Re: [PATCH] Mach-O: Remove ambiguity of mach-o-le and mach-o-x86-64

[Iain Sandoe <developer at sandoe-acoustics dot co dot uk>]

Hi...

On 15 Dec 2011, at 18:26, shinichiro hamaji wrote:

> When we use objdump for x86-64 mach-o files without explicit -b flag,
> it says "File format is ambiguous" because there are two matching
> formats (mach-o-le and mach-o-x86-64).

I was wondering recently whether there is any value to retaining this  
generic mach-o-le when we have specific vecs for the actual targets?

(I have a local implementation for the powerpc* vecs, to post at some  
stage too).
Iain

> It's especially not useful when
> we run objdump for a fat binary which contains x86-64. Actually, I
> couldn't find any way to dump contents of x86-64 part of a fat binary.
> It seems i386 is OK as i386 mach-o doesn't match mach-o-le. This tiny
> patch does the same thing for x86-64:
> http://shinh.skr.jp/t/mach-o-64-le.patch
>
> bfd/
> 2011-12-16  Shinichiro Hamaji  <shinichiro.hamaji@gmail.com>
>
> 	* mach-o.c (bfd_mach_o_header_p): Don't handle x86-64 mach-o
> 	when cputype is not specified. This case should be handled by
> 	mach-o-x86-64.c.
> diff --git a/bfd/mach-o.c b/bfd/mach-o.c
> index 325cf61..a455b09 100644
> --- a/bfd/mach-o.c
> +++ b/bfd/mach-o.c
> @@ -3266,7 +3266,8 @@ bfd_mach_o_header_p (bfd *abfd,
>       switch (header.cputype)
>         {
>         case BFD_MACH_O_CPU_TYPE_I386:
> -          /* Handled by mach-o-i386 */
> +        case BFD_MACH_O_CPU_TYPE_X86_64:
> +          /* Handled by mach-o-i386 or mach-o-x86-64 */
>           goto wrong;
>         default:
>           break;