The current proposal for virtio support in Xen currently comes with a standalone DEMU based IOREQ server for virtio-block devices. As we want to develop a multitude of virtio based devices we want to implement a new Xen aware machine model for QEMU which can be wired up to accept the IOREQ requests for any virtio-mmio devices that can be currently handled by QEMU.

As the way Xen and QEMU interact differs from the other hypervisors we also need to come up with some sort of way for Xen to define what the virtio-mmio devices are and pass the relevant information to QEMU. There is already IOREQ functionality in QEMU for supporting HVM machines so this should mostly be re-plumbing support in the QEMU build.

QEMU should also be able to act as the stub setup to pass IOREQ events to a separate vhost-user daemon as a stepping stone to a future Xen aware vhost-user stub program.

Deliverables

• QEMU -m xenvirt model

• Expanding Xen tooling to specify additional virtio-mmio buses and
  instantiate on QEMU

• Allow QEMU to pass through IOREQs to vhost-user daemons

Out of Scope (Optional)

• A standalone Xen/vhost-user stub is for future cards

Risks and Assumptions (Optional)

• Risk - Mitigation

• Assumptions

Acceptance Criteria Example and Notes (Optional)

Legend: