This is holding up submitting to Factory. In ad-hock tests, using my own
personal Windows10 guest on my laptop running Tumbleweed, I see hangs using the
v2.12 qemu sitting in the OBS Virtualization project. I've tried debugging it
but it has been elusive so far.

A similar report has been made here:
https://bugs.launchpad.net/qemu/+bug/1769189

As I report there, increasing the vcpu count does seem to increase the
likelihood that the bug is encountered, and it seems the guest is expecting an
interrupt from the virtual ahci device, since if I manually inject one, the
guest un-hangs.