SOURCES: glibc-ppc-include-mathinline.patch (NEW) - fix build on P...

[Jakub Bogusz]

On Sun, Oct 08, 2006 at 09:24:58PM +0200, sparky at pld-linux.org wrote:
> On Sun, Oct 08, 2006 at 08:34:44PM +0200, Jakub Bogusz wrote:
> > On Fri, Oct 06, 2006 at 07:34:12PM +0200, sparky at pld-linux.org wrote:
> > > On Fri, Oct 06, 2006 at 07:26:20PM +0200, sparky wrote:
> > > > Author: sparky                       Date: Fri Oct  6 17:26:20 2006 GMT
> > > > Module: SOURCES                       Tag: HEAD
> > > > ---- Log message:
> > > > - fix build on PowerPC
> > > 
> > > I can't understand why build passed on th-ppc. For me on forge and
> > > gzowski it failed with same problem:
> > > 
> > > ../sysdeps/powerpc/fpu/e_sqrt.c: In function '__ieee754_sqrt':
> > > ../sysdeps/powerpc/fpu/e_sqrt.c:173: error: '__CPU_HAS_FSQRT' undeclared (first use in this function)
> > > 
> > > maybe th builder is not using appropiate setarch ?
> > 
> > No, in case of such issue __CPU_HAS_FSQRT would be just defined to 1.
> > 
> > What flags did you use? -O0 or -Os?
> > __USE_EXTERN_INLINES is defined in features.h with the following condition:
> > 
> > #v+
> > /* Decide whether we can define 'extern inline' functions in headers.  */
> > #if __GNUC_PREREQ (2, 7) && defined __OPTIMIZE__ \
> >     && !defined __OPTIMIZE_SIZE__ && !defined __NO_INLINE__
> > # define __USE_EXTERN_INLINES   1
> > #endif
> > #v-
> 
> I see. I was using -Os.
> But my fix is probably not the best one,

__USE_EXTERN_INLINES shouldn't be defined when compiling with -Os

> perhaps __CPU_HAS_FSQRT
> definicions should be moved somewere else (to math.h) ?

Not <math.h>, as it's system-independent file. Maybe <bits/mathdefs.h>.
You can try to report it to <libc-alpha at sources.redhat.com>.
But it's likely you'll get some "only -O2 is supported for glibc
compilation" reply.


-- 
Jakub Bogusz    http://qboosh.cs.net.pl/