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/netinet/if_ether.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  * Copyright (c) 1982, 1986, 1988, 1993
    3  *      The Regents of the University of California.  All rights reserved.
    4  *
    5  * Redistribution and use in source and binary forms, with or without
    6  * modification, are permitted provided that the following conditions
    7  * are met:
    8  * 1. Redistributions of source code must retain the above copyright
    9  *    notice, this list of conditions and the following disclaimer.
   10  * 2. Redistributions in binary form must reproduce the above copyright
   11  *    notice, this list of conditions and the following disclaimer in the
   12  *    documentation and/or other materials provided with the distribution.
   13  * 4. Neither the name of the University nor the names of its contributors
   14  *    may be used to endorse or promote products derived from this software
   15  *    without specific prior written permission.
   16  *
   17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
   18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
   19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
   20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
   21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
   22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
   23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
   24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
   25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
   26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
   27  * SUCH DAMAGE.
   28  *
   29  *      @(#)if_ether.c  8.1 (Berkeley) 6/10/93
   30  */
   31 
   32 /*
   33  * Ethernet address resolution protocol.
   34  * TODO:
   35  *      add "inuse/lock" bit (or ref. count) along with valid bit
   36  */
   37 
   38 #include <sys/cdefs.h>
   39 __FBSDID("$FreeBSD$");
   40 
   41 #include "opt_inet.h"
   42 #include "opt_mac.h"
   43 #include "opt_carp.h"
   44 
   45 #include <sys/param.h>
   46 #include <sys/kernel.h>
   47 #include <sys/queue.h>
   48 #include <sys/sysctl.h>
   49 #include <sys/systm.h>
   50 #include <sys/mbuf.h>
   51 #include <sys/malloc.h>
   52 #include <sys/proc.h>
   53 #include <sys/socket.h>
   54 #include <sys/syslog.h>
   55 
   56 #include <net/if.h>
   57 #include <net/if_dl.h>
   58 #include <net/if_types.h>
   59 #include <net/route.h>
   60 #include <net/netisr.h>
   61 #include <net/if_llc.h>
   62 #include <net/ethernet.h>
   63 
   64 #include <netinet/in.h>
   65 #include <netinet/in_var.h>
   66 #include <netinet/if_ether.h>
   67 
   68 #include <net/if_arc.h>
   69 #include <net/iso88025.h>
   70 
   71 #ifdef DEV_CARP
   72 #include <netinet/ip_carp.h>
   73 #endif
   74 
   75 #include <security/mac/mac_framework.h>
   76 
   77 #define SIN(s) ((struct sockaddr_in *)s)
   78 #define SDL(s) ((struct sockaddr_dl *)s)
   79 
   80 SYSCTL_DECL(_net_link_ether);
   81 SYSCTL_NODE(_net_link_ether, PF_INET, inet, CTLFLAG_RW, 0, "");
   82 
   83 /* timer values */
   84 static int arpt_keep = (20*60); /* once resolved, good for 20 more minutes */
   85 
   86 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, max_age, CTLFLAG_RW, 
   87            &arpt_keep, 0, "ARP entry lifetime in seconds");
   88 
   89 #define rt_expire rt_rmx.rmx_expire
   90 
   91 struct llinfo_arp {
   92         struct  callout la_timer;
   93         struct  rtentry *la_rt;
   94         struct  mbuf *la_hold;  /* last packet until resolved/timeout */
   95         u_short la_preempt;     /* countdown for pre-expiry arps */
   96         u_short la_asked;       /* # requests sent */
   97 };
   98 
   99 static struct   ifqueue arpintrq;
  100 static int      arp_allocated;
  101 
  102 static int      arp_maxtries = 5;
  103 static int      useloopback = 1; /* use loopback interface for local traffic */
  104 static int      arp_proxyall = 0;
  105 
  106 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, maxtries, CTLFLAG_RW,
  107            &arp_maxtries, 0, "ARP resolution attempts before returning error");
  108 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, useloopback, CTLFLAG_RW,
  109            &useloopback, 0, "Use the loopback interface for local traffic");
  110 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, proxyall, CTLFLAG_RW,
  111            &arp_proxyall, 0, "Enable proxy ARP for all suitable requests");
  112 
  113 static void     arp_init(void);
  114 static void     arp_rtrequest(int, struct rtentry *, struct rt_addrinfo *);
  115 static void     arprequest(struct ifnet *,
  116                         struct in_addr *, struct in_addr *, u_char *);
  117 static void     arpintr(struct mbuf *);
  118 static void     arptimer(void *);
  119 static struct rtentry
  120                 *arplookup(u_long, int, int, int);
  121 #ifdef INET
  122 static void     in_arpinput(struct mbuf *);
  123 #endif
  124 
  125 /*
  126  * Timeout routine.
  127  */
  128 static void
  129 arptimer(void *arg)
  130 {
  131         struct rtentry *rt = (struct rtentry *)arg;
  132 
  133         RT_LOCK_ASSERT(rt);
  134         /*
  135          * The lock is needed to close a theoretical race
  136          * between spontaneous expiry and intentional removal.
  137          * We still got an extra reference on rtentry, so can
  138          * safely pass pointers to its contents.
  139          */
  140         RT_UNLOCK(rt);
  141 
  142         in_rtrequest(RTM_DELETE, rt_key(rt), NULL, rt_mask(rt), 0, NULL,
  143             rt->rt_fibnum);
  144 }
  145 
  146 /*
  147  * Parallel to llc_rtrequest.
  148  */
  149 static void
  150 arp_rtrequest(int req, struct rtentry *rt, struct rt_addrinfo *info)
  151 {
  152         struct sockaddr *gate;
  153         struct llinfo_arp *la;
  154         static struct sockaddr_dl null_sdl = {sizeof(null_sdl), AF_LINK};
  155         struct in_ifaddr *ia;
  156         struct ifaddr *ifa;
  157 
  158         RT_LOCK_ASSERT(rt);
  159 
  160         if (rt->rt_flags & RTF_GATEWAY)
  161                 return;
  162         gate = rt->rt_gateway;
  163         la = (struct llinfo_arp *)rt->rt_llinfo;
  164         switch (req) {
  165 
  166         case RTM_ADD:
  167                 /*
  168                  * XXX: If this is a manually added route to interface
  169                  * such as older version of routed or gated might provide,
  170                  * restore cloning bit.
  171                  */
  172                 if ((rt->rt_flags & RTF_HOST) == 0 &&
  173                     rt_mask(rt) != NULL &&
  174                     SIN(rt_mask(rt))->sin_addr.s_addr != 0xffffffff)
  175                         rt->rt_flags |= RTF_CLONING;
  176                 if (rt->rt_flags & RTF_CLONING) {
  177                         /*
  178                          * Case 1: This route should come from a route to iface.
  179                          */
  180                         rt_setgate(rt, rt_key(rt),
  181                                         (struct sockaddr *)&null_sdl);
  182                         gate = rt->rt_gateway;
  183                         SDL(gate)->sdl_type = rt->rt_ifp->if_type;
  184                         SDL(gate)->sdl_index = rt->rt_ifp->if_index;
  185                         rt->rt_expire = time_uptime;
  186                         break;
  187                 }
  188                 /* Announce a new entry if requested. */
  189                 if (rt->rt_flags & RTF_ANNOUNCE)
  190                         arprequest(rt->rt_ifp,
  191                             &SIN(rt_key(rt))->sin_addr,
  192                             &SIN(rt_key(rt))->sin_addr,
  193                             (u_char *)LLADDR(SDL(gate)));
  194                 /*FALLTHROUGH*/
  195         case RTM_RESOLVE:
  196                 if (gate->sa_family != AF_LINK ||
  197                     gate->sa_len < sizeof(null_sdl)) {
  198                         log(LOG_DEBUG, "%s: bad gateway %s%s\n", __func__,
  199                             inet_ntoa(SIN(rt_key(rt))->sin_addr),
  200                             (gate->sa_family != AF_LINK) ?
  201                             " (!AF_LINK)": "");
  202                         break;
  203                 }
  204                 SDL(gate)->sdl_type = rt->rt_ifp->if_type;
  205                 SDL(gate)->sdl_index = rt->rt_ifp->if_index;
  206                 if (la != 0)
  207                         break; /* This happens on a route change */
  208                 /*
  209                  * Case 2:  This route may come from cloning, or a manual route
  210                  * add with a LL address.
  211                  */
  212                 R_Zalloc(la, struct llinfo_arp *, sizeof(*la));
  213                 rt->rt_llinfo = (caddr_t)la;
  214                 if (la == 0) {
  215                         log(LOG_DEBUG, "%s: malloc failed\n", __func__);
  216                         break;
  217                 }
  218                 arp_allocated++;
  219                 /*
  220                  * We are storing a route entry outside of radix tree. So,
  221                  * it can be found and accessed by other means than radix
  222                  * lookup. The routing code assumes that any rtentry detached
  223                  * from radix can be destroyed safely. To prevent this, we
  224                  * add an additional reference.
  225                  */
  226                 RT_ADDREF(rt);
  227                 la->la_rt = rt;
  228                 rt->rt_flags |= RTF_LLINFO;
  229                 callout_init_mtx(&la->la_timer, &rt->rt_mtx,
  230                     CALLOUT_RETURNUNLOCKED);
  231 
  232 #ifdef INET
  233                 /*
  234                  * This keeps the multicast addresses from showing up
  235                  * in `arp -a' listings as unresolved.  It's not actually
  236                  * functional.  Then the same for broadcast.
  237                  */
  238                 if (IN_MULTICAST(ntohl(SIN(rt_key(rt))->sin_addr.s_addr)) &&
  239                     rt->rt_ifp->if_type != IFT_ARCNET) {
  240                         ETHER_MAP_IP_MULTICAST(&SIN(rt_key(rt))->sin_addr,
  241                                                LLADDR(SDL(gate)));
  242                         SDL(gate)->sdl_alen = 6;
  243                         rt->rt_expire = 0;
  244                 }
  245                 if (in_broadcast(SIN(rt_key(rt))->sin_addr, rt->rt_ifp)) {
  246                         memcpy(LLADDR(SDL(gate)), rt->rt_ifp->if_broadcastaddr,
  247                                rt->rt_ifp->if_addrlen);
  248                         SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen;
  249                         rt->rt_expire = 0;
  250                 }
  251 #endif
  252 
  253                 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
  254                         if (ia->ia_ifp == rt->rt_ifp &&
  255                             SIN(rt_key(rt))->sin_addr.s_addr ==
  256                             (IA_SIN(ia))->sin_addr.s_addr)
  257                                 break;
  258                 }
  259                 if (ia) {
  260                     /*
  261                      * This test used to be
  262                      *  if (loif.if_flags & IFF_UP)
  263                      * It allowed local traffic to be forced
  264                      * through the hardware by configuring the loopback down.
  265                      * However, it causes problems during network configuration
  266                      * for boards that can't receive packets they send.
  267                      * It is now necessary to clear "useloopback" and remove
  268                      * the route to force traffic out to the hardware.
  269                      */
  270                         rt->rt_expire = 0;
  271                         bcopy(IF_LLADDR(rt->rt_ifp), LLADDR(SDL(gate)),
  272                               SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen);
  273                         if (useloopback) {
  274                                 rt->rt_ifp = loif;
  275                                 rt->rt_rmx.rmx_mtu = loif->if_mtu;
  276                         }
  277 
  278                     /*
  279                      * make sure to set rt->rt_ifa to the interface
  280                      * address we are using, otherwise we will have trouble
  281                      * with source address selection.
  282                      */
  283                         ifa = &ia->ia_ifa;
  284                         if (ifa != rt->rt_ifa) {
  285                                 IFAFREE(rt->rt_ifa);
  286                                 IFAREF(ifa);
  287                                 rt->rt_ifa = ifa;
  288                         }
  289                 }
  290                 break;
  291 
  292         case RTM_DELETE:
  293                 if (la == NULL) /* XXX: at least CARP does this. */
  294                         break;
  295                 callout_stop(&la->la_timer);
  296                 rt->rt_llinfo = NULL;
  297                 rt->rt_flags &= ~RTF_LLINFO;
  298                 RT_REMREF(rt);
  299                 if (la->la_hold)
  300                         m_freem(la->la_hold);
  301                 Free((caddr_t)la);
  302         }
  303 }
  304 
  305 /*
  306  * Broadcast an ARP request. Caller specifies:
  307  *      - arp header source ip address
  308  *      - arp header target ip address
  309  *      - arp header source ethernet address
  310  */
  311 static void
  312 arprequest(struct ifnet *ifp, struct in_addr *sip, struct in_addr *tip,
  313     u_char *enaddr)
  314 {
  315         struct mbuf *m;
  316         struct arphdr *ah;
  317         struct sockaddr sa;
  318 
  319         if ((m = m_gethdr(M_DONTWAIT, MT_DATA)) == NULL)
  320                 return;
  321         m->m_len = sizeof(*ah) + 2*sizeof(struct in_addr) +
  322                 2*ifp->if_data.ifi_addrlen;
  323         m->m_pkthdr.len = m->m_len;
  324         MH_ALIGN(m, m->m_len);
  325         ah = mtod(m, struct arphdr *);
  326         bzero((caddr_t)ah, m->m_len);
  327 #ifdef MAC
  328         mac_create_mbuf_linklayer(ifp, m);
  329 #endif
  330         ah->ar_pro = htons(ETHERTYPE_IP);
  331         ah->ar_hln = ifp->if_addrlen;           /* hardware address length */
  332         ah->ar_pln = sizeof(struct in_addr);    /* protocol address length */
  333         ah->ar_op = htons(ARPOP_REQUEST);
  334         bcopy((caddr_t)enaddr, (caddr_t)ar_sha(ah), ah->ar_hln);
  335         bcopy((caddr_t)sip, (caddr_t)ar_spa(ah), ah->ar_pln);
  336         bcopy((caddr_t)tip, (caddr_t)ar_tpa(ah), ah->ar_pln);
  337         sa.sa_family = AF_ARP;
  338         sa.sa_len = 2;
  339         m->m_flags |= M_BCAST;
  340         (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)0);
  341 
  342         return;
  343 }
  344 
  345 /*
  346  * Resolve an IP address into an ethernet address.
  347  * On input:
  348  *    ifp is the interface we use
  349  *    dst is the next hop,
  350  *    rt0 is the route to the final destination (possibly useless)
  351  *    m is the mbuf
  352  *    desten is where we want the address.
  353  *
  354  * On success, desten is filled in and the function returns 0;
  355  * If the packet must be held pending resolution, we return EWOULDBLOCK
  356  * On other errors, we return the corresponding error code.
  357  */
  358 int
  359 arpresolve(struct ifnet *ifp, struct rtentry *rt0, struct mbuf *m,
  360     struct sockaddr *dst, u_char *desten)
  361 {
  362         struct llinfo_arp *la = NULL;
  363         struct rtentry *rt = NULL;
  364         struct sockaddr_dl *sdl;
  365         int error;
  366         int fibnum = -1;
  367 
  368         if (m) {
  369                 
  370                 if (m->m_flags & M_BCAST) {     /* broadcast */
  371                         (void)memcpy(desten, ifp->if_broadcastaddr, ifp->if_addrlen);
  372                         return (0);
  373                 }
  374                 if (m->m_flags & M_MCAST && ifp->if_type != IFT_ARCNET) {/* multicast */
  375                         ETHER_MAP_IP_MULTICAST(&SIN(dst)->sin_addr, desten);
  376                         return (0);
  377                 }
  378                 fibnum = M_GETFIB(m);
  379         }
  380 
  381         if (rt0 != NULL) {
  382                 /* Look for a cached arp (ll) entry. */
  383                 error = rt_check(&rt, &rt0, dst);
  384                 if (error) {
  385                         m_freem(m);
  386                         return error;
  387                 }
  388                 la = (struct llinfo_arp *)rt->rt_llinfo;
  389                 if (la == NULL)
  390                         RT_UNLOCK(rt);
  391         }
  392 
  393         /*
  394          * If we had no mbuf and no route, then hope the caller
  395          * has a fib in mind because we are running out of ideas.
  396          * I think this should not happen in current code.
  397          * (kmacy would know).
  398          */
  399         if (fibnum == -1)
  400                 fibnum = curthread->td_proc->p_fibnum; /* last gasp */
  401 
  402         if (la == NULL) {
  403                 /*
  404                  * We enter this block if rt0 was NULL,
  405                  * or if rt found by rt_check() didn't have llinfo.
  406                  * We should get a cloned route from the local interface,
  407                  * so it should have an ll entry.
  408                  * It may be incomplete but that's ok.
  409                  */
  410                 rt = arplookup(SIN(dst)->sin_addr.s_addr, 1, 0, fibnum);
  411                 if (rt == NULL) {
  412                         log(LOG_DEBUG,
  413                             "arpresolve: can't allocate route for %s\n",
  414                             inet_ntoa(SIN(dst)->sin_addr));
  415                         m_freem(m);
  416                         return (EINVAL); /* XXX */
  417                 }
  418                 la = (struct llinfo_arp *)rt->rt_llinfo;
  419                 if (la == NULL) {
  420                         RT_UNLOCK(rt);
  421                         log(LOG_DEBUG,
  422                             "arpresolve: can't allocate llinfo for %s\n",
  423                             inet_ntoa(SIN(dst)->sin_addr));
  424                         m_freem(m);
  425                         return (EINVAL); /* XXX */
  426                 }
  427         }
  428         sdl = SDL(rt->rt_gateway);
  429         /*
  430          * Check the address family and length is valid, the address
  431          * is resolved; otherwise, try to resolve.
  432          */
  433         if ((rt->rt_expire == 0 || rt->rt_expire > time_uptime) &&
  434             sdl->sdl_family == AF_LINK && sdl->sdl_alen != 0) {
  435 
  436                 bcopy(LLADDR(sdl), desten, sdl->sdl_alen);
  437 
  438                 /*
  439                  * If entry has an expiry time and it is approaching,
  440                  * send an ARP request.
  441                  */
  442                 if ((rt->rt_expire != 0) &&
  443                     (time_uptime + la->la_preempt > rt->rt_expire)) {
  444                         struct in_addr sin = 
  445                             SIN(rt->rt_ifa->ifa_addr)->sin_addr;
  446 
  447                         la->la_preempt--;
  448                         RT_UNLOCK(rt);
  449                         arprequest(ifp, &sin, &SIN(dst)->sin_addr,
  450                             IF_LLADDR(ifp));
  451                         return (0);
  452                 } 
  453 
  454                 RT_UNLOCK(rt);
  455                 return (0);
  456         }
  457         /*
  458          * If ARP is disabled or static on this interface, stop.
  459          * XXX
  460          * Probably should not allocate empty llinfo struct if we are
  461          * not going to be sending out an arp request.
  462          */
  463         if (ifp->if_flags & (IFF_NOARP | IFF_STATICARP)) {
  464                 RT_UNLOCK(rt);
  465                 m_freem(m);
  466                 return (EINVAL);
  467         }
  468         /*
  469          * There is an arptab entry, but no ethernet address
  470          * response yet.  Replace the held mbuf with this
  471          * latest one.
  472          */
  473         if (m) {
  474                 if (la->la_hold)
  475                         m_freem(la->la_hold);
  476                 la->la_hold = m;
  477         }
  478                 
  479         KASSERT(rt->rt_expire > 0, ("sending ARP request for static entry"));
  480 
  481         /*
  482          * Return EWOULDBLOCK if we have tried less than arp_maxtries. It
  483          * will be masked by ether_output(). Return EHOSTDOWN/EHOSTUNREACH
  484          * if we have already sent arp_maxtries ARP requests. Retransmit the
  485          * ARP request, but not faster than one request per second.
  486          */
  487         if (la->la_asked < arp_maxtries)
  488                 error = EWOULDBLOCK;    /* First request. */
  489         else
  490                 error = (rt == rt0) ? EHOSTDOWN : EHOSTUNREACH;
  491 
  492         if (la->la_asked == 0 || rt->rt_expire != time_uptime) {
  493                 struct in_addr sin =
  494                     SIN(rt->rt_ifa->ifa_addr)->sin_addr;
  495 
  496                 rt->rt_expire = time_uptime;
  497                 callout_reset(&la->la_timer, hz, arptimer, rt);
  498                 la->la_asked++;
  499                 RT_UNLOCK(rt);
  500 
  501                 arprequest(ifp, &sin, &SIN(dst)->sin_addr,
  502                     IF_LLADDR(ifp));
  503         } else
  504                 RT_UNLOCK(rt);
  505 
  506         return (error);
  507 }
  508 
  509 /*
  510  * Common length and type checks are done here,
  511  * then the protocol-specific routine is called.
  512  */
  513 static void
  514 arpintr(struct mbuf *m)
  515 {
  516         struct arphdr *ar;
  517 
  518         if (m->m_len < sizeof(struct arphdr) &&
  519             ((m = m_pullup(m, sizeof(struct arphdr))) == NULL)) {
  520                 log(LOG_ERR, "arp: runt packet -- m_pullup failed\n");
  521                 return;
  522         }
  523         ar = mtod(m, struct arphdr *);
  524 
  525         if (ntohs(ar->ar_hrd) != ARPHRD_ETHER &&
  526             ntohs(ar->ar_hrd) != ARPHRD_IEEE802 &&
  527             ntohs(ar->ar_hrd) != ARPHRD_ARCNET &&
  528             ntohs(ar->ar_hrd) != ARPHRD_IEEE1394) {
  529                 log(LOG_ERR, "arp: unknown hardware address format (0x%2D)\n",
  530                     (unsigned char *)&ar->ar_hrd, "");
  531                 m_freem(m);
  532                 return;
  533         }
  534 
  535         if (m->m_len < arphdr_len(ar)) {
  536                 if ((m = m_pullup(m, arphdr_len(ar))) == NULL) {
  537                         log(LOG_ERR, "arp: runt packet\n");
  538                         m_freem(m);
  539                         return;
  540                 }
  541                 ar = mtod(m, struct arphdr *);
  542         }
  543 
  544         switch (ntohs(ar->ar_pro)) {
  545 #ifdef INET
  546         case ETHERTYPE_IP:
  547                 in_arpinput(m);
  548                 return;
  549 #endif
  550         }
  551         m_freem(m);
  552 }
  553 
  554 #ifdef INET
  555 /*
  556  * ARP for Internet protocols on 10 Mb/s Ethernet.
  557  * Algorithm is that given in RFC 826.
  558  * In addition, a sanity check is performed on the sender
  559  * protocol address, to catch impersonators.
  560  * We no longer handle negotiations for use of trailer protocol:
  561  * Formerly, ARP replied for protocol type ETHERTYPE_TRAIL sent
  562  * along with IP replies if we wanted trailers sent to us,
  563  * and also sent them in response to IP replies.
  564  * This allowed either end to announce the desire to receive
  565  * trailer packets.
  566  * We no longer reply to requests for ETHERTYPE_TRAIL protocol either,
  567  * but formerly didn't normally send requests.
  568  */
  569 static int log_arp_wrong_iface = 1;
  570 static int log_arp_movements = 1;
  571 static int log_arp_permanent_modify = 1;
  572 
  573 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, log_arp_wrong_iface, CTLFLAG_RW,
  574         &log_arp_wrong_iface, 0,
  575         "log arp packets arriving on the wrong interface");
  576 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, log_arp_movements, CTLFLAG_RW,
  577         &log_arp_movements, 0,
  578         "log arp replies from MACs different than the one in the cache");
  579 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, log_arp_permanent_modify, CTLFLAG_RW,
  580         &log_arp_permanent_modify, 0,
  581         "log arp replies from MACs different than the one in the permanent arp entry");
  582 
  583 
  584 static void
  585 in_arpinput(struct mbuf *m)
  586 {
  587         struct arphdr *ah;
  588         struct ifnet *ifp = m->m_pkthdr.rcvif;
  589         struct llinfo_arp *la;
  590         struct rtentry *rt;
  591         struct ifaddr *ifa;
  592         struct in_ifaddr *ia;
  593         struct sockaddr_dl *sdl;
  594         struct sockaddr sa;
  595         struct in_addr isaddr, itaddr, myaddr;
  596         struct mbuf *hold;
  597         u_int8_t *enaddr = NULL;
  598         int op, rif_len;
  599         int req_len;
  600         int bridged = 0, is_bridge = 0;
  601         u_int fibnum;
  602         u_int goodfib = 0;
  603         int firstpass = 1;
  604 #ifdef DEV_CARP
  605         int carp_match = 0;
  606 #endif
  607 
  608         if (ifp->if_bridge)
  609                 bridged = 1;
  610         if (ifp->if_type == IFT_BRIDGE)
  611                 is_bridge = 1;
  612 
  613         req_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
  614         if (m->m_len < req_len && (m = m_pullup(m, req_len)) == NULL) {
  615                 log(LOG_ERR, "in_arp: runt packet -- m_pullup failed\n");
  616                 return;
  617         }
  618 
  619         ah = mtod(m, struct arphdr *);
  620         op = ntohs(ah->ar_op);
  621         (void)memcpy(&isaddr, ar_spa(ah), sizeof (isaddr));
  622         (void)memcpy(&itaddr, ar_tpa(ah), sizeof (itaddr));
  623 
  624         /*
  625          * For a bridge, we want to check the address irrespective
  626          * of the receive interface. (This will change slightly
  627          * when we have clusters of interfaces).
  628          * If the interface does not match, but the recieving interface
  629          * is part of carp, we call carp_iamatch to see if this is a
  630          * request for the virtual host ip.
  631          * XXX: This is really ugly!
  632          */
  633         LIST_FOREACH(ia, INADDR_HASH(itaddr.s_addr), ia_hash) {
  634                 if (((bridged && ia->ia_ifp->if_bridge != NULL) ||
  635                     (ia->ia_ifp == ifp)) &&
  636                     itaddr.s_addr == ia->ia_addr.sin_addr.s_addr)
  637                         goto match;
  638 #ifdef DEV_CARP
  639                 if (ifp->if_carp != NULL &&
  640                     carp_iamatch(ifp->if_carp, ia, &isaddr, &enaddr) &&
  641                     itaddr.s_addr == ia->ia_addr.sin_addr.s_addr) {
  642                         carp_match = 1;
  643                         goto match;
  644                 }
  645 #endif
  646         }
  647         LIST_FOREACH(ia, INADDR_HASH(isaddr.s_addr), ia_hash)
  648                 if (((bridged && ia->ia_ifp->if_bridge != NULL) ||
  649                     (ia->ia_ifp == ifp)) &&
  650                     isaddr.s_addr == ia->ia_addr.sin_addr.s_addr)
  651                         goto match;
  652 
  653 #define BDG_MEMBER_MATCHES_ARP(addr, ifp, ia)                           \
  654   (ia->ia_ifp->if_bridge == ifp->if_softc &&                            \
  655   !bcmp(IF_LLADDR(ia->ia_ifp), IF_LLADDR(ifp), ifp->if_addrlen) &&      \
  656   addr == ia->ia_addr.sin_addr.s_addr)
  657         /*
  658          * Check the case when bridge shares its MAC address with
  659          * some of its children, so packets are claimed by bridge
  660          * itself (bridge_input() does it first), but they are really
  661          * meant to be destined to the bridge member.
  662          */
  663         if (is_bridge) {
  664                 LIST_FOREACH(ia, INADDR_HASH(itaddr.s_addr), ia_hash) {
  665                         if (BDG_MEMBER_MATCHES_ARP(itaddr.s_addr, ifp, ia)) {
  666                                 ifp = ia->ia_ifp;
  667                                 goto match;
  668                         }
  669                 }
  670         }
  671 #undef BDG_MEMBER_MATCHES_ARP
  672 
  673         /*
  674          * No match, use the first inet address on the receive interface
  675          * as a dummy address for the rest of the function.
  676          */
  677         TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link)
  678                 if (ifa->ifa_addr->sa_family == AF_INET) {
  679                         ia = ifatoia(ifa);
  680                         goto match;
  681                 }
  682         /*
  683          * If bridging, fall back to using any inet address.
  684          */
  685         if (!bridged || (ia = TAILQ_FIRST(&in_ifaddrhead)) == NULL)
  686                 goto drop;
  687 match:
  688         if (!enaddr)
  689                 enaddr = (u_int8_t *)IF_LLADDR(ifp);
  690         myaddr = ia->ia_addr.sin_addr;
  691         if (!bcmp(ar_sha(ah), enaddr, ifp->if_addrlen))
  692                 goto drop;      /* it's from me, ignore it. */
  693         if (!bcmp(ar_sha(ah), ifp->if_broadcastaddr, ifp->if_addrlen)) {
  694                 log(LOG_ERR,
  695                     "arp: link address is broadcast for IP address %s!\n",
  696                     inet_ntoa(isaddr));
  697                 goto drop;
  698         }
  699         /*
  700          * Warn if another host is using the same IP address, but only if the
  701          * IP address isn't 0.0.0.0, which is used for DHCP only, in which
  702          * case we suppress the warning to avoid false positive complaints of
  703          * potential misconfiguration.
  704          */
  705         if (!bridged && isaddr.s_addr == myaddr.s_addr && myaddr.s_addr != 0) {
  706                 log(LOG_ERR,
  707                    "arp: %*D is using my IP address %s on %s!\n",
  708                    ifp->if_addrlen, (u_char *)ar_sha(ah), ":",
  709                    inet_ntoa(isaddr), ifp->if_xname);
  710                 itaddr = myaddr;
  711                 goto reply;
  712         }
  713         if (ifp->if_flags & IFF_STATICARP)
  714                 goto reply;
  715         /*
  716          * We look for any FIBs that has this address to find
  717          * the interface etc.
  718          * For sanity checks that are FIB independent we abort the loop.
  719          */
  720         for (fibnum = 0; fibnum < rt_numfibs; fibnum++) {
  721                 rt = arplookup(isaddr.s_addr,
  722                     itaddr.s_addr == myaddr.s_addr, 0, fibnum);
  723                 if (rt == NULL)
  724                         continue;
  725                 
  726                 sdl = SDL(rt->rt_gateway);
  727                 if (firstpass) {
  728                         struct sockaddr_in sin;
  729 
  730                         sin.sin_len = sizeof(struct sockaddr_in);
  731                         sin.sin_family = AF_INET;
  732                         sin.sin_addr.s_addr = isaddr.s_addr;
  733                         EVENTHANDLER_INVOKE(route_arp_update_event, rt,
  734                             ar_sha(ah), (struct sockaddr *)&sin);
  735                 }
  736                 la = (struct llinfo_arp *)rt->rt_llinfo;
  737                 if (la == NULL) {
  738                         RT_UNLOCK(rt);
  739                         continue;
  740                 }
  741 
  742                 if (firstpass) {
  743                         /* The following is not an error when doing bridging. */
  744                         if (!bridged && rt->rt_ifp != ifp
  745 #ifdef DEV_CARP
  746                             && (ifp->if_type != IFT_CARP || !carp_match)
  747 #endif
  748                             ) {
  749                                 if (log_arp_wrong_iface)
  750                                         log(LOG_ERR, "arp: %s is on %s "
  751                                                 "but got reply from %*D "
  752                                                 "on %s\n",
  753                                             inet_ntoa(isaddr),
  754                                             rt->rt_ifp->if_xname,
  755                                             ifp->if_addrlen,
  756                                             (u_char *)ar_sha(ah), ":",
  757                                             ifp->if_xname);
  758                                 RT_UNLOCK(rt);
  759                                 break;
  760                         }
  761                         if (sdl->sdl_alen &&
  762                             bcmp(ar_sha(ah), LLADDR(sdl), sdl->sdl_alen)) {
  763                                 if (rt->rt_expire) {
  764                                     if (log_arp_movements)
  765                                         log(LOG_INFO,
  766                                             "arp: %s moved from %*D to %*D "
  767                                             "on %s\n",
  768                                             inet_ntoa(isaddr),
  769                                             ifp->if_addrlen,
  770                                             (u_char *)LLADDR(sdl), ":",
  771                                             ifp->if_addrlen,
  772                                             (u_char *)ar_sha(ah), ":",
  773                                             ifp->if_xname);
  774                                 } else {
  775                                         RT_UNLOCK(rt);
  776                                         if (log_arp_permanent_modify)
  777                                                 log(LOG_ERR,
  778                                                     "arp: %*D attempts to "
  779                                                     "modify permanent entry "
  780                                                     "for %s on %s\n",
  781                                                     ifp->if_addrlen,
  782                                                     (u_char *)ar_sha(ah), ":",
  783                                                     inet_ntoa(isaddr),
  784                                                     ifp->if_xname);
  785                                         break;
  786                                 }
  787                         }
  788                         /*
  789                          * sanity check for the address length.
  790                          * XXX this does not work for protocols
  791                          * with variable address length. -is
  792                          */
  793                         if (sdl->sdl_alen &&
  794                             sdl->sdl_alen != ah->ar_hln) {
  795                                 log(LOG_WARNING,
  796                                     "arp from %*D: new addr len %d, was %d",
  797                                     ifp->if_addrlen, (u_char *) ar_sha(ah),
  798                                     ":", ah->ar_hln, sdl->sdl_alen);
  799                         }
  800                         if (ifp->if_addrlen != ah->ar_hln) {
  801                                 log(LOG_WARNING,
  802                                     "arp from %*D: addr len: "
  803                                     "new %d, i/f %d (ignored)",
  804                                     ifp->if_addrlen, (u_char *) ar_sha(ah),
  805                                     ":", ah->ar_hln, ifp->if_addrlen);
  806                                 RT_UNLOCK(rt);
  807                                 break;
  808                         }
  809                         firstpass = 0;
  810                         goodfib = fibnum;
  811                 }
  812 
  813                 /* Copy in the information received. */
  814                 (void)memcpy(LLADDR(sdl), ar_sha(ah),
  815                     sdl->sdl_alen = ah->ar_hln);
  816                 /*
  817                  * If we receive an arp from a token-ring station over
  818                  * a token-ring nic then try to save the source routing info.
  819                  * XXXMRT Only minimal Token Ring support for MRT.
  820                  * Only do this on the first pass as if modifies the mbuf.
  821                  */
  822                 if (ifp->if_type == IFT_ISO88025) {
  823                         struct iso88025_header *th = NULL;
  824                         struct iso88025_sockaddr_dl_data *trld;
  825 
  826                         /* force the fib loop to end after this pass */
  827                         fibnum = rt_numfibs - 1;
  828 
  829                         th = (struct iso88025_header *)m->m_pkthdr.header;
  830                         trld = SDL_ISO88025(sdl);
  831                         rif_len = TR_RCF_RIFLEN(th->rcf);
  832                         if ((th->iso88025_shost[0] & TR_RII) &&
  833                             (rif_len > 2)) {
  834                                 trld->trld_rcf = th->rcf;
  835                                 trld->trld_rcf ^= htons(TR_RCF_DIR);
  836                                 memcpy(trld->trld_route, th->rd, rif_len - 2);
  837                                 trld->trld_rcf &= ~htons(TR_RCF_BCST_MASK);
  838                                 /*
  839                                  * Set up source routing information for
  840                                  * reply packet (XXX)
  841                                  */
  842                                 m->m_data -= rif_len;
  843                                 m->m_len  += rif_len;
  844                                 m->m_pkthdr.len += rif_len;
  845                         } else {
  846                                 th->iso88025_shost[0] &= ~TR_RII;
  847                                 trld->trld_rcf = 0;
  848                         }
  849                         m->m_data -= 8;
  850                         m->m_len  += 8;
  851                         m->m_pkthdr.len += 8;
  852                         th->rcf = trld->trld_rcf;
  853                 }
  854 
  855                 if (rt->rt_expire) {
  856                         rt->rt_expire = time_uptime + arpt_keep;
  857                         callout_reset(&la->la_timer, hz * arpt_keep,
  858                             arptimer, rt);
  859                 }
  860                 la->la_asked = 0;
  861                 la->la_preempt = arp_maxtries;
  862                 hold = la->la_hold;
  863                 la->la_hold = NULL;
  864                 RT_UNLOCK(rt);
  865                 if (hold != NULL)
  866                         (*ifp->if_output)(ifp, hold, rt_key(rt), rt);
  867         } /* end of FIB loop */
  868 reply:
  869 
  870         /*
  871          * Decide if we have to respond to something.
  872          */
  873         if (op != ARPOP_REQUEST)
  874                 goto drop;
  875         if (itaddr.s_addr == myaddr.s_addr) {
  876                 /* Shortcut.. the receiving interface is the target. */
  877                 (void)memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
  878                 (void)memcpy(ar_sha(ah), enaddr, ah->ar_hln);
  879         } else {
  880                 /* It's not asking for our address. But it still may
  881                  * be something we should answer.
  882                  *
  883                  * XXX MRT
  884                  * We assume that link level info is independent of
  885                  * the table used and so we use whichever we can and don't
  886                  * have a better option.
  887                  */
  888                 /* Have we been asked to proxy for the target. */
  889                 rt = arplookup(itaddr.s_addr, 0, SIN_PROXY, goodfib);
  890                 if (rt == NULL) {
  891                         /* Nope, only intersted now if proxying everything. */
  892                         struct sockaddr_in sin;
  893 
  894                         if (!arp_proxyall)
  895                                 goto drop;
  896 
  897                         bzero(&sin, sizeof sin);
  898                         sin.sin_family = AF_INET;
  899                         sin.sin_len = sizeof sin;
  900                         sin.sin_addr = itaddr;
  901 
  902                         /* XXX MRT use table 0 for arp reply  */
  903                         rt = in_rtalloc1((struct sockaddr *)&sin, 0, 0UL, 0);
  904                         if (!rt)
  905                                 goto drop;
  906                         /*
  907                          * Don't send proxies for nodes on the same interface
  908                          * as this one came out of, or we'll get into a fight
  909                          * over who claims what Ether address.
  910                          */
  911                         if (rt->rt_ifp == ifp) {
  912                                 RTFREE_LOCKED(rt);
  913                                 goto drop;
  914                         }
  915                         (void)memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
  916                         (void)memcpy(ar_sha(ah), enaddr, ah->ar_hln);
  917                         RTFREE_LOCKED(rt);
  918 
  919                         /*
  920                          * Also check that the node which sent the ARP packet
  921                          * is on the the interface we expect it to be on. This
  922                          * avoids ARP chaos if an interface is connected to the
  923                          * wrong network.
  924                          */
  925                         sin.sin_addr = isaddr;
  926 
  927                         /* XXX MRT use table 0 for arp checks */
  928                         rt = in_rtalloc1((struct sockaddr *)&sin, 0, 0UL, 0);
  929                         if (!rt)
  930                                 goto drop;
  931                         if (rt->rt_ifp != ifp) {
  932                                 log(LOG_INFO, "arp_proxy: ignoring request"
  933                                     " from %s via %s, expecting %s\n",
  934                                     inet_ntoa(isaddr), ifp->if_xname,
  935                                     rt->rt_ifp->if_xname);
  936                                 RTFREE_LOCKED(rt);
  937                                 goto drop;
  938                         }
  939                         RTFREE_LOCKED(rt);
  940 
  941 #ifdef DEBUG_PROXY
  942                         printf("arp: proxying for %s\n",
  943                                inet_ntoa(itaddr));
  944 #endif
  945                 } else {
  946                         /*
  947                          * Return proxied ARP replies only on the interface
  948                          * or bridge cluster where this network resides.
  949                          * Otherwise we may conflict with the host we are
  950                          * proxying for.
  951                          */
  952                         if (rt->rt_ifp != ifp &&
  953                             (rt->rt_ifp->if_bridge != ifp->if_bridge ||
  954                             ifp->if_bridge == NULL)) {
  955                                 RT_UNLOCK(rt);
  956                                 goto drop;
  957                         }
  958                         sdl = SDL(rt->rt_gateway);
  959                         (void)memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
  960                         (void)memcpy(ar_sha(ah), LLADDR(sdl), ah->ar_hln);
  961                         RT_UNLOCK(rt);
  962                 }
  963         }
  964 
  965         if (itaddr.s_addr == myaddr.s_addr &&
  966             IN_LINKLOCAL(ntohl(itaddr.s_addr))) {
  967                 /* RFC 3927 link-local IPv4; always reply by broadcast. */
  968 #ifdef DEBUG_LINKLOCAL
  969                 printf("arp: sending reply for link-local addr %s\n",
  970                     inet_ntoa(itaddr));
  971 #endif
  972                 m->m_flags |= M_BCAST;
  973                 m->m_flags &= ~M_MCAST;
  974         } else {
  975                 /* default behaviour; never reply by broadcast. */
  976                 m->m_flags &= ~(M_BCAST|M_MCAST);
  977         }
  978         (void)memcpy(ar_tpa(ah), ar_spa(ah), ah->ar_pln);
  979         (void)memcpy(ar_spa(ah), &itaddr, ah->ar_pln);
  980         ah->ar_op = htons(ARPOP_REPLY);
  981         ah->ar_pro = htons(ETHERTYPE_IP); /* let's be sure! */
  982         m->m_len = sizeof(*ah) + (2 * ah->ar_pln) + (2 * ah->ar_hln);   
  983         m->m_pkthdr.len = m->m_len;   
  984         sa.sa_family = AF_ARP;
  985         sa.sa_len = 2;
  986         (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)0);
  987         return;
  988 
  989 drop:
  990         m_freem(m);
  991 }
  992 #endif
  993 
  994 /*
  995  * Lookup or enter a new address in arptab.
  996  */
  997 static struct rtentry *
  998 arplookup(u_long addr, int create, int proxy, int fibnum)
  999 {
 1000         struct rtentry *rt;
 1001         struct sockaddr_inarp sin;
 1002         const char *why = 0;
 1003 
 1004         bzero(&sin, sizeof(sin));
 1005         sin.sin_len = sizeof(sin);
 1006         sin.sin_family = AF_INET;
 1007         sin.sin_addr.s_addr = addr;
 1008         if (proxy)
 1009                 sin.sin_other = SIN_PROXY;
 1010         rt = in_rtalloc1((struct sockaddr *)&sin, create, 0UL, fibnum);
 1011         if (rt == 0)
 1012                 return (0);
 1013 
 1014         if (rt->rt_flags & RTF_GATEWAY)
 1015                 why = "host is not on local network";
 1016         else if ((rt->rt_flags & RTF_LLINFO) == 0)
 1017                 why = "could not allocate llinfo";
 1018         else if (rt->rt_gateway->sa_family != AF_LINK)
 1019                 why = "gateway route is not ours";
 1020 
 1021         if (why) {
 1022 #define ISDYNCLONE(_rt) \
 1023         (((_rt)->rt_flags & (RTF_STATIC | RTF_WASCLONED)) == RTF_WASCLONED)
 1024                 if (create)
 1025                         log(LOG_DEBUG, "arplookup %s failed: %s\n",
 1026                             inet_ntoa(sin.sin_addr), why);
 1027                 /*
 1028                  * If there are no references to this Layer 2 route,
 1029                  * and it is a cloned route, and not static, and
 1030                  * arplookup() is creating the route, then purge
 1031                  * it from the routing table as it is probably bogus.
 1032                  */
 1033                 if (rt->rt_refcnt == 1 && ISDYNCLONE(rt))
 1034                         rtexpunge(rt);
 1035                 RTFREE_LOCKED(rt);
 1036                 return (0);
 1037 #undef ISDYNCLONE
 1038         } else {
 1039                 RT_REMREF(rt);
 1040                 return (rt);
 1041         }
 1042 }
 1043 
 1044 void
 1045 arp_ifinit(struct ifnet *ifp, struct ifaddr *ifa)
 1046 {
 1047         if (ntohl(IA_SIN(ifa)->sin_addr.s_addr) != INADDR_ANY)
 1048                 arprequest(ifp, &IA_SIN(ifa)->sin_addr,
 1049                                 &IA_SIN(ifa)->sin_addr, IF_LLADDR(ifp));
 1050         ifa->ifa_rtrequest = arp_rtrequest;
 1051         ifa->ifa_flags |= RTF_CLONING;
 1052 }
 1053 
 1054 void
 1055 arp_ifinit2(struct ifnet *ifp, struct ifaddr *ifa, u_char *enaddr)
 1056 {
 1057         if (ntohl(IA_SIN(ifa)->sin_addr.s_addr) != INADDR_ANY)
 1058                 arprequest(ifp, &IA_SIN(ifa)->sin_addr,
 1059                                 &IA_SIN(ifa)->sin_addr, enaddr);
 1060         ifa->ifa_rtrequest = arp_rtrequest;
 1061         ifa->ifa_flags |= RTF_CLONING;
 1062 }
 1063 
 1064 static void
 1065 arp_init(void)
 1066 {
 1067 
 1068         arpintrq.ifq_maxlen = 50;
 1069         mtx_init(&arpintrq.ifq_mtx, "arp_inq", NULL, MTX_DEF);
 1070         netisr_register(NETISR_ARP, arpintr, &arpintrq, NETISR_MPSAFE);
 1071 }
 1072 SYSINIT(arp, SI_SUB_PROTO_DOMAIN, SI_ORDER_ANY, arp_init, 0);

Cache object: 74ecf5534216bc51fd22a55e7f6ae633


[ 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.