[people/arne_f/kernel.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)) {
		ret = PTR_ERR(leaf);
		goto fail;
	}

	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, 0);
	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, 0);
	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 btrfs_ordered_extent *ordered;
	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++;
again:
		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);

		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);

		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;
	}

	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
	namelen = strlen(vol_args->name);

	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;
	}

	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
	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;
	}

	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;
	}
	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
	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;
	}
	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
	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);
8e8a1e31 JB	81	if (IS_ERR(leaf)) {
	82	ret = PTR_ERR(leaf);
	83	goto fail;
	84	}
f46b5a66 CH	85
	86	btrfs_set_header_nritems(leaf, 0);
	87	btrfs_set_header_level(leaf, 0);
	88	btrfs_set_header_bytenr(leaf, leaf->start);
	89	btrfs_set_header_generation(leaf, trans->transid);
	90	btrfs_set_header_owner(leaf, objectid);
	91
	92	write_extent_buffer(leaf, root->fs_info->fsid,
	93	(unsigned long)btrfs_header_fsid(leaf),
	94	BTRFS_FSID_SIZE);
	95	btrfs_mark_buffer_dirty(leaf);
	96
	97	inode_item = &root_item.inode;
	98	memset(inode_item, 0, sizeof(*inode_item));
	99	inode_item->generation = cpu_to_le64(1);
	100	inode_item->size = cpu_to_le64(3);
	101	inode_item->nlink = cpu_to_le32(1);
	102	inode_item->nblocks = cpu_to_le64(1);
	103	inode_item->mode = cpu_to_le32(S_IFDIR \| 0755);
	104
	105	btrfs_set_root_bytenr(&root_item, leaf->start);
	106	btrfs_set_root_level(&root_item, 0);
	107	btrfs_set_root_refs(&root_item, 1);
	108	btrfs_set_root_used(&root_item, 0);
	109
	110	memset(&root_item.drop_progress, 0, sizeof(root_item.drop_progress));
	111	root_item.drop_level = 0;
	112
925baedd	113	btrfs_tree_unlock(leaf);
f46b5a66 CH	114	free_extent_buffer(leaf);
	115	leaf = NULL;
	116
	117	btrfs_set_root_dirid(&root_item, new_dirid);
	118
	119	key.objectid = objectid;
	120	key.offset = 1;
	121	btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
	122	ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key,
	123	&root_item);
	124	if (ret)
	125	goto fail;
	126
	127	/*
	128	* insert the directory item
	129	*/
	130	key.offset = (u64)-1;
	131	dir = root->fs_info->sb->s_root->d_inode;
	132	ret = btrfs_insert_dir_item(trans, root->fs_info->tree_root,
	133	name, namelen, dir->i_ino, &key,
aec7477b	134	BTRFS_FT_DIR, 0);
f46b5a66 CH	135	if (ret)
	136	goto fail;
	137
	138	ret = btrfs_insert_inode_ref(trans, root->fs_info->tree_root,
	139	name, namelen, objectid,
aec7477b	140	root->fs_info->sb->s_root->d_inode->i_ino, 0);
f46b5a66 CH	141	if (ret)
	142	goto fail;
	143
	144	ret = btrfs_commit_transaction(trans, root);
	145	if (ret)
	146	goto fail_commit;
	147
	148	new_root = btrfs_read_fs_root(root->fs_info, &key, name, namelen);
	149	BUG_ON(!new_root);
	150
	151	trans = btrfs_start_transaction(new_root, 1);
	152	BUG_ON(!trans);
	153
	154	ret = btrfs_create_subvol_root(new_root, trans, new_dirid,
	155	BTRFS_I(dir)->block_group);
	156	if (ret)
	157	goto fail;
	158
	159	/* Invalidate existing dcache entry for new subvolume. */
	160	btrfs_invalidate_dcache_root(root, name, namelen);
	161
	162	fail:
	163	nr = trans->blocks_used;
	164	err = btrfs_commit_transaction(trans, new_root);
	165	if (err && !ret)
	166	ret = err;
	167	fail_commit:
f46b5a66	168	btrfs_btree_balance_dirty(root, nr);
f46b5a66 CH	169	return ret;
	170	}
	171
	172	static int create_snapshot(struct btrfs_root root, char name, int namelen)
	173	{
	174	struct btrfs_pending_snapshot *pending_snapshot;
	175	struct btrfs_trans_handle *trans;
	176	int ret;
	177	int err;
	178	unsigned long nr = 0;
	179
	180	if (!root->ref_cows)
	181	return -EINVAL;
	182
f46b5a66 CH	183	ret = btrfs_check_free_space(root, 1, 0);
	184	if (ret)
	185	goto fail_unlock;
	186
	187	pending_snapshot = kmalloc(sizeof(*pending_snapshot), GFP_NOFS);
	188	if (!pending_snapshot) {
	189	ret = -ENOMEM;
	190	goto fail_unlock;
	191	}
	192	pending_snapshot->name = kmalloc(namelen + 1, GFP_NOFS);
	193	if (!pending_snapshot->name) {
	194	ret = -ENOMEM;
	195	kfree(pending_snapshot);
	196	goto fail_unlock;
	197	}
	198	memcpy(pending_snapshot->name, name, namelen);
	199	pending_snapshot->name[namelen] = '\0';
	200	trans = btrfs_start_transaction(root, 1);
	201	BUG_ON(!trans);
	202	pending_snapshot->root = root;
	203	list_add(&pending_snapshot->list,
	204	&trans->transaction->pending_snapshots);
	205	ret = btrfs_update_inode(trans, root, root->inode);
	206	err = btrfs_commit_transaction(trans, root);
	207
	208	fail_unlock:
f46b5a66	209	btrfs_btree_balance_dirty(root, nr);
f46b5a66 CH	210	return ret;
	211	}
	212
	213	int btrfs_defrag_file(struct file *file)
	214	{
	215	struct inode *inode = fdentry(file)->d_inode;
	216	struct btrfs_root *root = BTRFS_I(inode)->root;
	217	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
3eaa2885	218	struct btrfs_ordered_extent *ordered;
f46b5a66 CH	219	struct page *page;
	220	unsigned long last_index;
	221	unsigned long ra_pages = root->fs_info->bdi.ra_pages;
	222	unsigned long total_read = 0;
	223	u64 page_start;
	224	u64 page_end;
	225	unsigned long i;
	226	int ret;
	227
f46b5a66	228	ret = btrfs_check_free_space(root, inode->i_size, 0);
f46b5a66 CH	229	if (ret)
	230	return -ENOSPC;
	231
	232	mutex_lock(&inode->i_mutex);
	233	last_index = inode->i_size >> PAGE_CACHE_SHIFT;
	234	for (i = 0; i <= last_index; i++) {
	235	if (total_read % ra_pages == 0) {
	236	btrfs_force_ra(inode->i_mapping, &file->f_ra, file, i,
	237	min(last_index, i + ra_pages - 1));
	238	}
	239	total_read++;
3eaa2885	240	again:
f46b5a66 CH	241	page = grab_cache_page(inode->i_mapping, i);
	242	if (!page)
	243	goto out_unlock;
	244	if (!PageUptodate(page)) {
	245	btrfs_readpage(NULL, page);
	246	lock_page(page);
	247	if (!PageUptodate(page)) {
	248	unlock_page(page);
	249	page_cache_release(page);
	250	goto out_unlock;
	251	}
	252	}
	253
f46b5a66	254	wait_on_page_writeback(page);
f46b5a66 CH	255
	256	page_start = (u64)page->index << PAGE_CACHE_SHIFT;
	257	page_end = page_start + PAGE_CACHE_SIZE - 1;
f46b5a66	258	lock_extent(io_tree, page_start, page_end, GFP_NOFS);
3eaa2885 CM	259
	260	ordered = btrfs_lookup_ordered_extent(inode, page_start);
	261	if (ordered) {
	262	unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
	263	unlock_page(page);
	264	page_cache_release(page);
	265	btrfs_start_ordered_extent(inode, ordered, 1);
	266	btrfs_put_ordered_extent(ordered);
	267	goto again;
	268	}
	269	set_page_extent_mapped(page);
	270
f46b5a66 CH	271	set_extent_delalloc(io_tree, page_start,
	272	page_end, GFP_NOFS);
	273
	274	unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
	275	set_page_dirty(page);
	276	unlock_page(page);
	277	page_cache_release(page);
	278	balance_dirty_pages_ratelimited_nr(inode->i_mapping, 1);
	279	}
	280
	281	out_unlock:
	282	mutex_unlock(&inode->i_mutex);
	283	return 0;
	284	}
	285
	286	/*
	287	* Called inside transaction, so use GFP_NOFS
	288	*/
	289
	290	static int btrfs_ioctl_resize(struct btrfs_root root, void __user arg)
	291	{
	292	u64 new_size;
	293	u64 old_size;
	294	u64 devid = 1;
	295	struct btrfs_ioctl_vol_args *vol_args;
	296	struct btrfs_trans_handle *trans;
	297	struct btrfs_device *device = NULL;
	298	char *sizestr;
	299	char *devstr = NULL;
	300	int ret = 0;
	301	int namelen;
	302	int mod = 0;
	303
	304	vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
	305
	306	if (!vol_args)
	307	return -ENOMEM;
	308
	309	if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
	310	ret = -EFAULT;
	311	goto out;
	312	}
5516e595 MF	313
5516e595 MF	314	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
f46b5a66	315	namelen = strlen(vol_args->name);
f46b5a66	316
7d9eb12c	317	mutex_lock(&root->fs_info->volume_mutex);
f46b5a66 CH	318	sizestr = vol_args->name;
	319	devstr = strchr(sizestr, ':');
	320	if (devstr) {
	321	char *end;
	322	sizestr = devstr + 1;
	323	*devstr = '\0';
	324	devstr = vol_args->name;
	325	devid = simple_strtoull(devstr, &end, 10);
	326	printk(KERN_INFO "resizing devid %llu\n", devid);
	327	}
	328	device = btrfs_find_device(root, devid, NULL);
	329	if (!device) {
	330	printk(KERN_INFO "resizer unable to find device %llu\n", devid);
	331	ret = -EINVAL;
	332	goto out_unlock;
	333	}
	334	if (!strcmp(sizestr, "max"))
	335	new_size = device->bdev->bd_inode->i_size;
	336	else {
	337	if (sizestr[0] == '-') {
	338	mod = -1;
	339	sizestr++;
	340	} else if (sizestr[0] == '+') {
	341	mod = 1;
	342	sizestr++;
	343	}
	344	new_size = btrfs_parse_size(sizestr);
	345	if (new_size == 0) {
	346	ret = -EINVAL;
	347	goto out_unlock;
	348	}
	349	}
	350
	351	old_size = device->total_bytes;
	352
	353	if (mod < 0) {
	354	if (new_size > old_size) {
	355	ret = -EINVAL;
	356	goto out_unlock;
	357	}
	358	new_size = old_size - new_size;
	359	} else if (mod > 0) {
	360	new_size = old_size + new_size;
	361	}
	362
	363	if (new_size < 256 * 1024 * 1024) {
	364	ret = -EINVAL;
	365	goto out_unlock;
	366	}
	367	if (new_size > device->bdev->bd_inode->i_size) {
	368	ret = -EFBIG;
	369	goto out_unlock;
	370	}
	371
	372	do_div(new_size, root->sectorsize);
	373	new_size *= root->sectorsize;
	374
	375	printk(KERN_INFO "new size for %s is %llu\n",
	376	device->name, (unsigned long long)new_size);
	377
	378	if (new_size > old_size) {
	379	trans = btrfs_start_transaction(root, 1);
	380	ret = btrfs_grow_device(trans, device, new_size);
	381	btrfs_commit_transaction(trans, root);
382	} else {
383	ret = btrfs_shrink_device(device, new_size);
384	}
385
386	out_unlock:
7d9eb12c	387	mutex_unlock(&root->fs_info->volume_mutex);
f46b5a66 CH	388	out:
	389	kfree(vol_args);
	390	return ret;
	391	}
	392
	393	static noinline int btrfs_ioctl_snap_create(struct btrfs_root *root,
	394	void __user *arg)
	395	{
	396	struct btrfs_ioctl_vol_args *vol_args;
	397	struct btrfs_dir_item *di;
	398	struct btrfs_path *path;
	399	u64 root_dirid;
	400	int namelen;
	401	int ret;
	402
	403	vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
	404
	405	if (!vol_args)
	406	return -ENOMEM;
	407
	408	if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
	409	ret = -EFAULT;
	410	goto out;
	411	}
	412
5516e595	413	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
f46b5a66	414	namelen = strlen(vol_args->name);
f46b5a66 CH	415	if (strchr(vol_args->name, '/')) {
	416	ret = -EINVAL;
	417	goto out;
	418	}
	419
	420	path = btrfs_alloc_path();
	421	if (!path) {
	422	ret = -ENOMEM;
	423	goto out;
	424	}
	425
	426	root_dirid = root->fs_info->sb->s_root->d_inode->i_ino,
f46b5a66 CH	427	di = btrfs_lookup_dir_item(NULL, root->fs_info->tree_root,
	428	path, root_dirid,
	429	vol_args->name, namelen, 0);
f46b5a66 CH	430	btrfs_free_path(path);
	431
	432	if (di && !IS_ERR(di)) {
	433	ret = -EEXIST;
	434	goto out;
	435	}
	436
	437	if (IS_ERR(di)) {
	438	ret = PTR_ERR(di);
	439	goto out;
	440	}
	441
a2135011	442	mutex_lock(&root->fs_info->drop_mutex);
f46b5a66 CH	443	if (root == root->fs_info->tree_root)
	444	ret = create_subvol(root, vol_args->name, namelen);
	445	else
	446	ret = create_snapshot(root, vol_args->name, namelen);
a2135011	447	mutex_unlock(&root->fs_info->drop_mutex);
f46b5a66 CH	448	out:
	449	kfree(vol_args);
	450	return ret;
	451	}
	452
	453	static int btrfs_ioctl_defrag(struct file *file)
	454	{
	455	struct inode *inode = fdentry(file)->d_inode;
	456	struct btrfs_root *root = BTRFS_I(inode)->root;
	457
	458	switch (inode->i_mode & S_IFMT) {
	459	case S_IFDIR:
f46b5a66 CH	460	btrfs_defrag_root(root, 0);
f46b5a66 CH	461	btrfs_defrag_root(root->fs_info->extent_root, 0);
f46b5a66 CH	462	break;
	463	case S_IFREG:
	464	btrfs_defrag_file(file);
	465	break;
	466	}
	467
	468	return 0;
	469	}
	470
	471	long btrfs_ioctl_add_dev(struct btrfs_root root, void __user arg)
	472	{
	473	struct btrfs_ioctl_vol_args *vol_args;
	474	int ret;
	475
	476	vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
	477
	478	if (!vol_args)
	479	return -ENOMEM;
	480
	481	if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
	482	ret = -EFAULT;
	483	goto out;
	484	}
5516e595	485	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
f46b5a66 CH	486	ret = btrfs_init_new_device(root, vol_args->name);
	487
	488	out:
	489	kfree(vol_args);
	490	return ret;
	491	}
	492
	493	long btrfs_ioctl_rm_dev(struct btrfs_root root, void __user arg)
	494	{
	495	struct btrfs_ioctl_vol_args *vol_args;
	496	int ret;
	497
	498	vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
	499
	500	if (!vol_args)
	501	return -ENOMEM;
	502
	503	if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
	504	ret = -EFAULT;
	505	goto out;
	506	}
5516e595	507	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
f46b5a66 CH	508	ret = btrfs_rm_device(root, vol_args->name);
	509
	510	out:
	511	kfree(vol_args);
	512	return ret;
	513	}
	514
	515	int dup_item_to_inode(struct btrfs_trans_handle *trans,
	516	struct btrfs_root *root,
	517	struct btrfs_path *path,
	518	struct extent_buffer *leaf,
	519	int slot,
	520	struct btrfs_key *key,
	521	u64 destino)
	522	{
	523	char *dup;
	524	int len = btrfs_item_size_nr(leaf, slot);
	525	struct btrfs_key ckey = *key;
	526	int ret = 0;
	527
	528	dup = kmalloc(len, GFP_NOFS);
	529	if (!dup)
	530	return -ENOMEM;
	531
	532	read_extent_buffer(leaf, dup, btrfs_item_ptr_offset(leaf, slot), len);
	533	btrfs_release_path(root, path);
	534
	535	ckey.objectid = destino;
	536	ret = btrfs_insert_item(trans, root, &ckey, dup, len);
	537	kfree(dup);
	538	return ret;
	539	}
	540
	541	long btrfs_ioctl_clone(struct file *file, unsigned long src_fd)
	542	{
	543	struct inode *inode = fdentry(file)->d_inode;
	544	struct btrfs_root *root = BTRFS_I(inode)->root;
	545	struct file *src_file;
	546	struct inode *src;
	547	struct btrfs_trans_handle *trans;
	548	int ret;
	549	u64 pos;
	550	struct btrfs_path *path;
	551	struct btrfs_key key;
	552	struct extent_buffer *leaf;
	553	u32 nritems;
	554	int slot;
	555
	556	src_file = fget(src_fd);
	557	if (!src_file)
	558	return -EBADF;
	559	src = src_file->f_dentry->d_inode;
	560
	561	ret = -EXDEV;
	562	if (src->i_sb != inode->i_sb)
	563	goto out_fput;
	564
	565	if (inode < src) {
	566	mutex_lock(&inode->i_mutex);
	567	mutex_lock(&src->i_mutex);
	568	} else {
	569	mutex_lock(&src->i_mutex);
	570	mutex_lock(&inode->i_mutex);
	571	}
572
573	ret = -ENOTEMPTY;
574	if (inode->i_size)
575	goto out_unlock;
576
577	/* do any pending delalloc/csum calc on src, one way or
578	another, and lock file content */
579	while (1) {
580	filemap_write_and_wait(src->i_mapping);
581	lock_extent(&BTRFS_I(src)->io_tree, 0, (u64)-1, GFP_NOFS);
582	if (BTRFS_I(src)->delalloc_bytes == 0)
583	break;
584	unlock_extent(&BTRFS_I(src)->io_tree, 0, (u64)-1, GFP_NOFS);
585	}
586
f46b5a66 CH	587	trans = btrfs_start_transaction(root, 0);
	588	path = btrfs_alloc_path();
	589	if (!path) {
	590	ret = -ENOMEM;
	591	goto out;
	592	}
	593	key.offset = 0;
	594	key.type = BTRFS_EXTENT_DATA_KEY;
	595	key.objectid = src->i_ino;
	596	pos = 0;
	597	path->reada = 2;
	598
	599	while (1) {
	600	/*
	601	* note the key will change type as we walk through the
	602	* tree.
	603	*/
	604	ret = btrfs_search_slot(trans, root, &key, path, 0, 0);
	605	if (ret < 0)
	606	goto out;
	607
	608	if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
	609	ret = btrfs_next_leaf(root, path);
	610	if (ret < 0)
	611	goto out;
	612	if (ret > 0)
	613	break;
	614	}
	615	leaf = path->nodes[0];
	616	slot = path->slots[0];
	617	btrfs_item_key_to_cpu(leaf, &key, slot);
	618	nritems = btrfs_header_nritems(leaf);
	619
	620	if (btrfs_key_type(&key) > BTRFS_CSUM_ITEM_KEY \|\|
	621	key.objectid != src->i_ino)
	622	break;
	623
	624	if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) {
	625	struct btrfs_file_extent_item *extent;
	626	int found_type;
	627	pos = key.offset;
	628	extent = btrfs_item_ptr(leaf, slot,
	629	struct btrfs_file_extent_item);
	630	found_type = btrfs_file_extent_type(leaf, extent);
	631	if (found_type == BTRFS_FILE_EXTENT_REG) {
	632	u64 len = btrfs_file_extent_num_bytes(leaf,
	633	extent);
	634	u64 ds = btrfs_file_extent_disk_bytenr(leaf,
	635	extent);
	636	u64 dl = btrfs_file_extent_disk_num_bytes(leaf,
	637	extent);
	638	u64 off = btrfs_file_extent_offset(leaf,
	639	extent);
	640	btrfs_insert_file_extent(trans, root,
	641	inode->i_ino, pos,
	642	ds, dl, len, off);
	643	/* ds == 0 means there's a hole */
	644	if (ds != 0) {
	645	btrfs_inc_extent_ref(trans, root,
	646	ds, dl,
	647	root->root_key.objectid,
	648	trans->transid,
	649	inode->i_ino, pos);
	650	}
651	pos = key.offset + len;
652	} else if (found_type == BTRFS_FILE_EXTENT_INLINE) {
653	ret = dup_item_to_inode(trans, root, path,
654	leaf, slot, &key,
655	inode->i_ino);
656	if (ret)
657	goto out;
658	pos = key.offset + btrfs_item_size_nr(leaf,
659	slot);
660	}
661	} else if (btrfs_key_type(&key) == BTRFS_CSUM_ITEM_KEY) {
662	ret = dup_item_to_inode(trans, root, path, leaf,
663	slot, &key, inode->i_ino);
664
665	if (ret)
666	goto out;
667	}
668	key.offset++;
669	btrfs_release_path(root, path);
670	}
671
672	ret = 0;
673	out:
674	btrfs_free_path(path);
675
676	inode->i_blocks = src->i_blocks;
677	i_size_write(inode, src->i_size);
678	btrfs_update_inode(trans, root, inode);
679
680	unlock_extent(&BTRFS_I(src)->io_tree, 0, (u64)-1, GFP_NOFS);
681
682	btrfs_end_transaction(trans, root);
f46b5a66 CH	683
	684	out_unlock:
	685	mutex_unlock(&src->i_mutex);
	686	mutex_unlock(&inode->i_mutex);
	687	out_fput:
	688	fput(src_file);
	689	return ret;
	690	}
	691
	692	/*
	693	* there are many ways the trans_start and trans_end ioctls can lead
	694	* to deadlocks. They should only be used by applications that
	695	* basically own the machine, and have a very in depth understanding
	696	* of all the possible deadlocks and enospc problems.
	697	*/
	698	long btrfs_ioctl_trans_start(struct file *file)
	699	{
	700	struct inode *inode = fdentry(file)->d_inode;
	701	struct btrfs_root *root = BTRFS_I(inode)->root;
	702	struct btrfs_trans_handle *trans;
	703	int ret = 0;
	704
df5b5520 CH	705	if (!capable(CAP_SYS_ADMIN))
	706	return -EPERM;
	707
f46b5a66 CH	708	if (file->private_data) {
	709	ret = -EINPROGRESS;
	710	goto out;
	711	}
	712	trans = btrfs_start_transaction(root, 0);
	713	if (trans)
	714	file->private_data = trans;
	715	else
	716	ret = -ENOMEM;
	717	/printk(KERN_INFO "btrfs_ioctl_trans_start on %p\n", file);/
	718	out:
f46b5a66 CH	719	return ret;
	720	}
	721
	722	/*
	723	* there are many ways the trans_start and trans_end ioctls can lead
	724	* to deadlocks. They should only be used by applications that
	725	* basically own the machine, and have a very in depth understanding
	726	* of all the possible deadlocks and enospc problems.
	727	*/
	728	long btrfs_ioctl_trans_end(struct file *file)
	729	{
	730	struct inode *inode = fdentry(file)->d_inode;
	731	struct btrfs_root *root = BTRFS_I(inode)->root;
	732	struct btrfs_trans_handle *trans;
	733	int ret = 0;
	734
f46b5a66 CH	735	trans = file->private_data;
	736	if (!trans) {
	737	ret = -EINVAL;
	738	goto out;
	739	}
	740	btrfs_end_transaction(trans, root);
	741	file->private_data = 0;
	742	out:
f46b5a66 CH	743	return ret;
	744	}
	745
	746	long btrfs_ioctl(struct file *file, unsigned int
	747	cmd, unsigned long arg)
	748	{
	749	struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
	750
	751	switch (cmd) {
	752	case BTRFS_IOC_SNAP_CREATE:
	753	return btrfs_ioctl_snap_create(root, (void __user *)arg);
	754	case BTRFS_IOC_DEFRAG:
	755	return btrfs_ioctl_defrag(file);
	756	case BTRFS_IOC_RESIZE:
	757	return btrfs_ioctl_resize(root, (void __user *)arg);
	758	case BTRFS_IOC_ADD_DEV:
	759	return btrfs_ioctl_add_dev(root, (void __user *)arg);
	760	case BTRFS_IOC_RM_DEV:
	761	return btrfs_ioctl_rm_dev(root, (void __user *)arg);
	762	case BTRFS_IOC_BALANCE:
	763	return btrfs_balance(root->fs_info->dev_root);
	764	case BTRFS_IOC_CLONE:
	765	return btrfs_ioctl_clone(file, arg);
	766	case BTRFS_IOC_TRANS_START:
	767	return btrfs_ioctl_trans_start(file);
	768	case BTRFS_IOC_TRANS_END:
	769	return btrfs_ioctl_trans_end(file);
	770	case BTRFS_IOC_SYNC:
	771	btrfs_sync_fs(file->f_dentry->d_sb, 1);
	772	return 0;
	773	}
	774
	775	return -ENOTTY;
	776	}