Re: console class in kvm

["Michael S. Tsirkin" <mst@redhat.com>]

On Mon, Mar 26, 2012 at 02:29:58PM +0200, Avi Kivity wrote:
> On 03/26/2012 01:19 PM, Michael S. Tsirkin wrote:
> > On Mon, Mar 26, 2012 at 12:18:54PM +0200, Avi Kivity wrote:
> > > On 03/26/2012 11:48 AM, Michael S. Tsirkin wrote:
> > > > kvm used to carry this commit:
> > > 
> > > Used to?  Which commit reverts this?
> >
> > A merge from qemu.git I would guess.
> > git log does not seem to show the culprit, I don't know
> > how to find it.
> 
> It was e2478f504fff20ad (git log  -SPCI_CLASS_OTHERS master -m -p).
> 
> 
> > > > commit 4667e6ec0df770867095d8093562d93c94d96ca2
> > > > Author: Avi Kivity <avi@redhat.com>
> > > > Date:   Thu Feb 12 11:43:17 2009 +0200
> > > >
> > > >     Change virtio-console to PCI_CLASS_OTHERS
> > > >     
> > > >     As a PCI_CLASS_DISPLAY_OTHER, it reduces primary display somehow on Windows XP
> > > >     (possibly Windows disables acceleration since it fails to find a driver).
> > > >     
> > > >     Signed-off-by: Avi Kivity <avi@redhat.com>
> > > >
> > > > This seems to have been dropped. Is the issue gone?
> > > > Could relevant parties speak up please?
> > > > Do we want to merge this commit into qemu.git?
> > > 
> > > It's an impossible compatibility problem now.  I have this sinking
> > > feeling that we need to create yet another driver property.
> >
> > That's easy, we have a class property for this already.
> 
> Yes, d6beee9938.

Exactly. qemu-kvm used to set the class to CLASS_OTHER while
the current code sets it to PCI_CLASS_COMMUNICATION_OTHER.
Do we want support for CLASS_OTHER or is it ok to drop it?

> 
> -- 
> error compiling committee.c: too many arguments to function