[Intel-gfx] [v5 2/2] drm/i915/icl: Enable Plane Input CSC for YUV to RGB Conversion

Matt Roper matthew.d.roper at intel.com
Wed Oct 31 16:40:47 UTC 2018


On Wed, Oct 31, 2018 at 05:34:19AM -0700, Shankar, Uma wrote:
> 
> 
> >-----Original Message-----
> >From: Roper, Matthew D
> >Sent: Tuesday, October 30, 2018 4:59 AM
> >To: Shankar, Uma <uma.shankar at intel.com>
> >Cc: intel-gfx at lists.freedesktop.org; Syrjala, Ville <ville.syrjala at intel.com>;
> >Lankhorst, Maarten <maarten.lankhorst at intel.com>
> >Subject: Re: [Intel-gfx] [v5 2/2] drm/i915/icl: Enable Plane Input CSC for YUV to
> >RGB Conversion
> >
> >On Fri, Oct 26, 2018 at 03:31:57PM +0530, Uma Shankar wrote:
> >> Plane input CSC needs to be enabled to convert frambuffers from YUV to
> >> RGB. This is needed for bottom 3 planes on ICL, rest of the planes
> >> have hardcoded conversion and taken care by the legacy code.
> >>
> >> This patch defines the co-efficient values for YUV to RGB conversion
> >> in BT709 and BT601 formats. It programs the coefficients and enables
> >> the plane input csc unit in hardware.
> >>
> >> Note: This is currently untested and floated to get an early feedback
> >> on the design and implementation for this feature. In parallel, I will
> >> test this on actual ICL hardware and confirm with planar formats.
> >>
> >> v2: Addressed Maarten's and Ville's review comments and added the
> >> coefficients in a 2D array instead of independent Macros.
> >>
> >> v3: Added individual coefficient matrix (9 values) instead of 6
> >> register values as per Maarten's comment. Also addresed a shift issue
> >> with B channel coefficient.
> >>
> >> v4: Added support for Limited Range Color Handling
> >>
> >> v5: Fixed Matt and Maarten's review comments.
> >>
> >> Signed-off-by: Uma Shankar <uma.shankar at intel.com>
> >> ---
> >>  drivers/gpu/drm/i915/intel_color.c   | 79
> >++++++++++++++++++++++++++++++++++++
> >>  drivers/gpu/drm/i915/intel_display.c | 23 ++++++++---
> >>  drivers/gpu/drm/i915/intel_drv.h     |  2 +
> >>  3 files changed, 98 insertions(+), 6 deletions(-)
> >>
> >> diff --git a/drivers/gpu/drm/i915/intel_color.c
> >> b/drivers/gpu/drm/i915/intel_color.c
> >> index 5127da2..681cd13 100644
> >> --- a/drivers/gpu/drm/i915/intel_color.c
> >> +++ b/drivers/gpu/drm/i915/intel_color.c
> >> @@ -57,6 +57,15 @@
> >>  #define CSC_RGB_TO_YUV_RV_GV 0xbce89ad8  #define
> >CSC_RGB_TO_YUV_BV
> >> 0x1e080000
> >>
> >> +#define  ROFF(x)          (((x) & 0xffff) << 16)
> >> +#define  GOFF(x)          (((x) & 0xffff) << 0)
> >> +#define  BOFF(x)          (((x) & 0xffff) << 16)
> >> +
> >> +/* Preoffset values for YUV to RGB Conversion */
> >> +#define PREOFF_YUV_TO_RGB_HI		0x1800
> >> +#define PREOFF_YUV_TO_RGB_ME		0x1F00
> >> +#define PREOFF_YUV_TO_RGB_LO		0x1800
> >> +
> >>  /*
> >>   * Extract the CSC coefficient from a CTM coefficient (in U32.32 fixed point
> >>   * format). This macro takes the coefficient we want transformed and
> >> the @@ -643,6 +652,76 @@ int intel_color_check(struct drm_crtc *crtc,
> >>  	return -EINVAL;
> >>  }
> >>
> >> +void icl_program_input_csc_coeff(const struct intel_crtc_state *crtc_state,
> >> +				 const struct intel_plane_state *plane_state) {
> >> +	struct drm_i915_private *dev_priv =
> >> +		to_i915(plane_state->base.plane->dev);
> >> +	struct intel_crtc *crtc = to_intel_crtc(crtc_state->base.crtc);
> >> +	enum pipe pipe = crtc->pipe;
> >> +	struct intel_plane *intel_plane =
> >> +		to_intel_plane(plane_state->base.plane);
> >> +	enum plane_id plane = intel_plane->id;
> >> +
> >> +	static const u16 input_csc_matrix[][9] = {
> >
> >Can you add comments to these indicating the human-readable values they
> >translate to?
> 
> Sure Matt, will add that.
> 
> >> +		/* BT.601 full range YCbCr -> full range RGB */
> >> +		[DRM_COLOR_YCBCR_BT601] = {
> >> +			0x7AF8, 0x7800, 0x0,
> >> +			0x8B28, 0x7800, 0x9AC0,
> >> +			0x0, 0x7800, 0x7DD8,
> >> +		},
> >> +		/* BT.709 full range YCbCr -> full range RGB */
> >> +		[DRM_COLOR_YCBCR_BT709] = {
> >> +			0x7C98, 0x7800, 0x0,
> >> +			0x9EF8, 0x7800, 0xABF8,
> >> +			0x0, 0x7800,  0x7ED8,
> >> +		},
> >> +	};
> >> +
> >> +	/* Matrix for Limited Range to Full Range Conversion */
> >> +	static const u16 input_csc_matrix_lr[][9] = {
> >> +		/* BT.601 Limted range YCbCr -> full range RGB */
> >> +		[DRM_COLOR_YCBCR_BT601] = {
> >> +			0x7CC8, 0x7950, 0x0,
> >> +			0x8CB8, 0x7918, 0x9C40,
> >> +			0x0, 0x7918, 0x7FC8,
> >
> >Are these obtained by scaling the first row (Y-based) by 256/219 and the other
> >two rows (Cb and Cr) by 256/224?  If so, it looks like you've always rounded
> >down, whereas in some cases rounding up gives you a closer value (and matches
> >how the bspec seems to have chosen the full range encodings for their example).
> 
> Yes, this is how it's done. But the only reason of delta is that you have taken 8 bit max
> value as 256, but I am taking it as 255. I feel it should be (1 << bpc ie 8) - 1. If you put that, 
> the values which you got will match to what I have here. Please let me know if that's ok.

Ah, okay, that makes sense.  I do get the same values as you for the
BT.601 table that way, but I still get a slight rounding difference in
one cell of the BT.709 matrix down below.

> 
> And thanks for detailed review and useful pointers.
> 
> Regards,
> Uma Shankar
> 
> >        [ 0x7CD0, 0x7958,    0x0 ]
> >        [ 0x8CC0, 0x7928, 0x9C48 ]
> >        [    0x0, 0x7928, 0x7FD8 ]
> >
> >Our encodings of the 1.0 value on the second two rows seems to deviate slightly
> >more for some reason; not sure why that is.
> >
> >For completeness, here's how I came up with 0x7928:
> >
> >        1 * 256/224 = 1.142857143
> >        Sign bit = 0
> >        Exponent bits = 0b111
> >        Mantissa bits = round(1.142857143 << 8)
> >                      = round(292.571428571)
> >                      = 293
> >                      = 0b100100101
> >        Reserved bits = 0b000
> >
> >        Result = 0111 1001 0010 1000
> >               = 0x7928
> >
> >If you did floor() instead of round() for the mantissa, you'd get 292, which would
> >translate to 0x7920 instead.
> >
> >> +		},
> >> +		/* BT.709 Limited range YCbCr -> full range RGB */
> >> +		[DRM_COLOR_YCBCR_BT709] = {
> >> +			0x7EA8, 0x7950, 0x0,
> >> +			0x8888, 0x7918, 0xADA8,
> >> +			0x0, 0x7918,  0x6870,

For BU I get 0xADA0 instead of 0xADA8.

        -.187 * 255/224 = -0.212879464
        Sign bit = 1
        Exponent bits = 0b010
        Mantissa bits = round(0.212879464 << (9+2))
                      = round(435.977142272)
                      = 436
                      = 0b110110100
        Reserved bits = 0b000
        Result = 1010 1101 1010 0000
               = 0xADA0

All of the other values match now though.


Matt


> >
> >For these I get
> >
> >        [ 0x7EB8, 0x7958,      0 ]
> >        [ 0x8890, 0x7928, 0xADB0 ]
> >        [    0x0, 0x7928, 0x6878 ]
> >
> >So all the numbers are still pretty close to what you have.
> >
> >
> >Matt
> >
> >> +		},
> >> +	};
> >> +	const u16 *csc;
> >> +
> >> +	if (plane_state->base.color_range ==
> >DRM_COLOR_YCBCR_FULL_RANGE)
> >> +		csc = input_csc_matrix[plane_state->base.color_encoding];
> >> +	else
> >> +		csc = input_csc_matrix_lr[plane_state->base.color_encoding];
> >> +
> >> +	I915_WRITE(PLANE_INPUT_CSC_COEFF(pipe, plane, 0), ROFF(csc[0]) |
> >> +		   GOFF(csc[1]));
> >> +	I915_WRITE(PLANE_INPUT_CSC_COEFF(pipe, plane, 1), BOFF(csc[2]));
> >> +	I915_WRITE(PLANE_INPUT_CSC_COEFF(pipe, plane, 2), ROFF(csc[3]) |
> >> +		   GOFF(csc[4]));
> >> +	I915_WRITE(PLANE_INPUT_CSC_COEFF(pipe, plane, 3), BOFF(csc[5]));
> >> +	I915_WRITE(PLANE_INPUT_CSC_COEFF(pipe, plane, 4), ROFF(csc[6]) |
> >> +		   GOFF(csc[7]));
> >> +	I915_WRITE(PLANE_INPUT_CSC_COEFF(pipe, plane, 5), BOFF(csc[8]));
> >> +
> >> +	I915_WRITE(PLANE_INPUT_CSC_PREOFF(pipe, plane, 0),
> >> +		   PREOFF_YUV_TO_RGB_HI);
> >> +	I915_WRITE(PLANE_INPUT_CSC_PREOFF(pipe, plane, 1),
> >> +		   PREOFF_YUV_TO_RGB_ME);
> >> +	I915_WRITE(PLANE_INPUT_CSC_PREOFF(pipe, plane, 2),
> >> +		   PREOFF_YUV_TO_RGB_LO);
> >> +
> >> +	I915_WRITE(PLANE_INPUT_CSC_POSTOFF(pipe, plane, 0), 0x0);
> >> +	I915_WRITE(PLANE_INPUT_CSC_POSTOFF(pipe, plane, 1), 0x0);
> >> +	I915_WRITE(PLANE_INPUT_CSC_POSTOFF(pipe, plane, 2), 0x0); }
> >> +
> >>  void intel_color_init(struct drm_crtc *crtc)  {
> >>  	struct drm_i915_private *dev_priv = to_i915(crtc->dev); diff --git
> >> a/drivers/gpu/drm/i915/intel_display.c
> >> b/drivers/gpu/drm/i915/intel_display.c
> >> index fe045ab..d16a064 100644
> >> --- a/drivers/gpu/drm/i915/intel_display.c
> >> +++ b/drivers/gpu/drm/i915/intel_display.c
> >> @@ -3666,6 +3666,7 @@ u32 glk_plane_color_ctl(const struct intel_crtc_state
> >*crtc_state,
> >>  	struct drm_i915_private *dev_priv =
> >>  		to_i915(plane_state->base.plane->dev);
> >>  	const struct drm_framebuffer *fb = plane_state->base.fb;
> >> +	struct intel_plane *plane = to_intel_plane(plane_state->base.plane);
> >>  	u32 plane_color_ctl = 0;
> >>
> >>  	if (INTEL_GEN(dev_priv) < 11) {
> >> @@ -3676,13 +3677,23 @@ u32 glk_plane_color_ctl(const struct
> >intel_crtc_state *crtc_state,
> >>  	plane_color_ctl |= glk_plane_color_ctl_alpha(plane_state);
> >>
> >>  	if (fb->format->is_yuv) {
> >> -		if (plane_state->base.color_encoding ==
> >DRM_COLOR_YCBCR_BT709)
> >> -			plane_color_ctl |=
> >PLANE_COLOR_CSC_MODE_YUV709_TO_RGB709;
> >> -		else
> >> -			plane_color_ctl |=
> >PLANE_COLOR_CSC_MODE_YUV601_TO_RGB709;
> >> +		if (!icl_is_hdr_plane(plane)) {
> >> +			if (plane_state->base.color_encoding ==
> >> +					DRM_COLOR_YCBCR_BT709)
> >> +				plane_color_ctl |=
> >> +
> >	PLANE_COLOR_CSC_MODE_YUV709_TO_RGB709;
> >> +			else
> >> +				plane_color_ctl |=
> >> +
> >	PLANE_COLOR_CSC_MODE_YUV601_TO_RGB709;
> >>
> >> -		if (plane_state->base.color_range ==
> >DRM_COLOR_YCBCR_FULL_RANGE)
> >> -			plane_color_ctl |=
> >PLANE_COLOR_YUV_RANGE_CORRECTION_DISABLE;
> >> +			if (plane_state->base.color_range ==
> >> +					DRM_COLOR_YCBCR_FULL_RANGE)
> >> +				plane_color_ctl |=
> >> +
> >	PLANE_COLOR_YUV_RANGE_CORRECTION_DISABLE;
> >> +		} else {
> >> +			icl_program_input_csc_coeff(crtc_state, plane_state);
> >> +			plane_color_ctl |= PLANE_COLOR_INPUT_CSC_ENABLE;
> >> +		}
> >>  	}
> >>
> >>  	return plane_color_ctl;
> >> diff --git a/drivers/gpu/drm/i915/intel_drv.h
> >> b/drivers/gpu/drm/i915/intel_drv.h
> >> index db24308..bd9e946 100644
> >> --- a/drivers/gpu/drm/i915/intel_drv.h
> >> +++ b/drivers/gpu/drm/i915/intel_drv.h
> >> @@ -2285,6 +2285,8 @@ int intel_plane_atomic_check_with_state(const
> >> struct intel_crtc_state *old_crtc_  int intel_color_check(struct
> >> drm_crtc *crtc, struct drm_crtc_state *state);  void
> >> intel_color_set_csc(struct drm_crtc_state *crtc_state);  void
> >> intel_color_load_luts(struct drm_crtc_state *crtc_state);
> >> +void icl_program_input_csc_coeff(const struct intel_crtc_state *crtc_state,
> >> +				 const struct intel_plane_state *plane_state);
> >>
> >>  /* intel_lspcon.c */
> >>  bool lspcon_init(struct intel_digital_port *intel_dig_port);
> >> --
> >> 1.9.1
> >>
> >> _______________________________________________
> >> Intel-gfx mailing list
> >> Intel-gfx at lists.freedesktop.org
> >> https://lists.freedesktop.org/mailman/listinfo/intel-gfx
> >
> >--
> >Matt Roper
> >Graphics Software Engineer
> >IoTG Platform Enabling & Development
> >Intel Corporation
> >(916) 356-2795

-- 
Matt Roper
Graphics Software Engineer
IoTG Platform Enabling & Development
Intel Corporation
(916) 356-2795


More information about the Intel-gfx mailing list