[apple/xnu.git] / bsd / netinet6 / ip6_forward.c

/*	$FreeBSD: src/sys/netinet6/ip6_forward.c,v 1.16 2002/10/16 02:25:05 sam Exp $	*/
/*	$KAME: ip6_forward.c,v 1.69 2001/05/17 03:48:30 itojun Exp $	*/

/*
 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Neither the name of the project nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */


#include <sys/param.h>
#include <sys/systm.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/domain.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/errno.h>
#include <sys/time.h>
#include <sys/kernel.h>
#include <sys/syslog.h>

#include <net/if.h>
#include <net/route.h>

#include <netinet/in.h>
#include <netinet/in_var.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/ip_var.h>
#include <netinet6/in6_var.h>
#include <netinet/ip6.h>
#include <netinet6/ip6_var.h>
#include <netinet/icmp6.h>
#include <netinet6/nd6.h>

#include <netinet/in_pcb.h>

#if IPSEC
#include <netinet6/ipsec.h>
#if INET6
#include <netinet6/ipsec6.h>
#endif
#include <netkey/key.h>
extern int ipsec_bypass;
extern lck_mtx_t *sadb_mutex;
extern lck_mtx_t *ip6_mutex;
#endif /* IPSEC */

#include <netinet6/ip6_fw.h>

#include <net/net_osdep.h>

struct	route_in6 ip6_forward_rt;

/*
 * Forward a packet.  If some error occurs return the sender
 * an icmp packet.  Note we can't always generate a meaningful
 * icmp message because icmp doesn't have a large enough repertoire
 * of codes and types.
 *
 * If not forwarding, just drop the packet.  This could be confusing
 * if ipforwarding was zero but some routing protocol was advancing
 * us as a gateway to somewhere.  However, we must let the routing
 * protocol deal with that.
 *
 */

void
ip6_forward(m, srcrt, locked)
	struct mbuf *m;
	int srcrt;
	int locked;
{
	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
	struct sockaddr_in6 *dst;
	struct rtentry *rt;
	int error, type = 0, code = 0;
	struct mbuf *mcopy = NULL;
	struct ifnet *origifp;	/* maybe unnecessary */
#if IPSEC
	struct secpolicy *sp = NULL;
#endif
	struct timeval timenow;

	getmicrotime(&timenow);


#if IPSEC
	/*
	 * Check AH/ESP integrity.
	 */
	/*
	 * Don't increment ip6s_cantforward because this is the check
	 * before forwarding packet actually.
	 */
	if (ipsec_bypass == 0) {
		lck_mtx_lock(sadb_mutex);
		if (ipsec6_in_reject(m, NULL)) {
			ipsec6stat.in_polvio++;
			lck_mtx_unlock(sadb_mutex);
			m_freem(m);
			return;
		}
		lck_mtx_unlock(sadb_mutex);
	}
#endif /*IPSEC*/

	/*
	 * Do not forward packets to multicast destination (should be handled
	 * by ip6_mforward().
	 * Do not forward packets with unspecified source.  It was discussed
	 * in July 2000, on ipngwg mailing list.
	 */
	if ((m->m_flags & (M_BCAST|M_MCAST)) != 0 ||
	    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
		ip6stat.ip6s_cantforward++;
		/* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
		if (ip6_log_time + ip6_log_interval < timenow.tv_sec) {
			ip6_log_time = timenow.tv_sec;
			log(LOG_DEBUG,
			    "cannot forward "
			    "from %s to %s nxt %d received on %s\n",
			    ip6_sprintf(&ip6->ip6_src),
			    ip6_sprintf(&ip6->ip6_dst),
			    ip6->ip6_nxt,
			    if_name(m->m_pkthdr.rcvif));
		}
		m_freem(m);
		return;
	}

	if (ip6->ip6_hlim <= IPV6_HLIMDEC) {
		/* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
		if (locked)
			lck_mtx_unlock(ip6_mutex);
		icmp6_error(m, ICMP6_TIME_EXCEEDED,
				ICMP6_TIME_EXCEED_TRANSIT, 0);
		if (locked)
			lck_mtx_lock(ip6_mutex);
		return;
	}
	ip6->ip6_hlim -= IPV6_HLIMDEC;

	/*
	 * Save at most ICMPV6_PLD_MAXLEN (= the min IPv6 MTU -
	 * size of IPv6 + ICMPv6 headers) bytes of the packet in case
	 * we need to generate an ICMP6 message to the src.
	 * Thanks to M_EXT, in most cases copy will not occur.
	 *
	 * It is important to save it before IPsec processing as IPsec
	 * processing may modify the mbuf.
	 */
	mcopy = m_copy(m, 0, imin(m->m_pkthdr.len, ICMPV6_PLD_MAXLEN));

#if IPSEC
	if (ipsec_bypass != 0)
		goto skip_ipsec;
	lck_mtx_lock(sadb_mutex);
	/* get a security policy for this packet */
	sp = ipsec6_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND, IP_FORWARDING,
	    &error);
	if (sp == NULL) {
		ipsec6stat.out_inval++;
		ip6stat.ip6s_cantforward++;
		if (mcopy) {
#if 0
			/* XXX: what icmp ? */
#else
			m_freem(mcopy);
#endif
		}
		lck_mtx_unlock(sadb_mutex);
		m_freem(m);
		return;
	}

	error = 0;

	/* check policy */
	switch (sp->policy) {
	case IPSEC_POLICY_DISCARD:
		/*
		 * This packet is just discarded.
		 */
		ipsec6stat.out_polvio++;
		ip6stat.ip6s_cantforward++;
		key_freesp(sp);
		if (mcopy) {
#if 0
			/* XXX: what icmp ? */
#else
			m_freem(mcopy);
#endif
		}
		lck_mtx_unlock(sadb_mutex);
		m_freem(m);
		return;

	case IPSEC_POLICY_BYPASS:
	case IPSEC_POLICY_NONE:
		/* no need to do IPsec. */
		key_freesp(sp);
		lck_mtx_unlock(sadb_mutex);
		goto skip_ipsec;

	case IPSEC_POLICY_IPSEC:
		if (sp->req == NULL) {
			/* XXX should be panic ? */
			printf("ip6_forward: No IPsec request specified.\n");
			ip6stat.ip6s_cantforward++;
			key_freesp(sp);
			if (mcopy) {
#if 0
				/* XXX: what icmp ? */
#else
				m_freem(mcopy);
#endif
			}
			lck_mtx_unlock(sadb_mutex);
			m_freem(m);
			return;
		}
		/* do IPsec */
		break;

	case IPSEC_POLICY_ENTRUST:
	default:
		/* should be panic ?? */
		printf("ip6_forward: Invalid policy found. %d\n", sp->policy);
		key_freesp(sp);
		lck_mtx_unlock(sadb_mutex);
		goto skip_ipsec;
	}

    {
	struct ipsec_output_state state;

	/*
	 * All the extension headers will become inaccessible
	 * (since they can be encrypted).
	 * Don't panic, we need no more updates to extension headers
	 * on inner IPv6 packet (since they are now encapsulated).
	 *
	 * IPv6 [ESP|AH] IPv6 [extension headers] payload
	 */
	bzero(&state, sizeof(state));
	state.m = m;
	state.ro = NULL;	/* update at ipsec6_output_tunnel() */
	state.dst = NULL;	/* update at ipsec6_output_tunnel() */

	if (locked)
			lck_mtx_unlock(ip6_mutex);
	error = ipsec6_output_tunnel(&state, sp, 0);
	if (locked) {
			lck_mtx_unlock(sadb_mutex);
			lck_mtx_lock(ip6_mutex);
			lck_mtx_lock(sadb_mutex);
	}

	m = state.m;
	key_freesp(sp);

	if (error) {
		/* mbuf is already reclaimed in ipsec6_output_tunnel. */
		switch (error) {
		case EHOSTUNREACH:
		case ENETUNREACH:
		case EMSGSIZE:
		case ENOBUFS:
		case ENOMEM:
			break;
		default:
			printf("ip6_output (ipsec): error code %d\n", error);
			/* fall through */
		case ENOENT:
			/* don't show these error codes to the user */
			break;
		}
		ip6stat.ip6s_cantforward++;
		if (mcopy) {
#if 0
			/* XXX: what icmp ? */
#else
			m_freem(mcopy);
#endif
		}
		lck_mtx_unlock(sadb_mutex);
		m_freem(m);
		return;
	}
    }
	lck_mtx_unlock(sadb_mutex);
    skip_ipsec:
#endif /* IPSEC */

	dst = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
	if (!srcrt) {
		/*
		 * ip6_forward_rt.ro_dst.sin6_addr is equal to ip6->ip6_dst
		 */
		if (ip6_forward_rt.ro_rt == 0 ||
		    (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) == 0) {
			if (ip6_forward_rt.ro_rt) {
				rtfree(ip6_forward_rt.ro_rt);
				ip6_forward_rt.ro_rt = 0;
			}
			/* this probably fails but give it a try again */
			rtalloc_ign((struct route *)&ip6_forward_rt,
				    RTF_PRCLONING);
		}

		if (ip6_forward_rt.ro_rt == 0) {
			ip6stat.ip6s_noroute++;
			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_noroute);
			if (mcopy) {
				if (locked)
					lck_mtx_unlock(ip6_mutex);
				icmp6_error(mcopy, ICMP6_DST_UNREACH,
					    ICMP6_DST_UNREACH_NOROUTE, 0);
				if (locked)
					lck_mtx_lock(ip6_mutex);
			}
			m_freem(m);
			return;
		}
	} else if ((rt = ip6_forward_rt.ro_rt) == 0 ||
		 !IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &dst->sin6_addr)) {
		if (ip6_forward_rt.ro_rt) {
			rtfree(ip6_forward_rt.ro_rt);
			ip6_forward_rt.ro_rt = 0;
		}
		bzero(dst, sizeof(*dst));
		dst->sin6_len = sizeof(struct sockaddr_in6);
		dst->sin6_family = AF_INET6;
		dst->sin6_addr = ip6->ip6_dst;

  		rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING);
		if (ip6_forward_rt.ro_rt == 0) {
			ip6stat.ip6s_noroute++;
			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_noroute);
			if (mcopy) {
				if (locked)
					lck_mtx_unlock(ip6_mutex);
				icmp6_error(mcopy, ICMP6_DST_UNREACH,
					    ICMP6_DST_UNREACH_NOROUTE, 0);
				if (locked)
					lck_mtx_lock(ip6_mutex);
			}
			m_freem(m);
			return;
		}
	}
	rt = ip6_forward_rt.ro_rt;

	/*
	 * Scope check: if a packet can't be delivered to its destination
	 * for the reason that the destination is beyond the scope of the
	 * source address, discard the packet and return an icmp6 destination
	 * unreachable error with Code 2 (beyond scope of source address).
	 * [draft-ietf-ipngwg-icmp-v3-02.txt, Section 3.1]
	 */
	if (in6_addr2scopeid(m->m_pkthdr.rcvif, &ip6->ip6_src) !=
	    in6_addr2scopeid(rt->rt_ifp, &ip6->ip6_src)) {
		ip6stat.ip6s_cantforward++;
		ip6stat.ip6s_badscope++;
		in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard);

		if (ip6_log_time + ip6_log_interval < timenow.tv_sec) {
			ip6_log_time = timenow.tv_sec;
			log(LOG_DEBUG,
			    "cannot forward "
			    "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
			    ip6_sprintf(&ip6->ip6_src),
			    ip6_sprintf(&ip6->ip6_dst),
			    ip6->ip6_nxt,
			    if_name(m->m_pkthdr.rcvif), if_name(rt->rt_ifp));
		}
		if (mcopy) {
			if (locked)
				lck_mtx_unlock(ip6_mutex);
			icmp6_error(mcopy, ICMP6_DST_UNREACH,
				    ICMP6_DST_UNREACH_BEYONDSCOPE, 0);
			if (locked)
				lck_mtx_lock(ip6_mutex);
		}
		m_freem(m);
		return;
	}

	if (m->m_pkthdr.len > rt->rt_ifp->if_mtu) {
		in6_ifstat_inc(rt->rt_ifp, ifs6_in_toobig);
		if (mcopy) {
			u_long mtu;
#if IPSEC
			struct secpolicy *sp;
			int ipsecerror;
			size_t ipsechdrsiz;
#endif

			mtu = rt->rt_ifp->if_mtu;
#if IPSEC
			/*
			 * When we do IPsec tunnel ingress, we need to play
			 * with if_mtu value (decrement IPsec header size
			 * from mtu value).  The code is much simpler than v4
			 * case, as we have the outgoing interface for
			 * encapsulated packet as "rt->rt_ifp".
			 */
			lck_mtx_lock(sadb_mutex);
			sp = ipsec6_getpolicybyaddr(mcopy, IPSEC_DIR_OUTBOUND,
				IP_FORWARDING, &ipsecerror);
			if (sp) {
				ipsechdrsiz = ipsec6_hdrsiz(mcopy,
					IPSEC_DIR_OUTBOUND, NULL);
				if (ipsechdrsiz < mtu)
					mtu -= ipsechdrsiz;
			}
			lck_mtx_unlock(sadb_mutex);
			/*
			 * if mtu becomes less than minimum MTU,
			 * tell minimum MTU (and I'll need to fragment it).
			 */
			if (mtu < IPV6_MMTU)
				mtu = IPV6_MMTU;
#endif
			if (locked)
				lck_mtx_unlock(ip6_mutex);
			icmp6_error(mcopy, ICMP6_PACKET_TOO_BIG, 0, mtu);
			if (locked)
				lck_mtx_lock(ip6_mutex);
		}
		m_freem(m);
		return;
 	}

	if (rt->rt_flags & RTF_GATEWAY)
		dst = (struct sockaddr_in6 *)rt->rt_gateway;

	/*
	 * If we are to forward the packet using the same interface
	 * as one we got the packet from, perhaps we should send a redirect
	 * to sender to shortcut a hop.
	 * Only send redirect if source is sending directly to us,
	 * and if packet was not source routed (or has any options).
	 * Also, don't send redirect if forwarding using a route
	 * modified by a redirect.
	 */
	if (rt->rt_ifp == m->m_pkthdr.rcvif && !srcrt &&
	    (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0) {
		if ((rt->rt_ifp->if_flags & IFF_POINTOPOINT) != 0) {
			/*
			 * If the incoming interface is equal to the outgoing
			 * one, and the link attached to the interface is
			 * point-to-point, then it will be highly probable
			 * that a routing loop occurs. Thus, we immediately
			 * drop the packet and send an ICMPv6 error message.
			 *
			 * type/code is based on suggestion by Rich Draves.
			 * not sure if it is the best pick.
			 */
			if (locked)
				lck_mtx_unlock(ip6_mutex);
			icmp6_error(mcopy, ICMP6_DST_UNREACH,
				    ICMP6_DST_UNREACH_ADDR, 0);
			if (locked)
				lck_mtx_lock(ip6_mutex);
			m_freem(m);
			return;
		}
		type = ND_REDIRECT;
	}

	/*
	 * Check with the firewall...
	 */
	if (ip6_fw_enable && ip6_fw_chk_ptr) {
		u_short port = 0;
		/* If ipfw says divert, we have to just drop packet */
		if (ip6_fw_chk_ptr(&ip6, rt->rt_ifp, &port, &m)) {
			m_freem(m);
			goto freecopy;
		}
		if (!m)
			goto freecopy;
	}

	/*
	 * Fake scoped addresses. Note that even link-local source or
	 * destinaion can appear, if the originating node just sends the
	 * packet to us (without address resolution for the destination).
	 * Since both icmp6_error and icmp6_redirect_output fill the embedded
	 * link identifiers, we can do this stuff after making a copy for
	 * returning an error.
	 */
	if ((rt->rt_ifp->if_flags & IFF_LOOPBACK) != 0) {
		/*
		 * See corresponding comments in ip6_output.
		 * XXX: but is it possible that ip6_forward() sends a packet
		 *      to a loopback interface? I don't think so, and thus
		 *      I bark here. (jinmei@kame.net)
		 * XXX: it is common to route invalid packets to loopback.
		 *	also, the codepath will be visited on use of ::1 in
		 *	rthdr. (itojun)
		 */
#if 1
		if (0)
#else
		if ((rt->rt_flags & (RTF_BLACKHOLE|RTF_REJECT)) == 0)
#endif
		{
			printf("ip6_forward: outgoing interface is loopback. "
				"src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
				ip6_sprintf(&ip6->ip6_src),
				ip6_sprintf(&ip6->ip6_dst),
				ip6->ip6_nxt, if_name(m->m_pkthdr.rcvif),
				if_name(rt->rt_ifp));
		}

		/* we can just use rcvif in forwarding. */
		origifp = m->m_pkthdr.rcvif;
	}
	else
		origifp = rt->rt_ifp;
#ifndef SCOPEDROUTING
	/*
	 * clear embedded scope identifiers if necessary.
	 * in6_clearscope will touch the addresses only when necessary.
	 */
	in6_clearscope(&ip6->ip6_src);
	in6_clearscope(&ip6->ip6_dst);
#endif

	error = nd6_output(rt->rt_ifp, origifp, m, dst, rt, locked);
	if (error) {
		in6_ifstat_inc(rt->rt_ifp, ifs6_out_discard);
		ip6stat.ip6s_cantforward++;
	} else {
		ip6stat.ip6s_forward++;
		in6_ifstat_inc(rt->rt_ifp, ifs6_out_forward);
		if (type)
			ip6stat.ip6s_redirectsent++;
		else {
			if (mcopy)
				goto freecopy;
		}
	}
	if (mcopy == NULL)
		return;

	switch (error) {
	case 0:
#if 1
		if (type == ND_REDIRECT) {
			icmp6_redirect_output(mcopy, rt);
			return;
		}
#endif
		goto freecopy;

	case EMSGSIZE:
		/* xxx MTU is constant in PPP? */
		goto freecopy;

	case ENOBUFS:
		/* Tell source to slow down like source quench in IP? */
		goto freecopy;

	case ENETUNREACH:	/* shouldn't happen, checked above */
	case EHOSTUNREACH:
	case ENETDOWN:
	case EHOSTDOWN:
	default:
		type = ICMP6_DST_UNREACH;
		code = ICMP6_DST_UNREACH_ADDR;
		break;
	}
	if (locked)
		lck_mtx_unlock(ip6_mutex);
	icmp6_error(mcopy, type, code, 0);
	if (locked)
		lck_mtx_lock(ip6_mutex);
	return;

 freecopy:
	m_freem(mcopy);
	return;
}
Commit	Line	Data
55e303ae	1	/* $FreeBSD: src/sys/netinet6/ip6_forward.c,v 1.16 2002/10/16 02:25:05 sam Exp $ */
9bccf70c	2	/* $KAME: ip6_forward.c,v 1.69 2001/05/17 03:48:30 itojun Exp $ */
1c79356b A	3
	4	/*
	5	* Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
	6	* All rights reserved.
	7	*
	8	* Redistribution and use in source and binary forms, with or without
	9	* modification, are permitted provided that the following conditions
	10	* are met:
	11	* 1. Redistributions of source code must retain the above copyright
	12	* notice, this list of conditions and the following disclaimer.
	13	* 2. Redistributions in binary form must reproduce the above copyright
	14	* notice, this list of conditions and the following disclaimer in the
	15	* documentation and/or other materials provided with the distribution.
	16	* 3. Neither the name of the project nor the names of its contributors
	17	* may be used to endorse or promote products derived from this software
	18	* without specific prior written permission.
	19	*
	20	* THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
	21	* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
	22	* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
	23	* ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
	24	* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
	25	* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
	26	* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
	27	* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
	28	* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
	29	* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
	30	* SUCH DAMAGE.
	31	*/
	32
1c79356b A	33
	34	#include <sys/param.h>
	35	#include <sys/systm.h>
	36	#include <sys/malloc.h>
	37	#include <sys/mbuf.h>
	38	#include <sys/domain.h>
	39	#include <sys/protosw.h>
	40	#include <sys/socket.h>
	41	#include <sys/errno.h>
	42	#include <sys/time.h>
	43	#include <sys/kernel.h>
	44	#include <sys/syslog.h>
	45
	46	#include <net/if.h>
	47	#include <net/route.h>
	48
	49	#include <netinet/in.h>
	50	#include <netinet/in_var.h>
9bccf70c A	51	#include <netinet/in_systm.h>
9bccf70c A	52	#include <netinet/ip.h>
1c79356b	53	#include <netinet/ip_var.h>
9bccf70c	54	#include <netinet6/in6_var.h>
1c79356b A	55	#include <netinet/ip6.h>
	56	#include <netinet6/ip6_var.h>
	57	#include <netinet/icmp6.h>
	58	#include <netinet6/nd6.h>
	59
9bccf70c A	60	#include <netinet/in_pcb.h>
	61
	62	#if IPSEC
1c79356b	63	#include <netinet6/ipsec.h>
9bccf70c A	64	#if INET6
	65	#include <netinet6/ipsec6.h>
	66	#endif
1c79356b	67	#include <netkey/key.h>
9bccf70c	68	extern int ipsec_bypass;
91447636 A	69	extern lck_mtx_t *sadb_mutex;
91447636 A	70	extern lck_mtx_t *ip6_mutex;
9bccf70c	71	#endif /* IPSEC */
1c79356b	72
1c79356b	73	#include <netinet6/ip6_fw.h>
1c79356b A	74
	75	#include <net/net_osdep.h>
	76
	77	struct route_in6 ip6_forward_rt;
	78
	79	/*
	80	* Forward a packet. If some error occurs return the sender
	81	* an icmp packet. Note we can't always generate a meaningful
	82	* icmp message because icmp doesn't have a large enough repertoire
	83	* of codes and types.
	84	*
	85	* If not forwarding, just drop the packet. This could be confusing
	86	* if ipforwarding was zero but some routing protocol was advancing
	87	* us as a gateway to somewhere. However, we must let the routing
	88	* protocol deal with that.
	89	*
	90	*/
	91
	92	void
91447636	93	ip6_forward(m, srcrt, locked)
1c79356b A	94	struct mbuf *m;
1c79356b A	95	int srcrt;
91447636	96	int locked;
1c79356b A	97	{
1c79356b A	98	struct ip6_hdr ip6 = mtod(m, struct ip6_hdr );
9bccf70c A	99	struct sockaddr_in6 *dst;
9bccf70c A	100	struct rtentry *rt;
1c79356b A	101	int error, type = 0, code = 0;
1c79356b A	102	struct mbuf *mcopy = NULL;
9bccf70c A	103	struct ifnet origifp; / maybe unnecessary */
9bccf70c A	104	#if IPSEC
1c79356b A	105	struct secpolicy *sp = NULL;
1c79356b A	106	#endif
91447636 A	107	struct timeval timenow;
	108
	109	getmicrotime(&timenow);
	110
1c79356b	111
9bccf70c	112	#if IPSEC
1c79356b A	113	/*
	114	* Check AH/ESP integrity.
	115	*/
	116	/*
	117	* Don't increment ip6s_cantforward because this is the check
	118	* before forwarding packet actually.
	119	*/
91447636 A	120	if (ipsec_bypass == 0) {
	121	lck_mtx_lock(sadb_mutex);
	122	if (ipsec6_in_reject(m, NULL)) {
	123	ipsec6stat.in_polvio++;
	124	lck_mtx_unlock(sadb_mutex);
	125	m_freem(m);
	126	return;
	127	}
	128	lck_mtx_unlock(sadb_mutex);
1c79356b	129	}
9bccf70c	130	#endif /IPSEC/
1c79356b	131
9bccf70c A	132	/*
	133	* Do not forward packets to multicast destination (should be handled
	134	* by ip6_mforward().
	135	* Do not forward packets with unspecified source. It was discussed
	136	* in July 2000, on ipngwg mailing list.
	137	*/
1c79356b	138	if ((m->m_flags & (M_BCAST\|M_MCAST)) != 0 \|\|
9bccf70c A	139	IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) \|\|
9bccf70c A	140	IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1c79356b A	141	ip6stat.ip6s_cantforward++;
1c79356b A	142	/* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
91447636 A	143	if (ip6_log_time + ip6_log_interval < timenow.tv_sec) {
91447636 A	144	ip6_log_time = timenow.tv_sec;
1c79356b A	145	log(LOG_DEBUG,
	146	"cannot forward "
	147	"from %s to %s nxt %d received on %s\n",
	148	ip6_sprintf(&ip6->ip6_src),
	149	ip6_sprintf(&ip6->ip6_dst),
	150	ip6->ip6_nxt,
	151	if_name(m->m_pkthdr.rcvif));
	152	}
	153	m_freem(m);
	154	return;
	155	}
	156
	157	if (ip6->ip6_hlim <= IPV6_HLIMDEC) {
	158	/* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
91447636 A	159	if (locked)
91447636 A	160	lck_mtx_unlock(ip6_mutex);
1c79356b A	161	icmp6_error(m, ICMP6_TIME_EXCEEDED,
1c79356b A	162	ICMP6_TIME_EXCEED_TRANSIT, 0);
91447636 A	163	if (locked)
91447636 A	164	lck_mtx_lock(ip6_mutex);
1c79356b A	165	return;
	166	}
	167	ip6->ip6_hlim -= IPV6_HLIMDEC;
	168
	169	/*
	170	* Save at most ICMPV6_PLD_MAXLEN (= the min IPv6 MTU -
	171	* size of IPv6 + ICMPv6 headers) bytes of the packet in case
	172	* we need to generate an ICMP6 message to the src.
	173	* Thanks to M_EXT, in most cases copy will not occur.
	174	*
	175	* It is important to save it before IPsec processing as IPsec
	176	* processing may modify the mbuf.
	177	*/
	178	mcopy = m_copy(m, 0, imin(m->m_pkthdr.len, ICMPV6_PLD_MAXLEN));
	179
9bccf70c A	180	#if IPSEC
	181	if (ipsec_bypass != 0)
	182	goto skip_ipsec;
91447636	183	lck_mtx_lock(sadb_mutex);
1c79356b	184	/* get a security policy for this packet */
9bccf70c A	185	sp = ipsec6_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND, IP_FORWARDING,
9bccf70c A	186	&error);
1c79356b A	187	if (sp == NULL) {
	188	ipsec6stat.out_inval++;
	189	ip6stat.ip6s_cantforward++;
	190	if (mcopy) {
	191	#if 0
	192	/* XXX: what icmp ? */
	193	#else
	194	m_freem(mcopy);
	195	#endif
	196	}
91447636	197	lck_mtx_unlock(sadb_mutex);
1c79356b A	198	m_freem(m);
	199	return;
	200	}
	201
	202	error = 0;
	203
	204	/* check policy */
	205	switch (sp->policy) {
	206	case IPSEC_POLICY_DISCARD:
	207	/*
	208	* This packet is just discarded.
	209	*/
	210	ipsec6stat.out_polvio++;
	211	ip6stat.ip6s_cantforward++;
	212	key_freesp(sp);
	213	if (mcopy) {
	214	#if 0
	215	/* XXX: what icmp ? */
	216	#else
	217	m_freem(mcopy);
	218	#endif
	219	}
91447636	220	lck_mtx_unlock(sadb_mutex);
1c79356b A	221	m_freem(m);
	222	return;
	223
	224	case IPSEC_POLICY_BYPASS:
	225	case IPSEC_POLICY_NONE:
	226	/* no need to do IPsec. */
	227	key_freesp(sp);
91447636	228	lck_mtx_unlock(sadb_mutex);
1c79356b	229	goto skip_ipsec;
9bccf70c	230
1c79356b A	231	case IPSEC_POLICY_IPSEC:
	232	if (sp->req == NULL) {
	233	/* XXX should be panic ? */
	234	printf("ip6_forward: No IPsec request specified.\n");
	235	ip6stat.ip6s_cantforward++;
	236	key_freesp(sp);
	237	if (mcopy) {
	238	#if 0
	239	/* XXX: what icmp ? */
	240	#else
	241	m_freem(mcopy);
	242	#endif
	243	}
91447636	244	lck_mtx_unlock(sadb_mutex);
1c79356b A	245	m_freem(m);
	246	return;
	247	}
	248	/* do IPsec */
	249	break;
	250
	251	case IPSEC_POLICY_ENTRUST:
	252	default:
	253	/* should be panic ?? */
	254	printf("ip6_forward: Invalid policy found. %d\n", sp->policy);
	255	key_freesp(sp);
91447636	256	lck_mtx_unlock(sadb_mutex);
1c79356b A	257	goto skip_ipsec;
	258	}
	259
	260	{
	261	struct ipsec_output_state state;
	262
	263	/*
	264	* All the extension headers will become inaccessible
	265	* (since they can be encrypted).
	266	* Don't panic, we need no more updates to extension headers
	267	* on inner IPv6 packet (since they are now encapsulated).
	268	*
	269	* IPv6 [ESP\|AH] IPv6 [extension headers] payload
	270	*/
	271	bzero(&state, sizeof(state));
	272	state.m = m;
	273	state.ro = NULL; /* update at ipsec6_output_tunnel() */
	274	state.dst = NULL; /* update at ipsec6_output_tunnel() */
	275
13fec989 A	276	if (locked)
13fec989 A	277	lck_mtx_unlock(ip6_mutex);
1c79356b	278	error = ipsec6_output_tunnel(&state, sp, 0);
13fec989 A	279	if (locked) {
	280	lck_mtx_unlock(sadb_mutex);
	281	lck_mtx_lock(ip6_mutex);
	282	lck_mtx_lock(sadb_mutex);
	283	}
1c79356b A	284
1c79356b A	285	m = state.m;
1c79356b A	286	key_freesp(sp);
	287
	288	if (error) {
	289	/* mbuf is already reclaimed in ipsec6_output_tunnel. */
	290	switch (error) {
	291	case EHOSTUNREACH:
	292	case ENETUNREACH:
	293	case EMSGSIZE:
	294	case ENOBUFS:
	295	case ENOMEM:
	296	break;
	297	default:
	298	printf("ip6_output (ipsec): error code %d\n", error);
55e303ae	299	/* fall through */
1c79356b A	300	case ENOENT:
	301	/* don't show these error codes to the user */
	302	break;
	303	}
	304	ip6stat.ip6s_cantforward++;
	305	if (mcopy) {
	306	#if 0
	307	/* XXX: what icmp ? */
	308	#else
	309	m_freem(mcopy);
	310	#endif
	311	}
91447636	312	lck_mtx_unlock(sadb_mutex);
1c79356b A	313	m_freem(m);
	314	return;
	315	}
	316	}
91447636	317	lck_mtx_unlock(sadb_mutex);
1c79356b	318	skip_ipsec:
9bccf70c A	319	#endif /* IPSEC */
	320
	321	dst = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
1c79356b A	322	if (!srcrt) {
	323	/*
	324	* ip6_forward_rt.ro_dst.sin6_addr is equal to ip6->ip6_dst
	325	*/
	326	if (ip6_forward_rt.ro_rt == 0 \|\|
	327	(ip6_forward_rt.ro_rt->rt_flags & RTF_UP) == 0) {
	328	if (ip6_forward_rt.ro_rt) {
9bccf70c	329	rtfree(ip6_forward_rt.ro_rt);
1c79356b A	330	ip6_forward_rt.ro_rt = 0;
	331	}
	332	/* this probably fails but give it a try again */
1c79356b A	333	rtalloc_ign((struct route *)&ip6_forward_rt,
1c79356b A	334	RTF_PRCLONING);
1c79356b	335	}
9bccf70c	336
1c79356b A	337	if (ip6_forward_rt.ro_rt == 0) {
1c79356b A	338	ip6stat.ip6s_noroute++;
9bccf70c	339	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_noroute);
1c79356b	340	if (mcopy) {
91447636 A	341	if (locked)
91447636 A	342	lck_mtx_unlock(ip6_mutex);
1c79356b A	343	icmp6_error(mcopy, ICMP6_DST_UNREACH,
1c79356b A	344	ICMP6_DST_UNREACH_NOROUTE, 0);
91447636 A	345	if (locked)
91447636 A	346	lck_mtx_lock(ip6_mutex);
1c79356b A	347	}
	348	m_freem(m);
	349	return;
	350	}
	351	} else if ((rt = ip6_forward_rt.ro_rt) == 0 \|\|
	352	!IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &dst->sin6_addr)) {
	353	if (ip6_forward_rt.ro_rt) {
9bccf70c	354	rtfree(ip6_forward_rt.ro_rt);
1c79356b A	355	ip6_forward_rt.ro_rt = 0;
	356	}
	357	bzero(dst, sizeof(*dst));
	358	dst->sin6_len = sizeof(struct sockaddr_in6);
	359	dst->sin6_family = AF_INET6;
	360	dst->sin6_addr = ip6->ip6_dst;
	361
1c79356b	362	rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING);
1c79356b A	363	if (ip6_forward_rt.ro_rt == 0) {
1c79356b A	364	ip6stat.ip6s_noroute++;
9bccf70c	365	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_noroute);
1c79356b	366	if (mcopy) {
91447636 A	367	if (locked)
91447636 A	368	lck_mtx_unlock(ip6_mutex);
1c79356b A	369	icmp6_error(mcopy, ICMP6_DST_UNREACH,
1c79356b A	370	ICMP6_DST_UNREACH_NOROUTE, 0);
91447636 A	371	if (locked)
91447636 A	372	lck_mtx_lock(ip6_mutex);
1c79356b A	373	}
	374	m_freem(m);
	375	return;
	376	}
	377	}
	378	rt = ip6_forward_rt.ro_rt;
	379
	380	/*
	381	* Scope check: if a packet can't be delivered to its destination
	382	* for the reason that the destination is beyond the scope of the
	383	* source address, discard the packet and return an icmp6 destination
	384	* unreachable error with Code 2 (beyond scope of source address).
55e303ae	385	* [draft-ietf-ipngwg-icmp-v3-02.txt, Section 3.1]
1c79356b A	386	*/
	387	if (in6_addr2scopeid(m->m_pkthdr.rcvif, &ip6->ip6_src) !=
	388	in6_addr2scopeid(rt->rt_ifp, &ip6->ip6_src)) {
	389	ip6stat.ip6s_cantforward++;
	390	ip6stat.ip6s_badscope++;
	391	in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard);
	392
91447636 A	393	if (ip6_log_time + ip6_log_interval < timenow.tv_sec) {
91447636 A	394	ip6_log_time = timenow.tv_sec;
1c79356b A	395	log(LOG_DEBUG,
	396	"cannot forward "
	397	"src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
	398	ip6_sprintf(&ip6->ip6_src),
	399	ip6_sprintf(&ip6->ip6_dst),
	400	ip6->ip6_nxt,
	401	if_name(m->m_pkthdr.rcvif), if_name(rt->rt_ifp));
	402	}
91447636 A	403	if (mcopy) {
	404	if (locked)
	405	lck_mtx_unlock(ip6_mutex);
1c79356b A	406	icmp6_error(mcopy, ICMP6_DST_UNREACH,
1c79356b A	407	ICMP6_DST_UNREACH_BEYONDSCOPE, 0);
91447636 A	408	if (locked)
	409	lck_mtx_lock(ip6_mutex);
	410	}
1c79356b A	411	m_freem(m);
	412	return;
	413	}
	414
	415	if (m->m_pkthdr.len > rt->rt_ifp->if_mtu) {
	416	in6_ifstat_inc(rt->rt_ifp, ifs6_in_toobig);
	417	if (mcopy) {
	418	u_long mtu;
9bccf70c	419	#if IPSEC
1c79356b A	420	struct secpolicy *sp;
	421	int ipsecerror;
	422	size_t ipsechdrsiz;
	423	#endif
	424
	425	mtu = rt->rt_ifp->if_mtu;
55e303ae	426	#if IPSEC
1c79356b A	427	/*
	428	* When we do IPsec tunnel ingress, we need to play
	429	* with if_mtu value (decrement IPsec header size
	430	* from mtu value). The code is much simpler than v4
	431	* case, as we have the outgoing interface for
	432	* encapsulated packet as "rt->rt_ifp".
	433	*/
91447636	434	lck_mtx_lock(sadb_mutex);
1c79356b A	435	sp = ipsec6_getpolicybyaddr(mcopy, IPSEC_DIR_OUTBOUND,
	436	IP_FORWARDING, &ipsecerror);
	437	if (sp) {
	438	ipsechdrsiz = ipsec6_hdrsiz(mcopy,
	439	IPSEC_DIR_OUTBOUND, NULL);
	440	if (ipsechdrsiz < mtu)
	441	mtu -= ipsechdrsiz;
	442	}
91447636	443	lck_mtx_unlock(sadb_mutex);
1c79356b A	444	/*
	445	* if mtu becomes less than minimum MTU,
	446	* tell minimum MTU (and I'll need to fragment it).
	447	*/
	448	if (mtu < IPV6_MMTU)
	449	mtu = IPV6_MMTU;
	450	#endif
91447636 A	451	if (locked)
91447636 A	452	lck_mtx_unlock(ip6_mutex);
1c79356b	453	icmp6_error(mcopy, ICMP6_PACKET_TOO_BIG, 0, mtu);
91447636 A	454	if (locked)
91447636 A	455	lck_mtx_lock(ip6_mutex);
1c79356b A	456	}
	457	m_freem(m);
	458	return;
	459	}
	460
	461	if (rt->rt_flags & RTF_GATEWAY)
	462	dst = (struct sockaddr_in6 *)rt->rt_gateway;
	463
	464	/*
	465	* If we are to forward the packet using the same interface
	466	* as one we got the packet from, perhaps we should send a redirect
	467	* to sender to shortcut a hop.
	468	* Only send redirect if source is sending directly to us,
	469	* and if packet was not source routed (or has any options).
	470	* Also, don't send redirect if forwarding using a route
	471	* modified by a redirect.
	472	*/
	473	if (rt->rt_ifp == m->m_pkthdr.rcvif && !srcrt &&
9bccf70c A	474	(rt->rt_flags & (RTF_DYNAMIC\|RTF_MODIFIED)) == 0) {
	475	if ((rt->rt_ifp->if_flags & IFF_POINTOPOINT) != 0) {
	476	/*
	477	* If the incoming interface is equal to the outgoing
	478	* one, and the link attached to the interface is
	479	* point-to-point, then it will be highly probable
	480	* that a routing loop occurs. Thus, we immediately
	481	* drop the packet and send an ICMPv6 error message.
	482	*
	483	* type/code is based on suggestion by Rich Draves.
	484	* not sure if it is the best pick.
	485	*/
91447636 A	486	if (locked)
91447636 A	487	lck_mtx_unlock(ip6_mutex);
9bccf70c A	488	icmp6_error(mcopy, ICMP6_DST_UNREACH,
9bccf70c A	489	ICMP6_DST_UNREACH_ADDR, 0);
91447636 A	490	if (locked)
91447636 A	491	lck_mtx_lock(ip6_mutex);
9bccf70c A	492	m_freem(m);
	493	return;
	494	}
1c79356b	495	type = ND_REDIRECT;
9bccf70c	496	}
1c79356b	497
1c79356b A	498	/*
	499	* Check with the firewall...
	500	*/
9bccf70c	501	if (ip6_fw_enable && ip6_fw_chk_ptr) {
1c79356b A	502	u_short port = 0;
1c79356b A	503	/* If ipfw says divert, we have to just drop packet */
91447636	504	if (ip6_fw_chk_ptr(&ip6, rt->rt_ifp, &port, &m)) {
1c79356b A	505	m_freem(m);
	506	goto freecopy;
	507	}
	508	if (!m)
	509	goto freecopy;
	510	}
1c79356b	511
9bccf70c A	512	/*
	513	* Fake scoped addresses. Note that even link-local source or
	514	* destinaion can appear, if the originating node just sends the
	515	* packet to us (without address resolution for the destination).
	516	* Since both icmp6_error and icmp6_redirect_output fill the embedded
	517	* link identifiers, we can do this stuff after making a copy for
	518	* returning an error.
	519	*/
	520	if ((rt->rt_ifp->if_flags & IFF_LOOPBACK) != 0) {
	521	/*
	522	* See corresponding comments in ip6_output.
	523	* XXX: but is it possible that ip6_forward() sends a packet
	524	* to a loopback interface? I don't think so, and thus
	525	* I bark here. (jinmei@kame.net)
	526	* XXX: it is common to route invalid packets to loopback.
	527	* also, the codepath will be visited on use of ::1 in
	528	* rthdr. (itojun)
	529	*/
	530	#if 1
	531	if (0)
1c79356b	532	#else
9bccf70c A	533	if ((rt->rt_flags & (RTF_BLACKHOLE\|RTF_REJECT)) == 0)
	534	#endif
	535	{
	536	printf("ip6_forward: outgoing interface is loopback. "
55e303ae A	537	"src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
	538	ip6_sprintf(&ip6->ip6_src),
	539	ip6_sprintf(&ip6->ip6_dst),
	540	ip6->ip6_nxt, if_name(m->m_pkthdr.rcvif),
	541	if_name(rt->rt_ifp));
9bccf70c A	542	}
	543
	544	/* we can just use rcvif in forwarding. */
	545	origifp = m->m_pkthdr.rcvif;
	546	}
	547	else
	548	origifp = rt->rt_ifp;
	549	#ifndef SCOPEDROUTING
	550	/*
	551	* clear embedded scope identifiers if necessary.
	552	* in6_clearscope will touch the addresses only when necessary.
	553	*/
	554	in6_clearscope(&ip6->ip6_src);
	555	in6_clearscope(&ip6->ip6_dst);
	556	#endif
	557
91447636	558	error = nd6_output(rt->rt_ifp, origifp, m, dst, rt, locked);
1c79356b A	559	if (error) {
	560	in6_ifstat_inc(rt->rt_ifp, ifs6_out_discard);
	561	ip6stat.ip6s_cantforward++;
	562	} else {
	563	ip6stat.ip6s_forward++;
	564	in6_ifstat_inc(rt->rt_ifp, ifs6_out_forward);
	565	if (type)
	566	ip6stat.ip6s_redirectsent++;
	567	else {
	568	if (mcopy)
	569	goto freecopy;
	570	}
	571	}
	572	if (mcopy == NULL)
	573	return;
	574
	575	switch (error) {
	576	case 0:
	577	#if 1
	578	if (type == ND_REDIRECT) {
	579	icmp6_redirect_output(mcopy, rt);
	580	return;
	581	}
	582	#endif
	583	goto freecopy;
	584
	585	case EMSGSIZE:
	586	/* xxx MTU is constant in PPP? */
	587	goto freecopy;
	588
	589	case ENOBUFS:
	590	/* Tell source to slow down like source quench in IP? */
	591	goto freecopy;
	592
	593	case ENETUNREACH: /* shouldn't happen, checked above */
	594	case EHOSTUNREACH:
	595	case ENETDOWN:
	596	case EHOSTDOWN:
	597	default:
	598	type = ICMP6_DST_UNREACH;
	599	code = ICMP6_DST_UNREACH_ADDR;
	600	break;
	601	}
91447636 A	602	if (locked)
91447636 A	603	lck_mtx_unlock(ip6_mutex);
1c79356b	604	icmp6_error(mcopy, type, code, 0);
91447636 A	605	if (locked)
91447636 A	606	lck_mtx_lock(ip6_mutex);
1c79356b A	607	return;
	608
	609	freecopy:
	610	m_freem(mcopy);
	611	return;
	612	}