It seems to me that we leak the memory allocated to 'value' in
btrfs_get_acl() if the call to posix_acl_from_xattr() fails.
Here's a patch that attempts to correct that problem.
Signed-off-by: Jesper Juhl <jj@chaosbits.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
                size = __btrfs_getxattr(inode, name, value, size);
                if (size > 0) {
                        acl = posix_acl_from_xattr(value, size);
-                       if (IS_ERR(acl))
+                       if (IS_ERR(acl)) {
+                               kfree(value);
                                return acl;
+                       }
                        set_cached_acl(inode, type, acl);
                }
                kfree(value);