[igt-dev] [PATCH i-g-t] tests/kms_flip: Restrict the hang tests execution to first and last pipe combinations

[B, Jeevan]

> -----Original Message-----
> From: Kamil Konieczny <kamil.konieczny at linux.intel.com>
> Sent: Wednesday, July 26, 2023 8:33 PM
> To: igt-dev at lists.freedesktop.org
> Cc: B, Jeevan <jeevan.b at intel.com>
> Subject: Re: [igt-dev] [PATCH i-g-t] tests/kms_flip: Restrict the hang tests
> execution to first and last pipe combinations
> 
> Hi Jeevan,
> 
> We already have patch with: "Restrict the hang tests execution to first and
> last pipe combinations"
> so maybe change your message to: Restrict even further hang tests
> 
> On 2023-07-18 at 16:30:06 +0530, Jeevan B wrote:
> > restrict the hang tests execution to single pipe combinations still
> - ^
> Start from uppercase, s/restrict/Restrict/
> 
> > causing timed-out so limiting the combination to first and last pipe.
> 
> Please write which tests where so long and give some log example why it
> helped, how much less it took?
> 
> >
> > Signed-off-by: Jeevan B <jeevan.b at intel.com>
> > ---
> >  tests/kms_flip.c | 6 +++++-
> >  1 file changed, 5 insertions(+), 1 deletion(-)
> >
> > diff --git a/tests/kms_flip.c b/tests/kms_flip.c index
> > ef2234f74..ee4e8a43d 100755
> > --- a/tests/kms_flip.c
> > +++ b/tests/kms_flip.c
> > @@ -1729,10 +1729,14 @@ static void run_pair(int duration, int flags)
> >  					crtc_idxs[0] = n;
> >  					crtc_idxs[1] = m;
> >
> > -					/* Limit the execution to PIPE_A
> combination for hang tests */
> > +					/* Limit the execution to 2 CRTCs
> > +					 * (first and last combination) for
> hang tests */
> 
> Write multiline comment starting with empty /* like:
> 					/*
> 					 * Limit the execution to 2 CRTCs
> 					 * (first and last combination) for
> hang tests
> 					 */
> 
> >  					if ((flags & TEST_HANG) && !all_pipes
> &&
> >  					    (n != 0 && n != resources-
> >count_crtcs))
> >  						continue;
> > +					if ((flags & TEST_HANG) && !all_pipes
> &&
> > +					    m != resources->count_crtcs - 1)
> > +						continue;
> >
> 
> Why not putting it all together?
>   					if ((flags & TEST_HANG) && !all_pipes
> &&
>   					    n != 0 && n != resources-
> >count_crtcs &&
> 					    m != resources->count_crtcs - 1)
>   						continue;
> 
> I made some simple test with single monitor connected and did not see any
> difference. In what configuration did you test it?
This is for dual display test and HANG subtest specific. So with the pervious patch time was reduced to ~120s from ~400s but in CI still tests were getting timed-out
So now we have tests running on only first and last pipe combination. 
EX: 
Without patch   display 1 & 2: AB,AC & AD 
		display 2 & 3: AB, AC & AD
		display 1 & 3: AB, AC & AD
Now with patch it will only run on AD combination for all display combination.  

> 
> Regards,
> Kamil
> 
> >  					run_test_on_crtc_set(&o, crtc_idxs,
> >  							     RUN_PAIR,
> > --
> > 2.41.0
> >