0];
		slot = path->slots[0];
		btrfs_item_key_to_cpu(leaf, &key, slot);
		nritems = btrfs_header_nritems(leaf);

		if (btrfs_key_type(&key) > BTRFS_CSUM_ITEM_KEY ||
		    key.objectid != src->i_ino)
			break;

		if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) {
			struct btrfs_file_extent_item *extent;
			int found_type;
			pos = key.offset;
			extent = btrfs_item_ptr(leaf, slot,
						struct btrfs_file_extent_item);
			found_type = btrfs_file_extent_type(leaf, extent);
			if (found_type == BTRFS_FILE_EXTENT_REG) {
				u64 len = btrfs_file_extent_num_bytes(leaf,
								      extent);
				u64 ds = btrfs_file_extent_disk_bytenr(leaf,
								       extent);
				u64 dl = btrfs_file_extent_disk_num_bytes(leaf,
								 extent);
				u64 off = btrfs_file_extent_offset(leaf,
								   extent);
				btrfs_insert_file_extent(trans, root,
							 inode->i_ino, pos,
							 ds, dl, len, off);
				/* ds == 0 means there's a hole */
				if (ds != 0) {
					btrfs_inc_extent_ref(trans, root,
						     ds, dl,
						     root->root_key.objectid,
						     trans->transid,
						     inode->i_ino, pos);
				}
				pos = key.offset + len;
			} else if (found_type == BTRFS_FILE_EXTENT_INLINE) {
				ret = dup_item_to_inode(trans, root, path,
							leaf, slot, &key,
							inode->i_ino);
				if (ret)
					goto out;
				pos = key.offset + btrfs_item_size_nr(leaf,
								      slot);
			}
		} else if (btrfs_key_type(&key) == BTRFS_CSUM_ITEM_KEY) {
			ret = dup_item_to_inode(trans, root, path, leaf,
						slot, &key, inode->i_ino);

			if (ret)
				goto out;
		}
		key.offset++;
		btrfs_release_path(root, path);
	}

	ret = 0;
out:
	btrfs_free_path(path);

	inode->i_blocks = src->i_blocks;
	i_size_write(inode, src->i_size);
	btrfs_update_inode(trans, root, inode);

	unlock_extent(&BTRFS_I(src)->io_tree, 0, (u64)-1, GFP_NOFS);

	btrfs_end_transaction(trans, root);

out_unlock:
	mutex_unlock(&src->i_mutex);
	mutex_unlock(&inode->i_mutex);
out_fput:
	fput(src_file);
	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_start(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;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	if (file->private_data) {
		ret = -EINPROGRESS;
		goto out;
	}
	trans = btrfs_start_transaction(root, 0);
	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);
	file->private_data = 0;
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;

	switch (cmd) {
	case BTRFS_IOC_SNAP_CREATE:
		return btrfs_ioctl_snap_create(root, (void __user *)arg);
	case BTRFS_IOC_DEFRAG:
		return btrfs_ioctl_defrag(file);
	case BTRFS_IOC_RESIZE:
		return btrfs_ioctl_resize(root, (void __user *)arg);
	case BTRFS_IOC_ADD_DEV:
		return btrfs_ioctl_add_dev(root, (void __user *)arg);
	case BTRFS_IOC_RM_DEV:
		return btrfs_ioctl_rm_dev(root, (void __user *)arg);
	case BTRFS_IOC_BALANCE:
		return btrfs_balance(root->fs_info->dev_root);
	case BTRFS_IOC_CLONE:
		return btrfs_ioctl_clone(file, arg);
	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;
}
Commit	Line	Data
f46b5a66 CH	1	/*
	2	* Copyright (C) 2007 Oracle. All rights reserved.
	3	*
	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.
	7	*
	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.
	12	*
	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.
	17	*/
	18
	19	#include <linux/kernel.h>
	20	#include <linux/bio.h>
	21	#include <linux/buffer_head.h>
	22	#include <linux/file.h>
	23	#include <linux/fs.h>
	24	#include <linux/pagemap.h>
	25	#include <linux/highmem.h>
	26	#include <linux/time.h>
	27	#include <linux/init.h>
	28	#include <linux/string.h>
	29	#include <linux/smp_lock.h>
	30	#include <linux/backing-dev.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/bit_spinlock.h>
	37	#include <linux/version.h>
	38	#include <linux/xattr.h>
	39	#include "ctree.h"
	40	#include "disk-io.h"
	41	#include "transaction.h"
	42	#include "btrfs_inode.h"
	43	#include "ioctl.h"
	44	#include "print-tree.h"
	45	#include "volumes.h"
925baedd	46	#include "locking.h"
f46b5a66 CH	47
	48
	49
	50	static noinline int create_subvol(struct btrfs_root root, char name,
	51	int namelen)
	52	{
	53	struct btrfs_trans_handle *trans;
	54	struct btrfs_key key;
	55	struct btrfs_root_item root_item;
	56	struct btrfs_inode_item *inode_item;
	57	struct extent_buffer *leaf;
	58	struct btrfs_root *new_root = root;
	59	struct inode *dir;
	60	int ret;
	61	int err;
	62	u64 objectid;
	63	u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
	64	unsigned long nr = 1;
	65
f46b5a66 CH	66	ret = btrfs_check_free_space(root, 1, 0);
	67	if (ret)
	68	goto fail_commit;
	69
	70	trans = btrfs_start_transaction(root, 1);
	71	BUG_ON(!trans);
	72
	73	ret = btrfs_find_free_objectid(trans, root->fs_info->tree_root,
	74	0, &objectid);
	75	if (ret)
	76	goto fail;
	77
925baedd CM	78	leaf = btrfs_alloc_free_block(trans, root, root->leafsize,
	79	objectid, trans->transid, 0, 0,
	80	0, 0);
f46b5a66 CH	81	if (IS_ERR(leaf))
	82	return PTR_ERR(leaf);
	83
	84	btrfs_set_header_nritems(leaf, 0);
	85	btrfs_set_header_level(leaf, 0);
	86	btrfs_set_header_bytenr(leaf, leaf->start);
	87	btrfs_set_header_generation(leaf, trans->transid);
	88	btrfs_set_header_owner(leaf, objectid);
	89
	90	write_extent_buffer(leaf, root->fs_info->fsid,
	91	(unsigned long)btrfs_header_fsid(leaf),
	92	BTRFS_FSID_SIZE);
	93	btrfs_mark_buffer_dirty(leaf);
	94
	95	inode_item = &root_item.inode;
	96	memset(inode_item, 0, sizeof(*inode_item));
	97	inode_item->generation = cpu_to_le64(1);
	98	inode_item->size = cpu_to_le64(3);
	99	inode_item->nlink = cpu_to_le32(1);
	100	inode_item->nblocks = cpu_to_le64(1);
	101	inode_item->mode = cpu_to_le32(S_IFDIR \| 0755);
	102
	103	btrfs_set_root_bytenr(&root_item, leaf->start);
	104	btrfs_set_root_level(&root_item, 0);
	105	btrfs_set_root_refs(&root_item, 1);
	106	btrfs_set_root_used(&root_item, 0);
	107
	108	memset(&root_item.drop_progress, 0, sizeof(root_item.drop_progress));
	109	root_item.drop_level = 0;
	110
925baedd	111	btrfs_tree_unlock(leaf);
f46b5a66 CH	112	free_extent_buffer(leaf);
	113	leaf = NULL;
	114
	115	btrfs_set_root_dirid(&root_item, new_dirid);
	116
	117	key.objectid = objectid;
	118	key.offset = 1;
	119	btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
	120	ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key,
	121	&root_item);
	122	if (ret)
	123	goto fail;
	124
	125	/*
	126	* insert the directory item
	127	*/
	128	key.offset = (u64)-1;
	129	dir = root->fs_info->sb->s_root->d_inode;
	130	ret = btrfs_insert_dir_item(trans, root->fs_info->tree_root,
	131	name, namelen, dir->i_ino, &key,
	132	BTRFS_FT_DIR);
	133	if (ret)
	134	goto fail;
	135
	136	ret = btrfs_insert_inode_ref(trans, root->fs_info->tree_root,
	137	name, namelen, objectid,
	138	root->fs_info->sb->s_root->d_inode->i_ino);
	139	if (ret)
	140	goto fail;
	141
	142	ret = btrfs_commit_transaction(trans, root);
	143	if (ret)
	144	goto fail_commit;
	145
	146	new_root = btrfs_read_fs_root(root->fs_info, &key, name, namelen);
	147	BUG_ON(!new_root);
	148
	149	trans = btrfs_start_transaction(new_root, 1);
	150	BUG_ON(!trans);
	151
	152	ret = btrfs_create_subvol_root(new_root, trans, new_dirid,
	153	BTRFS_I(dir)->block_group);
	154	if (ret)
	155	goto fail;
	156
	157	/* Invalidate existing dcache entry for new subvolume. */
	158	btrfs_invalidate_dcache_root(root, name, namelen);
	159
	160	fail:
	161	nr = trans->blocks_used;
	162	err = btrfs_commit_transaction(trans, new_root);
	163	if (err && !ret)
	164	ret = err;
	165	fail_commit:
f46b5a66	166	btrfs_btree_balance_dirty(root, nr);
f46b5a66 CH	167	return ret;
	168	}
	169
	170	static int create_snapshot(struct btrfs_root root, char name, int namelen)
	171	{
	172	struct btrfs_pending_snapshot *pending_snapshot;
	173	struct btrfs_trans_handle *trans;
	174	int ret;
	175	int err;
	176	unsigned long nr = 0;
	177
	178	if (!root->ref_cows)
	179	return -EINVAL;
	180
f46b5a66 CH	181	ret = btrfs_check_free_space(root, 1, 0);
	182	if (ret)
	183	goto fail_unlock;
	184
	185	pending_snapshot = kmalloc(sizeof(*pending_snapshot), GFP_NOFS);
	186	if (!pending_snapshot) {
	187	ret = -ENOMEM;
	188	goto fail_unlock;
	189	}
	190	pending_snapshot->name = kmalloc(namelen + 1, GFP_NOFS);
	191	if (!pending_snapshot->name) {
	192	ret = -ENOMEM;
	193	kfree(pending_snapshot);
	194	goto fail_unlock;
	195	}
	196	memcpy(pending_snapshot->name, name, namelen);
	197	pending_snapshot->name[namelen] = '\0';
	198	trans = btrfs_start_transaction(root, 1);
	199	BUG_ON(!trans);
	200	pending_snapshot->root = root;
	201	list_add(&pending_snapshot->list,
	202	&trans->transaction->pending_snapshots);
	203	ret = btrfs_update_inode(trans, root, root->inode);
	204	err = btrfs_commit_transaction(trans, root);
	205
	206	fail_unlock:
f46b5a66	207	btrfs_btree_balance_dirty(root, nr);
f46b5a66 CH	208	return ret;
	209	}
	210
	211	int btrfs_defrag_file(struct file *file)
	212	{
	213	struct inode *inode = fdentry(file)->d_inode;
	214	struct btrfs_root *root = BTRFS_I(inode)->root;
	215	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	216	struct page *page;
	217	unsigned long last_index;
	218	unsigned long ra_pages = root->fs_info->bdi.ra_pages;
	219	unsigned long total_read = 0;
	220	u64 page_start;
	221	u64 page_end;
	222	unsigned long i;
	223	int ret;
	224
f46b5a66	225	ret = btrfs_check_free_space(root, inode->i_size, 0);
f46b5a66 CH	226	if (ret)
	227	return -ENOSPC;
	228
	229	mutex_lock(&inode->i_mutex);
	230	last_index = inode->i_size >> PAGE_CACHE_SHIFT;
	231	for (i = 0; i <= last_index; i++) {
	232	if (total_read % ra_pages == 0) {
	233	btrfs_force_ra(inode->i_mapping, &file->f_ra, file, i,
	234	min(last_index, i + ra_pages - 1));
	235	}
	236	total_read++;
	237	page = grab_cache_page(inode->i_mapping, i);
	238	if (!page)
	239	goto out_unlock;
	240	if (!PageUptodate(page)) {
	241	btrfs_readpage(NULL, page);
	242	lock_page(page);
	243	if (!PageUptodate(page)) {
	244	unlock_page(page);
	245	page_cache_release(page);
	246	goto out_unlock;
	247	}
	248	}
	249
	250	#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,18)
	251	ClearPageDirty(page);
	252	#else
	253	cancel_dirty_page(page, PAGE_CACHE_SIZE);
	254	#endif
	255	wait_on_page_writeback(page);
	256	set_page_extent_mapped(page);
	257
	258	page_start = (u64)page->index << PAGE_CACHE_SHIFT;
	259	page_end = page_start + PAGE_CACHE_SIZE - 1;
	260
	261	lock_extent(io_tree, page_start, page_end, GFP_NOFS);
	262	set_extent_delalloc(io_tree, page_start,
	263	page_end, GFP_NOFS);
	264
	265	unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
	266	set_page_dirty(page);
	267	unlock_page(page);
	268	page_cache_release(page);
	269	balance_dirty_pages_ratelimited_nr(inode->i_mapping, 1);
	270	}
	271
	272	out_unlock:
	273	mutex_unlock(&inode->i_mutex);
	274	return 0;
	275	}
	276
	277	/*
	278	* Called inside transaction, so use GFP_NOFS
	279	*/
	280
	281	static int btrfs_ioctl_resize(struct btrfs_root root, void __user arg)
	282	{
	283	u64 new_size;
	284	u64 old_size;
	285	u64 devid = 1;
	286	struct btrfs_ioctl_vol_args *vol_args;
	287	struct btrfs_trans_handle *trans;
	288	struct btrfs_device *device = NULL;
	289	char *sizestr;
290	char *devstr = NULL;
291	int ret = 0;
292	int namelen;
293	int mod = 0;
294
295	vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
296
297	if (!vol_args)
298	return -ENOMEM;
299
300	if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
301	ret = -EFAULT;
302	goto out;
303	}
304	namelen = strlen(vol_args->name);
305	if (namelen > BTRFS_VOL_NAME_MAX) {
306	ret = -EINVAL;
307	goto out;
308	}
309
7d9eb12c	310	mutex_lock(&root->fs_info->volume_mutex);
f46b5a66 CH	311	sizestr = vol_args->name;
	312	devstr = strchr(sizestr, ':');
	313	if (devstr) {
	314	char *end;
	315	sizestr = devstr + 1;
	316	*devstr = '\0';
	317	devstr = vol_args->name;
	318	devid = simple_strtoull(devstr, &end, 10);
	319	printk(KERN_INFO "resizing devid %llu\n", devid);
	320	}
	321	device = btrfs_find_device(root, devid, NULL);
	322	if (!device) {
	323	printk(KERN_INFO "resizer unable to find device %llu\n", devid);
	324	ret = -EINVAL;
	325	goto out_unlock;
	326	}
	327	if (!strcmp(sizestr, "max"))
	328	new_size = device->bdev->bd_inode->i_size;
	329	else {
	330	if (sizestr[0] == '-') {
	331	mod = -1;
	332	sizestr++;
	333	} else if (sizestr[0] == '+') {
	334	mod = 1;
	335	sizestr++;
	336	}
	337	new_size = btrfs_parse_size(sizestr);
	338	if (new_size == 0) {
	339	ret = -EINVAL;
	340	goto out_unlock;
	341	}
	342	}
	343
	344	old_size = device->total_bytes;
	345
	346	if (mod < 0) {
	347	if (new_size > old_size) {
	348	ret = -EINVAL;
	349	goto out_unlock;
	350	}
	351	new_size = old_size - new_size;
	352	} else if (mod > 0) {
	353	new_size = old_size + new_size;
	354	}
	355
	356	if (new_size < 256 * 1024 * 1024) {
	357	ret = -EINVAL;
	358	goto out_unlock;
	359	}
	360	if (new_size > device->bdev->bd_inode->i_size) {
	361	ret = -EFBIG;
	362	goto out_unlock;
	363	}
	364
	365	do_div(new_size, root->sectorsize);
	366	new_size *= root->sectorsize;
	367
	368	printk(KERN_INFO "new size for %s is %llu\n",
	369	device->name, (unsigned long long)new_size);
	370
	371	if (new_size > old_size) {
	372	trans = btrfs_start_transaction(root, 1);
	373	ret = btrfs_grow_device(trans, device, new_size);
	374	btrfs_commit_transaction(trans, root);
375	} else {
376	ret = btrfs_shrink_device(device, new_size);
377	}
378
379	out_unlock:
7d9eb12c	380	mutex_unlock(&root->fs_info->volume_mutex);
f46b5a66 CH	381	out:
	382	kfree(vol_args);
	383	return ret;
	384	}
	385
	386	static noinline int btrfs_ioctl_snap_create(struct btrfs_root *root,
	387	void __user *arg)
	388	{
	389	struct btrfs_ioctl_vol_args *vol_args;
	390	struct btrfs_dir_item *di;
	391	struct btrfs_path *path;
	392	u64 root_dirid;
	393	int namelen;
	394	int ret;
	395
	396	vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
	397
	398	if (!vol_args)
	399	return -ENOMEM;
	400
	401	if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
	402	ret = -EFAULT;
	403	goto out;
	404	}
	405
	406	namelen = strlen(vol_args->name);
	407	if (namelen > BTRFS_VOL_NAME_MAX) {
	408	ret = -EINVAL;
	409	goto out;
	410	}
	411	if (strchr(vol_args->name, '/')) {
	412	ret = -EINVAL;
	413	goto out;
	414	}
	415
	416	path = btrfs_alloc_path();
	417	if (!path) {
	418	ret = -ENOMEM;
	419	goto out;
	420	}
	421
	422	root_dirid = root->fs_info->sb->s_root->d_inode->i_ino,
f46b5a66 CH	423	di = btrfs_lookup_dir_item(NULL, root->fs_info->tree_root,
	424	path, root_dirid,
	425	vol_args->name, namelen, 0);
f46b5a66 CH	426	btrfs_free_path(path);
	427
	428	if (di && !IS_ERR(di)) {
	429	ret = -EEXIST;
	430	goto out;
	431	}
	432
	433	if (IS_ERR(di)) {
	434	ret = PTR_ERR(di);
	435	goto out;
	436	}
	437
a2135011	438	mutex_lock(&root->fs_info->drop_mutex);
f46b5a66 CH	439	if (root == root->fs_info->tree_root)
	440	ret = create_subvol(root, vol_args->name, namelen);
	441	else
	442	ret = create_snapshot(root, vol_args->name, namelen);
a2135011	443	mutex_unlock(&root->fs_info->drop_mutex);
f46b5a66 CH	444	out:
	445	kfree(vol_args);
	446	return ret;
	447	}
	448
	449	static int btrfs_ioctl_defrag(struct file *file)
	450	{
	451	struct inode *inode = fdentry(file)->d_inode;
	452	struct btrfs_root *root = BTRFS_I(inode)->root;
	453
	454	switch (inode->i_mode & S_IFMT) {
	455	case S_IFDIR:
f46b5a66 CH	456	btrfs_defrag_root(root, 0);
f46b5a66 CH	457	btrfs_defrag_root(root->fs_info->extent_root, 0);
f46b5a66 CH	458	break;
	459	case S_IFREG:
	460	btrfs_defrag_file(file);
	461	break;
	462	}
	463
	464	return 0;
	465	}
	466
	467	long btrfs_ioctl_add_dev(struct btrfs_root root, void __user arg)
	468	{
	469	struct btrfs_ioctl_vol_args *vol_args;
	470	int ret;
	471
	472	vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
	473
	474	if (!vol_args)
	475	return -ENOMEM;
	476
	477	if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
	478	ret = -EFAULT;
	479	goto out;
	480	}
	481	ret = btrfs_init_new_device(root, vol_args->name);
	482
	483	out:
	484	kfree(vol_args);
	485	return ret;
	486	}
	487
	488	long btrfs_ioctl_rm_dev(struct btrfs_root root, void __user arg)
	489	{
	490	struct btrfs_ioctl_vol_args *vol_args;
	491	int ret;
	492
	493	vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
	494
	495	if (!vol_args)
	496	return -ENOMEM;
	497
	498	if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
	499	ret = -EFAULT;
	500	goto out;
	501	}
	502	ret = btrfs_rm_device(root, vol_args->name);
	503
	504	out:
	505	kfree(vol_args);
	506	return ret;
	507	}
	508
	509	int dup_item_to_inode(struct btrfs_trans_handle *trans,
	510	struct btrfs_root *root,
	511	struct btrfs_path *path,
	512	struct extent_buffer *leaf,
	513	int slot,
	514	struct btrfs_key *key,
	515	u64 destino)
	516	{
	517	char *dup;
	518	int len = btrfs_item_size_nr(leaf, slot);
	519	struct btrfs_key ckey = *key;
	520	int ret = 0;
	521
522	dup = kmalloc(len, GFP_NOFS);
523	if (!dup)
524	return -ENOMEM;
525
526	read_extent_buffer(leaf, dup, btrfs_item_ptr_offset(leaf, slot), len);
527	btrfs_release_path(root, path);
528
529	ckey.objectid = destino;
530	ret = btrfs_insert_item(trans, root, &ckey, dup, len);
531	kfree(dup);
532	return ret;
533	}
534
535	long btrfs_ioctl_clone(struct file *file, unsigned long src_fd)
536	{
537	struct inode *inode = fdentry(file)->d_inode;
538	struct btrfs_root *root = BTRFS_I(inode)->root;
539	struct file *src_file;
540	struct inode *src;
541	struct btrfs_trans_handle *trans;
542	int ret;
543	u64 pos;
544	struct btrfs_path *path;
545	struct btrfs_key key;
546	struct extent_buffer *leaf;
547	u32 nritems;
548	int slot;
549
550	src_file = fget(src_fd);
551	if (!src_file)
552	return -EBADF;
553	src = src_file->f_dentry->d_inode;
554
555	ret = -EXDEV;
556	if (src->i_sb != inode->i_sb)
557	goto out_fput;
558
559	if (inode < src) {
560	mutex_lock(&inode->i_mutex);
561	mutex_lock(&src->i_mutex);
562	} else {
563	mutex_lock(&src->i_mutex);
564	mutex_lock(&inode->i_mutex);
565	}
566
567	ret = -ENOTEMPTY;
568	if (inode->i_size)
569	goto out_unlock;
570
571	/* do any pending delalloc/csum calc on src, one way or
572	another, and lock file content */
573	while (1) {
574	filemap_write_and_wait(src->i_mapping);
575	lock_extent(&BTRFS_I(src)->io_tree, 0, (u64)-1, GFP_NOFS);
576	if (BTRFS_I(src)->delalloc_bytes == 0)
577	break;
578	unlock_extent(&BTRFS_I(src)->io_tree, 0, (u64)-1, GFP_NOFS);
579	}
580
f46b5a66 CH	581	trans = btrfs_start_transaction(root, 0);
	582	path = btrfs_alloc_path();
	583	if (!path) {
	584	ret = -ENOMEM;
	585	goto out;
	586	}
	587	key.offset = 0;
	588	key.type = BTRFS_EXTENT_DATA_KEY;
	589	key.objectid = src->i_ino;
	590	pos = 0;
	591	path->reada = 2;
	592
	593	while (1) {
	594	/*
	595	* note the key will change type as we walk through the
	596	* tree.
	597	*/
	598	ret = btrfs_search_slot(trans, root, &key, path, 0, 0);
	599	if (ret < 0)
	600	goto out;
	601
	602	if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
	603	ret = btrfs_next_leaf(root, path);
	604	if (ret < 0)
	605	goto out;
	606	if (ret > 0)
	607	break;
	608	}
	609	leaf = path->nodes[0];
	610	slot = path->slots[0];
	611	btrfs_item_key_to_cpu(leaf, &key, slot);
	612	nritems = btrfs_header_nritems(leaf);
	613
	614	if (btrfs_key_type(&key) > BTRFS_CSUM_ITEM_KEY \|\|
	615	key.objectid != src->i_ino)
	616	break;
	617
	618	if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) {
	619	struct btrfs_file_extent_item *extent;
	620	int found_type;
	621	pos = key.offset;
	622	extent = btrfs_item_ptr(leaf, slot,
	623	struct btrfs_file_extent_item);
	624	found_type = btrfs_file_extent_type(leaf, extent);
	625	if (found_type == BTRFS_FILE_EXTENT_REG) {
	626	u64 len = btrfs_file_extent_num_bytes(leaf,
	627	extent);
	628	u64 ds = btrfs_file_extent_disk_bytenr(leaf,
	629	extent);
	630	u64 dl = btrfs_file_extent_disk_num_bytes(leaf,
	631	extent);
	632	u64 off = btrfs_file_extent_offset(leaf,
	633	extent);
	634	btrfs_insert_file_extent(trans, root,
	635	inode->i_ino, pos,
	636	ds, dl, len, off);
	637	/* ds == 0 means there's a hole */
	638	if (ds != 0) {
	639	btrfs_inc_extent_ref(trans, root,
	640	ds, dl,
	641	root->root_key.objectid,
	642	trans->transid,
	643	inode->i_ino, pos);
	644	}
645	pos = key.offset + len;
646	} else if (found_type == BTRFS_FILE_EXTENT_INLINE) {
647	ret = dup_item_to_inode(trans, root, path,
648	leaf, slot, &key,
649	inode->i_ino);
650	if (ret)
651	goto out;
652	pos = key.offset + btrfs_item_size_nr(leaf,
653	slot);
654	}
655	} else if (btrfs_key_type(&key) == BTRFS_CSUM_ITEM_KEY) {
656	ret = dup_item_to_inode(trans, root, path, leaf,
657	slot, &key, inode->i_ino);
658
659	if (ret)
660	goto out;
661	}
662	key.offset++;
663	btrfs_release_path(root, path);
664	}
665
666	ret = 0;
667	out:
668	btrfs_free_path(path);
669
670	inode->i_blocks = src->i_blocks;
671	i_size_write(inode, src->i_size);
672	btrfs_update_inode(trans, root, inode);
673
674	unlock_extent(&BTRFS_I(src)->io_tree, 0, (u64)-1, GFP_NOFS);
675
676	btrfs_end_transaction(trans, root);
f46b5a66 CH	677
	678	out_unlock:
	679	mutex_unlock(&src->i_mutex);
	680	mutex_unlock(&inode->i_mutex);
	681	out_fput:
	682	fput(src_file);
	683	return ret;
	684	}
	685
	686	/*
	687	* there are many ways the trans_start and trans_end ioctls can lead
	688	* to deadlocks. They should only be used by applications that
	689	* basically own the machine, and have a very in depth understanding
	690	* of all the possible deadlocks and enospc problems.
	691	*/
	692	long btrfs_ioctl_trans_start(struct file *file)
	693	{
	694	struct inode *inode = fdentry(file)->d_inode;
	695	struct btrfs_root *root = BTRFS_I(inode)->root;
	696	struct btrfs_trans_handle *trans;
	697	int ret = 0;
	698
df5b5520 CH	699	if (!capable(CAP_SYS_ADMIN))
	700	return -EPERM;
	701
f46b5a66 CH	702	if (file->private_data) {
	703	ret = -EINPROGRESS;
	704	goto out;
	705	}
	706	trans = btrfs_start_transaction(root, 0);
	707	if (trans)
	708	file->private_data = trans;
	709	else
	710	ret = -ENOMEM;
	711	/printk(KERN_INFO "btrfs_ioctl_trans_start on %p\n", file);/
	712	out:
f46b5a66 CH	713	return ret;
	714	}
	715
	716	/*
	717	* there are many ways the trans_start and trans_end ioctls can lead
	718	* to deadlocks. They should only be used by applications that
	719	* basically own the machine, and have a very in depth understanding
	720	* of all the possible deadlocks and enospc problems.
	721	*/
	722	long btrfs_ioctl_trans_end(struct file *file)
	723	{
	724	struct inode *inode = fdentry(file)->d_inode;
	725	struct btrfs_root *root = BTRFS_I(inode)->root;
	726	struct btrfs_trans_handle *trans;
	727	int ret = 0;
	728
f46b5a66 CH	729	trans = file->private_data;
	730	if (!trans) {
	731	ret = -EINVAL;
	732	goto out;
	733	}
	734	btrfs_end_transaction(trans, root);
	735	file->private_data = 0;
	736	out:
f46b5a66 CH	737	return ret;
	738	}
	739
	740	long btrfs_ioctl(struct file *file, unsigned int
	741	cmd, unsigned long arg)
	742	{
	743	struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
	744
	745	switch (cmd) {
	746	case BTRFS_IOC_SNAP_CREATE:
	747	return btrfs_ioctl_snap_create(root, (void __user *)arg);
	748	case BTRFS_IOC_DEFRAG:
	749	return btrfs_ioctl_defrag(file);
	750	case BTRFS_IOC_RESIZE:
	751	return btrfs_ioctl_resize(root, (void __user *)arg);
	752	case BTRFS_IOC_ADD_DEV:
	753	return btrfs_ioctl_add_dev(root, (void __user *)arg);
	754	case BTRFS_IOC_RM_DEV:
	755	return btrfs_ioctl_rm_dev(root, (void __user *)arg);
	756	case BTRFS_IOC_BALANCE:
	757	return btrfs_balance(root->fs_info->dev_root);
	758	case BTRFS_IOC_CLONE:
	759	return btrfs_ioctl_clone(file, arg);
	760	case BTRFS_IOC_TRANS_START:
	761	return btrfs_ioctl_trans_start(file);
	762	case BTRFS_IOC_TRANS_END:
	763	return btrfs_ioctl_trans_end(file);
	764	case BTRFS_IOC_SYNC:
	765	btrfs_sync_fs(file->f_dentry->d_sb, 1);
	766	return 0;
	767	}
	768
	769	return -ENOTTY;
	770	}