linux-2.6.0-test 5 crashes reproducable with a tun / ipv6 setup
From: Thomas Glanzmann
Date: Sun Sep 21 2003 - 17:09:28 EST
Hello together,
I am using linux-2.6.0-test5 together with a tun / openvpn[1] setup.
openvpn is a userland application which connects two tun devices on
different machines using udp or tcp packets. We use it at FAU[2] to
provide IPv6 tunnels to our computer science students.
linux-2.6.0-test5 crashes if I try to ping an IPv6 client via a tun
device, when the IPv6 address on the clients (peers) side is
misconfigured. If correctly configured everything works as expected.
skput:under: c039b9db:206 put:14 dev:lo------------[ cut here ]------------
kernel BUG at net/core/skbuff.c:103!
invalid operand: 0000 [#1]
CPU: 3
EIP: 0060:[<c0341ae3>] Not tainted
EFLAGS: 00010246
EIP is at skb_under_panic+0x2b/0x38
eax: 0000002a ebx: c039b9db ecx: c379d0c0 edx: c0483944
esi: d46f6ba8 edi: d3e49800 ebp: d4a1be80 esp: f6b19d64
ds: 007b es: 007b ss: 0068
Process openvpn (pid: 648, threadinfo=f6b18000 task=f6a1ece0)
Stack: c045be80 c039b9db 000000ce 0000000e c04cd9e0 d46f6b80 c039b9e3 d4a1be80
0000000e c039b9db c04de90c f72d6d80 f6b19e24 f6b19e34 f6b18000 c039ba8a
d4a1be80 c03acfdc d4a1be80 f554be1a d4a1b180 f6d10980 d3e49858 d4a1be80
Call Trace:
[<c039b9db>] ip6_output2+0x1af/0x228
[<c039b9e3>] ip6_output2+0x1b7/0x228
[<c039b9db>] ip6_output2+0x1af/0x228
[<c039ba8a>] ip6_output+0x36/0x3c
[<c03acfdc>] ndisc_send_redirect+0x39c/0x480
[<c039d043>] ip6_forward+0x157/0x310
[<c039ea49>] ipv6_rcv+0x199/0x230
[<c03461db>] netif_receive_skb+0x20b/0x260
[<c03462b9>] process_backlog+0x89/0x118
[<c03463ce>] net_rx_action+0x86/0x140
[<c01249eb>] do_softirq+0x6b/0xd8
[<c02baa56>] tun_chr_writev+0x18e/0x1a0
[<c015415b>] do_readv_writev+0x1ab/0x254
[<c02baa68>] tun_chr_write+0x0/0x30
[<c016502e>] select_bits_free+0xa/0x10
[<c0165495>] sys_select+0x461/0x470
[<c0154297>] vfs_writev+0x4b/0x50
[<c0154319>] sys_writev+0x31/0x4c
[<c01090cf>] syscall_call+0x7/0xb
Code: 0f 0b 67 00 63 be 45 c0 83 c4 14 5b c3 55 57 56 53 8b 5c 24
<0>Kernel panic: Fatal exception in interrupt
In interrupt handler - not syncing
ping6 <clients ipv6address here> from the 'server' machine triggers the
bug. This is reproducable if something is unclear feel free to ask. :-)
Greetings,
Thomas
[1] OpenVPN
http://openvpn.sf.net/
[2] FAU - Friedrich Alexander University
http://www.uni-erlangen.de/
--
Thomas Glanzmann ++49 (0) 9131 85-27574 Department of Computer Science III
Martensstrasse 3 D-91058 Erlangen Germany University of Erlangen-Nuremberg
http://www3.informatik.uni-erlangen.de/Research/FAUmachine/
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/