Re: [PATCH 10/27] drm/atomic-helper: nonblocking commit support

On Wed, Jun 8, 2016 at 6:19 PM, Daniel Vetter daniel@ffwll.ch wrote:

But the main one is kms_flip - this is one shockingly nasty testcase ;-)

Also note that kms_flip wasn't just run on i915, but also on other drivers in this conversion (not all iirc), thanks to collabora's work to port igt testcase to generic kms. Again it mostly uncovered bugs in drivers (after I debugged some silly gotchas in the helpers) with a combination of kms_flip sanity checks and the runtime checks in the helpers themselves. I think it's fair to claim that if you have an atomic driver and switch over to these helpers + run kms_flip, you're driver will have gained a few bugfixes until it manages to pass - all of them needed some ;-) -Daniel