Re: [PATCH 2/2] ARM: dt: tegra: cardhu: register core regulator tps65911

[Stephen Warren <swarren@wwwdotorg.org>]

On 05/22/2012 11:56 AM, Laxman Dewangan wrote:
> On Tuesday 22 May 2012 10:49 PM, Stephen Warren wrote:
>> On 05/22/2012 11:09 AM, Laxman Dewangan wrote:
>>> On Tuesday 22 May 2012 10:10 PM, Stephen Warren wrote:
>>>> On 05/22/2012 07:05 AM, Laxman Dewangan wrote:
>>>>> Add device info for the PMIC device tps65911 in tegra-cardhu
>>>>> dts file. This device supports the multiple regulator rails,
>>>>> gpio, interrupts.
...
>>>> Nitpicky, but the labels might be more logical as reg_vdd1 rather than
>>>> vdd1_reg, but not a big deal.
>>>>
>>>> So, please replace the line above with:
>>>>
>>>>      reg_vdd1: regulator@0 {
>>>>          reg = <0>;
>>>
>>> Why do we really require the reg at all?
>>> I dont think any usage of doing this.

Oh, perhaps you meant the reg property not "reg_" in the label name?

It is required because the parent node has #address-cells and
#size-cells and because the node name itself has a unit address ("@nnn").

>> I guess if these regulators are enabled at boot and always on, we don't
>> even need the labels for now; we could add labels later as/when drivers
>> begin to dynamically control the regulators.
> 
> I think we should provide the label here whether it is always on or not.
> The driver who uses the rails will not aware that rail is always on or not.
> Second thing is that this gives uniformity and whenever any consumer get
> added, we will not touch this part, only will be change in the driver
> specific part.

Yes, if drivers are referring to these nodes, you do need the label.