Subject: Re: kernel stack traceback
To: Paul H. Anderson <>
From: Bill Studenmund <>
List: tech-kern
Date: 03/19/1999 11:36:06
On Fri, 19 Mar 1999, Paul H. Anderson wrote:

> On Thu, 18 Mar 1999, Bill Studenmund wrote:
> I can relatively easily reproduce the crash.  This morning, I did newfs on
> my ccd filesystem (-b 32768 -f 8192) and ran a stress test that caused a
> panic after about an hour.  The locks below come from this second panic
> (not the one that gave the above traceback).  The second panic looks like
> the first (same message), but it is still at the ddb prompt on the serial
> console, and the alpha doesn't do kernel stack tracebacks yet, so I can't
> confirm that it is exactly the same - tell me what you want to get, and
> I'll do it.
> > Does ddb (the in-kernel debugger) work on alpha? If so, call
> > printlockedvnodes() before trying to reboot. I think it should tell you
> > where/how the node got locked.
> Here's the locks:
> charger# sh
> ... some minor output deleted
> successful open, length=0, mapdata=0
> panic: lockmgr: locking against myself
> Stopped in tar at       Debugger+0x4:   ret     zero,(ra)
> db> call printlockedvnodes

I'd say have it drop core, and then get a stack trace from gdb. :-)

Can you send me that script?

Take care,