Re: Glibc issue, was: Trying to include floating point trapping in fortran program

[Toon Moene <toon at moene dot indiv dot nluug dot nl>]

Toon Moene wrote:

> Duncan Galloway wrote:

> > I'm trying to compile a program and include the c code listed in the
> > document
> >
> >         http://gcc.gnu.org/onlinedocs/g77_20.html#SEC604
> >
> > This is supposed to trap floating point errors, overflows, underflows

> IIRC, this solution only worked with glibc-2.0.x; with glibc-2.1 and
> higher you need to do this in a different way.  Unfortunately, I do not
> have a glibc-2.1 system handy, so I can't explain how.
> 
> Hope someone else on this list can help,

I got the following suggestion from Alfredo Ferrari INFN - Milan in
private email:

<QUOTE>
the following piece of code works nicely on my RedHat systems with both 
Glibc2.0 and 2.1 in order to trap floating point exceptions in fortran
codes. 
I hope it is useful... It is a long time I am not trying it on a
Glibc2.0
system, I hope it is still ok, with RH6.0,6.1 and 6.2 it works like a
charm.

                    Cheers
                 Alfredo Ferrari


#include <features.h>
#if __GLIBC_MINOR__ > 0

#include <fenv.h>

#else

#include <fpu_control.h>

#endif

static void __attribute__ ((constructor))
     trapfpe ()
{
#if __GLIBC_MINOR__ > 0

  feenabletraps ((FE_INVALID | FE_DIVBYZERO | FE_OVERFLOW));

#else

  __setfpucw (_FPU_DEFAULT &
              ~(_FPU_MASK_IM | _FPU_MASK_ZM | _FPU_MASK_OM));

#endif
}

void fpenab_ () {
  trapfpe();
  }

/*
  Enable exception traps for the exceptions specified by traps.
  */
int feenabletraps(int traps)
{
  unsigned short cw;

  asm volatile ("fnstcw %0":"=m" (cw));
  cw &= ~traps;
  asm volatile ("fldcw %0"::"m" (cw));
  return 0;
}
</QUOTE>

Hope this helps,

-- 
Toon Moene - mailto:toon@moene.indiv.nluug.nl - phoneto: +31 346 214290
Saturnushof 14, 3738 XG  Maartensdijk, The Netherlands
GNU Fortran 77: http://gcc.gnu.org/onlinedocs/g77_news.html
GNU Fortran 95: http://g95.sourceforge.net/ (under construction)