Re: [RFC] [PATCH] Cgroup based OOM killer controller

[Evgeniy Polyakov <zbr@ioremap.net>]

On Tue, Jan 27, 2009 at 09:10:53PM +0530, Balbir Singh (balbir@linux.vnet.ibm.com) wrote:
> > Having some special application which will monitor /dev/mem_notify and
> > kill processes based on its own hueristics is a good idea, but when it
> > fails to do its work (or does not exist) system has to have ability to
> > make a progress and invoke a main oom-killer.
> 
> The last part is what we've discussed in the mini-summit. There should
> be OOM kill notification to user space and if that fails let the kernel
> invoke the OOM killer. The exact interface for notification is not
> interesting, one could use netlink if that works well.

Yes, that's exactly what I would like to see.
Btw, netlink will not be a good idea, since it requires additional (and
quite big) allocation. I believe just reading the char device (or maybe
having a syscall) is enough, but its up to the implementation.

-- 
	Evgeniy Polyakov