[Intel-gfx] [PATCH 1/3] drm/i915: Merge BDW pipe gamma and degamma table code

Ander Conselvan De Oliveira conselvan2 at gmail.com
Fri Feb 10 10:14:45 UTC 2017


On Thu, 2017-02-09 at 16:47 +0200, Ville Syrjälä wrote:
> On Fri, Feb 03, 2017 at 03:11:13PM +0200, Ander Conselvan de Oliveira wrote:
> > The only difference between the code loading the pipe gamma and degamma
> > tables in BDW is that the gamma code also writes the registers that hold
> > the maximum values. So we can use the gamma code for the degamma table,
> > at the expense of writing the maximum value register twice, with
> > potenttially wrong values in the first time.
> > 
> > v2: Pass PAL_PREC_SPLIT_MODE from the caller. (Ville)
> > Cc: Ville Syrjälä <ville.syrjala at linux.intel.com>
> > Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira at inte
> > l.com>
> > ---
> >  drivers/gpu/drm/i915/intel_color.c | 59 ++++++++++-------------------------
> > ---
> >  1 file changed, 15 insertions(+), 44 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/intel_color.c
> > b/drivers/gpu/drm/i915/intel_color.c
> > index 0627eee..82e1809 100644
> > --- a/drivers/gpu/drm/i915/intel_color.c
> > +++ b/drivers/gpu/drm/i915/intel_color.c
> > @@ -340,54 +340,20 @@ static void haswell_load_luts(struct drm_crtc_state
> > *crtc_state)
> >  		hsw_enable_ips(intel_crtc);
> >  }
> >  
> > -static void bdw_load_degamma_lut(struct drm_crtc_state *state)
> > +static void bdw_load_lut(struct drm_crtc_state *state, u32 offset,
> 
> Maybe just pass the intel_crtc instead of the state? I presume we don't
> need anything else from the state really?
> 
> > +			 struct drm_color_lut *lut, u32 lut_size,
> > +			 u32 flags)
> >  {
> >  	struct drm_i915_private *dev_priv = to_i915(state->crtc->dev);
> >  	enum pipe pipe = to_intel_crtc(state->crtc)->pipe;
> > -	uint32_t i, lut_size = INTEL_INFO(dev_priv)-
> > >color.degamma_lut_size;
> > -
> > -	I915_WRITE(PREC_PAL_INDEX(pipe),
> > -		   PAL_PREC_SPLIT_MODE | PAL_PREC_AUTO_INCREMENT);
> > -
> > -	if (state->degamma_lut) {
> > -		struct drm_color_lut *lut =
> > -			(struct drm_color_lut *) state->degamma_lut->data;
> > -
> > -		for (i = 0; i < lut_size; i++) {
> > -			uint32_t word =
> > -			drm_color_lut_extract(lut[i].red, 10) << 20 |
> > -			drm_color_lut_extract(lut[i].green, 10) << 10 |
> > -			drm_color_lut_extract(lut[i].blue, 10);
> > -
> > -			I915_WRITE(PREC_PAL_DATA(pipe), word);
> > -		}
> > -	} else {
> > -		for (i = 0; i < lut_size; i++) {
> > -			uint32_t v = (i * ((1 << 10) - 1)) / (lut_size -
> > 1);
> > -
> > -			I915_WRITE(PREC_PAL_DATA(pipe),
> > -				   (v << 20) | (v << 10) | v);
> > -		}
> > -	}
> > -}
> > -
> > -static void bdw_load_gamma_lut(struct drm_crtc_state *state, u32 offset)
> > -{
> > -	struct drm_i915_private *dev_priv = to_i915(state->crtc->dev);
> > -	enum pipe pipe = to_intel_crtc(state->crtc)->pipe;
> > -	uint32_t i, lut_size = INTEL_INFO(dev_priv)->color.gamma_lut_size;
> > +	uint32_t i;
> >  
> >  	WARN_ON(offset & ~PAL_PREC_INDEX_VALUE_MASK);
> >  
> >  	I915_WRITE(PREC_PAL_INDEX(pipe),
> > -		   (offset ? PAL_PREC_SPLIT_MODE : 0) |
> > -		   PAL_PREC_AUTO_INCREMENT |
> > -		   offset);
> > -
> > -	if (state->gamma_lut) {
> > -		struct drm_color_lut *lut =
> > -			(struct drm_color_lut *) state->gamma_lut->data;
> > +		   flags | PAL_PREC_AUTO_INCREMENT | offset);
> >  
> > +	if (lut) {
> >  		for (i = 0; i < lut_size; i++) {
> >  			uint32_t word =
> >  			(drm_color_lut_extract(lut[i].red, 10) << 20) |
> > @@ -430,9 +396,13 @@ static void broadwell_load_luts(struct drm_crtc_state
> > *state)
> >  		return;
> >  	}
> >  
> > -	bdw_load_degamma_lut(state);
> > -	bdw_load_gamma_lut(state,
> > -			   INTEL_INFO(dev_priv)->color.degamma_lut_size);
> > +	bdw_load_lut(state, 0, (struct drm_color_lut *) state->degamma_lut,
> 
> ->data ?
> 
> Such explicit casts are a good recipe for bugs I think, so would be nice
> if someone could make this blob stuff more type safe.

Yikes! Totally missed that. Indeed the cast is a really bad idea.

> Dunno, maybe some kind of 'void *blob_get_data()' type of thing, or
> maybe decouple the blob data from the blob once again so that we'd have
> 'void *data' in there.

For this patch, how about a 

static struct drm_color_lut *
blob_data_as_lut(struct drm_property_blob *blob)
{
        if (!blob)
                return NULL;

        return (struct drm_color_lut *) blob->data;
}

I think it is better than passing a void*.

Ander

> 
> > +		     INTEL_INFO(dev_priv)->color.degamma_lut_size,
> > +		     PAL_PREC_SPLIT_MODE);
> > +	bdw_load_lut(state, INTEL_INFO(dev_priv)->color.degamma_lut_size,
> > +		     (struct drm_color_lut *) state->gamma_lut,
> > +		     INTEL_INFO(dev_priv)->color.gamma_lut_size,
> > +		     PAL_PREC_SPLIT_MODE);
> >  
> >  	intel_state->gamma_mode = GAMMA_MODE_MODE_SPLIT;
> >  	I915_WRITE(GAMMA_MODE(pipe), GAMMA_MODE_MODE_SPLIT);
> > @@ -489,7 +459,8 @@ static void glk_load_luts(struct drm_crtc_state *state)
> >  	}
> >  
> >  	glk_load_degamma_lut(state);
> > -	bdw_load_gamma_lut(state, 0);
> > +	bdw_load_lut(state, 0, (struct drm_color_lut *) state->gamma_lut,
> > +		     INTEL_INFO(dev_priv)->color.gamma_lut_size, 0);
> >  
> >  	intel_state->gamma_mode = GAMMA_MODE_MODE_10BIT;
> >  	I915_WRITE(GAMMA_MODE(pipe), GAMMA_MODE_MODE_10BIT);
> > -- 
> > 2.9.3
> 
> 


More information about the Intel-gfx mailing list