Re: [patch] x86_64 darwin work

[Geoffrey Keating <geoffk at apple dot com>]

On 07/09/2006, at 8:45 PM, Eric Christopher wrote:

> #define JUMP_TABLES_IN_TEXT_SECTION \
> -  (!TARGET_64BIT && flag_pic && !HAVE_AS_GOTOFF_IN_DATA)
> +  (flag_pic && ((TARGET_MACHO && TARGET_64BIT) \
> +   || (!TARGET_64BIT && !HAVE_AS_GOTOFF_IN_DATA)))

Indenting.

> @@ -48,13 +74,18 @@ Boston, MA 02110-1301, USA.  */
>    %{g: %{!fno-eliminate-unused-debug-symbols: -feliminate-unused- 
> debug-symbols }}"
>
>  #undef ASM_SPEC
> -#define ASM_SPEC "-arch i386 -force_cpusubtype_ALL"
> +#define ASM_SPEC "%{m64: -arch x86_64 -force_cpusubtype_ALL} \
> +                  %{!m64: -arch i386 -force_cpusubtype_ALL}"

This should use '-arch %(darwin_arch)' rather than replicating that  
logic.  In fact, that's what %(darwin_arch) is for; look at the  
default in config/darwin.h.

> +#define DARWIN_ARCH_SPEC "%{m64:x86_64;:i386}"
> +#define DARWIN_SUBARCH_SPEC "                                   \
> +  %{m64: x86_64}                                                \
> +  %{!m64: i386}"

Just do
#define DARWIN_SUBARCH_SPEC DARWIN_ARCH_SPEC

> +/* Functions shipped in the ppc64 version of libgcc_s.1.dylib

'in the ppc64 and x86_64'

> +#if defined (__ppc64__) || defined (__x86_64__)
> +/* Many of these functions have probably never been used by anyone
> +   anywhere on ppc64, but it's hard to prove this, so they're defined

'anywhere on these targets'

With those changed, OK.

Attachment: smime.p7s
Description: S/MIME cryptographic signature