1 /*      $FreeBSD: src/sys/netinet6/ip6_forward.c,v 1.16 2002/10/16 02:25:05 sam Exp $   */ 
   2 /*      $KAME: ip6_forward.c,v 1.69 2001/05/17 03:48:30 itojun Exp $    */ 
   5  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 
   8  * Redistribution and use in source and binary forms, with or without 
   9  * modification, are permitted provided that the following conditions 
  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. 
  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 
  34 #include <sys/param.h> 
  35 #include <sys/systm.h> 
  36 #include <sys/malloc.h> 
  38 #include <sys/domain.h> 
  39 #include <sys/protosw.h> 
  40 #include <sys/socket.h> 
  41 #include <sys/errno.h> 
  43 #include <sys/kernel.h> 
  44 #include <sys/syslog.h> 
  47 #include <net/route.h> 
  49 #include <netinet/in.h> 
  50 #include <netinet/in_var.h> 
  51 #include <netinet/in_systm.h> 
  52 #include <netinet/ip.h> 
  53 #include <netinet/ip_var.h> 
  54 #include <netinet6/in6_var.h> 
  55 #include <netinet/ip6.h> 
  56 #include <netinet6/ip6_var.h> 
  57 #include <netinet/icmp6.h> 
  58 #include <netinet6/nd6.h> 
  60 #include <netinet/in_pcb.h> 
  63 #include <netinet6/ipsec.h> 
  65 #include <netinet6/ipsec6.h> 
  67 #include <netkey/key.h> 
  68 extern int ipsec_bypass
; 
  69 extern lck_mtx_t 
*sadb_mutex
; 
  70 extern lck_mtx_t 
*ip6_mutex
; 
  73 #include <netinet6/ip6_fw.h> 
  75 #include <net/net_osdep.h> 
  77 struct  route_in6 ip6_forward_rt
; 
  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 
  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. 
  93 ip6_forward(m
, srcrt
, locked
) 
  98         struct ip6_hdr 
*ip6 
= mtod(m
, struct ip6_hdr 
*); 
  99         struct sockaddr_in6 
*dst
; 
 101         int error
, type 
= 0, code 
= 0; 
 102         struct mbuf 
*mcopy 
= NULL
; 
 103         struct ifnet 
*origifp
;  /* maybe unnecessary */ 
 105         struct secpolicy 
*sp 
= NULL
; 
 107         struct timeval timenow
; 
 109         getmicrotime(&timenow
); 
 114          * Check AH/ESP integrity. 
 117          * Don't increment ip6s_cantforward because this is the check 
 118          * before forwarding packet actually. 
 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
); 
 128                 lck_mtx_unlock(sadb_mutex
); 
 133          * Do not forward packets to multicast destination (should be handled 
 135          * Do not forward packets with unspecified source.  It was discussed 
 136          * in July 2000, on ipngwg mailing list. 
 138         if ((m
->m_flags 
& (M_BCAST
|M_MCAST
)) != 0 || 
 139             IN6_IS_ADDR_MULTICAST(&ip6
->ip6_dst
) || 
 140             IN6_IS_ADDR_UNSPECIFIED(&ip6
->ip6_src
)) { 
 141                 ip6stat
.ip6s_cantforward
++; 
 142                 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */ 
 143                 if (ip6_log_time 
+ ip6_log_interval 
< timenow
.tv_sec
) { 
 144                         ip6_log_time 
= timenow
.tv_sec
; 
 147                             "from %s to %s nxt %d received on %s\n", 
 148                             ip6_sprintf(&ip6
->ip6_src
), 
 149                             ip6_sprintf(&ip6
->ip6_dst
), 
 151                             if_name(m
->m_pkthdr
.rcvif
)); 
 157         if (ip6
->ip6_hlim 
<= IPV6_HLIMDEC
) { 
 158                 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */ 
 160                         lck_mtx_unlock(ip6_mutex
); 
 161                 icmp6_error(m
, ICMP6_TIME_EXCEEDED
, 
 162                                 ICMP6_TIME_EXCEED_TRANSIT
, 0); 
 164                         lck_mtx_lock(ip6_mutex
); 
 167         ip6
->ip6_hlim 
-= IPV6_HLIMDEC
; 
 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. 
 175          * It is important to save it before IPsec processing as IPsec 
 176          * processing may modify the mbuf. 
 178         mcopy 
= m_copy(m
, 0, imin(m
->m_pkthdr
.len
, ICMPV6_PLD_MAXLEN
)); 
 181         if (ipsec_bypass 
!= 0) 
 183         lck_mtx_lock(sadb_mutex
); 
 184         /* get a security policy for this packet */ 
 185         sp 
= ipsec6_getpolicybyaddr(m
, IPSEC_DIR_OUTBOUND
, IP_FORWARDING
, 
 188                 ipsec6stat
.out_inval
++; 
 189                 ip6stat
.ip6s_cantforward
++; 
 192                         /* XXX: what icmp ? */ 
 197                 lck_mtx_unlock(sadb_mutex
); 
 205         switch (sp
->policy
) { 
 206         case IPSEC_POLICY_DISCARD
: 
 208                  * This packet is just discarded. 
 210                 ipsec6stat
.out_polvio
++; 
 211                 ip6stat
.ip6s_cantforward
++; 
 215                         /* XXX: what icmp ? */ 
 220                 lck_mtx_unlock(sadb_mutex
); 
 224         case IPSEC_POLICY_BYPASS
: 
 225         case IPSEC_POLICY_NONE
: 
 226                 /* no need to do IPsec. */ 
 228                 lck_mtx_unlock(sadb_mutex
); 
 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
++; 
 239                                 /* XXX: what icmp ? */ 
 244                         lck_mtx_unlock(sadb_mutex
); 
 251         case IPSEC_POLICY_ENTRUST
: 
 253                 /* should be panic ?? */ 
 254                 printf("ip6_forward: Invalid policy found. %d\n", sp
->policy
); 
 256                 lck_mtx_unlock(sadb_mutex
); 
 261         struct ipsec_output_state state
; 
 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). 
 269          * IPv6 [ESP|AH] IPv6 [extension headers] payload 
 271         bzero(&state
, sizeof(state
)); 
 273         state
.ro 
= NULL
;        /* update at ipsec6_output_tunnel() */ 
 274         state
.dst 
= NULL
;       /* update at ipsec6_output_tunnel() */ 
 277                         lck_mtx_unlock(ip6_mutex
); 
 278         error 
= ipsec6_output_tunnel(&state
, sp
, 0); 
 280                         lck_mtx_unlock(sadb_mutex
); 
 281                         lck_mtx_lock(ip6_mutex
); 
 282                         lck_mtx_lock(sadb_mutex
); 
 289                 /* mbuf is already reclaimed in ipsec6_output_tunnel. */ 
 298                         printf("ip6_output (ipsec): error code %d\n", error
); 
 301                         /* don't show these error codes to the user */ 
 304                 ip6stat
.ip6s_cantforward
++; 
 307                         /* XXX: what icmp ? */ 
 312                 lck_mtx_unlock(sadb_mutex
); 
 317         lck_mtx_unlock(sadb_mutex
); 
 321         dst 
= (struct sockaddr_in6 
*)&ip6_forward_rt
.ro_dst
; 
 324                  * ip6_forward_rt.ro_dst.sin6_addr is equal to ip6->ip6_dst 
 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
) { 
 329                                 rtfree(ip6_forward_rt
.ro_rt
); 
 330                                 ip6_forward_rt
.ro_rt 
= 0; 
 332                         /* this probably fails but give it a try again */ 
 333                         rtalloc_ign((struct route 
*)&ip6_forward_rt
, 
 337                 if (ip6_forward_rt
.ro_rt 
== 0) { 
 338                         ip6stat
.ip6s_noroute
++; 
 339                         in6_ifstat_inc(m
->m_pkthdr
.rcvif
, ifs6_in_noroute
); 
 342                                         lck_mtx_unlock(ip6_mutex
); 
 343                                 icmp6_error(mcopy
, ICMP6_DST_UNREACH
, 
 344                                             ICMP6_DST_UNREACH_NOROUTE
, 0); 
 346                                         lck_mtx_lock(ip6_mutex
); 
 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
) { 
 354                         rtfree(ip6_forward_rt
.ro_rt
); 
 355                         ip6_forward_rt
.ro_rt 
= 0; 
 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
; 
 362                 rtalloc_ign((struct route 
*)&ip6_forward_rt
, RTF_PRCLONING
); 
 363                 if (ip6_forward_rt
.ro_rt 
== 0) { 
 364                         ip6stat
.ip6s_noroute
++; 
 365                         in6_ifstat_inc(m
->m_pkthdr
.rcvif
, ifs6_in_noroute
); 
 368                                         lck_mtx_unlock(ip6_mutex
); 
 369                                 icmp6_error(mcopy
, ICMP6_DST_UNREACH
, 
 370                                             ICMP6_DST_UNREACH_NOROUTE
, 0); 
 372                                         lck_mtx_lock(ip6_mutex
); 
 378         rt 
= ip6_forward_rt
.ro_rt
; 
 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). 
 385          * [draft-ietf-ipngwg-icmp-v3-02.txt, Section 3.1] 
 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
); 
 393                 if (ip6_log_time 
+ ip6_log_interval 
< timenow
.tv_sec
) { 
 394                         ip6_log_time 
= timenow
.tv_sec
; 
 397                             "src %s, dst %s, nxt %d, rcvif %s, outif %s\n", 
 398                             ip6_sprintf(&ip6
->ip6_src
), 
 399                             ip6_sprintf(&ip6
->ip6_dst
), 
 401                             if_name(m
->m_pkthdr
.rcvif
), if_name(rt
->rt_ifp
)); 
 405                                 lck_mtx_unlock(ip6_mutex
); 
 406                         icmp6_error(mcopy
, ICMP6_DST_UNREACH
, 
 407                                     ICMP6_DST_UNREACH_BEYONDSCOPE
, 0); 
 409                                 lck_mtx_lock(ip6_mutex
); 
 415         if (m
->m_pkthdr
.len 
> rt
->rt_ifp
->if_mtu
) { 
 416                 in6_ifstat_inc(rt
->rt_ifp
, ifs6_in_toobig
); 
 420                         struct secpolicy 
*sp
; 
 425                         mtu 
= rt
->rt_ifp
->if_mtu
; 
 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". 
 434                         lck_mtx_lock(sadb_mutex
); 
 435                         sp 
= ipsec6_getpolicybyaddr(mcopy
, IPSEC_DIR_OUTBOUND
, 
 436                                 IP_FORWARDING
, &ipsecerror
); 
 438                                 ipsechdrsiz 
= ipsec6_hdrsiz(mcopy
, 
 439                                         IPSEC_DIR_OUTBOUND
, NULL
); 
 440                                 if (ipsechdrsiz 
< mtu
) 
 443                         lck_mtx_unlock(sadb_mutex
); 
 445                          * if mtu becomes less than minimum MTU, 
 446                          * tell minimum MTU (and I'll need to fragment it). 
 452                                 lck_mtx_unlock(ip6_mutex
); 
 453                         icmp6_error(mcopy
, ICMP6_PACKET_TOO_BIG
, 0, mtu
); 
 455                                 lck_mtx_lock(ip6_mutex
); 
 461         if (rt
->rt_flags 
& RTF_GATEWAY
) 
 462                 dst 
= (struct sockaddr_in6 
*)rt
->rt_gateway
; 
 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. 
 473         if (rt
->rt_ifp 
== m
->m_pkthdr
.rcvif 
&& !srcrt 
&& 
 474             (rt
->rt_flags 
& (RTF_DYNAMIC
|RTF_MODIFIED
)) == 0) { 
 475                 if ((rt
->rt_ifp
->if_flags 
& IFF_POINTOPOINT
) != 0) { 
 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. 
 483                          * type/code is based on suggestion by Rich Draves. 
 484                          * not sure if it is the best pick. 
 487                                 lck_mtx_unlock(ip6_mutex
); 
 488                         icmp6_error(mcopy
, ICMP6_DST_UNREACH
, 
 489                                     ICMP6_DST_UNREACH_ADDR
, 0); 
 491                                 lck_mtx_lock(ip6_mutex
); 
 499          * Check with the firewall... 
 501         if (ip6_fw_enable 
&& ip6_fw_chk_ptr
) { 
 503                 /* If ipfw says divert, we have to just drop packet */ 
 504                 if (ip6_fw_chk_ptr(&ip6
, rt
->rt_ifp
, &port
, &m
)) { 
 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. 
 520         if ((rt
->rt_ifp
->if_flags 
& IFF_LOOPBACK
) != 0) { 
 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 
 533                 if ((rt
->rt_flags 
& (RTF_BLACKHOLE
|RTF_REJECT
)) == 0) 
 536                         printf("ip6_forward: outgoing interface is loopback. " 
 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
)); 
 544                 /* we can just use rcvif in forwarding. */ 
 545                 origifp 
= m
->m_pkthdr
.rcvif
; 
 548                 origifp 
= rt
->rt_ifp
; 
 549 #ifndef SCOPEDROUTING 
 551          * clear embedded scope identifiers if necessary. 
 552          * in6_clearscope will touch the addresses only when necessary. 
 554         in6_clearscope(&ip6
->ip6_src
); 
 555         in6_clearscope(&ip6
->ip6_dst
); 
 558         error 
= nd6_output(rt
->rt_ifp
, origifp
, m
, dst
, rt
, locked
); 
 560                 in6_ifstat_inc(rt
->rt_ifp
, ifs6_out_discard
); 
 561                 ip6stat
.ip6s_cantforward
++; 
 563                 ip6stat
.ip6s_forward
++; 
 564                 in6_ifstat_inc(rt
->rt_ifp
, ifs6_out_forward
); 
 566                         ip6stat
.ip6s_redirectsent
++; 
 578                 if (type 
== ND_REDIRECT
) { 
 579                         icmp6_redirect_output(mcopy
, rt
); 
 586                 /* xxx MTU is constant in PPP? */ 
 590                 /* Tell source to slow down like source quench in IP? */ 
 593         case ENETUNREACH
:       /* shouldn't happen, checked above */ 
 598                 type 
= ICMP6_DST_UNREACH
; 
 599                 code 
= ICMP6_DST_UNREACH_ADDR
; 
 603                 lck_mtx_unlock(ip6_mutex
); 
 604         icmp6_error(mcopy
, type
, code
, 0); 
 606                 lck_mtx_lock(ip6_mutex
);