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/in6_src.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 /*-
    2  * SPDX-License-Identifier: BSD-3-Clause
    3  *
    4  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
    5  * All rights reserved.
    6  *
    7  * Redistribution and use in source and binary forms, with or without
    8  * modification, are permitted provided that the following conditions
    9  * are met:
   10  * 1. Redistributions of source code must retain the above copyright
   11  *    notice, this list of conditions and the following disclaimer.
   12  * 2. Redistributions in binary form must reproduce the above copyright
   13  *    notice, this list of conditions and the following disclaimer in the
   14  *    documentation and/or other materials provided with the distribution.
   15  * 3. Neither the name of the project nor the names of its contributors
   16  *    may be used to endorse or promote products derived from this software
   17  *    without specific prior written permission.
   18  *
   19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
   20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
   21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
   22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
   23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
   24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
   25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
   26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
   27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
   28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
   29  * SUCH DAMAGE.
   30  *
   31  *      $KAME: in6_src.c,v 1.132 2003/08/26 04:42:27 keiichi Exp $
   32  */
   33 
   34 /*-
   35  * Copyright (c) 1982, 1986, 1991, 1993
   36  *      The Regents of the University of California.  All rights reserved.
   37  *
   38  * Redistribution and use in source and binary forms, with or without
   39  * modification, are permitted provided that the following conditions
   40  * are met:
   41  * 1. Redistributions of source code must retain the above copyright
   42  *    notice, this list of conditions and the following disclaimer.
   43  * 2. Redistributions in binary form must reproduce the above copyright
   44  *    notice, this list of conditions and the following disclaimer in the
   45  *    documentation and/or other materials provided with the distribution.
   46  * 3. Neither the name of the University nor the names of its contributors
   47  *    may be used to endorse or promote products derived from this software
   48  *    without specific prior written permission.
   49  *
   50  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
   51  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
   52  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
   53  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
   54  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
   55  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
   56  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
   57  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
   58  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
   59  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
   60  * SUCH DAMAGE.
   61  *
   62  *      @(#)in_pcb.c    8.2 (Berkeley) 1/4/94
   63  */
   64 
   65 #include <sys/cdefs.h>
   66 __FBSDID("$FreeBSD$");
   67 
   68 #include "opt_inet.h"
   69 #include "opt_inet6.h"
   70 #include "opt_mpath.h"
   71 
   72 #include <sys/param.h>
   73 #include <sys/systm.h>
   74 #include <sys/lock.h>
   75 #include <sys/malloc.h>
   76 #include <sys/mbuf.h>
   77 #include <sys/priv.h>
   78 #include <sys/protosw.h>
   79 #include <sys/socket.h>
   80 #include <sys/socketvar.h>
   81 #include <sys/sockio.h>
   82 #include <sys/sysctl.h>
   83 #include <sys/errno.h>
   84 #include <sys/time.h>
   85 #include <sys/jail.h>
   86 #include <sys/kernel.h>
   87 #include <sys/rmlock.h>
   88 #include <sys/sx.h>
   89 
   90 #include <net/if.h>
   91 #include <net/if_var.h>
   92 #include <net/if_dl.h>
   93 #include <net/route.h>
   94 #include <net/if_llatbl.h>
   95 #ifdef RADIX_MPATH
   96 #include <net/radix_mpath.h>
   97 #endif
   98 
   99 #include <netinet/in.h>
  100 #include <netinet/in_var.h>
  101 #include <netinet/in_systm.h>
  102 #include <netinet/ip.h>
  103 #include <netinet/in_pcb.h>
  104 #include <netinet/ip_var.h>
  105 #include <netinet/udp.h>
  106 #include <netinet/udp_var.h>
  107 
  108 #include <netinet6/in6_var.h>
  109 #include <netinet/ip6.h>
  110 #include <netinet6/in6_fib.h>
  111 #include <netinet6/in6_pcb.h>
  112 #include <netinet6/ip6_var.h>
  113 #include <netinet6/scope6_var.h>
  114 #include <netinet6/nd6.h>
  115 
  116 static struct mtx addrsel_lock;
  117 #define ADDRSEL_LOCK_INIT()     mtx_init(&addrsel_lock, "addrsel_lock", NULL, MTX_DEF)
  118 #define ADDRSEL_LOCK()          mtx_lock(&addrsel_lock)
  119 #define ADDRSEL_UNLOCK()        mtx_unlock(&addrsel_lock)
  120 #define ADDRSEL_LOCK_ASSERT()   mtx_assert(&addrsel_lock, MA_OWNED)
  121 
  122 static struct sx addrsel_sxlock;
  123 #define ADDRSEL_SXLOCK_INIT()   sx_init(&addrsel_sxlock, "addrsel_sxlock")
  124 #define ADDRSEL_SLOCK()         sx_slock(&addrsel_sxlock)
  125 #define ADDRSEL_SUNLOCK()       sx_sunlock(&addrsel_sxlock)
  126 #define ADDRSEL_XLOCK()         sx_xlock(&addrsel_sxlock)
  127 #define ADDRSEL_XUNLOCK()       sx_xunlock(&addrsel_sxlock)
  128 
  129 #define ADDR_LABEL_NOTAPP (-1)
  130 VNET_DEFINE_STATIC(struct in6_addrpolicy, defaultaddrpolicy);
  131 #define V_defaultaddrpolicy             VNET(defaultaddrpolicy)
  132 
  133 VNET_DEFINE(int, ip6_prefer_tempaddr) = 0;
  134 
  135 static int selectroute(struct sockaddr_in6 *, struct ip6_pktopts *,
  136         struct ip6_moptions *, struct route_in6 *, struct ifnet **,
  137         struct rtentry **, int, u_int);
  138 static int in6_selectif(struct sockaddr_in6 *, struct ip6_pktopts *,
  139         struct ip6_moptions *, struct ifnet **,
  140         struct ifnet *, u_int);
  141 static int in6_selectsrc(uint32_t, struct sockaddr_in6 *,
  142         struct ip6_pktopts *, struct inpcb *, struct ucred *,
  143         struct ifnet **, struct in6_addr *);
  144 
  145 static struct in6_addrpolicy *lookup_addrsel_policy(struct sockaddr_in6 *);
  146 
  147 static void init_policy_queue(void);
  148 static int add_addrsel_policyent(struct in6_addrpolicy *);
  149 static int delete_addrsel_policyent(struct in6_addrpolicy *);
  150 static int walk_addrsel_policy(int (*)(struct in6_addrpolicy *, void *),
  151         void *);
  152 static int dump_addrsel_policyent(struct in6_addrpolicy *, void *);
  153 static struct in6_addrpolicy *match_addrsel_policy(struct sockaddr_in6 *);
  154 
  155 /*
  156  * Return an IPv6 address, which is the most appropriate for a given
  157  * destination and user specified options.
  158  * If necessary, this function lookups the routing table and returns
  159  * an entry to the caller for later use.
  160  */
  161 #define REPLACE(r) do {\
  162         IP6STAT_INC(ip6s_sources_rule[(r)]); \
  163         /* { \
  164         char ip6buf[INET6_ADDRSTRLEN], ip6b[INET6_ADDRSTRLEN]; \
  165         printf("in6_selectsrc: replace %s with %s by %d\n", ia_best ? ip6_sprintf(ip6buf, &ia_best->ia_addr.sin6_addr) : "none", ip6_sprintf(ip6b, &ia->ia_addr.sin6_addr), (r)); \
  166         } */ \
  167         goto replace; \
  168 } while(0)
  169 #define NEXT(r) do {\
  170         /* { \
  171         char ip6buf[INET6_ADDRSTRLEN], ip6b[INET6_ADDRSTRLEN]; \
  172         printf("in6_selectsrc: keep %s against %s by %d\n", ia_best ? ip6_sprintf(ip6buf, &ia_best->ia_addr.sin6_addr) : "none", ip6_sprintf(ip6b, &ia->ia_addr.sin6_addr), (r)); \
  173         } */ \
  174         goto next;              /* XXX: we can't use 'continue' here */ \
  175 } while(0)
  176 #define BREAK(r) do { \
  177         IP6STAT_INC(ip6s_sources_rule[(r)]); \
  178         goto out;               /* XXX: we can't use 'break' here */ \
  179 } while(0)
  180 
  181 static int
  182 in6_selectsrc(uint32_t fibnum, struct sockaddr_in6 *dstsock,
  183     struct ip6_pktopts *opts, struct inpcb *inp, struct ucred *cred,
  184     struct ifnet **ifpp, struct in6_addr *srcp)
  185 {
  186         struct rm_priotracker in6_ifa_tracker;
  187         struct in6_addr dst, tmp;
  188         struct ifnet *ifp = NULL, *oifp = NULL;
  189         struct in6_ifaddr *ia = NULL, *ia_best = NULL;
  190         struct in6_pktinfo *pi = NULL;
  191         int dst_scope = -1, best_scope = -1, best_matchlen = -1;
  192         struct in6_addrpolicy *dst_policy = NULL, *best_policy = NULL;
  193         u_int32_t odstzone;
  194         int prefer_tempaddr;
  195         int error;
  196         struct ip6_moptions *mopts;
  197 
  198         KASSERT(srcp != NULL, ("%s: srcp is NULL", __func__));
  199 
  200         dst = dstsock->sin6_addr; /* make a copy for local operation */
  201         if (ifpp) {
  202                 /*
  203                  * Save a possibly passed in ifp for in6_selectsrc. Only
  204                  * neighbor discovery code should use this feature, where
  205                  * we may know the interface but not the FIB number holding
  206                  * the connected subnet in case someone deleted it from the
  207                  * default FIB and we need to check the interface.
  208                  */
  209                 if (*ifpp != NULL)
  210                         oifp = *ifpp;
  211                 *ifpp = NULL;
  212         }
  213 
  214         if (inp != NULL) {
  215                 INP_LOCK_ASSERT(inp);
  216                 mopts = inp->in6p_moptions;
  217         } else {
  218                 mopts = NULL;
  219         }
  220 
  221         /*
  222          * If the source address is explicitly specified by the caller,
  223          * check if the requested source address is indeed a unicast address
  224          * assigned to the node, and can be used as the packet's source
  225          * address.  If everything is okay, use the address as source.
  226          */
  227         if (opts && (pi = opts->ip6po_pktinfo) &&
  228             !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) {
  229                 /* get the outgoing interface */
  230                 if ((error = in6_selectif(dstsock, opts, mopts, &ifp, oifp,
  231                     fibnum))
  232                     != 0)
  233                         return (error);
  234 
  235                 /*
  236                  * determine the appropriate zone id of the source based on
  237                  * the zone of the destination and the outgoing interface.
  238                  * If the specified address is ambiguous wrt the scope zone,
  239                  * the interface must be specified; otherwise, ifa_ifwithaddr()
  240                  * will fail matching the address.
  241                  */
  242                 tmp = pi->ipi6_addr;
  243                 if (ifp) {
  244                         error = in6_setscope(&tmp, ifp, &odstzone);
  245                         if (error)
  246                                 return (error);
  247                 }
  248                 if (cred != NULL && (error = prison_local_ip6(cred,
  249                     &tmp, (inp->inp_flags & IN6P_IPV6_V6ONLY) != 0)) != 0)
  250                         return (error);
  251 
  252                 /*
  253                  * If IPV6_BINDANY socket option is set, we allow to specify
  254                  * non local addresses as source address in IPV6_PKTINFO
  255                  * ancillary data.
  256                  */
  257                 if ((inp->inp_flags & INP_BINDANY) == 0) {
  258                         ia = in6ifa_ifwithaddr(&tmp, 0 /* XXX */);
  259                         if (ia == NULL || (ia->ia6_flags & (IN6_IFF_ANYCAST |
  260                             IN6_IFF_NOTREADY))) {
  261                                 if (ia != NULL)
  262                                         ifa_free(&ia->ia_ifa);
  263                                 return (EADDRNOTAVAIL);
  264                         }
  265                         bcopy(&ia->ia_addr.sin6_addr, srcp, sizeof(*srcp));
  266                         ifa_free(&ia->ia_ifa);
  267                 } else
  268                         bcopy(&tmp, srcp, sizeof(*srcp));
  269                 pi->ipi6_addr = tmp; /* XXX: this overrides pi */
  270                 if (ifpp)
  271                         *ifpp = ifp;
  272                 return (0);
  273         }
  274 
  275         /*
  276          * Otherwise, if the socket has already bound the source, just use it.
  277          */
  278         if (inp != NULL && !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
  279                 if (cred != NULL &&
  280                     (error = prison_local_ip6(cred, &inp->in6p_laddr,
  281                     ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
  282                         return (error);
  283                 bcopy(&inp->in6p_laddr, srcp, sizeof(*srcp));
  284                 return (0);
  285         }
  286 
  287         /*
  288          * Bypass source address selection and use the primary jail IP
  289          * if requested.
  290          */
  291         if (cred != NULL && !prison_saddrsel_ip6(cred, srcp))
  292                 return (0);
  293 
  294         /*
  295          * If the address is not specified, choose the best one based on
  296          * the outgoing interface and the destination address.
  297          */
  298         /* get the outgoing interface */
  299         if ((error = in6_selectif(dstsock, opts, mopts, &ifp, oifp,
  300             (inp != NULL) ? inp->inp_inc.inc_fibnum : fibnum)) != 0)
  301                 return (error);
  302 
  303 #ifdef DIAGNOSTIC
  304         if (ifp == NULL)        /* this should not happen */
  305                 panic("in6_selectsrc: NULL ifp");
  306 #endif
  307         error = in6_setscope(&dst, ifp, &odstzone);
  308         if (error)
  309                 return (error);
  310 
  311         IN6_IFADDR_RLOCK(&in6_ifa_tracker);
  312         CK_STAILQ_FOREACH(ia, &V_in6_ifaddrhead, ia_link) {
  313                 int new_scope = -1, new_matchlen = -1;
  314                 struct in6_addrpolicy *new_policy = NULL;
  315                 u_int32_t srczone, osrczone, dstzone;
  316                 struct in6_addr src;
  317                 struct ifnet *ifp1 = ia->ia_ifp;
  318 
  319                 /*
  320                  * We'll never take an address that breaks the scope zone
  321                  * of the destination.  We also skip an address if its zone
  322                  * does not contain the outgoing interface.
  323                  * XXX: we should probably use sin6_scope_id here.
  324                  */
  325                 if (in6_setscope(&dst, ifp1, &dstzone) ||
  326                     odstzone != dstzone) {
  327                         continue;
  328                 }
  329                 src = ia->ia_addr.sin6_addr;
  330                 if (in6_setscope(&src, ifp, &osrczone) ||
  331                     in6_setscope(&src, ifp1, &srczone) ||
  332                     osrczone != srczone) {
  333                         continue;
  334                 }
  335 
  336                 /* avoid unusable addresses */
  337                 if ((ia->ia6_flags &
  338                      (IN6_IFF_NOTREADY | IN6_IFF_ANYCAST | IN6_IFF_DETACHED))) {
  339                                 continue;
  340                 }
  341                 if (!V_ip6_use_deprecated && IFA6_IS_DEPRECATED(ia))
  342                         continue;
  343 
  344                 /* If jailed only take addresses of the jail into account. */
  345                 if (cred != NULL &&
  346                     prison_check_ip6(cred, &ia->ia_addr.sin6_addr) != 0)
  347                         continue;
  348 
  349                 /* Rule 1: Prefer same address */
  350                 if (IN6_ARE_ADDR_EQUAL(&dst, &ia->ia_addr.sin6_addr)) {
  351                         ia_best = ia;
  352                         BREAK(1); /* there should be no better candidate */
  353                 }
  354 
  355                 if (ia_best == NULL)
  356                         REPLACE(0);
  357 
  358                 /* Rule 2: Prefer appropriate scope */
  359                 if (dst_scope < 0)
  360                         dst_scope = in6_addrscope(&dst);
  361                 new_scope = in6_addrscope(&ia->ia_addr.sin6_addr);
  362                 if (IN6_ARE_SCOPE_CMP(best_scope, new_scope) < 0) {
  363                         if (IN6_ARE_SCOPE_CMP(best_scope, dst_scope) < 0)
  364                                 REPLACE(2);
  365                         NEXT(2);
  366                 } else if (IN6_ARE_SCOPE_CMP(new_scope, best_scope) < 0) {
  367                         if (IN6_ARE_SCOPE_CMP(new_scope, dst_scope) < 0)
  368                                 NEXT(2);
  369                         REPLACE(2);
  370                 }
  371 
  372                 /*
  373                  * Rule 3: Avoid deprecated addresses.  Note that the case of
  374                  * !ip6_use_deprecated is already rejected above.
  375                  */
  376                 if (!IFA6_IS_DEPRECATED(ia_best) && IFA6_IS_DEPRECATED(ia))
  377                         NEXT(3);
  378                 if (IFA6_IS_DEPRECATED(ia_best) && !IFA6_IS_DEPRECATED(ia))
  379                         REPLACE(3);
  380 
  381                 /* Rule 4: Prefer home addresses */
  382                 /*
  383                  * XXX: This is a TODO.  We should probably merge the MIP6
  384                  * case above.
  385                  */
  386 
  387                 /* Rule 5: Prefer outgoing interface */
  388                 if (!(ND_IFINFO(ifp)->flags & ND6_IFF_NO_PREFER_IFACE)) {
  389                         if (ia_best->ia_ifp == ifp && ia->ia_ifp != ifp)
  390                                 NEXT(5);
  391                         if (ia_best->ia_ifp != ifp && ia->ia_ifp == ifp)
  392                                 REPLACE(5);
  393                 }
  394 
  395                 /*
  396                  * Rule 6: Prefer matching label
  397                  * Note that best_policy should be non-NULL here.
  398                  */
  399                 if (dst_policy == NULL)
  400                         dst_policy = lookup_addrsel_policy(dstsock);
  401                 if (dst_policy->label != ADDR_LABEL_NOTAPP) {
  402                         new_policy = lookup_addrsel_policy(&ia->ia_addr);
  403                         if (dst_policy->label == best_policy->label &&
  404                             dst_policy->label != new_policy->label)
  405                                 NEXT(6);
  406                         if (dst_policy->label != best_policy->label &&
  407                             dst_policy->label == new_policy->label)
  408                                 REPLACE(6);
  409                 }
  410 
  411                 /*
  412                  * Rule 7: Prefer public addresses.
  413                  * We allow users to reverse the logic by configuring
  414                  * a sysctl variable, so that privacy conscious users can
  415                  * always prefer temporary addresses.
  416                  */
  417                 if (opts == NULL ||
  418                     opts->ip6po_prefer_tempaddr == IP6PO_TEMPADDR_SYSTEM) {
  419                         prefer_tempaddr = V_ip6_prefer_tempaddr;
  420                 } else if (opts->ip6po_prefer_tempaddr ==
  421                     IP6PO_TEMPADDR_NOTPREFER) {
  422                         prefer_tempaddr = 0;
  423                 } else
  424                         prefer_tempaddr = 1;
  425                 if (!(ia_best->ia6_flags & IN6_IFF_TEMPORARY) &&
  426                     (ia->ia6_flags & IN6_IFF_TEMPORARY)) {
  427                         if (prefer_tempaddr)
  428                                 REPLACE(7);
  429                         else
  430                                 NEXT(7);
  431                 }
  432                 if ((ia_best->ia6_flags & IN6_IFF_TEMPORARY) &&
  433                     !(ia->ia6_flags & IN6_IFF_TEMPORARY)) {
  434                         if (prefer_tempaddr)
  435                                 NEXT(7);
  436                         else
  437                                 REPLACE(7);
  438                 }
  439 
  440                 /*
  441                  * Rule 8: prefer addresses on alive interfaces.
  442                  * This is a KAME specific rule.
  443                  */
  444                 if ((ia_best->ia_ifp->if_flags & IFF_UP) &&
  445                     !(ia->ia_ifp->if_flags & IFF_UP))
  446                         NEXT(8);
  447                 if (!(ia_best->ia_ifp->if_flags & IFF_UP) &&
  448                     (ia->ia_ifp->if_flags & IFF_UP))
  449                         REPLACE(8);
  450 
  451                 /*
  452                  * Rule 9: prefer address with better virtual status.
  453                  */
  454                 if (ifa_preferred(&ia_best->ia_ifa, &ia->ia_ifa))
  455                         REPLACE(9);
  456                 if (ifa_preferred(&ia->ia_ifa, &ia_best->ia_ifa))
  457                         NEXT(9);
  458 
  459                 /*
  460                  * Rule 10: prefer address with `prefer_source' flag.
  461                  */
  462                 if ((ia_best->ia6_flags & IN6_IFF_PREFER_SOURCE) == 0 &&
  463                     (ia->ia6_flags & IN6_IFF_PREFER_SOURCE) != 0)
  464                         REPLACE(10);
  465                 if ((ia_best->ia6_flags & IN6_IFF_PREFER_SOURCE) != 0 &&
  466                     (ia->ia6_flags & IN6_IFF_PREFER_SOURCE) == 0)
  467                         NEXT(10);
  468 
  469                 /*
  470                  * Rule 14: Use longest matching prefix.
  471                  * Note: in the address selection draft, this rule is
  472                  * documented as "Rule 8".  However, since it is also
  473                  * documented that this rule can be overridden, we assign
  474                  * a large number so that it is easy to assign smaller numbers
  475                  * to more preferred rules.
  476                  */
  477                 new_matchlen = in6_matchlen(&ia->ia_addr.sin6_addr, &dst);
  478                 if (best_matchlen < new_matchlen)
  479                         REPLACE(14);
  480                 if (new_matchlen < best_matchlen)
  481                         NEXT(14);
  482 
  483                 /* Rule 15 is reserved. */
  484 
  485                 /*
  486                  * Last resort: just keep the current candidate.
  487                  * Or, do we need more rules?
  488                  */
  489                 continue;
  490 
  491           replace:
  492                 ia_best = ia;
  493                 best_scope = (new_scope >= 0 ? new_scope :
  494                               in6_addrscope(&ia_best->ia_addr.sin6_addr));
  495                 best_policy = (new_policy ? new_policy :
  496                                lookup_addrsel_policy(&ia_best->ia_addr));
  497                 best_matchlen = (new_matchlen >= 0 ? new_matchlen :
  498                                  in6_matchlen(&ia_best->ia_addr.sin6_addr,
  499                                               &dst));
  500 
  501           next:
  502                 continue;
  503 
  504           out:
  505                 break;
  506         }
  507 
  508         if ((ia = ia_best) == NULL) {
  509                 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
  510                 IP6STAT_INC(ip6s_sources_none);
  511                 return (EADDRNOTAVAIL);
  512         }
  513 
  514         /*
  515          * At this point at least one of the addresses belonged to the jail
  516          * but it could still be, that we want to further restrict it, e.g.
  517          * theoratically IN6_IS_ADDR_LOOPBACK.
  518          * It must not be IN6_IS_ADDR_UNSPECIFIED anymore.
  519          * prison_local_ip6() will fix an IN6_IS_ADDR_LOOPBACK but should
  520          * let all others previously selected pass.
  521          * Use tmp to not change ::1 on lo0 to the primary jail address.
  522          */
  523         tmp = ia->ia_addr.sin6_addr;
  524         if (cred != NULL && prison_local_ip6(cred, &tmp, (inp != NULL &&
  525             (inp->inp_flags & IN6P_IPV6_V6ONLY) != 0)) != 0) {
  526                 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
  527                 IP6STAT_INC(ip6s_sources_none);
  528                 return (EADDRNOTAVAIL);
  529         }
  530 
  531         if (ifpp)
  532                 *ifpp = ifp;
  533 
  534         bcopy(&tmp, srcp, sizeof(*srcp));
  535         if (ia->ia_ifp == ifp)
  536                 IP6STAT_INC(ip6s_sources_sameif[best_scope]);
  537         else
  538                 IP6STAT_INC(ip6s_sources_otherif[best_scope]);
  539         if (dst_scope == best_scope)
  540                 IP6STAT_INC(ip6s_sources_samescope[best_scope]);
  541         else
  542                 IP6STAT_INC(ip6s_sources_otherscope[best_scope]);
  543         if (IFA6_IS_DEPRECATED(ia))
  544                 IP6STAT_INC(ip6s_sources_deprecated[best_scope]);
  545         IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
  546         return (0);
  547 }
  548 
  549 /*
  550  * Select source address based on @inp, @dstsock and @opts.
  551  * Stores selected address to @srcp. If @scope_ambiguous is set,
  552  * embed scope from selected outgoing interface. If @hlim pointer
  553  * is provided, stores calculated hop limit there.
  554  * Returns 0 on success.
  555  */
  556 int
  557 in6_selectsrc_socket(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts,
  558     struct inpcb *inp, struct ucred *cred, int scope_ambiguous,
  559     struct in6_addr *srcp, int *hlim)
  560 {
  561         struct ifnet *retifp;
  562         uint32_t fibnum;
  563         int error;
  564 
  565         fibnum = inp->inp_inc.inc_fibnum;
  566         retifp = NULL;
  567 
  568         error = in6_selectsrc(fibnum, dstsock, opts, inp, cred, &retifp, srcp);
  569         if (error != 0)
  570                 return (error);
  571 
  572         if (hlim != NULL)
  573                 *hlim = in6_selecthlim(inp, retifp);
  574 
  575         if (retifp == NULL || scope_ambiguous == 0)
  576                 return (0);
  577 
  578         /*
  579          * Application should provide a proper zone ID or the use of
  580          * default zone IDs should be enabled.  Unfortunately, some
  581          * applications do not behave as it should, so we need a
  582          * workaround.  Even if an appropriate ID is not determined
  583          * (when it's required), if we can determine the outgoing
  584          * interface. determine the zone ID based on the interface.
  585          */
  586         error = in6_setscope(&dstsock->sin6_addr, retifp, NULL);
  587 
  588         return (error);
  589 }
  590 
  591 /*
  592  * Select source address based on @fibnum, @dst and @scopeid.
  593  * Stores selected address to @srcp.
  594  * Returns 0 on success.
  595  *
  596  * Used by non-socket based consumers (ND code mostly)
  597  */
  598 int
  599 in6_selectsrc_addr(uint32_t fibnum, const struct in6_addr *dst,
  600     uint32_t scopeid, struct ifnet *ifp, struct in6_addr *srcp,
  601     int *hlim)
  602 {
  603         struct ifnet *retifp;
  604         struct sockaddr_in6 dst_sa;
  605         int error;
  606 
  607         retifp = ifp;
  608         bzero(&dst_sa, sizeof(dst_sa));
  609         dst_sa.sin6_family = AF_INET6;
  610         dst_sa.sin6_len = sizeof(dst_sa);
  611         dst_sa.sin6_addr = *dst;
  612         dst_sa.sin6_scope_id = scopeid;
  613         sa6_embedscope(&dst_sa, 0);
  614 
  615         error = in6_selectsrc(fibnum, &dst_sa, NULL, NULL, NULL, &retifp, srcp);
  616         if (hlim != NULL)
  617                 *hlim = in6_selecthlim(NULL, retifp);
  618 
  619         return (error);
  620 }
  621 
  622 /*
  623  * clone - meaningful only for bsdi and freebsd
  624  */
  625 static int
  626 selectroute(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts,
  627     struct ip6_moptions *mopts, struct route_in6 *ro,
  628     struct ifnet **retifp, struct rtentry **retrt, int norouteok, u_int fibnum)
  629 {
  630         int error = 0;
  631         struct ifnet *ifp = NULL;
  632         struct rtentry *rt = NULL;
  633         struct sockaddr_in6 *sin6_next;
  634         struct in6_pktinfo *pi = NULL;
  635         struct in6_addr *dst = &dstsock->sin6_addr;
  636         uint32_t zoneid;
  637 #if 0
  638         char ip6buf[INET6_ADDRSTRLEN];
  639 
  640         if (dstsock->sin6_addr.s6_addr32[0] == 0 &&
  641             dstsock->sin6_addr.s6_addr32[1] == 0 &&
  642             !IN6_IS_ADDR_LOOPBACK(&dstsock->sin6_addr)) {
  643                 printf("in6_selectroute: strange destination %s\n",
  644                        ip6_sprintf(ip6buf, &dstsock->sin6_addr));
  645         } else {
  646                 printf("in6_selectroute: destination = %s%%%d\n",
  647                        ip6_sprintf(ip6buf, &dstsock->sin6_addr),
  648                        dstsock->sin6_scope_id); /* for debug */
  649         }
  650 #endif
  651 
  652         /* If the caller specify the outgoing interface explicitly, use it. */
  653         if (opts && (pi = opts->ip6po_pktinfo) != NULL && pi->ipi6_ifindex) {
  654                 /* XXX boundary check is assumed to be already done. */
  655                 ifp = ifnet_byindex(pi->ipi6_ifindex);
  656                 if (ifp != NULL &&
  657                     (norouteok || retrt == NULL ||
  658                     IN6_IS_ADDR_MULTICAST(dst))) {
  659                         /*
  660                          * we do not have to check or get the route for
  661                          * multicast.
  662                          */
  663                         goto done;
  664                 } else
  665                         goto getroute;
  666         }
  667         /*
  668          * If the destination address is a multicast address and the outgoing
  669          * interface for the address is specified by the caller, use it.
  670          */
  671         if (IN6_IS_ADDR_MULTICAST(dst) &&
  672             mopts != NULL && (ifp = mopts->im6o_multicast_ifp) != NULL) {
  673                 goto done; /* we do not need a route for multicast. */
  674         }
  675         /*
  676          * If destination address is LLA or link- or node-local multicast,
  677          * use it's embedded scope zone id to determine outgoing interface.
  678          */
  679         if (IN6_IS_ADDR_MC_LINKLOCAL(dst) ||
  680             IN6_IS_ADDR_MC_NODELOCAL(dst)) {
  681                 zoneid = ntohs(in6_getscope(dst));
  682                 if (zoneid > 0) {
  683                         ifp = in6_getlinkifnet(zoneid);
  684                         goto done;
  685                 }
  686         }
  687 
  688   getroute:
  689         /*
  690          * If the next hop address for the packet is specified by the caller,
  691          * use it as the gateway.
  692          */
  693         if (opts && opts->ip6po_nexthop) {
  694                 struct route_in6 *ron;
  695 
  696                 sin6_next = satosin6(opts->ip6po_nexthop);
  697                 if (IN6_IS_ADDR_LINKLOCAL(&sin6_next->sin6_addr)) {
  698                         /*
  699                          * Next hop is LLA, thus it should be neighbor.
  700                          * Determine outgoing interface by zone index.
  701                          */
  702                         zoneid = ntohs(in6_getscope(&sin6_next->sin6_addr));
  703                         if (zoneid > 0) {
  704                                 ifp = in6_getlinkifnet(zoneid);
  705                                 goto done;
  706                         }
  707                 }
  708                 ron = &opts->ip6po_nextroute;
  709                 /* Use a cached route if it exists and is valid. */
  710                 if (ron->ro_rt != NULL && (
  711                     (ron->ro_rt->rt_flags & RTF_UP) == 0 ||
  712                     ron->ro_dst.sin6_family != AF_INET6 ||
  713                     !IN6_ARE_ADDR_EQUAL(&ron->ro_dst.sin6_addr,
  714                         &sin6_next->sin6_addr)))
  715                         RO_RTFREE(ron);
  716                 if (ron->ro_rt == NULL) {
  717                         ron->ro_dst = *sin6_next;
  718                         in6_rtalloc(ron, fibnum); /* multi path case? */
  719                 }
  720                 /*
  721                  * The node identified by that address must be a
  722                  * neighbor of the sending host.
  723                  */
  724                 if (ron->ro_rt == NULL ||
  725                     (ron->ro_rt->rt_flags & RTF_GATEWAY) != 0)
  726                         error = EHOSTUNREACH;
  727                 else {
  728                         rt = ron->ro_rt;
  729                         ifp = rt->rt_ifp;
  730                 }
  731                 goto done;
  732         }
  733 
  734         /*
  735          * Use a cached route if it exists and is valid, else try to allocate
  736          * a new one.  Note that we should check the address family of the
  737          * cached destination, in case of sharing the cache with IPv4.
  738          */
  739         if (ro) {
  740                 if (ro->ro_rt &&
  741                     (!(ro->ro_rt->rt_flags & RTF_UP) ||
  742                      ((struct sockaddr *)(&ro->ro_dst))->sa_family != AF_INET6 ||
  743                      !IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr,
  744                      dst))) {
  745                         RTFREE(ro->ro_rt);
  746                         ro->ro_rt = (struct rtentry *)NULL;
  747                 }
  748                 if (ro->ro_rt == (struct rtentry *)NULL) {
  749                         struct sockaddr_in6 *sa6;
  750 
  751                         /* No route yet, so try to acquire one */
  752                         bzero(&ro->ro_dst, sizeof(struct sockaddr_in6));
  753                         sa6 = (struct sockaddr_in6 *)&ro->ro_dst;
  754                         *sa6 = *dstsock;
  755                         sa6->sin6_scope_id = 0;
  756 
  757 #ifdef RADIX_MPATH
  758                                 rtalloc_mpath_fib((struct route *)ro,
  759                                     ntohl(sa6->sin6_addr.s6_addr32[3]), fibnum);
  760 #else                   
  761                                 ro->ro_rt = in6_rtalloc1((struct sockaddr *)
  762                                     &ro->ro_dst, 0, 0UL, fibnum);
  763                                 if (ro->ro_rt)
  764                                         RT_UNLOCK(ro->ro_rt);
  765 #endif
  766                 }
  767                                 
  768                 /*
  769                  * do not care about the result if we have the nexthop
  770                  * explicitly specified.
  771                  */
  772                 if (opts && opts->ip6po_nexthop)
  773                         goto done;
  774 
  775                 if (ro->ro_rt) {
  776                         ifp = ro->ro_rt->rt_ifp;
  777 
  778                         if (ifp == NULL) { /* can this really happen? */
  779                                 RTFREE(ro->ro_rt);
  780                                 ro->ro_rt = NULL;
  781                         }
  782                 }
  783                 if (ro->ro_rt == NULL)
  784                         error = EHOSTUNREACH;
  785                 rt = ro->ro_rt;
  786 
  787                 /*
  788                  * Check if the outgoing interface conflicts with
  789                  * the interface specified by ipi6_ifindex (if specified).
  790                  * Note that loopback interface is always okay.
  791                  * (this may happen when we are sending a packet to one of
  792                  *  our own addresses.)
  793                  */
  794                 if (ifp && opts && opts->ip6po_pktinfo &&
  795                     opts->ip6po_pktinfo->ipi6_ifindex) {
  796                         if (!(ifp->if_flags & IFF_LOOPBACK) &&
  797                             ifp->if_index !=
  798                             opts->ip6po_pktinfo->ipi6_ifindex) {
  799                                 error = EHOSTUNREACH;
  800                                 goto done;
  801                         }
  802                 }
  803         }
  804 
  805   done:
  806         if (ifp == NULL && rt == NULL) {
  807                 /*
  808                  * This can happen if the caller did not pass a cached route
  809                  * nor any other hints.  We treat this case an error.
  810                  */
  811                 error = EHOSTUNREACH;
  812         }
  813         if (error == EHOSTUNREACH)
  814                 IP6STAT_INC(ip6s_noroute);
  815 
  816         if (retifp != NULL) {
  817                 *retifp = ifp;
  818 
  819                 /*
  820                  * Adjust the "outgoing" interface.  If we're going to loop 
  821                  * the packet back to ourselves, the ifp would be the loopback 
  822                  * interface. However, we'd rather know the interface associated 
  823                  * to the destination address (which should probably be one of 
  824                  * our own addresses.)
  825                  */
  826                 if (rt) {
  827                         if ((rt->rt_ifp->if_flags & IFF_LOOPBACK) &&
  828                             (rt->rt_gateway->sa_family == AF_LINK))
  829                                 *retifp = 
  830                                         ifnet_byindex(((struct sockaddr_dl *)
  831                                                        rt->rt_gateway)->sdl_index);
  832                 }
  833         }
  834 
  835         if (retrt != NULL)
  836                 *retrt = rt;    /* rt may be NULL */
  837 
  838         return (error);
  839 }
  840 
  841 static int
  842 in6_selectif(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts,
  843     struct ip6_moptions *mopts, struct ifnet **retifp,
  844     struct ifnet *oifp, u_int fibnum)
  845 {
  846         int error;
  847         struct route_in6 sro;
  848         struct rtentry *rt = NULL;
  849         int rt_flags;
  850 
  851         KASSERT(retifp != NULL, ("%s: retifp is NULL", __func__));
  852 
  853         bzero(&sro, sizeof(sro));
  854         rt_flags = 0;
  855 
  856         error = selectroute(dstsock, opts, mopts, &sro, retifp, &rt, 1, fibnum);
  857 
  858         if (rt)
  859                 rt_flags = rt->rt_flags;
  860         if (rt && rt == sro.ro_rt)
  861                 RTFREE(rt);
  862 
  863         if (error != 0) {
  864                 /* Help ND. See oifp comment in in6_selectsrc(). */
  865                 if (oifp != NULL && fibnum == RT_DEFAULT_FIB) {
  866                         *retifp = oifp;
  867                         error = 0;
  868                 }
  869                 return (error);
  870         }
  871 
  872         /*
  873          * do not use a rejected or black hole route.
  874          * XXX: this check should be done in the L2 output routine.
  875          * However, if we skipped this check here, we'd see the following
  876          * scenario:
  877          * - install a rejected route for a scoped address prefix
  878          *   (like fe80::/10)
  879          * - send a packet to a destination that matches the scoped prefix,
  880          *   with ambiguity about the scope zone.
  881          * - pick the outgoing interface from the route, and disambiguate the
  882          *   scope zone with the interface.
  883          * - ip6_output() would try to get another route with the "new"
  884          *   destination, which may be valid.
  885          * - we'd see no error on output.
  886          * Although this may not be very harmful, it should still be confusing.
  887          * We thus reject the case here.
  888          */
  889 
  890         if (rt_flags & (RTF_REJECT | RTF_BLACKHOLE)) {
  891                 error = (rt_flags & RTF_HOST ? EHOSTUNREACH : ENETUNREACH);
  892                 return (error);
  893         }
  894 
  895         return (0);
  896 }
  897 
  898 /*
  899  * Public wrapper function to selectroute().
  900  *
  901  * XXX-BZ in6_selectroute() should and will grow the FIB argument. The
  902  * in6_selectroute_fib() function is only there for backward compat on stable.
  903  */
  904 int
  905 in6_selectroute(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts,
  906     struct ip6_moptions *mopts, struct route_in6 *ro,
  907     struct ifnet **retifp, struct rtentry **retrt)
  908 {
  909 
  910         return (selectroute(dstsock, opts, mopts, ro, retifp,
  911             retrt, 0, RT_DEFAULT_FIB));
  912 }
  913 
  914 #ifndef BURN_BRIDGES
  915 int
  916 in6_selectroute_fib(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts,
  917     struct ip6_moptions *mopts, struct route_in6 *ro,
  918     struct ifnet **retifp, struct rtentry **retrt, u_int fibnum)
  919 {
  920 
  921         return (selectroute(dstsock, opts, mopts, ro, retifp,
  922             retrt, 0, fibnum));
  923 }
  924 #endif
  925 
  926 /*
  927  * Default hop limit selection. The precedence is as follows:
  928  * 1. Hoplimit value specified via ioctl.
  929  * 2. (If the outgoing interface is detected) the current
  930  *     hop limit of the interface specified by router advertisement.
  931  * 3. The system default hoplimit.
  932  */
  933 int
  934 in6_selecthlim(struct inpcb *inp, struct ifnet *ifp)
  935 {
  936 
  937         if (inp && inp->in6p_hops >= 0)
  938                 return (inp->in6p_hops);
  939         else if (ifp)
  940                 return (ND_IFINFO(ifp)->chlim);
  941         else if (inp && !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) {
  942                 struct nhop6_basic nh6;
  943                 struct in6_addr dst;
  944                 uint32_t fibnum, scopeid;
  945                 int hlim;
  946 
  947                 fibnum = inp->inp_inc.inc_fibnum;
  948                 in6_splitscope(&inp->in6p_faddr, &dst, &scopeid);
  949                 if (fib6_lookup_nh_basic(fibnum, &dst, scopeid, 0, 0, &nh6)==0){
  950                         hlim = ND_IFINFO(nh6.nh_ifp)->chlim;
  951                         return (hlim);
  952                 }
  953         }
  954         return (V_ip6_defhlim);
  955 }
  956 
  957 /*
  958  * XXX: this is borrowed from in6_pcbbind(). If possible, we should
  959  * share this function by all *bsd*...
  960  */
  961 int
  962 in6_pcbsetport(struct in6_addr *laddr, struct inpcb *inp, struct ucred *cred)
  963 {
  964         struct socket *so = inp->inp_socket;
  965         u_int16_t lport = 0;
  966         int error, lookupflags = 0;
  967 #ifdef INVARIANTS
  968         struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
  969 #endif
  970 
  971         INP_WLOCK_ASSERT(inp);
  972         INP_HASH_WLOCK_ASSERT(pcbinfo);
  973 
  974         error = prison_local_ip6(cred, laddr,
  975             ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0));
  976         if (error)
  977                 return(error);
  978 
  979         /* XXX: this is redundant when called from in6_pcbbind */
  980         if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT|SO_REUSEPORT_LB)) == 0)
  981                 lookupflags = INPLOOKUP_WILDCARD;
  982 
  983         inp->inp_flags |= INP_ANONPORT;
  984 
  985         error = in_pcb_lport(inp, NULL, &lport, cred, lookupflags);
  986         if (error != 0)
  987                 return (error);
  988 
  989         inp->inp_lport = lport;
  990         if (in_pcbinshash(inp) != 0) {
  991                 inp->in6p_laddr = in6addr_any;
  992                 inp->inp_lport = 0;
  993                 return (EAGAIN);
  994         }
  995 
  996         return (0);
  997 }
  998 
  999 void
 1000 addrsel_policy_init(void)
 1001 {
 1002 
 1003         init_policy_queue();
 1004 
 1005         /* initialize the "last resort" policy */
 1006         bzero(&V_defaultaddrpolicy, sizeof(V_defaultaddrpolicy));
 1007         V_defaultaddrpolicy.label = ADDR_LABEL_NOTAPP;
 1008 
 1009         if (!IS_DEFAULT_VNET(curvnet))
 1010                 return;
 1011 
 1012         ADDRSEL_LOCK_INIT();
 1013         ADDRSEL_SXLOCK_INIT();
 1014 }
 1015 
 1016 static struct in6_addrpolicy *
 1017 lookup_addrsel_policy(struct sockaddr_in6 *key)
 1018 {
 1019         struct in6_addrpolicy *match = NULL;
 1020 
 1021         ADDRSEL_LOCK();
 1022         match = match_addrsel_policy(key);
 1023 
 1024         if (match == NULL)
 1025                 match = &V_defaultaddrpolicy;
 1026         else
 1027                 match->use++;
 1028         ADDRSEL_UNLOCK();
 1029 
 1030         return (match);
 1031 }
 1032 
 1033 /*
 1034  * Subroutines to manage the address selection policy table via sysctl.
 1035  */
 1036 struct walkarg {
 1037         struct sysctl_req *w_req;
 1038 };
 1039 
 1040 static int in6_src_sysctl(SYSCTL_HANDLER_ARGS);
 1041 SYSCTL_DECL(_net_inet6_ip6);
 1042 static SYSCTL_NODE(_net_inet6_ip6, IPV6CTL_ADDRCTLPOLICY, addrctlpolicy,
 1043         CTLFLAG_RD, in6_src_sysctl, "");
 1044 
 1045 static int
 1046 in6_src_sysctl(SYSCTL_HANDLER_ARGS)
 1047 {
 1048         struct walkarg w;
 1049 
 1050         if (req->newptr)
 1051                 return EPERM;
 1052 
 1053         bzero(&w, sizeof(w));
 1054         w.w_req = req;
 1055 
 1056         return (walk_addrsel_policy(dump_addrsel_policyent, &w));
 1057 }
 1058 
 1059 int
 1060 in6_src_ioctl(u_long cmd, caddr_t data)
 1061 {
 1062         struct in6_addrpolicy ent0;
 1063 
 1064         if (cmd != SIOCAADDRCTL_POLICY && cmd != SIOCDADDRCTL_POLICY)
 1065                 return (EOPNOTSUPP); /* check for safety */
 1066 
 1067         ent0 = *(struct in6_addrpolicy *)data;
 1068 
 1069         if (ent0.label == ADDR_LABEL_NOTAPP)
 1070                 return (EINVAL);
 1071         /* check if the prefix mask is consecutive. */
 1072         if (in6_mask2len(&ent0.addrmask.sin6_addr, NULL) < 0)
 1073                 return (EINVAL);
 1074         /* clear trailing garbages (if any) of the prefix address. */
 1075         IN6_MASK_ADDR(&ent0.addr.sin6_addr, &ent0.addrmask.sin6_addr);
 1076         ent0.use = 0;
 1077 
 1078         switch (cmd) {
 1079         case SIOCAADDRCTL_POLICY:
 1080                 return (add_addrsel_policyent(&ent0));
 1081         case SIOCDADDRCTL_POLICY:
 1082                 return (delete_addrsel_policyent(&ent0));
 1083         }
 1084 
 1085         return (0);             /* XXX: compromise compilers */
 1086 }
 1087 
 1088 /*
 1089  * The followings are implementation of the policy table using a
 1090  * simple tail queue.
 1091  * XXX such details should be hidden.
 1092  * XXX implementation using binary tree should be more efficient.
 1093  */
 1094 struct addrsel_policyent {
 1095         TAILQ_ENTRY(addrsel_policyent) ape_entry;
 1096         struct in6_addrpolicy ape_policy;
 1097 };
 1098 
 1099 TAILQ_HEAD(addrsel_policyhead, addrsel_policyent);
 1100 
 1101 VNET_DEFINE_STATIC(struct addrsel_policyhead, addrsel_policytab);
 1102 #define V_addrsel_policytab             VNET(addrsel_policytab)
 1103 
 1104 static void
 1105 init_policy_queue(void)
 1106 {
 1107 
 1108         TAILQ_INIT(&V_addrsel_policytab);
 1109 }
 1110 
 1111 static int
 1112 add_addrsel_policyent(struct in6_addrpolicy *newpolicy)
 1113 {
 1114         struct addrsel_policyent *new, *pol;
 1115 
 1116         new = malloc(sizeof(*new), M_IFADDR,
 1117                M_WAITOK);
 1118         ADDRSEL_XLOCK();
 1119         ADDRSEL_LOCK();
 1120 
 1121         /* duplication check */
 1122         TAILQ_FOREACH(pol, &V_addrsel_policytab, ape_entry) {
 1123                 if (IN6_ARE_ADDR_EQUAL(&newpolicy->addr.sin6_addr,
 1124                                        &pol->ape_policy.addr.sin6_addr) &&
 1125                     IN6_ARE_ADDR_EQUAL(&newpolicy->addrmask.sin6_addr,
 1126                                        &pol->ape_policy.addrmask.sin6_addr)) {
 1127                         ADDRSEL_UNLOCK();
 1128                         ADDRSEL_XUNLOCK();
 1129                         free(new, M_IFADDR);
 1130                         return (EEXIST);        /* or override it? */
 1131                 }
 1132         }
 1133 
 1134         bzero(new, sizeof(*new));
 1135 
 1136         /* XXX: should validate entry */
 1137         new->ape_policy = *newpolicy;
 1138 
 1139         TAILQ_INSERT_TAIL(&V_addrsel_policytab, new, ape_entry);
 1140         ADDRSEL_UNLOCK();
 1141         ADDRSEL_XUNLOCK();
 1142 
 1143         return (0);
 1144 }
 1145 
 1146 static int
 1147 delete_addrsel_policyent(struct in6_addrpolicy *key)
 1148 {
 1149         struct addrsel_policyent *pol;
 1150 
 1151         ADDRSEL_XLOCK();
 1152         ADDRSEL_LOCK();
 1153 
 1154         /* search for the entry in the table */
 1155         TAILQ_FOREACH(pol, &V_addrsel_policytab, ape_entry) {
 1156                 if (IN6_ARE_ADDR_EQUAL(&key->addr.sin6_addr,
 1157                     &pol->ape_policy.addr.sin6_addr) &&
 1158                     IN6_ARE_ADDR_EQUAL(&key->addrmask.sin6_addr,
 1159                     &pol->ape_policy.addrmask.sin6_addr)) {
 1160                         break;
 1161                 }
 1162         }
 1163         if (pol == NULL) {
 1164                 ADDRSEL_UNLOCK();
 1165                 ADDRSEL_XUNLOCK();
 1166                 return (ESRCH);
 1167         }
 1168 
 1169         TAILQ_REMOVE(&V_addrsel_policytab, pol, ape_entry);
 1170         ADDRSEL_UNLOCK();
 1171         ADDRSEL_XUNLOCK();
 1172         free(pol, M_IFADDR);
 1173 
 1174         return (0);
 1175 }
 1176 
 1177 static int
 1178 walk_addrsel_policy(int (*callback)(struct in6_addrpolicy *, void *), void *w)
 1179 {
 1180         struct addrsel_policyent *pol;
 1181         int error = 0;
 1182 
 1183         ADDRSEL_SLOCK();
 1184         TAILQ_FOREACH(pol, &V_addrsel_policytab, ape_entry) {
 1185                 if ((error = (*callback)(&pol->ape_policy, w)) != 0) {
 1186                         ADDRSEL_SUNLOCK();
 1187                         return (error);
 1188                 }
 1189         }
 1190         ADDRSEL_SUNLOCK();
 1191         return (error);
 1192 }
 1193 
 1194 static int
 1195 dump_addrsel_policyent(struct in6_addrpolicy *pol, void *arg)
 1196 {
 1197         int error = 0;
 1198         struct walkarg *w = arg;
 1199 
 1200         error = SYSCTL_OUT(w->w_req, pol, sizeof(*pol));
 1201 
 1202         return (error);
 1203 }
 1204 
 1205 static struct in6_addrpolicy *
 1206 match_addrsel_policy(struct sockaddr_in6 *key)
 1207 {
 1208         struct addrsel_policyent *pent;
 1209         struct in6_addrpolicy *bestpol = NULL, *pol;
 1210         int matchlen, bestmatchlen = -1;
 1211         u_char *mp, *ep, *k, *p, m;
 1212 
 1213         TAILQ_FOREACH(pent, &V_addrsel_policytab, ape_entry) {
 1214                 matchlen = 0;
 1215 
 1216                 pol = &pent->ape_policy;
 1217                 mp = (u_char *)&pol->addrmask.sin6_addr;
 1218                 ep = mp + 16;   /* XXX: scope field? */
 1219                 k = (u_char *)&key->sin6_addr;
 1220                 p = (u_char *)&pol->addr.sin6_addr;
 1221                 for (; mp < ep && *mp; mp++, k++, p++) {
 1222                         m = *mp;
 1223                         if ((*k & m) != *p)
 1224                                 goto next; /* not match */
 1225                         if (m == 0xff) /* short cut for a typical case */
 1226                                 matchlen += 8;
 1227                         else {
 1228                                 while (m >= 0x80) {
 1229                                         matchlen++;
 1230                                         m <<= 1;
 1231                                 }
 1232                         }
 1233                 }
 1234 
 1235                 /* matched.  check if this is better than the current best. */
 1236                 if (bestpol == NULL ||
 1237                     matchlen > bestmatchlen) {
 1238                         bestpol = pol;
 1239                         bestmatchlen = matchlen;
 1240                 }
 1241 
 1242           next:
 1243                 continue;
 1244         }
 1245 
 1246         return (bestpol);
 1247 }

Cache object: ca83811841d10084942ca690645b7e25


[ 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.