The Design and Implementation of the FreeBSD Operating System, Second Edition
Now available: The Design and Implementation of the FreeBSD Operating System (Second Edition)


[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]

FreeBSD/Linux Kernel Cross Reference
sys/bsd/netinet6/ip6_output.c

Version: -  FREEBSD  -  FREEBSD-13-STABLE  -  FREEBSD-13-0  -  FREEBSD-12-STABLE  -  FREEBSD-12-0  -  FREEBSD-11-STABLE  -  FREEBSD-11-0  -  FREEBSD-10-STABLE  -  FREEBSD-10-0  -  FREEBSD-9-STABLE  -  FREEBSD-9-0  -  FREEBSD-8-STABLE  -  FREEBSD-8-0  -  FREEBSD-7-STABLE  -  FREEBSD-7-0  -  FREEBSD-6-STABLE  -  FREEBSD-6-0  -  FREEBSD-5-STABLE  -  FREEBSD-5-0  -  FREEBSD-4-STABLE  -  FREEBSD-3-STABLE  -  FREEBSD22  -  l41  -  OPENBSD  -  linux-2.6  -  MK84  -  PLAN9  -  xnu-8792 
SearchContext: -  none  -  3  -  10 

    1 /*      $FreeBSD: src/sys/netinet6/ip6_output.c,v 1.43 2002/10/31 19:45:48 ume Exp $    */
    2 /*      $KAME: ip6_output.c,v 1.279 2002/01/26 06:12:30 jinmei Exp $    */
    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 
   33 /*
   34  * Copyright (c) 1982, 1986, 1988, 1990, 1993
   35  *      The Regents of the University of California.  All rights reserved.
   36  *
   37  * Redistribution and use in source and binary forms, with or without
   38  * modification, are permitted provided that the following conditions
   39  * are met:
   40  * 1. Redistributions of source code must retain the above copyright
   41  *    notice, this list of conditions and the following disclaimer.
   42  * 2. Redistributions in binary form must reproduce the above copyright
   43  *    notice, this list of conditions and the following disclaimer in the
   44  *    documentation and/or other materials provided with the distribution.
   45  * 3. All advertising materials mentioning features or use of this software
   46  *    must display the following acknowledgement:
   47  *      This product includes software developed by the University of
   48  *      California, Berkeley and its contributors.
   49  * 4. Neither the name of the University nor the names of its contributors
   50  *    may be used to endorse or promote products derived from this software
   51  *    without specific prior written permission.
   52  *
   53  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
   54  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
   55  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
   56  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
   57  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
   58  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
   59  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
   60  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
   61  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
   62  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
   63  * SUCH DAMAGE.
   64  *
   65  *      @(#)ip_output.c 8.3 (Berkeley) 1/21/94
   66  */
   67 
   68 
   69 #include <sys/param.h>
   70 #include <sys/malloc.h>
   71 #include <sys/mbuf.h>
   72 #include <sys/errno.h>
   73 #include <sys/protosw.h>
   74 #include <sys/socket.h>
   75 #include <sys/socketvar.h>
   76 #include <sys/systm.h>
   77 #include <sys/kernel.h>
   78 #include <sys/proc.h>
   79 
   80 #include <net/if.h>
   81 #include <net/route.h>
   82 
   83 #include <netinet/in.h>
   84 #include <netinet/in_var.h>
   85 #include <netinet/ip_var.h>
   86 #include <netinet6/in6_var.h>
   87 #include <netinet/ip6.h>
   88 #include <netinet/icmp6.h>
   89 #include <netinet6/ip6_var.h>
   90 #include <netinet/in_pcb.h>
   91 #include <netinet6/nd6.h>
   92 
   93 #if IPSEC
   94 #include <netinet6/ipsec.h>
   95 #if INET6
   96 #include <netinet6/ipsec6.h>
   97 #endif
   98 #include <netkey/key.h>
   99 extern int ipsec_bypass;
  100 #endif /* IPSEC */
  101 
  102 #include <netinet6/ip6_fw.h>
  103 
  104 #include <net/net_osdep.h>
  105 
  106 #ifndef __APPLE__
  107 static MALLOC_DEFINE(M_IPMOPTS, "ip6_moptions", "internet multicast options");
  108 #endif
  109 
  110 
  111 static u_long  lo_dl_tag = 0;
  112 extern u_long  route_generation;
  113 
  114 
  115 struct ip6_exthdrs {
  116         struct mbuf *ip6e_ip6;
  117         struct mbuf *ip6e_hbh;
  118         struct mbuf *ip6e_dest1;
  119         struct mbuf *ip6e_rthdr;
  120         struct mbuf *ip6e_dest2;
  121 };
  122 
  123 static int ip6_pcbopts __P((struct ip6_pktopts **, struct mbuf *,
  124                             struct socket *, struct sockopt *sopt));
  125 static int ip6_setmoptions __P((int, struct inpcb *, struct mbuf *));
  126 static int ip6_getmoptions __P((int, struct ip6_moptions *, struct mbuf **));
  127 static int ip6_copyexthdr __P((struct mbuf **, caddr_t, int));
  128 static int ip6_insertfraghdr __P((struct mbuf *, struct mbuf *, int,
  129                                   struct ip6_frag **));
  130 static int ip6_insert_jumboopt __P((struct ip6_exthdrs *, u_int32_t));
  131 static int ip6_splithdr __P((struct mbuf *, struct ip6_exthdrs *));
  132 
  133 extern int ip_createmoptions(struct ip_moptions **imop);
  134 extern int ip_addmembership(struct ip_moptions *imo, struct ip_mreq *mreq);
  135 extern int ip_dropmembership(struct ip_moptions *imo, struct ip_mreq *mreq);
  136 
  137 /*
  138  * IP6 output. The packet in mbuf chain m contains a skeletal IP6
  139  * header (with pri, len, nxt, hlim, src, dst).
  140  * This function may modify ver and hlim only.
  141  * The mbuf chain containing the packet will be freed.
  142  * The mbuf opt, if present, will not be freed.
  143  *
  144  * type of "mtu": rt_rmx.rmx_mtu is u_long, ifnet.ifr_mtu is int, and
  145  * nd_ifinfo.linkmtu is u_int32_t.  so we use u_long to hold largest one,
  146  * which is rt_rmx.rmx_mtu.
  147  */
  148 int
  149 ip6_output(m0, opt, ro, flags, im6o, ifpp)
  150         struct mbuf *m0;
  151         struct ip6_pktopts *opt;
  152         struct route_in6 *ro;
  153         int flags;
  154         struct ip6_moptions *im6o;
  155         struct ifnet **ifpp;            /* XXX: just for statistics */
  156 {
  157         struct ip6_hdr *ip6, *mhip6;
  158         struct ifnet *ifp, *origifp;
  159         struct mbuf *m = m0;
  160         int hlen, tlen, len, off;
  161         struct route_in6 ip6route;
  162         struct sockaddr_in6 *dst;
  163         int error = 0;
  164         struct in6_ifaddr *ia = NULL;
  165         u_long mtu;
  166         u_int32_t optlen = 0, plen = 0, unfragpartlen = 0;
  167         struct ip6_exthdrs exthdrs;
  168         struct in6_addr finaldst;
  169         struct route_in6 *ro_pmtu = NULL;
  170         int hdrsplit = 0;
  171         int needipsec = 0;
  172 #if IPSEC
  173         int needipsectun = 0;
  174         struct socket *so = NULL;
  175         struct secpolicy *sp = NULL;
  176 
  177         /* for AH processing. stupid to have "socket" variable in IP layer... */
  178         if (ipsec_bypass == 0)
  179         {
  180                 so = ipsec_getsocket(m);
  181                 (void)ipsec_setsocket(m, NULL);
  182         }
  183         ip6 = mtod(m, struct ip6_hdr *);
  184 #endif /* IPSEC */
  185 
  186 #define MAKE_EXTHDR(hp, mp)                                             \
  187     do {                                                                \
  188         if (hp) {                                                       \
  189                 struct ip6_ext *eh = (struct ip6_ext *)(hp);            \
  190                 error = ip6_copyexthdr((mp), (caddr_t)(hp),             \
  191                                        ((eh)->ip6e_len + 1) << 3);      \
  192                 if (error)                                              \
  193                         goto freehdrs;                                  \
  194         }                                                               \
  195     } while (0)
  196         
  197         bzero(&exthdrs, sizeof(exthdrs));
  198         
  199         if (opt) {
  200                 /* Hop-by-Hop options header */
  201                 MAKE_EXTHDR(opt->ip6po_hbh, &exthdrs.ip6e_hbh);
  202                 /* Destination options header(1st part) */
  203                 MAKE_EXTHDR(opt->ip6po_dest1, &exthdrs.ip6e_dest1);
  204                 /* Routing header */
  205                 MAKE_EXTHDR(opt->ip6po_rthdr, &exthdrs.ip6e_rthdr);
  206                 /* Destination options header(2nd part) */
  207                 MAKE_EXTHDR(opt->ip6po_dest2, &exthdrs.ip6e_dest2);
  208         }
  209 
  210 #if IPSEC
  211         if (ipsec_bypass != 0)
  212                 goto skip_ipsec;
  213 
  214         /* get a security policy for this packet */
  215         if (so == NULL)
  216                 sp = ipsec6_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND, 0, &error);
  217         else
  218                 sp = ipsec6_getpolicybysock(m, IPSEC_DIR_OUTBOUND, so, &error);
  219 
  220         if (sp == NULL) {
  221                 ipsec6stat.out_inval++;
  222                 goto freehdrs;
  223         }
  224 
  225         error = 0;
  226 
  227         /* check policy */
  228         switch (sp->policy) {
  229         case IPSEC_POLICY_DISCARD:
  230                 /*
  231                  * This packet is just discarded.
  232                  */
  233                 ipsec6stat.out_polvio++;
  234                 goto freehdrs;
  235 
  236         case IPSEC_POLICY_BYPASS:
  237         case IPSEC_POLICY_NONE:
  238                 /* no need to do IPsec. */
  239                 needipsec = 0;
  240                 break;
  241         
  242         case IPSEC_POLICY_IPSEC:
  243                 if (sp->req == NULL) {
  244                         /* acquire a policy */
  245                         error = key_spdacquire(sp);
  246                         goto freehdrs;
  247                 }
  248                 needipsec = 1;
  249                 break;
  250 
  251         case IPSEC_POLICY_ENTRUST:
  252         default:
  253                 printf("ip6_output: Invalid policy found. %d\n", sp->policy);
  254         }
  255         skip_ipsec:
  256 #endif /* IPSEC */
  257 
  258         /*
  259          * Calculate the total length of the extension header chain.
  260          * Keep the length of the unfragmentable part for fragmentation.
  261          */
  262         optlen = 0;
  263         if (exthdrs.ip6e_hbh) optlen += exthdrs.ip6e_hbh->m_len;
  264         if (exthdrs.ip6e_dest1) optlen += exthdrs.ip6e_dest1->m_len;
  265         if (exthdrs.ip6e_rthdr) optlen += exthdrs.ip6e_rthdr->m_len;
  266         unfragpartlen = optlen + sizeof(struct ip6_hdr);
  267         /* NOTE: we don't add AH/ESP length here. do that later. */
  268         if (exthdrs.ip6e_dest2) optlen += exthdrs.ip6e_dest2->m_len;
  269 
  270         /*
  271          * If we need IPsec, or there is at least one extension header,
  272          * separate IP6 header from the payload.
  273          */
  274         if ((needipsec || optlen) && !hdrsplit) {
  275                 if ((error = ip6_splithdr(m, &exthdrs)) != 0) {
  276                         m = NULL;
  277                         goto freehdrs;
  278                 }
  279                 m = exthdrs.ip6e_ip6;
  280                 hdrsplit++;
  281         }
  282 
  283         /* adjust pointer */
  284         ip6 = mtod(m, struct ip6_hdr *);
  285 
  286         /* adjust mbuf packet header length */
  287         m->m_pkthdr.len += optlen;
  288         plen = m->m_pkthdr.len - sizeof(*ip6);
  289 
  290         /* If this is a jumbo payload, insert a jumbo payload option. */
  291         if (plen > IPV6_MAXPACKET) {
  292                 if (!hdrsplit) {
  293                         if ((error = ip6_splithdr(m, &exthdrs)) != 0) {
  294                                 m = NULL;
  295                                 goto freehdrs;
  296                         }
  297                         m = exthdrs.ip6e_ip6;
  298                         hdrsplit++;
  299                 }
  300                 /* adjust pointer */
  301                 ip6 = mtod(m, struct ip6_hdr *);
  302                 if ((error = ip6_insert_jumboopt(&exthdrs, plen)) != 0)
  303                         goto freehdrs;
  304                 ip6->ip6_plen = 0;
  305         } else
  306                 ip6->ip6_plen = htons(plen);
  307 
  308         /*
  309          * Concatenate headers and fill in next header fields.
  310          * Here we have, on "m"
  311          *      IPv6 payload
  312          * and we insert headers accordingly.  Finally, we should be getting:
  313          *      IPv6 hbh dest1 rthdr ah* [esp* dest2 payload]
  314          *
  315          * during the header composing process, "m" points to IPv6 header.
  316          * "mprev" points to an extension header prior to esp.
  317          */
  318         {
  319                 u_char *nexthdrp = &ip6->ip6_nxt;
  320                 struct mbuf *mprev = m;
  321 
  322                 /*
  323                  * we treat dest2 specially.  this makes IPsec processing
  324                  * much easier.  the goal here is to make mprev point the
  325                  * mbuf prior to dest2.
  326                  *
  327                  * result: IPv6 dest2 payload
  328                  * m and mprev will point to IPv6 header.
  329                  */
  330                 if (exthdrs.ip6e_dest2) {
  331                         if (!hdrsplit)
  332                                 panic("assumption failed: hdr not split");
  333                         exthdrs.ip6e_dest2->m_next = m->m_next;
  334                         m->m_next = exthdrs.ip6e_dest2;
  335                         *mtod(exthdrs.ip6e_dest2, u_char *) = ip6->ip6_nxt;
  336                         ip6->ip6_nxt = IPPROTO_DSTOPTS;
  337                 }
  338 
  339 #define MAKE_CHAIN(m, mp, p, i)\
  340     do {\
  341         if (m) {\
  342                 if (!hdrsplit) \
  343                         panic("assumption failed: hdr not split"); \
  344                 *mtod((m), u_char *) = *(p);\
  345                 *(p) = (i);\
  346                 p = mtod((m), u_char *);\
  347                 (m)->m_next = (mp)->m_next;\
  348                 (mp)->m_next = (m);\
  349                 (mp) = (m);\
  350         }\
  351     } while (0)
  352                 /*
  353                  * result: IPv6 hbh dest1 rthdr dest2 payload
  354                  * m will point to IPv6 header.  mprev will point to the
  355                  * extension header prior to dest2 (rthdr in the above case).
  356                  */
  357                 MAKE_CHAIN(exthdrs.ip6e_hbh, mprev,
  358                            nexthdrp, IPPROTO_HOPOPTS);
  359                 MAKE_CHAIN(exthdrs.ip6e_dest1, mprev,
  360                            nexthdrp, IPPROTO_DSTOPTS);
  361                 MAKE_CHAIN(exthdrs.ip6e_rthdr, mprev,
  362                            nexthdrp, IPPROTO_ROUTING);
  363 
  364 #if IPSEC
  365                 if (!needipsec)
  366                         goto skip_ipsec2;
  367 
  368                 /*
  369                  * pointers after IPsec headers are not valid any more.
  370                  * other pointers need a great care too.
  371                  * (IPsec routines should not mangle mbufs prior to AH/ESP)
  372                  */
  373                 exthdrs.ip6e_dest2 = NULL;
  374 
  375             {
  376                 struct ip6_rthdr *rh = NULL;
  377                 int segleft_org = 0;
  378                 struct ipsec_output_state state;
  379 
  380                 if (exthdrs.ip6e_rthdr) {
  381                         rh = mtod(exthdrs.ip6e_rthdr, struct ip6_rthdr *);
  382                         segleft_org = rh->ip6r_segleft;
  383                         rh->ip6r_segleft = 0;
  384                 }
  385 
  386                 bzero(&state, sizeof(state));
  387                 state.m = m;
  388                 error = ipsec6_output_trans(&state, nexthdrp, mprev, sp, flags,
  389                         &needipsectun);
  390                 m = state.m;
  391                 if (error) {
  392                         /* mbuf is already reclaimed in ipsec6_output_trans. */
  393                         m = NULL;
  394                         switch (error) {
  395                         case EHOSTUNREACH:
  396                         case ENETUNREACH:
  397                         case EMSGSIZE:
  398                         case ENOBUFS:
  399                         case ENOMEM:
  400                                 break;
  401                         default:
  402                                 printf("ip6_output (ipsec): error code %d\n", error);
  403                                 /* fall through */
  404                         case ENOENT:
  405                                 /* don't show these error codes to the user */
  406                                 error = 0;
  407                                 break;
  408                         }
  409                         goto bad;
  410                 }
  411                 if (exthdrs.ip6e_rthdr) {
  412                         /* ah6_output doesn't modify mbuf chain */
  413                         rh->ip6r_segleft = segleft_org;
  414                 }
  415             }
  416 skip_ipsec2:;
  417 #endif
  418         }
  419 
  420         /*
  421          * If there is a routing header, replace destination address field
  422          * with the first hop of the routing header.
  423          */
  424         if (exthdrs.ip6e_rthdr) {
  425                 struct ip6_rthdr *rh =
  426                         (struct ip6_rthdr *)(mtod(exthdrs.ip6e_rthdr,
  427                                                   struct ip6_rthdr *));
  428                 struct ip6_rthdr0 *rh0;
  429 
  430                 finaldst = ip6->ip6_dst;
  431                 switch (rh->ip6r_type) {
  432                 case IPV6_RTHDR_TYPE_0:
  433                          rh0 = (struct ip6_rthdr0 *)rh;
  434                          ip6->ip6_dst = rh0->ip6r0_addr[0];
  435                          bcopy((caddr_t)&rh0->ip6r0_addr[1],
  436                                (caddr_t)&rh0->ip6r0_addr[0],
  437                                sizeof(struct in6_addr)*(rh0->ip6r0_segleft - 1)
  438                                  );
  439                          rh0->ip6r0_addr[rh0->ip6r0_segleft - 1] = finaldst;
  440                          break;
  441                 default:        /* is it possible? */
  442                          error = EINVAL;
  443                          goto bad;
  444                 }
  445         }
  446 
  447         /* Source address validation */
  448         if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src) &&
  449             (flags & IPV6_DADOUTPUT) == 0) {
  450                 error = EOPNOTSUPP;
  451                 ip6stat.ip6s_badscope++;
  452                 goto bad;
  453         }
  454         if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src)) {
  455                 error = EOPNOTSUPP;
  456                 ip6stat.ip6s_badscope++;
  457                 goto bad;
  458         }
  459 
  460         ip6stat.ip6s_localout++;
  461 
  462         /*
  463          * Route packet.
  464          */
  465         if (ro == 0) {
  466                 ro = &ip6route;
  467                 bzero((caddr_t)ro, sizeof(*ro));
  468         }
  469         ro_pmtu = ro;
  470         if (opt && opt->ip6po_rthdr)
  471                 ro = &opt->ip6po_route;
  472         dst = (struct sockaddr_in6 *)&ro->ro_dst;
  473         /*
  474          * If there is a cached route,
  475          * check that it is to the same destination
  476          * and is still up. If not, free it and try again.
  477          */
  478         if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 ||
  479                          dst->sin6_family != AF_INET6 ||
  480                          !IN6_ARE_ADDR_EQUAL(&dst->sin6_addr, &ip6->ip6_dst) ||
  481                          ro->ro_rt->generation_id != route_generation)) {
  482                 rtfree(ro->ro_rt);
  483                 ro->ro_rt = (struct rtentry *)0;
  484         }
  485         if (ro->ro_rt == 0) {
  486                 bzero(dst, sizeof(*dst));
  487                 dst->sin6_family = AF_INET6;
  488                 dst->sin6_len = sizeof(struct sockaddr_in6);
  489                 dst->sin6_addr = ip6->ip6_dst;
  490 #if SCOPEDROUTING
  491                 /* XXX: sin6_scope_id should already be fixed at this point */
  492                 if (IN6_IS_SCOPE_LINKLOCAL(&dst->sin6_addr))
  493                         dst->sin6_scope_id = ntohs(dst->sin6_addr.s6_addr16[1]);
  494 #endif
  495         }
  496 #if IPSEC
  497         if (needipsec && needipsectun) {
  498                 struct ipsec_output_state state;
  499 
  500                 /*
  501                  * All the extension headers will become inaccessible
  502                  * (since they can be encrypted).
  503                  * Don't panic, we need no more updates to extension headers
  504                  * on inner IPv6 packet (since they are now encapsulated).
  505                  *
  506                  * IPv6 [ESP|AH] IPv6 [extension headers] payload
  507                  */
  508                 bzero(&exthdrs, sizeof(exthdrs));
  509                 exthdrs.ip6e_ip6 = m;
  510 
  511                 bzero(&state, sizeof(state));
  512                 state.m = m;
  513                 state.ro = (struct route *)ro;
  514                 state.dst = (struct sockaddr *)dst;
  515 
  516                 error = ipsec6_output_tunnel(&state, sp, flags);
  517 
  518                 m = state.m;
  519                 ro = (struct route_in6 *)state.ro;
  520                 dst = (struct sockaddr_in6 *)state.dst;
  521                 if (error) {
  522                         /* mbuf is already reclaimed in ipsec6_output_tunnel. */
  523                         m0 = m = NULL;
  524                         m = NULL;
  525                         switch (error) {
  526                         case EHOSTUNREACH:
  527                         case ENETUNREACH:
  528                         case EMSGSIZE:
  529                         case ENOBUFS:
  530                         case ENOMEM:
  531                                 break;
  532                         default:
  533                                 printf("ip6_output (ipsec): error code %d\n", error);
  534                                 /* fall through */
  535                         case ENOENT:
  536                                 /* don't show these error codes to the user */
  537                                 error = 0;
  538                                 break;
  539                         }
  540                         goto bad;
  541                 }
  542 
  543                 exthdrs.ip6e_ip6 = m;
  544         }
  545 #endif /* IPSEC */
  546 
  547         if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
  548                 /* Unicast */
  549 
  550 #define ifatoia6(ifa)   ((struct in6_ifaddr *)(ifa))
  551 #define sin6tosa(sin6)  ((struct sockaddr *)(sin6))
  552                 /* xxx
  553                  * interface selection comes here
  554                  * if an interface is specified from an upper layer,
  555                  * ifp must point it.
  556                  */
  557                 if (ro->ro_rt == 0) {
  558                         /*
  559                          * non-bsdi always clone routes, if parent is
  560                          * PRF_CLONING.
  561                          */
  562                         rtalloc((struct route *)ro);
  563                 }
  564                 if (ro->ro_rt == 0) {
  565                         ip6stat.ip6s_noroute++;
  566                         error = EHOSTUNREACH;
  567                         /* XXX in6_ifstat_inc(ifp, ifs6_out_discard); */
  568                         goto bad;
  569                 }
  570                 ia = ifatoia6(ro->ro_rt->rt_ifa);
  571                 ifp = ro->ro_rt->rt_ifp;
  572                 ro->ro_rt->rt_use++;
  573                 if (ro->ro_rt->rt_flags & RTF_GATEWAY)
  574                         dst = (struct sockaddr_in6 *)ro->ro_rt->rt_gateway;
  575                 m->m_flags &= ~(M_BCAST | M_MCAST);     /* just in case */
  576 
  577                 in6_ifstat_inc(ifp, ifs6_out_request);
  578 
  579                 /*
  580                  * Check if the outgoing interface conflicts with
  581                  * the interface specified by ifi6_ifindex (if specified).
  582                  * Note that loopback interface is always okay.
  583                  * (this may happen when we are sending a packet to one of
  584                  *  our own addresses.)
  585                  */
  586                 if (opt && opt->ip6po_pktinfo
  587                  && opt->ip6po_pktinfo->ipi6_ifindex) {
  588                         if (!(ifp->if_flags & IFF_LOOPBACK)
  589                          && ifp->if_index != opt->ip6po_pktinfo->ipi6_ifindex) {
  590                                 ip6stat.ip6s_noroute++;
  591                                 in6_ifstat_inc(ifp, ifs6_out_discard);
  592                                 error = EHOSTUNREACH;
  593                                 goto bad;
  594                         }
  595                 }
  596 
  597                 if (opt && opt->ip6po_hlim != -1)
  598                         ip6->ip6_hlim = opt->ip6po_hlim & 0xff;
  599         } else {
  600                 /* Multicast */
  601                 struct  in6_multi *in6m;
  602 
  603                 m->m_flags = (m->m_flags & ~M_BCAST) | M_MCAST;
  604 
  605                 /*
  606                  * See if the caller provided any multicast options
  607                  */
  608                 ifp = NULL;
  609                 if (im6o != NULL) {
  610                         ip6->ip6_hlim = im6o->im6o_multicast_hlim;
  611                         if (im6o->im6o_multicast_ifp != NULL)
  612                                 ifp = im6o->im6o_multicast_ifp;
  613                 } else
  614                         ip6->ip6_hlim = ip6_defmcasthlim;
  615 
  616                 /*
  617                  * See if the caller provided the outgoing interface
  618                  * as an ancillary data.
  619                  * Boundary check for ifindex is assumed to be already done.
  620                  */
  621                 if (opt && opt->ip6po_pktinfo && opt->ip6po_pktinfo->ipi6_ifindex)
  622                         ifp = ifindex2ifnet[opt->ip6po_pktinfo->ipi6_ifindex];
  623 
  624                 /*
  625                  * If the destination is a node-local scope multicast,
  626                  * the packet should be loop-backed only.
  627                  */
  628                 if (IN6_IS_ADDR_MC_NODELOCAL(&ip6->ip6_dst)) {
  629                         /*
  630                          * If the outgoing interface is already specified,
  631                          * it should be a loopback interface.
  632                          */
  633                         if (ifp && (ifp->if_flags & IFF_LOOPBACK) == 0) {
  634                                 ip6stat.ip6s_badscope++;
  635                                 error = ENETUNREACH; /* XXX: better error? */
  636                                 /* XXX correct ifp? */
  637                                 in6_ifstat_inc(ifp, ifs6_out_discard);
  638                                 goto bad;
  639                         } else {
  640                                 ifp = &loif[0];
  641                         }
  642                 }
  643 
  644                 if (opt && opt->ip6po_hlim != -1)
  645                         ip6->ip6_hlim = opt->ip6po_hlim & 0xff;
  646 
  647                 /*
  648                  * If caller did not provide an interface lookup a
  649                  * default in the routing table.  This is either a
  650                  * default for the speicfied group (i.e. a host
  651                  * route), or a multicast default (a route for the
  652                  * ``net'' ff00::/8).
  653                  */
  654                 if (ifp == NULL) {
  655                         if (ro->ro_rt == 0) {
  656                                 ro->ro_rt = rtalloc1((struct sockaddr *)
  657                                                 &ro->ro_dst, 0, 0UL);
  658                         }
  659                         if (ro->ro_rt == 0) {
  660                                 ip6stat.ip6s_noroute++;
  661                                 error = EHOSTUNREACH;
  662                                 /* XXX in6_ifstat_inc(ifp, ifs6_out_discard) */
  663                                 goto bad;
  664                         }
  665                         ia = ifatoia6(ro->ro_rt->rt_ifa);
  666                         ifp = ro->ro_rt->rt_ifp;
  667                         ro->ro_rt->rt_use++;
  668                 }
  669 
  670                 if ((flags & IPV6_FORWARDING) == 0)
  671                         in6_ifstat_inc(ifp, ifs6_out_request);
  672                 in6_ifstat_inc(ifp, ifs6_out_mcast);
  673 
  674                 /*
  675                  * Confirm that the outgoing interface supports multicast.
  676                  */
  677                 if ((ifp->if_flags & IFF_MULTICAST) == 0) {
  678                         ip6stat.ip6s_noroute++;
  679                         in6_ifstat_inc(ifp, ifs6_out_discard);
  680                         error = ENETUNREACH;
  681                         goto bad;
  682                 }
  683                 IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m);
  684                 if (in6m != NULL &&
  685                    (im6o == NULL || im6o->im6o_multicast_loop)) {
  686                         /*
  687                          * If we belong to the destination multicast group
  688                          * on the outgoing interface, and the caller did not
  689                          * forbid loopback, loop back a copy.
  690                          */
  691                         ip6_mloopback(ifp, m, dst);
  692                 } else {
  693                         /*
  694                          * If we are acting as a multicast router, perform
  695                          * multicast forwarding as if the packet had just
  696                          * arrived on the interface to which we are about
  697                          * to send.  The multicast forwarding function
  698                          * recursively calls this function, using the
  699                          * IPV6_FORWARDING flag to prevent infinite recursion.
  700                          *
  701                          * Multicasts that are looped back by ip6_mloopback(),
  702                          * above, will be forwarded by the ip6_input() routine,
  703                          * if necessary.
  704                          */
  705                         if (ip6_mrouter && (flags & IPV6_FORWARDING) == 0) {
  706                                 if (ip6_mforward(ip6, ifp, m) != NULL) {
  707                                         m_freem(m);
  708                                         goto done;
  709                                 }
  710                         }
  711                 }
  712                 /*
  713                  * Multicasts with a hoplimit of zero may be looped back,
  714                  * above, but must not be transmitted on a network.
  715                  * Also, multicasts addressed to the loopback interface
  716                  * are not sent -- the above call to ip6_mloopback() will
  717                  * loop back a copy if this host actually belongs to the
  718                  * destination group on the loopback interface.
  719                  */
  720                 if (ip6->ip6_hlim == 0 || (ifp->if_flags & IFF_LOOPBACK)) {
  721                         m_freem(m);
  722                         goto done;
  723                 }
  724         }
  725 
  726         /*
  727          * Fill the outgoing inteface to tell the upper layer
  728          * to increment per-interface statistics.
  729          */
  730         if (ifpp)
  731                 *ifpp = ifp;
  732 
  733         /*
  734          * Determine path MTU.
  735          */
  736         if (ro_pmtu != ro) {
  737                 /* The first hop and the final destination may differ. */
  738                 struct sockaddr_in6 *sin6_fin =
  739                         (struct sockaddr_in6 *)&ro_pmtu->ro_dst;
  740                 if (ro_pmtu->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 ||
  741                                        !IN6_ARE_ADDR_EQUAL(&sin6_fin->sin6_addr,
  742                                                            &finaldst))) {
  743                         rtfree(ro_pmtu->ro_rt);
  744                         ro_pmtu->ro_rt = (struct rtentry *)0;
  745                 }
  746                 if (ro_pmtu->ro_rt == 0) {
  747                         bzero(sin6_fin, sizeof(*sin6_fin));
  748                         sin6_fin->sin6_family = AF_INET6;
  749                         sin6_fin->sin6_len = sizeof(struct sockaddr_in6);
  750                         sin6_fin->sin6_addr = finaldst;
  751 
  752                         rtalloc((struct route *)ro_pmtu);
  753                 }
  754         }
  755         if (ro_pmtu->ro_rt != NULL) {
  756                 u_int32_t ifmtu = nd_ifinfo[ifp->if_index].linkmtu;
  757 
  758                 mtu = ro_pmtu->ro_rt->rt_rmx.rmx_mtu;
  759                 if (mtu > ifmtu || mtu == 0) {
  760                         /*
  761                          * The MTU on the route is larger than the MTU on
  762                          * the interface!  This shouldn't happen, unless the
  763                          * MTU of the interface has been changed after the
  764                          * interface was brought up.  Change the MTU in the
  765                          * route to match the interface MTU (as long as the
  766                          * field isn't locked).
  767                          *
  768                          * if MTU on the route is 0, we need to fix the MTU.
  769                          * this case happens with path MTU discovery timeouts.
  770                          */
  771                          mtu = ifmtu;
  772                          if ((ro_pmtu->ro_rt->rt_rmx.rmx_locks & RTV_MTU) == 0)
  773                                  ro_pmtu->ro_rt->rt_rmx.rmx_mtu = mtu; /* XXX */
  774                 }
  775         } else {
  776                 mtu = nd_ifinfo[ifp->if_index].linkmtu;
  777         }
  778 
  779         /*
  780          * advanced API (IPV6_USE_MIN_MTU) overrides mtu setting
  781          */
  782         if ((flags & IPV6_MINMTU) != 0 && mtu > IPV6_MMTU)
  783                 mtu = IPV6_MMTU;
  784 
  785         /* Fake scoped addresses */
  786         if ((ifp->if_flags & IFF_LOOPBACK) != 0) {
  787                 /*
  788                  * If source or destination address is a scoped address, and
  789                  * the packet is going to be sent to a loopback interface,
  790                  * we should keep the original interface.
  791                  */
  792 
  793                 /*
  794                  * XXX: this is a very experimental and temporary solution.
  795                  * We eventually have sockaddr_in6 and use the sin6_scope_id
  796                  * field of the structure here.
  797                  * We rely on the consistency between two scope zone ids
  798                  * of source and destination, which should already be assured.
  799                  * Larger scopes than link will be supported in the future. 
  800                  */
  801                 origifp = NULL;
  802                 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
  803                         origifp = ifindex2ifnet[ntohs(ip6->ip6_src.s6_addr16[1])];
  804                 else if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst))
  805                         origifp = ifindex2ifnet[ntohs(ip6->ip6_dst.s6_addr16[1])];
  806                 /*
  807                  * XXX: origifp can be NULL even in those two cases above.
  808                  * For example, if we remove the (only) link-local address
  809                  * from the loopback interface, and try to send a link-local
  810                  * address without link-id information.  Then the source
  811                  * address is ::1, and the destination address is the
  812                  * link-local address with its s6_addr16[1] being zero.
  813                  * What is worse, if the packet goes to the loopback interface
  814                  * by a default rejected route, the null pointer would be
  815                  * passed to looutput, and the kernel would hang.
  816                  * The following last resort would prevent such disaster.
  817                  */
  818                 if (origifp == NULL)
  819                         origifp = ifp;
  820         }
  821         else
  822                 origifp = ifp;
  823 #ifndef SCOPEDROUTING
  824         /*
  825          * clear embedded scope identifiers if necessary.
  826          * in6_clearscope will touch the addresses only when necessary.
  827          */
  828         in6_clearscope(&ip6->ip6_src);
  829         in6_clearscope(&ip6->ip6_dst);
  830 #endif
  831 
  832         /*
  833          * Check with the firewall...
  834          */
  835         if (ip6_fw_enable && ip6_fw_chk_ptr) {
  836                 u_short port = 0;
  837                 m->m_pkthdr.rcvif = NULL;       /* XXX */
  838                 /* If ipfw says divert, we have to just drop packet */
  839                 if ((*ip6_fw_chk_ptr)(&ip6, ifp, &port, &m)) {
  840                         m_freem(m);
  841                         goto done;
  842                 }
  843                 if (!m) {
  844                         error = EACCES;
  845                         goto done;
  846                 }
  847         }
  848 
  849         /*
  850          * If the outgoing packet contains a hop-by-hop options header,
  851          * it must be examined and processed even by the source node.
  852          * (RFC 2460, section 4.)
  853          */
  854         if (exthdrs.ip6e_hbh) {
  855                 struct ip6_hbh *hbh = mtod(exthdrs.ip6e_hbh, struct ip6_hbh *);
  856                 u_int32_t dummy1; /* XXX unused */
  857                 u_int32_t dummy2; /* XXX unused */
  858 
  859 #if DIAGNOSTIC
  860                 if ((hbh->ip6h_len + 1) << 3 > exthdrs.ip6e_hbh->m_len)
  861                         panic("ip6e_hbh is not continuous");
  862 #endif
  863                 /*
  864                  *  XXX: if we have to send an ICMPv6 error to the sender,
  865                  *       we need the M_LOOP flag since icmp6_error() expects
  866                  *       the IPv6 and the hop-by-hop options header are
  867                  *       continuous unless the flag is set.
  868                  */
  869                 m->m_flags |= M_LOOP;
  870                 m->m_pkthdr.rcvif = ifp;
  871                 if (ip6_process_hopopts(m,
  872                                         (u_int8_t *)(hbh + 1),
  873                                         ((hbh->ip6h_len + 1) << 3) -
  874                                         sizeof(struct ip6_hbh),
  875                                         &dummy1, &dummy2) < 0) {
  876                         /* m was already freed at this point */
  877                         error = EINVAL;/* better error? */
  878                         goto done;
  879                 }
  880                 m->m_flags &= ~M_LOOP; /* XXX */
  881                 m->m_pkthdr.rcvif = NULL;
  882         }
  883 
  884         /*
  885          * Send the packet to the outgoing interface.
  886          * If necessary, do IPv6 fragmentation before sending.
  887          */
  888         tlen = m->m_pkthdr.len;
  889         if (tlen <= mtu
  890 #if notyet
  891             /*
  892              * On any link that cannot convey a 1280-octet packet in one piece,
  893              * link-specific fragmentation and reassembly must be provided at
  894              * a layer below IPv6. [RFC 2460, sec.5]
  895              * Thus if the interface has ability of link-level fragmentation,
  896              * we can just send the packet even if the packet size is
  897              * larger than the link's MTU.
  898              * XXX: IFF_FRAGMENTABLE (or such) flag has not been defined yet...
  899              */
  900         
  901             || ifp->if_flags & IFF_FRAGMENTABLE
  902 #endif
  903             )
  904         {
  905                 /* Record statistics for this interface address. */
  906                 if (ia && !(flags & IPV6_FORWARDING)) {
  907 #ifndef __APPLE__
  908                         ia->ia_ifa.if_opackets++;
  909                         ia->ia_ifa.if_obytes += m->m_pkthdr.len;
  910 #endif
  911                 }
  912 #ifdef IPSEC
  913                 /* clean ipsec history once it goes out of the node */
  914                 ipsec_delaux(m);
  915 #endif
  916 
  917                 error = nd6_output(ifp, origifp, m, dst, ro->ro_rt);
  918                 goto done;
  919         } else if (mtu < IPV6_MMTU) {
  920                 /*
  921                  * note that path MTU is never less than IPV6_MMTU
  922                  * (see icmp6_input).
  923                  */
  924                 error = EMSGSIZE;
  925                 in6_ifstat_inc(ifp, ifs6_out_fragfail);
  926                 goto bad;
  927         } else if (ip6->ip6_plen == 0) { /* jumbo payload cannot be fragmented */
  928                 error = EMSGSIZE;
  929                 in6_ifstat_inc(ifp, ifs6_out_fragfail);
  930                 goto bad;
  931         } else {
  932                 struct mbuf **mnext, *m_frgpart;
  933                 struct ip6_frag *ip6f;
  934                 u_int32_t id = htonl(ip6_id++);
  935                 u_char nextproto;
  936 
  937                 /*
  938                  * Too large for the destination or interface;
  939                  * fragment if possible.
  940                  * Must be able to put at least 8 bytes per fragment.
  941                  */
  942                 hlen = unfragpartlen;
  943                 if (mtu > IPV6_MAXPACKET)
  944                         mtu = IPV6_MAXPACKET;
  945 
  946                 len = (mtu - hlen - sizeof(struct ip6_frag)) & ~7;
  947                 if (len < 8) {
  948                         error = EMSGSIZE;
  949                         in6_ifstat_inc(ifp, ifs6_out_fragfail);
  950                         goto bad;
  951                 }
  952 
  953                 mnext = &m->m_nextpkt;
  954 
  955                 /*
  956                  * Change the next header field of the last header in the
  957                  * unfragmentable part.
  958                  */
  959                 if (exthdrs.ip6e_rthdr) {
  960                         nextproto = *mtod(exthdrs.ip6e_rthdr, u_char *);
  961                         *mtod(exthdrs.ip6e_rthdr, u_char *) = IPPROTO_FRAGMENT;
  962                 } else if (exthdrs.ip6e_dest1) {
  963                         nextproto = *mtod(exthdrs.ip6e_dest1, u_char *);
  964                         *mtod(exthdrs.ip6e_dest1, u_char *) = IPPROTO_FRAGMENT;
  965                 } else if (exthdrs.ip6e_hbh) {
  966                         nextproto = *mtod(exthdrs.ip6e_hbh, u_char *);
  967                         *mtod(exthdrs.ip6e_hbh, u_char *) = IPPROTO_FRAGMENT;
  968                 } else {
  969                         nextproto = ip6->ip6_nxt;
  970                         ip6->ip6_nxt = IPPROTO_FRAGMENT;
  971                 }
  972 
  973                 /*
  974                  * Loop through length of segment after first fragment,
  975                  * make new header and copy data of each part and link onto
  976                  * chain.
  977                  */
  978                 m0 = m;
  979                 for (off = hlen; off < tlen; off += len) {
  980                         MGETHDR(m, M_DONTWAIT, MT_HEADER);
  981                         if (!m) {
  982                                 error = ENOBUFS;
  983                                 ip6stat.ip6s_odropped++;
  984                                 goto sendorfree;
  985                         }
  986                         m->m_pkthdr.rcvif = NULL;
  987                         m->m_flags = m0->m_flags & M_COPYFLAGS;
  988                         *mnext = m;
  989                         mnext = &m->m_nextpkt;
  990                         m->m_data += max_linkhdr;
  991                         mhip6 = mtod(m, struct ip6_hdr *);
  992                         *mhip6 = *ip6;
  993                         m->m_len = sizeof(*mhip6);
  994                         error = ip6_insertfraghdr(m0, m, hlen, &ip6f);
  995                         if (error) {
  996                                 ip6stat.ip6s_odropped++;
  997                                 goto sendorfree;
  998                         }
  999                         ip6f->ip6f_offlg = htons((u_short)((off - hlen) & ~7));
 1000                         if (off + len >= tlen)
 1001                                 len = tlen - off;
 1002                         else
 1003                                 ip6f->ip6f_offlg |= IP6F_MORE_FRAG;
 1004                         mhip6->ip6_plen = htons((u_short)(len + hlen +
 1005                                                           sizeof(*ip6f) -
 1006                                                           sizeof(struct ip6_hdr)));
 1007                         if ((m_frgpart = m_copy(m0, off, len)) == 0) {
 1008                                 error = ENOBUFS;
 1009                                 ip6stat.ip6s_odropped++;
 1010                                 goto sendorfree;
 1011                         }
 1012                         m_cat(m, m_frgpart);
 1013                         m->m_pkthdr.len = len + hlen + sizeof(*ip6f);
 1014                         m->m_pkthdr.rcvif = (struct ifnet *)0;
 1015                         ip6f->ip6f_reserved = 0;
 1016                         ip6f->ip6f_ident = id;
 1017                         ip6f->ip6f_nxt = nextproto;
 1018                         ip6stat.ip6s_ofragments++;
 1019                         in6_ifstat_inc(ifp, ifs6_out_fragcreat);
 1020                 }
 1021 
 1022                 in6_ifstat_inc(ifp, ifs6_out_fragok);
 1023         }
 1024 
 1025         /*
 1026          * Remove leading garbages.
 1027          */
 1028 sendorfree:
 1029         m = m0->m_nextpkt;
 1030         m0->m_nextpkt = 0;
 1031         m_freem(m0);
 1032         for (m0 = m; m; m = m0) {
 1033                 m0 = m->m_nextpkt;
 1034                 m->m_nextpkt = 0;
 1035                 if (error == 0) {
 1036                         /* Record statistics for this interface address. */
 1037                         if (ia) {
 1038 #ifndef __APPLE__
 1039                                 ia->ia_ifa.if_opackets++;
 1040                                 ia->ia_ifa.if_obytes += m->m_pkthdr.len;
 1041 #endif
 1042                         }
 1043 #if IPSEC
 1044                         /* clean ipsec history once it goes out of the node */
 1045                         ipsec_delaux(m);
 1046 #endif
 1047                         error = nd6_output(ifp, origifp, m, dst, ro->ro_rt);
 1048                 } else
 1049                         m_freem(m);
 1050         }
 1051 
 1052         if (error == 0)
 1053                 ip6stat.ip6s_fragmented++;
 1054 
 1055 done:
 1056         if (ro == &ip6route && ro->ro_rt) { /* brace necessary for rtfree */
 1057                 rtfree(ro->ro_rt);
 1058         } else if (ro_pmtu == &ip6route && ro_pmtu->ro_rt) {
 1059                 rtfree(ro_pmtu->ro_rt);
 1060         }
 1061 
 1062 #if IPSEC
 1063         if (sp != NULL)
 1064                 key_freesp(sp);
 1065 #endif /* IPSEC */
 1066 
 1067         return(error);
 1068 
 1069 freehdrs:
 1070         m_freem(exthdrs.ip6e_hbh);      /* m_freem will check if mbuf is 0 */
 1071         m_freem(exthdrs.ip6e_dest1);
 1072         m_freem(exthdrs.ip6e_rthdr);
 1073         m_freem(exthdrs.ip6e_dest2);
 1074         /* fall through */
 1075 bad:
 1076         m_freem(m);
 1077         goto done;
 1078 }
 1079 
 1080 static int
 1081 ip6_copyexthdr(mp, hdr, hlen)
 1082         struct mbuf **mp;
 1083         caddr_t hdr;
 1084         int hlen;
 1085 {
 1086         struct mbuf *m;
 1087 
 1088         if (hlen > MCLBYTES)
 1089                 return(ENOBUFS); /* XXX */
 1090 
 1091         MGET(m, M_DONTWAIT, MT_DATA);
 1092         if (!m)
 1093                 return(ENOBUFS);
 1094 
 1095         if (hlen > MLEN) {
 1096                 MCLGET(m, M_DONTWAIT);
 1097                 if ((m->m_flags & M_EXT) == 0) {
 1098                         m_free(m);
 1099                         return(ENOBUFS);
 1100                 }
 1101         }
 1102         m->m_len = hlen;
 1103         if (hdr)
 1104                 bcopy(hdr, mtod(m, caddr_t), hlen);
 1105 
 1106         *mp = m;
 1107         return(0);
 1108 }
 1109 
 1110 /*
 1111  * Insert jumbo payload option.
 1112  */
 1113 static int
 1114 ip6_insert_jumboopt(exthdrs, plen)
 1115         struct ip6_exthdrs *exthdrs;
 1116         u_int32_t plen;
 1117 {
 1118         struct mbuf *mopt;
 1119         u_char *optbuf;
 1120         u_int32_t v;
 1121 
 1122 #define JUMBOOPTLEN     8       /* length of jumbo payload option and padding */
 1123 
 1124         /*
 1125          * If there is no hop-by-hop options header, allocate new one.
 1126          * If there is one but it doesn't have enough space to store the
 1127          * jumbo payload option, allocate a cluster to store the whole options.
 1128          * Otherwise, use it to store the options.
 1129          */
 1130         if (exthdrs->ip6e_hbh == 0) {
 1131                 MGET(mopt, M_DONTWAIT, MT_DATA);
 1132                 if (mopt == 0)
 1133                         return(ENOBUFS);
 1134                 mopt->m_len = JUMBOOPTLEN;
 1135                 optbuf = mtod(mopt, u_char *);
 1136                 optbuf[1] = 0;  /* = ((JUMBOOPTLEN) >> 3) - 1 */
 1137                 exthdrs->ip6e_hbh = mopt;
 1138         } else {
 1139                 struct ip6_hbh *hbh;
 1140 
 1141                 mopt = exthdrs->ip6e_hbh;
 1142                 if (M_TRAILINGSPACE(mopt) < JUMBOOPTLEN) {
 1143                         /*
 1144                          * XXX assumption:
 1145                          * - exthdrs->ip6e_hbh is not referenced from places
 1146                          *   other than exthdrs.
 1147                          * - exthdrs->ip6e_hbh is not an mbuf chain.
 1148                          */
 1149                         int oldoptlen = mopt->m_len;
 1150                         struct mbuf *n;
 1151 
 1152                         /*
 1153                          * XXX: give up if the whole (new) hbh header does
 1154                          * not fit even in an mbuf cluster.
 1155                          */
 1156                         if (oldoptlen + JUMBOOPTLEN > MCLBYTES)
 1157                                 return(ENOBUFS);
 1158 
 1159                         /*
 1160                          * As a consequence, we must always prepare a cluster
 1161                          * at this point.
 1162                          */
 1163                         MGET(n, M_DONTWAIT, MT_DATA);
 1164                         if (n) {
 1165                                 MCLGET(n, M_DONTWAIT);
 1166                                 if ((n->m_flags & M_EXT) == 0) {
 1167                                         m_freem(n);
 1168                                         n = NULL;
 1169                                 }
 1170                         }
 1171                         if (!n)
 1172                                 return(ENOBUFS);
 1173                         n->m_len = oldoptlen + JUMBOOPTLEN;
 1174                         bcopy(mtod(mopt, caddr_t), mtod(n, caddr_t),
 1175                               oldoptlen);
 1176                         optbuf = mtod(n, caddr_t) + oldoptlen;
 1177                         m_freem(mopt);
 1178                         mopt = exthdrs->ip6e_hbh = n;
 1179                 } else {
 1180                         optbuf = mtod(mopt, u_char *) + mopt->m_len;
 1181                         mopt->m_len += JUMBOOPTLEN;
 1182                 }
 1183                 optbuf[0] = IP6OPT_PADN;
 1184                 optbuf[1] = 1;
 1185 
 1186                 /*
 1187                  * Adjust the header length according to the pad and
 1188                  * the jumbo payload option.
 1189                  */
 1190                 hbh = mtod(mopt, struct ip6_hbh *);
 1191                 hbh->ip6h_len += (JUMBOOPTLEN >> 3);
 1192         }
 1193 
 1194         /* fill in the option. */
 1195         optbuf[2] = IP6OPT_JUMBO;
 1196         optbuf[3] = 4;
 1197         v = (u_int32_t)htonl(plen + JUMBOOPTLEN);
 1198         bcopy(&v, &optbuf[4], sizeof(u_int32_t));
 1199 
 1200         /* finally, adjust the packet header length */
 1201         exthdrs->ip6e_ip6->m_pkthdr.len += JUMBOOPTLEN;
 1202 
 1203         return(0);
 1204 #undef JUMBOOPTLEN
 1205 }
 1206 
 1207 /*
 1208  * Insert fragment header and copy unfragmentable header portions.
 1209  */
 1210 static int
 1211 ip6_insertfraghdr(m0, m, hlen, frghdrp)
 1212         struct mbuf *m0, *m;
 1213         int hlen;
 1214         struct ip6_frag **frghdrp;
 1215 {
 1216         struct mbuf *n, *mlast;
 1217 
 1218         if (hlen > sizeof(struct ip6_hdr)) {
 1219                 n = m_copym(m0, sizeof(struct ip6_hdr),
 1220                             hlen - sizeof(struct ip6_hdr), M_DONTWAIT);
 1221                 if (n == 0)
 1222                         return(ENOBUFS);
 1223                 m->m_next = n;
 1224         } else
 1225                 n = m;
 1226 
 1227         /* Search for the last mbuf of unfragmentable part. */
 1228         for (mlast = n; mlast->m_next; mlast = mlast->m_next)
 1229                 ;
 1230 
 1231         if ((mlast->m_flags & M_EXT) == 0 &&
 1232             M_TRAILINGSPACE(mlast) >= sizeof(struct ip6_frag)) {
 1233                 /* use the trailing space of the last mbuf for the fragment hdr */
 1234                 *frghdrp =
 1235                         (struct ip6_frag *)(mtod(mlast, caddr_t) + mlast->m_len);
 1236                 mlast->m_len += sizeof(struct ip6_frag);
 1237                 m->m_pkthdr.len += sizeof(struct ip6_frag);
 1238         } else {
 1239                 /* allocate a new mbuf for the fragment header */
 1240                 struct mbuf *mfrg;
 1241 
 1242                 MGET(mfrg, M_DONTWAIT, MT_DATA);
 1243                 if (mfrg == 0)
 1244                         return(ENOBUFS);
 1245                 mfrg->m_len = sizeof(struct ip6_frag);
 1246                 *frghdrp = mtod(mfrg, struct ip6_frag *);
 1247                 mlast->m_next = mfrg;
 1248         }
 1249 
 1250         return(0);
 1251 }
 1252 
 1253 extern int load_ipfw();
 1254 
 1255 /*
 1256  * IP6 socket option processing.
 1257  */
 1258 int
 1259 ip6_ctloutput(so, sopt)
 1260         struct socket *so;
 1261         struct sockopt *sopt;
 1262 {
 1263         int privileged;
 1264         struct inpcb *in6p = sotoinpcb(so);
 1265         int error, optval;
 1266         int level, op, optname;
 1267         int optlen;
 1268         struct proc *p;
 1269 
 1270         if (sopt == NULL) 
 1271                 panic("ip6_ctloutput: arg soopt is NULL");
 1272         else {
 1273                 level = sopt->sopt_level;
 1274                 op = sopt->sopt_dir;
 1275                 optname = sopt->sopt_name;
 1276                 optlen = sopt->sopt_valsize;
 1277                 p = sopt->sopt_p;
 1278         }
 1279         error = optval = 0;
 1280 
 1281         privileged = (p == 0 || suser(p->p_ucred, &p->p_acflag)) ? 0 : 1;
 1282 
 1283         if (level == IPPROTO_IPV6) {
 1284                 switch (op) {
 1285 
 1286                 case SOPT_SET:
 1287                         switch (optname) {
 1288                         case IPV6_PKTOPTIONS:
 1289                         {
 1290                                 struct mbuf *m;
 1291 
 1292                                 error = soopt_getm(sopt, &m); /* XXX */
 1293                                 if (error != NULL)
 1294                                         break;
 1295                                 error = soopt_mcopyin(sopt, m); /* XXX */
 1296                                 if (error != NULL)
 1297                                         break;
 1298                                 error = ip6_pcbopts(&in6p->in6p_outputopts,
 1299                                                     m, so, sopt);
 1300                                 m_freem(m); /* XXX */
 1301                                 break;
 1302                         }
 1303 
 1304                         /*
 1305                          * Use of some Hop-by-Hop options or some
 1306                          * Destination options, might require special
 1307                          * privilege.  That is, normal applications
 1308                          * (without special privilege) might be forbidden
 1309                          * from setting certain options in outgoing packets,
 1310                          * and might never see certain options in received
 1311                          * packets. [RFC 2292 Section 6]
 1312                          * KAME specific note:
 1313                          *  KAME prevents non-privileged users from sending or
 1314                          *  receiving ANY hbh/dst options in order to avoid
 1315                          *  overhead of parsing options in the kernel.
 1316                          */
 1317                         case IPV6_UNICAST_HOPS:
 1318                         case IPV6_CHECKSUM:
 1319                         case IPV6_FAITH:
 1320 
 1321                         case IPV6_V6ONLY:
 1322                                 if (optlen != sizeof(int)) {
 1323                                         error = EINVAL;
 1324                                         break;
 1325                                 }
 1326                                 error = sooptcopyin(sopt, &optval,
 1327                                         sizeof optval, sizeof optval);
 1328                                 if (error)
 1329                                         break;
 1330                                 switch (optname) {
 1331 
 1332                                 case IPV6_UNICAST_HOPS:
 1333                                         if (optval < -1 || optval >= 256)
 1334                                                 error = EINVAL;
 1335                                         else {
 1336                                                 /* -1 = kernel default */
 1337                                                 in6p->in6p_hops = optval;
 1338 
 1339                                                 if ((in6p->in6p_vflag &
 1340                                                      INP_IPV4) != 0)
 1341                                                         in6p->inp_ip_ttl = optval;
 1342                                         }
 1343                                         break;
 1344 #define OPTSET(bit) \
 1345 do { \
 1346         if (optval) \
 1347                 in6p->in6p_flags |= (bit); \
 1348         else \
 1349                 in6p->in6p_flags &= ~(bit); \
 1350 } while (0)
 1351 #define OPTBIT(bit) (in6p->in6p_flags & (bit) ? 1 : 0)
 1352 
 1353                                 case IPV6_CHECKSUM:
 1354                                         in6p->in6p_cksum = optval;
 1355                                         break;
 1356 
 1357                                 case IPV6_FAITH:
 1358                                         OPTSET(IN6P_FAITH);
 1359                                         break;
 1360 
 1361                                 case IPV6_V6ONLY:
 1362                                         /*
 1363                                          * make setsockopt(IPV6_V6ONLY)
 1364                                          * available only prior to bind(2).
 1365                                          * see ipng mailing list, Jun 22 2001.
 1366                                          */
 1367                                         if (in6p->in6p_lport ||
 1368                                             !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr))
 1369                                         {
 1370                                                 error = EINVAL;
 1371                                                 break;
 1372                                         }
 1373                                         OPTSET(IN6P_IPV6_V6ONLY);
 1374                                         if (optval)
 1375                                                 in6p->in6p_vflag &= ~INP_IPV4;
 1376                                         else
 1377                                                 in6p->in6p_vflag |= INP_IPV4;
 1378                                         break;
 1379                                 }
 1380                                 break;
 1381 
 1382                         case IPV6_PKTINFO:
 1383                         case IPV6_HOPLIMIT:
 1384                         case IPV6_HOPOPTS:
 1385                         case IPV6_DSTOPTS:
 1386                         case IPV6_RTHDR:
 1387                                 /* RFC 2292 */
 1388                                 if (optlen != sizeof(int)) {
 1389                                         error = EINVAL;
 1390                                         break;
 1391                                 }
 1392                                 error = sooptcopyin(sopt, &optval,
 1393                                         sizeof optval, sizeof optval);
 1394                                 if (error)
 1395                                         break;
 1396                                 switch (optname) {
 1397                                 case IPV6_PKTINFO:
 1398                                         OPTSET(IN6P_PKTINFO);
 1399                                         break;
 1400                                 case IPV6_HOPLIMIT:
 1401                                         OPTSET(IN6P_HOPLIMIT);
 1402                                         break;
 1403                                 case IPV6_HOPOPTS:
 1404                                         /*
 1405                                          * Check super-user privilege.
 1406                                          * See comments for IPV6_RECVHOPOPTS.
 1407                                          */
 1408                                         if (!privileged)
 1409                                                 return(EPERM);
 1410                                         OPTSET(IN6P_HOPOPTS);
 1411                                         break;
 1412                                 case IPV6_DSTOPTS:
 1413                                         if (!privileged)
 1414                                                 return(EPERM);
 1415                                         OPTSET(IN6P_DSTOPTS|IN6P_RTHDRDSTOPTS); /* XXX */
 1416                                         break;
 1417                                 case IPV6_RTHDR:
 1418                                         OPTSET(IN6P_RTHDR);
 1419                                         break;
 1420                                 }
 1421                                 break;
 1422 #undef OPTSET
 1423 
 1424                         case IPV6_MULTICAST_IF:
 1425                         case IPV6_MULTICAST_HOPS:
 1426                         case IPV6_MULTICAST_LOOP:
 1427                         case IPV6_JOIN_GROUP:
 1428                         case IPV6_LEAVE_GROUP:
 1429                             {
 1430                                 struct mbuf *m;
 1431                                 if (sopt->sopt_valsize > MLEN) {
 1432                                         error = EMSGSIZE;
 1433                                         break;
 1434                                 }
 1435                                 /* XXX */
 1436                                 MGET(m, sopt->sopt_p ? M_WAIT : M_DONTWAIT, MT_HEADER);
 1437                                 if (m == 0) {
 1438                                         error = ENOBUFS;
 1439                                         break;
 1440                                 }
 1441                                 m->m_len = sopt->sopt_valsize;
 1442                                 error = sooptcopyin(sopt, mtod(m, char *),
 1443                                                     m->m_len, m->m_len);
 1444                                 error = ip6_setmoptions(sopt->sopt_name, in6p, m);
 1445                                 (void)m_free(m);
 1446                             }
 1447                                 break;
 1448 
 1449                         case IPV6_PORTRANGE:
 1450                                 error = sooptcopyin(sopt, &optval,
 1451                                     sizeof optval, sizeof optval);
 1452                                 if (error)
 1453                                         break;
 1454 
 1455                                 switch (optval) {
 1456                                 case IPV6_PORTRANGE_DEFAULT:
 1457                                         in6p->in6p_flags &= ~(IN6P_LOWPORT);
 1458                                         in6p->in6p_flags &= ~(IN6P_HIGHPORT);
 1459                                         break;
 1460 
 1461                                 case IPV6_PORTRANGE_HIGH:
 1462                                         in6p->in6p_flags &= ~(IN6P_LOWPORT);
 1463                                         in6p->in6p_flags |= IN6P_HIGHPORT;
 1464                                         break;
 1465 
 1466                                 case IPV6_PORTRANGE_LOW:
 1467                                         in6p->in6p_flags &= ~(IN6P_HIGHPORT);
 1468                                         in6p->in6p_flags |= IN6P_LOWPORT;
 1469                                         break;
 1470 
 1471                                 default:
 1472                                         error = EINVAL;
 1473                                         break;
 1474                                 }
 1475                                 break;
 1476 
 1477 #if IPSEC
 1478                         case IPV6_IPSEC_POLICY:
 1479                             {
 1480                                 caddr_t req = NULL;
 1481                                 size_t len = 0;
 1482                                 struct mbuf *m;
 1483 
 1484                                 if ((error = soopt_getm(sopt, &m)) != 0) /* XXX */
 1485                                         break;
 1486                                 if ((error = soopt_mcopyin(sopt, m)) != 0) /* XXX */
 1487                                         break;
 1488                                 if (m) {
 1489                                         req = mtod(m, caddr_t);
 1490                                         len = m->m_len;
 1491                                 }
 1492                                 error = ipsec6_set_policy(in6p, optname, req,
 1493                                                           len, privileged);
 1494                                 m_freem(m);
 1495                             }
 1496                                 break;
 1497 #endif /* KAME IPSEC */
 1498 
 1499                         case IPV6_FW_ADD:
 1500                         case IPV6_FW_DEL:
 1501                         case IPV6_FW_FLUSH:
 1502                         case IPV6_FW_ZERO:
 1503                                 {
 1504                                 if (ip6_fw_ctl_ptr == NULL && load_ipfw() != 0)
 1505                                         return EINVAL;
 1506 
 1507                                 error = (*ip6_fw_ctl_ptr)(sopt);
 1508                                 }
 1509                                 break;
 1510 
 1511                         default:
 1512                                 error = ENOPROTOOPT;
 1513                                 break;
 1514                         }
 1515                         break;
 1516 
 1517                 case SOPT_GET:
 1518                         switch (optname) {
 1519 
 1520                         case IPV6_PKTOPTIONS:
 1521                                 if (in6p->in6p_options) {
 1522                                         struct mbuf *m;
 1523                                         m = m_copym(in6p->in6p_options,
 1524                                             0, M_COPYALL, M_WAIT);
 1525                                         error = soopt_mcopyout(sopt, m);
 1526                                         if (error == 0)
 1527                                                 m_freem(m);
 1528                                 } else
 1529                                         sopt->sopt_valsize = 0;
 1530                                 break;
 1531 
 1532                         case IPV6_UNICAST_HOPS:
 1533                         case IPV6_CHECKSUM:
 1534 
 1535                         case IPV6_FAITH:
 1536                         case IPV6_V6ONLY:
 1537                         case IPV6_PORTRANGE:
 1538                                 switch (optname) {
 1539 
 1540                                 case IPV6_UNICAST_HOPS:
 1541                                         optval = in6p->in6p_hops;
 1542                                         break;
 1543 
 1544                                 case IPV6_CHECKSUM:
 1545                                         optval = in6p->in6p_cksum;
 1546                                         break;
 1547 
 1548                                 case IPV6_FAITH:
 1549                                         optval = OPTBIT(IN6P_FAITH);
 1550                                         break;
 1551 
 1552                                 case IPV6_V6ONLY:
 1553                                         optval = OPTBIT(IN6P_IPV6_V6ONLY);
 1554                                         break;
 1555 
 1556                                 case IPV6_PORTRANGE:
 1557                                     {
 1558                                         int flags;
 1559                                         flags = in6p->in6p_flags;
 1560                                         if (flags & IN6P_HIGHPORT)
 1561                                                 optval = IPV6_PORTRANGE_HIGH;
 1562                                         else if (flags & IN6P_LOWPORT)
 1563                                                 optval = IPV6_PORTRANGE_LOW;
 1564                                         else
 1565                                                 optval = 0;
 1566                                         break;
 1567                                     }
 1568                                 }
 1569                                 error = sooptcopyout(sopt, &optval,
 1570                                         sizeof optval);
 1571                                 break;
 1572 
 1573                         case IPV6_PKTINFO:
 1574                         case IPV6_HOPLIMIT:
 1575                         case IPV6_HOPOPTS:
 1576                         case IPV6_RTHDR:
 1577                         case IPV6_DSTOPTS:
 1578                                 if (optname == IPV6_HOPOPTS ||
 1579                                     optname == IPV6_DSTOPTS ||
 1580                                     !privileged)
 1581                                         return(EPERM);
 1582                                 switch (optname) {
 1583                                 case IPV6_PKTINFO:
 1584                                         optval = OPTBIT(IN6P_PKTINFO);
 1585                                         break;
 1586                                 case IPV6_HOPLIMIT:
 1587                                         optval = OPTBIT(IN6P_HOPLIMIT);
 1588                                         break;
 1589                                 case IPV6_HOPOPTS:
 1590                                         if (!privileged)
 1591                                                 return(EPERM);
 1592                                         optval = OPTBIT(IN6P_HOPOPTS);
 1593                                         break;
 1594                                 case IPV6_RTHDR:
 1595                                         optval = OPTBIT(IN6P_RTHDR);
 1596                                         break;
 1597                                 case IPV6_DSTOPTS:
 1598                                         if (!privileged)
 1599                                                 return(EPERM);
 1600                                         optval = OPTBIT(IN6P_DSTOPTS|IN6P_RTHDRDSTOPTS);
 1601                                         break;
 1602                                 }
 1603                                 error = sooptcopyout(sopt, &optval,
 1604                                         sizeof optval);
 1605                                 break;
 1606 
 1607                         case IPV6_MULTICAST_IF:
 1608                         case IPV6_MULTICAST_HOPS:
 1609                         case IPV6_MULTICAST_LOOP:
 1610                         case IPV6_JOIN_GROUP:
 1611                         case IPV6_LEAVE_GROUP:
 1612                             {
 1613                                 struct mbuf *m;
 1614                                 error = ip6_getmoptions(sopt->sopt_name,
 1615                                                 in6p->in6p_moptions, &m);
 1616                                 if (error == 0)
 1617                                         error = sooptcopyout(sopt,
 1618                                                 mtod(m, char *), m->m_len);
 1619                                 m_freem(m);
 1620                             }
 1621                                 break;
 1622 
 1623 #if IPSEC
 1624                         case IPV6_IPSEC_POLICY:
 1625                           {
 1626                                 caddr_t req = NULL;
 1627                                 size_t len = 0;
 1628                                 struct mbuf *m = NULL;
 1629                                 struct mbuf **mp = &m;
 1630 
 1631                                 error = soopt_getm(sopt, &m); /* XXX */
 1632                                 if (error != NULL)
 1633                                         break;
 1634                                 error = soopt_mcopyin(sopt, m); /* XXX */
 1635                                 if (error != NULL)
 1636                                         break;
 1637                                 if (m) {
 1638                                         req = mtod(m, caddr_t);
 1639                                         len = m->m_len;
 1640                                 }
 1641                                 error = ipsec6_get_policy(in6p, req, len, mp);
 1642                                 if (error == 0)
 1643                                         error = soopt_mcopyout(sopt, m); /*XXX*/
 1644                                 if (error == 0 && m)
 1645                                         m_freem(m);
 1646                                 break;
 1647                           }
 1648 #endif /* KAME IPSEC */
 1649 
 1650                         case IPV6_FW_GET:
 1651                                 {
 1652                                 if (ip6_fw_ctl_ptr == NULL && load_ipfw() != 0)
 1653                                         return EINVAL;
 1654 
 1655                                 error = (*ip6_fw_ctl_ptr)(sopt);
 1656                                 }
 1657                                 break;
 1658 
 1659                         default:
 1660                                 error = ENOPROTOOPT;
 1661                                 break;
 1662                         }
 1663                         break;
 1664                 }
 1665         } else {
 1666                 error = EINVAL;
 1667         }
 1668         return(error);
 1669 }
 1670 
 1671 /*
 1672  * Set up IP6 options in pcb for insertion in output packets or
 1673  * specifying behavior of outgoing packets.
 1674  */
 1675 static int
 1676 ip6_pcbopts(pktopt, m, so, sopt)
 1677         struct ip6_pktopts **pktopt;
 1678         struct mbuf *m;
 1679         struct socket *so;
 1680         struct sockopt *sopt;
 1681 {
 1682         struct ip6_pktopts *opt = *pktopt;
 1683         int error = 0;
 1684         struct proc *p = sopt->sopt_p;
 1685         int priv = 0;
 1686 
 1687         /* turn off any old options. */
 1688         if (opt) {
 1689 #if DIAGNOSTIC
 1690                 if (opt->ip6po_pktinfo || opt->ip6po_nexthop ||
 1691                     opt->ip6po_hbh || opt->ip6po_dest1 || opt->ip6po_dest2 ||
 1692                     opt->ip6po_rhinfo.ip6po_rhi_rthdr)
 1693                         printf("ip6_pcbopts: all specified options are cleared.\n");
 1694 #endif
 1695                 ip6_clearpktopts(opt, 1, -1);
 1696         } else
 1697                 opt = _MALLOC(sizeof(*opt), M_IP6OPT, M_WAITOK);
 1698         *pktopt = NULL;
 1699 
 1700         if (!m || m->m_len == 0) {
 1701                 /*
 1702                  * Only turning off any previous options, regardless of
 1703                  * whether the opt is just created or given.
 1704                  */
 1705                 if (opt)
 1706                         FREE(opt, M_IP6OPT);
 1707                 return(0);
 1708         }
 1709 
 1710         /*  set options specified by user. */
 1711         if (p && !suser(p->p_ucred, &p->p_acflag))
 1712                 priv = 1;
 1713         if ((error = ip6_setpktoptions(m, opt, priv, 1)) != 0) {
 1714                 ip6_clearpktopts(opt, 1, -1); /* XXX: discard all options */
 1715                 FREE(opt, M_IP6OPT);
 1716                 return(error);
 1717         }
 1718         *pktopt = opt;
 1719         return(0);
 1720 }
 1721 
 1722 /*
 1723  * initialize ip6_pktopts.  beware that there are non-zero default values in
 1724  * the struct.
 1725  */
 1726 void
 1727 init_ip6pktopts(opt)
 1728         struct ip6_pktopts *opt;
 1729 {
 1730 
 1731         bzero(opt, sizeof(*opt));
 1732         opt->ip6po_hlim = -1;   /* -1 means default hop limit */
 1733 }
 1734 
 1735 void
 1736 ip6_clearpktopts(pktopt, needfree, optname)
 1737         struct ip6_pktopts *pktopt;
 1738         int needfree, optname;
 1739 {
 1740         if (pktopt == NULL)
 1741                 return;
 1742 
 1743         if (optname == -1) {
 1744                 if (needfree && pktopt->ip6po_pktinfo)
 1745                         FREE(pktopt->ip6po_pktinfo, M_IP6OPT);
 1746                 pktopt->ip6po_pktinfo = NULL;
 1747         }
 1748         if (optname == -1)
 1749                 pktopt->ip6po_hlim = -1;
 1750         if (optname == -1) {
 1751                 if (needfree && pktopt->ip6po_nexthop)
 1752                         FREE(pktopt->ip6po_nexthop, M_IP6OPT);
 1753                 pktopt->ip6po_nexthop = NULL;
 1754         }
 1755         if (optname == -1) {
 1756                 if (needfree && pktopt->ip6po_hbh)
 1757                         FREE(pktopt->ip6po_hbh, M_IP6OPT);
 1758                 pktopt->ip6po_hbh = NULL;
 1759         }
 1760         if (optname == -1) {
 1761                 if (needfree && pktopt->ip6po_dest1)
 1762                         FREE(pktopt->ip6po_dest1, M_IP6OPT);
 1763                 pktopt->ip6po_dest1 = NULL;
 1764         }
 1765         if (optname == -1) {
 1766                 if (needfree && pktopt->ip6po_rhinfo.ip6po_rhi_rthdr)
 1767                         FREE(pktopt->ip6po_rhinfo.ip6po_rhi_rthdr, M_IP6OPT);
 1768                 pktopt->ip6po_rhinfo.ip6po_rhi_rthdr = NULL;
 1769                 if (pktopt->ip6po_route.ro_rt) {
 1770                         rtfree(pktopt->ip6po_route.ro_rt);
 1771                         pktopt->ip6po_route.ro_rt = NULL;
 1772                 }
 1773         }
 1774         if (optname == -1) {
 1775                 if (needfree && pktopt->ip6po_dest2)
 1776                         FREE(pktopt->ip6po_dest2, M_IP6OPT);
 1777                 pktopt->ip6po_dest2 = NULL;
 1778         }
 1779 }
 1780 
 1781 #define PKTOPT_EXTHDRCPY(type) \
 1782 do {\
 1783         if (src->type) {\
 1784                 int hlen =\
 1785                         (((struct ip6_ext *)src->type)->ip6e_len + 1) << 3;\
 1786                 dst->type = _MALLOC(hlen, M_IP6OPT, canwait);\
 1787                 if (dst->type == NULL && canwait == M_NOWAIT)\
 1788                         goto bad;\
 1789                 bcopy(src->type, dst->type, hlen);\
 1790         }\
 1791 } while (0)
 1792 
 1793 struct ip6_pktopts *
 1794 ip6_copypktopts(src, canwait)
 1795         struct ip6_pktopts *src;
 1796         int canwait;
 1797 {
 1798         struct ip6_pktopts *dst;
 1799 
 1800         if (src == NULL) {
 1801                 printf("ip6_clearpktopts: invalid argument\n");
 1802                 return(NULL);
 1803         }
 1804 
 1805         dst = _MALLOC(sizeof(*dst), M_IP6OPT, canwait);
 1806         if (dst == NULL && canwait == M_NOWAIT)
 1807                 return (NULL);
 1808         bzero(dst, sizeof(*dst));
 1809 
 1810         dst->ip6po_hlim = src->ip6po_hlim;
 1811         if (src->ip6po_pktinfo) {
 1812                 dst->ip6po_pktinfo = _MALLOC(sizeof(*dst->ip6po_pktinfo),
 1813                                             M_IP6OPT, canwait);
 1814                 if (dst->ip6po_pktinfo == NULL && canwait == M_NOWAIT)
 1815                         goto bad;
 1816                 *dst->ip6po_pktinfo = *src->ip6po_pktinfo;
 1817         }
 1818         if (src->ip6po_nexthop) {
 1819                 dst->ip6po_nexthop = _MALLOC(src->ip6po_nexthop->sa_len,
 1820                                             M_IP6OPT, canwait);
 1821                 if (dst->ip6po_nexthop == NULL && canwait == M_NOWAIT)
 1822                         goto bad;
 1823                 bcopy(src->ip6po_nexthop, dst->ip6po_nexthop,
 1824                       src->ip6po_nexthop->sa_len);
 1825         }
 1826         PKTOPT_EXTHDRCPY(ip6po_hbh);
 1827         PKTOPT_EXTHDRCPY(ip6po_dest1);
 1828         PKTOPT_EXTHDRCPY(ip6po_dest2);
 1829         PKTOPT_EXTHDRCPY(ip6po_rthdr); /* not copy the cached route */
 1830         return(dst);
 1831 
 1832   bad:
 1833         if (dst->ip6po_pktinfo) FREE(dst->ip6po_pktinfo, M_IP6OPT);
 1834         if (dst->ip6po_nexthop) FREE(dst->ip6po_nexthop, M_IP6OPT);
 1835         if (dst->ip6po_hbh) FREE(dst->ip6po_hbh, M_IP6OPT);
 1836         if (dst->ip6po_dest1) FREE(dst->ip6po_dest1, M_IP6OPT);
 1837         if (dst->ip6po_dest2) FREE(dst->ip6po_dest2, M_IP6OPT);
 1838         if (dst->ip6po_rthdr) FREE(dst->ip6po_rthdr, M_IP6OPT);
 1839         FREE(dst, M_IP6OPT);
 1840         return(NULL);
 1841 }
 1842 #undef PKTOPT_EXTHDRCPY
 1843 
 1844 void
 1845 ip6_freepcbopts(pktopt)
 1846         struct ip6_pktopts *pktopt;
 1847 {
 1848         if (pktopt == NULL)
 1849                 return;
 1850 
 1851         ip6_clearpktopts(pktopt, 1, -1);
 1852 
 1853         FREE(pktopt, M_IP6OPT);
 1854 }
 1855 
 1856 /*
 1857  * Set the IP6 multicast options in response to user setsockopt().
 1858  */
 1859 static int
 1860 ip6_setmoptions(optname, in6p, m)
 1861         int optname;
 1862         struct inpcb* in6p;
 1863         struct mbuf *m;
 1864 {
 1865         int error = 0;
 1866         u_int loop, ifindex;
 1867         struct ipv6_mreq *mreq;
 1868         struct ifnet *ifp;
 1869         struct ip6_moptions **im6op = &in6p->in6p_moptions;
 1870         struct ip6_moptions *im6o = *im6op;
 1871         struct ip_moptions *imo;
 1872         struct route_in6 ro;
 1873         struct sockaddr_in6 *dst;
 1874         struct in6_multi_mship *imm;
 1875         struct proc *p = current_proc();        /* XXX */
 1876 
 1877         if (im6o == NULL) {
 1878                 /*
 1879                  * No multicast option buffer attached to the pcb;
 1880                  * allocate one and initialize to default values.
 1881                  */
 1882                 im6o = (struct ip6_moptions *)
 1883                         _MALLOC(sizeof(*im6o), M_IPMOPTS, M_WAITOK);
 1884 
 1885                 if (im6o == NULL)
 1886                         return(ENOBUFS);
 1887                 *im6op = im6o;
 1888                 im6o->im6o_multicast_ifp = NULL;
 1889                 im6o->im6o_multicast_hlim = ip6_defmcasthlim;
 1890                 im6o->im6o_multicast_loop = IPV6_DEFAULT_MULTICAST_LOOP;
 1891                 LIST_INIT(&im6o->im6o_memberships);
 1892         }
 1893         
 1894         if (in6p->inp_moptions == NULL) {
 1895                 /*
 1896                  * No IPv4 multicast option buffer attached to the pcb;
 1897                  * call ip_createmoptions to allocate one and initialize
 1898                  * to default values.
 1899                  */
 1900                 error = ip_createmoptions(&in6p->inp_moptions);
 1901                 if (error != 0)
 1902                         return error;
 1903         }
 1904         imo = in6p->inp_moptions;
 1905 
 1906         switch (optname) {
 1907 
 1908         case IPV6_MULTICAST_IF:
 1909                 /*
 1910                  * Select the interface for outgoing multicast packets.
 1911                  */
 1912                 if (m == NULL || m->m_len != sizeof(u_int)) {
 1913                         error = EINVAL;
 1914                         break;
 1915                 }
 1916                 bcopy(mtod(m, u_int *), &ifindex, sizeof(ifindex));
 1917                 if (ifindex < 0 || if_index < ifindex) {
 1918                         error = ENXIO;  /* XXX EINVAL? */
 1919                         break;
 1920                 }
 1921                 ifp = ifindex2ifnet[ifindex];
 1922                 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) {
 1923                         error = EADDRNOTAVAIL;
 1924                         break;
 1925                 }
 1926                 im6o->im6o_multicast_ifp = ifp;
 1927                 imo->imo_multicast_ifp = ifp;
 1928                 break;
 1929 
 1930         case IPV6_MULTICAST_HOPS:
 1931             {
 1932                 /*
 1933                  * Set the IP6 hoplimit for outgoing multicast packets.
 1934                  */
 1935                 int optval;
 1936                 if (m == NULL || m->m_len != sizeof(int)) {
 1937                         error = EINVAL;
 1938                         break;
 1939                 }
 1940                 bcopy(mtod(m, u_int *), &optval, sizeof(optval));
 1941                 if (optval < -1 || optval >= 256)
 1942                         error = EINVAL;
 1943                 else if (optval == -1) {
 1944                         im6o->im6o_multicast_hlim = ip6_defmcasthlim;
 1945                         imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL;
 1946                 } else {
 1947                         im6o->im6o_multicast_hlim = optval;
 1948                         imo->imo_multicast_ttl = optval;
 1949                 }
 1950                 break;
 1951             }
 1952 
 1953         case IPV6_MULTICAST_LOOP:
 1954                 /*
 1955                  * Set the loopback flag for outgoing multicast packets.
 1956                  * Must be zero or one.
 1957                  */
 1958                 if (m == NULL || m->m_len != sizeof(u_int)) {
 1959                         error = EINVAL;
 1960                         break;
 1961                 }
 1962                 bcopy(mtod(m, u_int *), &loop, sizeof(loop));
 1963                 if (loop > 1) {
 1964                         error = EINVAL;
 1965                         break;
 1966                 }
 1967                 im6o->im6o_multicast_loop = loop;
 1968                 imo->imo_multicast_loop = loop;
 1969                 break;
 1970 
 1971         case IPV6_JOIN_GROUP:
 1972                 /*
 1973                  * Add a multicast group membership.
 1974                  * Group must be a valid IP6 multicast address.
 1975                  */
 1976                 if (m == NULL || m->m_len != sizeof(struct ipv6_mreq)) {
 1977                         error = EINVAL;
 1978                         break;
 1979                 }
 1980                 mreq = mtod(m, struct ipv6_mreq *);
 1981                 /*
 1982                  * If the interface is specified, validate it.
 1983                  */
 1984                 if (mreq->ipv6mr_interface < 0
 1985                  || if_index < mreq->ipv6mr_interface) {
 1986                         error = ENXIO;  /* XXX EINVAL? */
 1987                         break;
 1988                 }
 1989                 
 1990                 if (IN6_IS_ADDR_UNSPECIFIED(&mreq->ipv6mr_multiaddr)) {
 1991                         /*
 1992                          * We use the unspecified address to specify to accept
 1993                          * all multicast addresses. Only super user is allowed
 1994                          * to do this.
 1995                          */
 1996                         if (suser(p->p_ucred, &p->p_acflag))
 1997                         {
 1998                                 error = EACCES;
 1999                                 break;
 2000                         }
 2001                 } else if (IN6_IS_ADDR_V4MAPPED(&mreq->ipv6mr_multiaddr)) {
 2002                         struct ip_mreq v4req;
 2003                         
 2004                         v4req.imr_multiaddr.s_addr = mreq->ipv6mr_multiaddr.s6_addr32[3];
 2005                         v4req.imr_interface.s_addr = INADDR_ANY;
 2006                         
 2007                         /* Find an IPv4 address on the specified interface. */
 2008                         if (mreq->ipv6mr_interface != 0) {
 2009                                 struct in_ifaddr *ifa;
 2010 
 2011                                 ifp = ifindex2ifnet[mreq->ipv6mr_interface];
 2012 
 2013                                 TAILQ_FOREACH(ifa, &in_ifaddrhead, ia_link) {
 2014                                         if (ifa->ia_ifp == ifp) {
 2015                                                 v4req.imr_interface = IA_SIN(ifa)->sin_addr;
 2016                                                 break;
 2017                                         }
 2018                                 }
 2019                                 
 2020                                 if (v4req.imr_multiaddr.s_addr == 0) {
 2021                                         /* Interface has no IPv4 address. */
 2022                                         error = EINVAL;
 2023                                         break;
 2024                                 }
 2025                         }
 2026                         
 2027                         error = ip_addmembership(imo, &v4req);
 2028                         break;
 2029                 } else if (!IN6_IS_ADDR_MULTICAST(&mreq->ipv6mr_multiaddr)) {
 2030                         error = EINVAL;
 2031                         break;
 2032                 }
 2033                 /*
 2034                  * If no interface was explicitly specified, choose an
 2035                  * appropriate one according to the given multicast address.
 2036                  */
 2037                 if (mreq->ipv6mr_interface == 0) {
 2038                         /*
 2039                          * If the multicast address is in node-local scope,
 2040                          * the interface should be a loopback interface.
 2041                          * Otherwise, look up the routing table for the
 2042                          * address, and choose the outgoing interface.
 2043                          *   XXX: is it a good approach?
 2044                          */
 2045                         if (IN6_IS_ADDR_MC_NODELOCAL(&mreq->ipv6mr_multiaddr)) {
 2046                                 ifp = &loif[0];
 2047                         } else {
 2048                                 ro.ro_rt = NULL;
 2049                                 dst = (struct sockaddr_in6 *)&ro.ro_dst;
 2050                                 bzero(dst, sizeof(*dst));
 2051                                 dst->sin6_len = sizeof(struct sockaddr_in6);
 2052                                 dst->sin6_family = AF_INET6;
 2053                                 dst->sin6_addr = mreq->ipv6mr_multiaddr;
 2054                                 rtalloc((struct route *)&ro);
 2055                                 if (ro.ro_rt == NULL) {
 2056                                         error = EADDRNOTAVAIL;
 2057                                         break;
 2058                                 }
 2059                                 ifp = ro.ro_rt->rt_ifp;
 2060                                 rtfree(ro.ro_rt);
 2061                         }
 2062                 } else
 2063                         ifp = ifindex2ifnet[mreq->ipv6mr_interface];
 2064 
 2065                 /*
 2066                  * See if we found an interface, and confirm that it
 2067                  * supports multicast
 2068                  */
 2069                 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) {
 2070                         error = EADDRNOTAVAIL;
 2071                         break;
 2072                 }
 2073                 /*
 2074                  * Put interface index into the multicast address,
 2075                  * if the address has link-local scope.
 2076                  */
 2077                 if (IN6_IS_ADDR_MC_LINKLOCAL(&mreq->ipv6mr_multiaddr)) {
 2078                         mreq->ipv6mr_multiaddr.s6_addr16[1]
 2079                                 = htons(mreq->ipv6mr_interface);
 2080                 }
 2081                 /*
 2082                  * See if the membership already exists.
 2083                  */
 2084                 for (imm = im6o->im6o_memberships.lh_first;
 2085                      imm != NULL; imm = imm->i6mm_chain.le_next)
 2086                         if (imm->i6mm_maddr->in6m_ifp == ifp &&
 2087                             IN6_ARE_ADDR_EQUAL(&imm->i6mm_maddr->in6m_addr,
 2088                                                &mreq->ipv6mr_multiaddr))
 2089                                 break;
 2090                 if (imm != NULL) {
 2091                         error = EADDRINUSE;
 2092                         break;
 2093                 }
 2094                 /*
 2095                  * Everything looks good; add a new record to the multicast
 2096                  * address list for the given interface.
 2097                  */
 2098                 imm = _MALLOC(sizeof(*imm), M_IPMADDR, M_WAITOK);
 2099                 if (imm == NULL) {
 2100                         error = ENOBUFS;
 2101                         break;
 2102                 }
 2103                 if ((imm->i6mm_maddr =
 2104                      in6_addmulti(&mreq->ipv6mr_multiaddr, ifp, &error)) == NULL) {
 2105                         FREE(imm, M_IPMADDR);
 2106                         break;
 2107                 }
 2108                 LIST_INSERT_HEAD(&im6o->im6o_memberships, imm, i6mm_chain);
 2109                 break;
 2110 
 2111         case IPV6_LEAVE_GROUP:
 2112                 /*
 2113                  * Drop a multicast group membership.
 2114                  * Group must be a valid IP6 multicast address.
 2115                  */
 2116                 if (m == NULL || m->m_len != sizeof(struct ipv6_mreq)) {
 2117                         error = EINVAL;
 2118                         break;
 2119                 }
 2120                 mreq = mtod(m, struct ipv6_mreq *);
 2121                 /*
 2122                  * If an interface address was specified, get a pointer
 2123                  * to its ifnet structure.
 2124                  */
 2125                 if (mreq->ipv6mr_interface < 0
 2126                  || if_index < mreq->ipv6mr_interface) {
 2127                         error = ENXIO;  /* XXX EINVAL? */
 2128                         break;
 2129                 }
 2130                 ifp = ifindex2ifnet[mreq->ipv6mr_interface];
 2131                 
 2132                 if (IN6_IS_ADDR_UNSPECIFIED(&mreq->ipv6mr_multiaddr)) {
 2133                         if (suser(p->p_ucred, &p->p_acflag)) {
 2134                                 error = EACCES;
 2135                                 break;
 2136                         }
 2137                 } else if (IN6_IS_ADDR_V4MAPPED(&mreq->ipv6mr_multiaddr)) {
 2138                         struct ip_mreq v4req;
 2139                         
 2140                         v4req.imr_multiaddr.s_addr = mreq->ipv6mr_multiaddr.s6_addr32[3];
 2141                         v4req.imr_interface.s_addr = INADDR_ANY;
 2142                         
 2143                         if (ifp != NULL) {
 2144                                 struct in_ifaddr *ifa;
 2145                                 
 2146                                 TAILQ_FOREACH(ifa, &in_ifaddrhead, ia_link) {
 2147                                         if (ifa->ia_ifp == ifp) {
 2148                                                 v4req.imr_interface = IA_SIN(ifa)->sin_addr;
 2149                                                 break;
 2150                                         }
 2151                                 }
 2152                         }
 2153                         
 2154                         error = ip_dropmembership(imo, &v4req);
 2155                         break;
 2156                 } else if (!IN6_IS_ADDR_MULTICAST(&mreq->ipv6mr_multiaddr)) {
 2157                         error = EINVAL;
 2158                         break;
 2159                 }
 2160                 /*
 2161                  * Put interface index into the multicast address,
 2162                  * if the address has link-local scope.
 2163                  */
 2164                 if (IN6_IS_ADDR_MC_LINKLOCAL(&mreq->ipv6mr_multiaddr)) {
 2165                         mreq->ipv6mr_multiaddr.s6_addr16[1]
 2166                                 = htons(mreq->ipv6mr_interface);
 2167                 }
 2168                 /*
 2169                  * Find the membership in the membership list.
 2170                  */
 2171                 for (imm = im6o->im6o_memberships.lh_first;
 2172                      imm != NULL; imm = imm->i6mm_chain.le_next) {
 2173                         if ((ifp == NULL ||
 2174                              imm->i6mm_maddr->in6m_ifp == ifp) &&
 2175                             IN6_ARE_ADDR_EQUAL(&imm->i6mm_maddr->in6m_addr,
 2176                                                &mreq->ipv6mr_multiaddr))
 2177                                 break;
 2178                 }
 2179                 if (imm == NULL) {
 2180                         /* Unable to resolve interface */
 2181                         error = EADDRNOTAVAIL;
 2182                         break;
 2183                 }
 2184                 /*
 2185                  * Give up the multicast address record to which the
 2186                  * membership points.
 2187                  */
 2188                 LIST_REMOVE(imm, i6mm_chain);
 2189                 in6_delmulti(imm->i6mm_maddr);
 2190                 FREE(imm, M_IPMADDR);
 2191                 break;
 2192 
 2193         default:
 2194                 error = EOPNOTSUPP;
 2195                 break;
 2196         }
 2197 
 2198         /*
 2199          * If all options have default values, no need to keep the mbuf.
 2200          */
 2201         if (im6o->im6o_multicast_ifp == NULL &&
 2202             im6o->im6o_multicast_hlim == ip6_defmcasthlim &&
 2203             im6o->im6o_multicast_loop == IPV6_DEFAULT_MULTICAST_LOOP &&
 2204             im6o->im6o_memberships.lh_first == NULL) {
 2205                 FREE(*im6op, M_IPMOPTS);
 2206                 *im6op = NULL;
 2207         }
 2208         if (imo->imo_multicast_ifp == NULL &&
 2209             imo->imo_multicast_vif == -1 &&
 2210             imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL &&
 2211             imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP &&
 2212             imo->imo_num_memberships == 0) {
 2213                 ip_freemoptions(imo);
 2214                 in6p->inp_moptions = 0;
 2215         }
 2216 
 2217         return(error);
 2218 }
 2219 
 2220 /*
 2221  * Return the IP6 multicast options in response to user getsockopt().
 2222  */
 2223 static int
 2224 ip6_getmoptions(optname, im6o, mp)
 2225         int optname;
 2226         struct ip6_moptions *im6o;
 2227         struct mbuf **mp;
 2228 {
 2229         u_int *hlim, *loop, *ifindex;
 2230 
 2231         *mp = m_get(M_WAIT, MT_HEADER);         /*XXX*/
 2232 
 2233         switch (optname) {
 2234 
 2235         case IPV6_MULTICAST_IF:
 2236                 ifindex = mtod(*mp, u_int *);
 2237                 (*mp)->m_len = sizeof(u_int);
 2238                 if (im6o == NULL || im6o->im6o_multicast_ifp == NULL)
 2239                         *ifindex = 0;
 2240                 else
 2241                         *ifindex = im6o->im6o_multicast_ifp->if_index;
 2242                 return(0);
 2243 
 2244         case IPV6_MULTICAST_HOPS:
 2245                 hlim = mtod(*mp, u_int *);
 2246                 (*mp)->m_len = sizeof(u_int);
 2247                 if (im6o == NULL)
 2248                         *hlim = ip6_defmcasthlim;
 2249                 else
 2250                         *hlim = im6o->im6o_multicast_hlim;
 2251                 return(0);
 2252 
 2253         case IPV6_MULTICAST_LOOP:
 2254                 loop = mtod(*mp, u_int *);
 2255                 (*mp)->m_len = sizeof(u_int);
 2256                 if (im6o == NULL)
 2257                         *loop = ip6_defmcasthlim;
 2258                 else
 2259                         *loop = im6o->im6o_multicast_loop;
 2260                 return(0);
 2261 
 2262         default:
 2263                 return(EOPNOTSUPP);
 2264         }
 2265 }
 2266 
 2267 /*
 2268  * Discard the IP6 multicast options.
 2269  */
 2270 void
 2271 ip6_freemoptions(im6o)
 2272         struct ip6_moptions *im6o;
 2273 {
 2274         struct in6_multi_mship *imm;
 2275 
 2276         if (im6o == NULL)
 2277                 return;
 2278 
 2279         while ((imm = im6o->im6o_memberships.lh_first) != NULL) {
 2280                 LIST_REMOVE(imm, i6mm_chain);
 2281                 if (imm->i6mm_maddr)
 2282                         in6_delmulti(imm->i6mm_maddr);
 2283                 FREE(imm, M_IPMADDR);
 2284         }
 2285         FREE(im6o, M_IPMOPTS);
 2286 }
 2287 
 2288 /*
 2289  * Set IPv6 outgoing packet options based on advanced API.
 2290  */
 2291 int
 2292 ip6_setpktoptions(control, opt, priv, needcopy)
 2293         struct mbuf *control;
 2294         struct ip6_pktopts *opt;
 2295         int priv, needcopy;
 2296 {
 2297         struct cmsghdr *cm = 0;
 2298 
 2299         if (control == 0 || opt == 0)
 2300                 return(EINVAL);
 2301 
 2302         init_ip6pktopts(opt);
 2303 
 2304         /*
 2305          * XXX: Currently, we assume all the optional information is stored
 2306          * in a single mbuf.
 2307          */
 2308         if (control->m_next)
 2309                 return(EINVAL);
 2310 
 2311         for (; control->m_len; control->m_data += CMSG_ALIGN(cm->cmsg_len),
 2312                      control->m_len -= CMSG_ALIGN(cm->cmsg_len)) {
 2313                 cm = mtod(control, struct cmsghdr *);
 2314                 if (cm->cmsg_len == 0 || cm->cmsg_len > control->m_len)
 2315                         return(EINVAL);
 2316                 if (cm->cmsg_level != IPPROTO_IPV6)
 2317                         continue;
 2318 
 2319                 /*
 2320                  * XXX should check if RFC2292 API is mixed with 2292bis API
 2321                  */
 2322                 switch (cm->cmsg_type) {
 2323                 case IPV6_PKTINFO:
 2324                         if (cm->cmsg_len != CMSG_LEN(sizeof(struct in6_pktinfo)))
 2325                                 return(EINVAL);
 2326                         if (needcopy) {
 2327                                 /* XXX: Is it really WAITOK? */
 2328                                 opt->ip6po_pktinfo =
 2329                                         _MALLOC(sizeof(struct in6_pktinfo),
 2330                                                M_IP6OPT, M_WAITOK);
 2331                                 bcopy(CMSG_DATA(cm), opt->ip6po_pktinfo,
 2332                                     sizeof(struct in6_pktinfo));
 2333                         } else
 2334                                 opt->ip6po_pktinfo =
 2335                                         (struct in6_pktinfo *)CMSG_DATA(cm);
 2336                         if (opt->ip6po_pktinfo->ipi6_ifindex &&
 2337                             IN6_IS_ADDR_LINKLOCAL(&opt->ip6po_pktinfo->ipi6_addr))
 2338                                 opt->ip6po_pktinfo->ipi6_addr.s6_addr16[1] =
 2339                                         htons(opt->ip6po_pktinfo->ipi6_ifindex);
 2340 
 2341                         if (opt->ip6po_pktinfo->ipi6_ifindex > if_index
 2342                          || opt->ip6po_pktinfo->ipi6_ifindex < 0) {
 2343                                 return(ENXIO);
 2344                         }
 2345 
 2346                         /*
 2347                          * Check if the requested source address is indeed a
 2348                          * unicast address assigned to the node, and can be
 2349                          * used as the packet's source address.
 2350                          */
 2351                         if (!IN6_IS_ADDR_UNSPECIFIED(&opt->ip6po_pktinfo->ipi6_addr)) {
 2352                                 struct in6_ifaddr *ia6;
 2353                                 struct sockaddr_in6 sin6;
 2354 
 2355                                 bzero(&sin6, sizeof(sin6));
 2356                                 sin6.sin6_len = sizeof(sin6);
 2357                                 sin6.sin6_family = AF_INET6;
 2358                                 sin6.sin6_addr =
 2359                                         opt->ip6po_pktinfo->ipi6_addr;
 2360                                 ia6 = (struct in6_ifaddr *)ifa_ifwithaddr(sin6tosa(&sin6));
 2361                                 if (ia6 == NULL ||
 2362                                     (ia6->ia6_flags & (IN6_IFF_ANYCAST |
 2363                                                        IN6_IFF_NOTREADY)) != 0)
 2364                                         return(EADDRNOTAVAIL);
 2365                         }
 2366                         break;
 2367 
 2368                 case IPV6_HOPLIMIT:
 2369                         if (cm->cmsg_len != CMSG_LEN(sizeof(int)))
 2370                                 return(EINVAL);
 2371 
 2372                         opt->ip6po_hlim = *(int *)CMSG_DATA(cm);
 2373                         if (opt->ip6po_hlim < -1 || opt->ip6po_hlim > 255)
 2374                                 return(EINVAL);
 2375                         break;
 2376 
 2377                 case IPV6_NEXTHOP:
 2378                         if (!priv)
 2379                                 return(EPERM);
 2380 
 2381                         if (cm->cmsg_len < sizeof(u_char) ||
 2382                             /* check if cmsg_len is large enough for sa_len */
 2383                             cm->cmsg_len < CMSG_LEN(*CMSG_DATA(cm)))
 2384                                 return(EINVAL);
 2385 
 2386                         if (needcopy) {
 2387                                 opt->ip6po_nexthop =
 2388                                         _MALLOC(*CMSG_DATA(cm),
 2389                                                M_IP6OPT, M_WAITOK);
 2390                                 bcopy(CMSG_DATA(cm),
 2391                                       opt->ip6po_nexthop,
 2392                                       *CMSG_DATA(cm));
 2393                         } else
 2394                                 opt->ip6po_nexthop =
 2395                                         (struct sockaddr *)CMSG_DATA(cm);
 2396                         break;
 2397 
 2398                 case IPV6_HOPOPTS:
 2399                 {
 2400                         struct ip6_hbh *hbh;
 2401                         int hbhlen;
 2402 
 2403                         if (cm->cmsg_len < CMSG_LEN(sizeof(struct ip6_hbh)))
 2404                                 return(EINVAL);
 2405                         hbh = (struct ip6_hbh *)CMSG_DATA(cm);
 2406                         hbhlen = (hbh->ip6h_len + 1) << 3;
 2407                         if (cm->cmsg_len != CMSG_LEN(hbhlen))
 2408                                 return(EINVAL);
 2409 
 2410                         if (needcopy) {
 2411                                 opt->ip6po_hbh =
 2412                                         _MALLOC(hbhlen, M_IP6OPT, M_WAITOK);
 2413                                 bcopy(hbh, opt->ip6po_hbh, hbhlen);
 2414                         } else
 2415                                 opt->ip6po_hbh = hbh;
 2416                         break;
 2417                 }
 2418 
 2419                 case IPV6_DSTOPTS:
 2420                 {
 2421                         struct ip6_dest *dest, **newdest;
 2422                         int destlen;
 2423 
 2424                         if (cm->cmsg_len < CMSG_LEN(sizeof(struct ip6_dest)))
 2425                                 return(EINVAL);
 2426                         dest = (struct ip6_dest *)CMSG_DATA(cm);
 2427                         destlen = (dest->ip6d_len + 1) << 3;
 2428                         if (cm->cmsg_len != CMSG_LEN(destlen))
 2429                                 return(EINVAL);
 2430 
 2431                         /* 
 2432                          * The old advacned API is ambiguous on this
 2433                          * point. Our approach is to determine the
 2434                          * position based according to the existence
 2435                          * of a routing header. Note, however, that
 2436                          * this depends on the order of the extension
 2437                          * headers in the ancillary data; the 1st part
 2438                          * of the destination options header must
 2439                          * appear before the routing header in the
 2440                          * ancillary data, too.
 2441                          * RFC2292bis solved the ambiguity by
 2442                          * introducing separate cmsg types.
 2443                          */
 2444                         if (opt->ip6po_rthdr == NULL)
 2445                                 newdest = &opt->ip6po_dest1;
 2446                         else
 2447                                 newdest = &opt->ip6po_dest2;
 2448 
 2449                         if (needcopy) {
 2450                                 *newdest = _MALLOC(destlen, M_IP6OPT, M_WAITOK);
 2451                                 bcopy(dest, *newdest, destlen);
 2452                         } else
 2453                                 *newdest = dest;
 2454 
 2455                         break;
 2456                 }
 2457 
 2458                 case IPV6_RTHDR:
 2459                 {
 2460                         struct ip6_rthdr *rth;
 2461                         int rthlen;
 2462 
 2463                         if (cm->cmsg_len < CMSG_LEN(sizeof(struct ip6_rthdr)))
 2464                                 return(EINVAL);
 2465                         rth = (struct ip6_rthdr *)CMSG_DATA(cm);
 2466                         rthlen = (rth->ip6r_len + 1) << 3;
 2467                         if (cm->cmsg_len != CMSG_LEN(rthlen))
 2468                                 return(EINVAL);
 2469 
 2470                         switch (rth->ip6r_type) {
 2471                         case IPV6_RTHDR_TYPE_0:
 2472                                 /* must contain one addr */
 2473                                 if (rth->ip6r_len == 0)
 2474                                         return(EINVAL);
 2475                                 /* length must be even */
 2476                                 if (rth->ip6r_len % 2)
 2477                                         return(EINVAL);
 2478                                 if (rth->ip6r_len / 2 != rth->ip6r_segleft)
 2479                                         return(EINVAL);
 2480                                 break;
 2481                         default:
 2482                                 return(EINVAL); /* not supported */
 2483                         }
 2484 
 2485                         if (needcopy) {
 2486                                 opt->ip6po_rthdr = _MALLOC(rthlen, M_IP6OPT,
 2487                                                           M_WAITOK);
 2488                                 bcopy(rth, opt->ip6po_rthdr, rthlen);
 2489                         } else
 2490                                 opt->ip6po_rthdr = rth;
 2491 
 2492                         break;
 2493                 }
 2494 
 2495                 default:
 2496                         return(ENOPROTOOPT);
 2497                 }
 2498         }
 2499 
 2500         return(0);
 2501 }
 2502 
 2503 /*
 2504  * Routine called from ip6_output() to loop back a copy of an IP6 multicast
 2505  * packet to the input queue of a specified interface.  Note that this
 2506  * calls the output routine of the loopback "driver", but with an interface
 2507  * pointer that might NOT be &loif -- easier than replicating that code here.
 2508  */
 2509 void
 2510 ip6_mloopback(ifp, m, dst)
 2511         struct ifnet *ifp;
 2512         struct mbuf *m;
 2513         struct sockaddr_in6 *dst;
 2514 {
 2515         struct mbuf *copym;
 2516         struct ip6_hdr *ip6;
 2517 
 2518         copym = m_copy(m, 0, M_COPYALL);
 2519         if (copym == NULL)
 2520                 return;
 2521 
 2522         /*
 2523          * Make sure to deep-copy IPv6 header portion in case the data
 2524          * is in an mbuf cluster, so that we can safely override the IPv6
 2525          * header portion later.
 2526          */
 2527         if ((copym->m_flags & M_EXT) != 0 ||
 2528             copym->m_len < sizeof(struct ip6_hdr)) {
 2529                 copym = m_pullup(copym, sizeof(struct ip6_hdr));
 2530                 if (copym == NULL)
 2531                         return;
 2532         }
 2533 
 2534 #if DIAGNOSTIC
 2535         if (copym->m_len < sizeof(*ip6)) {
 2536                 m_freem(copym);
 2537                 return;
 2538         }
 2539 #endif
 2540 
 2541         ip6 = mtod(copym, struct ip6_hdr *);
 2542 #ifndef SCOPEDROUTING
 2543         /*
 2544          * clear embedded scope identifiers if necessary.
 2545          * in6_clearscope will touch the addresses only when necessary.
 2546          */
 2547         in6_clearscope(&ip6->ip6_src);
 2548         in6_clearscope(&ip6->ip6_dst);
 2549 #endif
 2550 
 2551 #ifdef __APPLE__
 2552 
 2553         /* Makes sure the HW checksum flags are cleaned before sending the packet */
 2554 
 2555         copym->m_pkthdr.rcvif = (struct ifnet *)0;
 2556         copym->m_pkthdr.csum_data = 0;
 2557         copym->m_pkthdr.csum_flags = 0;
 2558 
 2559         if (lo_dl_tag == 0)
 2560                 dlil_find_dltag(APPLE_IF_FAM_LOOPBACK, 0, PF_INET, &lo_dl_tag);
 2561 
 2562         if (lo_dl_tag) {
 2563                 copym->m_pkthdr.rcvif = ifp;
 2564                 dlil_output(lo_dl_tag, copym, 0, (struct sockaddr *)dst, 0);
 2565         } else
 2566                 m_free(copym);
 2567 #else
 2568         (void)if_simloop(ifp, copym, dst->sin6_family, NULL);
 2569 #endif
 2570 }
 2571 
 2572 /*
 2573  * Chop IPv6 header off from the payload.
 2574  */
 2575 static int
 2576 ip6_splithdr(m, exthdrs)
 2577         struct mbuf *m;
 2578         struct ip6_exthdrs *exthdrs;
 2579 {
 2580         struct mbuf *mh;
 2581         struct ip6_hdr *ip6;
 2582 
 2583         ip6 = mtod(m, struct ip6_hdr *);
 2584         if (m->m_len > sizeof(*ip6)) {
 2585                 MGETHDR(mh, M_DONTWAIT, MT_HEADER);
 2586                 if (mh == 0) {
 2587                         m_freem(m);
 2588                         return ENOBUFS;
 2589                 }
 2590                 M_COPY_PKTHDR(mh, m);
 2591                 MH_ALIGN(mh, sizeof(*ip6));
 2592                 m->m_flags &= ~M_PKTHDR;
 2593                 m->m_len -= sizeof(*ip6);
 2594                 m->m_data += sizeof(*ip6);
 2595                 mh->m_next = m;
 2596                 m = mh;
 2597                 m->m_len = sizeof(*ip6);
 2598                 bcopy((caddr_t)ip6, mtod(m, caddr_t), sizeof(*ip6));
 2599         }
 2600         exthdrs->ip6e_ip6 = m;
 2601         return 0;
 2602 }
 2603 
 2604 /*
 2605  * Compute IPv6 extension header length.
 2606  */
 2607 int
 2608 ip6_optlen(in6p)
 2609         struct in6pcb *in6p;
 2610 {
 2611         int len;
 2612 
 2613         if (!in6p->in6p_outputopts)
 2614                 return 0;
 2615 
 2616         len = 0;
 2617 #define elen(x) \
 2618     (((struct ip6_ext *)(x)) ? (((struct ip6_ext *)(x))->ip6e_len + 1) << 3 : 0)
 2619 
 2620         len += elen(in6p->in6p_outputopts->ip6po_hbh);
 2621         if (in6p->in6p_outputopts->ip6po_rthdr)
 2622                 /* dest1 is valid with rthdr only */
 2623                 len += elen(in6p->in6p_outputopts->ip6po_dest1);
 2624         len += elen(in6p->in6p_outputopts->ip6po_rthdr);
 2625         len += elen(in6p->in6p_outputopts->ip6po_dest2);
 2626         return len;
 2627 #undef elen
 2628 }

Cache object: 9d1fa0a487b2d0cacf8c81db786b4876


[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]


This page is part of the FreeBSD/Linux Linux Kernel Cross-Reference, and was automatically generated using a modified version of the LXR engine.