[PATCH v2] drm/amd/display: don't ignore alpha property on pre-multiplied mode

[PATCH v2] drm/amd/display: don't ignore alpha property on pre-multiplied mode

[Melissa Wen]

"Pre-multiplied" is the default pixel blend mode for KMS/DRM, as
documented in supported_modes of drm_plane_create_blend_mode_property():
https://cgit.freedesktop.org/drm/drm-misc/tree/drivers/gpu/drm/drm_blend.c

In this mode, both 'pixel alpha' and 'plane alpha' participate in the
calculation, as described by the pixel blend mode formula in KMS/DRM
documentation:

out.rgb = plane_alpha * fg.rgb +
          (1 - (plane_alpha * fg.alpha)) * bg.rgb

Considering the blend config mechanisms we have in the driver so far,
the alpha mode that better fits this blend mode is the
_PER_PIXEL_ALPHA_COMBINED_GLOBAL_GAIN, where the value for global_gain
is the plane alpha (global_alpha).

With this change, alpha property stops to be ignored. It also addresses
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1734

v2:
 * keep the 8-bit value for global_alpha_value (Nicholas)
 * correct the logical ordering for combined global gain (Nicholas)
 * apply to dcn10 too (Nicholas)

Signed-off-by: Melissa Wen <mwen@igalia.com>
---
 .../drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c  | 14 +++++++++-----
 drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 14 +++++++++-----
 2 files changed, 18 insertions(+), 10 deletions(-)

diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
index ad757b59e00e..b1034e6258c8 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
@@ -2528,14 +2528,18 @@ void dcn10_update_mpcc(struct dc *dc, struct pipe_ctx *pipe_ctx)
 	struct mpc *mpc = dc->res_pool->mpc;
 	struct mpc_tree *mpc_tree_params = &(pipe_ctx->stream_res.opp->mpc_tree_params);
 
-	if (per_pixel_alpha)
-		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA;
-	else
-		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_GLOBAL_ALPHA;
-
 	blnd_cfg.overlap_only = false;
 	blnd_cfg.global_gain = 0xff;
 
+	if (per_pixel_alpha && pipe_ctx->plane_state->global_alpha) {
+		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA_COMBINED_GLOBAL_GAIN;
+		blnd_cfg.global_gain = pipe_ctx->plane_state->global_alpha_value;
+	} else if (per_pixel_alpha) {
+		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA;
+	} else {
+		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_GLOBAL_ALPHA;
+	}
+
 	if (pipe_ctx->plane_state->global_alpha)
 		blnd_cfg.global_alpha = pipe_ctx->plane_state->global_alpha_value;
 	else
diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
index 4290eaf11a04..b627c41713cc 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
@@ -2344,14 +2344,18 @@ void dcn20_update_mpcc(struct dc *dc, struct pipe_ctx *pipe_ctx)
 	struct mpc *mpc = dc->res_pool->mpc;
 	struct mpc_tree *mpc_tree_params = &(pipe_ctx->stream_res.opp->mpc_tree_params);
 
-	if (per_pixel_alpha)
-		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA;
-	else
-		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_GLOBAL_ALPHA;
-
 	blnd_cfg.overlap_only = false;
 	blnd_cfg.global_gain = 0xff;
 
+	if (per_pixel_alpha && pipe_ctx->plane_state->global_alpha) {
+		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA_COMBINED_GLOBAL_GAIN;
+		blnd_cfg.global_gain = pipe_ctx->plane_state->global_alpha_value;
+	} else if (per_pixel_alpha) {
+		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA;
+	} else {
+		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_GLOBAL_ALPHA;
+	}
+
 	if (pipe_ctx->plane_state->global_alpha)
 		blnd_cfg.global_alpha = pipe_ctx->plane_state->global_alpha_value;
 	else
-- 
2.35.1

Re: [PATCH v2] drm/amd/display: don't ignore alpha property on pre-multiplied mode

[Rodrigo Siqueira Jordao]

On 2022-03-29 16:18, Melissa Wen wrote:
> "Pre-multiplied" is the default pixel blend mode for KMS/DRM, as
> documented in supported_modes of drm_plane_create_blend_mode_property():
> https://cgit.freedesktop.org/drm/drm-misc/tree/drivers/gpu/drm/drm_blend.c
> 
> In this mode, both 'pixel alpha' and 'plane alpha' participate in the
> calculation, as described by the pixel blend mode formula in KMS/DRM
> documentation:
> 
> out.rgb = plane_alpha * fg.rgb +
>            (1 - (plane_alpha * fg.alpha)) * bg.rgb
> 
> Considering the blend config mechanisms we have in the driver so far,
> the alpha mode that better fits this blend mode is the
> _PER_PIXEL_ALPHA_COMBINED_GLOBAL_GAIN, where the value for global_gain
> is the plane alpha (global_alpha).
> 
> With this change, alpha property stops to be ignored. It also addresses
> Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1734
> 
> v2:
>   * keep the 8-bit value for global_alpha_value (Nicholas)
>   * correct the logical ordering for combined global gain (Nicholas)
>   * apply to dcn10 too (Nicholas)
> 
> Signed-off-by: Melissa Wen <mwen@igalia.com>
> ---
>   .../drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c  | 14 +++++++++-----
>   drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 14 +++++++++-----
>   2 files changed, 18 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
> index ad757b59e00e..b1034e6258c8 100644
> --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
> +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
> @@ -2528,14 +2528,18 @@ void dcn10_update_mpcc(struct dc *dc, struct pipe_ctx *pipe_ctx)
>   	struct mpc *mpc = dc->res_pool->mpc;
>   	struct mpc_tree *mpc_tree_params = &(pipe_ctx->stream_res.opp->mpc_tree_params);
>   
> -	if (per_pixel_alpha)
> -		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA;
> -	else
> -		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_GLOBAL_ALPHA;
> -
>   	blnd_cfg.overlap_only = false;
>   	blnd_cfg.global_gain = 0xff;
>   
> +	if (per_pixel_alpha && pipe_ctx->plane_state->global_alpha) {
> +		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA_COMBINED_GLOBAL_GAIN;
> +		blnd_cfg.global_gain = pipe_ctx->plane_state->global_alpha_value;
> +	} else if (per_pixel_alpha) {
> +		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA;
> +	} else {
> +		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_GLOBAL_ALPHA;
> +	}
> +
>   	if (pipe_ctx->plane_state->global_alpha)
>   		blnd_cfg.global_alpha = pipe_ctx->plane_state->global_alpha_value;
>   	else
> diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
> index 4290eaf11a04..b627c41713cc 100644
> --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
> +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
> @@ -2344,14 +2344,18 @@ void dcn20_update_mpcc(struct dc *dc, struct pipe_ctx *pipe_ctx)
>   	struct mpc *mpc = dc->res_pool->mpc;
>   	struct mpc_tree *mpc_tree_params = &(pipe_ctx->stream_res.opp->mpc_tree_params);
>   
> -	if (per_pixel_alpha)
> -		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA;
> -	else
> -		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_GLOBAL_ALPHA;
> -
>   	blnd_cfg.overlap_only = false;
>   	blnd_cfg.global_gain = 0xff;
>   
> +	if (per_pixel_alpha && pipe_ctx->plane_state->global_alpha) {
> +		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA_COMBINED_GLOBAL_GAIN;
> +		blnd_cfg.global_gain = pipe_ctx->plane_state->global_alpha_value;
> +	} else if (per_pixel_alpha) {
> +		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA;
> +	} else {
> +		blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_GLOBAL_ALPHA;
> +	}
> +

Hi Melissa,

Thanks a lot for this patch. I run your patch in our CI, and everything 
looks good from the IGT test result.

In this sense:
Tested-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>

However, I think it is still necessary to have someone else review.

Harry, Nick, or Zhan, what do you think about this change?

Thanks
Siqueira

>   	if (pipe_ctx->plane_state->global_alpha)
>   		blnd_cfg.global_alpha = pipe_ctx->plane_state->global_alpha_value;
>   	else

Re: [PATCH v2] drm/amd/display: don't ignore alpha property on pre-multiplied mode

[Harry Wentland]

On 2022-03-30 08:30, Rodrigo Siqueira Jordao wrote:
> 
> 
> On 2022-03-29 16:18, Melissa Wen wrote:
>> "Pre-multiplied" is the default pixel blend mode for KMS/DRM, as
>> documented in supported_modes of drm_plane_create_blend_mode_property():
>> https://cgit.freedesktop.org/drm/drm-misc/tree/drivers/gpu/drm/drm_blend.c
>>
>> In this mode, both 'pixel alpha' and 'plane alpha' participate in the
>> calculation, as described by the pixel blend mode formula in KMS/DRM
>> documentation:
>>
>> out.rgb = plane_alpha * fg.rgb +
>>            (1 - (plane_alpha * fg.alpha)) * bg.rgb
>>
>> Considering the blend config mechanisms we have in the driver so far,
>> the alpha mode that better fits this blend mode is the
>> _PER_PIXEL_ALPHA_COMBINED_GLOBAL_GAIN, where the value for global_gain
>> is the plane alpha (global_alpha).
>>
>> With this change, alpha property stops to be ignored. It also addresses
>> Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1734
>>
>> v2:
>>   * keep the 8-bit value for global_alpha_value (Nicholas)
>>   * correct the logical ordering for combined global gain (Nicholas)
>>   * apply to dcn10 too (Nicholas)
>>
>> Signed-off-by: Melissa Wen <mwen@igalia.com>
>> ---
>>   .../drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c  | 14 +++++++++-----
>>   drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c | 14 +++++++++-----
>>   2 files changed, 18 insertions(+), 10 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
>> index ad757b59e00e..b1034e6258c8 100644
>> --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
>> +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
>> @@ -2528,14 +2528,18 @@ void dcn10_update_mpcc(struct dc *dc, struct pipe_ctx *pipe_ctx)
>>       struct mpc *mpc = dc->res_pool->mpc;
>>       struct mpc_tree *mpc_tree_params = &(pipe_ctx->stream_res.opp->mpc_tree_params);
>>   -    if (per_pixel_alpha)
>> -        blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA;
>> -    else
>> -        blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_GLOBAL_ALPHA;
>> -
>>       blnd_cfg.overlap_only = false;
>>       blnd_cfg.global_gain = 0xff;
>>   +    if (per_pixel_alpha && pipe_ctx->plane_state->global_alpha) {
>> +        blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA_COMBINED_GLOBAL_GAIN;
>> +        blnd_cfg.global_gain = pipe_ctx->plane_state->global_alpha_value;
>> +    } else if (per_pixel_alpha) {
>> +        blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA;
>> +    } else {
>> +        blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_GLOBAL_ALPHA;
>> +    }
>> +
>>       if (pipe_ctx->plane_state->global_alpha)
>>           blnd_cfg.global_alpha = pipe_ctx->plane_state->global_alpha_value;
>>       else
>> diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
>> index 4290eaf11a04..b627c41713cc 100644
>> --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
>> +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
>> @@ -2344,14 +2344,18 @@ void dcn20_update_mpcc(struct dc *dc, struct pipe_ctx *pipe_ctx)
>>       struct mpc *mpc = dc->res_pool->mpc;
>>       struct mpc_tree *mpc_tree_params = &(pipe_ctx->stream_res.opp->mpc_tree_params);
>>   -    if (per_pixel_alpha)
>> -        blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA;
>> -    else
>> -        blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_GLOBAL_ALPHA;
>> -
>>       blnd_cfg.overlap_only = false;
>>       blnd_cfg.global_gain = 0xff;
>>   +    if (per_pixel_alpha && pipe_ctx->plane_state->global_alpha) {
>> +        blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA_COMBINED_GLOBAL_GAIN;
>> +        blnd_cfg.global_gain = pipe_ctx->plane_state->global_alpha_value;
>> +    } else if (per_pixel_alpha) {
>> +        blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA;
>> +    } else {
>> +        blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_GLOBAL_ALPHA;
>> +    }
>> +
> 
> Hi Melissa,
> 
> Thanks a lot for this patch. I run your patch in our CI, and everything looks good from the IGT test result.
> 
> In this sense:
> Tested-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
> 
> However, I think it is still necessary to have someone else review.
> 
> Harry, Nick, or Zhan, what do you think about this change?
> 

Looks good to me.

Reviewed-by: Harry Wentland <harry.wentland@amd.com>

Harry

> Thanks
> Siqueira
> 
>>       if (pipe_ctx->plane_state->global_alpha)
>>           blnd_cfg.global_alpha = pipe_ctx->plane_state->global_alpha_value;
>>       else
> 

Re: [PATCH v2] drm/amd/display: don't ignore alpha property on pre-multiplied mode

[Simon Ser]

I've tested this patch and it fixes my bug [1]. Thanks!

Tested-by: Simon Ser <contact@emersion.fr>

[1]: https://gitlab.freedesktop.org/drm/amd/-/issues/1734

Re: [PATCH v2] drm/amd/display: don't ignore alpha property on pre-multiplied mode

[Rodrigo Siqueira Jordao]

Patch merged to amd-staging-drm-next.

Thanks a lot!

On 2022-04-05 15:32, Simon Ser wrote:
> I've tested this patch and it fixes my bug [1]. Thanks!
> 
> Tested-by: Simon Ser <contact@emersion.fr>
> 
> [1]: https://gitlab.freedesktop.org/drm/amd/-/issues/1734>