ip_encap.c revision 78064 
1/*	$FreeBSD: head/sys/netinet/ip_encap.c 78064 2001-06-11 12:39:29Z ume $	*/
2/*	$KAME: ip_encap.c,v 1.41 2001/03/15 08:35:08 itojun Exp $	*/
3
4/*
5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 *    notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 *    notice, this list of conditions and the following disclaimer in the
15 *    documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the project nor the names of its contributors
17 *    may be used to endorse or promote products derived from this software
18 *    without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 */
32/*
33 * My grandfather said that there's a devil inside tunnelling technology...
34 *
35 * We have surprisingly many protocols that want packets with IP protocol
36 * #4 or #41.  Here's a list of protocols that want protocol #41:
37 *	RFC1933 configured tunnel
38 *	RFC1933 automatic tunnel
39 *	RFC2401 IPsec tunnel
40 *	RFC2473 IPv6 generic packet tunnelling
41 *	RFC2529 6over4 tunnel
42 *	mobile-ip6 (uses RFC2473)
43 *	6to4 tunnel
44 * Here's a list of protocol that want protocol #4:
45 *	RFC1853 IPv4-in-IPv4 tunnelling
46 *	RFC2003 IPv4 encapsulation within IPv4
47 *	RFC2344 reverse tunnelling for mobile-ip4
48 *	RFC2401 IPsec tunnel
49 * Well, what can I say.  They impose different en/decapsulation mechanism
50 * from each other, so they need separate protocol handler.  The only one
51 * we can easily determine by protocol # is IPsec, which always has
52 * AH/ESP/IPComp header right after outer IP header.
53 *
54 * So, clearly good old protosw does not work for protocol #4 and #41.
55 * The code will let you match protocol via src/dst address pair.
56 */
57/* XXX is M_NETADDR correct? */
58
59#include "opt_mrouting.h"
60#include "opt_inet.h"
61#include "opt_inet6.h"
62
63#include <sys/param.h>
64#include <sys/systm.h>
65#include <sys/socket.h>
66#include <sys/sockio.h>
67#include <sys/mbuf.h>
68#include <sys/errno.h>
69#include <sys/protosw.h>
70#include <sys/queue.h>
71
72#include <net/if.h>
73#include <net/route.h>
74
75#include <netinet/in.h>
76#include <netinet/in_systm.h>
77#include <netinet/ip.h>
78#include <netinet/ip_var.h>
79#include <netinet/ip_encap.h>
80#ifdef MROUTING
81#include <netinet/ip_mroute.h>
82#endif /* MROUTING */
83#include <netinet/ipprotosw.h>
84
85#ifdef INET6
86#include <netinet/ip6.h>
87#include <netinet6/ip6_var.h>
88#include <netinet6/ip6protosw.h>
89#endif
90
91#include <machine/stdarg.h>
92
93#include <net/net_osdep.h>
94
95#include <sys/kernel.h>
96#include <sys/malloc.h>
97static MALLOC_DEFINE(M_NETADDR, "Export Host", "Export host address structure");
98
99static void encap_add __P((struct encaptab *));
100static int mask_match __P((const struct encaptab *, const struct sockaddr *,
101		const struct sockaddr *));
102static void encap_fillarg __P((struct mbuf *, const struct encaptab *));
103
104#ifndef LIST_HEAD_INITIALIZER
105/* rely upon BSS initialization */
106LIST_HEAD(, encaptab) encaptab;
107#else
108LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(&encaptab);
109#endif
110
111void
112encap_init()
113{
114	static int initialized = 0;
115
116	if (initialized)
117		return;
118	initialized++;
119#if 0
120	/*
121	 * we cannot use LIST_INIT() here, since drivers may want to call
122	 * encap_attach(), on driver attach.  encap_init() will be called
123	 * on AF_INET{,6} initialization, which happens after driver
124	 * initialization - using LIST_INIT() here can nuke encap_attach()
125	 * from drivers.
126	 */
127	LIST_INIT(&encaptab);
128#endif
129}
130
131#ifdef INET
132void
133#if __STDC__
134encap4_input(struct mbuf *m, ...)
135#else
136encap4_input(m, va_alist)
137	struct mbuf *m;
138	va_dcl
139#endif
140{
141	int off, proto;
142	struct ip *ip;
143	struct sockaddr_in s, d;
144	const struct ipprotosw *psw;
145	struct encaptab *ep, *match;
146	va_list ap;
147	int prio, matchprio;
148
149	va_start(ap, m);
150	off = va_arg(ap, int);
151	proto = va_arg(ap, int);
152	va_end(ap);
153
154	ip = mtod(m, struct ip *);
155
156	bzero(&s, sizeof(s));
157	s.sin_family = AF_INET;
158	s.sin_len = sizeof(struct sockaddr_in);
159	s.sin_addr = ip->ip_src;
160	bzero(&d, sizeof(d));
161	d.sin_family = AF_INET;
162	d.sin_len = sizeof(struct sockaddr_in);
163	d.sin_addr = ip->ip_dst;
164
165	match = NULL;
166	matchprio = 0;
167	LIST_FOREACH(ep, &encaptab, chain) {
168		if (ep->af != AF_INET)
169			continue;
170		if (ep->proto >= 0 && ep->proto != proto)
171			continue;
172		if (ep->func)
173			prio = (*ep->func)(m, off, proto, ep->arg);
174		else {
175			/*
176			 * it's inbound traffic, we need to match in reverse
177			 * order
178			 */
179			prio = mask_match(ep, (struct sockaddr *)&d,
180			    (struct sockaddr *)&s);
181		}
182
183		/*
184		 * We prioritize the matches by using bit length of the
185		 * matches.  mask_match() and user-supplied matching function
186		 * should return the bit length of the matches (for example,
187		 * if both src/dst are matched for IPv4, 64 should be returned).
188		 * 0 or negative return value means "it did not match".
189		 *
190		 * The question is, since we have two "mask" portion, we
191		 * cannot really define total order between entries.
192		 * For example, which of these should be preferred?
193		 * mask_match() returns 48 (32 + 16) for both of them.
194		 *	src=3ffe::/16, dst=3ffe:501::/32
195		 *	src=3ffe:501::/32, dst=3ffe::/16
196		 *
197		 * We need to loop through all the possible candidates
198		 * to get the best match - the search takes O(n) for
199		 * n attachments (i.e. interfaces).
200		 */
201		if (prio <= 0)
202			continue;
203		if (prio > matchprio) {
204			matchprio = prio;
205			match = ep;
206		}
207	}
208
209	if (match) {
210		/* found a match, "match" has the best one */
211		psw = (const struct ipprotosw *)match->psw;
212		if (psw && psw->pr_input) {
213			encap_fillarg(m, match);
214			(*psw->pr_input)(m, off, proto);
215		} else
216			m_freem(m);
217		return;
218	}
219
220	/* for backward compatibility */
221# ifdef MROUTING
222#  define COMPATFUNC	ipip_input
223# endif /*MROUTING*/
224
225#ifdef COMPATFUNC
226	if (proto == IPPROTO_IPV4) {
227		COMPATFUNC(m, off, proto);
228		return;
229	}
230#endif
231
232	/* last resort: inject to raw socket */
233	rip_input(m, off, proto);
234}
235#endif
236
237#ifdef INET6
238int
239encap6_input(mp, offp, proto)
240	struct mbuf **mp;
241	int *offp;
242	int proto;
243{
244	struct mbuf *m = *mp;
245	struct ip6_hdr *ip6;
246	struct sockaddr_in6 s, d;
247	const struct ip6protosw *psw;
248	struct encaptab *ep, *match;
249	int prio, matchprio;
250
251	ip6 = mtod(m, struct ip6_hdr *);
252
253	bzero(&s, sizeof(s));
254	s.sin6_family = AF_INET6;
255	s.sin6_len = sizeof(struct sockaddr_in6);
256	s.sin6_addr = ip6->ip6_src;
257	bzero(&d, sizeof(d));
258	d.sin6_family = AF_INET6;
259	d.sin6_len = sizeof(struct sockaddr_in6);
260	d.sin6_addr = ip6->ip6_dst;
261
262	match = NULL;
263	matchprio = 0;
264	LIST_FOREACH(ep, &encaptab, chain) {
265		if (ep->af != AF_INET6)
266			continue;
267		if (ep->proto >= 0 && ep->proto != proto)
268			continue;
269		if (ep->func)
270			prio = (*ep->func)(m, *offp, proto, ep->arg);
271		else {
272			/*
273			 * it's inbound traffic, we need to match in reverse
274			 * order
275			 */
276			prio = mask_match(ep, (struct sockaddr *)&d,
277			    (struct sockaddr *)&s);
278		}
279
280		/* see encap4_input() for issues here */
281		if (prio <= 0)
282			continue;
283		if (prio > matchprio) {
284			matchprio = prio;
285			match = ep;
286		}
287	}
288
289	if (match) {
290		/* found a match */
291		psw = (const struct ip6protosw *)match->psw;
292		if (psw && psw->pr_input) {
293			encap_fillarg(m, match);
294			return (*psw->pr_input)(mp, offp, proto);
295		} else {
296			m_freem(m);
297			return IPPROTO_DONE;
298		}
299	}
300
301	/* last resort: inject to raw socket */
302	return rip6_input(mp, offp, proto);
303}
304#endif
305
306static void
307encap_add(ep)
308	struct encaptab *ep;
309{
310
311	LIST_INSERT_HEAD(&encaptab, ep, chain);
312}
313
314/*
315 * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
316 * length of mask (sm and dm) is assumed to be same as sp/dp.
317 * Return value will be necessary as input (cookie) for encap_detach().
318 */
319const struct encaptab *
320encap_attach(af, proto, sp, sm, dp, dm, psw, arg)
321	int af;
322	int proto;
323	const struct sockaddr *sp, *sm;
324	const struct sockaddr *dp, *dm;
325	const struct protosw *psw;
326	void *arg;
327{
328	struct encaptab *ep;
329	int error;
330	int s;
331
332	s = splnet();
333	/* sanity check on args */
334	if (sp->sa_len > sizeof(ep->src) || dp->sa_len > sizeof(ep->dst)) {
335		error = EINVAL;
336		goto fail;
337	}
338	if (sp->sa_len != dp->sa_len) {
339		error = EINVAL;
340		goto fail;
341	}
342	if (af != sp->sa_family || af != dp->sa_family) {
343		error = EINVAL;
344		goto fail;
345	}
346
347	/* check if anyone have already attached with exactly same config */
348	LIST_FOREACH(ep, &encaptab, chain) {
349		if (ep->af != af)
350			continue;
351		if (ep->proto != proto)
352			continue;
353		if (ep->src.ss_len != sp->sa_len ||
354		    bcmp(&ep->src, sp, sp->sa_len) != 0 ||
355		    bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
356			continue;
357		if (ep->dst.ss_len != dp->sa_len ||
358		    bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
359		    bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
360			continue;
361
362		error = EEXIST;
363		goto fail;
364	}
365
366	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
367	if (ep == NULL) {
368		error = ENOBUFS;
369		goto fail;
370	}
371	bzero(ep, sizeof(*ep));
372
373	ep->af = af;
374	ep->proto = proto;
375	bcopy(sp, &ep->src, sp->sa_len);
376	bcopy(sm, &ep->srcmask, sp->sa_len);
377	bcopy(dp, &ep->dst, dp->sa_len);
378	bcopy(dm, &ep->dstmask, dp->sa_len);
379	ep->psw = psw;
380	ep->arg = arg;
381
382	encap_add(ep);
383
384	error = 0;
385	splx(s);
386	return ep;
387
388fail:
389	splx(s);
390	return NULL;
391}
392
393const struct encaptab *
394encap_attach_func(af, proto, func, psw, arg)
395	int af;
396	int proto;
397	int (*func) __P((const struct mbuf *, int, int, void *));
398	const struct protosw *psw;
399	void *arg;
400{
401	struct encaptab *ep;
402	int error;
403	int s;
404
405	s = splnet();
406	/* sanity check on args */
407	if (!func) {
408		error = EINVAL;
409		goto fail;
410	}
411
412	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
413	if (ep == NULL) {
414		error = ENOBUFS;
415		goto fail;
416	}
417	bzero(ep, sizeof(*ep));
418
419	ep->af = af;
420	ep->proto = proto;
421	ep->func = func;
422	ep->psw = psw;
423	ep->arg = arg;
424
425	encap_add(ep);
426
427	error = 0;
428	splx(s);
429	return ep;
430
431fail:
432	splx(s);
433	return NULL;
434}
435
436int
437encap_detach(cookie)
438	const struct encaptab *cookie;
439{
440	const struct encaptab *ep = cookie;
441	struct encaptab *p;
442
443	LIST_FOREACH(p, &encaptab, chain) {
444		if (p == ep) {
445			LIST_REMOVE(p, chain);
446			free(p, M_NETADDR);	/*XXX*/
447			return 0;
448		}
449	}
450
451	return EINVAL;
452}
453
454static int
455mask_match(ep, sp, dp)
456	const struct encaptab *ep;
457	const struct sockaddr *sp;
458	const struct sockaddr *dp;
459{
460	struct sockaddr_storage s;
461	struct sockaddr_storage d;
462	int i;
463	const u_int8_t *p, *q;
464	u_int8_t *r;
465	int matchlen;
466
467	if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d))
468		return 0;
469	if (sp->sa_family != ep->af || dp->sa_family != ep->af)
470		return 0;
471	if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
472		return 0;
473
474	matchlen = 0;
475
476	p = (const u_int8_t *)sp;
477	q = (const u_int8_t *)&ep->srcmask;
478	r = (u_int8_t *)&s;
479	for (i = 0 ; i < sp->sa_len; i++) {
480		r[i] = p[i] & q[i];
481		/* XXX estimate */
482		matchlen += (q[i] ? 8 : 0);
483	}
484
485	p = (const u_int8_t *)dp;
486	q = (const u_int8_t *)&ep->dstmask;
487	r = (u_int8_t *)&d;
488	for (i = 0 ; i < dp->sa_len; i++) {
489		r[i] = p[i] & q[i];
490		/* XXX rough estimate */
491		matchlen += (q[i] ? 8 : 0);
492	}
493
494	/* need to overwrite len/family portion as we don't compare them */
495	s.ss_len = sp->sa_len;
496	s.ss_family = sp->sa_family;
497	d.ss_len = dp->sa_len;
498	d.ss_family = dp->sa_family;
499
500	if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
501	    bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
502		return matchlen;
503	} else
504		return 0;
505}
506
507static void
508encap_fillarg(m, ep)
509	struct mbuf *m;
510	const struct encaptab *ep;
511{
512#if 0
513	m->m_pkthdr.aux = ep->arg;
514#else
515	struct mbuf *n;
516
517	n = m_aux_add(m, AF_INET, IPPROTO_IPV4);
518	if (n) {
519		*mtod(n, void **) = ep->arg;
520		n->m_len = sizeof(void *);
521	}
522#endif
523}
524
525void *
526encap_getarg(m)
527	struct mbuf *m;
528{
529	void *p;
530#if 0
531	p = m->m_pkthdr.aux;
532	m->m_pkthdr.aux = NULL;
533	return p;
534#else
535	struct mbuf *n;
536
537	p = NULL;
538	n = m_aux_find(m, AF_INET, IPPROTO_IPV4);
539	if (n) {
540		if (n->m_len == sizeof(void *))
541			p = *mtod(n, void **);
542		m_aux_delete(m, n);
543	}
544	return p;
545#endif
546}
547