linux-mips-fnet
[Top] [All Lists]

Re: Hi, Linux on DECsystem 5100

To: linux-mips@fnet.fr
Subject: Re: Hi, Linux on DECsystem 5100
From: "Paul M. Antoine" <paul@softway.com.au>
Date: Mon, 08 Jun 1998 12:10:03 +1000
Organization: Softway Pty Ltd
References: <199806071443.QAA10233@jordan.numerik>
Michael Engel wrote:

> > Unfortunately we also had a corporate image change that meant I had
> > to remove my DECstation pages, which has just made things worse, as
> > it appears to some that I've dropped off the planet! :-)
> 
> What horrible things are happening Down Under ???

That's definitely a three-beer story! :-)

> I should probably put up a notice on the web server that mentions
> that you are still alive :-).

Indeed.

> > Meanwhile, fortunately, Harald has taken up the role of de facto
> > DECstation maintainer. I'm here, and I'm listening and sometimes
> > hacking.  When Harald gets the latest 2.1.xx sources up, I'll try
> > hacking some more.
> 
> 2.1.99 is up along with a set of patched binutils-2.8.1 compiled for
> Linux-i386/libc5. There is also a precompiled netbootable kernel image
> at http://decstation.unix-ag.org/kernels/

I've downloaded stuff, and have had some problems.  The 2.1.99 kernel
binary seems to be corrupt (I can't gzip -d the file), and my web
browser reports that the gcc 2.7.2 Intel cross-dev package URL is
incorrect and the file cannot be found - the URL in the link is:

        http://decstation.unix-ag.org/software/gcc-i386.tar.gz

Apart from that, the source tree comes down fine, and the binutils
package (2.8.1) is also fine.

> I just got around to test 2.1.99 on my Personal DS 5k/25 with graphics
> console. Something is still crashing the system while probing for
> TurboChannel devices (in tc_probe), I'm not sure if the dbe
> exception handler gets initialised correctly - it probably never gets
> called.

This one has caused me problems in the past, but I thought Harald had
fixed it.

> It might also be that the REX PROM routines return a wrong address
> for the TC slots, I'll have to check that out.

Could be, but the PROM self-test routines would rely on these too...

> When removing the TC probing, the kernel runs up to loading the
> initial RAM disk which is - according to Harald - as far as it can
> run at the moment.

He's just posted a patch that gets things further!

> I'll send more details to Harald later, now I've got to get my slides
> ready for my Linux lecture at University on Tuesday ...

Good luck!  What aspect of the wonderful world of Linux is your lecture
going to cover - the "joys" of cross-dev hacking perhaps?? :-)

Paul

<Prev in Thread] Current Thread [Next in Thread>