ip_encap.c revision 112148 
1/*	$FreeBSD: head/sys/netinet/ip_encap.c 112148 2003-03-12 14:45:22Z sam $	*/
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 <net/net_osdep.h>
91
92#include <sys/kernel.h>
93#include <sys/malloc.h>
94static MALLOC_DEFINE(M_NETADDR, "Export Host", "Export host address structure");
95
96static void encap_add(struct encaptab *);
97static int mask_match(const struct encaptab *, const struct sockaddr *,
98		const struct sockaddr *);
99static void encap_fillarg(struct mbuf *, const struct encaptab *);
100
101#ifndef LIST_HEAD_INITIALIZER
102/* rely upon BSS initialization */
103LIST_HEAD(, encaptab) encaptab;
104#else
105LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(&encaptab);
106#endif
107
108void
109encap_init()
110{
111	static int initialized = 0;
112
113	if (initialized)
114		return;
115	initialized++;
116#if 0
117	/*
118	 * we cannot use LIST_INIT() here, since drivers may want to call
119	 * encap_attach(), on driver attach.  encap_init() will be called
120	 * on AF_INET{,6} initialization, which happens after driver
121	 * initialization - using LIST_INIT() here can nuke encap_attach()
122	 * from drivers.
123	 */
124	LIST_INIT(&encaptab);
125#endif
126}
127
128#ifdef INET
129void
130encap4_input(m, off)
131	struct mbuf *m;
132	int off;
133{
134	struct ip *ip;
135	int proto;
136	struct sockaddr_in s, d;
137	const struct protosw *psw;
138	struct encaptab *ep, *match;
139	int prio, matchprio;
140
141	ip = mtod(m, struct ip *);
142	proto = ip->ip_p;
143
144	bzero(&s, sizeof(s));
145	s.sin_family = AF_INET;
146	s.sin_len = sizeof(struct sockaddr_in);
147	s.sin_addr = ip->ip_src;
148	bzero(&d, sizeof(d));
149	d.sin_family = AF_INET;
150	d.sin_len = sizeof(struct sockaddr_in);
151	d.sin_addr = ip->ip_dst;
152
153	match = NULL;
154	matchprio = 0;
155	LIST_FOREACH(ep, &encaptab, chain) {
156		if (ep->af != AF_INET)
157			continue;
158		if (ep->proto >= 0 && ep->proto != proto)
159			continue;
160		if (ep->func)
161			prio = (*ep->func)(m, off, proto, ep->arg);
162		else {
163			/*
164			 * it's inbound traffic, we need to match in reverse
165			 * order
166			 */
167			prio = mask_match(ep, (struct sockaddr *)&d,
168			    (struct sockaddr *)&s);
169		}
170
171		/*
172		 * We prioritize the matches by using bit length of the
173		 * matches.  mask_match() and user-supplied matching function
174		 * should return the bit length of the matches (for example,
175		 * if both src/dst are matched for IPv4, 64 should be returned).
176		 * 0 or negative return value means "it did not match".
177		 *
178		 * The question is, since we have two "mask" portion, we
179		 * cannot really define total order between entries.
180		 * For example, which of these should be preferred?
181		 * mask_match() returns 48 (32 + 16) for both of them.
182		 *	src=3ffe::/16, dst=3ffe:501::/32
183		 *	src=3ffe:501::/32, dst=3ffe::/16
184		 *
185		 * We need to loop through all the possible candidates
186		 * to get the best match - the search takes O(n) for
187		 * n attachments (i.e. interfaces).
188		 */
189		if (prio <= 0)
190			continue;
191		if (prio > matchprio) {
192			matchprio = prio;
193			match = ep;
194		}
195	}
196
197	if (match) {
198		/* found a match, "match" has the best one */
199		psw = match->psw;
200		if (psw && psw->pr_input) {
201			encap_fillarg(m, match);
202			(*psw->pr_input)(m, off);
203		} else
204			m_freem(m);
205		return;
206	}
207
208	/* last resort: inject to raw socket */
209	rip_input(m, off);
210}
211#endif
212
213#ifdef INET6
214int
215encap6_input(mp, offp, proto)
216	struct mbuf **mp;
217	int *offp;
218	int proto;
219{
220	struct mbuf *m = *mp;
221	struct ip6_hdr *ip6;
222	struct sockaddr_in6 s, d;
223	const struct ip6protosw *psw;
224	struct encaptab *ep, *match;
225	int prio, matchprio;
226
227	ip6 = mtod(m, struct ip6_hdr *);
228
229	bzero(&s, sizeof(s));
230	s.sin6_family = AF_INET6;
231	s.sin6_len = sizeof(struct sockaddr_in6);
232	s.sin6_addr = ip6->ip6_src;
233	bzero(&d, sizeof(d));
234	d.sin6_family = AF_INET6;
235	d.sin6_len = sizeof(struct sockaddr_in6);
236	d.sin6_addr = ip6->ip6_dst;
237
238	match = NULL;
239	matchprio = 0;
240	LIST_FOREACH(ep, &encaptab, chain) {
241		if (ep->af != AF_INET6)
242			continue;
243		if (ep->proto >= 0 && ep->proto != proto)
244			continue;
245		if (ep->func)
246			prio = (*ep->func)(m, *offp, proto, ep->arg);
247		else {
248			/*
249			 * it's inbound traffic, we need to match in reverse
250			 * order
251			 */
252			prio = mask_match(ep, (struct sockaddr *)&d,
253			    (struct sockaddr *)&s);
254		}
255
256		/* see encap4_input() for issues here */
257		if (prio <= 0)
258			continue;
259		if (prio > matchprio) {
260			matchprio = prio;
261			match = ep;
262		}
263	}
264
265	if (match) {
266		/* found a match */
267		psw = (const struct ip6protosw *)match->psw;
268		if (psw && psw->pr_input) {
269			encap_fillarg(m, match);
270			return (*psw->pr_input)(mp, offp, proto);
271		} else {
272			m_freem(m);
273			return IPPROTO_DONE;
274		}
275	}
276
277	/* last resort: inject to raw socket */
278	return rip6_input(mp, offp, proto);
279}
280#endif
281
282static void
283encap_add(ep)
284	struct encaptab *ep;
285{
286
287	LIST_INSERT_HEAD(&encaptab, ep, chain);
288}
289
290/*
291 * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
292 * length of mask (sm and dm) is assumed to be same as sp/dp.
293 * Return value will be necessary as input (cookie) for encap_detach().
294 */
295const struct encaptab *
296encap_attach(af, proto, sp, sm, dp, dm, psw, arg)
297	int af;
298	int proto;
299	const struct sockaddr *sp, *sm;
300	const struct sockaddr *dp, *dm;
301	const struct protosw *psw;
302	void *arg;
303{
304	struct encaptab *ep;
305	int error;
306	int s;
307
308	s = splnet();
309	/* sanity check on args */
310	if (sp->sa_len > sizeof(ep->src) || dp->sa_len > sizeof(ep->dst)) {
311		error = EINVAL;
312		goto fail;
313	}
314	if (sp->sa_len != dp->sa_len) {
315		error = EINVAL;
316		goto fail;
317	}
318	if (af != sp->sa_family || af != dp->sa_family) {
319		error = EINVAL;
320		goto fail;
321	}
322
323	/* check if anyone have already attached with exactly same config */
324	LIST_FOREACH(ep, &encaptab, chain) {
325		if (ep->af != af)
326			continue;
327		if (ep->proto != proto)
328			continue;
329		if (ep->src.ss_len != sp->sa_len ||
330		    bcmp(&ep->src, sp, sp->sa_len) != 0 ||
331		    bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
332			continue;
333		if (ep->dst.ss_len != dp->sa_len ||
334		    bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
335		    bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
336			continue;
337
338		error = EEXIST;
339		goto fail;
340	}
341
342	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
343	if (ep == NULL) {
344		error = ENOBUFS;
345		goto fail;
346	}
347	bzero(ep, sizeof(*ep));
348
349	ep->af = af;
350	ep->proto = proto;
351	bcopy(sp, &ep->src, sp->sa_len);
352	bcopy(sm, &ep->srcmask, sp->sa_len);
353	bcopy(dp, &ep->dst, dp->sa_len);
354	bcopy(dm, &ep->dstmask, dp->sa_len);
355	ep->psw = psw;
356	ep->arg = arg;
357
358	encap_add(ep);
359
360	error = 0;
361	splx(s);
362	return ep;
363
364fail:
365	splx(s);
366	return NULL;
367}
368
369const struct encaptab *
370encap_attach_func(af, proto, func, psw, arg)
371	int af;
372	int proto;
373	int (*func)(const struct mbuf *, int, int, void *);
374	const struct protosw *psw;
375	void *arg;
376{
377	struct encaptab *ep;
378	int error;
379	int s;
380
381	s = splnet();
382	/* sanity check on args */
383	if (!func) {
384		error = EINVAL;
385		goto fail;
386	}
387
388	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
389	if (ep == NULL) {
390		error = ENOBUFS;
391		goto fail;
392	}
393	bzero(ep, sizeof(*ep));
394
395	ep->af = af;
396	ep->proto = proto;
397	ep->func = func;
398	ep->psw = psw;
399	ep->arg = arg;
400
401	encap_add(ep);
402
403	error = 0;
404	splx(s);
405	return ep;
406
407fail:
408	splx(s);
409	return NULL;
410}
411
412int
413encap_detach(cookie)
414	const struct encaptab *cookie;
415{
416	const struct encaptab *ep = cookie;
417	struct encaptab *p;
418
419	LIST_FOREACH(p, &encaptab, chain) {
420		if (p == ep) {
421			LIST_REMOVE(p, chain);
422			free(p, M_NETADDR);	/*XXX*/
423			return 0;
424		}
425	}
426
427	return EINVAL;
428}
429
430static int
431mask_match(ep, sp, dp)
432	const struct encaptab *ep;
433	const struct sockaddr *sp;
434	const struct sockaddr *dp;
435{
436	struct sockaddr_storage s;
437	struct sockaddr_storage d;
438	int i;
439	const u_int8_t *p, *q;
440	u_int8_t *r;
441	int matchlen;
442
443	if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d))
444		return 0;
445	if (sp->sa_family != ep->af || dp->sa_family != ep->af)
446		return 0;
447	if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
448		return 0;
449
450	matchlen = 0;
451
452	p = (const u_int8_t *)sp;
453	q = (const u_int8_t *)&ep->srcmask;
454	r = (u_int8_t *)&s;
455	for (i = 0 ; i < sp->sa_len; i++) {
456		r[i] = p[i] & q[i];
457		/* XXX estimate */
458		matchlen += (q[i] ? 8 : 0);
459	}
460
461	p = (const u_int8_t *)dp;
462	q = (const u_int8_t *)&ep->dstmask;
463	r = (u_int8_t *)&d;
464	for (i = 0 ; i < dp->sa_len; i++) {
465		r[i] = p[i] & q[i];
466		/* XXX rough estimate */
467		matchlen += (q[i] ? 8 : 0);
468	}
469
470	/* need to overwrite len/family portion as we don't compare them */
471	s.ss_len = sp->sa_len;
472	s.ss_family = sp->sa_family;
473	d.ss_len = dp->sa_len;
474	d.ss_family = dp->sa_family;
475
476	if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
477	    bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
478		return matchlen;
479	} else
480		return 0;
481}
482
483static void
484encap_fillarg(m, ep)
485	struct mbuf *m;
486	const struct encaptab *ep;
487{
488	struct m_tag *tag;
489
490	tag = m_tag_get(PACKET_TAG_ENCAP, sizeof (void*), M_NOWAIT);
491	if (tag) {
492		*(void**)(tag+1) = ep->arg;
493		m_tag_prepend(m, tag);
494	}
495}
496
497void *
498encap_getarg(m)
499	struct mbuf *m;
500{
501	void *p = NULL;
502	struct m_tag *tag;
503
504	tag = m_tag_find(m, PACKET_TAG_ENCAP, NULL);
505	if (tag) {
506		p = *(void**)(tag+1);
507		m_tag_delete(m, tag);
508	}
509	return p;
510}
511