xref: /freebsd/sys/netinet6/in6_gif.c (revision ee7b0571c2c18bdec848ed2044223cc88db29bd8) 
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 		itos = 0;
180 		break;
181 
182 	default:
183 #ifdef DEBUG
184 		printf("in6_gif_output: warning: unknown family %d passed\n",
185 			family);
186 #endif
187 		m_freem(m);
188 		return EAFNOSUPPORT;
189 	}
190 
191 	/* prepend new IP header */
192 	len = sizeof(struct ip6_hdr);
193 #ifndef __NO_STRICT_ALIGNMENT
194 	if (family == AF_LINK)
195 		len += ETHERIP_ALIGN;
196 #endif
197 	M_PREPEND(m, len, M_NOWAIT);
198 	if (m != NULL && m->m_len < len)
199 		m = m_pullup(m, len);
200 	if (m == NULL) {
201 		printf("ENOBUFS in in6_gif_output %d\n", __LINE__);
202 		return ENOBUFS;
203 	}
204 #ifndef __NO_STRICT_ALIGNMENT
205 	if (family == AF_LINK) {
206 		len = mtod(m, vm_offset_t) & 3;
207 		KASSERT(len == 0 || len == ETHERIP_ALIGN,
208 		    ("in6_gif_output: unexpected misalignment"));
209 		m->m_data += len;
210 		m->m_len -= ETHERIP_ALIGN;
211 	}
212 #endif
213 
214 	ip6 = mtod(m, struct ip6_hdr *);
215 	ip6->ip6_flow	= 0;
216 	ip6->ip6_vfc	&= ~IPV6_VERSION_MASK;
217 	ip6->ip6_vfc	|= IPV6_VERSION;
218 	ip6->ip6_plen	= htons((u_short)m->m_pkthdr.len);
219 	ip6->ip6_nxt	= proto;
220 	ip6->ip6_hlim	= V_ip6_gif_hlim;
221 	ip6->ip6_src	= sin6_src->sin6_addr;
222 	/* bidirectional configured tunnel mode */
223 	if (!IN6_IS_ADDR_UNSPECIFIED(&sin6_dst->sin6_addr))
224 		ip6->ip6_dst = sin6_dst->sin6_addr;
225 	else  {
226 		m_freem(m);
227 		return ENETUNREACH;
228 	}
229 	ip_ecn_ingress((ifp->if_flags & IFF_LINK1) ? ECN_ALLOWED : ECN_NOCARE,
230 		       &otos, &itos);
231 	ip6->ip6_flow &= ~htonl(0xff << 20);
232 	ip6->ip6_flow |= htonl((u_int32_t)otos << 20);
233 
234 	M_SETFIB(m, sc->gif_fibnum);
235 
236 	if (dst->sin6_family != sin6_dst->sin6_family ||
237 	     !IN6_ARE_ADDR_EQUAL(&dst->sin6_addr, &sin6_dst->sin6_addr)) {
238 		/* cache route doesn't match */
239 		bzero(dst, sizeof(*dst));
240 		dst->sin6_family = sin6_dst->sin6_family;
241 		dst->sin6_len = sizeof(struct sockaddr_in6);
242 		dst->sin6_addr = sin6_dst->sin6_addr;
243 		if (sc->gif_ro6.ro_rt) {
244 			RTFREE(sc->gif_ro6.ro_rt);
245 			sc->gif_ro6.ro_rt = NULL;
246 		}
247 #if 0
248 		GIF2IFP(sc)->if_mtu = GIF_MTU;
249 #endif
250 	}
251 
252 	if (sc->gif_ro6.ro_rt == NULL) {
253 		in6_rtalloc(&sc->gif_ro6, sc->gif_fibnum);
254 		if (sc->gif_ro6.ro_rt == NULL) {
255 			m_freem(m);
256 			return ENETUNREACH;
257 		}
258 
259 		/* if it constitutes infinite encapsulation, punt. */
260 		if (sc->gif_ro.ro_rt->rt_ifp == ifp) {
261 			m_freem(m);
262 			return ENETUNREACH;	/*XXX*/
263 		}
264 #if 0
265 		ifp->if_mtu = sc->gif_ro6.ro_rt->rt_ifp->if_mtu
266 			- sizeof(struct ip6_hdr);
267 #endif
268 	}
269 
270 	m->m_flags &= ~(M_BCAST|M_MCAST);
271 #ifdef IPV6_MINMTU
272 	/*
273 	 * force fragmentation to minimum MTU, to avoid path MTU discovery.
274 	 * it is too painful to ask for resend of inner packet, to achieve
275 	 * path MTU discovery for encapsulated packets.
276 	 */
277 	error = ip6_output(m, 0, &sc->gif_ro6, IPV6_MINMTU, 0, NULL, NULL);
278 #else
279 	error = ip6_output(m, 0, &sc->gif_ro6, 0, 0, NULL, NULL);
280 #endif
281 
282 	if (!(GIF2IFP(sc)->if_flags & IFF_LINK0) &&
283 	    sc->gif_ro6.ro_rt != NULL) {
284 		RTFREE(sc->gif_ro6.ro_rt);
285 		sc->gif_ro6.ro_rt = NULL;
286 	}
287 
288 	return (error);
289 }
290 
291 int
292 in6_gif_input(struct mbuf **mp, int *offp, int proto)
293 {
294 	struct mbuf *m = *mp;
295 	struct ifnet *gifp = NULL;
296 	struct gif_softc *sc;
297 	struct ip6_hdr *ip6;
298 	int af = 0;
299 	u_int32_t otos;
300 
301 	ip6 = mtod(m, struct ip6_hdr *);
302 
303 	sc = (struct gif_softc *)encap_getarg(m);
304 	if (sc == NULL) {
305 		m_freem(m);
306 		IP6STAT_INC(ip6s_nogif);
307 		return IPPROTO_DONE;
308 	}
309 
310 	gifp = GIF2IFP(sc);
311 	if (gifp == NULL || (gifp->if_flags & IFF_UP) == 0) {
312 		m_freem(m);
313 		IP6STAT_INC(ip6s_nogif);
314 		return IPPROTO_DONE;
315 	}
316 
317 	otos = ip6->ip6_flow;
318 	m_adj(m, *offp);
319 
320 	switch (proto) {
321 #ifdef INET
322 	case IPPROTO_IPV4:
323 	    {
324 		struct ip *ip;
325 		u_int8_t otos8;
326 		af = AF_INET;
327 		otos8 = (ntohl(otos) >> 20) & 0xff;
328 		if (m->m_len < sizeof(*ip)) {
329 			m = m_pullup(m, sizeof(*ip));
330 			if (!m)
331 				return IPPROTO_DONE;
332 		}
333 		ip = mtod(m, struct ip *);
334 		if (ip_ecn_egress((gifp->if_flags & IFF_LINK1) ?
335 				  ECN_ALLOWED : ECN_NOCARE,
336 				  &otos8, &ip->ip_tos) == 0) {
337 			m_freem(m);
338 			return IPPROTO_DONE;
339 		}
340 		break;
341 	    }
342 #endif /* INET */
343 #ifdef INET6
344 	case IPPROTO_IPV6:
345 	    {
346 		struct ip6_hdr *ip6;
347 		af = AF_INET6;
348 		if (m->m_len < sizeof(*ip6)) {
349 			m = m_pullup(m, sizeof(*ip6));
350 			if (!m)
351 				return IPPROTO_DONE;
352 		}
353 		ip6 = mtod(m, struct ip6_hdr *);
354 		if (ip6_ecn_egress((gifp->if_flags & IFF_LINK1) ?
355 				   ECN_ALLOWED : ECN_NOCARE,
356 				   &otos, &ip6->ip6_flow) == 0) {
357 			m_freem(m);
358 			return IPPROTO_DONE;
359 		}
360 		break;
361 	    }
362 #endif
363 	case IPPROTO_ETHERIP:
364 		af = AF_LINK;
365 		break;
366 
367 	default:
368 		IP6STAT_INC(ip6s_nogif);
369 		m_freem(m);
370 		return IPPROTO_DONE;
371 	}
372 
373 	gif_input(m, af, gifp);
374 	return IPPROTO_DONE;
375 }
376 
377 /*
378  * validate outer address.
379  */
380 static int
381 gif_validate6(const struct ip6_hdr *ip6, struct gif_softc *sc,
382     struct ifnet *ifp)
383 {
384 	struct sockaddr_in6 *src, *dst;
385 
386 	src = (struct sockaddr_in6 *)sc->gif_psrc;
387 	dst = (struct sockaddr_in6 *)sc->gif_pdst;
388 
389 	/*
390 	 * Check for address match.  Note that the check is for an incoming
391 	 * packet.  We should compare the *source* address in our configuration
392 	 * and the *destination* address of the packet, and vice versa.
393 	 */
394 	if (!IN6_ARE_ADDR_EQUAL(&src->sin6_addr, &ip6->ip6_dst) ||
395 	    !IN6_ARE_ADDR_EQUAL(&dst->sin6_addr, &ip6->ip6_src))
396 		return 0;
397 
398 	/* martian filters on outer source - done in ip6_input */
399 
400 	/* ingress filters on outer source */
401 	if ((GIF2IFP(sc)->if_flags & IFF_LINK2) == 0 && ifp) {
402 		struct sockaddr_in6 sin6;
403 		struct rtentry *rt;
404 
405 		bzero(&sin6, sizeof(sin6));
406 		sin6.sin6_family = AF_INET6;
407 		sin6.sin6_len = sizeof(struct sockaddr_in6);
408 		sin6.sin6_addr = ip6->ip6_src;
409 		sin6.sin6_scope_id = 0; /* XXX */
410 
411 		rt = in6_rtalloc1((struct sockaddr *)&sin6, 0, 0UL,
412 		    sc->gif_fibnum);
413 		if (!rt || rt->rt_ifp != ifp) {
414 #if 0
415 			char ip6buf[INET6_ADDRSTRLEN];
416 			log(LOG_WARNING, "%s: packet from %s dropped "
417 			    "due to ingress filter\n", if_name(GIF2IFP(sc)),
418 			    ip6_sprintf(ip6buf, &sin6.sin6_addr));
419 #endif
420 			if (rt)
421 				RTFREE_LOCKED(rt);
422 			return 0;
423 		}
424 		RTFREE_LOCKED(rt);
425 	}
426 
427 	return 128 * 2;
428 }
429 
430 /*
431  * we know that we are in IFF_UP, outer address available, and outer family
432  * matched the physical addr family.  see gif_encapcheck().
433  * sanity check for arg should have been done in the caller.
434  */
435 int
436 gif_encapcheck6(const struct mbuf *m, int off, int proto, void *arg)
437 {
438 	struct ip6_hdr ip6;
439 	struct gif_softc *sc;
440 	struct ifnet *ifp;
441 
442 	/* sanity check done in caller */
443 	sc = (struct gif_softc *)arg;
444 
445 	/* LINTED const cast */
446 	m_copydata(m, 0, sizeof(ip6), (caddr_t)&ip6);
447 	ifp = ((m->m_flags & M_PKTHDR) != 0) ? m->m_pkthdr.rcvif : NULL;
448 
449 	return gif_validate6(&ip6, sc, ifp);
450 }
451 
452 int
453 in6_gif_attach(struct gif_softc *sc)
454 {
455 	sc->encap_cookie6 = encap_attach_func(AF_INET6, -1, gif_encapcheck,
456 	    (void *)&in6_gif_protosw, sc);
457 	if (sc->encap_cookie6 == NULL)
458 		return EEXIST;
459 	return 0;
460 }
461 
462 int
463 in6_gif_detach(struct gif_softc *sc)
464 {
465 	int error;
466 
467 	error = encap_detach(sc->encap_cookie6);
468 	if (error == 0)
469 		sc->encap_cookie6 = NULL;
470 	return error;
471 }
472