Re: [PATCH 3/3] hwmon: (ina2xx) Add support for ina260

[Michal Simek <michal.simek@xilinx.com>]

On 19. 05. 20 16:14, Guenter Roeck wrote:
> On 5/18/20 10:21 PM, Michal Simek wrote:
>> On 26. 02. 20 3:16, Guenter Roeck wrote:
>>> On 2/24/20 3:26 PM, Franz Forstmayr wrote:
>>>> Add initial support for INA260 power monitor with integrated shunt.
>>>> Registers are different from other INA2xx devices, that's why a small
>>>> translation table is used.
>>>>
>>>> Signed-off-by: Franz Forstmayr <forstmayr.franz@gmail.com>
>>>
>>> I think the chip is sufficiently different to other chips that a separate
>>> driver would make much more sense than adding support to the existing
>>> driver.
>>> There is no calibration, registers are different, the retry logic is
>>> not needed. A new driver could use the with_info API and would be much
>>> simpler while at the same time not messing up the existing driver.
>>
>> Isn't it also better to switch to IIO framework?
>> As we discussed in past there are two ina226 drivers. One in hwmon and
>> second based on IIO framework (more advance one?) and would be good to
>> deprecate hwmon one.
> 
> "More advanced" is relative. The ina2xx driver in iio doesn't support
> alert limits (which is queued in the hwmon driver for 5.8), and the
> iio->hwmon bridge doesn't support it either. On top of that, there are
> existing users of the hwmon driver, which would have to be converted
> first. As for ina260, it would be up to the implementer to determine
> if alert limit support is needed or not, and which API would be
> appropriate for the intended use case.

Good to know. If ina260 is done as separate driver I am fine with it.

Thanks,
Michal