Date: Sat, 3 Oct 2015 19:52:51 +0200 From: =?UTF-8?Q?Ermal_Lu=C3=A7i?= <eri@freebsd.org> To: Nikos Vassiliadis <nvass@gmx.com> Cc: freebsd-net <freebsd-net@freebsd.org> Subject: Re: carp on if_bridge deadlock Message-ID: <CAPBZQG3=%2BboSOy%2BfSrj5_g9SkUm0xjqf97zKp1tC08QOe44hUg@mail.gmail.com> In-Reply-To: <560F00E6.5010503@gmx.com> References: <560F00E6.5010503@gmx.com>
next in thread | previous in thread | raw e-mail | index | archive | help
This should apply https://reviews.freebsd.org/D3133 Somehow it is still pending on gnn@ for some reason! On Sat, Oct 3, 2015 at 12:10 AM, Nikos Vassiliadis <nvass@gmx.com> wrote: > Hi, > > I am trying to use carp over an if_bridge and am getting > this LOR: > >> login: lock order reversal: >> 1st 0xfffff8000848a018 if_bridge (if_bridge) @ >> /usr/src/sys/modules/if_bridge/../../net/if_bridge.c:2315 >> 2nd 0xfffff80003a58778 carp_if (carp_if) @ >> /usr/src/sys/modules/carp/../../netinet/ip_carp.c:1118 >> KDB: stack backtrace: >> db_trace_self_wrapper() at db_trace_self_wrapper+0x2b/frame >> 0xfffffe0092c3b6a0 >> witness_checkorder() at witness_checkorder+0xe7a/frame 0xfffffe0092c3b720 >> __mtx_lock_flags() at __mtx_lock_flags+0xa8/frame 0xfffffe0092c3b770 >> carp_forus() at carp_forus+0x7a/frame 0xfffffe0092c3b7b0 >> bridge_input() at bridge_input+0x338/frame 0xfffffe0092c3b820 >> ether_nh_input() at ether_nh_input+0x2ab/frame 0xfffffe0092c3b860 >> netisr_dispatch_src() at netisr_dispatch_src+0x86/frame 0xfffffe0092c3b8d0 >> ether_input() at ether_input+0x4f/frame 0xfffffe0092c3b900 >> vtnet_rxq_eof() at vtnet_rxq_eof+0x845/frame 0xfffffe0092c3b9b0 >> vtnet_rx_vq_intr() at vtnet_rx_vq_intr+0x4e/frame 0xfffffe0092c3b9e0 >> intr_event_execute_handlers() at intr_event_execute_handlers+0xe4/frame >> 0xfffffe0092c3ba20 >> ithread_loop() at ithread_loop+0xa6/frame 0xfffffe0092c3ba70 >> fork_exit() at fork_exit+0x84/frame 0xfffffe0092c3bab0 >> fork_trampoline() at fork_trampoline+0xe/frame 0xfffffe0092c3bab0 >> --- trap 0, rip = 0, rsp = 0, rbp = 0 --- >> > > Eventually all network activity will stop and this will happen: > >> load: 0.33 cmd: ifconfig 665 [*carp_if] 228.29r 0.00u 0.01s 0% 2716k >> > Or: > >> load: 2.86 cmd: ifconfig 720 [running] 104.92r 0.00u 36.53s 100% 2716k >> > > A single ping to the carp address is enough to trigger the problem. > The debugger says: > >> db> show all locks >> Process 669 (sysctl) thread 0xfffff800084269a0 (100082) >> exclusive sleep mutex Giant (Giant) r = 0 (0xffffffff81cc4e90) locked @ >> /usr/src/sys/kern/kern_sysctl.c:164 >> Process 668 (ifconfig) thread 0xfffff8000842e9a0 (100079) >> exclusive sx carp_sx (carp_sx) r = 0 (0xffffffff820b0ac0) locked @ >> /usr/src/sys/modules/carp/../../netinet/ip_carp.c:1644 >> Process 12 (intr) thread 0xfffff800038a89a0 (100010) >> exclusive sleep mutex carp_softc (carp_softc) r = 0 (0xfffff80008266508) >> locked @ /usr/src/sys/kern/kern_mutex.c:158 >> Process 12 (intr) thread 0xfffff80003a679a0 (100031) >> exclusive sleep mutex carp_if (carp_if) r = 0 (0xfffff8000825b078) locked >> @ /usr/src/sys/modules/carp/../../netinet/ip_carp.c:1118 >> exclusive sleep mutex if_bridge (if_bridge) r = 0 (0xfffff80008467018) >> locked @ /usr/src/sys/modules/if_bridge/../../net/if_bridge.c:2315 >> db> >> > > There is a related PR > https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=200319 > But I couldn't apply cleanly the patch to HEAD or 10-STABLE. > > Thanks for any insights, > Nikos > > _______________________________________________ > freebsd-net@freebsd.org mailing list > https://lists.freebsd.org/mailman/listinfo/freebsd-net > To unsubscribe, send any mail to "freebsd-net-unsubscribe@freebsd.org" > > -- > Ermal >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAPBZQG3=%2BboSOy%2BfSrj5_g9SkUm0xjqf97zKp1tC08QOe44hUg>