ifq.h revision 256525
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 256525 2013-10-15 11:37:57Z 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/counter.h>
87#include <sys/lock.h>		/* XXX */
88#include <sys/mutex.h>		/* XXX */
89#include <sys/rwlock.h>		/* XXX */
90#include <sys/sx.h>		/* XXX */
91#include <sys/event.h>		/* XXX */
92#include <sys/_task.h>
93
94#define	IF_DUNIT_NONE	-1
95
96#include <altq/if_altq.h>
97
98TAILQ_HEAD(ifnethead, ifnet);	/* we use TAILQs so that the order of */
99TAILQ_HEAD(ifaddrhead, ifaddr);	/* instantiation is preserved in the list */
100TAILQ_HEAD(ifmultihead, ifmultiaddr);
101TAILQ_HEAD(ifgrouphead, ifg_group);
102
103#ifdef _KERNEL
104VNET_DECLARE(struct pfil_head, link_pfil_hook);	/* packet filter hooks */
105#define	V_link_pfil_hook	VNET(link_pfil_hook)
106#endif /* _KERNEL */
107
108/*
109 * Structure defining a queue for a network interface.
110 */
111struct	ifqueue {
112	struct	mbuf *ifq_head;
113	struct	mbuf *ifq_tail;
114	int	ifq_len;
115	int	ifq_maxlen;
116	int	ifq_drops;
117	struct	mtx ifq_mtx;
118};
119
120/*
121 * Structure defining a network interface.
122 *
123 * (Would like to call this struct ``if'', but C isn't PL/1.)
124 */
125
126struct ifnet {
127	void	*if_softc;		/* pointer to driver state */
128	void	*if_l2com;		/* pointer to protocol bits */
129	struct vnet *if_vnet;		/* pointer to network stack instance */
130	TAILQ_ENTRY(ifnet) if_link; 	/* all struct ifnets are chained */
131	char	if_xname[IFNAMSIZ];	/* external name (name + unit) */
132	const char *if_dname;		/* driver name */
133	int	if_dunit;		/* unit or IF_DUNIT_NONE */
134	u_int	if_refcount;		/* reference count */
135	struct	ifaddrhead if_addrhead;	/* linked list of addresses per if */
136		/*
137		 * if_addrhead is the list of all addresses associated to
138		 * an interface.
139		 * Some code in the kernel assumes that first element
140		 * of the list has type AF_LINK, and contains sockaddr_dl
141		 * addresses which store the link-level address and the name
142		 * of the interface.
143		 * However, access to the AF_LINK address through this
144		 * field is deprecated. Use if_addr or ifaddr_byindex() instead.
145		 */
146	int	if_pcount;		/* number of promiscuous listeners */
147	struct	carp_if *if_carp;	/* carp interface structure */
148	struct	bpf_if *if_bpf;		/* packet filter structure */
149	u_short	if_index;		/* numeric abbreviation for this if  */
150	short	if_index_reserved;	/* spare space to grow if_index */
151	struct  ifvlantrunk *if_vlantrunk; /* pointer to 802.1q data */
152	int	if_flags;		/* up/down, broadcast, etc. */
153	int	if_capabilities;	/* interface features & capabilities */
154	int	if_capenable;		/* enabled features & capabilities */
155	void	*if_linkmib;		/* link-type-specific MIB data */
156	size_t	if_linkmiblen;		/* length of above data */
157	struct	if_data if_data;
158	struct	ifmultihead if_multiaddrs; /* multicast addresses configured */
159	int	if_amcount;		/* number of all-multicast requests */
160/* procedure handles */
161	int	(*if_output)		/* output routine (enqueue) */
162		(struct ifnet *, struct mbuf *, const struct sockaddr *,
163		     struct route *);
164	void	(*if_input)		/* input routine (from h/w driver) */
165		(struct ifnet *, struct mbuf *);
166	void	(*if_start)		/* initiate output routine */
167		(struct ifnet *);
168	int	(*if_ioctl)		/* ioctl routine */
169		(struct ifnet *, u_long, caddr_t);
170	void	(*if_init)		/* Init routine */
171		(void *);
172	int	(*if_resolvemulti)	/* validate/resolve multicast */
173		(struct ifnet *, struct sockaddr **, struct sockaddr *);
174	void	(*if_qflush)		/* flush any queues */
175		(struct ifnet *);
176	int	(*if_transmit)		/* initiate output routine */
177		(struct ifnet *, struct mbuf *);
178	void	(*if_reassign)		/* reassign to vnet routine */
179		(struct ifnet *, struct vnet *, char *);
180	struct	vnet *if_home_vnet;	/* where this ifnet originates from */
181	struct	ifaddr	*if_addr;	/* pointer to link-level address */
182	void	*if_llsoftc;		/* link layer softc */
183	int	if_drv_flags;		/* driver-managed status flags */
184	struct  ifaltq if_snd;		/* output queue (includes altq) */
185	const u_int8_t *if_broadcastaddr; /* linklevel broadcast bytestring */
186
187	void	*if_bridge;		/* bridge glue */
188
189	struct	label *if_label;	/* interface MAC label */
190
191	/* these are only used by IPv6 */
192	void	*if_unused[2];
193	void	*if_afdata[AF_MAX];
194	int	if_afdata_initialized;
195	struct	rwlock if_afdata_lock;
196	struct	task if_linktask;	/* task for link change events */
197	struct	rwlock if_addr_lock;	/* lock to protect address lists */
198
199	LIST_ENTRY(ifnet) if_clones;	/* interfaces of a cloner */
200	TAILQ_HEAD(, ifg_list) if_groups; /* linked list of groups per if */
201					/* protected by if_addr_lock */
202	void	*if_pf_kif;
203	void	*if_lagg;		/* lagg glue */
204	char	*if_description;	/* interface description */
205	u_int	if_fib;			/* interface FIB */
206	u_char	if_alloctype;		/* if_type at time of allocation */
207
208	u_int	if_hw_tsomax;		/* tso burst length limit, the minimum
209					 * is (IP_MAXPACKET / 8).
210					 * XXXAO: Have to find a better place
211					 * for it eventually. */
212
213	/*
214	 * Spare fields are added so that we can modify sensitive data
215	 * structures without changing the kernel binary interface, and must
216	 * be used with care where binary compatibility is required.
217	 */
218	char	if_cspare[3];
219	int	if_ispare[4];
220	void	*if_pspare[8];		/* 1 netmap, 7 TDB */
221};
222
223typedef void if_init_f_t(void *);
224
225/*
226 * XXX These aliases are terribly dangerous because they could apply
227 * to anything.
228 */
229#define	if_mtu		if_data.ifi_mtu
230#define	if_type		if_data.ifi_type
231#define if_physical	if_data.ifi_physical
232#define	if_addrlen	if_data.ifi_addrlen
233#define	if_hdrlen	if_data.ifi_hdrlen
234#define	if_metric	if_data.ifi_metric
235#define	if_link_state	if_data.ifi_link_state
236#define	if_baudrate	if_data.ifi_baudrate
237#define	if_baudrate_pf	if_data.ifi_baudrate_pf
238#define	if_hwassist	if_data.ifi_hwassist
239#define	if_ipackets	if_data.ifi_ipackets
240#define	if_ierrors	if_data.ifi_ierrors
241#define	if_opackets	if_data.ifi_opackets
242#define	if_oerrors	if_data.ifi_oerrors
243#define	if_collisions	if_data.ifi_collisions
244#define	if_ibytes	if_data.ifi_ibytes
245#define	if_obytes	if_data.ifi_obytes
246#define	if_imcasts	if_data.ifi_imcasts
247#define	if_omcasts	if_data.ifi_omcasts
248#define	if_iqdrops	if_data.ifi_iqdrops
249#define	if_noproto	if_data.ifi_noproto
250#define	if_lastchange	if_data.ifi_lastchange
251
252/* for compatibility with other BSDs */
253#define	if_addrlist	if_addrhead
254#define	if_list		if_link
255#define	if_name(ifp)	((ifp)->if_xname)
256
257/*
258 * Locks for address lists on the network interface.
259 */
260#define	IF_ADDR_LOCK_INIT(if)	rw_init(&(if)->if_addr_lock, "if_addr_lock")
261#define	IF_ADDR_LOCK_DESTROY(if)	rw_destroy(&(if)->if_addr_lock)
262#define	IF_ADDR_WLOCK(if)	rw_wlock(&(if)->if_addr_lock)
263#define	IF_ADDR_WUNLOCK(if)	rw_wunlock(&(if)->if_addr_lock)
264#define	IF_ADDR_RLOCK(if)	rw_rlock(&(if)->if_addr_lock)
265#define	IF_ADDR_RUNLOCK(if)	rw_runlock(&(if)->if_addr_lock)
266#define	IF_ADDR_LOCK_ASSERT(if)	rw_assert(&(if)->if_addr_lock, RA_LOCKED)
267#define	IF_ADDR_WLOCK_ASSERT(if) rw_assert(&(if)->if_addr_lock, RA_WLOCKED)
268
269/*
270 * Function variations on locking macros intended to be used by loadable
271 * kernel modules in order to divorce them from the internals of address list
272 * locking.
273 */
274void	if_addr_rlock(struct ifnet *ifp);	/* if_addrhead */
275void	if_addr_runlock(struct ifnet *ifp);	/* if_addrhead */
276void	if_maddr_rlock(struct ifnet *ifp);	/* if_multiaddrs */
277void	if_maddr_runlock(struct ifnet *ifp);	/* if_multiaddrs */
278
279/*
280 * Output queues (ifp->if_snd) and slow device input queues (*ifp->if_slowq)
281 * are queues of messages stored on ifqueue structures
282 * (defined above).  Entries are added to and deleted from these structures
283 * by these macros.
284 */
285#define IF_LOCK(ifq)		mtx_lock(&(ifq)->ifq_mtx)
286#define IF_UNLOCK(ifq)		mtx_unlock(&(ifq)->ifq_mtx)
287#define	IF_LOCK_ASSERT(ifq)	mtx_assert(&(ifq)->ifq_mtx, MA_OWNED)
288#define	_IF_QFULL(ifq)		((ifq)->ifq_len >= (ifq)->ifq_maxlen)
289#define	_IF_DROP(ifq)		((ifq)->ifq_drops++)
290#define	_IF_QLEN(ifq)		((ifq)->ifq_len)
291
292#define	_IF_ENQUEUE(ifq, m) do { 				\
293	(m)->m_nextpkt = NULL;					\
294	if ((ifq)->ifq_tail == NULL) 				\
295		(ifq)->ifq_head = m; 				\
296	else 							\
297		(ifq)->ifq_tail->m_nextpkt = m; 		\
298	(ifq)->ifq_tail = m; 					\
299	(ifq)->ifq_len++; 					\
300} while (0)
301
302#define IF_ENQUEUE(ifq, m) do {					\
303	IF_LOCK(ifq); 						\
304	_IF_ENQUEUE(ifq, m); 					\
305	IF_UNLOCK(ifq); 					\
306} while (0)
307
308#define	_IF_PREPEND(ifq, m) do {				\
309	(m)->m_nextpkt = (ifq)->ifq_head; 			\
310	if ((ifq)->ifq_tail == NULL) 				\
311		(ifq)->ifq_tail = (m); 				\
312	(ifq)->ifq_head = (m); 					\
313	(ifq)->ifq_len++; 					\
314} while (0)
315
316#define IF_PREPEND(ifq, m) do {		 			\
317	IF_LOCK(ifq); 						\
318	_IF_PREPEND(ifq, m); 					\
319	IF_UNLOCK(ifq); 					\
320} while (0)
321
322#define	_IF_DEQUEUE(ifq, m) do { 				\
323	(m) = (ifq)->ifq_head; 					\
324	if (m) { 						\
325		if (((ifq)->ifq_head = (m)->m_nextpkt) == NULL)	\
326			(ifq)->ifq_tail = NULL; 		\
327		(m)->m_nextpkt = NULL; 				\
328		(ifq)->ifq_len--; 				\
329	} 							\
330} while (0)
331
332#define IF_DEQUEUE(ifq, m) do { 				\
333	IF_LOCK(ifq); 						\
334	_IF_DEQUEUE(ifq, m); 					\
335	IF_UNLOCK(ifq); 					\
336} while (0)
337
338#define	_IF_DEQUEUE_ALL(ifq, m) do {				\
339	(m) = (ifq)->ifq_head;					\
340	(ifq)->ifq_head = (ifq)->ifq_tail = NULL;		\
341	(ifq)->ifq_len = 0;					\
342} while (0)
343
344#define	IF_DEQUEUE_ALL(ifq, m) do {				\
345	IF_LOCK(ifq); 						\
346	_IF_DEQUEUE_ALL(ifq, m);				\
347	IF_UNLOCK(ifq); 					\
348} while (0)
349
350#define	_IF_POLL(ifq, m)	((m) = (ifq)->ifq_head)
351#define	IF_POLL(ifq, m)		_IF_POLL(ifq, m)
352
353#define _IF_DRAIN(ifq) do { 					\
354	struct mbuf *m; 					\
355	for (;;) { 						\
356		_IF_DEQUEUE(ifq, m); 				\
357		if (m == NULL) 					\
358			break; 					\
359		m_freem(m); 					\
360	} 							\
361} while (0)
362
363#define IF_DRAIN(ifq) do {					\
364	IF_LOCK(ifq);						\
365	_IF_DRAIN(ifq);						\
366	IF_UNLOCK(ifq);						\
367} while(0)
368
369#ifdef _KERNEL
370/* interface link layer address change event */
371typedef void (*iflladdr_event_handler_t)(void *, struct ifnet *);
372EVENTHANDLER_DECLARE(iflladdr_event, iflladdr_event_handler_t);
373/* interface address change event */
374typedef void (*ifaddr_event_handler_t)(void *, struct ifnet *);
375EVENTHANDLER_DECLARE(ifaddr_event, ifaddr_event_handler_t);
376/* new interface arrival event */
377typedef void (*ifnet_arrival_event_handler_t)(void *, struct ifnet *);
378EVENTHANDLER_DECLARE(ifnet_arrival_event, ifnet_arrival_event_handler_t);
379/* interface departure event */
380typedef void (*ifnet_departure_event_handler_t)(void *, struct ifnet *);
381EVENTHANDLER_DECLARE(ifnet_departure_event, ifnet_departure_event_handler_t);
382/* Interface link state change event */
383typedef void (*ifnet_link_event_handler_t)(void *, struct ifnet *, int);
384EVENTHANDLER_DECLARE(ifnet_link_event, ifnet_link_event_handler_t);
385
386/*
387 * interface groups
388 */
389struct ifg_group {
390	char				 ifg_group[IFNAMSIZ];
391	u_int				 ifg_refcnt;
392	void				*ifg_pf_kif;
393	TAILQ_HEAD(, ifg_member)	 ifg_members;
394	TAILQ_ENTRY(ifg_group)		 ifg_next;
395};
396
397struct ifg_member {
398	TAILQ_ENTRY(ifg_member)	 ifgm_next;
399	struct ifnet		*ifgm_ifp;
400};
401
402struct ifg_list {
403	struct ifg_group	*ifgl_group;
404	TAILQ_ENTRY(ifg_list)	 ifgl_next;
405};
406
407/* group attach event */
408typedef void (*group_attach_event_handler_t)(void *, struct ifg_group *);
409EVENTHANDLER_DECLARE(group_attach_event, group_attach_event_handler_t);
410/* group detach event */
411typedef void (*group_detach_event_handler_t)(void *, struct ifg_group *);
412EVENTHANDLER_DECLARE(group_detach_event, group_detach_event_handler_t);
413/* group change event */
414typedef void (*group_change_event_handler_t)(void *, const char *);
415EVENTHANDLER_DECLARE(group_change_event, group_change_event_handler_t);
416
417#define	IF_AFDATA_LOCK_INIT(ifp)	\
418	rw_init(&(ifp)->if_afdata_lock, "if_afdata")
419
420#define	IF_AFDATA_WLOCK(ifp)	rw_wlock(&(ifp)->if_afdata_lock)
421#define	IF_AFDATA_RLOCK(ifp)	rw_rlock(&(ifp)->if_afdata_lock)
422#define	IF_AFDATA_WUNLOCK(ifp)	rw_wunlock(&(ifp)->if_afdata_lock)
423#define	IF_AFDATA_RUNLOCK(ifp)	rw_runlock(&(ifp)->if_afdata_lock)
424#define	IF_AFDATA_LOCK(ifp)	IF_AFDATA_WLOCK(ifp)
425#define	IF_AFDATA_UNLOCK(ifp)	IF_AFDATA_WUNLOCK(ifp)
426#define	IF_AFDATA_TRYLOCK(ifp)	rw_try_wlock(&(ifp)->if_afdata_lock)
427#define	IF_AFDATA_DESTROY(ifp)	rw_destroy(&(ifp)->if_afdata_lock)
428
429#define	IF_AFDATA_LOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_LOCKED)
430#define	IF_AFDATA_RLOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_RLOCKED)
431#define	IF_AFDATA_WLOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_WLOCKED)
432#define	IF_AFDATA_UNLOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_UNLOCKED)
433
434int	if_handoff(struct ifqueue *ifq, struct mbuf *m, struct ifnet *ifp,
435	    int adjust);
436#define	IF_HANDOFF(ifq, m, ifp)			\
437	if_handoff((struct ifqueue *)ifq, m, ifp, 0)
438#define	IF_HANDOFF_ADJ(ifq, m, ifp, adj)	\
439	if_handoff((struct ifqueue *)ifq, m, ifp, adj)
440
441void	if_start(struct ifnet *);
442
443#define	IFQ_ENQUEUE(ifq, m, err)					\
444do {									\
445	IF_LOCK(ifq);							\
446	if (ALTQ_IS_ENABLED(ifq))					\
447		ALTQ_ENQUEUE(ifq, m, NULL, err);			\
448	else {								\
449		if (_IF_QFULL(ifq)) {					\
450			m_freem(m);					\
451			(err) = ENOBUFS;				\
452		} else {						\
453			_IF_ENQUEUE(ifq, m);				\
454			(err) = 0;					\
455		}							\
456	}								\
457	if (err)							\
458		(ifq)->ifq_drops++;					\
459	IF_UNLOCK(ifq);							\
460} while (0)
461
462#define	IFQ_DEQUEUE_NOLOCK(ifq, m)					\
463do {									\
464	if (TBR_IS_ENABLED(ifq))					\
465		(m) = tbr_dequeue_ptr(ifq, ALTDQ_REMOVE);		\
466	else if (ALTQ_IS_ENABLED(ifq))					\
467		ALTQ_DEQUEUE(ifq, m);					\
468	else								\
469		_IF_DEQUEUE(ifq, m);					\
470} while (0)
471
472#define	IFQ_DEQUEUE(ifq, m)						\
473do {									\
474	IF_LOCK(ifq);							\
475	IFQ_DEQUEUE_NOLOCK(ifq, m);					\
476	IF_UNLOCK(ifq);							\
477} while (0)
478
479#define	IFQ_POLL_NOLOCK(ifq, m)						\
480do {									\
481	if (TBR_IS_ENABLED(ifq))					\
482		(m) = tbr_dequeue_ptr(ifq, ALTDQ_POLL);			\
483	else if (ALTQ_IS_ENABLED(ifq))					\
484		ALTQ_POLL(ifq, m);					\
485	else								\
486		_IF_POLL(ifq, m);					\
487} while (0)
488
489#define	IFQ_POLL(ifq, m)						\
490do {									\
491	IF_LOCK(ifq);							\
492	IFQ_POLL_NOLOCK(ifq, m);					\
493	IF_UNLOCK(ifq);							\
494} while (0)
495
496#define	IFQ_PURGE_NOLOCK(ifq)						\
497do {									\
498	if (ALTQ_IS_ENABLED(ifq)) {					\
499		ALTQ_PURGE(ifq);					\
500	} else								\
501		_IF_DRAIN(ifq);						\
502} while (0)
503
504#define	IFQ_PURGE(ifq)							\
505do {									\
506	IF_LOCK(ifq);							\
507	IFQ_PURGE_NOLOCK(ifq);						\
508	IF_UNLOCK(ifq);							\
509} while (0)
510
511#define	IFQ_SET_READY(ifq)						\
512	do { ((ifq)->altq_flags |= ALTQF_READY); } while (0)
513
514#define	IFQ_LOCK(ifq)			IF_LOCK(ifq)
515#define	IFQ_UNLOCK(ifq)			IF_UNLOCK(ifq)
516#define	IFQ_LOCK_ASSERT(ifq)		IF_LOCK_ASSERT(ifq)
517#define	IFQ_IS_EMPTY(ifq)		((ifq)->ifq_len == 0)
518#define	IFQ_INC_LEN(ifq)		((ifq)->ifq_len++)
519#define	IFQ_DEC_LEN(ifq)		(--(ifq)->ifq_len)
520#define	IFQ_INC_DROPS(ifq)		((ifq)->ifq_drops++)
521#define	IFQ_SET_MAXLEN(ifq, len)	((ifq)->ifq_maxlen = (len))
522
523/*
524 * The IFF_DRV_OACTIVE test should really occur in the device driver, not in
525 * the handoff logic, as that flag is locked by the device driver.
526 */
527#define	IFQ_HANDOFF_ADJ(ifp, m, adj, err)				\
528do {									\
529	int len;							\
530	short mflags;							\
531									\
532	len = (m)->m_pkthdr.len;					\
533	mflags = (m)->m_flags;						\
534	IFQ_ENQUEUE(&(ifp)->if_snd, m, err);				\
535	if ((err) == 0) {						\
536		(ifp)->if_obytes += len + (adj);			\
537		if (mflags & M_MCAST)					\
538			(ifp)->if_omcasts++;				\
539		if (((ifp)->if_drv_flags & IFF_DRV_OACTIVE) == 0)	\
540			if_start(ifp);					\
541	}								\
542} while (0)
543
544#define	IFQ_HANDOFF(ifp, m, err)					\
545	IFQ_HANDOFF_ADJ(ifp, m, 0, err)
546
547#define	IFQ_DRV_DEQUEUE(ifq, m)						\
548do {									\
549	(m) = (ifq)->ifq_drv_head;					\
550	if (m) {							\
551		if (((ifq)->ifq_drv_head = (m)->m_nextpkt) == NULL)	\
552			(ifq)->ifq_drv_tail = NULL;			\
553		(m)->m_nextpkt = NULL;					\
554		(ifq)->ifq_drv_len--;					\
555	} else {							\
556		IFQ_LOCK(ifq);						\
557		IFQ_DEQUEUE_NOLOCK(ifq, m);				\
558		while ((ifq)->ifq_drv_len < (ifq)->ifq_drv_maxlen) {	\
559			struct mbuf *m0;				\
560			IFQ_DEQUEUE_NOLOCK(ifq, m0);			\
561			if (m0 == NULL)					\
562				break;					\
563			m0->m_nextpkt = NULL;				\
564			if ((ifq)->ifq_drv_tail == NULL)		\
565				(ifq)->ifq_drv_head = m0;		\
566			else						\
567				(ifq)->ifq_drv_tail->m_nextpkt = m0;	\
568			(ifq)->ifq_drv_tail = m0;			\
569			(ifq)->ifq_drv_len++;				\
570		}							\
571		IFQ_UNLOCK(ifq);					\
572	}								\
573} while (0)
574
575#define	IFQ_DRV_PREPEND(ifq, m)						\
576do {									\
577	(m)->m_nextpkt = (ifq)->ifq_drv_head;				\
578	if ((ifq)->ifq_drv_tail == NULL)				\
579		(ifq)->ifq_drv_tail = (m);				\
580	(ifq)->ifq_drv_head = (m);					\
581	(ifq)->ifq_drv_len++;						\
582} while (0)
583
584#define	IFQ_DRV_IS_EMPTY(ifq)						\
585	(((ifq)->ifq_drv_len == 0) && ((ifq)->ifq_len == 0))
586
587#define	IFQ_DRV_PURGE(ifq)						\
588do {									\
589	struct mbuf *m, *n = (ifq)->ifq_drv_head;			\
590	while((m = n) != NULL) {					\
591		n = m->m_nextpkt;					\
592		m_freem(m);						\
593	}								\
594	(ifq)->ifq_drv_head = (ifq)->ifq_drv_tail = NULL;		\
595	(ifq)->ifq_drv_len = 0;						\
596	IFQ_PURGE(ifq);							\
597} while (0)
598
599#ifdef _KERNEL
600static __inline void
601if_initbaudrate(struct ifnet *ifp, uintmax_t baud)
602{
603
604	ifp->if_baudrate_pf = 0;
605	while (baud > (u_long)(~0UL)) {
606		baud /= 10;
607		ifp->if_baudrate_pf++;
608	}
609	ifp->if_baudrate = baud;
610}
611
612static __inline int
613drbr_enqueue(struct ifnet *ifp, struct buf_ring *br, struct mbuf *m)
614{
615	int error = 0;
616
617#ifdef ALTQ
618	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
619		IFQ_ENQUEUE(&ifp->if_snd, m, error);
620		return (error);
621	}
622#endif
623	error = buf_ring_enqueue(br, m);
624	if (error)
625		m_freem(m);
626
627	return (error);
628}
629
630static __inline void
631drbr_putback(struct ifnet *ifp, struct buf_ring *br, struct mbuf *new)
632{
633	/*
634	 * The top of the list needs to be swapped
635	 * for this one.
636	 */
637#ifdef ALTQ
638	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd)) {
639		/*
640		 * Peek in altq case dequeued it
641		 * so put it back.
642		 */
643		IFQ_DRV_PREPEND(&ifp->if_snd, new);
644		return;
645	}
646#endif
647	buf_ring_putback_sc(br, new);
648}
649
650static __inline struct mbuf *
651drbr_peek(struct ifnet *ifp, struct buf_ring *br)
652{
653#ifdef ALTQ
654	struct mbuf *m;
655	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd)) {
656		/*
657		 * Pull it off like a dequeue
658		 * since drbr_advance() does nothing
659		 * for altq and drbr_putback() will
660		 * use the old prepend function.
661		 */
662		IFQ_DEQUEUE(&ifp->if_snd, m);
663		return (m);
664	}
665#endif
666	return(buf_ring_peek(br));
667}
668
669static __inline void
670drbr_flush(struct ifnet *ifp, struct buf_ring *br)
671{
672	struct mbuf *m;
673
674#ifdef ALTQ
675	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd))
676		IFQ_PURGE(&ifp->if_snd);
677#endif
678	while ((m = buf_ring_dequeue_sc(br)) != NULL)
679		m_freem(m);
680}
681
682static __inline void
683drbr_free(struct buf_ring *br, struct malloc_type *type)
684{
685
686	drbr_flush(NULL, br);
687	buf_ring_free(br, type);
688}
689
690static __inline struct mbuf *
691drbr_dequeue(struct ifnet *ifp, struct buf_ring *br)
692{
693#ifdef ALTQ
694	struct mbuf *m;
695
696	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd)) {
697		IFQ_DEQUEUE(&ifp->if_snd, m);
698		return (m);
699	}
700#endif
701	return (buf_ring_dequeue_sc(br));
702}
703
704static __inline void
705drbr_advance(struct ifnet *ifp, struct buf_ring *br)
706{
707#ifdef ALTQ
708	/* Nothing to do here since peek dequeues in altq case */
709	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd))
710		return;
711#endif
712	return (buf_ring_advance_sc(br));
713}
714
715
716static __inline struct mbuf *
717drbr_dequeue_cond(struct ifnet *ifp, struct buf_ring *br,
718    int (*func) (struct mbuf *, void *), void *arg)
719{
720	struct mbuf *m;
721#ifdef ALTQ
722	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
723		IFQ_LOCK(&ifp->if_snd);
724		IFQ_POLL_NOLOCK(&ifp->if_snd, m);
725		if (m != NULL && func(m, arg) == 0) {
726			IFQ_UNLOCK(&ifp->if_snd);
727			return (NULL);
728		}
729		IFQ_DEQUEUE_NOLOCK(&ifp->if_snd, m);
730		IFQ_UNLOCK(&ifp->if_snd);
731		return (m);
732	}
733#endif
734	m = buf_ring_peek(br);
735	if (m == NULL || func(m, arg) == 0)
736		return (NULL);
737
738	return (buf_ring_dequeue_sc(br));
739}
740
741static __inline int
742drbr_empty(struct ifnet *ifp, struct buf_ring *br)
743{
744#ifdef ALTQ
745	if (ALTQ_IS_ENABLED(&ifp->if_snd))
746		return (IFQ_IS_EMPTY(&ifp->if_snd));
747#endif
748	return (buf_ring_empty(br));
749}
750
751static __inline int
752drbr_needs_enqueue(struct ifnet *ifp, struct buf_ring *br)
753{
754#ifdef ALTQ
755	if (ALTQ_IS_ENABLED(&ifp->if_snd))
756		return (1);
757#endif
758	return (!buf_ring_empty(br));
759}
760
761static __inline int
762drbr_inuse(struct ifnet *ifp, struct buf_ring *br)
763{
764#ifdef ALTQ
765	if (ALTQ_IS_ENABLED(&ifp->if_snd))
766		return (ifp->if_snd.ifq_len);
767#endif
768	return (buf_ring_count(br));
769}
770#endif
771/*
772 * 72 was chosen below because it is the size of a TCP/IP
773 * header (40) + the minimum mss (32).
774 */
775#define	IF_MINMTU	72
776#define	IF_MAXMTU	65535
777
778#define	TOEDEV(ifp)	((ifp)->if_llsoftc)
779
780#endif /* _KERNEL */
781
782/*
783 * The ifaddr structure contains information about one address
784 * of an interface.  They are maintained by the different address families,
785 * are allocated and attached when an address is set, and are linked
786 * together so all addresses for an interface can be located.
787 *
788 * NOTE: a 'struct ifaddr' is always at the beginning of a larger
789 * chunk of malloc'ed memory, where we store the three addresses
790 * (ifa_addr, ifa_dstaddr and ifa_netmask) referenced here.
791 */
792#if defined(_KERNEL) || defined(_WANT_IFADDR)
793struct ifaddr {
794	struct	sockaddr *ifa_addr;	/* address of interface */
795	struct	sockaddr *ifa_dstaddr;	/* other end of p-to-p link */
796#define	ifa_broadaddr	ifa_dstaddr	/* broadcast address interface */
797	struct	sockaddr *ifa_netmask;	/* used to determine subnet */
798	struct	ifnet *ifa_ifp;		/* back-pointer to interface */
799	struct	carp_softc *ifa_carp;	/* pointer to CARP data */
800	TAILQ_ENTRY(ifaddr) ifa_link;	/* queue macro glue */
801	void	(*ifa_rtrequest)	/* check or clean routes (+ or -)'d */
802		(int, struct rtentry *, struct rt_addrinfo *);
803	u_short	ifa_flags;		/* mostly rt_flags for cloning */
804	u_int	ifa_refcnt;		/* references to this structure */
805	int	ifa_metric;		/* cost of going out this interface */
806	int (*ifa_claim_addr)		/* check if an addr goes to this if */
807		(struct ifaddr *, struct sockaddr *);
808
809	counter_u64_t	ifa_ipackets;
810	counter_u64_t	ifa_opackets;
811	counter_u64_t	ifa_ibytes;
812	counter_u64_t	ifa_obytes;
813};
814#endif
815
816#ifdef _KERNEL
817#define	IFA_ROUTE	RTF_UP		/* route installed */
818#define	IFA_RTSELF	RTF_HOST	/* loopback route to self installed */
819
820/* For compatibility with other BSDs. SCTP uses it. */
821#define	ifa_list	ifa_link
822
823struct ifaddr *	ifa_alloc(size_t size, int flags);
824void	ifa_free(struct ifaddr *ifa);
825void	ifa_ref(struct ifaddr *ifa);
826#endif /* _KERNEL */
827
828/*
829 * Multicast address structure.  This is analogous to the ifaddr
830 * structure except that it keeps track of multicast addresses.
831 */
832struct ifmultiaddr {
833	TAILQ_ENTRY(ifmultiaddr) ifma_link; /* queue macro glue */
834	struct	sockaddr *ifma_addr; 	/* address this membership is for */
835	struct	sockaddr *ifma_lladdr;	/* link-layer translation, if any */
836	struct	ifnet *ifma_ifp;	/* back-pointer to interface */
837	u_int	ifma_refcount;		/* reference count */
838	void	*ifma_protospec;	/* protocol-specific state, if any */
839	struct	ifmultiaddr *ifma_llifma; /* pointer to ifma for ifma_lladdr */
840};
841
842#ifdef _KERNEL
843
844extern	struct rwlock ifnet_rwlock;
845extern	struct sx ifnet_sxlock;
846
847#define	IFNET_LOCK_INIT() do {						\
848	rw_init_flags(&ifnet_rwlock, "ifnet_rw",  RW_RECURSE);		\
849	sx_init_flags(&ifnet_sxlock, "ifnet_sx",  SX_RECURSE);		\
850} while(0)
851
852#define	IFNET_WLOCK() do {						\
853	sx_xlock(&ifnet_sxlock);					\
854	rw_wlock(&ifnet_rwlock);					\
855} while (0)
856
857#define	IFNET_WUNLOCK() do {						\
858	rw_wunlock(&ifnet_rwlock);					\
859	sx_xunlock(&ifnet_sxlock);					\
860} while (0)
861
862/*
863 * To assert the ifnet lock, you must know not only whether it's for read or
864 * write, but also whether it was acquired with sleep support or not.
865 */
866#define	IFNET_RLOCK_ASSERT()		sx_assert(&ifnet_sxlock, SA_SLOCKED)
867#define	IFNET_RLOCK_NOSLEEP_ASSERT()	rw_assert(&ifnet_rwlock, RA_RLOCKED)
868#define	IFNET_WLOCK_ASSERT() do {					\
869	sx_assert(&ifnet_sxlock, SA_XLOCKED);				\
870	rw_assert(&ifnet_rwlock, RA_WLOCKED);				\
871} while (0)
872
873#define	IFNET_RLOCK()		sx_slock(&ifnet_sxlock)
874#define	IFNET_RLOCK_NOSLEEP()	rw_rlock(&ifnet_rwlock)
875#define	IFNET_RUNLOCK()		sx_sunlock(&ifnet_sxlock)
876#define	IFNET_RUNLOCK_NOSLEEP()	rw_runlock(&ifnet_rwlock)
877
878/*
879 * Look up an ifnet given its index; the _ref variant also acquires a
880 * reference that must be freed using if_rele().  It is almost always a bug
881 * to call ifnet_byindex() instead if ifnet_byindex_ref().
882 */
883struct ifnet	*ifnet_byindex(u_short idx);
884struct ifnet	*ifnet_byindex_locked(u_short idx);
885struct ifnet	*ifnet_byindex_ref(u_short idx);
886
887/*
888 * Given the index, ifaddr_byindex() returns the one and only
889 * link-level ifaddr for the interface. You are not supposed to use
890 * it to traverse the list of addresses associated to the interface.
891 */
892struct ifaddr	*ifaddr_byindex(u_short idx);
893
894VNET_DECLARE(struct ifnethead, ifnet);
895VNET_DECLARE(struct ifgrouphead, ifg_head);
896VNET_DECLARE(int, if_index);
897VNET_DECLARE(struct ifnet *, loif);	/* first loopback interface */
898VNET_DECLARE(int, useloopback);
899
900#define	V_ifnet		VNET(ifnet)
901#define	V_ifg_head	VNET(ifg_head)
902#define	V_if_index	VNET(if_index)
903#define	V_loif		VNET(loif)
904#define	V_useloopback	VNET(useloopback)
905
906extern	int ifqmaxlen;
907
908int	if_addgroup(struct ifnet *, const char *);
909int	if_delgroup(struct ifnet *, const char *);
910int	if_addmulti(struct ifnet *, struct sockaddr *, struct ifmultiaddr **);
911int	if_allmulti(struct ifnet *, int);
912struct	ifnet* if_alloc(u_char);
913void	if_attach(struct ifnet *);
914void	if_dead(struct ifnet *);
915int	if_delmulti(struct ifnet *, struct sockaddr *);
916void	if_delmulti_ifma(struct ifmultiaddr *);
917void	if_detach(struct ifnet *);
918void	if_vmove(struct ifnet *, struct vnet *);
919void	if_purgeaddrs(struct ifnet *);
920void	if_delallmulti(struct ifnet *);
921void	if_down(struct ifnet *);
922struct ifmultiaddr *
923	if_findmulti(struct ifnet *, struct sockaddr *);
924void	if_free(struct ifnet *);
925void	if_initname(struct ifnet *, const char *, int);
926void	if_link_state_change(struct ifnet *, int);
927int	if_printf(struct ifnet *, const char *, ...) __printflike(2, 3);
928void	if_qflush(struct ifnet *);
929void	if_ref(struct ifnet *);
930void	if_rele(struct ifnet *);
931int	if_setlladdr(struct ifnet *, const u_char *, int);
932void	if_up(struct ifnet *);
933int	ifioctl(struct socket *, u_long, caddr_t, struct thread *);
934int	ifpromisc(struct ifnet *, int);
935struct	ifnet *ifunit(const char *);
936struct	ifnet *ifunit_ref(const char *);
937
938void	ifq_init(struct ifaltq *, struct ifnet *ifp);
939void	ifq_delete(struct ifaltq *);
940
941int	ifa_add_loopback_route(struct ifaddr *, struct sockaddr *);
942int	ifa_del_loopback_route(struct ifaddr *, struct sockaddr *);
943
944struct	ifaddr *ifa_ifwithaddr(struct sockaddr *);
945int		ifa_ifwithaddr_check(struct sockaddr *);
946struct	ifaddr *ifa_ifwithbroadaddr(struct sockaddr *);
947struct	ifaddr *ifa_ifwithdstaddr(struct sockaddr *);
948struct	ifaddr *ifa_ifwithnet(struct sockaddr *, int);
949struct	ifaddr *ifa_ifwithroute(int, struct sockaddr *, struct sockaddr *);
950struct	ifaddr *ifa_ifwithroute_fib(int, struct sockaddr *, struct sockaddr *, u_int);
951struct	ifaddr *ifaof_ifpforaddr(struct sockaddr *, struct ifnet *);
952int	ifa_preferred(struct ifaddr *, struct ifaddr *);
953
954int	if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen);
955
956typedef	void *if_com_alloc_t(u_char type, struct ifnet *ifp);
957typedef	void if_com_free_t(void *com, u_char type);
958void	if_register_com_alloc(u_char type, if_com_alloc_t *a, if_com_free_t *f);
959void	if_deregister_com_alloc(u_char type);
960
961#define IF_LLADDR(ifp)							\
962    LLADDR((struct sockaddr_dl *)((ifp)->if_addr->ifa_addr))
963
964#ifdef DEVICE_POLLING
965enum poll_cmd {	POLL_ONLY, POLL_AND_CHECK_STATUS };
966
967typedef	int poll_handler_t(struct ifnet *ifp, enum poll_cmd cmd, int count);
968int    ether_poll_register(poll_handler_t *h, struct ifnet *ifp);
969int    ether_poll_deregister(struct ifnet *ifp);
970#endif /* DEVICE_POLLING */
971
972#endif /* _KERNEL */
973
974#endif /* !_NET_IF_VAR_H_ */
975