ip_encap.c revision 335031 
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: stable/11/sys/netinet/ip_encap.c 335031 2018-06-13 07:14:34Z ae $");
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/lock.h>
69#include <sys/mutex.h>
70#include <sys/socket.h>
71#include <sys/sockio.h>
72#include <sys/mbuf.h>
73#include <sys/errno.h>
74#include <sys/protosw.h>
75#include <sys/queue.h>
76
77#include <net/if.h>
78#include <net/route.h>
79
80#include <netinet/in.h>
81#include <netinet/in_systm.h>
82#include <netinet/ip.h>
83#include <netinet/ip_var.h>
84#include <netinet/ip_encap.h>
85
86#ifdef INET6
87#include <netinet/ip6.h>
88#include <netinet6/ip6_var.h>
89#endif
90
91#include <machine/stdarg.h>
92
93#include <sys/kernel.h>
94#include <sys/malloc.h>
95static MALLOC_DEFINE(M_NETADDR, "encap_export_host", "Export host address structure");
96
97static void encap_add(struct encaptab *);
98static int mask_match(const struct encaptab *, const struct sockaddr *,
99		const struct sockaddr *);
100static void encap_fillarg(struct mbuf *, void *);
101
102/*
103 * All global variables in ip_encap.c are locked using encapmtx.
104 */
105static struct mtx encapmtx;
106MTX_SYSINIT(encapmtx, &encapmtx, "encapmtx", MTX_DEF);
107static LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(encaptab);
108
109#ifdef INET
110int
111encap4_input(struct mbuf **mp, int *offp, int proto)
112{
113	struct ip *ip;
114	struct mbuf *m;
115	struct sockaddr_in s, d;
116	const struct protosw *psw;
117	struct encaptab *ep, *match;
118	void *arg;
119	int matchprio, off, prio;
120
121	m = *mp;
122	off = *offp;
123	ip = mtod(m, struct ip *);
124
125	bzero(&s, sizeof(s));
126	s.sin_family = AF_INET;
127	s.sin_len = sizeof(struct sockaddr_in);
128	s.sin_addr = ip->ip_src;
129	bzero(&d, sizeof(d));
130	d.sin_family = AF_INET;
131	d.sin_len = sizeof(struct sockaddr_in);
132	d.sin_addr = ip->ip_dst;
133
134	arg = NULL;
135	psw = NULL;
136	match = NULL;
137	matchprio = 0;
138	mtx_lock(&encapmtx);
139	LIST_FOREACH(ep, &encaptab, chain) {
140		if (ep->af != AF_INET)
141			continue;
142		if (ep->proto >= 0 && ep->proto != proto)
143			continue;
144		if (ep->func)
145			prio = (*ep->func)(m, off, proto, ep->arg);
146		else {
147			/*
148			 * it's inbound traffic, we need to match in reverse
149			 * order
150			 */
151			prio = mask_match(ep, (struct sockaddr *)&d,
152			    (struct sockaddr *)&s);
153		}
154
155		/*
156		 * We prioritize the matches by using bit length of the
157		 * matches.  mask_match() and user-supplied matching function
158		 * should return the bit length of the matches (for example,
159		 * if both src/dst are matched for IPv4, 64 should be returned).
160		 * 0 or negative return value means "it did not match".
161		 *
162		 * The question is, since we have two "mask" portion, we
163		 * cannot really define total order between entries.
164		 * For example, which of these should be preferred?
165		 * mask_match() returns 48 (32 + 16) for both of them.
166		 *	src=3ffe::/16, dst=3ffe:501::/32
167		 *	src=3ffe:501::/32, dst=3ffe::/16
168		 *
169		 * We need to loop through all the possible candidates
170		 * to get the best match - the search takes O(n) for
171		 * n attachments (i.e. interfaces).
172		 */
173		if (prio <= 0)
174			continue;
175		if (prio > matchprio) {
176			matchprio = prio;
177			match = ep;
178		}
179	}
180	if (match != NULL) {
181		psw = match->psw;
182		arg = match->arg;
183	}
184	mtx_unlock(&encapmtx);
185
186	if (match != NULL) {
187		/* found a match, "match" has the best one */
188		if (psw != NULL && psw->pr_input != NULL) {
189			encap_fillarg(m, arg);
190			(*psw->pr_input)(mp, offp, proto);
191		} else
192			m_freem(m);
193		return (IPPROTO_DONE);
194	}
195
196	/* last resort: inject to raw socket */
197	return (rip_input(mp, offp, proto));
198}
199#endif
200
201#ifdef INET6
202int
203encap6_input(struct mbuf **mp, int *offp, int proto)
204{
205	struct mbuf *m = *mp;
206	struct ip6_hdr *ip6;
207	struct sockaddr_in6 s, d;
208	const struct protosw *psw;
209	struct encaptab *ep, *match;
210	void *arg;
211	int prio, matchprio;
212
213	ip6 = mtod(m, struct ip6_hdr *);
214
215	bzero(&s, sizeof(s));
216	s.sin6_family = AF_INET6;
217	s.sin6_len = sizeof(struct sockaddr_in6);
218	s.sin6_addr = ip6->ip6_src;
219	bzero(&d, sizeof(d));
220	d.sin6_family = AF_INET6;
221	d.sin6_len = sizeof(struct sockaddr_in6);
222	d.sin6_addr = ip6->ip6_dst;
223
224	arg = NULL;
225	psw = NULL;
226	match = NULL;
227	matchprio = 0;
228	mtx_lock(&encapmtx);
229	LIST_FOREACH(ep, &encaptab, chain) {
230		if (ep->af != AF_INET6)
231			continue;
232		if (ep->proto >= 0 && ep->proto != proto)
233			continue;
234		if (ep->func)
235			prio = (*ep->func)(m, *offp, proto, ep->arg);
236		else {
237			/*
238			 * it's inbound traffic, we need to match in reverse
239			 * order
240			 */
241			prio = mask_match(ep, (struct sockaddr *)&d,
242			    (struct sockaddr *)&s);
243		}
244
245		/* see encap4_input() for issues here */
246		if (prio <= 0)
247			continue;
248		if (prio > matchprio) {
249			matchprio = prio;
250			match = ep;
251		}
252	}
253	if (match != NULL) {
254		psw = match->psw;
255		arg = match->arg;
256	}
257	mtx_unlock(&encapmtx);
258
259	if (match != NULL) {
260		/* found a match */
261		if (psw != NULL && psw->pr_input != NULL) {
262			encap_fillarg(m, arg);
263			return (*psw->pr_input)(mp, offp, proto);
264		} else {
265			m_freem(m);
266			return (IPPROTO_DONE);
267		}
268	}
269
270	/* last resort: inject to raw socket */
271	return rip6_input(mp, offp, proto);
272}
273#endif
274
275/*lint -sem(encap_add, custodial(1)) */
276static void
277encap_add(struct encaptab *ep)
278{
279
280	mtx_assert(&encapmtx, MA_OWNED);
281	LIST_INSERT_HEAD(&encaptab, ep, chain);
282}
283
284/*
285 * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
286 * length of mask (sm and dm) is assumed to be same as sp/dp.
287 * Return value will be necessary as input (cookie) for encap_detach().
288 */
289const struct encaptab *
290encap_attach(int af, int proto, const struct sockaddr *sp,
291    const struct sockaddr *sm, const struct sockaddr *dp,
292    const struct sockaddr *dm, const struct protosw *psw, void *arg)
293{
294	struct encaptab *ep;
295
296	/* sanity check on args */
297	if (sp->sa_len > sizeof(ep->src) || dp->sa_len > sizeof(ep->dst))
298		return (NULL);
299	if (sp->sa_len != dp->sa_len)
300		return (NULL);
301	if (af != sp->sa_family || af != dp->sa_family)
302		return (NULL);
303
304	/* check if anyone have already attached with exactly same config */
305	mtx_lock(&encapmtx);
306	LIST_FOREACH(ep, &encaptab, chain) {
307		if (ep->af != af)
308			continue;
309		if (ep->proto != proto)
310			continue;
311		if (ep->src.ss_len != sp->sa_len ||
312		    bcmp(&ep->src, sp, sp->sa_len) != 0 ||
313		    bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
314			continue;
315		if (ep->dst.ss_len != dp->sa_len ||
316		    bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
317		    bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
318			continue;
319
320		mtx_unlock(&encapmtx);
321		return (NULL);
322	}
323
324	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
325	if (ep == NULL) {
326		mtx_unlock(&encapmtx);
327		return (NULL);
328	}
329	bzero(ep, sizeof(*ep));
330
331	ep->af = af;
332	ep->proto = proto;
333	bcopy(sp, &ep->src, sp->sa_len);
334	bcopy(sm, &ep->srcmask, sp->sa_len);
335	bcopy(dp, &ep->dst, dp->sa_len);
336	bcopy(dm, &ep->dstmask, dp->sa_len);
337	ep->psw = psw;
338	ep->arg = arg;
339
340	encap_add(ep);
341	mtx_unlock(&encapmtx);
342	return (ep);
343}
344
345const struct encaptab *
346encap_attach_func(int af, int proto,
347    int (*func)(const struct mbuf *, int, int, void *),
348    const struct protosw *psw, void *arg)
349{
350	struct encaptab *ep;
351
352	/* sanity check on args */
353	if (!func)
354		return (NULL);
355
356	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
357	if (ep == NULL)
358		return (NULL);
359	bzero(ep, sizeof(*ep));
360
361	ep->af = af;
362	ep->proto = proto;
363	ep->func = func;
364	ep->psw = psw;
365	ep->arg = arg;
366
367	mtx_lock(&encapmtx);
368	encap_add(ep);
369	mtx_unlock(&encapmtx);
370	return (ep);
371}
372
373int
374encap_detach(const struct encaptab *cookie)
375{
376	const struct encaptab *ep = cookie;
377	struct encaptab *p;
378
379	mtx_lock(&encapmtx);
380	LIST_FOREACH(p, &encaptab, chain) {
381		if (p == ep) {
382			LIST_REMOVE(p, chain);
383			mtx_unlock(&encapmtx);
384			free(p, M_NETADDR);	/*XXX*/
385			return 0;
386		}
387	}
388	mtx_unlock(&encapmtx);
389
390	return EINVAL;
391}
392
393static int
394mask_match(const struct encaptab *ep, const struct sockaddr *sp,
395    const struct sockaddr *dp)
396{
397	struct sockaddr_storage s;
398	struct sockaddr_storage d;
399	int i;
400	const u_int8_t *p, *q;
401	u_int8_t *r;
402	int matchlen;
403
404	if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d))
405		return 0;
406	if (sp->sa_family != ep->af || dp->sa_family != ep->af)
407		return 0;
408	if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
409		return 0;
410
411	matchlen = 0;
412
413	p = (const u_int8_t *)sp;
414	q = (const u_int8_t *)&ep->srcmask;
415	r = (u_int8_t *)&s;
416	for (i = 0 ; i < sp->sa_len; i++) {
417		r[i] = p[i] & q[i];
418		/* XXX estimate */
419		matchlen += (q[i] ? 8 : 0);
420	}
421
422	p = (const u_int8_t *)dp;
423	q = (const u_int8_t *)&ep->dstmask;
424	r = (u_int8_t *)&d;
425	for (i = 0 ; i < dp->sa_len; i++) {
426		r[i] = p[i] & q[i];
427		/* XXX rough estimate */
428		matchlen += (q[i] ? 8 : 0);
429	}
430
431	/* need to overwrite len/family portion as we don't compare them */
432	s.ss_len = sp->sa_len;
433	s.ss_family = sp->sa_family;
434	d.ss_len = dp->sa_len;
435	d.ss_family = dp->sa_family;
436
437	if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
438	    bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
439		return matchlen;
440	} else
441		return 0;
442}
443
444static void
445encap_fillarg(struct mbuf *m, void *arg)
446{
447	struct m_tag *tag;
448
449	if (arg != NULL) {
450		tag = m_tag_get(PACKET_TAG_ENCAP, sizeof(void *), M_NOWAIT);
451		if (tag != NULL) {
452			*(void**)(tag+1) = arg;
453			m_tag_prepend(m, tag);
454		}
455	}
456}
457
458void *
459encap_getarg(struct mbuf *m)
460{
461	void *p = NULL;
462	struct m_tag *tag;
463
464	tag = m_tag_find(m, PACKET_TAG_ENCAP, NULL);
465	if (tag) {
466		p = *(void**)(tag+1);
467		m_tag_delete(m, tag);
468	}
469	return p;
470}
471