[igt-dev] [PATCH] igt_chameleon: Assert if there are no connected ports

[Petri Latvala]

On Thu, Dec 01, 2022 at 11:37:19AM -0500, Mark Yacoub wrote:
> On Thu, Dec 1, 2022 at 9:38 AM Petri Latvala <petri.latvala at intel.com> wrote:
> >
> > On Wed, Nov 30, 2022 at 12:42:17PM -0500, Mark Yacoub wrote:
> > > [Why]
> > > The tests are meant to run on devices that have a chameleon connected.
> > > If autodiscovery finds no port, it means something is off and the DUT is
> > > not behaving well.
> > >
> > > [How]
> > > Add an assert if no port is mapped.
> > >
> > > Signed-off-by: Mark Yacoub <markyacoub at chromium.org>
> > > ---
> > >  lib/igt_chamelium.c | 4 ++++
> > >  1 file changed, 4 insertions(+)
> > >
> > > diff --git a/lib/igt_chamelium.c b/lib/igt_chamelium.c
> > > index 79920de1..4e7a7a8f 100644
> > > --- a/lib/igt_chamelium.c
> > > +++ b/lib/igt_chamelium.c
> > > @@ -2749,6 +2749,10 @@ unplug_port:
> > >       igt_debug("Auto-discovery took %fms and found %i connector(s)\n",
> > >                 (float)elapsed_ns / (1000 * 1000), chamelium->port_count);
> > >
> > > +     igt_assert_f(
> > > +             is_any_port_mapped,
> > > +             "No port is mapped. We don't expect to have no connected connectors.\n");
> > > +
> >
> > If this situation (chamelium is configured but is not suitable for
> > testing) ever happens, in my opinion an igt_abort would be more
> > appropriate. Similar to the case where chamelium is configured but
> > cannot be reached. What do you think?
> >
> So the case I saw is volteer failing to connect any of its ports
> because of some bug it has that if it doesn't disconnect cleanly it
> stops working without extra steps. I thought when we catch this, this
> might be a good candidate for a real bug to file so it's a legit DUT
> failure, hence the assert, rather than abort because test conditions
> weren't met.
> They're both good to me, so whatever you think is best.

An abort result should be considered a bad result by CI systems. And a
DUT failure should look different from a kernel failure in my
opinion. Let's use abort here.


-- 
Petri Latvala