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/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 /*      $FreeBSD$       */
    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 #include "opt_ip6fw.h"
   69 #include "opt_inet.h"
   70 #include "opt_inet6.h"
   71 #include "opt_ipsec.h"
   72 
   73 #include <sys/param.h>
   74 #include <sys/malloc.h>
   75 #include <sys/mbuf.h>
   76 #include <sys/errno.h>
   77 #include <sys/protosw.h>
   78 #include <sys/socket.h>
   79 #include <sys/socketvar.h>
   80 #include <sys/systm.h>
   81 #include <sys/kernel.h>
   82 #include <sys/proc.h>
   83 
   84 #include <net/if.h>
   85 #include <net/route.h>
   86 
   87 #include <netinet/in.h>
   88 #include <netinet/in_var.h>
   89 #include <netinet6/in6_var.h>
   90 #include <netinet/ip6.h>
   91 #include <netinet/icmp6.h>
   92 #include <netinet6/ip6_var.h>
   93 #include <netinet/in_pcb.h>
   94 #include <netinet6/nd6.h>
   95 #include <netinet6/ip6protosw.h>
   96 
   97 #ifdef IPSEC
   98 #include <netinet6/ipsec.h>
   99 #ifdef INET6
  100 #include <netinet6/ipsec6.h>
  101 #endif
  102 #include <netkey/key.h>
  103 #endif /* IPSEC */
  104 
  105 #ifdef FAST_IPSEC
  106 #include <netipsec/ipsec.h>
  107 #include <netipsec/ipsec6.h>
  108 #include <netipsec/key.h>
  109 #endif /* FAST_IPSEC */
  110 
  111 #include <netinet6/ip6_fw.h>
  112 
  113 #include <net/net_osdep.h>
  114 
  115 extern int (*fr_checkp) __P((struct ip *, int, struct ifnet *, int, struct mbuf **));
  116 
  117 static MALLOC_DEFINE(M_IPMOPTS, "ip6_moptions", "internet multicast options");
  118 
  119 struct ip6_exthdrs {
  120         struct mbuf *ip6e_ip6;
  121         struct mbuf *ip6e_hbh;
  122         struct mbuf *ip6e_dest1;
  123         struct mbuf *ip6e_rthdr;
  124         struct mbuf *ip6e_dest2;
  125 };
  126 
  127 static int ip6_pcbopts __P((struct ip6_pktopts **, struct mbuf *,
  128                             struct socket *, struct sockopt *sopt));
  129 static int ip6_setmoptions __P((int, struct ip6_moptions **, struct mbuf *));
  130 static int ip6_getmoptions __P((int, struct ip6_moptions *, struct mbuf **));
  131 static int ip6_copyexthdr __P((struct mbuf **, caddr_t, int));
  132 static int ip6_insertfraghdr __P((struct mbuf *, struct mbuf *, int,
  133                                   struct ip6_frag **));
  134 static int ip6_insert_jumboopt __P((struct ip6_exthdrs *, u_int32_t));
  135 static int ip6_splithdr __P((struct mbuf *, struct ip6_exthdrs *));
  136 static int ip6_getpmtu __P((struct route_in6 *, struct route_in6 *,
  137         struct ifnet *, struct in6_addr *, u_long *, int *));
  138 
  139 /*
  140  * IP6 output. The packet in mbuf chain m contains a skeletal IP6
  141  * header (with pri, len, nxt, hlim, src, dst).
  142  * This function may modify ver and hlim only.
  143  * The mbuf chain containing the packet will be freed.
  144  * The mbuf opt, if present, will not be freed.
  145  *
  146  * type of "mtu": rt_rmx.rmx_mtu is u_long, ifnet.ifr_mtu is int, and
  147  * nd_ifinfo.linkmtu is u_int32_t.  so we use u_long to hold largest one,
  148  * which is rt_rmx.rmx_mtu.
  149  */
  150 int
  151 ip6_output(m0, opt, ro, flags, im6o, ifpp, inp)
  152         struct mbuf *m0;
  153         struct ip6_pktopts *opt;
  154         struct route_in6 *ro;
  155         int flags;
  156         struct ip6_moptions *im6o;
  157         struct ifnet **ifpp;            /* XXX: just for statistics */
  158         struct inpcb *inp;
  159 {
  160         struct ip6_hdr *ip6, *mhip6;
  161         struct ifnet *ifp, *origifp;
  162         struct mbuf *m = m0;
  163         int hlen, tlen, len, off;
  164         struct route_in6 ip6route;
  165         struct sockaddr_in6 *dst;
  166         int error = 0;
  167         struct in6_ifaddr *ia = NULL;
  168         u_long mtu;
  169         int alwaysfrag, dontfrag;
  170         u_int32_t optlen = 0, plen = 0, unfragpartlen = 0;
  171         struct ip6_exthdrs exthdrs;
  172         struct in6_addr finaldst;
  173         struct route_in6 *ro_pmtu = NULL;
  174         int hdrsplit = 0;
  175         int needipsec = 0;
  176 #ifdef IPSEC
  177         int needipsectun = 0;
  178         struct secpolicy *sp = NULL;
  179         struct socket *so = inp ? inp->inp_socket : NULL;
  180 
  181         ip6 = mtod(m, struct ip6_hdr *);
  182 #endif /* IPSEC */
  183 #ifdef FAST_IPSEC
  184         int needipsectun = 0;
  185         struct secpolicy *sp = NULL;
  186 
  187         ip6 = mtod(m, struct ip6_hdr *);
  188 #endif /* FAST_IPSEC */
  189 
  190 #define MAKE_EXTHDR(hp, mp)                                             \
  191     do {                                                                \
  192         if (hp) {                                                       \
  193                 struct ip6_ext *eh = (struct ip6_ext *)(hp);            \
  194                 error = ip6_copyexthdr((mp), (caddr_t)(hp),             \
  195                                        ((eh)->ip6e_len + 1) << 3);      \
  196                 if (error)                                              \
  197                         goto freehdrs;                                  \
  198         }                                                               \
  199     } while (0)
  200         
  201         bzero(&exthdrs, sizeof(exthdrs));
  202         
  203         if (opt) {
  204                 /* Hop-by-Hop options header */
  205                 MAKE_EXTHDR(opt->ip6po_hbh, &exthdrs.ip6e_hbh);
  206                 /* Destination options header(1st part) */
  207                 MAKE_EXTHDR(opt->ip6po_dest1, &exthdrs.ip6e_dest1);
  208                 /* Routing header */
  209                 MAKE_EXTHDR(opt->ip6po_rthdr, &exthdrs.ip6e_rthdr);
  210                 /* Destination options header(2nd part) */
  211                 MAKE_EXTHDR(opt->ip6po_dest2, &exthdrs.ip6e_dest2);
  212         }
  213 
  214 #ifdef IPSEC
  215         /* get a security policy for this packet */
  216         if (so == NULL)
  217                 sp = ipsec6_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND, 0, &error);
  218         else
  219                 sp = ipsec6_getpolicybysock(m, IPSEC_DIR_OUTBOUND, so, &error);
  220 
  221         if (sp == NULL) {
  222                 ipsec6stat.out_inval++;
  223                 goto freehdrs;
  224         }
  225 
  226         error = 0;
  227 
  228         /* check policy */
  229         switch (sp->policy) {
  230         case IPSEC_POLICY_DISCARD:
  231                 /*
  232                  * This packet is just discarded.
  233                  */
  234                 ipsec6stat.out_polvio++;
  235                 goto freehdrs;
  236 
  237         case IPSEC_POLICY_BYPASS:
  238         case IPSEC_POLICY_NONE:
  239                 /* no need to do IPsec. */
  240                 needipsec = 0;
  241                 break;
  242         
  243         case IPSEC_POLICY_IPSEC:
  244                 if (sp->req == NULL) {
  245                         /* acquire a policy */
  246                         error = key_spdacquire(sp);
  247                         goto freehdrs;
  248                 }
  249                 needipsec = 1;
  250                 break;
  251 
  252         case IPSEC_POLICY_ENTRUST:
  253         default:
  254                 printf("ip6_output: Invalid policy found. %d\n", sp->policy);
  255         }
  256 #endif /* IPSEC */
  257 #ifdef FAST_IPSEC
  258         /* get a security policy for this packet */
  259         if (inp == NULL)
  260                 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND, 0, &error);
  261         else
  262                 sp = ipsec_getpolicybysock(m, IPSEC_DIR_OUTBOUND, inp, &error);
  263 
  264         if (sp == NULL) {
  265                 newipsecstat.ips_out_inval++;
  266                 goto freehdrs;
  267         }
  268 
  269         error = 0;
  270 
  271         /* check policy */
  272         switch (sp->policy) {
  273         case IPSEC_POLICY_DISCARD:
  274                 /*
  275                  * This packet is just discarded.
  276                  */
  277                 newipsecstat.ips_out_polvio++;
  278                 goto freehdrs;
  279 
  280         case IPSEC_POLICY_BYPASS:
  281         case IPSEC_POLICY_NONE:
  282                 /* no need to do IPsec. */
  283                 needipsec = 0;
  284                 break;
  285         
  286         case IPSEC_POLICY_IPSEC:
  287                 if (sp->req == NULL) {
  288                         /* acquire a policy */
  289                         error = key_spdacquire(sp);
  290                         goto freehdrs;
  291                 }
  292                 needipsec = 1;
  293                 break;
  294 
  295         case IPSEC_POLICY_ENTRUST:
  296         default:
  297                 printf("ip6_output: Invalid policy found. %d\n", sp->policy);
  298         }
  299 #endif /* FAST_IPSEC */
  300 
  301         /*
  302          * Calculate the total length of the extension header chain.
  303          * Keep the length of the unfragmentable part for fragmentation.
  304          */
  305         optlen = 0;
  306         if (exthdrs.ip6e_hbh) optlen += exthdrs.ip6e_hbh->m_len;
  307         if (exthdrs.ip6e_dest1) optlen += exthdrs.ip6e_dest1->m_len;
  308         if (exthdrs.ip6e_rthdr) optlen += exthdrs.ip6e_rthdr->m_len;
  309         unfragpartlen = optlen + sizeof(struct ip6_hdr);
  310         /* NOTE: we don't add AH/ESP length here. do that later. */
  311         if (exthdrs.ip6e_dest2) optlen += exthdrs.ip6e_dest2->m_len;
  312 
  313         /*
  314          * If we need IPsec, or there is at least one extension header,
  315          * separate IP6 header from the payload.
  316          */
  317         if ((needipsec || optlen) && !hdrsplit) {
  318                 if ((error = ip6_splithdr(m, &exthdrs)) != 0) {
  319                         m = NULL;
  320                         goto freehdrs;
  321                 }
  322                 m = exthdrs.ip6e_ip6;
  323                 hdrsplit++;
  324         }
  325 
  326         /* adjust pointer */
  327         ip6 = mtod(m, struct ip6_hdr *);
  328 
  329         /* adjust mbuf packet header length */
  330         m->m_pkthdr.len += optlen;
  331         plen = m->m_pkthdr.len - sizeof(*ip6);
  332 
  333         /* If this is a jumbo payload, insert a jumbo payload option. */
  334         if (plen > IPV6_MAXPACKET) {
  335                 if (!hdrsplit) {
  336                         if ((error = ip6_splithdr(m, &exthdrs)) != 0) {
  337                                 m = NULL;
  338                                 goto freehdrs;
  339                         }
  340                         m = exthdrs.ip6e_ip6;
  341                         hdrsplit++;
  342                 }
  343                 /* adjust pointer */
  344                 ip6 = mtod(m, struct ip6_hdr *);
  345                 if ((error = ip6_insert_jumboopt(&exthdrs, plen)) != 0)
  346                         goto freehdrs;
  347                 ip6->ip6_plen = 0;
  348         } else
  349                 ip6->ip6_plen = htons(plen);
  350 
  351         /*
  352          * Concatenate headers and fill in next header fields.
  353          * Here we have, on "m"
  354          *      IPv6 payload
  355          * and we insert headers accordingly.  Finally, we should be getting:
  356          *      IPv6 hbh dest1 rthdr ah* [esp* dest2 payload]
  357          *
  358          * during the header composing process, "m" points to IPv6 header.
  359          * "mprev" points to an extension header prior to esp.
  360          */
  361         {
  362                 u_char *nexthdrp = &ip6->ip6_nxt;
  363                 struct mbuf *mprev = m;
  364 
  365                 /*
  366                  * we treat dest2 specially.  this makes IPsec processing
  367                  * much easier.  the goal here is to make mprev point the
  368                  * mbuf prior to dest2.
  369                  *
  370                  * result: IPv6 dest2 payload
  371                  * m and mprev will point to IPv6 header.
  372                  */
  373                 if (exthdrs.ip6e_dest2) {
  374                         if (!hdrsplit)
  375                                 panic("assumption failed: hdr not split");
  376                         exthdrs.ip6e_dest2->m_next = m->m_next;
  377                         m->m_next = exthdrs.ip6e_dest2;
  378                         *mtod(exthdrs.ip6e_dest2, u_char *) = ip6->ip6_nxt;
  379                         ip6->ip6_nxt = IPPROTO_DSTOPTS;
  380                 }
  381 
  382 #define MAKE_CHAIN(m, mp, p, i)\
  383     do {\
  384         if (m) {\
  385                 if (!hdrsplit) \
  386                         panic("assumption failed: hdr not split"); \
  387                 *mtod((m), u_char *) = *(p);\
  388                 *(p) = (i);\
  389                 p = mtod((m), u_char *);\
  390                 (m)->m_next = (mp)->m_next;\
  391                 (mp)->m_next = (m);\
  392                 (mp) = (m);\
  393         }\
  394     } while (0)
  395                 /*
  396                  * result: IPv6 hbh dest1 rthdr dest2 payload
  397                  * m will point to IPv6 header.  mprev will point to the
  398                  * extension header prior to dest2 (rthdr in the above case).
  399                  */
  400                 MAKE_CHAIN(exthdrs.ip6e_hbh, mprev,
  401                            nexthdrp, IPPROTO_HOPOPTS);
  402                 MAKE_CHAIN(exthdrs.ip6e_dest1, mprev,
  403                            nexthdrp, IPPROTO_DSTOPTS);
  404                 MAKE_CHAIN(exthdrs.ip6e_rthdr, mprev,
  405                            nexthdrp, IPPROTO_ROUTING);
  406 
  407 #if defined(IPSEC) || defined(FAST_IPSEC)
  408                 if (!needipsec)
  409                         goto skip_ipsec2;
  410 
  411                 /*
  412                  * pointers after IPsec headers are not valid any more.
  413                  * other pointers need a great care too.
  414                  * (IPsec routines should not mangle mbufs prior to AH/ESP)
  415                  */
  416                 exthdrs.ip6e_dest2 = NULL;
  417 
  418             {
  419                 struct ip6_rthdr *rh = NULL;
  420                 int segleft_org = 0;
  421                 struct ipsec_output_state state;
  422 
  423                 if (exthdrs.ip6e_rthdr) {
  424                         rh = mtod(exthdrs.ip6e_rthdr, struct ip6_rthdr *);
  425                         segleft_org = rh->ip6r_segleft;
  426                         rh->ip6r_segleft = 0;
  427                 }
  428 
  429                 bzero(&state, sizeof(state));
  430                 state.m = m;
  431                 error = ipsec6_output_trans(&state, nexthdrp, mprev, sp, flags,
  432                         &needipsectun);
  433                 m = state.m;
  434                 if (error) {
  435                         /* mbuf is already reclaimed in ipsec6_output_trans. */
  436                         m = NULL;
  437                         switch (error) {
  438                         case EHOSTUNREACH:
  439                         case ENETUNREACH:
  440                         case EMSGSIZE:
  441                         case ENOBUFS:
  442                         case ENOMEM:
  443                                 break;
  444                         default:
  445                                 printf("ip6_output (ipsec): error code %d\n", error);
  446                                 /* fall through */
  447                         case ENOENT:
  448                                 /* don't show these error codes to the user */
  449                                 error = 0;
  450                                 break;
  451                         }
  452                         goto bad;
  453                 }
  454                 if (exthdrs.ip6e_rthdr) {
  455                         /* ah6_output doesn't modify mbuf chain */
  456                         rh->ip6r_segleft = segleft_org;
  457                 }
  458             }
  459 skip_ipsec2:;
  460 #endif
  461         }
  462 
  463         /*
  464          * If there is a routing header, replace destination address field
  465          * with the first hop of the routing header.
  466          */
  467         if (exthdrs.ip6e_rthdr) {
  468                 struct ip6_rthdr *rh =
  469                         (struct ip6_rthdr *)(mtod(exthdrs.ip6e_rthdr,
  470                                                   struct ip6_rthdr *));
  471                 struct ip6_rthdr0 *rh0;
  472 
  473                 finaldst = ip6->ip6_dst;
  474                 switch (rh->ip6r_type) {
  475                 case IPV6_RTHDR_TYPE_0:
  476                          rh0 = (struct ip6_rthdr0 *)rh;
  477                          ip6->ip6_dst = rh0->ip6r0_addr[0];
  478                          bcopy((caddr_t)&rh0->ip6r0_addr[1],
  479                                (caddr_t)&rh0->ip6r0_addr[0],
  480                                sizeof(struct in6_addr)*(rh0->ip6r0_segleft - 1)
  481                                  );
  482                          rh0->ip6r0_addr[rh0->ip6r0_segleft - 1] = finaldst;
  483                          break;
  484                 default:        /* is it possible? */
  485                          error = EINVAL;
  486                          goto bad;
  487                 }
  488         }
  489 
  490         /* Source address validation */
  491         if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src) &&
  492             (flags & IPV6_DADOUTPUT) == 0) {
  493                 error = EOPNOTSUPP;
  494                 ip6stat.ip6s_badscope++;
  495                 goto bad;
  496         }
  497         if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src)) {
  498                 error = EOPNOTSUPP;
  499                 ip6stat.ip6s_badscope++;
  500                 goto bad;
  501         }
  502 
  503         ip6stat.ip6s_localout++;
  504 
  505         /*
  506          * Route packet.
  507          */
  508         if (ro == 0) {
  509                 ro = &ip6route;
  510                 bzero((caddr_t)ro, sizeof(*ro));
  511         }
  512         ro_pmtu = ro;
  513         if (opt && opt->ip6po_rthdr)
  514                 ro = &opt->ip6po_route;
  515         dst = (struct sockaddr_in6 *)&ro->ro_dst;
  516         /*
  517          * If there is a cached route,
  518          * check that it is to the same destination
  519          * and is still up. If not, free it and try again.
  520          */
  521         if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 ||
  522                          dst->sin6_family != AF_INET6 ||
  523                          !IN6_ARE_ADDR_EQUAL(&dst->sin6_addr, &ip6->ip6_dst))) {
  524                 RTFREE(ro->ro_rt);
  525                 ro->ro_rt = (struct rtentry *)0;
  526         }
  527         if (ro->ro_rt == 0) {
  528                 bzero(dst, sizeof(*dst));
  529                 dst->sin6_family = AF_INET6;
  530                 dst->sin6_len = sizeof(struct sockaddr_in6);
  531                 dst->sin6_addr = ip6->ip6_dst;
  532 #ifdef SCOPEDROUTING
  533                 /* XXX: sin6_scope_id should already be fixed at this point */
  534                 if (IN6_IS_SCOPE_LINKLOCAL(&dst->sin6_addr))
  535                         dst->sin6_scope_id = ntohs(dst->sin6_addr.s6_addr16[1]);
  536 #endif
  537         }
  538 #if defined(IPSEC) || defined(FAST_IPSEC)
  539         if (needipsec && needipsectun) {
  540                 struct ipsec_output_state state;
  541 
  542                 /*
  543                  * All the extension headers will become inaccessible
  544                  * (since they can be encrypted).
  545                  * Don't panic, we need no more updates to extension headers
  546                  * on inner IPv6 packet (since they are now encapsulated).
  547                  *
  548                  * IPv6 [ESP|AH] IPv6 [extension headers] payload
  549                  */
  550                 bzero(&exthdrs, sizeof(exthdrs));
  551                 exthdrs.ip6e_ip6 = m;
  552 
  553                 bzero(&state, sizeof(state));
  554                 state.m = m;
  555                 state.ro = (struct route *)ro;
  556                 state.dst = (struct sockaddr *)dst;
  557 
  558                 error = ipsec6_output_tunnel(&state, sp, flags);
  559 
  560                 m = state.m;
  561                 ro = (struct route_in6 *)state.ro;
  562                 dst = (struct sockaddr_in6 *)state.dst;
  563                 if (error) {
  564                         /* mbuf is already reclaimed in ipsec6_output_tunnel. */
  565                         m0 = m = NULL;
  566                         m = NULL;
  567                         switch (error) {
  568                         case EHOSTUNREACH:
  569                         case ENETUNREACH:
  570                         case EMSGSIZE:
  571                         case ENOBUFS:
  572                         case ENOMEM:
  573                                 break;
  574                         default:
  575                                 printf("ip6_output (ipsec): error code %d\n", error);
  576                                 /* fall through */
  577                         case ENOENT:
  578                                 /* don't show these error codes to the user */
  579                                 error = 0;
  580                                 break;
  581                         }
  582                         goto bad;
  583                 }
  584 
  585                 exthdrs.ip6e_ip6 = m;
  586         }
  587 #endif /* IPSEC */
  588 
  589         if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
  590                 /* Unicast */
  591 
  592 #define ifatoia6(ifa)   ((struct in6_ifaddr *)(ifa))
  593 #define sin6tosa(sin6)  ((struct sockaddr *)(sin6))
  594                 /* xxx
  595                  * interface selection comes here
  596                  * if an interface is specified from an upper layer,
  597                  * ifp must point it.
  598                  */
  599                 if (ro->ro_rt == 0) {
  600                         /*
  601                          * non-bsdi always clone routes, if parent is
  602                          * PRF_CLONING.
  603                          */
  604                         rtalloc((struct route *)ro);
  605                 }
  606                 if (ro->ro_rt == 0) {
  607                         ip6stat.ip6s_noroute++;
  608                         error = EHOSTUNREACH;
  609                         /* XXX in6_ifstat_inc(ifp, ifs6_out_discard); */
  610                         goto bad;
  611                 }
  612                 ia = ifatoia6(ro->ro_rt->rt_ifa);
  613                 ifp = ro->ro_rt->rt_ifp;
  614                 ro->ro_rt->rt_use++;
  615                 if (ro->ro_rt->rt_flags & RTF_GATEWAY)
  616                         dst = (struct sockaddr_in6 *)ro->ro_rt->rt_gateway;
  617                 m->m_flags &= ~(M_BCAST | M_MCAST);     /* just in case */
  618 
  619                 in6_ifstat_inc(ifp, ifs6_out_request);
  620 
  621                 /*
  622                  * Check if the outgoing interface conflicts with
  623                  * the interface specified by ifi6_ifindex (if specified).
  624                  * Note that loopback interface is always okay.
  625                  * (this may happen when we are sending a packet to one of
  626                  *  our own addresses.)
  627                  */
  628                 if (opt && opt->ip6po_pktinfo
  629                  && opt->ip6po_pktinfo->ipi6_ifindex) {
  630                         if (!(ifp->if_flags & IFF_LOOPBACK)
  631                          && ifp->if_index != opt->ip6po_pktinfo->ipi6_ifindex) {
  632                                 ip6stat.ip6s_noroute++;
  633                                 in6_ifstat_inc(ifp, ifs6_out_discard);
  634                                 error = EHOSTUNREACH;
  635                                 goto bad;
  636                         }
  637                 }
  638 
  639                 if (opt && opt->ip6po_hlim != -1)
  640                         ip6->ip6_hlim = opt->ip6po_hlim & 0xff;
  641         } else {
  642                 /* Multicast */
  643                 struct  in6_multi *in6m;
  644 
  645                 m->m_flags = (m->m_flags & ~M_BCAST) | M_MCAST;
  646 
  647                 /*
  648                  * See if the caller provided any multicast options
  649                  */
  650                 ifp = NULL;
  651                 if (im6o != NULL) {
  652                         ip6->ip6_hlim = im6o->im6o_multicast_hlim;
  653                         if (im6o->im6o_multicast_ifp != NULL)
  654                                 ifp = im6o->im6o_multicast_ifp;
  655                 } else
  656                         ip6->ip6_hlim = ip6_defmcasthlim;
  657 
  658                 /*
  659                  * See if the caller provided the outgoing interface
  660                  * as an ancillary data.
  661                  * Boundary check for ifindex is assumed to be already done.
  662                  */
  663                 if (opt && opt->ip6po_pktinfo && opt->ip6po_pktinfo->ipi6_ifindex)
  664                         ifp = ifindex2ifnet[opt->ip6po_pktinfo->ipi6_ifindex];
  665 
  666                 /*
  667                  * If the destination is a node-local scope multicast,
  668                  * the packet should be loop-backed only.
  669                  */
  670                 if (IN6_IS_ADDR_MC_NODELOCAL(&ip6->ip6_dst)) {
  671                         /*
  672                          * If the outgoing interface is already specified,
  673                          * it should be a loopback interface.
  674                          */
  675                         if (ifp && (ifp->if_flags & IFF_LOOPBACK) == 0) {
  676                                 ip6stat.ip6s_badscope++;
  677                                 error = ENETUNREACH; /* XXX: better error? */
  678                                 /* XXX correct ifp? */
  679                                 in6_ifstat_inc(ifp, ifs6_out_discard);
  680                                 goto bad;
  681                         } else {
  682                                 ifp = &loif[0];
  683                         }
  684                 }
  685 
  686                 if (opt && opt->ip6po_hlim != -1)
  687                         ip6->ip6_hlim = opt->ip6po_hlim & 0xff;
  688 
  689                 /*
  690                  * If caller did not provide an interface lookup a
  691                  * default in the routing table.  This is either a
  692                  * default for the speicfied group (i.e. a host
  693                  * route), or a multicast default (a route for the
  694                  * ``net'' ff00::/8).
  695                  */
  696                 if (ifp == NULL) {
  697                         if (ro->ro_rt == 0) {
  698                                 ro->ro_rt = rtalloc1((struct sockaddr *)
  699                                                 &ro->ro_dst, 0, 0UL);
  700                         }
  701                         if (ro->ro_rt == 0) {
  702                                 ip6stat.ip6s_noroute++;
  703                                 error = EHOSTUNREACH;
  704                                 /* XXX in6_ifstat_inc(ifp, ifs6_out_discard) */
  705                                 goto bad;
  706                         }
  707                         ia = ifatoia6(ro->ro_rt->rt_ifa);
  708                         ifp = ro->ro_rt->rt_ifp;
  709                         ro->ro_rt->rt_use++;
  710                 }
  711 
  712                 if ((flags & IPV6_FORWARDING) == 0)
  713                         in6_ifstat_inc(ifp, ifs6_out_request);
  714                 in6_ifstat_inc(ifp, ifs6_out_mcast);
  715 
  716                 /*
  717                  * Confirm that the outgoing interface supports multicast.
  718                  */
  719                 if ((ifp->if_flags & IFF_MULTICAST) == 0) {
  720                         ip6stat.ip6s_noroute++;
  721                         in6_ifstat_inc(ifp, ifs6_out_discard);
  722                         error = ENETUNREACH;
  723                         goto bad;
  724                 }
  725                 IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m);
  726                 if (in6m != NULL &&
  727                    (im6o == NULL || im6o->im6o_multicast_loop)) {
  728                         /*
  729                          * If we belong to the destination multicast group
  730                          * on the outgoing interface, and the caller did not
  731                          * forbid loopback, loop back a copy.
  732                          */
  733                         ip6_mloopback(ifp, m, dst);
  734                 } else {
  735                         /*
  736                          * If we are acting as a multicast router, perform
  737                          * multicast forwarding as if the packet had just
  738                          * arrived on the interface to which we are about
  739                          * to send.  The multicast forwarding function
  740                          * recursively calls this function, using the
  741                          * IPV6_FORWARDING flag to prevent infinite recursion.
  742                          *
  743                          * Multicasts that are looped back by ip6_mloopback(),
  744                          * above, will be forwarded by the ip6_input() routine,
  745                          * if necessary.
  746                          */
  747                         if (ip6_mrouter && (flags & IPV6_FORWARDING) == 0) {
  748                                 if (ip6_mforward(ip6, ifp, m) != 0) {
  749                                         m_freem(m);
  750                                         goto done;
  751                                 }
  752                         }
  753                 }
  754                 /*
  755                  * Multicasts with a hoplimit of zero may be looped back,
  756                  * above, but must not be transmitted on a network.
  757                  * Also, multicasts addressed to the loopback interface
  758                  * are not sent -- the above call to ip6_mloopback() will
  759                  * loop back a copy if this host actually belongs to the
  760                  * destination group on the loopback interface.
  761                  */
  762                 if (ip6->ip6_hlim == 0 || (ifp->if_flags & IFF_LOOPBACK)) {
  763                         m_freem(m);
  764                         goto done;
  765                 }
  766         }
  767 
  768         /*
  769          * Fill the outgoing inteface to tell the upper layer
  770          * to increment per-interface statistics.
  771          */
  772         if (ifpp)
  773                 *ifpp = ifp;
  774 
  775         /* Determine path MTU. */
  776         if ((error = ip6_getpmtu(ro_pmtu, ro, ifp, &finaldst, &mtu,
  777             &alwaysfrag)) != 0)
  778                 goto bad;
  779 
  780         /*
  781          * The caller of this function may specify to use the minimum MTU
  782          * in some cases.
  783          */
  784         if (mtu > IPV6_MMTU) {
  785                 if ((flags & IPV6_MINMTU))
  786                         mtu = IPV6_MMTU;
  787         }
  788 
  789         /* Fake scoped addresses */
  790         if ((ifp->if_flags & IFF_LOOPBACK) != 0) {
  791                 /*
  792                  * If source or destination address is a scoped address, and
  793                  * the packet is going to be sent to a loopback interface,
  794                  * we should keep the original interface.
  795                  */
  796 
  797                 /*
  798                  * XXX: this is a very experimental and temporary solution.
  799                  * We eventually have sockaddr_in6 and use the sin6_scope_id
  800                  * field of the structure here.
  801                  * We rely on the consistency between two scope zone ids
  802                  * of source and destination, which should already be assured.
  803                  * Larger scopes than link will be supported in the future. 
  804                  */
  805                 origifp = NULL;
  806                 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
  807                         origifp = ifindex2ifnet[ntohs(ip6->ip6_src.s6_addr16[1])];
  808                 else if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst))
  809                         origifp = ifindex2ifnet[ntohs(ip6->ip6_dst.s6_addr16[1])];
  810                 /*
  811                  * XXX: origifp can be NULL even in those two cases above.
  812                  * For example, if we remove the (only) link-local address
  813                  * from the loopback interface, and try to send a link-local
  814                  * address without link-id information.  Then the source
  815                  * address is ::1, and the destination address is the
  816                  * link-local address with its s6_addr16[1] being zero.
  817                  * What is worse, if the packet goes to the loopback interface
  818                  * by a default rejected route, the null pointer would be
  819                  * passed to looutput, and the kernel would hang.
  820                  * The following last resort would prevent such disaster.
  821                  */
  822                 if (origifp == NULL)
  823                         origifp = ifp;
  824         }
  825         else
  826                 origifp = ifp;
  827 #ifndef SCOPEDROUTING
  828         /*
  829          * clear embedded scope identifiers if necessary.
  830          * in6_clearscope will touch the addresses only when necessary.
  831          */
  832         in6_clearscope(&ip6->ip6_src);
  833         in6_clearscope(&ip6->ip6_dst);
  834 #endif
  835 
  836         /*
  837          * Check with the firewall...
  838          */
  839         if (ip6_fw_enable && ip6_fw_chk_ptr) {
  840                 u_short port = 0;
  841                 m->m_pkthdr.rcvif = NULL;       /* XXX */
  842                 /* If ipfw says divert, we have to just drop packet */
  843                 if ((*ip6_fw_chk_ptr)(&ip6, ifp, &port, &m)) {
  844                         m_freem(m);
  845                         goto done;
  846                 }
  847                 if (!m) {
  848                         error = EACCES;
  849                         goto done;
  850                 }
  851         }
  852 
  853         /*
  854          * If the outgoing packet contains a hop-by-hop options header,
  855          * it must be examined and processed even by the source node.
  856          * (RFC 2460, section 4.)
  857          */
  858         if (exthdrs.ip6e_hbh) {
  859                 struct ip6_hbh *hbh = mtod(exthdrs.ip6e_hbh, struct ip6_hbh *);
  860                 u_int32_t dummy1; /* XXX unused */
  861                 u_int32_t dummy2; /* XXX unused */
  862 
  863 #ifdef DIAGNOSTIC
  864                 if ((hbh->ip6h_len + 1) << 3 > exthdrs.ip6e_hbh->m_len)
  865                         panic("ip6e_hbh is not continuous");
  866 #endif
  867                 /*
  868                  *  XXX: if we have to send an ICMPv6 error to the sender,
  869                  *       we need the M_LOOP flag since icmp6_error() expects
  870                  *       the IPv6 and the hop-by-hop options header are
  871                  *       continuous unless the flag is set.
  872                  */
  873                 m->m_flags |= M_LOOP;
  874                 m->m_pkthdr.rcvif = ifp;
  875                 if (ip6_process_hopopts(m,
  876                                         (u_int8_t *)(hbh + 1),
  877                                         ((hbh->ip6h_len + 1) << 3) -
  878                                         sizeof(struct ip6_hbh),
  879                                         &dummy1, &dummy2) < 0) {
  880                         /* m was already freed at this point */
  881                         error = EINVAL;/* better error? */
  882                         goto done;
  883                 }
  884                 m->m_flags &= ~M_LOOP; /* XXX */
  885                 m->m_pkthdr.rcvif = NULL;
  886         }
  887 
  888         /*
  889          * Check if we want to allow this packet to be processed.
  890          * Consider it to be bad if not.
  891          */
  892         if (fr_checkp) {
  893                 struct  mbuf    *m1 = m;
  894 
  895                 if ((*fr_checkp)((struct ip *)ip6, sizeof(*ip6), ifp, 1, &m1))
  896                         goto done;
  897                 m = m1;
  898                 if (m == NULL)
  899                         goto done;
  900                 ip6 = mtod(m, struct ip6_hdr *);
  901         }
  902 
  903         /*
  904          * Send the packet to the outgoing interface.
  905          * If necessary, do IPv6 fragmentation before sending.
  906          *
  907          * the logic here is rather complex:
  908          * 1: normal case (dontfrag == 0, alwaysfrag == 0)
  909          * 1-a: send as is if tlen <= path mtu
  910          * 1-b: fragment if tlen > path mtu
  911          *
  912          * 2: if user asks us not to fragment (dontfrag == 1)
  913          * 2-a: send as is if tlen <= interface mtu
  914          * 2-b: error if tlen > interface mtu
  915          *
  916          * 3: if we always need to attach fragment header (alwaysfrag == 1)
  917          *      always fragment
  918          *
  919          * 4: if dontfrag == 1 && alwaysfrag == 1
  920          *      error, as we cannot handle this conflicting request
  921          */
  922         tlen = m->m_pkthdr.len;
  923 
  924         dontfrag = 0;
  925         if (dontfrag && alwaysfrag) {   /* case 4 */
  926                 /* conflicting request - can't transmit */
  927                 error = EMSGSIZE;
  928                 goto bad;
  929         }
  930         if (dontfrag && tlen > nd_ifinfo[ifp->if_index].linkmtu) {      /* case 2-b */
  931                 /*
  932                  * Even if the DONTFRAG option is specified, we cannot send the
  933                  * packet when the data length is larger than the MTU of the
  934                  * outgoing interface.
  935                  * Notify the error by sending IPV6_PATHMTU ancillary data as
  936                  * well as returning an error code (the latter is not described
  937                  * in the API spec.)
  938                  */
  939                 u_int32_t mtu32;
  940                 struct ip6ctlparam ip6cp;
  941 
  942                 mtu32 = (u_int32_t)mtu;
  943                 bzero(&ip6cp, sizeof(ip6cp));
  944                 ip6cp.ip6c_cmdarg = (void *)&mtu32;
  945                 pfctlinput2(PRC_MSGSIZE, (struct sockaddr *)&ro_pmtu->ro_dst,
  946                     (void *)&ip6cp);
  947 
  948                 error = EMSGSIZE;
  949                 goto bad;
  950         }
  951 
  952         /*
  953          * transmit packet without fragmentation
  954          */
  955         if (dontfrag || (!alwaysfrag && tlen <= mtu)) { /* case 1-a and 2-a */
  956                 struct in6_ifaddr *ia6;
  957 
  958                 ip6 = mtod(m, struct ip6_hdr *);
  959                 ia6 = in6_ifawithifp(ifp, &ip6->ip6_src);
  960                 if (ia6) {
  961                         /* Record statistics for this interface address. */
  962                         ia6->ia_ifa.if_opackets++;
  963                         ia6->ia_ifa.if_obytes += m->m_pkthdr.len;
  964                 }
  965 #ifdef IPSEC
  966                 /* clean ipsec history once it goes out of the node */
  967                 ipsec_delaux(m);
  968 #endif
  969                 error = nd6_output(ifp, origifp, m, dst, ro->ro_rt);
  970                 goto done;
  971         }
  972 
  973         /*
  974          * try to fragment the packet.  case 1-b and 3
  975          */
  976         if (mtu < IPV6_MMTU) {
  977                 /* path MTU cannot be less than IPV6_MMTU */
  978                 error = EMSGSIZE;
  979                 in6_ifstat_inc(ifp, ifs6_out_fragfail);
  980                 goto bad;
  981         } else if (ip6->ip6_plen == 0) {
  982                 /* jumbo payload cannot be fragmented */
  983                 error = EMSGSIZE;
  984                 in6_ifstat_inc(ifp, ifs6_out_fragfail);
  985                 goto bad;
  986         } else {
  987                 struct mbuf **mnext, *m_frgpart;
  988                 struct ip6_frag *ip6f;
  989                 u_int32_t id = htonl(ip6_id++);
  990                 u_char nextproto;
  991                 struct ip6ctlparam ip6cp;
  992                 u_int32_t mtu32;
  993 
  994                 /*
  995                  * Too large for the destination or interface;
  996                  * fragment if possible.
  997                  * Must be able to put at least 8 bytes per fragment.
  998                  */
  999                 hlen = unfragpartlen;
 1000                 if (mtu > IPV6_MAXPACKET)
 1001                         mtu = IPV6_MAXPACKET;
 1002 
 1003                 /* Notify a proper path MTU to applications. */
 1004                 mtu32 = (u_int32_t)mtu;
 1005                 bzero(&ip6cp, sizeof(ip6cp));
 1006                 ip6cp.ip6c_cmdarg = (void *)&mtu32;
 1007                 pfctlinput2(PRC_MSGSIZE, (struct sockaddr *)&ro_pmtu->ro_dst,
 1008                     (void *)&ip6cp);
 1009 
 1010                 len = (mtu - hlen - sizeof(struct ip6_frag)) & ~7;
 1011                 if (len < 8) {
 1012                         error = EMSGSIZE;
 1013                         in6_ifstat_inc(ifp, ifs6_out_fragfail);
 1014                         goto bad;
 1015                 }
 1016 
 1017                 mnext = &m->m_nextpkt;
 1018 
 1019                 /*
 1020                  * Change the next header field of the last header in the
 1021                  * unfragmentable part.
 1022                  */
 1023                 if (exthdrs.ip6e_rthdr) {
 1024                         nextproto = *mtod(exthdrs.ip6e_rthdr, u_char *);
 1025                         *mtod(exthdrs.ip6e_rthdr, u_char *) = IPPROTO_FRAGMENT;
 1026                 } else if (exthdrs.ip6e_dest1) {
 1027                         nextproto = *mtod(exthdrs.ip6e_dest1, u_char *);
 1028                         *mtod(exthdrs.ip6e_dest1, u_char *) = IPPROTO_FRAGMENT;
 1029                 } else if (exthdrs.ip6e_hbh) {
 1030                         nextproto = *mtod(exthdrs.ip6e_hbh, u_char *);
 1031                         *mtod(exthdrs.ip6e_hbh, u_char *) = IPPROTO_FRAGMENT;
 1032                 } else {
 1033                         nextproto = ip6->ip6_nxt;
 1034                         ip6->ip6_nxt = IPPROTO_FRAGMENT;
 1035                 }
 1036 
 1037                 /*
 1038                  * Loop through length of segment after first fragment,
 1039                  * make new header and copy data of each part and link onto
 1040                  * chain.
 1041                  */
 1042                 m0 = m;
 1043                 for (off = hlen; off < tlen; off += len) {
 1044                         MGETHDR(m, M_DONTWAIT, MT_HEADER);
 1045                         if (!m) {
 1046                                 error = ENOBUFS;
 1047                                 ip6stat.ip6s_odropped++;
 1048                                 goto sendorfree;
 1049                         }
 1050                         m->m_pkthdr.rcvif = NULL;
 1051                         m->m_flags = m0->m_flags & M_COPYFLAGS;
 1052                         *mnext = m;
 1053                         mnext = &m->m_nextpkt;
 1054                         m->m_data += max_linkhdr;
 1055                         mhip6 = mtod(m, struct ip6_hdr *);
 1056                         *mhip6 = *ip6;
 1057                         m->m_len = sizeof(*mhip6);
 1058                         error = ip6_insertfraghdr(m0, m, hlen, &ip6f);
 1059                         if (error) {
 1060                                 ip6stat.ip6s_odropped++;
 1061                                 goto sendorfree;
 1062                         }
 1063                         ip6f->ip6f_offlg = htons((u_short)((off - hlen) & ~7));
 1064                         if (off + len >= tlen)
 1065                                 len = tlen - off;
 1066                         else
 1067                                 ip6f->ip6f_offlg |= IP6F_MORE_FRAG;
 1068                         mhip6->ip6_plen = htons((u_short)(len + hlen +
 1069                                                           sizeof(*ip6f) -
 1070                                                           sizeof(struct ip6_hdr)));
 1071                         if ((m_frgpart = m_copy(m0, off, len)) == 0) {
 1072                                 error = ENOBUFS;
 1073                                 ip6stat.ip6s_odropped++;
 1074                                 goto sendorfree;
 1075                         }
 1076                         m_cat(m, m_frgpart);
 1077                         m->m_pkthdr.len = len + hlen + sizeof(*ip6f);
 1078                         m->m_pkthdr.rcvif = (struct ifnet *)0;
 1079                         ip6f->ip6f_reserved = 0;
 1080                         ip6f->ip6f_ident = id;
 1081                         ip6f->ip6f_nxt = nextproto;
 1082                         ip6stat.ip6s_ofragments++;
 1083                         in6_ifstat_inc(ifp, ifs6_out_fragcreat);
 1084                 }
 1085 
 1086                 in6_ifstat_inc(ifp, ifs6_out_fragok);
 1087         }
 1088 
 1089         /*
 1090          * Remove leading garbages.
 1091          */
 1092 sendorfree:
 1093         m = m0->m_nextpkt;
 1094         m0->m_nextpkt = 0;
 1095         m_freem(m0);
 1096         for (m0 = m; m; m = m0) {
 1097                 m0 = m->m_nextpkt;
 1098                 m->m_nextpkt = 0;
 1099                 if (error == 0) {
 1100                         /* Record statistics for this interface address. */
 1101                         if (ia) {
 1102                                 ia->ia_ifa.if_opackets++;
 1103                                 ia->ia_ifa.if_obytes += m->m_pkthdr.len;
 1104                         }
 1105 #ifdef IPSEC
 1106                         /* clean ipsec history once it goes out of the node */
 1107                         ipsec_delaux(m);
 1108 #endif
 1109                         error = nd6_output(ifp, origifp, m, dst, ro->ro_rt);
 1110                 } else
 1111                         m_freem(m);
 1112         }
 1113 
 1114         if (error == 0)
 1115                 ip6stat.ip6s_fragmented++;
 1116 
 1117 done:
 1118         if (ro == &ip6route && ro->ro_rt) { /* brace necessary for RTFREE */
 1119                 RTFREE(ro->ro_rt);
 1120         } else if (ro_pmtu == &ip6route && ro_pmtu->ro_rt) {
 1121                 RTFREE(ro_pmtu->ro_rt);
 1122         }
 1123 
 1124 #ifdef IPSEC
 1125         if (sp != NULL)
 1126                 key_freesp(sp);
 1127 #endif /* IPSEC */
 1128 #ifdef FAST_IPSEC
 1129         if (sp != NULL)
 1130                 KEY_FREESP(&sp);
 1131 #endif /* FAST_IPSEC */
 1132 
 1133         return(error);
 1134 
 1135 freehdrs:
 1136         m_freem(exthdrs.ip6e_hbh);      /* m_freem will check if mbuf is 0 */
 1137         m_freem(exthdrs.ip6e_dest1);
 1138         m_freem(exthdrs.ip6e_rthdr);
 1139         m_freem(exthdrs.ip6e_dest2);
 1140         /* fall through */
 1141 bad:
 1142         m_freem(m);
 1143         goto done;
 1144 }
 1145 
 1146 static int
 1147 ip6_copyexthdr(mp, hdr, hlen)
 1148         struct mbuf **mp;
 1149         caddr_t hdr;
 1150         int hlen;
 1151 {
 1152         struct mbuf *m;
 1153 
 1154         if (hlen > MCLBYTES)
 1155                 return(ENOBUFS); /* XXX */
 1156 
 1157         MGET(m, M_DONTWAIT, MT_DATA);
 1158         if (!m)
 1159                 return(ENOBUFS);
 1160 
 1161         if (hlen > MLEN) {
 1162                 MCLGET(m, M_DONTWAIT);
 1163                 if ((m->m_flags & M_EXT) == 0) {
 1164                         m_free(m);
 1165                         return(ENOBUFS);
 1166                 }
 1167         }
 1168         m->m_len = hlen;
 1169         if (hdr)
 1170                 bcopy(hdr, mtod(m, caddr_t), hlen);
 1171 
 1172         *mp = m;
 1173         return(0);
 1174 }
 1175 
 1176 /*
 1177  * Insert jumbo payload option.
 1178  */
 1179 static int
 1180 ip6_insert_jumboopt(exthdrs, plen)
 1181         struct ip6_exthdrs *exthdrs;
 1182         u_int32_t plen;
 1183 {
 1184         struct mbuf *mopt;
 1185         u_char *optbuf;
 1186         u_int32_t v;
 1187 
 1188 #define JUMBOOPTLEN     8       /* length of jumbo payload option and padding */
 1189 
 1190         /*
 1191          * If there is no hop-by-hop options header, allocate new one.
 1192          * If there is one but it doesn't have enough space to store the
 1193          * jumbo payload option, allocate a cluster to store the whole options.
 1194          * Otherwise, use it to store the options.
 1195          */
 1196         if (exthdrs->ip6e_hbh == 0) {
 1197                 MGET(mopt, M_DONTWAIT, MT_DATA);
 1198                 if (mopt == 0)
 1199                         return(ENOBUFS);
 1200                 mopt->m_len = JUMBOOPTLEN;
 1201                 optbuf = mtod(mopt, u_char *);
 1202                 optbuf[1] = 0;  /* = ((JUMBOOPTLEN) >> 3) - 1 */
 1203                 exthdrs->ip6e_hbh = mopt;
 1204         } else {
 1205                 struct ip6_hbh *hbh;
 1206 
 1207                 mopt = exthdrs->ip6e_hbh;
 1208                 if (M_TRAILINGSPACE(mopt) < JUMBOOPTLEN) {
 1209                         /*
 1210                          * XXX assumption:
 1211                          * - exthdrs->ip6e_hbh is not referenced from places
 1212                          *   other than exthdrs.
 1213                          * - exthdrs->ip6e_hbh is not an mbuf chain.
 1214                          */
 1215                         int oldoptlen = mopt->m_len;
 1216                         struct mbuf *n;
 1217 
 1218                         /*
 1219                          * XXX: give up if the whole (new) hbh header does
 1220                          * not fit even in an mbuf cluster.
 1221                          */
 1222                         if (oldoptlen + JUMBOOPTLEN > MCLBYTES)
 1223                                 return(ENOBUFS);
 1224 
 1225                         /*
 1226                          * As a consequence, we must always prepare a cluster
 1227                          * at this point.
 1228                          */
 1229                         MGET(n, M_DONTWAIT, MT_DATA);
 1230                         if (n) {
 1231                                 MCLGET(n, M_DONTWAIT);
 1232                                 if ((n->m_flags & M_EXT) == 0) {
 1233                                         m_freem(n);
 1234                                         n = NULL;
 1235                                 }
 1236                         }
 1237                         if (!n)
 1238                                 return(ENOBUFS);
 1239                         n->m_len = oldoptlen + JUMBOOPTLEN;
 1240                         bcopy(mtod(mopt, caddr_t), mtod(n, caddr_t),
 1241                               oldoptlen);
 1242                         optbuf = mtod(n, caddr_t) + oldoptlen;
 1243                         m_freem(mopt);
 1244                         mopt = exthdrs->ip6e_hbh = n;
 1245                 } else {
 1246                         optbuf = mtod(mopt, u_char *) + mopt->m_len;
 1247                         mopt->m_len += JUMBOOPTLEN;
 1248                 }
 1249                 optbuf[0] = IP6OPT_PADN;
 1250                 optbuf[1] = 1;
 1251 
 1252                 /*
 1253                  * Adjust the header length according to the pad and
 1254                  * the jumbo payload option.
 1255                  */
 1256                 hbh = mtod(mopt, struct ip6_hbh *);
 1257                 hbh->ip6h_len += (JUMBOOPTLEN >> 3);
 1258         }
 1259 
 1260         /* fill in the option. */
 1261         optbuf[2] = IP6OPT_JUMBO;
 1262         optbuf[3] = 4;
 1263         v = (u_int32_t)htonl(plen + JUMBOOPTLEN);
 1264         bcopy(&v, &optbuf[4], sizeof(u_int32_t));
 1265 
 1266         /* finally, adjust the packet header length */
 1267         exthdrs->ip6e_ip6->m_pkthdr.len += JUMBOOPTLEN;
 1268 
 1269         return(0);
 1270 #undef JUMBOOPTLEN
 1271 }
 1272 
 1273 /*
 1274  * Insert fragment header and copy unfragmentable header portions.
 1275  */
 1276 static int
 1277 ip6_insertfraghdr(m0, m, hlen, frghdrp)
 1278         struct mbuf *m0, *m;
 1279         int hlen;
 1280         struct ip6_frag **frghdrp;
 1281 {
 1282         struct mbuf *n, *mlast;
 1283 
 1284         if (hlen > sizeof(struct ip6_hdr)) {
 1285                 n = m_copym(m0, sizeof(struct ip6_hdr),
 1286                             hlen - sizeof(struct ip6_hdr), M_DONTWAIT);
 1287                 if (n == 0)
 1288                         return(ENOBUFS);
 1289                 m->m_next = n;
 1290         } else
 1291                 n = m;
 1292 
 1293         /* Search for the last mbuf of unfragmentable part. */
 1294         for (mlast = n; mlast->m_next; mlast = mlast->m_next)
 1295                 ;
 1296 
 1297         if ((mlast->m_flags & M_EXT) == 0 &&
 1298             M_TRAILINGSPACE(mlast) >= sizeof(struct ip6_frag)) {
 1299                 /* use the trailing space of the last mbuf for the fragment hdr */
 1300                 *frghdrp =
 1301                         (struct ip6_frag *)(mtod(mlast, caddr_t) + mlast->m_len);
 1302                 mlast->m_len += sizeof(struct ip6_frag);
 1303                 m->m_pkthdr.len += sizeof(struct ip6_frag);
 1304         } else {
 1305                 /* allocate a new mbuf for the fragment header */
 1306                 struct mbuf *mfrg;
 1307 
 1308                 MGET(mfrg, M_DONTWAIT, MT_DATA);
 1309                 if (mfrg == 0)
 1310                         return(ENOBUFS);
 1311                 mfrg->m_len = sizeof(struct ip6_frag);
 1312                 *frghdrp = mtod(mfrg, struct ip6_frag *);
 1313                 mlast->m_next = mfrg;
 1314         }
 1315 
 1316         return(0);
 1317 }
 1318 
 1319 static int
 1320 ip6_getpmtu(ro_pmtu, ro, ifp, dst, mtup, alwaysfragp)
 1321         struct route_in6 *ro_pmtu, *ro;
 1322         struct ifnet *ifp;
 1323         struct in6_addr *dst;
 1324         u_long *mtup;
 1325         int *alwaysfragp;
 1326 {
 1327         u_int32_t mtu = 0;
 1328         int alwaysfrag = 0;
 1329         int error = 0;
 1330 
 1331         if (ro_pmtu != ro) {
 1332                 /* The first hop and the final destination may differ. */
 1333                 struct sockaddr_in6 *sa6_dst =
 1334                         (struct sockaddr_in6 *)&ro_pmtu->ro_dst;
 1335                 if (ro_pmtu->ro_rt &&
 1336                     ((ro_pmtu->ro_rt->rt_flags & RTF_UP) == 0 ||
 1337                      !IN6_ARE_ADDR_EQUAL(&sa6_dst->sin6_addr, dst))) {
 1338                         RTFREE(ro_pmtu->ro_rt);
 1339                         ro_pmtu->ro_rt = (struct rtentry *)NULL;
 1340                 }
 1341                 if (ro_pmtu->ro_rt == NULL) {
 1342                         bzero(sa6_dst, sizeof(*sa6_dst));
 1343                         sa6_dst->sin6_family = AF_INET6;
 1344                         sa6_dst->sin6_len = sizeof(struct sockaddr_in6);
 1345                         sa6_dst->sin6_addr = *dst;
 1346 
 1347                         rtalloc((struct route *)ro_pmtu);
 1348                 }
 1349         }
 1350         if (ro_pmtu->ro_rt) {
 1351                 u_int32_t ifmtu;
 1352 
 1353                 if (ifp == NULL)
 1354                         ifp = ro_pmtu->ro_rt->rt_ifp;
 1355                 ifmtu = nd_ifinfo[ifp->if_index].linkmtu;
 1356                 mtu = ro_pmtu->ro_rt->rt_rmx.rmx_mtu;
 1357                 if (mtu == 0)
 1358                         mtu = ifmtu;
 1359                 else if (mtu < IPV6_MMTU) {
 1360                         /*
 1361                          * RFC2460 section 5, last paragraph:
 1362                          * if we record ICMPv6 too big message with
 1363                          * mtu < IPV6_MMTU, transmit packets sized IPV6_MMTU
 1364                          * or smaller, with fragment header attached.
 1365                          * (fragment header is needed regardless from the
 1366                          * packet size, for translators to identify packets)
 1367                          */
 1368                         alwaysfrag = 1;
 1369                         mtu = IPV6_MMTU;
 1370                 } else if (mtu > ifmtu) {
 1371                         /*
 1372                          * The MTU on the route is larger than the MTU on
 1373                          * the interface!  This shouldn't happen, unless the
 1374                          * MTU of the interface has been changed after the
 1375                          * interface was brought up.  Change the MTU in the
 1376                          * route to match the interface MTU (as long as the
 1377                          * field isn't locked).
 1378                          */
 1379                         mtu = ifmtu;
 1380                         if (!(ro_pmtu->ro_rt->rt_rmx.rmx_locks & RTV_MTU))
 1381                                 ro_pmtu->ro_rt->rt_rmx.rmx_mtu = mtu;
 1382                 }
 1383         } else if (ifp) {
 1384                 mtu = nd_ifinfo[ifp->if_index].linkmtu;
 1385         } else
 1386                 error = EHOSTUNREACH; /* XXX */
 1387 
 1388         *mtup = mtu;
 1389         if (alwaysfragp)
 1390                 *alwaysfragp = alwaysfrag;
 1391         return (error);
 1392 }
 1393 
 1394 /*
 1395  * IP6 socket option processing.
 1396  */
 1397 int
 1398 ip6_ctloutput(so, sopt)
 1399         struct socket *so;
 1400         struct sockopt *sopt;
 1401 {
 1402         int privileged;
 1403         struct inpcb *in6p = sotoinpcb(so);
 1404         int error, optval;
 1405         int level, op, optname;
 1406         int optlen;
 1407         struct proc *p;
 1408 
 1409         if (sopt) {
 1410                 level = sopt->sopt_level;
 1411                 op = sopt->sopt_dir;
 1412                 optname = sopt->sopt_name;
 1413                 optlen = sopt->sopt_valsize;
 1414                 p = sopt->sopt_p;
 1415         } else {
 1416                 panic("ip6_ctloutput: arg soopt is NULL");
 1417         }
 1418         error = optval = 0;
 1419 
 1420         privileged = (p == 0 || suser(p)) ? 0 : 1;
 1421 
 1422         if (level == IPPROTO_IPV6) {
 1423                 switch (op) {
 1424 
 1425                 case SOPT_SET:
 1426                         switch (optname) {
 1427                         case IPV6_PKTOPTIONS:
 1428                         {
 1429                                 struct mbuf *m;
 1430 
 1431                                 error = soopt_getm(sopt, &m); /* XXX */
 1432                                 if (error != NULL)
 1433                                         break;
 1434                                 error = soopt_mcopyin(sopt, m); /* XXX */
 1435                                 if (error != NULL)
 1436                                         break;
 1437                                 error = ip6_pcbopts(&in6p->in6p_outputopts,
 1438                                                     m, so, sopt);
 1439                                 m_freem(m); /* XXX */
 1440                                 break;
 1441                         }
 1442 
 1443                         /*
 1444                          * Use of some Hop-by-Hop options or some
 1445                          * Destination options, might require special
 1446                          * privilege.  That is, normal applications
 1447                          * (without special privilege) might be forbidden
 1448                          * from setting certain options in outgoing packets,
 1449                          * and might never see certain options in received
 1450                          * packets. [RFC 2292 Section 6]
 1451                          * KAME specific note:
 1452                          *  KAME prevents non-privileged users from sending or
 1453                          *  receiving ANY hbh/dst options in order to avoid
 1454                          *  overhead of parsing options in the kernel.
 1455                          */
 1456                         case IPV6_UNICAST_HOPS:
 1457                         case IPV6_CHECKSUM:
 1458                         case IPV6_FAITH:
 1459 
 1460                         case IPV6_V6ONLY:
 1461                                 if (optlen != sizeof(int)) {
 1462                                         error = EINVAL;
 1463                                         break;
 1464                                 }
 1465                                 error = sooptcopyin(sopt, &optval,
 1466                                         sizeof optval, sizeof optval);
 1467                                 if (error)
 1468                                         break;
 1469                                 switch (optname) {
 1470 
 1471                                 case IPV6_UNICAST_HOPS:
 1472                                         if (optval < -1 || optval >= 256)
 1473                                                 error = EINVAL;
 1474                                         else {
 1475                                                 /* -1 = kernel default */
 1476                                                 in6p->in6p_hops = optval;
 1477 
 1478                                                 if ((in6p->in6p_vflag &
 1479                                                      INP_IPV4) != 0)
 1480                                                         in6p->inp_ip_ttl = optval;
 1481                                         }
 1482                                         break;
 1483 #define OPTSET(bit) \
 1484 do { \
 1485         if (optval) \
 1486                 in6p->in6p_flags |= (bit); \
 1487         else \
 1488                 in6p->in6p_flags &= ~(bit); \
 1489 } while (0)
 1490 #define OPTBIT(bit) (in6p->in6p_flags & (bit) ? 1 : 0)
 1491 
 1492                                 case IPV6_CHECKSUM:
 1493                                         in6p->in6p_cksum = optval;
 1494                                         break;
 1495 
 1496                                 case IPV6_FAITH:
 1497                                         OPTSET(IN6P_FAITH);
 1498                                         break;
 1499 
 1500                                 case IPV6_V6ONLY:
 1501                                         /*
 1502                                          * make setsockopt(IPV6_V6ONLY)
 1503                                          * available only prior to bind(2).
 1504                                          * see ipng mailing list, Jun 22 2001.
 1505                                          */
 1506                                         if (in6p->in6p_lport ||
 1507                                             !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr))
 1508                                         {
 1509                                                 error = EINVAL;
 1510                                                 break;
 1511                                         }
 1512                                         OPTSET(IN6P_IPV6_V6ONLY);
 1513                                         if (optval)
 1514                                                 in6p->in6p_vflag &= ~INP_IPV4;
 1515                                         else
 1516                                                 in6p->in6p_vflag |= INP_IPV4;
 1517                                         break;
 1518                                 }
 1519                                 break;
 1520 
 1521                         case IPV6_PKTINFO:
 1522                         case IPV6_HOPLIMIT:
 1523                         case IPV6_HOPOPTS:
 1524                         case IPV6_DSTOPTS:
 1525                         case IPV6_RTHDR:
 1526                                 /* RFC 2292 */
 1527                                 if (optlen != sizeof(int)) {
 1528                                         error = EINVAL;
 1529                                         break;
 1530                                 }
 1531                                 error = sooptcopyin(sopt, &optval,
 1532                                         sizeof optval, sizeof optval);
 1533                                 if (error)
 1534                                         break;
 1535                                 switch (optname) {
 1536                                 case IPV6_PKTINFO:
 1537                                         OPTSET(IN6P_PKTINFO);
 1538                                         break;
 1539                                 case IPV6_HOPLIMIT:
 1540                                         OPTSET(IN6P_HOPLIMIT);
 1541                                         break;
 1542                                 case IPV6_HOPOPTS:
 1543                                         /*
 1544                                          * Check super-user privilege.
 1545                                          * See comments for IPV6_RECVHOPOPTS.
 1546                                          */
 1547                                         if (!privileged)
 1548                                                 return(EPERM);
 1549                                         OPTSET(IN6P_HOPOPTS);
 1550                                         break;
 1551                                 case IPV6_DSTOPTS:
 1552                                         if (!privileged)
 1553                                                 return(EPERM);
 1554                                         OPTSET(IN6P_DSTOPTS|IN6P_RTHDRDSTOPTS); /* XXX */
 1555                                         break;
 1556                                 case IPV6_RTHDR:
 1557                                         OPTSET(IN6P_RTHDR);
 1558                                         break;
 1559                                 }
 1560                                 break;
 1561 #undef OPTSET
 1562 
 1563                         case IPV6_MULTICAST_IF:
 1564                         case IPV6_MULTICAST_HOPS:
 1565                         case IPV6_MULTICAST_LOOP:
 1566                         case IPV6_JOIN_GROUP:
 1567                         case IPV6_LEAVE_GROUP:
 1568                             {
 1569                                 struct mbuf *m;
 1570                                 if (sopt->sopt_valsize > MLEN) {
 1571                                         error = EMSGSIZE;
 1572                                         break;
 1573                                 }
 1574                                 /* XXX */
 1575                                 MGET(m, sopt->sopt_p ? M_WAIT : M_DONTWAIT, MT_HEADER);
 1576                                 if (m == 0) {
 1577                                         error = ENOBUFS;
 1578                                         break;
 1579                                 }
 1580                                 m->m_len = sopt->sopt_valsize;
 1581                                 error = sooptcopyin(sopt, mtod(m, char *),
 1582                                                     m->m_len, m->m_len);
 1583                                 error = ip6_setmoptions(sopt->sopt_name,
 1584                                                         &in6p->in6p_moptions,
 1585                                                         m);
 1586                                 (void)m_free(m);
 1587                             }
 1588                                 break;
 1589 
 1590                         case IPV6_PORTRANGE:
 1591                                 error = sooptcopyin(sopt, &optval,
 1592                                     sizeof optval, sizeof optval);
 1593                                 if (error)
 1594                                         break;
 1595 
 1596                                 switch (optval) {
 1597                                 case IPV6_PORTRANGE_DEFAULT:
 1598                                         in6p->in6p_flags &= ~(IN6P_LOWPORT);
 1599                                         in6p->in6p_flags &= ~(IN6P_HIGHPORT);
 1600                                         break;
 1601 
 1602                                 case IPV6_PORTRANGE_HIGH:
 1603                                         in6p->in6p_flags &= ~(IN6P_LOWPORT);
 1604                                         in6p->in6p_flags |= IN6P_HIGHPORT;
 1605                                         break;
 1606 
 1607                                 case IPV6_PORTRANGE_LOW:
 1608                                         in6p->in6p_flags &= ~(IN6P_HIGHPORT);
 1609                                         in6p->in6p_flags |= IN6P_LOWPORT;
 1610                                         break;
 1611 
 1612                                 default:
 1613                                         error = EINVAL;
 1614                                         break;
 1615                                 }
 1616                                 break;
 1617 
 1618 #if defined(IPSEC) || defined(FAST_IPSEC)
 1619                         case IPV6_IPSEC_POLICY:
 1620                             {
 1621                                 caddr_t req = NULL;
 1622                                 size_t len = 0;
 1623                                 struct mbuf *m;
 1624 
 1625                                 if ((error = soopt_getm(sopt, &m)) != 0) /* XXX */
 1626                                         break;
 1627                                 if ((error = soopt_mcopyin(sopt, m)) != 0) /* XXX */
 1628                                         break;
 1629                                 if (m) {
 1630                                         req = mtod(m, caddr_t);
 1631                                         len = m->m_len;
 1632                                 }
 1633                                 error = ipsec6_set_policy(in6p, optname, req,
 1634                                                           len, privileged);
 1635                                 m_freem(m);
 1636                             }
 1637                                 break;
 1638 #endif /* KAME IPSEC */
 1639 
 1640                         case IPV6_FW_ADD:
 1641                         case IPV6_FW_DEL:
 1642                         case IPV6_FW_FLUSH:
 1643                         case IPV6_FW_ZERO:
 1644                             {
 1645                                 struct mbuf *m;
 1646                                 struct mbuf **mp = &m;
 1647 
 1648                                 if (ip6_fw_ctl_ptr == NULL)
 1649                                         return EINVAL;
 1650                                 /* XXX */
 1651                                 if ((error = soopt_getm(sopt, &m)) != 0)
 1652                                         break;
 1653                                 /* XXX */
 1654                                 if ((error = soopt_mcopyin(sopt, m)) != 0)
 1655                                         break;
 1656                                 error = (*ip6_fw_ctl_ptr)(optname, mp);
 1657                                 m = *mp;
 1658                             }
 1659                                 break;
 1660 
 1661                         default:
 1662                                 error = ENOPROTOOPT;
 1663                                 break;
 1664                         }
 1665                         break;
 1666 
 1667                 case SOPT_GET:
 1668                         switch (optname) {
 1669 
 1670                         case IPV6_PKTOPTIONS:
 1671                                 if (in6p->in6p_options) {
 1672                                         struct mbuf *m;
 1673                                         m = m_copym(in6p->in6p_options,
 1674                                             0, M_COPYALL, M_WAIT);
 1675                                         error = soopt_mcopyout(sopt, m);
 1676                                         if (error == 0)
 1677                                                 m_freem(m);
 1678                                 } else
 1679                                         sopt->sopt_valsize = 0;
 1680                                 break;
 1681 
 1682                         case IPV6_UNICAST_HOPS:
 1683                         case IPV6_CHECKSUM:
 1684 
 1685                         case IPV6_FAITH:
 1686                         case IPV6_V6ONLY:
 1687                         case IPV6_PORTRANGE:
 1688                                 switch (optname) {
 1689 
 1690                                 case IPV6_UNICAST_HOPS:
 1691                                         optval = in6p->in6p_hops;
 1692                                         break;
 1693 
 1694                                 case IPV6_CHECKSUM:
 1695                                         optval = in6p->in6p_cksum;
 1696                                         break;
 1697 
 1698                                 case IPV6_FAITH:
 1699                                         optval = OPTBIT(IN6P_FAITH);
 1700                                         break;
 1701 
 1702                                 case IPV6_V6ONLY:
 1703                                         optval = OPTBIT(IN6P_IPV6_V6ONLY);
 1704                                         break;
 1705 
 1706                                 case IPV6_PORTRANGE:
 1707                                     {
 1708                                         int flags;
 1709                                         flags = in6p->in6p_flags;
 1710                                         if (flags & IN6P_HIGHPORT)
 1711                                                 optval = IPV6_PORTRANGE_HIGH;
 1712                                         else if (flags & IN6P_LOWPORT)
 1713                                                 optval = IPV6_PORTRANGE_LOW;
 1714                                         else
 1715                                                 optval = 0;
 1716                                         break;
 1717                                     }
 1718                                 }
 1719                                 error = sooptcopyout(sopt, &optval,
 1720                                         sizeof optval);
 1721                                 break;
 1722 
 1723                         case IPV6_PKTINFO:
 1724                         case IPV6_HOPLIMIT:
 1725                         case IPV6_HOPOPTS:
 1726                         case IPV6_RTHDR:
 1727                         case IPV6_DSTOPTS:
 1728                                 if (optname == IPV6_HOPOPTS ||
 1729                                     optname == IPV6_DSTOPTS ||
 1730                                     !privileged)
 1731                                         return(EPERM);
 1732                                 switch (optname) {
 1733                                 case IPV6_PKTINFO:
 1734                                         optval = OPTBIT(IN6P_PKTINFO);
 1735                                         break;
 1736                                 case IPV6_HOPLIMIT:
 1737                                         optval = OPTBIT(IN6P_HOPLIMIT);
 1738                                         break;
 1739                                 case IPV6_HOPOPTS:
 1740                                         if (!privileged)
 1741                                                 return(EPERM);
 1742                                         optval = OPTBIT(IN6P_HOPOPTS);
 1743                                         break;
 1744                                 case IPV6_RTHDR:
 1745                                         optval = OPTBIT(IN6P_RTHDR);
 1746                                         break;
 1747                                 case IPV6_DSTOPTS:
 1748                                         if (!privileged)
 1749                                                 return(EPERM);
 1750                                         optval = OPTBIT(IN6P_DSTOPTS|IN6P_RTHDRDSTOPTS);
 1751                                         break;
 1752                                 }
 1753                                 error = sooptcopyout(sopt, &optval,
 1754                                         sizeof optval);
 1755                                 break;
 1756 
 1757                         case IPV6_MULTICAST_IF:
 1758                         case IPV6_MULTICAST_HOPS:
 1759                         case IPV6_MULTICAST_LOOP:
 1760                         case IPV6_JOIN_GROUP:
 1761                         case IPV6_LEAVE_GROUP:
 1762                             {
 1763                                 struct mbuf *m;
 1764                                 error = ip6_getmoptions(sopt->sopt_name,
 1765                                                 in6p->in6p_moptions, &m);
 1766                                 if (error == 0)
 1767                                         error = sooptcopyout(sopt,
 1768                                                 mtod(m, char *), m->m_len);
 1769                                 m_freem(m);
 1770                             }
 1771                                 break;
 1772 
 1773 #if defined(IPSEC) || defined(FAST_IPSEC)
 1774                         case IPV6_IPSEC_POLICY:
 1775                           {
 1776                                 caddr_t req = NULL;
 1777                                 size_t len = 0;
 1778                                 struct mbuf *m = NULL;
 1779                                 struct mbuf **mp = &m;
 1780 
 1781                                 error = soopt_getm(sopt, &m); /* XXX */
 1782                                 if (error != NULL)
 1783                                         break;
 1784                                 error = soopt_mcopyin(sopt, m); /* XXX */
 1785                                 if (error != NULL)
 1786                                         break;
 1787                                 if (m) {
 1788                                         req = mtod(m, caddr_t);
 1789                                         len = m->m_len;
 1790                                 }
 1791                                 error = ipsec6_get_policy(in6p, req, len, mp);
 1792                                 if (error == 0)
 1793                                         error = soopt_mcopyout(sopt, m); /*XXX*/
 1794                                 if (error == 0 && m)
 1795                                         m_freem(m);
 1796                                 break;
 1797                           }
 1798 #endif /* KAME IPSEC */
 1799 
 1800                         case IPV6_FW_GET:
 1801                           {
 1802                                 struct mbuf *m;
 1803                                 struct mbuf **mp = &m;
 1804 
 1805                                 if (ip6_fw_ctl_ptr == NULL)
 1806                                 {
 1807                                         return EINVAL;
 1808                                 }
 1809                                 error = (*ip6_fw_ctl_ptr)(optname, mp);
 1810                                 if (error == 0)
 1811                                         error = soopt_mcopyout(sopt, m); /* XXX */
 1812                                 if (error == 0 && m)
 1813                                         m_freem(m);
 1814                           }
 1815                                 break;
 1816 
 1817                         default:
 1818                                 error = ENOPROTOOPT;
 1819                                 break;
 1820                         }
 1821                         break;
 1822                 }
 1823         } else {
 1824                 error = EINVAL;
 1825         }
 1826         return(error);
 1827 }
 1828 
 1829 /*
 1830  * Set up IP6 options in pcb for insertion in output packets or
 1831  * specifying behavior of outgoing packets.
 1832  */
 1833 static int
 1834 ip6_pcbopts(pktopt, m, so, sopt)
 1835         struct ip6_pktopts **pktopt;
 1836         struct mbuf *m;
 1837         struct socket *so;
 1838         struct sockopt *sopt;
 1839 {
 1840         struct ip6_pktopts *opt = *pktopt;
 1841         int error = 0;
 1842         struct proc *p = sopt->sopt_p;
 1843         int priv = 0;
 1844 
 1845         /* turn off any old options. */
 1846         if (opt) {
 1847 #ifdef DIAGNOSTIC
 1848                 if (opt->ip6po_pktinfo || opt->ip6po_nexthop ||
 1849                     opt->ip6po_hbh || opt->ip6po_dest1 || opt->ip6po_dest2 ||
 1850                     opt->ip6po_rhinfo.ip6po_rhi_rthdr)
 1851                         printf("ip6_pcbopts: all specified options are cleared.\n");
 1852 #endif
 1853                 ip6_clearpktopts(opt, 1, -1);
 1854         } else
 1855                 opt = malloc(sizeof(*opt), M_IP6OPT, M_WAITOK);
 1856         *pktopt = NULL;
 1857 
 1858         if (!m || m->m_len == 0) {
 1859                 /*
 1860                  * Only turning off any previous options, regardless of
 1861                  * whether the opt is just created or given.
 1862                  */
 1863                 free(opt, M_IP6OPT);
 1864                 return(0);
 1865         }
 1866 
 1867         /*  set options specified by user. */
 1868         if (p && !suser(p))
 1869                 priv = 1;
 1870         if ((error = ip6_setpktoptions(m, opt, priv, 1)) != 0) {
 1871                 ip6_clearpktopts(opt, 1, -1); /* XXX: discard all options */
 1872                 free(opt, M_IP6OPT);
 1873                 return(error);
 1874         }
 1875         *pktopt = opt;
 1876         return(0);
 1877 }
 1878 
 1879 /*
 1880  * initialize ip6_pktopts.  beware that there are non-zero default values in
 1881  * the struct.
 1882  */
 1883 void
 1884 init_ip6pktopts(opt)
 1885         struct ip6_pktopts *opt;
 1886 {
 1887 
 1888         bzero(opt, sizeof(*opt));
 1889         opt->ip6po_hlim = -1;   /* -1 means default hop limit */
 1890 }
 1891 
 1892 void
 1893 ip6_clearpktopts(pktopt, needfree, optname)
 1894         struct ip6_pktopts *pktopt;
 1895         int needfree, optname;
 1896 {
 1897         if (pktopt == NULL)
 1898                 return;
 1899 
 1900         if (optname == -1) {
 1901                 if (needfree && pktopt->ip6po_pktinfo)
 1902                         free(pktopt->ip6po_pktinfo, M_IP6OPT);
 1903                 pktopt->ip6po_pktinfo = NULL;
 1904         }
 1905         if (optname == -1)
 1906                 pktopt->ip6po_hlim = -1;
 1907         if (optname == -1) {
 1908                 if (needfree && pktopt->ip6po_nexthop)
 1909                         free(pktopt->ip6po_nexthop, M_IP6OPT);
 1910                 pktopt->ip6po_nexthop = NULL;
 1911         }
 1912         if (optname == -1) {
 1913                 if (needfree && pktopt->ip6po_hbh)
 1914                         free(pktopt->ip6po_hbh, M_IP6OPT);
 1915                 pktopt->ip6po_hbh = NULL;
 1916         }
 1917         if (optname == -1) {
 1918                 if (needfree && pktopt->ip6po_dest1)
 1919                         free(pktopt->ip6po_dest1, M_IP6OPT);
 1920                 pktopt->ip6po_dest1 = NULL;
 1921         }
 1922         if (optname == -1) {
 1923                 if (needfree && pktopt->ip6po_rhinfo.ip6po_rhi_rthdr)
 1924                         free(pktopt->ip6po_rhinfo.ip6po_rhi_rthdr, M_IP6OPT);
 1925                 pktopt->ip6po_rhinfo.ip6po_rhi_rthdr = NULL;
 1926                 if (pktopt->ip6po_route.ro_rt) {
 1927                         RTFREE(pktopt->ip6po_route.ro_rt);
 1928                         pktopt->ip6po_route.ro_rt = NULL;
 1929                 }
 1930         }
 1931         if (optname == -1) {
 1932                 if (needfree && pktopt->ip6po_dest2)
 1933                         free(pktopt->ip6po_dest2, M_IP6OPT);
 1934                 pktopt->ip6po_dest2 = NULL;
 1935         }
 1936 }
 1937 
 1938 #define PKTOPT_EXTHDRCPY(type) \
 1939 do {\
 1940         if (src->type) {\
 1941                 int hlen =\
 1942                         (((struct ip6_ext *)src->type)->ip6e_len + 1) << 3;\
 1943                 dst->type = malloc(hlen, M_IP6OPT, canwait);\
 1944                 if (dst->type == NULL && canwait == M_NOWAIT)\
 1945                         goto bad;\
 1946                 bcopy(src->type, dst->type, hlen);\
 1947         }\
 1948 } while (0)
 1949 
 1950 struct ip6_pktopts *
 1951 ip6_copypktopts(src, canwait)
 1952         struct ip6_pktopts *src;
 1953         int canwait;
 1954 {
 1955         struct ip6_pktopts *dst;
 1956 
 1957         if (src == NULL) {
 1958                 printf("ip6_clearpktopts: invalid argument\n");
 1959                 return(NULL);
 1960         }
 1961 
 1962         dst = malloc(sizeof(*dst), M_IP6OPT, canwait);
 1963         if (dst == NULL && canwait == M_NOWAIT)
 1964                 return (NULL);
 1965         bzero(dst, sizeof(*dst));
 1966 
 1967         dst->ip6po_hlim = src->ip6po_hlim;
 1968         if (src->ip6po_pktinfo) {
 1969                 dst->ip6po_pktinfo = malloc(sizeof(*dst->ip6po_pktinfo),
 1970                                             M_IP6OPT, canwait);
 1971                 if (dst->ip6po_pktinfo == NULL && canwait == M_NOWAIT)
 1972                         goto bad;
 1973                 *dst->ip6po_pktinfo = *src->ip6po_pktinfo;
 1974         }
 1975         if (src->ip6po_nexthop) {
 1976                 dst->ip6po_nexthop = malloc(src->ip6po_nexthop->sa_len,
 1977                                             M_IP6OPT, canwait);
 1978                 if (dst->ip6po_nexthop == NULL && canwait == M_NOWAIT)
 1979                         goto bad;
 1980                 bcopy(src->ip6po_nexthop, dst->ip6po_nexthop,
 1981                       src->ip6po_nexthop->sa_len);
 1982         }
 1983         PKTOPT_EXTHDRCPY(ip6po_hbh);
 1984         PKTOPT_EXTHDRCPY(ip6po_dest1);
 1985         PKTOPT_EXTHDRCPY(ip6po_dest2);
 1986         PKTOPT_EXTHDRCPY(ip6po_rthdr); /* not copy the cached route */
 1987         return(dst);
 1988 
 1989   bad:
 1990         if (dst->ip6po_pktinfo) free(dst->ip6po_pktinfo, M_IP6OPT);
 1991         if (dst->ip6po_nexthop) free(dst->ip6po_nexthop, M_IP6OPT);
 1992         if (dst->ip6po_hbh) free(dst->ip6po_hbh, M_IP6OPT);
 1993         if (dst->ip6po_dest1) free(dst->ip6po_dest1, M_IP6OPT);
 1994         if (dst->ip6po_dest2) free(dst->ip6po_dest2, M_IP6OPT);
 1995         if (dst->ip6po_rthdr) free(dst->ip6po_rthdr, M_IP6OPT);
 1996         free(dst, M_IP6OPT);
 1997         return(NULL);
 1998 }
 1999 #undef PKTOPT_EXTHDRCPY
 2000 
 2001 void
 2002 ip6_freepcbopts(pktopt)
 2003         struct ip6_pktopts *pktopt;
 2004 {
 2005         if (pktopt == NULL)
 2006                 return;
 2007 
 2008         ip6_clearpktopts(pktopt, 1, -1);
 2009 
 2010         free(pktopt, M_IP6OPT);
 2011 }
 2012 
 2013 /*
 2014  * Set the IP6 multicast options in response to user setsockopt().
 2015  */
 2016 static int
 2017 ip6_setmoptions(optname, im6op, m)
 2018         int optname;
 2019         struct ip6_moptions **im6op;
 2020         struct mbuf *m;
 2021 {
 2022         int error = 0;
 2023         u_int loop, ifindex;
 2024         struct ipv6_mreq *mreq;
 2025         struct ifnet *ifp;
 2026         struct ip6_moptions *im6o = *im6op;
 2027         struct route_in6 ro;
 2028         struct sockaddr_in6 *dst;
 2029         struct in6_multi_mship *imm;
 2030         struct proc *p = curproc;       /* XXX */
 2031 
 2032         if (im6o == NULL) {
 2033                 /*
 2034                  * No multicast option buffer attached to the pcb;
 2035                  * allocate one and initialize to default values.
 2036                  */
 2037                 im6o = (struct ip6_moptions *)
 2038                         malloc(sizeof(*im6o), M_IPMOPTS, M_WAITOK);
 2039 
 2040                 if (im6o == NULL)
 2041                         return(ENOBUFS);
 2042                 *im6op = im6o;
 2043                 im6o->im6o_multicast_ifp = NULL;
 2044                 im6o->im6o_multicast_hlim = ip6_defmcasthlim;
 2045                 im6o->im6o_multicast_loop = IPV6_DEFAULT_MULTICAST_LOOP;
 2046                 LIST_INIT(&im6o->im6o_memberships);
 2047         }
 2048 
 2049         switch (optname) {
 2050 
 2051         case IPV6_MULTICAST_IF:
 2052                 /*
 2053                  * Select the interface for outgoing multicast packets.
 2054                  */
 2055                 if (m == NULL || m->m_len != sizeof(u_int)) {
 2056                         error = EINVAL;
 2057                         break;
 2058                 }
 2059                 bcopy(mtod(m, u_int *), &ifindex, sizeof(ifindex));
 2060                 if (ifindex < 0 || if_index < ifindex) {
 2061                         error = ENXIO;  /* XXX EINVAL? */
 2062                         break;
 2063                 }
 2064                 ifp = ifindex2ifnet[ifindex];
 2065                 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) {
 2066                         error = EADDRNOTAVAIL;
 2067                         break;
 2068                 }
 2069                 im6o->im6o_multicast_ifp = ifp;
 2070                 break;
 2071 
 2072         case IPV6_MULTICAST_HOPS:
 2073             {
 2074                 /*
 2075                  * Set the IP6 hoplimit for outgoing multicast packets.
 2076                  */
 2077                 int optval;
 2078                 if (m == NULL || m->m_len != sizeof(int)) {
 2079                         error = EINVAL;
 2080                         break;
 2081                 }
 2082                 bcopy(mtod(m, u_int *), &optval, sizeof(optval));
 2083                 if (optval < -1 || optval >= 256)
 2084                         error = EINVAL;
 2085                 else if (optval == -1)
 2086                         im6o->im6o_multicast_hlim = ip6_defmcasthlim;
 2087                 else
 2088                         im6o->im6o_multicast_hlim = optval;
 2089                 break;
 2090             }
 2091 
 2092         case IPV6_MULTICAST_LOOP:
 2093                 /*
 2094                  * Set the loopback flag for outgoing multicast packets.
 2095                  * Must be zero or one.
 2096                  */
 2097                 if (m == NULL || m->m_len != sizeof(u_int)) {
 2098                         error = EINVAL;
 2099                         break;
 2100                 }
 2101                 bcopy(mtod(m, u_int *), &loop, sizeof(loop));
 2102                 if (loop > 1) {
 2103                         error = EINVAL;
 2104                         break;
 2105                 }
 2106                 im6o->im6o_multicast_loop = loop;
 2107                 break;
 2108 
 2109         case IPV6_JOIN_GROUP:
 2110                 /*
 2111                  * Add a multicast group membership.
 2112                  * Group must be a valid IP6 multicast address.
 2113                  */
 2114                 if (m == NULL || m->m_len != sizeof(struct ipv6_mreq)) {
 2115                         error = EINVAL;
 2116                         break;
 2117                 }
 2118                 mreq = mtod(m, struct ipv6_mreq *);
 2119                 if (IN6_IS_ADDR_UNSPECIFIED(&mreq->ipv6mr_multiaddr)) {
 2120                         /*
 2121                          * We use the unspecified address to specify to accept
 2122                          * all multicast addresses. Only super user is allowed
 2123                          * to do this.
 2124                          */
 2125                         if (suser(p))
 2126                         {
 2127                                 error = EACCES;
 2128                                 break;
 2129                         }
 2130                 } else if (!IN6_IS_ADDR_MULTICAST(&mreq->ipv6mr_multiaddr)) {
 2131                         error = EINVAL;
 2132                         break;
 2133                 }
 2134 
 2135                 /*
 2136                  * If the interface is specified, validate it.
 2137                  */
 2138                 if (mreq->ipv6mr_interface < 0 ||
 2139                     if_index < mreq->ipv6mr_interface) {
 2140                         error = ENXIO;  /* XXX EINVAL? */
 2141                         break;
 2142                 }
 2143                 /*
 2144                  * If no interface was explicitly specified, choose an
 2145                  * appropriate one according to the given multicast address.
 2146                  */
 2147                 if (mreq->ipv6mr_interface == 0) {
 2148                         /*
 2149                          * If the multicast address is in node-local scope,
 2150                          * the interface should be a loopback interface.
 2151                          * Otherwise, look up the routing table for the
 2152                          * address, and choose the outgoing interface.
 2153                          *   XXX: is it a good approach?
 2154                          */
 2155                         if (IN6_IS_ADDR_MC_NODELOCAL(&mreq->ipv6mr_multiaddr)) {
 2156                                 ifp = &loif[0];
 2157                         } else {
 2158                                 ro.ro_rt = NULL;
 2159                                 dst = (struct sockaddr_in6 *)&ro.ro_dst;
 2160                                 bzero(dst, sizeof(*dst));
 2161                                 dst->sin6_len = sizeof(struct sockaddr_in6);
 2162                                 dst->sin6_family = AF_INET6;
 2163                                 dst->sin6_addr = mreq->ipv6mr_multiaddr;
 2164                                 rtalloc((struct route *)&ro);
 2165                                 if (ro.ro_rt == NULL) {
 2166                                         error = EADDRNOTAVAIL;
 2167                                         break;
 2168                                 }
 2169                                 ifp = ro.ro_rt->rt_ifp;
 2170                                 rtfree(ro.ro_rt);
 2171                         }
 2172                 } else
 2173                         ifp = ifindex2ifnet[mreq->ipv6mr_interface];
 2174 
 2175                 /*
 2176                  * See if we found an interface, and confirm that it
 2177                  * supports multicast
 2178                  */
 2179                 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) {
 2180                         error = EADDRNOTAVAIL;
 2181                         break;
 2182                 }
 2183                 /*
 2184                  * Put interface index into the multicast address,
 2185                  * if the address has link-local scope.
 2186                  */
 2187                 if (IN6_IS_ADDR_MC_LINKLOCAL(&mreq->ipv6mr_multiaddr)) {
 2188                         mreq->ipv6mr_multiaddr.s6_addr16[1] =
 2189                             htons(ifp->if_index);
 2190                 }
 2191                 /*
 2192                  * See if the membership already exists.
 2193                  */
 2194                 for (imm = im6o->im6o_memberships.lh_first;
 2195                      imm != NULL; imm = imm->i6mm_chain.le_next)
 2196                         if (imm->i6mm_maddr->in6m_ifp == ifp &&
 2197                             IN6_ARE_ADDR_EQUAL(&imm->i6mm_maddr->in6m_addr,
 2198                                                &mreq->ipv6mr_multiaddr))
 2199                                 break;
 2200                 if (imm != NULL) {
 2201                         error = EADDRINUSE;
 2202                         break;
 2203                 }
 2204                 /*
 2205                  * Everything looks good; add a new record to the multicast
 2206                  * address list for the given interface.
 2207                  */
 2208                 imm = malloc(sizeof(*imm), M_IPMADDR, M_WAITOK);
 2209                 if (imm == NULL) {
 2210                         error = ENOBUFS;
 2211                         break;
 2212                 }
 2213                 if ((imm->i6mm_maddr =
 2214                      in6_addmulti(&mreq->ipv6mr_multiaddr, ifp, &error)) == NULL) {
 2215                         free(imm, M_IPMADDR);
 2216                         break;
 2217                 }
 2218                 LIST_INSERT_HEAD(&im6o->im6o_memberships, imm, i6mm_chain);
 2219                 break;
 2220 
 2221         case IPV6_LEAVE_GROUP:
 2222                 /*
 2223                  * Drop a multicast group membership.
 2224                  * Group must be a valid IP6 multicast address.
 2225                  */
 2226                 if (m == NULL || m->m_len != sizeof(struct ipv6_mreq)) {
 2227                         error = EINVAL;
 2228                         break;
 2229                 }
 2230                 mreq = mtod(m, struct ipv6_mreq *);
 2231                 if (IN6_IS_ADDR_UNSPECIFIED(&mreq->ipv6mr_multiaddr)) {
 2232                         if (suser(p)) {
 2233                                 error = EACCES;
 2234                                 break;
 2235                         }
 2236                 } else if (!IN6_IS_ADDR_MULTICAST(&mreq->ipv6mr_multiaddr)) {
 2237                         error = EINVAL;
 2238                         break;
 2239                 }
 2240                 /*
 2241                  * If an interface address was specified, get a pointer
 2242                  * to its ifnet structure.
 2243                  */
 2244                 if (mreq->ipv6mr_interface < 0
 2245                  || if_index < mreq->ipv6mr_interface) {
 2246                         error = ENXIO;  /* XXX EINVAL? */
 2247                         break;
 2248                 }
 2249                 ifp = ifindex2ifnet[mreq->ipv6mr_interface];
 2250                 /*
 2251                  * Put interface index into the multicast address,
 2252                  * if the address has link-local scope.
 2253                  */
 2254                 if (IN6_IS_ADDR_MC_LINKLOCAL(&mreq->ipv6mr_multiaddr)) {
 2255                         mreq->ipv6mr_multiaddr.s6_addr16[1]
 2256                                 = htons(mreq->ipv6mr_interface);
 2257                 }
 2258                 /*
 2259                  * Find the membership in the membership list.
 2260                  */
 2261                 for (imm = im6o->im6o_memberships.lh_first;
 2262                      imm != NULL; imm = imm->i6mm_chain.le_next) {
 2263                         if ((ifp == NULL ||
 2264                              imm->i6mm_maddr->in6m_ifp == ifp) &&
 2265                             IN6_ARE_ADDR_EQUAL(&imm->i6mm_maddr->in6m_addr,
 2266                                                &mreq->ipv6mr_multiaddr))
 2267                                 break;
 2268                 }
 2269                 if (imm == NULL) {
 2270                         /* Unable to resolve interface */
 2271                         error = EADDRNOTAVAIL;
 2272                         break;
 2273                 }
 2274                 /*
 2275                  * Give up the multicast address record to which the
 2276                  * membership points.
 2277                  */
 2278                 LIST_REMOVE(imm, i6mm_chain);
 2279                 in6_delmulti(imm->i6mm_maddr);
 2280                 free(imm, M_IPMADDR);
 2281                 break;
 2282 
 2283         default:
 2284                 error = EOPNOTSUPP;
 2285                 break;
 2286         }
 2287 
 2288         /*
 2289          * If all options have default values, no need to keep the mbuf.
 2290          */
 2291         if (im6o->im6o_multicast_ifp == NULL &&
 2292             im6o->im6o_multicast_hlim == ip6_defmcasthlim &&
 2293             im6o->im6o_multicast_loop == IPV6_DEFAULT_MULTICAST_LOOP &&
 2294             im6o->im6o_memberships.lh_first == NULL) {
 2295                 free(*im6op, M_IPMOPTS);
 2296                 *im6op = NULL;
 2297         }
 2298 
 2299         return(error);
 2300 }
 2301 
 2302 /*
 2303  * Return the IP6 multicast options in response to user getsockopt().
 2304  */
 2305 static int
 2306 ip6_getmoptions(optname, im6o, mp)
 2307         int optname;
 2308         struct ip6_moptions *im6o;
 2309         struct mbuf **mp;
 2310 {
 2311         u_int *hlim, *loop, *ifindex;
 2312 
 2313         *mp = m_get(M_WAIT, MT_HEADER);         /* XXX */
 2314 
 2315         switch (optname) {
 2316 
 2317         case IPV6_MULTICAST_IF:
 2318                 ifindex = mtod(*mp, u_int *);
 2319                 (*mp)->m_len = sizeof(u_int);
 2320                 if (im6o == NULL || im6o->im6o_multicast_ifp == NULL)
 2321                         *ifindex = 0;
 2322                 else
 2323                         *ifindex = im6o->im6o_multicast_ifp->if_index;
 2324                 return(0);
 2325 
 2326         case IPV6_MULTICAST_HOPS:
 2327                 hlim = mtod(*mp, u_int *);
 2328                 (*mp)->m_len = sizeof(u_int);
 2329                 if (im6o == NULL)
 2330                         *hlim = ip6_defmcasthlim;
 2331                 else
 2332                         *hlim = im6o->im6o_multicast_hlim;
 2333                 return(0);
 2334 
 2335         case IPV6_MULTICAST_LOOP:
 2336                 loop = mtod(*mp, u_int *);
 2337                 (*mp)->m_len = sizeof(u_int);
 2338                 if (im6o == NULL)
 2339                         *loop = ip6_defmcasthlim;
 2340                 else
 2341                         *loop = im6o->im6o_multicast_loop;
 2342                 return(0);
 2343 
 2344         default:
 2345                 return(EOPNOTSUPP);
 2346         }
 2347 }
 2348 
 2349 /*
 2350  * Discard the IP6 multicast options.
 2351  */
 2352 void
 2353 ip6_freemoptions(im6o)
 2354         struct ip6_moptions *im6o;
 2355 {
 2356         struct in6_multi_mship *imm;
 2357 
 2358         if (im6o == NULL)
 2359                 return;
 2360 
 2361         while ((imm = im6o->im6o_memberships.lh_first) != NULL) {
 2362                 LIST_REMOVE(imm, i6mm_chain);
 2363                 if (imm->i6mm_maddr)
 2364                         in6_delmulti(imm->i6mm_maddr);
 2365                 free(imm, M_IPMADDR);
 2366         }
 2367         free(im6o, M_IPMOPTS);
 2368 }
 2369 
 2370 /*
 2371  * Set IPv6 outgoing packet options based on advanced API.
 2372  */
 2373 int
 2374 ip6_setpktoptions(control, opt, priv, needcopy)
 2375         struct mbuf *control;
 2376         struct ip6_pktopts *opt;
 2377         int priv, needcopy;
 2378 {
 2379         struct cmsghdr *cm = 0;
 2380 
 2381         if (control == 0 || opt == 0)
 2382                 return(EINVAL);
 2383 
 2384         init_ip6pktopts(opt);
 2385 
 2386         /*
 2387          * XXX: Currently, we assume all the optional information is stored
 2388          * in a single mbuf.
 2389          */
 2390         if (control->m_next)
 2391                 return(EINVAL);
 2392 
 2393         for (; control->m_len; control->m_data += CMSG_ALIGN(cm->cmsg_len),
 2394                      control->m_len -= CMSG_ALIGN(cm->cmsg_len)) {
 2395                 cm = mtod(control, struct cmsghdr *);
 2396                 if (cm->cmsg_len == 0 || cm->cmsg_len > control->m_len)
 2397                         return(EINVAL);
 2398                 if (cm->cmsg_level != IPPROTO_IPV6)
 2399                         continue;
 2400 
 2401                 /*
 2402                  * XXX should check if RFC2292 API is mixed with 2292bis API
 2403                  */
 2404                 switch (cm->cmsg_type) {
 2405                 case IPV6_PKTINFO:
 2406                         if (cm->cmsg_len != CMSG_LEN(sizeof(struct in6_pktinfo)))
 2407                                 return(EINVAL);
 2408                         if (needcopy) {
 2409                                 /* XXX: Is it really WAITOK? */
 2410                                 opt->ip6po_pktinfo =
 2411                                         malloc(sizeof(struct in6_pktinfo),
 2412                                                M_IP6OPT, M_WAITOK);
 2413                                 bcopy(CMSG_DATA(cm), opt->ip6po_pktinfo,
 2414                                     sizeof(struct in6_pktinfo));
 2415                         } else
 2416                                 opt->ip6po_pktinfo =
 2417                                         (struct in6_pktinfo *)CMSG_DATA(cm);
 2418                         if (opt->ip6po_pktinfo->ipi6_ifindex &&
 2419                             IN6_IS_ADDR_LINKLOCAL(&opt->ip6po_pktinfo->ipi6_addr))
 2420                                 opt->ip6po_pktinfo->ipi6_addr.s6_addr16[1] =
 2421                                         htons(opt->ip6po_pktinfo->ipi6_ifindex);
 2422 
 2423                         if (opt->ip6po_pktinfo->ipi6_ifindex > if_index
 2424                          || opt->ip6po_pktinfo->ipi6_ifindex < 0) {
 2425                                 return(ENXIO);
 2426                         }
 2427 
 2428                         /*
 2429                          * Check if the requested source address is indeed a
 2430                          * unicast address assigned to the node, and can be
 2431                          * used as the packet's source address.
 2432                          */
 2433                         if (!IN6_IS_ADDR_UNSPECIFIED(&opt->ip6po_pktinfo->ipi6_addr)) {
 2434                                 struct in6_ifaddr *ia6;
 2435                                 struct sockaddr_in6 sin6;
 2436 
 2437                                 bzero(&sin6, sizeof(sin6));
 2438                                 sin6.sin6_len = sizeof(sin6);
 2439                                 sin6.sin6_family = AF_INET6;
 2440                                 sin6.sin6_addr =
 2441                                         opt->ip6po_pktinfo->ipi6_addr;
 2442                                 ia6 = (struct in6_ifaddr *)ifa_ifwithaddr(sin6tosa(&sin6));
 2443                                 if (ia6 == NULL ||
 2444                                     (ia6->ia6_flags & (IN6_IFF_ANYCAST |
 2445                                                        IN6_IFF_NOTREADY)) != 0)
 2446                                         return(EADDRNOTAVAIL);
 2447                         }
 2448                         break;
 2449 
 2450                 case IPV6_HOPLIMIT:
 2451                         if (cm->cmsg_len != CMSG_LEN(sizeof(int)))
 2452                                 return(EINVAL);
 2453 
 2454                         opt->ip6po_hlim = *(int *)CMSG_DATA(cm);
 2455                         if (opt->ip6po_hlim < -1 || opt->ip6po_hlim > 255)
 2456                                 return(EINVAL);
 2457                         break;
 2458 
 2459                 case IPV6_NEXTHOP:
 2460                         if (!priv)
 2461                                 return(EPERM);
 2462 
 2463                         if (cm->cmsg_len < sizeof(u_char) ||
 2464                             /* check if cmsg_len is large enough for sa_len */
 2465                             cm->cmsg_len < CMSG_LEN(*CMSG_DATA(cm)))
 2466                                 return(EINVAL);
 2467 
 2468                         if (needcopy) {
 2469                                 opt->ip6po_nexthop =
 2470                                         malloc(*CMSG_DATA(cm),
 2471                                                M_IP6OPT, M_WAITOK);
 2472                                 bcopy(CMSG_DATA(cm),
 2473                                       opt->ip6po_nexthop,
 2474                                       *CMSG_DATA(cm));
 2475                         } else
 2476                                 opt->ip6po_nexthop =
 2477                                         (struct sockaddr *)CMSG_DATA(cm);
 2478                         break;
 2479 
 2480                 case IPV6_HOPOPTS:
 2481                 {
 2482                         struct ip6_hbh *hbh;
 2483                         int hbhlen;
 2484 
 2485                         if (cm->cmsg_len < CMSG_LEN(sizeof(struct ip6_hbh)))
 2486                                 return(EINVAL);
 2487                         hbh = (struct ip6_hbh *)CMSG_DATA(cm);
 2488                         hbhlen = (hbh->ip6h_len + 1) << 3;
 2489                         if (cm->cmsg_len != CMSG_LEN(hbhlen))
 2490                                 return(EINVAL);
 2491 
 2492                         if (needcopy) {
 2493                                 opt->ip6po_hbh =
 2494                                         malloc(hbhlen, M_IP6OPT, M_WAITOK);
 2495                                 bcopy(hbh, opt->ip6po_hbh, hbhlen);
 2496                         } else
 2497                                 opt->ip6po_hbh = hbh;
 2498                         break;
 2499                 }
 2500 
 2501                 case IPV6_DSTOPTS:
 2502                 {
 2503                         struct ip6_dest *dest, **newdest;
 2504                         int destlen;
 2505 
 2506                         if (cm->cmsg_len < CMSG_LEN(sizeof(struct ip6_dest)))
 2507                                 return(EINVAL);
 2508                         dest = (struct ip6_dest *)CMSG_DATA(cm);
 2509                         destlen = (dest->ip6d_len + 1) << 3;
 2510                         if (cm->cmsg_len != CMSG_LEN(destlen))
 2511                                 return(EINVAL);
 2512 
 2513                         /* 
 2514                          * The old advacned API is ambiguous on this
 2515                          * point. Our approach is to determine the
 2516                          * position based according to the existence
 2517                          * of a routing header. Note, however, that
 2518                          * this depends on the order of the extension
 2519                          * headers in the ancillary data; the 1st part
 2520                          * of the destination options header must
 2521                          * appear before the routing header in the
 2522                          * ancillary data, too.
 2523                          * RFC2292bis solved the ambiguity by
 2524                          * introducing separate cmsg types.
 2525                          */
 2526                         if (opt->ip6po_rthdr == NULL)
 2527                                 newdest = &opt->ip6po_dest1;
 2528                         else
 2529                                 newdest = &opt->ip6po_dest2;
 2530 
 2531                         if (needcopy) {
 2532                                 *newdest = malloc(destlen, M_IP6OPT, M_WAITOK);
 2533                                 bcopy(dest, *newdest, destlen);
 2534                         } else
 2535                                 *newdest = dest;
 2536 
 2537                         break;
 2538                 }
 2539 
 2540                 case IPV6_RTHDR:
 2541                 {
 2542                         struct ip6_rthdr *rth;
 2543                         int rthlen;
 2544 
 2545                         if (cm->cmsg_len < CMSG_LEN(sizeof(struct ip6_rthdr)))
 2546                                 return(EINVAL);
 2547                         rth = (struct ip6_rthdr *)CMSG_DATA(cm);
 2548                         rthlen = (rth->ip6r_len + 1) << 3;
 2549                         if (cm->cmsg_len != CMSG_LEN(rthlen))
 2550                                 return(EINVAL);
 2551 
 2552                         switch (rth->ip6r_type) {
 2553                         case IPV6_RTHDR_TYPE_0:
 2554                                 /* must contain one addr */
 2555                                 if (rth->ip6r_len == 0)
 2556                                         return(EINVAL);
 2557                                 /* length must be even */
 2558                                 if (rth->ip6r_len % 2)
 2559                                         return(EINVAL);
 2560                                 if (rth->ip6r_len / 2 != rth->ip6r_segleft)
 2561                                         return(EINVAL);
 2562                                 break;
 2563                         default:
 2564                                 return(EINVAL); /* not supported */
 2565                         }
 2566 
 2567                         if (needcopy) {
 2568                                 opt->ip6po_rthdr = malloc(rthlen, M_IP6OPT,
 2569                                                           M_WAITOK);
 2570                                 bcopy(rth, opt->ip6po_rthdr, rthlen);
 2571                         } else
 2572                                 opt->ip6po_rthdr = rth;
 2573 
 2574                         break;
 2575                 }
 2576 
 2577                 default:
 2578                         return(ENOPROTOOPT);
 2579                 }
 2580         }
 2581 
 2582         return(0);
 2583 }
 2584 
 2585 /*
 2586  * Routine called from ip6_output() to loop back a copy of an IP6 multicast
 2587  * packet to the input queue of a specified interface.  Note that this
 2588  * calls the output routine of the loopback "driver", but with an interface
 2589  * pointer that might NOT be &loif -- easier than replicating that code here.
 2590  */
 2591 void
 2592 ip6_mloopback(ifp, m, dst)
 2593         struct ifnet *ifp;
 2594         struct mbuf *m;
 2595         struct sockaddr_in6 *dst;
 2596 {
 2597         struct mbuf *copym;
 2598         struct ip6_hdr *ip6;
 2599 
 2600         copym = m_copy(m, 0, M_COPYALL);
 2601         if (copym == NULL)
 2602                 return;
 2603 
 2604         /*
 2605          * Make sure to deep-copy IPv6 header portion in case the data
 2606          * is in an mbuf cluster, so that we can safely override the IPv6
 2607          * header portion later.
 2608          */
 2609         if ((copym->m_flags & M_EXT) != 0 ||
 2610             copym->m_len < sizeof(struct ip6_hdr)) {
 2611                 copym = m_pullup(copym, sizeof(struct ip6_hdr));
 2612                 if (copym == NULL)
 2613                         return;
 2614         }
 2615 
 2616 #ifdef DIAGNOSTIC
 2617         if (copym->m_len < sizeof(*ip6)) {
 2618                 m_freem(copym);
 2619                 return;
 2620         }
 2621 #endif
 2622 
 2623         ip6 = mtod(copym, struct ip6_hdr *);
 2624 #ifndef SCOPEDROUTING
 2625         /*
 2626          * clear embedded scope identifiers if necessary.
 2627          * in6_clearscope will touch the addresses only when necessary.
 2628          */
 2629         in6_clearscope(&ip6->ip6_src);
 2630         in6_clearscope(&ip6->ip6_dst);
 2631 #endif
 2632 
 2633         (void)if_simloop(ifp, copym, dst->sin6_family, NULL);
 2634 }
 2635 
 2636 /*
 2637  * Chop IPv6 header off from the payload.
 2638  */
 2639 static int
 2640 ip6_splithdr(m, exthdrs)
 2641         struct mbuf *m;
 2642         struct ip6_exthdrs *exthdrs;
 2643 {
 2644         struct mbuf *mh;
 2645         struct ip6_hdr *ip6;
 2646 
 2647         ip6 = mtod(m, struct ip6_hdr *);
 2648         if (m->m_len > sizeof(*ip6)) {
 2649                 MGETHDR(mh, M_DONTWAIT, MT_HEADER);
 2650                 if (mh == 0) {
 2651                         m_freem(m);
 2652                         return ENOBUFS;
 2653                 }
 2654                 M_MOVE_PKTHDR(mh, m);
 2655                 MH_ALIGN(mh, sizeof(*ip6));
 2656                 m->m_len -= sizeof(*ip6);
 2657                 m->m_data += sizeof(*ip6);
 2658                 mh->m_next = m;
 2659                 m = mh;
 2660                 m->m_len = sizeof(*ip6);
 2661                 bcopy((caddr_t)ip6, mtod(m, caddr_t), sizeof(*ip6));
 2662         }
 2663         exthdrs->ip6e_ip6 = m;
 2664         return 0;
 2665 }
 2666 
 2667 /*
 2668  * Compute IPv6 extension header length.
 2669  */
 2670 int
 2671 ip6_optlen(in6p)
 2672         struct in6pcb *in6p;
 2673 {
 2674         int len;
 2675 
 2676         if (!in6p->in6p_outputopts)
 2677                 return 0;
 2678 
 2679         len = 0;
 2680 #define elen(x) \
 2681     (((struct ip6_ext *)(x)) ? (((struct ip6_ext *)(x))->ip6e_len + 1) << 3 : 0)
 2682 
 2683         len += elen(in6p->in6p_outputopts->ip6po_hbh);
 2684         if (in6p->in6p_outputopts->ip6po_rthdr)
 2685                 /* dest1 is valid with rthdr only */
 2686                 len += elen(in6p->in6p_outputopts->ip6po_dest1);
 2687         len += elen(in6p->in6p_outputopts->ip6po_rthdr);
 2688         len += elen(in6p->in6p_outputopts->ip6po_dest2);
 2689         return len;
 2690 #undef elen
 2691 }

Cache object: fbfe516220aa9a4517eaadc64a1386af


[ 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.