Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: 8BIT
Subject: RE: [RFT][PATCH] i386 per cpu IDT (2.6.12-rc1-mm1)
Date: Mon, 26 Sep 2005 11:23:31 -0700
Message-ID: <7F740D512C7C1046AB53446D37200173055345A1@scsmsx402.amr.corp.intel.com>
Thread-Topic: [RFT][PATCH] i386 per cpu IDT (2.6.12-rc1-mm1)
Thread-Index: AcXB+ksihk3CZlInRyWMhqUbHJLPPwAyA8EQ
From: "Nakajima, Jun" <jun.nakajima@intel.com>
To: "Zwane Mwaikambo" <zwane@arm.linux.org.uk>,
       "Protasevich, Natalie" <Natalie.Protasevich@UNISYS.com>
Cc: "Linux Kernel" <linux-kernel@vger.kernel.org>, "Andi Kleen" <ak@suse.de>,
       "Raj, Ashok" <ashok.raj@intel.com>, <bjorn.helgaas@hp.com>
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1209
Lines: 40

Zwane Mwaikambo wrote:

Zwane, Hi

> Apologies for the long periods between updates, i've been doing some
> relocating.
> 
> Changes since last post:
> 
> * Current interrupt handling domain is still on a node basis,
> although i 
> have moved over to dynamically allocated per cpu IDTs.

I think it might be better if you define some cpu group where the cpus
share the same IDT. Then you can handle big SMP machines as well; it's a
kind of software partitioning limited to I/O device interrupts. That
will be helpful for virtulization like Xen.

> 
> * MSI now allocates vectors per node too, i've introduced a policy
> whereupon the node its allocated on depends on where the code is
> running. 
> I'd like to move towards a policy where we allocate the vector on the
> node 
> the bus/device belongs to, objections? This code is totally untested
> as i 
> don't have any MSI capable devices.
> 
> Thanks,
> 	Zwane
> 

Jun
---
Intel Open Source Technology Center
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/