Subject: Re: NetBSD 1.4.1 - Clock stops when doing I/O
To: Michael R. Zucca <mrz5149@acm.org>
From: Allen Briggs <briggs@ninthwonder.com>
List: port-mac68k
Date: 01/05/2000 22:56:06
> It is definitely possible. I've looked at the linux-mac68k sources recently
> and they do ADB, serial, and floppy via the IOPs. If you don't attempt to
> reprogram the IOPs yourself the Apple supplied IOP software works on a
> simple message passing protocol. Using those drivers as a "developers
> guide" it should be easy enough to write our own.

Well...  It's supposed to work.  I haven't actually tried Linux
itself, but I was getting totally wedged on my Q950 with the code
that I ended up with, and I couldn't tell how it was functionally
_any_ different from the Linux code.  My next step, I think, was
to make a fresh MacOS installation and boot linux from that--make
sure that things are functional there, and then try again to get
it working...  I went ahead and dumped my latest code into the
-current tree so that someone else wouldn't have to start from
ground zero on it.

> I would imagine performance might be improved by reprogramming the IOPs to
> use a deeper (greater than one) fifo with larger queue elements. Of course,
> we'd need a 6502 assembler in the NetBSD distribution to do it but I think
> that's the least of the problems associated with such a project :-)

That would be cool.  It would probably require some more knowledge
about the operating environment within the IOP, though I'm not sure
about that.

BTW, what's the current status of the color/graphics code?  I haven't
noticed/heard anything about it in a while.

-allen