From mboxrd@z Thu Jan  1 00:00:00 1970
From: Thomas Gleixner <tglx@linutronix.de>
Subject: Re: [PATCH 2/2] Device assignment: Fix MSI IRQ affinity setting
Date: Fri, 25 May 2012 01:19:22 +0200 (CEST)
Message-ID: <alpine.LFD.2.02.1205250115560.3231@ionos>
References: <1337878924-39069-1-git-send-email-richard@nod.at> <1337878924-39069-2-git-send-email-richard@nod.at> <1337883627.4714.32.camel@ul30vt> <alpine.LFD.2.02.1205242323230.3231@ionos> <4FBEADCB.2010900@siemens.com> <20120524221757.GF31110@redhat.com>
 <alpine.LFD.2.02.1205250106350.3231@ionos>
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Cc: Jan Kiszka <jan.kiszka@siemens.com>,
	Alex Williamson <alex.williamson@redhat.com>,
	Richard Weinberger <richard@nod.at>, kvm@vger.kernel.org,
	avi@redhat.com, Marcelo Tosatti <mtosatti@redhat.com>,
	Bjorn Helgaas <bhelgaas@google.com>
To: "Michael S. Tsirkin" <mst@redhat.com>
Return-path: <kvm-owner@vger.kernel.org>
Received: from www.linutronix.de ([62.245.132.108]:59220 "EHLO
	Galois.linutronix.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1753395Ab2EXXTZ (ORCPT <rfc822;kvm@vger.kernel.org>);
	Thu, 24 May 2012 19:19:25 -0400
In-Reply-To: <alpine.LFD.2.02.1205250106350.3231@ionos>
Sender: kvm-owner@vger.kernel.org
List-ID: <kvm.vger.kernel.org>

On Fri, 25 May 2012, Thomas Gleixner wrote:

> On Fri, 25 May 2012, Michael S. Tsirkin wrote:
> > On Thu, May 24, 2012 at 06:53:15PM -0300, Jan Kiszka wrote:
> > > On 2012-05-24 18:39, Thomas Gleixner wrote:
> > > > On Thu, 24 May 2012, Alex Williamson wrote:
> > > >> On Thu, 2012-05-24 at 18:02 +0100, Richard Weinberger wrote:
> > > >>> +            if (address == msi_start + PCI_MSI_DATA_32)
> > > >>> +                handle_cfg_write_msi(pci_dev, assigned_dev);
> > > >>
> > > >> Why didn't we just use range_covers_byte(address, len, pci_dev->msi_cap
> > > >> + PCI_MSI_DATA_32) to start with?  But how does this handle the enable
> > > >> bit?
> > > > 
> > > > The problem with the current implementation is that it only changes
> > > > the routing if the msi entry goes from masked to unmasked state.
> > > > 
> > > > Linux does not mask the entries on affinity changes and never did,
> > > > neither for MSI nor for MSI-X.
> > > > 
> > > > I know it's probably not according to the spec, but we can't fix that
> > > > retroactively.
> > > 
> > > For MSI, this is allowed. For MSI-X, this would clearly be a Linux bug,
> > > waiting for hardware to dislike this spec violation.
> > > 
> > > However, if this is the current behavior of such a prominent guest, I
> > > guess we have to stop optimizing the QEMU MSI-X code that it only
> > > updates routings on mask changes. Possibly other OSes get this wrong too...
> > 
> > Very strange, a clear spec violation. I'll have to dig in the source to
> > verify this.
> 
> Stop digging. MSI-X is correct.

This was based off an older version of qemu-kvm, where the routing for
MSI-X was broken for other reasons. But that seems to be fixed now.

I use the age excuse :)

Thanks,

	tglx