Re: Please test new SILO - problems with CDROM booting may be fixed
* Jurij Smakov ([EMAIL PROTECTED]) wrote: > David Miller has just posted a silo fix, addressing the long-standing > problem of it being unable to boot the system from CDROM on a variety of > hardware. So if you've got a machine which dies with 'Fast Data Access > MMU Miss', or 'FP Disabled', or 'Illegal instruction' while trying to boot > from CDROM, please try the iso image at > > http://www.wooyd.org/debian/silo/sparc-mini.iso > > It features the patched-up silo, and might work for you. Please report > your experiences to the list. Finally! This booted up to "Booting Linux..." for me, making it *much* farther than previous SILO's. No 'MMU Miss' error! This was on my SunFire v880 (which had always previously given an 'MMU Miss' error). I'm guessing it probably did boot the kernel and just ended up finding the video card, so I'll have to pop by the lab sometime and see what's on the screen. :) Thanks! Stephen signature.asc Description: Digital signature
Re: Unofficial CD featuring SILO 1.4.11 available
* Michael-John Turner ([EMAIL PROTECTED]) wrote: > On Mon, Apr 10, 2006 at 02:09:32PM -0400, Stephen Frost wrote: > > I know you asked me to try this out a long time ago but I finally got > > around to actually doing it. :) Here's a boot attempt using OpenBSD: > > Hmm... Is this not due to OpenBSD not support US-III CPUs? I think the idea was to check if it got past the MMU-miss thingy, which I *think* the test indicated that it did... > I think FreeBSD is the only *BSD with (rudimentary) support for the US-III. > NetBSD and OpenBSD only support US-I and US-II. Hmm, might be interesting to test that, tho I thought that FreeBSD didn't support it either, honestly... Thanks, Stephen signature.asc Description: Digital signature
Re: Unofficial CD featuring SILO 1.4.11 available
Jurij,
I know you asked me to try this out a long time ago but I finally got
around to actually doing it. :) Here's a boot attempt using OpenBSD:
Sun Fire 880, No Keyboard
Copyright 1998-2002 Sun Microsystems, Inc. All rights reserved.
OpenBoot 4.5, 4096 MB memory installed, Serial #51056707.
Ethernet address 0:3:ba:b:10:43, Host ID: 830b1043.
{2} ok boot cdrom
Boot device: /[EMAIL PROTECTED],70/[EMAIL PROTECTED]/[EMAIL
PROTECTED],0:f File and args:
OpenBSD IEEE 1275 Bootblock 1.1
..>> OpenBSD 3.8 (obj) #1: Thu Sep 1 17:32:37 MDT 2005
[EMAIL PROTECTED]:/usr/src/sys/arch/sparc64/stand/ofwboot/obj
: trying bsd...
Sun Fire 880, No Keyboard
Copyright 1998-2002 Sun Microsystems, Inc. All rights reserved.
OpenBoot 4.5, 4096 MB memory installed, Serial #51056707.
Ethernet address 0:3:ba:b:10:43, Host ID: 830b1043.
{2} ok
It seems to get *farther*, at least, and it doesn't report back the
same error (just reboots). Dunno if that helps or not.. This was
using:
-rw-r--r-- 1 sfrost sfrost 4259840 Feb 5 21:47 cd38.iso
===# md5sum cd38.iso
10911d8b58dd97b118ccd4522c761985 cd38.iso
===#
Thanks,
Stephen
signature.asc
Description: Digital signature
Re: Unofficial CD featuring SILO 1.4.11 available
Greetings,
I tried this out on my:
Sun Fire 880, No Keyboard
Copyright 1998-2002 Sun Microsystems, Inc. All rights reserved.
OpenBoot 4.5, 4096 MB memory installed, Serial #51056707.
Ethernet address 0:3:ba:b:10:43, Host ID: 830b1043.
Didn't work so hot tho, much the same as the others:
Rebooting with command: boot cdrom
Boot device: /[EMAIL PROTECTED],70/[EMAIL PROTECTED]/[EMAIL
PROTECTED],0:f File and args:
SILO Version 1.4.11
Fast Data Access MMU Miss
{2} ok
Sorry. :/
Thanks,
Stephen
signature.asc
Description: Digital signature
Re: Developer accessible SPARC machine
* Ben Collins ([EMAIL PROTECTED]) wrote: > On Wed, 2006-01-04 at 00:22 -0800, Steve Langasek wrote: > > On Tue, Jan 03, 2006 at 04:35:30PM -0800, Andrew Pollock wrote: > > > I note that one of the issues with the Sparc port is the the lack of a > > > developer accessible machine. > > > > At present, vore.debian.org is back on line; the underlying issue, though, > > seems to be that vore, like the buildds, won't necessarily *stay* on-line > > due to some hard-to-pin kernel bugs that keep taking the systems down. > > Vore is stable now. What was the issue? I'm guessing kernel bug, so what change/version resolved the problem? Can it build openoffice, etc, now? Has there been any progress on the SunFire (and similar) 'Data MMU Fast Path Miss' or whatever that error is when trying to boot from cd? Thanks! Stephen signature.asc Description: Digital signature
Re: sparc meeting requirements for etch
* Frans Pop ([EMAIL PROTECTED]) wrote: > On Thursday 22 September 2005 16:03, Ben Collins wrote: > > What we lack is someone pointing to an IP address saying, saying > > "Here's an 8-way UltraSPARC 4000 hosted on a multihomed T3, here's root > > access, we wont touch the box, and it will be there for as long as you > > need it. I, the company president, personally ok use of our resources > > for Debian." > > So who has the contacts to arrange something like this (although I guess > slightly lower specs would still be acceptable)? > As I understand it debian-sparc is effectively dead if an extra machine is > not found on this basis. I've been working this issue through my management folks. I need to write up a briefing on why it'd be good for us to host it and I've been slacking on that- sorry. :/ I'll try to write the briefing and get on the director's schedule next week. I'd need the actual sparc box but it sounds like we've got the hardware. I do have an ultra1 I could use but it sounds like there's much better available. Assuming management approves, it'd be on a multihomed network on a fibre ring, I think we have something like 15Mbs from our ISPs. It'd be in one of the labs I'm the lab manager for behind 2 badge readers with very limited access and probably in a locked rack (assuming it's a rack-mount machine). Anyway, I've been working with vorlon and dilinger (iirc) on it and hopefully will have something positive to talk about next week. Thanks, Stephen signature.asc Description: Digital signature
Re: fast data access mmu miss after harddisk upgrade
> > --- Jurij Smakov <[EMAIL PROTECTED]> wrote: > > > This is a pretty old heisenbug, which cannot be reproduced reliably > > > (except on SunFire 280R). It has been reported numerous times before, Just to comment on this specifically- I seem to be able to reproduce it pretty reliably on a SunFire v880. I'm happy to test things on it and/or provide futher info about the problem. I may be able to allow remote access via serial port if it'd be useful. I'd certainly like to help get this apparently long-standing problem solved.. Stephen signature.asc Description: Digital signature
