[Intel-gfx] [PATCH] drm/i915: Tile F plane format support

[Lisovskiy, Stanislav]

On Thu, Sep 23, 2021 at 01:44:11PM +0300, Lisovskiy, Stanislav wrote:
> On Thu, Sep 23, 2021 at 01:28:21PM +0300, Jani Nikula wrote:
> > On Thu, 23 Sep 2021, Stanislav Lisovskiy <stanislav.lisovskiy at intel.com> wrote:
> > > @@ -1941,6 +1951,10 @@ static bool gen12_plane_format_mod_supported(struct drm_plane *_plane,
> > >  		if (IS_ADLP_DISPLAY_STEP(dev_priv, STEP_A0, STEP_B0))
> > >  			return false;
> > >  		break;
> > > +	case I915_FORMAT_MOD_F_TILED:
> > > +		if (!HAS_FTILE(dev_priv))
> > > +			return false;
> > > +		fallthrough;
> > >  	default:
> > >  		return false;
> > >  	}
> > 
> > Seems odd.
> 
> I agree, however this wasn't even added by me.
> This patch got changed in ridiculous ways since last time,
> I ever touched it. 
> Currently we have it internally exactly same way(wondering why)
> Unfortunately didn't pay attention to this, was assuming
> that if its r-bed and pushed - can trust it.

Actually checked - seems to be result of wrong merge conflict resolution
by me.

It was initially that way:

-	case DRM_FORMAT_MOD_LINEAR:
-	case I915_FORMAT_MOD_X_TILED:
-		break;
 	case I915_FORMAT_MOD_Y_TILED:
 		if (IS_DG2(dev_priv))
 			return false;
@@ -1928,6 +1934,13 @@  static bool gen12_plane_format_mod_supported(struct drm_plane *_plane,
 		if (IS_DG2(dev_priv))
 			return false;
 		break;
+	case I915_FORMAT_MOD_F_TILED:
+		if (!HAS_FTILE(dev_priv))
+			return false;
+		fallthrough;
+	case DRM_FORMAT_MOD_LINEAR:
+	case I915_FORMAT_MOD_X_TILED:
+		break;

So my bad here. Need to put this back.

Stan


> 
> Stan
> 
> 
> > 
> > BR,
> > Jani.
> > 
> > 
> > -- 
> > Jani Nikula, Intel Open Source Graphics Center