This is a repost of bug from 2006-06-13 18:27, which got lost due to the Bugzilla database failure. Description of problem: On a HP DL385 Dual Core Opteron server running Fedora Core 4, I've upgraded the kernel to version 2.6.16-1.2111_FC4smp last sunday. Now it outputs the following error in kernel log every 20-30 minutes: Jun 13 04:53:11 hostname kernel: Debug: sleeping function called from invalid context at include/linux/rwsem.h:43 Jun 13 04:53:11 hostname kernel: in_atomic():1, irqs_disabled():0 Jun 13 04:53:11 hostname kernel: Jun 13 04:53:11 hostname kernel: Call Trace: <IRQ> <ffffffff8015b35a>{audit_log_exit+416} Jun 13 04:53:11 hostname kernel: <ffffffff8013a4f6>{getnstimeofday+16} <ffffffff8015c773>{audit_free+282} Jun 13 04:53:11 hostname kernel: <ffffffff8013280b>{__put_task_struct_cb+205} <ffffffff80147b0e>{__rcu_process_callbacks+303} Jun 13 04:53:11 hostname kernel: <ffffffff80147bbe>{rcu_process_callbacks+35} <ffffffff8013ad8d>{tasklet_action+102} Jun 13 04:53:11 hostname kernel: <ffffffff8013aedd>{__do_softirq+96} <ffffffff8010be46>{call_softirq+30} Jun 13 04:53:11 hostname kernel: <ffffffff8010cdcc>{do_softirq+44} <ffffffff8010b7a0>{apic_timer_interrupt+132} <EOI> Jun 13 04:53:11 hostname kernel: <ffffffff881921a0>{:ip_conntrack:udp_invert_tuple+0} Jun 13 04:53:11 hostname kernel: <ffffffff801f0a5b>{security_port_sid+68} <ffffffff801f0a35>{security_port_sid+30} Jun 13 04:53:11 hostname kernel: <ffffffff801e86eb>{selinux_ip_postroute_last+428} <ffffffff80305911>{nf_iterate+89} Jun 13 04:53:11 hostname kernel: <ffffffff803110f2>{ip_finish_output+0} <ffffffff80305a41>{nf_hook_slow+95} Jun 13 04:53:11 hostname kernel: <ffffffff803110f2>{ip_finish_output+0} <ffffffff8030e5e0>{dst_output+0} Jun 13 04:53:11 hostname kernel: <ffffffff803117f3>{ip_output+166} <ffffffff80310697>{ip_push_pending_frames+912} Jun 13 04:53:11 hostname kernel: <ffffffff8032a7b4>{udp_push_pending_frames+593} <ffffffff8032b256>{udp_sendmsg+1520} Jun 13 04:53:11 hostname kernel: <ffffffff802e473c>{sock_sendmsg+240} <ffffffff8014a390>{autoremove_wake_function+0} Jun 13 04:53:11 hostname kernel: <ffffffff8016c8bd>{do_wp_page+720} <ffffffff8015bb00>{audit_sockaddr+83} Jun 13 04:53:11 hostname kernel: <ffffffff802e4ace>{sys_sendto+235} <ffffffff8015c52b>{audit_syscall_entry+305} Jun 13 04:53:11 hostname kernel: <ffffffff8010ab71>{tracesys+209} Version-Release number of selected component (if applicable): kernel-2.6.16-1.2111_FC4.x86_64 The system works fine, but the message looks suspicious. The message looks the same like in bug 180108, but: 1) I'm not using fuse or any other non-typical filesystems 2) Bug 180108 was supposedly fixed in version 2.6.16-1.2096_FC4, and I'm running a newer one.
the audit layer got a large reworking in 2.6.17. I'm pretty confident updating to the update I pushed out last week will fix it, as I was able to trigger a similar stack trace myself during .17 development. Feel free to reopen if you manage to trigger the same stack trace under the latest errata. Thanks.
I cannot see the SMP 2.6.17 kernel anywhere: http://download.fedora.redhat.com/pub/fedora/linux/core/updates/4/x86_64/ There's only kernel-2.6.17-1.2139_FC4.x86_64.rpm, no kernel-smp.