ppc64 build failure - asm code - bswap

[Dominik 'Rathann' Mierzejewski]

On Tuesday, 23 September 2008 at 15:22, David Timms wrote:
> Adrian Reber wrote:
> >On Mon, Sep 22, 2008 at 12:36:11AM +1000, David Timms wrote:
> >>So it seems that bswap might not suit ppc64 hardware. If that is the  
> >>case then:
> >># ppc64 build does not seem to support asm opcode bswap
> >>ExcludeArch: ppc64
> >>would seem appropriate an appropriate addition to the .spec. Does that  
> >>make sense ?
> >>
> >>Note that it did build OK on i386|x86_64|ppc. I would have imagined that  
> >>ppc64 opcodes would be a superset of ppc, so there might be something  
> >>else going on ?
> >
> >The following is the your problem:
> >
> >%ifarch ppc x86_64
> >ASM_X86=0
> >%else
> >ASM_X86=1
> >%endif
> >
> >
> >You forgot to mention ppc64 in that list and ppc64 can just not handle
> >intel assembler just like ppc. 
> I don't have much experience with this sort of thing. Is x86_64 not able 
> to do x86 assembler ?

If you use the 32bit legacy mode, yes, but you can't use ia32 asm on amd64
without some changes.

> Given this was a package import that I didn't create the gcube.spec 
> initially, should the "ifarch ppc x86_64" have been "ppc ppc64" ?

I think it is better to do it the other way, that is:
%ifarch %{ix86}
ASM_X86=1
%else
ASM_X86=0
%endif
You don't have to worry about alpha, sparc(64) or ia64 that way.

Regards,
R.

-- 
Fedora http://fedoraproject.org/wiki/User:Rathann
RPMFusion http://rpmfusion.org | MPlayer http://mplayerhq.hu
"Faith manages."
        -- Delenn to Lennier in Babylon 5:"Confessions and Lamentations"