From: Marcelo Ricardo Leitner Date: Fri, 22 Dec 2017 19:04:37 +0000 Subject: Re: INFO: task hung in bpf_exit_net Message-Id: <20171222190437.GC6123@localhost.localdomain> List-Id: References: <001a1143fd00a8cc790560b0b552@google.com> <20171222182807.GC32765@localhost.localdomain> In-Reply-To: <20171222182807.GC32765@localhost.localdomain> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Dmitry Vyukov Cc: David Ahern , syzbot , LKML , Ingo Molnar , Peter Zijlstra , syzkaller-bugs@googlegroups.com, David Miller , Florian Westphal , Daniel Borkmann , Xin Long , jakub.kicinski@netronome.com, mschiffer@universe-factory.net, Vladislav Yasevich , Jiri Benc , netdev , Neil Horman , linux-sctp@vger.kernel.org On Fri, Dec 22, 2017 at 04:28:07PM -0200, Marcelo Ricardo Leitner wrote: > On Fri, Dec 22, 2017 at 11:58:08AM +0100, Dmitry Vyukov wrote: > ... > > > Same with this one, perhaps related to / fixed by: > > > http://patchwork.ozlabs.org/patch/850957/ > > > > > > > > > > > Looking at the log, this one seems to be an infinite loop in SCTP code > > with console output in it. Kernel is busy printing gazilion of: > > > > [ 176.491099] sctp: sctp_transport_update_pmtu: Reported pmtu 508 too > > low, using default minimum of 512 > > ** 110 printk messages dropped ** > > [ 176.503409] sctp: sctp_transport_update_pmtu: Reported pmtu 508 too > > low, using default minimum of 512 > > ** 103 printk messages dropped ** > > ... > > [ 246.742374] sctp: sctp_transport_update_pmtu: Reported pmtu 508 too > > low, using default minimum of 512 > > [ 246.742484] sctp: sctp_transport_update_pmtu: Reported pmtu 508 too > > low, using default minimum of 512 > > [ 246.742590] sctp: sctp_transport_update_pmtu: Reported pmtu 508 too > > low, using default minimum of 512 > > > > Looks like a different issue. > > > > Oh. I guess this is caused by the interface having a MTU smaller than > SCTP_DEFAULT_MINSEGMENT (512), as the icmp frag needed handler > (sctp_icmp_frag_needed) will trigger an instant retransmission. > But as the MTU is smaller, SCTP won't update it, but will issue the > retransmission anyway. > > I will test this soon. Should be fairly easy to trigger it. Reproduced it. netns A veth0(1500) - veth1(1500) B veth2(508) - veth3(508) C When A sends a sctp packet bigger than 508, it triggers the issue as B will reply a icmp frag needed with a size that sctp won't accept but will retransmit anyway. Marcelo