Distribution: Hardware Environment: Alpha 164LX 21164a 600MHz Software Environment: Debian unstable Problem Description: During bootup, I am getting multiple instances ofitems like this, typically when a daemon starts up: Jan 12 17:23:38 erwin kernel: RTNL: assertion failed at net/ipv6/addrconf.c (1992) Trace shows: Jan 12 17:23:38 erwin kernel: Trace: Jan 12 17:23:38 erwin kernel: [sk_free+324/336] sk_free+0x144/0x150 Jan 12 17:23:38 erwin kernel: [inet_release+128/160] inet_release+0x80/0xa0 Jan 12 17:23:38 erwin kernel: [sock_release+172/240] sock_release+0xac/0xf0 Jan 12 17:23:38 erwin kernel: [sock_close+48/112] sock_close+0x30/0x70 Jan 12 17:23:38 erwin kernel: [__fput+412/448] __fput+0x19c/0x1c0 Jan 12 17:23:38 erwin kernel: [fput+72/96] fput+0x48/0x60 Jan 12 17:23:38 erwin kernel: [filp_close+132/240] filp_close+0x84/0xf0 Jan 12 17:23:38 erwin kernel: [close_files+172/224] close_files+0xac/0xe0 Jan 12 17:23:38 erwin kernel: [put_files_struct+84/224] put_files_struct+0x54/0xe0 Jan 12 17:23:39 erwin kernel: [do_exit+508/1424] do_exit+0x1fc/0x590 Jan 12 17:23:39 erwin kernel: [do_group_exit+60/176] do_group_exit+0x3c/0xb0 Jan 12 17:23:39 erwin kernel: [entSys+164/192] entSys+0xa4/0xc0 The system appears to function normally after bootup. However, when it comes time to halt the system, I cannot. It stalls with: unregister_netdevice: waiting for sit1 to become free. Usage count = 1 This machine is acting as a firewall. I have two Ethernet devices in it; one for the outside and one for the inside. The outside speaks only IPv4. Inside speaks IPv4 and IPv6. IPv6 is routed to the outside via sit1, using the public 6to4 router address. I will attach dmesg logs, ksymoops output, configurations, etc. This behavior was observed on both 2.6.9 and 2.6.10. Please ask if there is any other information you would need to resolve this problem. I am quite happy to provide anything that may be of use.
Created attachment 4378 [details] 2.6.9 config file
Created attachment 4379 [details] dmesg log, 2.6.9
Created attachment 4380 [details] 2.6.10 dmesg log
ksymoops did not appear to add anything useful to this. Again, please ask if I can help with debugging.
I've seen the "halting" occur on a Linux 2.6.11 SMP machine as well, the following message printed to the console when attempting to reboot: unregister_netdevice: waiting for sit1 to become free. Usage count = 1 I do not see the oops though on bootup. I believe this problem can only be reproduced on certain machines. I see it occur on a dual PIII-550mhz system, but not on a dual P4 2.4Ghz Xeon. To cause this problem, just enable both sit0 and sit1, and try to reboot. When it fails on a system, it fails quite reliably.
Is this issue still present in kernel 2.6.18?
I do think this is acient and it is already fixed.