d7bbed6455
to be soft-float. the floating-point usage in ld.elf_so is only because of stdarg functions like printf(), and gcc 4.5 is good about not actually executing the FP instructions in such functions unless FP values are passed to them, which ld.elf_so doesn't do.