Subject: m88k code donation
To: None <port-m88k@netbsd.org>
From: Toru Nishimura <locore32@gaea.ocn.ne.jp>
List: port-m88k
Date: 10/23/2003 13:39:37
Hi,

Some of you remember I worked on m88k years ago.   Since
then I'm out of touch from the hardware and abandoned to
make effort for it.   I spent a significant time to look and analyze
Mach/luna88k and OpenBSD/mvme88k, and made sketchy code
fragments.   I donate the entire files to NetBSD hoping others
will take it as stair steps.

- pmap.c
Near to complete.  It's another occurrence of 10+10+12 address
partitioning MMU and pmap.c ends up with pretty close to i386.
(in fact, closer to my own MIPS pmap.c)

- trap.c
m88k processor is peculiar in the way to solve exception.  The
pipeline context of exception is exposed out of processor and
trap() handler has to decode the condition to prepare exception
return.  I left uncoded around them.   Another vacancy is
floating point trap handler.

- locore.s
Sketch work, but it should contain complete description what
to fill before working code for real hardware.

- other files in include/ and m88k/ for completeness...

Allen, if you're interested, I would like you take over my code.

I ran 2 (4?) processor luna88k for a while and recognized it was
significantly faster then R3000.  The box had Mach2.5.  Analysis
revealed it drives single dedicated CPU (#0) to run kernel.

Toru Nishimura/ALKYL Technology