Subject: Re: Keyboard can cause boot troubles
To: None <email@example.com>
From: Greg Earle <earle@isolar.Tujunga.CA.US>
Date: 12/08/1994 03:05:01
> It seems that if you plug in a newer (i.e., SPARClassic style)
> keyboard into an older sun4c (like my SPARCstation SLC 4/20) at boot time,
> the machine can hang at:
> found boot device sd0
> One has to break to the PROM and then just type 'go' to get it out of the rut.
Uh, is this with -current or with 1.0?
I am using a Type-5 keyboard with a hoary old wheezing IPC and have had no
trouble of this sort whatsoever under 1.0.
Speaking of hangs/wedges/crashes etc., I have found that using LBX can
reliably crash the machine almost all the time with the same problem as before;
i.e. it drops into the PROM with
Instruction Access Exception
Yesterday there was a new one, namely
Memory Address Alignment Error (or somesuch)
I've now built a kernel with "options DDB" enabled (as well as - you guessed
it - "options UCONSOLE" bwah hah hah (-: ).
Before I go mucking about with it and checking out this new boot blocks
situation etc., however, a question - can DDB catch a Watchdog reset?!?
It's been a long while since I've seen one under SunOS, but in my recollection
they were usually caused by double bus errors (mostly due to flaky hardware;
e.g., I once had a Sun-3/160 that got a rash of 'em all of a sudden ... ). I
thought those dropped you straight into the Boot PROM; i.e. I thought those
were caught "down there", kinda like a machine-level SIGKILL :-)
In other words, is it worth my while to use a DDB-enabled kernel to try and
catch this kind of problem, if it results in being dropped into the PROM?
(I never remembered whether booting under "kadb" in SunOS would allow one to
"catch" a "Watchdog reset", to be honest, but I would have doubted it.)