xref: /freebsd/sys/netinet6/in6_gif.c (revision ab0b9f6b3073e6c4d1dfbf07444d7db67a189a96) 
1 /*-
2  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
3  * All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 3. Neither the name of the project nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  *
29  *	$KAME: in6_gif.c,v 1.49 2001/05/14 14:02:17 itojun Exp $
30  */
31 
32 #include <sys/cdefs.h>
33 __FBSDID("$FreeBSD$");
34 
35 #include "opt_inet.h"
36 #include "opt_inet6.h"
37 
38 #include <sys/param.h>
39 #include <sys/systm.h>
40 #include <sys/socket.h>
41 #include <sys/sockio.h>
42 #include <sys/mbuf.h>
43 #include <sys/errno.h>
44 #include <sys/kernel.h>
45 #include <sys/queue.h>
46 #include <sys/syslog.h>
47 #include <sys/sysctl.h>
48 #include <sys/protosw.h>
49 #include <sys/malloc.h>
50 
51 #include <net/if.h>
52 #include <net/if_var.h>
53 #include <net/route.h>
54 #include <net/vnet.h>
55 
56 #include <netinet/in.h>
57 #include <netinet/in_systm.h>
58 #ifdef INET
59 #include <netinet/ip.h>
60 #endif
61 #include <netinet/ip_encap.h>
62 #ifdef INET6
63 #include <netinet/ip6.h>
64 #include <netinet6/ip6_var.h>
65 #include <netinet6/in6_gif.h>
66 #include <netinet6/in6_var.h>
67 #endif
68 #include <netinet6/ip6protosw.h>
69 #include <netinet/ip_ecn.h>
70 #ifdef INET6
71 #include <netinet6/ip6_ecn.h>
72 #endif
73 
74 #include <net/if_gif.h>
75 
76 VNET_DEFINE(int, ip6_gif_hlim) = GIF_HLIM;
77 #define	V_ip6_gif_hlim			VNET(ip6_gif_hlim)
78 
79 SYSCTL_DECL(_net_inet6_ip6);
80 SYSCTL_VNET_INT(_net_inet6_ip6, IPV6CTL_GIF_HLIM, gifhlim, CTLFLAG_RW,
81     &VNET_NAME(ip6_gif_hlim), 0, "");
82 
83 static int gif_validate6(const struct ip6_hdr *, struct gif_softc *,
84 			 struct ifnet *);
85 
86 extern  struct domain inet6domain;
87 struct ip6protosw in6_gif_protosw = {
88 	.pr_type =	SOCK_RAW,
89 	.pr_domain =	&inet6domain,
90 	.pr_protocol =	0,			/* IPPROTO_IPV[46] */
91 	.pr_flags =	PR_ATOMIC|PR_ADDR,
92 	.pr_input =	in6_gif_input,
93 	.pr_output =	rip6_output,
94 	.pr_ctloutput =	rip6_ctloutput,
95 	.pr_usrreqs =	&rip6_usrreqs
96 };
97 
98 int
99 in6_gif_output(struct ifnet *ifp,
100     int family,			/* family of the packet to be encapsulate */
101     struct mbuf *m)
102 {
103 	struct gif_softc *sc = ifp->if_softc;
104 	struct sockaddr_in6 *dst = (struct sockaddr_in6 *)&sc->gif_ro6.ro_dst;
105 	struct sockaddr_in6 *sin6_src = (struct sockaddr_in6 *)sc->gif_psrc;
106 	struct sockaddr_in6 *sin6_dst = (struct sockaddr_in6 *)sc->gif_pdst;
107 	struct ip6_hdr *ip6;
108 	struct etherip_header eiphdr;
109 	int error, len, proto;
110 	u_int8_t itos, otos;
111 
112 	GIF_LOCK_ASSERT(sc);
113 
114 	if (sin6_src == NULL || sin6_dst == NULL ||
115 	    sin6_src->sin6_family != AF_INET6 ||
116 	    sin6_dst->sin6_family != AF_INET6) {
117 		m_freem(m);
118 		return EAFNOSUPPORT;
119 	}
120 
121 	switch (family) {
122 #ifdef INET
123 	case AF_INET:
124 	    {
125 		struct ip *ip;
126 
127 		proto = IPPROTO_IPV4;
128 		if (m->m_len < sizeof(*ip)) {
129 			m = m_pullup(m, sizeof(*ip));
130 			if (!m)
131 				return ENOBUFS;
132 		}
133 		ip = mtod(m, struct ip *);
134 		itos = ip->ip_tos;
135 		break;
136 	    }
137 #endif
138 #ifdef INET6
139 	case AF_INET6:
140 	    {
141 		struct ip6_hdr *ip6;
142 		proto = IPPROTO_IPV6;
143 		if (m->m_len < sizeof(*ip6)) {
144 			m = m_pullup(m, sizeof(*ip6));
145 			if (!m)
146 				return ENOBUFS;
147 		}
148 		ip6 = mtod(m, struct ip6_hdr *);
149 		itos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
150 		break;
151 	    }
152 #endif
153 	case AF_LINK:
154 		proto = IPPROTO_ETHERIP;
155 
156 		/*
157 		 * GIF_SEND_REVETHIP (disabled by default) intentionally
158 		 * sends an EtherIP packet with revered version field in
159 		 * the header.  This is a knob for backward compatibility
160 		 * with FreeBSD 7.2R or prior.
161 		 */
162 		if ((sc->gif_options & GIF_SEND_REVETHIP)) {
163 			eiphdr.eip_ver = 0;
164 			eiphdr.eip_resvl = ETHERIP_VERSION;
165 			eiphdr.eip_resvh = 0;
166 		} else {
167 			eiphdr.eip_ver = ETHERIP_VERSION;
168 			eiphdr.eip_resvl = 0;
169 			eiphdr.eip_resvh = 0;
170 		}
171 		/* prepend Ethernet-in-IP header */
172 		M_PREPEND(m, sizeof(struct etherip_header), M_NOWAIT);
173 		if (m && m->m_len < sizeof(struct etherip_header))
174 			m = m_pullup(m, sizeof(struct etherip_header));
175 		if (m == NULL)
176 			return ENOBUFS;
177 		bcopy(&eiphdr, mtod(m, struct etherip_header *),
178 		    sizeof(struct etherip_header));
179 		break;
180 
181 	default:
182 #ifdef DEBUG
183 		printf("in6_gif_output: warning: unknown family %d passed\n",
184 			family);
185 #endif
186 		m_freem(m);
187 		return EAFNOSUPPORT;
188 	}
189 
190 	/* prepend new IP header */
191 	len = sizeof(struct ip6_hdr);
192 #ifndef __NO_STRICT_ALIGNMENT
193 	if (family == AF_LINK)
194 		len += ETHERIP_ALIGN;
195 #endif
196 	M_PREPEND(m, len, M_NOWAIT);
197 	if (m != NULL && m->m_len < len)
198 		m = m_pullup(m, len);
199 	if (m == NULL) {
200 		printf("ENOBUFS in in6_gif_output %d\n", __LINE__);
201 		return ENOBUFS;
202 	}
203 #ifndef __NO_STRICT_ALIGNMENT
204 	if (family == AF_LINK) {
205 		len = mtod(m, vm_offset_t) & 3;
206 		KASSERT(len == 0 || len == ETHERIP_ALIGN,
207 		    ("in6_gif_output: unexpected misalignment"));
208 		m->m_data += len;
209 		m->m_len -= ETHERIP_ALIGN;
210 	}
211 #endif
212 
213 	ip6 = mtod(m, struct ip6_hdr *);
214 	ip6->ip6_flow	= 0;
215 	ip6->ip6_vfc	&= ~IPV6_VERSION_MASK;
216 	ip6->ip6_vfc	|= IPV6_VERSION;
217 	ip6->ip6_plen	= htons((u_short)m->m_pkthdr.len);
218 	ip6->ip6_nxt	= proto;
219 	ip6->ip6_hlim	= V_ip6_gif_hlim;
220 	ip6->ip6_src	= sin6_src->sin6_addr;
221 	/* bidirectional configured tunnel mode */
222 	if (!IN6_IS_ADDR_UNSPECIFIED(&sin6_dst->sin6_addr))
223 		ip6->ip6_dst = sin6_dst->sin6_addr;
224 	else  {
225 		m_freem(m);
226 		return ENETUNREACH;
227 	}
228 	ip_ecn_ingress((ifp->if_flags & IFF_LINK1) ? ECN_ALLOWED : ECN_NOCARE,
229 		       &otos, &itos);
230 	ip6->ip6_flow &= ~htonl(0xff << 20);
231 	ip6->ip6_flow |= htonl((u_int32_t)otos << 20);
232 
233 	M_SETFIB(m, sc->gif_fibnum);
234 
235 	if (dst->sin6_family != sin6_dst->sin6_family ||
236 	     !IN6_ARE_ADDR_EQUAL(&dst->sin6_addr, &sin6_dst->sin6_addr)) {
237 		/* cache route doesn't match */
238 		bzero(dst, sizeof(*dst));
239 		dst->sin6_family = sin6_dst->sin6_family;
240 		dst->sin6_len = sizeof(struct sockaddr_in6);
241 		dst->sin6_addr = sin6_dst->sin6_addr;
242 		if (sc->gif_ro6.ro_rt) {
243 			RTFREE(sc->gif_ro6.ro_rt);
244 			sc->gif_ro6.ro_rt = NULL;
245 		}
246 #if 0
247 		GIF2IFP(sc)->if_mtu = GIF_MTU;
248 #endif
249 	}
250 
251 	if (sc->gif_ro6.ro_rt == NULL) {
252 		in6_rtalloc(&sc->gif_ro6, sc->gif_fibnum);
253 		if (sc->gif_ro6.ro_rt == NULL) {
254 			m_freem(m);
255 			return ENETUNREACH;
256 		}
257 
258 		/* if it constitutes infinite encapsulation, punt. */
259 		if (sc->gif_ro.ro_rt->rt_ifp == ifp) {
260 			m_freem(m);
261 			return ENETUNREACH;	/*XXX*/
262 		}
263 #if 0
264 		ifp->if_mtu = sc->gif_ro6.ro_rt->rt_ifp->if_mtu
265 			- sizeof(struct ip6_hdr);
266 #endif
267 	}
268 
269 #ifdef IPV6_MINMTU
270 	/*
271 	 * force fragmentation to minimum MTU, to avoid path MTU discovery.
272 	 * it is too painful to ask for resend of inner packet, to achieve
273 	 * path MTU discovery for encapsulated packets.
274 	 */
275 	error = ip6_output(m, 0, &sc->gif_ro6, IPV6_MINMTU, 0, NULL, NULL);
276 #else
277 	error = ip6_output(m, 0, &sc->gif_ro6, 0, 0, NULL, NULL);
278 #endif
279 
280 	if (!(GIF2IFP(sc)->if_flags & IFF_LINK0) &&
281 	    sc->gif_ro6.ro_rt != NULL) {
282 		RTFREE(sc->gif_ro6.ro_rt);
283 		sc->gif_ro6.ro_rt = NULL;
284 	}
285 
286 	return (error);
287 }
288 
289 int
290 in6_gif_input(struct mbuf **mp, int *offp, int proto)
291 {
292 	struct mbuf *m = *mp;
293 	struct ifnet *gifp = NULL;
294 	struct gif_softc *sc;
295 	struct ip6_hdr *ip6;
296 	int af = 0;
297 	u_int32_t otos;
298 
299 	ip6 = mtod(m, struct ip6_hdr *);
300 
301 	sc = (struct gif_softc *)encap_getarg(m);
302 	if (sc == NULL) {
303 		m_freem(m);
304 		IP6STAT_INC(ip6s_nogif);
305 		return IPPROTO_DONE;
306 	}
307 
308 	gifp = GIF2IFP(sc);
309 	if (gifp == NULL || (gifp->if_flags & IFF_UP) == 0) {
310 		m_freem(m);
311 		IP6STAT_INC(ip6s_nogif);
312 		return IPPROTO_DONE;
313 	}
314 
315 	otos = ip6->ip6_flow;
316 	m_adj(m, *offp);
317 
318 	switch (proto) {
319 #ifdef INET
320 	case IPPROTO_IPV4:
321 	    {
322 		struct ip *ip;
323 		u_int8_t otos8;
324 		af = AF_INET;
325 		otos8 = (ntohl(otos) >> 20) & 0xff;
326 		if (m->m_len < sizeof(*ip)) {
327 			m = m_pullup(m, sizeof(*ip));
328 			if (!m)
329 				return IPPROTO_DONE;
330 		}
331 		ip = mtod(m, struct ip *);
332 		if (ip_ecn_egress((gifp->if_flags & IFF_LINK1) ?
333 				  ECN_ALLOWED : ECN_NOCARE,
334 				  &otos8, &ip->ip_tos) == 0) {
335 			m_freem(m);
336 			return IPPROTO_DONE;
337 		}
338 		break;
339 	    }
340 #endif /* INET */
341 #ifdef INET6
342 	case IPPROTO_IPV6:
343 	    {
344 		struct ip6_hdr *ip6;
345 		af = AF_INET6;
346 		if (m->m_len < sizeof(*ip6)) {
347 			m = m_pullup(m, sizeof(*ip6));
348 			if (!m)
349 				return IPPROTO_DONE;
350 		}
351 		ip6 = mtod(m, struct ip6_hdr *);
352 		if (ip6_ecn_egress((gifp->if_flags & IFF_LINK1) ?
353 				   ECN_ALLOWED : ECN_NOCARE,
354 				   &otos, &ip6->ip6_flow) == 0) {
355 			m_freem(m);
356 			return IPPROTO_DONE;
357 		}
358 		break;
359 	    }
360 #endif
361 	case IPPROTO_ETHERIP:
362 		af = AF_LINK;
363 		break;
364 
365 	default:
366 		IP6STAT_INC(ip6s_nogif);
367 		m_freem(m);
368 		return IPPROTO_DONE;
369 	}
370 
371 	gif_input(m, af, gifp);
372 	return IPPROTO_DONE;
373 }
374 
375 /*
376  * validate outer address.
377  */
378 static int
379 gif_validate6(const struct ip6_hdr *ip6, struct gif_softc *sc,
380     struct ifnet *ifp)
381 {
382 	struct sockaddr_in6 *src, *dst;
383 
384 	src = (struct sockaddr_in6 *)sc->gif_psrc;
385 	dst = (struct sockaddr_in6 *)sc->gif_pdst;
386 
387 	/*
388 	 * Check for address match.  Note that the check is for an incoming
389 	 * packet.  We should compare the *source* address in our configuration
390 	 * and the *destination* address of the packet, and vice versa.
391 	 */
392 	if (!IN6_ARE_ADDR_EQUAL(&src->sin6_addr, &ip6->ip6_dst) ||
393 	    !IN6_ARE_ADDR_EQUAL(&dst->sin6_addr, &ip6->ip6_src))
394 		return 0;
395 
396 	/* martian filters on outer source - done in ip6_input */
397 
398 	/* ingress filters on outer source */
399 	if ((GIF2IFP(sc)->if_flags & IFF_LINK2) == 0 && ifp) {
400 		struct sockaddr_in6 sin6;
401 		struct rtentry *rt;
402 
403 		bzero(&sin6, sizeof(sin6));
404 		sin6.sin6_family = AF_INET6;
405 		sin6.sin6_len = sizeof(struct sockaddr_in6);
406 		sin6.sin6_addr = ip6->ip6_src;
407 		sin6.sin6_scope_id = 0; /* XXX */
408 
409 		rt = in6_rtalloc1((struct sockaddr *)&sin6, 0, 0UL,
410 		    sc->gif_fibnum);
411 		if (!rt || rt->rt_ifp != ifp) {
412 #if 0
413 			char ip6buf[INET6_ADDRSTRLEN];
414 			log(LOG_WARNING, "%s: packet from %s dropped "
415 			    "due to ingress filter\n", if_name(GIF2IFP(sc)),
416 			    ip6_sprintf(ip6buf, &sin6.sin6_addr));
417 #endif
418 			if (rt)
419 				RTFREE_LOCKED(rt);
420 			return 0;
421 		}
422 		RTFREE_LOCKED(rt);
423 	}
424 
425 	return 128 * 2;
426 }
427 
428 /*
429  * we know that we are in IFF_UP, outer address available, and outer family
430  * matched the physical addr family.  see gif_encapcheck().
431  * sanity check for arg should have been done in the caller.
432  */
433 int
434 gif_encapcheck6(const struct mbuf *m, int off, int proto, void *arg)
435 {
436 	struct ip6_hdr ip6;
437 	struct gif_softc *sc;
438 	struct ifnet *ifp;
439 
440 	/* sanity check done in caller */
441 	sc = (struct gif_softc *)arg;
442 
443 	/* LINTED const cast */
444 	m_copydata(m, 0, sizeof(ip6), (caddr_t)&ip6);
445 	ifp = ((m->m_flags & M_PKTHDR) != 0) ? m->m_pkthdr.rcvif : NULL;
446 
447 	return gif_validate6(&ip6, sc, ifp);
448 }
449 
450 int
451 in6_gif_attach(struct gif_softc *sc)
452 {
453 	sc->encap_cookie6 = encap_attach_func(AF_INET6, -1, gif_encapcheck,
454 	    (void *)&in6_gif_protosw, sc);
455 	if (sc->encap_cookie6 == NULL)
456 		return EEXIST;
457 	return 0;
458 }
459 
460 int
461 in6_gif_detach(struct gif_softc *sc)
462 {
463 	int error;
464 
465 	error = encap_detach(sc->encap_cookie6);
466 	if (error == 0)
467 		sc->encap_cookie6 = NULL;
468 	return error;
469 }
470