Re: [PATCH 2/5] clk: bcm2835: enable management of PCM clock

[Geert Uytterhoeven <geert@linux-m68k.org>]

Hi Arnd,

On Sat, Jan 9, 2016 at 9:56 PM, Arnd Bergmann <arnd@arndb.de> wrote:
> On Saturday 09 January 2016 09:25:54 kernel@martin.sperl.org wrote:
>> --- a/include/dt-bindings/clock/bcm2835.h
>> +++ b/include/dt-bindings/clock/bcm2835.h
>> @@ -44,5 +44,6 @@
>>  #define BCM2835_CLOCK_EMMC             28
>>  #define BCM2835_CLOCK_PERI_IMAGE       29
>>  #define BCM2835_CLOCK_PWM              30
>> +#define BCM2835_CLOCK_PCM              31
>>
>> -#define BCM2835_CLOCK_COUNT            31
>> +#define BCM2835_CLOCK_COUNT            32
>
> The last line contains an incompatible change, please don't do that.
> If you have to add another clock, do that after the BCM2835_CLOCK_COUNT
> definition to avoid changing dts files that use that number.

While I agree this changes dts files (in an unexpected way?), not updating
BCM2835_CLOCK_COUNT makes it definition useless. Which teaches that
having such definitions in DT headers is not a good idea in the first place...

Hence it can better be replaced (it seems to be unused in dts files, but you
can keep the definition to be 100% sure) by an ARRAY_SIZE() in the C driver.
This requires changing the driver to e.g. initialize clks[] in
bcm2835_clk_probe() based on a table instead of explicit code.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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