[Bug c/12070] New: ICE converting between double and long double with -msoft-float

["taylor at candd dot org" <gcc-bugzilla at gcc dot gnu dot org>]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12070

           Summary: ICE converting between double and long double with -
                    msoft-float
           Product: gcc
           Version: 3.3.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: taylor at candd dot org
                CC: gcc-bugs at gcc dot gnu dot org
GCC target triplet: i686-pc-linux-gnu

The following simple function (extracted from libstdc++-v3/libmath/stubs.c)
will get an internal compiler error.

extern double atan2(double, double);

long double
atan2l(long double x, long double y)
{
  return atan2((double) x, (double) y);
}

The ICE occurs with x86 native on GNU/Linux with GCC 3.3.1 when invoked with
-msoft-float.  It does not ICE with GCC 2.95.3.  I haven't yet tried the HEAD
of the CVS trunk.