[libvirt] Libvirt error in Openstack Tempest attaching disk on arm64 system

Clark Laughlin clark.laughlin at linaro.org
Tue Aug 4 20:48:16 UTC 2015


I was just recently told I should try using virtio-scsi on arm64 -- that
QEMU 2.4.0 has everything needed for block device hotplug on arm64 using
virtio-scsi.  I see that nova appears to have support as of Feb 2014 (
https://review.openstack.org/#/c/70263/).  I am currently using the
distro-provided libvirt on trusty and vivid.  Would this be new enough to
have what I need?

Thank you,
Clark

On Wed, Jul 22, 2015 at 3:58 PM, Daniel P. Berrange <berrange at redhat.com>
wrote:

> On Wed, Jul 22, 2015 at 02:53:31PM -0500, Clark Laughlin wrote:
> > I am running Openstack Tempest on an arm64 platform and am seeing some
> > test failures related to attaching volumes to an instance.  This is an
> > example of the disk XML generated by one of the tests:
> >
> > <disk type="block" device="disk"><driver name="qemu" type="raw"
> > cache="none"/<source
> >
> dev="/dev/disk/by-path/ip-10.7.1.2:3260-iscsi-iqn.2010-10.org.openstack:volume-5a204339-80cb-4d06-aecf-2a8a2c970b0e-lun-1"/><target
> > bus="virtio"
> dev="vdb"/><serial>5a204339-80cb-4d06-aecf-2a8a2c970b0e</serial></disk>
> >
> > The test is failing with the error "XML error: No PCI buses
> > available".  I am trying to find the relevent source locations for
> > this functionality in either in the nova libvirt driver or in the
> > libvirt source itself.  I am not sure why I am getting an error about
> > no PCI buses when the bus specified in the XML is "virtio".
> >
> > I would appreciate any pointers / help.
>
> virtio is just a guest/host device communication protocol that
> can be run over multiple different transport. On x86 virtio has
> always used PCI, but on s390 it uses either s390 or ccw bus,
> and on arm7/aarch64 it uses mmio. I'm not sure that the mmio
> bus supports hotplug, which could be why you see the error in
> question. Very latest upstream QEMU does now support PCI with
> aarch64 so in the near future it should have parity of functionality
> with x86
>
> 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
> :|
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/libvir-list/attachments/20150804/da0befc4/attachment-0001.htm>


More information about the libvir-list mailing list