in6_rmx.c revision 283291
1180394Sgnn/*-
2180394Sgnn * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
3180394Sgnn * All rights reserved.
4180394Sgnn *
5180394Sgnn * Redistribution and use in source and binary forms, with or without
6180394Sgnn * modification, are permitted provided that the following conditions
7180394Sgnn * are met:
8180394Sgnn * 1. Redistributions of source code must retain the above copyright
9180394Sgnn *    notice, this list of conditions and the following disclaimer.
10180394Sgnn * 2. Redistributions in binary form must reproduce the above copyright
11180394Sgnn *    notice, this list of conditions and the following disclaimer in the
12180394Sgnn *    documentation and/or other materials provided with the distribution.
13180394Sgnn * 3. Neither the name of the project nor the names of its contributors
14180394Sgnn *    may be used to endorse or promote products derived from this software
15180394Sgnn *    without specific prior written permission.
16180394Sgnn *
17180394Sgnn * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18180394Sgnn * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19180394Sgnn * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20180394Sgnn * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21180394Sgnn * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22180394Sgnn * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23180394Sgnn * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24180394Sgnn * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25180394Sgnn * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26180394Sgnn * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27180394Sgnn * SUCH DAMAGE.
28180394Sgnn *
29180394Sgnn *	$KAME: in6_rmx.c,v 1.11 2001/07/26 06:53:16 jinmei Exp $
30180394Sgnn */
31180394Sgnn
32180394Sgnn/*-
33180394Sgnn * Copyright 1994, 1995 Massachusetts Institute of Technology
34180394Sgnn *
35180394Sgnn * Permission to use, copy, modify, and distribute this software and
36180394Sgnn * its documentation for any purpose and without fee is hereby
37180394Sgnn * granted, provided that both the above copyright notice and this
38180394Sgnn * permission notice appear in all copies, that both the above
39180394Sgnn * copyright notice and this permission notice appear in all
40180394Sgnn * supporting documentation, and that the name of M.I.T. not be used
41180394Sgnn * in advertising or publicity pertaining to distribution of the
42180394Sgnn * software without specific, written prior permission.  M.I.T. makes
43180394Sgnn * no representations about the suitability of this software for any
44180394Sgnn * purpose.  It is provided "as is" without express or implied
45180394Sgnn * warranty.
46180394Sgnn *
47180394Sgnn * THIS SOFTWARE IS PROVIDED BY M.I.T. ``AS IS''.  M.I.T. DISCLAIMS
48180394Sgnn * ALL EXPRESS OR IMPLIED WARRANTIES WITH REGARD TO THIS SOFTWARE,
49180394Sgnn * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
50180394Sgnn * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT
51180394Sgnn * SHALL M.I.T. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
52180394Sgnn * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
53180394Sgnn * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
54180394Sgnn * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
55180394Sgnn * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
56180394Sgnn * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
57180394Sgnn * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
58180394Sgnn * SUCH DAMAGE.
59180394Sgnn *
60180394Sgnn */
61180394Sgnn
62180394Sgnn#include <sys/cdefs.h>
63180394Sgnn__FBSDID("$FreeBSD: head/sys/netinet6/in6_rmx.c 283291 2015-05-22 17:05:21Z jkim $");
64180394Sgnn
65180394Sgnn#include <sys/param.h>
66180394Sgnn#include <sys/systm.h>
67180394Sgnn#include <sys/kernel.h>
68180394Sgnn#include <sys/lock.h>
69180394Sgnn#include <sys/queue.h>
70180394Sgnn#include <sys/socket.h>
71180394Sgnn#include <sys/socketvar.h>
72180394Sgnn#include <sys/mbuf.h>
73180394Sgnn#include <sys/rwlock.h>
74180394Sgnn#include <sys/syslog.h>
75180394Sgnn#include <sys/callout.h>
76180394Sgnn
77180394Sgnn#include <net/if.h>
78180394Sgnn#include <net/if_var.h>
79180394Sgnn#include <net/route.h>
80180394Sgnn
81180394Sgnn#include <netinet/in.h>
82180394Sgnn#include <netinet/ip_var.h>
83180394Sgnn#include <netinet/in_var.h>
84180394Sgnn
85180394Sgnn#include <netinet/ip6.h>
86180394Sgnn#include <netinet6/ip6_var.h>
87180394Sgnn
88180394Sgnn#include <netinet/icmp6.h>
89180394Sgnn#include <netinet6/nd6.h>
90180394Sgnn
91261426Sgnn#include <netinet/tcp.h>
92180394Sgnn#include <netinet/tcp_seq.h>
93180394Sgnn#include <netinet/tcp_timer.h>
94180394Sgnn#include <netinet/tcp_var.h>
95180394Sgnn
96180394Sgnnextern int	in6_inithead(void **head, int off);
97180394Sgnn#ifdef VIMAGE
98180394Sgnnextern int	in6_detachhead(void **head, int off);
99180394Sgnn#endif
100180394Sgnn
101180394Sgnn/*
102180394Sgnn * Do what we need to do when inserting a route.
103180394Sgnn */
104180394Sgnnstatic struct radix_node *
105180394Sgnnin6_addroute(void *v_arg, void *n_arg, struct radix_node_head *head,
106180394Sgnn    struct radix_node *treenodes)
107180394Sgnn{
108180394Sgnn	struct rtentry *rt = (struct rtentry *)treenodes;
109180394Sgnn	struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)rt_key(rt);
110180394Sgnn	struct radix_node *ret;
111180394Sgnn
112180394Sgnn	RADIX_NODE_HEAD_WLOCK_ASSERT(head);
113180394Sgnn	if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
114180394Sgnn		rt->rt_flags |= RTF_MULTICAST;
115180394Sgnn
116180394Sgnn	/*
117180394Sgnn	 * A little bit of help for both IPv6 output and input:
118180394Sgnn	 *   For local addresses, we make sure that RTF_LOCAL is set,
119180394Sgnn	 *   with the thought that this might one day be used to speed up
120180394Sgnn	 *   ip_input().
121180394Sgnn	 *
122180394Sgnn	 * We also mark routes to multicast addresses as such, because
123180394Sgnn	 * it's easy to do and might be useful (but this is much more
124180394Sgnn	 * dubious since it's so easy to inspect the address).  (This
125180394Sgnn	 * is done above.)
126180394Sgnn	 *
127180394Sgnn	 * XXX
128180394Sgnn	 * should elaborate the code.
129180394Sgnn	 */
130180394Sgnn	if (rt->rt_flags & RTF_HOST) {
131180394Sgnn		if (IN6_ARE_ADDR_EQUAL(&satosin6(rt->rt_ifa->ifa_addr)
132180394Sgnn					->sin6_addr,
133180394Sgnn				       &sin6->sin6_addr)) {
134180394Sgnn			rt->rt_flags |= RTF_LOCAL;
135180394Sgnn		}
136180394Sgnn	}
137180394Sgnn
138180394Sgnn	if (rt->rt_ifp != NULL) {
139180394Sgnn
140180394Sgnn		/*
141261426Sgnn		 * Check route MTU:
142261426Sgnn		 * inherit interface MTU if not set or
143261426Sgnn		 * check if MTU is too large.
144261426Sgnn		 */
145180394Sgnn		if (rt->rt_mtu == 0) {
146180394Sgnn			rt->rt_mtu = IN6_LINKMTU(rt->rt_ifp);
147180394Sgnn		} else if (rt->rt_mtu > IN6_LINKMTU(rt->rt_ifp))
148180394Sgnn			rt->rt_mtu = IN6_LINKMTU(rt->rt_ifp);
149180394Sgnn	}
150180394Sgnn
151180394Sgnn	ret = rn_addroute(v_arg, n_arg, head, treenodes);
152180394Sgnn	if (ret == NULL) {
153180394Sgnn		struct rtentry *rt2;
154180394Sgnn		/*
155180394Sgnn		 * We are trying to add a net route, but can't.
156180394Sgnn		 * The following case should be allowed, so we'll make a
157180394Sgnn		 * special check for this:
158180394Sgnn		 *	Two IPv6 addresses with the same prefix is assigned
159180394Sgnn		 *	to a single interrface.
160180394Sgnn		 *	# ifconfig if0 inet6 3ffe:0501::1 prefix 64 alias (*1)
161180394Sgnn		 *	# ifconfig if0 inet6 3ffe:0501::2 prefix 64 alias (*2)
162180394Sgnn		 *	In this case, (*1) and (*2) want to add the same
163180394Sgnn		 *	net route entry, 3ffe:0501:: -> if0.
164180394Sgnn		 *	This case should not raise an error.
165261426Sgnn		 */
166180394Sgnn		rt2 = in6_rtalloc1((struct sockaddr *)sin6, 0, RTF_RNH_LOCKED,
167261426Sgnn		    rt->rt_fibnum);
168180394Sgnn		if (rt2) {
169180394Sgnn			if (((rt2->rt_flags & (RTF_HOST|RTF_GATEWAY)) == 0)
170261426Sgnn			 && rt2->rt_gateway
171180394Sgnn			 && rt2->rt_gateway->sa_family == AF_LINK
172180394Sgnn			 && rt2->rt_ifp == rt->rt_ifp) {
173180394Sgnn				ret = rt2->rt_nodes;
174180394Sgnn			}
175180394Sgnn			RTFREE_LOCKED(rt2);
176180394Sgnn		}
177180394Sgnn	}
178180394Sgnn	return (ret);
179180394Sgnn}
180180394Sgnn
181180394Sgnn/*
182180394Sgnn * Age old PMTUs.
183180394Sgnn */
184261426Sgnnstruct mtuex_arg {
185261426Sgnn	struct radix_node_head *rnh;
186261426Sgnn	time_t nextstop;
187180394Sgnn};
188180394Sgnnstatic VNET_DEFINE(struct callout, rtq_mtutimer);
189180394Sgnn#define	V_rtq_mtutimer			VNET(rtq_mtutimer)
190180394Sgnn
191180394Sgnnstatic int
192180394Sgnnin6_mtuexpire(struct radix_node *rn, void *rock)
193261426Sgnn{
194180394Sgnn	struct rtentry *rt = (struct rtentry *)rn;
195180394Sgnn	struct mtuex_arg *ap = rock;
196
197	/* sanity */
198	if (!rt)
199		panic("rt == NULL in in6_mtuexpire");
200
201	if (rt->rt_expire && !(rt->rt_flags & RTF_PROBEMTU)) {
202		if (rt->rt_expire <= time_uptime) {
203			rt->rt_flags |= RTF_PROBEMTU;
204		} else {
205			ap->nextstop = lmin(ap->nextstop, rt->rt_expire);
206		}
207	}
208
209	return 0;
210}
211
212#define	MTUTIMO_DEFAULT	(60*1)
213
214static void
215in6_mtutimo_one(struct radix_node_head *rnh)
216{
217	struct mtuex_arg arg;
218
219	arg.rnh = rnh;
220	arg.nextstop = time_uptime + MTUTIMO_DEFAULT;
221	RADIX_NODE_HEAD_LOCK(rnh);
222	rnh->rnh_walktree(rnh, in6_mtuexpire, &arg);
223	RADIX_NODE_HEAD_UNLOCK(rnh);
224}
225
226static void
227in6_mtutimo(void *rock)
228{
229	CURVNET_SET_QUIET((struct vnet *) rock);
230	struct radix_node_head *rnh;
231	struct timeval atv;
232	u_int fibnum;
233
234	for (fibnum = 0; fibnum < rt_numfibs; fibnum++) {
235		rnh = rt_tables_get_rnh(fibnum, AF_INET6);
236		if (rnh != NULL)
237			in6_mtutimo_one(rnh);
238	}
239
240	atv.tv_sec = MTUTIMO_DEFAULT;
241	atv.tv_usec = 0;
242	callout_reset(&V_rtq_mtutimer, tvtohz(&atv), in6_mtutimo, rock);
243	CURVNET_RESTORE();
244}
245
246/*
247 * Initialize our routing tree.
248 */
249static VNET_DEFINE(int, _in6_rt_was_here);
250#define	V__in6_rt_was_here	VNET(_in6_rt_was_here)
251
252int
253in6_inithead(void **head, int off)
254{
255	struct radix_node_head *rnh;
256
257	if (!rn_inithead(head, offsetof(struct sockaddr_in6, sin6_addr) << 3))
258		return (0);
259
260	rnh = *head;
261	RADIX_NODE_HEAD_LOCK_INIT(rnh);
262
263	rnh->rnh_addaddr = in6_addroute;
264
265	if (V__in6_rt_was_here == 0) {
266		callout_init(&V_rtq_mtutimer, 1);
267		in6_mtutimo(curvnet);	/* kick off timeout first time */
268		V__in6_rt_was_here = 1;
269	}
270
271	return (1);
272}
273
274#ifdef VIMAGE
275int
276in6_detachhead(void **head, int off)
277{
278
279	callout_drain(&V_rtq_mtutimer);
280	return (1);
281}
282#endif
283
284/*
285 * Extended API for IPv6 FIB support.
286 */
287void
288in6_rtredirect(struct sockaddr *dst, struct sockaddr *gw, struct sockaddr *nm,
289    int flags, struct sockaddr *src, u_int fibnum)
290{
291
292	rtredirect_fib(dst, gw, nm, flags, src, fibnum);
293}
294
295int
296in6_rtrequest(int req, struct sockaddr *dst, struct sockaddr *gw,
297    struct sockaddr *mask, int flags, struct rtentry **ret_nrt, u_int fibnum)
298{
299
300	return (rtrequest_fib(req, dst, gw, mask, flags, ret_nrt, fibnum));
301}
302
303void
304in6_rtalloc(struct route_in6 *ro, u_int fibnum)
305{
306
307	rtalloc_ign_fib((struct route *)ro, 0ul, fibnum);
308}
309
310void
311in6_rtalloc_ign(struct route_in6 *ro, u_long ignflags, u_int fibnum)
312{
313
314	rtalloc_ign_fib((struct route *)ro, ignflags, fibnum);
315}
316
317struct rtentry *
318in6_rtalloc1(struct sockaddr *dst, int report, u_long ignflags, u_int fibnum)
319{
320
321	return (rtalloc1_fib(dst, report, ignflags, fibnum));
322}
323