From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.8b4) Gecko/20050915 Fedora/1.5-0.5.0.beta1 Firefox/1.4 Description of problem: I've been getting oopses on reboot with the last few rawhide kernels. What I see on screen after the box takes too long to reboot goes like this. Part of the message has already scrolled out, unfortunately. Typos are all mine. Call Trace:<ffffffff88308c29>{:ip_conntrack:ip_conntrack_cleanup+20} <ffffffff88306ab4>{:ip_conntrack:init_or_cleanup+676} <ffffffff80157798>{sys_delete_module+490} <ffffffff801118aa>{syscall_trace_enter+217> <ffffffff8010eb6c>{tracesys+209} Code: f0 ff 0a 0f 94 c0 84 c0 74 0f 48 89 d7 ff 52 00 48 8b 35 d7 RIP <ffffffff88308b8a>{:ip_conntrack:ip_conntrack_flush+74} RSP <ffff81001b1b9eb8> <0>Kernel panic - not syncing: Oops Another trace appears below, with panic+133, vgacon_cursor+219, redraw_screen+399, vgacon_cursor+219, oops_end+71, die+66, do_general_protection+257, error_exit+0, :ip_conntrack:ip_conntrack_flush+74, :ip_conntrack:ip_conntrack_flush+11, :ip_conntrack:ip_conntrack_cleanup+20, :ip_conntrack:init_or_cleanup+676, sys_delete_module+490, syscall_trace_enter+217, tracesys+209 Version-Release number of selected component (if applicable): kernel-2.6.13-1.1592_FC5 (and also 1588_FC5) How reproducible: Sometimes Steps to Reproduce: 1.Reboot an Athlon64 notebook running IPv4 only, iptables, a single network interface plus a bunch of virtual tun interfaces, some used by vpnc, some used by openvpn. Actual Results: The oops above Expected Results: Should reboot cleanly Additional info:
Are you running an SMP kernel by chance?
in rawhide, there is no UP x86-64 kernel any more.
Ok, that's a good clue then. I think the per-cpu walking in ip_ct_event_cache_flush() is suspect. The "for_each_cpu()" should be "for_each_online_cpu()". Let me consult the netfilter folks, who unfortunately right now are not very responsive since the netfilter summit is about this time. :-)
Is this still happening on the latest builds ?
Nope, it appears to have been fixed a while ago.