Re: [Xen-devel] Re: [Qemu-devel] [PATCH 1/7] xen: groundwork for xen support

["Daniel P. Berrange" <berrange@redhat.com>]

On Tue, Jul 29, 2008 at 09:38:01AM +0200, Gerd Hoffmann wrote:
> Samuel Thibault wrote:
> > Anthony Liguori, le Mon 28 Jul 2008 09:04:54 -0500, a écrit :
> >>> +/* xen_machine.c */
> >>> +extern QEMUMachine xenpv_machine;
> >>> +extern QEMUMachine xenfv_machine;
> >> Why does xenfv need its own machine type?
> > 
> > IIRC that's at least because it adds its own Xen platform -specific PCI card.
> 
> That is only one tiny bit of the differences between paravirtual and
> fully virtualized machines, there is plenty more.  The question is
> whenever qemu should try figure itself whenever the virtual machine it
> serves is fully virtualized or whenever we'll explicitly tell it qemu.
> 
> I think it is better to explicitly tell qemu whenever we want create a
> paravirtual or a full virtualized machine.  First, because this is how
> it is handled right now in xen (via -M [ xenpv | xenfv ] ), and second,
> it leaves the door open to have qemu also create the domain.

The setup process for 'xenpv' vs 'xenfv' is really very different. As I
mentioned in the other mail, 'xenfv' is really much closer to 'pc'
machine type, then 'xenpv'.  Thus I think its better to keep them
explicitly separate - they call out to shared code where needed anyway,
so there's no serious code duplication problem there.

Daniel
-- 
|: Red Hat, Engineering, London   -o-   http://people.redhat.com/berrange/ :|
|: http://libvirt.org  -o-  http://virt-manager.org  -o-  http://ovirt.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505  -o-  F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|