headers
Tom Vier | 3 May 2003 23:57
Picon

can't get sgivol to work

i'm trying to move everything to another drive, but i can't seem to make it
bootable. i've run sgivol -w boot /usr/mdec/boot sd1, but arc doesn't want
to boot it.

when i try to boot, it seems to exec the loader, but i get this:

Boot: scsi(1)disk(5)partition(0)netbsd
devopen: scsi(1)disk(5)partition(0) type scsi file netbsd
open scsi(1)disk(5)partition(0)netbsd: No such file or directory
Boot failed! Halting...

/netbsd is definetly there, however.

i have netbsd installed and booting from sd0. sd1 is the drive i'm moving
everything to.

sd0 disklabel:
8 partitions:
#        size    offset     fstype  [fsize bsize cpg/sgs]
 a:   7947440      3280     4.2BSD   1024  8192    54   # (Cyl.    2 - 4847)
 b:    524800   7950720       swap                      # (Cyl. 4848 - 5167)
 c:   8482080         0     unused      0     0         # (Cyl.    0 - 5171)
 d:      3280         0       boot                      # (Cyl.    0 - 1)

sd1 disklabel:
4 partitions:
#        size    offset     fstype  [fsize bsize cpg/sgs]
 a:   1541967      3605     4.2BSD   1024  8192   144   # (Cyl.    7 -
3001*)
 b:    524288   1545572       swap                      # (Cyl. 3001*-
(Continue reading)

Permalink | Reply | 2 comments |
headers
Rafal Boni | 4 May 2003 05:28
Picon

Re: can't get sgivol to work

In message <20030503215708.GA1479 <at> zero>, you write: 

-> i'm trying to move everything to another drive, but i can't seem to make it
-> bootable. i've run sgivol -w boot /usr/mdec/boot sd1, but arc doesn't want
-> to boot it.
-> 
-> when i try to boot, it seems to exec the loader, but i get this:
-> 
-> Boot: scsi(1)disk(5)partition(0)netbsd
-> devopen: scsi(1)disk(5)partition(0) type scsi file netbsd
-> open scsi(1)disk(5)partition(0)netbsd: No such file or directory
-> Boot failed! Halting...

Is sd1 indeed id 5 on SCSI bus 1 -- whichever SCSI bus that happens to
be?  Even if that isn't a problem here, it may cause issues with the
kernel detecting the boot device later...

[..]
-> sd1 disklabel:
-> 4 partitions:
-> #        size    offset     fstype  [fsize bsize cpg/sgs]
->  a:   1541967      3605     4.2BSD   1024  8192   144   # (Cyl.    7 -
-> 3001*)
->  b:    524288   1545572       swap                      # (Cyl. 3001*-
-> 4019*)
->  c:   2069860         0     unused      0     0         # (Cyl.    0 -
-> 4019*)
->  d:      3605         0       boot                      # (Cyl.    0 - 6)
-> 
->
(Continue reading)

Permalink | Reply | 1 comment |
headers
Tom Vier | 5 May 2003 03:54
Picon

Re: can't get sgivol to work

On Sat, May 03, 2003 at 11:28:50PM -0400, Rafal Boni wrote:
> I wonder whether your ARCS variables simply aren't set right (not that
> I'm anywhere near a master of ARCS :-), but here's what my Challenge S
> says:

turns out the problem was a and b didn't weren't cylinder-aligned. thanks,
though.

--

-- 
Tom Vier <tmv <at> comcast.net>
DSA Key ID 0xE6CB97DA
Permalink | Reply | 0 comments |
headers
Christian Johansson | 8 May 2003 17:38
Picon

OT: Single ended disks in Onyx system?

Hi!

Sorry about this question being OT, but since I was on this list anyway
I thought there may be someone here who knows the answer.

I just got an SGI Onyx RE2 system and the system disk was a bit
small for IRIX 6.5 (NetBSD not supported, right?). The problem
is that the disk seems to be differential and my question is: 
May I stick an "normal" single ended disk in that drive sled without
problems, or will there be flames and smoke?

Regards

/Christian Johansson
Permalink | Reply | 3 comments |
headers
Ethan | 8 May 2003 22:35
Favicon

Re: OT: Single ended disks in Onyx system?


> Hi!
> Sorry about this question being OT, but since I was on this list anyway
> I thought there may be someone here who knows the answer.
> I just got an SGI Onyx RE2 system and the system disk was a bit
> small for IRIX 6.5 (NetBSD not supported, right?). The problem
> is that the disk seems to be differential and my question is:
> May I stick an "normal" single ended disk in that drive sled without
> problems, or will there be flames and smoke?
> Regards
> /Christian Johansson

The base scsi chain is normally single ended, and the 2ndary one is often
differential. On the IO4 card, there are these small adaptors that plug
into the board (that in turn provide the scsi connection) that make the
scsi channel either SE or diff. One is green (SE I believe) and the other
is red (Diff).

Most desksides come with a SE chain for the SCSI cdrom.

Also, the pull out sleds have two connectors (and maybe some jumpers). In
base systems one is often SE and the other Diff. But check it before you
wreck it.

				-- Ethan
Permalink | Reply | 0 comments |
headers
Christian Smith | 9 May 2003 15:59

Re: OT: Single ended disks in Onyx system?

On Thu, 8 May 2003, Christian Johansson wrote:

>Hi!
>
>Sorry about this question being OT, but since I was on this list anyway
>I thought there may be someone here who knows the answer.
>
>I just got an SGI Onyx RE2 system and the system disk was a bit
>small for IRIX 6.5 (NetBSD not supported, right?). The problem
>is that the disk seems to be differential and my question is: 
>May I stick an "normal" single ended disk in that drive sled without
>problems, or will there be flames and smoke?

Assuming we're talking LVD (Low Voltage Differential,) then the drive
should be fine, and the bus and any other LVD devices will fall back to
being SE. LVD (MSE) SCSI was specifically designed with this compatibility
in mind, in contrast to HVD (High Voltage Differential) with was
completely incompatible with SE SCSI.

For more info, check out:
http://directory.google.com/Top/Computers/Hardware/Buses/SCSI/FAQs

>
>Regards
>
>/Christian Johansson
>

Christian
(Continue reading)

Permalink | Reply | 0 comments |
headers
Johan Danielsson | 9 May 2003 18:36
Picon
Picon
Favicon

Re: OT: Single ended disks in Onyx system?

Christian Johansson <Christian.Johansson <at> isd.se> writes:

> I just got an SGI Onyx RE2 system

It doesn't happen to have serial number S31240?

/Johan
Permalink | Reply | 0 comments |
headers
Christian Smith | 22 May 2003 13:45

Re: IP12 (Indigo R3000) boots multiuser

On Tue, 29 Apr 2003, Steve Rumble wrote:

>I've placed an ip12 kernel at
>http://www.cataclysmic.org/sgimips/netbsd.ip12

By the inability to read the MAC address from sq0, I guess this kernel 
is not ethernet capable, hence my failure to boot with NFS root?

Has there been any progress recently on this port? I'm itching to get my 
Indigo running NetBSD, but don't fancy trashing my Irix installation to 
make room (could get another disk, I suppose.)

>
>-Steve
>

Christian

--

-- 
"You have not converted a man because you have silenced him." -John Morley

----------------------------------------------------------------------
    /"\ 
    \ /    ASCII RIBBON CAMPAIGN - AGAINST HTML MAIL 
     X                           - AGAINST MS ATTACHMENTS
    / \
Permalink | Reply | 4 comments |
headers
Steve Rumble | 23 May 2003 01:24

Re: IP12 (Indigo R3000) boots multiuser

On Thu, 22 May 2003 12:45:51 +0100 (BST)
Christian Smith <csmith <at> micromuse.com> wrote:

> By the inability to read the MAC address from sq0, I guess this kernel
> is not ethernet capable, hence my failure to boot with NFS root?

It's certainly ethernet capable, but I mustn't have mentioned the lack
of nvram support and its ramifications. The ARCS prom SGIs can read the
mac address using arcbios support. The IP12 is not such a machine and
that kernel is from before such a time as I'd written the eeprom
code. I did complete it just last weekend, however. Recent source is
available at:
http://www.cataclysmic.org/sgimips/mips_sgimips-20030518-IP12-IP20.tar.gz
I'll also throw a new IP12 kernel up there
(www.cataclysmic.org/sgimips/) as well as a 1.6T userland build; just
check the dates.

> Has there been any progress recently on this port? I'm itching to get
> my Indigo running NetBSD, but don't fancy trashing my Irix
> installation to make room (could get another disk, I suppose.)

IP12 and IP20 do have some wd33c93 issues as well as a rather
significant seeq bug (or, perhaps a series of bugs) which really hinders
performance and causes hangs on certain nfs operations. If using NFS,
please experiment with the mount_nfs -r and -w options to limit read and
write size. One could include defaults in the kernel with
NFS_RSIZE/NFS_WSIZE, but that's just kludgery. On my IP20 using a
value of 1024 allowed for a full NFS boot. My IP20 also has a
propensity towards segfaulting userland binaries, especially ls. Rafal
has suggested a cache issue, but I'm not qualified to venture a guess.
(Continue reading)

Permalink | Reply | 2 comments |
headers
Rafal Boni | 23 May 2003 02:42
Picon

Re: IP12 (Indigo R3000) boots multiuser

In message <20030522182433.09a5e0ac.steve <at> paintballresource.org>, you write: 

[...]
-> Both systems can boot to a disk-based system, however a bootloader
-> doesn't exist for IP12. I've heard that the sgi volume header is around
-> 2 megabytes, so one could probably just squeeze a kernel in there pretty
-> easily. I haven't yet experimented with that. It's certainly no solution
-> as a very minimal kernel still occupies 1.6 megabytes.

I did a lot of this in the initial stages of getting IP22 support going;
it's a PITA (I had to remove sash and/or ide to fit a kernel one of my
IP22 machines), but only if you insist on keeping the IRIX install on
disk untouched -- otherwise, just reformat the disk and make a bigger
volhdr with fx(1).

If you *do* want to keep IRIX on the disk, just save the copies of the
bits you removed from the volhdr on the IRIX volume first (IIRC, it is
possible to boot IRIX without sash and ide is certainly not necessary).

--rafal

----
Rafal Boni                                                     rafal <at> attbi.com
  We are all worms.  But I do believe I am a glowworm.  -- Winston Churchill
Permalink | Reply | 0 comments |

Gmane