On Sun, May 20, 2012 at 04:09:43PM +0530, Laxman Dewangan wrote:
> On Sunday 20 May 2012 02:31 PM, Mark Brown wrote:

> >No.  This is happening because the device tree doesn't have any supplies
> >mapped for the regulators.  This is nothing at all to do with where the
> >code looks for the supplies, no matter where it looks there's nothing to
> >find.

> No, we should not put the regulator mapping under parent, need to
> have under "regulator" otherwise we need to fix the issue in dt
> parsing where first it looks for "regulator" and then parse the rail
> mapping.

What is this issue and why should we not fix it?

> Now when compare to driver mc13892-regulator.c, the
> tps65910-regulator is almost same like this.
> The driver mc13892-regulator.c have following code in probe:

...

> I want to have similar fix in my tps65910-regulator.c.

So why can't you do what mc13892 is doing?

> I am sorry that I am not able to explain the issue correctly. I think
> I will take help from Stephen Warren here to first explain him  and
> then I will come back for core changes.

OK, I guess.  I think a key thing here is that these shouldn't be any
different to any other supply.  Adding something that is specific to
regulator-regulator supplies doesn't do that so is a clear sign that
something has been missed.