On Thu, Jan 05, 2017 at 10:50:33AM +0100, Quentin Schulz wrote:
> >>>> +static int axp20x_remove(struct platform_device *pdev)
> >>>> +{
> >>>> +       struct axp20x_adc_iio *info;
> >>>> +       struct iio_dev *indio_dev = platform_get_drvdata(pdev);
> >>>> +
> >>>> +       info = iio_priv(indio_dev);
> >>>
> >>> Nit: you could just reverse the 2 declarations above and join this
> >>> line after struct axp20x_adc_iio *info;
> >>>
> >>>> +       regmap_write(info->regmap, AXP20X_ADC_EN1, 0);
> >>>> +       regmap_write(info->regmap, AXP20X_ADC_EN2, 0);
> >>>
> >>> The existing VBUS power supply driver enables the VBUS ADC bits itself,
> >>> and does not check them later on. This means if one were to remove this
> >>> axp20x-adc module, the voltage/current readings in the VBUS power supply
> >>> would be invalid. Some sort of workaround would be needed here in this
> >>> driver of the VBUS driver.
> >>>
> >>
> >> That would be one reason to migrate the VBUS driver to use the IIO
> >> channels, wouldn't it?
> > 
> > It is, preferably without changing the device tree.
> > 
> 
> Yes, of course.

Note that you can have something like a test for the iio channel
property in the VBUS driver, and keep the old behaviour in the case
where we wouldn't have it.

Maxime

-- 
Maxime Ripard, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com