RE: [Xen-devel] Re: [PATCH] Handle MSI irq storm

["Shan, Haitao" <haitao.shan@xxxxxxxxx>]

Oh, I just saw this second mail from you.
Yes. That's the same irq storm as before which was left unhandled at that time.
I did not see this IRQ storm to happen when this NIC is assigned to PV guests. 
So I guess maybe it is because PV guests tend to be more responsive than HVM 
guests so that there is less time window for IRQ storm?
But actually the difficulty with PV guests is I have not thought of a good 
point to identify the end of interrupt handling, as described in my last mail. 
Do you have any idea on this? Thanks.

Shan Haitao

-----Original Message-----
From: Keir Fraser [mailto:keir.fraser@xxxxxxxxxxxxx] 
Sent: 2008年7月3日 16:18
To: Shan, Haitao
Cc: xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: Re: [Xen-devel] Re: [PATCH] Handle MSI irq storm

On 3/7/08 09:10, "Keir Fraser" <keir.fraser@xxxxxxxxxxxxx> wrote:

> I thought that there already was storm-avoidance logic included, because of
> the other storm scenario you used to be able to reproduce?

Oh, actually, do you mean that this is still the same storm you could repro
before? And we do not have avoidance logic yet because you stopped being
able to repro the issue for a while?

Okay if so, but still I think there are some issues to deal with in the
patch. The avoidance logic for MSI delivery to a PV guest looks suspicious,
as I said in my previous email -- can the patch actually avoid storm
delivery to a PV guest?

 -- Keir



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel