]> git.saurik.com Git - apple/xnu.git/blobdiff - bsd/netinet6/raw_ip6.c
xnu-4570.1.46.tar.gz
[apple/xnu.git] / bsd / netinet6 / raw_ip6.c
index 169b7992dc2cb5d706d54048761461938c43537a..d7c6f689fb2dcad1301e6c6144c98984c742ab25 100644 (file)
@@ -1,8 +1,8 @@
 /*
- * Copyright (c) 2000-2011 Apple Inc. All rights reserved.
+ * Copyright (c) 2000-2016 Apple Inc. All rights reserved.
  *
  * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
- * 
+ *
  * This file contains Original Code and/or Modifications of Original Code
  * as defined in and that are subject to the Apple Public Source License
  * Version 2.0 (the 'License'). You may not use this file except in
  * unlawful or unlicensed copies of an Apple operating system, or to
  * circumvent, violate, or enable the circumvention or violation of, any
  * terms of an Apple operating system software license agreement.
- * 
+ *
  * Please obtain a copy of the License at
  * http://www.opensource.apple.com/apsl/ and read it before using this file.
- * 
+ *
  * The Original Code and all software distributed under the License are
  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
@@ -22,7 +22,7 @@
  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  * Please see the License for the specific language governing rights and
  * limitations under the License.
- * 
+ *
  * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  */
 /*
@@ -93,6 +93,7 @@
 #include <sys/param.h>
 #include <sys/malloc.h>
 #include <sys/proc.h>
+#include <sys/mcache.h>
 #include <sys/mbuf.h>
 #include <sys/socket.h>
 #include <sys/protosw.h>
 #include <sys/systm.h>
 
 #include <net/if.h>
+#include <net/net_api_stats.h>
 #include <net/route.h>
 #include <net/if_types.h>
 
 #include <netinet/in.h>
 #include <netinet/in_var.h>
 #include <netinet/in_systm.h>
+#include <netinet/in_tclass.h>
 #include <netinet/ip6.h>
 #include <netinet6/ip6_var.h>
-#include <netinet6/ip6_mroute.h>
 #include <netinet/icmp6.h>
 #include <netinet/in_pcb.h>
 #include <netinet6/in6_pcb.h>
 #include <netinet6/nd6.h>
 #include <netinet6/ip6protosw.h>
-#if ENABLE_DEFAULT_SCOPE
 #include <netinet6/scope6_var.h>
-#endif
 #include <netinet6/raw_ip6.h>
-#include <netinet6/ip6_fw.h>
 
 #if IPSEC
 #include <netinet6/ipsec.h>
 #include <netinet6/ipsec6.h>
-extern int ipsec_bypass;
 #endif /*IPSEC*/
 
-#define        satosin6(sa)    ((struct sockaddr_in6 *)(sa))
-#define        ifatoia6(ifa)   ((struct in6_ifaddr *)(ifa))
+#if NECP
+#include <net/necp.h>
+#endif
 
 /*
  * Raw interface to IP6 protocol.
@@ -149,29 +148,26 @@ struct rip6stat rip6stat;
 int
 rip6_input(
        struct  mbuf **mp,
-       int     *offp)
+       int     *offp,
+       int     proto)
 {
        struct mbuf *m = *mp;
-       register struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
-       register struct inpcb *in6p;
+       struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
+       struct inpcb *in6p;
        struct inpcb *last = 0;
        struct mbuf *opts = NULL;
        struct sockaddr_in6 rip6src;
-       int proto = ip6->ip6_nxt;
+       int ret;
+       struct ifnet *ifp = m->m_pkthdr.rcvif;
 
-       rip6stat.rip6s_ipackets++;
+       /* Expect 32-bit aligned data pointer on strict-align platforms */
+       MBUF_STRICT_DATA_ALIGNMENT_CHECK_32(m);
 
-#if defined(NFAITH) && 0 < NFAITH
-       if (faithprefix(&ip6->ip6_dst)) {
-               /* XXX send icmp6 host/port unreach? */
-               m_freem(m);
-               return IPPROTO_DONE;
-       }
-#endif
+       rip6stat.rip6s_ipackets++;
 
        init_sin6(&rip6src, m); /* general init */
 
-       lck_rw_lock_shared(ripcbinfo.mtx);
+       lck_rw_lock_shared(ripcbinfo.ipi_lock);
        LIST_FOREACH(in6p, &ripcb, inp_list) {
                if ((in6p->in6p_vflag & INP_IPV6) == 0)
                        continue;
@@ -184,7 +180,11 @@ rip6_input(
                if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
                    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
                        continue;
-               if (in6p->in6p_cksum != -1) {
+
+               if (inp_restricted_recv(in6p, ifp))
+                       continue;
+
+               if (proto == IPPROTO_ICMPV6 || in6p->in6p_cksum != -1) {
                        rip6stat.rip6s_isum++;
                        if (in6_cksum(m, ip6->ip6_nxt, *offp,
                            m->m_pkthdr.len - *offp)) {
@@ -195,22 +195,28 @@ rip6_input(
                if (last) {
                        struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
 
-#if IPSEC
-                       /*
-                        * Check AH/ESP integrity.
-                        */
-                       if (ipsec_bypass == 0 && n && ipsec6_in_reject_so(n, last->inp_socket)) {
-                                       m_freem(n);
-                                       IPSEC_STAT_INCREMENT(ipsec6stat.in_polvio);
-                                       /* do not inject data into pcb */
+#if NECP
+                       if (n && !necp_socket_is_allowed_to_send_recv_v6(in6p, 0, 0,
+                               &ip6->ip6_dst, &ip6->ip6_src, ifp, NULL, NULL)) {
+                               m_freem(n);
+                               /* do not inject data into pcb */
                        } else
-#endif /*IPSEC*/
+#endif /* NECP */
                        if (n) {
-                               if (last->in6p_flags & IN6P_CONTROLOPTS ||
-                                   last->in6p_socket->so_options & SO_TIMESTAMP)
-                                       ip6_savecontrol(last, &opts, ip6, n);
+                               if ((last->in6p_flags & INP_CONTROLOPTS) != 0 ||
+                                   (last->in6p_socket->so_options & SO_TIMESTAMP) != 0 ||
+                                   (last->in6p_socket->so_options & SO_TIMESTAMP_MONOTONIC) != 0) {
+                                       ret = ip6_savecontrol(last, n, &opts);
+                                       if (ret != 0) {
+                                               m_freem(n);
+                                               m_freem(opts);
+                                               last = in6p;
+                                               continue;
+                                       }
+                               }
                                /* strip intermediate headers */
                                m_adj(n, *offp);
+                               so_recv_data_stat(last->in6p_socket, m, 0);
                                if (sbappendaddr(&last->in6p_socket->so_rcv,
                                                (struct sockaddr *)&rip6src,
                                                 n, opts, NULL) == 0) {
@@ -222,24 +228,31 @@ rip6_input(
                }
                last = in6p;
        }
-       lck_rw_done(ripcbinfo.mtx);
-#if IPSEC
-       /*
-        * Check AH/ESP integrity.
-        */
-       if (ipsec_bypass == 0 && last && ipsec6_in_reject_so(m, last->inp_socket)) {
-                       m_freem(m);
-                       IPSEC_STAT_INCREMENT(ipsec6stat.in_polvio);
-                       ip6stat.ip6s_delivered--;
-                       /* do not inject data into pcb */
+
+#if NECP
+       if (last && !necp_socket_is_allowed_to_send_recv_v6(in6p, 0, 0,
+               &ip6->ip6_dst, &ip6->ip6_src, ifp, NULL, NULL)) {
+               m_freem(m);
+               ip6stat.ip6s_delivered--;
+               /* do not inject data into pcb */
        } else
-#endif /*IPSEC*/
+#endif /* NECP */
        if (last) {
-               if (last->in6p_flags & IN6P_CONTROLOPTS ||
-                   last->in6p_socket->so_options & SO_TIMESTAMP)
-                       ip6_savecontrol(last, &opts, ip6, m);
+               if ((last->in6p_flags & INP_CONTROLOPTS) != 0 ||
+                   (last->in6p_socket->so_options & SO_TIMESTAMP) != 0 ||
+                   (last->in6p_socket->so_options & SO_TIMESTAMP_MONOTONIC) != 0) {
+                       ret = ip6_savecontrol(last, m, &opts);
+                       if (ret != 0) {
+                               m_freem(m);
+                               m_freem(opts);
+                               ip6stat.ip6s_delivered--;
+                               goto unlock;
+                       }
+
+               }
                /* strip intermediate headers */
                m_adj(m, *offp);
+               so_recv_data_stat(last->in6p_socket, m, 0);
                if (sbappendaddr(&last->in6p_socket->so_rcv,
                                (struct sockaddr *)&rip6src, m, opts, NULL) == 0) {
                        rip6stat.rip6s_fullsock++;
@@ -259,6 +272,10 @@ rip6_input(
                }
                ip6stat.ip6s_delivered--;
        }
+
+unlock:
+       lck_rw_done(ripcbinfo.ipi_lock);
+
        return IPPROTO_DONE;
 }
 
@@ -266,10 +283,12 @@ void
 rip6_ctlinput(
        int cmd,
        struct sockaddr *sa,
-       void *d)
+       void *d,
+       __unused struct ifnet *ifp)
 {
        struct ip6_hdr *ip6;
        struct mbuf *m;
+       void *cmdarg = NULL;
        int off = 0;
        struct ip6ctlparam *ip6cp = NULL;
        const struct sockaddr_in6 *sa6_src = NULL;
@@ -281,9 +300,10 @@ rip6_ctlinput(
 
        if ((unsigned)cmd >= PRC_NCMDS)
                return;
-       if (PRC_IS_REDIRECT(cmd))
-               notify = in6_rtchange, d = NULL;
-       else if (cmd == PRC_HOSTDEAD)
+       if (PRC_IS_REDIRECT(cmd)) {
+               notify = in6_rtchange;
+               d = NULL;
+       } else if (cmd == PRC_HOSTDEAD)
                d = NULL;
        else if (inet6ctlerrmap[cmd] == 0)
                return;
@@ -294,6 +314,7 @@ rip6_ctlinput(
                m = ip6cp->ip6c_m;
                ip6 = ip6cp->ip6c_ip6;
                off = ip6cp->ip6c_off;
+               cmdarg = ip6cp->ip6c_cmdarg;
                sa6_src = ip6cp->ip6c_src;
        } else {
                m = NULL;
@@ -302,7 +323,7 @@ rip6_ctlinput(
        }
 
        (void) in6_pcbnotify(&ripcbinfo, sa, 0, (const struct sockaddr *)sa6_src,
-                            0, cmd, notify);
+                            0, cmd, cmdarg, notify);
 }
 
 /*
@@ -311,46 +332,80 @@ rip6_ctlinput(
  */
 int
 rip6_output(
-       register struct mbuf *m,
+       struct mbuf *m,
        struct socket *so,
        struct sockaddr_in6 *dstsock,
-       struct mbuf *control)
+       struct mbuf *control,
+       int israw)
 {
        struct in6_addr *dst;
        struct ip6_hdr *ip6;
        struct inpcb *in6p;
        u_int   plen = m->m_pkthdr.len;
        int error = 0;
-       struct ip6_pktopts opt, *optp = 0;
+       struct ip6_pktopts opt, *optp = NULL;
+       struct ip6_moptions *im6o = NULL;
        struct ifnet *oifp = NULL;
        int type = 0, code = 0;         /* for ICMPv6 output statistics only */
-       int priv = 0;
-#if PKT_PRIORITY
-       mbuf_traffic_class_t mtc = MBUF_TC_NONE;
-#endif /* PKT_PRIORITY */
+       int sotc = SO_TC_UNSPEC;
+       int netsvctype = _NET_SERVICE_TYPE_UNSPEC;
+       struct ip6_out_args ip6oa =
+           { IFSCOPE_NONE, { 0 }, IP6OAF_SELECT_SRCIF, 0, 0, 0 };
+       int flags = IPV6_OUTARGS;
 
        in6p = sotoin6pcb(so);
 
-       priv = 0;
-       if (so->so_uid == 0)
-               priv = 1;
+       if (in6p == NULL
+#if NECP
+               || (necp_socket_should_use_flow_divert(in6p))
+#endif /* NECP */
+               ) {
+               if (in6p == NULL)
+                       error = EINVAL;
+               else
+                       error = EPROTOTYPE;
+               goto bad;
+       }
+       if (dstsock != NULL && IN6_IS_ADDR_V4MAPPED(&dstsock->sin6_addr)) {
+               error = EINVAL;
+               goto bad;
+       }
+
+       if (in6p->inp_flags & INP_BOUND_IF) {
+               ip6oa.ip6oa_boundif = in6p->inp_boundifp->if_index;
+               ip6oa.ip6oa_flags |= IP6OAF_BOUND_IF;
+       }
+       if (INP_NO_CELLULAR(in6p))
+               ip6oa.ip6oa_flags |= IP6OAF_NO_CELLULAR;
+       if (INP_NO_EXPENSIVE(in6p))
+               ip6oa.ip6oa_flags |= IP6OAF_NO_EXPENSIVE;
+       if (INP_AWDL_UNRESTRICTED(in6p))
+               ip6oa.ip6oa_flags |= IP6OAF_AWDL_UNRESTRICTED;
+       if (INP_INTCOPROC_ALLOWED(in6p))
+               ip6oa.ip6oa_flags |= IP6OAF_INTCOPROC_ALLOWED;
+
        dst = &dstsock->sin6_addr;
        if (control) {
-#if PKT_PRIORITY
-               mtc = mbuf_traffic_class_from_control(control);
-#endif /* PKT_PRIORITY */
+               sotc = so_tc_from_control(control, &netsvctype);
 
-               if ((error = ip6_setpktoptions(control, &opt, priv, 0)) != 0)
+               if ((error = ip6_setpktopts(control, &opt, NULL,
+                   SOCK_PROTO(so))) != 0)
                        goto bad;
                optp = &opt;
        } else
                optp = in6p->in6p_outputopts;
+       if (sotc == SO_TC_UNSPEC) {
+               sotc = so->so_traffic_class;
+               netsvctype = so->so_netsvctype;
+       }
+       ip6oa.ip6oa_sotc = sotc;
+       ip6oa.ip6oa_netsvctype = netsvctype;
 
        /*
         * For an ICMPv6 packet, we should know its type and code
         * to update statistics.
         */
-       if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
+       if (SOCK_PROTO(so) == IPPROTO_ICMPV6) {
                struct icmp6_hdr *icmp6;
                if (m->m_len < sizeof(struct icmp6_hdr) &&
                    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
@@ -362,7 +417,16 @@ rip6_output(
                code = icmp6->icmp6_code;
        }
 
-       M_PREPEND(m, sizeof(*ip6), M_WAIT);
+       if (in6p->inp_flowhash == 0)
+               in6p->inp_flowhash = inp_calc_flowhash(in6p);
+       /* update flowinfo - RFC 6437 */
+       if (in6p->inp_flow == 0 && in6p->in6p_flags & IN6P_AUTOFLOWLABEL) {
+               in6p->inp_flow &= ~IPV6_FLOWLABEL_MASK;
+               in6p->inp_flow |=
+                   (htonl(in6p->inp_flowhash) & IPV6_FLOWLABEL_MASK);
+       }
+
+       M_PREPEND(m, sizeof(*ip6), M_WAIT, 1);
        if (m == NULL) {
                error = ENOBUFS;
                goto bad;
@@ -374,6 +438,8 @@ rip6_output(
         */
        ip6->ip6_dst = *dst;
 
+       im6o = in6p->in6p_moptions;
+
        /*
         * If the scope of the destination is link-local, embed the interface
         * index in the address.
@@ -382,7 +448,13 @@ rip6_output(
         */
        if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) {
                struct in6_pktinfo *pi;
+               struct ifnet *im6o_multicast_ifp = NULL;
 
+               if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) && im6o != NULL) {
+                       IM6O_LOCK(im6o);
+                       im6o_multicast_ifp = im6o->im6o_multicast_ifp;
+                       IM6O_UNLOCK(im6o);
+               }
                /*
                 * XXX Boundary check is assumed to be already done in
                 * ip6_setpktoptions().
@@ -391,14 +463,16 @@ rip6_output(
                if (optp && (pi = optp->ip6po_pktinfo) && pi->ipi6_ifindex) {
                        ip6->ip6_dst.s6_addr16[1] = htons(pi->ipi6_ifindex);
                        oifp = ifindex2ifnet[pi->ipi6_ifindex];
+                       if (oifp != NULL)
+                               ifnet_reference(oifp);
                } else if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
-                        in6p->in6p_moptions &&
-                        in6p->in6p_moptions->im6o_multicast_ifp) {
-                       oifp = in6p->in6p_moptions->im6o_multicast_ifp;
+                   im6o != NULL && im6o_multicast_ifp != NULL) {
+                       oifp = im6o_multicast_ifp;
+                       ifnet_reference(oifp);
                        ip6->ip6_dst.s6_addr16[1] = htons(oifp->if_index);
                } else if (dstsock->sin6_scope_id) {
-                       /* 
-                        * boundary check 
+                       /*
+                        * boundary check
                         *
                         * Sinced stsock->sin6_scope_id is unsigned, we don't
                         * need to check if it's < 0
@@ -421,14 +495,18 @@ rip6_output(
                struct in6_addr *in6a;
                struct in6_addr storage;
                u_short index = 0;
-               if ((in6a = in6_selectsrc(dstsock, optp,
-                                         in6p->in6p_moptions,
-                                         &in6p->in6p_route,
-                                         &in6p->in6p_laddr,
-                                         &storage, &error)) == 0) {
+
+               if (israw != 0 && optp && optp->ip6po_pktinfo && !IN6_IS_ADDR_UNSPECIFIED(&optp->ip6po_pktinfo->ipi6_addr)) {
+                       in6a = &optp->ip6po_pktinfo->ipi6_addr;
+                       flags |= IPV6_FLAG_NOSRCIFSEL;
+               } else if ((in6a = in6_selectsrc(dstsock, optp, in6p,
+                   &in6p->in6p_route, NULL, &storage, ip6oa.ip6oa_boundif,
+                   &error)) == 0) {
                        if (error == 0)
                                error = EADDRNOTAVAIL;
                        goto bad;
+               } else {
+                       ip6oa.ip6oa_flags |= IP6OAF_BOUND_SRCADDR;
                }
                ip6->ip6_src = *in6a;
                if (in6p->in6p_route.ro_rt != NULL) {
@@ -436,34 +514,37 @@ rip6_output(
                        if (in6p->in6p_route.ro_rt->rt_ifp != NULL)
                                index = in6p->in6p_route.ro_rt->rt_ifp->if_index;
                        RT_UNLOCK(in6p->in6p_route.ro_rt);
+                       if (oifp != NULL)
+                               ifnet_release(oifp);
                        ifnet_head_lock_shared();
                        if (index == 0 || if_index < index) {
                                panic("bad if_index on interface from route");
                        }
                        oifp = ifindex2ifnet[index];
+                       if (oifp != NULL)
+                               ifnet_reference(oifp);
                        ifnet_head_done();
                }
        }
        ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
-               (in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK);
+               (in6p->inp_flow & IPV6_FLOWINFO_MASK);
        ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
                (IPV6_VERSION & IPV6_VERSION_MASK);
        /* ip6_plen will be filled in ip6_output, so not fill it here. */
        ip6->ip6_nxt = in6p->in6p_ip6_nxt;
        ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
 
-       if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
-           in6p->in6p_cksum != -1) {
+       if (SOCK_PROTO(so) == IPPROTO_ICMPV6 || in6p->in6p_cksum != -1) {
                struct mbuf *n;
                int off;
                u_int16_t *p;
 
                /* compute checksum */
-               if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
+               if (SOCK_PROTO(so) == IPPROTO_ICMPV6)
                        off = offsetof(struct icmp6_hdr, icmp6_cksum);
                else
                        off = in6p->in6p_cksum;
-               if (plen < off + 1) {
+               if (plen < (unsigned int)(off + 1)) {
                        error = EINVAL;
                        goto bad;
                }
@@ -476,46 +557,135 @@ rip6_output(
                }
                if (!n)
                        goto bad;
-               p = (u_int16_t *)(mtod(n, caddr_t) + off);
+               p = (u_int16_t *)(void *)(mtod(n, caddr_t) + off);
                *p = 0;
                *p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
        }
 
+#if NECP
+       {
+               necp_kernel_policy_id policy_id;
+               u_int32_t route_rule_id;
+
+               /*
+                * We need a route to perform NECP route rule checks
+                */
+               if (net_qos_policy_restricted != 0 &&
+                   ROUTE_UNUSABLE(&in6p->in6p_route)) {
+                       struct sockaddr_in6 to;
+                       struct sockaddr_in6 from;
+
+                       ROUTE_RELEASE(&in6p->in6p_route);
+
+                       bzero(&from, sizeof(struct sockaddr_in6));
+                       from.sin6_family = AF_INET6;
+                       from.sin6_len = sizeof(struct sockaddr_in6);
+                       from.sin6_addr = ip6->ip6_src;
+
+                       bzero(&to, sizeof(struct sockaddr_in6));
+                       to.sin6_family = AF_INET6;
+                       to.sin6_len = sizeof(struct sockaddr_in6);
+                       to.sin6_addr = ip6->ip6_dst;
+
+                       in6p->in6p_route.ro_dst.sin6_family = AF_INET6;
+                       in6p->in6p_route.ro_dst.sin6_len = sizeof(struct sockaddr_in6);
+                       ((struct sockaddr_in6 *)(void *)&in6p->in6p_route.ro_dst)->sin6_addr =
+                               ip6->ip6_dst;
+
+                       rtalloc_scoped((struct route *)&in6p->in6p_route, ip6oa.ip6oa_boundif);
+
+                       inp_update_necp_policy(in6p, (struct sockaddr *)&from,
+                           (struct sockaddr *)&to, ip6oa.ip6oa_boundif);
+                       in6p->inp_policyresult.results.qos_marking_gencount = 0;
+               }
+
+               if (!necp_socket_is_allowed_to_send_recv_v6(in6p, 0, 0,
+                       &ip6->ip6_src, &ip6->ip6_dst, NULL, &policy_id, &route_rule_id)) {
+                       error = EHOSTUNREACH;
+                       goto bad;
+               }
+
+               necp_mark_packet_from_socket(m, in6p, policy_id, route_rule_id);
+
+               if (net_qos_policy_restricted != 0) {
+                       necp_socket_update_qos_marking(in6p, in6p->in6p_route.ro_rt,
+                           NULL, route_rule_id);
+               }
+       }
+#endif /* NECP */
+       if ((so->so_flags1 & SOF1_QOSMARKING_ALLOWED))
+               ip6oa.ip6oa_flags |= IP6OAF_QOSMARKING_ALLOWED;
+
 #if IPSEC
-       if (ipsec_bypass == 0 && ipsec_setsocket(m, so) != 0) {
+       if (in6p->in6p_sp != NULL && ipsec_setsocket(m, so) != 0) {
                error = ENOBUFS;
                goto bad;
        }
 #endif /*IPSEC*/
 
-       if (in6p->in6p_route.ro_rt != NULL &&
-           in6p->in6p_route.ro_rt->generation_id != route_generation) {
-               rtfree(in6p->in6p_route.ro_rt);
-               in6p->in6p_route.ro_rt = NULL;
+       if (ROUTE_UNUSABLE(&in6p->in6p_route))
+               ROUTE_RELEASE(&in6p->in6p_route);
+
+       if (oifp != NULL) {
+               ifnet_release(oifp);
+               oifp = NULL;
        }
 
-#if PKT_PRIORITY
-       set_traffic_class(m, so, mtc);
-#endif /* PKT_PRIORITY */
+       set_packet_service_class(m, so, sotc, PKT_SCF_IPV6);
+       m->m_pkthdr.pkt_flowsrc = FLOWSRC_INPCB;
+       m->m_pkthdr.pkt_flowid = in6p->inp_flowhash;
+       m->m_pkthdr.pkt_flags |= (PKTF_FLOW_ID | PKTF_FLOW_LOCALSRC |
+           PKTF_FLOW_RAWSOCK);
+       m->m_pkthdr.pkt_proto = in6p->in6p_ip6_nxt;
+
+       if (im6o != NULL)
+               IM6O_ADDREF(im6o);
+
+       error = ip6_output(m, optp, &in6p->in6p_route, flags, im6o,
+           &oifp, &ip6oa);
 
-       error = ip6_output(m, optp, &in6p->in6p_route, 0,
-                          in6p->in6p_moptions, &oifp, 0);
+       if (im6o != NULL)
+               IM6O_REMREF(im6o);
+
+       if (in6p->in6p_route.ro_rt != NULL) {
+               struct rtentry *rt = in6p->in6p_route.ro_rt;
+               struct ifnet *outif;
+
+               if ((rt->rt_flags & RTF_MULTICAST) ||
+                   in6p->in6p_socket == NULL ||
+                   !(in6p->in6p_socket->so_state & SS_ISCONNECTED)) {
+                       rt = NULL;      /* unusable */
+               }
+               /*
+                * Always discard the cached route for unconnected
+                * socket or if it is a multicast route.
+                */
+               if (rt == NULL)
+                       ROUTE_RELEASE(&in6p->in6p_route);
+
+               /*
+                * If this is a connected socket and the destination
+                * route is not multicast, update outif with that of
+                * the route interface index used by IP.
+                */
+               if (rt != NULL &&
+                   (outif = rt->rt_ifp) != in6p->in6p_last_outifp) {
+                       in6p->in6p_last_outifp = outif;
+               }
+       } else {
+               ROUTE_RELEASE(&in6p->in6p_route);
+       }
 
-#if IFNET_ROUTE_REFCNT
        /*
-        * Always discard the cached route for unconnected socket
-        * or if it is a multicast route.
+        * If output interface was cellular/expensive, and this socket is
+        * denied access to it, generate an event.
         */
-       if (in6p->in6p_route.ro_rt != NULL &&
-           ((in6p->in6p_route.ro_rt->rt_flags & RTF_MULTICAST) ||
-           in6p->in6p_socket == NULL ||
-           in6p->in6p_socket->so_state != SS_ISCONNECTED)) {
-               rtfree(in6p->in6p_route.ro_rt);
-               in6p->in6p_route.ro_rt = NULL;
-       }
-#endif /* IFNET_ROUTE_REFCNT */
+       if (error != 0 && (ip6oa.ip6oa_retflags & IP6OARF_IFDENIED) &&
+           (INP_NO_CELLULAR(in6p) || INP_NO_EXPENSIVE(in6p)))
+               soevent(in6p->inp_socket, (SO_FILT_HINT_LOCKED|
+                   SO_FILT_HINT_IFDENIED));
 
-       if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
+       if (SOCK_PROTO(so) == IPPROTO_ICMPV6) {
                if (oifp)
                        icmp6_ifoutstat_inc(oifp, type, code);
                icmp6stat.icp6s_outhist[type]++;
@@ -524,31 +694,24 @@ rip6_output(
 
        goto freectl;
 
- bad:
-       if (m)
+bad:
+       if (m != NULL)
                m_freem(m);
 
- freectl:
-       if (optp == &opt && optp->ip6po_rthdr && optp->ip6po_route.ro_rt) {
-               rtfree(optp->ip6po_route.ro_rt);
-               optp->ip6po_route.ro_rt = NULL;
-       }
-       if (control) {
+freectl:
+       if (optp == &opt && optp->ip6po_rthdr)
+               ROUTE_RELEASE(&optp->ip6po_route);
+
+       if (control != NULL) {
                if (optp == &opt)
-                       ip6_clearpktopts(optp, 0, -1);
+                       ip6_clearpktopts(optp, -1);
                m_freem(control);
        }
+       if (oifp != NULL)
+               ifnet_release(oifp);
        return(error);
 }
 
-#if IPFW2
-__private_extern__ void
-load_ip6fw(void)
-{
-       ip6_fw_init();
-}
-#endif
-
 /*
  * Raw IPv6 socket option processing.
  */
@@ -557,15 +720,17 @@ rip6_ctloutput(
        struct socket *so,
        struct sockopt *sopt)
 {
-       int error;
+       int error, optval;
 
+       /* Allow <SOL_SOCKET,SO_FLUSH> at this level */
        if (sopt->sopt_level == IPPROTO_ICMPV6)
                /*
                 * XXX: is it better to call icmp6_ctloutput() directly
                 * from protosw?
                 */
                return(icmp6_ctloutput(so, sopt));
-       else if (sopt->sopt_level != IPPROTO_IPV6)
+       else if (sopt->sopt_level != IPPROTO_IPV6 &&
+           !(sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_FLUSH))
                return (EINVAL);
 
        error = 0;
@@ -573,29 +738,9 @@ rip6_ctloutput(
        switch (sopt->sopt_dir) {
        case SOPT_GET:
                switch (sopt->sopt_name) {
-#if IPFW2
-               case IPV6_FW_ADD:
-               case IPV6_FW_GET:
-                       if (ip6_fw_ctl_ptr == 0)
-                               load_ip6fw();
-                       if (ip6_fw_ctl_ptr)
-                               error = ip6_fw_ctl_ptr(sopt);
-                       else
-                               error = ENOPROTOOPT;
-                       break;
-#endif
-
-               case MRT6_INIT:
-               case MRT6_DONE:
-               case MRT6_ADD_MIF:
-               case MRT6_DEL_MIF:
-               case MRT6_ADD_MFC:
-               case MRT6_DEL_MFC:
-               case MRT6_PIM:
-#if MROUTING
-                       error = ip6_mrouter_get(so, sopt);
+               case IPV6_CHECKSUM:
+                       error = ip6_raw_ctloutput(so, sopt);
                        break;
-#endif
                default:
                        error = ip6_ctloutput(so, sopt);
                        break;
@@ -604,31 +749,18 @@ rip6_ctloutput(
 
        case SOPT_SET:
                switch (sopt->sopt_name) {
-#if IPFW2
-               case IPV6_FW_ADD:
-               case IPV6_FW_DEL:
-               case IPV6_FW_FLUSH:
-               case IPV6_FW_ZERO:
-                       if (ip6_fw_ctl_ptr == 0)
-                               load_ip6fw();
-                       if (ip6_fw_ctl_ptr)
-                               error = ip6_fw_ctl_ptr(sopt);
-                       else
-                               error = ENOPROTOOPT;
+               case IPV6_CHECKSUM:
+                       error = ip6_raw_ctloutput(so, sopt);
                        break;
-#endif
 
-               case MRT6_INIT:
-               case MRT6_DONE:
-               case MRT6_ADD_MIF:
-               case MRT6_DEL_MIF:
-               case MRT6_ADD_MFC:
-               case MRT6_DEL_MFC:
-               case MRT6_PIM:
-#if MROUTING
-                       error = ip6_mrouter_set(so, sopt);
+               case SO_FLUSH:
+                       if ((error = sooptcopyin(sopt, &optval, sizeof (optval),
+                           sizeof (optval))) != 0)
+                               break;
+
+                       error = inp_flush(sotoinpcb(so), optval);
                        break;
-#endif
+
                default:
                        error = ip6_ctloutput(so, sopt);
                        break;
@@ -679,10 +811,6 @@ rip6_detach(struct socket *so)
        if (inp == 0)
                panic("rip6_detach");
        /* xxx: RSVP */
-#if MROUTING
-       if (so == ip6_mrouter)
-               ip6_mrouter_done();
-#endif
        if (inp->in6p_icmp6filt) {
                FREE(inp->in6p_icmp6filt, M_PCB);
                inp->in6p_icmp6filt = NULL;
@@ -710,56 +838,93 @@ rip6_disconnect(struct socket *so)
 }
 
 static int
-rip6_bind(struct socket *so, struct sockaddr *nam, __unused struct proc *p)
+rip6_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
 {
+#pragma unused(p)
        struct inpcb *inp = sotoinpcb(so);
-       struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
-       struct ifaddr *ia = NULL;
+       struct sockaddr_in6 sin6;
+       struct ifaddr *ifa = NULL;
+       struct ifnet *outif = NULL;
+       int error;
 
-       if (nam->sa_len != sizeof(*addr))
-               return EINVAL;
+       if (inp == NULL
+#if NECP
+               || (necp_socket_should_use_flow_divert(inp))
+#endif /* NECP */
+               )
+               return (inp == NULL ? EINVAL : EPROTOTYPE);
 
-       if (TAILQ_EMPTY(&ifnet_head) || addr->sin6_family != AF_INET6)
-               return EADDRNOTAVAIL;
-#if ENABLE_DEFAULT_SCOPE
-       if (addr->sin6_scope_id == 0) { /* not change if specified  */
-               addr->sin6_scope_id = scope6_addr2default(&addr->sin6_addr);
-       }
-#endif
-       if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
-           (ia = ifa_ifwithaddr((struct sockaddr *)addr)) == 0)
-               return EADDRNOTAVAIL;
-       if (ia &&
-           ((struct in6_ifaddr *)ia)->ia6_flags &
-           (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
-            IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
-           if (ia) ifafree(ia);
-               return(EADDRNOTAVAIL);
+       if (nam->sa_len != sizeof (struct sockaddr_in6))
+               return (EINVAL);
+
+       if (TAILQ_EMPTY(&ifnet_head) || SIN6(nam)->sin6_family != AF_INET6)
+               return (EADDRNOTAVAIL);
+
+       bzero(&sin6, sizeof (sin6));
+       *(&sin6) = *SIN6(nam);
+
+       if ((error = sa6_embedscope(&sin6, ip6_use_defzone)) != 0)
+               return (error);
+
+       /* Sanitize local copy for address searches */
+       sin6.sin6_flowinfo = 0;
+       sin6.sin6_scope_id = 0;
+       sin6.sin6_port = 0;
+
+       if (!IN6_IS_ADDR_UNSPECIFIED(&sin6.sin6_addr) &&
+           (ifa = ifa_ifwithaddr(SA(&sin6))) == 0)
+               return (EADDRNOTAVAIL);
+       if (ifa != NULL) {
+               IFA_LOCK(ifa);
+               if (((struct in6_ifaddr *)ifa)->ia6_flags &
+                   (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
+                    IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
+                       IFA_UNLOCK(ifa);
+                       IFA_REMREF(ifa);
+                       return (EADDRNOTAVAIL);
+               }
+               outif = ifa->ifa_ifp;
+               IFA_UNLOCK(ifa);
+               IFA_REMREF(ifa);
        }
-       if (ia != NULL)
-               ifafree(ia);
-       inp->in6p_laddr = addr->sin6_addr;
-       return 0;
+       inp->in6p_laddr = sin6.sin6_addr;
+       inp->in6p_last_outifp = outif;
+
+       return (0);
 }
 
 static int
 rip6_connect(struct socket *so, struct sockaddr *nam, __unused struct proc *p)
 {
        struct inpcb *inp = sotoinpcb(so);
-       struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
+       struct sockaddr_in6 *addr = (struct sockaddr_in6 *)(void *)nam;
        struct in6_addr *in6a = NULL;
        struct in6_addr storage;
        int error = 0;
 #if ENABLE_DEFAULT_SCOPE
        struct sockaddr_in6 tmp;
 #endif
-
+       unsigned int ifscope;
+       struct ifnet *outif = NULL;
+
+       if (inp == NULL
+#if NECP
+               || (necp_socket_should_use_flow_divert(inp))
+#endif /* NECP */
+               )
+               return (inp == NULL ? EINVAL : EPROTOTYPE);
        if (nam->sa_len != sizeof(*addr))
                return EINVAL;
        if (TAILQ_EMPTY(&ifnet_head))
                return EADDRNOTAVAIL;
        if (addr->sin6_family != AF_INET6)
                return EAFNOSUPPORT;
+
+       if (!(so->so_flags1 & SOF1_CONNECT_COUNTED)) {
+               so->so_flags1 |= SOF1_CONNECT_COUNTED;
+               INC_ATOMIC_INT64_LIM(net_api_stats.nas_socket_inet6_dgram_connected);
+       }
+
 #if ENABLE_DEFAULT_SCOPE
        if (addr->sin6_scope_id == 0) { /* not change if specified  */
                /* avoid overwrites */
@@ -768,14 +933,25 @@ rip6_connect(struct socket *so, struct sockaddr *nam, __unused struct proc *p)
                addr->sin6_scope_id = scope6_addr2default(&addr->sin6_addr);
        }
 #endif
+
+       /* KAME hack: embed scopeid */
+       if (in6_embedscope(&SIN6(nam)->sin6_addr, SIN6(nam), inp, NULL, NULL) != 0)
+               return (EINVAL);
+
+       ifscope = (inp->inp_flags & INP_BOUND_IF) ?
+           inp->inp_boundifp->if_index : IFSCOPE_NONE;
+
        /* Source address selection. XXX: need pcblookup? */
-       in6a = in6_selectsrc(addr, inp->in6p_outputopts,
-                            inp->in6p_moptions, &inp->in6p_route,
-                            &inp->in6p_laddr, &storage, &error);
+       in6a = in6_selectsrc(addr, inp->in6p_outputopts, inp, &inp->in6p_route,
+           NULL, &storage, ifscope, &error);
        if (in6a == NULL)
                return (error ? error : EADDRNOTAVAIL);
        inp->in6p_laddr = *in6a;
        inp->in6p_faddr = addr->sin6_addr;
+       if (inp->in6p_route.ro_rt != NULL)
+               outif = inp->in6p_route.ro_rt->rt_ifp;
+       inp->in6p_last_outifp = outif;
+
        soisconnected(so);
        return 0;
 }
@@ -788,18 +964,32 @@ rip6_shutdown(struct socket *so)
 }
 
 static int
-rip6_send(struct socket *so, __unused int flags, struct mbuf *m, struct sockaddr *nam,
-        struct mbuf *control, __unused struct proc *p)
+rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
+    struct mbuf *control, struct proc *p)
 {
+#pragma unused(flags, p)
        struct inpcb *inp = sotoinpcb(so);
        struct sockaddr_in6 tmp;
-       struct sockaddr_in6 *dst;
+       struct sockaddr_in6 *dst = (struct sockaddr_in6 *)(void *)nam;
+       int error = 0;
+
+       if (inp == NULL
+#if NECP
+               || (necp_socket_should_use_flow_divert(inp))
+#endif /* NECP */
+               ) {
+               if (inp == NULL)
+                       error = EINVAL;
+               else
+                       error = EPROTOTYPE;
+               goto bad;
+       }
 
        /* always copy sockaddr to avoid overwrites */
        if (so->so_state & SS_ISCONNECTED) {
-               if (nam) {
-                       m_freem(m);
-                       return EISCONN;
+               if (nam != NULL) {
+                       error = EISCONN;
+                       goto bad;
                }
                /* XXX */
                bzero(&tmp, sizeof(tmp));
@@ -810,10 +1000,10 @@ rip6_send(struct socket *so, __unused int flags, struct mbuf *m, struct sockaddr
                dst = &tmp;
        } else {
                if (nam == NULL) {
-                       m_freem(m);
-                       return ENOTCONN;
+                       error = ENOTCONN;
+                       goto bad;
                }
-               tmp = *(struct sockaddr_in6 *)nam;
+               tmp = *(struct sockaddr_in6 *)(void *)nam;
                dst = &tmp;
        }
 #if ENABLE_DEFAULT_SCOPE
@@ -821,24 +1011,47 @@ rip6_send(struct socket *so, __unused int flags, struct mbuf *m, struct sockaddr
                dst->sin6_scope_id = scope6_addr2default(&dst->sin6_addr);
        }
 #endif
-       return rip6_output(m, so, dst, control);
+       return (rip6_output(m, so, dst, control, 1));
+
+bad:
+       VERIFY(error != 0);
+
+       if (m != NULL)
+               m_freem(m);
+       if (control != NULL)
+               m_freem(control);
+
+       return (error);
 }
 
 struct pr_usrreqs rip6_usrreqs = {
-       rip6_abort, pru_accept_notsupp, rip6_attach, rip6_bind, rip6_connect,
-       pru_connect2_notsupp, in6_control, rip6_detach, rip6_disconnect,
-       pru_listen_notsupp, in6_setpeeraddr, pru_rcvd_notsupp,
-       pru_rcvoob_notsupp, rip6_send, pru_sense_null, rip6_shutdown,
-       in6_setsockaddr, sosend, soreceive, pru_sopoll_notsupp
+       .pru_abort =            rip6_abort,
+       .pru_attach =           rip6_attach,
+       .pru_bind =             rip6_bind,
+       .pru_connect =          rip6_connect,
+       .pru_control =          in6_control,
+       .pru_detach =           rip6_detach,
+       .pru_disconnect =       rip6_disconnect,
+       .pru_peeraddr =         in6_getpeeraddr,
+       .pru_send =             rip6_send,
+       .pru_shutdown =         rip6_shutdown,
+       .pru_sockaddr =         in6_getsockaddr,
+       .pru_sosend =           sosend,
+       .pru_soreceive =        soreceive,
 };
 
 __private_extern__ struct pr_usrreqs icmp6_dgram_usrreqs = {
-        rip6_abort, pru_accept_notsupp, icmp6_dgram_attach, rip6_bind, rip6_connect,
-        pru_connect2_notsupp, in6_control, rip6_detach, rip6_disconnect,
-        pru_listen_notsupp, in6_setpeeraddr, pru_rcvd_notsupp,
-        pru_rcvoob_notsupp, icmp6_dgram_send, pru_sense_null, rip6_shutdown,
-        in6_setsockaddr, sosend, soreceive, pru_sopoll_notsupp
+       .pru_abort =            rip6_abort,
+       .pru_attach =           icmp6_dgram_attach,
+       .pru_bind =             rip6_bind,
+       .pru_connect =          rip6_connect,
+       .pru_control =          in6_control,
+       .pru_detach =           rip6_detach,
+       .pru_disconnect =       rip6_disconnect,
+       .pru_peeraddr =         in6_getpeeraddr,
+       .pru_send =             icmp6_dgram_send,
+       .pru_shutdown =         rip6_shutdown,
+       .pru_sockaddr =         in6_getsockaddr,
+       .pru_sosend =           sosend,
+       .pru_soreceive =        soreceive,
 };
-
-
-