xref: /netbsd-src/sys/net/if.c (revision 3ad526a38650991367c07de424cf644e1c8a250e)

/*	$NetBSD: if.c,v 1.531 2024/12/16 05:18:37 ozaki-r Exp $	*/

/*-
 * Copyright (c) 1999, 2000, 2001, 2008 The NetBSD Foundation, Inc.
 * All rights reserved.
 *
 * This code is derived from software contributed to The NetBSD Foundation
 * by William Studenmund and Jason R. Thorpe.
 *
 * 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.
 *
 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. 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 FOUNDATION 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.
 */

/*
 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
 * 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. Neither the name of the project 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 PROJECT 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 PROJECT 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.
 */

/*
 * Copyright (c) 1980, 1986, 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. 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.
 *
 *	@(#)if.c	8.5 (Berkeley) 1/9/95
 */

#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: if.c,v 1.531 2024/12/16 05:18:37 ozaki-r Exp $");

#if defined(_KERNEL_OPT)
#include "opt_inet.h"
#include "opt_ipsec.h"
#include "opt_atalk.h"
#include "opt_wlan.h"
#include "opt_net_mpsafe.h"
#include "opt_mrouting.h"
#endif

#include <sys/param.h>
#include <sys/mbuf.h>
#include <sys/systm.h>
#include <sys/callout.h>
#include <sys/proc.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/domain.h>
#include <sys/protosw.h>
#include <sys/kernel.h>
#include <sys/ioctl.h>
#include <sys/sysctl.h>
#include <sys/syslog.h>
#include <sys/kauth.h>
#include <sys/kmem.h>
#include <sys/xcall.h>
#include <sys/cpu.h>
#include <sys/intr.h>
#include <sys/module_hook.h>
#include <sys/compat_stub.h>
#include <sys/msan.h>
#include <sys/hook.h>

#include <net/if.h>
#include <net/if_dl.h>
#include <net/if_ether.h>
#include <net/if_media.h>
#include <net80211/ieee80211.h>
#include <net80211/ieee80211_ioctl.h>
#include <net/if_types.h>
#include <net/route.h>
#include <sys/module.h>
#ifdef NETATALK
#include <netatalk/at_extern.h>
#include <netatalk/at.h>
#endif
#include <net/pfil.h>
#include <netinet/in.h>
#include <netinet/in_var.h>
#include <netinet/ip_encap.h>
#include <net/bpf.h>

#ifdef INET6
#include <netinet6/in6_var.h>
#include <netinet6/nd6.h>
#endif

#include "ether.h"

#include "bridge.h"
#if NBRIDGE > 0
#include <net/if_bridgevar.h>
#endif

#include "carp.h"
#if NCARP > 0
#include <netinet/ip_carp.h>
#endif

#include <compat/sys/sockio.h>

MALLOC_DEFINE(M_IFADDR, "ifaddr", "interface address");
MALLOC_DEFINE(M_IFMADDR, "ether_multi", "link-level multicast address");

/*
 * XXX reusing (ifp)->if_snd->ifq_lock rather than having another spin mutex
 * for each ifnet.  It doesn't matter because:
 * - if IFEF_MPSAFE is enabled, if_snd isn't used and lock contentions on
 *   ifq_lock don't happen
 * - if IFEF_MPSAFE is disabled, there is no lock contention on ifq_lock
 *   because if_snd, if_link_state_change and if_link_state_change_process
 *   are all called with KERNEL_LOCK
 */
#define IF_LINK_STATE_CHANGE_LOCK(ifp)		\
	mutex_enter((ifp)->if_snd.ifq_lock)
#define IF_LINK_STATE_CHANGE_UNLOCK(ifp)	\
	mutex_exit((ifp)->if_snd.ifq_lock)

/*
 * Global list of interfaces.
 */
/* DEPRECATED. Remove it once kvm(3) users disappeared */
struct ifnet_head		ifnet_list;

struct pslist_head		ifnet_pslist;
static ifnet_t **		ifindex2ifnet = NULL;
static u_int			if_index = 1;
static size_t			if_indexlim = 0;
static uint64_t			index_gen;
/* Mutex to protect the above objects. */
kmutex_t			ifnet_mtx __cacheline_aligned;
static struct psref_class	*ifnet_psref_class __read_mostly;
static pserialize_t		ifnet_psz;
static struct workqueue		*ifnet_link_state_wq __read_mostly;

static struct workqueue		*if_slowtimo_wq __read_mostly;

static kmutex_t			if_clone_mtx;

struct ifnet *lo0ifp;
int	ifqmaxlen = IFQ_MAXLEN;

struct psref_class		*ifa_psref_class __read_mostly;

static int	if_delroute_matcher(struct rtentry *, void *);

static bool if_is_unit(const char *);
static struct if_clone *if_clone_lookup(const char *, int *);

static LIST_HEAD(, if_clone) if_cloners = LIST_HEAD_INITIALIZER(if_cloners);
static int if_cloners_count;

/* Packet filtering hook for interfaces. */
pfil_head_t *			if_pfil __read_mostly;

static kauth_listener_t if_listener;

static int doifioctl(struct socket *, u_long, void *, struct lwp *);
static void sysctl_sndq_setup(struct sysctllog **, const char *,
    struct ifaltq *);
static void if_slowtimo_intr(void *);
static void if_slowtimo_work(struct work *, void *);
static int sysctl_if_watchdog(SYSCTLFN_PROTO);
static void sysctl_watchdog_setup(struct ifnet *);
static void if_attachdomain1(struct ifnet *);
static int ifconf(u_long, void *);
static int if_transmit(struct ifnet *, struct mbuf *);
static int if_clone_create(const char *);
static int if_clone_destroy(const char *);
static void if_link_state_change_work(struct work *, void *);
static void if_up_locked(struct ifnet *);
static void _if_down(struct ifnet *);
static void if_down_deactivated(struct ifnet *);

struct if_percpuq {
	struct ifnet	*ipq_ifp;
	void		*ipq_si;
	struct percpu	*ipq_ifqs;	/* struct ifqueue */
};

static struct mbuf *if_percpuq_dequeue(struct if_percpuq *);

static void if_percpuq_drops(void *, void *, struct cpu_info *);
static int sysctl_percpuq_drops_handler(SYSCTLFN_PROTO);
static void sysctl_percpuq_setup(struct sysctllog **, const char *,
    struct if_percpuq *);

struct if_deferred_start {
	struct ifnet	*ids_ifp;
	void		(*ids_if_start)(struct ifnet *);
	void		*ids_si;
};

static void if_deferred_start_softint(void *);
static void if_deferred_start_common(struct ifnet *);
static void if_deferred_start_destroy(struct ifnet *);

struct if_slowtimo_data {
	kmutex_t		isd_lock;
	struct callout		isd_ch;
	struct work		isd_work;
	struct ifnet		*isd_ifp;
	bool			isd_queued;
	bool			isd_dying;
	bool			isd_trigger;
};

/*
 * Hook for if_vlan - needed by if_agr
 */
struct if_vlan_vlan_input_hook_t if_vlan_vlan_input_hook;

static void if_sysctl_setup(struct sysctllog **);

static int
if_listener_cb(kauth_cred_t cred, kauth_action_t action, void *cookie,
    void *arg0, void *arg1, void *arg2, void *arg3)
{
	int result;
	enum kauth_network_req req;

	result = KAUTH_RESULT_DEFER;
	req = (enum kauth_network_req)(uintptr_t)arg1;

	if (action != KAUTH_NETWORK_INTERFACE)
		return result;

	if ((req == KAUTH_REQ_NETWORK_INTERFACE_GET) ||
	    (req == KAUTH_REQ_NETWORK_INTERFACE_SET))
		result = KAUTH_RESULT_ALLOW;

	return result;
}

/*
 * Network interface utility routines.
 *
 * Routines with ifa_ifwith* names take sockaddr *'s as
 * parameters.
 */
void
ifinit(void)
{

#if (defined(INET) || defined(INET6))
	encapinit();
#endif

	if_listener = kauth_listen_scope(KAUTH_SCOPE_NETWORK,
	    if_listener_cb, NULL);

	/* interfaces are available, inform socket code */
	ifioctl = doifioctl;
}

/*
 * XXX Initialization before configure().
 * XXX hack to get pfil_add_hook working in autoconf.
 */
void
ifinit1(void)
{
	int error __diagused;

#ifdef NET_MPSAFE
	printf("NET_MPSAFE enabled\n");
#endif

	mutex_init(&if_clone_mtx, MUTEX_DEFAULT, IPL_NONE);

	TAILQ_INIT(&ifnet_list);
	mutex_init(&ifnet_mtx, MUTEX_DEFAULT, IPL_NONE);
	ifnet_psz = pserialize_create();
	ifnet_psref_class = psref_class_create("ifnet", IPL_SOFTNET);
	ifa_psref_class = psref_class_create("ifa", IPL_SOFTNET);
	error = workqueue_create(&ifnet_link_state_wq, "iflnkst",
	    if_link_state_change_work, NULL, PRI_SOFTNET, IPL_NET,
	    WQ_MPSAFE);
	KASSERT(error == 0);
	PSLIST_INIT(&ifnet_pslist);

	error = workqueue_create(&if_slowtimo_wq, "ifwdog",
	    if_slowtimo_work, NULL, PRI_SOFTNET, IPL_SOFTCLOCK, WQ_MPSAFE);
	KASSERTMSG(error == 0, "error=%d", error);

	if_indexlim = 8;

	if_pfil = pfil_head_create(PFIL_TYPE_IFNET, NULL);
	KASSERT(if_pfil != NULL);

#if NETHER > 0 || defined(NETATALK) || defined(WLAN)
	etherinit();
#endif
}

/* XXX must be after domaininit() */
void
ifinit_post(void)
{

	if_sysctl_setup(NULL);
}

ifnet_t *
if_alloc(u_char type)
{

	return kmem_zalloc(sizeof(ifnet_t), KM_SLEEP);
}

void
if_free(ifnet_t *ifp)
{

	kmem_free(ifp, sizeof(ifnet_t));
}

void
if_initname(struct ifnet *ifp, const char *name, int unit)
{

	(void)snprintf(ifp->if_xname, sizeof(ifp->if_xname),
	    "%s%d", name, unit);
}

/*
 * Null routines used while an interface is going away.  These routines
 * just return an error.
 */

int
if_nulloutput(struct ifnet *ifp, struct mbuf *m,
    const struct sockaddr *so, const struct rtentry *rt)
{

	return ENXIO;
}

void
if_nullinput(struct ifnet *ifp, struct mbuf *m)
{

	/* Nothing. */
}

void
if_nullstart(struct ifnet *ifp)
{

	/* Nothing. */
}

int
if_nulltransmit(struct ifnet *ifp, struct mbuf *m)
{

	m_freem(m);
	return ENXIO;
}

int
if_nullioctl(struct ifnet *ifp, u_long cmd, void *data)
{

	return ENXIO;
}

int
if_nullinit(struct ifnet *ifp)
{

	return ENXIO;
}

void
if_nullstop(struct ifnet *ifp, int disable)
{

	/* Nothing. */
}

void
if_nullslowtimo(struct ifnet *ifp)
{

	/* Nothing. */
}

void
if_nulldrain(struct ifnet *ifp)
{

	/* Nothing. */
}

void
if_set_sadl(struct ifnet *ifp, const void *lla, u_char addrlen, bool factory)
{
	struct ifaddr *ifa;
	struct sockaddr_dl *sdl;

	ifp->if_addrlen = addrlen;
	if_alloc_sadl(ifp);
	ifa = ifp->if_dl;
	sdl = satosdl(ifa->ifa_addr);

	(void)sockaddr_dl_setaddr(sdl, sdl->sdl_len, lla, ifp->if_addrlen);
	if (factory) {
		KASSERT(ifp->if_hwdl == NULL);
		ifp->if_hwdl = ifp->if_dl;
		ifaref(ifp->if_hwdl);
	}
	/* TBD routing socket */
}

struct ifaddr *
if_dl_create(const struct ifnet *ifp, const struct sockaddr_dl **sdlp)
{
	unsigned socksize, ifasize;
	int addrlen, namelen;
	struct sockaddr_dl *mask, *sdl;
	struct ifaddr *ifa;

	namelen = strlen(ifp->if_xname);
	addrlen = ifp->if_addrlen;
	socksize = roundup(sockaddr_dl_measure(namelen, addrlen),
	    sizeof(long));
	ifasize = sizeof(*ifa) + 2 * socksize;
	ifa = malloc(ifasize, M_IFADDR, M_WAITOK | M_ZERO);

	sdl = (struct sockaddr_dl *)(ifa + 1);
	mask = (struct sockaddr_dl *)(socksize + (char *)sdl);

	sockaddr_dl_init(sdl, socksize, ifp->if_index, ifp->if_type,
	    ifp->if_xname, namelen, NULL, addrlen);
	mask->sdl_family = AF_LINK;
	mask->sdl_len = sockaddr_dl_measure(namelen, 0);
	memset(&mask->sdl_data[0], 0xff, namelen);
	ifa->ifa_rtrequest = link_rtrequest;
	ifa->ifa_addr = (struct sockaddr *)sdl;
	ifa->ifa_netmask = (struct sockaddr *)mask;
	ifa_psref_init(ifa);

	*sdlp = sdl;

	return ifa;
}

static void
if_sadl_setrefs(struct ifnet *ifp, struct ifaddr *ifa)
{
	const struct sockaddr_dl *sdl;

	ifp->if_dl = ifa;
	ifaref(ifa);
	sdl = satosdl(ifa->ifa_addr);
	ifp->if_sadl = sdl;
}

/*
 * Allocate the link level name for the specified interface.  This
 * is an attachment helper.  It must be called after ifp->if_addrlen
 * is initialized, which may not be the case when if_attach() is
 * called.
 */
void
if_alloc_sadl(struct ifnet *ifp)
{
	struct ifaddr *ifa;
	const struct sockaddr_dl *sdl;

	/*
	 * If the interface already has a link name, release it
	 * now.  This is useful for interfaces that can change
	 * link types, and thus switch link names often.
	 */
	if (ifp->if_sadl != NULL)
		if_free_sadl(ifp, 0);

	ifa = if_dl_create(ifp, &sdl);

	ifa_insert(ifp, ifa);
	if_sadl_setrefs(ifp, ifa);
}

static void
if_deactivate_sadl(struct ifnet *ifp)
{
	struct ifaddr *ifa;

	KASSERT(ifp->if_dl != NULL);

	ifa = ifp->if_dl;

	ifp->if_sadl = NULL;

	ifp->if_dl = NULL;
	ifafree(ifa);
}

static void
if_replace_sadl(struct ifnet *ifp, struct ifaddr *ifa)
{
	struct ifaddr *old;

	KASSERT(ifp->if_dl != NULL);

	old = ifp->if_dl;

	ifaref(ifa);
	/* XXX Update if_dl and if_sadl atomically */
	ifp->if_dl = ifa;
	ifp->if_sadl = satosdl(ifa->ifa_addr);

	ifafree(old);
}

void
if_activate_sadl(struct ifnet *ifp, struct ifaddr *ifa0,
    const struct sockaddr_dl *sdl)
{
	struct ifaddr *ifa;
	const int bound = curlwp_bind();

	KASSERT(ifa_held(ifa0));

	const int s = splsoftnet();

	if_replace_sadl(ifp, ifa0);

	int ss = pserialize_read_enter();
	IFADDR_READER_FOREACH(ifa, ifp) {
		struct psref psref;
		ifa_acquire(ifa, &psref);
		pserialize_read_exit(ss);

		rtinit(ifa, RTM_LLINFO_UPD, 0);

		ss = pserialize_read_enter();
		ifa_release(ifa, &psref);
	}
	pserialize_read_exit(ss);

	splx(s);
	curlwp_bindx(bound);
}

/*
 * Free the link level name for the specified interface.  This is
 * a detach helper.  This is called from if_detach().
 */
void
if_free_sadl(struct ifnet *ifp, int factory)
{
	struct ifaddr *ifa;

	if (factory && ifp->if_hwdl != NULL) {
		ifa = ifp->if_hwdl;
		ifp->if_hwdl = NULL;
		ifafree(ifa);
	}

	ifa = ifp->if_dl;
	if (ifa == NULL) {
		KASSERT(ifp->if_sadl == NULL);
		return;
	}

	KASSERT(ifp->if_sadl != NULL);

	const int s = splsoftnet();
	KASSERT(ifa->ifa_addr->sa_family == AF_LINK);
	ifa_remove(ifp, ifa);
	if_deactivate_sadl(ifp);
	splx(s);
}

static void
if_getindex(ifnet_t *ifp)
{
	bool hitlimit = false;
	char xnamebuf[HOOKNAMSIZ];

	ifp->if_index_gen = index_gen++;
	snprintf(xnamebuf, sizeof(xnamebuf), "%s-lshk", ifp->if_xname);
	ifp->if_linkstate_hooks = simplehook_create(IPL_NET,
	    xnamebuf);

	ifp->if_index = if_index;
	if (ifindex2ifnet == NULL) {
		if_index++;
		goto skip;
	}
	while (if_byindex(ifp->if_index)) {
		/*
		 * If we hit USHRT_MAX, we skip back to 0 since
		 * there are a number of places where the value
		 * of if_index or if_index itself is compared
		 * to or stored in an unsigned short.  By
		 * jumping back, we won't botch those assignments
		 * or comparisons.
		 */
		if (++if_index == 0) {
			if_index = 1;
		} else if (if_index == USHRT_MAX) {
			/*
			 * However, if we have to jump back to
			 * zero *twice* without finding an empty
			 * slot in ifindex2ifnet[], then there
			 * there are too many (>65535) interfaces.
			 */
			if (hitlimit)
				panic("too many interfaces");
			hitlimit = true;
			if_index = 1;
		}
		ifp->if_index = if_index;
	}
skip:
	/*
	 * ifindex2ifnet is indexed by if_index. Since if_index will
	 * grow dynamically, it should grow too.
	 */
	if (ifindex2ifnet == NULL || ifp->if_index >= if_indexlim) {
		size_t m, n, oldlim;
		void *q;

		oldlim = if_indexlim;
		while (ifp->if_index >= if_indexlim)
			if_indexlim <<= 1;

		/* grow ifindex2ifnet */
		m = oldlim * sizeof(struct ifnet *);
		n = if_indexlim * sizeof(struct ifnet *);
		q = malloc(n, M_IFADDR, M_WAITOK | M_ZERO);
		if (ifindex2ifnet != NULL) {
			memcpy(q, ifindex2ifnet, m);
			free(ifindex2ifnet, M_IFADDR);
		}
		ifindex2ifnet = (struct ifnet **)q;
	}
	ifindex2ifnet[ifp->if_index] = ifp;
}

/*
 * Initialize an interface and assign an index for it.
 *
 * It must be called prior to a device specific attach routine
 * (e.g., ether_ifattach and ieee80211_ifattach) or if_alloc_sadl,
 * and be followed by if_register:
 *
 *     if_initialize(ifp);
 *     ether_ifattach(ifp, enaddr);
 *     if_register(ifp);
 */
void
if_initialize(ifnet_t *ifp)
{

	KASSERT(if_indexlim > 0);
	TAILQ_INIT(&ifp->if_addrlist);

	/*
	 * Link level name is allocated later by a separate call to
	 * if_alloc_sadl().
	 */

	if (ifp->if_snd.ifq_maxlen == 0)
		ifp->if_snd.ifq_maxlen = ifqmaxlen;

	ifp->if_broadcastaddr = 0; /* reliably crash if used uninitialized */

	ifp->if_link_state = LINK_STATE_UNKNOWN;
	ifp->if_link_queue = -1; /* all bits set, see link_state_change() */
	ifp->if_link_scheduled = false;

	ifp->if_capenable = 0;
	ifp->if_csum_flags_tx = 0;
	ifp->if_csum_flags_rx = 0;

#ifdef ALTQ
	ifp->if_snd.altq_type = 0;
	ifp->if_snd.altq_disc = NULL;
	ifp->if_snd.altq_flags &= ALTQF_CANTCHANGE;
	ifp->if_snd.altq_tbr  = NULL;
	ifp->if_snd.altq_ifp  = ifp;
#endif

	IFQ_LOCK_INIT(&ifp->if_snd);

	ifp->if_pfil = pfil_head_create(PFIL_TYPE_IFNET, ifp);
	pfil_run_ifhooks(if_pfil, PFIL_IFNET_ATTACH, ifp);

	IF_AFDATA_LOCK_INIT(ifp);

	PSLIST_ENTRY_INIT(ifp, if_pslist_entry);
	PSLIST_INIT(&ifp->if_addr_pslist);
	psref_target_init(&ifp->if_psref, ifnet_psref_class);
	ifp->if_ioctl_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE);
	LIST_INIT(&ifp->if_multiaddrs);
	if_stats_init(ifp);

	IFNET_GLOBAL_LOCK();
	if_getindex(ifp);
	IFNET_GLOBAL_UNLOCK();
}

/*
 * Register an interface to the list of "active" interfaces.
 */
void
if_register(ifnet_t *ifp)
{
	/*
	 * If the driver has not supplied its own if_ioctl or if_stop,
	 * then supply the default.
	 */
	if (ifp->if_ioctl == NULL)
		ifp->if_ioctl = ifioctl_common;
	if (ifp->if_stop == NULL)
		ifp->if_stop = if_nullstop;

	sysctl_sndq_setup(&ifp->if_sysctl_log, ifp->if_xname, &ifp->if_snd);

	if (!STAILQ_EMPTY(&domains))
		if_attachdomain1(ifp);

	/* Announce the interface. */
	rt_ifannouncemsg(ifp, IFAN_ARRIVAL);

	if (ifp->if_slowtimo != NULL) {
		struct if_slowtimo_data *isd;

		isd = kmem_zalloc(sizeof(*isd), KM_SLEEP);
		mutex_init(&isd->isd_lock, MUTEX_DEFAULT, IPL_SOFTCLOCK);
		callout_init(&isd->isd_ch, CALLOUT_MPSAFE);
		callout_setfunc(&isd->isd_ch, if_slowtimo_intr, ifp);
		isd->isd_ifp = ifp;

		ifp->if_slowtimo_data = isd;

		if_slowtimo_intr(ifp);

		sysctl_watchdog_setup(ifp);
	}

	if (ifp->if_transmit == NULL || ifp->if_transmit == if_nulltransmit)
		ifp->if_transmit = if_transmit;

	IFNET_GLOBAL_LOCK();
	TAILQ_INSERT_TAIL(&ifnet_list, ifp, if_list);
	IFNET_WRITER_INSERT_TAIL(ifp);
	IFNET_GLOBAL_UNLOCK();
}

/*
 * The if_percpuq framework
 *
 * It allows network device drivers to execute the network stack
 * in softint (so called softint-based if_input). It utilizes
 * softint and percpu ifqueue. It doesn't distribute any packets
 * between CPUs, unlike pktqueue(9).
 *
 * Currently we support two options for device drivers to apply the framework:
 * - Use it implicitly with less changes
 *   - If you use if_attach in driver's _attach function and if_input in
 *     driver's Rx interrupt handler, a packet is queued and a softint handles
 *     the packet implicitly
 * - Use it explicitly in each driver (recommended)
 *   - You can use if_percpuq_* directly in your driver
 *   - In this case, you need to allocate struct if_percpuq in driver's softc
 *   - See wm(4) as a reference implementation
 */

static void
if_percpuq_softint(void *arg)
{
	struct if_percpuq *ipq = arg;
	struct ifnet *ifp = ipq->ipq_ifp;
	struct mbuf *m;

	while ((m = if_percpuq_dequeue(ipq)) != NULL) {
		if_statinc(ifp, if_ipackets);
		bpf_mtap(ifp, m, BPF_D_IN);

		ifp->_if_input(ifp, m);
	}
}

static void
if_percpuq_init_ifq(void *p, void *arg __unused, struct cpu_info *ci __unused)
{
	struct ifqueue *const ifq = p;

	memset(ifq, 0, sizeof(*ifq));
	ifq->ifq_maxlen = IFQ_MAXLEN;
}

struct if_percpuq *
if_percpuq_create(struct ifnet *ifp)
{
	struct if_percpuq *ipq;
	u_int flags = SOFTINT_NET;

	flags |= if_is_mpsafe(ifp) ? SOFTINT_MPSAFE : 0;

	ipq = kmem_zalloc(sizeof(*ipq), KM_SLEEP);
	ipq->ipq_ifp = ifp;
	ipq->ipq_si = softint_establish(flags, if_percpuq_softint, ipq);
	ipq->ipq_ifqs = percpu_alloc(sizeof(struct ifqueue));
	percpu_foreach(ipq->ipq_ifqs, &if_percpuq_init_ifq, NULL);

	sysctl_percpuq_setup(&ifp->if_sysctl_log, ifp->if_xname, ipq);

	return ipq;
}

static struct mbuf *
if_percpuq_dequeue(struct if_percpuq *ipq)
{
	struct mbuf *m;
	struct ifqueue *ifq;

	const int s = splnet();
	ifq = percpu_getref(ipq->ipq_ifqs);
	IF_DEQUEUE(ifq, m);
	percpu_putref(ipq->ipq_ifqs);
	splx(s);

	return m;
}

static void
if_percpuq_purge_ifq(void *p, void *arg __unused, struct cpu_info *ci __unused)
{
	struct ifqueue *const ifq = p;

	IF_PURGE(ifq);
}

void
if_percpuq_destroy(struct if_percpuq *ipq)
{

	/* if_detach may already destroy it */
	if (ipq == NULL)
		return;

	softint_disestablish(ipq->ipq_si);
	percpu_foreach(ipq->ipq_ifqs, &if_percpuq_purge_ifq, NULL);
	percpu_free(ipq->ipq_ifqs, sizeof(struct ifqueue));
	kmem_free(ipq, sizeof(*ipq));
}

void
if_percpuq_enqueue(struct if_percpuq *ipq, struct mbuf *m)
{
	struct ifqueue *ifq;

	KASSERT(ipq != NULL);

	const int s = splnet();
	ifq = percpu_getref(ipq->ipq_ifqs);
	if (IF_QFULL(ifq)) {
		IF_DROP(ifq);
		percpu_putref(ipq->ipq_ifqs);
		if_statinc(ipq->ipq_ifp, if_iqdrops);
		m_freem(m);
		goto out;
	}
	IF_ENQUEUE(ifq, m);
	percpu_putref(ipq->ipq_ifqs);

	softint_schedule(ipq->ipq_si);
out:
	splx(s);
}

static void
if_percpuq_drops(void *p, void *arg, struct cpu_info *ci __unused)
{
	struct ifqueue *const ifq = p;
	uint64_t *sum = arg;

	*sum += ifq->ifq_drops;
}

static int
sysctl_percpuq_drops_handler(SYSCTLFN_ARGS)
{
	struct sysctlnode node;
	struct if_percpuq *ipq;
	uint64_t sum = 0;
	int error;

	node = *rnode;
	ipq = node.sysctl_data;

	percpu_foreach(ipq->ipq_ifqs, if_percpuq_drops, &sum);

	node.sysctl_data = &sum;
	error = sysctl_lookup(SYSCTLFN_CALL(&node));
	if (error != 0 || newp == NULL)
		return error;

	return 0;
}

static void
sysctl_percpuq_setup(struct sysctllog **clog, const char* ifname,
    struct if_percpuq *ipq)
{
	const struct sysctlnode *cnode, *rnode;

	if (sysctl_createv(clog, 0, NULL, &rnode,
		       CTLFLAG_PERMANENT,
		       CTLTYPE_NODE, "interfaces",
		       SYSCTL_DESCR("Per-interface controls"),
		       NULL, 0, NULL, 0,
		       CTL_NET, CTL_CREATE, CTL_EOL) != 0)
		goto bad;

	if (sysctl_createv(clog, 0, &rnode, &rnode,
		       CTLFLAG_PERMANENT,
		       CTLTYPE_NODE, ifname,
		       SYSCTL_DESCR("Interface controls"),
		       NULL, 0, NULL, 0,
		       CTL_CREATE, CTL_EOL) != 0)
		goto bad;

	if (sysctl_createv(clog, 0, &rnode, &rnode,
		       CTLFLAG_PERMANENT,
		       CTLTYPE_NODE, "rcvq",
		       SYSCTL_DESCR("Interface input queue controls"),
		       NULL, 0, NULL, 0,
		       CTL_CREATE, CTL_EOL) != 0)
		goto bad;

#ifdef NOTYET
	/* XXX Should show each per-CPU queue length? */
	if (sysctl_createv(clog, 0, &rnode, &rnode,
		       CTLFLAG_PERMANENT,
		       CTLTYPE_INT, "len",
		       SYSCTL_DESCR("Current input queue length"),
		       sysctl_percpuq_len, 0, NULL, 0,
		       CTL_CREATE, CTL_EOL) != 0)
		goto bad;

	if (sysctl_createv(clog, 0, &rnode, &cnode,
		       CTLFLAG_PERMANENT | CTLFLAG_READWRITE,
		       CTLTYPE_INT, "maxlen",
		       SYSCTL_DESCR("Maximum allowed input queue length"),
		       sysctl_percpuq_maxlen_handler, 0, (void *)ipq, 0,
		       CTL_CREATE, CTL_EOL) != 0)
		goto bad;
#endif

	if (sysctl_createv(clog, 0, &rnode, &cnode,
		       CTLFLAG_PERMANENT,
		       CTLTYPE_QUAD, "drops",
		       SYSCTL_DESCR("Total packets dropped due to full input queue"),
		       sysctl_percpuq_drops_handler, 0, (void *)ipq, 0,
		       CTL_CREATE, CTL_EOL) != 0)
		goto bad;

	return;
bad:
	printf("%s: could not attach sysctl nodes\n", ifname);
	return;
}

/*
 * The deferred if_start framework
 *
 * The common APIs to defer if_start to softint when if_start is requested
 * from a device driver running in hardware interrupt context.
 */
/*
 * Call ifp->if_start (or equivalent) in a dedicated softint for
 * deferred if_start.
 */
static void
if_deferred_start_softint(void *arg)
{
	struct if_deferred_start *ids = arg;
	struct ifnet *ifp = ids->ids_ifp;

	ids->ids_if_start(ifp);
}

/*
 * The default callback function for deferred if_start.
 */
static void
if_deferred_start_common(struct ifnet *ifp)
{
	const int s = splnet();
	if_start_lock(ifp);
	splx(s);
}

static inline bool
if_snd_is_used(struct ifnet *ifp)
{

	return ALTQ_IS_ENABLED(&ifp->if_snd) ||
	    ifp->if_transmit == if_transmit ||
	    ifp->if_transmit == NULL ||
	    ifp->if_transmit == if_nulltransmit;
}

/*
 * Schedule deferred if_start.
 */
void
if_schedule_deferred_start(struct ifnet *ifp)
{

	KASSERT(ifp->if_deferred_start != NULL);

	if (if_snd_is_used(ifp) && IFQ_IS_EMPTY(&ifp->if_snd))
		return;

	softint_schedule(ifp->if_deferred_start->ids_si);
}

/*
 * Create an instance of deferred if_start. A driver should call the function
 * only if the driver needs deferred if_start. Drivers can setup their own
 * deferred if_start function via 2nd argument.
 */
void
if_deferred_start_init(struct ifnet *ifp, void (*func)(struct ifnet *))
{
	struct if_deferred_start *ids;
	u_int flags = SOFTINT_NET;

	flags |= if_is_mpsafe(ifp) ? SOFTINT_MPSAFE : 0;

	ids = kmem_zalloc(sizeof(*ids), KM_SLEEP);
	ids->ids_ifp = ifp;
	ids->ids_si = softint_establish(flags, if_deferred_start_softint, ids);
	if (func != NULL)
		ids->ids_if_start = func;
	else
		ids->ids_if_start = if_deferred_start_common;

	ifp->if_deferred_start = ids;
}

static void
if_deferred_start_destroy(struct ifnet *ifp)
{

	if (ifp->if_deferred_start == NULL)
		return;

	softint_disestablish(ifp->if_deferred_start->ids_si);
	kmem_free(ifp->if_deferred_start, sizeof(*ifp->if_deferred_start));
	ifp->if_deferred_start = NULL;
}

/*
 * The common interface input routine that is called by device drivers,
 * which should be used only when the driver's rx handler already runs
 * in softint.
 */
void
if_input(struct ifnet *ifp, struct mbuf *m)
{

	KASSERT(ifp->if_percpuq == NULL);
	KASSERT(!cpu_intr_p());

	if_statinc(ifp, if_ipackets);
	bpf_mtap(ifp, m, BPF_D_IN);

	ifp->_if_input(ifp, m);
}

/*
 * DEPRECATED. Use if_initialize and if_register instead.
 * See the above comment of if_initialize.
 *
 * Note that it implicitly enables if_percpuq to make drivers easy to
 * migrate softint-based if_input without much changes. If you don't
 * want to enable it, use if_initialize instead.
 */
void
if_attach(ifnet_t *ifp)
{

	if_initialize(ifp);
	ifp->if_percpuq = if_percpuq_create(ifp);
	if_register(ifp);
}

void
if_attachdomain(void)
{
	struct ifnet *ifp;
	const int bound = curlwp_bind();

	int s = pserialize_read_enter();
	IFNET_READER_FOREACH(ifp) {
		struct psref psref;
		psref_acquire(&psref, &ifp->if_psref, ifnet_psref_class);
		pserialize_read_exit(s);
		if_attachdomain1(ifp);
		s = pserialize_read_enter();
		psref_release(&psref, &ifp->if_psref, ifnet_psref_class);
	}
	pserialize_read_exit(s);
	curlwp_bindx(bound);
}

static void
if_attachdomain1(struct ifnet *ifp)
{
	struct domain *dp;
	const int s = splsoftnet();

	/* address family dependent data region */
	memset(ifp->if_afdata, 0, sizeof(ifp->if_afdata));
	DOMAIN_FOREACH(dp) {
		if (dp->dom_ifattach != NULL)
			ifp->if_afdata[dp->dom_family] =
			    (*dp->dom_ifattach)(ifp);
	}

	splx(s);
}

/*
 * Deactivate an interface.  This points all of the procedure
 * handles at error stubs.  May be called from interrupt context.
 */
void
if_deactivate(struct ifnet *ifp)
{
	const int s = splsoftnet();

	ifp->if_output	 = if_nulloutput;
	ifp->_if_input	 = if_nullinput;
	ifp->if_start	 = if_nullstart;
	ifp->if_transmit = if_nulltransmit;
	ifp->if_ioctl	 = if_nullioctl;
	ifp->if_init	 = if_nullinit;
	ifp->if_stop	 = if_nullstop;
	if (ifp->if_slowtimo)
		ifp->if_slowtimo = if_nullslowtimo;
	ifp->if_drain	 = if_nulldrain;

	/* No more packets may be enqueued. */
	ifp->if_snd.ifq_maxlen = 0;

	splx(s);
}

bool
if_is_deactivated(const struct ifnet *ifp)
{

	return ifp->if_output == if_nulloutput;
}

void
if_purgeaddrs(struct ifnet *ifp, int family,
    void (*purgeaddr)(struct ifaddr *))
{
	struct ifaddr *ifa, *nifa;
	int s;

	s = pserialize_read_enter();
	for (ifa = IFADDR_READER_FIRST(ifp); ifa; ifa = nifa) {
		nifa = IFADDR_READER_NEXT(ifa);
		if (ifa->ifa_addr->sa_family != family)
			continue;
		pserialize_read_exit(s);

		(*purgeaddr)(ifa);

		s = pserialize_read_enter();
	}
	pserialize_read_exit(s);
}

#ifdef IFAREF_DEBUG
static struct ifaddr **ifa_list;
static int ifa_list_size;

/* Depends on only one if_attach runs at once */
static void
if_build_ifa_list(struct ifnet *ifp)
{
	struct ifaddr *ifa;
	int i;

	KASSERT(ifa_list == NULL);
	KASSERT(ifa_list_size == 0);

	IFADDR_READER_FOREACH(ifa, ifp)
		ifa_list_size++;

	ifa_list = kmem_alloc(sizeof(*ifa) * ifa_list_size, KM_SLEEP);
	i = 0;
	IFADDR_READER_FOREACH(ifa, ifp) {
		ifa_list[i++] = ifa;
		ifaref(ifa);
	}
}

static void
if_check_and_free_ifa_list(struct ifnet *ifp)
{
	int i;
	struct ifaddr *ifa;

	if (ifa_list == NULL)
		return;

	for (i = 0; i < ifa_list_size; i++) {
		char buf[64];

		ifa = ifa_list[i];
		sockaddr_format(ifa->ifa_addr, buf, sizeof(buf));
		if (ifa->ifa_refcnt > 1) {
			log(LOG_WARNING,
			    "ifa(%s) still referenced (refcnt=%d)\n",
			    buf, ifa->ifa_refcnt - 1);
		} else
			log(LOG_DEBUG,
			    "ifa(%s) not referenced (refcnt=%d)\n",
			    buf, ifa->ifa_refcnt - 1);
		ifafree(ifa);
	}

	kmem_free(ifa_list, sizeof(*ifa) * ifa_list_size);
	ifa_list = NULL;
	ifa_list_size = 0;
}
#endif

/*
 * Detach an interface from the list of "active" interfaces,
 * freeing any resources as we go along.
 *
 * NOTE: This routine must be called with a valid thread context,
 * as it may block.
 */
void
if_detach(struct ifnet *ifp)
{
	struct socket so;
	struct ifaddr *ifa;
#ifdef IFAREF_DEBUG
	struct ifaddr *last_ifa = NULL;
#endif
	struct domain *dp;
	const struct protosw *pr;
	int i, family, purged;

#ifdef IFAREF_DEBUG
	if_build_ifa_list(ifp);
#endif
	/*
	 * XXX It's kind of lame that we have to have the
	 * XXX socket structure...
	 */
	memset(&so, 0, sizeof(so));

	const int s = splnet();

	sysctl_teardown(&ifp->if_sysctl_log);

	IFNET_LOCK(ifp);

	/*
	 * Unset all queued link states and pretend a
	 * link state change is scheduled.
	 * This stops any more link state changes occurring for this
	 * interface while it's being detached so it's safe
	 * to drain the workqueue.
	 */
	IF_LINK_STATE_CHANGE_LOCK(ifp);
	ifp->if_link_queue = -1; /* all bits set, see link_state_change() */
	ifp->if_link_scheduled = true;
	IF_LINK_STATE_CHANGE_UNLOCK(ifp);
	workqueue_wait(ifnet_link_state_wq, &ifp->if_link_work);

	if_deactivate(ifp);
	IFNET_UNLOCK(ifp);

	/*
	 * Unlink from the list and wait for all readers to leave
	 * from pserialize read sections.  Note that we can't do
	 * psref_target_destroy here.  See below.
	 */
	IFNET_GLOBAL_LOCK();
	ifindex2ifnet[ifp->if_index] = NULL;
	TAILQ_REMOVE(&ifnet_list, ifp, if_list);
	IFNET_WRITER_REMOVE(ifp);
	pserialize_perform(ifnet_psz);
	IFNET_GLOBAL_UNLOCK();

	if (ifp->if_slowtimo != NULL) {
		struct if_slowtimo_data *isd = ifp->if_slowtimo_data;

		mutex_enter(&isd->isd_lock);
		isd->isd_dying = true;
		mutex_exit(&isd->isd_lock);
		callout_halt(&isd->isd_ch, NULL);
		workqueue_wait(if_slowtimo_wq, &isd->isd_work);
		callout_destroy(&isd->isd_ch);
		mutex_destroy(&isd->isd_lock);
		kmem_free(isd, sizeof(*isd));

		ifp->if_slowtimo_data = NULL; /* paraonia */
		ifp->if_slowtimo = NULL;      /* paranoia */
	}
	if_deferred_start_destroy(ifp);

	/*
	 * Do an if_down() to give protocols a chance to do something.
	 */
	if_down_deactivated(ifp);

#ifdef ALTQ
	if (ALTQ_IS_ENABLED(&ifp->if_snd))
		altq_disable(&ifp->if_snd);
	if (ALTQ_IS_ATTACHED(&ifp->if_snd))
		altq_detach(&ifp->if_snd);
#endif

#if NCARP > 0
	/* Remove the interface from any carp group it is a part of.  */
	if (ifp->if_carp != NULL && ifp->if_type != IFT_CARP)
		carp_ifdetach(ifp);
#endif

	/*
	 * Ensure that all packets on protocol input pktqueues have been
	 * processed, or, at least, removed from the queues.
	 *
	 * A cross-call will ensure that the interrupts have completed.
	 * FIXME: not quite..
	 */
	pktq_ifdetach();
	xc_barrier(0);

	/*
	 * Rip all the addresses off the interface.  This should make
	 * all of the routes go away.
	 *
	 * pr_usrreq calls can remove an arbitrary number of ifaddrs
	 * from the list, including our "cursor", ifa.  For safety,
	 * and to honor the TAILQ abstraction, I just restart the
	 * loop after each removal.  Note that the loop will exit
	 * when all of the remaining ifaddrs belong to the AF_LINK
	 * family.  I am counting on the historical fact that at
	 * least one pr_usrreq in each address domain removes at
	 * least one ifaddr.
	 */
again:
	/*
	 * At this point, no other one tries to remove ifa in the list,
	 * so we don't need to take a lock or psref.  Avoid using
	 * IFADDR_READER_FOREACH to pass over an inspection of contract
	 * violations of pserialize.
	 */
	IFADDR_WRITER_FOREACH(ifa, ifp) {
		family = ifa->ifa_addr->sa_family;
#ifdef IFAREF_DEBUG
		printf("if_detach: ifaddr %p, family %d, refcnt %d\n",
		    ifa, family, ifa->ifa_refcnt);
		if (last_ifa != NULL && ifa == last_ifa)
			panic("if_detach: loop detected");
		last_ifa = ifa;
#endif
		if (family == AF_LINK)
			continue;
		dp = pffinddomain(family);
		KASSERTMSG(dp != NULL, "no domain for AF %d", family);
		/*
		 * XXX These PURGEIF calls are redundant with the
		 * purge-all-families calls below, but are left in for
		 * now both to make a smaller change, and to avoid
		 * unplanned interactions with clearing of
		 * ifp->if_addrlist.
		 */
		purged = 0;
		for (pr = dp->dom_protosw;
		     pr < dp->dom_protoswNPROTOSW; pr++) {
			so.so_proto = pr;
			if (pr->pr_usrreqs) {
				(void) (*pr->pr_usrreqs->pr_purgeif)(&so, ifp);
				purged = 1;
			}
		}
		if (purged == 0) {
			/*
			 * XXX What's really the best thing to do
			 * XXX here?  --thorpej@NetBSD.org
			 */
			printf("if_detach: WARNING: AF %d not purged\n",
			    family);
			ifa_remove(ifp, ifa);
		}
		goto again;
	}

	if_free_sadl(ifp, 1);

restart:
	IFADDR_WRITER_FOREACH(ifa, ifp) {
		family = ifa->ifa_addr->sa_family;
		KASSERT(family == AF_LINK);
		ifa_remove(ifp, ifa);
		goto restart;
	}

	/* Delete stray routes from the routing table. */
	for (i = 0; i <= AF_MAX; i++)
		rt_delete_matched_entries(i, if_delroute_matcher, ifp, false);

	DOMAIN_FOREACH(dp) {
		if (dp->dom_ifdetach != NULL && ifp->if_afdata[dp->dom_family])
		{
			void *p = ifp->if_afdata[dp->dom_family];
			if (p) {
				ifp->if_afdata[dp->dom_family] = NULL;
				(*dp->dom_ifdetach)(ifp, p);
			}
		}

		/*
		 * One would expect multicast memberships (INET and
		 * INET6) on UDP sockets to be purged by the PURGEIF
		 * calls above, but if all addresses were removed from
		 * the interface prior to destruction, the calls will
		 * not be made (e.g. ppp, for which pppd(8) generally
		 * removes addresses before destroying the interface).
		 * Because there is no invariant that multicast
		 * memberships only exist for interfaces with IPv4
		 * addresses, we must call PURGEIF regardless of
		 * addresses.  (Protocols which might store ifnet
		 * pointers are marked with PR_PURGEIF.)
		 */
		for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
		{
			so.so_proto = pr;
			if (pr->pr_usrreqs && pr->pr_flags & PR_PURGEIF)
				(void)(*pr->pr_usrreqs->pr_purgeif)(&so, ifp);
		}
	}

	/*
	 * Must be done after the above pr_purgeif because if_psref may be
	 * still used in pr_purgeif.
	 */
	psref_target_destroy(&ifp->if_psref, ifnet_psref_class);
	PSLIST_ENTRY_DESTROY(ifp, if_pslist_entry);

	pfil_run_ifhooks(if_pfil, PFIL_IFNET_DETACH, ifp);
	(void)pfil_head_destroy(ifp->if_pfil);

	/* Announce that the interface is gone. */
	rt_ifannouncemsg(ifp, IFAN_DEPARTURE);

	IF_AFDATA_LOCK_DESTROY(ifp);

	if (ifp->if_percpuq != NULL) {
		if_percpuq_destroy(ifp->if_percpuq);
		ifp->if_percpuq = NULL;
	}

	mutex_obj_free(ifp->if_ioctl_lock);
	ifp->if_ioctl_lock = NULL;
	mutex_obj_free(ifp->if_snd.ifq_lock);
	if_stats_fini(ifp);
	KASSERT(!simplehook_has_hooks(ifp->if_linkstate_hooks));
	simplehook_destroy(ifp->if_linkstate_hooks);

	splx(s);

#ifdef IFAREF_DEBUG
	if_check_and_free_ifa_list(ifp);
#endif
}

/*
 * Callback for a radix tree walk to delete all references to an
 * ifnet.
 */
static int
if_delroute_matcher(struct rtentry *rt, void *v)
{
	struct ifnet *ifp = (struct ifnet *)v;

	if (rt->rt_ifp == ifp)
		return 1;
	else
		return 0;
}

/*
 * Create a clone network interface.
 */
static int
if_clone_create(const char *name)
{
	struct if_clone *ifc;
	struct ifnet *ifp;
	struct psref psref;
	int unit;

	KASSERT(mutex_owned(&if_clone_mtx));

	ifc = if_clone_lookup(name, &unit);
	if (ifc == NULL)
		return EINVAL;

	ifp = if_get(name, &psref);
	if (ifp != NULL) {
		if_put(ifp, &psref);
		return EEXIST;
	}

	return (*ifc->ifc_create)(ifc, unit);
}

/*
 * Destroy a clone network interface.
 */
static int
if_clone_destroy(const char *name)
{
	struct if_clone *ifc;
	struct ifnet *ifp;
	struct psref psref;
	int error;
	int (*if_ioctlfn)(struct ifnet *, u_long, void *);

	KASSERT(mutex_owned(&if_clone_mtx));

	ifc = if_clone_lookup(name, NULL);
	if (ifc == NULL)
		return EINVAL;

	if (ifc->ifc_destroy == NULL)
		return EOPNOTSUPP;

	ifp = if_get(name, &psref);
	if (ifp == NULL)
		return ENXIO;

	/* We have to disable ioctls here */
	IFNET_LOCK(ifp);
	if_ioctlfn = ifp->if_ioctl;
	ifp->if_ioctl = if_nullioctl;
	IFNET_UNLOCK(ifp);

	/*
	 * We cannot call ifc_destroy with holding ifp.
	 * Releasing ifp here is safe thanks to if_clone_mtx.
	 */
	if_put(ifp, &psref);

	error = (*ifc->ifc_destroy)(ifp);

	if (error != 0) {
		/* We have to restore if_ioctl on error */
		IFNET_LOCK(ifp);
		ifp->if_ioctl = if_ioctlfn;
		IFNET_UNLOCK(ifp);
	}

	return error;
}

static bool
if_is_unit(const char *name)
{

	while (*name != '\0') {
		if (*name < '0' || *name > '9')
			return false;
		name++;
	}

	return true;
}

/*
 * Look up a network interface cloner.
 */
static struct if_clone *
if_clone_lookup(const char *name, int *unitp)
{
	struct if_clone *ifc;
	const char *cp;
	char *dp, ifname[IFNAMSIZ + 3];
	int unit;

	KASSERT(mutex_owned(&if_clone_mtx));

	strcpy(ifname, "if_");
	/* separate interface name from unit */
	/* TODO: search unit number from backward */
	for (dp = ifname + 3, cp = name; cp - name < IFNAMSIZ &&
	    *cp && !if_is_unit(cp);)
		*dp++ = *cp++;

	if (cp == name || cp - name == IFNAMSIZ || !*cp)
		return NULL;	/* No name or unit number */
	*dp++ = '\0';

again:
	LIST_FOREACH(ifc, &if_cloners, ifc_list) {
		if (strcmp(ifname + 3, ifc->ifc_name) == 0)
			break;
	}

	if (ifc == NULL) {
		int error;
		if (*ifname == '\0')
			return NULL;
		mutex_exit(&if_clone_mtx);
		error = module_autoload(ifname, MODULE_CLASS_DRIVER);
		mutex_enter(&if_clone_mtx);
		if (error)
			return NULL;
		*ifname = '\0';
		goto again;
	}

	unit = 0;
	while (cp - name < IFNAMSIZ && *cp) {
		if (*cp < '0' || *cp > '9' || unit >= INT_MAX / 10) {
			/* Bogus unit number. */
			return NULL;
		}
		unit = (unit * 10) + (*cp++ - '0');
	}

	if (unitp != NULL)
		*unitp = unit;
	return ifc;
}

/*
 * Register a network interface cloner.
 */
void
if_clone_attach(struct if_clone *ifc)
{

	mutex_enter(&if_clone_mtx);
	LIST_INSERT_HEAD(&if_cloners, ifc, ifc_list);
	if_cloners_count++;
	mutex_exit(&if_clone_mtx);
}

/*
 * Unregister a network interface cloner.
 */
void
if_clone_detach(struct if_clone *ifc)
{

	mutex_enter(&if_clone_mtx);
	LIST_REMOVE(ifc, ifc_list);
	if_cloners_count--;
	mutex_exit(&if_clone_mtx);
}

/*
 * Provide list of interface cloners to userspace.
 */
int
if_clone_list(int buf_count, char *buffer, int *total)
{
	char outbuf[IFNAMSIZ], *dst;
	struct if_clone *ifc;
	int count, error = 0;

	mutex_enter(&if_clone_mtx);
	*total = if_cloners_count;
	if ((dst = buffer) == NULL) {
		/* Just asking how many there are. */
		goto out;
	}

	if (buf_count < 0) {
		error = EINVAL;
		goto out;
	}

	count = (if_cloners_count < buf_count) ? if_cloners_count : buf_count;

	for (ifc = LIST_FIRST(&if_cloners); ifc != NULL && count != 0;
	     ifc = LIST_NEXT(ifc, ifc_list), count--, dst += IFNAMSIZ) {
		(void)strncpy(outbuf, ifc->ifc_name, sizeof(outbuf));
		if (outbuf[sizeof(outbuf) - 1] != '\0') {
			error = ENAMETOOLONG;
			goto out;
		}
		error = copyout(outbuf, dst, sizeof(outbuf));
		if (error != 0)
			break;
	}

out:
	mutex_exit(&if_clone_mtx);
	return error;
}

void
ifa_psref_init(struct ifaddr *ifa)
{

	psref_target_init(&ifa->ifa_psref, ifa_psref_class);
}

void
ifaref(struct ifaddr *ifa)
{

	atomic_inc_uint(&ifa->ifa_refcnt);
}

void
ifafree(struct ifaddr *ifa)
{
	KASSERT(ifa != NULL);
	KASSERTMSG(ifa->ifa_refcnt > 0, "ifa_refcnt=%d", ifa->ifa_refcnt);

	membar_release();
	if (atomic_dec_uint_nv(&ifa->ifa_refcnt) != 0)
		return;
	membar_acquire();
	free(ifa, M_IFADDR);
}

bool
ifa_is_destroying(struct ifaddr *ifa)
{

	return ISSET(ifa->ifa_flags, IFA_DESTROYING);
}

void
ifa_insert(struct ifnet *ifp, struct ifaddr *ifa)
{

	ifa->ifa_ifp = ifp;

	/*
	 * Check MP-safety for IFEF_MPSAFE drivers.
	 * Check !IFF_RUNNING for initialization routines that normally don't
	 * take IFNET_LOCK but it's safe because there is no competitor.
	 * XXX there are false positive cases because IFF_RUNNING can be off on
	 * if_stop.
	 */
	KASSERT(!if_is_mpsafe(ifp) || !ISSET(ifp->if_flags, IFF_RUNNING) ||
	    IFNET_LOCKED(ifp));

	TAILQ_INSERT_TAIL(&ifp->if_addrlist, ifa, ifa_list);
	IFADDR_ENTRY_INIT(ifa);
	IFADDR_WRITER_INSERT_TAIL(ifp, ifa);

	ifaref(ifa);
}

void
ifa_remove(struct ifnet *ifp, struct ifaddr *ifa)
{

	KASSERT(ifa->ifa_ifp == ifp);
	/*
	 * Check MP-safety for IFEF_MPSAFE drivers.
	 * if_is_deactivated indicates ifa_remove is called from if_detach
	 * where it is safe even if IFNET_LOCK isn't held.
	 */
	KASSERT(!if_is_mpsafe(ifp) || if_is_deactivated(ifp) ||
	    IFNET_LOCKED(ifp));

	TAILQ_REMOVE(&ifp->if_addrlist, ifa, ifa_list);
	IFADDR_WRITER_REMOVE(ifa);
#ifdef NET_MPSAFE
	IFNET_GLOBAL_LOCK();
	pserialize_perform(ifnet_psz);
	IFNET_GLOBAL_UNLOCK();
#endif

#ifdef NET_MPSAFE
	psref_target_destroy(&ifa->ifa_psref, ifa_psref_class);
#endif
	IFADDR_ENTRY_DESTROY(ifa);
	ifafree(ifa);
}

void
ifa_acquire(struct ifaddr *ifa, struct psref *psref)
{

	PSREF_DEBUG_FILL_RETURN_ADDRESS(psref);
	psref_acquire(psref, &ifa->ifa_psref, ifa_psref_class);
}

void
ifa_release(struct ifaddr *ifa, struct psref *psref)
{

	if (ifa == NULL)
		return;

	psref_release(psref, &ifa->ifa_psref, ifa_psref_class);
}

bool
ifa_held(struct ifaddr *ifa)
{

	return psref_held(&ifa->ifa_psref, ifa_psref_class);
}

static inline int
equal(const struct sockaddr *sa1, const struct sockaddr *sa2)
{

	return sockaddr_cmp(sa1, sa2) == 0;
}

/*
 * Locate an interface based on a complete address.
 */
/*ARGSUSED*/
struct ifaddr *
ifa_ifwithaddr(const struct sockaddr *addr)
{
	struct ifnet *ifp;
	struct ifaddr *ifa;

	IFNET_READER_FOREACH(ifp) {
		if (if_is_deactivated(ifp))
			continue;
		IFADDR_READER_FOREACH(ifa, ifp) {
			if (ifa->ifa_addr->sa_family != addr->sa_family)
				continue;
			if (equal(addr, ifa->ifa_addr))
				return ifa;
			if ((ifp->if_flags & IFF_BROADCAST) &&
			    ifa->ifa_broadaddr &&
			    /* IP6 doesn't have broadcast */
			    ifa->ifa_broadaddr->sa_len != 0 &&
			    equal(ifa->ifa_broadaddr, addr))
				return ifa;
		}
	}
	return NULL;
}

struct ifaddr *
ifa_ifwithaddr_psref(const struct sockaddr *addr, struct psref *psref)
{
	struct ifaddr *ifa;
	int s = pserialize_read_enter();

	ifa = ifa_ifwithaddr(addr);
	if (ifa != NULL)
		ifa_acquire(ifa, psref);
	pserialize_read_exit(s);

	return ifa;
}

/*
 * Locate the point to point interface with a given destination address.
 */
/*ARGSUSED*/
struct ifaddr *
ifa_ifwithdstaddr(const struct sockaddr *addr)
{
	struct ifnet *ifp;
	struct ifaddr *ifa;

	IFNET_READER_FOREACH(ifp) {
		if (if_is_deactivated(ifp))
			continue;
		if ((ifp->if_flags & IFF_POINTOPOINT) == 0)
			continue;
		IFADDR_READER_FOREACH(ifa, ifp) {
			if (ifa->ifa_addr->sa_family != addr->sa_family ||
			    ifa->ifa_dstaddr == NULL)
				continue;
			if (equal(addr, ifa->ifa_dstaddr))
				return ifa;
		}
	}

	return NULL;
}

struct ifaddr *
ifa_ifwithdstaddr_psref(const struct sockaddr *addr, struct psref *psref)
{
	struct ifaddr *ifa;
	int s;

	s = pserialize_read_enter();
	ifa = ifa_ifwithdstaddr(addr);
	if (ifa != NULL)
		ifa_acquire(ifa, psref);
	pserialize_read_exit(s);

	return ifa;
}

/*
 * Find an interface on a specific network.  If many, choice
 * is most specific found.
 */
struct ifaddr *
ifa_ifwithnet(const struct sockaddr *addr)
{
	struct ifnet *ifp;
	struct ifaddr *ifa, *ifa_maybe = NULL;
	const struct sockaddr_dl *sdl;
	u_int af = addr->sa_family;
	const char *addr_data = addr->sa_data, *cplim;

	if (af == AF_LINK) {
		sdl = satocsdl(addr);
		if (sdl->sdl_index && sdl->sdl_index < if_indexlim &&
		    ifindex2ifnet[sdl->sdl_index] &&
		    !if_is_deactivated(ifindex2ifnet[sdl->sdl_index])) {
			return ifindex2ifnet[sdl->sdl_index]->if_dl;
		}
	}
#ifdef NETATALK
	if (af == AF_APPLETALK) {
		const struct sockaddr_at *sat, *sat2;
		sat = (const struct sockaddr_at *)addr;
		IFNET_READER_FOREACH(ifp) {
			if (if_is_deactivated(ifp))
				continue;
			ifa = at_ifawithnet((const struct sockaddr_at *)addr,
			    ifp);
			if (ifa == NULL)
				continue;
			sat2 = (struct sockaddr_at *)ifa->ifa_addr;
			if (sat2->sat_addr.s_net == sat->sat_addr.s_net)
				return ifa; /* exact match */
			if (ifa_maybe == NULL) {
				/* else keep the if with the right range */
				ifa_maybe = ifa;
			}
		}
		return ifa_maybe;
	}
#endif
	IFNET_READER_FOREACH(ifp) {
		if (if_is_deactivated(ifp))
			continue;
		IFADDR_READER_FOREACH(ifa, ifp) {
			const char *cp, *cp2, *cp3;

			if (ifa->ifa_addr->sa_family != af ||
			    ifa->ifa_netmask == NULL)
 next:				continue;
			cp = addr_data;
			cp2 = ifa->ifa_addr->sa_data;
			cp3 = ifa->ifa_netmask->sa_data;
			cplim = (const char *)ifa->ifa_netmask +
			    ifa->ifa_netmask->sa_len;
			while (cp3 < cplim) {
				if ((*cp++ ^ *cp2++) & *cp3++) {
					/* want to continue for() loop */
					goto next;
				}
			}
			if (ifa_maybe == NULL ||
			    rt_refines(ifa->ifa_netmask,
				       ifa_maybe->ifa_netmask))
				ifa_maybe = ifa;
		}
	}
	return ifa_maybe;
}

struct ifaddr *
ifa_ifwithnet_psref(const struct sockaddr *addr, struct psref *psref)
{
	struct ifaddr *ifa;
	int s;

	s = pserialize_read_enter();
	ifa = ifa_ifwithnet(addr);
	if (ifa != NULL)
		ifa_acquire(ifa, psref);
	pserialize_read_exit(s);

	return ifa;
}

/*
 * Find the interface of the address.
 */
struct ifaddr *
ifa_ifwithladdr(const struct sockaddr *addr)
{
	struct ifaddr *ia;

	if ((ia = ifa_ifwithaddr(addr)) || (ia = ifa_ifwithdstaddr(addr)) ||
	    (ia = ifa_ifwithnet(addr)))
		return ia;
	return NULL;
}

struct ifaddr *
ifa_ifwithladdr_psref(const struct sockaddr *addr, struct psref *psref)
{
	struct ifaddr *ifa;
	int s;

	s = pserialize_read_enter();
	ifa = ifa_ifwithladdr(addr);
	if (ifa != NULL)
		ifa_acquire(ifa, psref);
	pserialize_read_exit(s);

	return ifa;
}

/*
 * Find an interface using a specific address family
 */
struct ifaddr *
ifa_ifwithaf(int af)
{
	struct ifnet *ifp;
	struct ifaddr *ifa = NULL;
	int s;

	s = pserialize_read_enter();
	IFNET_READER_FOREACH(ifp) {
		if (if_is_deactivated(ifp))
			continue;
		IFADDR_READER_FOREACH(ifa, ifp) {
			if (ifa->ifa_addr->sa_family == af)
				goto out;
		}
	}
out:
	pserialize_read_exit(s);
	return ifa;
}

/*
 * Find an interface address specific to an interface best matching
 * a given address.
 */
struct ifaddr *
ifaof_ifpforaddr(const struct sockaddr *addr, struct ifnet *ifp)
{
	struct ifaddr *ifa;
	const char *cp, *cp2, *cp3;
	const char *cplim;
	struct ifaddr *ifa_maybe = 0;
	u_int af = addr->sa_family;

	if (if_is_deactivated(ifp))
		return NULL;

	if (af >= AF_MAX)
		return NULL;

	IFADDR_READER_FOREACH(ifa, ifp) {
		if (ifa->ifa_addr->sa_family != af)
			continue;
		ifa_maybe = ifa;
		if (ifa->ifa_netmask == NULL) {
			if (equal(addr, ifa->ifa_addr) ||
			    (ifa->ifa_dstaddr &&
			     equal(addr, ifa->ifa_dstaddr)))
				return ifa;
			continue;
		}
		cp = addr->sa_data;
		cp2 = ifa->ifa_addr->sa_data;
		cp3 = ifa->ifa_netmask->sa_data;
		cplim = ifa->ifa_netmask->sa_len + (char *)ifa->ifa_netmask;
		for (; cp3 < cplim; cp3++) {
			if ((*cp++ ^ *cp2++) & *cp3)
				break;
		}
		if (cp3 == cplim)
			return ifa;
	}
	return ifa_maybe;
}

struct ifaddr *
ifaof_ifpforaddr_psref(const struct sockaddr *addr, struct ifnet *ifp,
    struct psref *psref)
{
	struct ifaddr *ifa;
	int s;

	s = pserialize_read_enter();
	ifa = ifaof_ifpforaddr(addr, ifp);
	if (ifa != NULL)
		ifa_acquire(ifa, psref);
	pserialize_read_exit(s);

	return ifa;
}

/*
 * Default action when installing a route with a Link Level gateway.
 * Lookup an appropriate real ifa to point to.
 * This should be moved to /sys/net/link.c eventually.
 */
void
link_rtrequest(int cmd, struct rtentry *rt, const struct rt_addrinfo *info)
{
	struct ifaddr *ifa;
	const struct sockaddr *dst;
	struct ifnet *ifp;
	struct psref psref;

	if (cmd != RTM_ADD || ISSET(info->rti_flags, RTF_DONTCHANGEIFA))
		return;
	ifp = rt->rt_ifa->ifa_ifp;
	dst = rt_getkey(rt);
	if ((ifa = ifaof_ifpforaddr_psref(dst, ifp, &psref)) != NULL) {
		rt_replace_ifa(rt, ifa);
		if (ifa->ifa_rtrequest && ifa->ifa_rtrequest != link_rtrequest)
			ifa->ifa_rtrequest(cmd, rt, info);
		ifa_release(ifa, &psref);
	}
}

/*
 * bitmask macros to manage a densely packed link_state change queue.
 * Because we need to store LINK_STATE_UNKNOWN(0), LINK_STATE_DOWN(1) and
 * LINK_STATE_UP(2) we need 2 bits for each state change.
 * As a state change to store is 0, treat all bits set as an unset item.
 */
#define LQ_ITEM_BITS		2
#define LQ_ITEM_MASK		((1 << LQ_ITEM_BITS) - 1)
#define LQ_MASK(i)		(LQ_ITEM_MASK << (i) * LQ_ITEM_BITS)
#define LINK_STATE_UNSET	LQ_ITEM_MASK
#define LQ_ITEM(q, i)		(((q) & LQ_MASK((i))) >> (i) * LQ_ITEM_BITS)
#define LQ_STORE(q, i, v)						      \
	do {								      \
		(q) &= ~LQ_MASK((i));					      \
		(q) |= (v) << (i) * LQ_ITEM_BITS;			      \
	} while (0 /* CONSTCOND */)
#define LQ_MAX(q)		((sizeof((q)) * NBBY) / LQ_ITEM_BITS)
#define LQ_POP(q, v)							      \
	do {								      \
		(v) = LQ_ITEM((q), 0);					      \
		(q) >>= LQ_ITEM_BITS;					      \
		(q) |= LINK_STATE_UNSET << (LQ_MAX((q)) - 1) * LQ_ITEM_BITS;  \
	} while (0 /* CONSTCOND */)
#define LQ_PUSH(q, v)							      \
	do {								      \
		(q) >>= LQ_ITEM_BITS;					      \
		(q) |= (v) << (LQ_MAX((q)) - 1) * LQ_ITEM_BITS;		      \
	} while (0 /* CONSTCOND */)
#define LQ_FIND_UNSET(q, i)						      \
	for ((i) = 0; i < LQ_MAX((q)); (i)++) {				      \
		if (LQ_ITEM((q), (i)) == LINK_STATE_UNSET)		      \
			break;						      \
	}

/*
 * Handle a change in the interface link state and
 * queue notifications.
 */
void
if_link_state_change(struct ifnet *ifp, int link_state)
{
	int idx;

	/* Ensure change is to a valid state */
	switch (link_state) {
	case LINK_STATE_UNKNOWN:	/* FALLTHROUGH */
	case LINK_STATE_DOWN:		/* FALLTHROUGH */
	case LINK_STATE_UP:
		break;
	default:
#ifdef DEBUG
		printf("%s: invalid link state %d\n",
		    ifp->if_xname, link_state);
#endif
		return;
	}

	IF_LINK_STATE_CHANGE_LOCK(ifp);

	/* Find the last unset event in the queue. */
	LQ_FIND_UNSET(ifp->if_link_queue, idx);

	if (idx == 0) {
		/*
		 * There is no queue of link state changes.
		 * As we have the lock we can safely compare against the
		 * current link state and return if the same.
		 * Otherwise, if scheduled is true then the interface is being
		 * detached and the queue is being drained so we need
		 * to avoid queuing more work.
		 */
		 if (ifp->if_link_state == link_state ||
		     ifp->if_link_scheduled)
			goto out;
	} else {
		/* Ensure link_state doesn't match the last queued state. */
		if (LQ_ITEM(ifp->if_link_queue, idx - 1)
		    == (uint8_t)link_state)
			goto out;
	}

	/* Handle queue overflow. */
	if (idx == LQ_MAX(ifp->if_link_queue)) {
		uint8_t lost;

		/*
		 * The DOWN state must be protected from being pushed off
		 * the queue to ensure that userland will always be
		 * in a sane state.
		 * Because DOWN is protected, there is no need to protect
		 * UNKNOWN.
		 * It should be invalid to change from any other state to
		 * UNKNOWN anyway ...
		 */
		lost = LQ_ITEM(ifp->if_link_queue, 0);
		LQ_PUSH(ifp->if_link_queue, (uint8_t)link_state);
		if (lost == LINK_STATE_DOWN) {
			lost = LQ_ITEM(ifp->if_link_queue, 0);
			LQ_STORE(ifp->if_link_queue, 0, LINK_STATE_DOWN);
		}
		printf("%s: lost link state change %s\n",
		    ifp->if_xname,
		    lost == LINK_STATE_UP ? "UP" :
		    lost == LINK_STATE_DOWN ? "DOWN" :
		    "UNKNOWN");
	} else
		LQ_STORE(ifp->if_link_queue, idx, (uint8_t)link_state);

	if (ifp->if_link_scheduled)
		goto out;

	ifp->if_link_scheduled = true;
	workqueue_enqueue(ifnet_link_state_wq, &ifp->if_link_work, NULL);

out:
	IF_LINK_STATE_CHANGE_UNLOCK(ifp);
}

/*
 * Handle interface link state change notifications.
 */
static void
if_link_state_change_process(struct ifnet *ifp, int link_state)
{
	struct domain *dp;
	const int s = splnet();
	bool notify;

	KASSERT(!cpu_intr_p());

	IF_LINK_STATE_CHANGE_LOCK(ifp);

	/* Ensure the change is still valid. */
	if (ifp->if_link_state == link_state) {
		IF_LINK_STATE_CHANGE_UNLOCK(ifp);
		splx(s);
		return;
	}

#ifdef DEBUG
	log(LOG_DEBUG, "%s: link state %s (was %s)\n", ifp->if_xname,
		link_state == LINK_STATE_UP ? "UP" :
		link_state == LINK_STATE_DOWN ? "DOWN" :
		"UNKNOWN",
		ifp->if_link_state == LINK_STATE_UP ? "UP" :
		ifp->if_link_state == LINK_STATE_DOWN ? "DOWN" :
		"UNKNOWN");
#endif

	/*
	 * When going from UNKNOWN to UP, we need to mark existing
	 * addresses as tentative and restart DAD as we may have
	 * erroneously not found a duplicate.
	 *
	 * This needs to happen before rt_ifmsg to avoid a race where
	 * listeners would have an address and expect it to work right
	 * away.
	 */
	notify = (link_state == LINK_STATE_UP &&
	    ifp->if_link_state == LINK_STATE_UNKNOWN);
	ifp->if_link_state = link_state;
	/* The following routines may sleep so release the spin mutex */
	IF_LINK_STATE_CHANGE_UNLOCK(ifp);

	KERNEL_LOCK_UNLESS_NET_MPSAFE();
	if (notify) {
		DOMAIN_FOREACH(dp) {
			if (dp->dom_if_link_state_change != NULL)
				dp->dom_if_link_state_change(ifp,
				    LINK_STATE_DOWN);
		}
	}

	/* Notify that the link state has changed. */
	rt_ifmsg(ifp);

	simplehook_dohooks(ifp->if_linkstate_hooks);

	DOMAIN_FOREACH(dp) {
		if (dp->dom_if_link_state_change != NULL)
			dp->dom_if_link_state_change(ifp, link_state);
	}
	KERNEL_UNLOCK_UNLESS_NET_MPSAFE();
	splx(s);
}

/*
 * Process the interface link state change queue.
 */
static void
if_link_state_change_work(struct work *work, void *arg)
{
	struct ifnet *ifp = container_of(work, struct ifnet, if_link_work);
	uint8_t state;

	KERNEL_LOCK_UNLESS_NET_MPSAFE();
	const int s = splnet();

	/*
	 * Pop a link state change from the queue and process it.
	 * If there is nothing to process then if_detach() has been called.
	 * We keep if_link_scheduled = true so the queue can safely drain
	 * without more work being queued.
	 */
	IF_LINK_STATE_CHANGE_LOCK(ifp);
	LQ_POP(ifp->if_link_queue, state);
	IF_LINK_STATE_CHANGE_UNLOCK(ifp);
	if (state == LINK_STATE_UNSET)
		goto out;

	if_link_state_change_process(ifp, state);

	/* If there is a link state change to come, schedule it. */
	IF_LINK_STATE_CHANGE_LOCK(ifp);
	if (LQ_ITEM(ifp->if_link_queue, 0) != LINK_STATE_UNSET) {
		ifp->if_link_scheduled = true;
		workqueue_enqueue(ifnet_link_state_wq, &ifp->if_link_work,
		    NULL);
	} else
		ifp->if_link_scheduled = false;
	IF_LINK_STATE_CHANGE_UNLOCK(ifp);

out:
	splx(s);
	KERNEL_UNLOCK_UNLESS_NET_MPSAFE();
}

void *
if_linkstate_change_establish(struct ifnet *ifp, void (*fn)(void *), void *arg)
{
	khook_t *hk;

	hk = simplehook_establish(ifp->if_linkstate_hooks, fn, arg);

	return (void *)hk;
}

void
if_linkstate_change_disestablish(struct ifnet *ifp, void *vhook,
    kmutex_t *lock)
{

	simplehook_disestablish(ifp->if_linkstate_hooks, vhook, lock);
}

/*
 * Used to mark addresses on an interface as DETATCHED or TENTATIVE
 * and thus start Duplicate Address Detection without changing the
 * real link state.
 */
void
if_domain_link_state_change(struct ifnet *ifp, int link_state)
{
	struct domain *dp;

	const int s = splnet();
	KERNEL_LOCK_UNLESS_NET_MPSAFE();

	DOMAIN_FOREACH(dp) {
		if (dp->dom_if_link_state_change != NULL)
			dp->dom_if_link_state_change(ifp, link_state);
	}

	splx(s);
	KERNEL_UNLOCK_UNLESS_NET_MPSAFE();
}

/*
 * Default action when installing a local route on a point-to-point
 * interface.
 */
void
p2p_rtrequest(int req, struct rtentry *rt,
    __unused const struct rt_addrinfo *info)
{
	struct ifnet *ifp = rt->rt_ifp;
	struct ifaddr *ifa, *lo0ifa;
	int s = pserialize_read_enter();

	switch (req) {
	case RTM_ADD:
		if ((rt->rt_flags & RTF_LOCAL) == 0)
			break;

		rt->rt_ifp = lo0ifp;

		if (ISSET(info->rti_flags, RTF_DONTCHANGEIFA))
			break;

		IFADDR_READER_FOREACH(ifa, ifp) {
			if (equal(rt_getkey(rt), ifa->ifa_addr))
				break;
		}
		if (ifa == NULL)
			break;

		/*
		 * Ensure lo0 has an address of the same family.
		 */
		IFADDR_READER_FOREACH(lo0ifa, lo0ifp) {
			if (lo0ifa->ifa_addr->sa_family ==
			    ifa->ifa_addr->sa_family)
				break;
		}
		if (lo0ifa == NULL)
			break;

		/*
		 * Make sure to set rt->rt_ifa to the interface
		 * address we are using, otherwise we will have trouble
		 * with source address selection.
		 */
		if (ifa != rt->rt_ifa)
			rt_replace_ifa(rt, ifa);
		break;
	case RTM_DELETE:
	default:
		break;
	}
	pserialize_read_exit(s);
}

static void
_if_down(struct ifnet *ifp)
{
	struct ifaddr *ifa;
	struct domain *dp;
	struct psref psref;

	ifp->if_flags &= ~IFF_UP;
	nanotime(&ifp->if_lastchange);

	const int bound = curlwp_bind();
	int s = pserialize_read_enter();
	IFADDR_READER_FOREACH(ifa, ifp) {
		ifa_acquire(ifa, &psref);
		pserialize_read_exit(s);

		pfctlinput(PRC_IFDOWN, ifa->ifa_addr);

		s = pserialize_read_enter();
		ifa_release(ifa, &psref);
	}
	pserialize_read_exit(s);
	curlwp_bindx(bound);

	IFQ_PURGE(&ifp->if_snd);
#if NCARP > 0
	if (ifp->if_carp)
		carp_carpdev_state(ifp);
#endif
	rt_ifmsg(ifp);
	DOMAIN_FOREACH(dp) {
		if (dp->dom_if_down)
			dp->dom_if_down(ifp);
	}
}

static void
if_down_deactivated(struct ifnet *ifp)
{

	KASSERT(if_is_deactivated(ifp));
	_if_down(ifp);
}

void
if_down_locked(struct ifnet *ifp)
{

	KASSERT(IFNET_LOCKED(ifp));
	_if_down(ifp);
}

/*
 * Mark an interface down and notify protocols of
 * the transition.
 * NOTE: must be called at splsoftnet or equivalent.
 */
void
if_down(struct ifnet *ifp)
{

	IFNET_LOCK(ifp);
	if_down_locked(ifp);
	IFNET_UNLOCK(ifp);
}

/*
 * Must be called with holding if_ioctl_lock.
 */
static void
if_up_locked(struct ifnet *ifp)
{
#ifdef notyet
	struct ifaddr *ifa;
#endif
	struct domain *dp;

	KASSERT(IFNET_LOCKED(ifp));

	KASSERT(!if_is_deactivated(ifp));
	ifp->if_flags |= IFF_UP;
	nanotime(&ifp->if_lastchange);
#ifdef notyet
	/* this has no effect on IP, and will kill all ISO connections XXX */
	IFADDR_READER_FOREACH(ifa, ifp)
		pfctlinput(PRC_IFUP, ifa->ifa_addr);
#endif
#if NCARP > 0
	if (ifp->if_carp)
		carp_carpdev_state(ifp);
#endif
	rt_ifmsg(ifp);
	DOMAIN_FOREACH(dp) {
		if (dp->dom_if_up)
			dp->dom_if_up(ifp);
	}
}

/*
 * Handle interface slowtimo timer routine.  Called
 * from softclock, we decrement timer (if set) and
 * call the appropriate interface routine on expiration.
 */
static bool
if_slowtimo_countdown(struct ifnet *ifp)
{
	bool fire = false;
	const int s = splnet();

	KERNEL_LOCK(1, NULL);
	if (ifp->if_timer != 0 && --ifp->if_timer == 0)
		fire = true;
	KERNEL_UNLOCK_ONE(NULL);
	splx(s);

	return fire;
}

static void
if_slowtimo_intr(void *arg)
{
	struct ifnet *ifp = arg;
	struct if_slowtimo_data *isd = ifp->if_slowtimo_data;

	mutex_enter(&isd->isd_lock);
	if (!isd->isd_dying) {
		if (isd->isd_trigger || if_slowtimo_countdown(ifp)) {
			if (!isd->isd_queued) {
				isd->isd_queued = true;
				workqueue_enqueue(if_slowtimo_wq,
				    &isd->isd_work, NULL);
			}
		} else
			callout_schedule(&isd->isd_ch, hz / IFNET_SLOWHZ);
	}
	mutex_exit(&isd->isd_lock);
}

static void
if_slowtimo_work(struct work *work, void *arg)
{
	struct if_slowtimo_data *isd =
	    container_of(work, struct if_slowtimo_data, isd_work);
	struct ifnet *ifp = isd->isd_ifp;
	const int s = splnet();

	KERNEL_LOCK(1, NULL);
	(*ifp->if_slowtimo)(ifp);
	KERNEL_UNLOCK_ONE(NULL);
	splx(s);

	mutex_enter(&isd->isd_lock);
	if (isd->isd_trigger) {
		isd->isd_trigger = false;
		printf("%s: watchdog triggered\n", ifp->if_xname);
	}
	isd->isd_queued = false;
	if (!isd->isd_dying)
		callout_schedule(&isd->isd_ch, hz / IFNET_SLOWHZ);
	mutex_exit(&isd->isd_lock);
}

static int
sysctl_if_watchdog(SYSCTLFN_ARGS)
{
	struct sysctlnode node = *rnode;
	struct ifnet *ifp = node.sysctl_data;
	struct if_slowtimo_data *isd = ifp->if_slowtimo_data;
	int arg = 0;
	int error;

	node.sysctl_data = &arg;
	error = sysctl_lookup(SYSCTLFN_CALL(&node));
	if (error || newp == NULL)
		return error;
	if (arg) {
		mutex_enter(&isd->isd_lock);
		KASSERT(!isd->isd_dying);
		isd->isd_trigger = true;
		callout_schedule(&isd->isd_ch, 0);
		mutex_exit(&isd->isd_lock);
	}

	return 0;
}

static void
sysctl_watchdog_setup(struct ifnet *ifp)
{
	struct sysctllog **clog = &ifp->if_sysctl_log;
	const struct sysctlnode *rnode;

	if (sysctl_createv(clog, 0, NULL, &rnode,
		CTLFLAG_PERMANENT, CTLTYPE_NODE, "interfaces",
		SYSCTL_DESCR("Per-interface controls"),
		NULL, 0, NULL, 0,
		CTL_NET, CTL_CREATE, CTL_EOL) != 0)
		goto bad;
	if (sysctl_createv(clog, 0, &rnode, &rnode,
		CTLFLAG_PERMANENT, CTLTYPE_NODE, ifp->if_xname,
		SYSCTL_DESCR("Interface controls"),
		NULL, 0, NULL, 0,
		CTL_CREATE, CTL_EOL) != 0)
		goto bad;
	if (sysctl_createv(clog, 0, &rnode, &rnode,
		CTLFLAG_PERMANENT, CTLTYPE_NODE, "watchdog",
		SYSCTL_DESCR("Interface watchdog controls"),
		NULL, 0, NULL, 0,
		CTL_CREATE, CTL_EOL) != 0)
		goto bad;
	if (sysctl_createv(clog, 0, &rnode, NULL,
		CTLFLAG_PERMANENT|CTLFLAG_READWRITE, CTLTYPE_INT, "trigger",
		SYSCTL_DESCR("Trigger watchdog timeout"),
		sysctl_if_watchdog, 0, (int *)ifp, 0,
		CTL_CREATE, CTL_EOL) != 0)
		goto bad;

	return;

bad:
	printf("%s: could not attach sysctl watchdog nodes\n", ifp->if_xname);
}

/*
 * Mark an interface up and notify protocols of
 * the transition.
 * NOTE: must be called at splsoftnet or equivalent.
 */
void
if_up(struct ifnet *ifp)
{

	IFNET_LOCK(ifp);
	if_up_locked(ifp);
	IFNET_UNLOCK(ifp);
}

/*
 * Set/clear promiscuous mode on interface ifp based on the truth value
 * of pswitch.  The calls are reference counted so that only the first
 * "on" request actually has an effect, as does the final "off" request.
 * Results are undefined if the "off" and "on" requests are not matched.
 */
int
ifpromisc_locked(struct ifnet *ifp, int pswitch)
{
	int pcount, ret = 0;
	u_short nflags;

	KASSERT(IFNET_LOCKED(ifp));

	pcount = ifp->if_pcount;
	if (pswitch) {
		/*
		 * Allow the device to be "placed" into promiscuous
		 * mode even if it is not configured up.  It will
		 * consult IFF_PROMISC when it is brought up.
		 */
		if (ifp->if_pcount++ != 0)
			goto out;
		nflags = ifp->if_flags | IFF_PROMISC;
	} else {
		if (--ifp->if_pcount > 0)
			goto out;
		nflags = ifp->if_flags & ~IFF_PROMISC;
	}
	ret = if_flags_set(ifp, nflags);
	/* Restore interface state if not successful. */
	if (ret != 0)
		ifp->if_pcount = pcount;

out:
	return ret;
}

int
ifpromisc(struct ifnet *ifp, int pswitch)
{
	int e;

	IFNET_LOCK(ifp);
	e = ifpromisc_locked(ifp, pswitch);
	IFNET_UNLOCK(ifp);

	return e;
}

/*
 * if_ioctl(ifp, cmd, data)
 *
 *	Apply an ioctl command to the interface.  Returns 0 on success,
 *	nonzero errno(3) number on failure.
 *
 *	For SIOCADDMULTI/SIOCDELMULTI, caller need not hold locks -- it
 *	is the driver's responsibility to take any internal locks.
 *	(Kernel logic should generally invoke these only through
 *	if_mcast_op.)
 *
 *	For all other ioctls, caller must hold ifp->if_ioctl_lock,
 *	a.k.a. IFNET_LOCK.  May sleep.
 */
int
if_ioctl(struct ifnet *ifp, u_long cmd, void *data)
{

	switch (cmd) {
	case SIOCADDMULTI:
	case SIOCDELMULTI:
		break;
	default:
		KASSERTMSG(IFNET_LOCKED(ifp), "%s", ifp->if_xname);
	}

	return (*ifp->if_ioctl)(ifp, cmd, data);
}

/*
 * if_init(ifp)
 *
 *	Prepare the hardware underlying ifp to process packets
 *	according to its current configuration.  Returns 0 on success,
 *	nonzero errno(3) number on failure.
 *
 *	May sleep.  Caller must hold ifp->if_ioctl_lock, a.k.a
 *	IFNET_LOCK.
 */
int
if_init(struct ifnet *ifp)
{

	KASSERTMSG(IFNET_LOCKED(ifp), "%s", ifp->if_xname);

	return (*ifp->if_init)(ifp);
}

/*
 * if_stop(ifp, disable)
 *
 *	Stop the hardware underlying ifp from processing packets.
 *
 *	If disable is true, ... XXX(?)
 *
 *	May sleep.  Caller must hold ifp->if_ioctl_lock, a.k.a
 *	IFNET_LOCK.
 */
void
if_stop(struct ifnet *ifp, int disable)
{

	KASSERTMSG(IFNET_LOCKED(ifp), "%s", ifp->if_xname);

	(*ifp->if_stop)(ifp, disable);
}

/*
 * Map interface name to
 * interface structure pointer.
 */
struct ifnet *
ifunit(const char *name)
{
	struct ifnet *ifp;
	const char *cp = name;
	u_int unit = 0;
	u_int i;

	/*
	 * If the entire name is a number, treat it as an ifindex.
	 */
	for (i = 0; i < IFNAMSIZ && *cp >= '0' && *cp <= '9'; i++, cp++)
		unit = unit * 10 + (*cp - '0');

	/*
	 * If the number took all of the name, then it's a valid ifindex.
	 */
	if (i == IFNAMSIZ || (cp != name && *cp == '\0'))
		return if_byindex(unit);

	ifp = NULL;
	const int s = pserialize_read_enter();
	IFNET_READER_FOREACH(ifp) {
		if (if_is_deactivated(ifp))
			continue;
		if (strcmp(ifp->if_xname, name) == 0)
			goto out;
	}
out:
	pserialize_read_exit(s);
	return ifp;
}

/*
 * Get a reference of an ifnet object by an interface name.
 * The returned reference is protected by psref(9). The caller
 * must release a returned reference by if_put after use.
 */
struct ifnet *
if_get(const char *name, struct psref *psref)
{
	struct ifnet *ifp;
	const char *cp = name;
	u_int unit = 0;
	u_int i;

	/*
	 * If the entire name is a number, treat it as an ifindex.
	 */
	for (i = 0; i < IFNAMSIZ && *cp >= '0' && *cp <= '9'; i++, cp++)
		unit = unit * 10 + (*cp - '0');

	/*
	 * If the number took all of the name, then it's a valid ifindex.
	 */
	if (i == IFNAMSIZ || (cp != name && *cp == '\0'))
		return if_get_byindex(unit, psref);

	ifp = NULL;
	const int s = pserialize_read_enter();
	IFNET_READER_FOREACH(ifp) {
		if (if_is_deactivated(ifp))
			continue;
		if (strcmp(ifp->if_xname, name) == 0) {
			PSREF_DEBUG_FILL_RETURN_ADDRESS(psref);
			psref_acquire(psref, &ifp->if_psref,
			    ifnet_psref_class);
			goto out;
		}
	}
out:
	pserialize_read_exit(s);
	return ifp;
}

/*
 * Release a reference of an ifnet object given by if_get, if_get_byindex
 * or if_get_bylla.
 */
void
if_put(const struct ifnet *ifp, struct psref *psref)
{

	if (ifp == NULL)
		return;

	psref_release(psref, &ifp->if_psref, ifnet_psref_class);
}

/*
 * Return ifp having idx. Return NULL if not found.  Normally if_byindex
 * should be used.
 */
ifnet_t *
_if_byindex(u_int idx)
{

	return (__predict_true(idx < if_indexlim)) ? ifindex2ifnet[idx] : NULL;
}

/*
 * Return ifp having idx. Return NULL if not found or the found ifp is
 * already deactivated.
 */
ifnet_t *
if_byindex(u_int idx)
{
	ifnet_t *ifp;

	ifp = _if_byindex(idx);
	if (ifp != NULL && if_is_deactivated(ifp))
		ifp = NULL;
	return ifp;
}

/*
 * Get a reference of an ifnet object by an interface index.
 * The returned reference is protected by psref(9). The caller
 * must release a returned reference by if_put after use.
 */
ifnet_t *
if_get_byindex(u_int idx, struct psref *psref)
{
	ifnet_t *ifp;

	const int s = pserialize_read_enter();
	ifp = if_byindex(idx);
	if (__predict_true(ifp != NULL)) {
		PSREF_DEBUG_FILL_RETURN_ADDRESS(psref);
		psref_acquire(psref, &ifp->if_psref, ifnet_psref_class);
	}
	pserialize_read_exit(s);

	return ifp;
}

ifnet_t *
if_get_bylla(const void *lla, unsigned char lla_len, struct psref *psref)
{
	ifnet_t *ifp;

	const int s = pserialize_read_enter();
	IFNET_READER_FOREACH(ifp) {
		if (if_is_deactivated(ifp))
			continue;
		if (ifp->if_addrlen != lla_len)
			continue;
		if (memcmp(lla, CLLADDR(ifp->if_sadl), lla_len) == 0) {
			psref_acquire(psref, &ifp->if_psref,
			    ifnet_psref_class);
			break;
		}
	}
	pserialize_read_exit(s);

	return ifp;
}

/*
 * Note that it's safe only if the passed ifp is guaranteed to not be freed,
 * for example using pserialize or the ifp is already held or some other
 * object is held which guarantes the ifp to not be freed indirectly.
 */
void
if_acquire(struct ifnet *ifp, struct psref *psref)
{

	KASSERT(ifp->if_index != 0);
	psref_acquire(psref, &ifp->if_psref, ifnet_psref_class);
}

bool
if_held(struct ifnet *ifp)
{

	return psref_held(&ifp->if_psref, ifnet_psref_class);
}

/*
 * Some tunnel interfaces can nest, e.g. IPv4 over IPv4 gif(4) tunnel over
 * IPv4. Check the tunnel nesting count.
 * Return > 0, if tunnel nesting count is more than limit.
 * Return 0, if tunnel nesting count is equal or less than limit.
 */
int
if_tunnel_check_nesting(struct ifnet *ifp, struct mbuf *m, int limit)
{
	struct m_tag *mtag;
	int *count;

	mtag = m_tag_find(m, PACKET_TAG_TUNNEL_INFO);
	if (mtag != NULL) {
		count = (int *)(mtag + 1);
		if (++(*count) > limit) {
			log(LOG_NOTICE,
			    "%s: recursively called too many times(%d)\n",
			    ifp->if_xname, *count);
			return EIO;
		}
	} else {
		mtag = m_tag_get(PACKET_TAG_TUNNEL_INFO, sizeof(*count),
		    M_NOWAIT);
		if (mtag != NULL) {
			m_tag_prepend(m, mtag);
			count = (int *)(mtag + 1);
			*count = 0;
		} else {
			log(LOG_DEBUG, "%s: m_tag_get() failed, "
			    "recursion calls are not prevented.\n",
			    ifp->if_xname);
		}
	}

	return 0;
}

static void
if_tunnel_ro_init_pc(void *p, void *arg __unused, struct cpu_info *ci __unused)
{
	struct tunnel_ro *tro = p;

	tro->tr_ro = kmem_zalloc(sizeof(*tro->tr_ro), KM_SLEEP);
	tro->tr_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE);
}

static void
if_tunnel_ro_fini_pc(void *p, void *arg __unused, struct cpu_info *ci __unused)
{
	struct tunnel_ro *tro = p;

	rtcache_free(tro->tr_ro);
	kmem_free(tro->tr_ro, sizeof(*tro->tr_ro));

	mutex_obj_free(tro->tr_lock);
}

percpu_t *
if_tunnel_alloc_ro_percpu(void)
{

	return percpu_create(sizeof(struct tunnel_ro),
	    if_tunnel_ro_init_pc, if_tunnel_ro_fini_pc, NULL);
}

void
if_tunnel_free_ro_percpu(percpu_t *ro_percpu)
{

	percpu_free(ro_percpu, sizeof(struct tunnel_ro));
}


static void
if_tunnel_rtcache_free_pc(void *p, void *arg __unused,
    struct cpu_info *ci __unused)
{
	struct tunnel_ro *tro = p;

	mutex_enter(tro->tr_lock);
	rtcache_free(tro->tr_ro);
	mutex_exit(tro->tr_lock);
}

void if_tunnel_ro_percpu_rtcache_free(percpu_t *ro_percpu)
{

	percpu_foreach(ro_percpu, if_tunnel_rtcache_free_pc, NULL);
}

void
if_export_if_data(ifnet_t * const ifp, struct if_data *ifi, bool zero_stats)
{

	/* Collect the volatile stats first; this zeros *ifi. */
	if_stats_to_if_data(ifp, ifi, zero_stats);

	ifi->ifi_type = ifp->if_type;
	ifi->ifi_addrlen = ifp->if_addrlen;
	ifi->ifi_hdrlen = ifp->if_hdrlen;
	ifi->ifi_link_state = ifp->if_link_state;
	ifi->ifi_mtu = ifp->if_mtu;
	ifi->ifi_metric = ifp->if_metric;
	ifi->ifi_baudrate = ifp->if_baudrate;
	ifi->ifi_lastchange = ifp->if_lastchange;
}

/* common */
int
ifioctl_common(struct ifnet *ifp, u_long cmd, void *data)
{
	struct ifreq *ifr;
	struct ifcapreq *ifcr;
	struct ifdatareq *ifdr;
	unsigned short flags;
	char *descr;
	int error;

	switch (cmd) {
	case SIOCSIFCAP:
		ifcr = data;
		if ((ifcr->ifcr_capenable & ~ifp->if_capabilities) != 0)
			return EINVAL;

		if (ifcr->ifcr_capenable == ifp->if_capenable)
			return 0;

		ifp->if_capenable = ifcr->ifcr_capenable;

		/* Pre-compute the checksum flags mask. */
		ifp->if_csum_flags_tx = 0;
		ifp->if_csum_flags_rx = 0;
		if (ifp->if_capenable & IFCAP_CSUM_IPv4_Tx)
			ifp->if_csum_flags_tx |= M_CSUM_IPv4;
		if (ifp->if_capenable & IFCAP_CSUM_IPv4_Rx)
			ifp->if_csum_flags_rx |= M_CSUM_IPv4;

		if (ifp->if_capenable & IFCAP_CSUM_TCPv4_Tx)
			ifp->if_csum_flags_tx |= M_CSUM_TCPv4;
		if (ifp->if_capenable & IFCAP_CSUM_TCPv4_Rx)
			ifp->if_csum_flags_rx |= M_CSUM_TCPv4;

		if (ifp->if_capenable & IFCAP_CSUM_UDPv4_Tx)
			ifp->if_csum_flags_tx |= M_CSUM_UDPv4;
		if (ifp->if_capenable & IFCAP_CSUM_UDPv4_Rx)
			ifp->if_csum_flags_rx |= M_CSUM_UDPv4;

		if (ifp->if_capenable & IFCAP_CSUM_TCPv6_Tx)
			ifp->if_csum_flags_tx |= M_CSUM_TCPv6;
		if (ifp->if_capenable & IFCAP_CSUM_TCPv6_Rx)
			ifp->if_csum_flags_rx |= M_CSUM_TCPv6;

		if (ifp->if_capenable & IFCAP_CSUM_UDPv6_Tx)
			ifp->if_csum_flags_tx |= M_CSUM_UDPv6;
		if (ifp->if_capenable & IFCAP_CSUM_UDPv6_Rx)
			ifp->if_csum_flags_rx |= M_CSUM_UDPv6;

		if (ifp->if_capenable & IFCAP_TSOv4)
			ifp->if_csum_flags_tx |= M_CSUM_TSOv4;
		if (ifp->if_capenable & IFCAP_TSOv6)
			ifp->if_csum_flags_tx |= M_CSUM_TSOv6;

#if NBRIDGE > 0
		if (ifp->if_bridge != NULL)
			bridge_calc_csum_flags(ifp->if_bridge);
#endif

		if (ifp->if_flags & IFF_UP)
			return ENETRESET;
		return 0;
	case SIOCSIFFLAGS:
		ifr = data;
		/*
		 * If if_is_mpsafe(ifp), KERNEL_LOCK isn't held here, but if_up
		 * and if_down aren't MP-safe yet, so we must hold the lock.
		 */
		KERNEL_LOCK_IF_IFP_MPSAFE(ifp);
		if (ifp->if_flags & IFF_UP && (ifr->ifr_flags & IFF_UP) == 0) {
			const int s = splsoftnet();
			if_down_locked(ifp);
			splx(s);
		}
		if (ifr->ifr_flags & IFF_UP && (ifp->if_flags & IFF_UP) == 0) {
			const int s = splsoftnet();
			if_up_locked(ifp);
			splx(s);
		}
		KERNEL_UNLOCK_IF_IFP_MPSAFE(ifp);
		flags = (ifp->if_flags & IFF_CANTCHANGE) |
		    (ifr->ifr_flags &~ IFF_CANTCHANGE);
		if (ifp->if_flags != flags) {
			ifp->if_flags = flags;
			/* Notify that the flags have changed. */
			rt_ifmsg(ifp);
		}
		break;
	case SIOCGIFFLAGS:
		ifr = data;
		ifr->ifr_flags = ifp->if_flags;
		break;

	case SIOCGIFMETRIC:
		ifr = data;
		ifr->ifr_metric = ifp->if_metric;
		break;

	case SIOCGIFMTU:
		ifr = data;
		ifr->ifr_mtu = ifp->if_mtu;
		break;

	case SIOCGIFDLT:
		ifr = data;
		ifr->ifr_dlt = ifp->if_dlt;
		break;

	case SIOCGIFCAP:
		ifcr = data;
		ifcr->ifcr_capabilities = ifp->if_capabilities;
		ifcr->ifcr_capenable = ifp->if_capenable;
		break;

	case SIOCSIFMETRIC:
		ifr = data;
		ifp->if_metric = ifr->ifr_metric;
		break;

	case SIOCGIFDATA:
		ifdr = data;
		if_export_if_data(ifp, &ifdr->ifdr_data, false);
		break;

	case SIOCGIFINDEX:
		ifr = data;
		ifr->ifr_index = ifp->if_index;
		break;

	case SIOCZIFDATA:
		ifdr = data;
		if_export_if_data(ifp, &ifdr->ifdr_data, true);
		getnanotime(&ifp->if_lastchange);
		break;
	case SIOCSIFMTU:
		ifr = data;
		if (ifp->if_mtu == ifr->ifr_mtu)
			break;
		ifp->if_mtu = ifr->ifr_mtu;
		return ENETRESET;
	case SIOCSIFDESCR:
		error = kauth_authorize_network(kauth_cred_get(),
		    KAUTH_NETWORK_INTERFACE,
		    KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp, KAUTH_ARG(cmd),
		    NULL);
		if (error)
			return error;

		ifr = data;

		if (ifr->ifr_buflen > IFDESCRSIZE)
			return ENAMETOOLONG;

		if (ifr->ifr_buf == NULL || ifr->ifr_buflen == 0) {
			/* unset description */
			descr = NULL;
		} else {
			descr = kmem_zalloc(IFDESCRSIZE, KM_SLEEP);
			/*
			 * copy (IFDESCRSIZE - 1) bytes to ensure
			 * terminating nul
			 */
			error = copyin(ifr->ifr_buf, descr, IFDESCRSIZE - 1);
			if (error) {
				kmem_free(descr, IFDESCRSIZE);
				return error;
			}
		}

		if (ifp->if_description != NULL)
			kmem_free(ifp->if_description, IFDESCRSIZE);

		ifp->if_description = descr;
		break;

	case SIOCGIFDESCR:
		ifr = data;
		descr = ifp->if_description;

		if (descr == NULL)
			return ENOMSG;

		if (ifr->ifr_buflen < IFDESCRSIZE)
			return EINVAL;

		error = copyout(descr, ifr->ifr_buf, IFDESCRSIZE);
		if (error)
			return error;
		break;

	default:
		return ENOTTY;
	}
	return 0;
}

int
ifaddrpref_ioctl(struct socket *so, u_long cmd, void *data, struct ifnet *ifp)
{
	struct if_addrprefreq *ifap = (struct if_addrprefreq *)data;
	struct ifaddr *ifa;
	const struct sockaddr *any, *sa;
	union {
		struct sockaddr sa;
		struct sockaddr_storage ss;
	} u, v;
	int s, error = 0;

	switch (cmd) {
	case SIOCSIFADDRPREF:
		error = kauth_authorize_network(kauth_cred_get(),
		    KAUTH_NETWORK_INTERFACE,
		    KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp, KAUTH_ARG(cmd),
		    NULL);
		if (error)
			return error;
		break;
	case SIOCGIFADDRPREF:
		break;
	default:
		return EOPNOTSUPP;
	}

	/* sanity checks */
	if (data == NULL || ifp == NULL) {
		panic("invalid argument to %s", __func__);
		/*NOTREACHED*/
	}

	/* address must be specified on ADD and DELETE */
	sa = sstocsa(&ifap->ifap_addr);
	if (sa->sa_family != sofamily(so))
		return EINVAL;
	if ((any = sockaddr_any(sa)) == NULL || sa->sa_len != any->sa_len)
		return EINVAL;

	sockaddr_externalize(&v.sa, sizeof(v.ss), sa);

	s = pserialize_read_enter();
	IFADDR_READER_FOREACH(ifa, ifp) {
		if (ifa->ifa_addr->sa_family != sa->sa_family)
			continue;
		sockaddr_externalize(&u.sa, sizeof(u.ss), ifa->ifa_addr);
		if (sockaddr_cmp(&u.sa, &v.sa) == 0)
			break;
	}
	if (ifa == NULL) {
		error = EADDRNOTAVAIL;
		goto out;
	}

	switch (cmd) {
	case SIOCSIFADDRPREF:
		ifa->ifa_preference = ifap->ifap_preference;
		goto out;
	case SIOCGIFADDRPREF:
		/* fill in the if_laddrreq structure */
		(void)sockaddr_copy(sstosa(&ifap->ifap_addr),
		    sizeof(ifap->ifap_addr), ifa->ifa_addr);
		ifap->ifap_preference = ifa->ifa_preference;
		goto out;
	default:
		error = EOPNOTSUPP;
	}
out:
	pserialize_read_exit(s);
	return error;
}

/*
 * Interface ioctls.
 */
static int
doifioctl(struct socket *so, u_long cmd, void *data, struct lwp *l)
{
	struct ifnet *ifp;
	struct ifreq *ifr;
	int error = 0;
	u_long ocmd = cmd;
	u_short oif_flags;
	struct ifreq ifrb;
	struct oifreq *oifr = NULL;
	int r;
	struct psref psref;
	bool do_if43_post = false;
	bool do_ifm80_post = false;

	switch (cmd) {
	case SIOCGIFCONF:
		return ifconf(cmd, data);
	case SIOCINITIFADDR:
		return EPERM;
	default:
		MODULE_HOOK_CALL(uipc_syscalls_40_hook, (cmd, data), enosys(),
		    error);
		if (error != ENOSYS)
			return error;
		MODULE_HOOK_CALL(uipc_syscalls_50_hook, (l, cmd, data),
		    enosys(), error);
		if (error != ENOSYS)
			return error;
		error = 0;
		break;
	}

	ifr = data;
	/* Pre-conversion */
	MODULE_HOOK_CALL(if_cvtcmd_43_hook, (&cmd, ocmd), enosys(), error);
	if (cmd != ocmd) {
		oifr = data;
		data = ifr = &ifrb;
		IFREQO2N_43(oifr, ifr);
		do_if43_post = true;
	}
	MODULE_HOOK_CALL(ifmedia_80_pre_hook, (ifr, &cmd, &do_ifm80_post),
	    enosys(), error);

	switch (cmd) {
	case SIOCIFCREATE:
	case SIOCIFDESTROY: {
		const int bound = curlwp_bind();
		if (l != NULL) {
			ifp = if_get(ifr->ifr_name, &psref);
			error = kauth_authorize_network(l->l_cred,
			    KAUTH_NETWORK_INTERFACE,
			    KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp,
			    KAUTH_ARG(cmd), NULL);
			if (ifp != NULL)
				if_put(ifp, &psref);
			if (error != 0) {
				curlwp_bindx(bound);
				return error;
			}
		}
		KERNEL_LOCK_UNLESS_NET_MPSAFE();
		mutex_enter(&if_clone_mtx);
		r = (cmd == SIOCIFCREATE) ?
			if_clone_create(ifr->ifr_name) :
			if_clone_destroy(ifr->ifr_name);
		mutex_exit(&if_clone_mtx);
		KERNEL_UNLOCK_UNLESS_NET_MPSAFE();
		curlwp_bindx(bound);
		return r;
	    }
	case SIOCIFGCLONERS: {
		struct if_clonereq *req = (struct if_clonereq *)data;
		return if_clone_list(req->ifcr_count, req->ifcr_buffer,
		    &req->ifcr_total);
	    }
	}

	if ((cmd & IOC_IN) == 0 || IOCPARM_LEN(cmd) < sizeof(ifr->ifr_name))
		return EINVAL;

	const int bound = curlwp_bind();
	ifp = if_get(ifr->ifr_name, &psref);
	if (ifp == NULL) {
		curlwp_bindx(bound);
		return ENXIO;
	}

	switch (cmd) {
	case SIOCALIFADDR:
	case SIOCDLIFADDR:
	case SIOCSIFADDRPREF:
	case SIOCSIFFLAGS:
	case SIOCSIFCAP:
	case SIOCSIFMETRIC:
	case SIOCZIFDATA:
	case SIOCSIFMTU:
	case SIOCSIFPHYADDR:
	case SIOCDIFPHYADDR:
#ifdef INET6
	case SIOCSIFPHYADDR_IN6:
#endif
	case SIOCSLIFPHYADDR:
	case SIOCADDMULTI:
	case SIOCDELMULTI:
	case SIOCSETHERCAP:
	case SIOCSIFMEDIA:
	case SIOCSDRVSPEC:
	case SIOCG80211:
	case SIOCS80211:
	case SIOCS80211NWID:
	case SIOCS80211NWKEY:
	case SIOCS80211POWER:
	case SIOCS80211BSSID:
	case SIOCS80211CHANNEL:
	case SIOCSLINKSTR:
		if (l != NULL) {
			error = kauth_authorize_network(l->l_cred,
			    KAUTH_NETWORK_INTERFACE,
			    KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp,
			    KAUTH_ARG(cmd), NULL);
			if (error != 0)
				goto out;
		}
	}

	oif_flags = ifp->if_flags;

	KERNEL_LOCK_UNLESS_IFP_MPSAFE(ifp);
	IFNET_LOCK(ifp);

	error = if_ioctl(ifp, cmd, data);
	if (error != ENOTTY)
		;
	else if (so->so_proto == NULL)
		error = EOPNOTSUPP;
	else {
		KERNEL_LOCK_IF_IFP_MPSAFE(ifp);
		MODULE_HOOK_CALL(if_ifioctl_43_hook,
			     (so, ocmd, cmd, data, l), enosys(), error);
		if (error == ENOSYS)
			error = (*so->so_proto->pr_usrreqs->pr_ioctl)(so,
			    cmd, data, ifp);
		KERNEL_UNLOCK_IF_IFP_MPSAFE(ifp);
	}

	if (((oif_flags ^ ifp->if_flags) & IFF_UP) != 0) {
		if ((ifp->if_flags & IFF_UP) != 0) {
			const int s = splsoftnet();
			if_up_locked(ifp);
			splx(s);
		}
	}

	/* Post-conversion */
	if (do_ifm80_post && (error == 0))
		MODULE_HOOK_CALL(ifmedia_80_post_hook, (ifr, cmd),
		    enosys(), error);
	if (do_if43_post)
		IFREQN2O_43(oifr, ifr);

	IFNET_UNLOCK(ifp);
	KERNEL_UNLOCK_UNLESS_IFP_MPSAFE(ifp);
out:
	if_put(ifp, &psref);
	curlwp_bindx(bound);
	return error;
}

/*
 * Return interface configuration
 * of system.  List may be used
 * in later ioctl's (above) to get
 * other information.
 *
 * Each record is a struct ifreq.  Before the addition of
 * sockaddr_storage, the API rule was that sockaddr flavors that did
 * not fit would extend beyond the struct ifreq, with the next struct
 * ifreq starting sa_len beyond the struct sockaddr.  Because the
 * union in struct ifreq includes struct sockaddr_storage, every kind
 * of sockaddr must fit.  Thus, there are no longer any overlength
 * records.
 *
 * Records are added to the user buffer if they fit, and ifc_len is
 * adjusted to the length that was written.  Thus, the user is only
 * assured of getting the complete list if ifc_len on return is at
 * least sizeof(struct ifreq) less than it was on entry.
 *
 * If the user buffer pointer is NULL, this routine copies no data and
 * returns the amount of space that would be needed.
 *
 * Invariants:
 * ifrp points to the next part of the user's buffer to be used.  If
 * ifrp != NULL, space holds the number of bytes remaining that we may
 * write at ifrp.  Otherwise, space holds the number of bytes that
 * would have been written had there been adequate space.
 */
/*ARGSUSED*/
static int
ifconf(u_long cmd, void *data)
{
	struct ifconf *ifc = (struct ifconf *)data;
	struct ifnet *ifp;
	struct ifaddr *ifa;
	struct ifreq ifr, *ifrp = NULL;
	int space = 0, error = 0;
	const int sz = (int)sizeof(struct ifreq);
	const bool docopy = ifc->ifc_req != NULL;
	struct psref psref;

	if (docopy) {
		if (ifc->ifc_len < 0)
			return EINVAL;

		space = ifc->ifc_len;
		ifrp = ifc->ifc_req;
	}
	memset(&ifr, 0, sizeof(ifr));

	const int bound = curlwp_bind();
	int s = pserialize_read_enter();
	IFNET_READER_FOREACH(ifp) {
		psref_acquire(&psref, &ifp->if_psref, ifnet_psref_class);
		pserialize_read_exit(s);

		(void)strncpy(ifr.ifr_name, ifp->if_xname,
		    sizeof(ifr.ifr_name));
		if (ifr.ifr_name[sizeof(ifr.ifr_name) - 1] != '\0') {
			error = ENAMETOOLONG;
			goto release_exit;
		}
		if (IFADDR_READER_EMPTY(ifp)) {
			/* Interface with no addresses - send zero sockaddr. */
			memset(&ifr.ifr_addr, 0, sizeof(ifr.ifr_addr));
			if (!docopy) {
				space += sz;
				goto next;
			}
			if (space >= sz) {
				error = copyout(&ifr, ifrp, sz);
				if (error != 0)
					goto release_exit;
				ifrp++;
				space -= sz;
			}
		}

		s = pserialize_read_enter();
		IFADDR_READER_FOREACH(ifa, ifp) {
			struct sockaddr *sa = ifa->ifa_addr;
			/* all sockaddrs must fit in sockaddr_storage */
			KASSERT(sa->sa_len <= sizeof(ifr.ifr_ifru));

			if (!docopy) {
				space += sz;
				continue;
			}
			memcpy(&ifr.ifr_space, sa, sa->sa_len);
			pserialize_read_exit(s);

			if (space >= sz) {
				error = copyout(&ifr, ifrp, sz);
				if (error != 0)
					goto release_exit;
				ifrp++; space -= sz;
			}
			s = pserialize_read_enter();
		}
		pserialize_read_exit(s);

next:
		s = pserialize_read_enter();
		psref_release(&psref, &ifp->if_psref, ifnet_psref_class);
	}
	pserialize_read_exit(s);
	curlwp_bindx(bound);

	if (docopy) {
		KASSERT(0 <= space && space <= ifc->ifc_len);
		ifc->ifc_len -= space;
	} else {
		KASSERT(space >= 0);
		ifc->ifc_len = space;
	}
	return 0;

release_exit:
	psref_release(&psref, &ifp->if_psref, ifnet_psref_class);
	curlwp_bindx(bound);
	return error;
}

int
ifreq_setaddr(u_long cmd, struct ifreq *ifr, const struct sockaddr *sa)
{
	uint8_t len = sizeof(ifr->ifr_ifru.ifru_space);
	struct ifreq ifrb;
	struct oifreq *oifr = NULL;
	u_long ocmd = cmd;
	int hook;

	MODULE_HOOK_CALL(if_cvtcmd_43_hook, (&cmd, ocmd), enosys(), hook);
	if (hook != ENOSYS) {
		if (cmd != ocmd) {
			oifr = (struct oifreq *)(void *)ifr;
			ifr = &ifrb;
			IFREQO2N_43(oifr, ifr);
				len = sizeof(oifr->ifr_addr);
		}
	}

	if (len < sa->sa_len)
		return EFBIG;

	memset(&ifr->ifr_addr, 0, len);
	sockaddr_copy(&ifr->ifr_addr, len, sa);

	if (cmd != ocmd)
		IFREQN2O_43(oifr, ifr);
	return 0;
}

/*
 * wrapper function for the drivers which doesn't have if_transmit().
 */
static int
if_transmit(struct ifnet *ifp, struct mbuf *m)
{
	int error;
	size_t pktlen = m->m_pkthdr.len;
	bool mcast = (m->m_flags & M_MCAST) != 0;

	const int s = splnet();

	IFQ_ENQUEUE(&ifp->if_snd, m, error);
	if (error != 0) {
		/* mbuf is already freed */
		goto out;
	}

	net_stat_ref_t nsr = IF_STAT_GETREF(ifp);
	if_statadd_ref(ifp, nsr, if_obytes, pktlen);
	if (mcast)
		if_statinc_ref(ifp, nsr, if_omcasts);
	IF_STAT_PUTREF(ifp);

	if ((ifp->if_flags & IFF_OACTIVE) == 0)
		if_start_lock(ifp);
out:
	splx(s);

	return error;
}

int
if_transmit_lock(struct ifnet *ifp, struct mbuf *m)
{
	int error;

	kmsan_check_mbuf(m);

#ifdef ALTQ
	KERNEL_LOCK(1, NULL);
	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
		error = if_transmit(ifp, m);
		KERNEL_UNLOCK_ONE(NULL);
	} else {
		KERNEL_UNLOCK_ONE(NULL);
		error = (*ifp->if_transmit)(ifp, m);
		/* mbuf is already freed */
	}
#else /* !ALTQ */
	error = (*ifp->if_transmit)(ifp, m);
	/* mbuf is already freed */
#endif /* !ALTQ */

	return error;
}

/*
 * Queue message on interface, and start output if interface
 * not yet active.
 */
int
ifq_enqueue(struct ifnet *ifp, struct mbuf *m)
{

	return if_transmit_lock(ifp, m);
}

/*
 * Queue message on interface, possibly using a second fast queue
 */
int
ifq_enqueue2(struct ifnet *ifp, struct ifqueue *ifq, struct mbuf *m)
{
	int error = 0;

	if (ifq != NULL
#ifdef ALTQ
	    && ALTQ_IS_ENABLED(&ifp->if_snd) == 0
#endif
	    ) {
		if (IF_QFULL(ifq)) {
			IF_DROP(&ifp->if_snd);
			m_freem(m);
			if (error == 0)
				error = ENOBUFS;
		} else
			IF_ENQUEUE(ifq, m);
	} else
		IFQ_ENQUEUE(&ifp->if_snd, m, error);
	if (error != 0) {
		if_statinc(ifp, if_oerrors);
		return error;
	}
	return 0;
}

int
if_addr_init(ifnet_t *ifp, struct ifaddr *ifa, const bool src)
{
	int rc;

	KASSERT(IFNET_LOCKED(ifp));
	if (ifp->if_initaddr != NULL)
		rc = (*ifp->if_initaddr)(ifp, ifa, src);
	else if (src || (rc = if_ioctl(ifp, SIOCSIFDSTADDR, ifa)) == ENOTTY)
		rc = if_ioctl(ifp, SIOCINITIFADDR, ifa);

	return rc;
}

int
if_do_dad(struct ifnet *ifp)
{
	if ((ifp->if_flags & IFF_LOOPBACK) != 0)
		return 0;

	switch (ifp->if_type) {
	case IFT_FAITH:
		/*
		 * These interfaces do not have the IFF_LOOPBACK flag,
		 * but loop packets back.  We do not have to do DAD on such
		 * interfaces.  We should even omit it, because loop-backed
		 * responses would confuse the DAD procedure.
		 */
		return 0;
	default:
		/*
		 * Our DAD routine requires the interface up and running.
		 * However, some interfaces can be up before the RUNNING
		 * status.  Additionally, users may try to assign addresses
		 * before the interface becomes up (or running).
		 * We simply skip DAD in such a case as a work around.
		 * XXX: we should rather mark "tentative" on such addresses,
		 * and do DAD after the interface becomes ready.
		 */
		if ((ifp->if_flags & (IFF_UP | IFF_RUNNING)) !=
		    (IFF_UP | IFF_RUNNING))
			return 0;

		return 1;
	}
}

/*
 * if_flags_set(ifp, flags)
 *
 *	Ask ifp to change ifp->if_flags to flags, as if with the
 *	SIOCSIFFLAGS ioctl command.
 *
 *	May sleep.  Caller must hold ifp->if_ioctl_lock, a.k.a
 *	IFNET_LOCK.
 */
int
if_flags_set(ifnet_t *ifp, const u_short flags)
{
	int rc;

	KASSERT(IFNET_LOCKED(ifp));

	if (ifp->if_setflags != NULL)
		rc = (*ifp->if_setflags)(ifp, flags);
	else {
		u_short cantflags, chgdflags;
		struct ifreq ifr;

		chgdflags = ifp->if_flags ^ flags;
		cantflags = chgdflags & IFF_CANTCHANGE;

		if (cantflags != 0)
			ifp->if_flags ^= cantflags;

		/*
		 * Traditionally, we do not call if_ioctl after
		 * setting/clearing only IFF_PROMISC if the interface
		 * isn't IFF_UP.  Uphold that tradition.
		 */
		if (chgdflags == IFF_PROMISC && (ifp->if_flags & IFF_UP) == 0)
			return 0;

		memset(&ifr, 0, sizeof(ifr));

		ifr.ifr_flags = flags & ~IFF_CANTCHANGE;
		rc = if_ioctl(ifp, SIOCSIFFLAGS, &ifr);

		if (rc != 0 && cantflags != 0)
			ifp->if_flags ^= cantflags;
	}

	return rc;
}

/*
 * if_mcast_op(ifp, cmd, sa)
 *
 *	Apply a multicast command, SIOCADDMULTI/SIOCDELMULTI, to the
 *	interface.  Returns 0 on success, nonzero errno(3) number on
 *	failure.
 *
 *	May sleep.
 *
 *	Use this, not if_ioctl, for the multicast commands.
 */
int
if_mcast_op(ifnet_t *ifp, const unsigned long cmd, const struct sockaddr *sa)
{
	int rc;
	struct ifreq ifr;

	switch (cmd) {
	case SIOCADDMULTI:
	case SIOCDELMULTI:
		break;
	default:
		panic("invalid ifnet multicast command: 0x%lx", cmd);
	}

	ifreq_setaddr(cmd, &ifr, sa);
	rc = if_ioctl(ifp, cmd, &ifr);

	return rc;
}

static void
sysctl_sndq_setup(struct sysctllog **clog, const char *ifname,
    struct ifaltq *ifq)
{
	const struct sysctlnode *cnode, *rnode;

	if (sysctl_createv(clog, 0, NULL, &rnode,
		       CTLFLAG_PERMANENT,
		       CTLTYPE_NODE, "interfaces",
		       SYSCTL_DESCR("Per-interface controls"),
		       NULL, 0, NULL, 0,
		       CTL_NET, CTL_CREATE, CTL_EOL) != 0)
		goto bad;

	if (sysctl_createv(clog, 0, &rnode, &rnode,
		       CTLFLAG_PERMANENT,
		       CTLTYPE_NODE, ifname,
		       SYSCTL_DESCR("Interface controls"),
		       NULL, 0, NULL, 0,
		       CTL_CREATE, CTL_EOL) != 0)
		goto bad;

	if (sysctl_createv(clog, 0, &rnode, &rnode,
		       CTLFLAG_PERMANENT,
		       CTLTYPE_NODE, "sndq",
		       SYSCTL_DESCR("Interface output queue controls"),
		       NULL, 0, NULL, 0,
		       CTL_CREATE, CTL_EOL) != 0)
		goto bad;

	if (sysctl_createv(clog, 0, &rnode, &cnode,
		       CTLFLAG_PERMANENT,
		       CTLTYPE_INT, "len",
		       SYSCTL_DESCR("Current output queue length"),
		       NULL, 0, &ifq->ifq_len, 0,
		       CTL_CREATE, CTL_EOL) != 0)
		goto bad;

	if (sysctl_createv(clog, 0, &rnode, &cnode,
		       CTLFLAG_PERMANENT | CTLFLAG_READWRITE,
		       CTLTYPE_INT, "maxlen",
		       SYSCTL_DESCR("Maximum allowed output queue length"),
		       NULL, 0, &ifq->ifq_maxlen, 0,
		       CTL_CREATE, CTL_EOL) != 0)
		goto bad;

	if (sysctl_createv(clog, 0, &rnode, &cnode,
		       CTLFLAG_PERMANENT,
		       CTLTYPE_QUAD, "drops",
		       SYSCTL_DESCR("Packets dropped due to full output queue"),
		       NULL, 0, &ifq->ifq_drops, 0,
		       CTL_CREATE, CTL_EOL) != 0)
		goto bad;

	return;
bad:
	printf("%s: could not attach sysctl nodes\n", ifname);
	return;
}

static int
if_sdl_sysctl(SYSCTLFN_ARGS)
{
	struct ifnet *ifp;
	const struct sockaddr_dl *sdl;
	struct psref psref;
	int error = 0;

	if (namelen != 1)
		return EINVAL;

	const int bound = curlwp_bind();
	ifp = if_get_byindex(name[0], &psref);
	if (ifp == NULL) {
		error = ENODEV;
		goto out0;
	}

	sdl = ifp->if_sadl;
	if (sdl == NULL) {
		*oldlenp = 0;
		goto out1;
	}

	if (oldp == NULL) {
		*oldlenp = sdl->sdl_alen;
		goto out1;
	}

	if (*oldlenp >= sdl->sdl_alen)
		*oldlenp = sdl->sdl_alen;
	error = sysctl_copyout(l, &sdl->sdl_data[sdl->sdl_nlen],
	    oldp, *oldlenp);
out1:
	if_put(ifp, &psref);
out0:
	curlwp_bindx(bound);
	return error;
}

static void
if_sysctl_setup(struct sysctllog **clog)
{
	const struct sysctlnode *rnode = NULL;

	sysctl_createv(clog, 0, NULL, &rnode,
		       CTLFLAG_PERMANENT,
		       CTLTYPE_NODE, "sdl",
		       SYSCTL_DESCR("Get active link-layer address"),
		       if_sdl_sysctl, 0, NULL, 0,
		       CTL_NET, CTL_CREATE, CTL_EOL);
}