[PATCH v5 21/24] drm/msm/dpu: Implement 10-bit color alpha for v12.0 DPU

[Dmitry Baryshkov]

On Fri, May 23, 2025 at 10:02:03AM +0300, Abel Vesa wrote:
> On 25-05-23 09:55:00, Abel Vesa wrote:
> > On 25-04-30 15:00:51, Krzysztof Kozlowski wrote:
> > > v12.0 DPU on SM8750 comes with 10-bit color alpha.  Add register
> > > differences and new implementations of setup_alpha_out(),
> > > setup_border_color() and setup_blend_config().
> > > 
> > > Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov at oss.qualcomm.com>
> > > Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski at linaro.org>
> > > 
> > > ---
> > > 
> > > Changes in v4:
> > > 1. Lowercase hex, use spaces for define indentation
> > > 2. _dpu_crtc_setup_blend_cfg(): pass mdss_ver instead of ctl
> > > 
> > > Changes in v3:
> > > 1. New patch, split from previous big DPU v12.0.
> > > ---
> > >  drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c  | 19 ++++---
> > >  drivers/gpu/drm/msm/disp/dpu1/dpu_hw_lm.c | 84 +++++++++++++++++++++++++++++--
> > >  2 files changed, 94 insertions(+), 9 deletions(-)
> > > 
> > > diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
> > > index a4b0fe0d9899b32141928f0b6a16503a49b3c27a..90f47fc15ee5708795701d78a1380f4ab01c1427 100644
> > > --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
> > > +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
> > > @@ -320,14 +320,20 @@ static bool dpu_crtc_get_scanout_position(struct drm_crtc *crtc,
> > >  }
> > >  
> > >  static void _dpu_crtc_setup_blend_cfg(struct dpu_crtc_mixer *mixer,
> > > -		struct dpu_plane_state *pstate, const struct msm_format *format)
> > > +				      struct dpu_plane_state *pstate,
> > > +				      const struct msm_format *format,
> > > +				      const struct dpu_mdss_version *mdss_ver)
> > >  {
> > >  	struct dpu_hw_mixer *lm = mixer->hw_lm;
> > >  	uint32_t blend_op;
> > > -	uint32_t fg_alpha, bg_alpha;
> > > +	uint32_t fg_alpha, bg_alpha, max_alpha;
> > >  
> > >  	fg_alpha = pstate->base.alpha >> 8;
> > 
> > For the 10-bit alpha, you need to shift here by 5 instead of 8.
> 
> Typo. "6 instead of 8".

Granted there would be a next iteration of this patch, I'd suggest to
modify _dpu_crtc_setup_blend_cfg() to always use 16-bit values and pass
them down to LM's setup_blend_config() callback. Then LM can perform
version-specific shifts, utilizing either 8 bits or 10 bits of alpha
channel values.

-- 
With best wishes
Dmitry