Its there to stop the kernel using any floating point code
	(except in specifically written assembler). If the kernel
	uses floating point then it has to save and restore the
	entire FPU state on each context switch into and out of the
	kernel and this is too much of an overhead (given that there
	is so little fp usage inside the kernel).

	-msoft-float _shouldnt_ be needed, as there should not be
	any code that uses floating point (except for the aforementioned
	specifically written assembler), but its there as a safety net.

	This question gets asked regularly on just about all the
	ports, maybe someone should add a comment to the Makefile?

		David/abs	david@{mono.org,southern.com,mhm-internet.com}

System Manager: Southern Studios Ltd, PO Box 59, London N22 1AR.
Satisfied User: NetBSD, free Un*x {i386,sparc,mac68k,+more} 'www.netbsd.org'.
  System Admin: MHM Internet, 14 Barley Mow Passage, Chiswick, London W4 4PH.
         SysOP: Monochrome, Largest UK Internet BBS - 'telnet mono.org'..

On Sun, 29 Dec 1996, David Carrel wrote:

> Why is "-msoft-float" always in the CFLAGS for kernel builds?  I edited it
> out of the Makefile and the kernel builds fine and seems to be running
> fine.  It seems to me, that for machines with FP hardware having
> -msoft-float would just produce a slower kernel.  Am I missing something.
> Is there just so little FP used in the kernel that it's left in the
> Makefile for generality??
> 
> Dave
>