Re: [PATCH] drm/msm/dp: Make it possible to enable the test pattern
From: Bjorn Andersson
Date: Tue Jun 29 2021 - 10:46:10 EST
On Mon 28 Jun 20:07 CDT 2021, abhinavk@xxxxxxxxxxxxxx wrote:
> On 2021-06-28 17:55, Bjorn Andersson wrote:
> > On Mon 28 Jun 19:31 CDT 2021, abhinavk@xxxxxxxxxxxxxx wrote:
> >
> > > Hi Bjorn
> > >
> > > On 2021-06-28 17:22, Bjorn Andersson wrote:
> > > > The debugfs interface contains the knobs to make the DisplayPort
> > > > controller output a test pattern, unfortunately there's nothing
> > > > currently that actually enables the defined test pattern.
> > > >
> > > > Fixes: de3ee25473ba ("drm/msm/dp: add debugfs nodes for video pattern
> > > > tests")
> > > > Signed-off-by: Bjorn Andersson <bjorn.andersson@xxxxxxxxxx>
> > >
> > > This is not how this debugfs node works. This is meant to be used
> > > while
> > > running
> > > DP compliance video pattern test.
> > >
> > > https://gitlab.freedesktop.org/drm/igt-gpu-tools/-/blob/master/tools/msm_dp_compliance.c
> > >
> > > While the compliance test is being run with this msm_dp_compliance app
> > > running,
> > > it will draw the test pattern when it gets the "test_active" from the
> > > driver.
> > >
> > > The test pattern which this app draws is as per the requirements of
> > > the
> > > compliance test
> > > as the test equipment will match the CRC of the pattern which is
> > > drawn.
> > >
> > > The API dp_panel_tpg_config() which you are trying to call here
> > > draws the DP
> > > test pattern
> > > from the DP controller hardware but not the pattern which the
> > > compliance
> > > test expects.
> > >
> >
> > So clearly not an oversight, but rather me not understanding how to use
> > the test pattern.
> >
> > You say that I should run msm_dp_compliance while the test is running,
> > so how do I run the test?
>
> There are two test patterns with different purposes. The one which the
> msm_dp_compliance
> draws is strictly for the DP compliance test and it needs even the DPU to
> draw the frame because
> it sets up the display pipeline and just draws the buffer.
>
> That is not what you are looking for here.
>
> So rather than trying to run msm_dp_compliance on your setup, just try
> calling dp_panel_tpg_config().
> We typically just call this API, right after the link training is done.
> But if you really need a debugfs node for this, you can write up a separate
> debugfs for it
> Something like:
>
> echo 1 > dp/tpg/en
>
Having the ability to turn on the test pattern was very useful to me and
I would use this next time I'm adding DP support on a new platform. So
adding some way of invoking that API without a lot of extra effort seems
useful.
> Lets not disturb this one.
>
Agreed.
Thanks,
Bjorn
> >
> > > Its just a debug API to call when required during bringup/debug
> > > purposes.
> > >
> >
> > Yes, I was trying to isolate the DP code from some misconfiguration in
> > the DPU during bringup and with this fix the debugfs interface became
> > useful.
>
> >
> > Regards,
> > Bjorn
> >
> > > Hence this is not the place to call it as it will end up breaking CTS.
> > >
> > > Thanks
> > >
> > > Abhinav
> > >
> > > > ---
> > > > drivers/gpu/drm/msm/dp/dp_debug.c | 2 ++
> > > > 1 file changed, 2 insertions(+)
> > > >
> > > > diff --git a/drivers/gpu/drm/msm/dp/dp_debug.c
> > > > b/drivers/gpu/drm/msm/dp/dp_debug.c
> > > > index 2f6247e80e9d..82911af44905 100644
> > > > --- a/drivers/gpu/drm/msm/dp/dp_debug.c
> > > > +++ b/drivers/gpu/drm/msm/dp/dp_debug.c
> > > > @@ -305,6 +305,8 @@ static ssize_t dp_test_active_write(struct file
> > > > *file,
> > > > debug->panel->video_test = true;
> > > > else
> > > > debug->panel->video_test = false;
> > > > +
> > > > + dp_panel_tpg_config(debug->panel, debug->panel->video_test);
> > > > }
> > > > }
> > > > drm_connector_list_iter_end(&conn_iter);