Re: [U-Boot] [PATCH v6 11/20] tegra: fdt: Add clock bindings for Tegra2 Seaboard

[Albert ARIBAUD <albert.u.boot-LhW3hqR2+23R7s880joybQ@public.gmane.org>]

Le 28/02/2012 19:46, Simon Glass a écrit :
> Hi Stephen,
>
> On Tue, Feb 28, 2012 at 10:41 AM, Stephen Warren<swarren-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>  wrote:
>> Simon Glass wrote at Tuesday, February 28, 2012 11:37 AM:
>>> On Tue, Feb 28, 2012 at 10:31 AM, Stephen Warren<swarren-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>  wrote:
>>>> Simon Glass wrote at Tuesday, February 28, 2012 10:38 AM:
>>>> ...
>>>>> I am going to add your binding, less the #clock-cells which U-Boot
>>>>> currently can't support because it conflicts with the C preprocessor
>>>>> (at some point I may look at a patch to use sed or some other means of
>>>>> avoiding this).
>>>>
>>>> Out of curiosity, why does the C preprocessor come into it? Is U-Boot's
>>>> build process running cpp on the .dts files or something? That's non-
>>>> standard, although perhaps it could be a useful standard...
>>>
>>> Yes, but at the moment we only use it for '/include/ ARCH_CPU_DTS'.
>>
>> Uggh. That's going to make the device tree files look different between
>> the kernel and U-Boot:-( With # disallowed in particular, it's going to
>> prevent U-Boot from /ever/ using the correct protocols for parsing the
>> device tree. This seems like an extremely bad idea.
>
> Until we change it in U-Boot, you mean. We could move to sed or pre-
> and post-process the file to remove and re-insert the #.

Rather, to convert # signs into something that the DTS cannot contain 
and the compiler can withstand (and it should be printable ASCII, too). 
is '##' a good candidate?

If so, a forward conversion would e.g. map '/include/' to '#include' and 
any '#' to '##', and the reverse conversion would turn all '##' to '#'.

But something that simple is bound to be wrong in some way...

Amicalement,
-- 
Albert.