]>
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> | |
d9f5b2f5 | 30 | #include <linux/nexthop.h> |
718e3744 | 31 | |
32 | /* Hack for GNU libc version 2. */ | |
33 | #ifndef MSG_TRUNC | |
34 | #define MSG_TRUNC 0x20 | |
35 | #endif /* MSG_TRUNC */ | |
36 | ||
37 | #include "linklist.h" | |
38 | #include "if.h" | |
39 | #include "log.h" | |
40 | #include "prefix.h" | |
41 | #include "connected.h" | |
42 | #include "table.h" | |
26e2ae36 | 43 | #include "memory.h" |
4a1ab8e4 | 44 | #include "zebra_memory.h" |
718e3744 | 45 | #include "rib.h" |
e04ab74d | 46 | #include "thread.h" |
edd7c245 | 47 | #include "privs.h" |
fb018d25 | 48 | #include "nexthop.h" |
78104b9b | 49 | #include "vrf.h" |
5e6a74d8 | 50 | #include "vty.h" |
40c7bdb0 | 51 | #include "mpls.h" |
13d60d35 | 52 | #include "vxlan.h" |
8d03bc50 | 53 | #include "printfrr.h" |
718e3744 | 54 | |
bf094f69 | 55 | #include "zebra/zapi_msg.h" |
fe18ee2d | 56 | #include "zebra/zebra_ns.h" |
7c551956 | 57 | #include "zebra/zebra_vrf.h" |
6621ca86 | 58 | #include "zebra/rt.h" |
718e3744 | 59 | #include "zebra/redistribute.h" |
60 | #include "zebra/interface.h" | |
61 | #include "zebra/debug.h" | |
12f6fb97 | 62 | #include "zebra/rtadv.h" |
567b877d | 63 | #include "zebra/zebra_ptm.h" |
40c7bdb0 | 64 | #include "zebra/zebra_mpls.h" |
1fdc9eae | 65 | #include "zebra/kernel_netlink.h" |
66 | #include "zebra/rt_netlink.h" | |
d9f5b2f5 | 67 | #include "zebra/zebra_nhg.h" |
e3be0432 | 68 | #include "zebra/zebra_mroute.h" |
2232a77c | 69 | #include "zebra/zebra_vxlan.h" |
364fed6b | 70 | #include "zebra/zebra_errors.h" |
e3be0432 | 71 | |
40c7bdb0 | 72 | #ifndef AF_MPLS |
73 | #define AF_MPLS 28 | |
74 | #endif | |
75 | ||
2232a77c | 76 | static vlanid_t filter_vlan = 0; |
77 | ||
fec211ad | 78 | static bool supports_nh; |
81505946 | 79 | |
d62a17ae | 80 | struct gw_family_t { |
d7c0a89a QY |
81 | uint16_t filler; |
82 | uint16_t family; | |
d62a17ae | 83 | union g_addr gate; |
40c7bdb0 | 84 | }; |
85 | ||
2b64873d DL |
86 | static const char ipv4_ll_buf[16] = "169.254.0.1"; |
87 | static struct in_addr ipv4_ll; | |
8755598a DS |
88 | |
89 | /* | |
90 | * The ipv4_ll data structure is used for all 5549 | |
91 | * additions to the kernel. Let's figure out the | |
92 | * correct value one time instead for every | |
93 | * install/remove of a 5549 type route | |
94 | */ | |
d62a17ae | 95 | void rt_netlink_init(void) |
8755598a | 96 | { |
d62a17ae | 97 | inet_pton(AF_INET, ipv4_ll_buf, &ipv4_ll); |
8755598a DS |
98 | } |
99 | ||
931fa60c MS |
100 | /* |
101 | * Mapping from dataplane neighbor flags to netlink flags | |
102 | */ | |
103 | static uint8_t neigh_flags_to_netlink(uint8_t dplane_flags) | |
104 | { | |
105 | uint8_t flags = 0; | |
106 | ||
107 | if (dplane_flags & DPLANE_NTF_EXT_LEARNED) | |
108 | flags |= NTF_EXT_LEARNED; | |
109 | if (dplane_flags & DPLANE_NTF_ROUTER) | |
110 | flags |= NTF_ROUTER; | |
111 | ||
112 | return flags; | |
113 | } | |
114 | ||
115 | /* | |
116 | * Mapping from dataplane neighbor state to netlink state | |
117 | */ | |
118 | static uint16_t neigh_state_to_netlink(uint16_t dplane_state) | |
119 | { | |
120 | uint16_t state = 0; | |
121 | ||
122 | if (dplane_state & DPLANE_NUD_REACHABLE) | |
123 | state |= NUD_REACHABLE; | |
124 | if (dplane_state & DPLANE_NUD_STALE) | |
125 | state |= NUD_STALE; | |
126 | if (dplane_state & DPLANE_NUD_NOARP) | |
127 | state |= NUD_NOARP; | |
128 | if (dplane_state & DPLANE_NUD_PROBE) | |
129 | state |= NUD_PROBE; | |
130 | ||
131 | return state; | |
132 | } | |
133 | ||
134 | ||
23b1f334 DD |
135 | static inline int is_selfroute(int proto) |
136 | { | |
d62a17ae | 137 | if ((proto == RTPROT_BGP) || (proto == RTPROT_OSPF) |
d4d71f11 | 138 | || (proto == RTPROT_ZSTATIC) || (proto == RTPROT_ZEBRA) |
d62a17ae | 139 | || (proto == RTPROT_ISIS) || (proto == RTPROT_RIPNG) |
140 | || (proto == RTPROT_NHRP) || (proto == RTPROT_EIGRP) | |
915902cb | 141 | || (proto == RTPROT_LDP) || (proto == RTPROT_BABEL) |
0761368a | 142 | || (proto == RTPROT_RIP) || (proto == RTPROT_SHARP) |
da82f6b4 | 143 | || (proto == RTPROT_PBR) || (proto == RTPROT_OPENFABRIC)) { |
d62a17ae | 144 | return 1; |
145 | } | |
146 | ||
147 | return 0; | |
23b1f334 DD |
148 | } |
149 | ||
915902cb | 150 | static inline int zebra2proto(int proto) |
23b1f334 | 151 | { |
d62a17ae | 152 | switch (proto) { |
153 | case ZEBRA_ROUTE_BABEL: | |
154 | proto = RTPROT_BABEL; | |
155 | break; | |
156 | case ZEBRA_ROUTE_BGP: | |
157 | proto = RTPROT_BGP; | |
158 | break; | |
159 | case ZEBRA_ROUTE_OSPF: | |
160 | case ZEBRA_ROUTE_OSPF6: | |
161 | proto = RTPROT_OSPF; | |
162 | break; | |
163 | case ZEBRA_ROUTE_STATIC: | |
d4d71f11 | 164 | proto = RTPROT_ZSTATIC; |
d62a17ae | 165 | break; |
166 | case ZEBRA_ROUTE_ISIS: | |
167 | proto = RTPROT_ISIS; | |
168 | break; | |
169 | case ZEBRA_ROUTE_RIP: | |
170 | proto = RTPROT_RIP; | |
171 | break; | |
172 | case ZEBRA_ROUTE_RIPNG: | |
173 | proto = RTPROT_RIPNG; | |
174 | break; | |
175 | case ZEBRA_ROUTE_NHRP: | |
176 | proto = RTPROT_NHRP; | |
177 | break; | |
178 | case ZEBRA_ROUTE_EIGRP: | |
179 | proto = RTPROT_EIGRP; | |
180 | break; | |
181 | case ZEBRA_ROUTE_LDP: | |
182 | proto = RTPROT_LDP; | |
183 | break; | |
8a71d93d DS |
184 | case ZEBRA_ROUTE_SHARP: |
185 | proto = RTPROT_SHARP; | |
186 | break; | |
0761368a DS |
187 | case ZEBRA_ROUTE_PBR: |
188 | proto = RTPROT_PBR; | |
189 | break; | |
da82f6b4 CF |
190 | case ZEBRA_ROUTE_OPENFABRIC: |
191 | proto = RTPROT_OPENFABRIC; | |
192 | break; | |
a56ec5c0 | 193 | case ZEBRA_ROUTE_TABLE: |
38e40db1 | 194 | case ZEBRA_ROUTE_NHG: |
a56ec5c0 DS |
195 | proto = RTPROT_ZEBRA; |
196 | break; | |
d62a17ae | 197 | default: |
0761368a DS |
198 | /* |
199 | * When a user adds a new protocol this will show up | |
200 | * to let them know to do something about it. This | |
201 | * is intentionally a warn because we should see | |
202 | * this as part of development of a new protocol | |
203 | */ | |
9df414fe QY |
204 | zlog_debug( |
205 | "%s: Please add this protocol(%d) to proper rt_netlink.c handling", | |
206 | __PRETTY_FUNCTION__, proto); | |
d62a17ae | 207 | proto = RTPROT_ZEBRA; |
208 | break; | |
209 | } | |
210 | ||
211 | return proto; | |
23b1f334 DD |
212 | } |
213 | ||
38e40db1 | 214 | static inline int proto2zebra(int proto, int family, bool is_nexthop) |
915902cb DS |
215 | { |
216 | switch (proto) { | |
217 | case RTPROT_BABEL: | |
218 | proto = ZEBRA_ROUTE_BABEL; | |
219 | break; | |
220 | case RTPROT_BGP: | |
221 | proto = ZEBRA_ROUTE_BGP; | |
222 | break; | |
223 | case RTPROT_OSPF: | |
996c9314 LB |
224 | proto = (family == AFI_IP) ? ZEBRA_ROUTE_OSPF |
225 | : ZEBRA_ROUTE_OSPF6; | |
915902cb DS |
226 | break; |
227 | case RTPROT_ISIS: | |
228 | proto = ZEBRA_ROUTE_ISIS; | |
229 | break; | |
230 | case RTPROT_RIP: | |
231 | proto = ZEBRA_ROUTE_RIP; | |
232 | break; | |
233 | case RTPROT_RIPNG: | |
234 | proto = ZEBRA_ROUTE_RIPNG; | |
235 | break; | |
236 | case RTPROT_NHRP: | |
237 | proto = ZEBRA_ROUTE_NHRP; | |
238 | break; | |
239 | case RTPROT_EIGRP: | |
240 | proto = ZEBRA_ROUTE_EIGRP; | |
241 | break; | |
242 | case RTPROT_LDP: | |
243 | proto = ZEBRA_ROUTE_LDP; | |
244 | break; | |
245 | case RTPROT_STATIC: | |
d4d71f11 | 246 | case RTPROT_ZSTATIC: |
915902cb DS |
247 | proto = ZEBRA_ROUTE_STATIC; |
248 | break; | |
0761368a DS |
249 | case RTPROT_SHARP: |
250 | proto = ZEBRA_ROUTE_SHARP; | |
251 | break; | |
252 | case RTPROT_PBR: | |
253 | proto = ZEBRA_ROUTE_PBR; | |
254 | break; | |
da82f6b4 CF |
255 | case RTPROT_OPENFABRIC: |
256 | proto = ZEBRA_ROUTE_OPENFABRIC; | |
257 | break; | |
38e40db1 SW |
258 | case RTPROT_ZEBRA: |
259 | if (is_nexthop) { | |
260 | proto = ZEBRA_ROUTE_NHG; | |
261 | break; | |
262 | } | |
263 | /* Intentional fall thru */ | |
915902cb | 264 | default: |
0761368a DS |
265 | /* |
266 | * When a user adds a new protocol this will show up | |
267 | * to let them know to do something about it. This | |
268 | * is intentionally a warn because we should see | |
269 | * this as part of development of a new protocol | |
270 | */ | |
9df414fe QY |
271 | zlog_debug( |
272 | "%s: Please add this protocol(%d) to proper rt_netlink.c handling", | |
273 | __PRETTY_FUNCTION__, proto); | |
915902cb DS |
274 | proto = ZEBRA_ROUTE_KERNEL; |
275 | break; | |
276 | } | |
277 | return proto; | |
278 | } | |
279 | ||
12f6fb97 DS |
280 | /* |
281 | Pending: create an efficient table_id (in a tree/hash) based lookup) | |
282 | */ | |
d7c0a89a | 283 | static vrf_id_t vrf_lookup_by_table(uint32_t table_id, ns_id_t ns_id) |
12f6fb97 | 284 | { |
d62a17ae | 285 | struct vrf *vrf; |
286 | struct zebra_vrf *zvrf; | |
12f6fb97 | 287 | |
a2addae8 | 288 | RB_FOREACH (vrf, vrf_id_head, &vrfs_by_id) { |
78dd30b2 PG |
289 | zvrf = vrf->info; |
290 | if (zvrf == NULL) | |
d62a17ae | 291 | continue; |
78dd30b2 PG |
292 | /* case vrf with netns : match the netnsid */ |
293 | if (vrf_is_backend_netns()) { | |
294 | if (ns_id == zvrf_id(zvrf)) | |
295 | return zvrf_id(zvrf); | |
296 | } else { | |
297 | /* VRF is VRF_BACKEND_VRF_LITE */ | |
298 | if (zvrf->table_id != table_id) | |
299 | continue; | |
300 | return zvrf_id(zvrf); | |
301 | } | |
d62a17ae | 302 | } |
12f6fb97 | 303 | |
d62a17ae | 304 | return VRF_DEFAULT; |
12f6fb97 DS |
305 | } |
306 | ||
87da6a60 SW |
307 | /** |
308 | * @parse_encap_mpls() - Parses encapsulated mpls attributes | |
309 | * @tb: Pointer to rtattr to look for nested items in. | |
310 | * @labels: Pointer to store labels in. | |
311 | * | |
312 | * Return: Number of mpls labels found. | |
313 | */ | |
314 | static int parse_encap_mpls(struct rtattr *tb, mpls_label_t *labels) | |
315 | { | |
316 | struct rtattr *tb_encap[MPLS_IPTUNNEL_MAX + 1] = {0}; | |
317 | mpls_lse_t *lses = NULL; | |
318 | int num_labels = 0; | |
319 | uint32_t ttl = 0; | |
320 | uint32_t bos = 0; | |
321 | uint32_t exp = 0; | |
322 | mpls_label_t label = 0; | |
323 | ||
324 | netlink_parse_rtattr_nested(tb_encap, MPLS_IPTUNNEL_MAX, tb); | |
325 | lses = (mpls_lse_t *)RTA_DATA(tb_encap[MPLS_IPTUNNEL_DST]); | |
326 | while (!bos && num_labels < MPLS_MAX_LABELS) { | |
327 | mpls_lse_decode(lses[num_labels], &label, &ttl, &exp, &bos); | |
328 | labels[num_labels++] = label; | |
329 | } | |
330 | ||
331 | return num_labels; | |
332 | } | |
333 | ||
77a44d94 SW |
334 | static struct nexthop |
335 | parse_nexthop_unicast(ns_id_t ns_id, struct rtmsg *rtm, struct rtattr **tb, | |
336 | enum blackhole_type bh_type, int index, void *prefsrc, | |
20822f9d | 337 | void *gate, afi_t afi, vrf_id_t vrf_id) |
77a44d94 SW |
338 | { |
339 | struct interface *ifp = NULL; | |
340 | struct nexthop nh = {0}; | |
341 | mpls_label_t labels[MPLS_MAX_LABELS] = {0}; | |
342 | int num_labels = 0; | |
343 | ||
20822f9d | 344 | vrf_id_t nh_vrf_id = vrf_id; |
77a44d94 SW |
345 | size_t sz = (afi == AFI_IP) ? 4 : 16; |
346 | ||
347 | if (bh_type == BLACKHOLE_UNSPEC) { | |
348 | if (index && !gate) | |
349 | nh.type = NEXTHOP_TYPE_IFINDEX; | |
350 | else if (index && gate) | |
351 | nh.type = (afi == AFI_IP) ? NEXTHOP_TYPE_IPV4_IFINDEX | |
352 | : NEXTHOP_TYPE_IPV6_IFINDEX; | |
353 | else if (!index && gate) | |
354 | nh.type = (afi == AFI_IP) ? NEXTHOP_TYPE_IPV4 | |
355 | : NEXTHOP_TYPE_IPV6; | |
356 | else { | |
357 | nh.type = NEXTHOP_TYPE_BLACKHOLE; | |
358 | nh.bh_type = bh_type; | |
359 | } | |
360 | } else { | |
361 | nh.type = NEXTHOP_TYPE_BLACKHOLE; | |
362 | nh.bh_type = bh_type; | |
363 | } | |
364 | nh.ifindex = index; | |
365 | if (prefsrc) | |
366 | memcpy(&nh.src, prefsrc, sz); | |
367 | if (gate) | |
368 | memcpy(&nh.gate, gate, sz); | |
369 | ||
370 | if (index) { | |
371 | ifp = if_lookup_by_index_per_ns(zebra_ns_lookup(ns_id), index); | |
372 | if (ifp) | |
373 | nh_vrf_id = ifp->vrf_id; | |
374 | } | |
375 | nh.vrf_id = nh_vrf_id; | |
376 | ||
377 | if (tb[RTA_ENCAP] && tb[RTA_ENCAP_TYPE] | |
378 | && *(uint16_t *)RTA_DATA(tb[RTA_ENCAP_TYPE]) | |
379 | == LWTUNNEL_ENCAP_MPLS) { | |
380 | num_labels = parse_encap_mpls(tb[RTA_ENCAP], labels); | |
381 | } | |
382 | ||
383 | if (rtm->rtm_flags & RTNH_F_ONLINK) | |
384 | SET_FLAG(nh.flags, NEXTHOP_FLAG_ONLINK); | |
385 | ||
386 | if (num_labels) | |
387 | nexthop_add_labels(&nh, ZEBRA_LSP_STATIC, num_labels, labels); | |
388 | ||
389 | return nh; | |
390 | } | |
391 | ||
20822f9d | 392 | static uint8_t parse_multipath_nexthops_unicast(ns_id_t ns_id, |
0eb97b86 | 393 | struct nexthop_group *ng, |
20822f9d SW |
394 | struct rtmsg *rtm, |
395 | struct rtnexthop *rtnh, | |
396 | struct rtattr **tb, | |
397 | void *prefsrc, vrf_id_t vrf_id) | |
398 | { | |
399 | void *gate = NULL; | |
400 | struct interface *ifp = NULL; | |
401 | int index = 0; | |
402 | /* MPLS labels */ | |
403 | mpls_label_t labels[MPLS_MAX_LABELS] = {0}; | |
404 | int num_labels = 0; | |
405 | struct rtattr *rtnh_tb[RTA_MAX + 1] = {}; | |
406 | ||
407 | int len = RTA_PAYLOAD(tb[RTA_MULTIPATH]); | |
408 | vrf_id_t nh_vrf_id = vrf_id; | |
409 | ||
20822f9d SW |
410 | for (;;) { |
411 | struct nexthop *nh = NULL; | |
412 | ||
413 | if (len < (int)sizeof(*rtnh) || rtnh->rtnh_len > len) | |
414 | break; | |
415 | ||
416 | index = rtnh->rtnh_ifindex; | |
417 | if (index) { | |
418 | /* | |
419 | * Yes we are looking this up | |
420 | * for every nexthop and just | |
421 | * using the last one looked | |
422 | * up right now | |
423 | */ | |
424 | ifp = if_lookup_by_index_per_ns(zebra_ns_lookup(ns_id), | |
425 | index); | |
426 | if (ifp) | |
427 | nh_vrf_id = ifp->vrf_id; | |
428 | else { | |
429 | flog_warn( | |
430 | EC_ZEBRA_UNKNOWN_INTERFACE, | |
431 | "%s: Unknown interface %u specified, defaulting to VRF_DEFAULT", | |
432 | __PRETTY_FUNCTION__, index); | |
433 | nh_vrf_id = VRF_DEFAULT; | |
434 | } | |
435 | } else | |
436 | nh_vrf_id = vrf_id; | |
437 | ||
438 | if (rtnh->rtnh_len > sizeof(*rtnh)) { | |
439 | memset(rtnh_tb, 0, sizeof(rtnh_tb)); | |
440 | ||
441 | netlink_parse_rtattr(rtnh_tb, RTA_MAX, RTNH_DATA(rtnh), | |
442 | rtnh->rtnh_len - sizeof(*rtnh)); | |
443 | if (rtnh_tb[RTA_GATEWAY]) | |
444 | gate = RTA_DATA(rtnh_tb[RTA_GATEWAY]); | |
445 | if (rtnh_tb[RTA_ENCAP] && rtnh_tb[RTA_ENCAP_TYPE] | |
446 | && *(uint16_t *)RTA_DATA(rtnh_tb[RTA_ENCAP_TYPE]) | |
447 | == LWTUNNEL_ENCAP_MPLS) { | |
448 | num_labels = parse_encap_mpls( | |
449 | rtnh_tb[RTA_ENCAP], labels); | |
450 | } | |
451 | } | |
452 | ||
f3354e16 SW |
453 | if (gate && rtm->rtm_family == AF_INET) { |
454 | if (index) | |
0eb97b86 MS |
455 | nh = nexthop_from_ipv4_ifindex( |
456 | gate, prefsrc, index, nh_vrf_id); | |
f3354e16 | 457 | else |
0eb97b86 MS |
458 | nh = nexthop_from_ipv4(gate, prefsrc, |
459 | nh_vrf_id); | |
f3354e16 SW |
460 | } else if (gate && rtm->rtm_family == AF_INET6) { |
461 | if (index) | |
0eb97b86 MS |
462 | nh = nexthop_from_ipv6_ifindex( |
463 | gate, index, nh_vrf_id); | |
f3354e16 | 464 | else |
0eb97b86 | 465 | nh = nexthop_from_ipv6(gate, nh_vrf_id); |
20822f9d | 466 | } else |
0eb97b86 | 467 | nh = nexthop_from_ifindex(index, nh_vrf_id); |
20822f9d SW |
468 | |
469 | if (nh) { | |
df7fb580 DS |
470 | nh->weight = rtnh->rtnh_hops + 1; |
471 | ||
20822f9d SW |
472 | if (num_labels) |
473 | nexthop_add_labels(nh, ZEBRA_LSP_STATIC, | |
474 | num_labels, labels); | |
475 | ||
476 | if (rtnh->rtnh_flags & RTNH_F_ONLINK) | |
477 | SET_FLAG(nh->flags, NEXTHOP_FLAG_ONLINK); | |
0eb97b86 MS |
478 | |
479 | /* Add to temporary list */ | |
480 | nexthop_group_add_sorted(ng, nh); | |
20822f9d SW |
481 | } |
482 | ||
483 | if (rtnh->rtnh_len == 0) | |
484 | break; | |
485 | ||
486 | len -= NLMSG_ALIGN(rtnh->rtnh_len); | |
487 | rtnh = RTNH_NEXT(rtnh); | |
488 | } | |
489 | ||
0eb97b86 | 490 | uint8_t nhop_num = nexthop_group_nexthop_num(ng); |
20822f9d SW |
491 | |
492 | return nhop_num; | |
493 | } | |
494 | ||
718e3744 | 495 | /* Looking up routing table by netlink interface. */ |
2414abd3 | 496 | static int netlink_route_change_read_unicast(struct nlmsghdr *h, ns_id_t ns_id, |
d62a17ae | 497 | int startup) |
718e3744 | 498 | { |
d62a17ae | 499 | int len; |
500 | struct rtmsg *rtm; | |
501 | struct rtattr *tb[RTA_MAX + 1]; | |
d7c0a89a | 502 | uint8_t flags = 0; |
d62a17ae | 503 | struct prefix p; |
792fa92e | 504 | struct prefix_ipv6 src_p = {}; |
78dd30b2 | 505 | vrf_id_t vrf_id; |
d62a17ae | 506 | |
507 | char anyaddr[16] = {0}; | |
508 | ||
915902cb | 509 | int proto = ZEBRA_ROUTE_KERNEL; |
d62a17ae | 510 | int index = 0; |
511 | int table; | |
512 | int metric = 0; | |
d7c0a89a | 513 | uint32_t mtu = 0; |
25715c7e | 514 | uint8_t distance = 0; |
4e40b6d6 | 515 | route_tag_t tag = 0; |
fcc89a9c | 516 | uint32_t nhe_id = 0; |
d62a17ae | 517 | |
518 | void *dest = NULL; | |
519 | void *gate = NULL; | |
520 | void *prefsrc = NULL; /* IPv4 preferred source host address */ | |
521 | void *src = NULL; /* IPv6 srcdest source prefix */ | |
e655a03c | 522 | enum blackhole_type bh_type = BLACKHOLE_UNSPEC; |
d62a17ae | 523 | |
524 | rtm = NLMSG_DATA(h); | |
525 | ||
526 | if (startup && h->nlmsg_type != RTM_NEWROUTE) | |
527 | return 0; | |
e655a03c DL |
528 | switch (rtm->rtm_type) { |
529 | case RTN_UNICAST: | |
530 | break; | |
531 | case RTN_BLACKHOLE: | |
532 | bh_type = BLACKHOLE_NULL; | |
533 | break; | |
534 | case RTN_UNREACHABLE: | |
535 | bh_type = BLACKHOLE_REJECT; | |
536 | break; | |
537 | case RTN_PROHIBIT: | |
538 | bh_type = BLACKHOLE_ADMINPROHIB; | |
539 | break; | |
540 | default: | |
8c8f250b DS |
541 | if (IS_ZEBRA_DEBUG_KERNEL) |
542 | zlog_debug("Route rtm_type: %s(%d) intentionally ignoring", | |
543 | nl_rttype_to_str(rtm->rtm_type), | |
544 | rtm->rtm_type); | |
d62a17ae | 545 | return 0; |
e655a03c | 546 | } |
d62a17ae | 547 | |
548 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct rtmsg)); | |
9bdf8618 DS |
549 | if (len < 0) { |
550 | zlog_err("%s: Message received from netlink is of a broken size %d %zu", | |
551 | __PRETTY_FUNCTION__, h->nlmsg_len, | |
552 | (size_t)NLMSG_LENGTH(sizeof(struct rtmsg))); | |
d62a17ae | 553 | return -1; |
9bdf8618 | 554 | } |
d62a17ae | 555 | |
556 | memset(tb, 0, sizeof tb); | |
557 | netlink_parse_rtattr(tb, RTA_MAX, RTM_RTA(rtm), len); | |
558 | ||
559 | if (rtm->rtm_flags & RTM_F_CLONED) | |
560 | return 0; | |
561 | if (rtm->rtm_protocol == RTPROT_REDIRECT) | |
562 | return 0; | |
563 | if (rtm->rtm_protocol == RTPROT_KERNEL) | |
564 | return 0; | |
565 | ||
566 | if (!startup && is_selfroute(rtm->rtm_protocol) | |
6ab5222f DS |
567 | && h->nlmsg_type == RTM_NEWROUTE) { |
568 | if (IS_ZEBRA_DEBUG_KERNEL) | |
569 | zlog_debug("Route type: %d Received that we think we have originated, ignoring", | |
570 | rtm->rtm_protocol); | |
d62a17ae | 571 | return 0; |
6ab5222f | 572 | } |
d62a17ae | 573 | |
574 | /* We don't care about change notifications for the MPLS table. */ | |
575 | /* TODO: Revisit this. */ | |
576 | if (rtm->rtm_family == AF_MPLS) | |
577 | return 0; | |
578 | ||
579 | /* Table corresponding to route. */ | |
580 | if (tb[RTA_TABLE]) | |
581 | table = *(int *)RTA_DATA(tb[RTA_TABLE]); | |
582 | else | |
583 | table = rtm->rtm_table; | |
584 | ||
585 | /* Map to VRF */ | |
78dd30b2 | 586 | vrf_id = vrf_lookup_by_table(table, ns_id); |
d62a17ae | 587 | if (vrf_id == VRF_DEFAULT) { |
588 | if (!is_zebra_valid_kernel_table(table) | |
589 | && !is_zebra_main_routing_table(table)) | |
590 | return 0; | |
591 | } | |
592 | ||
593 | /* Route which inserted by Zebra. */ | |
915902cb | 594 | if (is_selfroute(rtm->rtm_protocol)) { |
d62a17ae | 595 | flags |= ZEBRA_FLAG_SELFROUTE; |
38e40db1 | 596 | proto = proto2zebra(rtm->rtm_protocol, rtm->rtm_family, false); |
915902cb | 597 | } |
d62a17ae | 598 | if (tb[RTA_OIF]) |
599 | index = *(int *)RTA_DATA(tb[RTA_OIF]); | |
600 | ||
601 | if (tb[RTA_DST]) | |
602 | dest = RTA_DATA(tb[RTA_DST]); | |
603 | else | |
604 | dest = anyaddr; | |
605 | ||
606 | if (tb[RTA_SRC]) | |
607 | src = RTA_DATA(tb[RTA_SRC]); | |
608 | else | |
609 | src = anyaddr; | |
610 | ||
611 | if (tb[RTA_PREFSRC]) | |
612 | prefsrc = RTA_DATA(tb[RTA_PREFSRC]); | |
613 | ||
614 | if (tb[RTA_GATEWAY]) | |
615 | gate = RTA_DATA(tb[RTA_GATEWAY]); | |
616 | ||
fcc89a9c SW |
617 | if (tb[RTA_NH_ID]) |
618 | nhe_id = *(uint32_t *)RTA_DATA(tb[RTA_NH_ID]); | |
619 | ||
f19435a8 DS |
620 | if (tb[RTA_PRIORITY]) |
621 | metric = *(int *)RTA_DATA(tb[RTA_PRIORITY]); | |
d62a17ae | 622 | |
4e40b6d6 KK |
623 | #if defined(SUPPORT_REALMS) |
624 | if (tb[RTA_FLOW]) | |
625 | tag = *(uint32_t *)RTA_DATA(tb[RTA_FLOW]); | |
626 | #endif | |
627 | ||
f19435a8 DS |
628 | if (tb[RTA_METRICS]) { |
629 | struct rtattr *mxrta[RTAX_MAX + 1]; | |
d62a17ae | 630 | |
f19435a8 | 631 | memset(mxrta, 0, sizeof mxrta); |
996c9314 | 632 | netlink_parse_rtattr(mxrta, RTAX_MAX, RTA_DATA(tb[RTA_METRICS]), |
f19435a8 | 633 | RTA_PAYLOAD(tb[RTA_METRICS])); |
d62a17ae | 634 | |
f19435a8 | 635 | if (mxrta[RTAX_MTU]) |
d7c0a89a | 636 | mtu = *(uint32_t *)RTA_DATA(mxrta[RTAX_MTU]); |
d62a17ae | 637 | } |
638 | ||
639 | if (rtm->rtm_family == AF_INET) { | |
640 | p.family = AF_INET; | |
930571d2 | 641 | if (rtm->rtm_dst_len > IPV4_MAX_BITLEN) { |
e17d9b2d | 642 | zlog_err( |
75829703 | 643 | "Invalid destination prefix length: %u received from kernel route change", |
930571d2 | 644 | rtm->rtm_dst_len); |
e17d9b2d | 645 | return -1; |
930571d2 | 646 | } |
d62a17ae | 647 | memcpy(&p.u.prefix4, dest, 4); |
648 | p.prefixlen = rtm->rtm_dst_len; | |
649 | ||
1f610a1f CF |
650 | if (rtm->rtm_src_len != 0) { |
651 | char buf[PREFIX_STRLEN]; | |
9df414fe | 652 | flog_warn( |
e914ccbe | 653 | EC_ZEBRA_UNSUPPORTED_V4_SRCDEST, |
9df414fe QY |
654 | "unsupported IPv4 sourcedest route (dest %s vrf %u)", |
655 | prefix2str(&p, buf, sizeof(buf)), vrf_id); | |
1f610a1f CF |
656 | return 0; |
657 | } | |
930571d2 | 658 | |
1f610a1f CF |
659 | /* Force debug below to not display anything for source */ |
660 | src_p.prefixlen = 0; | |
d62a17ae | 661 | } else if (rtm->rtm_family == AF_INET6) { |
662 | p.family = AF_INET6; | |
930571d2 | 663 | if (rtm->rtm_dst_len > IPV6_MAX_BITLEN) { |
e17d9b2d | 664 | zlog_err( |
75829703 | 665 | "Invalid destination prefix length: %u received from kernel route change", |
930571d2 | 666 | rtm->rtm_dst_len); |
e17d9b2d | 667 | return -1; |
930571d2 | 668 | } |
d62a17ae | 669 | memcpy(&p.u.prefix6, dest, 16); |
670 | p.prefixlen = rtm->rtm_dst_len; | |
671 | ||
672 | src_p.family = AF_INET6; | |
930571d2 | 673 | if (rtm->rtm_src_len > IPV6_MAX_BITLEN) { |
e17d9b2d | 674 | zlog_err( |
75829703 | 675 | "Invalid source prefix length: %u received from kernel route change", |
930571d2 | 676 | rtm->rtm_src_len); |
e17d9b2d | 677 | return -1; |
930571d2 | 678 | } |
d62a17ae | 679 | memcpy(&src_p.prefix, src, 16); |
680 | src_p.prefixlen = rtm->rtm_src_len; | |
681 | } | |
682 | ||
25715c7e DS |
683 | /* |
684 | * For ZEBRA_ROUTE_KERNEL types: | |
685 | * | |
686 | * The metric/priority of the route received from the kernel | |
687 | * is a 32 bit number. We are going to interpret the high | |
688 | * order byte as the Admin Distance and the low order 3 bytes | |
689 | * as the metric. | |
690 | * | |
691 | * This will allow us to do two things: | |
692 | * 1) Allow the creation of kernel routes that can be | |
693 | * overridden by zebra. | |
694 | * 2) Allow the old behavior for 'most' kernel route types | |
695 | * if a user enters 'ip route ...' v4 routes get a metric | |
696 | * of 0 and v6 routes get a metric of 1024. Both of these | |
697 | * values will end up with a admin distance of 0, which | |
698 | * will cause them to win for the purposes of zebra. | |
699 | */ | |
700 | if (proto == ZEBRA_ROUTE_KERNEL) { | |
701 | distance = (metric >> 24) & 0xFF; | |
996c9314 | 702 | metric = (metric & 0x00FFFFFF); |
25715c7e DS |
703 | } |
704 | ||
d62a17ae | 705 | if (IS_ZEBRA_DEBUG_KERNEL) { |
706 | char buf[PREFIX_STRLEN]; | |
707 | char buf2[PREFIX_STRLEN]; | |
45df4e96 | 708 | zlog_debug("%s %s%s%s vrf %u(%u) metric: %d Admin Distance: %d", |
996c9314 LB |
709 | nl_msg_type_to_str(h->nlmsg_type), |
710 | prefix2str(&p, buf, sizeof(buf)), | |
711 | src_p.prefixlen ? " from " : "", | |
712 | src_p.prefixlen | |
713 | ? prefix2str(&src_p, buf2, sizeof(buf2)) | |
714 | : "", | |
45df4e96 | 715 | vrf_id, table, metric, distance); |
d62a17ae | 716 | } |
717 | ||
718 | afi_t afi = AFI_IP; | |
719 | if (rtm->rtm_family == AF_INET6) | |
720 | afi = AFI_IP6; | |
721 | ||
722 | if (h->nlmsg_type == RTM_NEWROUTE) { | |
8795f904 | 723 | |
fd36be7e | 724 | if (!tb[RTA_MULTIPATH]) { |
77a44d94 | 725 | struct nexthop nh = {0}; |
8795f904 | 726 | |
77a44d94 SW |
727 | if (!nhe_id) { |
728 | nh = parse_nexthop_unicast( | |
729 | ns_id, rtm, tb, bh_type, index, prefsrc, | |
20822f9d | 730 | gate, afi, vrf_id); |
87da6a60 | 731 | } |
4a7371e9 | 732 | rib_add(afi, SAFI_UNICAST, vrf_id, proto, 0, flags, &p, |
8032b717 SW |
733 | &src_p, &nh, nhe_id, table, metric, mtu, |
734 | distance, tag); | |
fd36be7e | 735 | } else { |
d62a17ae | 736 | /* This is a multipath route */ |
d62a17ae | 737 | struct route_entry *re; |
0eb97b86 | 738 | struct nexthop_group *ng = NULL; |
d62a17ae | 739 | struct rtnexthop *rtnh = |
740 | (struct rtnexthop *)RTA_DATA(tb[RTA_MULTIPATH]); | |
d62a17ae | 741 | |
742 | re = XCALLOC(MTYPE_RE, sizeof(struct route_entry)); | |
915902cb | 743 | re->type = proto; |
25715c7e | 744 | re->distance = distance; |
d62a17ae | 745 | re->flags = flags; |
746 | re->metric = metric; | |
747 | re->mtu = mtu; | |
748 | re->vrf_id = vrf_id; | |
749 | re->table = table; | |
98572489 | 750 | re->uptime = monotime(NULL); |
4e40b6d6 | 751 | re->tag = tag; |
bbb322f2 | 752 | re->nhe_id = nhe_id; |
3c04071d | 753 | |
20822f9d | 754 | if (!nhe_id) { |
0eb97b86 MS |
755 | uint8_t nhop_num; |
756 | ||
757 | /* Use temporary list of nexthops; parse | |
758 | * message payload's nexthops. | |
759 | */ | |
760 | ng = nexthop_group_new(); | |
761 | nhop_num = | |
20822f9d | 762 | parse_multipath_nexthops_unicast( |
0eb97b86 | 763 | ns_id, ng, rtm, rtnh, tb, |
20822f9d SW |
764 | prefsrc, vrf_id); |
765 | ||
766 | zserv_nexthop_num_warn( | |
767 | __func__, (const struct prefix *)&p, | |
768 | nhop_num); | |
0eb97b86 MS |
769 | |
770 | if (nhop_num == 0) { | |
771 | nexthop_group_delete(&ng); | |
772 | ng = NULL; | |
773 | } | |
d62a17ae | 774 | } |
775 | ||
0eb97b86 | 776 | if (nhe_id || ng) |
1f610a1f | 777 | rib_add_multipath(afi, SAFI_UNICAST, &p, |
0eb97b86 | 778 | &src_p, re, ng); |
20822f9d SW |
779 | else |
780 | XFREE(MTYPE_RE, re); | |
d62a17ae | 781 | } |
782 | } else { | |
bc541126 SW |
783 | if (nhe_id) { |
784 | rib_delete(afi, SAFI_UNICAST, vrf_id, proto, 0, flags, | |
785 | &p, &src_p, NULL, nhe_id, table, metric, | |
786 | distance, true); | |
787 | } else { | |
788 | if (!tb[RTA_MULTIPATH]) { | |
789 | struct nexthop nh; | |
760f39dc HS |
790 | |
791 | nh = parse_nexthop_unicast( | |
792 | ns_id, rtm, tb, bh_type, index, prefsrc, | |
793 | gate, afi, vrf_id); | |
bc541126 SW |
794 | rib_delete(afi, SAFI_UNICAST, vrf_id, proto, 0, |
795 | flags, &p, &src_p, &nh, 0, table, | |
796 | metric, distance, true); | |
8ba5bd58 | 797 | } else { |
bc541126 SW |
798 | /* XXX: need to compare the entire list of |
799 | * nexthops here for NLM_F_APPEND stupidity */ | |
800 | rib_delete(afi, SAFI_UNICAST, vrf_id, proto, 0, | |
801 | flags, &p, &src_p, NULL, 0, table, | |
802 | metric, distance, true); | |
8ba5bd58 | 803 | } |
d62a17ae | 804 | } |
805 | } | |
806 | ||
807 | return 0; | |
718e3744 | 808 | } |
809 | ||
e3be0432 DS |
810 | static struct mcast_route_data *mroute = NULL; |
811 | ||
2414abd3 | 812 | static int netlink_route_change_read_multicast(struct nlmsghdr *h, |
d62a17ae | 813 | ns_id_t ns_id, int startup) |
565fdc75 | 814 | { |
d62a17ae | 815 | int len; |
816 | struct rtmsg *rtm; | |
817 | struct rtattr *tb[RTA_MAX + 1]; | |
818 | struct mcast_route_data *m; | |
819 | struct mcast_route_data mr; | |
820 | int iif = 0; | |
821 | int count; | |
822 | int oif[256]; | |
823 | int oif_count = 0; | |
824 | char sbuf[40]; | |
825 | char gbuf[40]; | |
826 | char oif_list[256] = "\0"; | |
78dd30b2 | 827 | vrf_id_t vrf; |
43b5cc5e | 828 | int table; |
d62a17ae | 829 | |
830 | if (mroute) | |
831 | m = mroute; | |
832 | else { | |
833 | memset(&mr, 0, sizeof(mr)); | |
834 | m = &mr; | |
835 | } | |
836 | ||
837 | rtm = NLMSG_DATA(h); | |
838 | ||
839 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct rtmsg)); | |
840 | ||
841 | memset(tb, 0, sizeof tb); | |
842 | netlink_parse_rtattr(tb, RTA_MAX, RTM_RTA(rtm), len); | |
90d82769 | 843 | |
43b5cc5e DS |
844 | if (tb[RTA_TABLE]) |
845 | table = *(int *)RTA_DATA(tb[RTA_TABLE]); | |
846 | else | |
847 | table = rtm->rtm_table; | |
848 | ||
78dd30b2 | 849 | vrf = vrf_lookup_by_table(table, ns_id); |
43b5cc5e | 850 | |
d62a17ae | 851 | if (tb[RTA_IIF]) |
852 | iif = *(int *)RTA_DATA(tb[RTA_IIF]); | |
853 | ||
854 | if (tb[RTA_SRC]) | |
bd8b9272 | 855 | m->sg.src = *(struct in_addr *)RTA_DATA(tb[RTA_SRC]); |
d62a17ae | 856 | |
857 | if (tb[RTA_DST]) | |
bd8b9272 | 858 | m->sg.grp = *(struct in_addr *)RTA_DATA(tb[RTA_DST]); |
d62a17ae | 859 | |
62819462 | 860 | if (tb[RTA_EXPIRES]) |
d62a17ae | 861 | m->lastused = *(unsigned long long *)RTA_DATA(tb[RTA_EXPIRES]); |
862 | ||
863 | if (tb[RTA_MULTIPATH]) { | |
864 | struct rtnexthop *rtnh = | |
865 | (struct rtnexthop *)RTA_DATA(tb[RTA_MULTIPATH]); | |
866 | ||
867 | len = RTA_PAYLOAD(tb[RTA_MULTIPATH]); | |
868 | for (;;) { | |
869 | if (len < (int)sizeof(*rtnh) || rtnh->rtnh_len > len) | |
870 | break; | |
871 | ||
872 | oif[oif_count] = rtnh->rtnh_ifindex; | |
873 | oif_count++; | |
874 | ||
3c04071d SW |
875 | if (rtnh->rtnh_len == 0) |
876 | break; | |
877 | ||
d62a17ae | 878 | len -= NLMSG_ALIGN(rtnh->rtnh_len); |
879 | rtnh = RTNH_NEXT(rtnh); | |
880 | } | |
881 | } | |
882 | ||
883 | if (IS_ZEBRA_DEBUG_KERNEL) { | |
822c9af2 SW |
884 | struct interface *ifp = NULL; |
885 | struct zebra_vrf *zvrf = NULL; | |
886 | ||
0af35d90 RW |
887 | strlcpy(sbuf, inet_ntoa(m->sg.src), sizeof(sbuf)); |
888 | strlcpy(gbuf, inet_ntoa(m->sg.grp), sizeof(gbuf)); | |
d62a17ae | 889 | for (count = 0; count < oif_count; count++) { |
890 | ifp = if_lookup_by_index(oif[count], vrf); | |
891 | char temp[256]; | |
892 | ||
5b4256ca DS |
893 | sprintf(temp, "%s(%d) ", ifp ? ifp->name : "Unknown", |
894 | oif[count]); | |
eab4a5c2 | 895 | strlcat(oif_list, temp, sizeof(oif_list)); |
d62a17ae | 896 | } |
822c9af2 | 897 | zvrf = zebra_vrf_lookup_by_id(vrf); |
d62a17ae | 898 | ifp = if_lookup_by_index(iif, vrf); |
822c9af2 SW |
899 | zlog_debug( |
900 | "MCAST VRF: %s(%d) %s (%s,%s) IIF: %s(%d) OIF: %s jiffies: %lld", | |
901 | (zvrf ? zvrf->vrf->name : "Unknown"), vrf, | |
902 | nl_msg_type_to_str(h->nlmsg_type), sbuf, gbuf, | |
903 | ifp ? ifp->name : "Unknown", iif, oif_list, | |
904 | m->lastused); | |
90d82769 | 905 | } |
d62a17ae | 906 | return 0; |
565fdc75 DS |
907 | } |
908 | ||
2414abd3 | 909 | int netlink_route_change(struct nlmsghdr *h, ns_id_t ns_id, int startup) |
565fdc75 | 910 | { |
d62a17ae | 911 | int len; |
d62a17ae | 912 | struct rtmsg *rtm; |
913 | ||
914 | rtm = NLMSG_DATA(h); | |
915 | ||
916 | if (!(h->nlmsg_type == RTM_NEWROUTE || h->nlmsg_type == RTM_DELROUTE)) { | |
917 | /* If this is not route add/delete message print warning. */ | |
9165c5f5 | 918 | zlog_debug("Kernel message: %s NS %u", |
87b5d1b0 | 919 | nl_msg_type_to_str(h->nlmsg_type), ns_id); |
d62a17ae | 920 | return 0; |
921 | } | |
922 | ||
c25e2f1a DS |
923 | if (!(rtm->rtm_family == AF_INET || |
924 | rtm->rtm_family == AF_INET6 || | |
925 | rtm->rtm_family == RTNL_FAMILY_IPMR )) { | |
9df414fe | 926 | flog_warn( |
e914ccbe | 927 | EC_ZEBRA_UNKNOWN_FAMILY, |
87b5d1b0 DS |
928 | "Invalid address family: %u received from kernel route change: %s", |
929 | rtm->rtm_family, nl_msg_type_to_str(h->nlmsg_type)); | |
8a1b681c SW |
930 | return 0; |
931 | } | |
932 | ||
d62a17ae | 933 | /* Connected route. */ |
934 | if (IS_ZEBRA_DEBUG_KERNEL) | |
78dd30b2 | 935 | zlog_debug("%s %s %s proto %s NS %u", |
d62a17ae | 936 | nl_msg_type_to_str(h->nlmsg_type), |
937 | nl_family_to_str(rtm->rtm_family), | |
938 | nl_rttype_to_str(rtm->rtm_type), | |
78dd30b2 | 939 | nl_rtproto_to_str(rtm->rtm_protocol), ns_id); |
d62a17ae | 940 | |
d62a17ae | 941 | |
942 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct rtmsg)); | |
9bdf8618 DS |
943 | if (len < 0) { |
944 | zlog_err("%s: Message received from netlink is of a broken size: %d %zu", | |
945 | __PRETTY_FUNCTION__, | |
946 | h->nlmsg_len, | |
947 | (size_t)NLMSG_LENGTH(sizeof(struct rtmsg))); | |
d62a17ae | 948 | return -1; |
9bdf8618 | 949 | } |
d62a17ae | 950 | |
e655a03c | 951 | if (rtm->rtm_type == RTN_MULTICAST) |
2414abd3 | 952 | netlink_route_change_read_multicast(h, ns_id, startup); |
e655a03c | 953 | else |
2414abd3 | 954 | netlink_route_change_read_unicast(h, ns_id, startup); |
d62a17ae | 955 | return 0; |
565fdc75 DS |
956 | } |
957 | ||
289602d7 | 958 | /* Request for specific route information from the kernel */ |
d62a17ae | 959 | static int netlink_request_route(struct zebra_ns *zns, int family, int type) |
289602d7 | 960 | { |
d62a17ae | 961 | struct { |
962 | struct nlmsghdr n; | |
963 | struct rtmsg rtm; | |
964 | } req; | |
965 | ||
966 | /* Form the request, specifying filter (rtattr) if needed. */ | |
967 | memset(&req, 0, sizeof(req)); | |
968 | req.n.nlmsg_type = type; | |
718f9b0f | 969 | req.n.nlmsg_flags = NLM_F_ROOT | NLM_F_MATCH | NLM_F_REQUEST; |
d62a17ae | 970 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg)); |
971 | req.rtm.rtm_family = family; | |
972 | ||
973 | return netlink_request(&zns->netlink_cmd, &req.n); | |
289602d7 | 974 | } |
975 | ||
718e3744 | 976 | /* Routing table read function using netlink interface. Only called |
977 | bootstrap time. */ | |
d62a17ae | 978 | int netlink_route_read(struct zebra_ns *zns) |
718e3744 | 979 | { |
d62a17ae | 980 | int ret; |
85a75f1e MS |
981 | struct zebra_dplane_info dp_info; |
982 | ||
983 | zebra_dplane_info_from_zns(&dp_info, zns, true /*is_cmd*/); | |
d62a17ae | 984 | |
985 | /* Get IPv4 routing table. */ | |
986 | ret = netlink_request_route(zns, AF_INET, RTM_GETROUTE); | |
987 | if (ret < 0) | |
988 | return ret; | |
989 | ret = netlink_parse_info(netlink_route_change_read_unicast, | |
85a75f1e | 990 | &zns->netlink_cmd, &dp_info, 0, 1); |
d62a17ae | 991 | if (ret < 0) |
992 | return ret; | |
993 | ||
994 | /* Get IPv6 routing table. */ | |
995 | ret = netlink_request_route(zns, AF_INET6, RTM_GETROUTE); | |
996 | if (ret < 0) | |
997 | return ret; | |
998 | ret = netlink_parse_info(netlink_route_change_read_unicast, | |
85a75f1e | 999 | &zns->netlink_cmd, &dp_info, 0, 1); |
d62a17ae | 1000 | if (ret < 0) |
1001 | return ret; | |
1002 | ||
1003 | return 0; | |
718e3744 | 1004 | } |
1005 | ||
d7c0a89a QY |
1006 | static void _netlink_route_nl_add_gateway_info(uint8_t route_family, |
1007 | uint8_t gw_family, | |
d62a17ae | 1008 | struct nlmsghdr *nlmsg, |
1009 | size_t req_size, int bytelen, | |
81793ac1 | 1010 | const struct nexthop *nexthop) |
40c7bdb0 | 1011 | { |
d62a17ae | 1012 | if (route_family == AF_MPLS) { |
1013 | struct gw_family_t gw_fam; | |
1014 | ||
1015 | gw_fam.family = gw_family; | |
1016 | if (gw_family == AF_INET) | |
1017 | memcpy(&gw_fam.gate.ipv4, &nexthop->gate.ipv4, bytelen); | |
1018 | else | |
1019 | memcpy(&gw_fam.gate.ipv6, &nexthop->gate.ipv6, bytelen); | |
1020 | addattr_l(nlmsg, req_size, RTA_VIA, &gw_fam.family, | |
1021 | bytelen + 2); | |
1022 | } else { | |
1023 | if (gw_family == AF_INET) | |
1024 | addattr_l(nlmsg, req_size, RTA_GATEWAY, | |
1025 | &nexthop->gate.ipv4, bytelen); | |
1026 | else | |
1027 | addattr_l(nlmsg, req_size, RTA_GATEWAY, | |
1028 | &nexthop->gate.ipv6, bytelen); | |
1029 | } | |
40c7bdb0 | 1030 | } |
1031 | ||
d7c0a89a QY |
1032 | static void _netlink_route_rta_add_gateway_info(uint8_t route_family, |
1033 | uint8_t gw_family, | |
d62a17ae | 1034 | struct rtattr *rta, |
1035 | struct rtnexthop *rtnh, | |
1036 | size_t req_size, int bytelen, | |
81793ac1 | 1037 | const struct nexthop *nexthop) |
40c7bdb0 | 1038 | { |
d62a17ae | 1039 | if (route_family == AF_MPLS) { |
1040 | struct gw_family_t gw_fam; | |
1041 | ||
1042 | gw_fam.family = gw_family; | |
1043 | if (gw_family == AF_INET) | |
1044 | memcpy(&gw_fam.gate.ipv4, &nexthop->gate.ipv4, bytelen); | |
1045 | else | |
1046 | memcpy(&gw_fam.gate.ipv6, &nexthop->gate.ipv6, bytelen); | |
1047 | rta_addattr_l(rta, req_size, RTA_VIA, &gw_fam.family, | |
1048 | bytelen + 2); | |
1049 | rtnh->rtnh_len += RTA_LENGTH(bytelen + 2); | |
1050 | } else { | |
1051 | if (gw_family == AF_INET) | |
1052 | rta_addattr_l(rta, req_size, RTA_GATEWAY, | |
1053 | &nexthop->gate.ipv4, bytelen); | |
1054 | else | |
1055 | rta_addattr_l(rta, req_size, RTA_GATEWAY, | |
1056 | &nexthop->gate.ipv6, bytelen); | |
1057 | rtnh->rtnh_len += sizeof(struct rtattr) + bytelen; | |
1058 | } | |
40c7bdb0 | 1059 | } |
1060 | ||
b7537db6 SW |
1061 | static int build_label_stack(struct mpls_label_stack *nh_label, |
1062 | mpls_lse_t *out_lse, char *label_buf, | |
1063 | size_t label_buf_size) | |
1064 | { | |
1065 | char label_buf1[20]; | |
1066 | int num_labels = 0; | |
1067 | ||
1068 | for (int i = 0; nh_label && i < nh_label->num_labels; i++) { | |
1069 | if (nh_label->label[i] == MPLS_LABEL_IMPLICIT_NULL) | |
1070 | continue; | |
1071 | ||
1072 | if (IS_ZEBRA_DEBUG_KERNEL) { | |
1073 | if (!num_labels) | |
1074 | sprintf(label_buf, "label %u", | |
1075 | nh_label->label[i]); | |
1076 | else { | |
1077 | sprintf(label_buf1, "/%u", nh_label->label[i]); | |
1078 | strlcat(label_buf, label_buf1, label_buf_size); | |
1079 | } | |
1080 | } | |
1081 | ||
1082 | out_lse[num_labels] = | |
1083 | mpls_lse_encode(nh_label->label[i], 0, 0, 0); | |
1084 | num_labels++; | |
1085 | } | |
1086 | ||
1087 | return num_labels; | |
1088 | } | |
1089 | ||
fa713d9e CF |
1090 | /* This function takes a nexthop as argument and adds |
1091 | * the appropriate netlink attributes to an existing | |
1092 | * netlink message. | |
1093 | * | |
1094 | * @param routedesc: Human readable description of route type | |
1095 | * (direct/recursive, single-/multipath) | |
1096 | * @param bytelen: Length of addresses in bytes. | |
1097 | * @param nexthop: Nexthop information | |
1098 | * @param nlmsg: nlmsghdr structure to fill in. | |
1099 | * @param req_size: The size allocated for the message. | |
1100 | */ | |
d62a17ae | 1101 | static void _netlink_route_build_singlepath(const char *routedesc, int bytelen, |
81793ac1 | 1102 | const struct nexthop *nexthop, |
d62a17ae | 1103 | struct nlmsghdr *nlmsg, |
1104 | struct rtmsg *rtmsg, | |
1105 | size_t req_size, int cmd) | |
fa713d9e | 1106 | { |
b7537db6 | 1107 | |
d62a17ae | 1108 | mpls_lse_t out_lse[MPLS_MAX_LABELS]; |
9a62e84b | 1109 | char label_buf[256]; |
b7537db6 SW |
1110 | int num_labels = 0; |
1111 | ||
1112 | assert(nexthop); | |
d62a17ae | 1113 | |
1114 | /* | |
1115 | * label_buf is *only* currently used within debugging. | |
1116 | * As such when we assign it we are guarding it inside | |
1117 | * a debug test. If you want to change this make sure | |
1118 | * you fix this assumption | |
1119 | */ | |
1120 | label_buf[0] = '\0'; | |
d62a17ae | 1121 | |
b7537db6 SW |
1122 | num_labels = build_label_stack(nexthop->nh_label, out_lse, label_buf, |
1123 | sizeof(label_buf)); | |
fa712963 RW |
1124 | |
1125 | if (num_labels) { | |
1126 | /* Set the BoS bit */ | |
1127 | out_lse[num_labels - 1] |= htonl(1 << MPLS_LS_S_SHIFT); | |
1128 | ||
1129 | if (rtmsg->rtm_family == AF_MPLS) | |
1130 | addattr_l(nlmsg, req_size, RTA_NEWDST, &out_lse, | |
1131 | num_labels * sizeof(mpls_lse_t)); | |
1132 | else { | |
1133 | struct rtattr *nest; | |
d7c0a89a | 1134 | uint16_t encap = LWTUNNEL_ENCAP_MPLS; |
fa712963 RW |
1135 | |
1136 | addattr_l(nlmsg, req_size, RTA_ENCAP_TYPE, &encap, | |
d7c0a89a | 1137 | sizeof(uint16_t)); |
fa712963 RW |
1138 | nest = addattr_nest(nlmsg, req_size, RTA_ENCAP); |
1139 | addattr_l(nlmsg, req_size, MPLS_IPTUNNEL_DST, &out_lse, | |
1140 | num_labels * sizeof(mpls_lse_t)); | |
1141 | addattr_nest_end(nlmsg, nest); | |
66d42727 | 1142 | } |
0aabccc0 | 1143 | } |
fa713d9e | 1144 | |
d62a17ae | 1145 | if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ONLINK)) |
1146 | rtmsg->rtm_flags |= RTNH_F_ONLINK; | |
1147 | ||
1148 | if (rtmsg->rtm_family == AF_INET | |
1149 | && (nexthop->type == NEXTHOP_TYPE_IPV6 | |
1150 | || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX)) { | |
1151 | rtmsg->rtm_flags |= RTNH_F_ONLINK; | |
1152 | addattr_l(nlmsg, req_size, RTA_GATEWAY, &ipv4_ll, 4); | |
1153 | addattr32(nlmsg, req_size, RTA_OIF, nexthop->ifindex); | |
1154 | ||
1155 | if (nexthop->rmap_src.ipv4.s_addr && (cmd == RTM_NEWROUTE)) | |
1156 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1157 | &nexthop->rmap_src.ipv4, bytelen); | |
1158 | else if (nexthop->src.ipv4.s_addr && (cmd == RTM_NEWROUTE)) | |
1159 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1160 | &nexthop->src.ipv4, bytelen); | |
1161 | ||
1162 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1163 | zlog_debug( | |
1164 | " 5549: _netlink_route_build_singlepath() (%s): " | |
7556c3fd | 1165 | "nexthop via %s %s if %u(%u)", |
d62a17ae | 1166 | routedesc, ipv4_ll_buf, label_buf, |
7556c3fd | 1167 | nexthop->ifindex, nexthop->vrf_id); |
d62a17ae | 1168 | return; |
0aabccc0 DD |
1169 | } |
1170 | ||
d62a17ae | 1171 | if (nexthop->type == NEXTHOP_TYPE_IPV4 |
1172 | || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX) { | |
1173 | /* Send deletes to the kernel without specifying the next-hop */ | |
1174 | if (cmd != RTM_DELROUTE) | |
1175 | _netlink_route_nl_add_gateway_info( | |
1176 | rtmsg->rtm_family, AF_INET, nlmsg, req_size, | |
1177 | bytelen, nexthop); | |
1178 | ||
1179 | if (cmd == RTM_NEWROUTE) { | |
1180 | if (nexthop->rmap_src.ipv4.s_addr) | |
1181 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1182 | &nexthop->rmap_src.ipv4, bytelen); | |
1183 | else if (nexthop->src.ipv4.s_addr) | |
1184 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1185 | &nexthop->src.ipv4, bytelen); | |
1186 | } | |
1187 | ||
1188 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1189 | zlog_debug( | |
1190 | "netlink_route_multipath() (%s): " | |
7556c3fd | 1191 | "nexthop via %s %s if %u(%u)", |
d62a17ae | 1192 | routedesc, inet_ntoa(nexthop->gate.ipv4), |
7556c3fd | 1193 | label_buf, nexthop->ifindex, nexthop->vrf_id); |
0aabccc0 | 1194 | } |
fa713d9e | 1195 | |
d62a17ae | 1196 | if (nexthop->type == NEXTHOP_TYPE_IPV6 |
1197 | || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) { | |
1198 | _netlink_route_nl_add_gateway_info(rtmsg->rtm_family, AF_INET6, | |
1199 | nlmsg, req_size, bytelen, | |
1200 | nexthop); | |
1201 | ||
1202 | if (cmd == RTM_NEWROUTE) { | |
1203 | if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->rmap_src.ipv6)) | |
1204 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1205 | &nexthop->rmap_src.ipv6, bytelen); | |
1206 | else if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->src.ipv6)) | |
1207 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1208 | &nexthop->src.ipv6, bytelen); | |
1209 | } | |
fa713d9e | 1210 | |
d62a17ae | 1211 | if (IS_ZEBRA_DEBUG_KERNEL) |
1212 | zlog_debug( | |
1213 | "netlink_route_multipath() (%s): " | |
7556c3fd | 1214 | "nexthop via %s %s if %u(%u)", |
d62a17ae | 1215 | routedesc, inet6_ntoa(nexthop->gate.ipv6), |
7556c3fd | 1216 | label_buf, nexthop->ifindex, nexthop->vrf_id); |
d62a17ae | 1217 | } |
5e210522 DS |
1218 | |
1219 | /* | |
1220 | * We have the ifindex so we should always send it | |
1221 | * This is especially useful if we are doing route | |
1222 | * leaking. | |
1223 | */ | |
1224 | if (nexthop->type != NEXTHOP_TYPE_BLACKHOLE) | |
d62a17ae | 1225 | addattr32(nlmsg, req_size, RTA_OIF, nexthop->ifindex); |
1226 | ||
275565fb | 1227 | if (nexthop->type == NEXTHOP_TYPE_IFINDEX) { |
d62a17ae | 1228 | if (cmd == RTM_NEWROUTE) { |
1229 | if (nexthop->rmap_src.ipv4.s_addr) | |
1230 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1231 | &nexthop->rmap_src.ipv4, bytelen); | |
1232 | else if (nexthop->src.ipv4.s_addr) | |
1233 | addattr_l(nlmsg, req_size, RTA_PREFSRC, | |
1234 | &nexthop->src.ipv4, bytelen); | |
1235 | } | |
fa713d9e | 1236 | |
d62a17ae | 1237 | if (IS_ZEBRA_DEBUG_KERNEL) |
1238 | zlog_debug( | |
1239 | "netlink_route_multipath() (%s): " | |
7556c3fd DS |
1240 | "nexthop via if %u(%u)", |
1241 | routedesc, nexthop->ifindex, nexthop->vrf_id); | |
0aabccc0 | 1242 | } |
fa713d9e CF |
1243 | } |
1244 | ||
1245 | /* This function takes a nexthop as argument and | |
1246 | * appends to the given rtattr/rtnexthop pair the | |
1247 | * representation of the nexthop. If the nexthop | |
1248 | * defines a preferred source, the src parameter | |
1249 | * will be modified to point to that src, otherwise | |
1250 | * it will be kept unmodified. | |
1251 | * | |
1252 | * @param routedesc: Human readable description of route type | |
1253 | * (direct/recursive, single-/multipath) | |
1254 | * @param bytelen: Length of addresses in bytes. | |
1255 | * @param nexthop: Nexthop information | |
1256 | * @param rta: rtnetlink attribute structure | |
1257 | * @param rtnh: pointer to an rtnetlink nexthop structure | |
1258 | * @param src: pointer pointing to a location where | |
1259 | * the prefsrc should be stored. | |
1260 | */ | |
d62a17ae | 1261 | static void _netlink_route_build_multipath(const char *routedesc, int bytelen, |
81793ac1 | 1262 | const struct nexthop *nexthop, |
d62a17ae | 1263 | struct rtattr *rta, |
1264 | struct rtnexthop *rtnh, | |
1265 | struct rtmsg *rtmsg, | |
81793ac1 | 1266 | const union g_addr **src) |
fa713d9e | 1267 | { |
d62a17ae | 1268 | mpls_lse_t out_lse[MPLS_MAX_LABELS]; |
9a62e84b | 1269 | char label_buf[256]; |
b7537db6 | 1270 | int num_labels = 0; |
d62a17ae | 1271 | |
1272 | rtnh->rtnh_len = sizeof(*rtnh); | |
1273 | rtnh->rtnh_flags = 0; | |
1274 | rtnh->rtnh_hops = 0; | |
1275 | rta->rta_len += rtnh->rtnh_len; | |
1276 | ||
b7537db6 SW |
1277 | assert(nexthop); |
1278 | ||
d62a17ae | 1279 | /* |
1280 | * label_buf is *only* currently used within debugging. | |
1281 | * As such when we assign it we are guarding it inside | |
1282 | * a debug test. If you want to change this make sure | |
1283 | * you fix this assumption | |
1284 | */ | |
1285 | label_buf[0] = '\0'; | |
d62a17ae | 1286 | |
b7537db6 SW |
1287 | num_labels = build_label_stack(nexthop->nh_label, out_lse, label_buf, |
1288 | sizeof(label_buf)); | |
fa712963 RW |
1289 | |
1290 | if (num_labels) { | |
1291 | /* Set the BoS bit */ | |
1292 | out_lse[num_labels - 1] |= htonl(1 << MPLS_LS_S_SHIFT); | |
1293 | ||
1294 | if (rtmsg->rtm_family == AF_MPLS) { | |
1295 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, RTA_NEWDST, | |
1296 | &out_lse, | |
1297 | num_labels * sizeof(mpls_lse_t)); | |
1298 | rtnh->rtnh_len += | |
1299 | RTA_LENGTH(num_labels * sizeof(mpls_lse_t)); | |
1300 | } else { | |
1301 | struct rtattr *nest; | |
d7c0a89a | 1302 | uint16_t encap = LWTUNNEL_ENCAP_MPLS; |
fa712963 RW |
1303 | int len = rta->rta_len; |
1304 | ||
1305 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, RTA_ENCAP_TYPE, | |
d7c0a89a | 1306 | &encap, sizeof(uint16_t)); |
fa712963 RW |
1307 | nest = rta_nest(rta, NL_PKT_BUF_SIZE, RTA_ENCAP); |
1308 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, MPLS_IPTUNNEL_DST, | |
1309 | &out_lse, | |
1310 | num_labels * sizeof(mpls_lse_t)); | |
1311 | rta_nest_end(rta, nest); | |
1312 | rtnh->rtnh_len += rta->rta_len - len; | |
66d42727 | 1313 | } |
d62a17ae | 1314 | } |
1315 | ||
1316 | if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ONLINK)) | |
1317 | rtnh->rtnh_flags |= RTNH_F_ONLINK; | |
1318 | ||
1319 | if (rtmsg->rtm_family == AF_INET | |
1320 | && (nexthop->type == NEXTHOP_TYPE_IPV6 | |
1321 | || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX)) { | |
1322 | bytelen = 4; | |
1323 | rtnh->rtnh_flags |= RTNH_F_ONLINK; | |
1324 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, RTA_GATEWAY, &ipv4_ll, | |
1325 | bytelen); | |
1326 | rtnh->rtnh_len += sizeof(struct rtattr) + bytelen; | |
1327 | rtnh->rtnh_ifindex = nexthop->ifindex; | |
1328 | ||
1329 | if (nexthop->rmap_src.ipv4.s_addr) | |
1330 | *src = &nexthop->rmap_src; | |
1331 | else if (nexthop->src.ipv4.s_addr) | |
1332 | *src = &nexthop->src; | |
1333 | ||
1334 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1335 | zlog_debug( | |
1336 | " 5549: netlink_route_build_multipath() (%s): " | |
1337 | "nexthop via %s %s if %u", | |
1338 | routedesc, ipv4_ll_buf, label_buf, | |
1339 | nexthop->ifindex); | |
1340 | return; | |
1341 | } | |
1342 | ||
1343 | if (nexthop->type == NEXTHOP_TYPE_IPV4 | |
1344 | || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX) { | |
1345 | _netlink_route_rta_add_gateway_info(rtmsg->rtm_family, AF_INET, | |
1346 | rta, rtnh, NL_PKT_BUF_SIZE, | |
1347 | bytelen, nexthop); | |
1348 | if (nexthop->rmap_src.ipv4.s_addr) | |
1349 | *src = &nexthop->rmap_src; | |
1350 | else if (nexthop->src.ipv4.s_addr) | |
1351 | *src = &nexthop->src; | |
1352 | ||
1353 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1354 | zlog_debug( | |
1355 | "netlink_route_multipath() (%s): " | |
1356 | "nexthop via %s %s if %u", | |
1357 | routedesc, inet_ntoa(nexthop->gate.ipv4), | |
1358 | label_buf, nexthop->ifindex); | |
1359 | } | |
1360 | if (nexthop->type == NEXTHOP_TYPE_IPV6 | |
1361 | || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) { | |
1362 | _netlink_route_rta_add_gateway_info(rtmsg->rtm_family, AF_INET6, | |
1363 | rta, rtnh, NL_PKT_BUF_SIZE, | |
1364 | bytelen, nexthop); | |
1365 | ||
1366 | if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->rmap_src.ipv6)) | |
1367 | *src = &nexthop->rmap_src; | |
1368 | else if (!IN6_IS_ADDR_UNSPECIFIED(&nexthop->src.ipv6)) | |
1369 | *src = &nexthop->src; | |
1370 | ||
1371 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1372 | zlog_debug( | |
1373 | "netlink_route_multipath() (%s): " | |
1374 | "nexthop via %s %s if %u", | |
1375 | routedesc, inet6_ntoa(nexthop->gate.ipv6), | |
1376 | label_buf, nexthop->ifindex); | |
1377 | } | |
5e210522 DS |
1378 | |
1379 | /* | |
1380 | * We have figured out the ifindex so we should always send it | |
1381 | * This is especially useful if we are doing route | |
1382 | * leaking. | |
1383 | */ | |
1384 | if (nexthop->type != NEXTHOP_TYPE_BLACKHOLE) | |
1385 | rtnh->rtnh_ifindex = nexthop->ifindex; | |
1386 | ||
d62a17ae | 1387 | /* ifindex */ |
275565fb | 1388 | if (nexthop->type == NEXTHOP_TYPE_IFINDEX) { |
d62a17ae | 1389 | if (nexthop->rmap_src.ipv4.s_addr) |
1390 | *src = &nexthop->rmap_src; | |
1391 | else if (nexthop->src.ipv4.s_addr) | |
1392 | *src = &nexthop->src; | |
1393 | ||
1394 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1395 | zlog_debug( | |
1396 | "netlink_route_multipath() (%s): " | |
1397 | "nexthop via if %u", | |
1398 | routedesc, nexthop->ifindex); | |
d62a17ae | 1399 | } |
df7fb580 DS |
1400 | |
1401 | if (nexthop->weight) | |
1402 | rtnh->rtnh_hops = nexthop->weight - 1; | |
fa713d9e CF |
1403 | } |
1404 | ||
d62a17ae | 1405 | static inline void _netlink_mpls_build_singlepath(const char *routedesc, |
81793ac1 | 1406 | const zebra_nhlfe_t *nhlfe, |
d62a17ae | 1407 | struct nlmsghdr *nlmsg, |
1408 | struct rtmsg *rtmsg, | |
1409 | size_t req_size, int cmd) | |
40c7bdb0 | 1410 | { |
d62a17ae | 1411 | int bytelen; |
d7c0a89a | 1412 | uint8_t family; |
40c7bdb0 | 1413 | |
d62a17ae | 1414 | family = NHLFE_FAMILY(nhlfe); |
1415 | bytelen = (family == AF_INET ? 4 : 16); | |
1416 | _netlink_route_build_singlepath(routedesc, bytelen, nhlfe->nexthop, | |
1417 | nlmsg, rtmsg, req_size, cmd); | |
40c7bdb0 | 1418 | } |
1419 | ||
1420 | ||
1421 | static inline void | |
81793ac1 | 1422 | _netlink_mpls_build_multipath(const char *routedesc, const zebra_nhlfe_t *nhlfe, |
d62a17ae | 1423 | struct rtattr *rta, struct rtnexthop *rtnh, |
81793ac1 | 1424 | struct rtmsg *rtmsg, const union g_addr **src) |
40c7bdb0 | 1425 | { |
d62a17ae | 1426 | int bytelen; |
d7c0a89a | 1427 | uint8_t family; |
40c7bdb0 | 1428 | |
d62a17ae | 1429 | family = NHLFE_FAMILY(nhlfe); |
1430 | bytelen = (family == AF_INET ? 4 : 16); | |
1431 | _netlink_route_build_multipath(routedesc, bytelen, nhlfe->nexthop, rta, | |
1432 | rtnh, rtmsg, src); | |
40c7bdb0 | 1433 | } |
1434 | ||
1435 | ||
fa713d9e CF |
1436 | /* Log debug information for netlink_route_multipath |
1437 | * if debug logging is enabled. | |
1438 | * | |
1439 | * @param cmd: Netlink command which is to be processed | |
1440 | * @param p: Prefix for which the change is due | |
fa713d9e | 1441 | * @param family: Address family which the change concerns |
45df4e96 DS |
1442 | * @param zvrf: The vrf we are in |
1443 | * @param tableid: The table we are working on | |
fa713d9e | 1444 | */ |
86391e56 MS |
1445 | static void _netlink_route_debug(int cmd, const struct prefix *p, |
1446 | int family, vrf_id_t vrfid, | |
7556c3fd | 1447 | uint32_t tableid) |
fa713d9e | 1448 | { |
d62a17ae | 1449 | if (IS_ZEBRA_DEBUG_KERNEL) { |
1450 | char buf[PREFIX_STRLEN]; | |
1451 | zlog_debug( | |
45df4e96 DS |
1452 | "netlink_route_multipath(): %s %s vrf %u(%u)", |
1453 | nl_msg_type_to_str(cmd), | |
1454 | prefix2str(p, buf, sizeof(buf)), | |
86391e56 | 1455 | vrfid, tableid); |
d62a17ae | 1456 | } |
1457 | } | |
1458 | ||
e8968ccb SW |
1459 | static void _netlink_nexthop_debug(int cmd, uint32_t id) |
1460 | { | |
1461 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1462 | zlog_debug("netlink_nexthop(): %s, id=%u", | |
1463 | nl_msg_type_to_str(cmd), id); | |
1464 | } | |
1465 | ||
d7c0a89a | 1466 | static void _netlink_mpls_debug(int cmd, uint32_t label, const char *routedesc) |
40c7bdb0 | 1467 | { |
d62a17ae | 1468 | if (IS_ZEBRA_DEBUG_KERNEL) |
1469 | zlog_debug("netlink_mpls_multipath() (%s): %s %u/20", routedesc, | |
1470 | nl_msg_type_to_str(cmd), label); | |
fa713d9e CF |
1471 | } |
1472 | ||
d62a17ae | 1473 | static int netlink_neigh_update(int cmd, int ifindex, uint32_t addr, char *lla, |
5895d33f | 1474 | int llalen, ns_id_t ns_id) |
5c610faf | 1475 | { |
f3dbec60 | 1476 | uint8_t protocol = RTPROT_ZEBRA; |
d62a17ae | 1477 | struct { |
1478 | struct nlmsghdr n; | |
1479 | struct ndmsg ndm; | |
1480 | char buf[256]; | |
1481 | } req; | |
5c610faf | 1482 | |
5895d33f | 1483 | struct zebra_ns *zns = zebra_ns_lookup(ns_id); |
8f7d9fc0 | 1484 | |
5605ecfc | 1485 | memset(&req, 0, sizeof(req)); |
5c610faf | 1486 | |
d62a17ae | 1487 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); |
1488 | req.n.nlmsg_flags = NLM_F_CREATE | NLM_F_REQUEST; | |
1489 | req.n.nlmsg_type = cmd; // RTM_NEWNEIGH or RTM_DELNEIGH | |
1490 | req.n.nlmsg_pid = zns->netlink_cmd.snl.nl_pid; | |
a55ba23f | 1491 | |
d62a17ae | 1492 | req.ndm.ndm_family = AF_INET; |
1493 | req.ndm.ndm_state = NUD_PERMANENT; | |
1494 | req.ndm.ndm_ifindex = ifindex; | |
1495 | req.ndm.ndm_type = RTN_UNICAST; | |
5c610faf | 1496 | |
f3dbec60 DS |
1497 | addattr_l(&req.n, sizeof(req), |
1498 | NDA_PROTOCOL, &protocol, sizeof(protocol)); | |
d62a17ae | 1499 | addattr_l(&req.n, sizeof(req), NDA_DST, &addr, 4); |
1500 | addattr_l(&req.n, sizeof(req), NDA_LLADDR, lla, llalen); | |
5c610faf | 1501 | |
d62a17ae | 1502 | return netlink_talk(netlink_talk_filter, &req.n, &zns->netlink_cmd, zns, |
1503 | 0); | |
5c610faf DS |
1504 | } |
1505 | ||
7cdb1a84 MS |
1506 | /* |
1507 | * Routing table change via netlink interface, using a dataplane context object | |
1508 | */ | |
25779064 | 1509 | static int netlink_route_multipath(int cmd, struct zebra_dplane_ctx *ctx) |
7cdb1a84 MS |
1510 | { |
1511 | int bytelen; | |
7cdb1a84 MS |
1512 | struct nexthop *nexthop = NULL; |
1513 | unsigned int nexthop_num; | |
1514 | int family; | |
1515 | const char *routedesc; | |
1516 | int setsrc = 0; | |
1517 | union g_addr src; | |
1518 | const struct prefix *p, *src_p; | |
1519 | uint32_t table_id; | |
1520 | ||
1521 | struct { | |
1522 | struct nlmsghdr n; | |
1523 | struct rtmsg r; | |
1524 | char buf[NL_PKT_BUF_SIZE]; | |
1525 | } req; | |
1526 | ||
1527 | p = dplane_ctx_get_dest(ctx); | |
1528 | src_p = dplane_ctx_get_src(ctx); | |
1529 | ||
1530 | family = PREFIX_FAMILY(p); | |
1531 | ||
5709131c | 1532 | memset(&req, 0, sizeof(req) - NL_PKT_BUF_SIZE); |
7cdb1a84 MS |
1533 | |
1534 | bytelen = (family == AF_INET ? 4 : 16); | |
1535 | ||
1536 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg)); | |
1537 | req.n.nlmsg_flags = NLM_F_CREATE | NLM_F_REQUEST; | |
1538 | ||
334734a8 DS |
1539 | if ((cmd == RTM_NEWROUTE) && |
1540 | ((p->family == AF_INET) || v6_rr_semantics)) | |
1541 | req.n.nlmsg_flags |= NLM_F_REPLACE; | |
7cdb1a84 MS |
1542 | |
1543 | req.n.nlmsg_type = cmd; | |
1544 | ||
1545 | req.n.nlmsg_pid = dplane_ctx_get_ns(ctx)->nls.snl.nl_pid; | |
1546 | ||
1547 | req.r.rtm_family = family; | |
1548 | req.r.rtm_dst_len = p->prefixlen; | |
1549 | req.r.rtm_src_len = src_p ? src_p->prefixlen : 0; | |
1550 | req.r.rtm_scope = RT_SCOPE_UNIVERSE; | |
1551 | ||
5709131c | 1552 | if (cmd == RTM_DELROUTE) |
7cdb1a84 | 1553 | req.r.rtm_protocol = zebra2proto(dplane_ctx_get_old_type(ctx)); |
5709131c | 1554 | else |
7cdb1a84 | 1555 | req.r.rtm_protocol = zebra2proto(dplane_ctx_get_type(ctx)); |
7cdb1a84 MS |
1556 | |
1557 | /* | |
1558 | * blackhole routes are not RTN_UNICAST, they are | |
1559 | * RTN_ BLACKHOLE|UNREACHABLE|PROHIBIT | |
1560 | * so setting this value as a RTN_UNICAST would | |
1561 | * cause the route lookup of just the prefix | |
1562 | * to fail. So no need to specify this for | |
1563 | * the RTM_DELROUTE case | |
1564 | */ | |
1565 | if (cmd != RTM_DELROUTE) | |
1566 | req.r.rtm_type = RTN_UNICAST; | |
1567 | ||
5709131c | 1568 | addattr_l(&req.n, sizeof(req), RTA_DST, &p->u.prefix, bytelen); |
7cdb1a84 | 1569 | if (src_p) |
5709131c | 1570 | addattr_l(&req.n, sizeof(req), RTA_SRC, &src_p->u.prefix, |
7cdb1a84 MS |
1571 | bytelen); |
1572 | ||
1573 | /* Metric. */ | |
1574 | /* Hardcode the metric for all routes coming from zebra. Metric isn't | |
1575 | * used | |
1576 | * either by the kernel or by zebra. Its purely for calculating best | |
1577 | * path(s) | |
1578 | * by the routing protocol and for communicating with protocol peers. | |
1579 | */ | |
5709131c | 1580 | addattr32(&req.n, sizeof(req), RTA_PRIORITY, NL_DEFAULT_ROUTE_METRIC); |
7cdb1a84 MS |
1581 | |
1582 | #if defined(SUPPORT_REALMS) | |
1583 | { | |
1584 | route_tag_t tag; | |
1585 | ||
5709131c | 1586 | if (cmd == RTM_DELROUTE) |
7cdb1a84 | 1587 | tag = dplane_ctx_get_old_tag(ctx); |
5709131c | 1588 | else |
7cdb1a84 | 1589 | tag = dplane_ctx_get_tag(ctx); |
7cdb1a84 MS |
1590 | |
1591 | if (tag > 0 && tag <= 255) | |
5709131c | 1592 | addattr32(&req.n, sizeof(req), RTA_FLOW, tag); |
7cdb1a84 MS |
1593 | } |
1594 | #endif | |
1595 | /* Table corresponding to this route. */ | |
1596 | table_id = dplane_ctx_get_table(ctx); | |
1597 | if (table_id < 256) | |
1598 | req.r.rtm_table = table_id; | |
1599 | else { | |
1600 | req.r.rtm_table = RT_TABLE_UNSPEC; | |
5709131c | 1601 | addattr32(&req.n, sizeof(req), RTA_TABLE, table_id); |
7cdb1a84 MS |
1602 | } |
1603 | ||
1604 | _netlink_route_debug(cmd, p, family, dplane_ctx_get_vrf(ctx), table_id); | |
1605 | ||
1606 | /* | |
1607 | * If we are not updating the route and we have received | |
1608 | * a route delete, then all we need to fill in is the | |
1609 | * prefix information to tell the kernel to schwack | |
1610 | * it. | |
1611 | */ | |
1612 | if (cmd == RTM_DELROUTE) | |
1613 | goto skip; | |
1614 | ||
1615 | if (dplane_ctx_get_mtu(ctx) || dplane_ctx_get_nh_mtu(ctx)) { | |
1616 | char buf[NL_PKT_BUF_SIZE]; | |
1617 | struct rtattr *rta = (void *)buf; | |
1618 | uint32_t mtu = dplane_ctx_get_mtu(ctx); | |
1619 | uint32_t nexthop_mtu = dplane_ctx_get_nh_mtu(ctx); | |
5709131c | 1620 | |
7cdb1a84 MS |
1621 | if (!mtu || (nexthop_mtu && nexthop_mtu < mtu)) |
1622 | mtu = nexthop_mtu; | |
1623 | rta->rta_type = RTA_METRICS; | |
1624 | rta->rta_len = RTA_LENGTH(0); | |
5709131c MS |
1625 | rta_addattr_l(rta, NL_PKT_BUF_SIZE, |
1626 | RTAX_MTU, &mtu, sizeof(mtu)); | |
7cdb1a84 MS |
1627 | addattr_l(&req.n, NL_PKT_BUF_SIZE, RTA_METRICS, RTA_DATA(rta), |
1628 | RTA_PAYLOAD(rta)); | |
1629 | } | |
1630 | ||
81505946 | 1631 | if (supports_nh) { |
de3f5488 SW |
1632 | /* Kernel supports nexthop objects */ |
1633 | addattr32(&req.n, sizeof(req), RTA_NH_ID, | |
1634 | dplane_ctx_get_nhe_id(ctx)); | |
1635 | goto skip; | |
1636 | } | |
1637 | ||
7cdb1a84 | 1638 | /* Count overall nexthops so we can decide whether to use singlepath |
5709131c MS |
1639 | * or multipath case. |
1640 | */ | |
7cdb1a84 MS |
1641 | nexthop_num = 0; |
1642 | for (ALL_NEXTHOPS_PTR(dplane_ctx_get_ng(ctx), nexthop)) { | |
1643 | if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_RECURSIVE)) | |
1644 | continue; | |
1645 | if (cmd == RTM_NEWROUTE && !NEXTHOP_IS_ACTIVE(nexthop->flags)) | |
1646 | continue; | |
1647 | ||
1648 | nexthop_num++; | |
1649 | } | |
1650 | ||
1651 | /* Singlepath case. */ | |
220f0f42 | 1652 | if (nexthop_num == 1) { |
7cdb1a84 MS |
1653 | nexthop_num = 0; |
1654 | for (ALL_NEXTHOPS_PTR(dplane_ctx_get_ng(ctx), nexthop)) { | |
1655 | /* | |
1656 | * So we want to cover 2 types of blackhole | |
1657 | * routes here: | |
1658 | * 1) A normal blackhole route( ala from a static | |
1659 | * install. | |
1660 | * 2) A recursively resolved blackhole route | |
1661 | */ | |
1662 | if (nexthop->type == NEXTHOP_TYPE_BLACKHOLE) { | |
1663 | switch (nexthop->bh_type) { | |
1664 | case BLACKHOLE_ADMINPROHIB: | |
1665 | req.r.rtm_type = RTN_PROHIBIT; | |
1666 | break; | |
1667 | case BLACKHOLE_REJECT: | |
1668 | req.r.rtm_type = RTN_UNREACHABLE; | |
1669 | break; | |
1670 | default: | |
1671 | req.r.rtm_type = RTN_BLACKHOLE; | |
1672 | break; | |
1673 | } | |
1674 | goto skip; | |
1675 | } | |
1676 | if (CHECK_FLAG(nexthop->flags, | |
1677 | NEXTHOP_FLAG_RECURSIVE)) { | |
5709131c MS |
1678 | |
1679 | if (setsrc) | |
1680 | continue; | |
1681 | ||
1682 | if (family == AF_INET) { | |
1683 | if (nexthop->rmap_src.ipv4.s_addr | |
1684 | != 0) { | |
1685 | src.ipv4 = | |
1686 | nexthop->rmap_src.ipv4; | |
1687 | setsrc = 1; | |
1688 | } else if (nexthop->src.ipv4.s_addr | |
1689 | != 0) { | |
1690 | src.ipv4 = | |
1691 | nexthop->src.ipv4; | |
1692 | setsrc = 1; | |
1693 | } | |
1694 | } else if (family == AF_INET6) { | |
1695 | if (!IN6_IS_ADDR_UNSPECIFIED( | |
1696 | &nexthop->rmap_src.ipv6)) { | |
1697 | src.ipv6 = | |
1698 | nexthop->rmap_src.ipv6; | |
1699 | setsrc = 1; | |
1700 | } else if ( | |
1701 | !IN6_IS_ADDR_UNSPECIFIED( | |
1702 | &nexthop->src.ipv6)) { | |
1703 | src.ipv6 = | |
1704 | nexthop->src.ipv6; | |
1705 | setsrc = 1; | |
7cdb1a84 MS |
1706 | } |
1707 | } | |
f183e380 | 1708 | continue; |
7cdb1a84 MS |
1709 | } |
1710 | ||
1711 | if ((cmd == RTM_NEWROUTE | |
1712 | && NEXTHOP_IS_ACTIVE(nexthop->flags))) { | |
1713 | routedesc = nexthop->rparent | |
1714 | ? "recursive, single-path" | |
1715 | : "single-path"; | |
1716 | ||
1717 | _netlink_route_build_singlepath( | |
1718 | routedesc, bytelen, nexthop, &req.n, | |
5709131c | 1719 | &req.r, sizeof(req), cmd); |
7cdb1a84 MS |
1720 | nexthop_num++; |
1721 | break; | |
1722 | } | |
1723 | } | |
1724 | if (setsrc && (cmd == RTM_NEWROUTE)) { | |
1725 | if (family == AF_INET) | |
5709131c | 1726 | addattr_l(&req.n, sizeof(req), RTA_PREFSRC, |
7cdb1a84 MS |
1727 | &src.ipv4, bytelen); |
1728 | else if (family == AF_INET6) | |
5709131c | 1729 | addattr_l(&req.n, sizeof(req), RTA_PREFSRC, |
7cdb1a84 MS |
1730 | &src.ipv6, bytelen); |
1731 | } | |
1732 | } else { /* Multipath case */ | |
1733 | char buf[NL_PKT_BUF_SIZE]; | |
1734 | struct rtattr *rta = (void *)buf; | |
1735 | struct rtnexthop *rtnh; | |
81793ac1 | 1736 | const union g_addr *src1 = NULL; |
7cdb1a84 MS |
1737 | |
1738 | rta->rta_type = RTA_MULTIPATH; | |
1739 | rta->rta_len = RTA_LENGTH(0); | |
1740 | rtnh = RTA_DATA(rta); | |
1741 | ||
1742 | nexthop_num = 0; | |
1743 | for (ALL_NEXTHOPS_PTR(dplane_ctx_get_ng(ctx), nexthop)) { | |
7cdb1a84 MS |
1744 | if (CHECK_FLAG(nexthop->flags, |
1745 | NEXTHOP_FLAG_RECURSIVE)) { | |
1746 | /* This only works for IPv4 now */ | |
5709131c MS |
1747 | if (setsrc) |
1748 | continue; | |
1749 | ||
1750 | if (family == AF_INET) { | |
1751 | if (nexthop->rmap_src.ipv4.s_addr | |
1752 | != 0) { | |
1753 | src.ipv4 = | |
1754 | nexthop->rmap_src.ipv4; | |
1755 | setsrc = 1; | |
1756 | } else if (nexthop->src.ipv4.s_addr | |
1757 | != 0) { | |
1758 | src.ipv4 = | |
1759 | nexthop->src.ipv4; | |
1760 | setsrc = 1; | |
1761 | } | |
1762 | } else if (family == AF_INET6) { | |
1763 | if (!IN6_IS_ADDR_UNSPECIFIED( | |
1764 | &nexthop->rmap_src.ipv6)) { | |
1765 | src.ipv6 = | |
1766 | nexthop->rmap_src.ipv6; | |
1767 | setsrc = 1; | |
1768 | } else if ( | |
1769 | !IN6_IS_ADDR_UNSPECIFIED( | |
1770 | &nexthop->src.ipv6)) { | |
1771 | src.ipv6 = | |
1772 | nexthop->src.ipv6; | |
1773 | setsrc = 1; | |
7cdb1a84 MS |
1774 | } |
1775 | } | |
78e54ded MS |
1776 | |
1777 | continue; | |
7cdb1a84 MS |
1778 | } |
1779 | ||
1780 | if ((cmd == RTM_NEWROUTE | |
1781 | && NEXTHOP_IS_ACTIVE(nexthop->flags))) { | |
1782 | routedesc = nexthop->rparent | |
1783 | ? "recursive, multipath" | |
1784 | : "multipath"; | |
1785 | nexthop_num++; | |
1786 | ||
1787 | _netlink_route_build_multipath( | |
1788 | routedesc, bytelen, nexthop, rta, rtnh, | |
1789 | &req.r, &src1); | |
1790 | rtnh = RTNH_NEXT(rtnh); | |
1791 | ||
1792 | if (!setsrc && src1) { | |
1793 | if (family == AF_INET) | |
1794 | src.ipv4 = src1->ipv4; | |
1795 | else if (family == AF_INET6) | |
1796 | src.ipv6 = src1->ipv6; | |
1797 | ||
1798 | setsrc = 1; | |
1799 | } | |
1800 | } | |
1801 | } | |
1802 | if (setsrc && (cmd == RTM_NEWROUTE)) { | |
1803 | if (family == AF_INET) | |
5709131c | 1804 | addattr_l(&req.n, sizeof(req), RTA_PREFSRC, |
7cdb1a84 MS |
1805 | &src.ipv4, bytelen); |
1806 | else if (family == AF_INET6) | |
5709131c | 1807 | addattr_l(&req.n, sizeof(req), RTA_PREFSRC, |
7cdb1a84 MS |
1808 | &src.ipv6, bytelen); |
1809 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1810 | zlog_debug("Setting source"); | |
1811 | } | |
1812 | ||
1813 | if (rta->rta_len > RTA_LENGTH(0)) | |
1814 | addattr_l(&req.n, NL_PKT_BUF_SIZE, RTA_MULTIPATH, | |
1815 | RTA_DATA(rta), RTA_PAYLOAD(rta)); | |
1816 | } | |
1817 | ||
1818 | /* If there is no useful nexthop then return. */ | |
1819 | if (nexthop_num == 0) { | |
1820 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1821 | zlog_debug( | |
1822 | "netlink_route_multipath(): No useful nexthop."); | |
1823 | return 0; | |
1824 | } | |
1825 | ||
1826 | skip: | |
7cdb1a84 MS |
1827 | /* Talk to netlink socket. */ |
1828 | return netlink_talk_info(netlink_talk_filter, &req.n, | |
1829 | dplane_ctx_get_ns(ctx), 0); | |
1830 | } | |
1831 | ||
43b5cc5e | 1832 | int kernel_get_ipmr_sg_stats(struct zebra_vrf *zvrf, void *in) |
e3be0432 | 1833 | { |
5523c156 | 1834 | uint32_t actual_table; |
d62a17ae | 1835 | int suc = 0; |
1836 | struct mcast_route_data *mr = (struct mcast_route_data *)in; | |
bd8b9272 DS |
1837 | struct { |
1838 | struct nlmsghdr n; | |
1839 | struct ndmsg ndm; | |
1840 | char buf[256]; | |
1841 | } req; | |
e3be0432 | 1842 | |
d62a17ae | 1843 | mroute = mr; |
5895d33f | 1844 | struct zebra_ns *zns; |
bd8b9272 | 1845 | |
009f8ad5 | 1846 | zns = zvrf->zns; |
5605ecfc | 1847 | memset(&req, 0, sizeof(req)); |
bd8b9272 DS |
1848 | |
1849 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
1850 | req.n.nlmsg_flags = NLM_F_REQUEST; | |
1851 | req.n.nlmsg_pid = zns->netlink_cmd.snl.nl_pid; | |
1852 | ||
1853 | req.ndm.ndm_family = RTNL_FAMILY_IPMR; | |
1854 | req.n.nlmsg_type = RTM_GETROUTE; | |
1855 | ||
1856 | addattr_l(&req.n, sizeof(req), RTA_IIF, &mroute->ifindex, 4); | |
1857 | addattr_l(&req.n, sizeof(req), RTA_OIF, &mroute->ifindex, 4); | |
1858 | addattr_l(&req.n, sizeof(req), RTA_SRC, &mroute->sg.src.s_addr, 4); | |
1859 | addattr_l(&req.n, sizeof(req), RTA_DST, &mroute->sg.grp.s_addr, 4); | |
5523c156 DS |
1860 | /* |
1861 | * What? | |
1862 | * | |
1863 | * So during the namespace cleanup we started storing | |
1864 | * the zvrf table_id for the default table as RT_TABLE_MAIN | |
1865 | * which is what the normal routing table for ip routing is. | |
1866 | * This change caused this to break our lookups of sg data | |
1867 | * because prior to this change the zvrf->table_id was 0 | |
1868 | * and when the pim multicast kernel code saw a 0, | |
1869 | * it was auto-translated to RT_TABLE_DEFAULT. But since | |
1870 | * we are now passing in RT_TABLE_MAIN there is no auto-translation | |
1871 | * and the kernel goes screw you and the delicious cookies you | |
1872 | * are trying to give me. So now we have this little hack. | |
1873 | */ | |
1874 | actual_table = (zvrf->table_id == RT_TABLE_MAIN) ? RT_TABLE_DEFAULT : | |
1875 | zvrf->table_id; | |
1876 | addattr_l(&req.n, sizeof(req), RTA_TABLE, &actual_table, 4); | |
e3be0432 | 1877 | |
bd8b9272 DS |
1878 | suc = netlink_talk(netlink_route_change_read_multicast, &req.n, |
1879 | &zns->netlink_cmd, zns, 0); | |
e3be0432 | 1880 | |
bd8b9272 | 1881 | mroute = NULL; |
d62a17ae | 1882 | return suc; |
e3be0432 DS |
1883 | } |
1884 | ||
8d03bc50 SW |
1885 | /* Char length to debug ID with */ |
1886 | #define ID_LENGTH 10 | |
1887 | ||
565ce0d3 | 1888 | static void _netlink_nexthop_build_group(struct nlmsghdr *n, size_t req_size, |
8d03bc50 | 1889 | uint32_t id, |
e22e8001 | 1890 | const struct nh_grp *z_grp, |
0c8215cb | 1891 | const uint8_t count) |
565ce0d3 | 1892 | { |
565ce0d3 | 1893 | struct nexthop_grp grp[count]; |
8d03bc50 SW |
1894 | /* Need space for max group size, "/", and null term */ |
1895 | char buf[(MULTIPATH_NUM * (ID_LENGTH + 1)) + 1]; | |
1896 | char buf1[ID_LENGTH + 2]; | |
1897 | ||
1898 | buf[0] = '\0'; | |
565ce0d3 SW |
1899 | |
1900 | memset(grp, 0, sizeof(grp)); | |
1901 | ||
1902 | if (count) { | |
0c8215cb | 1903 | for (int i = 0; i < count; i++) { |
e22e8001 | 1904 | grp[i].id = z_grp[i].id; |
df7fb580 | 1905 | grp[i].weight = z_grp[i].weight - 1; |
8d03bc50 SW |
1906 | |
1907 | if (IS_ZEBRA_DEBUG_KERNEL) { | |
1908 | if (i == 0) | |
1909 | snprintf(buf, sizeof(buf1), "group %u", | |
1910 | grp[i].id); | |
1911 | else { | |
1912 | snprintf(buf1, sizeof(buf1), "/%u", | |
1913 | grp[i].id); | |
1914 | strlcat(buf, buf1, sizeof(buf)); | |
1915 | } | |
1916 | } | |
565ce0d3 | 1917 | } |
0c8215cb | 1918 | addattr_l(n, req_size, NHA_GROUP, grp, count * sizeof(*grp)); |
565ce0d3 | 1919 | } |
8d03bc50 SW |
1920 | |
1921 | if (IS_ZEBRA_DEBUG_KERNEL) | |
1922 | zlog_debug("%s: ID (%u): %s", __func__, id, buf); | |
565ce0d3 SW |
1923 | } |
1924 | ||
f820d025 SW |
1925 | /** |
1926 | * netlink_nexthop() - Nexthop change via the netlink interface | |
1927 | * | |
1928 | * @ctx: Dataplane ctx | |
1929 | * | |
1930 | * Return: Result status | |
1931 | */ | |
1932 | static int netlink_nexthop(int cmd, struct zebra_dplane_ctx *ctx) | |
1933 | { | |
f820d025 SW |
1934 | struct { |
1935 | struct nlmsghdr n; | |
1936 | struct nhmsg nhm; | |
1937 | char buf[NL_PKT_BUF_SIZE]; | |
1938 | } req; | |
1939 | ||
8d03bc50 SW |
1940 | mpls_lse_t out_lse[MPLS_MAX_LABELS]; |
1941 | char label_buf[256]; | |
1942 | int num_labels = 0; | |
1943 | size_t req_size = sizeof(req); | |
1944 | ||
81505946 SW |
1945 | /* Nothing to do if the kernel doesn't support nexthop objects */ |
1946 | if (!supports_nh) | |
1947 | return 0; | |
1948 | ||
8d03bc50 SW |
1949 | label_buf[0] = '\0'; |
1950 | ||
1951 | memset(&req, 0, req_size); | |
f820d025 | 1952 | |
f820d025 SW |
1953 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct nhmsg)); |
1954 | req.n.nlmsg_flags = NLM_F_CREATE | NLM_F_REQUEST; | |
9a1588c4 SW |
1955 | |
1956 | if (cmd == RTM_NEWNEXTHOP) | |
1957 | req.n.nlmsg_flags |= NLM_F_REPLACE; | |
1958 | ||
f820d025 | 1959 | req.n.nlmsg_type = cmd; |
4f096395 | 1960 | req.n.nlmsg_pid = dplane_ctx_get_ns(ctx)->nls.snl.nl_pid; |
f820d025 SW |
1961 | |
1962 | req.nhm.nh_family = AF_UNSPEC; | |
fec211ad | 1963 | /* TODO: Scope? */ |
f820d025 | 1964 | |
0c8215cb SW |
1965 | uint32_t id = dplane_ctx_get_nhe_id(ctx); |
1966 | ||
1967 | if (!id) { | |
f820d025 SW |
1968 | flog_err( |
1969 | EC_ZEBRA_NHG_FIB_UPDATE, | |
1970 | "Failed trying to update a nexthop group in the kernel that does not have an ID"); | |
1971 | return -1; | |
1972 | } | |
1973 | ||
8d03bc50 | 1974 | addattr32(&req.n, req_size, NHA_ID, id); |
f820d025 SW |
1975 | |
1976 | if (cmd == RTM_NEWNEXTHOP) { | |
e22e8001 | 1977 | if (dplane_ctx_get_nhe_nh_grp_count(ctx)) |
0c8215cb | 1978 | _netlink_nexthop_build_group( |
8d03bc50 | 1979 | &req.n, req_size, id, |
e22e8001 SW |
1980 | dplane_ctx_get_nhe_nh_grp(ctx), |
1981 | dplane_ctx_get_nhe_nh_grp_count(ctx)); | |
0c8215cb SW |
1982 | else { |
1983 | const struct nexthop *nh = | |
1984 | dplane_ctx_get_nhe_ng(ctx)->nexthop; | |
1985 | afi_t afi = dplane_ctx_get_nhe_afi(ctx); | |
e8b0e420 | 1986 | |
0c8215cb | 1987 | if (afi == AFI_IP) |
565ce0d3 | 1988 | req.nhm.nh_family = AF_INET; |
0c8215cb | 1989 | else if (afi == AFI_IP6) |
565ce0d3 | 1990 | req.nhm.nh_family = AF_INET6; |
f820d025 | 1991 | |
565ce0d3 | 1992 | switch (nh->type) { |
a6e6a6d8 | 1993 | case NEXTHOP_TYPE_IPV4: |
565ce0d3 | 1994 | case NEXTHOP_TYPE_IPV4_IFINDEX: |
8d03bc50 | 1995 | addattr_l(&req.n, req_size, NHA_GATEWAY, |
565ce0d3 SW |
1996 | &nh->gate.ipv4, IPV4_MAX_BYTELEN); |
1997 | break; | |
a6e6a6d8 | 1998 | case NEXTHOP_TYPE_IPV6: |
565ce0d3 | 1999 | case NEXTHOP_TYPE_IPV6_IFINDEX: |
8d03bc50 | 2000 | addattr_l(&req.n, req_size, NHA_GATEWAY, |
565ce0d3 SW |
2001 | &nh->gate.ipv6, IPV6_MAX_BYTELEN); |
2002 | break; | |
2003 | case NEXTHOP_TYPE_BLACKHOLE: | |
8d03bc50 SW |
2004 | addattr_l(&req.n, req_size, NHA_BLACKHOLE, NULL, |
2005 | 0); | |
2006 | /* Blackhole shouldn't have anymore attributes | |
2007 | */ | |
2008 | goto nexthop_done; | |
565ce0d3 SW |
2009 | case NEXTHOP_TYPE_IFINDEX: |
2010 | /* Don't need anymore info for this */ | |
2011 | break; | |
a6e6a6d8 SW |
2012 | } |
2013 | ||
2014 | if (!nh->ifindex) { | |
565ce0d3 SW |
2015 | flog_err( |
2016 | EC_ZEBRA_NHG_FIB_UPDATE, | |
2017 | "Context received for kernel nexthop update without an interface"); | |
2018 | return -1; | |
565ce0d3 SW |
2019 | } |
2020 | ||
8d03bc50 SW |
2021 | addattr32(&req.n, req_size, NHA_OIF, nh->ifindex); |
2022 | ||
62d2ecb2 SW |
2023 | if (CHECK_FLAG(nh->flags, NEXTHOP_FLAG_ONLINK)) |
2024 | req.nhm.nh_flags |= RTNH_F_ONLINK; | |
2025 | ||
8d03bc50 SW |
2026 | num_labels = |
2027 | build_label_stack(nh->nh_label, out_lse, | |
2028 | label_buf, sizeof(label_buf)); | |
2029 | ||
2030 | if (num_labels) { | |
2031 | /* Set the BoS bit */ | |
2032 | out_lse[num_labels - 1] |= | |
2033 | htonl(1 << MPLS_LS_S_SHIFT); | |
2034 | ||
2035 | /* | |
2036 | * TODO: MPLS unsupported for now in kernel. | |
2037 | */ | |
2038 | if (req.nhm.nh_family == AF_MPLS) | |
2039 | goto nexthop_done; | |
2040 | #if 0 | |
2041 | addattr_l(&req.n, req_size, NHA_NEWDST, | |
2042 | &out_lse, | |
2043 | num_labels | |
2044 | * sizeof(mpls_lse_t)); | |
2045 | #endif | |
2046 | else { | |
2047 | struct rtattr *nest; | |
2048 | uint16_t encap = LWTUNNEL_ENCAP_MPLS; | |
2049 | ||
2050 | addattr_l(&req.n, req_size, | |
2051 | NHA_ENCAP_TYPE, &encap, | |
2052 | sizeof(uint16_t)); | |
2053 | nest = addattr_nest(&req.n, req_size, | |
2054 | NHA_ENCAP); | |
2055 | addattr_l(&req.n, req_size, | |
2056 | MPLS_IPTUNNEL_DST, &out_lse, | |
2057 | num_labels | |
2058 | * sizeof(mpls_lse_t)); | |
2059 | addattr_nest_end(&req.n, nest); | |
2060 | } | |
2061 | } | |
2062 | ||
2063 | nexthop_done: | |
2064 | if (IS_ZEBRA_DEBUG_KERNEL) { | |
2065 | char buf[NEXTHOP_STRLEN]; | |
2066 | ||
2067 | snprintfrr(buf, sizeof(buf), "%pNHv", nh); | |
2068 | zlog_debug("%s: ID (%u): %s (%u) %s ", __func__, | |
2069 | id, buf, nh->vrf_id, label_buf); | |
2070 | } | |
f820d025 SW |
2071 | } |
2072 | ||
38e40db1 | 2073 | req.nhm.nh_protocol = zebra2proto(dplane_ctx_get_nhe_type(ctx)); |
f820d025 | 2074 | |
f820d025 SW |
2075 | } else if (cmd != RTM_DELNEXTHOP) { |
2076 | flog_err( | |
2077 | EC_ZEBRA_NHG_FIB_UPDATE, | |
2078 | "Nexthop group kernel update command (%d) does not exist", | |
2079 | cmd); | |
2080 | return -1; | |
2081 | } | |
2082 | ||
0c8215cb | 2083 | _netlink_nexthop_debug(cmd, id); |
f820d025 SW |
2084 | |
2085 | return netlink_talk_info(netlink_talk_filter, &req.n, | |
2086 | dplane_ctx_get_ns(ctx), 0); | |
2087 | } | |
2088 | ||
2089 | /** | |
2090 | * kernel_nexthop_update() - Update/delete a nexthop from the kernel | |
2091 | * | |
2092 | * @ctx: Dataplane context | |
2093 | * | |
2094 | * Return: Dataplane result flag | |
2095 | */ | |
2096 | enum zebra_dplane_result kernel_nexthop_update(struct zebra_dplane_ctx *ctx) | |
2097 | { | |
98cda54a SW |
2098 | int cmd = 0; |
2099 | int ret = 0; | |
f820d025 SW |
2100 | |
2101 | switch (dplane_ctx_get_op(ctx)) { | |
2102 | case DPLANE_OP_NH_DELETE: | |
2103 | cmd = RTM_DELNEXTHOP; | |
2104 | break; | |
2105 | case DPLANE_OP_NH_INSTALL: | |
2106 | case DPLANE_OP_NH_UPDATE: | |
2107 | cmd = RTM_NEWNEXTHOP; | |
2108 | break; | |
2109 | case DPLANE_OP_ROUTE_INSTALL: | |
2110 | case DPLANE_OP_ROUTE_UPDATE: | |
2111 | case DPLANE_OP_ROUTE_DELETE: | |
2112 | case DPLANE_OP_ROUTE_NOTIFY: | |
2113 | case DPLANE_OP_LSP_INSTALL: | |
2114 | case DPLANE_OP_LSP_UPDATE: | |
2115 | case DPLANE_OP_LSP_DELETE: | |
2116 | case DPLANE_OP_LSP_NOTIFY: | |
2117 | case DPLANE_OP_PW_INSTALL: | |
2118 | case DPLANE_OP_PW_UNINSTALL: | |
2119 | case DPLANE_OP_SYS_ROUTE_ADD: | |
2120 | case DPLANE_OP_SYS_ROUTE_DELETE: | |
2121 | case DPLANE_OP_ADDR_INSTALL: | |
2122 | case DPLANE_OP_ADDR_UNINSTALL: | |
2123 | case DPLANE_OP_MAC_INSTALL: | |
2124 | case DPLANE_OP_MAC_DELETE: | |
40a2a6cd SW |
2125 | case DPLANE_OP_NEIGH_INSTALL: |
2126 | case DPLANE_OP_NEIGH_UPDATE: | |
2127 | case DPLANE_OP_NEIGH_DELETE: | |
2128 | case DPLANE_OP_VTEP_ADD: | |
2129 | case DPLANE_OP_VTEP_DELETE: | |
f820d025 SW |
2130 | case DPLANE_OP_NONE: |
2131 | flog_err( | |
2132 | EC_ZEBRA_NHG_FIB_UPDATE, | |
2133 | "Context received for kernel nexthop update with incorrect OP code (%u)", | |
2134 | dplane_ctx_get_op(ctx)); | |
2135 | return ZEBRA_DPLANE_REQUEST_FAILURE; | |
f820d025 SW |
2136 | } |
2137 | ||
2138 | ret = netlink_nexthop(cmd, ctx); | |
2139 | ||
2140 | return (ret == 0 ? ZEBRA_DPLANE_REQUEST_SUCCESS | |
2141 | : ZEBRA_DPLANE_REQUEST_FAILURE); | |
2142 | } | |
2143 | ||
7cdb1a84 MS |
2144 | /* |
2145 | * Update or delete a prefix from the kernel, | |
2146 | * using info from a dataplane context. | |
2147 | */ | |
25779064 | 2148 | enum zebra_dplane_result kernel_route_update(struct zebra_dplane_ctx *ctx) |
7cdb1a84 MS |
2149 | { |
2150 | int cmd, ret; | |
2151 | const struct prefix *p = dplane_ctx_get_dest(ctx); | |
f183e380 | 2152 | struct nexthop *nexthop; |
7cdb1a84 MS |
2153 | |
2154 | if (dplane_ctx_get_op(ctx) == DPLANE_OP_ROUTE_DELETE) { | |
2155 | cmd = RTM_DELROUTE; | |
2156 | } else if (dplane_ctx_get_op(ctx) == DPLANE_OP_ROUTE_INSTALL) { | |
2157 | cmd = RTM_NEWROUTE; | |
2158 | } else if (dplane_ctx_get_op(ctx) == DPLANE_OP_ROUTE_UPDATE) { | |
2159 | ||
2160 | if (p->family == AF_INET || v6_rr_semantics) { | |
2161 | /* Single 'replace' operation */ | |
2162 | cmd = RTM_NEWROUTE; | |
fe5f21af DS |
2163 | |
2164 | /* | |
2165 | * With route replace semantics in place | |
2166 | * for v4 routes and the new route is a system | |
2167 | * route we do not install anything. | |
2168 | * The problem here is that the new system | |
2169 | * route should cause us to withdraw from | |
2170 | * the kernel the old non-system route | |
2171 | */ | |
2172 | if (RSYSTEM_ROUTE(dplane_ctx_get_type(ctx)) && | |
2173 | !RSYSTEM_ROUTE(dplane_ctx_get_old_type(ctx))) | |
2174 | (void)netlink_route_multipath(RTM_DELROUTE, | |
2175 | ctx); | |
7cdb1a84 MS |
2176 | } else { |
2177 | /* | |
2178 | * So v6 route replace semantics are not in | |
2179 | * the kernel at this point as I understand it. | |
2180 | * so let's do a delete then an add. | |
2181 | * In the future once v6 route replace semantics | |
2182 | * are in we can figure out what to do here to | |
2183 | * allow working with old and new kernels. | |
2184 | * | |
2185 | * I'm also intentionally ignoring the failure case | |
2186 | * of the route delete. If that happens yeah we're | |
2187 | * screwed. | |
2188 | */ | |
3cdba47a DS |
2189 | if (!RSYSTEM_ROUTE(dplane_ctx_get_old_type(ctx))) |
2190 | (void)netlink_route_multipath(RTM_DELROUTE, | |
2191 | ctx); | |
7cdb1a84 MS |
2192 | cmd = RTM_NEWROUTE; |
2193 | } | |
2194 | ||
2195 | } else { | |
2196 | return ZEBRA_DPLANE_REQUEST_FAILURE; | |
2197 | } | |
2198 | ||
3cdba47a DS |
2199 | if (!RSYSTEM_ROUTE(dplane_ctx_get_type(ctx))) |
2200 | ret = netlink_route_multipath(cmd, ctx); | |
2201 | else | |
2202 | ret = 0; | |
f183e380 MS |
2203 | if ((cmd == RTM_NEWROUTE) && (ret == 0)) { |
2204 | /* Update installed nexthops to signal which have been | |
2205 | * installed. | |
2206 | */ | |
2207 | for (ALL_NEXTHOPS_PTR(dplane_ctx_get_ng(ctx), nexthop)) { | |
2208 | if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_RECURSIVE)) | |
2209 | continue; | |
2210 | ||
2211 | if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ACTIVE)) { | |
2212 | SET_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB); | |
f183e380 MS |
2213 | } |
2214 | } | |
2215 | } | |
7cdb1a84 MS |
2216 | |
2217 | return (ret == 0 ? | |
2218 | ZEBRA_DPLANE_REQUEST_SUCCESS : ZEBRA_DPLANE_REQUEST_FAILURE); | |
2219 | } | |
2220 | ||
d9f5b2f5 SW |
2221 | /** |
2222 | * netlink_nexthop_process_nh() - Parse the gatway/if info from a new nexthop | |
2223 | * | |
2224 | * @tb: Netlink RTA data | |
2225 | * @family: Address family in the nhmsg | |
8c0a24c1 | 2226 | * @ifp: Interface connected - this should be NULL, we fill it in |
d9f5b2f5 SW |
2227 | * @ns_id: Namspace id |
2228 | * | |
2229 | * Return: New nexthop | |
2230 | */ | |
e22e8001 SW |
2231 | static struct nexthop netlink_nexthop_process_nh(struct rtattr **tb, |
2232 | unsigned char family, | |
2233 | struct interface **ifp, | |
2234 | ns_id_t ns_id) | |
d9f5b2f5 | 2235 | { |
e22e8001 | 2236 | struct nexthop nh = {}; |
d9f5b2f5 | 2237 | void *gate = NULL; |
8e401b25 | 2238 | enum nexthop_types_t type = 0; |
e22e8001 SW |
2239 | int if_index = 0; |
2240 | size_t sz = 0; | |
7134ba70 | 2241 | struct interface *ifp_lookup; |
d9f5b2f5 SW |
2242 | |
2243 | if_index = *(int *)RTA_DATA(tb[NHA_OIF]); | |
2244 | ||
8e401b25 | 2245 | |
d9f5b2f5 SW |
2246 | if (tb[NHA_GATEWAY]) { |
2247 | switch (family) { | |
2248 | case AF_INET: | |
8e401b25 | 2249 | type = NEXTHOP_TYPE_IPV4_IFINDEX; |
d9f5b2f5 SW |
2250 | sz = 4; |
2251 | break; | |
2252 | case AF_INET6: | |
8e401b25 | 2253 | type = NEXTHOP_TYPE_IPV6_IFINDEX; |
d9f5b2f5 SW |
2254 | sz = 16; |
2255 | break; | |
2256 | default: | |
2257 | flog_warn( | |
2258 | EC_ZEBRA_BAD_NHG_MESSAGE, | |
c4239c05 | 2259 | "Nexthop gateway with bad address family (%d) received from kernel", |
d9f5b2f5 | 2260 | family); |
e22e8001 | 2261 | return nh; |
d9f5b2f5 SW |
2262 | } |
2263 | gate = RTA_DATA(tb[NHA_GATEWAY]); | |
e22e8001 | 2264 | } else |
8e401b25 | 2265 | type = NEXTHOP_TYPE_IFINDEX; |
d9f5b2f5 | 2266 | |
8e401b25 | 2267 | if (type) |
e22e8001 | 2268 | nh.type = type; |
8e401b25 SW |
2269 | |
2270 | if (gate) | |
e22e8001 | 2271 | memcpy(&(nh.gate), gate, sz); |
8e401b25 SW |
2272 | |
2273 | if (if_index) | |
e22e8001 | 2274 | nh.ifindex = if_index; |
8e401b25 | 2275 | |
7134ba70 DS |
2276 | ifp_lookup = |
2277 | if_lookup_by_index_per_ns(zebra_ns_lookup(ns_id), nh.ifindex); | |
2278 | ||
e22e8001 | 2279 | if (ifp) |
7134ba70 DS |
2280 | *ifp = ifp_lookup; |
2281 | if (ifp_lookup) | |
2282 | nh.vrf_id = ifp_lookup->vrf_id; | |
e22e8001 | 2283 | else { |
d9f5b2f5 SW |
2284 | flog_warn( |
2285 | EC_ZEBRA_UNKNOWN_INTERFACE, | |
2286 | "%s: Unknown nexthop interface %u received, defaulting to VRF_DEFAULT", | |
e22e8001 | 2287 | __PRETTY_FUNCTION__, nh.ifindex); |
d9f5b2f5 | 2288 | |
e22e8001 | 2289 | nh.vrf_id = VRF_DEFAULT; |
d9f5b2f5 SW |
2290 | } |
2291 | ||
2292 | if (tb[NHA_ENCAP] && tb[NHA_ENCAP_TYPE]) { | |
2293 | uint16_t encap_type = *(uint16_t *)RTA_DATA(tb[NHA_ENCAP_TYPE]); | |
2294 | int num_labels = 0; | |
6e728764 | 2295 | |
d9f5b2f5 SW |
2296 | mpls_label_t labels[MPLS_MAX_LABELS] = {0}; |
2297 | ||
e22e8001 | 2298 | if (encap_type == LWTUNNEL_ENCAP_MPLS) |
d9f5b2f5 | 2299 | num_labels = parse_encap_mpls(tb[NHA_ENCAP], labels); |
d9f5b2f5 | 2300 | |
e22e8001 SW |
2301 | if (num_labels) |
2302 | nexthop_add_labels(&nh, ZEBRA_LSP_STATIC, num_labels, | |
d9f5b2f5 | 2303 | labels); |
d9f5b2f5 SW |
2304 | } |
2305 | ||
2306 | return nh; | |
2307 | } | |
2308 | ||
85f5e761 | 2309 | static int netlink_nexthop_process_group(struct rtattr **tb, |
5a935f79 | 2310 | struct nh_grp *z_grp, int z_grp_size) |
d9f5b2f5 | 2311 | { |
e22e8001 SW |
2312 | uint8_t count = 0; |
2313 | /* linux/nexthop.h group struct */ | |
d9f5b2f5 SW |
2314 | struct nexthop_grp *n_grp = NULL; |
2315 | ||
85f5e761 | 2316 | n_grp = (struct nexthop_grp *)RTA_DATA(tb[NHA_GROUP]); |
d9f5b2f5 SW |
2317 | count = (RTA_PAYLOAD(tb[NHA_GROUP]) / sizeof(*n_grp)); |
2318 | ||
2319 | if (!count || (count * sizeof(*n_grp)) != RTA_PAYLOAD(tb[NHA_GROUP])) { | |
2320 | flog_warn(EC_ZEBRA_BAD_NHG_MESSAGE, | |
2321 | "Invalid nexthop group received from the kernel"); | |
85f5e761 | 2322 | return count; |
d9f5b2f5 SW |
2323 | } |
2324 | ||
38e40db1 | 2325 | #if 0 |
d9f5b2f5 | 2326 | // TODO: Need type for something? |
85f5e761 SW |
2327 | zlog_debug("Nexthop group type: %d", |
2328 | *((uint16_t *)RTA_DATA(tb[NHA_GROUP_TYPE]))); | |
d9f5b2f5 | 2329 | |
38e40db1 | 2330 | #endif |
d9f5b2f5 | 2331 | |
5a935f79 | 2332 | for (int i = 0; ((i < count) && (i < z_grp_size)); i++) { |
e22e8001 | 2333 | z_grp[i].id = n_grp[i].id; |
df7fb580 | 2334 | z_grp[i].weight = n_grp[i].weight + 1; |
85f5e761 | 2335 | } |
d9f5b2f5 SW |
2336 | return count; |
2337 | } | |
2338 | ||
2339 | /** | |
2340 | * netlink_nexthop_change() - Read in change about nexthops from the kernel | |
2341 | * | |
2342 | * @h: Netlink message header | |
2343 | * @ns_id: Namspace id | |
2344 | * @startup: Are we reading under startup conditions? | |
2345 | * | |
2346 | * Return: Result status | |
2347 | */ | |
2348 | int netlink_nexthop_change(struct nlmsghdr *h, ns_id_t ns_id, int startup) | |
2349 | { | |
2350 | int len; | |
2351 | /* nexthop group id */ | |
2352 | uint32_t id; | |
2353 | unsigned char family; | |
38e40db1 | 2354 | int type; |
e8b0e420 | 2355 | afi_t afi = AFI_UNSPEC; |
bbb3940e | 2356 | vrf_id_t vrf_id = 0; |
8c0a24c1 | 2357 | struct interface *ifp = NULL; |
d9f5b2f5 | 2358 | struct nhmsg *nhm = NULL; |
e22e8001 SW |
2359 | struct nexthop nh = {}; |
2360 | struct nh_grp grp[MULTIPATH_NUM] = {}; | |
85f5e761 | 2361 | /* Count of nexthops in group array */ |
e22e8001 | 2362 | uint8_t grp_count = 0; |
e22e8001 | 2363 | struct rtattr *tb[NHA_MAX + 1] = {}; |
d9f5b2f5 | 2364 | |
d9f5b2f5 SW |
2365 | nhm = NLMSG_DATA(h); |
2366 | ||
2367 | if (startup && h->nlmsg_type != RTM_NEWNEXTHOP) | |
2368 | return 0; | |
2369 | ||
2370 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct nhmsg)); | |
2371 | if (len < 0) { | |
2372 | zlog_warn( | |
2373 | "%s: Message received from netlink is of a broken size %d %zu", | |
2374 | __PRETTY_FUNCTION__, h->nlmsg_len, | |
2375 | (size_t)NLMSG_LENGTH(sizeof(struct nhmsg))); | |
2376 | return -1; | |
2377 | } | |
2378 | ||
d9f5b2f5 SW |
2379 | netlink_parse_rtattr(tb, NHA_MAX, RTM_NHA(nhm), len); |
2380 | ||
2381 | ||
2382 | if (!tb[NHA_ID]) { | |
2383 | flog_warn( | |
2384 | EC_ZEBRA_BAD_NHG_MESSAGE, | |
2385 | "Nexthop group without an ID received from the kernel"); | |
2386 | return -1; | |
2387 | } | |
2388 | ||
2389 | /* We use the ID key'd nhg table for kernel updates */ | |
2390 | id = *((uint32_t *)RTA_DATA(tb[NHA_ID])); | |
d9f5b2f5 | 2391 | |
e8b0e420 | 2392 | family = nhm->nh_family; |
e8b0e420 SW |
2393 | afi = family2afi(family); |
2394 | ||
38e40db1 SW |
2395 | type = proto2zebra(nhm->nh_protocol, 0, true); |
2396 | ||
fdee485a SW |
2397 | if (IS_ZEBRA_DEBUG_KERNEL) |
2398 | zlog_debug("%s ID (%u) %s NS %u", | |
2399 | nl_msg_type_to_str(h->nlmsg_type), id, | |
2400 | nl_family_to_str(family), ns_id); | |
2401 | ||
2402 | ||
d9f5b2f5 SW |
2403 | if (h->nlmsg_type == RTM_NEWNEXTHOP) { |
2404 | if (tb[NHA_GROUP]) { | |
2405 | /** | |
2406 | * If this is a group message its only going to have | |
2407 | * an array of nexthop IDs associated with it | |
2408 | */ | |
5a935f79 SW |
2409 | grp_count = netlink_nexthop_process_group( |
2410 | tb, grp, array_size(grp)); | |
85f5e761 SW |
2411 | } else { |
2412 | if (tb[NHA_BLACKHOLE]) { | |
2413 | /** | |
2414 | * This nexthop is just for blackhole-ing | |
2415 | * traffic, it should not have an OIF, GATEWAY, | |
2416 | * or ENCAP | |
2417 | */ | |
e22e8001 SW |
2418 | nh.type = NEXTHOP_TYPE_BLACKHOLE; |
2419 | nh.bh_type = BLACKHOLE_UNSPEC; | |
2420 | } else if (tb[NHA_OIF]) | |
85f5e761 SW |
2421 | /** |
2422 | * This is a true new nexthop, so we need | |
2423 | * to parse the gateway and device info | |
2424 | */ | |
2425 | nh = netlink_nexthop_process_nh(tb, family, | |
2426 | &ifp, ns_id); | |
e22e8001 SW |
2427 | else { |
2428 | ||
8e401b25 SW |
2429 | flog_warn( |
2430 | EC_ZEBRA_BAD_NHG_MESSAGE, | |
2431 | "Invalid Nexthop message received from the kernel with ID (%u)", | |
2432 | id); | |
2433 | return -1; | |
2434 | } | |
e22e8001 SW |
2435 | SET_FLAG(nh.flags, NEXTHOP_FLAG_ACTIVE); |
2436 | if (nhm->nh_flags & RTNH_F_ONLINK) | |
2437 | SET_FLAG(nh.flags, NEXTHOP_FLAG_ONLINK); | |
2438 | vrf_id = nh.vrf_id; | |
d9f5b2f5 SW |
2439 | } |
2440 | ||
38e40db1 SW |
2441 | if (zebra_nhg_kernel_find(id, &nh, grp, grp_count, vrf_id, afi, |
2442 | type, startup)) | |
e22e8001 | 2443 | return -1; |
8e401b25 | 2444 | |
9a1588c4 SW |
2445 | } else if (h->nlmsg_type == RTM_DELNEXTHOP) |
2446 | zebra_nhg_kernel_del(id); | |
d9f5b2f5 | 2447 | |
d9f5b2f5 SW |
2448 | return 0; |
2449 | } | |
2450 | ||
2451 | /** | |
2452 | * netlink_request_nexthop() - Request nextop information from the kernel | |
2453 | * @zns: Zebra namespace | |
2454 | * @family: AF_* netlink family | |
2455 | * @type: RTM_* route type | |
2456 | * | |
2457 | * Return: Result status | |
2458 | */ | |
2459 | static int netlink_request_nexthop(struct zebra_ns *zns, int family, int type) | |
2460 | { | |
2461 | struct { | |
2462 | struct nlmsghdr n; | |
2463 | struct nhmsg nhm; | |
2464 | } req; | |
2465 | ||
2466 | /* Form the request, specifying filter (rtattr) if needed. */ | |
2467 | memset(&req, 0, sizeof(req)); | |
2468 | req.n.nlmsg_type = type; | |
2469 | req.n.nlmsg_flags = NLM_F_ROOT | NLM_F_MATCH | NLM_F_REQUEST; | |
2470 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct nhmsg)); | |
2471 | req.nhm.nh_family = family; | |
2472 | ||
2473 | return netlink_request(&zns->netlink_cmd, &req.n); | |
2474 | } | |
2475 | ||
7d5bb02b | 2476 | |
d9f5b2f5 SW |
2477 | /** |
2478 | * netlink_nexthop_read() - Nexthop read function using netlink interface | |
2479 | * | |
2480 | * @zns: Zebra name space | |
2481 | * | |
2482 | * Return: Result status | |
2483 | * Only called at bootstrap time. | |
2484 | */ | |
2485 | int netlink_nexthop_read(struct zebra_ns *zns) | |
2486 | { | |
2487 | int ret; | |
2488 | struct zebra_dplane_info dp_info; | |
2489 | ||
2490 | zebra_dplane_info_from_zns(&dp_info, zns, true /*is_cmd*/); | |
2491 | ||
2492 | /* Get nexthop objects */ | |
2493 | ret = netlink_request_nexthop(zns, AF_UNSPEC, RTM_GETNEXTHOP); | |
2494 | if (ret < 0) | |
2495 | return ret; | |
2496 | ret = netlink_parse_info(netlink_nexthop_change, &zns->netlink_cmd, | |
2497 | &dp_info, 0, 1); | |
81505946 SW |
2498 | |
2499 | if (!ret) | |
2500 | /* If we succesfully read in nexthop objects, | |
2501 | * this kernel must support them. | |
2502 | */ | |
2503 | supports_nh = true; | |
2504 | else if (IS_ZEBRA_DEBUG_KERNEL) | |
2505 | zlog_debug("Nexthop objects not supported on this kernel"); | |
2506 | ||
60e0eaee | 2507 | return ret; |
d9f5b2f5 SW |
2508 | } |
2509 | ||
2510 | ||
d62a17ae | 2511 | int kernel_neigh_update(int add, int ifindex, uint32_t addr, char *lla, |
5895d33f | 2512 | int llalen, ns_id_t ns_id) |
6b8a5694 | 2513 | { |
d62a17ae | 2514 | return netlink_neigh_update(add ? RTM_NEWNEIGH : RTM_DELNEIGH, ifindex, |
5895d33f | 2515 | addr, lla, llalen, ns_id); |
6b8a5694 | 2516 | } |
718e3744 | 2517 | |
13d60d35 | 2518 | /* |
2519 | * Add remote VTEP to the flood list for this VxLAN interface (VNI). This | |
2520 | * is done by adding an FDB entry with a MAC of 00:00:00:00:00:00. | |
2521 | */ | |
0bbd4ff4 MS |
2522 | static int netlink_vxlan_flood_update_ctx(const struct zebra_dplane_ctx *ctx, |
2523 | int cmd) | |
13d60d35 | 2524 | { |
f3dbec60 | 2525 | uint8_t protocol = RTPROT_ZEBRA; |
d62a17ae | 2526 | struct { |
2527 | struct nlmsghdr n; | |
2528 | struct ndmsg ndm; | |
2529 | char buf[256]; | |
2530 | } req; | |
d7c0a89a | 2531 | uint8_t dst_mac[6] = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0}; |
0bbd4ff4 | 2532 | const struct ipaddr *addr; |
d62a17ae | 2533 | |
5605ecfc | 2534 | memset(&req, 0, sizeof(req)); |
d62a17ae | 2535 | |
2536 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
2537 | req.n.nlmsg_flags = NLM_F_REQUEST; | |
2538 | if (cmd == RTM_NEWNEIGH) | |
2539 | req.n.nlmsg_flags |= (NLM_F_CREATE | NLM_F_APPEND); | |
2540 | req.n.nlmsg_type = cmd; | |
2541 | req.ndm.ndm_family = PF_BRIDGE; | |
2542 | req.ndm.ndm_state = NUD_NOARP | NUD_PERMANENT; | |
fec211ad | 2543 | req.ndm.ndm_flags |= NTF_SELF; /* Handle by "self", not "master" */ |
d62a17ae | 2544 | |
2545 | ||
f3dbec60 DS |
2546 | addattr_l(&req.n, sizeof(req), |
2547 | NDA_PROTOCOL, &protocol, sizeof(protocol)); | |
d62a17ae | 2548 | addattr_l(&req.n, sizeof(req), NDA_LLADDR, &dst_mac, 6); |
0bbd4ff4 | 2549 | req.ndm.ndm_ifindex = dplane_ctx_get_ifindex(ctx); |
13d60d35 | 2550 | |
0bbd4ff4 | 2551 | addr = dplane_ctx_neigh_get_ipaddr(ctx); |
13d60d35 | 2552 | |
0bbd4ff4 | 2553 | addattr_l(&req.n, sizeof(req), NDA_DST, &(addr->ipaddr_v4), 4); |
13d60d35 | 2554 | |
0bbd4ff4 MS |
2555 | return netlink_talk_info(netlink_talk_filter, &req.n, |
2556 | dplane_ctx_get_ns(ctx), 0); | |
13d60d35 | 2557 | } |
2558 | ||
2232a77c | 2559 | #ifndef NDA_RTA |
d62a17ae | 2560 | #define NDA_RTA(r) \ |
2561 | ((struct rtattr *)(((char *)(r)) + NLMSG_ALIGN(sizeof(struct ndmsg)))) | |
2232a77c | 2562 | #endif |
2563 | ||
2414abd3 | 2564 | static int netlink_macfdb_change(struct nlmsghdr *h, int len, ns_id_t ns_id) |
2232a77c | 2565 | { |
d62a17ae | 2566 | struct ndmsg *ndm; |
2567 | struct interface *ifp; | |
2568 | struct zebra_if *zif; | |
d62a17ae | 2569 | struct rtattr *tb[NDA_MAX + 1]; |
2570 | struct interface *br_if; | |
2571 | struct ethaddr mac; | |
2572 | vlanid_t vid = 0; | |
2573 | struct prefix vtep_ip; | |
2574 | int vid_present = 0, dst_present = 0; | |
2575 | char buf[ETHER_ADDR_STRLEN]; | |
2576 | char vid_buf[20]; | |
2577 | char dst_buf[30]; | |
a37f4598 | 2578 | bool sticky; |
d62a17ae | 2579 | |
2580 | ndm = NLMSG_DATA(h); | |
2581 | ||
2853fed6 | 2582 | /* We only process macfdb notifications if EVPN is enabled */ |
2583 | if (!is_evpn_enabled()) | |
2584 | return 0; | |
2585 | ||
d62a17ae | 2586 | /* The interface should exist. */ |
5895d33f | 2587 | ifp = if_lookup_by_index_per_ns(zebra_ns_lookup(ns_id), |
d62a17ae | 2588 | ndm->ndm_ifindex); |
28bd0652 DS |
2589 | if (!ifp || !ifp->info) { |
2590 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2591 | zlog_debug("\t%s without associated interface: %u", | |
2592 | __PRETTY_FUNCTION__, ndm->ndm_ifindex); | |
d62a17ae | 2593 | return 0; |
28bd0652 | 2594 | } |
d62a17ae | 2595 | |
2596 | /* The interface should be something we're interested in. */ | |
28bd0652 DS |
2597 | if (!IS_ZEBRA_IF_BRIDGE_SLAVE(ifp)) { |
2598 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2599 | zlog_debug("\t%s Not interested in %s, not a slave", | |
2600 | __PRETTY_FUNCTION__, ifp->name); | |
d62a17ae | 2601 | return 0; |
28bd0652 | 2602 | } |
d62a17ae | 2603 | |
2604 | /* Drop "permanent" entries. */ | |
28bd0652 DS |
2605 | if (ndm->ndm_state & NUD_PERMANENT) { |
2606 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2607 | zlog_debug("\t%s Entry is PERMANENT, dropping", | |
2608 | __PRETTY_FUNCTION__); | |
d62a17ae | 2609 | return 0; |
28bd0652 | 2610 | } |
d62a17ae | 2611 | |
2612 | zif = (struct zebra_if *)ifp->info; | |
2613 | if ((br_if = zif->brslave_info.br_if) == NULL) { | |
28bd0652 DS |
2614 | if (IS_ZEBRA_DEBUG_KERNEL) |
2615 | zlog_debug( | |
2616 | "%s family %s IF %s(%u) brIF %u - no bridge master", | |
2617 | nl_msg_type_to_str(h->nlmsg_type), | |
2618 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
2619 | ndm->ndm_ifindex, | |
2620 | zif->brslave_info.bridge_ifindex); | |
d62a17ae | 2621 | return 0; |
2622 | } | |
2623 | ||
2624 | /* Parse attributes and extract fields of interest. */ | |
2625 | memset(tb, 0, sizeof tb); | |
2626 | netlink_parse_rtattr(tb, NDA_MAX, NDA_RTA(ndm), len); | |
2627 | ||
2628 | if (!tb[NDA_LLADDR]) { | |
28bd0652 DS |
2629 | if (IS_ZEBRA_DEBUG_KERNEL) |
2630 | zlog_debug("%s family %s IF %s(%u) brIF %u - no LLADDR", | |
2631 | nl_msg_type_to_str(h->nlmsg_type), | |
2632 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
2633 | ndm->ndm_ifindex, | |
2634 | zif->brslave_info.bridge_ifindex); | |
d62a17ae | 2635 | return 0; |
2636 | } | |
2637 | ||
ff8b7eb8 | 2638 | if (RTA_PAYLOAD(tb[NDA_LLADDR]) != ETH_ALEN) { |
28bd0652 DS |
2639 | if (IS_ZEBRA_DEBUG_KERNEL) |
2640 | zlog_debug( | |
2641 | "%s family %s IF %s(%u) brIF %u - LLADDR is not MAC, len %lu", | |
2642 | nl_msg_type_to_str(h->nlmsg_type), | |
2643 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
2644 | ndm->ndm_ifindex, | |
2645 | zif->brslave_info.bridge_ifindex, | |
2646 | (unsigned long)RTA_PAYLOAD(tb[NDA_LLADDR])); | |
d62a17ae | 2647 | return 0; |
2648 | } | |
2649 | ||
ff8b7eb8 | 2650 | memcpy(&mac, RTA_DATA(tb[NDA_LLADDR]), ETH_ALEN); |
d62a17ae | 2651 | |
2652 | if ((NDA_VLAN <= NDA_MAX) && tb[NDA_VLAN]) { | |
2653 | vid_present = 1; | |
d7c0a89a | 2654 | vid = *(uint16_t *)RTA_DATA(tb[NDA_VLAN]); |
d62a17ae | 2655 | sprintf(vid_buf, " VLAN %u", vid); |
2656 | } | |
2657 | ||
2658 | if (tb[NDA_DST]) { | |
2659 | /* TODO: Only IPv4 supported now. */ | |
2660 | dst_present = 1; | |
2661 | vtep_ip.family = AF_INET; | |
2662 | vtep_ip.prefixlen = IPV4_MAX_BITLEN; | |
2663 | memcpy(&(vtep_ip.u.prefix4.s_addr), RTA_DATA(tb[NDA_DST]), | |
2664 | IPV4_MAX_BYTELEN); | |
2665 | sprintf(dst_buf, " dst %s", inet_ntoa(vtep_ip.u.prefix4)); | |
2666 | } | |
2667 | ||
a37f4598 | 2668 | sticky = !!(ndm->ndm_state & NUD_NOARP); |
d62a17ae | 2669 | |
2670 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2671 | zlog_debug("Rx %s family %s IF %s(%u)%s %sMAC %s%s", | |
2672 | nl_msg_type_to_str(h->nlmsg_type), | |
2673 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
2674 | ndm->ndm_ifindex, vid_present ? vid_buf : "", | |
2675 | sticky ? "sticky " : "", | |
2676 | prefix_mac2str(&mac, buf, sizeof(buf)), | |
2677 | dst_present ? dst_buf : ""); | |
2678 | ||
28bd0652 DS |
2679 | if (filter_vlan && vid != filter_vlan) { |
2680 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2681 | zlog_debug("\tFiltered due to filter vlan: %d", | |
2682 | filter_vlan); | |
d62a17ae | 2683 | return 0; |
28bd0652 | 2684 | } |
d62a17ae | 2685 | |
2686 | /* If add or update, do accordingly if learnt on a "local" interface; if | |
2687 | * the notification is over VxLAN, this has to be related to | |
2688 | * multi-homing, | |
2689 | * so perform an implicit delete of any local entry (if it exists). | |
2690 | */ | |
2691 | if (h->nlmsg_type == RTM_NEWNEIGH) { | |
d62a17ae | 2692 | if (IS_ZEBRA_IF_VXLAN(ifp)) |
2693 | return zebra_vxlan_check_del_local_mac(ifp, br_if, &mac, | |
2694 | vid); | |
2695 | ||
2696 | return zebra_vxlan_local_mac_add_update(ifp, br_if, &mac, vid, | |
2697 | sticky); | |
2698 | } | |
2699 | ||
2700 | /* This is a delete notification. | |
2701 | * 1. For a MAC over VxLan, check if it needs to be refreshed(readded) | |
2702 | * 2. For a MAC over "local" interface, delete the mac | |
2703 | * Note: We will get notifications from both bridge driver and VxLAN | |
2704 | * driver. | |
2705 | * Ignore the notification from VxLan driver as it is also generated | |
2706 | * when mac moves from remote to local. | |
2707 | */ | |
28bd0652 DS |
2708 | if (dst_present) { |
2709 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2710 | zlog_debug("\tNo Destination Present"); | |
d62a17ae | 2711 | return 0; |
28bd0652 | 2712 | } |
d62a17ae | 2713 | |
2714 | if (IS_ZEBRA_IF_VXLAN(ifp)) | |
2715 | return zebra_vxlan_check_readd_remote_mac(ifp, br_if, &mac, | |
2716 | vid); | |
2717 | ||
2718 | return zebra_vxlan_local_mac_del(ifp, br_if, &mac, vid); | |
2232a77c | 2719 | } |
2720 | ||
2414abd3 | 2721 | static int netlink_macfdb_table(struct nlmsghdr *h, ns_id_t ns_id, int startup) |
2232a77c | 2722 | { |
d62a17ae | 2723 | int len; |
2724 | struct ndmsg *ndm; | |
2232a77c | 2725 | |
d62a17ae | 2726 | if (h->nlmsg_type != RTM_NEWNEIGH) |
2727 | return 0; | |
2232a77c | 2728 | |
d62a17ae | 2729 | /* Length validity. */ |
2730 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct ndmsg)); | |
2731 | if (len < 0) | |
2732 | return -1; | |
2232a77c | 2733 | |
d62a17ae | 2734 | /* We are interested only in AF_BRIDGE notifications. */ |
2735 | ndm = NLMSG_DATA(h); | |
2736 | if (ndm->ndm_family != AF_BRIDGE) | |
2737 | return 0; | |
2232a77c | 2738 | |
2414abd3 | 2739 | return netlink_macfdb_change(h, len, ns_id); |
2232a77c | 2740 | } |
2741 | ||
2742 | /* Request for MAC FDB information from the kernel */ | |
85a75f1e MS |
2743 | static int netlink_request_macs(struct nlsock *netlink_cmd, int family, |
2744 | int type, ifindex_t master_ifindex) | |
2232a77c | 2745 | { |
d62a17ae | 2746 | struct { |
2747 | struct nlmsghdr n; | |
2748 | struct ifinfomsg ifm; | |
2749 | char buf[256]; | |
2750 | } req; | |
2751 | ||
2752 | /* Form the request, specifying filter (rtattr) if needed. */ | |
2753 | memset(&req, 0, sizeof(req)); | |
2754 | req.n.nlmsg_type = type; | |
718f9b0f | 2755 | req.n.nlmsg_flags = NLM_F_ROOT | NLM_F_MATCH | NLM_F_REQUEST; |
d62a17ae | 2756 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ifinfomsg)); |
2757 | req.ifm.ifi_family = family; | |
2758 | if (master_ifindex) | |
2759 | addattr32(&req.n, sizeof(req), IFLA_MASTER, master_ifindex); | |
2760 | ||
85a75f1e | 2761 | return netlink_request(netlink_cmd, &req.n); |
2232a77c | 2762 | } |
2763 | ||
2764 | /* | |
2765 | * MAC forwarding database read using netlink interface. This is invoked | |
2766 | * at startup. | |
2767 | */ | |
d62a17ae | 2768 | int netlink_macfdb_read(struct zebra_ns *zns) |
2232a77c | 2769 | { |
d62a17ae | 2770 | int ret; |
85a75f1e MS |
2771 | struct zebra_dplane_info dp_info; |
2772 | ||
2773 | zebra_dplane_info_from_zns(&dp_info, zns, true /*is_cmd*/); | |
d62a17ae | 2774 | |
2775 | /* Get bridge FDB table. */ | |
85a75f1e MS |
2776 | ret = netlink_request_macs(&zns->netlink_cmd, AF_BRIDGE, RTM_GETNEIGH, |
2777 | 0); | |
d62a17ae | 2778 | if (ret < 0) |
2779 | return ret; | |
2780 | /* We are reading entire table. */ | |
2781 | filter_vlan = 0; | |
85a75f1e MS |
2782 | ret = netlink_parse_info(netlink_macfdb_table, &zns->netlink_cmd, |
2783 | &dp_info, 0, 1); | |
d62a17ae | 2784 | |
2785 | return ret; | |
2232a77c | 2786 | } |
2787 | ||
2788 | /* | |
2789 | * MAC forwarding database read using netlink interface. This is for a | |
2790 | * specific bridge and matching specific access VLAN (if VLAN-aware bridge). | |
2791 | */ | |
d62a17ae | 2792 | int netlink_macfdb_read_for_bridge(struct zebra_ns *zns, struct interface *ifp, |
2793 | struct interface *br_if) | |
2232a77c | 2794 | { |
d62a17ae | 2795 | struct zebra_if *br_zif; |
2796 | struct zebra_if *zif; | |
2797 | struct zebra_l2info_vxlan *vxl; | |
85a75f1e | 2798 | struct zebra_dplane_info dp_info; |
d62a17ae | 2799 | int ret = 0; |
2800 | ||
85a75f1e | 2801 | zebra_dplane_info_from_zns(&dp_info, zns, true /*is_cmd*/); |
d62a17ae | 2802 | |
2803 | /* Save VLAN we're filtering on, if needed. */ | |
2804 | br_zif = (struct zebra_if *)br_if->info; | |
2805 | zif = (struct zebra_if *)ifp->info; | |
2806 | vxl = &zif->l2info.vxl; | |
2807 | if (IS_ZEBRA_IF_BRIDGE_VLAN_AWARE(br_zif)) | |
2808 | filter_vlan = vxl->access_vlan; | |
2809 | ||
2810 | /* Get bridge FDB table for specific bridge - we do the VLAN filtering. | |
2811 | */ | |
85a75f1e | 2812 | ret = netlink_request_macs(&zns->netlink_cmd, AF_BRIDGE, RTM_GETNEIGH, |
d62a17ae | 2813 | br_if->ifindex); |
2814 | if (ret < 0) | |
2815 | return ret; | |
85a75f1e MS |
2816 | ret = netlink_parse_info(netlink_macfdb_table, &zns->netlink_cmd, |
2817 | &dp_info, 0, 0); | |
d62a17ae | 2818 | |
2819 | /* Reset VLAN filter. */ | |
2820 | filter_vlan = 0; | |
2821 | return ret; | |
2232a77c | 2822 | } |
2823 | ||
67fb9374 CS |
2824 | |
2825 | /* Request for MAC FDB for a specific MAC address in VLAN from the kernel */ | |
2826 | static int netlink_request_specific_mac_in_bridge(struct zebra_ns *zns, | |
2827 | int family, | |
2828 | int type, | |
2829 | struct interface *br_if, | |
2830 | struct ethaddr *mac, | |
2831 | vlanid_t vid) | |
2832 | { | |
2833 | struct { | |
2834 | struct nlmsghdr n; | |
2835 | struct ndmsg ndm; | |
2836 | char buf[256]; | |
2837 | } req; | |
2838 | struct zebra_if *br_zif; | |
2839 | char buf[ETHER_ADDR_STRLEN]; | |
2840 | ||
2841 | memset(&req, 0, sizeof(req)); | |
2842 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
2843 | req.n.nlmsg_type = type; /* RTM_GETNEIGH */ | |
2844 | req.n.nlmsg_flags = NLM_F_REQUEST; | |
2845 | req.ndm.ndm_family = family; /* AF_BRIDGE */ | |
2846 | /* req.ndm.ndm_state = NUD_REACHABLE; */ | |
2847 | ||
2848 | addattr_l(&req.n, sizeof(req), NDA_LLADDR, mac, 6); | |
2849 | ||
2850 | br_zif = (struct zebra_if *)br_if->info; | |
2851 | if (IS_ZEBRA_IF_BRIDGE_VLAN_AWARE(br_zif) && vid > 0) | |
2852 | addattr16(&req.n, sizeof(req), NDA_VLAN, vid); | |
2853 | ||
2854 | addattr32(&req.n, sizeof(req), NDA_MASTER, br_if->ifindex); | |
2855 | ||
2856 | if (IS_ZEBRA_DEBUG_KERNEL) | |
2857 | zlog_debug("%s: Tx family %s IF %s(%u) MAC %s vid %u", | |
2858 | __PRETTY_FUNCTION__, | |
2859 | nl_family_to_str(req.ndm.ndm_family), br_if->name, | |
2860 | br_if->ifindex, | |
2861 | prefix_mac2str(mac, buf, sizeof(buf)), vid); | |
2862 | ||
2863 | return netlink_request(&zns->netlink_cmd, &req.n); | |
2864 | } | |
2865 | ||
2866 | int netlink_macfdb_read_specific_mac(struct zebra_ns *zns, | |
2867 | struct interface *br_if, | |
2868 | struct ethaddr *mac, vlanid_t vid) | |
2869 | { | |
2870 | int ret = 0; | |
2871 | struct zebra_dplane_info dp_info; | |
2872 | ||
2873 | zebra_dplane_info_from_zns(&dp_info, zns, true /*is_cmd*/); | |
2874 | ||
2875 | /* Get bridge FDB table for specific bridge - we do the VLAN filtering. | |
2876 | */ | |
2877 | ret = netlink_request_specific_mac_in_bridge(zns, AF_BRIDGE, | |
2878 | RTM_GETNEIGH, | |
2879 | br_if, mac, vid); | |
2880 | if (ret < 0) | |
2881 | return ret; | |
2882 | ||
2883 | ret = netlink_parse_info(netlink_macfdb_table, &zns->netlink_cmd, | |
2884 | &dp_info, 1, 0); | |
2885 | ||
2886 | return ret; | |
2887 | } | |
036d93c0 MS |
2888 | |
2889 | /* | |
2890 | * Netlink-specific handler for MAC updates using dataplane context object. | |
2891 | */ | |
2892 | static enum zebra_dplane_result | |
2893 | netlink_macfdb_update_ctx(struct zebra_dplane_ctx *ctx) | |
2232a77c | 2894 | { |
f3dbec60 | 2895 | uint8_t protocol = RTPROT_ZEBRA; |
d62a17ae | 2896 | struct { |
2897 | struct nlmsghdr n; | |
2898 | struct ndmsg ndm; | |
2899 | char buf[256]; | |
2900 | } req; | |
036d93c0 | 2901 | int ret; |
d62a17ae | 2902 | int dst_alen; |
36590076 | 2903 | int vid_present = 0; |
036d93c0 MS |
2904 | int cmd; |
2905 | struct in_addr vtep_ip; | |
2906 | vlanid_t vid; | |
2907 | ||
2908 | if (dplane_ctx_get_op(ctx) == DPLANE_OP_MAC_INSTALL) | |
2909 | cmd = RTM_NEWNEIGH; | |
2910 | else | |
2911 | cmd = RTM_DELNEIGH; | |
2912 | ||
5605ecfc | 2913 | memset(&req, 0, sizeof(req)); |
d62a17ae | 2914 | |
2915 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
2916 | req.n.nlmsg_flags = NLM_F_REQUEST; | |
2917 | if (cmd == RTM_NEWNEIGH) | |
2918 | req.n.nlmsg_flags |= (NLM_F_CREATE | NLM_F_REPLACE); | |
2919 | req.n.nlmsg_type = cmd; | |
2920 | req.ndm.ndm_family = AF_BRIDGE; | |
2921 | req.ndm.ndm_flags |= NTF_SELF | NTF_MASTER; | |
2922 | req.ndm.ndm_state = NUD_REACHABLE; | |
2923 | ||
036d93c0 | 2924 | if (dplane_ctx_mac_is_sticky(ctx)) |
d62a17ae | 2925 | req.ndm.ndm_state |= NUD_NOARP; |
2926 | else | |
2927 | req.ndm.ndm_flags |= NTF_EXT_LEARNED; | |
2928 | ||
f3dbec60 DS |
2929 | addattr_l(&req.n, sizeof(req), |
2930 | NDA_PROTOCOL, &protocol, sizeof(protocol)); | |
036d93c0 MS |
2931 | addattr_l(&req.n, sizeof(req), NDA_LLADDR, |
2932 | dplane_ctx_mac_get_addr(ctx), 6); | |
478566d6 | 2933 | req.ndm.ndm_ifindex = dplane_ctx_get_ifindex(ctx); |
036d93c0 | 2934 | |
d63c1b18 | 2935 | dst_alen = 4; // TODO: hardcoded |
036d93c0 | 2936 | vtep_ip = *(dplane_ctx_mac_get_vtep_ip(ctx)); |
d63c1b18 | 2937 | addattr_l(&req.n, sizeof(req), NDA_DST, &vtep_ip, dst_alen); |
036d93c0 | 2938 | |
478566d6 MS |
2939 | vid = dplane_ctx_mac_get_vlan(ctx); |
2940 | ||
2941 | if (vid > 0) { | |
d62a17ae | 2942 | addattr16(&req.n, sizeof(req), NDA_VLAN, vid); |
2943 | vid_present = 1; | |
d62a17ae | 2944 | } |
478566d6 MS |
2945 | addattr32(&req.n, sizeof(req), NDA_MASTER, |
2946 | dplane_ctx_mac_get_br_ifindex(ctx)); | |
d62a17ae | 2947 | |
036d93c0 MS |
2948 | if (IS_ZEBRA_DEBUG_KERNEL) { |
2949 | char ipbuf[PREFIX_STRLEN]; | |
2950 | char buf[ETHER_ADDR_STRLEN]; | |
2951 | char dst_buf[PREFIX_STRLEN + 10]; | |
478566d6 MS |
2952 | char vid_buf[20]; |
2953 | ||
2954 | if (vid_present) | |
2955 | snprintf(vid_buf, sizeof(vid_buf), " VLAN %u", vid); | |
2956 | else | |
2957 | vid_buf[0] = '\0'; | |
036d93c0 MS |
2958 | |
2959 | inet_ntop(AF_INET, &vtep_ip, ipbuf, sizeof(ipbuf)); | |
2960 | snprintf(dst_buf, sizeof(dst_buf), " dst %s", ipbuf); | |
2961 | prefix_mac2str(dplane_ctx_mac_get_addr(ctx), buf, sizeof(buf)); | |
2962 | ||
d62a17ae | 2963 | zlog_debug("Tx %s family %s IF %s(%u)%s %sMAC %s%s", |
2964 | nl_msg_type_to_str(cmd), | |
478566d6 MS |
2965 | nl_family_to_str(req.ndm.ndm_family), |
2966 | dplane_ctx_get_ifname(ctx), | |
2967 | dplane_ctx_get_ifindex(ctx), vid_buf, | |
036d93c0 MS |
2968 | dplane_ctx_mac_is_sticky(ctx) ? "sticky " : "", |
2969 | buf, dst_buf); | |
2970 | } | |
d62a17ae | 2971 | |
036d93c0 MS |
2972 | ret = netlink_talk_info(netlink_talk_filter, &req.n, |
2973 | dplane_ctx_get_ns(ctx), 0); | |
2974 | if (ret == 0) | |
2975 | return ZEBRA_DPLANE_REQUEST_SUCCESS; | |
2976 | else | |
2977 | return ZEBRA_DPLANE_REQUEST_FAILURE; | |
2232a77c | 2978 | } |
2979 | ||
f17b99ed DS |
2980 | /* |
2981 | * In the event the kernel deletes ipv4 link-local neighbor entries created for | |
2982 | * 5549 support, re-install them. | |
2983 | */ | |
2984 | static void netlink_handle_5549(struct ndmsg *ndm, struct zebra_if *zif, | |
9b036974 DS |
2985 | struct interface *ifp, struct ipaddr *ip, |
2986 | bool handle_failed) | |
f17b99ed DS |
2987 | { |
2988 | if (ndm->ndm_family != AF_INET) | |
2989 | return; | |
2990 | ||
2991 | if (!zif->v6_2_v4_ll_neigh_entry) | |
2992 | return; | |
2993 | ||
2994 | if (ipv4_ll.s_addr != ip->ip._v4_addr.s_addr) | |
2995 | return; | |
2996 | ||
9b036974 DS |
2997 | if (handle_failed && ndm->ndm_state & NUD_FAILED) { |
2998 | zlog_info("Neighbor Entry for %s has entered a failed state, not reinstalling", | |
2999 | ifp->name); | |
3000 | return; | |
3001 | } | |
3002 | ||
f17b99ed DS |
3003 | if_nbr_ipv6ll_to_ipv4ll_neigh_update(ifp, &zif->v6_2_v4_ll_addr6, true); |
3004 | } | |
3005 | ||
d62a17ae | 3006 | #define NUD_VALID \ |
3007 | (NUD_PERMANENT | NUD_NOARP | NUD_REACHABLE | NUD_PROBE | NUD_STALE \ | |
3008 | | NUD_DELAY) | |
2232a77c | 3009 | |
2414abd3 | 3010 | static int netlink_ipneigh_change(struct nlmsghdr *h, int len, ns_id_t ns_id) |
2232a77c | 3011 | { |
d62a17ae | 3012 | struct ndmsg *ndm; |
3013 | struct interface *ifp; | |
3014 | struct zebra_if *zif; | |
d62a17ae | 3015 | struct rtattr *tb[NDA_MAX + 1]; |
3016 | struct interface *link_if; | |
3017 | struct ethaddr mac; | |
3018 | struct ipaddr ip; | |
3019 | char buf[ETHER_ADDR_STRLEN]; | |
3020 | char buf2[INET6_ADDRSTRLEN]; | |
3021 | int mac_present = 0; | |
a37f4598 | 3022 | bool is_ext; |
3023 | bool is_router; | |
d62a17ae | 3024 | |
3025 | ndm = NLMSG_DATA(h); | |
3026 | ||
3027 | /* The interface should exist. */ | |
5895d33f | 3028 | ifp = if_lookup_by_index_per_ns(zebra_ns_lookup(ns_id), |
d62a17ae | 3029 | ndm->ndm_ifindex); |
2853fed6 | 3030 | if (!ifp || !ifp->info) |
d62a17ae | 3031 | return 0; |
3032 | ||
20089ae2 DS |
3033 | zif = (struct zebra_if *)ifp->info; |
3034 | ||
3035 | /* Parse attributes and extract fields of interest. */ | |
3036 | memset(tb, 0, sizeof tb); | |
3037 | netlink_parse_rtattr(tb, NDA_MAX, NDA_RTA(ndm), len); | |
3038 | ||
3039 | if (!tb[NDA_DST]) { | |
9df414fe QY |
3040 | zlog_debug("%s family %s IF %s(%u) - no DST", |
3041 | nl_msg_type_to_str(h->nlmsg_type), | |
3042 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
3043 | ndm->ndm_ifindex); | |
d62a17ae | 3044 | return 0; |
20089ae2 DS |
3045 | } |
3046 | ||
3047 | memset(&ip, 0, sizeof(struct ipaddr)); | |
3048 | ip.ipa_type = (ndm->ndm_family == AF_INET) ? IPADDR_V4 : IPADDR_V6; | |
3049 | memcpy(&ip.ip.addr, RTA_DATA(tb[NDA_DST]), RTA_PAYLOAD(tb[NDA_DST])); | |
3050 | ||
f17b99ed DS |
3051 | /* if kernel deletes our rfc5549 neighbor entry, re-install it */ |
3052 | if (h->nlmsg_type == RTM_DELNEIGH && (ndm->ndm_state & NUD_PERMANENT)) { | |
9b036974 | 3053 | netlink_handle_5549(ndm, zif, ifp, &ip, false); |
28bd0652 DS |
3054 | if (IS_ZEBRA_DEBUG_KERNEL) |
3055 | zlog_debug( | |
3056 | "\tNeighbor Entry Received is a 5549 entry, finished"); | |
20089ae2 DS |
3057 | return 0; |
3058 | } | |
d62a17ae | 3059 | |
f17b99ed | 3060 | /* if kernel marks our rfc5549 neighbor entry invalid, re-install it */ |
9b036974 DS |
3061 | if (h->nlmsg_type == RTM_NEWNEIGH && !(ndm->ndm_state & NUD_VALID)) |
3062 | netlink_handle_5549(ndm, zif, ifp, &ip, true); | |
f17b99ed | 3063 | |
d62a17ae | 3064 | /* The neighbor is present on an SVI. From this, we locate the |
3065 | * underlying | |
3066 | * bridge because we're only interested in neighbors on a VxLAN bridge. | |
3067 | * The bridge is located based on the nature of the SVI: | |
3068 | * (a) In the case of a VLAN-aware bridge, the SVI is a L3 VLAN | |
3069 | * interface | |
3070 | * and is linked to the bridge | |
3071 | * (b) In the case of a VLAN-unaware bridge, the SVI is the bridge | |
3072 | * inteface | |
3073 | * itself | |
3074 | */ | |
3075 | if (IS_ZEBRA_IF_VLAN(ifp)) { | |
5895d33f | 3076 | link_if = if_lookup_by_index_per_ns(zebra_ns_lookup(ns_id), |
71349e03 | 3077 | zif->link_ifindex); |
d62a17ae | 3078 | if (!link_if) |
3079 | return 0; | |
3080 | } else if (IS_ZEBRA_IF_BRIDGE(ifp)) | |
3081 | link_if = ifp; | |
28bd0652 DS |
3082 | else { |
3083 | if (IS_ZEBRA_DEBUG_KERNEL) | |
3084 | zlog_debug( | |
3085 | "\tNeighbor Entry received is not on a VLAN or a BRIDGE, ignoring"); | |
d62a17ae | 3086 | return 0; |
28bd0652 | 3087 | } |
d62a17ae | 3088 | |
d62a17ae | 3089 | memset(&mac, 0, sizeof(struct ethaddr)); |
d62a17ae | 3090 | if (h->nlmsg_type == RTM_NEWNEIGH) { |
3091 | if (tb[NDA_LLADDR]) { | |
ff8b7eb8 | 3092 | if (RTA_PAYLOAD(tb[NDA_LLADDR]) != ETH_ALEN) { |
28bd0652 DS |
3093 | if (IS_ZEBRA_DEBUG_KERNEL) |
3094 | zlog_debug( | |
3095 | "%s family %s IF %s(%u) - LLADDR is not MAC, len %lu", | |
3096 | nl_msg_type_to_str( | |
3097 | h->nlmsg_type), | |
3098 | nl_family_to_str( | |
3099 | ndm->ndm_family), | |
3100 | ifp->name, ndm->ndm_ifindex, | |
3101 | (unsigned long)RTA_PAYLOAD( | |
3102 | tb[NDA_LLADDR])); | |
d62a17ae | 3103 | return 0; |
3104 | } | |
3105 | ||
3106 | mac_present = 1; | |
ff8b7eb8 | 3107 | memcpy(&mac, RTA_DATA(tb[NDA_LLADDR]), ETH_ALEN); |
d62a17ae | 3108 | } |
3109 | ||
a37f4598 | 3110 | is_ext = !!(ndm->ndm_flags & NTF_EXT_LEARNED); |
3111 | is_router = !!(ndm->ndm_flags & NTF_ROUTER); | |
d62a17ae | 3112 | |
3113 | if (IS_ZEBRA_DEBUG_KERNEL) | |
3114 | zlog_debug( | |
3115 | "Rx %s family %s IF %s(%u) IP %s MAC %s state 0x%x flags 0x%x", | |
3116 | nl_msg_type_to_str(h->nlmsg_type), | |
3117 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
3118 | ndm->ndm_ifindex, | |
3119 | ipaddr2str(&ip, buf2, sizeof(buf2)), | |
3120 | mac_present | |
3121 | ? prefix_mac2str(&mac, buf, sizeof(buf)) | |
3122 | : "", | |
3123 | ndm->ndm_state, ndm->ndm_flags); | |
3124 | ||
3125 | /* If the neighbor state is valid for use, process as an add or | |
3126 | * update | |
3127 | * else process as a delete. Note that the delete handling may | |
3128 | * result | |
3129 | * in re-adding the neighbor if it is a valid "remote" neighbor. | |
3130 | */ | |
3131 | if (ndm->ndm_state & NUD_VALID) | |
ee69da27 | 3132 | return zebra_vxlan_handle_kernel_neigh_update( |
d62a17ae | 3133 | ifp, link_if, &ip, &mac, ndm->ndm_state, |
a37f4598 | 3134 | is_ext, is_router); |
d62a17ae | 3135 | |
ee69da27 | 3136 | return zebra_vxlan_handle_kernel_neigh_del(ifp, link_if, &ip); |
d62a17ae | 3137 | } |
3138 | ||
3139 | if (IS_ZEBRA_DEBUG_KERNEL) | |
3140 | zlog_debug("Rx %s family %s IF %s(%u) IP %s", | |
3141 | nl_msg_type_to_str(h->nlmsg_type), | |
3142 | nl_family_to_str(ndm->ndm_family), ifp->name, | |
3143 | ndm->ndm_ifindex, | |
3144 | ipaddr2str(&ip, buf2, sizeof(buf2))); | |
3145 | ||
3146 | /* Process the delete - it may result in re-adding the neighbor if it is | |
3147 | * a valid "remote" neighbor. | |
3148 | */ | |
ee69da27 | 3149 | return zebra_vxlan_handle_kernel_neigh_del(ifp, link_if, &ip); |
2232a77c | 3150 | } |
3151 | ||
2414abd3 | 3152 | static int netlink_neigh_table(struct nlmsghdr *h, ns_id_t ns_id, int startup) |
2232a77c | 3153 | { |
d62a17ae | 3154 | int len; |
3155 | struct ndmsg *ndm; | |
2232a77c | 3156 | |
d62a17ae | 3157 | if (h->nlmsg_type != RTM_NEWNEIGH) |
3158 | return 0; | |
2232a77c | 3159 | |
d62a17ae | 3160 | /* Length validity. */ |
3161 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct ndmsg)); | |
3162 | if (len < 0) | |
3163 | return -1; | |
2232a77c | 3164 | |
d62a17ae | 3165 | /* We are interested only in AF_INET or AF_INET6 notifications. */ |
3166 | ndm = NLMSG_DATA(h); | |
3167 | if (ndm->ndm_family != AF_INET && ndm->ndm_family != AF_INET6) | |
3168 | return 0; | |
2232a77c | 3169 | |
2414abd3 | 3170 | return netlink_neigh_change(h, len); |
2232a77c | 3171 | } |
3172 | ||
3173 | /* Request for IP neighbor information from the kernel */ | |
85a75f1e MS |
3174 | static int netlink_request_neigh(struct nlsock *netlink_cmd, int family, |
3175 | int type, ifindex_t ifindex) | |
2232a77c | 3176 | { |
d62a17ae | 3177 | struct { |
3178 | struct nlmsghdr n; | |
3179 | struct ndmsg ndm; | |
3180 | char buf[256]; | |
3181 | } req; | |
3182 | ||
3183 | /* Form the request, specifying filter (rtattr) if needed. */ | |
3184 | memset(&req, 0, sizeof(req)); | |
3185 | req.n.nlmsg_type = type; | |
718f9b0f | 3186 | req.n.nlmsg_flags = NLM_F_ROOT | NLM_F_MATCH | NLM_F_REQUEST; |
d62a17ae | 3187 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); |
3188 | req.ndm.ndm_family = family; | |
3189 | if (ifindex) | |
3190 | addattr32(&req.n, sizeof(req), NDA_IFINDEX, ifindex); | |
3191 | ||
85a75f1e | 3192 | return netlink_request(netlink_cmd, &req.n); |
2232a77c | 3193 | } |
3194 | ||
3195 | /* | |
3196 | * IP Neighbor table read using netlink interface. This is invoked | |
3197 | * at startup. | |
3198 | */ | |
d62a17ae | 3199 | int netlink_neigh_read(struct zebra_ns *zns) |
2232a77c | 3200 | { |
d62a17ae | 3201 | int ret; |
85a75f1e MS |
3202 | struct zebra_dplane_info dp_info; |
3203 | ||
3204 | zebra_dplane_info_from_zns(&dp_info, zns, true /*is_cmd*/); | |
2232a77c | 3205 | |
d62a17ae | 3206 | /* Get IP neighbor table. */ |
85a75f1e MS |
3207 | ret = netlink_request_neigh(&zns->netlink_cmd, AF_UNSPEC, RTM_GETNEIGH, |
3208 | 0); | |
d62a17ae | 3209 | if (ret < 0) |
3210 | return ret; | |
85a75f1e MS |
3211 | ret = netlink_parse_info(netlink_neigh_table, &zns->netlink_cmd, |
3212 | &dp_info, 0, 1); | |
2232a77c | 3213 | |
d62a17ae | 3214 | return ret; |
2232a77c | 3215 | } |
3216 | ||
3217 | /* | |
3218 | * IP Neighbor table read using netlink interface. This is for a specific | |
3219 | * VLAN device. | |
3220 | */ | |
d62a17ae | 3221 | int netlink_neigh_read_for_vlan(struct zebra_ns *zns, struct interface *vlan_if) |
2232a77c | 3222 | { |
d62a17ae | 3223 | int ret = 0; |
85a75f1e MS |
3224 | struct zebra_dplane_info dp_info; |
3225 | ||
3226 | zebra_dplane_info_from_zns(&dp_info, zns, true /*is_cmd*/); | |
2232a77c | 3227 | |
85a75f1e | 3228 | ret = netlink_request_neigh(&zns->netlink_cmd, AF_UNSPEC, RTM_GETNEIGH, |
d62a17ae | 3229 | vlan_if->ifindex); |
3230 | if (ret < 0) | |
3231 | return ret; | |
85a75f1e MS |
3232 | ret = netlink_parse_info(netlink_neigh_table, &zns->netlink_cmd, |
3233 | &dp_info, 0, 0); | |
2232a77c | 3234 | |
d62a17ae | 3235 | return ret; |
2232a77c | 3236 | } |
3237 | ||
67fb9374 CS |
3238 | /* |
3239 | * Request for a specific IP in VLAN (SVI) device from IP Neighbor table, | |
3240 | * read using netlink interface. | |
3241 | */ | |
3242 | static int netlink_request_specific_neigh_in_vlan(struct zebra_ns *zns, | |
3243 | int type, struct ipaddr *ip, | |
3244 | ifindex_t ifindex) | |
3245 | { | |
3246 | struct { | |
3247 | struct nlmsghdr n; | |
3248 | struct ndmsg ndm; | |
3249 | char buf[256]; | |
3250 | } req; | |
3251 | int ipa_len; | |
3252 | ||
3253 | /* Form the request, specifying filter (rtattr) if needed. */ | |
3254 | memset(&req, 0, sizeof(req)); | |
3255 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); | |
3256 | req.n.nlmsg_flags = NLM_F_REQUEST; | |
3257 | req.n.nlmsg_type = type; /* RTM_GETNEIGH */ | |
3258 | req.ndm.ndm_ifindex = ifindex; | |
3259 | ||
3260 | if (IS_IPADDR_V4(ip)) { | |
3261 | ipa_len = IPV4_MAX_BYTELEN; | |
3262 | req.ndm.ndm_family = AF_INET; | |
3263 | ||
3264 | } else { | |
3265 | ipa_len = IPV6_MAX_BYTELEN; | |
3266 | req.ndm.ndm_family = AF_INET6; | |
3267 | } | |
3268 | ||
3269 | addattr_l(&req.n, sizeof(req), NDA_DST, &ip->ip.addr, ipa_len); | |
3270 | ||
3271 | return netlink_request(&zns->netlink_cmd, &req.n); | |
3272 | } | |
3273 | ||
3274 | int netlink_neigh_read_specific_ip(struct ipaddr *ip, | |
3275 | struct interface *vlan_if) | |
3276 | { | |
3277 | int ret = 0; | |
3278 | struct zebra_ns *zns; | |
3279 | struct zebra_vrf *zvrf = zebra_vrf_lookup_by_id(vlan_if->vrf_id); | |
3280 | char buf[INET6_ADDRSTRLEN]; | |
3281 | struct zebra_dplane_info dp_info; | |
3282 | ||
3283 | zns = zvrf->zns; | |
3284 | ||
3285 | zebra_dplane_info_from_zns(&dp_info, zns, true /*is_cmd*/); | |
3286 | ||
3287 | if (IS_ZEBRA_DEBUG_KERNEL) | |
3288 | zlog_debug("%s: neigh request IF %s(%u) IP %s vrf_id %u", | |
3289 | __PRETTY_FUNCTION__, vlan_if->name, | |
3290 | vlan_if->ifindex, | |
3291 | ipaddr2str(ip, buf, sizeof(buf)), | |
3292 | vlan_if->vrf_id); | |
3293 | ||
3294 | ret = netlink_request_specific_neigh_in_vlan(zns, RTM_GETNEIGH, ip, | |
3295 | vlan_if->ifindex); | |
3296 | if (ret < 0) | |
3297 | return ret; | |
3298 | ||
3299 | ret = netlink_parse_info(netlink_neigh_table, &zns->netlink_cmd, | |
3300 | &dp_info, 1, 0); | |
3301 | ||
3302 | return ret; | |
3303 | } | |
3304 | ||
2414abd3 | 3305 | int netlink_neigh_change(struct nlmsghdr *h, ns_id_t ns_id) |
2232a77c | 3306 | { |
d62a17ae | 3307 | int len; |
3308 | struct ndmsg *ndm; | |
2232a77c | 3309 | |
d62a17ae | 3310 | if (!(h->nlmsg_type == RTM_NEWNEIGH || h->nlmsg_type == RTM_DELNEIGH)) |
3311 | return 0; | |
2232a77c | 3312 | |
d62a17ae | 3313 | /* Length validity. */ |
3314 | len = h->nlmsg_len - NLMSG_LENGTH(sizeof(struct ndmsg)); | |
9bdf8618 DS |
3315 | if (len < 0) { |
3316 | zlog_err("%s: Message received from netlink is of a broken size %d %zu", | |
3317 | __PRETTY_FUNCTION__, h->nlmsg_len, | |
3318 | (size_t)NLMSG_LENGTH(sizeof(struct ndmsg))); | |
d62a17ae | 3319 | return -1; |
9bdf8618 | 3320 | } |
2232a77c | 3321 | |
d62a17ae | 3322 | /* Is this a notification for the MAC FDB or IP neighbor table? */ |
3323 | ndm = NLMSG_DATA(h); | |
3324 | if (ndm->ndm_family == AF_BRIDGE) | |
2414abd3 | 3325 | return netlink_macfdb_change(h, len, ns_id); |
2232a77c | 3326 | |
d62a17ae | 3327 | if (ndm->ndm_type != RTN_UNICAST) |
3328 | return 0; | |
2232a77c | 3329 | |
d62a17ae | 3330 | if (ndm->ndm_family == AF_INET || ndm->ndm_family == AF_INET6) |
2414abd3 | 3331 | return netlink_ipneigh_change(h, len, ns_id); |
8a1b681c | 3332 | else { |
9df414fe | 3333 | flog_warn( |
e914ccbe | 3334 | EC_ZEBRA_UNKNOWN_FAMILY, |
87b5d1b0 DS |
3335 | "Invalid address family: %u received from kernel neighbor change: %s", |
3336 | ndm->ndm_family, nl_msg_type_to_str(h->nlmsg_type)); | |
8a1b681c SW |
3337 | return 0; |
3338 | } | |
2232a77c | 3339 | |
d62a17ae | 3340 | return 0; |
2232a77c | 3341 | } |
3342 | ||
931fa60c MS |
3343 | /* |
3344 | * Utility neighbor-update function, using info from dplane context. | |
3345 | */ | |
3346 | static int netlink_neigh_update_ctx(const struct zebra_dplane_ctx *ctx, | |
3347 | int cmd) | |
2232a77c | 3348 | { |
f3dbec60 | 3349 | uint8_t protocol = RTPROT_ZEBRA; |
d62a17ae | 3350 | struct { |
3351 | struct nlmsghdr n; | |
3352 | struct ndmsg ndm; | |
3353 | char buf[256]; | |
3354 | } req; | |
3355 | int ipa_len; | |
d62a17ae | 3356 | char buf[INET6_ADDRSTRLEN]; |
3357 | char buf2[ETHER_ADDR_STRLEN]; | |
931fa60c MS |
3358 | const struct ipaddr *ip; |
3359 | const struct ethaddr *mac; | |
3360 | uint8_t flags; | |
3361 | uint16_t state; | |
d62a17ae | 3362 | |
5605ecfc | 3363 | memset(&req, 0, sizeof(req)); |
d62a17ae | 3364 | |
931fa60c MS |
3365 | ip = dplane_ctx_neigh_get_ipaddr(ctx); |
3366 | mac = dplane_ctx_neigh_get_mac(ctx); | |
3367 | if (is_zero_mac(mac)) | |
3368 | mac = NULL; | |
3369 | ||
3370 | flags = neigh_flags_to_netlink(dplane_ctx_neigh_get_flags(ctx)); | |
3371 | state = neigh_state_to_netlink(dplane_ctx_neigh_get_state(ctx)); | |
3372 | ||
d62a17ae | 3373 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ndmsg)); |
3374 | req.n.nlmsg_flags = NLM_F_REQUEST; | |
3375 | if (cmd == RTM_NEWNEIGH) | |
3376 | req.n.nlmsg_flags |= (NLM_F_CREATE | NLM_F_REPLACE); | |
3377 | req.n.nlmsg_type = cmd; // RTM_NEWNEIGH or RTM_DELNEIGH | |
3378 | req.ndm.ndm_family = IS_IPADDR_V4(ip) ? AF_INET : AF_INET6; | |
68e33151 | 3379 | req.ndm.ndm_state = state; |
931fa60c | 3380 | req.ndm.ndm_ifindex = dplane_ctx_get_ifindex(ctx); |
d62a17ae | 3381 | req.ndm.ndm_type = RTN_UNICAST; |
68e33151 | 3382 | req.ndm.ndm_flags = flags; |
d62a17ae | 3383 | |
f3dbec60 DS |
3384 | addattr_l(&req.n, sizeof(req), |
3385 | NDA_PROTOCOL, &protocol, sizeof(protocol)); | |
d62a17ae | 3386 | ipa_len = IS_IPADDR_V4(ip) ? IPV4_MAX_BYTELEN : IPV6_MAX_BYTELEN; |
3387 | addattr_l(&req.n, sizeof(req), NDA_DST, &ip->ip.addr, ipa_len); | |
3388 | if (mac) | |
3389 | addattr_l(&req.n, sizeof(req), NDA_LLADDR, mac, 6); | |
3390 | ||
3391 | if (IS_ZEBRA_DEBUG_KERNEL) | |
6fe2b0e6 | 3392 | zlog_debug("Tx %s family %s IF %s(%u) Neigh %s MAC %s flags 0x%x state 0x%x", |
d62a17ae | 3393 | nl_msg_type_to_str(cmd), |
931fa60c MS |
3394 | nl_family_to_str(req.ndm.ndm_family), |
3395 | dplane_ctx_get_ifname(ctx), | |
3396 | dplane_ctx_get_ifindex(ctx), | |
3397 | ipaddr2str(ip, buf, sizeof(buf)), | |
d62a17ae | 3398 | mac ? prefix_mac2str(mac, buf2, sizeof(buf2)) |
931fa60c MS |
3399 | : "null", |
3400 | flags, state); | |
d62a17ae | 3401 | |
931fa60c MS |
3402 | return netlink_talk_info(netlink_talk_filter, &req.n, |
3403 | dplane_ctx_get_ns(ctx), 0); | |
2232a77c | 3404 | } |
3405 | ||
036d93c0 MS |
3406 | /* |
3407 | * Update MAC, using dataplane context object. | |
3408 | */ | |
3409 | enum zebra_dplane_result kernel_mac_update_ctx(struct zebra_dplane_ctx *ctx) | |
2232a77c | 3410 | { |
036d93c0 | 3411 | return netlink_macfdb_update_ctx(ctx); |
2232a77c | 3412 | } |
3413 | ||
931fa60c | 3414 | enum zebra_dplane_result kernel_neigh_update_ctx(struct zebra_dplane_ctx *ctx) |
2232a77c | 3415 | { |
931fa60c | 3416 | int ret = -1; |
2232a77c | 3417 | |
931fa60c MS |
3418 | switch (dplane_ctx_get_op(ctx)) { |
3419 | case DPLANE_OP_NEIGH_INSTALL: | |
3420 | case DPLANE_OP_NEIGH_UPDATE: | |
3421 | ret = netlink_neigh_update_ctx(ctx, RTM_NEWNEIGH); | |
3422 | break; | |
3423 | case DPLANE_OP_NEIGH_DELETE: | |
3424 | ret = netlink_neigh_update_ctx(ctx, RTM_DELNEIGH); | |
3425 | break; | |
0bbd4ff4 MS |
3426 | case DPLANE_OP_VTEP_ADD: |
3427 | ret = netlink_vxlan_flood_update_ctx(ctx, RTM_NEWNEIGH); | |
3428 | break; | |
3429 | case DPLANE_OP_VTEP_DELETE: | |
3430 | ret = netlink_vxlan_flood_update_ctx(ctx, RTM_DELNEIGH); | |
3431 | break; | |
931fa60c MS |
3432 | default: |
3433 | break; | |
3434 | } | |
2232a77c | 3435 | |
931fa60c MS |
3436 | return (ret == 0 ? |
3437 | ZEBRA_DPLANE_REQUEST_SUCCESS : ZEBRA_DPLANE_REQUEST_FAILURE); | |
6fe2b0e6 CS |
3438 | } |
3439 | ||
16c628de MS |
3440 | /* |
3441 | * MPLS label forwarding table change via netlink interface, using dataplane | |
3442 | * context information. | |
3443 | */ | |
fc608372 | 3444 | int netlink_mpls_multipath(int cmd, struct zebra_dplane_ctx *ctx) |
16c628de MS |
3445 | { |
3446 | mpls_lse_t lse; | |
81793ac1 | 3447 | const zebra_nhlfe_t *nhlfe; |
16c628de MS |
3448 | struct nexthop *nexthop = NULL; |
3449 | unsigned int nexthop_num; | |
3450 | const char *routedesc; | |
3451 | int route_type; | |
3452 | ||
3453 | struct { | |
3454 | struct nlmsghdr n; | |
3455 | struct rtmsg r; | |
3456 | char buf[NL_PKT_BUF_SIZE]; | |
3457 | } req; | |
3458 | ||
3459 | memset(&req, 0, sizeof(req) - NL_PKT_BUF_SIZE); | |
3460 | ||
3461 | /* | |
3462 | * Count # nexthops so we can decide whether to use singlepath | |
3463 | * or multipath case. | |
3464 | */ | |
3465 | nexthop_num = 0; | |
3466 | for (nhlfe = dplane_ctx_get_nhlfe(ctx); nhlfe; nhlfe = nhlfe->next) { | |
3467 | nexthop = nhlfe->nexthop; | |
3468 | if (!nexthop) | |
3469 | continue; | |
3470 | if (cmd == RTM_NEWROUTE) { | |
3471 | /* Count all selected NHLFEs */ | |
3472 | if (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_SELECTED) | |
3473 | && CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ACTIVE)) | |
3474 | nexthop_num++; | |
3475 | } else { /* DEL */ | |
3476 | /* Count all installed NHLFEs */ | |
3477 | if (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_INSTALLED) | |
3478 | && CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB)) | |
3479 | nexthop_num++; | |
3480 | } | |
3481 | } | |
3482 | ||
3483 | if ((nexthop_num == 0) || | |
3484 | (!dplane_ctx_get_best_nhlfe(ctx) && (cmd != RTM_DELROUTE))) | |
3485 | return 0; | |
3486 | ||
3487 | req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg)); | |
3488 | req.n.nlmsg_flags = NLM_F_CREATE | NLM_F_REQUEST; | |
3489 | req.n.nlmsg_type = cmd; | |
3490 | req.n.nlmsg_pid = dplane_ctx_get_ns(ctx)->nls.snl.nl_pid; | |
3491 | ||
3492 | req.r.rtm_family = AF_MPLS; | |
3493 | req.r.rtm_table = RT_TABLE_MAIN; | |
3494 | req.r.rtm_dst_len = MPLS_LABEL_LEN_BITS; | |
3495 | req.r.rtm_scope = RT_SCOPE_UNIVERSE; | |
3496 | req.r.rtm_type = RTN_UNICAST; | |
3497 | ||
3498 | if (cmd == RTM_NEWROUTE) { | |
3499 | /* We do a replace to handle update. */ | |
3500 | req.n.nlmsg_flags |= NLM_F_REPLACE; | |
3501 | ||
3502 | /* set the protocol value if installing */ | |
3503 | route_type = re_type_from_lsp_type( | |
3504 | dplane_ctx_get_best_nhlfe(ctx)->type); | |
3505 | req.r.rtm_protocol = zebra2proto(route_type); | |
3506 | } | |
3507 | ||
3508 | /* Fill destination */ | |
3509 | lse = mpls_lse_encode(dplane_ctx_get_in_label(ctx), 0, 0, 1); | |
3510 | addattr_l(&req.n, sizeof(req), RTA_DST, &lse, sizeof(mpls_lse_t)); | |
3511 | ||
3512 | /* Fill nexthops (paths) based on single-path or multipath. The paths | |
3513 | * chosen depend on the operation. | |
3514 | */ | |
fc608372 | 3515 | if (nexthop_num == 1) { |
16c628de MS |
3516 | routedesc = "single-path"; |
3517 | _netlink_mpls_debug(cmd, dplane_ctx_get_in_label(ctx), | |
3518 | routedesc); | |
3519 | ||
3520 | nexthop_num = 0; | |
3521 | for (nhlfe = dplane_ctx_get_nhlfe(ctx); | |
3522 | nhlfe; nhlfe = nhlfe->next) { | |
3523 | nexthop = nhlfe->nexthop; | |
3524 | if (!nexthop) | |
3525 | continue; | |
3526 | ||
3527 | if ((cmd == RTM_NEWROUTE | |
3528 | && (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_SELECTED) | |
3529 | && CHECK_FLAG(nexthop->flags, | |
3530 | NEXTHOP_FLAG_ACTIVE))) | |
3531 | || (cmd == RTM_DELROUTE | |
3532 | && (CHECK_FLAG(nhlfe->flags, | |
3533 | NHLFE_FLAG_INSTALLED) | |
3534 | && CHECK_FLAG(nexthop->flags, | |
3535 | NEXTHOP_FLAG_FIB)))) { | |
3536 | /* Add the gateway */ | |
3537 | _netlink_mpls_build_singlepath( | |
3538 | routedesc, nhlfe, | |
3539 | &req.n, &req.r, | |
3540 | sizeof(req), cmd); | |
3541 | ||
3542 | nexthop_num++; | |
3543 | break; | |
3544 | } | |
3545 | } | |
3546 | } else { /* Multipath case */ | |
3547 | char buf[NL_PKT_BUF_SIZE]; | |
3548 | struct rtattr *rta = (void *)buf; | |
3549 | struct rtnexthop *rtnh; | |
81793ac1 | 3550 | const union g_addr *src1 = NULL; |
16c628de MS |
3551 | |
3552 | rta->rta_type = RTA_MULTIPATH; | |
3553 | rta->rta_len = RTA_LENGTH(0); | |
3554 | rtnh = RTA_DATA(rta); | |
3555 | ||
3556 | routedesc = "multipath"; | |
3557 | _netlink_mpls_debug(cmd, dplane_ctx_get_in_label(ctx), | |
3558 | routedesc); | |
3559 | ||
3560 | nexthop_num = 0; | |
3561 | for (nhlfe = dplane_ctx_get_nhlfe(ctx); | |
3562 | nhlfe; nhlfe = nhlfe->next) { | |
3563 | nexthop = nhlfe->nexthop; | |
3564 | if (!nexthop) | |
3565 | continue; | |
3566 | ||
16c628de MS |
3567 | if ((cmd == RTM_NEWROUTE |
3568 | && (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_SELECTED) | |
3569 | && CHECK_FLAG(nexthop->flags, | |
3570 | NEXTHOP_FLAG_ACTIVE))) | |
3571 | || (cmd == RTM_DELROUTE | |
3572 | && (CHECK_FLAG(nhlfe->flags, | |
3573 | NHLFE_FLAG_INSTALLED) | |
3574 | && CHECK_FLAG(nexthop->flags, | |
3575 | NEXTHOP_FLAG_FIB)))) { | |
3576 | nexthop_num++; | |
3577 | ||
3578 | /* Build the multipath */ | |
3579 | _netlink_mpls_build_multipath(routedesc, nhlfe, | |
3580 | rta, rtnh, &req.r, | |
3581 | &src1); | |
3582 | rtnh = RTNH_NEXT(rtnh); | |
3583 | } | |
3584 | } | |
3585 | ||
3586 | /* Add the multipath */ | |
3587 | if (rta->rta_len > RTA_LENGTH(0)) | |
3588 | addattr_l(&req.n, NL_PKT_BUF_SIZE, RTA_MULTIPATH, | |
3589 | RTA_DATA(rta), RTA_PAYLOAD(rta)); | |
3590 | } | |
3591 | ||
3592 | /* Talk to netlink socket. */ | |
3593 | return netlink_talk_info(netlink_talk_filter, &req.n, | |
3594 | dplane_ctx_get_ns(ctx), 0); | |
3595 | } | |
ddfeb486 | 3596 | #endif /* HAVE_NETLINK */ |