]> git.saurik.com Git - apple/xnu.git/blobdiff - bsd/netinet/ip_output.c
xnu-1228.3.13.tar.gz
[apple/xnu.git] / bsd / netinet / ip_output.c
index 44f89c3cfb9af8f3c60bdaf3333ac1cf389e75e2..db39fe174d3ca7277baad8bd55a0b98a89bb8c62 100644 (file)
@@ -1,23 +1,29 @@
 /*
- * Copyright (c) 2000-2004 Apple Computer, Inc. All rights reserved.
+ * Copyright (c) 2000-2007 Apple Inc. All rights reserved.
  *
- * @APPLE_LICENSE_HEADER_START@
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
  * 
- * The contents of this file constitute Original Code as defined in and
- * are subject to the Apple Public Source License Version 1.1 (the
- * "License").  You may not use this file except in compliance with the
- * License.  Please obtain a copy of the License at
- * http://www.apple.com/publicsource and read it before using this file.
+ * 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
+ * compliance with the License. The rights granted to you under the License
+ * may not be used to create, or enable the creation or redistribution of,
+ * 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.
  * 
- * This Original Code and all software distributed under the License are
- * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
+ * 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,
  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT.  Please see the
- * License for the specific language governing rights and limitations
- * under the License.
+ * 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_LICENSE_HEADER_END@
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  */
 /*
  * Copyright (c) 1982, 1986, 1988, 1990, 1993
  *     @(#)ip_output.c 8.3 (Berkeley) 1/21/94
  * $FreeBSD: src/sys/netinet/ip_output.c,v 1.99.2.16 2001/07/19 06:37:26 kris Exp $
  */
+/*
+ * NOTICE: This file was modified by SPARTA, Inc. in 2005 to introduce
+ * support for mandatory and extensible security protections.  This notice
+ * is included in support of clause 2.2 (b) of the Apple Public License,
+ * Version 2.0.
+ */
 
 #define _IP_VHL
 
 
 #include <netinet/kpi_ipfilter_var.h>
 
+#if CONFIG_MACF_NET
+#include <security/mac_framework.h>
+#endif
+
 #include "faith.h"
 
 #include <net/dlil.h>
 #include <sys/kdebug.h>
+#include <libkern/OSAtomic.h>
 
 #define DBG_LAYER_BEG          NETDBG_CODE(DBG_NETIP, 1)
 #define DBG_LAYER_END          NETDBG_CODE(DBG_NETIP, 3)
 #define DBG_FNC_IP_OUTPUT      NETDBG_CODE(DBG_NETIP, (1 << 8) | 1)
 #define DBG_FNC_IPSEC4_OUTPUT  NETDBG_CODE(DBG_NETIP, (2 << 8) | 1)
 
+#define        SWAP16(v) ((((v) & 0xff) << 8) | ((v) >> 8))
 
 #if IPSEC
 #include <netinet6/ipsec.h>
                                                  (ntohl(a.s_addr))&0xFF);
 #endif
 
-#if IPSEC
-extern lck_mtx_t    *sadb_mutex;
-#endif
 
 u_short ip_id;
 
@@ -129,16 +144,17 @@ static int        ip_getmoptions(struct sockopt *, struct ip_moptions *);
 static int     ip_pcbopts(int, struct mbuf **, struct mbuf *);
 static int     ip_setmoptions(struct sockopt *, struct ip_moptions **);
 
+static void ip_out_cksum_stats(int, u_int32_t);
+
 int ip_createmoptions(struct ip_moptions **imop);
 int ip_addmembership(struct ip_moptions *imo, struct ip_mreq *mreq);
 int ip_dropmembership(struct ip_moptions *imo, struct ip_mreq *mreq);
 int    ip_optcopy(struct ip *, struct ip *);
+void in_delayed_cksum_offset(struct mbuf *, int );
+void in_cksum_offset(struct mbuf* , size_t );
+
 extern int (*fr_checkp)(struct ip *, int, struct ifnet *, int, struct mbuf **);
-#ifdef __APPLE__
-extern struct mbuf*    m_dup(register struct mbuf *m, int how);
-#endif
 
-extern int apple_hwcksum_tx;
 extern u_long  route_generation;
 
 extern struct protosw inetsw[];
@@ -154,6 +170,11 @@ extern int ipsec_bypass;
 static int     ip_maxchainsent = 0;
 SYSCTL_INT(_net_inet_ip, OID_AUTO, maxchainsent, CTLFLAG_RW,
     &ip_maxchainsent, 0, "use dlil_output_list");
+#if DEBUG
+static int forge_ce = 0;
+SYSCTL_INT(_net_inet_ip, OID_AUTO, forge_ce, CTLFLAG_RW,
+    &forge_ce, 0, "Forge ECN CE");
+#endif /* DEBUG */
 /*
  * IP output.  The packet in mbuf chain m contains a skeletal IP
  * header (with len, off, ttl, proto, tos, src, dst).
@@ -166,13 +187,36 @@ ip_output(
        struct mbuf *opt,
        struct route *ro,
        int flags,
-       struct ip_moptions *imo)
+       struct ip_moptions *imo,
+       struct ifnet *ifp)
 {
        int error;
-       error = ip_output_list(m0, 0, opt, ro, flags, imo);
+       error = ip_output_list(m0, 0, opt, ro, flags, imo, ifp);
        return error;
 }
 
+/*
+ * Returns:    0                       Success
+ *             ENOMEM
+ *             EADDRNOTAVAIL
+ *             ENETUNREACH
+ *             EHOSTUNREACH
+ *             EACCES
+ *             EMSGSIZE
+ *             ENOBUFS
+ *     ipsec4_getpolicybyaddr:???      [IPSEC 4th argument, contents modified]
+ *     ipsec4_getpolicybysock:???      [IPSEC 4th argument, contents modified]
+ *     key_spdacquire:???              [IPSEC]
+ *     ipsec4_output:???               [IPSEC]
+ *     <fr_checkp>:???                 [firewall]
+ *     ip_dn_io_ptr:???                [dummynet]
+ *     dlil_output:???                 [DLIL]
+ *     dlil_output_list:???            [DLIL]
+ *
+ * Notes:      The ipsec4_getpolicyby{addr|sock} function error returns are
+ *             only used as the error return from this function where one of
+ *             these functions fails to return a policy.
+ */
 int
 ip_output_list(
        struct mbuf *m0,
@@ -180,13 +224,19 @@ ip_output_list(
        struct mbuf *opt,
        struct route *ro,
        int flags,
-       struct ip_moptions *imo)
+       struct ip_moptions *imo,
+#if CONFIG_FORCE_OUT_IFP
+       struct ifnet *pdp_ifp
+#else
+       __unused struct ifnet *unused_ifp
+#endif
+       )
 {
        struct ip *ip, *mhip;
        struct ifnet *ifp = NULL;
        struct mbuf *m = m0;
        int hlen = sizeof (struct ip);
-       int len, off, error = 0;
+       int len = 0, off, error = 0;
        struct sockaddr_in *dst = NULL;
        struct in_ifaddr *ia = NULL;
        int isbroadcast, sw_csum;
@@ -199,22 +249,24 @@ ip_output_list(
 #if IPFIREWALL_FORWARD
        int fwd_rewrite_src = 0;
 #endif
+#if IPFIREWALL
        struct ip_fw_args args;
+#endif
        int didfilter = 0;
        ipfilter_t inject_filter_ref = 0;
        struct m_tag    *tag;
-       struct route    dn_route;
+       struct route    saved_route;
        struct mbuf * packetlist;
        int pktcnt = 0;
        
-       lck_mtx_lock(ip_mutex); 
 
        KERNEL_DEBUG(DBG_FNC_IP_OUTPUT | DBG_FUNC_START, 0,0,0,0,0);
 
        packetlist = m0;
+#if IPFIREWALL
+       args.next_hop = NULL;
        args.eh = NULL;
        args.rule = NULL;
-       args.next_hop = NULL;
        args.divert_rule = 0;                   /* divert cookie */
        
        /* Grab info from mtags prepended to the chain */
@@ -225,8 +277,8 @@ ip_output_list(
                dn_tag = (struct dn_pkt_tag *)(tag+1);
                args.rule = dn_tag->rule;
                opt = NULL;
-               dn_route = dn_tag->ro;
-               ro = &dn_route;
+               saved_route = dn_tag->ro;
+               ro = &saved_route;
                        
                imo = NULL;
                dst = dn_tag->dn_dst;
@@ -237,6 +289,7 @@ ip_output_list(
        }
 #endif /* DUMMYNET */
 
+#if IPDIVERT
        if ((tag = m_tag_locate(m0, KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_DIVERT, NULL)) != NULL) {
                struct divert_tag       *div_tag;
                
@@ -245,6 +298,8 @@ ip_output_list(
 
                m_tag_delete(m0, tag);
        }
+#endif /* IPDIVERT */
+
        if ((tag = m_tag_locate(m0, KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_IPFORWARD, NULL)) != NULL) {
                struct ip_fwd_tag       *ipfwd_tag;
                
@@ -253,6 +308,7 @@ ip_output_list(
                
                m_tag_delete(m0, tag);
        }
+#endif /* IPFIREWALL */
 
        m = m0;
        
@@ -264,21 +320,25 @@ ip_output_list(
                      mtod(m, struct ip *)->ip_p);
 #endif
 
+#if IPFIREWALL
        if (args.rule != NULL) {        /* dummynet already saw us */
             ip = mtod(m, struct ip *);
             hlen = IP_VHL_HL(ip->ip_vhl) << 2 ;
-            if (ro->ro_rt != NULL)
+           lck_mtx_lock(rt_mtx);
+            if (ro->ro_rt != NULL) 
                 ia = (struct in_ifaddr *)ro->ro_rt->rt_ifa;
-                       if (ia)
-                               ifaref(&ia->ia_ifa);
+           if (ia)
+               ifaref(&ia->ia_ifa);
+           lck_mtx_unlock(rt_mtx);
 #if IPSEC
-                       if (ipsec_bypass == 0 && (flags & IP_NOIPSEC) == 0) {
-                               so = ipsec_getsocket(m);
-                               (void)ipsec_setsocket(m, NULL);
-                       }
+           if (ipsec_bypass == 0 && (flags & IP_NOIPSEC) == 0) {
+               so = ipsec_getsocket(m);
+               (void)ipsec_setsocket(m, NULL);
+               }
 #endif
             goto sendit;
        }
+#endif /* IPFIREWALL */
 
 #if IPSEC
        if (ipsec_bypass == 0 && (flags & IP_NOIPSEC) == 0) {
@@ -298,7 +358,11 @@ loopit:
                hlen = len;
        }
        ip = mtod(m, struct ip *);
+#if IPFIREWALL
        pkt_dst = args.next_hop ? args.next_hop->sin_addr : ip->ip_dst;
+#else
+       pkt_dst = ip->ip_dst;
+#endif
 
        /*
         * Fill in IP header.
@@ -311,14 +375,24 @@ loopit:
 #else
                ip->ip_id = htons(ip_id++);
 #endif
-               ipstat.ips_localout++;
+               OSAddAtomic(1, (SInt32*)&ipstat.ips_localout);
        } else {
                hlen = IP_VHL_HL(ip->ip_vhl) << 2;
        }
+       
+#if DEBUG
+       /* For debugging, we let the stack forge congestion */
+       if (forge_ce != 0 &&
+               ((ip->ip_tos & IPTOS_ECN_MASK) == IPTOS_ECN_ECT1 ||
+                (ip->ip_tos & IPTOS_ECN_MASK) == IPTOS_ECN_ECT0)) {
+               ip->ip_tos = (ip->ip_tos & ~IPTOS_ECN_MASK) | IPTOS_ECN_CE;
+               forge_ce--;
+       }
+#endif /* DEBUG */
 
        KERNEL_DEBUG(DBG_LAYER_BEG, ip->ip_dst.s_addr, 
                     ip->ip_src.s_addr, ip->ip_p, ip->ip_off, ip->ip_len);
-
+       
        dst = (struct sockaddr_in *)&ro->ro_dst;
 
        /*
@@ -329,21 +403,26 @@ loopit:
         * cache with IPv6.
         */
 
-       {
-               if (ro->ro_rt && (ro->ro_rt->generation_id != route_generation) &&
-                       ((flags & (IP_ROUTETOIF | IP_FORWARDING)) == 0) && (ip->ip_src.s_addr != INADDR_ANY) &&
-                       (ifa_foraddr(ip->ip_src.s_addr) == 0)) {
+       lck_mtx_lock(rt_mtx);
+       if (ro->ro_rt != NULL) {
+               if (ro->ro_rt->generation_id != route_generation &&
+                   ((flags & (IP_ROUTETOIF | IP_FORWARDING)) == 0) &&
+                   (ip->ip_src.s_addr != INADDR_ANY) &&
+                   (ifa_foraddr(ip->ip_src.s_addr) == 0)) {
                        error = EADDRNOTAVAIL;
+                       lck_mtx_unlock(rt_mtx);
                        goto bad;
                }
+               if ((ro->ro_rt->rt_flags & RTF_UP) == 0 ||
+                   dst->sin_family != AF_INET ||
+                   dst->sin_addr.s_addr != pkt_dst.s_addr) {
+                       rtfree_locked(ro->ro_rt);
+                       ro->ro_rt = NULL;
+               }
+               if (ro->ro_rt && ro->ro_rt->generation_id != route_generation)
+                       ro->ro_rt->generation_id = route_generation;
        }
-       if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 ||
-          dst->sin_family != AF_INET ||
-          dst->sin_addr.s_addr != pkt_dst.s_addr)) {
-               rtfree(ro->ro_rt);
-               ro->ro_rt = (struct rtentry *)0;
-       }
-       if (ro->ro_rt == 0) {
+       if (ro->ro_rt == NULL) {
                bzero(dst, sizeof(*dst));
                dst->sin_family = AF_INET;
                dst->sin_len = sizeof(*dst);
@@ -360,8 +439,9 @@ loopit:
                        ifafree(&ia->ia_ifa);
                if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == 0) {
                        if ((ia = ifatoia(ifa_ifwithnet(sintosa(dst)))) == 0) {
-                               ipstat.ips_noroute++;
+                               OSAddAtomic(1, (SInt32*)&ipstat.ips_noroute);
                                error = ENETUNREACH;
+                               lck_mtx_unlock(rt_mtx);
                                goto bad;
                        }
                }
@@ -369,6 +449,21 @@ loopit:
                ip->ip_ttl = 1;
                isbroadcast = in_broadcast(dst->sin_addr, ifp);
        } else {
+
+#if CONFIG_FORCE_OUT_IFP
+               /* Check if this packet should be forced out a specific interface */
+               if (ro->ro_rt == 0 && pdp_ifp != NULL) {
+                       pdp_context_route_locked(pdp_ifp, ro);
+                       
+                       if (ro->ro_rt == NULL) {
+                               OSAddAtomic(1, (UInt32*)&ipstat.ips_noroute);
+                               error = EHOSTUNREACH;
+                               lck_mtx_unlock(rt_mtx);
+                               goto bad;
+                       }
+               }
+#endif
+               
                /*
                 * If this is the case, we probably don't want to allocate
                 * a protocol-cloned route since we didn't get one from the
@@ -378,13 +473,37 @@ loopit:
                 * the link layer, as this is probably required in all cases
                 * for correct operation (as it is for ARP).
                 */
-               if (ro->ro_rt == 0)
-                       rtalloc_ign(ro, RTF_PRCLONING);
+               
+               if (ro->ro_rt == 0) {
+                       unsigned long ign = RTF_PRCLONING;
+                       /*
+                        * We make an exception here: if the destination
+                        * address is INADDR_BROADCAST, allocate a protocol-
+                        * cloned host route so that we end up with a route
+                        * marked with the RTF_BROADCAST flag.  Otherwise,
+                        * we would end up referring to the default route,
+                        * instead of creating a cloned host route entry.
+                        * That would introduce inconsistencies between ULPs
+                        * that allocate a route and those that don't.  The
+                        * RTF_BROADCAST route is important since we'd want
+                        * to send out undirected IP broadcast packets using
+                        * link-level broadcast address.
+                        *
+                        * This exception will no longer be necessary when
+                        * the RTF_PRCLONING scheme is no longer present.
+                        */
+                       if (dst->sin_addr.s_addr == INADDR_BROADCAST)
+                               ign &= ~RTF_PRCLONING;
+
+                       rtalloc_ign_locked(ro, ign);
+               }
                if (ro->ro_rt == 0) {
-                       ipstat.ips_noroute++;
+                       OSAddAtomic(1, (SInt32*)&ipstat.ips_noroute);
                        error = EHOSTUNREACH;
+                       lck_mtx_unlock(rt_mtx);
                        goto bad;
                }
+               
                if (ia)
                        ifafree(&ia->ia_ifa);
                ia = ifatoia(ro->ro_rt->rt_ifa);
@@ -399,6 +518,7 @@ loopit:
                else
                        isbroadcast = in_broadcast(dst->sin_addr, ifp);
        }
+       lck_mtx_unlock(rt_mtx);
        if (IN_MULTICAST(ntohl(pkt_dst.s_addr))) {
                struct in_multi *inm;
 
@@ -417,10 +537,12 @@ loopit:
                        if (imo->imo_multicast_ifp != NULL) {
                                ifp = imo->imo_multicast_ifp;
                        }
+#if MROUTING
                        if (imo->imo_multicast_vif != -1 && 
                                ((flags & IP_RAWOUTPUT) == 0 || ip->ip_src.s_addr == INADDR_ANY))
                                ip->ip_src.s_addr =
                                        ip_mcast_src(imo->imo_multicast_vif);
+#endif /* MROUTING */
                } else
                        if ((flags & IP_RAWOUTPUT) == 0) ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL;
                /*
@@ -428,7 +550,7 @@ loopit:
                 */
                if ((imo == NULL) || (imo->imo_multicast_vif == -1)) {
                        if ((ifp->if_flags & IFF_MULTICAST) == 0) {
-                               ipstat.ips_noroute++;
+                               OSAddAtomic(1, (SInt32*)&ipstat.ips_noroute);
                                error = ENETUNREACH;
                                goto bad;
                        }
@@ -439,13 +561,14 @@ loopit:
                 */
                if (ip->ip_src.s_addr == INADDR_ANY) {
                        register struct in_ifaddr *ia1;
-
+                       lck_mtx_lock(rt_mtx);   
                        TAILQ_FOREACH(ia1, &in_ifaddrhead, ia_link)
                                if (ia1->ia_ifp == ifp) {
                                        ip->ip_src = IA_SIN(ia1)->sin_addr;
                                        
                                        break;
                                }
+                       lck_mtx_unlock(rt_mtx); 
                        if (ip->ip_src.s_addr == INADDR_ANY) {
                                error = ENETUNREACH;
                                goto bad;
@@ -474,8 +597,12 @@ loopit:
                                        ipf_pktopts.ippo_mcast_loop = imo->imo_multicast_loop;
                                }
                                
-                               lck_mtx_unlock(ip_mutex);
                                ipf_ref();
+                               
+                               /* 4135317 - always pass network byte order to filter */
+                               HTONS(ip->ip_len);
+                               HTONS(ip->ip_off);
+                               
                                TAILQ_FOREACH(filter, &ipv4_filters, ipf_link) {
                                        if (seen == 0) {
                                                if ((struct ipfilter *)inject_filter_ref == filter)
@@ -489,17 +616,22 @@ loopit:
                                                }
                                                if (result != 0) {
                                                        ipf_unref();
-                                                       lck_mtx_lock(ip_mutex);
                                                        goto bad;
                                                }
                                        }
                                }
-                               lck_mtx_lock(ip_mutex);
+                               
+                               /* set back to host byte order */
+                               ip = mtod(m, struct ip *);
+                               NTOHS(ip->ip_len);
+                               NTOHS(ip->ip_off);
+                               
                                ipf_unref();
                                didfilter = 1;
                        }
                        ip_mloopback(ifp, m, dst, hlen);
                }
+#if MROUTING
                else {
                        /*
                         * If we are acting as a multicast router, perform
@@ -524,11 +656,11 @@ loopit:
                                  imo = NULL;
                                if (ip_mforward(ip, ifp, m, imo) != 0) {
                                        m_freem(m);
-                                       lck_mtx_unlock(ip_mutex);
                                        goto done;
                                }
                        }
                }
+#endif /* MROUTING */
 
                /*
                 * Multicasts with a time-to-live of zero may be looped-
@@ -540,7 +672,6 @@ loopit:
                 */
                if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) {
                        m_freem(m);
-                       lck_mtx_unlock(ip_mutex);
                        goto done;
                }
 
@@ -600,13 +731,16 @@ sendit:
                }
         }
 
-injectit:
        if (!didfilter && !TAILQ_EMPTY(&ipv4_filters)) {
                struct ipfilter *filter;
                int seen = (inject_filter_ref == 0);
                
-               lck_mtx_unlock(ip_mutex);
                ipf_ref();
+               
+               /* 4135317 - always pass network byte order to filter */
+               HTONS(ip->ip_len);
+               HTONS(ip->ip_off);
+               
                TAILQ_FOREACH(filter, &ipv4_filters, ipf_link) {
                        if (seen == 0) {
                                if ((struct ipfilter *)inject_filter_ref == filter)
@@ -620,13 +754,17 @@ injectit:
                                }
                                if (result != 0) {
                                        ipf_unref();
-                                       lck_mtx_lock(ip_mutex);
                                        goto bad;
                                }
                        }
                }
+               
+               /* set back to host byte order */
+               ip = mtod(m, struct ip *);
+               NTOHS(ip->ip_len);
+               NTOHS(ip->ip_off);
+               
                ipf_unref();
-               lck_mtx_lock(ip_mutex);
        }
 
 #if IPSEC
@@ -637,7 +775,6 @@ injectit:
 
        KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_START, 0,0,0,0,0);
 
-       lck_mtx_lock(sadb_mutex);
 
        /* get SP for this packet */
        if (so == NULL)
@@ -646,9 +783,8 @@ injectit:
                sp = ipsec4_getpolicybysock(m, IPSEC_DIR_OUTBOUND, so, &error);
 
        if (sp == NULL) {
-               ipsecstat.out_inval++;
+               IPSEC_STAT_INCREMENT(ipsecstat.out_inval);              
                KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_END, 0,0,0,0,0);
-               lck_mtx_unlock(sadb_mutex);
                goto bad;
        }
 
@@ -657,19 +793,18 @@ injectit:
        /* check policy */
        switch (sp->policy) {
        case IPSEC_POLICY_DISCARD:
+       case IPSEC_POLICY_GENERATE:
                /*
                 * This packet is just discarded.
                 */
-               ipsecstat.out_polvio++;
+               IPSEC_STAT_INCREMENT(ipsecstat.out_polvio);
                KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_END, 1,0,0,0,0);
-               lck_mtx_unlock(sadb_mutex);
                goto bad;
 
        case IPSEC_POLICY_BYPASS:
        case IPSEC_POLICY_NONE:
                /* no need to do IPsec. */
                KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_END, 2,0,0,0,0);
-               lck_mtx_unlock(sadb_mutex);
                goto skip_ipsec;
        
        case IPSEC_POLICY_IPSEC:
@@ -677,7 +812,6 @@ injectit:
                        /* acquire a policy */
                        error = key_spdacquire(sp);
                        KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_END, 3,0,0,0,0);
-                       lck_mtx_unlock(sadb_mutex);
                        goto bad;
                }
                break;
@@ -711,10 +845,7 @@ injectit:
        HTONS(ip->ip_len);
        HTONS(ip->ip_off);
 
-       lck_mtx_unlock(ip_mutex);
        error = ipsec4_output(&state, sp, flags);
-       lck_mtx_unlock(sadb_mutex);
-       lck_mtx_lock(ip_mutex);
     
        m0 = m = state.m;
        
@@ -764,13 +895,15 @@ injectit:
 #endif
        /* Check that there wasn't a route change and src is still valid */
 
-       if (ro->ro_rt->generation_id != route_generation) {
+       lck_mtx_lock(rt_mtx);
+       if (ro->ro_rt && ro->ro_rt->generation_id != route_generation) {
                if (ifa_foraddr(ip->ip_src.s_addr) == 0 && ((flags & (IP_ROUTETOIF | IP_FORWARDING)) == 0)) {
                        error = EADDRNOTAVAIL;
+                       lck_mtx_unlock(rt_mtx);
                        KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_END, 5,0,0,0,0);
                        goto bad;
                }
-               rtfree(ro->ro_rt);
+               rtfree_locked(ro->ro_rt);
                ro->ro_rt = NULL;
        }
 
@@ -779,6 +912,7 @@ injectit:
                        printf("ip_output: "
                                "can't update route after IPsec processing\n");
                        error = EHOSTUNREACH;   /*XXX*/ 
+                       lck_mtx_unlock(rt_mtx);
                        KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_END, 6,0,0,0,0);
                        goto bad;
                }
@@ -790,6 +924,7 @@ injectit:
                        ifaref(&ia->ia_ifa);
                ifp = ro->ro_rt->rt_ifp;
        }
+       lck_mtx_unlock(rt_mtx);
 
        /* make it flipped, again. */
        NTOHS(ip->ip_len);
@@ -800,8 +935,12 @@ injectit:
        if (!TAILQ_EMPTY(&ipv4_filters)) {
                struct ipfilter *filter;
                
-               lck_mtx_unlock(ip_mutex);
                ipf_ref();
+               
+               /* 4135317 - always pass network byte order to filter */
+               HTONS(ip->ip_len);
+               HTONS(ip->ip_off);
+               
                TAILQ_FOREACH(filter, &ipv4_filters, ipf_link) {
                        if (filter->ipf_filter.ipf_output) {
                                errno_t result;
@@ -812,17 +951,22 @@ injectit:
                                }
                                if (result != 0) {
                                        ipf_unref();
-                                       lck_mtx_lock(ip_mutex);
                                        goto bad;
                                }
                        }
                }
+               
+               /* set back to host byte order */
+               ip = mtod(m, struct ip *);
+               NTOHS(ip->ip_len);
+               NTOHS(ip->ip_off);
+               
                ipf_unref();
-               lck_mtx_lock(ip_mutex);
        }
 skip_ipsec:
 #endif /*IPSEC*/
 
+#if IPFIREWALL
        /*
         * IpHack's section.
         * - Xlate: translate packet's addr/port (NAT).
@@ -834,7 +978,6 @@ skip_ipsec:
                struct  mbuf    *m1 = m;
 
                if ((error = (*fr_checkp)(ip, hlen, ifp, 1, &m1)) || !m1) {
-                       lck_mtx_unlock(ip_mutex);
                        goto done;
                }
                ip = mtod(m0 = m = m1, struct ip *);
@@ -850,7 +993,6 @@ skip_ipsec:
                args.m = m;
                args.next_hop = dst;
                args.oif = ifp;
-               lck_mtx_unlock(ip_mutex);
                off = ip_fw_chk_ptr(&args);
                m = args.m;
                dst = args.next_hop;
@@ -877,8 +1019,8 @@ skip_ipsec:
                        goto done ;
                }
                ip = mtod(m, struct ip *);
+               
                if (off == 0 && dst == old) {/* common case */
-                       lck_mtx_lock(ip_mutex);
                        goto pass ;
                }
 #if DUMMYNET
@@ -901,7 +1043,6 @@ skip_ipsec:
                    goto done;
                }
 #endif /* DUMMYNET */
-               lck_mtx_lock(ip_mutex);
 #if IPDIVERT
                if (off != 0 && (off & IP_FW_PORT_DYNT_FLAG) == 0) {
                        struct mbuf *clone = NULL;
@@ -932,7 +1073,6 @@ skip_ipsec:
                                ip = mtod(m, struct ip *);
                                goto pass;
                        }
-                       lck_mtx_unlock(ip_mutex);
                        goto done;
                }
 #endif
@@ -1019,7 +1159,6 @@ skip_ipsec:
                                HTONS(ip->ip_len);
                                HTONS(ip->ip_off);
                                
-                               lck_mtx_unlock(ip_mutex);
                                
                                /*  we need to call dlil_output to run filters
                                 *      and resync to avoid recursion loops.
@@ -1041,11 +1180,13 @@ skip_ipsec:
                        bcopy(dst, &ro_fwd->ro_dst, sizeof(*dst));
 
                        ro_fwd->ro_rt = 0;
-                       rtalloc_ign(ro_fwd, RTF_PRCLONING);
+                       lck_mtx_lock(rt_mtx);
+                       rtalloc_ign_locked(ro_fwd, RTF_PRCLONING);
 
                        if (ro_fwd->ro_rt == 0) {
-                               ipstat.ips_noroute++;
+                               OSAddAtomic(1, (SInt32*)&ipstat.ips_noroute);
                                error = EHOSTUNREACH;
+                               lck_mtx_unlock(rt_mtx);
                                goto bad;
                        }
 
@@ -1059,9 +1200,10 @@ skip_ipsec:
                                    (ro_fwd->ro_rt->rt_flags & RTF_BROADCAST);
                        else
                                isbroadcast = in_broadcast(dst->sin_addr, ifp);
-                       rtfree(ro->ro_rt);
+                       rtfree_locked(ro->ro_rt);
                        ro->ro_rt = ro_fwd->ro_rt;
                        dst = (struct sockaddr_in *)&ro_fwd->ro_dst;
+                       lck_mtx_unlock(rt_mtx);
 
                        /*
                         * If we added a default src ip earlier,
@@ -1079,9 +1221,9 @@ skip_ipsec:
                  */
                m_freem(m);
                error = EACCES; /* not sure this is the right error msg */
-               lck_mtx_unlock(ip_mutex);
                goto done;
        }
+#endif /* IPFIREWALL */
 
 pass:
 #if __APPLE__
@@ -1089,7 +1231,7 @@ pass:
        if ((ifp->if_flags & IFF_LOOPBACK) == 0 &&
                 ((ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
                  (ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET)) {
-               ipstat.ips_badaddr++;
+               OSAddAtomic(1, (SInt32*)&ipstat.ips_badaddr);
                m_freem(m);
                /* 
                 * Do not simply drop the packet just like a firewall -- we want the 
@@ -1099,7 +1241,6 @@ pass:
                 * loopback as the source address.
                 */
                error = ENETUNREACH;
-               lck_mtx_unlock(ip_mutex);
                goto done;
        }
 #endif
@@ -1127,6 +1268,9 @@ pass:
                        /* let the software handle any UDP or TCP checksums */
                        sw_csum |= (CSUM_DELAY_DATA & m->m_pkthdr.csum_flags);
                }
+       } else if (apple_hwcksum_tx == 0) {
+               sw_csum |= (CSUM_DELAY_DATA | CSUM_DELAY_IP) &
+                   m->m_pkthdr.csum_flags;
        }
        
        if (sw_csum & CSUM_DELAY_DATA) {
@@ -1134,8 +1278,13 @@ pass:
                sw_csum &= ~CSUM_DELAY_DATA;
                m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
        }
-       
-       m->m_pkthdr.csum_flags &= IF_HWASSIST_CSUM_FLAGS(ifp->if_hwassist);
+
+       if (apple_hwcksum_tx != 0) {
+               m->m_pkthdr.csum_flags &=
+                   IF_HWASSIST_CSUM_FLAGS(ifp->if_hwassist);
+       } else {
+               m->m_pkthdr.csum_flags = 0;
+       }
 
        /*
         * If small enough for interface, or the interface will take
@@ -1143,6 +1292,8 @@ pass:
         */
        if ((u_short)ip->ip_len <= ifp->if_mtu ||
            ifp->if_hwassist & CSUM_FRAGMENT) {
+               struct rtentry *rte;
+
                HTONS(ip->ip_len);
                HTONS(ip->ip_off);
                ip->ip_sum = 0;
@@ -1164,20 +1315,30 @@ pass:
                        ipsec_delaux(m);
 #endif
                if (packetchain == 0) {
-                       lck_mtx_unlock(ip_mutex);
-                       error = dlil_output(ifp, PF_INET, m, (void *) ro->ro_rt,
-                                   (struct sockaddr *)dst, 0);
-                       goto done;      
+                       lck_mtx_lock(rt_mtx);
+                       if ((rte = ro->ro_rt) != NULL)
+                               rtref(rte);
+                       lck_mtx_unlock(rt_mtx);
+                       error = ifnet_output(ifp, PF_INET, m, rte,
+                           (struct sockaddr *)dst);
+                       if (rte != NULL)
+                               rtfree(rte);
+                       goto done;
                }
                else { /* packet chaining allows us to reuse the route for all packets */
                        m = m->m_nextpkt;
                        if (m == NULL) {
                                if (pktcnt > ip_maxchainsent)
                                        ip_maxchainsent = pktcnt;
+                               lck_mtx_lock(rt_mtx);
+                               if ((rte = ro->ro_rt) != NULL)
+                                       rtref(rte);
+                               lck_mtx_unlock(rt_mtx);
                                //send
-                               lck_mtx_unlock(ip_mutex);
-                               error = dlil_output_list(ifp, PF_INET, packetlist, (void *) ro->ro_rt,
-                                       (struct sockaddr *)dst, 0);
+                               error = ifnet_output(ifp, PF_INET, packetlist,
+                                   rte, (struct sockaddr *)dst);
+                               if (rte != NULL)
+                                       rtfree(rte);
                                pktcnt = 0;
                                goto done;
        
@@ -1200,12 +1361,15 @@ pass:
                 * them, there is no way for one to update all its
                 * routes when the MTU is changed.
                 */
-               if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST))
+               
+               lck_mtx_lock(rt_mtx);
+               if (ro->ro_rt && (ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST))
                    && !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU)
                    && (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) {
                        ro->ro_rt->rt_rmx.rmx_mtu = ifp->if_mtu;
                }
-               ipstat.ips_cantfrag++;
+               lck_mtx_unlock(rt_mtx);
+               OSAddAtomic(1, (SInt32*)&ipstat.ips_cantfrag);
                goto bad;
        }
        len = (ifp->if_mtu - hlen) &~ 7;
@@ -1221,10 +1385,6 @@ pass:
        if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA &&
            (ifp->if_hwassist & CSUM_IP_FRAGS) == 0) {
                in_delayed_cksum(m);
-               if (m == NULL) {
-                       lck_mtx_unlock(ip_mutex);
-                       return(ENOMEM);
-               }
                m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
        }
 
@@ -1241,10 +1401,10 @@ pass:
        m0 = m;
        mhlen = sizeof (struct ip);
        for (off = hlen + len; off < (u_short)ip->ip_len; off += len) {
-               MGETHDR(m, M_DONTWAIT, MT_HEADER);
+               MGETHDR(m, M_DONTWAIT, MT_HEADER);      /* MAC-OK */
                if (m == 0) {
                        error = ENOBUFS;
-                       ipstat.ips_odropped++;
+                       OSAddAtomic(1, (SInt32*)&ipstat.ips_odropped);
                        goto sendorfree;
                }
                m->m_flags |= (m0->m_flags & M_MCAST) | M_FRAG;
@@ -1268,13 +1428,16 @@ pass:
                if (m->m_next == 0) {
                        (void) m_free(m);
                        error = ENOBUFS;        /* ??? */
-                       ipstat.ips_odropped++;
+                       OSAddAtomic(1, (SInt32*)&ipstat.ips_odropped);
                        goto sendorfree;
                }
                m->m_pkthdr.len = mhlen + len;
                m->m_pkthdr.rcvif = 0;
                m->m_pkthdr.csum_flags = m0->m_pkthdr.csum_flags;
                m->m_pkthdr.socket_id = m0->m_pkthdr.socket_id;
+#if CONFIG_MACF_NET
+               mac_netinet_fragment(m0, m);
+#endif
                HTONS(mhip->ip_off);
                mhip->ip_sum = 0;
                if (sw_csum & CSUM_DELAY_IP) {
@@ -1284,7 +1447,7 @@ pass:
                mnext = &m->m_nextpkt;
                nfrags++;
        }
-       ipstat.ips_ofragments += nfrags;
+       OSAddAtomic(nfrags, (SInt32*)&ipstat.ips_ofragments);
 
        /* set first/last markers for fragment chain */
        m->m_flags |= M_LASTFRAG;
@@ -1310,7 +1473,6 @@ sendorfree:
        KERNEL_DEBUG(DBG_LAYER_END, ip->ip_dst.s_addr, 
                     ip->ip_src.s_addr, ip->ip_p, ip->ip_off, ip->ip_len);
 
-       lck_mtx_unlock(ip_mutex);
        for (m = m0; m; m = m0) {
                m0 = m->m_nextpkt;
                m->m_nextpkt = 0;
@@ -1320,6 +1482,7 @@ sendorfree:
                        ipsec_delaux(m);
 #endif
                if (error == 0) {
+                       struct rtentry *rte;
 #ifndef __APPLE__
                        /* Record statistics for this interface address. */
                        if (ia != NULL) {
@@ -1327,16 +1490,22 @@ sendorfree:
                                ia->ia_ifa.if_obytes += m->m_pkthdr.len;
                        }
 #endif
-                   if ((packetchain != 0)  && (pktcnt > 0))    
-                               panic("ip_output: mix of packet in packetlist is wrong=%x", packetlist);
-                   error = dlil_output(ifp, PF_INET, m, (void *) ro->ro_rt,
-                                       (struct sockaddr *)dst, 0);
+                       if ((packetchain != 0)  && (pktcnt > 0))
+                               panic("ip_output: mix of packet in packetlist is wrong=%p", packetlist);
+                       lck_mtx_lock(rt_mtx);
+                       if ((rte = ro->ro_rt) != NULL)
+                               rtref(rte);
+                       lck_mtx_unlock(rt_mtx);
+                       error = ifnet_output(ifp, PF_INET, m, rte,
+                           (struct sockaddr *)dst);
+                       if (rte != NULL)
+                               rtfree(rte);
                } else
                        m_freem(m);
        }
 
        if (error == 0)
-               ipstat.ips_fragmented++;
+               OSAddAtomic(1, (SInt32*)&ipstat.ips_fragmented);
     }
 done:
        if (ia) {
@@ -1352,9 +1521,7 @@ done:
        if (sp != NULL) {
                KEYDEBUG(KEYDEBUG_IPSEC_STAMP,
                        printf("DP ip_output call free SP:%x\n", sp));
-               lck_mtx_lock(sadb_mutex);
-               key_freesp(sp);
-               lck_mtx_unlock(sadb_mutex);
+               key_freesp(sp, KEY_SADB_UNLOCKED);
        }
        }
 #endif /* IPSEC */
@@ -1363,33 +1530,53 @@ done:
        return (error);
 bad:
        m_freem(m0);
-       lck_mtx_unlock(ip_mutex);
        goto done;
 }
 
+static void
+ip_out_cksum_stats(int proto, u_int32_t len)
+{
+       switch (proto) {
+       case IPPROTO_TCP:
+               tcp_out_cksum_stats(len);
+               break;
+       case IPPROTO_UDP:
+               udp_out_cksum_stats(len);
+               break;
+       default:
+               /* keep only TCP or UDP stats for now */
+               break;
+       }
+}
+
 void
-in_delayed_cksum_offset(struct mbuf *m, int ip_offset)
+in_delayed_cksum_offset(struct mbuf *m0, int ip_offset)
 {
        struct ip *ip;
-       u_short csum, offset;
+       unsigned char buf[sizeof(struct ip)];
+       u_short csum, offset, ip_len;
+       struct mbuf *m = m0;
        
-       while (ip_offset > m->m_len) {
+       while (ip_offset >= m->m_len) {
                ip_offset -= m->m_len;
                m = m->m_next;
-               if (m) {
+               if (m == NULL) {
                        printf("in_delayed_cksum_withoffset failed - ip_offset wasn't in the packet\n");
                        return;
                }
        }
        
+       /* Sometimes the IP header is not contiguous, yes this can happen! */
        if (ip_offset + sizeof(struct ip) > m->m_len) {
-               printf("delayed m_pullup, m->len: %d  off: %d\n",
+#if DEBUG              
+               printf("delayed m_pullup, m->len: %ld  off: %d\n",
                        m->m_len, ip_offset);
-               /*
-                * XXX
-                * this shouldn't happen
-                */
-               m = m_pullup(m, ip_offset + sizeof(struct ip));
+#endif
+               m_copydata(m, ip_offset, sizeof(struct ip), (caddr_t) buf);
+               
+               ip = (struct ip *)buf;
+       } else {
+               ip = (struct ip*)(m->m_data + ip_offset);
        }
        
        /* Gross */
@@ -1398,36 +1585,60 @@ in_delayed_cksum_offset(struct mbuf *m, int ip_offset)
                m->m_data += ip_offset;
        }
        
-       ip = mtod(m, struct ip*);
        offset = IP_VHL_HL(ip->ip_vhl) << 2 ;
-       csum = in_cksum_skip(m, ip->ip_len, offset);
-       if (m->m_pkthdr.csum_flags & CSUM_UDP && csum == 0)
+
+       /*
+        * We could be in the context of an IP or interface filter; in the
+        * former case, ip_len would be in host (correct) order while for
+        * the latter it would be in network order.  Because of this, we
+        * attempt to interpret the length field by comparing it against
+        * the actual packet length.  If the comparison fails, byte swap
+        * the length and check again.  If it still fails, then the packet
+        * is bogus and we give up.
+        */
+       ip_len = ip->ip_len;
+       if (ip_len != (m0->m_pkthdr.len - ip_offset)) {
+               ip_len = SWAP16(ip_len);
+               if (ip_len != (m0->m_pkthdr.len - ip_offset)) {
+                       printf("in_delayed_cksum_offset: ip_len %d (%d) "
+                           "doesn't match actual length %d\n", ip->ip_len,
+                           ip_len, (m0->m_pkthdr.len - ip_offset));
+                       return;
+               }
+       }
+
+       csum = in_cksum_skip(m, ip_len, offset);
+
+       /* Update stats */
+       ip_out_cksum_stats(ip->ip_p, ip_len - offset);
+
+       if (m0->m_pkthdr.csum_flags & CSUM_UDP && csum == 0)
                csum = 0xffff;
-       offset += m->m_pkthdr.csum_data & 0xFFFF;        /* checksum offset */
-       
+       offset += m0->m_pkthdr.csum_data & 0xFFFF;        /* checksum offset */
+
        /* Gross */
        if (ip_offset) {
                if (M_LEADINGSPACE(m) < ip_offset)
-                       panic("in_delayed_cksum_withoffset - chain modified!\n");
+                       panic("in_delayed_cksum_offset - chain modified!\n");
                m->m_len += ip_offset;
                m->m_data -= ip_offset;
        }
 
-       if (offset > ip->ip_len) /* bogus offset */
+       if (offset > ip_len) /* bogus offset */
                return;
 
+       /* Insert the checksum in the existing chain */
        if (offset + ip_offset + sizeof(u_short) > m->m_len) {
-               printf("delayed m_pullup, m->len: %d  off: %d  p: %d\n",
+               char tmp[2];
+               
+#if DEBUG
+               printf("delayed m_copyback, m->len: %ld  off: %d  p: %d\n",
                    m->m_len, offset + ip_offset, ip->ip_p);
-               /*
-                * XXX
-                * this shouldn't happen, but if it does, the
-                * correct behavior may be to insert the checksum
-                * in the existing chain instead of rearranging it.
-                */
-               m = m_pullup(m, offset + ip_offset + sizeof(u_short));
-       }
-       *(u_short *)(m->m_data + offset + ip_offset) = csum;
+#endif
+               *(u_short *)tmp = csum;
+               m_copyback(m, offset + ip_offset, 2, tmp);
+       } else
+               *(u_short *)(m->m_data + offset + ip_offset) = csum;
 }
 
 void
@@ -1441,24 +1652,33 @@ in_cksum_offset(struct mbuf* m, size_t ip_offset)
 {
        struct ip* ip = NULL;
        int hlen = 0;
+       unsigned char buf[sizeof(struct ip)];
+       int swapped = 0;
        
-       while (ip_offset > m->m_len) {
+       while (ip_offset >= m->m_len) {
                ip_offset -= m->m_len;
                m = m->m_next;
-               if (m) {
+               if (m == NULL) {
                        printf("in_cksum_offset failed - ip_offset wasn't in the packet\n");
                        return;
                }
        }
        
+       /* Sometimes the IP header is not contiguous, yes this can happen! */
        if (ip_offset + sizeof(struct ip) > m->m_len) {
-               printf("in_cksum_offset - delayed m_pullup, m->len: %d  off: %d\n",
+
+#if DEBUG
+               printf("in_cksum_offset - delayed m_pullup, m->len: %ld  off: %lu\n",
                        m->m_len, ip_offset);
-               /*
-                * XXX
-                * this shouldn't happen
-                */
-               m = m_pullup(m, ip_offset + sizeof(struct ip));
+#endif 
+               m_copydata(m, ip_offset, sizeof(struct ip), (caddr_t) buf);
+
+               ip = (struct ip *)buf;
+               ip->ip_sum = 0;
+               m_copyback(m, ip_offset + offsetof(struct ip, ip_sum), 2, (caddr_t)&ip->ip_sum);
+       } else {
+               ip = (struct ip*)(m->m_data + ip_offset);
+               ip->ip_sum = 0;
        }
        
        /* Gross */
@@ -1467,17 +1687,38 @@ in_cksum_offset(struct mbuf* m, size_t ip_offset)
                m->m_data += ip_offset;
        }
 
-       ip = mtod(m, struct ip*);
-
 #ifdef _IP_VHL
        hlen = IP_VHL_HL(ip->ip_vhl) << 2;
 #else
        hlen = ip->ip_hl << 2;
 #endif
-       
+       /*
+        * We could be in the context of an IP or interface filter; in the
+        * former case, ip_len would be in host order while for the latter
+        * it would be in network (correct) order.  Because of this, we
+        * attempt to interpret the length field by comparing it against
+        * the actual packet length.  If the comparison fails, byte swap
+        * the length and check again.  If it still fails, then the packet
+        * is bogus and we give up.
+        */
+       if (ntohs(ip->ip_len) != (m->m_pkthdr.len - ip_offset)) {
+               ip->ip_len = SWAP16(ip->ip_len);
+               swapped = 1;
+               if (ntohs(ip->ip_len) != (m->m_pkthdr.len - ip_offset)) {
+                       ip->ip_len = SWAP16(ip->ip_len);
+                       printf("in_cksum_offset: ip_len %d (%d) "
+                           "doesn't match actual length %lu\n",
+                           ip->ip_len, SWAP16(ip->ip_len),
+                           (m->m_pkthdr.len - ip_offset));
+                       return;
+               }
+       }
+
        ip->ip_sum = 0;
        ip->ip_sum = in_cksum(m, hlen);
-       
+       if (swapped)
+               ip->ip_len = SWAP16(ip->ip_len);
+
        /* Gross */
        if (ip_offset) {
                if (M_LEADINGSPACE(m) < ip_offset)
@@ -1485,6 +1726,18 @@ in_cksum_offset(struct mbuf* m, size_t ip_offset)
                m->m_len += ip_offset;
                m->m_data -= ip_offset;
        }
+
+       /* Insert the checksum in the existing chain if IP header not contiguous */
+       if (ip_offset + sizeof(struct ip) > m->m_len) {
+               char tmp[2];
+
+#if DEBUG
+               printf("in_cksum_offset m_copyback, m->len: %lu  off: %lu  p: %d\n",
+                   m->m_len, ip_offset + offsetof(struct ip, ip_sum), ip->ip_p);
+#endif
+               *(u_short *)tmp = ip->ip_sum;
+               m_copyback(m, ip_offset + offsetof(struct ip, ip_sum), 2, tmp);
+       }
 }
 
 /*
@@ -1511,10 +1764,13 @@ ip_insertoptions(m, opt, phlen)
        if (p->ipopt_dst.s_addr)
                ip->ip_dst = p->ipopt_dst;
        if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) {
-               MGETHDR(n, M_DONTWAIT, MT_HEADER);
+               MGETHDR(n, M_DONTWAIT, MT_HEADER);      /* MAC-OK */
                if (n == 0)
                        return (m);
                n->m_pkthdr.rcvif = 0;
+#if CONFIG_MACF_NET
+               mac_mbuf_label_copy(m, n);
+#endif
                n->m_pkthdr.len = m->m_pkthdr.len + optlen;
                m->m_len -= sizeof(struct ip);
                m->m_data += sizeof(struct ip);
@@ -1685,6 +1941,61 @@ ip_ctloutput(so, sopt)
                        break;
 #undef OPTSET
 
+#if CONFIG_FORCE_OUT_IFP               
+               case IP_FORCE_OUT_IFP: {
+                       char    ifname[IFNAMSIZ];
+                       ifnet_t ifp;
+                       
+                       /* Verify interface name parameter is sane */
+                       if (sopt->sopt_valsize > sizeof(ifname)) {
+                               error = EINVAL;
+                               break;
+                       }
+                       
+                       /* Copy the interface name */
+                       if (sopt->sopt_valsize != 0) {
+                               error = sooptcopyin(sopt, ifname, sizeof(ifname), sopt->sopt_valsize);
+                               if (error)
+                                       break;
+                       }
+                       
+                       if (sopt->sopt_valsize == 0 || ifname[0] == 0) {
+                               // Set pdp_ifp to NULL
+                               inp->pdp_ifp = NULL;
+                               
+                               // Flush the route
+                               if (inp->inp_route.ro_rt) {
+                                       rtfree(inp->inp_route.ro_rt);
+                                       inp->inp_route.ro_rt = NULL;
+                               }
+                               
+                               break;
+                       }
+                       
+                       /* Verify name is NULL terminated */
+                       if (ifname[sopt->sopt_valsize - 1] != 0) {
+                               error = EINVAL;
+                               break;
+                       }
+                       
+                       if (ifnet_find_by_name(ifname, &ifp) != 0) {
+                               error = ENXIO;
+                               break;
+                       }
+                       
+                       /* Won't actually free. Since we don't release this later, we should do it now. */
+                       ifnet_release(ifp);
+                       
+                       /* This only works for point-to-point interfaces */
+                       if ((ifp->if_flags & IFF_POINTOPOINT) == 0) {
+                               error = ENOTSUP;
+                               break;
+                       }
+                       
+                       inp->pdp_ifp = ifp;
+               }
+               break;
+#endif
                case IP_MULTICAST_IF:
                case IP_MULTICAST_VIF:
                case IP_MULTICAST_TTL:
@@ -1746,14 +2057,29 @@ ip_ctloutput(so, sopt)
                                len = m->m_len;
                        }
                        optname = sopt->sopt_name;
-                       lck_mtx_lock(sadb_mutex);
                        error = ipsec4_set_policy(inp, optname, req, len, priv);
-                       lck_mtx_unlock(sadb_mutex);
                        m_freem(m);
                        break;
                }
 #endif /*IPSEC*/
 
+#if TRAFFIC_MGT
+               case IP_TRAFFIC_MGT_BACKGROUND:
+               {
+                       unsigned        background = 0;
+                       error = sooptcopyin(sopt, &background, sizeof(background), sizeof(background));
+                       if (error) 
+                               break;
+
+                       if (background) 
+                               so->so_traffic_mgt_flags |= TRAFFIC_MGT_SO_BACKGROUND;
+                       else 
+                               so->so_traffic_mgt_flags &= ~TRAFFIC_MGT_SO_BACKGROUND;
+
+                       break;
+               }
+#endif /* TRAFFIC_MGT */
+
                default:
                        error = ENOPROTOOPT;
                        break;
@@ -1854,9 +2180,7 @@ ip_ctloutput(so, sopt)
                                req = mtod(m, caddr_t);
                                len = m->m_len;
                        }
-                       lck_mtx_lock(sadb_mutex);
                        error = ipsec4_get_policy(sotoinpcb(so), req, len, &m);
-                       lck_mtx_unlock(sadb_mutex);
                        if (error == 0)
                                error = soopt_mcopyout(sopt, m); /* XXX */
                        if (error == 0)
@@ -1865,6 +2189,15 @@ ip_ctloutput(so, sopt)
                }
 #endif /*IPSEC*/
 
+#if TRAFFIC_MGT
+               case IP_TRAFFIC_MGT_BACKGROUND:
+               {
+                       unsigned        background = so->so_traffic_mgt_flags;
+                       return (sooptcopyout(sopt, &background, sizeof(background)));
+                       break;
+               }
+#endif /* TRAFFIC_MGT */
+
                default:
                        error = ENOPROTOOPT;
                        break;
@@ -1880,10 +2213,10 @@ ip_ctloutput(so, sopt)
  * with destination address if source routed.
  */
 static int
-ip_pcbopts(optname, pcbopt, m)
-       int optname;
-       struct mbuf **pcbopt;
-       register struct mbuf *m;
+ip_pcbopts(
+       __unused int optname,
+       struct mbuf **pcbopt,
+       register struct mbuf *m)
 {
        register int cnt, optlen;
        register u_char *cp;
@@ -2046,6 +2379,7 @@ ip_setmoptions(sopt, imop)
 
        switch (sopt->sopt_name) {
        /* store an index number for the vif you wanna use in the send */
+#if MROUTING
        case IP_MULTICAST_VIF:
                if (legal_vif_num == 0) {
                        error = EOPNOTSUPP;
@@ -2060,6 +2394,7 @@ ip_setmoptions(sopt, imop)
                }
                imo->imo_multicast_vif = i;
                break;
+#endif /* MROUTING */
 
        case IP_MULTICAST_IF:
                /*
@@ -2176,7 +2511,7 @@ ip_setmoptions(sopt, imop)
         * If all options have default values, no need to keep the mbuf.
         */
        if (imo->imo_multicast_ifp == NULL &&
-           imo->imo_multicast_vif == -1 &&
+           imo->imo_multicast_vif == (u_long)-1 &&
            imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL &&
            imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP &&
            imo->imo_num_memberships == 0) {
@@ -2239,15 +2574,17 @@ ip_addmembership(
                dst->sin_len = sizeof(*dst);
                dst->sin_family = AF_INET;
                dst->sin_addr = mreq->imr_multiaddr;
-               rtalloc(&ro);
+               lck_mtx_lock(rt_mtx);
+               rtalloc_ign_locked(&ro, 0UL);
                if (ro.ro_rt != NULL) {
                        ifp = ro.ro_rt->rt_ifp;
-                       rtfree(ro.ro_rt);
+                       rtfree_locked(ro.ro_rt);
                }
                else {
                        /* If there's no default route, try using loopback */
                        mreq->imr_interface.s_addr = INADDR_LOOPBACK;
                }
+               lck_mtx_unlock(rt_mtx);
        }
        
        if (ifp == NULL) {
@@ -2368,6 +2705,7 @@ ip_getmoptions(sopt, imo)
 
        error = 0;
        switch (sopt->sopt_name) {
+#if MROUTING
        case IP_MULTICAST_VIF: 
                if (imo != NULL)
                        optval = imo->imo_multicast_vif;
@@ -2375,6 +2713,7 @@ ip_getmoptions(sopt, imo)
                        optval = -1;
                error = sooptcopyout(sopt, &optval, sizeof optval);
                break;
+#endif /* MROUTING */
 
        case IP_MULTICAST_IF:
                if (imo == NULL || imo->imo_multicast_ifp == NULL)
@@ -2451,83 +2790,83 @@ ip_mloopback(ifp, m, dst, hlen)
 {
        register struct ip *ip;
        struct mbuf *copym;
+       int sw_csum = (apple_hwcksum_tx == 0);
 
        copym = m_copy(m, 0, M_COPYALL);
        if (copym != NULL && (copym->m_flags & M_EXT || copym->m_len < hlen))
                copym = m_pullup(copym, hlen);
-       if (copym != NULL) {
-               /*
-                * We don't bother to fragment if the IP length is greater
-                * than the interface's MTU.  Can this possibly matter?
-                */
-               ip = mtod(copym, struct ip *);
-               HTONS(ip->ip_len);
-               HTONS(ip->ip_off);
-               ip->ip_sum = 0;
-               ip->ip_sum = in_cksum(copym, hlen);
-               /*
-                * NB:
-                * It's not clear whether there are any lingering
-                * reentrancy problems in other areas which might
-                * be exposed by using ip_input directly (in
-                * particular, everything which modifies the packet
-                * in-place).  Yet another option is using the
-                * protosw directly to deliver the looped back
-                * packet.  For the moment, we'll err on the side
-                * of safety by using if_simloop().
-                */
+
+       if (copym == NULL)
+               return;
+
+       /*
+        * We don't bother to fragment if the IP length is greater
+        * than the interface's MTU.  Can this possibly matter?
+        */
+       ip = mtod(copym, struct ip *);
+       HTONS(ip->ip_len);
+       HTONS(ip->ip_off);
+       ip->ip_sum = 0;
+       ip->ip_sum = in_cksum(copym, hlen);
+       /*
+        * NB:
+        * It's not clear whether there are any lingering
+        * reentrancy problems in other areas which might
+        * be exposed by using ip_input directly (in
+        * particular, everything which modifies the packet
+        * in-place).  Yet another option is using the
+        * protosw directly to deliver the looped back
+        * packet.  For the moment, we'll err on the side
+        * of safety by using if_simloop().
+        */
 #if 1 /* XXX */
-               if (dst->sin_family != AF_INET) {
-                       printf("ip_mloopback: bad address family %d\n",
-                                               dst->sin_family);
-                       dst->sin_family = AF_INET;
-               }
+       if (dst->sin_family != AF_INET) {
+               printf("ip_mloopback: bad address family %d\n",
+                                       dst->sin_family);
+               dst->sin_family = AF_INET;
+       }
 #endif
 
-
         /*
-        * Mark checksum as valid or calculate checksum for loopback.
-        * 
-        * This is done this way because we have to embed the ifp of
-        * the interface we will send the original copy of the packet
-        * out on in the mbuf. ip_input will check if_hwassist of the
-        * embedded ifp and ignore all csum_flags if if_hwassist is 0.
-        * The UDP checksum has not been calculated yet.
-        */
-        if (copym->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
-            if (IF_HWASSIST_CSUM_FLAGS(ifp->if_hwassist)) {
-                copym->m_pkthdr.csum_flags |=
-                    CSUM_DATA_VALID | CSUM_PSEUDO_HDR |
-                    CSUM_IP_CHECKED | CSUM_IP_VALID;
-                copym->m_pkthdr.csum_data = 0xffff;
-            } else {
-               NTOHS(ip->ip_len);
-                in_delayed_cksum(copym);
-               HTONS(ip->ip_len);
-           }
-        }
-
-
-               /*
-                * TedW: 
-                * We need to send all loopback traffic down to dlil in case 
-                * a filter has tapped-in.
-                */
-
-               /*
-                * Stuff the 'real' ifp into the pkthdr, to be used in matching
-                *  in ip_input(); we need the loopback ifp/dl_tag passed as args
-                *  to make the loopback driver compliant with the data link
-                *  requirements.
-                */
-               if (lo_ifp) {
-                       copym->m_pkthdr.rcvif = ifp;
-                   dlil_output(lo_ifp, PF_INET, copym, 0, (struct sockaddr *) dst, 0);
+         * Mark checksum as valid or calculate checksum for loopback.
+         *
+         * This is done this way because we have to embed the ifp of
+         * the interface we will send the original copy of the packet
+         * out on in the mbuf. ip_input will check if_hwassist of the
+         * embedded ifp and ignore all csum_flags if if_hwassist is 0.
+         * The UDP checksum has not been calculated yet.
+         */
+        if (sw_csum || (copym->m_pkthdr.csum_flags & CSUM_DELAY_DATA)) {
+               if (!sw_csum && IF_HWASSIST_CSUM_FLAGS(ifp->if_hwassist)) {
+                       copym->m_pkthdr.csum_flags |=
+                           CSUM_DATA_VALID | CSUM_PSEUDO_HDR |
+                           CSUM_IP_CHECKED | CSUM_IP_VALID;
+                       copym->m_pkthdr.csum_data = 0xffff;
                } else {
-                   printf("Warning: ip_output call to dlil_find_dltag failed!\n");
-                   m_freem(copym);
+                       NTOHS(ip->ip_len);
+                       in_delayed_cksum(copym);
+                       HTONS(ip->ip_len);
                }
+        }
+
+       /*
+        * TedW: 
+        * We need to send all loopback traffic down to dlil in case 
+        * a filter has tapped-in.
+        */
 
-/*             if_simloop(ifp, copym, (struct sockaddr *)dst, 0);*/
+       /*
+        * Stuff the 'real' ifp into the pkthdr, to be used in matching
+        *  in ip_input(); we need the loopback ifp/dl_tag passed as args
+        *  to make the loopback driver compliant with the data link
+        *  requirements.
+        */
+       if (lo_ifp) {
+               copym->m_pkthdr.rcvif = ifp;
+               dlil_output(lo_ifp, PF_INET, copym, 0,
+                   (struct sockaddr *) dst, 0);
+       } else {
+               printf("Warning: ip_output call to dlil_find_dltag failed!\n");
+               m_freem(copym);
        }
 }