Demi and I figured out the last known blocker for her Cloud Hypervisor
generic vhost-user device[1] — reads and writes of the VIRTIO
configuration space weren't being forwarded to the backend — and she's
now seen it work with virtiofsd.  Another of her Cloud Hypervisor PRs
was just merged[2], so I'm hoping generic vhost-user will be quite soon
too, and then we'll be able to start easily adopting more device types
in Spectrum, starting with virtio-sound.  She also submitted a few
small Spectrum improvements[3][4][5][6].

[1]: https://github.com/cloud-hypervisor/cloud-hypervisor/pull/7221
[2]: https://github.com/cloud-hypervisor/cloud-hypervisor/pull/7223
[3]: https://spectrum-os.org/lists/archives/spectrum-devel/20250807-no-require-kvm-v1-1-5dec3fbaef7b@gmail.com/
[4]: https://spectrum-os.org/lists/archives/spectrum-devel/20250807-no-erofs-xattrs-v1-1-957d518c0861@gmail.com/
[5]: https://spectrum-os.org/lists/archives/spectrum-devel/20250808-envrc-v1-1-450c87b192f9@gmail.com/
[6]: https://spectrum-os.org/lists/archives/spectrum-devel/20250807-host-dev-fd-v1-1-3d08c0f5eab3@gmail.com/

Yureka has continued working on her program that will forward traffic on
network devices out of driver VMs to the host userspace router, and the
code is now public[7].  It now handles multiple physical interfaces per
driver VM, which will be important if we want to use both of them and
they're in the same IOMMU group.

[7]: https://cyberchaos.dev/yuka/xdp-forwarder

As for me, it was mostly the usual Nixpkgs maintenance — start of a new
staging cycle — and preparation for the new grant.  I also fixed a few
Spectrum bugs — a couple of compatibility fixes for applications, and a
harmless error during the build.  I've been at WHY2025 this week too,
and having some good conversations about future opportunities for
Spectrum.  I'll be here until Wednesday morning — if you are too and
would like to talk about Spectrum, or get a sticker, DECT QYLS. :)