[ovs-discuss] SIGSEGV in nl_attr_get_size() - OVS 2.0.2

Ben Pfaff blp at nicira.com
Sun Jan 11 18:05:41 UTC 2015


On Thu, Jan 08, 2015 at 09:15:17PM +0000, James Page wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA256
> 
> Hi Ben
> 
> On 07/01/15 16:18, Ben Pfaff wrote:
> > Thanks.  In the past I've tried but I've been unable to track down
> > the root cause.  I also suspected some kind of compiler bug or
> > strange interaction.
> 
> I've still not determined whether this is an optimizer issue, but we
> have now seen this problem across all 6 server in our Ubuntu QA cloud;
> this appears to be a problem that builds up over time - the tenant
> that sees this first is our automated testing account which
> creates/deletes alot of instances over time.
> 
> Across all nodes I see:
> 
> 2015-01-08T13:36:07.822Z|80336|dpif|WARN|system at ovs-system: failed to
> flow_del (No such file or directory)
> skb_priority(0),in_port(4),skb_mark(0),eth(src=ae:c4:93:ba:85:3e,dst=33:33:ff:5f:c4:32),eth_type(0x86dd),ipv6(src=::,dst=ff02::1:ff5f:c432,label=0,proto=58,tclass=0,hlimit=255,frag=no),icmpv6(type=135,code=0),nd(target=fe80::8412:d1ff:fe5f:c432)
> 
> message alot; which is related to
> https://github.com/openvswitch/ovs/commit/3601bd879 which we don't
> have in the 3.13 kernel for trusty.
> 
> I'm wondering whether this is related - the stacktrace relates to flow
> misses and has a large parameter list in the handle_flow_misses call.
> 
> The optimized code may just be obfuscating the actual problem.

Any chance you could try a build without optimization, or without
optimization in the source files at issue?



More information about the discuss mailing list