On Sat, 19 Apr 1997, Kyle Farrell wrote:

> Hi all-
> 
> I'm in the process of compiling a kernel and noticed that in the CFLAGS of
> the makefile "-msoft-float" is used.  My machine is a TT030, with a 68882
> FPU.
> 
> Should the soft-float flag be given in the makefile?  Am i missing
> something in the kernel configure file?  I don't have FPU_EMULATE
> specified.  I assume the compiler is using 6888{1,2} instructions, but the
> soft-float flag makes me curious. 

	There is virtually nothing in the kernel that could use an FPU,
	while saving/restoring FPU state is relatively expensive.

	If the kernel made any use of the FPU the state would have to be
	stored/restored twice for every system call (once on entry to the
	kernel, once on exit). This is definitely a Bad Thing :)

	On m68k -msoft-float is used to ensure that no FPU code is used
	'by accident' in the kernel. On other ports (alpha?) more
	stringent flags are needed to ensure the FPU is untouched within
	the kernel :)
	
                David/abs               abs@anim.dreamworks.com

    --- Oxycodone (Percocet) is definitely great stuff...  However on ---
    --- balance its not worth having wisdom teeth extracted to obtain ---