Re: [libvirt] [PATCH 2/6] Use -machine option for PPC64
On Thu, Dec 20, 2012 at 6:28 PM, Daniel P. Berrange <berrange(a)redhat.com>wrote:
On Thu, Dec 20, 2012 at 02:02:23PM +0800, Li Zhang wrote:
> From: Li Zhang <zhlcindy(a)linux.vnet.ibm.com>
>
> On PPC64 platform, it sets USB option in -machine option. It is
> set as the following:
> -machine pseries,usb=off
>
> This will disable USB, which avoids creating one USB controller
> implicitly in the platform's initialization. Libvirt will add one
> USB controller if there is no definition in XML file. Users also can
> add one specific USB controller in the XML file as the following:
>
> <controller type='usb' index='0'
model='pci-ohci'>
> </controller>
NACK, current behaviour matches libvirt on x86, where we also
add a USB controller by default.
Yes, I saw the code in Libvirt which adds a USB controller by default.
My problem is that:
In QEMU for pseries, default VGA is enabled, one USB controller is created
and USB keyboard and mouse are created.
If libvirt also creates USB controller with -device xxxx, it will produce
the error:
qemu-system-ppc64: -device pci-ohci,id=usb,bus=pci,addr=0x1.0x2: PCI:
single function device can't be populated in function 1.2
qemu-system-ppc64: -device pci-ohci,id=usb,bus=pci,addr=0x1.0x2: Device
'pci-ohci' could not be initialized
2012-12-20 15:16:54.985+0000: shutting down
To avoid the conflict, we need to use -machine name,usb=off to disable USB
option.
My patch doesn't create the default USB controller, it only help disable
USB option.
Thanks. :)
The mechanism for disabling the default USB controlle ris
to use
<controller type='usb' index='0' model='none'>
PPC should follow this same behaviour
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:|
--
Best Regards
-Li
Attachments:
- attachment.html (text/html — 3.5 KB)