On Thu, 31 Mar 2022 13:18:56 +0530 Naresh Kamboju wrote: > Hi Jakub, > > > Can you check where the process is stuck and it's state? > > /proc/$pid/stack and run that thru scripts/decode_stacktrace > > > > Steps to reproduce: > - cd /opt/kselftests/default-in-kernel/net > - ./tls & > - tests_pid=$! > - echo $tests_pid > - sleep 90 > - cat /proc/$tests_pid/stack | tee tests_pid_stack.log > - cat tests_pid_stack.log > > [<0>] do_wait+0x191/0x3a0 > [<0>] kernel_wait4+0xaf/0x160 > [<0>] __do_sys_wait4+0x85/0x90 > [<0>] __x64_sys_wait4+0x1c/0x20 > [<0>] do_syscall_64+0x5c/0x80 > [<0>] entry_SYSCALL_64_after_hwframe+0x44/0xae The kernel test harness runs the tests in a separate process, this is the stack trace for the main process. We'll need a stack trace for the child, so: main_pid=$! tests_pid=$(pgrep -P $main_pid) Also since each test runs in a separate child, we can try to catch the one that got stuck for longer than 30 sec: pp=$(pgrep -P $main_pid) while true; do sleep 30 p=$(pgrep -P $main_pid) if [ $p != $pp ]; then pp=$p echo "New PID $p, continue waiting..." continue fi echo "PID $p is stuck!" cat /proc/$p/stack done > Detail test log can be found here in this link [1] > > I do not see any output from > ./scripts/decode_stacktrace.sh stack-dump.txt > > > - Naresh > > [1] https://lkft.validation.linaro.org/scheduler/job/4812800#L2256
