Re: [PATCH] Revert "Input: atmel_mxt_ts - disable interrupt for 50ms after reset"

From: Nick Dyer
Date: Fri Apr 08 2016 - 05:10:16 EST


On 2016-04-07 23:52, Tom Rini wrote:
> This reverts commit 885f3fb9fa1f9e185e8a4e905157087495734349 due to this
> change breaking the touchpad on the Chromebook Pixel 2015 on resume from
> sleep or warm resets.
>
> Cc: Olof Johansson <olof@xxxxxxxxx>
> Cc: Nick Dyer <nick.dyer@xxxxxxxxxxx>
> Cc: Dmitry Torokhov <dmitry.torokhov@xxxxxxxxx>
> Cc: Henrik Rydberg <rydberg@xxxxxxxxxxx>
> Signed-off-by: Tom Rini <trini@xxxxxxxxxxxx>

Hi Tom-

Sorry that this has caused a problem. I suggest we try and find a 3rd
option other than reverting this patch, because otherwise we will cause
problems on other platforms.

I have a Pixel 2 here - can you advise how to reproduce?

> ---
> drivers/input/touchscreen/atmel_mxt_ts.c | 9 +--------
> 1 file changed, 1 insertion(+), 8 deletions(-)
>
> diff --git a/drivers/input/touchscreen/atmel_mxt_ts.c b/drivers/input/touchscreen/atmel_mxt_ts.c
> index 2160512..9b92b60 100644
> --- a/drivers/input/touchscreen/atmel_mxt_ts.c
> +++ b/drivers/input/touchscreen/atmel_mxt_ts.c
> @@ -1098,9 +1098,7 @@ static int mxt_soft_reset(struct mxt_data *data)
> struct device *dev = &data->client->dev;
> int ret = 0;
>
> - dev_info(dev, "Resetting device\n");
> -
> - disable_irq(data->irq);
> + dev_info(dev, "Resetting chip\n");
>
> reinit_completion(&data->reset_completion);
>
> @@ -1108,11 +1106,6 @@ static int mxt_soft_reset(struct mxt_data *data)
> if (ret)
> return ret;
>
> - /* Ignore CHG line for 100ms after reset */
> - msleep(100);
> -
> - enable_irq(data->irq);

I suspect what is going on here is that Pixel 2 is using edge triggering
and we are missing the edge during the 100ms delay. Could you try just
changing this from "enable_irq" to
mxt_acquire_irq(data);

> -
> ret = mxt_wait_for_completion(data, &data->reset_completion,
> MXT_RESET_TIMEOUT);
> if (ret)
>