Re: [Intel-gfx] [PATCH 07/14] drm/i915: Clean up pre-skl primary plane registers

["Ville Syrjälä" <ville.syrjala@linux.intel.com>]

On Tue, Jan 18, 2022 at 01:40:41PM +0000, Souza, Jose wrote:
> On Tue, 2022-01-18 at 02:55 +0200, Ville Syrjälä wrote:
> > On Wed, Jan 12, 2022 at 08:12:31PM +0000, Souza, Jose wrote:
> > > On Wed, 2021-12-01 at 17:25 +0200, Ville Syrjala wrote:
> > > > @@ -427,9 +427,9 @@ static void i9xx_plane_update_noarm(struct intel_plane *plane,
> > > >  		 * program whatever is there.
> > > >  		 */
> > > >  		intel_de_write_fw(dev_priv, DSPPOS(i9xx_plane),
> > > > -				  (crtc_y << 16) | crtc_x);
> > > > +				  DSP_POS_Y(crtc_y) | DSP_POS_X(crtc_x));
> > > >  		intel_de_write_fw(dev_priv, DSPSIZE(i9xx_plane),
> > > > -				  ((crtc_h - 1) << 16) | (crtc_w - 1));
> > > > +				  DSP_HEIGHT(crtc_h - 1) | DSP_POS_X(crtc_w - 1));
> > > 
> > > DSP_HEIGHT(crtc_h - 1) | DSP_WIDTH(crtc_w - 1));
> > 
> > Whoops. Thanks for cathcing that.
> > 
> > <snip>
> > > > +#define   DSP_ENABLE			REG_BIT(31)
> > > 
> > > I really don't like DSP, it is broadly used acronym to Digital Signal Processors.
> > > Would prefer to have DISPLAY or DISP.
> > 
> > The registers are called DSP<foo>, so the spec makes the case for DSP_.
> > The problem with DISP_/etc. is that the namespace then makes it a bit
> > hard to figure out what register the defines belong to.
> > 
> > > 
> > > Anyways, DSP_ENABLE should have also have plane on it.
> > 
> > DSP==display plane. Any more would be redundant.
> 
> Damn, even worst, thought it was DiSPlay.

Well I guess it might be that too. I think the old docs tend to use
"display A" vs. "display plane A" etc. interchangeably when talking
about planes.

> But if this is the BSpec name, go ahead with it. 

I guess I could be convinced to use DISP_ just to raise a few less
eyebrows. Just a bit sad that the namespace won't match the register
name then. But I suppose we have that sort of thing going on in other
places too.

-- 
Ville Syrjälä
Intel