root/net/route.c

/* [<][>][^][v][top][bottom][index][help] */

DEFINITIONS

This source file includes following definitions.
  1. encap_findgwifa
  2. rtable_init
  3. route_init
  4. rtable_add
  5. rtable_exists
  6. rtalloc_noclone
  7. okaytoclone
  8. rtalloc2
  9. rtalloc
  10. rtalloc1
  11. rtfree
  12. ifafree
  13. rtredirect
  14. rtdeletemsg
  15. rtflushclone1
  16. rtflushclone
  17. rtioctl
  18. ifa_ifwithroute
  19. rtrequest
  20. rt_getifa
  21. rtrequest1
  22. rt_setgate
  23. rt_maskedcopy
  24. rtinit
  25. rt_timer_init
  26. rt_timer_queue_create
  27. rt_timer_queue_change
  28. rt_timer_queue_destroy
  29. rt_timer_count
  30. rt_timer_remove_all
  31. rt_timer_add
  32. rt_gettable
  33. rt_lookup
  34. rt_timer_timer
  35. rtlabel_name2id
  36. rtlabel_id2name
  37. rtlabel_unref
  38. rt_if_remove
  39. rt_if_remove_rtdelete

    1 /*      $OpenBSD: route.c,v 1.84 2007/06/14 18:31:49 reyk Exp $ */
    2 /*      $NetBSD: route.c,v 1.14 1996/02/13 22:00:46 christos 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) 1980, 1986, 1991, 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  *      @(#)route.c     8.2 (Berkeley) 11/15/93
   62  */
   63 
   64 /*
   65  *      @(#)COPYRIGHT   1.1 (NRL) 17 January 1995
   66  * 
   67  * NRL grants permission for redistribution and use in source and binary
   68  * forms, with or without modification, of the software and documentation
   69  * created at NRL provided that the following conditions are met:
   70  * 
   71  * 1. Redistributions of source code must retain the above copyright
   72  *    notice, this list of conditions and the following disclaimer.
   73  * 2. Redistributions in binary form must reproduce the above copyright
   74  *    notice, this list of conditions and the following disclaimer in the
   75  *    documentation and/or other materials provided with the distribution.
   76  * 3. All advertising materials mentioning features or use of this software
   77  *    must display the following acknowledgements:
   78  *      This product includes software developed by the University of
   79  *      California, Berkeley and its contributors.
   80  *      This product includes software developed at the Information
   81  *      Technology Division, US Naval Research Laboratory.
   82  * 4. Neither the name of the NRL nor the names of its contributors
   83  *    may be used to endorse or promote products derived from this software
   84  *    without specific prior written permission.
   85  * 
   86  * THE SOFTWARE PROVIDED BY NRL IS PROVIDED BY NRL AND CONTRIBUTORS ``AS
   87  * IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
   88  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
   89  * PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL NRL OR
   90  * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
   91  * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
   92  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
   93  * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
   94  * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
   95  * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
   96  * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
   97  * 
   98  * The views and conclusions contained in the software and documentation
   99  * are those of the authors and should not be interpreted as representing
  100  * official policies, either expressed or implied, of the US Naval
  101  * Research Laboratory (NRL).
  102  */
  103 
  104 #include <sys/param.h>
  105 #include <sys/systm.h>
  106 #include <sys/proc.h>
  107 #include <sys/mbuf.h>
  108 #include <sys/socket.h>
  109 #include <sys/socketvar.h>
  110 #include <sys/domain.h>
  111 #include <sys/protosw.h>
  112 #include <sys/ioctl.h>
  113 #include <sys/kernel.h>
  114 #include <sys/queue.h>
  115 #include <sys/pool.h>
  116 
  117 #include <net/if.h>
  118 #include <net/route.h>
  119 #include <net/raw_cb.h>
  120 
  121 #include <netinet/in.h>
  122 #include <netinet/in_var.h>
  123 
  124 #ifdef IPSEC
  125 #include <netinet/ip_ipsp.h>
  126 
  127 extern struct ifnet encif;
  128 struct ifaddr   *encap_findgwifa(struct sockaddr *);
  129 #endif
  130 
  131 #define SA(p) ((struct sockaddr *)(p))
  132 
  133 struct  route_cb           route_cb;
  134 struct  rtstat             rtstat;
  135 struct  radix_node_head ***rt_tables;
  136 u_int8_t                   af2rtafidx[AF_MAX+1];
  137 u_int8_t                   rtafidx_max;
  138 u_int                      rtbl_id_max = 0;
  139 
  140 int                     rttrash;        /* routes not in table but not freed */
  141 
  142 struct pool             rtentry_pool;   /* pool for rtentry structures */
  143 struct pool             rttimer_pool;   /* pool for rttimer structures */
  144 
  145 int     rtable_init(struct radix_node_head ***);
  146 int     okaytoclone(u_int, int);
  147 int     rtdeletemsg(struct rtentry *, u_int);
  148 int     rtflushclone1(struct radix_node *, void *);
  149 void    rtflushclone(struct radix_node_head *, struct rtentry *);
  150 int     rt_if_remove_rtdelete(struct radix_node *, void *);
  151 
  152 #define LABELID_MAX     50000
  153 
  154 struct rt_label {
  155         TAILQ_ENTRY(rt_label)   rtl_entry;
  156         char                    rtl_name[RTLABEL_LEN];
  157         u_int16_t               rtl_id;
  158         int                     rtl_ref;
  159 };
  160 
  161 TAILQ_HEAD(rt_labels, rt_label) rt_labels = TAILQ_HEAD_INITIALIZER(rt_labels);
  162 
  163 #ifdef IPSEC
  164 struct ifaddr *
  165 encap_findgwifa(struct sockaddr *gw)
  166 {
  167         return (TAILQ_FIRST(&encif.if_addrlist));
  168 }
  169 #endif
  170 
  171 int
  172 rtable_init(struct radix_node_head ***table)
  173 {
  174         void            **p;
  175         struct domain    *dom;
  176 
  177         if ((p = malloc(sizeof(void *) * (rtafidx_max + 1), M_RTABLE,
  178             M_NOWAIT)) == NULL)
  179                 return (-1);
  180         bzero(p, sizeof(void *) * (rtafidx_max + 1));
  181 
  182         /* 2nd pass: attach */
  183         for (dom = domains; dom != NULL; dom = dom->dom_next)
  184                 if (dom->dom_rtattach)
  185                         dom->dom_rtattach(&p[af2rtafidx[dom->dom_family]],
  186                             dom->dom_rtoffset);
  187 
  188         *table = (struct radix_node_head **)p;
  189         return (0);
  190 }
  191 
  192 void
  193 route_init()
  194 {
  195         struct domain    *dom;
  196 
  197         pool_init(&rtentry_pool, sizeof(struct rtentry), 0, 0, 0, "rtentpl",
  198             NULL);
  199         rn_init();      /* initialize all zeroes, all ones, mask table */
  200 
  201         bzero(af2rtafidx, sizeof(af2rtafidx));
  202         rtafidx_max = 1;        /* must have NULL at index 0, so start at 1 */
  203 
  204         /* find out how many tables to allocate */
  205         for (dom = domains; dom != NULL; dom = dom->dom_next)
  206                 if (dom->dom_rtattach)
  207                         af2rtafidx[dom->dom_family] = rtafidx_max++;
  208 
  209         if (rtable_add(0) == -1)
  210                 panic("route_init rtable_add");
  211 }
  212 
  213 int
  214 rtable_add(u_int id)    /* must be called at splsoftnet */
  215 {
  216         void    *p;
  217 
  218         if (id > RT_TABLEID_MAX)
  219                 return (-1);
  220 
  221         if (id == 0 || id > rtbl_id_max) {
  222                 size_t  newlen = sizeof(void *) * (id+1);
  223 
  224                 if ((p = malloc(newlen, M_RTABLE, M_NOWAIT)) == NULL)
  225                         return (-1);
  226                 bzero(p, newlen);
  227                 if (id > 0) {
  228                         bcopy(rt_tables, p, sizeof(void *) * (rtbl_id_max+1));
  229                         free(rt_tables, M_RTABLE);
  230                 }
  231                 rt_tables = p;
  232                 rtbl_id_max = id;
  233         }
  234 
  235         if (rt_tables[id] != NULL)      /* already exists */
  236                 return (-1);
  237 
  238         return (rtable_init(&rt_tables[id]));
  239 }
  240 
  241 int
  242 rtable_exists(u_int id) /* verify table with that ID exists */
  243 {
  244         if (id > RT_TABLEID_MAX)
  245                 return (0);
  246 
  247         if (id > rtbl_id_max)
  248                 return (0);
  249 
  250         if (rt_tables[id] == NULL)      /* should not happen */
  251                 return (0);
  252 
  253         return (1);
  254 }
  255 
  256 #include "pf.h"
  257 #if NPF > 0
  258 void
  259 rtalloc_noclone(struct route *ro, int howstrict)
  260 {
  261         if (ro->ro_rt && ro->ro_rt->rt_ifp && (ro->ro_rt->rt_flags & RTF_UP))
  262                 return;         /* XXX */
  263         ro->ro_rt = rtalloc2(&ro->ro_dst, 1, howstrict);
  264 }
  265 
  266 int
  267 okaytoclone(u_int flags, int howstrict)
  268 {
  269         if (howstrict == ALL_CLONING)
  270                 return (1);
  271         if (howstrict == ONNET_CLONING && !(flags & RTF_GATEWAY))
  272                 return (1);
  273         return (0);
  274 }
  275 
  276 struct rtentry *
  277 rtalloc2(struct sockaddr *dst, int report, int howstrict)
  278 {
  279         struct radix_node_head  *rnh;
  280         struct rtentry          *rt;
  281         struct radix_node       *rn;
  282         struct rtentry          *newrt = 0;
  283         struct rt_addrinfo       info;
  284         int                      s = splnet(), err = 0, msgtype = RTM_MISS;
  285 
  286         rnh = rt_gettable(dst->sa_family, 0);
  287         if (rnh && (rn = rnh->rnh_matchaddr((caddr_t)dst, rnh)) &&
  288             ((rn->rn_flags & RNF_ROOT) == 0)) {
  289                 newrt = rt = (struct rtentry *)rn;
  290                 if (report && (rt->rt_flags & RTF_CLONING) &&
  291                     okaytoclone(rt->rt_flags, howstrict)) {
  292                         err = rtrequest(RTM_RESOLVE, dst, SA(0), SA(0), 0,
  293                             &newrt, 0);
  294                         if (err) {
  295                                 newrt = rt;
  296                                 rt->rt_refcnt++;
  297                                 goto miss;
  298                         }
  299                         if ((rt = newrt) && (rt->rt_flags & RTF_XRESOLVE)) {
  300                                 msgtype = RTM_RESOLVE;
  301                                 goto miss;
  302                         }
  303                 } else
  304                         rt->rt_refcnt++;
  305         } else {
  306                 rtstat.rts_unreach++;
  307 miss:
  308                 if (report) {
  309                         bzero((caddr_t)&info, sizeof(info));
  310                         info.rti_info[RTAX_DST] = dst;
  311                         rt_missmsg(msgtype, &info, 0, NULL, err, 0);
  312                 }
  313         }
  314         splx(s);
  315         return (newrt);
  316 }
  317 #endif /* NPF > 0 */
  318 
  319 /*
  320  * Packet routing routines.
  321  */
  322 void
  323 rtalloc(struct route *ro)
  324 {
  325         if (ro->ro_rt && ro->ro_rt->rt_ifp && (ro->ro_rt->rt_flags & RTF_UP))
  326                 return;                          /* XXX */
  327         ro->ro_rt = rtalloc1(&ro->ro_dst, 1, 0);
  328 }
  329 
  330 struct rtentry *
  331 rtalloc1(struct sockaddr *dst, int report, u_int tableid)
  332 {
  333         struct radix_node_head  *rnh;
  334         struct rtentry          *rt;
  335         struct radix_node       *rn;
  336         struct rtentry          *newrt = 0;
  337         struct rt_addrinfo       info;
  338         int                      s = splsoftnet(), err = 0, msgtype = RTM_MISS;
  339 
  340         rnh = rt_gettable(dst->sa_family, tableid);
  341         if (rnh && (rn = rnh->rnh_matchaddr((caddr_t)dst, rnh)) &&
  342             ((rn->rn_flags & RNF_ROOT) == 0)) {
  343                 newrt = rt = (struct rtentry *)rn;
  344                 if (report && (rt->rt_flags & RTF_CLONING)) {
  345                         err = rtrequest(RTM_RESOLVE, dst, SA(NULL),
  346                             SA(NULL), 0, &newrt, tableid);
  347                         if (err) {
  348                                 newrt = rt;
  349                                 rt->rt_refcnt++;
  350                                 goto miss;
  351                         }
  352                         if ((rt = newrt) && (rt->rt_flags & RTF_XRESOLVE)) {
  353                                 msgtype = RTM_RESOLVE;
  354                                 goto miss;
  355                         }
  356                         /* Inform listeners of the new route */
  357                         bzero(&info, sizeof(info));
  358                         info.rti_info[RTAX_DST] = rt_key(rt);
  359                         info.rti_info[RTAX_NETMASK] = rt_mask(rt);
  360                         info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
  361                         if (rt->rt_ifp != NULL) {
  362                                 info.rti_info[RTAX_IFP] =
  363                                     TAILQ_FIRST(&rt->rt_ifp->if_addrlist)->ifa_addr;
  364                                 info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr;
  365                         }
  366                         rt_missmsg(RTM_ADD, &info, rt->rt_flags,
  367                             rt->rt_ifp, 0, tableid);
  368                 } else
  369                         rt->rt_refcnt++;
  370         } else {
  371                 if (dst->sa_family != PF_KEY)
  372                         rtstat.rts_unreach++;
  373         /*
  374          * IP encapsulation does lots of lookups where we don't need nor want
  375          * the RTM_MISSes that would be generated.  It causes RTM_MISS storms
  376          * sent upward breaking user-level routing queries.
  377          */
  378 miss:
  379                 if (report && dst->sa_family != PF_KEY) {
  380                         bzero((caddr_t)&info, sizeof(info));
  381                         info.rti_info[RTAX_DST] = dst;
  382                         rt_missmsg(msgtype, &info, 0, NULL, err, tableid);
  383                 }
  384         }
  385         splx(s);
  386         return (newrt);
  387 }
  388 
  389 void
  390 rtfree(struct rtentry *rt)
  391 {
  392         struct ifaddr   *ifa;
  393 
  394         if (rt == NULL)
  395                 panic("rtfree");
  396 
  397         rt->rt_refcnt--;
  398 
  399         if (rt->rt_refcnt <= 0 && (rt->rt_flags & RTF_UP) == 0) {
  400                 if (rt->rt_nodes->rn_flags & (RNF_ACTIVE | RNF_ROOT))
  401                         panic("rtfree 2");
  402                 rttrash--;
  403                 if (rt->rt_refcnt < 0) {
  404                         printf("rtfree: %p not freed (neg refs)\n", rt);
  405                         return;
  406                 }
  407                 rt_timer_remove_all(rt);
  408                 ifa = rt->rt_ifa;
  409                 if (ifa)
  410                         IFAFREE(ifa);
  411                 rtlabel_unref(rt->rt_labelid);
  412                 Free(rt_key(rt));
  413                 pool_put(&rtentry_pool, rt);
  414         }
  415 }
  416 
  417 void
  418 ifafree(struct ifaddr *ifa)
  419 {
  420         if (ifa == NULL)
  421                 panic("ifafree");
  422         if (ifa->ifa_refcnt == 0)
  423                 free(ifa, M_IFADDR);
  424         else
  425                 ifa->ifa_refcnt--;
  426 }
  427 
  428 /*
  429  * Force a routing table entry to the specified
  430  * destination to go through the given gateway.
  431  * Normally called as a result of a routing redirect
  432  * message from the network layer.
  433  *
  434  * N.B.: must be called at splsoftnet
  435  */
  436 void
  437 rtredirect(struct sockaddr *dst, struct sockaddr *gateway,
  438     struct sockaddr *netmask, int flags, struct sockaddr *src,
  439     struct rtentry **rtp)
  440 {
  441         struct rtentry          *rt;
  442         int                      error = 0;
  443         u_int32_t               *stat = NULL;
  444         struct rt_addrinfo       info;
  445         struct ifaddr           *ifa;
  446         struct ifnet            *ifp = NULL;
  447 
  448         splassert(IPL_SOFTNET);
  449 
  450         /* verify the gateway is directly reachable */
  451         if ((ifa = ifa_ifwithnet(gateway)) == NULL) {
  452                 error = ENETUNREACH;
  453                 goto out;
  454         }
  455         ifp = ifa->ifa_ifp;
  456         rt = rtalloc1(dst, 0, 0);
  457         /*
  458          * If the redirect isn't from our current router for this dst,
  459          * it's either old or wrong.  If it redirects us to ourselves,
  460          * we have a routing loop, perhaps as a result of an interface
  461          * going down recently.
  462          */
  463 #define equal(a1, a2) \
  464         ((a1)->sa_len == (a2)->sa_len && \
  465          bcmp((caddr_t)(a1), (caddr_t)(a2), (a1)->sa_len) == 0)
  466         if (!(flags & RTF_DONE) && rt &&
  467              (!equal(src, rt->rt_gateway) || rt->rt_ifa != ifa))
  468                 error = EINVAL;
  469         else if (ifa_ifwithaddr(gateway) != NULL)
  470                 error = EHOSTUNREACH;
  471         if (error)
  472                 goto done;
  473         /*
  474          * Create a new entry if we just got back a wildcard entry
  475          * or the lookup failed.  This is necessary for hosts
  476          * which use routing redirects generated by smart gateways
  477          * to dynamically build the routing tables.
  478          */
  479         if ((rt == NULL) || (rt_mask(rt) && rt_mask(rt)->sa_len < 2))
  480                 goto create;
  481         /*
  482          * Don't listen to the redirect if it's
  483          * for a route to an interface. 
  484          */
  485         if (rt->rt_flags & RTF_GATEWAY) {
  486                 if (((rt->rt_flags & RTF_HOST) == 0) && (flags & RTF_HOST)) {
  487                         /*
  488                          * Changing from route to net => route to host.
  489                          * Create new route, rather than smashing route to net.
  490                          */
  491 create:
  492                         if (rt)
  493                                 rtfree(rt);
  494                         flags |= RTF_GATEWAY | RTF_DYNAMIC;
  495                         bzero(&info, sizeof(info));
  496                         info.rti_info[RTAX_DST] = dst;
  497                         info.rti_info[RTAX_GATEWAY] = gateway;
  498                         info.rti_info[RTAX_NETMASK] = netmask;
  499                         info.rti_ifa = ifa;
  500                         info.rti_flags = flags;
  501                         rt = NULL;
  502                         error = rtrequest1(RTM_ADD, &info, &rt, 0);
  503                         if (rt != NULL)
  504                                 flags = rt->rt_flags;
  505                         stat = &rtstat.rts_dynamic;
  506                 } else {
  507                         /*
  508                          * Smash the current notion of the gateway to
  509                          * this destination.  Should check about netmask!!!
  510                          */
  511                         rt->rt_flags |= RTF_MODIFIED;
  512                         flags |= RTF_MODIFIED;
  513                         stat = &rtstat.rts_newgateway;
  514                         rt_setgate(rt, rt_key(rt), gateway, 0);
  515                 }
  516         } else
  517                 error = EHOSTUNREACH;
  518 done:
  519         if (rt) {
  520                 if (rtp && !error)
  521                         *rtp = rt;
  522                 else
  523                         rtfree(rt);
  524         }
  525 out:
  526         if (error)
  527                 rtstat.rts_badredirect++;
  528         else if (stat != NULL)
  529                 (*stat)++;
  530         bzero((caddr_t)&info, sizeof(info));
  531         info.rti_info[RTAX_DST] = dst;
  532         info.rti_info[RTAX_GATEWAY] = gateway;
  533         info.rti_info[RTAX_NETMASK] = netmask;
  534         info.rti_info[RTAX_AUTHOR] = src;
  535         rt_missmsg(RTM_REDIRECT, &info, flags, ifp, error, 0);
  536 }
  537 
  538 /*
  539  * Delete a route and generate a message
  540  */
  541 int
  542 rtdeletemsg(struct rtentry *rt, u_int tableid)
  543 {
  544         int                     error;
  545         struct rt_addrinfo      info;
  546         struct ifnet            *ifp;
  547 
  548         /*
  549          * Request the new route so that the entry is not actually
  550          * deleted.  That will allow the information being reported to
  551          * be accurate (and consistent with route_output()).
  552          */
  553         bzero((caddr_t)&info, sizeof(info));
  554         info.rti_info[RTAX_DST] = rt_key(rt);
  555         info.rti_info[RTAX_NETMASK] = rt_mask(rt);
  556         info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
  557         info.rti_flags = rt->rt_flags;
  558         ifp = rt->rt_ifp;
  559         error = rtrequest1(RTM_DELETE, &info, &rt, tableid);
  560 
  561         rt_missmsg(RTM_DELETE, &info, info.rti_flags, ifp, error, tableid);
  562 
  563         /* Adjust the refcount */
  564         if (error == 0 && rt->rt_refcnt <= 0) {
  565                 rt->rt_refcnt++;
  566                 rtfree(rt);
  567         }
  568         return (error);
  569 }
  570 
  571 int
  572 rtflushclone1(struct radix_node *rn, void *arg)
  573 {
  574         struct rtentry  *rt, *parent;
  575 
  576         rt = (struct rtentry *)rn;
  577         parent = (struct rtentry *)arg;
  578         if ((rt->rt_flags & RTF_CLONED) != 0 && rt->rt_parent == parent)
  579                 rtdeletemsg(rt, 0);
  580         return 0;
  581 }
  582 
  583 void
  584 rtflushclone(struct radix_node_head *rnh, struct rtentry *parent)
  585 {
  586 
  587 #ifdef DIAGNOSTIC
  588         if (!parent || (parent->rt_flags & RTF_CLONING) == 0)
  589                 panic("rtflushclone: called with a non-cloning route");
  590         if (!rnh->rnh_walktree)
  591                 panic("rtflushclone: no rnh_walktree");
  592 #endif
  593         rnh->rnh_walktree(rnh, rtflushclone1, (void *)parent);
  594 }
  595 
  596 int
  597 rtioctl(u_long req, caddr_t data, struct proc *p)
  598 {
  599         return (EOPNOTSUPP);
  600 }
  601 
  602 struct ifaddr *
  603 ifa_ifwithroute(int flags, struct sockaddr *dst, struct sockaddr *gateway)
  604 {
  605         struct ifaddr   *ifa;
  606 
  607 #ifdef IPSEC
  608         /*
  609          * If the destination is a PF_KEY address, we'll look
  610          * for the existence of a encap interface number or address
  611          * in the options list of the gateway. By default, we'll return
  612          * enc0.
  613          */
  614         if (dst && (dst->sa_family == PF_KEY))
  615                 return (encap_findgwifa(gateway));
  616 #endif
  617 
  618         if ((flags & RTF_GATEWAY) == 0) {
  619                 /*
  620                  * If we are adding a route to an interface,
  621                  * and the interface is a pt to pt link
  622                  * we should search for the destination
  623                  * as our clue to the interface.  Otherwise
  624                  * we can use the local address.
  625                  */
  626                 ifa = NULL;
  627                 if (flags & RTF_HOST)
  628                         ifa = ifa_ifwithdstaddr(dst);
  629                 if (ifa == NULL)
  630                         ifa = ifa_ifwithaddr(gateway);
  631         } else {
  632                 /*
  633                  * If we are adding a route to a remote net
  634                  * or host, the gateway may still be on the
  635                  * other end of a pt to pt link.
  636                  */
  637                 ifa = ifa_ifwithdstaddr(gateway);
  638         }
  639         if (ifa == NULL)
  640                 ifa = ifa_ifwithnet(gateway);
  641         if (ifa == NULL) {
  642                 struct rtentry  *rt = rtalloc1(gateway, 0, 0);
  643                 if (rt == NULL)
  644                         return (NULL);
  645                 rt->rt_refcnt--;
  646                 /* The gateway must be local if the same address family. */
  647                 if ((rt->rt_flags & RTF_GATEWAY) &&
  648                     rt_key(rt)->sa_family == dst->sa_family)
  649                         return (0);
  650                 if ((ifa = rt->rt_ifa) == NULL)
  651                         return (NULL);
  652         }
  653         if (ifa->ifa_addr->sa_family != dst->sa_family) {
  654                 struct ifaddr   *oifa = ifa;
  655                 ifa = ifaof_ifpforaddr(dst, ifa->ifa_ifp);
  656                 if (ifa == NULL)
  657                         ifa = oifa;
  658         }
  659         return (ifa);
  660 }
  661 
  662 #define ROUNDUP(a) (a>0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
  663 
  664 int
  665 rtrequest(int req, struct sockaddr *dst, struct sockaddr *gateway,
  666     struct sockaddr *netmask, int flags, struct rtentry **ret_nrt,
  667     u_int tableid)
  668 {
  669         struct rt_addrinfo      info;
  670 
  671         bzero(&info, sizeof(info));
  672         info.rti_flags = flags;
  673         info.rti_info[RTAX_DST] = dst;
  674         info.rti_info[RTAX_GATEWAY] = gateway;
  675         info.rti_info[RTAX_NETMASK] = netmask;
  676         return (rtrequest1(req, &info, ret_nrt, tableid));
  677 }
  678 
  679 int
  680 rt_getifa(struct rt_addrinfo *info)
  681 {
  682         struct ifaddr   *ifa;
  683         int              error = 0;
  684 
  685         /*
  686          * ifp may be specified by sockaddr_dl when protocol address
  687          * is ambiguous
  688          */
  689         if (info->rti_ifp == NULL && info->rti_info[RTAX_IFP] != NULL
  690             && info->rti_info[RTAX_IFP]->sa_family == AF_LINK &&
  691             (ifa = ifa_ifwithnet((struct sockaddr *)info->rti_info[RTAX_IFP]))
  692             != NULL)
  693                 info->rti_ifp = ifa->ifa_ifp;
  694 
  695         if (info->rti_ifa == NULL && info->rti_info[RTAX_IFA] != NULL)
  696                 info->rti_ifa = ifa_ifwithaddr(info->rti_info[RTAX_IFA]);
  697 
  698         if (info->rti_ifa == NULL) {
  699                 struct sockaddr *sa;
  700 
  701                 if ((sa = info->rti_info[RTAX_IFA]) == NULL)
  702                         if ((sa = info->rti_info[RTAX_GATEWAY]) == NULL)
  703                                 sa = info->rti_info[RTAX_DST];
  704 
  705                 if (sa != NULL && info->rti_ifp != NULL)
  706                         info->rti_ifa = ifaof_ifpforaddr(sa, info->rti_ifp);
  707                 else if (info->rti_info[RTAX_DST] != NULL &&
  708                     info->rti_info[RTAX_GATEWAY] != NULL)
  709                         info->rti_ifa = ifa_ifwithroute(info->rti_flags,
  710                             info->rti_info[RTAX_DST],
  711                             info->rti_info[RTAX_GATEWAY]);
  712                 else if (sa != NULL)
  713                         info->rti_ifa = ifa_ifwithroute(info->rti_flags,
  714                             sa, sa);
  715         }
  716         if ((ifa = info->rti_ifa) != NULL) {
  717                 if (info->rti_ifp == NULL)
  718                         info->rti_ifp = ifa->ifa_ifp;
  719         } else
  720                 error = ENETUNREACH;
  721         return (error);
  722 }
  723 
  724 int
  725 rtrequest1(int req, struct rt_addrinfo *info, struct rtentry **ret_nrt,
  726     u_int tableid)
  727 {
  728         int                      s = splsoftnet(); int error = 0;
  729         struct rtentry          *rt, *crt;
  730         struct radix_node       *rn;
  731         struct radix_node_head  *rnh;
  732         struct ifaddr           *ifa;
  733         struct sockaddr         *ndst;
  734         struct sockaddr_rtlabel *sa_rl;
  735 #define senderr(x) { error = x ; goto bad; }
  736 
  737         if ((rnh = rt_gettable(info->rti_info[RTAX_DST]->sa_family, tableid)) ==
  738             NULL)
  739                 senderr(EAFNOSUPPORT);
  740         if (info->rti_flags & RTF_HOST)
  741                 info->rti_info[RTAX_NETMASK] = NULL;
  742         switch (req) {
  743         case RTM_DELETE:
  744                 if ((rn = rnh->rnh_lookup(info->rti_info[RTAX_DST],
  745                     info->rti_info[RTAX_NETMASK], rnh)) == NULL)
  746                         senderr(ESRCH);
  747                 rt = (struct rtentry *)rn;
  748 #ifndef SMALL_KERNEL
  749                 /*
  750                  * if we got multipath routes, we require users to specify
  751                  * a matching RTAX_GATEWAY.
  752                  */
  753                 if (rn_mpath_capable(rnh)) {
  754                         rt = rt_mpath_matchgate(rt,
  755                             info->rti_info[RTAX_GATEWAY]);
  756                         rn = (struct radix_node *)rt;
  757                         if (!rt)
  758                                 senderr(ESRCH);
  759                 }
  760 #endif
  761                 if ((rn = rnh->rnh_deladdr(info->rti_info[RTAX_DST],
  762                     info->rti_info[RTAX_NETMASK], rnh, rn)) == NULL)
  763                         senderr(ESRCH);
  764                 rt = (struct rtentry *)rn;
  765 
  766                 /* clean up any cloned children */
  767                 if ((rt->rt_flags & RTF_CLONING) != 0)
  768                         rtflushclone(rnh, rt);
  769 
  770                 if (rn->rn_flags & (RNF_ACTIVE | RNF_ROOT))
  771                         panic ("rtrequest delete");
  772 
  773                 if (rt->rt_gwroute) {
  774                         rt = rt->rt_gwroute; RTFREE(rt);
  775                         (rt = (struct rtentry *)rn)->rt_gwroute = NULL;
  776                 }
  777 
  778                 if (rt->rt_parent) {
  779                         rt->rt_parent->rt_refcnt--;
  780                         rt->rt_parent = NULL;
  781                 }
  782 
  783 #ifndef SMALL_KERNEL
  784                 if (rn_mpath_capable(rnh)) {
  785                         if ((rn = rnh->rnh_lookup(info->rti_info[RTAX_DST],
  786                             info->rti_info[RTAX_NETMASK], rnh)) != NULL &&
  787                             rn_mpath_next(rn) == NULL)
  788                                 ((struct rtentry *)rn)->rt_flags &= ~RTF_MPATH;
  789                 }
  790 #endif
  791 
  792                 rt->rt_flags &= ~RTF_UP;
  793                 if ((ifa = rt->rt_ifa) && ifa->ifa_rtrequest)
  794                         ifa->ifa_rtrequest(RTM_DELETE, rt, info);
  795                 rttrash++;
  796 
  797                 if (ret_nrt)
  798                         *ret_nrt = rt;
  799                 else if (rt->rt_refcnt <= 0) {
  800                         rt->rt_refcnt++;
  801                         rtfree(rt);
  802                 }
  803                 break;
  804 
  805         case RTM_RESOLVE:
  806                 if (ret_nrt == NULL || (rt = *ret_nrt) == NULL)
  807                         senderr(EINVAL);
  808                 if ((rt->rt_flags & RTF_CLONING) == 0)
  809                         senderr(EINVAL);
  810                 ifa = rt->rt_ifa;
  811                 info->rti_flags = rt->rt_flags & ~(RTF_CLONING | RTF_STATIC);
  812                 info->rti_flags |= RTF_CLONED;
  813                 info->rti_info[RTAX_GATEWAY] = rt->rt_gateway;
  814                 if ((info->rti_info[RTAX_NETMASK] = rt->rt_genmask) == NULL)
  815                         info->rti_flags |= RTF_HOST;
  816                 goto makeroute;
  817 
  818         case RTM_ADD:
  819                 if (info->rti_ifa == 0 && (error = rt_getifa(info)))
  820                         senderr(error);
  821                 ifa = info->rti_ifa;
  822 makeroute:
  823                 rt = pool_get(&rtentry_pool, PR_NOWAIT);
  824                 if (rt == NULL)
  825                         senderr(ENOBUFS);
  826                 Bzero(rt, sizeof(*rt));
  827                 rt->rt_flags = RTF_UP | info->rti_flags;
  828                 LIST_INIT(&rt->rt_timer);
  829                 if (rt_setgate(rt, info->rti_info[RTAX_DST],
  830                     info->rti_info[RTAX_GATEWAY], tableid)) {
  831                         pool_put(&rtentry_pool, rt);
  832                         senderr(ENOBUFS);
  833                 }
  834                 ndst = rt_key(rt);
  835                 if (info->rti_info[RTAX_NETMASK] != NULL) {
  836                         rt_maskedcopy(info->rti_info[RTAX_DST], ndst,
  837                             info->rti_info[RTAX_NETMASK]);
  838                 } else
  839                         Bcopy(info->rti_info[RTAX_DST], ndst,
  840                             info->rti_info[RTAX_DST]->sa_len);
  841 #ifndef SMALL_KERNEL
  842                 /* do not permit exactly the same dst/mask/gw pair */
  843                 if (rn_mpath_capable(rnh) &&
  844                     rt_mpath_conflict(rnh, rt, info->rti_info[RTAX_NETMASK],
  845                     info->rti_flags & RTF_MPATH)) {
  846                         if (rt->rt_gwroute)
  847                                 rtfree(rt->rt_gwroute);
  848                         Free(rt_key(rt));
  849                         pool_put(&rtentry_pool, rt);
  850                         senderr(EEXIST);
  851                 }
  852 #endif
  853 
  854                 if (info->rti_info[RTAX_LABEL] != NULL) {
  855                         sa_rl = (struct sockaddr_rtlabel *)
  856                             info->rti_info[RTAX_LABEL];
  857                         rt->rt_labelid = rtlabel_name2id(sa_rl->sr_label);
  858                 }
  859 
  860                 ifa->ifa_refcnt++;
  861                 rt->rt_ifa = ifa;
  862                 rt->rt_ifp = ifa->ifa_ifp;
  863                 if (req == RTM_RESOLVE) {
  864                         /*
  865                          * Copy both metrics and a back pointer to the cloned
  866                          * route's parent.
  867                          */
  868                         rt->rt_rmx = (*ret_nrt)->rt_rmx; /* copy metrics */
  869                         rt->rt_parent = *ret_nrt;        /* Back ptr. to parent. */
  870                         rt->rt_parent->rt_refcnt++;
  871                 }
  872                 rn = rnh->rnh_addaddr((caddr_t)ndst,
  873                     (caddr_t)info->rti_info[RTAX_NETMASK], rnh, rt->rt_nodes);
  874                 if (rn == NULL && (crt = rtalloc1(ndst, 0, tableid)) != NULL) {
  875                         /* overwrite cloned route */
  876                         if ((crt->rt_flags & RTF_CLONED) != 0) {
  877                                 rtdeletemsg(crt, tableid);
  878                                 rn = rnh->rnh_addaddr((caddr_t)ndst,
  879                                     (caddr_t)info->rti_info[RTAX_NETMASK],
  880                                     rnh, rt->rt_nodes);
  881                         }
  882                         RTFREE(crt);
  883                 }
  884                 if (rn == 0) {
  885                         IFAFREE(ifa);
  886                         if ((rt->rt_flags & RTF_CLONED) != 0 && rt->rt_parent)
  887                                 rtfree(rt->rt_parent);
  888                         if (rt->rt_gwroute)
  889                                 rtfree(rt->rt_gwroute);
  890                         Free(rt_key(rt));
  891                         pool_put(&rtentry_pool, rt);
  892                         senderr(EEXIST);
  893                 }
  894 
  895 #ifndef SMALL_KERNEL
  896                 if (rn_mpath_capable(rnh) &&
  897                     (rn = rnh->rnh_lookup(info->rti_info[RTAX_DST],
  898                     info->rti_info[RTAX_NETMASK], rnh)) != NULL) {
  899                         if (rn_mpath_next(rn) == NULL)
  900                                 ((struct rtentry *)rn)->rt_flags &= ~RTF_MPATH;
  901                         else
  902                                 ((struct rtentry *)rn)->rt_flags |= RTF_MPATH;
  903                 }
  904 #endif
  905 
  906                 if (ifa->ifa_rtrequest)
  907                         ifa->ifa_rtrequest(req, rt, info);
  908                 if (ret_nrt) {
  909                         *ret_nrt = rt;
  910                         rt->rt_refcnt++;
  911                 }
  912                 if ((rt->rt_flags & RTF_CLONING) != 0) {
  913                         /* clean up any cloned children */
  914                         rtflushclone(rnh, rt);
  915                 }
  916 
  917                 if_group_routechange(info->rti_info[RTAX_DST],
  918                         info->rti_info[RTAX_NETMASK]);
  919                 break;
  920         }
  921 bad:
  922         splx(s);
  923         return (error);
  924 }
  925 
  926 int
  927 rt_setgate(struct rtentry *rt0, struct sockaddr *dst, struct sockaddr *gate,
  928     u_int tableid)
  929 {
  930         caddr_t new, old;
  931         int     dlen = ROUNDUP(dst->sa_len), glen = ROUNDUP(gate->sa_len);
  932         struct rtentry  *rt = rt0;
  933 
  934         if (rt->rt_gateway == NULL || glen > ROUNDUP(rt->rt_gateway->sa_len)) {
  935                 old = (caddr_t)rt_key(rt);
  936                 R_Malloc(new, caddr_t, dlen + glen);
  937                 if (new == NULL)
  938                         return 1;
  939                 rt->rt_nodes->rn_key = new;
  940         } else {
  941                 new = rt->rt_nodes->rn_key;
  942                 old = NULL;
  943         }
  944         Bcopy(gate, (rt->rt_gateway = (struct sockaddr *)(new + dlen)), glen);
  945         if (old) {
  946                 Bcopy(dst, new, dlen);
  947                 Free(old);
  948         }
  949         if (rt->rt_gwroute != NULL) {
  950                 rt = rt->rt_gwroute;
  951                 RTFREE(rt);
  952                 rt = rt0;
  953                 rt->rt_gwroute = NULL;
  954         }
  955         if (rt->rt_flags & RTF_GATEWAY) {
  956                 rt->rt_gwroute = rtalloc1(gate, 1, tableid);
  957                 /*
  958                  * If we switched gateways, grab the MTU from the new
  959                  * gateway route if the current MTU is 0 or greater
  960                  * than the MTU of gateway.
  961                  * Note that, if the MTU of gateway is 0, we will reset the
  962                  * MTU of the route to run PMTUD again from scratch. XXX
  963                  */
  964                 if (rt->rt_gwroute && !(rt->rt_rmx.rmx_locks & RTV_MTU) &&
  965                     rt->rt_rmx.rmx_mtu &&
  966                     rt->rt_rmx.rmx_mtu > rt->rt_gwroute->rt_rmx.rmx_mtu) {
  967                         rt->rt_rmx.rmx_mtu = rt->rt_gwroute->rt_rmx.rmx_mtu;
  968                 }
  969         }
  970         return (0);
  971 }
  972 
  973 void
  974 rt_maskedcopy(struct sockaddr *src, struct sockaddr *dst,
  975     struct sockaddr *netmask)
  976 {
  977         u_char  *cp1 = (u_char *)src;
  978         u_char  *cp2 = (u_char *)dst;
  979         u_char  *cp3 = (u_char *)netmask;
  980         u_char  *cplim = cp2 + *cp3;
  981         u_char  *cplim2 = cp2 + *cp1;
  982 
  983         *cp2++ = *cp1++; *cp2++ = *cp1++; /* copies sa_len & sa_family */
  984         cp3 += 2;
  985         if (cplim > cplim2)
  986                 cplim = cplim2;
  987         while (cp2 < cplim)
  988                 *cp2++ = *cp1++ & *cp3++;
  989         if (cp2 < cplim2)
  990                 bzero((caddr_t)cp2, (unsigned)(cplim2 - cp2));
  991 }
  992 
  993 /*
  994  * Set up a routing table entry, normally
  995  * for an interface.
  996  */
  997 int
  998 rtinit(struct ifaddr *ifa, int cmd, int flags)
  999 {
 1000         struct rtentry          *rt;
 1001         struct sockaddr         *dst, *deldst;
 1002         struct mbuf             *m = NULL;
 1003         struct rtentry          *nrt = NULL;
 1004         int                      error;
 1005         struct rt_addrinfo       info;
 1006         struct sockaddr_rtlabel  sa_rl;
 1007         const char              *label;
 1008 
 1009         dst = flags & RTF_HOST ? ifa->ifa_dstaddr : ifa->ifa_addr;
 1010         if (cmd == RTM_DELETE) {
 1011                 if ((flags & RTF_HOST) == 0 && ifa->ifa_netmask) {
 1012                         m = m_get(M_DONTWAIT, MT_SONAME);
 1013                         if (m == NULL)
 1014                                 return (ENOBUFS);
 1015                         deldst = mtod(m, struct sockaddr *);
 1016                         rt_maskedcopy(dst, deldst, ifa->ifa_netmask);
 1017                         dst = deldst;
 1018                 }
 1019                 if ((rt = rtalloc1(dst, 0, 0)) != NULL) {
 1020                         rt->rt_refcnt--;
 1021                         if (rt->rt_ifa != ifa) {
 1022                                 if (m != NULL)
 1023                                         (void) m_free(m);
 1024                                 return (flags & RTF_HOST ? EHOSTUNREACH
 1025                                                         : ENETUNREACH);
 1026                         }
 1027                 }
 1028         }
 1029         bzero(&info, sizeof(info));
 1030         info.rti_ifa = ifa;
 1031         info.rti_flags = flags | ifa->ifa_flags;
 1032         info.rti_info[RTAX_DST] = dst;
 1033         info.rti_info[RTAX_GATEWAY] = ifa->ifa_addr;
 1034         if (ifa->ifa_ifp->if_rtlabelid) {
 1035                 label = rtlabel_id2name(ifa->ifa_ifp->if_rtlabelid);
 1036                 bzero(&sa_rl, sizeof(sa_rl));
 1037                 sa_rl.sr_len = sizeof(sa_rl);
 1038                 sa_rl.sr_family = AF_UNSPEC;
 1039                 strlcpy(sa_rl.sr_label, label, sizeof(sa_rl.sr_label));
 1040                 info.rti_info[RTAX_LABEL] = (struct sockaddr *)&sa_rl;
 1041         }
 1042 
 1043         /*
 1044          * XXX here, it seems that we are assuming that ifa_netmask is NULL
 1045          * for RTF_HOST.  bsdi4 passes NULL explicitly (via intermediate
 1046          * variable) when RTF_HOST is 1.  still not sure if i can safely
 1047          * change it to meet bsdi4 behavior.
 1048          */
 1049         info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
 1050         error = rtrequest1(cmd, &info, &nrt, 0);
 1051         if (cmd == RTM_DELETE && error == 0 && (rt = nrt) != NULL) {
 1052                 rt_newaddrmsg(cmd, ifa, error, nrt);
 1053                 if (rt->rt_refcnt <= 0) {
 1054                         rt->rt_refcnt++;
 1055                         rtfree(rt);
 1056                 }
 1057         }
 1058         if (cmd == RTM_ADD && error == 0 && (rt = nrt) != NULL) {
 1059                 rt->rt_refcnt--;
 1060                 if (rt->rt_ifa != ifa) {
 1061                         printf("rtinit: wrong ifa (%p) was (%p)\n",
 1062                             ifa, rt->rt_ifa);
 1063                         if (rt->rt_ifa->ifa_rtrequest)
 1064                                 rt->rt_ifa->ifa_rtrequest(RTM_DELETE, rt, NULL);
 1065                         IFAFREE(rt->rt_ifa);
 1066                         rt->rt_ifa = ifa;
 1067                         rt->rt_ifp = ifa->ifa_ifp;
 1068                         ifa->ifa_refcnt++;
 1069                         if (ifa->ifa_rtrequest)
 1070                                 ifa->ifa_rtrequest(RTM_ADD, rt, NULL);
 1071                 }
 1072                 rt_newaddrmsg(cmd, ifa, error, nrt);
 1073         }
 1074         return (error);
 1075 }
 1076 
 1077 /*
 1078  * Route timer routines.  These routes allow functions to be called
 1079  * for various routes at any time.  This is useful in supporting
 1080  * path MTU discovery and redirect route deletion.
 1081  *
 1082  * This is similar to some BSDI internal functions, but it provides
 1083  * for multiple queues for efficiency's sake...
 1084  */
 1085 
 1086 LIST_HEAD(, rttimer_queue)      rttimer_queue_head;
 1087 static int                      rt_init_done = 0;
 1088 
 1089 #define RTTIMER_CALLOUT(r)      {                               \
 1090         if (r->rtt_func != NULL) {                              \
 1091                 (*r->rtt_func)(r->rtt_rt, r);                   \
 1092         } else {                                                \
 1093                 rtrequest((int) RTM_DELETE,                     \
 1094                     (struct sockaddr *)rt_key(r->rtt_rt),       \
 1095                     0, 0, 0, 0, 0);                             \
 1096         }                                                       \
 1097 }
 1098 
 1099 /* 
 1100  * Some subtle order problems with domain initialization mean that
 1101  * we cannot count on this being run from rt_init before various
 1102  * protocol initializations are done.  Therefore, we make sure
 1103  * that this is run when the first queue is added...
 1104  */
 1105 
 1106 void
 1107 rt_timer_init()
 1108 {
 1109         static struct timeout   rt_timer_timeout;
 1110 
 1111         if (rt_init_done)
 1112                 panic("rt_timer_init: already initialized");
 1113 
 1114         pool_init(&rttimer_pool, sizeof(struct rttimer), 0, 0, 0, "rttmrpl",
 1115             NULL);
 1116 
 1117         LIST_INIT(&rttimer_queue_head);
 1118         timeout_set(&rt_timer_timeout, rt_timer_timer, &rt_timer_timeout);
 1119         timeout_add(&rt_timer_timeout, hz);     /* every second */
 1120         rt_init_done = 1;
 1121 }
 1122 
 1123 struct rttimer_queue *
 1124 rt_timer_queue_create(u_int timeout)
 1125 {
 1126         struct rttimer_queue    *rtq;
 1127 
 1128         if (rt_init_done == 0)
 1129                 rt_timer_init();
 1130 
 1131         R_Malloc(rtq, struct rttimer_queue *, sizeof *rtq);
 1132         if (rtq == NULL)
 1133                 return (NULL);
 1134         Bzero(rtq, sizeof *rtq);
 1135 
 1136         rtq->rtq_timeout = timeout;
 1137         rtq->rtq_count = 0;
 1138         TAILQ_INIT(&rtq->rtq_head);
 1139         LIST_INSERT_HEAD(&rttimer_queue_head, rtq, rtq_link);
 1140 
 1141         return (rtq);
 1142 }
 1143 
 1144 void
 1145 rt_timer_queue_change(struct rttimer_queue *rtq, long timeout)
 1146 {
 1147         rtq->rtq_timeout = timeout;
 1148 }
 1149 
 1150 void
 1151 rt_timer_queue_destroy(struct rttimer_queue *rtq, int destroy)
 1152 {
 1153         struct rttimer  *r;
 1154 
 1155         while ((r = TAILQ_FIRST(&rtq->rtq_head)) != NULL) {
 1156                 LIST_REMOVE(r, rtt_link);
 1157                 TAILQ_REMOVE(&rtq->rtq_head, r, rtt_next);
 1158                 if (destroy)
 1159                         RTTIMER_CALLOUT(r);
 1160                 pool_put(&rttimer_pool, r);
 1161                 if (rtq->rtq_count > 0)
 1162                         rtq->rtq_count--;
 1163                 else
 1164                         printf("rt_timer_queue_destroy: rtq_count reached 0\n");
 1165         }
 1166 
 1167         LIST_REMOVE(rtq, rtq_link);
 1168 
 1169         /*
 1170          * Caller is responsible for freeing the rttimer_queue structure.
 1171          */
 1172 }
 1173 
 1174 unsigned long
 1175 rt_timer_count(struct rttimer_queue *rtq)
 1176 {
 1177         return (rtq->rtq_count);
 1178 }
 1179 
 1180 void
 1181 rt_timer_remove_all(struct rtentry *rt)
 1182 {
 1183         struct rttimer  *r;
 1184 
 1185         while ((r = LIST_FIRST(&rt->rt_timer)) != NULL) {
 1186                 LIST_REMOVE(r, rtt_link);
 1187                 TAILQ_REMOVE(&r->rtt_queue->rtq_head, r, rtt_next);
 1188                 if (r->rtt_queue->rtq_count > 0)
 1189                         r->rtt_queue->rtq_count--;
 1190                 else
 1191                         printf("rt_timer_remove_all: rtq_count reached 0\n");
 1192                 pool_put(&rttimer_pool, r);
 1193         }
 1194 }
 1195 
 1196 int
 1197 rt_timer_add(struct rtentry *rt, void (*func)(struct rtentry *,
 1198     struct rttimer *), struct rttimer_queue *queue)
 1199 {
 1200         struct rttimer  *r;
 1201         long             current_time;
 1202 
 1203         current_time = time_uptime;
 1204         rt->rt_rmx.rmx_expire = time_second + queue->rtq_timeout;
 1205 
 1206         /*
 1207          * If there's already a timer with this action, destroy it before
 1208          * we add a new one.
 1209          */
 1210         for (r = LIST_FIRST(&rt->rt_timer); r != NULL;
 1211              r = LIST_NEXT(r, rtt_link)) {
 1212                 if (r->rtt_func == func) {
 1213                         LIST_REMOVE(r, rtt_link);
 1214                         TAILQ_REMOVE(&r->rtt_queue->rtq_head, r, rtt_next);
 1215                         if (r->rtt_queue->rtq_count > 0)
 1216                                 r->rtt_queue->rtq_count--;
 1217                         else
 1218                                 printf("rt_timer_add: rtq_count reached 0\n");
 1219                         pool_put(&rttimer_pool, r);
 1220                         break;  /* only one per list, so we can quit... */
 1221                 }
 1222         }
 1223 
 1224         r = pool_get(&rttimer_pool, PR_NOWAIT);
 1225         if (r == NULL)
 1226                 return (ENOBUFS);
 1227         Bzero(r, sizeof(*r));
 1228 
 1229         r->rtt_rt = rt;
 1230         r->rtt_time = current_time;
 1231         r->rtt_func = func;
 1232         r->rtt_queue = queue;
 1233         LIST_INSERT_HEAD(&rt->rt_timer, r, rtt_link);
 1234         TAILQ_INSERT_TAIL(&queue->rtq_head, r, rtt_next);
 1235         r->rtt_queue->rtq_count++;
 1236 
 1237         return (0);
 1238 }
 1239 
 1240 struct radix_node_head *
 1241 rt_gettable(sa_family_t af, u_int id)
 1242 {
 1243         return (rt_tables[id][af2rtafidx[af]]);
 1244 }
 1245 
 1246 struct radix_node *
 1247 rt_lookup(struct sockaddr *dst, struct sockaddr *mask, int tableid)
 1248 {
 1249         struct radix_node_head  *rnh;
 1250 
 1251         if ((rnh = rt_gettable(dst->sa_family, tableid)) == NULL)
 1252                 return (NULL);
 1253 
 1254         return (rnh->rnh_lookup(dst, mask, rnh));
 1255 }
 1256 
 1257 /* ARGSUSED */
 1258 void
 1259 rt_timer_timer(void *arg)
 1260 {
 1261         struct timeout          *to = (struct timeout *)arg;
 1262         struct rttimer_queue    *rtq;
 1263         struct rttimer          *r;
 1264         long                     current_time;
 1265         int                      s;
 1266 
 1267         current_time = time_uptime;
 1268 
 1269         s = splsoftnet();
 1270         for (rtq = LIST_FIRST(&rttimer_queue_head); rtq != NULL;
 1271              rtq = LIST_NEXT(rtq, rtq_link)) {
 1272                 while ((r = TAILQ_FIRST(&rtq->rtq_head)) != NULL &&
 1273                     (r->rtt_time + rtq->rtq_timeout) < current_time) {
 1274                         LIST_REMOVE(r, rtt_link);
 1275                         TAILQ_REMOVE(&rtq->rtq_head, r, rtt_next);
 1276                         RTTIMER_CALLOUT(r);
 1277                         pool_put(&rttimer_pool, r);
 1278                         if (rtq->rtq_count > 0)
 1279                                 rtq->rtq_count--;
 1280                         else
 1281                                 printf("rt_timer_timer: rtq_count reached 0\n");
 1282                 }
 1283         }
 1284         splx(s);
 1285 
 1286         timeout_add(to, hz);            /* every second */
 1287 }
 1288 
 1289 u_int16_t
 1290 rtlabel_name2id(char *name)
 1291 {
 1292         struct rt_label         *label, *p = NULL;
 1293         u_int16_t                new_id = 1;
 1294 
 1295         if (!name[0])
 1296                 return (0);
 1297 
 1298         TAILQ_FOREACH(label, &rt_labels, rtl_entry)
 1299                 if (strcmp(name, label->rtl_name) == 0) {
 1300                         label->rtl_ref++;
 1301                         return (label->rtl_id);
 1302                 }
 1303 
 1304         /*
 1305          * to avoid fragmentation, we do a linear search from the beginning
 1306          * and take the first free slot we find. if there is none or the list
 1307          * is empty, append a new entry at the end.
 1308          */
 1309 
 1310         if (!TAILQ_EMPTY(&rt_labels))
 1311                 for (p = TAILQ_FIRST(&rt_labels); p != NULL &&
 1312                     p->rtl_id == new_id; p = TAILQ_NEXT(p, rtl_entry))
 1313                         new_id = p->rtl_id + 1;
 1314 
 1315         if (new_id > LABELID_MAX)
 1316                 return (0);
 1317 
 1318         label = (struct rt_label *)malloc(sizeof(struct rt_label),
 1319             M_TEMP, M_NOWAIT);
 1320         if (label == NULL)
 1321                 return (0);
 1322         bzero(label, sizeof(struct rt_label));
 1323         strlcpy(label->rtl_name, name, sizeof(label->rtl_name));
 1324         label->rtl_id = new_id;
 1325         label->rtl_ref++;
 1326 
 1327         if (p != NULL)  /* insert new entry before p */
 1328                 TAILQ_INSERT_BEFORE(p, label, rtl_entry);
 1329         else            /* either list empty or no free slot in between */
 1330                 TAILQ_INSERT_TAIL(&rt_labels, label, rtl_entry);
 1331 
 1332         return (label->rtl_id);
 1333 }
 1334 
 1335 const char *
 1336 rtlabel_id2name(u_int16_t id)
 1337 {
 1338         struct rt_label *label;
 1339 
 1340         TAILQ_FOREACH(label, &rt_labels, rtl_entry)
 1341                 if (label->rtl_id == id)
 1342                         return (label->rtl_name);
 1343 
 1344         return (NULL);
 1345 }
 1346 
 1347 void
 1348 rtlabel_unref(u_int16_t id)
 1349 {
 1350         struct rt_label *p, *next;
 1351 
 1352         if (id == 0)
 1353                 return;
 1354 
 1355         for (p = TAILQ_FIRST(&rt_labels); p != NULL; p = next) {
 1356                 next = TAILQ_NEXT(p, rtl_entry);
 1357                 if (id == p->rtl_id) {
 1358                         if (--p->rtl_ref == 0) {
 1359                                 TAILQ_REMOVE(&rt_labels, p, rtl_entry);
 1360                                 free(p, M_TEMP);
 1361                         }
 1362                         break;
 1363                 }
 1364         }
 1365 }
 1366 
 1367 void
 1368 rt_if_remove(struct ifnet *ifp)
 1369 {
 1370         int                      i;
 1371         struct radix_node_head  *rnh;
 1372 
 1373         for (i = 1; i <= AF_MAX; i++)
 1374                 if ((rnh = rt_gettable(i, 0)) != NULL)
 1375                         while ((*rnh->rnh_walktree)(rnh,
 1376                             rt_if_remove_rtdelete, ifp) == EAGAIN)
 1377                                 ;       /* nothing */
 1378 }
 1379 
 1380 /*
 1381  * Note that deleting a RTF_CLONING route can trigger the
 1382  * deletion of more entries, so we need to cancel the walk
 1383  * and return EAGAIN.  The caller should restart the walk
 1384  * as long as EAGAIN is returned.
 1385  */
 1386 int
 1387 rt_if_remove_rtdelete(struct radix_node *rn, void *vifp)
 1388 {
 1389         struct ifnet    *ifp = vifp;
 1390         struct rtentry  *rt = (struct rtentry *)rn;
 1391 
 1392         if (rt->rt_ifp == ifp) {
 1393                 int     cloning = (rt->rt_flags & RTF_CLONING);
 1394 
 1395                 if (rtrequest(RTM_DELETE, rt_key(rt), rt->rt_gateway,
 1396                     rt_mask(rt), 0, NULL, 0) == 0 && cloning)
 1397                         return (EAGAIN);
 1398         }
 1399 
 1400         /*
 1401          * XXX There should be no need to check for rt_ifa belonging to this
 1402          * interface, because then rt_ifp is set, right?
 1403          */
 1404 
 1405         return (0);
 1406 }

/* [<][>][^][v][top][bottom][index][help] */