ifq.h revision 208553
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 208553 2010-05-25 20:42:35Z qingli $
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  ifvlantrunk;
73struct	route;
74struct	vnet;
75#endif
76
77#include <sys/queue.h>		/* get TAILQ macros */
78
79#ifdef _KERNEL
80#include <sys/mbuf.h>
81#include <sys/eventhandler.h>
82#include <sys/buf_ring.h>
83#include <net/vnet.h>
84#endif /* _KERNEL */
85#include <sys/lock.h>		/* XXX */
86#include <sys/mutex.h>		/* XXX */
87#include <sys/rwlock.h>		/* XXX */
88#include <sys/sx.h>		/* XXX */
89#include <sys/event.h>		/* XXX */
90#include <sys/_task.h>
91
92#define	IF_DUNIT_NONE	-1
93
94#include <altq/if_altq.h>
95
96TAILQ_HEAD(ifnethead, ifnet);	/* we use TAILQs so that the order of */
97TAILQ_HEAD(ifaddrhead, ifaddr);	/* instantiation is preserved in the list */
98TAILQ_HEAD(ifprefixhead, ifprefix);
99TAILQ_HEAD(ifmultihead, ifmultiaddr);
100TAILQ_HEAD(ifgrouphead, ifg_group);
101
102/*
103 * Structure defining a queue for a network interface.
104 */
105struct	ifqueue {
106	struct	mbuf *ifq_head;
107	struct	mbuf *ifq_tail;
108	int	ifq_len;
109	int	ifq_maxlen;
110	int	ifq_drops;
111	struct	mtx ifq_mtx;
112};
113
114/*
115 * Structure defining a network interface.
116 *
117 * (Would like to call this struct ``if'', but C isn't PL/1.)
118 */
119
120struct ifnet {
121	void	*if_softc;		/* pointer to driver state */
122	void	*if_l2com;		/* pointer to protocol bits */
123	struct vnet *if_vnet;		/* pointer to network stack instance */
124	TAILQ_ENTRY(ifnet) if_link; 	/* all struct ifnets are chained */
125	char	if_xname[IFNAMSIZ];	/* external name (name + unit) */
126	const char *if_dname;		/* driver name */
127	int	if_dunit;		/* unit or IF_DUNIT_NONE */
128	u_int	if_refcount;		/* reference count */
129	struct	ifaddrhead if_addrhead;	/* linked list of addresses per if */
130		/*
131		 * if_addrhead is the list of all addresses associated to
132		 * an interface.
133		 * Some code in the kernel assumes that first element
134		 * of the list has type AF_LINK, and contains sockaddr_dl
135		 * addresses which store the link-level address and the name
136		 * of the interface.
137		 * However, access to the AF_LINK address through this
138		 * field is deprecated. Use if_addr or ifaddr_byindex() instead.
139		 */
140	int	if_pcount;		/* number of promiscuous listeners */
141	struct	carp_if *if_carp;	/* carp interface structure */
142	struct	bpf_if *if_bpf;		/* packet filter structure */
143	u_short	if_index;		/* numeric abbreviation for this if  */
144	short	if_index_reserved;	/* spare space to grow if_index */
145	struct  ifvlantrunk *if_vlantrunk; /* pointer to 802.1q data */
146	int	if_flags;		/* up/down, broadcast, etc. */
147	int	if_capabilities;	/* interface features & capabilities */
148	int	if_capenable;		/* enabled features & capabilities */
149	void	*if_linkmib;		/* link-type-specific MIB data */
150	size_t	if_linkmiblen;		/* length of above data */
151	struct	if_data if_data;
152	struct	ifmultihead if_multiaddrs; /* multicast addresses configured */
153	int	if_amcount;		/* number of all-multicast requests */
154/* procedure handles */
155	int	(*if_output)		/* output routine (enqueue) */
156		(struct ifnet *, struct mbuf *, struct sockaddr *,
157		     struct route *);
158	void	(*if_input)		/* input routine (from h/w driver) */
159		(struct ifnet *, struct mbuf *);
160	void	(*if_start)		/* initiate output routine */
161		(struct ifnet *);
162	int	(*if_ioctl)		/* ioctl routine */
163		(struct ifnet *, u_long, caddr_t);
164	void	(*if_init)		/* Init routine */
165		(void *);
166	int	(*if_resolvemulti)	/* validate/resolve multicast */
167		(struct ifnet *, struct sockaddr **, struct sockaddr *);
168	void	(*if_qflush)		/* flush any queues */
169		(struct ifnet *);
170	int	(*if_transmit)		/* initiate output routine */
171		(struct ifnet *, struct mbuf *);
172	void	(*if_reassign)		/* reassign to vnet routine */
173		(struct ifnet *, struct vnet *, char *);
174	struct	vnet *if_home_vnet;	/* where this ifnet originates from */
175	struct	ifaddr	*if_addr;	/* pointer to link-level address */
176	void	*if_llsoftc;		/* link layer softc */
177	int	if_drv_flags;		/* driver-managed status flags */
178	struct  ifaltq if_snd;		/* output queue (includes altq) */
179	const u_int8_t *if_broadcastaddr; /* linklevel broadcast bytestring */
180
181	void	*if_bridge;		/* bridge glue */
182
183	struct	label *if_label;	/* interface MAC label */
184
185	/* these are only used by IPv6 */
186	struct	ifprefixhead if_prefixhead; /* list of prefixes per if */
187	void	*if_afdata[AF_MAX];
188	int	if_afdata_initialized;
189	struct	rwlock if_afdata_lock;
190	struct	task if_linktask;	/* task for link change events */
191	struct	mtx if_addr_mtx;	/* mutex to protect address lists */
192
193	LIST_ENTRY(ifnet) if_clones;	/* interfaces of a cloner */
194	TAILQ_HEAD(, ifg_list) if_groups; /* linked list of groups per if */
195					/* protected by if_addr_mtx */
196	void	*if_pf_kif;
197	void	*if_lagg;		/* lagg glue */
198	u_char	 if_alloctype;		/* if_type at time of allocation */
199
200	/*
201	 * Spare fields are added so that we can modify sensitive data
202	 * structures without changing the kernel binary interface, and must
203	 * be used with care where binary compatibility is required.
204	 */
205	char	 if_cspare[3];
206	char	*if_description;	/* interface description */
207	void	*if_pspare[7];
208	int	if_ispare[4];
209};
210
211typedef void if_init_f_t(void *);
212
213/*
214 * XXX These aliases are terribly dangerous because they could apply
215 * to anything.
216 */
217#define	if_mtu		if_data.ifi_mtu
218#define	if_type		if_data.ifi_type
219#define if_physical	if_data.ifi_physical
220#define	if_addrlen	if_data.ifi_addrlen
221#define	if_hdrlen	if_data.ifi_hdrlen
222#define	if_metric	if_data.ifi_metric
223#define	if_link_state	if_data.ifi_link_state
224#define	if_baudrate	if_data.ifi_baudrate
225#define	if_hwassist	if_data.ifi_hwassist
226#define	if_ipackets	if_data.ifi_ipackets
227#define	if_ierrors	if_data.ifi_ierrors
228#define	if_opackets	if_data.ifi_opackets
229#define	if_oerrors	if_data.ifi_oerrors
230#define	if_collisions	if_data.ifi_collisions
231#define	if_ibytes	if_data.ifi_ibytes
232#define	if_obytes	if_data.ifi_obytes
233#define	if_imcasts	if_data.ifi_imcasts
234#define	if_omcasts	if_data.ifi_omcasts
235#define	if_iqdrops	if_data.ifi_iqdrops
236#define	if_noproto	if_data.ifi_noproto
237#define	if_lastchange	if_data.ifi_lastchange
238
239/* for compatibility with other BSDs */
240#define	if_addrlist	if_addrhead
241#define	if_list		if_link
242#define	if_name(ifp)	((ifp)->if_xname)
243
244/*
245 * Locks for address lists on the network interface.
246 */
247#define	IF_ADDR_LOCK_INIT(if)	mtx_init(&(if)->if_addr_mtx,		\
248				    "if_addr_mtx", NULL, MTX_DEF)
249#define	IF_ADDR_LOCK_DESTROY(if)	mtx_destroy(&(if)->if_addr_mtx)
250#define	IF_ADDR_LOCK(if)	mtx_lock(&(if)->if_addr_mtx)
251#define	IF_ADDR_UNLOCK(if)	mtx_unlock(&(if)->if_addr_mtx)
252#define	IF_ADDR_LOCK_ASSERT(if)	mtx_assert(&(if)->if_addr_mtx, MA_OWNED)
253
254/*
255 * Function variations on locking macros intended to be used by loadable
256 * kernel modules in order to divorce them from the internals of address list
257 * locking.
258 */
259void	if_addr_rlock(struct ifnet *ifp);	/* if_addrhead */
260void	if_addr_runlock(struct ifnet *ifp);	/* if_addrhead */
261void	if_maddr_rlock(struct ifnet *ifp);	/* if_multiaddrs */
262void	if_maddr_runlock(struct ifnet *ifp);	/* if_multiaddrs */
263
264/*
265 * Output queues (ifp->if_snd) and slow device input queues (*ifp->if_slowq)
266 * are queues of messages stored on ifqueue structures
267 * (defined above).  Entries are added to and deleted from these structures
268 * by these macros, which should be called with ipl raised to splimp().
269 */
270#define IF_LOCK(ifq)		mtx_lock(&(ifq)->ifq_mtx)
271#define IF_UNLOCK(ifq)		mtx_unlock(&(ifq)->ifq_mtx)
272#define	IF_LOCK_ASSERT(ifq)	mtx_assert(&(ifq)->ifq_mtx, MA_OWNED)
273#define	_IF_QFULL(ifq)		((ifq)->ifq_len >= (ifq)->ifq_maxlen)
274#define	_IF_DROP(ifq)		((ifq)->ifq_drops++)
275#define	_IF_QLEN(ifq)		((ifq)->ifq_len)
276
277#define	_IF_ENQUEUE(ifq, m) do { 				\
278	(m)->m_nextpkt = NULL;					\
279	if ((ifq)->ifq_tail == NULL) 				\
280		(ifq)->ifq_head = m; 				\
281	else 							\
282		(ifq)->ifq_tail->m_nextpkt = m; 		\
283	(ifq)->ifq_tail = m; 					\
284	(ifq)->ifq_len++; 					\
285} while (0)
286
287#define IF_ENQUEUE(ifq, m) do {					\
288	IF_LOCK(ifq); 						\
289	_IF_ENQUEUE(ifq, m); 					\
290	IF_UNLOCK(ifq); 					\
291} while (0)
292
293#define	_IF_PREPEND(ifq, m) do {				\
294	(m)->m_nextpkt = (ifq)->ifq_head; 			\
295	if ((ifq)->ifq_tail == NULL) 				\
296		(ifq)->ifq_tail = (m); 				\
297	(ifq)->ifq_head = (m); 					\
298	(ifq)->ifq_len++; 					\
299} while (0)
300
301#define IF_PREPEND(ifq, m) do {		 			\
302	IF_LOCK(ifq); 						\
303	_IF_PREPEND(ifq, m); 					\
304	IF_UNLOCK(ifq); 					\
305} while (0)
306
307#define	_IF_DEQUEUE(ifq, m) do { 				\
308	(m) = (ifq)->ifq_head; 					\
309	if (m) { 						\
310		if (((ifq)->ifq_head = (m)->m_nextpkt) == NULL)	\
311			(ifq)->ifq_tail = NULL; 		\
312		(m)->m_nextpkt = NULL; 				\
313		(ifq)->ifq_len--; 				\
314	} 							\
315} while (0)
316
317#define IF_DEQUEUE(ifq, m) do { 				\
318	IF_LOCK(ifq); 						\
319	_IF_DEQUEUE(ifq, m); 					\
320	IF_UNLOCK(ifq); 					\
321} while (0)
322
323#define	_IF_POLL(ifq, m)	((m) = (ifq)->ifq_head)
324#define	IF_POLL(ifq, m)		_IF_POLL(ifq, m)
325
326#define _IF_DRAIN(ifq) do { 					\
327	struct mbuf *m; 					\
328	for (;;) { 						\
329		_IF_DEQUEUE(ifq, m); 				\
330		if (m == NULL) 					\
331			break; 					\
332		m_freem(m); 					\
333	} 							\
334} while (0)
335
336#define IF_DRAIN(ifq) do {					\
337	IF_LOCK(ifq);						\
338	_IF_DRAIN(ifq);						\
339	IF_UNLOCK(ifq);						\
340} while(0)
341
342#ifdef _KERNEL
343/* interface link layer address change event */
344typedef void (*iflladdr_event_handler_t)(void *, struct ifnet *);
345EVENTHANDLER_DECLARE(iflladdr_event, iflladdr_event_handler_t);
346/* interface address change event */
347typedef void (*ifaddr_event_handler_t)(void *, struct ifnet *);
348EVENTHANDLER_DECLARE(ifaddr_event, ifaddr_event_handler_t);
349/* new interface arrival event */
350typedef void (*ifnet_arrival_event_handler_t)(void *, struct ifnet *);
351EVENTHANDLER_DECLARE(ifnet_arrival_event, ifnet_arrival_event_handler_t);
352/* interface departure event */
353typedef void (*ifnet_departure_event_handler_t)(void *, struct ifnet *);
354EVENTHANDLER_DECLARE(ifnet_departure_event, ifnet_departure_event_handler_t);
355
356/*
357 * interface groups
358 */
359struct ifg_group {
360	char				 ifg_group[IFNAMSIZ];
361	u_int				 ifg_refcnt;
362	void				*ifg_pf_kif;
363	TAILQ_HEAD(, ifg_member)	 ifg_members;
364	TAILQ_ENTRY(ifg_group)		 ifg_next;
365};
366
367struct ifg_member {
368	TAILQ_ENTRY(ifg_member)	 ifgm_next;
369	struct ifnet		*ifgm_ifp;
370};
371
372struct ifg_list {
373	struct ifg_group	*ifgl_group;
374	TAILQ_ENTRY(ifg_list)	 ifgl_next;
375};
376
377/* group attach event */
378typedef void (*group_attach_event_handler_t)(void *, struct ifg_group *);
379EVENTHANDLER_DECLARE(group_attach_event, group_attach_event_handler_t);
380/* group detach event */
381typedef void (*group_detach_event_handler_t)(void *, struct ifg_group *);
382EVENTHANDLER_DECLARE(group_detach_event, group_detach_event_handler_t);
383/* group change event */
384typedef void (*group_change_event_handler_t)(void *, const char *);
385EVENTHANDLER_DECLARE(group_change_event, group_change_event_handler_t);
386
387#define	IF_AFDATA_LOCK_INIT(ifp)	\
388	rw_init(&(ifp)->if_afdata_lock, "if_afdata")
389
390#define	IF_AFDATA_WLOCK(ifp)	rw_wlock(&(ifp)->if_afdata_lock)
391#define	IF_AFDATA_RLOCK(ifp)	rw_rlock(&(ifp)->if_afdata_lock)
392#define	IF_AFDATA_WUNLOCK(ifp)	rw_wunlock(&(ifp)->if_afdata_lock)
393#define	IF_AFDATA_RUNLOCK(ifp)	rw_runlock(&(ifp)->if_afdata_lock)
394#define	IF_AFDATA_LOCK(ifp)	IF_AFDATA_WLOCK(ifp)
395#define	IF_AFDATA_UNLOCK(ifp)	IF_AFDATA_WUNLOCK(ifp)
396#define	IF_AFDATA_TRYLOCK(ifp)	rw_try_wlock(&(ifp)->if_afdata_lock)
397#define	IF_AFDATA_DESTROY(ifp)	rw_destroy(&(ifp)->if_afdata_lock)
398
399#define	IF_AFDATA_LOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_LOCKED)
400#define	IF_AFDATA_UNLOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_UNLOCKED)
401
402int	if_handoff(struct ifqueue *ifq, struct mbuf *m, struct ifnet *ifp,
403	    int adjust);
404#define	IF_HANDOFF(ifq, m, ifp)			\
405	if_handoff((struct ifqueue *)ifq, m, ifp, 0)
406#define	IF_HANDOFF_ADJ(ifq, m, ifp, adj)	\
407	if_handoff((struct ifqueue *)ifq, m, ifp, adj)
408
409void	if_start(struct ifnet *);
410
411#define	IFQ_ENQUEUE(ifq, m, err)					\
412do {									\
413	IF_LOCK(ifq);							\
414	if (ALTQ_IS_ENABLED(ifq))					\
415		ALTQ_ENQUEUE(ifq, m, NULL, err);			\
416	else {								\
417		if (_IF_QFULL(ifq)) {					\
418			m_freem(m);					\
419			(err) = ENOBUFS;				\
420		} else {						\
421			_IF_ENQUEUE(ifq, m);				\
422			(err) = 0;					\
423		}							\
424	}								\
425	if (err)							\
426		(ifq)->ifq_drops++;					\
427	IF_UNLOCK(ifq);							\
428} while (0)
429
430#define	IFQ_DEQUEUE_NOLOCK(ifq, m)					\
431do {									\
432	if (TBR_IS_ENABLED(ifq))					\
433		(m) = tbr_dequeue_ptr(ifq, ALTDQ_REMOVE);		\
434	else if (ALTQ_IS_ENABLED(ifq))					\
435		ALTQ_DEQUEUE(ifq, m);					\
436	else								\
437		_IF_DEQUEUE(ifq, m);					\
438} while (0)
439
440#define	IFQ_DEQUEUE(ifq, m)						\
441do {									\
442	IF_LOCK(ifq);							\
443	IFQ_DEQUEUE_NOLOCK(ifq, m);					\
444	IF_UNLOCK(ifq);							\
445} while (0)
446
447#define	IFQ_POLL_NOLOCK(ifq, m)						\
448do {									\
449	if (TBR_IS_ENABLED(ifq))					\
450		(m) = tbr_dequeue_ptr(ifq, ALTDQ_POLL);			\
451	else if (ALTQ_IS_ENABLED(ifq))					\
452		ALTQ_POLL(ifq, m);					\
453	else								\
454		_IF_POLL(ifq, m);					\
455} while (0)
456
457#define	IFQ_POLL(ifq, m)						\
458do {									\
459	IF_LOCK(ifq);							\
460	IFQ_POLL_NOLOCK(ifq, m);					\
461	IF_UNLOCK(ifq);							\
462} while (0)
463
464#define	IFQ_PURGE_NOLOCK(ifq)						\
465do {									\
466	if (ALTQ_IS_ENABLED(ifq)) {					\
467		ALTQ_PURGE(ifq);					\
468	} else								\
469		_IF_DRAIN(ifq);						\
470} while (0)
471
472#define	IFQ_PURGE(ifq)							\
473do {									\
474	IF_LOCK(ifq);							\
475	IFQ_PURGE_NOLOCK(ifq);						\
476	IF_UNLOCK(ifq);							\
477} while (0)
478
479#define	IFQ_SET_READY(ifq)						\
480	do { ((ifq)->altq_flags |= ALTQF_READY); } while (0)
481
482#define	IFQ_LOCK(ifq)			IF_LOCK(ifq)
483#define	IFQ_UNLOCK(ifq)			IF_UNLOCK(ifq)
484#define	IFQ_LOCK_ASSERT(ifq)		IF_LOCK_ASSERT(ifq)
485#define	IFQ_IS_EMPTY(ifq)		((ifq)->ifq_len == 0)
486#define	IFQ_INC_LEN(ifq)		((ifq)->ifq_len++)
487#define	IFQ_DEC_LEN(ifq)		(--(ifq)->ifq_len)
488#define	IFQ_INC_DROPS(ifq)		((ifq)->ifq_drops++)
489#define	IFQ_SET_MAXLEN(ifq, len)	((ifq)->ifq_maxlen = (len))
490
491/*
492 * The IFF_DRV_OACTIVE test should really occur in the device driver, not in
493 * the handoff logic, as that flag is locked by the device driver.
494 */
495#define	IFQ_HANDOFF_ADJ(ifp, m, adj, err)				\
496do {									\
497	int len;							\
498	short mflags;							\
499									\
500	len = (m)->m_pkthdr.len;					\
501	mflags = (m)->m_flags;						\
502	IFQ_ENQUEUE(&(ifp)->if_snd, m, err);				\
503	if ((err) == 0) {						\
504		(ifp)->if_obytes += len + (adj);			\
505		if (mflags & M_MCAST)					\
506			(ifp)->if_omcasts++;				\
507		if (((ifp)->if_drv_flags & IFF_DRV_OACTIVE) == 0)	\
508			if_start(ifp);					\
509	}								\
510} while (0)
511
512#define	IFQ_HANDOFF(ifp, m, err)					\
513	IFQ_HANDOFF_ADJ(ifp, m, 0, err)
514
515#define	IFQ_DRV_DEQUEUE(ifq, m)						\
516do {									\
517	(m) = (ifq)->ifq_drv_head;					\
518	if (m) {							\
519		if (((ifq)->ifq_drv_head = (m)->m_nextpkt) == NULL)	\
520			(ifq)->ifq_drv_tail = NULL;			\
521		(m)->m_nextpkt = NULL;					\
522		(ifq)->ifq_drv_len--;					\
523	} else {							\
524		IFQ_LOCK(ifq);						\
525		IFQ_DEQUEUE_NOLOCK(ifq, m);				\
526		while ((ifq)->ifq_drv_len < (ifq)->ifq_drv_maxlen) {	\
527			struct mbuf *m0;				\
528			IFQ_DEQUEUE_NOLOCK(ifq, m0);			\
529			if (m0 == NULL)					\
530				break;					\
531			m0->m_nextpkt = NULL;				\
532			if ((ifq)->ifq_drv_tail == NULL)		\
533				(ifq)->ifq_drv_head = m0;		\
534			else						\
535				(ifq)->ifq_drv_tail->m_nextpkt = m0;	\
536			(ifq)->ifq_drv_tail = m0;			\
537			(ifq)->ifq_drv_len++;				\
538		}							\
539		IFQ_UNLOCK(ifq);					\
540	}								\
541} while (0)
542
543#define	IFQ_DRV_PREPEND(ifq, m)						\
544do {									\
545	(m)->m_nextpkt = (ifq)->ifq_drv_head;				\
546	if ((ifq)->ifq_drv_tail == NULL)				\
547		(ifq)->ifq_drv_tail = (m);				\
548	(ifq)->ifq_drv_head = (m);					\
549	(ifq)->ifq_drv_len++;						\
550} while (0)
551
552#define	IFQ_DRV_IS_EMPTY(ifq)						\
553	(((ifq)->ifq_drv_len == 0) && ((ifq)->ifq_len == 0))
554
555#define	IFQ_DRV_PURGE(ifq)						\
556do {									\
557	struct mbuf *m, *n = (ifq)->ifq_drv_head;			\
558	while((m = n) != NULL) {					\
559		n = m->m_nextpkt;					\
560		m_freem(m);						\
561	}								\
562	(ifq)->ifq_drv_head = (ifq)->ifq_drv_tail = NULL;		\
563	(ifq)->ifq_drv_len = 0;						\
564	IFQ_PURGE(ifq);							\
565} while (0)
566
567#ifdef _KERNEL
568static __inline void
569drbr_stats_update(struct ifnet *ifp, int len, int mflags)
570{
571#ifndef NO_SLOW_STATS
572	ifp->if_obytes += len;
573	if (mflags & M_MCAST)
574		ifp->if_omcasts++;
575#endif
576}
577
578static __inline int
579drbr_enqueue(struct ifnet *ifp, struct buf_ring *br, struct mbuf *m)
580{
581	int error = 0;
582	int len = m->m_pkthdr.len;
583	int mflags = m->m_flags;
584
585#ifdef ALTQ
586	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
587		IFQ_ENQUEUE(&ifp->if_snd, m, error);
588		return (error);
589	}
590#endif
591	if ((error = buf_ring_enqueue_bytes(br, m, len)) == ENOBUFS) {
592		br->br_drops++;
593		m_freem(m);
594	} else
595		drbr_stats_update(ifp, len, mflags);
596
597	return (error);
598}
599
600static __inline void
601drbr_flush(struct ifnet *ifp, struct buf_ring *br)
602{
603	struct mbuf *m;
604
605#ifdef ALTQ
606	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd))
607		IFQ_PURGE(&ifp->if_snd);
608#endif
609	while ((m = buf_ring_dequeue_sc(br)) != NULL)
610		m_freem(m);
611}
612
613static __inline void
614drbr_free(struct buf_ring *br, struct malloc_type *type)
615{
616
617	drbr_flush(NULL, br);
618	buf_ring_free(br, type);
619}
620
621static __inline struct mbuf *
622drbr_dequeue(struct ifnet *ifp, struct buf_ring *br)
623{
624#ifdef ALTQ
625	struct mbuf *m;
626
627	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
628		IFQ_DEQUEUE(&ifp->if_snd, m);
629		return (m);
630	}
631#endif
632	return (buf_ring_dequeue_sc(br));
633}
634
635static __inline struct mbuf *
636drbr_dequeue_cond(struct ifnet *ifp, struct buf_ring *br,
637    int (*func) (struct mbuf *, void *), void *arg)
638{
639	struct mbuf *m;
640#ifdef ALTQ
641	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
642		IFQ_LOCK(&ifp->if_snd);
643		IFQ_POLL_NOLOCK(&ifp->if_snd, m);
644		if (m != NULL && func(m, arg) == 0) {
645			IFQ_UNLOCK(&ifp->if_snd);
646			return (NULL);
647		}
648		IFQ_DEQUEUE_NOLOCK(&ifp->if_snd, m);
649		IFQ_UNLOCK(&ifp->if_snd);
650		return (m);
651	}
652#endif
653	m = buf_ring_peek(br);
654	if (m == NULL || func(m, arg) == 0)
655		return (NULL);
656
657	return (buf_ring_dequeue_sc(br));
658}
659
660static __inline int
661drbr_empty(struct ifnet *ifp, struct buf_ring *br)
662{
663#ifdef ALTQ
664	if (ALTQ_IS_ENABLED(&ifp->if_snd))
665		return (IFQ_IS_EMPTY(&ifp->if_snd));
666#endif
667	return (buf_ring_empty(br));
668}
669
670static __inline int
671drbr_needs_enqueue(struct ifnet *ifp, struct buf_ring *br)
672{
673#ifdef ALTQ
674	if (ALTQ_IS_ENABLED(&ifp->if_snd))
675		return (1);
676#endif
677	return (!buf_ring_empty(br));
678}
679
680static __inline int
681drbr_inuse(struct ifnet *ifp, struct buf_ring *br)
682{
683#ifdef ALTQ
684	if (ALTQ_IS_ENABLED(&ifp->if_snd))
685		return (ifp->if_snd.ifq_len);
686#endif
687	return (buf_ring_count(br));
688}
689#endif
690/*
691 * 72 was chosen below because it is the size of a TCP/IP
692 * header (40) + the minimum mss (32).
693 */
694#define	IF_MINMTU	72
695#define	IF_MAXMTU	65535
696
697#endif /* _KERNEL */
698
699/*
700 * The ifaddr structure contains information about one address
701 * of an interface.  They are maintained by the different address families,
702 * are allocated and attached when an address is set, and are linked
703 * together so all addresses for an interface can be located.
704 *
705 * NOTE: a 'struct ifaddr' is always at the beginning of a larger
706 * chunk of malloc'ed memory, where we store the three addresses
707 * (ifa_addr, ifa_dstaddr and ifa_netmask) referenced here.
708 */
709struct ifaddr {
710	struct	sockaddr *ifa_addr;	/* address of interface */
711	struct	sockaddr *ifa_dstaddr;	/* other end of p-to-p link */
712#define	ifa_broadaddr	ifa_dstaddr	/* broadcast address interface */
713	struct	sockaddr *ifa_netmask;	/* used to determine subnet */
714	struct	if_data if_data;	/* not all members are meaningful */
715	struct	ifnet *ifa_ifp;		/* back-pointer to interface */
716	TAILQ_ENTRY(ifaddr) ifa_link;	/* queue macro glue */
717	void	(*ifa_rtrequest)	/* check or clean routes (+ or -)'d */
718		(int, struct rtentry *, struct rt_addrinfo *);
719	u_short	ifa_flags;		/* mostly rt_flags for cloning */
720	u_int	ifa_refcnt;		/* references to this structure */
721	int	ifa_metric;		/* cost of going out this interface */
722	int (*ifa_claim_addr)		/* check if an addr goes to this if */
723		(struct ifaddr *, struct sockaddr *);
724	struct mtx ifa_mtx;
725};
726#define	IFA_ROUTE	RTF_UP		/* route installed */
727#define IFA_RTSELF	RTF_HOST	/* loopback route to self installed */
728
729/* for compatibility with other BSDs */
730#define	ifa_list	ifa_link
731
732#ifdef _KERNEL
733#define	IFA_LOCK(ifa)		mtx_lock(&(ifa)->ifa_mtx)
734#define	IFA_UNLOCK(ifa)		mtx_unlock(&(ifa)->ifa_mtx)
735
736void	ifa_free(struct ifaddr *ifa);
737void	ifa_init(struct ifaddr *ifa);
738void	ifa_ref(struct ifaddr *ifa);
739#endif
740
741/*
742 * The prefix structure contains information about one prefix
743 * of an interface.  They are maintained by the different address families,
744 * are allocated and attached when a prefix or an address is set,
745 * and are linked together so all prefixes for an interface can be located.
746 */
747struct ifprefix {
748	struct	sockaddr *ifpr_prefix;	/* prefix of interface */
749	struct	ifnet *ifpr_ifp;	/* back-pointer to interface */
750	TAILQ_ENTRY(ifprefix) ifpr_list; /* queue macro glue */
751	u_char	ifpr_plen;		/* prefix length in bits */
752	u_char	ifpr_type;		/* protocol dependent prefix type */
753};
754
755/*
756 * Multicast address structure.  This is analogous to the ifaddr
757 * structure except that it keeps track of multicast addresses.
758 */
759struct ifmultiaddr {
760	TAILQ_ENTRY(ifmultiaddr) ifma_link; /* queue macro glue */
761	struct	sockaddr *ifma_addr; 	/* address this membership is for */
762	struct	sockaddr *ifma_lladdr;	/* link-layer translation, if any */
763	struct	ifnet *ifma_ifp;	/* back-pointer to interface */
764	u_int	ifma_refcount;		/* reference count */
765	void	*ifma_protospec;	/* protocol-specific state, if any */
766	struct	ifmultiaddr *ifma_llifma; /* pointer to ifma for ifma_lladdr */
767};
768
769#ifdef _KERNEL
770
771extern	struct rwlock ifnet_rwlock;
772extern	struct sx ifnet_sxlock;
773
774#define	IFNET_LOCK_INIT() do {						\
775	rw_init_flags(&ifnet_rwlock, "ifnet_rw",  RW_RECURSE);		\
776	sx_init_flags(&ifnet_sxlock, "ifnet_sx",  SX_RECURSE);		\
777} while(0)
778
779#define	IFNET_WLOCK() do {						\
780	sx_xlock(&ifnet_sxlock);					\
781	rw_wlock(&ifnet_rwlock);					\
782} while (0)
783
784#define	IFNET_WUNLOCK() do {						\
785	rw_wunlock(&ifnet_rwlock);					\
786	sx_xunlock(&ifnet_sxlock);					\
787} while (0)
788
789/*
790 * To assert the ifnet lock, you must know not only whether it's for read or
791 * write, but also whether it was acquired with sleep support or not.
792 */
793#define	IFNET_RLOCK_ASSERT()		sx_assert(&ifnet_sxlock, SA_SLOCKED)
794#define	IFNET_RLOCK_NOSLEEP_ASSERT()	rw_assert(&ifnet_rwlock, RA_RLOCKED)
795#define	IFNET_WLOCK_ASSERT() do {					\
796	sx_assert(&ifnet_sxlock, SA_XLOCKED);				\
797	rw_assert(&ifnet_rwlock, RA_WLOCKED);				\
798} while (0)
799
800#define	IFNET_RLOCK()		sx_slock(&ifnet_sxlock)
801#define	IFNET_RLOCK_NOSLEEP()	rw_rlock(&ifnet_rwlock)
802#define	IFNET_RUNLOCK()		sx_sunlock(&ifnet_sxlock)
803#define	IFNET_RUNLOCK_NOSLEEP()	rw_runlock(&ifnet_rwlock)
804
805/*
806 * Look up an ifnet given its index; the _ref variant also acquires a
807 * reference that must be freed using if_rele().  It is almost always a bug
808 * to call ifnet_byindex() instead if ifnet_byindex_ref().
809 */
810struct ifnet	*ifnet_byindex(u_short idx);
811struct ifnet	*ifnet_byindex_locked(u_short idx);
812struct ifnet	*ifnet_byindex_ref(u_short idx);
813
814/*
815 * Given the index, ifaddr_byindex() returns the one and only
816 * link-level ifaddr for the interface. You are not supposed to use
817 * it to traverse the list of addresses associated to the interface.
818 */
819struct ifaddr	*ifaddr_byindex(u_short idx);
820
821VNET_DECLARE(struct ifnethead, ifnet);
822VNET_DECLARE(struct ifgrouphead, ifg_head);
823VNET_DECLARE(int, if_index);
824VNET_DECLARE(struct ifnet *, loif);	/* first loopback interface */
825VNET_DECLARE(int, useloopback);
826
827#define	V_ifnet		VNET(ifnet)
828#define	V_ifg_head	VNET(ifg_head)
829#define	V_if_index	VNET(if_index)
830#define	V_loif		VNET(loif)
831#define	V_useloopback	VNET(useloopback)
832
833extern	int ifqmaxlen;
834
835int	if_addgroup(struct ifnet *, const char *);
836int	if_delgroup(struct ifnet *, const char *);
837int	if_addmulti(struct ifnet *, struct sockaddr *, struct ifmultiaddr **);
838int	if_allmulti(struct ifnet *, int);
839struct	ifnet* if_alloc(u_char);
840void	if_attach(struct ifnet *);
841void	if_dead(struct ifnet *);
842int	if_delmulti(struct ifnet *, struct sockaddr *);
843void	if_delmulti_ifma(struct ifmultiaddr *);
844void	if_detach(struct ifnet *);
845void	if_vmove(struct ifnet *, struct vnet *);
846void	if_purgeaddrs(struct ifnet *);
847void	if_delallmulti(struct ifnet *);
848void	if_down(struct ifnet *);
849struct ifmultiaddr *
850	if_findmulti(struct ifnet *, struct sockaddr *);
851void	if_free(struct ifnet *);
852void	if_free_type(struct ifnet *, u_char);
853void	if_initname(struct ifnet *, const char *, int);
854void	if_link_state_change(struct ifnet *, int);
855int	if_printf(struct ifnet *, const char *, ...) __printflike(2, 3);
856void	if_qflush(struct ifnet *);
857void	if_ref(struct ifnet *);
858void	if_rele(struct ifnet *);
859int	if_setlladdr(struct ifnet *, const u_char *, int);
860void	if_up(struct ifnet *);
861int	ifioctl(struct socket *, u_long, caddr_t, struct thread *);
862int	ifpromisc(struct ifnet *, int);
863struct	ifnet *ifunit(const char *);
864struct	ifnet *ifunit_ref(const char *);
865
866void	ifq_init(struct ifaltq *, struct ifnet *ifp);
867void	ifq_delete(struct ifaltq *);
868
869int	ifa_add_loopback_route(struct ifaddr *, struct sockaddr *);
870int	ifa_del_loopback_route(struct ifaddr *, struct sockaddr *);
871
872struct	ifaddr *ifa_ifwithaddr(struct sockaddr *);
873int		ifa_ifwithaddr_check(struct sockaddr *);
874struct	ifaddr *ifa_ifwithbroadaddr(struct sockaddr *);
875struct	ifaddr *ifa_ifwithdstaddr(struct sockaddr *);
876struct	ifaddr *ifa_ifwithnet(struct sockaddr *, int);
877struct	ifaddr *ifa_ifwithroute(int, struct sockaddr *, struct sockaddr *);
878struct	ifaddr *ifa_ifwithroute_fib(int, struct sockaddr *, struct sockaddr *, u_int);
879
880struct	ifaddr *ifaof_ifpforaddr(struct sockaddr *, struct ifnet *);
881
882int	if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen);
883
884typedef	void *if_com_alloc_t(u_char type, struct ifnet *ifp);
885typedef	void if_com_free_t(void *com, u_char type);
886void	if_register_com_alloc(u_char type, if_com_alloc_t *a, if_com_free_t *f);
887void	if_deregister_com_alloc(u_char type);
888
889#define IF_LLADDR(ifp)							\
890    LLADDR((struct sockaddr_dl *)((ifp)->if_addr->ifa_addr))
891
892#ifdef DEVICE_POLLING
893enum poll_cmd {	POLL_ONLY, POLL_AND_CHECK_STATUS };
894
895typedef	int poll_handler_t(struct ifnet *ifp, enum poll_cmd cmd, int count);
896int    ether_poll_register(poll_handler_t *h, struct ifnet *ifp);
897int    ether_poll_deregister(struct ifnet *ifp);
898#endif /* DEVICE_POLLING */
899
900#endif /* _KERNEL */
901
902#endif /* !_NET_IF_VAR_H_ */
903