On Sat, May 19, 2012 at 10:44:00PM +0530, Laxman Dewangan wrote:
> On Saturday 19 May 2012 10:11 PM, Mark Brown wrote:
> >On Sat, May 19, 2012 at 07:44:06PM +0530, Laxman Dewangan wrote:

> >>When registering the regulator driver, use the rdev->dev for
> >>getting the regulator device of given supply instead of parent
> >>device.

> >You're providing no motivation for this and it's difficult to see how it
> >improves things.  The class device is dynamically numbered so it's not
> >suitable for specifying supplies on a non-DT system and for a DT system
> >it's not obvious to me that we would want to involve the class device in
> >anything, it requires an additional layer of indirection but that's
> >about it.

> If I dont do this then it will not enter in the following case for
> getting the regulator_dev of supply regulator because dev->of_node
> is null, the tps65910-regulator driver have not set the
> pdev->dev.ofnode.

But why, why should it do that and how is this related to your patch?
You're *really* not explaining anything clearly with what you're doing
with device tree...  it's not clear what you're trying to do here or
that you've understood what's there currently.