[PATCH v2 06/10] mmc: omap_hsmmc: add support for pbias configuration in dt

Balaji T K balajitk at ti.com
Fri Jun 14 01:01:52 EST 2013


On Thursday 13 June 2013 04:17 PM, Lee Jones wrote:
> On Thu, 13 Jun 2013, Linus Walleij wrote:
>
>> On Thu, Jun 6, 2013 at 9:14 PM, Balaji T K <balajitk at ti.com> wrote:
>>
>>> PBIAS register configuration is based on the regulator voltage
>>> which supplies these pbias cells, sd i/o pads.
>>> With PBIAS register address and bit definitions different across
>>> omap[3,4,5], Simplify PBIAS configuration under three different
>>> regulator voltage levels - O V, 1.8 V, 3 V. Corresponding pinctrl states
>>> are defined as pbias_off, pbias_1v8, pbias_3v.
>>>
>>> pinctrl state mmc_init is used for configuring speed mode, loopback clock
>>> (in devconf0/devconf1/prog_io1 register for omap3) and pull strength
>>> configuration (in control_mmc1 for omap4)
>>>
>>> Signed-off-by: Balaji T K <balajitk at ti.com>
>>
>> You *need* Lee Jones and Mark Brown to review this.
>> Maybe Laurent has something to add too.
>>
>> Ux500 had the very same thing, and there this was solved using
>> a GPIO regulator for "vqmmc" a level-shifter. I vaguely remember
>> Laurent doing something similar with the SH stuff.
>
> I haven't seem much of this patch-set, but this certainly looks like
> it should be handled by a GPIO regulator instead of pinctrl. States
> are easily declared in a 'struct gpio_regulator_state', which the
> framework then uses to set the correct pins for the required voltage.
>

Thanks for the pointer, but wondering why is it named as gpio-regulator
and how it is different from fixed-regulator.
After going through git log description, I understand that voltage/current level
for a particular regulator is controlled by a set of pad/pin on the POWER IC
and pad/pin may be usually connected to gpio pins if it is needs to be
configurable and ground/pulled for constant voltage.

Collection of gpios logic level are modeled as state for particular voltage.
But gpio is not used in my case.

> And yes, 'vqmmc' is a good place to store the this regulator.
>



More information about the devicetree-discuss mailing list