xref: /freebsd/sys/netinet/ip_encap.c (revision 5b9c547c) 
1 /*	$KAME: ip_encap.c,v 1.41 2001/03/15 08:35:08 itojun Exp $	*/
2 
3 /*-
4  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the project nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 /*
32  * My grandfather said that there's a devil inside tunnelling technology...
33  *
34  * We have surprisingly many protocols that want packets with IP protocol
35  * #4 or #41.  Here's a list of protocols that want protocol #41:
36  *	RFC1933 configured tunnel
37  *	RFC1933 automatic tunnel
38  *	RFC2401 IPsec tunnel
39  *	RFC2473 IPv6 generic packet tunnelling
40  *	RFC2529 6over4 tunnel
41  *	mobile-ip6 (uses RFC2473)
42  *	RFC3056 6to4 tunnel
43  *	isatap 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 <sys/cdefs.h>
60 __FBSDID("$FreeBSD$");
61 
62 #include "opt_mrouting.h"
63 #include "opt_inet.h"
64 #include "opt_inet6.h"
65 
66 #include <sys/param.h>
67 #include <sys/systm.h>
68 #include <sys/socket.h>
69 #include <sys/sockio.h>
70 #include <sys/mbuf.h>
71 #include <sys/errno.h>
72 #include <sys/protosw.h>
73 #include <sys/queue.h>
74 
75 #include <net/if.h>
76 #include <net/route.h>
77 
78 #include <netinet/in.h>
79 #include <netinet/in_systm.h>
80 #include <netinet/ip.h>
81 #include <netinet/ip_var.h>
82 #include <netinet/ip_encap.h>
83 
84 #ifdef INET6
85 #include <netinet/ip6.h>
86 #include <netinet6/ip6_var.h>
87 #endif
88 
89 #include <machine/stdarg.h>
90 
91 #include <sys/kernel.h>
92 #include <sys/malloc.h>
93 static MALLOC_DEFINE(M_NETADDR, "encap_export_host", "Export host address structure");
94 
95 static void encap_add(struct encaptab *);
96 static int mask_match(const struct encaptab *, const struct sockaddr *,
97 		const struct sockaddr *);
98 static void encap_fillarg(struct mbuf *, const struct encaptab *);
99 
100 /*
101  * All global variables in ip_encap.c are locked using encapmtx.
102  */
103 static struct mtx encapmtx;
104 MTX_SYSINIT(encapmtx, &encapmtx, "encapmtx", MTX_DEF);
105 LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(encaptab);
106 
107 /*
108  * We currently keey encap_init() for source code compatibility reasons --
109  * it's referenced by KAME pieces in netinet6.
110  */
111 void
112 encap_init(void)
113 {
114 }
115 
116 #ifdef INET
117 int
118 encap4_input(struct mbuf **mp, int *offp, int proto)
119 {
120 	struct ip *ip;
121 	struct mbuf *m;
122 	struct sockaddr_in s, d;
123 	const struct protosw *psw;
124 	struct encaptab *ep, *match;
125 	int matchprio, off, prio;
126 
127 	m = *mp;
128 	off = *offp;
129 	ip = mtod(m, struct ip *);
130 	*mp = NULL;
131 
132 	bzero(&s, sizeof(s));
133 	s.sin_family = AF_INET;
134 	s.sin_len = sizeof(struct sockaddr_in);
135 	s.sin_addr = ip->ip_src;
136 	bzero(&d, sizeof(d));
137 	d.sin_family = AF_INET;
138 	d.sin_len = sizeof(struct sockaddr_in);
139 	d.sin_addr = ip->ip_dst;
140 
141 	match = NULL;
142 	matchprio = 0;
143 	mtx_lock(&encapmtx);
144 	LIST_FOREACH(ep, &encaptab, chain) {
145 		if (ep->af != AF_INET)
146 			continue;
147 		if (ep->proto >= 0 && ep->proto != proto)
148 			continue;
149 		if (ep->func)
150 			prio = (*ep->func)(m, off, proto, ep->arg);
151 		else {
152 			/*
153 			 * it's inbound traffic, we need to match in reverse
154 			 * order
155 			 */
156 			prio = mask_match(ep, (struct sockaddr *)&d,
157 			    (struct sockaddr *)&s);
158 		}
159 
160 		/*
161 		 * We prioritize the matches by using bit length of the
162 		 * matches.  mask_match() and user-supplied matching function
163 		 * should return the bit length of the matches (for example,
164 		 * if both src/dst are matched for IPv4, 64 should be returned).
165 		 * 0 or negative return value means "it did not match".
166 		 *
167 		 * The question is, since we have two "mask" portion, we
168 		 * cannot really define total order between entries.
169 		 * For example, which of these should be preferred?
170 		 * mask_match() returns 48 (32 + 16) for both of them.
171 		 *	src=3ffe::/16, dst=3ffe:501::/32
172 		 *	src=3ffe:501::/32, dst=3ffe::/16
173 		 *
174 		 * We need to loop through all the possible candidates
175 		 * to get the best match - the search takes O(n) for
176 		 * n attachments (i.e. interfaces).
177 		 */
178 		if (prio <= 0)
179 			continue;
180 		if (prio > matchprio) {
181 			matchprio = prio;
182 			match = ep;
183 		}
184 	}
185 	mtx_unlock(&encapmtx);
186 
187 	if (match) {
188 		/* found a match, "match" has the best one */
189 		psw = match->psw;
190 		if (psw && psw->pr_input) {
191 			encap_fillarg(m, match);
192 			*mp = m;
193 			(*psw->pr_input)(mp, offp, proto);
194 		} else
195 			m_freem(m);
196 		return (IPPROTO_DONE);
197 	}
198 
199 	/* last resort: inject to raw socket */
200 	*mp = m;
201 	return (rip_input(mp, offp, proto));
202 }
203 #endif
204 
205 #ifdef INET6
206 int
207 encap6_input(struct mbuf **mp, int *offp, int proto)
208 {
209 	struct mbuf *m = *mp;
210 	struct ip6_hdr *ip6;
211 	struct sockaddr_in6 s, d;
212 	const struct protosw *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 = 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)) */
274 static void
275 encap_add(struct encaptab *ep)
276 {
277 
278 	mtx_assert(&encapmtx, MA_OWNED);
279 	LIST_INSERT_HEAD(&encaptab, ep, chain);
280 }
281 
282 /*
283  * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
284  * length of mask (sm and dm) is assumed to be same as sp/dp.
285  * Return value will be necessary as input (cookie) for encap_detach().
286  */
287 const struct encaptab *
288 encap_attach(int af, int proto, const struct sockaddr *sp,
289     const struct sockaddr *sm, const struct sockaddr *dp,
290     const struct sockaddr *dm, const struct protosw *psw, void *arg)
291 {
292 	struct encaptab *ep;
293 
294 	/* sanity check on args */
295 	if (sp->sa_len > sizeof(ep->src) || dp->sa_len > sizeof(ep->dst))
296 		return (NULL);
297 	if (sp->sa_len != dp->sa_len)
298 		return (NULL);
299 	if (af != sp->sa_family || af != dp->sa_family)
300 		return (NULL);
301 
302 	/* check if anyone have already attached with exactly same config */
303 	mtx_lock(&encapmtx);
304 	LIST_FOREACH(ep, &encaptab, chain) {
305 		if (ep->af != af)
306 			continue;
307 		if (ep->proto != proto)
308 			continue;
309 		if (ep->src.ss_len != sp->sa_len ||
310 		    bcmp(&ep->src, sp, sp->sa_len) != 0 ||
311 		    bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
312 			continue;
313 		if (ep->dst.ss_len != dp->sa_len ||
314 		    bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
315 		    bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
316 			continue;
317 
318 		mtx_unlock(&encapmtx);
319 		return (NULL);
320 	}
321 
322 	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
323 	if (ep == NULL) {
324 		mtx_unlock(&encapmtx);
325 		return (NULL);
326 	}
327 	bzero(ep, sizeof(*ep));
328 
329 	ep->af = af;
330 	ep->proto = proto;
331 	bcopy(sp, &ep->src, sp->sa_len);
332 	bcopy(sm, &ep->srcmask, sp->sa_len);
333 	bcopy(dp, &ep->dst, dp->sa_len);
334 	bcopy(dm, &ep->dstmask, dp->sa_len);
335 	ep->psw = psw;
336 	ep->arg = arg;
337 
338 	encap_add(ep);
339 	mtx_unlock(&encapmtx);
340 	return (ep);
341 }
342 
343 const struct encaptab *
344 encap_attach_func(int af, int proto,
345     int (*func)(const struct mbuf *, int, int, void *),
346     const struct protosw *psw, void *arg)
347 {
348 	struct encaptab *ep;
349 
350 	/* sanity check on args */
351 	if (!func)
352 		return (NULL);
353 
354 	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
355 	if (ep == NULL)
356 		return (NULL);
357 	bzero(ep, sizeof(*ep));
358 
359 	ep->af = af;
360 	ep->proto = proto;
361 	ep->func = func;
362 	ep->psw = psw;
363 	ep->arg = arg;
364 
365 	mtx_lock(&encapmtx);
366 	encap_add(ep);
367 	mtx_unlock(&encapmtx);
368 	return (ep);
369 }
370 
371 int
372 encap_detach(const struct encaptab *cookie)
373 {
374 	const struct encaptab *ep = cookie;
375 	struct encaptab *p;
376 
377 	mtx_lock(&encapmtx);
378 	LIST_FOREACH(p, &encaptab, chain) {
379 		if (p == ep) {
380 			LIST_REMOVE(p, chain);
381 			mtx_unlock(&encapmtx);
382 			free(p, M_NETADDR);	/*XXX*/
383 			return 0;
384 		}
385 	}
386 	mtx_unlock(&encapmtx);
387 
388 	return EINVAL;
389 }
390 
391 static int
392 mask_match(const struct encaptab *ep, const struct sockaddr *sp,
393     const struct sockaddr *dp)
394 {
395 	struct sockaddr_storage s;
396 	struct sockaddr_storage d;
397 	int i;
398 	const u_int8_t *p, *q;
399 	u_int8_t *r;
400 	int matchlen;
401 
402 	if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d))
403 		return 0;
404 	if (sp->sa_family != ep->af || dp->sa_family != ep->af)
405 		return 0;
406 	if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
407 		return 0;
408 
409 	matchlen = 0;
410 
411 	p = (const u_int8_t *)sp;
412 	q = (const u_int8_t *)&ep->srcmask;
413 	r = (u_int8_t *)&s;
414 	for (i = 0 ; i < sp->sa_len; i++) {
415 		r[i] = p[i] & q[i];
416 		/* XXX estimate */
417 		matchlen += (q[i] ? 8 : 0);
418 	}
419 
420 	p = (const u_int8_t *)dp;
421 	q = (const u_int8_t *)&ep->dstmask;
422 	r = (u_int8_t *)&d;
423 	for (i = 0 ; i < dp->sa_len; i++) {
424 		r[i] = p[i] & q[i];
425 		/* XXX rough estimate */
426 		matchlen += (q[i] ? 8 : 0);
427 	}
428 
429 	/* need to overwrite len/family portion as we don't compare them */
430 	s.ss_len = sp->sa_len;
431 	s.ss_family = sp->sa_family;
432 	d.ss_len = dp->sa_len;
433 	d.ss_family = dp->sa_family;
434 
435 	if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
436 	    bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
437 		return matchlen;
438 	} else
439 		return 0;
440 }
441 
442 static void
443 encap_fillarg(struct mbuf *m, const struct encaptab *ep)
444 {
445 	struct m_tag *tag;
446 
447 	tag = m_tag_get(PACKET_TAG_ENCAP, sizeof (void*), M_NOWAIT);
448 	if (tag) {
449 		*(void**)(tag+1) = ep->arg;
450 		m_tag_prepend(m, tag);
451 	}
452 }
453 
454 void *
455 encap_getarg(struct mbuf *m)
456 {
457 	void *p = NULL;
458 	struct m_tag *tag;
459 
460 	tag = m_tag_find(m, PACKET_TAG_ENCAP, NULL);
461 	if (tag) {
462 		p = *(void**)(tag+1);
463 		m_tag_delete(m, tag);
464 	}
465 	return p;
466 }
467