Re: [PATCH 2/2] Input: atmel_mxt_ts: atmel_mxt_ts: Fix event loss

[Dmitry Torokhov <dmitry.torokhov@gmail.com>]

On Wed, Jun 23, 2021 at 03:56:37PM +0200, Loic Poulain wrote:
> If both touch events and release are part of the same report,
> userspace will not consider it as a touch-down & touch-up but as
> a non-action. That can happen on resume when 'buffered' events are
> dequeued in a row.
> 
> Make sure that release always causes previous events to be synced
> before being reported.
> 
> Signed-off-by: Loic Poulain <loic.poulain@linaro.org>
> ---
>  drivers/input/touchscreen/atmel_mxt_ts.c | 7 +++++++
>  1 file changed, 7 insertions(+)
> 
> diff --git a/drivers/input/touchscreen/atmel_mxt_ts.c b/drivers/input/touchscreen/atmel_mxt_ts.c
> index 807f449..e05ec30 100644
> --- a/drivers/input/touchscreen/atmel_mxt_ts.c
> +++ b/drivers/input/touchscreen/atmel_mxt_ts.c
> @@ -990,6 +990,13 @@ static void mxt_proc_t100_message(struct mxt_data *data, u8 *message)
>  		input_report_abs(input_dev, ABS_MT_DISTANCE, distance);
>  		input_report_abs(input_dev, ABS_MT_ORIENTATION, orientation);
>  	} else {
> +		/*
> +		 * Always sync input before reporting release, to be sure
> +		 * previous event(s) are taking into account by user side.
> +		 */
> +		if (data->update_input)
> +			mxt_input_sync(data);

That means we sync for every contact release, whereas I think ideal
would be to only sync when we observe touch-down and touch-up in the
same slot.

Let's also add Peter to the conversation...

> +
>  		dev_dbg(dev, "[%u] release\n", id);
>  
>  		/* close out slot */
> -- 
> 2.7.4
> 

Thanks.

-- 
Dmitry