2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
4 * Copyright (c) 2020 Alexander V. Chernikov
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
15 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
16 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
17 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
18 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
19 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
20 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
21 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
22 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
23 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
24 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
28 #include <sys/cdefs.h>
29 __FBSDID("$FreeBSD$");
31 #include "opt_inet6.h"
32 #include "opt_route.h"
34 #include <sys/param.h>
36 #include <sys/systm.h>
37 #include <sys/malloc.h>
39 #include <sys/socket.h>
40 #include <sys/sysctl.h>
41 #include <sys/syslog.h>
42 #include <sys/sysproto.h>
44 #include <sys/domain.h>
45 #include <sys/kernel.h>
47 #include <sys/rmlock.h>
50 #include <net/if_var.h>
51 #include <net/if_dl.h>
52 #include <net/route.h>
53 #include <net/route/route_ctl.h>
54 #include <net/route/route_var.h>
55 #include <net/route/nhop_utils.h>
56 #include <net/route/nhop.h>
57 #include <net/route/nhop_var.h>
59 #include <netinet/in_fib.h>
62 #include <netinet6/in6_fib.h>
67 * RIB helper functions.
71 rib_walk_ext_locked(struct rib_head *rnh, rib_walktree_f_t *wa_f,
72 rib_walk_hook_f_t *hook_f, void *arg)
75 hook_f(rnh, RIB_WALK_HOOK_PRE, arg);
76 rnh->rnh_walktree(&rnh->head, (walktree_f_t *)wa_f, arg);
78 hook_f(rnh, RIB_WALK_HOOK_POST, arg);
82 * Calls @wa_f with @arg for each entry in the table specified by
85 * @ss_t callback is called before and after the tree traversal
86 * while holding table lock.
88 * Table is traversed under read lock unless @wlock is set.
91 rib_walk_ext_internal(struct rib_head *rnh, bool wlock, rib_walktree_f_t *wa_f,
92 rib_walk_hook_f_t *hook_f, void *arg)
100 rib_walk_ext_locked(rnh, wa_f, hook_f, arg);
108 rib_walk_ext(uint32_t fibnum, int family, bool wlock, rib_walktree_f_t *wa_f,
109 rib_walk_hook_f_t *hook_f, void *arg)
111 struct rib_head *rnh;
113 if ((rnh = rt_tables_get_rnh(fibnum, family)) != NULL)
114 rib_walk_ext_internal(rnh, wlock, wa_f, hook_f, arg);
118 * Calls @wa_f with @arg for each entry in the table specified by
121 * Table is traversed under read lock unless @wlock is set.
124 rib_walk(uint32_t fibnum, int family, bool wlock, rib_walktree_f_t *wa_f,
128 rib_walk_ext(fibnum, family, wlock, wa_f, NULL, arg);
132 * Iterates over all existing fibs in system calling
133 * @hook_f function before/after traversing each fib.
134 * Calls @wa_f function for each element in current fib.
135 * If af is not AF_UNSPEC, iterates over fibs in particular
139 rib_foreach_table_walk(int family, bool wlock, rib_walktree_f_t *wa_f,
140 rib_walk_hook_f_t *hook_f, void *arg)
143 for (uint32_t fibnum = 0; fibnum < rt_numfibs; fibnum++) {
144 /* Do we want some specific family? */
145 if (family != AF_UNSPEC) {
146 rib_walk_ext(fibnum, family, wlock, wa_f, hook_f, arg);
150 for (int i = 1; i <= AF_MAX; i++)
151 rib_walk_ext(fibnum, i, wlock, wa_f, hook_f, arg);
156 * Iterates over all existing fibs in system and deletes each element
157 * for which @filter_f function returns non-zero value.
158 * If @family is not AF_UNSPEC, iterates over fibs in particular
162 rib_foreach_table_walk_del(int family, rib_filter_f_t *filter_f, void *arg)
165 for (uint32_t fibnum = 0; fibnum < rt_numfibs; fibnum++) {
166 /* Do we want some specific family? */
167 if (family != AF_UNSPEC) {
168 rib_walk_del(fibnum, family, filter_f, arg, 0);
172 for (int i = 1; i <= AF_MAX; i++)
173 rib_walk_del(fibnum, i, filter_f, arg, 0);
179 * Wrapper for the control plane functions for performing af-agnostic
181 * @fibnum: fib to perform the lookup.
182 * @dst: sockaddr with family and addr filled in. IPv6 addresses needs to be in
184 * @flags: fib(9) flags.
185 * @flowid: flow id for path selection in multipath use case.
187 * Returns nhop_object or NULL.
189 * Requires NET_EPOCH.
193 rib_lookup(uint32_t fibnum, const struct sockaddr *dst, uint32_t flags,
196 struct nhop_object *nh;
200 switch (dst->sa_family) {
204 const struct sockaddr_in *a = (const struct sockaddr_in *)dst;
205 nh = fib4_lookup(fibnum, a->sin_addr, 0, flags, flowid);
212 const struct sockaddr_in6 *a = (const struct sockaddr_in6*)dst;
213 nh = fib6_lookup(fibnum, &a->sin6_addr, a->sin6_scope_id,
225 decompose_change_notification(struct rib_cmd_info *rc, route_notification_t *cb,
228 uint32_t num_old, num_new;
229 uint32_t nh_idx_old, nh_idx_new;
230 struct weightened_nhop *wn_old, *wn_new;
231 struct weightened_nhop tmp = { NULL, 0 };
232 uint32_t idx_old = 0, idx_new = 0;
234 struct rib_cmd_info rc_del = { .rc_cmd = RTM_DELETE, .rc_rt = rc->rc_rt };
235 struct rib_cmd_info rc_add = { .rc_cmd = RTM_ADD, .rc_rt = rc->rc_rt };
237 if (NH_IS_NHGRP(rc->rc_nh_old)) {
238 wn_old = nhgrp_get_nhops((struct nhgrp_object *)rc->rc_nh_old, &num_old);
240 tmp.nh = rc->rc_nh_old;
241 tmp.weight = rc->rc_nh_weight;
245 if (NH_IS_NHGRP(rc->rc_nh_new)) {
246 wn_new = nhgrp_get_nhops((struct nhgrp_object *)rc->rc_nh_new, &num_new);
248 tmp.nh = rc->rc_nh_new;
249 tmp.weight = rc->rc_nh_weight;
254 /* Use the fact that each @wn array is sorted */
256 * Want to convert into set of add and delete operations
257 * [1] -> [1, 2] = A{2}
258 * [2] -> [1, 2] = A{1}
259 * [1, 2, 4]->[1, 3, 4] = A{2}, D{3}
260 * [1, 2, 4]->[1, 4] = D{2}
261 * [1, 2, 4] -> [3, 4] = D{1}, C{2,3} OR C{1,3}, D{2} OR D{1},D{2},A{3}
266 while ((idx_old < num_old) && (idx_new < num_new)) {
267 nh_idx_old = wn_old[idx_old].nh->nh_priv->nh_idx;
268 nh_idx_new = wn_new[idx_new].nh->nh_priv->nh_idx;
270 if (nh_idx_old == nh_idx_new) {
271 if (wn_old[idx_old].weight != wn_new[idx_new].weight) {
272 /* Update weight by providing del/add notifications */
273 rc_del.rc_nh_old = wn_old[idx_old].nh;
274 rc_del.rc_nh_weight = wn_old[idx_old].weight;
277 rc_add.rc_nh_new = wn_new[idx_new].nh;
278 rc_add.rc_nh_weight = wn_new[idx_new].weight;
283 } else if (nh_idx_old < nh_idx_new) {
285 * [1, ~2~, 4], [1, ~3~, 4]
286 * [1, ~2~, 5], [1, ~3~, 4]
287 * [1, ~2~], [1, ~3~, 4]
289 if ((idx_old + 1 >= num_old) ||
290 (wn_old[idx_old + 1].nh->nh_priv->nh_idx > nh_idx_new)) {
291 /* Add new unless the next old item is still <= new */
292 rc_add.rc_nh_new = wn_new[idx_new].nh;
293 rc_add.rc_nh_weight = wn_new[idx_new].weight;
297 /* In any case, delete current old */
298 rc_del.rc_nh_old = wn_old[idx_old].nh;
299 rc_del.rc_nh_weight = wn_old[idx_old].weight;
304 * nh_idx_old > nh_idx_new
306 * [1, ~3~, 4], [1, ~2~, 4]
307 * [1, ~3~, 5], [1, ~2~, 4]
308 * [1, ~3~, 4], [1, ~2~]
310 if ((idx_new + 1 >= num_new) ||
311 (wn_new[idx_new + 1].nh->nh_priv->nh_idx > nh_idx_old)) {
312 /* No next item or next item is > current one */
313 rc_add.rc_nh_new = wn_new[idx_new].nh;
314 rc_add.rc_nh_weight = wn_new[idx_new].weight;
318 /* In any case, delete current old */
319 rc_del.rc_nh_old = wn_old[idx_old].nh;
320 rc_del.rc_nh_weight = wn_old[idx_old].weight;
326 while (idx_old < num_old) {
327 rc_del.rc_nh_old = wn_old[idx_old].nh;
328 rc_del.rc_nh_weight = wn_old[idx_old].weight;
333 while (idx_new < num_new) {
334 rc_add.rc_nh_new = wn_new[idx_new].nh;
335 rc_add.rc_nh_weight = wn_new[idx_new].weight;
342 * Decompose multipath cmd info @rc into a list of add/del/change
343 * single-path operations, calling @cb callback for each operation.
344 * Assumes at least one of the nexthops in @rc is multipath.
347 rib_decompose_notification(struct rib_cmd_info *rc, route_notification_t *cb,
350 struct weightened_nhop *wn;
352 struct rib_cmd_info rc_new;
355 DPRINTF("cb=%p cmd=%d nh_old=%p nh_new=%p",
356 cb, rc->cmd, rc->nh_old, rc->nh_new);
357 switch (rc->rc_cmd) {
359 if (!NH_IS_NHGRP(rc->rc_nh_new))
361 wn = nhgrp_get_nhops((struct nhgrp_object *)rc->rc_nh_new, &num_nhops);
362 for (uint32_t i = 0; i < num_nhops; i++) {
363 rc_new.rc_nh_new = wn[i].nh;
364 rc_new.rc_nh_weight = wn[i].weight;
369 if (!NH_IS_NHGRP(rc->rc_nh_old))
371 wn = nhgrp_get_nhops((struct nhgrp_object *)rc->rc_nh_old, &num_nhops);
372 for (uint32_t i = 0; i < num_nhops; i++) {
373 rc_new.rc_nh_old = wn[i].nh;
374 rc_new.rc_nh_weight = wn[i].weight;
379 if (!NH_IS_NHGRP(rc->rc_nh_old) && !NH_IS_NHGRP(rc->rc_nh_new))
381 decompose_change_notification(rc, cb, cbdata);