Subject: Re: Anyone home
To: Toru Nishimura <email@example.com>
From: Matthew Hudson <firstname.lastname@example.org>
Date: 05/09/2000 10:52:57
I'm supposed to be working on ELF, but with work and
other personal situations I haven't had much time. If
there is anyone interested I can tar up my source pool
and they can continue the work. Someone with much more
experience working on binutils than I have really needs
to get there hands on this.
Toru Nishimura wrote:
> >> So, who is doing what?
> > I'm trying to find time to work on the AViiONs. I've also received
> > a luna88k box, but I've not gotten it hooked up yet. The person or
> > persons who were/are working on the luna68k port have mentioned (and
> > sparked some discussion about details regarding) a luna88k port. I
> > don't know what their schedule or status is.
> During last weekend, I went back to my NetBSD/luna88k project, and
> last night I could figure out what I have wanted to know. Here goes
> what I learned from Mach/luna88k source codes.
> LUNA88K can have upto 4 CPUs and has 4 different 32bit registers in
> parallel to indicate which device requests for service at a certain
> memory location. Device interrupts is served only by master #0 CPU.
> Slaves can take hardclock() and IPI requests. After brainstorming of
> interrupt service glue logic, I think I got the code for NetBSD/luna88k
> at last. The code is MP ready, but in this moment, I have no plan to
> spin up slave CPUs, and kernel will be compiled for uni-processor
> configuration with #define cpu_number() (0).
> I still no access to m88k UM (users manual). From a help from
> conventional computer science textbook for general public, I learned
> some about m88k. Logic of exception service (enter & return) using
> SXIP, SFIP, SNIP registers seem obscure, but managable, I think.
> Scratch build of pmap.c seems not a terribly hard work (no code was
> written in this area). Another concern (what I don't learn yet) is how
> to implement delayed lazy FPU context switch.
> Anyway, NetBSD/luna88k got a bit concrate ficture, but before true
> ELF/m88k toolchain the real output will not happen. If someone wants
> to see my fragmented codes, please drop me a message privately.
> Tohru Nishimura
> Information Technology Centre
> Nara Institute of Science and Technology