]>
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 | |
bf094f69 | 53 | #include "zebra/zapi_msg.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 { |
d7c0a89a QY |
75 | uint16_t filler; |
76 | uint16_t family; | |
d62a17ae | 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) |
d4d71f11 | 97 | || (proto == RTPROT_ZSTATIC) || (proto == RTPROT_ZEBRA) |
d62a17ae | 98 | || (proto == RTPROT_ISIS) || (proto == RTPROT_RIPNG) |
99 | || (proto == RTPROT_NHRP) || (proto == RTPROT_EIGRP) | |
915902cb | 100 | || (proto == RTPROT_LDP) || (proto == RTPROT_BABEL) |
0761368a DS |
101 | || (proto == RTPROT_RIP) || (proto == RTPROT_SHARP) |
102 | || (proto == RTPROT_PBR)) { | |
d62a17ae | 103 | return 1; |
104 | } | |
105 | ||
106 | return 0; | |
23b1f334 DD |
107 | } |
108 | ||
915902cb | 109 | static inline int zebra2proto(int proto) |
23b1f334 | 110 | { |
d62a17ae | 111 | switch (proto) { |
112 | case ZEBRA_ROUTE_BABEL: | |
113 | proto = RTPROT_BABEL; | |
114 | break; | |
115 | case ZEBRA_ROUTE_BGP: | |
116 | proto = RTPROT_BGP; | |
117 | break; | |
118 | case ZEBRA_ROUTE_OSPF: | |
119 | case ZEBRA_ROUTE_OSPF6: | |
120 | proto = RTPROT_OSPF; | |
121 | break; | |
122 | case ZEBRA_ROUTE_STATIC: | |
d4d71f11 | 123 | proto = RTPROT_ZSTATIC; |
d62a17ae | 124 | break; |
125 | case ZEBRA_ROUTE_ISIS: | |
126 | proto = RTPROT_ISIS; | |
127 | break; | |
128 | case ZEBRA_ROUTE_RIP: | |
129 | proto = RTPROT_RIP; | |
130 | break; | |
131 | case ZEBRA_ROUTE_RIPNG: | |
132 | proto = RTPROT_RIPNG; | |
133 | break; | |
134 | case ZEBRA_ROUTE_NHRP: | |
135 | proto = RTPROT_NHRP; | |
136 | break; | |
137 | case ZEBRA_ROUTE_EIGRP: | |
138 | proto = RTPROT_EIGRP; | |
139 | break; | |
140 | case ZEBRA_ROUTE_LDP: | |
141 | proto = RTPROT_LDP; | |
142 | break; | |
8a71d93d DS |
143 | case ZEBRA_ROUTE_SHARP: |
144 | proto = RTPROT_SHARP; | |
145 | break; | |
0761368a DS |
146 | case ZEBRA_ROUTE_PBR: |
147 | proto = RTPROT_PBR; | |
148 | break; | |
d62a17ae | 149 | default: |
0761368a DS |
150 | /* |
151 | * When a user adds a new protocol this will show up | |
152 | * to let them know to do something about it. This | |
153 | * is intentionally a warn because we should see | |
154 | * this as part of development of a new protocol | |
155 | */ | |
156 | zlog_warn("%s: Please add this protocol(%d) to proper rt_netlink.c handling", | |
157 | __PRETTY_FUNCTION__, proto); | |
d62a17ae | 158 | proto = RTPROT_ZEBRA; |
159 | break; | |
160 | } | |
161 | ||
162 | return proto; | |
23b1f334 DD |
163 | } |
164 | ||
915902cb DS |
165 | static inline int proto2zebra(int proto, int family) |
166 | { | |
167 | switch (proto) { | |
168 | case RTPROT_BABEL: | |
169 | proto = ZEBRA_ROUTE_BABEL; | |
170 | break; | |
171 | case RTPROT_BGP: | |
172 | proto = ZEBRA_ROUTE_BGP; | |
173 | break; | |
174 | case RTPROT_OSPF: | |
996c9314 LB |
175 | proto = (family == AFI_IP) ? ZEBRA_ROUTE_OSPF |
176 | : ZEBRA_ROUTE_OSPF6; | |
915902cb DS |
177 | break; |
178 | case RTPROT_ISIS: | |
179 | proto = ZEBRA_ROUTE_ISIS; | |
180 | break; | |
181 | case RTPROT_RIP: | |
182 | proto = ZEBRA_ROUTE_RIP; | |
183 | break; | |
184 | case RTPROT_RIPNG: | |
185 | proto = ZEBRA_ROUTE_RIPNG; | |
186 | break; | |
187 | case RTPROT_NHRP: | |
188 | proto = ZEBRA_ROUTE_NHRP; | |
189 | break; | |
190 | case RTPROT_EIGRP: | |
191 | proto = ZEBRA_ROUTE_EIGRP; | |
192 | break; | |
193 | case RTPROT_LDP: | |
194 | proto = ZEBRA_ROUTE_LDP; | |
195 | break; | |
196 | case RTPROT_STATIC: | |
d4d71f11 | 197 | case RTPROT_ZSTATIC: |
915902cb DS |
198 | proto = ZEBRA_ROUTE_STATIC; |
199 | break; | |
0761368a DS |
200 | case RTPROT_SHARP: |
201 | proto = ZEBRA_ROUTE_SHARP; | |
202 | break; | |
203 | case RTPROT_PBR: | |
204 | proto = ZEBRA_ROUTE_PBR; | |
205 | break; | |
915902cb | 206 | default: |
0761368a DS |
207 | /* |
208 | * When a user adds a new protocol this will show up | |
209 | * to let them know to do something about it. This | |
210 | * is intentionally a warn because we should see | |
211 | * this as part of development of a new protocol | |
212 | */ | |
213 | zlog_warn("%s: Please add this protocol(%d) to proper rt_netlink.c handling", | |
214 | __PRETTY_FUNCTION__, | |
215 | proto); | |
915902cb DS |
216 | proto = ZEBRA_ROUTE_KERNEL; |
217 | break; | |
218 | } | |
219 | return proto; | |
220 | } | |
221 | ||
12f6fb97 DS |
222 | /* |
223 | Pending: create an efficient table_id (in a tree/hash) based lookup) | |
224 | */ | |
d7c0a89a | 225 | static vrf_id_t vrf_lookup_by_table(uint32_t table_id, ns_id_t ns_id) |
12f6fb97 | 226 | { |
d62a17ae | 227 | struct vrf *vrf; |
228 | struct zebra_vrf *zvrf; | |
12f6fb97 | 229 | |
a2addae8 | 230 | RB_FOREACH (vrf, vrf_id_head, &vrfs_by_id) { |
78dd30b2 PG |
231 | zvrf = vrf->info; |
232 | if (zvrf == NULL) | |
d62a17ae | 233 | continue; |
78dd30b2 PG |
234 | /* case vrf with netns : match the netnsid */ |
235 | if (vrf_is_backend_netns()) { | |
236 | if (ns_id == zvrf_id(zvrf)) | |
237 | return zvrf_id(zvrf); | |
238 | } else { | |
239 | /* VRF is VRF_BACKEND_VRF_LITE */ | |
240 | if (zvrf->table_id != table_id) | |
241 | continue; | |
242 | return zvrf_id(zvrf); | |
243 | } | |
d62a17ae | 244 | } |
12f6fb97 | 245 | |
d62a17ae | 246 | return VRF_DEFAULT; |
12f6fb97 DS |
247 | } |
248 | ||
718e3744 | 249 | /* Looking up routing table by netlink interface. */ |
2414abd3 | 250 | static int netlink_route_change_read_unicast(struct nlmsghdr *h, ns_id_t ns_id, |
d62a17ae | 251 | int startup) |
718e3744 | 252 | { |
d62a17ae | 253 | int len; |
254 | struct rtmsg *rtm; | |
255 | struct rtattr *tb[RTA_MAX + 1]; | |
d7c0a89a | 256 | uint8_t flags = 0; |
d62a17ae | 257 | struct prefix p; |
792fa92e | 258 | struct prefix_ipv6 src_p = {}; |
78dd30b2 | 259 | vrf_id_t vrf_id; |
d62a17ae | 260 | |
261 | char anyaddr[16] = {0}; | |
262 | ||
915902cb | 263 | int proto = ZEBRA_ROUTE_KERNEL; |
d62a17ae | 264 | int index = 0; |
265 | int table; | |
266 | int metric = 0; | |
d7c0a89a | 267 | uint32_t mtu = 0; |
25715c7e | 268 | uint8_t distance = 0; |
4e40b6d6 | 269 | route_tag_t tag = 0; |
d62a17ae | 270 | |
271 | void *dest = NULL; | |
272 | void *gate = NULL; | |
273 | void *prefsrc = NULL; /* IPv4 preferred source host address */ | |
274 | void *src = NULL; /* IPv6 srcdest source prefix */ | |
e655a03c | 275 | enum blackhole_type bh_type = BLACKHOLE_UNSPEC; |
d62a17ae | 276 | |
277 | rtm = NLMSG_DATA(h); | |
278 | ||
279 | if (startup && h->nlmsg_type != RTM_NEWROUTE) | |
280 | return 0; | |
e655a03c DL |
281 | switch (rtm->rtm_type) { |
282 | case RTN_UNICAST: | |
283 | break; | |
284 | case RTN_BLACKHOLE: | |
285 | bh_type = BLACKHOLE_NULL; | |
286 | break; | |
287 | case RTN_UNREACHABLE: | |
288 | bh_type = BLACKHOLE_REJECT; | |
289 | break; | |
290 | case RTN_PROHIBIT: | |
291 | bh_type = BLACKHOLE_ADMINPROHIB; | |
292 | break; | |
293 | default: | |
d62a17ae | 294 | return 0; |
e655a03c | 295 | } |
d62a17ae | 296 | |
297 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct rtmsg)); | |
298 | if (len < 0) | |
299 | return -1; | |
300 | ||
301 | memset(tb, 0, sizeof tb); | |
302 | netlink_parse_rtattr(tb, RTA_MAX, RTM_RTA(rtm), len); | |
303 | ||
304 | if (rtm->rtm_flags & RTM_F_CLONED) | |
305 | return 0; | |
306 | if (rtm->rtm_protocol == RTPROT_REDIRECT) | |
307 | return 0; | |
308 | if (rtm->rtm_protocol == RTPROT_KERNEL) | |
309 | return 0; | |
310 | ||
311 | if (!startup && is_selfroute(rtm->rtm_protocol) | |
6ab5222f DS |
312 | && h->nlmsg_type == RTM_NEWROUTE) { |
313 | if (IS_ZEBRA_DEBUG_KERNEL) | |
314 | zlog_debug("Route type: %d Received that we think we have originated, ignoring", | |
315 | rtm->rtm_protocol); | |
d62a17ae | 316 | return 0; |
6ab5222f | 317 | } |
d62a17ae | 318 | |
319 | /* We don't care about change notifications for the MPLS table. */ | |
320 | /* TODO: Revisit this. */ | |
321 | if (rtm->rtm_family == AF_MPLS) | |
322 | return 0; | |
323 | ||
324 | /* Table corresponding to route. */ | |
325 | if (tb[RTA_TABLE]) | |
326 | table = *(int *)RTA_DATA(tb[RTA_TABLE]); | |
327 | else | |
328 | table = rtm->rtm_table; | |
329 | ||
330 | /* Map to VRF */ | |
78dd30b2 | 331 | vrf_id = vrf_lookup_by_table(table, ns_id); |
d62a17ae | 332 | if (vrf_id == VRF_DEFAULT) { |
333 | if (!is_zebra_valid_kernel_table(table) | |
334 | && !is_zebra_main_routing_table(table)) | |
335 | return 0; | |
336 | } | |
337 | ||
338 | /* Route which inserted by Zebra. */ | |
915902cb | 339 | if (is_selfroute(rtm->rtm_protocol)) { |
d62a17ae | 340 | flags |= ZEBRA_FLAG_SELFROUTE; |
915902cb DS |
341 | proto = proto2zebra(rtm->rtm_protocol, rtm->rtm_family); |
342 | } | |
d62a17ae | 343 | if (tb[RTA_OIF]) |
344 | index = *(int *)RTA_DATA(tb[RTA_OIF]); | |
345 | ||
346 | if (tb[RTA_DST]) | |
347 | dest = RTA_DATA(tb[RTA_DST]); | |
348 | else | |
349 | dest = anyaddr; | |
350 | ||
351 | if (tb[RTA_SRC]) | |
352 | src = RTA_DATA(tb[RTA_SRC]); | |
353 | else | |
354 | src = anyaddr; | |
355 | ||
356 | if (tb[RTA_PREFSRC]) | |
357 | prefsrc = RTA_DATA(tb[RTA_PREFSRC]); | |
358 | ||
359 | if (tb[RTA_GATEWAY]) | |
360 | gate = RTA_DATA(tb[RTA_GATEWAY]); | |
361 | ||
f19435a8 DS |
362 | if (tb[RTA_PRIORITY]) |
363 | metric = *(int *)RTA_DATA(tb[RTA_PRIORITY]); | |
d62a17ae | 364 | |
4e40b6d6 KK |
365 | #if defined(SUPPORT_REALMS) |
366 | if (tb[RTA_FLOW]) | |
367 | tag = *(uint32_t *)RTA_DATA(tb[RTA_FLOW]); | |
368 | #endif | |
369 | ||
f19435a8 DS |
370 | if (tb[RTA_METRICS]) { |
371 | struct rtattr *mxrta[RTAX_MAX + 1]; | |
d62a17ae | 372 | |
f19435a8 | 373 | memset(mxrta, 0, sizeof mxrta); |
996c9314 | 374 | netlink_parse_rtattr(mxrta, RTAX_MAX, RTA_DATA(tb[RTA_METRICS]), |
f19435a8 | 375 | RTA_PAYLOAD(tb[RTA_METRICS])); |
d62a17ae | 376 | |
f19435a8 | 377 | if (mxrta[RTAX_MTU]) |
d7c0a89a | 378 | mtu = *(uint32_t *)RTA_DATA(mxrta[RTAX_MTU]); |
d62a17ae | 379 | } |
380 | ||
381 | if (rtm->rtm_family == AF_INET) { | |
382 | p.family = AF_INET; | |
383 | memcpy(&p.u.prefix4, dest, 4); | |
384 | p.prefixlen = rtm->rtm_dst_len; | |
385 | ||
386 | src_p.prefixlen = | |
387 | 0; // Forces debug below to not display anything | |
388 | } else if (rtm->rtm_family == AF_INET6) { | |
389 | p.family = AF_INET6; | |
390 | memcpy(&p.u.prefix6, dest, 16); | |
391 | p.prefixlen = rtm->rtm_dst_len; | |
392 | ||
393 | src_p.family = AF_INET6; | |
394 | memcpy(&src_p.prefix, src, 16); | |
395 | src_p.prefixlen = rtm->rtm_src_len; | |
396 | } | |
397 | ||
398 | if (rtm->rtm_src_len != 0) { | |
399 | char buf[PREFIX_STRLEN]; | |
400 | zlog_warn( | |
401 | "unsupported IPv[4|6] sourcedest route (dest %s vrf %u)", | |
402 | prefix2str(&p, buf, sizeof(buf)), vrf_id); | |
403 | return 0; | |
404 | } | |
405 | ||
25715c7e DS |
406 | /* |
407 | * For ZEBRA_ROUTE_KERNEL types: | |
408 | * | |
409 | * The metric/priority of the route received from the kernel | |
410 | * is a 32 bit number. We are going to interpret the high | |
411 | * order byte as the Admin Distance and the low order 3 bytes | |
412 | * as the metric. | |
413 | * | |
414 | * This will allow us to do two things: | |
415 | * 1) Allow the creation of kernel routes that can be | |
416 | * overridden by zebra. | |
417 | * 2) Allow the old behavior for 'most' kernel route types | |
418 | * if a user enters 'ip route ...' v4 routes get a metric | |
419 | * of 0 and v6 routes get a metric of 1024. Both of these | |
420 | * values will end up with a admin distance of 0, which | |
421 | * will cause them to win for the purposes of zebra. | |
422 | */ | |
423 | if (proto == ZEBRA_ROUTE_KERNEL) { | |
424 | distance = (metric >> 24) & 0xFF; | |
996c9314 | 425 | metric = (metric & 0x00FFFFFF); |
25715c7e DS |
426 | } |
427 | ||
d62a17ae | 428 | if (IS_ZEBRA_DEBUG_KERNEL) { |
429 | char buf[PREFIX_STRLEN]; | |
430 | char buf2[PREFIX_STRLEN]; | |
45df4e96 | 431 | zlog_debug("%s %s%s%s vrf %u(%u) metric: %d Admin Distance: %d", |
996c9314 LB |
432 | nl_msg_type_to_str(h->nlmsg_type), |
433 | prefix2str(&p, buf, sizeof(buf)), | |
434 | src_p.prefixlen ? " from " : "", | |
435 | src_p.prefixlen | |
436 | ? prefix2str(&src_p, buf2, sizeof(buf2)) | |
437 | : "", | |
45df4e96 | 438 | vrf_id, table, metric, distance); |
d62a17ae | 439 | } |
440 | ||
441 | afi_t afi = AFI_IP; | |
442 | if (rtm->rtm_family == AF_INET6) | |
443 | afi = AFI_IP6; | |
444 | ||
445 | if (h->nlmsg_type == RTM_NEWROUTE) { | |
8795f904 DS |
446 | struct interface *ifp; |
447 | vrf_id_t nh_vrf_id = vrf_id; | |
448 | ||
fd36be7e DL |
449 | if (!tb[RTA_MULTIPATH]) { |
450 | struct nexthop nh; | |
451 | size_t sz = (afi == AFI_IP) ? 4 : 16; | |
452 | ||
453 | memset(&nh, 0, sizeof(nh)); | |
af760ec1 DS |
454 | |
455 | if (bh_type == BLACKHOLE_UNSPEC) { | |
456 | if (index && !gate) | |
457 | nh.type = NEXTHOP_TYPE_IFINDEX; | |
458 | else if (index && gate) | |
996c9314 LB |
459 | nh.type = |
460 | (afi == AFI_IP) | |
461 | ? NEXTHOP_TYPE_IPV4_IFINDEX | |
462 | : NEXTHOP_TYPE_IPV6_IFINDEX; | |
af760ec1 | 463 | else if (!index && gate) |
1dca2eaa RW |
464 | nh.type = (afi == AFI_IP) |
465 | ? NEXTHOP_TYPE_IPV4 | |
466 | : NEXTHOP_TYPE_IPV6; | |
af760ec1 DS |
467 | else { |
468 | nh.type = NEXTHOP_TYPE_BLACKHOLE; | |
469 | nh.bh_type = bh_type; | |
470 | } | |
471 | } else { | |
fd36be7e | 472 | nh.type = NEXTHOP_TYPE_BLACKHOLE; |
e655a03c DL |
473 | nh.bh_type = bh_type; |
474 | } | |
fd36be7e DL |
475 | nh.ifindex = index; |
476 | if (prefsrc) | |
477 | memcpy(&nh.src, prefsrc, sz); | |
478 | if (gate) | |
479 | memcpy(&nh.gate, gate, sz); | |
915902cb | 480 | |
8795f904 | 481 | if (index) { |
996c9314 | 482 | ifp = if_lookup_by_index(index, VRF_UNKNOWN); |
8795f904 DS |
483 | if (ifp) |
484 | nh_vrf_id = ifp->vrf_id; | |
485 | } | |
4a7371e9 | 486 | nh.vrf_id = nh_vrf_id; |
8795f904 | 487 | |
4a7371e9 DS |
488 | rib_add(afi, SAFI_UNICAST, vrf_id, proto, 0, flags, &p, |
489 | NULL, &nh, table, metric, mtu, distance, tag); | |
fd36be7e | 490 | } else { |
d62a17ae | 491 | /* This is a multipath route */ |
492 | ||
493 | struct route_entry *re; | |
494 | struct rtnexthop *rtnh = | |
495 | (struct rtnexthop *)RTA_DATA(tb[RTA_MULTIPATH]); | |
496 | ||
497 | len = RTA_PAYLOAD(tb[RTA_MULTIPATH]); | |
498 | ||
499 | re = XCALLOC(MTYPE_RE, sizeof(struct route_entry)); | |
915902cb | 500 | re->type = proto; |
25715c7e | 501 | re->distance = distance; |
d62a17ae | 502 | re->flags = flags; |
503 | re->metric = metric; | |
504 | re->mtu = mtu; | |
505 | re->vrf_id = vrf_id; | |
506 | re->table = table; | |
507 | re->nexthop_num = 0; | |
508 | re->uptime = time(NULL); | |
4e40b6d6 | 509 | re->tag = tag; |
d62a17ae | 510 | |
511 | for (;;) { | |
4a7371e9 | 512 | vrf_id_t nh_vrf_id; |
d62a17ae | 513 | if (len < (int)sizeof(*rtnh) |
514 | || rtnh->rtnh_len > len) | |
515 | break; | |
516 | ||
517 | index = rtnh->rtnh_ifindex; | |
8795f904 DS |
518 | if (index) { |
519 | /* | |
520 | * Yes we are looking this up | |
521 | * for every nexthop and just | |
522 | * using the last one looked | |
523 | * up right now | |
524 | */ | |
525 | ifp = if_lookup_by_index(index, | |
526 | VRF_UNKNOWN); | |
527 | if (ifp) | |
4a7371e9 DS |
528 | nh_vrf_id = ifp->vrf_id; |
529 | else { | |
530 | zlog_warn( | |
531 | "%s: Unknown interface %u specified, defaulting to VRF_DEFAULT", | |
532 | __PRETTY_FUNCTION__, | |
533 | index); | |
534 | nh_vrf_id = VRF_DEFAULT; | |
535 | } | |
536 | } else | |
537 | nh_vrf_id = vrf_id; | |
538 | ||
d62a17ae | 539 | gate = 0; |
540 | if (rtnh->rtnh_len > sizeof(*rtnh)) { | |
541 | memset(tb, 0, sizeof(tb)); | |
542 | netlink_parse_rtattr( | |
543 | tb, RTA_MAX, RTNH_DATA(rtnh), | |
544 | rtnh->rtnh_len - sizeof(*rtnh)); | |
545 | if (tb[RTA_GATEWAY]) | |
546 | gate = RTA_DATA( | |
547 | tb[RTA_GATEWAY]); | |
548 | } | |
549 | ||
550 | if (gate) { | |
551 | if (rtm->rtm_family == AF_INET) { | |
552 | if (index) | |
553 | route_entry_nexthop_ipv4_ifindex_add( | |
554 | re, gate, | |
4a7371e9 DS |
555 | prefsrc, index, |
556 | nh_vrf_id); | |
d62a17ae | 557 | else |
558 | route_entry_nexthop_ipv4_add( | |
559 | re, gate, | |
4a7371e9 DS |
560 | prefsrc, |
561 | nh_vrf_id); | |
d62a17ae | 562 | } else if (rtm->rtm_family |
563 | == AF_INET6) { | |
564 | if (index) | |
565 | route_entry_nexthop_ipv6_ifindex_add( | |
4a7371e9 DS |
566 | re, gate, index, |
567 | nh_vrf_id); | |
d62a17ae | 568 | else |
569 | route_entry_nexthop_ipv6_add( | |
4a7371e9 DS |
570 | re, gate, |
571 | nh_vrf_id); | |
d62a17ae | 572 | } |
573 | } else | |
4a7371e9 DS |
574 | route_entry_nexthop_ifindex_add( |
575 | re, index, nh_vrf_id); | |
d62a17ae | 576 | |
577 | len -= NLMSG_ALIGN(rtnh->rtnh_len); | |
578 | rtnh = RTNH_NEXT(rtnh); | |
579 | } | |
580 | ||
581 | zserv_nexthop_num_warn(__func__, | |
582 | (const struct prefix *)&p, | |
583 | re->nexthop_num); | |
584 | if (re->nexthop_num == 0) | |
585 | XFREE(MTYPE_RE, re); | |
586 | else | |
996c9314 LB |
587 | rib_add_multipath(afi, SAFI_UNICAST, &p, NULL, |
588 | re); | |
d62a17ae | 589 | } |
590 | } else { | |
fd36be7e DL |
591 | if (!tb[RTA_MULTIPATH]) { |
592 | struct nexthop nh; | |
593 | size_t sz = (afi == AFI_IP) ? 4 : 16; | |
594 | ||
595 | memset(&nh, 0, sizeof(nh)); | |
8ba5bd58 RW |
596 | if (bh_type == BLACKHOLE_UNSPEC) { |
597 | if (index && !gate) | |
598 | nh.type = NEXTHOP_TYPE_IFINDEX; | |
599 | else if (index && gate) | |
600 | nh.type = | |
601 | (afi == AFI_IP) | |
602 | ? NEXTHOP_TYPE_IPV4_IFINDEX | |
603 | : NEXTHOP_TYPE_IPV6_IFINDEX; | |
604 | else if (!index && gate) | |
605 | nh.type = (afi == AFI_IP) | |
606 | ? NEXTHOP_TYPE_IPV4 | |
60466a63 | 607 | : NEXTHOP_TYPE_IPV6; |
8ba5bd58 RW |
608 | else { |
609 | nh.type = NEXTHOP_TYPE_BLACKHOLE; | |
610 | nh.bh_type = BLACKHOLE_UNSPEC; | |
611 | } | |
612 | } else { | |
fd36be7e | 613 | nh.type = NEXTHOP_TYPE_BLACKHOLE; |
8ba5bd58 RW |
614 | nh.bh_type = bh_type; |
615 | } | |
fd36be7e DL |
616 | nh.ifindex = index; |
617 | if (gate) | |
618 | memcpy(&nh.gate, gate, sz); | |
996c9314 | 619 | rib_delete(afi, SAFI_UNICAST, vrf_id, proto, 0, flags, |
22e63104 | 620 | &p, NULL, &nh, table, metric, true); |
fd36be7e DL |
621 | } else { |
622 | /* XXX: need to compare the entire list of nexthops | |
623 | * here for NLM_F_APPEND stupidity */ | |
996c9314 | 624 | rib_delete(afi, SAFI_UNICAST, vrf_id, proto, 0, flags, |
22e63104 | 625 | &p, NULL, NULL, table, metric, true); |
d62a17ae | 626 | } |
627 | } | |
628 | ||
629 | return 0; | |
718e3744 | 630 | } |
631 | ||
e3be0432 DS |
632 | static struct mcast_route_data *mroute = NULL; |
633 | ||
2414abd3 | 634 | static int netlink_route_change_read_multicast(struct nlmsghdr *h, |
d62a17ae | 635 | ns_id_t ns_id, int startup) |
565fdc75 | 636 | { |
d62a17ae | 637 | int len; |
638 | struct rtmsg *rtm; | |
639 | struct rtattr *tb[RTA_MAX + 1]; | |
640 | struct mcast_route_data *m; | |
641 | struct mcast_route_data mr; | |
642 | int iif = 0; | |
643 | int count; | |
644 | int oif[256]; | |
645 | int oif_count = 0; | |
646 | char sbuf[40]; | |
647 | char gbuf[40]; | |
648 | char oif_list[256] = "\0"; | |
78dd30b2 | 649 | vrf_id_t vrf; |
43b5cc5e | 650 | int table; |
d62a17ae | 651 | |
652 | if (mroute) | |
653 | m = mroute; | |
654 | else { | |
655 | memset(&mr, 0, sizeof(mr)); | |
656 | m = &mr; | |
657 | } | |
658 | ||
659 | rtm = NLMSG_DATA(h); | |
660 | ||
661 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct rtmsg)); | |
662 | ||
663 | memset(tb, 0, sizeof tb); | |
664 | netlink_parse_rtattr(tb, RTA_MAX, RTM_RTA(rtm), len); | |
90d82769 | 665 | |
43b5cc5e DS |
666 | if (tb[RTA_TABLE]) |
667 | table = *(int *)RTA_DATA(tb[RTA_TABLE]); | |
668 | else | |
669 | table = rtm->rtm_table; | |
670 | ||
78dd30b2 | 671 | vrf = vrf_lookup_by_table(table, ns_id); |
43b5cc5e | 672 | |
d62a17ae | 673 | if (tb[RTA_IIF]) |
674 | iif = *(int *)RTA_DATA(tb[RTA_IIF]); | |
675 | ||
676 | if (tb[RTA_SRC]) | |
bd8b9272 | 677 | m->sg.src = *(struct in_addr *)RTA_DATA(tb[RTA_SRC]); |
d62a17ae | 678 | |
679 | if (tb[RTA_DST]) | |
bd8b9272 | 680 | m->sg.grp = *(struct in_addr *)RTA_DATA(tb[RTA_DST]); |
d62a17ae | 681 | |
682 | if ((RTA_EXPIRES <= RTA_MAX) && tb[RTA_EXPIRES]) | |
683 | m->lastused = *(unsigned long long *)RTA_DATA(tb[RTA_EXPIRES]); | |
684 | ||
685 | if (tb[RTA_MULTIPATH]) { | |
686 | struct rtnexthop *rtnh = | |
687 | (struct rtnexthop *)RTA_DATA(tb[RTA_MULTIPATH]); | |
688 | ||
689 | len = RTA_PAYLOAD(tb[RTA_MULTIPATH]); | |
690 | for (;;) { | |
691 | if (len < (int)sizeof(*rtnh) || rtnh->rtnh_len > len) | |
692 | break; | |
693 | ||
694 | oif[oif_count] = rtnh->rtnh_ifindex; | |
695 | oif_count++; | |
696 | ||
697 | len -= NLMSG_ALIGN(rtnh->rtnh_len); | |
698 | rtnh = RTNH_NEXT(rtnh); | |
699 | } | |
700 | } | |
701 | ||
702 | if (IS_ZEBRA_DEBUG_KERNEL) { | |
703 | struct interface *ifp; | |
0af35d90 RW |
704 | strlcpy(sbuf, inet_ntoa(m->sg.src), sizeof(sbuf)); |
705 | strlcpy(gbuf, inet_ntoa(m->sg.grp), sizeof(gbuf)); | |
d62a17ae | 706 | for (count = 0; count < oif_count; count++) { |
707 | ifp = if_lookup_by_index(oif[count], vrf); | |
708 | char temp[256]; | |
709 | ||
710 | sprintf(temp, "%s ", ifp->name); | |
711 | strcat(oif_list, temp); | |
712 | } | |
43b5cc5e | 713 | struct zebra_vrf *zvrf = zebra_vrf_lookup_by_id(vrf); |
d62a17ae | 714 | ifp = if_lookup_by_index(iif, vrf); |
43b5cc5e DS |
715 | zlog_debug( |
716 | "MCAST VRF: %s(%d) %s (%s,%s) IIF: %s OIF: %s jiffies: %lld", | |
717 | zvrf->vrf->name, vrf, nl_msg_type_to_str(h->nlmsg_type), | |
718 | sbuf, gbuf, ifp->name, oif_list, m->lastused); | |
90d82769 | 719 | } |
d62a17ae | 720 | return 0; |
565fdc75 DS |
721 | } |
722 | ||
2414abd3 | 723 | int netlink_route_change(struct nlmsghdr *h, ns_id_t ns_id, int startup) |
565fdc75 | 724 | { |
d62a17ae | 725 | int len; |
d62a17ae | 726 | struct rtmsg *rtm; |
727 | ||
728 | rtm = NLMSG_DATA(h); | |
729 | ||
730 | if (!(h->nlmsg_type == RTM_NEWROUTE || h->nlmsg_type == RTM_DELROUTE)) { | |
731 | /* If this is not route add/delete message print warning. */ | |
78dd30b2 | 732 | zlog_warn("Kernel message: %d NS %u\n", h->nlmsg_type, ns_id); |
d62a17ae | 733 | return 0; |
734 | } | |
735 | ||
736 | /* Connected route. */ | |
737 | if (IS_ZEBRA_DEBUG_KERNEL) | |
78dd30b2 | 738 | zlog_debug("%s %s %s proto %s NS %u", |
d62a17ae | 739 | nl_msg_type_to_str(h->nlmsg_type), |
740 | nl_family_to_str(rtm->rtm_family), | |
741 | nl_rttype_to_str(rtm->rtm_type), | |
78dd30b2 | 742 | nl_rtproto_to_str(rtm->rtm_protocol), ns_id); |
d62a17ae | 743 | |
744 | /* We don't care about change notifications for the MPLS table. */ | |
745 | /* TODO: Revisit this. */ | |
746 | if (rtm->rtm_family == AF_MPLS) | |
747 | return 0; | |
748 | ||
749 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct rtmsg)); | |
750 | if (len < 0) | |
751 | return -1; | |
752 | ||
e655a03c | 753 | if (rtm->rtm_type == RTN_MULTICAST) |
2414abd3 | 754 | netlink_route_change_read_multicast(h, ns_id, startup); |
e655a03c | 755 | else |
2414abd3 | 756 | netlink_route_change_read_unicast(h, ns_id, startup); |
d62a17ae | 757 | return 0; |
565fdc75 DS |
758 | } |
759 | ||
289602d7 | 760 | /* Request for specific route information from the kernel */ |
d62a17ae | 761 | static int netlink_request_route(struct zebra_ns *zns, int family, int type) |
289602d7 | 762 | { |
d62a17ae | 763 | struct { |
764 | struct nlmsghdr n; | |
765 | struct rtmsg rtm; | |
766 | } req; | |
767 | ||
768 | /* Form the request, specifying filter (rtattr) if needed. */ | |
769 | memset(&req, 0, sizeof(req)); | |
770 | req.n.nlmsg_type = type; | |
771 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg)); | |
772 | req.rtm.rtm_family = family; | |
773 | ||
774 | return netlink_request(&zns->netlink_cmd, &req.n); | |
289602d7 | 775 | } |
776 | ||
718e3744 | 777 | /* Routing table read function using netlink interface. Only called |
778 | bootstrap time. */ | |
d62a17ae | 779 | int netlink_route_read(struct zebra_ns *zns) |
718e3744 | 780 | { |
d62a17ae | 781 | int ret; |
782 | ||
783 | /* Get IPv4 routing table. */ | |
784 | ret = netlink_request_route(zns, AF_INET, RTM_GETROUTE); | |
785 | if (ret < 0) | |
786 | return ret; | |
787 | ret = netlink_parse_info(netlink_route_change_read_unicast, | |
788 | &zns->netlink_cmd, zns, 0, 1); | |
789 | if (ret < 0) | |
790 | return ret; | |
791 | ||
792 | /* Get IPv6 routing table. */ | |
793 | ret = netlink_request_route(zns, AF_INET6, RTM_GETROUTE); | |
794 | if (ret < 0) | |
795 | return ret; | |
796 | ret = netlink_parse_info(netlink_route_change_read_unicast, | |
797 | &zns->netlink_cmd, zns, 0, 1); | |
798 | if (ret < 0) | |
799 | return ret; | |
800 | ||
801 | return 0; | |
718e3744 | 802 | } |
803 | ||
d7c0a89a QY |
804 | static void _netlink_route_nl_add_gateway_info(uint8_t route_family, |
805 | uint8_t gw_family, | |
d62a17ae | 806 | struct nlmsghdr *nlmsg, |
807 | size_t req_size, int bytelen, | |
808 | struct nexthop *nexthop) | |
40c7bdb0 | 809 | { |
d62a17ae | 810 | if (route_family == AF_MPLS) { |
811 | struct gw_family_t gw_fam; | |
812 | ||
813 | gw_fam.family = gw_family; | |
814 | if (gw_family == AF_INET) | |
815 | memcpy(&gw_fam.gate.ipv4, &nexthop->gate.ipv4, bytelen); | |
816 | else | |
817 | memcpy(&gw_fam.gate.ipv6, &nexthop->gate.ipv6, bytelen); | |
818 | addattr_l(nlmsg, req_size, RTA_VIA, &gw_fam.family, | |
819 | bytelen + 2); | |
820 | } else { | |
821 | if (gw_family == AF_INET) | |
822 | addattr_l(nlmsg, req_size, RTA_GATEWAY, | |
823 | &nexthop->gate.ipv4, bytelen); | |
824 | else | |
825 | addattr_l(nlmsg, req_size, RTA_GATEWAY, | |
826 | &nexthop->gate.ipv6, bytelen); | |
827 | } | |
40c7bdb0 | 828 | } |
829 | ||
d7c0a89a QY |
830 | static void _netlink_route_rta_add_gateway_info(uint8_t route_family, |
831 | uint8_t gw_family, | |
d62a17ae | 832 | struct rtattr *rta, |
833 | struct rtnexthop *rtnh, | |
834 | size_t req_size, int bytelen, | |
835 | struct nexthop *nexthop) | |
40c7bdb0 | 836 | { |
d62a17ae | 837 | if (route_family == AF_MPLS) { |
838 | struct gw_family_t gw_fam; | |
839 | ||
840 | gw_fam.family = gw_family; | |
841 | if (gw_family == AF_INET) | |
842 | memcpy(&gw_fam.gate.ipv4, &nexthop->gate.ipv4, bytelen); | |
843 | else | |
844 | memcpy(&gw_fam.gate.ipv6, &nexthop->gate.ipv6, bytelen); | |
845 | rta_addattr_l(rta, req_size, RTA_VIA, &gw_fam.family, | |
846 | bytelen + 2); | |
847 | rtnh->rtnh_len += RTA_LENGTH(bytelen + 2); | |
848 | } else { | |
849 | if (gw_family == AF_INET) | |
850 | rta_addattr_l(rta, req_size, RTA_GATEWAY, | |
851 | &nexthop->gate.ipv4, bytelen); | |
852 | else | |
853 | rta_addattr_l(rta, req_size, RTA_GATEWAY, | |
854 | &nexthop->gate.ipv6, bytelen); | |
855 | rtnh->rtnh_len += sizeof(struct rtattr) + bytelen; | |
856 | } | |
40c7bdb0 | 857 | } |
858 | ||
fa713d9e CF |
859 | /* This function takes a nexthop as argument and adds |
860 | * the appropriate netlink attributes to an existing | |
861 | * netlink message. | |
862 | * | |
863 | * @param routedesc: Human readable description of route type | |
864 | * (direct/recursive, single-/multipath) | |
865 | * @param bytelen: Length of addresses in bytes. | |
866 | * @param nexthop: Nexthop information | |
867 | * @param nlmsg: nlmsghdr structure to fill in. | |
868 | * @param req_size: The size allocated for the message. | |
869 | */ | |
d62a17ae | 870 | static void _netlink_route_build_singlepath(const char *routedesc, int bytelen, |
871 | struct nexthop *nexthop, | |
872 | struct nlmsghdr *nlmsg, | |
873 | struct rtmsg *rtmsg, | |
874 | size_t req_size, int cmd) | |
fa713d9e | 875 | { |
8ecdb26e | 876 | struct mpls_label_stack *nh_label; |
d62a17ae | 877 | mpls_lse_t out_lse[MPLS_MAX_LABELS]; |
fa712963 | 878 | int num_labels = 0; |
9a62e84b | 879 | char label_buf[256]; |
d62a17ae | 880 | |
881 | /* | |
882 | * label_buf is *only* currently used within debugging. | |
883 | * As such when we assign it we are guarding it inside | |
884 | * a debug test. If you want to change this make sure | |
885 | * you fix this assumption | |
886 | */ | |
887 | label_buf[0] = '\0'; | |
d62a17ae | 888 | |
fa712963 RW |
889 | assert(nexthop); |
890 | for (struct nexthop *nh = nexthop; nh; nh = nh->rparent) { | |
d62a17ae | 891 | char label_buf1[20]; |
892 | ||
fa712963 RW |
893 | nh_label = nh->nh_label; |
894 | if (!nh_label || !nh_label->num_labels) | |
895 | continue; | |
896 | ||
897 | for (int i = 0; i < nh_label->num_labels; i++) { | |
898 | if (nh_label->label[i] == MPLS_LABEL_IMPLICIT_NULL) | |
899 | continue; | |
900 | ||
901 | if (IS_ZEBRA_DEBUG_KERNEL) { | |
902 | if (!num_labels) | |
903 | sprintf(label_buf, "label %u", | |
904 | nh_label->label[i]); | |
905 | else { | |
906 | sprintf(label_buf1, "/%u", | |
907 | nh_label->label[i]); | |
908 | strlcat(label_buf, label_buf1, | |
909 | sizeof(label_buf)); | |
d62a17ae | 910 | } |
d62a17ae | 911 | } |
fa712963 RW |
912 | |
913 | out_lse[num_labels] = | |
914 | mpls_lse_encode(nh_label->label[i], 0, 0, 0); | |
915 | num_labels++; | |
d62a17ae | 916 | } |
fa712963 RW |
917 | } |
918 | ||
919 | if (num_labels) { | |
920 | /* Set the BoS bit */ | |
921 | out_lse[num_labels - 1] |= htonl(1 << MPLS_LS_S_SHIFT); | |
922 | ||
923 | if (rtmsg->rtm_family == AF_MPLS) | |
924 | addattr_l(nlmsg, req_size, RTA_NEWDST, &out_lse, | |
925 | num_labels * sizeof(mpls_lse_t)); | |
926 | else { | |
927 | struct rtattr *nest; | |
d7c0a89a | 928 | uint16_t encap = LWTUNNEL_ENCAP_MPLS; |
fa712963 RW |
929 | |
930 | addattr_l(nlmsg, req_size, RTA_ENCAP_TYPE, &encap, | |
d7c0a89a | 931 | sizeof(uint16_t)); |
fa712963 RW |
932 | nest = addattr_nest(nlmsg, req_size, RTA_ENCAP); |
933 | addattr_l(nlmsg, req_size, MPLS_IPTUNNEL_DST, &out_lse, | |
934 | num_labels * sizeof(mpls_lse_t)); | |
935 | addattr_nest_end(nlmsg, nest); | |
66d42727 | 936 | } |
0aabccc0 | 937 | } |
fa713d9e | 938 | |
d62a17ae | 939 | if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ONLINK)) |
940 | rtmsg->rtm_flags |= RTNH_F_ONLINK; | |
941 | ||
942 | if (rtmsg->rtm_family == AF_INET | |
943 | && (nexthop->type == NEXTHOP_TYPE_IPV6 | |
944 | || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX)) { | |
945 | rtmsg->rtm_flags |= RTNH_F_ONLINK; | |
946 | addattr_l(nlmsg, req_size, RTA_GATEWAY, &ipv4_ll, 4); | |
947 | addattr32(nlmsg, req_size, RTA_OIF, nexthop->ifindex); | |
948 | ||
949 | if (nexthop->rmap_src.ipv4.s_addr && (cmd == RTM_NEWROUTE)) | |
950 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
951 | &nexthop->rmap_src.ipv4, bytelen); | |
952 | else if (nexthop->src.ipv4.s_addr && (cmd == RTM_NEWROUTE)) | |
953 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
954 | &nexthop->src.ipv4, bytelen); | |
955 | ||
956 | if (IS_ZEBRA_DEBUG_KERNEL) | |
957 | zlog_debug( | |
958 | " 5549: _netlink_route_build_singlepath() (%s): " | |
7556c3fd | 959 | "nexthop via %s %s if %u(%u)", |
d62a17ae | 960 | routedesc, ipv4_ll_buf, label_buf, |
7556c3fd | 961 | nexthop->ifindex, nexthop->vrf_id); |
d62a17ae | 962 | return; |
0aabccc0 DD |
963 | } |
964 | ||
d62a17ae | 965 | if (nexthop->type == NEXTHOP_TYPE_IPV4 |
966 | || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX) { | |
967 | /* Send deletes to the kernel without specifying the next-hop */ | |
968 | if (cmd != RTM_DELROUTE) | |
969 | _netlink_route_nl_add_gateway_info( | |
970 | rtmsg->rtm_family, AF_INET, nlmsg, req_size, | |
971 | bytelen, nexthop); | |
972 | ||
973 | if (cmd == RTM_NEWROUTE) { | |
974 | if (nexthop->rmap_src.ipv4.s_addr) | |
975 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
976 | &nexthop->rmap_src.ipv4, bytelen); | |
977 | else if (nexthop->src.ipv4.s_addr) | |
978 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
979 | &nexthop->src.ipv4, bytelen); | |
980 | } | |
981 | ||
982 | if (IS_ZEBRA_DEBUG_KERNEL) | |
983 | zlog_debug( | |
984 | "netlink_route_multipath() (%s): " | |
7556c3fd | 985 | "nexthop via %s %s if %u(%u)", |
d62a17ae | 986 | routedesc, inet_ntoa(nexthop->gate.ipv4), |
7556c3fd | 987 | label_buf, nexthop->ifindex, nexthop->vrf_id); |
0aabccc0 | 988 | } |
fa713d9e | 989 | |
d62a17ae | 990 | if (nexthop->type == NEXTHOP_TYPE_IPV6 |
991 | || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) { | |
992 | _netlink_route_nl_add_gateway_info(rtmsg->rtm_family, AF_INET6, | |
993 | nlmsg, req_size, bytelen, | |
994 | nexthop); | |
995 | ||
996 | if (cmd == RTM_NEWROUTE) { | |
997 | if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->rmap_src.ipv6)) | |
998 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
999 | &nexthop->rmap_src.ipv6, bytelen); | |
1000 | else if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->src.ipv6)) | |
1001 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1002 | &nexthop->src.ipv6, bytelen); | |
1003 | } | |
fa713d9e | 1004 | |
d62a17ae | 1005 | if (IS_ZEBRA_DEBUG_KERNEL) |
1006 | zlog_debug( | |
1007 | "netlink_route_multipath() (%s): " | |
7556c3fd | 1008 | "nexthop via %s %s if %u(%u)", |
d62a17ae | 1009 | routedesc, inet6_ntoa(nexthop->gate.ipv6), |
7556c3fd | 1010 | label_buf, nexthop->ifindex, nexthop->vrf_id); |
d62a17ae | 1011 | } |
5e210522 DS |
1012 | |
1013 | /* | |
1014 | * We have the ifindex so we should always send it | |
1015 | * This is especially useful if we are doing route | |
1016 | * leaking. | |
1017 | */ | |
1018 | if (nexthop->type != NEXTHOP_TYPE_BLACKHOLE) | |
d62a17ae | 1019 | addattr32(nlmsg, req_size, RTA_OIF, nexthop->ifindex); |
1020 | ||
5e210522 DS |
1021 | if (nexthop->type == NEXTHOP_TYPE_IFINDEX |
1022 | || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX) { | |
d62a17ae | 1023 | if (cmd == RTM_NEWROUTE) { |
1024 | if (nexthop->rmap_src.ipv4.s_addr) | |
1025 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1026 | &nexthop->rmap_src.ipv4, bytelen); | |
1027 | else if (nexthop->src.ipv4.s_addr) | |
1028 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1029 | &nexthop->src.ipv4, bytelen); | |
1030 | } | |
fa713d9e | 1031 | |
d62a17ae | 1032 | if (IS_ZEBRA_DEBUG_KERNEL) |
1033 | zlog_debug( | |
1034 | "netlink_route_multipath() (%s): " | |
7556c3fd DS |
1035 | "nexthop via if %u(%u)", |
1036 | routedesc, nexthop->ifindex, nexthop->vrf_id); | |
0aabccc0 DD |
1037 | } |
1038 | ||
d62a17ae | 1039 | if (nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) { |
d62a17ae | 1040 | if (cmd == RTM_NEWROUTE) { |
1041 | if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->rmap_src.ipv6)) | |
1042 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1043 | &nexthop->rmap_src.ipv6, bytelen); | |
1044 | else if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->src.ipv6)) | |
1045 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1046 | &nexthop->src.ipv6, bytelen); | |
1047 | } | |
1048 | ||
1049 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1050 | zlog_debug( | |
1051 | "netlink_route_multipath() (%s): " | |
7556c3fd DS |
1052 | "nexthop via if %u(%u)", |
1053 | routedesc, nexthop->ifindex, nexthop->vrf_id); | |
d62a17ae | 1054 | } |
fa713d9e CF |
1055 | } |
1056 | ||
1057 | /* This function takes a nexthop as argument and | |
1058 | * appends to the given rtattr/rtnexthop pair the | |
1059 | * representation of the nexthop. If the nexthop | |
1060 | * defines a preferred source, the src parameter | |
1061 | * will be modified to point to that src, otherwise | |
1062 | * it will be kept unmodified. | |
1063 | * | |
1064 | * @param routedesc: Human readable description of route type | |
1065 | * (direct/recursive, single-/multipath) | |
1066 | * @param bytelen: Length of addresses in bytes. | |
1067 | * @param nexthop: Nexthop information | |
1068 | * @param rta: rtnetlink attribute structure | |
1069 | * @param rtnh: pointer to an rtnetlink nexthop structure | |
1070 | * @param src: pointer pointing to a location where | |
1071 | * the prefsrc should be stored. | |
1072 | */ | |
d62a17ae | 1073 | static void _netlink_route_build_multipath(const char *routedesc, int bytelen, |
1074 | struct nexthop *nexthop, | |
1075 | struct rtattr *rta, | |
1076 | struct rtnexthop *rtnh, | |
1077 | struct rtmsg *rtmsg, | |
1078 | union g_addr **src) | |
fa713d9e | 1079 | { |
8ecdb26e | 1080 | struct mpls_label_stack *nh_label; |
d62a17ae | 1081 | mpls_lse_t out_lse[MPLS_MAX_LABELS]; |
fa712963 | 1082 | int num_labels = 0; |
9a62e84b | 1083 | char label_buf[256]; |
d62a17ae | 1084 | |
1085 | rtnh->rtnh_len = sizeof(*rtnh); | |
1086 | rtnh->rtnh_flags = 0; | |
1087 | rtnh->rtnh_hops = 0; | |
1088 | rta->rta_len += rtnh->rtnh_len; | |
1089 | ||
1090 | /* | |
1091 | * label_buf is *only* currently used within debugging. | |
1092 | * As such when we assign it we are guarding it inside | |
1093 | * a debug test. If you want to change this make sure | |
1094 | * you fix this assumption | |
1095 | */ | |
1096 | label_buf[0] = '\0'; | |
d62a17ae | 1097 | |
fa712963 RW |
1098 | assert(nexthop); |
1099 | for (struct nexthop *nh = nexthop; nh; nh = nh->rparent) { | |
d62a17ae | 1100 | char label_buf1[20]; |
1101 | ||
fa712963 RW |
1102 | nh_label = nh->nh_label; |
1103 | if (!nh_label || !nh_label->num_labels) | |
1104 | continue; | |
1105 | ||
1106 | for (int i = 0; i < nh_label->num_labels; i++) { | |
1107 | if (nh_label->label[i] == MPLS_LABEL_IMPLICIT_NULL) | |
1108 | continue; | |
1109 | ||
1110 | if (IS_ZEBRA_DEBUG_KERNEL) { | |
1111 | if (!num_labels) | |
1112 | sprintf(label_buf, "label %u", | |
1113 | nh_label->label[i]); | |
1114 | else { | |
1115 | sprintf(label_buf1, "/%u", | |
1116 | nh_label->label[i]); | |
1117 | strlcat(label_buf, label_buf1, | |
1118 | sizeof(label_buf)); | |
d62a17ae | 1119 | } |
d62a17ae | 1120 | } |
fa712963 RW |
1121 | |
1122 | out_lse[num_labels] = | |
1123 | mpls_lse_encode(nh_label->label[i], 0, 0, 0); | |
1124 | num_labels++; | |
d62a17ae | 1125 | } |
fa712963 RW |
1126 | } |
1127 | ||
1128 | if (num_labels) { | |
1129 | /* Set the BoS bit */ | |
1130 | out_lse[num_labels - 1] |= htonl(1 << MPLS_LS_S_SHIFT); | |
1131 | ||
1132 | if (rtmsg->rtm_family == AF_MPLS) { | |
1133 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, RTA_NEWDST, | |
1134 | &out_lse, | |
1135 | num_labels * sizeof(mpls_lse_t)); | |
1136 | rtnh->rtnh_len += | |
1137 | RTA_LENGTH(num_labels * sizeof(mpls_lse_t)); | |
1138 | } else { | |
1139 | struct rtattr *nest; | |
d7c0a89a | 1140 | uint16_t encap = LWTUNNEL_ENCAP_MPLS; |
fa712963 RW |
1141 | int len = rta->rta_len; |
1142 | ||
1143 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, RTA_ENCAP_TYPE, | |
d7c0a89a | 1144 | &encap, sizeof(uint16_t)); |
fa712963 RW |
1145 | nest = rta_nest(rta, NL_PKT_BUF_SIZE, RTA_ENCAP); |
1146 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, MPLS_IPTUNNEL_DST, | |
1147 | &out_lse, | |
1148 | num_labels * sizeof(mpls_lse_t)); | |
1149 | rta_nest_end(rta, nest); | |
1150 | rtnh->rtnh_len += rta->rta_len - len; | |
66d42727 | 1151 | } |
d62a17ae | 1152 | } |
1153 | ||
1154 | if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ONLINK)) | |
1155 | rtnh->rtnh_flags |= RTNH_F_ONLINK; | |
1156 | ||
1157 | if (rtmsg->rtm_family == AF_INET | |
1158 | && (nexthop->type == NEXTHOP_TYPE_IPV6 | |
1159 | || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX)) { | |
1160 | bytelen = 4; | |
1161 | rtnh->rtnh_flags |= RTNH_F_ONLINK; | |
1162 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, RTA_GATEWAY, &ipv4_ll, | |
1163 | bytelen); | |
1164 | rtnh->rtnh_len += sizeof(struct rtattr) + bytelen; | |
1165 | rtnh->rtnh_ifindex = nexthop->ifindex; | |
1166 | ||
1167 | if (nexthop->rmap_src.ipv4.s_addr) | |
1168 | *src = &nexthop->rmap_src; | |
1169 | else if (nexthop->src.ipv4.s_addr) | |
1170 | *src = &nexthop->src; | |
1171 | ||
1172 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1173 | zlog_debug( | |
1174 | " 5549: netlink_route_build_multipath() (%s): " | |
1175 | "nexthop via %s %s if %u", | |
1176 | routedesc, ipv4_ll_buf, label_buf, | |
1177 | nexthop->ifindex); | |
1178 | return; | |
1179 | } | |
1180 | ||
1181 | if (nexthop->type == NEXTHOP_TYPE_IPV4 | |
1182 | || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX) { | |
1183 | _netlink_route_rta_add_gateway_info(rtmsg->rtm_family, AF_INET, | |
1184 | rta, rtnh, NL_PKT_BUF_SIZE, | |
1185 | bytelen, nexthop); | |
1186 | if (nexthop->rmap_src.ipv4.s_addr) | |
1187 | *src = &nexthop->rmap_src; | |
1188 | else if (nexthop->src.ipv4.s_addr) | |
1189 | *src = &nexthop->src; | |
1190 | ||
1191 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1192 | zlog_debug( | |
1193 | "netlink_route_multipath() (%s): " | |
1194 | "nexthop via %s %s if %u", | |
1195 | routedesc, inet_ntoa(nexthop->gate.ipv4), | |
1196 | label_buf, nexthop->ifindex); | |
1197 | } | |
1198 | if (nexthop->type == NEXTHOP_TYPE_IPV6 | |
1199 | || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) { | |
1200 | _netlink_route_rta_add_gateway_info(rtmsg->rtm_family, AF_INET6, | |
1201 | rta, rtnh, NL_PKT_BUF_SIZE, | |
1202 | bytelen, nexthop); | |
1203 | ||
1204 | if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->rmap_src.ipv6)) | |
1205 | *src = &nexthop->rmap_src; | |
1206 | else if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->src.ipv6)) | |
1207 | *src = &nexthop->src; | |
1208 | ||
1209 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1210 | zlog_debug( | |
1211 | "netlink_route_multipath() (%s): " | |
1212 | "nexthop via %s %s if %u", | |
1213 | routedesc, inet6_ntoa(nexthop->gate.ipv6), | |
1214 | label_buf, nexthop->ifindex); | |
1215 | } | |
5e210522 DS |
1216 | |
1217 | /* | |
1218 | * We have figured out the ifindex so we should always send it | |
1219 | * This is especially useful if we are doing route | |
1220 | * leaking. | |
1221 | */ | |
1222 | if (nexthop->type != NEXTHOP_TYPE_BLACKHOLE) | |
1223 | rtnh->rtnh_ifindex = nexthop->ifindex; | |
1224 | ||
d62a17ae | 1225 | /* ifindex */ |
1226 | if (nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX | |
1227 | || nexthop->type == NEXTHOP_TYPE_IFINDEX) { | |
d62a17ae | 1228 | if (nexthop->rmap_src.ipv4.s_addr) |
1229 | *src = &nexthop->rmap_src; | |
1230 | else if (nexthop->src.ipv4.s_addr) | |
1231 | *src = &nexthop->src; | |
1232 | ||
1233 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1234 | zlog_debug( | |
1235 | "netlink_route_multipath() (%s): " | |
1236 | "nexthop via if %u", | |
1237 | routedesc, nexthop->ifindex); | |
d62a17ae | 1238 | } |
fa713d9e CF |
1239 | } |
1240 | ||
d62a17ae | 1241 | static inline void _netlink_mpls_build_singlepath(const char *routedesc, |
1242 | zebra_nhlfe_t *nhlfe, | |
1243 | struct nlmsghdr *nlmsg, | |
1244 | struct rtmsg *rtmsg, | |
1245 | size_t req_size, int cmd) | |
40c7bdb0 | 1246 | { |
d62a17ae | 1247 | int bytelen; |
d7c0a89a | 1248 | uint8_t family; |
40c7bdb0 | 1249 | |
d62a17ae | 1250 | family = NHLFE_FAMILY(nhlfe); |
1251 | bytelen = (family == AF_INET ? 4 : 16); | |
1252 | _netlink_route_build_singlepath(routedesc, bytelen, nhlfe->nexthop, | |
1253 | nlmsg, rtmsg, req_size, cmd); | |
40c7bdb0 | 1254 | } |
1255 | ||
1256 | ||
1257 | static inline void | |
d62a17ae | 1258 | _netlink_mpls_build_multipath(const char *routedesc, zebra_nhlfe_t *nhlfe, |
1259 | struct rtattr *rta, struct rtnexthop *rtnh, | |
1260 | struct rtmsg *rtmsg, union g_addr **src) | |
40c7bdb0 | 1261 | { |
d62a17ae | 1262 | int bytelen; |
d7c0a89a | 1263 | uint8_t family; |
40c7bdb0 | 1264 | |
d62a17ae | 1265 | family = NHLFE_FAMILY(nhlfe); |
1266 | bytelen = (family == AF_INET ? 4 : 16); | |
1267 | _netlink_route_build_multipath(routedesc, bytelen, nhlfe->nexthop, rta, | |
1268 | rtnh, rtmsg, src); | |
40c7bdb0 | 1269 | } |
1270 | ||
1271 | ||
fa713d9e CF |
1272 | /* Log debug information for netlink_route_multipath |
1273 | * if debug logging is enabled. | |
1274 | * | |
1275 | * @param cmd: Netlink command which is to be processed | |
1276 | * @param p: Prefix for which the change is due | |
fa713d9e | 1277 | * @param family: Address family which the change concerns |
45df4e96 DS |
1278 | * @param zvrf: The vrf we are in |
1279 | * @param tableid: The table we are working on | |
fa713d9e | 1280 | */ |
d62a17ae | 1281 | static void _netlink_route_debug(int cmd, struct prefix *p, |
7556c3fd DS |
1282 | int family, struct zebra_vrf *zvrf, |
1283 | uint32_t tableid) | |
fa713d9e | 1284 | { |
d62a17ae | 1285 | if (IS_ZEBRA_DEBUG_KERNEL) { |
1286 | char buf[PREFIX_STRLEN]; | |
1287 | zlog_debug( | |
45df4e96 DS |
1288 | "netlink_route_multipath(): %s %s vrf %u(%u)", |
1289 | nl_msg_type_to_str(cmd), | |
1290 | prefix2str(p, buf, sizeof(buf)), | |
1291 | zvrf_id(zvrf), tableid); | |
d62a17ae | 1292 | } |
1293 | } | |
1294 | ||
d7c0a89a | 1295 | static void _netlink_mpls_debug(int cmd, uint32_t label, const char *routedesc) |
40c7bdb0 | 1296 | { |
d62a17ae | 1297 | if (IS_ZEBRA_DEBUG_KERNEL) |
1298 | zlog_debug("netlink_mpls_multipath() (%s): %s %u/20", routedesc, | |
1299 | nl_msg_type_to_str(cmd), label); | |
fa713d9e CF |
1300 | } |
1301 | ||
d62a17ae | 1302 | static int netlink_neigh_update(int cmd, int ifindex, uint32_t addr, char *lla, |
5895d33f | 1303 | int llalen, ns_id_t ns_id) |
5c610faf | 1304 | { |
d62a17ae | 1305 | struct { |
1306 | struct nlmsghdr n; | |
1307 | struct ndmsg ndm; | |
1308 | char buf[256]; | |
1309 | } req; | |
5c610faf | 1310 | |
5895d33f | 1311 | struct zebra_ns *zns = zebra_ns_lookup(ns_id); |
8f7d9fc0 | 1312 | |
5605ecfc | 1313 | memset(&req, 0, sizeof(req)); |
5c610faf | 1314 | |
d62a17ae | 1315 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); |
1316 | req.n.nlmsg_flags = NLM_F_CREATE | NLM_F_REQUEST; | |
1317 | req.n.nlmsg_type = cmd; // RTM_NEWNEIGH or RTM_DELNEIGH | |
1318 | req.n.nlmsg_pid = zns->netlink_cmd.snl.nl_pid; | |
a55ba23f | 1319 | |
d62a17ae | 1320 | req.ndm.ndm_family = AF_INET; |
1321 | req.ndm.ndm_state = NUD_PERMANENT; | |
1322 | req.ndm.ndm_ifindex = ifindex; | |
1323 | req.ndm.ndm_type = RTN_UNICAST; | |
5c610faf | 1324 | |
d62a17ae | 1325 | addattr_l(&req.n, sizeof(req), NDA_DST, &addr, 4); |
1326 | addattr_l(&req.n, sizeof(req), NDA_LLADDR, lla, llalen); | |
5c610faf | 1327 | |
d62a17ae | 1328 | return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns, |
1329 | 0); | |
5c610faf DS |
1330 | } |
1331 | ||
718e3744 | 1332 | /* Routing table change via netlink interface. */ |
6ae24471 | 1333 | /* Update flag indicates whether this is a "replace" or not. */ |
d62a17ae | 1334 | static int netlink_route_multipath(int cmd, struct prefix *p, |
1335 | struct prefix *src_p, struct route_entry *re, | |
1336 | int update) | |
718e3744 | 1337 | { |
d62a17ae | 1338 | int bytelen; |
1339 | struct sockaddr_nl snl; | |
1340 | struct nexthop *nexthop = NULL; | |
1341 | unsigned int nexthop_num; | |
d62a17ae | 1342 | int family = PREFIX_FAMILY(p); |
1343 | const char *routedesc; | |
1344 | int setsrc = 0; | |
1345 | union g_addr src; | |
1346 | ||
1347 | struct { | |
1348 | struct nlmsghdr n; | |
1349 | struct rtmsg r; | |
1350 | char buf[NL_PKT_BUF_SIZE]; | |
1351 | } req; | |
1352 | ||
9a76375f | 1353 | struct zebra_ns *zns; |
d62a17ae | 1354 | struct zebra_vrf *zvrf = vrf_info_lookup(re->vrf_id); |
1355 | ||
009f8ad5 | 1356 | zns = zvrf->zns; |
d62a17ae | 1357 | memset(&req, 0, sizeof req - NL_PKT_BUF_SIZE); |
1358 | ||
1359 | bytelen = (family == AF_INET ? 4 : 16); | |
1360 | ||
1361 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg)); | |
1362 | req.n.nlmsg_flags = NLM_F_CREATE | NLM_F_REQUEST; | |
1363 | if ((cmd == RTM_NEWROUTE) && update) | |
1364 | req.n.nlmsg_flags |= NLM_F_REPLACE; | |
1365 | req.n.nlmsg_type = cmd; | |
1366 | req.n.nlmsg_pid = zns->netlink_cmd.snl.nl_pid; | |
1367 | ||
1368 | req.r.rtm_family = family; | |
1369 | req.r.rtm_dst_len = p->prefixlen; | |
1370 | req.r.rtm_src_len = src_p ? src_p->prefixlen : 0; | |
915902cb | 1371 | req.r.rtm_protocol = zebra2proto(re->type); |
d62a17ae | 1372 | req.r.rtm_scope = RT_SCOPE_UNIVERSE; |
c766824c DS |
1373 | |
1374 | /* | |
1375 | * blackhole routes are not RTN_UNICAST, they are | |
1376 | * RTN_ BLACKHOLE|UNREACHABLE|PROHIBIT | |
1377 | * so setting this value as a RTN_UNICAST would | |
1378 | * cause the route lookup of just the prefix | |
1379 | * to fail. So no need to specify this for | |
1380 | * the RTM_DELROUTE case | |
1381 | */ | |
1382 | if (cmd != RTM_DELROUTE) | |
1383 | req.r.rtm_type = RTN_UNICAST; | |
d62a17ae | 1384 | |
d62a17ae | 1385 | addattr_l(&req.n, sizeof req, RTA_DST, &p->u.prefix, bytelen); |
1386 | if (src_p) | |
1387 | addattr_l(&req.n, sizeof req, RTA_SRC, &src_p->u.prefix, | |
1388 | bytelen); | |
1389 | ||
1390 | /* Metric. */ | |
1391 | /* Hardcode the metric for all routes coming from zebra. Metric isn't | |
1392 | * used | |
1393 | * either by the kernel or by zebra. Its purely for calculating best | |
1394 | * path(s) | |
1395 | * by the routing protocol and for communicating with protocol peers. | |
1396 | */ | |
1397 | addattr32(&req.n, sizeof req, RTA_PRIORITY, NL_DEFAULT_ROUTE_METRIC); | |
4e40b6d6 KK |
1398 | #if defined(SUPPORT_REALMS) |
1399 | if (re->tag > 0 && re->tag <= 255) | |
1400 | addattr32(&req.n, sizeof req, RTA_FLOW, re->tag); | |
1401 | #endif | |
d62a17ae | 1402 | /* Table corresponding to this route. */ |
1403 | if (re->table < 256) | |
1404 | req.r.rtm_table = re->table; | |
1405 | else { | |
1406 | req.r.rtm_table = RT_TABLE_UNSPEC; | |
1407 | addattr32(&req.n, sizeof req, RTA_TABLE, re->table); | |
0aabccc0 | 1408 | } |
718e3744 | 1409 | |
45df4e96 DS |
1410 | _netlink_route_debug(cmd, p, family, zvrf, re->table); |
1411 | ||
c1d63a93 DS |
1412 | /* |
1413 | * If we are not updating the route and we have received | |
1414 | * a route delete, then all we need to fill in is the | |
1415 | * prefix information to tell the kernel to schwack | |
1416 | * it. | |
1417 | */ | |
1418 | if (!update && cmd == RTM_DELROUTE) | |
a8309422 DL |
1419 | goto skip; |
1420 | ||
d62a17ae | 1421 | if (re->mtu || re->nexthop_mtu) { |
1422 | char buf[NL_PKT_BUF_SIZE]; | |
1423 | struct rtattr *rta = (void *)buf; | |
d7c0a89a | 1424 | uint32_t mtu = re->mtu; |
d62a17ae | 1425 | if (!mtu || (re->nexthop_mtu && re->nexthop_mtu < mtu)) |
1426 | mtu = re->nexthop_mtu; | |
1427 | rta->rta_type = RTA_METRICS; | |
1428 | rta->rta_len = RTA_LENGTH(0); | |
1429 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, RTAX_MTU, &mtu, sizeof mtu); | |
1430 | addattr_l(&req.n, NL_PKT_BUF_SIZE, RTA_METRICS, RTA_DATA(rta), | |
1431 | RTA_PAYLOAD(rta)); | |
1432 | } | |
1433 | ||
d62a17ae | 1434 | /* Count overall nexthops so we can decide whether to use singlepath |
1435 | * or multipath case. */ | |
1436 | nexthop_num = 0; | |
7ee30f28 | 1437 | for (ALL_NEXTHOPS(re->ng, nexthop)) { |
d62a17ae | 1438 | if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_RECURSIVE)) |
1439 | continue; | |
996c9314 | 1440 | if (cmd == RTM_NEWROUTE && !NEXTHOP_IS_ACTIVE(nexthop->flags)) |
d62a17ae | 1441 | continue; |
1442 | if (cmd == RTM_DELROUTE | |
1443 | && !CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB)) | |
1444 | continue; | |
1445 | ||
1446 | nexthop_num++; | |
1447 | } | |
1448 | ||
1449 | /* Singlepath case. */ | |
1450 | if (nexthop_num == 1 || multipath_num == 1) { | |
1451 | nexthop_num = 0; | |
7ee30f28 | 1452 | for (ALL_NEXTHOPS(re->ng, nexthop)) { |
1f1d24a8 DS |
1453 | /* |
1454 | * So we want to cover 2 types of blackhole | |
1455 | * routes here: | |
1456 | * 1) A normal blackhole route( ala from a static | |
1457 | * install. | |
1458 | * 2) A recursively resolved blackhole route | |
1459 | */ | |
1460 | if (nexthop->type == NEXTHOP_TYPE_BLACKHOLE) { | |
1461 | switch (nexthop->bh_type) { | |
1462 | case BLACKHOLE_ADMINPROHIB: | |
1463 | req.r.rtm_type = RTN_PROHIBIT; | |
1464 | break; | |
1465 | case BLACKHOLE_REJECT: | |
1466 | req.r.rtm_type = RTN_UNREACHABLE; | |
1467 | break; | |
1468 | default: | |
1469 | req.r.rtm_type = RTN_BLACKHOLE; | |
1470 | break; | |
1471 | } | |
1472 | goto skip; | |
1473 | } | |
d62a17ae | 1474 | if (CHECK_FLAG(nexthop->flags, |
1475 | NEXTHOP_FLAG_RECURSIVE)) { | |
1476 | if (!setsrc) { | |
1477 | if (family == AF_INET) { | |
1478 | if (nexthop->rmap_src.ipv4 | |
1479 | .s_addr | |
1480 | != 0) { | |
1481 | src.ipv4 = | |
1482 | nexthop->rmap_src | |
1483 | .ipv4; | |
1484 | setsrc = 1; | |
1485 | } else if (nexthop->src.ipv4 | |
1486 | .s_addr | |
1487 | != 0) { | |
1488 | src.ipv4 = | |
1489 | nexthop->src | |
1490 | .ipv4; | |
1491 | setsrc = 1; | |
1492 | } | |
1493 | } else if (family == AF_INET6) { | |
1494 | if (!IN6_IS_ADDR_UNSPECIFIED( | |
1495 | &nexthop->rmap_src | |
1496 | .ipv6)) { | |
1497 | src.ipv6 = | |
1498 | nexthop->rmap_src | |
1499 | .ipv6; | |
1500 | setsrc = 1; | |
1501 | } else if ( | |
1502 | !IN6_IS_ADDR_UNSPECIFIED( | |
1503 | &nexthop->src | |
1504 | .ipv6)) { | |
1505 | src.ipv6 = | |
1506 | nexthop->src | |
1507 | .ipv6; | |
1508 | setsrc = 1; | |
1509 | } | |
1510 | } | |
1511 | } | |
1512 | continue; | |
1513 | } | |
1514 | ||
1515 | if ((cmd == RTM_NEWROUTE | |
25b9cb0c | 1516 | && NEXTHOP_IS_ACTIVE(nexthop->flags)) |
d62a17ae | 1517 | || (cmd == RTM_DELROUTE |
1518 | && CHECK_FLAG(nexthop->flags, | |
1519 | NEXTHOP_FLAG_FIB))) { | |
1520 | routedesc = nexthop->rparent | |
8b1450b9 RW |
1521 | ? "recursive, single-path" |
1522 | : "single-path"; | |
d62a17ae | 1523 | |
d62a17ae | 1524 | _netlink_route_build_singlepath( |
1525 | routedesc, bytelen, nexthop, &req.n, | |
1526 | &req.r, sizeof req, cmd); | |
1527 | nexthop_num++; | |
1528 | break; | |
1529 | } | |
1530 | } | |
1531 | if (setsrc && (cmd == RTM_NEWROUTE)) { | |
1532 | if (family == AF_INET) | |
1533 | addattr_l(&req.n, sizeof req, RTA_PREFSRC, | |
1534 | &src.ipv4, bytelen); | |
1535 | else if (family == AF_INET6) | |
1536 | addattr_l(&req.n, sizeof req, RTA_PREFSRC, | |
1537 | &src.ipv6, bytelen); | |
1538 | } | |
1539 | } else { | |
1540 | char buf[NL_PKT_BUF_SIZE]; | |
1541 | struct rtattr *rta = (void *)buf; | |
1542 | struct rtnexthop *rtnh; | |
1543 | union g_addr *src1 = NULL; | |
1544 | ||
1545 | rta->rta_type = RTA_MULTIPATH; | |
1546 | rta->rta_len = RTA_LENGTH(0); | |
1547 | rtnh = RTA_DATA(rta); | |
1548 | ||
1549 | nexthop_num = 0; | |
7ee30f28 | 1550 | for (ALL_NEXTHOPS(re->ng, nexthop)) { |
d62a17ae | 1551 | if (nexthop_num >= multipath_num) |
1552 | break; | |
1553 | ||
1554 | if (CHECK_FLAG(nexthop->flags, | |
1555 | NEXTHOP_FLAG_RECURSIVE)) { | |
1556 | /* This only works for IPv4 now */ | |
1557 | if (!setsrc) { | |
1558 | if (family == AF_INET) { | |
1559 | if (nexthop->rmap_src.ipv4 | |
1560 | .s_addr | |
1561 | != 0) { | |
1562 | src.ipv4 = | |
1563 | nexthop->rmap_src | |
1564 | .ipv4; | |
1565 | setsrc = 1; | |
1566 | } else if (nexthop->src.ipv4 | |
1567 | .s_addr | |
1568 | != 0) { | |
1569 | src.ipv4 = | |
1570 | nexthop->src | |
1571 | .ipv4; | |
1572 | setsrc = 1; | |
1573 | } | |
1574 | } else if (family == AF_INET6) { | |
1575 | if (!IN6_IS_ADDR_UNSPECIFIED( | |
1576 | &nexthop->rmap_src | |
1577 | .ipv6)) { | |
1578 | src.ipv6 = | |
1579 | nexthop->rmap_src | |
1580 | .ipv6; | |
1581 | setsrc = 1; | |
1582 | } else if ( | |
1583 | !IN6_IS_ADDR_UNSPECIFIED( | |
1584 | &nexthop->src | |
1585 | .ipv6)) { | |
1586 | src.ipv6 = | |
1587 | nexthop->src | |
1588 | .ipv6; | |
1589 | setsrc = 1; | |
1590 | } | |
1591 | } | |
1592 | } | |
1593 | continue; | |
1594 | } | |
1595 | ||
1596 | if ((cmd == RTM_NEWROUTE | |
25b9cb0c | 1597 | && NEXTHOP_IS_ACTIVE(nexthop->flags)) |
d62a17ae | 1598 | || (cmd == RTM_DELROUTE |
1599 | && CHECK_FLAG(nexthop->flags, | |
1600 | NEXTHOP_FLAG_FIB))) { | |
1601 | routedesc = nexthop->rparent | |
8b1450b9 RW |
1602 | ? "recursive, multipath" |
1603 | : "multipath"; | |
d62a17ae | 1604 | nexthop_num++; |
1605 | ||
d62a17ae | 1606 | _netlink_route_build_multipath( |
1607 | routedesc, bytelen, nexthop, rta, rtnh, | |
1608 | &req.r, &src1); | |
1609 | rtnh = RTNH_NEXT(rtnh); | |
1610 | ||
1611 | if (!setsrc && src1) { | |
1612 | if (family == AF_INET) | |
1613 | src.ipv4 = src1->ipv4; | |
1614 | else if (family == AF_INET6) | |
1615 | src.ipv6 = src1->ipv6; | |
1616 | ||
1617 | setsrc = 1; | |
1618 | } | |
1619 | } | |
1620 | } | |
1621 | if (setsrc && (cmd == RTM_NEWROUTE)) { | |
1622 | if (family == AF_INET) | |
1623 | addattr_l(&req.n, sizeof req, RTA_PREFSRC, | |
1624 | &src.ipv4, bytelen); | |
1625 | else if (family == AF_INET6) | |
1626 | addattr_l(&req.n, sizeof req, RTA_PREFSRC, | |
1627 | &src.ipv6, bytelen); | |
1628 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1629 | zlog_debug("Setting source"); | |
1630 | } | |
1631 | ||
1632 | if (rta->rta_len > RTA_LENGTH(0)) | |
1633 | addattr_l(&req.n, NL_PKT_BUF_SIZE, RTA_MULTIPATH, | |
1634 | RTA_DATA(rta), RTA_PAYLOAD(rta)); | |
1635 | } | |
718e3744 | 1636 | |
d62a17ae | 1637 | /* If there is no useful nexthop then return. */ |
1638 | if (nexthop_num == 0) { | |
1639 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1640 | zlog_debug( | |
1641 | "netlink_route_multipath(): No useful nexthop."); | |
1642 | return 0; | |
1643 | } | |
718e3744 | 1644 | |
7021c425 | 1645 | skip: |
718e3744 | 1646 | |
d62a17ae | 1647 | /* Destination netlink address. */ |
1648 | memset(&snl, 0, sizeof snl); | |
1649 | snl.nl_family = AF_NETLINK; | |
718e3744 | 1650 | |
d62a17ae | 1651 | /* Talk to netlink socket. */ |
1652 | return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns, | |
1653 | 0); | |
718e3744 | 1654 | } |
1655 | ||
43b5cc5e | 1656 | int kernel_get_ipmr_sg_stats(struct zebra_vrf *zvrf, void *in) |
e3be0432 | 1657 | { |
d62a17ae | 1658 | int suc = 0; |
1659 | struct mcast_route_data *mr = (struct mcast_route_data *)in; | |
bd8b9272 DS |
1660 | struct { |
1661 | struct nlmsghdr n; | |
1662 | struct ndmsg ndm; | |
1663 | char buf[256]; | |
1664 | } req; | |
e3be0432 | 1665 | |
d62a17ae | 1666 | mroute = mr; |
5895d33f | 1667 | struct zebra_ns *zns; |
bd8b9272 | 1668 | |
009f8ad5 | 1669 | zns = zvrf->zns; |
5605ecfc | 1670 | memset(&req, 0, sizeof(req)); |
bd8b9272 DS |
1671 | |
1672 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
1673 | req.n.nlmsg_flags = NLM_F_REQUEST; | |
1674 | req.n.nlmsg_pid = zns->netlink_cmd.snl.nl_pid; | |
1675 | ||
1676 | req.ndm.ndm_family = RTNL_FAMILY_IPMR; | |
1677 | req.n.nlmsg_type = RTM_GETROUTE; | |
1678 | ||
1679 | addattr_l(&req.n, sizeof(req), RTA_IIF, &mroute->ifindex, 4); | |
1680 | addattr_l(&req.n, sizeof(req), RTA_OIF, &mroute->ifindex, 4); | |
1681 | addattr_l(&req.n, sizeof(req), RTA_SRC, &mroute->sg.src.s_addr, 4); | |
1682 | addattr_l(&req.n, sizeof(req), RTA_DST, &mroute->sg.grp.s_addr, 4); | |
1683 | addattr_l(&req.n, sizeof(req), RTA_TABLE, &zvrf->table_id, 4); | |
e3be0432 | 1684 | |
bd8b9272 DS |
1685 | suc = netlink_talk(netlink_route_change_read_multicast, &req.n, |
1686 | &zns->netlink_cmd, zns, 0); | |
e3be0432 | 1687 | |
bd8b9272 | 1688 | mroute = NULL; |
d62a17ae | 1689 | return suc; |
e3be0432 DS |
1690 | } |
1691 | ||
7d974ba3 DS |
1692 | void kernel_route_rib(struct route_node *rn, struct prefix *p, |
1693 | struct prefix *src_p, struct route_entry *old, | |
1694 | struct route_entry *new) | |
718e3744 | 1695 | { |
0c555cc6 DS |
1696 | int ret = 0; |
1697 | ||
0af35d90 RW |
1698 | assert(old || new); |
1699 | ||
0c555cc6 | 1700 | if (new) { |
6b093863 | 1701 | if (p->family == AF_INET || v6_rr_semantics) |
0c555cc6 DS |
1702 | ret = netlink_route_multipath(RTM_NEWROUTE, p, src_p, |
1703 | new, (old) ? 1 : 0); | |
1704 | else { | |
1705 | /* | |
1706 | * So v6 route replace semantics are not in | |
1707 | * the kernel at this point as I understand it. | |
1708 | * So let's do a delete than an add. | |
1709 | * In the future once v6 route replace semantics | |
1710 | * are in we can figure out what to do here to | |
1711 | * allow working with old and new kernels. | |
1712 | * | |
1713 | * I'm also intentionally ignoring the failure case | |
1714 | * of the route delete. If that happens yeah we're | |
1715 | * screwed. | |
1716 | */ | |
1717 | if (old) | |
996c9314 LB |
1718 | netlink_route_multipath(RTM_DELROUTE, p, src_p, |
1719 | old, 0); | |
1720 | ret = netlink_route_multipath(RTM_NEWROUTE, p, src_p, | |
1721 | new, 0); | |
0c555cc6 | 1722 | } |
7d974ba3 | 1723 | kernel_route_rib_pass_fail(rn, p, new, |
996c9314 LB |
1724 | (!ret) ? SOUTHBOUND_INSTALL_SUCCESS |
1725 | : SOUTHBOUND_INSTALL_FAILURE); | |
0c555cc6 DS |
1726 | return; |
1727 | } | |
718e3744 | 1728 | |
0c555cc6 DS |
1729 | if (old) { |
1730 | ret = netlink_route_multipath(RTM_DELROUTE, p, src_p, old, 0); | |
e7fcb843 | 1731 | |
7d974ba3 | 1732 | kernel_route_rib_pass_fail(rn, p, old, |
996c9314 LB |
1733 | (!ret) ? SOUTHBOUND_DELETE_SUCCESS |
1734 | : SOUTHBOUND_DELETE_FAILURE); | |
0c555cc6 | 1735 | } |
718e3744 | 1736 | } |
1737 | ||
d62a17ae | 1738 | int kernel_neigh_update(int add, int ifindex, uint32_t addr, char *lla, |
5895d33f | 1739 | int llalen, ns_id_t ns_id) |
6b8a5694 | 1740 | { |
d62a17ae | 1741 | return netlink_neigh_update(add ? RTM_NEWNEIGH : RTM_DELNEIGH, ifindex, |
5895d33f | 1742 | addr, lla, llalen, ns_id); |
6b8a5694 | 1743 | } |
718e3744 | 1744 | |
13d60d35 | 1745 | /* |
1746 | * Add remote VTEP to the flood list for this VxLAN interface (VNI). This | |
1747 | * is done by adding an FDB entry with a MAC of 00:00:00:00:00:00. | |
1748 | */ | |
d62a17ae | 1749 | static int netlink_vxlan_flood_list_update(struct interface *ifp, |
1750 | struct in_addr *vtep_ip, int cmd) | |
13d60d35 | 1751 | { |
5895d33f | 1752 | struct zebra_ns *zns; |
d62a17ae | 1753 | struct { |
1754 | struct nlmsghdr n; | |
1755 | struct ndmsg ndm; | |
1756 | char buf[256]; | |
1757 | } req; | |
d7c0a89a | 1758 | uint8_t dst_mac[6] = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0}; |
009f8ad5 | 1759 | struct zebra_vrf *zvrf = zebra_vrf_lookup_by_id(ifp->vrf_id); |
d62a17ae | 1760 | |
009f8ad5 | 1761 | zns = zvrf->zns; |
5605ecfc | 1762 | memset(&req, 0, sizeof(req)); |
d62a17ae | 1763 | |
1764 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
1765 | req.n.nlmsg_flags = NLM_F_REQUEST; | |
1766 | if (cmd == RTM_NEWNEIGH) | |
1767 | req.n.nlmsg_flags |= (NLM_F_CREATE | NLM_F_APPEND); | |
1768 | req.n.nlmsg_type = cmd; | |
1769 | req.ndm.ndm_family = PF_BRIDGE; | |
1770 | req.ndm.ndm_state = NUD_NOARP | NUD_PERMANENT; | |
1771 | req.ndm.ndm_flags |= NTF_SELF; // Handle by "self", not "master" | |
1772 | ||
1773 | ||
1774 | addattr_l(&req.n, sizeof(req), NDA_LLADDR, &dst_mac, 6); | |
1775 | req.ndm.ndm_ifindex = ifp->ifindex; | |
1776 | addattr_l(&req.n, sizeof(req), NDA_DST, &vtep_ip->s_addr, 4); | |
1777 | ||
1778 | return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns, | |
1779 | 0); | |
13d60d35 | 1780 | } |
1781 | ||
1782 | /* | |
d62a17ae | 1783 | * Add remote VTEP for this VxLAN interface (VNI). In Linux, this involves |
1784 | * adding | |
13d60d35 | 1785 | * a "flood" MAC FDB entry. |
1786 | */ | |
d62a17ae | 1787 | int kernel_add_vtep(vni_t vni, struct interface *ifp, struct in_addr *vtep_ip) |
13d60d35 | 1788 | { |
d62a17ae | 1789 | if (IS_ZEBRA_DEBUG_VXLAN) |
1790 | zlog_debug("Install %s into flood list for VNI %u intf %s(%u)", | |
1791 | inet_ntoa(*vtep_ip), vni, ifp->name, ifp->ifindex); | |
13d60d35 | 1792 | |
d62a17ae | 1793 | return netlink_vxlan_flood_list_update(ifp, vtep_ip, RTM_NEWNEIGH); |
13d60d35 | 1794 | } |
1795 | ||
1796 | /* | |
1797 | * Remove remote VTEP for this VxLAN interface (VNI). In Linux, this involves | |
1798 | * deleting the "flood" MAC FDB entry. | |
1799 | */ | |
d62a17ae | 1800 | int kernel_del_vtep(vni_t vni, struct interface *ifp, struct in_addr *vtep_ip) |
13d60d35 | 1801 | { |
d62a17ae | 1802 | if (IS_ZEBRA_DEBUG_VXLAN) |
1803 | zlog_debug( | |
1804 | "Uninstall %s from flood list for VNI %u intf %s(%u)", | |
1805 | inet_ntoa(*vtep_ip), vni, ifp->name, ifp->ifindex); | |
13d60d35 | 1806 | |
d62a17ae | 1807 | return netlink_vxlan_flood_list_update(ifp, vtep_ip, RTM_DELNEIGH); |
13d60d35 | 1808 | } |
1809 | ||
2232a77c | 1810 | #ifndef NDA_RTA |
d62a17ae | 1811 | #define NDA_RTA(r) \ |
1812 | ((struct rtattr *)(((char *)(r)) + NLMSG_ALIGN(sizeof(struct ndmsg)))) | |
2232a77c | 1813 | #endif |
1814 | ||
2414abd3 | 1815 | static int netlink_macfdb_change(struct nlmsghdr *h, int len, ns_id_t ns_id) |
2232a77c | 1816 | { |
d62a17ae | 1817 | struct ndmsg *ndm; |
1818 | struct interface *ifp; | |
1819 | struct zebra_if *zif; | |
d62a17ae | 1820 | struct rtattr *tb[NDA_MAX + 1]; |
1821 | struct interface *br_if; | |
1822 | struct ethaddr mac; | |
1823 | vlanid_t vid = 0; | |
1824 | struct prefix vtep_ip; | |
1825 | int vid_present = 0, dst_present = 0; | |
1826 | char buf[ETHER_ADDR_STRLEN]; | |
1827 | char vid_buf[20]; | |
1828 | char dst_buf[30]; | |
d7c0a89a | 1829 | uint8_t sticky = 0; |
d62a17ae | 1830 | |
1831 | ndm = NLMSG_DATA(h); | |
1832 | ||
2853fed6 | 1833 | /* We only process macfdb notifications if EVPN is enabled */ |
1834 | if (!is_evpn_enabled()) | |
1835 | return 0; | |
1836 | ||
d62a17ae | 1837 | /* The interface should exist. */ |
5895d33f | 1838 | ifp = if_lookup_by_index_per_ns(zebra_ns_lookup(ns_id), |
d62a17ae | 1839 | ndm->ndm_ifindex); |
2853fed6 | 1840 | if (!ifp || !ifp->info) |
d62a17ae | 1841 | return 0; |
1842 | ||
1843 | /* The interface should be something we're interested in. */ | |
1844 | if (!IS_ZEBRA_IF_BRIDGE_SLAVE(ifp)) | |
1845 | return 0; | |
1846 | ||
1847 | /* Drop "permanent" entries. */ | |
1848 | if (ndm->ndm_state & NUD_PERMANENT) | |
1849 | return 0; | |
1850 | ||
1851 | zif = (struct zebra_if *)ifp->info; | |
1852 | if ((br_if = zif->brslave_info.br_if) == NULL) { | |
1853 | zlog_warn("%s family %s IF %s(%u) brIF %u - no bridge master", | |
1854 | nl_msg_type_to_str(h->nlmsg_type), | |
1855 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
1856 | ndm->ndm_ifindex, zif->brslave_info.bridge_ifindex); | |
1857 | return 0; | |
1858 | } | |
1859 | ||
1860 | /* Parse attributes and extract fields of interest. */ | |
1861 | memset(tb, 0, sizeof tb); | |
1862 | netlink_parse_rtattr(tb, NDA_MAX, NDA_RTA(ndm), len); | |
1863 | ||
1864 | if (!tb[NDA_LLADDR]) { | |
1865 | zlog_warn("%s family %s IF %s(%u) brIF %u - no LLADDR", | |
1866 | nl_msg_type_to_str(h->nlmsg_type), | |
1867 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
1868 | ndm->ndm_ifindex, zif->brslave_info.bridge_ifindex); | |
1869 | return 0; | |
1870 | } | |
1871 | ||
ff8b7eb8 | 1872 | if (RTA_PAYLOAD(tb[NDA_LLADDR]) != ETH_ALEN) { |
d62a17ae | 1873 | zlog_warn( |
df0b13cf | 1874 | "%s family %s IF %s(%u) brIF %u - LLADDR is not MAC, len %lu", |
d62a17ae | 1875 | nl_msg_type_to_str(h->nlmsg_type), |
1876 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
1877 | ndm->ndm_ifindex, zif->brslave_info.bridge_ifindex, | |
df0b13cf | 1878 | (unsigned long)RTA_PAYLOAD(tb[NDA_LLADDR])); |
d62a17ae | 1879 | return 0; |
1880 | } | |
1881 | ||
ff8b7eb8 | 1882 | memcpy(&mac, RTA_DATA(tb[NDA_LLADDR]), ETH_ALEN); |
d62a17ae | 1883 | |
1884 | if ((NDA_VLAN <= NDA_MAX) && tb[NDA_VLAN]) { | |
1885 | vid_present = 1; | |
d7c0a89a | 1886 | vid = *(uint16_t *)RTA_DATA(tb[NDA_VLAN]); |
d62a17ae | 1887 | sprintf(vid_buf, " VLAN %u", vid); |
1888 | } | |
1889 | ||
1890 | if (tb[NDA_DST]) { | |
1891 | /* TODO: Only IPv4 supported now. */ | |
1892 | dst_present = 1; | |
1893 | vtep_ip.family = AF_INET; | |
1894 | vtep_ip.prefixlen = IPV4_MAX_BITLEN; | |
1895 | memcpy(&(vtep_ip.u.prefix4.s_addr), RTA_DATA(tb[NDA_DST]), | |
1896 | IPV4_MAX_BYTELEN); | |
1897 | sprintf(dst_buf, " dst %s", inet_ntoa(vtep_ip.u.prefix4)); | |
1898 | } | |
1899 | ||
1900 | sticky = (ndm->ndm_state & NUD_NOARP) ? 1 : 0; | |
1901 | ||
1902 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1903 | zlog_debug("Rx %s family %s IF %s(%u)%s %sMAC %s%s", | |
1904 | nl_msg_type_to_str(h->nlmsg_type), | |
1905 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
1906 | ndm->ndm_ifindex, vid_present ? vid_buf : "", | |
1907 | sticky ? "sticky " : "", | |
1908 | prefix_mac2str(&mac, buf, sizeof(buf)), | |
1909 | dst_present ? dst_buf : ""); | |
1910 | ||
1911 | if (filter_vlan && vid != filter_vlan) | |
1912 | return 0; | |
1913 | ||
1914 | /* If add or update, do accordingly if learnt on a "local" interface; if | |
1915 | * the notification is over VxLAN, this has to be related to | |
1916 | * multi-homing, | |
1917 | * so perform an implicit delete of any local entry (if it exists). | |
1918 | */ | |
1919 | if (h->nlmsg_type == RTM_NEWNEIGH) { | |
1920 | /* Drop "permanent" entries. */ | |
1921 | if (ndm->ndm_state & NUD_PERMANENT) | |
1922 | return 0; | |
1923 | ||
1924 | if (IS_ZEBRA_IF_VXLAN(ifp)) | |
1925 | return zebra_vxlan_check_del_local_mac(ifp, br_if, &mac, | |
1926 | vid); | |
1927 | ||
1928 | return zebra_vxlan_local_mac_add_update(ifp, br_if, &mac, vid, | |
1929 | sticky); | |
1930 | } | |
1931 | ||
1932 | /* This is a delete notification. | |
1933 | * 1. For a MAC over VxLan, check if it needs to be refreshed(readded) | |
1934 | * 2. For a MAC over "local" interface, delete the mac | |
1935 | * Note: We will get notifications from both bridge driver and VxLAN | |
1936 | * driver. | |
1937 | * Ignore the notification from VxLan driver as it is also generated | |
1938 | * when mac moves from remote to local. | |
1939 | */ | |
1940 | if (dst_present) | |
1941 | return 0; | |
1942 | ||
1943 | if (IS_ZEBRA_IF_VXLAN(ifp)) | |
1944 | return zebra_vxlan_check_readd_remote_mac(ifp, br_if, &mac, | |
1945 | vid); | |
1946 | ||
1947 | return zebra_vxlan_local_mac_del(ifp, br_if, &mac, vid); | |
2232a77c | 1948 | } |
1949 | ||
2414abd3 | 1950 | static int netlink_macfdb_table(struct nlmsghdr *h, ns_id_t ns_id, int startup) |
2232a77c | 1951 | { |
d62a17ae | 1952 | int len; |
1953 | struct ndmsg *ndm; | |
2232a77c | 1954 | |
d62a17ae | 1955 | if (h->nlmsg_type != RTM_NEWNEIGH) |
1956 | return 0; | |
2232a77c | 1957 | |
d62a17ae | 1958 | /* Length validity. */ |
1959 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct ndmsg)); | |
1960 | if (len < 0) | |
1961 | return -1; | |
2232a77c | 1962 | |
d62a17ae | 1963 | /* We are interested only in AF_BRIDGE notifications. */ |
1964 | ndm = NLMSG_DATA(h); | |
1965 | if (ndm->ndm_family != AF_BRIDGE) | |
1966 | return 0; | |
2232a77c | 1967 | |
2414abd3 | 1968 | return netlink_macfdb_change(h, len, ns_id); |
2232a77c | 1969 | } |
1970 | ||
1971 | /* Request for MAC FDB information from the kernel */ | |
d62a17ae | 1972 | static int netlink_request_macs(struct zebra_ns *zns, int family, int type, |
1973 | ifindex_t master_ifindex) | |
2232a77c | 1974 | { |
d62a17ae | 1975 | struct { |
1976 | struct nlmsghdr n; | |
1977 | struct ifinfomsg ifm; | |
1978 | char buf[256]; | |
1979 | } req; | |
1980 | ||
1981 | /* Form the request, specifying filter (rtattr) if needed. */ | |
1982 | memset(&req, 0, sizeof(req)); | |
1983 | req.n.nlmsg_type = type; | |
1984 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ifinfomsg)); | |
1985 | req.ifm.ifi_family = family; | |
1986 | if (master_ifindex) | |
1987 | addattr32(&req.n, sizeof(req), IFLA_MASTER, master_ifindex); | |
1988 | ||
1989 | return netlink_request(&zns->netlink_cmd, &req.n); | |
2232a77c | 1990 | } |
1991 | ||
1992 | /* | |
1993 | * MAC forwarding database read using netlink interface. This is invoked | |
1994 | * at startup. | |
1995 | */ | |
d62a17ae | 1996 | int netlink_macfdb_read(struct zebra_ns *zns) |
2232a77c | 1997 | { |
d62a17ae | 1998 | int ret; |
1999 | ||
2000 | /* Get bridge FDB table. */ | |
2001 | ret = netlink_request_macs(zns, AF_BRIDGE, RTM_GETNEIGH, 0); | |
2002 | if (ret < 0) | |
2003 | return ret; | |
2004 | /* We are reading entire table. */ | |
2005 | filter_vlan = 0; | |
2006 | ret = netlink_parse_info(netlink_macfdb_table, &zns->netlink_cmd, zns, | |
2007 | 0, 1); | |
2008 | ||
2009 | return ret; | |
2232a77c | 2010 | } |
2011 | ||
2012 | /* | |
2013 | * MAC forwarding database read using netlink interface. This is for a | |
2014 | * specific bridge and matching specific access VLAN (if VLAN-aware bridge). | |
2015 | */ | |
d62a17ae | 2016 | int netlink_macfdb_read_for_bridge(struct zebra_ns *zns, struct interface *ifp, |
2017 | struct interface *br_if) | |
2232a77c | 2018 | { |
d62a17ae | 2019 | struct zebra_if *br_zif; |
2020 | struct zebra_if *zif; | |
2021 | struct zebra_l2info_vxlan *vxl; | |
2022 | int ret = 0; | |
2023 | ||
2024 | ||
2025 | /* Save VLAN we're filtering on, if needed. */ | |
2026 | br_zif = (struct zebra_if *)br_if->info; | |
2027 | zif = (struct zebra_if *)ifp->info; | |
2028 | vxl = &zif->l2info.vxl; | |
2029 | if (IS_ZEBRA_IF_BRIDGE_VLAN_AWARE(br_zif)) | |
2030 | filter_vlan = vxl->access_vlan; | |
2031 | ||
2032 | /* Get bridge FDB table for specific bridge - we do the VLAN filtering. | |
2033 | */ | |
2034 | ret = netlink_request_macs(zns, AF_BRIDGE, RTM_GETNEIGH, | |
2035 | br_if->ifindex); | |
2036 | if (ret < 0) | |
2037 | return ret; | |
2038 | ret = netlink_parse_info(netlink_macfdb_table, &zns->netlink_cmd, zns, | |
2039 | 0, 0); | |
2040 | ||
2041 | /* Reset VLAN filter. */ | |
2042 | filter_vlan = 0; | |
2043 | return ret; | |
2232a77c | 2044 | } |
2045 | ||
d62a17ae | 2046 | static int netlink_macfdb_update(struct interface *ifp, vlanid_t vid, |
2047 | struct ethaddr *mac, struct in_addr vtep_ip, | |
d7c0a89a | 2048 | int local, int cmd, uint8_t sticky) |
2232a77c | 2049 | { |
5895d33f | 2050 | struct zebra_ns *zns; |
d62a17ae | 2051 | struct { |
2052 | struct nlmsghdr n; | |
2053 | struct ndmsg ndm; | |
2054 | char buf[256]; | |
2055 | } req; | |
2056 | int dst_alen; | |
2057 | struct zebra_if *zif; | |
2058 | struct interface *br_if; | |
2059 | struct zebra_if *br_zif; | |
2060 | char buf[ETHER_ADDR_STRLEN]; | |
2061 | int vid_present = 0, dst_present = 0; | |
2062 | char vid_buf[20]; | |
2063 | char dst_buf[30]; | |
009f8ad5 | 2064 | struct zebra_vrf *zvrf = zebra_vrf_lookup_by_id(ifp->vrf_id); |
d62a17ae | 2065 | |
009f8ad5 | 2066 | zns = zvrf->zns; |
d62a17ae | 2067 | zif = ifp->info; |
2068 | if ((br_if = zif->brslave_info.br_if) == NULL) { | |
2069 | zlog_warn("MAC %s on IF %s(%u) - no mapping to bridge", | |
2070 | (cmd == RTM_NEWNEIGH) ? "add" : "del", ifp->name, | |
2071 | ifp->ifindex); | |
2072 | return -1; | |
2073 | } | |
2074 | ||
5605ecfc | 2075 | memset(&req, 0, sizeof(req)); |
d62a17ae | 2076 | |
2077 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
2078 | req.n.nlmsg_flags = NLM_F_REQUEST; | |
2079 | if (cmd == RTM_NEWNEIGH) | |
2080 | req.n.nlmsg_flags |= (NLM_F_CREATE | NLM_F_REPLACE); | |
2081 | req.n.nlmsg_type = cmd; | |
2082 | req.ndm.ndm_family = AF_BRIDGE; | |
2083 | req.ndm.ndm_flags |= NTF_SELF | NTF_MASTER; | |
2084 | req.ndm.ndm_state = NUD_REACHABLE; | |
2085 | ||
2086 | if (sticky) | |
2087 | req.ndm.ndm_state |= NUD_NOARP; | |
2088 | else | |
2089 | req.ndm.ndm_flags |= NTF_EXT_LEARNED; | |
2090 | ||
2091 | addattr_l(&req.n, sizeof(req), NDA_LLADDR, mac, 6); | |
2092 | req.ndm.ndm_ifindex = ifp->ifindex; | |
2093 | if (!local) { | |
2094 | dst_alen = 4; // TODO: hardcoded | |
2095 | addattr_l(&req.n, sizeof(req), NDA_DST, &vtep_ip, dst_alen); | |
2096 | dst_present = 1; | |
2097 | sprintf(dst_buf, " dst %s", inet_ntoa(vtep_ip)); | |
2098 | } | |
2099 | br_zif = (struct zebra_if *)br_if->info; | |
2100 | if (IS_ZEBRA_IF_BRIDGE_VLAN_AWARE(br_zif) && vid > 0) { | |
2101 | addattr16(&req.n, sizeof(req), NDA_VLAN, vid); | |
2102 | vid_present = 1; | |
2103 | sprintf(vid_buf, " VLAN %u", vid); | |
2104 | } | |
2105 | addattr32(&req.n, sizeof(req), NDA_MASTER, br_if->ifindex); | |
2106 | ||
2107 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2108 | zlog_debug("Tx %s family %s IF %s(%u)%s %sMAC %s%s", | |
2109 | nl_msg_type_to_str(cmd), | |
2110 | nl_family_to_str(req.ndm.ndm_family), ifp->name, | |
2111 | ifp->ifindex, vid_present ? vid_buf : "", | |
2112 | sticky ? "sticky " : "", | |
2113 | prefix_mac2str(mac, buf, sizeof(buf)), | |
2114 | dst_present ? dst_buf : ""); | |
2115 | ||
2116 | return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns, | |
2117 | 0); | |
2232a77c | 2118 | } |
2119 | ||
d62a17ae | 2120 | #define NUD_VALID \ |
2121 | (NUD_PERMANENT | NUD_NOARP | NUD_REACHABLE | NUD_PROBE | NUD_STALE \ | |
2122 | | NUD_DELAY) | |
2232a77c | 2123 | |
2414abd3 | 2124 | static int netlink_ipneigh_change(struct nlmsghdr *h, int len, ns_id_t ns_id) |
2232a77c | 2125 | { |
d62a17ae | 2126 | struct ndmsg *ndm; |
2127 | struct interface *ifp; | |
2128 | struct zebra_if *zif; | |
d62a17ae | 2129 | struct rtattr *tb[NDA_MAX + 1]; |
2130 | struct interface *link_if; | |
2131 | struct ethaddr mac; | |
2132 | struct ipaddr ip; | |
2133 | char buf[ETHER_ADDR_STRLEN]; | |
2134 | char buf2[INET6_ADDRSTRLEN]; | |
2135 | int mac_present = 0; | |
d7c0a89a | 2136 | uint8_t ext_learned; |
d62a17ae | 2137 | |
2138 | ndm = NLMSG_DATA(h); | |
2139 | ||
2140 | /* The interface should exist. */ | |
5895d33f | 2141 | ifp = if_lookup_by_index_per_ns(zebra_ns_lookup(ns_id), |
d62a17ae | 2142 | ndm->ndm_ifindex); |
2853fed6 | 2143 | if (!ifp || !ifp->info) |
d62a17ae | 2144 | return 0; |
2145 | ||
20089ae2 DS |
2146 | zif = (struct zebra_if *)ifp->info; |
2147 | ||
2148 | /* Parse attributes and extract fields of interest. */ | |
2149 | memset(tb, 0, sizeof tb); | |
2150 | netlink_parse_rtattr(tb, NDA_MAX, NDA_RTA(ndm), len); | |
2151 | ||
2152 | if (!tb[NDA_DST]) { | |
2153 | zlog_warn("%s family %s IF %s(%u) - no DST", | |
2154 | nl_msg_type_to_str(h->nlmsg_type), | |
2155 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
2156 | ndm->ndm_ifindex); | |
d62a17ae | 2157 | return 0; |
20089ae2 DS |
2158 | } |
2159 | ||
2160 | memset(&ip, 0, sizeof(struct ipaddr)); | |
2161 | ip.ipa_type = (ndm->ndm_family == AF_INET) ? IPADDR_V4 : IPADDR_V6; | |
2162 | memcpy(&ip.ip.addr, RTA_DATA(tb[NDA_DST]), RTA_PAYLOAD(tb[NDA_DST])); | |
2163 | ||
2164 | /* Drop some "permanent" entries. */ | |
2165 | if (ndm->ndm_state & NUD_PERMANENT) { | |
2166 | char buf[16] = "169.254.0.1"; | |
2167 | struct in_addr ipv4_ll; | |
2168 | ||
2169 | if (ndm->ndm_family != AF_INET) | |
2170 | return 0; | |
2171 | ||
2172 | if (!zif->v6_2_v4_ll_neigh_entry) | |
2173 | return 0; | |
2174 | ||
2175 | if (h->nlmsg_type != RTM_DELNEIGH) | |
2176 | return 0; | |
2177 | ||
2178 | inet_pton(AF_INET, buf, &ipv4_ll); | |
2179 | if (ipv4_ll.s_addr != ip.ip._v4_addr.s_addr) | |
2180 | return 0; | |
2181 | ||
2182 | if_nbr_ipv6ll_to_ipv4ll_neigh_update( | |
2183 | ifp, &zif->v6_2_v4_ll_addr6, true); | |
2184 | return 0; | |
2185 | } | |
d62a17ae | 2186 | |
d62a17ae | 2187 | /* The neighbor is present on an SVI. From this, we locate the |
2188 | * underlying | |
2189 | * bridge because we're only interested in neighbors on a VxLAN bridge. | |
2190 | * The bridge is located based on the nature of the SVI: | |
2191 | * (a) In the case of a VLAN-aware bridge, the SVI is a L3 VLAN | |
2192 | * interface | |
2193 | * and is linked to the bridge | |
2194 | * (b) In the case of a VLAN-unaware bridge, the SVI is the bridge | |
2195 | * inteface | |
2196 | * itself | |
2197 | */ | |
2198 | if (IS_ZEBRA_IF_VLAN(ifp)) { | |
5895d33f | 2199 | link_if = if_lookup_by_index_per_ns(zebra_ns_lookup(ns_id), |
71349e03 | 2200 | zif->link_ifindex); |
d62a17ae | 2201 | if (!link_if) |
2202 | return 0; | |
2203 | } else if (IS_ZEBRA_IF_BRIDGE(ifp)) | |
2204 | link_if = ifp; | |
2205 | else | |
2206 | return 0; | |
2207 | ||
d62a17ae | 2208 | memset(&mac, 0, sizeof(struct ethaddr)); |
d62a17ae | 2209 | if (h->nlmsg_type == RTM_NEWNEIGH) { |
2210 | if (tb[NDA_LLADDR]) { | |
ff8b7eb8 | 2211 | if (RTA_PAYLOAD(tb[NDA_LLADDR]) != ETH_ALEN) { |
d62a17ae | 2212 | zlog_warn( |
df0b13cf | 2213 | "%s family %s IF %s(%u) - LLADDR is not MAC, len %lu", |
d62a17ae | 2214 | nl_msg_type_to_str(h->nlmsg_type), |
2215 | nl_family_to_str(ndm->ndm_family), | |
2216 | ifp->name, ndm->ndm_ifindex, | |
996c9314 LB |
2217 | (unsigned long)RTA_PAYLOAD( |
2218 | tb[NDA_LLADDR])); | |
d62a17ae | 2219 | return 0; |
2220 | } | |
2221 | ||
2222 | mac_present = 1; | |
ff8b7eb8 | 2223 | memcpy(&mac, RTA_DATA(tb[NDA_LLADDR]), ETH_ALEN); |
d62a17ae | 2224 | } |
2225 | ||
2226 | ext_learned = (ndm->ndm_flags & NTF_EXT_LEARNED) ? 1 : 0; | |
2227 | ||
2228 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2229 | zlog_debug( | |
2230 | "Rx %s family %s IF %s(%u) IP %s MAC %s state 0x%x flags 0x%x", | |
2231 | nl_msg_type_to_str(h->nlmsg_type), | |
2232 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
2233 | ndm->ndm_ifindex, | |
2234 | ipaddr2str(&ip, buf2, sizeof(buf2)), | |
2235 | mac_present | |
2236 | ? prefix_mac2str(&mac, buf, sizeof(buf)) | |
2237 | : "", | |
2238 | ndm->ndm_state, ndm->ndm_flags); | |
2239 | ||
2240 | /* If the neighbor state is valid for use, process as an add or | |
2241 | * update | |
2242 | * else process as a delete. Note that the delete handling may | |
2243 | * result | |
2244 | * in re-adding the neighbor if it is a valid "remote" neighbor. | |
2245 | */ | |
2246 | if (ndm->ndm_state & NUD_VALID) | |
ee69da27 | 2247 | return zebra_vxlan_handle_kernel_neigh_update( |
d62a17ae | 2248 | ifp, link_if, &ip, &mac, ndm->ndm_state, |
2249 | ext_learned); | |
2250 | ||
ee69da27 | 2251 | return zebra_vxlan_handle_kernel_neigh_del(ifp, link_if, &ip); |
d62a17ae | 2252 | } |
2253 | ||
2254 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2255 | zlog_debug("Rx %s family %s IF %s(%u) IP %s", | |
2256 | nl_msg_type_to_str(h->nlmsg_type), | |
2257 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
2258 | ndm->ndm_ifindex, | |
2259 | ipaddr2str(&ip, buf2, sizeof(buf2))); | |
2260 | ||
2261 | /* Process the delete - it may result in re-adding the neighbor if it is | |
2262 | * a valid "remote" neighbor. | |
2263 | */ | |
ee69da27 | 2264 | return zebra_vxlan_handle_kernel_neigh_del(ifp, link_if, &ip); |
2232a77c | 2265 | } |
2266 | ||
2414abd3 | 2267 | static int netlink_neigh_table(struct nlmsghdr *h, ns_id_t ns_id, int startup) |
2232a77c | 2268 | { |
d62a17ae | 2269 | int len; |
2270 | struct ndmsg *ndm; | |
2232a77c | 2271 | |
d62a17ae | 2272 | if (h->nlmsg_type != RTM_NEWNEIGH) |
2273 | return 0; | |
2232a77c | 2274 | |
d62a17ae | 2275 | /* Length validity. */ |
2276 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct ndmsg)); | |
2277 | if (len < 0) | |
2278 | return -1; | |
2232a77c | 2279 | |
d62a17ae | 2280 | /* We are interested only in AF_INET or AF_INET6 notifications. */ |
2281 | ndm = NLMSG_DATA(h); | |
2282 | if (ndm->ndm_family != AF_INET && ndm->ndm_family != AF_INET6) | |
2283 | return 0; | |
2232a77c | 2284 | |
2414abd3 | 2285 | return netlink_neigh_change(h, len); |
2232a77c | 2286 | } |
2287 | ||
2288 | /* Request for IP neighbor information from the kernel */ | |
d62a17ae | 2289 | static int netlink_request_neigh(struct zebra_ns *zns, int family, int type, |
2290 | ifindex_t ifindex) | |
2232a77c | 2291 | { |
d62a17ae | 2292 | struct { |
2293 | struct nlmsghdr n; | |
2294 | struct ndmsg ndm; | |
2295 | char buf[256]; | |
2296 | } req; | |
2297 | ||
2298 | /* Form the request, specifying filter (rtattr) if needed. */ | |
2299 | memset(&req, 0, sizeof(req)); | |
2300 | req.n.nlmsg_type = type; | |
2301 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
2302 | req.ndm.ndm_family = family; | |
2303 | if (ifindex) | |
2304 | addattr32(&req.n, sizeof(req), NDA_IFINDEX, ifindex); | |
2305 | ||
2306 | return netlink_request(&zns->netlink_cmd, &req.n); | |
2232a77c | 2307 | } |
2308 | ||
2309 | /* | |
2310 | * IP Neighbor table read using netlink interface. This is invoked | |
2311 | * at startup. | |
2312 | */ | |
d62a17ae | 2313 | int netlink_neigh_read(struct zebra_ns *zns) |
2232a77c | 2314 | { |
d62a17ae | 2315 | int ret; |
2232a77c | 2316 | |
d62a17ae | 2317 | /* Get IP neighbor table. */ |
2318 | ret = netlink_request_neigh(zns, AF_UNSPEC, RTM_GETNEIGH, 0); | |
2319 | if (ret < 0) | |
2320 | return ret; | |
2321 | ret = netlink_parse_info(netlink_neigh_table, &zns->netlink_cmd, zns, 0, | |
2322 | 1); | |
2232a77c | 2323 | |
d62a17ae | 2324 | return ret; |
2232a77c | 2325 | } |
2326 | ||
2327 | /* | |
2328 | * IP Neighbor table read using netlink interface. This is for a specific | |
2329 | * VLAN device. | |
2330 | */ | |
d62a17ae | 2331 | int netlink_neigh_read_for_vlan(struct zebra_ns *zns, struct interface *vlan_if) |
2232a77c | 2332 | { |
d62a17ae | 2333 | int ret = 0; |
2232a77c | 2334 | |
d62a17ae | 2335 | ret = netlink_request_neigh(zns, AF_UNSPEC, RTM_GETNEIGH, |
2336 | vlan_if->ifindex); | |
2337 | if (ret < 0) | |
2338 | return ret; | |
2339 | ret = netlink_parse_info(netlink_neigh_table, &zns->netlink_cmd, zns, 0, | |
2340 | 0); | |
2232a77c | 2341 | |
d62a17ae | 2342 | return ret; |
2232a77c | 2343 | } |
2344 | ||
2414abd3 | 2345 | int netlink_neigh_change(struct nlmsghdr *h, ns_id_t ns_id) |
2232a77c | 2346 | { |
d62a17ae | 2347 | int len; |
2348 | struct ndmsg *ndm; | |
2232a77c | 2349 | |
d62a17ae | 2350 | if (!(h->nlmsg_type == RTM_NEWNEIGH || h->nlmsg_type == RTM_DELNEIGH)) |
2351 | return 0; | |
2232a77c | 2352 | |
d62a17ae | 2353 | /* Length validity. */ |
2354 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct ndmsg)); | |
2355 | if (len < 0) | |
2356 | return -1; | |
2232a77c | 2357 | |
d62a17ae | 2358 | /* Is this a notification for the MAC FDB or IP neighbor table? */ |
2359 | ndm = NLMSG_DATA(h); | |
2360 | if (ndm->ndm_family == AF_BRIDGE) | |
2414abd3 | 2361 | return netlink_macfdb_change(h, len, ns_id); |
2232a77c | 2362 | |
d62a17ae | 2363 | if (ndm->ndm_type != RTN_UNICAST) |
2364 | return 0; | |
2232a77c | 2365 | |
d62a17ae | 2366 | if (ndm->ndm_family == AF_INET || ndm->ndm_family == AF_INET6) |
2414abd3 | 2367 | return netlink_ipneigh_change(h, len, ns_id); |
2232a77c | 2368 | |
d62a17ae | 2369 | return 0; |
2232a77c | 2370 | } |
2371 | ||
d62a17ae | 2372 | static int netlink_neigh_update2(struct interface *ifp, struct ipaddr *ip, |
d7c0a89a | 2373 | struct ethaddr *mac, uint32_t flags, int cmd) |
2232a77c | 2374 | { |
d62a17ae | 2375 | struct { |
2376 | struct nlmsghdr n; | |
2377 | struct ndmsg ndm; | |
2378 | char buf[256]; | |
2379 | } req; | |
2380 | int ipa_len; | |
2381 | ||
5895d33f | 2382 | struct zebra_ns *zns; |
d62a17ae | 2383 | char buf[INET6_ADDRSTRLEN]; |
2384 | char buf2[ETHER_ADDR_STRLEN]; | |
009f8ad5 | 2385 | struct zebra_vrf *zvrf = zebra_vrf_lookup_by_id(ifp->vrf_id); |
d62a17ae | 2386 | |
009f8ad5 | 2387 | zns = zvrf->zns; |
5605ecfc | 2388 | memset(&req, 0, sizeof(req)); |
d62a17ae | 2389 | |
2390 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
2391 | req.n.nlmsg_flags = NLM_F_REQUEST; | |
2392 | if (cmd == RTM_NEWNEIGH) | |
2393 | req.n.nlmsg_flags |= (NLM_F_CREATE | NLM_F_REPLACE); | |
2394 | req.n.nlmsg_type = cmd; // RTM_NEWNEIGH or RTM_DELNEIGH | |
2395 | req.ndm.ndm_family = IS_IPADDR_V4(ip) ? AF_INET : AF_INET6; | |
2396 | req.ndm.ndm_state = flags; | |
2397 | req.ndm.ndm_ifindex = ifp->ifindex; | |
2398 | req.ndm.ndm_type = RTN_UNICAST; | |
2399 | req.ndm.ndm_flags = NTF_EXT_LEARNED; | |
2400 | ||
2401 | ||
2402 | ipa_len = IS_IPADDR_V4(ip) ? IPV4_MAX_BYTELEN : IPV6_MAX_BYTELEN; | |
2403 | addattr_l(&req.n, sizeof(req), NDA_DST, &ip->ip.addr, ipa_len); | |
2404 | if (mac) | |
2405 | addattr_l(&req.n, sizeof(req), NDA_LLADDR, mac, 6); | |
2406 | ||
2407 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2408 | zlog_debug("Tx %s family %s IF %s(%u) Neigh %s MAC %s", | |
2409 | nl_msg_type_to_str(cmd), | |
2410 | nl_family_to_str(req.ndm.ndm_family), ifp->name, | |
2411 | ifp->ifindex, ipaddr2str(ip, buf, sizeof(buf)), | |
2412 | mac ? prefix_mac2str(mac, buf2, sizeof(buf2)) | |
2413 | : "null"); | |
2414 | ||
2415 | return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns, | |
2416 | 0); | |
2232a77c | 2417 | } |
2418 | ||
d62a17ae | 2419 | int kernel_add_mac(struct interface *ifp, vlanid_t vid, struct ethaddr *mac, |
d7c0a89a | 2420 | struct in_addr vtep_ip, uint8_t sticky) |
2232a77c | 2421 | { |
d62a17ae | 2422 | return netlink_macfdb_update(ifp, vid, mac, vtep_ip, 0, RTM_NEWNEIGH, |
2423 | sticky); | |
2232a77c | 2424 | } |
2425 | ||
d62a17ae | 2426 | int kernel_del_mac(struct interface *ifp, vlanid_t vid, struct ethaddr *mac, |
2427 | struct in_addr vtep_ip, int local) | |
2232a77c | 2428 | { |
d62a17ae | 2429 | return netlink_macfdb_update(ifp, vid, mac, vtep_ip, local, |
2430 | RTM_DELNEIGH, 0); | |
2232a77c | 2431 | } |
2432 | ||
d62a17ae | 2433 | int kernel_add_neigh(struct interface *ifp, struct ipaddr *ip, |
2434 | struct ethaddr *mac) | |
2232a77c | 2435 | { |
e9d2cbde | 2436 | return netlink_neigh_update2(ifp, ip, mac, NUD_NOARP, RTM_NEWNEIGH); |
2232a77c | 2437 | } |
2438 | ||
d62a17ae | 2439 | int kernel_del_neigh(struct interface *ifp, struct ipaddr *ip) |
2232a77c | 2440 | { |
d62a17ae | 2441 | return netlink_neigh_update2(ifp, ip, NULL, 0, RTM_DELNEIGH); |
2232a77c | 2442 | } |
2443 | ||
40c7bdb0 | 2444 | /* |
2445 | * MPLS label forwarding table change via netlink interface. | |
2446 | */ | |
d62a17ae | 2447 | int netlink_mpls_multipath(int cmd, zebra_lsp_t *lsp) |
40c7bdb0 | 2448 | { |
d62a17ae | 2449 | mpls_lse_t lse; |
2450 | zebra_nhlfe_t *nhlfe; | |
2451 | struct nexthop *nexthop = NULL; | |
2452 | unsigned int nexthop_num; | |
2453 | const char *routedesc; | |
2454 | struct zebra_ns *zns = zebra_ns_lookup(NS_DEFAULT); | |
805444ce | 2455 | int route_type; |
d62a17ae | 2456 | |
2457 | struct { | |
2458 | struct nlmsghdr n; | |
2459 | struct rtmsg r; | |
2460 | char buf[NL_PKT_BUF_SIZE]; | |
2461 | } req; | |
2462 | ||
2463 | memset(&req, 0, sizeof req - NL_PKT_BUF_SIZE); | |
2464 | ||
d62a17ae | 2465 | /* |
2466 | * Count # nexthops so we can decide whether to use singlepath | |
2467 | * or multipath case. | |
2468 | */ | |
2469 | nexthop_num = 0; | |
2470 | for (nhlfe = lsp->nhlfe_list; nhlfe; nhlfe = nhlfe->next) { | |
2471 | nexthop = nhlfe->nexthop; | |
2472 | if (!nexthop) | |
2473 | continue; | |
2474 | if (cmd == RTM_NEWROUTE) { | |
2475 | /* Count all selected NHLFEs */ | |
2476 | if (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_SELECTED) | |
2477 | && CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ACTIVE)) | |
2478 | nexthop_num++; | |
2479 | } else /* DEL */ | |
2480 | { | |
2481 | /* Count all installed NHLFEs */ | |
2482 | if (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_INSTALLED) | |
2483 | && CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB)) | |
2484 | nexthop_num++; | |
2485 | } | |
2486 | } | |
2487 | ||
8dc8a4b6 | 2488 | if ((nexthop_num == 0) || (!lsp->best_nhlfe && (cmd != RTM_DELROUTE))) |
d62a17ae | 2489 | return 0; |
2490 | ||
2491 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg)); | |
2492 | req.n.nlmsg_flags = NLM_F_CREATE | NLM_F_REQUEST; | |
2493 | req.n.nlmsg_type = cmd; | |
2494 | req.n.nlmsg_pid = zns->netlink_cmd.snl.nl_pid; | |
2495 | ||
2496 | req.r.rtm_family = AF_MPLS; | |
2497 | req.r.rtm_table = RT_TABLE_MAIN; | |
2498 | req.r.rtm_dst_len = MPLS_LABEL_LEN_BITS; | |
d62a17ae | 2499 | req.r.rtm_scope = RT_SCOPE_UNIVERSE; |
2500 | req.r.rtm_type = RTN_UNICAST; | |
2501 | ||
8dc8a4b6 | 2502 | if (cmd == RTM_NEWROUTE) { |
d62a17ae | 2503 | /* We do a replace to handle update. */ |
2504 | req.n.nlmsg_flags |= NLM_F_REPLACE; | |
2505 | ||
8dc8a4b6 DS |
2506 | /* set the protocol value if installing */ |
2507 | route_type = re_type_from_lsp_type(lsp->best_nhlfe->type); | |
2508 | req.r.rtm_protocol = zebra2proto(route_type); | |
2509 | } | |
2510 | ||
d62a17ae | 2511 | /* Fill destination */ |
2512 | lse = mpls_lse_encode(lsp->ile.in_label, 0, 0, 1); | |
2513 | addattr_l(&req.n, sizeof req, RTA_DST, &lse, sizeof(mpls_lse_t)); | |
2514 | ||
2515 | /* Fill nexthops (paths) based on single-path or multipath. The paths | |
2516 | * chosen depend on the operation. | |
2517 | */ | |
2518 | if (nexthop_num == 1 || multipath_num == 1) { | |
8b1450b9 | 2519 | routedesc = "single-path"; |
d62a17ae | 2520 | _netlink_mpls_debug(cmd, lsp->ile.in_label, routedesc); |
2521 | ||
2522 | nexthop_num = 0; | |
2523 | for (nhlfe = lsp->nhlfe_list; nhlfe; nhlfe = nhlfe->next) { | |
2524 | nexthop = nhlfe->nexthop; | |
2525 | if (!nexthop) | |
2526 | continue; | |
2527 | ||
2528 | if ((cmd == RTM_NEWROUTE | |
2529 | && (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_SELECTED) | |
2530 | && CHECK_FLAG(nexthop->flags, | |
2531 | NEXTHOP_FLAG_ACTIVE))) | |
2532 | || (cmd == RTM_DELROUTE | |
2533 | && (CHECK_FLAG(nhlfe->flags, | |
2534 | NHLFE_FLAG_INSTALLED) | |
2535 | && CHECK_FLAG(nexthop->flags, | |
2536 | NEXTHOP_FLAG_FIB)))) { | |
2537 | /* Add the gateway */ | |
2538 | _netlink_mpls_build_singlepath(routedesc, nhlfe, | |
2539 | &req.n, &req.r, | |
2540 | sizeof req, cmd); | |
d62a17ae | 2541 | nexthop_num++; |
2542 | break; | |
2543 | } | |
2544 | } | |
2545 | } else /* Multipath case */ | |
2546 | { | |
2547 | char buf[NL_PKT_BUF_SIZE]; | |
2548 | struct rtattr *rta = (void *)buf; | |
2549 | struct rtnexthop *rtnh; | |
2550 | union g_addr *src1 = NULL; | |
2551 | ||
2552 | rta->rta_type = RTA_MULTIPATH; | |
2553 | rta->rta_len = RTA_LENGTH(0); | |
2554 | rtnh = RTA_DATA(rta); | |
2555 | ||
8b1450b9 | 2556 | routedesc = "multipath"; |
d62a17ae | 2557 | _netlink_mpls_debug(cmd, lsp->ile.in_label, routedesc); |
2558 | ||
2559 | nexthop_num = 0; | |
2560 | for (nhlfe = lsp->nhlfe_list; nhlfe; nhlfe = nhlfe->next) { | |
2561 | nexthop = nhlfe->nexthop; | |
2562 | if (!nexthop) | |
2563 | continue; | |
2564 | ||
2565 | if (nexthop_num >= multipath_num) | |
2566 | break; | |
2567 | ||
2568 | if ((cmd == RTM_NEWROUTE | |
2569 | && (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_SELECTED) | |
2570 | && CHECK_FLAG(nexthop->flags, | |
2571 | NEXTHOP_FLAG_ACTIVE))) | |
2572 | || (cmd == RTM_DELROUTE | |
2573 | && (CHECK_FLAG(nhlfe->flags, | |
2574 | NHLFE_FLAG_INSTALLED) | |
2575 | && CHECK_FLAG(nexthop->flags, | |
2576 | NEXTHOP_FLAG_FIB)))) { | |
2577 | nexthop_num++; | |
2578 | ||
2579 | /* Build the multipath */ | |
2580 | _netlink_mpls_build_multipath(routedesc, nhlfe, | |
2581 | rta, rtnh, &req.r, | |
2582 | &src1); | |
2583 | rtnh = RTNH_NEXT(rtnh); | |
d62a17ae | 2584 | } |
2585 | } | |
2586 | ||
2587 | /* Add the multipath */ | |
2588 | if (rta->rta_len > RTA_LENGTH(0)) | |
2589 | addattr_l(&req.n, NL_PKT_BUF_SIZE, RTA_MULTIPATH, | |
2590 | RTA_DATA(rta), RTA_PAYLOAD(rta)); | |
2591 | } | |
2592 | ||
2593 | /* Talk to netlink socket. */ | |
2594 | return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns, | |
2595 | 0); | |
40c7bdb0 | 2596 | } |
ddfeb486 | 2597 | #endif /* HAVE_NETLINK */ |