Commit 9b49c9b9 authored by Chris Mason's avatar Chris Mason

Btrfs: Fix allocation completions in tree log replay

After a crash, the tree log code uses btrfs_alloc_logged_extent to
record allocations of data extents that it finds in the log tree.  These
come in basically random order, which does not fit how
btrfs_remove_free_space() expects to be called.

btrfs_remove_free_space was changed to support recording an extent
allocation in the middle of a region of free space.
Signed-off-by: default avatarChris Mason <chris.mason@oracle.com>
parent 60582d1e
......@@ -306,11 +306,45 @@ int btrfs_remove_free_space(struct btrfs_block_group_cache *block_group,
ret = link_free_space(block_group, info);
BUG_ON(ret);
} else if (info && info->offset < offset &&
info->offset + info->bytes >= offset + bytes) {
u64 old_start = info->offset;
/*
* we're freeing space in the middle of the info,
* this can happen during tree log replay
*
* first unlink the old info and then
* insert it again after the hole we're creating
*/
unlink_free_space(block_group, info);
if (offset + bytes < info->offset + info->bytes) {
u64 old_end = info->offset + info->bytes;
info->offset = offset + bytes;
info->bytes = old_end - info->offset;
ret = link_free_space(block_group, info);
BUG_ON(ret);
} else {
/* the hole we're creating ends at the end
* of the info struct, just free the info
*/
kfree(info);
}
/* step two, insert a new info struct to cover anything
* before the hole
*/
spin_unlock(&block_group->lock);
ret = btrfs_add_free_space(block_group, old_start,
offset - old_start);
BUG_ON(ret);
goto out_nolock;
} else {
WARN_ON(1);
}
out:
spin_unlock(&block_group->lock);
out_nolock:
return ret;
}
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment