[Intel-gfx] [PATCH i-g-t v3] kms_rotation_crc: 90 degree flip test is not a stress test

[Chris Wilson]

Quoting Tvrtko Ursulin (2017-09-08 15:54:18)
> 
> On 08/09/2017 15:06, Chris Wilson wrote:
> > Quoting Tvrtko Ursulin (2017-09-08 12:24:07)
> >> From: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
> >> +                       /*
> >> +                        * If flips are requested flip away and back before
> >> +                        * checking CRC.
> > 
> > And back? We only check of the original framebuffer and not the rotated?
> > Or am I missing the point...
> 
> Yes, hm, it would be really bad if both flips silently did nothing, but 
> the last CRC would still match. So yes, still a holey hole there.
> 
> (Note we are not changing rotation with the flip here, just flipping 
> between two equally rotated fbs.)

Although with pageflips mapped through to a nonblocking modeset, there
shouldn't be any problems now with flipping to a new rotation, stride,
format, etc. I guess they are all checked somewhere, but I also guess
that somewhere is here ;)
-Chris