Re: Deadlock due to "cpuset: Make cpuset hotplug synchronous"

From: Tejun Heo
Date: Tue Apr 07 2020 - 09:35:25 EST


On Tue, Apr 07, 2020 at 02:07:26PM +0100, Qais Yousef wrote:
> On 04/06/20 10:34, Tejun Heo wrote:
> > On Mon, Apr 06, 2020 at 11:55:23AM +0100, Qais Yousef wrote:
> > > What would be the right approach to get a fix in then? We have been skipping
> > > this test for a while and we'd like to enable it but this failure is a
> > > blocking issue.
> >
> > Update the test so that it accounts for the async nature of the operation?
>
> With a sleep() between the hp ops and the read? Isn't this a nasty hack?

Some operations are just async. Why is that inherently wrong?

> I was just trying to find out what can be done for 5.7. But based on your
> response this is sadly a dead end. For no reason I can comprehend.

Because it's creating real issues while trying to address a non-real problem.
It's dumb. Fix the non-real part.

Thanks.

--
tejun