[PATCH] iio: dln2: Check return value of devm_iio_trigger_register()

[PATCH] iio: dln2: Check return value of devm_iio_trigger_register()

[Lars-Peter Clausen]

Registering a trigger can fail and the return value of
devm_iio_trigger_register() must be checked. Otherwise undefined behavior
can occur when the trigger is used.

Fixes: 7c0299e879dd ("iio: adc: Add support for DLN2 ADC")
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
---
 drivers/iio/adc/dln2-adc.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/drivers/iio/adc/dln2-adc.c b/drivers/iio/adc/dln2-adc.c
index 16407664182c..846c92296221 100644
--- a/drivers/iio/adc/dln2-adc.c
+++ b/drivers/iio/adc/dln2-adc.c
@@ -656,7 +656,11 @@ static int dln2_adc_probe(struct platform_device *pdev)
 		return -ENOMEM;
 	}
 	iio_trigger_set_drvdata(dln2->trig, dln2);
-	devm_iio_trigger_register(dev, dln2->trig);
+	ret = devm_iio_trigger_register(dev, dln2->trig);
+	if (ret) {
+		dev_err(dev, "failed to register trigger: %d\n", ret);
+		return ret;
+	}
 	iio_trigger_set_immutable(indio_dev, dln2->trig);
 
 	ret = devm_iio_triggered_buffer_setup(dev, indio_dev, NULL,
-- 
2.20.1

Re: [PATCH] iio: dln2: Check return value of devm_iio_trigger_register()

[Jonathan Cameron]

On Mon,  1 Nov 2021 14:30:43 +0100
Lars-Peter Clausen <lars@metafoo.de> wrote:

> Registering a trigger can fail and the return value of
> devm_iio_trigger_register() must be checked. Otherwise undefined behavior
> can occur when the trigger is used.
> 
> Fixes: 7c0299e879dd ("iio: adc: Add support for DLN2 ADC")
> Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>

Seems obvious enough to pick up and won't go upstream until after the
merge window anyway so time for any feedback to come in.

Anyhow, applied to the fixes-togreg branch of iio.git and
marked for stable.

Thanks,

Jonathan

> ---
>  drivers/iio/adc/dln2-adc.c | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/iio/adc/dln2-adc.c b/drivers/iio/adc/dln2-adc.c
> index 16407664182c..846c92296221 100644
> --- a/drivers/iio/adc/dln2-adc.c
> +++ b/drivers/iio/adc/dln2-adc.c
> @@ -656,7 +656,11 @@ static int dln2_adc_probe(struct platform_device *pdev)
>  		return -ENOMEM;
>  	}
>  	iio_trigger_set_drvdata(dln2->trig, dln2);
> -	devm_iio_trigger_register(dev, dln2->trig);
> +	ret = devm_iio_trigger_register(dev, dln2->trig);
> +	if (ret) {
> +		dev_err(dev, "failed to register trigger: %d\n", ret);
> +		return ret;
> +	}
>  	iio_trigger_set_immutable(indio_dev, dln2->trig);
>  
>  	ret = devm_iio_triggered_buffer_setup(dev, indio_dev, NULL,

Re: [PATCH] iio: dln2: Check return value of devm_iio_trigger_register()

[Andy Shevchenko]

On Mon, Nov 1, 2021 at 3:31 PM Lars-Peter Clausen <lars@metafoo.de> wrote:
>
> Registering a trigger can fail and the return value of
> devm_iio_trigger_register() must be checked. Otherwise undefined behavior
> can occur when the trigger is used.

...

> +       ret = devm_iio_trigger_register(dev, dln2->trig);

> +       if (ret) {
> +               dev_err(dev, "failed to register trigger: %d\n", ret);
> +               return ret;

It's fine to use dev_err_ptobe() in ->probe() even for known error codes.
Hence,

if (ret)
 return dev_err_probe(dev, ret, "failed to register trigger\n");

> +       }

Should it be a separate patch?

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH] iio: dln2: Check return value of devm_iio_trigger_register()

[Lars-Peter Clausen]

On 11/4/21 11:35 AM, Andy Shevchenko wrote:
> On Mon, Nov 1, 2021 at 3:31 PM Lars-Peter Clausen <lars@metafoo.de> wrote:
>> Registering a trigger can fail and the return value of
>> devm_iio_trigger_register() must be checked. Otherwise undefined behavior
>> can occur when the trigger is used.
> ...
>
>> +       ret = devm_iio_trigger_register(dev, dln2->trig);
>> +       if (ret) {
>> +               dev_err(dev, "failed to register trigger: %d\n", ret);
>> +               return ret;
> It's fine to use dev_err_ptobe() in ->probe() even for known error codes.
> Hence,
>
> if (ret)
>   return dev_err_probe(dev, ret, "failed to register trigger\n");
>
>> +       }
> Should it be a separate patch?
>
I'm a big fan of dev_err_probe() using it in all my new drivers for 
everything.

But here I'm trying to stay with the style of the existing driver. And 
we also want to have this as a fix that can potentially be backported.

The conversion to dev_err_probe() is best done for the whole driver in a 
separate feature patch.