Btrfs: don't leak reloc root nodes on error
authorJosef Bacik <jbacik@fb.com>
Fri, 2 Sep 2016 19:25:43 +0000 (15:25 -0400)
committerDavid Sterba <dsterba@suse.com>
Mon, 26 Sep 2016 16:08:44 +0000 (18:08 +0200)
We don't track the reloc roots in any sort of normal way, so the only way the
root/commit_root nodes get free'd is if the relocation finishes successfully and
the reloc root is deleted.  Fix this by free'ing them in free_reloc_roots.
Thanks,

Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
fs/btrfs/relocation.c

index 8e7c78ba77d903aabcb898f84fbe0a70ee1420a9..3c0973a6857257814cbfb8597d291d41d7936bba 100644 (file)
@@ -2357,6 +2357,10 @@ void free_reloc_roots(struct list_head *list)
        while (!list_empty(list)) {
                reloc_root = list_entry(list->next, struct btrfs_root,
                                        root_list);
+               free_extent_buffer(reloc_root->node);
+               free_extent_buffer(reloc_root->commit_root);
+               reloc_root->node = NULL;
+               reloc_root->commit_root = NULL;
                __del_reloc_root(reloc_root);
        }
 }