ifq.h revision 228571
1/*-
2 * Copyright (c) 1982, 1986, 1989, 1993
3 *	The Regents of the University of California.  All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 * 4. Neither the name of the University nor the names of its contributors
14 *    may be used to endorse or promote products derived from this software
15 *    without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 *	From: @(#)if.h	8.1 (Berkeley) 6/10/93
30 * $FreeBSD: head/sys/net/if_var.h 228571 2011-12-16 12:16:56Z glebius $
31 */
32
33#ifndef	_NET_IF_VAR_H_
34#define	_NET_IF_VAR_H_
35
36/*
37 * Structures defining a network interface, providing a packet
38 * transport mechanism (ala level 0 of the PUP protocols).
39 *
40 * Each interface accepts output datagrams of a specified maximum
41 * length, and provides higher level routines with input datagrams
42 * received from its medium.
43 *
44 * Output occurs when the routine if_output is called, with three parameters:
45 *	(*ifp->if_output)(ifp, m, dst, rt)
46 * Here m is the mbuf chain to be sent and dst is the destination address.
47 * The output routine encapsulates the supplied datagram if necessary,
48 * and then transmits it on its medium.
49 *
50 * On input, each interface unwraps the data received by it, and either
51 * places it on the input queue of an internetwork datagram routine
52 * and posts the associated software interrupt, or passes the datagram to a raw
53 * packet input routine.
54 *
55 * Routines exist for locating interfaces by their addresses
56 * or for locating an interface on a certain network, as well as more general
57 * routing and gateway routines maintaining information used to locate
58 * interfaces.  These routines live in the files if.c and route.c
59 */
60
61#ifdef __STDC__
62/*
63 * Forward structure declarations for function prototypes [sic].
64 */
65struct	mbuf;
66struct	thread;
67struct	rtentry;
68struct	rt_addrinfo;
69struct	socket;
70struct	ether_header;
71struct	carp_if;
72struct	carp_softc;
73struct  ifvlantrunk;
74struct	route;
75struct	vnet;
76#endif
77
78#include <sys/queue.h>		/* get TAILQ macros */
79
80#ifdef _KERNEL
81#include <sys/mbuf.h>
82#include <sys/eventhandler.h>
83#include <sys/buf_ring.h>
84#include <net/vnet.h>
85#endif /* _KERNEL */
86#include <sys/lock.h>		/* XXX */
87#include <sys/mutex.h>		/* XXX */
88#include <sys/rwlock.h>		/* XXX */
89#include <sys/sx.h>		/* XXX */
90#include <sys/event.h>		/* XXX */
91#include <sys/_task.h>
92
93#define	IF_DUNIT_NONE	-1
94
95#include <altq/if_altq.h>
96
97TAILQ_HEAD(ifnethead, ifnet);	/* we use TAILQs so that the order of */
98TAILQ_HEAD(ifaddrhead, ifaddr);	/* instantiation is preserved in the list */
99TAILQ_HEAD(ifprefixhead, ifprefix);
100TAILQ_HEAD(ifmultihead, ifmultiaddr);
101TAILQ_HEAD(ifgrouphead, ifg_group);
102
103/*
104 * Structure defining a queue for a network interface.
105 */
106struct	ifqueue {
107	struct	mbuf *ifq_head;
108	struct	mbuf *ifq_tail;
109	int	ifq_len;
110	int	ifq_maxlen;
111	int	ifq_drops;
112	struct	mtx ifq_mtx;
113};
114
115/*
116 * Structure defining a network interface.
117 *
118 * (Would like to call this struct ``if'', but C isn't PL/1.)
119 */
120
121struct ifnet {
122	void	*if_softc;		/* pointer to driver state */
123	void	*if_l2com;		/* pointer to protocol bits */
124	struct vnet *if_vnet;		/* pointer to network stack instance */
125	TAILQ_ENTRY(ifnet) if_link; 	/* all struct ifnets are chained */
126	char	if_xname[IFNAMSIZ];	/* external name (name + unit) */
127	const char *if_dname;		/* driver name */
128	int	if_dunit;		/* unit or IF_DUNIT_NONE */
129	u_int	if_refcount;		/* reference count */
130	struct	ifaddrhead if_addrhead;	/* linked list of addresses per if */
131		/*
132		 * if_addrhead is the list of all addresses associated to
133		 * an interface.
134		 * Some code in the kernel assumes that first element
135		 * of the list has type AF_LINK, and contains sockaddr_dl
136		 * addresses which store the link-level address and the name
137		 * of the interface.
138		 * However, access to the AF_LINK address through this
139		 * field is deprecated. Use if_addr or ifaddr_byindex() instead.
140		 */
141	int	if_pcount;		/* number of promiscuous listeners */
142	struct	carp_if *if_carp;	/* carp interface structure */
143	struct	bpf_if *if_bpf;		/* packet filter structure */
144	u_short	if_index;		/* numeric abbreviation for this if  */
145	short	if_index_reserved;	/* spare space to grow if_index */
146	struct  ifvlantrunk *if_vlantrunk; /* pointer to 802.1q data */
147	int	if_flags;		/* up/down, broadcast, etc. */
148	int	if_capabilities;	/* interface features & capabilities */
149	int	if_capenable;		/* enabled features & capabilities */
150	void	*if_linkmib;		/* link-type-specific MIB data */
151	size_t	if_linkmiblen;		/* length of above data */
152	struct	if_data if_data;
153	struct	ifmultihead if_multiaddrs; /* multicast addresses configured */
154	int	if_amcount;		/* number of all-multicast requests */
155/* procedure handles */
156	int	(*if_output)		/* output routine (enqueue) */
157		(struct ifnet *, struct mbuf *, struct sockaddr *,
158		     struct route *);
159	void	(*if_input)		/* input routine (from h/w driver) */
160		(struct ifnet *, struct mbuf *);
161	void	(*if_start)		/* initiate output routine */
162		(struct ifnet *);
163	int	(*if_ioctl)		/* ioctl routine */
164		(struct ifnet *, u_long, caddr_t);
165	void	(*if_init)		/* Init routine */
166		(void *);
167	int	(*if_resolvemulti)	/* validate/resolve multicast */
168		(struct ifnet *, struct sockaddr **, struct sockaddr *);
169	void	(*if_qflush)		/* flush any queues */
170		(struct ifnet *);
171	int	(*if_transmit)		/* initiate output routine */
172		(struct ifnet *, struct mbuf *);
173	void	(*if_reassign)		/* reassign to vnet routine */
174		(struct ifnet *, struct vnet *, char *);
175	struct	vnet *if_home_vnet;	/* where this ifnet originates from */
176	struct	ifaddr	*if_addr;	/* pointer to link-level address */
177	void	*if_llsoftc;		/* link layer softc */
178	int	if_drv_flags;		/* driver-managed status flags */
179	struct  ifaltq if_snd;		/* output queue (includes altq) */
180	const u_int8_t *if_broadcastaddr; /* linklevel broadcast bytestring */
181
182	void	*if_bridge;		/* bridge glue */
183
184	struct	label *if_label;	/* interface MAC label */
185
186	/* these are only used by IPv6 */
187	struct	ifprefixhead if_prefixhead; /* list of prefixes per if */
188	void	*if_afdata[AF_MAX];
189	int	if_afdata_initialized;
190	struct	rwlock if_afdata_lock;
191	struct	task if_linktask;	/* task for link change events */
192	struct	mtx if_addr_mtx;	/* mutex to protect address lists */
193
194	LIST_ENTRY(ifnet) if_clones;	/* interfaces of a cloner */
195	TAILQ_HEAD(, ifg_list) if_groups; /* linked list of groups per if */
196					/* protected by if_addr_mtx */
197	void	*if_pf_kif;
198	void	*if_lagg;		/* lagg glue */
199	char	*if_description;	/* interface description */
200	u_int	if_fib;			/* interface FIB */
201	u_char	if_alloctype;		/* if_type at time of allocation */
202
203	/*
204	 * Spare fields are added so that we can modify sensitive data
205	 * structures without changing the kernel binary interface, and must
206	 * be used with care where binary compatibility is required.
207	 */
208	char	if_cspare[3];
209	int	if_ispare[4];
210	void	*if_pspare[8];		/* 1 netmap, 7 TDB */
211};
212
213typedef void if_init_f_t(void *);
214
215/*
216 * XXX These aliases are terribly dangerous because they could apply
217 * to anything.
218 */
219#define	if_mtu		if_data.ifi_mtu
220#define	if_type		if_data.ifi_type
221#define if_physical	if_data.ifi_physical
222#define	if_addrlen	if_data.ifi_addrlen
223#define	if_hdrlen	if_data.ifi_hdrlen
224#define	if_metric	if_data.ifi_metric
225#define	if_link_state	if_data.ifi_link_state
226#define	if_baudrate	if_data.ifi_baudrate
227#define	if_hwassist	if_data.ifi_hwassist
228#define	if_ipackets	if_data.ifi_ipackets
229#define	if_ierrors	if_data.ifi_ierrors
230#define	if_opackets	if_data.ifi_opackets
231#define	if_oerrors	if_data.ifi_oerrors
232#define	if_collisions	if_data.ifi_collisions
233#define	if_ibytes	if_data.ifi_ibytes
234#define	if_obytes	if_data.ifi_obytes
235#define	if_imcasts	if_data.ifi_imcasts
236#define	if_omcasts	if_data.ifi_omcasts
237#define	if_iqdrops	if_data.ifi_iqdrops
238#define	if_noproto	if_data.ifi_noproto
239#define	if_lastchange	if_data.ifi_lastchange
240
241/* for compatibility with other BSDs */
242#define	if_addrlist	if_addrhead
243#define	if_list		if_link
244#define	if_name(ifp)	((ifp)->if_xname)
245
246/*
247 * Locks for address lists on the network interface.
248 */
249#define	IF_ADDR_LOCK_INIT(if)	mtx_init(&(if)->if_addr_mtx,		\
250				    "if_addr_mtx", NULL, MTX_DEF)
251#define	IF_ADDR_LOCK_DESTROY(if)	mtx_destroy(&(if)->if_addr_mtx)
252#define	IF_ADDR_LOCK(if)	mtx_lock(&(if)->if_addr_mtx)
253#define	IF_ADDR_UNLOCK(if)	mtx_unlock(&(if)->if_addr_mtx)
254#define	IF_ADDR_LOCK_ASSERT(if)	mtx_assert(&(if)->if_addr_mtx, MA_OWNED)
255
256/*
257 * Function variations on locking macros intended to be used by loadable
258 * kernel modules in order to divorce them from the internals of address list
259 * locking.
260 */
261void	if_addr_rlock(struct ifnet *ifp);	/* if_addrhead */
262void	if_addr_runlock(struct ifnet *ifp);	/* if_addrhead */
263void	if_maddr_rlock(struct ifnet *ifp);	/* if_multiaddrs */
264void	if_maddr_runlock(struct ifnet *ifp);	/* if_multiaddrs */
265
266/*
267 * Output queues (ifp->if_snd) and slow device input queues (*ifp->if_slowq)
268 * are queues of messages stored on ifqueue structures
269 * (defined above).  Entries are added to and deleted from these structures
270 * by these macros, which should be called with ipl raised to splimp().
271 */
272#define IF_LOCK(ifq)		mtx_lock(&(ifq)->ifq_mtx)
273#define IF_UNLOCK(ifq)		mtx_unlock(&(ifq)->ifq_mtx)
274#define	IF_LOCK_ASSERT(ifq)	mtx_assert(&(ifq)->ifq_mtx, MA_OWNED)
275#define	_IF_QFULL(ifq)		((ifq)->ifq_len >= (ifq)->ifq_maxlen)
276#define	_IF_DROP(ifq)		((ifq)->ifq_drops++)
277#define	_IF_QLEN(ifq)		((ifq)->ifq_len)
278
279#define	_IF_ENQUEUE(ifq, m) do { 				\
280	(m)->m_nextpkt = NULL;					\
281	if ((ifq)->ifq_tail == NULL) 				\
282		(ifq)->ifq_head = m; 				\
283	else 							\
284		(ifq)->ifq_tail->m_nextpkt = m; 		\
285	(ifq)->ifq_tail = m; 					\
286	(ifq)->ifq_len++; 					\
287} while (0)
288
289#define IF_ENQUEUE(ifq, m) do {					\
290	IF_LOCK(ifq); 						\
291	_IF_ENQUEUE(ifq, m); 					\
292	IF_UNLOCK(ifq); 					\
293} while (0)
294
295#define	_IF_PREPEND(ifq, m) do {				\
296	(m)->m_nextpkt = (ifq)->ifq_head; 			\
297	if ((ifq)->ifq_tail == NULL) 				\
298		(ifq)->ifq_tail = (m); 				\
299	(ifq)->ifq_head = (m); 					\
300	(ifq)->ifq_len++; 					\
301} while (0)
302
303#define IF_PREPEND(ifq, m) do {		 			\
304	IF_LOCK(ifq); 						\
305	_IF_PREPEND(ifq, m); 					\
306	IF_UNLOCK(ifq); 					\
307} while (0)
308
309#define	_IF_DEQUEUE(ifq, m) do { 				\
310	(m) = (ifq)->ifq_head; 					\
311	if (m) { 						\
312		if (((ifq)->ifq_head = (m)->m_nextpkt) == NULL)	\
313			(ifq)->ifq_tail = NULL; 		\
314		(m)->m_nextpkt = NULL; 				\
315		(ifq)->ifq_len--; 				\
316	} 							\
317} while (0)
318
319#define IF_DEQUEUE(ifq, m) do { 				\
320	IF_LOCK(ifq); 						\
321	_IF_DEQUEUE(ifq, m); 					\
322	IF_UNLOCK(ifq); 					\
323} while (0)
324
325#define	_IF_DEQUEUE_ALL(ifq, m) do {				\
326	(m) = (ifq)->ifq_head;					\
327	(ifq)->ifq_head = (ifq)->ifq_tail = NULL;		\
328	(ifq)->ifq_len = 0;					\
329} while (0)
330
331#define	IF_DEQUEUE_ALL(ifq, m) do {				\
332	IF_LOCK(ifq); 						\
333	_IF_DEQUEUE_ALL(ifq, m);				\
334	IF_UNLOCK(ifq); 					\
335} while (0)
336
337#define	_IF_POLL(ifq, m)	((m) = (ifq)->ifq_head)
338#define	IF_POLL(ifq, m)		_IF_POLL(ifq, m)
339
340#define _IF_DRAIN(ifq) do { 					\
341	struct mbuf *m; 					\
342	for (;;) { 						\
343		_IF_DEQUEUE(ifq, m); 				\
344		if (m == NULL) 					\
345			break; 					\
346		m_freem(m); 					\
347	} 							\
348} while (0)
349
350#define IF_DRAIN(ifq) do {					\
351	IF_LOCK(ifq);						\
352	_IF_DRAIN(ifq);						\
353	IF_UNLOCK(ifq);						\
354} while(0)
355
356#ifdef _KERNEL
357/* interface link layer address change event */
358typedef void (*iflladdr_event_handler_t)(void *, struct ifnet *);
359EVENTHANDLER_DECLARE(iflladdr_event, iflladdr_event_handler_t);
360/* interface address change event */
361typedef void (*ifaddr_event_handler_t)(void *, struct ifnet *);
362EVENTHANDLER_DECLARE(ifaddr_event, ifaddr_event_handler_t);
363/* new interface arrival event */
364typedef void (*ifnet_arrival_event_handler_t)(void *, struct ifnet *);
365EVENTHANDLER_DECLARE(ifnet_arrival_event, ifnet_arrival_event_handler_t);
366/* interface departure event */
367typedef void (*ifnet_departure_event_handler_t)(void *, struct ifnet *);
368EVENTHANDLER_DECLARE(ifnet_departure_event, ifnet_departure_event_handler_t);
369/* Interface link state change event */
370typedef void (*ifnet_link_event_handler_t)(void *, struct ifnet *, int);
371EVENTHANDLER_DECLARE(ifnet_link_event, ifnet_link_event_handler_t);
372
373/*
374 * interface groups
375 */
376struct ifg_group {
377	char				 ifg_group[IFNAMSIZ];
378	u_int				 ifg_refcnt;
379	void				*ifg_pf_kif;
380	TAILQ_HEAD(, ifg_member)	 ifg_members;
381	TAILQ_ENTRY(ifg_group)		 ifg_next;
382};
383
384struct ifg_member {
385	TAILQ_ENTRY(ifg_member)	 ifgm_next;
386	struct ifnet		*ifgm_ifp;
387};
388
389struct ifg_list {
390	struct ifg_group	*ifgl_group;
391	TAILQ_ENTRY(ifg_list)	 ifgl_next;
392};
393
394/* group attach event */
395typedef void (*group_attach_event_handler_t)(void *, struct ifg_group *);
396EVENTHANDLER_DECLARE(group_attach_event, group_attach_event_handler_t);
397/* group detach event */
398typedef void (*group_detach_event_handler_t)(void *, struct ifg_group *);
399EVENTHANDLER_DECLARE(group_detach_event, group_detach_event_handler_t);
400/* group change event */
401typedef void (*group_change_event_handler_t)(void *, const char *);
402EVENTHANDLER_DECLARE(group_change_event, group_change_event_handler_t);
403
404#define	IF_AFDATA_LOCK_INIT(ifp)	\
405	rw_init(&(ifp)->if_afdata_lock, "if_afdata")
406
407#define	IF_AFDATA_WLOCK(ifp)	rw_wlock(&(ifp)->if_afdata_lock)
408#define	IF_AFDATA_RLOCK(ifp)	rw_rlock(&(ifp)->if_afdata_lock)
409#define	IF_AFDATA_WUNLOCK(ifp)	rw_wunlock(&(ifp)->if_afdata_lock)
410#define	IF_AFDATA_RUNLOCK(ifp)	rw_runlock(&(ifp)->if_afdata_lock)
411#define	IF_AFDATA_LOCK(ifp)	IF_AFDATA_WLOCK(ifp)
412#define	IF_AFDATA_UNLOCK(ifp)	IF_AFDATA_WUNLOCK(ifp)
413#define	IF_AFDATA_TRYLOCK(ifp)	rw_try_wlock(&(ifp)->if_afdata_lock)
414#define	IF_AFDATA_DESTROY(ifp)	rw_destroy(&(ifp)->if_afdata_lock)
415
416#define	IF_AFDATA_LOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_LOCKED)
417#define	IF_AFDATA_UNLOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_UNLOCKED)
418
419int	if_handoff(struct ifqueue *ifq, struct mbuf *m, struct ifnet *ifp,
420	    int adjust);
421#define	IF_HANDOFF(ifq, m, ifp)			\
422	if_handoff((struct ifqueue *)ifq, m, ifp, 0)
423#define	IF_HANDOFF_ADJ(ifq, m, ifp, adj)	\
424	if_handoff((struct ifqueue *)ifq, m, ifp, adj)
425
426void	if_start(struct ifnet *);
427
428#define	IFQ_ENQUEUE(ifq, m, err)					\
429do {									\
430	IF_LOCK(ifq);							\
431	if (ALTQ_IS_ENABLED(ifq))					\
432		ALTQ_ENQUEUE(ifq, m, NULL, err);			\
433	else {								\
434		if (_IF_QFULL(ifq)) {					\
435			m_freem(m);					\
436			(err) = ENOBUFS;				\
437		} else {						\
438			_IF_ENQUEUE(ifq, m);				\
439			(err) = 0;					\
440		}							\
441	}								\
442	if (err)							\
443		(ifq)->ifq_drops++;					\
444	IF_UNLOCK(ifq);							\
445} while (0)
446
447#define	IFQ_DEQUEUE_NOLOCK(ifq, m)					\
448do {									\
449	if (TBR_IS_ENABLED(ifq))					\
450		(m) = tbr_dequeue_ptr(ifq, ALTDQ_REMOVE);		\
451	else if (ALTQ_IS_ENABLED(ifq))					\
452		ALTQ_DEQUEUE(ifq, m);					\
453	else								\
454		_IF_DEQUEUE(ifq, m);					\
455} while (0)
456
457#define	IFQ_DEQUEUE(ifq, m)						\
458do {									\
459	IF_LOCK(ifq);							\
460	IFQ_DEQUEUE_NOLOCK(ifq, m);					\
461	IF_UNLOCK(ifq);							\
462} while (0)
463
464#define	IFQ_POLL_NOLOCK(ifq, m)						\
465do {									\
466	if (TBR_IS_ENABLED(ifq))					\
467		(m) = tbr_dequeue_ptr(ifq, ALTDQ_POLL);			\
468	else if (ALTQ_IS_ENABLED(ifq))					\
469		ALTQ_POLL(ifq, m);					\
470	else								\
471		_IF_POLL(ifq, m);					\
472} while (0)
473
474#define	IFQ_POLL(ifq, m)						\
475do {									\
476	IF_LOCK(ifq);							\
477	IFQ_POLL_NOLOCK(ifq, m);					\
478	IF_UNLOCK(ifq);							\
479} while (0)
480
481#define	IFQ_PURGE_NOLOCK(ifq)						\
482do {									\
483	if (ALTQ_IS_ENABLED(ifq)) {					\
484		ALTQ_PURGE(ifq);					\
485	} else								\
486		_IF_DRAIN(ifq);						\
487} while (0)
488
489#define	IFQ_PURGE(ifq)							\
490do {									\
491	IF_LOCK(ifq);							\
492	IFQ_PURGE_NOLOCK(ifq);						\
493	IF_UNLOCK(ifq);							\
494} while (0)
495
496#define	IFQ_SET_READY(ifq)						\
497	do { ((ifq)->altq_flags |= ALTQF_READY); } while (0)
498
499#define	IFQ_LOCK(ifq)			IF_LOCK(ifq)
500#define	IFQ_UNLOCK(ifq)			IF_UNLOCK(ifq)
501#define	IFQ_LOCK_ASSERT(ifq)		IF_LOCK_ASSERT(ifq)
502#define	IFQ_IS_EMPTY(ifq)		((ifq)->ifq_len == 0)
503#define	IFQ_INC_LEN(ifq)		((ifq)->ifq_len++)
504#define	IFQ_DEC_LEN(ifq)		(--(ifq)->ifq_len)
505#define	IFQ_INC_DROPS(ifq)		((ifq)->ifq_drops++)
506#define	IFQ_SET_MAXLEN(ifq, len)	((ifq)->ifq_maxlen = (len))
507
508/*
509 * The IFF_DRV_OACTIVE test should really occur in the device driver, not in
510 * the handoff logic, as that flag is locked by the device driver.
511 */
512#define	IFQ_HANDOFF_ADJ(ifp, m, adj, err)				\
513do {									\
514	int len;							\
515	short mflags;							\
516									\
517	len = (m)->m_pkthdr.len;					\
518	mflags = (m)->m_flags;						\
519	IFQ_ENQUEUE(&(ifp)->if_snd, m, err);				\
520	if ((err) == 0) {						\
521		(ifp)->if_obytes += len + (adj);			\
522		if (mflags & M_MCAST)					\
523			(ifp)->if_omcasts++;				\
524		if (((ifp)->if_drv_flags & IFF_DRV_OACTIVE) == 0)	\
525			if_start(ifp);					\
526	}								\
527} while (0)
528
529#define	IFQ_HANDOFF(ifp, m, err)					\
530	IFQ_HANDOFF_ADJ(ifp, m, 0, err)
531
532#define	IFQ_DRV_DEQUEUE(ifq, m)						\
533do {									\
534	(m) = (ifq)->ifq_drv_head;					\
535	if (m) {							\
536		if (((ifq)->ifq_drv_head = (m)->m_nextpkt) == NULL)	\
537			(ifq)->ifq_drv_tail = NULL;			\
538		(m)->m_nextpkt = NULL;					\
539		(ifq)->ifq_drv_len--;					\
540	} else {							\
541		IFQ_LOCK(ifq);						\
542		IFQ_DEQUEUE_NOLOCK(ifq, m);				\
543		while ((ifq)->ifq_drv_len < (ifq)->ifq_drv_maxlen) {	\
544			struct mbuf *m0;				\
545			IFQ_DEQUEUE_NOLOCK(ifq, m0);			\
546			if (m0 == NULL)					\
547				break;					\
548			m0->m_nextpkt = NULL;				\
549			if ((ifq)->ifq_drv_tail == NULL)		\
550				(ifq)->ifq_drv_head = m0;		\
551			else						\
552				(ifq)->ifq_drv_tail->m_nextpkt = m0;	\
553			(ifq)->ifq_drv_tail = m0;			\
554			(ifq)->ifq_drv_len++;				\
555		}							\
556		IFQ_UNLOCK(ifq);					\
557	}								\
558} while (0)
559
560#define	IFQ_DRV_PREPEND(ifq, m)						\
561do {									\
562	(m)->m_nextpkt = (ifq)->ifq_drv_head;				\
563	if ((ifq)->ifq_drv_tail == NULL)				\
564		(ifq)->ifq_drv_tail = (m);				\
565	(ifq)->ifq_drv_head = (m);					\
566	(ifq)->ifq_drv_len++;						\
567} while (0)
568
569#define	IFQ_DRV_IS_EMPTY(ifq)						\
570	(((ifq)->ifq_drv_len == 0) && ((ifq)->ifq_len == 0))
571
572#define	IFQ_DRV_PURGE(ifq)						\
573do {									\
574	struct mbuf *m, *n = (ifq)->ifq_drv_head;			\
575	while((m = n) != NULL) {					\
576		n = m->m_nextpkt;					\
577		m_freem(m);						\
578	}								\
579	(ifq)->ifq_drv_head = (ifq)->ifq_drv_tail = NULL;		\
580	(ifq)->ifq_drv_len = 0;						\
581	IFQ_PURGE(ifq);							\
582} while (0)
583
584#ifdef _KERNEL
585static __inline void
586drbr_stats_update(struct ifnet *ifp, int len, int mflags)
587{
588#ifndef NO_SLOW_STATS
589	ifp->if_obytes += len;
590	if (mflags & M_MCAST)
591		ifp->if_omcasts++;
592#endif
593}
594
595static __inline int
596drbr_enqueue(struct ifnet *ifp, struct buf_ring *br, struct mbuf *m)
597{
598	int error = 0;
599	int len = m->m_pkthdr.len;
600	int mflags = m->m_flags;
601
602#ifdef ALTQ
603	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
604		IFQ_ENQUEUE(&ifp->if_snd, m, error);
605		return (error);
606	}
607#endif
608	if ((error = buf_ring_enqueue_bytes(br, m, len)) == ENOBUFS) {
609		br->br_drops++;
610		m_freem(m);
611	} else
612		drbr_stats_update(ifp, len, mflags);
613
614	return (error);
615}
616
617static __inline void
618drbr_flush(struct ifnet *ifp, struct buf_ring *br)
619{
620	struct mbuf *m;
621
622#ifdef ALTQ
623	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd))
624		IFQ_PURGE(&ifp->if_snd);
625#endif
626	while ((m = buf_ring_dequeue_sc(br)) != NULL)
627		m_freem(m);
628}
629
630static __inline void
631drbr_free(struct buf_ring *br, struct malloc_type *type)
632{
633
634	drbr_flush(NULL, br);
635	buf_ring_free(br, type);
636}
637
638static __inline struct mbuf *
639drbr_dequeue(struct ifnet *ifp, struct buf_ring *br)
640{
641#ifdef ALTQ
642	struct mbuf *m;
643
644	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
645		IFQ_DEQUEUE(&ifp->if_snd, m);
646		return (m);
647	}
648#endif
649	return (buf_ring_dequeue_sc(br));
650}
651
652static __inline struct mbuf *
653drbr_dequeue_cond(struct ifnet *ifp, struct buf_ring *br,
654    int (*func) (struct mbuf *, void *), void *arg)
655{
656	struct mbuf *m;
657#ifdef ALTQ
658	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
659		IFQ_LOCK(&ifp->if_snd);
660		IFQ_POLL_NOLOCK(&ifp->if_snd, m);
661		if (m != NULL && func(m, arg) == 0) {
662			IFQ_UNLOCK(&ifp->if_snd);
663			return (NULL);
664		}
665		IFQ_DEQUEUE_NOLOCK(&ifp->if_snd, m);
666		IFQ_UNLOCK(&ifp->if_snd);
667		return (m);
668	}
669#endif
670	m = buf_ring_peek(br);
671	if (m == NULL || func(m, arg) == 0)
672		return (NULL);
673
674	return (buf_ring_dequeue_sc(br));
675}
676
677static __inline int
678drbr_empty(struct ifnet *ifp, struct buf_ring *br)
679{
680#ifdef ALTQ
681	if (ALTQ_IS_ENABLED(&ifp->if_snd))
682		return (IFQ_IS_EMPTY(&ifp->if_snd));
683#endif
684	return (buf_ring_empty(br));
685}
686
687static __inline int
688drbr_needs_enqueue(struct ifnet *ifp, struct buf_ring *br)
689{
690#ifdef ALTQ
691	if (ALTQ_IS_ENABLED(&ifp->if_snd))
692		return (1);
693#endif
694	return (!buf_ring_empty(br));
695}
696
697static __inline int
698drbr_inuse(struct ifnet *ifp, struct buf_ring *br)
699{
700#ifdef ALTQ
701	if (ALTQ_IS_ENABLED(&ifp->if_snd))
702		return (ifp->if_snd.ifq_len);
703#endif
704	return (buf_ring_count(br));
705}
706#endif
707/*
708 * 72 was chosen below because it is the size of a TCP/IP
709 * header (40) + the minimum mss (32).
710 */
711#define	IF_MINMTU	72
712#define	IF_MAXMTU	65535
713
714#endif /* _KERNEL */
715
716/*
717 * The ifaddr structure contains information about one address
718 * of an interface.  They are maintained by the different address families,
719 * are allocated and attached when an address is set, and are linked
720 * together so all addresses for an interface can be located.
721 *
722 * NOTE: a 'struct ifaddr' is always at the beginning of a larger
723 * chunk of malloc'ed memory, where we store the three addresses
724 * (ifa_addr, ifa_dstaddr and ifa_netmask) referenced here.
725 */
726struct ifaddr {
727	struct	sockaddr *ifa_addr;	/* address of interface */
728	struct	sockaddr *ifa_dstaddr;	/* other end of p-to-p link */
729#define	ifa_broadaddr	ifa_dstaddr	/* broadcast address interface */
730	struct	sockaddr *ifa_netmask;	/* used to determine subnet */
731	struct	if_data if_data;	/* not all members are meaningful */
732	struct	ifnet *ifa_ifp;		/* back-pointer to interface */
733	struct	carp_softc *ifa_carp;	/* pointer to CARP data */
734	TAILQ_ENTRY(ifaddr) ifa_link;	/* queue macro glue */
735	void	(*ifa_rtrequest)	/* check or clean routes (+ or -)'d */
736		(int, struct rtentry *, struct rt_addrinfo *);
737	u_short	ifa_flags;		/* mostly rt_flags for cloning */
738	u_int	ifa_refcnt;		/* references to this structure */
739	int	ifa_metric;		/* cost of going out this interface */
740	int (*ifa_claim_addr)		/* check if an addr goes to this if */
741		(struct ifaddr *, struct sockaddr *);
742	struct mtx ifa_mtx;
743};
744#define	IFA_ROUTE	RTF_UP		/* route installed */
745#define IFA_RTSELF	RTF_HOST	/* loopback route to self installed */
746
747/* for compatibility with other BSDs */
748#define	ifa_list	ifa_link
749
750#ifdef _KERNEL
751#define	IFA_LOCK(ifa)		mtx_lock(&(ifa)->ifa_mtx)
752#define	IFA_UNLOCK(ifa)		mtx_unlock(&(ifa)->ifa_mtx)
753
754void	ifa_free(struct ifaddr *ifa);
755void	ifa_init(struct ifaddr *ifa);
756void	ifa_ref(struct ifaddr *ifa);
757#endif
758
759/*
760 * The prefix structure contains information about one prefix
761 * of an interface.  They are maintained by the different address families,
762 * are allocated and attached when a prefix or an address is set,
763 * and are linked together so all prefixes for an interface can be located.
764 */
765struct ifprefix {
766	struct	sockaddr *ifpr_prefix;	/* prefix of interface */
767	struct	ifnet *ifpr_ifp;	/* back-pointer to interface */
768	TAILQ_ENTRY(ifprefix) ifpr_list; /* queue macro glue */
769	u_char	ifpr_plen;		/* prefix length in bits */
770	u_char	ifpr_type;		/* protocol dependent prefix type */
771};
772
773/*
774 * Multicast address structure.  This is analogous to the ifaddr
775 * structure except that it keeps track of multicast addresses.
776 */
777struct ifmultiaddr {
778	TAILQ_ENTRY(ifmultiaddr) ifma_link; /* queue macro glue */
779	struct	sockaddr *ifma_addr; 	/* address this membership is for */
780	struct	sockaddr *ifma_lladdr;	/* link-layer translation, if any */
781	struct	ifnet *ifma_ifp;	/* back-pointer to interface */
782	u_int	ifma_refcount;		/* reference count */
783	void	*ifma_protospec;	/* protocol-specific state, if any */
784	struct	ifmultiaddr *ifma_llifma; /* pointer to ifma for ifma_lladdr */
785};
786
787#ifdef _KERNEL
788
789extern	struct rwlock ifnet_rwlock;
790extern	struct sx ifnet_sxlock;
791
792#define	IFNET_LOCK_INIT() do {						\
793	rw_init_flags(&ifnet_rwlock, "ifnet_rw",  RW_RECURSE);		\
794	sx_init_flags(&ifnet_sxlock, "ifnet_sx",  SX_RECURSE);		\
795} while(0)
796
797#define	IFNET_WLOCK() do {						\
798	sx_xlock(&ifnet_sxlock);					\
799	rw_wlock(&ifnet_rwlock);					\
800} while (0)
801
802#define	IFNET_WUNLOCK() do {						\
803	rw_wunlock(&ifnet_rwlock);					\
804	sx_xunlock(&ifnet_sxlock);					\
805} while (0)
806
807/*
808 * To assert the ifnet lock, you must know not only whether it's for read or
809 * write, but also whether it was acquired with sleep support or not.
810 */
811#define	IFNET_RLOCK_ASSERT()		sx_assert(&ifnet_sxlock, SA_SLOCKED)
812#define	IFNET_RLOCK_NOSLEEP_ASSERT()	rw_assert(&ifnet_rwlock, RA_RLOCKED)
813#define	IFNET_WLOCK_ASSERT() do {					\
814	sx_assert(&ifnet_sxlock, SA_XLOCKED);				\
815	rw_assert(&ifnet_rwlock, RA_WLOCKED);				\
816} while (0)
817
818#define	IFNET_RLOCK()		sx_slock(&ifnet_sxlock)
819#define	IFNET_RLOCK_NOSLEEP()	rw_rlock(&ifnet_rwlock)
820#define	IFNET_RUNLOCK()		sx_sunlock(&ifnet_sxlock)
821#define	IFNET_RUNLOCK_NOSLEEP()	rw_runlock(&ifnet_rwlock)
822
823/*
824 * Look up an ifnet given its index; the _ref variant also acquires a
825 * reference that must be freed using if_rele().  It is almost always a bug
826 * to call ifnet_byindex() instead if ifnet_byindex_ref().
827 */
828struct ifnet	*ifnet_byindex(u_short idx);
829struct ifnet	*ifnet_byindex_locked(u_short idx);
830struct ifnet	*ifnet_byindex_ref(u_short idx);
831
832/*
833 * Given the index, ifaddr_byindex() returns the one and only
834 * link-level ifaddr for the interface. You are not supposed to use
835 * it to traverse the list of addresses associated to the interface.
836 */
837struct ifaddr	*ifaddr_byindex(u_short idx);
838
839VNET_DECLARE(struct ifnethead, ifnet);
840VNET_DECLARE(struct ifgrouphead, ifg_head);
841VNET_DECLARE(int, if_index);
842VNET_DECLARE(struct ifnet *, loif);	/* first loopback interface */
843VNET_DECLARE(int, useloopback);
844
845#define	V_ifnet		VNET(ifnet)
846#define	V_ifg_head	VNET(ifg_head)
847#define	V_if_index	VNET(if_index)
848#define	V_loif		VNET(loif)
849#define	V_useloopback	VNET(useloopback)
850
851extern	int ifqmaxlen;
852
853int	if_addgroup(struct ifnet *, const char *);
854int	if_delgroup(struct ifnet *, const char *);
855int	if_addmulti(struct ifnet *, struct sockaddr *, struct ifmultiaddr **);
856int	if_allmulti(struct ifnet *, int);
857struct	ifnet* if_alloc(u_char);
858void	if_attach(struct ifnet *);
859void	if_dead(struct ifnet *);
860int	if_delmulti(struct ifnet *, struct sockaddr *);
861void	if_delmulti_ifma(struct ifmultiaddr *);
862void	if_detach(struct ifnet *);
863void	if_vmove(struct ifnet *, struct vnet *);
864void	if_purgeaddrs(struct ifnet *);
865void	if_delallmulti(struct ifnet *);
866void	if_down(struct ifnet *);
867struct ifmultiaddr *
868	if_findmulti(struct ifnet *, struct sockaddr *);
869void	if_free(struct ifnet *);
870void	if_initname(struct ifnet *, const char *, int);
871void	if_link_state_change(struct ifnet *, int);
872int	if_printf(struct ifnet *, const char *, ...) __printflike(2, 3);
873void	if_qflush(struct ifnet *);
874void	if_ref(struct ifnet *);
875void	if_rele(struct ifnet *);
876int	if_setlladdr(struct ifnet *, const u_char *, int);
877void	if_up(struct ifnet *);
878int	ifioctl(struct socket *, u_long, caddr_t, struct thread *);
879int	ifpromisc(struct ifnet *, int);
880struct	ifnet *ifunit(const char *);
881struct	ifnet *ifunit_ref(const char *);
882
883void	ifq_init(struct ifaltq *, struct ifnet *ifp);
884void	ifq_delete(struct ifaltq *);
885
886int	ifa_add_loopback_route(struct ifaddr *, struct sockaddr *);
887int	ifa_del_loopback_route(struct ifaddr *, struct sockaddr *);
888
889struct	ifaddr *ifa_ifwithaddr(struct sockaddr *);
890int		ifa_ifwithaddr_check(struct sockaddr *);
891struct	ifaddr *ifa_ifwithbroadaddr(struct sockaddr *);
892struct	ifaddr *ifa_ifwithdstaddr(struct sockaddr *);
893struct	ifaddr *ifa_ifwithnet(struct sockaddr *, int);
894struct	ifaddr *ifa_ifwithroute(int, struct sockaddr *, struct sockaddr *);
895struct	ifaddr *ifa_ifwithroute_fib(int, struct sockaddr *, struct sockaddr *, u_int);
896
897struct	ifaddr *ifaof_ifpforaddr(struct sockaddr *, struct ifnet *);
898
899int	if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen);
900
901typedef	void *if_com_alloc_t(u_char type, struct ifnet *ifp);
902typedef	void if_com_free_t(void *com, u_char type);
903void	if_register_com_alloc(u_char type, if_com_alloc_t *a, if_com_free_t *f);
904void	if_deregister_com_alloc(u_char type);
905
906#define IF_LLADDR(ifp)							\
907    LLADDR((struct sockaddr_dl *)((ifp)->if_addr->ifa_addr))
908
909#ifdef DEVICE_POLLING
910enum poll_cmd {	POLL_ONLY, POLL_AND_CHECK_STATUS };
911
912typedef	int poll_handler_t(struct ifnet *ifp, enum poll_cmd cmd, int count);
913int    ether_poll_register(poll_handler_t *h, struct ifnet *ifp);
914int    ether_poll_deregister(struct ifnet *ifp);
915#endif /* DEVICE_POLLING */
916
917#endif /* _KERNEL */
918
919#endif /* !_NET_IF_VAR_H_ */
920