On Tue, Sep 19, 2000 at 09:52:15AM -0700, Richard Henderson wrote:
> > Instead of *= 0.5, try /= 2.0
>
> Yes indeed you've found a bug in the kernel's FP emulation.
> I'll see about fixing it.
Rather than fix the old udiv128 function, which was trying to do
128/128 bit division, I've pulled in a subroutine from libgcc
that does 128/64 bit division, which is all we need here.
So it should be a bit faster than the old routine, for what ever
that's worth to someone who is already trapping into the kernel...
r~
This archive was generated by hypermail 2b29 : Sat Sep 23 2000 - 21:00:22 EST