Re: [PATCH v2] iio: max5487: Add support for Maxim digital potentiometers

From: Joachim Eastwood
Date: Sun Apr 10 2016 - 09:10:20 EST


On 10 April 2016 at 14:47, Joachim Eastwood <manabian@xxxxxxxxx> wrote:
> Hi Cristina,
>
> On 9 April 2016 at 10:24, Cristina Moraru <cristina.moraru09@xxxxxxxxx> wrote:
>> Add implementation for Maxim MAX5487, MAX5488, MAX5489
>> digital potentiometers.
>>
>> Datasheet:
>> http://datasheets.maximintegrated.com/en/ds/MAX5487-MAX5489.pdf
>>
>> Signed-off-by: Cristina Moraru <cristina.moraru09@xxxxxxxxx>
>> CC: Daniel Baluta <daniel.baluta@xxxxxxxxx>
>> ---
> ...
>> +static int max5487_read_raw(struct iio_dev *indio_dev,
>> + struct iio_chan_spec const *chan,
>> + int *val, int *val2, long mask)
>> +{
>> + struct max5487_data *data = iio_priv(indio_dev);
>> +
>> + if (mask != IIO_CHAN_INFO_SCALE)
>> + return -EINVAL;
>> +
>> + *val = 1000 * data->kohms;
>> + *val2 = MAX5487_MAX_POS;
>
> Newline before return.
>
>> + return IIO_VAL_FRACTIONAL;
>> +}
>> +
>> +static int max5487_write_raw(struct iio_dev *indio_dev,
>> + struct iio_chan_spec const *chan,
>> + int val, int val2, long mask)
>> +{
>> + struct max5487_data *data = iio_priv(indio_dev);
>> +
>> + switch (mask) {
>> + case IIO_CHAN_INFO_RAW:
>> + if (val < 0 || val > MAX5487_MAX_POS)
>> + return -EINVAL;
>> + return regmap_write(data->regmap, chan->address, val);
>> + default:
>> + return -EINVAL;
>> + }
>> + return -EINVAL;
>
> To be consistent with your max5487_read_raw() function you could do a:
> if (mask != IIO_CHAN_INFO_RAW)
> return -EINVAL;
>
>
>> +static const struct iio_info max5487_info = {
>> + .read_raw = &max5487_read_raw,
>> + .write_raw = &max5487_write_raw,
>
> Address operator should be unnecessary on functions.
>
>
>> + data->regmap = devm_regmap_init_spi(spi, &max5487_regmap_config);
>> + if (IS_ERR(data->regmap))
>> + return PTR_ERR(data->regmap);
>
> Nothing wrong with using regmap here, but since you are only using
> simple regmap_write()'s you might as well have used spi_write()
> directly. I am not telling you to switch, but I don't see the point of
> using regmap here.

Looking again: it seem that spi.h doesn't have a function that do
write(cmd, data) which regmap does. So I guess that is one reason for
using regmap. But it wouldn't be hard to create a write(cmd,
data)-function for spi either. Just wrap spi_write() and have a local
buf var. I am a bit surprised that spi.h doesn't have such a function
as it should be quite a common pattern for spi chips.

>
> Which reminds me; for regmap you need to select REGMAP_SPI in your
> Kconfig entry.
>
>
> regards,
> Joachim Eastwood