1/* $KAME: radix_mpath.c,v 1.17 2004/11/08 10:29:39 itojun Exp $ */
2
3/*
4 * Copyright (C) 2001 WIDE Project.
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
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.
18 *
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
29 * SUCH DAMAGE.
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
33 * PROPERTIES.
34 */
35
36#include <sys/cdefs.h>
| 1/* $KAME: radix_mpath.c,v 1.17 2004/11/08 10:29:39 itojun Exp $ */
2
3/*
4 * Copyright (C) 2001 WIDE Project.
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
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.
18 *
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
29 * SUCH DAMAGE.
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
33 * PROPERTIES.
34 */
35
36#include <sys/cdefs.h>
|
37__FBSDID("$FreeBSD: head/sys/net/radix_mpath.c 178454 2008-04-24 05:04:52Z qingli $");
| 37__FBSDID("$FreeBSD: head/sys/net/radix_mpath.c 178888 2008-05-09 23:03:00Z julian $");
|
38
39#include "opt_inet6.h"
40
41#include <sys/param.h>
42#include <sys/systm.h>
43#include <sys/malloc.h>
44#include <sys/socket.h>
45#include <sys/domain.h>
46#include <sys/syslog.h>
47#include <net/radix.h>
48#include <net/radix_mpath.h>
49#include <net/route.h>
50#include <net/if.h>
51#include <net/if_var.h>
52
53/*
54 * give some jitter to hash, to avoid synchronization between routers
55 */
56static u_int32_t hashjitter;
57
58int
59rn_mpath_capable(struct radix_node_head *rnh)
60{
61
62 return rnh->rnh_multipath;
63}
64
65struct radix_node *
66rn_mpath_next(struct radix_node *rn)
67{
68 struct radix_node *next;
69
70 if (!rn->rn_dupedkey)
71 return NULL;
72 next = rn->rn_dupedkey;
73 if (rn->rn_mask == next->rn_mask)
74 return next;
75 else
76 return NULL;
77}
78
79int
80rn_mpath_count(struct radix_node *rn)
81{
82 int i;
83
84 i = 1;
85 while ((rn = rn_mpath_next(rn)) != NULL)
86 i++;
87 return i;
88}
89
90struct rtentry *
91rt_mpath_matchgate(struct rtentry *rt, struct sockaddr *gate)
92{
93 struct radix_node *rn;
94
95 if (!rn_mpath_next((struct radix_node *)rt))
96 return rt;
97
98 if (!gate)
99 return NULL;
100
101 /* beyond here, we use rn as the master copy */
102 rn = (struct radix_node *)rt;
103 do {
104 rt = (struct rtentry *)rn;
105 /*
106 * we are removing an address alias that has
107 * the same prefix as another address
108 * we need to compare the interface address because
109 * rt_gateway is a special sockadd_dl structure
110 */
111 if (rt->rt_gateway->sa_family == AF_LINK) {
112 if (!memcmp(rt->rt_ifa->ifa_addr, gate, gate->sa_len))
113 break;
114 } else {
115 if (rt->rt_gateway->sa_len == gate->sa_len &&
116 !memcmp(rt->rt_gateway, gate, gate->sa_len))
117 break;
118 }
119 } while ((rn = rn_mpath_next(rn)) != NULL);
120
121 return (struct rtentry *)rn;
122}
123
124/*
125 * go through the chain and unlink "rt" from the list
126 * the caller will free "rt"
127 */
128int
129rt_mpath_deldup(struct rtentry *headrt, struct rtentry *rt)
130{
131 struct radix_node *t, *tt;
132
133 if (!headrt || !rt)
134 return (0);
135 t = (struct radix_node *)headrt;
136 tt = rn_mpath_next(t);
137 while (tt) {
138 if (tt == (struct radix_node *)rt) {
139 t->rn_dupedkey = tt->rn_dupedkey;
140 tt->rn_dupedkey = NULL;
141 tt->rn_flags &= ~RNF_ACTIVE;
142 tt[1].rn_flags &= ~RNF_ACTIVE;
143 return (1);
144 }
145 t = tt;
146 tt = rn_mpath_next((struct radix_node *)t);
147 }
148 return (0);
149}
150
151/*
152 * check if we have the same key/mask/gateway on the table already.
153 */
154int
155rt_mpath_conflict(struct radix_node_head *rnh, struct rtentry *rt,
156 struct sockaddr *netmask)
157{
158 struct radix_node *rn, *rn1;
159 struct rtentry *rt1;
160 char *p, *q, *eq;
161 int same, l, skip;
162
163 rn = (struct radix_node *)rt;
164 rn1 = rnh->rnh_lookup(rt_key(rt), netmask, rnh);
165 if (!rn1 || rn1->rn_flags & RNF_ROOT)
166 return 0;
167
168 /*
169 * unlike other functions we have in this file, we have to check
170 * all key/mask/gateway as rnh_lookup can match less specific entry.
171 */
172 rt1 = (struct rtentry *)rn1;
173
174 /* compare key. */
175 if (rt_key(rt1)->sa_len != rt_key(rt)->sa_len ||
176 bcmp(rt_key(rt1), rt_key(rt), rt_key(rt1)->sa_len))
177 goto different;
178
179 /* key was the same. compare netmask. hairy... */
180 if (rt_mask(rt1) && netmask) {
181 skip = rnh->rnh_treetop->rn_offset;
182 if (rt_mask(rt1)->sa_len > netmask->sa_len) {
183 /*
184 * as rt_mask(rt1) is made optimal by radix.c,
185 * there must be some 1-bits on rt_mask(rt1)
186 * after netmask->sa_len. therefore, in
187 * this case, the entries are different.
188 */
189 if (rt_mask(rt1)->sa_len > skip)
190 goto different;
191 else {
192 /* no bits to compare, i.e. same*/
193 goto maskmatched;
194 }
195 }
196
197 l = rt_mask(rt1)->sa_len;
198 if (skip > l) {
199 /* no bits to compare, i.e. same */
200 goto maskmatched;
201 }
202 p = (char *)rt_mask(rt1);
203 q = (char *)netmask;
204 if (bcmp(p + skip, q + skip, l - skip))
205 goto different;
206 /*
207 * need to go through all the bit, as netmask is not
208 * optimal and can contain trailing 0s
209 */
210 eq = (char *)netmask + netmask->sa_len;
211 q += l;
212 same = 1;
213 while (eq > q)
214 if (*q++) {
215 same = 0;
216 break;
217 }
218 if (!same)
219 goto different;
220 } else if (!rt_mask(rt1) && !netmask)
221 ; /* no mask to compare, i.e. same */
222 else {
223 /* one has mask and the other does not, different */
224 goto different;
225 }
226
227maskmatched:
228
229 /* key/mask were the same. compare gateway for all multipaths */
230 do {
231 rt1 = (struct rtentry *)rn1;
232
233 /* sanity: no use in comparing the same thing */
234 if (rn1 == rn)
235 continue;
236
237 if (rt1->rt_gateway->sa_family == AF_LINK) {
238 if (rt1->rt_ifa->ifa_addr->sa_len != rt->rt_ifa->ifa_addr->sa_len ||
239 bcmp(rt1->rt_ifa->ifa_addr, rt->rt_ifa->ifa_addr,
240 rt1->rt_ifa->ifa_addr->sa_len))
241 continue;
242 } else {
243 if (rt1->rt_gateway->sa_len != rt->rt_gateway->sa_len ||
244 bcmp(rt1->rt_gateway, rt->rt_gateway,
245 rt1->rt_gateway->sa_len))
246 continue;
247 }
248
249 /* all key/mask/gateway are the same. conflicting entry. */
250 return EEXIST;
251 } while ((rn1 = rn_mpath_next(rn1)) != NULL);
252
253different:
254 return 0;
255}
256
257void
| 38
39#include "opt_inet6.h"
40
41#include <sys/param.h>
42#include <sys/systm.h>
43#include <sys/malloc.h>
44#include <sys/socket.h>
45#include <sys/domain.h>
46#include <sys/syslog.h>
47#include <net/radix.h>
48#include <net/radix_mpath.h>
49#include <net/route.h>
50#include <net/if.h>
51#include <net/if_var.h>
52
53/*
54 * give some jitter to hash, to avoid synchronization between routers
55 */
56static u_int32_t hashjitter;
57
58int
59rn_mpath_capable(struct radix_node_head *rnh)
60{
61
62 return rnh->rnh_multipath;
63}
64
65struct radix_node *
66rn_mpath_next(struct radix_node *rn)
67{
68 struct radix_node *next;
69
70 if (!rn->rn_dupedkey)
71 return NULL;
72 next = rn->rn_dupedkey;
73 if (rn->rn_mask == next->rn_mask)
74 return next;
75 else
76 return NULL;
77}
78
79int
80rn_mpath_count(struct radix_node *rn)
81{
82 int i;
83
84 i = 1;
85 while ((rn = rn_mpath_next(rn)) != NULL)
86 i++;
87 return i;
88}
89
90struct rtentry *
91rt_mpath_matchgate(struct rtentry *rt, struct sockaddr *gate)
92{
93 struct radix_node *rn;
94
95 if (!rn_mpath_next((struct radix_node *)rt))
96 return rt;
97
98 if (!gate)
99 return NULL;
100
101 /* beyond here, we use rn as the master copy */
102 rn = (struct radix_node *)rt;
103 do {
104 rt = (struct rtentry *)rn;
105 /*
106 * we are removing an address alias that has
107 * the same prefix as another address
108 * we need to compare the interface address because
109 * rt_gateway is a special sockadd_dl structure
110 */
111 if (rt->rt_gateway->sa_family == AF_LINK) {
112 if (!memcmp(rt->rt_ifa->ifa_addr, gate, gate->sa_len))
113 break;
114 } else {
115 if (rt->rt_gateway->sa_len == gate->sa_len &&
116 !memcmp(rt->rt_gateway, gate, gate->sa_len))
117 break;
118 }
119 } while ((rn = rn_mpath_next(rn)) != NULL);
120
121 return (struct rtentry *)rn;
122}
123
124/*
125 * go through the chain and unlink "rt" from the list
126 * the caller will free "rt"
127 */
128int
129rt_mpath_deldup(struct rtentry *headrt, struct rtentry *rt)
130{
131 struct radix_node *t, *tt;
132
133 if (!headrt || !rt)
134 return (0);
135 t = (struct radix_node *)headrt;
136 tt = rn_mpath_next(t);
137 while (tt) {
138 if (tt == (struct radix_node *)rt) {
139 t->rn_dupedkey = tt->rn_dupedkey;
140 tt->rn_dupedkey = NULL;
141 tt->rn_flags &= ~RNF_ACTIVE;
142 tt[1].rn_flags &= ~RNF_ACTIVE;
143 return (1);
144 }
145 t = tt;
146 tt = rn_mpath_next((struct radix_node *)t);
147 }
148 return (0);
149}
150
151/*
152 * check if we have the same key/mask/gateway on the table already.
153 */
154int
155rt_mpath_conflict(struct radix_node_head *rnh, struct rtentry *rt,
156 struct sockaddr *netmask)
157{
158 struct radix_node *rn, *rn1;
159 struct rtentry *rt1;
160 char *p, *q, *eq;
161 int same, l, skip;
162
163 rn = (struct radix_node *)rt;
164 rn1 = rnh->rnh_lookup(rt_key(rt), netmask, rnh);
165 if (!rn1 || rn1->rn_flags & RNF_ROOT)
166 return 0;
167
168 /*
169 * unlike other functions we have in this file, we have to check
170 * all key/mask/gateway as rnh_lookup can match less specific entry.
171 */
172 rt1 = (struct rtentry *)rn1;
173
174 /* compare key. */
175 if (rt_key(rt1)->sa_len != rt_key(rt)->sa_len ||
176 bcmp(rt_key(rt1), rt_key(rt), rt_key(rt1)->sa_len))
177 goto different;
178
179 /* key was the same. compare netmask. hairy... */
180 if (rt_mask(rt1) && netmask) {
181 skip = rnh->rnh_treetop->rn_offset;
182 if (rt_mask(rt1)->sa_len > netmask->sa_len) {
183 /*
184 * as rt_mask(rt1) is made optimal by radix.c,
185 * there must be some 1-bits on rt_mask(rt1)
186 * after netmask->sa_len. therefore, in
187 * this case, the entries are different.
188 */
189 if (rt_mask(rt1)->sa_len > skip)
190 goto different;
191 else {
192 /* no bits to compare, i.e. same*/
193 goto maskmatched;
194 }
195 }
196
197 l = rt_mask(rt1)->sa_len;
198 if (skip > l) {
199 /* no bits to compare, i.e. same */
200 goto maskmatched;
201 }
202 p = (char *)rt_mask(rt1);
203 q = (char *)netmask;
204 if (bcmp(p + skip, q + skip, l - skip))
205 goto different;
206 /*
207 * need to go through all the bit, as netmask is not
208 * optimal and can contain trailing 0s
209 */
210 eq = (char *)netmask + netmask->sa_len;
211 q += l;
212 same = 1;
213 while (eq > q)
214 if (*q++) {
215 same = 0;
216 break;
217 }
218 if (!same)
219 goto different;
220 } else if (!rt_mask(rt1) && !netmask)
221 ; /* no mask to compare, i.e. same */
222 else {
223 /* one has mask and the other does not, different */
224 goto different;
225 }
226
227maskmatched:
228
229 /* key/mask were the same. compare gateway for all multipaths */
230 do {
231 rt1 = (struct rtentry *)rn1;
232
233 /* sanity: no use in comparing the same thing */
234 if (rn1 == rn)
235 continue;
236
237 if (rt1->rt_gateway->sa_family == AF_LINK) {
238 if (rt1->rt_ifa->ifa_addr->sa_len != rt->rt_ifa->ifa_addr->sa_len ||
239 bcmp(rt1->rt_ifa->ifa_addr, rt->rt_ifa->ifa_addr,
240 rt1->rt_ifa->ifa_addr->sa_len))
241 continue;
242 } else {
243 if (rt1->rt_gateway->sa_len != rt->rt_gateway->sa_len ||
244 bcmp(rt1->rt_gateway, rt->rt_gateway,
245 rt1->rt_gateway->sa_len))
246 continue;
247 }
248
249 /* all key/mask/gateway are the same. conflicting entry. */
250 return EEXIST;
251 } while ((rn1 = rn_mpath_next(rn1)) != NULL);
252
253different:
254 return 0;
255}
256
257void
|
258rtalloc_mpath(struct route *ro, int hash)
| 258rtalloc_mpath_fib(struct route *ro, int hash, u_int fibnum)
|
259{
260 struct radix_node *rn0, *rn;
261 int n;
262
263 /*
264 * XXX we don't attempt to lookup cached route again; what should
265 * be done for sendto(3) case?
266 */
267 if (ro->ro_rt && ro->ro_rt->rt_ifp && (ro->ro_rt->rt_flags & RTF_UP))
268 return; /* XXX */
| 259{
260 struct radix_node *rn0, *rn;
261 int n;
262
263 /*
264 * XXX we don't attempt to lookup cached route again; what should
265 * be done for sendto(3) case?
266 */
267 if (ro->ro_rt && ro->ro_rt->rt_ifp && (ro->ro_rt->rt_flags & RTF_UP))
268 return; /* XXX */
|
269 ro->ro_rt = rtalloc1(&ro->ro_dst, 1, 0UL);
| 269 ro->ro_rt = rtalloc1_fib(&ro->ro_dst, 1, 0UL, fibnum);
|
270
271 /* if the route does not exist or it is not multipath, don't care */
272 if (ro->ro_rt == NULL)
273 return;
274 if (rn_mpath_next((struct radix_node *)ro->ro_rt) == NULL) {
275 RT_UNLOCK(ro->ro_rt);
276 return;
277 }
278
279 /* beyond here, we use rn as the master copy */
280 rn0 = rn = (struct radix_node *)ro->ro_rt;
281 n = rn_mpath_count(rn0);
282
283 /* gw selection by Modulo-N Hash (RFC2991) XXX need improvement? */
284 hash += hashjitter;
285 hash %= n;
286 while (hash-- > 0 && rn) {
287 /* stay within the multipath routes */
288 if (rn->rn_dupedkey && rn->rn_mask != rn->rn_dupedkey->rn_mask)
289 break;
290 rn = rn->rn_dupedkey;
291 }
292
293 /* XXX try filling rt_gwroute and avoid unreachable gw */
294
295 /* if gw selection fails, use the first match (default) */
296 if (!rn) {
297 RT_UNLOCK(ro->ro_rt);
298 return;
299 }
300
301 rtfree(ro->ro_rt);
302 ro->ro_rt = (struct rtentry *)rn;
303 RT_LOCK(ro->ro_rt);
304 RT_ADDREF(ro->ro_rt);
305 RT_UNLOCK(ro->ro_rt);
306}
307
308extern int in6_inithead(void **head, int off);
309extern int in_inithead(void **head, int off);
310
311int
312rn4_mpath_inithead(void **head, int off)
313{
314 struct radix_node_head *rnh;
315
316 hashjitter = arc4random();
317 if (in_inithead(head, off) == 1) {
318 rnh = (struct radix_node_head *)*head;
319 rnh->rnh_multipath = 1;
320 return 1;
321 } else
322 return 0;
323}
324
325#ifdef INET6
326int
327rn6_mpath_inithead(void **head, int off)
328{
329 struct radix_node_head *rnh;
330
331 hashjitter = arc4random();
332 if (in6_inithead(head, off) == 1) {
333 rnh = (struct radix_node_head *)*head;
334 rnh->rnh_multipath = 1;
335 return 1;
336 } else
337 return 0;
338}
339
340#endif
| 270
271 /* if the route does not exist or it is not multipath, don't care */
272 if (ro->ro_rt == NULL)
273 return;
274 if (rn_mpath_next((struct radix_node *)ro->ro_rt) == NULL) {
275 RT_UNLOCK(ro->ro_rt);
276 return;
277 }
278
279 /* beyond here, we use rn as the master copy */
280 rn0 = rn = (struct radix_node *)ro->ro_rt;
281 n = rn_mpath_count(rn0);
282
283 /* gw selection by Modulo-N Hash (RFC2991) XXX need improvement? */
284 hash += hashjitter;
285 hash %= n;
286 while (hash-- > 0 && rn) {
287 /* stay within the multipath routes */
288 if (rn->rn_dupedkey && rn->rn_mask != rn->rn_dupedkey->rn_mask)
289 break;
290 rn = rn->rn_dupedkey;
291 }
292
293 /* XXX try filling rt_gwroute and avoid unreachable gw */
294
295 /* if gw selection fails, use the first match (default) */
296 if (!rn) {
297 RT_UNLOCK(ro->ro_rt);
298 return;
299 }
300
301 rtfree(ro->ro_rt);
302 ro->ro_rt = (struct rtentry *)rn;
303 RT_LOCK(ro->ro_rt);
304 RT_ADDREF(ro->ro_rt);
305 RT_UNLOCK(ro->ro_rt);
306}
307
308extern int in6_inithead(void **head, int off);
309extern int in_inithead(void **head, int off);
310
311int
312rn4_mpath_inithead(void **head, int off)
313{
314 struct radix_node_head *rnh;
315
316 hashjitter = arc4random();
317 if (in_inithead(head, off) == 1) {
318 rnh = (struct radix_node_head *)*head;
319 rnh->rnh_multipath = 1;
320 return 1;
321 } else
322 return 0;
323}
324
325#ifdef INET6
326int
327rn6_mpath_inithead(void **head, int off)
328{
329 struct radix_node_head *rnh;
330
331 hashjitter = arc4random();
332 if (in6_inithead(head, off) == 1) {
333 rnh = (struct radix_node_head *)*head;
334 rnh->rnh_multipath = 1;
335 return 1;
336 } else
337 return 0;
338}
339
340#endif
|