[igt-dev] [PATCH i-g-t] tests/kms_flip: Keep signal helpers enabled after a failed interruptible subtest

[Anshuman Gupta]

On 2020-05-21 at 19:52:34 +0530, Shankar, Uma wrote:
> 
> 
> > -----Original Message-----
> > From: igt-dev <igt-dev-bounces at lists.freedesktop.org> On Behalf Of Anshuman
> > Gupta
> > Sent: Thursday, May 21, 2020 6:33 PM
> > To: Deak, Imre <imre.deak at intel.com>
> > Cc: igt-dev at lists.freedesktop.org; Gupta, Nidhi1 <nidhi1.gupta at intel.com>
> > Subject: Re: [igt-dev] [PATCH i-g-t] tests/kms_flip: Keep signal helpers enabled
> > after a failed interruptible subtest
> > 
> > On 2020-05-19 at 18:29:56 +0530, Imre Deak wrote:
> > > For the duration of all the interruptible subtests the signal helper
> > > is assumed to stay running after each subtest, both in case the
> > > subtest passes or aborts with a failure. So make sure we don't leave
> > > the signal helper suspended in case igt_assert() aborts.
> > >
> > > References: https://gitlab.freedesktop.org/drm/intel/issues/1883
> > > Fixes: 60e8be7ccc72 ("tests/kms_flip: Retry test in case of a DP/HDMI
> > > link reset")
> > > Cc: Uma Shankar <uma.shankar at intel.com>
> > > Cc: Nidhi Gupta <nidhi1.gupta at intel.com>
> > > Signed-off-by: Imre Deak <imre.deak at intel.com>
> > Looks good to me,
> > But i have some doubts regarding the reported failure in issue 1883.
> > > ---
> > >  tests/kms_flip.c | 7 ++++++-
> > >  1 file changed, 6 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/tests/kms_flip.c b/tests/kms_flip.c index
> > > 46bfc5858..7bf6958af 100755
> > > --- a/tests/kms_flip.c
> > > +++ b/tests/kms_flip.c
> > > @@ -1325,10 +1325,15 @@ retry:
> > >  	 * reset event, which the driver signals with a hotplug event.
> > >  	 */
> > >  	if (!state_ok) {
> > > +		bool hotplug_detected;
> > > +
> > >  		igt_suspend_signal_helper();
> > > -		igt_assert(!retried && igt_hotplug_detected(mon, 3));
> > AFAIU correctly the test didn't received any HOTPLUG uevent due to any
> > wait_event_interruptible() waitqueue in driver blocking it till 3 seconds, if this
> > waitqueue in kernel get interrupted by SIGCONT signal, which is primarly signal
> > helper in this igt, then igt_hotplug_detected() will detect a hotplug. Hope my
> > understading here is correct here.
> 
> Here since we have suspended the signal handler which will ignore or mask any interrupts,
> we need to ensure to resume it before we assert.
IMHO igt_suspend_signal_helper send a SIGSTOP to signal helper process, which will stop
the signal helper execution, which was actually signalling SIGCONT to parent process at 500Hz
intervel.
I am not able to see if it masks any interrupt here. 
Please correct me if i am wrong here.
I am interested here to know why it has not received HOTPLUG event for a period of 3 second,
what was really blocking it from kernel driver.
Thanks,
Anshuman Gupta.

> 
> > Thanks,
> > Anshuman Gupta.
> > > +		if (!retried)
> > > +			hotplug_detected = igt_hotplug_detected(mon, 3);
> > >  		igt_resume_signal_helper();
> > >
> > > +		igt_assert(!retried && hotplug_detected);
> > > +
> > >  		igt_debug("Retrying after a hotplug event\n");
> > >  		retried = true;
> > >  		memset(&o->vblank_state, 0, sizeof(o->vblank_state));
> > > --
> > > 2.23.1
> > >
> > > _______________________________________________
> > > igt-dev mailing list
> > > igt-dev at lists.freedesktop.org
> > > https://lists.freedesktop.org/mailman/listinfo/igt-dev
> > _______________________________________________
> > igt-dev mailing list
> > igt-dev at lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/igt-dev