On Tue 21-08-12 13:40:45, Glauber Costa wrote:

On 08/21/2012 01:35 PM, Michal Hocko wrote:

[...]

quoted

I am asking because this should trigger memcg-oom
but that one will usually pick up something else than the fork bomb
which would have a small memory footprint. But that needs to be handled
on the oom level obviously.

Sure, but keep in mind that the main protection is against tasks *not*
in this memcg.

Yes and that's is good step forward. I just wanted to mention that we
still have the problem inside the subhierarchy. The changelog was not
specific enough.
-- 
Michal Hocko
SUSE Labs