[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH 0/3] VMX: Properly handle pi descriptor and per-cpu blocking list

On Tue, 2016-05-24 at 13:33 +0000, Wu, Feng wrote:
> > From: Wu, Feng
> > > From: Dario Faggioli [mailto:dario.faggioli@xxxxxxxxxx]
> > > 
> > > If a
> > > vCPU is blocker, there is nothing to do, and in fact, nothing
> > > happens
> > > (as vcpu_sleep_nosync() and vcpu_wake() are NOP in that case).
> > What do you mean by saying ' vcpu_sleep_nosync() and vcpu_wake()
> > are NOP '?
> I think I understand what you meant above now.
Right. In any case, see the email I've just sent, with a detailed
breakdown of the situation and of what actually happens.

> Do you think the following idea makes sense?
> When a pCPU is unplugged, we can just remove the vcpus on the
> associated per-cpu blocking list, then we can choose another online
> cpu, set the right NDST value, and put the vCPU the new per-cpu list?
Well, this does make sense to me, but the point is how you do that. I
mean, how do you get to execute some PI adjustment code, from cpu-
teardown code?

Right now, for what seemed to be necessary until now, we have the
arch_vcpu_block(). Do we need more? If yes where?

From looking only at schedule.c, we already have arch_move_irqs(), can
we take advantage of it?

<<This happens because I choose it to happen!>> (Raistlin Majere)
Dario Faggioli, Ph.D, http://about.me/dario.faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)

Attachment: signature.asc
Description: This is a digitally signed message part

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.