[libvirt] [PATCH]Support virtio-scsi disk XML with tag 'model'

[Chen HanXiao]

> -----Original Message-----
> From: Daniel P. Berrange [mailto:berrange at redhat.com]
> Sent: Monday, November 12, 2012 6:31 PM
> To: Chen Hanxiao
> Cc: libvir-list at redhat.com
> Subject: Re: [libvirt] [PATCH]Support virtio-scsi disk XML with tag 'model'
> 
> On Mon, Nov 12, 2012 at 05:18:41PM +0800, Chen Hanxiao wrote:
> > From: ChenHanxiao <chenhanxiao at cn.fujitsu.com>
> >
> > If we add a virtio-disk, we need to add a SCSI controller with
> > model 'virtio-scsi'.
> > This patch allows libvirt to analyze disks XML with tag 'model':
> >
> > <disk type='block' device='disk'>
> >     <driver name='qemu' type='raw'/>
> >     <source dev='/dev/sdb'/>
> >     <target dev='sda' bus='scsi' model = 'virtio-scsi'/>
> > </disk>
> >
> > If we got a disks XML with bus='scsi' model = 'virtio-scsi',
> > we'll add a SCSI controller with model 'virtio-scsi' automatically.
> >
> > Signed-off-by: ChenHanxiao <chenhanxiao at cn.fujitsu.com>
> 
> NACK duplicating the bus information on the <disk> is just
> not very pleasant. Applications should really be adding
> <controllers> explicitly - the auto-add of controllers is
> just to provide back-compat for applications that were
> written to add IDE disks before the <controller> concept
> was introduced. New apps should not rely on this.

Adding SATA, SCSI, IDE disk for instance,
Did this means that functions in libvirt such as:
virDomainDefAddImplicitControllers
virDomainDefAddDiskControllersForType
is not helper function for adding controller of specific devices
but to make the old code compatible with newly introduced
controller concept?  

Regards
> 
> Daniel
> --
> |: http://berrange.com      -o-
> http://www.flickr.com/photos/dberrange/ :|
> |: http://libvirt.org              -o-
> http://virt-manager.org :|
> |: http://autobuild.org       -o-
> http://search.cpan.org/~danberr/ :|
> |: http://entangle-photo.org       -o-
> http://live.gnome.org/gtk-vnc :|