Subject: Re: crash!
To: Nathan Gelbard <gelbard@ENGR.ORST.EDU>
From: Steve Peurifoy <sp128@ibm.net>
List: port-hp300
Date: 10/06/1997 18:42:12
> NetBSD/hp300 1.2e + pieces of current <g>
> 
> I awoke this morning to find
> 
> panic: bad dir
> Stopped at	_Debugger0x6: unlk	a6
> db>
> 
> What does this mean, and how can I avoid it in the future?
> I had just installed perl the night before, and everything
> seemed to be working smoothly.

Is this with HP-IB or SCSI disk(s)?  What models, just out of
curiosity?

The hp300 SCSI driver is pretty fragile and there are at least two
problems in it that can cause this sort of FS corruption if your
hardware configuration is susceptible.  The first has already been
reported (PR hp300/3769), although the PR understates the potential
seriousness of the problem.  The second has to do with re-trying
target aborted commands (the driver should, but doesn't).

I only saw the latter when I tried using an MO drive on the same
bus with my system disk.  I don't know why the disk (a Seagate Hawk)
started aborting a few transfers here and there when the MO drive was
added but I'm going to try to investigate further after I (1) add
aborted command retry to the driver, and (2) have a reliable means
to recover the system.  I had to re-install my / and /usr partitions
last time.

Of course, the cause of your problem could be something entirely
different...

I can send-pr the driver change when it's done if anyone cares.

-Steve