- Bird-users - Mailing Lists

Error sending route to kernel on FreeBSD (ARM)
by Eugene Sevastyanov 01 Sep '17

01 Sep '17

Hi! I tried to run a bird (v1.6.3) under FreeBSD on ARM architecture. An error occurred: <ERR> KRT: Error sending route XX.XX.XX.XX/XX to kernel: Invalid argument My config: log "/var/log/bird.log" all; router id xx.xx.xx.xx; #debug protocols all; protocol device { scan time 10; } protocol kernel { import none; export all; scan time 15; } protocol bgp ULNIX { local as 65507; neighbor xx.xx.xx.xx as xxx; source address xx.xx.xx.xx; import all; import limit 5000 action restart; } Studying the source code, found out that the problem with aligning the structure (file: /sysdep/bsd/krt-sock.c): struct ks_msg { struct rt_msghdr rtm; struct sockaddr_storage buf[RTAX_MAX]; }; After making changes (see patch in attached), I managed to run the bird. -- Regards Eugene Sevastyanov

2 2

Fix IPv6 ECMP handling with 4.11+ kernel
by Vincent Bernat 01 Sep '17

01 Sep '17

Hey! Starting from kernel 4.11 (commit beb1afac518d), IPv6 are now notified using RTA_MULTIPATH, like IPv4 routes. Those routes are not handled correctly by BIRD. We handle them correctly. This also enable to parse alien routes correctly. Route modifications is still done in the old way as for insertion/deletion, this is not helpful to optimize in those cases and for replace, IPv4 case is not optimized either. It should be possible to detect appropriate support for RTA_MULTIPATH when receiving an IPv6 route with this attribute, but I don't see how it would be helpful, so I didn't do it (simpler code this way). I did some quick tests and routes are removed/added correctly: Deleted 2001:db8:a3::/64 proto bird metric 1024 nexthop via 2001:db8:ff::5 dev vti5 weight 1 nexthop via 2001:db8:ff::7 dev vti6 weight 1 2001:db8:a3::/64 via 2001:db8:ff::7 dev vti6 proto bird metric 1024 pref medium Deleted 2001:db8:a3::/64 via 2001:db8:ff::7 dev vti6 proto bird metric 1024 pref medium 2001:db8:a3::/64 via 2001:db8:ff::5 dev vti5 proto bird metric 1024 pref medium 2001:db8:a3::/64 proto bird metric 1024 nexthop via 2001:db8:ff::7 dev vti6 weight 1 nexthop via 2001:db8:ff::5 dev vti5 weight 1 Also, alien routes are correctly parsed when next-hops are correctly ordered (I didn't check if this restriction is also present for IPv4 or if Linux is always sending IPv4 multipath routes with next-hops correctly ordered): 2001:db8:a4::/64 metric 1024 nexthop via 2001:db8:ff::5 dev vti5 weight 1 nexthop via 2001:db8:ff::7 dev vti6 weight 1 2001:db8:a4::/64 multipath [kernel1 22:30:54] * (10) via 2001:db8:ff::5 on vti5 weight 1 via 2001:db8:ff::7 on vti6 weight 1 I'll try to investigate that unless someone already knows the answer. -- Make input easy to proofread. - The Elements of Programming Style (Kernighan & Plauger)

3 8

Version 1.6.2
by Ondrej Filip 31 Aug '17

31 Aug '17

Dear users, a new version of the 1.6.x branch is out: Version 1.6.2 (2016-09-29) o Fixes serious bug introduced in the previous version Upgrade from 1.6.1 is highly recommended. Enjoy it! Ondrej

4 6

how to change ospf interface cost dynamically
by d tbsky 30 Aug '17

30 Aug '17

hi: I have two ospf interfaces between two routers. sometimes I want to use one of the interfaces to transfer data, and sometimes I want to use another. I think I can change configuration of interface cost and issue bird reload every time I need to switch interface. but I wonder if there is smarter way. any possibility I can change some interface parameters by linux system commands, so bird would change ospf cost dynamically? or can I tell bird to change interface cost dynamically? thanks a lot for suggestion!

1 0

[RFC] filter profiling
by Lennert Buytenhek 29 Aug '17

29 Aug '17

Hi! Attached is a crude hack to allow profiling filter processing. When you're done implementing all of your filter logic and are wondering exactly why it now takes bird so friggin' long to propagate routes while sitting for several minutes at 100% CPU on the anemic embedded CPUs you're trying to run it on, tools like 'perf' aren't too useful, since they'll just tell you that all CPU time is being spent in interpret() or related functions and won't tell you which filter language statements are responsible for how much of the CPU time. The attached patch is a very rough way of attempting to enable you to figure out what's eating your cycles. The basic idea is that it sets up a SIGUSR2 handler that will write to stderr the config file name and line currently being processed, and the prefix being operated on if there is one. You can then send bird SIGUSR2 in a tight loop whenever it seems to be spinning at 100% CPU and the resulting output will give you a clue of what's going on. This is a very dirty hack, and it does some things that probably aren't entirely safe, like calling into snprintf from a signal handler, and it shouldn't be merged in this form, but for our purposes, it seems to be working well, and seems to be able to provide useful insights into some of our self-inflicted filter performance problems. Feedback appreciated! (Better ideas also appreciated. :D) Cheers, Lennert diff --git a/filter/f-util.c b/filter/f-util.c index def2b248..452dbcb2 100644 --- a/filter/f-util.c +++ b/filter/f-util.c @@ -20,6 +20,7 @@ f_new_inst(void) ret->code = ret->aux = 0; ret->arg1 = ret->arg2 = ret->next = NULL; ret->lineno = ifs->lino; + ret->file_name = ifs->file_name; return ret; } diff --git a/filter/filter.c b/filter/filter.c index f18970e0..d58c6cb4 100644 --- a/filter/filter.c +++ b/filter/filter.c @@ -33,6 +33,10 @@ #undef LOCAL_DEBUG +#include <stdio.h> +#include <signal.h> +#include <unistd.h> + #include "nest/bird.h" #include "lib/lists.h" #include "lib/resource.h" @@ -551,7 +555,7 @@ val_format(struct f_val v, buffer *buf) } } -static struct rte **f_rte; +static struct rte ** volatile f_rte; static struct rta *f_old_rta; static struct ea_list **f_tmp_attrs; static struct linpool *f_pool; @@ -630,8 +634,10 @@ static struct tbf rl_runtime_err = TBF_DEFAULT_LOG_LIMITS; * &f_val structures are copied around, so there are no problems with * memory managment. */ +static struct f_val interpret(struct f_inst *what); + static struct f_val -interpret(struct f_inst *what) +__interpret(struct f_inst *what) { struct symbol *sym; struct f_val v1, v2, res, *vp; @@ -1515,6 +1521,70 @@ interpret(struct f_inst *what) return res; } +#define barrier() __asm__ __volatile__("": : :"memory") + +static char *file_name; +static int lino; + +static struct f_val +interpret(struct f_inst *what) +{ + struct f_val ret; + + if (what != NULL) { + lino = what->lineno; + barrier(); + file_name = what->file_name; + } + + ret = __interpret(what); + + if (what != NULL) { + file_name = NULL; + barrier(); + lino = -1; + } + + return ret; +} + +static void handle_sigusr2(int sig UNUSED) +{ + char prefix[64]; + char buf[1024]; + + if (file_name == NULL) { +// write(2, "not interpreting an f_inst right now\n", 37); + return; + } + + prefix[0] = 0; +#ifndef IPV6 + if (f_rte != NULL) { + struct fib_node *n = &(*f_rte)->net->n; + ip_addr addr = n->prefix; + + snprintf(prefix, sizeof(prefix), "%d.%d.%d.%d/%d", + (addr >> 24) & 0xff, (addr >> 16) & 0xff, + (addr >> 8) & 0xff, addr & 0xff, n->pxlen); + } +#endif + + snprintf(buf, sizeof(buf), "%s:%d %s\n", file_name, lino, prefix); + + write(2, buf, strlen(buf)); +} + +void filter_profile_init(void) +{ + struct sigaction sa; + + bzero(&sa, sizeof(sa)); + sa.sa_handler = handle_sigusr2; + sa.sa_flags = SA_RESTART; + sigaction(SIGUSR2, &sa, NULL); +} + #undef ARG #define ARG(x,y) \ if (!i_same(f1->y, f2->y)) \ @@ -1720,6 +1790,7 @@ f_run(struct filter *filter, struct rte **rte, struct ea_list **tmp_attrs, struc rta_free(f_old_rta); } + f_rte = NULL; if (res.type != T_RETURN) { log_rl(&rl_runtime_err, L_ERR "Filter %s did not return accept nor reject. Make up your mind", filter->name); @@ -1747,6 +1818,8 @@ f_eval_rte(struct f_inst *expr, struct rte **rte, struct linpool *tmp_pool) /* Note that in this function we assume that rte->attrs is private / uncached */ struct f_val res = interpret(expr); + f_rte = NULL; + /* Hack to include EAF_TEMP attributes to the main list */ (*rte)->attrs->eattrs = ea_append(tmp_attrs, (*rte)->attrs->eattrs); diff --git a/filter/filter.h b/filter/filter.h index 049ceb76..4349fdb3 100644 --- a/filter/filter.h +++ b/filter/filter.h @@ -26,6 +26,7 @@ struct f_inst { /* Instruction */ int i; void *p; } a2; + char *file_name; int lineno; }; diff --git a/sysdep/unix/main.c b/sysdep/unix/main.c index 8aa19fce..4ef84196 100644 --- a/sysdep/unix/main.c +++ b/sysdep/unix/main.c @@ -870,6 +870,7 @@ main(int argc, char **argv) write_pid_file(); signal_init(); + filter_profile_init(); config_commit(conf, RECONFIG_HARD, 0);

2 5

{ } in BGP.as_path
by Mikhail Mayorov 25 Aug '17

25 Aug '17

Hi all! What is mean '{}' in as_path? PAPA.rndpapa bird # birdcl show route where net = '216.126.32.0/20' all BIRD 1.6.3 ready. 216.126.32.0/20 via 95.174.97.33 on vlan8 [bird_ipv4 25.08.2017 17:54:37] * (100/0) [i] Type: BGP unicast univ BGP.origin: IGP BGP.as_path: 9002 3356 13703 { 20340 } BGP.next_hop: 95.174.97.33 BGP.med: 999 BGP.local_pref: 100 BGP.aggregator: 208.52.178.233 AS13703 BGP.community: (9002,9002) (9002,64657) (49037,0) (49037,10001) -- Regards, Mikhail V. Majorov

2 1

IPv6 BIRD filters
by Marek Królikowski 25 Aug '17

25 Aug '17

Hello Guys, Anyone create maybe a good filters for BIRD IPv6 upstreams? I just add this to my import filter: import filter { if bgp_path.len > 49 then reject; if (net.len < 16) || (net.len > 48) then reject; accept; }; But i saw i get strange routing: root@it-bgp:~# birdc6 show route 2404:1710::/28 BIRD 1.6.3 ready. 2404:1710::/28 via 2a02:XXXX:XXXX:1 on eth0 [Italy 13:52:47 from 2a02:XXXX:XXXX::2] * (100/?) [AS65332i] Anyone got any good filter to block something like this? Best Regards Marek

2 1

IPv6 prefix set
by Damien Clabaut 22 Aug '17

22 Aug '17

Hello, I am trying to use prefix-set inside a filter in IPv6: filter my_filter6 prefix set my_prefixset6; { my_prefixset6 = [aaaa:bbbb:100:ff::/56]; if (net ~ my_prefixset6) then { bgp_local_pref = 104; } accept; } But I get the following error: birdc6 BIRD 1.6.3 ready. bird> configure check Invalid network prefix: aaaa:bbbb:100:ff::/56. Am I doing anything wrong ? Also, the prefix set is dynamically updated via script, is there a correct way to handle the case where it is empty ? Or should the scripts comment the definition of prefix-set and the complete "if" block ? Regards, -- Damien Clabaut R&D vRouter ovh.qc.ca

3 3

Propagate default route when BGP session is up
by Robert Sander 22 Aug '17

22 Aug '17

Hi, how do I propagate a default route into OSPF when a specific BGP session is up and not propagate it when it is down? Currently I have these config snippets: protocol static static_default { preference 10; route 0.0.0.0/0 unreachable; } function export_ospf() { if source = RTS_BGP then { return false; } if proto = "static_bgp" then { return false; } return true; } protocol ospf { rfc1583compat on; import all; export filter { if export_ospf() then accept; reject; }; area 0 { interface "ospf" { cost 10; hello 10; retransmit 5; dead 45; }; }; } protocol bgp ispa1 { description "ISP A1"; import all; export filter { if proto = "static_bgp" then accept; reject; }; local 1.2.3.6 as 1; neighbor 1.2.3.5 as 2; password "xyz"; } I can manually "birdc disable static_default" to not propagate the default route but how can this be automated when the BGP session to ispa1 is down? Regards -- Robert Sander Heinlein Support GmbH Schwedter Str. 8/9b, 10119 Berlin http://www.heinlein-support.de Tel: 030 / 405051-43 Fax: 030 / 405051-19 Zwangsangaben lt. §35a GmbHG: HRB 93818 B / Amtsgericht Berlin-Charlottenburg, Geschäftsführer: Peer Heinlein -- Sitz: Berlin

2 1

bgp_path.first
by nixx 22 Aug '17

22 Aug '17

Hello. i'm trying to configure bird for working with uplinks and downlinks with this example: https://gitlab.labs.nic.cz/labs/bird/wikis/BGP_filtering i'm using truncated function: function rt_export() { if proto = "static_bgp" then return true; if source != RTS_BGP then return false; return bgp_path.first ~ [ downlink1AS, downlink2AS ]; } But this incorrectly works - bird trying to export my full-view to uplink. Then, when i changed to if ( bgp_path.len = 2 ) && ( bgp_path.first = myAS ) then return bgp_path ~ [ downlink1AS, downlink2AS ]; ...it works. What i'm (or my bird) doing wrong? Or maybe this example is not correct?

3 3