[PATCH] Input: goodix - disable IRQs while suspended

[PATCH] Input: goodix - disable IRQs while suspended

[Hans de Goede]

We should not try to do any i2c transfers before the controller is
resumed (which happens before our resume method gets called).

So we need to disable our IRQ while suspended to enforce this. The
code paths for devices with GPIOs for the int and reset pins already
disable the IRQ the through goodix_free_irq().

This commit also disables the IRQ while suspended for devices without
GPIOs for the int and reset pins.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
---
 drivers/input/touchscreen/goodix.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/input/touchscreen/goodix.c b/drivers/input/touchscreen/goodix.c
index 69d0b8cbc71f..ecec8eb17f28 100644
--- a/drivers/input/touchscreen/goodix.c
+++ b/drivers/input/touchscreen/goodix.c
@@ -878,8 +878,10 @@ static int __maybe_unused goodix_suspend(struct device *dev)
 	int error;
 
 	/* We need gpio pins to suspend/resume */
-	if (!ts->gpiod_int || !ts->gpiod_rst)
+	if (!ts->gpiod_int || !ts->gpiod_rst) {
+		disable_irq(client->irq);
 		return 0;
+	}
 
 	wait_for_completion(&ts->firmware_loading_complete);
 
@@ -919,8 +921,10 @@ static int __maybe_unused goodix_resume(struct device *dev)
 	struct goodix_ts_data *ts = i2c_get_clientdata(client);
 	int error;
 
-	if (!ts->gpiod_int || !ts->gpiod_rst)
+	if (!ts->gpiod_int || !ts->gpiod_rst) {
+		enable_irq(client->irq);
 		return 0;
+	}
 
 	/*
 	 * Exit sleep mode by outputting HIGH level to INT pin
-- 
2.14.3

Re: [PATCH] Input: goodix - disable IRQs while suspended

[Bastien Nocera]

On Wed, 2018-01-10 at 12:34 +0100, Hans de Goede wrote:
> We should not try to do any i2c transfers before the controller is
> resumed (which happens before our resume method gets called).
> 
> So we need to disable our IRQ while suspended to enforce this. The
> code paths for devices with GPIOs for the int and reset pins already
> disable the IRQ the through goodix_free_irq().
> 
> This commit also disables the IRQ while suspended for devices without
> GPIOs for the int and reset pins.

Do disable_irq() and enable_irq() require to be paired, and if so, are
goodix_suspend() and goodix_resume() guaranteed to be paired?

Other than that, looks good to me. Please add the name of the device(s)
you tested this on in the changelog and add your tested-by, to show
it's not a "compile-only check" kind of commit.

Cheers

> Signed-off-by: Hans de Goede <hdegoede@redhat.com>
> ---
>  drivers/input/touchscreen/goodix.c | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/input/touchscreen/goodix.c
> b/drivers/input/touchscreen/goodix.c
> index 69d0b8cbc71f..ecec8eb17f28 100644
> --- a/drivers/input/touchscreen/goodix.c
> +++ b/drivers/input/touchscreen/goodix.c
> @@ -878,8 +878,10 @@ static int __maybe_unused goodix_suspend(struct
> device *dev)
>  	int error;
>  
>  	/* We need gpio pins to suspend/resume */
> -	if (!ts->gpiod_int || !ts->gpiod_rst)
> +	if (!ts->gpiod_int || !ts->gpiod_rst) {
> +		disable_irq(client->irq);
>  		return 0;
> +	}
>  
>  	wait_for_completion(&ts->firmware_loading_complete);
>  
> @@ -919,8 +921,10 @@ static int __maybe_unused goodix_resume(struct
> device *dev)
>  	struct goodix_ts_data *ts = i2c_get_clientdata(client);
>  	int error;
>  
> -	if (!ts->gpiod_int || !ts->gpiod_rst)
> +	if (!ts->gpiod_int || !ts->gpiod_rst) {
> +		enable_irq(client->irq);
>  		return 0;
> +	}
>  
>  	/*
>  	 * Exit sleep mode by outputting HIGH level to INT pin

Re: [PATCH] Input: goodix - disable IRQs while suspended

[Hans de Goede]

Hi,

On 10-01-18 15:14, Bastien Nocera wrote:
> On Wed, 2018-01-10 at 12:34 +0100, Hans de Goede wrote:
>> We should not try to do any i2c transfers before the controller is
>> resumed (which happens before our resume method gets called).
>>
>> So we need to disable our IRQ while suspended to enforce this. The
>> code paths for devices with GPIOs for the int and reset pins already
>> disable the IRQ the through goodix_free_irq().
>>
>> This commit also disables the IRQ while suspended for devices without
>> GPIOs for the int and reset pins.
> 
> Do disable_irq() and enable_irq() require to be paired,

Yes.

> and if so, are
> goodix_suspend() and goodix_resume() guaranteed to be paired?

Yes I believe they are, many i2c drivers call disable / enable_irq
from there suspend/resume handler.

> Other than that, looks good to me. Please add the name of the device(s)
> you tested this on in the changelog and add your tested-by, to show
> it's not a "compile-only check" kind of commit.

Ok, v2 with amended commit messages coming up.

Regards,

Hans

> 
> Cheers
> 
>> Signed-off-by: Hans de Goede <hdegoede@redhat.com>
>> ---
>>   drivers/input/touchscreen/goodix.c | 8 ++++++--
>>   1 file changed, 6 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/input/touchscreen/goodix.c
>> b/drivers/input/touchscreen/goodix.c
>> index 69d0b8cbc71f..ecec8eb17f28 100644
>> --- a/drivers/input/touchscreen/goodix.c
>> +++ b/drivers/input/touchscreen/goodix.c
>> @@ -878,8 +878,10 @@ static int __maybe_unused goodix_suspend(struct
>> device *dev)
>>   	int error;
>>   
>>   	/* We need gpio pins to suspend/resume */
>> -	if (!ts->gpiod_int || !ts->gpiod_rst)
>> +	if (!ts->gpiod_int || !ts->gpiod_rst) {
>> +		disable_irq(client->irq);
>>   		return 0;
>> +	}
>>   
>>   	wait_for_completion(&ts->firmware_loading_complete);
>>   
>> @@ -919,8 +921,10 @@ static int __maybe_unused goodix_resume(struct
>> device *dev)
>>   	struct goodix_ts_data *ts = i2c_get_clientdata(client);
>>   	int error;
>>   
>> -	if (!ts->gpiod_int || !ts->gpiod_rst)
>> +	if (!ts->gpiod_int || !ts->gpiod_rst) {
>> +		enable_irq(client->irq);
>>   		return 0;
>> +	}
>>   
>>   	/*
>>   	 * Exit sleep mode by outputting HIGH level to INT pin