1/*	$NetBSD: if.h,v 1.304 2022/11/25 08:39:32 knakahara Exp $	*/
   2
   3/*-
   4 * Copyright (c) 1999, 2000, 2001 The NetBSD Foundation, Inc.
   5 * All rights reserved.
   6 *
   7 * This code is derived from software contributed to The NetBSD Foundation
   8 * by William Studenmund and Jason R. Thorpe.
   9 *
  10 * Redistribution and use in source and binary forms, with or without
  11 * modification, are permitted provided that the following conditions
  12 * are met:
  13 * 1. Redistributions of source code must retain the above copyright
  14 *    notice, this list of conditions and the following disclaimer.
  15 * 2. Redistributions in binary form must reproduce the above copyright
  16 *    notice, this list of conditions and the following disclaimer in the
  17 *    documentation and/or other materials provided with the distribution.
  18 *
  19 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
  20 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
  21 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
  22 * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
  23 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
  24 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
  25 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
  26 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
  27 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
  28 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
  29 * POSSIBILITY OF SUCH DAMAGE.
  30 */
  31
  32/*
  33 * Copyright (c) 1982, 1986, 1989, 1993
  34 *	The Regents of the University of California.  All rights reserved.
  35 *
  36 * Redistribution and use in source and binary forms, with or without
  37 * modification, are permitted provided that the following conditions
  38 * are met:
  39 * 1. Redistributions of source code must retain the above copyright
  40 *    notice, this list of conditions and the following disclaimer.
  41 * 2. Redistributions in binary form must reproduce the above copyright
  42 *    notice, this list of conditions and the following disclaimer in the
  43 *    documentation and/or other materials provided with the distribution.
  44 * 3. Neither the name of the University nor the names of its contributors
  45 *    may be used to endorse or promote products derived from this software
  46 *    without specific prior written permission.
  47 *
  48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  51 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  58 * SUCH DAMAGE.
  59 *
  60 *	@(#)if.h	8.3 (Berkeley) 2/9/95
  61 */
  62
  63#ifndef _NET_IF_H_
  64#define _NET_IF_H_
  65
  66#if !defined(_KERNEL) && !defined(_STANDALONE)
  67#include <stdbool.h>
  68#endif
  69
  70#include <sys/featuretest.h>
  71
  72/*
  73 * Length of interface external name, including terminating '\0'.
  74 * Note: this is the same size as a generic device's external name.
  75 */
  76#define IF_NAMESIZE 16
  77
  78/*
  79 * Length of interface description, including terminating '\0'.
  80 */
  81#define	IFDESCRSIZE	64
  82
  83#if defined(_NETBSD_SOURCE)
  84
  85#include <sys/socket.h>
  86#include <sys/queue.h>
  87#include <sys/mutex.h>
  88#include <sys/hook.h>
  89
  90#include <net/dlt.h>
  91#include <net/pfil.h>
  92#ifdef _KERNEL
  93#include <net/pktqueue.h>
  94#include <sys/pslist.h>
  95#include <sys/pserialize.h>
  96#include <sys/psref.h>
  97#include <sys/module_hook.h>
  98#endif
  99
 100/*
 101 * Always include ALTQ glue here -- we use the ALTQ interface queue
 102 * structure even when ALTQ is not configured into the kernel so that
 103 * the size of struct ifnet does not changed based on the option.  The
 104 * ALTQ queue structure is API-compatible with the legacy ifqueue.
 105 */
 106#include <altq/if_altq.h>
 107
 108/*
 109 * Structures defining a network interface, providing a packet
 110 * transport mechanism (ala level 0 of the PUP protocols).
 111 *
 112 * Each interface accepts output datagrams of a specified maximum
 113 * length, and provides higher level routines with input datagrams
 114 * received from its medium.
 115 *
 116 * Output occurs when the routine if_output is called, with four parameters:
 117 *	(*ifp->if_output)(ifp, m, dst, rt)
 118 * Here m is the mbuf chain to be sent and dst is the destination address.
 119 * The output routine encapsulates the supplied datagram if necessary,
 120 * and then transmits it on its medium.
 121 *
 122 * On input, each interface unwraps the data received by it, and either
 123 * places it on the input queue of a internetwork datagram routine
 124 * and posts the associated software interrupt, or passes the datagram to a raw
 125 * packet input routine.
 126 *
 127 * Routines exist for locating interfaces by their addresses
 128 * or for locating a interface on a certain network, as well as more general
 129 * routing and gateway routines maintaining information used to locate
 130 * interfaces.  These routines live in the files if.c and route.c
 131 */
 132#include <sys/time.h>
 133
 134#if defined(_KERNEL_OPT)
 135#include "opt_compat_netbsd.h"
 136#include "opt_gateway.h"
 137#endif
 138
 139struct mbuf;
 140struct proc;
 141struct rtentry;
 142struct socket;
 143struct ether_header;
 144struct ifaddr;
 145struct ifnet;
 146struct rt_addrinfo;
 147
 148#define	IFNAMSIZ	IF_NAMESIZE
 149
 150/*
 151 * Structure describing a `cloning' interface.
 152 */
 153struct if_clone {
 154	LIST_ENTRY(if_clone) ifc_list;	/* on list of cloners */
 155	const char *ifc_name;		/* name of device, e.g. `gif' */
 156	size_t ifc_namelen;		/* length of name */
 157
 158	int	(*ifc_create)(struct if_clone *, int);
 159	int	(*ifc_destroy)(struct ifnet *);
 160};
 161
 162#define	IF_CLONE_INITIALIZER(name, create, destroy)			\
 163	{ { NULL, NULL }, name, sizeof(name) - 1, create, destroy }
 164
 165/*
 166 * Structure used to query names of interface cloners.
 167 */
 168struct if_clonereq {
 169	int	ifcr_total;		/* total cloners (out) */
 170	int	ifcr_count;		/* room for this many in user buffer */
 171	char	*ifcr_buffer;		/* buffer for cloner names */
 172};
 173
 174/*
 175 * Structure defining statistics and other data kept regarding a network
 176 * interface.
 177 *
 178 * Only used for exporting data from the interface.
 179 */
 180struct if_data {
 181	/* generic interface information */
 182	u_char	ifi_type;		/* ethernet, tokenring, etc. */
 183	u_char	ifi_addrlen;		/* media address length */
 184	u_char	ifi_hdrlen;		/* media header length */
 185	int	ifi_link_state;		/* current link state */
 186	uint64_t ifi_mtu;		/* maximum transmission unit */
 187	uint64_t ifi_metric;		/* routing metric (external only) */
 188	uint64_t ifi_baudrate;		/* linespeed */
 189	/* volatile statistics */
 190	uint64_t ifi_ipackets;		/* packets received on interface */
 191	uint64_t ifi_ierrors;		/* input errors on interface */
 192	uint64_t ifi_opackets;		/* packets sent on interface */
 193	uint64_t ifi_oerrors;		/* output errors on interface */
 194	uint64_t ifi_collisions;	/* collisions on csma interfaces */
 195	uint64_t ifi_ibytes;		/* total number of octets received */
 196	uint64_t ifi_obytes;		/* total number of octets sent */
 197	uint64_t ifi_imcasts;		/* packets received via multicast */
 198	uint64_t ifi_omcasts;		/* packets sent via multicast */
 199	uint64_t ifi_iqdrops;		/* dropped on input, this interface */
 200	uint64_t ifi_noproto;		/* destined for unsupported protocol */
 201	struct	timespec ifi_lastchange;/* last operational state change */
 202};
 203
 204/*
 205 * Values for if_link_state.
 206 */
 207#define	LINK_STATE_UNKNOWN	0	/* link invalid/unknown */
 208#define	LINK_STATE_DOWN		1	/* link is down */
 209#define	LINK_STATE_UP		2	/* link is up */
 210
 211/*
 212 * Status bit descriptions for the various interface types.
 213 */
 214struct if_status_description {
 215	unsigned char	ifs_type;
 216	unsigned char	ifs_state;
 217	const char	*ifs_string;
 218};
 219
 220#define LINK_STATE_DESC_MATCH(_ifs, _t, _s)				\
 221	(((_ifs)->ifs_type == (_t) || (_ifs)->ifs_type == 0) &&		\
 222	    (_ifs)->ifs_state == (_s))
 223
 224#define LINK_STATE_DESCRIPTIONS {					\
 225	{ IFT_ETHER, LINK_STATE_DOWN, "no carrier" },			\
 226	{ IFT_IEEE80211, LINK_STATE_DOWN, "no network" },		\
 227	{ IFT_PPP, LINK_STATE_DOWN, "no carrier" },			\
 228	{ IFT_CARP, LINK_STATE_DOWN, "backup" },			\
 229	{ IFT_CARP, LINK_STATE_UP, "master" },				\
 230	{ 0, LINK_STATE_UP, "active" },					\
 231	{ 0, LINK_STATE_UNKNOWN, "unknown" },				\
 232	{ 0, LINK_STATE_DOWN, "down" },					\
 233	{ 0, 0, NULL }							\
 234}
 235
 236/*
 237 * Structure defining a queue for a network interface.
 238 */
 239struct ifqueue {
 240	struct		mbuf *ifq_head;
 241	struct		mbuf *ifq_tail;
 242	int		ifq_len;
 243	int		ifq_maxlen;
 244	uint64_t	ifq_drops;
 245	kmutex_t	*ifq_lock;
 246};
 247
 248#ifdef _KERNEL
 249#include <sys/percpu.h>
 250#include <sys/callout.h>
 251#include <sys/rwlock.h>
 252#include <sys/workqueue.h>
 253
 254#endif /* _KERNEL */
 255
 256/*
 257 * Structure defining a queue for a network interface.
 258 *
 259 * (Would like to call this struct ``if'', but C isn't PL/1.)
 260 */
 261TAILQ_HEAD(ifnet_head, ifnet);		/* the actual queue head */
 262
 263struct bridge_softc;
 264struct bridge_iflist;
 265struct callout;
 266struct krwlock;
 267struct if_percpuq;
 268struct if_deferred_start;
 269struct in6_multi;
 270
 271typedef unsigned short if_index_t;
 272
 273/*
 274 * Interface.  Field markings and the corresponding locks:
 275 *
 276 * i:	IFNET_LOCK (a.k.a., if_ioctl_lock)
 277 * q:	ifq_lock (struct ifaltq)
 278 * a:	if_afdata_lock
 279 * 6:	in6_multilock (global lock)
 280 * ::	unlocked, stable
 281 * ?:	unknown, maybe unsafe
 282 *
 283 * Lock order: IFNET_LOCK => in6_multilock => if_afdata_lock => ifq_lock
 284 *   Note that currently if_afdata_lock and ifq_lock aren't held
 285 *   at the same time, but define the order anyway.
 286 *
 287 * Lock order of IFNET_LOCK with other locks:
 288 *     softnet_lock => solock => IFNET_LOCK => ND6_LOCK, in_multilock
 289 */
 290typedef struct ifnet {
 291	void		*if_softc;	/* :: lower-level data for this if */
 292	/* DEPRECATED. Keep it to avoid breaking kvm(3) users */
 293	TAILQ_ENTRY(ifnet)
 294			if_list;	/* i: all struct ifnets are chained */
 295	TAILQ_HEAD(, ifaddr)
 296			if_addrlist;	/* i: linked list of addresses per if */
 297	char		if_xname[IFNAMSIZ];
 298					/* :: external name (name + unit) */
 299	int		if_pcount;	/* i: number of promiscuous listeners */
 300	struct bpf_if	*if_bpf;	/* :: packet filter structure */
 301	if_index_t	if_index;	/* :: numeric abbreviation for this if */
 302	short		if_timer;	/* ?: time 'til if_slowtimo called */
 303	unsigned short	if_flags;	/* i: up/down, broadcast, etc. */
 304	short		if_extflags;	/* :: if_output MP-safe, etc. */
 305	u_char		if_type;	/* :: ethernet, tokenring, etc. */
 306	u_char		if_addrlen;	/* :: media address length */
 307	u_char		if_hdrlen;	/* :: media header length */
 308	/* XXX audit :? fields here. */
 309	int		if_link_state;	/* :? current link state */
 310	uint64_t	if_mtu;		/* :? maximum transmission unit */
 311	uint64_t	if_metric;	/* :? routing metric (external only) */
 312	uint64_t	if_baudrate;	/* :? linespeed */
 313	struct timespec	if_lastchange;	/* :? last operational state change */
 314#ifdef _KERNEL
 315	percpu_t	*if_stats;	/* :: statistics */
 316#else
 317	void		*if_stats;	/* opaque to user-space */
 318#endif /* _KERNEL */
 319	/*
 320	 * Procedure handles.  If you add more of these, don't forget the
 321	 * corresponding NULL stub in if.c.
 322	 */
 323	int		(*if_output)	/* :: output routine (enqueue) */
 324			    (struct ifnet *, struct mbuf *, const struct sockaddr *,
 325			     const struct rtentry *);
 326	void		(*_if_input)	/* :: input routine (from h/w driver) */
 327			    (struct ifnet *, struct mbuf *);
 328	void		(*if_start)	/* :: initiate output routine */
 329			    (struct ifnet *);
 330	int		(*if_transmit)	/* :: output routine, must be MP-safe */
 331			    (struct ifnet *, struct mbuf *);
 332	int		(*if_ioctl)	/* :: ioctl routine */
 333			    (struct ifnet *, u_long, void *);
 334	int		(*if_init)	/* :: init routine */
 335			    (struct ifnet *);
 336	void		(*if_stop)	/* :: stop routine */
 337			    (struct ifnet *, int);
 338	void		(*if_slowtimo)	/* :: timer routine */
 339			    (struct ifnet *);
 340#define	if_watchdog	if_slowtimo
 341	void		(*if_drain)	/* :: routine to release resources */
 342			    (struct ifnet *);
 343	void		(*if_bpf_mtap)	/* :: bpf routine */
 344			    (struct bpf_if *, struct mbuf *, u_int);
 345	struct ifaltq	if_snd;		/* q: output queue (includes altq) */
 346	struct ifaddr	*if_dl;		/* i: identity of this interface. */
 347	const struct sockaddr_dl
 348			*if_sadl;	/* i: pointer to sockaddr_dl of if_dl */
 349	/*
 350	 * May be NULL.  If not NULL, it is the address assigned
 351	 * to the interface by the manufacturer, so it very likely
 352	 * to be unique.  It MUST NOT be deleted.  It is highly
 353	 * suitable for deriving the EUI64 for the interface.
 354	 */
 355	struct ifaddr	*if_hwdl;	/* i: h/w identity */
 356	const uint8_t	*if_broadcastaddr;
 357					/* :: linklevel broadcast bytestring */
 358	struct bridge_softc
 359			*if_bridge;	/* i: bridge glue */
 360	struct bridge_iflist
 361			*if_bridgeif;	/* i: shortcut to interface list entry */
 362	int		if_dlt;		/* :: data link type (<net/dlt.h>) */
 363	pfil_head_t *	if_pfil;	/* :: filtering point */
 364	uint64_t	if_capabilities;
 365					/* i: interface capabilities */
 366	uint64_t	if_capenable;	/* i: capabilities enabled */
 367	union {
 368		void *		carp_s;	/* carp structure (used by !carp ifs) */
 369		struct ifnet	*carp_d;/* ptr to carpdev (used by carp ifs) */
 370	}		if_carp_ptr;	/* ?: */
 371#define if_carp		if_carp_ptr.carp_s
 372#define if_carpdev	if_carp_ptr.carp_d
 373	/*
 374	 * These are pre-computed based on an interfaces enabled
 375	 * capabilities, for speed elsewhere.
 376	 */
 377	int		if_csum_flags_tx;
 378					/* i: M_CSUM_* flags for Tx */
 379	int		if_csum_flags_rx;
 380					/* i: M_CSUM_* flags for Rx */
 381
 382	void		*if_afdata[AF_MAX];
 383					/* a: */
 384	struct mowner	*if_mowner;	/* ?: who owns mbufs for this interface */
 385
 386	void		*if_lagg;	/* :: lagg or agr structure */
 387	void		*if_npf_private;/* ?: associated NPF context */
 388
 389	/*
 390	 * pf specific data, used only when #if NPF > 0.
 391	 */
 392	void		*if_pf_kif;	/* ?: pf interface abstraction */
 393	void		*if_pf_groups;	/* ?: pf interface groups */
 394	/*
 395	 * During an ifnet's lifetime, it has only one if_index, but
 396	 * an if_index is not sufficient to identify an ifnet
 397	 * because during the lifetime of the system, many ifnets may occupy a
 398	 * given if_index.  Let us tell different ifnets at the same
 399	 * if_index apart by their if_index_gen, a unique number that each ifnet
 400	 * is assigned when it if_attach()s.  Now, the kernel can use the
 401	 * pair (if_index, if_index_gen) as a weak reference to an ifnet.
 402	 */
 403	uint64_t	if_index_gen;	/* :: generation number for the ifnet
 404					 * at if_index: if two ifnets' index
 405					 * and generation number are both the
 406					 * same, they are the same ifnet.
 407					 */
 408	struct sysctllog
 409			*if_sysctl_log;	/* :: */
 410	int		(*if_initaddr)  /* :: */
 411			    (struct ifnet *, struct ifaddr *, bool);
 412	int		(*if_setflags)	/* :: */
 413			    (struct ifnet *, const u_short);
 414	kmutex_t	*if_ioctl_lock;	/* :: */
 415	char		*if_description;	/* i: interface description */
 416#ifdef _KERNEL /* XXX kvm(3) */
 417	struct if_slowtimo_data *if_slowtimo_data; /* :: */
 418	struct krwlock	*if_afdata_lock;/* :: */
 419	struct if_percpuq
 420			*if_percpuq;	/* :: we should remove it in the future */
 421	struct work	if_link_work;	/* q: linkage on link state work queue */
 422	uint16_t	if_link_queue;	/* q: masked link state change queue */
 423					/* q: is link state work scheduled? */
 424	bool		if_link_scheduled;
 425	struct pslist_entry
 426			if_pslist_entry;/* i: */
 427	struct psref_target
 428			if_psref;	/* :: */
 429	struct pslist_head
 430			if_addr_pslist;	/* i: */
 431	struct if_deferred_start
 432			*if_deferred_start;
 433					/* :: */
 434	/* XXX should be protocol independent */
 435	LIST_HEAD(, in6_multi)
 436			if_multiaddrs;	/* 6: */
 437	khook_list_t	*if_linkstate_hooks;	/* :: */
 438#endif
 439} ifnet_t;
 440
 441#include <net/if_stats.h>
 442
 443#define	if_name(ifp)	((ifp)->if_xname)
 444
 445#define	IFF_UP		0x0001		/* interface is up */
 446#define	IFF_BROADCAST	0x0002		/* broadcast address valid */
 447#define	IFF_DEBUG	0x0004		/* turn on debugging */
 448#define	IFF_LOOPBACK	0x0008		/* is a loopback net */
 449#define	IFF_POINTOPOINT	0x0010		/* interface is point-to-point link */
 450#if 0
 451/*			0x0020		   was IFF_NOTRAILERS */
 452#else
 453/*
 454 * sys/compat/svr4 is remvoed on 19 Dec 2018.
 455 * And then, IFF_NOTRAILERS itself is removed by if.h:r1.268 on 5 Feb 2019.
 456 */
 457#define	IFF_UNNUMBERED	0x0020		/* explicit unnumbered */
 458#endif
 459#define	IFF_RUNNING	0x0040		/* resources allocated */
 460#define	IFF_NOARP	0x0080		/* no address resolution protocol */
 461#define	IFF_PROMISC	0x0100		/* receive all packets */
 462#define	IFF_ALLMULTI	0x0200		/* receive all multicast packets */
 463#define	IFF_OACTIVE	0x0400		/* transmission in progress */
 464#define	IFF_SIMPLEX	0x0800		/* can't hear own transmissions */
 465#define	IFF_LINK0	0x1000		/* per link layer defined bit */
 466#define	IFF_LINK1	0x2000		/* per link layer defined bit */
 467#define	IFF_LINK2	0x4000		/* per link layer defined bit */
 468#define	IFF_MULTICAST	0x8000		/* supports multicast */
 469
 470#define	IFEF_MPSAFE			__BIT(0)	/* handlers can run in parallel (see below) */
 471
 472/*
 473 * The guidelines for converting an interface to IFEF_MPSAFE are as follows
 474 *
 475 * Enabling IFEF_MPSAFE on an interface suppresses taking KERNEL_LOCK when
 476 * calling the following handlers:
 477 * - if_start
 478 *   - Note that if_transmit is always called without KERNEL_LOCK
 479 * - if_output
 480 * - if_ioctl
 481 * - if_init
 482 * - if_stop
 483 *
 484 * This means that an interface with IFEF_MPSAFE must make the above handlers
 485 * MP-safe or take KERNEL_LOCK by itself inside handlers that aren't MP-safe
 486 * yet.
 487 *
 488 * There are some additional restrictions to access member variables of struct
 489 * ifnet:
 490 * - if_flags
 491 *   - Must be updated with holding IFNET_LOCK
 492 *   - You cannot use the flag in Tx/Rx paths anymore because there is no
 493 *     synchronization on the flag except for IFNET_LOCK
 494 *   - Note that IFNET_LOCK can't be taken in softint because it's known
 495 *     that it causes a deadlock
 496 *     - Some synchronization mechanisms such as pserialize_perform are called
 497 *       with IFNET_LOCK and also require context switches on every CPUs
 498 *       that mean softints finish so trying to take IFNET_LOCK in softint
 499 *       might block on IFNET_LOCK and prevent such synchronization mechanisms
 500 *       from being completed
 501 *     - Currently the deadlock occurs only if NET_MPSAFE is enabled, however,
 502 *       we should deal with the restriction because NET_MPSAFE will be enabled
 503 *       by default in the future
 504 * - if_watchdog and if_timer
 505 *   - The watchdog framework works only for non-IFEF_MPSAFE interfaces
 506 *     that rely on KERNEL_LOCK
 507 *   - Interfaces with IFEF_MPSAFE have to provide its own watchdog mechanism
 508 *     if needed
 509 *     - Keep if_watchdog NULL when calling if_attach
 510 */
 511
 512#ifdef _KERNEL
 513static __inline bool
 514if_is_mpsafe(struct ifnet *ifp)
 515{
 516
 517	return ((ifp->if_extflags & IFEF_MPSAFE) != 0);
 518}
 519
 520static __inline int
 521if_output_lock(struct ifnet *cifp, struct ifnet *ifp, struct mbuf *m,
 522    const struct sockaddr *dst, const struct rtentry *rt)
 523{
 524
 525	if (if_is_mpsafe(cifp)) {
 526		return (*cifp->if_output)(ifp, m, dst, rt);
 527	} else {
 528		int ret;
 529
 530		KERNEL_LOCK(1, NULL);
 531		ret = (*cifp->if_output)(ifp, m, dst, rt);
 532		KERNEL_UNLOCK_ONE(NULL);
 533		return ret;
 534	}
 535}
 536
 537static __inline void
 538if_start_lock(struct ifnet *ifp)
 539{
 540
 541	if (if_is_mpsafe(ifp)) {
 542		(*ifp->if_start)(ifp);
 543	} else {
 544		KERNEL_LOCK(1, NULL);
 545		(*ifp->if_start)(ifp);
 546		KERNEL_UNLOCK_ONE(NULL);
 547	}
 548}
 549
 550#define KERNEL_LOCK_IF_IFP_MPSAFE(ifp)					\
 551	do { if (if_is_mpsafe(ifp)) { KERNEL_LOCK(1, NULL); } } while (0)
 552#define KERNEL_UNLOCK_IF_IFP_MPSAFE(ifp)				\
 553	do { if (if_is_mpsafe(ifp)) { KERNEL_UNLOCK_ONE(NULL); } } while (0)
 554
 555#define KERNEL_LOCK_UNLESS_IFP_MPSAFE(ifp)				\
 556	do { if (!if_is_mpsafe(ifp)) { KERNEL_LOCK(1, NULL); } } while (0)
 557#define KERNEL_UNLOCK_UNLESS_IFP_MPSAFE(ifp)				\
 558	do { if (!if_is_mpsafe(ifp)) { KERNEL_UNLOCK_ONE(NULL); } } while (0)
 559
 560#ifdef _KERNEL_OPT
 561#include "opt_net_mpsafe.h"
 562#endif
 563
 564/* XXX explore a better place to define */
 565#ifdef NET_MPSAFE
 566
 567#define KERNEL_LOCK_UNLESS_NET_MPSAFE()		do { } while (0)
 568#define KERNEL_UNLOCK_UNLESS_NET_MPSAFE()	do { } while (0)
 569
 570#define SOFTNET_LOCK_UNLESS_NET_MPSAFE()	do { } while (0)
 571#define SOFTNET_UNLOCK_UNLESS_NET_MPSAFE()	do { } while (0)
 572
 573#define SOFTNET_LOCK_IF_NET_MPSAFE()					\
 574	do { mutex_enter(softnet_lock); } while (0)
 575#define SOFTNET_UNLOCK_IF_NET_MPSAFE()					\
 576	do { mutex_exit(softnet_lock); } while (0)
 577
 578#else /* NET_MPSAFE */
 579
 580#define KERNEL_LOCK_UNLESS_NET_MPSAFE()					\
 581	do { KERNEL_LOCK(1, NULL); } while (0)
 582#define KERNEL_UNLOCK_UNLESS_NET_MPSAFE()				\
 583	do { KERNEL_UNLOCK_ONE(NULL); } while (0)
 584
 585#define SOFTNET_LOCK_UNLESS_NET_MPSAFE()				\
 586	do { mutex_enter(softnet_lock); } while (0)
 587#define SOFTNET_UNLOCK_UNLESS_NET_MPSAFE()				\
 588	do { mutex_exit(softnet_lock); } while (0)
 589
 590#define SOFTNET_LOCK_IF_NET_MPSAFE()		do { } while (0)
 591#define SOFTNET_UNLOCK_IF_NET_MPSAFE()		do { } while (0)
 592
 593#endif /* NET_MPSAFE */
 594
 595#define SOFTNET_KERNEL_LOCK_UNLESS_NET_MPSAFE()				\
 596	do {								\
 597		SOFTNET_LOCK_UNLESS_NET_MPSAFE();			\
 598		KERNEL_LOCK_UNLESS_NET_MPSAFE();			\
 599	} while (0)
 600
 601#define SOFTNET_KERNEL_UNLOCK_UNLESS_NET_MPSAFE()			\
 602	do {								\
 603		KERNEL_UNLOCK_UNLESS_NET_MPSAFE();			\
 604		SOFTNET_UNLOCK_UNLESS_NET_MPSAFE();			\
 605	} while (0)
 606
 607#endif /* _KERNEL */
 608
 609#define	IFFBITS \
 610    "\020\1UP\2BROADCAST\3DEBUG\4LOOPBACK\5POINTOPOINT\6UNNUMBERED" \
 611    "\7RUNNING\10NOARP\11PROMISC\12ALLMULTI\13OACTIVE\14SIMPLEX" \
 612    "\15LINK0\16LINK1\17LINK2\20MULTICAST"
 613
 614/* flags set internally only: */
 615#define	IFF_CANTCHANGE \
 616	(IFF_BROADCAST|IFF_POINTOPOINT|IFF_RUNNING|IFF_OACTIVE|\
 617	    IFF_SIMPLEX|IFF_MULTICAST|IFF_ALLMULTI|IFF_PROMISC)
 618
 619/*
 620 * Some convenience macros used for setting ifi_baudrate.
 621 */
 622#define	IF_Kbps(x)	((x) * 1000ULL)			/* kilobits/sec. */
 623#define	IF_Mbps(x)	(IF_Kbps((x) * 1000ULL))	/* megabits/sec. */
 624#define	IF_Gbps(x)	(IF_Mbps((x) * 1000ULL))	/* gigabits/sec. */
 625
 626/* Capabilities that interfaces can advertise. */
 627					/* 0x01 .. 0x40 were previously used */
 628#define	IFCAP_TSOv4		0x00080	/* can do TCPv4 segmentation offload */
 629#define	IFCAP_CSUM_IPv4_Rx	0x00100	/* can do IPv4 header checksums (Rx) */
 630#define	IFCAP_CSUM_IPv4_Tx	0x00200	/* can do IPv4 header checksums (Tx) */
 631#define	IFCAP_CSUM_TCPv4_Rx	0x00400	/* can do IPv4/TCP checksums (Rx) */
 632#define	IFCAP_CSUM_TCPv4_Tx	0x00800	/* can do IPv4/TCP checksums (Tx) */
 633#define	IFCAP_CSUM_UDPv4_Rx	0x01000	/* can do IPv4/UDP checksums (Rx) */
 634#define	IFCAP_CSUM_UDPv4_Tx	0x02000	/* can do IPv4/UDP checksums (Tx) */
 635#define	IFCAP_CSUM_TCPv6_Rx	0x04000	/* can do IPv6/TCP checksums (Rx) */
 636#define	IFCAP_CSUM_TCPv6_Tx	0x08000	/* can do IPv6/TCP checksums (Tx) */
 637#define	IFCAP_CSUM_UDPv6_Rx	0x10000	/* can do IPv6/UDP checksums (Rx) */
 638#define	IFCAP_CSUM_UDPv6_Tx	0x20000	/* can do IPv6/UDP checksums (Tx) */
 639#define	IFCAP_TSOv6		0x40000	/* can do TCPv6 segmentation offload */
 640#define	IFCAP_LRO		0x80000	/* can do Large Receive Offload */
 641#define	IFCAP_MASK		0xfff80 /* currently valid capabilities */
 642
 643#define	IFCAPBITS		\
 644	"\020"			\
 645	"\10TSO4"		\
 646	"\11IP4CSUM_Rx"		\
 647	"\12IP4CSUM_Tx"		\
 648	"\13TCP4CSUM_Rx"	\
 649	"\14TCP4CSUM_Tx"	\
 650	"\15UDP4CSUM_Rx"	\
 651	"\16UDP4CSUM_Tx"	\
 652	"\17TCP6CSUM_Rx"	\
 653	"\20TCP6CSUM_Tx"	\
 654	"\21UDP6CSUM_Rx"	\
 655	"\22UDP6CSUM_Tx"	\
 656	"\23TSO6"		\
 657	"\24LRO"		\
 658
 659#define	IF_AFDATA_LOCK_INIT(ifp)	\
 660	do {(ifp)->if_afdata_lock = rw_obj_alloc();} while (0)
 661
 662#define	IF_AFDATA_LOCK_DESTROY(ifp)	rw_obj_free((ifp)->if_afdata_lock)
 663
 664#define	IF_AFDATA_WLOCK(ifp)	rw_enter((ifp)->if_afdata_lock, RW_WRITER)
 665#define	IF_AFDATA_RLOCK(ifp)	rw_enter((ifp)->if_afdata_lock, RW_READER)
 666#define	IF_AFDATA_WUNLOCK(ifp)	rw_exit((ifp)->if_afdata_lock)
 667#define	IF_AFDATA_RUNLOCK(ifp)	rw_exit((ifp)->if_afdata_lock)
 668#define	IF_AFDATA_LOCK(ifp)	IF_AFDATA_WLOCK(ifp)
 669#define	IF_AFDATA_UNLOCK(ifp)	IF_AFDATA_WUNLOCK(ifp)
 670#define	IF_AFDATA_TRYLOCK(ifp)	rw_tryenter((ifp)->if_afdata_lock, RW_WRITER)
 671
 672#define	IF_AFDATA_LOCK_ASSERT(ifp)	\
 673	KASSERT(rw_lock_held((ifp)->if_afdata_lock))
 674#define	IF_AFDATA_RLOCK_ASSERT(ifp)	\
 675	KASSERT(rw_read_held((ifp)->if_afdata_lock))
 676#define	IF_AFDATA_WLOCK_ASSERT(ifp)	\
 677	KASSERT(rw_write_held((ifp)->if_afdata_lock))
 678
 679/*
 680 * Output queues (ifp->if_snd) and internetwork datagram level (pup level 1)
 681 * input routines have queues of messages stored on ifqueue structures
 682 * (defined above).  Entries are added to and deleted from these structures
 683 * by these macros, which should be called with ipl raised to splnet().
 684 */
 685#define	IF_QFULL(ifq)		((ifq)->ifq_len >= (ifq)->ifq_maxlen)
 686#define	IF_DROP(ifq)		((ifq)->ifq_drops++)
 687#define	IF_ENQUEUE(ifq, m) do { \
 688	(m)->m_nextpkt = 0; \
 689	if ((ifq)->ifq_tail == 0) \
 690		(ifq)->ifq_head = m; \
 691	else \
 692		(ifq)->ifq_tail->m_nextpkt = m; \
 693	(ifq)->ifq_tail = m; \
 694	(ifq)->ifq_len++; \
 695} while (/*CONSTCOND*/0)
 696#define	IF_PREPEND(ifq, m) do { \
 697	(m)->m_nextpkt = (ifq)->ifq_head; \
 698	if ((ifq)->ifq_tail == 0) \
 699		(ifq)->ifq_tail = (m); \
 700	(ifq)->ifq_head = (m); \
 701	(ifq)->ifq_len++; \
 702} while (/*CONSTCOND*/0)
 703#define	IF_DEQUEUE(ifq, m) do { \
 704	(m) = (ifq)->ifq_head; \
 705	if (m) { \
 706		if (((ifq)->ifq_head = (m)->m_nextpkt) == 0) \
 707			(ifq)->ifq_tail = 0; \
 708		(m)->m_nextpkt = 0; \
 709		(ifq)->ifq_len--; \
 710	} \
 711} while (/*CONSTCOND*/0)
 712#define	IF_POLL(ifq, m)		((m) = (ifq)->ifq_head)
 713#define	IF_PURGE(ifq)							\
 714do {									\
 715	struct mbuf *__m0;						\
 716									\
 717	for (;;) {							\
 718		IF_DEQUEUE((ifq), __m0);				\
 719		if (__m0 == NULL)					\
 720			break;						\
 721		else							\
 722			m_freem(__m0);					\
 723	}								\
 724} while (/*CONSTCOND*/ 0)
 725#define	IF_IS_EMPTY(ifq)	((ifq)->ifq_len == 0)
 726
 727#ifndef IFQ_MAXLEN
 728#define	IFQ_MAXLEN	256
 729#endif
 730#define	IFNET_SLOWHZ	1		/* granularity is 1 second */
 731
 732/*
 733 * Structure defining statistics and other data kept regarding an address
 734 * on a network interface.
 735 */
 736struct ifaddr_data {
 737	int64_t	ifad_inbytes;
 738	int64_t	ifad_outbytes;
 739};
 740
 741/*
 742 * The ifaddr structure contains information about one address
 743 * of an interface.  They are maintained by the different address families,
 744 * are allocated and attached when an address is set, and are linked
 745 * together so all addresses for an interface can be located.
 746 */
 747struct ifaddr {
 748	struct	sockaddr *ifa_addr;	/* address of interface */
 749	struct	sockaddr *ifa_dstaddr;	/* other end of p-to-p link */
 750#define	ifa_broadaddr	ifa_dstaddr	/* broadcast address interface */
 751	struct	sockaddr *ifa_netmask;	/* used to determine subnet */
 752	struct	ifnet *ifa_ifp;		/* back-pointer to interface */
 753	TAILQ_ENTRY(ifaddr) ifa_list;	/* list of addresses for interface */
 754	struct	ifaddr_data	ifa_data;	/* statistics on the address */
 755	void	(*ifa_rtrequest)	/* check or clean routes (+ or -)'d */
 756		        (int, struct rtentry *, const struct rt_addrinfo *);
 757	u_int	ifa_flags;		/* mostly rt_flags for cloning */
 758	int	ifa_refcnt;		/* count of references */
 759	int	ifa_metric;		/* cost of going out this interface */
 760	struct ifaddr	*(*ifa_getifa)(struct ifaddr *,
 761			               const struct sockaddr *);
 762	uint32_t	*ifa_seqno;
 763	int16_t	ifa_preference;	/* preference level for this address */
 764#ifdef _KERNEL
 765	struct pslist_entry     ifa_pslist_entry;
 766	struct psref_target	ifa_psref;
 767#endif
 768};
 769#define	IFA_ROUTE	RTF_UP	/* (0x01) route installed */
 770#define	IFA_DESTROYING	0x2
 771
 772/*
 773 * Message format for use in obtaining information about interfaces from
 774 * sysctl and the routing socket.  We need to force 64-bit alignment if we
 775 * aren't using compatibility definitions.
 776 */
 777#if !defined(_KERNEL) || !defined(COMPAT_RTSOCK)
 778#define	__align64	__aligned(sizeof(uint64_t))
 779#else
 780#define	__align64
 781#endif
 782struct if_msghdr {
 783	u_short	ifm_msglen __align64;
 784				/* to skip over non-understood messages */
 785	u_char	ifm_version;	/* future binary compatibility */
 786	u_char	ifm_type;	/* message type */
 787	int	ifm_addrs;	/* like rtm_addrs */
 788	int	ifm_flags;	/* value of if_flags */
 789	u_short	ifm_index;	/* index for associated ifp */
 790	struct	if_data ifm_data __align64;
 791				/* statistics and other data about if */
 792};
 793
 794/*
 795 * Message format for use in obtaining information about interface addresses
 796 * from sysctl and the routing socket.
 797 */
 798struct ifa_msghdr {
 799	u_short	ifam_msglen __align64;
 800				/* to skip over non-understood messages */
 801	u_char	ifam_version;	/* future binary compatibility */
 802	u_char	ifam_type;	/* message type */
 803	u_short	ifam_index;	/* index for associated ifp */
 804	int	ifam_flags;	/* value of ifa_flags */
 805	int	ifam_addrs;	/* like rtm_addrs */
 806	pid_t	ifam_pid;	/* identify sender */
 807	int	ifam_addrflags;	/* family specific address flags */
 808	int	ifam_metric;	/* value of ifa_metric */
 809};
 810
 811/*
 812 * Message format announcing the arrival or departure of a network interface.
 813 */
 814struct if_announcemsghdr {
 815	u_short	ifan_msglen __align64;
 816				/* to skip over non-understood messages */
 817	u_char	ifan_version;	/* future binary compatibility */
 818	u_char	ifan_type;	/* message type */
 819	u_short	ifan_index;	/* index for associated ifp */
 820	char	ifan_name[IFNAMSIZ]; /* if name, e.g. "en0" */
 821	u_short	ifan_what;	/* what type of announcement */
 822};
 823
 824#define	IFAN_ARRIVAL	0	/* interface arrival */
 825#define	IFAN_DEPARTURE	1	/* interface departure */
 826
 827#undef __align64
 828
 829/*
 830 * Interface request structure used for socket
 831 * ioctl's.  All interface ioctl's must have parameter
 832 * definitions which begin with ifr_name.  The
 833 * remainder may be interface specific.
 834 */
 835struct	ifreq {
 836	char	ifr_name[IFNAMSIZ];		/* if name, e.g. "en0" */
 837	union {
 838		struct	sockaddr ifru_addr;
 839		struct	sockaddr ifru_dstaddr;
 840		struct	sockaddr ifru_broadaddr;
 841		struct	sockaddr_storage ifru_space;
 842		short	ifru_flags;
 843		int	ifru_addrflags;
 844		int	ifru_metric;
 845		int	ifru_mtu;
 846		int	ifru_dlt;
 847		u_int	ifru_value;
 848		void *	ifru_data;
 849		struct {
 850			uint32_t	b_buflen;
 851			void		*b_buf;
 852		} ifru_b;
 853	} ifr_ifru;
 854#define	ifr_addr	ifr_ifru.ifru_addr	/* address */
 855#define	ifr_dstaddr	ifr_ifru.ifru_dstaddr	/* other end of p-to-p link */
 856#define	ifr_broadaddr	ifr_ifru.ifru_broadaddr	/* broadcast address */
 857#define	ifr_space	ifr_ifru.ifru_space	/* sockaddr_storage */
 858#define	ifr_flags	ifr_ifru.ifru_flags	/* flags */
 859#define	ifr_addrflags	ifr_ifru.ifru_addrflags	/* addr flags */
 860#define	ifr_metric	ifr_ifru.ifru_metric	/* metric */
 861#define	ifr_mtu		ifr_ifru.ifru_mtu	/* mtu */
 862#define	ifr_dlt		ifr_ifru.ifru_dlt	/* data link type (DLT_*) */
 863#define	ifr_value	ifr_ifru.ifru_value	/* generic value */
 864#define	ifr_media	ifr_ifru.ifru_metric	/* media options (overload) */
 865#define	ifr_data	ifr_ifru.ifru_data	/* for use by interface
 866						 * XXX deprecated
 867						 */
 868#define	ifr_buf		ifr_ifru.ifru_b.b_buf	/* new interface ioctls */
 869#define	ifr_buflen	ifr_ifru.ifru_b.b_buflen
 870#define	ifr_index	ifr_ifru.ifru_value	/* interface index, BSD */
 871#define	ifr_ifindex	ifr_index		/* interface index, linux */
 872};
 873
 874#ifdef _KERNEL
 875#define	ifreq_setdstaddr	ifreq_setaddr
 876#define	ifreq_setbroadaddr	ifreq_setaddr
 877#define	ifreq_getdstaddr	ifreq_getaddr
 878#define	ifreq_getbroadaddr	ifreq_getaddr
 879
 880static __inline const struct sockaddr *
 881/*ARGSUSED*/
 882ifreq_getaddr(u_long cmd, const struct ifreq *ifr)
 883{
 884	return &ifr->ifr_addr;
 885}
 886#endif /* _KERNEL */
 887
 888struct ifcapreq {
 889	char		ifcr_name[IFNAMSIZ];	/* if name, e.g. "en0" */
 890	uint64_t	ifcr_capabilities;	/* supported capabiliites */
 891	uint64_t	ifcr_capenable;		/* capabilities enabled */
 892};
 893
 894struct ifaliasreq {
 895	char	ifra_name[IFNAMSIZ];		/* if name, e.g. "en0" */
 896	struct	sockaddr ifra_addr;
 897	struct	sockaddr ifra_dstaddr;
 898#define	ifra_broadaddr	ifra_dstaddr
 899	struct	sockaddr ifra_mask;
 900};
 901
 902struct ifdatareq {
 903	char	ifdr_name[IFNAMSIZ];		/* if name, e.g. "en0" */
 904	struct	if_data ifdr_data;
 905};
 906
 907struct ifmediareq {
 908	char	ifm_name[IFNAMSIZ];	/* if name, e.g. "en0" */
 909	int	ifm_current;		/* IFMWD: current media options */
 910	int	ifm_mask;		/* IFMWD: don't care mask */
 911	int	ifm_status;		/* media status */
 912	int	ifm_active;		/* IFMWD: active options */
 913	int	ifm_count;		/* # entries in ifm_ulist
 914					   array */
 915	int	*ifm_ulist;		/* array of ifmedia word */
 916};
 917
 918
 919struct  ifdrv {
 920	char		ifd_name[IFNAMSIZ];	/* if name, e.g. "en0" */
 921	unsigned long	ifd_cmd;
 922	size_t		ifd_len;
 923	void		*ifd_data;
 924};
 925#define IFLINKSTR_QUERYLEN	0x01
 926#define IFLINKSTR_UNSET		0x02
 927
 928/*
 929 * Structure used in SIOCGIFCONF request.
 930 * Used to retrieve interface configuration
 931 * for machine (useful for programs which
 932 * must know all networks accessible).
 933 */
 934struct	ifconf {
 935	int	ifc_len;		/* size of associated buffer */
 936	union {
 937		void *	ifcu_buf;
 938		struct	ifreq *ifcu_req;
 939	} ifc_ifcu;
 940#define	ifc_buf	ifc_ifcu.ifcu_buf	/* buffer address */
 941#define	ifc_req	ifc_ifcu.ifcu_req	/* array of structures returned */
 942};
 943
 944/*
 945 * Structure for SIOC[AGD]LIFADDR
 946 */
 947struct if_laddrreq {
 948	char iflr_name[IFNAMSIZ];
 949	unsigned int flags;
 950#define IFLR_PREFIX	0x8000	/* in: prefix given  out: kernel fills id */
 951#define IFLR_ACTIVE	0x4000	/* in/out: link-layer address activation */
 952#define IFLR_FACTORY	0x2000	/* in/out: factory link-layer address */
 953	unsigned int prefixlen;		/* in/out */
 954	struct sockaddr_storage addr;	/* in/out */
 955	struct sockaddr_storage dstaddr; /* out */
 956};
 957
 958/*
 959 * Structure for SIOC[SG]IFADDRPREF
 960 */
 961struct if_addrprefreq {
 962	char			ifap_name[IFNAMSIZ];
 963	int16_t			ifap_preference;	/* in/out */
 964	struct sockaddr_storage	ifap_addr;		/* in/out */
 965};
 966
 967#include <net/if_arp.h>
 968
 969#endif /* _NETBSD_SOURCE */
 970
 971#ifdef _KERNEL
 972#ifdef ALTQ
 973#define IFQ_ENQUEUE(ifq, m, err)					\
 974do {									\
 975	mutex_enter((ifq)->ifq_lock);					\
 976	if (ALTQ_IS_ENABLED(ifq))					\
 977		ALTQ_ENQUEUE((ifq), (m), (err));			\
 978	else {								\
 979		if (IF_QFULL(ifq)) {					\
 980			m_freem(m);					\
 981			(err) = ENOBUFS;				\
 982		} else {						\
 983			IF_ENQUEUE((ifq), (m));				\
 984			(err) = 0;					\
 985		}							\
 986	}								\
 987	if ((err))							\
 988		(ifq)->ifq_drops++;					\
 989	mutex_exit((ifq)->ifq_lock);					\
 990} while (/*CONSTCOND*/ 0)
 991
 992#define IFQ_DEQUEUE(ifq, m)						\
 993do {									\
 994	mutex_enter((ifq)->ifq_lock);					\
 995	if (TBR_IS_ENABLED(ifq))					\
 996		(m) = tbr_dequeue((ifq), ALTDQ_REMOVE);			\
 997	else if (ALTQ_IS_ENABLED(ifq))					\
 998		ALTQ_DEQUEUE((ifq), (m));				\
 999	else								\
1000		IF_DEQUEUE((ifq), (m));					\
1001	mutex_exit((ifq)->ifq_lock);					\
1002} while (/*CONSTCOND*/ 0)
1003
1004#define	IFQ_POLL(ifq, m)						\
1005do {									\
1006	mutex_enter((ifq)->ifq_lock);					\
1007	if (TBR_IS_ENABLED(ifq))					\
1008		(m) = tbr_dequeue((ifq), ALTDQ_POLL);			\
1009	else if (ALTQ_IS_ENABLED(ifq))					\
1010		ALTQ_POLL((ifq), (m));					\
1011	else								\
1012		IF_POLL((ifq), (m));					\
1013	mutex_exit((ifq)->ifq_lock);					\
1014} while (/*CONSTCOND*/ 0)
1015
1016#define	IFQ_PURGE(ifq)							\
1017do {									\
1018	mutex_enter((ifq)->ifq_lock);					\
1019	if (ALTQ_IS_ENABLED(ifq))					\
1020		ALTQ_PURGE(ifq);					\
1021	else								\
1022		IF_PURGE(ifq);						\
1023	mutex_exit((ifq)->ifq_lock);					\
1024} while (/*CONSTCOND*/ 0)
1025
1026#define	IFQ_SET_READY(ifq)						\
1027do {									\
1028	(ifq)->altq_flags |= ALTQF_READY;				\
1029} while (/*CONSTCOND*/ 0)
1030
1031#define	IFQ_CLASSIFY(ifq, m, af)					\
1032do {									\
1033	KASSERT(((m)->m_flags & M_PKTHDR) != 0);			\
1034	mutex_enter((ifq)->ifq_lock);					\
1035	if (ALTQ_IS_ENABLED(ifq)) {					\
1036		if (ALTQ_NEEDS_CLASSIFY(ifq))				\
1037			(m)->m_pkthdr.pattr_class = (*(ifq)->altq_classify) \
1038				((ifq)->altq_clfier, (m), (af));	\
1039		(m)->m_pkthdr.pattr_af = (af);				\
1040		(m)->m_pkthdr.pattr_hdr = mtod((m), void *);		\
1041	}								\
1042	mutex_exit((ifq)->ifq_lock);					\
1043} while (/*CONSTCOND*/ 0)
1044#else /* ! ALTQ */
1045#define	IFQ_ENQUEUE(ifq, m, err)					\
1046do {									\
1047	mutex_enter((ifq)->ifq_lock);					\
1048	if (IF_QFULL(ifq)) {						\
1049		m_freem(m);						\
1050		(err) = ENOBUFS;					\
1051	} else {							\
1052		IF_ENQUEUE((ifq), (m));					\
1053		(err) = 0;						\
1054	}								\
1055	if (err)							\
1056		(ifq)->ifq_drops++;					\
1057	mutex_exit((ifq)->ifq_lock);					\
1058} while (/*CONSTCOND*/ 0)
1059
1060#define	IFQ_DEQUEUE(ifq, m)						\
1061do {									\
1062	mutex_enter((ifq)->ifq_lock);					\
1063	IF_DEQUEUE((ifq), (m));						\
1064	mutex_exit((ifq)->ifq_lock);					\
1065} while (/*CONSTCOND*/ 0)
1066
1067#define	IFQ_POLL(ifq, m)						\
1068do {									\
1069	mutex_enter((ifq)->ifq_lock);					\
1070	IF_POLL((ifq), (m));						\
1071	mutex_exit((ifq)->ifq_lock);					\
1072} while (/*CONSTCOND*/ 0)
1073
1074#define	IFQ_PURGE(ifq)							\
1075do {									\
1076	mutex_enter((ifq)->ifq_lock);					\
1077	IF_PURGE(ifq);							\
1078	mutex_exit((ifq)->ifq_lock);					\
1079} while (/*CONSTCOND*/ 0)
1080
1081#define	IFQ_SET_READY(ifq)	/* nothing */
1082
1083#define	IFQ_CLASSIFY(ifq, m, af) /* nothing */
1084
1085#endif /* ALTQ */
1086
1087#define IFQ_LOCK_INIT(ifq)	(ifq)->ifq_lock =			\
1088	    mutex_obj_alloc(MUTEX_DEFAULT, IPL_NET)
1089#define IFQ_LOCK_DESTROY(ifq)	mutex_obj_free((ifq)->ifq_lock)
1090#define IFQ_LOCK(ifq)		mutex_enter((ifq)->ifq_lock)
1091#define IFQ_UNLOCK(ifq)		mutex_exit((ifq)->ifq_lock)
1092
1093#define	IFQ_IS_EMPTY(ifq)		IF_IS_EMPTY(ifq)
1094#define	IFQ_INC_LEN(ifq)		((ifq)->ifq_len++)
1095#define	IFQ_DEC_LEN(ifq)		(--(ifq)->ifq_len)
1096#define	IFQ_INC_DROPS(ifq)		((ifq)->ifq_drops++)
1097#define	IFQ_SET_MAXLEN(ifq, len)	((ifq)->ifq_maxlen = (len))
1098
1099#include <sys/mallocvar.h>
1100MALLOC_DECLARE(M_IFADDR);
1101MALLOC_DECLARE(M_IFMADDR);
1102
1103int ifreq_setaddr(u_long, struct ifreq *, const struct sockaddr *);
1104
1105struct ifnet *if_alloc(u_char);
1106void if_free(struct ifnet *);
1107void if_initname(struct ifnet *, const char *, int);
1108struct ifaddr *if_dl_create(const struct ifnet *, const struct sockaddr_dl **);
1109void if_activate_sadl(struct ifnet *, struct ifaddr *,
1110    const struct sockaddr_dl *);
1111void	if_set_sadl(struct ifnet *, const void *, u_char, bool);
1112void	if_alloc_sadl(struct ifnet *);
1113void	if_free_sadl(struct ifnet *, int);
1114void	if_initialize(struct ifnet *);
1115void	if_register(struct ifnet *);
1116void	if_attach(struct ifnet *); /* Deprecated. Use if_initialize and if_register */
1117void	if_attachdomain(void);
1118void	if_deactivate(struct ifnet *);
1119bool	if_is_deactivated(const struct ifnet *);
1120void	if_export_if_data(struct ifnet *, struct if_data *, bool);
1121void	if_purgeaddrs(struct ifnet *, int, void (*)(struct ifaddr *));
1122void	if_detach(struct ifnet *);
1123void	if_down(struct ifnet *);
1124void	if_down_locked(struct ifnet *);
1125void	if_link_state_change(struct ifnet *, int);
1126void	if_domain_link_state_change(struct ifnet *, int);
1127void	if_up(struct ifnet *);
1128void	ifinit(void);
1129void	ifinit1(void);
1130void	ifinit_post(void);
1131int	ifaddrpref_ioctl(struct socket *, u_long, void *, struct ifnet *);
1132extern int (*ifioctl)(struct socket *, u_long, void *, struct lwp *);
1133int	ifioctl_common(struct ifnet *, u_long, void *);
1134int	ifpromisc(struct ifnet *, int);
1135int	ifpromisc_locked(struct ifnet *, int);
1136int	if_addr_init(ifnet_t *, struct ifaddr *, bool);
1137int	if_do_dad(struct ifnet *);
1138int	if_mcast_op(ifnet_t *, const unsigned long, const struct sockaddr *);
1139int	if_flags_set(struct ifnet *, const u_short);
1140int	if_clone_list(int, char *, int *);
1141
1142int	if_ioctl(struct ifnet *, u_long, void *);
1143int	if_init(struct ifnet *);
1144void	if_stop(struct ifnet *, int);
1145
1146struct	ifnet *ifunit(const char *);
1147struct	ifnet *if_get(const char *, struct psref *);
1148ifnet_t *if_byindex(u_int);
1149ifnet_t *_if_byindex(u_int);
1150ifnet_t *if_get_byindex(u_int, struct psref *);
1151ifnet_t *if_get_bylla(const void *, unsigned char, struct psref *);
1152void	if_put(const struct ifnet *, struct psref *);
1153void	if_acquire(struct ifnet *, struct psref *);
1154#define	if_release	if_put
1155
1156int if_tunnel_check_nesting(struct ifnet *, struct mbuf *, int);
1157percpu_t *if_tunnel_alloc_ro_percpu(void);
1158void if_tunnel_free_ro_percpu(percpu_t *);
1159void if_tunnel_ro_percpu_rtcache_free(percpu_t *);
1160
1161struct tunnel_ro {
1162	struct route *tr_ro;
1163	kmutex_t *tr_lock;
1164};
1165
1166static inline void
1167if_tunnel_get_ro(percpu_t *ro_percpu, struct route **ro, kmutex_t **lock)
1168{
1169	struct tunnel_ro *tro;
1170
1171	tro = percpu_getref(ro_percpu);
1172	*ro = tro->tr_ro;
1173	*lock = tro->tr_lock;
1174	mutex_enter(*lock);
1175}
1176
1177static inline void
1178if_tunnel_put_ro(percpu_t *ro_percpu, kmutex_t *lock)
1179{
1180
1181	mutex_exit(lock);
1182	percpu_putref(ro_percpu);
1183}
1184
1185static __inline if_index_t
1186if_get_index(const struct ifnet *ifp)
1187{
1188
1189	return ifp != NULL ? ifp->if_index : 0;
1190}
1191
1192bool	if_held(struct ifnet *);
1193
1194void	if_input(struct ifnet *, struct mbuf *);
1195
1196struct if_percpuq *
1197	if_percpuq_create(struct ifnet *);
1198void	if_percpuq_destroy(struct if_percpuq *);
1199void
1200	if_percpuq_enqueue(struct if_percpuq *, struct mbuf *);
1201
1202void	if_deferred_start_init(struct ifnet *, void (*)(struct ifnet *));
1203void	if_schedule_deferred_start(struct ifnet *);
1204
1205void ifa_insert(struct ifnet *, struct ifaddr *);
1206void ifa_remove(struct ifnet *, struct ifaddr *);
1207
1208void	ifa_psref_init(struct ifaddr *);
1209void	ifa_acquire(struct ifaddr *, struct psref *);
1210void	ifa_release(struct ifaddr *, struct psref *);
1211bool	ifa_held(struct ifaddr *);
1212bool	ifa_is_destroying(struct ifaddr *);
1213
1214void	ifaref(struct ifaddr *);
1215void	ifafree(struct ifaddr *);
1216
1217struct	ifaddr *ifa_ifwithaddr(const struct sockaddr *);
1218struct	ifaddr *ifa_ifwithaddr_psref(const struct sockaddr *, struct psref *);
1219struct	ifaddr *ifa_ifwithaf(int);
1220struct	ifaddr *ifa_ifwithdstaddr(const struct sockaddr *);
1221struct	ifaddr *ifa_ifwithdstaddr_psref(const struct sockaddr *,
1222	    struct psref *);
1223struct	ifaddr *ifa_ifwithnet(const struct sockaddr *);
1224struct	ifaddr *ifa_ifwithnet_psref(const struct sockaddr *, struct psref *);
1225struct	ifaddr *ifa_ifwithladdr(const struct sockaddr *);
1226struct	ifaddr *ifa_ifwithladdr_psref(const struct sockaddr *, struct psref *);
1227struct	ifaddr *ifaof_ifpforaddr(const struct sockaddr *, struct ifnet *);
1228struct	ifaddr *ifaof_ifpforaddr_psref(const struct sockaddr *, struct ifnet *,
1229	    struct psref *);
1230void	link_rtrequest(int, struct rtentry *, const struct rt_addrinfo *);
1231void	p2p_rtrequest(int, struct rtentry *, const struct rt_addrinfo *);
1232
1233void	if_clone_attach(struct if_clone *);
1234void	if_clone_detach(struct if_clone *);
1235
1236int	if_transmit_lock(struct ifnet *, struct mbuf *);
1237
1238int	ifq_enqueue(struct ifnet *, struct mbuf *);
1239int	ifq_enqueue2(struct ifnet *, struct ifqueue *, struct mbuf *);
1240
1241int	loioctl(struct ifnet *, u_long, void *);
1242void	loopattach(int);
1243void	loopinit(void);
1244int	looutput(struct ifnet *,
1245	   struct mbuf *, const struct sockaddr *, const struct rtentry *);
1246
1247void *	if_linkstate_change_establish(struct ifnet *,
1248	    void (*)(void *), void *);
1249void	if_linkstate_change_disestablish(struct ifnet *,
1250	    void *, kmutex_t *);
1251
1252/*
1253 * These are exported because they're an easy way to tell if
1254 * an interface is going away without having to burn a flag.
1255 */
1256int	if_nulloutput(struct ifnet *, struct mbuf *,
1257	    const struct sockaddr *, const struct rtentry *);
1258void	if_nullinput(struct ifnet *, struct mbuf *);
1259void	if_nullstart(struct ifnet *);
1260int	if_nulltransmit(struct ifnet *, struct mbuf *);
1261int	if_nullioctl(struct ifnet *, u_long, void *);
1262int	if_nullinit(struct ifnet *);
1263void	if_nullstop(struct ifnet *, int);
1264void	if_nullslowtimo(struct ifnet *);
1265#define	if_nullwatchdog	if_nullslowtimo
1266void	if_nulldrain(struct ifnet *);
1267#else
1268struct if_nameindex {
1269	unsigned int	if_index;	/* 1, 2, ... */
1270	char		*if_name;	/* null terminated name: "le0", ... */
1271};
1272
1273#include <sys/cdefs.h>
1274__BEGIN_DECLS
1275unsigned int if_nametoindex(const char *);
1276char *	if_indextoname(unsigned int, char *);
1277struct	if_nameindex * if_nameindex(void);
1278void	if_freenameindex(struct if_nameindex *);
1279__END_DECLS
1280#endif /* _KERNEL */ /* XXX really ALTQ? */
1281
1282#ifdef _KERNEL
1283
1284#define	IFADDR_FIRST(__ifp)		TAILQ_FIRST(&(__ifp)->if_addrlist)
1285#define	IFADDR_NEXT(__ifa)		TAILQ_NEXT((__ifa), ifa_list)
1286#define	IFADDR_FOREACH(__ifa, __ifp)	TAILQ_FOREACH(__ifa, \
1287					    &(__ifp)->if_addrlist, ifa_list)
1288#define	IFADDR_FOREACH_SAFE(__ifa, __ifp, __nifa) \
1289					    TAILQ_FOREACH_SAFE(__ifa, \
1290					    &(__ifp)->if_addrlist, ifa_list, __nifa)
1291#define	IFADDR_EMPTY(__ifp)		TAILQ_EMPTY(&(__ifp)->if_addrlist)
1292
1293#define IFADDR_ENTRY_INIT(__ifa)					\
1294	PSLIST_ENTRY_INIT((__ifa), ifa_pslist_entry)
1295#define IFADDR_ENTRY_DESTROY(__ifa)					\
1296	PSLIST_ENTRY_DESTROY((__ifa), ifa_pslist_entry)
1297#define IFADDR_READER_EMPTY(__ifp)					\
1298	(PSLIST_READER_FIRST(&(__ifp)->if_addr_pslist, struct ifaddr,	\
1299	                     ifa_pslist_entry) == NULL)
1300#define IFADDR_READER_FIRST(__ifp)					\
1301	PSLIST_READER_FIRST(&(__ifp)->if_addr_pslist, struct ifaddr,	\
1302	                    ifa_pslist_entry)
1303#define IFADDR_READER_NEXT(__ifa)					\
1304	PSLIST_READER_NEXT((__ifa), struct ifaddr, ifa_pslist_entry)
1305#define IFADDR_READER_FOREACH(__ifa, __ifp)				\
1306	PSLIST_READER_FOREACH((__ifa), &(__ifp)->if_addr_pslist, struct ifaddr,\
1307	                      ifa_pslist_entry)
1308#define IFADDR_WRITER_INSERT_HEAD(__ifp, __ifa)				\
1309	PSLIST_WRITER_INSERT_HEAD(&(__ifp)->if_addr_pslist, (__ifa),	\
1310	                          ifa_pslist_entry)
1311#define IFADDR_WRITER_REMOVE(__ifa)					\
1312	PSLIST_WRITER_REMOVE((__ifa), ifa_pslist_entry)
1313#define IFADDR_WRITER_FOREACH(__ifa, __ifp)				\
1314	PSLIST_WRITER_FOREACH((__ifa), &(__ifp)->if_addr_pslist, struct ifaddr,\
1315	                      ifa_pslist_entry)
1316#define IFADDR_WRITER_NEXT(__ifp)					\
1317	PSLIST_WRITER_NEXT((__ifp), struct ifaddr, ifa_pslist_entry)
1318#define IFADDR_WRITER_INSERT_AFTER(__ifp, __new)			\
1319	PSLIST_WRITER_INSERT_AFTER((__ifp), (__new), ifa_pslist_entry)
1320#define IFADDR_WRITER_EMPTY(__ifp)					\
1321	(PSLIST_WRITER_FIRST(&(__ifp)->if_addr_pslist, struct ifaddr,	\
1322	                     ifa_pslist_entry) == NULL)
1323#define IFADDR_WRITER_INSERT_TAIL(__ifp, __new)				\
1324	do {								\
1325		if (IFADDR_WRITER_EMPTY(__ifp)) {			\
1326			IFADDR_WRITER_INSERT_HEAD((__ifp), (__new));	\
1327		} else {						\
1328			struct ifaddr *__ifa;				\
1329			IFADDR_WRITER_FOREACH(__ifa, (__ifp)) {		\
1330				if (IFADDR_WRITER_NEXT(__ifa) == NULL) {\
1331					IFADDR_WRITER_INSERT_AFTER(__ifa,\
1332					    (__new));			\
1333					break;				\
1334				}					\
1335			}						\
1336		}							\
1337	} while (0)
1338
1339#define	IFNET_GLOBAL_LOCK()			mutex_enter(&ifnet_mtx)
1340#define	IFNET_GLOBAL_UNLOCK()			mutex_exit(&ifnet_mtx)
1341#define	IFNET_GLOBAL_LOCKED()			mutex_owned(&ifnet_mtx)
1342
1343#define IFNET_READER_EMPTY() \
1344	(PSLIST_READER_FIRST(&ifnet_pslist, struct ifnet, if_pslist_entry) == NULL)
1345#define IFNET_READER_FIRST() \
1346	PSLIST_READER_FIRST(&ifnet_pslist, struct ifnet, if_pslist_entry)
1347#define IFNET_READER_NEXT(__ifp) \
1348	PSLIST_READER_NEXT((__ifp), struct ifnet, if_pslist_entry)
1349#define IFNET_READER_FOREACH(__ifp) \
1350	PSLIST_READER_FOREACH((__ifp), &ifnet_pslist, struct ifnet, \
1351	                      if_pslist_entry)
1352#define IFNET_WRITER_INSERT_HEAD(__ifp) \
1353	PSLIST_WRITER_INSERT_HEAD(&ifnet_pslist, (__ifp), if_pslist_entry)
1354#define IFNET_WRITER_REMOVE(__ifp) \
1355	PSLIST_WRITER_REMOVE((__ifp), if_pslist_entry)
1356#define IFNET_WRITER_FOREACH(__ifp) \
1357	PSLIST_WRITER_FOREACH((__ifp), &ifnet_pslist, struct ifnet, \
1358	                      if_pslist_entry)
1359#define IFNET_WRITER_NEXT(__ifp) \
1360	PSLIST_WRITER_NEXT((__ifp), struct ifnet, if_pslist_entry)
1361#define IFNET_WRITER_INSERT_AFTER(__ifp, __new) \
1362	PSLIST_WRITER_INSERT_AFTER((__ifp), (__new), if_pslist_entry)
1363#define IFNET_WRITER_EMPTY() \
1364	(PSLIST_WRITER_FIRST(&ifnet_pslist, struct ifnet, if_pslist_entry) == NULL)
1365#define IFNET_WRITER_INSERT_TAIL(__new)					\
1366	do {								\
1367		if (IFNET_WRITER_EMPTY()) {				\
1368			IFNET_WRITER_INSERT_HEAD(__new);		\
1369		} else {						\
1370			struct ifnet *__ifp;				\
1371			IFNET_WRITER_FOREACH(__ifp) {			\
1372				if (IFNET_WRITER_NEXT(__ifp) == NULL) {	\
1373					IFNET_WRITER_INSERT_AFTER(__ifp,\
1374					    (__new));			\
1375					break;				\
1376				}					\
1377			}						\
1378		}							\
1379	} while (0)
1380
1381#define IFNET_LOCK(ifp)		mutex_enter((ifp)->if_ioctl_lock)
1382#define IFNET_UNLOCK(ifp)	mutex_exit((ifp)->if_ioctl_lock)
1383#define IFNET_LOCKED(ifp)	mutex_owned((ifp)->if_ioctl_lock)
1384
1385#define IFNET_ASSERT_UNLOCKED(ifp)	\
1386	KDASSERT(mutex_ownable((ifp)->if_ioctl_lock))
1387
1388extern struct pslist_head ifnet_pslist;
1389extern kmutex_t ifnet_mtx;
1390
1391extern struct ifnet *lo0ifp;
1392
1393/*
1394 * ifq sysctl support
1395 */
1396int	sysctl_ifq(int *name, u_int namelen, void *oldp,
1397		       size_t *oldlenp, void *newp, size_t newlen,
1398		       struct ifqueue *ifq);
1399/* symbolic names for terminal (per-protocol) CTL_IFQ_ nodes */
1400#define IFQCTL_LEN	1
1401#define IFQCTL_MAXLEN	2
1402#define IFQCTL_PEAK	3
1403#define IFQCTL_DROPS	4
1404
1405/*
1406 * Hook for if_vlan - needed by if_agr
1407 */
1408MODULE_HOOK(if_vlan_vlan_input_hook,
1409    struct mbuf *, (struct ifnet *, struct mbuf *));
1410
1411#endif /* _KERNEL */
1412
1413#endif /* !_NET_IF_H_ */