Re: [RFC PATCH v3 7/7] btrfs: enable swap file support

[Omar Sandoval <osandov@osandov.com>]

On Fri, Dec 12, 2014 at 11:51:22AM +0100, David Sterba wrote:
> On Tue, Dec 09, 2014 at 05:45:48PM -0800, Omar Sandoval wrote:
> > +static void __clear_swapfile_extents(struct inode *inode)
> > +{
> > +	u64 isize = inode->i_size;
> > +	struct extent_map *em;
> > +	u64 start, len;
> > +
> > +	start = 0;
> > +	while (start < isize) {
> > +		len = isize - start;
> > +		em = btrfs_get_extent(inode, NULL, 0, start, len, 0);
> > +		if (IS_ERR(em))
> > +			return;
> 
> This could transiently fail if there's no memory to allocate the em, and
> would leak the following extents.
> 
This leak I was aware of, and at the time I didn't see a good way to get
around it. After all, if we can't get the current extent, there's no way
to iterate through the rest of them. Now I see that instead of doing
this at the btrfs_get_extent level, I can just go through all of the
extent_maps in the extent_map_tree.

> > +
> > +		clear_bit(EXTENT_FLAG_SWAPFILE, &em->flags);
> > +
> > +		start = extent_map_end(em);
> > +		free_extent_map(em);
> > +	}
> > +}
> > +
> > +static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file,
> > +			       sector_t *span)
> > +{
> > +	struct inode *inode = file_inode(file);
> > +	struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
> > +	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
> > +	int ret = 0;
> > +	u64 isize = inode->i_size;
> > +	struct extent_state *cached_state = NULL;
> > +	struct extent_map *em;
> > +	u64 start, len;
> > +
> > +	if (BTRFS_I(inode)->flags & BTRFS_INODE_COMPRESS) {
> > +		/* Can't do direct I/O on a compressed file. */
> > +		btrfs_err(fs_info, "swapfile is compressed");
> > +		return -EINVAL;
> > +	}
> > +	if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW)) {
> > +		/*
> > +		 * Going through the copy-on-write path while swapping pages
> > +		 * in/out and doing a bunch of allocations could stress the
> > +		 * memory management code that got us there in the first place,
> > +		 * and that's sure to be a bad time.
> > +		 */
> > +		btrfs_err(fs_info, "swapfile is copy-on-write");
> > +		return -EINVAL;
> > +	}
> > +
> > +	lock_extent_bits(io_tree, 0, isize - 1, 0, &cached_state);
> > +
> > +	/*
> > +	 * All of the extents must be allocated and support direct I/O. Inline
> > +	 * extents and compressed extents fall back to buffered I/O, so those
> > +	 * are no good. Additionally, all of the extents must be safe for nocow.
> > +	 */
> > +	atomic_inc(&BTRFS_I(inode)->root->nr_swapfiles);
> > +	start = 0;
> > +	while (start < isize) {
> > +		len = isize - start;
> > +		em = btrfs_get_extent(inode, NULL, 0, start, len, 0);
> > +		if (IS_ERR(em)) {
> 
> 		IS_ERR_OR_NULL(em)
> 
> From now on the em is valid and has to be free_extent_map()ed ...
> 
> > +			ret = PTR_ERR(em);
> > +			goto out;
> > +		}
> > +
> > +		if (test_bit(EXTENT_FLAG_VACANCY, &em->flags) ||
> > +		    em->block_start == EXTENT_MAP_HOLE) {
> > +			btrfs_err(fs_info, "swapfile has holes");
> > +			ret = -EINVAL;
> 
> ... and all the error branches would miss it.
> 
> > +			goto out;
> > +		}
> > +		if (em->block_start == EXTENT_MAP_INLINE) {
> > +			/*
> > +			 * It's unlikely we'll ever actually find ourselves
> > +			 * here, as a file small enough to fit inline won't be
> > +			 * big enough to store more than the swap header, but in
> > +			 * case something changes in the future, let's catch it
> > +			 * here rather than later.
> > +			 */
> > +			btrfs_err(fs_info, "swapfile is inline");
> > +			ret = -EINVAL;
> 
> here
> 
> > +			goto out;
> > +		}
> > +		if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) {
> > +			btrfs_err(fs_info, "swapfile is compresed");
> > +			ret = -EINVAL;
> 
> here
> 
> > +			goto out;
> > +		}
> > +		ret = can_nocow_extent(inode, start, &len, NULL, NULL, NULL);
> > +		if (ret < 0) {
> 
> here
> 
> > +			goto out;
> > +		} else if (ret == 1) {
> > +			ret = 0;
> > +		} else {
> > +			btrfs_err(fs_info, "swapfile has extent requiring COW (%llu-%llu)",
> > +				  start, start + len - 1);
> > +			ret = -EINVAL;
> 
> here
> 
> > +			goto out;
> > +		}
> > +
> > +		set_bit(EXTENT_FLAG_SWAPFILE, &em->flags);
> > +
> > +		start = extent_map_end(em);
> > +		free_extent_map(em);
> > +	}
> > +
> > +out:
> > +	if (ret) {
> 
> should be fixed by:
> 
> 		if (!IS_ERR_OR_NULL(em))
> 			free_extent_map(em);
> 
> > +		__clear_swapfile_extents(inode);
> > +		atomic_dec(&BTRFS_I(inode)->root->nr_swapfiles);
> > +	}
> > +	unlock_extent_cached(io_tree, 0, isize - 1, &cached_state, GFP_NOFS);
> > +	return ret;
> > +}
This leak I completely missed. Thanks.

-- 
Omar

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>