]> git.saurik.com Git - apple/xnu.git/blobdiff - bsd/netinet/raw_ip.c
xnu-7195.50.7.100.1.tar.gz
[apple/xnu.git] / bsd / netinet / raw_ip.c
index f06517a26748f28f8ef99df3b2d0ca57a1fec3bf..a140e89253d546f4543a4f27de7168d4b4691e89 100644 (file)
@@ -1,8 +1,8 @@
 /*
- * Copyright (c) 2000-2012 Apple Inc. All rights reserved.
+ * Copyright (c) 2000-2020 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@
  */
 /*
 #include <pexpert/pexpert.h>
 
 #include <net/if.h>
+#include <net/net_api_stats.h>
 #include <net/route.h>
+#include <net/content_filter.h>
 
 #define _IP_VHL
 #include <netinet/in.h>
 #include <netinet/in_systm.h>
+#include <netinet/in_tclass.h>
 #include <netinet/ip.h>
 #include <netinet/in_pcb.h>
 #include <netinet/in_var.h>
 #include <netinet/ip_var.h>
-#include <netinet/ip_mroute.h>
 
-#if INET6
 #include <netinet6/in6_pcb.h>
-#endif /* INET6 */
 
-#include <netinet/ip_fw.h>
 
 #if IPSEC
 #include <netinet6/ipsec.h>
 
 #if DUMMYNET
 #include <netinet/ip_dummynet.h>
-#endif
-
-#if CONFIG_MACF_NET
-#include <security/mac_framework.h>
-#endif /* MAC_NET */
+#endif /* DUMMYNET */
 
-int load_ipfw(void);
 int rip_detach(struct socket *);
 int rip_abort(struct socket *);
 int rip_disconnect(struct socket *);
 int rip_bind(struct socket *, struct sockaddr *, struct proc *);
 int rip_connect(struct socket *, struct sockaddr *, struct proc *);
 int rip_shutdown(struct socket *);
-#if IPSEC
-extern int ipsec_bypass;
-#endif
 
-struct inpcbhead ripcb;
-struct inpcbinfo ripcbinfo;
+struct  inpcbhead ripcb;
+struct  inpcbinfo ripcbinfo;
 
-/* control hooks for ipfw and dummynet */
-#if IPFIREWALL
-ip_fw_ctl_t *ip_fw_ctl_ptr;
-#endif /* IPFIREWALL */
+/* control hooks for dummynet */
 #if DUMMYNET
 ip_dn_ctl_t *ip_dn_ctl_ptr;
 #endif /* DUMMYNET */
@@ -139,8 +126,8 @@ ip_dn_ctl_t *ip_dn_ctl_ptr;
 /*
  * Nominal space allocated to a raw ip socket.
  */
-#define        RIPSNDQ         8192
-#define        RIPRCVQ         8192
+#define RIPSNDQ         8192
+#define RIPRCVQ         8192
 
 /*
  * Raw interface to IP protocol.
@@ -150,106 +137,121 @@ ip_dn_ctl_t *ip_dn_ctl_ptr;
  * Initialize raw connection block q.
  */
 void
-rip_init()
+rip_init(struct protosw *pp, struct domain *dp)
 {
-       struct inpcbinfo *pcbinfo;
+#pragma unused(dp)
+       static int rip_initialized = 0;
+       struct inpcbinfo *pcbinfo;
+
+       VERIFY((pp->pr_flags & (PR_INITIALIZED | PR_ATTACHED)) == PR_ATTACHED);
+
+       if (rip_initialized) {
+               return;
+       }
+       rip_initialized = 1;
 
        LIST_INIT(&ripcb);
-       ripcbinfo.listhead = &ripcb;
+       ripcbinfo.ipi_listhead = &ripcb;
        /*
         * XXX We don't use the hash list for raw IP, but it's easier
         * to allocate a one entry hash list than it is to check all
-        * over the place for hashbase == NULL.
+        * over the place for ipi_hashbase == NULL.
         */
-       ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask);
-       ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask);
+       ripcbinfo.ipi_hashbase = hashinit(1, M_PCB, &ripcbinfo.ipi_hashmask);
+       ripcbinfo.ipi_porthashbase = hashinit(1, M_PCB, &ripcbinfo.ipi_porthashmask);
 
-       ripcbinfo.ipi_zone = (void *) zinit(sizeof(struct inpcb),
-                                           (4096 * sizeof(struct inpcb)), 
-                                           4096, "ripzone");
+       ripcbinfo.ipi_zone = zone_create("ripzone", sizeof(struct inpcb),
+           ZC_NONE);
 
        pcbinfo = &ripcbinfo;
-        /*
+       /*
         * allocate lock group attribute and group for udp pcb mutexes
         */
-       pcbinfo->mtx_grp_attr = lck_grp_attr_alloc_init();
+       pcbinfo->ipi_lock_grp_attr = lck_grp_attr_alloc_init();
+       pcbinfo->ipi_lock_grp = lck_grp_alloc_init("ripcb", pcbinfo->ipi_lock_grp_attr);
 
-       pcbinfo->mtx_grp = lck_grp_alloc_init("ripcb", pcbinfo->mtx_grp_attr);
-               
        /*
         * allocate the lock attribute for udp pcb mutexes
         */
-       pcbinfo->mtx_attr = lck_attr_alloc_init();
-
-       if ((pcbinfo->mtx = lck_rw_alloc_init(pcbinfo->mtx_grp, pcbinfo->mtx_attr)) == NULL)
-               return; /* pretty much dead if this fails... */
+       pcbinfo->ipi_lock_attr = lck_attr_alloc_init();
+       if ((pcbinfo->ipi_lock = lck_rw_alloc_init(pcbinfo->ipi_lock_grp,
+           pcbinfo->ipi_lock_attr)) == NULL) {
+               panic("%s: unable to allocate PCB lock\n", __func__);
+               /* NOTREACHED */
+       }
 
+       in_pcbinfo_attach(&ripcbinfo);
 }
 
-static struct  sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET , 0, {0}, {0,0,0,0,0,0,0,0,} };
+static struct   sockaddr_in ripsrc = {
+       .sin_len = sizeof(ripsrc),
+       .sin_family = AF_INET,
+       .sin_port = 0,
+       .sin_addr = { .s_addr = 0 },
+       .sin_zero = {0, 0, 0, 0, 0, 0, 0, 0, }
+};
+
 /*
  * Setup generic address and protocol structures
  * for raw_input routine, then pass them along with
  * mbuf chain.
  */
 void
-rip_input(m, iphlen)
-       struct mbuf *m;
-       int iphlen;
+rip_input(struct mbuf *m, int iphlen)
 {
-       register struct ip *ip = mtod(m, struct ip *);
-       register struct inpcb *inp;
+       struct ip *ip = mtod(m, struct ip *);
+       struct inpcb *inp;
        struct inpcb *last = 0;
        struct mbuf *opts = 0;
        int skipit = 0, ret = 0;
+       struct ifnet *ifp = m->m_pkthdr.rcvif;
 
        /* Expect 32-bit aligned data pointer on strict-align platforms */
        MBUF_STRICT_DATA_ALIGNMENT_CHECK_32(m);
 
        ripsrc.sin_addr = ip->ip_src;
-       lck_rw_lock_shared(ripcbinfo.mtx);
+       lck_rw_lock_shared(ripcbinfo.ipi_lock);
        LIST_FOREACH(inp, &ripcb, inp_list) {
-#if INET6
-               if ((inp->inp_vflag & INP_IPV4) == 0)
+               if ((inp->inp_vflag & INP_IPV4) == 0) {
                        continue;
-#endif
-               if (inp->inp_ip_p && (inp->inp_ip_p != ip->ip_p))
+               }
+               if (inp->inp_ip_p && (inp->inp_ip_p != ip->ip_p)) {
                        continue;
+               }
                if (inp->inp_laddr.s_addr &&
-                  inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
+                   inp->inp_laddr.s_addr != ip->ip_dst.s_addr) {
                        continue;
+               }
                if (inp->inp_faddr.s_addr &&
-                  inp->inp_faddr.s_addr != ip->ip_src.s_addr)
+                   inp->inp_faddr.s_addr != ip->ip_src.s_addr) {
                        continue;
+               }
+               if (inp_restricted_recv(inp, ifp)) {
+                       continue;
+               }
                if (last) {
                        struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
-               
+
                        skipit = 0;
-#if IPSEC
-                       /* check AH/ESP integrity. */
-                       if (ipsec_bypass == 0 && n) {
-                               if (ipsec4_in_reject_so(n, last->inp_socket)) {
-                                       m_freem(n);
-                                       IPSEC_STAT_INCREMENT(ipsecstat.in_polvio);
-                                       /* do not inject data to pcb */
-                                       skipit = 1;
-                               }
-                       } 
-#endif /*IPSEC*/
-#if CONFIG_MACF_NET
-                       if (n && skipit == 0) {
-                               if (mac_inpcb_check_deliver(last, n, AF_INET,
-                                   SOCK_RAW) != 0) {
-                                       m_freem(n);
-                                       skipit = 1;
-                               }
+
+#if NECP
+                       if (n && !necp_socket_is_allowed_to_send_recv_v4(last, 0, 0,
+                           &ip->ip_dst, &ip->ip_src, ifp, 0, NULL, NULL, NULL, NULL)) {
+                               m_freem(n);
+                               /* do not inject data to pcb */
+                               skipit = 1;
                        }
-#endif
+#endif /* NECP */
                        if (n && skipit == 0) {
                                int error = 0;
                                if ((last->inp_flags & INP_CONTROLOPTS) != 0 ||
+#if CONTENT_FILTER
+                                   /* Content Filter needs to see local address */
+                                   (last->inp_socket->so_cfil_db != NULL) ||
+#endif
                                    (last->inp_socket->so_options & SO_TIMESTAMP) != 0 ||
-                                   (last->inp_socket->so_options & SO_TIMESTAMP_MONOTONIC) != 0) {
+                                   (last->inp_socket->so_options & SO_TIMESTAMP_MONOTONIC) != 0 ||
+                                   (last->inp_socket->so_options & SO_TIMESTAMP_CONTINUOUS) != 0) {
                                        ret = ip_savecontrol(last, &opts, ip, n);
                                        if (ret != 0) {
                                                m_freem(n);
@@ -258,7 +260,14 @@ rip_input(m, iphlen)
                                                continue;
                                        }
                                }
-                               if (last->inp_flags & INP_STRIPHDR) {
+                               if (last->inp_flags & INP_STRIPHDR
+#if CONTENT_FILTER
+                                   /*
+                                    * If socket is subject to Content Filter, delay stripping until reinject
+                                    */
+                                   && (last->inp_socket->so_cfil_db == NULL)
+#endif
+                                   ) {
                                        n->m_len -= iphlen;
                                        n->m_pkthdr.len -= iphlen;
                                        n->m_data += iphlen;
@@ -271,7 +280,7 @@ rip_input(m, iphlen)
                                } else {
                                        if (error) {
                                                /* should notify about lost packet */
-                                               kprintf("rip_input can't append to socket\n");
+                                               ipstat.ips_raw_sappend_fail++;
                                        }
                                }
                                opts = 0;
@@ -281,49 +290,50 @@ rip_input(m, iphlen)
        }
 
        skipit = 0;
-#if IPSEC
-       /* check AH/ESP integrity. */
-       if (ipsec_bypass == 0 && last) {
-               if (ipsec4_in_reject_so(m, last->inp_socket)) {
-                       m_freem(m);
-                       IPSEC_STAT_INCREMENT(ipsecstat.in_polvio);
-                       OSAddAtomic(1, &ipstat.ips_delivered);
-                       /* do not inject data to pcb */
-                       skipit = 1;
-               }
-       } 
-#endif /*IPSEC*/
-#if CONFIG_MACF_NET
-       if (last && skipit == 0) {
-               if (mac_inpcb_check_deliver(last, m, AF_INET, SOCK_RAW) != 0) {
-                       skipit = 1;
-                       m_freem(m);
-               }
+#if NECP
+       if (last && !necp_socket_is_allowed_to_send_recv_v4(last, 0, 0,
+           &ip->ip_dst, &ip->ip_src, ifp, 0, NULL, NULL, NULL, NULL)) {
+               m_freem(m);
+               OSAddAtomic(1, &ipstat.ips_delivered);
+               /* do not inject data to pcb */
+               skipit = 1;
        }
-#endif
+#endif /* NECP */
        if (skipit == 0) {
                if (last) {
                        if ((last->inp_flags & INP_CONTROLOPTS) != 0 ||
-                               (last->inp_socket->so_options & SO_TIMESTAMP) != 0 ||
-                               (last->inp_socket->so_options & SO_TIMESTAMP_MONOTONIC) != 0) {
+#if CONTENT_FILTER
+                           /* Content Filter needs to see local address */
+                           (last->inp_socket->so_cfil_db != NULL) ||
+#endif
+                           (last->inp_socket->so_options & SO_TIMESTAMP) != 0 ||
+                           (last->inp_socket->so_options & SO_TIMESTAMP_MONOTONIC) != 0 ||
+                           (last->inp_socket->so_options & SO_TIMESTAMP_CONTINUOUS) != 0) {
                                ret = ip_savecontrol(last, &opts, ip, m);
                                if (ret != 0) {
                                        m_freem(m);
                                        m_freem(opts);
-                                       goto unlock;            
+                                       goto unlock;
                                }
                        }
-                       if (last->inp_flags & INP_STRIPHDR) {
+                       if (last->inp_flags & INP_STRIPHDR
+#if CONTENT_FILTER
+                           /*
+                            * If socket is subject to Content Filter, delay stripping until reinject
+                            */
+                           && (last->inp_socket->so_cfil_db == NULL)
+#endif
+                           ) {
                                m->m_len -= iphlen;
                                m->m_pkthdr.len -= iphlen;
                                m->m_data += iphlen;
                        }
                        so_recv_data_stat(last->inp_socket, m, 0);
                        if (sbappendaddr(&last->inp_socket->so_rcv,
-                               (struct sockaddr *)&ripsrc, m, opts, NULL) != 0) {
+                           (struct sockaddr *)&ripsrc, m, opts, NULL) != 0) {
                                sorwakeup(last->inp_socket);
                        } else {
-                               kprintf("rip_input(2) can't append to socket\n");
+                               ipstat.ips_raw_sappend_fail++;
                        }
                } else {
                        m_freem(m);
@@ -333,10 +343,10 @@ rip_input(m, iphlen)
        }
 unlock:
        /*
-        * Keep the list locked because socket filter may force the socket lock 
+        * Keep the list locked because socket filter may force the socket lock
         * to be released when calling sbappendaddr() -- see rdar://7627704
         */
-       lck_rw_done(ripcbinfo.mtx);
+       lck_rw_done(ripcbinfo.ipi_lock);
 }
 
 /*
@@ -350,18 +360,108 @@ rip_output(
        u_int32_t dst,
        struct mbuf *control)
 {
-       register struct ip *ip;
-       register struct inpcb *inp = sotoinpcb(so);
+       struct ip *ip;
+       struct inpcb *inp = sotoinpcb(so);
        int flags = (so->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST;
-       struct ip_out_args ipoa = { IFSCOPE_NONE, { 0 }, IPOAF_SELECT_SRCIF };
+       int inp_flags = inp ? inp->inp_flags : 0;
+       struct ip_out_args ipoa;
        struct ip_moptions *imo;
+       int tos = IPTOS_UNSPEC;
        int error = 0;
-       mbuf_svc_class_t msc = MBUF_SC_UNSPEC;
+#if CONTENT_FILTER
+       struct m_tag *cfil_tag = NULL;
+       bool cfil_faddr_use = false;
+       uint32_t cfil_so_state_change_cnt = 0;
+       uint32_t cfil_so_options = 0;
+       int cfil_inp_flags = 0;
+       struct sockaddr *cfil_faddr = NULL;
+       struct sockaddr_in *cfil_sin;
+#endif
+
+#if CONTENT_FILTER
+       /*
+        * If socket is subject to Content Filter and no addr is passed in,
+        * retrieve CFIL saved state from mbuf and use it if necessary.
+        */
+       if (so->so_cfil_db && dst == INADDR_ANY) {
+               cfil_tag = cfil_dgram_get_socket_state(m, &cfil_so_state_change_cnt, &cfil_so_options, &cfil_faddr, &cfil_inp_flags);
+               if (cfil_tag) {
+                       cfil_sin = SIN(cfil_faddr);
+                       flags = (cfil_so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST;
+                       inp_flags = cfil_inp_flags;
+                       if (inp && inp->inp_faddr.s_addr == INADDR_ANY) {
+                               /*
+                                * Socket is unconnected, simply use the saved faddr as 'addr' to go through
+                                * the connect/disconnect logic.
+                                */
+                               dst = cfil_sin->sin_addr.s_addr;
+                       } else if ((so->so_state_change_cnt != cfil_so_state_change_cnt) &&
+                           (inp->inp_fport != cfil_sin->sin_port ||
+                           inp->inp_faddr.s_addr != cfil_sin->sin_addr.s_addr)) {
+                               /*
+                                * Socket is connected but socket state and dest addr/port changed.
+                                * We need to use the saved faddr and socket options.
+                                */
+                               cfil_faddr_use = true;
+                       }
+                       m_tag_free(cfil_tag);
+               }
+       }
+#endif
+
+       if (so->so_state & SS_ISCONNECTED) {
+               if (dst != INADDR_ANY) {
+                       if (m != NULL) {
+                               m_freem(m);
+                       }
+                       if (control != NULL) {
+                               m_freem(control);
+                       }
+                       return EISCONN;
+               }
+               dst = cfil_faddr_use ? cfil_sin->sin_addr.s_addr : inp->inp_faddr.s_addr;
+       } else {
+               if (dst == INADDR_ANY) {
+                       if (m != NULL) {
+                               m_freem(m);
+                       }
+                       if (control != NULL) {
+                               m_freem(control);
+                       }
+                       return ENOTCONN;
+               }
+       }
+
+       bzero(&ipoa, sizeof(ipoa));
+       ipoa.ipoa_boundif = IFSCOPE_NONE;
+       ipoa.ipoa_flags = IPOAF_SELECT_SRCIF;
+
+       int sotc = SO_TC_UNSPEC;
+       int netsvctype = _NET_SERVICE_TYPE_UNSPEC;
+
 
        if (control != NULL) {
-               msc = mbuf_service_class_from_control(control);
+               tos = so_tos_from_control(control);
+               sotc = so_tc_from_control(control, &netsvctype);
 
                m_freem(control);
+               control = NULL;
+       }
+       if (sotc == SO_TC_UNSPEC) {
+               sotc = so->so_traffic_class;
+               netsvctype = so->so_netsvctype;
+       }
+
+       if (inp == NULL
+#if NECP
+           || (necp_socket_should_use_flow_divert(inp))
+#endif /* NECP */
+           ) {
+               if (m != NULL) {
+                       m_freem(m);
+               }
+               VERIFY(control == NULL);
+               return inp == NULL ? EINVAL : EPROTOTYPE;
        }
 
        flags |= IP_OUTARGS;
@@ -370,158 +470,252 @@ rip_output(
                ipoa.ipoa_boundif = inp->inp_boundifp->if_index;
                ipoa.ipoa_flags |= IPOAF_BOUND_IF;
        }
-       if (inp->inp_flags & INP_NO_IFT_CELLULAR)
+       if (INP_NO_CELLULAR(inp)) {
                ipoa.ipoa_flags |=  IPOAF_NO_CELLULAR;
+       }
+       if (INP_NO_EXPENSIVE(inp)) {
+               ipoa.ipoa_flags |=  IPOAF_NO_EXPENSIVE;
+       }
+       if (INP_NO_CONSTRAINED(inp)) {
+               ipoa.ipoa_flags |=  IPOAF_NO_CONSTRAINED;
+       }
+       if (INP_AWDL_UNRESTRICTED(inp)) {
+               ipoa.ipoa_flags |=  IPOAF_AWDL_UNRESTRICTED;
+       }
+       ipoa.ipoa_sotc = sotc;
+       ipoa.ipoa_netsvctype = netsvctype;
 
-       if (inp->inp_flowhash == 0)
+       if (inp->inp_flowhash == 0) {
                inp->inp_flowhash = inp_calc_flowhash(inp);
+       }
 
        /*
         * If the user handed us a complete IP packet, use it.
         * Otherwise, allocate an mbuf for a header and fill it in.
         */
-       if ((inp->inp_flags & INP_HDRINCL) == 0) {
+       if ((inp_flags & INP_HDRINCL) == 0) {
                if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
                        m_freem(m);
-                       return(EMSGSIZE);
+                       return EMSGSIZE;
                }
-               M_PREPEND(m, sizeof(struct ip), M_WAIT);
-               if (m == NULL)
+               M_PREPEND(m, sizeof(struct ip), M_WAIT, 1);
+               if (m == NULL) {
                        return ENOBUFS;
+               }
                ip = mtod(m, struct ip *);
-               ip->ip_tos = inp->inp_ip_tos;
-               ip->ip_off = 0;
+               if (tos != IPTOS_UNSPEC) {
+                       ip->ip_tos = (uint8_t)(tos & IPTOS_MASK);
+               } else {
+                       ip->ip_tos = inp->inp_ip_tos;
+               }
+               if (inp->inp_flags2 & INP2_DONTFRAG) {
+                       ip->ip_off = IP_DF;
+               } else {
+                       ip->ip_off = 0;
+               }
                ip->ip_p = inp->inp_ip_p;
-               ip->ip_len = m->m_pkthdr.len;
+               ip->ip_len = (uint16_t)m->m_pkthdr.len;
                ip->ip_src = inp->inp_laddr;
                ip->ip_dst.s_addr = dst;
                ip->ip_ttl = inp->inp_ip_ttl;
        } else {
                if (m->m_pkthdr.len > IP_MAXPACKET) {
                        m_freem(m);
-                       return(EMSGSIZE);
+                       return EMSGSIZE;
                }
                ip = mtod(m, struct ip *);
                /* don't allow both user specified and setsockopt options,
-                  and don't allow packet length sizes that will crash */
-               if (((IP_VHL_HL(ip->ip_vhl) != (sizeof (*ip) >> 2))
-                    && inp->inp_options)
+                *  and don't allow packet length sizes that will crash */
+               if (((IP_VHL_HL(ip->ip_vhl) != (sizeof(*ip) >> 2))
+                   && inp->inp_options)
                    || (ip->ip_len > m->m_pkthdr.len)
                    || (ip->ip_len < (IP_VHL_HL(ip->ip_vhl) << 2))) {
                        m_freem(m);
                        return EINVAL;
                }
-               if (ip->ip_id == 0)
-#if RANDOM_IP_ID
+               if (ip->ip_id == 0 && !(rfc6864 && IP_OFF_IS_ATOMIC(ntohs(ip->ip_off)))) {
                        ip->ip_id = ip_randomid();
-#else
-                       ip->ip_id = htons(ip_id++);
-#endif
+               }
                /* XXX prevent ip_output from overwriting header fields */
                flags |= IP_RAWOUTPUT;
                OSAddAtomic(1, &ipstat.ips_rawout);
        }
 
-       if (inp->inp_laddr.s_addr != INADDR_ANY)
+       if (inp->inp_laddr.s_addr != INADDR_ANY) {
                ipoa.ipoa_flags |= IPOAF_BOUND_SRCADDR;
+       }
+
+#if NECP
+       {
+               necp_kernel_policy_id policy_id;
+               necp_kernel_policy_id skip_policy_id;
+               u_int32_t route_rule_id;
+               u_int32_t pass_flags;
+
+               /*
+                * We need a route to perform NECP route rule checks
+                */
+               if ((net_qos_policy_restricted != 0 &&
+                   ROUTE_UNUSABLE(&inp->inp_route))
+#if CONTENT_FILTER
+                   || cfil_faddr_use
+#endif
+                   ) {
+                       struct sockaddr_in to;
+                       struct sockaddr_in from;
+                       struct in_addr laddr = ip->ip_src;
+
+                       ROUTE_RELEASE(&inp->inp_route);
+
+                       bzero(&from, sizeof(struct sockaddr_in));
+                       from.sin_family = AF_INET;
+                       from.sin_len = sizeof(struct sockaddr_in);
+                       from.sin_addr = laddr;
+
+                       bzero(&to, sizeof(struct sockaddr_in));
+                       to.sin_family = AF_INET;
+                       to.sin_len = sizeof(struct sockaddr_in);
+                       to.sin_addr.s_addr = ip->ip_dst.s_addr;
+
+                       if ((error = in_pcbladdr(inp, (struct sockaddr *)&to,
+                           &laddr, ipoa.ipoa_boundif, NULL, 1)) != 0) {
+                               printf("%s in_pcbladdr(%p) error %d\n",
+                                   __func__, inp, error);
+                               m_freem(m);
+                               return error;
+                       }
+
+                       inp_update_necp_policy(inp, (struct sockaddr *)&from,
+                           (struct sockaddr *)&to, ipoa.ipoa_boundif);
+                       inp->inp_policyresult.results.qos_marking_gencount = 0;
+               }
+
+               if (!necp_socket_is_allowed_to_send_recv_v4(inp, 0, 0,
+                   &ip->ip_src, &ip->ip_dst, NULL, 0, &policy_id, &route_rule_id, &skip_policy_id, &pass_flags)) {
+                       m_freem(m);
+                       return EHOSTUNREACH;
+               }
+
+               necp_mark_packet_from_socket(m, inp, policy_id, route_rule_id, skip_policy_id, pass_flags);
+
+               if (net_qos_policy_restricted != 0) {
+                       struct ifnet *rt_ifp = NULL;
+
+                       if (inp->inp_route.ro_rt != NULL) {
+                               rt_ifp = inp->inp_route.ro_rt->rt_ifp;
+                       }
+
+                       necp_socket_update_qos_marking(inp, inp->inp_route.ro_rt, route_rule_id);
+               }
+       }
+#endif /* NECP */
+       if ((so->so_flags1 & SOF1_QOSMARKING_ALLOWED)) {
+               ipoa.ipoa_flags |= IPOAF_QOSMARKING_ALLOWED;
+       }
 
 #if IPSEC
-       if (ipsec_bypass == 0 && ipsec_setsocket(m, so) != 0) {
+       if (inp->inp_sp != NULL && ipsec_setsocket(m, so) != 0) {
                m_freem(m);
                return ENOBUFS;
        }
 #endif /*IPSEC*/
 
-       if (inp->inp_route.ro_rt != NULL &&
-           inp->inp_route.ro_rt->generation_id != route_generation) {
-               rtfree(inp->inp_route.ro_rt);
-               inp->inp_route.ro_rt = NULL;
+       if (ROUTE_UNUSABLE(&inp->inp_route)) {
+               ROUTE_RELEASE(&inp->inp_route);
        }
 
-       set_packet_service_class(m, so, msc, 0);
-       m->m_pkthdr.m_flowhash = inp->inp_flowhash;
-       m->m_pkthdr.m_fhflags |= PF_TAG_FLOWHASH;
-
-#if CONFIG_MACF_NET
-       mac_mbuf_label_associate_inpcb(inp, m);
-#endif
+       set_packet_service_class(m, so, sotc, 0);
+       m->m_pkthdr.pkt_flowsrc = FLOWSRC_INPCB;
+       m->m_pkthdr.pkt_flowid = inp->inp_flowhash;
+       m->m_pkthdr.pkt_flags |= (PKTF_FLOW_ID | PKTF_FLOW_LOCALSRC |
+           PKTF_FLOW_RAWSOCK);
+       m->m_pkthdr.pkt_proto = inp->inp_ip_p;
+       m->m_pkthdr.tx_rawip_pid = so->last_pid;
+       m->m_pkthdr.tx_rawip_e_pid = so->e_pid;
+       if (so->so_flags & SOF_DELEGATED) {
+               m->m_pkthdr.tx_rawip_e_pid = so->e_pid;
+       } else {
+               m->m_pkthdr.tx_rawip_e_pid = 0;
+       }
 
        imo = inp->inp_moptions;
-       if (imo != NULL)
+       if (imo != NULL) {
                IMO_ADDREF(imo);
+       }
        /*
         * The domain lock is held across ip_output, so it is okay
         * to pass the PCB cached route pointer directly to IP and
         * the modules beneath it.
         */
+       // TODO: PASS DOWN ROUTE RULE ID
        error = ip_output(m, inp->inp_options, &inp->inp_route, flags,
            imo, &ipoa);
 
-       if (imo != NULL)
+       if (imo != NULL) {
                IMO_REMREF(imo);
+       }
 
        if (inp->inp_route.ro_rt != NULL) {
                struct rtentry *rt = inp->inp_route.ro_rt;
                struct ifnet *outif;
 
-               if ((rt->rt_flags & (RTF_MULTICAST|RTF_BROADCAST)) ||
+               if ((rt->rt_flags & (RTF_MULTICAST | RTF_BROADCAST)) ||
                    inp->inp_socket == NULL ||
+#if CONTENT_FILTER
+                   /* Discard temporary route for cfil case */
+                   cfil_faddr_use ||
+#endif
                    !(inp->inp_socket->so_state & SS_ISCONNECTED)) {
-                       rt = NULL;      /* unusable */
+                       rt = NULL;      /* unusable */
                }
                /*
                 * Always discard the cached route for unconnected
                 * socket or if it is a multicast route.
                 */
                if (rt == NULL) {
-                       rtfree(inp->inp_route.ro_rt);
-                       inp->inp_route.ro_rt = NULL;
+                       ROUTE_RELEASE(&inp->inp_route);
                }
+
                /*
                 * If this is a connected socket and the destination
                 * route is unicast, update outif with that of the
                 * route interface used by IP.
                 */
-               if (rt != NULL && (outif = rt->rt_ifp) != inp->inp_last_outifp)
+               if (rt != NULL &&
+                   (outif = rt->rt_ifp) != inp->inp_last_outifp) {
                        inp->inp_last_outifp = outif;
+               }
+       } else {
+               ROUTE_RELEASE(&inp->inp_route);
        }
 
-       return (error);
-}
+       /*
+        * If output interface was cellular/expensive/constrained, and this socket is
+        * denied access to it, generate an event.
+        */
+       if (error != 0 && (ipoa.ipoa_retflags & IPOARF_IFDENIED) &&
+           (INP_NO_CELLULAR(inp) || INP_NO_EXPENSIVE(inp) || INP_NO_CONSTRAINED(inp))) {
+               soevent(so, (SO_FILT_HINT_LOCKED | SO_FILT_HINT_IFDENIED));
+       }
 
-#if IPFIREWALL
-int
-load_ipfw(void)
-{
-       kern_return_t   err;
-       
-       ipfw_init();
-       
-#if DUMMYNET
-       if (!DUMMYNET_LOADED)
-               ip_dn_init();
-#endif /* DUMMYNET */
-       err = 0;
-       
-       return err == 0 && ip_fw_ctl_ptr == NULL ? -1 : err;
+       return error;
 }
-#endif /* IPFIREWALL */
+
 
 /*
  * Raw IP socket option processing.
  */
 int
-rip_ctloutput(so, sopt)
-       struct socket *so;
-       struct sockopt *sopt;
+rip_ctloutput(struct socket *so, struct sockopt *sopt)
 {
-       struct  inpcb *inp = sotoinpcb(so);
-       int     error, optval;
+       struct  inpcb *inp = sotoinpcb(so);
+       int     error, optval;
 
        /* Allow <SOL_SOCKET,SO_FLUSH> at this level */
        if (sopt->sopt_level != IPPROTO_IP &&
-           !(sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_FLUSH))
-               return (EINVAL);
+           !(sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_FLUSH)) {
+               return EINVAL;
+       }
 
        error = 0;
 
@@ -538,43 +732,19 @@ rip_ctloutput(so, sopt)
                        error = sooptcopyout(sopt, &optval, sizeof optval);
                        break;
 
-#if IPFIREWALL
-               case IP_FW_ADD:
-               case IP_FW_GET:
-               case IP_OLD_FW_ADD:
-               case IP_OLD_FW_GET:
-                       if (ip_fw_ctl_ptr == 0)
-                               error = load_ipfw();
-                       if (ip_fw_ctl_ptr && error == 0)
-                               error = ip_fw_ctl_ptr(sopt);
-                       else
-                               error = ENOPROTOOPT;
-                       break;
-#endif /* IPFIREWALL */
 
 #if DUMMYNET
                case IP_DUMMYNET_GET:
-                       if (!DUMMYNET_LOADED)
+                       if (!DUMMYNET_LOADED) {
                                ip_dn_init();
-                       if (DUMMYNET_LOADED)
+                       }
+                       if (DUMMYNET_LOADED) {
                                error = ip_dn_ctl_ptr(sopt);
-                       else
+                       } else {
                                error = ENOPROTOOPT;
-                       break ;
-#endif /* DUMMYNET */
-
-#if MROUTING
-               case MRT_INIT:
-               case MRT_DONE:
-               case MRT_ADD_VIF:
-               case MRT_DEL_VIF:
-               case MRT_ADD_MFC:
-               case MRT_DEL_MFC:
-               case MRT_VERSION:
-               case MRT_ASSERT:
-                       error = ip_mrouter_get(so, sopt);
+                       }
                        break;
-#endif /* MROUTING */
+#endif /* DUMMYNET */
 
                default:
                        error = ip_ctloutput(so, sopt);
@@ -586,93 +756,51 @@ rip_ctloutput(so, sopt)
                switch (sopt->sopt_name) {
                case IP_HDRINCL:
                        error = sooptcopyin(sopt, &optval, sizeof optval,
-                                           sizeof optval);
-                       if (error)
+                           sizeof optval);
+                       if (error) {
                                break;
-                       if (optval)
+                       }
+                       if (optval) {
                                inp->inp_flags |= INP_HDRINCL;
-                       else
+                       } else {
                                inp->inp_flags &= ~INP_HDRINCL;
+                       }
                        break;
 
                case IP_STRIPHDR:
                        error = sooptcopyin(sopt, &optval, sizeof optval,
                            sizeof optval);
-                       if (error)
+                       if (error) {
                                break;
-                       if (optval)
+                       }
+                       if (optval) {
                                inp->inp_flags |= INP_STRIPHDR;
-                       else
+                       } else {
                                inp->inp_flags &= ~INP_STRIPHDR;
+                       }
                        break;
 
-#if IPFIREWALL
-               case IP_FW_ADD:
-               case IP_FW_DEL:
-               case IP_FW_FLUSH:
-               case IP_FW_ZERO:
-               case IP_FW_RESETLOG:
-               case IP_OLD_FW_ADD:
-               case IP_OLD_FW_DEL:
-               case IP_OLD_FW_FLUSH:
-               case IP_OLD_FW_ZERO:
-               case IP_OLD_FW_RESETLOG:
-                       if (ip_fw_ctl_ptr == 0)
-                               error = load_ipfw();
-                       if (ip_fw_ctl_ptr && error == 0)
-                               error = ip_fw_ctl_ptr(sopt);
-                       else
-                               error = ENOPROTOOPT;
-                       break;
-#endif /* IPFIREWALL */
 
 #if DUMMYNET
                case IP_DUMMYNET_CONFIGURE:
                case IP_DUMMYNET_DEL:
                case IP_DUMMYNET_FLUSH:
-                       if (!DUMMYNET_LOADED)
+                       if (!DUMMYNET_LOADED) {
                                ip_dn_init();
-                       if (DUMMYNET_LOADED)
+                       }
+                       if (DUMMYNET_LOADED) {
                                error = ip_dn_ctl_ptr(sopt);
-                       else
-                               error = ENOPROTOOPT ;
-                       break ;
-#endif
-
-#if MROUTING
-               case IP_RSVP_ON:
-                       error = ip_rsvp_init(so);
-                       break;
-
-               case IP_RSVP_OFF:
-                       error = ip_rsvp_done();
-                       break;
-
-                       /* XXX - should be combined */
-               case IP_RSVP_VIF_ON:
-                       error = ip_rsvp_vif_init(so, sopt);
-                       break;
-
-               case IP_RSVP_VIF_OFF:
-                       error = ip_rsvp_vif_done(so, sopt);
-                       break;
-
-               case MRT_INIT:
-               case MRT_DONE:
-               case MRT_ADD_VIF:
-               case MRT_DEL_VIF:
-               case MRT_ADD_MFC:
-               case MRT_DEL_MFC:
-               case MRT_VERSION:
-               case MRT_ASSERT:
-                       error = ip_mrouter_set(so, sopt);
+                       } else {
+                               error = ENOPROTOOPT;
+                       }
                        break;
-#endif /* MROUTING */
+#endif /* DUMMYNET */
 
                case SO_FLUSH:
-                       if ((error = sooptcopyin(sopt, &optval, sizeof (optval),
-                           sizeof (optval))) != 0)
+                       if ((error = sooptcopyin(sopt, &optval, sizeof(optval),
+                           sizeof(optval))) != 0) {
                                break;
+                       }
 
                        error = inp_flush(inp, optval);
                        break;
@@ -684,7 +812,7 @@ rip_ctloutput(so, sopt)
                break;
        }
 
-       return (error);
+       return error;
 }
 
 /*
@@ -698,18 +826,19 @@ void
 rip_ctlinput(
        int cmd,
        struct sockaddr *sa,
-       __unused void *vip)
+       __unused void *vip,
+       __unused struct ifnet *ifp)
 {
-       struct in_ifaddr *ia;
-       struct ifnet *ifp;
-       int err;
+       struct in_ifaddr *ia = NULL;
+       struct ifnet *iaifp = NULL;
+       int err = 0;
        int flags, done = 0;
 
        switch (cmd) {
        case PRC_IFDOWN:
                lck_rw_lock_shared(in_ifaddr_rwlock);
                for (ia = in_ifaddrhead.tqh_first; ia;
-                    ia = ia->ia_link.tqe_next) {
+                   ia = ia->ia_link.tqe_next) {
                        IFA_LOCK(&ia->ia_ifa);
                        if (ia->ia_ifa.ifa_addr == sa &&
                            (ia->ia_flags & IFA_ROUTE)) {
@@ -735,14 +864,15 @@ rip_ctlinput(
                        }
                        IFA_UNLOCK(&ia->ia_ifa);
                }
-               if (!done)
+               if (!done) {
                        lck_rw_done(in_ifaddr_rwlock);
+               }
                break;
 
        case PRC_IFUP:
                lck_rw_lock_shared(in_ifaddr_rwlock);
                for (ia = in_ifaddrhead.tqh_first; ia;
-                    ia = ia->ia_link.tqe_next) {
+                   ia = ia->ia_link.tqe_next) {
                        IFA_LOCK(&ia->ia_ifa);
                        if (ia->ia_ifa.ifa_addr == sa) {
                                /* keep it locked */
@@ -752,8 +882,9 @@ rip_ctlinput(
                }
                if (ia == NULL || (ia->ia_flags & IFA_ROUTE) ||
                    (ia->ia_ifa.ifa_debug & IFD_NOTREADY)) {
-                       if (ia != NULL)
+                       if (ia != NULL) {
                                IFA_UNLOCK(&ia->ia_ifa);
+                       }
                        lck_rw_done(in_ifaddr_rwlock);
                        return;
                }
@@ -762,11 +893,12 @@ rip_ctlinput(
                lck_rw_done(in_ifaddr_rwlock);
 
                flags = RTF_UP;
-               ifp = ia->ia_ifa.ifa_ifp;
+               iaifp = ia->ia_ifa.ifa_ifp;
 
-               if ((ifp->if_flags & IFF_LOOPBACK)
-                   || (ifp->if_flags & IFF_POINTOPOINT))
+               if ((iaifp->if_flags & IFF_LOOPBACK)
+                   || (iaifp->if_flags & IFF_POINTOPOINT)) {
                        flags |= RTF_HOST;
+               }
 
                err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
                if (err == 0) {
@@ -779,13 +911,15 @@ rip_ctlinput(
        }
 }
 
-u_int32_t      rip_sendspace = RIPSNDQ;
-u_int32_t      rip_recvspace = RIPRCVQ;
+u_int32_t       rip_sendspace = RIPSNDQ;
+u_int32_t       rip_recvspace = RIPRCVQ;
 
 SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW | CTLFLAG_LOCKED,
     &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
 SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW | CTLFLAG_LOCKED,
     &rip_recvspace, 0, "Maximum incoming raw IP datagram size");
+SYSCTL_UINT(_net_inet_raw, OID_AUTO, pcbcount, CTLFLAG_RD | CTLFLAG_LOCKED,
+    &ripcbinfo.ipi_count, 0, "Number of active PCBs");
 
 static int
 rip_attach(struct socket *so, int proto, struct proc *p)
@@ -794,21 +928,26 @@ rip_attach(struct socket *so, int proto, struct proc *p)
        int error;
 
        inp = sotoinpcb(so);
-       if (inp)
+       if (inp) {
                panic("rip_attach");
-       if ((so->so_state & SS_PRIV) == 0)
-               return (EPERM);
+       }
+       if ((so->so_state & SS_PRIV) == 0) {
+               return EPERM;
+       }
 
        error = soreserve(so, rip_sendspace, rip_recvspace);
-       if (error)
+       if (error) {
                return error;
+       }
        error = in_pcballoc(so, &ripcbinfo, p);
-       if (error)
+       if (error) {
                return error;
+       }
        inp = (struct inpcb *)so->so_pcb;
        inp->inp_vflag |= INP_IPV4;
-       inp->inp_ip_p = proto;
-       inp->inp_ip_ttl = ip_defttl;
+       VERIFY(proto <= UINT8_MAX);
+       inp->inp_ip_p = (u_char)proto;
+       inp->inp_ip_ttl = (u_char)ip_defttl;
        return 0;
 }
 
@@ -818,15 +957,9 @@ rip_detach(struct socket *so)
        struct inpcb *inp;
 
        inp = sotoinpcb(so);
-       if (inp == 0)
+       if (inp == 0) {
                panic("rip_detach");
-#if MROUTING
-       if (so == ip_mrouter)
-               ip_mrouter_done();
-       ip_rsvp_force_done(so);
-       if (so == ip_rsvpd)
-               ip_rsvp_done();
-#endif /* MROUTING */
+       }
        in_pcbdetach(inp);
        return 0;
 }
@@ -841,36 +974,60 @@ rip_abort(struct socket *so)
 __private_extern__ int
 rip_disconnect(struct socket *so)
 {
-       if ((so->so_state & SS_ISCONNECTED) == 0)
+       if ((so->so_state & SS_ISCONNECTED) == 0) {
                return ENOTCONN;
+       }
        return rip_abort(so);
 }
 
 __private_extern__ int
-rip_bind(struct socket *so, struct sockaddr *nam, __unused struct proc *p)
+rip_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
 {
+#pragma unused(p)
        struct inpcb *inp = sotoinpcb(so);
-       struct sockaddr_in *addr = (struct sockaddr_in *)(void *)nam;
+       struct sockaddr_in sin;
        struct ifaddr *ifa = NULL;
        struct ifnet *outif = NULL;
 
-       if (nam->sa_len != sizeof(*addr))
+       if (inp == NULL
+#if NECP
+           || (necp_socket_should_use_flow_divert(inp))
+#endif /* NECP */
+           ) {
+               return inp == NULL ? EINVAL : EPROTOTYPE;
+       }
+
+       if (nam->sa_len != sizeof(struct sockaddr_in)) {
                return EINVAL;
+       }
 
-       if (TAILQ_EMPTY(&ifnet_head) || ((addr->sin_family != AF_INET) &&
-                                   (addr->sin_family != AF_IMPLINK)) ||
-           (addr->sin_addr.s_addr &&
-            (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == 0)) {
+       /* Sanitized local copy for interface address searches */
+       bzero(&sin, sizeof(sin));
+       sin.sin_family = AF_INET;
+       sin.sin_len = sizeof(struct sockaddr_in);
+       sin.sin_addr.s_addr = SIN(nam)->sin_addr.s_addr;
+
+       if (TAILQ_EMPTY(&ifnet_head) ||
+           (sin.sin_family != AF_INET && sin.sin_family != AF_IMPLINK) ||
+           (sin.sin_addr.s_addr && (ifa = ifa_ifwithaddr(SA(&sin))) == 0)) {
                return EADDRNOTAVAIL;
-       }
-       else if (ifa) {
+       } else if (ifa) {
+               /*
+                * Opportunistically determine the outbound
+                * interface that may be used; this may not
+                * hold true if we end up using a route
+                * going over a different interface, e.g.
+                * when sending to a local address.  This
+                * will get updated again after sending.
+                */
                IFA_LOCK(ifa);
                outif = ifa->ifa_ifp;
                IFA_UNLOCK(ifa);
                IFA_REMREF(ifa);
        }
-       inp->inp_laddr = addr->sin_addr;
+       inp->inp_laddr = sin.sin_addr;
        inp->inp_last_outifp = outif;
+
        return 0;
 }
 
@@ -880,13 +1037,29 @@ rip_connect(struct socket *so, struct sockaddr *nam, __unused  struct proc *p)
        struct inpcb *inp = sotoinpcb(so);
        struct sockaddr_in *addr = (struct sockaddr_in *)(void *)nam;
 
-       if (nam->sa_len != sizeof(*addr))
+       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))
+       }
+       if (TAILQ_EMPTY(&ifnet_head)) {
                return EADDRNOTAVAIL;
+       }
        if ((addr->sin_family != AF_INET) &&
-           (addr->sin_family != AF_IMPLINK))
+           (addr->sin_family != AF_IMPLINK)) {
                return EAFNOSUPPORT;
+       }
+
+       if (!(so->so_flags1 & SOF1_CONNECT_COUNTED)) {
+               so->so_flags1 |= SOF1_CONNECT_COUNTED;
+               INC_ATOMIC_INT64_LIM(net_api_stats.nas_socket_inet_dgram_connected);
+       }
+
        inp->inp_faddr = addr->sin_addr;
        soisconnected(so);
 
@@ -901,30 +1074,47 @@ rip_shutdown(struct socket *so)
 }
 
 __private_extern__ int
-rip_send(struct socket *so, __unused int flags, struct mbuf *m, struct sockaddr *nam,
-       struct mbuf *control, __unused struct proc *p)
+rip_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);
-       register u_int32_t dst;
+       u_int32_t dst = INADDR_ANY;
+       int error = 0;
 
-       if (so->so_state & SS_ISCONNECTED) {
-               if (nam) {
-                       m_freem(m);
-                       return EISCONN;
-               }
-               dst = inp->inp_faddr.s_addr;
-       } else {
-               if (nam == NULL) {
-                       m_freem(m);
-                       return ENOTCONN;
+       if (inp == NULL
+#if NECP
+           || (necp_socket_should_use_flow_divert(inp) && (error = EPROTOTYPE))
+#endif /* NECP */
+           ) {
+               if (inp == NULL) {
+                       error = EINVAL;
+               } else {
+                       error = EPROTOTYPE;
                }
+               goto bad;
+       }
+
+       if (nam != NULL) {
                dst = ((struct sockaddr_in *)(void *)nam)->sin_addr.s_addr;
        }
        return rip_output(m, so, dst, control);
+
+bad:
+       VERIFY(error != 0);
+
+       if (m != NULL) {
+               m_freem(m);
+       }
+       if (control != NULL) {
+               m_freem(control);
+       }
+
+       return error;
 }
 
 /* note: rip_unlock is called from different protos  instead of the generic socket_unlock,
- * it will handle the socket dealloc on last reference 
+ * it will handle the socket dealloc on last reference
  * */
 int
 rip_unlock(struct socket *so, int refcount, void *debug)
@@ -932,10 +1122,11 @@ rip_unlock(struct socket *so, int refcount, void *debug)
        void *lr_saved;
        struct inpcb *inp = sotoinpcb(so);
 
-       if (debug == NULL)
+       if (debug == NULL) {
                lr_saved = __builtin_return_address(0);
-       else
+       } else {
                lr_saved = debug;
+       }
 
        if (refcount) {
                if (so->so_usecount <= 0) {
@@ -947,24 +1138,23 @@ rip_unlock(struct socket *so, int refcount, void *debug)
                if (so->so_usecount == 0 && (inp->inp_wantcnt == WNT_STOPUSING)) {
                        /* cleanup after last reference */
                        lck_mtx_unlock(so->so_proto->pr_domain->dom_mtx);
-                       lck_rw_lock_exclusive(ripcbinfo.mtx);
+                       lck_rw_lock_exclusive(ripcbinfo.ipi_lock);
                        if (inp->inp_state != INPCB_STATE_DEAD) {
-#if INET6
-                               if (INP_CHECK_SOCKAF(so, AF_INET6))
+                               if (SOCK_CHECK_DOM(so, PF_INET6)) {
                                        in6_pcbdetach(inp);
-                               else
-#endif /* INET6 */
-                               in_pcbdetach(inp);
+                               } else {
+                                       in_pcbdetach(inp);
+                               }
                        }
                        in_pcbdispose(inp);
-                       lck_rw_done(ripcbinfo.mtx);
-                       return(0);
+                       lck_rw_done(ripcbinfo.ipi_lock);
+                       return 0;
                }
        }
        so->unlock_lr[so->next_unlock_lr] = lr_saved;
-       so->next_unlock_lr = (so->next_unlock_lr+1) % SO_LCKDBG_MAX;
+       so->next_unlock_lr = (so->next_unlock_lr + 1) % SO_LCKDBG_MAX;
        lck_mtx_unlock(so->so_proto->pr_domain->dom_mtx);
-       return(0);
+       return 0;
 }
 
 static int
@@ -980,17 +1170,17 @@ rip_pcblist SYSCTL_HANDLER_ARGS
         * The process of preparing the TCB list is too time-consuming and
         * resource-intensive to repeat twice on every request.
         */
-       lck_rw_lock_exclusive(ripcbinfo.mtx);
+       lck_rw_lock_exclusive(ripcbinfo.ipi_lock);
        if (req->oldptr == USER_ADDR_NULL) {
                n = ripcbinfo.ipi_count;
                req->oldidx = 2 * (sizeof xig)
-                       + (n + n/8) * sizeof(struct xinpcb);
-               lck_rw_done(ripcbinfo.mtx);
+                   + (n + n / 8) * sizeof(struct xinpcb);
+               lck_rw_done(ripcbinfo.ipi_lock);
                return 0;
        }
 
        if (req->newptr != USER_ADDR_NULL) {
-               lck_rw_done(ripcbinfo.mtx);
+               lck_rw_done(ripcbinfo.ipi_lock);
                return EPERM;
        }
 
@@ -999,7 +1189,7 @@ rip_pcblist SYSCTL_HANDLER_ARGS
         */
        gencnt = ripcbinfo.ipi_gencnt;
        n = ripcbinfo.ipi_count;
-       
+
        bzero(&xig, sizeof(xig));
        xig.xig_len = sizeof xig;
        xig.xig_count = n;
@@ -1007,27 +1197,28 @@ rip_pcblist SYSCTL_HANDLER_ARGS
        xig.xig_sogen = so_gencnt;
        error = SYSCTL_OUT(req, &xig, sizeof xig);
        if (error) {
-               lck_rw_done(ripcbinfo.mtx);
+               lck_rw_done(ripcbinfo.ipi_lock);
                return error;
        }
-    /*
-     * We are done if there is no pcb
-     */
-    if (n == 0) {
-       lck_rw_done(ripcbinfo.mtx);
-        return 0; 
-    }
+       /*
+        * We are done if there is no pcb
+        */
+       if (n == 0) {
+               lck_rw_done(ripcbinfo.ipi_lock);
+               return 0;
+       }
 
        inp_list = _MALLOC(n * sizeof *inp_list, M_TEMP, M_WAITOK);
        if (inp_list == 0) {
-               lck_rw_done(ripcbinfo.mtx);
+               lck_rw_done(ripcbinfo.ipi_lock);
                return ENOMEM;
        }
-       
-       for (inp = ripcbinfo.listhead->lh_first, i = 0; inp && i < n;
-            inp = inp->inp_list.le_next) {
-               if (inp->inp_gencnt <= gencnt && inp->inp_state != INPCB_STATE_DEAD)
+
+       for (inp = ripcbinfo.ipi_listhead->lh_first, i = 0; inp && i < n;
+           inp = inp->inp_list.le_next) {
+               if (inp->inp_gencnt <= gencnt && inp->inp_state != INPCB_STATE_DEAD) {
                        inp_list[i++] = inp;
+               }
        }
        n = i;
 
@@ -1041,8 +1232,9 @@ rip_pcblist SYSCTL_HANDLER_ARGS
                        xi.xi_len = sizeof xi;
                        /* XXX should avoid extra copy */
                        inpcb_to_compat(inp, &xi.xi_inp);
-                       if (inp->inp_socket)
+                       if (inp->inp_socket) {
                                sotoxsocket(inp->inp_socket, &xi.xi_socket);
+                       }
                        error = SYSCTL_OUT(req, &xi, sizeof xi);
                }
        }
@@ -1062,117 +1254,121 @@ rip_pcblist SYSCTL_HANDLER_ARGS
                error = SYSCTL_OUT(req, &xig, sizeof xig);
        }
        FREE(inp_list, M_TEMP);
-       lck_rw_done(ripcbinfo.mtx);
+       lck_rw_done(ripcbinfo.ipi_lock);
        return error;
 }
 
-SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0,
-           rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
+SYSCTL_PROC(_net_inet_raw, OID_AUTO /*XXX*/, pcblist,
+    CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0,
+    rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
 
-#if !CONFIG_EMBEDDED
+#if XNU_TARGET_OS_OSX
 
 static int
 rip_pcblist64 SYSCTL_HANDLER_ARGS
 {
 #pragma unused(oidp, arg1, arg2)
-        int error, i, n;
-        struct inpcb *inp, **inp_list;
-        inp_gen_t gencnt;
-        struct xinpgen xig;
-
-        /*
-         * The process of preparing the TCB list is too time-consuming and
-         * resource-intensive to repeat twice on every request.
-         */
-        lck_rw_lock_exclusive(ripcbinfo.mtx);
-        if (req->oldptr == USER_ADDR_NULL) {
-                n = ripcbinfo.ipi_count;
-                req->oldidx = 2 * (sizeof xig)
-                        + (n + n/8) * sizeof(struct xinpcb64);
-                lck_rw_done(ripcbinfo.mtx);
-                return 0;
-        }
-
-        if (req->newptr != USER_ADDR_NULL) {
-                lck_rw_done(ripcbinfo.mtx);
-                return EPERM;
-        }
-
-        /*
-         * OK, now we're committed to doing something.
-         */
-        gencnt = ripcbinfo.ipi_gencnt;
-        n = ripcbinfo.ipi_count;
-
-        bzero(&xig, sizeof(xig));
-        xig.xig_len = sizeof xig;
-        xig.xig_count = n;
-        xig.xig_gen = gencnt;
-        xig.xig_sogen = so_gencnt;
-        error = SYSCTL_OUT(req, &xig, sizeof xig);
-        if (error) {
-                lck_rw_done(ripcbinfo.mtx);
-                return error;
-        }
-    /*
-     * We are done if there is no pcb
-     */
-    if (n == 0) {
-        lck_rw_done(ripcbinfo.mtx);
-        return 0;
-    }
-
-        inp_list = _MALLOC(n * sizeof *inp_list, M_TEMP, M_WAITOK);
-        if (inp_list == 0) {
-                lck_rw_done(ripcbinfo.mtx);
-                return ENOMEM;
-        }
-
-        for (inp = ripcbinfo.listhead->lh_first, i = 0; inp && i < n;
-             inp = inp->inp_list.le_next) {
-                if (inp->inp_gencnt <= gencnt && inp->inp_state != INPCB_STATE_DEAD)
-                        inp_list[i++] = inp;
-        }
-        n = i;
-
-        error = 0;
-        for (i = 0; i < n; i++) {
-                inp = inp_list[i];
-                if (inp->inp_gencnt <= gencnt && inp->inp_state != INPCB_STATE_DEAD) {
-                        struct xinpcb64 xi;
-
-                        bzero(&xi, sizeof(xi));
-                        xi.xi_len = sizeof xi;
-                        inpcb_to_xinpcb64(inp, &xi);
-                        if (inp->inp_socket)
-                                sotoxsocket64(inp->inp_socket, &xi.xi_socket);
-                        error = SYSCTL_OUT(req, &xi, sizeof xi);
-                }
-        }
-        if (!error) {
-                /*
-                 * Give the user an updated idea of our state.
-                 * If the generation differs from what we told
-                 * her before, she knows that something happened
-                 * while we were processing this request, and it
-                 * might be necessary to retry.
-                 */
-                bzero(&xig, sizeof(xig));
-                xig.xig_len = sizeof xig;
-                xig.xig_gen = ripcbinfo.ipi_gencnt;
-                xig.xig_sogen = so_gencnt;
-                xig.xig_count = ripcbinfo.ipi_count;
-                error = SYSCTL_OUT(req, &xig, sizeof xig);
-        }
-        FREE(inp_list, M_TEMP);
-        lck_rw_done(ripcbinfo.mtx);
-        return error;
+       int error, i, n;
+       struct inpcb *inp, **inp_list;
+       inp_gen_t gencnt;
+       struct xinpgen xig;
+
+       /*
+        * The process of preparing the TCB list is too time-consuming and
+        * resource-intensive to repeat twice on every request.
+        */
+       lck_rw_lock_exclusive(ripcbinfo.ipi_lock);
+       if (req->oldptr == USER_ADDR_NULL) {
+               n = ripcbinfo.ipi_count;
+               req->oldidx = 2 * (sizeof xig)
+                   + (n + n / 8) * sizeof(struct xinpcb64);
+               lck_rw_done(ripcbinfo.ipi_lock);
+               return 0;
+       }
+
+       if (req->newptr != USER_ADDR_NULL) {
+               lck_rw_done(ripcbinfo.ipi_lock);
+               return EPERM;
+       }
+
+       /*
+        * OK, now we're committed to doing something.
+        */
+       gencnt = ripcbinfo.ipi_gencnt;
+       n = ripcbinfo.ipi_count;
+
+       bzero(&xig, sizeof(xig));
+       xig.xig_len = sizeof xig;
+       xig.xig_count = n;
+       xig.xig_gen = gencnt;
+       xig.xig_sogen = so_gencnt;
+       error = SYSCTL_OUT(req, &xig, sizeof xig);
+       if (error) {
+               lck_rw_done(ripcbinfo.ipi_lock);
+               return error;
+       }
+       /*
+        * We are done if there is no pcb
+        */
+       if (n == 0) {
+               lck_rw_done(ripcbinfo.ipi_lock);
+               return 0;
+       }
+
+       inp_list = _MALLOC(n * sizeof *inp_list, M_TEMP, M_WAITOK);
+       if (inp_list == 0) {
+               lck_rw_done(ripcbinfo.ipi_lock);
+               return ENOMEM;
+       }
+
+       for (inp = ripcbinfo.ipi_listhead->lh_first, i = 0; inp && i < n;
+           inp = inp->inp_list.le_next) {
+               if (inp->inp_gencnt <= gencnt && inp->inp_state != INPCB_STATE_DEAD) {
+                       inp_list[i++] = inp;
+               }
+       }
+       n = i;
+
+       error = 0;
+       for (i = 0; i < n; i++) {
+               inp = inp_list[i];
+               if (inp->inp_gencnt <= gencnt && inp->inp_state != INPCB_STATE_DEAD) {
+                       struct xinpcb64 xi;
+
+                       bzero(&xi, sizeof(xi));
+                       xi.xi_len = sizeof xi;
+                       inpcb_to_xinpcb64(inp, &xi);
+                       if (inp->inp_socket) {
+                               sotoxsocket64(inp->inp_socket, &xi.xi_socket);
+                       }
+                       error = SYSCTL_OUT(req, &xi, sizeof xi);
+               }
+       }
+       if (!error) {
+               /*
+                * Give the user an updated idea of our state.
+                * If the generation differs from what we told
+                * her before, she knows that something happened
+                * while we were processing this request, and it
+                * might be necessary to retry.
+                */
+               bzero(&xig, sizeof(xig));
+               xig.xig_len = sizeof xig;
+               xig.xig_gen = ripcbinfo.ipi_gencnt;
+               xig.xig_sogen = so_gencnt;
+               xig.xig_count = ripcbinfo.ipi_count;
+               error = SYSCTL_OUT(req, &xig, sizeof xig);
+       }
+       FREE(inp_list, M_TEMP);
+       lck_rw_done(ripcbinfo.ipi_lock);
+       return error;
 }
 
-SYSCTL_PROC(_net_inet_raw, OID_AUTO, pcblist64, CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0,
-            rip_pcblist64, "S,xinpcb64", "List of active raw IP sockets");
+SYSCTL_PROC(_net_inet_raw, OID_AUTO, pcblist64,
+    CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0,
+    rip_pcblist64, "S,xinpcb64", "List of active raw IP sockets");
 
-#endif /* !CONFIG_EMBEDDED */
+#endif /* XNU_TARGET_OS_OSX */
 
 
 static int
@@ -1180,20 +1376,29 @@ rip_pcblist_n SYSCTL_HANDLER_ARGS
 {
 #pragma unused(oidp, arg1, arg2)
        int error = 0;
-       
+
        error = get_pcblist_n(IPPROTO_IP, req, &ripcbinfo);
-       
+
        return error;
 }
 
-SYSCTL_PROC(_net_inet_raw, OID_AUTO, pcblist_n, CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0,
-            rip_pcblist_n, "S,xinpcb_n", "List of active raw IP sockets");
+SYSCTL_PROC(_net_inet_raw, OID_AUTO, pcblist_n,
+    CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0,
+    rip_pcblist_n, "S,xinpcb_n", "List of active raw IP sockets");
 
 struct pr_usrreqs rip_usrreqs = {
-       rip_abort, pru_accept_notsupp, rip_attach, rip_bind, rip_connect,
-       pru_connect2_notsupp, in_control, rip_detach, rip_disconnect,
-       pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp,
-       pru_rcvoob_notsupp, rip_send, pru_sense_null, rip_shutdown,
-       in_setsockaddr, sosend, soreceive, pru_sopoll_notsupp
+       .pru_abort =            rip_abort,
+       .pru_attach =           rip_attach,
+       .pru_bind =             rip_bind,
+       .pru_connect =          rip_connect,
+       .pru_control =          in_control,
+       .pru_detach =           rip_detach,
+       .pru_disconnect =       rip_disconnect,
+       .pru_peeraddr =         in_getpeeraddr,
+       .pru_send =             rip_send,
+       .pru_shutdown =         rip_shutdown,
+       .pru_sockaddr =         in_getsockaddr,
+       .pru_sosend =           sosend,
+       .pru_soreceive =        soreceive,
 };
 /* DSEP Review Done pl-20051213-v02 @3253 */