ip_encap.c revision 160981 
1/*	$FreeBSD: head/sys/netinet/ip_encap.c 160981 2006-08-04 21:27:40Z brooks $	*/
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 *	RFC3056 6to4 tunnel
44 *	isatap tunnel
45 * Here's a list of protocol that want protocol #4:
46 *	RFC1853 IPv4-in-IPv4 tunnelling
47 *	RFC2003 IPv4 encapsulation within IPv4
48 *	RFC2344 reverse tunnelling for mobile-ip4
49 *	RFC2401 IPsec tunnel
50 * Well, what can I say.  They impose different en/decapsulation mechanism
51 * from each other, so they need separate protocol handler.  The only one
52 * we can easily determine by protocol # is IPsec, which always has
53 * AH/ESP/IPComp header right after outer IP header.
54 *
55 * So, clearly good old protosw does not work for protocol #4 and #41.
56 * The code will let you match protocol via src/dst address pair.
57 */
58/* XXX is M_NETADDR correct? */
59
60#include "opt_mrouting.h"
61#include "opt_inet.h"
62#include "opt_inet6.h"
63
64#include <sys/param.h>
65#include <sys/systm.h>
66#include <sys/socket.h>
67#include <sys/sockio.h>
68#include <sys/mbuf.h>
69#include <sys/errno.h>
70#include <sys/protosw.h>
71#include <sys/queue.h>
72
73#include <net/if.h>
74#include <net/route.h>
75
76#include <netinet/in.h>
77#include <netinet/in_systm.h>
78#include <netinet/ip.h>
79#include <netinet/ip_var.h>
80#include <netinet/ip_encap.h>
81
82#ifdef INET6
83#include <netinet/ip6.h>
84#include <netinet6/ip6_var.h>
85#include <netinet6/ip6protosw.h>
86#endif
87
88#include <machine/stdarg.h>
89
90#include <sys/kernel.h>
91#include <sys/malloc.h>
92static MALLOC_DEFINE(M_NETADDR, "encap_export_host", "Export host address structure");
93
94static void encap_add(struct encaptab *);
95static int mask_match(const struct encaptab *, const struct sockaddr *,
96		const struct sockaddr *);
97static void encap_fillarg(struct mbuf *, const struct encaptab *);
98
99/*
100 * All global variables in ip_encap.c are locked using encapmtx.
101 */
102static struct mtx encapmtx;
103MTX_SYSINIT(encapmtx, &encapmtx, "encapmtx", MTX_DEF);
104LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(&encaptab);
105
106/*
107 * We currently keey encap_init() for source code compatibility reasons --
108 * it's referenced by KAME pieces in netinet6.
109 */
110void
111encap_init()
112{
113}
114
115#ifdef INET
116void
117encap4_input(m, off)
118	struct mbuf *m;
119	int off;
120{
121	struct ip *ip;
122	int proto;
123	struct sockaddr_in s, d;
124	const struct protosw *psw;
125	struct encaptab *ep, *match;
126	int prio, matchprio;
127
128	ip = mtod(m, struct ip *);
129	proto = ip->ip_p;
130
131	bzero(&s, sizeof(s));
132	s.sin_family = AF_INET;
133	s.sin_len = sizeof(struct sockaddr_in);
134	s.sin_addr = ip->ip_src;
135	bzero(&d, sizeof(d));
136	d.sin_family = AF_INET;
137	d.sin_len = sizeof(struct sockaddr_in);
138	d.sin_addr = ip->ip_dst;
139
140	match = NULL;
141	matchprio = 0;
142	mtx_lock(&encapmtx);
143	LIST_FOREACH(ep, &encaptab, chain) {
144		if (ep->af != AF_INET)
145			continue;
146		if (ep->proto >= 0 && ep->proto != proto)
147			continue;
148		if (ep->func)
149			prio = (*ep->func)(m, off, proto, ep->arg);
150		else {
151			/*
152			 * it's inbound traffic, we need to match in reverse
153			 * order
154			 */
155			prio = mask_match(ep, (struct sockaddr *)&d,
156			    (struct sockaddr *)&s);
157		}
158
159		/*
160		 * We prioritize the matches by using bit length of the
161		 * matches.  mask_match() and user-supplied matching function
162		 * should return the bit length of the matches (for example,
163		 * if both src/dst are matched for IPv4, 64 should be returned).
164		 * 0 or negative return value means "it did not match".
165		 *
166		 * The question is, since we have two "mask" portion, we
167		 * cannot really define total order between entries.
168		 * For example, which of these should be preferred?
169		 * mask_match() returns 48 (32 + 16) for both of them.
170		 *	src=3ffe::/16, dst=3ffe:501::/32
171		 *	src=3ffe:501::/32, dst=3ffe::/16
172		 *
173		 * We need to loop through all the possible candidates
174		 * to get the best match - the search takes O(n) for
175		 * n attachments (i.e. interfaces).
176		 */
177		if (prio <= 0)
178			continue;
179		if (prio > matchprio) {
180			matchprio = prio;
181			match = ep;
182		}
183	}
184	mtx_unlock(&encapmtx);
185
186	if (match) {
187		/* found a match, "match" has the best one */
188		psw = match->psw;
189		if (psw && psw->pr_input) {
190			encap_fillarg(m, match);
191			(*psw->pr_input)(m, off);
192		} else
193			m_freem(m);
194		return;
195	}
196
197	/* last resort: inject to raw socket */
198	rip_input(m, off);
199}
200#endif
201
202#ifdef INET6
203int
204encap6_input(mp, offp, proto)
205	struct mbuf **mp;
206	int *offp;
207	int proto;
208{
209	struct mbuf *m = *mp;
210	struct ip6_hdr *ip6;
211	struct sockaddr_in6 s, d;
212	const struct ip6protosw *psw;
213	struct encaptab *ep, *match;
214	int prio, matchprio;
215
216	ip6 = mtod(m, struct ip6_hdr *);
217
218	bzero(&s, sizeof(s));
219	s.sin6_family = AF_INET6;
220	s.sin6_len = sizeof(struct sockaddr_in6);
221	s.sin6_addr = ip6->ip6_src;
222	bzero(&d, sizeof(d));
223	d.sin6_family = AF_INET6;
224	d.sin6_len = sizeof(struct sockaddr_in6);
225	d.sin6_addr = ip6->ip6_dst;
226
227	match = NULL;
228	matchprio = 0;
229	mtx_lock(&encapmtx);
230	LIST_FOREACH(ep, &encaptab, chain) {
231		if (ep->af != AF_INET6)
232			continue;
233		if (ep->proto >= 0 && ep->proto != proto)
234			continue;
235		if (ep->func)
236			prio = (*ep->func)(m, *offp, proto, ep->arg);
237		else {
238			/*
239			 * it's inbound traffic, we need to match in reverse
240			 * order
241			 */
242			prio = mask_match(ep, (struct sockaddr *)&d,
243			    (struct sockaddr *)&s);
244		}
245
246		/* see encap4_input() for issues here */
247		if (prio <= 0)
248			continue;
249		if (prio > matchprio) {
250			matchprio = prio;
251			match = ep;
252		}
253	}
254	mtx_unlock(&encapmtx);
255
256	if (match) {
257		/* found a match */
258		psw = (const struct ip6protosw *)match->psw;
259		if (psw && psw->pr_input) {
260			encap_fillarg(m, match);
261			return (*psw->pr_input)(mp, offp, proto);
262		} else {
263			m_freem(m);
264			return IPPROTO_DONE;
265		}
266	}
267
268	/* last resort: inject to raw socket */
269	return rip6_input(mp, offp, proto);
270}
271#endif
272
273/*lint -sem(encap_add, custodial(1)) */
274static void
275encap_add(ep)
276	struct encaptab *ep;
277{
278
279	mtx_assert(&encapmtx, MA_OWNED);
280	LIST_INSERT_HEAD(&encaptab, ep, chain);
281}
282
283/*
284 * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
285 * length of mask (sm and dm) is assumed to be same as sp/dp.
286 * Return value will be necessary as input (cookie) for encap_detach().
287 */
288const struct encaptab *
289encap_attach(af, proto, sp, sm, dp, dm, psw, arg)
290	int af;
291	int proto;
292	const struct sockaddr *sp, *sm;
293	const struct sockaddr *dp, *dm;
294	const struct protosw *psw;
295	void *arg;
296{
297	struct encaptab *ep;
298
299	/* sanity check on args */
300	if (sp->sa_len > sizeof(ep->src) || dp->sa_len > sizeof(ep->dst))
301		return (NULL);
302	if (sp->sa_len != dp->sa_len)
303		return (NULL);
304	if (af != sp->sa_family || af != dp->sa_family)
305		return (NULL);
306
307	/* check if anyone have already attached with exactly same config */
308	mtx_lock(&encapmtx);
309	LIST_FOREACH(ep, &encaptab, chain) {
310		if (ep->af != af)
311			continue;
312		if (ep->proto != proto)
313			continue;
314		if (ep->src.ss_len != sp->sa_len ||
315		    bcmp(&ep->src, sp, sp->sa_len) != 0 ||
316		    bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
317			continue;
318		if (ep->dst.ss_len != dp->sa_len ||
319		    bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
320		    bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
321			continue;
322
323		mtx_unlock(&encapmtx);
324		return (NULL);
325	}
326
327	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
328	if (ep == NULL) {
329		mtx_unlock(&encapmtx);
330		return (NULL);
331	}
332	bzero(ep, sizeof(*ep));
333
334	ep->af = af;
335	ep->proto = proto;
336	bcopy(sp, &ep->src, sp->sa_len);
337	bcopy(sm, &ep->srcmask, sp->sa_len);
338	bcopy(dp, &ep->dst, dp->sa_len);
339	bcopy(dm, &ep->dstmask, dp->sa_len);
340	ep->psw = psw;
341	ep->arg = arg;
342
343	encap_add(ep);
344	mtx_unlock(&encapmtx);
345	return (ep);
346}
347
348const struct encaptab *
349encap_attach_func(af, proto, func, psw, arg)
350	int af;
351	int proto;
352	int (*func)(const struct mbuf *, int, int, void *);
353	const struct protosw *psw;
354	void *arg;
355{
356	struct encaptab *ep;
357
358	/* sanity check on args */
359	if (!func)
360		return (NULL);
361
362	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
363	if (ep == NULL)
364		return (NULL);
365	bzero(ep, sizeof(*ep));
366
367	ep->af = af;
368	ep->proto = proto;
369	ep->func = func;
370	ep->psw = psw;
371	ep->arg = arg;
372
373	mtx_lock(&encapmtx);
374	encap_add(ep);
375	mtx_unlock(&encapmtx);
376	return (ep);
377}
378
379int
380encap_detach(cookie)
381	const struct encaptab *cookie;
382{
383	const struct encaptab *ep = cookie;
384	struct encaptab *p;
385
386	mtx_lock(&encapmtx);
387	LIST_FOREACH(p, &encaptab, chain) {
388		if (p == ep) {
389			LIST_REMOVE(p, chain);
390			mtx_unlock(&encapmtx);
391			free(p, M_NETADDR);	/*XXX*/
392			return 0;
393		}
394	}
395	mtx_unlock(&encapmtx);
396
397	return EINVAL;
398}
399
400static int
401mask_match(ep, sp, dp)
402	const struct encaptab *ep;
403	const struct sockaddr *sp;
404	const struct sockaddr *dp;
405{
406	struct sockaddr_storage s;
407	struct sockaddr_storage d;
408	int i;
409	const u_int8_t *p, *q;
410	u_int8_t *r;
411	int matchlen;
412
413	if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d))
414		return 0;
415	if (sp->sa_family != ep->af || dp->sa_family != ep->af)
416		return 0;
417	if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
418		return 0;
419
420	matchlen = 0;
421
422	p = (const u_int8_t *)sp;
423	q = (const u_int8_t *)&ep->srcmask;
424	r = (u_int8_t *)&s;
425	for (i = 0 ; i < sp->sa_len; i++) {
426		r[i] = p[i] & q[i];
427		/* XXX estimate */
428		matchlen += (q[i] ? 8 : 0);
429	}
430
431	p = (const u_int8_t *)dp;
432	q = (const u_int8_t *)&ep->dstmask;
433	r = (u_int8_t *)&d;
434	for (i = 0 ; i < dp->sa_len; i++) {
435		r[i] = p[i] & q[i];
436		/* XXX rough estimate */
437		matchlen += (q[i] ? 8 : 0);
438	}
439
440	/* need to overwrite len/family portion as we don't compare them */
441	s.ss_len = sp->sa_len;
442	s.ss_family = sp->sa_family;
443	d.ss_len = dp->sa_len;
444	d.ss_family = dp->sa_family;
445
446	if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
447	    bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
448		return matchlen;
449	} else
450		return 0;
451}
452
453static void
454encap_fillarg(m, ep)
455	struct mbuf *m;
456	const struct encaptab *ep;
457{
458	struct m_tag *tag;
459
460	tag = m_tag_get(PACKET_TAG_ENCAP, sizeof (void*), M_NOWAIT);
461	if (tag) {
462		*(void**)(tag+1) = ep->arg;
463		m_tag_prepend(m, tag);
464	}
465}
466
467void *
468encap_getarg(m)
469	struct mbuf *m;
470{
471	void *p = NULL;
472	struct m_tag *tag;
473
474	tag = m_tag_find(m, PACKET_TAG_ENCAP, NULL);
475	if (tag) {
476		p = *(void**)(tag+1);
477		m_tag_delete(m, tag);
478	}
479	return p;
480}
481