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);
83 static void htable_unlink_entry(struct llentry *lle);
84 static void htable_link_entry(struct lltable *llt, struct llentry *lle);
85 static int htable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f,
89 * Dump lle state for a specific address family.
92 lltable_dump_af(struct lltable *llt, struct sysctl_req *wr)
94 struct epoch_tracker et;
97 LLTABLE_LIST_LOCK_ASSERT();
99 if (llt->llt_ifp->if_flags & IFF_LOOPBACK)
104 error = lltable_foreach_lle(llt,
105 (llt_foreach_cb_t *)llt->llt_dump_entry, wr);
112 * Dump arp state for a specific address family.
115 lltable_sysctl_dumparp(int af, struct sysctl_req *wr)
120 LLTABLE_LIST_RLOCK();
121 SLIST_FOREACH(llt, &V_lltables, llt_link) {
122 if (llt->llt_af == af) {
123 error = lltable_dump_af(llt, wr);
129 LLTABLE_LIST_RUNLOCK();
134 * Common function helpers for chained hash table.
138 * Runs specified callback for each entry in @llt.
139 * Caller does the locking.
143 htable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg)
145 struct llentry *lle, *next;
150 for (i = 0; i < llt->llt_hsize; i++) {
151 CK_LIST_FOREACH_SAFE(lle, &llt->lle_head[i], lle_next, next) {
152 error = f(llt, lle, farg);
162 htable_link_entry(struct lltable *llt, struct llentry *lle)
164 struct llentries *lleh;
167 if ((lle->la_flags & LLE_LINKED) != 0)
170 IF_AFDATA_WLOCK_ASSERT(llt->llt_ifp);
172 hashidx = llt->llt_hash(lle, llt->llt_hsize);
173 lleh = &llt->lle_head[hashidx];
176 lle->lle_head = lleh;
177 lle->la_flags |= LLE_LINKED;
178 CK_LIST_INSERT_HEAD(lleh, lle, lle_next);
182 htable_unlink_entry(struct llentry *lle)
185 if ((lle->la_flags & LLE_LINKED) != 0) {
186 IF_AFDATA_WLOCK_ASSERT(lle->lle_tbl->llt_ifp);
187 CK_LIST_REMOVE(lle, lle_next);
188 lle->la_flags &= ~(LLE_VALID | LLE_LINKED);
191 lle->lle_head = NULL;
196 struct prefix_match_data {
197 const struct sockaddr *addr;
198 const struct sockaddr *mask;
199 struct llentries dchain;
204 htable_prefix_free_cb(struct lltable *llt, struct llentry *lle, void *farg)
206 struct prefix_match_data *pmd;
208 pmd = (struct prefix_match_data *)farg;
210 if (llt->llt_match_prefix(pmd->addr, pmd->mask, pmd->flags, lle)) {
212 CK_LIST_INSERT_HEAD(&pmd->dchain, lle, lle_chain);
219 htable_prefix_free(struct lltable *llt, const struct sockaddr *addr,
220 const struct sockaddr *mask, u_int flags)
222 struct llentry *lle, *next;
223 struct prefix_match_data pmd;
225 bzero(&pmd, sizeof(pmd));
229 CK_LIST_INIT(&pmd.dchain);
231 IF_AFDATA_WLOCK(llt->llt_ifp);
232 /* Push matching lles to chain */
233 lltable_foreach_lle(llt, htable_prefix_free_cb, &pmd);
235 llentries_unlink(llt, &pmd.dchain);
236 IF_AFDATA_WUNLOCK(llt->llt_ifp);
238 CK_LIST_FOREACH_SAFE(lle, &pmd.dchain, lle_chain, next)
239 lltable_free_entry(llt, lle);
243 htable_free_tbl(struct lltable *llt)
246 free(llt->lle_head, M_LLTABLE);
247 free(llt, M_LLTABLE);
251 llentries_unlink(struct lltable *llt, struct llentries *head)
253 struct llentry *lle, *next;
255 CK_LIST_FOREACH_SAFE(lle, head, lle_chain, next)
256 llt->llt_unlink_entry(lle);
260 * Helper function used to drop all mbufs in hold queue.
262 * Returns the number of held packets, if any, that were dropped.
265 lltable_drop_entry_queue(struct llentry *lle)
270 LLE_WLOCK_ASSERT(lle);
273 while ((lle->la_numheld > 0) && (lle->la_hold != NULL)) {
274 next = lle->la_hold->m_nextpkt;
275 m_freem(lle->la_hold);
281 KASSERT(lle->la_numheld == 0,
282 ("%s: la_numheld %d > 0, pkts_droped %zd", __func__,
283 lle->la_numheld, pkts_dropped));
285 return (pkts_dropped);
289 lltable_set_entry_addr(struct ifnet *ifp, struct llentry *lle,
290 const char *linkhdr, size_t linkhdrsize, int lladdr_off)
293 memcpy(lle->r_linkdata, linkhdr, linkhdrsize);
294 lle->r_hdrlen = linkhdrsize;
295 lle->ll_addr = &lle->r_linkdata[lladdr_off];
296 lle->la_flags |= LLE_VALID;
297 lle->r_flags |= RLLE_VALID;
301 * Tries to update @lle link-level address.
302 * Since update requires AFDATA WLOCK, function
303 * drops @lle lock, acquires AFDATA lock and then acquires
304 * @lle lock to maintain lock order.
306 * Returns 1 on success.
309 lltable_try_set_entry_addr(struct ifnet *ifp, struct llentry *lle,
310 const char *linkhdr, size_t linkhdrsize, int lladdr_off)
313 /* Perform real LLE update */
314 /* use afdata WLOCK to update fields */
315 LLE_WLOCK_ASSERT(lle);
318 IF_AFDATA_WLOCK(ifp);
322 * Since we droppped LLE lock, other thread might have deleted
323 * this lle. Check and return
325 if ((lle->la_flags & LLE_DELETED) != 0) {
326 IF_AFDATA_WUNLOCK(ifp);
327 LLE_FREE_LOCKED(lle);
332 lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize, lladdr_off);
334 IF_AFDATA_WUNLOCK(ifp);
342 * Helper function used to pre-compute full/partial link-layer
343 * header data suitable for feeding into if_output().
346 lltable_calc_llheader(struct ifnet *ifp, int family, char *lladdr,
347 char *buf, size_t *bufsize, int *lladdr_off)
349 struct if_encap_req ereq;
352 bzero(buf, *bufsize);
353 bzero(&ereq, sizeof(ereq));
355 ereq.bufsize = *bufsize;
356 ereq.rtype = IFENCAP_LL;
357 ereq.family = family;
358 ereq.lladdr = lladdr;
359 ereq.lladdr_len = ifp->if_addrlen;
360 error = ifp->if_requestencap(ifp, &ereq);
362 *bufsize = ereq.bufsize;
363 *lladdr_off = ereq.lladdr_off;
370 * Update link-layer header for given @lle after
371 * interface lladdr was changed.
374 llentry_update_ifaddr(struct lltable *llt, struct llentry *lle, void *farg)
377 u_char linkhdr[LLE_MAX_LINKHDR];
382 ifp = (struct ifnet *)farg;
384 lladdr = lle->ll_addr;
387 if ((lle->la_flags & LLE_VALID) == 0) {
392 if ((lle->la_flags & LLE_IFADDR) != 0)
393 lladdr = IF_LLADDR(ifp);
395 linkhdrsize = sizeof(linkhdr);
396 lltable_calc_llheader(ifp, llt->llt_af, lladdr, linkhdr, &linkhdrsize,
398 memcpy(lle->r_linkdata, linkhdr, linkhdrsize);
405 * Update all calculated headers for given @llt
408 lltable_update_ifaddr(struct lltable *llt)
411 if (llt->llt_ifp->if_flags & IFF_LOOPBACK)
414 IF_AFDATA_WLOCK(llt->llt_ifp);
415 lltable_foreach_lle(llt, llentry_update_ifaddr, llt->llt_ifp);
416 IF_AFDATA_WUNLOCK(llt->llt_ifp);
421 * Performs generic cleanup routines and frees lle.
423 * Called for non-linked entries, with callouts and
424 * other AF-specific cleanups performed.
426 * @lle must be passed WLOCK'ed
428 * Returns the number of held packets, if any, that were dropped.
431 llentry_free(struct llentry *lle)
435 LLE_WLOCK_ASSERT(lle);
437 KASSERT((lle->la_flags & LLE_LINKED) == 0, ("freeing linked lle"));
439 pkts_dropped = lltable_drop_entry_queue(lle);
442 if (callout_stop(&lle->lle_timer) > 0)
444 LLE_FREE_LOCKED(lle);
446 return (pkts_dropped);
450 * Free all entries from given table and free itself.
454 lltable_free_cb(struct lltable *llt, struct llentry *lle, void *farg)
456 struct llentries *dchain;
458 dchain = (struct llentries *)farg;
461 CK_LIST_INSERT_HEAD(dchain, lle, lle_chain);
467 * Free all entries from given table and free itself.
470 lltable_free(struct lltable *llt)
472 struct llentry *lle, *next;
473 struct llentries dchain;
475 KASSERT(llt != NULL, ("%s: llt is NULL", __func__));
479 CK_LIST_INIT(&dchain);
480 IF_AFDATA_WLOCK(llt->llt_ifp);
481 /* Push all lles to @dchain */
482 lltable_foreach_lle(llt, lltable_free_cb, &dchain);
483 llentries_unlink(llt, &dchain);
484 IF_AFDATA_WUNLOCK(llt->llt_ifp);
486 CK_LIST_FOREACH_SAFE(lle, &dchain, lle_chain, next) {
490 llt->llt_free_tbl(llt);
494 * Deletes an address from given lltable.
495 * Used for userland interaction to remove
496 * individual entries. Skips entries added by OS.
499 lltable_delete_addr(struct lltable *llt, u_int flags,
500 const struct sockaddr *l3addr)
506 IF_AFDATA_WLOCK(ifp);
507 lle = lla_lookup(llt, LLE_EXCLUSIVE, l3addr);
510 IF_AFDATA_WUNLOCK(ifp);
513 if ((lle->la_flags & LLE_IFADDR) != 0 && (flags & LLE_IFADDR) == 0) {
514 IF_AFDATA_WUNLOCK(ifp);
519 lltable_unlink_entry(llt, lle);
520 IF_AFDATA_WUNLOCK(ifp);
522 llt->llt_delete_entry(llt, lle);
528 lltable_prefix_free(int af, struct sockaddr *addr, struct sockaddr *mask,
533 LLTABLE_LIST_RLOCK();
534 SLIST_FOREACH(llt, &V_lltables, llt_link) {
535 if (llt->llt_af != af)
538 llt->llt_prefix_free(llt, addr, mask, flags);
540 LLTABLE_LIST_RUNLOCK();
544 lltable_allocate_htbl(uint32_t hsize)
549 llt = malloc(sizeof(struct lltable), M_LLTABLE, M_WAITOK | M_ZERO);
550 llt->llt_hsize = hsize;
551 llt->lle_head = malloc(sizeof(struct llentries) * hsize,
552 M_LLTABLE, M_WAITOK | M_ZERO);
554 for (i = 0; i < llt->llt_hsize; i++)
555 CK_LIST_INIT(&llt->lle_head[i]);
557 /* Set some default callbacks */
558 llt->llt_link_entry = htable_link_entry;
559 llt->llt_unlink_entry = htable_unlink_entry;
560 llt->llt_prefix_free = htable_prefix_free;
561 llt->llt_foreach_entry = htable_foreach_lle;
562 llt->llt_free_tbl = htable_free_tbl;
568 * Links lltable to global llt list.
571 lltable_link(struct lltable *llt)
574 LLTABLE_LIST_WLOCK();
575 SLIST_INSERT_HEAD(&V_lltables, llt, llt_link);
576 LLTABLE_LIST_WUNLOCK();
580 lltable_unlink(struct lltable *llt)
583 LLTABLE_LIST_WLOCK();
584 SLIST_REMOVE(&V_lltables, llt, lltable, llt_link);
585 LLTABLE_LIST_WUNLOCK();
590 * External methods used by lltable consumers
594 lltable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg)
597 return (llt->llt_foreach_entry(llt, f, farg));
601 lltable_alloc_entry(struct lltable *llt, u_int flags,
602 const struct sockaddr *l3addr)
605 return (llt->llt_alloc_entry(llt, flags, l3addr));
609 lltable_free_entry(struct lltable *llt, struct llentry *lle)
612 llt->llt_free_entry(llt, lle);
616 lltable_link_entry(struct lltable *llt, struct llentry *lle)
619 llt->llt_link_entry(llt, lle);
623 lltable_unlink_entry(struct lltable *llt, struct llentry *lle)
626 llt->llt_unlink_entry(lle);
630 lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa)
635 llt->llt_fill_sa_entry(lle, sa);
639 lltable_get_ifp(const struct lltable *llt)
642 return (llt->llt_ifp);
646 lltable_get_af(const struct lltable *llt)
649 return (llt->llt_af);
653 * Called in route_output when rtm_flags contains RTF_LLDATA.
656 lla_rt_output(struct rt_msghdr *rtm, struct rt_addrinfo *info)
658 struct sockaddr_dl *dl =
659 (struct sockaddr_dl *)info->rti_info[RTAX_GATEWAY];
660 struct sockaddr *dst = (struct sockaddr *)info->rti_info[RTAX_DST];
663 struct llentry *lle, *lle_tmp;
664 uint8_t linkhdr[LLE_MAX_LINKHDR];
670 KASSERT(dl != NULL && dl->sdl_family == AF_LINK,
671 ("%s: invalid dl\n", __func__));
673 ifp = ifnet_byindex(dl->sdl_index);
675 log(LOG_INFO, "%s: invalid ifp (sdl_index %d)\n",
676 __func__, dl->sdl_index);
680 /* XXX linked list may be too expensive */
681 LLTABLE_LIST_RLOCK();
682 SLIST_FOREACH(llt, &V_lltables, llt_link) {
683 if (llt->llt_af == dst->sa_family &&
687 LLTABLE_LIST_RUNLOCK();
688 KASSERT(llt != NULL, ("Yep, ugly hacks are bad\n"));
692 switch (rtm->rtm_type) {
696 if (rtm->rtm_rmx.rmx_expire == 0)
697 laflags = LLE_STATIC;
698 lle = lltable_alloc_entry(llt, laflags, dst);
702 linkhdrsize = sizeof(linkhdr);
703 if (lltable_calc_llheader(ifp, dst->sa_family, LLADDR(dl),
704 linkhdr, &linkhdrsize, &lladdr_off) != 0)
706 lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize,
708 if ((rtm->rtm_flags & RTF_ANNOUNCE))
709 lle->la_flags |= LLE_PUB;
710 lle->la_expire = rtm->rtm_rmx.rmx_expire;
712 laflags = lle->la_flags;
714 /* Try to link new entry */
716 IF_AFDATA_WLOCK(ifp);
718 lle_tmp = lla_lookup(llt, LLE_EXCLUSIVE, dst);
719 if (lle_tmp != NULL) {
720 /* Check if we are trying to replace immutable entry */
721 if ((lle_tmp->la_flags & LLE_IFADDR) != 0) {
722 IF_AFDATA_WUNLOCK(ifp);
723 LLE_WUNLOCK(lle_tmp);
724 lltable_free_entry(llt, lle);
727 /* Unlink existing entry from table */
728 lltable_unlink_entry(llt, lle_tmp);
730 lltable_link_entry(llt, lle);
731 IF_AFDATA_WUNLOCK(ifp);
733 if (lle_tmp != NULL) {
734 EVENTHANDLER_INVOKE(lle_event, lle_tmp,LLENTRY_EXPIRED);
735 lltable_free_entry(llt, lle_tmp);
739 * By invoking LLE handler here we might get
740 * two events on static LLE entry insertion
741 * in routing socket. However, since we might have
742 * other subscribers we need to generate this event.
744 EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_RESOLVED);
748 if ((laflags & LLE_PUB) && dst->sa_family == AF_INET)
750 &((struct sockaddr_in *)dst)->sin_addr,
751 &((struct sockaddr_in *)dst)->sin_addr,
752 (u_char *)LLADDR(dl));
758 return (lltable_delete_addr(llt, 0, dst));
770 struct sockaddr l3_addr;
774 llatbl_lle_show(struct llentry_sa *la)
780 db_printf("lle=%p\n", lle);
781 db_printf(" lle_next=%p\n", lle->lle_next.cle_next);
782 db_printf(" lle_lock=%p\n", &lle->lle_lock);
783 db_printf(" lle_tbl=%p\n", lle->lle_tbl);
784 db_printf(" lle_head=%p\n", lle->lle_head);
785 db_printf(" la_hold=%p\n", lle->la_hold);
786 db_printf(" la_numheld=%d\n", lle->la_numheld);
787 db_printf(" la_expire=%ju\n", (uintmax_t)lle->la_expire);
788 db_printf(" la_flags=0x%04x\n", lle->la_flags);
789 db_printf(" la_asked=%u\n", lle->la_asked);
790 db_printf(" la_preempt=%u\n", lle->la_preempt);
791 db_printf(" ln_state=%d\n", lle->ln_state);
792 db_printf(" ln_router=%u\n", lle->ln_router);
793 db_printf(" ln_ntick=%ju\n", (uintmax_t)lle->ln_ntick);
794 db_printf(" lle_refcnt=%d\n", lle->lle_refcnt);
795 bcopy(lle->ll_addr, octet, sizeof(octet));
796 db_printf(" ll_addr=%02x:%02x:%02x:%02x:%02x:%02x\n",
797 octet[0], octet[1], octet[2], octet[3], octet[4], octet[5]);
798 db_printf(" lle_timer=%p\n", &lle->lle_timer);
800 switch (la->l3_addr.sa_family) {
804 struct sockaddr_in *sin;
805 char l3s[INET_ADDRSTRLEN];
807 sin = (struct sockaddr_in *)&la->l3_addr;
808 inet_ntoa_r(sin->sin_addr, l3s);
809 db_printf(" l3_addr=%s\n", l3s);
816 struct sockaddr_in6 *sin6;
817 char l3s[INET6_ADDRSTRLEN];
819 sin6 = (struct sockaddr_in6 *)&la->l3_addr;
820 ip6_sprintf(l3s, &sin6->sin6_addr);
821 db_printf(" l3_addr=%s\n", l3s);
826 db_printf(" l3_addr=N/A (af=%d)\n", la->l3_addr.sa_family);
831 DB_SHOW_COMMAND(llentry, db_show_llentry)
835 db_printf("usage: show llentry <struct llentry *>\n");
839 llatbl_lle_show((struct llentry_sa *)addr);
843 llatbl_llt_show(struct lltable *llt)
848 db_printf("llt=%p llt_af=%d llt_ifp=%p\n",
849 llt, llt->llt_af, llt->llt_ifp);
851 for (i = 0; i < llt->llt_hsize; i++) {
852 CK_LIST_FOREACH(lle, &llt->lle_head[i], lle_next) {
854 llatbl_lle_show((struct llentry_sa *)lle);
861 DB_SHOW_COMMAND(lltable, db_show_lltable)
865 db_printf("usage: show lltable <struct lltable *>\n");
869 llatbl_llt_show((struct lltable *)addr);
872 DB_SHOW_ALL_COMMAND(lltables, db_show_all_lltables)
874 VNET_ITERATOR_DECL(vnet_iter);
877 VNET_FOREACH(vnet_iter) {
878 CURVNET_SET_QUIET(vnet_iter);
880 db_printf("vnet=%p\n", curvnet);
882 SLIST_FOREACH(llt, &V_lltables, llt_link) {
883 db_printf("llt=%p llt_af=%d llt_ifp=%p(%s)\n",
884 llt, llt->llt_af, llt->llt_ifp,
885 (llt->llt_ifp != NULL) ?
886 llt->llt_ifp->if_xname : "?");
887 if (have_addr && addr != 0) /* verbose */
888 llatbl_llt_show(llt);