WARNING - OLD ARCHIVES

This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
   
 
 
Xen 
 
Home Products Support Community News
 
   
 

xen-ppc-devel

Re: [XenPPC] [pushed][ppc] process all interrupts that may be in the PIC

To: Hollis Blanchard <hollisb@xxxxxxxxxx>
Subject: Re: [XenPPC] [pushed][ppc] process all interrupts that may be in the PIC
From: Jimi Xenidis <jimix@xxxxxxxxxxxxxx>
Date: Wed, 29 Mar 2006 13:15:44 -0500
Cc: xen-ppc-devel@xxxxxxxxxxxxxxxxxxx
Delivery-date: Wed, 29 Mar 2006 18:16:40 +0000
Envelope-to: www-data@xxxxxxxxxxxxxxxxxxx
In-reply-to: <200603291158.19810.hollisb@xxxxxxxxxx>
List-help: <mailto:xen-ppc-devel-request@lists.xensource.com?subject=help>
List-id: Xen PPC development <xen-ppc-devel.lists.xensource.com>
List-post: <mailto:xen-ppc-devel@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-ppc-devel>, <mailto:xen-ppc-devel-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-ppc-devel>, <mailto:xen-ppc-devel-request@lists.xensource.com?subject=unsubscribe>
References: <E1FOeZF-0008Hm-6K@xxxxxxxxxxxxxxxxxxxxx> <200603291158.19810.hollisb@xxxxxxxxxx>
Sender: xen-ppc-devel-bounces@xxxxxxxxxxxxxxxxxxx

On Mar 29, 2006, at 12:58 PM, Hollis Blanchard wrote:

On Wednesday 29 March 2006 11:34, jimix@xxxxxxxxxxxxxx wrote:
changeset:   9665:adf9242cac67
user:        jimix@xxxxxxxxxxxxxxxxxxxxx
date:        Wed Mar 29 12:31:15 2006 -0500
summary:     [ppc] process all interrupts that may be in the PIC

Linux used to loop until there were no more interrupts pending, but it moved away from that behavior. Anton explained there were two reasons for that: 1) statistically speaking, there is almost never more than one interrupt
pendings, and
2) the cost of an MMIO to find out if there is another interrupt exceeds the
cost of taking another EE exception

Things a quite a bit different with Xen:
  1. All IRQs have the same priority.
2. IRQs routed to domains are EOI'd by the domain, thats a looong time, longer then an MMIO 3. We are trying to fill as many event channel bits as possible with this logic.

Accordingly, I don't like this patch.

Tho, I sort of agree with the Anton's strategy for Linux, I'm pretty sure the loop is worth it for Xen.
When we actually get more running we can revisit and profile it.
-JX


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

<Prev in Thread] Current Thread [Next in Thread>