[PATCH] platform/x86: adv_swbutton: disable wakeup in .remove() and the error path of .probe()

[PATCH] platform/x86: adv_swbutton: disable wakeup in .remove() and the error path of .probe()

[Joe Hattori]

Current code leaves the device's wakeup enabled in the error path of
.probe() and .remove(), which results in a memory leak. Therefore, add
the device_init_wakeup(&device->dev, false) calls.

Fixes: 3d904005f686 ("platform/x86: add support for Advantech software defined button")
Signed-off-by: Joe Hattori <joe@pf.is.s.u-tokyo.ac.jp>
---
 drivers/platform/x86/adv_swbutton.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/platform/x86/adv_swbutton.c b/drivers/platform/x86/adv_swbutton.c
index 6fa60f3fc53c..523836be6d4b 100644
--- a/drivers/platform/x86/adv_swbutton.c
+++ b/drivers/platform/x86/adv_swbutton.c
@@ -84,6 +84,7 @@ static int adv_swbutton_probe(struct platform_device *device)
 					     device);
 	if (ACPI_FAILURE(status)) {
 		dev_err(&device->dev, "Error installing notify handler\n");
+		device_init_wakeup(&device->dev, false);
 		return -EIO;
 	}
 
@@ -96,6 +97,7 @@ static void adv_swbutton_remove(struct platform_device *device)
 
 	acpi_remove_notify_handler(handle, ACPI_DEVICE_NOTIFY,
 				   adv_swbutton_notify);
+	device_init_wakeup(&device->dev, false);
 }
 
 static const struct acpi_device_id button_device_ids[] = {
-- 
2.34.1

Re: [PATCH] platform/x86: adv_swbutton: disable wakeup in .remove() and the error path of .probe()

[Ilpo Järvinen]

On Fri, 6 Dec 2024, Joe Hattori wrote:

> Current code leaves the device's wakeup enabled in the error path of
> .probe() and .remove(), which results in a memory leak. Therefore, add
> the device_init_wakeup(&device->dev, false) calls.
> 
> Fixes: 3d904005f686 ("platform/x86: add support for Advantech software defined button")
> Signed-off-by: Joe Hattori <joe@pf.is.s.u-tokyo.ac.jp>
> ---
>  drivers/platform/x86/adv_swbutton.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/platform/x86/adv_swbutton.c b/drivers/platform/x86/adv_swbutton.c
> index 6fa60f3fc53c..523836be6d4b 100644
> --- a/drivers/platform/x86/adv_swbutton.c
> +++ b/drivers/platform/x86/adv_swbutton.c
> @@ -84,6 +84,7 @@ static int adv_swbutton_probe(struct platform_device *device)
>  					     device);
>  	if (ACPI_FAILURE(status)) {
>  		dev_err(&device->dev, "Error installing notify handler\n");
> +		device_init_wakeup(&device->dev, false);
>  		return -EIO;
>  	}
>  
> @@ -96,6 +97,7 @@ static void adv_swbutton_remove(struct platform_device *device)
>  
>  	acpi_remove_notify_handler(handle, ACPI_DEVICE_NOTIFY,
>  				   adv_swbutton_notify);
> +	device_init_wakeup(&device->dev, false);

Is the non-symmetric order here intentional?

How about using devm_add_action_or_reset() instead?

-- 
 i.

Re: [PATCH] platform/x86: adv_swbutton: disable wakeup in .remove() and the error path of .probe()

[Joe Hattori]

Hi Ilpo,

Thank you for your review.

On 12/10/24 22:52, Ilpo Järvinen wrote:
> On Fri, 6 Dec 2024, Joe Hattori wrote:
> 
>> Current code leaves the device's wakeup enabled in the error path of
>> .probe() and .remove(), which results in a memory leak. Therefore, add
>> the device_init_wakeup(&device->dev, false) calls.
>>
>> Fixes: 3d904005f686 ("platform/x86: add support for Advantech software defined button")
>> Signed-off-by: Joe Hattori <joe@pf.is.s.u-tokyo.ac.jp>
>> ---
>>   drivers/platform/x86/adv_swbutton.c | 2 ++
>>   1 file changed, 2 insertions(+)
>>
>> diff --git a/drivers/platform/x86/adv_swbutton.c b/drivers/platform/x86/adv_swbutton.c
>> index 6fa60f3fc53c..523836be6d4b 100644
>> --- a/drivers/platform/x86/adv_swbutton.c
>> +++ b/drivers/platform/x86/adv_swbutton.c
>> @@ -84,6 +84,7 @@ static int adv_swbutton_probe(struct platform_device *device)
>>   					     device);
>>   	if (ACPI_FAILURE(status)) {
>>   		dev_err(&device->dev, "Error installing notify handler\n");
>> +		device_init_wakeup(&device->dev, false);
>>   		return -EIO;
>>   	}
>>   
>> @@ -96,6 +97,7 @@ static void adv_swbutton_remove(struct platform_device *device)
>>   
>>   	acpi_remove_notify_handler(handle, ACPI_DEVICE_NOTIFY,
>>   				   adv_swbutton_notify);
>> +	device_init_wakeup(&device->dev, false);
> 
> Is the non-symmetric order here intentional?
> 
> How about using devm_add_action_or_reset() instead?
>

Agreed, just sent a v2 patch using devm_add_action_or_reset(), so please 
take a look at it.

Best,
Joe