>>I think the problem is a little different. It's not that we're
>>manipulating the partent, it's we want to manipulate the clone.
>
>	there's also a behavior change that affects the manipulation of cloned
>	child.  if you have an environment to repeat the symptom, pls try it
>	with the latest kernel.

	i mean, the behavior change is already in the tree.

itojun