headers
Patrick M. Hausen | 3 Jul 2002 10:00
Picon
Favicon

Q: bootstrapping a "virgin" RaQ 1?

Hi all!

I started to experiment with NetBSD/Cobalt lately and
now I'm kind of stuck in the installation process.

Setting up my FreeBSD machine as a server to netboot
the RaQ and have it run from NFS root was painless as
can be - so now I'm running NetBSD 1.5.2. Great ;-)

The problem is getting the OS on that bloody hard disk.

1. boot netbsd.GENERIC.gz -> trigger DMA "bug", system hangs

2. boot netbsd.GENERIC.noatadma.gz -> system is running fine from
   NFS root

3. fdisk -> panic

I have found:

http://mail-index.netbsd.org/port-cobalt/2001/08/23/0000.html

and various other mails stating the same problem. Since I can
run the system from NFS I could recompile. Is there any patch
to the kernel or fdisk letting me get beyond step 3?

Getting -current sources from a daily snapshot and calling
./build.sh failed early in the compilation process.
I'm used to the "make buildworld, buildkernel, installkernel, installworld,
mergemaster" steps in FreeBSD. Is NetBSD fundamentally different here?
(Continue reading)

Permalink | Reply | 1 comment |
headers
Kevin Lahey | 3 Jul 2002 18:02

Re: Q: bootstrapping a "virgin" RaQ 1?

In message <200207030800.g6380Gr86457 <at> hugo10.ka.punkt.de>
"Patrick M. Hausen" writes:
>
>Getting -current sources from a daily snapshot and calling
>./build.sh failed early in the compilation process.
>I'm used to the "make buildworld, buildkernel, installkernel, installworld,
>mergemaster" steps in FreeBSD. Is NetBSD fundamentally different here?

I've never had any trouble on my i386 systems with the build.sh
process, but the Cobalt has always been a bit tougher.  I've often 
had trouble with sh;  lately I've taken to temporarily replacing
/bin/sh with bash.

Once I do that, builds of both the regular source tree and 
of pkgsrc seem to be much more successful.  The only downside
of this replacement is that the startup scripts in /etc/rc.d
don't seem to groove on bash, so I have to make sure I reinstate
the normal /bin/sh before rebooting.

I've been too lazy to PR this, let alone figure out the 
problem, so I'm not exactly pointing fingers here.  :-)

Good luck,

Kevin
kml <at> patheticgeek.net
Permalink | Reply | 0 comments |
headers
marius | 7 Jul 2002 15:14
Picon

Re: Q: bootstrapping a "virgin" RaQ 1?

On Wed, Jul 03, 2002 at 10:00:15AM +0200, Patrick M. Hausen wrote:
> Hi all!
> 
> I started to experiment with NetBSD/Cobalt lately and
> now I'm kind of stuck in the installation process.
> 
> Setting up my FreeBSD machine as a server to netboot
> the RaQ and have it run from NFS root was painless as
> can be - so now I'm running NetBSD 1.5.2. Great ;-)
> 
> The problem is getting the OS on that bloody hard disk.
> 
> 1. boot netbsd.GENERIC.gz -> trigger DMA "bug", system hangs

this is due to cobalt 2700 hardware (raq1/qube1) not being dma coherent

> 
> 2. boot netbsd.GENERIC.noatadma.gz -> system is running fine from
>    NFS root
> 
> 3. fdisk -> panic
> 

this happens on all cobalt mips models under netbsd when ata dma is
disabled for reasons unknown to me; maybe this is fixed on -current,
don't know

you could netboot your raq1 with nbsd 1.5.2 and recompile a kernel with
the attached patch applied (bus.c is located in src/sys/arch/cobalt/cobalt)
_and_ ata dma enabled; afterwards you should be able to fdisk and install
(Continue reading)

Permalink | Reply | 0 comments |
headers
Jamie Heilman | 13 Jul 2002 02:47

make build error again, w/1.5.3 kernel

I just ran into the same thing Thang T. Mai did in April[1], only I was
running a 1.5.3 kernel built from sources pulled on the 11th of this
month.  Everything dynamically linked against libc bombs now.  What
next?  I had no critical data on this qube2 so I can always reinstall
without any pain, but I'm not sure where to go from here, does 1.6
have the same problem?

[1] http://mail-index.netbsd.org/port-cobalt/2002/04/23/0000.html

--

-- 
Jamie Heilman                   http://audible.transient.net/~jamie/
"...thats the metaphorical equivalent of flopping your wedding tackle 
 into a lion's mouth and flicking his lovespuds with a wet towel, pure 
 insanity..."						-Rimmer
Permalink | Reply | 0 comments |
headers
Kevin Lahey | 13 Jul 2002 07:53

Re: make build error again, w/1.5.3 kernel

In message <20020713004723.GF1383 <at> audible.transient.net>Jamie Heilman writes
>I just ran into the same thing Thang T. Mai did in April[1], only I was
>running a 1.5.3 kernel built from sources pulled on the 11th of this
>month.  Everything dynamically linked against libc bombs now.  What
>next?  I had no critical data on this qube2 so I can always reinstall
>without any pain, but I'm not sure where to go from here, does 1.6
>have the same problem?
>
>[1] http://mail-index.netbsd.org/port-cobalt/2002/04/23/0000.html

I've been bit by this several times.  I generally set DESTDIR
to an NFS mounted partition, then NFS boot off of that partition
and do the install into the disk paritition the Qube.  I'm not
really sure why the Cobalt port is so sensitive to this, as I've 
never had these problems on i386 boxes.

OTOH, none of my i386 boxes have that cool glowing green bar.  :-)
I think that the fact that we have a port to the Qube at all
is way, way cool.  It looks great sitting in the corner of
my home office, sucking down a mere 30W and happily providing
802.11b...

My recent upgrade from 1.5-something-current to 1.6C to pick
up the security fixes went fairly smoothly, so perhaps that'll
work out for you without using a non-/ DESTDIR.  I wouldn't bet 
the farm on it, though.

Good luck,

Kevin
(Continue reading)

Permalink | Reply | 1 comment |
headers
Jamie Heilman | 15 Jul 2002 01:46

Re: make build error again, w/1.5.3 kernel

> I've been bit by this several times.  I generally set DESTDIR
> to an NFS mounted partition, then NFS boot off of that partition
> and do the install into the disk paritition the Qube.  I'm not
> really sure why the Cobalt port is so sensitive to this, as I've 
> never had these problems on i386 boxes.

Well I just reinstalled from the 1.6 snapshots, so I won't need to
rebuild for a while I guess.  I can do the DESTDIR swizzle if need be,
but does anyone know why a normal make build chokes like this?  Seems
like it must be some ordering dependancy.

--

-- 
Jamie Heilman                   http://audible.transient.net/~jamie/
"Paranoia is a disease unto itself, and may I add, the person standing
 next to you may not be who they appear to be, so take precaution."
						-Sathington Willoughby
Permalink | Reply | 0 comments |
headers
Thierry | 23 Jul 2002 07:49
Picon

pppoe and qube2

I had like to know if someone had a running config with pppoe.
My box is running netbsd 1.6A GENERIC from chs.
The system is up and running after a real fighting party with the boot...
I don't find how to boot ffs without bfd /boot/vmlinux.gz-a command.
What 's the right configuration or how to compile the kernel without 
core.

Thank's

Thierry
Permalink | Reply | 0 comments |
headers
Bob Bostwick | 31 Jul 2002 15:28

(unknown)

Anyone else experiencing problems with netbsd FTP site?  I try to login
anonymous, but it never accepts my password, no matter what I use.  Am I
supposed to use a special password for anonymous?

Thanks,
Permalink | Reply | 0 comments |

Gmane