When a timeout occurs in a running test, what we see in the logs is: 'ERROR: Exit status is 9' It would be helpful if ptest-runner2 actually reported the reason for the error status as timeout related (otherwise someone might erroneously think the exit code is coming from the ptest itself and not the ptest-runner). See https://bugzilla.yoctoproject.org/show_bug.cgi?id=14258#c6 And the log: https://autobuilder.yoctoproject.org/typhoon/#/builders/82/builds/1998/steps/12/logs/stdio 'ERROR: Exit status is 9\n' 'DURATION: 2904\n', 'strace': ['kill_child.test']} The code in question: http://git.yoctoproject.org/cgit/cgit.cgi/ptest-runner2/tree/utils.c#n529 It seems that there is a different path to timeout that is not being trapped by: http://git.yoctoproject.org/cgit/cgit.cgi/ptest-runner2/tree/utils.c#n533
Might be related: https://lists.yoctoproject.org/g/yocto/message/54739 https://lists.yoctoproject.org/g/yocto/message/54740
I think so. Can you try the patches locally?
Anibal, If you can do this great, if not, ping me and I'll take a look.
I was tempted to use this library: http://libexplain.sourceforge.net/ https://linux.die.net/man/3/explain_waitpid but we probably shouldn't add complexity to ptest-runner. What do you think Anibal?
The library isn't going to help much. We know that error 9 is killing ptest on timeout, what we need is as much information as possible about why ptest got stuck.
(In reply to comment #5) > The library isn't going to help much. We know that error 9 is killing ptest > on timeout, what we need is as much information as possible about why ptest > got stuck. Yes I think the unique way is to see the logs and figure out why got stuck or reproduce the issue locally.
I think this has already been fixed? https://git.yoctoproject.org/ptest-runner2/log/
Fixed as described above.