[Intel-gfx] [PATCH 7/8] drm/i915/gen9+: Program watermarks as a separate step during evasion

[Maarten Lankhorst]

Op 12-10-16 om 19:15 schreef Lyude:
> Accidentally sent original view twice and found one more issue after
> looking at the rest of them, sorry about that!
>
> On Wed, 2016-10-12 at 13:04 -0400, Lyude wrote:
>> Loving this patch so far! Would it be possible to get this split into
>> two separate patches though? One for removing skl_results and one for
>> programming watermarks as a separate step.
Yeah the small hunk has to be moved to patch #5.

I can't split this patch up. skl_results becomes useless after moving the
programming to a separate step. The old way of doing it was storing
intel_state->wm_results in dev_priv->wm.skl_results and using it in the
plane callbacks.

With the watermark update in its own function, this becomes useless
because the pointer to watermarks can be retrieved from state directly.

~Maarten