Trying to run f44 or rawhide s390x cloud image on libvirt/qemu s390x, hanging network-less

[Frank Ch. Eigler]

Hi -

I'm hoping to operate a slow but useful s390x emulator for our team's
testing needs.  Our infrastructure for this is libvirt/qemu based, and
it has worked okay with a Fedora 44 s390x image off of the
fedora-secondary distribution sites.

I could never get a Rawhide image running kernel 7.X working: it would
always become unresponsive shortly after the qemu-guest-agent service
loads, and never comes up on the simulated network.  Then, two days
ago, the same kernel was backported into F44, and now I can't get the
updated-F44 VM disk image past that point either.

Is this sort of thing a common / known / interesting problem?

- FChE

[Dan Horák]

Hi Frank,

On Sun, 5 Jul 2026 13:51:15 -0400
"Frank Ch. Eigler via s390x"  wrote:
> Hi -
>
> I'm hoping to operate a slow but useful s390x emulator for our team's
> testing needs.  Our infrastructure for this is libvirt/qemu based, and
> it has worked okay with a Fedora 44 s390x image off of the
> fedora-secondary distribution sites.
>
> I could never get a Rawhide image running kernel 7.X working: it would
> always become unresponsive shortly after the qemu-guest-agent service
> loads, and never comes up on the simulated network.  Then, two days
> ago, the same kernel was backported into F44, and now I can't get the
> updated-F44 VM disk image past that point either.
>
> Is this sort of thing a common / known / interesting problem?

I don't think it's a know problem, but IMO it's interesting enough to
be reported (likely) to qemu upstream. Are you saying that with 7.0 it's
OK, but with 7.1 it misbehaves?

Dan

[Frank Ch. Eigler]

Hi -
> [...]
> > Is this sort of thing a common / known / interesting problem?
>
> I don't think it's a know problem, but IMO it's interesting enough to
> be reported (likely) to qemu upstream. Are you saying that with 7.0 it's
> OK, but with 7.1 it misbehaves?

With the F44 cloud image (running a 6.12ish kernel), it worked, but as soon
as that image internally self-updated to a 7.2-ish kernel, or whether a
brand new Rawhide cloud image (running same 7.2-ish kernel) attempts to boot,
there's a hang a few minutes into the sequence.  (There are a couple of
TIME / timed-out-waiting-for-devince errors about a minute before the hang
itself.)

- FChE