Patch for i386 libm-test-ulps

[Andreas Jaeger]

Ulrich Drepper <drepper@redhat.com> writes:

> Andreas Jaeger <aj@suse.de> writes:
> 
> > On both my P3 and my Athlon with both gcc 2.95.2 and gcc from CVS, I
> > got some problems libm-test after the constants have been changed to
> > long double.
> > 
> > I've committed the appended patch,
> 
> I prefer to not have this.  I have a PIII as well and don't see any
> problems.  So this must be a compiler issue.  If gcc 2.95 generates
> wrong code it cannot be used.  If (what is more likely) it generates
> code which causes the loss of precision the C code must be made more
> stable so that a different compiler makes no difference.
> 
> Please undo the patch and look at the code difference between 2.95 and
> what ever other gcc version.

I'll undo it for now and will look what the differences are.  What
compiler flags are you using?  It might be that this is one issue why
you don't see it and I see it.

Andreas
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.inka.de
    http://www.suse.de/~aj