xref: /freebsd/sys/netinet/ip_encap.c (revision 325151a3) 
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 *, void *);
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 static 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 	void *arg;
126 	int matchprio, off, prio;
127 
128 	m = *mp;
129 	off = *offp;
130 	ip = mtod(m, struct ip *);
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 	arg = NULL;
142 	psw = NULL;
143 	match = NULL;
144 	matchprio = 0;
145 	mtx_lock(&encapmtx);
146 	LIST_FOREACH(ep, &encaptab, chain) {
147 		if (ep->af != AF_INET)
148 			continue;
149 		if (ep->proto >= 0 && ep->proto != proto)
150 			continue;
151 		if (ep->func)
152 			prio = (*ep->func)(m, off, proto, ep->arg);
153 		else {
154 			/*
155 			 * it's inbound traffic, we need to match in reverse
156 			 * order
157 			 */
158 			prio = mask_match(ep, (struct sockaddr *)&d,
159 			    (struct sockaddr *)&s);
160 		}
161 
162 		/*
163 		 * We prioritize the matches by using bit length of the
164 		 * matches.  mask_match() and user-supplied matching function
165 		 * should return the bit length of the matches (for example,
166 		 * if both src/dst are matched for IPv4, 64 should be returned).
167 		 * 0 or negative return value means "it did not match".
168 		 *
169 		 * The question is, since we have two "mask" portion, we
170 		 * cannot really define total order between entries.
171 		 * For example, which of these should be preferred?
172 		 * mask_match() returns 48 (32 + 16) for both of them.
173 		 *	src=3ffe::/16, dst=3ffe:501::/32
174 		 *	src=3ffe:501::/32, dst=3ffe::/16
175 		 *
176 		 * We need to loop through all the possible candidates
177 		 * to get the best match - the search takes O(n) for
178 		 * n attachments (i.e. interfaces).
179 		 */
180 		if (prio <= 0)
181 			continue;
182 		if (prio > matchprio) {
183 			matchprio = prio;
184 			match = ep;
185 		}
186 	}
187 	if (match != NULL) {
188 		psw = match->psw;
189 		arg = match->arg;
190 	}
191 	mtx_unlock(&encapmtx);
192 
193 	if (match != NULL) {
194 		/* found a match, "match" has the best one */
195 		if (psw != NULL && psw->pr_input != NULL) {
196 			encap_fillarg(m, arg);
197 			(*psw->pr_input)(mp, offp, proto);
198 		} else
199 			m_freem(m);
200 		return (IPPROTO_DONE);
201 	}
202 
203 	/* last resort: inject to raw socket */
204 	return (rip_input(mp, offp, proto));
205 }
206 #endif
207 
208 #ifdef INET6
209 int
210 encap6_input(struct mbuf **mp, int *offp, int proto)
211 {
212 	struct mbuf *m = *mp;
213 	struct ip6_hdr *ip6;
214 	struct sockaddr_in6 s, d;
215 	const struct protosw *psw;
216 	struct encaptab *ep, *match;
217 	void *arg;
218 	int prio, matchprio;
219 
220 	ip6 = mtod(m, struct ip6_hdr *);
221 
222 	bzero(&s, sizeof(s));
223 	s.sin6_family = AF_INET6;
224 	s.sin6_len = sizeof(struct sockaddr_in6);
225 	s.sin6_addr = ip6->ip6_src;
226 	bzero(&d, sizeof(d));
227 	d.sin6_family = AF_INET6;
228 	d.sin6_len = sizeof(struct sockaddr_in6);
229 	d.sin6_addr = ip6->ip6_dst;
230 
231 	arg = NULL;
232 	psw = NULL;
233 	match = NULL;
234 	matchprio = 0;
235 	mtx_lock(&encapmtx);
236 	LIST_FOREACH(ep, &encaptab, chain) {
237 		if (ep->af != AF_INET6)
238 			continue;
239 		if (ep->proto >= 0 && ep->proto != proto)
240 			continue;
241 		if (ep->func)
242 			prio = (*ep->func)(m, *offp, proto, ep->arg);
243 		else {
244 			/*
245 			 * it's inbound traffic, we need to match in reverse
246 			 * order
247 			 */
248 			prio = mask_match(ep, (struct sockaddr *)&d,
249 			    (struct sockaddr *)&s);
250 		}
251 
252 		/* see encap4_input() for issues here */
253 		if (prio <= 0)
254 			continue;
255 		if (prio > matchprio) {
256 			matchprio = prio;
257 			match = ep;
258 		}
259 	}
260 	if (match != NULL) {
261 		psw = match->psw;
262 		arg = match->arg;
263 	}
264 	mtx_unlock(&encapmtx);
265 
266 	if (match != NULL) {
267 		/* found a match */
268 		if (psw != NULL && psw->pr_input != NULL) {
269 			encap_fillarg(m, arg);
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 
282 /*lint -sem(encap_add, custodial(1)) */
283 static void
284 encap_add(struct encaptab *ep)
285 {
286 
287 	mtx_assert(&encapmtx, MA_OWNED);
288 	LIST_INSERT_HEAD(&encaptab, ep, chain);
289 }
290 
291 /*
292  * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
293  * length of mask (sm and dm) is assumed to be same as sp/dp.
294  * Return value will be necessary as input (cookie) for encap_detach().
295  */
296 const struct encaptab *
297 encap_attach(int af, int proto, const struct sockaddr *sp,
298     const struct sockaddr *sm, const struct sockaddr *dp,
299     const struct sockaddr *dm, const struct protosw *psw, void *arg)
300 {
301 	struct encaptab *ep;
302 
303 	/* sanity check on args */
304 	if (sp->sa_len > sizeof(ep->src) || dp->sa_len > sizeof(ep->dst))
305 		return (NULL);
306 	if (sp->sa_len != dp->sa_len)
307 		return (NULL);
308 	if (af != sp->sa_family || af != dp->sa_family)
309 		return (NULL);
310 
311 	/* check if anyone have already attached with exactly same config */
312 	mtx_lock(&encapmtx);
313 	LIST_FOREACH(ep, &encaptab, chain) {
314 		if (ep->af != af)
315 			continue;
316 		if (ep->proto != proto)
317 			continue;
318 		if (ep->src.ss_len != sp->sa_len ||
319 		    bcmp(&ep->src, sp, sp->sa_len) != 0 ||
320 		    bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
321 			continue;
322 		if (ep->dst.ss_len != dp->sa_len ||
323 		    bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
324 		    bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
325 			continue;
326 
327 		mtx_unlock(&encapmtx);
328 		return (NULL);
329 	}
330 
331 	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
332 	if (ep == NULL) {
333 		mtx_unlock(&encapmtx);
334 		return (NULL);
335 	}
336 	bzero(ep, sizeof(*ep));
337 
338 	ep->af = af;
339 	ep->proto = proto;
340 	bcopy(sp, &ep->src, sp->sa_len);
341 	bcopy(sm, &ep->srcmask, sp->sa_len);
342 	bcopy(dp, &ep->dst, dp->sa_len);
343 	bcopy(dm, &ep->dstmask, dp->sa_len);
344 	ep->psw = psw;
345 	ep->arg = arg;
346 
347 	encap_add(ep);
348 	mtx_unlock(&encapmtx);
349 	return (ep);
350 }
351 
352 const struct encaptab *
353 encap_attach_func(int af, int proto,
354     int (*func)(const struct mbuf *, int, int, void *),
355     const struct protosw *psw, void *arg)
356 {
357 	struct encaptab *ep;
358 
359 	/* sanity check on args */
360 	if (!func)
361 		return (NULL);
362 
363 	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
364 	if (ep == NULL)
365 		return (NULL);
366 	bzero(ep, sizeof(*ep));
367 
368 	ep->af = af;
369 	ep->proto = proto;
370 	ep->func = func;
371 	ep->psw = psw;
372 	ep->arg = arg;
373 
374 	mtx_lock(&encapmtx);
375 	encap_add(ep);
376 	mtx_unlock(&encapmtx);
377 	return (ep);
378 }
379 
380 int
381 encap_detach(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 
400 static int
401 mask_match(const struct encaptab *ep, const struct sockaddr *sp,
402     const struct sockaddr *dp)
403 {
404 	struct sockaddr_storage s;
405 	struct sockaddr_storage d;
406 	int i;
407 	const u_int8_t *p, *q;
408 	u_int8_t *r;
409 	int matchlen;
410 
411 	if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d))
412 		return 0;
413 	if (sp->sa_family != ep->af || dp->sa_family != ep->af)
414 		return 0;
415 	if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
416 		return 0;
417 
418 	matchlen = 0;
419 
420 	p = (const u_int8_t *)sp;
421 	q = (const u_int8_t *)&ep->srcmask;
422 	r = (u_int8_t *)&s;
423 	for (i = 0 ; i < sp->sa_len; i++) {
424 		r[i] = p[i] & q[i];
425 		/* XXX estimate */
426 		matchlen += (q[i] ? 8 : 0);
427 	}
428 
429 	p = (const u_int8_t *)dp;
430 	q = (const u_int8_t *)&ep->dstmask;
431 	r = (u_int8_t *)&d;
432 	for (i = 0 ; i < dp->sa_len; i++) {
433 		r[i] = p[i] & q[i];
434 		/* XXX rough estimate */
435 		matchlen += (q[i] ? 8 : 0);
436 	}
437 
438 	/* need to overwrite len/family portion as we don't compare them */
439 	s.ss_len = sp->sa_len;
440 	s.ss_family = sp->sa_family;
441 	d.ss_len = dp->sa_len;
442 	d.ss_family = dp->sa_family;
443 
444 	if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
445 	    bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
446 		return matchlen;
447 	} else
448 		return 0;
449 }
450 
451 static void
452 encap_fillarg(struct mbuf *m, void *arg)
453 {
454 	struct m_tag *tag;
455 
456 	if (arg != NULL) {
457 		tag = m_tag_get(PACKET_TAG_ENCAP, sizeof(void *), M_NOWAIT);
458 		if (tag != NULL) {
459 			*(void**)(tag+1) = arg;
460 			m_tag_prepend(m, tag);
461 		}
462 	}
463 }
464 
465 void *
466 encap_getarg(struct mbuf *m)
467 {
468 	void *p = NULL;
469 	struct m_tag *tag;
470 
471 	tag = m_tag_find(m, PACKET_TAG_ENCAP, NULL);
472 	if (tag) {
473 		p = *(void**)(tag+1);
474 		m_tag_delete(m, tag);
475 	}
476 	return p;
477 }
478