[Intel-gfx] [PATCH] drm/i915: Introduce concept of a sub-platform

[Ville Syrjälä]

On Fri, Mar 15, 2019 at 05:55:19PM +0200, Ville Syrjälä wrote:
> On Fri, Mar 15, 2019 at 02:21:57PM +0000, Tvrtko Ursulin wrote:
> > 
> > On 15/03/2019 14:09, Ville Syrjälä wrote:
> > > On Fri, Mar 15, 2019 at 12:26:33PM +0000, Tvrtko Ursulin wrote:
> > >> -#define IS_KBL_ULX(dev_priv)	(INTEL_DEVID(dev_priv) == 0x590E || \
> > >> -				 INTEL_DEVID(dev_priv) == 0x5915 || \
> > >> -				 INTEL_DEVID(dev_priv) == 0x591E)
> > >> -#define IS_AML_ULX(dev_priv)	(INTEL_DEVID(dev_priv) == 0x591C || \
> > >> -				 INTEL_DEVID(dev_priv) == 0x87C0)
> > >> +#define IS_HSW_ULX(dev_priv) \
> > >> +	IS_SUBPLATFORM(dev_priv, INTEL_HASWELL, INTEL_SUBPLATFORM_ULX)
> > >> +#define IS_SKL_ULT(dev_priv) \
> > >> +	IS_SUBPLATFORM(dev_priv, INTEL_SKYLAKE, INTEL_SUBPLATFORM_ULT)
> > >> +#define IS_SKL_ULX(dev_priv) \
> > >> +	IS_SUBPLATFORM(dev_priv, INTEL_SKYLAKE, INTEL_SUBPLATFORM_ULX)
> > >> +#define IS_KBL_ULT(dev_priv) \
> > >> +	IS_SUBPLATFORM(dev_priv, INTEL_KABYLAKE, INTEL_SUBPLATFORM_ULT)
> > >> +#define IS_KBL_ULX(dev_priv) \
> > >> +	IS_SUBPLATFORM(dev_priv, INTEL_KABYLAKE, INTEL_SUBPLATFORM_ULX)
> > >> +#define IS_AML_ULX(dev_priv) \
> > >> +	IS_SUBPLATFORM(dev_priv, INTEL_KABYLAKE, INTEL_SUBPLATFORM_AML_ULX)
> > > 
> > > Why is AML_ULX different from normal ULX?
> > 
> > As far as I could it is a different devid which we officially call 
> > Amberlake, but did not add it as separate platform.
> > 
> > Given how this macro is used, which is always in conjuction with 
> > IS_KBL_ULX, I considered just removing it and adding the devid to 
> > KBL_ULX, but did not want to interferre with other people's ideas.
> 
> This whole AML_ULX makes no sense. The "non-ULX" AML is also
> docucemented to be a Y SKU, so AFAICS it should use the exact
> same codepaths as KBL Y (aka. KBL_ULX). Cc:ing Jose who added this
> second AML variant...

Actually cc this time.

> 
> I'm really tempted to just go and nuke all CFL,AML,etc. from the
> code to make it actually match the spec.
> 

-- 
Ville Syrjälä
Intel