[RESEND] [PATCH] Better vxWorks support in longlong.h

[Mark Haigh]

Kevin Ryde wrote:

>Mark Haigh <mhaigh at sonicwall.com> writes:
>  
>
>>However, this usage is ambiguous because vxWorks always defines
>>PPC, regardless of the actual target architecture.
>>    
>>
>
>Oh, it's a constant one compares CPU_FAMILY to is it?
>
>  
>
Yep, that's correct.

>>The patch here is as minimal and uninvasive as possible.
>>    
>>
>
>Thanks, but in the next release we're going to drive it from the
>configured --host, so as to avoid compiler and/or system nonsense.
>
>In the meantime, is it enough just to remove the "defined(PPC)" for a
>non-PPC vxworks?
>
>  
>
That works.

Slightly better and nearly as painless is removing defined(PPC), as you 
suggest, and adding checks for defined(__PPC__)  (GCC 3.x PPC Linux) and 
defined(__ppc) (GCC 3.4 PPC VxWorks).

Then you simultaneously fix all non-PPC vxWorks builds, and keep support 
for PPC working for GCC 3.x PPC Linux and GCC 3.4 PPC VxWorks.  The only 
caveat is 2.x GCC PPC Linux.

That fix would be great, as it could be put into the GCC 3.4 tree also.


Thanks,

Mark F. Haigh
mfhaigh at acm.org