[LTP] [PATCH/RFC] tst_process_state_wait: wait for schedstats to settle when state == S

[Cyril Hrubis <chrubis@suse.cz>]

Hi!
> > hb->lock is locked at this point, and requeue takes it too, so I'm not
> > sure what makes it fail. I've seen testcase fail in at least
> > 2 different ways now. Here's the other one:
> 
> Looking at the test shouldn't we wait for the processes to be sleeping
> after the futex_cmp_requeue() syscall?
> 
> i.e.
> 
> diff --git a/testcases/kernel/syscalls/futex/futex_cmp_requeue01.c b/testcases/kernel/syscalls/futex/futex_cmp_requeue01.c
> index f5264c2ba..41f5e88a6 100644
> --- a/testcases/kernel/syscalls/futex/futex_cmp_requeue01.c
> +++ b/testcases/kernel/syscalls/futex/futex_cmp_requeue01.c
> @@ -71,6 +71,9 @@ static void verify_futex_cmp_requeue(unsigned int n)
>                         TST_RET, exp_ret);
>         }
> 
> +       for (i = 0; i < tc->num_waiters; i++)
> +               TST_PROCESS_STATE_WAIT(pid[i], 'S');
> +
>         num_requeues = futex_wake(&futexes[1], tc->num_waiters, 0);
>         num_waits = futex_wake(&futexes[0], tc->num_waiters, 0);

And this obviously wouldn't work since we are waking up some of them in
a subset of the testcases.

-- 
Cyril Hrubis
chrubis@suse.cz