[Intel-gfx] [PATCH i-g-t] tests/kms_frontbuffer_tracking: Skip tests when PSR cannot be enabled

[Pandiyan, Dhinakaran]

On Thu, 2017-10-26 at 17:35 -0200, Paulo Zanoni wrote:
> Em Qui, 2017-10-26 às 12:32 -0700, Rodrigo Vivi escreveu:
> > On Thu, Oct 26, 2017 at 01:29:57PM +0000, Paulo Zanoni wrote:
> > > Em Qua, 2017-10-25 às 17:37 -0700, Dhinakaran Pandiyan escreveu:
> > > > The frontbuffer_tracking PSR tests fail if PSR cannot be
> > > > activated
> > > > when
> > > > there is sink support. But, there are several other requirements
> > > > related to
> > > > mode timings that have to be satisfied before PSR can be enabled.
> > > > No
> > > > reason
> > > > to fail these tests when PSR cannot be enabled.
> > > > 
> > > > Cc: Paulo Zanoni <paulo.r.zanoni at intel.com>
> > > > Cc: Rodrigo Vivi <rodrigo.vivi at intel.com>
> > > > Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan at intel.com
> > > > >
> > > > ---
> > > >  tests/kms_frontbuffer_tracking.c | 8 ++++----
> > > >  1 file changed, 4 insertions(+), 4 deletions(-)
> > > > 
> > > > diff --git a/tests/kms_frontbuffer_tracking.c
> > > > b/tests/kms_frontbuffer_tracking.c
> > > > index a068c8af..13c948de 100644
> > > > --- a/tests/kms_frontbuffer_tracking.c
> > > > +++ b/tests/kms_frontbuffer_tracking.c
> > > > @@ -1548,12 +1548,12 @@ static void teardown_fbc(void)
> > > >  {
> > > >  }
> > > >  
> > > > -static bool psr_sink_has_support(void)
> > > > +static bool psr_has_support(void)
> > > >  {
> > > >  	char buf[256];
> > > >  
> > > >  	debugfs_read("i915_edp_psr_status", buf);
> > > > -	return strstr(buf, "Sink_Support: yes\n");
> > > > +	return strstr(buf, "Enabled: yes\n");
> > > 
> > > What if PSR can be enabled but is just not enabled due to a bug?

My reasoning was psr.enabled will be true as long as intel_psr_enable()
is called. But, yeah there is a chance that intel_psr_enable() is not
called.

> > 
> > then psr basic should be the one failing...
> 
> But if we apply this patch we'll skip it instead of fail.
> 
> > 
> > > 
> > > I suppose we could print on debugfs whether source+sink support
> > > equals
> > > yes.
> > 
> > or maybe here instead of sink: yes or enable: yes
> > we could check
> > sink: yes && source: yes
> > 

That's how I wrote this initially, but source_ok appears to be in the
wrong place in the driver, i.e., source_ok is set inside psr_enable()
and hence has the same effect as checking for enabled. Moving source_ok
= true inside psr_compute_config() and changing the IGT to test for
sink_support && source_ok will solve our problem.


> > if sink yes and source yes and psr is still disabled when we
> > expect that enabled than we have the bug paulo mentioned.
> > 
> > > 
> > > >  }
> > > >  
> > > >  static void setup_psr(void)
> > > > @@ -1564,8 +1564,8 @@ static void setup_psr(void)
> > > >  		return;
> > > >  	}
> > > >  
> > > > -	if (!psr_sink_has_support()) {
> > > > -		igt_info("Can't test PSR: not supported by
> > > > sink.\n");
> > > > +	if (!psr_has_support()) {
> > > > +		igt_info("Can't test PSR: not supported.\n");
> > > >  		return;
> > > >  	}
> > > >  	psr.can_test = true;