]>
Commit | Line | Data |
---|---|---|
718e3744 | 1 | /* Kernel routing table updates using netlink over GNU/Linux system. |
2 | * Copyright (C) 1997, 98, 99 Kunihiro Ishiguro | |
3 | * | |
4 | * This file is part of GNU Zebra. | |
5 | * | |
6 | * GNU Zebra is free software; you can redistribute it and/or modify it | |
7 | * under the terms of the GNU General Public License as published by the | |
8 | * Free Software Foundation; either version 2, or (at your option) any | |
9 | * later version. | |
10 | * | |
11 | * GNU Zebra is distributed in the hope that it will be useful, but | |
12 | * WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
14 | * General Public License for more details. | |
15 | * | |
896014f4 DL |
16 | * You should have received a copy of the GNU General Public License along |
17 | * with this program; see the file COPYING; if not, write to the Free Software | |
18 | * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA | |
718e3744 | 19 | */ |
20 | ||
21 | #include <zebra.h> | |
ddfeb486 DL |
22 | |
23 | #ifdef HAVE_NETLINK | |
24 | ||
8ccc7e80 | 25 | #include <net/if_arp.h> |
ba777396 RW |
26 | #include <linux/lwtunnel.h> |
27 | #include <linux/mpls_iptunnel.h> | |
28 | #include <linux/neighbour.h> | |
29 | #include <linux/rtnetlink.h> | |
718e3744 | 30 | |
31 | /* Hack for GNU libc version 2. */ | |
32 | #ifndef MSG_TRUNC | |
33 | #define MSG_TRUNC 0x20 | |
34 | #endif /* MSG_TRUNC */ | |
35 | ||
36 | #include "linklist.h" | |
37 | #include "if.h" | |
38 | #include "log.h" | |
39 | #include "prefix.h" | |
40 | #include "connected.h" | |
41 | #include "table.h" | |
26e2ae36 | 42 | #include "memory.h" |
4a1ab8e4 | 43 | #include "zebra_memory.h" |
718e3744 | 44 | #include "rib.h" |
e04ab74d | 45 | #include "thread.h" |
edd7c245 | 46 | #include "privs.h" |
fb018d25 | 47 | #include "nexthop.h" |
78104b9b | 48 | #include "vrf.h" |
5e6a74d8 | 49 | #include "vty.h" |
40c7bdb0 | 50 | #include "mpls.h" |
13d60d35 | 51 | #include "vxlan.h" |
718e3744 | 52 | |
53 | #include "zebra/zserv.h" | |
fe18ee2d | 54 | #include "zebra/zebra_ns.h" |
7c551956 | 55 | #include "zebra/zebra_vrf.h" |
6621ca86 | 56 | #include "zebra/rt.h" |
718e3744 | 57 | #include "zebra/redistribute.h" |
58 | #include "zebra/interface.h" | |
59 | #include "zebra/debug.h" | |
12f6fb97 | 60 | #include "zebra/rtadv.h" |
567b877d | 61 | #include "zebra/zebra_ptm.h" |
40c7bdb0 | 62 | #include "zebra/zebra_mpls.h" |
1fdc9eae | 63 | #include "zebra/kernel_netlink.h" |
64 | #include "zebra/rt_netlink.h" | |
e3be0432 | 65 | #include "zebra/zebra_mroute.h" |
2232a77c | 66 | #include "zebra/zebra_vxlan.h" |
e3be0432 | 67 | |
40c7bdb0 | 68 | #ifndef AF_MPLS |
69 | #define AF_MPLS 28 | |
70 | #endif | |
71 | ||
2232a77c | 72 | static vlanid_t filter_vlan = 0; |
73 | ||
d62a17ae | 74 | struct gw_family_t { |
75 | u_int16_t filler; | |
76 | u_int16_t family; | |
77 | union g_addr gate; | |
40c7bdb0 | 78 | }; |
79 | ||
8755598a DS |
80 | char ipv4_ll_buf[16] = "169.254.0.1"; |
81 | struct in_addr ipv4_ll; | |
82 | ||
83 | /* | |
84 | * The ipv4_ll data structure is used for all 5549 | |
85 | * additions to the kernel. Let's figure out the | |
86 | * correct value one time instead for every | |
87 | * install/remove of a 5549 type route | |
88 | */ | |
d62a17ae | 89 | void rt_netlink_init(void) |
8755598a | 90 | { |
d62a17ae | 91 | inet_pton(AF_INET, ipv4_ll_buf, &ipv4_ll); |
8755598a DS |
92 | } |
93 | ||
23b1f334 DD |
94 | static inline int is_selfroute(int proto) |
95 | { | |
d62a17ae | 96 | if ((proto == RTPROT_BGP) || (proto == RTPROT_OSPF) |
97 | || (proto == RTPROT_STATIC) || (proto == RTPROT_ZEBRA) | |
98 | || (proto == RTPROT_ISIS) || (proto == RTPROT_RIPNG) | |
99 | || (proto == RTPROT_NHRP) || (proto == RTPROT_EIGRP) | |
915902cb | 100 | || (proto == RTPROT_LDP) || (proto == RTPROT_BABEL) |
8a71d93d | 101 | || (proto == RTPROT_RIP) || (proto == RTPROT_SHARP)) { |
d62a17ae | 102 | return 1; |
103 | } | |
104 | ||
105 | return 0; | |
23b1f334 DD |
106 | } |
107 | ||
915902cb | 108 | static inline int zebra2proto(int proto) |
23b1f334 | 109 | { |
d62a17ae | 110 | switch (proto) { |
111 | case ZEBRA_ROUTE_BABEL: | |
112 | proto = RTPROT_BABEL; | |
113 | break; | |
114 | case ZEBRA_ROUTE_BGP: | |
115 | proto = RTPROT_BGP; | |
116 | break; | |
117 | case ZEBRA_ROUTE_OSPF: | |
118 | case ZEBRA_ROUTE_OSPF6: | |
119 | proto = RTPROT_OSPF; | |
120 | break; | |
121 | case ZEBRA_ROUTE_STATIC: | |
122 | proto = RTPROT_STATIC; | |
123 | break; | |
124 | case ZEBRA_ROUTE_ISIS: | |
125 | proto = RTPROT_ISIS; | |
126 | break; | |
127 | case ZEBRA_ROUTE_RIP: | |
128 | proto = RTPROT_RIP; | |
129 | break; | |
130 | case ZEBRA_ROUTE_RIPNG: | |
131 | proto = RTPROT_RIPNG; | |
132 | break; | |
133 | case ZEBRA_ROUTE_NHRP: | |
134 | proto = RTPROT_NHRP; | |
135 | break; | |
136 | case ZEBRA_ROUTE_EIGRP: | |
137 | proto = RTPROT_EIGRP; | |
138 | break; | |
139 | case ZEBRA_ROUTE_LDP: | |
140 | proto = RTPROT_LDP; | |
141 | break; | |
8a71d93d DS |
142 | case ZEBRA_ROUTE_SHARP: |
143 | proto = RTPROT_SHARP; | |
144 | break; | |
d62a17ae | 145 | default: |
146 | proto = RTPROT_ZEBRA; | |
147 | break; | |
148 | } | |
149 | ||
150 | return proto; | |
23b1f334 DD |
151 | } |
152 | ||
915902cb DS |
153 | static inline int proto2zebra(int proto, int family) |
154 | { | |
155 | switch (proto) { | |
156 | case RTPROT_BABEL: | |
157 | proto = ZEBRA_ROUTE_BABEL; | |
158 | break; | |
159 | case RTPROT_BGP: | |
160 | proto = ZEBRA_ROUTE_BGP; | |
161 | break; | |
162 | case RTPROT_OSPF: | |
163 | proto = (family == AFI_IP) ? | |
164 | ZEBRA_ROUTE_OSPF : ZEBRA_ROUTE_OSPF6; | |
165 | break; | |
166 | case RTPROT_ISIS: | |
167 | proto = ZEBRA_ROUTE_ISIS; | |
168 | break; | |
169 | case RTPROT_RIP: | |
170 | proto = ZEBRA_ROUTE_RIP; | |
171 | break; | |
172 | case RTPROT_RIPNG: | |
173 | proto = ZEBRA_ROUTE_RIPNG; | |
174 | break; | |
175 | case RTPROT_NHRP: | |
176 | proto = ZEBRA_ROUTE_NHRP; | |
177 | break; | |
178 | case RTPROT_EIGRP: | |
179 | proto = ZEBRA_ROUTE_EIGRP; | |
180 | break; | |
181 | case RTPROT_LDP: | |
182 | proto = ZEBRA_ROUTE_LDP; | |
183 | break; | |
184 | case RTPROT_STATIC: | |
185 | proto = ZEBRA_ROUTE_STATIC; | |
186 | break; | |
187 | default: | |
188 | proto = ZEBRA_ROUTE_KERNEL; | |
189 | break; | |
190 | } | |
191 | return proto; | |
192 | } | |
193 | ||
12f6fb97 DS |
194 | /* |
195 | Pending: create an efficient table_id (in a tree/hash) based lookup) | |
196 | */ | |
d62a17ae | 197 | static vrf_id_t vrf_lookup_by_table(u_int32_t table_id) |
12f6fb97 | 198 | { |
d62a17ae | 199 | struct vrf *vrf; |
200 | struct zebra_vrf *zvrf; | |
12f6fb97 | 201 | |
a2addae8 | 202 | RB_FOREACH (vrf, vrf_id_head, &vrfs_by_id) { |
d62a17ae | 203 | if ((zvrf = vrf->info) == NULL || (zvrf->table_id != table_id)) |
204 | continue; | |
12f6fb97 | 205 | |
d62a17ae | 206 | return zvrf_id(zvrf); |
207 | } | |
12f6fb97 | 208 | |
d62a17ae | 209 | return VRF_DEFAULT; |
12f6fb97 DS |
210 | } |
211 | ||
718e3744 | 212 | /* Looking up routing table by netlink interface. */ |
d62a17ae | 213 | static int netlink_route_change_read_unicast(struct sockaddr_nl *snl, |
214 | struct nlmsghdr *h, ns_id_t ns_id, | |
215 | int startup) | |
718e3744 | 216 | { |
d62a17ae | 217 | int len; |
218 | struct rtmsg *rtm; | |
219 | struct rtattr *tb[RTA_MAX + 1]; | |
220 | u_char flags = 0; | |
221 | struct prefix p; | |
792fa92e | 222 | struct prefix_ipv6 src_p = {}; |
d62a17ae | 223 | vrf_id_t vrf_id = VRF_DEFAULT; |
224 | ||
225 | char anyaddr[16] = {0}; | |
226 | ||
915902cb | 227 | int proto = ZEBRA_ROUTE_KERNEL; |
d62a17ae | 228 | int index = 0; |
229 | int table; | |
230 | int metric = 0; | |
231 | u_int32_t mtu = 0; | |
25715c7e | 232 | uint8_t distance = 0; |
4e40b6d6 | 233 | route_tag_t tag = 0; |
d62a17ae | 234 | |
235 | void *dest = NULL; | |
236 | void *gate = NULL; | |
237 | void *prefsrc = NULL; /* IPv4 preferred source host address */ | |
238 | void *src = NULL; /* IPv6 srcdest source prefix */ | |
e655a03c | 239 | enum blackhole_type bh_type = BLACKHOLE_UNSPEC; |
d62a17ae | 240 | |
241 | rtm = NLMSG_DATA(h); | |
242 | ||
243 | if (startup && h->nlmsg_type != RTM_NEWROUTE) | |
244 | return 0; | |
e655a03c DL |
245 | switch (rtm->rtm_type) { |
246 | case RTN_UNICAST: | |
247 | break; | |
248 | case RTN_BLACKHOLE: | |
249 | bh_type = BLACKHOLE_NULL; | |
250 | break; | |
251 | case RTN_UNREACHABLE: | |
252 | bh_type = BLACKHOLE_REJECT; | |
253 | break; | |
254 | case RTN_PROHIBIT: | |
255 | bh_type = BLACKHOLE_ADMINPROHIB; | |
256 | break; | |
257 | default: | |
d62a17ae | 258 | return 0; |
e655a03c | 259 | } |
d62a17ae | 260 | |
261 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct rtmsg)); | |
262 | if (len < 0) | |
263 | return -1; | |
264 | ||
265 | memset(tb, 0, sizeof tb); | |
266 | netlink_parse_rtattr(tb, RTA_MAX, RTM_RTA(rtm), len); | |
267 | ||
268 | if (rtm->rtm_flags & RTM_F_CLONED) | |
269 | return 0; | |
270 | if (rtm->rtm_protocol == RTPROT_REDIRECT) | |
271 | return 0; | |
272 | if (rtm->rtm_protocol == RTPROT_KERNEL) | |
273 | return 0; | |
274 | ||
275 | if (!startup && is_selfroute(rtm->rtm_protocol) | |
276 | && h->nlmsg_type == RTM_NEWROUTE) | |
277 | return 0; | |
278 | ||
279 | /* We don't care about change notifications for the MPLS table. */ | |
280 | /* TODO: Revisit this. */ | |
281 | if (rtm->rtm_family == AF_MPLS) | |
282 | return 0; | |
283 | ||
284 | /* Table corresponding to route. */ | |
285 | if (tb[RTA_TABLE]) | |
286 | table = *(int *)RTA_DATA(tb[RTA_TABLE]); | |
287 | else | |
288 | table = rtm->rtm_table; | |
289 | ||
290 | /* Map to VRF */ | |
291 | vrf_id = vrf_lookup_by_table(table); | |
292 | if (vrf_id == VRF_DEFAULT) { | |
293 | if (!is_zebra_valid_kernel_table(table) | |
294 | && !is_zebra_main_routing_table(table)) | |
295 | return 0; | |
296 | } | |
297 | ||
298 | /* Route which inserted by Zebra. */ | |
915902cb | 299 | if (is_selfroute(rtm->rtm_protocol)) { |
d62a17ae | 300 | flags |= ZEBRA_FLAG_SELFROUTE; |
915902cb DS |
301 | proto = proto2zebra(rtm->rtm_protocol, rtm->rtm_family); |
302 | } | |
d62a17ae | 303 | if (tb[RTA_OIF]) |
304 | index = *(int *)RTA_DATA(tb[RTA_OIF]); | |
305 | ||
306 | if (tb[RTA_DST]) | |
307 | dest = RTA_DATA(tb[RTA_DST]); | |
308 | else | |
309 | dest = anyaddr; | |
310 | ||
311 | if (tb[RTA_SRC]) | |
312 | src = RTA_DATA(tb[RTA_SRC]); | |
313 | else | |
314 | src = anyaddr; | |
315 | ||
316 | if (tb[RTA_PREFSRC]) | |
317 | prefsrc = RTA_DATA(tb[RTA_PREFSRC]); | |
318 | ||
319 | if (tb[RTA_GATEWAY]) | |
320 | gate = RTA_DATA(tb[RTA_GATEWAY]); | |
321 | ||
f19435a8 DS |
322 | if (tb[RTA_PRIORITY]) |
323 | metric = *(int *)RTA_DATA(tb[RTA_PRIORITY]); | |
d62a17ae | 324 | |
4e40b6d6 KK |
325 | #if defined(SUPPORT_REALMS) |
326 | if (tb[RTA_FLOW]) | |
327 | tag = *(uint32_t *)RTA_DATA(tb[RTA_FLOW]); | |
328 | #endif | |
329 | ||
f19435a8 DS |
330 | if (tb[RTA_METRICS]) { |
331 | struct rtattr *mxrta[RTAX_MAX + 1]; | |
d62a17ae | 332 | |
f19435a8 DS |
333 | memset(mxrta, 0, sizeof mxrta); |
334 | netlink_parse_rtattr(mxrta, RTAX_MAX, | |
335 | RTA_DATA(tb[RTA_METRICS]), | |
336 | RTA_PAYLOAD(tb[RTA_METRICS])); | |
d62a17ae | 337 | |
f19435a8 DS |
338 | if (mxrta[RTAX_MTU]) |
339 | mtu = *(u_int32_t *)RTA_DATA(mxrta[RTAX_MTU]); | |
d62a17ae | 340 | } |
341 | ||
342 | if (rtm->rtm_family == AF_INET) { | |
343 | p.family = AF_INET; | |
344 | memcpy(&p.u.prefix4, dest, 4); | |
345 | p.prefixlen = rtm->rtm_dst_len; | |
346 | ||
347 | src_p.prefixlen = | |
348 | 0; // Forces debug below to not display anything | |
349 | } else if (rtm->rtm_family == AF_INET6) { | |
350 | p.family = AF_INET6; | |
351 | memcpy(&p.u.prefix6, dest, 16); | |
352 | p.prefixlen = rtm->rtm_dst_len; | |
353 | ||
354 | src_p.family = AF_INET6; | |
355 | memcpy(&src_p.prefix, src, 16); | |
356 | src_p.prefixlen = rtm->rtm_src_len; | |
357 | } | |
358 | ||
359 | if (rtm->rtm_src_len != 0) { | |
360 | char buf[PREFIX_STRLEN]; | |
361 | zlog_warn( | |
362 | "unsupported IPv[4|6] sourcedest route (dest %s vrf %u)", | |
363 | prefix2str(&p, buf, sizeof(buf)), vrf_id); | |
364 | return 0; | |
365 | } | |
366 | ||
25715c7e DS |
367 | /* |
368 | * For ZEBRA_ROUTE_KERNEL types: | |
369 | * | |
370 | * The metric/priority of the route received from the kernel | |
371 | * is a 32 bit number. We are going to interpret the high | |
372 | * order byte as the Admin Distance and the low order 3 bytes | |
373 | * as the metric. | |
374 | * | |
375 | * This will allow us to do two things: | |
376 | * 1) Allow the creation of kernel routes that can be | |
377 | * overridden by zebra. | |
378 | * 2) Allow the old behavior for 'most' kernel route types | |
379 | * if a user enters 'ip route ...' v4 routes get a metric | |
380 | * of 0 and v6 routes get a metric of 1024. Both of these | |
381 | * values will end up with a admin distance of 0, which | |
382 | * will cause them to win for the purposes of zebra. | |
383 | */ | |
384 | if (proto == ZEBRA_ROUTE_KERNEL) { | |
385 | distance = (metric >> 24) & 0xFF; | |
386 | metric = (metric & 0x00FFFFFF); | |
387 | } | |
388 | ||
d62a17ae | 389 | if (IS_ZEBRA_DEBUG_KERNEL) { |
390 | char buf[PREFIX_STRLEN]; | |
391 | char buf2[PREFIX_STRLEN]; | |
392 | zlog_debug( | |
25715c7e | 393 | "%s %s%s%s vrf %u metric: %d Admin Distance: %d", nl_msg_type_to_str(h->nlmsg_type), |
d62a17ae | 394 | prefix2str(&p, buf, sizeof(buf)), |
395 | src_p.prefixlen ? " from " : "", | |
396 | src_p.prefixlen ? prefix2str(&src_p, buf2, sizeof(buf2)) | |
397 | : "", | |
25715c7e | 398 | vrf_id, metric, distance); |
d62a17ae | 399 | } |
400 | ||
401 | afi_t afi = AFI_IP; | |
402 | if (rtm->rtm_family == AF_INET6) | |
403 | afi = AFI_IP6; | |
404 | ||
405 | if (h->nlmsg_type == RTM_NEWROUTE) { | |
8795f904 DS |
406 | struct interface *ifp; |
407 | vrf_id_t nh_vrf_id = vrf_id; | |
408 | ||
fd36be7e DL |
409 | if (!tb[RTA_MULTIPATH]) { |
410 | struct nexthop nh; | |
411 | size_t sz = (afi == AFI_IP) ? 4 : 16; | |
412 | ||
413 | memset(&nh, 0, sizeof(nh)); | |
af760ec1 DS |
414 | |
415 | if (bh_type == BLACKHOLE_UNSPEC) { | |
416 | if (index && !gate) | |
417 | nh.type = NEXTHOP_TYPE_IFINDEX; | |
418 | else if (index && gate) | |
1dca2eaa | 419 | nh.type = (afi == AFI_IP) |
af760ec1 DS |
420 | ? NEXTHOP_TYPE_IPV4_IFINDEX |
421 | : NEXTHOP_TYPE_IPV6_IFINDEX; | |
422 | else if (!index && gate) | |
1dca2eaa RW |
423 | nh.type = (afi == AFI_IP) |
424 | ? NEXTHOP_TYPE_IPV4 | |
425 | : NEXTHOP_TYPE_IPV6; | |
af760ec1 DS |
426 | else { |
427 | nh.type = NEXTHOP_TYPE_BLACKHOLE; | |
428 | nh.bh_type = bh_type; | |
429 | } | |
430 | } else { | |
fd36be7e | 431 | nh.type = NEXTHOP_TYPE_BLACKHOLE; |
e655a03c DL |
432 | nh.bh_type = bh_type; |
433 | } | |
fd36be7e DL |
434 | nh.ifindex = index; |
435 | if (prefsrc) | |
436 | memcpy(&nh.src, prefsrc, sz); | |
437 | if (gate) | |
438 | memcpy(&nh.gate, gate, sz); | |
915902cb | 439 | |
8795f904 DS |
440 | if (index) { |
441 | ifp = if_lookup_by_index(index, | |
442 | VRF_UNKNOWN); | |
443 | if (ifp) | |
444 | nh_vrf_id = ifp->vrf_id; | |
445 | } | |
4a7371e9 | 446 | nh.vrf_id = nh_vrf_id; |
8795f904 | 447 | |
4a7371e9 DS |
448 | rib_add(afi, SAFI_UNICAST, vrf_id, proto, 0, flags, &p, |
449 | NULL, &nh, table, metric, mtu, distance, tag); | |
fd36be7e | 450 | } else { |
d62a17ae | 451 | /* This is a multipath route */ |
452 | ||
453 | struct route_entry *re; | |
454 | struct rtnexthop *rtnh = | |
455 | (struct rtnexthop *)RTA_DATA(tb[RTA_MULTIPATH]); | |
456 | ||
457 | len = RTA_PAYLOAD(tb[RTA_MULTIPATH]); | |
458 | ||
459 | re = XCALLOC(MTYPE_RE, sizeof(struct route_entry)); | |
915902cb | 460 | re->type = proto; |
25715c7e | 461 | re->distance = distance; |
d62a17ae | 462 | re->flags = flags; |
463 | re->metric = metric; | |
464 | re->mtu = mtu; | |
465 | re->vrf_id = vrf_id; | |
466 | re->table = table; | |
467 | re->nexthop_num = 0; | |
468 | re->uptime = time(NULL); | |
4e40b6d6 | 469 | re->tag = tag; |
d62a17ae | 470 | |
471 | for (;;) { | |
4a7371e9 | 472 | vrf_id_t nh_vrf_id; |
d62a17ae | 473 | if (len < (int)sizeof(*rtnh) |
474 | || rtnh->rtnh_len > len) | |
475 | break; | |
476 | ||
477 | index = rtnh->rtnh_ifindex; | |
8795f904 DS |
478 | if (index) { |
479 | /* | |
480 | * Yes we are looking this up | |
481 | * for every nexthop and just | |
482 | * using the last one looked | |
483 | * up right now | |
484 | */ | |
485 | ifp = if_lookup_by_index(index, | |
486 | VRF_UNKNOWN); | |
487 | if (ifp) | |
4a7371e9 DS |
488 | nh_vrf_id = ifp->vrf_id; |
489 | else { | |
490 | zlog_warn( | |
491 | "%s: Unknown interface %u specified, defaulting to VRF_DEFAULT", | |
492 | __PRETTY_FUNCTION__, | |
493 | index); | |
494 | nh_vrf_id = VRF_DEFAULT; | |
495 | } | |
496 | } else | |
497 | nh_vrf_id = vrf_id; | |
498 | ||
d62a17ae | 499 | gate = 0; |
500 | if (rtnh->rtnh_len > sizeof(*rtnh)) { | |
501 | memset(tb, 0, sizeof(tb)); | |
502 | netlink_parse_rtattr( | |
503 | tb, RTA_MAX, RTNH_DATA(rtnh), | |
504 | rtnh->rtnh_len - sizeof(*rtnh)); | |
505 | if (tb[RTA_GATEWAY]) | |
506 | gate = RTA_DATA( | |
507 | tb[RTA_GATEWAY]); | |
508 | } | |
509 | ||
510 | if (gate) { | |
511 | if (rtm->rtm_family == AF_INET) { | |
512 | if (index) | |
513 | route_entry_nexthop_ipv4_ifindex_add( | |
514 | re, gate, | |
4a7371e9 DS |
515 | prefsrc, index, |
516 | nh_vrf_id); | |
d62a17ae | 517 | else |
518 | route_entry_nexthop_ipv4_add( | |
519 | re, gate, | |
4a7371e9 DS |
520 | prefsrc, |
521 | nh_vrf_id); | |
d62a17ae | 522 | } else if (rtm->rtm_family |
523 | == AF_INET6) { | |
524 | if (index) | |
525 | route_entry_nexthop_ipv6_ifindex_add( | |
4a7371e9 DS |
526 | re, gate, index, |
527 | nh_vrf_id); | |
d62a17ae | 528 | else |
529 | route_entry_nexthop_ipv6_add( | |
4a7371e9 DS |
530 | re, gate, |
531 | nh_vrf_id); | |
d62a17ae | 532 | } |
533 | } else | |
4a7371e9 DS |
534 | route_entry_nexthop_ifindex_add( |
535 | re, index, nh_vrf_id); | |
d62a17ae | 536 | |
537 | len -= NLMSG_ALIGN(rtnh->rtnh_len); | |
538 | rtnh = RTNH_NEXT(rtnh); | |
539 | } | |
540 | ||
541 | zserv_nexthop_num_warn(__func__, | |
542 | (const struct prefix *)&p, | |
543 | re->nexthop_num); | |
544 | if (re->nexthop_num == 0) | |
545 | XFREE(MTYPE_RE, re); | |
546 | else | |
7990990e | 547 | rib_add_multipath(afi, SAFI_UNICAST, &p, |
d62a17ae | 548 | NULL, re); |
549 | } | |
550 | } else { | |
fd36be7e DL |
551 | if (!tb[RTA_MULTIPATH]) { |
552 | struct nexthop nh; | |
553 | size_t sz = (afi == AFI_IP) ? 4 : 16; | |
554 | ||
555 | memset(&nh, 0, sizeof(nh)); | |
8ba5bd58 RW |
556 | if (bh_type == BLACKHOLE_UNSPEC) { |
557 | if (index && !gate) | |
558 | nh.type = NEXTHOP_TYPE_IFINDEX; | |
559 | else if (index && gate) | |
560 | nh.type = | |
561 | (afi == AFI_IP) | |
562 | ? NEXTHOP_TYPE_IPV4_IFINDEX | |
563 | : NEXTHOP_TYPE_IPV6_IFINDEX; | |
564 | else if (!index && gate) | |
565 | nh.type = (afi == AFI_IP) | |
566 | ? NEXTHOP_TYPE_IPV4 | |
60466a63 | 567 | : NEXTHOP_TYPE_IPV6; |
8ba5bd58 RW |
568 | else { |
569 | nh.type = NEXTHOP_TYPE_BLACKHOLE; | |
570 | nh.bh_type = BLACKHOLE_UNSPEC; | |
571 | } | |
572 | } else { | |
fd36be7e | 573 | nh.type = NEXTHOP_TYPE_BLACKHOLE; |
8ba5bd58 RW |
574 | nh.bh_type = bh_type; |
575 | } | |
fd36be7e DL |
576 | nh.ifindex = index; |
577 | if (gate) | |
578 | memcpy(&nh.gate, gate, sz); | |
d62a17ae | 579 | rib_delete(afi, SAFI_UNICAST, vrf_id, |
915902cb | 580 | proto, 0, flags, &p, NULL, &nh, |
6134fd82 | 581 | table, metric, true, NULL); |
fd36be7e DL |
582 | } else { |
583 | /* XXX: need to compare the entire list of nexthops | |
584 | * here for NLM_F_APPEND stupidity */ | |
585 | rib_delete(afi, SAFI_UNICAST, vrf_id, | |
915902cb | 586 | proto, 0, flags, &p, NULL, NULL, |
6134fd82 | 587 | table, metric, true, NULL); |
d62a17ae | 588 | } |
589 | } | |
590 | ||
591 | return 0; | |
718e3744 | 592 | } |
593 | ||
e3be0432 DS |
594 | static struct mcast_route_data *mroute = NULL; |
595 | ||
d62a17ae | 596 | static int netlink_route_change_read_multicast(struct sockaddr_nl *snl, |
597 | struct nlmsghdr *h, | |
598 | ns_id_t ns_id, int startup) | |
565fdc75 | 599 | { |
d62a17ae | 600 | int len; |
601 | struct rtmsg *rtm; | |
602 | struct rtattr *tb[RTA_MAX + 1]; | |
603 | struct mcast_route_data *m; | |
604 | struct mcast_route_data mr; | |
605 | int iif = 0; | |
606 | int count; | |
607 | int oif[256]; | |
608 | int oif_count = 0; | |
609 | char sbuf[40]; | |
610 | char gbuf[40]; | |
611 | char oif_list[256] = "\0"; | |
612 | vrf_id_t vrf = ns_id; | |
43b5cc5e | 613 | int table; |
d62a17ae | 614 | |
615 | if (mroute) | |
616 | m = mroute; | |
617 | else { | |
618 | memset(&mr, 0, sizeof(mr)); | |
619 | m = &mr; | |
620 | } | |
621 | ||
622 | rtm = NLMSG_DATA(h); | |
623 | ||
624 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct rtmsg)); | |
625 | ||
626 | memset(tb, 0, sizeof tb); | |
627 | netlink_parse_rtattr(tb, RTA_MAX, RTM_RTA(rtm), len); | |
90d82769 | 628 | |
43b5cc5e DS |
629 | if (tb[RTA_TABLE]) |
630 | table = *(int *)RTA_DATA(tb[RTA_TABLE]); | |
631 | else | |
632 | table = rtm->rtm_table; | |
633 | ||
634 | vrf = vrf_lookup_by_table(table); | |
635 | ||
d62a17ae | 636 | if (tb[RTA_IIF]) |
637 | iif = *(int *)RTA_DATA(tb[RTA_IIF]); | |
638 | ||
639 | if (tb[RTA_SRC]) | |
bd8b9272 | 640 | m->sg.src = *(struct in_addr *)RTA_DATA(tb[RTA_SRC]); |
d62a17ae | 641 | |
642 | if (tb[RTA_DST]) | |
bd8b9272 | 643 | m->sg.grp = *(struct in_addr *)RTA_DATA(tb[RTA_DST]); |
d62a17ae | 644 | |
645 | if ((RTA_EXPIRES <= RTA_MAX) && tb[RTA_EXPIRES]) | |
646 | m->lastused = *(unsigned long long *)RTA_DATA(tb[RTA_EXPIRES]); | |
647 | ||
648 | if (tb[RTA_MULTIPATH]) { | |
649 | struct rtnexthop *rtnh = | |
650 | (struct rtnexthop *)RTA_DATA(tb[RTA_MULTIPATH]); | |
651 | ||
652 | len = RTA_PAYLOAD(tb[RTA_MULTIPATH]); | |
653 | for (;;) { | |
654 | if (len < (int)sizeof(*rtnh) || rtnh->rtnh_len > len) | |
655 | break; | |
656 | ||
657 | oif[oif_count] = rtnh->rtnh_ifindex; | |
658 | oif_count++; | |
659 | ||
660 | len -= NLMSG_ALIGN(rtnh->rtnh_len); | |
661 | rtnh = RTNH_NEXT(rtnh); | |
662 | } | |
663 | } | |
664 | ||
665 | if (IS_ZEBRA_DEBUG_KERNEL) { | |
666 | struct interface *ifp; | |
0af35d90 RW |
667 | strlcpy(sbuf, inet_ntoa(m->sg.src), sizeof(sbuf)); |
668 | strlcpy(gbuf, inet_ntoa(m->sg.grp), sizeof(gbuf)); | |
d62a17ae | 669 | for (count = 0; count < oif_count; count++) { |
670 | ifp = if_lookup_by_index(oif[count], vrf); | |
671 | char temp[256]; | |
672 | ||
673 | sprintf(temp, "%s ", ifp->name); | |
674 | strcat(oif_list, temp); | |
675 | } | |
43b5cc5e | 676 | struct zebra_vrf *zvrf = zebra_vrf_lookup_by_id(vrf); |
d62a17ae | 677 | ifp = if_lookup_by_index(iif, vrf); |
43b5cc5e DS |
678 | zlog_debug( |
679 | "MCAST VRF: %s(%d) %s (%s,%s) IIF: %s OIF: %s jiffies: %lld", | |
680 | zvrf->vrf->name, vrf, nl_msg_type_to_str(h->nlmsg_type), | |
681 | sbuf, gbuf, ifp->name, oif_list, m->lastused); | |
90d82769 | 682 | } |
d62a17ae | 683 | return 0; |
565fdc75 DS |
684 | } |
685 | ||
d62a17ae | 686 | int netlink_route_change(struct sockaddr_nl *snl, struct nlmsghdr *h, |
687 | ns_id_t ns_id, int startup) | |
565fdc75 | 688 | { |
d62a17ae | 689 | int len; |
690 | vrf_id_t vrf_id = ns_id; | |
691 | struct rtmsg *rtm; | |
692 | ||
693 | rtm = NLMSG_DATA(h); | |
694 | ||
695 | if (!(h->nlmsg_type == RTM_NEWROUTE || h->nlmsg_type == RTM_DELROUTE)) { | |
696 | /* If this is not route add/delete message print warning. */ | |
697 | zlog_warn("Kernel message: %d vrf %u\n", h->nlmsg_type, vrf_id); | |
698 | return 0; | |
699 | } | |
700 | ||
701 | /* Connected route. */ | |
702 | if (IS_ZEBRA_DEBUG_KERNEL) | |
703 | zlog_debug("%s %s %s proto %s vrf %u", | |
704 | nl_msg_type_to_str(h->nlmsg_type), | |
705 | nl_family_to_str(rtm->rtm_family), | |
706 | nl_rttype_to_str(rtm->rtm_type), | |
707 | nl_rtproto_to_str(rtm->rtm_protocol), vrf_id); | |
708 | ||
709 | /* We don't care about change notifications for the MPLS table. */ | |
710 | /* TODO: Revisit this. */ | |
711 | if (rtm->rtm_family == AF_MPLS) | |
712 | return 0; | |
713 | ||
714 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct rtmsg)); | |
715 | if (len < 0) | |
716 | return -1; | |
717 | ||
e655a03c | 718 | if (rtm->rtm_type == RTN_MULTICAST) |
d62a17ae | 719 | netlink_route_change_read_multicast(snl, h, ns_id, startup); |
e655a03c DL |
720 | else |
721 | netlink_route_change_read_unicast(snl, h, ns_id, startup); | |
d62a17ae | 722 | return 0; |
565fdc75 DS |
723 | } |
724 | ||
289602d7 | 725 | /* Request for specific route information from the kernel */ |
d62a17ae | 726 | static int netlink_request_route(struct zebra_ns *zns, int family, int type) |
289602d7 | 727 | { |
d62a17ae | 728 | struct { |
729 | struct nlmsghdr n; | |
730 | struct rtmsg rtm; | |
731 | } req; | |
732 | ||
733 | /* Form the request, specifying filter (rtattr) if needed. */ | |
734 | memset(&req, 0, sizeof(req)); | |
735 | req.n.nlmsg_type = type; | |
736 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg)); | |
737 | req.rtm.rtm_family = family; | |
738 | ||
739 | return netlink_request(&zns->netlink_cmd, &req.n); | |
289602d7 | 740 | } |
741 | ||
718e3744 | 742 | /* Routing table read function using netlink interface. Only called |
743 | bootstrap time. */ | |
d62a17ae | 744 | int netlink_route_read(struct zebra_ns *zns) |
718e3744 | 745 | { |
d62a17ae | 746 | int ret; |
747 | ||
748 | /* Get IPv4 routing table. */ | |
749 | ret = netlink_request_route(zns, AF_INET, RTM_GETROUTE); | |
750 | if (ret < 0) | |
751 | return ret; | |
752 | ret = netlink_parse_info(netlink_route_change_read_unicast, | |
753 | &zns->netlink_cmd, zns, 0, 1); | |
754 | if (ret < 0) | |
755 | return ret; | |
756 | ||
757 | /* Get IPv6 routing table. */ | |
758 | ret = netlink_request_route(zns, AF_INET6, RTM_GETROUTE); | |
759 | if (ret < 0) | |
760 | return ret; | |
761 | ret = netlink_parse_info(netlink_route_change_read_unicast, | |
762 | &zns->netlink_cmd, zns, 0, 1); | |
763 | if (ret < 0) | |
764 | return ret; | |
765 | ||
766 | return 0; | |
718e3744 | 767 | } |
768 | ||
d62a17ae | 769 | static void _netlink_route_nl_add_gateway_info(u_char route_family, |
770 | u_char gw_family, | |
771 | struct nlmsghdr *nlmsg, | |
772 | size_t req_size, int bytelen, | |
773 | struct nexthop *nexthop) | |
40c7bdb0 | 774 | { |
d62a17ae | 775 | if (route_family == AF_MPLS) { |
776 | struct gw_family_t gw_fam; | |
777 | ||
778 | gw_fam.family = gw_family; | |
779 | if (gw_family == AF_INET) | |
780 | memcpy(&gw_fam.gate.ipv4, &nexthop->gate.ipv4, bytelen); | |
781 | else | |
782 | memcpy(&gw_fam.gate.ipv6, &nexthop->gate.ipv6, bytelen); | |
783 | addattr_l(nlmsg, req_size, RTA_VIA, &gw_fam.family, | |
784 | bytelen + 2); | |
785 | } else { | |
786 | if (gw_family == AF_INET) | |
787 | addattr_l(nlmsg, req_size, RTA_GATEWAY, | |
788 | &nexthop->gate.ipv4, bytelen); | |
789 | else | |
790 | addattr_l(nlmsg, req_size, RTA_GATEWAY, | |
791 | &nexthop->gate.ipv6, bytelen); | |
792 | } | |
40c7bdb0 | 793 | } |
794 | ||
d62a17ae | 795 | static void _netlink_route_rta_add_gateway_info(u_char route_family, |
796 | u_char gw_family, | |
797 | struct rtattr *rta, | |
798 | struct rtnexthop *rtnh, | |
799 | size_t req_size, int bytelen, | |
800 | struct nexthop *nexthop) | |
40c7bdb0 | 801 | { |
d62a17ae | 802 | if (route_family == AF_MPLS) { |
803 | struct gw_family_t gw_fam; | |
804 | ||
805 | gw_fam.family = gw_family; | |
806 | if (gw_family == AF_INET) | |
807 | memcpy(&gw_fam.gate.ipv4, &nexthop->gate.ipv4, bytelen); | |
808 | else | |
809 | memcpy(&gw_fam.gate.ipv6, &nexthop->gate.ipv6, bytelen); | |
810 | rta_addattr_l(rta, req_size, RTA_VIA, &gw_fam.family, | |
811 | bytelen + 2); | |
812 | rtnh->rtnh_len += RTA_LENGTH(bytelen + 2); | |
813 | } else { | |
814 | if (gw_family == AF_INET) | |
815 | rta_addattr_l(rta, req_size, RTA_GATEWAY, | |
816 | &nexthop->gate.ipv4, bytelen); | |
817 | else | |
818 | rta_addattr_l(rta, req_size, RTA_GATEWAY, | |
819 | &nexthop->gate.ipv6, bytelen); | |
820 | rtnh->rtnh_len += sizeof(struct rtattr) + bytelen; | |
821 | } | |
40c7bdb0 | 822 | } |
823 | ||
fa713d9e CF |
824 | /* This function takes a nexthop as argument and adds |
825 | * the appropriate netlink attributes to an existing | |
826 | * netlink message. | |
827 | * | |
828 | * @param routedesc: Human readable description of route type | |
829 | * (direct/recursive, single-/multipath) | |
830 | * @param bytelen: Length of addresses in bytes. | |
831 | * @param nexthop: Nexthop information | |
832 | * @param nlmsg: nlmsghdr structure to fill in. | |
833 | * @param req_size: The size allocated for the message. | |
834 | */ | |
d62a17ae | 835 | static void _netlink_route_build_singlepath(const char *routedesc, int bytelen, |
836 | struct nexthop *nexthop, | |
837 | struct nlmsghdr *nlmsg, | |
838 | struct rtmsg *rtmsg, | |
839 | size_t req_size, int cmd) | |
fa713d9e | 840 | { |
8ecdb26e | 841 | struct mpls_label_stack *nh_label; |
d62a17ae | 842 | mpls_lse_t out_lse[MPLS_MAX_LABELS]; |
9a62e84b | 843 | char label_buf[256]; |
d62a17ae | 844 | |
845 | /* | |
846 | * label_buf is *only* currently used within debugging. | |
847 | * As such when we assign it we are guarding it inside | |
848 | * a debug test. If you want to change this make sure | |
849 | * you fix this assumption | |
850 | */ | |
851 | label_buf[0] = '\0'; | |
852 | /* outgoing label - either as NEWDST (in the case of LSR) or as ENCAP | |
853 | * (in the case of LER) | |
854 | */ | |
855 | nh_label = nexthop->nh_label; | |
856 | if (rtmsg->rtm_family == AF_MPLS) { | |
857 | assert(nh_label); | |
858 | assert(nh_label->num_labels == 1); | |
859 | } | |
860 | ||
861 | if (nh_label && nh_label->num_labels) { | |
862 | int i, num_labels = 0; | |
863 | u_int32_t bos; | |
864 | char label_buf1[20]; | |
865 | ||
866 | for (i = 0; i < nh_label->num_labels; i++) { | |
70e98a7f | 867 | if (nh_label->label[i] != MPLS_LABEL_IMPLICIT_NULL) { |
d62a17ae | 868 | bos = ((i == (nh_label->num_labels - 1)) ? 1 |
869 | : 0); | |
870 | out_lse[i] = mpls_lse_encode(nh_label->label[i], | |
871 | 0, 0, bos); | |
872 | if (IS_ZEBRA_DEBUG_KERNEL) { | |
873 | if (!num_labels) | |
9a62e84b | 874 | sprintf(label_buf, "label %u", |
d62a17ae | 875 | nh_label->label[i]); |
876 | else { | |
9a62e84b | 877 | sprintf(label_buf1, "/%u", |
d62a17ae | 878 | nh_label->label[i]); |
9a62e84b RW |
879 | strlcat(label_buf, label_buf1, |
880 | sizeof(label_buf)); | |
d62a17ae | 881 | } |
882 | } | |
883 | num_labels++; | |
884 | } | |
885 | } | |
886 | if (num_labels) { | |
887 | if (rtmsg->rtm_family == AF_MPLS) | |
888 | addattr_l(nlmsg, req_size, RTA_NEWDST, &out_lse, | |
889 | num_labels * sizeof(mpls_lse_t)); | |
890 | else { | |
891 | struct rtattr *nest; | |
892 | u_int16_t encap = LWTUNNEL_ENCAP_MPLS; | |
893 | ||
894 | addattr_l(nlmsg, req_size, RTA_ENCAP_TYPE, | |
895 | &encap, sizeof(u_int16_t)); | |
896 | nest = addattr_nest(nlmsg, req_size, RTA_ENCAP); | |
897 | addattr_l(nlmsg, req_size, MPLS_IPTUNNEL_DST, | |
898 | &out_lse, | |
899 | num_labels * sizeof(mpls_lse_t)); | |
900 | addattr_nest_end(nlmsg, nest); | |
901 | } | |
66d42727 | 902 | } |
0aabccc0 | 903 | } |
fa713d9e | 904 | |
d62a17ae | 905 | if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ONLINK)) |
906 | rtmsg->rtm_flags |= RTNH_F_ONLINK; | |
907 | ||
908 | if (rtmsg->rtm_family == AF_INET | |
909 | && (nexthop->type == NEXTHOP_TYPE_IPV6 | |
910 | || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX)) { | |
911 | rtmsg->rtm_flags |= RTNH_F_ONLINK; | |
912 | addattr_l(nlmsg, req_size, RTA_GATEWAY, &ipv4_ll, 4); | |
913 | addattr32(nlmsg, req_size, RTA_OIF, nexthop->ifindex); | |
914 | ||
915 | if (nexthop->rmap_src.ipv4.s_addr && (cmd == RTM_NEWROUTE)) | |
916 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
917 | &nexthop->rmap_src.ipv4, bytelen); | |
918 | else if (nexthop->src.ipv4.s_addr && (cmd == RTM_NEWROUTE)) | |
919 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
920 | &nexthop->src.ipv4, bytelen); | |
921 | ||
922 | if (IS_ZEBRA_DEBUG_KERNEL) | |
923 | zlog_debug( | |
924 | " 5549: _netlink_route_build_singlepath() (%s): " | |
925 | "nexthop via %s %s if %u", | |
926 | routedesc, ipv4_ll_buf, label_buf, | |
927 | nexthop->ifindex); | |
928 | return; | |
0aabccc0 DD |
929 | } |
930 | ||
d62a17ae | 931 | if (nexthop->type == NEXTHOP_TYPE_IPV4 |
932 | || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX) { | |
933 | /* Send deletes to the kernel without specifying the next-hop */ | |
934 | if (cmd != RTM_DELROUTE) | |
935 | _netlink_route_nl_add_gateway_info( | |
936 | rtmsg->rtm_family, AF_INET, nlmsg, req_size, | |
937 | bytelen, nexthop); | |
938 | ||
939 | if (cmd == RTM_NEWROUTE) { | |
940 | if (nexthop->rmap_src.ipv4.s_addr) | |
941 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
942 | &nexthop->rmap_src.ipv4, bytelen); | |
943 | else if (nexthop->src.ipv4.s_addr) | |
944 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
945 | &nexthop->src.ipv4, bytelen); | |
946 | } | |
947 | ||
948 | if (IS_ZEBRA_DEBUG_KERNEL) | |
949 | zlog_debug( | |
950 | "netlink_route_multipath() (%s): " | |
951 | "nexthop via %s %s if %u", | |
952 | routedesc, inet_ntoa(nexthop->gate.ipv4), | |
953 | label_buf, nexthop->ifindex); | |
0aabccc0 | 954 | } |
fa713d9e | 955 | |
d62a17ae | 956 | if (nexthop->type == NEXTHOP_TYPE_IPV6 |
957 | || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) { | |
958 | _netlink_route_nl_add_gateway_info(rtmsg->rtm_family, AF_INET6, | |
959 | nlmsg, req_size, bytelen, | |
960 | nexthop); | |
961 | ||
962 | if (cmd == RTM_NEWROUTE) { | |
963 | if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->rmap_src.ipv6)) | |
964 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
965 | &nexthop->rmap_src.ipv6, bytelen); | |
966 | else if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->src.ipv6)) | |
967 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
968 | &nexthop->src.ipv6, bytelen); | |
969 | } | |
fa713d9e | 970 | |
d62a17ae | 971 | if (IS_ZEBRA_DEBUG_KERNEL) |
972 | zlog_debug( | |
973 | "netlink_route_multipath() (%s): " | |
974 | "nexthop via %s %s if %u", | |
975 | routedesc, inet6_ntoa(nexthop->gate.ipv6), | |
976 | label_buf, nexthop->ifindex); | |
977 | } | |
5e210522 DS |
978 | |
979 | /* | |
980 | * We have the ifindex so we should always send it | |
981 | * This is especially useful if we are doing route | |
982 | * leaking. | |
983 | */ | |
984 | if (nexthop->type != NEXTHOP_TYPE_BLACKHOLE) | |
d62a17ae | 985 | addattr32(nlmsg, req_size, RTA_OIF, nexthop->ifindex); |
986 | ||
5e210522 DS |
987 | if (nexthop->type == NEXTHOP_TYPE_IFINDEX |
988 | || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX) { | |
d62a17ae | 989 | if (cmd == RTM_NEWROUTE) { |
990 | if (nexthop->rmap_src.ipv4.s_addr) | |
991 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
992 | &nexthop->rmap_src.ipv4, bytelen); | |
993 | else if (nexthop->src.ipv4.s_addr) | |
994 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
995 | &nexthop->src.ipv4, bytelen); | |
996 | } | |
fa713d9e | 997 | |
d62a17ae | 998 | if (IS_ZEBRA_DEBUG_KERNEL) |
999 | zlog_debug( | |
1000 | "netlink_route_multipath() (%s): " | |
1001 | "nexthop via if %u", | |
1002 | routedesc, nexthop->ifindex); | |
0aabccc0 DD |
1003 | } |
1004 | ||
d62a17ae | 1005 | if (nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) { |
d62a17ae | 1006 | if (cmd == RTM_NEWROUTE) { |
1007 | if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->rmap_src.ipv6)) | |
1008 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1009 | &nexthop->rmap_src.ipv6, bytelen); | |
1010 | else if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->src.ipv6)) | |
1011 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1012 | &nexthop->src.ipv6, bytelen); | |
1013 | } | |
1014 | ||
1015 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1016 | zlog_debug( | |
1017 | "netlink_route_multipath() (%s): " | |
1018 | "nexthop via if %u", | |
1019 | routedesc, nexthop->ifindex); | |
1020 | } | |
fa713d9e CF |
1021 | } |
1022 | ||
1023 | /* This function takes a nexthop as argument and | |
1024 | * appends to the given rtattr/rtnexthop pair the | |
1025 | * representation of the nexthop. If the nexthop | |
1026 | * defines a preferred source, the src parameter | |
1027 | * will be modified to point to that src, otherwise | |
1028 | * it will be kept unmodified. | |
1029 | * | |
1030 | * @param routedesc: Human readable description of route type | |
1031 | * (direct/recursive, single-/multipath) | |
1032 | * @param bytelen: Length of addresses in bytes. | |
1033 | * @param nexthop: Nexthop information | |
1034 | * @param rta: rtnetlink attribute structure | |
1035 | * @param rtnh: pointer to an rtnetlink nexthop structure | |
1036 | * @param src: pointer pointing to a location where | |
1037 | * the prefsrc should be stored. | |
1038 | */ | |
d62a17ae | 1039 | static void _netlink_route_build_multipath(const char *routedesc, int bytelen, |
1040 | struct nexthop *nexthop, | |
1041 | struct rtattr *rta, | |
1042 | struct rtnexthop *rtnh, | |
1043 | struct rtmsg *rtmsg, | |
1044 | union g_addr **src) | |
fa713d9e | 1045 | { |
8ecdb26e | 1046 | struct mpls_label_stack *nh_label; |
d62a17ae | 1047 | mpls_lse_t out_lse[MPLS_MAX_LABELS]; |
9a62e84b | 1048 | char label_buf[256]; |
d62a17ae | 1049 | |
1050 | rtnh->rtnh_len = sizeof(*rtnh); | |
1051 | rtnh->rtnh_flags = 0; | |
1052 | rtnh->rtnh_hops = 0; | |
1053 | rta->rta_len += rtnh->rtnh_len; | |
1054 | ||
1055 | /* | |
1056 | * label_buf is *only* currently used within debugging. | |
1057 | * As such when we assign it we are guarding it inside | |
1058 | * a debug test. If you want to change this make sure | |
1059 | * you fix this assumption | |
1060 | */ | |
1061 | label_buf[0] = '\0'; | |
1062 | /* outgoing label - either as NEWDST (in the case of LSR) or as ENCAP | |
1063 | * (in the case of LER) | |
1064 | */ | |
1065 | nh_label = nexthop->nh_label; | |
1066 | if (rtmsg->rtm_family == AF_MPLS) { | |
1067 | assert(nh_label); | |
1068 | assert(nh_label->num_labels == 1); | |
1069 | } | |
1070 | ||
1071 | if (nh_label && nh_label->num_labels) { | |
1072 | int i, num_labels = 0; | |
1073 | u_int32_t bos; | |
1074 | char label_buf1[20]; | |
1075 | ||
1076 | for (i = 0; i < nh_label->num_labels; i++) { | |
70e98a7f | 1077 | if (nh_label->label[i] != MPLS_LABEL_IMPLICIT_NULL) { |
d62a17ae | 1078 | bos = ((i == (nh_label->num_labels - 1)) ? 1 |
1079 | : 0); | |
1080 | out_lse[i] = mpls_lse_encode(nh_label->label[i], | |
1081 | 0, 0, bos); | |
1082 | if (IS_ZEBRA_DEBUG_KERNEL) { | |
1083 | if (!num_labels) | |
9a62e84b | 1084 | sprintf(label_buf, "label %u", |
d62a17ae | 1085 | nh_label->label[i]); |
1086 | else { | |
9a62e84b | 1087 | sprintf(label_buf1, "/%u", |
d62a17ae | 1088 | nh_label->label[i]); |
9a62e84b RW |
1089 | strlcat(label_buf, label_buf1, |
1090 | sizeof(label_buf)); | |
d62a17ae | 1091 | } |
1092 | } | |
1093 | num_labels++; | |
1094 | } | |
1095 | } | |
1096 | if (num_labels) { | |
1097 | if (rtmsg->rtm_family == AF_MPLS) { | |
1098 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, RTA_NEWDST, | |
1099 | &out_lse, | |
1100 | num_labels * sizeof(mpls_lse_t)); | |
1101 | rtnh->rtnh_len += RTA_LENGTH( | |
1102 | num_labels * sizeof(mpls_lse_t)); | |
1103 | } else { | |
1104 | struct rtattr *nest; | |
1105 | u_int16_t encap = LWTUNNEL_ENCAP_MPLS; | |
1106 | int len = rta->rta_len; | |
1107 | ||
1108 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, | |
1109 | RTA_ENCAP_TYPE, &encap, | |
1110 | sizeof(u_int16_t)); | |
1111 | nest = rta_nest(rta, NL_PKT_BUF_SIZE, | |
1112 | RTA_ENCAP); | |
1113 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, | |
1114 | MPLS_IPTUNNEL_DST, &out_lse, | |
1115 | num_labels * sizeof(mpls_lse_t)); | |
1116 | rta_nest_end(rta, nest); | |
1117 | rtnh->rtnh_len += rta->rta_len - len; | |
1118 | } | |
66d42727 | 1119 | } |
d62a17ae | 1120 | } |
1121 | ||
1122 | if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ONLINK)) | |
1123 | rtnh->rtnh_flags |= RTNH_F_ONLINK; | |
1124 | ||
1125 | if (rtmsg->rtm_family == AF_INET | |
1126 | && (nexthop->type == NEXTHOP_TYPE_IPV6 | |
1127 | || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX)) { | |
1128 | bytelen = 4; | |
1129 | rtnh->rtnh_flags |= RTNH_F_ONLINK; | |
1130 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, RTA_GATEWAY, &ipv4_ll, | |
1131 | bytelen); | |
1132 | rtnh->rtnh_len += sizeof(struct rtattr) + bytelen; | |
1133 | rtnh->rtnh_ifindex = nexthop->ifindex; | |
1134 | ||
1135 | if (nexthop->rmap_src.ipv4.s_addr) | |
1136 | *src = &nexthop->rmap_src; | |
1137 | else if (nexthop->src.ipv4.s_addr) | |
1138 | *src = &nexthop->src; | |
1139 | ||
1140 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1141 | zlog_debug( | |
1142 | " 5549: netlink_route_build_multipath() (%s): " | |
1143 | "nexthop via %s %s if %u", | |
1144 | routedesc, ipv4_ll_buf, label_buf, | |
1145 | nexthop->ifindex); | |
1146 | return; | |
1147 | } | |
1148 | ||
1149 | if (nexthop->type == NEXTHOP_TYPE_IPV4 | |
1150 | || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX) { | |
1151 | _netlink_route_rta_add_gateway_info(rtmsg->rtm_family, AF_INET, | |
1152 | rta, rtnh, NL_PKT_BUF_SIZE, | |
1153 | bytelen, nexthop); | |
1154 | if (nexthop->rmap_src.ipv4.s_addr) | |
1155 | *src = &nexthop->rmap_src; | |
1156 | else if (nexthop->src.ipv4.s_addr) | |
1157 | *src = &nexthop->src; | |
1158 | ||
1159 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1160 | zlog_debug( | |
1161 | "netlink_route_multipath() (%s): " | |
1162 | "nexthop via %s %s if %u", | |
1163 | routedesc, inet_ntoa(nexthop->gate.ipv4), | |
1164 | label_buf, nexthop->ifindex); | |
1165 | } | |
1166 | if (nexthop->type == NEXTHOP_TYPE_IPV6 | |
1167 | || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) { | |
1168 | _netlink_route_rta_add_gateway_info(rtmsg->rtm_family, AF_INET6, | |
1169 | rta, rtnh, NL_PKT_BUF_SIZE, | |
1170 | bytelen, nexthop); | |
1171 | ||
1172 | if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->rmap_src.ipv6)) | |
1173 | *src = &nexthop->rmap_src; | |
1174 | else if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->src.ipv6)) | |
1175 | *src = &nexthop->src; | |
1176 | ||
1177 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1178 | zlog_debug( | |
1179 | "netlink_route_multipath() (%s): " | |
1180 | "nexthop via %s %s if %u", | |
1181 | routedesc, inet6_ntoa(nexthop->gate.ipv6), | |
1182 | label_buf, nexthop->ifindex); | |
1183 | } | |
5e210522 DS |
1184 | |
1185 | /* | |
1186 | * We have figured out the ifindex so we should always send it | |
1187 | * This is especially useful if we are doing route | |
1188 | * leaking. | |
1189 | */ | |
1190 | if (nexthop->type != NEXTHOP_TYPE_BLACKHOLE) | |
1191 | rtnh->rtnh_ifindex = nexthop->ifindex; | |
1192 | ||
d62a17ae | 1193 | /* ifindex */ |
1194 | if (nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX | |
1195 | || nexthop->type == NEXTHOP_TYPE_IFINDEX) { | |
d62a17ae | 1196 | if (nexthop->rmap_src.ipv4.s_addr) |
1197 | *src = &nexthop->rmap_src; | |
1198 | else if (nexthop->src.ipv4.s_addr) | |
1199 | *src = &nexthop->src; | |
1200 | ||
1201 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1202 | zlog_debug( | |
1203 | "netlink_route_multipath() (%s): " | |
1204 | "nexthop via if %u", | |
1205 | routedesc, nexthop->ifindex); | |
1206 | } else if (nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) { | |
d62a17ae | 1207 | if (IS_ZEBRA_DEBUG_KERNEL) |
1208 | zlog_debug( | |
1209 | "netlink_route_multipath() (%s): " | |
1210 | "nexthop via if %u", | |
1211 | routedesc, nexthop->ifindex); | |
1212 | } else { | |
1213 | rtnh->rtnh_ifindex = 0; | |
1214 | } | |
fa713d9e CF |
1215 | } |
1216 | ||
d62a17ae | 1217 | static inline void _netlink_mpls_build_singlepath(const char *routedesc, |
1218 | zebra_nhlfe_t *nhlfe, | |
1219 | struct nlmsghdr *nlmsg, | |
1220 | struct rtmsg *rtmsg, | |
1221 | size_t req_size, int cmd) | |
40c7bdb0 | 1222 | { |
d62a17ae | 1223 | int bytelen; |
1224 | u_char family; | |
40c7bdb0 | 1225 | |
d62a17ae | 1226 | family = NHLFE_FAMILY(nhlfe); |
1227 | bytelen = (family == AF_INET ? 4 : 16); | |
1228 | _netlink_route_build_singlepath(routedesc, bytelen, nhlfe->nexthop, | |
1229 | nlmsg, rtmsg, req_size, cmd); | |
40c7bdb0 | 1230 | } |
1231 | ||
1232 | ||
1233 | static inline void | |
d62a17ae | 1234 | _netlink_mpls_build_multipath(const char *routedesc, zebra_nhlfe_t *nhlfe, |
1235 | struct rtattr *rta, struct rtnexthop *rtnh, | |
1236 | struct rtmsg *rtmsg, union g_addr **src) | |
40c7bdb0 | 1237 | { |
d62a17ae | 1238 | int bytelen; |
1239 | u_char family; | |
40c7bdb0 | 1240 | |
d62a17ae | 1241 | family = NHLFE_FAMILY(nhlfe); |
1242 | bytelen = (family == AF_INET ? 4 : 16); | |
1243 | _netlink_route_build_multipath(routedesc, bytelen, nhlfe->nexthop, rta, | |
1244 | rtnh, rtmsg, src); | |
40c7bdb0 | 1245 | } |
1246 | ||
1247 | ||
fa713d9e CF |
1248 | /* Log debug information for netlink_route_multipath |
1249 | * if debug logging is enabled. | |
1250 | * | |
1251 | * @param cmd: Netlink command which is to be processed | |
1252 | * @param p: Prefix for which the change is due | |
1253 | * @param nexthop: Nexthop which is currently processed | |
1254 | * @param routedesc: Semantic annotation for nexthop | |
1255 | * (recursive, multipath, etc.) | |
1256 | * @param family: Address family which the change concerns | |
1257 | */ | |
d62a17ae | 1258 | static void _netlink_route_debug(int cmd, struct prefix *p, |
1259 | struct nexthop *nexthop, const char *routedesc, | |
1260 | int family, struct zebra_vrf *zvrf) | |
fa713d9e | 1261 | { |
d62a17ae | 1262 | if (IS_ZEBRA_DEBUG_KERNEL) { |
1263 | char buf[PREFIX_STRLEN]; | |
1264 | zlog_debug( | |
1265 | "netlink_route_multipath() (%s): %s %s vrf %u type %s", | |
1266 | routedesc, nl_msg_type_to_str(cmd), | |
1267 | prefix2str(p, buf, sizeof(buf)), zvrf_id(zvrf), | |
1268 | (nexthop) ? nexthop_type_to_str(nexthop->type) : "UNK"); | |
1269 | } | |
1270 | } | |
1271 | ||
1272 | static void _netlink_mpls_debug(int cmd, u_int32_t label, const char *routedesc) | |
40c7bdb0 | 1273 | { |
d62a17ae | 1274 | if (IS_ZEBRA_DEBUG_KERNEL) |
1275 | zlog_debug("netlink_mpls_multipath() (%s): %s %u/20", routedesc, | |
1276 | nl_msg_type_to_str(cmd), label); | |
fa713d9e CF |
1277 | } |
1278 | ||
d62a17ae | 1279 | static int netlink_neigh_update(int cmd, int ifindex, uint32_t addr, char *lla, |
1280 | int llalen) | |
5c610faf | 1281 | { |
d62a17ae | 1282 | struct { |
1283 | struct nlmsghdr n; | |
1284 | struct ndmsg ndm; | |
1285 | char buf[256]; | |
1286 | } req; | |
5c610faf | 1287 | |
d62a17ae | 1288 | struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT); |
8f7d9fc0 | 1289 | |
d62a17ae | 1290 | memset(&req.n, 0, sizeof(req.n)); |
1291 | memset(&req.ndm, 0, sizeof(req.ndm)); | |
5c610faf | 1292 | |
d62a17ae | 1293 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); |
1294 | req.n.nlmsg_flags = NLM_F_CREATE | NLM_F_REQUEST; | |
1295 | req.n.nlmsg_type = cmd; // RTM_NEWNEIGH or RTM_DELNEIGH | |
1296 | req.n.nlmsg_pid = zns->netlink_cmd.snl.nl_pid; | |
a55ba23f | 1297 | |
d62a17ae | 1298 | req.ndm.ndm_family = AF_INET; |
1299 | req.ndm.ndm_state = NUD_PERMANENT; | |
1300 | req.ndm.ndm_ifindex = ifindex; | |
1301 | req.ndm.ndm_type = RTN_UNICAST; | |
5c610faf | 1302 | |
d62a17ae | 1303 | addattr_l(&req.n, sizeof(req), NDA_DST, &addr, 4); |
1304 | addattr_l(&req.n, sizeof(req), NDA_LLADDR, lla, llalen); | |
5c610faf | 1305 | |
d62a17ae | 1306 | return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns, |
1307 | 0); | |
5c610faf DS |
1308 | } |
1309 | ||
718e3744 | 1310 | /* Routing table change via netlink interface. */ |
6ae24471 | 1311 | /* Update flag indicates whether this is a "replace" or not. */ |
d62a17ae | 1312 | static int netlink_route_multipath(int cmd, struct prefix *p, |
1313 | struct prefix *src_p, struct route_entry *re, | |
1314 | int update) | |
718e3744 | 1315 | { |
d62a17ae | 1316 | int bytelen; |
1317 | struct sockaddr_nl snl; | |
1318 | struct nexthop *nexthop = NULL; | |
1319 | unsigned int nexthop_num; | |
a8309422 | 1320 | int discard = 0; |
d62a17ae | 1321 | int family = PREFIX_FAMILY(p); |
1322 | const char *routedesc; | |
1323 | int setsrc = 0; | |
1324 | union g_addr src; | |
1325 | ||
1326 | struct { | |
1327 | struct nlmsghdr n; | |
1328 | struct rtmsg r; | |
1329 | char buf[NL_PKT_BUF_SIZE]; | |
1330 | } req; | |
1331 | ||
1332 | struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT); | |
1333 | struct zebra_vrf *zvrf = vrf_info_lookup(re->vrf_id); | |
1334 | ||
1335 | memset(&req, 0, sizeof req - NL_PKT_BUF_SIZE); | |
1336 | ||
1337 | bytelen = (family == AF_INET ? 4 : 16); | |
1338 | ||
1339 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg)); | |
1340 | req.n.nlmsg_flags = NLM_F_CREATE | NLM_F_REQUEST; | |
1341 | if ((cmd == RTM_NEWROUTE) && update) | |
1342 | req.n.nlmsg_flags |= NLM_F_REPLACE; | |
1343 | req.n.nlmsg_type = cmd; | |
1344 | req.n.nlmsg_pid = zns->netlink_cmd.snl.nl_pid; | |
1345 | ||
1346 | req.r.rtm_family = family; | |
1347 | req.r.rtm_dst_len = p->prefixlen; | |
1348 | req.r.rtm_src_len = src_p ? src_p->prefixlen : 0; | |
915902cb | 1349 | req.r.rtm_protocol = zebra2proto(re->type); |
d62a17ae | 1350 | req.r.rtm_scope = RT_SCOPE_UNIVERSE; |
a8309422 | 1351 | req.r.rtm_type = RTN_UNICAST; |
d62a17ae | 1352 | |
d62a17ae | 1353 | addattr_l(&req.n, sizeof req, RTA_DST, &p->u.prefix, bytelen); |
1354 | if (src_p) | |
1355 | addattr_l(&req.n, sizeof req, RTA_SRC, &src_p->u.prefix, | |
1356 | bytelen); | |
1357 | ||
1358 | /* Metric. */ | |
1359 | /* Hardcode the metric for all routes coming from zebra. Metric isn't | |
1360 | * used | |
1361 | * either by the kernel or by zebra. Its purely for calculating best | |
1362 | * path(s) | |
1363 | * by the routing protocol and for communicating with protocol peers. | |
1364 | */ | |
1365 | addattr32(&req.n, sizeof req, RTA_PRIORITY, NL_DEFAULT_ROUTE_METRIC); | |
4e40b6d6 KK |
1366 | #if defined(SUPPORT_REALMS) |
1367 | if (re->tag > 0 && re->tag <= 255) | |
1368 | addattr32(&req.n, sizeof req, RTA_FLOW, re->tag); | |
1369 | #endif | |
d62a17ae | 1370 | /* Table corresponding to this route. */ |
1371 | if (re->table < 256) | |
1372 | req.r.rtm_table = re->table; | |
1373 | else { | |
1374 | req.r.rtm_table = RT_TABLE_UNSPEC; | |
1375 | addattr32(&req.n, sizeof req, RTA_TABLE, re->table); | |
0aabccc0 | 1376 | } |
718e3744 | 1377 | |
a8309422 DL |
1378 | if (discard) |
1379 | goto skip; | |
1380 | ||
d62a17ae | 1381 | if (re->mtu || re->nexthop_mtu) { |
1382 | char buf[NL_PKT_BUF_SIZE]; | |
1383 | struct rtattr *rta = (void *)buf; | |
1384 | u_int32_t mtu = re->mtu; | |
1385 | if (!mtu || (re->nexthop_mtu && re->nexthop_mtu < mtu)) | |
1386 | mtu = re->nexthop_mtu; | |
1387 | rta->rta_type = RTA_METRICS; | |
1388 | rta->rta_len = RTA_LENGTH(0); | |
1389 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, RTAX_MTU, &mtu, sizeof mtu); | |
1390 | addattr_l(&req.n, NL_PKT_BUF_SIZE, RTA_METRICS, RTA_DATA(rta), | |
1391 | RTA_PAYLOAD(rta)); | |
1392 | } | |
1393 | ||
d62a17ae | 1394 | /* Count overall nexthops so we can decide whether to use singlepath |
1395 | * or multipath case. */ | |
1396 | nexthop_num = 0; | |
1397 | for (ALL_NEXTHOPS(re->nexthop, nexthop)) { | |
1398 | if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_RECURSIVE)) | |
1399 | continue; | |
1400 | if (cmd == RTM_NEWROUTE | |
25b9cb0c | 1401 | && !NEXTHOP_IS_ACTIVE(nexthop->flags)) |
d62a17ae | 1402 | continue; |
1403 | if (cmd == RTM_DELROUTE | |
1404 | && !CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB)) | |
1405 | continue; | |
1406 | ||
1407 | nexthop_num++; | |
1408 | } | |
1409 | ||
1410 | /* Singlepath case. */ | |
1411 | if (nexthop_num == 1 || multipath_num == 1) { | |
1412 | nexthop_num = 0; | |
1413 | for (ALL_NEXTHOPS(re->nexthop, nexthop)) { | |
1f1d24a8 DS |
1414 | /* |
1415 | * So we want to cover 2 types of blackhole | |
1416 | * routes here: | |
1417 | * 1) A normal blackhole route( ala from a static | |
1418 | * install. | |
1419 | * 2) A recursively resolved blackhole route | |
1420 | */ | |
1421 | if (nexthop->type == NEXTHOP_TYPE_BLACKHOLE) { | |
1422 | switch (nexthop->bh_type) { | |
1423 | case BLACKHOLE_ADMINPROHIB: | |
1424 | req.r.rtm_type = RTN_PROHIBIT; | |
1425 | break; | |
1426 | case BLACKHOLE_REJECT: | |
1427 | req.r.rtm_type = RTN_UNREACHABLE; | |
1428 | break; | |
1429 | default: | |
1430 | req.r.rtm_type = RTN_BLACKHOLE; | |
1431 | break; | |
1432 | } | |
1433 | goto skip; | |
1434 | } | |
d62a17ae | 1435 | if (CHECK_FLAG(nexthop->flags, |
1436 | NEXTHOP_FLAG_RECURSIVE)) { | |
1437 | if (!setsrc) { | |
1438 | if (family == AF_INET) { | |
1439 | if (nexthop->rmap_src.ipv4 | |
1440 | .s_addr | |
1441 | != 0) { | |
1442 | src.ipv4 = | |
1443 | nexthop->rmap_src | |
1444 | .ipv4; | |
1445 | setsrc = 1; | |
1446 | } else if (nexthop->src.ipv4 | |
1447 | .s_addr | |
1448 | != 0) { | |
1449 | src.ipv4 = | |
1450 | nexthop->src | |
1451 | .ipv4; | |
1452 | setsrc = 1; | |
1453 | } | |
1454 | } else if (family == AF_INET6) { | |
1455 | if (!IN6_IS_ADDR_UNSPECIFIED( | |
1456 | &nexthop->rmap_src | |
1457 | .ipv6)) { | |
1458 | src.ipv6 = | |
1459 | nexthop->rmap_src | |
1460 | .ipv6; | |
1461 | setsrc = 1; | |
1462 | } else if ( | |
1463 | !IN6_IS_ADDR_UNSPECIFIED( | |
1464 | &nexthop->src | |
1465 | .ipv6)) { | |
1466 | src.ipv6 = | |
1467 | nexthop->src | |
1468 | .ipv6; | |
1469 | setsrc = 1; | |
1470 | } | |
1471 | } | |
1472 | } | |
1473 | continue; | |
1474 | } | |
1475 | ||
1476 | if ((cmd == RTM_NEWROUTE | |
25b9cb0c | 1477 | && NEXTHOP_IS_ACTIVE(nexthop->flags)) |
d62a17ae | 1478 | || (cmd == RTM_DELROUTE |
1479 | && CHECK_FLAG(nexthop->flags, | |
1480 | NEXTHOP_FLAG_FIB))) { | |
1481 | routedesc = nexthop->rparent | |
8b1450b9 RW |
1482 | ? "recursive, single-path" |
1483 | : "single-path"; | |
d62a17ae | 1484 | |
1485 | _netlink_route_debug(cmd, p, nexthop, routedesc, | |
1486 | family, zvrf); | |
1487 | _netlink_route_build_singlepath( | |
1488 | routedesc, bytelen, nexthop, &req.n, | |
1489 | &req.r, sizeof req, cmd); | |
1490 | nexthop_num++; | |
1491 | break; | |
1492 | } | |
1493 | } | |
1494 | if (setsrc && (cmd == RTM_NEWROUTE)) { | |
1495 | if (family == AF_INET) | |
1496 | addattr_l(&req.n, sizeof req, RTA_PREFSRC, | |
1497 | &src.ipv4, bytelen); | |
1498 | else if (family == AF_INET6) | |
1499 | addattr_l(&req.n, sizeof req, RTA_PREFSRC, | |
1500 | &src.ipv6, bytelen); | |
1501 | } | |
1502 | } else { | |
1503 | char buf[NL_PKT_BUF_SIZE]; | |
1504 | struct rtattr *rta = (void *)buf; | |
1505 | struct rtnexthop *rtnh; | |
1506 | union g_addr *src1 = NULL; | |
1507 | ||
1508 | rta->rta_type = RTA_MULTIPATH; | |
1509 | rta->rta_len = RTA_LENGTH(0); | |
1510 | rtnh = RTA_DATA(rta); | |
1511 | ||
1512 | nexthop_num = 0; | |
1513 | for (ALL_NEXTHOPS(re->nexthop, nexthop)) { | |
1514 | if (nexthop_num >= multipath_num) | |
1515 | break; | |
1516 | ||
1517 | if (CHECK_FLAG(nexthop->flags, | |
1518 | NEXTHOP_FLAG_RECURSIVE)) { | |
1519 | /* This only works for IPv4 now */ | |
1520 | if (!setsrc) { | |
1521 | if (family == AF_INET) { | |
1522 | if (nexthop->rmap_src.ipv4 | |
1523 | .s_addr | |
1524 | != 0) { | |
1525 | src.ipv4 = | |
1526 | nexthop->rmap_src | |
1527 | .ipv4; | |
1528 | setsrc = 1; | |
1529 | } else if (nexthop->src.ipv4 | |
1530 | .s_addr | |
1531 | != 0) { | |
1532 | src.ipv4 = | |
1533 | nexthop->src | |
1534 | .ipv4; | |
1535 | setsrc = 1; | |
1536 | } | |
1537 | } else if (family == AF_INET6) { | |
1538 | if (!IN6_IS_ADDR_UNSPECIFIED( | |
1539 | &nexthop->rmap_src | |
1540 | .ipv6)) { | |
1541 | src.ipv6 = | |
1542 | nexthop->rmap_src | |
1543 | .ipv6; | |
1544 | setsrc = 1; | |
1545 | } else if ( | |
1546 | !IN6_IS_ADDR_UNSPECIFIED( | |
1547 | &nexthop->src | |
1548 | .ipv6)) { | |
1549 | src.ipv6 = | |
1550 | nexthop->src | |
1551 | .ipv6; | |
1552 | setsrc = 1; | |
1553 | } | |
1554 | } | |
1555 | } | |
1556 | continue; | |
1557 | } | |
1558 | ||
1559 | if ((cmd == RTM_NEWROUTE | |
25b9cb0c | 1560 | && NEXTHOP_IS_ACTIVE(nexthop->flags)) |
d62a17ae | 1561 | || (cmd == RTM_DELROUTE |
1562 | && CHECK_FLAG(nexthop->flags, | |
1563 | NEXTHOP_FLAG_FIB))) { | |
1564 | routedesc = nexthop->rparent | |
8b1450b9 RW |
1565 | ? "recursive, multipath" |
1566 | : "multipath"; | |
d62a17ae | 1567 | nexthop_num++; |
1568 | ||
1569 | _netlink_route_debug(cmd, p, nexthop, routedesc, | |
1570 | family, zvrf); | |
1571 | _netlink_route_build_multipath( | |
1572 | routedesc, bytelen, nexthop, rta, rtnh, | |
1573 | &req.r, &src1); | |
1574 | rtnh = RTNH_NEXT(rtnh); | |
1575 | ||
1576 | if (!setsrc && src1) { | |
1577 | if (family == AF_INET) | |
1578 | src.ipv4 = src1->ipv4; | |
1579 | else if (family == AF_INET6) | |
1580 | src.ipv6 = src1->ipv6; | |
1581 | ||
1582 | setsrc = 1; | |
1583 | } | |
1584 | } | |
1585 | } | |
1586 | if (setsrc && (cmd == RTM_NEWROUTE)) { | |
1587 | if (family == AF_INET) | |
1588 | addattr_l(&req.n, sizeof req, RTA_PREFSRC, | |
1589 | &src.ipv4, bytelen); | |
1590 | else if (family == AF_INET6) | |
1591 | addattr_l(&req.n, sizeof req, RTA_PREFSRC, | |
1592 | &src.ipv6, bytelen); | |
1593 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1594 | zlog_debug("Setting source"); | |
1595 | } | |
1596 | ||
1597 | if (rta->rta_len > RTA_LENGTH(0)) | |
1598 | addattr_l(&req.n, NL_PKT_BUF_SIZE, RTA_MULTIPATH, | |
1599 | RTA_DATA(rta), RTA_PAYLOAD(rta)); | |
1600 | } | |
718e3744 | 1601 | |
d62a17ae | 1602 | /* If there is no useful nexthop then return. */ |
1603 | if (nexthop_num == 0) { | |
1604 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1605 | zlog_debug( | |
1606 | "netlink_route_multipath(): No useful nexthop."); | |
1607 | return 0; | |
1608 | } | |
718e3744 | 1609 | |
7021c425 | 1610 | skip: |
718e3744 | 1611 | |
d62a17ae | 1612 | /* Destination netlink address. */ |
1613 | memset(&snl, 0, sizeof snl); | |
1614 | snl.nl_family = AF_NETLINK; | |
718e3744 | 1615 | |
d62a17ae | 1616 | /* Talk to netlink socket. */ |
1617 | return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns, | |
1618 | 0); | |
718e3744 | 1619 | } |
1620 | ||
43b5cc5e | 1621 | int kernel_get_ipmr_sg_stats(struct zebra_vrf *zvrf, void *in) |
e3be0432 | 1622 | { |
d62a17ae | 1623 | int suc = 0; |
1624 | struct mcast_route_data *mr = (struct mcast_route_data *)in; | |
bd8b9272 DS |
1625 | struct { |
1626 | struct nlmsghdr n; | |
1627 | struct ndmsg ndm; | |
1628 | char buf[256]; | |
1629 | } req; | |
e3be0432 | 1630 | |
d62a17ae | 1631 | mroute = mr; |
bd8b9272 DS |
1632 | struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT); |
1633 | ||
1634 | memset(&req.n, 0, sizeof(req.n)); | |
1635 | memset(&req.ndm, 0, sizeof(req.ndm)); | |
1636 | ||
1637 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
1638 | req.n.nlmsg_flags = NLM_F_REQUEST; | |
1639 | req.n.nlmsg_pid = zns->netlink_cmd.snl.nl_pid; | |
1640 | ||
1641 | req.ndm.ndm_family = RTNL_FAMILY_IPMR; | |
1642 | req.n.nlmsg_type = RTM_GETROUTE; | |
1643 | ||
1644 | addattr_l(&req.n, sizeof(req), RTA_IIF, &mroute->ifindex, 4); | |
1645 | addattr_l(&req.n, sizeof(req), RTA_OIF, &mroute->ifindex, 4); | |
1646 | addattr_l(&req.n, sizeof(req), RTA_SRC, &mroute->sg.src.s_addr, 4); | |
1647 | addattr_l(&req.n, sizeof(req), RTA_DST, &mroute->sg.grp.s_addr, 4); | |
1648 | addattr_l(&req.n, sizeof(req), RTA_TABLE, &zvrf->table_id, 4); | |
e3be0432 | 1649 | |
bd8b9272 DS |
1650 | suc = netlink_talk(netlink_route_change_read_multicast, &req.n, |
1651 | &zns->netlink_cmd, zns, 0); | |
e3be0432 | 1652 | |
bd8b9272 | 1653 | mroute = NULL; |
d62a17ae | 1654 | return suc; |
e3be0432 DS |
1655 | } |
1656 | ||
7d974ba3 DS |
1657 | void kernel_route_rib(struct route_node *rn, struct prefix *p, |
1658 | struct prefix *src_p, struct route_entry *old, | |
1659 | struct route_entry *new) | |
718e3744 | 1660 | { |
0c555cc6 DS |
1661 | int ret = 0; |
1662 | ||
0af35d90 RW |
1663 | assert(old || new); |
1664 | ||
0c555cc6 DS |
1665 | if (new) { |
1666 | if (p->family == AF_INET) | |
1667 | ret = netlink_route_multipath(RTM_NEWROUTE, p, src_p, | |
1668 | new, (old) ? 1 : 0); | |
1669 | else { | |
1670 | /* | |
1671 | * So v6 route replace semantics are not in | |
1672 | * the kernel at this point as I understand it. | |
1673 | * So let's do a delete than an add. | |
1674 | * In the future once v6 route replace semantics | |
1675 | * are in we can figure out what to do here to | |
1676 | * allow working with old and new kernels. | |
1677 | * | |
1678 | * I'm also intentionally ignoring the failure case | |
1679 | * of the route delete. If that happens yeah we're | |
1680 | * screwed. | |
1681 | */ | |
1682 | if (old) | |
1683 | netlink_route_multipath(RTM_DELROUTE, p, | |
1684 | src_p, old, 0); | |
1685 | ret = netlink_route_multipath(RTM_NEWROUTE, p, | |
1686 | src_p, new, 0); | |
1687 | } | |
7d974ba3 | 1688 | kernel_route_rib_pass_fail(rn, p, new, |
0c555cc6 DS |
1689 | (!ret) ? |
1690 | SOUTHBOUND_INSTALL_SUCCESS : | |
1691 | SOUTHBOUND_INSTALL_FAILURE); | |
1692 | return; | |
1693 | } | |
718e3744 | 1694 | |
0c555cc6 DS |
1695 | if (old) { |
1696 | ret = netlink_route_multipath(RTM_DELROUTE, p, src_p, old, 0); | |
e7fcb843 | 1697 | |
7d974ba3 | 1698 | kernel_route_rib_pass_fail(rn, p, old, |
0c555cc6 DS |
1699 | (!ret) ? |
1700 | SOUTHBOUND_DELETE_SUCCESS : | |
1701 | SOUTHBOUND_DELETE_FAILURE); | |
1702 | } | |
718e3744 | 1703 | } |
1704 | ||
d62a17ae | 1705 | int kernel_neigh_update(int add, int ifindex, uint32_t addr, char *lla, |
1706 | int llalen) | |
6b8a5694 | 1707 | { |
d62a17ae | 1708 | return netlink_neigh_update(add ? RTM_NEWNEIGH : RTM_DELNEIGH, ifindex, |
1709 | addr, lla, llalen); | |
6b8a5694 | 1710 | } |
718e3744 | 1711 | |
13d60d35 | 1712 | /* |
1713 | * Add remote VTEP to the flood list for this VxLAN interface (VNI). This | |
1714 | * is done by adding an FDB entry with a MAC of 00:00:00:00:00:00. | |
1715 | */ | |
d62a17ae | 1716 | static int netlink_vxlan_flood_list_update(struct interface *ifp, |
1717 | struct in_addr *vtep_ip, int cmd) | |
13d60d35 | 1718 | { |
d62a17ae | 1719 | struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT); |
1720 | struct { | |
1721 | struct nlmsghdr n; | |
1722 | struct ndmsg ndm; | |
1723 | char buf[256]; | |
1724 | } req; | |
1725 | u_char dst_mac[6] = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0}; | |
1726 | ||
1727 | memset(&req.n, 0, sizeof(req.n)); | |
1728 | memset(&req.ndm, 0, sizeof(req.ndm)); | |
1729 | ||
1730 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
1731 | req.n.nlmsg_flags = NLM_F_REQUEST; | |
1732 | if (cmd == RTM_NEWNEIGH) | |
1733 | req.n.nlmsg_flags |= (NLM_F_CREATE | NLM_F_APPEND); | |
1734 | req.n.nlmsg_type = cmd; | |
1735 | req.ndm.ndm_family = PF_BRIDGE; | |
1736 | req.ndm.ndm_state = NUD_NOARP | NUD_PERMANENT; | |
1737 | req.ndm.ndm_flags |= NTF_SELF; // Handle by "self", not "master" | |
1738 | ||
1739 | ||
1740 | addattr_l(&req.n, sizeof(req), NDA_LLADDR, &dst_mac, 6); | |
1741 | req.ndm.ndm_ifindex = ifp->ifindex; | |
1742 | addattr_l(&req.n, sizeof(req), NDA_DST, &vtep_ip->s_addr, 4); | |
1743 | ||
1744 | return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns, | |
1745 | 0); | |
13d60d35 | 1746 | } |
1747 | ||
1748 | /* | |
d62a17ae | 1749 | * Add remote VTEP for this VxLAN interface (VNI). In Linux, this involves |
1750 | * adding | |
13d60d35 | 1751 | * a "flood" MAC FDB entry. |
1752 | */ | |
d62a17ae | 1753 | int kernel_add_vtep(vni_t vni, struct interface *ifp, struct in_addr *vtep_ip) |
13d60d35 | 1754 | { |
d62a17ae | 1755 | if (IS_ZEBRA_DEBUG_VXLAN) |
1756 | zlog_debug("Install %s into flood list for VNI %u intf %s(%u)", | |
1757 | inet_ntoa(*vtep_ip), vni, ifp->name, ifp->ifindex); | |
13d60d35 | 1758 | |
d62a17ae | 1759 | return netlink_vxlan_flood_list_update(ifp, vtep_ip, RTM_NEWNEIGH); |
13d60d35 | 1760 | } |
1761 | ||
1762 | /* | |
1763 | * Remove remote VTEP for this VxLAN interface (VNI). In Linux, this involves | |
1764 | * deleting the "flood" MAC FDB entry. | |
1765 | */ | |
d62a17ae | 1766 | int kernel_del_vtep(vni_t vni, struct interface *ifp, struct in_addr *vtep_ip) |
13d60d35 | 1767 | { |
d62a17ae | 1768 | if (IS_ZEBRA_DEBUG_VXLAN) |
1769 | zlog_debug( | |
1770 | "Uninstall %s from flood list for VNI %u intf %s(%u)", | |
1771 | inet_ntoa(*vtep_ip), vni, ifp->name, ifp->ifindex); | |
13d60d35 | 1772 | |
d62a17ae | 1773 | return netlink_vxlan_flood_list_update(ifp, vtep_ip, RTM_DELNEIGH); |
13d60d35 | 1774 | } |
1775 | ||
2232a77c | 1776 | #ifndef NDA_RTA |
d62a17ae | 1777 | #define NDA_RTA(r) \ |
1778 | ((struct rtattr *)(((char *)(r)) + NLMSG_ALIGN(sizeof(struct ndmsg)))) | |
2232a77c | 1779 | #endif |
1780 | ||
d62a17ae | 1781 | static int netlink_macfdb_change(struct sockaddr_nl *snl, struct nlmsghdr *h, |
1782 | int len) | |
2232a77c | 1783 | { |
d62a17ae | 1784 | struct ndmsg *ndm; |
1785 | struct interface *ifp; | |
1786 | struct zebra_if *zif; | |
d62a17ae | 1787 | struct rtattr *tb[NDA_MAX + 1]; |
1788 | struct interface *br_if; | |
1789 | struct ethaddr mac; | |
1790 | vlanid_t vid = 0; | |
1791 | struct prefix vtep_ip; | |
1792 | int vid_present = 0, dst_present = 0; | |
1793 | char buf[ETHER_ADDR_STRLEN]; | |
1794 | char vid_buf[20]; | |
1795 | char dst_buf[30]; | |
1796 | u_char sticky = 0; | |
1797 | ||
1798 | ndm = NLMSG_DATA(h); | |
1799 | ||
2853fed6 | 1800 | /* We only process macfdb notifications if EVPN is enabled */ |
1801 | if (!is_evpn_enabled()) | |
1802 | return 0; | |
1803 | ||
d62a17ae | 1804 | /* The interface should exist. */ |
1805 | ifp = if_lookup_by_index_per_ns(zebra_ns_lookup(NS_DEFAULT), | |
1806 | ndm->ndm_ifindex); | |
2853fed6 | 1807 | if (!ifp || !ifp->info) |
d62a17ae | 1808 | return 0; |
1809 | ||
1810 | /* The interface should be something we're interested in. */ | |
1811 | if (!IS_ZEBRA_IF_BRIDGE_SLAVE(ifp)) | |
1812 | return 0; | |
1813 | ||
1814 | /* Drop "permanent" entries. */ | |
1815 | if (ndm->ndm_state & NUD_PERMANENT) | |
1816 | return 0; | |
1817 | ||
1818 | zif = (struct zebra_if *)ifp->info; | |
1819 | if ((br_if = zif->brslave_info.br_if) == NULL) { | |
1820 | zlog_warn("%s family %s IF %s(%u) brIF %u - no bridge master", | |
1821 | nl_msg_type_to_str(h->nlmsg_type), | |
1822 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
1823 | ndm->ndm_ifindex, zif->brslave_info.bridge_ifindex); | |
1824 | return 0; | |
1825 | } | |
1826 | ||
1827 | /* Parse attributes and extract fields of interest. */ | |
1828 | memset(tb, 0, sizeof tb); | |
1829 | netlink_parse_rtattr(tb, NDA_MAX, NDA_RTA(ndm), len); | |
1830 | ||
1831 | if (!tb[NDA_LLADDR]) { | |
1832 | zlog_warn("%s family %s IF %s(%u) brIF %u - no LLADDR", | |
1833 | nl_msg_type_to_str(h->nlmsg_type), | |
1834 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
1835 | ndm->ndm_ifindex, zif->brslave_info.bridge_ifindex); | |
1836 | return 0; | |
1837 | } | |
1838 | ||
ff8b7eb8 | 1839 | if (RTA_PAYLOAD(tb[NDA_LLADDR]) != ETH_ALEN) { |
d62a17ae | 1840 | zlog_warn( |
df0b13cf | 1841 | "%s family %s IF %s(%u) brIF %u - LLADDR is not MAC, len %lu", |
d62a17ae | 1842 | nl_msg_type_to_str(h->nlmsg_type), |
1843 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
1844 | ndm->ndm_ifindex, zif->brslave_info.bridge_ifindex, | |
df0b13cf | 1845 | (unsigned long)RTA_PAYLOAD(tb[NDA_LLADDR])); |
d62a17ae | 1846 | return 0; |
1847 | } | |
1848 | ||
ff8b7eb8 | 1849 | memcpy(&mac, RTA_DATA(tb[NDA_LLADDR]), ETH_ALEN); |
d62a17ae | 1850 | |
1851 | if ((NDA_VLAN <= NDA_MAX) && tb[NDA_VLAN]) { | |
1852 | vid_present = 1; | |
1853 | vid = *(u_int16_t *)RTA_DATA(tb[NDA_VLAN]); | |
1854 | sprintf(vid_buf, " VLAN %u", vid); | |
1855 | } | |
1856 | ||
1857 | if (tb[NDA_DST]) { | |
1858 | /* TODO: Only IPv4 supported now. */ | |
1859 | dst_present = 1; | |
1860 | vtep_ip.family = AF_INET; | |
1861 | vtep_ip.prefixlen = IPV4_MAX_BITLEN; | |
1862 | memcpy(&(vtep_ip.u.prefix4.s_addr), RTA_DATA(tb[NDA_DST]), | |
1863 | IPV4_MAX_BYTELEN); | |
1864 | sprintf(dst_buf, " dst %s", inet_ntoa(vtep_ip.u.prefix4)); | |
1865 | } | |
1866 | ||
1867 | sticky = (ndm->ndm_state & NUD_NOARP) ? 1 : 0; | |
1868 | ||
1869 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1870 | zlog_debug("Rx %s family %s IF %s(%u)%s %sMAC %s%s", | |
1871 | nl_msg_type_to_str(h->nlmsg_type), | |
1872 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
1873 | ndm->ndm_ifindex, vid_present ? vid_buf : "", | |
1874 | sticky ? "sticky " : "", | |
1875 | prefix_mac2str(&mac, buf, sizeof(buf)), | |
1876 | dst_present ? dst_buf : ""); | |
1877 | ||
1878 | if (filter_vlan && vid != filter_vlan) | |
1879 | return 0; | |
1880 | ||
1881 | /* If add or update, do accordingly if learnt on a "local" interface; if | |
1882 | * the notification is over VxLAN, this has to be related to | |
1883 | * multi-homing, | |
1884 | * so perform an implicit delete of any local entry (if it exists). | |
1885 | */ | |
1886 | if (h->nlmsg_type == RTM_NEWNEIGH) { | |
1887 | /* Drop "permanent" entries. */ | |
1888 | if (ndm->ndm_state & NUD_PERMANENT) | |
1889 | return 0; | |
1890 | ||
1891 | if (IS_ZEBRA_IF_VXLAN(ifp)) | |
1892 | return zebra_vxlan_check_del_local_mac(ifp, br_if, &mac, | |
1893 | vid); | |
1894 | ||
1895 | return zebra_vxlan_local_mac_add_update(ifp, br_if, &mac, vid, | |
1896 | sticky); | |
1897 | } | |
1898 | ||
1899 | /* This is a delete notification. | |
1900 | * 1. For a MAC over VxLan, check if it needs to be refreshed(readded) | |
1901 | * 2. For a MAC over "local" interface, delete the mac | |
1902 | * Note: We will get notifications from both bridge driver and VxLAN | |
1903 | * driver. | |
1904 | * Ignore the notification from VxLan driver as it is also generated | |
1905 | * when mac moves from remote to local. | |
1906 | */ | |
1907 | if (dst_present) | |
1908 | return 0; | |
1909 | ||
1910 | if (IS_ZEBRA_IF_VXLAN(ifp)) | |
1911 | return zebra_vxlan_check_readd_remote_mac(ifp, br_if, &mac, | |
1912 | vid); | |
1913 | ||
1914 | return zebra_vxlan_local_mac_del(ifp, br_if, &mac, vid); | |
2232a77c | 1915 | } |
1916 | ||
d62a17ae | 1917 | static int netlink_macfdb_table(struct sockaddr_nl *snl, struct nlmsghdr *h, |
1918 | ns_id_t ns_id, int startup) | |
2232a77c | 1919 | { |
d62a17ae | 1920 | int len; |
1921 | struct ndmsg *ndm; | |
2232a77c | 1922 | |
d62a17ae | 1923 | if (h->nlmsg_type != RTM_NEWNEIGH) |
1924 | return 0; | |
2232a77c | 1925 | |
d62a17ae | 1926 | /* Length validity. */ |
1927 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct ndmsg)); | |
1928 | if (len < 0) | |
1929 | return -1; | |
2232a77c | 1930 | |
d62a17ae | 1931 | /* We are interested only in AF_BRIDGE notifications. */ |
1932 | ndm = NLMSG_DATA(h); | |
1933 | if (ndm->ndm_family != AF_BRIDGE) | |
1934 | return 0; | |
2232a77c | 1935 | |
d62a17ae | 1936 | return netlink_macfdb_change(snl, h, len); |
2232a77c | 1937 | } |
1938 | ||
1939 | /* Request for MAC FDB information from the kernel */ | |
d62a17ae | 1940 | static int netlink_request_macs(struct zebra_ns *zns, int family, int type, |
1941 | ifindex_t master_ifindex) | |
2232a77c | 1942 | { |
d62a17ae | 1943 | struct { |
1944 | struct nlmsghdr n; | |
1945 | struct ifinfomsg ifm; | |
1946 | char buf[256]; | |
1947 | } req; | |
1948 | ||
1949 | /* Form the request, specifying filter (rtattr) if needed. */ | |
1950 | memset(&req, 0, sizeof(req)); | |
1951 | req.n.nlmsg_type = type; | |
1952 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ifinfomsg)); | |
1953 | req.ifm.ifi_family = family; | |
1954 | if (master_ifindex) | |
1955 | addattr32(&req.n, sizeof(req), IFLA_MASTER, master_ifindex); | |
1956 | ||
1957 | return netlink_request(&zns->netlink_cmd, &req.n); | |
2232a77c | 1958 | } |
1959 | ||
1960 | /* | |
1961 | * MAC forwarding database read using netlink interface. This is invoked | |
1962 | * at startup. | |
1963 | */ | |
d62a17ae | 1964 | int netlink_macfdb_read(struct zebra_ns *zns) |
2232a77c | 1965 | { |
d62a17ae | 1966 | int ret; |
1967 | ||
1968 | /* Get bridge FDB table. */ | |
1969 | ret = netlink_request_macs(zns, AF_BRIDGE, RTM_GETNEIGH, 0); | |
1970 | if (ret < 0) | |
1971 | return ret; | |
1972 | /* We are reading entire table. */ | |
1973 | filter_vlan = 0; | |
1974 | ret = netlink_parse_info(netlink_macfdb_table, &zns->netlink_cmd, zns, | |
1975 | 0, 1); | |
1976 | ||
1977 | return ret; | |
2232a77c | 1978 | } |
1979 | ||
1980 | /* | |
1981 | * MAC forwarding database read using netlink interface. This is for a | |
1982 | * specific bridge and matching specific access VLAN (if VLAN-aware bridge). | |
1983 | */ | |
d62a17ae | 1984 | int netlink_macfdb_read_for_bridge(struct zebra_ns *zns, struct interface *ifp, |
1985 | struct interface *br_if) | |
2232a77c | 1986 | { |
d62a17ae | 1987 | struct zebra_if *br_zif; |
1988 | struct zebra_if *zif; | |
1989 | struct zebra_l2info_vxlan *vxl; | |
1990 | int ret = 0; | |
1991 | ||
1992 | ||
1993 | /* Save VLAN we're filtering on, if needed. */ | |
1994 | br_zif = (struct zebra_if *)br_if->info; | |
1995 | zif = (struct zebra_if *)ifp->info; | |
1996 | vxl = &zif->l2info.vxl; | |
1997 | if (IS_ZEBRA_IF_BRIDGE_VLAN_AWARE(br_zif)) | |
1998 | filter_vlan = vxl->access_vlan; | |
1999 | ||
2000 | /* Get bridge FDB table for specific bridge - we do the VLAN filtering. | |
2001 | */ | |
2002 | ret = netlink_request_macs(zns, AF_BRIDGE, RTM_GETNEIGH, | |
2003 | br_if->ifindex); | |
2004 | if (ret < 0) | |
2005 | return ret; | |
2006 | ret = netlink_parse_info(netlink_macfdb_table, &zns->netlink_cmd, zns, | |
2007 | 0, 0); | |
2008 | ||
2009 | /* Reset VLAN filter. */ | |
2010 | filter_vlan = 0; | |
2011 | return ret; | |
2232a77c | 2012 | } |
2013 | ||
d62a17ae | 2014 | static int netlink_macfdb_update(struct interface *ifp, vlanid_t vid, |
2015 | struct ethaddr *mac, struct in_addr vtep_ip, | |
2016 | int local, int cmd, u_char sticky) | |
2232a77c | 2017 | { |
d62a17ae | 2018 | struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT); |
2019 | struct { | |
2020 | struct nlmsghdr n; | |
2021 | struct ndmsg ndm; | |
2022 | char buf[256]; | |
2023 | } req; | |
2024 | int dst_alen; | |
2025 | struct zebra_if *zif; | |
2026 | struct interface *br_if; | |
2027 | struct zebra_if *br_zif; | |
2028 | char buf[ETHER_ADDR_STRLEN]; | |
2029 | int vid_present = 0, dst_present = 0; | |
2030 | char vid_buf[20]; | |
2031 | char dst_buf[30]; | |
2032 | ||
2033 | zif = ifp->info; | |
2034 | if ((br_if = zif->brslave_info.br_if) == NULL) { | |
2035 | zlog_warn("MAC %s on IF %s(%u) - no mapping to bridge", | |
2036 | (cmd == RTM_NEWNEIGH) ? "add" : "del", ifp->name, | |
2037 | ifp->ifindex); | |
2038 | return -1; | |
2039 | } | |
2040 | ||
2041 | memset(&req.n, 0, sizeof(req.n)); | |
2042 | memset(&req.ndm, 0, sizeof(req.ndm)); | |
2043 | ||
2044 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
2045 | req.n.nlmsg_flags = NLM_F_REQUEST; | |
2046 | if (cmd == RTM_NEWNEIGH) | |
2047 | req.n.nlmsg_flags |= (NLM_F_CREATE | NLM_F_REPLACE); | |
2048 | req.n.nlmsg_type = cmd; | |
2049 | req.ndm.ndm_family = AF_BRIDGE; | |
2050 | req.ndm.ndm_flags |= NTF_SELF | NTF_MASTER; | |
2051 | req.ndm.ndm_state = NUD_REACHABLE; | |
2052 | ||
2053 | if (sticky) | |
2054 | req.ndm.ndm_state |= NUD_NOARP; | |
2055 | else | |
2056 | req.ndm.ndm_flags |= NTF_EXT_LEARNED; | |
2057 | ||
2058 | addattr_l(&req.n, sizeof(req), NDA_LLADDR, mac, 6); | |
2059 | req.ndm.ndm_ifindex = ifp->ifindex; | |
2060 | if (!local) { | |
2061 | dst_alen = 4; // TODO: hardcoded | |
2062 | addattr_l(&req.n, sizeof(req), NDA_DST, &vtep_ip, dst_alen); | |
2063 | dst_present = 1; | |
2064 | sprintf(dst_buf, " dst %s", inet_ntoa(vtep_ip)); | |
2065 | } | |
2066 | br_zif = (struct zebra_if *)br_if->info; | |
2067 | if (IS_ZEBRA_IF_BRIDGE_VLAN_AWARE(br_zif) && vid > 0) { | |
2068 | addattr16(&req.n, sizeof(req), NDA_VLAN, vid); | |
2069 | vid_present = 1; | |
2070 | sprintf(vid_buf, " VLAN %u", vid); | |
2071 | } | |
2072 | addattr32(&req.n, sizeof(req), NDA_MASTER, br_if->ifindex); | |
2073 | ||
2074 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2075 | zlog_debug("Tx %s family %s IF %s(%u)%s %sMAC %s%s", | |
2076 | nl_msg_type_to_str(cmd), | |
2077 | nl_family_to_str(req.ndm.ndm_family), ifp->name, | |
2078 | ifp->ifindex, vid_present ? vid_buf : "", | |
2079 | sticky ? "sticky " : "", | |
2080 | prefix_mac2str(mac, buf, sizeof(buf)), | |
2081 | dst_present ? dst_buf : ""); | |
2082 | ||
2083 | return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns, | |
2084 | 0); | |
2232a77c | 2085 | } |
2086 | ||
d62a17ae | 2087 | #define NUD_VALID \ |
2088 | (NUD_PERMANENT | NUD_NOARP | NUD_REACHABLE | NUD_PROBE | NUD_STALE \ | |
2089 | | NUD_DELAY) | |
2232a77c | 2090 | |
d62a17ae | 2091 | static int netlink_ipneigh_change(struct sockaddr_nl *snl, struct nlmsghdr *h, |
2092 | int len) | |
2232a77c | 2093 | { |
d62a17ae | 2094 | struct ndmsg *ndm; |
2095 | struct interface *ifp; | |
2096 | struct zebra_if *zif; | |
d62a17ae | 2097 | struct rtattr *tb[NDA_MAX + 1]; |
2098 | struct interface *link_if; | |
2099 | struct ethaddr mac; | |
2100 | struct ipaddr ip; | |
2101 | char buf[ETHER_ADDR_STRLEN]; | |
2102 | char buf2[INET6_ADDRSTRLEN]; | |
2103 | int mac_present = 0; | |
2104 | u_char ext_learned; | |
2105 | ||
2106 | ndm = NLMSG_DATA(h); | |
2107 | ||
2853fed6 | 2108 | /* We only process neigh notifications if EVPN is enabled */ |
2109 | if (!is_evpn_enabled()) | |
2110 | return 0; | |
2111 | ||
d62a17ae | 2112 | /* The interface should exist. */ |
2113 | ifp = if_lookup_by_index_per_ns(zebra_ns_lookup(NS_DEFAULT), | |
2114 | ndm->ndm_ifindex); | |
2853fed6 | 2115 | if (!ifp || !ifp->info) |
d62a17ae | 2116 | return 0; |
2117 | ||
2118 | /* Drop "permanent" entries. */ | |
2119 | if (ndm->ndm_state & NUD_PERMANENT) | |
2120 | return 0; | |
2121 | ||
2122 | zif = (struct zebra_if *)ifp->info; | |
2123 | /* The neighbor is present on an SVI. From this, we locate the | |
2124 | * underlying | |
2125 | * bridge because we're only interested in neighbors on a VxLAN bridge. | |
2126 | * The bridge is located based on the nature of the SVI: | |
2127 | * (a) In the case of a VLAN-aware bridge, the SVI is a L3 VLAN | |
2128 | * interface | |
2129 | * and is linked to the bridge | |
2130 | * (b) In the case of a VLAN-unaware bridge, the SVI is the bridge | |
2131 | * inteface | |
2132 | * itself | |
2133 | */ | |
2134 | if (IS_ZEBRA_IF_VLAN(ifp)) { | |
71349e03 MK |
2135 | link_if = if_lookup_by_index_per_ns(zebra_ns_lookup(NS_DEFAULT), |
2136 | zif->link_ifindex); | |
d62a17ae | 2137 | if (!link_if) |
2138 | return 0; | |
2139 | } else if (IS_ZEBRA_IF_BRIDGE(ifp)) | |
2140 | link_if = ifp; | |
2141 | else | |
2142 | return 0; | |
2143 | ||
2144 | /* Parse attributes and extract fields of interest. */ | |
2145 | memset(tb, 0, sizeof tb); | |
2146 | netlink_parse_rtattr(tb, NDA_MAX, NDA_RTA(ndm), len); | |
2147 | ||
2148 | if (!tb[NDA_DST]) { | |
2149 | zlog_warn("%s family %s IF %s(%u) - no DST", | |
2150 | nl_msg_type_to_str(h->nlmsg_type), | |
2151 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
2152 | ndm->ndm_ifindex); | |
2153 | return 0; | |
2154 | } | |
2155 | memset(&mac, 0, sizeof(struct ethaddr)); | |
2156 | memset(&ip, 0, sizeof(struct ipaddr)); | |
2157 | ip.ipa_type = (ndm->ndm_family == AF_INET) ? IPADDR_V4 : IPADDR_V6; | |
2158 | memcpy(&ip.ip.addr, RTA_DATA(tb[NDA_DST]), RTA_PAYLOAD(tb[NDA_DST])); | |
2159 | ||
2160 | if (h->nlmsg_type == RTM_NEWNEIGH) { | |
2161 | if (tb[NDA_LLADDR]) { | |
ff8b7eb8 | 2162 | if (RTA_PAYLOAD(tb[NDA_LLADDR]) != ETH_ALEN) { |
d62a17ae | 2163 | zlog_warn( |
df0b13cf | 2164 | "%s family %s IF %s(%u) - LLADDR is not MAC, len %lu", |
d62a17ae | 2165 | nl_msg_type_to_str(h->nlmsg_type), |
2166 | nl_family_to_str(ndm->ndm_family), | |
2167 | ifp->name, ndm->ndm_ifindex, | |
df0b13cf | 2168 | (unsigned long)RTA_PAYLOAD(tb[NDA_LLADDR])); |
d62a17ae | 2169 | return 0; |
2170 | } | |
2171 | ||
2172 | mac_present = 1; | |
ff8b7eb8 | 2173 | memcpy(&mac, RTA_DATA(tb[NDA_LLADDR]), ETH_ALEN); |
d62a17ae | 2174 | } |
2175 | ||
2176 | ext_learned = (ndm->ndm_flags & NTF_EXT_LEARNED) ? 1 : 0; | |
2177 | ||
2178 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2179 | zlog_debug( | |
2180 | "Rx %s family %s IF %s(%u) IP %s MAC %s state 0x%x flags 0x%x", | |
2181 | nl_msg_type_to_str(h->nlmsg_type), | |
2182 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
2183 | ndm->ndm_ifindex, | |
2184 | ipaddr2str(&ip, buf2, sizeof(buf2)), | |
2185 | mac_present | |
2186 | ? prefix_mac2str(&mac, buf, sizeof(buf)) | |
2187 | : "", | |
2188 | ndm->ndm_state, ndm->ndm_flags); | |
2189 | ||
2190 | /* If the neighbor state is valid for use, process as an add or | |
2191 | * update | |
2192 | * else process as a delete. Note that the delete handling may | |
2193 | * result | |
2194 | * in re-adding the neighbor if it is a valid "remote" neighbor. | |
2195 | */ | |
2196 | if (ndm->ndm_state & NUD_VALID) | |
2197 | return zebra_vxlan_local_neigh_add_update( | |
2198 | ifp, link_if, &ip, &mac, ndm->ndm_state, | |
2199 | ext_learned); | |
2200 | ||
2201 | return zebra_vxlan_local_neigh_del(ifp, link_if, &ip); | |
2202 | } | |
2203 | ||
2204 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2205 | zlog_debug("Rx %s family %s IF %s(%u) IP %s", | |
2206 | nl_msg_type_to_str(h->nlmsg_type), | |
2207 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
2208 | ndm->ndm_ifindex, | |
2209 | ipaddr2str(&ip, buf2, sizeof(buf2))); | |
2210 | ||
2211 | /* Process the delete - it may result in re-adding the neighbor if it is | |
2212 | * a valid "remote" neighbor. | |
2213 | */ | |
2214 | return zebra_vxlan_local_neigh_del(ifp, link_if, &ip); | |
2232a77c | 2215 | } |
2216 | ||
d62a17ae | 2217 | static int netlink_neigh_table(struct sockaddr_nl *snl, struct nlmsghdr *h, |
2218 | ns_id_t ns_id, int startup) | |
2232a77c | 2219 | { |
d62a17ae | 2220 | int len; |
2221 | struct ndmsg *ndm; | |
2232a77c | 2222 | |
d62a17ae | 2223 | if (h->nlmsg_type != RTM_NEWNEIGH) |
2224 | return 0; | |
2232a77c | 2225 | |
d62a17ae | 2226 | /* Length validity. */ |
2227 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct ndmsg)); | |
2228 | if (len < 0) | |
2229 | return -1; | |
2232a77c | 2230 | |
d62a17ae | 2231 | /* We are interested only in AF_INET or AF_INET6 notifications. */ |
2232 | ndm = NLMSG_DATA(h); | |
2233 | if (ndm->ndm_family != AF_INET && ndm->ndm_family != AF_INET6) | |
2234 | return 0; | |
2232a77c | 2235 | |
d62a17ae | 2236 | return netlink_neigh_change(snl, h, len); |
2232a77c | 2237 | } |
2238 | ||
2239 | /* Request for IP neighbor information from the kernel */ | |
d62a17ae | 2240 | static int netlink_request_neigh(struct zebra_ns *zns, int family, int type, |
2241 | ifindex_t ifindex) | |
2232a77c | 2242 | { |
d62a17ae | 2243 | struct { |
2244 | struct nlmsghdr n; | |
2245 | struct ndmsg ndm; | |
2246 | char buf[256]; | |
2247 | } req; | |
2248 | ||
2249 | /* Form the request, specifying filter (rtattr) if needed. */ | |
2250 | memset(&req, 0, sizeof(req)); | |
2251 | req.n.nlmsg_type = type; | |
2252 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
2253 | req.ndm.ndm_family = family; | |
2254 | if (ifindex) | |
2255 | addattr32(&req.n, sizeof(req), NDA_IFINDEX, ifindex); | |
2256 | ||
2257 | return netlink_request(&zns->netlink_cmd, &req.n); | |
2232a77c | 2258 | } |
2259 | ||
2260 | /* | |
2261 | * IP Neighbor table read using netlink interface. This is invoked | |
2262 | * at startup. | |
2263 | */ | |
d62a17ae | 2264 | int netlink_neigh_read(struct zebra_ns *zns) |
2232a77c | 2265 | { |
d62a17ae | 2266 | int ret; |
2232a77c | 2267 | |
d62a17ae | 2268 | /* Get IP neighbor table. */ |
2269 | ret = netlink_request_neigh(zns, AF_UNSPEC, RTM_GETNEIGH, 0); | |
2270 | if (ret < 0) | |
2271 | return ret; | |
2272 | ret = netlink_parse_info(netlink_neigh_table, &zns->netlink_cmd, zns, 0, | |
2273 | 1); | |
2232a77c | 2274 | |
d62a17ae | 2275 | return ret; |
2232a77c | 2276 | } |
2277 | ||
2278 | /* | |
2279 | * IP Neighbor table read using netlink interface. This is for a specific | |
2280 | * VLAN device. | |
2281 | */ | |
d62a17ae | 2282 | int netlink_neigh_read_for_vlan(struct zebra_ns *zns, struct interface *vlan_if) |
2232a77c | 2283 | { |
d62a17ae | 2284 | int ret = 0; |
2232a77c | 2285 | |
d62a17ae | 2286 | ret = netlink_request_neigh(zns, AF_UNSPEC, RTM_GETNEIGH, |
2287 | vlan_if->ifindex); | |
2288 | if (ret < 0) | |
2289 | return ret; | |
2290 | ret = netlink_parse_info(netlink_neigh_table, &zns->netlink_cmd, zns, 0, | |
2291 | 0); | |
2232a77c | 2292 | |
d62a17ae | 2293 | return ret; |
2232a77c | 2294 | } |
2295 | ||
d62a17ae | 2296 | int netlink_neigh_change(struct sockaddr_nl *snl, struct nlmsghdr *h, |
2297 | ns_id_t ns_id) | |
2232a77c | 2298 | { |
d62a17ae | 2299 | int len; |
2300 | struct ndmsg *ndm; | |
2232a77c | 2301 | |
d62a17ae | 2302 | if (!(h->nlmsg_type == RTM_NEWNEIGH || h->nlmsg_type == RTM_DELNEIGH)) |
2303 | return 0; | |
2232a77c | 2304 | |
d62a17ae | 2305 | /* Length validity. */ |
2306 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct ndmsg)); | |
2307 | if (len < 0) | |
2308 | return -1; | |
2232a77c | 2309 | |
d62a17ae | 2310 | /* Is this a notification for the MAC FDB or IP neighbor table? */ |
2311 | ndm = NLMSG_DATA(h); | |
2312 | if (ndm->ndm_family == AF_BRIDGE) | |
2313 | return netlink_macfdb_change(snl, h, len); | |
2232a77c | 2314 | |
d62a17ae | 2315 | if (ndm->ndm_type != RTN_UNICAST) |
2316 | return 0; | |
2232a77c | 2317 | |
d62a17ae | 2318 | if (ndm->ndm_family == AF_INET || ndm->ndm_family == AF_INET6) |
2319 | return netlink_ipneigh_change(snl, h, len); | |
2232a77c | 2320 | |
d62a17ae | 2321 | return 0; |
2232a77c | 2322 | } |
2323 | ||
d62a17ae | 2324 | static int netlink_neigh_update2(struct interface *ifp, struct ipaddr *ip, |
2325 | struct ethaddr *mac, u_int32_t flags, int cmd) | |
2232a77c | 2326 | { |
d62a17ae | 2327 | struct { |
2328 | struct nlmsghdr n; | |
2329 | struct ndmsg ndm; | |
2330 | char buf[256]; | |
2331 | } req; | |
2332 | int ipa_len; | |
2333 | ||
2334 | struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT); | |
2335 | char buf[INET6_ADDRSTRLEN]; | |
2336 | char buf2[ETHER_ADDR_STRLEN]; | |
2337 | ||
2338 | memset(&req.n, 0, sizeof(req.n)); | |
2339 | memset(&req.ndm, 0, sizeof(req.ndm)); | |
2340 | ||
2341 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
2342 | req.n.nlmsg_flags = NLM_F_REQUEST; | |
2343 | if (cmd == RTM_NEWNEIGH) | |
2344 | req.n.nlmsg_flags |= (NLM_F_CREATE | NLM_F_REPLACE); | |
2345 | req.n.nlmsg_type = cmd; // RTM_NEWNEIGH or RTM_DELNEIGH | |
2346 | req.ndm.ndm_family = IS_IPADDR_V4(ip) ? AF_INET : AF_INET6; | |
2347 | req.ndm.ndm_state = flags; | |
2348 | req.ndm.ndm_ifindex = ifp->ifindex; | |
2349 | req.ndm.ndm_type = RTN_UNICAST; | |
2350 | req.ndm.ndm_flags = NTF_EXT_LEARNED; | |
2351 | ||
2352 | ||
2353 | ipa_len = IS_IPADDR_V4(ip) ? IPV4_MAX_BYTELEN : IPV6_MAX_BYTELEN; | |
2354 | addattr_l(&req.n, sizeof(req), NDA_DST, &ip->ip.addr, ipa_len); | |
2355 | if (mac) | |
2356 | addattr_l(&req.n, sizeof(req), NDA_LLADDR, mac, 6); | |
2357 | ||
2358 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2359 | zlog_debug("Tx %s family %s IF %s(%u) Neigh %s MAC %s", | |
2360 | nl_msg_type_to_str(cmd), | |
2361 | nl_family_to_str(req.ndm.ndm_family), ifp->name, | |
2362 | ifp->ifindex, ipaddr2str(ip, buf, sizeof(buf)), | |
2363 | mac ? prefix_mac2str(mac, buf2, sizeof(buf2)) | |
2364 | : "null"); | |
2365 | ||
2366 | return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns, | |
2367 | 0); | |
2232a77c | 2368 | } |
2369 | ||
d62a17ae | 2370 | int kernel_add_mac(struct interface *ifp, vlanid_t vid, struct ethaddr *mac, |
2371 | struct in_addr vtep_ip, u_char sticky) | |
2232a77c | 2372 | { |
d62a17ae | 2373 | return netlink_macfdb_update(ifp, vid, mac, vtep_ip, 0, RTM_NEWNEIGH, |
2374 | sticky); | |
2232a77c | 2375 | } |
2376 | ||
d62a17ae | 2377 | int kernel_del_mac(struct interface *ifp, vlanid_t vid, struct ethaddr *mac, |
2378 | struct in_addr vtep_ip, int local) | |
2232a77c | 2379 | { |
d62a17ae | 2380 | return netlink_macfdb_update(ifp, vid, mac, vtep_ip, local, |
2381 | RTM_DELNEIGH, 0); | |
2232a77c | 2382 | } |
2383 | ||
d62a17ae | 2384 | int kernel_add_neigh(struct interface *ifp, struct ipaddr *ip, |
2385 | struct ethaddr *mac) | |
2232a77c | 2386 | { |
d62a17ae | 2387 | return netlink_neigh_update2(ifp, ip, mac, NUD_REACHABLE, RTM_NEWNEIGH); |
2232a77c | 2388 | } |
2389 | ||
d62a17ae | 2390 | int kernel_del_neigh(struct interface *ifp, struct ipaddr *ip) |
2232a77c | 2391 | { |
d62a17ae | 2392 | return netlink_neigh_update2(ifp, ip, NULL, 0, RTM_DELNEIGH); |
2232a77c | 2393 | } |
2394 | ||
40c7bdb0 | 2395 | /* |
2396 | * MPLS label forwarding table change via netlink interface. | |
2397 | */ | |
d62a17ae | 2398 | int netlink_mpls_multipath(int cmd, zebra_lsp_t *lsp) |
40c7bdb0 | 2399 | { |
d62a17ae | 2400 | mpls_lse_t lse; |
2401 | zebra_nhlfe_t *nhlfe; | |
2402 | struct nexthop *nexthop = NULL; | |
2403 | unsigned int nexthop_num; | |
2404 | const char *routedesc; | |
2405 | struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT); | |
805444ce | 2406 | int route_type; |
d62a17ae | 2407 | |
2408 | struct { | |
2409 | struct nlmsghdr n; | |
2410 | struct rtmsg r; | |
2411 | char buf[NL_PKT_BUF_SIZE]; | |
2412 | } req; | |
2413 | ||
2414 | memset(&req, 0, sizeof req - NL_PKT_BUF_SIZE); | |
2415 | ||
d62a17ae | 2416 | /* |
2417 | * Count # nexthops so we can decide whether to use singlepath | |
2418 | * or multipath case. | |
2419 | */ | |
2420 | nexthop_num = 0; | |
2421 | for (nhlfe = lsp->nhlfe_list; nhlfe; nhlfe = nhlfe->next) { | |
2422 | nexthop = nhlfe->nexthop; | |
2423 | if (!nexthop) | |
2424 | continue; | |
2425 | if (cmd == RTM_NEWROUTE) { | |
2426 | /* Count all selected NHLFEs */ | |
2427 | if (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_SELECTED) | |
2428 | && CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ACTIVE)) | |
2429 | nexthop_num++; | |
2430 | } else /* DEL */ | |
2431 | { | |
2432 | /* Count all installed NHLFEs */ | |
2433 | if (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_INSTALLED) | |
2434 | && CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB)) | |
2435 | nexthop_num++; | |
2436 | } | |
2437 | } | |
2438 | ||
8dc8a4b6 | 2439 | if ((nexthop_num == 0) || (!lsp->best_nhlfe && (cmd != RTM_DELROUTE))) |
d62a17ae | 2440 | return 0; |
2441 | ||
2442 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg)); | |
2443 | req.n.nlmsg_flags = NLM_F_CREATE | NLM_F_REQUEST; | |
2444 | req.n.nlmsg_type = cmd; | |
2445 | req.n.nlmsg_pid = zns->netlink_cmd.snl.nl_pid; | |
2446 | ||
2447 | req.r.rtm_family = AF_MPLS; | |
2448 | req.r.rtm_table = RT_TABLE_MAIN; | |
2449 | req.r.rtm_dst_len = MPLS_LABEL_LEN_BITS; | |
d62a17ae | 2450 | req.r.rtm_scope = RT_SCOPE_UNIVERSE; |
2451 | req.r.rtm_type = RTN_UNICAST; | |
2452 | ||
8dc8a4b6 | 2453 | if (cmd == RTM_NEWROUTE) { |
d62a17ae | 2454 | /* We do a replace to handle update. */ |
2455 | req.n.nlmsg_flags |= NLM_F_REPLACE; | |
2456 | ||
8dc8a4b6 DS |
2457 | /* set the protocol value if installing */ |
2458 | route_type = re_type_from_lsp_type(lsp->best_nhlfe->type); | |
2459 | req.r.rtm_protocol = zebra2proto(route_type); | |
2460 | } | |
2461 | ||
d62a17ae | 2462 | /* Fill destination */ |
2463 | lse = mpls_lse_encode(lsp->ile.in_label, 0, 0, 1); | |
2464 | addattr_l(&req.n, sizeof req, RTA_DST, &lse, sizeof(mpls_lse_t)); | |
2465 | ||
2466 | /* Fill nexthops (paths) based on single-path or multipath. The paths | |
2467 | * chosen depend on the operation. | |
2468 | */ | |
2469 | if (nexthop_num == 1 || multipath_num == 1) { | |
8b1450b9 | 2470 | routedesc = "single-path"; |
d62a17ae | 2471 | _netlink_mpls_debug(cmd, lsp->ile.in_label, routedesc); |
2472 | ||
2473 | nexthop_num = 0; | |
2474 | for (nhlfe = lsp->nhlfe_list; nhlfe; nhlfe = nhlfe->next) { | |
2475 | nexthop = nhlfe->nexthop; | |
2476 | if (!nexthop) | |
2477 | continue; | |
2478 | ||
2479 | if ((cmd == RTM_NEWROUTE | |
2480 | && (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_SELECTED) | |
2481 | && CHECK_FLAG(nexthop->flags, | |
2482 | NEXTHOP_FLAG_ACTIVE))) | |
2483 | || (cmd == RTM_DELROUTE | |
2484 | && (CHECK_FLAG(nhlfe->flags, | |
2485 | NHLFE_FLAG_INSTALLED) | |
2486 | && CHECK_FLAG(nexthop->flags, | |
2487 | NEXTHOP_FLAG_FIB)))) { | |
2488 | /* Add the gateway */ | |
2489 | _netlink_mpls_build_singlepath(routedesc, nhlfe, | |
2490 | &req.n, &req.r, | |
2491 | sizeof req, cmd); | |
d62a17ae | 2492 | nexthop_num++; |
2493 | break; | |
2494 | } | |
2495 | } | |
2496 | } else /* Multipath case */ | |
2497 | { | |
2498 | char buf[NL_PKT_BUF_SIZE]; | |
2499 | struct rtattr *rta = (void *)buf; | |
2500 | struct rtnexthop *rtnh; | |
2501 | union g_addr *src1 = NULL; | |
2502 | ||
2503 | rta->rta_type = RTA_MULTIPATH; | |
2504 | rta->rta_len = RTA_LENGTH(0); | |
2505 | rtnh = RTA_DATA(rta); | |
2506 | ||
8b1450b9 | 2507 | routedesc = "multipath"; |
d62a17ae | 2508 | _netlink_mpls_debug(cmd, lsp->ile.in_label, routedesc); |
2509 | ||
2510 | nexthop_num = 0; | |
2511 | for (nhlfe = lsp->nhlfe_list; nhlfe; nhlfe = nhlfe->next) { | |
2512 | nexthop = nhlfe->nexthop; | |
2513 | if (!nexthop) | |
2514 | continue; | |
2515 | ||
2516 | if (nexthop_num >= multipath_num) | |
2517 | break; | |
2518 | ||
2519 | if ((cmd == RTM_NEWROUTE | |
2520 | && (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_SELECTED) | |
2521 | && CHECK_FLAG(nexthop->flags, | |
2522 | NEXTHOP_FLAG_ACTIVE))) | |
2523 | || (cmd == RTM_DELROUTE | |
2524 | && (CHECK_FLAG(nhlfe->flags, | |
2525 | NHLFE_FLAG_INSTALLED) | |
2526 | && CHECK_FLAG(nexthop->flags, | |
2527 | NEXTHOP_FLAG_FIB)))) { | |
2528 | nexthop_num++; | |
2529 | ||
2530 | /* Build the multipath */ | |
2531 | _netlink_mpls_build_multipath(routedesc, nhlfe, | |
2532 | rta, rtnh, &req.r, | |
2533 | &src1); | |
2534 | rtnh = RTNH_NEXT(rtnh); | |
d62a17ae | 2535 | } |
2536 | } | |
2537 | ||
2538 | /* Add the multipath */ | |
2539 | if (rta->rta_len > RTA_LENGTH(0)) | |
2540 | addattr_l(&req.n, NL_PKT_BUF_SIZE, RTA_MULTIPATH, | |
2541 | RTA_DATA(rta), RTA_PAYLOAD(rta)); | |
2542 | } | |
2543 | ||
2544 | /* Talk to netlink socket. */ | |
2545 | return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns, | |
2546 | 0); | |
40c7bdb0 | 2547 | } |
ddfeb486 | 2548 | #endif /* HAVE_NETLINK */ |