Re: panic during 'cvs -R -d /home/dcvs/ checkout src'

[Chris Pressey <cpressey@xxxxxxxxxxxxxxx>]

On Mon, 26 Jan 2004 20:35:41 -0800 (PST)
Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx> wrote:

> :The machine's still sitting at that prompt, and I can leave it there
> :pretty much indefinately, so if anyone can make suggestions on how to
> :get useful information out of DDB, please do!  Meanwhile I'm going to
> :read up on chapter 10 of the FreeBSD Developer's Handbook because
> it's:been quite a while since I've used a debugger in any
> seriousness...:
> :-Chris
> 
>     Ok, first thing to do is 'show regi'  (show registers).  Pick the
>     stack pointer and write it down.

esp	0xc9123bdc

>     Then do a 'trace'.

  db> trace
  Debugger(c9123bfc,c017142b,c02d3b4d,1010002,1000040) at Debugger+0x34
  dumpmag(73252228,a2922,0,0,0) at dumpmag+0x1b9

I'm guessing that's not helpful.

>     If the trace doesn't look good, try tracing with
>     the stack pointer plus an offset of 4 and of 8.  You should get a
>     good trace.   e.g. if the stack pointer were 0xdaaaaa08  you could
>     try'trace 0xdaaaaa0c' and 'trace 0xdaaaaa10'.

  db> trace 0xc9123be0
  _end(c017142b,c02d3b4d,1010002,1000040) at 0xc9123bfc
  _end(6f726577,645f6666,79616c65,72706b00,735f636f) at 0xc9123bfc

Surely that's also not useful.

Third time's the charm.  I hope.

  db> trace 0xc9123be4
  panic(c02d3b4d) at panic+0x97
  panic(c02d1fc0,10002,c94adfc0,c94ae02c,c9123c2c) at panic+0x97
  lockmgr(c94c1a90,1010002,c94ae02c,c4824b00,c9123c38) at lockmgr+0x360
  vop_stdlock(c9123c50,c9123c60,c01aa898,c9123c50,c94adfc0) at
	vop_stdlock+0x3b
  ufs_vnoperate(c9123c50) at ufs_vnoperate+0x15
  vn_lock(c94adfc0,10002,c4824b00,c94adfc0,c94ae02c) at vn_lock+0x90
  vget(c94adfc0,10002,c4824b00) at vget+0x97
  ufs_ihashget(c4c45f78,c505,c9740620,c9740620,c4824b00) at
	ufs_ihashget+0xaf
  ffs_vget(c9103530,c505,c9123cf4) at ffs_vget+0x22
  handle_workitem_remove(c9740620,c4824b00,0,0,0) at
	handle_workitem_remove+0x28
  process_worklist_item(0,0,0,0,c0b17ce4) at process_worklist_item+0x10d
  softdep_process_worklist(0) at softdep_process_worklist+0x106
  sched_sync(0,0,0,0,0) at sched_sync+0x17c
  kthread_exit() at kthread_exit
  db>

>     If you can get a crash dump that would be good to, typing 'panic'
>     and hitting return once or twice will get you there. but if you
>     haven't already set it up there nothing you can do about it now
>     and that will reboot the machine.

Yeah, unfortunately not... I'll set it up for next time and see if I
can't reproduce this by running cvs checkout again.

>     I've seen this type of crash before, but have not yet been able to
>     track it down.
> 
> 					-Matt
> 					Matthew Dillon 
> 					<dillon@xxxxxxxxxxxxx>

Well, I hope this helps in some small way.  Thanks for guiding me
through it.

-Chris