if_var.h revision 274043
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: stable/10/sys/net/if_var.h 274043 2014-11-03 12:38:29Z hselasky $
31 */
32
33#ifndef	_NET_IF_VAR_H_
34#define	_NET_IF_VAR_H_
35
36/*
37 * Structures defining a network interface, providing a packet
38 * transport mechanism (ala level 0 of the PUP protocols).
39 *
40 * Each interface accepts output datagrams of a specified maximum
41 * length, and provides higher level routines with input datagrams
42 * received from its medium.
43 *
44 * Output occurs when the routine if_output is called, with three parameters:
45 *	(*ifp->if_output)(ifp, m, dst, rt)
46 * Here m is the mbuf chain to be sent and dst is the destination address.
47 * The output routine encapsulates the supplied datagram if necessary,
48 * and then transmits it on its medium.
49 *
50 * On input, each interface unwraps the data received by it, and either
51 * places it on the input queue of an internetwork datagram routine
52 * and posts the associated software interrupt, or passes the datagram to a raw
53 * packet input routine.
54 *
55 * Routines exist for locating interfaces by their addresses
56 * or for locating an interface on a certain network, as well as more general
57 * routing and gateway routines maintaining information used to locate
58 * interfaces.  These routines live in the files if.c and route.c
59 */
60
61#ifdef __STDC__
62/*
63 * Forward structure declarations for function prototypes [sic].
64 */
65struct	mbuf;
66struct	thread;
67struct	rtentry;
68struct	rt_addrinfo;
69struct	socket;
70struct	ether_header;
71struct	carp_if;
72struct	carp_softc;
73struct  ifvlantrunk;
74struct	route;
75struct	vnet;
76#endif
77
78#include <sys/queue.h>		/* get TAILQ macros */
79
80#ifdef _KERNEL
81#include <sys/mbuf.h>
82#include <sys/eventhandler.h>
83#include <sys/buf_ring.h>
84#include <net/vnet.h>
85#endif /* _KERNEL */
86#include <sys/lock.h>		/* XXX */
87#include <sys/mutex.h>		/* XXX */
88#include <sys/rwlock.h>		/* XXX */
89#include <sys/sx.h>		/* XXX */
90#include <sys/event.h>		/* XXX */
91#include <sys/_task.h>
92
93#define	IF_DUNIT_NONE	-1
94
95#include <altq/if_altq.h>
96
97TAILQ_HEAD(ifnethead, ifnet);	/* we use TAILQs so that the order of */
98TAILQ_HEAD(ifaddrhead, ifaddr);	/* instantiation is preserved in the list */
99TAILQ_HEAD(ifmultihead, ifmultiaddr);
100TAILQ_HEAD(ifgrouphead, ifg_group);
101
102#ifdef _KERNEL
103VNET_DECLARE(struct pfil_head, link_pfil_hook);	/* packet filter hooks */
104#define	V_link_pfil_hook	VNET(link_pfil_hook)
105#endif /* _KERNEL */
106
107/*
108 * Structure defining a queue for a network interface.
109 */
110struct	ifqueue {
111	struct	mbuf *ifq_head;
112	struct	mbuf *ifq_tail;
113	int	ifq_len;
114	int	ifq_maxlen;
115	int	ifq_drops;
116	struct	mtx ifq_mtx;
117};
118
119struct ifnet_hw_tsomax {
120	u_int	tsomaxbytes;	/* TSO total burst length limit in bytes */
121	u_int	tsomaxsegcount;	/* TSO maximum segment count */
122	u_int	tsomaxsegsize;	/* TSO maximum segment size in bytes */
123};
124
125/*
126 * Structure defining a network interface.
127 *
128 * (Would like to call this struct ``if'', but C isn't PL/1.)
129 */
130
131struct ifnet {
132	void	*if_softc;		/* pointer to driver state */
133	void	*if_l2com;		/* pointer to protocol bits */
134	struct vnet *if_vnet;		/* pointer to network stack instance */
135	TAILQ_ENTRY(ifnet) if_link; 	/* all struct ifnets are chained */
136	char	if_xname[IFNAMSIZ];	/* external name (name + unit) */
137	const char *if_dname;		/* driver name */
138	int	if_dunit;		/* unit or IF_DUNIT_NONE */
139	u_int	if_refcount;		/* reference count */
140	struct	ifaddrhead if_addrhead;	/* linked list of addresses per if */
141		/*
142		 * if_addrhead is the list of all addresses associated to
143		 * an interface.
144		 * Some code in the kernel assumes that first element
145		 * of the list has type AF_LINK, and contains sockaddr_dl
146		 * addresses which store the link-level address and the name
147		 * of the interface.
148		 * However, access to the AF_LINK address through this
149		 * field is deprecated. Use if_addr or ifaddr_byindex() instead.
150		 */
151	int	if_pcount;		/* number of promiscuous listeners */
152	struct	carp_if *if_carp;	/* carp interface structure */
153	struct	bpf_if *if_bpf;		/* packet filter structure */
154	u_short	if_index;		/* numeric abbreviation for this if  */
155	short	if_index_reserved;	/* spare space to grow if_index */
156	struct  ifvlantrunk *if_vlantrunk; /* pointer to 802.1q data */
157	int	if_flags;		/* up/down, broadcast, etc. */
158	int	if_capabilities;	/* interface features & capabilities */
159	int	if_capenable;		/* enabled features & capabilities */
160	void	*if_linkmib;		/* link-type-specific MIB data */
161	size_t	if_linkmiblen;		/* length of above data */
162	struct	if_data if_data;
163	struct	ifmultihead if_multiaddrs; /* multicast addresses configured */
164	int	if_amcount;		/* number of all-multicast requests */
165/* procedure handles */
166	int	(*if_output)		/* output routine (enqueue) */
167		(struct ifnet *, struct mbuf *, const struct sockaddr *,
168		     struct route *);
169	void	(*if_input)		/* input routine (from h/w driver) */
170		(struct ifnet *, struct mbuf *);
171	void	(*if_start)		/* initiate output routine */
172		(struct ifnet *);
173	int	(*if_ioctl)		/* ioctl routine */
174		(struct ifnet *, u_long, caddr_t);
175	void	(*if_init)		/* Init routine */
176		(void *);
177	int	(*if_resolvemulti)	/* validate/resolve multicast */
178		(struct ifnet *, struct sockaddr **, struct sockaddr *);
179	void	(*if_qflush)		/* flush any queues */
180		(struct ifnet *);
181	int	(*if_transmit)		/* initiate output routine */
182		(struct ifnet *, struct mbuf *);
183	void	(*if_reassign)		/* reassign to vnet routine */
184		(struct ifnet *, struct vnet *, char *);
185	struct	vnet *if_home_vnet;	/* where this ifnet originates from */
186	struct	ifaddr	*if_addr;	/* pointer to link-level address */
187	void	*if_llsoftc;		/* link layer softc */
188	int	if_drv_flags;		/* driver-managed status flags */
189	struct  ifaltq if_snd;		/* output queue (includes altq) */
190	const u_int8_t *if_broadcastaddr; /* linklevel broadcast bytestring */
191
192	void	*if_bridge;		/* bridge glue */
193
194	struct	label *if_label;	/* interface MAC label */
195
196	/* these are only used by IPv6 */
197	void	*if_unused[2];
198	void	*if_afdata[AF_MAX];
199	int	if_afdata_initialized;
200	struct	rwlock if_afdata_lock;
201	struct	task if_linktask;	/* task for link change events */
202	struct	rwlock if_addr_lock;	/* lock to protect address lists */
203
204	LIST_ENTRY(ifnet) if_clones;	/* interfaces of a cloner */
205	TAILQ_HEAD(, ifg_list) if_groups; /* linked list of groups per if */
206					/* protected by if_addr_lock */
207	void	*if_pf_kif;
208	void	*if_lagg;		/* lagg glue */
209	char	*if_description;	/* interface description */
210	u_int	if_fib;			/* interface FIB */
211	u_char	if_alloctype;		/* if_type at time of allocation */
212
213	u_int	if_hw_tsomax;		/* TSO total burst length
214					 * limit in bytes. A value of
215					 * zero means no limit. Have
216					 * to find a better place for
217					 * it eventually. */
218
219	/*
220	 * Spare fields are added so that we can modify sensitive data
221	 * structures without changing the kernel binary interface, and must
222	 * be used with care where binary compatibility is required.
223	 */
224	char	if_cspare[3];
225	int	if_ispare[2];
226
227	/*
228	 * TSO fields for segment limits. If a field is zero below,
229	 * there is no limit:
230	 */
231	u_int	if_hw_tsomaxsegcount;	/* TSO maximum segment count */
232	u_int	if_hw_tsomaxsegsize;	/* TSO maximum segment size in bytes */
233	void	*if_pspare[8];		/* 1 netmap, 7 TDB */
234};
235
236typedef void if_init_f_t(void *);
237
238/*
239 * XXX These aliases are terribly dangerous because they could apply
240 * to anything.
241 */
242#define	if_mtu		if_data.ifi_mtu
243#define	if_type		if_data.ifi_type
244#define if_physical	if_data.ifi_physical
245#define	if_addrlen	if_data.ifi_addrlen
246#define	if_hdrlen	if_data.ifi_hdrlen
247#define	if_metric	if_data.ifi_metric
248#define	if_link_state	if_data.ifi_link_state
249#define	if_baudrate	if_data.ifi_baudrate
250#define	if_baudrate_pf	if_data.ifi_baudrate_pf
251#define	if_hwassist	if_data.ifi_hwassist
252#define	if_ipackets	if_data.ifi_ipackets
253#define	if_ierrors	if_data.ifi_ierrors
254#define	if_opackets	if_data.ifi_opackets
255#define	if_oerrors	if_data.ifi_oerrors
256#define	if_collisions	if_data.ifi_collisions
257#define	if_ibytes	if_data.ifi_ibytes
258#define	if_obytes	if_data.ifi_obytes
259#define	if_imcasts	if_data.ifi_imcasts
260#define	if_omcasts	if_data.ifi_omcasts
261#define	if_iqdrops	if_data.ifi_iqdrops
262#define	if_noproto	if_data.ifi_noproto
263#define	if_lastchange	if_data.ifi_lastchange
264
265/* for compatibility with other BSDs */
266#define	if_addrlist	if_addrhead
267#define	if_list		if_link
268#define	if_name(ifp)	((ifp)->if_xname)
269
270/*
271 * Locks for address lists on the network interface.
272 */
273#define	IF_ADDR_LOCK_INIT(if)	rw_init(&(if)->if_addr_lock, "if_addr_lock")
274#define	IF_ADDR_LOCK_DESTROY(if)	rw_destroy(&(if)->if_addr_lock)
275#define	IF_ADDR_WLOCK(if)	rw_wlock(&(if)->if_addr_lock)
276#define	IF_ADDR_WUNLOCK(if)	rw_wunlock(&(if)->if_addr_lock)
277#define	IF_ADDR_RLOCK(if)	rw_rlock(&(if)->if_addr_lock)
278#define	IF_ADDR_RUNLOCK(if)	rw_runlock(&(if)->if_addr_lock)
279#define	IF_ADDR_LOCK_ASSERT(if)	rw_assert(&(if)->if_addr_lock, RA_LOCKED)
280#define	IF_ADDR_WLOCK_ASSERT(if) rw_assert(&(if)->if_addr_lock, RA_WLOCKED)
281
282/*
283 * Function variations on locking macros intended to be used by loadable
284 * kernel modules in order to divorce them from the internals of address list
285 * locking.
286 */
287void	if_addr_rlock(struct ifnet *ifp);	/* if_addrhead */
288void	if_addr_runlock(struct ifnet *ifp);	/* if_addrhead */
289void	if_maddr_rlock(struct ifnet *ifp);	/* if_multiaddrs */
290void	if_maddr_runlock(struct ifnet *ifp);	/* if_multiaddrs */
291
292/*
293 * Output queues (ifp->if_snd) and slow device input queues (*ifp->if_slowq)
294 * are queues of messages stored on ifqueue structures
295 * (defined above).  Entries are added to and deleted from these structures
296 * by these macros.
297 */
298#define IF_LOCK(ifq)		mtx_lock(&(ifq)->ifq_mtx)
299#define IF_UNLOCK(ifq)		mtx_unlock(&(ifq)->ifq_mtx)
300#define	IF_LOCK_ASSERT(ifq)	mtx_assert(&(ifq)->ifq_mtx, MA_OWNED)
301#define	_IF_QFULL(ifq)		((ifq)->ifq_len >= (ifq)->ifq_maxlen)
302#define	_IF_DROP(ifq)		((ifq)->ifq_drops++)
303#define	_IF_QLEN(ifq)		((ifq)->ifq_len)
304
305#define	_IF_ENQUEUE(ifq, m) do { 				\
306	(m)->m_nextpkt = NULL;					\
307	if ((ifq)->ifq_tail == NULL) 				\
308		(ifq)->ifq_head = m; 				\
309	else 							\
310		(ifq)->ifq_tail->m_nextpkt = m; 		\
311	(ifq)->ifq_tail = m; 					\
312	(ifq)->ifq_len++; 					\
313} while (0)
314
315#define IF_ENQUEUE(ifq, m) do {					\
316	IF_LOCK(ifq); 						\
317	_IF_ENQUEUE(ifq, m); 					\
318	IF_UNLOCK(ifq); 					\
319} while (0)
320
321#define	_IF_PREPEND(ifq, m) do {				\
322	(m)->m_nextpkt = (ifq)->ifq_head; 			\
323	if ((ifq)->ifq_tail == NULL) 				\
324		(ifq)->ifq_tail = (m); 				\
325	(ifq)->ifq_head = (m); 					\
326	(ifq)->ifq_len++; 					\
327} while (0)
328
329#define IF_PREPEND(ifq, m) do {		 			\
330	IF_LOCK(ifq); 						\
331	_IF_PREPEND(ifq, m); 					\
332	IF_UNLOCK(ifq); 					\
333} while (0)
334
335#define	_IF_DEQUEUE(ifq, m) do { 				\
336	(m) = (ifq)->ifq_head; 					\
337	if (m) { 						\
338		if (((ifq)->ifq_head = (m)->m_nextpkt) == NULL)	\
339			(ifq)->ifq_tail = NULL; 		\
340		(m)->m_nextpkt = NULL; 				\
341		(ifq)->ifq_len--; 				\
342	} 							\
343} while (0)
344
345#define IF_DEQUEUE(ifq, m) do { 				\
346	IF_LOCK(ifq); 						\
347	_IF_DEQUEUE(ifq, m); 					\
348	IF_UNLOCK(ifq); 					\
349} while (0)
350
351#define	_IF_DEQUEUE_ALL(ifq, m) do {				\
352	(m) = (ifq)->ifq_head;					\
353	(ifq)->ifq_head = (ifq)->ifq_tail = NULL;		\
354	(ifq)->ifq_len = 0;					\
355} while (0)
356
357#define	IF_DEQUEUE_ALL(ifq, m) do {				\
358	IF_LOCK(ifq); 						\
359	_IF_DEQUEUE_ALL(ifq, m);				\
360	IF_UNLOCK(ifq); 					\
361} while (0)
362
363#define	_IF_POLL(ifq, m)	((m) = (ifq)->ifq_head)
364#define	IF_POLL(ifq, m)		_IF_POLL(ifq, m)
365
366#define _IF_DRAIN(ifq) do { 					\
367	struct mbuf *m; 					\
368	for (;;) { 						\
369		_IF_DEQUEUE(ifq, m); 				\
370		if (m == NULL) 					\
371			break; 					\
372		m_freem(m); 					\
373	} 							\
374} while (0)
375
376#define IF_DRAIN(ifq) do {					\
377	IF_LOCK(ifq);						\
378	_IF_DRAIN(ifq);						\
379	IF_UNLOCK(ifq);						\
380} while(0)
381
382#ifdef _KERNEL
383/* interface link layer address change event */
384typedef void (*iflladdr_event_handler_t)(void *, struct ifnet *);
385EVENTHANDLER_DECLARE(iflladdr_event, iflladdr_event_handler_t);
386/* interface address change event */
387typedef void (*ifaddr_event_handler_t)(void *, struct ifnet *);
388EVENTHANDLER_DECLARE(ifaddr_event, ifaddr_event_handler_t);
389/* new interface arrival event */
390typedef void (*ifnet_arrival_event_handler_t)(void *, struct ifnet *);
391EVENTHANDLER_DECLARE(ifnet_arrival_event, ifnet_arrival_event_handler_t);
392/* interface departure event */
393typedef void (*ifnet_departure_event_handler_t)(void *, struct ifnet *);
394EVENTHANDLER_DECLARE(ifnet_departure_event, ifnet_departure_event_handler_t);
395/* Interface link state change event */
396typedef void (*ifnet_link_event_handler_t)(void *, struct ifnet *, int);
397EVENTHANDLER_DECLARE(ifnet_link_event, ifnet_link_event_handler_t);
398
399/*
400 * interface groups
401 */
402struct ifg_group {
403	char				 ifg_group[IFNAMSIZ];
404	u_int				 ifg_refcnt;
405	void				*ifg_pf_kif;
406	TAILQ_HEAD(, ifg_member)	 ifg_members;
407	TAILQ_ENTRY(ifg_group)		 ifg_next;
408};
409
410struct ifg_member {
411	TAILQ_ENTRY(ifg_member)	 ifgm_next;
412	struct ifnet		*ifgm_ifp;
413};
414
415struct ifg_list {
416	struct ifg_group	*ifgl_group;
417	TAILQ_ENTRY(ifg_list)	 ifgl_next;
418};
419
420/* group attach event */
421typedef void (*group_attach_event_handler_t)(void *, struct ifg_group *);
422EVENTHANDLER_DECLARE(group_attach_event, group_attach_event_handler_t);
423/* group detach event */
424typedef void (*group_detach_event_handler_t)(void *, struct ifg_group *);
425EVENTHANDLER_DECLARE(group_detach_event, group_detach_event_handler_t);
426/* group change event */
427typedef void (*group_change_event_handler_t)(void *, const char *);
428EVENTHANDLER_DECLARE(group_change_event, group_change_event_handler_t);
429
430#define	IF_AFDATA_LOCK_INIT(ifp)	\
431	rw_init(&(ifp)->if_afdata_lock, "if_afdata")
432
433#define	IF_AFDATA_WLOCK(ifp)	rw_wlock(&(ifp)->if_afdata_lock)
434#define	IF_AFDATA_RLOCK(ifp)	rw_rlock(&(ifp)->if_afdata_lock)
435#define	IF_AFDATA_WUNLOCK(ifp)	rw_wunlock(&(ifp)->if_afdata_lock)
436#define	IF_AFDATA_RUNLOCK(ifp)	rw_runlock(&(ifp)->if_afdata_lock)
437#define	IF_AFDATA_LOCK(ifp)	IF_AFDATA_WLOCK(ifp)
438#define	IF_AFDATA_UNLOCK(ifp)	IF_AFDATA_WUNLOCK(ifp)
439#define	IF_AFDATA_TRYLOCK(ifp)	rw_try_wlock(&(ifp)->if_afdata_lock)
440#define	IF_AFDATA_DESTROY(ifp)	rw_destroy(&(ifp)->if_afdata_lock)
441
442#define	IF_AFDATA_LOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_LOCKED)
443#define	IF_AFDATA_RLOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_RLOCKED)
444#define	IF_AFDATA_WLOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_WLOCKED)
445#define	IF_AFDATA_UNLOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_UNLOCKED)
446
447int	if_handoff(struct ifqueue *ifq, struct mbuf *m, struct ifnet *ifp,
448	    int adjust);
449#define	IF_HANDOFF(ifq, m, ifp)			\
450	if_handoff((struct ifqueue *)ifq, m, ifp, 0)
451#define	IF_HANDOFF_ADJ(ifq, m, ifp, adj)	\
452	if_handoff((struct ifqueue *)ifq, m, ifp, adj)
453
454void	if_start(struct ifnet *);
455
456#define	IFQ_ENQUEUE(ifq, m, err)					\
457do {									\
458	IF_LOCK(ifq);							\
459	if (ALTQ_IS_ENABLED(ifq))					\
460		ALTQ_ENQUEUE(ifq, m, NULL, err);			\
461	else {								\
462		if (_IF_QFULL(ifq)) {					\
463			m_freem(m);					\
464			(err) = ENOBUFS;				\
465		} else {						\
466			_IF_ENQUEUE(ifq, m);				\
467			(err) = 0;					\
468		}							\
469	}								\
470	if (err)							\
471		(ifq)->ifq_drops++;					\
472	IF_UNLOCK(ifq);							\
473} while (0)
474
475#define	IFQ_DEQUEUE_NOLOCK(ifq, m)					\
476do {									\
477	if (TBR_IS_ENABLED(ifq))					\
478		(m) = tbr_dequeue_ptr(ifq, ALTDQ_REMOVE);		\
479	else if (ALTQ_IS_ENABLED(ifq))					\
480		ALTQ_DEQUEUE(ifq, m);					\
481	else								\
482		_IF_DEQUEUE(ifq, m);					\
483} while (0)
484
485#define	IFQ_DEQUEUE(ifq, m)						\
486do {									\
487	IF_LOCK(ifq);							\
488	IFQ_DEQUEUE_NOLOCK(ifq, m);					\
489	IF_UNLOCK(ifq);							\
490} while (0)
491
492#define	IFQ_POLL_NOLOCK(ifq, m)						\
493do {									\
494	if (TBR_IS_ENABLED(ifq))					\
495		(m) = tbr_dequeue_ptr(ifq, ALTDQ_POLL);			\
496	else if (ALTQ_IS_ENABLED(ifq))					\
497		ALTQ_POLL(ifq, m);					\
498	else								\
499		_IF_POLL(ifq, m);					\
500} while (0)
501
502#define	IFQ_POLL(ifq, m)						\
503do {									\
504	IF_LOCK(ifq);							\
505	IFQ_POLL_NOLOCK(ifq, m);					\
506	IF_UNLOCK(ifq);							\
507} while (0)
508
509#define	IFQ_PURGE_NOLOCK(ifq)						\
510do {									\
511	if (ALTQ_IS_ENABLED(ifq)) {					\
512		ALTQ_PURGE(ifq);					\
513	} else								\
514		_IF_DRAIN(ifq);						\
515} while (0)
516
517#define	IFQ_PURGE(ifq)							\
518do {									\
519	IF_LOCK(ifq);							\
520	IFQ_PURGE_NOLOCK(ifq);						\
521	IF_UNLOCK(ifq);							\
522} while (0)
523
524#define	IFQ_SET_READY(ifq)						\
525	do { ((ifq)->altq_flags |= ALTQF_READY); } while (0)
526
527#define	IFQ_LOCK(ifq)			IF_LOCK(ifq)
528#define	IFQ_UNLOCK(ifq)			IF_UNLOCK(ifq)
529#define	IFQ_LOCK_ASSERT(ifq)		IF_LOCK_ASSERT(ifq)
530#define	IFQ_IS_EMPTY(ifq)		((ifq)->ifq_len == 0)
531#define	IFQ_INC_LEN(ifq)		((ifq)->ifq_len++)
532#define	IFQ_DEC_LEN(ifq)		(--(ifq)->ifq_len)
533#define	IFQ_INC_DROPS(ifq)		((ifq)->ifq_drops++)
534#define	IFQ_SET_MAXLEN(ifq, len)	((ifq)->ifq_maxlen = (len))
535
536/*
537 * The IFF_DRV_OACTIVE test should really occur in the device driver, not in
538 * the handoff logic, as that flag is locked by the device driver.
539 */
540#define	IFQ_HANDOFF_ADJ(ifp, m, adj, err)				\
541do {									\
542	int len;							\
543	short mflags;							\
544									\
545	len = (m)->m_pkthdr.len;					\
546	mflags = (m)->m_flags;						\
547	IFQ_ENQUEUE(&(ifp)->if_snd, m, err);				\
548	if ((err) == 0) {						\
549		(ifp)->if_obytes += len + (adj);			\
550		if (mflags & M_MCAST)					\
551			(ifp)->if_omcasts++;				\
552		if (((ifp)->if_drv_flags & IFF_DRV_OACTIVE) == 0)	\
553			if_start(ifp);					\
554	}								\
555} while (0)
556
557#define	IFQ_HANDOFF(ifp, m, err)					\
558	IFQ_HANDOFF_ADJ(ifp, m, 0, err)
559
560#define	IFQ_DRV_DEQUEUE(ifq, m)						\
561do {									\
562	(m) = (ifq)->ifq_drv_head;					\
563	if (m) {							\
564		if (((ifq)->ifq_drv_head = (m)->m_nextpkt) == NULL)	\
565			(ifq)->ifq_drv_tail = NULL;			\
566		(m)->m_nextpkt = NULL;					\
567		(ifq)->ifq_drv_len--;					\
568	} else {							\
569		IFQ_LOCK(ifq);						\
570		IFQ_DEQUEUE_NOLOCK(ifq, m);				\
571		while ((ifq)->ifq_drv_len < (ifq)->ifq_drv_maxlen) {	\
572			struct mbuf *m0;				\
573			IFQ_DEQUEUE_NOLOCK(ifq, m0);			\
574			if (m0 == NULL)					\
575				break;					\
576			m0->m_nextpkt = NULL;				\
577			if ((ifq)->ifq_drv_tail == NULL)		\
578				(ifq)->ifq_drv_head = m0;		\
579			else						\
580				(ifq)->ifq_drv_tail->m_nextpkt = m0;	\
581			(ifq)->ifq_drv_tail = m0;			\
582			(ifq)->ifq_drv_len++;				\
583		}							\
584		IFQ_UNLOCK(ifq);					\
585	}								\
586} while (0)
587
588#define	IFQ_DRV_PREPEND(ifq, m)						\
589do {									\
590	(m)->m_nextpkt = (ifq)->ifq_drv_head;				\
591	if ((ifq)->ifq_drv_tail == NULL)				\
592		(ifq)->ifq_drv_tail = (m);				\
593	(ifq)->ifq_drv_head = (m);					\
594	(ifq)->ifq_drv_len++;						\
595} while (0)
596
597#define	IFQ_DRV_IS_EMPTY(ifq)						\
598	(((ifq)->ifq_drv_len == 0) && ((ifq)->ifq_len == 0))
599
600#define	IFQ_DRV_PURGE(ifq)						\
601do {									\
602	struct mbuf *m, *n = (ifq)->ifq_drv_head;			\
603	while((m = n) != NULL) {					\
604		n = m->m_nextpkt;					\
605		m_freem(m);						\
606	}								\
607	(ifq)->ifq_drv_head = (ifq)->ifq_drv_tail = NULL;		\
608	(ifq)->ifq_drv_len = 0;						\
609	IFQ_PURGE(ifq);							\
610} while (0)
611
612#ifdef _KERNEL
613static __inline void
614if_initbaudrate(struct ifnet *ifp, uintmax_t baud)
615{
616
617	ifp->if_baudrate_pf = 0;
618	while (baud > (u_long)(~0UL)) {
619		baud /= 10;
620		ifp->if_baudrate_pf++;
621	}
622	ifp->if_baudrate = baud;
623}
624
625static __inline int
626drbr_enqueue(struct ifnet *ifp, struct buf_ring *br, struct mbuf *m)
627{
628	int error = 0;
629
630#ifdef ALTQ
631	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
632		IFQ_ENQUEUE(&ifp->if_snd, m, error);
633		return (error);
634	}
635#endif
636	error = buf_ring_enqueue(br, m);
637	if (error)
638		m_freem(m);
639
640	return (error);
641}
642
643static __inline void
644drbr_putback(struct ifnet *ifp, struct buf_ring *br, struct mbuf *new)
645{
646	/*
647	 * The top of the list needs to be swapped
648	 * for this one.
649	 */
650#ifdef ALTQ
651	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd)) {
652		/*
653		 * Peek in altq case dequeued it
654		 * so put it back.
655		 */
656		IFQ_DRV_PREPEND(&ifp->if_snd, new);
657		return;
658	}
659#endif
660	buf_ring_putback_sc(br, new);
661}
662
663static __inline struct mbuf *
664drbr_peek(struct ifnet *ifp, struct buf_ring *br)
665{
666#ifdef ALTQ
667	struct mbuf *m;
668	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd)) {
669		/*
670		 * Pull it off like a dequeue
671		 * since drbr_advance() does nothing
672		 * for altq and drbr_putback() will
673		 * use the old prepend function.
674		 */
675		IFQ_DEQUEUE(&ifp->if_snd, m);
676		return (m);
677	}
678#endif
679	return(buf_ring_peek(br));
680}
681
682static __inline void
683drbr_flush(struct ifnet *ifp, struct buf_ring *br)
684{
685	struct mbuf *m;
686
687#ifdef ALTQ
688	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd))
689		IFQ_PURGE(&ifp->if_snd);
690#endif
691	while ((m = buf_ring_dequeue_sc(br)) != NULL)
692		m_freem(m);
693}
694
695static __inline void
696drbr_free(struct buf_ring *br, struct malloc_type *type)
697{
698
699	drbr_flush(NULL, br);
700	buf_ring_free(br, type);
701}
702
703static __inline struct mbuf *
704drbr_dequeue(struct ifnet *ifp, struct buf_ring *br)
705{
706#ifdef ALTQ
707	struct mbuf *m;
708
709	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd)) {
710		IFQ_DEQUEUE(&ifp->if_snd, m);
711		return (m);
712	}
713#endif
714	return (buf_ring_dequeue_sc(br));
715}
716
717static __inline void
718drbr_advance(struct ifnet *ifp, struct buf_ring *br)
719{
720#ifdef ALTQ
721	/* Nothing to do here since peek dequeues in altq case */
722	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd))
723		return;
724#endif
725	return (buf_ring_advance_sc(br));
726}
727
728
729static __inline struct mbuf *
730drbr_dequeue_cond(struct ifnet *ifp, struct buf_ring *br,
731    int (*func) (struct mbuf *, void *), void *arg)
732{
733	struct mbuf *m;
734#ifdef ALTQ
735	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
736		IFQ_LOCK(&ifp->if_snd);
737		IFQ_POLL_NOLOCK(&ifp->if_snd, m);
738		if (m != NULL && func(m, arg) == 0) {
739			IFQ_UNLOCK(&ifp->if_snd);
740			return (NULL);
741		}
742		IFQ_DEQUEUE_NOLOCK(&ifp->if_snd, m);
743		IFQ_UNLOCK(&ifp->if_snd);
744		return (m);
745	}
746#endif
747	m = buf_ring_peek(br);
748	if (m == NULL || func(m, arg) == 0)
749		return (NULL);
750
751	return (buf_ring_dequeue_sc(br));
752}
753
754static __inline int
755drbr_empty(struct ifnet *ifp, struct buf_ring *br)
756{
757#ifdef ALTQ
758	if (ALTQ_IS_ENABLED(&ifp->if_snd))
759		return (IFQ_IS_EMPTY(&ifp->if_snd));
760#endif
761	return (buf_ring_empty(br));
762}
763
764static __inline int
765drbr_needs_enqueue(struct ifnet *ifp, struct buf_ring *br)
766{
767#ifdef ALTQ
768	if (ALTQ_IS_ENABLED(&ifp->if_snd))
769		return (1);
770#endif
771	return (!buf_ring_empty(br));
772}
773
774static __inline int
775drbr_inuse(struct ifnet *ifp, struct buf_ring *br)
776{
777#ifdef ALTQ
778	if (ALTQ_IS_ENABLED(&ifp->if_snd))
779		return (ifp->if_snd.ifq_len);
780#endif
781	return (buf_ring_count(br));
782}
783#endif
784/*
785 * 72 was chosen below because it is the size of a TCP/IP
786 * header (40) + the minimum mss (32).
787 */
788#define	IF_MINMTU	72
789#define	IF_MAXMTU	65535
790
791#define	TOEDEV(ifp)	((ifp)->if_llsoftc)
792
793#endif /* _KERNEL */
794
795/*
796 * The ifaddr structure contains information about one address
797 * of an interface.  They are maintained by the different address families,
798 * are allocated and attached when an address is set, and are linked
799 * together so all addresses for an interface can be located.
800 *
801 * NOTE: a 'struct ifaddr' is always at the beginning of a larger
802 * chunk of malloc'ed memory, where we store the three addresses
803 * (ifa_addr, ifa_dstaddr and ifa_netmask) referenced here.
804 */
805struct ifaddr {
806	struct	sockaddr *ifa_addr;	/* address of interface */
807	struct	sockaddr *ifa_dstaddr;	/* other end of p-to-p link */
808#define	ifa_broadaddr	ifa_dstaddr	/* broadcast address interface */
809	struct	sockaddr *ifa_netmask;	/* used to determine subnet */
810	struct	if_data if_data;	/* not all members are meaningful */
811	struct	ifnet *ifa_ifp;		/* back-pointer to interface */
812	struct	carp_softc *ifa_carp;	/* pointer to CARP data */
813	TAILQ_ENTRY(ifaddr) ifa_link;	/* queue macro glue */
814	void	(*ifa_rtrequest)	/* check or clean routes (+ or -)'d */
815		(int, struct rtentry *, struct rt_addrinfo *);
816	u_short	ifa_flags;		/* mostly rt_flags for cloning */
817	u_int	ifa_refcnt;		/* references to this structure */
818	int	ifa_metric;		/* cost of going out this interface */
819	int (*ifa_claim_addr)		/* check if an addr goes to this if */
820		(struct ifaddr *, struct sockaddr *);
821	struct mtx ifa_mtx;
822};
823#define	IFA_ROUTE	RTF_UP		/* route installed */
824#define IFA_RTSELF	RTF_HOST	/* loopback route to self installed */
825
826/* for compatibility with other BSDs */
827#define	ifa_list	ifa_link
828
829#ifdef _KERNEL
830#define	IFA_LOCK(ifa)		mtx_lock(&(ifa)->ifa_mtx)
831#define	IFA_UNLOCK(ifa)		mtx_unlock(&(ifa)->ifa_mtx)
832
833void	ifa_free(struct ifaddr *ifa);
834void	ifa_init(struct ifaddr *ifa);
835void	ifa_ref(struct ifaddr *ifa);
836#endif
837
838/*
839 * Multicast address structure.  This is analogous to the ifaddr
840 * structure except that it keeps track of multicast addresses.
841 */
842struct ifmultiaddr {
843	TAILQ_ENTRY(ifmultiaddr) ifma_link; /* queue macro glue */
844	struct	sockaddr *ifma_addr; 	/* address this membership is for */
845	struct	sockaddr *ifma_lladdr;	/* link-layer translation, if any */
846	struct	ifnet *ifma_ifp;	/* back-pointer to interface */
847	u_int	ifma_refcount;		/* reference count */
848	void	*ifma_protospec;	/* protocol-specific state, if any */
849	struct	ifmultiaddr *ifma_llifma; /* pointer to ifma for ifma_lladdr */
850};
851
852#ifdef _KERNEL
853
854extern	struct rwlock ifnet_rwlock;
855extern	struct sx ifnet_sxlock;
856
857#define	IFNET_LOCK_INIT() do {						\
858	rw_init_flags(&ifnet_rwlock, "ifnet_rw",  RW_RECURSE);		\
859	sx_init_flags(&ifnet_sxlock, "ifnet_sx",  SX_RECURSE);		\
860} while(0)
861
862#define	IFNET_WLOCK() do {						\
863	sx_xlock(&ifnet_sxlock);					\
864	rw_wlock(&ifnet_rwlock);					\
865} while (0)
866
867#define	IFNET_WUNLOCK() do {						\
868	rw_wunlock(&ifnet_rwlock);					\
869	sx_xunlock(&ifnet_sxlock);					\
870} while (0)
871
872/*
873 * To assert the ifnet lock, you must know not only whether it's for read or
874 * write, but also whether it was acquired with sleep support or not.
875 */
876#define	IFNET_RLOCK_ASSERT()		sx_assert(&ifnet_sxlock, SA_SLOCKED)
877#define	IFNET_RLOCK_NOSLEEP_ASSERT()	rw_assert(&ifnet_rwlock, RA_RLOCKED)
878#define	IFNET_WLOCK_ASSERT() do {					\
879	sx_assert(&ifnet_sxlock, SA_XLOCKED);				\
880	rw_assert(&ifnet_rwlock, RA_WLOCKED);				\
881} while (0)
882
883#define	IFNET_RLOCK()		sx_slock(&ifnet_sxlock)
884#define	IFNET_RLOCK_NOSLEEP()	rw_rlock(&ifnet_rwlock)
885#define	IFNET_RUNLOCK()		sx_sunlock(&ifnet_sxlock)
886#define	IFNET_RUNLOCK_NOSLEEP()	rw_runlock(&ifnet_rwlock)
887
888/*
889 * Look up an ifnet given its index; the _ref variant also acquires a
890 * reference that must be freed using if_rele().  It is almost always a bug
891 * to call ifnet_byindex() instead if ifnet_byindex_ref().
892 */
893struct ifnet	*ifnet_byindex(u_short idx);
894struct ifnet	*ifnet_byindex_locked(u_short idx);
895struct ifnet	*ifnet_byindex_ref(u_short idx);
896
897/*
898 * Given the index, ifaddr_byindex() returns the one and only
899 * link-level ifaddr for the interface. You are not supposed to use
900 * it to traverse the list of addresses associated to the interface.
901 */
902struct ifaddr	*ifaddr_byindex(u_short idx);
903
904VNET_DECLARE(struct ifnethead, ifnet);
905VNET_DECLARE(struct ifgrouphead, ifg_head);
906VNET_DECLARE(int, if_index);
907VNET_DECLARE(struct ifnet *, loif);	/* first loopback interface */
908VNET_DECLARE(int, useloopback);
909
910#define	V_ifnet		VNET(ifnet)
911#define	V_ifg_head	VNET(ifg_head)
912#define	V_if_index	VNET(if_index)
913#define	V_loif		VNET(loif)
914#define	V_useloopback	VNET(useloopback)
915
916extern	int ifqmaxlen;
917
918int	if_addgroup(struct ifnet *, const char *);
919int	if_delgroup(struct ifnet *, const char *);
920int	if_addmulti(struct ifnet *, struct sockaddr *, struct ifmultiaddr **);
921int	if_allmulti(struct ifnet *, int);
922struct	ifnet* if_alloc(u_char);
923void	if_attach(struct ifnet *);
924void	if_dead(struct ifnet *);
925int	if_delmulti(struct ifnet *, struct sockaddr *);
926void	if_delmulti_ifma(struct ifmultiaddr *);
927void	if_detach(struct ifnet *);
928void	if_vmove(struct ifnet *, struct vnet *);
929void	if_purgeaddrs(struct ifnet *);
930void	if_delallmulti(struct ifnet *);
931void	if_down(struct ifnet *);
932struct ifmultiaddr *
933	if_findmulti(struct ifnet *, struct sockaddr *);
934void	if_free(struct ifnet *);
935void	if_initname(struct ifnet *, const char *, int);
936void	if_link_state_change(struct ifnet *, int);
937int	if_printf(struct ifnet *, const char *, ...) __printflike(2, 3);
938void	if_qflush(struct ifnet *);
939void	if_ref(struct ifnet *);
940void	if_rele(struct ifnet *);
941int	if_setlladdr(struct ifnet *, const u_char *, int);
942void	if_up(struct ifnet *);
943int	ifioctl(struct socket *, u_long, caddr_t, struct thread *);
944int	ifpromisc(struct ifnet *, int);
945struct	ifnet *ifunit(const char *);
946struct	ifnet *ifunit_ref(const char *);
947
948void	ifq_init(struct ifaltq *, struct ifnet *ifp);
949void	ifq_delete(struct ifaltq *);
950
951int	ifa_add_loopback_route(struct ifaddr *, struct sockaddr *);
952int	ifa_del_loopback_route(struct ifaddr *, struct sockaddr *);
953
954struct	ifaddr *ifa_ifwithaddr(struct sockaddr *);
955int		ifa_ifwithaddr_check(struct sockaddr *);
956struct	ifaddr *ifa_ifwithbroadaddr(struct sockaddr *);
957struct	ifaddr *ifa_ifwithdstaddr(struct sockaddr *);
958struct	ifaddr *ifa_ifwithdstaddr_fib(struct sockaddr *, int);
959struct	ifaddr *ifa_ifwithnet(struct sockaddr *, int);
960struct	ifaddr *ifa_ifwithnet_fib(struct sockaddr *, int, int);
961struct	ifaddr *ifa_ifwithroute(int, struct sockaddr *, struct sockaddr *);
962struct	ifaddr *ifa_ifwithroute_fib(int, struct sockaddr *, struct sockaddr *, u_int);
963struct	ifaddr *ifaof_ifpforaddr(struct sockaddr *, struct ifnet *);
964int	ifa_preferred(struct ifaddr *, struct ifaddr *);
965
966int	if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen);
967
968typedef	void *if_com_alloc_t(u_char type, struct ifnet *ifp);
969typedef	void if_com_free_t(void *com, u_char type);
970void	if_register_com_alloc(u_char type, if_com_alloc_t *a, if_com_free_t *f);
971void	if_deregister_com_alloc(u_char type);
972
973#define IF_LLADDR(ifp)							\
974    LLADDR((struct sockaddr_dl *)((ifp)->if_addr->ifa_addr))
975
976#ifdef DEVICE_POLLING
977enum poll_cmd {	POLL_ONLY, POLL_AND_CHECK_STATUS };
978
979typedef	int poll_handler_t(struct ifnet *ifp, enum poll_cmd cmd, int count);
980int    ether_poll_register(poll_handler_t *h, struct ifnet *ifp);
981int    ether_poll_deregister(struct ifnet *ifp);
982#endif /* DEVICE_POLLING */
983
984/* TSO */
985void if_hw_tsomax_common(struct ifnet *, struct ifnet_hw_tsomax *);
986int if_hw_tsomax_update(struct ifnet *, struct ifnet_hw_tsomax *);
987
988#endif /* _KERNEL */
989
990#endif /* !_NET_IF_VAR_H_ */
991