1/*
2 * Copyright (c) 2004 Luigi Rizzo, Alessandro Cerri. All rights reserved.
3 * Copyright (c) 2004-2008 Qing Li. All rights reserved.
4 * Copyright (c) 2008 Kip Macy. All rights reserved.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 * 1. Redistributions of source code must retain the above copyright
10 *    notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 *    notice, this list of conditions and the following disclaimer in the
13 *    documentation and/or other materials provided with the distribution.
14 *
15 * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS ``AS IS'' AND
16 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
17 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
18 * ARE DISCLAIMED.  IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE
19 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
20 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
21 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
22 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
23 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
24 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
25 * SUCH DAMAGE.
26 */
27#include <sys/cdefs.h>
28__FBSDID("$FreeBSD$");
29
30#include "opt_ddb.h"
31#include "opt_inet.h"
32#include "opt_inet6.h"
33
34#include <sys/param.h>
35#include <sys/systm.h>
36#include <sys/malloc.h>
37#include <sys/mbuf.h>
38#include <sys/syslog.h>
39#include <sys/sysctl.h>
40#include <sys/socket.h>
41#include <sys/kernel.h>
42#include <sys/lock.h>
43#include <sys/mutex.h>
44#include <sys/rwlock.h>
45
46#ifdef DDB
47#include <ddb/ddb.h>
48#endif
49
50#include <vm/uma.h>
51
52#include <netinet/in.h>
53#include <net/if_llatbl.h>
54#include <net/if.h>
55#include <net/if_dl.h>
56#include <net/if_var.h>
57#include <net/route.h>
58#include <net/vnet.h>
59#include <netinet/if_ether.h>
60#include <netinet6/in6_var.h>
61#include <netinet6/nd6.h>
62
63MALLOC_DEFINE(M_LLTABLE, "lltable", "link level address tables");
64
65static VNET_DEFINE(SLIST_HEAD(, lltable), lltables) =
66    SLIST_HEAD_INITIALIZER(lltables);
67#define	V_lltables	VNET(lltables)
68
69struct rwlock lltable_rwlock;
70RW_SYSINIT(lltable_rwlock, &lltable_rwlock, "lltable_rwlock");
71
72/*
73 * Dump arp state for a specific address family.
74 */
75int
76lltable_sysctl_dumparp(int af, struct sysctl_req *wr)
77{
78	struct lltable *llt;
79	int error = 0;
80
81	LLTABLE_RLOCK();
82	SLIST_FOREACH(llt, &V_lltables, llt_link) {
83		if (llt->llt_af == af) {
84			error = llt->llt_dump(llt, wr);
85			if (error != 0)
86				goto done;
87		}
88	}
89done:
90	LLTABLE_RUNLOCK();
91	return (error);
92}
93
94/*
95 * Deletes an address from the address table.
96 * This function is called by the timer functions
97 * such as arptimer() and nd6_llinfo_timer(), and
98 * the caller does the locking.
99 *
100 * Returns the number of held packets, if any, that were dropped.
101 */
102size_t
103llentry_free(struct llentry *lle)
104{
105	size_t pkts_dropped;
106	struct mbuf *next;
107
108	IF_AFDATA_WLOCK_ASSERT(lle->lle_tbl->llt_ifp);
109	LLE_WLOCK_ASSERT(lle);
110
111	LIST_REMOVE(lle, lle_next);
112	lle->la_flags &= ~(LLE_VALID | LLE_LINKED);
113
114	pkts_dropped = 0;
115	while ((lle->la_numheld > 0) && (lle->la_hold != NULL)) {
116		next = lle->la_hold->m_nextpkt;
117		m_freem(lle->la_hold);
118		lle->la_hold = next;
119		lle->la_numheld--;
120		pkts_dropped++;
121	}
122
123	KASSERT(lle->la_numheld == 0,
124		("%s: la_numheld %d > 0, pkts_droped %zd", __func__,
125		 lle->la_numheld, pkts_dropped));
126
127	LLE_FREE_LOCKED(lle);
128
129	return (pkts_dropped);
130}
131
132/*
133 * (al)locate an llentry for address dst (equivalent to rtalloc for new-arp).
134 *
135 * If found the llentry * is returned referenced and unlocked.
136 */
137struct llentry *
138llentry_alloc(struct ifnet *ifp, struct lltable *lt,
139    struct sockaddr_storage *dst)
140{
141	struct llentry *la;
142
143	IF_AFDATA_RLOCK(ifp);
144	la = lla_lookup(lt, LLE_EXCLUSIVE, (struct sockaddr *)dst);
145	IF_AFDATA_RUNLOCK(ifp);
146	if ((la == NULL) &&
147	    (ifp->if_flags & (IFF_NOARP | IFF_STATICARP)) == 0) {
148		IF_AFDATA_WLOCK(ifp);
149		la = lla_lookup(lt, (LLE_CREATE | LLE_EXCLUSIVE),
150		    (struct sockaddr *)dst);
151		IF_AFDATA_WUNLOCK(ifp);
152	}
153
154	if (la != NULL) {
155		LLE_ADDREF(la);
156		LLE_WUNLOCK(la);
157	}
158
159	return (la);
160}
161
162/*
163 * Free all entries from given table and free itself.
164 */
165void
166lltable_free(struct lltable *llt)
167{
168	struct llentry *lle, *next;
169	int i;
170
171	KASSERT(llt != NULL, ("%s: llt is NULL", __func__));
172
173	LLTABLE_WLOCK();
174	SLIST_REMOVE(&V_lltables, llt, lltable, llt_link);
175	LLTABLE_WUNLOCK();
176
177	IF_AFDATA_WLOCK(llt->llt_ifp);
178	for (i = 0; i < LLTBL_HASHTBL_SIZE; i++) {
179		LIST_FOREACH_SAFE(lle, &llt->lle_head[i], lle_next, next) {
180			LLE_WLOCK(lle);
181			if (callout_stop(&lle->la_timer))
182				LLE_REMREF(lle);
183			llentry_free(lle);
184		}
185	}
186	IF_AFDATA_WUNLOCK(llt->llt_ifp);
187
188	free(llt, M_LLTABLE);
189}
190
191#if 0
192void
193lltable_drain(int af)
194{
195	struct lltable	*llt;
196	struct llentry	*lle;
197	register int i;
198
199	LLTABLE_RLOCK();
200	SLIST_FOREACH(llt, &V_lltables, llt_link) {
201		if (llt->llt_af != af)
202			continue;
203
204		for (i=0; i < LLTBL_HASHTBL_SIZE; i++) {
205			LIST_FOREACH(lle, &llt->lle_head[i], lle_next) {
206				LLE_WLOCK(lle);
207				if (lle->la_hold) {
208					m_freem(lle->la_hold);
209					lle->la_hold = NULL;
210				}
211				LLE_WUNLOCK(lle);
212			}
213		}
214	}
215	LLTABLE_RUNLOCK();
216}
217#endif
218
219void
220lltable_prefix_free(int af, struct sockaddr *prefix, struct sockaddr *mask,
221    u_int flags)
222{
223	struct lltable *llt;
224
225	LLTABLE_RLOCK();
226	SLIST_FOREACH(llt, &V_lltables, llt_link) {
227		if (llt->llt_af != af)
228			continue;
229
230		llt->llt_prefix_free(llt, prefix, mask, flags);
231	}
232	LLTABLE_RUNLOCK();
233}
234
235
236
237/*
238 * Create a new lltable.
239 */
240struct lltable *
241lltable_init(struct ifnet *ifp, int af)
242{
243	struct lltable *llt;
244	register int i;
245
246	llt = malloc(sizeof(struct lltable), M_LLTABLE, M_WAITOK);
247
248	llt->llt_af = af;
249	llt->llt_ifp = ifp;
250	for (i = 0; i < LLTBL_HASHTBL_SIZE; i++)
251		LIST_INIT(&llt->lle_head[i]);
252
253	LLTABLE_WLOCK();
254	SLIST_INSERT_HEAD(&V_lltables, llt, llt_link);
255	LLTABLE_WUNLOCK();
256
257	return (llt);
258}
259
260/*
261 * Called in route_output when adding/deleting a route to an interface.
262 */
263int
264lla_rt_output(struct rt_msghdr *rtm, struct rt_addrinfo *info)
265{
266	struct sockaddr_dl *dl =
267	    (struct sockaddr_dl *)info->rti_info[RTAX_GATEWAY];
268	struct sockaddr *dst = (struct sockaddr *)info->rti_info[RTAX_DST];
269	struct ifnet *ifp;
270	struct lltable *llt;
271	struct llentry *lle;
272	u_int laflags = 0, flags = 0;
273	int error = 0;
274
275	KASSERT(dl != NULL && dl->sdl_family == AF_LINK,
276	    ("%s: invalid dl\n", __func__));
277
278	ifp = ifnet_byindex(dl->sdl_index);
279	if (ifp == NULL) {
280		log(LOG_INFO, "%s: invalid ifp (sdl_index %d)\n",
281		    __func__, dl->sdl_index);
282		return EINVAL;
283	}
284
285	switch (rtm->rtm_type) {
286	case RTM_ADD:
287		if (rtm->rtm_flags & RTF_ANNOUNCE)
288			flags |= LLE_PUB;
289		flags |= LLE_CREATE;
290		break;
291
292	case RTM_DELETE:
293		flags |= LLE_DELETE;
294		break;
295
296	case RTM_CHANGE:
297		break;
298
299	default:
300		return EINVAL; /* XXX not implemented yet */
301	}
302
303	/* XXX linked list may be too expensive */
304	LLTABLE_RLOCK();
305	SLIST_FOREACH(llt, &V_lltables, llt_link) {
306		if (llt->llt_af == dst->sa_family &&
307		    llt->llt_ifp == ifp)
308			break;
309	}
310	LLTABLE_RUNLOCK();
311	KASSERT(llt != NULL, ("Yep, ugly hacks are bad\n"));
312
313	if (flags & LLE_CREATE)
314		flags |= LLE_EXCLUSIVE;
315
316	IF_AFDATA_LOCK(ifp);
317	lle = lla_lookup(llt, flags, dst);
318	IF_AFDATA_UNLOCK(ifp);
319	if (LLE_IS_VALID(lle)) {
320		if (flags & LLE_CREATE) {
321			/*
322			 * If we delay the delete, then a subsequent
323			 * "arp add" should look up this entry, reset the
324			 * LLE_DELETED flag, and reset the expiration timer
325			 */
326			bcopy(LLADDR(dl), &lle->ll_addr, ifp->if_addrlen);
327			lle->la_flags |= (flags & LLE_PUB);
328			lle->la_flags |= LLE_VALID;
329			lle->la_flags &= ~LLE_DELETED;
330#ifdef INET6
331			/*
332			 * ND6
333			 */
334			if (dst->sa_family == AF_INET6)
335				lle->ln_state = ND6_LLINFO_REACHABLE;
336#endif
337			/*
338			 * NB: arp and ndp always set (RTF_STATIC | RTF_HOST)
339			 */
340
341			if (rtm->rtm_rmx.rmx_expire == 0) {
342				lle->la_flags |= LLE_STATIC;
343				lle->la_expire = 0;
344			} else
345				lle->la_expire = rtm->rtm_rmx.rmx_expire;
346			laflags = lle->la_flags;
347			LLE_WUNLOCK(lle);
348#ifdef INET
349			/* gratuitous ARP */
350			if ((laflags & LLE_PUB) && dst->sa_family == AF_INET)
351				arprequest(ifp,
352				    &((struct sockaddr_in *)dst)->sin_addr,
353				    &((struct sockaddr_in *)dst)->sin_addr,
354				    (u_char *)LLADDR(dl));
355#endif
356		} else {
357			if (flags & LLE_EXCLUSIVE)
358				LLE_WUNLOCK(lle);
359			else
360				LLE_RUNLOCK(lle);
361		}
362	} else if ((lle == NULL) && (flags & LLE_DELETE))
363		error = EINVAL;
364
365
366	return (error);
367}
368
369#ifdef DDB
370struct llentry_sa {
371	struct llentry		base;
372	struct sockaddr		l3_addr;
373};
374
375static void
376llatbl_lle_show(struct llentry_sa *la)
377{
378	struct llentry *lle;
379	uint8_t octet[6];
380
381	lle = &la->base;
382	db_printf("lle=%p\n", lle);
383	db_printf(" lle_next=%p\n", lle->lle_next.le_next);
384	db_printf(" lle_lock=%p\n", &lle->lle_lock);
385	db_printf(" lle_tbl=%p\n", lle->lle_tbl);
386	db_printf(" lle_head=%p\n", lle->lle_head);
387	db_printf(" la_hold=%p\n", lle->la_hold);
388	db_printf(" la_numheld=%d\n", lle->la_numheld);
389	db_printf(" la_expire=%ju\n", (uintmax_t)lle->la_expire);
390	db_printf(" la_flags=0x%04x\n", lle->la_flags);
391	db_printf(" la_asked=%u\n", lle->la_asked);
392	db_printf(" la_preempt=%u\n", lle->la_preempt);
393	db_printf(" ln_byhint=%u\n", lle->ln_byhint);
394	db_printf(" ln_state=%d\n", lle->ln_state);
395	db_printf(" ln_router=%u\n", lle->ln_router);
396	db_printf(" ln_ntick=%ju\n", (uintmax_t)lle->ln_ntick);
397	db_printf(" lle_refcnt=%d\n", lle->lle_refcnt);
398	bcopy(&lle->ll_addr.mac16, octet, sizeof(octet));
399	db_printf(" ll_addr=%02x:%02x:%02x:%02x:%02x:%02x\n",
400	    octet[0], octet[1], octet[2], octet[3], octet[4], octet[5]);
401	db_printf(" la_timer=%p\n", &lle->la_timer);
402
403	switch (la->l3_addr.sa_family) {
404#ifdef INET
405	case AF_INET:
406	{
407		struct sockaddr_in *sin;
408		char l3s[INET_ADDRSTRLEN];
409
410		sin = (struct sockaddr_in *)&la->l3_addr;
411		inet_ntoa_r(sin->sin_addr, l3s);
412		db_printf(" l3_addr=%s\n", l3s);
413		break;
414	}
415#endif
416#ifdef INET6
417	case AF_INET6:
418	{
419		struct sockaddr_in6 *sin6;
420		char l3s[INET6_ADDRSTRLEN];
421
422		sin6 = (struct sockaddr_in6 *)&la->l3_addr;
423		ip6_sprintf(l3s, &sin6->sin6_addr);
424		db_printf(" l3_addr=%s\n", l3s);
425		break;
426	}
427#endif
428	default:
429		db_printf(" l3_addr=N/A (af=%d)\n", la->l3_addr.sa_family);
430		break;
431	}
432}
433
434DB_SHOW_COMMAND(llentry, db_show_llentry)
435{
436
437	if (!have_addr) {
438		db_printf("usage: show llentry <struct llentry *>\n");
439		return;
440	}
441
442	llatbl_lle_show((struct llentry_sa *)addr);
443}
444
445static void
446llatbl_llt_show(struct lltable *llt)
447{
448	int i;
449	struct llentry *lle;
450
451	db_printf("llt=%p llt_af=%d llt_ifp=%p\n",
452	    llt, llt->llt_af, llt->llt_ifp);
453
454	for (i = 0; i < LLTBL_HASHTBL_SIZE; i++) {
455		LIST_FOREACH(lle, &llt->lle_head[i], lle_next) {
456
457			llatbl_lle_show((struct llentry_sa *)lle);
458			if (db_pager_quit)
459				return;
460		}
461	}
462}
463
464DB_SHOW_COMMAND(lltable, db_show_lltable)
465{
466
467	if (!have_addr) {
468		db_printf("usage: show lltable <struct lltable *>\n");
469		return;
470	}
471
472	llatbl_llt_show((struct lltable *)addr);
473}
474
475DB_SHOW_ALL_COMMAND(lltables, db_show_all_lltables)
476{
477	VNET_ITERATOR_DECL(vnet_iter);
478	struct lltable *llt;
479
480	VNET_FOREACH(vnet_iter) {
481		CURVNET_SET_QUIET(vnet_iter);
482#ifdef VIMAGE
483		db_printf("vnet=%p\n", curvnet);
484#endif
485		SLIST_FOREACH(llt, &V_lltables, llt_link) {
486			db_printf("llt=%p llt_af=%d llt_ifp=%p(%s)\n",
487			    llt, llt->llt_af, llt->llt_ifp,
488			    (llt->llt_ifp != NULL) ?
489				llt->llt_ifp->if_xname : "?");
490			if (have_addr && addr != 0) /* verbose */
491				llatbl_llt_show(llt);
492			if (db_pager_quit) {
493				CURVNET_RESTORE();
494				return;
495			}
496		}
497		CURVNET_RESTORE();
498	}
499}
500#endif
501