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>
52 #include <net/if_var.h>
55 * give some jitter to hash, to avoid synchronization between routers
57 static uint32_t hashjitter;
60 rn_mpath_capable(struct radix_node_head *rnh)
63 return rnh->rnh_multipath;
67 rn_mpath_next(struct radix_node *rn)
69 struct radix_node *next;
73 next = rn->rn_dupedkey;
74 if (rn->rn_mask == next->rn_mask)
81 rn_mpath_count(struct radix_node *rn)
87 rt = (struct rtentry *)rn;
88 i += rt->rt_rmx.rmx_weight;
89 rn = rn_mpath_next(rn);
95 rt_mpath_matchgate(struct rtentry *rt, struct sockaddr *gate)
97 struct radix_node *rn;
99 if (!rn_mpath_next((struct radix_node *)rt))
105 /* beyond here, we use rn as the master copy */
106 rn = (struct radix_node *)rt;
108 rt = (struct rtentry *)rn;
110 * we are removing an address alias that has
111 * the same prefix as another address
112 * we need to compare the interface address because
113 * rt_gateway is a special sockadd_dl structure
115 if (rt->rt_gateway->sa_family == AF_LINK) {
116 if (!memcmp(rt->rt_ifa->ifa_addr, gate, gate->sa_len))
119 if (rt->rt_gateway->sa_len == gate->sa_len &&
120 !memcmp(rt->rt_gateway, gate, gate->sa_len))
123 } while ((rn = rn_mpath_next(rn)) != NULL);
125 return (struct rtentry *)rn;
129 * go through the chain and unlink "rt" from the list
130 * the caller will free "rt"
133 rt_mpath_deldup(struct rtentry *headrt, struct rtentry *rt)
135 struct radix_node *t, *tt;
139 t = (struct radix_node *)headrt;
140 tt = rn_mpath_next(t);
142 if (tt == (struct radix_node *)rt) {
143 t->rn_dupedkey = tt->rn_dupedkey;
144 tt->rn_dupedkey = NULL;
145 tt->rn_flags &= ~RNF_ACTIVE;
146 tt[1].rn_flags &= ~RNF_ACTIVE;
150 tt = rn_mpath_next((struct radix_node *)t);
156 * check if we have the same key/mask/gateway on the table already.
159 rt_mpath_conflict(struct radix_node_head *rnh, struct rtentry *rt,
160 struct sockaddr *netmask)
162 struct radix_node *rn, *rn1;
167 rn = (struct radix_node *)rt;
168 rn1 = rnh->rnh_lookup(rt_key(rt), netmask, rnh);
169 if (!rn1 || rn1->rn_flags & RNF_ROOT)
173 * unlike other functions we have in this file, we have to check
174 * all key/mask/gateway as rnh_lookup can match less specific entry.
176 rt1 = (struct rtentry *)rn1;
179 if (rt_key(rt1)->sa_len != rt_key(rt)->sa_len ||
180 bcmp(rt_key(rt1), rt_key(rt), rt_key(rt1)->sa_len))
183 /* key was the same. compare netmask. hairy... */
184 if (rt_mask(rt1) && netmask) {
185 skip = rnh->rnh_treetop->rn_offset;
186 if (rt_mask(rt1)->sa_len > netmask->sa_len) {
188 * as rt_mask(rt1) is made optimal by radix.c,
189 * there must be some 1-bits on rt_mask(rt1)
190 * after netmask->sa_len. therefore, in
191 * this case, the entries are different.
193 if (rt_mask(rt1)->sa_len > skip)
196 /* no bits to compare, i.e. same*/
201 l = rt_mask(rt1)->sa_len;
203 /* no bits to compare, i.e. same */
206 p = (char *)rt_mask(rt1);
208 if (bcmp(p + skip, q + skip, l - skip))
211 * need to go through all the bit, as netmask is not
212 * optimal and can contain trailing 0s
214 eq = (char *)netmask + netmask->sa_len;
224 } else if (!rt_mask(rt1) && !netmask)
225 ; /* no mask to compare, i.e. same */
227 /* one has mask and the other does not, different */
233 /* key/mask were the same. compare gateway for all multipaths */
235 rt1 = (struct rtentry *)rn1;
237 /* sanity: no use in comparing the same thing */
241 if (rt1->rt_gateway->sa_family == AF_LINK) {
242 if (rt1->rt_ifa->ifa_addr->sa_len != rt->rt_ifa->ifa_addr->sa_len ||
243 bcmp(rt1->rt_ifa->ifa_addr, rt->rt_ifa->ifa_addr,
244 rt1->rt_ifa->ifa_addr->sa_len))
247 if (rt1->rt_gateway->sa_len != rt->rt_gateway->sa_len ||
248 bcmp(rt1->rt_gateway, rt->rt_gateway,
249 rt1->rt_gateway->sa_len))
253 /* all key/mask/gateway are the same. conflicting entry. */
255 } while ((rn1 = rn_mpath_next(rn1)) != NULL);
262 rtalloc_mpath_fib(struct route *ro, uint32_t hash, u_int fibnum)
264 struct radix_node *rn0, *rn;
270 * XXX we don't attempt to lookup cached route again; what should
271 * be done for sendto(3) case?
273 if (ro->ro_rt && ro->ro_rt->rt_ifp && (ro->ro_rt->rt_flags & RTF_UP)
274 && RT_LINK_IS_UP(ro->ro_rt->rt_ifp))
276 ro->ro_rt = rtalloc1_fib(&ro->ro_dst, 1, 0, fibnum);
278 /* if the route does not exist or it is not multipath, don't care */
279 if (ro->ro_rt == NULL)
281 if (rn_mpath_next((struct radix_node *)ro->ro_rt) == NULL) {
282 RT_UNLOCK(ro->ro_rt);
286 /* beyond here, we use rn as the master copy */
287 rn0 = rn = (struct radix_node *)ro->ro_rt;
288 n = rn_mpath_count(rn0);
290 /* gw selection by Modulo-N Hash (RFC2991) XXX need improvement? */
293 for (weight = abs((int32_t)hash), rt = ro->ro_rt;
294 weight >= rt->rt_rmx.rmx_weight && rn;
295 weight -= rt->rt_rmx.rmx_weight) {
297 /* stay within the multipath routes */
298 if (rn->rn_dupedkey && rn->rn_mask != rn->rn_dupedkey->rn_mask)
300 rn = rn->rn_dupedkey;
301 rt = (struct rtentry *)rn;
303 /* XXX try filling rt_gwroute and avoid unreachable gw */
305 /* gw selection has failed - there must be only zero weight routes */
307 RT_UNLOCK(ro->ro_rt);
311 if (ro->ro_rt != rt) {
312 RTFREE_LOCKED(ro->ro_rt);
313 ro->ro_rt = (struct rtentry *)rn;
315 RT_ADDREF(ro->ro_rt);
318 RT_UNLOCK(ro->ro_rt);
321 extern int in6_inithead(void **head, int off);
322 extern int in_inithead(void **head, int off);
326 rn4_mpath_inithead(void **head, int off)
328 struct radix_node_head *rnh;
330 hashjitter = arc4random();
331 if (in_inithead(head, off) == 1) {
332 rnh = (struct radix_node_head *)*head;
333 rnh->rnh_multipath = 1;
342 rn6_mpath_inithead(void **head, int off)
344 struct radix_node_head *rnh;
346 hashjitter = arc4random();
347 if (in6_inithead(head, off) == 1) {
348 rnh = (struct radix_node_head *)*head;
349 rnh->rnh_multipath = 1;