On 08/27/14 11:24, Bjorn Andersson wrote:

On Tue, Jul 29, 2014 at 11:06 PM, Stephen Boyd [off-list ref] wrote:

quoted

On 07/29, Rob Herring wrote:

[..]

quoted

quoted

You might as well do of_property_read_u32 in the below example.

Fair enough. The example is probably too simple. Things are
sometimes slightly more complicated and a simple
of_property_read_u32() isn't going to work in the case of
multiple reg values or when reg-names is in play.

But do we have such cases in the Qualcomm PMICs?

At the least we have reg-names users.

The only case I've hit so far is for gpios and mpps, where it feels
like reg should be base, size and not simply base reg of first gpio -
but that's a different thing.

Also, so far it seems like most drivers just code the base address in
the driver, as we have very specific compatibles.

It would be nice if the drivers didn't have to do this. It annoys me
that platform drivers need to know that they're using DT to figure out
things that are standard platform features: registers, irqs, etc.

How about we stop trying so hard to make this "perfect" and just merge
something close to Josh's original proposal and ignore this problem?
Currently all we're doing is delaying any possibility of getting
drivers for the individual blocks merged.
If we have the dt bindings require the reg to be there, we can discuss
and change this all we want later!

What's Josh's original proposal? We've already punted on this for SSBI
PMICs and just required them to put registers in DT for use some day and
I don't see anyone blocking individual SPMI pmic drivers from merging
over this. So I guess I agree with you that we can move forward with the
other drivers. I'd still like to see us make this better though, so it
seems worthwhile to have the discussion and get to some conclusion.

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation