Re: [PATCH] include/asm-arm/arch-omap/omap24xx.h

[Dirk Behme <dirk.behme@googlemail.com>]

Paul Walmsley wrote:
> On Thu, 1 Nov 2007, Tony Lindgren wrote:
> 
> 
>>* Tony Lindgren <tony@atomide.com> [071031 08:22]:
>>
>>>* Paul Walmsley <paul@pwsan.com> [071025 09:01]:
>>>
>>>
>>>>Hello Dirk,
>>>>
>>>>On Thu, 25 Oct 2007, Dirk Behme wrote:
>>>>
>>>>
>>>>>Mmh, I like to fix it in include/asm-arm/arch-omap/omap24xx.h, so I prefer 
>>>>>[1].
>>>>>
>>>>>With [1] we
>>>>>
>>>>>- fix it where it was introduced
>>>>>- have an additional #ifdef only in a header file instead of 
>>>>>plat-omap/devices.c, which is IMHO better
>>>>>- fix it for everybody now and in the future who directly or indirectly 
>>>>>includes omap24xx.h.
>>>>
>>>>Hopefully this is just a workaround until we can get rid of these 
>>>>conditional defines for multiboot kernels.  Generally speaking, would you 
>>>>propose doing the same thing for the other conditional defines in 
>>>>include/asm-arm/arch-omap/omap24xx.h, such as OMAP2_32KSYNCT_BASE, 
>>>>OMAP2_SDRC_BASE, etc.?
>>>
>>>Sounds like the real fix is to set these addresses in the init code
>>>somewhere based on the omap type.
>  
> Yeah, agreed.

Me too.

>>I'll push a temporary fix based on Luis' patch, see below.
> 
> 
> OMAP3430 builds now generate redefinition warnings for OMAP2_CTRL_BASE; 
> patch to work around is below.
...
> From: Paul Walmsley <paul@pwsan.com>
> Subject: Work around OMAP2_CTRL_BASE warnings in 3430 build
> 
> Work around OMAP2_CTRL_BASE redefinition warnings for OMAP3430 builds. 
> 
> Signed-off-by: Paul Walmsley <paul@pwsan.com>
> 
> ---
>  include/asm/arch/omap24xx.h |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> Index: linux-omap/include/asm/arch/omap24xx.h
> ===================================================================
> --- linux-omap.orig/include/asm/arch/omap24xx.h	2007-11-04 20:55:00.000000000 -0700
> +++ linux-omap/include/asm/arch/omap24xx.h	2007-11-04 21:32:31.000000000 -0700
> @@ -101,7 +101,7 @@
>  #define OMAP2_VA_IC_BASE	IO_ADDRESS(OMAP24XX_IC_BASE)
>  #define OMAP2_CTRL_BASE		OMAP243X_CTRL_BASE
>  
> -#else
> +#elif !defined(CONFIG_ARCH_OMAP34XX)
>  /* FIXME: These will go away soon */
>  #define OMAP2_CTRL_BASE		L4_24XX_BASE
>  #endif

Until we have real fix like above: Instead of now trying to cover all 
situations where OMAP2_CTRL_BASE is undefined or redefinded (maybe 
there are some more?), I wonder why not simply make it self-dependent?

#if !defined(OMAP2_CTRL_BASE)
/* FIXME: These will go away soon */
#define OMAP2_CTRL_BASE		L4_24XX_BASE
#endif

Regards

Dirk

http://linux.omap.com/pipermail/linux-omap-open-source/2007-September/011393.html