🚀 go-pugleaf

We have in program in C that performs a test of the floating point
hardware.

Because of the C standard, the conversion from floating point to
integer is truncated. But we need to test this in the four available
rounding modes. For the i386-version of the program, I was able to
find some code with inline assembly that online performs the
conversion, and doesn't set the rounding mode to zero. The code is as
follows:

static inline int RoundToInt(double d) {
  int i;

  asm ("fistpl %0" : "=m" (i) : "t" (d) : "%st");

  return i;
}

But I'm not able to generate assembly code for the Sparc-version of
the program. Is there someone who can help me out with this?

We use gcc 2.8.1 on our system.

Thanks in advance,
Peter.