Subject: Re: NetBSD without MMU ?
To: netbsd-ports@netbsd.org, , <tech-ports@netbsd.org>
From: Greg A. Woods <woods@weird.com>
List: tech-ports
Date: 04/12/2002 13:18:21
[ On Friday, April 12, 2002 at 12:02:17 (-0400), Todd Vierling wrote: ]
> Subject: Re: NetBSD without MMU ?
>
> It's still evil from a modern CPU standpoint.  8-)

Maybe evil for modern CPUs with MMUs, but my limited understanding of
some of those MMU-less CPUs we might be talking about here are much more
ammenable to running position independent code (eg. the 8088 in the way
MINIX uses it! ;-)....  You can also use a relocating loader in the
kernel instead, of course, though unless your coding model is extremely
restricted that'll prevent you from moving the process (i.e. doing
compaction) once it's been loaded.

(Obviously there's still a whole "new" (or rather very old) approach
needed for memory management, and processes can't grow beyond some fixed
size that's allocated for them on exec, but isn't that what makes
programming such "embedded systems" so much more fun and interesting?  :-)

I think the real win for using a full MMU with hardware page protection
support even for an embedded system is of course the benefits of that
protection support.

-- 
								Greg A. Woods

+1 416 218-0098;  <gwoods@acm.org>;  <g.a.woods@ieee.org>;  <woods@robohack.ca>
Planix, Inc. <woods@planix.com>; VE3TCP; Secrets of the Weird <woods@weird.com>