Re: [PATCH] drm/msm/dpu: correct LM pairing for SM6150

[Abhinav Kumar]

On 12/16/2024 12:20 AM, Dmitry Baryshkov wrote:
> According to the vendor devicetree on SM6150 LM_0 is paired with LM_2
> rather than LM_1. Correct pairing indices.
>
> Fixes: cb2f9144693b ("drm/msm/dpu: Add SM6150 support")
> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@xxxxxxxxxx>
> ---
>   drivers/gpu/drm/msm/disp/dpu1/catalog/dpu_5_3_sm6150.h | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/msm/disp/dpu1/catalog/dpu_5_3_sm6150.h b/drivers/gpu/drm/msm/disp/dpu1/catalog/dpu_5_3_sm6150.h
> index 621a2140f675fa28b3a7fcd8573e59b306cd6832..81eb274cc7000a3b70b0f6650088ddcd24648eab 100644
> --- a/drivers/gpu/drm/msm/disp/dpu1/catalog/dpu_5_3_sm6150.h
> +++ b/drivers/gpu/drm/msm/disp/dpu1/catalog/dpu_5_3_sm6150.h
> @@ -116,20 +116,20 @@ static const struct dpu_lm_cfg sm6150_lm[] = {
>   		.sblk = &sdm845_lm_sblk,
>   		.pingpong = PINGPONG_0,
>   		.dspp = DSPP_0,
> -		.lm_pair = LM_1,
> +		.lm_pair = LM_2,
>   	}, {
>   		.name = "lm_1", .id = LM_1,
>   		.base = 0x45000, .len = 0x320,
>   		.features = MIXER_QCM2290_MASK,
>   		.sblk = &sdm845_lm_sblk,
>   		.pingpong = PINGPONG_1,
> -		.lm_pair = LM_0,
>   	}, {
>   		.name = "lm_2", .id = LM_2,
>   		.base = 0x46000, .len = 0x320,
>   		.features = MIXER_QCM2290_MASK,
>   		.sblk = &sdm845_lm_sblk,
>   		.pingpong = PINGPONG_2,
> +		.lm_pair = LM_0,
>   	},
>   };

Have a basic question here. We check the peer only if we will have more 
than one LM needed in the topology but sm6150 does not have 3dmux, so 
the number of LMs will not go beyond one.

318 		/* Valid primary mixer found, find matching peers */
319 		if (lm_count < reqs->topology.num_lm) {

It seems like this fix will be unused or does not really matter.

Downstream has a different implementation for lm_pair, its used even to 
decide the LM pair for CWB mux. Upstream has a simpler implementation of 
just doing that in the code of using ODD LMs for ODD CWB muxes and even 
LMs for even CWB muxes. So fix is okay but not needed.

>   
>
> ---
> base-commit: a3d570eace66b4016f2692a6f1045742ee70c6b1
> change-id: 20241216-dpu-fix-sm6150-17f0739f8fe0
>
> Best regards,