xref: /freebsd/sys/netinet/in_gif.c (revision 10b59a9b4add0320d52c15ce057dd697261e7dfc) 
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 VNET_DEFINE(int, ip_gif_ttl) = GIF_TTL;
89 #define	V_ip_gif_ttl		VNET(ip_gif_ttl)
90 SYSCTL_VNET_INT(_net_inet_ip, IPCTL_GIF_TTL, gifttl, CTLFLAG_RW,
91 	&VNET_NAME(ip_gif_ttl), 0, "");
92 
93 int
94 in_gif_output(struct ifnet *ifp, int family, struct mbuf *m)
95 {
96 	struct gif_softc *sc = ifp->if_softc;
97 	struct sockaddr_in *dst = (struct sockaddr_in *)&sc->gif_ro.ro_dst;
98 	struct sockaddr_in *sin_src = (struct sockaddr_in *)sc->gif_psrc;
99 	struct sockaddr_in *sin_dst = (struct sockaddr_in *)sc->gif_pdst;
100 	struct ip iphdr;	/* capsule IP header, host byte ordered */
101 	struct etherip_header eiphdr;
102 	int error, len, proto;
103 	u_int8_t tos;
104 
105 	GIF_LOCK_ASSERT(sc);
106 
107 	if (sin_src == NULL || sin_dst == NULL ||
108 	    sin_src->sin_family != AF_INET ||
109 	    sin_dst->sin_family != AF_INET) {
110 		m_freem(m);
111 		return EAFNOSUPPORT;
112 	}
113 
114 	switch (family) {
115 #ifdef INET
116 	case AF_INET:
117 	    {
118 		struct ip *ip;
119 
120 		proto = IPPROTO_IPV4;
121 		if (m->m_len < sizeof(*ip)) {
122 			m = m_pullup(m, sizeof(*ip));
123 			if (!m)
124 				return ENOBUFS;
125 		}
126 		ip = mtod(m, struct ip *);
127 		tos = ip->ip_tos;
128 		break;
129 	    }
130 #endif /* INET */
131 #ifdef INET6
132 	case AF_INET6:
133 	    {
134 		struct ip6_hdr *ip6;
135 		proto = IPPROTO_IPV6;
136 		if (m->m_len < sizeof(*ip6)) {
137 			m = m_pullup(m, sizeof(*ip6));
138 			if (!m)
139 				return ENOBUFS;
140 		}
141 		ip6 = mtod(m, struct ip6_hdr *);
142 		tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
143 		break;
144 	    }
145 #endif /* INET6 */
146 	case AF_LINK:
147  		proto = IPPROTO_ETHERIP;
148 
149 		/*
150 		 * GIF_SEND_REVETHIP (disabled by default) intentionally
151 		 * sends an EtherIP packet with revered version field in
152 		 * the header.  This is a knob for backward compatibility
153 		 * with FreeBSD 7.2R or prior.
154 		 */
155 		if ((sc->gif_options & GIF_SEND_REVETHIP)) {
156  			eiphdr.eip_ver = 0;
157  			eiphdr.eip_resvl = ETHERIP_VERSION;
158  			eiphdr.eip_resvh = 0;
159 		} else {
160  			eiphdr.eip_ver = ETHERIP_VERSION;
161  			eiphdr.eip_resvl = 0;
162  			eiphdr.eip_resvh = 0;
163 		}
164  		/* prepend Ethernet-in-IP header */
165  		M_PREPEND(m, sizeof(struct etherip_header), M_DONTWAIT);
166  		if (m && m->m_len < sizeof(struct etherip_header))
167  			m = m_pullup(m, sizeof(struct etherip_header));
168  		if (m == NULL)
169  			return ENOBUFS;
170  		bcopy(&eiphdr, mtod(m, struct etherip_header *),
171 		    sizeof(struct etherip_header));
172 		break;
173 
174 	default:
175 #ifdef DEBUG
176 		printf("in_gif_output: warning: unknown family %d passed\n",
177 			family);
178 #endif
179 		m_freem(m);
180 		return EAFNOSUPPORT;
181 	}
182 
183 	bzero(&iphdr, sizeof(iphdr));
184 	iphdr.ip_src = sin_src->sin_addr;
185 	/* bidirectional configured tunnel mode */
186 	if (sin_dst->sin_addr.s_addr != INADDR_ANY)
187 		iphdr.ip_dst = sin_dst->sin_addr;
188 	else {
189 		m_freem(m);
190 		return ENETUNREACH;
191 	}
192 	iphdr.ip_p = proto;
193 	/* version will be set in ip_output() */
194 	iphdr.ip_ttl = V_ip_gif_ttl;
195 	iphdr.ip_len = m->m_pkthdr.len + sizeof(struct ip);
196 	ip_ecn_ingress((ifp->if_flags & IFF_LINK1) ? ECN_ALLOWED : ECN_NOCARE,
197 		       &iphdr.ip_tos, &tos);
198 
199 	/* prepend new IP header */
200 	len = sizeof(struct ip);
201 #ifndef __NO_STRICT_ALIGNMENT
202 	if (family == AF_LINK)
203 		len += ETHERIP_ALIGN;
204 #endif
205 	M_PREPEND(m, len, M_DONTWAIT);
206 	if (m != NULL && m->m_len < len)
207 		m = m_pullup(m, len);
208 	if (m == NULL) {
209 		printf("ENOBUFS in in_gif_output %d\n", __LINE__);
210 		return ENOBUFS;
211 	}
212 #ifndef __NO_STRICT_ALIGNMENT
213 	if (family == AF_LINK) {
214 		len = mtod(m, vm_offset_t) & 3;
215 		KASSERT(len == 0 || len == ETHERIP_ALIGN,
216 		    ("in_gif_output: unexpected misalignment"));
217 		m->m_data += len;
218 		m->m_len -= ETHERIP_ALIGN;
219 	}
220 #endif
221 	bcopy(&iphdr, mtod(m, struct ip *), sizeof(struct ip));
222 
223 	M_SETFIB(m, sc->gif_fibnum);
224 
225 	if (dst->sin_family != sin_dst->sin_family ||
226 	    dst->sin_addr.s_addr != sin_dst->sin_addr.s_addr) {
227 		/* cache route doesn't match */
228 		bzero(dst, sizeof(*dst));
229 		dst->sin_family = sin_dst->sin_family;
230 		dst->sin_len = sizeof(struct sockaddr_in);
231 		dst->sin_addr = sin_dst->sin_addr;
232 		if (sc->gif_ro.ro_rt) {
233 			RTFREE(sc->gif_ro.ro_rt);
234 			sc->gif_ro.ro_rt = NULL;
235 		}
236 #if 0
237 		GIF2IFP(sc)->if_mtu = GIF_MTU;
238 #endif
239 	}
240 
241 	if (sc->gif_ro.ro_rt == NULL) {
242 		in_rtalloc_ign(&sc->gif_ro, 0, sc->gif_fibnum);
243 		if (sc->gif_ro.ro_rt == NULL) {
244 			m_freem(m);
245 			return ENETUNREACH;
246 		}
247 
248 		/* if it constitutes infinite encapsulation, punt. */
249 		if (sc->gif_ro.ro_rt->rt_ifp == ifp) {
250 			m_freem(m);
251 			return ENETUNREACH;	/* XXX */
252 		}
253 #if 0
254 		ifp->if_mtu = sc->gif_ro.ro_rt->rt_ifp->if_mtu
255 			- sizeof(struct ip);
256 #endif
257 	}
258 
259 	m_addr_changed(m);
260 
261 	error = ip_output(m, NULL, &sc->gif_ro, 0, NULL, NULL);
262 
263 	if (!(GIF2IFP(sc)->if_flags & IFF_LINK0) &&
264 	    sc->gif_ro.ro_rt != NULL) {
265 		RTFREE(sc->gif_ro.ro_rt);
266 		sc->gif_ro.ro_rt = NULL;
267 	}
268 
269 	return (error);
270 }
271 
272 void
273 in_gif_input(struct mbuf *m, int off)
274 {
275 	struct ifnet *gifp = NULL;
276 	struct gif_softc *sc;
277 	struct ip *ip;
278 	int af;
279 	u_int8_t otos;
280 	int proto;
281 
282 	ip = mtod(m, struct ip *);
283 	proto = ip->ip_p;
284 
285 	sc = (struct gif_softc *)encap_getarg(m);
286 	if (sc == NULL) {
287 		m_freem(m);
288 		KMOD_IPSTAT_INC(ips_nogif);
289 		return;
290 	}
291 
292 	gifp = GIF2IFP(sc);
293 	if (gifp == NULL || (gifp->if_flags & IFF_UP) == 0) {
294 		m_freem(m);
295 		KMOD_IPSTAT_INC(ips_nogif);
296 		return;
297 	}
298 
299 	otos = ip->ip_tos;
300 	m_adj(m, off);
301 
302 	switch (proto) {
303 #ifdef INET
304 	case IPPROTO_IPV4:
305 	    {
306 		struct ip *ip;
307 		af = AF_INET;
308 		if (m->m_len < sizeof(*ip)) {
309 			m = m_pullup(m, sizeof(*ip));
310 			if (!m)
311 				return;
312 		}
313 		ip = mtod(m, struct ip *);
314 		if (ip_ecn_egress((gifp->if_flags & IFF_LINK1) ?
315 				  ECN_ALLOWED : ECN_NOCARE,
316 				  &otos, &ip->ip_tos) == 0) {
317 			m_freem(m);
318 			return;
319 		}
320 		break;
321 	    }
322 #endif
323 #ifdef INET6
324 	case IPPROTO_IPV6:
325 	    {
326 		struct ip6_hdr *ip6;
327 		u_int8_t itos, oitos;
328 
329 		af = AF_INET6;
330 		if (m->m_len < sizeof(*ip6)) {
331 			m = m_pullup(m, sizeof(*ip6));
332 			if (!m)
333 				return;
334 		}
335 		ip6 = mtod(m, struct ip6_hdr *);
336 		itos = oitos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
337 		if (ip_ecn_egress((gifp->if_flags & IFF_LINK1) ?
338 				  ECN_ALLOWED : ECN_NOCARE,
339 				  &otos, &itos) == 0) {
340 			m_freem(m);
341 			return;
342 		}
343 		if (itos != oitos) {
344 			ip6->ip6_flow &= ~htonl(0xff << 20);
345 			ip6->ip6_flow |= htonl((u_int32_t)itos << 20);
346 		}
347 		break;
348 	    }
349 #endif /* INET6 */
350  	case IPPROTO_ETHERIP:
351  		af = AF_LINK;
352  		break;
353 
354 	default:
355 		KMOD_IPSTAT_INC(ips_nogif);
356 		m_freem(m);
357 		return;
358 	}
359 	gif_input(m, af, gifp);
360 	return;
361 }
362 
363 /*
364  * validate outer address.
365  */
366 static int
367 gif_validate4(const struct ip *ip, struct gif_softc *sc, struct ifnet *ifp)
368 {
369 	struct sockaddr_in *src, *dst;
370 	struct in_ifaddr *ia4;
371 
372 	src = (struct sockaddr_in *)sc->gif_psrc;
373 	dst = (struct sockaddr_in *)sc->gif_pdst;
374 
375 	/* check for address match */
376 	if (src->sin_addr.s_addr != ip->ip_dst.s_addr ||
377 	    dst->sin_addr.s_addr != ip->ip_src.s_addr)
378 		return 0;
379 
380 	/* martian filters on outer source - NOT done in ip_input! */
381 	if (IN_MULTICAST(ntohl(ip->ip_src.s_addr)))
382 		return 0;
383 	switch ((ntohl(ip->ip_src.s_addr) & 0xff000000) >> 24) {
384 	case 0: case 127: case 255:
385 		return 0;
386 	}
387 
388 	/* reject packets with broadcast on source */
389 	/* XXXRW: should use hash lists? */
390 	IN_IFADDR_RLOCK();
391 	TAILQ_FOREACH(ia4, &V_in_ifaddrhead, ia_link) {
392 		if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0)
393 			continue;
394 		if (ip->ip_src.s_addr == ia4->ia_broadaddr.sin_addr.s_addr) {
395 			IN_IFADDR_RUNLOCK();
396 			return 0;
397 		}
398 	}
399 	IN_IFADDR_RUNLOCK();
400 
401 	/* ingress filters on outer source */
402 	if ((GIF2IFP(sc)->if_flags & IFF_LINK2) == 0 && ifp) {
403 		struct sockaddr_in sin;
404 		struct rtentry *rt;
405 
406 		bzero(&sin, sizeof(sin));
407 		sin.sin_family = AF_INET;
408 		sin.sin_len = sizeof(struct sockaddr_in);
409 		sin.sin_addr = ip->ip_src;
410 		/* XXX MRT  check for the interface we would use on output */
411 		rt = in_rtalloc1((struct sockaddr *)&sin, 0,
412 		    0UL, sc->gif_fibnum);
413 		if (!rt || rt->rt_ifp != ifp) {
414 #if 0
415 			log(LOG_WARNING, "%s: packet from 0x%x dropped "
416 			    "due to ingress filter\n", if_name(GIF2IFP(sc)),
417 			    (u_int32_t)ntohl(sin.sin_addr.s_addr));
418 #endif
419 			if (rt)
420 				RTFREE_LOCKED(rt);
421 			return 0;
422 		}
423 		RTFREE_LOCKED(rt);
424 	}
425 
426 	return 32 * 2;
427 }
428 
429 /*
430  * we know that we are in IFF_UP, outer address available, and outer family
431  * matched the physical addr family.  see gif_encapcheck().
432  */
433 int
434 gif_encapcheck4(const struct mbuf *m, int off, int proto, void *arg)
435 {
436 	struct ip ip;
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(ip), (caddr_t)&ip);
445 	ifp = ((m->m_flags & M_PKTHDR) != 0) ? m->m_pkthdr.rcvif : NULL;
446 
447 	return gif_validate4(&ip, sc, ifp);
448 }
449 
450 int
451 in_gif_attach(struct gif_softc *sc)
452 {
453 	sc->encap_cookie4 = encap_attach_func(AF_INET, -1, gif_encapcheck,
454 	    &in_gif_protosw, sc);
455 	if (sc->encap_cookie4 == NULL)
456 		return EEXIST;
457 	return 0;
458 }
459 
460 int
461 in_gif_detach(struct gif_softc *sc)
462 {
463 	int error;
464 
465 	error = encap_detach(sc->encap_cookie4);
466 	if (error == 0)
467 		sc->encap_cookie4 = NULL;
468 	return error;
469 }
470