Re: [RFC 2/2] hwmon: lm90: add thermal_zone temperature sensor support

[Guenter Roeck <linux@roeck-us.net>]

On Mon, Feb 06, 2017 at 05:01:31PM +0100, Christian Lamparter wrote:
> On Sunday, February 5, 2017 7:10:53 PM CET Guenter Roeck wrote:
> > On 02/05/2017 01:03 PM, Christian Lamparter wrote:
> > > This patch adds thermal_zone temperature sensor support
> > > to the lm90 module. The feature has to be enabled
> > > separately via the Kconfig option:
> > > 	CONFIG_SENSORS_LM90_THERMAL_DEVICE
> > >
> > > The LM90 supports two (three for MAX6695 and MAX6696)
> > > temperature sensors. The local sensor is integrated
> > > into the LM90 chip. The remote sensors are connected
> > > to external temperature sensing diodes.
> > >
> > > Cc: Wei Ni <wni@nvidia.com>
> > > Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
> > 
> > Since the hwnmon core now supports thermal registration, and since the lm90
> > driver has already been converted to using the new hwmon API, I would
> > rather like to understand why using the hwmon core for this purpose is
> > insufficient, and I would prefer to address the deficiencies in the hwmon
> > core and not in the driver.
> 
> Hey, that's great, I completely missed that! Yes, this makes the changes
> to lm90.c obsolete.
> 
> However, what about the device-tree updates in 1/2? I'm asking because
> the hwmon device node still needs the #thermal-sensor-cells property
> defined for thermal_zone_of_sensor_register(). Without it, the sensor
> will be skipped and the thermal-zones will do nothing (no regulation).
> I can respin the device-tree patch (local and remote need to trade 
> places). What do you think?
> 
Sure, no problem with that. Does supporting this require changes in the hwmon
core ?

Thanks,
Guenter