On Sun, May 20, 2012 at 02:43:18AM +0530, Laxman Dewangan wrote:

> For mapping, the node should start from "regulators", not from pmu
> on this example.

What makes you say this?  I'm really not even sure what it means.
How does a node "start" from something?  Supply mappings are direct
links between consumers and regulators.

> Here my understanding is that config->of_node should contain the
> node information of the regulator being registered only. In DT case,
> it should not be null.

Right, but this is unrelated to what we're doing when the regulator is a
consumer.  Then we just do the same thing as regulator_get(dev, name).

> >I still don't see any change needed here, from the above it simply looks
> >like the supplies aren't set up.

> Unfortunately,
> My regulator_get is failing if I dont correct the above logic to
> have proper config.of_node.

But this seems like it is unrelated to the patch we're discussing!  Your
patch does nothing to config.of_node, it changes the device used to look
up the supply.  To repeat yet again:

| context of the class device we create.  I can't think of any situation
| where I'd expect that to make matters any better - the class device
| should certainly never appear in the device tree and isn't going to have
| a stable name for non-DT systems either.

*Please* engage with this, especially the non-DT part.  You need to
explain how what you're saying is related to the patch you posted, you
keep talking about a "proper" config.of_node and saying this happens to
make your system work but this isn't visibily related to the patch you
posted.

What is not "proper" about the of_node that was supplied for the
regulator being registered?  In what way is this related to the device
used by the regulator functioning as a consumer to request a supply?