It seems gcc requires libgcc to do that, but libgcc is not available.
Right now the code casts to a double, but it is not good to use the
FPU in the kernel. I do have a disgusting hack that does the job in
4 divisions and 3 multiplications (hmmm, is div() available?), but
the hack takes almost 4 times as long on a Pentium.
Just how bad are FPU operations in the kernel? I seem to remember
that the FPU memcpy patch had to take special precautions to avoid
corrupting data.