xref: /freebsd/sys/netinet/ip_encap.c (revision 7d958cc9aa0aa5303daef28fbba829197b580d5e) 
1 /*	$KAME: ip_encap.c,v 1.41 2001/03/15 08:35:08 itojun Exp $	*/
2 
3 /*-
4  * SPDX-License-Identifier: BSD-3-Clause
5  *
6  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
7  * All rights reserved.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. Neither the name of the project nor the names of its contributors
18  *    may be used to endorse or promote products derived from this software
19  *    without specific prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31  * SUCH DAMAGE.
32  */
33 /*
34  * My grandfather said that there's a devil inside tunnelling technology...
35  *
36  * We have surprisingly many protocols that want packets with IP protocol
37  * #4 or #41.  Here's a list of protocols that want protocol #41:
38  *	RFC1933 configured tunnel
39  *	RFC1933 automatic tunnel
40  *	RFC2401 IPsec tunnel
41  *	RFC2473 IPv6 generic packet tunnelling
42  *	RFC2529 6over4 tunnel
43  *	mobile-ip6 (uses RFC2473)
44  *	RFC3056 6to4 tunnel
45  *	isatap tunnel
46  * Here's a list of protocol that want protocol #4:
47  *	RFC1853 IPv4-in-IPv4 tunnelling
48  *	RFC2003 IPv4 encapsulation within IPv4
49  *	RFC2344 reverse tunnelling for mobile-ip4
50  *	RFC2401 IPsec tunnel
51  * Well, what can I say.  They impose different en/decapsulation mechanism
52  * from each other, so they need separate protocol handler.  The only one
53  * we can easily determine by protocol # is IPsec, which always has
54  * AH/ESP/IPComp header right after outer IP header.
55  *
56  * So, clearly good old protosw does not work for protocol #4 and #41.
57  * The code will let you match protocol via src/dst address pair.
58  */
59 /* XXX is M_NETADDR correct? */
60 
61 #include <sys/cdefs.h>
62 __FBSDID("$FreeBSD$");
63 
64 #include "opt_mrouting.h"
65 #include "opt_inet.h"
66 #include "opt_inet6.h"
67 
68 #include <sys/param.h>
69 #include <sys/systm.h>
70 #include <sys/lock.h>
71 #include <sys/mutex.h>
72 #include <sys/socket.h>
73 #include <sys/sockio.h>
74 #include <sys/mbuf.h>
75 #include <sys/errno.h>
76 #include <sys/protosw.h>
77 #include <sys/queue.h>
78 
79 #include <net/if.h>
80 #include <net/route.h>
81 
82 #include <netinet/in.h>
83 #include <netinet/in_systm.h>
84 #include <netinet/ip.h>
85 #include <netinet/ip_var.h>
86 #include <netinet/ip_encap.h>
87 
88 #ifdef INET6
89 #include <netinet/ip6.h>
90 #include <netinet6/ip6_var.h>
91 #endif
92 
93 #include <machine/stdarg.h>
94 
95 #include <sys/kernel.h>
96 #include <sys/malloc.h>
97 static MALLOC_DEFINE(M_NETADDR, "encap_export_host", "Export host address structure");
98 
99 static void encap_add(struct encaptab *);
100 static int mask_match(const struct encaptab *, const struct sockaddr *,
101 		const struct sockaddr *);
102 static void encap_fillarg(struct mbuf *, void *);
103 
104 /*
105  * All global variables in ip_encap.c are locked using encapmtx.
106  */
107 static struct mtx encapmtx;
108 MTX_SYSINIT(encapmtx, &encapmtx, "encapmtx", MTX_DEF);
109 static LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(encaptab);
110 
111 #ifdef INET
112 int
113 encap4_input(struct mbuf **mp, int *offp, int proto)
114 {
115 	struct ip *ip;
116 	struct mbuf *m;
117 	struct sockaddr_in s, d;
118 	const struct protosw *psw;
119 	struct encaptab *ep, *match;
120 	void *arg;
121 	int matchprio, off, prio;
122 
123 	m = *mp;
124 	off = *offp;
125 	ip = mtod(m, struct ip *);
126 
127 	bzero(&s, sizeof(s));
128 	s.sin_family = AF_INET;
129 	s.sin_len = sizeof(struct sockaddr_in);
130 	s.sin_addr = ip->ip_src;
131 	bzero(&d, sizeof(d));
132 	d.sin_family = AF_INET;
133 	d.sin_len = sizeof(struct sockaddr_in);
134 	d.sin_addr = ip->ip_dst;
135 
136 	arg = NULL;
137 	psw = NULL;
138 	match = NULL;
139 	matchprio = 0;
140 	mtx_lock(&encapmtx);
141 	LIST_FOREACH(ep, &encaptab, chain) {
142 		if (ep->af != AF_INET)
143 			continue;
144 		if (ep->proto >= 0 && ep->proto != proto)
145 			continue;
146 		if (ep->func)
147 			prio = (*ep->func)(m, off, proto, ep->arg);
148 		else {
149 			/*
150 			 * it's inbound traffic, we need to match in reverse
151 			 * order
152 			 */
153 			prio = mask_match(ep, (struct sockaddr *)&d,
154 			    (struct sockaddr *)&s);
155 		}
156 
157 		/*
158 		 * We prioritize the matches by using bit length of the
159 		 * matches.  mask_match() and user-supplied matching function
160 		 * should return the bit length of the matches (for example,
161 		 * if both src/dst are matched for IPv4, 64 should be returned).
162 		 * 0 or negative return value means "it did not match".
163 		 *
164 		 * The question is, since we have two "mask" portion, we
165 		 * cannot really define total order between entries.
166 		 * For example, which of these should be preferred?
167 		 * mask_match() returns 48 (32 + 16) for both of them.
168 		 *	src=3ffe::/16, dst=3ffe:501::/32
169 		 *	src=3ffe:501::/32, dst=3ffe::/16
170 		 *
171 		 * We need to loop through all the possible candidates
172 		 * to get the best match - the search takes O(n) for
173 		 * n attachments (i.e. interfaces).
174 		 */
175 		if (prio <= 0)
176 			continue;
177 		if (prio > matchprio) {
178 			matchprio = prio;
179 			match = ep;
180 		}
181 	}
182 	if (match != NULL) {
183 		psw = match->psw;
184 		arg = match->arg;
185 	}
186 	mtx_unlock(&encapmtx);
187 
188 	if (match != NULL) {
189 		/* found a match, "match" has the best one */
190 		if (psw != NULL && psw->pr_input != NULL) {
191 			encap_fillarg(m, arg);
192 			(*psw->pr_input)(mp, offp, proto);
193 		} else
194 			m_freem(m);
195 		return (IPPROTO_DONE);
196 	}
197 
198 	/* last resort: inject to raw socket */
199 	return (rip_input(mp, offp, proto));
200 }
201 #endif
202 
203 #ifdef INET6
204 int
205 encap6_input(struct mbuf **mp, int *offp, int proto)
206 {
207 	struct mbuf *m = *mp;
208 	struct ip6_hdr *ip6;
209 	struct sockaddr_in6 s, d;
210 	const struct protosw *psw;
211 	struct encaptab *ep, *match;
212 	void *arg;
213 	int prio, matchprio;
214 
215 	ip6 = mtod(m, struct ip6_hdr *);
216 
217 	bzero(&s, sizeof(s));
218 	s.sin6_family = AF_INET6;
219 	s.sin6_len = sizeof(struct sockaddr_in6);
220 	s.sin6_addr = ip6->ip6_src;
221 	bzero(&d, sizeof(d));
222 	d.sin6_family = AF_INET6;
223 	d.sin6_len = sizeof(struct sockaddr_in6);
224 	d.sin6_addr = ip6->ip6_dst;
225 
226 	arg = NULL;
227 	psw = NULL;
228 	match = NULL;
229 	matchprio = 0;
230 	mtx_lock(&encapmtx);
231 	LIST_FOREACH(ep, &encaptab, chain) {
232 		if (ep->af != AF_INET6)
233 			continue;
234 		if (ep->proto >= 0 && ep->proto != proto)
235 			continue;
236 		if (ep->func)
237 			prio = (*ep->func)(m, *offp, proto, ep->arg);
238 		else {
239 			/*
240 			 * it's inbound traffic, we need to match in reverse
241 			 * order
242 			 */
243 			prio = mask_match(ep, (struct sockaddr *)&d,
244 			    (struct sockaddr *)&s);
245 		}
246 
247 		/* see encap4_input() for issues here */
248 		if (prio <= 0)
249 			continue;
250 		if (prio > matchprio) {
251 			matchprio = prio;
252 			match = ep;
253 		}
254 	}
255 	if (match != NULL) {
256 		psw = match->psw;
257 		arg = match->arg;
258 	}
259 	mtx_unlock(&encapmtx);
260 
261 	if (match != NULL) {
262 		/* found a match */
263 		if (psw != NULL && psw->pr_input != NULL) {
264 			encap_fillarg(m, arg);
265 			return (*psw->pr_input)(mp, offp, proto);
266 		} else {
267 			m_freem(m);
268 			return (IPPROTO_DONE);
269 		}
270 	}
271 
272 	/* last resort: inject to raw socket */
273 	return rip6_input(mp, offp, proto);
274 }
275 #endif
276 
277 /*lint -sem(encap_add, custodial(1)) */
278 static void
279 encap_add(struct encaptab *ep)
280 {
281 
282 	mtx_assert(&encapmtx, MA_OWNED);
283 	LIST_INSERT_HEAD(&encaptab, ep, chain);
284 }
285 
286 /*
287  * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
288  * length of mask (sm and dm) is assumed to be same as sp/dp.
289  * Return value will be necessary as input (cookie) for encap_detach().
290  */
291 const struct encaptab *
292 encap_attach(int af, int proto, const struct sockaddr *sp,
293     const struct sockaddr *sm, const struct sockaddr *dp,
294     const struct sockaddr *dm, const struct protosw *psw, void *arg)
295 {
296 	struct encaptab *ep;
297 
298 	/* sanity check on args */
299 	if (sp->sa_len > sizeof(ep->src) || dp->sa_len > sizeof(ep->dst))
300 		return (NULL);
301 	if (sp->sa_len != dp->sa_len)
302 		return (NULL);
303 	if (af != sp->sa_family || af != dp->sa_family)
304 		return (NULL);
305 
306 	/* check if anyone have already attached with exactly same config */
307 	mtx_lock(&encapmtx);
308 	LIST_FOREACH(ep, &encaptab, chain) {
309 		if (ep->af != af)
310 			continue;
311 		if (ep->proto != proto)
312 			continue;
313 		if (ep->src.ss_len != sp->sa_len ||
314 		    bcmp(&ep->src, sp, sp->sa_len) != 0 ||
315 		    bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
316 			continue;
317 		if (ep->dst.ss_len != dp->sa_len ||
318 		    bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
319 		    bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
320 			continue;
321 
322 		mtx_unlock(&encapmtx);
323 		return (NULL);
324 	}
325 
326 	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
327 	if (ep == NULL) {
328 		mtx_unlock(&encapmtx);
329 		return (NULL);
330 	}
331 	bzero(ep, sizeof(*ep));
332 
333 	ep->af = af;
334 	ep->proto = proto;
335 	bcopy(sp, &ep->src, sp->sa_len);
336 	bcopy(sm, &ep->srcmask, sp->sa_len);
337 	bcopy(dp, &ep->dst, dp->sa_len);
338 	bcopy(dm, &ep->dstmask, dp->sa_len);
339 	ep->psw = psw;
340 	ep->arg = arg;
341 
342 	encap_add(ep);
343 	mtx_unlock(&encapmtx);
344 	return (ep);
345 }
346 
347 const struct encaptab *
348 encap_attach_func(int af, int proto,
349     int (*func)(const struct mbuf *, int, int, void *),
350     const struct protosw *psw, void *arg)
351 {
352 	struct encaptab *ep;
353 
354 	/* sanity check on args */
355 	if (!func)
356 		return (NULL);
357 
358 	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
359 	if (ep == NULL)
360 		return (NULL);
361 	bzero(ep, sizeof(*ep));
362 
363 	ep->af = af;
364 	ep->proto = proto;
365 	ep->func = func;
366 	ep->psw = psw;
367 	ep->arg = arg;
368 
369 	mtx_lock(&encapmtx);
370 	encap_add(ep);
371 	mtx_unlock(&encapmtx);
372 	return (ep);
373 }
374 
375 int
376 encap_detach(const struct encaptab *cookie)
377 {
378 	const struct encaptab *ep = cookie;
379 	struct encaptab *p;
380 
381 	mtx_lock(&encapmtx);
382 	LIST_FOREACH(p, &encaptab, chain) {
383 		if (p == ep) {
384 			LIST_REMOVE(p, chain);
385 			mtx_unlock(&encapmtx);
386 			free(p, M_NETADDR);	/*XXX*/
387 			return 0;
388 		}
389 	}
390 	mtx_unlock(&encapmtx);
391 
392 	return EINVAL;
393 }
394 
395 static int
396 mask_match(const struct encaptab *ep, const struct sockaddr *sp,
397     const struct sockaddr *dp)
398 {
399 	struct sockaddr_storage s;
400 	struct sockaddr_storage d;
401 	int i;
402 	const u_int8_t *p, *q;
403 	u_int8_t *r;
404 	int matchlen;
405 
406 	if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d))
407 		return 0;
408 	if (sp->sa_family != ep->af || dp->sa_family != ep->af)
409 		return 0;
410 	if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
411 		return 0;
412 
413 	matchlen = 0;
414 
415 	p = (const u_int8_t *)sp;
416 	q = (const u_int8_t *)&ep->srcmask;
417 	r = (u_int8_t *)&s;
418 	for (i = 0 ; i < sp->sa_len; i++) {
419 		r[i] = p[i] & q[i];
420 		/* XXX estimate */
421 		matchlen += (q[i] ? 8 : 0);
422 	}
423 
424 	p = (const u_int8_t *)dp;
425 	q = (const u_int8_t *)&ep->dstmask;
426 	r = (u_int8_t *)&d;
427 	for (i = 0 ; i < dp->sa_len; i++) {
428 		r[i] = p[i] & q[i];
429 		/* XXX rough estimate */
430 		matchlen += (q[i] ? 8 : 0);
431 	}
432 
433 	/* need to overwrite len/family portion as we don't compare them */
434 	s.ss_len = sp->sa_len;
435 	s.ss_family = sp->sa_family;
436 	d.ss_len = dp->sa_len;
437 	d.ss_family = dp->sa_family;
438 
439 	if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
440 	    bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
441 		return matchlen;
442 	} else
443 		return 0;
444 }
445 
446 static void
447 encap_fillarg(struct mbuf *m, void *arg)
448 {
449 	struct m_tag *tag;
450 
451 	if (arg != NULL) {
452 		tag = m_tag_get(PACKET_TAG_ENCAP, sizeof(void *), M_NOWAIT);
453 		if (tag != NULL) {
454 			*(void**)(tag+1) = arg;
455 			m_tag_prepend(m, tag);
456 		}
457 	}
458 }
459 
460 void *
461 encap_getarg(struct mbuf *m)
462 {
463 	void *p = NULL;
464 	struct m_tag *tag;
465 
466 	tag = m_tag_find(m, PACKET_TAG_ENCAP, NULL);
467 	if (tag) {
468 		p = *(void**)(tag+1);
469 		m_tag_delete(m, tag);
470 	}
471 	return p;
472 }
473