1 /* $KAME: radix_mpath.c,v 1.17 2004/11/08 10:29:39 itojun Exp $ */
4 * Copyright (C) 2001 WIDE Project.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the project nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * THE AUTHORS DO NOT GUARANTEE THAT THIS SOFTWARE DOES NOT INFRINGE
31 * ANY OTHERS' INTELLECTUAL PROPERTIES. IN NO EVENT SHALL THE AUTHORS
32 * BE LIABLE FOR ANY INFRINGEMENT OF ANY OTHERS' INTELLECTUAL
36 #include <sys/cdefs.h>
37 __FBSDID("$FreeBSD$");
40 #include "opt_inet6.h"
42 #include <sys/param.h>
43 #include <sys/systm.h>
44 #include <sys/malloc.h>
45 #include <sys/socket.h>
46 #include <sys/domain.h>
47 #include <sys/syslog.h>
48 #include <net/radix.h>
49 #include <net/radix_mpath.h>
50 #include <net/route.h>
51 #include <net/route_var.h>
53 #include <net/if_var.h>
56 * give some jitter to hash, to avoid synchronization between routers
58 static uint32_t hashjitter;
61 rt_mpath_capable(struct rib_head *rnh)
64 return rnh->rnh_multipath;
68 rn_mpath_capable(struct radix_head *rh)
71 return (rt_mpath_capable((struct rib_head *)rh));
75 rn_mpath_next(struct radix_node *rn)
77 struct radix_node *next;
81 next = rn->rn_dupedkey;
82 if (rn->rn_mask == next->rn_mask)
89 rn_mpath_count(struct radix_node *rn)
95 rt = (struct rtentry *)rn;
97 rn = rn_mpath_next(rn);
103 rt_mpath_matchgate(struct rtentry *rt, struct sockaddr *gate)
105 struct radix_node *rn;
107 if (!gate || !rt->rt_gateway)
110 /* beyond here, we use rn as the master copy */
111 rn = (struct radix_node *)rt;
113 rt = (struct rtentry *)rn;
115 * we are removing an address alias that has
116 * the same prefix as another address
117 * we need to compare the interface address because
118 * rt_gateway is a special sockadd_dl structure
120 if (rt->rt_gateway->sa_family == AF_LINK) {
121 if (!memcmp(rt->rt_ifa->ifa_addr, gate, gate->sa_len))
126 * Check for other options:
127 * 1) Routes with 'real' IPv4/IPv6 gateway
128 * 2) Loopback host routes (another AF_LINK/sockadd_dl check)
130 if (rt->rt_gateway->sa_len == gate->sa_len &&
131 !memcmp(rt->rt_gateway, gate, gate->sa_len))
133 } while ((rn = rn_mpath_next(rn)) != NULL);
135 return (struct rtentry *)rn;
139 * go through the chain and unlink "rt" from the list
140 * the caller will free "rt"
143 rt_mpath_deldup(struct rtentry *headrt, struct rtentry *rt)
145 struct radix_node *t, *tt;
149 t = (struct radix_node *)headrt;
150 tt = rn_mpath_next(t);
152 if (tt == (struct radix_node *)rt) {
153 t->rn_dupedkey = tt->rn_dupedkey;
154 tt->rn_dupedkey = NULL;
155 tt->rn_flags &= ~RNF_ACTIVE;
156 tt[1].rn_flags &= ~RNF_ACTIVE;
160 tt = rn_mpath_next((struct radix_node *)t);
166 * check if we have the same key/mask/gateway on the table already.
167 * Assume @rt rt_key host bits are cleared according to @netmask
170 rt_mpath_conflict(struct rib_head *rnh, struct rtentry *rt,
171 struct sockaddr *netmask)
173 struct radix_node *rn, *rn1;
176 rn = (struct radix_node *)rt;
177 rn1 = rnh->rnh_lookup(rt_key(rt), netmask, &rnh->head);
178 if (!rn1 || rn1->rn_flags & RNF_ROOT)
181 /* key/mask are the same. compare gateway for all multipaths */
183 rt1 = (struct rtentry *)rn1;
185 /* sanity: no use in comparing the same thing */
189 if (rt1->rt_gateway->sa_family == AF_LINK) {
190 if (rt1->rt_ifa->ifa_addr->sa_len != rt->rt_ifa->ifa_addr->sa_len ||
191 bcmp(rt1->rt_ifa->ifa_addr, rt->rt_ifa->ifa_addr,
192 rt1->rt_ifa->ifa_addr->sa_len))
195 if (rt1->rt_gateway->sa_len != rt->rt_gateway->sa_len ||
196 bcmp(rt1->rt_gateway, rt->rt_gateway,
197 rt1->rt_gateway->sa_len))
201 /* all key/mask/gateway are the same. conflicting entry. */
203 } while ((rn1 = rn_mpath_next(rn1)) != NULL);
208 static struct rtentry *
209 rt_mpath_selectrte(struct rtentry *rte, uint32_t hash)
211 struct radix_node *rn0, *rn;
212 uint32_t total_weight;
216 /* beyond here, we use rn as the master copy */
217 rn0 = rn = (struct radix_node *)rte;
220 /* gw selection by Modulo-N Hash (RFC2991) XXX need improvement? */
221 total_weight = rn_mpath_count(rn0);
223 hash %= total_weight;
224 for (weight = abs((int32_t)hash);
225 rt != NULL && weight >= rt->rt_weight;
226 weight -= (rt == NULL) ? 0 : rt->rt_weight) {
228 /* stay within the multipath routes */
229 if (rn->rn_dupedkey && rn->rn_mask != rn->rn_dupedkey->rn_mask)
231 rn = rn->rn_dupedkey;
232 rt = (struct rtentry *)rn;
239 rt_mpath_select(struct rtentry *rte, uint32_t hash)
241 if (rn_mpath_next((struct radix_node *)rte) == NULL)
244 return (rt_mpath_selectrte(rte, hash));
248 rtalloc_mpath_fib(struct route *ro, uint32_t hash, u_int fibnum)
253 * XXX we don't attempt to lookup cached route again; what should
254 * be done for sendto(3) case?
256 if (ro->ro_rt && ro->ro_rt->rt_ifp && (ro->ro_rt->rt_flags & RTF_UP)
257 && RT_LINK_IS_UP(ro->ro_rt->rt_ifp))
259 ro->ro_rt = rtalloc1_fib(&ro->ro_dst, 1, 0, fibnum);
261 /* if the route does not exist or it is not multipath, don't care */
262 if (ro->ro_rt == NULL)
264 if (rn_mpath_next((struct radix_node *)ro->ro_rt) == NULL) {
265 RT_UNLOCK(ro->ro_rt);
269 rt = rt_mpath_selectrte(ro->ro_rt, hash);
270 /* XXX try filling rt_gwroute and avoid unreachable gw */
272 /* gw selection has failed - there must be only zero weight routes */
274 RT_UNLOCK(ro->ro_rt);
278 if (ro->ro_rt != rt) {
279 RTFREE_LOCKED(ro->ro_rt);
282 RT_ADDREF(ro->ro_rt);
285 RT_UNLOCK(ro->ro_rt);
288 extern int in6_inithead(void **head, int off);
289 extern int in_inithead(void **head, int off);
293 rn4_mpath_inithead(void **head, int off)
295 struct rib_head *rnh;
297 hashjitter = arc4random();
298 if (in_inithead(head, off) == 1) {
299 rnh = (struct rib_head *)*head;
300 rnh->rnh_multipath = 1;
309 rn6_mpath_inithead(void **head, int off)
311 struct rib_head *rnh;
313 hashjitter = arc4random();
314 if (in6_inithead(head, off) == 1) {
315 rnh = (struct rib_head *)*head;
316 rnh->rnh_multipath = 1;