Richard Henderson writes: > On Fri, Apr 29, 2005 at 07:13:16PM +0100, Andrew Haley wrote: > > I agree. It's very little extra overhead for the non-varargs case, > > and it'll make varargs functions work. > > Surely we can compute the correct value... Sure, but is that reason to rejct the patch? The value in %ax at the present time is garbage. Andrew.