Re: [PATCH v6 2/2] iio: adc: add support for Nuvoton NCT7201

[Andy Shevchenko <andy.shevchenko@gmail.com>]

Sun, Apr 20, 2025 at 09:03:05PM +0800, Yu-Hsian Yang kirjoitti:
> Andy Shevchenko <andriy.shevchenko@linux.intel.com> 於 2025年4月16日 週三 下午5:34寫道：
> > On Wed, Apr 16, 2025 at 04:17:34PM +0800, Eason Yang wrote:

...

> > > +#define NCT7201_REG_VIN(i)                           (i)
> >
> > This doesn't do anything useful. Why do you need this rather useless macro?
> >
> 
> Actually here we should define NCT7201_REG_VIN(i) as (0x00 + i),
> We simply it as (i).

Please, don't. Use the full form which makes a big difference to the perception
of this macro.

...

> > > +struct nct7201_chip_info {
> > > +     struct device *dev;
> >
> > This can be derived from the respective regmap. No need to have it here.
> >
> > > +     struct regmap *regmap;
> > > +     struct regmap *regmap16;
> > > +     int num_vin_channels;
> > > +     u16 vin_mask;
> > > +};
> 
> Use regmap->dev is okay if use regmap API.

No, you are not supposed to use regmap->dev (and you can't), you need to call a
getter API and get the device pointer.

> But if we need to print message not from regmap API,

I don't get this. What do you mean? The example you showed prints a message for
the device. The same device that was used to create a regmap.

> how suggestions to do in this case?

Use the one that you can retrieve from regmap.

...

So, you have commented on some with agreement, and left uncommented a lot. Does
it mean you are agree on all points? The rule of thumb do not comment on the cases
you are fully agree with. Current email just makes a confusion.

-- 
With Best Regards,
Andy Shevchenko