Subject: kern/26736: 'looutput: no header mbuf' with today's 2.0beta
To: None <gnats-bugs@gnats.NetBSD.org>
From: None <smb@research.att.com>
List: netbsd-bugs
Date: 08/22/2004 23:52:33
>Number: 26736
>Category: kern
>Synopsis: 'looutput: no header mbuf' when named starts
>Confidential: no
>Severity: critical
>Priority: high
>Responsible: kern-bug-people
>State: open
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Mon Aug 23 03:53:00 UTC 2004
>Closed-Date:
>Last-Modified:
>Originator: Steven M. Bellovin
>Release: NetBSD 2.0_BETA
>Organization:
>Environment:
System: NetBSD hudson.machshav.com 2.0_BETA NetBSD 2.0_BETA (HUDSON) #0: Sun Aug 22 22:48:24 EDT 2004 smb@hudson.machshav.com:/usr/BUILD/obj/sys/arch/i386/compile/HUDSON i386
Architecture: i386
Machine: i386
>Description:
With a kernel from today's 2.0beta source, I get a
panic ('looutput: no header mbuf') when named starts up.
I believe that this is the same as kern/26666, but I'm now seeing
it on 2.0beta as well as on -current. It is almost certainly tied
to ipfilter, since on -current I can bypass the problem by not
starting ipfilter. I do not see the problem on a 2.0beta kernel
from 8 August. (Of course, I really want the newer ipfilter, since
it has icmpmapid, which the 8 August version does not.) Note that
I'm seeing this with only ipnat enabled, and without starting
ipfilter per se. (On -current, I'm seeing it with ipfilter and
not ipnat.)
>How-To-Repeat:
See above. I can send my ipnat.conf if you want. I do not have
a core dump for 2.0beta, possibly because the machine is RAID-based.
>Fix:
The only bypass I know is to avoid running ipfilter -- but on
this machine, I need to run it...
>Release-Note:
>Audit-Trail:
>Unformatted: