Subject: Upgrade to a releng build gone bad: How does one boot single user from the boot prom?
To: None <port-sgimips@NetBSD.org>
From: Tillman Hodgson <tillman@seekingfire.com>
List: port-sgimips
Date: 07/30/2004 12:21:13
Howdy folks,

Well, my test O2 upgraded to a recent -current build just fine, but the
real O2 choked. It boots the old kernel and then ends up with this:

Jul 30 18:12:34 init: can't exec getty '/usr/libexec/getty' for port /dev/console: No such file or directory
Jul 30 18:13:05 init: can't exec getty '/usr/libexec/getty' for port /dev/console: No such file or directory

etc etc. I'm assuming that a change in /dev or /etc/ttys from the new
world (and the accompanying etcupdate) makes it unhappy.

When I try to boot the new kernel (by changing the
OSLoadFilename variable):

> boot
Cannot load scsi(0)disk(1)rdisk(0)partition(8)/boot.
Text start 0x2000, size 0xb7b0 doesn't fit in a FreeMemory area.
Unable to execute scsi(0)disk(1)rdisk(0)partition(8)/boot:  not enough space
Unable to load scsi(0)disk(1)rdisk(0)partition(8)/boot: not enough space

That doesn't look happy.

What does the kernname environment variable mean? I've seen it in my
google searches and it looks relevent, though I thought that
OSLoadFilename was correct variable for that sort of thing.

Have the boot PROM env variable requirements changed between 1.6.x and
2.0-current at all? And how does one boot single user from the boot PROM
anyway?

Thanks muchly,

-T


-- 
"What is Zen?"
"Not always so."
    - Shunryu Suzuki