xref: /freebsd/sys/netinet/ip_encap.c (revision 0a48773f8c6108ad8755065c5cdc13fcdc0728b2)
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  * Copyright (c) 2018 Andrey V. Elsukov <ae@FreeBSD.org>
8  * All rights reserved.
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  * 1. Redistributions of source code must retain the above copyright
14  *    notice, this list of conditions and the following disclaimer.
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in the
17  *    documentation and/or other materials provided with the distribution.
18  * 3. Neither the name of the project nor the names of its contributors
19  *    may be used to endorse or promote products derived from this software
20  *    without specific prior written permission.
21  *
22  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
23  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
26  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32  * SUCH DAMAGE.
33  */
34 /*
35  * My grandfather said that there's a devil inside tunnelling technology...
36  *
37  * We have surprisingly many protocols that want packets with IP protocol
38  * #4 or #41.  Here's a list of protocols that want protocol #41:
39  *	RFC1933 configured tunnel
40  *	RFC1933 automatic tunnel
41  *	RFC2401 IPsec tunnel
42  *	RFC2473 IPv6 generic packet tunnelling
43  *	RFC2529 6over4 tunnel
44  *	mobile-ip6 (uses RFC2473)
45  *	RFC3056 6to4 tunnel
46  *	isatap tunnel
47  * Here's a list of protocol that want protocol #4:
48  *	RFC1853 IPv4-in-IPv4 tunnelling
49  *	RFC2003 IPv4 encapsulation within IPv4
50  *	RFC2344 reverse tunnelling for mobile-ip4
51  *	RFC2401 IPsec tunnel
52  * Well, what can I say.  They impose different en/decapsulation mechanism
53  * from each other, so they need separate protocol handler.  The only one
54  * we can easily determine by protocol # is IPsec, which always has
55  * AH/ESP/IPComp header right after outer IP header.
56  *
57  * So, clearly good old protosw does not work for protocol #4 and #41.
58  * The code will let you match protocol via src/dst address pair.
59  */
60 
61 #include <sys/cdefs.h>
62 __FBSDID("$FreeBSD$");
63 
64 #include "opt_inet.h"
65 #include "opt_inet6.h"
66 
67 #include <sys/param.h>
68 #include <sys/systm.h>
69 #include <sys/eventhandler.h>
70 #include <sys/kernel.h>
71 #include <sys/lock.h>
72 #include <sys/malloc.h>
73 #include <sys/mutex.h>
74 #include <sys/mbuf.h>
75 #include <sys/errno.h>
76 #include <sys/socket.h>
77 
78 #include <net/if.h>
79 #include <net/if_var.h>
80 
81 #include <netinet/in.h>
82 #include <netinet/ip_var.h>
83 #include <netinet/ip_encap.h>
84 
85 #ifdef INET6
86 #include <netinet6/ip6_var.h>
87 #endif
88 
89 static MALLOC_DEFINE(M_NETADDR, "encap_export_host",
90     "Export host address structure");
91 
92 struct encaptab {
93 	CK_LIST_ENTRY(encaptab) chain;
94 	int		proto;
95 	int		min_length;
96 	int		exact_match;
97 	void		*arg;
98 
99 	encap_lookup_t	lookup;
100 	encap_check_t	check;
101 	encap_input_t	input;
102 };
103 
104 struct srcaddrtab {
105 	CK_LIST_ENTRY(srcaddrtab) chain;
106 
107 	encap_srcaddr_t	srcaddr;
108 	void		*arg;
109 };
110 
111 CK_LIST_HEAD(encaptab_head, encaptab);
112 CK_LIST_HEAD(srcaddrtab_head, srcaddrtab);
113 #ifdef INET
114 static struct encaptab_head ipv4_encaptab = CK_LIST_HEAD_INITIALIZER();
115 static struct srcaddrtab_head ipv4_srcaddrtab = CK_LIST_HEAD_INITIALIZER();
116 #endif
117 #ifdef INET6
118 static struct encaptab_head ipv6_encaptab = CK_LIST_HEAD_INITIALIZER();
119 static struct srcaddrtab_head ipv6_srcaddrtab = CK_LIST_HEAD_INITIALIZER();
120 #endif
121 
122 static struct mtx encapmtx, srcaddrmtx;
123 MTX_SYSINIT(encapmtx, &encapmtx, "encapmtx", MTX_DEF);
124 MTX_SYSINIT(srcaddrmtx, &srcaddrmtx, "srcaddrmtx", MTX_DEF);
125 #define	ENCAP_WLOCK()		mtx_lock(&encapmtx)
126 #define	ENCAP_WUNLOCK()		mtx_unlock(&encapmtx)
127 #define	ENCAP_RLOCK_TRACKER	struct epoch_tracker encap_et
128 #define	ENCAP_RLOCK()		\
129     epoch_enter_preempt(net_epoch_preempt, &encap_et)
130 #define	ENCAP_RUNLOCK()		\
131     epoch_exit_preempt(net_epoch_preempt, &encap_et)
132 #define	ENCAP_WAIT()		epoch_wait_preempt(net_epoch_preempt)
133 
134 #define	SRCADDR_WLOCK()		mtx_lock(&srcaddrmtx)
135 #define	SRCADDR_WUNLOCK()	mtx_unlock(&srcaddrmtx)
136 #define	SRCADDR_RLOCK_TRACKER	struct epoch_tracker srcaddr_et
137 #define	SRCADDR_RLOCK()		\
138     epoch_enter_preempt(net_epoch_preempt, &srcaddr_et)
139 #define	SRCADDR_RUNLOCK()	\
140     epoch_exit_preempt(net_epoch_preempt, &srcaddr_et)
141 #define	SRCADDR_WAIT()		epoch_wait_preempt(net_epoch_preempt)
142 
143 /*
144  * ifaddr_event_ext handler.
145  *
146  * Tunnelling interfaces may request the kernel to notify when
147  * some interface addresses appears or disappears. Usually tunnelling
148  * interface must use an address configured on the local machine as
149  * ingress address to be able receive datagramms and do not send
150  * spoofed packets.
151  */
152 static void
153 srcaddr_change_event(void *arg __unused, struct ifnet *ifp,
154     struct ifaddr *ifa, int event)
155 {
156 	SRCADDR_RLOCK_TRACKER;
157 	struct srcaddrtab_head *head;
158 	struct srcaddrtab *p;
159 
160 	/* Support for old ifaddr_event. */
161 	EVENTHANDLER_INVOKE(ifaddr_event, ifp);
162 
163 	switch (ifa->ifa_addr->sa_family) {
164 #ifdef INET
165 	case AF_INET:
166 		head = &ipv4_srcaddrtab;
167 		break;
168 #endif
169 #ifdef INET6
170 	case AF_INET6:
171 		head = &ipv6_srcaddrtab;
172 		break;
173 #endif
174 	default:
175 		/* ignore event */
176 		return;
177 	}
178 
179 	SRCADDR_RLOCK();
180 	CK_LIST_FOREACH(p, head, chain) {
181 		(*p->srcaddr)(p->arg, ifa->ifa_addr, event);
182 	}
183 	SRCADDR_RUNLOCK();
184 }
185 EVENTHANDLER_DEFINE(ifaddr_event_ext, srcaddr_change_event, NULL, 0);
186 
187 static struct srcaddrtab *
188 encap_register_srcaddr(struct srcaddrtab_head *head, encap_srcaddr_t func,
189     void *arg, int mflags)
190 {
191 	struct srcaddrtab *p, *tmp;
192 
193 	if (func == NULL)
194 		return (NULL);
195 	p = malloc(sizeof(*p), M_NETADDR, mflags);
196 	if (p == NULL)
197 		return (NULL);
198 	p->srcaddr = func;
199 	p->arg = arg;
200 
201 	SRCADDR_WLOCK();
202 	CK_LIST_FOREACH(tmp, head, chain) {
203 		if (func == tmp->srcaddr && arg == tmp->arg)
204 			break;
205 	}
206 	if (tmp == NULL)
207 		CK_LIST_INSERT_HEAD(head, p, chain);
208 	SRCADDR_WUNLOCK();
209 
210 	if (tmp != NULL) {
211 		free(p, M_NETADDR);
212 		p = tmp;
213 	}
214 	return (p);
215 }
216 
217 static int
218 encap_unregister_srcaddr(struct srcaddrtab_head *head,
219     const struct srcaddrtab *cookie)
220 {
221 	struct srcaddrtab *p;
222 
223 	SRCADDR_WLOCK();
224 	CK_LIST_FOREACH(p, head, chain) {
225 		if (p == cookie) {
226 			CK_LIST_REMOVE(p, chain);
227 			SRCADDR_WUNLOCK();
228 			SRCADDR_WAIT();
229 			free(p, M_NETADDR);
230 			return (0);
231 		}
232 	}
233 	SRCADDR_WUNLOCK();
234 	return (EINVAL);
235 }
236 
237 static struct encaptab *
238 encap_attach(struct encaptab_head *head, const struct encap_config *cfg,
239     void *arg, int mflags)
240 {
241 	struct encaptab *ep, *tmp;
242 
243 	if (cfg == NULL || cfg->input == NULL ||
244 	    (cfg->check == NULL && cfg->lookup == NULL) ||
245 	    (cfg->lookup != NULL && cfg->exact_match != ENCAP_DRV_LOOKUP) ||
246 	    (cfg->exact_match == ENCAP_DRV_LOOKUP && cfg->lookup == NULL))
247 		return (NULL);
248 
249 	ep = malloc(sizeof(*ep), M_NETADDR, mflags);
250 	if (ep == NULL)
251 		return (NULL);
252 
253 	ep->proto = cfg->proto;
254 	ep->min_length = cfg->min_length;
255 	ep->exact_match = cfg->exact_match;
256 	ep->arg = arg;
257 	ep->lookup = cfg->exact_match == ENCAP_DRV_LOOKUP ? cfg->lookup: NULL;
258 	ep->check = cfg->exact_match != ENCAP_DRV_LOOKUP ? cfg->check: NULL;
259 	ep->input = cfg->input;
260 
261 	ENCAP_WLOCK();
262 	CK_LIST_FOREACH(tmp, head, chain) {
263 		if (tmp->exact_match <= ep->exact_match)
264 			break;
265 	}
266 	if (tmp == NULL)
267 		CK_LIST_INSERT_HEAD(head, ep, chain);
268 	else
269 		CK_LIST_INSERT_BEFORE(tmp, ep, chain);
270 	ENCAP_WUNLOCK();
271 	return (ep);
272 }
273 
274 static int
275 encap_detach(struct encaptab_head *head, const struct encaptab *cookie)
276 {
277 	struct encaptab *ep;
278 
279 	ENCAP_WLOCK();
280 	CK_LIST_FOREACH(ep, head, chain) {
281 		if (ep == cookie) {
282 			CK_LIST_REMOVE(ep, chain);
283 			ENCAP_WUNLOCK();
284 			ENCAP_WAIT();
285 			free(ep, M_NETADDR);
286 			return (0);
287 		}
288 	}
289 	ENCAP_WUNLOCK();
290 	return (EINVAL);
291 }
292 
293 static int
294 encap_input(struct encaptab_head *head, struct mbuf *m, int off, int proto)
295 {
296 	ENCAP_RLOCK_TRACKER;
297 	struct encaptab *ep, *match;
298 	void *arg;
299 	int matchprio, ret;
300 
301 	match = NULL;
302 	matchprio = 0;
303 
304 	ENCAP_RLOCK();
305 	CK_LIST_FOREACH(ep, head, chain) {
306 		if (ep->proto >= 0 && ep->proto != proto)
307 			continue;
308 		if (ep->min_length > m->m_pkthdr.len)
309 			continue;
310 		if (ep->exact_match == ENCAP_DRV_LOOKUP)
311 			ret = (*ep->lookup)(m, off, proto, &arg);
312 		else
313 			ret = (*ep->check)(m, off, proto, ep->arg);
314 		if (ret <= 0)
315 			continue;
316 		if (ret > matchprio) {
317 			match = ep;
318 			if (ep->exact_match != ENCAP_DRV_LOOKUP)
319 				arg = ep->arg;
320 			/*
321 			 * No need to continue the search, we got the
322 			 * exact match.
323 			 */
324 			if (ret >= ep->exact_match)
325 				break;
326 			matchprio = ret;
327 		}
328 	}
329 
330 	if (match != NULL) {
331 		/* found a match, "match" has the best one */
332 		ret = (*match->input)(m, off, proto, arg);
333 		ENCAP_RUNLOCK();
334 		MPASS(ret == IPPROTO_DONE);
335 		return (IPPROTO_DONE);
336 	}
337 	ENCAP_RUNLOCK();
338 	return (0);
339 }
340 
341 #ifdef INET
342 const struct srcaddrtab *
343 ip_encap_register_srcaddr(encap_srcaddr_t func, void *arg, int mflags)
344 {
345 
346 	return (encap_register_srcaddr(&ipv4_srcaddrtab, func, arg, mflags));
347 }
348 
349 int
350 ip_encap_unregister_srcaddr(const struct srcaddrtab *cookie)
351 {
352 
353 	return (encap_unregister_srcaddr(&ipv4_srcaddrtab, cookie));
354 }
355 
356 const struct encaptab *
357 ip_encap_attach(const struct encap_config *cfg, void *arg, int mflags)
358 {
359 
360 	return (encap_attach(&ipv4_encaptab, cfg, arg, mflags));
361 }
362 
363 int
364 ip_encap_detach(const struct encaptab *cookie)
365 {
366 
367 	return (encap_detach(&ipv4_encaptab, cookie));
368 }
369 
370 int
371 encap4_input(struct mbuf **mp, int *offp, int proto)
372 {
373 
374 	if (encap_input(&ipv4_encaptab, *mp, *offp, proto) != IPPROTO_DONE)
375 		return (rip_input(mp, offp, proto));
376 	return (IPPROTO_DONE);
377 }
378 #endif /* INET */
379 
380 #ifdef INET6
381 const struct srcaddrtab *
382 ip6_encap_register_srcaddr(encap_srcaddr_t func, void *arg, int mflags)
383 {
384 
385 	return (encap_register_srcaddr(&ipv6_srcaddrtab, func, arg, mflags));
386 }
387 
388 int
389 ip6_encap_unregister_srcaddr(const struct srcaddrtab *cookie)
390 {
391 
392 	return (encap_unregister_srcaddr(&ipv6_srcaddrtab, cookie));
393 }
394 
395 const struct encaptab *
396 ip6_encap_attach(const struct encap_config *cfg, void *arg, int mflags)
397 {
398 
399 	return (encap_attach(&ipv6_encaptab, cfg, arg, mflags));
400 }
401 
402 int
403 ip6_encap_detach(const struct encaptab *cookie)
404 {
405 
406 	return (encap_detach(&ipv6_encaptab, cookie));
407 }
408 
409 int
410 encap6_input(struct mbuf **mp, int *offp, int proto)
411 {
412 
413 	if (encap_input(&ipv6_encaptab, *mp, *offp, proto) != IPPROTO_DONE)
414 		return (rip6_input(mp, offp, proto));
415 	return (IPPROTO_DONE);
416 }
417 #endif /* INET6 */
418