Re: [V7 34/45] drm/amd/display: add shaper and blend colorops for 1D Curve Custom LUT

[Harry Wentland <harry.wentland@amd.com>]

On 2025-02-12 18:44, Leo Li wrote:
> 
> 
> On 2024-12-19 23:33, Alex Hung wrote:
>> This patch adds colorops for custom 1D LUTs in the SHAPER and
>> BLND HW blocks.
>>
>> With this change the following IGT tests pass:
>> kms_colorop --run plane-XR30-XR30-srgb_inv_eotf_lut
>> kms_colorop --run plane-XR30-XR30-srgb_inv_eotf_lut-srgb_eotf_lut
>>
>> The color pipeline now consists of the following colorops:
>> 1. 1D curve colorop
>> 2. 1D curve colorop
>> 3. 1D LUT
>> 4. 1D curve colorop
>> 5. 1D LUT
>>
>> The 1D curve colorops support sRGB, BT2020, and PQ scaled to 125.0.
>>
>> Signed-off-by: Alex Hung <alex.hung@amd.com>
>> Signed-off-by: Harry Wentland <harry.wentland@amd.com>
>> ---
>> v7:
>>   - Initialize uint32_t blend_size = 0 by default (kernel test robot)
>>   - Modify state->size to colorop->lut_size
>>
>>   .../amd/display/amdgpu_dm/amdgpu_dm_color.c   | 166 ++++++++++--------
>>   .../amd/display/amdgpu_dm/amdgpu_dm_colorop.c |  32 ++++
>>   2 files changed, 120 insertions(+), 78 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c
>> index 1765402bc122..0bea52eede39 100644
>> --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c
>> +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c
>> @@ -1211,38 +1211,6 @@ __set_dm_plane_colorop_degamma(struct drm_plane_state *plane_state,
>>       return __set_colorop_in_tf_1d_curve(dc_plane_state, colorop_state);
>>   }
>>   -static int
>> -__set_colorop_in_shaper_1d_curve(struct dc_plane_state *dc_plane_state,
>> -               struct drm_colorop_state *colorop_state)
>> -{
>> -    struct dc_transfer_func *tf = &dc_plane_state->in_shaper_func;
>> -    struct drm_colorop *colorop = colorop_state->colorop;
>> -    struct drm_device *drm = colorop->dev;
>> -
>> -    if (colorop->type != DRM_COLOROP_1D_CURVE)
>> -        return -EINVAL;
>> -
>> -    if (!(BIT(colorop_state->curve_1d_type) & amdgpu_dm_supported_shaper_tfs))
>> -        return -EINVAL;
>> -
>> -    if (colorop_state->bypass) {
>> -        tf->type = TF_TYPE_BYPASS;
>> -        tf->tf = TRANSFER_FUNCTION_LINEAR;
>> -        return 0;
>> -    }
>> -
>> -    drm_dbg(drm, "Shaper colorop with ID: %d\n", colorop->base.id);
>> -
>> -    if (colorop->type == DRM_COLOROP_1D_CURVE) {
>> -        tf->type = TF_TYPE_DISTRIBUTED_POINTS;
>> -        tf->tf = amdgpu_colorop_tf_to_dc_tf(colorop_state->curve_1d_type);
>> -        tf->sdr_ref_white_level = SDR_WHITE_LEVEL_INIT_VALUE;
>> -        return __set_output_tf(tf, 0, 0, false);
>> -    }
>> -
>> -    return -EINVAL;
>> -}
>> -
>>   static int
>>   __set_dm_plane_colorop_shaper(struct drm_plane_state *plane_state,
>>                     struct dc_plane_state *dc_plane_state,
>> @@ -1251,64 +1219,61 @@ __set_dm_plane_colorop_shaper(struct drm_plane_state *plane_state,
>>       struct drm_colorop *old_colorop;
>>       struct drm_colorop_state *colorop_state = NULL, *new_colorop_state;
>>       struct drm_atomic_state *state = plane_state->state;
>> +    enum dc_transfer_func_predefined default_tf = TRANSFER_FUNCTION_LINEAR;
>> +    struct dc_transfer_func *tf = &dc_plane_state->in_shaper_func;
>> +    const struct drm_color_lut *shaper_lut;
>> +    struct drm_device *dev = colorop->dev;
>> +    uint32_t shaper_size;
>>       int i = 0;
>>   +    /* 1D Curve - SHAPER TF */
>>       old_colorop = colorop;
>> -
>> -    /* 2nd op: 1d curve - shaper */
>>       for_each_new_colorop_in_state(state, colorop, new_colorop_state, i) {
>>           if (new_colorop_state->colorop == old_colorop &&
>>               (BIT(new_colorop_state->curve_1d_type) & amdgpu_dm_supported_shaper_tfs)) {
>>               colorop_state = new_colorop_state;
>>               break;
>>           }
>> -
>> -        if (new_colorop_state->colorop == old_colorop) {
>> -            colorop_state = new_colorop_state;
>> -            break;
>> -        }
>>       }
>>   -    if (!colorop_state)
>> -        return -EINVAL;
>> -
>> -    return __set_colorop_in_shaper_1d_curve(dc_plane_state, colorop_state);
>> -}
>> -
>> -
>> -static int
>> -__set_colorop_1d_curve_blend_tf_lut(struct dc_plane_state *dc_plane_state,
>> -                  struct drm_colorop_state *colorop_state)
>> -{
>> -
>> -    struct dc_transfer_func *tf = &dc_plane_state->blend_tf;
>> -    struct drm_colorop *colorop = colorop_state->colorop;
>> -    struct drm_device *drm = colorop->dev;
>> -    const struct drm_color_lut *blend_lut;
>> -    uint32_t blend_size = 0;
>> -
>> -    if (colorop->type != DRM_COLOROP_1D_CURVE)
>> -        return -EINVAL;
>> +    if (colorop_state && !colorop_state->bypass && colorop->type == DRM_COLOROP_1D_CURVE) {
>> +        drm_dbg(dev, "Shaper TF colorop with ID: %d\n", colorop->base.id);
>> +        tf->type = TF_TYPE_DISTRIBUTED_POINTS;
>> +        tf->tf = default_tf = amdgpu_colorop_tf_to_dc_tf(colorop_state->curve_1d_type);
>> +        tf->sdr_ref_white_level = SDR_WHITE_LEVEL_INIT_VALUE;
>> +        __set_output_tf(tf, 0, 0, false);
>> +    }
> 
> __set_output_tf() could fail silently -- would that be problematic?
> 
> Looks like this question applies to similar changes below.
> 
>>   -    if (!(BIT(colorop_state->curve_1d_type) & amdgpu_dm_supported_blnd_tfs))
>> +    /* 1D LUT - SHAPER LUT */
>> +    colorop = old_colorop->next;
>> +    if (!colorop) {
>> +        drm_dbg(dev, "no Shaper LUT colorop found\n");
>>           return -EINVAL;
>> -
>> -    if (colorop_state->bypass) {
>> -        tf->type = TF_TYPE_BYPASS;
>> -        tf->tf = TRANSFER_FUNCTION_LINEAR;
>> -        return 0;
>>       }
>>   -    drm_dbg(drm, "Blend colorop with ID: %d\n", colorop->base.id);
>> +    old_colorop = colorop;
>> +    for_each_new_colorop_in_state(state, colorop, new_colorop_state, i) {
>> +        if (new_colorop_state->colorop == old_colorop &&
>> +            new_colorop_state->colorop->type == DRM_COLOROP_1D_LUT) {
>> +            colorop_state = new_colorop_state;
>> +            break;
>> +        }
>> +    }
>>   -    if (colorop->type == DRM_COLOROP_1D_CURVE) {
>> +    if (colorop_state && !colorop_state->bypass && colorop->type == DRM_COLOROP_1D_LUT) {
>> +        drm_dbg(dev, "Shaper LUT colorop with ID: %d\n", colorop->base.id);
>>           tf->type = TF_TYPE_DISTRIBUTED_POINTS;
>> -        tf->tf = amdgpu_colorop_tf_to_dc_tf(colorop_state->curve_1d_type);
>> +        tf->tf = default_tf;
> 
> Not sure if IIUC --the shaper in HW can either be a custom 1D LUT, or a
> predefined 1D CURVE, but cannot be both enabled at the same time, right?
> 

HW has only a custom LUT but the color module combines the pre-defined
transfer function with a custom LUT. That's why we're exposing
a pre-defined TF, followed by a custom LUT.

If we could only support one or the other we would need to expose
multiple color pipelines, ones with TFs, and ones with LUTs.

Harry

> If so, what would be the expected outcome if both LUT and CURVE colorops are not in bypass? It looks from the code that we'd prefer the LUT.
> 
> Likewise, this question applies to similar changes below.
> 
> - Leo
> 
>>           tf->sdr_ref_white_level = SDR_WHITE_LEVEL_INIT_VALUE;
>> -        return __set_input_tf(NULL, tf, blend_lut, blend_size);
>> +        shaper_lut = __extract_blob_lut(colorop_state->data, &shaper_size);
>> +        shaper_size = shaper_lut != NULL ? shaper_size : 0;
>> +
>> +        /* Custom LUT size must be the same as supported size */
>> +        if (shaper_size == colorop->lut_size)
>> +            __set_output_tf(tf, shaper_lut, shaper_size, false);
>>       }
>>   -    return -EINVAL;
>> +    return 0;
>>   }
>>     static int
>> @@ -1319,28 +1284,63 @@ __set_dm_plane_colorop_blend(struct drm_plane_state *plane_state,
>>       struct drm_colorop *old_colorop;
>>       struct drm_colorop_state *colorop_state = NULL, *new_colorop_state;
>>       struct drm_atomic_state *state = plane_state->state;
>> +    enum dc_transfer_func_predefined default_tf = TRANSFER_FUNCTION_LINEAR;
>> +    struct dc_transfer_func *tf = &dc_plane_state->blend_tf;
>> +    const struct drm_color_lut *blend_lut;
>> +    struct drm_device *dev = colorop->dev;
>> +    uint32_t blend_size;
>>       int i = 0;
>>   +    /* 1D Curve - BLND TF */
>>       old_colorop = colorop;
>> -
>> -    /* 3nd op: 1d curve - blend */
>>       for_each_new_colorop_in_state(state, colorop, new_colorop_state, i) {
>>           if (new_colorop_state->colorop == old_colorop &&
>>               (BIT(new_colorop_state->curve_1d_type) & amdgpu_dm_supported_blnd_tfs)) {
>>               colorop_state = new_colorop_state;
>>               break;
>>           }
>> +    }
>> +
>> +    if (colorop_state && !colorop_state->bypass && colorop->type == DRM_COLOROP_1D_CURVE &&
>> +        (BIT(colorop_state->curve_1d_type) & amdgpu_dm_supported_blnd_tfs)) {
>> +        drm_dbg(dev, "Blend TF colorop with ID: %d\n", colorop->base.id);
>> +        tf->type = TF_TYPE_DISTRIBUTED_POINTS;
>> +        tf->tf = default_tf = amdgpu_colorop_tf_to_dc_tf(colorop_state->curve_1d_type);
>> +        tf->sdr_ref_white_level = SDR_WHITE_LEVEL_INIT_VALUE;
>> +        __set_input_tf(NULL, tf, blend_lut, blend_size);
>> +    }
>>   -        if (new_colorop_state->colorop == old_colorop) {
>> +    /* 1D Curve - BLND LUT */
>> +    colorop = old_colorop->next;
>> +    if (!colorop) {
>> +        drm_dbg(dev, "no Blend LUT colorop found\n");
>> +        return -EINVAL;
>> +    }
>> +
>> +    old_colorop = colorop;
>> +    for_each_new_colorop_in_state(state, colorop, new_colorop_state, i) {
>> +        if (new_colorop_state->colorop == old_colorop &&
>> +            new_colorop_state->colorop->type == DRM_COLOROP_1D_LUT) {
>>               colorop_state = new_colorop_state;
>>               break;
>>           }
>>       }
>>   -    if (!colorop_state)
>> -        return -EINVAL;
>> +    if (colorop_state && !colorop_state->bypass && colorop->type == DRM_COLOROP_1D_LUT &&
>> +        (BIT(colorop_state->curve_1d_type) & amdgpu_dm_supported_blnd_tfs)) {
>> +        drm_dbg(dev, "Blend LUT colorop with ID: %d\n", colorop->base.id);
>> +        tf->type = TF_TYPE_DISTRIBUTED_POINTS;
>> +        tf->tf = default_tf;
>> +        tf->sdr_ref_white_level = SDR_WHITE_LEVEL_INIT_VALUE;
>> +        blend_lut = __extract_blob_lut(colorop_state->data, &blend_size);
>> +        blend_size = blend_lut != NULL ? blend_size : 0;
>> +
>> +        /* Custom LUT size must be the same as supported size */
>> +        if (blend_size == colorop->lut_size)
>> +            __set_input_tf(NULL, tf, blend_lut, blend_size);
>> +    }
>>   -    return __set_colorop_1d_curve_blend_tf_lut(dc_plane_state, colorop_state);
>> +    return 0;
>>   }
>>     static int
>> @@ -1409,7 +1409,7 @@ amdgpu_dm_plane_set_colorop_properties(struct drm_plane_state *plane_state,
>>       if (ret)
>>           return ret;
>>   -    /* 1D Curve - SHAPER TF */
>> +    /* 1D Curve & LUT - SHAPER TF & LUT */
>>       colorop = colorop->next;
>>       if (!colorop) {
>>           drm_dbg(dev, "no Shaper TF colorop found\n");
>> @@ -1420,7 +1420,12 @@ amdgpu_dm_plane_set_colorop_properties(struct drm_plane_state *plane_state,
>>       if (ret)
>>           return ret;
>>   -    /* 1D Curve - BLND TF */
>> +    /* Shaper LUT colorop is already handled, just skip here */
>> +    colorop = colorop->next;
>> +    if (!colorop)
>> +        return -EINVAL;
>> +
>> +    /* 1D Curve & LUT - BLND TF & LUT */
>>       colorop = colorop->next;
>>       if (!colorop) {
>>           drm_dbg(dev, "no Blend TF colorop found\n");
>> @@ -1431,6 +1436,11 @@ amdgpu_dm_plane_set_colorop_properties(struct drm_plane_state *plane_state,
>>       if (ret)
>>           return ret;
>>   +    /* BLND LUT colorop is already handled, just skip here */
>> +    colorop = colorop->next;
>> +    if (!colorop)
>> +        return -EINVAL;
>> +
>>       return 0;
>>   }
>>   diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_colorop.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_colorop.c
>> index ff5828a1e8cd..8a5e15083f11 100644
>> --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_colorop.c
>> +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_colorop.c
>> @@ -29,6 +29,7 @@
>>   #include <drm/drm_property.h>
>>   #include <drm/drm_colorop.h>
>>   +#include "amdgpu.h"
>>   #include "amdgpu_dm_colorop.h"
>>     const u64 amdgpu_dm_supported_degam_tfs =
>> @@ -90,6 +91,22 @@ int amdgpu_dm_initialize_default_pipeline(struct drm_plane *plane, struct drm_pr
>>         i++;
>>   +    /* 1D LUT - SHAPER LUT */
>> +    ops[i] = kzalloc(sizeof(struct drm_colorop), GFP_KERNEL);
>> +    if (!ops[i]) {
>> +        DRM_ERROR("KMS: Failed to allocate colorop\n");
>> +        ret = -ENOMEM;
>> +        goto cleanup;
>> +    }
>> +
>> +    ret = drm_colorop_curve_1d_lut_init(dev, ops[i], plane, MAX_COLOR_LUT_ENTRIES);
>> +    if (ret)
>> +        goto cleanup;
>> +
>> +    drm_colorop_set_next_property(ops[i-1], ops[i]);
>> +
>> +    i++;
>> +
>>       /* 1D curve - BLND TF */
>>       ops[i] = kzalloc(sizeof(struct drm_colorop), GFP_KERNEL);
>>       if (!ops[i]) {
>> @@ -104,6 +121,21 @@ int amdgpu_dm_initialize_default_pipeline(struct drm_plane *plane, struct drm_pr
>>         drm_colorop_set_next_property(ops[i-1], ops[i]);
>>   +    i++;
>> +
>> +    /* 1D LUT - BLND LUT */
>> +    ops[i] = kzalloc(sizeof(struct drm_colorop), GFP_KERNEL);
>> +    if (!ops[i]) {
>> +        DRM_ERROR("KMS: Failed to allocate colorop\n");
>> +        ret = -ENOMEM;
>> +        goto cleanup;
>> +    }
>> +
>> +    ret = drm_colorop_curve_1d_lut_init(dev, ops[i], plane, MAX_COLOR_LUT_ENTRIES);
>> +    if (ret)
>> +        goto cleanup;
>> +
>> +    drm_colorop_set_next_property(ops[i-1], ops[i]);
>>       return 0;
>>     cleanup:
>