Subject: Re: panic with named on -current (panic: looutput: no header mbuf)
To: Steven M. Bellovin <email@example.com>
From: Pavel Cahyna <firstname.lastname@example.org>
Date: 08/16/2004 10:25:52
>>>>I got a nice panic today, triggered by named, sources are from 14. Aug,
>>>>around 18:00 CEST. Everything is fine when I do not enable/try to run named!
>>>I've seen and reported the same thing. I have no fix, though I wonder
>>>if it's a combination of ipv6 and some unusual hardware -- my machine
>>>has a bge0 NIC.
>>I think that it has to do with ipv6. Is there a pr for it?
> I have not filed one, since I'm not able to verify or provide further
> information for a few weeks. But I doubt that it's only ipv6 that
> triggers this or we'd see a lot more people experiencing the problem.
I had a similar problem, which also involves ICMPv6 and it looks
like some mbuf corruption. When I try flood-ping6 to a link-local
address, the target machine usually panics quite fast.
------ cut here
panic: m_copym0 overrun
cpu_Debugger(181c,0,80fe,c0d0af00,c0d0b400) at netbsd:cpu_Debugger+0x4
panic(c06ad358,c0d0b420,5d0,c0d0b400,0) at netbsd:panic+0x11d
m_copym0(c0d0b400,0,3b9aca00,1,0) at netbsd:m_copym0+0x325
m_copym(c0d0a400,0,3b9aca00,1,c0d14600) at netbsd:m_copym+0x19
icmp6_input(c0aa4e50,c0aa4e1c,3a,0,6e) at netbsd:icmp6_input+0x4ac
ip6_input(c0d0a400,c0be1700,0,7,0) at netbsd:ip6_input+0x418
ip6intr(a0440010,30,5f10010,10,c0aa1000) at netbsd:ip6intr+0x71
DDB lost frame for netbsd:Xsoftnet+0x4d, trying 0xc0aa4e70
Xsoftnet() at netbsd:Xsoftnet+0x4d
------ cut here
Target machine is running 2.0_BETA, from quite old sources (that's why
I haven't reported it yet). The NIC is wi in hostap mode.