From mboxrd@z Thu Jan 1 00:00:00 1970 From: Keir Fraser Subject: Re: irq_guest_eoi_timer interaction with MSI Date: Tue, 25 Nov 2008 10:47:31 +0000 Message-ID: References: <492BDD85.76E4.0078.0@novell.com> Mime-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <492BDD85.76E4.0078.0@novell.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Jan Beulich Cc: xen-devel@lists.xensource.com List-Id: xen-devel@lists.xenproject.org On 25/11/08 10:12, "Jan Beulich" wrote: >>>> Keir Fraser 24.11.08 18:02 >>> >> Setting the need-a-hypercall bit looks racey. Don't you need to set the bit, >> then check the guest didn't unmask meanwhile? > > What would the action be in that case? Try send_guest_pirq() a second > time, and not arm the timer (along with clearing all the bits again) if one > of the guests is now able to accept it? Would seem too complicated to me > (the more that then the HVM case would need to be taken care of > explicitly), given that all we try to avoid is a 1ms gap in delivering the > next event. But it would be doable of course. Yes, I think it would be something like that. As you say, the patch you propose is a strict improvement on the current state of affairs, so we can just go with that and extend it only if it proves necessary. -- Keir