2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
4 * Copyright (c) 2004 Luigi Rizzo, Alessandro Cerri. All rights reserved.
5 * Copyright (c) 2004-2008 Qing Li. All rights reserved.
6 * Copyright (c) 2008 Kip Macy. All rights reserved.
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
17 * THIS SOFTWARE IS PROVIDED BY AUTHOR 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 AUTHOR 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
29 #include <sys/cdefs.h>
30 __FBSDID("$FreeBSD$");
34 #include "opt_inet6.h"
36 #include <sys/param.h>
37 #include <sys/systm.h>
38 #include <sys/eventhandler.h>
39 #include <sys/malloc.h>
41 #include <sys/syslog.h>
42 #include <sys/sysctl.h>
43 #include <sys/socket.h>
44 #include <sys/kernel.h>
46 #include <sys/mutex.h>
47 #include <sys/rwlock.h>
55 #include <netinet/in.h>
56 #include <net/if_llatbl.h>
58 #include <net/if_dl.h>
59 #include <net/if_var.h>
60 #include <net/route.h>
62 #include <netinet/if_ether.h>
63 #include <netinet6/in6_var.h>
64 #include <netinet6/nd6.h>
66 MALLOC_DEFINE(M_LLTABLE, "lltable", "link level address tables");
68 VNET_DEFINE_STATIC(SLIST_HEAD(, lltable), lltables) =
69 SLIST_HEAD_INITIALIZER(lltables);
70 #define V_lltables VNET(lltables)
72 static struct rwlock lltable_list_lock;
73 RW_SYSINIT(lltable_list_lock, &lltable_list_lock, "lltable_list_lock");
74 #define LLTABLE_LIST_RLOCK() rw_rlock(&lltable_list_lock)
75 #define LLTABLE_LIST_RUNLOCK() rw_runlock(&lltable_list_lock)
76 #define LLTABLE_LIST_WLOCK() rw_wlock(&lltable_list_lock)
77 #define LLTABLE_LIST_WUNLOCK() rw_wunlock(&lltable_list_lock)
78 #define LLTABLE_LIST_LOCK_ASSERT() rw_assert(&lltable_list_lock, RA_LOCKED)
80 static void lltable_unlink(struct lltable *llt);
81 static void llentries_unlink(struct lltable *llt, struct llentries *head);
84 * Dump lle state for a specific address family.
87 lltable_dump_af(struct lltable *llt, struct sysctl_req *wr)
89 struct epoch_tracker et;
92 LLTABLE_LIST_LOCK_ASSERT();
94 if (llt->llt_ifp->if_flags & IFF_LOOPBACK)
99 error = lltable_foreach_lle(llt,
100 (llt_foreach_cb_t *)llt->llt_dump_entry, wr);
107 * Dump arp state for a specific address family.
110 lltable_sysctl_dumparp(int af, struct sysctl_req *wr)
115 LLTABLE_LIST_RLOCK();
116 SLIST_FOREACH(llt, &V_lltables, llt_link) {
117 if (llt->llt_af == af) {
118 error = lltable_dump_af(llt, wr);
124 LLTABLE_LIST_RUNLOCK();
129 * Common function helpers for chained hash table.
133 * Runs specified callback for each entry in @llt.
134 * Caller does the locking.
138 htable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg)
140 struct llentry *lle, *next;
145 for (i = 0; i < llt->llt_hsize; i++) {
146 CK_LIST_FOREACH_SAFE(lle, &llt->lle_head[i], lle_next, next) {
147 error = f(llt, lle, farg);
157 htable_link_entry(struct lltable *llt, struct llentry *lle)
159 struct llentries *lleh;
162 if ((lle->la_flags & LLE_LINKED) != 0)
165 IF_AFDATA_WLOCK_ASSERT(llt->llt_ifp);
167 hashidx = llt->llt_hash(lle, llt->llt_hsize);
168 lleh = &llt->lle_head[hashidx];
171 lle->lle_head = lleh;
172 lle->la_flags |= LLE_LINKED;
173 CK_LIST_INSERT_HEAD(lleh, lle, lle_next);
177 htable_unlink_entry(struct llentry *lle)
180 if ((lle->la_flags & LLE_LINKED) == 0)
183 IF_AFDATA_WLOCK_ASSERT(lle->lle_tbl->llt_ifp);
184 CK_LIST_REMOVE(lle, lle_next);
185 lle->la_flags &= ~(LLE_VALID | LLE_LINKED);
188 lle->lle_head = NULL;
192 struct prefix_match_data {
193 const struct sockaddr *addr;
194 const struct sockaddr *mask;
195 struct llentries dchain;
200 htable_prefix_free_cb(struct lltable *llt, struct llentry *lle, void *farg)
202 struct prefix_match_data *pmd;
204 pmd = (struct prefix_match_data *)farg;
206 if (llt->llt_match_prefix(pmd->addr, pmd->mask, pmd->flags, lle)) {
208 CK_LIST_INSERT_HEAD(&pmd->dchain, lle, lle_chain);
215 htable_prefix_free(struct lltable *llt, const struct sockaddr *addr,
216 const struct sockaddr *mask, u_int flags)
218 struct llentry *lle, *next;
219 struct prefix_match_data pmd;
221 bzero(&pmd, sizeof(pmd));
225 CK_LIST_INIT(&pmd.dchain);
227 IF_AFDATA_WLOCK(llt->llt_ifp);
228 /* Push matching lles to chain */
229 lltable_foreach_lle(llt, htable_prefix_free_cb, &pmd);
231 llentries_unlink(llt, &pmd.dchain);
232 IF_AFDATA_WUNLOCK(llt->llt_ifp);
234 CK_LIST_FOREACH_SAFE(lle, &pmd.dchain, lle_chain, next)
235 lltable_free_entry(llt, lle);
239 htable_free_tbl(struct lltable *llt)
242 free(llt->lle_head, M_LLTABLE);
243 free(llt, M_LLTABLE);
247 llentries_unlink(struct lltable *llt, struct llentries *head)
249 struct llentry *lle, *next;
251 CK_LIST_FOREACH_SAFE(lle, head, lle_chain, next)
252 llt->llt_unlink_entry(lle);
256 * Helper function used to drop all mbufs in hold queue.
258 * Returns the number of held packets, if any, that were dropped.
261 lltable_drop_entry_queue(struct llentry *lle)
266 LLE_WLOCK_ASSERT(lle);
269 while ((lle->la_numheld > 0) && (lle->la_hold != NULL)) {
270 next = lle->la_hold->m_nextpkt;
271 m_freem(lle->la_hold);
277 KASSERT(lle->la_numheld == 0,
278 ("%s: la_numheld %d > 0, pkts_droped %zd", __func__,
279 lle->la_numheld, pkts_dropped));
281 return (pkts_dropped);
285 lltable_set_entry_addr(struct ifnet *ifp, struct llentry *lle,
286 const char *linkhdr, size_t linkhdrsize, int lladdr_off)
289 memcpy(lle->r_linkdata, linkhdr, linkhdrsize);
290 lle->r_hdrlen = linkhdrsize;
291 lle->ll_addr = &lle->r_linkdata[lladdr_off];
292 lle->la_flags |= LLE_VALID;
293 lle->r_flags |= RLLE_VALID;
297 * Tries to update @lle link-level address.
298 * Since update requires AFDATA WLOCK, function
299 * drops @lle lock, acquires AFDATA lock and then acquires
300 * @lle lock to maintain lock order.
302 * Returns 1 on success.
305 lltable_try_set_entry_addr(struct ifnet *ifp, struct llentry *lle,
306 const char *linkhdr, size_t linkhdrsize, int lladdr_off)
309 /* Perform real LLE update */
310 /* use afdata WLOCK to update fields */
311 LLE_WLOCK_ASSERT(lle);
314 IF_AFDATA_WLOCK(ifp);
318 * Since we droppped LLE lock, other thread might have deleted
319 * this lle. Check and return
321 if ((lle->la_flags & LLE_DELETED) != 0) {
322 IF_AFDATA_WUNLOCK(ifp);
323 LLE_FREE_LOCKED(lle);
328 lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize, lladdr_off);
330 IF_AFDATA_WUNLOCK(ifp);
338 * Helper function used to pre-compute full/partial link-layer
339 * header data suitable for feeding into if_output().
342 lltable_calc_llheader(struct ifnet *ifp, int family, char *lladdr,
343 char *buf, size_t *bufsize, int *lladdr_off)
345 struct if_encap_req ereq;
348 bzero(buf, *bufsize);
349 bzero(&ereq, sizeof(ereq));
351 ereq.bufsize = *bufsize;
352 ereq.rtype = IFENCAP_LL;
353 ereq.family = family;
354 ereq.lladdr = lladdr;
355 ereq.lladdr_len = ifp->if_addrlen;
356 error = ifp->if_requestencap(ifp, &ereq);
358 *bufsize = ereq.bufsize;
359 *lladdr_off = ereq.lladdr_off;
366 * Update link-layer header for given @lle after
367 * interface lladdr was changed.
370 llentry_update_ifaddr(struct lltable *llt, struct llentry *lle, void *farg)
373 u_char linkhdr[LLE_MAX_LINKHDR];
378 ifp = (struct ifnet *)farg;
380 lladdr = lle->ll_addr;
383 if ((lle->la_flags & LLE_VALID) == 0) {
388 if ((lle->la_flags & LLE_IFADDR) != 0)
389 lladdr = IF_LLADDR(ifp);
391 linkhdrsize = sizeof(linkhdr);
392 lltable_calc_llheader(ifp, llt->llt_af, lladdr, linkhdr, &linkhdrsize,
394 memcpy(lle->r_linkdata, linkhdr, linkhdrsize);
401 * Update all calculated headers for given @llt
404 lltable_update_ifaddr(struct lltable *llt)
407 if (llt->llt_ifp->if_flags & IFF_LOOPBACK)
410 IF_AFDATA_WLOCK(llt->llt_ifp);
411 lltable_foreach_lle(llt, llentry_update_ifaddr, llt->llt_ifp);
412 IF_AFDATA_WUNLOCK(llt->llt_ifp);
417 * Performs generic cleanup routines and frees lle.
419 * Called for non-linked entries, with callouts and
420 * other AF-specific cleanups performed.
422 * @lle must be passed WLOCK'ed
424 * Returns the number of held packets, if any, that were dropped.
427 llentry_free(struct llentry *lle)
431 LLE_WLOCK_ASSERT(lle);
433 KASSERT((lle->la_flags & LLE_LINKED) == 0, ("freeing linked lle"));
435 pkts_dropped = lltable_drop_entry_queue(lle);
438 if (callout_stop(&lle->lle_timer) > 0)
440 LLE_FREE_LOCKED(lle);
442 return (pkts_dropped);
446 * Free all entries from given table and free itself.
450 lltable_free_cb(struct lltable *llt, struct llentry *lle, void *farg)
452 struct llentries *dchain;
454 dchain = (struct llentries *)farg;
457 CK_LIST_INSERT_HEAD(dchain, lle, lle_chain);
463 * Free all entries from given table and free itself.
466 lltable_free(struct lltable *llt)
468 struct llentry *lle, *next;
469 struct llentries dchain;
471 KASSERT(llt != NULL, ("%s: llt is NULL", __func__));
475 CK_LIST_INIT(&dchain);
476 IF_AFDATA_WLOCK(llt->llt_ifp);
477 /* Push all lles to @dchain */
478 lltable_foreach_lle(llt, lltable_free_cb, &dchain);
479 llentries_unlink(llt, &dchain);
480 IF_AFDATA_WUNLOCK(llt->llt_ifp);
482 CK_LIST_FOREACH_SAFE(lle, &dchain, lle_chain, next) {
486 llt->llt_free_tbl(llt);
490 * Deletes an address from given lltable.
491 * Used for userland interaction to remove
492 * individual entries. Skips entries added by OS.
495 lltable_delete_addr(struct lltable *llt, u_int flags,
496 const struct sockaddr *l3addr)
502 IF_AFDATA_WLOCK(ifp);
503 lle = lla_lookup(llt, LLE_EXCLUSIVE, l3addr);
506 IF_AFDATA_WUNLOCK(ifp);
509 if ((lle->la_flags & LLE_IFADDR) != 0 && (flags & LLE_IFADDR) == 0) {
510 IF_AFDATA_WUNLOCK(ifp);
515 lltable_unlink_entry(llt, lle);
516 IF_AFDATA_WUNLOCK(ifp);
518 llt->llt_delete_entry(llt, lle);
524 lltable_prefix_free(int af, struct sockaddr *addr, struct sockaddr *mask,
529 LLTABLE_LIST_RLOCK();
530 SLIST_FOREACH(llt, &V_lltables, llt_link) {
531 if (llt->llt_af != af)
534 llt->llt_prefix_free(llt, addr, mask, flags);
536 LLTABLE_LIST_RUNLOCK();
540 lltable_allocate_htbl(uint32_t hsize)
545 llt = malloc(sizeof(struct lltable), M_LLTABLE, M_WAITOK | M_ZERO);
546 llt->llt_hsize = hsize;
547 llt->lle_head = malloc(sizeof(struct llentries) * hsize,
548 M_LLTABLE, M_WAITOK | M_ZERO);
550 for (i = 0; i < llt->llt_hsize; i++)
551 CK_LIST_INIT(&llt->lle_head[i]);
553 /* Set some default callbacks */
554 llt->llt_link_entry = htable_link_entry;
555 llt->llt_unlink_entry = htable_unlink_entry;
556 llt->llt_prefix_free = htable_prefix_free;
557 llt->llt_foreach_entry = htable_foreach_lle;
558 llt->llt_free_tbl = htable_free_tbl;
564 * Links lltable to global llt list.
567 lltable_link(struct lltable *llt)
570 LLTABLE_LIST_WLOCK();
571 SLIST_INSERT_HEAD(&V_lltables, llt, llt_link);
572 LLTABLE_LIST_WUNLOCK();
576 lltable_unlink(struct lltable *llt)
579 LLTABLE_LIST_WLOCK();
580 SLIST_REMOVE(&V_lltables, llt, lltable, llt_link);
581 LLTABLE_LIST_WUNLOCK();
586 * External methods used by lltable consumers
590 lltable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg)
593 return (llt->llt_foreach_entry(llt, f, farg));
597 lltable_alloc_entry(struct lltable *llt, u_int flags,
598 const struct sockaddr *l3addr)
601 return (llt->llt_alloc_entry(llt, flags, l3addr));
605 lltable_free_entry(struct lltable *llt, struct llentry *lle)
608 llt->llt_free_entry(llt, lle);
612 lltable_link_entry(struct lltable *llt, struct llentry *lle)
615 llt->llt_link_entry(llt, lle);
619 lltable_unlink_entry(struct lltable *llt, struct llentry *lle)
622 llt->llt_unlink_entry(lle);
626 lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa)
631 llt->llt_fill_sa_entry(lle, sa);
635 lltable_get_ifp(const struct lltable *llt)
638 return (llt->llt_ifp);
642 lltable_get_af(const struct lltable *llt)
645 return (llt->llt_af);
649 * Called in route_output when rtm_flags contains RTF_LLDATA.
652 lla_rt_output(struct rt_msghdr *rtm, struct rt_addrinfo *info)
654 struct sockaddr_dl *dl =
655 (struct sockaddr_dl *)info->rti_info[RTAX_GATEWAY];
656 struct sockaddr *dst = (struct sockaddr *)info->rti_info[RTAX_DST];
659 struct llentry *lle, *lle_tmp;
660 uint8_t linkhdr[LLE_MAX_LINKHDR];
666 if (dl == NULL || dl->sdl_family != AF_LINK)
669 ifp = ifnet_byindex(dl->sdl_index);
671 log(LOG_INFO, "%s: invalid ifp (sdl_index %d)\n",
672 __func__, dl->sdl_index);
676 /* XXX linked list may be too expensive */
677 LLTABLE_LIST_RLOCK();
678 SLIST_FOREACH(llt, &V_lltables, llt_link) {
679 if (llt->llt_af == dst->sa_family &&
683 LLTABLE_LIST_RUNLOCK();
689 switch (rtm->rtm_type) {
693 if (rtm->rtm_rmx.rmx_expire == 0)
694 laflags = LLE_STATIC;
695 lle = lltable_alloc_entry(llt, laflags, dst);
699 linkhdrsize = sizeof(linkhdr);
700 if (lltable_calc_llheader(ifp, dst->sa_family, LLADDR(dl),
701 linkhdr, &linkhdrsize, &lladdr_off) != 0)
703 lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize,
705 if ((rtm->rtm_flags & RTF_ANNOUNCE))
706 lle->la_flags |= LLE_PUB;
707 lle->la_expire = rtm->rtm_rmx.rmx_expire;
709 laflags = lle->la_flags;
711 /* Try to link new entry */
713 IF_AFDATA_WLOCK(ifp);
715 lle_tmp = lla_lookup(llt, LLE_EXCLUSIVE, dst);
716 if (lle_tmp != NULL) {
717 /* Check if we are trying to replace immutable entry */
718 if ((lle_tmp->la_flags & LLE_IFADDR) != 0) {
719 IF_AFDATA_WUNLOCK(ifp);
720 LLE_WUNLOCK(lle_tmp);
721 lltable_free_entry(llt, lle);
724 /* Unlink existing entry from table */
725 lltable_unlink_entry(llt, lle_tmp);
727 lltable_link_entry(llt, lle);
728 IF_AFDATA_WUNLOCK(ifp);
730 if (lle_tmp != NULL) {
731 EVENTHANDLER_INVOKE(lle_event, lle_tmp,LLENTRY_EXPIRED);
732 lltable_free_entry(llt, lle_tmp);
736 * By invoking LLE handler here we might get
737 * two events on static LLE entry insertion
738 * in routing socket. However, since we might have
739 * other subscribers we need to generate this event.
741 EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_RESOLVED);
745 if ((laflags & LLE_PUB) && dst->sa_family == AF_INET)
747 &((struct sockaddr_in *)dst)->sin_addr,
748 &((struct sockaddr_in *)dst)->sin_addr,
749 (u_char *)LLADDR(dl));
755 return (lltable_delete_addr(llt, 0, dst));
767 struct sockaddr l3_addr;
771 llatbl_lle_show(struct llentry_sa *la)
777 db_printf("lle=%p\n", lle);
778 db_printf(" lle_next=%p\n", lle->lle_next.cle_next);
779 db_printf(" lle_lock=%p\n", &lle->lle_lock);
780 db_printf(" lle_tbl=%p\n", lle->lle_tbl);
781 db_printf(" lle_head=%p\n", lle->lle_head);
782 db_printf(" la_hold=%p\n", lle->la_hold);
783 db_printf(" la_numheld=%d\n", lle->la_numheld);
784 db_printf(" la_expire=%ju\n", (uintmax_t)lle->la_expire);
785 db_printf(" la_flags=0x%04x\n", lle->la_flags);
786 db_printf(" la_asked=%u\n", lle->la_asked);
787 db_printf(" la_preempt=%u\n", lle->la_preempt);
788 db_printf(" ln_state=%d\n", lle->ln_state);
789 db_printf(" ln_router=%u\n", lle->ln_router);
790 db_printf(" ln_ntick=%ju\n", (uintmax_t)lle->ln_ntick);
791 db_printf(" lle_refcnt=%d\n", lle->lle_refcnt);
792 bcopy(lle->ll_addr, octet, sizeof(octet));
793 db_printf(" ll_addr=%02x:%02x:%02x:%02x:%02x:%02x\n",
794 octet[0], octet[1], octet[2], octet[3], octet[4], octet[5]);
795 db_printf(" lle_timer=%p\n", &lle->lle_timer);
797 switch (la->l3_addr.sa_family) {
801 struct sockaddr_in *sin;
802 char l3s[INET_ADDRSTRLEN];
804 sin = (struct sockaddr_in *)&la->l3_addr;
805 inet_ntoa_r(sin->sin_addr, l3s);
806 db_printf(" l3_addr=%s\n", l3s);
813 struct sockaddr_in6 *sin6;
814 char l3s[INET6_ADDRSTRLEN];
816 sin6 = (struct sockaddr_in6 *)&la->l3_addr;
817 ip6_sprintf(l3s, &sin6->sin6_addr);
818 db_printf(" l3_addr=%s\n", l3s);
823 db_printf(" l3_addr=N/A (af=%d)\n", la->l3_addr.sa_family);
828 DB_SHOW_COMMAND(llentry, db_show_llentry)
832 db_printf("usage: show llentry <struct llentry *>\n");
836 llatbl_lle_show((struct llentry_sa *)addr);
840 llatbl_llt_show(struct lltable *llt)
845 db_printf("llt=%p llt_af=%d llt_ifp=%p\n",
846 llt, llt->llt_af, llt->llt_ifp);
848 for (i = 0; i < llt->llt_hsize; i++) {
849 CK_LIST_FOREACH(lle, &llt->lle_head[i], lle_next) {
851 llatbl_lle_show((struct llentry_sa *)lle);
858 DB_SHOW_COMMAND(lltable, db_show_lltable)
862 db_printf("usage: show lltable <struct lltable *>\n");
866 llatbl_llt_show((struct lltable *)addr);
869 DB_SHOW_ALL_COMMAND(lltables, db_show_all_lltables)
871 VNET_ITERATOR_DECL(vnet_iter);
874 VNET_FOREACH(vnet_iter) {
875 CURVNET_SET_QUIET(vnet_iter);
877 db_printf("vnet=%p\n", curvnet);
879 SLIST_FOREACH(llt, &V_lltables, llt_link) {
880 db_printf("llt=%p llt_af=%d llt_ifp=%p(%s)\n",
881 llt, llt->llt_af, llt->llt_ifp,
882 (llt->llt_ifp != NULL) ?
883 llt->llt_ifp->if_xname : "?");
884 if (have_addr && addr != 0) /* verbose */
885 llatbl_llt_show(llt);