ifq.h revision 229873
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 229873 2012-01-09 19:34:12Z jhb $
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(ifprefixhead, ifprefix);
100TAILQ_HEAD(ifmultihead, ifmultiaddr);
101TAILQ_HEAD(ifgrouphead, ifg_group);
102
103/*
104 * Structure defining a queue for a network interface.
105 */
106struct	ifqueue {
107	struct	mbuf *ifq_head;
108	struct	mbuf *ifq_tail;
109	int	ifq_len;
110	int	ifq_maxlen;
111	int	ifq_drops;
112	struct	mtx ifq_mtx;
113};
114
115/*
116 * Structure defining a network interface.
117 *
118 * (Would like to call this struct ``if'', but C isn't PL/1.)
119 */
120
121struct ifnet {
122	void	*if_softc;		/* pointer to driver state */
123	void	*if_l2com;		/* pointer to protocol bits */
124	struct vnet *if_vnet;		/* pointer to network stack instance */
125	TAILQ_ENTRY(ifnet) if_link; 	/* all struct ifnets are chained */
126	char	if_xname[IFNAMSIZ];	/* external name (name + unit) */
127	const char *if_dname;		/* driver name */
128	int	if_dunit;		/* unit or IF_DUNIT_NONE */
129	u_int	if_refcount;		/* reference count */
130	struct	ifaddrhead if_addrhead;	/* linked list of addresses per if */
131		/*
132		 * if_addrhead is the list of all addresses associated to
133		 * an interface.
134		 * Some code in the kernel assumes that first element
135		 * of the list has type AF_LINK, and contains sockaddr_dl
136		 * addresses which store the link-level address and the name
137		 * of the interface.
138		 * However, access to the AF_LINK address through this
139		 * field is deprecated. Use if_addr or ifaddr_byindex() instead.
140		 */
141	int	if_pcount;		/* number of promiscuous listeners */
142	struct	carp_if *if_carp;	/* carp interface structure */
143	struct	bpf_if *if_bpf;		/* packet filter structure */
144	u_short	if_index;		/* numeric abbreviation for this if  */
145	short	if_index_reserved;	/* spare space to grow if_index */
146	struct  ifvlantrunk *if_vlantrunk; /* pointer to 802.1q data */
147	int	if_flags;		/* up/down, broadcast, etc. */
148	int	if_capabilities;	/* interface features & capabilities */
149	int	if_capenable;		/* enabled features & capabilities */
150	void	*if_linkmib;		/* link-type-specific MIB data */
151	size_t	if_linkmiblen;		/* length of above data */
152	struct	if_data if_data;
153	struct	ifmultihead if_multiaddrs; /* multicast addresses configured */
154	int	if_amcount;		/* number of all-multicast requests */
155/* procedure handles */
156	int	(*if_output)		/* output routine (enqueue) */
157		(struct ifnet *, struct mbuf *, struct sockaddr *,
158		     struct route *);
159	void	(*if_input)		/* input routine (from h/w driver) */
160		(struct ifnet *, struct mbuf *);
161	void	(*if_start)		/* initiate output routine */
162		(struct ifnet *);
163	int	(*if_ioctl)		/* ioctl routine */
164		(struct ifnet *, u_long, caddr_t);
165	void	(*if_init)		/* Init routine */
166		(void *);
167	int	(*if_resolvemulti)	/* validate/resolve multicast */
168		(struct ifnet *, struct sockaddr **, struct sockaddr *);
169	void	(*if_qflush)		/* flush any queues */
170		(struct ifnet *);
171	int	(*if_transmit)		/* initiate output routine */
172		(struct ifnet *, struct mbuf *);
173	void	(*if_reassign)		/* reassign to vnet routine */
174		(struct ifnet *, struct vnet *, char *);
175	struct	vnet *if_home_vnet;	/* where this ifnet originates from */
176	struct	ifaddr	*if_addr;	/* pointer to link-level address */
177	void	*if_llsoftc;		/* link layer softc */
178	int	if_drv_flags;		/* driver-managed status flags */
179	struct  ifaltq if_snd;		/* output queue (includes altq) */
180	const u_int8_t *if_broadcastaddr; /* linklevel broadcast bytestring */
181
182	void	*if_bridge;		/* bridge glue */
183
184	struct	label *if_label;	/* interface MAC label */
185
186	/* these are only used by IPv6 */
187	struct	ifprefixhead if_prefixhead; /* list of prefixes per if */
188	void	*if_afdata[AF_MAX];
189	int	if_afdata_initialized;
190	struct	rwlock if_afdata_lock;
191	struct	task if_linktask;	/* task for link change events */
192	struct	rwlock if_addr_lock;	/* lock to protect address lists */
193
194	LIST_ENTRY(ifnet) if_clones;	/* interfaces of a cloner */
195	TAILQ_HEAD(, ifg_list) if_groups; /* linked list of groups per if */
196					/* protected by if_addr_lock */
197	void	*if_pf_kif;
198	void	*if_lagg;		/* lagg glue */
199	char	*if_description;	/* interface description */
200	u_int	if_fib;			/* interface FIB */
201	u_char	if_alloctype;		/* if_type at time of allocation */
202
203	/*
204	 * Spare fields are added so that we can modify sensitive data
205	 * structures without changing the kernel binary interface, and must
206	 * be used with care where binary compatibility is required.
207	 */
208	char	if_cspare[3];
209	int	if_ispare[4];
210	void	*if_pspare[8];		/* 1 netmap, 7 TDB */
211};
212
213typedef void if_init_f_t(void *);
214
215/*
216 * XXX These aliases are terribly dangerous because they could apply
217 * to anything.
218 */
219#define	if_mtu		if_data.ifi_mtu
220#define	if_type		if_data.ifi_type
221#define if_physical	if_data.ifi_physical
222#define	if_addrlen	if_data.ifi_addrlen
223#define	if_hdrlen	if_data.ifi_hdrlen
224#define	if_metric	if_data.ifi_metric
225#define	if_link_state	if_data.ifi_link_state
226#define	if_baudrate	if_data.ifi_baudrate
227#define	if_hwassist	if_data.ifi_hwassist
228#define	if_ipackets	if_data.ifi_ipackets
229#define	if_ierrors	if_data.ifi_ierrors
230#define	if_opackets	if_data.ifi_opackets
231#define	if_oerrors	if_data.ifi_oerrors
232#define	if_collisions	if_data.ifi_collisions
233#define	if_ibytes	if_data.ifi_ibytes
234#define	if_obytes	if_data.ifi_obytes
235#define	if_imcasts	if_data.ifi_imcasts
236#define	if_omcasts	if_data.ifi_omcasts
237#define	if_iqdrops	if_data.ifi_iqdrops
238#define	if_noproto	if_data.ifi_noproto
239#define	if_lastchange	if_data.ifi_lastchange
240
241/* for compatibility with other BSDs */
242#define	if_addrlist	if_addrhead
243#define	if_list		if_link
244#define	if_name(ifp)	((ifp)->if_xname)
245
246/*
247 * Locks for address lists on the network interface.
248 */
249#define	IF_ADDR_LOCK_INIT(if)	rw_init(&(if)->if_addr_lock, "if_addr_lock")
250#define	IF_ADDR_LOCK_DESTROY(if)	rw_destroy(&(if)->if_addr_lock)
251#define	IF_ADDR_WLOCK(if)	rw_wlock(&(if)->if_addr_lock)
252#define	IF_ADDR_WUNLOCK(if)	rw_wunlock(&(if)->if_addr_lock)
253#define	IF_ADDR_RLOCK(if)	rw_rlock(&(if)->if_addr_lock)
254#define	IF_ADDR_RUNLOCK(if)	rw_runlock(&(if)->if_addr_lock)
255#define	IF_ADDR_LOCK_ASSERT(if)	rw_assert(&(if)->if_addr_lock, RA_LOCKED)
256#define	IF_ADDR_WLOCK_ASSERT(if) rw_assert(&(if)->if_addr_lock, RA_WLOCKED)
257/* XXX: Compat. */
258#define	IF_ADDR_LOCK(if)	IF_ADDR_WLOCK(if)
259#define	IF_ADDR_UNLOCK(if)	IF_ADDR_WUNLOCK(if)
260
261/*
262 * Function variations on locking macros intended to be used by loadable
263 * kernel modules in order to divorce them from the internals of address list
264 * locking.
265 */
266void	if_addr_rlock(struct ifnet *ifp);	/* if_addrhead */
267void	if_addr_runlock(struct ifnet *ifp);	/* if_addrhead */
268void	if_maddr_rlock(struct ifnet *ifp);	/* if_multiaddrs */
269void	if_maddr_runlock(struct ifnet *ifp);	/* if_multiaddrs */
270
271/*
272 * Output queues (ifp->if_snd) and slow device input queues (*ifp->if_slowq)
273 * are queues of messages stored on ifqueue structures
274 * (defined above).  Entries are added to and deleted from these structures
275 * by these macros, which should be called with ipl raised to splimp().
276 */
277#define IF_LOCK(ifq)		mtx_lock(&(ifq)->ifq_mtx)
278#define IF_UNLOCK(ifq)		mtx_unlock(&(ifq)->ifq_mtx)
279#define	IF_LOCK_ASSERT(ifq)	mtx_assert(&(ifq)->ifq_mtx, MA_OWNED)
280#define	_IF_QFULL(ifq)		((ifq)->ifq_len >= (ifq)->ifq_maxlen)
281#define	_IF_DROP(ifq)		((ifq)->ifq_drops++)
282#define	_IF_QLEN(ifq)		((ifq)->ifq_len)
283
284#define	_IF_ENQUEUE(ifq, m) do { 				\
285	(m)->m_nextpkt = NULL;					\
286	if ((ifq)->ifq_tail == NULL) 				\
287		(ifq)->ifq_head = m; 				\
288	else 							\
289		(ifq)->ifq_tail->m_nextpkt = m; 		\
290	(ifq)->ifq_tail = m; 					\
291	(ifq)->ifq_len++; 					\
292} while (0)
293
294#define IF_ENQUEUE(ifq, m) do {					\
295	IF_LOCK(ifq); 						\
296	_IF_ENQUEUE(ifq, m); 					\
297	IF_UNLOCK(ifq); 					\
298} while (0)
299
300#define	_IF_PREPEND(ifq, m) do {				\
301	(m)->m_nextpkt = (ifq)->ifq_head; 			\
302	if ((ifq)->ifq_tail == NULL) 				\
303		(ifq)->ifq_tail = (m); 				\
304	(ifq)->ifq_head = (m); 					\
305	(ifq)->ifq_len++; 					\
306} while (0)
307
308#define IF_PREPEND(ifq, m) do {		 			\
309	IF_LOCK(ifq); 						\
310	_IF_PREPEND(ifq, m); 					\
311	IF_UNLOCK(ifq); 					\
312} while (0)
313
314#define	_IF_DEQUEUE(ifq, m) do { 				\
315	(m) = (ifq)->ifq_head; 					\
316	if (m) { 						\
317		if (((ifq)->ifq_head = (m)->m_nextpkt) == NULL)	\
318			(ifq)->ifq_tail = NULL; 		\
319		(m)->m_nextpkt = NULL; 				\
320		(ifq)->ifq_len--; 				\
321	} 							\
322} while (0)
323
324#define IF_DEQUEUE(ifq, m) do { 				\
325	IF_LOCK(ifq); 						\
326	_IF_DEQUEUE(ifq, m); 					\
327	IF_UNLOCK(ifq); 					\
328} while (0)
329
330#define	_IF_DEQUEUE_ALL(ifq, m) do {				\
331	(m) = (ifq)->ifq_head;					\
332	(ifq)->ifq_head = (ifq)->ifq_tail = NULL;		\
333	(ifq)->ifq_len = 0;					\
334} while (0)
335
336#define	IF_DEQUEUE_ALL(ifq, m) do {				\
337	IF_LOCK(ifq); 						\
338	_IF_DEQUEUE_ALL(ifq, m);				\
339	IF_UNLOCK(ifq); 					\
340} while (0)
341
342#define	_IF_POLL(ifq, m)	((m) = (ifq)->ifq_head)
343#define	IF_POLL(ifq, m)		_IF_POLL(ifq, m)
344
345#define _IF_DRAIN(ifq) do { 					\
346	struct mbuf *m; 					\
347	for (;;) { 						\
348		_IF_DEQUEUE(ifq, m); 				\
349		if (m == NULL) 					\
350			break; 					\
351		m_freem(m); 					\
352	} 							\
353} while (0)
354
355#define IF_DRAIN(ifq) do {					\
356	IF_LOCK(ifq);						\
357	_IF_DRAIN(ifq);						\
358	IF_UNLOCK(ifq);						\
359} while(0)
360
361#ifdef _KERNEL
362/* interface link layer address change event */
363typedef void (*iflladdr_event_handler_t)(void *, struct ifnet *);
364EVENTHANDLER_DECLARE(iflladdr_event, iflladdr_event_handler_t);
365/* interface address change event */
366typedef void (*ifaddr_event_handler_t)(void *, struct ifnet *);
367EVENTHANDLER_DECLARE(ifaddr_event, ifaddr_event_handler_t);
368/* new interface arrival event */
369typedef void (*ifnet_arrival_event_handler_t)(void *, struct ifnet *);
370EVENTHANDLER_DECLARE(ifnet_arrival_event, ifnet_arrival_event_handler_t);
371/* interface departure event */
372typedef void (*ifnet_departure_event_handler_t)(void *, struct ifnet *);
373EVENTHANDLER_DECLARE(ifnet_departure_event, ifnet_departure_event_handler_t);
374/* Interface link state change event */
375typedef void (*ifnet_link_event_handler_t)(void *, struct ifnet *, int);
376EVENTHANDLER_DECLARE(ifnet_link_event, ifnet_link_event_handler_t);
377
378/*
379 * interface groups
380 */
381struct ifg_group {
382	char				 ifg_group[IFNAMSIZ];
383	u_int				 ifg_refcnt;
384	void				*ifg_pf_kif;
385	TAILQ_HEAD(, ifg_member)	 ifg_members;
386	TAILQ_ENTRY(ifg_group)		 ifg_next;
387};
388
389struct ifg_member {
390	TAILQ_ENTRY(ifg_member)	 ifgm_next;
391	struct ifnet		*ifgm_ifp;
392};
393
394struct ifg_list {
395	struct ifg_group	*ifgl_group;
396	TAILQ_ENTRY(ifg_list)	 ifgl_next;
397};
398
399/* group attach event */
400typedef void (*group_attach_event_handler_t)(void *, struct ifg_group *);
401EVENTHANDLER_DECLARE(group_attach_event, group_attach_event_handler_t);
402/* group detach event */
403typedef void (*group_detach_event_handler_t)(void *, struct ifg_group *);
404EVENTHANDLER_DECLARE(group_detach_event, group_detach_event_handler_t);
405/* group change event */
406typedef void (*group_change_event_handler_t)(void *, const char *);
407EVENTHANDLER_DECLARE(group_change_event, group_change_event_handler_t);
408
409#define	IF_AFDATA_LOCK_INIT(ifp)	\
410	rw_init(&(ifp)->if_afdata_lock, "if_afdata")
411
412#define	IF_AFDATA_WLOCK(ifp)	rw_wlock(&(ifp)->if_afdata_lock)
413#define	IF_AFDATA_RLOCK(ifp)	rw_rlock(&(ifp)->if_afdata_lock)
414#define	IF_AFDATA_WUNLOCK(ifp)	rw_wunlock(&(ifp)->if_afdata_lock)
415#define	IF_AFDATA_RUNLOCK(ifp)	rw_runlock(&(ifp)->if_afdata_lock)
416#define	IF_AFDATA_LOCK(ifp)	IF_AFDATA_WLOCK(ifp)
417#define	IF_AFDATA_UNLOCK(ifp)	IF_AFDATA_WUNLOCK(ifp)
418#define	IF_AFDATA_TRYLOCK(ifp)	rw_try_wlock(&(ifp)->if_afdata_lock)
419#define	IF_AFDATA_DESTROY(ifp)	rw_destroy(&(ifp)->if_afdata_lock)
420
421#define	IF_AFDATA_LOCK_ASSERT(ifp)	rw_assert(&(ifp)->if_afdata_lock, RA_LOCKED)
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
591drbr_stats_update(struct ifnet *ifp, int len, int mflags)
592{
593#ifndef NO_SLOW_STATS
594	ifp->if_obytes += len;
595	if (mflags & M_MCAST)
596		ifp->if_omcasts++;
597#endif
598}
599
600static __inline int
601drbr_enqueue(struct ifnet *ifp, struct buf_ring *br, struct mbuf *m)
602{
603	int error = 0;
604	int len = m->m_pkthdr.len;
605	int mflags = m->m_flags;
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	if ((error = buf_ring_enqueue_bytes(br, m, len)) == ENOBUFS) {
614		br->br_drops++;
615		m_freem(m);
616	} else
617		drbr_stats_update(ifp, len, mflags);
618
619	return (error);
620}
621
622static __inline void
623drbr_flush(struct ifnet *ifp, struct buf_ring *br)
624{
625	struct mbuf *m;
626
627#ifdef ALTQ
628	if (ifp != NULL && ALTQ_IS_ENABLED(&ifp->if_snd))
629		IFQ_PURGE(&ifp->if_snd);
630#endif
631	while ((m = buf_ring_dequeue_sc(br)) != NULL)
632		m_freem(m);
633}
634
635static __inline void
636drbr_free(struct buf_ring *br, struct malloc_type *type)
637{
638
639	drbr_flush(NULL, br);
640	buf_ring_free(br, type);
641}
642
643static __inline struct mbuf *
644drbr_dequeue(struct ifnet *ifp, struct buf_ring *br)
645{
646#ifdef ALTQ
647	struct mbuf *m;
648
649	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
650		IFQ_DEQUEUE(&ifp->if_snd, m);
651		return (m);
652	}
653#endif
654	return (buf_ring_dequeue_sc(br));
655}
656
657static __inline struct mbuf *
658drbr_dequeue_cond(struct ifnet *ifp, struct buf_ring *br,
659    int (*func) (struct mbuf *, void *), void *arg)
660{
661	struct mbuf *m;
662#ifdef ALTQ
663	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
664		IFQ_LOCK(&ifp->if_snd);
665		IFQ_POLL_NOLOCK(&ifp->if_snd, m);
666		if (m != NULL && func(m, arg) == 0) {
667			IFQ_UNLOCK(&ifp->if_snd);
668			return (NULL);
669		}
670		IFQ_DEQUEUE_NOLOCK(&ifp->if_snd, m);
671		IFQ_UNLOCK(&ifp->if_snd);
672		return (m);
673	}
674#endif
675	m = buf_ring_peek(br);
676	if (m == NULL || func(m, arg) == 0)
677		return (NULL);
678
679	return (buf_ring_dequeue_sc(br));
680}
681
682static __inline int
683drbr_empty(struct ifnet *ifp, struct buf_ring *br)
684{
685#ifdef ALTQ
686	if (ALTQ_IS_ENABLED(&ifp->if_snd))
687		return (IFQ_IS_EMPTY(&ifp->if_snd));
688#endif
689	return (buf_ring_empty(br));
690}
691
692static __inline int
693drbr_needs_enqueue(struct ifnet *ifp, struct buf_ring *br)
694{
695#ifdef ALTQ
696	if (ALTQ_IS_ENABLED(&ifp->if_snd))
697		return (1);
698#endif
699	return (!buf_ring_empty(br));
700}
701
702static __inline int
703drbr_inuse(struct ifnet *ifp, struct buf_ring *br)
704{
705#ifdef ALTQ
706	if (ALTQ_IS_ENABLED(&ifp->if_snd))
707		return (ifp->if_snd.ifq_len);
708#endif
709	return (buf_ring_count(br));
710}
711#endif
712/*
713 * 72 was chosen below because it is the size of a TCP/IP
714 * header (40) + the minimum mss (32).
715 */
716#define	IF_MINMTU	72
717#define	IF_MAXMTU	65535
718
719#endif /* _KERNEL */
720
721/*
722 * The ifaddr structure contains information about one address
723 * of an interface.  They are maintained by the different address families,
724 * are allocated and attached when an address is set, and are linked
725 * together so all addresses for an interface can be located.
726 *
727 * NOTE: a 'struct ifaddr' is always at the beginning of a larger
728 * chunk of malloc'ed memory, where we store the three addresses
729 * (ifa_addr, ifa_dstaddr and ifa_netmask) referenced here.
730 */
731struct ifaddr {
732	struct	sockaddr *ifa_addr;	/* address of interface */
733	struct	sockaddr *ifa_dstaddr;	/* other end of p-to-p link */
734#define	ifa_broadaddr	ifa_dstaddr	/* broadcast address interface */
735	struct	sockaddr *ifa_netmask;	/* used to determine subnet */
736	struct	if_data if_data;	/* not all members are meaningful */
737	struct	ifnet *ifa_ifp;		/* back-pointer to interface */
738	struct	carp_softc *ifa_carp;	/* pointer to CARP data */
739	TAILQ_ENTRY(ifaddr) ifa_link;	/* queue macro glue */
740	void	(*ifa_rtrequest)	/* check or clean routes (+ or -)'d */
741		(int, struct rtentry *, struct rt_addrinfo *);
742	u_short	ifa_flags;		/* mostly rt_flags for cloning */
743	u_int	ifa_refcnt;		/* references to this structure */
744	int	ifa_metric;		/* cost of going out this interface */
745	int (*ifa_claim_addr)		/* check if an addr goes to this if */
746		(struct ifaddr *, struct sockaddr *);
747	struct mtx ifa_mtx;
748};
749#define	IFA_ROUTE	RTF_UP		/* route installed */
750#define IFA_RTSELF	RTF_HOST	/* loopback route to self installed */
751
752/* for compatibility with other BSDs */
753#define	ifa_list	ifa_link
754
755#ifdef _KERNEL
756#define	IFA_LOCK(ifa)		mtx_lock(&(ifa)->ifa_mtx)
757#define	IFA_UNLOCK(ifa)		mtx_unlock(&(ifa)->ifa_mtx)
758
759void	ifa_free(struct ifaddr *ifa);
760void	ifa_init(struct ifaddr *ifa);
761void	ifa_ref(struct ifaddr *ifa);
762#endif
763
764/*
765 * The prefix structure contains information about one prefix
766 * of an interface.  They are maintained by the different address families,
767 * are allocated and attached when a prefix or an address is set,
768 * and are linked together so all prefixes for an interface can be located.
769 */
770struct ifprefix {
771	struct	sockaddr *ifpr_prefix;	/* prefix of interface */
772	struct	ifnet *ifpr_ifp;	/* back-pointer to interface */
773	TAILQ_ENTRY(ifprefix) ifpr_list; /* queue macro glue */
774	u_char	ifpr_plen;		/* prefix length in bits */
775	u_char	ifpr_type;		/* protocol dependent prefix type */
776};
777
778/*
779 * Multicast address structure.  This is analogous to the ifaddr
780 * structure except that it keeps track of multicast addresses.
781 */
782struct ifmultiaddr {
783	TAILQ_ENTRY(ifmultiaddr) ifma_link; /* queue macro glue */
784	struct	sockaddr *ifma_addr; 	/* address this membership is for */
785	struct	sockaddr *ifma_lladdr;	/* link-layer translation, if any */
786	struct	ifnet *ifma_ifp;	/* back-pointer to interface */
787	u_int	ifma_refcount;		/* reference count */
788	void	*ifma_protospec;	/* protocol-specific state, if any */
789	struct	ifmultiaddr *ifma_llifma; /* pointer to ifma for ifma_lladdr */
790};
791
792#ifdef _KERNEL
793
794extern	struct rwlock ifnet_rwlock;
795extern	struct sx ifnet_sxlock;
796
797#define	IFNET_LOCK_INIT() do {						\
798	rw_init_flags(&ifnet_rwlock, "ifnet_rw",  RW_RECURSE);		\
799	sx_init_flags(&ifnet_sxlock, "ifnet_sx",  SX_RECURSE);		\
800} while(0)
801
802#define	IFNET_WLOCK() do {						\
803	sx_xlock(&ifnet_sxlock);					\
804	rw_wlock(&ifnet_rwlock);					\
805} while (0)
806
807#define	IFNET_WUNLOCK() do {						\
808	rw_wunlock(&ifnet_rwlock);					\
809	sx_xunlock(&ifnet_sxlock);					\
810} while (0)
811
812/*
813 * To assert the ifnet lock, you must know not only whether it's for read or
814 * write, but also whether it was acquired with sleep support or not.
815 */
816#define	IFNET_RLOCK_ASSERT()		sx_assert(&ifnet_sxlock, SA_SLOCKED)
817#define	IFNET_RLOCK_NOSLEEP_ASSERT()	rw_assert(&ifnet_rwlock, RA_RLOCKED)
818#define	IFNET_WLOCK_ASSERT() do {					\
819	sx_assert(&ifnet_sxlock, SA_XLOCKED);				\
820	rw_assert(&ifnet_rwlock, RA_WLOCKED);				\
821} while (0)
822
823#define	IFNET_RLOCK()		sx_slock(&ifnet_sxlock)
824#define	IFNET_RLOCK_NOSLEEP()	rw_rlock(&ifnet_rwlock)
825#define	IFNET_RUNLOCK()		sx_sunlock(&ifnet_sxlock)
826#define	IFNET_RUNLOCK_NOSLEEP()	rw_runlock(&ifnet_rwlock)
827
828/*
829 * Look up an ifnet given its index; the _ref variant also acquires a
830 * reference that must be freed using if_rele().  It is almost always a bug
831 * to call ifnet_byindex() instead if ifnet_byindex_ref().
832 */
833struct ifnet	*ifnet_byindex(u_short idx);
834struct ifnet	*ifnet_byindex_locked(u_short idx);
835struct ifnet	*ifnet_byindex_ref(u_short idx);
836
837/*
838 * Given the index, ifaddr_byindex() returns the one and only
839 * link-level ifaddr for the interface. You are not supposed to use
840 * it to traverse the list of addresses associated to the interface.
841 */
842struct ifaddr	*ifaddr_byindex(u_short idx);
843
844VNET_DECLARE(struct ifnethead, ifnet);
845VNET_DECLARE(struct ifgrouphead, ifg_head);
846VNET_DECLARE(int, if_index);
847VNET_DECLARE(struct ifnet *, loif);	/* first loopback interface */
848VNET_DECLARE(int, useloopback);
849
850#define	V_ifnet		VNET(ifnet)
851#define	V_ifg_head	VNET(ifg_head)
852#define	V_if_index	VNET(if_index)
853#define	V_loif		VNET(loif)
854#define	V_useloopback	VNET(useloopback)
855
856extern	int ifqmaxlen;
857
858int	if_addgroup(struct ifnet *, const char *);
859int	if_delgroup(struct ifnet *, const char *);
860int	if_addmulti(struct ifnet *, struct sockaddr *, struct ifmultiaddr **);
861int	if_allmulti(struct ifnet *, int);
862struct	ifnet* if_alloc(u_char);
863void	if_attach(struct ifnet *);
864void	if_dead(struct ifnet *);
865int	if_delmulti(struct ifnet *, struct sockaddr *);
866void	if_delmulti_ifma(struct ifmultiaddr *);
867void	if_detach(struct ifnet *);
868void	if_vmove(struct ifnet *, struct vnet *);
869void	if_purgeaddrs(struct ifnet *);
870void	if_delallmulti(struct ifnet *);
871void	if_down(struct ifnet *);
872struct ifmultiaddr *
873	if_findmulti(struct ifnet *, struct sockaddr *);
874void	if_free(struct ifnet *);
875void	if_initname(struct ifnet *, const char *, int);
876void	if_link_state_change(struct ifnet *, int);
877int	if_printf(struct ifnet *, const char *, ...) __printflike(2, 3);
878void	if_qflush(struct ifnet *);
879void	if_ref(struct ifnet *);
880void	if_rele(struct ifnet *);
881int	if_setlladdr(struct ifnet *, const u_char *, int);
882void	if_up(struct ifnet *);
883int	ifioctl(struct socket *, u_long, caddr_t, struct thread *);
884int	ifpromisc(struct ifnet *, int);
885struct	ifnet *ifunit(const char *);
886struct	ifnet *ifunit_ref(const char *);
887
888void	ifq_init(struct ifaltq *, struct ifnet *ifp);
889void	ifq_delete(struct ifaltq *);
890
891int	ifa_add_loopback_route(struct ifaddr *, struct sockaddr *);
892int	ifa_del_loopback_route(struct ifaddr *, struct sockaddr *);
893
894struct	ifaddr *ifa_ifwithaddr(struct sockaddr *);
895int		ifa_ifwithaddr_check(struct sockaddr *);
896struct	ifaddr *ifa_ifwithbroadaddr(struct sockaddr *);
897struct	ifaddr *ifa_ifwithdstaddr(struct sockaddr *);
898struct	ifaddr *ifa_ifwithnet(struct sockaddr *, int);
899struct	ifaddr *ifa_ifwithroute(int, struct sockaddr *, struct sockaddr *);
900struct	ifaddr *ifa_ifwithroute_fib(int, struct sockaddr *, struct sockaddr *, u_int);
901
902struct	ifaddr *ifaof_ifpforaddr(struct sockaddr *, struct ifnet *);
903
904int	if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen);
905
906typedef	void *if_com_alloc_t(u_char type, struct ifnet *ifp);
907typedef	void if_com_free_t(void *com, u_char type);
908void	if_register_com_alloc(u_char type, if_com_alloc_t *a, if_com_free_t *f);
909void	if_deregister_com_alloc(u_char type);
910
911#define IF_LLADDR(ifp)							\
912    LLADDR((struct sockaddr_dl *)((ifp)->if_addr->ifa_addr))
913
914#ifdef DEVICE_POLLING
915enum poll_cmd {	POLL_ONLY, POLL_AND_CHECK_STATUS };
916
917typedef	int poll_handler_t(struct ifnet *ifp, enum poll_cmd cmd, int count);
918int    ether_poll_register(poll_handler_t *h, struct ifnet *ifp);
919int    ether_poll_deregister(struct ifnet *ifp);
920#endif /* DEVICE_POLLING */
921
922#endif /* _KERNEL */
923
924#endif /* !_NET_IF_VAR_H_ */
925