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.
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
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.
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
27 #include <sys/cdefs.h>
28 __FBSDID("$FreeBSD$");
32 #include "opt_inet6.h"
34 #include <sys/param.h>
35 #include <sys/systm.h>
36 #include <sys/malloc.h>
38 #include <sys/syslog.h>
39 #include <sys/sysctl.h>
40 #include <sys/socket.h>
41 #include <sys/kernel.h>
43 #include <sys/mutex.h>
44 #include <sys/rwlock.h>
52 #include <netinet/in.h>
53 #include <net/if_llatbl.h>
55 #include <net/if_dl.h>
56 #include <net/if_var.h>
57 #include <net/route.h>
59 #include <netinet/if_ether.h>
60 #include <netinet6/in6_var.h>
61 #include <netinet6/nd6.h>
63 MALLOC_DEFINE(M_LLTABLE, "lltable", "link level address tables");
65 static VNET_DEFINE(SLIST_HEAD(, lltable), lltables);
66 #define V_lltables VNET(lltables)
68 static void vnet_lltable_init(void);
70 struct rwlock lltable_rwlock;
71 RW_SYSINIT(lltable_rwlock, &lltable_rwlock, "lltable_rwlock");
73 static void lltable_unlink(struct lltable *llt);
74 static void llentries_unlink(struct lltable *llt, struct llentries *head);
76 static void htable_unlink_entry(struct llentry *lle);
77 static void htable_link_entry(struct lltable *llt, struct llentry *lle);
78 static int htable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f,
82 * Dump lle state for a specific address family.
85 lltable_dump_af(struct lltable *llt, struct sysctl_req *wr)
89 LLTABLE_LOCK_ASSERT();
91 if (llt->llt_ifp->if_flags & IFF_LOOPBACK)
95 IF_AFDATA_RLOCK(llt->llt_ifp);
96 error = lltable_foreach_lle(llt,
97 (llt_foreach_cb_t *)llt->llt_dump_entry, wr);
98 IF_AFDATA_RUNLOCK(llt->llt_ifp);
104 * Dump arp state for a specific address family.
107 lltable_sysctl_dumparp(int af, struct sysctl_req *wr)
113 SLIST_FOREACH(llt, &V_lltables, llt_link) {
114 if (llt->llt_af == af) {
115 error = lltable_dump_af(llt, wr);
126 * Common function helpers for chained hash table.
130 * Runs specified callback for each entry in @llt.
131 * Caller does the locking.
135 htable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg)
137 struct llentry *lle, *next;
142 for (i = 0; i < llt->llt_hsize; i++) {
143 LIST_FOREACH_SAFE(lle, &llt->lle_head[i], lle_next, next) {
144 error = f(llt, lle, farg);
154 htable_link_entry(struct lltable *llt, struct llentry *lle)
156 struct llentries *lleh;
159 if ((lle->la_flags & LLE_LINKED) != 0)
162 IF_AFDATA_WLOCK_ASSERT(llt->llt_ifp);
164 hashidx = llt->llt_hash(lle, llt->llt_hsize);
165 lleh = &llt->lle_head[hashidx];
168 lle->lle_head = lleh;
169 lle->la_flags |= LLE_LINKED;
170 LIST_INSERT_HEAD(lleh, lle, lle_next);
174 htable_unlink_entry(struct llentry *lle)
177 if ((lle->la_flags & LLE_LINKED) != 0) {
178 IF_AFDATA_WLOCK_ASSERT(lle->lle_tbl->llt_ifp);
179 LIST_REMOVE(lle, lle_next);
180 lle->la_flags &= ~(LLE_VALID | LLE_LINKED);
183 lle->lle_head = NULL;
188 struct prefix_match_data {
189 const struct sockaddr *addr;
190 const struct sockaddr *mask;
191 struct llentries dchain;
196 htable_prefix_free_cb(struct lltable *llt, struct llentry *lle, void *farg)
198 struct prefix_match_data *pmd;
200 pmd = (struct prefix_match_data *)farg;
202 if (llt->llt_match_prefix(pmd->addr, pmd->mask, pmd->flags, lle)) {
204 LIST_INSERT_HEAD(&pmd->dchain, lle, lle_chain);
211 htable_prefix_free(struct lltable *llt, const struct sockaddr *addr,
212 const struct sockaddr *mask, u_int flags)
214 struct llentry *lle, *next;
215 struct prefix_match_data pmd;
217 bzero(&pmd, sizeof(pmd));
221 LIST_INIT(&pmd.dchain);
223 IF_AFDATA_WLOCK(llt->llt_ifp);
224 /* Push matching lles to chain */
225 lltable_foreach_lle(llt, htable_prefix_free_cb, &pmd);
227 llentries_unlink(llt, &pmd.dchain);
228 IF_AFDATA_WUNLOCK(llt->llt_ifp);
230 LIST_FOREACH_SAFE(lle, &pmd.dchain, lle_chain, next)
231 lltable_free_entry(llt, lle);
235 htable_free_tbl(struct lltable *llt)
238 free(llt->lle_head, M_LLTABLE);
239 free(llt, M_LLTABLE);
243 llentries_unlink(struct lltable *llt, struct llentries *head)
245 struct llentry *lle, *next;
247 LIST_FOREACH_SAFE(lle, head, lle_chain, next)
248 llt->llt_unlink_entry(lle);
252 * Helper function used to drop all mbufs in hold queue.
254 * Returns the number of held packets, if any, that were dropped.
257 lltable_drop_entry_queue(struct llentry *lle)
262 LLE_WLOCK_ASSERT(lle);
265 while ((lle->la_numheld > 0) && (lle->la_hold != NULL)) {
266 next = lle->la_hold->m_nextpkt;
267 m_freem(lle->la_hold);
273 KASSERT(lle->la_numheld == 0,
274 ("%s: la_numheld %d > 0, pkts_droped %zd", __func__,
275 lle->la_numheld, pkts_dropped));
277 return (pkts_dropped);
282 * Performes generic cleanup routines and frees lle.
284 * Called for non-linked entries, with callouts and
285 * other AF-specific cleanups performed.
287 * @lle must be passed WLOCK'ed
289 * Returns the number of held packets, if any, that were dropped.
292 llentry_free(struct llentry *lle)
296 LLE_WLOCK_ASSERT(lle);
298 KASSERT((lle->la_flags & LLE_LINKED) == 0, ("freeing linked lle"));
300 pkts_dropped = lltable_drop_entry_queue(lle);
302 LLE_FREE_LOCKED(lle);
304 return (pkts_dropped);
308 * (al)locate an llentry for address dst (equivalent to rtalloc for new-arp).
310 * If found the llentry * is returned referenced and unlocked.
313 llentry_alloc(struct ifnet *ifp, struct lltable *lt,
314 struct sockaddr_storage *dst)
316 struct llentry *la, *la_tmp;
318 IF_AFDATA_RLOCK(ifp);
319 la = lla_lookup(lt, LLE_EXCLUSIVE, (struct sockaddr *)dst);
320 IF_AFDATA_RUNLOCK(ifp);
328 if ((ifp->if_flags & (IFF_NOARP | IFF_STATICARP)) == 0) {
329 la = lltable_alloc_entry(lt, 0, (struct sockaddr *)dst);
332 IF_AFDATA_WLOCK(ifp);
334 /* Prefer any existing LLE over newly-created one */
335 la_tmp = lla_lookup(lt, LLE_EXCLUSIVE, (struct sockaddr *)dst);
337 lltable_link_entry(lt, la);
338 IF_AFDATA_WUNLOCK(ifp);
339 if (la_tmp != NULL) {
340 lltable_free_entry(lt, la);
351 * Free all entries from given table and free itself.
355 lltable_free_cb(struct lltable *llt, struct llentry *lle, void *farg)
357 struct llentries *dchain;
359 dchain = (struct llentries *)farg;
362 LIST_INSERT_HEAD(dchain, lle, lle_chain);
368 * Free all entries from given table and free itself.
371 lltable_free(struct lltable *llt)
373 struct llentry *lle, *next;
374 struct llentries dchain;
376 KASSERT(llt != NULL, ("%s: llt is NULL", __func__));
381 IF_AFDATA_WLOCK(llt->llt_ifp);
382 /* Push all lles to @dchain */
383 lltable_foreach_lle(llt, lltable_free_cb, &dchain);
384 llentries_unlink(llt, &dchain);
385 IF_AFDATA_WUNLOCK(llt->llt_ifp);
387 LIST_FOREACH_SAFE(lle, &dchain, lle_chain, next) {
388 if (callout_stop(&lle->lle_timer))
393 llt->llt_free_tbl(llt);
398 lltable_drain(int af)
405 SLIST_FOREACH(llt, &V_lltables, llt_link) {
406 if (llt->llt_af != af)
409 for (i=0; i < llt->llt_hsize; i++) {
410 LIST_FOREACH(lle, &llt->lle_head[i], lle_next) {
413 m_freem(lle->la_hold);
425 * Deletes an address from given lltable.
426 * Used for userland interaction to remove
427 * individual entries. Skips entries added by OS.
430 lltable_delete_addr(struct lltable *llt, u_int flags,
431 const struct sockaddr *l3addr)
437 IF_AFDATA_WLOCK(ifp);
438 lle = lla_lookup(llt, LLE_EXCLUSIVE, l3addr);
441 IF_AFDATA_WUNLOCK(ifp);
444 if ((lle->la_flags & LLE_IFADDR) != 0 && (flags & LLE_IFADDR) == 0) {
445 IF_AFDATA_WUNLOCK(ifp);
450 lltable_unlink_entry(llt, lle);
451 IF_AFDATA_WUNLOCK(ifp);
453 llt->llt_delete_entry(llt, lle);
459 lltable_prefix_free(int af, struct sockaddr *addr, struct sockaddr *mask,
465 SLIST_FOREACH(llt, &V_lltables, llt_link) {
466 if (llt->llt_af != af)
469 llt->llt_prefix_free(llt, addr, mask, flags);
475 lltable_allocate_htbl(uint32_t hsize)
480 llt = malloc(sizeof(struct lltable), M_LLTABLE, M_WAITOK | M_ZERO);
481 llt->llt_hsize = hsize;
482 llt->lle_head = malloc(sizeof(struct llentries) * hsize,
483 M_LLTABLE, M_WAITOK | M_ZERO);
485 for (i = 0; i < llt->llt_hsize; i++)
486 LIST_INIT(&llt->lle_head[i]);
488 /* Set some default callbacks */
489 llt->llt_link_entry = htable_link_entry;
490 llt->llt_unlink_entry = htable_unlink_entry;
491 llt->llt_prefix_free = htable_prefix_free;
492 llt->llt_foreach_entry = htable_foreach_lle;
493 llt->llt_free_tbl = htable_free_tbl;
499 * Links lltable to global llt list.
502 lltable_link(struct lltable *llt)
506 SLIST_INSERT_HEAD(&V_lltables, llt, llt_link);
511 lltable_unlink(struct lltable *llt)
515 SLIST_REMOVE(&V_lltables, llt, lltable, llt_link);
521 * External methods used by lltable consumers
525 lltable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg)
528 return (llt->llt_foreach_entry(llt, f, farg));
532 lltable_alloc_entry(struct lltable *llt, u_int flags,
533 const struct sockaddr *l3addr)
536 return (llt->llt_alloc_entry(llt, flags, l3addr));
540 lltable_free_entry(struct lltable *llt, struct llentry *lle)
543 llt->llt_free_entry(llt, lle);
547 lltable_link_entry(struct lltable *llt, struct llentry *lle)
550 llt->llt_link_entry(llt, lle);
554 lltable_unlink_entry(struct lltable *llt, struct llentry *lle)
557 llt->llt_unlink_entry(lle);
561 lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa)
566 llt->llt_fill_sa_entry(lle, sa);
570 lltable_get_ifp(const struct lltable *llt)
573 return (llt->llt_ifp);
577 lltable_get_af(const struct lltable *llt)
580 return (llt->llt_af);
584 * Called in route_output when rtm_flags contains RTF_LLDATA.
587 lla_rt_output(struct rt_msghdr *rtm, struct rt_addrinfo *info)
589 struct sockaddr_dl *dl =
590 (struct sockaddr_dl *)info->rti_info[RTAX_GATEWAY];
591 struct sockaddr *dst = (struct sockaddr *)info->rti_info[RTAX_DST];
594 struct llentry *lle, *lle_tmp;
598 KASSERT(dl != NULL && dl->sdl_family == AF_LINK,
599 ("%s: invalid dl\n", __func__));
601 ifp = ifnet_byindex(dl->sdl_index);
603 log(LOG_INFO, "%s: invalid ifp (sdl_index %d)\n",
604 __func__, dl->sdl_index);
608 /* XXX linked list may be too expensive */
610 SLIST_FOREACH(llt, &V_lltables, llt_link) {
611 if (llt->llt_af == dst->sa_family &&
616 KASSERT(llt != NULL, ("Yep, ugly hacks are bad\n"));
620 switch (rtm->rtm_type) {
624 if (rtm->rtm_rmx.rmx_expire == 0)
625 laflags = LLE_STATIC;
626 lle = lltable_alloc_entry(llt, laflags, dst);
630 bcopy(LLADDR(dl), &lle->ll_addr, ifp->if_addrlen);
631 if ((rtm->rtm_flags & RTF_ANNOUNCE))
632 lle->la_flags |= LLE_PUB;
633 lle->la_flags |= LLE_VALID;
634 lle->la_expire = rtm->rtm_rmx.rmx_expire;
636 laflags = lle->la_flags;
638 /* Try to link new entry */
640 IF_AFDATA_WLOCK(ifp);
642 lle_tmp = lla_lookup(llt, LLE_EXCLUSIVE, dst);
643 if (lle_tmp != NULL) {
644 /* Check if we are trying to replace immutable entry */
645 if ((lle_tmp->la_flags & LLE_IFADDR) != 0) {
646 IF_AFDATA_WUNLOCK(ifp);
647 LLE_WUNLOCK(lle_tmp);
648 lltable_free_entry(llt, lle);
651 /* Unlink existing entry from table */
652 lltable_unlink_entry(llt, lle_tmp);
654 lltable_link_entry(llt, lle);
655 IF_AFDATA_WUNLOCK(ifp);
657 if (lle_tmp != NULL) {
658 EVENTHANDLER_INVOKE(lle_event, lle_tmp,LLENTRY_EXPIRED);
659 lltable_free_entry(llt, lle_tmp);
663 * By invoking LLE handler here we might get
664 * two events on static LLE entry insertion
665 * in routing socket. However, since we might have
666 * other subscribers we need to generate this event.
668 EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_RESOLVED);
672 if ((laflags & LLE_PUB) && dst->sa_family == AF_INET)
674 &((struct sockaddr_in *)dst)->sin_addr,
675 &((struct sockaddr_in *)dst)->sin_addr,
676 (u_char *)LLADDR(dl));
682 return (lltable_delete_addr(llt, 0, dst));
695 SLIST_INIT(&V_lltables);
697 VNET_SYSINIT(vnet_lltable_init, SI_SUB_PSEUDO, SI_ORDER_FIRST,
698 vnet_lltable_init, NULL);
703 struct sockaddr l3_addr;
707 llatbl_lle_show(struct llentry_sa *la)
713 db_printf("lle=%p\n", lle);
714 db_printf(" lle_next=%p\n", lle->lle_next.le_next);
715 db_printf(" lle_lock=%p\n", &lle->lle_lock);
716 db_printf(" lle_tbl=%p\n", lle->lle_tbl);
717 db_printf(" lle_head=%p\n", lle->lle_head);
718 db_printf(" la_hold=%p\n", lle->la_hold);
719 db_printf(" la_numheld=%d\n", lle->la_numheld);
720 db_printf(" la_expire=%ju\n", (uintmax_t)lle->la_expire);
721 db_printf(" la_flags=0x%04x\n", lle->la_flags);
722 db_printf(" la_asked=%u\n", lle->la_asked);
723 db_printf(" la_preempt=%u\n", lle->la_preempt);
724 db_printf(" ln_byhint=%u\n", lle->ln_byhint);
725 db_printf(" ln_state=%d\n", lle->ln_state);
726 db_printf(" ln_router=%u\n", lle->ln_router);
727 db_printf(" ln_ntick=%ju\n", (uintmax_t)lle->ln_ntick);
728 db_printf(" lle_refcnt=%d\n", lle->lle_refcnt);
729 bcopy(&lle->ll_addr.mac16, octet, sizeof(octet));
730 db_printf(" ll_addr=%02x:%02x:%02x:%02x:%02x:%02x\n",
731 octet[0], octet[1], octet[2], octet[3], octet[4], octet[5]);
732 db_printf(" lle_timer=%p\n", &lle->lle_timer);
734 switch (la->l3_addr.sa_family) {
738 struct sockaddr_in *sin;
739 char l3s[INET_ADDRSTRLEN];
741 sin = (struct sockaddr_in *)&la->l3_addr;
742 inet_ntoa_r(sin->sin_addr, l3s);
743 db_printf(" l3_addr=%s\n", l3s);
750 struct sockaddr_in6 *sin6;
751 char l3s[INET6_ADDRSTRLEN];
753 sin6 = (struct sockaddr_in6 *)&la->l3_addr;
754 ip6_sprintf(l3s, &sin6->sin6_addr);
755 db_printf(" l3_addr=%s\n", l3s);
760 db_printf(" l3_addr=N/A (af=%d)\n", la->l3_addr.sa_family);
765 DB_SHOW_COMMAND(llentry, db_show_llentry)
769 db_printf("usage: show llentry <struct llentry *>\n");
773 llatbl_lle_show((struct llentry_sa *)addr);
777 llatbl_llt_show(struct lltable *llt)
782 db_printf("llt=%p llt_af=%d llt_ifp=%p\n",
783 llt, llt->llt_af, llt->llt_ifp);
785 for (i = 0; i < llt->llt_hsize; i++) {
786 LIST_FOREACH(lle, &llt->lle_head[i], lle_next) {
788 llatbl_lle_show((struct llentry_sa *)lle);
795 DB_SHOW_COMMAND(lltable, db_show_lltable)
799 db_printf("usage: show lltable <struct lltable *>\n");
803 llatbl_llt_show((struct lltable *)addr);
806 DB_SHOW_ALL_COMMAND(lltables, db_show_all_lltables)
808 VNET_ITERATOR_DECL(vnet_iter);
811 VNET_FOREACH(vnet_iter) {
812 CURVNET_SET_QUIET(vnet_iter);
814 db_printf("vnet=%p\n", curvnet);
816 SLIST_FOREACH(llt, &V_lltables, llt_link) {
817 db_printf("llt=%p llt_af=%d llt_ifp=%p(%s)\n",
818 llt, llt->llt_af, llt->llt_ifp,
819 (llt->llt_ifp != NULL) ?
820 llt->llt_ifp->if_xname : "?");
821 if (have_addr && addr != 0) /* verbose */
822 llatbl_llt_show(llt);