Floating point math library in kernel?

Hi,
I am now developing some kernel modules in our own embedded target board.
The processor
we are using doesn't support FPU, so I am considering to use kernel S/W
emulation. After compiling
our kernel modules with -msoft-float parameters, the command "insmod xxxx.o"
returned some
unsolved symbols regarding floating point math function calls, such as
sin( ) or floor( ) etc.
I guess our embedded Linux kernel doesn't support the FP math lib for
linking, could anyone be
kind to tell me how to solve it? Thanks a lot.

Regards,

Sam