xref: /freebsd/sys/netinet/in_gif.c (revision a0e793cbf1951d07fc47a0d9ea389d7dacba5213)
1 /*	$KAME: in_gif.c,v 1.54 2001/05/14 14:02:16 itojun Exp $	*/
2 
3 /*-
4  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the project nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 
32 #include <sys/cdefs.h>
33 __FBSDID("$FreeBSD$");
34 
35 #include "opt_mrouting.h"
36 #include "opt_inet.h"
37 #include "opt_inet6.h"
38 
39 #include <sys/param.h>
40 #include <sys/systm.h>
41 #include <sys/socket.h>
42 #include <sys/sockio.h>
43 #include <sys/mbuf.h>
44 #include <sys/errno.h>
45 #include <sys/kernel.h>
46 #include <sys/sysctl.h>
47 #include <sys/protosw.h>
48 #include <sys/malloc.h>
49 
50 #include <net/if.h>
51 #include <net/if_var.h>
52 #include <net/route.h>
53 #include <net/vnet.h>
54 
55 #include <netinet/in.h>
56 #include <netinet/in_systm.h>
57 #include <netinet/ip.h>
58 #include <netinet/ip_var.h>
59 #include <netinet/in_gif.h>
60 #include <netinet/in_var.h>
61 #include <netinet/ip_encap.h>
62 #include <netinet/ip_ecn.h>
63 
64 #ifdef INET6
65 #include <netinet/ip6.h>
66 #endif
67 
68 #ifdef MROUTING
69 #include <netinet/ip_mroute.h>
70 #endif /* MROUTING */
71 
72 #include <net/if_gif.h>
73 
74 static int gif_validate4(const struct ip *, struct gif_softc *,
75 	struct ifnet *);
76 
77 extern  struct domain inetdomain;
78 struct protosw in_gif_protosw = {
79 	.pr_type =		SOCK_RAW,
80 	.pr_domain =		&inetdomain,
81 	.pr_protocol =		0/* IPPROTO_IPV[46] */,
82 	.pr_flags =		PR_ATOMIC|PR_ADDR,
83 	.pr_input =		in_gif_input,
84 	.pr_output =		(pr_output_t *)rip_output,
85 	.pr_ctloutput =		rip_ctloutput,
86 	.pr_usrreqs =		&rip_usrreqs
87 };
88 
89 VNET_DEFINE(int, ip_gif_ttl) = GIF_TTL;
90 #define	V_ip_gif_ttl		VNET(ip_gif_ttl)
91 SYSCTL_VNET_INT(_net_inet_ip, IPCTL_GIF_TTL, gifttl, CTLFLAG_RW,
92 	&VNET_NAME(ip_gif_ttl), 0, "");
93 
94 int
95 in_gif_output(struct ifnet *ifp, int family, struct mbuf *m)
96 {
97 	struct gif_softc *sc = ifp->if_softc;
98 	struct sockaddr_in *dst = (struct sockaddr_in *)&sc->gif_ro.ro_dst;
99 	struct sockaddr_in *sin_src = (struct sockaddr_in *)sc->gif_psrc;
100 	struct sockaddr_in *sin_dst = (struct sockaddr_in *)sc->gif_pdst;
101 	struct ip iphdr;	/* capsule IP header, host byte ordered */
102 	struct etherip_header eiphdr;
103 	int error, len, proto;
104 	u_int8_t tos;
105 
106 	GIF_LOCK_ASSERT(sc);
107 
108 	if (sin_src == NULL || sin_dst == NULL ||
109 	    sin_src->sin_family != AF_INET ||
110 	    sin_dst->sin_family != AF_INET) {
111 		m_freem(m);
112 		return EAFNOSUPPORT;
113 	}
114 
115 	switch (family) {
116 #ifdef INET
117 	case AF_INET:
118 	    {
119 		struct ip *ip;
120 
121 		proto = IPPROTO_IPV4;
122 		if (m->m_len < sizeof(*ip)) {
123 			m = m_pullup(m, sizeof(*ip));
124 			if (!m)
125 				return ENOBUFS;
126 		}
127 		ip = mtod(m, struct ip *);
128 		tos = ip->ip_tos;
129 		break;
130 	    }
131 #endif /* INET */
132 #ifdef INET6
133 	case AF_INET6:
134 	    {
135 		struct ip6_hdr *ip6;
136 		proto = IPPROTO_IPV6;
137 		if (m->m_len < sizeof(*ip6)) {
138 			m = m_pullup(m, sizeof(*ip6));
139 			if (!m)
140 				return ENOBUFS;
141 		}
142 		ip6 = mtod(m, struct ip6_hdr *);
143 		tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
144 		break;
145 	    }
146 #endif /* INET6 */
147 	case AF_LINK:
148  		proto = IPPROTO_ETHERIP;
149 
150 		/*
151 		 * GIF_SEND_REVETHIP (disabled by default) intentionally
152 		 * sends an EtherIP packet with revered version field in
153 		 * the header.  This is a knob for backward compatibility
154 		 * with FreeBSD 7.2R or prior.
155 		 */
156 		if ((sc->gif_options & GIF_SEND_REVETHIP)) {
157  			eiphdr.eip_ver = 0;
158  			eiphdr.eip_resvl = ETHERIP_VERSION;
159  			eiphdr.eip_resvh = 0;
160 		} else {
161  			eiphdr.eip_ver = ETHERIP_VERSION;
162  			eiphdr.eip_resvl = 0;
163  			eiphdr.eip_resvh = 0;
164 		}
165  		/* prepend Ethernet-in-IP header */
166  		M_PREPEND(m, sizeof(struct etherip_header), M_NOWAIT);
167  		if (m && m->m_len < sizeof(struct etherip_header))
168  			m = m_pullup(m, sizeof(struct etherip_header));
169  		if (m == NULL)
170  			return ENOBUFS;
171  		bcopy(&eiphdr, mtod(m, struct etherip_header *),
172 		    sizeof(struct etherip_header));
173 		tos = 0;
174 		break;
175 
176 	default:
177 #ifdef DEBUG
178 		printf("in_gif_output: warning: unknown family %d passed\n",
179 			family);
180 #endif
181 		m_freem(m);
182 		return EAFNOSUPPORT;
183 	}
184 
185 	bzero(&iphdr, sizeof(iphdr));
186 	iphdr.ip_src = sin_src->sin_addr;
187 	/* bidirectional configured tunnel mode */
188 	if (sin_dst->sin_addr.s_addr != INADDR_ANY)
189 		iphdr.ip_dst = sin_dst->sin_addr;
190 	else {
191 		m_freem(m);
192 		return ENETUNREACH;
193 	}
194 	iphdr.ip_p = proto;
195 	/* version will be set in ip_output() */
196 	iphdr.ip_ttl = V_ip_gif_ttl;
197 	iphdr.ip_len = htons(m->m_pkthdr.len + sizeof(struct ip));
198 	ip_ecn_ingress((ifp->if_flags & IFF_LINK1) ? ECN_ALLOWED : ECN_NOCARE,
199 		       &iphdr.ip_tos, &tos);
200 
201 	/* prepend new IP header */
202 	len = sizeof(struct ip);
203 #ifndef __NO_STRICT_ALIGNMENT
204 	if (family == AF_LINK)
205 		len += ETHERIP_ALIGN;
206 #endif
207 	M_PREPEND(m, len, M_NOWAIT);
208 	if (m != NULL && m->m_len < len)
209 		m = m_pullup(m, len);
210 	if (m == NULL) {
211 		printf("ENOBUFS in in_gif_output %d\n", __LINE__);
212 		return ENOBUFS;
213 	}
214 #ifndef __NO_STRICT_ALIGNMENT
215 	if (family == AF_LINK) {
216 		len = mtod(m, vm_offset_t) & 3;
217 		KASSERT(len == 0 || len == ETHERIP_ALIGN,
218 		    ("in_gif_output: unexpected misalignment"));
219 		m->m_data += len;
220 		m->m_len -= ETHERIP_ALIGN;
221 	}
222 #endif
223 	bcopy(&iphdr, mtod(m, struct ip *), sizeof(struct ip));
224 
225 	M_SETFIB(m, sc->gif_fibnum);
226 
227 	if (dst->sin_family != sin_dst->sin_family ||
228 	    dst->sin_addr.s_addr != sin_dst->sin_addr.s_addr) {
229 		/* cache route doesn't match */
230 		bzero(dst, sizeof(*dst));
231 		dst->sin_family = sin_dst->sin_family;
232 		dst->sin_len = sizeof(struct sockaddr_in);
233 		dst->sin_addr = sin_dst->sin_addr;
234 		if (sc->gif_ro.ro_rt) {
235 			RTFREE(sc->gif_ro.ro_rt);
236 			sc->gif_ro.ro_rt = NULL;
237 		}
238 #if 0
239 		GIF2IFP(sc)->if_mtu = GIF_MTU;
240 #endif
241 	}
242 
243 	if (sc->gif_ro.ro_rt == NULL) {
244 		in_rtalloc_ign(&sc->gif_ro, 0, sc->gif_fibnum);
245 		if (sc->gif_ro.ro_rt == NULL) {
246 			m_freem(m);
247 			return ENETUNREACH;
248 		}
249 
250 		/* if it constitutes infinite encapsulation, punt. */
251 		if (sc->gif_ro.ro_rt->rt_ifp == ifp) {
252 			m_freem(m);
253 			return ENETUNREACH;	/* XXX */
254 		}
255 #if 0
256 		ifp->if_mtu = sc->gif_ro.ro_rt->rt_ifp->if_mtu
257 			- sizeof(struct ip);
258 #endif
259 	}
260 
261 	m->m_flags &= ~(M_BCAST|M_MCAST);
262 	error = ip_output(m, NULL, &sc->gif_ro, 0, NULL, NULL);
263 
264 	if (!(GIF2IFP(sc)->if_flags & IFF_LINK0) &&
265 	    sc->gif_ro.ro_rt != NULL) {
266 		RTFREE(sc->gif_ro.ro_rt);
267 		sc->gif_ro.ro_rt = NULL;
268 	}
269 
270 	return (error);
271 }
272 
273 int
274 in_gif_input(struct mbuf **mp, int *offp, int proto)
275 {
276 	struct mbuf *m;
277 	struct ifnet *gifp = NULL;
278 	struct gif_softc *sc;
279 	struct ip *ip;
280 	int af;
281 	int off;
282 	u_int8_t otos;
283 
284 	m = *mp;
285 	ip = mtod(m, struct ip *);
286 	off = *offp;
287 	*mp = NULL;
288 
289 	sc = (struct gif_softc *)encap_getarg(m);
290 	if (sc == NULL) {
291 		m_freem(m);
292 		KMOD_IPSTAT_INC(ips_nogif);
293 		return (IPPROTO_DONE);
294 	}
295 
296 	gifp = GIF2IFP(sc);
297 	if (gifp == NULL || (gifp->if_flags & IFF_UP) == 0) {
298 		m_freem(m);
299 		KMOD_IPSTAT_INC(ips_nogif);
300 		return (IPPROTO_DONE);
301 	}
302 
303 	otos = ip->ip_tos;
304 	m_adj(m, off);
305 
306 	switch (proto) {
307 #ifdef INET
308 	case IPPROTO_IPV4:
309 	    {
310 		struct ip *ip;
311 		af = AF_INET;
312 		if (m->m_len < sizeof(*ip)) {
313 			m = m_pullup(m, sizeof(*ip));
314 			if (!m)
315 				return (IPPROTO_DONE);
316 		}
317 		ip = mtod(m, struct ip *);
318 		if (ip_ecn_egress((gifp->if_flags & IFF_LINK1) ?
319 				  ECN_ALLOWED : ECN_NOCARE,
320 				  &otos, &ip->ip_tos) == 0) {
321 			m_freem(m);
322 			return (IPPROTO_DONE);
323 		}
324 		break;
325 	    }
326 #endif
327 #ifdef INET6
328 	case IPPROTO_IPV6:
329 	    {
330 		struct ip6_hdr *ip6;
331 		u_int8_t itos, oitos;
332 
333 		af = AF_INET6;
334 		if (m->m_len < sizeof(*ip6)) {
335 			m = m_pullup(m, sizeof(*ip6));
336 			if (!m)
337 				return (IPPROTO_DONE);
338 		}
339 		ip6 = mtod(m, struct ip6_hdr *);
340 		itos = oitos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
341 		if (ip_ecn_egress((gifp->if_flags & IFF_LINK1) ?
342 				  ECN_ALLOWED : ECN_NOCARE,
343 				  &otos, &itos) == 0) {
344 			m_freem(m);
345 			return (IPPROTO_DONE);
346 		}
347 		if (itos != oitos) {
348 			ip6->ip6_flow &= ~htonl(0xff << 20);
349 			ip6->ip6_flow |= htonl((u_int32_t)itos << 20);
350 		}
351 		break;
352 	    }
353 #endif /* INET6 */
354  	case IPPROTO_ETHERIP:
355  		af = AF_LINK;
356  		break;
357 
358 	default:
359 		KMOD_IPSTAT_INC(ips_nogif);
360 		m_freem(m);
361 		return (IPPROTO_DONE);
362 	}
363 	gif_input(m, af, gifp);
364 	return (IPPROTO_DONE);
365 }
366 
367 /*
368  * validate outer address.
369  */
370 static int
371 gif_validate4(const struct ip *ip, struct gif_softc *sc, struct ifnet *ifp)
372 {
373 	struct sockaddr_in *src, *dst;
374 	struct in_ifaddr *ia4;
375 
376 	src = (struct sockaddr_in *)sc->gif_psrc;
377 	dst = (struct sockaddr_in *)sc->gif_pdst;
378 
379 	/* check for address match */
380 	if (src->sin_addr.s_addr != ip->ip_dst.s_addr ||
381 	    dst->sin_addr.s_addr != ip->ip_src.s_addr)
382 		return 0;
383 
384 	/* martian filters on outer source - NOT done in ip_input! */
385 	if (IN_MULTICAST(ntohl(ip->ip_src.s_addr)))
386 		return 0;
387 	switch ((ntohl(ip->ip_src.s_addr) & 0xff000000) >> 24) {
388 	case 0: case 127: case 255:
389 		return 0;
390 	}
391 
392 	/* reject packets with broadcast on source */
393 	/* XXXRW: should use hash lists? */
394 	IN_IFADDR_RLOCK();
395 	TAILQ_FOREACH(ia4, &V_in_ifaddrhead, ia_link) {
396 		if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0)
397 			continue;
398 		if (ip->ip_src.s_addr == ia4->ia_broadaddr.sin_addr.s_addr) {
399 			IN_IFADDR_RUNLOCK();
400 			return 0;
401 		}
402 	}
403 	IN_IFADDR_RUNLOCK();
404 
405 	/* ingress filters on outer source */
406 	if ((GIF2IFP(sc)->if_flags & IFF_LINK2) == 0 && ifp) {
407 		struct sockaddr_in sin;
408 		struct rtentry *rt;
409 
410 		bzero(&sin, sizeof(sin));
411 		sin.sin_family = AF_INET;
412 		sin.sin_len = sizeof(struct sockaddr_in);
413 		sin.sin_addr = ip->ip_src;
414 		/* XXX MRT  check for the interface we would use on output */
415 		rt = in_rtalloc1((struct sockaddr *)&sin, 0,
416 		    0UL, sc->gif_fibnum);
417 		if (!rt || rt->rt_ifp != ifp) {
418 #if 0
419 			log(LOG_WARNING, "%s: packet from 0x%x dropped "
420 			    "due to ingress filter\n", if_name(GIF2IFP(sc)),
421 			    (u_int32_t)ntohl(sin.sin_addr.s_addr));
422 #endif
423 			if (rt)
424 				RTFREE_LOCKED(rt);
425 			return 0;
426 		}
427 		RTFREE_LOCKED(rt);
428 	}
429 
430 	return 32 * 2;
431 }
432 
433 /*
434  * we know that we are in IFF_UP, outer address available, and outer family
435  * matched the physical addr family.  see gif_encapcheck().
436  */
437 int
438 gif_encapcheck4(const struct mbuf *m, int off, int proto, void *arg)
439 {
440 	struct ip ip;
441 	struct gif_softc *sc;
442 	struct ifnet *ifp;
443 
444 	/* sanity check done in caller */
445 	sc = (struct gif_softc *)arg;
446 
447 	/* LINTED const cast */
448 	m_copydata(m, 0, sizeof(ip), (caddr_t)&ip);
449 	ifp = ((m->m_flags & M_PKTHDR) != 0) ? m->m_pkthdr.rcvif : NULL;
450 
451 	return gif_validate4(&ip, sc, ifp);
452 }
453 
454 int
455 in_gif_attach(struct gif_softc *sc)
456 {
457 	sc->encap_cookie4 = encap_attach_func(AF_INET, -1, gif_encapcheck,
458 	    &in_gif_protosw, sc);
459 	if (sc->encap_cookie4 == NULL)
460 		return EEXIST;
461 	return 0;
462 }
463 
464 int
465 in_gif_detach(struct gif_softc *sc)
466 {
467 	int error;
468 
469 	error = encap_detach(sc->encap_cookie4);
470 	if (error == 0)
471 		sc->encap_cookie4 = NULL;
472 	return error;
473 }
474