xref: /freebsd/sys/net/route/route_helpers.c (revision d485c77f203fb0f4cdc08dea5ff81631b51d8809)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3  *
4  * Copyright (c) 2020 Alexander V. Chernikov
5  *
6  * Redistribution and use in source and binary forms, with or without
7  * modification, are permitted provided that the following conditions
8  * are met:
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.
14  *
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
25  * SUCH DAMAGE.
26  */
27 
28 #include <sys/cdefs.h>
29 __FBSDID("$FreeBSD$");
30 #include "opt_inet.h"
31 #include "opt_inet6.h"
32 #include "opt_route.h"
33 
34 #include <sys/param.h>
35 #include <sys/jail.h>
36 #include <sys/systm.h>
37 #include <sys/malloc.h>
38 #include <sys/mbuf.h>
39 #include <sys/socket.h>
40 #include <sys/sysctl.h>
41 #include <sys/syslog.h>
42 #include <sys/sysproto.h>
43 #include <sys/proc.h>
44 #include <sys/domain.h>
45 #include <sys/kernel.h>
46 #include <sys/lock.h>
47 #include <sys/rmlock.h>
48 
49 #include <net/if.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>
58 #ifdef INET
59 #include <netinet/in_fib.h>
60 #endif
61 #ifdef INET6
62 #include <netinet6/in6_fib.h>
63 #endif
64 #include <net/vnet.h>
65 
66 /*
67  * RIB helper functions.
68  */
69 
70 void
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)
73 {
74 	if (hook_f != NULL)
75 		hook_f(rnh, RIB_WALK_HOOK_PRE, arg);
76 	rnh->rnh_walktree(&rnh->head, (walktree_f_t *)wa_f, arg);
77 	if (hook_f != NULL)
78 		hook_f(rnh, RIB_WALK_HOOK_POST, arg);
79 }
80 
81 /*
82  * Calls @wa_f with @arg for each entry in the table specified by
83  * @af and @fibnum.
84  *
85  * @ss_t callback is called before and after the tree traversal
86  *  while holding table lock.
87  *
88  * Table is traversed under read lock unless @wlock is set.
89  */
90 void
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)
93 {
94 	RIB_RLOCK_TRACKER;
95 
96 	if (wlock)
97 		RIB_WLOCK(rnh);
98 	else
99 		RIB_RLOCK(rnh);
100 	rib_walk_ext_locked(rnh, wa_f, hook_f, arg);
101 	if (wlock)
102 		RIB_WUNLOCK(rnh);
103 	else
104 		RIB_RUNLOCK(rnh);
105 }
106 
107 void
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)
110 {
111 	struct rib_head *rnh;
112 
113 	if ((rnh = rt_tables_get_rnh(fibnum, family)) != NULL)
114 		rib_walk_ext_internal(rnh, wlock, wa_f, hook_f, arg);
115 }
116 
117 /*
118  * Calls @wa_f with @arg for each entry in the table specified by
119  * @af and @fibnum.
120  *
121  * Table is traversed under read lock unless @wlock is set.
122  */
123 void
124 rib_walk(uint32_t fibnum, int family, bool wlock, rib_walktree_f_t *wa_f,
125     void *arg)
126 {
127 
128 	rib_walk_ext(fibnum, family, wlock, wa_f, NULL, arg);
129 }
130 
131 /*
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
136  * address family.
137  */
138 void
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)
141 {
142 
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);
147 			continue;
148 		}
149 
150 		for (int i = 1; i <= AF_MAX; i++)
151 			rib_walk_ext(fibnum, i, wlock, wa_f, hook_f, arg);
152 	}
153 }
154 
155 /*
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
159  * address family.
160  */
161 void
162 rib_foreach_table_walk_del(int family, rib_filter_f_t *filter_f, void *arg)
163 {
164 
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);
169 			continue;
170 		}
171 
172 		for (int i = 1; i <= AF_MAX; i++)
173 			rib_walk_del(fibnum, i, filter_f, arg, 0);
174 	}
175 }
176 
177 
178 /*
179  * Wrapper for the control plane functions for performing af-agnostic
180  *  lookups.
181  * @fibnum: fib to perform the lookup.
182  * @dst: sockaddr with family and addr filled in. IPv6 addresses needs to be in
183  *  deembedded from.
184  * @flags: fib(9) flags.
185  * @flowid: flow id for path selection in multipath use case.
186  *
187  * Returns nhop_object or NULL.
188  *
189  * Requires NET_EPOCH.
190  *
191  */
192 struct nhop_object *
193 rib_lookup(uint32_t fibnum, const struct sockaddr *dst, uint32_t flags,
194     uint32_t flowid)
195 {
196 	struct nhop_object *nh;
197 
198 	nh = NULL;
199 
200 	switch (dst->sa_family) {
201 #ifdef INET
202 	case AF_INET:
203 	{
204 		const struct sockaddr_in *a = (const struct sockaddr_in *)dst;
205 		nh = fib4_lookup(fibnum, a->sin_addr, 0, flags, flowid);
206 		break;
207 	}
208 #endif
209 #ifdef INET6
210 	case AF_INET6:
211 	{
212 		const struct sockaddr_in6 *a = (const struct sockaddr_in6*)dst;
213 		nh = fib6_lookup(fibnum, &a->sin6_addr, a->sin6_scope_id,
214 		    flags, flowid);
215 		break;
216 	}
217 #endif
218 	}
219 
220 	return (nh);
221 }
222 
223 #ifdef ROUTE_MPATH
224 static void
225 decompose_change_notification(struct rib_cmd_info *rc, route_notification_t *cb,
226     void *cbdata)
227 {
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;
233 
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 };
236 
237 	if (NH_IS_NHGRP(rc->rc_nh_old)) {
238 		wn_old = nhgrp_get_nhops((struct nhgrp_object *)rc->rc_nh_old, &num_old);
239 	} else {
240 		tmp.nh = rc->rc_nh_old;
241 		tmp.weight = rc->rc_nh_weight;
242 		wn_old = &tmp;
243 		num_old = 1;
244 	}
245 	if (NH_IS_NHGRP(rc->rc_nh_new)) {
246 		wn_new = nhgrp_get_nhops((struct nhgrp_object *)rc->rc_nh_new, &num_new);
247 	} else {
248 		tmp.nh = rc->rc_nh_new;
249 		tmp.weight = rc->rc_nh_weight;
250 		wn_new = &tmp;
251 		num_new = 1;
252 	}
253 
254 	/* Use the fact that each @wn array is sorted */
255 	/*
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}
262 	 * [1, 2] -> [3, 4] =
263 	 *
264 	 */
265 	idx_old = 0;
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;
269 
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;
275 				cb(&rc_del, cbdata);
276 
277 				rc_add.rc_nh_new = wn_new[idx_new].nh;
278 				rc_add.rc_nh_weight = wn_new[idx_new].weight;
279 				cb(&rc_add, cbdata);
280 			}
281 			idx_old++;
282 			idx_new++;
283 		} else if (nh_idx_old < nh_idx_new) {
284 			/*
285 			 * [1, ~2~, 4], [1, ~3~, 4]
286 			 * [1, ~2~, 5], [1, ~3~, 4]
287 			 * [1, ~2~], [1, ~3~, 4]
288 			 */
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;
294 				cb(&rc_add, cbdata);
295 				idx_new++;
296 			}
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;
300 			cb(&rc_del, cbdata);
301 			idx_old++;
302 		} else {
303 			/*
304 			 * nh_idx_old > nh_idx_new
305 			 *
306 			 * [1, ~3~, 4], [1, ~2~, 4]
307 			 * [1, ~3~, 5], [1, ~2~, 4]
308 			 * [1, ~3~, 4], [1, ~2~]
309 			 */
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;
315 				cb(&rc_add, cbdata);
316 				idx_new++;
317 			}
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;
321 			cb(&rc_del, cbdata);
322 			idx_old++;
323 		}
324 	}
325 
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;
329 		cb(&rc_del, cbdata);
330 		idx_old++;
331 	}
332 
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;
336 		cb(&rc_add, cbdata);
337 		idx_new++;
338 	}
339 }
340 
341 /*
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.
345  */
346 void
347 rib_decompose_notification(struct rib_cmd_info *rc, route_notification_t *cb,
348     void *cbdata)
349 {
350 	struct weightened_nhop *wn;
351 	uint32_t num_nhops;
352 	struct rib_cmd_info rc_new;
353 
354 	rc_new = *rc;
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) {
358 	case RTM_ADD:
359 		if (!NH_IS_NHGRP(rc->rc_nh_new))
360 			return;
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;
365 			cb(&rc_new, cbdata);
366 		}
367 		break;
368 	case RTM_DELETE:
369 		if (!NH_IS_NHGRP(rc->rc_nh_old))
370 			return;
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;
375 			cb(&rc_new, cbdata);
376 		}
377 		break;
378 	case RTM_CHANGE:
379 		if (!NH_IS_NHGRP(rc->rc_nh_old) && !NH_IS_NHGRP(rc->rc_nh_new))
380 			return;
381 		decompose_change_notification(rc, cb, cbdata);
382 		break;
383 	}
384 }
385 #endif
386