On Thu, 19 Oct 2006 13:01:33 +0900 (JST)
Atsushi Nemoto <anemo@xxxxxxxxxxxxx> wrote:
> On Fri, 13 Oct 2006 14:39:01 +0200, Franck Bui-Huu <vagabon.xyz@xxxxxxxxx> wrote:
> > +#if defined(CONFIG_64BITS) && !defined(CONFIG_BUILD_ELF64)
> > +#define __page_offset(x) ((unsigned long)(x) < CKSEG0 ? PAGE_OFFSET : CKSEG0)
>
> CONFIG_64BIT, not CONFIG_64BITS. Sorry, my mistake.
>
> Also since CKSEG0 is defined with _LLCONST_ macro, the final type of
> __page_offset(), __pa(), __pa_sym() will be "unsigned long long", not
> "unsigned long". This raise a "comparison of distinct pointer types
> lacks a cast" warning on this line.
>
> reserved_end = max(init_initrd(), PFN_UP(__pa_symbol(&_end)));
>
> A qiuck and non-intrusive hack would be cast CKSEG0 with "unsigned
> long" here, but it might be preferred to change _LLCONST_ definition
> like this. What do you think?
>
>
> Subject: Use "long" for _ATYPE64_ and _LLCONST_ on 64-bit kernel.
>
> Signed-off-by: Atsushi Nemoto <anemo@xxxxxxxxxxxxx>
>
> diff --git a/include/asm-mips/addrspace.h b/include/asm-mips/addrspace.h
> index 45c706e..5005555 100644
> --- a/include/asm-mips/addrspace.h
> +++ b/include/asm-mips/addrspace.h
> @@ -23,9 +23,14 @@ #define _LLCONST_(x) x
> #else
> #define _ATYPE_ __PTRDIFF_TYPE__
> #define _ATYPE32_ int
> +#ifdef CONFIG_64BIT
> +#define _ATYPE64_ long
> +#define _LLCONST_(x) x ## L
^^ ^
The name is not corresponding to reality.
It's not so good.
Yoichi
