ifq.h revision 223735
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 223735 2011-07-03 12:22:02Z bz $
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	u_int	if_fib;			/* interface FIB */
200
201	/*
202	 * Spare fields are added so that we can modify sensitive data
203	 * structures without changing the kernel binary interface, and must
204	 * be used with care where binary compatibility is required.
205	 */
206	char	 if_cspare[3];
207	char	*if_description;	/* interface description */
208	void	*if_pspare[7];
209	int	if_ispare[4];
210};
211
212typedef void if_init_f_t(void *);
213
214/*
215 * XXX These aliases are terribly dangerous because they could apply
216 * to anything.
217 */
218#define	if_mtu		if_data.ifi_mtu
219#define	if_type		if_data.ifi_type
220#define if_physical	if_data.ifi_physical
221#define	if_addrlen	if_data.ifi_addrlen
222#define	if_hdrlen	if_data.ifi_hdrlen
223#define	if_metric	if_data.ifi_metric
224#define	if_link_state	if_data.ifi_link_state
225#define	if_baudrate	if_data.ifi_baudrate
226#define	if_hwassist	if_data.ifi_hwassist
227#define	if_ipackets	if_data.ifi_ipackets
228#define	if_ierrors	if_data.ifi_ierrors
229#define	if_opackets	if_data.ifi_opackets
230#define	if_oerrors	if_data.ifi_oerrors
231#define	if_collisions	if_data.ifi_collisions
232#define	if_ibytes	if_data.ifi_ibytes
233#define	if_obytes	if_data.ifi_obytes
234#define	if_imcasts	if_data.ifi_imcasts
235#define	if_omcasts	if_data.ifi_omcasts
236#define	if_iqdrops	if_data.ifi_iqdrops
237#define	if_noproto	if_data.ifi_noproto
238#define	if_lastchange	if_data.ifi_lastchange
239
240/* for compatibility with other BSDs */
241#define	if_addrlist	if_addrhead
242#define	if_list		if_link
243#define	if_name(ifp)	((ifp)->if_xname)
244
245/*
246 * Locks for address lists on the network interface.
247 */
248#define	IF_ADDR_LOCK_INIT(if)	mtx_init(&(if)->if_addr_mtx,		\
249				    "if_addr_mtx", NULL, MTX_DEF)
250#define	IF_ADDR_LOCK_DESTROY(if)	mtx_destroy(&(if)->if_addr_mtx)
251#define	IF_ADDR_LOCK(if)	mtx_lock(&(if)->if_addr_mtx)
252#define	IF_ADDR_UNLOCK(if)	mtx_unlock(&(if)->if_addr_mtx)
253#define	IF_ADDR_LOCK_ASSERT(if)	mtx_assert(&(if)->if_addr_mtx, MA_OWNED)
254
255/*
256 * Function variations on locking macros intended to be used by loadable
257 * kernel modules in order to divorce them from the internals of address list
258 * locking.
259 */
260void	if_addr_rlock(struct ifnet *ifp);	/* if_addrhead */
261void	if_addr_runlock(struct ifnet *ifp);	/* if_addrhead */
262void	if_maddr_rlock(struct ifnet *ifp);	/* if_multiaddrs */
263void	if_maddr_runlock(struct ifnet *ifp);	/* if_multiaddrs */
264
265/*
266 * Output queues (ifp->if_snd) and slow device input queues (*ifp->if_slowq)
267 * are queues of messages stored on ifqueue structures
268 * (defined above).  Entries are added to and deleted from these structures
269 * by these macros, which should be called with ipl raised to splimp().
270 */
271#define IF_LOCK(ifq)		mtx_lock(&(ifq)->ifq_mtx)
272#define IF_UNLOCK(ifq)		mtx_unlock(&(ifq)->ifq_mtx)
273#define	IF_LOCK_ASSERT(ifq)	mtx_assert(&(ifq)->ifq_mtx, MA_OWNED)
274#define	_IF_QFULL(ifq)		((ifq)->ifq_len >= (ifq)->ifq_maxlen)
275#define	_IF_DROP(ifq)		((ifq)->ifq_drops++)
276#define	_IF_QLEN(ifq)		((ifq)->ifq_len)
277
278#define	_IF_ENQUEUE(ifq, m) do { 				\
279	(m)->m_nextpkt = NULL;					\
280	if ((ifq)->ifq_tail == NULL) 				\
281		(ifq)->ifq_head = m; 				\
282	else 							\
283		(ifq)->ifq_tail->m_nextpkt = m; 		\
284	(ifq)->ifq_tail = m; 					\
285	(ifq)->ifq_len++; 					\
286} while (0)
287
288#define IF_ENQUEUE(ifq, m) do {					\
289	IF_LOCK(ifq); 						\
290	_IF_ENQUEUE(ifq, m); 					\
291	IF_UNLOCK(ifq); 					\
292} while (0)
293
294#define	_IF_PREPEND(ifq, m) do {				\
295	(m)->m_nextpkt = (ifq)->ifq_head; 			\
296	if ((ifq)->ifq_tail == NULL) 				\
297		(ifq)->ifq_tail = (m); 				\
298	(ifq)->ifq_head = (m); 					\
299	(ifq)->ifq_len++; 					\
300} while (0)
301
302#define IF_PREPEND(ifq, m) do {		 			\
303	IF_LOCK(ifq); 						\
304	_IF_PREPEND(ifq, m); 					\
305	IF_UNLOCK(ifq); 					\
306} while (0)
307
308#define	_IF_DEQUEUE(ifq, m) do { 				\
309	(m) = (ifq)->ifq_head; 					\
310	if (m) { 						\
311		if (((ifq)->ifq_head = (m)->m_nextpkt) == NULL)	\
312			(ifq)->ifq_tail = NULL; 		\
313		(m)->m_nextpkt = NULL; 				\
314		(ifq)->ifq_len--; 				\
315	} 							\
316} while (0)
317
318#define IF_DEQUEUE(ifq, m) do { 				\
319	IF_LOCK(ifq); 						\
320	_IF_DEQUEUE(ifq, m); 					\
321	IF_UNLOCK(ifq); 					\
322} while (0)
323
324#define	_IF_POLL(ifq, m)	((m) = (ifq)->ifq_head)
325#define	IF_POLL(ifq, m)		_IF_POLL(ifq, m)
326
327#define _IF_DRAIN(ifq) do { 					\
328	struct mbuf *m; 					\
329	for (;;) { 						\
330		_IF_DEQUEUE(ifq, m); 				\
331		if (m == NULL) 					\
332			break; 					\
333		m_freem(m); 					\
334	} 							\
335} while (0)
336
337#define IF_DRAIN(ifq) do {					\
338	IF_LOCK(ifq);						\
339	_IF_DRAIN(ifq);						\
340	IF_UNLOCK(ifq);						\
341} while(0)
342
343#ifdef _KERNEL
344/* interface link layer address change event */
345typedef void (*iflladdr_event_handler_t)(void *, struct ifnet *);
346EVENTHANDLER_DECLARE(iflladdr_event, iflladdr_event_handler_t);
347/* interface address change event */
348typedef void (*ifaddr_event_handler_t)(void *, struct ifnet *);
349EVENTHANDLER_DECLARE(ifaddr_event, ifaddr_event_handler_t);
350/* new interface arrival event */
351typedef void (*ifnet_arrival_event_handler_t)(void *, struct ifnet *);
352EVENTHANDLER_DECLARE(ifnet_arrival_event, ifnet_arrival_event_handler_t);
353/* interface departure event */
354typedef void (*ifnet_departure_event_handler_t)(void *, struct ifnet *);
355EVENTHANDLER_DECLARE(ifnet_departure_event, ifnet_departure_event_handler_t);
356/* Interface link state change event */
357typedef void (*ifnet_link_event_handler_t)(void *, struct ifnet *, int);
358EVENTHANDLER_DECLARE(ifnet_link_event, ifnet_link_event_handler_t);
359
360/*
361 * interface groups
362 */
363struct ifg_group {
364	char				 ifg_group[IFNAMSIZ];
365	u_int				 ifg_refcnt;
366	void				*ifg_pf_kif;
367	TAILQ_HEAD(, ifg_member)	 ifg_members;
368	TAILQ_ENTRY(ifg_group)		 ifg_next;
369};
370
371struct ifg_member {
372	TAILQ_ENTRY(ifg_member)	 ifgm_next;
373	struct ifnet		*ifgm_ifp;
374};
375
376struct ifg_list {
377	struct ifg_group	*ifgl_group;
378	TAILQ_ENTRY(ifg_list)	 ifgl_next;
379};
380
381/* group attach event */
382typedef void (*group_attach_event_handler_t)(void *, struct ifg_group *);
383EVENTHANDLER_DECLARE(group_attach_event, group_attach_event_handler_t);
384/* group detach event */
385typedef void (*group_detach_event_handler_t)(void *, struct ifg_group *);
386EVENTHANDLER_DECLARE(group_detach_event, group_detach_event_handler_t);
387/* group change event */
388typedef void (*group_change_event_handler_t)(void *, const char *);
389EVENTHANDLER_DECLARE(group_change_event, group_change_event_handler_t);
390
391#define	IF_AFDATA_LOCK_INIT(ifp)	\
392	rw_init(&(ifp)->if_afdata_lock, "if_afdata")
393
394#define	IF_AFDATA_WLOCK(ifp)	rw_wlock(&(ifp)->if_afdata_lock)
395#define	IF_AFDATA_RLOCK(ifp)	rw_rlock(&(ifp)->if_afdata_lock)
396#define	IF_AFDATA_WUNLOCK(ifp)	rw_wunlock(&(ifp)->if_afdata_lock)
397#define	IF_AFDATA_RUNLOCK(ifp)	rw_runlock(&(ifp)->if_afdata_lock)
398#define	IF_AFDATA_LOCK(ifp)	IF_AFDATA_WLOCK(ifp)
399#define	IF_AFDATA_UNLOCK(ifp)	IF_AFDATA_WUNLOCK(ifp)
400#define	IF_AFDATA_TRYLOCK(ifp)	rw_try_wlock(&(ifp)->if_afdata_lock)
401#define	IF_AFDATA_DESTROY(ifp)	rw_destroy(&(ifp)->if_afdata_lock)
402
403#define	IF_AFDATA_LOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_LOCKED)
404#define	IF_AFDATA_UNLOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_UNLOCKED)
405
406int	if_handoff(struct ifqueue *ifq, struct mbuf *m, struct ifnet *ifp,
407	    int adjust);
408#define	IF_HANDOFF(ifq, m, ifp)			\
409	if_handoff((struct ifqueue *)ifq, m, ifp, 0)
410#define	IF_HANDOFF_ADJ(ifq, m, ifp, adj)	\
411	if_handoff((struct ifqueue *)ifq, m, ifp, adj)
412
413void	if_start(struct ifnet *);
414
415#define	IFQ_ENQUEUE(ifq, m, err)					\
416do {									\
417	IF_LOCK(ifq);							\
418	if (ALTQ_IS_ENABLED(ifq))					\
419		ALTQ_ENQUEUE(ifq, m, NULL, err);			\
420	else {								\
421		if (_IF_QFULL(ifq)) {					\
422			m_freem(m);					\
423			(err) = ENOBUFS;				\
424		} else {						\
425			_IF_ENQUEUE(ifq, m);				\
426			(err) = 0;					\
427		}							\
428	}								\
429	if (err)							\
430		(ifq)->ifq_drops++;					\
431	IF_UNLOCK(ifq);							\
432} while (0)
433
434#define	IFQ_DEQUEUE_NOLOCK(ifq, m)					\
435do {									\
436	if (TBR_IS_ENABLED(ifq))					\
437		(m) = tbr_dequeue_ptr(ifq, ALTDQ_REMOVE);		\
438	else if (ALTQ_IS_ENABLED(ifq))					\
439		ALTQ_DEQUEUE(ifq, m);					\
440	else								\
441		_IF_DEQUEUE(ifq, m);					\
442} while (0)
443
444#define	IFQ_DEQUEUE(ifq, m)						\
445do {									\
446	IF_LOCK(ifq);							\
447	IFQ_DEQUEUE_NOLOCK(ifq, m);					\
448	IF_UNLOCK(ifq);							\
449} while (0)
450
451#define	IFQ_POLL_NOLOCK(ifq, m)						\
452do {									\
453	if (TBR_IS_ENABLED(ifq))					\
454		(m) = tbr_dequeue_ptr(ifq, ALTDQ_POLL);			\
455	else if (ALTQ_IS_ENABLED(ifq))					\
456		ALTQ_POLL(ifq, m);					\
457	else								\
458		_IF_POLL(ifq, m);					\
459} while (0)
460
461#define	IFQ_POLL(ifq, m)						\
462do {									\
463	IF_LOCK(ifq);							\
464	IFQ_POLL_NOLOCK(ifq, m);					\
465	IF_UNLOCK(ifq);							\
466} while (0)
467
468#define	IFQ_PURGE_NOLOCK(ifq)						\
469do {									\
470	if (ALTQ_IS_ENABLED(ifq)) {					\
471		ALTQ_PURGE(ifq);					\
472	} else								\
473		_IF_DRAIN(ifq);						\
474} while (0)
475
476#define	IFQ_PURGE(ifq)							\
477do {									\
478	IF_LOCK(ifq);							\
479	IFQ_PURGE_NOLOCK(ifq);						\
480	IF_UNLOCK(ifq);							\
481} while (0)
482
483#define	IFQ_SET_READY(ifq)						\
484	do { ((ifq)->altq_flags |= ALTQF_READY); } while (0)
485
486#define	IFQ_LOCK(ifq)			IF_LOCK(ifq)
487#define	IFQ_UNLOCK(ifq)			IF_UNLOCK(ifq)
488#define	IFQ_LOCK_ASSERT(ifq)		IF_LOCK_ASSERT(ifq)
489#define	IFQ_IS_EMPTY(ifq)		((ifq)->ifq_len == 0)
490#define	IFQ_INC_LEN(ifq)		((ifq)->ifq_len++)
491#define	IFQ_DEC_LEN(ifq)		(--(ifq)->ifq_len)
492#define	IFQ_INC_DROPS(ifq)		((ifq)->ifq_drops++)
493#define	IFQ_SET_MAXLEN(ifq, len)	((ifq)->ifq_maxlen = (len))
494
495/*
496 * The IFF_DRV_OACTIVE test should really occur in the device driver, not in
497 * the handoff logic, as that flag is locked by the device driver.
498 */
499#define	IFQ_HANDOFF_ADJ(ifp, m, adj, err)				\
500do {									\
501	int len;							\
502	short mflags;							\
503									\
504	len = (m)->m_pkthdr.len;					\
505	mflags = (m)->m_flags;						\
506	IFQ_ENQUEUE(&(ifp)->if_snd, m, err);				\
507	if ((err) == 0) {						\
508		(ifp)->if_obytes += len + (adj);			\
509		if (mflags & M_MCAST)					\
510			(ifp)->if_omcasts++;				\
511		if (((ifp)->if_drv_flags & IFF_DRV_OACTIVE) == 0)	\
512			if_start(ifp);					\
513	}								\
514} while (0)
515
516#define	IFQ_HANDOFF(ifp, m, err)					\
517	IFQ_HANDOFF_ADJ(ifp, m, 0, err)
518
519#define	IFQ_DRV_DEQUEUE(ifq, m)						\
520do {									\
521	(m) = (ifq)->ifq_drv_head;					\
522	if (m) {							\
523		if (((ifq)->ifq_drv_head = (m)->m_nextpkt) == NULL)	\
524			(ifq)->ifq_drv_tail = NULL;			\
525		(m)->m_nextpkt = NULL;					\
526		(ifq)->ifq_drv_len--;					\
527	} else {							\
528		IFQ_LOCK(ifq);						\
529		IFQ_DEQUEUE_NOLOCK(ifq, m);				\
530		while ((ifq)->ifq_drv_len < (ifq)->ifq_drv_maxlen) {	\
531			struct mbuf *m0;				\
532			IFQ_DEQUEUE_NOLOCK(ifq, m0);			\
533			if (m0 == NULL)					\
534				break;					\
535			m0->m_nextpkt = NULL;				\
536			if ((ifq)->ifq_drv_tail == NULL)		\
537				(ifq)->ifq_drv_head = m0;		\
538			else						\
539				(ifq)->ifq_drv_tail->m_nextpkt = m0;	\
540			(ifq)->ifq_drv_tail = m0;			\
541			(ifq)->ifq_drv_len++;				\
542		}							\
543		IFQ_UNLOCK(ifq);					\
544	}								\
545} while (0)
546
547#define	IFQ_DRV_PREPEND(ifq, m)						\
548do {									\
549	(m)->m_nextpkt = (ifq)->ifq_drv_head;				\
550	if ((ifq)->ifq_drv_tail == NULL)				\
551		(ifq)->ifq_drv_tail = (m);				\
552	(ifq)->ifq_drv_head = (m);					\
553	(ifq)->ifq_drv_len++;						\
554} while (0)
555
556#define	IFQ_DRV_IS_EMPTY(ifq)						\
557	(((ifq)->ifq_drv_len == 0) && ((ifq)->ifq_len == 0))
558
559#define	IFQ_DRV_PURGE(ifq)						\
560do {									\
561	struct mbuf *m, *n = (ifq)->ifq_drv_head;			\
562	while((m = n) != NULL) {					\
563		n = m->m_nextpkt;					\
564		m_freem(m);						\
565	}								\
566	(ifq)->ifq_drv_head = (ifq)->ifq_drv_tail = NULL;		\
567	(ifq)->ifq_drv_len = 0;						\
568	IFQ_PURGE(ifq);							\
569} while (0)
570
571#ifdef _KERNEL
572static __inline void
573drbr_stats_update(struct ifnet *ifp, int len, int mflags)
574{
575#ifndef NO_SLOW_STATS
576	ifp->if_obytes += len;
577	if (mflags & M_MCAST)
578		ifp->if_omcasts++;
579#endif
580}
581
582static __inline int
583drbr_enqueue(struct ifnet *ifp, struct buf_ring *br, struct mbuf *m)
584{
585	int error = 0;
586	int len = m->m_pkthdr.len;
587	int mflags = m->m_flags;
588
589#ifdef ALTQ
590	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
591		IFQ_ENQUEUE(&ifp->if_snd, m, error);
592		return (error);
593	}
594#endif
595	if ((error = buf_ring_enqueue_bytes(br, m, len)) == ENOBUFS) {
596		br->br_drops++;
597		m_freem(m);
598	} else
599		drbr_stats_update(ifp, len, mflags);
600
601	return (error);
602}
603
604static __inline void
605drbr_flush(struct ifnet *ifp, struct buf_ring *br)
606{
607	struct mbuf *m;
608
609#ifdef ALTQ
610	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd))
611		IFQ_PURGE(&ifp->if_snd);
612#endif
613	while ((m = buf_ring_dequeue_sc(br)) != NULL)
614		m_freem(m);
615}
616
617static __inline void
618drbr_free(struct buf_ring *br, struct malloc_type *type)
619{
620
621	drbr_flush(NULL, br);
622	buf_ring_free(br, type);
623}
624
625static __inline struct mbuf *
626drbr_dequeue(struct ifnet *ifp, struct buf_ring *br)
627{
628#ifdef ALTQ
629	struct mbuf *m;
630
631	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
632		IFQ_DEQUEUE(&ifp->if_snd, m);
633		return (m);
634	}
635#endif
636	return (buf_ring_dequeue_sc(br));
637}
638
639static __inline struct mbuf *
640drbr_dequeue_cond(struct ifnet *ifp, struct buf_ring *br,
641    int (*func) (struct mbuf *, void *), void *arg)
642{
643	struct mbuf *m;
644#ifdef ALTQ
645	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
646		IFQ_LOCK(&ifp->if_snd);
647		IFQ_POLL_NOLOCK(&ifp->if_snd, m);
648		if (m != NULL && func(m, arg) == 0) {
649			IFQ_UNLOCK(&ifp->if_snd);
650			return (NULL);
651		}
652		IFQ_DEQUEUE_NOLOCK(&ifp->if_snd, m);
653		IFQ_UNLOCK(&ifp->if_snd);
654		return (m);
655	}
656#endif
657	m = buf_ring_peek(br);
658	if (m == NULL || func(m, arg) == 0)
659		return (NULL);
660
661	return (buf_ring_dequeue_sc(br));
662}
663
664static __inline int
665drbr_empty(struct ifnet *ifp, struct buf_ring *br)
666{
667#ifdef ALTQ
668	if (ALTQ_IS_ENABLED(&ifp->if_snd))
669		return (IFQ_IS_EMPTY(&ifp->if_snd));
670#endif
671	return (buf_ring_empty(br));
672}
673
674static __inline int
675drbr_needs_enqueue(struct ifnet *ifp, struct buf_ring *br)
676{
677#ifdef ALTQ
678	if (ALTQ_IS_ENABLED(&ifp->if_snd))
679		return (1);
680#endif
681	return (!buf_ring_empty(br));
682}
683
684static __inline int
685drbr_inuse(struct ifnet *ifp, struct buf_ring *br)
686{
687#ifdef ALTQ
688	if (ALTQ_IS_ENABLED(&ifp->if_snd))
689		return (ifp->if_snd.ifq_len);
690#endif
691	return (buf_ring_count(br));
692}
693#endif
694/*
695 * 72 was chosen below because it is the size of a TCP/IP
696 * header (40) + the minimum mss (32).
697 */
698#define	IF_MINMTU	72
699#define	IF_MAXMTU	65535
700
701#endif /* _KERNEL */
702
703/*
704 * The ifaddr structure contains information about one address
705 * of an interface.  They are maintained by the different address families,
706 * are allocated and attached when an address is set, and are linked
707 * together so all addresses for an interface can be located.
708 *
709 * NOTE: a 'struct ifaddr' is always at the beginning of a larger
710 * chunk of malloc'ed memory, where we store the three addresses
711 * (ifa_addr, ifa_dstaddr and ifa_netmask) referenced here.
712 */
713struct ifaddr {
714	struct	sockaddr *ifa_addr;	/* address of interface */
715	struct	sockaddr *ifa_dstaddr;	/* other end of p-to-p link */
716#define	ifa_broadaddr	ifa_dstaddr	/* broadcast address interface */
717	struct	sockaddr *ifa_netmask;	/* used to determine subnet */
718	struct	if_data if_data;	/* not all members are meaningful */
719	struct	ifnet *ifa_ifp;		/* back-pointer to interface */
720	TAILQ_ENTRY(ifaddr) ifa_link;	/* queue macro glue */
721	void	(*ifa_rtrequest)	/* check or clean routes (+ or -)'d */
722		(int, struct rtentry *, struct rt_addrinfo *);
723	u_short	ifa_flags;		/* mostly rt_flags for cloning */
724	u_int	ifa_refcnt;		/* references to this structure */
725	int	ifa_metric;		/* cost of going out this interface */
726	int (*ifa_claim_addr)		/* check if an addr goes to this if */
727		(struct ifaddr *, struct sockaddr *);
728	struct mtx ifa_mtx;
729};
730#define	IFA_ROUTE	RTF_UP		/* route installed */
731#define IFA_RTSELF	RTF_HOST	/* loopback route to self installed */
732
733/* for compatibility with other BSDs */
734#define	ifa_list	ifa_link
735
736#ifdef _KERNEL
737#define	IFA_LOCK(ifa)		mtx_lock(&(ifa)->ifa_mtx)
738#define	IFA_UNLOCK(ifa)		mtx_unlock(&(ifa)->ifa_mtx)
739
740void	ifa_free(struct ifaddr *ifa);
741void	ifa_init(struct ifaddr *ifa);
742void	ifa_ref(struct ifaddr *ifa);
743#endif
744
745/*
746 * The prefix structure contains information about one prefix
747 * of an interface.  They are maintained by the different address families,
748 * are allocated and attached when a prefix or an address is set,
749 * and are linked together so all prefixes for an interface can be located.
750 */
751struct ifprefix {
752	struct	sockaddr *ifpr_prefix;	/* prefix of interface */
753	struct	ifnet *ifpr_ifp;	/* back-pointer to interface */
754	TAILQ_ENTRY(ifprefix) ifpr_list; /* queue macro glue */
755	u_char	ifpr_plen;		/* prefix length in bits */
756	u_char	ifpr_type;		/* protocol dependent prefix type */
757};
758
759/*
760 * Multicast address structure.  This is analogous to the ifaddr
761 * structure except that it keeps track of multicast addresses.
762 */
763struct ifmultiaddr {
764	TAILQ_ENTRY(ifmultiaddr) ifma_link; /* queue macro glue */
765	struct	sockaddr *ifma_addr; 	/* address this membership is for */
766	struct	sockaddr *ifma_lladdr;	/* link-layer translation, if any */
767	struct	ifnet *ifma_ifp;	/* back-pointer to interface */
768	u_int	ifma_refcount;		/* reference count */
769	void	*ifma_protospec;	/* protocol-specific state, if any */
770	struct	ifmultiaddr *ifma_llifma; /* pointer to ifma for ifma_lladdr */
771};
772
773#ifdef _KERNEL
774
775extern	struct rwlock ifnet_rwlock;
776extern	struct sx ifnet_sxlock;
777
778#define	IFNET_LOCK_INIT() do {						\
779	rw_init_flags(&ifnet_rwlock, "ifnet_rw",  RW_RECURSE);		\
780	sx_init_flags(&ifnet_sxlock, "ifnet_sx",  SX_RECURSE);		\
781} while(0)
782
783#define	IFNET_WLOCK() do {						\
784	sx_xlock(&ifnet_sxlock);					\
785	rw_wlock(&ifnet_rwlock);					\
786} while (0)
787
788#define	IFNET_WUNLOCK() do {						\
789	rw_wunlock(&ifnet_rwlock);					\
790	sx_xunlock(&ifnet_sxlock);					\
791} while (0)
792
793/*
794 * To assert the ifnet lock, you must know not only whether it's for read or
795 * write, but also whether it was acquired with sleep support or not.
796 */
797#define	IFNET_RLOCK_ASSERT()		sx_assert(&ifnet_sxlock, SA_SLOCKED)
798#define	IFNET_RLOCK_NOSLEEP_ASSERT()	rw_assert(&ifnet_rwlock, RA_RLOCKED)
799#define	IFNET_WLOCK_ASSERT() do {					\
800	sx_assert(&ifnet_sxlock, SA_XLOCKED);				\
801	rw_assert(&ifnet_rwlock, RA_WLOCKED);				\
802} while (0)
803
804#define	IFNET_RLOCK()		sx_slock(&ifnet_sxlock)
805#define	IFNET_RLOCK_NOSLEEP()	rw_rlock(&ifnet_rwlock)
806#define	IFNET_RUNLOCK()		sx_sunlock(&ifnet_sxlock)
807#define	IFNET_RUNLOCK_NOSLEEP()	rw_runlock(&ifnet_rwlock)
808
809/*
810 * Look up an ifnet given its index; the _ref variant also acquires a
811 * reference that must be freed using if_rele().  It is almost always a bug
812 * to call ifnet_byindex() instead if ifnet_byindex_ref().
813 */
814struct ifnet	*ifnet_byindex(u_short idx);
815struct ifnet	*ifnet_byindex_locked(u_short idx);
816struct ifnet	*ifnet_byindex_ref(u_short idx);
817
818/*
819 * Given the index, ifaddr_byindex() returns the one and only
820 * link-level ifaddr for the interface. You are not supposed to use
821 * it to traverse the list of addresses associated to the interface.
822 */
823struct ifaddr	*ifaddr_byindex(u_short idx);
824
825VNET_DECLARE(struct ifnethead, ifnet);
826VNET_DECLARE(struct ifgrouphead, ifg_head);
827VNET_DECLARE(int, if_index);
828VNET_DECLARE(struct ifnet *, loif);	/* first loopback interface */
829VNET_DECLARE(int, useloopback);
830
831#define	V_ifnet		VNET(ifnet)
832#define	V_ifg_head	VNET(ifg_head)
833#define	V_if_index	VNET(if_index)
834#define	V_loif		VNET(loif)
835#define	V_useloopback	VNET(useloopback)
836
837extern	int ifqmaxlen;
838
839int	if_addgroup(struct ifnet *, const char *);
840int	if_delgroup(struct ifnet *, const char *);
841int	if_addmulti(struct ifnet *, struct sockaddr *, struct ifmultiaddr **);
842int	if_allmulti(struct ifnet *, int);
843struct	ifnet* if_alloc(u_char);
844void	if_attach(struct ifnet *);
845void	if_dead(struct ifnet *);
846int	if_delmulti(struct ifnet *, struct sockaddr *);
847void	if_delmulti_ifma(struct ifmultiaddr *);
848void	if_detach(struct ifnet *);
849void	if_vmove(struct ifnet *, struct vnet *);
850void	if_purgeaddrs(struct ifnet *);
851void	if_delallmulti(struct ifnet *);
852void	if_down(struct ifnet *);
853struct ifmultiaddr *
854	if_findmulti(struct ifnet *, struct sockaddr *);
855void	if_free(struct ifnet *);
856void	if_free_type(struct ifnet *, u_char);
857void	if_initname(struct ifnet *, const char *, int);
858void	if_link_state_change(struct ifnet *, int);
859int	if_printf(struct ifnet *, const char *, ...) __printflike(2, 3);
860void	if_qflush(struct ifnet *);
861void	if_ref(struct ifnet *);
862void	if_rele(struct ifnet *);
863int	if_setlladdr(struct ifnet *, const u_char *, int);
864void	if_up(struct ifnet *);
865int	ifioctl(struct socket *, u_long, caddr_t, struct thread *);
866int	ifpromisc(struct ifnet *, int);
867struct	ifnet *ifunit(const char *);
868struct	ifnet *ifunit_ref(const char *);
869
870void	ifq_init(struct ifaltq *, struct ifnet *ifp);
871void	ifq_delete(struct ifaltq *);
872
873int	ifa_add_loopback_route(struct ifaddr *, struct sockaddr *);
874int	ifa_del_loopback_route(struct ifaddr *, struct sockaddr *);
875
876struct	ifaddr *ifa_ifwithaddr(struct sockaddr *);
877int		ifa_ifwithaddr_check(struct sockaddr *);
878struct	ifaddr *ifa_ifwithbroadaddr(struct sockaddr *);
879struct	ifaddr *ifa_ifwithdstaddr(struct sockaddr *);
880struct	ifaddr *ifa_ifwithnet(struct sockaddr *, int);
881struct	ifaddr *ifa_ifwithroute(int, struct sockaddr *, struct sockaddr *);
882struct	ifaddr *ifa_ifwithroute_fib(int, struct sockaddr *, struct sockaddr *, u_int);
883
884struct	ifaddr *ifaof_ifpforaddr(struct sockaddr *, struct ifnet *);
885
886int	if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen);
887
888typedef	void *if_com_alloc_t(u_char type, struct ifnet *ifp);
889typedef	void if_com_free_t(void *com, u_char type);
890void	if_register_com_alloc(u_char type, if_com_alloc_t *a, if_com_free_t *f);
891void	if_deregister_com_alloc(u_char type);
892
893#define IF_LLADDR(ifp)							\
894    LLADDR((struct sockaddr_dl *)((ifp)->if_addr->ifa_addr))
895
896#ifdef DEVICE_POLLING
897enum poll_cmd {	POLL_ONLY, POLL_AND_CHECK_STATUS };
898
899typedef	int poll_handler_t(struct ifnet *ifp, enum poll_cmd cmd, int count);
900int    ether_poll_register(poll_handler_t *h, struct ifnet *ifp);
901int    ether_poll_deregister(struct ifnet *ifp);
902#endif /* DEVICE_POLLING */
903
904#endif /* _KERNEL */
905
906#endif /* !_NET_IF_VAR_H_ */
907