ifq.h revision 257271
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 257271 2013-10-28 20:32:05Z 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
61struct	rtentry;		/* ifa_rtrequest */
62struct	rt_addrinfo;		/* ifa_rtrequest */
63struct	socket;
64struct	carp_if;
65struct	carp_softc;
66struct  ifvlantrunk;
67struct	route;			/* if_output */
68struct	vnet;
69
70#ifdef _KERNEL
71#include <sys/mbuf.h>		/* ifqueue only? */
72#include <sys/buf_ring.h>
73#include <net/vnet.h>
74#endif /* _KERNEL */
75#include <sys/counter.h>
76#include <sys/lock.h>		/* XXX */
77#include <sys/mutex.h>		/* struct ifqueue */
78#include <sys/rwlock.h>		/* XXX */
79#include <sys/sx.h>		/* XXX */
80#include <sys/_task.h>		/* if_link_task */
81
82#define	IF_DUNIT_NONE	-1
83
84#include <altq/if_altq.h>
85
86TAILQ_HEAD(ifnethead, ifnet);	/* we use TAILQs so that the order of */
87TAILQ_HEAD(ifaddrhead, ifaddr);	/* instantiation is preserved in the list */
88TAILQ_HEAD(ifmultihead, ifmultiaddr);
89TAILQ_HEAD(ifgrouphead, ifg_group);
90
91#ifdef _KERNEL
92VNET_DECLARE(struct pfil_head, link_pfil_hook);	/* packet filter hooks */
93#define	V_link_pfil_hook	VNET(link_pfil_hook)
94#endif /* _KERNEL */
95
96/*
97 * Structure defining a queue for a network interface.
98 */
99struct	ifqueue {
100	struct	mbuf *ifq_head;
101	struct	mbuf *ifq_tail;
102	int	ifq_len;
103	int	ifq_maxlen;
104	int	ifq_drops;
105	struct	mtx ifq_mtx;
106};
107
108/*
109 * Structure defining a network interface.
110 *
111 * (Would like to call this struct ``if'', but C isn't PL/1.)
112 */
113
114struct ifnet {
115	void	*if_softc;		/* pointer to driver state */
116	void	*if_l2com;		/* pointer to protocol bits */
117	struct vnet *if_vnet;		/* pointer to network stack instance */
118	TAILQ_ENTRY(ifnet) if_link; 	/* all struct ifnets are chained */
119	char	if_xname[IFNAMSIZ];	/* external name (name + unit) */
120	const char *if_dname;		/* driver name */
121	int	if_dunit;		/* unit or IF_DUNIT_NONE */
122	u_int	if_refcount;		/* reference count */
123	struct	ifaddrhead if_addrhead;	/* linked list of addresses per if */
124		/*
125		 * if_addrhead is the list of all addresses associated to
126		 * an interface.
127		 * Some code in the kernel assumes that first element
128		 * of the list has type AF_LINK, and contains sockaddr_dl
129		 * addresses which store the link-level address and the name
130		 * of the interface.
131		 * However, access to the AF_LINK address through this
132		 * field is deprecated. Use if_addr or ifaddr_byindex() instead.
133		 */
134	int	if_pcount;		/* number of promiscuous listeners */
135	struct	carp_if *if_carp;	/* carp interface structure */
136	struct	bpf_if *if_bpf;		/* packet filter structure */
137	u_short	if_index;		/* numeric abbreviation for this if  */
138	short	if_index_reserved;	/* spare space to grow if_index */
139	struct  ifvlantrunk *if_vlantrunk; /* pointer to 802.1q data */
140	int	if_flags;		/* up/down, broadcast, etc. */
141	int	if_capabilities;	/* interface features & capabilities */
142	int	if_capenable;		/* enabled features & capabilities */
143	void	*if_linkmib;		/* link-type-specific MIB data */
144	size_t	if_linkmiblen;		/* length of above data */
145	struct	if_data if_data;
146	struct	ifmultihead if_multiaddrs; /* multicast addresses configured */
147	int	if_amcount;		/* number of all-multicast requests */
148/* procedure handles */
149	int	(*if_output)		/* output routine (enqueue) */
150		(struct ifnet *, struct mbuf *, const struct sockaddr *,
151		     struct route *);
152	void	(*if_input)		/* input routine (from h/w driver) */
153		(struct ifnet *, struct mbuf *);
154	void	(*if_start)		/* initiate output routine */
155		(struct ifnet *);
156	int	(*if_ioctl)		/* ioctl routine */
157		(struct ifnet *, u_long, caddr_t);
158	void	(*if_init)		/* Init routine */
159		(void *);
160	int	(*if_resolvemulti)	/* validate/resolve multicast */
161		(struct ifnet *, struct sockaddr **, struct sockaddr *);
162	void	(*if_qflush)		/* flush any queues */
163		(struct ifnet *);
164	int	(*if_transmit)		/* initiate output routine */
165		(struct ifnet *, struct mbuf *);
166	void	(*if_reassign)		/* reassign to vnet routine */
167		(struct ifnet *, struct vnet *, char *);
168	struct	vnet *if_home_vnet;	/* where this ifnet originates from */
169	struct	ifaddr	*if_addr;	/* pointer to link-level address */
170	void	*if_llsoftc;		/* link layer softc */
171	int	if_drv_flags;		/* driver-managed status flags */
172	struct  ifaltq if_snd;		/* output queue (includes altq) */
173	const u_int8_t *if_broadcastaddr; /* linklevel broadcast bytestring */
174
175	void	*if_bridge;		/* bridge glue */
176
177	struct	label *if_label;	/* interface MAC label */
178
179	/* these are only used by IPv6 */
180	void	*if_unused[2];
181	void	*if_afdata[AF_MAX];
182	int	if_afdata_initialized;
183	struct	rwlock if_afdata_lock;
184	struct	task if_linktask;	/* task for link change events */
185	struct	rwlock if_addr_lock;	/* lock to protect address lists */
186
187	LIST_ENTRY(ifnet) if_clones;	/* interfaces of a cloner */
188	TAILQ_HEAD(, ifg_list) if_groups; /* linked list of groups per if */
189					/* protected by if_addr_lock */
190	void	*if_pf_kif;
191	void	*if_lagg;		/* lagg glue */
192	char	*if_description;	/* interface description */
193	u_int	if_fib;			/* interface FIB */
194	u_char	if_alloctype;		/* if_type at time of allocation */
195
196	u_int	if_hw_tsomax;		/* tso burst length limit, the minimum
197					 * is (IP_MAXPACKET / 8).
198					 * XXXAO: Have to find a better place
199					 * for it eventually. */
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	int	if_ispare[4];
208	void	*if_pspare[8];		/* 1 netmap, 7 TDB */
209};
210
211/*
212 * XXX These aliases are terribly dangerous because they could apply
213 * to anything.
214 */
215#define	if_mtu		if_data.ifi_mtu
216#define	if_type		if_data.ifi_type
217#define if_physical	if_data.ifi_physical
218#define	if_addrlen	if_data.ifi_addrlen
219#define	if_hdrlen	if_data.ifi_hdrlen
220#define	if_metric	if_data.ifi_metric
221#define	if_link_state	if_data.ifi_link_state
222#define	if_baudrate	if_data.ifi_baudrate
223#define	if_baudrate_pf	if_data.ifi_baudrate_pf
224#define	if_hwassist	if_data.ifi_hwassist
225#define	if_ipackets	if_data.ifi_ipackets
226#define	if_ierrors	if_data.ifi_ierrors
227#define	if_opackets	if_data.ifi_opackets
228#define	if_oerrors	if_data.ifi_oerrors
229#define	if_collisions	if_data.ifi_collisions
230#define	if_ibytes	if_data.ifi_ibytes
231#define	if_obytes	if_data.ifi_obytes
232#define	if_imcasts	if_data.ifi_imcasts
233#define	if_omcasts	if_data.ifi_omcasts
234#define	if_iqdrops	if_data.ifi_iqdrops
235#define	if_noproto	if_data.ifi_noproto
236#define	if_lastchange	if_data.ifi_lastchange
237
238/* for compatibility with other BSDs */
239#define	if_addrlist	if_addrhead
240#define	if_list		if_link
241#define	if_name(ifp)	((ifp)->if_xname)
242
243/*
244 * Locks for address lists on the network interface.
245 */
246#define	IF_ADDR_LOCK_INIT(if)	rw_init(&(if)->if_addr_lock, "if_addr_lock")
247#define	IF_ADDR_LOCK_DESTROY(if)	rw_destroy(&(if)->if_addr_lock)
248#define	IF_ADDR_WLOCK(if)	rw_wlock(&(if)->if_addr_lock)
249#define	IF_ADDR_WUNLOCK(if)	rw_wunlock(&(if)->if_addr_lock)
250#define	IF_ADDR_RLOCK(if)	rw_rlock(&(if)->if_addr_lock)
251#define	IF_ADDR_RUNLOCK(if)	rw_runlock(&(if)->if_addr_lock)
252#define	IF_ADDR_LOCK_ASSERT(if)	rw_assert(&(if)->if_addr_lock, RA_LOCKED)
253#define	IF_ADDR_WLOCK_ASSERT(if) rw_assert(&(if)->if_addr_lock, RA_WLOCKED)
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.
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_DEQUEUE_ALL(ifq, m) do {				\
325	(m) = (ifq)->ifq_head;					\
326	(ifq)->ifq_head = (ifq)->ifq_tail = NULL;		\
327	(ifq)->ifq_len = 0;					\
328} while (0)
329
330#define	IF_DEQUEUE_ALL(ifq, m) do {				\
331	IF_LOCK(ifq); 						\
332	_IF_DEQUEUE_ALL(ifq, m);				\
333	IF_UNLOCK(ifq); 					\
334} while (0)
335
336#define	_IF_POLL(ifq, m)	((m) = (ifq)->ifq_head)
337#define	IF_POLL(ifq, m)		_IF_POLL(ifq, m)
338
339#define _IF_DRAIN(ifq) do { 					\
340	struct mbuf *m; 					\
341	for (;;) { 						\
342		_IF_DEQUEUE(ifq, m); 				\
343		if (m == NULL) 					\
344			break; 					\
345		m_freem(m); 					\
346	} 							\
347} while (0)
348
349#define IF_DRAIN(ifq) do {					\
350	IF_LOCK(ifq);						\
351	_IF_DRAIN(ifq);						\
352	IF_UNLOCK(ifq);						\
353} while(0)
354
355#ifdef _KERNEL
356#ifdef _SYS_EVENTHANDLER_H_
357/* interface link layer address change event */
358typedef void (*iflladdr_event_handler_t)(void *, struct ifnet *);
359EVENTHANDLER_DECLARE(iflladdr_event, iflladdr_event_handler_t);
360/* interface address change event */
361typedef void (*ifaddr_event_handler_t)(void *, struct ifnet *);
362EVENTHANDLER_DECLARE(ifaddr_event, ifaddr_event_handler_t);
363/* new interface arrival event */
364typedef void (*ifnet_arrival_event_handler_t)(void *, struct ifnet *);
365EVENTHANDLER_DECLARE(ifnet_arrival_event, ifnet_arrival_event_handler_t);
366/* interface departure event */
367typedef void (*ifnet_departure_event_handler_t)(void *, struct ifnet *);
368EVENTHANDLER_DECLARE(ifnet_departure_event, ifnet_departure_event_handler_t);
369/* Interface link state change event */
370typedef void (*ifnet_link_event_handler_t)(void *, struct ifnet *, int);
371EVENTHANDLER_DECLARE(ifnet_link_event, ifnet_link_event_handler_t);
372#endif /* _SYS_EVENTHANDLER_H_ */
373
374/*
375 * interface groups
376 */
377struct ifg_group {
378	char				 ifg_group[IFNAMSIZ];
379	u_int				 ifg_refcnt;
380	void				*ifg_pf_kif;
381	TAILQ_HEAD(, ifg_member)	 ifg_members;
382	TAILQ_ENTRY(ifg_group)		 ifg_next;
383};
384
385struct ifg_member {
386	TAILQ_ENTRY(ifg_member)	 ifgm_next;
387	struct ifnet		*ifgm_ifp;
388};
389
390struct ifg_list {
391	struct ifg_group	*ifgl_group;
392	TAILQ_ENTRY(ifg_list)	 ifgl_next;
393};
394
395#ifdef _SYS_EVENTHANDLER_H_
396/* group attach event */
397typedef void (*group_attach_event_handler_t)(void *, struct ifg_group *);
398EVENTHANDLER_DECLARE(group_attach_event, group_attach_event_handler_t);
399/* group detach event */
400typedef void (*group_detach_event_handler_t)(void *, struct ifg_group *);
401EVENTHANDLER_DECLARE(group_detach_event, group_detach_event_handler_t);
402/* group change event */
403typedef void (*group_change_event_handler_t)(void *, const char *);
404EVENTHANDLER_DECLARE(group_change_event, group_change_event_handler_t);
405#endif /* _SYS_EVENTHANDLER_H_ */
406
407#define	IF_AFDATA_LOCK_INIT(ifp)	\
408	rw_init(&(ifp)->if_afdata_lock, "if_afdata")
409
410#define	IF_AFDATA_WLOCK(ifp)	rw_wlock(&(ifp)->if_afdata_lock)
411#define	IF_AFDATA_RLOCK(ifp)	rw_rlock(&(ifp)->if_afdata_lock)
412#define	IF_AFDATA_WUNLOCK(ifp)	rw_wunlock(&(ifp)->if_afdata_lock)
413#define	IF_AFDATA_RUNLOCK(ifp)	rw_runlock(&(ifp)->if_afdata_lock)
414#define	IF_AFDATA_LOCK(ifp)	IF_AFDATA_WLOCK(ifp)
415#define	IF_AFDATA_UNLOCK(ifp)	IF_AFDATA_WUNLOCK(ifp)
416#define	IF_AFDATA_TRYLOCK(ifp)	rw_try_wlock(&(ifp)->if_afdata_lock)
417#define	IF_AFDATA_DESTROY(ifp)	rw_destroy(&(ifp)->if_afdata_lock)
418
419#define	IF_AFDATA_LOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_LOCKED)
420#define	IF_AFDATA_RLOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_RLOCKED)
421#define	IF_AFDATA_WLOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_WLOCKED)
422#define	IF_AFDATA_UNLOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_UNLOCKED)
423
424int	if_handoff(struct ifqueue *ifq, struct mbuf *m, struct ifnet *ifp,
425	    int adjust);
426#define	IF_HANDOFF(ifq, m, ifp)			\
427	if_handoff((struct ifqueue *)ifq, m, ifp, 0)
428#define	IF_HANDOFF_ADJ(ifq, m, ifp, adj)	\
429	if_handoff((struct ifqueue *)ifq, m, ifp, adj)
430
431void	if_start(struct ifnet *);
432
433#define	IFQ_ENQUEUE(ifq, m, err)					\
434do {									\
435	IF_LOCK(ifq);							\
436	if (ALTQ_IS_ENABLED(ifq))					\
437		ALTQ_ENQUEUE(ifq, m, NULL, err);			\
438	else {								\
439		if (_IF_QFULL(ifq)) {					\
440			m_freem(m);					\
441			(err) = ENOBUFS;				\
442		} else {						\
443			_IF_ENQUEUE(ifq, m);				\
444			(err) = 0;					\
445		}							\
446	}								\
447	if (err)							\
448		(ifq)->ifq_drops++;					\
449	IF_UNLOCK(ifq);							\
450} while (0)
451
452#define	IFQ_DEQUEUE_NOLOCK(ifq, m)					\
453do {									\
454	if (TBR_IS_ENABLED(ifq))					\
455		(m) = tbr_dequeue_ptr(ifq, ALTDQ_REMOVE);		\
456	else if (ALTQ_IS_ENABLED(ifq))					\
457		ALTQ_DEQUEUE(ifq, m);					\
458	else								\
459		_IF_DEQUEUE(ifq, m);					\
460} while (0)
461
462#define	IFQ_DEQUEUE(ifq, m)						\
463do {									\
464	IF_LOCK(ifq);							\
465	IFQ_DEQUEUE_NOLOCK(ifq, m);					\
466	IF_UNLOCK(ifq);							\
467} while (0)
468
469#define	IFQ_POLL_NOLOCK(ifq, m)						\
470do {									\
471	if (TBR_IS_ENABLED(ifq))					\
472		(m) = tbr_dequeue_ptr(ifq, ALTDQ_POLL);			\
473	else if (ALTQ_IS_ENABLED(ifq))					\
474		ALTQ_POLL(ifq, m);					\
475	else								\
476		_IF_POLL(ifq, m);					\
477} while (0)
478
479#define	IFQ_POLL(ifq, m)						\
480do {									\
481	IF_LOCK(ifq);							\
482	IFQ_POLL_NOLOCK(ifq, m);					\
483	IF_UNLOCK(ifq);							\
484} while (0)
485
486#define	IFQ_PURGE_NOLOCK(ifq)						\
487do {									\
488	if (ALTQ_IS_ENABLED(ifq)) {					\
489		ALTQ_PURGE(ifq);					\
490	} else								\
491		_IF_DRAIN(ifq);						\
492} while (0)
493
494#define	IFQ_PURGE(ifq)							\
495do {									\
496	IF_LOCK(ifq);							\
497	IFQ_PURGE_NOLOCK(ifq);						\
498	IF_UNLOCK(ifq);							\
499} while (0)
500
501#define	IFQ_SET_READY(ifq)						\
502	do { ((ifq)->altq_flags |= ALTQF_READY); } while (0)
503
504#define	IFQ_LOCK(ifq)			IF_LOCK(ifq)
505#define	IFQ_UNLOCK(ifq)			IF_UNLOCK(ifq)
506#define	IFQ_LOCK_ASSERT(ifq)		IF_LOCK_ASSERT(ifq)
507#define	IFQ_IS_EMPTY(ifq)		((ifq)->ifq_len == 0)
508#define	IFQ_INC_LEN(ifq)		((ifq)->ifq_len++)
509#define	IFQ_DEC_LEN(ifq)		(--(ifq)->ifq_len)
510#define	IFQ_INC_DROPS(ifq)		((ifq)->ifq_drops++)
511#define	IFQ_SET_MAXLEN(ifq, len)	((ifq)->ifq_maxlen = (len))
512
513/*
514 * The IFF_DRV_OACTIVE test should really occur in the device driver, not in
515 * the handoff logic, as that flag is locked by the device driver.
516 */
517#define	IFQ_HANDOFF_ADJ(ifp, m, adj, err)				\
518do {									\
519	int len;							\
520	short mflags;							\
521									\
522	len = (m)->m_pkthdr.len;					\
523	mflags = (m)->m_flags;						\
524	IFQ_ENQUEUE(&(ifp)->if_snd, m, err);				\
525	if ((err) == 0) {						\
526		(ifp)->if_obytes += len + (adj);			\
527		if (mflags & M_MCAST)					\
528			(ifp)->if_omcasts++;				\
529		if (((ifp)->if_drv_flags & IFF_DRV_OACTIVE) == 0)	\
530			if_start(ifp);					\
531	}								\
532} while (0)
533
534#define	IFQ_HANDOFF(ifp, m, err)					\
535	IFQ_HANDOFF_ADJ(ifp, m, 0, err)
536
537#define	IFQ_DRV_DEQUEUE(ifq, m)						\
538do {									\
539	(m) = (ifq)->ifq_drv_head;					\
540	if (m) {							\
541		if (((ifq)->ifq_drv_head = (m)->m_nextpkt) == NULL)	\
542			(ifq)->ifq_drv_tail = NULL;			\
543		(m)->m_nextpkt = NULL;					\
544		(ifq)->ifq_drv_len--;					\
545	} else {							\
546		IFQ_LOCK(ifq);						\
547		IFQ_DEQUEUE_NOLOCK(ifq, m);				\
548		while ((ifq)->ifq_drv_len < (ifq)->ifq_drv_maxlen) {	\
549			struct mbuf *m0;				\
550			IFQ_DEQUEUE_NOLOCK(ifq, m0);			\
551			if (m0 == NULL)					\
552				break;					\
553			m0->m_nextpkt = NULL;				\
554			if ((ifq)->ifq_drv_tail == NULL)		\
555				(ifq)->ifq_drv_head = m0;		\
556			else						\
557				(ifq)->ifq_drv_tail->m_nextpkt = m0;	\
558			(ifq)->ifq_drv_tail = m0;			\
559			(ifq)->ifq_drv_len++;				\
560		}							\
561		IFQ_UNLOCK(ifq);					\
562	}								\
563} while (0)
564
565#define	IFQ_DRV_PREPEND(ifq, m)						\
566do {									\
567	(m)->m_nextpkt = (ifq)->ifq_drv_head;				\
568	if ((ifq)->ifq_drv_tail == NULL)				\
569		(ifq)->ifq_drv_tail = (m);				\
570	(ifq)->ifq_drv_head = (m);					\
571	(ifq)->ifq_drv_len++;						\
572} while (0)
573
574#define	IFQ_DRV_IS_EMPTY(ifq)						\
575	(((ifq)->ifq_drv_len == 0) && ((ifq)->ifq_len == 0))
576
577#define	IFQ_DRV_PURGE(ifq)						\
578do {									\
579	struct mbuf *m, *n = (ifq)->ifq_drv_head;			\
580	while((m = n) != NULL) {					\
581		n = m->m_nextpkt;					\
582		m_freem(m);						\
583	}								\
584	(ifq)->ifq_drv_head = (ifq)->ifq_drv_tail = NULL;		\
585	(ifq)->ifq_drv_len = 0;						\
586	IFQ_PURGE(ifq);							\
587} while (0)
588
589#ifdef _KERNEL
590static __inline void
591if_initbaudrate(struct ifnet *ifp, uintmax_t baud)
592{
593
594	ifp->if_baudrate_pf = 0;
595	while (baud > (u_long)(~0UL)) {
596		baud /= 10;
597		ifp->if_baudrate_pf++;
598	}
599	ifp->if_baudrate = baud;
600}
601
602static __inline int
603drbr_enqueue(struct ifnet *ifp, struct buf_ring *br, struct mbuf *m)
604{
605	int error = 0;
606
607#ifdef ALTQ
608	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
609		IFQ_ENQUEUE(&ifp->if_snd, m, error);
610		return (error);
611	}
612#endif
613	error = buf_ring_enqueue(br, m);
614	if (error)
615		m_freem(m);
616
617	return (error);
618}
619
620static __inline void
621drbr_putback(struct ifnet *ifp, struct buf_ring *br, struct mbuf *new)
622{
623	/*
624	 * The top of the list needs to be swapped
625	 * for this one.
626	 */
627#ifdef ALTQ
628	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd)) {
629		/*
630		 * Peek in altq case dequeued it
631		 * so put it back.
632		 */
633		IFQ_DRV_PREPEND(&ifp->if_snd, new);
634		return;
635	}
636#endif
637	buf_ring_putback_sc(br, new);
638}
639
640static __inline struct mbuf *
641drbr_peek(struct ifnet *ifp, struct buf_ring *br)
642{
643#ifdef ALTQ
644	struct mbuf *m;
645	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd)) {
646		/*
647		 * Pull it off like a dequeue
648		 * since drbr_advance() does nothing
649		 * for altq and drbr_putback() will
650		 * use the old prepend function.
651		 */
652		IFQ_DEQUEUE(&ifp->if_snd, m);
653		return (m);
654	}
655#endif
656	return(buf_ring_peek(br));
657}
658
659static __inline void
660drbr_flush(struct ifnet *ifp, struct buf_ring *br)
661{
662	struct mbuf *m;
663
664#ifdef ALTQ
665	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd))
666		IFQ_PURGE(&ifp->if_snd);
667#endif
668	while ((m = buf_ring_dequeue_sc(br)) != NULL)
669		m_freem(m);
670}
671
672static __inline void
673drbr_free(struct buf_ring *br, struct malloc_type *type)
674{
675
676	drbr_flush(NULL, br);
677	buf_ring_free(br, type);
678}
679
680static __inline struct mbuf *
681drbr_dequeue(struct ifnet *ifp, struct buf_ring *br)
682{
683#ifdef ALTQ
684	struct mbuf *m;
685
686	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd)) {
687		IFQ_DEQUEUE(&ifp->if_snd, m);
688		return (m);
689	}
690#endif
691	return (buf_ring_dequeue_sc(br));
692}
693
694static __inline void
695drbr_advance(struct ifnet *ifp, struct buf_ring *br)
696{
697#ifdef ALTQ
698	/* Nothing to do here since peek dequeues in altq case */
699	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd))
700		return;
701#endif
702	return (buf_ring_advance_sc(br));
703}
704
705
706static __inline struct mbuf *
707drbr_dequeue_cond(struct ifnet *ifp, struct buf_ring *br,
708    int (*func) (struct mbuf *, void *), void *arg)
709{
710	struct mbuf *m;
711#ifdef ALTQ
712	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
713		IFQ_LOCK(&ifp->if_snd);
714		IFQ_POLL_NOLOCK(&ifp->if_snd, m);
715		if (m != NULL && func(m, arg) == 0) {
716			IFQ_UNLOCK(&ifp->if_snd);
717			return (NULL);
718		}
719		IFQ_DEQUEUE_NOLOCK(&ifp->if_snd, m);
720		IFQ_UNLOCK(&ifp->if_snd);
721		return (m);
722	}
723#endif
724	m = buf_ring_peek(br);
725	if (m == NULL || func(m, arg) == 0)
726		return (NULL);
727
728	return (buf_ring_dequeue_sc(br));
729}
730
731static __inline int
732drbr_empty(struct ifnet *ifp, struct buf_ring *br)
733{
734#ifdef ALTQ
735	if (ALTQ_IS_ENABLED(&ifp->if_snd))
736		return (IFQ_IS_EMPTY(&ifp->if_snd));
737#endif
738	return (buf_ring_empty(br));
739}
740
741static __inline int
742drbr_needs_enqueue(struct ifnet *ifp, struct buf_ring *br)
743{
744#ifdef ALTQ
745	if (ALTQ_IS_ENABLED(&ifp->if_snd))
746		return (1);
747#endif
748	return (!buf_ring_empty(br));
749}
750
751static __inline int
752drbr_inuse(struct ifnet *ifp, struct buf_ring *br)
753{
754#ifdef ALTQ
755	if (ALTQ_IS_ENABLED(&ifp->if_snd))
756		return (ifp->if_snd.ifq_len);
757#endif
758	return (buf_ring_count(br));
759}
760#endif
761/*
762 * 72 was chosen below because it is the size of a TCP/IP
763 * header (40) + the minimum mss (32).
764 */
765#define	IF_MINMTU	72
766#define	IF_MAXMTU	65535
767
768#define	TOEDEV(ifp)	((ifp)->if_llsoftc)
769
770#endif /* _KERNEL */
771
772/*
773 * The ifaddr structure contains information about one address
774 * of an interface.  They are maintained by the different address families,
775 * are allocated and attached when an address is set, and are linked
776 * together so all addresses for an interface can be located.
777 *
778 * NOTE: a 'struct ifaddr' is always at the beginning of a larger
779 * chunk of malloc'ed memory, where we store the three addresses
780 * (ifa_addr, ifa_dstaddr and ifa_netmask) referenced here.
781 */
782#if defined(_KERNEL) || defined(_WANT_IFADDR)
783struct ifaddr {
784	struct	sockaddr *ifa_addr;	/* address of interface */
785	struct	sockaddr *ifa_dstaddr;	/* other end of p-to-p link */
786#define	ifa_broadaddr	ifa_dstaddr	/* broadcast address interface */
787	struct	sockaddr *ifa_netmask;	/* used to determine subnet */
788	struct	ifnet *ifa_ifp;		/* back-pointer to interface */
789	struct	carp_softc *ifa_carp;	/* pointer to CARP data */
790	TAILQ_ENTRY(ifaddr) ifa_link;	/* queue macro glue */
791	void	(*ifa_rtrequest)	/* check or clean routes (+ or -)'d */
792		(int, struct rtentry *, struct rt_addrinfo *);
793	u_short	ifa_flags;		/* mostly rt_flags for cloning */
794	u_int	ifa_refcnt;		/* references to this structure */
795	int	ifa_metric;		/* cost of going out this interface */
796	int (*ifa_claim_addr)		/* check if an addr goes to this if */
797		(struct ifaddr *, struct sockaddr *);
798
799	counter_u64_t	ifa_ipackets;
800	counter_u64_t	ifa_opackets;
801	counter_u64_t	ifa_ibytes;
802	counter_u64_t	ifa_obytes;
803};
804#endif
805
806#ifdef _KERNEL
807#define	IFA_ROUTE	RTF_UP		/* route installed */
808#define	IFA_RTSELF	RTF_HOST	/* loopback route to self installed */
809
810/* For compatibility with other BSDs. SCTP uses it. */
811#define	ifa_list	ifa_link
812
813struct ifaddr *	ifa_alloc(size_t size, int flags);
814void	ifa_free(struct ifaddr *ifa);
815void	ifa_ref(struct ifaddr *ifa);
816#endif /* _KERNEL */
817
818/*
819 * Multicast address structure.  This is analogous to the ifaddr
820 * structure except that it keeps track of multicast addresses.
821 */
822struct ifmultiaddr {
823	TAILQ_ENTRY(ifmultiaddr) ifma_link; /* queue macro glue */
824	struct	sockaddr *ifma_addr; 	/* address this membership is for */
825	struct	sockaddr *ifma_lladdr;	/* link-layer translation, if any */
826	struct	ifnet *ifma_ifp;	/* back-pointer to interface */
827	u_int	ifma_refcount;		/* reference count */
828	void	*ifma_protospec;	/* protocol-specific state, if any */
829	struct	ifmultiaddr *ifma_llifma; /* pointer to ifma for ifma_lladdr */
830};
831
832#ifdef _KERNEL
833
834extern	struct rwlock ifnet_rwlock;
835extern	struct sx ifnet_sxlock;
836
837#define	IFNET_LOCK_INIT() do {						\
838	rw_init_flags(&ifnet_rwlock, "ifnet_rw",  RW_RECURSE);		\
839	sx_init_flags(&ifnet_sxlock, "ifnet_sx",  SX_RECURSE);		\
840} while(0)
841
842#define	IFNET_WLOCK() do {						\
843	sx_xlock(&ifnet_sxlock);					\
844	rw_wlock(&ifnet_rwlock);					\
845} while (0)
846
847#define	IFNET_WUNLOCK() do {						\
848	rw_wunlock(&ifnet_rwlock);					\
849	sx_xunlock(&ifnet_sxlock);					\
850} while (0)
851
852/*
853 * To assert the ifnet lock, you must know not only whether it's for read or
854 * write, but also whether it was acquired with sleep support or not.
855 */
856#define	IFNET_RLOCK_ASSERT()		sx_assert(&ifnet_sxlock, SA_SLOCKED)
857#define	IFNET_RLOCK_NOSLEEP_ASSERT()	rw_assert(&ifnet_rwlock, RA_RLOCKED)
858#define	IFNET_WLOCK_ASSERT() do {					\
859	sx_assert(&ifnet_sxlock, SA_XLOCKED);				\
860	rw_assert(&ifnet_rwlock, RA_WLOCKED);				\
861} while (0)
862
863#define	IFNET_RLOCK()		sx_slock(&ifnet_sxlock)
864#define	IFNET_RLOCK_NOSLEEP()	rw_rlock(&ifnet_rwlock)
865#define	IFNET_RUNLOCK()		sx_sunlock(&ifnet_sxlock)
866#define	IFNET_RUNLOCK_NOSLEEP()	rw_runlock(&ifnet_rwlock)
867
868/*
869 * Look up an ifnet given its index; the _ref variant also acquires a
870 * reference that must be freed using if_rele().  It is almost always a bug
871 * to call ifnet_byindex() instead if ifnet_byindex_ref().
872 */
873struct ifnet	*ifnet_byindex(u_short idx);
874struct ifnet	*ifnet_byindex_locked(u_short idx);
875struct ifnet	*ifnet_byindex_ref(u_short idx);
876
877/*
878 * Given the index, ifaddr_byindex() returns the one and only
879 * link-level ifaddr for the interface. You are not supposed to use
880 * it to traverse the list of addresses associated to the interface.
881 */
882struct ifaddr	*ifaddr_byindex(u_short idx);
883
884VNET_DECLARE(struct ifnethead, ifnet);
885VNET_DECLARE(struct ifgrouphead, ifg_head);
886VNET_DECLARE(int, if_index);
887VNET_DECLARE(struct ifnet *, loif);	/* first loopback interface */
888VNET_DECLARE(int, useloopback);
889
890#define	V_ifnet		VNET(ifnet)
891#define	V_ifg_head	VNET(ifg_head)
892#define	V_if_index	VNET(if_index)
893#define	V_loif		VNET(loif)
894#define	V_useloopback	VNET(useloopback)
895
896extern	int ifqmaxlen;
897
898int	if_addgroup(struct ifnet *, const char *);
899int	if_delgroup(struct ifnet *, const char *);
900int	if_addmulti(struct ifnet *, struct sockaddr *, struct ifmultiaddr **);
901int	if_allmulti(struct ifnet *, int);
902struct	ifnet* if_alloc(u_char);
903void	if_attach(struct ifnet *);
904void	if_dead(struct ifnet *);
905int	if_delmulti(struct ifnet *, struct sockaddr *);
906void	if_delmulti_ifma(struct ifmultiaddr *);
907void	if_detach(struct ifnet *);
908void	if_vmove(struct ifnet *, struct vnet *);
909void	if_purgeaddrs(struct ifnet *);
910void	if_delallmulti(struct ifnet *);
911void	if_down(struct ifnet *);
912struct ifmultiaddr *
913	if_findmulti(struct ifnet *, struct sockaddr *);
914void	if_free(struct ifnet *);
915void	if_initname(struct ifnet *, const char *, int);
916void	if_link_state_change(struct ifnet *, int);
917int	if_printf(struct ifnet *, const char *, ...) __printflike(2, 3);
918void	if_qflush(struct ifnet *);
919void	if_ref(struct ifnet *);
920void	if_rele(struct ifnet *);
921int	if_setlladdr(struct ifnet *, const u_char *, int);
922void	if_up(struct ifnet *);
923int	ifioctl(struct socket *, u_long, caddr_t, struct thread *);
924int	ifpromisc(struct ifnet *, int);
925struct	ifnet *ifunit(const char *);
926struct	ifnet *ifunit_ref(const char *);
927
928void	ifq_init(struct ifaltq *, struct ifnet *ifp);
929void	ifq_delete(struct ifaltq *);
930
931int	ifa_add_loopback_route(struct ifaddr *, struct sockaddr *);
932int	ifa_del_loopback_route(struct ifaddr *, struct sockaddr *);
933
934struct	ifaddr *ifa_ifwithaddr(struct sockaddr *);
935int		ifa_ifwithaddr_check(struct sockaddr *);
936struct	ifaddr *ifa_ifwithbroadaddr(struct sockaddr *);
937struct	ifaddr *ifa_ifwithdstaddr(struct sockaddr *);
938struct	ifaddr *ifa_ifwithnet(struct sockaddr *, int);
939struct	ifaddr *ifa_ifwithroute(int, struct sockaddr *, struct sockaddr *);
940struct	ifaddr *ifa_ifwithroute_fib(int, struct sockaddr *, struct sockaddr *, u_int);
941struct	ifaddr *ifaof_ifpforaddr(struct sockaddr *, struct ifnet *);
942int	ifa_preferred(struct ifaddr *, struct ifaddr *);
943
944int	if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen);
945
946typedef	void *if_com_alloc_t(u_char type, struct ifnet *ifp);
947typedef	void if_com_free_t(void *com, u_char type);
948void	if_register_com_alloc(u_char type, if_com_alloc_t *a, if_com_free_t *f);
949void	if_deregister_com_alloc(u_char type);
950
951#define IF_LLADDR(ifp)							\
952    LLADDR((struct sockaddr_dl *)((ifp)->if_addr->ifa_addr))
953
954#ifdef DEVICE_POLLING
955enum poll_cmd {	POLL_ONLY, POLL_AND_CHECK_STATUS };
956
957typedef	int poll_handler_t(struct ifnet *ifp, enum poll_cmd cmd, int count);
958int    ether_poll_register(poll_handler_t *h, struct ifnet *ifp);
959int    ether_poll_deregister(struct ifnet *ifp);
960#endif /* DEVICE_POLLING */
961
962#endif /* _KERNEL */
963
964#endif /* !_NET_IF_VAR_H_ */
965