December 2020 - Bird-users

Bug: BFD confuses sessions with identical IPv6 link-local addresses
by Julian Brost 14 Jan '21

14 Jan '21

Hi, I just tried to enable BFD for some OSPF and BGP connections and ran into an issue which BFD sessions flapping up and down and Bird logging many messages like these: bfd1: Bad packet from fe80::2 - unknown session id (130079069) bfd1: Bad packet from fe80::2 - unknown session id (3102513000) bfd1: Bad packet from fe80::2 - unknown session id (3650438750) bfd1: Bad packet from fe80::2 - unknown session id (1597034259) This presumably happens due to this piece of code in bfd_rx_hook() in proto/bfd/packets.c: s = bfd_find_session_by_addr(p, sk->faddr); I use fe80::1/fe80::2 as addresses for most of my IPv6 peering connections and sk->faddr is just a 128 bit value, so it contains no interface identifier, which likely leads to Bird confusing multiple BFD sessions. Regards, Julian

3 5

[Bug] Static routes with recursive routes not going through filter when next hop changes
by Wang Shanker 14 Jan '21

14 Jan '21

Hi, all I recently spotted when filters used with recursive static routes can sometimes fail to work. The minimum case for reproducing this issue is as follows: ``` router id 192.168.0.1; table tab1; table tab2; protocol device { }; protocol static s1 { disabled; table tab1; route 2001:db8::1/128 via fe80::1%eth0; }; protocol static s2 { table tab2; igp table tab1; route 2001:db8::/32 recursive 2001:db8::1; import filter { print "Filter called"; print "dest=", dest; if dest = RTD_UNREACHABLE then { reject; } else { accept; } }; }; ``` Suppose in routing table `tab2`, we want the nexthop of 2001:db8::/32 follows 2001:db8::1 in table `tab1`, only when 2001:db8::1 is reachable. Hence, a filter rejecting unreachable routes is used. In `tab1`, a static protocol `s1` is used to emulate the routing changes there. In actual scenario, the nexthop of 2001:db8::1 may be decided by more complex routing protocols. When bird started, everything looks fine, both routing tables are empty: ``` # ./bird -c bird-test.conf -d -s bird.ctl & bird: Filter called bird: dest=(enum 34)3 bird: Started # ./birdc -s bird.ctl sh route table tab1 BIRD 1.6.8 ready. # ./birdc -s bird.ctl sh route table tab2 BIRD 1.6.8 ready. ``` However, things begin to go wrong when I trigger a change in table `tab1`: ``` # ./birdc -s bird.ctl enable s1 BIRD 1.6.8 ready. bird: Enabling protocol s1 s1: enabled # ./birdc -s bird.ctl sh route table tab1 BIRD 1.6.8 ready. 2001:db8::1/128 via fe80::1 on eth0 [s1 16:20:47] * (200) # ./birdc -s bird.ctl sh route table tab2 BIRD 1.6.8 ready. ``` It seems that the changed route does not going through the filter and failed to be injected into table `tab2`. This also happens when the route is available at start and is withdrawn later. ``` # We edit the configuration and remove the "disabled" statement in protocol s1 # ./bird -c bird-test.conf -d -s bird.ctl & bird: Filter called bird: dest=(enum 34)0 bird: Started # ./birdc -s bird.ctl sh route table tab1 BIRD 1.6.8 ready. 2001:db8::1/128 via fe80::1 on eth0 [s1 16:25:33] * (200) # ./birdc -s bird.ctl sh route table tab2 BIRD 1.6.8 ready. 2001:db8::/32 via fe80::1 on eth0 [s2 16:25:33] * (200) # ./birdc -s bird.ctl disable s1 BIRD 1.6.8 ready. bird: Disabling protocol s1 s1: disabled # ./birdc -s bird.ctl sh route table tab1 BIRD 1.6.8 ready. # ./birdc -s bird.ctl sh route table tab2 BIRD 1.6.8 ready. 2001:db8::/32 unreachable [s2 16:25:32] * (200) ``` Expected behavior: I suggest that when there are any changes to the nexthop of recursive routes, they should be re-filtered. Cheers, Miao Wang

2 1

IPv6 link-local address with scope
by Wang Shanker 14 Jan '21

14 Jan '21

Hi, all I encountered a problem when trying to configure a static IPv6 route the nexthop of which is a link-local address. ``` protocol static { route 2001:db8:1::/48 via fe80::1234%foo; } ``` The above configuration worked well for routing traffic to 2001:db8:1::/48 to the router fe80::1234 at the interface named foo. However, if the name of the interface contains special characters, for example, `foo-bar`, bird will complain about the syntax. I dig into the source code and found out that in https://gitlab.nic.cz/labs/bird/-/blob/legacy/conf/confbase.Y#L162-165, `ipa_scope` is defined like `%[SYM]`, so in above scenario, the address shall be expressed as `fe80::1234%'foo-bar'`. I believe it could be better to have a record of this in user's doc. Cheers, Miao Wang

2 1

[Bug] Mixed BFD session for peers with same IPv6 link-local addresses on different interfaces
by Wang Shanker 14 Jan '21

14 Jan '21

Hi, all In bird, BFD sessions are distinguished only by their peers' address, so if multiple routing protocols request BFD session for the same neighbor, they can share and be controlled by one session. Normally it won't be a problem. However, it is not the case when it come to the IPv6 link-local addresses. Since fe80:: addresses are link-local, the same address on different interfaces refers to different nodes. So when requesting BFD sessions for neighbors with link-local address, the destination interface should also be considered. This issue can be reproduced by the following configuration with the latest commit in legacy branch, and a similar configuration with the latest commit in master branch: ``` router id 192.168.0.1; protocol bfd bfd1{ interface "eth*"; }; protocol device { }; protocol static { route 2001:db8::/32 multipath via fe80::1%eth0 bfd via fe80::1%eth1 bfd ; }; ``` As shown in `show bfd session`, only one BFD session is created: ``` # ./birdc sh bfd sess BIRD 1.6.8 ready. bfd1: IP address Interface State Since Interval Timeout fe80::1 eth0 Down 01:33:24 1.000 0.000 ``` Expected behavior: two BFD sessions are created with both routers on both interfaces. Cheers, Miao Wang

2 1

Template sans "channels"
by Clemens Schrimpe 14 Jan '21

14 Jan '21

Ahoj (again) :-) While in the process of cleaning up "ancient" configs (basically rewriting them from scratch) I came across an odd discovery: A template definition (for a kernel protocol) demands the specification of one (and only one) channel - which, in my case, defeats the whole purpose of having a kernel template in the first place. I attempted to have a template with all the "standard behaviour" I wanted (persist, metric, table #, ...) for both IPv6 and legacy-IPv4 kernel tables and then tried to apply it like so: template kernel KERNEL { persist; learn; metric 20; kernel table 10; } protocol kernel KERNEL6 from KERNEL { ipv6; } protocol kernel KERNEL4 from KERNEL { ipv4; } ... because I thought the templating mechanism was "just" some syntactic sugar, similar to an "include" statement?! As the kernel template demands a (single) channel statement, which could then, upon invocation, not be "negated" again, i.e., by a no ipv4; statement I can only have two templates - one pre-destined for IPv6 and another for IPv4. What is the intention behind this? I do understand, that -in the end- the actual protocol defined through a template must have one (and only one in the case of a "kernel" protocol) channel definition and in my example this requirement would be fulfilled. Thoughts? Hints? Thanks again - Clemens

2 2

BGP "missing lladdr" option not implemented?
by Sébastien Parisot 14 Jan '21

14 Jan '21

Hi, I was looking at the "missing lladdr" options under BGP Channel configuration, but I can't find how it is implemented in the code. It seems to me that the only use of the missing_lladdr field in struct bgp_channel_config is in the configuration parser, bgp_postconfig to set its default value, and bgp_channel_reconfigure. I think it is like this since this particular commit : commit d15b0b0a1b494c14b139d2d28706d82cd6e2f139 Author: Ondrej Zajicek (work) <santiago(a)crfreenet.org> Date: Wed Dec 7 14:11:28 2016 +0100 BGP redesign Integrated and extensible BGP with generalized AFI handling, support for IPv4+IPv6 AFI and unicast+multicast SAFI. Was that intended or did I miss something? Thanks, Sébastien

2 1

Bird can't add static route.
by Mikhail Petrov 14 Jan '21

14 Jan '21

4 3

[PATCH] Nest: Read Babel metric as IGP metric
by James Lu 29 Dec '20

29 Dec '20

Hi all, When testing Babel with Bird 2.0.7, I noticed that unlike OSPF, Babel metrics are not read as an IGP metric for BGP purposes. I've attached a simple patch that adds this support - it seems to do the trick in my local testing. Best, James

4 8

BIRD 2.0.7 segfault in route filters
by Damian Zaremba 28 Dec '20

28 Dec '20

Hi, It appears there is a segfault in the route specific filters, this was initially observed during a `configure`, but appears to be easily reproducible with startup config. Environment: OS: Ubuntu Focal/Xenial (from latest image on DockerHub) GCC: 9.3.0 / 5.4.0 Bird: 2.0.7 (from https://bird.network.cz/download/bird-2.0.7.tar.gz) / 61dae32 from Git Build options: --enable-debug / no options Reproduction configuration file; ``` router id 192.0.2.0; protocol device {} protocol static { ipv4 {}; route 192.0.2.0/24 via "lo" { return false; }; } ``` Backtrace; ``` Reading symbols from bird... BIRD pretty printers loaded OK. BIRD pretty printers loaded OK. (gdb) r -d -c test.conf Starting program: /bird-2.0.7/bird -d -c test.conf warning: Error disabling address space randomization: Operation not permitted [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1". Filter line 0000558f4c848340 (len=2) Instruction FI_CONSTANT at line 8 value FALSE Instruction FI_RETURN at line 8 Filter line 0000558f4c848340 dump done Interpreting line. Filter line 0000558f4c848340 (len=2) Instruction FI_CONSTANT at line 8 value FALSE Instruction FI_RETURN at line 8 Filter line 0000558f4c848340 dump done Program received signal SIGSEGV, Segmentation fault. 0x0000558f4b269796 in interpret (fs=0x7efd219df730, line=0x558f4c848340, val=0x0) at filter/f-inst.c:890 890 fstk->vcnt = fstk->estk[fstk->ecnt].ventry - 1; (gdb) bt #0 0x0000558f4b269796 in interpret (fs=0x7efd219df730, line=0x558f4c848340, val=0x0) at filter/f-inst.c:890 #1 0x0000558f4b26da99 in f_eval_rte (expr=0x558f4c848340, rte=0x7ffc421efc40, tmp_pool=0x558f4c856060) at filter/filter.c:352 #2 0x0000558f4b2f9c5e in static_announce_rte (p=0x558f4c84e7b0, r=0x558f4c8481e0) at proto/static/static.c:106 #3 0x0000558f4b2fa13a in static_add_rte (p=0x558f4c84e7b0, r=0x558f4c8481e0) at proto/static/static.c:228 #4 0x0000558f4b2fa9d9 in static_start (P=0x558f4c84e7b0) at proto/static/static.c:432 #5 0x0000558f4b29501f in proto_rethink_goal (p=0x558f4c84e7b0) at nest/proto.c:1173 #6 0x0000558f4b294e7f in protos_commit (new=0x558f4c842400, old=0x0, force_reconfig=0, type=1) at nest/proto.c:1132 #7 0x0000558f4b261018 in config_do_commit (c=0x558f4c842400, type=1) at conf/conf.c:271 #8 0x0000558f4b2611ef in config_commit (c=0x558f4c842400, type=1, timeout=0) at conf/conf.c:361 #9 0x0000558f4b30b793 in main (argc=4, argv=0x7ffc421eff58) at sysdep/unix/main.c:908 ``` It would be useful to be able to reject routes from their filter (to enable suppression in function logic, either via return or reject), however in the worst case the syntax should result in an invalid configuration error rather than a crash. Let me know if you need any more info. Cheers - Damian

2 5

Incorrect iBGP next hop in bird2
by Юрий Иванов 22 Dec '20

22 Dec '20

Hi, I've very simple install. Server with two links vlan1 - 50.0.0.1/25 and vlan2 - 10.0.0.7/31 One static default route and one iBGP neighbor: protocol static default_route { ipv4; route 0.0.0.0/0 via 50.0.0.254; } protocol bgp nat_downlink1 { local 10.0.0.7 as 65001; neighbor 10.0.0.6 as 65001; #direct; ipv4 { next hop self; import all; export all; }; } Neighbor sends me a bunch of /32 host routes and all routes received from 10.0.0.6 neigbor has next hop 50.0.0.254. bird> sh route Table master4: 0.0.0.0/0 unicast [default_route 15:23:28.170] * (200) via 50.0.0.254 on vlan1 10.107.1.81/32 unicast [downlink1 16:59:19.320 from 10.0.0.6] * (100/?) [i] via 50.0.0.254 on vlan2 To fix this I can add direct option to me BGP neigbor, but I wonder why this even happens. Is there more correct eay to fix this?

2 1