Hey, Michal.

On Thu, Nov 01, 2012 at 04:15:56PM +0100, Michal Hocko wrote:

quoted

	if (!cgroup_lock_live_group(parent))
		return -ENODEV;

        cgrp = kzalloc(sizeof(*cgrp), GFP_KERNEL);
        if (!cgrp)
                return -ENOMEM;

this needs to drop the lock of course but it doesn't make it much more
complicated...

Yeah, now look at the error exit path of the function.  You have to
reorder them too so that it matches the changed order above, which is
all fine and dandy, but one way is not necessarily better than the
other, so there's no good reason to incur that churn.

Thanks.

-- 
tejun