[PATCH] sparc64-linux --with-cpu=ultrasparc fix

[Dan Kegel]

Jakub Jelinek wrote:
>>What's the right way to get gcc to default to building 64 bit executables
>>when targeting sparc64-linux [and --with-cpu=ultrasparc3]?
> 
> 
> There was TARGET_CPU_ultrasparc3 missing in linux64.h.
> I've commited the following fix to the trunk, but as it is not a regression,
> I'm not sure if Mark is ok with this for gcc-3_4-branch.
> 
> 2004-05-01  Jakub Jelinek  <jakub@redhat.com>
> 
> 	* config/sparc/linux64.h (TARGET_DEFAULT): Make 64-bit by default
> 	also for TARGET_CPU_DEFAULT == TARGET_CPU_ultrasparc3.
> 
> --- gcc/config/sparc/linux64.h.jj	2004-05-01 22:59:52.000000000 +0200
> +++ gcc/config/sparc/linux64.h	2004-05-01 23:00:41.126176529 +0200
> @@ -39,7 +39,9 @@ Boston, MA 02111-1307, USA.  */
>  #undef MD_EXEC_PREFIX
>  #undef MD_STARTFILE_PREFIX
>  
> -#if TARGET_CPU_DEFAULT == TARGET_CPU_v9 || TARGET_CPU_DEFAULT == TARGET_CPU_ultrasparc
> +#if TARGET_CPU_DEFAULT == TARGET_CPU_v9 \
> +    || TARGET_CPU_DEFAULT == TARGET_CPU_ultrasparc \
> +    || TARGET_CPU_DEFAULT == TARGET_CPU_ultrasparc3
>  /* A 64 bit v9 compiler with stack-bias,
>     in a Medium/Low code model environment.  */

Ah, thanks.  That got me past the problem.  Sadly, I seem to have run
into a new one, this time gcc-3.4.0 seems to be generating a bad
floating point register name when building glibc; I've filed a bug at
http://gcc.gnu.org/PR15247
with a minimal test case.  (Hope I didn't miss a known fix; I did look
through the gcc bug database.)
- Dan

-- 
My technical stuff: http://kegel.com
My politics: see http://www.misleader.org for examples of why I'm for regime change