[Intel-gfx] [PATCH i-g-t] tests/kms_flip: Add test to check suspend/resume

[Daniel Vetter]

On Mon, Jan 21, 2019 at 07:34:32AM -0200, Rodrigo Siqueira wrote:
> Hi,
> 
> On 01/18, Shayenne Moura wrote:
> > This patch adds one test to evaluate suspend/resume operations using kms_flip.
> > 
> > Signed-off-by: Shayenne Moura <shayenneluzmoura at gmail.com>
> > ---
> >  tests/kms_flip.c | 1 +
> >  1 file changed, 1 insertion(+)
> >  mode change 100644 => 100755 tests/kms_flip.c
> > 
> > diff --git a/tests/kms_flip.c b/tests/kms_flip.c
> > old mode 100644
> > new mode 100755
> > index f28272dd..3ca2fdfc
> > --- a/tests/kms_flip.c
> > +++ b/tests/kms_flip.c
> > @@ -1567,6 +1567,7 @@ int main(int argc, char **argv)
> >  		{ 10, TEST_DPMS_OFF | TEST_DPMS | TEST_VBLANK_RACE, "dpms-vs-vblank-race" },
> >  		{ 10, TEST_MODESET | TEST_VBLANK_RACE, "modeset-vs-vblank-race" },
> >  		{ 0, TEST_BO_TOOBIG | TEST_NO_2X_OUTPUT, "bo-too-big" },
> > +		{ 30, TEST_FLIP | TEST_SUSPEND, "flip-vs-suspend" },
> 
> I remember to follow a conversation in the IRC that you said that VKMS
> pass this test. I tried it here on my VM, and after running the test my
> system got freeze. My VM uses Arch Linux with Kernel 5.0.0-rc1. Did I
> miss something?
> 
> Also, I tested it on my host machine with i915 driver and I noticed that
> the test took much more than 30 seconds to finish. Is it right?

Yeah should probably reduce to 10 or something. The susend-resume will
take 10+ seconds at least, and we'lld do one per CRTC, so this is going to
take a while anyway. Note that the test timeout is per-CRTC iirc.
-Daniel

> 
> Thanks
> Best Regards
> 
> >  	};
> >  	int i;
> >  
> > -- 
> > 2.17.1
> > 
> 
> -- 
> Rodrigo Siqueira
> https://siqueira.tech
> Graduate Student
> Department of Computer Science
> University of São Paulo



-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch