[Intel-gfx] [PATCH] drm/i915/skl+: Enable gamma and CSC on bottom color.

[Daniel Vetter]

On Tue, Oct 20, 2015 at 11:16:20AM -0700, Matt Roper wrote:
> On Tue, Oct 20, 2015 at 09:04:56AM -0700, Bob Paauwe wrote:
> > To stay consisent with how we're programming all the other planes,
> > enable gamma and CSC on the bottom color.  Without this, we fail the
> > the kms_universal_plane functional tests because the black primary plane
> > is brighter (gamma corrected) than the disabled plane case.  If the bottom
> > color is also gamma/csc corrected, then the disiabled case will match the
> > black plane case.
> > 
> > testcase: igt/kms_universal_plane/universal-plane-pipe-[ABC]-functional
> > CC: Konduru, Chandra <chandra.konduru at intel.com>
> > cc: Kevin Strasser <kevin.strasser at linux.intel.com>
> > Signed-off-by: Bob Paauwe <bob.j.paauwe at intel.com>
> > ---
> >  drivers/gpu/drm/i915/i915_reg.h      | 10 ++++++++++
> >  drivers/gpu/drm/i915/intel_display.c |  7 +++++++
> >  2 files changed, 17 insertions(+)
> > 
> > diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
> > index 724f057..9ad330b 100644
> > --- a/drivers/gpu/drm/i915/i915_reg.h
> > +++ b/drivers/gpu/drm/i915/i915_reg.h
> > @@ -4479,6 +4479,16 @@ enum skl_disp_power_wells {
> >  #define   PIPEMISC_DITHER_TYPE_SP	(0<<2)
> >  #define PIPEMISC(pipe) _PIPE2(pipe, _PIPE_MISC_A)
> >  
> > +#define _PIPE_BOTTOM_COLOR_A           0x70034
> > +#define _PIPE_BOTTOM_COLOR_B           0x71034
> > +#define _PIPE_BOTTOM_COLOR_C           0x72034
> > +#define   PIPE_BOTTOM_GAMMA_ENABLE     (1<<31)
> > +#define   PIPE_BOTTOM_CSC_ENABLE       (1<<30)
> > +#define   PIPE_BOTTOM_COLOR_MASK       0x3FFFFFFF
> > +#define PIPE_BOTTOM_COLOR(pipe) _PIPE3(pipe, _PIPE_BOTTOM_COLOR_A, \
> > +                                      _PIPE_BOTTOM_COLOR_B, \
> > +                                      _PIPE_BOTTOM_COLOR_C)
> > +
> >  #define VLV_DPFLIPSTAT				(VLV_DISPLAY_BASE + 0x70028)
> >  #define   PIPEB_LINE_COMPARE_INT_EN		(1<<29)
> >  #define   PIPEB_HLINE_INT_EN			(1<<28)
> > diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
> > index 3b4bacf..7665730 100644
> > --- a/drivers/gpu/drm/i915/intel_display.c
> > +++ b/drivers/gpu/drm/i915/intel_display.c
> > @@ -4949,6 +4949,7 @@ static void haswell_crtc_enable(struct drm_crtc *crtc)
> >  	struct intel_crtc_state *pipe_config =
> >  		to_intel_crtc_state(crtc->state);
> >  	bool is_dsi = intel_pipe_has_type(intel_crtc, INTEL_OUTPUT_DSI);
> > +	u32 bottom;
> >  
> >  	if (WARN_ON(intel_crtc->active))
> >  		return;
> > @@ -5033,6 +5034,12 @@ static void haswell_crtc_enable(struct drm_crtc *crtc)
> >  		intel_wait_for_vblank(dev, hsw_workaround_pipe);
> >  		intel_wait_for_vblank(dev, hsw_workaround_pipe);
> >  	}
> > +
> > +	if (INTEL_INFO(dev)->gen >= 9) {
> > +		bottom = I915_READ(PIPE_BOTTOM_COLOR(pipe));
> 
> Do we really want to do the r/m/w of the current color (meaning that we
> could get something that isn't even black if our boot firmware messed
> with the register), or should we just set it to explicit black with the
> CSC/gamma bits turned on?  Eventually that hardcoded black would change
> once we land properties to allow userspace to change the color, but
> setting it explicitly to black today seems reasonable to me.

rmw considered harmful (mostly because it just hides bugs which then show
up when we cleanse all registers with runtime PM while all outputs are
dpms off). Just write black.

> Anyway, this does solve the i-g-t failures for me, so
> 
> Tested-by(BXT): Matt Roper <matthew.d.roper at intel.com>

As said in the other thread, I think an overall series with these fixes
for all recent-ish platforms would be great. Plus confirmation from hw
engineers.
-Daniel

> 
> 
> Matt
> 
> > +		bottom |= (PIPE_BOTTOM_CSC_ENABLE | PIPE_BOTTOM_GAMMA_ENABLE);
> > +		I915_WRITE(PIPE_BOTTOM_COLOR(pipe), bottom);
> > +	}
> >  }
> >  
> >  static void ironlake_pfit_disable(struct intel_crtc *crtc, bool force)
> > -- 
> > 2.4.3
> > 
> > _______________________________________________
> > Intel-gfx mailing list
> > Intel-gfx at lists.freedesktop.org
> > http://lists.freedesktop.org/mailman/listinfo/intel-gfx
> 
> -- 
> Matt Roper
> Graphics Software Engineer
> IoTG Platform Enabling & Development
> Intel Corporation
> (916) 356-2795
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch