Re: [PATCH v3 1/3] iio: adc: Cosmic Circuits 10001 ADC driver

[Lars-Peter Clausen <lars-Qo5EllUWu/uELgA04lAiVw@public.gmane.org>]

[...]
>>>
>>>>>>> +
>>>>>>> +	adc_dev->reg = devm_regulator_get(&pdev->dev, "vref");
>>>>>>> +	if (IS_ERR_OR_NULL(adc_dev->reg))
>>>>>>> +		return -EINVAL;
>>>>>> 	if (IS_ERR(adc_dev->reg))
>>>>>> 		return PTR_ERR(adc_dev->reg);
>>>>>
>>>>> Are you sure? What if devm_regulator_get() returns NULL?
>>>>>
>>>> I had a look at it, but couldn't figure out a condition in which it would return NULL. But I'd be happy to be informed of anything different.
>>>
>>> If CONFIG_REGULATOR=n you get NULL there (although I added a select REGULATOR
>>> on the v4 I just posted).
>>>
>> Indeed, thanks. Now I'm thinking if it would make sense to handle this special case in a special way (as it would be caused by an invalid kernel config). Well, I guess, I leave it up to you.
>
> Well, adding the select REGULATOR that shouldn't happen, so let's better
> drop the NULL check entirely.

Doing a 'select REGULATOR' can (and probably will sooner or later) create 
nasty dependency loops. User selectable config items should not be selected 
by another config item. If the driver really has a hard dependency on the 
regulator framework then 'depends on REGULATOR' is the right thing to do.

- Lars