ip_encap.c revision 82884 
1/*	$FreeBSD: head/sys/netinet/ip_encap.c 82884 2001-09-03 20:03:55Z julian $	*/
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
81#ifdef INET6
82#include <netinet/ip6.h>
83#include <netinet6/ip6_var.h>
84#include <netinet6/ip6protosw.h>
85#endif
86
87#include <machine/stdarg.h>
88
89#include <net/net_osdep.h>
90
91#include <sys/kernel.h>
92#include <sys/malloc.h>
93static MALLOC_DEFINE(M_NETADDR, "Export Host", "Export host address structure");
94
95static void encap_add __P((struct encaptab *));
96static int mask_match __P((const struct encaptab *, const struct sockaddr *,
97		const struct sockaddr *));
98static void encap_fillarg __P((struct mbuf *, const struct encaptab *));
99
100#ifndef LIST_HEAD_INITIALIZER
101/* rely upon BSS initialization */
102LIST_HEAD(, encaptab) encaptab;
103#else
104LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(&encaptab);
105#endif
106
107void
108encap_init()
109{
110	static int initialized = 0;
111
112	if (initialized)
113		return;
114	initialized++;
115#if 0
116	/*
117	 * we cannot use LIST_INIT() here, since drivers may want to call
118	 * encap_attach(), on driver attach.  encap_init() will be called
119	 * on AF_INET{,6} initialization, which happens after driver
120	 * initialization - using LIST_INIT() here can nuke encap_attach()
121	 * from drivers.
122	 */
123	LIST_INIT(&encaptab);
124#endif
125}
126
127#ifdef INET
128void
129#if __STDC__
130encap4_input(struct mbuf *m, ...)
131#else
132encap4_input(m, va_alist)
133	struct mbuf *m;
134	va_dcl
135#endif
136{
137	int off, proto;
138	struct ip *ip;
139	struct sockaddr_in s, d;
140	const struct protosw *psw;
141	struct encaptab *ep, *match;
142	va_list ap;
143	int prio, matchprio;
144
145	va_start(ap, m);
146	off = va_arg(ap, int);
147	va_end(ap);
148
149	ip = mtod(m, struct ip *);
150	proto = ip->ip_p;
151
152	bzero(&s, sizeof(s));
153	s.sin_family = AF_INET;
154	s.sin_len = sizeof(struct sockaddr_in);
155	s.sin_addr = ip->ip_src;
156	bzero(&d, sizeof(d));
157	d.sin_family = AF_INET;
158	d.sin_len = sizeof(struct sockaddr_in);
159	d.sin_addr = ip->ip_dst;
160
161	match = NULL;
162	matchprio = 0;
163	LIST_FOREACH(ep, &encaptab, chain) {
164		if (ep->af != AF_INET)
165			continue;
166		if (ep->proto >= 0 && ep->proto != proto)
167			continue;
168		if (ep->func)
169			prio = (*ep->func)(m, off, proto, ep->arg);
170		else {
171			/*
172			 * it's inbound traffic, we need to match in reverse
173			 * order
174			 */
175			prio = mask_match(ep, (struct sockaddr *)&d,
176			    (struct sockaddr *)&s);
177		}
178
179		/*
180		 * We prioritize the matches by using bit length of the
181		 * matches.  mask_match() and user-supplied matching function
182		 * should return the bit length of the matches (for example,
183		 * if both src/dst are matched for IPv4, 64 should be returned).
184		 * 0 or negative return value means "it did not match".
185		 *
186		 * The question is, since we have two "mask" portion, we
187		 * cannot really define total order between entries.
188		 * For example, which of these should be preferred?
189		 * mask_match() returns 48 (32 + 16) for both of them.
190		 *	src=3ffe::/16, dst=3ffe:501::/32
191		 *	src=3ffe:501::/32, dst=3ffe::/16
192		 *
193		 * We need to loop through all the possible candidates
194		 * to get the best match - the search takes O(n) for
195		 * n attachments (i.e. interfaces).
196		 */
197		if (prio <= 0)
198			continue;
199		if (prio > matchprio) {
200			matchprio = prio;
201			match = ep;
202		}
203	}
204
205	if (match) {
206		/* found a match, "match" has the best one */
207		psw = match->psw;
208		if (psw && psw->pr_input) {
209			encap_fillarg(m, match);
210			(*psw->pr_input)(m, off);
211		} else
212			m_freem(m);
213		return;
214	}
215
216	/* for backward compatibility - messy... */
217	if (proto == IPPROTO_IPV4) {
218		ipip_input(m, off);
219		return;
220	}
221
222	/* last resort: inject to raw socket */
223	rip_input(m, off);
224}
225#endif
226
227#ifdef INET6
228int
229encap6_input(mp, offp, proto)
230	struct mbuf **mp;
231	int *offp;
232	int proto;
233{
234	struct mbuf *m = *mp;
235	struct ip6_hdr *ip6;
236	struct sockaddr_in6 s, d;
237	const struct ip6protosw *psw;
238	struct encaptab *ep, *match;
239	int prio, matchprio;
240
241	ip6 = mtod(m, struct ip6_hdr *);
242
243	bzero(&s, sizeof(s));
244	s.sin6_family = AF_INET6;
245	s.sin6_len = sizeof(struct sockaddr_in6);
246	s.sin6_addr = ip6->ip6_src;
247	bzero(&d, sizeof(d));
248	d.sin6_family = AF_INET6;
249	d.sin6_len = sizeof(struct sockaddr_in6);
250	d.sin6_addr = ip6->ip6_dst;
251
252	match = NULL;
253	matchprio = 0;
254	LIST_FOREACH(ep, &encaptab, chain) {
255		if (ep->af != AF_INET6)
256			continue;
257		if (ep->proto >= 0 && ep->proto != proto)
258			continue;
259		if (ep->func)
260			prio = (*ep->func)(m, *offp, proto, ep->arg);
261		else {
262			/*
263			 * it's inbound traffic, we need to match in reverse
264			 * order
265			 */
266			prio = mask_match(ep, (struct sockaddr *)&d,
267			    (struct sockaddr *)&s);
268		}
269
270		/* see encap4_input() for issues here */
271		if (prio <= 0)
272			continue;
273		if (prio > matchprio) {
274			matchprio = prio;
275			match = ep;
276		}
277	}
278
279	if (match) {
280		/* found a match */
281		psw = (const struct ip6protosw *)match->psw;
282		if (psw && psw->pr_input) {
283			encap_fillarg(m, match);
284			return (*psw->pr_input)(mp, offp, proto);
285		} else {
286			m_freem(m);
287			return IPPROTO_DONE;
288		}
289	}
290
291	/* last resort: inject to raw socket */
292	return rip6_input(mp, offp, proto);
293}
294#endif
295
296static void
297encap_add(ep)
298	struct encaptab *ep;
299{
300
301	LIST_INSERT_HEAD(&encaptab, ep, chain);
302}
303
304/*
305 * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
306 * length of mask (sm and dm) is assumed to be same as sp/dp.
307 * Return value will be necessary as input (cookie) for encap_detach().
308 */
309const struct encaptab *
310encap_attach(af, proto, sp, sm, dp, dm, psw, arg)
311	int af;
312	int proto;
313	const struct sockaddr *sp, *sm;
314	const struct sockaddr *dp, *dm;
315	const struct protosw *psw;
316	void *arg;
317{
318	struct encaptab *ep;
319	int error;
320	int s;
321
322	s = splnet();
323	/* sanity check on args */
324	if (sp->sa_len > sizeof(ep->src) || dp->sa_len > sizeof(ep->dst)) {
325		error = EINVAL;
326		goto fail;
327	}
328	if (sp->sa_len != dp->sa_len) {
329		error = EINVAL;
330		goto fail;
331	}
332	if (af != sp->sa_family || af != dp->sa_family) {
333		error = EINVAL;
334		goto fail;
335	}
336
337	/* check if anyone have already attached with exactly same config */
338	LIST_FOREACH(ep, &encaptab, chain) {
339		if (ep->af != af)
340			continue;
341		if (ep->proto != proto)
342			continue;
343		if (ep->src.ss_len != sp->sa_len ||
344		    bcmp(&ep->src, sp, sp->sa_len) != 0 ||
345		    bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
346			continue;
347		if (ep->dst.ss_len != dp->sa_len ||
348		    bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
349		    bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
350			continue;
351
352		error = EEXIST;
353		goto fail;
354	}
355
356	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
357	if (ep == NULL) {
358		error = ENOBUFS;
359		goto fail;
360	}
361	bzero(ep, sizeof(*ep));
362
363	ep->af = af;
364	ep->proto = proto;
365	bcopy(sp, &ep->src, sp->sa_len);
366	bcopy(sm, &ep->srcmask, sp->sa_len);
367	bcopy(dp, &ep->dst, dp->sa_len);
368	bcopy(dm, &ep->dstmask, dp->sa_len);
369	ep->psw = psw;
370	ep->arg = arg;
371
372	encap_add(ep);
373
374	error = 0;
375	splx(s);
376	return ep;
377
378fail:
379	splx(s);
380	return NULL;
381}
382
383const struct encaptab *
384encap_attach_func(af, proto, func, psw, arg)
385	int af;
386	int proto;
387	int (*func) __P((const struct mbuf *, int, int, void *));
388	const struct protosw *psw;
389	void *arg;
390{
391	struct encaptab *ep;
392	int error;
393	int s;
394
395	s = splnet();
396	/* sanity check on args */
397	if (!func) {
398		error = EINVAL;
399		goto fail;
400	}
401
402	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
403	if (ep == NULL) {
404		error = ENOBUFS;
405		goto fail;
406	}
407	bzero(ep, sizeof(*ep));
408
409	ep->af = af;
410	ep->proto = proto;
411	ep->func = func;
412	ep->psw = psw;
413	ep->arg = arg;
414
415	encap_add(ep);
416
417	error = 0;
418	splx(s);
419	return ep;
420
421fail:
422	splx(s);
423	return NULL;
424}
425
426int
427encap_detach(cookie)
428	const struct encaptab *cookie;
429{
430	const struct encaptab *ep = cookie;
431	struct encaptab *p;
432
433	LIST_FOREACH(p, &encaptab, chain) {
434		if (p == ep) {
435			LIST_REMOVE(p, chain);
436			free(p, M_NETADDR);	/*XXX*/
437			return 0;
438		}
439	}
440
441	return EINVAL;
442}
443
444static int
445mask_match(ep, sp, dp)
446	const struct encaptab *ep;
447	const struct sockaddr *sp;
448	const struct sockaddr *dp;
449{
450	struct sockaddr_storage s;
451	struct sockaddr_storage d;
452	int i;
453	const u_int8_t *p, *q;
454	u_int8_t *r;
455	int matchlen;
456
457	if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d))
458		return 0;
459	if (sp->sa_family != ep->af || dp->sa_family != ep->af)
460		return 0;
461	if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
462		return 0;
463
464	matchlen = 0;
465
466	p = (const u_int8_t *)sp;
467	q = (const u_int8_t *)&ep->srcmask;
468	r = (u_int8_t *)&s;
469	for (i = 0 ; i < sp->sa_len; i++) {
470		r[i] = p[i] & q[i];
471		/* XXX estimate */
472		matchlen += (q[i] ? 8 : 0);
473	}
474
475	p = (const u_int8_t *)dp;
476	q = (const u_int8_t *)&ep->dstmask;
477	r = (u_int8_t *)&d;
478	for (i = 0 ; i < dp->sa_len; i++) {
479		r[i] = p[i] & q[i];
480		/* XXX rough estimate */
481		matchlen += (q[i] ? 8 : 0);
482	}
483
484	/* need to overwrite len/family portion as we don't compare them */
485	s.ss_len = sp->sa_len;
486	s.ss_family = sp->sa_family;
487	d.ss_len = dp->sa_len;
488	d.ss_family = dp->sa_family;
489
490	if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
491	    bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
492		return matchlen;
493	} else
494		return 0;
495}
496
497static void
498encap_fillarg(m, ep)
499	struct mbuf *m;
500	const struct encaptab *ep;
501{
502#if 0
503	m->m_pkthdr.aux = ep->arg;
504#else
505	struct mbuf *n;
506
507	n = m_aux_add(m, AF_INET, IPPROTO_IPV4);
508	if (n) {
509		*mtod(n, void **) = ep->arg;
510		n->m_len = sizeof(void *);
511	}
512#endif
513}
514
515void *
516encap_getarg(m)
517	struct mbuf *m;
518{
519	void *p;
520#if 0
521	p = m->m_pkthdr.aux;
522	m->m_pkthdr.aux = NULL;
523	return p;
524#else
525	struct mbuf *n;
526
527	p = NULL;
528	n = m_aux_find(m, AF_INET, IPPROTO_IPV4);
529	if (n) {
530		if (n->m_len == sizeof(void *))
531			p = *mtod(n, void **);
532		m_aux_delete(m, n);
533	}
534	return p;
535#endif
536}
537