Subject: Re: workqueue(9), per-CPU queues [was: Re: soc zfs: taskqueue / workqueue]
To: Mindaugas R. <rmind@NetBSD.org>
From: Martin Husemann <martin@duskware.de>
List: tech-kern
Date: 07/11/2007 15:45:29
On Wed, Jul 11, 2007 at 04:41:03PM +0300, Mindaugas R. wrote:
> Why entire struct workqueue, not only struct workqueue_queue? Such was my
> primary intention, but I am not sure how it should behave in CPUs detaching
> case?

I guess CPUs detaching could be handled by just setting a "gone" flag in the
cpuinfo, and clearing it (reviving the cpuid) if a cpu attaches at the same
slot.

If there is no unique slot ID for an arch, that arch can just reuse any
detached cpuinfos at first attach of another cpu.

Martin