Add s390x support to QEMU backend

[LKHN]

These virtual devices added for s390x virtualization:

• virtio-scsi as alias of virtio-scsi-ccw for SCSI controller
• virtio-gpu as alias of virtio-gpu-ccw for VirtIO GPU Device
• virtio-rng as alias of virtio-rng-ccw for VirtIO RNG
• virtio-tablet as alias of virtio-tablet-ccw for input mouse interaction
• virtio-keyboard as alias of virtio-keyboard-ccw for keyboard interaction

Disabled audio support since there is no audio virtual device available on this architecture (see qemu-kvm -device help)

[github-actions]

Great PR! Please pay attention to the following items before merging:

Files matching backend/**.pm:

• Consider running manual verification tests, especially for non-qemu backends

This is an automatically generated QA checklist based on modified files.

[Martchus]

I suppose all of this is generally not very problematic because we don't use the QEMU backend for s390x test at SUSE (so it won't break existing use cases). However, tests need to be adapted (to pass again) and supposedly also extended (to cover all newly introduced code paths).

[codecov]

Codecov Report

Merging #2309 (27c60d5) into master (0688448) will decrease coverage by 0.02%.
The diff coverage is 91.66%.

❗ Current head 27c60d5 differs from pull request most recent head 5f43d43. Consider uploading reports for the commit 5f43d43 to get more accurate results

@@            Coverage Diff             @@
##           master    #2309      +/-   ##
==========================================
- Coverage   94.94%   94.93%   -0.02%     
==========================================
  Files         155      155              
  Lines       15223    15224       +1     
==========================================
- Hits        14454    14453       -1     
- Misses        769      771       +2     

Impacted Files	Coverage Δ
backend/qemu.pm	99.68% <90.47%> (-0.32%)	⬇️
OpenQA/Qemu/Proc.pm	100.00% <100.00%> (ø)

... and 2 files with indirect coverage changes

[LKHN]

@Martchus Thanks for the review.

I fixed the not passing tests and applied your suggestion. Thanks a lot!

Another issue that I tried to extend the tests with asserting QEMU parameters for s390x. However it fails due the mixed QEMU parameters from the previous tests. How can I generate the QEMU parameters which are use only variables I set for s390x?

Test - LKHN@8680ff6#diff-2c6d71750f92f3631c01e08b2695ed0a936acb082240f62d5c6162ddb5c04ce7
Error - https://github.com/LKHN/os-autoinst/actions/runs/4874753836/jobs/8696080601

[Martchus]

Likely by just setting bmwqemu::vars{ARCH} = 's390x'; before invoking the function you need to test. You can also delete keys/values from that global hash map if they lead to cases you don't want to test.

[LKHN]

It looks like, it passes now but no response yet from the codecov -https://app.codecov.io/gh/os-autoinst/os-autoinst/commit/16e5a31424afd5f4dbfa2845c7be237d21aa9ad0

I tried to extend the unit tests to cover the testing of these QEMU options:

When: ARCH => 's390x', QEMU_VIDEO_DEVICE => 'virtio-gpu', OFW => 0

• EDID is present for virtio-gpu
• -boot option not present.
• virtio-keyboard is present.
• no audio options present.
• isa-fdc.fdtypeA=none is not present.
• virtio-rng is present.
• virtio-tablet is present.

What is left, is the testing of virtio-scsi as SCSICONTROLLER Which comes from OpenQA/Qemu/Proc.pm.

@Martchus @okurz Could tell me if this approach of the implementation tests is correct? If yes how can I test the SCSICONTROLLER too?

[Martchus]

but no response yet from the codecov

The CI check shows your diff is fully covered. I think that's good enough.

[LKHN]

@Martchus and @okurz , Thanks for the review, please let me know if we need any further changes.

[Martchus]

No further changes are required. The PR hasn't been merged automatically because not all OBS checks have reported back. However, https://build.opensuse.org/package/show/devel:openQA:GitHub:os-autoinst:os-autoinst:PR-2309/os-autoinst shows that all builds were successful. So I'm merging this manually.