[RFC PATCH 1/4] regulator: Introduce OMAP regulator to control PMIC over VC/VP

Nishanth Menon nm at ti.com
Tue Jun 11 02:16:59 EST 2013


+Paul.

On Mon, Jun 10, 2013 at 5:31 AM, Mark Brown <broonie at kernel.org> wrote:
> On Wed, May 22, 2013 at 01:18:34PM -0500, Nishanth Menon wrote:
>
> So, the biggest problem here has been patch 4 (having to have a hack to
> deploy this stuff is a bit worrying) plus the general not having a real
> driver thing.
Patch #4 in this series was a hack as it was not properly split up and
organized as a proper DTS series -it was meant as a proof of concept -
not entirely meant to indicate the remaining 1-3 patches were hacks
:).

>
>> +- ti,i2c-slave-address - I2C slave address of the PMIC
>> +- ti,i2c-voltage-register - I2C register address where voltage commands are
>> +     to be set.
>> +- ti,i2c-command-register - I2C register address where commands are to be set
>> +     when OMAP enters low power state. This may be the same as
>> +     ti,i2c-voltage-register depending on the PMIC.
>> +- ti,slew-rate-microvolt - worst case slew rate of rise / fall for voltage
>> +     transition in microvolts per microseconds (uV/uS)
>> +- step-size-micro-volts - Step size in micovolts as to what one step in voltage
>> +     selector increment translates to. See example.
>> +- regulator-min-microvolt - Minimum voltage in microvolts which is supported by
>> +     the PMIC in ti,step-size-microvolt increments. See example.
>> +- regulator-max-microvolt - Maximum voltage in microvolts which is supported
>> +     by the PMIC in ti,step-size-microvolt increments. See example.
>
> The other thing is this whole business of encoding the properties of the
> PMIC in the DT like this.  Paul Walmsley has started doing some work for
> some similiar hardware where instead of doing this the regulator is in
> the DT as normal and then the driver for the offloaded voltage scaling
> gets the information about the register layout from the regulator
> driver.  This is a bit neater overall and would cope with determining
> which method to use at runtime.

I think you mean http://marc.info/?t=137059249100003&r=1&w=2 series. I
will dig into it. if it is possible for Tegra and OMAP to use the same
framework and strategy to deal with these kind of h/w blocks, all the
more better.

Regards,
Nishanth Menon


More information about the devicetree-discuss mailing list