]> git.proxmox.com Git - mirror_frr.git/blame - zebra/rt_netlink.c
projects / mirror_frr.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
Merge pull request #1430 from chiragshah6/ospf_vrf_dev
[mirror_frr.git] / zebra / rt_netlink.c
CommitLineData
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 72static vlanid_t filter_vlan = 0;
73
d62a17ae 74struct gw_family_t {
75 u_int16_t filler;
76 u_int16_t family;
77 union g_addr gate;
40c7bdb0 78};
79
8755598a
DS		 80char ipv4_ll_buf[16] = "169.254.0.1";
81struct 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 89void rt_netlink_init(void)
8755598a 90{
d62a17ae 91 inet_pton(AF_INET, ipv4_ll_buf, &ipv4_ll);
8755598a
DS		 92}
93
23b1f334
DD		 94static 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
DS		 100 || (proto == RTPROT_LDP) || (proto == RTPROT_BABEL)
101 || (proto == RTPROT_RIP)) {
d62a17ae 102 return 1;
103 }
104
105 return 0;
23b1f334
DD		 106}
107
915902cb 108static 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;
142 default:
143 proto = RTPROT_ZEBRA;
144 break;
145 }
146
147 return proto;
23b1f334
DD		 148}
149
915902cb
DS		 150static inline int proto2zebra(int proto, int family)
151{
152 switch (proto) {
153 case RTPROT_BABEL:
154 proto = ZEBRA_ROUTE_BABEL;
155 break;
156 case RTPROT_BGP:
157 proto = ZEBRA_ROUTE_BGP;
158 break;
159 case RTPROT_OSPF:
160 proto = (family == AFI_IP) ?
161 ZEBRA_ROUTE_OSPF : ZEBRA_ROUTE_OSPF6;
162 break;
163 case RTPROT_ISIS:
164 proto = ZEBRA_ROUTE_ISIS;
165 break;
166 case RTPROT_RIP:
167 proto = ZEBRA_ROUTE_RIP;
168 break;
169 case RTPROT_RIPNG:
170 proto = ZEBRA_ROUTE_RIPNG;
171 break;
172 case RTPROT_NHRP:
173 proto = ZEBRA_ROUTE_NHRP;
174 break;
175 case RTPROT_EIGRP:
176 proto = ZEBRA_ROUTE_EIGRP;
177 break;
178 case RTPROT_LDP:
179 proto = ZEBRA_ROUTE_LDP;
180 break;
181 case RTPROT_STATIC:
182 proto = ZEBRA_ROUTE_STATIC;
183 break;
184 default:
185 proto = ZEBRA_ROUTE_KERNEL;
186 break;
187 }
188 return proto;
189}
190
12f6fb97
DS		 191/*
192Pending: create an efficient table_id (in a tree/hash) based lookup)
193 */
d62a17ae 194static vrf_id_t vrf_lookup_by_table(u_int32_t table_id)
12f6fb97 195{
d62a17ae 196 struct vrf *vrf;
197 struct zebra_vrf *zvrf;
12f6fb97 198
a2addae8 199 RB_FOREACH (vrf, vrf_id_head, &vrfs_by_id) {
d62a17ae 200 if ((zvrf = vrf->info) == NULL || (zvrf->table_id != table_id))
201 continue;
12f6fb97 202
d62a17ae 203 return zvrf_id(zvrf);
204 }
12f6fb97 205
d62a17ae 206 return VRF_DEFAULT;
12f6fb97
DS		 207}
208
718e3744 209/* Looking up routing table by netlink interface. */
d62a17ae 210static int netlink_route_change_read_unicast(struct sockaddr_nl *snl,
211 struct nlmsghdr *h, ns_id_t ns_id,
212 int startup)
718e3744 213{
d62a17ae 214 int len;
215 struct rtmsg *rtm;
216 struct rtattr *tb[RTA_MAX + 1];
217 u_char flags = 0;
218 struct prefix p;
792fa92e 219 struct prefix_ipv6 src_p = {};
d62a17ae 220 vrf_id_t vrf_id = VRF_DEFAULT;
221
222 char anyaddr[16] = {0};
223
915902cb 224 int proto = ZEBRA_ROUTE_KERNEL;
d62a17ae 225 int index = 0;
226 int table;
227 int metric = 0;
228 u_int32_t mtu = 0;
25715c7e 229 uint8_t distance = 0;
d62a17ae 230
231 void *dest = NULL;
232 void *gate = NULL;
233 void *prefsrc = NULL; /* IPv4 preferred source host address */
234 void *src = NULL; /* IPv6 srcdest source prefix */
e655a03c 235 enum blackhole_type bh_type = BLACKHOLE_UNSPEC;
d62a17ae 236
237 rtm = NLMSG_DATA(h);
238
239 if (startup && h->nlmsg_type != RTM_NEWROUTE)
240 return 0;
e655a03c
DL		 241 switch (rtm->rtm_type) {
242 case RTN_UNICAST:
243 break;
244 case RTN_BLACKHOLE:
245 bh_type = BLACKHOLE_NULL;
246 break;
247 case RTN_UNREACHABLE:
248 bh_type = BLACKHOLE_REJECT;
249 break;
250 case RTN_PROHIBIT:
251 bh_type = BLACKHOLE_ADMINPROHIB;
252 break;
253 default:
d62a17ae 254 return 0;
e655a03c 255 }
d62a17ae 256
257 len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct rtmsg));
258 if (len < 0)
259 return -1;
260
261 memset(tb, 0, sizeof tb);
262 netlink_parse_rtattr(tb, RTA_MAX, RTM_RTA(rtm), len);
263
264 if (rtm->rtm_flags & RTM_F_CLONED)
265 return 0;
266 if (rtm->rtm_protocol == RTPROT_REDIRECT)
267 return 0;
268 if (rtm->rtm_protocol == RTPROT_KERNEL)
269 return 0;
270
271 if (!startup && is_selfroute(rtm->rtm_protocol)
272 && h->nlmsg_type == RTM_NEWROUTE)
273 return 0;
274
275 /* We don't care about change notifications for the MPLS table. */
276 /* TODO: Revisit this. */
277 if (rtm->rtm_family == AF_MPLS)
278 return 0;
279
280 /* Table corresponding to route. */
281 if (tb[RTA_TABLE])
282 table = *(int *)RTA_DATA(tb[RTA_TABLE]);
283 else
284 table = rtm->rtm_table;
285
286 /* Map to VRF */
287 vrf_id = vrf_lookup_by_table(table);
288 if (vrf_id == VRF_DEFAULT) {
289 if (!is_zebra_valid_kernel_table(table)
290 && !is_zebra_main_routing_table(table))
291 return 0;
292 }
293
294 /* Route which inserted by Zebra. */
915902cb 295 if (is_selfroute(rtm->rtm_protocol)) {
d62a17ae 296 flags |= ZEBRA_FLAG_SELFROUTE;
915902cb
DS		 297 proto = proto2zebra(rtm->rtm_protocol, rtm->rtm_family);
298 }
d62a17ae 299 if (tb[RTA_OIF])
300 index = *(int *)RTA_DATA(tb[RTA_OIF]);
301
302 if (tb[RTA_DST])
303 dest = RTA_DATA(tb[RTA_DST]);
304 else
305 dest = anyaddr;
306
307 if (tb[RTA_SRC])
308 src = RTA_DATA(tb[RTA_SRC]);
309 else
310 src = anyaddr;
311
312 if (tb[RTA_PREFSRC])
313 prefsrc = RTA_DATA(tb[RTA_PREFSRC]);
314
315 if (tb[RTA_GATEWAY])
316 gate = RTA_DATA(tb[RTA_GATEWAY]);
317
f19435a8
DS		 318 if (tb[RTA_PRIORITY])
319 metric = *(int *)RTA_DATA(tb[RTA_PRIORITY]);
d62a17ae 320
f19435a8
DS		 321 if (tb[RTA_METRICS]) {
322 struct rtattr *mxrta[RTAX_MAX + 1];
d62a17ae 323
f19435a8
DS		 324 memset(mxrta, 0, sizeof mxrta);
325 netlink_parse_rtattr(mxrta, RTAX_MAX,
326 RTA_DATA(tb[RTA_METRICS]),
327 RTA_PAYLOAD(tb[RTA_METRICS]));
d62a17ae 328
f19435a8
DS		 329 if (mxrta[RTAX_MTU])
330 mtu = *(u_int32_t *)RTA_DATA(mxrta[RTAX_MTU]);
d62a17ae 331 }
332
333 if (rtm->rtm_family == AF_INET) {
334 p.family = AF_INET;
335 memcpy(&p.u.prefix4, dest, 4);
336 p.prefixlen = rtm->rtm_dst_len;
337
338 src_p.prefixlen =
339 0; // Forces debug below to not display anything
340 } else if (rtm->rtm_family == AF_INET6) {
341 p.family = AF_INET6;
342 memcpy(&p.u.prefix6, dest, 16);
343 p.prefixlen = rtm->rtm_dst_len;
344
345 src_p.family = AF_INET6;
346 memcpy(&src_p.prefix, src, 16);
347 src_p.prefixlen = rtm->rtm_src_len;
348 }
349
350 if (rtm->rtm_src_len != 0) {
351 char buf[PREFIX_STRLEN];
352 zlog_warn(
353 "unsupported IPv[4|6] sourcedest route (dest %s vrf %u)",
354 prefix2str(&p, buf, sizeof(buf)), vrf_id);
355 return 0;
356 }
357
25715c7e
DS		 358 /*
359 * For ZEBRA_ROUTE_KERNEL types:
360 *
361 * The metric/priority of the route received from the kernel
362 * is a 32 bit number. We are going to interpret the high
363 * order byte as the Admin Distance and the low order 3 bytes
364 * as the metric.
365 *
366 * This will allow us to do two things:
367 * 1) Allow the creation of kernel routes that can be
368 * overridden by zebra.
369 * 2) Allow the old behavior for 'most' kernel route types
370 * if a user enters 'ip route ...' v4 routes get a metric
371 * of 0 and v6 routes get a metric of 1024. Both of these
372 * values will end up with a admin distance of 0, which
373 * will cause them to win for the purposes of zebra.
374 */
375 if (proto == ZEBRA_ROUTE_KERNEL) {
376 distance = (metric >> 24) & 0xFF;
377 metric = (metric & 0x00FFFFFF);
378 }
379
d62a17ae 380 if (IS_ZEBRA_DEBUG_KERNEL) {
381 char buf[PREFIX_STRLEN];
382 char buf2[PREFIX_STRLEN];
383 zlog_debug(
25715c7e 384 "%s %s%s%s vrf %u metric: %d Admin Distance: %d", nl_msg_type_to_str(h->nlmsg_type),
d62a17ae 385 prefix2str(&p, buf, sizeof(buf)),
386 src_p.prefixlen ? " from " : "",
387 src_p.prefixlen ? prefix2str(&src_p, buf2, sizeof(buf2))
388 : "",
25715c7e 389 vrf_id, metric, distance);
d62a17ae 390 }
391
392 afi_t afi = AFI_IP;
393 if (rtm->rtm_family == AF_INET6)
394 afi = AFI_IP6;
395
396 if (h->nlmsg_type == RTM_NEWROUTE) {
fd36be7e
DL		 397 if (!tb[RTA_MULTIPATH]) {
398 struct nexthop nh;
399 size_t sz = (afi == AFI_IP) ? 4 : 16;
400
401 memset(&nh, 0, sizeof(nh));
af760ec1
DS		 402
403 if (bh_type == BLACKHOLE_UNSPEC) {
404 if (index && !gate)
405 nh.type = NEXTHOP_TYPE_IFINDEX;
406 else if (index && gate)
1dca2eaa 407 nh.type = (afi == AFI_IP)
af760ec1
DS		 408 ? NEXTHOP_TYPE_IPV4_IFINDEX
409 : NEXTHOP_TYPE_IPV6_IFINDEX;
410 else if (!index && gate)
1dca2eaa
RW		 411 nh.type = (afi == AFI_IP)
412 ? NEXTHOP_TYPE_IPV4
413 : NEXTHOP_TYPE_IPV6;
af760ec1
DS		 414 else {
415 nh.type = NEXTHOP_TYPE_BLACKHOLE;
416 nh.bh_type = bh_type;
417 }
418 } else {
fd36be7e 419 nh.type = NEXTHOP_TYPE_BLACKHOLE;
e655a03c
DL		 420 nh.bh_type = bh_type;
421 }
fd36be7e
DL		 422 nh.ifindex = index;
423 if (prefsrc)
424 memcpy(&nh.src, prefsrc, sz);
425 if (gate)
426 memcpy(&nh.gate, gate, sz);
915902cb
DS		 427
428 rib_add(afi, SAFI_UNICAST, vrf_id, proto,
25715c7e 429 0, flags, &p, NULL, &nh, table, metric, mtu, distance);
fd36be7e 430 } else {
d62a17ae 431 /* This is a multipath route */
432
433 struct route_entry *re;
434 struct rtnexthop *rtnh =
435 (struct rtnexthop *)RTA_DATA(tb[RTA_MULTIPATH]);
436
437 len = RTA_PAYLOAD(tb[RTA_MULTIPATH]);
438
439 re = XCALLOC(MTYPE_RE, sizeof(struct route_entry));
915902cb 440 re->type = proto;
25715c7e 441 re->distance = distance;
d62a17ae 442 re->flags = flags;
443 re->metric = metric;
444 re->mtu = mtu;
445 re->vrf_id = vrf_id;
446 re->table = table;
447 re->nexthop_num = 0;
448 re->uptime = time(NULL);
449
450 for (;;) {
451 if (len < (int)sizeof(*rtnh)
452 || rtnh->rtnh_len > len)
453 break;
454
455 index = rtnh->rtnh_ifindex;
456 gate = 0;
457 if (rtnh->rtnh_len > sizeof(*rtnh)) {
458 memset(tb, 0, sizeof(tb));
459 netlink_parse_rtattr(
460 tb, RTA_MAX, RTNH_DATA(rtnh),
461 rtnh->rtnh_len - sizeof(*rtnh));
462 if (tb[RTA_GATEWAY])
463 gate = RTA_DATA(
464 tb[RTA_GATEWAY]);
465 }
466
467 if (gate) {
468 if (rtm->rtm_family == AF_INET) {
469 if (index)
470 route_entry_nexthop_ipv4_ifindex_add(
471 re, gate,
472 prefsrc, index);
473 else
474 route_entry_nexthop_ipv4_add(
475 re, gate,
476 prefsrc);
477 } else if (rtm->rtm_family
478 == AF_INET6) {
479 if (index)
480 route_entry_nexthop_ipv6_ifindex_add(
481 re, gate,
482 index);
483 else
484 route_entry_nexthop_ipv6_add(
485 re, gate);
486 }
487 } else
488 route_entry_nexthop_ifindex_add(re,
489 index);
490
491 len -= NLMSG_ALIGN(rtnh->rtnh_len);
492 rtnh = RTNH_NEXT(rtnh);
493 }
494
495 zserv_nexthop_num_warn(__func__,
496 (const struct prefix *)&p,
497 re->nexthop_num);
498 if (re->nexthop_num == 0)
499 XFREE(MTYPE_RE, re);
500 else
7990990e 501 rib_add_multipath(afi, SAFI_UNICAST, &p,
d62a17ae 502 NULL, re);
503 }
504 } else {
fd36be7e
DL		 505 if (!tb[RTA_MULTIPATH]) {
506 struct nexthop nh;
507 size_t sz = (afi == AFI_IP) ? 4 : 16;
508
509 memset(&nh, 0, sizeof(nh));
8ba5bd58
RW		 510 if (bh_type == BLACKHOLE_UNSPEC) {
511 if (index && !gate)
512 nh.type = NEXTHOP_TYPE_IFINDEX;
513 else if (index && gate)
514 nh.type =
515 (afi == AFI_IP)
516 ? NEXTHOP_TYPE_IPV4_IFINDEX
517 : NEXTHOP_TYPE_IPV6_IFINDEX;
518 else if (!index && gate)
519 nh.type = (afi == AFI_IP)
520 ? NEXTHOP_TYPE_IPV4
60466a63 521 : NEXTHOP_TYPE_IPV6;
8ba5bd58
RW		 522 else {
523 nh.type = NEXTHOP_TYPE_BLACKHOLE;
524 nh.bh_type = BLACKHOLE_UNSPEC;
525 }
526 } else {
fd36be7e 527 nh.type = NEXTHOP_TYPE_BLACKHOLE;
8ba5bd58
RW		 528 nh.bh_type = bh_type;
529 }
fd36be7e
DL		 530 nh.ifindex = index;
531 if (gate)
532 memcpy(&nh.gate, gate, sz);
d62a17ae 533 rib_delete(afi, SAFI_UNICAST, vrf_id,
915902cb 534 proto, 0, flags, &p, NULL, &nh,
5dfeba19 535 table, metric, true);
fd36be7e
DL		 536 } else {
537 /* XXX: need to compare the entire list of nexthops
538 * here for NLM_F_APPEND stupidity */
539 rib_delete(afi, SAFI_UNICAST, vrf_id,
915902cb 540 proto, 0, flags, &p, NULL, NULL,
5dfeba19 541 table, metric, true);
d62a17ae 542 }
543 }
544
545 return 0;
718e3744 546}
547
e3be0432
DS		 548static struct mcast_route_data *mroute = NULL;
549
d62a17ae 550static int netlink_route_change_read_multicast(struct sockaddr_nl *snl,
551 struct nlmsghdr *h,
552 ns_id_t ns_id, int startup)
565fdc75 553{
d62a17ae 554 int len;
555 struct rtmsg *rtm;
556 struct rtattr *tb[RTA_MAX + 1];
557 struct mcast_route_data *m;
558 struct mcast_route_data mr;
559 int iif = 0;
560 int count;
561 int oif[256];
562 int oif_count = 0;
563 char sbuf[40];
564 char gbuf[40];
565 char oif_list[256] = "\0";
566 vrf_id_t vrf = ns_id;
43b5cc5e 567 int table;
d62a17ae 568
569 if (mroute)
570 m = mroute;
571 else {
572 memset(&mr, 0, sizeof(mr));
573 m = &mr;
574 }
575
576 rtm = NLMSG_DATA(h);
577
578 len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct rtmsg));
579
580 memset(tb, 0, sizeof tb);
581 netlink_parse_rtattr(tb, RTA_MAX, RTM_RTA(rtm), len);
90d82769 582
43b5cc5e
DS		 583 if (tb[RTA_TABLE])
584 table = *(int *)RTA_DATA(tb[RTA_TABLE]);
585 else
586 table = rtm->rtm_table;
587
588 vrf = vrf_lookup_by_table(table);
589
d62a17ae 590 if (tb[RTA_IIF])
591 iif = *(int *)RTA_DATA(tb[RTA_IIF]);
592
593 if (tb[RTA_SRC])
bd8b9272 594 m->sg.src = *(struct in_addr *)RTA_DATA(tb[RTA_SRC]);
d62a17ae 595
596 if (tb[RTA_DST])
bd8b9272 597 m->sg.grp = *(struct in_addr *)RTA_DATA(tb[RTA_DST]);
d62a17ae 598
599 if ((RTA_EXPIRES <= RTA_MAX) && tb[RTA_EXPIRES])
600 m->lastused = *(unsigned long long *)RTA_DATA(tb[RTA_EXPIRES]);
601
602 if (tb[RTA_MULTIPATH]) {
603 struct rtnexthop *rtnh =
604 (struct rtnexthop *)RTA_DATA(tb[RTA_MULTIPATH]);
605
606 len = RTA_PAYLOAD(tb[RTA_MULTIPATH]);
607 for (;;) {
608 if (len < (int)sizeof(*rtnh) || rtnh->rtnh_len > len)
609 break;
610
611 oif[oif_count] = rtnh->rtnh_ifindex;
612 oif_count++;
613
614 len -= NLMSG_ALIGN(rtnh->rtnh_len);
615 rtnh = RTNH_NEXT(rtnh);
616 }
617 }
618
619 if (IS_ZEBRA_DEBUG_KERNEL) {
620 struct interface *ifp;
0af35d90
RW		 621 strlcpy(sbuf, inet_ntoa(m->sg.src), sizeof(sbuf));
622 strlcpy(gbuf, inet_ntoa(m->sg.grp), sizeof(gbuf));
d62a17ae 623 for (count = 0; count < oif_count; count++) {
624 ifp = if_lookup_by_index(oif[count], vrf);
625 char temp[256];
626
627 sprintf(temp, "%s ", ifp->name);
628 strcat(oif_list, temp);
629 }
43b5cc5e 630 struct zebra_vrf *zvrf = zebra_vrf_lookup_by_id(vrf);
d62a17ae 631 ifp = if_lookup_by_index(iif, vrf);
43b5cc5e
DS		 632 zlog_debug(
633 "MCAST VRF: %s(%d) %s (%s,%s) IIF: %s OIF: %s jiffies: %lld",
634 zvrf->vrf->name, vrf, nl_msg_type_to_str(h->nlmsg_type),
635 sbuf, gbuf, ifp->name, oif_list, m->lastused);
90d82769 636 }
d62a17ae 637 return 0;
565fdc75
DS		 638}
639
d62a17ae 640int netlink_route_change(struct sockaddr_nl *snl, struct nlmsghdr *h,
641 ns_id_t ns_id, int startup)
565fdc75 642{
d62a17ae 643 int len;
644 vrf_id_t vrf_id = ns_id;
645 struct rtmsg *rtm;
646
647 rtm = NLMSG_DATA(h);
648
649 if (!(h->nlmsg_type == RTM_NEWROUTE || h->nlmsg_type == RTM_DELROUTE)) {
650 /* If this is not route add/delete message print warning. */
651 zlog_warn("Kernel message: %d vrf %u\n", h->nlmsg_type, vrf_id);
652 return 0;
653 }
654
655 /* Connected route. */
656 if (IS_ZEBRA_DEBUG_KERNEL)
657 zlog_debug("%s %s %s proto %s vrf %u",
658 nl_msg_type_to_str(h->nlmsg_type),
659 nl_family_to_str(rtm->rtm_family),
660 nl_rttype_to_str(rtm->rtm_type),
661 nl_rtproto_to_str(rtm->rtm_protocol), vrf_id);
662
663 /* We don't care about change notifications for the MPLS table. */
664 /* TODO: Revisit this. */
665 if (rtm->rtm_family == AF_MPLS)
666 return 0;
667
668 len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct rtmsg));
669 if (len < 0)
670 return -1;
671
e655a03c 672 if (rtm->rtm_type == RTN_MULTICAST)
d62a17ae 673 netlink_route_change_read_multicast(snl, h, ns_id, startup);
e655a03c
DL		 674 else
675 netlink_route_change_read_unicast(snl, h, ns_id, startup);
d62a17ae 676 return 0;
565fdc75
DS		 677}
678
289602d7 679/* Request for specific route information from the kernel */
d62a17ae 680static int netlink_request_route(struct zebra_ns *zns, int family, int type)
289602d7 681{
d62a17ae 682 struct {
683 struct nlmsghdr n;
684 struct rtmsg rtm;
685 } req;
686
687 /* Form the request, specifying filter (rtattr) if needed. */
688 memset(&req, 0, sizeof(req));
689 req.n.nlmsg_type = type;
690 req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg));
691 req.rtm.rtm_family = family;
692
693 return netlink_request(&zns->netlink_cmd, &req.n);
289602d7 694}
695
718e3744 696/* Routing table read function using netlink interface. Only called
697 bootstrap time. */
d62a17ae 698int netlink_route_read(struct zebra_ns *zns)
718e3744 699{
d62a17ae 700 int ret;
701
702 /* Get IPv4 routing table. */
703 ret = netlink_request_route(zns, AF_INET, RTM_GETROUTE);
704 if (ret < 0)
705 return ret;
706 ret = netlink_parse_info(netlink_route_change_read_unicast,
707 &zns->netlink_cmd, zns, 0, 1);
708 if (ret < 0)
709 return ret;
710
711 /* Get IPv6 routing table. */
712 ret = netlink_request_route(zns, AF_INET6, RTM_GETROUTE);
713 if (ret < 0)
714 return ret;
715 ret = netlink_parse_info(netlink_route_change_read_unicast,
716 &zns->netlink_cmd, zns, 0, 1);
717 if (ret < 0)
718 return ret;
719
720 return 0;
718e3744 721}
722
d62a17ae 723static void _netlink_route_nl_add_gateway_info(u_char route_family,
724 u_char gw_family,
725 struct nlmsghdr *nlmsg,
726 size_t req_size, int bytelen,
727 struct nexthop *nexthop)
40c7bdb0 728{
d62a17ae 729 if (route_family == AF_MPLS) {
730 struct gw_family_t gw_fam;
731
732 gw_fam.family = gw_family;
733 if (gw_family == AF_INET)
734 memcpy(&gw_fam.gate.ipv4, &nexthop->gate.ipv4, bytelen);
735 else
736 memcpy(&gw_fam.gate.ipv6, &nexthop->gate.ipv6, bytelen);
737 addattr_l(nlmsg, req_size, RTA_VIA, &gw_fam.family,
738 bytelen + 2);
739 } else {
740 if (gw_family == AF_INET)
741 addattr_l(nlmsg, req_size, RTA_GATEWAY,
742 &nexthop->gate.ipv4, bytelen);
743 else
744 addattr_l(nlmsg, req_size, RTA_GATEWAY,
745 &nexthop->gate.ipv6, bytelen);
746 }
40c7bdb0 747}
748
d62a17ae 749static void _netlink_route_rta_add_gateway_info(u_char route_family,
750 u_char gw_family,
751 struct rtattr *rta,
752 struct rtnexthop *rtnh,
753 size_t req_size, int bytelen,
754 struct nexthop *nexthop)
40c7bdb0 755{
d62a17ae 756 if (route_family == AF_MPLS) {
757 struct gw_family_t gw_fam;
758
759 gw_fam.family = gw_family;
760 if (gw_family == AF_INET)
761 memcpy(&gw_fam.gate.ipv4, &nexthop->gate.ipv4, bytelen);
762 else
763 memcpy(&gw_fam.gate.ipv6, &nexthop->gate.ipv6, bytelen);
764 rta_addattr_l(rta, req_size, RTA_VIA, &gw_fam.family,
765 bytelen + 2);
766 rtnh->rtnh_len += RTA_LENGTH(bytelen + 2);
767 } else {
768 if (gw_family == AF_INET)
769 rta_addattr_l(rta, req_size, RTA_GATEWAY,
770 &nexthop->gate.ipv4, bytelen);
771 else
772 rta_addattr_l(rta, req_size, RTA_GATEWAY,
773 &nexthop->gate.ipv6, bytelen);
774 rtnh->rtnh_len += sizeof(struct rtattr) + bytelen;
775 }
40c7bdb0 776}
777
fa713d9e
CF		 778/* This function takes a nexthop as argument and adds
779 * the appropriate netlink attributes to an existing
780 * netlink message.
781 *
782 * @param routedesc: Human readable description of route type
783 * (direct/recursive, single-/multipath)
784 * @param bytelen: Length of addresses in bytes.
785 * @param nexthop: Nexthop information
786 * @param nlmsg: nlmsghdr structure to fill in.
787 * @param req_size: The size allocated for the message.
788 */
d62a17ae 789static void _netlink_route_build_singlepath(const char *routedesc, int bytelen,
790 struct nexthop *nexthop,
791 struct nlmsghdr *nlmsg,
792 struct rtmsg *rtmsg,
793 size_t req_size, int cmd)
fa713d9e 794{
d62a17ae 795 struct nexthop_label *nh_label;
796 mpls_lse_t out_lse[MPLS_MAX_LABELS];
9a62e84b 797 char label_buf[256];
d62a17ae 798
799 /*
800 * label_buf is *only* currently used within debugging.
801 * As such when we assign it we are guarding it inside
802 * a debug test. If you want to change this make sure
803 * you fix this assumption
804 */
805 label_buf[0] = '\0';
806 /* outgoing label - either as NEWDST (in the case of LSR) or as ENCAP
807 * (in the case of LER)
808 */
809 nh_label = nexthop->nh_label;
810 if (rtmsg->rtm_family == AF_MPLS) {
811 assert(nh_label);
812 assert(nh_label->num_labels == 1);
813 }
814
815 if (nh_label && nh_label->num_labels) {
816 int i, num_labels = 0;
817 u_int32_t bos;
818 char label_buf1[20];
819
820 for (i = 0; i < nh_label->num_labels; i++) {
821 if (nh_label->label[i] != MPLS_IMP_NULL_LABEL) {
822 bos = ((i == (nh_label->num_labels - 1)) ? 1
823 : 0);
824 out_lse[i] = mpls_lse_encode(nh_label->label[i],
825 0, 0, bos);
826 if (IS_ZEBRA_DEBUG_KERNEL) {
827 if (!num_labels)
9a62e84b 828 sprintf(label_buf, "label %u",
d62a17ae 829 nh_label->label[i]);
830 else {
9a62e84b 831 sprintf(label_buf1, "/%u",
d62a17ae 832 nh_label->label[i]);
9a62e84b
RW		 833 strlcat(label_buf, label_buf1,
834 sizeof(label_buf));
d62a17ae 835 }
836 }
837 num_labels++;
838 }
839 }
840 if (num_labels) {
841 if (rtmsg->rtm_family == AF_MPLS)
842 addattr_l(nlmsg, req_size, RTA_NEWDST, &out_lse,
843 num_labels * sizeof(mpls_lse_t));
844 else {
845 struct rtattr *nest;
846 u_int16_t encap = LWTUNNEL_ENCAP_MPLS;
847
848 addattr_l(nlmsg, req_size, RTA_ENCAP_TYPE,
849 &encap, sizeof(u_int16_t));
850 nest = addattr_nest(nlmsg, req_size, RTA_ENCAP);
851 addattr_l(nlmsg, req_size, MPLS_IPTUNNEL_DST,
852 &out_lse,
853 num_labels * sizeof(mpls_lse_t));
854 addattr_nest_end(nlmsg, nest);
855 }
66d42727 856 }
0aabccc0 857 }
fa713d9e 858
d62a17ae 859 if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ONLINK))
860 rtmsg->rtm_flags |= RTNH_F_ONLINK;
861
862 if (rtmsg->rtm_family == AF_INET
863 && (nexthop->type == NEXTHOP_TYPE_IPV6
864 || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX)) {
865 rtmsg->rtm_flags |= RTNH_F_ONLINK;
866 addattr_l(nlmsg, req_size, RTA_GATEWAY, &ipv4_ll, 4);
867 addattr32(nlmsg, req_size, RTA_OIF, nexthop->ifindex);
868
869 if (nexthop->rmap_src.ipv4.s_addr && (cmd == RTM_NEWROUTE))
870 addattr_l(nlmsg, req_size, RTA_PREFSRC,
871 &nexthop->rmap_src.ipv4, bytelen);
872 else if (nexthop->src.ipv4.s_addr && (cmd == RTM_NEWROUTE))
873 addattr_l(nlmsg, req_size, RTA_PREFSRC,
874 &nexthop->src.ipv4, bytelen);
875
876 if (IS_ZEBRA_DEBUG_KERNEL)
877 zlog_debug(
878 " 5549: _netlink_route_build_singlepath() (%s): "
879 "nexthop via %s %s if %u",
880 routedesc, ipv4_ll_buf, label_buf,
881 nexthop->ifindex);
882 return;
0aabccc0
DD		 883 }
884
d62a17ae 885 if (nexthop->type == NEXTHOP_TYPE_IPV4
886 || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX) {
887 /* Send deletes to the kernel without specifying the next-hop */
888 if (cmd != RTM_DELROUTE)
889 _netlink_route_nl_add_gateway_info(
890 rtmsg->rtm_family, AF_INET, nlmsg, req_size,
891 bytelen, nexthop);
892
893 if (cmd == RTM_NEWROUTE) {
894 if (nexthop->rmap_src.ipv4.s_addr)
895 addattr_l(nlmsg, req_size, RTA_PREFSRC,
896 &nexthop->rmap_src.ipv4, bytelen);
897 else if (nexthop->src.ipv4.s_addr)
898 addattr_l(nlmsg, req_size, RTA_PREFSRC,
899 &nexthop->src.ipv4, bytelen);
900 }
901
902 if (IS_ZEBRA_DEBUG_KERNEL)
903 zlog_debug(
904 "netlink_route_multipath() (%s): "
905 "nexthop via %s %s if %u",
906 routedesc, inet_ntoa(nexthop->gate.ipv4),
907 label_buf, nexthop->ifindex);
0aabccc0 908 }
fa713d9e 909
d62a17ae 910 if (nexthop->type == NEXTHOP_TYPE_IPV6
911 || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) {
912 _netlink_route_nl_add_gateway_info(rtmsg->rtm_family, AF_INET6,
913 nlmsg, req_size, bytelen,
914 nexthop);
915
916 if (cmd == RTM_NEWROUTE) {
917 if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->rmap_src.ipv6))
918 addattr_l(nlmsg, req_size, RTA_PREFSRC,
919 &nexthop->rmap_src.ipv6, bytelen);
920 else if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->src.ipv6))
921 addattr_l(nlmsg, req_size, RTA_PREFSRC,
922 &nexthop->src.ipv6, bytelen);
923 }
fa713d9e 924
d62a17ae 925 if (IS_ZEBRA_DEBUG_KERNEL)
926 zlog_debug(
927 "netlink_route_multipath() (%s): "
928 "nexthop via %s %s if %u",
929 routedesc, inet6_ntoa(nexthop->gate.ipv6),
930 label_buf, nexthop->ifindex);
931 }
932 if (nexthop->type == NEXTHOP_TYPE_IFINDEX
933 || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX) {
934 addattr32(nlmsg, req_size, RTA_OIF, nexthop->ifindex);
935
936 if (cmd == RTM_NEWROUTE) {
937 if (nexthop->rmap_src.ipv4.s_addr)
938 addattr_l(nlmsg, req_size, RTA_PREFSRC,
939 &nexthop->rmap_src.ipv4, bytelen);
940 else if (nexthop->src.ipv4.s_addr)
941 addattr_l(nlmsg, req_size, RTA_PREFSRC,
942 &nexthop->src.ipv4, bytelen);
943 }
fa713d9e 944
d62a17ae 945 if (IS_ZEBRA_DEBUG_KERNEL)
946 zlog_debug(
947 "netlink_route_multipath() (%s): "
948 "nexthop via if %u",
949 routedesc, nexthop->ifindex);
0aabccc0
DD		 950 }
951
d62a17ae 952 if (nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) {
953 addattr32(nlmsg, req_size, RTA_OIF, nexthop->ifindex);
954
955 if (cmd == RTM_NEWROUTE) {
956 if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->rmap_src.ipv6))
957 addattr_l(nlmsg, req_size, RTA_PREFSRC,
958 &nexthop->rmap_src.ipv6, bytelen);
959 else if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->src.ipv6))
960 addattr_l(nlmsg, req_size, RTA_PREFSRC,
961 &nexthop->src.ipv6, bytelen);
962 }
963
964 if (IS_ZEBRA_DEBUG_KERNEL)
965 zlog_debug(
966 "netlink_route_multipath() (%s): "
967 "nexthop via if %u",
968 routedesc, nexthop->ifindex);
969 }
fa713d9e
CF		 970}
971
972/* This function takes a nexthop as argument and
973 * appends to the given rtattr/rtnexthop pair the
974 * representation of the nexthop. If the nexthop
975 * defines a preferred source, the src parameter
976 * will be modified to point to that src, otherwise
977 * it will be kept unmodified.
978 *
979 * @param routedesc: Human readable description of route type
980 * (direct/recursive, single-/multipath)
981 * @param bytelen: Length of addresses in bytes.
982 * @param nexthop: Nexthop information
983 * @param rta: rtnetlink attribute structure
984 * @param rtnh: pointer to an rtnetlink nexthop structure
985 * @param src: pointer pointing to a location where
986 * the prefsrc should be stored.
987 */
d62a17ae 988static void _netlink_route_build_multipath(const char *routedesc, int bytelen,
989 struct nexthop *nexthop,
990 struct rtattr *rta,
991 struct rtnexthop *rtnh,
992 struct rtmsg *rtmsg,
993 union g_addr **src)
fa713d9e 994{
d62a17ae 995 struct nexthop_label *nh_label;
996 mpls_lse_t out_lse[MPLS_MAX_LABELS];
9a62e84b 997 char label_buf[256];
d62a17ae 998
999 rtnh->rtnh_len = sizeof(*rtnh);
1000 rtnh->rtnh_flags = 0;
1001 rtnh->rtnh_hops = 0;
1002 rta->rta_len += rtnh->rtnh_len;
1003
1004 /*
1005 * label_buf is *only* currently used within debugging.
1006 * As such when we assign it we are guarding it inside
1007 * a debug test. If you want to change this make sure
1008 * you fix this assumption
1009 */
1010 label_buf[0] = '\0';
1011 /* outgoing label - either as NEWDST (in the case of LSR) or as ENCAP
1012 * (in the case of LER)
1013 */
1014 nh_label = nexthop->nh_label;
1015 if (rtmsg->rtm_family == AF_MPLS) {
1016 assert(nh_label);
1017 assert(nh_label->num_labels == 1);
1018 }
1019
1020 if (nh_label && nh_label->num_labels) {
1021 int i, num_labels = 0;
1022 u_int32_t bos;
1023 char label_buf1[20];
1024
1025 for (i = 0; i < nh_label->num_labels; i++) {
1026 if (nh_label->label[i] != MPLS_IMP_NULL_LABEL) {
1027 bos = ((i == (nh_label->num_labels - 1)) ? 1
1028 : 0);
1029 out_lse[i] = mpls_lse_encode(nh_label->label[i],
1030 0, 0, bos);
1031 if (IS_ZEBRA_DEBUG_KERNEL) {
1032 if (!num_labels)
9a62e84b 1033 sprintf(label_buf, "label %u",
d62a17ae 1034 nh_label->label[i]);
1035 else {
9a62e84b 1036 sprintf(label_buf1, "/%u",
d62a17ae 1037 nh_label->label[i]);
9a62e84b
RW		 1038 strlcat(label_buf, label_buf1,
1039 sizeof(label_buf));
d62a17ae 1040 }
1041 }
1042 num_labels++;
1043 }
1044 }
1045 if (num_labels) {
1046 if (rtmsg->rtm_family == AF_MPLS) {
1047 rta_addattr_l(rta, NL_PKT_BUF_SIZE, RTA_NEWDST,
1048 &out_lse,
1049 num_labels * sizeof(mpls_lse_t));
1050 rtnh->rtnh_len += RTA_LENGTH(
1051 num_labels * sizeof(mpls_lse_t));
1052 } else {
1053 struct rtattr *nest;
1054 u_int16_t encap = LWTUNNEL_ENCAP_MPLS;
1055 int len = rta->rta_len;
1056
1057 rta_addattr_l(rta, NL_PKT_BUF_SIZE,
1058 RTA_ENCAP_TYPE, &encap,
1059 sizeof(u_int16_t));
1060 nest = rta_nest(rta, NL_PKT_BUF_SIZE,
1061 RTA_ENCAP);
1062 rta_addattr_l(rta, NL_PKT_BUF_SIZE,
1063 MPLS_IPTUNNEL_DST, &out_lse,
1064 num_labels * sizeof(mpls_lse_t));
1065 rta_nest_end(rta, nest);
1066 rtnh->rtnh_len += rta->rta_len - len;
1067 }
66d42727 1068 }
d62a17ae 1069 }
1070
1071 if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ONLINK))
1072 rtnh->rtnh_flags |= RTNH_F_ONLINK;
1073
1074 if (rtmsg->rtm_family == AF_INET
1075 && (nexthop->type == NEXTHOP_TYPE_IPV6
1076 || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX)) {
1077 bytelen = 4;
1078 rtnh->rtnh_flags |= RTNH_F_ONLINK;
1079 rta_addattr_l(rta, NL_PKT_BUF_SIZE, RTA_GATEWAY, &ipv4_ll,
1080 bytelen);
1081 rtnh->rtnh_len += sizeof(struct rtattr) + bytelen;
1082 rtnh->rtnh_ifindex = nexthop->ifindex;
1083
1084 if (nexthop->rmap_src.ipv4.s_addr)
1085 *src = &nexthop->rmap_src;
1086 else if (nexthop->src.ipv4.s_addr)
1087 *src = &nexthop->src;
1088
1089 if (IS_ZEBRA_DEBUG_KERNEL)
1090 zlog_debug(
1091 " 5549: netlink_route_build_multipath() (%s): "
1092 "nexthop via %s %s if %u",
1093 routedesc, ipv4_ll_buf, label_buf,
1094 nexthop->ifindex);
1095 return;
1096 }
1097
1098 if (nexthop->type == NEXTHOP_TYPE_IPV4
1099 || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX) {
1100 _netlink_route_rta_add_gateway_info(rtmsg->rtm_family, AF_INET,
1101 rta, rtnh, NL_PKT_BUF_SIZE,
1102 bytelen, nexthop);
1103 if (nexthop->rmap_src.ipv4.s_addr)
1104 *src = &nexthop->rmap_src;
1105 else if (nexthop->src.ipv4.s_addr)
1106 *src = &nexthop->src;
1107
1108 if (IS_ZEBRA_DEBUG_KERNEL)
1109 zlog_debug(
1110 "netlink_route_multipath() (%s): "
1111 "nexthop via %s %s if %u",
1112 routedesc, inet_ntoa(nexthop->gate.ipv4),
1113 label_buf, nexthop->ifindex);
1114 }
1115 if (nexthop->type == NEXTHOP_TYPE_IPV6
1116 || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) {
1117 _netlink_route_rta_add_gateway_info(rtmsg->rtm_family, AF_INET6,
1118 rta, rtnh, NL_PKT_BUF_SIZE,
1119 bytelen, nexthop);
1120
1121 if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->rmap_src.ipv6))
1122 *src = &nexthop->rmap_src;
1123 else if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->src.ipv6))
1124 *src = &nexthop->src;
1125
1126 if (IS_ZEBRA_DEBUG_KERNEL)
1127 zlog_debug(
1128 "netlink_route_multipath() (%s): "
1129 "nexthop via %s %s if %u",
1130 routedesc, inet6_ntoa(nexthop->gate.ipv6),
1131 label_buf, nexthop->ifindex);
1132 }
1133 /* ifindex */
1134 if (nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX
1135 || nexthop->type == NEXTHOP_TYPE_IFINDEX) {
1136 rtnh->rtnh_ifindex = nexthop->ifindex;
1137
1138 if (nexthop->rmap_src.ipv4.s_addr)
1139 *src = &nexthop->rmap_src;
1140 else if (nexthop->src.ipv4.s_addr)
1141 *src = &nexthop->src;
1142
1143 if (IS_ZEBRA_DEBUG_KERNEL)
1144 zlog_debug(
1145 "netlink_route_multipath() (%s): "
1146 "nexthop via if %u",
1147 routedesc, nexthop->ifindex);
1148 } else if (nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) {
1149 rtnh->rtnh_ifindex = nexthop->ifindex;
1150
1151 if (IS_ZEBRA_DEBUG_KERNEL)
1152 zlog_debug(
1153 "netlink_route_multipath() (%s): "
1154 "nexthop via if %u",
1155 routedesc, nexthop->ifindex);
1156 } else {
1157 rtnh->rtnh_ifindex = 0;
1158 }
fa713d9e
CF		 1159}
1160
d62a17ae 1161static inline void _netlink_mpls_build_singlepath(const char *routedesc,
1162 zebra_nhlfe_t *nhlfe,
1163 struct nlmsghdr *nlmsg,
1164 struct rtmsg *rtmsg,
1165 size_t req_size, int cmd)
40c7bdb0 1166{
d62a17ae 1167 int bytelen;
1168 u_char family;
40c7bdb0 1169
d62a17ae 1170 family = NHLFE_FAMILY(nhlfe);
1171 bytelen = (family == AF_INET ? 4 : 16);
1172 _netlink_route_build_singlepath(routedesc, bytelen, nhlfe->nexthop,
1173 nlmsg, rtmsg, req_size, cmd);
40c7bdb0 1174}
1175
1176
1177static inline void
d62a17ae 1178_netlink_mpls_build_multipath(const char *routedesc, zebra_nhlfe_t *nhlfe,
1179 struct rtattr *rta, struct rtnexthop *rtnh,
1180 struct rtmsg *rtmsg, union g_addr **src)
40c7bdb0 1181{
d62a17ae 1182 int bytelen;
1183 u_char family;
40c7bdb0 1184
d62a17ae 1185 family = NHLFE_FAMILY(nhlfe);
1186 bytelen = (family == AF_INET ? 4 : 16);
1187 _netlink_route_build_multipath(routedesc, bytelen, nhlfe->nexthop, rta,
1188 rtnh, rtmsg, src);
40c7bdb0 1189}
1190
1191
fa713d9e
CF		 1192/* Log debug information for netlink_route_multipath
1193 * if debug logging is enabled.
1194 *
1195 * @param cmd: Netlink command which is to be processed
1196 * @param p: Prefix for which the change is due
1197 * @param nexthop: Nexthop which is currently processed
1198 * @param routedesc: Semantic annotation for nexthop
1199 * (recursive, multipath, etc.)
1200 * @param family: Address family which the change concerns
1201 */
d62a17ae 1202static void _netlink_route_debug(int cmd, struct prefix *p,
1203 struct nexthop *nexthop, const char *routedesc,
1204 int family, struct zebra_vrf *zvrf)
fa713d9e 1205{
d62a17ae 1206 if (IS_ZEBRA_DEBUG_KERNEL) {
1207 char buf[PREFIX_STRLEN];
1208 zlog_debug(
1209 "netlink_route_multipath() (%s): %s %s vrf %u type %s",
1210 routedesc, nl_msg_type_to_str(cmd),
1211 prefix2str(p, buf, sizeof(buf)), zvrf_id(zvrf),
1212 (nexthop) ? nexthop_type_to_str(nexthop->type) : "UNK");
1213 }
1214}
1215
1216static void _netlink_mpls_debug(int cmd, u_int32_t label, const char *routedesc)
40c7bdb0 1217{
d62a17ae 1218 if (IS_ZEBRA_DEBUG_KERNEL)
1219 zlog_debug("netlink_mpls_multipath() (%s): %s %u/20", routedesc,
1220 nl_msg_type_to_str(cmd), label);
fa713d9e
CF		 1221}
1222
d62a17ae 1223static int netlink_neigh_update(int cmd, int ifindex, uint32_t addr, char *lla,
1224 int llalen)
5c610faf 1225{
d62a17ae 1226 struct {
1227 struct nlmsghdr n;
1228 struct ndmsg ndm;
1229 char buf[256];
1230 } req;
5c610faf 1231
d62a17ae 1232 struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT);
8f7d9fc0 1233
d62a17ae 1234 memset(&req.n, 0, sizeof(req.n));
1235 memset(&req.ndm, 0, sizeof(req.ndm));
5c610faf 1236
d62a17ae 1237 req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg));
1238 req.n.nlmsg_flags = NLM_F_CREATE | NLM_F_REQUEST;
1239 req.n.nlmsg_type = cmd; // RTM_NEWNEIGH or RTM_DELNEIGH
1240 req.n.nlmsg_pid = zns->netlink_cmd.snl.nl_pid;
a55ba23f 1241
d62a17ae 1242 req.ndm.ndm_family = AF_INET;
1243 req.ndm.ndm_state = NUD_PERMANENT;
1244 req.ndm.ndm_ifindex = ifindex;
1245 req.ndm.ndm_type = RTN_UNICAST;
5c610faf 1246
d62a17ae 1247 addattr_l(&req.n, sizeof(req), NDA_DST, &addr, 4);
1248 addattr_l(&req.n, sizeof(req), NDA_LLADDR, lla, llalen);
5c610faf 1249
d62a17ae 1250 return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns,
1251 0);
5c610faf
DS		 1252}
1253
718e3744 1254/* Routing table change via netlink interface. */
6ae24471 1255/* Update flag indicates whether this is a "replace" or not. */
d62a17ae 1256static int netlink_route_multipath(int cmd, struct prefix *p,
1257 struct prefix *src_p, struct route_entry *re,
1258 int update)
718e3744 1259{
d62a17ae 1260 int bytelen;
1261 struct sockaddr_nl snl;
1262 struct nexthop *nexthop = NULL;
1263 unsigned int nexthop_num;
a8309422 1264 int discard = 0;
d62a17ae 1265 int family = PREFIX_FAMILY(p);
1266 const char *routedesc;
1267 int setsrc = 0;
1268 union g_addr src;
1269
1270 struct {
1271 struct nlmsghdr n;
1272 struct rtmsg r;
1273 char buf[NL_PKT_BUF_SIZE];
1274 } req;
1275
1276 struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT);
1277 struct zebra_vrf *zvrf = vrf_info_lookup(re->vrf_id);
1278
1279 memset(&req, 0, sizeof req - NL_PKT_BUF_SIZE);
1280
1281 bytelen = (family == AF_INET ? 4 : 16);
1282
1283 req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg));
1284 req.n.nlmsg_flags = NLM_F_CREATE | NLM_F_REQUEST;
1285 if ((cmd == RTM_NEWROUTE) && update)
1286 req.n.nlmsg_flags |= NLM_F_REPLACE;
1287 req.n.nlmsg_type = cmd;
1288 req.n.nlmsg_pid = zns->netlink_cmd.snl.nl_pid;
1289
1290 req.r.rtm_family = family;
1291 req.r.rtm_dst_len = p->prefixlen;
1292 req.r.rtm_src_len = src_p ? src_p->prefixlen : 0;
915902cb 1293 req.r.rtm_protocol = zebra2proto(re->type);
d62a17ae 1294 req.r.rtm_scope = RT_SCOPE_UNIVERSE;
a8309422 1295 req.r.rtm_type = RTN_UNICAST;
d62a17ae 1296
d62a17ae 1297 addattr_l(&req.n, sizeof req, RTA_DST, &p->u.prefix, bytelen);
1298 if (src_p)
1299 addattr_l(&req.n, sizeof req, RTA_SRC, &src_p->u.prefix,
1300 bytelen);
1301
1302 /* Metric. */
1303 /* Hardcode the metric for all routes coming from zebra. Metric isn't
1304 * used
1305 * either by the kernel or by zebra. Its purely for calculating best
1306 * path(s)
1307 * by the routing protocol and for communicating with protocol peers.
1308 */
1309 addattr32(&req.n, sizeof req, RTA_PRIORITY, NL_DEFAULT_ROUTE_METRIC);
1310
1311 /* Table corresponding to this route. */
1312 if (re->table < 256)
1313 req.r.rtm_table = re->table;
1314 else {
1315 req.r.rtm_table = RT_TABLE_UNSPEC;
1316 addattr32(&req.n, sizeof req, RTA_TABLE, re->table);
0aabccc0 1317 }
718e3744 1318
a8309422
DL		 1319 if (discard)
1320 goto skip;
1321
d62a17ae 1322 if (re->mtu || re->nexthop_mtu) {
1323 char buf[NL_PKT_BUF_SIZE];
1324 struct rtattr *rta = (void *)buf;
1325 u_int32_t mtu = re->mtu;
1326 if (!mtu || (re->nexthop_mtu && re->nexthop_mtu < mtu))
1327 mtu = re->nexthop_mtu;
1328 rta->rta_type = RTA_METRICS;
1329 rta->rta_len = RTA_LENGTH(0);
1330 rta_addattr_l(rta, NL_PKT_BUF_SIZE, RTAX_MTU, &mtu, sizeof mtu);
1331 addattr_l(&req.n, NL_PKT_BUF_SIZE, RTA_METRICS, RTA_DATA(rta),
1332 RTA_PAYLOAD(rta));
1333 }
1334
d62a17ae 1335 /* Count overall nexthops so we can decide whether to use singlepath
1336 * or multipath case. */
1337 nexthop_num = 0;
1338 for (ALL_NEXTHOPS(re->nexthop, nexthop)) {
1339 if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_RECURSIVE))
1340 continue;
1341 if (cmd == RTM_NEWROUTE
25b9cb0c 1342 && !NEXTHOP_IS_ACTIVE(nexthop->flags))
d62a17ae 1343 continue;
1344 if (cmd == RTM_DELROUTE
1345 && !CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB))
1346 continue;
1347
1348 nexthop_num++;
1349 }
1350
1351 /* Singlepath case. */
1352 if (nexthop_num == 1 || multipath_num == 1) {
1353 nexthop_num = 0;
1354 for (ALL_NEXTHOPS(re->nexthop, nexthop)) {
1f1d24a8
DS		 1355 /*
1356 * So we want to cover 2 types of blackhole
1357 * routes here:
1358 * 1) A normal blackhole route( ala from a static
1359 * install.
1360 * 2) A recursively resolved blackhole route
1361 */
1362 if (nexthop->type == NEXTHOP_TYPE_BLACKHOLE) {
1363 switch (nexthop->bh_type) {
1364 case BLACKHOLE_ADMINPROHIB:
1365 req.r.rtm_type = RTN_PROHIBIT;
1366 break;
1367 case BLACKHOLE_REJECT:
1368 req.r.rtm_type = RTN_UNREACHABLE;
1369 break;
1370 default:
1371 req.r.rtm_type = RTN_BLACKHOLE;
1372 break;
1373 }
1374 goto skip;
1375 }
d62a17ae 1376 if (CHECK_FLAG(nexthop->flags,
1377 NEXTHOP_FLAG_RECURSIVE)) {
1378 if (!setsrc) {
1379 if (family == AF_INET) {
1380 if (nexthop->rmap_src.ipv4
1381 .s_addr
1382 != 0) {
1383 src.ipv4 =
1384 nexthop->rmap_src
1385 .ipv4;
1386 setsrc = 1;
1387 } else if (nexthop->src.ipv4
1388 .s_addr
1389 != 0) {
1390 src.ipv4 =
1391 nexthop->src
1392 .ipv4;
1393 setsrc = 1;
1394 }
1395 } else if (family == AF_INET6) {
1396 if (!IN6_IS_ADDR_UNSPECIFIED(
1397 &nexthop->rmap_src
1398 .ipv6)) {
1399 src.ipv6 =
1400 nexthop->rmap_src
1401 .ipv6;
1402 setsrc = 1;
1403 } else if (
1404 !IN6_IS_ADDR_UNSPECIFIED(
1405 &nexthop->src
1406 .ipv6)) {
1407 src.ipv6 =
1408 nexthop->src
1409 .ipv6;
1410 setsrc = 1;
1411 }
1412 }
1413 }
1414 continue;
1415 }
1416
1417 if ((cmd == RTM_NEWROUTE
25b9cb0c 1418 && NEXTHOP_IS_ACTIVE(nexthop->flags))
d62a17ae 1419 || (cmd == RTM_DELROUTE
1420 && CHECK_FLAG(nexthop->flags,
1421 NEXTHOP_FLAG_FIB))) {
1422 routedesc = nexthop->rparent
8b1450b9
RW		 1423 ? "recursive, single-path"
1424 : "single-path";
d62a17ae 1425
1426 _netlink_route_debug(cmd, p, nexthop, routedesc,
1427 family, zvrf);
1428 _netlink_route_build_singlepath(
1429 routedesc, bytelen, nexthop, &req.n,
1430 &req.r, sizeof req, cmd);
1431 nexthop_num++;
1432 break;
1433 }
1434 }
1435 if (setsrc && (cmd == RTM_NEWROUTE)) {
1436 if (family == AF_INET)
1437 addattr_l(&req.n, sizeof req, RTA_PREFSRC,
1438 &src.ipv4, bytelen);
1439 else if (family == AF_INET6)
1440 addattr_l(&req.n, sizeof req, RTA_PREFSRC,
1441 &src.ipv6, bytelen);
1442 }
1443 } else {
1444 char buf[NL_PKT_BUF_SIZE];
1445 struct rtattr *rta = (void *)buf;
1446 struct rtnexthop *rtnh;
1447 union g_addr *src1 = NULL;
1448
1449 rta->rta_type = RTA_MULTIPATH;
1450 rta->rta_len = RTA_LENGTH(0);
1451 rtnh = RTA_DATA(rta);
1452
1453 nexthop_num = 0;
1454 for (ALL_NEXTHOPS(re->nexthop, nexthop)) {
1455 if (nexthop_num >= multipath_num)
1456 break;
1457
1458 if (CHECK_FLAG(nexthop->flags,
1459 NEXTHOP_FLAG_RECURSIVE)) {
1460 /* This only works for IPv4 now */
1461 if (!setsrc) {
1462 if (family == AF_INET) {
1463 if (nexthop->rmap_src.ipv4
1464 .s_addr
1465 != 0) {
1466 src.ipv4 =
1467 nexthop->rmap_src
1468 .ipv4;
1469 setsrc = 1;
1470 } else if (nexthop->src.ipv4
1471 .s_addr
1472 != 0) {
1473 src.ipv4 =
1474 nexthop->src
1475 .ipv4;
1476 setsrc = 1;
1477 }
1478 } else if (family == AF_INET6) {
1479 if (!IN6_IS_ADDR_UNSPECIFIED(
1480 &nexthop->rmap_src
1481 .ipv6)) {
1482 src.ipv6 =
1483 nexthop->rmap_src
1484 .ipv6;
1485 setsrc = 1;
1486 } else if (
1487 !IN6_IS_ADDR_UNSPECIFIED(
1488 &nexthop->src
1489 .ipv6)) {
1490 src.ipv6 =
1491 nexthop->src
1492 .ipv6;
1493 setsrc = 1;
1494 }
1495 }
1496 }
1497 continue;
1498 }
1499
1500 if ((cmd == RTM_NEWROUTE
25b9cb0c 1501 && NEXTHOP_IS_ACTIVE(nexthop->flags))
d62a17ae 1502 || (cmd == RTM_DELROUTE
1503 && CHECK_FLAG(nexthop->flags,
1504 NEXTHOP_FLAG_FIB))) {
1505 routedesc = nexthop->rparent
8b1450b9
RW		 1506 ? "recursive, multipath"
1507 : "multipath";
d62a17ae 1508 nexthop_num++;
1509
1510 _netlink_route_debug(cmd, p, nexthop, routedesc,
1511 family, zvrf);
1512 _netlink_route_build_multipath(
1513 routedesc, bytelen, nexthop, rta, rtnh,
1514 &req.r, &src1);
1515 rtnh = RTNH_NEXT(rtnh);
1516
1517 if (!setsrc && src1) {
1518 if (family == AF_INET)
1519 src.ipv4 = src1->ipv4;
1520 else if (family == AF_INET6)
1521 src.ipv6 = src1->ipv6;
1522
1523 setsrc = 1;
1524 }
1525 }
1526 }
1527 if (setsrc && (cmd == RTM_NEWROUTE)) {
1528 if (family == AF_INET)
1529 addattr_l(&req.n, sizeof req, RTA_PREFSRC,
1530 &src.ipv4, bytelen);
1531 else if (family == AF_INET6)
1532 addattr_l(&req.n, sizeof req, RTA_PREFSRC,
1533 &src.ipv6, bytelen);
1534 if (IS_ZEBRA_DEBUG_KERNEL)
1535 zlog_debug("Setting source");
1536 }
1537
1538 if (rta->rta_len > RTA_LENGTH(0))
1539 addattr_l(&req.n, NL_PKT_BUF_SIZE, RTA_MULTIPATH,
1540 RTA_DATA(rta), RTA_PAYLOAD(rta));
1541 }
718e3744 1542
d62a17ae 1543 /* If there is no useful nexthop then return. */
1544 if (nexthop_num == 0) {
1545 if (IS_ZEBRA_DEBUG_KERNEL)
1546 zlog_debug(
1547 "netlink_route_multipath(): No useful nexthop.");
1548 return 0;
1549 }
718e3744 1550
7021c425 1551skip:
718e3744 1552
d62a17ae 1553 /* Destination netlink address. */
1554 memset(&snl, 0, sizeof snl);
1555 snl.nl_family = AF_NETLINK;
718e3744 1556
d62a17ae 1557 /* Talk to netlink socket. */
1558 return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns,
1559 0);
718e3744 1560}
1561
43b5cc5e 1562int kernel_get_ipmr_sg_stats(struct zebra_vrf *zvrf, void *in)
e3be0432 1563{
d62a17ae 1564 int suc = 0;
1565 struct mcast_route_data *mr = (struct mcast_route_data *)in;
bd8b9272
DS		 1566 struct {
1567 struct nlmsghdr n;
1568 struct ndmsg ndm;
1569 char buf[256];
1570 } req;
e3be0432 1571
d62a17ae 1572 mroute = mr;
bd8b9272
DS		 1573 struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT);
1574
1575 memset(&req.n, 0, sizeof(req.n));
1576 memset(&req.ndm, 0, sizeof(req.ndm));
1577
1578 req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg));
1579 req.n.nlmsg_flags = NLM_F_REQUEST;
1580 req.n.nlmsg_pid = zns->netlink_cmd.snl.nl_pid;
1581
1582 req.ndm.ndm_family = RTNL_FAMILY_IPMR;
1583 req.n.nlmsg_type = RTM_GETROUTE;
1584
1585 addattr_l(&req.n, sizeof(req), RTA_IIF, &mroute->ifindex, 4);
1586 addattr_l(&req.n, sizeof(req), RTA_OIF, &mroute->ifindex, 4);
1587 addattr_l(&req.n, sizeof(req), RTA_SRC, &mroute->sg.src.s_addr, 4);
1588 addattr_l(&req.n, sizeof(req), RTA_DST, &mroute->sg.grp.s_addr, 4);
1589 addattr_l(&req.n, sizeof(req), RTA_TABLE, &zvrf->table_id, 4);
e3be0432 1590
bd8b9272
DS		 1591 suc = netlink_talk(netlink_route_change_read_multicast, &req.n,
1592 &zns->netlink_cmd, zns, 0);
e3be0432 1593
bd8b9272 1594 mroute = NULL;
d62a17ae 1595 return suc;
e3be0432
DS		 1596}
1597
d62a17ae 1598int kernel_route_rib(struct prefix *p, struct prefix *src_p,
1599 struct route_entry *old, struct route_entry *new)
718e3744 1600{
0af35d90
RW		 1601 assert(old || new);
1602
d62a17ae 1603 if (!old && new)
1604 return netlink_route_multipath(RTM_NEWROUTE, p, src_p, new, 0);
1605 if (old && !new)
1606 return netlink_route_multipath(RTM_DELROUTE, p, src_p, old, 0);
718e3744 1607
d62a17ae 1608 return netlink_route_multipath(RTM_NEWROUTE, p, src_p, new, 1);
718e3744 1609}
1610
d62a17ae 1611int kernel_neigh_update(int add, int ifindex, uint32_t addr, char *lla,
1612 int llalen)
6b8a5694 1613{
d62a17ae 1614 return netlink_neigh_update(add ? RTM_NEWNEIGH : RTM_DELNEIGH, ifindex,
1615 addr, lla, llalen);
6b8a5694 1616}
718e3744 1617
13d60d35 1618/*
1619 * Add remote VTEP to the flood list for this VxLAN interface (VNI). This
1620 * is done by adding an FDB entry with a MAC of 00:00:00:00:00:00.
1621 */
d62a17ae 1622static int netlink_vxlan_flood_list_update(struct interface *ifp,
1623 struct in_addr *vtep_ip, int cmd)
13d60d35 1624{
d62a17ae 1625 struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT);
1626 struct {
1627 struct nlmsghdr n;
1628 struct ndmsg ndm;
1629 char buf[256];
1630 } req;
1631 u_char dst_mac[6] = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0};
1632
1633 memset(&req.n, 0, sizeof(req.n));
1634 memset(&req.ndm, 0, sizeof(req.ndm));
1635
1636 req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg));
1637 req.n.nlmsg_flags = NLM_F_REQUEST;
1638 if (cmd == RTM_NEWNEIGH)
1639 req.n.nlmsg_flags |= (NLM_F_CREATE | NLM_F_APPEND);
1640 req.n.nlmsg_type = cmd;
1641 req.ndm.ndm_family = PF_BRIDGE;
1642 req.ndm.ndm_state = NUD_NOARP | NUD_PERMANENT;
1643 req.ndm.ndm_flags |= NTF_SELF; // Handle by "self", not "master"
1644
1645
1646 addattr_l(&req.n, sizeof(req), NDA_LLADDR, &dst_mac, 6);
1647 req.ndm.ndm_ifindex = ifp->ifindex;
1648 addattr_l(&req.n, sizeof(req), NDA_DST, &vtep_ip->s_addr, 4);
1649
1650 return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns,
1651 0);
13d60d35 1652}
1653
1654/*
d62a17ae 1655 * Add remote VTEP for this VxLAN interface (VNI). In Linux, this involves
1656 * adding
13d60d35 1657 * a "flood" MAC FDB entry.
1658 */
d62a17ae 1659int kernel_add_vtep(vni_t vni, struct interface *ifp, struct in_addr *vtep_ip)
13d60d35 1660{
d62a17ae 1661 if (IS_ZEBRA_DEBUG_VXLAN)
1662 zlog_debug("Install %s into flood list for VNI %u intf %s(%u)",
1663 inet_ntoa(*vtep_ip), vni, ifp->name, ifp->ifindex);
13d60d35 1664
d62a17ae 1665 return netlink_vxlan_flood_list_update(ifp, vtep_ip, RTM_NEWNEIGH);
13d60d35 1666}
1667
1668/*
1669 * Remove remote VTEP for this VxLAN interface (VNI). In Linux, this involves
1670 * deleting the "flood" MAC FDB entry.
1671 */
d62a17ae 1672int kernel_del_vtep(vni_t vni, struct interface *ifp, struct in_addr *vtep_ip)
13d60d35 1673{
d62a17ae 1674 if (IS_ZEBRA_DEBUG_VXLAN)
1675 zlog_debug(
1676 "Uninstall %s from flood list for VNI %u intf %s(%u)",
1677 inet_ntoa(*vtep_ip), vni, ifp->name, ifp->ifindex);
13d60d35 1678
d62a17ae 1679 return netlink_vxlan_flood_list_update(ifp, vtep_ip, RTM_DELNEIGH);
13d60d35 1680}
1681
2232a77c 1682#ifndef NDA_RTA
d62a17ae 1683#define NDA_RTA(r) \
1684 ((struct rtattr *)(((char *)(r)) + NLMSG_ALIGN(sizeof(struct ndmsg))))
2232a77c 1685#endif
1686
d62a17ae 1687static int netlink_macfdb_change(struct sockaddr_nl *snl, struct nlmsghdr *h,
1688 int len)
2232a77c 1689{
d62a17ae 1690 struct ndmsg *ndm;
1691 struct interface *ifp;
1692 struct zebra_if *zif;
d62a17ae 1693 struct rtattr *tb[NDA_MAX + 1];
1694 struct interface *br_if;
1695 struct ethaddr mac;
1696 vlanid_t vid = 0;
1697 struct prefix vtep_ip;
1698 int vid_present = 0, dst_present = 0;
1699 char buf[ETHER_ADDR_STRLEN];
1700 char vid_buf[20];
1701 char dst_buf[30];
1702 u_char sticky = 0;
1703
1704 ndm = NLMSG_DATA(h);
1705
2853fed6 1706 /* We only process macfdb notifications if EVPN is enabled */
1707 if (!is_evpn_enabled())
1708 return 0;
1709
d62a17ae 1710 /* The interface should exist. */
1711 ifp = if_lookup_by_index_per_ns(zebra_ns_lookup(NS_DEFAULT),
1712 ndm->ndm_ifindex);
2853fed6 1713 if (!ifp || !ifp->info)
d62a17ae 1714 return 0;
1715
1716 /* The interface should be something we're interested in. */
1717 if (!IS_ZEBRA_IF_BRIDGE_SLAVE(ifp))
1718 return 0;
1719
1720 /* Drop "permanent" entries. */
1721 if (ndm->ndm_state & NUD_PERMANENT)
1722 return 0;
1723
1724 zif = (struct zebra_if *)ifp->info;
1725 if ((br_if = zif->brslave_info.br_if) == NULL) {
1726 zlog_warn("%s family %s IF %s(%u) brIF %u - no bridge master",
1727 nl_msg_type_to_str(h->nlmsg_type),
1728 nl_family_to_str(ndm->ndm_family), ifp->name,
1729 ndm->ndm_ifindex, zif->brslave_info.bridge_ifindex);
1730 return 0;
1731 }
1732
1733 /* Parse attributes and extract fields of interest. */
1734 memset(tb, 0, sizeof tb);
1735 netlink_parse_rtattr(tb, NDA_MAX, NDA_RTA(ndm), len);
1736
1737 if (!tb[NDA_LLADDR]) {
1738 zlog_warn("%s family %s IF %s(%u) brIF %u - no LLADDR",
1739 nl_msg_type_to_str(h->nlmsg_type),
1740 nl_family_to_str(ndm->ndm_family), ifp->name,
1741 ndm->ndm_ifindex, zif->brslave_info.bridge_ifindex);
1742 return 0;
1743 }
1744
ff8b7eb8 1745 if (RTA_PAYLOAD(tb[NDA_LLADDR]) != ETH_ALEN) {
d62a17ae 1746 zlog_warn(
df0b13cf 1747 "%s family %s IF %s(%u) brIF %u - LLADDR is not MAC, len %lu",
d62a17ae 1748 nl_msg_type_to_str(h->nlmsg_type),
1749 nl_family_to_str(ndm->ndm_family), ifp->name,
1750 ndm->ndm_ifindex, zif->brslave_info.bridge_ifindex,
df0b13cf 1751 (unsigned long)RTA_PAYLOAD(tb[NDA_LLADDR]));
d62a17ae 1752 return 0;
1753 }
1754
ff8b7eb8 1755 memcpy(&mac, RTA_DATA(tb[NDA_LLADDR]), ETH_ALEN);
d62a17ae 1756
1757 if ((NDA_VLAN <= NDA_MAX) && tb[NDA_VLAN]) {
1758 vid_present = 1;
1759 vid = *(u_int16_t *)RTA_DATA(tb[NDA_VLAN]);
1760 sprintf(vid_buf, " VLAN %u", vid);
1761 }
1762
1763 if (tb[NDA_DST]) {
1764 /* TODO: Only IPv4 supported now. */
1765 dst_present = 1;
1766 vtep_ip.family = AF_INET;
1767 vtep_ip.prefixlen = IPV4_MAX_BITLEN;
1768 memcpy(&(vtep_ip.u.prefix4.s_addr), RTA_DATA(tb[NDA_DST]),
1769 IPV4_MAX_BYTELEN);
1770 sprintf(dst_buf, " dst %s", inet_ntoa(vtep_ip.u.prefix4));
1771 }
1772
1773 sticky = (ndm->ndm_state & NUD_NOARP) ? 1 : 0;
1774
1775 if (IS_ZEBRA_DEBUG_KERNEL)
1776 zlog_debug("Rx %s family %s IF %s(%u)%s %sMAC %s%s",
1777 nl_msg_type_to_str(h->nlmsg_type),
1778 nl_family_to_str(ndm->ndm_family), ifp->name,
1779 ndm->ndm_ifindex, vid_present ? vid_buf : "",
1780 sticky ? "sticky " : "",
1781 prefix_mac2str(&mac, buf, sizeof(buf)),
1782 dst_present ? dst_buf : "");
1783
1784 if (filter_vlan && vid != filter_vlan)
1785 return 0;
1786
1787 /* If add or update, do accordingly if learnt on a "local" interface; if
1788 * the notification is over VxLAN, this has to be related to
1789 * multi-homing,
1790 * so perform an implicit delete of any local entry (if it exists).
1791 */
1792 if (h->nlmsg_type == RTM_NEWNEIGH) {
1793 /* Drop "permanent" entries. */
1794 if (ndm->ndm_state & NUD_PERMANENT)
1795 return 0;
1796
1797 if (IS_ZEBRA_IF_VXLAN(ifp))
1798 return zebra_vxlan_check_del_local_mac(ifp, br_if, &mac,
1799 vid);
1800
1801 return zebra_vxlan_local_mac_add_update(ifp, br_if, &mac, vid,
1802 sticky);
1803 }
1804
1805 /* This is a delete notification.
1806 * 1. For a MAC over VxLan, check if it needs to be refreshed(readded)
1807 * 2. For a MAC over "local" interface, delete the mac
1808 * Note: We will get notifications from both bridge driver and VxLAN
1809 * driver.
1810 * Ignore the notification from VxLan driver as it is also generated
1811 * when mac moves from remote to local.
1812 */
1813 if (dst_present)
1814 return 0;
1815
1816 if (IS_ZEBRA_IF_VXLAN(ifp))
1817 return zebra_vxlan_check_readd_remote_mac(ifp, br_if, &mac,
1818 vid);
1819
1820 return zebra_vxlan_local_mac_del(ifp, br_if, &mac, vid);
2232a77c 1821}
1822
d62a17ae 1823static int netlink_macfdb_table(struct sockaddr_nl *snl, struct nlmsghdr *h,
1824 ns_id_t ns_id, int startup)
2232a77c 1825{
d62a17ae 1826 int len;
1827 struct ndmsg *ndm;
2232a77c 1828
d62a17ae 1829 if (h->nlmsg_type != RTM_NEWNEIGH)
1830 return 0;
2232a77c 1831
d62a17ae 1832 /* Length validity. */
1833 len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct ndmsg));
1834 if (len < 0)
1835 return -1;
2232a77c 1836
d62a17ae 1837 /* We are interested only in AF_BRIDGE notifications. */
1838 ndm = NLMSG_DATA(h);
1839 if (ndm->ndm_family != AF_BRIDGE)
1840 return 0;
2232a77c 1841
d62a17ae 1842 return netlink_macfdb_change(snl, h, len);
2232a77c 1843}
1844
1845/* Request for MAC FDB information from the kernel */
d62a17ae 1846static int netlink_request_macs(struct zebra_ns *zns, int family, int type,
1847 ifindex_t master_ifindex)
2232a77c 1848{
d62a17ae 1849 struct {
1850 struct nlmsghdr n;
1851 struct ifinfomsg ifm;
1852 char buf[256];
1853 } req;
1854
1855 /* Form the request, specifying filter (rtattr) if needed. */
1856 memset(&req, 0, sizeof(req));
1857 req.n.nlmsg_type = type;
1858 req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ifinfomsg));
1859 req.ifm.ifi_family = family;
1860 if (master_ifindex)
1861 addattr32(&req.n, sizeof(req), IFLA_MASTER, master_ifindex);
1862
1863 return netlink_request(&zns->netlink_cmd, &req.n);
2232a77c 1864}
1865
1866/*
1867 * MAC forwarding database read using netlink interface. This is invoked
1868 * at startup.
1869 */
d62a17ae 1870int netlink_macfdb_read(struct zebra_ns *zns)
2232a77c 1871{
d62a17ae 1872 int ret;
1873
1874 /* Get bridge FDB table. */
1875 ret = netlink_request_macs(zns, AF_BRIDGE, RTM_GETNEIGH, 0);
1876 if (ret < 0)
1877 return ret;
1878 /* We are reading entire table. */
1879 filter_vlan = 0;
1880 ret = netlink_parse_info(netlink_macfdb_table, &zns->netlink_cmd, zns,
1881 0, 1);
1882
1883 return ret;
2232a77c 1884}
1885
1886/*
1887 * MAC forwarding database read using netlink interface. This is for a
1888 * specific bridge and matching specific access VLAN (if VLAN-aware bridge).
1889 */
d62a17ae 1890int netlink_macfdb_read_for_bridge(struct zebra_ns *zns, struct interface *ifp,
1891 struct interface *br_if)
2232a77c 1892{
d62a17ae 1893 struct zebra_if *br_zif;
1894 struct zebra_if *zif;
1895 struct zebra_l2info_vxlan *vxl;
1896 int ret = 0;
1897
1898
1899 /* Save VLAN we're filtering on, if needed. */
1900 br_zif = (struct zebra_if *)br_if->info;
1901 zif = (struct zebra_if *)ifp->info;
1902 vxl = &zif->l2info.vxl;
1903 if (IS_ZEBRA_IF_BRIDGE_VLAN_AWARE(br_zif))
1904 filter_vlan = vxl->access_vlan;
1905
1906 /* Get bridge FDB table for specific bridge - we do the VLAN filtering.
1907 */
1908 ret = netlink_request_macs(zns, AF_BRIDGE, RTM_GETNEIGH,
1909 br_if->ifindex);
1910 if (ret < 0)
1911 return ret;
1912 ret = netlink_parse_info(netlink_macfdb_table, &zns->netlink_cmd, zns,
1913 0, 0);
1914
1915 /* Reset VLAN filter. */
1916 filter_vlan = 0;
1917 return ret;
2232a77c 1918}
1919
d62a17ae 1920static int netlink_macfdb_update(struct interface *ifp, vlanid_t vid,
1921 struct ethaddr *mac, struct in_addr vtep_ip,
1922 int local, int cmd, u_char sticky)
2232a77c 1923{
d62a17ae 1924 struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT);
1925 struct {
1926 struct nlmsghdr n;
1927 struct ndmsg ndm;
1928 char buf[256];
1929 } req;
1930 int dst_alen;
1931 struct zebra_if *zif;
1932 struct interface *br_if;
1933 struct zebra_if *br_zif;
1934 char buf[ETHER_ADDR_STRLEN];
1935 int vid_present = 0, dst_present = 0;
1936 char vid_buf[20];
1937 char dst_buf[30];
1938
1939 zif = ifp->info;
1940 if ((br_if = zif->brslave_info.br_if) == NULL) {
1941 zlog_warn("MAC %s on IF %s(%u) - no mapping to bridge",
1942 (cmd == RTM_NEWNEIGH) ? "add" : "del", ifp->name,
1943 ifp->ifindex);
1944 return -1;
1945 }
1946
1947 memset(&req.n, 0, sizeof(req.n));
1948 memset(&req.ndm, 0, sizeof(req.ndm));
1949
1950 req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg));
1951 req.n.nlmsg_flags = NLM_F_REQUEST;
1952 if (cmd == RTM_NEWNEIGH)
1953 req.n.nlmsg_flags |= (NLM_F_CREATE | NLM_F_REPLACE);
1954 req.n.nlmsg_type = cmd;
1955 req.ndm.ndm_family = AF_BRIDGE;
1956 req.ndm.ndm_flags |= NTF_SELF | NTF_MASTER;
1957 req.ndm.ndm_state = NUD_REACHABLE;
1958
1959 if (sticky)
1960 req.ndm.ndm_state |= NUD_NOARP;
1961 else
1962 req.ndm.ndm_flags |= NTF_EXT_LEARNED;
1963
1964 addattr_l(&req.n, sizeof(req), NDA_LLADDR, mac, 6);
1965 req.ndm.ndm_ifindex = ifp->ifindex;
1966 if (!local) {
1967 dst_alen = 4; // TODO: hardcoded
1968 addattr_l(&req.n, sizeof(req), NDA_DST, &vtep_ip, dst_alen);
1969 dst_present = 1;
1970 sprintf(dst_buf, " dst %s", inet_ntoa(vtep_ip));
1971 }
1972 br_zif = (struct zebra_if *)br_if->info;
1973 if (IS_ZEBRA_IF_BRIDGE_VLAN_AWARE(br_zif) && vid > 0) {
1974 addattr16(&req.n, sizeof(req), NDA_VLAN, vid);
1975 vid_present = 1;
1976 sprintf(vid_buf, " VLAN %u", vid);
1977 }
1978 addattr32(&req.n, sizeof(req), NDA_MASTER, br_if->ifindex);
1979
1980 if (IS_ZEBRA_DEBUG_KERNEL)
1981 zlog_debug("Tx %s family %s IF %s(%u)%s %sMAC %s%s",
1982 nl_msg_type_to_str(cmd),
1983 nl_family_to_str(req.ndm.ndm_family), ifp->name,
1984 ifp->ifindex, vid_present ? vid_buf : "",
1985 sticky ? "sticky " : "",
1986 prefix_mac2str(mac, buf, sizeof(buf)),
1987 dst_present ? dst_buf : "");
1988
1989 return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns,
1990 0);
2232a77c 1991}
1992
d62a17ae 1993#define NUD_VALID \
1994 (NUD_PERMANENT | NUD_NOARP | NUD_REACHABLE | NUD_PROBE | NUD_STALE \
1995 | NUD_DELAY)
2232a77c 1996
d62a17ae 1997static int netlink_ipneigh_change(struct sockaddr_nl *snl, struct nlmsghdr *h,
1998 int len)
2232a77c 1999{
d62a17ae 2000 struct ndmsg *ndm;
2001 struct interface *ifp;
2002 struct zebra_if *zif;
d62a17ae 2003 struct rtattr *tb[NDA_MAX + 1];
2004 struct interface *link_if;
2005 struct ethaddr mac;
2006 struct ipaddr ip;
2007 char buf[ETHER_ADDR_STRLEN];
2008 char buf2[INET6_ADDRSTRLEN];
2009 int mac_present = 0;
2010 u_char ext_learned;
2011
2012 ndm = NLMSG_DATA(h);
2013
2853fed6 2014 /* We only process neigh notifications if EVPN is enabled */
2015 if (!is_evpn_enabled())
2016 return 0;
2017
d62a17ae 2018 /* The interface should exist. */
2019 ifp = if_lookup_by_index_per_ns(zebra_ns_lookup(NS_DEFAULT),
2020 ndm->ndm_ifindex);
2853fed6 2021 if (!ifp || !ifp->info)
d62a17ae 2022 return 0;
2023
2024 /* Drop "permanent" entries. */
2025 if (ndm->ndm_state & NUD_PERMANENT)
2026 return 0;
2027
2028 zif = (struct zebra_if *)ifp->info;
2029 /* The neighbor is present on an SVI. From this, we locate the
2030 * underlying
2031 * bridge because we're only interested in neighbors on a VxLAN bridge.
2032 * The bridge is located based on the nature of the SVI:
2033 * (a) In the case of a VLAN-aware bridge, the SVI is a L3 VLAN
2034 * interface
2035 * and is linked to the bridge
2036 * (b) In the case of a VLAN-unaware bridge, the SVI is the bridge
2037 * inteface
2038 * itself
2039 */
2040 if (IS_ZEBRA_IF_VLAN(ifp)) {
71349e03
MK		 2041 link_if = if_lookup_by_index_per_ns(zebra_ns_lookup(NS_DEFAULT),
2042 zif->link_ifindex);
d62a17ae 2043 if (!link_if)
2044 return 0;
2045 } else if (IS_ZEBRA_IF_BRIDGE(ifp))
2046 link_if = ifp;
2047 else
2048 return 0;
2049
2050 /* Parse attributes and extract fields of interest. */
2051 memset(tb, 0, sizeof tb);
2052 netlink_parse_rtattr(tb, NDA_MAX, NDA_RTA(ndm), len);
2053
2054 if (!tb[NDA_DST]) {
2055 zlog_warn("%s family %s IF %s(%u) - no DST",
2056 nl_msg_type_to_str(h->nlmsg_type),
2057 nl_family_to_str(ndm->ndm_family), ifp->name,
2058 ndm->ndm_ifindex);
2059 return 0;
2060 }
2061 memset(&mac, 0, sizeof(struct ethaddr));
2062 memset(&ip, 0, sizeof(struct ipaddr));
2063 ip.ipa_type = (ndm->ndm_family == AF_INET) ? IPADDR_V4 : IPADDR_V6;
2064 memcpy(&ip.ip.addr, RTA_DATA(tb[NDA_DST]), RTA_PAYLOAD(tb[NDA_DST]));
2065
2066 if (h->nlmsg_type == RTM_NEWNEIGH) {
2067 if (tb[NDA_LLADDR]) {
ff8b7eb8 2068 if (RTA_PAYLOAD(tb[NDA_LLADDR]) != ETH_ALEN) {
d62a17ae 2069 zlog_warn(
df0b13cf 2070 "%s family %s IF %s(%u) - LLADDR is not MAC, len %lu",
d62a17ae 2071 nl_msg_type_to_str(h->nlmsg_type),
2072 nl_family_to_str(ndm->ndm_family),
2073 ifp->name, ndm->ndm_ifindex,
df0b13cf 2074 (unsigned long)RTA_PAYLOAD(tb[NDA_LLADDR]));
d62a17ae 2075 return 0;
2076 }
2077
2078 mac_present = 1;
ff8b7eb8 2079 memcpy(&mac, RTA_DATA(tb[NDA_LLADDR]), ETH_ALEN);
d62a17ae 2080 }
2081
2082 ext_learned = (ndm->ndm_flags & NTF_EXT_LEARNED) ? 1 : 0;
2083
2084 if (IS_ZEBRA_DEBUG_KERNEL)
2085 zlog_debug(
2086 "Rx %s family %s IF %s(%u) IP %s MAC %s state 0x%x flags 0x%x",
2087 nl_msg_type_to_str(h->nlmsg_type),
2088 nl_family_to_str(ndm->ndm_family), ifp->name,
2089 ndm->ndm_ifindex,
2090 ipaddr2str(&ip, buf2, sizeof(buf2)),
2091 mac_present
2092 ? prefix_mac2str(&mac, buf, sizeof(buf))
2093 : "",
2094 ndm->ndm_state, ndm->ndm_flags);
2095
2096 /* If the neighbor state is valid for use, process as an add or
2097 * update
2098 * else process as a delete. Note that the delete handling may
2099 * result
2100 * in re-adding the neighbor if it is a valid "remote" neighbor.
2101 */
2102 if (ndm->ndm_state & NUD_VALID)
2103 return zebra_vxlan_local_neigh_add_update(
2104 ifp, link_if, &ip, &mac, ndm->ndm_state,
2105 ext_learned);
2106
2107 return zebra_vxlan_local_neigh_del(ifp, link_if, &ip);
2108 }
2109
2110 if (IS_ZEBRA_DEBUG_KERNEL)
2111 zlog_debug("Rx %s family %s IF %s(%u) IP %s",
2112 nl_msg_type_to_str(h->nlmsg_type),
2113 nl_family_to_str(ndm->ndm_family), ifp->name,
2114 ndm->ndm_ifindex,
2115 ipaddr2str(&ip, buf2, sizeof(buf2)));
2116
2117 /* Process the delete - it may result in re-adding the neighbor if it is
2118 * a valid "remote" neighbor.
2119 */
2120 return zebra_vxlan_local_neigh_del(ifp, link_if, &ip);
2232a77c 2121}
2122
d62a17ae 2123static int netlink_neigh_table(struct sockaddr_nl *snl, struct nlmsghdr *h,
2124 ns_id_t ns_id, int startup)
2232a77c 2125{
d62a17ae 2126 int len;
2127 struct ndmsg *ndm;
2232a77c 2128
d62a17ae 2129 if (h->nlmsg_type != RTM_NEWNEIGH)
2130 return 0;
2232a77c 2131
d62a17ae 2132 /* Length validity. */
2133 len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct ndmsg));
2134 if (len < 0)
2135 return -1;
2232a77c 2136
d62a17ae 2137 /* We are interested only in AF_INET or AF_INET6 notifications. */
2138 ndm = NLMSG_DATA(h);
2139 if (ndm->ndm_family != AF_INET && ndm->ndm_family != AF_INET6)
2140 return 0;
2232a77c 2141
d62a17ae 2142 return netlink_neigh_change(snl, h, len);
2232a77c 2143}
2144
2145/* Request for IP neighbor information from the kernel */
d62a17ae 2146static int netlink_request_neigh(struct zebra_ns *zns, int family, int type,
2147 ifindex_t ifindex)
2232a77c 2148{
d62a17ae 2149 struct {
2150 struct nlmsghdr n;
2151 struct ndmsg ndm;
2152 char buf[256];
2153 } req;
2154
2155 /* Form the request, specifying filter (rtattr) if needed. */
2156 memset(&req, 0, sizeof(req));
2157 req.n.nlmsg_type = type;
2158 req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg));
2159 req.ndm.ndm_family = family;
2160 if (ifindex)
2161 addattr32(&req.n, sizeof(req), NDA_IFINDEX, ifindex);
2162
2163 return netlink_request(&zns->netlink_cmd, &req.n);
2232a77c 2164}
2165
2166/*
2167 * IP Neighbor table read using netlink interface. This is invoked
2168 * at startup.
2169 */
d62a17ae 2170int netlink_neigh_read(struct zebra_ns *zns)
2232a77c 2171{
d62a17ae 2172 int ret;
2232a77c 2173
d62a17ae 2174 /* Get IP neighbor table. */
2175 ret = netlink_request_neigh(zns, AF_UNSPEC, RTM_GETNEIGH, 0);
2176 if (ret < 0)
2177 return ret;
2178 ret = netlink_parse_info(netlink_neigh_table, &zns->netlink_cmd, zns, 0,
2179 1);
2232a77c 2180
d62a17ae 2181 return ret;
2232a77c 2182}
2183
2184/*
2185 * IP Neighbor table read using netlink interface. This is for a specific
2186 * VLAN device.
2187 */
d62a17ae 2188int netlink_neigh_read_for_vlan(struct zebra_ns *zns, struct interface *vlan_if)
2232a77c 2189{
d62a17ae 2190 int ret = 0;
2232a77c 2191
d62a17ae 2192 ret = netlink_request_neigh(zns, AF_UNSPEC, RTM_GETNEIGH,
2193 vlan_if->ifindex);
2194 if (ret < 0)
2195 return ret;
2196 ret = netlink_parse_info(netlink_neigh_table, &zns->netlink_cmd, zns, 0,
2197 0);
2232a77c 2198
d62a17ae 2199 return ret;
2232a77c 2200}
2201
d62a17ae 2202int netlink_neigh_change(struct sockaddr_nl *snl, struct nlmsghdr *h,
2203 ns_id_t ns_id)
2232a77c 2204{
d62a17ae 2205 int len;
2206 struct ndmsg *ndm;
2232a77c 2207
d62a17ae 2208 if (!(h->nlmsg_type == RTM_NEWNEIGH || h->nlmsg_type == RTM_DELNEIGH))
2209 return 0;
2232a77c 2210
d62a17ae 2211 /* Length validity. */
2212 len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct ndmsg));
2213 if (len < 0)
2214 return -1;
2232a77c 2215
d62a17ae 2216 /* Is this a notification for the MAC FDB or IP neighbor table? */
2217 ndm = NLMSG_DATA(h);
2218 if (ndm->ndm_family == AF_BRIDGE)
2219 return netlink_macfdb_change(snl, h, len);
2232a77c 2220
d62a17ae 2221 if (ndm->ndm_type != RTN_UNICAST)
2222 return 0;
2232a77c 2223
d62a17ae 2224 if (ndm->ndm_family == AF_INET || ndm->ndm_family == AF_INET6)
2225 return netlink_ipneigh_change(snl, h, len);
2232a77c 2226
d62a17ae 2227 return 0;
2232a77c 2228}
2229
d62a17ae 2230static int netlink_neigh_update2(struct interface *ifp, struct ipaddr *ip,
2231 struct ethaddr *mac, u_int32_t flags, int cmd)
2232a77c 2232{
d62a17ae 2233 struct {
2234 struct nlmsghdr n;
2235 struct ndmsg ndm;
2236 char buf[256];
2237 } req;
2238 int ipa_len;
2239
2240 struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT);
2241 char buf[INET6_ADDRSTRLEN];
2242 char buf2[ETHER_ADDR_STRLEN];
2243
2244 memset(&req.n, 0, sizeof(req.n));
2245 memset(&req.ndm, 0, sizeof(req.ndm));
2246
2247 req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg));
2248 req.n.nlmsg_flags = NLM_F_REQUEST;
2249 if (cmd == RTM_NEWNEIGH)
2250 req.n.nlmsg_flags |= (NLM_F_CREATE | NLM_F_REPLACE);
2251 req.n.nlmsg_type = cmd; // RTM_NEWNEIGH or RTM_DELNEIGH
2252 req.ndm.ndm_family = IS_IPADDR_V4(ip) ? AF_INET : AF_INET6;
2253 req.ndm.ndm_state = flags;
2254 req.ndm.ndm_ifindex = ifp->ifindex;
2255 req.ndm.ndm_type = RTN_UNICAST;
2256 req.ndm.ndm_flags = NTF_EXT_LEARNED;
2257
2258
2259 ipa_len = IS_IPADDR_V4(ip) ? IPV4_MAX_BYTELEN : IPV6_MAX_BYTELEN;
2260 addattr_l(&req.n, sizeof(req), NDA_DST, &ip->ip.addr, ipa_len);
2261 if (mac)
2262 addattr_l(&req.n, sizeof(req), NDA_LLADDR, mac, 6);
2263
2264 if (IS_ZEBRA_DEBUG_KERNEL)
2265 zlog_debug("Tx %s family %s IF %s(%u) Neigh %s MAC %s",
2266 nl_msg_type_to_str(cmd),
2267 nl_family_to_str(req.ndm.ndm_family), ifp->name,
2268 ifp->ifindex, ipaddr2str(ip, buf, sizeof(buf)),
2269 mac ? prefix_mac2str(mac, buf2, sizeof(buf2))
2270 : "null");
2271
2272 return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns,
2273 0);
2232a77c 2274}
2275
d62a17ae 2276int kernel_add_mac(struct interface *ifp, vlanid_t vid, struct ethaddr *mac,
2277 struct in_addr vtep_ip, u_char sticky)
2232a77c 2278{
d62a17ae 2279 return netlink_macfdb_update(ifp, vid, mac, vtep_ip, 0, RTM_NEWNEIGH,
2280 sticky);
2232a77c 2281}
2282
d62a17ae 2283int kernel_del_mac(struct interface *ifp, vlanid_t vid, struct ethaddr *mac,
2284 struct in_addr vtep_ip, int local)
2232a77c 2285{
d62a17ae 2286 return netlink_macfdb_update(ifp, vid, mac, vtep_ip, local,
2287 RTM_DELNEIGH, 0);
2232a77c 2288}
2289
d62a17ae 2290int kernel_add_neigh(struct interface *ifp, struct ipaddr *ip,
2291 struct ethaddr *mac)
2232a77c 2292{
d62a17ae 2293 return netlink_neigh_update2(ifp, ip, mac, NUD_REACHABLE, RTM_NEWNEIGH);
2232a77c 2294}
2295
d62a17ae 2296int kernel_del_neigh(struct interface *ifp, struct ipaddr *ip)
2232a77c 2297{
d62a17ae 2298 return netlink_neigh_update2(ifp, ip, NULL, 0, RTM_DELNEIGH);
2232a77c 2299}
2300
40c7bdb0 2301/*
2302 * MPLS label forwarding table change via netlink interface.
2303 */
d62a17ae 2304int netlink_mpls_multipath(int cmd, zebra_lsp_t *lsp)
40c7bdb0 2305{
d62a17ae 2306 mpls_lse_t lse;
2307 zebra_nhlfe_t *nhlfe;
2308 struct nexthop *nexthop = NULL;
2309 unsigned int nexthop_num;
2310 const char *routedesc;
2311 struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT);
805444ce 2312 int route_type;
d62a17ae 2313
2314 struct {
2315 struct nlmsghdr n;
2316 struct rtmsg r;
2317 char buf[NL_PKT_BUF_SIZE];
2318 } req;
2319
2320 memset(&req, 0, sizeof req - NL_PKT_BUF_SIZE);
2321
d62a17ae 2322 /*
2323 * Count # nexthops so we can decide whether to use singlepath
2324 * or multipath case.
2325 */
2326 nexthop_num = 0;
2327 for (nhlfe = lsp->nhlfe_list; nhlfe; nhlfe = nhlfe->next) {
2328 nexthop = nhlfe->nexthop;
2329 if (!nexthop)
2330 continue;
2331 if (cmd == RTM_NEWROUTE) {
2332 /* Count all selected NHLFEs */
2333 if (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_SELECTED)
2334 && CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ACTIVE))
2335 nexthop_num++;
2336 } else /* DEL */
2337 {
2338 /* Count all installed NHLFEs */
2339 if (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_INSTALLED)
2340 && CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB))
2341 nexthop_num++;
2342 }
2343 }
2344
8dc8a4b6 2345 if ((nexthop_num == 0) || (!lsp->best_nhlfe && (cmd != RTM_DELROUTE)))
d62a17ae 2346 return 0;
2347
2348 req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg));
2349 req.n.nlmsg_flags = NLM_F_CREATE | NLM_F_REQUEST;
2350 req.n.nlmsg_type = cmd;
2351 req.n.nlmsg_pid = zns->netlink_cmd.snl.nl_pid;
2352
2353 req.r.rtm_family = AF_MPLS;
2354 req.r.rtm_table = RT_TABLE_MAIN;
2355 req.r.rtm_dst_len = MPLS_LABEL_LEN_BITS;
d62a17ae 2356 req.r.rtm_scope = RT_SCOPE_UNIVERSE;
2357 req.r.rtm_type = RTN_UNICAST;
2358
8dc8a4b6 2359 if (cmd == RTM_NEWROUTE) {
d62a17ae 2360 /* We do a replace to handle update. */
2361 req.n.nlmsg_flags |= NLM_F_REPLACE;
2362
8dc8a4b6
DS		 2363 /* set the protocol value if installing */
2364 route_type = re_type_from_lsp_type(lsp->best_nhlfe->type);
2365 req.r.rtm_protocol = zebra2proto(route_type);
2366 }
2367
d62a17ae 2368 /* Fill destination */
2369 lse = mpls_lse_encode(lsp->ile.in_label, 0, 0, 1);
2370 addattr_l(&req.n, sizeof req, RTA_DST, &lse, sizeof(mpls_lse_t));
2371
2372 /* Fill nexthops (paths) based on single-path or multipath. The paths
2373 * chosen depend on the operation.
2374 */
2375 if (nexthop_num == 1 || multipath_num == 1) {
8b1450b9 2376 routedesc = "single-path";
d62a17ae 2377 _netlink_mpls_debug(cmd, lsp->ile.in_label, routedesc);
2378
2379 nexthop_num = 0;
2380 for (nhlfe = lsp->nhlfe_list; nhlfe; nhlfe = nhlfe->next) {
2381 nexthop = nhlfe->nexthop;
2382 if (!nexthop)
2383 continue;
2384
2385 if ((cmd == RTM_NEWROUTE
2386 && (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_SELECTED)
2387 && CHECK_FLAG(nexthop->flags,
2388 NEXTHOP_FLAG_ACTIVE)))
2389 || (cmd == RTM_DELROUTE
2390 && (CHECK_FLAG(nhlfe->flags,
2391 NHLFE_FLAG_INSTALLED)
2392 && CHECK_FLAG(nexthop->flags,
2393 NEXTHOP_FLAG_FIB)))) {
2394 /* Add the gateway */
2395 _netlink_mpls_build_singlepath(routedesc, nhlfe,
2396 &req.n, &req.r,
2397 sizeof req, cmd);
2398 if (cmd == RTM_NEWROUTE) {
2399 SET_FLAG(nhlfe->flags,
2400 NHLFE_FLAG_INSTALLED);
2401 SET_FLAG(nexthop->flags,
2402 NEXTHOP_FLAG_FIB);
2403 } else {
2404 UNSET_FLAG(nhlfe->flags,
2405 NHLFE_FLAG_INSTALLED);
2406 UNSET_FLAG(nexthop->flags,
2407 NEXTHOP_FLAG_FIB);
2408 }
2409 nexthop_num++;
2410 break;
2411 }
2412 }
2413 } else /* Multipath case */
2414 {
2415 char buf[NL_PKT_BUF_SIZE];
2416 struct rtattr *rta = (void *)buf;
2417 struct rtnexthop *rtnh;
2418 union g_addr *src1 = NULL;
2419
2420 rta->rta_type = RTA_MULTIPATH;
2421 rta->rta_len = RTA_LENGTH(0);
2422 rtnh = RTA_DATA(rta);
2423
8b1450b9 2424 routedesc = "multipath";
d62a17ae 2425 _netlink_mpls_debug(cmd, lsp->ile.in_label, routedesc);
2426
2427 nexthop_num = 0;
2428 for (nhlfe = lsp->nhlfe_list; nhlfe; nhlfe = nhlfe->next) {
2429 nexthop = nhlfe->nexthop;
2430 if (!nexthop)
2431 continue;
2432
2433 if (nexthop_num >= multipath_num)
2434 break;
2435
2436 if ((cmd == RTM_NEWROUTE
2437 && (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_SELECTED)
2438 && CHECK_FLAG(nexthop->flags,
2439 NEXTHOP_FLAG_ACTIVE)))
2440 || (cmd == RTM_DELROUTE
2441 && (CHECK_FLAG(nhlfe->flags,
2442 NHLFE_FLAG_INSTALLED)
2443 && CHECK_FLAG(nexthop->flags,
2444 NEXTHOP_FLAG_FIB)))) {
2445 nexthop_num++;
2446
2447 /* Build the multipath */
2448 _netlink_mpls_build_multipath(routedesc, nhlfe,
2449 rta, rtnh, &req.r,
2450 &src1);
2451 rtnh = RTNH_NEXT(rtnh);
2452
2453 if (cmd == RTM_NEWROUTE) {
2454 SET_FLAG(nhlfe->flags,
2455 NHLFE_FLAG_INSTALLED);
2456 SET_FLAG(nexthop->flags,
2457 NEXTHOP_FLAG_FIB);
2458 } else {
2459 UNSET_FLAG(nhlfe->flags,
2460 NHLFE_FLAG_INSTALLED);
2461 UNSET_FLAG(nexthop->flags,
2462 NEXTHOP_FLAG_FIB);
2463 }
2464 }
2465 }
2466
2467 /* Add the multipath */
2468 if (rta->rta_len > RTA_LENGTH(0))
2469 addattr_l(&req.n, NL_PKT_BUF_SIZE, RTA_MULTIPATH,
2470 RTA_DATA(rta), RTA_PAYLOAD(rta));
2471 }
2472
2473 /* Talk to netlink socket. */
2474 return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns,
2475 0);
40c7bdb0 2476}
ddfeb486 2477#endif /* HAVE_NETLINK */
FRRouting mirror