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 /*      $NetBSD: ip6_output.c,v 1.232 2023/01/27 09:33:43 ozaki-r Exp $ */
    2 /*      $KAME: ip6_output.c,v 1.172 2001/03/25 09:55:56 itojun 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. Neither the name of the University nor the names of its contributors
   46  *    may be used to endorse or promote products derived from this software
   47  *    without specific prior written permission.
   48  *
   49  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
   50  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
   51  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
   52  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
   53  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
   54  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
   55  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
   56  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
   57  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
   58  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
   59  * SUCH DAMAGE.
   60  *
   61  *      @(#)ip_output.c 8.3 (Berkeley) 1/21/94
   62  */
   63 
   64 #include <sys/cdefs.h>
   65 __KERNEL_RCSID(0, "$NetBSD: ip6_output.c,v 1.232 2023/01/27 09:33:43 ozaki-r Exp $");
   66 
   67 #ifdef _KERNEL_OPT
   68 #include "opt_inet.h"
   69 #include "opt_inet6.h"
   70 #include "opt_ipsec.h"
   71 #endif
   72 
   73 #include <sys/param.h>
   74 #include <sys/malloc.h>
   75 #include <sys/mbuf.h>
   76 #include <sys/errno.h>
   77 #include <sys/socket.h>
   78 #include <sys/socketvar.h>
   79 #include <sys/syslog.h>
   80 #include <sys/systm.h>
   81 #include <sys/proc.h>
   82 #include <sys/kauth.h>
   83 
   84 #include <net/if.h>
   85 #include <net/route.h>
   86 #include <net/pfil.h>
   87 
   88 #include <netinet/in.h>
   89 #include <netinet/in_var.h>
   90 #include <netinet/ip6.h>
   91 #include <netinet/ip_var.h>
   92 #include <netinet/icmp6.h>
   93 #include <netinet/in_offload.h>
   94 #include <netinet/portalgo.h>
   95 #include <netinet6/in6_offload.h>
   96 #include <netinet6/ip6_var.h>
   97 #include <netinet6/ip6_private.h>
   98 #include <netinet6/in6_pcb.h>
   99 #include <netinet6/nd6.h>
  100 #include <netinet6/ip6protosw.h>
  101 #include <netinet6/scope6_var.h>
  102 
  103 #ifdef IPSEC
  104 #include <netipsec/ipsec.h>
  105 #include <netipsec/ipsec6.h>
  106 #include <netipsec/key.h>
  107 #endif
  108 
  109 extern pfil_head_t *inet6_pfil_hook;    /* XXX */
  110 
  111 struct ip6_exthdrs {
  112         struct mbuf *ip6e_ip6;
  113         struct mbuf *ip6e_hbh;
  114         struct mbuf *ip6e_dest1;
  115         struct mbuf *ip6e_rthdr;
  116         struct mbuf *ip6e_dest2;
  117 };
  118 
  119 static int ip6_pcbopt(int, u_char *, int, struct ip6_pktopts **,
  120         kauth_cred_t, int);
  121 static int ip6_getpcbopt(struct ip6_pktopts *, int, struct sockopt *);
  122 static int ip6_setpktopt(int, u_char *, int, struct ip6_pktopts *, kauth_cred_t,
  123         int, int, int);
  124 static int ip6_setmoptions(const struct sockopt *, struct inpcb *);
  125 static int ip6_getmoptions(struct sockopt *, struct inpcb *);
  126 static int ip6_copyexthdr(struct mbuf **, void *, int);
  127 static int ip6_insertfraghdr(struct mbuf *, struct mbuf *, int,
  128         struct ip6_frag **);
  129 static int ip6_insert_jumboopt(struct ip6_exthdrs *, u_int32_t);
  130 static int ip6_splithdr(struct mbuf *, struct ip6_exthdrs *);
  131 static int ip6_getpmtu(struct rtentry *, struct ifnet *, u_long *, int *);
  132 static int copypktopts(struct ip6_pktopts *, struct ip6_pktopts *, int);
  133 static int ip6_ifaddrvalid(const struct in6_addr *, const struct in6_addr *);
  134 static int ip6_handle_rthdr(struct ip6_rthdr *, struct ip6_hdr *);
  135 
  136 #ifdef RFC2292
  137 static int ip6_pcbopts(struct ip6_pktopts **, struct socket *, struct sockopt *);
  138 #endif
  139 
  140 static int
  141 ip6_handle_rthdr(struct ip6_rthdr *rh, struct ip6_hdr *ip6)
  142 {
  143         int error = 0;
  144 
  145         switch (rh->ip6r_type) {
  146         case IPV6_RTHDR_TYPE_0:
  147                 /* Dropped, RFC5095. */
  148         default:        /* is it possible? */
  149                 error = EINVAL;
  150         }
  151 
  152         return error;
  153 }
  154 
  155 /*
  156  * Send an IP packet to a host.
  157  */
  158 int
  159 ip6_if_output(struct ifnet * const ifp, struct ifnet * const origifp,
  160     struct mbuf * const m, const struct sockaddr_in6 * const dst,
  161     const struct rtentry *rt)
  162 {
  163         int error = 0;
  164 
  165         if (rt != NULL) {
  166                 error = rt_check_reject_route(rt, ifp);
  167                 if (error != 0) {
  168                         IP6_STATINC(IP6_STAT_RTREJECT);
  169                         m_freem(m);
  170                         return error;
  171                 }
  172         }
  173 
  174         if ((ifp->if_flags & IFF_LOOPBACK) != 0)
  175                 error = if_output_lock(ifp, origifp, m, sin6tocsa(dst), rt);
  176         else
  177                 error = if_output_lock(ifp, ifp, m, sin6tocsa(dst), rt);
  178         return error;
  179 }
  180 
  181 /*
  182  * IP6 output. The packet in mbuf chain m contains a skeletal IP6
  183  * header (with pri, len, nxt, hlim, src, dst).
  184  *
  185  * This function may modify ver and hlim only. The mbuf chain containing the
  186  * packet will be freed. The mbuf opt, if present, will not be freed.
  187  *
  188  * Type of "mtu": rt_rmx.rmx_mtu is u_long, ifnet.ifr_mtu is int, and
  189  * nd_ifinfo.linkmtu is u_int32_t. So we use u_long to hold largest one,
  190  * which is rt_rmx.rmx_mtu.
  191  */
  192 int
  193 ip6_output(
  194     struct mbuf *m0,
  195     struct ip6_pktopts *opt,
  196     struct route *ro,
  197     int flags,
  198     struct ip6_moptions *im6o,
  199     struct inpcb *inp,
  200     struct ifnet **ifpp         /* XXX: just for statistics */
  201 )
  202 {
  203         struct ip6_hdr *ip6, *mhip6;
  204         struct ifnet *ifp = NULL, *origifp = NULL;
  205         struct mbuf *m = m0;
  206         int tlen, len, off;
  207         bool tso;
  208         struct route ip6route;
  209         struct rtentry *rt = NULL, *rt_pmtu;
  210         const struct sockaddr_in6 *dst;
  211         struct sockaddr_in6 src_sa, dst_sa;
  212         int error = 0;
  213         struct in6_ifaddr *ia = NULL;
  214         u_long mtu;
  215         int alwaysfrag, dontfrag;
  216         u_int32_t optlen = 0, plen = 0, unfragpartlen = 0;
  217         struct ip6_exthdrs exthdrs;
  218         struct in6_addr finaldst, src0, dst0;
  219         u_int32_t zone;
  220         struct route *ro_pmtu = NULL;
  221         int hdrsplit = 0;
  222         int needipsec = 0;
  223 #ifdef IPSEC
  224         struct secpolicy *sp = NULL;
  225 #endif
  226         struct psref psref, psref_ia;
  227         int bound = curlwp_bind();
  228         bool release_psref_ia = false;
  229 
  230 #ifdef DIAGNOSTIC
  231         if ((m->m_flags & M_PKTHDR) == 0)
  232                 panic("ip6_output: no HDR");
  233         if ((m->m_pkthdr.csum_flags &
  234             (M_CSUM_TCPv4|M_CSUM_UDPv4|M_CSUM_TSOv4)) != 0) {
  235                 panic("ip6_output: IPv4 checksum offload flags: %d",
  236                     m->m_pkthdr.csum_flags);
  237         }
  238         if ((m->m_pkthdr.csum_flags & (M_CSUM_TCPv6|M_CSUM_UDPv6)) ==
  239             (M_CSUM_TCPv6|M_CSUM_UDPv6)) {
  240                 panic("ip6_output: conflicting checksum offload flags: %d",
  241                     m->m_pkthdr.csum_flags);
  242         }
  243 #endif
  244 
  245         M_CSUM_DATA_IPv6_SET(m->m_pkthdr.csum_data, sizeof(struct ip6_hdr));
  246 
  247 #define MAKE_EXTHDR(hp, mp)                                             \
  248     do {                                                                \
  249         if (hp) {                                                       \
  250                 struct ip6_ext *eh = (struct ip6_ext *)(hp);            \
  251                 error = ip6_copyexthdr((mp), (void *)(hp),              \
  252                     ((eh)->ip6e_len + 1) << 3);                         \
  253                 if (error)                                              \
  254                         goto freehdrs;                                  \
  255         }                                                               \
  256     } while (/*CONSTCOND*/ 0)
  257 
  258         memset(&exthdrs, 0, sizeof(exthdrs));
  259         if (opt) {
  260                 /* Hop-by-Hop options header */
  261                 MAKE_EXTHDR(opt->ip6po_hbh, &exthdrs.ip6e_hbh);
  262                 /* Destination options header (1st part) */
  263                 MAKE_EXTHDR(opt->ip6po_dest1, &exthdrs.ip6e_dest1);
  264                 /* Routing header */
  265                 MAKE_EXTHDR(opt->ip6po_rthdr, &exthdrs.ip6e_rthdr);
  266                 /* Destination options header (2nd part) */
  267                 MAKE_EXTHDR(opt->ip6po_dest2, &exthdrs.ip6e_dest2);
  268         }
  269 
  270         /*
  271          * Calculate the total length of the extension header chain.
  272          * Keep the length of the unfragmentable part for fragmentation.
  273          */
  274         optlen = 0;
  275         if (exthdrs.ip6e_hbh)
  276                 optlen += exthdrs.ip6e_hbh->m_len;
  277         if (exthdrs.ip6e_dest1)
  278                 optlen += exthdrs.ip6e_dest1->m_len;
  279         if (exthdrs.ip6e_rthdr)
  280                 optlen += exthdrs.ip6e_rthdr->m_len;
  281         unfragpartlen = optlen + sizeof(struct ip6_hdr);
  282         /* NOTE: we don't add AH/ESP length here. do that later. */
  283         if (exthdrs.ip6e_dest2)
  284                 optlen += exthdrs.ip6e_dest2->m_len;
  285 
  286 #ifdef IPSEC
  287         if (ipsec_used) {
  288                 /* Check the security policy (SP) for the packet */
  289                 sp = ipsec6_check_policy(m, inp, flags, &needipsec, &error);
  290                 if (error != 0) {
  291                         /*
  292                          * Hack: -EINVAL is used to signal that a packet
  293                          * should be silently discarded.  This is typically
  294                          * because we asked key management for an SA and
  295                          * it was delayed (e.g. kicked up to IKE).
  296                          */
  297                         if (error == -EINVAL)
  298                                 error = 0;
  299                         IP6_STATINC(IP6_STAT_IPSECDROP_OUT);
  300                         goto freehdrs;
  301                 }
  302         }
  303 #endif
  304 
  305         if (needipsec &&
  306             (m->m_pkthdr.csum_flags & (M_CSUM_UDPv6|M_CSUM_TCPv6)) != 0) {
  307                 in6_undefer_cksum_tcpudp(m);
  308                 m->m_pkthdr.csum_flags &= ~(M_CSUM_UDPv6|M_CSUM_TCPv6);
  309         }
  310 
  311         /*
  312          * If we need IPsec, or there is at least one extension header,
  313          * separate IP6 header from the payload.
  314          */
  315         if ((needipsec || optlen) && !hdrsplit) {
  316                 if ((error = ip6_splithdr(m, &exthdrs)) != 0) {
  317                         IP6_STATINC(IP6_STAT_ODROPPED);
  318                         m = NULL;
  319                         goto freehdrs;
  320                 }
  321                 m = exthdrs.ip6e_ip6;
  322                 hdrsplit++;
  323         }
  324 
  325         /* adjust pointer */
  326         ip6 = mtod(m, struct ip6_hdr *);
  327 
  328         /* adjust mbuf packet header length */
  329         m->m_pkthdr.len += optlen;
  330         plen = m->m_pkthdr.len - sizeof(*ip6);
  331 
  332         /* If this is a jumbo payload, insert a jumbo payload option. */
  333         if (plen > IPV6_MAXPACKET) {
  334                 if (!hdrsplit) {
  335                         if ((error = ip6_splithdr(m, &exthdrs)) != 0) {
  336                                 IP6_STATINC(IP6_STAT_ODROPPED);
  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                         IP6_STATINC(IP6_STAT_ODROPPED);
  347                         goto freehdrs;
  348                 }
  349                 optlen += 8; /* XXX JUMBOOPTLEN */
  350                 ip6->ip6_plen = 0;
  351         } else
  352                 ip6->ip6_plen = htons(plen);
  353 
  354         /*
  355          * Concatenate headers and fill in next header fields.
  356          * Here we have, on "m"
  357          *      IPv6 payload
  358          * and we insert headers accordingly.  Finally, we should be getting:
  359          *      IPv6 hbh dest1 rthdr ah* [esp* dest2 payload]
  360          *
  361          * during the header composing process, "m" points to IPv6 header.
  362          * "mprev" points to an extension header prior to esp.
  363          */
  364         {
  365                 u_char *nexthdrp = &ip6->ip6_nxt;
  366                 struct mbuf *mprev = m;
  367 
  368                 /*
  369                  * we treat dest2 specially.  this makes IPsec processing
  370                  * much easier.  the goal here is to make mprev point the
  371                  * mbuf prior to dest2.
  372                  *
  373                  * result: IPv6 dest2 payload
  374                  * m and mprev will point to IPv6 header.
  375                  */
  376                 if (exthdrs.ip6e_dest2) {
  377                         if (!hdrsplit)
  378                                 panic("assumption failed: hdr not split");
  379                         exthdrs.ip6e_dest2->m_next = m->m_next;
  380                         m->m_next = exthdrs.ip6e_dest2;
  381                         *mtod(exthdrs.ip6e_dest2, u_char *) = ip6->ip6_nxt;
  382                         ip6->ip6_nxt = IPPROTO_DSTOPTS;
  383                 }
  384 
  385 #define MAKE_CHAIN(m, mp, p, i)\
  386     do {\
  387         if (m) {\
  388                 if (!hdrsplit) \
  389                         panic("assumption failed: hdr not split"); \
  390                 *mtod((m), u_char *) = *(p);\
  391                 *(p) = (i);\
  392                 p = mtod((m), u_char *);\
  393                 (m)->m_next = (mp)->m_next;\
  394                 (mp)->m_next = (m);\
  395                 (mp) = (m);\
  396         }\
  397     } while (/*CONSTCOND*/ 0)
  398                 /*
  399                  * result: IPv6 hbh dest1 rthdr dest2 payload
  400                  * m will point to IPv6 header.  mprev will point to the
  401                  * extension header prior to dest2 (rthdr in the above case).
  402                  */
  403                 MAKE_CHAIN(exthdrs.ip6e_hbh, mprev, nexthdrp, IPPROTO_HOPOPTS);
  404                 MAKE_CHAIN(exthdrs.ip6e_dest1, mprev, nexthdrp,
  405                     IPPROTO_DSTOPTS);
  406                 MAKE_CHAIN(exthdrs.ip6e_rthdr, mprev, nexthdrp,
  407                     IPPROTO_ROUTING);
  408 
  409                 M_CSUM_DATA_IPv6_SET(m->m_pkthdr.csum_data,
  410                     sizeof(struct ip6_hdr) + optlen);
  411         }
  412 
  413         /* Need to save for pmtu */
  414         finaldst = ip6->ip6_dst;
  415 
  416         /*
  417          * If there is a routing header, replace destination address field
  418          * with the first hop of the routing header.
  419          */
  420         if (exthdrs.ip6e_rthdr) {
  421                 struct ip6_rthdr *rh;
  422 
  423                 rh = mtod(exthdrs.ip6e_rthdr, struct ip6_rthdr *);
  424 
  425                 error = ip6_handle_rthdr(rh, ip6);
  426                 if (error != 0) {
  427                         IP6_STATINC(IP6_STAT_ODROPPED);
  428                         goto bad;
  429                 }
  430         }
  431 
  432         /* Source address validation */
  433         if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src) &&
  434             (flags & IPV6_UNSPECSRC) == 0) {
  435                 error = EOPNOTSUPP;
  436                 IP6_STATINC(IP6_STAT_BADSCOPE);
  437                 goto bad;
  438         }
  439         if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src)) {
  440                 error = EOPNOTSUPP;
  441                 IP6_STATINC(IP6_STAT_BADSCOPE);
  442                 goto bad;
  443         }
  444 
  445         IP6_STATINC(IP6_STAT_LOCALOUT);
  446 
  447         /*
  448          * Route packet.
  449          */
  450         /* initialize cached route */
  451         if (ro == NULL) {
  452                 memset(&ip6route, 0, sizeof(ip6route));
  453                 ro = &ip6route;
  454         }
  455         ro_pmtu = ro;
  456         if (opt && opt->ip6po_rthdr)
  457                 ro = &opt->ip6po_route;
  458 
  459         /*
  460          * if specified, try to fill in the traffic class field.
  461          * do not override if a non-zero value is already set.
  462          * we check the diffserv field and the ecn field separately.
  463          */
  464         if (opt && opt->ip6po_tclass >= 0) {
  465                 int mask = 0;
  466 
  467                 if ((ip6->ip6_flow & htonl(0xfc << 20)) == 0)
  468                         mask |= 0xfc;
  469                 if ((ip6->ip6_flow & htonl(0x03 << 20)) == 0)
  470                         mask |= 0x03;
  471                 if (mask != 0)
  472                         ip6->ip6_flow |= htonl((opt->ip6po_tclass & mask) << 20);
  473         }
  474 
  475         /* fill in or override the hop limit field, if necessary. */
  476         if (opt && opt->ip6po_hlim != -1)
  477                 ip6->ip6_hlim = opt->ip6po_hlim & 0xff;
  478         else if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
  479                 if (im6o != NULL)
  480                         ip6->ip6_hlim = im6o->im6o_multicast_hlim;
  481                 else
  482                         ip6->ip6_hlim = ip6_defmcasthlim;
  483         }
  484 
  485 #ifdef IPSEC
  486         if (needipsec) {
  487                 error = ipsec6_process_packet(m, sp->req, flags);
  488 
  489                 /*
  490                  * Preserve KAME behaviour: ENOENT can be returned
  491                  * when an SA acquire is in progress.  Don't propagate
  492                  * this to user-level; it confuses applications.
  493                  * XXX this will go away when the SADB is redone.
  494                  */
  495                 if (error == ENOENT)
  496                         error = 0;
  497 
  498                 goto done;
  499         }
  500 #endif
  501 
  502         /* adjust pointer */
  503         ip6 = mtod(m, struct ip6_hdr *);
  504 
  505         sockaddr_in6_init(&dst_sa, &ip6->ip6_dst, 0, 0, 0);
  506 
  507         /* We do not need a route for multicast */
  508         if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
  509                 struct in6_pktinfo *pi = NULL;
  510 
  511                 /*
  512                  * If the outgoing interface for the address is specified by
  513                  * the caller, use it.
  514                  */
  515                 if (opt && (pi = opt->ip6po_pktinfo) != NULL) {
  516                         /* XXX boundary check is assumed to be already done. */
  517                         ifp = if_get_byindex(pi->ipi6_ifindex, &psref);
  518                 } else if (im6o != NULL) {
  519                         ifp = if_get_byindex(im6o->im6o_multicast_if_index,
  520                             &psref);
  521                 }
  522         }
  523 
  524         if (ifp == NULL) {
  525                 error = in6_selectroute(&dst_sa, opt, &ro, &rt, true);
  526                 if (error != 0)
  527                         goto bad;
  528                 ifp = if_get_byindex(rt->rt_ifp->if_index, &psref);
  529         }
  530 
  531         if (rt == NULL) {
  532                 /*
  533                  * If in6_selectroute() does not return a route entry,
  534                  * dst may not have been updated.
  535                  */
  536                 error = rtcache_setdst(ro, sin6tosa(&dst_sa));
  537                 if (error) {
  538                         IP6_STATINC(IP6_STAT_ODROPPED);
  539                         goto bad;
  540                 }
  541         }
  542 
  543         /*
  544          * then rt (for unicast) and ifp must be non-NULL valid values.
  545          */
  546         if ((flags & IPV6_FORWARDING) == 0) {
  547                 /* XXX: the FORWARDING flag can be set for mrouting. */
  548                 in6_ifstat_inc(ifp, ifs6_out_request);
  549         }
  550         if (rt != NULL) {
  551                 ia = (struct in6_ifaddr *)(rt->rt_ifa);
  552                 rt->rt_use++;
  553         }
  554 
  555         /*
  556          * The outgoing interface must be in the zone of source and
  557          * destination addresses.  We should use ia_ifp to support the
  558          * case of sending packets to an address of our own.
  559          */
  560         if (ia != NULL) {
  561                 origifp = ia->ia_ifp;
  562                 if (if_is_deactivated(origifp)) {
  563                         IP6_STATINC(IP6_STAT_ODROPPED);
  564                         goto bad;
  565                 }
  566                 if_acquire(origifp, &psref_ia);
  567                 release_psref_ia = true;
  568         } else
  569                 origifp = ifp;
  570 
  571         src0 = ip6->ip6_src;
  572         if (in6_setscope(&src0, origifp, &zone))
  573                 goto badscope;
  574         sockaddr_in6_init(&src_sa, &ip6->ip6_src, 0, 0, 0);
  575         if (sa6_recoverscope(&src_sa) || zone != src_sa.sin6_scope_id)
  576                 goto badscope;
  577 
  578         dst0 = ip6->ip6_dst;
  579         if (in6_setscope(&dst0, origifp, &zone))
  580                 goto badscope;
  581         /* re-initialize to be sure */
  582         sockaddr_in6_init(&dst_sa, &ip6->ip6_dst, 0, 0, 0);
  583         if (sa6_recoverscope(&dst_sa) || zone != dst_sa.sin6_scope_id)
  584                 goto badscope;
  585 
  586         /* scope check is done. */
  587 
  588         /* Ensure we only send from a valid address. */
  589         if ((ifp->if_flags & IFF_LOOPBACK) == 0 &&
  590             (flags & IPV6_FORWARDING) == 0 &&
  591             (error = ip6_ifaddrvalid(&src0, &dst0)) != 0)
  592         {
  593                 char ip6buf[INET6_ADDRSTRLEN];
  594                 nd6log(LOG_ERR,
  595                     "refusing to send from invalid address %s (pid %d)\n",
  596                     IN6_PRINT(ip6buf, &src0), curproc->p_pid);
  597                 IP6_STATINC(IP6_STAT_ODROPPED);
  598                 in6_ifstat_inc(origifp, ifs6_out_discard);
  599                 if (error == 1)
  600                         /*
  601                          * Address exists, but is tentative or detached.
  602                          * We can't send from it because it's invalid,
  603                          * so we drop the packet.
  604                          */
  605                         error = 0;
  606                 else
  607                         error = EADDRNOTAVAIL;
  608                 goto bad;
  609         }
  610 
  611         if (rt != NULL && (rt->rt_flags & RTF_GATEWAY) &&
  612             !IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst))
  613                 dst = satocsin6(rt->rt_gateway);
  614         else
  615                 dst = satocsin6(rtcache_getdst(ro));
  616 
  617         /*
  618          * XXXXXX: original code follows:
  619          */
  620         if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst))
  621                 m->m_flags &= ~(M_BCAST | M_MCAST);     /* just in case */
  622         else {
  623                 bool ingroup;
  624 
  625                 m->m_flags = (m->m_flags & ~M_BCAST) | M_MCAST;
  626 
  627                 in6_ifstat_inc(ifp, ifs6_out_mcast);
  628 
  629                 /*
  630                  * Confirm that the outgoing interface supports multicast.
  631                  */
  632                 if (!(ifp->if_flags & IFF_MULTICAST)) {
  633                         IP6_STATINC(IP6_STAT_NOROUTE);
  634                         in6_ifstat_inc(ifp, ifs6_out_discard);
  635                         error = ENETUNREACH;
  636                         goto bad;
  637                 }
  638 
  639                 ingroup = in6_multi_group(&ip6->ip6_dst, ifp);
  640                 if (ingroup && (im6o == NULL || im6o->im6o_multicast_loop)) {
  641                         /*
  642                          * If we belong to the destination multicast group
  643                          * on the outgoing interface, and the caller did not
  644                          * forbid loopback, loop back a copy.
  645                          */
  646                         KASSERT(dst != NULL);
  647                         ip6_mloopback(ifp, m, dst);
  648                 } else {
  649                         /*
  650                          * If we are acting as a multicast router, perform
  651                          * multicast forwarding as if the packet had just
  652                          * arrived on the interface to which we are about
  653                          * to send.  The multicast forwarding function
  654                          * recursively calls this function, using the
  655                          * IPV6_FORWARDING flag to prevent infinite recursion.
  656                          *
  657                          * Multicasts that are looped back by ip6_mloopback(),
  658                          * above, will be forwarded by the ip6_input() routine,
  659                          * if necessary.
  660                          */
  661                         if (ip6_mrouter && (flags & IPV6_FORWARDING) == 0) {
  662                                 if (ip6_mforward(ip6, ifp, m) != 0) {
  663                                         m_freem(m);
  664                                         goto done;
  665                                 }
  666                         }
  667                 }
  668                 /*
  669                  * Multicasts with a hoplimit of zero may be looped back,
  670                  * above, but must not be transmitted on a network.
  671                  * Also, multicasts addressed to the loopback interface
  672                  * are not sent -- the above call to ip6_mloopback() will
  673                  * loop back a copy if this host actually belongs to the
  674                  * destination group on the loopback interface.
  675                  */
  676                 if (ip6->ip6_hlim == 0 || (ifp->if_flags & IFF_LOOPBACK) ||
  677                     IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst)) {
  678                         m_freem(m);
  679                         goto done;
  680                 }
  681         }
  682 
  683         /*
  684          * Fill the outgoing interface to tell the upper layer
  685          * to increment per-interface statistics.
  686          */
  687         if (ifpp)
  688                 *ifpp = ifp;
  689 
  690         /* Determine path MTU. */
  691         /*
  692          * ro_pmtu represent final destination while
  693          * ro might represent immediate destination.
  694          * Use ro_pmtu destination since MTU might differ.
  695          */
  696         if (ro_pmtu != ro) {
  697                 union {
  698                         struct sockaddr         dst;
  699                         struct sockaddr_in6     dst6;
  700                 } u;
  701 
  702                 /* ro_pmtu may not have a cache */
  703                 sockaddr_in6_init(&u.dst6, &finaldst, 0, 0, 0);
  704                 rt_pmtu = rtcache_lookup(ro_pmtu, &u.dst);
  705         } else
  706                 rt_pmtu = rt;
  707         error = ip6_getpmtu(rt_pmtu, ifp, &mtu, &alwaysfrag);
  708         if (rt_pmtu != NULL && rt_pmtu != rt)
  709                 rtcache_unref(rt_pmtu, ro_pmtu);
  710         KASSERT(error == 0); /* ip6_getpmtu never fail if ifp is passed */
  711 
  712         /*
  713          * The caller of this function may specify to use the minimum MTU
  714          * in some cases.
  715          * An advanced API option (IPV6_USE_MIN_MTU) can also override MTU
  716          * setting.  The logic is a bit complicated; by default, unicast
  717          * packets will follow path MTU while multicast packets will be sent at
  718          * the minimum MTU.  If IP6PO_MINMTU_ALL is specified, all packets
  719          * including unicast ones will be sent at the minimum MTU.  Multicast
  720          * packets will always be sent at the minimum MTU unless
  721          * IP6PO_MINMTU_DISABLE is explicitly specified.
  722          * See RFC 3542 for more details.
  723          */
  724         if (mtu > IPV6_MMTU) {
  725                 if ((flags & IPV6_MINMTU))
  726                         mtu = IPV6_MMTU;
  727                 else if (opt && opt->ip6po_minmtu == IP6PO_MINMTU_ALL)
  728                         mtu = IPV6_MMTU;
  729                 else if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
  730                          (opt == NULL ||
  731                           opt->ip6po_minmtu != IP6PO_MINMTU_DISABLE)) {
  732                         mtu = IPV6_MMTU;
  733                 }
  734         }
  735 
  736         /*
  737          * clear embedded scope identifiers if necessary.
  738          * in6_clearscope will touch the addresses only when necessary.
  739          */
  740         in6_clearscope(&ip6->ip6_src);
  741         in6_clearscope(&ip6->ip6_dst);
  742 
  743         /*
  744          * If the outgoing packet contains a hop-by-hop options header,
  745          * it must be examined and processed even by the source node.
  746          * (RFC 2460, section 4.)
  747          *
  748          * XXX Is this really necessary?
  749          */
  750         if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
  751                 u_int32_t dummy1; /* XXX unused */
  752                 u_int32_t dummy2; /* XXX unused */
  753                 int hoff = sizeof(struct ip6_hdr);
  754 
  755                 if (ip6_hopopts_input(&dummy1, &dummy2, &m, &hoff)) {
  756                         /* m was already freed at this point */
  757                         error = EINVAL;
  758                         goto done;
  759                 }
  760 
  761                 ip6 = mtod(m, struct ip6_hdr *);
  762         }
  763 
  764         /*
  765          * Run through list of hooks for output packets.
  766          */
  767         error = pfil_run_hooks(inet6_pfil_hook, &m, ifp, PFIL_OUT);
  768         if (error != 0 || m == NULL) {
  769                 IP6_STATINC(IP6_STAT_PFILDROP_OUT);
  770                 goto done;
  771         }
  772         ip6 = mtod(m, struct ip6_hdr *);
  773 
  774         /*
  775          * Send the packet to the outgoing interface.
  776          * If necessary, do IPv6 fragmentation before sending.
  777          *
  778          * the logic here is rather complex:
  779          * 1: normal case (dontfrag == 0, alwaysfrag == 0)
  780          * 1-a: send as is if tlen <= path mtu
  781          * 1-b: fragment if tlen > path mtu
  782          *
  783          * 2: if user asks us not to fragment (dontfrag == 1)
  784          * 2-a: send as is if tlen <= interface mtu
  785          * 2-b: error if tlen > interface mtu
  786          *
  787          * 3: if we always need to attach fragment header (alwaysfrag == 1)
  788          *      always fragment
  789          *
  790          * 4: if dontfrag == 1 && alwaysfrag == 1
  791          *      error, as we cannot handle this conflicting request
  792          */
  793         tlen = m->m_pkthdr.len;
  794         tso = (m->m_pkthdr.csum_flags & M_CSUM_TSOv6) != 0;
  795         if (opt && (opt->ip6po_flags & IP6PO_DONTFRAG))
  796                 dontfrag = 1;
  797         else
  798                 dontfrag = 0;
  799 
  800         if (dontfrag && alwaysfrag) {   /* case 4 */
  801                 /* conflicting request - can't transmit */
  802                 IP6_STATINC(IP6_STAT_CANTFRAG);
  803                 error = EMSGSIZE;
  804                 goto bad;
  805         }
  806         if (dontfrag && (!tso && tlen > ifp->if_mtu)) { /* case 2-b */
  807                 /*
  808                  * Even if the DONTFRAG option is specified, we cannot send the
  809                  * packet when the data length is larger than the MTU of the
  810                  * outgoing interface.
  811                  * Notify the error by sending IPV6_PATHMTU ancillary data as
  812                  * well as returning an error code (the latter is not described
  813                  * in the API spec.)
  814                  */
  815                 u_int32_t mtu32;
  816                 struct ip6ctlparam ip6cp;
  817 
  818                 mtu32 = (u_int32_t)mtu;
  819                 memset(&ip6cp, 0, sizeof(ip6cp));
  820                 ip6cp.ip6c_cmdarg = (void *)&mtu32;
  821                 pfctlinput2(PRC_MSGSIZE,
  822                     rtcache_getdst(ro_pmtu), &ip6cp);
  823 
  824                 IP6_STATINC(IP6_STAT_CANTFRAG);
  825                 error = EMSGSIZE;
  826                 goto bad;
  827         }
  828 
  829         /*
  830          * transmit packet without fragmentation
  831          */
  832         if (dontfrag || (!alwaysfrag && (tlen <= mtu || tso))) {
  833                 /* case 1-a and 2-a */
  834                 struct in6_ifaddr *ia6;
  835                 int sw_csum;
  836                 int s;
  837 
  838                 ip6 = mtod(m, struct ip6_hdr *);
  839                 s = pserialize_read_enter();
  840                 ia6 = in6_ifawithifp(ifp, &ip6->ip6_src);
  841                 if (ia6) {
  842                         /* Record statistics for this interface address. */
  843                         ia6->ia_ifa.ifa_data.ifad_outbytes += m->m_pkthdr.len;
  844                 }
  845                 pserialize_read_exit(s);
  846 
  847                 sw_csum = m->m_pkthdr.csum_flags & ~ifp->if_csum_flags_tx;
  848                 if ((sw_csum & (M_CSUM_UDPv6|M_CSUM_TCPv6)) != 0) {
  849                         if (IN6_NEED_CHECKSUM(ifp,
  850                             sw_csum & (M_CSUM_UDPv6|M_CSUM_TCPv6))) {
  851                                 in6_undefer_cksum_tcpudp(m);
  852                         }
  853                         m->m_pkthdr.csum_flags &= ~(M_CSUM_UDPv6|M_CSUM_TCPv6);
  854                 }
  855 
  856                 KASSERT(dst != NULL);
  857                 if (__predict_false(sw_csum & M_CSUM_TSOv6)) {
  858                         /*
  859                          * TSO6 is required by a packet, but disabled for
  860                          * the interface.
  861                          */
  862                         error = ip6_tso_output(ifp, origifp, m, dst, rt);
  863                 } else
  864                         error = ip6_if_output(ifp, origifp, m, dst, rt);
  865                 goto done;
  866         }
  867 
  868         if (tso) {
  869                 IP6_STATINC(IP6_STAT_CANTFRAG); /* XXX */
  870                 error = EINVAL; /* XXX */
  871                 goto bad;
  872         }
  873 
  874         /*
  875          * try to fragment the packet.  case 1-b and 3
  876          */
  877         if (mtu < IPV6_MMTU) {
  878                 /* path MTU cannot be less than IPV6_MMTU */
  879                 IP6_STATINC(IP6_STAT_CANTFRAG);
  880                 error = EMSGSIZE;
  881                 in6_ifstat_inc(ifp, ifs6_out_fragfail);
  882                 goto bad;
  883         } else if (ip6->ip6_plen == 0) {
  884                 /* jumbo payload cannot be fragmented */
  885                 IP6_STATINC(IP6_STAT_CANTFRAG);
  886                 error = EMSGSIZE;
  887                 in6_ifstat_inc(ifp, ifs6_out_fragfail);
  888                 goto bad;
  889         } else {
  890                 const uint32_t id = ip6_randomid();
  891                 struct mbuf **mnext, *m_frgpart;
  892                 const int hlen = unfragpartlen;
  893                 struct ip6_frag *ip6f;
  894                 u_char nextproto;
  895 
  896                 if (mtu > IPV6_MAXPACKET)
  897                         mtu = IPV6_MAXPACKET;
  898 
  899                 /*
  900                  * Must be able to put at least 8 bytes per fragment.
  901                  */
  902                 len = (mtu - hlen - sizeof(struct ip6_frag)) & ~7;
  903                 if (len < 8) {
  904                         IP6_STATINC(IP6_STAT_CANTFRAG);
  905                         error = EMSGSIZE;
  906                         in6_ifstat_inc(ifp, ifs6_out_fragfail);
  907                         goto bad;
  908                 }
  909 
  910                 mnext = &m->m_nextpkt;
  911 
  912                 /*
  913                  * Change the next header field of the last header in the
  914                  * unfragmentable part.
  915                  */
  916                 if (exthdrs.ip6e_rthdr) {
  917                         nextproto = *mtod(exthdrs.ip6e_rthdr, u_char *);
  918                         *mtod(exthdrs.ip6e_rthdr, u_char *) = IPPROTO_FRAGMENT;
  919                 } else if (exthdrs.ip6e_dest1) {
  920                         nextproto = *mtod(exthdrs.ip6e_dest1, u_char *);
  921                         *mtod(exthdrs.ip6e_dest1, u_char *) = IPPROTO_FRAGMENT;
  922                 } else if (exthdrs.ip6e_hbh) {
  923                         nextproto = *mtod(exthdrs.ip6e_hbh, u_char *);
  924                         *mtod(exthdrs.ip6e_hbh, u_char *) = IPPROTO_FRAGMENT;
  925                 } else {
  926                         nextproto = ip6->ip6_nxt;
  927                         ip6->ip6_nxt = IPPROTO_FRAGMENT;
  928                 }
  929 
  930                 if ((m->m_pkthdr.csum_flags & (M_CSUM_UDPv6|M_CSUM_TCPv6))
  931                     != 0) {
  932                         if (IN6_NEED_CHECKSUM(ifp,
  933                             m->m_pkthdr.csum_flags &
  934                             (M_CSUM_UDPv6|M_CSUM_TCPv6))) {
  935                                 in6_undefer_cksum_tcpudp(m);
  936                         }
  937                         m->m_pkthdr.csum_flags &= ~(M_CSUM_UDPv6|M_CSUM_TCPv6);
  938                 }
  939 
  940                 /*
  941                  * Loop through length of segment after first fragment,
  942                  * make new header and copy data of each part and link onto
  943                  * chain.
  944                  */
  945                 m0 = m;
  946                 for (off = hlen; off < tlen; off += len) {
  947                         struct mbuf *mlast;
  948 
  949                         MGETHDR(m, M_DONTWAIT, MT_HEADER);
  950                         if (!m) {
  951                                 error = ENOBUFS;
  952                                 IP6_STATINC(IP6_STAT_ODROPPED);
  953                                 goto sendorfree;
  954                         }
  955                         m_reset_rcvif(m);
  956                         m->m_flags = m0->m_flags & M_COPYFLAGS;
  957                         *mnext = m;
  958                         mnext = &m->m_nextpkt;
  959                         m->m_data += max_linkhdr;
  960                         mhip6 = mtod(m, struct ip6_hdr *);
  961                         *mhip6 = *ip6;
  962                         m->m_len = sizeof(*mhip6);
  963 
  964                         ip6f = NULL;
  965                         error = ip6_insertfraghdr(m0, m, hlen, &ip6f);
  966                         if (error) {
  967                                 IP6_STATINC(IP6_STAT_ODROPPED);
  968                                 goto sendorfree;
  969                         }
  970 
  971                         /* Fill in the Frag6 Header */
  972                         ip6f->ip6f_offlg = htons((u_int16_t)((off - hlen) & ~7));
  973                         if (off + len >= tlen)
  974                                 len = tlen - off;
  975                         else
  976                                 ip6f->ip6f_offlg |= IP6F_MORE_FRAG;
  977                         ip6f->ip6f_reserved = 0;
  978                         ip6f->ip6f_ident = id;
  979                         ip6f->ip6f_nxt = nextproto;
  980 
  981                         mhip6->ip6_plen = htons((u_int16_t)(len + hlen +
  982                             sizeof(*ip6f) - sizeof(struct ip6_hdr)));
  983                         if ((m_frgpart = m_copym(m0, off, len, M_DONTWAIT)) == NULL) {
  984                                 error = ENOBUFS;
  985                                 IP6_STATINC(IP6_STAT_ODROPPED);
  986                                 goto sendorfree;
  987                         }
  988                         for (mlast = m; mlast->m_next; mlast = mlast->m_next)
  989                                 ;
  990                         mlast->m_next = m_frgpart;
  991 
  992                         m->m_pkthdr.len = len + hlen + sizeof(*ip6f);
  993                         m_reset_rcvif(m);
  994                         IP6_STATINC(IP6_STAT_OFRAGMENTS);
  995                         in6_ifstat_inc(ifp, ifs6_out_fragcreat);
  996                 }
  997 
  998                 in6_ifstat_inc(ifp, ifs6_out_fragok);
  999         }
 1000 
 1001 sendorfree:
 1002         m = m0->m_nextpkt;
 1003         m0->m_nextpkt = 0;
 1004         m_freem(m0);
 1005         for (m0 = m; m; m = m0) {
 1006                 m0 = m->m_nextpkt;
 1007                 m->m_nextpkt = 0;
 1008                 if (error == 0) {
 1009                         struct in6_ifaddr *ia6;
 1010                         int s;
 1011                         ip6 = mtod(m, struct ip6_hdr *);
 1012                         s = pserialize_read_enter();
 1013                         ia6 = in6_ifawithifp(ifp, &ip6->ip6_src);
 1014                         if (ia6) {
 1015                                 /*
 1016                                  * Record statistics for this interface
 1017                                  * address.
 1018                                  */
 1019                                 ia6->ia_ifa.ifa_data.ifad_outbytes +=
 1020                                     m->m_pkthdr.len;
 1021                         }
 1022                         pserialize_read_exit(s);
 1023                         KASSERT(dst != NULL);
 1024                         error = ip6_if_output(ifp, origifp, m, dst, rt);
 1025                 } else
 1026                         m_freem(m);
 1027         }
 1028 
 1029         if (error == 0)
 1030                 IP6_STATINC(IP6_STAT_FRAGMENTED);
 1031 
 1032 done:
 1033         rtcache_unref(rt, ro);
 1034         if (ro == &ip6route)
 1035                 rtcache_free(&ip6route);
 1036 #ifdef IPSEC
 1037         if (sp != NULL)
 1038                 KEY_SP_UNREF(&sp);
 1039 #endif
 1040         if_put(ifp, &psref);
 1041         if (release_psref_ia)
 1042                 if_put(origifp, &psref_ia);
 1043         curlwp_bindx(bound);
 1044 
 1045         return error;
 1046 
 1047 freehdrs:
 1048         m_freem(exthdrs.ip6e_hbh);
 1049         m_freem(exthdrs.ip6e_dest1);
 1050         m_freem(exthdrs.ip6e_rthdr);
 1051         m_freem(exthdrs.ip6e_dest2);
 1052         /* FALLTHROUGH */
 1053 bad:
 1054         m_freem(m);
 1055         goto done;
 1056 
 1057 badscope:
 1058         IP6_STATINC(IP6_STAT_BADSCOPE);
 1059         in6_ifstat_inc(origifp, ifs6_out_discard);
 1060         if (error == 0)
 1061                 error = EHOSTUNREACH; /* XXX */
 1062         goto bad;
 1063 }
 1064 
 1065 static int
 1066 ip6_copyexthdr(struct mbuf **mp, void *hdr, int hlen)
 1067 {
 1068         struct mbuf *m;
 1069 
 1070         if (hlen > MCLBYTES)
 1071                 return ENOBUFS; /* XXX */
 1072 
 1073         MGET(m, M_DONTWAIT, MT_DATA);
 1074         if (!m)
 1075                 return ENOBUFS;
 1076 
 1077         if (hlen > MLEN) {
 1078                 MCLGET(m, M_DONTWAIT);
 1079                 if ((m->m_flags & M_EXT) == 0) {
 1080                         m_free(m);
 1081                         return ENOBUFS;
 1082                 }
 1083         }
 1084         m->m_len = hlen;
 1085         if (hdr)
 1086                 memcpy(mtod(m, void *), hdr, hlen);
 1087 
 1088         *mp = m;
 1089         return 0;
 1090 }
 1091 
 1092 /*
 1093  * Insert jumbo payload option.
 1094  */
 1095 static int
 1096 ip6_insert_jumboopt(struct ip6_exthdrs *exthdrs, u_int32_t plen)
 1097 {
 1098         struct mbuf *mopt;
 1099         u_int8_t *optbuf;
 1100         u_int32_t v;
 1101 
 1102 #define JUMBOOPTLEN     8       /* length of jumbo payload option and padding */
 1103 
 1104         /*
 1105          * If there is no hop-by-hop options header, allocate new one.
 1106          * If there is one but it doesn't have enough space to store the
 1107          * jumbo payload option, allocate a cluster to store the whole options.
 1108          * Otherwise, use it to store the options.
 1109          */
 1110         if (exthdrs->ip6e_hbh == NULL) {
 1111                 MGET(mopt, M_DONTWAIT, MT_DATA);
 1112                 if (mopt == 0)
 1113                         return (ENOBUFS);
 1114                 mopt->m_len = JUMBOOPTLEN;
 1115                 optbuf = mtod(mopt, u_int8_t *);
 1116                 optbuf[1] = 0;  /* = ((JUMBOOPTLEN) >> 3) - 1 */
 1117                 exthdrs->ip6e_hbh = mopt;
 1118         } else {
 1119                 struct ip6_hbh *hbh;
 1120 
 1121                 mopt = exthdrs->ip6e_hbh;
 1122                 if (M_TRAILINGSPACE(mopt) < JUMBOOPTLEN) {
 1123                         const int oldoptlen = mopt->m_len;
 1124                         struct mbuf *n;
 1125 
 1126                         /*
 1127                          * Assumptions:
 1128                          * - exthdrs->ip6e_hbh is not referenced from places
 1129                          *   other than exthdrs.
 1130                          * - exthdrs->ip6e_hbh is not an mbuf chain.
 1131                          */
 1132                         KASSERT(mopt->m_next == NULL);
 1133 
 1134                         /*
 1135                          * Give up if the whole (new) hbh header does not fit
 1136                          * even in an mbuf cluster.
 1137                          */
 1138                         if (oldoptlen + JUMBOOPTLEN > MCLBYTES)
 1139                                 return ENOBUFS;
 1140 
 1141                         /*
 1142                          * At this point, we must always prepare a cluster.
 1143                          */
 1144                         MGET(n, M_DONTWAIT, MT_DATA);
 1145                         if (n) {
 1146                                 MCLGET(n, M_DONTWAIT);
 1147                                 if ((n->m_flags & M_EXT) == 0) {
 1148                                         m_freem(n);
 1149                                         n = NULL;
 1150                                 }
 1151                         }
 1152                         if (!n)
 1153                                 return ENOBUFS;
 1154 
 1155                         n->m_len = oldoptlen + JUMBOOPTLEN;
 1156                         bcopy(mtod(mopt, void *), mtod(n, void *),
 1157                             oldoptlen);
 1158                         optbuf = mtod(n, u_int8_t *) + oldoptlen;
 1159                         m_freem(mopt);
 1160                         mopt = exthdrs->ip6e_hbh = n;
 1161                 } else {
 1162                         optbuf = mtod(mopt, u_int8_t *) + mopt->m_len;
 1163                         mopt->m_len += JUMBOOPTLEN;
 1164                 }
 1165                 optbuf[0] = IP6OPT_PADN;
 1166                 optbuf[1] = 0;
 1167 
 1168                 /*
 1169                  * Adjust the header length according to the pad and
 1170                  * the jumbo payload option.
 1171                  */
 1172                 hbh = mtod(mopt, struct ip6_hbh *);
 1173                 hbh->ip6h_len += (JUMBOOPTLEN >> 3);
 1174         }
 1175 
 1176         /* fill in the option. */
 1177         optbuf[2] = IP6OPT_JUMBO;
 1178         optbuf[3] = 4;
 1179         v = (u_int32_t)htonl(plen + JUMBOOPTLEN);
 1180         memcpy(&optbuf[4], &v, sizeof(u_int32_t));
 1181 
 1182         /* finally, adjust the packet header length */
 1183         exthdrs->ip6e_ip6->m_pkthdr.len += JUMBOOPTLEN;
 1184 
 1185         return 0;
 1186 #undef JUMBOOPTLEN
 1187 }
 1188 
 1189 /*
 1190  * Insert fragment header and copy unfragmentable header portions.
 1191  *
 1192  * *frghdrp will not be read, and it is guaranteed that either an
 1193  * error is returned or that *frghdrp will point to space allocated
 1194  * for the fragment header.
 1195  *
 1196  * On entry, m contains:
 1197  *     IPv6 Header
 1198  * On exit, it contains:
 1199  *     IPv6 Header -> Unfragmentable Part -> Frag6 Header
 1200  */
 1201 static int
 1202 ip6_insertfraghdr(struct mbuf *m0, struct mbuf *m, int hlen,
 1203         struct ip6_frag **frghdrp)
 1204 {
 1205         struct mbuf *n, *mlast;
 1206 
 1207         if (hlen > sizeof(struct ip6_hdr)) {
 1208                 n = m_copym(m0, sizeof(struct ip6_hdr),
 1209                     hlen - sizeof(struct ip6_hdr), M_DONTWAIT);
 1210                 if (n == NULL)
 1211                         return ENOBUFS;
 1212                 m->m_next = n;
 1213         } else
 1214                 n = m;
 1215 
 1216         /* Search for the last mbuf of unfragmentable part. */
 1217         for (mlast = n; mlast->m_next; mlast = mlast->m_next)
 1218                 ;
 1219 
 1220         if ((mlast->m_flags & M_EXT) == 0 &&
 1221             M_TRAILINGSPACE(mlast) >= sizeof(struct ip6_frag)) {
 1222                 /* use the trailing space of the last mbuf for the fragment hdr */
 1223                 *frghdrp = (struct ip6_frag *)(mtod(mlast, char *) +
 1224                     mlast->m_len);
 1225                 mlast->m_len += sizeof(struct ip6_frag);
 1226         } else {
 1227                 /* allocate a new mbuf for the fragment header */
 1228                 struct mbuf *mfrg;
 1229 
 1230                 MGET(mfrg, M_DONTWAIT, MT_DATA);
 1231                 if (mfrg == NULL)
 1232                         return ENOBUFS;
 1233                 mfrg->m_len = sizeof(struct ip6_frag);
 1234                 *frghdrp = mtod(mfrg, struct ip6_frag *);
 1235                 mlast->m_next = mfrg;
 1236         }
 1237 
 1238         return 0;
 1239 }
 1240 
 1241 static int
 1242 ip6_getpmtu(struct rtentry *rt, struct ifnet *ifp, u_long *mtup,
 1243     int *alwaysfragp)
 1244 {
 1245         u_int32_t mtu = 0;
 1246         int alwaysfrag = 0;
 1247         int error = 0;
 1248 
 1249         if (rt != NULL) {
 1250                 if (ifp == NULL)
 1251                         ifp = rt->rt_ifp;
 1252                 mtu = rt->rt_rmx.rmx_mtu;
 1253                 if (mtu == 0)
 1254                         mtu = ifp->if_mtu;
 1255                 else if (mtu < IPV6_MMTU) {
 1256                         /*
 1257                          * RFC2460 section 5, last paragraph:
 1258                          * if we record ICMPv6 too big message with
 1259                          * mtu < IPV6_MMTU, transmit packets sized IPV6_MMTU
 1260                          * or smaller, with fragment header attached.
 1261                          * (fragment header is needed regardless from the
 1262                          * packet size, for translators to identify packets)
 1263                          */
 1264                         alwaysfrag = 1;
 1265                         mtu = IPV6_MMTU;
 1266                 } else if (mtu > ifp->if_mtu) {
 1267                         /*
 1268                          * The MTU on the route is larger than the MTU on
 1269                          * the interface!  This shouldn't happen, unless the
 1270                          * MTU of the interface has been changed after the
 1271                          * interface was brought up.  Change the MTU in the
 1272                          * route to match the interface MTU (as long as the
 1273                          * field isn't locked).
 1274                          */
 1275                         mtu = ifp->if_mtu;
 1276                         if (!(rt->rt_rmx.rmx_locks & RTV_MTU))
 1277                                 rt->rt_rmx.rmx_mtu = mtu;
 1278                 }
 1279         } else if (ifp) {
 1280                 mtu = ifp->if_mtu;
 1281         } else
 1282                 error = EHOSTUNREACH; /* XXX */
 1283 
 1284         *mtup = mtu;
 1285         if (alwaysfragp)
 1286                 *alwaysfragp = alwaysfrag;
 1287         return (error);
 1288 }
 1289 
 1290 /*
 1291  * IP6 socket option processing.
 1292  */
 1293 int
 1294 ip6_ctloutput(int op, struct socket *so, struct sockopt *sopt)
 1295 {
 1296         int optdatalen, uproto;
 1297         void *optdata;
 1298         struct inpcb *inp = sotoinpcb(so);
 1299         struct ip_moptions **mopts;
 1300         int error, optval;
 1301         int level, optname;
 1302 
 1303         KASSERT(solocked(so));
 1304         KASSERT(sopt != NULL);
 1305 
 1306         level = sopt->sopt_level;
 1307         optname = sopt->sopt_name;
 1308 
 1309         error = optval = 0;
 1310         uproto = (int)so->so_proto->pr_protocol;
 1311 
 1312         switch (level) {
 1313         case IPPROTO_IP:
 1314                 switch (optname) {
 1315                 case IP_ADD_MEMBERSHIP:
 1316                 case IP_DROP_MEMBERSHIP:
 1317                 case IP_MULTICAST_IF:
 1318                 case IP_MULTICAST_LOOP:
 1319                 case IP_MULTICAST_TTL:
 1320                         mopts = &inp->inp_moptions;
 1321                         switch (op) {
 1322                         case PRCO_GETOPT:
 1323                                 return ip_getmoptions(*mopts, sopt);
 1324                         case PRCO_SETOPT:
 1325                                 return ip_setmoptions(mopts, sopt);
 1326                         default:
 1327                                 return EINVAL;
 1328                         }
 1329                 default:
 1330                         return ENOPROTOOPT;
 1331                 }
 1332         case IPPROTO_IPV6:
 1333                 break;
 1334         default:
 1335                 return ENOPROTOOPT;
 1336         }
 1337         switch (op) {
 1338         case PRCO_SETOPT:
 1339                 switch (optname) {
 1340 #ifdef RFC2292
 1341                 case IPV6_2292PKTOPTIONS:
 1342                         error = ip6_pcbopts(&in6p_outputopts(inp), so, sopt);
 1343                         break;
 1344 #endif
 1345 
 1346                 /*
 1347                  * Use of some Hop-by-Hop options or some
 1348                  * Destination options, might require special
 1349                  * privilege.  That is, normal applications
 1350                  * (without special privilege) might be forbidden
 1351                  * from setting certain options in outgoing packets,
 1352                  * and might never see certain options in received
 1353                  * packets. [RFC 2292 Section 6]
 1354                  * KAME specific note:
 1355                  *  KAME prevents non-privileged users from sending or
 1356                  *  receiving ANY hbh/dst options in order to avoid
 1357                  *  overhead of parsing options in the kernel.
 1358                  */
 1359                 case IPV6_RECVHOPOPTS:
 1360                 case IPV6_RECVDSTOPTS:
 1361                 case IPV6_RECVRTHDRDSTOPTS:
 1362                         error = kauth_authorize_network(
 1363                             kauth_cred_get(),
 1364                             KAUTH_NETWORK_IPV6, KAUTH_REQ_NETWORK_IPV6_HOPBYHOP,
 1365                             NULL, NULL, NULL);
 1366                         if (error)
 1367                                 break;
 1368                         /* FALLTHROUGH */
 1369                 case IPV6_UNICAST_HOPS:
 1370                 case IPV6_HOPLIMIT:
 1371                 case IPV6_FAITH:
 1372 
 1373                 case IPV6_RECVPKTINFO:
 1374                 case IPV6_RECVHOPLIMIT:
 1375                 case IPV6_RECVRTHDR:
 1376                 case IPV6_RECVPATHMTU:
 1377                 case IPV6_RECVTCLASS:
 1378                 case IPV6_V6ONLY:
 1379                 case IPV6_BINDANY:
 1380                         error = sockopt_getint(sopt, &optval);
 1381                         if (error)
 1382                                 break;
 1383                         switch (optname) {
 1384                         case IPV6_UNICAST_HOPS:
 1385                                 if (optval < -1 || optval >= 256)
 1386                                         error = EINVAL;
 1387                                 else {
 1388                                         /* -1 = kernel default */
 1389                                         in6p_hops6(inp) = optval;
 1390                                 }
 1391                                 break;
 1392 #define OPTSET(bit) \
 1393 do { \
 1394 if (optval) \
 1395         inp->inp_flags |= (bit); \
 1396 else \
 1397         inp->inp_flags &= ~(bit); \
 1398 } while (/*CONSTCOND*/ 0)
 1399 
 1400 #ifdef RFC2292
 1401 #define OPTSET2292(bit)                         \
 1402 do {                                            \
 1403 inp->inp_flags |= IN6P_RFC2292;         \
 1404 if (optval)                             \
 1405         inp->inp_flags |= (bit);        \
 1406 else                                    \
 1407         inp->inp_flags &= ~(bit);       \
 1408 } while (/*CONSTCOND*/ 0)
 1409 #endif
 1410 
 1411 #define OPTBIT(bit) (inp->inp_flags & (bit) ? 1 : 0)
 1412 
 1413                         case IPV6_RECVPKTINFO:
 1414 #ifdef RFC2292
 1415                                 /* cannot mix with RFC2292 */
 1416                                 if (OPTBIT(IN6P_RFC2292)) {
 1417                                         error = EINVAL;
 1418                                         break;
 1419                                 }
 1420 #endif
 1421                                 OPTSET(IN6P_PKTINFO);
 1422                                 break;
 1423 
 1424                         case IPV6_HOPLIMIT:
 1425                         {
 1426                                 struct ip6_pktopts **optp;
 1427 
 1428 #ifdef RFC2292
 1429                                 /* cannot mix with RFC2292 */
 1430                                 if (OPTBIT(IN6P_RFC2292)) {
 1431                                         error = EINVAL;
 1432                                         break;
 1433                                 }
 1434 #endif
 1435                                 optp = &in6p_outputopts(inp);
 1436                                 error = ip6_pcbopt(IPV6_HOPLIMIT,
 1437                                                    (u_char *)&optval,
 1438                                                    sizeof(optval),
 1439                                                    optp,
 1440                                                    kauth_cred_get(), uproto);
 1441                                 break;
 1442                         }
 1443 
 1444                         case IPV6_RECVHOPLIMIT:
 1445 #ifdef RFC2292
 1446                                 /* cannot mix with RFC2292 */
 1447                                 if (OPTBIT(IN6P_RFC2292)) {
 1448                                         error = EINVAL;
 1449                                         break;
 1450                                 }
 1451 #endif
 1452                                 OPTSET(IN6P_HOPLIMIT);
 1453                                 break;
 1454 
 1455                         case IPV6_RECVHOPOPTS:
 1456 #ifdef RFC2292
 1457                                 /* cannot mix with RFC2292 */
 1458                                 if (OPTBIT(IN6P_RFC2292)) {
 1459                                         error = EINVAL;
 1460                                         break;
 1461                                 }
 1462 #endif
 1463                                 OPTSET(IN6P_HOPOPTS);
 1464                                 break;
 1465 
 1466                         case IPV6_RECVDSTOPTS:
 1467 #ifdef RFC2292
 1468                                 /* cannot mix with RFC2292 */
 1469                                 if (OPTBIT(IN6P_RFC2292)) {
 1470                                         error = EINVAL;
 1471                                         break;
 1472                                 }
 1473 #endif
 1474                                 OPTSET(IN6P_DSTOPTS);
 1475                                 break;
 1476 
 1477                         case IPV6_RECVRTHDRDSTOPTS:
 1478 #ifdef RFC2292
 1479                                 /* cannot mix with RFC2292 */
 1480                                 if (OPTBIT(IN6P_RFC2292)) {
 1481                                         error = EINVAL;
 1482                                         break;
 1483                                 }
 1484 #endif
 1485                                 OPTSET(IN6P_RTHDRDSTOPTS);
 1486                                 break;
 1487 
 1488                         case IPV6_RECVRTHDR:
 1489 #ifdef RFC2292
 1490                                 /* cannot mix with RFC2292 */
 1491                                 if (OPTBIT(IN6P_RFC2292)) {
 1492                                         error = EINVAL;
 1493                                         break;
 1494                                 }
 1495 #endif
 1496                                 OPTSET(IN6P_RTHDR);
 1497                                 break;
 1498 
 1499                         case IPV6_FAITH:
 1500                                 OPTSET(IN6P_FAITH);
 1501                                 break;
 1502 
 1503                         case IPV6_RECVPATHMTU:
 1504                                 /*
 1505                                  * We ignore this option for TCP
 1506                                  * sockets.
 1507                                  * (RFC3542 leaves this case
 1508                                  * unspecified.)
 1509                                  */
 1510                                 if (uproto != IPPROTO_TCP)
 1511                                         OPTSET(IN6P_MTU);
 1512                                 break;
 1513 
 1514                         case IPV6_V6ONLY:
 1515                                 /*
 1516                                  * make setsockopt(IPV6_V6ONLY)
 1517                                  * available only prior to bind(2).
 1518                                  * see ipng mailing list, Jun 22 2001.
 1519                                  */
 1520                                 if (inp->inp_lport ||
 1521                                     !IN6_IS_ADDR_UNSPECIFIED(&in6p_laddr(inp))) {
 1522                                         error = EINVAL;
 1523                                         break;
 1524                                 }
 1525 #ifdef INET6_BINDV6ONLY
 1526                                 if (!optval)
 1527                                         error = EINVAL;
 1528 #else
 1529                                 OPTSET(IN6P_IPV6_V6ONLY);
 1530 #endif
 1531                                 break;
 1532 
 1533                         case IPV6_RECVTCLASS:
 1534 #ifdef RFC2292
 1535                                 /* cannot mix with RFC2292 XXX */
 1536                                 if (OPTBIT(IN6P_RFC2292)) {
 1537                                         error = EINVAL;
 1538                                         break;
 1539                                 }
 1540 #endif
 1541                                 OPTSET(IN6P_TCLASS);
 1542                                 break;
 1543 
 1544                         case IPV6_BINDANY:
 1545                                 error = kauth_authorize_network(
 1546                                     kauth_cred_get(), KAUTH_NETWORK_BIND,
 1547                                     KAUTH_REQ_NETWORK_BIND_ANYADDR, so, NULL,
 1548                                     NULL);
 1549                                 if (error)
 1550                                         break;
 1551                                 OPTSET(IN6P_BINDANY);
 1552                                 break;
 1553                         }
 1554                         break;
 1555 
 1556                 case IPV6_OTCLASS:
 1557                 {
 1558                         struct ip6_pktopts **optp;
 1559                         u_int8_t tclass;
 1560 
 1561                         error = sockopt_get(sopt, &tclass, sizeof(tclass));
 1562                         if (error)
 1563                                 break;
 1564                         optp = &in6p_outputopts(inp);
 1565                         error = ip6_pcbopt(optname,
 1566                                            (u_char *)&tclass,
 1567                                            sizeof(tclass),
 1568                                            optp,
 1569                                            kauth_cred_get(), uproto);
 1570                         break;
 1571                 }
 1572 
 1573                 case IPV6_TCLASS:
 1574                 case IPV6_DONTFRAG:
 1575                 case IPV6_USE_MIN_MTU:
 1576                 case IPV6_PREFER_TEMPADDR:
 1577                         error = sockopt_getint(sopt, &optval);
 1578                         if (error)
 1579                                 break;
 1580                         {
 1581                                 struct ip6_pktopts **optp;
 1582                                 optp = &in6p_outputopts(inp);
 1583                                 error = ip6_pcbopt(optname,
 1584                                                    (u_char *)&optval,
 1585                                                    sizeof(optval),
 1586                                                    optp,
 1587                                                    kauth_cred_get(), uproto);
 1588                                 break;
 1589                         }
 1590 
 1591 #ifdef RFC2292
 1592                 case IPV6_2292PKTINFO:
 1593                 case IPV6_2292HOPLIMIT:
 1594                 case IPV6_2292HOPOPTS:
 1595                 case IPV6_2292DSTOPTS:
 1596                 case IPV6_2292RTHDR:
 1597                         /* RFC 2292 */
 1598                         error = sockopt_getint(sopt, &optval);
 1599                         if (error)
 1600                                 break;
 1601 
 1602                         switch (optname) {
 1603                         case IPV6_2292PKTINFO:
 1604                                 OPTSET2292(IN6P_PKTINFO);
 1605                                 break;
 1606                         case IPV6_2292HOPLIMIT:
 1607                                 OPTSET2292(IN6P_HOPLIMIT);
 1608                                 break;
 1609                         case IPV6_2292HOPOPTS:
 1610                                 /*
 1611                                  * Check super-user privilege.
 1612                                  * See comments for IPV6_RECVHOPOPTS.
 1613                                  */
 1614                                 error = kauth_authorize_network(
 1615                                     kauth_cred_get(),
 1616                                     KAUTH_NETWORK_IPV6,
 1617                                     KAUTH_REQ_NETWORK_IPV6_HOPBYHOP, NULL,
 1618                                     NULL, NULL);
 1619                                 if (error)
 1620                                         return (error);
 1621                                 OPTSET2292(IN6P_HOPOPTS);
 1622                                 break;
 1623                         case IPV6_2292DSTOPTS:
 1624                                 error = kauth_authorize_network(
 1625                                     kauth_cred_get(),
 1626                                     KAUTH_NETWORK_IPV6,
 1627                                     KAUTH_REQ_NETWORK_IPV6_HOPBYHOP, NULL,
 1628                                     NULL, NULL);
 1629                                 if (error)
 1630                                         return (error);
 1631                                 OPTSET2292(IN6P_DSTOPTS|IN6P_RTHDRDSTOPTS); /* XXX */
 1632                                 break;
 1633                         case IPV6_2292RTHDR:
 1634                                 OPTSET2292(IN6P_RTHDR);
 1635                                 break;
 1636                         }
 1637                         break;
 1638 #endif
 1639                 case IPV6_PKTINFO:
 1640                 case IPV6_HOPOPTS:
 1641                 case IPV6_RTHDR:
 1642                 case IPV6_DSTOPTS:
 1643                 case IPV6_RTHDRDSTOPTS:
 1644                 case IPV6_NEXTHOP: {
 1645                         /* new advanced API (RFC3542) */
 1646                         void *optbuf;
 1647                         int optbuflen;
 1648                         struct ip6_pktopts **optp;
 1649 
 1650 #ifdef RFC2292
 1651                         /* cannot mix with RFC2292 */
 1652                         if (OPTBIT(IN6P_RFC2292)) {
 1653                                 error = EINVAL;
 1654                                 break;
 1655                         }
 1656 #endif
 1657 
 1658                         optbuflen = sopt->sopt_size;
 1659                         optbuf = malloc(optbuflen, M_IP6OPT, M_NOWAIT);
 1660                         if (optbuf == NULL) {
 1661                                 error = ENOBUFS;
 1662                                 break;
 1663                         }
 1664 
 1665                         error = sockopt_get(sopt, optbuf, optbuflen);
 1666                         if (error) {
 1667                                 free(optbuf, M_IP6OPT);
 1668                                 break;
 1669                         }
 1670                         optp = &in6p_outputopts(inp);
 1671                         error = ip6_pcbopt(optname, optbuf, optbuflen,
 1672                             optp, kauth_cred_get(), uproto);
 1673 
 1674                         free(optbuf, M_IP6OPT);
 1675                         break;
 1676                         }
 1677 #undef OPTSET
 1678 
 1679                 case IPV6_MULTICAST_IF:
 1680                 case IPV6_MULTICAST_HOPS:
 1681                 case IPV6_MULTICAST_LOOP:
 1682                 case IPV6_JOIN_GROUP:
 1683                 case IPV6_LEAVE_GROUP:
 1684                         error = ip6_setmoptions(sopt, inp);
 1685                         break;
 1686 
 1687                 case IPV6_PORTRANGE:
 1688                         error = sockopt_getint(sopt, &optval);
 1689                         if (error)
 1690                                 break;
 1691 
 1692                         switch (optval) {
 1693                         case IPV6_PORTRANGE_DEFAULT:
 1694                                 inp->inp_flags &= ~(IN6P_LOWPORT);
 1695                                 inp->inp_flags &= ~(IN6P_HIGHPORT);
 1696                                 break;
 1697 
 1698                         case IPV6_PORTRANGE_HIGH:
 1699                                 inp->inp_flags &= ~(IN6P_LOWPORT);
 1700                                 inp->inp_flags |= IN6P_HIGHPORT;
 1701                                 break;
 1702 
 1703                         case IPV6_PORTRANGE_LOW:
 1704                                 inp->inp_flags &= ~(IN6P_HIGHPORT);
 1705                                 inp->inp_flags |= IN6P_LOWPORT;
 1706                                 break;
 1707 
 1708                         default:
 1709                                 error = EINVAL;
 1710                                 break;
 1711                         }
 1712                         break;
 1713 
 1714                 case IPV6_PORTALGO:
 1715                         error = sockopt_getint(sopt, &optval);
 1716                         if (error)
 1717                                 break;
 1718 
 1719                         error = portalgo_algo_index_select(inp, optval);
 1720                         break;
 1721 
 1722 #if defined(IPSEC)
 1723                 case IPV6_IPSEC_POLICY:
 1724                         if (ipsec_enabled) {
 1725                                 error = ipsec_set_policy(inp,
 1726                                     sopt->sopt_data, sopt->sopt_size,
 1727                                     kauth_cred_get());
 1728                         } else
 1729                                 error = ENOPROTOOPT;
 1730                         break;
 1731 #endif /* IPSEC */
 1732 
 1733                 default:
 1734                         error = ENOPROTOOPT;
 1735                         break;
 1736                 }
 1737                 break;
 1738 
 1739         case PRCO_GETOPT:
 1740                 switch (optname) {
 1741 #ifdef RFC2292
 1742                 case IPV6_2292PKTOPTIONS:
 1743                         /*
 1744                          * RFC3542 (effectively) deprecated the
 1745                          * semantics of the 2292-style pktoptions.
 1746                          * Since it was not reliable in nature (i.e.,
 1747                          * applications had to expect the lack of some
 1748                          * information after all), it would make sense
 1749                          * to simplify this part by always returning
 1750                          * empty data.
 1751                          */
 1752                         break;
 1753 #endif
 1754 
 1755                 case IPV6_RECVHOPOPTS:
 1756                 case IPV6_RECVDSTOPTS:
 1757                 case IPV6_RECVRTHDRDSTOPTS:
 1758                 case IPV6_UNICAST_HOPS:
 1759                 case IPV6_RECVPKTINFO:
 1760                 case IPV6_RECVHOPLIMIT:
 1761                 case IPV6_RECVRTHDR:
 1762                 case IPV6_RECVPATHMTU:
 1763 
 1764                 case IPV6_FAITH:
 1765                 case IPV6_V6ONLY:
 1766                 case IPV6_PORTRANGE:
 1767                 case IPV6_RECVTCLASS:
 1768                 case IPV6_BINDANY:
 1769                         switch (optname) {
 1770 
 1771                         case IPV6_RECVHOPOPTS:
 1772                                 optval = OPTBIT(IN6P_HOPOPTS);
 1773                                 break;
 1774 
 1775                         case IPV6_RECVDSTOPTS:
 1776                                 optval = OPTBIT(IN6P_DSTOPTS);
 1777                                 break;
 1778 
 1779                         case IPV6_RECVRTHDRDSTOPTS:
 1780                                 optval = OPTBIT(IN6P_RTHDRDSTOPTS);
 1781                                 break;
 1782 
 1783                         case IPV6_UNICAST_HOPS:
 1784                                 optval = in6p_hops6(inp);
 1785                                 break;
 1786 
 1787                         case IPV6_RECVPKTINFO:
 1788                                 optval = OPTBIT(IN6P_PKTINFO);
 1789                                 break;
 1790 
 1791                         case IPV6_RECVHOPLIMIT:
 1792                                 optval = OPTBIT(IN6P_HOPLIMIT);
 1793                                 break;
 1794 
 1795                         case IPV6_RECVRTHDR:
 1796                                 optval = OPTBIT(IN6P_RTHDR);
 1797                                 break;
 1798 
 1799                         case IPV6_RECVPATHMTU:
 1800                                 optval = OPTBIT(IN6P_MTU);
 1801                                 break;
 1802 
 1803                         case IPV6_FAITH:
 1804                                 optval = OPTBIT(IN6P_FAITH);
 1805                                 break;
 1806 
 1807                         case IPV6_V6ONLY:
 1808                                 optval = OPTBIT(IN6P_IPV6_V6ONLY);
 1809                                 break;
 1810 
 1811                         case IPV6_PORTRANGE:
 1812                             {
 1813                                 int flags;
 1814                                 flags = inp->inp_flags;
 1815                                 if (flags & IN6P_HIGHPORT)
 1816                                         optval = IPV6_PORTRANGE_HIGH;
 1817                                 else if (flags & IN6P_LOWPORT)
 1818                                         optval = IPV6_PORTRANGE_LOW;
 1819                                 else
 1820                                         optval = 0;
 1821                                 break;
 1822                             }
 1823                         case IPV6_RECVTCLASS:
 1824                                 optval = OPTBIT(IN6P_TCLASS);
 1825                                 break;
 1826 
 1827                         case IPV6_BINDANY:
 1828                                 optval = OPTBIT(IN6P_BINDANY);
 1829                                 break;
 1830                         }
 1831 
 1832                         if (error)
 1833                                 break;
 1834                         error = sockopt_setint(sopt, optval);
 1835                         break;
 1836 
 1837                 case IPV6_PATHMTU:
 1838                     {
 1839                         u_long pmtu = 0;
 1840                         struct ip6_mtuinfo mtuinfo;
 1841                         struct route *ro = &inp->inp_route;
 1842                         struct rtentry *rt;
 1843                         union {
 1844                                 struct sockaddr         dst;
 1845                                 struct sockaddr_in6     dst6;
 1846                         } u;
 1847 
 1848                         if (!(so->so_state & SS_ISCONNECTED))
 1849                                 return (ENOTCONN);
 1850                         /*
 1851                          * XXX: we dot not consider the case of source
 1852                          * routing, or optional information to specify
 1853                          * the outgoing interface.
 1854                          */
 1855                         sockaddr_in6_init(&u.dst6, &in6p_faddr(inp), 0, 0, 0);
 1856                         rt = rtcache_lookup(ro, &u.dst);
 1857                         error = ip6_getpmtu(rt, NULL, &pmtu, NULL);
 1858                         rtcache_unref(rt, ro);
 1859                         if (error)
 1860                                 break;
 1861                         if (pmtu > IPV6_MAXPACKET)
 1862                                 pmtu = IPV6_MAXPACKET;
 1863 
 1864                         memset(&mtuinfo, 0, sizeof(mtuinfo));
 1865                         mtuinfo.ip6m_mtu = (u_int32_t)pmtu;
 1866                         optdata = (void *)&mtuinfo;
 1867                         optdatalen = sizeof(mtuinfo);
 1868                         if (optdatalen > MCLBYTES)
 1869                                 return (EMSGSIZE); /* XXX */
 1870                         error = sockopt_set(sopt, optdata, optdatalen);
 1871                         break;
 1872                     }
 1873 
 1874 #ifdef RFC2292
 1875                 case IPV6_2292PKTINFO:
 1876                 case IPV6_2292HOPLIMIT:
 1877                 case IPV6_2292HOPOPTS:
 1878                 case IPV6_2292RTHDR:
 1879                 case IPV6_2292DSTOPTS:
 1880                         switch (optname) {
 1881                         case IPV6_2292PKTINFO:
 1882                                 optval = OPTBIT(IN6P_PKTINFO);
 1883                                 break;
 1884                         case IPV6_2292HOPLIMIT:
 1885                                 optval = OPTBIT(IN6P_HOPLIMIT);
 1886                                 break;
 1887                         case IPV6_2292HOPOPTS:
 1888                                 optval = OPTBIT(IN6P_HOPOPTS);
 1889                                 break;
 1890                         case IPV6_2292RTHDR:
 1891                                 optval = OPTBIT(IN6P_RTHDR);
 1892                                 break;
 1893                         case IPV6_2292DSTOPTS:
 1894                                 optval = OPTBIT(IN6P_DSTOPTS|IN6P_RTHDRDSTOPTS);
 1895                                 break;
 1896                         }
 1897                         error = sockopt_setint(sopt, optval);
 1898                         break;
 1899 #endif
 1900                 case IPV6_PKTINFO:
 1901                 case IPV6_HOPOPTS:
 1902                 case IPV6_RTHDR:
 1903                 case IPV6_DSTOPTS:
 1904                 case IPV6_RTHDRDSTOPTS:
 1905                 case IPV6_NEXTHOP:
 1906                 case IPV6_OTCLASS:
 1907                 case IPV6_TCLASS:
 1908                 case IPV6_DONTFRAG:
 1909                 case IPV6_USE_MIN_MTU:
 1910                 case IPV6_PREFER_TEMPADDR:
 1911                         error = ip6_getpcbopt(in6p_outputopts(inp),
 1912                             optname, sopt);
 1913                         break;
 1914 
 1915                 case IPV6_MULTICAST_IF:
 1916                 case IPV6_MULTICAST_HOPS:
 1917                 case IPV6_MULTICAST_LOOP:
 1918                 case IPV6_JOIN_GROUP:
 1919                 case IPV6_LEAVE_GROUP:
 1920                         error = ip6_getmoptions(sopt, inp);
 1921                         break;
 1922 
 1923                 case IPV6_PORTALGO:
 1924                         optval = inp->inp_portalgo;
 1925                         error = sockopt_setint(sopt, optval);
 1926                         break;
 1927 
 1928 #if defined(IPSEC)
 1929                 case IPV6_IPSEC_POLICY:
 1930                         if (ipsec_used) {
 1931                                 struct mbuf *m = NULL;
 1932 
 1933                                 /*
 1934                                  * XXX: this will return EINVAL as sopt is
 1935                                  * empty
 1936                                  */
 1937                                 error = ipsec_get_policy(inp, sopt->sopt_data,
 1938                                     sopt->sopt_size, &m);
 1939                                 if (!error)
 1940                                         error = sockopt_setmbuf(sopt, m);
 1941                         } else
 1942                                 error = ENOPROTOOPT;
 1943                         break;
 1944 #endif /* IPSEC */
 1945 
 1946                 default:
 1947                         error = ENOPROTOOPT;
 1948                         break;
 1949                 }
 1950                 break;
 1951         }
 1952         return (error);
 1953 }
 1954 
 1955 int
 1956 ip6_raw_ctloutput(int op, struct socket *so, struct sockopt *sopt)
 1957 {
 1958         int error = 0, optval;
 1959         const int icmp6off = offsetof(struct icmp6_hdr, icmp6_cksum);
 1960         struct inpcb *inp = sotoinpcb(so);
 1961         int level, optname;
 1962 
 1963         KASSERT(sopt != NULL);
 1964 
 1965         level = sopt->sopt_level;
 1966         optname = sopt->sopt_name;
 1967 
 1968         if (level != IPPROTO_IPV6) {
 1969                 return ENOPROTOOPT;
 1970         }
 1971 
 1972         switch (optname) {
 1973         case IPV6_CHECKSUM:
 1974                 /*
 1975                  * For ICMPv6 sockets, no modification allowed for checksum
 1976                  * offset, permit "no change" values to help existing apps.
 1977                  *
 1978                  * XXX RFC3542 says: "An attempt to set IPV6_CHECKSUM
 1979                  * for an ICMPv6 socket will fail."  The current
 1980                  * behavior does not meet RFC3542.
 1981                  */
 1982                 switch (op) {
 1983                 case PRCO_SETOPT:
 1984                         error = sockopt_getint(sopt, &optval);
 1985                         if (error)
 1986                                 break;
 1987                         if ((optval % 2) != 0) {
 1988                                 /* the API assumes even offset values */
 1989                                 error = EINVAL;
 1990                         } else if (so->so_proto->pr_protocol ==
 1991                             IPPROTO_ICMPV6) {
 1992                                 if (optval != icmp6off)
 1993                                         error = EINVAL;
 1994                         } else
 1995                                 in6p_cksum(inp) = optval;
 1996                         break;
 1997 
 1998                 case PRCO_GETOPT:
 1999                         if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
 2000                                 optval = icmp6off;
 2001                         else
 2002                                 optval = in6p_cksum(inp);
 2003 
 2004                         error = sockopt_setint(sopt, optval);
 2005                         break;
 2006 
 2007                 default:
 2008                         error = EINVAL;
 2009                         break;
 2010                 }
 2011                 break;
 2012 
 2013         default:
 2014                 error = ENOPROTOOPT;
 2015                 break;
 2016         }
 2017 
 2018         return (error);
 2019 }
 2020 
 2021 #ifdef RFC2292
 2022 /*
 2023  * Set up IP6 options in pcb for insertion in output packets or
 2024  * specifying behavior of outgoing packets.
 2025  */
 2026 static int
 2027 ip6_pcbopts(struct ip6_pktopts **pktopt, struct socket *so,
 2028     struct sockopt *sopt)
 2029 {
 2030         struct ip6_pktopts *opt = *pktopt;
 2031         struct mbuf *m;
 2032         int error = 0;
 2033 
 2034         KASSERT(solocked(so));
 2035 
 2036         /* turn off any old options. */
 2037         if (opt) {
 2038 #ifdef DIAGNOSTIC
 2039             if (opt->ip6po_pktinfo || opt->ip6po_nexthop ||
 2040                 opt->ip6po_hbh || opt->ip6po_dest1 || opt->ip6po_dest2 ||
 2041                 opt->ip6po_rhinfo.ip6po_rhi_rthdr)
 2042                     printf("ip6_pcbopts: all specified options are cleared.\n");
 2043 #endif
 2044                 ip6_clearpktopts(opt, -1);
 2045         } else {
 2046                 opt = malloc(sizeof(*opt), M_IP6OPT, M_NOWAIT);
 2047                 if (opt == NULL)
 2048                         return (ENOBUFS);
 2049         }
 2050         *pktopt = NULL;
 2051 
 2052         if (sopt == NULL || sopt->sopt_size == 0) {
 2053                 /*
 2054                  * Only turning off any previous options, regardless of
 2055                  * whether the opt is just created or given.
 2056                  */
 2057                 free(opt, M_IP6OPT);
 2058                 return (0);
 2059         }
 2060 
 2061         /*  set options specified by user. */
 2062         m = sockopt_getmbuf(sopt);
 2063         if (m == NULL) {
 2064                 free(opt, M_IP6OPT);
 2065                 return (ENOBUFS);
 2066         }
 2067 
 2068         error = ip6_setpktopts(m, opt, NULL, kauth_cred_get(),
 2069             so->so_proto->pr_protocol);
 2070         m_freem(m);
 2071         if (error != 0) {
 2072                 ip6_clearpktopts(opt, -1); /* XXX: discard all options */
 2073                 free(opt, M_IP6OPT);
 2074                 return (error);
 2075         }
 2076         *pktopt = opt;
 2077         return (0);
 2078 }
 2079 #endif
 2080 
 2081 /*
 2082  * initialize ip6_pktopts.  beware that there are non-zero default values in
 2083  * the struct.
 2084  */
 2085 void
 2086 ip6_initpktopts(struct ip6_pktopts *opt)
 2087 {
 2088 
 2089         memset(opt, 0, sizeof(*opt));
 2090         opt->ip6po_hlim = -1;   /* -1 means default hop limit */
 2091         opt->ip6po_tclass = -1; /* -1 means default traffic class */
 2092         opt->ip6po_minmtu = IP6PO_MINMTU_MCASTONLY;
 2093         opt->ip6po_prefer_tempaddr = IP6PO_TEMPADDR_SYSTEM;
 2094 }
 2095 
 2096 #define sin6tosa(sin6)  ((struct sockaddr *)(sin6)) /* XXX */
 2097 static int
 2098 ip6_pcbopt(int optname, u_char *buf, int len, struct ip6_pktopts **pktopt,
 2099     kauth_cred_t cred, int uproto)
 2100 {
 2101         struct ip6_pktopts *opt;
 2102 
 2103         if (*pktopt == NULL) {
 2104                 *pktopt = malloc(sizeof(struct ip6_pktopts), M_IP6OPT,
 2105                     M_NOWAIT);
 2106                 if (*pktopt == NULL)
 2107                         return (ENOBUFS);
 2108 
 2109                 ip6_initpktopts(*pktopt);
 2110         }
 2111         opt = *pktopt;
 2112 
 2113         return (ip6_setpktopt(optname, buf, len, opt, cred, 1, 0, uproto));
 2114 }
 2115 
 2116 static int
 2117 ip6_getpcbopt(struct ip6_pktopts *pktopt, int optname, struct sockopt *sopt)
 2118 {
 2119         void *optdata = NULL;
 2120         int optdatalen = 0;
 2121         struct ip6_ext *ip6e;
 2122         int error = 0;
 2123         struct in6_pktinfo null_pktinfo;
 2124         int deftclass = 0, on;
 2125         int defminmtu = IP6PO_MINMTU_MCASTONLY;
 2126         int defpreftemp = IP6PO_TEMPADDR_SYSTEM;
 2127 
 2128         switch (optname) {
 2129         case IPV6_PKTINFO:
 2130                 if (pktopt && pktopt->ip6po_pktinfo)
 2131                         optdata = (void *)pktopt->ip6po_pktinfo;
 2132                 else {
 2133                         /* XXX: we don't have to do this every time... */
 2134                         memset(&null_pktinfo, 0, sizeof(null_pktinfo));
 2135                         optdata = (void *)&null_pktinfo;
 2136                 }
 2137                 optdatalen = sizeof(struct in6_pktinfo);
 2138                 break;
 2139         case IPV6_OTCLASS:
 2140                 /* XXX */
 2141                 return (EINVAL);
 2142         case IPV6_TCLASS:
 2143                 if (pktopt && pktopt->ip6po_tclass >= 0)
 2144                         optdata = (void *)&pktopt->ip6po_tclass;
 2145                 else
 2146                         optdata = (void *)&deftclass;
 2147                 optdatalen = sizeof(int);
 2148                 break;
 2149         case IPV6_HOPOPTS:
 2150                 if (pktopt && pktopt->ip6po_hbh) {
 2151                         optdata = (void *)pktopt->ip6po_hbh;
 2152                         ip6e = (struct ip6_ext *)pktopt->ip6po_hbh;
 2153                         optdatalen = (ip6e->ip6e_len + 1) << 3;
 2154                 }
 2155                 break;
 2156         case IPV6_RTHDR:
 2157                 if (pktopt && pktopt->ip6po_rthdr) {
 2158                         optdata = (void *)pktopt->ip6po_rthdr;
 2159                         ip6e = (struct ip6_ext *)pktopt->ip6po_rthdr;
 2160                         optdatalen = (ip6e->ip6e_len + 1) << 3;
 2161                 }
 2162                 break;
 2163         case IPV6_RTHDRDSTOPTS:
 2164                 if (pktopt && pktopt->ip6po_dest1) {
 2165                         optdata = (void *)pktopt->ip6po_dest1;
 2166                         ip6e = (struct ip6_ext *)pktopt->ip6po_dest1;
 2167                         optdatalen = (ip6e->ip6e_len + 1) << 3;
 2168                 }
 2169                 break;
 2170         case IPV6_DSTOPTS:
 2171                 if (pktopt && pktopt->ip6po_dest2) {
 2172                         optdata = (void *)pktopt->ip6po_dest2;
 2173                         ip6e = (struct ip6_ext *)pktopt->ip6po_dest2;
 2174                         optdatalen = (ip6e->ip6e_len + 1) << 3;
 2175                 }
 2176                 break;
 2177         case IPV6_NEXTHOP:
 2178                 if (pktopt && pktopt->ip6po_nexthop) {
 2179                         optdata = (void *)pktopt->ip6po_nexthop;
 2180                         optdatalen = pktopt->ip6po_nexthop->sa_len;
 2181                 }
 2182                 break;
 2183         case IPV6_USE_MIN_MTU:
 2184                 if (pktopt)
 2185                         optdata = (void *)&pktopt->ip6po_minmtu;
 2186                 else
 2187                         optdata = (void *)&defminmtu;
 2188                 optdatalen = sizeof(int);
 2189                 break;
 2190         case IPV6_DONTFRAG:
 2191                 if (pktopt && ((pktopt->ip6po_flags) & IP6PO_DONTFRAG))
 2192                         on = 1;
 2193                 else
 2194                         on = 0;
 2195                 optdata = (void *)&on;
 2196                 optdatalen = sizeof(on);
 2197                 break;
 2198         case IPV6_PREFER_TEMPADDR:
 2199                 if (pktopt)
 2200                         optdata = (void *)&pktopt->ip6po_prefer_tempaddr;
 2201                 else
 2202                         optdata = (void *)&defpreftemp;
 2203                 optdatalen = sizeof(int);
 2204                 break;
 2205         default:                /* should not happen */
 2206 #ifdef DIAGNOSTIC
 2207                 panic("ip6_getpcbopt: unexpected option\n");
 2208 #endif
 2209                 return (ENOPROTOOPT);
 2210         }
 2211 
 2212         error = sockopt_set(sopt, optdata, optdatalen);
 2213 
 2214         return (error);
 2215 }
 2216 
 2217 void
 2218 ip6_clearpktopts(struct ip6_pktopts *pktopt, int optname)
 2219 {
 2220         if (optname == -1 || optname == IPV6_PKTINFO) {
 2221                 if (pktopt->ip6po_pktinfo)
 2222                         free(pktopt->ip6po_pktinfo, M_IP6OPT);
 2223                 pktopt->ip6po_pktinfo = NULL;
 2224         }
 2225         if (optname == -1 || optname == IPV6_HOPLIMIT)
 2226                 pktopt->ip6po_hlim = -1;
 2227         if (optname == -1 || optname == IPV6_TCLASS)
 2228                 pktopt->ip6po_tclass = -1;
 2229         if (optname == -1 || optname == IPV6_NEXTHOP) {
 2230                 rtcache_free(&pktopt->ip6po_nextroute);
 2231                 if (pktopt->ip6po_nexthop)
 2232                         free(pktopt->ip6po_nexthop, M_IP6OPT);
 2233                 pktopt->ip6po_nexthop = NULL;
 2234         }
 2235         if (optname == -1 || optname == IPV6_HOPOPTS) {
 2236                 if (pktopt->ip6po_hbh)
 2237                         free(pktopt->ip6po_hbh, M_IP6OPT);
 2238                 pktopt->ip6po_hbh = NULL;
 2239         }
 2240         if (optname == -1 || optname == IPV6_RTHDRDSTOPTS) {
 2241                 if (pktopt->ip6po_dest1)
 2242                         free(pktopt->ip6po_dest1, M_IP6OPT);
 2243                 pktopt->ip6po_dest1 = NULL;
 2244         }
 2245         if (optname == -1 || optname == IPV6_RTHDR) {
 2246                 if (pktopt->ip6po_rhinfo.ip6po_rhi_rthdr)
 2247                         free(pktopt->ip6po_rhinfo.ip6po_rhi_rthdr, M_IP6OPT);
 2248                 pktopt->ip6po_rhinfo.ip6po_rhi_rthdr = NULL;
 2249                 rtcache_free(&pktopt->ip6po_route);
 2250         }
 2251         if (optname == -1 || optname == IPV6_DSTOPTS) {
 2252                 if (pktopt->ip6po_dest2)
 2253                         free(pktopt->ip6po_dest2, M_IP6OPT);
 2254                 pktopt->ip6po_dest2 = NULL;
 2255         }
 2256 }
 2257 
 2258 #define PKTOPT_EXTHDRCPY(type)                                  \
 2259 do {                                                            \
 2260         if (src->type) {                                        \
 2261                 int hlen = (((struct ip6_ext *)src->type)->ip6e_len + 1) << 3;\
 2262                 dst->type = malloc(hlen, M_IP6OPT, canwait);    \
 2263                 if (dst->type == NULL)                          \
 2264                         goto bad;                               \
 2265                 memcpy(dst->type, src->type, hlen);             \
 2266         }                                                       \
 2267 } while (/*CONSTCOND*/ 0)
 2268 
 2269 static int
 2270 copypktopts(struct ip6_pktopts *dst, struct ip6_pktopts *src, int canwait)
 2271 {
 2272         dst->ip6po_hlim = src->ip6po_hlim;
 2273         dst->ip6po_tclass = src->ip6po_tclass;
 2274         dst->ip6po_flags = src->ip6po_flags;
 2275         dst->ip6po_minmtu = src->ip6po_minmtu;
 2276         dst->ip6po_prefer_tempaddr = src->ip6po_prefer_tempaddr;
 2277         if (src->ip6po_pktinfo) {
 2278                 dst->ip6po_pktinfo = malloc(sizeof(*dst->ip6po_pktinfo),
 2279                     M_IP6OPT, canwait);
 2280                 if (dst->ip6po_pktinfo == NULL)
 2281                         goto bad;
 2282                 *dst->ip6po_pktinfo = *src->ip6po_pktinfo;
 2283         }
 2284         if (src->ip6po_nexthop) {
 2285                 dst->ip6po_nexthop = malloc(src->ip6po_nexthop->sa_len,
 2286                     M_IP6OPT, canwait);
 2287                 if (dst->ip6po_nexthop == NULL)
 2288                         goto bad;
 2289                 memcpy(dst->ip6po_nexthop, src->ip6po_nexthop,
 2290                     src->ip6po_nexthop->sa_len);
 2291         }
 2292         PKTOPT_EXTHDRCPY(ip6po_hbh);
 2293         PKTOPT_EXTHDRCPY(ip6po_dest1);
 2294         PKTOPT_EXTHDRCPY(ip6po_dest2);
 2295         PKTOPT_EXTHDRCPY(ip6po_rthdr); /* not copy the cached route */
 2296         return (0);
 2297 
 2298   bad:
 2299         if (dst->ip6po_pktinfo) free(dst->ip6po_pktinfo, M_IP6OPT);
 2300         if (dst->ip6po_nexthop) free(dst->ip6po_nexthop, M_IP6OPT);
 2301         if (dst->ip6po_hbh) free(dst->ip6po_hbh, M_IP6OPT);
 2302         if (dst->ip6po_dest1) free(dst->ip6po_dest1, M_IP6OPT);
 2303         if (dst->ip6po_dest2) free(dst->ip6po_dest2, M_IP6OPT);
 2304         if (dst->ip6po_rthdr) free(dst->ip6po_rthdr, M_IP6OPT);
 2305 
 2306         return (ENOBUFS);
 2307 }
 2308 #undef PKTOPT_EXTHDRCPY
 2309 
 2310 struct ip6_pktopts *
 2311 ip6_copypktopts(struct ip6_pktopts *src, int canwait)
 2312 {
 2313         int error;
 2314         struct ip6_pktopts *dst;
 2315 
 2316         dst = malloc(sizeof(*dst), M_IP6OPT, canwait);
 2317         if (dst == NULL)
 2318                 return (NULL);
 2319         ip6_initpktopts(dst);
 2320 
 2321         if ((error = copypktopts(dst, src, canwait)) != 0) {
 2322                 free(dst, M_IP6OPT);
 2323                 return (NULL);
 2324         }
 2325 
 2326         return (dst);
 2327 }
 2328 
 2329 void
 2330 ip6_freepcbopts(struct ip6_pktopts *pktopt)
 2331 {
 2332         if (pktopt == NULL)
 2333                 return;
 2334 
 2335         ip6_clearpktopts(pktopt, -1);
 2336 
 2337         free(pktopt, M_IP6OPT);
 2338 }
 2339 
 2340 int
 2341 ip6_get_membership(const struct sockopt *sopt, struct ifnet **ifp,
 2342     struct psref *psref, void *v, size_t l)
 2343 {
 2344         struct ipv6_mreq mreq;
 2345         int error;
 2346         struct in6_addr *ia = &mreq.ipv6mr_multiaddr;
 2347         struct in_addr *ia4 = (void *)&ia->s6_addr32[3];
 2348 
 2349         error = sockopt_get(sopt, &mreq, sizeof(mreq));
 2350         if (error != 0)
 2351                 return error;
 2352 
 2353         if (IN6_IS_ADDR_UNSPECIFIED(ia)) {
 2354                 /*
 2355                  * We use the unspecified address to specify to accept
 2356                  * all multicast addresses. Only super user is allowed
 2357                  * to do this.
 2358                  */
 2359                 if (kauth_authorize_network(kauth_cred_get(),
 2360                     KAUTH_NETWORK_IPV6,
 2361                     KAUTH_REQ_NETWORK_IPV6_JOIN_MULTICAST, NULL, NULL, NULL))
 2362                         return EACCES;
 2363         } else if (IN6_IS_ADDR_V4MAPPED(ia)) {
 2364                 // Don't bother if we are not going to use ifp.
 2365                 if (l == sizeof(*ia)) {
 2366                         memcpy(v, ia, l);
 2367                         return 0;
 2368                 }
 2369         } else if (!IN6_IS_ADDR_MULTICAST(ia)) {
 2370                 return EINVAL;
 2371         }
 2372 
 2373         /*
 2374          * If no interface was explicitly specified, choose an
 2375          * appropriate one according to the given multicast address.
 2376          */
 2377         if (mreq.ipv6mr_interface == 0) {
 2378                 struct rtentry *rt;
 2379                 union {
 2380                         struct sockaddr         dst;
 2381                         struct sockaddr_in      dst4;
 2382                         struct sockaddr_in6     dst6;
 2383                 } u;
 2384                 struct route ro;
 2385 
 2386                 /*
 2387                  * Look up the routing table for the
 2388                  * address, and choose the outgoing interface.
 2389                  *   XXX: is it a good approach?
 2390                  */
 2391                 memset(&ro, 0, sizeof(ro));
 2392                 if (IN6_IS_ADDR_V4MAPPED(ia))
 2393                         sockaddr_in_init(&u.dst4, ia4, 0);
 2394                 else
 2395                         sockaddr_in6_init(&u.dst6, ia, 0, 0, 0);
 2396                 error = rtcache_setdst(&ro, &u.dst);
 2397                 if (error != 0)
 2398                         return error;
 2399                 rt = rtcache_init(&ro);
 2400                 *ifp = rt != NULL ?
 2401                     if_get_byindex(rt->rt_ifp->if_index, psref) : NULL;
 2402                 rtcache_unref(rt, &ro);
 2403                 rtcache_free(&ro);
 2404         } else {
 2405                 /*
 2406                  * If the interface is specified, validate it.
 2407                  */
 2408                 *ifp = if_get_byindex(mreq.ipv6mr_interface, psref);
 2409                 if (*ifp == NULL)
 2410                         return ENXIO;   /* XXX EINVAL? */
 2411         }
 2412         if (sizeof(*ia) == l)
 2413                 memcpy(v, ia, l);
 2414         else
 2415                 memcpy(v, ia4, l);
 2416         return 0;
 2417 }
 2418 
 2419 /*
 2420  * Set the IP6 multicast options in response to user setsockopt().
 2421  */
 2422 static int
 2423 ip6_setmoptions(const struct sockopt *sopt, struct inpcb *inp)
 2424 {
 2425         int error = 0;
 2426         u_int loop, ifindex;
 2427         struct ipv6_mreq mreq;
 2428         struct in6_addr ia;
 2429         struct ifnet *ifp;
 2430         struct ip6_moptions *im6o = in6p_moptions(inp);
 2431         struct in6_multi_mship *imm;
 2432 
 2433         KASSERT(inp_locked(inp));
 2434 
 2435         if (im6o == NULL) {
 2436                 /*
 2437                  * No multicast option buffer attached to the pcb;
 2438                  * allocate one and initialize to default values.
 2439                  */
 2440                 im6o = malloc(sizeof(*im6o), M_IPMOPTS, M_NOWAIT);
 2441                 if (im6o == NULL)
 2442                         return (ENOBUFS);
 2443                 in6p_moptions(inp) = im6o;
 2444                 im6o->im6o_multicast_if_index = 0;
 2445                 im6o->im6o_multicast_hlim = ip6_defmcasthlim;
 2446                 im6o->im6o_multicast_loop = IPV6_DEFAULT_MULTICAST_LOOP;
 2447                 LIST_INIT(&im6o->im6o_memberships);
 2448         }
 2449 
 2450         switch (sopt->sopt_name) {
 2451 
 2452         case IPV6_MULTICAST_IF: {
 2453                 int s;
 2454                 /*
 2455                  * Select the interface for outgoing multicast packets.
 2456                  */
 2457                 error = sockopt_get(sopt, &ifindex, sizeof(ifindex));
 2458                 if (error != 0)
 2459                         break;
 2460 
 2461                 s = pserialize_read_enter();
 2462                 if (ifindex != 0) {
 2463                         if ((ifp = if_byindex(ifindex)) == NULL) {
 2464                                 pserialize_read_exit(s);
 2465                                 error = ENXIO;  /* XXX EINVAL? */
 2466                                 break;
 2467                         }
 2468                         if ((ifp->if_flags & IFF_MULTICAST) == 0) {
 2469                                 pserialize_read_exit(s);
 2470                                 error = EADDRNOTAVAIL;
 2471                                 break;
 2472                         }
 2473                 } else
 2474                         ifp = NULL;
 2475                 im6o->im6o_multicast_if_index = if_get_index(ifp);
 2476                 pserialize_read_exit(s);
 2477                 break;
 2478             }
 2479 
 2480         case IPV6_MULTICAST_HOPS:
 2481             {
 2482                 /*
 2483                  * Set the IP6 hoplimit for outgoing multicast packets.
 2484                  */
 2485                 int optval;
 2486 
 2487                 error = sockopt_getint(sopt, &optval);
 2488                 if (error != 0)
 2489                         break;
 2490 
 2491                 if (optval < -1 || optval >= 256)
 2492                         error = EINVAL;
 2493                 else if (optval == -1)
 2494                         im6o->im6o_multicast_hlim = ip6_defmcasthlim;
 2495                 else
 2496                         im6o->im6o_multicast_hlim = optval;
 2497                 break;
 2498             }
 2499 
 2500         case IPV6_MULTICAST_LOOP:
 2501                 /*
 2502                  * Set the loopback flag for outgoing multicast packets.
 2503                  * Must be zero or one.
 2504                  */
 2505                 error = sockopt_get(sopt, &loop, sizeof(loop));
 2506                 if (error != 0)
 2507                         break;
 2508                 if (loop > 1) {
 2509                         error = EINVAL;
 2510                         break;
 2511                 }
 2512                 im6o->im6o_multicast_loop = loop;
 2513                 break;
 2514 
 2515         case IPV6_JOIN_GROUP: {
 2516                 int bound;
 2517                 struct psref psref;
 2518                 /*
 2519                  * Add a multicast group membership.
 2520                  * Group must be a valid IP6 multicast address.
 2521                  */
 2522                 bound = curlwp_bind();
 2523                 ifp = NULL;
 2524                 error = ip6_get_membership(sopt, &ifp, &psref, &ia, sizeof(ia));
 2525                 if (error != 0) {
 2526                         KASSERT(ifp == NULL);
 2527                         curlwp_bindx(bound);
 2528                         return error;
 2529                 }
 2530 
 2531                 if (IN6_IS_ADDR_V4MAPPED(&ia)) {
 2532                         error = ip_setmoptions(&inp->inp_moptions, sopt);
 2533                         goto put_break;
 2534                 }
 2535                 /*
 2536                  * See if we found an interface, and confirm that it
 2537                  * supports multicast
 2538                  */
 2539                 if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) {
 2540                         error = EADDRNOTAVAIL;
 2541                         goto put_break;
 2542                 }
 2543 
 2544                 if (in6_setscope(&ia, ifp, NULL)) {
 2545                         error = EADDRNOTAVAIL; /* XXX: should not happen */
 2546                         goto put_break;
 2547                 }
 2548 
 2549                 /*
 2550                  * See if the membership already exists.
 2551                  */
 2552                 LIST_FOREACH(imm, &im6o->im6o_memberships, i6mm_chain) {
 2553                         if (imm->i6mm_maddr->in6m_ifp == ifp &&
 2554                             IN6_ARE_ADDR_EQUAL(&imm->i6mm_maddr->in6m_addr,
 2555                             &ia))
 2556                                 goto put_break;
 2557                 }
 2558                 if (imm != NULL) {
 2559                         error = EADDRINUSE;
 2560                         goto put_break;
 2561                 }
 2562                 /*
 2563                  * Everything looks good; add a new record to the multicast
 2564                  * address list for the given interface.
 2565                  */
 2566                 imm = in6_joingroup(ifp, &ia, &error, 0);
 2567                 if (imm == NULL)
 2568                         goto put_break;
 2569                 LIST_INSERT_HEAD(&im6o->im6o_memberships, imm, i6mm_chain);
 2570             put_break:
 2571                 if_put(ifp, &psref);
 2572                 curlwp_bindx(bound);
 2573                 break;
 2574             }
 2575 
 2576         case IPV6_LEAVE_GROUP: {
 2577                 /*
 2578                  * Drop a multicast group membership.
 2579                  * Group must be a valid IP6 multicast address.
 2580                  */
 2581                 error = sockopt_get(sopt, &mreq, sizeof(mreq));
 2582                 if (error != 0)
 2583                         break;
 2584 
 2585                 if (IN6_IS_ADDR_V4MAPPED(&mreq.ipv6mr_multiaddr)) {
 2586                         error = ip_setmoptions(&inp->inp_moptions, sopt);
 2587                         break;
 2588                 }
 2589                 /*
 2590                  * If an interface address was specified, get a pointer
 2591                  * to its ifnet structure.
 2592                  */
 2593                 if (mreq.ipv6mr_interface != 0) {
 2594                         if ((ifp = if_byindex(mreq.ipv6mr_interface)) == NULL) {
 2595                                 error = ENXIO;  /* XXX EINVAL? */
 2596                                 break;
 2597                         }
 2598                 } else
 2599                         ifp = NULL;
 2600 
 2601                 /* Fill in the scope zone ID */
 2602                 if (ifp) {
 2603                         if (in6_setscope(&mreq.ipv6mr_multiaddr, ifp, NULL)) {
 2604                                 /* XXX: should not happen */
 2605                                 error = EADDRNOTAVAIL;
 2606                                 break;
 2607                         }
 2608                 } else if (mreq.ipv6mr_interface != 0) {
 2609                         /*
 2610                          * XXX: This case would happens when the (positive)
 2611                          * index is in the valid range, but the corresponding
 2612                          * interface has been detached dynamically.  The above
 2613                          * check probably avoids such case to happen here, but
 2614                          * we check it explicitly for safety.
 2615                          */
 2616                         error = EADDRNOTAVAIL;
 2617                         break;
 2618                 } else {        /* ipv6mr_interface == 0 */
 2619                         struct sockaddr_in6 sa6_mc;
 2620 
 2621                         /*
 2622                          * The API spec says as follows:
 2623                          *  If the interface index is specified as 0, the
 2624                          *  system may choose a multicast group membership to
 2625                          *  drop by matching the multicast address only.
 2626                          * On the other hand, we cannot disambiguate the scope
 2627                          * zone unless an interface is provided.  Thus, we
 2628                          * check if there's ambiguity with the default scope
 2629                          * zone as the last resort.
 2630                          */
 2631                         sockaddr_in6_init(&sa6_mc, &mreq.ipv6mr_multiaddr,
 2632                             0, 0, 0);
 2633                         error = sa6_embedscope(&sa6_mc, ip6_use_defzone);
 2634                         if (error != 0)
 2635                                 break;
 2636                         mreq.ipv6mr_multiaddr = sa6_mc.sin6_addr;
 2637                 }
 2638 
 2639                 /*
 2640                  * Find the membership in the membership list.
 2641                  */
 2642                 LIST_FOREACH(imm, &im6o->im6o_memberships, i6mm_chain) {
 2643                         if ((ifp == NULL || imm->i6mm_maddr->in6m_ifp == ifp) &&
 2644                             IN6_ARE_ADDR_EQUAL(&imm->i6mm_maddr->in6m_addr,
 2645                             &mreq.ipv6mr_multiaddr))
 2646                                 break;
 2647                 }
 2648                 if (imm == NULL) {
 2649                         /* Unable to resolve interface */
 2650                         error = EADDRNOTAVAIL;
 2651                         break;
 2652                 }
 2653                 /*
 2654                  * Give up the multicast address record to which the
 2655                  * membership points.
 2656                  */
 2657                 LIST_REMOVE(imm, i6mm_chain);
 2658                 in6_leavegroup(imm);
 2659                 /* in6m_ifp should not leave thanks to inp_lock */
 2660                 break;
 2661             }
 2662 
 2663         default:
 2664                 error = EOPNOTSUPP;
 2665                 break;
 2666         }
 2667 
 2668         /*
 2669          * If all options have default values, no need to keep the mbuf.
 2670          */
 2671         if (im6o->im6o_multicast_if_index == 0 &&
 2672             im6o->im6o_multicast_hlim == ip6_defmcasthlim &&
 2673             im6o->im6o_multicast_loop == IPV6_DEFAULT_MULTICAST_LOOP &&
 2674             LIST_EMPTY(&im6o->im6o_memberships)) {
 2675                 free(in6p_moptions(inp), M_IPMOPTS);
 2676                 in6p_moptions(inp) = NULL;
 2677         }
 2678 
 2679         return (error);
 2680 }
 2681 
 2682 /*
 2683  * Return the IP6 multicast options in response to user getsockopt().
 2684  */
 2685 static int
 2686 ip6_getmoptions(struct sockopt *sopt, struct inpcb *inp)
 2687 {
 2688         u_int optval;
 2689         int error;
 2690         struct ip6_moptions *im6o = in6p_moptions(inp);
 2691 
 2692         switch (sopt->sopt_name) {
 2693         case IPV6_MULTICAST_IF:
 2694                 if (im6o == NULL || im6o->im6o_multicast_if_index == 0)
 2695                         optval = 0;
 2696                 else
 2697                         optval = im6o->im6o_multicast_if_index;
 2698 
 2699                 error = sockopt_set(sopt, &optval, sizeof(optval));
 2700                 break;
 2701 
 2702         case IPV6_MULTICAST_HOPS:
 2703                 if (im6o == NULL)
 2704                         optval = ip6_defmcasthlim;
 2705                 else
 2706                         optval = im6o->im6o_multicast_hlim;
 2707 
 2708                 error = sockopt_set(sopt, &optval, sizeof(optval));
 2709                 break;
 2710 
 2711         case IPV6_MULTICAST_LOOP:
 2712                 if (im6o == NULL)
 2713                         optval = IPV6_DEFAULT_MULTICAST_LOOP;
 2714                 else
 2715                         optval = im6o->im6o_multicast_loop;
 2716 
 2717                 error = sockopt_set(sopt, &optval, sizeof(optval));
 2718                 break;
 2719 
 2720         default:
 2721                 error = EOPNOTSUPP;
 2722         }
 2723 
 2724         return (error);
 2725 }
 2726 
 2727 /*
 2728  * Discard the IP6 multicast options.
 2729  */
 2730 void
 2731 ip6_freemoptions(struct ip6_moptions *im6o)
 2732 {
 2733         struct in6_multi_mship *imm, *nimm;
 2734 
 2735         if (im6o == NULL)
 2736                 return;
 2737 
 2738         /* The owner of im6o (inp) should be protected by solock */
 2739         LIST_FOREACH_SAFE(imm, &im6o->im6o_memberships, i6mm_chain, nimm) {
 2740                 LIST_REMOVE(imm, i6mm_chain);
 2741                 in6_leavegroup(imm);
 2742         }
 2743         free(im6o, M_IPMOPTS);
 2744 }
 2745 
 2746 /*
 2747  * Set IPv6 outgoing packet options based on advanced API.
 2748  */
 2749 int
 2750 ip6_setpktopts(struct mbuf *control, struct ip6_pktopts *opt,
 2751         struct ip6_pktopts *stickyopt, kauth_cred_t cred, int uproto)
 2752 {
 2753         struct cmsghdr *cm = 0;
 2754 
 2755         if (control == NULL || opt == NULL)
 2756                 return (EINVAL);
 2757 
 2758         ip6_initpktopts(opt);
 2759         if (stickyopt) {
 2760                 int error;
 2761 
 2762                 /*
 2763                  * If stickyopt is provided, make a local copy of the options
 2764                  * for this particular packet, then override them by ancillary
 2765                  * objects.
 2766                  * XXX: copypktopts() does not copy the cached route to a next
 2767                  * hop (if any).  This is not very good in terms of efficiency,
 2768                  * but we can allow this since this option should be rarely
 2769                  * used.
 2770                  */
 2771                 if ((error = copypktopts(opt, stickyopt, M_NOWAIT)) != 0)
 2772                         return (error);
 2773         }
 2774 
 2775         /*
 2776          * XXX: Currently, we assume all the optional information is stored
 2777          * in a single mbuf.
 2778          */
 2779         if (control->m_next)
 2780                 return (EINVAL);
 2781 
 2782         /* XXX if cm->cmsg_len is not aligned, control->m_len can become <0 */
 2783         for (; control->m_len > 0; control->m_data += CMSG_ALIGN(cm->cmsg_len),
 2784             control->m_len -= CMSG_ALIGN(cm->cmsg_len)) {
 2785                 int error;
 2786 
 2787                 if (control->m_len < CMSG_LEN(0))
 2788                         return (EINVAL);
 2789 
 2790                 cm = mtod(control, struct cmsghdr *);
 2791                 if (cm->cmsg_len < CMSG_LEN(0) || cm->cmsg_len > control->m_len)
 2792                         return (EINVAL);
 2793                 if (cm->cmsg_level != IPPROTO_IPV6)
 2794                         continue;
 2795 
 2796                 error = ip6_setpktopt(cm->cmsg_type, CMSG_DATA(cm),
 2797                     cm->cmsg_len - CMSG_LEN(0), opt, cred, 0, 1, uproto);
 2798                 if (error)
 2799                         return (error);
 2800         }
 2801 
 2802         return (0);
 2803 }
 2804 
 2805 /*
 2806  * Set a particular packet option, as a sticky option or an ancillary data
 2807  * item.  "len" can be 0 only when it's a sticky option.
 2808  * We have 4 cases of combination of "sticky" and "cmsg":
 2809  * "sticky=0, cmsg=0": impossible
 2810  * "sticky=0, cmsg=1": RFC2292 or RFC3542 ancillary data
 2811  * "sticky=1, cmsg=0": RFC3542 socket option
 2812  * "sticky=1, cmsg=1": RFC2292 socket option
 2813  */
 2814 static int
 2815 ip6_setpktopt(int optname, u_char *buf, int len, struct ip6_pktopts *opt,
 2816     kauth_cred_t cred, int sticky, int cmsg, int uproto)
 2817 {
 2818         int minmtupolicy;
 2819         int error;
 2820 
 2821         if (!sticky && !cmsg) {
 2822 #ifdef DIAGNOSTIC
 2823                 printf("ip6_setpktopt: impossible case\n");
 2824 #endif
 2825                 return (EINVAL);
 2826         }
 2827 
 2828         /*
 2829          * IPV6_2292xxx is for backward compatibility to RFC2292, and should
 2830          * not be specified in the context of RFC3542.  Conversely,
 2831          * RFC3542 types should not be specified in the context of RFC2292.
 2832          */
 2833         if (!cmsg) {
 2834                 switch (optname) {
 2835                 case IPV6_2292PKTINFO:
 2836                 case IPV6_2292HOPLIMIT:
 2837                 case IPV6_2292NEXTHOP:
 2838                 case IPV6_2292HOPOPTS:
 2839                 case IPV6_2292DSTOPTS:
 2840                 case IPV6_2292RTHDR:
 2841                 case IPV6_2292PKTOPTIONS:
 2842                         return (ENOPROTOOPT);
 2843                 }
 2844         }
 2845         if (sticky && cmsg) {
 2846                 switch (optname) {
 2847                 case IPV6_PKTINFO:
 2848                 case IPV6_HOPLIMIT:
 2849                 case IPV6_NEXTHOP:
 2850                 case IPV6_HOPOPTS:
 2851                 case IPV6_DSTOPTS:
 2852                 case IPV6_RTHDRDSTOPTS:
 2853                 case IPV6_RTHDR:
 2854                 case IPV6_USE_MIN_MTU:
 2855                 case IPV6_DONTFRAG:
 2856                 case IPV6_OTCLASS:
 2857                 case IPV6_TCLASS:
 2858                 case IPV6_PREFER_TEMPADDR: /* XXX not an RFC3542 option */
 2859                         return (ENOPROTOOPT);
 2860                 }
 2861         }
 2862 
 2863         switch (optname) {
 2864 #ifdef RFC2292
 2865         case IPV6_2292PKTINFO:
 2866 #endif
 2867         case IPV6_PKTINFO:
 2868         {
 2869                 struct in6_pktinfo *pktinfo;
 2870 
 2871                 if (len != sizeof(struct in6_pktinfo))
 2872                         return (EINVAL);
 2873 
 2874                 pktinfo = (struct in6_pktinfo *)buf;
 2875 
 2876                 /*
 2877                  * An application can clear any sticky IPV6_PKTINFO option by
 2878                  * doing a "regular" setsockopt with ipi6_addr being
 2879                  * in6addr_any and ipi6_ifindex being zero.
 2880                  * [RFC 3542, Section 6]
 2881                  */
 2882                 if (optname == IPV6_PKTINFO && opt->ip6po_pktinfo &&
 2883                     pktinfo->ipi6_ifindex == 0 &&
 2884                     IN6_IS_ADDR_UNSPECIFIED(&pktinfo->ipi6_addr)) {
 2885                         ip6_clearpktopts(opt, optname);
 2886                         break;
 2887                 }
 2888 
 2889                 if (uproto == IPPROTO_TCP && optname == IPV6_PKTINFO &&
 2890                     sticky && !IN6_IS_ADDR_UNSPECIFIED(&pktinfo->ipi6_addr)) {
 2891                         return (EINVAL);
 2892                 }
 2893 
 2894                 /* Validate the interface index if specified. */
 2895                 if (pktinfo->ipi6_ifindex) {
 2896                         struct ifnet *ifp;
 2897                         int s = pserialize_read_enter();
 2898                         ifp = if_byindex(pktinfo->ipi6_ifindex);
 2899                         if (ifp == NULL) {
 2900                                 pserialize_read_exit(s);
 2901                                 return ENXIO;
 2902                         }
 2903                         pserialize_read_exit(s);
 2904                 }
 2905 
 2906                 /*
 2907                  * We store the address anyway, and let in6_selectsrc()
 2908                  * validate the specified address.  This is because ipi6_addr
 2909                  * may not have enough information about its scope zone, and
 2910                  * we may need additional information (such as outgoing
 2911                  * interface or the scope zone of a destination address) to
 2912                  * disambiguate the scope.
 2913                  * XXX: the delay of the validation may confuse the
 2914                  * application when it is used as a sticky option.
 2915                  */
 2916                 if (opt->ip6po_pktinfo == NULL) {
 2917                         opt->ip6po_pktinfo = malloc(sizeof(*pktinfo),
 2918                             M_IP6OPT, M_NOWAIT);
 2919                         if (opt->ip6po_pktinfo == NULL)
 2920                                 return (ENOBUFS);
 2921                 }
 2922                 memcpy(opt->ip6po_pktinfo, pktinfo, sizeof(*pktinfo));
 2923                 break;
 2924         }
 2925 
 2926 #ifdef RFC2292
 2927         case IPV6_2292HOPLIMIT:
 2928 #endif
 2929         case IPV6_HOPLIMIT:
 2930         {
 2931                 int *hlimp;
 2932 
 2933                 /*
 2934                  * RFC 3542 deprecated the usage of sticky IPV6_HOPLIMIT
 2935                  * to simplify the ordering among hoplimit options.
 2936                  */
 2937                 if (optname == IPV6_HOPLIMIT && sticky)
 2938                         return (ENOPROTOOPT);
 2939 
 2940                 if (len != sizeof(int))
 2941                         return (EINVAL);
 2942                 hlimp = (int *)buf;
 2943                 if (*hlimp < -1 || *hlimp > 255)
 2944                         return (EINVAL);
 2945 
 2946                 opt->ip6po_hlim = *hlimp;
 2947                 break;
 2948         }
 2949 
 2950         case IPV6_OTCLASS:
 2951                 if (len != sizeof(u_int8_t))
 2952                         return (EINVAL);
 2953 
 2954                 opt->ip6po_tclass = *(u_int8_t *)buf;
 2955                 break;
 2956 
 2957         case IPV6_TCLASS:
 2958         {
 2959                 int tclass;
 2960 
 2961                 if (len != sizeof(int))
 2962                         return (EINVAL);
 2963                 tclass = *(int *)buf;
 2964                 if (tclass < -1 || tclass > 255)
 2965                         return (EINVAL);
 2966 
 2967                 opt->ip6po_tclass = tclass;
 2968                 break;
 2969         }
 2970 
 2971 #ifdef RFC2292
 2972         case IPV6_2292NEXTHOP:
 2973 #endif
 2974         case IPV6_NEXTHOP:
 2975                 error = kauth_authorize_network(cred,
 2976                     KAUTH_NETWORK_IPV6,
 2977                     KAUTH_REQ_NETWORK_IPV6_HOPBYHOP, NULL, NULL, NULL);
 2978                 if (error)
 2979                         return (error);
 2980 
 2981                 if (len == 0) { /* just remove the option */
 2982                         ip6_clearpktopts(opt, IPV6_NEXTHOP);
 2983                         break;
 2984                 }
 2985 
 2986                 /* check if cmsg_len is large enough for sa_len */
 2987                 if (len < sizeof(struct sockaddr) || len < *buf)
 2988                         return (EINVAL);
 2989 
 2990                 switch (((struct sockaddr *)buf)->sa_family) {
 2991                 case AF_INET6:
 2992                 {
 2993                         struct sockaddr_in6 *sa6 = (struct sockaddr_in6 *)buf;
 2994 
 2995                         if (sa6->sin6_len != sizeof(struct sockaddr_in6))
 2996                                 return (EINVAL);
 2997 
 2998                         if (IN6_IS_ADDR_UNSPECIFIED(&sa6->sin6_addr) ||
 2999                             IN6_IS_ADDR_MULTICAST(&sa6->sin6_addr)) {
 3000                                 return (EINVAL);
 3001                         }
 3002                         if ((error = sa6_embedscope(sa6, ip6_use_defzone))
 3003                             != 0) {
 3004                                 return (error);
 3005                         }
 3006                         break;
 3007                 }
 3008                 case AF_LINK:   /* eventually be supported? */
 3009                 default:
 3010                         return (EAFNOSUPPORT);
 3011                 }
 3012 
 3013                 /* turn off the previous option, then set the new option. */
 3014                 ip6_clearpktopts(opt, IPV6_NEXTHOP);
 3015                 opt->ip6po_nexthop = malloc(*buf, M_IP6OPT, M_NOWAIT);
 3016                 if (opt->ip6po_nexthop == NULL)
 3017                         return (ENOBUFS);
 3018                 memcpy(opt->ip6po_nexthop, buf, *buf);
 3019                 break;
 3020 
 3021 #ifdef RFC2292
 3022         case IPV6_2292HOPOPTS:
 3023 #endif
 3024         case IPV6_HOPOPTS:
 3025         {
 3026                 struct ip6_hbh *hbh;
 3027                 int hbhlen;
 3028 
 3029                 /*
 3030                  * XXX: We don't allow a non-privileged user to set ANY HbH
 3031                  * options, since per-option restriction has too much
 3032                  * overhead.
 3033                  */
 3034                 error = kauth_authorize_network(cred,
 3035                     KAUTH_NETWORK_IPV6,
 3036                     KAUTH_REQ_NETWORK_IPV6_HOPBYHOP, NULL, NULL, NULL);
 3037                 if (error)
 3038                         return (error);
 3039 
 3040                 if (len == 0) {
 3041                         ip6_clearpktopts(opt, IPV6_HOPOPTS);
 3042                         break;  /* just remove the option */
 3043                 }
 3044 
 3045                 /* message length validation */
 3046                 if (len < sizeof(struct ip6_hbh))
 3047                         return (EINVAL);
 3048                 hbh = (struct ip6_hbh *)buf;
 3049                 hbhlen = (hbh->ip6h_len + 1) << 3;
 3050                 if (len != hbhlen)
 3051                         return (EINVAL);
 3052 
 3053                 /* turn off the previous option, then set the new option. */
 3054                 ip6_clearpktopts(opt, IPV6_HOPOPTS);
 3055                 opt->ip6po_hbh = malloc(hbhlen, M_IP6OPT, M_NOWAIT);
 3056                 if (opt->ip6po_hbh == NULL)
 3057                         return (ENOBUFS);
 3058                 memcpy(opt->ip6po_hbh, hbh, hbhlen);
 3059 
 3060                 break;
 3061         }
 3062 
 3063 #ifdef RFC2292
 3064         case IPV6_2292DSTOPTS:
 3065 #endif
 3066         case IPV6_DSTOPTS:
 3067         case IPV6_RTHDRDSTOPTS:
 3068         {
 3069                 struct ip6_dest *dest, **newdest = NULL;
 3070                 int destlen;
 3071 
 3072                 /* XXX: see the comment for IPV6_HOPOPTS */
 3073                 error = kauth_authorize_network(cred,
 3074                     KAUTH_NETWORK_IPV6,
 3075                     KAUTH_REQ_NETWORK_IPV6_HOPBYHOP, NULL, NULL, NULL);
 3076                 if (error)
 3077                         return (error);
 3078 
 3079                 if (len == 0) {
 3080                         ip6_clearpktopts(opt, optname);
 3081                         break;  /* just remove the option */
 3082                 }
 3083 
 3084                 /* message length validation */
 3085                 if (len < sizeof(struct ip6_dest))
 3086                         return (EINVAL);
 3087                 dest = (struct ip6_dest *)buf;
 3088                 destlen = (dest->ip6d_len + 1) << 3;
 3089                 if (len != destlen)
 3090                         return (EINVAL);
 3091                 /*
 3092                  * Determine the position that the destination options header
 3093                  * should be inserted; before or after the routing header.
 3094                  */
 3095                 switch (optname) {
 3096                 case IPV6_2292DSTOPTS:
 3097                         /*
 3098                          * The old advanced API is ambiguous on this point.
 3099                          * Our approach is to determine the position based
 3100                          * according to the existence of a routing header.
 3101                          * Note, however, that this depends on the order of the
 3102                          * extension headers in the ancillary data; the 1st
 3103                          * part of the destination options header must appear
 3104                          * before the routing header in the ancillary data,
 3105                          * too.
 3106                          * RFC3542 solved the ambiguity by introducing
 3107                          * separate ancillary data or option types.
 3108                          */
 3109                         if (opt->ip6po_rthdr == NULL)
 3110                                 newdest = &opt->ip6po_dest1;
 3111                         else
 3112                                 newdest = &opt->ip6po_dest2;
 3113                         break;
 3114                 case IPV6_RTHDRDSTOPTS:
 3115                         newdest = &opt->ip6po_dest1;
 3116                         break;
 3117                 case IPV6_DSTOPTS:
 3118                         newdest = &opt->ip6po_dest2;
 3119                         break;
 3120                 }
 3121 
 3122                 /* turn off the previous option, then set the new option. */
 3123                 ip6_clearpktopts(opt, optname);
 3124                 *newdest = malloc(destlen, M_IP6OPT, M_NOWAIT);
 3125                 if (*newdest == NULL)
 3126                         return (ENOBUFS);
 3127                 memcpy(*newdest, dest, destlen);
 3128 
 3129                 break;
 3130         }
 3131 
 3132 #ifdef RFC2292
 3133         case IPV6_2292RTHDR:
 3134 #endif
 3135         case IPV6_RTHDR:
 3136         {
 3137                 struct ip6_rthdr *rth;
 3138                 int rthlen;
 3139 
 3140                 if (len == 0) {
 3141                         ip6_clearpktopts(opt, IPV6_RTHDR);
 3142                         break;  /* just remove the option */
 3143                 }
 3144 
 3145                 /* message length validation */
 3146                 if (len < sizeof(struct ip6_rthdr))
 3147                         return (EINVAL);
 3148                 rth = (struct ip6_rthdr *)buf;
 3149                 rthlen = (rth->ip6r_len + 1) << 3;
 3150                 if (len != rthlen)
 3151                         return (EINVAL);
 3152                 switch (rth->ip6r_type) {
 3153                 case IPV6_RTHDR_TYPE_0:
 3154                         /* Dropped, RFC5095. */
 3155                 default:
 3156                         return (EINVAL);        /* not supported */
 3157                 }
 3158                 /* turn off the previous option */
 3159                 ip6_clearpktopts(opt, IPV6_RTHDR);
 3160                 opt->ip6po_rthdr = malloc(rthlen, M_IP6OPT, M_NOWAIT);
 3161                 if (opt->ip6po_rthdr == NULL)
 3162                         return (ENOBUFS);
 3163                 memcpy(opt->ip6po_rthdr, rth, rthlen);
 3164                 break;
 3165         }
 3166 
 3167         case IPV6_USE_MIN_MTU:
 3168                 if (len != sizeof(int))
 3169                         return (EINVAL);
 3170                 minmtupolicy = *(int *)buf;
 3171                 if (minmtupolicy != IP6PO_MINMTU_MCASTONLY &&
 3172                     minmtupolicy != IP6PO_MINMTU_DISABLE &&
 3173                     minmtupolicy != IP6PO_MINMTU_ALL) {
 3174                         return (EINVAL);
 3175                 }
 3176                 opt->ip6po_minmtu = minmtupolicy;
 3177                 break;
 3178 
 3179         case IPV6_DONTFRAG:
 3180                 if (len != sizeof(int))
 3181                         return (EINVAL);
 3182 
 3183                 if (uproto == IPPROTO_TCP || *(int *)buf == 0) {
 3184                         /*
 3185                          * we ignore this option for TCP sockets.
 3186                          * (RFC3542 leaves this case unspecified.)
 3187                          */
 3188                         opt->ip6po_flags &= ~IP6PO_DONTFRAG;
 3189                 } else
 3190                         opt->ip6po_flags |= IP6PO_DONTFRAG;
 3191                 break;
 3192 
 3193         case IPV6_PREFER_TEMPADDR:
 3194         {
 3195                 int preftemp;
 3196 
 3197                 if (len != sizeof(int))
 3198                         return (EINVAL);
 3199                 preftemp = *(int *)buf;
 3200                 switch (preftemp) {
 3201                 case IP6PO_TEMPADDR_SYSTEM:
 3202                 case IP6PO_TEMPADDR_NOTPREFER:
 3203                 case IP6PO_TEMPADDR_PREFER:
 3204                         break;
 3205                 default:
 3206                         return (EINVAL);
 3207                 }
 3208                 opt->ip6po_prefer_tempaddr = preftemp;
 3209                 break;
 3210         }
 3211 
 3212         default:
 3213                 return (ENOPROTOOPT);
 3214         } /* end of switch */
 3215 
 3216         return (0);
 3217 }
 3218 
 3219 /*
 3220  * Routine called from ip6_output() to loop back a copy of an IP6 multicast
 3221  * packet to the input queue of a specified interface.  Note that this
 3222  * calls the output routine of the loopback "driver", but with an interface
 3223  * pointer that might NOT be lo0ifp -- easier than replicating that code here.
 3224  */
 3225 void
 3226 ip6_mloopback(struct ifnet *ifp, struct mbuf *m,
 3227         const struct sockaddr_in6 *dst)
 3228 {
 3229         struct mbuf *copym;
 3230         struct ip6_hdr *ip6;
 3231 
 3232         copym = m_copypacket(m, M_DONTWAIT);
 3233         if (copym == NULL)
 3234                 return;
 3235 
 3236         /*
 3237          * Make sure to deep-copy IPv6 header portion in case the data
 3238          * is in an mbuf cluster, so that we can safely override the IPv6
 3239          * header portion later.
 3240          */
 3241         if ((copym->m_flags & M_EXT) != 0 ||
 3242             copym->m_len < sizeof(struct ip6_hdr)) {
 3243                 copym = m_pullup(copym, sizeof(struct ip6_hdr));
 3244                 if (copym == NULL)
 3245                         return;
 3246         }
 3247 
 3248 #ifdef DIAGNOSTIC
 3249         if (copym->m_len < sizeof(*ip6)) {
 3250                 m_freem(copym);
 3251                 return;
 3252         }
 3253 #endif
 3254 
 3255         ip6 = mtod(copym, struct ip6_hdr *);
 3256         /*
 3257          * clear embedded scope identifiers if necessary.
 3258          * in6_clearscope will touch the addresses only when necessary.
 3259          */
 3260         in6_clearscope(&ip6->ip6_src);
 3261         in6_clearscope(&ip6->ip6_dst);
 3262 
 3263         (void)looutput(ifp, copym, (const struct sockaddr *)dst, NULL);
 3264 }
 3265 
 3266 /*
 3267  * Chop IPv6 header off from the payload.
 3268  */
 3269 static int
 3270 ip6_splithdr(struct mbuf *m, struct ip6_exthdrs *exthdrs)
 3271 {
 3272         struct mbuf *mh;
 3273         struct ip6_hdr *ip6;
 3274 
 3275         ip6 = mtod(m, struct ip6_hdr *);
 3276         if (m->m_len > sizeof(*ip6)) {
 3277                 MGETHDR(mh, M_DONTWAIT, MT_HEADER);
 3278                 if (mh == NULL) {
 3279                         m_freem(m);
 3280                         return ENOBUFS;
 3281                 }
 3282                 m_move_pkthdr(mh, m);
 3283                 m_align(mh, sizeof(*ip6));
 3284                 m->m_len -= sizeof(*ip6);
 3285                 m->m_data += sizeof(*ip6);
 3286                 mh->m_next = m;
 3287                 mh->m_len = sizeof(*ip6);
 3288                 memcpy(mtod(mh, void *), (void *)ip6, sizeof(*ip6));
 3289                 m = mh;
 3290         }
 3291         exthdrs->ip6e_ip6 = m;
 3292         return 0;
 3293 }
 3294 
 3295 /*
 3296  * Compute IPv6 extension header length.
 3297  */
 3298 int
 3299 ip6_optlen(struct inpcb *inp)
 3300 {
 3301         int len;
 3302 
 3303         if (!in6p_outputopts(inp))
 3304                 return 0;
 3305 
 3306         len = 0;
 3307 #define elen(x) \
 3308     (((struct ip6_ext *)(x)) ? (((struct ip6_ext *)(x))->ip6e_len + 1) << 3 : 0)
 3309 
 3310         len += elen(in6p_outputopts(inp)->ip6po_hbh);
 3311         len += elen(in6p_outputopts(inp)->ip6po_dest1);
 3312         len += elen(in6p_outputopts(inp)->ip6po_rthdr);
 3313         len += elen(in6p_outputopts(inp)->ip6po_dest2);
 3314         return len;
 3315 #undef elen
 3316 }
 3317 
 3318 /*
 3319  * Ensure sending address is valid.
 3320  * Returns 0 on success, -1 if an error should be sent back or 1
 3321  * if the packet could be dropped without error (protocol dependent).
 3322  */
 3323 static int
 3324 ip6_ifaddrvalid(const struct in6_addr *src, const struct in6_addr *dst)
 3325 {
 3326         struct sockaddr_in6 sin6;
 3327         int s, error;
 3328         struct ifaddr *ifa;
 3329         struct in6_ifaddr *ia6;
 3330 
 3331         if (IN6_IS_ADDR_UNSPECIFIED(src))
 3332                 return 0;
 3333 
 3334         memset(&sin6, 0, sizeof(sin6));
 3335         sin6.sin6_family = AF_INET6;
 3336         sin6.sin6_len = sizeof(sin6);
 3337         sin6.sin6_addr = *src;
 3338 
 3339         s = pserialize_read_enter();
 3340         ifa = ifa_ifwithaddr(sin6tosa(&sin6));
 3341         if ((ia6 = ifatoia6(ifa)) == NULL ||
 3342             ia6->ia6_flags & (IN6_IFF_ANYCAST | IN6_IFF_DUPLICATED))
 3343                 error = -1;
 3344         else if (ia6->ia6_flags & IN6_IFF_TENTATIVE)
 3345                 error = 1;
 3346         else if (ia6->ia6_flags & IN6_IFF_DETACHED &&
 3347             (sin6.sin6_addr = *dst, ifa_ifwithaddr(sin6tosa(&sin6)) == NULL))
 3348                 /* Allow internal traffic to DETACHED addresses */
 3349                 error = 1;
 3350         else
 3351                 error = 0;
 3352         pserialize_read_exit(s);
 3353 
 3354         return error;
 3355 }

Cache object: 6eae9336feff25afb94bebe8060f38b1


[ 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.