Hi,
This is one of several emails to follow up on Linaro's internal KWG sprint last week in Cambridge where a number of Project Stratos hackers discussed what next steps we have and started to think about future work. I am splitting the update into several emails so I can freely CC the relevant lists for each without too much cross-posting spam.
Intro =====
We've made good progress over the last year and have up-streamed a number of device models as vhost-user daemons. We have also gotten our first proof of concept build of the xen-vhost-master which has allowed us to reuse these backends on the Xen hypervisor.
https://github.com/vireshk/xen-vhost-master
Remaining Work ==============
Scope out the remainder of APIs needed for oxerun -------------------------------------------------
The current xen-vhost-master uses a combination of the native rust oxerun and a bindgen import of libxensys (https://github.com/vireshk/libxen-sys) and a number of xen libraries built directly in the xen-vhost-master repository.
Our intention for the Stratos work is to remove any C dependency for the rust backend and use native rust bindings to talk to the hypervisor control ioctl.
Identifying what is needed should be easy enough as we can see where in master repository C calls are being made. This work should be broken down into groups in JIRA so the work can be efficiently divided up.
Currently our focus for the rust-vmm repo is to support the vhost-user daemons but a wider conversation needs to be had with the community about the rest of the tooling involved in the creation and control of DomU guests. For Stratos we would like to explore the possibilities of bare metal monitor programs for dom0-less (or dom0-light?) setups.
Strategy for testing oxerun in the rust-vmm project ---------------------------------------------------
Currently the rust-vmm projects rely heavily on unit tests and a (mostly) x86 build farm. While building for non-x86 architectures isn't insurmountable doing blackbox testing on real hypervisors isn't currently supported. Given the low level nature of the interactions simply mocking the ioctl interface to the kernel will not likely sufficiently exercise things.
We need a way to execute tests on a real system with a real Xen hypervisor and dom0 setup. We can either:
- somehow add Xen hosts to the Buildkite runner pool for rust-vmm or - investigate using QEMU TCG as a portable system in a box to run Xen and guests
Currently this is blocking wider up-streaming of the oxerun code to https://github.com/rust-vmm/xen-sys in the same way other rust-vmm repos work.
See also ========
Other subjects discussed will be the subject of other emails today with different distribution lists. These are:
- Remaining work for vhost-device - Additional virtio devices - Integrating rust-vmm with QEMU
Happy reading ;-)