Re: [RFC PATCH 3/4] kernel/workqueue: Distinguish between general unbound and WQ_SYSFS cpumask changes

[Juri Lelli]

On 17/01/24 09:42, Tejun Heo wrote:
> Hello,
> 
> On Wed, Jan 17, 2024 at 02:32:34PM -0500, Waiman Long wrote:
> > My impression is that changing the workqueue cpumask of ordered unbound
> > workqueue may break the ordering guarantee momentarily. I was planning to
> 
> Ah, you're right. Changing cpumask would require changing the dfl_pwq and
> that can introduce extra concurrency and break ordering and it's exempt from
> unbound_cpumask updates. We likely need to add a mechanism for updating
> ordered wq's so that the new pwq doesn't become until the previous one is
> drained.

Thanks for the additional info! Guess I'll need to think more about this
and possibly coordinate the effort with Waiman.

Best,
Juri