xref: /freebsd/sys/netinet/ip_encap.c (revision 06c3fb2749bda94cb5201f81ffdb8fa6c3161b2e)
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 #include "opt_inet.h"
63 #include "opt_inet6.h"
64 
65 #include <sys/param.h>
66 #include <sys/systm.h>
67 #include <sys/eventhandler.h>
68 #include <sys/kernel.h>
69 #include <sys/lock.h>
70 #include <sys/malloc.h>
71 #include <sys/mutex.h>
72 #include <sys/mbuf.h>
73 #include <sys/errno.h>
74 #include <sys/socket.h>
75 
76 #include <net/if.h>
77 #include <net/if_var.h>
78 
79 #include <netinet/in.h>
80 #include <netinet/ip_var.h>
81 #include <netinet/ip_encap.h>
82 
83 #ifdef INET6
84 #include <netinet6/ip6_var.h>
85 #endif
86 
87 static MALLOC_DEFINE(M_NETADDR, "encap_export_host",
88     "Export host address structure");
89 
90 struct encaptab {
91 	CK_LIST_ENTRY(encaptab) chain;
92 	int		proto;
93 	int		min_length;
94 	int		exact_match;
95 	void		*arg;
96 
97 	encap_lookup_t	lookup;
98 	encap_check_t	check;
99 	encap_input_t	input;
100 };
101 
102 struct srcaddrtab {
103 	CK_LIST_ENTRY(srcaddrtab) chain;
104 
105 	encap_srcaddr_t	srcaddr;
106 	void		*arg;
107 };
108 
109 CK_LIST_HEAD(encaptab_head, encaptab);
110 CK_LIST_HEAD(srcaddrtab_head, srcaddrtab);
111 #ifdef INET
112 static struct encaptab_head ipv4_encaptab = CK_LIST_HEAD_INITIALIZER();
113 static struct srcaddrtab_head ipv4_srcaddrtab = CK_LIST_HEAD_INITIALIZER();
114 #endif
115 #ifdef INET6
116 static struct encaptab_head ipv6_encaptab = CK_LIST_HEAD_INITIALIZER();
117 static struct srcaddrtab_head ipv6_srcaddrtab = CK_LIST_HEAD_INITIALIZER();
118 #endif
119 
120 static struct mtx encapmtx, srcaddrmtx;
121 MTX_SYSINIT(encapmtx, &encapmtx, "encapmtx", MTX_DEF);
122 MTX_SYSINIT(srcaddrmtx, &srcaddrmtx, "srcaddrmtx", MTX_DEF);
123 #define	ENCAP_WLOCK()		mtx_lock(&encapmtx)
124 #define	ENCAP_WUNLOCK()		mtx_unlock(&encapmtx)
125 #define	ENCAP_RLOCK_TRACKER	struct epoch_tracker encap_et
126 #define	ENCAP_RLOCK()		NET_EPOCH_ENTER(encap_et)
127 #define	ENCAP_RUNLOCK()		NET_EPOCH_EXIT(encap_et)
128 #define	ENCAP_WAIT()		NET_EPOCH_WAIT()
129 
130 #define	SRCADDR_WLOCK()		mtx_lock(&srcaddrmtx)
131 #define	SRCADDR_WUNLOCK()	mtx_unlock(&srcaddrmtx)
132 #define	SRCADDR_RLOCK_TRACKER	struct epoch_tracker srcaddr_et
133 #define	SRCADDR_RLOCK()		\
134     epoch_enter_preempt(net_epoch_preempt, &srcaddr_et)
135 #define	SRCADDR_RUNLOCK()	\
136     epoch_exit_preempt(net_epoch_preempt, &srcaddr_et)
137 #define	SRCADDR_WAIT()		epoch_wait_preempt(net_epoch_preempt)
138 
139 /*
140  * ifaddr_event_ext handler.
141  *
142  * Tunnelling interfaces may request the kernel to notify when
143  * some interface addresses appears or disappears. Usually tunnelling
144  * interface must use an address configured on the local machine as
145  * ingress address to be able receive datagramms and do not send
146  * spoofed packets.
147  */
148 static void
149 srcaddr_change_event(void *arg __unused, struct ifnet *ifp,
150     struct ifaddr *ifa, int event)
151 {
152 	SRCADDR_RLOCK_TRACKER;
153 	struct srcaddrtab_head *head;
154 	struct srcaddrtab *p;
155 
156 	/* Support for old ifaddr_event. */
157 	EVENTHANDLER_INVOKE(ifaddr_event, ifp);
158 
159 	switch (ifa->ifa_addr->sa_family) {
160 #ifdef INET
161 	case AF_INET:
162 		head = &ipv4_srcaddrtab;
163 		break;
164 #endif
165 #ifdef INET6
166 	case AF_INET6:
167 		head = &ipv6_srcaddrtab;
168 		break;
169 #endif
170 	default:
171 		/* ignore event */
172 		return;
173 	}
174 
175 	SRCADDR_RLOCK();
176 	CK_LIST_FOREACH(p, head, chain) {
177 		(*p->srcaddr)(p->arg, ifa->ifa_addr, event);
178 	}
179 	SRCADDR_RUNLOCK();
180 }
181 EVENTHANDLER_DEFINE(ifaddr_event_ext, srcaddr_change_event, NULL, 0);
182 
183 static struct srcaddrtab *
184 encap_register_srcaddr(struct srcaddrtab_head *head, encap_srcaddr_t func,
185     void *arg, int mflags)
186 {
187 	struct srcaddrtab *p, *tmp;
188 
189 	if (func == NULL)
190 		return (NULL);
191 	p = malloc(sizeof(*p), M_NETADDR, mflags);
192 	if (p == NULL)
193 		return (NULL);
194 	p->srcaddr = func;
195 	p->arg = arg;
196 
197 	SRCADDR_WLOCK();
198 	CK_LIST_FOREACH(tmp, head, chain) {
199 		if (func == tmp->srcaddr && arg == tmp->arg)
200 			break;
201 	}
202 	if (tmp == NULL)
203 		CK_LIST_INSERT_HEAD(head, p, chain);
204 	SRCADDR_WUNLOCK();
205 
206 	if (tmp != NULL) {
207 		free(p, M_NETADDR);
208 		p = tmp;
209 	}
210 	return (p);
211 }
212 
213 static int
214 encap_unregister_srcaddr(struct srcaddrtab_head *head,
215     const struct srcaddrtab *cookie)
216 {
217 	struct srcaddrtab *p;
218 
219 	SRCADDR_WLOCK();
220 	CK_LIST_FOREACH(p, head, chain) {
221 		if (p == cookie) {
222 			CK_LIST_REMOVE(p, chain);
223 			SRCADDR_WUNLOCK();
224 			SRCADDR_WAIT();
225 			free(p, M_NETADDR);
226 			return (0);
227 		}
228 	}
229 	SRCADDR_WUNLOCK();
230 	return (EINVAL);
231 }
232 
233 static struct encaptab *
234 encap_attach(struct encaptab_head *head, const struct encap_config *cfg,
235     void *arg, int mflags)
236 {
237 	struct encaptab *ep, *tmp;
238 
239 	if (cfg == NULL || cfg->input == NULL ||
240 	    (cfg->check == NULL && cfg->lookup == NULL) ||
241 	    (cfg->lookup != NULL && cfg->exact_match != ENCAP_DRV_LOOKUP) ||
242 	    (cfg->exact_match == ENCAP_DRV_LOOKUP && cfg->lookup == NULL))
243 		return (NULL);
244 
245 	ep = malloc(sizeof(*ep), M_NETADDR, mflags);
246 	if (ep == NULL)
247 		return (NULL);
248 
249 	ep->proto = cfg->proto;
250 	ep->min_length = cfg->min_length;
251 	ep->exact_match = cfg->exact_match;
252 	ep->arg = arg;
253 	ep->lookup = cfg->exact_match == ENCAP_DRV_LOOKUP ? cfg->lookup: NULL;
254 	ep->check = cfg->exact_match != ENCAP_DRV_LOOKUP ? cfg->check: NULL;
255 	ep->input = cfg->input;
256 
257 	ENCAP_WLOCK();
258 	CK_LIST_FOREACH(tmp, head, chain) {
259 		if (tmp->exact_match <= ep->exact_match)
260 			break;
261 	}
262 	if (tmp == NULL)
263 		CK_LIST_INSERT_HEAD(head, ep, chain);
264 	else
265 		CK_LIST_INSERT_BEFORE(tmp, ep, chain);
266 	ENCAP_WUNLOCK();
267 	return (ep);
268 }
269 
270 static int
271 encap_detach(struct encaptab_head *head, const struct encaptab *cookie)
272 {
273 	struct encaptab *ep;
274 
275 	ENCAP_WLOCK();
276 	CK_LIST_FOREACH(ep, head, chain) {
277 		if (ep == cookie) {
278 			CK_LIST_REMOVE(ep, chain);
279 			ENCAP_WUNLOCK();
280 			ENCAP_WAIT();
281 			free(ep, M_NETADDR);
282 			return (0);
283 		}
284 	}
285 	ENCAP_WUNLOCK();
286 	return (EINVAL);
287 }
288 
289 static int
290 encap_input(struct encaptab_head *head, struct mbuf *m, int off, int proto)
291 {
292 	ENCAP_RLOCK_TRACKER;
293 	struct encaptab *ep, *match;
294 	void *arg;
295 	int matchprio, ret;
296 
297 	match = NULL;
298 	matchprio = 0;
299 
300 	ENCAP_RLOCK();
301 	CK_LIST_FOREACH(ep, head, chain) {
302 		if (ep->proto >= 0 && ep->proto != proto)
303 			continue;
304 		if (ep->min_length > m->m_pkthdr.len)
305 			continue;
306 		if (ep->exact_match == ENCAP_DRV_LOOKUP)
307 			ret = (*ep->lookup)(m, off, proto, &arg);
308 		else
309 			ret = (*ep->check)(m, off, proto, ep->arg);
310 		if (ret <= 0)
311 			continue;
312 		if (ret > matchprio) {
313 			match = ep;
314 			if (ep->exact_match != ENCAP_DRV_LOOKUP)
315 				arg = ep->arg;
316 			/*
317 			 * No need to continue the search, we got the
318 			 * exact match.
319 			 */
320 			if (ret >= ep->exact_match)
321 				break;
322 			matchprio = ret;
323 		}
324 	}
325 
326 	if (match != NULL) {
327 		/* found a match, "match" has the best one */
328 		ret = (*match->input)(m, off, proto, arg);
329 		ENCAP_RUNLOCK();
330 		MPASS(ret == IPPROTO_DONE);
331 		return (IPPROTO_DONE);
332 	}
333 	ENCAP_RUNLOCK();
334 	return (0);
335 }
336 
337 #ifdef INET
338 const struct srcaddrtab *
339 ip_encap_register_srcaddr(encap_srcaddr_t func, void *arg, int mflags)
340 {
341 
342 	return (encap_register_srcaddr(&ipv4_srcaddrtab, func, arg, mflags));
343 }
344 
345 int
346 ip_encap_unregister_srcaddr(const struct srcaddrtab *cookie)
347 {
348 
349 	return (encap_unregister_srcaddr(&ipv4_srcaddrtab, cookie));
350 }
351 
352 const struct encaptab *
353 ip_encap_attach(const struct encap_config *cfg, void *arg, int mflags)
354 {
355 
356 	return (encap_attach(&ipv4_encaptab, cfg, arg, mflags));
357 }
358 
359 int
360 ip_encap_detach(const struct encaptab *cookie)
361 {
362 
363 	return (encap_detach(&ipv4_encaptab, cookie));
364 }
365 
366 int
367 encap4_input(struct mbuf **mp, int *offp, int proto)
368 {
369 
370 	if (encap_input(&ipv4_encaptab, *mp, *offp, proto) != IPPROTO_DONE)
371 		return (rip_input(mp, offp, proto));
372 	return (IPPROTO_DONE);
373 }
374 #endif /* INET */
375 
376 #ifdef INET6
377 const struct srcaddrtab *
378 ip6_encap_register_srcaddr(encap_srcaddr_t func, void *arg, int mflags)
379 {
380 
381 	return (encap_register_srcaddr(&ipv6_srcaddrtab, func, arg, mflags));
382 }
383 
384 int
385 ip6_encap_unregister_srcaddr(const struct srcaddrtab *cookie)
386 {
387 
388 	return (encap_unregister_srcaddr(&ipv6_srcaddrtab, cookie));
389 }
390 
391 const struct encaptab *
392 ip6_encap_attach(const struct encap_config *cfg, void *arg, int mflags)
393 {
394 
395 	return (encap_attach(&ipv6_encaptab, cfg, arg, mflags));
396 }
397 
398 int
399 ip6_encap_detach(const struct encaptab *cookie)
400 {
401 
402 	return (encap_detach(&ipv6_encaptab, cookie));
403 }
404 
405 int
406 encap6_input(struct mbuf **mp, int *offp, int proto)
407 {
408 
409 	if (encap_input(&ipv6_encaptab, *mp, *offp, proto) != IPPROTO_DONE)
410 		return (rip6_input(mp, offp, proto));
411 	return (IPPROTO_DONE);
412 }
413 #endif /* INET6 */
414