Subject: Re: Random lockups under -current
To: None <port-i386@netbsd.org>
From: Geoff Wing <mason@primenet.com.au>
List: port-i386
Date: 09/18/1998 05:36:22
Dave Huang <khym@bga.com> typed:
:Has anyone else been experiencing random lockups? I've got a Pentium
:with 64MB RAM, 53c810 and 53c875 SCSI adapters running NetBSD-current,
:and for the past maybe 3 weeks or so, it's been locking up for no
:apparent reason. At first, I was in X and thought maybe something
:happened to drop it into ddb, so I quit running X and took DDB out of my
:kernel. Then when it locked up, I found I couldn't even switch VTs (I
:was using wscons for a while, but switched back to pcvt to see if that
:was the problem). I was usually in the wrong VT to see any kernel
:messages, but it did finally lock up while I was looking at vt0, and
:there weren't any messages at all :( I reset the machine, and a few
:times, it'd lock up again while fscking the filesystems, or later on in
:/etc/rc.
:I put DDB back into my kernel, and stayed out of X for a while, and I've
:found that when it locks up, I can't Ctrl-Alt-Esc into DDB. So, I'm
:afraid I can't give much more info about it... it just locks up :) It
:might be correlated with disk access, although it's hard to tell. I've
:never had it lock up while idle; it's usually during compiles and fscks.

Yes, I've had it on two different machines:
1) My old machine:
   i) Pentium-120 (80MB) ATA drives
  ii) Same with Adaptec 1540CP & Quantum SCSI added (_seemed_ to occur
      more frequently) - at one stage here I was nearly always rebooting
      about two-five minutes after it had finished processing /etc/rc
      (which was another thing to inspire me to spend stacks of money on
      my new machine :-) )
2) PentiumII-400 (256MB-ECC) w/different ATA drives (has occured once in 
   just over a month)

They don't respond to anything in the lock up.  They're not pingable.
At the time I thought it was from disk access - I presumed (from my minimal
knowledge of the internals here) that it had made a request to the HD and
was waiting for fulfillment.

It started happening four or five months back (I think).

Either way, it's hard/impossible to diagnose on a normal kernel.  
-- 
Geoff Wing   <gcw@pobox.com>            Mobile : 0412 162 441
Work URL: http://www.primenet.com.au/   Ego URL: http://pobox.com/~gcw/