[PATCH] clk: new driver for efm32 SoC

[Uwe Kleine-König]

Hello Mike,

On Mon, Oct 07, 2013 at 02:23:54PM -0700, Mike Turquette wrote:
> Quoting Uwe Kleine-König (2013-08-14 13:04:53)
> > Signed-off-by: Uwe Kleine-König <u.kleine-koenig at pengutronix.de>
> > ---
> > Hello,
> > 
> > I'm not sure if there is a naming convention for #defines in headers below
> > dt-bindings. I named my clocks clk_$namefromreferencemanual to have some
> > namespacing.
> > 
> > The clock stuff on the different variants of efm32 are similar, but
> > different enough to not be able to use the same driver, so this is only
> > for the Giant Gecko (efm32gg) variant, still I used "efm32" in some
> > places because the other variants would fit there, too. (For
> > include/dt-bindings/clock/efm32-cmu.h I would need some #ifdefs, is that
> > OK or should the name of the defines include efm32gg? (That would make
> > the names still longer and so more ugly.))
> 
> The DT experts should weigh in on having #ifdef's in the header, but I
> think we should avoid that. Better to explicitly list the clocks for the
> variant used. You can add "gg" to this or not, I don't care, but the
> future variants will need some suffix, or maybe their own header.
I thought the way to go is to move the clocks into the device tree. That
would get rid of the need to add #ifdefs. OTOH even if the corresponding
clocks are using different register offsets and bit locations, I don't
see a need to also take a different clock index, so currently I think it
should be possible to not need #ifdefs even if we're going with this
patch's approach.

So if you are happy with my patch as is, it would be great if you'd take
it. We could also switch to in-dt-defined clocks later when it becomes
easier while maintaining this interface for compatibility.

> I'm happy with this patch and can take it for 3.13. Care to add a
> changelog?
What do you want? I could add:

	This patch adds support for the clocks provided by the Clock
	Management Unit of Energy Micro's efm32 Giant Gecko SoCs
	including device tree bindings.

I will send a v2 that makes use of the CLK_OF_DECLARE macro and so saves
the external init function.

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |