virsh/qemu: The -accel and "-machine accel=" options are incompatible

[sjau]

Describe the bug

I did update my nixos unstable installation on monday, in which I also use virsh/qemu. However I noticed now that I can't start VMs aymore using virsh start VM

Steps To Reproduce

Steps to reproduce the behavior:

1. Update to latest unstable branch
2. Run a preexisting VM (not sure if new vms are affected also) using virsh start VM
3. You get an error:

error: internal error: process exited while connecting to monitor: 2022-01-12T06:56:56.144411Z qemu-system-x86_64: The -accel and "-machine accel=" options are incompatible

I looked at the xml file but I could not find but I could not find any accel or machine entries. Not sure where qemu gets this from.

Expected behavior

The VM should be started using virsh start VM

Screenshots

If applicable, add screenshots to help explain your problem.

Additional context

Add any other context about the problem here.

Notify maintainers

@eelco @qyliss

Metadata

Please run nix-shell -p nix-info --run "nix-info -m" and paste the result.

 - system: `"x86_64-linux"`
 - host os: `Linux 5.10.90, NixOS, 22.05 (Quokka)`
 - multi-user?: `yes`
 - sandbox: `yes`
 - version: `nix-env (Nix) 2.5.1`
 - channels(hyper): `"nixos-21.11pre326640.e544ee88fa4"`
 - channels(root): `"nixos-22.05pre344851.0ecf7d41481"`
 - nixpkgs: `/nix/var/nix/profiles/per-user/root/channels/nixos`

[sjau]

This seems to be related:

https://listman.redhat.com/archives/libvir-list/2021-January/msg00536.html

Reverting to 5b091d4 makes virsh/qemu run again.

[alyssais]

If I'm reading that right, a qemu-kvm script is no longer required, and can just be a symlink to the appropriate QEMU binary for the architecture. That's what Fedora's doing now, at least (verified in a container). Perhaps we should be looking at a similar change.

[peterhoeg]

libvirtd is effectively broken on unstable as a consequence of this. I'll take a look as soon as I can.