ip_encap.c revision 92723 
1/*	$FreeBSD: head/sys/netinet/ip_encap.c 92723 2002-03-19 21:25:46Z alfred $	*/
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(struct encaptab *);
96static int mask_match(const struct encaptab *, const struct sockaddr *,
97		const struct sockaddr *);
98static void encap_fillarg(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
129encap4_input(m, off)
130	struct mbuf *m;
131	int off;
132{
133	struct ip *ip;
134	int proto;
135	struct sockaddr_in s, d;
136	const struct protosw *psw;
137	struct encaptab *ep, *match;
138	int prio, matchprio;
139
140	ip = mtod(m, struct ip *);
141	proto = ip->ip_p;
142
143	bzero(&s, sizeof(s));
144	s.sin_family = AF_INET;
145	s.sin_len = sizeof(struct sockaddr_in);
146	s.sin_addr = ip->ip_src;
147	bzero(&d, sizeof(d));
148	d.sin_family = AF_INET;
149	d.sin_len = sizeof(struct sockaddr_in);
150	d.sin_addr = ip->ip_dst;
151
152	match = NULL;
153	matchprio = 0;
154	LIST_FOREACH(ep, &encaptab, chain) {
155		if (ep->af != AF_INET)
156			continue;
157		if (ep->proto >= 0 && ep->proto != proto)
158			continue;
159		if (ep->func)
160			prio = (*ep->func)(m, off, proto, ep->arg);
161		else {
162			/*
163			 * it's inbound traffic, we need to match in reverse
164			 * order
165			 */
166			prio = mask_match(ep, (struct sockaddr *)&d,
167			    (struct sockaddr *)&s);
168		}
169
170		/*
171		 * We prioritize the matches by using bit length of the
172		 * matches.  mask_match() and user-supplied matching function
173		 * should return the bit length of the matches (for example,
174		 * if both src/dst are matched for IPv4, 64 should be returned).
175		 * 0 or negative return value means "it did not match".
176		 *
177		 * The question is, since we have two "mask" portion, we
178		 * cannot really define total order between entries.
179		 * For example, which of these should be preferred?
180		 * mask_match() returns 48 (32 + 16) for both of them.
181		 *	src=3ffe::/16, dst=3ffe:501::/32
182		 *	src=3ffe:501::/32, dst=3ffe::/16
183		 *
184		 * We need to loop through all the possible candidates
185		 * to get the best match - the search takes O(n) for
186		 * n attachments (i.e. interfaces).
187		 */
188		if (prio <= 0)
189			continue;
190		if (prio > matchprio) {
191			matchprio = prio;
192			match = ep;
193		}
194	}
195
196	if (match) {
197		/* found a match, "match" has the best one */
198		psw = match->psw;
199		if (psw && psw->pr_input) {
200			encap_fillarg(m, match);
201			(*psw->pr_input)(m, off);
202		} else
203			m_freem(m);
204		return;
205	}
206
207	/* last resort: inject to raw socket */
208	rip_input(m, off);
209}
210#endif
211
212#ifdef INET6
213int
214encap6_input(mp, offp, proto)
215	struct mbuf **mp;
216	int *offp;
217	int proto;
218{
219	struct mbuf *m = *mp;
220	struct ip6_hdr *ip6;
221	struct sockaddr_in6 s, d;
222	const struct ip6protosw *psw;
223	struct encaptab *ep, *match;
224	int prio, matchprio;
225
226	ip6 = mtod(m, struct ip6_hdr *);
227
228	bzero(&s, sizeof(s));
229	s.sin6_family = AF_INET6;
230	s.sin6_len = sizeof(struct sockaddr_in6);
231	s.sin6_addr = ip6->ip6_src;
232	bzero(&d, sizeof(d));
233	d.sin6_family = AF_INET6;
234	d.sin6_len = sizeof(struct sockaddr_in6);
235	d.sin6_addr = ip6->ip6_dst;
236
237	match = NULL;
238	matchprio = 0;
239	LIST_FOREACH(ep, &encaptab, chain) {
240		if (ep->af != AF_INET6)
241			continue;
242		if (ep->proto >= 0 && ep->proto != proto)
243			continue;
244		if (ep->func)
245			prio = (*ep->func)(m, *offp, proto, ep->arg);
246		else {
247			/*
248			 * it's inbound traffic, we need to match in reverse
249			 * order
250			 */
251			prio = mask_match(ep, (struct sockaddr *)&d,
252			    (struct sockaddr *)&s);
253		}
254
255		/* see encap4_input() for issues here */
256		if (prio <= 0)
257			continue;
258		if (prio > matchprio) {
259			matchprio = prio;
260			match = ep;
261		}
262	}
263
264	if (match) {
265		/* found a match */
266		psw = (const struct ip6protosw *)match->psw;
267		if (psw && psw->pr_input) {
268			encap_fillarg(m, match);
269			return (*psw->pr_input)(mp, offp, proto);
270		} else {
271			m_freem(m);
272			return IPPROTO_DONE;
273		}
274	}
275
276	/* last resort: inject to raw socket */
277	return rip6_input(mp, offp, proto);
278}
279#endif
280
281static void
282encap_add(ep)
283	struct encaptab *ep;
284{
285
286	LIST_INSERT_HEAD(&encaptab, ep, chain);
287}
288
289/*
290 * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
291 * length of mask (sm and dm) is assumed to be same as sp/dp.
292 * Return value will be necessary as input (cookie) for encap_detach().
293 */
294const struct encaptab *
295encap_attach(af, proto, sp, sm, dp, dm, psw, arg)
296	int af;
297	int proto;
298	const struct sockaddr *sp, *sm;
299	const struct sockaddr *dp, *dm;
300	const struct protosw *psw;
301	void *arg;
302{
303	struct encaptab *ep;
304	int error;
305	int s;
306
307	s = splnet();
308	/* sanity check on args */
309	if (sp->sa_len > sizeof(ep->src) || dp->sa_len > sizeof(ep->dst)) {
310		error = EINVAL;
311		goto fail;
312	}
313	if (sp->sa_len != dp->sa_len) {
314		error = EINVAL;
315		goto fail;
316	}
317	if (af != sp->sa_family || af != dp->sa_family) {
318		error = EINVAL;
319		goto fail;
320	}
321
322	/* check if anyone have already attached with exactly same config */
323	LIST_FOREACH(ep, &encaptab, chain) {
324		if (ep->af != af)
325			continue;
326		if (ep->proto != proto)
327			continue;
328		if (ep->src.ss_len != sp->sa_len ||
329		    bcmp(&ep->src, sp, sp->sa_len) != 0 ||
330		    bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
331			continue;
332		if (ep->dst.ss_len != dp->sa_len ||
333		    bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
334		    bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
335			continue;
336
337		error = EEXIST;
338		goto fail;
339	}
340
341	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
342	if (ep == NULL) {
343		error = ENOBUFS;
344		goto fail;
345	}
346	bzero(ep, sizeof(*ep));
347
348	ep->af = af;
349	ep->proto = proto;
350	bcopy(sp, &ep->src, sp->sa_len);
351	bcopy(sm, &ep->srcmask, sp->sa_len);
352	bcopy(dp, &ep->dst, dp->sa_len);
353	bcopy(dm, &ep->dstmask, dp->sa_len);
354	ep->psw = psw;
355	ep->arg = arg;
356
357	encap_add(ep);
358
359	error = 0;
360	splx(s);
361	return ep;
362
363fail:
364	splx(s);
365	return NULL;
366}
367
368const struct encaptab *
369encap_attach_func(af, proto, func, psw, arg)
370	int af;
371	int proto;
372	int (*func)(const struct mbuf *, int, int, void *);
373	const struct protosw *psw;
374	void *arg;
375{
376	struct encaptab *ep;
377	int error;
378	int s;
379
380	s = splnet();
381	/* sanity check on args */
382	if (!func) {
383		error = EINVAL;
384		goto fail;
385	}
386
387	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
388	if (ep == NULL) {
389		error = ENOBUFS;
390		goto fail;
391	}
392	bzero(ep, sizeof(*ep));
393
394	ep->af = af;
395	ep->proto = proto;
396	ep->func = func;
397	ep->psw = psw;
398	ep->arg = arg;
399
400	encap_add(ep);
401
402	error = 0;
403	splx(s);
404	return ep;
405
406fail:
407	splx(s);
408	return NULL;
409}
410
411int
412encap_detach(cookie)
413	const struct encaptab *cookie;
414{
415	const struct encaptab *ep = cookie;
416	struct encaptab *p;
417
418	LIST_FOREACH(p, &encaptab, chain) {
419		if (p == ep) {
420			LIST_REMOVE(p, chain);
421			free(p, M_NETADDR);	/*XXX*/
422			return 0;
423		}
424	}
425
426	return EINVAL;
427}
428
429static int
430mask_match(ep, sp, dp)
431	const struct encaptab *ep;
432	const struct sockaddr *sp;
433	const struct sockaddr *dp;
434{
435	struct sockaddr_storage s;
436	struct sockaddr_storage d;
437	int i;
438	const u_int8_t *p, *q;
439	u_int8_t *r;
440	int matchlen;
441
442	if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d))
443		return 0;
444	if (sp->sa_family != ep->af || dp->sa_family != ep->af)
445		return 0;
446	if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
447		return 0;
448
449	matchlen = 0;
450
451	p = (const u_int8_t *)sp;
452	q = (const u_int8_t *)&ep->srcmask;
453	r = (u_int8_t *)&s;
454	for (i = 0 ; i < sp->sa_len; i++) {
455		r[i] = p[i] & q[i];
456		/* XXX estimate */
457		matchlen += (q[i] ? 8 : 0);
458	}
459
460	p = (const u_int8_t *)dp;
461	q = (const u_int8_t *)&ep->dstmask;
462	r = (u_int8_t *)&d;
463	for (i = 0 ; i < dp->sa_len; i++) {
464		r[i] = p[i] & q[i];
465		/* XXX rough estimate */
466		matchlen += (q[i] ? 8 : 0);
467	}
468
469	/* need to overwrite len/family portion as we don't compare them */
470	s.ss_len = sp->sa_len;
471	s.ss_family = sp->sa_family;
472	d.ss_len = dp->sa_len;
473	d.ss_family = dp->sa_family;
474
475	if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
476	    bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
477		return matchlen;
478	} else
479		return 0;
480}
481
482static void
483encap_fillarg(m, ep)
484	struct mbuf *m;
485	const struct encaptab *ep;
486{
487#if 0
488	m->m_pkthdr.aux = ep->arg;
489#else
490	struct mbuf *n;
491
492	n = m_aux_add(m, AF_INET, IPPROTO_IPV4);
493	if (n) {
494		*mtod(n, void **) = ep->arg;
495		n->m_len = sizeof(void *);
496	}
497#endif
498}
499
500void *
501encap_getarg(m)
502	struct mbuf *m;
503{
504	void *p;
505#if 0
506	p = m->m_pkthdr.aux;
507	m->m_pkthdr.aux = NULL;
508	return p;
509#else
510	struct mbuf *n;
511
512	p = NULL;
513	n = m_aux_find(m, AF_INET, IPPROTO_IPV4);
514	if (n) {
515		if (n->m_len == sizeof(void *))
516			p = *mtod(n, void **);
517		m_aux_delete(m, n);
518	}
519	return p;
520#endif
521}
522