xref: /freebsd/sys/netinet/in_gif.c (revision 830940567b49bb0c08dfaed40418999e76616909) 
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/route.h>
52 #include <net/vnet.h>
53 
54 #include <netinet/in.h>
55 #include <netinet/in_systm.h>
56 #include <netinet/ip.h>
57 #include <netinet/ip_var.h>
58 #include <netinet/in_gif.h>
59 #include <netinet/in_var.h>
60 #include <netinet/ip_encap.h>
61 #include <netinet/ip_ecn.h>
62 
63 #ifdef INET6
64 #include <netinet/ip6.h>
65 #endif
66 
67 #ifdef MROUTING
68 #include <netinet/ip_mroute.h>
69 #endif /* MROUTING */
70 
71 #include <net/if_gif.h>
72 
73 static int gif_validate4(const struct ip *, struct gif_softc *,
74 	struct ifnet *);
75 
76 extern  struct domain inetdomain;
77 struct protosw in_gif_protosw = {
78 	.pr_type =		SOCK_RAW,
79 	.pr_domain =		&inetdomain,
80 	.pr_protocol =		0/* IPPROTO_IPV[46] */,
81 	.pr_flags =		PR_ATOMIC|PR_ADDR,
82 	.pr_input =		in_gif_input,
83 	.pr_output =		(pr_output_t*)rip_output,
84 	.pr_ctloutput =		rip_ctloutput,
85 	.pr_usrreqs =		&rip_usrreqs
86 };
87 
88 SYSCTL_VNET_INT(_net_inet_ip, IPCTL_GIF_TTL, gifttl, CTLFLAG_RW,
89 	&VNET_NAME(ip_gif_ttl), 0, "");
90 
91 int
92 in_gif_output(struct ifnet *ifp, int family, struct mbuf *m)
93 {
94 	struct gif_softc *sc = ifp->if_softc;
95 	struct sockaddr_in *dst = (struct sockaddr_in *)&sc->gif_ro.ro_dst;
96 	struct sockaddr_in *sin_src = (struct sockaddr_in *)sc->gif_psrc;
97 	struct sockaddr_in *sin_dst = (struct sockaddr_in *)sc->gif_pdst;
98 	struct ip iphdr;	/* capsule IP header, host byte ordered */
99 	struct etherip_header eiphdr;
100 	int error, len, proto;
101 	u_int8_t tos;
102 
103 	GIF_LOCK_ASSERT(sc);
104 
105 	if (sin_src == NULL || sin_dst == NULL ||
106 	    sin_src->sin_family != AF_INET ||
107 	    sin_dst->sin_family != AF_INET) {
108 		m_freem(m);
109 		return EAFNOSUPPORT;
110 	}
111 
112 	switch (family) {
113 #ifdef INET
114 	case AF_INET:
115 	    {
116 		struct ip *ip;
117 
118 		proto = IPPROTO_IPV4;
119 		if (m->m_len < sizeof(*ip)) {
120 			m = m_pullup(m, sizeof(*ip));
121 			if (!m)
122 				return ENOBUFS;
123 		}
124 		ip = mtod(m, struct ip *);
125 		tos = ip->ip_tos;
126 		break;
127 	    }
128 #endif /* INET */
129 #ifdef INET6
130 	case AF_INET6:
131 	    {
132 		struct ip6_hdr *ip6;
133 		proto = IPPROTO_IPV6;
134 		if (m->m_len < sizeof(*ip6)) {
135 			m = m_pullup(m, sizeof(*ip6));
136 			if (!m)
137 				return ENOBUFS;
138 		}
139 		ip6 = mtod(m, struct ip6_hdr *);
140 		tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
141 		break;
142 	    }
143 #endif /* INET6 */
144 	case AF_LINK:
145  		proto = IPPROTO_ETHERIP;
146 
147 		/*
148 		 * GIF_SEND_REVETHIP (disabled by default) intentionally
149 		 * sends an EtherIP packet with revered version field in
150 		 * the header.  This is a knob for backward compatibility
151 		 * with FreeBSD 7.2R or prior.
152 		 */
153 		if ((sc->gif_options & GIF_SEND_REVETHIP)) {
154  			eiphdr.eip_ver = 0;
155  			eiphdr.eip_resvl = ETHERIP_VERSION;
156  			eiphdr.eip_resvh = 0;
157 		} else {
158  			eiphdr.eip_ver = ETHERIP_VERSION;
159  			eiphdr.eip_resvl = 0;
160  			eiphdr.eip_resvh = 0;
161 		}
162  		/* prepend Ethernet-in-IP header */
163  		M_PREPEND(m, sizeof(struct etherip_header), M_DONTWAIT);
164  		if (m && m->m_len < sizeof(struct etherip_header))
165  			m = m_pullup(m, sizeof(struct etherip_header));
166  		if (m == NULL)
167  			return ENOBUFS;
168  		bcopy(&eiphdr, mtod(m, struct etherip_header *),
169 		    sizeof(struct etherip_header));
170 		break;
171 
172 	default:
173 #ifdef DEBUG
174 		printf("in_gif_output: warning: unknown family %d passed\n",
175 			family);
176 #endif
177 		m_freem(m);
178 		return EAFNOSUPPORT;
179 	}
180 
181 	bzero(&iphdr, sizeof(iphdr));
182 	iphdr.ip_src = sin_src->sin_addr;
183 	/* bidirectional configured tunnel mode */
184 	if (sin_dst->sin_addr.s_addr != INADDR_ANY)
185 		iphdr.ip_dst = sin_dst->sin_addr;
186 	else {
187 		m_freem(m);
188 		return ENETUNREACH;
189 	}
190 	iphdr.ip_p = proto;
191 	/* version will be set in ip_output() */
192 	iphdr.ip_ttl = V_ip_gif_ttl;
193 	iphdr.ip_len = m->m_pkthdr.len + sizeof(struct ip);
194 	ip_ecn_ingress((ifp->if_flags & IFF_LINK1) ? ECN_ALLOWED : ECN_NOCARE,
195 		       &iphdr.ip_tos, &tos);
196 
197 	/* prepend new IP header */
198 	len = sizeof(struct ip);
199 #ifndef __NO_STRICT_ALIGNMENT
200 	if (family == AF_LINK)
201 		len += ETHERIP_ALIGN;
202 #endif
203 	M_PREPEND(m, len, M_DONTWAIT);
204 	if (m != NULL && m->m_len < len)
205 		m = m_pullup(m, len);
206 	if (m == NULL) {
207 		printf("ENOBUFS in in_gif_output %d\n", __LINE__);
208 		return ENOBUFS;
209 	}
210 #ifndef __NO_STRICT_ALIGNMENT
211 	if (family == AF_LINK) {
212 		len = mtod(m, vm_offset_t) & 3;
213 		KASSERT(len == 0 || len == ETHERIP_ALIGN,
214 		    ("in_gif_output: unexpected misalignment"));
215 		m->m_data += len;
216 		m->m_len -= ETHERIP_ALIGN;
217 	}
218 #endif
219 	bcopy(&iphdr, mtod(m, struct ip *), sizeof(struct ip));
220 
221 	M_SETFIB(m, sc->gif_fibnum);
222 
223 	if (dst->sin_family != sin_dst->sin_family ||
224 	    dst->sin_addr.s_addr != sin_dst->sin_addr.s_addr) {
225 		/* cache route doesn't match */
226 		bzero(dst, sizeof(*dst));
227 		dst->sin_family = sin_dst->sin_family;
228 		dst->sin_len = sizeof(struct sockaddr_in);
229 		dst->sin_addr = sin_dst->sin_addr;
230 		if (sc->gif_ro.ro_rt) {
231 			RTFREE(sc->gif_ro.ro_rt);
232 			sc->gif_ro.ro_rt = NULL;
233 		}
234 #if 0
235 		GIF2IFP(sc)->if_mtu = GIF_MTU;
236 #endif
237 	}
238 
239 	if (sc->gif_ro.ro_rt == NULL) {
240 		in_rtalloc_ign(&sc->gif_ro, 0, sc->gif_fibnum);
241 		if (sc->gif_ro.ro_rt == NULL) {
242 			m_freem(m);
243 			return ENETUNREACH;
244 		}
245 
246 		/* if it constitutes infinite encapsulation, punt. */
247 		if (sc->gif_ro.ro_rt->rt_ifp == ifp) {
248 			m_freem(m);
249 			return ENETUNREACH;	/* XXX */
250 		}
251 #if 0
252 		ifp->if_mtu = sc->gif_ro.ro_rt->rt_ifp->if_mtu
253 			- sizeof(struct ip);
254 #endif
255 	}
256 
257 	error = ip_output(m, NULL, &sc->gif_ro, 0, NULL, NULL);
258 
259 	if (!(GIF2IFP(sc)->if_flags & IFF_LINK0) &&
260 	    sc->gif_ro.ro_rt != NULL) {
261 		RTFREE(sc->gif_ro.ro_rt);
262 		sc->gif_ro.ro_rt = NULL;
263 	}
264 
265 	return (error);
266 }
267 
268 void
269 in_gif_input(struct mbuf *m, int off)
270 {
271 	struct ifnet *gifp = NULL;
272 	struct gif_softc *sc;
273 	struct ip *ip;
274 	int af;
275 	u_int8_t otos;
276 	int proto;
277 
278 	ip = mtod(m, struct ip *);
279 	proto = ip->ip_p;
280 
281 	sc = (struct gif_softc *)encap_getarg(m);
282 	if (sc == NULL) {
283 		m_freem(m);
284 		KMOD_IPSTAT_INC(ips_nogif);
285 		return;
286 	}
287 
288 	gifp = GIF2IFP(sc);
289 	if (gifp == NULL || (gifp->if_flags & IFF_UP) == 0) {
290 		m_freem(m);
291 		KMOD_IPSTAT_INC(ips_nogif);
292 		return;
293 	}
294 
295 	otos = ip->ip_tos;
296 	m_adj(m, off);
297 
298 	switch (proto) {
299 #ifdef INET
300 	case IPPROTO_IPV4:
301 	    {
302 		struct ip *ip;
303 		af = AF_INET;
304 		if (m->m_len < sizeof(*ip)) {
305 			m = m_pullup(m, sizeof(*ip));
306 			if (!m)
307 				return;
308 		}
309 		ip = mtod(m, struct ip *);
310 		if (ip_ecn_egress((gifp->if_flags & IFF_LINK1) ?
311 				  ECN_ALLOWED : ECN_NOCARE,
312 				  &otos, &ip->ip_tos) == 0) {
313 			m_freem(m);
314 			return;
315 		}
316 		break;
317 	    }
318 #endif
319 #ifdef INET6
320 	case IPPROTO_IPV6:
321 	    {
322 		struct ip6_hdr *ip6;
323 		u_int8_t itos, oitos;
324 
325 		af = AF_INET6;
326 		if (m->m_len < sizeof(*ip6)) {
327 			m = m_pullup(m, sizeof(*ip6));
328 			if (!m)
329 				return;
330 		}
331 		ip6 = mtod(m, struct ip6_hdr *);
332 		itos = oitos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
333 		if (ip_ecn_egress((gifp->if_flags & IFF_LINK1) ?
334 				  ECN_ALLOWED : ECN_NOCARE,
335 				  &otos, &itos) == 0) {
336 			m_freem(m);
337 			return;
338 		}
339 		if (itos != oitos) {
340 			ip6->ip6_flow &= ~htonl(0xff << 20);
341 			ip6->ip6_flow |= htonl((u_int32_t)itos << 20);
342 		}
343 		break;
344 	    }
345 #endif /* INET6 */
346  	case IPPROTO_ETHERIP:
347  		af = AF_LINK;
348  		break;
349 
350 	default:
351 		KMOD_IPSTAT_INC(ips_nogif);
352 		m_freem(m);
353 		return;
354 	}
355 	gif_input(m, af, gifp);
356 	return;
357 }
358 
359 /*
360  * validate outer address.
361  */
362 static int
363 gif_validate4(const struct ip *ip, struct gif_softc *sc, struct ifnet *ifp)
364 {
365 	struct sockaddr_in *src, *dst;
366 	struct in_ifaddr *ia4;
367 
368 	src = (struct sockaddr_in *)sc->gif_psrc;
369 	dst = (struct sockaddr_in *)sc->gif_pdst;
370 
371 	/* check for address match */
372 	if (src->sin_addr.s_addr != ip->ip_dst.s_addr ||
373 	    dst->sin_addr.s_addr != ip->ip_src.s_addr)
374 		return 0;
375 
376 	/* martian filters on outer source - NOT done in ip_input! */
377 	if (IN_MULTICAST(ntohl(ip->ip_src.s_addr)))
378 		return 0;
379 	switch ((ntohl(ip->ip_src.s_addr) & 0xff000000) >> 24) {
380 	case 0: case 127: case 255:
381 		return 0;
382 	}
383 
384 	/* reject packets with broadcast on source */
385 	/* XXXRW: should use hash lists? */
386 	IN_IFADDR_RLOCK();
387 	TAILQ_FOREACH(ia4, &V_in_ifaddrhead, ia_link) {
388 		if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0)
389 			continue;
390 		if (ip->ip_src.s_addr == ia4->ia_broadaddr.sin_addr.s_addr) {
391 			IN_IFADDR_RUNLOCK();
392 			return 0;
393 		}
394 	}
395 	IN_IFADDR_RUNLOCK();
396 
397 	/* ingress filters on outer source */
398 	if ((GIF2IFP(sc)->if_flags & IFF_LINK2) == 0 && ifp) {
399 		struct sockaddr_in sin;
400 		struct rtentry *rt;
401 
402 		bzero(&sin, sizeof(sin));
403 		sin.sin_family = AF_INET;
404 		sin.sin_len = sizeof(struct sockaddr_in);
405 		sin.sin_addr = ip->ip_src;
406 		/* XXX MRT  check for the interface we would use on output */
407 		rt = in_rtalloc1((struct sockaddr *)&sin, 0,
408 		    0UL, sc->gif_fibnum);
409 		if (!rt || rt->rt_ifp != ifp) {
410 #if 0
411 			log(LOG_WARNING, "%s: packet from 0x%x dropped "
412 			    "due to ingress filter\n", if_name(GIF2IFP(sc)),
413 			    (u_int32_t)ntohl(sin.sin_addr.s_addr));
414 #endif
415 			if (rt)
416 				RTFREE_LOCKED(rt);
417 			return 0;
418 		}
419 		RTFREE_LOCKED(rt);
420 	}
421 
422 	return 32 * 2;
423 }
424 
425 /*
426  * we know that we are in IFF_UP, outer address available, and outer family
427  * matched the physical addr family.  see gif_encapcheck().
428  */
429 int
430 gif_encapcheck4(const struct mbuf *m, int off, int proto, void *arg)
431 {
432 	struct ip ip;
433 	struct gif_softc *sc;
434 	struct ifnet *ifp;
435 
436 	/* sanity check done in caller */
437 	sc = (struct gif_softc *)arg;
438 
439 	/* LINTED const cast */
440 	m_copydata(m, 0, sizeof(ip), (caddr_t)&ip);
441 	ifp = ((m->m_flags & M_PKTHDR) != 0) ? m->m_pkthdr.rcvif : NULL;
442 
443 	return gif_validate4(&ip, sc, ifp);
444 }
445 
446 int
447 in_gif_attach(struct gif_softc *sc)
448 {
449 	sc->encap_cookie4 = encap_attach_func(AF_INET, -1, gif_encapcheck,
450 	    &in_gif_protosw, sc);
451 	if (sc->encap_cookie4 == NULL)
452 		return EEXIST;
453 	return 0;
454 }
455 
456 int
457 in_gif_detach(struct gif_softc *sc)
458 {
459 	int error;
460 
461 	error = encap_detach(sc->encap_cookie4);
462 	if (error == 0)
463 		sc->encap_cookie4 = NULL;
464 	return error;
465 }
466