[PATCH v15 07/11] drm/mediatek: Add retry to prevent misjudgment for sink devices

[AngeloGioacchino Del Regno]

Il 27/07/22 06:50, Bo-Chen Chen ha scritto:
> For some DP dungles, we need to train more than onece to confirm that we
> don't misjudge the status of sink device.

Please fix the typos in your commit title and description.
title: misjudgment -> misjudgement
desc: dungles->dongles; onece->once

> 
> Signed-off-by: Bo-Chen Chen <rex-bc.chen at mediatek.com>
> ---
>   drivers/gpu/drm/mediatek/mtk_dp.c | 21 ++++++++++++++++++---
>   1 file changed, 18 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpu/drm/mediatek/mtk_dp.c b/drivers/gpu/drm/mediatek/mtk_dp.c
> index ce817cb59445..80d7d6488105 100644
> --- a/drivers/gpu/drm/mediatek/mtk_dp.c
> +++ b/drivers/gpu/drm/mediatek/mtk_dp.c
> @@ -42,6 +42,7 @@
>   #define MTK_DP_CHECK_SINK_CAP_TIMEOUT_COUNT 3
>   #define MTK_DP_TBC_BUF_READ_START_ADDR 0x08
>   #define MTK_DP_TRAIN_DOWNSCALE_RETRY 8
> +#define MTK_DP_TRAIN_CLEAR_RETRY 50
>   
>   struct mtk_dp_train_info {
>   	bool tps3;
> @@ -1431,11 +1432,25 @@ static int mtk_dp_video_config(struct mtk_dp *mtk_dp)
>   
>   static int mtk_dp_training(struct mtk_dp *mtk_dp)
>   {
> +	short max_retry = MTK_DP_TRAIN_CLEAR_RETRY;
>   	int ret;
>   
> -	ret = mtk_dp_train_start(mtk_dp);
> -	if (ret)
> -		return ret;
> +	/*
> +	 * We do retry to confirm that we don't misjudge the sink status.
> +	 * If it is still failed, we can confirm there are some issues for the
> +	 * sink device.
> +	 */
> +	do {
> +		ret = mtk_dp_train_start(mtk_dp);
> +		if (!ret)
> +			break;
> +	} while (--max_retry);
> +
> +	dev_info(mtk_dp->dev, "dp training clear retry times: %d\n",
> +		 MTK_DP_TRAIN_CLEAR_RETRY - max_retry);

dev_dbg() here.

...after which,

Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno at collabora.com>

> +
> +	if (!max_retry)
> +		return -ETIMEDOUT;
>   
>   	ret = mtk_dp_video_config(mtk_dp);
>   	if (ret)