OpenBSD Journal

Hardware request for c2k4 SMP efforts, use i386 snapshots for testing.

Contributed by grey on from the damage control dept.

We let the cat out of the bag early with regards to the SMP work, but in hopes of mending the damage, we hope our readers can find machines that match the criteria outlined in the hardware request found here.

If you test out your hardware and run into the problems mentioned, and you can express ship by Friday for c2k4, it will be appreciated.

For testing purposes on this front, there are new OpenBSD/i386 snapshots that will come in handy as mentioned here.

Please don't blindly ask what kind of hardware this happens on as there don't appear to be concrete details on that currently. Test the snapshots for yourself and see if you encounter the described issues and can help, ideally any machines found that exhibit this behaviour are small enough to ship easily unlike what was mentioned here.

The complete request is as follows:

List:       openbsd-tech
Subject:    We are looking for a particular SMP machine
From:       Theo de Raadt 
Date:       2004-06-13 18:53:34
Message-ID: <200406131853.i5DIrYSf003092 () cvs ! openbsd ! org>

As some of you have noticed, our SMP branch is working on almost all
MP machines.  The test reports we have received have been really helpful.
Thanks.

There are currently about 6 known major problems, all of which fall
into one particular area of the code: interrupt routing on a few very
very badly designed machines.  Engineers should be shot, but as it is,
we might as well make our code deal with this issue.

The two most major issues:

    1. "no MP mapping found".

    2. Interrupts are disabled during bootup, leading to cardbus/pcmcia
       and audio link rate probe issues; perhaps other issues too.

There is a hackathon upcoming in Calgary where nearly 50 developers
will be spending a week hacking on lots of things, including these
problems.

Unfortunately, none of the attending developers has a machine
available which exhibits this problem!  We've been hunting everywhere!

So we are hoping that before this coming Friday someone will step up
who is willing to express ship us a machine (or motherboard + cpus +
memory, assuming that it uses a standard power supply) which exhibits
the problematic issue.

We are looking for a machine that displays "no MP mapping found"
messages during boot of GENERIC.MP.  If you have one available, please
drop me a note saying what you can do (include as much of a dmesg as
you can) and I'll pick the best of the worst and take it off your
hands...  It is not going to work with GENERIC.MP until we fix this
(and once we fix it, it will be a poor performer because the board is
badly designed)

Thanks a lot.

(Comments are closed)


Latest Articles

Credits

Copyright © - Daniel Hartmeier. All rights reserved. Articles and comments are copyright their respective authors, submission implies license to publish on this web site. Contents of the archive prior to as well as images and HTML templates were copied from the fabulous original deadly.org with Jose's and Jim's kind permission. This journal runs as CGI with httpd(8) on OpenBSD, the source code is BSD licensed. undeadly \Un*dead"ly\, a. Not subject to death; immortal. [Obs.]