Bug 2375004
| Summary: | qemu tests fail with Python 3.14 | ||
|---|---|---|---|
| Product: | [Fedora] Fedora | Reporter: | Yaakov Selkowitz <yselkowi> |
| Component: | qemu | Assignee: | Richard W.M. Jones <rjones> |
| Status: | CLOSED RAWHIDE | QA Contact: | Fedora Extras Quality Assurance <extras-qa> |
| Severity: | medium | Docs Contact: | |
| Priority: | unspecified | ||
| Version: | rawhide | CC: | berrange, cfergeau, crobinso, ksurma, mcascell, mhroncok, pbonzini, philmd, rjones, suraj.ghimire7, virt-maint |
| Target Milestone: | --- | ||
| Target Release: | --- | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | --- | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2025-07-15 18:30:34 UTC | Type: | --- |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
| Bug Depends On: | |||
| Bug Blocks: | 2322407, 2339432 | ||
|
Description
Yaakov Selkowitz
2025-06-26 17:14:04 UTC
https://docs.python.org/3.14/whatsnew/3.14.html#id7 > Removed implicit creation of event loop by asyncio.get_event_loop(). It now raises a RuntimeError if there is no current event loop. https://github.com/python/cpython/issues/126353 This is now blocking a stoname bump in libcbor. https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/thread/5CO7XNDVNYEQPREU4U2GCYNPII2WK2YT/ Please, respond. I don't see any upstream fix yet. Can you suggest a fix? I'm neither an expert in Python asyncio nor in QMP. The code in python/qemu/qmp/legacy.py doesn't exactly match any of the suggestions in the Python docs. I'd try replacing:
self._aloop = asyncio.get_event_loop()
With:
try:
self._aloop = asyncio.get_event_loop()
except RuntimeError:
self._aloop = asyncio.new_event_loop()
This should effectively revert to the behavior of Python < 3.14, as get_event_loop used to return a new event loop when there was no current event loop.
What puzzles me is why we've never seen any deprecation warnings with python < 3.13, because if we were triggering the fallback 'create a new loop' code previously, we should have seen warnings. This makes me wonder if we've got further problems than just the headline event loop issue. OK I reproduced this locally and implemented the fix as suggested by Miro above. > What puzzles me is why we've never seen any deprecation warnings with python < 3.13
I am not sure the warning would propagate through the tests.
Fedora build: https://koji.fedoraproject.org/koji/taskinfo?taskID=134838089 Upstream patch: https://www.mail-archive.com/qemu-devel@nongnu.org/msg1127519.html (In reply to Daniel Berrangé from comment #5) > What puzzles me is why we've never seen any deprecation warnings with python > < 3.13, because if we were triggering the fallback 'create a new loop' code > previously, we should have seen warnings. This makes me wonder if we've got > further problems than just the headline event loop issue. Urgh, python only emits warnings if triggered *directly* by the __main__ module, so any usage from imports never shows warnings. This is spectacularly unhelpful at alerting people to usage of deprecated features :-( New Fedora build: https://koji.fedoraproject.org/koji/taskinfo?taskID=134838848 This failed to build on i686 with what might be a random failure. I will retry the build when all the other arches have finished. 54/310 qemu:qtest+qtest-arm / qtest-arm/aspeed_smc-test ERROR 43.15s killed by signal 6 SIGABRT It also failed on ppc64 with a test timeout. Here is a new build (unmodified): https://koji.fedoraproject.org/koji/taskinfo?taskID=134842457 The ppc64 test failure was on 'test-replication' which normally takes only 2 seconds, so a 60 second timeout is a clear hang. I notice upstream there is report that this test causes 'tsan' thread safety warning, so further looks like a bug. I've filed this to track it: https://gitlab.com/qemu-project/qemu/-/issues/3035 I cherry picked another patch from upstream to improve the stability of a test that failed on x86_64 (thanks Dan!) New build: https://koji.fedoraproject.org/koji/taskinfo?taskID=134842989 I added a further patch to increase the test-replication timeout, for: https://gitlab.com/qemu-project/qemu/-/issues/3035 New build: https://koji.fedoraproject.org/koji/taskinfo?taskID=134845860 Finally fixed in Rawhide - qemu-10.0.2-5.fc43 Thank you. For c10s (for merging back to ELN): https://gitlab.com/redhat/centos-stream/rpms/qemu-kvm/-/merge_requests/237 |