From mboxrd@z Thu Jan 1 00:00:00 1970 From: Keir Fraser Subject: Re: 4.0/4.1 requests Date: Thu, 08 Sep 2011 12:11:13 +0100 Message-ID: References: <4E689D9A.1020405@citrix.com> Mime-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <4E689D9A.1020405@citrix.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: Andrew Cooper Cc: "xen-devel@lists.xensource.com" , Jan Beulich List-Id: xen-devel@lists.xenproject.org On 08/09/2011 11:48, "Andrew Cooper" wrote: >>> Hi Keir, >>> >>> without the old IO-APIC part addressed, I wonder whether we should >>> really have the backport of 23805:7048810180de ("IRQ: manually EOI >>> migrating line interrupts") in both trees. >> It does fix a real bug. Perhaps Andrew could hack up a patch to make it >> dependent on IO-APIC version? > Now that I am not racing against a release deadline, this is a > possibility. It probably means falling through to the "fake an EOI for > line level interrupt code", as the Status bit is RO in the IO-APIC (The > IRR bit is RW but both need updating) Falling back to the old behaviour would be acceptable. I think we're talking only very old systems here. > Does anyone know which revision of the IO-APIC was the first with an EOI > register? Jan's patches which use the IOAPIC EOI register have a version check. You can copy that. -- Keir > If not, I think I have some document trawling to do.