xnu-4570.51.1.tar.gz

[apple/xnu.git] / bsd / netinet / ip_output.c

/*
 * Copyright (c) 2000-2017 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
 * 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.
 *
 * 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, 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@
 */
/*
 * Copyright (c) 1982, 1986, 1988, 1990, 1993
 *	The Regents of the University of California.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *	This product includes software developed by the University of
 *	California, Berkeley and its contributors.
 * 4. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *	@(#)ip_output.c	8.3 (Berkeley) 1/21/94
 */
/*
 * 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 <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <kern/locks.h>
#include <sys/sysctl.h>
#include <sys/mcache.h>
#include <sys/kdebug.h>

#include <machine/endian.h>
#include <pexpert/pexpert.h>
#include <mach/sdt.h>

#include <libkern/OSAtomic.h>
#include <libkern/OSByteOrder.h>

#include <net/if.h>
#include <net/if_dl.h>
#include <net/if_types.h>
#include <net/route.h>
#include <net/ntstat.h>
#include <net/net_osdep.h>
#include <net/dlil.h>
#include <net/net_perf.h>

#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/in_pcb.h>
#include <netinet/in_var.h>
#include <netinet/ip_var.h>
#include <netinet/kpi_ipfilter_var.h>
#include <netinet/in_tclass.h>

#if CONFIG_MACF_NET
#include <security/mac_framework.h>
#endif /* CONFIG_MACF_NET */

#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)

#if IPSEC
#include <netinet6/ipsec.h>
#include <netkey/key.h>
#if IPSEC_DEBUG
#include <netkey/key_debug.h>
#else
#define	KEYDEBUG(lev, arg)
#endif
#endif /* IPSEC */

#if NECP
#include <net/necp.h>
#endif /* NECP */

#if IPFIREWALL
#include <netinet/ip_fw.h>
#if IPDIVERT
#include <netinet/ip_divert.h>
#endif /* IPDIVERT */
#endif /* IPFIREWALL */

#if DUMMYNET
#include <netinet/ip_dummynet.h>
#endif

#if PF
#include <net/pfvar.h>
#endif /* PF */

#if IPFIREWALL_FORWARD && IPFIREWALL_FORWARD_DEBUG
#define	print_ip(a)	\
	printf("%ld.%ld.%ld.%ld", (ntohl(a.s_addr) >> 24) & 0xFF,	\
	    (ntohl(a.s_addr) >> 16) & 0xFF,				\
	    (ntohl(a.s_addr) >> 8) & 0xFF,				\
	    (ntohl(a.s_addr)) & 0xFF);
#endif /* IPFIREWALL_FORWARD && IPFIREWALL_FORWARD_DEBUG */

u_short ip_id;

static int sysctl_reset_ip_output_stats SYSCTL_HANDLER_ARGS;
static int sysctl_ip_output_measure_bins SYSCTL_HANDLER_ARGS;
static int sysctl_ip_output_getperf SYSCTL_HANDLER_ARGS;
static void ip_out_cksum_stats(int, u_int32_t);
static struct mbuf *ip_insertoptions(struct mbuf *, struct mbuf *, int *);
static int ip_optcopy(struct ip *, struct ip *);
static int ip_pcbopts(int, struct mbuf **, struct mbuf *);
static void imo_trace(struct ip_moptions *, int);
static void ip_mloopback(struct ifnet *, struct ifnet *, struct mbuf *,
    struct sockaddr_in *, int);
static struct ifaddr *in_selectsrcif(struct ip *, struct route *, unsigned int);

extern struct ip_linklocal_stat ip_linklocal_stat;

/* temporary: for testing */
#if IPSEC
extern int ipsec_bypass;
#endif

static int ip_maxchainsent = 0;
SYSCTL_INT(_net_inet_ip, OID_AUTO, maxchainsent,
	CTLFLAG_RW | CTLFLAG_LOCKED, &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 | CTLFLAG_LOCKED, &forge_ce, 0,
	"Forge ECN CE");
#endif /* DEBUG */

static int ip_select_srcif_debug = 0;
SYSCTL_INT(_net_inet_ip, OID_AUTO, select_srcif_debug,
	CTLFLAG_RW | CTLFLAG_LOCKED, &ip_select_srcif_debug, 0,
	"log source interface selection debug info");

static int ip_output_measure = 0;
SYSCTL_PROC(_net_inet_ip, OID_AUTO, output_perf,
	CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
	&ip_output_measure, 0, sysctl_reset_ip_output_stats, "I",
	"Do time measurement");

static uint64_t ip_output_measure_bins = 0;
SYSCTL_PROC(_net_inet_ip, OID_AUTO, output_perf_bins,
	CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED, &ip_output_measure_bins, 0,
	sysctl_ip_output_measure_bins, "I",
	"bins for chaining performance data histogram");

static net_perf_t net_perf;
SYSCTL_PROC(_net_inet_ip, OID_AUTO, output_perf_data,
	CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED,
	0, 0, sysctl_ip_output_getperf, "S,net_perf",
	"IP output performance data (struct net_perf, net/net_perf.h)");

__private_extern__ int rfc6864 = 1;
SYSCTL_INT(_net_inet_ip, OID_AUTO, rfc6864, CTLFLAG_RW | CTLFLAG_LOCKED,
	&rfc6864, 0, "updated ip id field behavior");

#define	IMO_TRACE_HIST_SIZE	32	/* size of trace history */

/* For gdb */
__private_extern__ unsigned int imo_trace_hist_size = IMO_TRACE_HIST_SIZE;

struct ip_moptions_dbg {
	struct ip_moptions	imo;			/* ip_moptions */
	u_int16_t		imo_refhold_cnt;	/* # of IMO_ADDREF */
	u_int16_t		imo_refrele_cnt;	/* # of IMO_REMREF */
	/*
	 * Alloc and free callers.
	 */
	ctrace_t		imo_alloc;
	ctrace_t		imo_free;
	/*
	 * Circular lists of IMO_ADDREF and IMO_REMREF callers.
	 */
	ctrace_t		imo_refhold[IMO_TRACE_HIST_SIZE];
	ctrace_t		imo_refrele[IMO_TRACE_HIST_SIZE];
};

#if DEBUG
static unsigned int imo_debug = 1;	/* debugging (enabled) */
#else
static unsigned int imo_debug;		/* debugging (disabled) */
#endif /* !DEBUG */
static unsigned int imo_size;		/* size of zone element */
static struct zone *imo_zone;		/* zone for ip_moptions */

#define	IMO_ZONE_MAX		64		/* maximum elements in zone */
#define	IMO_ZONE_NAME		"ip_moptions"	/* zone name */

/*
 * IP output.  The packet in mbuf chain m contains a skeletal IP
 * header (with len, off, ttl, proto, tos, src, dst).
 * The mbuf chain containing the packet will be freed.
 * The mbuf opt, if present, will not be freed.
 */
int
ip_output(struct mbuf *m0, struct mbuf *opt, struct route *ro, int flags,
    struct ip_moptions *imo, struct ip_out_args *ipoa)
{
	return (ip_output_list(m0, 0, opt, ro, flags, imo, ipoa));
}

/*
 * IP output.  The packet in mbuf chain m contains a skeletal IP
 * header (with len, off, ttl, proto, tos, src, dst).
 * The mbuf chain containing the packet will be freed.
 * The mbuf opt, if present, will not be freed.
 *
 * Route ro MUST be non-NULL; if ro->ro_rt is valid, route lookup would be
 * skipped and ro->ro_rt would be used.  Otherwise the result of route
 * lookup is stored in ro->ro_rt.
 *
 * In the IP forwarding case, the packet will arrive with options already
 * inserted, so must have a NULL opt pointer.
 */
int
ip_output_list(struct mbuf *m0, int packetchain, struct mbuf *opt,
    struct route *ro, int flags, struct ip_moptions *imo,
    struct ip_out_args *ipoa)
{
	struct ip *ip;
	struct ifnet *ifp = NULL;		/* not refcnt'd */
	struct mbuf *m = m0, *prevnxt = NULL, **mppn = &prevnxt;
	int hlen = sizeof (struct ip);
	int len = 0, error = 0;
	struct sockaddr_in *dst = NULL;
	struct in_ifaddr *ia = NULL, *src_ia = NULL;
	struct in_addr pkt_dst;
	struct ipf_pktopts *ippo = NULL;
	ipfilter_t inject_filter_ref = NULL;
	struct mbuf *packetlist;
	uint32_t sw_csum, pktcnt = 0, scnt = 0, bytecnt = 0;
	uint32_t packets_processed = 0;
	unsigned int ifscope = IFSCOPE_NONE;
	struct flowadv *adv = NULL;
	struct timeval start_tv;
#if IPSEC
	struct socket *so = NULL;
	struct secpolicy *sp = NULL;
#endif /* IPSEC */
#if NECP
	necp_kernel_policy_result necp_result = 0;
	necp_kernel_policy_result_parameter necp_result_parameter;
	necp_kernel_policy_id necp_matched_policy_id = 0;
#endif /* NECP */
#if IPFIREWALL
	int ipfwoff;
	struct sockaddr_in *next_hop_from_ipfwd_tag = NULL;
#endif /* IPFIREWALL */
#if IPFIREWALL || DUMMYNET
	struct m_tag *tag;
#endif /* IPFIREWALL || DUMMYNET */
#if DUMMYNET
	struct ip_out_args saved_ipoa;
	struct sockaddr_in dst_buf;
#endif /* DUMMYNET */
	struct {
#if IPSEC
		struct ipsec_output_state ipsec_state;
#endif /* IPSEC */
#if NECP
		struct route necp_route;
#endif /* NECP */
#if IPFIREWALL || DUMMYNET
		struct ip_fw_args args;
#endif /* IPFIREWALL || DUMMYNET */
#if IPFIREWALL_FORWARD
		struct route sro_fwd;
#endif /* IPFIREWALL_FORWARD */
#if DUMMYNET
		struct route saved_route;
#endif /* DUMMYNET */
		struct ipf_pktopts ipf_pktopts;
	} ipobz;
#define	ipsec_state	ipobz.ipsec_state
#define	necp_route	ipobz.necp_route
#define	args		ipobz.args
#define	sro_fwd		ipobz.sro_fwd
#define	saved_route	ipobz.saved_route
#define	ipf_pktopts	ipobz.ipf_pktopts
	union {
		struct {
			boolean_t select_srcif : 1;	/* set once */
			boolean_t srcbound : 1;		/* set once */
			boolean_t nocell : 1;		/* set once */
			boolean_t isbroadcast : 1;
			boolean_t didfilter : 1;
			boolean_t noexpensive : 1;	/* set once */
			boolean_t awdl_unrestricted : 1;	/* set once */
#if IPFIREWALL_FORWARD
			boolean_t fwd_rewrite_src : 1;
#endif /* IPFIREWALL_FORWARD */
		};
		uint32_t raw;
	} ipobf = { .raw = 0 };

/*
 * Here we check for restrictions when sending frames.
 * N.B.: IPv4 over internal co-processor interfaces is not allowed.
 */
#define	IP_CHECK_RESTRICTIONS(_ifp, _ipobf) 				\
	(((_ipobf).nocell && IFNET_IS_CELLULAR(_ifp)) ||		\
	 ((_ipobf).noexpensive && IFNET_IS_EXPENSIVE(_ifp)) ||		\
          (IFNET_IS_INTCOPROC(_ifp)) ||					\
	 (!(_ipobf).awdl_unrestricted && IFNET_IS_AWDL_RESTRICTED(_ifp)))

	if (ip_output_measure)
		net_perf_start_time(&net_perf, &start_tv);
	KERNEL_DEBUG(DBG_FNC_IP_OUTPUT | DBG_FUNC_START, 0, 0, 0, 0, 0);

	VERIFY(m0->m_flags & M_PKTHDR);
	packetlist = m0;

	/* zero out {ipsec_state, args, sro_fwd, saved_route, ipf_pktops} */
	bzero(&ipobz, sizeof (ipobz));
	ippo = &ipf_pktopts;

#if IPFIREWALL || DUMMYNET
	if (SLIST_EMPTY(&m0->m_pkthdr.tags))
		goto ipfw_tags_done;

	/* Grab info from mtags prepended to the chain */
#if DUMMYNET
	if ((tag = m_tag_locate(m0, KERNEL_MODULE_TAG_ID,
	    KERNEL_TAG_TYPE_DUMMYNET, NULL)) != NULL) {
		struct dn_pkt_tag	*dn_tag;

		dn_tag = (struct dn_pkt_tag *)(tag+1);
		args.fwa_ipfw_rule = dn_tag->dn_ipfw_rule;
		args.fwa_pf_rule = dn_tag->dn_pf_rule;
		opt = NULL;
		saved_route = dn_tag->dn_ro;
		ro = &saved_route;

		imo = NULL;
		bcopy(&dn_tag->dn_dst, &dst_buf, sizeof (dst_buf));
		dst = &dst_buf;
		ifp = dn_tag->dn_ifp;
		flags = dn_tag->dn_flags;
		if ((dn_tag->dn_flags & IP_OUTARGS)) {
			saved_ipoa = dn_tag->dn_ipoa;
			ipoa = &saved_ipoa;
		}

		m_tag_delete(m0, tag);
	}
#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;

		div_tag = (struct divert_tag *)(tag+1);
		args.fwa_divert_rule = div_tag->cookie;

		m_tag_delete(m0, tag);
	}
#endif /* IPDIVERT */

#if IPFIREWALL
	if ((tag = m_tag_locate(m0, KERNEL_MODULE_TAG_ID,
	    KERNEL_TAG_TYPE_IPFORWARD, NULL)) != NULL) {
		struct ip_fwd_tag	*ipfwd_tag;

		ipfwd_tag = (struct ip_fwd_tag *)(tag+1);
		next_hop_from_ipfwd_tag = ipfwd_tag->next_hop;

		m_tag_delete(m0, tag);
	}
#endif /* IPFIREWALL */

ipfw_tags_done:
#endif /* IPFIREWALL || DUMMYNET */

	m = m0;
	m->m_pkthdr.pkt_flags &= ~(PKTF_LOOP|PKTF_IFAINFO);

#if IPSEC
	if (ipsec_bypass == 0 && !(flags & IP_NOIPSEC)) {
		/* If packet is bound to an interface, check bound policies */
		if ((flags & IP_OUTARGS) && (ipoa != NULL) &&
			(ipoa->ipoa_flags & IPOAF_BOUND_IF) &&
			ipoa->ipoa_boundif != IFSCOPE_NONE) {
			if (ipsec4_getpolicybyinterface(m, IPSEC_DIR_OUTBOUND,
				&flags, ipoa, &sp) != 0)
				goto bad;
		}
	}
#endif /* IPSEC */

	VERIFY(ro != NULL);

	if (flags & IP_OUTARGS) {
		/*
		 * In the forwarding case, only the ifscope value is used,
		 * as source interface selection doesn't take place.
		 */
		if ((ipobf.select_srcif = (!(flags & IP_FORWARDING) &&
		    (ipoa->ipoa_flags & IPOAF_SELECT_SRCIF)))) {
			ipf_pktopts.ippo_flags |= IPPOF_SELECT_SRCIF;
		}

		if ((ipoa->ipoa_flags & IPOAF_BOUND_IF) &&
		    ipoa->ipoa_boundif != IFSCOPE_NONE) {
			ifscope = ipoa->ipoa_boundif;
			ipf_pktopts.ippo_flags |=
			    (IPPOF_BOUND_IF | (ifscope << IPPOF_SHIFT_IFSCOPE));
		}

		/* double negation needed for bool bit field */
		ipobf.srcbound = !!(ipoa->ipoa_flags & IPOAF_BOUND_SRCADDR);
		if (ipobf.srcbound)
			ipf_pktopts.ippo_flags |= IPPOF_BOUND_SRCADDR;
	} else {
		ipobf.select_srcif = FALSE;
		ipobf.srcbound = FALSE;
		ifscope = IFSCOPE_NONE;
		if (flags & IP_OUTARGS) {
			ipoa->ipoa_boundif = IFSCOPE_NONE;
			ipoa->ipoa_flags &= ~(IPOAF_SELECT_SRCIF |
			    IPOAF_BOUND_IF | IPOAF_BOUND_SRCADDR);
		}
	}

	if (flags & IP_OUTARGS) {
		if (ipoa->ipoa_flags & IPOAF_NO_CELLULAR) {
			ipobf.nocell = TRUE;
			ipf_pktopts.ippo_flags |= IPPOF_NO_IFT_CELLULAR;
		}
		if (ipoa->ipoa_flags & IPOAF_NO_EXPENSIVE) {
			ipobf.noexpensive = TRUE;
			ipf_pktopts.ippo_flags |= IPPOF_NO_IFF_EXPENSIVE;
		}
		if (ipoa->ipoa_flags & IPOAF_AWDL_UNRESTRICTED)
			ipobf.awdl_unrestricted = TRUE;
		adv = &ipoa->ipoa_flowadv;
		adv->code = FADV_SUCCESS;
		ipoa->ipoa_retflags = 0;
	}

#if IPSEC
	if (ipsec_bypass == 0 && !(flags & IP_NOIPSEC)) {
		so = ipsec_getsocket(m);
		if (so != NULL) {
			(void) ipsec_setsocket(m, NULL);
		}
	}
#endif /* IPSEC */

#if DUMMYNET
	if (args.fwa_ipfw_rule != NULL || args.fwa_pf_rule != NULL) {
		/* dummynet already saw us */
		ip = mtod(m, struct ip *);
		hlen = IP_VHL_HL(ip->ip_vhl) << 2;
		pkt_dst = ip->ip_dst;
		if (ro->ro_rt != NULL) {
			RT_LOCK_SPIN(ro->ro_rt);
			ia = (struct in_ifaddr *)ro->ro_rt->rt_ifa;
			if (ia) {
				/* Become a regular mutex */
				RT_CONVERT_LOCK(ro->ro_rt);
				IFA_ADDREF(&ia->ia_ifa);
			}
			RT_UNLOCK(ro->ro_rt);
		}

#if IPFIREWALL
		if (args.fwa_ipfw_rule != NULL)
			goto skip_ipsec;
#endif /* IPFIREWALL  */
		if (args.fwa_pf_rule != NULL)
			goto sendit;
	}
#endif /* DUMMYNET */

loopit:
	packets_processed++;
	ipobf.isbroadcast = FALSE;
	ipobf.didfilter = FALSE;
#if IPFIREWALL_FORWARD
	ipobf.fwd_rewrite_src = FALSE;
#endif /* IPFIREWALL_FORWARD */

	VERIFY(m->m_flags & M_PKTHDR);
	/*
	 * No need to proccess packet twice if we've already seen it.
	 */
	if (!SLIST_EMPTY(&m->m_pkthdr.tags))
		inject_filter_ref = ipf_get_inject_filter(m);
	else
		inject_filter_ref = NULL;

	if (opt) {
		m = ip_insertoptions(m, opt, &len);
		hlen = len;
		/* Update the chain */
		if (m != m0) {
			if (m0 == packetlist)
				packetlist = m;
			m0 = m;
		}
	}
	ip = mtod(m, struct ip *);

#if IPFIREWALL
	/*
	 * rdar://8542331
	 *
	 * When dealing with a packet chain, we need to reset "next_hop"
	 * because "dst" may have been changed to the gateway address below
	 * for the previous packet of the chain. This could cause the route
	 * to be inavertandly changed to the route to the gateway address
	 * (instead of the route to the destination).
	 */
	args.fwa_next_hop = next_hop_from_ipfwd_tag;
	pkt_dst = args.fwa_next_hop ? args.fwa_next_hop->sin_addr : ip->ip_dst;
#else /* !IPFIREWALL */
	pkt_dst = ip->ip_dst;
#endif /* !IPFIREWALL */

	/*
	 * We must not send if the packet is destined to network zero.
	 * RFC1122 3.2.1.3 (a) and (b).
	 */
	if (IN_ZERONET(ntohl(pkt_dst.s_addr))) {
		error = EHOSTUNREACH;
		goto bad;
	}

	/*
	 * Fill in IP header.
	 */
	if (!(flags & (IP_FORWARDING|IP_RAWOUTPUT))) {
		ip->ip_vhl = IP_MAKE_VHL(IPVERSION, hlen >> 2);
		ip->ip_off &= IP_DF;
		if (rfc6864 && IP_OFF_IS_ATOMIC(ip->ip_off)) {
			// Per RFC6864, value of ip_id is undefined for atomic ip packets
			ip->ip_id = 0;
		} else {
			ip->ip_id = ip_randomid();
		}
		OSAddAtomic(1, &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 = SIN(&ro->ro_dst);

	/*
	 * If there is a cached route,
	 * check that it is to the same destination
	 * and is still up.  If not, free it and try again.
	 * The address family should also be checked in case of sharing the
	 * cache with IPv6.
	 */

	if (ro->ro_rt != NULL) {
		if (ROUTE_UNUSABLE(ro) && ip->ip_src.s_addr != INADDR_ANY &&
		    !(flags & (IP_ROUTETOIF | IP_FORWARDING))) {
			src_ia = ifa_foraddr(ip->ip_src.s_addr);
			if (src_ia == NULL) {
				error = EADDRNOTAVAIL;
				goto bad;
			}
			IFA_REMREF(&src_ia->ia_ifa);
			src_ia = NULL;
		}
		/*
		 * Test rt_flags without holding rt_lock for performance
		 * reasons; if the route is down it will hopefully be
		 * caught by the layer below (since it uses this route
		 * as a hint) or during the next transmit.
		 */
		if (ROUTE_UNUSABLE(ro) || dst->sin_family != AF_INET ||
		    dst->sin_addr.s_addr != pkt_dst.s_addr)
			ROUTE_RELEASE(ro);

		/*
		 * If we're doing source interface selection, we may not
		 * want to use this route; only synch up the generation
		 * count otherwise.
		 */
		if (!ipobf.select_srcif && ro->ro_rt != NULL &&
		    RT_GENID_OUTOFSYNC(ro->ro_rt))
			RT_GENID_SYNC(ro->ro_rt);
	}
	if (ro->ro_rt == NULL) {
		bzero(dst, sizeof (*dst));
		dst->sin_family = AF_INET;
		dst->sin_len = sizeof (*dst);
		dst->sin_addr = pkt_dst;
	}
	/*
	 * If routing to interface only,
	 * short circuit routing lookup.
	 */
	if (flags & IP_ROUTETOIF) {
		if (ia != NULL)
			IFA_REMREF(&ia->ia_ifa);
		if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == NULL) {
			ia = ifatoia(ifa_ifwithnet(sintosa(dst)));
			if (ia == NULL) {
				OSAddAtomic(1, &ipstat.ips_noroute);
				error = ENETUNREACH;
				/* XXX IPv6 APN fallback notification?? */
				goto bad;
			}
		}
		ifp = ia->ia_ifp;
		ip->ip_ttl = 1;
		ipobf.isbroadcast = in_broadcast(dst->sin_addr, ifp);
		/*
		 * For consistency with other cases below.  Loopback
		 * multicast case is handled separately by ip_mloopback().
		 */
		if ((ifp->if_flags & IFF_LOOPBACK) &&
		    !IN_MULTICAST(ntohl(pkt_dst.s_addr))) {
			m->m_pkthdr.rcvif = ifp;
			ip_setsrcifaddr_info(m, ifp->if_index, NULL);
			ip_setdstifaddr_info(m, ifp->if_index, NULL);
		}
	} else if (IN_MULTICAST(ntohl(pkt_dst.s_addr)) &&
	    imo != NULL && (ifp = imo->imo_multicast_ifp) != NULL) {
		/*
		 * Bypass the normal routing lookup for multicast
		 * packets if the interface is specified.
		 */
		ipobf.isbroadcast = FALSE;
		if (ia != NULL)
			IFA_REMREF(&ia->ia_ifa);

		/* Macro takes reference on ia */
		IFP_TO_IA(ifp, ia);
	} else {
		struct ifaddr *ia0 = NULL;
		boolean_t cloneok = FALSE;
		/*
		 * Perform source interface selection; the source IP address
		 * must belong to one of the addresses of the interface used
		 * by the route.  For performance reasons, do this only if
		 * there is no route, or if the routing table has changed,
		 * or if we haven't done source interface selection on this
		 * route (for this PCB instance) before.
		 */
		if (ipobf.select_srcif &&
		    ip->ip_src.s_addr != INADDR_ANY && (ROUTE_UNUSABLE(ro) ||
		    !(ro->ro_flags & ROF_SRCIF_SELECTED))) {
			/* Find the source interface */
			ia0 = in_selectsrcif(ip, ro, ifscope);

			/*
			 * If the source address belongs to a restricted
			 * interface and the caller forbids our using
			 * interfaces of such type, pretend that there is no
			 * route.
			 */
			if (ia0 != NULL &&
			    IP_CHECK_RESTRICTIONS(ia0->ifa_ifp, ipobf)) {
				IFA_REMREF(ia0);
				ia0 = NULL;
				error = EHOSTUNREACH;
				if (flags & IP_OUTARGS)
					ipoa->ipoa_retflags |= IPOARF_IFDENIED;
				goto bad;
			}

			/*
			 * If the source address is spoofed (in the case of
			 * IP_RAWOUTPUT on an unbounded socket), or if this
			 * is destined for local/loopback, just let it go out
			 * using the interface of the route.  Otherwise,
			 * there's no interface having such an address,
			 * so bail out.
			 */
			if (ia0 == NULL && (!(flags & IP_RAWOUTPUT) ||
			    ipobf.srcbound) && ifscope != lo_ifp->if_index) {
				error = EADDRNOTAVAIL;
				goto bad;
			}

			/*
			 * If the caller didn't explicitly specify the scope,
			 * pick it up from the source interface.  If the cached
			 * route was wrong and was blown away as part of source
			 * interface selection, don't mask out RTF_PRCLONING
			 * since that route may have been allocated by the ULP,
			 * unless the IP header was created by the caller or
			 * the destination is IPv4 LLA.  The check for the
			 * latter is needed because IPv4 LLAs are never scoped
			 * in the current implementation, and we don't want to
			 * replace the resolved IPv4 LLA route with one whose
			 * gateway points to that of the default gateway on
			 * the primary interface of the system.
			 */
			if (ia0 != NULL) {
				if (ifscope == IFSCOPE_NONE)
					ifscope = ia0->ifa_ifp->if_index;
				cloneok = (!(flags & IP_RAWOUTPUT) &&
				    !(IN_LINKLOCAL(ntohl(ip->ip_dst.s_addr))));
			}
		}

		/*
		 * If this is the case, we probably don't want to allocate
		 * a protocol-cloned route since we didn't get one from the
		 * ULP.  This lets TCP do its thing, while not burdening
		 * forwarding or ICMP with the overhead of cloning a route.
		 * Of course, we still want to do any cloning requested by
		 * the link layer, as this is probably required in all cases
		 * for correct operation (as it is for ARP).
		 */
		if (ro->ro_rt == NULL) {
			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. Another exception
			 * is for ULP-created routes that got blown away by
			 * source interface selection (see above).
			 *
			 * These exceptions will no longer be necessary when
			 * the RTF_PRCLONING scheme is no longer present.
			 */
			if (cloneok || dst->sin_addr.s_addr == INADDR_BROADCAST)
				ign &= ~RTF_PRCLONING;

			/*
			 * Loosen the route lookup criteria if the ifscope
			 * corresponds to the loopback interface; this is
			 * needed to support Application Layer Gateways
			 * listening on loopback, in conjunction with packet
			 * filter redirection rules.  The final source IP
			 * address will be rewritten by the packet filter
			 * prior to the RFC1122 loopback check below.
			 */
			if (ifscope == lo_ifp->if_index)
				rtalloc_ign(ro, ign);
			else
				rtalloc_scoped_ign(ro, ign, ifscope);

			/*
			 * If the route points to a cellular/expensive interface
			 * and the caller forbids our using interfaces of such type,
			 * pretend that there is no route.
			 */
			if (ro->ro_rt != NULL) {
				RT_LOCK_SPIN(ro->ro_rt);
				if (IP_CHECK_RESTRICTIONS(ro->ro_rt->rt_ifp,
				    ipobf)) {
					RT_UNLOCK(ro->ro_rt);
					ROUTE_RELEASE(ro);
					if (flags & IP_OUTARGS) {
						ipoa->ipoa_retflags |=
						    IPOARF_IFDENIED;
					}
				} else {
					RT_UNLOCK(ro->ro_rt);
				}
			}
		}

		if (ro->ro_rt == NULL) {
			OSAddAtomic(1, &ipstat.ips_noroute);
			error = EHOSTUNREACH;
			if (ia0 != NULL) {
				IFA_REMREF(ia0);
				ia0 = NULL;
			}
			goto bad;
		}

		if (ia != NULL)
			IFA_REMREF(&ia->ia_ifa);
		RT_LOCK_SPIN(ro->ro_rt);
		ia = ifatoia(ro->ro_rt->rt_ifa);
		if (ia != NULL) {
			/* Become a regular mutex */
			RT_CONVERT_LOCK(ro->ro_rt);
			IFA_ADDREF(&ia->ia_ifa);
		}
		/*
		 * Note: ia_ifp may not be the same as rt_ifp; the latter
		 * is what we use for determining outbound i/f, mtu, etc.
		 */
		ifp = ro->ro_rt->rt_ifp;
		ro->ro_rt->rt_use++;
		if (ro->ro_rt->rt_flags & RTF_GATEWAY) {
			dst = SIN(ro->ro_rt->rt_gateway);
		}
		if (ro->ro_rt->rt_flags & RTF_HOST) {
			/* double negation needed for bool bit field */
			ipobf.isbroadcast =
			    !!(ro->ro_rt->rt_flags & RTF_BROADCAST);
		} else {
			/* Become a regular mutex */
			RT_CONVERT_LOCK(ro->ro_rt);
			ipobf.isbroadcast = in_broadcast(dst->sin_addr, ifp);
		}
		/*
		 * For consistency with IPv6, as well as to ensure that
		 * IP_RECVIF is set correctly for packets that are sent
		 * to one of the local addresses.  ia (rt_ifa) would have
		 * been fixed up by rt_setif for local routes.  This
		 * would make it appear as if the packet arrives on the
		 * interface which owns the local address.  Loopback
		 * multicast case is handled separately by ip_mloopback().
		 */
		if (ia != NULL && (ifp->if_flags & IFF_LOOPBACK) &&
		    !IN_MULTICAST(ntohl(pkt_dst.s_addr))) {
			uint32_t srcidx;

			m->m_pkthdr.rcvif = ia->ia_ifa.ifa_ifp;

			if (ia0 != NULL)
				srcidx = ia0->ifa_ifp->if_index;
			else if ((ro->ro_flags & ROF_SRCIF_SELECTED) &&
			    ro->ro_srcia != NULL)
				srcidx = ro->ro_srcia->ifa_ifp->if_index;
			else
				srcidx = 0;

			ip_setsrcifaddr_info(m, srcidx, NULL);
			ip_setdstifaddr_info(m, 0, ia);
		}
		RT_UNLOCK(ro->ro_rt);
		if (ia0 != NULL) {
			IFA_REMREF(ia0);
			ia0 = NULL;
		}
	}

	if (IN_MULTICAST(ntohl(pkt_dst.s_addr))) {
		struct ifnet *srcifp = NULL;
		struct in_multi *inm;
		u_int32_t vif = 0;
		u_int8_t ttl = IP_DEFAULT_MULTICAST_TTL;
		u_int8_t loop = IP_DEFAULT_MULTICAST_LOOP;

		m->m_flags |= M_MCAST;
		/*
		 * IP destination address is multicast.  Make sure "dst"
		 * still points to the address in "ro".  (It may have been
		 * changed to point to a gateway address, above.)
		 */
		dst = SIN(&ro->ro_dst);
		/*
		 * See if the caller provided any multicast options
		 */
		if (imo != NULL) {
			IMO_LOCK(imo);
			vif = imo->imo_multicast_vif;
			ttl = imo->imo_multicast_ttl;
			loop = imo->imo_multicast_loop;
			if (!(flags & IP_RAWOUTPUT))
				ip->ip_ttl = ttl;
			if (imo->imo_multicast_ifp != NULL)
				ifp = imo->imo_multicast_ifp;
			IMO_UNLOCK(imo);
		} else if (!(flags & IP_RAWOUTPUT)) {
			vif = -1;
			ip->ip_ttl = ttl;
		}
		/*
		 * Confirm that the outgoing interface supports multicast.
		 */
		if (imo == NULL || vif == -1) {
			if (!(ifp->if_flags & IFF_MULTICAST)) {
				OSAddAtomic(1, &ipstat.ips_noroute);
				error = ENETUNREACH;
				goto bad;
			}
		}
		/*
		 * If source address not specified yet, use address
		 * of outgoing interface.
		 */
		if (ip->ip_src.s_addr == INADDR_ANY) {
			struct in_ifaddr *ia1;
			lck_rw_lock_shared(in_ifaddr_rwlock);
			TAILQ_FOREACH(ia1, &in_ifaddrhead, ia_link) {
				IFA_LOCK_SPIN(&ia1->ia_ifa);
				if (ia1->ia_ifp == ifp) {
					ip->ip_src = IA_SIN(ia1)->sin_addr;
					srcifp = ifp;
					IFA_UNLOCK(&ia1->ia_ifa);
					break;
				}
				IFA_UNLOCK(&ia1->ia_ifa);
			}
			lck_rw_done(in_ifaddr_rwlock);
			if (ip->ip_src.s_addr == INADDR_ANY) {
				error = ENETUNREACH;
				goto bad;
			}
		}

		in_multihead_lock_shared();
		IN_LOOKUP_MULTI(&pkt_dst, ifp, inm);
		in_multihead_lock_done();
		if (inm != NULL && (imo == NULL || loop)) {
			/*
			 * If we belong to the destination multicast group
			 * on the outgoing interface, and the caller did not
			 * forbid loopback, loop back a copy.
			 */
			if (!TAILQ_EMPTY(&ipv4_filters)) {
				struct ipfilter	*filter;
				int seen = (inject_filter_ref == NULL);

				if (imo != NULL) {
					ipf_pktopts.ippo_flags |=
					    IPPOF_MCAST_OPTS;
					ipf_pktopts.ippo_mcast_ifnet = ifp;
					ipf_pktopts.ippo_mcast_ttl = ttl;
					ipf_pktopts.ippo_mcast_loop = loop;
				}

				ipf_ref();

				/*
				 * 4135317 - always pass network byte
				 * order to filter
				 */
#if BYTE_ORDER != BIG_ENDIAN
				HTONS(ip->ip_len);
				HTONS(ip->ip_off);
#endif
				TAILQ_FOREACH(filter, &ipv4_filters, ipf_link) {
					if (seen == 0) {
						if ((struct ipfilter *)
						    inject_filter_ref == filter)
							seen = 1;
					} else if (filter->ipf_filter.
					    ipf_output != NULL) {
						errno_t result;
						result = filter->ipf_filter.
						    ipf_output(filter->
						    ipf_filter.cookie,
						    (mbuf_t *)&m, ippo);
						if (result == EJUSTRETURN) {
							ipf_unref();
							INM_REMREF(inm);
							goto done;
						}
						if (result != 0) {
							ipf_unref();
							INM_REMREF(inm);
							goto bad;
						}
					}
				}

				/* set back to host byte order */
				ip = mtod(m, struct ip *);
#if BYTE_ORDER != BIG_ENDIAN
				NTOHS(ip->ip_len);
				NTOHS(ip->ip_off);
#endif
				ipf_unref();
				ipobf.didfilter = TRUE;
			}
			ip_mloopback(srcifp, ifp, m, dst, hlen);
		}
		if (inm != NULL)
			INM_REMREF(inm);
		/*
		 * Multicasts with a time-to-live of zero may be looped-
		 * back, above, but must not be transmitted on a network.
		 * Also, multicasts addressed to the loopback interface
		 * are not sent -- the above call to ip_mloopback() will
		 * loop back a copy if this host actually belongs to the
		 * destination group on the loopback interface.
		 */
		if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) {
			m_freem(m);
			goto done;
		}

		goto sendit;
	}
	/*
	 * If source address not specified yet, use address
	 * of outgoing interface.
	 */
	if (ip->ip_src.s_addr == INADDR_ANY) {
		IFA_LOCK_SPIN(&ia->ia_ifa);
		ip->ip_src = IA_SIN(ia)->sin_addr;
		IFA_UNLOCK(&ia->ia_ifa);
#if IPFIREWALL_FORWARD
		/*
		 * Keep note that we did this - if the firewall changes
		 * the next-hop, our interface may change, changing the
		 * default source IP. It's a shame so much effort happens
		 * twice. Oh well.
		 */
		ipobf.fwd_rewrite_src = TRUE;
#endif /* IPFIREWALL_FORWARD */
	}

	/*
	 * Look for broadcast address and
	 * and verify user is allowed to send
	 * such a packet.
	 */
	if (ipobf.isbroadcast) {
		if (!(ifp->if_flags & IFF_BROADCAST)) {
			error = EADDRNOTAVAIL;
			goto bad;
		}
		if (!(flags & IP_ALLOWBROADCAST)) {
			error = EACCES;
			goto bad;
		}
		/* don't allow broadcast messages to be fragmented */
		if ((u_short)ip->ip_len > ifp->if_mtu) {
			error = EMSGSIZE;
			goto bad;
		}
		m->m_flags |= M_BCAST;
	} else {
		m->m_flags &= ~M_BCAST;
	}

sendit:
#if PF
	/* Invoke outbound packet filter */
	if (PF_IS_ENABLED) {
		int rc;

		m0 = m; /* Save for later */
#if DUMMYNET
		args.fwa_m = m;
		args.fwa_next_hop = dst;
		args.fwa_oif = ifp;
		args.fwa_ro = ro;
		args.fwa_dst = dst;
		args.fwa_oflags = flags;
		if (flags & IP_OUTARGS)
			args.fwa_ipoa = ipoa;
		rc = pf_af_hook(ifp, mppn, &m, AF_INET, FALSE, &args);
#else /* DUMMYNET */
		rc = pf_af_hook(ifp, mppn, &m, AF_INET, FALSE, NULL);
#endif /* DUMMYNET */
		if (rc != 0 || m == NULL) {
			/* Move to the next packet */
			m = *mppn;

			/* Skip ahead if first packet in list got dropped */
			if (packetlist == m0)
				packetlist = m;

			if (m != NULL) {
				m0 = m;
				/* Next packet in the chain */
				goto loopit;
			} else if (packetlist != NULL) {
				/* No more packet; send down the chain */
				goto sendchain;
			}
			/* Nothing left; we're done */
			goto done;
		}
		m0 = m;
		ip = mtod(m, struct ip *);
		pkt_dst = ip->ip_dst;
		hlen = IP_VHL_HL(ip->ip_vhl) << 2;
	}
#endif /* PF */
	/*
	 * Force IP TTL to 255 following draft-ietf-zeroconf-ipv4-linklocal.txt
	 */
	if (IN_LINKLOCAL(ntohl(ip->ip_src.s_addr)) ||
	    IN_LINKLOCAL(ntohl(ip->ip_dst.s_addr))) {
		ip_linklocal_stat.iplls_out_total++;
		if (ip->ip_ttl != MAXTTL) {
			ip_linklocal_stat.iplls_out_badttl++;
			ip->ip_ttl = MAXTTL;
		}
	}

	if (!ipobf.didfilter && !TAILQ_EMPTY(&ipv4_filters)) {
		struct ipfilter	*filter;
		int seen = (inject_filter_ref == NULL);
		ipf_pktopts.ippo_flags &= ~IPPOF_MCAST_OPTS;

		/*
		 * Check that a TSO frame isn't passed to a filter.
		 * This could happen if a filter is inserted while
		 * TCP is sending the TSO packet.
		 */
		if (m->m_pkthdr.csum_flags & CSUM_TSO_IPV4) {
			error = EMSGSIZE;
			goto bad;
		}

		ipf_ref();

		/* 4135317 - always pass network byte order to filter */
#if BYTE_ORDER != BIG_ENDIAN
		HTONS(ip->ip_len);
		HTONS(ip->ip_off);
#endif
		TAILQ_FOREACH(filter, &ipv4_filters, ipf_link) {
			if (seen == 0) {
				if ((struct ipfilter *)inject_filter_ref ==
				    filter)
					seen = 1;
			} else if (filter->ipf_filter.ipf_output) {
				errno_t result;
				result = filter->ipf_filter.
				    ipf_output(filter->ipf_filter.cookie,
				    (mbuf_t *)&m, ippo);
				if (result == EJUSTRETURN) {
					ipf_unref();
					goto done;
				}
				if (result != 0) {
					ipf_unref();
					goto bad;
				}
			}
		}
		/* set back to host byte order */
		ip = mtod(m, struct ip *);
#if BYTE_ORDER != BIG_ENDIAN
		NTOHS(ip->ip_len);
		NTOHS(ip->ip_off);
#endif
		ipf_unref();
	}

#if NECP
	/* Process Network Extension Policy. Will Pass, Drop, or Rebind packet. */
	necp_matched_policy_id = necp_ip_output_find_policy_match (m,
		flags, (flags & IP_OUTARGS) ? ipoa : NULL, &necp_result, &necp_result_parameter);
	if (necp_matched_policy_id) {
		necp_mark_packet_from_ip(m, necp_matched_policy_id);
		switch (necp_result) {
			case NECP_KERNEL_POLICY_RESULT_PASS:
				/* Check if the interface is allowed */
				if (!necp_packet_is_allowed_over_interface(m, ifp)) {
					error = EHOSTUNREACH;
					OSAddAtomic(1, &ipstat.ips_necp_policy_drop);
					goto bad;
				}
				goto skip_ipsec;
			case NECP_KERNEL_POLICY_RESULT_DROP:
			case NECP_KERNEL_POLICY_RESULT_SOCKET_DIVERT:
				/* Flow divert packets should be blocked at the IP layer */
				error = EHOSTUNREACH;
				OSAddAtomic(1, &ipstat.ips_necp_policy_drop);
				goto bad;
			case NECP_KERNEL_POLICY_RESULT_IP_TUNNEL: {
				/* Verify that the packet is being routed to the tunnel */
				struct ifnet *policy_ifp = necp_get_ifnet_from_result_parameter(&necp_result_parameter);
				if (policy_ifp == ifp) {
					/* Check if the interface is allowed */
					if (!necp_packet_is_allowed_over_interface(m, ifp)) {
						error = EHOSTUNREACH;
						OSAddAtomic(1, &ipstat.ips_necp_policy_drop);
						goto bad;
					}
					goto skip_ipsec;
				} else {
					if (necp_packet_can_rebind_to_ifnet(m, policy_ifp, &necp_route, AF_INET)) {
						/* Check if the interface is allowed */
						if (!necp_packet_is_allowed_over_interface(m, policy_ifp)) {
							error = EHOSTUNREACH;
							OSAddAtomic(1, &ipstat.ips_necp_policy_drop);
							goto bad;
						}

						/* Set ifp to the tunnel interface, since it is compatible with the packet */
						ifp = policy_ifp;
						ro = &necp_route;
						goto skip_ipsec;
					} else {
						error = ENETUNREACH;
						OSAddAtomic(1, &ipstat.ips_necp_policy_drop);
						goto bad;
					}
				}
			}
			default:
				break;
		}
	}
	/* Catch-all to check if the interface is allowed */
	if (!necp_packet_is_allowed_over_interface(m, ifp)) {
		error = EHOSTUNREACH;
		OSAddAtomic(1, &ipstat.ips_necp_policy_drop);
		goto bad;
	}
#endif /* NECP */

#if IPSEC
	if (ipsec_bypass != 0 || (flags & IP_NOIPSEC))
		goto skip_ipsec;

	KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_START, 0, 0, 0, 0, 0);

	if (sp == NULL) {
		/* get SP for this packet */
		if (so != NULL) {
			sp = ipsec4_getpolicybysock(m, IPSEC_DIR_OUTBOUND,
				so, &error);
		} else {
			sp = ipsec4_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND,
				flags, &error);
		}
		if (sp == NULL) {
			IPSEC_STAT_INCREMENT(ipsecstat.out_inval);
			KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_END,
						 0, 0, 0, 0, 0);
			goto bad;
		}
	}

	error = 0;

	/* check policy */
	switch (sp->policy) {
	case IPSEC_POLICY_DISCARD:
	case IPSEC_POLICY_GENERATE:
		/*
		 * This packet is just discarded.
		 */
		IPSEC_STAT_INCREMENT(ipsecstat.out_polvio);
		KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_END,
		    1, 0, 0, 0, 0);
		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);
		goto skip_ipsec;

	case IPSEC_POLICY_IPSEC:
		if (sp->req == NULL) {
			/* acquire a policy */
			error = key_spdacquire(sp);
			KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_END,
			    3, 0, 0, 0, 0);
			goto bad;
		}
		if (sp->ipsec_if) {
			/* Verify the redirect to ipsec interface */
			if (sp->ipsec_if == ifp) {
				goto skip_ipsec;
			}
			goto bad;
		}
		break;

	case IPSEC_POLICY_ENTRUST:
	default:
		printf("ip_output: Invalid policy found. %d\n", sp->policy);
	}
	{
	ipsec_state.m = m;
	if (flags & IP_ROUTETOIF) {
		bzero(&ipsec_state.ro, sizeof (ipsec_state.ro));
	} else {
		route_copyout((struct route *)&ipsec_state.ro, ro, sizeof (struct route));
	}
	ipsec_state.dst = SA(dst);

	ip->ip_sum = 0;

	/*
	 * XXX
	 * delayed checksums are not currently compatible with IPsec
	 */
	if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
		in_delayed_cksum(m);

#if BYTE_ORDER != BIG_ENDIAN
	HTONS(ip->ip_len);
	HTONS(ip->ip_off);
#endif

	DTRACE_IP6(send, struct mbuf *, m, struct inpcb *, NULL,
	    struct ip *, ip, struct ifnet *, ifp,
	    struct ip *, ip, struct ip6_hdr *, NULL);

	error = ipsec4_output(&ipsec_state, sp, flags);
	if (ipsec_state.tunneled == 6) {
		m0 = m = NULL;
		error = 0;
		goto bad;
	}

	m0 = m = ipsec_state.m;

#if DUMMYNET
	/*
	 * If we're about to use the route in ipsec_state
	 * and this came from dummynet, cleaup now.
	 */
	if (ro == &saved_route &&
	    (!(flags & IP_ROUTETOIF) || ipsec_state.tunneled))
		ROUTE_RELEASE(ro);
#endif /* DUMMYNET */

	if (flags & IP_ROUTETOIF) {
		/*
		 * if we have tunnel mode SA, we may need to ignore
		 * IP_ROUTETOIF.
		 */
		if (ipsec_state.tunneled) {
			flags &= ~IP_ROUTETOIF;
			ro = (struct route *)&ipsec_state.ro;
		}
	} else {
		ro = (struct route *)&ipsec_state.ro;
	}
	dst = SIN(ipsec_state.dst);
	if (error) {
		/* mbuf is already reclaimed in ipsec4_output. */
		m0 = NULL;
		switch (error) {
		case EHOSTUNREACH:
		case ENETUNREACH:
		case EMSGSIZE:
		case ENOBUFS:
		case ENOMEM:
			break;
		default:
			printf("ip4_output (ipsec): error code %d\n", error);
			/* FALLTHRU */
		case ENOENT:
			/* don't show these error codes to the user */
			error = 0;
			break;
		}
		KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_END,
		    4, 0, 0, 0, 0);
		goto bad;
	}
	}

	/* be sure to update variables that are affected by ipsec4_output() */
	ip = mtod(m, struct ip *);

#ifdef _IP_VHL
	hlen = IP_VHL_HL(ip->ip_vhl) << 2;
#else /* !_IP_VHL */
	hlen = ip->ip_hl << 2;
#endif /* !_IP_VHL */
	/* Check that there wasn't a route change and src is still valid */
	if (ROUTE_UNUSABLE(ro)) {
		ROUTE_RELEASE(ro);
		VERIFY(src_ia == NULL);
		if (ip->ip_src.s_addr != INADDR_ANY &&
		    !(flags & (IP_ROUTETOIF | IP_FORWARDING)) &&
		    (src_ia = ifa_foraddr(ip->ip_src.s_addr)) == NULL) {
			error = EADDRNOTAVAIL;
			KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_END,
			    5, 0, 0, 0, 0);
			goto bad;
		}
		if (src_ia != NULL) {
			IFA_REMREF(&src_ia->ia_ifa);
			src_ia = NULL;
		}
	}

	if (ro->ro_rt == NULL) {
		if (!(flags & IP_ROUTETOIF)) {
			printf("%s: can't update route after "
			    "IPsec processing\n", __func__);
			error = EHOSTUNREACH;	/* XXX */
			KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_END,
			    6, 0, 0, 0, 0);
			goto bad;
		}
	} else {
		if (ia != NULL)
			IFA_REMREF(&ia->ia_ifa);
		RT_LOCK_SPIN(ro->ro_rt);
		ia = ifatoia(ro->ro_rt->rt_ifa);
		if (ia != NULL) {
			/* Become a regular mutex */
			RT_CONVERT_LOCK(ro->ro_rt);
			IFA_ADDREF(&ia->ia_ifa);
		}
		ifp = ro->ro_rt->rt_ifp;
		RT_UNLOCK(ro->ro_rt);
	}

	/* make it flipped, again. */
#if BYTE_ORDER != BIG_ENDIAN
	NTOHS(ip->ip_len);
	NTOHS(ip->ip_off);
#endif
	KERNEL_DEBUG(DBG_FNC_IPSEC4_OUTPUT | DBG_FUNC_END,
	    7, 0xff, 0xff, 0xff, 0xff);

	/* Pass to filters again */
	if (!TAILQ_EMPTY(&ipv4_filters)) {
		struct ipfilter	*filter;

		ipf_pktopts.ippo_flags &= ~IPPOF_MCAST_OPTS;

		/*
		 * Check that a TSO frame isn't passed to a filter.
		 * This could happen if a filter is inserted while
		 * TCP is sending the TSO packet.
		 */
		if (m->m_pkthdr.csum_flags & CSUM_TSO_IPV4) {
			error = EMSGSIZE;
			goto bad;
		}

		ipf_ref();

		/* 4135317 - always pass network byte order to filter */
#if BYTE_ORDER != BIG_ENDIAN
		HTONS(ip->ip_len);
		HTONS(ip->ip_off);
#endif
		TAILQ_FOREACH(filter, &ipv4_filters, ipf_link) {
			if (filter->ipf_filter.ipf_output) {
				errno_t result;
				result = filter->ipf_filter.
				    ipf_output(filter->ipf_filter.cookie,
				    (mbuf_t *)&m, ippo);
				if (result == EJUSTRETURN) {
					ipf_unref();
					goto done;
				}
				if (result != 0) {
					ipf_unref();
					goto bad;
				}
			}
		}
		/* set back to host byte order */
		ip = mtod(m, struct ip *);
#if BYTE_ORDER != BIG_ENDIAN
		NTOHS(ip->ip_len);
		NTOHS(ip->ip_off);
#endif
		ipf_unref();
	}
skip_ipsec:
#endif /* IPSEC */

#if IPFIREWALL
	/*
	 * Check with the firewall...
	 * but not if we are already being fwd'd from a firewall.
	 */
	if (fw_enable && IPFW_LOADED && !args.fwa_next_hop) {
		struct sockaddr_in *old = dst;

		args.fwa_m = m;
		args.fwa_next_hop = dst;
		args.fwa_oif = ifp;
		ipfwoff = ip_fw_chk_ptr(&args);
		m = args.fwa_m;
		dst = args.fwa_next_hop;

		/*
		 * On return we must do the following:
		 *   IP_FW_PORT_DENY_FLAG	  -> drop the pkt (XXX new)
		 *   1<=off<= 0xffff		  -> DIVERT
		 *   (off & IP_FW_PORT_DYNT_FLAG) -> send to a DUMMYNET pipe
		 *   (off & IP_FW_PORT_TEE_FLAG)  -> TEE the packet
		 *   dst != old			  -> IPFIREWALL_FORWARD
		 *   off==0, dst==old		  -> accept
		 * If some of the above modules is not compiled in, then
		 * we should't have to check the corresponding condition
		 * (because the ipfw control socket should not accept
		 * unsupported rules), but better play safe and drop
		 * packets in case of doubt.
		 */
		m0 = m;
		if ((ipfwoff & IP_FW_PORT_DENY_FLAG) || m == NULL) {
			if (m)
				m_freem(m);
			error = EACCES;
			goto done;
		}
		ip = mtod(m, struct ip *);

		if (ipfwoff == 0 && dst == old) {	/* common case */
			goto pass;
		}
#if DUMMYNET
		if (DUMMYNET_LOADED && (ipfwoff & IP_FW_PORT_DYNT_FLAG) != 0) {
			/*
			 * pass the pkt to dummynet. Need to include
			 * pipe number, m, ifp, ro, dst because these are
			 * not recomputed in the next pass.
			 * All other parameters have been already used and
			 * so they are not needed anymore.
			 * XXX note: if the ifp or ro entry are deleted
			 * while a pkt is in dummynet, we are in trouble!
			 */
			args.fwa_ro = ro;
			args.fwa_dst = dst;
			args.fwa_oflags = flags;
			if (flags & IP_OUTARGS)
				args.fwa_ipoa = ipoa;

			error = ip_dn_io_ptr(m, ipfwoff & 0xffff, DN_TO_IP_OUT,
			    &args, DN_CLIENT_IPFW);
			goto done;
		}
#endif /* DUMMYNET */
#if IPDIVERT
		if (ipfwoff != 0 && (ipfwoff & IP_FW_PORT_DYNT_FLAG) == 0) {
			struct mbuf *clone = NULL;

			/* Clone packet if we're doing a 'tee' */
			if ((ipfwoff & IP_FW_PORT_TEE_FLAG) != 0)
				clone = m_dup(m, M_DONTWAIT);
			/*
			 * XXX
			 * delayed checksums are not currently compatible
			 * with divert sockets.
			 */
			if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
				in_delayed_cksum(m);

			/* Restore packet header fields to original values */

#if BYTE_ORDER != BIG_ENDIAN
			HTONS(ip->ip_len);
			HTONS(ip->ip_off);
#endif

			/* Deliver packet to divert input routine */
			divert_packet(m, 0, ipfwoff & 0xffff,
			    args.fwa_divert_rule);

			/* If 'tee', continue with original packet */
			if (clone != NULL) {
				m0 = m = clone;
				ip = mtod(m, struct ip *);
				goto pass;
			}
			goto done;
		}
#endif /* IPDIVERT */
#if IPFIREWALL_FORWARD
		/*
		 * Here we check dst to make sure it's directly reachable on
		 * the interface we previously thought it was.
		 * If it isn't (which may be likely in some situations) we have
		 * to re-route it (ie, find a route for the next-hop and the
		 * associated interface) and set them here. This is nested
		 * forwarding which in most cases is undesirable, except where
		 * such control is nigh impossible. So we do it here.
		 * And I'm babbling.
		 */
		if (ipfwoff == 0 && old != dst) {
			struct in_ifaddr *ia_fw;
			struct route *ro_fwd = &sro_fwd;

#if IPFIREWALL_FORWARD_DEBUG
			printf("IPFIREWALL_FORWARD: New dst ip: ");
			print_ip(dst->sin_addr);
			printf("\n");
#endif /* IPFIREWALL_FORWARD_DEBUG */
			/*
			 * We need to figure out if we have been forwarded
			 * to a local socket. If so then we should somehow
			 * "loop back" to ip_input, and get directed to the
			 * PCB as if we had received this packet. This is
			 * because it may be dificult to identify the packets
			 * you want to forward until they are being output
			 * and have selected an interface. (e.g. locally
			 * initiated packets) If we used the loopback inteface,
			 * we would not be able to control what happens
			 * as the packet runs through ip_input() as
			 * it is done through a ISR.
			 */
			lck_rw_lock_shared(in_ifaddr_rwlock);
			TAILQ_FOREACH(ia_fw, &in_ifaddrhead, ia_link) {
				/*
				 * If the addr to forward to is one
				 * of ours, we pretend to
				 * be the destination for this packet.
				 */
				IFA_LOCK_SPIN(&ia_fw->ia_ifa);
				if (IA_SIN(ia_fw)->sin_addr.s_addr ==
				    dst->sin_addr.s_addr) {
					IFA_UNLOCK(&ia_fw->ia_ifa);
					break;
				}
				IFA_UNLOCK(&ia_fw->ia_ifa);
			}
			lck_rw_done(in_ifaddr_rwlock);
			if (ia_fw) {
				/* tell ip_input "dont filter" */
				struct m_tag		*fwd_tag;
				struct ip_fwd_tag	*ipfwd_tag;

				fwd_tag = m_tag_create(KERNEL_MODULE_TAG_ID,
				    KERNEL_TAG_TYPE_IPFORWARD,
				    sizeof (*ipfwd_tag), M_NOWAIT, m);
				if (fwd_tag == NULL) {
					error = ENOBUFS;
					goto bad;
				}

				ipfwd_tag = (struct ip_fwd_tag *)(fwd_tag+1);
				ipfwd_tag->next_hop = args.fwa_next_hop;

				m_tag_prepend(m, fwd_tag);

				if (m->m_pkthdr.rcvif == NULL)
					m->m_pkthdr.rcvif = lo_ifp;

#if BYTE_ORDER != BIG_ENDIAN
				HTONS(ip->ip_len);
				HTONS(ip->ip_off);
#endif
				mbuf_outbound_finalize(m, PF_INET, 0);

				/*
				 * we need to call dlil_output to run filters
				 * and resync to avoid recursion loops.
				 */
				if (lo_ifp) {
					dlil_output(lo_ifp, PF_INET, m, NULL,
					    SA(dst), 0, adv);
				} else {
					printf("%s: no loopback ifp for "
					    "forwarding!!!\n", __func__);
				}
				goto done;
			}
			/*
			 * Some of the logic for this was nicked from above.
			 *
			 * This rewrites the cached route in a local PCB.
			 * Is this what we want to do?
			 */
			ROUTE_RELEASE(ro_fwd);
			bcopy(dst, &ro_fwd->ro_dst, sizeof (*dst));

			rtalloc_ign(ro_fwd, RTF_PRCLONING, false);

			if (ro_fwd->ro_rt == NULL) {
				OSAddAtomic(1, &ipstat.ips_noroute);
				error = EHOSTUNREACH;
				goto bad;
			}

			RT_LOCK_SPIN(ro_fwd->ro_rt);
			ia_fw = ifatoia(ro_fwd->ro_rt->rt_ifa);
			if (ia_fw != NULL) {
				/* Become a regular mutex */
				RT_CONVERT_LOCK(ro_fwd->ro_rt);
				IFA_ADDREF(&ia_fw->ia_ifa);
			}
			ifp = ro_fwd->ro_rt->rt_ifp;
			ro_fwd->ro_rt->rt_use++;
			if (ro_fwd->ro_rt->rt_flags & RTF_GATEWAY)
				dst = SIN(ro_fwd->ro_rt->rt_gateway);
			if (ro_fwd->ro_rt->rt_flags & RTF_HOST) {
				/* double negation needed for bool bit field */
				ipobf.isbroadcast =
				    !!(ro_fwd->ro_rt->rt_flags & RTF_BROADCAST);
			} else {
				/* Become a regular mutex */
				RT_CONVERT_LOCK(ro_fwd->ro_rt);
				ipobf.isbroadcast =
				    in_broadcast(dst->sin_addr, ifp);
			}
			RT_UNLOCK(ro_fwd->ro_rt);
			ROUTE_RELEASE(ro);
			ro->ro_rt = ro_fwd->ro_rt;
			ro_fwd->ro_rt = NULL;
			dst = SIN(&ro_fwd->ro_dst);

			/*
			 * If we added a default src ip earlier,
			 * which would have been gotten from the-then
			 * interface, do it again, from the new one.
			 */
			if (ia_fw != NULL) {
				if (ipobf.fwd_rewrite_src) {
					IFA_LOCK_SPIN(&ia_fw->ia_ifa);
					ip->ip_src = IA_SIN(ia_fw)->sin_addr;
					IFA_UNLOCK(&ia_fw->ia_ifa);
				}
				IFA_REMREF(&ia_fw->ia_ifa);
			}
			goto pass;
		}
#endif /* IPFIREWALL_FORWARD */
		/*
		 * if we get here, none of the above matches, and
		 * we have to drop the pkt
		 */
		m_freem(m);
		error = EACCES; /* not sure this is the right error msg */
		goto done;
	}

pass:
#endif /* IPFIREWALL */

	/* 127/8 must not appear on wire - RFC1122 */
	if (!(ifp->if_flags & IFF_LOOPBACK) &&
	    ((ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
	    (ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET)) {
		OSAddAtomic(1, &ipstat.ips_badaddr);
		error = EADDRNOTAVAIL;
		goto bad;
	}

	if (ipoa != NULL) {
		u_int8_t dscp = ip->ip_tos >> IPTOS_DSCP_SHIFT;

		error = set_packet_qos(m, ifp,
		    ipoa->ipoa_flags & IPOAF_QOSMARKING_ALLOWED ? TRUE : FALSE,
		    ipoa->ipoa_sotc, ipoa->ipoa_netsvctype, &dscp);
		if (error == 0) {
			ip->ip_tos &= IPTOS_ECN_MASK;
			ip->ip_tos |= dscp << IPTOS_DSCP_SHIFT;
		} else {
			printf("%s if_dscp_for_mbuf() error %d\n", __func__, error);
			error = 0;
		}
	}

	/*
	 * Some Wi-Fi AP implementations do not correctly handle multicast IP
	 * packets with DSCP bits set -- see radr://9331522 -- so as a
	 * workaround we clear the DSCP bits and set the service class to BE
	 */
	if (IN_MULTICAST(ntohl(pkt_dst.s_addr))	&& IFNET_IS_WIFI_INFRA(ifp)) {
		ip->ip_tos &= IPTOS_ECN_MASK;
		mbuf_set_service_class(m, MBUF_SC_BE);
	}

	ip_output_checksum(ifp, m, (IP_VHL_HL(ip->ip_vhl) << 2),
	    ip->ip_len, &sw_csum);

	/*
	 * If small enough for interface, or the interface will take
	 * care of the fragmentation for us, can just send directly.
	 */
	if ((u_short)ip->ip_len <= ifp->if_mtu || TSO_IPV4_OK(ifp, m) ||
	    (!(ip->ip_off & IP_DF) && (ifp->if_hwassist & CSUM_FRAGMENT))) {
#if BYTE_ORDER != BIG_ENDIAN
		HTONS(ip->ip_len);
		HTONS(ip->ip_off);
#endif

		ip->ip_sum = 0;
		if (sw_csum & CSUM_DELAY_IP) {
			ip->ip_sum = ip_cksum_hdr_out(m, hlen);
			sw_csum &= ~CSUM_DELAY_IP;
			m->m_pkthdr.csum_flags &= ~CSUM_DELAY_IP;
		}

#if IPSEC
		/* clean ipsec history once it goes out of the node */
		if (ipsec_bypass == 0 && !(flags & IP_NOIPSEC))
			ipsec_delaux(m);
#endif /* IPSEC */
		if ((m->m_pkthdr.csum_flags & CSUM_TSO_IPV4) &&
		    (m->m_pkthdr.tso_segsz > 0))
			scnt += m->m_pkthdr.len / m->m_pkthdr.tso_segsz;
		else
			scnt++;

		if (packetchain == 0) {
			if (ro->ro_rt != NULL && nstat_collect)
				nstat_route_tx(ro->ro_rt, scnt,
				    m->m_pkthdr.len, 0);

			error = dlil_output(ifp, PF_INET, m, ro->ro_rt,
			    SA(dst), 0, adv);
			if (dlil_verbose && error) {
				printf("dlil_output error on interface %s: %d\n",
					ifp->if_xname, error);
			}
			scnt = 0;
			goto done;
		} else {
			/*
			 * packet chaining allows us to reuse the
			 * route for all packets
			 */
			bytecnt += m->m_pkthdr.len;
			mppn = &m->m_nextpkt;
			m = m->m_nextpkt;
			if (m == NULL) {
#if PF
sendchain:
#endif /* PF */
				if (pktcnt > ip_maxchainsent)
					ip_maxchainsent = pktcnt;
				if (ro->ro_rt != NULL && nstat_collect)
					nstat_route_tx(ro->ro_rt, scnt,
					    bytecnt, 0);

				error = dlil_output(ifp, PF_INET, packetlist,
				    ro->ro_rt, SA(dst), 0, adv);
				if (dlil_verbose && error) {
					printf("dlil_output error on interface %s: %d\n",
					ifp->if_xname, error);
				}
				pktcnt = 0;
				scnt = 0;
				bytecnt = 0;
				goto done;

			}
			m0 = m;
			pktcnt++;
			goto loopit;
		}
	}
	/*
	 * Too large for interface; fragment if possible.
	 * Must be able to put at least 8 bytes per fragment.
	 * Balk when DF bit is set or the interface didn't support TSO.
	 */
	if ((ip->ip_off & IP_DF) || pktcnt > 0 ||
	    (m->m_pkthdr.csum_flags & CSUM_TSO_IPV4)) {
		error = EMSGSIZE;
		/*
		 * This case can happen if the user changed the MTU
		 * of an interface after enabling IP on it.  Because
		 * most netifs don't keep track of routes pointing to
		 * them, there is no way for one to update all its
		 * routes when the MTU is changed.
		 */
		if (ro->ro_rt) {
			RT_LOCK_SPIN(ro->ro_rt);
			if ((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;
			}
			RT_UNLOCK(ro->ro_rt);
		}
		if (pktcnt > 0) {
			m0 = packetlist;
		}
		OSAddAtomic(1, &ipstat.ips_cantfrag);
		goto bad;
	}

	error = ip_fragment(m, ifp, ifp->if_mtu, sw_csum);
	if (error != 0) {
		m0 = m = NULL;
		goto bad;
	}

	KERNEL_DEBUG(DBG_LAYER_END, ip->ip_dst.s_addr,
	    ip->ip_src.s_addr, ip->ip_p, ip->ip_off, ip->ip_len);

	for (m = m0; m; m = m0) {
		m0 = m->m_nextpkt;
		m->m_nextpkt = 0;
#if IPSEC
		/* clean ipsec history once it goes out of the node */
		if (ipsec_bypass == 0 && !(flags & IP_NOIPSEC))
			ipsec_delaux(m);
#endif /* IPSEC */
		if (error == 0) {
			if ((packetchain != 0) && (pktcnt > 0)) {
				panic("%s: mix of packet in packetlist is "
				    "wrong=%p", __func__, packetlist);
				/* NOTREACHED */
			}
			if (ro->ro_rt != NULL && nstat_collect) {
				nstat_route_tx(ro->ro_rt, 1,
				    m->m_pkthdr.len, 0);
			}
			error = dlil_output(ifp, PF_INET, m, ro->ro_rt,
			    SA(dst), 0, adv);
			if (dlil_verbose && error) {
				printf("dlil_output error on interface %s: %d\n",
					   ifp->if_xname, error);
			}
		} else {
			m_freem(m);
		}
	}

	if (error == 0)
		OSAddAtomic(1, &ipstat.ips_fragmented);

done:
	if (ia != NULL) {
		IFA_REMREF(&ia->ia_ifa);
		ia = NULL;
	}
#if IPSEC
	ROUTE_RELEASE(&ipsec_state.ro);
	if (sp != NULL) {
		KEYDEBUG(KEYDEBUG_IPSEC_STAMP,
		    printf("DP ip_output call free SP:%x\n", sp));
		key_freesp(sp, KEY_SADB_UNLOCKED);
	}
#endif /* IPSEC */
#if NECP
	ROUTE_RELEASE(&necp_route);
#endif /* NECP */
#if DUMMYNET
	ROUTE_RELEASE(&saved_route);
#endif /* DUMMYNET */
#if IPFIREWALL_FORWARD
	ROUTE_RELEASE(&sro_fwd);
#endif /* IPFIREWALL_FORWARD */

	KERNEL_DEBUG(DBG_FNC_IP_OUTPUT | DBG_FUNC_END, error, 0, 0, 0, 0);
	if (ip_output_measure) {
		net_perf_measure_time(&net_perf, &start_tv, packets_processed);
		net_perf_histogram(&net_perf, packets_processed);
	}
	return (error);
bad:
	if (pktcnt > 0)
		m0 = packetlist;
	m_freem_list(m0);
	goto done;

#undef ipsec_state
#undef args
#undef sro_fwd
#undef saved_route
#undef ipf_pktopts
#undef IP_CHECK_RESTRICTIONS
}

int
ip_fragment(struct mbuf *m, struct ifnet *ifp, unsigned long mtu, int sw_csum)
{
	struct ip *ip, *mhip;
	int len, hlen, mhlen, firstlen, off, error = 0;
	struct mbuf **mnext = &m->m_nextpkt, *m0;
	int nfrags = 1;

	ip = mtod(m, struct ip *);
#ifdef _IP_VHL
	hlen = IP_VHL_HL(ip->ip_vhl) << 2;
#else /* !_IP_VHL */
	hlen = ip->ip_hl << 2;
#endif /* !_IP_VHL */

	firstlen = len = (mtu - hlen) &~ 7;
	if (len < 8) {
		m_freem(m);
		return (EMSGSIZE);
	}

	/*
	 * if the interface will not calculate checksums on
	 * fragmented packets, then do it here.
	 */
	if ((m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) &&
	    !(ifp->if_hwassist & CSUM_IP_FRAGS))
		in_delayed_cksum(m);

	/*
	 * Loop through length of segment after first fragment,
	 * make new header and copy data of each part and link onto chain.
	 */
	m0 = m;
	mhlen = sizeof (struct ip);
	for (off = hlen + len; off < (u_short)ip->ip_len; off += len) {
		MGETHDR(m, M_DONTWAIT, MT_HEADER);	/* MAC-OK */
		if (m == NULL) {
			error = ENOBUFS;
			OSAddAtomic(1, &ipstat.ips_odropped);
			goto sendorfree;
		}
		m->m_flags |= (m0->m_flags & M_MCAST) | M_FRAG;
		m->m_data += max_linkhdr;
		mhip = mtod(m, struct ip *);
		*mhip = *ip;
		if (hlen > sizeof (struct ip)) {
			mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip);
			mhip->ip_vhl = IP_MAKE_VHL(IPVERSION, mhlen >> 2);
		}
		m->m_len = mhlen;
		mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF);
		if (ip->ip_off & IP_MF)
			mhip->ip_off |= IP_MF;
		if (off + len >= (u_short)ip->ip_len)
			len = (u_short)ip->ip_len - off;
		else
			mhip->ip_off |= IP_MF;
		mhip->ip_len = htons((u_short)(len + mhlen));
		m->m_next = m_copy(m0, off, len);
		if (m->m_next == NULL) {
			(void) m_free(m);
			error = ENOBUFS;	/* ??? */
			OSAddAtomic(1, &ipstat.ips_odropped);
			goto sendorfree;
		}
		m->m_pkthdr.len = mhlen + len;
		m->m_pkthdr.rcvif = NULL;
		m->m_pkthdr.csum_flags = m0->m_pkthdr.csum_flags;

		M_COPY_CLASSIFIER(m, m0);
		M_COPY_PFTAG(m, m0);

#if CONFIG_MACF_NET
		mac_netinet_fragment(m0, m);
#endif /* CONFIG_MACF_NET */

#if BYTE_ORDER != BIG_ENDIAN
		HTONS(mhip->ip_off);
#endif

		mhip->ip_sum = 0;
		if (sw_csum & CSUM_DELAY_IP) {
			mhip->ip_sum = ip_cksum_hdr_out(m, mhlen);
			m->m_pkthdr.csum_flags &= ~CSUM_DELAY_IP;
		}
		*mnext = m;
		mnext = &m->m_nextpkt;
		nfrags++;
	}
	OSAddAtomic(nfrags, &ipstat.ips_ofragments);

	/* set first/last markers for fragment chain */
	m->m_flags |= M_LASTFRAG;
	m0->m_flags |= M_FIRSTFRAG | M_FRAG;
	m0->m_pkthdr.csum_data = nfrags;

	/*
	 * Update first fragment by trimming what's been copied out
	 * and updating header, then send each fragment (in order).
	 */
	m = m0;
	m_adj(m, hlen + firstlen - (u_short)ip->ip_len);
	m->m_pkthdr.len = hlen + firstlen;
	ip->ip_len = htons((u_short)m->m_pkthdr.len);
	ip->ip_off |= IP_MF;

#if BYTE_ORDER != BIG_ENDIAN
	HTONS(ip->ip_off);
#endif

	ip->ip_sum = 0;
	if (sw_csum & CSUM_DELAY_IP) {
		ip->ip_sum = ip_cksum_hdr_out(m, hlen);
		m->m_pkthdr.csum_flags &= ~CSUM_DELAY_IP;
	}
sendorfree:
	if (error)
		m_freem_list(m0);

	return (error);
}

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;
	}
}

/*
 * Process a delayed payload checksum calculation (outbound path.)
 *
 * hoff is the number of bytes beyond the mbuf data pointer which
 * points to the IP header.
 *
 * Returns a bitmask representing all the work done in software.
 */
uint32_t
in_finalize_cksum(struct mbuf *m, uint32_t hoff, uint32_t csum_flags)
{
	unsigned char buf[15 << 2] __attribute__((aligned(8)));
	struct ip *ip;
	uint32_t offset, _hlen, mlen, hlen, len, sw_csum;
	uint16_t csum, ip_len;

	_CASSERT(sizeof (csum) == sizeof (uint16_t));
	VERIFY(m->m_flags & M_PKTHDR);

	sw_csum = (csum_flags & m->m_pkthdr.csum_flags);

	if ((sw_csum &= (CSUM_DELAY_IP | CSUM_DELAY_DATA)) == 0)
		goto done;

	mlen = m->m_pkthdr.len;				/* total mbuf len */

	/* sanity check (need at least simple IP header) */
	if (mlen < (hoff + sizeof (*ip))) {
		panic("%s: mbuf %p pkt len (%u) < hoff+ip_hdr "
		    "(%u+%u)\n", __func__, m, mlen, hoff,
		    (uint32_t)sizeof (*ip));
		/* NOTREACHED */
	}

	/*
	 * In case the IP header is not contiguous, or not 32-bit aligned,
	 * or if we're computing the IP header checksum, copy it to a local
	 * buffer.  Copy only the simple IP header here (IP options case
	 * is handled below.)
	 */
	if ((sw_csum & CSUM_DELAY_IP) || (hoff + sizeof (*ip)) > m->m_len ||
	    !IP_HDR_ALIGNED_P(mtod(m, caddr_t) + hoff)) {
		m_copydata(m, hoff, sizeof (*ip), (caddr_t)buf);
		ip = (struct ip *)(void *)buf;
		_hlen = sizeof (*ip);
	} else {
		ip = (struct ip *)(void *)(m->m_data + hoff);
		_hlen = 0;
	}

	hlen = IP_VHL_HL(ip->ip_vhl) << 2;		/* IP header len */

	/* sanity check */
	if (mlen < (hoff + hlen)) {
		panic("%s: mbuf %p pkt too short (%d) for IP header (%u), "
		    "hoff %u", __func__, m, mlen, hlen, hoff);
		/* NOTREACHED */
	}

	/*
	 * 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, use the actual
	 * packet length.  This also covers the trailing bytes case.
	 */
	ip_len = ip->ip_len;
	if (ip_len != (mlen - hoff)) {
		ip_len = OSSwapInt16(ip_len);
		if (ip_len != (mlen - hoff)) {
			printf("%s: mbuf 0x%llx proto %d IP len %d (%x) "
			    "[swapped %d (%x)] doesn't match actual packet "
			    "length; %d is used instead\n", __func__,
			    (uint64_t)VM_KERNEL_ADDRPERM(m), ip->ip_p,
			    ip->ip_len, ip->ip_len, ip_len, ip_len,
			    (mlen - hoff));
			ip_len = mlen - hoff;
		}
	}

	len = ip_len - hlen;				/* csum span */

	if (sw_csum & CSUM_DELAY_DATA) {
		uint16_t ulpoff;

		/*
		 * offset is added to the lower 16-bit value of csum_data,
		 * which is expected to contain the ULP offset; therefore
		 * CSUM_PARTIAL offset adjustment must be undone.
		 */
		if ((m->m_pkthdr.csum_flags & (CSUM_PARTIAL|CSUM_DATA_VALID)) ==
		    (CSUM_PARTIAL|CSUM_DATA_VALID)) {
			/*
			 * Get back the original ULP offset (this will
			 * undo the CSUM_PARTIAL logic in ip_output.)
			 */
			m->m_pkthdr.csum_data = (m->m_pkthdr.csum_tx_stuff -
			    m->m_pkthdr.csum_tx_start);
		}

		ulpoff = (m->m_pkthdr.csum_data & 0xffff); /* ULP csum offset */
		offset = hoff + hlen;			/* ULP header */

		if (mlen < (ulpoff + sizeof (csum))) {
			panic("%s: mbuf %p pkt len (%u) proto %d invalid ULP "
			    "cksum offset (%u) cksum flags 0x%x\n", __func__,
			    m, mlen, ip->ip_p, ulpoff, m->m_pkthdr.csum_flags);
			/* NOTREACHED */
		}

		csum = inet_cksum(m, 0, offset, len);

		/* Update stats */
		ip_out_cksum_stats(ip->ip_p, len);

		/* RFC1122 4.1.3.4 */
		if (csum == 0 &&
		    (m->m_pkthdr.csum_flags & (CSUM_UDP|CSUM_ZERO_INVERT)))
			csum = 0xffff;

		/* Insert the checksum in the ULP csum field */
		offset += ulpoff;
		if (offset + sizeof (csum) > m->m_len) {
			m_copyback(m, offset, sizeof (csum), &csum);
		} else if (IP_HDR_ALIGNED_P(mtod(m, char *) + hoff)) {
			*(uint16_t *)(void *)(mtod(m, char *) + offset) = csum;
		} else {
			bcopy(&csum, (mtod(m, char *) + offset), sizeof (csum));
		}
		m->m_pkthdr.csum_flags &= ~(CSUM_DELAY_DATA | CSUM_DATA_VALID |
		    CSUM_PARTIAL | CSUM_ZERO_INVERT);
	}

	if (sw_csum & CSUM_DELAY_IP) {
		/* IP header must be in the local buffer */
		VERIFY(_hlen == sizeof (*ip));
		if (_hlen != hlen) {
			VERIFY(hlen <= sizeof (buf));
			m_copydata(m, hoff, hlen, (caddr_t)buf);
			ip = (struct ip *)(void *)buf;
			_hlen = hlen;
		}

		/*
		 * Compute the IP header checksum as if the IP length
		 * is the length which we believe is "correct"; see
		 * how ip_len gets calculated above.  Note that this
		 * is done on the local copy and not on the real one.
		 */
		ip->ip_len = htons(ip_len);
		ip->ip_sum = 0;
		csum = in_cksum_hdr_opt(ip);

		/* Update stats */
		ipstat.ips_snd_swcsum++;
		ipstat.ips_snd_swcsum_bytes += hlen;

		/*
		 * Insert only the checksum in the existing IP header
		 * csum field; all other fields are left unchanged.
		 */
		offset = hoff + offsetof(struct ip, ip_sum);
		if (offset + sizeof (csum) > m->m_len) {
			m_copyback(m, offset, sizeof (csum), &csum);
		} else if (IP_HDR_ALIGNED_P(mtod(m, char *) + hoff)) {
			*(uint16_t *)(void *)(mtod(m, char *) + offset) = csum;
		} else {
			bcopy(&csum, (mtod(m, char *) + offset), sizeof (csum));
		}
		m->m_pkthdr.csum_flags &= ~CSUM_DELAY_IP;
	}

done:
	return (sw_csum);
}

/*
 * Insert IP options into preformed packet.
 * Adjust IP destination as required for IP source routing,
 * as indicated by a non-zero in_addr at the start of the options.
 *
 * XXX This routine assumes that the packet has no options in place.
 */
static struct mbuf *
ip_insertoptions(struct mbuf *m, struct mbuf *opt, int *phlen)
{
	struct ipoption *p = mtod(opt, struct ipoption *);
	struct mbuf *n;
	struct ip *ip = mtod(m, struct ip *);
	unsigned optlen;

	optlen = opt->m_len - sizeof (p->ipopt_dst);
	if (optlen + (u_short)ip->ip_len > IP_MAXPACKET)
		return (m);		/* XXX should fail */
	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);	/* MAC-OK */
		if (n == NULL)
			return (m);
		n->m_pkthdr.rcvif = 0;
#if CONFIG_MACF_NET
		mac_mbuf_label_copy(m, n);
#endif /* CONFIG_MACF_NET */
		n->m_pkthdr.len = m->m_pkthdr.len + optlen;
		m->m_len -= sizeof (struct ip);
		m->m_data += sizeof (struct ip);
		n->m_next = m;
		m = n;
		m->m_len = optlen + sizeof (struct ip);
		m->m_data += max_linkhdr;
		(void) memcpy(mtod(m, void *), ip, sizeof (struct ip));
	} else {
		m->m_data -= optlen;
		m->m_len += optlen;
		m->m_pkthdr.len += optlen;
		ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof (struct ip));
	}
	ip = mtod(m, struct ip *);
	bcopy(p->ipopt_list, ip + 1, optlen);
	*phlen = sizeof (struct ip) + optlen;
	ip->ip_vhl = IP_MAKE_VHL(IPVERSION, *phlen >> 2);
	ip->ip_len += optlen;
	return (m);
}

/*
 * Copy options from ip to jp,
 * omitting those not copied during fragmentation.
 */
static int
ip_optcopy(struct ip *ip, struct ip *jp)
{
	u_char *cp, *dp;
	int opt, optlen, cnt;

	cp = (u_char *)(ip + 1);
	dp = (u_char *)(jp + 1);
	cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip);
	for (; cnt > 0; cnt -= optlen, cp += optlen) {
		opt = cp[0];
		if (opt == IPOPT_EOL)
			break;
		if (opt == IPOPT_NOP) {
			/* Preserve for IP mcast tunnel's LSRR alignment. */
			*dp++ = IPOPT_NOP;
			optlen = 1;
			continue;
		}
#if DIAGNOSTIC
		if (cnt < IPOPT_OLEN + sizeof (*cp)) {
			panic("malformed IPv4 option passed to ip_optcopy");
			/* NOTREACHED */
		}
#endif
		optlen = cp[IPOPT_OLEN];
#if DIAGNOSTIC
		if (optlen < IPOPT_OLEN + sizeof (*cp) || optlen > cnt) {
			panic("malformed IPv4 option passed to ip_optcopy");
			/* NOTREACHED */
		}
#endif
		/* bogus lengths should have been caught by ip_dooptions */
		if (optlen > cnt)
			optlen = cnt;
		if (IPOPT_COPIED(opt)) {
			bcopy(cp, dp, optlen);
			dp += optlen;
		}
	}
	for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++)
		*dp++ = IPOPT_EOL;
	return (optlen);
}

/*
 * IP socket option processing.
 */
int
ip_ctloutput(struct socket *so, struct sockopt *sopt)
{
	struct	inpcb *inp = sotoinpcb(so);
	int	error, optval;

	error = optval = 0;
	if (sopt->sopt_level != IPPROTO_IP)
		return (EINVAL);

	switch (sopt->sopt_dir) {
	case SOPT_SET:
		switch (sopt->sopt_name) {
#ifdef notyet
		case IP_RETOPTS:
#endif
		case IP_OPTIONS: {
			struct mbuf *m;

			if (sopt->sopt_valsize > MLEN) {
				error = EMSGSIZE;
				break;
			}
			MGET(m, sopt->sopt_p != kernproc ? M_WAIT : M_DONTWAIT,
			    MT_HEADER);
			if (m == NULL) {
				error = ENOBUFS;
				break;
			}
			m->m_len = sopt->sopt_valsize;
			error = sooptcopyin(sopt, mtod(m, char *),
			    m->m_len, m->m_len);
			if (error) {
				m_freem(m);
				break;
			}

			return (ip_pcbopts(sopt->sopt_name,
			    &inp->inp_options, m));
		}

		case IP_TOS:
		case IP_TTL:
		case IP_RECVOPTS:
		case IP_RECVRETOPTS:
		case IP_RECVDSTADDR:
		case IP_RECVIF:
		case IP_RECVTTL:
		case IP_RECVPKTINFO:
		case IP_RECVTOS:
			error = sooptcopyin(sopt, &optval, sizeof (optval),
			    sizeof (optval));
			if (error)
				break;

			switch (sopt->sopt_name) {
			case IP_TOS:
				inp->inp_ip_tos = optval;
				break;

			case IP_TTL:
				inp->inp_ip_ttl = optval;
				break;
#define	OPTSET(bit) \
	if (optval) \
		inp->inp_flags |= bit; \
	else \
		inp->inp_flags &= ~bit;

			case IP_RECVOPTS:
				OPTSET(INP_RECVOPTS);
				break;

			case IP_RECVRETOPTS:
				OPTSET(INP_RECVRETOPTS);
				break;

			case IP_RECVDSTADDR:
				OPTSET(INP_RECVDSTADDR);
				break;

			case IP_RECVIF:
				OPTSET(INP_RECVIF);
				break;

			case IP_RECVTTL:
				OPTSET(INP_RECVTTL);
				break;

			case IP_RECVPKTINFO:
				OPTSET(INP_PKTINFO);
				break;

			case IP_RECVTOS:
				OPTSET(INP_RECVTOS);
				break;
 #undef OPTSET
			}
			break;
		/*
		 * Multicast socket options are processed by the in_mcast
		 * module.
		 */
		case IP_MULTICAST_IF:
		case IP_MULTICAST_IFINDEX:
		case IP_MULTICAST_VIF:
		case IP_MULTICAST_TTL:
		case IP_MULTICAST_LOOP:
		case IP_ADD_MEMBERSHIP:
		case IP_DROP_MEMBERSHIP:
		case IP_ADD_SOURCE_MEMBERSHIP:
		case IP_DROP_SOURCE_MEMBERSHIP:
		case IP_BLOCK_SOURCE:
		case IP_UNBLOCK_SOURCE:
		case IP_MSFILTER:
		case MCAST_JOIN_GROUP:
		case MCAST_LEAVE_GROUP:
		case MCAST_JOIN_SOURCE_GROUP:
		case MCAST_LEAVE_SOURCE_GROUP:
		case MCAST_BLOCK_SOURCE:
		case MCAST_UNBLOCK_SOURCE:
			error = inp_setmoptions(inp, sopt);
			break;

		case IP_PORTRANGE:
			error = sooptcopyin(sopt, &optval, sizeof (optval),
			    sizeof (optval));
			if (error)
				break;

			switch (optval) {
			case IP_PORTRANGE_DEFAULT:
				inp->inp_flags &= ~(INP_LOWPORT);
				inp->inp_flags &= ~(INP_HIGHPORT);
				break;

			case IP_PORTRANGE_HIGH:
				inp->inp_flags &= ~(INP_LOWPORT);
				inp->inp_flags |= INP_HIGHPORT;
				break;

			case IP_PORTRANGE_LOW:
				inp->inp_flags &= ~(INP_HIGHPORT);
				inp->inp_flags |= INP_LOWPORT;
				break;

			default:
				error = EINVAL;
				break;
			}
			break;

#if IPSEC
		case IP_IPSEC_POLICY: {
			caddr_t req = NULL;
			size_t len = 0;
			int priv;
			struct mbuf *m;
			int optname;

			if ((error = soopt_getm(sopt, &m)) != 0) /* XXX */
				break;
			if ((error = soopt_mcopyin(sopt, m)) != 0) /* XXX */
				break;
			priv = (proc_suser(sopt->sopt_p) == 0);
			if (m) {
				req = mtod(m, caddr_t);
				len = m->m_len;
			}
			optname = sopt->sopt_name;
			error = ipsec4_set_policy(inp, optname, req, len, priv);
			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) {
				socket_set_traffic_mgt_flags_locked(so,
				    TRAFFIC_MGT_SO_BACKGROUND);
			} else {
				socket_clear_traffic_mgt_flags_locked(so,
				    TRAFFIC_MGT_SO_BACKGROUND);
			}

			break;
		}
#endif /* TRAFFIC_MGT */

		/*
		 * On a multihomed system, scoped routing can be used to
		 * restrict the source interface used for sending packets.
		 * The socket option IP_BOUND_IF binds a particular AF_INET
		 * socket to an interface such that data sent on the socket
		 * is restricted to that interface.  This is unlike the
		 * SO_DONTROUTE option where the routing table is bypassed;
		 * therefore it allows for a greater flexibility and control
		 * over the system behavior, and does not place any restriction
		 * on the destination address type (e.g.  unicast, multicast,
		 * or broadcast if applicable) or whether or not the host is
		 * directly reachable.  Note that in the multicast transmit
		 * case, IP_MULTICAST_{IF,IFINDEX} takes precedence over
		 * IP_BOUND_IF, since the former practically bypasses the
		 * routing table; in this case, IP_BOUND_IF sets the default
		 * interface used for sending multicast packets in the absence
		 * of an explicit multicast transmit interface.
		 */
		case IP_BOUND_IF:
			/* This option is settable only for IPv4 */
			if (!(inp->inp_vflag & INP_IPV4)) {
				error = EINVAL;
				break;
			}

			error = sooptcopyin(sopt, &optval, sizeof (optval),
			    sizeof (optval));

			if (error)
				break;

			error = inp_bindif(inp, optval, NULL);
			break;

		case IP_NO_IFT_CELLULAR:
			/* This option is settable only for IPv4 */
			if (!(inp->inp_vflag & INP_IPV4)) {
				error = EINVAL;
				break;
			}

			error = sooptcopyin(sopt, &optval, sizeof (optval),
			    sizeof (optval));

			if (error)
				break;

			/* once set, it cannot be unset */
			if (!optval && INP_NO_CELLULAR(inp)) {
				error = EINVAL;
				break;
			}

			error = so_set_restrictions(so,
			    SO_RESTRICT_DENY_CELLULAR);
			break;

		case IP_OUT_IF:
			/* This option is not settable */
			error = EINVAL;
			break;

		default:
			error = ENOPROTOOPT;
			break;
		}
		break;

	case SOPT_GET:
		switch (sopt->sopt_name) {
		case IP_OPTIONS:
		case IP_RETOPTS:
			if (inp->inp_options) {
				error = sooptcopyout(sopt,
				    mtod(inp->inp_options, char *),
				    inp->inp_options->m_len);
			} else {
				sopt->sopt_valsize = 0;
			}
			break;

		case IP_TOS:
		case IP_TTL:
		case IP_RECVOPTS:
		case IP_RECVRETOPTS:
		case IP_RECVDSTADDR:
		case IP_RECVIF:
		case IP_RECVTTL:
		case IP_PORTRANGE:
		case IP_RECVPKTINFO:
		case IP_RECVTOS:
			switch (sopt->sopt_name) {
			case IP_TOS:
				optval = inp->inp_ip_tos;
				break;

			case IP_TTL:
				optval = inp->inp_ip_ttl;
				break;

#define	OPTBIT(bit)	(inp->inp_flags & bit ? 1 : 0)

			case IP_RECVOPTS:
				optval = OPTBIT(INP_RECVOPTS);
				break;

			case IP_RECVRETOPTS:
				optval = OPTBIT(INP_RECVRETOPTS);
				break;

			case IP_RECVDSTADDR:
				optval = OPTBIT(INP_RECVDSTADDR);
				break;

			case IP_RECVIF:
				optval = OPTBIT(INP_RECVIF);
				break;

			case IP_RECVTTL:
				optval = OPTBIT(INP_RECVTTL);
				break;

			case IP_PORTRANGE:
				if (inp->inp_flags & INP_HIGHPORT)
					optval = IP_PORTRANGE_HIGH;
				else if (inp->inp_flags & INP_LOWPORT)
					optval = IP_PORTRANGE_LOW;
				else
					optval = 0;
				break;

			case IP_RECVPKTINFO:
				optval = OPTBIT(INP_PKTINFO);
				break;

			case IP_RECVTOS:
				optval = OPTBIT(INP_RECVTOS);
				break;
			}
			error = sooptcopyout(sopt, &optval, sizeof (optval));
			break;

		case IP_MULTICAST_IF:
		case IP_MULTICAST_IFINDEX:
		case IP_MULTICAST_VIF:
		case IP_MULTICAST_TTL:
		case IP_MULTICAST_LOOP:
		case IP_MSFILTER:
			error = inp_getmoptions(inp, sopt);
			break;

#if IPSEC
		case IP_IPSEC_POLICY: {
			error = 0; /* This option is no longer supported */
			break;
		}
#endif /* IPSEC */

#if TRAFFIC_MGT
		case IP_TRAFFIC_MGT_BACKGROUND: {
			unsigned background = (so->so_flags1 &
			    SOF1_TRAFFIC_MGT_SO_BACKGROUND) ? 1 : 0;
			return (sooptcopyout(sopt, &background,
			    sizeof (background)));
		}
#endif /* TRAFFIC_MGT */

		case IP_BOUND_IF:
			if (inp->inp_flags & INP_BOUND_IF)
				optval = inp->inp_boundifp->if_index;
			error = sooptcopyout(sopt, &optval, sizeof (optval));
			break;

		case IP_NO_IFT_CELLULAR:
			optval = INP_NO_CELLULAR(inp) ? 1 : 0;
			error = sooptcopyout(sopt, &optval, sizeof (optval));
			break;

		case IP_OUT_IF:
			optval = (inp->inp_last_outifp != NULL) ?
			    inp->inp_last_outifp->if_index : 0;
			error = sooptcopyout(sopt, &optval, sizeof (optval));
			break;

		default:
			error = ENOPROTOOPT;
			break;
		}
		break;
	}
	return (error);
}

/*
 * Set up IP options in pcb for insertion in output packets.
 * Store in mbuf with pointer in pcbopt, adding pseudo-option
 * with destination address if source routed.
 */
static int
ip_pcbopts(int optname, struct mbuf **pcbopt, struct mbuf *m)
{
#pragma unused(optname)
	int cnt, optlen;
	u_char *cp;
	u_char opt;

	/* turn off any old options */
	if (*pcbopt)
		(void) m_free(*pcbopt);
	*pcbopt = 0;
	if (m == (struct mbuf *)0 || m->m_len == 0) {
		/*
		 * Only turning off any previous options.
		 */
		if (m)
			(void) m_free(m);
		return (0);
	}

	if (m->m_len % sizeof (int32_t))
		goto bad;

	/*
	 * IP first-hop destination address will be stored before
	 * actual options; move other options back
	 * and clear it when none present.
	 */
	if (m->m_data + m->m_len + sizeof (struct in_addr) >= &m->m_dat[MLEN])
		goto bad;
	cnt = m->m_len;
	m->m_len += sizeof (struct in_addr);
	cp = mtod(m, u_char *) + sizeof (struct in_addr);
	ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt);
	bzero(mtod(m, caddr_t), sizeof (struct in_addr));

	for (; cnt > 0; cnt -= optlen, cp += optlen) {
		opt = cp[IPOPT_OPTVAL];
		if (opt == IPOPT_EOL)
			break;
		if (opt == IPOPT_NOP)
			optlen = 1;
		else {
			if (cnt < IPOPT_OLEN + sizeof (*cp))
				goto bad;
			optlen = cp[IPOPT_OLEN];
			if (optlen < IPOPT_OLEN + sizeof (*cp) || optlen > cnt)
				goto bad;
		}
		switch (opt) {

		default:
			break;

		case IPOPT_LSRR:
		case IPOPT_SSRR:
			/*
			 * user process specifies route as:
			 *	->A->B->C->D
			 * D must be our final destination (but we can't
			 * check that since we may not have connected yet).
			 * A is first hop destination, which doesn't appear in
			 * actual IP option, but is stored before the options.
			 */
			if (optlen < IPOPT_MINOFF - 1 + sizeof (struct in_addr))
				goto bad;
			m->m_len -= sizeof (struct in_addr);
			cnt -= sizeof (struct in_addr);
			optlen -= sizeof (struct in_addr);
			cp[IPOPT_OLEN] = optlen;
			/*
			 * Move first hop before start of options.
			 */
			bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t),
			    sizeof (struct in_addr));
			/*
			 * Then copy rest of options back
			 * to close up the deleted entry.
			 */
			ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] +
			    sizeof (struct in_addr)),
			    (caddr_t)&cp[IPOPT_OFFSET+1],
			    (unsigned)cnt + sizeof (struct in_addr));
			break;
		}
	}
	if (m->m_len > MAX_IPOPTLEN + sizeof (struct in_addr))
		goto bad;
	*pcbopt = m;
	return (0);

bad:
	(void) m_free(m);
	return (EINVAL);
}

void
ip_moptions_init(void)
{
	PE_parse_boot_argn("ifa_debug", &imo_debug, sizeof (imo_debug));

	imo_size = (imo_debug == 0) ? sizeof (struct ip_moptions) :
	    sizeof (struct ip_moptions_dbg);

	imo_zone = zinit(imo_size, IMO_ZONE_MAX * imo_size, 0,
	    IMO_ZONE_NAME);
	if (imo_zone == NULL) {
		panic("%s: failed allocating %s", __func__, IMO_ZONE_NAME);
		/* NOTREACHED */
	}
	zone_change(imo_zone, Z_EXPAND, TRUE);
}

void
imo_addref(struct ip_moptions *imo, int locked)
{
	if (!locked)
		IMO_LOCK(imo);
	else
		IMO_LOCK_ASSERT_HELD(imo);

	if (++imo->imo_refcnt == 0) {
		panic("%s: imo %p wraparound refcnt\n", __func__, imo);
		/* NOTREACHED */
	} else if (imo->imo_trace != NULL) {
		(*imo->imo_trace)(imo, TRUE);
	}

	if (!locked)
		IMO_UNLOCK(imo);
}

void
imo_remref(struct ip_moptions *imo)
{
	int i;

	IMO_LOCK(imo);
	if (imo->imo_refcnt == 0) {
		panic("%s: imo %p negative refcnt", __func__, imo);
		/* NOTREACHED */
	} else if (imo->imo_trace != NULL) {
		(*imo->imo_trace)(imo, FALSE);
	}

	--imo->imo_refcnt;
	if (imo->imo_refcnt > 0) {
		IMO_UNLOCK(imo);
		return;
	}

	for (i = 0; i < imo->imo_num_memberships; ++i) {
		struct in_mfilter *imf;

		imf = imo->imo_mfilters ? &imo->imo_mfilters[i] : NULL;
		if (imf != NULL)
			imf_leave(imf);

		(void) in_leavegroup(imo->imo_membership[i], imf);

		if (imf != NULL)
			imf_purge(imf);

		INM_REMREF(imo->imo_membership[i]);
		imo->imo_membership[i] = NULL;
	}
	imo->imo_num_memberships = 0;
	if (imo->imo_mfilters != NULL) {
		FREE(imo->imo_mfilters, M_INMFILTER);
		imo->imo_mfilters = NULL;
	}
	if (imo->imo_membership != NULL) {
		FREE(imo->imo_membership, M_IPMOPTS);
		imo->imo_membership = NULL;
	}
	IMO_UNLOCK(imo);

	lck_mtx_destroy(&imo->imo_lock, ifa_mtx_grp);

	if (!(imo->imo_debug & IFD_ALLOC)) {
		panic("%s: imo %p cannot be freed", __func__, imo);
		/* NOTREACHED */
	}
	zfree(imo_zone, imo);
}

static void
imo_trace(struct ip_moptions *imo, int refhold)
{
	struct ip_moptions_dbg *imo_dbg = (struct ip_moptions_dbg *)imo;
	ctrace_t *tr;
	u_int32_t idx;
	u_int16_t *cnt;

	if (!(imo->imo_debug & IFD_DEBUG)) {
		panic("%s: imo %p has no debug structure", __func__, imo);
		/* NOTREACHED */
	}
	if (refhold) {
		cnt = &imo_dbg->imo_refhold_cnt;
		tr = imo_dbg->imo_refhold;
	} else {
		cnt = &imo_dbg->imo_refrele_cnt;
		tr = imo_dbg->imo_refrele;
	}

	idx = atomic_add_16_ov(cnt, 1) % IMO_TRACE_HIST_SIZE;
	ctrace_record(&tr[idx]);
}

struct ip_moptions *
ip_allocmoptions(int how)
{
	struct ip_moptions *imo;

	imo = (how == M_WAITOK) ? zalloc(imo_zone) : zalloc_noblock(imo_zone);
	if (imo != NULL) {
		bzero(imo, imo_size);
		lck_mtx_init(&imo->imo_lock, ifa_mtx_grp, ifa_mtx_attr);
		imo->imo_debug |= IFD_ALLOC;
		if (imo_debug != 0) {
			imo->imo_debug |= IFD_DEBUG;
			imo->imo_trace = imo_trace;
		}
		IMO_ADDREF(imo);
	}

	return (imo);
}

/*
 * Routine called from ip_output() to loop back a copy of an IP multicast
 * packet to the input queue of a specified interface.  Note that this
 * calls the output routine of the loopback "driver", but with an interface
 * pointer that might NOT be a loopback interface -- evil, but easier than
 * replicating that code here.
 */
static void
ip_mloopback(struct ifnet *srcifp, struct ifnet *origifp, struct mbuf *m,
    struct sockaddr_in *dst, int hlen)
{
	struct mbuf *copym;
	struct ip *ip;

	if (lo_ifp == NULL)
		return;

	/*
	 * Copy the packet header as it's needed for the checksum
	 * Make sure to deep-copy IP header portion in case the data
	 * is in an mbuf cluster, so that we can safely override the IP
	 * header portion later.
	 */
	copym = m_copym_mode(m, 0, M_COPYALL, M_DONTWAIT, M_COPYM_COPY_HDR);
	if (copym != NULL && ((copym->m_flags & M_EXT) || copym->m_len < hlen))
		copym = m_pullup(copym, hlen);

	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 *);
#if BYTE_ORDER != BIG_ENDIAN
	HTONS(ip->ip_len);
	HTONS(ip->ip_off);
#endif
	ip->ip_sum = 0;
	ip->ip_sum = ip_cksum_hdr_out(copym, hlen);

	/*
	 * Mark checksum as valid unless receive checksum offload is
	 * disabled; if so, compute checksum in software.  If the
	 * interface itself is lo0, this will be overridden by if_loop.
	 */
	if (hwcksum_rx) {
		copym->m_pkthdr.csum_flags &= ~(CSUM_PARTIAL|CSUM_ZERO_INVERT);
		copym->m_pkthdr.csum_flags |=
		    CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
		copym->m_pkthdr.csum_data = 0xffff;
	} else if (copym->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
#if BYTE_ORDER != BIG_ENDIAN
		NTOHS(ip->ip_len);
#endif
		in_delayed_cksum(copym);
#if BYTE_ORDER != BIG_ENDIAN
		HTONS(ip->ip_len);
#endif
	}

	/*
	 * 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.
	 */
	copym->m_pkthdr.rcvif = origifp;

	/*
	 * Also record the source interface (which owns the source address).
	 * This is basically a stripped down version of ifa_foraddr().
	 */
	if (srcifp == NULL) {
		struct in_ifaddr *ia;

		lck_rw_lock_shared(in_ifaddr_rwlock);
		TAILQ_FOREACH(ia, INADDR_HASH(ip->ip_src.s_addr), ia_hash) {
			IFA_LOCK_SPIN(&ia->ia_ifa);
			if (IA_SIN(ia)->sin_addr.s_addr == ip->ip_src.s_addr) {
				srcifp = ia->ia_ifp;
				IFA_UNLOCK(&ia->ia_ifa);
				break;
			}
			IFA_UNLOCK(&ia->ia_ifa);
		}
		lck_rw_done(in_ifaddr_rwlock);
	}
	if (srcifp != NULL)
		ip_setsrcifaddr_info(copym, srcifp->if_index, NULL);
	ip_setdstifaddr_info(copym, origifp->if_index, NULL);

	dlil_output(lo_ifp, PF_INET, copym, NULL, SA(dst), 0, NULL);
}

/*
 * Given a source IP address (and route, if available), determine the best
 * interface to send the packet from.  Checking for (and updating) the
 * ROF_SRCIF_SELECTED flag in the pcb-supplied route placeholder is done
 * without any locks based on the assumption that ip_output() is single-
 * threaded per-pcb, i.e. for any given pcb there can only be one thread
 * performing output at the IP layer.
 *
 * This routine is analogous to in6_selectroute() for IPv6.
 */
static struct ifaddr *
in_selectsrcif(struct ip *ip, struct route *ro, unsigned int ifscope)
{
	struct ifaddr *ifa = NULL;
	struct in_addr src = ip->ip_src;
	struct in_addr dst = ip->ip_dst;
	struct ifnet *rt_ifp;
	char s_src[MAX_IPv4_STR_LEN], s_dst[MAX_IPv4_STR_LEN];

	VERIFY(src.s_addr != INADDR_ANY);

	if (ip_select_srcif_debug) {
		(void) inet_ntop(AF_INET, &src.s_addr, s_src, sizeof (s_src));
		(void) inet_ntop(AF_INET, &dst.s_addr, s_dst, sizeof (s_dst));
	}

	if (ro->ro_rt != NULL)
		RT_LOCK(ro->ro_rt);

	rt_ifp = (ro->ro_rt != NULL) ? ro->ro_rt->rt_ifp : NULL;

	/*
	 * Given the source IP address, find a suitable source interface
	 * to use for transmission; if the caller has specified a scope,
	 * optimize the search by looking at the addresses only for that
	 * interface.  This is still suboptimal, however, as we need to
	 * traverse the per-interface list.
	 */
	if (ifscope != IFSCOPE_NONE || ro->ro_rt != NULL) {
		unsigned int scope = ifscope;

		/*
		 * If no scope is specified and the route is stale (pointing
		 * to a defunct interface) use the current primary interface;
		 * this happens when switching between interfaces configured
		 * with the same IP address.  Otherwise pick up the scope
		 * information from the route; the ULP may have looked up a
		 * correct route and we just need to verify it here and mark
		 * it with the ROF_SRCIF_SELECTED flag below.
		 */
		if (scope == IFSCOPE_NONE) {
			scope = rt_ifp->if_index;
			if (scope != get_primary_ifscope(AF_INET) &&
			    ROUTE_UNUSABLE(ro))
				scope = get_primary_ifscope(AF_INET);
		}

		ifa = (struct ifaddr *)ifa_foraddr_scoped(src.s_addr, scope);

		if (ifa == NULL && ip->ip_p != IPPROTO_UDP &&
		    ip->ip_p != IPPROTO_TCP && ipforwarding) {
			/*
			 * If forwarding is enabled, and if the packet isn't
			 * TCP or UDP, check if the source address belongs
			 * to one of our own interfaces; if so, demote the
			 * interface scope and do a route lookup right below.
			 */
			ifa = (struct ifaddr *)ifa_foraddr(src.s_addr);
			if (ifa != NULL) {
				IFA_REMREF(ifa);
				ifa = NULL;
				ifscope = IFSCOPE_NONE;
			}
		}

		if (ip_select_srcif_debug && ifa != NULL) {
			if (ro->ro_rt != NULL) {
				printf("%s->%s ifscope %d->%d ifa_if %s "
				    "ro_if %s\n", s_src, s_dst, ifscope,
				    scope, if_name(ifa->ifa_ifp),
				    if_name(rt_ifp));
			} else {
				printf("%s->%s ifscope %d->%d ifa_if %s\n",
				    s_src, s_dst, ifscope, scope,
				    if_name(ifa->ifa_ifp));
			}
		}
	}

	/*
	 * Slow path; search for an interface having the corresponding source
	 * IP address if the scope was not specified by the caller, and:
	 *
	 *   1) There currently isn't any route, or,
	 *   2) The interface used by the route does not own that source
	 *	IP address; in this case, the route will get blown away
	 *	and we'll do a more specific scoped search using the newly
	 *	found interface.
	 */
	if (ifa == NULL && ifscope == IFSCOPE_NONE) {
		ifa = (struct ifaddr *)ifa_foraddr(src.s_addr);

		/*
		 * If we have the IP address, but not the route, we don't
		 * really know whether or not it belongs to the correct
		 * interface (it could be shared across multiple interfaces.)
		 * The only way to find out is to do a route lookup.
		 */
		if (ifa != NULL && ro->ro_rt == NULL) {
			struct rtentry *rt;
			struct sockaddr_in sin;
			struct ifaddr *oifa = NULL;

			bzero(&sin, sizeof (sin));
			sin.sin_family = AF_INET;
			sin.sin_len = sizeof (sin);
			sin.sin_addr = dst;

			lck_mtx_lock(rnh_lock);
			if ((rt = rt_lookup(TRUE, SA(&sin), NULL,
			    rt_tables[AF_INET], IFSCOPE_NONE)) != NULL) {
				RT_LOCK(rt);
				/*
				 * If the route uses a different interface,
				 * use that one instead.  The IP address of
				 * the ifaddr that we pick up here is not
				 * relevant.
				 */
				if (ifa->ifa_ifp != rt->rt_ifp) {
					oifa = ifa;
					ifa = rt->rt_ifa;
					IFA_ADDREF(ifa);
					RT_UNLOCK(rt);
				} else {
					RT_UNLOCK(rt);
				}
				rtfree_locked(rt);
			}
			lck_mtx_unlock(rnh_lock);

			if (oifa != NULL) {
				struct ifaddr *iifa;

				/*
				 * See if the interface pointed to by the
				 * route is configured with the source IP
				 * address of the packet.
				 */
				iifa = (struct ifaddr *)ifa_foraddr_scoped(
				    src.s_addr, ifa->ifa_ifp->if_index);

				if (iifa != NULL) {
					/*
					 * Found it; drop the original one
					 * as well as the route interface
					 * address, and use this instead.
					 */
					IFA_REMREF(oifa);
					IFA_REMREF(ifa);
					ifa = iifa;
				} else if (!ipforwarding ||
				    (rt->rt_flags & RTF_GATEWAY)) {
					/*
					 * This interface doesn't have that
					 * source IP address; drop the route
					 * interface address and just use the
					 * original one, and let the caller
					 * do a scoped route lookup.
					 */
					IFA_REMREF(ifa);
					ifa = oifa;
				} else {
					/*
					 * Forwarding is enabled and the source
					 * address belongs to one of our own
					 * interfaces which isn't the outgoing
					 * interface, and we have a route, and
					 * the destination is on a network that
					 * is directly attached (onlink); drop
					 * the original one and use the route
					 * interface address instead.
					 */
					IFA_REMREF(oifa);
				}
			}
		} else if (ifa != NULL && ro->ro_rt != NULL &&
		    !(ro->ro_rt->rt_flags & RTF_GATEWAY) &&
		    ifa->ifa_ifp != ro->ro_rt->rt_ifp && ipforwarding) {
			/*
			 * Forwarding is enabled and the source address belongs
			 * to one of our own interfaces which isn't the same
			 * as the interface used by the known route; drop the
			 * original one and use the route interface address.
			 */
			IFA_REMREF(ifa);
			ifa = ro->ro_rt->rt_ifa;
			IFA_ADDREF(ifa);
		}

		if (ip_select_srcif_debug && ifa != NULL) {
			printf("%s->%s ifscope %d ifa_if %s\n",
			    s_src, s_dst, ifscope, if_name(ifa->ifa_ifp));
		}
	}

	if (ro->ro_rt != NULL)
		RT_LOCK_ASSERT_HELD(ro->ro_rt);
	/*
	 * If there is a non-loopback route with the wrong interface, or if
	 * there is no interface configured with such an address, blow it
	 * away.  Except for local/loopback, we look for one with a matching
	 * interface scope/index.
	 */
	if (ro->ro_rt != NULL &&
	    (ifa == NULL || (ifa->ifa_ifp != rt_ifp && rt_ifp != lo_ifp) ||
	    !(ro->ro_rt->rt_flags & RTF_UP))) {
		if (ip_select_srcif_debug) {
			if (ifa != NULL) {
				printf("%s->%s ifscope %d ro_if %s != "
				    "ifa_if %s (cached route cleared)\n",
				    s_src, s_dst, ifscope, if_name(rt_ifp),
				    if_name(ifa->ifa_ifp));
			} else {
				printf("%s->%s ifscope %d ro_if %s "
				    "(no ifa_if found)\n",
				    s_src, s_dst, ifscope, if_name(rt_ifp));
			}
		}

		RT_UNLOCK(ro->ro_rt);
		ROUTE_RELEASE(ro);

		/*
		 * If the destination is IPv4 LLA and the route's interface
		 * doesn't match the source interface, then the source IP
		 * address is wrong; it most likely belongs to the primary
		 * interface associated with the IPv4 LL subnet.  Drop the
		 * packet rather than letting it go out and return an error
		 * to the ULP.  This actually applies not only to IPv4 LL
		 * but other shared subnets; for now we explicitly test only
		 * for the former case and save the latter for future.
		 */
		if (IN_LINKLOCAL(ntohl(dst.s_addr)) &&
		    !IN_LINKLOCAL(ntohl(src.s_addr)) && ifa != NULL) {
			IFA_REMREF(ifa);
			ifa = NULL;
		}
	}

	if (ip_select_srcif_debug && ifa == NULL) {
		printf("%s->%s ifscope %d (neither ro_if/ifa_if found)\n",
		    s_src, s_dst, ifscope);
	}

	/*
	 * If there is a route, mark it accordingly.  If there isn't one,
	 * we'll get here again during the next transmit (possibly with a
	 * route) and the flag will get set at that point.  For IPv4 LLA
	 * destination, mark it only if the route has been fully resolved;
	 * otherwise we want to come back here again when the route points
	 * to the interface over which the ARP reply arrives on.
	 */
	if (ro->ro_rt != NULL && (!IN_LINKLOCAL(ntohl(dst.s_addr)) ||
	    (ro->ro_rt->rt_gateway->sa_family == AF_LINK &&
	    SDL(ro->ro_rt->rt_gateway)->sdl_alen != 0))) {
		if (ifa != NULL)
			IFA_ADDREF(ifa);	/* for route */
		if (ro->ro_srcia != NULL)
			IFA_REMREF(ro->ro_srcia);
		ro->ro_srcia = ifa;
		ro->ro_flags |= ROF_SRCIF_SELECTED;
		RT_GENID_SYNC(ro->ro_rt);
	}

	if (ro->ro_rt != NULL)
		RT_UNLOCK(ro->ro_rt);

	return (ifa);
}

void
ip_output_checksum(struct ifnet *ifp, struct mbuf *m, int hlen, int ip_len,
    uint32_t *sw_csum)
{
	int tso = TSO_IPV4_OK(ifp, m);
	uint32_t hwcap = ifp->if_hwassist;

	m->m_pkthdr.csum_flags |= CSUM_IP;

	if (!hwcksum_tx) {
		/* do all in software; hardware checksum offload is disabled */
		*sw_csum = (CSUM_DELAY_DATA | CSUM_DELAY_IP) &
		    m->m_pkthdr.csum_flags;
	} else {
		/* do in software what the hardware cannot */
		*sw_csum = m->m_pkthdr.csum_flags &
		    ~IF_HWASSIST_CSUM_FLAGS(hwcap);
	}

	if (hlen != sizeof (struct ip)) {
		*sw_csum |= ((CSUM_DELAY_DATA | CSUM_DELAY_IP) &
		    m->m_pkthdr.csum_flags);
	} else if (!(*sw_csum & CSUM_DELAY_DATA) && (hwcap & CSUM_PARTIAL)) {
		/*
		 * Partial checksum offload, if non-IP fragment, and TCP only
		 * (no UDP support, as the hardware may not be able to convert
		 * +0 to -0 (0xffff) per RFC1122 4.1.3.4. unless the interface
		 * supports "invert zero" capability.)
		 */
		if (hwcksum_tx && !tso &&
		    ((m->m_pkthdr.csum_flags & CSUM_TCP) ||
		    ((hwcap & CSUM_ZERO_INVERT) &&
		    (m->m_pkthdr.csum_flags & CSUM_ZERO_INVERT))) &&
		    ip_len <= ifp->if_mtu) {
			uint16_t start = sizeof (struct ip);
			uint16_t ulpoff = m->m_pkthdr.csum_data & 0xffff;
			m->m_pkthdr.csum_flags |=
			    (CSUM_DATA_VALID | CSUM_PARTIAL);
			m->m_pkthdr.csum_tx_stuff = (ulpoff + start);
			m->m_pkthdr.csum_tx_start = start;
			/* do IP hdr chksum in software */
			*sw_csum = CSUM_DELAY_IP;
		} else {
			*sw_csum |= (CSUM_DELAY_DATA & m->m_pkthdr.csum_flags);
		}
	}

	if (*sw_csum & CSUM_DELAY_DATA) {
		in_delayed_cksum(m);
		*sw_csum &= ~CSUM_DELAY_DATA;
	}

	if (hwcksum_tx) {
		/*
		 * Drop off bits that aren't supported by hardware;
		 * also make sure to preserve non-checksum related bits.
		 */
		m->m_pkthdr.csum_flags =
		    ((m->m_pkthdr.csum_flags &
		    (IF_HWASSIST_CSUM_FLAGS(hwcap) | CSUM_DATA_VALID)) |
		    (m->m_pkthdr.csum_flags & ~IF_HWASSIST_CSUM_MASK));
	} else {
		/* drop all bits; hardware checksum offload is disabled */
		m->m_pkthdr.csum_flags = 0;
	}
}

/*
 * GRE protocol output for PPP/PPTP
 */
int
ip_gre_output(struct mbuf *m)
{
	struct route ro;
	int error;

	bzero(&ro, sizeof (ro));

	error = ip_output(m, NULL, &ro, 0, NULL, NULL);

	ROUTE_RELEASE(&ro);

	return (error);
}

static int
sysctl_reset_ip_output_stats SYSCTL_HANDLER_ARGS
{
#pragma unused(arg1, arg2)
	int error, i;

	i = ip_output_measure;
	error = sysctl_handle_int(oidp, &i, 0, req);
	if (error || req->newptr == USER_ADDR_NULL)
		goto done;
	/* impose bounds */
	if (i < 0 || i > 1) {
		error = EINVAL;
		goto done;
	}
	if (ip_output_measure != i && i == 1) {
		net_perf_initialize(&net_perf, ip_output_measure_bins);
	}
	ip_output_measure = i;
done:
	return (error);
}

static int
sysctl_ip_output_measure_bins SYSCTL_HANDLER_ARGS
{
#pragma unused(arg1, arg2)
	int error;
	uint64_t i;

	i = ip_output_measure_bins;
	error = sysctl_handle_quad(oidp, &i, 0, req);
	if (error || req->newptr == USER_ADDR_NULL)
		goto done;
	/* validate data */
	if (!net_perf_validate_bins(i)) {
		error = EINVAL;
		goto done;
	}
	ip_output_measure_bins = i;
done:
	return (error);
}

static int
sysctl_ip_output_getperf SYSCTL_HANDLER_ARGS
{
#pragma unused(oidp, arg1, arg2)
	if (req->oldptr == USER_ADDR_NULL)
		req->oldlen = (size_t)sizeof (struct ipstat);

	return (SYSCTL_OUT(req, &net_perf, MIN(sizeof (net_perf), req->oldlen)));
}