ipv6 broken for bridge

[Markus Heberling]

ipv6 seems to be broken on bridges in oneiric:

root@office:/home/markus# uname -a
Linux office 3.0.0-12-virtual #20-Ubuntu SMP Fri Oct 7 18:25:20 UTC 2011 i686 i686 i386 GNU/Linux
root@office:/home/markus# brctl addbr test
root@office:/home/markus# ip -6 addr add 2001:DB8:4309:2388::2/64 dev test
root@office:/home/markus# ifconfig test up
root@office:/home/markus# ifconfig test
test Link encap:Ethernet HWaddr fa:12:c8:34:79:69
          inet6 addr: 2001:db8:4309:2388::2/64 Scope:Global
          UP BROADCAST MULTICAST MTU:1500 Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:0 (0.0 B) TX bytes:0 (0.0 B)
root@office:/home/markus# ping6 2001:DB8:4309:2388::2
PING 2001:DB8:4309:2388::2(2001:db8:4309:2388::2) 56 data bytes
^C
--- 2001:DB8:4309:2388::2 ping statistics ---
7 packets transmitted, 0 received, 100% packet loss, time 5999ms

root@office:/home/markus#

The address is correctly assigned to the interface, but it can't be pinged.

This used to work in natty:

root@yavin:/home/markus# uname -a
Linux yavin 2.6.38-12-virtual #51-Ubuntu SMP Wed Sep 28 17:57:49 UTC 2011 i686 i686 i386 GNU/Linux
root@yavin:/home/markus# brctl addbr test
root@yavin:/home/markus# ip -6 addr add 2001:DB8:4309:2388::2/64 dev test
root@yavin:/home/markus# ifconfig test up
root@yavin:/home/markus# ifconfig test
test Link encap:Ethernet HWaddr 0a:38:75:60:be:f1
          inet6 addr: fe80::838:75ff:fe60:bef1/64 Scope:Link
          inet6 addr: 2001:db8:4309:2388::2/64 Scope:Global
          UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:17 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:0 (0.0 B) TX bytes:1446 (1.4 KB)
root@yavin:/home/markus# ping6 2001:DB8:4309:2388::2
PING 2001:DB8:4309:2388::2(2001:db8:4309:2388::2) 56 data bytes
64 bytes from 2001:db8:4309:2388::2: icmp_seq=1 ttl=64 time=0.035 ms
64 bytes from 2001:db8:4309:2388::2: icmp_seq=2 ttl=64 time=0.034 ms
64 bytes from 2001:db8:4309:2388::2: icmp_seq=3 ttl=64 time=0.034 ms
^C
--- 2001:DB8:4309:2388::2 ping statistics ---
3 packets transmitted, 3 received, 0% packet loss, time 1999ms
rtt min/avg/max/mdev = 0.034/0.034/0.035/0.004 ms

root@yavin:/home/markus#

[Markus Heberling (markus-tisoft)]

Just tried it in the 12.04 daily build. It works fine there. So seems to be a Ubuntu 11.10 problem only.

Don't want to put 12.04 daily build on my production machine though. :)

[Joseph Salisbury (jsalisbury)]

Do you know if this issue exists in 10.04, which is a LTS release? We generally recommend running a LTS release for production machines. It would also be good to know if this is a regression in 11.10.

[Sarah Newman (srn-f)]

Hi,

I am using 12.04 and that particular sequence appears to be expected behavior due to duplicate address detection in ipv6 leading to the ipv6 address being tentative. I see similar behavior on an ethernet port which has never physically been connected to anything.

Before attaching anything to the bridge, it has NO-CARRIER and tentative ipv6 address, ping does not work:

$ ip addr show br0
26: br0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN
    link/ether 00:16:3e:ff:00:01 brd ff:ff:ff:ff:ff:ff
    inet6 2605:2700:0:18::1/64 scope global tentative
       valid_lft forever preferred_lft forever

After attaching something which is itself up:
$ ip addr show br0
26: br0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP
    link/ether 00:16:3e:ff:00:01 brd ff:ff:ff:ff:ff:ff
    inet6 2605:2700:0:18::1/64 scope global
       valid_lft forever preferred_lft forever
    inet6 fe80::216:3eff:feff:1/64 scope link
       valid_lft forever preferred_lft forever

ping works at this point.

Furthermore, after removing the interface, NO-CARRIER is back but the address is still no longer tentative and is still pingable:

$ ip addr show br0
26: br0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN
    link/ether 00:16:3e:ff:00:01 brd ff:ff:ff:ff:ff:ff
    inet6 2605:2700:0:18::1/64 scope global
       valid_lft forever preferred_lft forever
    inet6 fe80::216:3eff:feff:1/64 scope link
       valid_lft forever preferred_lft forever
$ ping6 -c1 2605:2700:0:18::1
PING 2605:2700:0:18::1(2605:2700:0:18::1) 56 data bytes
64 bytes from 2605:2700:0:18::1: icmp_seq=1 ttl=64 time=0.047 ms

With 3.5.0-43-generic, I set net.ipv6.conf.all.dad_transmits and net.ipv6.conf.all.accept_dad to 0 and that doesn't cause the addresses to no longer be tentative. It seems like redhat may have had a similar bug https://bugzilla.redhat.com/show_bug.cgi?id=709271 but it is not clear what they did to fix it. I have not yet tried more recent kernels to see if the problem is fixed there.