Hi Chris,
On Fri, Mar 17, 2017 at 2:48 PM, Chris Brandt <Chris.Brandt@xxxxxxxxxxx> wrote:
> On Friday, March 17, 2017, Geert Uytterhoeven wrote:
>> So in the absence of an "rtc0" clock in the device node (you don't have
>> "clock-names" properties in the rtc devvice node yet), it will fall back
>> to clk_get_sys(), and will find the global "rtc0" clock. Unless you call
>> it "rtc"...
>>
>> Most drivers using a single clock just pass NULL instead of a name, so it
>> will match the first clock found.
>>
>> I think the simplest solution is to check if your device is instantiated
>> from DT (pdev->dev.of_node != NULL), and pass NULL (or "fck", when you add
>> multiple clocks to the DT bindings) to devm_clk_get() if that's the case.
>
> OK, so leave current SuperH code the way it is, but add the case for
> (pdev->dev.of_node != NULL) and do things the 'new way' so I can have
> an RTC clock name of "rtc".
Indeed.
> Question: Is your idea to add all the clocks (X1, X3, EXTAL), but put the
> first one as the peripheral clock (the one that runs the register interface 'p0')?
Which one is the peripheral clock doesn't matter if you use clock-names.
The convention is to use "fck" for the functional clock of the device.
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
