Re: [PATCH] Add a new PIIX option to control PCI hot unplugging of devices on non-root buses

[Igor Mammedov <imammedo@redhat.com>]

On Fri, 24 Apr 2020 14:44:48 -0400
Eduardo Habkost <ehabkost@redhat.com> wrote:

> On Fri, Apr 24, 2020 at 03:23:56PM +0000, Ani Sinha wrote:
> > 
> >   
> > > On Apr 22, 2020, at 4:15 PM, Ani Sinha <ani.sinha@nutanix.com> wrote:
> > > 
> > > 
> > >   
> > >> On Apr 21, 2020, at 8:32 PM, Daniel P. Berrangé <berrange@redhat.com> wrote:
> > >> 
> > >> On Tue, Apr 21, 2020 at 02:45:04PM +0000, Ani Sinha wrote:  
> > >>> 
> > >>>   
> > >>>> On Apr 20, 2020, at 8:32 PM, Michael S. Tsirkin <mst@redhat.com> wrote:
> > >>>> 
> > >>>> But I for one would like to focus on keeping PIIX stable
> > >>>> and focus development on q35.  Not bloating PIIX with lots of new
> > >>>> features is IMHO a good way to do that.  
> > >>> 
> > >>> Does this mean this patch is a no-go then? :(  
> > >> 
> > >> I'd support this patch, as I don't think it can really be described as
> > >> bloat or destabalizing. It is just adding a simple property to
> > >> conditionalize existing functionality.  Telling people to switch to Q35
> > >> is unreasonable as it is not a simple 1-1 conversion from existing use
> > >> of PIIX. Q35 has much higher complexity in its configuration, has higher
> > >> memory overhead per VM too, and lacks certain features of PIIX too.  
> > > 
> > > Cool. How do we go forward from here?
> > >   
> > 
> > We would really appreciate if we can add this extra knob in
> > Qemu. Maybe someone else also in the community will find this
> > useful. We don’t want to maintain this patch internally forever
> > but rather prefer we maintain this as a Qemu community.  
> 
> Michael, I agree with Daniel here and I don't think we should
> start refusing PIIX features if they are useful for a portion of
> the QEMU community.
> 
> Would you reconsider and merge this patch?

I put this patch on my review queue (hopefully next week I'd be able to get to it)