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/ip_output.c

Version: -  FREEBSD  -  FREEBSD-13-STABLE  -  FREEBSD-13-0  -  FREEBSD-12-STABLE  -  FREEBSD-12-0  -  FREEBSD-11-STABLE  -  FREEBSD-11-0  -  FREEBSD-10-STABLE  -  FREEBSD-10-0  -  FREEBSD-9-STABLE  -  FREEBSD-9-0  -  FREEBSD-8-STABLE  -  FREEBSD-8-0  -  FREEBSD-7-STABLE  -  FREEBSD-7-0  -  FREEBSD-6-STABLE  -  FREEBSD-6-0  -  FREEBSD-5-STABLE  -  FREEBSD-5-0  -  FREEBSD-4-STABLE  -  FREEBSD-3-STABLE  -  FREEBSD22  -  l41  -  OPENBSD  -  linux-2.6  -  MK84  -  PLAN9  -  xnu-8792 
SearchContext: -  none  -  3  -  10 

    1 /*-
    2  * Copyright (c) 1982, 1986, 1988, 1990, 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  *      @(#)ip_output.c 8.3 (Berkeley) 1/21/94
   30  */
   31 
   32 #include <sys/cdefs.h>
   33 __FBSDID("$FreeBSD: releng/9.1/sys/netinet/ip_output.c 238713 2012-07-23 09:19:14Z glebius $");
   34 
   35 #include "opt_ipfw.h"
   36 #include "opt_ipsec.h"
   37 #include "opt_route.h"
   38 #include "opt_mbuf_stress_test.h"
   39 #include "opt_mpath.h"
   40 #include "opt_sctp.h"
   41 
   42 #include <sys/param.h>
   43 #include <sys/systm.h>
   44 #include <sys/kernel.h>
   45 #include <sys/malloc.h>
   46 #include <sys/mbuf.h>
   47 #include <sys/priv.h>
   48 #include <sys/proc.h>
   49 #include <sys/protosw.h>
   50 #include <sys/socket.h>
   51 #include <sys/socketvar.h>
   52 #include <sys/sysctl.h>
   53 #include <sys/ucred.h>
   54 
   55 #include <net/if.h>
   56 #include <net/if_llatbl.h>
   57 #include <net/netisr.h>
   58 #include <net/pfil.h>
   59 #include <net/route.h>
   60 #include <net/flowtable.h>
   61 #ifdef RADIX_MPATH
   62 #include <net/radix_mpath.h>
   63 #endif
   64 #include <net/vnet.h>
   65 
   66 #include <netinet/in.h>
   67 #include <netinet/in_systm.h>
   68 #include <netinet/ip.h>
   69 #include <netinet/in_pcb.h>
   70 #include <netinet/in_var.h>
   71 #include <netinet/ip_var.h>
   72 #include <netinet/ip_options.h>
   73 #ifdef SCTP
   74 #include <netinet/sctp.h>
   75 #include <netinet/sctp_crc32.h>
   76 #endif
   77 
   78 #ifdef IPSEC
   79 #include <netinet/ip_ipsec.h>
   80 #include <netipsec/ipsec.h>
   81 #endif /* IPSEC*/
   82 
   83 #include <machine/in_cksum.h>
   84 
   85 #include <security/mac/mac_framework.h>
   86 
   87 VNET_DEFINE(u_short, ip_id);
   88 
   89 #ifdef MBUF_STRESS_TEST
   90 static int mbuf_frag_size = 0;
   91 SYSCTL_INT(_net_inet_ip, OID_AUTO, mbuf_frag_size, CTLFLAG_RW,
   92         &mbuf_frag_size, 0, "Fragment outgoing mbufs to this size");
   93 #endif
   94 
   95 static void     ip_mloopback
   96         (struct ifnet *, struct mbuf *, struct sockaddr_in *, int);
   97 
   98 
   99 extern int in_mcast_loop;
  100 extern  struct protosw inetsw[];
  101 
  102 /*
  103  * IP output.  The packet in mbuf chain m contains a skeletal IP
  104  * header (with len, off, ttl, proto, tos, src, dst).
  105  * ip_len and ip_off are in host format.
  106  * The mbuf chain containing the packet will be freed.
  107  * The mbuf opt, if present, will not be freed.
  108  * In the IP forwarding case, the packet will arrive with options already
  109  * inserted, so must have a NULL opt pointer.
  110  */
  111 int
  112 ip_output(struct mbuf *m, struct mbuf *opt, struct route *ro, int flags,
  113     struct ip_moptions *imo, struct inpcb *inp)
  114 {
  115         struct ip *ip;
  116         struct ifnet *ifp = NULL;       /* keep compiler happy */
  117         struct mbuf *m0;
  118         int hlen = sizeof (struct ip);
  119         int mtu;
  120         int n;  /* scratchpad */
  121         int error = 0;
  122         int nortfree = 0;
  123         struct sockaddr_in *dst;
  124         struct in_ifaddr *ia;
  125         int isbroadcast, sw_csum;
  126         struct route iproute;
  127         struct rtentry *rte;    /* cache for ro->ro_rt */
  128         struct in_addr odst;
  129 #ifdef IPFIREWALL_FORWARD
  130         struct m_tag *fwd_tag = NULL;
  131 #endif
  132 #ifdef IPSEC
  133         int no_route_but_check_spd = 0;
  134 #endif
  135         M_ASSERTPKTHDR(m);
  136 
  137         if (inp != NULL) {
  138                 INP_LOCK_ASSERT(inp);
  139                 M_SETFIB(m, inp->inp_inc.inc_fibnum);
  140                 if (inp->inp_flags & (INP_HW_FLOWID|INP_SW_FLOWID)) {
  141                         m->m_pkthdr.flowid = inp->inp_flowid;
  142                         m->m_flags |= M_FLOWID;
  143                 }
  144         }
  145 
  146         if (ro == NULL) {
  147                 ro = &iproute;
  148                 bzero(ro, sizeof (*ro));
  149 
  150 #ifdef FLOWTABLE
  151                 {
  152                         struct flentry *fle;
  153                         
  154                         /*
  155                          * The flow table returns route entries valid for up to 30
  156                          * seconds; we rely on the remainder of ip_output() taking no
  157                          * longer than that long for the stability of ro_rt.  The
  158                          * flow ID assignment must have happened before this point.
  159                          */
  160                         if ((fle = flowtable_lookup_mbuf(V_ip_ft, m, AF_INET)) != NULL) {
  161                                 flow_to_route(fle, ro);
  162                                 nortfree = 1;
  163                         }
  164                 }
  165 #endif
  166         }
  167 
  168         if (opt) {
  169                 int len = 0;
  170                 m = ip_insertoptions(m, opt, &len);
  171                 if (len != 0)
  172                         hlen = len; /* ip->ip_hl is updated above */
  173         }
  174         ip = mtod(m, struct ip *);
  175 
  176         /*
  177          * Fill in IP header.  If we are not allowing fragmentation,
  178          * then the ip_id field is meaningless, but we don't set it
  179          * to zero.  Doing so causes various problems when devices along
  180          * the path (routers, load balancers, firewalls, etc.) illegally
  181          * disable DF on our packet.  Note that a 16-bit counter
  182          * will wrap around in less than 10 seconds at 100 Mbit/s on a
  183          * medium with MTU 1500.  See Steven M. Bellovin, "A Technique
  184          * for Counting NATted Hosts", Proc. IMW'02, available at
  185          * <http://www.cs.columbia.edu/~smb/papers/fnat.pdf>.
  186          */
  187         if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) {
  188                 ip->ip_v = IPVERSION;
  189                 ip->ip_hl = hlen >> 2;
  190                 ip->ip_id = ip_newid();
  191                 IPSTAT_INC(ips_localout);
  192         } else {
  193                 /* Header already set, fetch hlen from there */
  194                 hlen = ip->ip_hl << 2;
  195         }
  196 
  197         dst = (struct sockaddr_in *)&ro->ro_dst;
  198 again:
  199         ia = NULL;
  200         /*
  201          * If there is a cached route,
  202          * check that it is to the same destination
  203          * and is still up.  If not, free it and try again.
  204          * The address family should also be checked in case of sharing the
  205          * cache with IPv6.
  206          */
  207         rte = ro->ro_rt;
  208         if (rte && ((rte->rt_flags & RTF_UP) == 0 ||
  209                     rte->rt_ifp == NULL ||
  210                     !RT_LINK_IS_UP(rte->rt_ifp) ||
  211                           dst->sin_family != AF_INET ||
  212                           dst->sin_addr.s_addr != ip->ip_dst.s_addr)) {
  213                 if (!nortfree)
  214                         RTFREE(rte);
  215                 rte = ro->ro_rt = (struct rtentry *)NULL;
  216                 ro->ro_lle = (struct llentry *)NULL;
  217         }
  218 #ifdef IPFIREWALL_FORWARD
  219         if (rte == NULL && fwd_tag == NULL) {
  220 #else
  221         if (rte == NULL) {
  222 #endif
  223                 bzero(dst, sizeof(*dst));
  224                 dst->sin_family = AF_INET;
  225                 dst->sin_len = sizeof(*dst);
  226                 dst->sin_addr = ip->ip_dst;
  227         }
  228         /*
  229          * If routing to interface only, short circuit routing lookup.
  230          * The use of an all-ones broadcast address implies this; an
  231          * interface is specified by the broadcast address of an interface,
  232          * or the destination address of a ptp interface.
  233          */
  234         if (flags & IP_SENDONES) {
  235                 if ((ia = ifatoia(ifa_ifwithbroadaddr(sintosa(dst)))) == NULL &&
  236                     (ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == NULL) {
  237                         IPSTAT_INC(ips_noroute);
  238                         error = ENETUNREACH;
  239                         goto bad;
  240                 }
  241                 ip->ip_dst.s_addr = INADDR_BROADCAST;
  242                 dst->sin_addr = ip->ip_dst;
  243                 ifp = ia->ia_ifp;
  244                 ip->ip_ttl = 1;
  245                 isbroadcast = 1;
  246         } else if (flags & IP_ROUTETOIF) {
  247                 if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == NULL &&
  248                     (ia = ifatoia(ifa_ifwithnet(sintosa(dst), 0))) == NULL) {
  249                         IPSTAT_INC(ips_noroute);
  250                         error = ENETUNREACH;
  251                         goto bad;
  252                 }
  253                 ifp = ia->ia_ifp;
  254                 ip->ip_ttl = 1;
  255                 isbroadcast = in_broadcast(dst->sin_addr, ifp);
  256         } else if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) &&
  257             imo != NULL && imo->imo_multicast_ifp != NULL) {
  258                 /*
  259                  * Bypass the normal routing lookup for multicast
  260                  * packets if the interface is specified.
  261                  */
  262                 ifp = imo->imo_multicast_ifp;
  263                 IFP_TO_IA(ifp, ia);
  264                 isbroadcast = 0;        /* fool gcc */
  265         } else {
  266                 /*
  267                  * We want to do any cloning requested by the link layer,
  268                  * as this is probably required in all cases for correct
  269                  * operation (as it is for ARP).
  270                  */
  271                 if (rte == NULL) {
  272 #ifdef RADIX_MPATH
  273                         rtalloc_mpath_fib(ro,
  274                             ntohl(ip->ip_src.s_addr ^ ip->ip_dst.s_addr),
  275                             inp ? inp->inp_inc.inc_fibnum : M_GETFIB(m));
  276 #else
  277                         in_rtalloc_ign(ro, 0,
  278                             inp ? inp->inp_inc.inc_fibnum : M_GETFIB(m));
  279 #endif
  280                         rte = ro->ro_rt;
  281                 }
  282                 if (rte == NULL ||
  283                     rte->rt_ifp == NULL ||
  284                     !RT_LINK_IS_UP(rte->rt_ifp)) {
  285 #ifdef IPSEC
  286                         /*
  287                          * There is no route for this packet, but it is
  288                          * possible that a matching SPD entry exists.
  289                          */
  290                         no_route_but_check_spd = 1;
  291                         mtu = 0; /* Silence GCC warning. */
  292                         goto sendit;
  293 #endif
  294                         IPSTAT_INC(ips_noroute);
  295                         error = EHOSTUNREACH;
  296                         goto bad;
  297                 }
  298                 ia = ifatoia(rte->rt_ifa);
  299                 ifa_ref(&ia->ia_ifa);
  300                 ifp = rte->rt_ifp;
  301                 rte->rt_rmx.rmx_pksent++;
  302                 if (rte->rt_flags & RTF_GATEWAY)
  303                         dst = (struct sockaddr_in *)rte->rt_gateway;
  304                 if (rte->rt_flags & RTF_HOST)
  305                         isbroadcast = (rte->rt_flags & RTF_BROADCAST);
  306                 else
  307                         isbroadcast = in_broadcast(dst->sin_addr, ifp);
  308         }
  309         /*
  310          * Calculate MTU.  If we have a route that is up, use that,
  311          * otherwise use the interface's MTU.
  312          */
  313         if (rte != NULL && (rte->rt_flags & (RTF_UP|RTF_HOST))) {
  314                 /*
  315                  * This case can happen if the user changed the MTU
  316                  * of an interface after enabling IP on it.  Because
  317                  * most netifs don't keep track of routes pointing to
  318                  * them, there is no way for one to update all its
  319                  * routes when the MTU is changed.
  320                  */
  321                 if (rte->rt_rmx.rmx_mtu > ifp->if_mtu)
  322                         rte->rt_rmx.rmx_mtu = ifp->if_mtu;
  323                 mtu = rte->rt_rmx.rmx_mtu;
  324         } else {
  325                 mtu = ifp->if_mtu;
  326         }
  327         /* Catch a possible divide by zero later. */
  328         KASSERT(mtu > 0, ("%s: mtu %d <= 0, rte=%p (rt_flags=0x%08x) ifp=%p",
  329             __func__, mtu, rte, (rte != NULL) ? rte->rt_flags : 0, ifp));
  330         if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
  331                 m->m_flags |= M_MCAST;
  332                 /*
  333                  * IP destination address is multicast.  Make sure "dst"
  334                  * still points to the address in "ro".  (It may have been
  335                  * changed to point to a gateway address, above.)
  336                  */
  337                 dst = (struct sockaddr_in *)&ro->ro_dst;
  338                 /*
  339                  * See if the caller provided any multicast options
  340                  */
  341                 if (imo != NULL) {
  342                         ip->ip_ttl = imo->imo_multicast_ttl;
  343                         if (imo->imo_multicast_vif != -1)
  344                                 ip->ip_src.s_addr =
  345                                     ip_mcast_src ?
  346                                     ip_mcast_src(imo->imo_multicast_vif) :
  347                                     INADDR_ANY;
  348                 } else
  349                         ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL;
  350                 /*
  351                  * Confirm that the outgoing interface supports multicast.
  352                  */
  353                 if ((imo == NULL) || (imo->imo_multicast_vif == -1)) {
  354                         if ((ifp->if_flags & IFF_MULTICAST) == 0) {
  355                                 IPSTAT_INC(ips_noroute);
  356                                 error = ENETUNREACH;
  357                                 goto bad;
  358                         }
  359                 }
  360                 /*
  361                  * If source address not specified yet, use address
  362                  * of outgoing interface.
  363                  */
  364                 if (ip->ip_src.s_addr == INADDR_ANY) {
  365                         /* Interface may have no addresses. */
  366                         if (ia != NULL)
  367                                 ip->ip_src = IA_SIN(ia)->sin_addr;
  368                 }
  369 
  370                 if ((imo == NULL && in_mcast_loop) ||
  371                     (imo && imo->imo_multicast_loop)) {
  372                         /*
  373                          * Loop back multicast datagram if not expressly
  374                          * forbidden to do so, even if we are not a member
  375                          * of the group; ip_input() will filter it later,
  376                          * thus deferring a hash lookup and mutex acquisition
  377                          * at the expense of a cheap copy using m_copym().
  378                          */
  379                         ip_mloopback(ifp, m, dst, hlen);
  380                 } else {
  381                         /*
  382                          * If we are acting as a multicast router, perform
  383                          * multicast forwarding as if the packet had just
  384                          * arrived on the interface to which we are about
  385                          * to send.  The multicast forwarding function
  386                          * recursively calls this function, using the
  387                          * IP_FORWARDING flag to prevent infinite recursion.
  388                          *
  389                          * Multicasts that are looped back by ip_mloopback(),
  390                          * above, will be forwarded by the ip_input() routine,
  391                          * if necessary.
  392                          */
  393                         if (V_ip_mrouter && (flags & IP_FORWARDING) == 0) {
  394                                 /*
  395                                  * If rsvp daemon is not running, do not
  396                                  * set ip_moptions. This ensures that the packet
  397                                  * is multicast and not just sent down one link
  398                                  * as prescribed by rsvpd.
  399                                  */
  400                                 if (!V_rsvp_on)
  401                                         imo = NULL;
  402                                 if (ip_mforward &&
  403                                     ip_mforward(ip, ifp, m, imo) != 0) {
  404                                         m_freem(m);
  405                                         goto done;
  406                                 }
  407                         }
  408                 }
  409 
  410                 /*
  411                  * Multicasts with a time-to-live of zero may be looped-
  412                  * back, above, but must not be transmitted on a network.
  413                  * Also, multicasts addressed to the loopback interface
  414                  * are not sent -- the above call to ip_mloopback() will
  415                  * loop back a copy. ip_input() will drop the copy if
  416                  * this host does not belong to the destination group on
  417                  * the loopback interface.
  418                  */
  419                 if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) {
  420                         m_freem(m);
  421                         goto done;
  422                 }
  423 
  424                 goto sendit;
  425         }
  426 
  427         /*
  428          * If the source address is not specified yet, use the address
  429          * of the outoing interface.
  430          */
  431         if (ip->ip_src.s_addr == INADDR_ANY) {
  432                 /* Interface may have no addresses. */
  433                 if (ia != NULL) {
  434                         ip->ip_src = IA_SIN(ia)->sin_addr;
  435                 }
  436         }
  437 
  438         /*
  439          * Verify that we have any chance at all of being able to queue the
  440          * packet or packet fragments, unless ALTQ is enabled on the given
  441          * interface in which case packetdrop should be done by queueing.
  442          */
  443         n = ip->ip_len / mtu + 1; /* how many fragments ? */
  444         if (
  445 #ifdef ALTQ
  446             (!ALTQ_IS_ENABLED(&ifp->if_snd)) &&
  447 #endif /* ALTQ */
  448             (ifp->if_snd.ifq_len + n) >= ifp->if_snd.ifq_maxlen ) {
  449                 error = ENOBUFS;
  450                 IPSTAT_INC(ips_odropped);
  451                 ifp->if_snd.ifq_drops += n;
  452                 goto bad;
  453         }
  454 
  455         /*
  456          * Look for broadcast address and
  457          * verify user is allowed to send
  458          * such a packet.
  459          */
  460         if (isbroadcast) {
  461                 if ((ifp->if_flags & IFF_BROADCAST) == 0) {
  462                         error = EADDRNOTAVAIL;
  463                         goto bad;
  464                 }
  465                 if ((flags & IP_ALLOWBROADCAST) == 0) {
  466                         error = EACCES;
  467                         goto bad;
  468                 }
  469                 /* don't allow broadcast messages to be fragmented */
  470                 if (ip->ip_len > mtu) {
  471                         error = EMSGSIZE;
  472                         goto bad;
  473                 }
  474                 m->m_flags |= M_BCAST;
  475         } else {
  476                 m->m_flags &= ~M_BCAST;
  477         }
  478 
  479 sendit:
  480 #ifdef IPSEC
  481         switch(ip_ipsec_output(&m, inp, &flags, &error)) {
  482         case 1:
  483                 goto bad;
  484         case -1:
  485                 goto done;
  486         case 0:
  487         default:
  488                 break;  /* Continue with packet processing. */
  489         }
  490         /*
  491          * Check if there was a route for this packet; return error if not.
  492          */
  493         if (no_route_but_check_spd) {
  494                 IPSTAT_INC(ips_noroute);
  495                 error = EHOSTUNREACH;
  496                 goto bad;
  497         }
  498         /* Update variables that are affected by ipsec4_output(). */
  499         ip = mtod(m, struct ip *);
  500         hlen = ip->ip_hl << 2;
  501 #endif /* IPSEC */
  502 
  503         /* Jump over all PFIL processing if hooks are not active. */
  504         if (!PFIL_HOOKED(&V_inet_pfil_hook))
  505                 goto passout;
  506 
  507         /* Run through list of hooks for output packets. */
  508         odst.s_addr = ip->ip_dst.s_addr;
  509         error = pfil_run_hooks(&V_inet_pfil_hook, &m, ifp, PFIL_OUT, inp);
  510         if (error != 0 || m == NULL)
  511                 goto done;
  512 
  513         ip = mtod(m, struct ip *);
  514 
  515         /* See if destination IP address was changed by packet filter. */
  516         if (odst.s_addr != ip->ip_dst.s_addr) {
  517                 m->m_flags |= M_SKIP_FIREWALL;
  518                 /* If destination is now ourself drop to ip_input(). */
  519                 if (in_localip(ip->ip_dst)) {
  520                         m->m_flags |= M_FASTFWD_OURS;
  521                         if (m->m_pkthdr.rcvif == NULL)
  522                                 m->m_pkthdr.rcvif = V_loif;
  523                         if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
  524                                 m->m_pkthdr.csum_flags |=
  525                                     CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
  526                                 m->m_pkthdr.csum_data = 0xffff;
  527                         }
  528                         m->m_pkthdr.csum_flags |=
  529                             CSUM_IP_CHECKED | CSUM_IP_VALID;
  530 #ifdef SCTP
  531                         if (m->m_pkthdr.csum_flags & CSUM_SCTP)
  532                                 m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID;
  533 #endif
  534                         error = netisr_queue(NETISR_IP, m);
  535                         goto done;
  536                 } else {
  537                         if (ia != NULL)
  538                                 ifa_free(&ia->ia_ifa);
  539                         goto again;     /* Redo the routing table lookup. */
  540                 }
  541         }
  542 
  543 #ifdef IPFIREWALL_FORWARD
  544         /* See if local, if yes, send it to netisr with IP_FASTFWD_OURS. */
  545         if (m->m_flags & M_FASTFWD_OURS) {
  546                 if (m->m_pkthdr.rcvif == NULL)
  547                         m->m_pkthdr.rcvif = V_loif;
  548                 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
  549                         m->m_pkthdr.csum_flags |=
  550                             CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
  551                         m->m_pkthdr.csum_data = 0xffff;
  552                 }
  553 #ifdef SCTP
  554                 if (m->m_pkthdr.csum_flags & CSUM_SCTP)
  555                         m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID;
  556 #endif
  557                 m->m_pkthdr.csum_flags |=
  558                             CSUM_IP_CHECKED | CSUM_IP_VALID;
  559 
  560                 error = netisr_queue(NETISR_IP, m);
  561                 goto done;
  562         }
  563         /* Or forward to some other address? */
  564         fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL);
  565         if (fwd_tag) {
  566                 dst = (struct sockaddr_in *)&ro->ro_dst;
  567                 bcopy((fwd_tag+1), dst, sizeof(struct sockaddr_in));
  568                 m->m_flags |= M_SKIP_FIREWALL;
  569                 m_tag_delete(m, fwd_tag);
  570                 if (ia != NULL)
  571                         ifa_free(&ia->ia_ifa);
  572                 goto again;
  573         }
  574 #endif /* IPFIREWALL_FORWARD */
  575 
  576 passout:
  577         /* 127/8 must not appear on wire - RFC1122. */
  578         if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
  579             (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
  580                 if ((ifp->if_flags & IFF_LOOPBACK) == 0) {
  581                         IPSTAT_INC(ips_badaddr);
  582                         error = EADDRNOTAVAIL;
  583                         goto bad;
  584                 }
  585         }
  586 
  587         m->m_pkthdr.csum_flags |= CSUM_IP;
  588         sw_csum = m->m_pkthdr.csum_flags & ~ifp->if_hwassist;
  589         if (sw_csum & CSUM_DELAY_DATA) {
  590                 in_delayed_cksum(m);
  591                 sw_csum &= ~CSUM_DELAY_DATA;
  592         }
  593 #ifdef SCTP
  594         if (sw_csum & CSUM_SCTP) {
  595                 sctp_delayed_cksum(m, (uint32_t)(ip->ip_hl << 2));
  596                 sw_csum &= ~CSUM_SCTP;
  597         }
  598 #endif
  599         m->m_pkthdr.csum_flags &= ifp->if_hwassist;
  600 
  601         /*
  602          * If small enough for interface, or the interface will take
  603          * care of the fragmentation for us, we can just send directly.
  604          */
  605         if (ip->ip_len <= mtu ||
  606             (m->m_pkthdr.csum_flags & ifp->if_hwassist & CSUM_TSO) != 0 ||
  607             ((ip->ip_off & IP_DF) == 0 && (ifp->if_hwassist & CSUM_FRAGMENT))) {
  608                 ip->ip_len = htons(ip->ip_len);
  609                 ip->ip_off = htons(ip->ip_off);
  610                 ip->ip_sum = 0;
  611                 if (sw_csum & CSUM_DELAY_IP)
  612                         ip->ip_sum = in_cksum(m, hlen);
  613 
  614                 /*
  615                  * Record statistics for this interface address.
  616                  * With CSUM_TSO the byte/packet count will be slightly
  617                  * incorrect because we count the IP+TCP headers only
  618                  * once instead of for every generated packet.
  619                  */
  620                 if (!(flags & IP_FORWARDING) && ia) {
  621                         if (m->m_pkthdr.csum_flags & CSUM_TSO)
  622                                 ia->ia_ifa.if_opackets +=
  623                                     m->m_pkthdr.len / m->m_pkthdr.tso_segsz;
  624                         else
  625                                 ia->ia_ifa.if_opackets++;
  626                         ia->ia_ifa.if_obytes += m->m_pkthdr.len;
  627                 }
  628 #ifdef MBUF_STRESS_TEST
  629                 if (mbuf_frag_size && m->m_pkthdr.len > mbuf_frag_size)
  630                         m = m_fragment(m, M_DONTWAIT, mbuf_frag_size);
  631 #endif
  632                 /*
  633                  * Reset layer specific mbuf flags
  634                  * to avoid confusing lower layers.
  635                  */
  636                 m->m_flags &= ~(M_PROTOFLAGS);
  637                 error = (*ifp->if_output)(ifp, m,
  638                                 (struct sockaddr *)dst, ro);
  639                 goto done;
  640         }
  641 
  642         /* Balk when DF bit is set or the interface didn't support TSO. */
  643         if ((ip->ip_off & IP_DF) || (m->m_pkthdr.csum_flags & CSUM_TSO)) {
  644                 error = EMSGSIZE;
  645                 IPSTAT_INC(ips_cantfrag);
  646                 goto bad;
  647         }
  648 
  649         /*
  650          * Too large for interface; fragment if possible. If successful,
  651          * on return, m will point to a list of packets to be sent.
  652          */
  653         error = ip_fragment(ip, &m, mtu, ifp->if_hwassist, sw_csum);
  654         if (error)
  655                 goto bad;
  656         for (; m; m = m0) {
  657                 m0 = m->m_nextpkt;
  658                 m->m_nextpkt = 0;
  659                 if (error == 0) {
  660                         /* Record statistics for this interface address. */
  661                         if (ia != NULL) {
  662                                 ia->ia_ifa.if_opackets++;
  663                                 ia->ia_ifa.if_obytes += m->m_pkthdr.len;
  664                         }
  665                         /*
  666                          * Reset layer specific mbuf flags
  667                          * to avoid confusing upper layers.
  668                          */
  669                         m->m_flags &= ~(M_PROTOFLAGS);
  670 
  671                         error = (*ifp->if_output)(ifp, m,
  672                             (struct sockaddr *)dst, ro);
  673                 } else
  674                         m_freem(m);
  675         }
  676 
  677         if (error == 0)
  678                 IPSTAT_INC(ips_fragmented);
  679 
  680 done:
  681         if (ro == &iproute && ro->ro_rt && !nortfree) {
  682                 RTFREE(ro->ro_rt);
  683         }
  684         if (ia != NULL)
  685                 ifa_free(&ia->ia_ifa);
  686         return (error);
  687 bad:
  688         m_freem(m);
  689         goto done;
  690 }
  691 
  692 /*
  693  * Create a chain of fragments which fit the given mtu. m_frag points to the
  694  * mbuf to be fragmented; on return it points to the chain with the fragments.
  695  * Return 0 if no error. If error, m_frag may contain a partially built
  696  * chain of fragments that should be freed by the caller.
  697  *
  698  * if_hwassist_flags is the hw offload capabilities (see if_data.ifi_hwassist)
  699  * sw_csum contains the delayed checksums flags (e.g., CSUM_DELAY_IP).
  700  */
  701 int
  702 ip_fragment(struct ip *ip, struct mbuf **m_frag, int mtu,
  703     u_long if_hwassist_flags, int sw_csum)
  704 {
  705         int error = 0;
  706         int hlen = ip->ip_hl << 2;
  707         int len = (mtu - hlen) & ~7;    /* size of payload in each fragment */
  708         int off;
  709         struct mbuf *m0 = *m_frag;      /* the original packet          */
  710         int firstlen;
  711         struct mbuf **mnext;
  712         int nfrags;
  713 
  714         if (ip->ip_off & IP_DF) {       /* Fragmentation not allowed */
  715                 IPSTAT_INC(ips_cantfrag);
  716                 return EMSGSIZE;
  717         }
  718 
  719         /*
  720          * Must be able to put at least 8 bytes per fragment.
  721          */
  722         if (len < 8)
  723                 return EMSGSIZE;
  724 
  725         /*
  726          * If the interface will not calculate checksums on
  727          * fragmented packets, then do it here.
  728          */
  729         if (m0->m_pkthdr.csum_flags & CSUM_DELAY_DATA &&
  730             (if_hwassist_flags & CSUM_IP_FRAGS) == 0) {
  731                 in_delayed_cksum(m0);
  732                 m0->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
  733         }
  734 #ifdef SCTP
  735         if (m0->m_pkthdr.csum_flags & CSUM_SCTP &&
  736             (if_hwassist_flags & CSUM_IP_FRAGS) == 0) {
  737                 sctp_delayed_cksum(m0, hlen);
  738                 m0->m_pkthdr.csum_flags &= ~CSUM_SCTP;
  739         }
  740 #endif
  741         if (len > PAGE_SIZE) {
  742                 /* 
  743                  * Fragment large datagrams such that each segment 
  744                  * contains a multiple of PAGE_SIZE amount of data, 
  745                  * plus headers. This enables a receiver to perform 
  746                  * page-flipping zero-copy optimizations.
  747                  *
  748                  * XXX When does this help given that sender and receiver
  749                  * could have different page sizes, and also mtu could
  750                  * be less than the receiver's page size ?
  751                  */
  752                 int newlen;
  753                 struct mbuf *m;
  754 
  755                 for (m = m0, off = 0; m && (off+m->m_len) <= mtu; m = m->m_next)
  756                         off += m->m_len;
  757 
  758                 /*
  759                  * firstlen (off - hlen) must be aligned on an 
  760                  * 8-byte boundary
  761                  */
  762                 if (off < hlen)
  763                         goto smart_frag_failure;
  764                 off = ((off - hlen) & ~7) + hlen;
  765                 newlen = (~PAGE_MASK) & mtu;
  766                 if ((newlen + sizeof (struct ip)) > mtu) {
  767                         /* we failed, go back the default */
  768 smart_frag_failure:
  769                         newlen = len;
  770                         off = hlen + len;
  771                 }
  772                 len = newlen;
  773 
  774         } else {
  775                 off = hlen + len;
  776         }
  777 
  778         firstlen = off - hlen;
  779         mnext = &m0->m_nextpkt;         /* pointer to next packet */
  780 
  781         /*
  782          * Loop through length of segment after first fragment,
  783          * make new header and copy data of each part and link onto chain.
  784          * Here, m0 is the original packet, m is the fragment being created.
  785          * The fragments are linked off the m_nextpkt of the original
  786          * packet, which after processing serves as the first fragment.
  787          */
  788         for (nfrags = 1; off < ip->ip_len; off += len, nfrags++) {
  789                 struct ip *mhip;        /* ip header on the fragment */
  790                 struct mbuf *m;
  791                 int mhlen = sizeof (struct ip);
  792 
  793                 MGETHDR(m, M_DONTWAIT, MT_DATA);
  794                 if (m == NULL) {
  795                         error = ENOBUFS;
  796                         IPSTAT_INC(ips_odropped);
  797                         goto done;
  798                 }
  799                 m->m_flags |= (m0->m_flags & M_MCAST) | M_FRAG;
  800                 /*
  801                  * In the first mbuf, leave room for the link header, then
  802                  * copy the original IP header including options. The payload
  803                  * goes into an additional mbuf chain returned by m_copym().
  804                  */
  805                 m->m_data += max_linkhdr;
  806                 mhip = mtod(m, struct ip *);
  807                 *mhip = *ip;
  808                 if (hlen > sizeof (struct ip)) {
  809                         mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip);
  810                         mhip->ip_v = IPVERSION;
  811                         mhip->ip_hl = mhlen >> 2;
  812                 }
  813                 m->m_len = mhlen;
  814                 /* XXX do we need to add ip->ip_off below ? */
  815                 mhip->ip_off = ((off - hlen) >> 3) + ip->ip_off;
  816                 if (off + len >= ip->ip_len) {  /* last fragment */
  817                         len = ip->ip_len - off;
  818                         m->m_flags |= M_LASTFRAG;
  819                 } else
  820                         mhip->ip_off |= IP_MF;
  821                 mhip->ip_len = htons((u_short)(len + mhlen));
  822                 m->m_next = m_copym(m0, off, len, M_DONTWAIT);
  823                 if (m->m_next == NULL) {        /* copy failed */
  824                         m_free(m);
  825                         error = ENOBUFS;        /* ??? */
  826                         IPSTAT_INC(ips_odropped);
  827                         goto done;
  828                 }
  829                 m->m_pkthdr.len = mhlen + len;
  830                 m->m_pkthdr.rcvif = NULL;
  831 #ifdef MAC
  832                 mac_netinet_fragment(m0, m);
  833 #endif
  834                 m->m_pkthdr.csum_flags = m0->m_pkthdr.csum_flags;
  835                 mhip->ip_off = htons(mhip->ip_off);
  836                 mhip->ip_sum = 0;
  837                 if (sw_csum & CSUM_DELAY_IP)
  838                         mhip->ip_sum = in_cksum(m, mhlen);
  839                 *mnext = m;
  840                 mnext = &m->m_nextpkt;
  841         }
  842         IPSTAT_ADD(ips_ofragments, nfrags);
  843 
  844         /* set first marker for fragment chain */
  845         m0->m_flags |= M_FIRSTFRAG | M_FRAG;
  846         m0->m_pkthdr.csum_data = nfrags;
  847 
  848         /*
  849          * Update first fragment by trimming what's been copied out
  850          * and updating header.
  851          */
  852         m_adj(m0, hlen + firstlen - ip->ip_len);
  853         m0->m_pkthdr.len = hlen + firstlen;
  854         ip->ip_len = htons((u_short)m0->m_pkthdr.len);
  855         ip->ip_off |= IP_MF;
  856         ip->ip_off = htons(ip->ip_off);
  857         ip->ip_sum = 0;
  858         if (sw_csum & CSUM_DELAY_IP)
  859                 ip->ip_sum = in_cksum(m0, hlen);
  860 
  861 done:
  862         *m_frag = m0;
  863         return error;
  864 }
  865 
  866 void
  867 in_delayed_cksum(struct mbuf *m)
  868 {
  869         struct ip *ip;
  870         u_short csum, offset;
  871 
  872         ip = mtod(m, struct ip *);
  873         offset = ip->ip_hl << 2 ;
  874         csum = in_cksum_skip(m, ip->ip_len, offset);
  875         if (m->m_pkthdr.csum_flags & CSUM_UDP && csum == 0)
  876                 csum = 0xffff;
  877         offset += m->m_pkthdr.csum_data;        /* checksum offset */
  878 
  879         if (offset + sizeof(u_short) > m->m_len) {
  880                 printf("delayed m_pullup, m->len: %d  off: %d  p: %d\n",
  881                     m->m_len, offset, ip->ip_p);
  882                 /*
  883                  * XXX
  884                  * this shouldn't happen, but if it does, the
  885                  * correct behavior may be to insert the checksum
  886                  * in the appropriate next mbuf in the chain.
  887                  */
  888                 return;
  889         }
  890         *(u_short *)(m->m_data + offset) = csum;
  891 }
  892 
  893 /*
  894  * IP socket option processing.
  895  */
  896 int
  897 ip_ctloutput(struct socket *so, struct sockopt *sopt)
  898 {
  899         struct  inpcb *inp = sotoinpcb(so);
  900         int     error, optval;
  901 
  902         error = optval = 0;
  903         if (sopt->sopt_level != IPPROTO_IP) {
  904                 error = EINVAL;
  905 
  906                 if (sopt->sopt_level == SOL_SOCKET &&
  907                     sopt->sopt_dir == SOPT_SET) {
  908                         switch (sopt->sopt_name) {
  909                         case SO_REUSEADDR:
  910                                 INP_WLOCK(inp);
  911                                 if (IN_MULTICAST(ntohl(inp->inp_laddr.s_addr))) {
  912                                         if ((so->so_options &
  913                                             (SO_REUSEADDR | SO_REUSEPORT)) != 0)
  914                                                 inp->inp_flags2 |= INP_REUSEPORT;
  915                                         else
  916                                                 inp->inp_flags2 &= ~INP_REUSEPORT;
  917                                 }
  918                                 INP_WUNLOCK(inp);
  919                                 error = 0;
  920                                 break;
  921                         case SO_REUSEPORT:
  922                                 INP_WLOCK(inp);
  923                                 if ((so->so_options & SO_REUSEPORT) != 0)
  924                                         inp->inp_flags2 |= INP_REUSEPORT;
  925                                 else
  926                                         inp->inp_flags2 &= ~INP_REUSEPORT;
  927                                 INP_WUNLOCK(inp);
  928                                 error = 0;
  929                                 break;
  930                         case SO_SETFIB:
  931                                 INP_WLOCK(inp);
  932                                 inp->inp_inc.inc_fibnum = so->so_fibnum;
  933                                 INP_WUNLOCK(inp);
  934                                 error = 0;
  935                                 break;
  936                         default:
  937                                 break;
  938                         }
  939                 }
  940                 return (error);
  941         }
  942 
  943         switch (sopt->sopt_dir) {
  944         case SOPT_SET:
  945                 switch (sopt->sopt_name) {
  946                 case IP_OPTIONS:
  947 #ifdef notyet
  948                 case IP_RETOPTS:
  949 #endif
  950                 {
  951                         struct mbuf *m;
  952                         if (sopt->sopt_valsize > MLEN) {
  953                                 error = EMSGSIZE;
  954                                 break;
  955                         }
  956                         MGET(m, sopt->sopt_td ? M_WAIT : M_DONTWAIT, MT_DATA);
  957                         if (m == NULL) {
  958                                 error = ENOBUFS;
  959                                 break;
  960                         }
  961                         m->m_len = sopt->sopt_valsize;
  962                         error = sooptcopyin(sopt, mtod(m, char *), m->m_len,
  963                                             m->m_len);
  964                         if (error) {
  965                                 m_free(m);
  966                                 break;
  967                         }
  968                         INP_WLOCK(inp);
  969                         error = ip_pcbopts(inp, sopt->sopt_name, m);
  970                         INP_WUNLOCK(inp);
  971                         return (error);
  972                 }
  973 
  974                 case IP_BINDANY:
  975                         if (sopt->sopt_td != NULL) {
  976                                 error = priv_check(sopt->sopt_td,
  977                                     PRIV_NETINET_BINDANY);
  978                                 if (error)
  979                                         break;
  980                         }
  981                         /* FALLTHROUGH */
  982                 case IP_TOS:
  983                 case IP_TTL:
  984                 case IP_MINTTL:
  985                 case IP_RECVOPTS:
  986                 case IP_RECVRETOPTS:
  987                 case IP_RECVDSTADDR:
  988                 case IP_RECVTTL:
  989                 case IP_RECVIF:
  990                 case IP_FAITH:
  991                 case IP_ONESBCAST:
  992                 case IP_DONTFRAG:
  993                 case IP_RECVTOS:
  994                         error = sooptcopyin(sopt, &optval, sizeof optval,
  995                                             sizeof optval);
  996                         if (error)
  997                                 break;
  998 
  999                         switch (sopt->sopt_name) {
 1000                         case IP_TOS:
 1001                                 inp->inp_ip_tos = optval;
 1002                                 break;
 1003 
 1004                         case IP_TTL:
 1005                                 inp->inp_ip_ttl = optval;
 1006                                 break;
 1007 
 1008                         case IP_MINTTL:
 1009                                 if (optval >= 0 && optval <= MAXTTL)
 1010                                         inp->inp_ip_minttl = optval;
 1011                                 else
 1012                                         error = EINVAL;
 1013                                 break;
 1014 
 1015 #define OPTSET(bit) do {                                                \
 1016         INP_WLOCK(inp);                                                 \
 1017         if (optval)                                                     \
 1018                 inp->inp_flags |= bit;                                  \
 1019         else                                                            \
 1020                 inp->inp_flags &= ~bit;                                 \
 1021         INP_WUNLOCK(inp);                                               \
 1022 } while (0)
 1023 
 1024                         case IP_RECVOPTS:
 1025                                 OPTSET(INP_RECVOPTS);
 1026                                 break;
 1027 
 1028                         case IP_RECVRETOPTS:
 1029                                 OPTSET(INP_RECVRETOPTS);
 1030                                 break;
 1031 
 1032                         case IP_RECVDSTADDR:
 1033                                 OPTSET(INP_RECVDSTADDR);
 1034                                 break;
 1035 
 1036                         case IP_RECVTTL:
 1037                                 OPTSET(INP_RECVTTL);
 1038                                 break;
 1039 
 1040                         case IP_RECVIF:
 1041                                 OPTSET(INP_RECVIF);
 1042                                 break;
 1043 
 1044                         case IP_FAITH:
 1045                                 OPTSET(INP_FAITH);
 1046                                 break;
 1047 
 1048                         case IP_ONESBCAST:
 1049                                 OPTSET(INP_ONESBCAST);
 1050                                 break;
 1051                         case IP_DONTFRAG:
 1052                                 OPTSET(INP_DONTFRAG);
 1053                                 break;
 1054                         case IP_BINDANY:
 1055                                 OPTSET(INP_BINDANY);
 1056                                 break;
 1057                         case IP_RECVTOS:
 1058                                 OPTSET(INP_RECVTOS);
 1059                                 break;
 1060                         }
 1061                         break;
 1062 #undef OPTSET
 1063 
 1064                 /*
 1065                  * Multicast socket options are processed by the in_mcast
 1066                  * module.
 1067                  */
 1068                 case IP_MULTICAST_IF:
 1069                 case IP_MULTICAST_VIF:
 1070                 case IP_MULTICAST_TTL:
 1071                 case IP_MULTICAST_LOOP:
 1072                 case IP_ADD_MEMBERSHIP:
 1073                 case IP_DROP_MEMBERSHIP:
 1074                 case IP_ADD_SOURCE_MEMBERSHIP:
 1075                 case IP_DROP_SOURCE_MEMBERSHIP:
 1076                 case IP_BLOCK_SOURCE:
 1077                 case IP_UNBLOCK_SOURCE:
 1078                 case IP_MSFILTER:
 1079                 case MCAST_JOIN_GROUP:
 1080                 case MCAST_LEAVE_GROUP:
 1081                 case MCAST_JOIN_SOURCE_GROUP:
 1082                 case MCAST_LEAVE_SOURCE_GROUP:
 1083                 case MCAST_BLOCK_SOURCE:
 1084                 case MCAST_UNBLOCK_SOURCE:
 1085                         error = inp_setmoptions(inp, sopt);
 1086                         break;
 1087 
 1088                 case IP_PORTRANGE:
 1089                         error = sooptcopyin(sopt, &optval, sizeof optval,
 1090                                             sizeof optval);
 1091                         if (error)
 1092                                 break;
 1093 
 1094                         INP_WLOCK(inp);
 1095                         switch (optval) {
 1096                         case IP_PORTRANGE_DEFAULT:
 1097                                 inp->inp_flags &= ~(INP_LOWPORT);
 1098                                 inp->inp_flags &= ~(INP_HIGHPORT);
 1099                                 break;
 1100 
 1101                         case IP_PORTRANGE_HIGH:
 1102                                 inp->inp_flags &= ~(INP_LOWPORT);
 1103                                 inp->inp_flags |= INP_HIGHPORT;
 1104                                 break;
 1105 
 1106                         case IP_PORTRANGE_LOW:
 1107                                 inp->inp_flags &= ~(INP_HIGHPORT);
 1108                                 inp->inp_flags |= INP_LOWPORT;
 1109                                 break;
 1110 
 1111                         default:
 1112                                 error = EINVAL;
 1113                                 break;
 1114                         }
 1115                         INP_WUNLOCK(inp);
 1116                         break;
 1117 
 1118 #ifdef IPSEC
 1119                 case IP_IPSEC_POLICY:
 1120                 {
 1121                         caddr_t req;
 1122                         struct mbuf *m;
 1123 
 1124                         if ((error = soopt_getm(sopt, &m)) != 0) /* XXX */
 1125                                 break;
 1126                         if ((error = soopt_mcopyin(sopt, m)) != 0) /* XXX */
 1127                                 break;
 1128                         req = mtod(m, caddr_t);
 1129                         error = ipsec_set_policy(inp, sopt->sopt_name, req,
 1130                             m->m_len, (sopt->sopt_td != NULL) ?
 1131                             sopt->sopt_td->td_ucred : NULL);
 1132                         m_freem(m);
 1133                         break;
 1134                 }
 1135 #endif /* IPSEC */
 1136 
 1137                 default:
 1138                         error = ENOPROTOOPT;
 1139                         break;
 1140                 }
 1141                 break;
 1142 
 1143         case SOPT_GET:
 1144                 switch (sopt->sopt_name) {
 1145                 case IP_OPTIONS:
 1146                 case IP_RETOPTS:
 1147                         if (inp->inp_options)
 1148                                 error = sooptcopyout(sopt, 
 1149                                                      mtod(inp->inp_options,
 1150                                                           char *),
 1151                                                      inp->inp_options->m_len);
 1152                         else
 1153                                 sopt->sopt_valsize = 0;
 1154                         break;
 1155 
 1156                 case IP_TOS:
 1157                 case IP_TTL:
 1158                 case IP_MINTTL:
 1159                 case IP_RECVOPTS:
 1160                 case IP_RECVRETOPTS:
 1161                 case IP_RECVDSTADDR:
 1162                 case IP_RECVTTL:
 1163                 case IP_RECVIF:
 1164                 case IP_PORTRANGE:
 1165                 case IP_FAITH:
 1166                 case IP_ONESBCAST:
 1167                 case IP_DONTFRAG:
 1168                 case IP_BINDANY:
 1169                 case IP_RECVTOS:
 1170                         switch (sopt->sopt_name) {
 1171 
 1172                         case IP_TOS:
 1173                                 optval = inp->inp_ip_tos;
 1174                                 break;
 1175 
 1176                         case IP_TTL:
 1177                                 optval = inp->inp_ip_ttl;
 1178                                 break;
 1179 
 1180                         case IP_MINTTL:
 1181                                 optval = inp->inp_ip_minttl;
 1182                                 break;
 1183 
 1184 #define OPTBIT(bit)     (inp->inp_flags & bit ? 1 : 0)
 1185 
 1186                         case IP_RECVOPTS:
 1187                                 optval = OPTBIT(INP_RECVOPTS);
 1188                                 break;
 1189 
 1190                         case IP_RECVRETOPTS:
 1191                                 optval = OPTBIT(INP_RECVRETOPTS);
 1192                                 break;
 1193 
 1194                         case IP_RECVDSTADDR:
 1195                                 optval = OPTBIT(INP_RECVDSTADDR);
 1196                                 break;
 1197 
 1198                         case IP_RECVTTL:
 1199                                 optval = OPTBIT(INP_RECVTTL);
 1200                                 break;
 1201 
 1202                         case IP_RECVIF:
 1203                                 optval = OPTBIT(INP_RECVIF);
 1204                                 break;
 1205 
 1206                         case IP_PORTRANGE:
 1207                                 if (inp->inp_flags & INP_HIGHPORT)
 1208                                         optval = IP_PORTRANGE_HIGH;
 1209                                 else if (inp->inp_flags & INP_LOWPORT)
 1210                                         optval = IP_PORTRANGE_LOW;
 1211                                 else
 1212                                         optval = 0;
 1213                                 break;
 1214 
 1215                         case IP_FAITH:
 1216                                 optval = OPTBIT(INP_FAITH);
 1217                                 break;
 1218 
 1219                         case IP_ONESBCAST:
 1220                                 optval = OPTBIT(INP_ONESBCAST);
 1221                                 break;
 1222                         case IP_DONTFRAG:
 1223                                 optval = OPTBIT(INP_DONTFRAG);
 1224                                 break;
 1225                         case IP_BINDANY:
 1226                                 optval = OPTBIT(INP_BINDANY);
 1227                                 break;
 1228                         case IP_RECVTOS:
 1229                                 optval = OPTBIT(INP_RECVTOS);
 1230                                 break;
 1231                         }
 1232                         error = sooptcopyout(sopt, &optval, sizeof optval);
 1233                         break;
 1234 
 1235                 /*
 1236                  * Multicast socket options are processed by the in_mcast
 1237                  * module.
 1238                  */
 1239                 case IP_MULTICAST_IF:
 1240                 case IP_MULTICAST_VIF:
 1241                 case IP_MULTICAST_TTL:
 1242                 case IP_MULTICAST_LOOP:
 1243                 case IP_MSFILTER:
 1244                         error = inp_getmoptions(inp, sopt);
 1245                         break;
 1246 
 1247 #ifdef IPSEC
 1248                 case IP_IPSEC_POLICY:
 1249                 {
 1250                         struct mbuf *m = NULL;
 1251                         caddr_t req = NULL;
 1252                         size_t len = 0;
 1253 
 1254                         if (m != 0) {
 1255                                 req = mtod(m, caddr_t);
 1256                                 len = m->m_len;
 1257                         }
 1258                         error = ipsec_get_policy(sotoinpcb(so), req, len, &m);
 1259                         if (error == 0)
 1260                                 error = soopt_mcopyout(sopt, m); /* XXX */
 1261                         if (error == 0)
 1262                                 m_freem(m);
 1263                         break;
 1264                 }
 1265 #endif /* IPSEC */
 1266 
 1267                 default:
 1268                         error = ENOPROTOOPT;
 1269                         break;
 1270                 }
 1271                 break;
 1272         }
 1273         return (error);
 1274 }
 1275 
 1276 /*
 1277  * Routine called from ip_output() to loop back a copy of an IP multicast
 1278  * packet to the input queue of a specified interface.  Note that this
 1279  * calls the output routine of the loopback "driver", but with an interface
 1280  * pointer that might NOT be a loopback interface -- evil, but easier than
 1281  * replicating that code here.
 1282  */
 1283 static void
 1284 ip_mloopback(struct ifnet *ifp, struct mbuf *m, struct sockaddr_in *dst,
 1285     int hlen)
 1286 {
 1287         register struct ip *ip;
 1288         struct mbuf *copym;
 1289 
 1290         /*
 1291          * Make a deep copy of the packet because we're going to
 1292          * modify the pack in order to generate checksums.
 1293          */
 1294         copym = m_dup(m, M_DONTWAIT);
 1295         if (copym != NULL && (copym->m_flags & M_EXT || copym->m_len < hlen))
 1296                 copym = m_pullup(copym, hlen);
 1297         if (copym != NULL) {
 1298                 /* If needed, compute the checksum and mark it as valid. */
 1299                 if (copym->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
 1300                         in_delayed_cksum(copym);
 1301                         copym->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
 1302                         copym->m_pkthdr.csum_flags |=
 1303                             CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
 1304                         copym->m_pkthdr.csum_data = 0xffff;
 1305                 }
 1306                 /*
 1307                  * We don't bother to fragment if the IP length is greater
 1308                  * than the interface's MTU.  Can this possibly matter?
 1309                  */
 1310                 ip = mtod(copym, struct ip *);
 1311                 ip->ip_len = htons(ip->ip_len);
 1312                 ip->ip_off = htons(ip->ip_off);
 1313                 ip->ip_sum = 0;
 1314                 ip->ip_sum = in_cksum(copym, hlen);
 1315 #if 1 /* XXX */
 1316                 if (dst->sin_family != AF_INET) {
 1317                         printf("ip_mloopback: bad address family %d\n",
 1318                                                 dst->sin_family);
 1319                         dst->sin_family = AF_INET;
 1320                 }
 1321 #endif
 1322                 if_simloop(ifp, copym, dst->sin_family, 0);
 1323         }
 1324 }

Cache object: a03ecd3255914c99cf08f873597f7d05


[ 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.