When processes terminate, we really want all of the child processes to terminate too. This was not happening for worker processes which spawned their own multiprocessing pools, leading to build hangs. This change ensures any sigterm gets passed to the whole process group. In local tests, this resolved some hanging process workloads I could generate. It does rely on signals being delivered in a timely fashion and there is a multiprocessing bug we have to work around there.
Signed-off-by: Richard Purdie <richard.pur...@linuxfoundation.org> diff --git a/bitbake/bin/bitbake-worker b/bitbake/bin/bitbake-worker index 05e0cf6..d1ff5b3 100755 --- a/bitbake/bin/bitbake-worker +++ b/bitbake/bin/bitbake-worker @@ -81,6 +81,11 @@ def workerlog_write(msg): lf.write(msg) lf.flush() +def sigterm_handler(signum, frame): + signal.signal(signal.SIGTERM, signal.SIG_DFL) + os.killpg(0, signal.SIGTERM) + sys.exit() + def fork_off_task(cfg, data, workerdata, fn, task, taskname, appends, taskdepdata, quieterrors=False): # We need to setup the environment BEFORE the fork, since # a fork() or exec*() activates PSEUDO... @@ -132,7 +137,7 @@ def fork_off_task(cfg, data, workerdata, fn, task, taskname, appends, taskdepdat global worker_pipe pipein.close() - signal.signal(signal.SIGTERM, signal.SIG_DFL) + signal.signal(signal.SIGTERM, sigterm_handler) # Save out the PID so that the event can include it the # events -- _______________________________________________ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core