Anyone know how/if I can detect in runtime which software float method is being used, kernel emulation or -msoft-float ?