Re: [PATCH v4 2/2] iio: mcp4531: Driver for Microchip digital potentiometers

[Lars-Peter Clausen]

On 09/23/2015 09:39 AM, Peter Rosin wrote:
> On 2015-09-22 20:22, Jonathan Cameron wrote:
>> On 22 September 2015 16:36:49 BST, Peter Rosin <peda@xxxxxxxxxxxxxx> wrote:
> 
> *snip* lots of cosmetics that I'll fix in the next spin...
> 
>>> +
>>> +static int mcp4531_read_raw(struct iio_dev *indio_dev,
>>> +			    struct iio_chan_spec const *chan,
>>> +			    int *val, int *val2, long mask)
>>> +{
>>> +	int err;
>>> +
>>> +	switch (mask) {
>>> +	case IIO_CHAN_INFO_RAW:
>>> +		err = mcp4531_get_value(indio_dev, chan->address, val);
>>> +		if (err < 0)
>>> +			return err;
>>> +		return IIO_VAL_INT;
>>> +	}
>> No scale known? Be nice to relate to real world values.  I gather this may need
>>  platform data of some type (can't open data sheet on phone for some reason)
> 
> ...but I need more input on this.
> 
> Yes, it would be nice and yes it requires platform data, as there is
> no way to retrieve if the driver is dealing with a 5, 10, 50 or 100 kOhms
> pot/rheostat.

You should use the full product name in the ID table for device, it includes
what the resistance is.


> You would also have to expose both the resistance between
> A to W and between W to B, as you have no way to know how it's been
> connected (or you'd need more platform data). Ok, for rheostats you
> know how it's connected, since A isn't available, but for pots there
> are two resistance values controlled with a single value X i.e.
> X*scale/max and (max-X)*scale/max (ignoring the 75 Ohm wiper resistance).

I don't think that is a problem. The scale is still the same for both of
them. Even if you don't provide the scale the user still has to know which
of the terminals uses the inversed and which one uses the non-inversed raw
value for their resistance. So nothing really changes in that regard by
providing the scale or not.

> But, I'd rather not go there, as I don't need any of it. Is it Ok to not
> provide scaling?

Having scaling is not required, but strongly encouraged. It allows creating
generic userspace applications that do not require specific knowledge about
the part they are talking to. Which is one of the main points for having a
framework. Otherwise you could just go with a driver with a custom ABI, or
even implement all the interface logic in userspace using /dev/i2c.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/