Re: [PATCH 3/4] hwmon: Add a simple driver to read the MXS SoC temperature

[Lars-Peter Clausen <lars@metafoo.de>]

On 06/27/2013 09:26 PM, Alexandre Belloni wrote:
> Hi,
> 
> On 27/06/2013 16:27, Guenter Roeck wrote:
>> On Thu, Jun 27, 2013 at 11:17:32AM +0200, Maxime Ripard wrote:
>>> On Wed, Jun 26, 2013 at 07:39:27AM -0700, Guenter Roeck wrote:
>>>> On Wed, Jun 26, 2013 at 10:51:12AM +0200, Alexandre Belloni wrote:
>>>>> The low resolution ADC of the mxs is able to read an internal temperature
>>>>> sensor, expose that using hwmon.
>>>>>
>>>>> Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
>>>>> ---
>>>> Wouldn't it make more sense to use iio-hwmon and improve it if necessary ?
>>> Actually, I wonder if we should not just put the hwmon driver
>>> capabilities directly into the mxs-lradc driver, just like it's already
>>> been done in this driver for the touchscreen support.
>>>
>>> The probing of this hwmon driver doesn't really belong to the DT, it's
>>> not really realistic to probe it from the machine definition, and it
>>> really is the IP that is wired that way.
>>>
>> Merging iio-hwmon functionality into an adc driver seems just as bad
>> (or even worse) as copying it into a new driver.
>>
>> If the lradc driver knows that the ADC channels are temperature sensors, it
>> should register them with the iio subsystem as IIO_TEMP type. Then you should
>> be able to use iio_hwmon as is.
> 
> They are already registered as IIO_TEMP but only implement read_raw. Also,
> 
> iio_hwmon_read_val() is using iio_read_channel_processed() and that will
> basically only read one of the 2 channels. As I documented, you actually
> need to read both channel 8 and channel 9 and then compute the value in
> Kelvins. I'm not sure how you want me to do that in the current framework.

What are these two channels actually measuring? Is the value of a single
channel meaningful on it's own? If not it might make sense to update the IIO
driver to just have one temperature channel.

- Lars