Subject: Re: Another softdep related panic
To: None <current-users@netbsd.org>
From: Bernd Sieker <bsieker@techfak.uni-bielefeld.de>
List: current-users
Date: 02/10/2000 18:16:04
On 09.02.00, 22:55:28, Steve Peurifoy wrote:
> 
> In preparation for building from more recent sources, I did
> 
> cd /usr/src
> rm -rf *
> 
> on my HP9000/375 test machine running 1.4P from 2000/01/14.  I had
> softdeps enabled on /usr.  A short time later I was rewarded with
> 
>   panic: handle_workitem_remove: bad file delta
>   Stopped in ioflush at   _cpu_Debugger+0x6:      unlk    a6
>   db> t
>   _cpu_Debugger(0,6f,0,63192c,519f28) + 6
>   _panic(fad51,1cc73c0,0,121b54,13d17c) + 52
>   _handle_workitem_remove(1cc73c0) + 62
>   _softdep_process_worklist(0) + 10e
>   _sched_sync(0) + 1b2
>   _proc_trampoline() + 2
>   db>
> 
> Lots o' stuff was reconnected by fsck upon reboot.

I regularly get the following panic when shutting down the system
(1.4R, not yet tested with 1.4S) with softdeps enabled (As of this
time I use softdeps only on non-critical file systems, But I'm quite
impressed with its performance :) ):

This happens after syncinc the disks, so fsck does not find any
inconsistencies at bootup, it just takes some minutes more than I
like.

  syncing disks... 6 5 done
  panic: insmntque into dying filesystem

> 
> -Steve

-- 
Bernd Sieker

Unix, the solution to the W2k problem.