pá 14. 8. 2026 v 16:19 odesílatel Steven Rostedt <[email protected]> napsal: > > On Tue, 11 Aug 2026 09:22:54 +0200 > Tomas Glozar <[email protected]> wrote: > > > tools/tracing/rtla/tests/scripts/lib/get_workload_pids.sh is for some > > reason not finding the user threads (as the tests for -H and -k > > succeeded) on your system. The threads themselves had to be running, > > as timerlat is reporting samples from them. The script and the related > > tests cases were added in 7.2 already, this pull request does not > > touch them so it's most likely a pre-existing issue. > > > > Either way, I will look into that, thank you for testing. It might be > > environment-dependent, as the test relies on the test script (executed > > as an --on-threshold action) being a direct child of the main rtla > > process - which only holds when the system() shell executes the script > > directly, not as a subprocess. > > I just pushed this to the for-next branch. I'll let you figure out if there > needs to be any more updates. > > -- Steve >
Thanks. As this is now a known bug in the test (and is already in 7.2 rather than a 7.3 regression in the pull request), I was not rushing to fix that and rather continued to focus on my patchset. Of course, I'm trying to keep tests passing, now that the majority of high-frequency intermittent failures are fixed, so that people running tests are not desensitized to failures. Tomas
