[GitHub/mt8127/android_kernel_alcatel_ttab.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, 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;
	}
	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,
aec7477b	132	BTRFS_FT_DIR, 0);
f46b5a66 CH	133	if (ret)
	134	goto fail;
	135
	136	ret = btrfs_insert_inode_ref(trans, root->fs_info->tree_root,
	137	name, namelen, objectid,
aec7477b	138	root->fs_info->sb->s_root->d_inode->i_ino, 0);
f46b5a66 CH	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;
3eaa2885	216	struct btrfs_ordered_extent *ordered;
f46b5a66 CH	217	struct page *page;
	218	unsigned long last_index;
	219	unsigned long ra_pages = root->fs_info->bdi.ra_pages;
	220	unsigned long total_read = 0;
	221	u64 page_start;
	222	u64 page_end;
	223	unsigned long i;
	224	int ret;
	225
f46b5a66	226	ret = btrfs_check_free_space(root, inode->i_size, 0);
f46b5a66 CH	227	if (ret)
	228	return -ENOSPC;
	229
	230	mutex_lock(&inode->i_mutex);
	231	last_index = inode->i_size >> PAGE_CACHE_SHIFT;
	232	for (i = 0; i <= last_index; i++) {
	233	if (total_read % ra_pages == 0) {
	234	btrfs_force_ra(inode->i_mapping, &file->f_ra, file, i,
	235	min(last_index, i + ra_pages - 1));
	236	}
	237	total_read++;
3eaa2885	238	again:
f46b5a66 CH	239	page = grab_cache_page(inode->i_mapping, i);
	240	if (!page)
	241	goto out_unlock;
	242	if (!PageUptodate(page)) {
	243	btrfs_readpage(NULL, page);
	244	lock_page(page);
	245	if (!PageUptodate(page)) {
	246	unlock_page(page);
	247	page_cache_release(page);
	248	goto out_unlock;
	249	}
	250	}
	251
f46b5a66	252	wait_on_page_writeback(page);
f46b5a66 CH	253
	254	page_start = (u64)page->index << PAGE_CACHE_SHIFT;
	255	page_end = page_start + PAGE_CACHE_SIZE - 1;
f46b5a66	256	lock_extent(io_tree, page_start, page_end, GFP_NOFS);
3eaa2885 CM	257
	258	ordered = btrfs_lookup_ordered_extent(inode, page_start);
	259	if (ordered) {
	260	unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
	261	unlock_page(page);
	262	page_cache_release(page);
	263	btrfs_start_ordered_extent(inode, ordered, 1);
	264	btrfs_put_ordered_extent(ordered);
	265	goto again;
	266	}
	267	set_page_extent_mapped(page);
	268
f46b5a66 CH	269	set_extent_delalloc(io_tree, page_start,
	270	page_end, GFP_NOFS);
	271
	272	unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
	273	set_page_dirty(page);
	274	unlock_page(page);
	275	page_cache_release(page);
	276	balance_dirty_pages_ratelimited_nr(inode->i_mapping, 1);
	277	}
	278
	279	out_unlock:
	280	mutex_unlock(&inode->i_mutex);
	281	return 0;
	282	}
	283
	284	/*
	285	* Called inside transaction, so use GFP_NOFS
	286	*/
	287
	288	static int btrfs_ioctl_resize(struct btrfs_root root, void __user arg)
	289	{
	290	u64 new_size;
	291	u64 old_size;
	292	u64 devid = 1;
	293	struct btrfs_ioctl_vol_args *vol_args;
	294	struct btrfs_trans_handle *trans;
	295	struct btrfs_device *device = NULL;
	296	char *sizestr;
	297	char *devstr = NULL;
	298	int ret = 0;
	299	int namelen;
	300	int mod = 0;
	301
	302	vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
	303
	304	if (!vol_args)
	305	return -ENOMEM;
	306
	307	if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
	308	ret = -EFAULT;
	309	goto out;
	310	}
	311	namelen = strlen(vol_args->name);
	312	if (namelen > BTRFS_VOL_NAME_MAX) {
	313	ret = -EINVAL;
	314	goto out;
	315	}
	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
	413	namelen = strlen(vol_args->name);
	414	if (namelen > BTRFS_VOL_NAME_MAX) {
	415	ret = -EINVAL;
	416	goto out;
	417	}
	418	if (strchr(vol_args->name, '/')) {
	419	ret = -EINVAL;
	420	goto out;
	421	}
	422
	423	path = btrfs_alloc_path();
	424	if (!path) {
	425	ret = -ENOMEM;
	426	goto out;
	427	}
	428
	429	root_dirid = root->fs_info->sb->s_root->d_inode->i_ino,
f46b5a66 CH	430	di = btrfs_lookup_dir_item(NULL, root->fs_info->tree_root,
	431	path, root_dirid,
	432	vol_args->name, namelen, 0);
f46b5a66 CH	433	btrfs_free_path(path);
	434
	435	if (di && !IS_ERR(di)) {
	436	ret = -EEXIST;
	437	goto out;
	438	}
	439
	440	if (IS_ERR(di)) {
	441	ret = PTR_ERR(di);
	442	goto out;
	443	}
	444
a2135011	445	mutex_lock(&root->fs_info->drop_mutex);
f46b5a66 CH	446	if (root == root->fs_info->tree_root)
	447	ret = create_subvol(root, vol_args->name, namelen);
	448	else
	449	ret = create_snapshot(root, vol_args->name, namelen);
a2135011	450	mutex_unlock(&root->fs_info->drop_mutex);
f46b5a66 CH	451	out:
	452	kfree(vol_args);
	453	return ret;
	454	}
	455
	456	static int btrfs_ioctl_defrag(struct file *file)
	457	{
	458	struct inode *inode = fdentry(file)->d_inode;
	459	struct btrfs_root *root = BTRFS_I(inode)->root;
	460
	461	switch (inode->i_mode & S_IFMT) {
	462	case S_IFDIR:
f46b5a66 CH	463	btrfs_defrag_root(root, 0);
f46b5a66 CH	464	btrfs_defrag_root(root->fs_info->extent_root, 0);
f46b5a66 CH	465	break;
	466	case S_IFREG:
	467	btrfs_defrag_file(file);
	468	break;
	469	}
	470
	471	return 0;
	472	}
	473
	474	long btrfs_ioctl_add_dev(struct btrfs_root root, void __user arg)
	475	{
	476	struct btrfs_ioctl_vol_args *vol_args;
	477	int ret;
	478
	479	vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
	480
	481	if (!vol_args)
	482	return -ENOMEM;
	483
	484	if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
	485	ret = -EFAULT;
	486	goto out;
	487	}
	488	ret = btrfs_init_new_device(root, vol_args->name);
	489
	490	out:
	491	kfree(vol_args);
	492	return ret;
	493	}
	494
	495	long btrfs_ioctl_rm_dev(struct btrfs_root root, void __user arg)
	496	{
	497	struct btrfs_ioctl_vol_args *vol_args;
	498	int ret;
	499
	500	vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
	501
	502	if (!vol_args)
	503	return -ENOMEM;
	504
	505	if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
	506	ret = -EFAULT;
	507	goto out;
	508	}
	509	ret = btrfs_rm_device(root, vol_args->name);
	510
	511	out:
	512	kfree(vol_args);
	513	return ret;
	514	}
	515
	516	int dup_item_to_inode(struct btrfs_trans_handle *trans,
	517	struct btrfs_root *root,
	518	struct btrfs_path *path,
	519	struct extent_buffer *leaf,
	520	int slot,
	521	struct btrfs_key *key,
	522	u64 destino)
	523	{
	524	char *dup;
	525	int len = btrfs_item_size_nr(leaf, slot);
	526	struct btrfs_key ckey = *key;
	527	int ret = 0;
	528
529	dup = kmalloc(len, GFP_NOFS);
530	if (!dup)
531	return -ENOMEM;
532
533	read_extent_buffer(leaf, dup, btrfs_item_ptr_offset(leaf, slot), len);
534	btrfs_release_path(root, path);
535
536	ckey.objectid = destino;
537	ret = btrfs_insert_item(trans, root, &ckey, dup, len);
538	kfree(dup);
539	return ret;
540	}
541
542	long btrfs_ioctl_clone(struct file *file, unsigned long src_fd)
543	{
544	struct inode *inode = fdentry(file)->d_inode;
545	struct btrfs_root *root = BTRFS_I(inode)->root;
546	struct file *src_file;
547	struct inode *src;
548	struct btrfs_trans_handle *trans;
549	int ret;
550	u64 pos;
551	struct btrfs_path *path;
552	struct btrfs_key key;
553	struct extent_buffer *leaf;
554	u32 nritems;
555	int slot;
556
557	src_file = fget(src_fd);
558	if (!src_file)
559	return -EBADF;
560	src = src_file->f_dentry->d_inode;
561
562	ret = -EXDEV;
563	if (src->i_sb != inode->i_sb)
564	goto out_fput;
565
566	if (inode < src) {
567	mutex_lock(&inode->i_mutex);
568	mutex_lock(&src->i_mutex);
569	} else {
570	mutex_lock(&src->i_mutex);
571	mutex_lock(&inode->i_mutex);
572	}
573
574	ret = -ENOTEMPTY;
575	if (inode->i_size)
576	goto out_unlock;
577
578	/* do any pending delalloc/csum calc on src, one way or
579	another, and lock file content */
580	while (1) {
581	filemap_write_and_wait(src->i_mapping);
582	lock_extent(&BTRFS_I(src)->io_tree, 0, (u64)-1, GFP_NOFS);
583	if (BTRFS_I(src)->delalloc_bytes == 0)
584	break;
585	unlock_extent(&BTRFS_I(src)->io_tree, 0, (u64)-1, GFP_NOFS);
586	}
587
f46b5a66 CH	588	trans = btrfs_start_transaction(root, 0);
	589	path = btrfs_alloc_path();
	590	if (!path) {
	591	ret = -ENOMEM;
	592	goto out;
	593	}
	594	key.offset = 0;
	595	key.type = BTRFS_EXTENT_DATA_KEY;
	596	key.objectid = src->i_ino;
	597	pos = 0;
	598	path->reada = 2;
	599
	600	while (1) {
	601	/*
	602	* note the key will change type as we walk through the
	603	* tree.
	604	*/
	605	ret = btrfs_search_slot(trans, root, &key, path, 0, 0);
	606	if (ret < 0)
	607	goto out;
	608
	609	if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
	610	ret = btrfs_next_leaf(root, path);
	611	if (ret < 0)
	612	goto out;
	613	if (ret > 0)
	614	break;
	615	}
	616	leaf = path->nodes[0];
	617	slot = path->slots[0];
	618	btrfs_item_key_to_cpu(leaf, &key, slot);
	619	nritems = btrfs_header_nritems(leaf);
	620
	621	if (btrfs_key_type(&key) > BTRFS_CSUM_ITEM_KEY \|\|
	622	key.objectid != src->i_ino)
	623	break;
	624
	625	if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) {
	626	struct btrfs_file_extent_item *extent;
	627	int found_type;
	628	pos = key.offset;
	629	extent = btrfs_item_ptr(leaf, slot,
	630	struct btrfs_file_extent_item);
	631	found_type = btrfs_file_extent_type(leaf, extent);
	632	if (found_type == BTRFS_FILE_EXTENT_REG) {
	633	u64 len = btrfs_file_extent_num_bytes(leaf,
	634	extent);
	635	u64 ds = btrfs_file_extent_disk_bytenr(leaf,
	636	extent);
	637	u64 dl = btrfs_file_extent_disk_num_bytes(leaf,
	638	extent);
	639	u64 off = btrfs_file_extent_offset(leaf,
	640	extent);
	641	btrfs_insert_file_extent(trans, root,
	642	inode->i_ino, pos,
	643	ds, dl, len, off);
	644	/* ds == 0 means there's a hole */
	645	if (ds != 0) {
	646	btrfs_inc_extent_ref(trans, root,
	647	ds, dl,
	648	root->root_key.objectid,
	649	trans->transid,
	650	inode->i_ino, pos);
	651	}
652	pos = key.offset + len;
653	} else if (found_type == BTRFS_FILE_EXTENT_INLINE) {
654	ret = dup_item_to_inode(trans, root, path,
655	leaf, slot, &key,
656	inode->i_ino);
657	if (ret)
658	goto out;
659	pos = key.offset + btrfs_item_size_nr(leaf,
660	slot);
661	}
662	} else if (btrfs_key_type(&key) == BTRFS_CSUM_ITEM_KEY) {
663	ret = dup_item_to_inode(trans, root, path, leaf,
664	slot, &key, inode->i_ino);
665
666	if (ret)
667	goto out;
668	}
669	key.offset++;
670	btrfs_release_path(root, path);
671	}
672
673	ret = 0;
674	out:
675	btrfs_free_path(path);
676
677	inode->i_blocks = src->i_blocks;
678	i_size_write(inode, src->i_size);
679	btrfs_update_inode(trans, root, inode);
680
681	unlock_extent(&BTRFS_I(src)->io_tree, 0, (u64)-1, GFP_NOFS);
682
683	btrfs_end_transaction(trans, root);
f46b5a66 CH	684
	685	out_unlock:
	686	mutex_unlock(&src->i_mutex);
	687	mutex_unlock(&inode->i_mutex);
	688	out_fput:
	689	fput(src_file);
	690	return ret;
	691	}
	692
	693	/*
	694	* there are many ways the trans_start and trans_end ioctls can lead
	695	* to deadlocks. They should only be used by applications that
	696	* basically own the machine, and have a very in depth understanding
	697	* of all the possible deadlocks and enospc problems.
	698	*/
	699	long btrfs_ioctl_trans_start(struct file *file)
	700	{
	701	struct inode *inode = fdentry(file)->d_inode;
	702	struct btrfs_root *root = BTRFS_I(inode)->root;
	703	struct btrfs_trans_handle *trans;
	704	int ret = 0;
	705
df5b5520 CH	706	if (!capable(CAP_SYS_ADMIN))
	707	return -EPERM;
	708
f46b5a66 CH	709	if (file->private_data) {
	710	ret = -EINPROGRESS;
	711	goto out;
	712	}
	713	trans = btrfs_start_transaction(root, 0);
	714	if (trans)
	715	file->private_data = trans;
	716	else
	717	ret = -ENOMEM;
	718	/printk(KERN_INFO "btrfs_ioctl_trans_start on %p\n", file);/
	719	out:
f46b5a66 CH	720	return ret;
	721	}
	722
	723	/*
	724	* there are many ways the trans_start and trans_end ioctls can lead
	725	* to deadlocks. They should only be used by applications that
	726	* basically own the machine, and have a very in depth understanding
	727	* of all the possible deadlocks and enospc problems.
	728	*/
	729	long btrfs_ioctl_trans_end(struct file *file)
	730	{
	731	struct inode *inode = fdentry(file)->d_inode;
	732	struct btrfs_root *root = BTRFS_I(inode)->root;
	733	struct btrfs_trans_handle *trans;
	734	int ret = 0;
	735
f46b5a66 CH	736	trans = file->private_data;
	737	if (!trans) {
	738	ret = -EINVAL;
	739	goto out;
	740	}
	741	btrfs_end_transaction(trans, root);
	742	file->private_data = 0;
	743	out:
f46b5a66 CH	744	return ret;
	745	}
	746
	747	long btrfs_ioctl(struct file *file, unsigned int
	748	cmd, unsigned long arg)
	749	{
	750	struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
	751
	752	switch (cmd) {
	753	case BTRFS_IOC_SNAP_CREATE:
	754	return btrfs_ioctl_snap_create(root, (void __user *)arg);
	755	case BTRFS_IOC_DEFRAG:
	756	return btrfs_ioctl_defrag(file);
	757	case BTRFS_IOC_RESIZE:
	758	return btrfs_ioctl_resize(root, (void __user *)arg);
	759	case BTRFS_IOC_ADD_DEV:
	760	return btrfs_ioctl_add_dev(root, (void __user *)arg);
	761	case BTRFS_IOC_RM_DEV:
	762	return btrfs_ioctl_rm_dev(root, (void __user *)arg);
	763	case BTRFS_IOC_BALANCE:
	764	return btrfs_balance(root->fs_info->dev_root);
	765	case BTRFS_IOC_CLONE:
	766	return btrfs_ioctl_clone(file, arg);
	767	case BTRFS_IOC_TRANS_START:
	768	return btrfs_ioctl_trans_start(file);
	769	case BTRFS_IOC_TRANS_END:
	770	return btrfs_ioctl_trans_end(file);
	771	case BTRFS_IOC_SYNC:
	772	btrfs_sync_fs(file->f_dentry->d_sb, 1);
	773	return 0;
	774	}
	775
	776	return -ENOTTY;
	777	}