Steven J. Hill wrote:
Unless you compile your code with -msoft-float *and* also have a version
of libgcc compiled with -mlong-calls -mno-abicalls -G0. If you do it
that way, floating point works fine in the kernel (as long as you don't
try to call sprintf with floating point parameters).
I won't even concede that solution. It's bad practice and design to have
floating point in the kernel.
I agree that floating point in the kernel is bad practice. However
under some circumstances, the most expedient solution does not conform
to best practice.
David Daney
