Hi,
We are actually trying to use virtio devices with Virt-mmio bus.
The idea is to use theses patches I mentioned in a previous mail
(http://lists.gnu.org/archive/html/qemu-devel/2012-09/msg04913.html).
We are currently drawing the plan of how we can complete theses patches
to have it upstreamed.
Actually we have two major topics.
Backends / transport separation :
We need a good separation between backends and virtio-pci/virtio-mmio,
actually it seems to be the case for virt-mmio but not for virtio-pci as
the old device file is temporary kept for compability ( virtio-pci-new.c ).
Quality of code :
We need to be sure that all the devices are happy with the QOM design,
any hints about that ?
For the moment your plan is :
Implementing virtio-xxx-pci which create virtio-pci and virtio-xxx ( as
suggested by Evgeny Voevodin ).
Looking at the command line issues.
Refactor remaining back-ends ( virtio-scsi, virto-serial, ... ) and
update s390.
Check the QOM design of all the modification ( virtio-mmio, transport,
... ).
Is that making sense ? Any suggestion ?
I'll reporting back by the end of the week.
Fred
