Re: [v2][PATCH 2/3] drm/i915/display: Extract icl_read_luts()

[Jani Nikula <jani.nikula@intel.com>]

On Tue, 17 Sep 2019, Swati Sharma <swati2.sharma@intel.com> wrote:
> For icl+, have hw read out to create hw blob of gamma
> lut values. icl+ platforms supports multi segmented gamma
> mode, add hw lut creation for this mode.
>
> This will be used to validate gamma programming using dsb
> (display state buffer) which is a tgl feature.
>
> v2: -readout code for multisegmented gamma has to come
>      up with some intermediate entries that aren't preserved
>      in hardware (Jani N)
>     -linear interpolation (Ville)
>     -moved common code to check gamma_enable to specific funcs,
>      since icl doesn't support that
>
> Signed-off-by: Swati Sharma <swati2.sharma@intel.com>
> ---
>  drivers/gpu/drm/i915/display/intel_color.c | 243 ++++++++++++++++++++++++++---
>  drivers/gpu/drm/i915/i915_reg.h            |   7 +
>  2 files changed, 230 insertions(+), 20 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/display/intel_color.c b/drivers/gpu/drm/i915/display/intel_color.c
> index b1f0f7e..0008011 100644
> --- a/drivers/gpu/drm/i915/display/intel_color.c
> +++ b/drivers/gpu/drm/i915/display/intel_color.c
> @@ -1370,6 +1370,9 @@ static int icl_color_check(struct intel_crtc_state *crtc_state)
>  
>  static int i9xx_gamma_precision(const struct intel_crtc_state *crtc_state)
>  {
> +	if (!crtc_state->gamma_enable)
> +		return 0;
> +

Why are you moving these checks back to the individual functions?

>  	switch (crtc_state->gamma_mode) {
>  	case GAMMA_MODE_MODE_8BIT:
>  		return 8;
> @@ -1383,6 +1386,9 @@ static int i9xx_gamma_precision(const struct intel_crtc_state *crtc_state)
>  
>  static int ilk_gamma_precision(const struct intel_crtc_state *crtc_state)
>  {
> +	if (!crtc_state->gamma_enable)
> +		return 0;
> +
>  	if ((crtc_state->csc_mode & CSC_POSITION_BEFORE_GAMMA) == 0)
>  		return 0;
>  
> @@ -1399,6 +1405,9 @@ static int ilk_gamma_precision(const struct intel_crtc_state *crtc_state)
>  
>  static int chv_gamma_precision(const struct intel_crtc_state *crtc_state)
>  {
> +	if (!crtc_state->gamma_enable)
> +		return 0;
> +
>  	if (crtc_state->cgm_mode & CGM_PIPE_MODE_GAMMA)
>  		return 10;
>  	else
> @@ -1407,6 +1416,9 @@ static int chv_gamma_precision(const struct intel_crtc_state *crtc_state)
>  
>  static int glk_gamma_precision(const struct intel_crtc_state *crtc_state)
>  {
> +	if (!crtc_state->gamma_enable)
> +		return 0;
> +
>  	switch (crtc_state->gamma_mode) {
>  	case GAMMA_MODE_MODE_8BIT:
>  		return 8;
> @@ -1418,21 +1430,39 @@ static int glk_gamma_precision(const struct intel_crtc_state *crtc_state)
>  	}
>  }
>  
> +static int icl_gamma_precision(const struct intel_crtc_state *crtc_state)
> +{
> +	if ((crtc_state->gamma_mode & POST_CSC_GAMMA_ENABLE) == 0)
> +		return 0;
> +
> +	switch (crtc_state->gamma_mode & GAMMA_MODE_MODE_MASK) {
> +	case GAMMA_MODE_MODE_8BIT:
> +		return 8;
> +	case GAMMA_MODE_MODE_10BIT:
> +		return 10;
> +	case GAMMA_MODE_MODE_12BIT_MULTI_SEGMENTED:
> +		return 16;  
> +	default:
> +		MISSING_CASE(crtc_state->gamma_mode);
> +		return 0;
> +	}
> +
> +}
> +
>  int intel_color_get_gamma_bit_precision(const struct intel_crtc_state *crtc_state)
>  {
>  	struct intel_crtc *crtc = to_intel_crtc(crtc_state->base.crtc);
>  	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
>  
> -	if (!crtc_state->gamma_enable)
> -		return 0;
> -

Why?

>  	if (HAS_GMCH(dev_priv)) {
>  		if (IS_CHERRYVIEW(dev_priv))
>  			return chv_gamma_precision(crtc_state);
>  		else
>  			return i9xx_gamma_precision(crtc_state);
>  	} else {
> -		if (IS_CANNONLAKE(dev_priv) || IS_GEMINILAKE(dev_priv))
> +		if (INTEL_GEN(dev_priv) >= 11)
> +			return icl_gamma_precision(crtc_state);
> +		else if (IS_CANNONLAKE(dev_priv) || IS_GEMINILAKE(dev_priv))
>  			return glk_gamma_precision(crtc_state);
>  		else if (IS_IRONLAKE(dev_priv))
>  			return ilk_gamma_precision(crtc_state);
> @@ -1463,6 +1493,30 @@ static bool intel_color_lut_entry_equal(struct drm_color_lut *lut1,
>  	return true;
>  }
>  
> +static bool intel_color_lut_entry_multi_equal(struct drm_color_lut *lut1,
> +					      struct drm_color_lut *lut2,
> +					      int lut_size, u32 err)
> +{
> +	int i;
> +
> +	for (i = 0; i < 9; i++) {
> +		if (!err_check(&lut1[i], &lut2[i], err))
> +			return false;
> +	}
> +
> +	for (i = 1; i <  257; i++) {
                       ^
                       extra space

> +		if (!err_check(&lut1[i * 8], &lut2[i * 8], err))
> +			return false;
> +	}

i == 8 will be checked twice.

> +
> +	for (i = 0; i < 256; i++) {
> +		if (!err_check(&lut1[i * 8 * 128], &lut2[i * 8 * 128], err))
> +			return false;
> +	}

i == 0 will be checked twice.

I note that these indices match the programming part, so maybe better to
keep them as they are here. No harm done I guess.

> +
> +	return true;
> +}
> +
>  bool intel_color_lut_equal(struct drm_property_blob *blob1,
>  			   struct drm_property_blob *blob2,
>  			   u32 gamma_mode, u32 bit_precision)
> @@ -1481,16 +1535,8 @@ bool intel_color_lut_equal(struct drm_property_blob *blob1,
>  	lut_size2 = drm_color_lut_size(blob2);
>  
>  	/* check sw and hw lut size */
> -	switch (gamma_mode) {
> -	case GAMMA_MODE_MODE_8BIT:
> -	case GAMMA_MODE_MODE_10BIT:
> -		if (lut_size1 != lut_size2)
> -			return false;
> -		break;
> -	default:
> -		MISSING_CASE(gamma_mode);
> -			return false;
> -	}
> +	if (lut_size1 != lut_size2)
> +		return false;
>  
>  	lut1 = blob1->data;
>  	lut2 = blob2->data;
> @@ -1498,13 +1544,18 @@ bool intel_color_lut_equal(struct drm_property_blob *blob1,
>  	err = 0xffff >> bit_precision;
>  
>  	/* check sw and hw lut entry to be equal */
> -	switch (gamma_mode) {
> +	switch (gamma_mode & GAMMA_MODE_MODE_MASK) {
>  	case GAMMA_MODE_MODE_8BIT:
>  	case GAMMA_MODE_MODE_10BIT:
>  		if (!intel_color_lut_entry_equal(lut1, lut2,
>  						 lut_size2, err))
>  			return false;
>  		break;
> +	case GAMMA_MODE_MODE_12BIT_MULTI_SEGMENTED:
> +		if (!intel_color_lut_entry_multi_equal(lut1, lut2,
> +						       lut_size2, err))
> +			return false;
> +		break;
>  	default:
>  		MISSING_CASE(gamma_mode);
>  			return false;
> @@ -1744,6 +1795,157 @@ static void glk_read_luts(struct intel_crtc_state *crtc_state)
>  		crtc_state->base.gamma_lut = glk_read_lut_10(crtc_state, PAL_PREC_INDEX_VALUE(0));
>  }
>  
> +static struct drm_color_lut *
> +icl_compute_interpolated_gamma_blob(struct drm_color_lut *tmp_lut,
> +				    struct drm_color_lut *lut, u32 lut_size)
> +{

I think you should just pass in the the actual lut, and not use a temp
lut at all. See my comments below for icl_read_lut_multi_seg() function.

> +	__u16 a_red, b_red, a_green, b_green, a_blue, b_blue;
> +	__u16 red_step, green_step, blue_step;

u16, not __u16.

> +	int i, j, k, m, n;
> +
> +	for (i = 0, k = 0; k < 9; i++, k++) {
> +		lut[i].red = tmp_lut[k].red;
> +		lut[i].green = tmp_lut[k].green;
> +		lut[i].blue = tmp_lut[k].blue;
> +	}

With a single lut, you can skip this.

> +
> +	for (k = 9; k < 264; k++) {
> +		a_red = tmp_lut[k].red;
> +		b_red = tmp_lut[k + 1].red;
> +		red_step = (b_red - a_red) / 8;
> +
> +		a_green = tmp_lut[k].green;
> +		b_green = tmp_lut[k + 1].green;
> +		green_step = (b_green - a_green) / 8;
> +
> +		a_blue = tmp_lut[k].blue;
> +		b_blue = tmp_lut[k + 1].blue;
> +		blue_step = (b_blue - a_blue) / 8;
> +
> +		for (j = 0; j < 8; j++) {
> +			lut[i].red = lut[i - 1].red + red_step;
> +			lut[i].green = lut[i - 1].green + green_step;
> +			lut[i].blue = lut[i - 1].blue + blue_step;
> +
> +			i++;
> +		}
> +	}

This would be written in a way to only cover the values that need to be
interpolated.

	for (i = 1; i < 257 - 1; i++) {
        	start = i * 8;
                end = (i + 1) * 8;
                steps = end - start;

                for (j = start + 1; j < end; j++) {
                }
        }

Fill in the gaps. For me I think this is easier to grasp and compare
against the readout and write. I find it very hard to check the ranges
in the loops.

> +
> +	for (k = 265; k < 521; k++) {
> +		a_red = tmp_lut[k].red;
> +		b_red = tmp_lut[k + 1].red;
> +		red_step = ((b_red - a_red) / 127) / 8;
> +
> +		a_green = tmp_lut[k].green;
> +		b_green = tmp_lut[k + 1].green;
> +		green_step = ((b_green - a_green) / 127) / 8;
> +
> +		a_blue = tmp_lut[k].blue;
> +		b_blue = tmp_lut[k + 1].blue;
> +		blue_step = ((b_blue - a_blue) / 127) / 8;
> +
> +		for (m = 0; m < 127; m++) {
> +			for (n = 0; n < 8; n++) {
> +				lut[i].red = lut[i - 1].red + red_step;
> +				lut[i].green = lut[i - 1].green + green_step;
> +				lut[i].blue = lut[i - 1].blue + blue_step;
> +
> +				i++;
> +			}
> +		}
> +	}

Similarly:

	for (i = 0; i < 256 - 1; i++) {
        	start = i * 8 * 128;
                end = (i + 1) * 8 * 128;
                steps = end - start;

                for (j = start + 1; j < end; j++) {
                }
        }


> +
> +	return lut;
> +}
> +
> +static struct drm_property_blob *
> +icl_read_lut_multi_seg(const struct intel_crtc_state *crtc_state)
> +{
> +	struct intel_crtc *crtc = to_intel_crtc(crtc_state->base.crtc);
> +	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
> +	int lut_size = INTEL_INFO(dev_priv)->color.gamma_lut_size;
> +	int tmp_lut_size = 522;
> +	enum pipe pipe = crtc->pipe;
> +	struct drm_property_blob *blob, *tmp_blob;
> +	struct drm_color_lut *blob_data, *tmp_blob_data;
> +	u32 i, val1, val2;
> +
> +	blob = drm_property_create_blob(&dev_priv->drm,
> +					sizeof(struct drm_color_lut) * lut_size,
> +					NULL);
> +	tmp_blob = drm_property_create_blob(&dev_priv->drm,
> +					    sizeof(struct drm_color_lut) * tmp_lut_size,
> +					    NULL);

You end up leaking the temporary blob. But I don't think you really need
the temporary blob at all. Read on.

> +	if (IS_ERR(blob) || IS_ERR(tmp_blob))
> +		return NULL;
> +
> +	blob_data = blob->data;
> +	tmp_blob_data = tmp_blob->data;
> +
> +	I915_WRITE(PREC_PAL_MULTI_SEG_INDEX(pipe), PAL_PREC_AUTO_INCREMENT);
> +
> +	for (i = 0; i < 9; i++) {
> +		val1 = I915_READ(PREC_PAL_MULTI_SEG_DATA(pipe));
> +		val2 = I915_READ(PREC_PAL_MULTI_SEG_DATA(pipe));
> +
> +		tmp_blob_data[i].red = REG_FIELD_GET(PAL_PREC_MULTI_SEG_RED_UDW_MASK, val2) << 6 |
> +				       REG_FIELD_GET(PAL_PREC_MULTI_SEG_RED_LDW_MASK, val1);
> +		tmp_blob_data[i].green = REG_FIELD_GET(PAL_PREC_MULTI_SEG_GREEN_UDW_MASK, val2) << 6 |
> +					 REG_FIELD_GET(PAL_PREC_MULTI_SEG_GREEN_LDW_MASK, val1);
> +		tmp_blob_data[i].blue = REG_FIELD_GET(PAL_PREC_MULTI_SEG_BLUE_UDW_MASK, val2) << 6 |
> +					REG_FIELD_GET(PAL_PREC_MULTI_SEG_BLUE_LDW_MASK, val1);
> +	}
> +
> +	I915_WRITE(PREC_PAL_INDEX(pipe), PAL_PREC_AUTO_INCREMENT);
> +
> +	for (i = 1; i < 257; i++) {
> +		val1 = I915_READ(PREC_PAL_DATA(pipe));
> +		val2 = I915_READ(PREC_PAL_DATA(pipe));
> +
> +		tmp_blob_data[i + 8].red = REG_FIELD_GET(PAL_PREC_MULTI_SEG_RED_UDW_MASK, val2) << 6 |
> +					   REG_FIELD_GET(PAL_PREC_MULTI_SEG_RED_LDW_MASK, val1);
> +		tmp_blob_data[i + 8].green = REG_FIELD_GET(PAL_PREC_MULTI_SEG_GREEN_UDW_MASK, val2) << 6 |
> +					     REG_FIELD_GET(PAL_PREC_MULTI_SEG_GREEN_LDW_MASK, val1);
> +		tmp_blob_data[i + 8].blue = REG_FIELD_GET(PAL_PREC_MULTI_SEG_BLUE_UDW_MASK, val2) << 6 |
> +					    REG_FIELD_GET(PAL_PREC_MULTI_SEG_BLUE_LDW_MASK, val1);
> +	}
> +
> +	for (i = 0; i < 256; i++) {
> +		val1 = I915_READ(PREC_PAL_DATA(pipe));
> +		val2 = I915_READ(PREC_PAL_DATA(pipe));
> +
> +		tmp_blob_data[i + 265].red = REG_FIELD_GET(PAL_PREC_MULTI_SEG_RED_UDW_MASK, val2) << 6 |
> +					     REG_FIELD_GET(PAL_PREC_MULTI_SEG_RED_LDW_MASK, val1);
> +		tmp_blob_data[i + 265].green = REG_FIELD_GET(PAL_PREC_MULTI_SEG_GREEN_UDW_MASK, val2) << 6 |
> +					       REG_FIELD_GET(PAL_PREC_MULTI_SEG_GREEN_LDW_MASK, val1);
> +		tmp_blob_data[i + 265].blue = REG_FIELD_GET(PAL_PREC_MULTI_SEG_BLUE_UDW_MASK, val2) << 6 |
> +					      REG_FIELD_GET(PAL_PREC_MULTI_SEG_BLUE_LDW_MASK, val1);
> +	}
> +
> +	tmp_blob_data[521].red = REG_FIELD_GET(PREC_PAL_GC_MAX_RGB_MASK,
> +					       I915_READ(PREC_PAL_GC_MAX(pipe, 0)));
> +	tmp_blob_data[521].green = REG_FIELD_GET(PREC_PAL_GC_MAX_RGB_MASK,
> +						 I915_READ(PREC_PAL_GC_MAX(pipe, 1)));
> +	tmp_blob_data[521].blue = REG_FIELD_GET(PREC_PAL_GC_MAX_RGB_MASK,
> +						I915_READ(PREC_PAL_GC_MAX(pipe, 1)));

In the above, I think you should just read the values directly to the
right locations in the actual blob. Then all the indices match the
programming side, and it's easier to review.

> +
> +	blob_data = icl_compute_interpolated_gamma_blob(tmp_blob_data, blob_data, lut_size);

And then you fill in the gaps in the interpolation function.

> +
> +	return blob;
> +}
> +
> +static void icl_read_luts(struct intel_crtc_state *crtc_state)
> +{
> +	if ((crtc_state->gamma_mode & GAMMA_MODE_MODE_MASK) ==
> +	    GAMMA_MODE_MODE_8BIT)
> +		crtc_state->base.gamma_lut = i9xx_read_lut_8(crtc_state);
> +	else if ((crtc_state->gamma_mode & GAMMA_MODE_MODE_MASK) ==
> +		 GAMMA_MODE_MODE_12BIT_MULTI_SEGMENTED)
> +		crtc_state->base.gamma_lut = icl_read_lut_multi_seg(crtc_state);
> +	else
> +		crtc_state->base.gamma_lut = glk_read_lut_10(crtc_state, PAL_PREC_INDEX_VALUE(0));
> +}
> +
>  void intel_color_init(struct intel_crtc *crtc)
>  {
>  	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
> @@ -1785,16 +1987,17 @@ void intel_color_init(struct intel_crtc *crtc)
>  		else
>  			dev_priv->display.color_commit = ilk_color_commit;
>  
> -		if (INTEL_GEN(dev_priv) >= 11)
> +		if (INTEL_GEN(dev_priv) >= 11) {
>  			dev_priv->display.load_luts = icl_load_luts;
> -		else if (IS_CANNONLAKE(dev_priv) || IS_GEMINILAKE(dev_priv)) {
> +			dev_priv->display.read_luts = icl_read_luts;
> +		} else if (IS_CANNONLAKE(dev_priv) || IS_GEMINILAKE(dev_priv)) {
>  			dev_priv->display.load_luts = glk_load_luts;
>  			dev_priv->display.read_luts = glk_read_luts;
> -		} else if (INTEL_GEN(dev_priv) >= 8)
> +		} else if (INTEL_GEN(dev_priv) >= 8) {
>  			dev_priv->display.load_luts = bdw_load_luts;
> -		else if (INTEL_GEN(dev_priv) >= 7)
> +		} else if (INTEL_GEN(dev_priv) >= 7) {
>  			dev_priv->display.load_luts = ivb_load_luts;
> -		else {
> +		} else {

Shouldn't the cleanup be part of patch 1?

>  			dev_priv->display.load_luts = ilk_load_luts;
>  			dev_priv->display.read_luts = ilk_read_luts;
>  		}
> diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
> index bf37ece..844dd62 100644
> --- a/drivers/gpu/drm/i915/i915_reg.h
> +++ b/drivers/gpu/drm/i915/i915_reg.h
> @@ -10378,6 +10378,7 @@ enum skl_power_gate {
>  
>  #define PREC_PAL_INDEX(pipe)		_MMIO_PIPE(pipe, _PAL_PREC_INDEX_A, _PAL_PREC_INDEX_B)
>  #define PREC_PAL_DATA(pipe)		_MMIO_PIPE(pipe, _PAL_PREC_DATA_A, _PAL_PREC_DATA_B)
> +#define PREC_PAL_GC_MAX_RGB_MASK	 REG_GENMASK(15, 0)
>  #define PREC_PAL_GC_MAX(pipe, i)	_MMIO(_PIPE(pipe, _PAL_PREC_GC_MAX_A, _PAL_PREC_GC_MAX_B) + (i) * 4)
>  #define PREC_PAL_EXT_GC_MAX(pipe, i)	_MMIO(_PIPE(pipe, _PAL_PREC_EXT_GC_MAX_A, _PAL_PREC_EXT_GC_MAX_B) + (i) * 4)
>  #define PREC_PAL_EXT2_GC_MAX(pipe, i)	_MMIO(_PIPE(pipe, _PAL_PREC_EXT2_GC_MAX_A, _PAL_PREC_EXT2_GC_MAX_B) + (i) * 4)
> @@ -10401,6 +10402,12 @@ enum skl_power_gate {
>  
>  #define _PAL_PREC_MULTI_SEG_DATA_A	0x4A40C
>  #define _PAL_PREC_MULTI_SEG_DATA_B	0x4AC0C
> +#define  PAL_PREC_MULTI_SEG_RED_LDW_MASK   REG_GENMASK(29, 24)
> +#define  PAL_PREC_MULTI_SEG_RED_UDW_MASK   REG_GENMASK(29, 20)
> +#define  PAL_PREC_MULTI_SEG_GREEN_LDW_MASK REG_GENMASK(19, 14)
> +#define  PAL_PREC_MULTI_SEG_GREEN_UDW_MASK REG_GENMASK(19, 10)
> +#define  PAL_PREC_MULTI_SEG_BLUE_LDW_MASK  REG_GENMASK(9, 4)
> +#define  PAL_PREC_MULTI_SEG_BLUE_UDW_MASK  REG_GENMASK(9, 0)
>  
>  #define PREC_PAL_MULTI_SEG_INDEX(pipe)	_MMIO_PIPE(pipe, \
>  					_PAL_PREC_MULTI_SEG_INDEX_A, \

-- 
Jani Nikula, Intel Open Source Graphics Center
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx