This breaks compilation on a hardfloat toolchain as we would need AFLAGS += -msoft-float aswell. Since we do not use floating point we can equally well remove the option completely. Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>