On Tue, 11 Mar 1997 18:06:05 -0800 (PST),
David Bushong <dbushong@cory.EECS.Berkeley.EDU> wrote:
> OK, I'm compiling a kernel, and I leave out FPE and FPSP, since I have an
> '030 with fpu (IIcx).
> 
> a) that's correct, right?

Correct.  If you wish, you could even leave out such options as
MC68040 and MC68020.

> b) why do I still get -msoft-float in my compile lines
>    i) am i supposed too and am misunderstanding the meaning of msoft-float?

I believe you are supposed to.

>    ii) if not, how do I get rid of it?

Why?  It hurts *very* little if at all.

find /sys -name '*.[ch]' -print | xargs egrep \
	'^[ 	]*(typedef|.*[,;{}()])?[ 	]*(double|float)'

This revealed there was no floating point operation of interest to us
mac68k users except a few in netiso/{clnp.h,clnp_flag.c}, though this
pipeline might miss something...  Even the load average is calculated
in fixed point.

Cheers,

Ken