[libvirt] [PATCH] PCI passthrough docs, tests and minor change

[Daniel P. Berrange]

On Mon, Jan 12, 2009 at 12:55:14PM +0100, Daniel Veillard wrote:
> On Mon, Jan 12, 2009 at 11:46:38AM +0000, Daniel P. Berrange wrote:
> > >   I also noticed that we parse (and save) and extra PCI domain
> > > argument, but it's not actually used when calling qemu, so I
> > > assume it's a missing feature for QEmu and just decided to patch
> > > the code to not save the extra value when not defined (i.e. 0).
> > NACK to this chunk - we should always output the 'domain' attribute
> > even when it is zero - it should only be optional when parsing the
> > XML. The fact that QEMU doesn't use it is just an impl artifact of
> > QEMU.
> 
>   Then it need to be documented too, no way around since it will
> always show up in dumps. And the fact it's not actually used must be
> documented too. IMHO this just increase the long maintainance for
> something you don't have any garantee to ever use.

IMHO it is a bug that QEMU just hardcodes domain ID as 0000 because this
is assuming a machine only ever has one PCI domain, which is just a bogus
assumption.

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 :|