xref: /freebsd/sys/netinet/ip_encap.c (revision af23369a6deaaeb612ab266eb88b8bb8d560c322)
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()		NET_EPOCH_ENTER(encap_et)
129 #define	ENCAP_RUNLOCK()		NET_EPOCH_EXIT(encap_et)
130 #define	ENCAP_WAIT()		NET_EPOCH_WAIT()
131 
132 #define	SRCADDR_WLOCK()		mtx_lock(&srcaddrmtx)
133 #define	SRCADDR_WUNLOCK()	mtx_unlock(&srcaddrmtx)
134 #define	SRCADDR_RLOCK_TRACKER	struct epoch_tracker srcaddr_et
135 #define	SRCADDR_RLOCK()		\
136     epoch_enter_preempt(net_epoch_preempt, &srcaddr_et)
137 #define	SRCADDR_RUNLOCK()	\
138     epoch_exit_preempt(net_epoch_preempt, &srcaddr_et)
139 #define	SRCADDR_WAIT()		epoch_wait_preempt(net_epoch_preempt)
140 
141 /*
142  * ifaddr_event_ext handler.
143  *
144  * Tunnelling interfaces may request the kernel to notify when
145  * some interface addresses appears or disappears. Usually tunnelling
146  * interface must use an address configured on the local machine as
147  * ingress address to be able receive datagramms and do not send
148  * spoofed packets.
149  */
150 static void
151 srcaddr_change_event(void *arg __unused, struct ifnet *ifp,
152     struct ifaddr *ifa, int event)
153 {
154 	SRCADDR_RLOCK_TRACKER;
155 	struct srcaddrtab_head *head;
156 	struct srcaddrtab *p;
157 
158 	/* Support for old ifaddr_event. */
159 	EVENTHANDLER_INVOKE(ifaddr_event, ifp);
160 
161 	switch (ifa->ifa_addr->sa_family) {
162 #ifdef INET
163 	case AF_INET:
164 		head = &ipv4_srcaddrtab;
165 		break;
166 #endif
167 #ifdef INET6
168 	case AF_INET6:
169 		head = &ipv6_srcaddrtab;
170 		break;
171 #endif
172 	default:
173 		/* ignore event */
174 		return;
175 	}
176 
177 	SRCADDR_RLOCK();
178 	CK_LIST_FOREACH(p, head, chain) {
179 		(*p->srcaddr)(p->arg, ifa->ifa_addr, event);
180 	}
181 	SRCADDR_RUNLOCK();
182 }
183 EVENTHANDLER_DEFINE(ifaddr_event_ext, srcaddr_change_event, NULL, 0);
184 
185 static struct srcaddrtab *
186 encap_register_srcaddr(struct srcaddrtab_head *head, encap_srcaddr_t func,
187     void *arg, int mflags)
188 {
189 	struct srcaddrtab *p, *tmp;
190 
191 	if (func == NULL)
192 		return (NULL);
193 	p = malloc(sizeof(*p), M_NETADDR, mflags);
194 	if (p == NULL)
195 		return (NULL);
196 	p->srcaddr = func;
197 	p->arg = arg;
198 
199 	SRCADDR_WLOCK();
200 	CK_LIST_FOREACH(tmp, head, chain) {
201 		if (func == tmp->srcaddr && arg == tmp->arg)
202 			break;
203 	}
204 	if (tmp == NULL)
205 		CK_LIST_INSERT_HEAD(head, p, chain);
206 	SRCADDR_WUNLOCK();
207 
208 	if (tmp != NULL) {
209 		free(p, M_NETADDR);
210 		p = tmp;
211 	}
212 	return (p);
213 }
214 
215 static int
216 encap_unregister_srcaddr(struct srcaddrtab_head *head,
217     const struct srcaddrtab *cookie)
218 {
219 	struct srcaddrtab *p;
220 
221 	SRCADDR_WLOCK();
222 	CK_LIST_FOREACH(p, head, chain) {
223 		if (p == cookie) {
224 			CK_LIST_REMOVE(p, chain);
225 			SRCADDR_WUNLOCK();
226 			SRCADDR_WAIT();
227 			free(p, M_NETADDR);
228 			return (0);
229 		}
230 	}
231 	SRCADDR_WUNLOCK();
232 	return (EINVAL);
233 }
234 
235 static struct encaptab *
236 encap_attach(struct encaptab_head *head, const struct encap_config *cfg,
237     void *arg, int mflags)
238 {
239 	struct encaptab *ep, *tmp;
240 
241 	if (cfg == NULL || cfg->input == NULL ||
242 	    (cfg->check == NULL && cfg->lookup == NULL) ||
243 	    (cfg->lookup != NULL && cfg->exact_match != ENCAP_DRV_LOOKUP) ||
244 	    (cfg->exact_match == ENCAP_DRV_LOOKUP && cfg->lookup == NULL))
245 		return (NULL);
246 
247 	ep = malloc(sizeof(*ep), M_NETADDR, mflags);
248 	if (ep == NULL)
249 		return (NULL);
250 
251 	ep->proto = cfg->proto;
252 	ep->min_length = cfg->min_length;
253 	ep->exact_match = cfg->exact_match;
254 	ep->arg = arg;
255 	ep->lookup = cfg->exact_match == ENCAP_DRV_LOOKUP ? cfg->lookup: NULL;
256 	ep->check = cfg->exact_match != ENCAP_DRV_LOOKUP ? cfg->check: NULL;
257 	ep->input = cfg->input;
258 
259 	ENCAP_WLOCK();
260 	CK_LIST_FOREACH(tmp, head, chain) {
261 		if (tmp->exact_match <= ep->exact_match)
262 			break;
263 	}
264 	if (tmp == NULL)
265 		CK_LIST_INSERT_HEAD(head, ep, chain);
266 	else
267 		CK_LIST_INSERT_BEFORE(tmp, ep, chain);
268 	ENCAP_WUNLOCK();
269 	return (ep);
270 }
271 
272 static int
273 encap_detach(struct encaptab_head *head, const struct encaptab *cookie)
274 {
275 	struct encaptab *ep;
276 
277 	ENCAP_WLOCK();
278 	CK_LIST_FOREACH(ep, head, chain) {
279 		if (ep == cookie) {
280 			CK_LIST_REMOVE(ep, chain);
281 			ENCAP_WUNLOCK();
282 			ENCAP_WAIT();
283 			free(ep, M_NETADDR);
284 			return (0);
285 		}
286 	}
287 	ENCAP_WUNLOCK();
288 	return (EINVAL);
289 }
290 
291 static int
292 encap_input(struct encaptab_head *head, struct mbuf *m, int off, int proto)
293 {
294 	ENCAP_RLOCK_TRACKER;
295 	struct encaptab *ep, *match;
296 	void *arg;
297 	int matchprio, ret;
298 
299 	match = NULL;
300 	matchprio = 0;
301 
302 	ENCAP_RLOCK();
303 	CK_LIST_FOREACH(ep, head, chain) {
304 		if (ep->proto >= 0 && ep->proto != proto)
305 			continue;
306 		if (ep->min_length > m->m_pkthdr.len)
307 			continue;
308 		if (ep->exact_match == ENCAP_DRV_LOOKUP)
309 			ret = (*ep->lookup)(m, off, proto, &arg);
310 		else
311 			ret = (*ep->check)(m, off, proto, ep->arg);
312 		if (ret <= 0)
313 			continue;
314 		if (ret > matchprio) {
315 			match = ep;
316 			if (ep->exact_match != ENCAP_DRV_LOOKUP)
317 				arg = ep->arg;
318 			/*
319 			 * No need to continue the search, we got the
320 			 * exact match.
321 			 */
322 			if (ret >= ep->exact_match)
323 				break;
324 			matchprio = ret;
325 		}
326 	}
327 
328 	if (match != NULL) {
329 		/* found a match, "match" has the best one */
330 		ret = (*match->input)(m, off, proto, arg);
331 		ENCAP_RUNLOCK();
332 		MPASS(ret == IPPROTO_DONE);
333 		return (IPPROTO_DONE);
334 	}
335 	ENCAP_RUNLOCK();
336 	return (0);
337 }
338 
339 #ifdef INET
340 const struct srcaddrtab *
341 ip_encap_register_srcaddr(encap_srcaddr_t func, void *arg, int mflags)
342 {
343 
344 	return (encap_register_srcaddr(&ipv4_srcaddrtab, func, arg, mflags));
345 }
346 
347 int
348 ip_encap_unregister_srcaddr(const struct srcaddrtab *cookie)
349 {
350 
351 	return (encap_unregister_srcaddr(&ipv4_srcaddrtab, cookie));
352 }
353 
354 const struct encaptab *
355 ip_encap_attach(const struct encap_config *cfg, void *arg, int mflags)
356 {
357 
358 	return (encap_attach(&ipv4_encaptab, cfg, arg, mflags));
359 }
360 
361 int
362 ip_encap_detach(const struct encaptab *cookie)
363 {
364 
365 	return (encap_detach(&ipv4_encaptab, cookie));
366 }
367 
368 int
369 encap4_input(struct mbuf **mp, int *offp, int proto)
370 {
371 
372 	if (encap_input(&ipv4_encaptab, *mp, *offp, proto) != IPPROTO_DONE)
373 		return (rip_input(mp, offp, proto));
374 	return (IPPROTO_DONE);
375 }
376 #endif /* INET */
377 
378 #ifdef INET6
379 const struct srcaddrtab *
380 ip6_encap_register_srcaddr(encap_srcaddr_t func, void *arg, int mflags)
381 {
382 
383 	return (encap_register_srcaddr(&ipv6_srcaddrtab, func, arg, mflags));
384 }
385 
386 int
387 ip6_encap_unregister_srcaddr(const struct srcaddrtab *cookie)
388 {
389 
390 	return (encap_unregister_srcaddr(&ipv6_srcaddrtab, cookie));
391 }
392 
393 const struct encaptab *
394 ip6_encap_attach(const struct encap_config *cfg, void *arg, int mflags)
395 {
396 
397 	return (encap_attach(&ipv6_encaptab, cfg, arg, mflags));
398 }
399 
400 int
401 ip6_encap_detach(const struct encaptab *cookie)
402 {
403 
404 	return (encap_detach(&ipv6_encaptab, cookie));
405 }
406 
407 int
408 encap6_input(struct mbuf **mp, int *offp, int proto)
409 {
410 
411 	if (encap_input(&ipv6_encaptab, *mp, *offp, proto) != IPPROTO_DONE)
412 		return (rip6_input(mp, offp, proto));
413 	return (IPPROTO_DONE);
414 }
415 #endif /* INET6 */
416