Attention is currently required from: arehbein, pespin.
1 comment:
Patchset:
This doesn't prove that the problem is local or related to my patch. […]
First of all, I think we can agree that not knowing the cause for these timestamps is unsatisfactory. I tried to reproduce the issue, but after fixing issues with the leak sanitizer I wasn't able to see wrong timestamps.
You can see my commits in daniel/stream_tests_lsan
Secondly, in https://gerrit.osmocom.org/c/libosmo-netif/+/33198/16#message-034bf24dc42f2c001406a6f258b1500be103d49e Pau seems to be in favor of disabling timestamps in tests in general. I think this is a good idea - especially with the io_uring backend which certainly causes different amount of mainloop invocations than poll.
Another solution would be to avoid calling osmo_gettimeofday_override_add() inside the osmo_select_main() loop, but only when e.g. a read/write callback is called.
To view, visit change 34224. To unsubscribe, or for help writing mail filters, visit settings.