Re: [PATCH 5/6] CPM_UART: unify clock sources

[Kumar Gala <galak@kernel.crashing.org>]

On Jun 24, 2006, at 10:21 AM, Vitaly Bordug wrote:

> Kumar,
>
> Thanks for feedback! My comment below.
>
> On Sat, 24 Jun 2006 08:46:54 -0500
> Kumar Gala <galak@kernel.crashing.org> wrote:
>
> [snip]
>>> +
>>> +#ifndef FS_PD_H
>>> +#define FS_PD_H
>>> +
>>> +#define GET_BAUDRATE()	(((bd_t *) __res)->bi_baudrate ?
>>> ((bd_t *) __res)->bi_baudrate : -1)
>>> +#define FS_UART_CLK()	(((bd_t *) __res)->bi_intfreq)
>>> +
>>> +#endif
>>
>> We shouldn't hide these differences in a header like this.  As we
>> slowly try to removing things from arch/ppc.  Why not just add
>> platform data for these two pieces of data and set it up like we do
>> other platform data.
>>
> That was my first guess as well, but:
> The aim is yet moving to powerpc not to break existing ppc/ stuff.
> So if alternatively platform_data may be utilized, all relevant ppc
> stuff that uses cpm uart should be updated. While doing the driver
> "platformize" trick, I made it possible to keep legacy behaviour
> (pd-less), this is not going to afford such a thing.
>
> IOW, I don't see much sense in updating ppc/ BSP files, just to keep
> the right way. It does not look very neat now, but does the job of
> moving FW and keeping existing stuff sane. If there is better approach
> envisioned, I'll definitely follow..

Fair enough, then I would suggest getting ride of all this header/ 
MACRO indirection and just implement these as functions with  
different implementations in arch/powerpc vs arch/ppc.

- k