Re: Crashes with 874bbfe600a6 in 3.18.25
From: Tejun Heo
Date: Fri Feb 05 2016 - 15:55:07 EST
Hello, Mike.
On Fri, Feb 05, 2016 at 09:47:11PM +0100, Mike Galbraith wrote:
> That very point is what makes it wrong for the workqueue code to ever
> target a work item. The instant it does target selection, correctness
> may be at stake, it doesn't know, thus it must assume the full onus,
> which it has neither the knowledge not the time to do. That's how we
> exploded on node = -1, trying to help out the user by doing his job,
I have a hard time seeing the NUMA_NO_NODE bug as something that
indicative of anything. It is a dumb bug from mm side which puts
everyone using cpu_to_node() at risk.
> but then not doing the whole job. IMHO, a better plan is to let the
> user screw it up all by himself.
What are you suggesting?
Thanks.
--
tejun