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) {
181 IF_AFDATA_WLOCK_ASSERT(lle->lle_tbl->llt_ifp);
182 CK_LIST_REMOVE(lle, lle_next);
183 lle->la_flags &= ~(LLE_VALID | LLE_LINKED);
186 lle->lle_head = NULL;
191 struct prefix_match_data {
192 const struct sockaddr *addr;
193 const struct sockaddr *mask;
194 struct llentries dchain;
199 htable_prefix_free_cb(struct lltable *llt, struct llentry *lle, void *farg)
201 struct prefix_match_data *pmd;
203 pmd = (struct prefix_match_data *)farg;
205 if (llt->llt_match_prefix(pmd->addr, pmd->mask, pmd->flags, lle)) {
207 CK_LIST_INSERT_HEAD(&pmd->dchain, lle, lle_chain);
214 htable_prefix_free(struct lltable *llt, const struct sockaddr *addr,
215 const struct sockaddr *mask, u_int flags)
217 struct llentry *lle, *next;
218 struct prefix_match_data pmd;
220 bzero(&pmd, sizeof(pmd));
224 CK_LIST_INIT(&pmd.dchain);
226 IF_AFDATA_WLOCK(llt->llt_ifp);
227 /* Push matching lles to chain */
228 lltable_foreach_lle(llt, htable_prefix_free_cb, &pmd);
230 llentries_unlink(llt, &pmd.dchain);
231 IF_AFDATA_WUNLOCK(llt->llt_ifp);
233 CK_LIST_FOREACH_SAFE(lle, &pmd.dchain, lle_chain, next)
234 lltable_free_entry(llt, lle);
238 htable_free_tbl(struct lltable *llt)
241 free(llt->lle_head, M_LLTABLE);
242 free(llt, M_LLTABLE);
246 llentries_unlink(struct lltable *llt, struct llentries *head)
248 struct llentry *lle, *next;
250 CK_LIST_FOREACH_SAFE(lle, head, lle_chain, next)
251 llt->llt_unlink_entry(lle);
255 * Helper function used to drop all mbufs in hold queue.
257 * Returns the number of held packets, if any, that were dropped.
260 lltable_drop_entry_queue(struct llentry *lle)
265 LLE_WLOCK_ASSERT(lle);
268 while ((lle->la_numheld > 0) && (lle->la_hold != NULL)) {
269 next = lle->la_hold->m_nextpkt;
270 m_freem(lle->la_hold);
276 KASSERT(lle->la_numheld == 0,
277 ("%s: la_numheld %d > 0, pkts_droped %zd", __func__,
278 lle->la_numheld, pkts_dropped));
280 return (pkts_dropped);
284 lltable_set_entry_addr(struct ifnet *ifp, struct llentry *lle,
285 const char *linkhdr, size_t linkhdrsize, int lladdr_off)
288 memcpy(lle->r_linkdata, linkhdr, linkhdrsize);
289 lle->r_hdrlen = linkhdrsize;
290 lle->ll_addr = &lle->r_linkdata[lladdr_off];
291 lle->la_flags |= LLE_VALID;
292 lle->r_flags |= RLLE_VALID;
296 * Tries to update @lle link-level address.
297 * Since update requires AFDATA WLOCK, function
298 * drops @lle lock, acquires AFDATA lock and then acquires
299 * @lle lock to maintain lock order.
301 * Returns 1 on success.
304 lltable_try_set_entry_addr(struct ifnet *ifp, struct llentry *lle,
305 const char *linkhdr, size_t linkhdrsize, int lladdr_off)
308 /* Perform real LLE update */
309 /* use afdata WLOCK to update fields */
310 LLE_WLOCK_ASSERT(lle);
313 IF_AFDATA_WLOCK(ifp);
317 * Since we droppped LLE lock, other thread might have deleted
318 * this lle. Check and return
320 if ((lle->la_flags & LLE_DELETED) != 0) {
321 IF_AFDATA_WUNLOCK(ifp);
322 LLE_FREE_LOCKED(lle);
327 lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize, lladdr_off);
329 IF_AFDATA_WUNLOCK(ifp);
337 * Helper function used to pre-compute full/partial link-layer
338 * header data suitable for feeding into if_output().
341 lltable_calc_llheader(struct ifnet *ifp, int family, char *lladdr,
342 char *buf, size_t *bufsize, int *lladdr_off)
344 struct if_encap_req ereq;
347 bzero(buf, *bufsize);
348 bzero(&ereq, sizeof(ereq));
350 ereq.bufsize = *bufsize;
351 ereq.rtype = IFENCAP_LL;
352 ereq.family = family;
353 ereq.lladdr = lladdr;
354 ereq.lladdr_len = ifp->if_addrlen;
355 error = ifp->if_requestencap(ifp, &ereq);
357 *bufsize = ereq.bufsize;
358 *lladdr_off = ereq.lladdr_off;
365 * Update link-layer header for given @lle after
366 * interface lladdr was changed.
369 llentry_update_ifaddr(struct lltable *llt, struct llentry *lle, void *farg)
372 u_char linkhdr[LLE_MAX_LINKHDR];
377 ifp = (struct ifnet *)farg;
379 lladdr = lle->ll_addr;
382 if ((lle->la_flags & LLE_VALID) == 0) {
387 if ((lle->la_flags & LLE_IFADDR) != 0)
388 lladdr = IF_LLADDR(ifp);
390 linkhdrsize = sizeof(linkhdr);
391 lltable_calc_llheader(ifp, llt->llt_af, lladdr, linkhdr, &linkhdrsize,
393 memcpy(lle->r_linkdata, linkhdr, linkhdrsize);
400 * Update all calculated headers for given @llt
403 lltable_update_ifaddr(struct lltable *llt)
406 if (llt->llt_ifp->if_flags & IFF_LOOPBACK)
409 IF_AFDATA_WLOCK(llt->llt_ifp);
410 lltable_foreach_lle(llt, llentry_update_ifaddr, llt->llt_ifp);
411 IF_AFDATA_WUNLOCK(llt->llt_ifp);
416 * Performs generic cleanup routines and frees lle.
418 * Called for non-linked entries, with callouts and
419 * other AF-specific cleanups performed.
421 * @lle must be passed WLOCK'ed
423 * Returns the number of held packets, if any, that were dropped.
426 llentry_free(struct llentry *lle)
430 LLE_WLOCK_ASSERT(lle);
432 KASSERT((lle->la_flags & LLE_LINKED) == 0, ("freeing linked lle"));
434 pkts_dropped = lltable_drop_entry_queue(lle);
437 if (callout_stop(&lle->lle_timer) > 0)
439 LLE_FREE_LOCKED(lle);
441 return (pkts_dropped);
445 * Free all entries from given table and free itself.
449 lltable_free_cb(struct lltable *llt, struct llentry *lle, void *farg)
451 struct llentries *dchain;
453 dchain = (struct llentries *)farg;
456 CK_LIST_INSERT_HEAD(dchain, lle, lle_chain);
462 * Free all entries from given table and free itself.
465 lltable_free(struct lltable *llt)
467 struct llentry *lle, *next;
468 struct llentries dchain;
470 KASSERT(llt != NULL, ("%s: llt is NULL", __func__));
474 CK_LIST_INIT(&dchain);
475 IF_AFDATA_WLOCK(llt->llt_ifp);
476 /* Push all lles to @dchain */
477 lltable_foreach_lle(llt, lltable_free_cb, &dchain);
478 llentries_unlink(llt, &dchain);
479 IF_AFDATA_WUNLOCK(llt->llt_ifp);
481 CK_LIST_FOREACH_SAFE(lle, &dchain, lle_chain, next) {
485 llt->llt_free_tbl(llt);
489 * Deletes an address from given lltable.
490 * Used for userland interaction to remove
491 * individual entries. Skips entries added by OS.
494 lltable_delete_addr(struct lltable *llt, u_int flags,
495 const struct sockaddr *l3addr)
501 IF_AFDATA_WLOCK(ifp);
502 lle = lla_lookup(llt, LLE_EXCLUSIVE, l3addr);
505 IF_AFDATA_WUNLOCK(ifp);
508 if ((lle->la_flags & LLE_IFADDR) != 0 && (flags & LLE_IFADDR) == 0) {
509 IF_AFDATA_WUNLOCK(ifp);
514 lltable_unlink_entry(llt, lle);
515 IF_AFDATA_WUNLOCK(ifp);
517 llt->llt_delete_entry(llt, lle);
523 lltable_prefix_free(int af, struct sockaddr *addr, struct sockaddr *mask,
528 LLTABLE_LIST_RLOCK();
529 SLIST_FOREACH(llt, &V_lltables, llt_link) {
530 if (llt->llt_af != af)
533 llt->llt_prefix_free(llt, addr, mask, flags);
535 LLTABLE_LIST_RUNLOCK();
539 lltable_allocate_htbl(uint32_t hsize)
544 llt = malloc(sizeof(struct lltable), M_LLTABLE, M_WAITOK | M_ZERO);
545 llt->llt_hsize = hsize;
546 llt->lle_head = malloc(sizeof(struct llentries) * hsize,
547 M_LLTABLE, M_WAITOK | M_ZERO);
549 for (i = 0; i < llt->llt_hsize; i++)
550 CK_LIST_INIT(&llt->lle_head[i]);
552 /* Set some default callbacks */
553 llt->llt_link_entry = htable_link_entry;
554 llt->llt_unlink_entry = htable_unlink_entry;
555 llt->llt_prefix_free = htable_prefix_free;
556 llt->llt_foreach_entry = htable_foreach_lle;
557 llt->llt_free_tbl = htable_free_tbl;
563 * Links lltable to global llt list.
566 lltable_link(struct lltable *llt)
569 LLTABLE_LIST_WLOCK();
570 SLIST_INSERT_HEAD(&V_lltables, llt, llt_link);
571 LLTABLE_LIST_WUNLOCK();
575 lltable_unlink(struct lltable *llt)
578 LLTABLE_LIST_WLOCK();
579 SLIST_REMOVE(&V_lltables, llt, lltable, llt_link);
580 LLTABLE_LIST_WUNLOCK();
585 * External methods used by lltable consumers
589 lltable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg)
592 return (llt->llt_foreach_entry(llt, f, farg));
596 lltable_alloc_entry(struct lltable *llt, u_int flags,
597 const struct sockaddr *l3addr)
600 return (llt->llt_alloc_entry(llt, flags, l3addr));
604 lltable_free_entry(struct lltable *llt, struct llentry *lle)
607 llt->llt_free_entry(llt, lle);
611 lltable_link_entry(struct lltable *llt, struct llentry *lle)
614 llt->llt_link_entry(llt, lle);
618 lltable_unlink_entry(struct lltable *llt, struct llentry *lle)
621 llt->llt_unlink_entry(lle);
625 lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa)
630 llt->llt_fill_sa_entry(lle, sa);
634 lltable_get_ifp(const struct lltable *llt)
637 return (llt->llt_ifp);
641 lltable_get_af(const struct lltable *llt)
644 return (llt->llt_af);
648 * Called in route_output when rtm_flags contains RTF_LLDATA.
651 lla_rt_output(struct rt_msghdr *rtm, struct rt_addrinfo *info)
653 struct sockaddr_dl *dl =
654 (struct sockaddr_dl *)info->rti_info[RTAX_GATEWAY];
655 struct sockaddr *dst = (struct sockaddr *)info->rti_info[RTAX_DST];
658 struct llentry *lle, *lle_tmp;
659 uint8_t linkhdr[LLE_MAX_LINKHDR];
665 KASSERT(dl != NULL && dl->sdl_family == AF_LINK,
666 ("%s: invalid dl\n", __func__));
668 ifp = ifnet_byindex(dl->sdl_index);
670 log(LOG_INFO, "%s: invalid ifp (sdl_index %d)\n",
671 __func__, dl->sdl_index);
675 /* XXX linked list may be too expensive */
676 LLTABLE_LIST_RLOCK();
677 SLIST_FOREACH(llt, &V_lltables, llt_link) {
678 if (llt->llt_af == dst->sa_family &&
682 LLTABLE_LIST_RUNLOCK();
683 KASSERT(llt != NULL, ("Yep, ugly hacks are bad\n"));
687 switch (rtm->rtm_type) {
691 if (rtm->rtm_rmx.rmx_expire == 0)
692 laflags = LLE_STATIC;
693 lle = lltable_alloc_entry(llt, laflags, dst);
697 linkhdrsize = sizeof(linkhdr);
698 if (lltable_calc_llheader(ifp, dst->sa_family, LLADDR(dl),
699 linkhdr, &linkhdrsize, &lladdr_off) != 0)
701 lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize,
703 if ((rtm->rtm_flags & RTF_ANNOUNCE))
704 lle->la_flags |= LLE_PUB;
705 lle->la_expire = rtm->rtm_rmx.rmx_expire;
707 laflags = lle->la_flags;
709 /* Try to link new entry */
711 IF_AFDATA_WLOCK(ifp);
713 lle_tmp = lla_lookup(llt, LLE_EXCLUSIVE, dst);
714 if (lle_tmp != NULL) {
715 /* Check if we are trying to replace immutable entry */
716 if ((lle_tmp->la_flags & LLE_IFADDR) != 0) {
717 IF_AFDATA_WUNLOCK(ifp);
718 LLE_WUNLOCK(lle_tmp);
719 lltable_free_entry(llt, lle);
722 /* Unlink existing entry from table */
723 lltable_unlink_entry(llt, lle_tmp);
725 lltable_link_entry(llt, lle);
726 IF_AFDATA_WUNLOCK(ifp);
728 if (lle_tmp != NULL) {
729 EVENTHANDLER_INVOKE(lle_event, lle_tmp,LLENTRY_EXPIRED);
730 lltable_free_entry(llt, lle_tmp);
734 * By invoking LLE handler here we might get
735 * two events on static LLE entry insertion
736 * in routing socket. However, since we might have
737 * other subscribers we need to generate this event.
739 EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_RESOLVED);
743 if ((laflags & LLE_PUB) && dst->sa_family == AF_INET)
745 &((struct sockaddr_in *)dst)->sin_addr,
746 &((struct sockaddr_in *)dst)->sin_addr,
747 (u_char *)LLADDR(dl));
753 return (lltable_delete_addr(llt, 0, dst));
765 struct sockaddr l3_addr;
769 llatbl_lle_show(struct llentry_sa *la)
775 db_printf("lle=%p\n", lle);
776 db_printf(" lle_next=%p\n", lle->lle_next.cle_next);
777 db_printf(" lle_lock=%p\n", &lle->lle_lock);
778 db_printf(" lle_tbl=%p\n", lle->lle_tbl);
779 db_printf(" lle_head=%p\n", lle->lle_head);
780 db_printf(" la_hold=%p\n", lle->la_hold);
781 db_printf(" la_numheld=%d\n", lle->la_numheld);
782 db_printf(" la_expire=%ju\n", (uintmax_t)lle->la_expire);
783 db_printf(" la_flags=0x%04x\n", lle->la_flags);
784 db_printf(" la_asked=%u\n", lle->la_asked);
785 db_printf(" la_preempt=%u\n", lle->la_preempt);
786 db_printf(" ln_state=%d\n", lle->ln_state);
787 db_printf(" ln_router=%u\n", lle->ln_router);
788 db_printf(" ln_ntick=%ju\n", (uintmax_t)lle->ln_ntick);
789 db_printf(" lle_refcnt=%d\n", lle->lle_refcnt);
790 bcopy(lle->ll_addr, octet, sizeof(octet));
791 db_printf(" ll_addr=%02x:%02x:%02x:%02x:%02x:%02x\n",
792 octet[0], octet[1], octet[2], octet[3], octet[4], octet[5]);
793 db_printf(" lle_timer=%p\n", &lle->lle_timer);
795 switch (la->l3_addr.sa_family) {
799 struct sockaddr_in *sin;
800 char l3s[INET_ADDRSTRLEN];
802 sin = (struct sockaddr_in *)&la->l3_addr;
803 inet_ntoa_r(sin->sin_addr, l3s);
804 db_printf(" l3_addr=%s\n", l3s);
811 struct sockaddr_in6 *sin6;
812 char l3s[INET6_ADDRSTRLEN];
814 sin6 = (struct sockaddr_in6 *)&la->l3_addr;
815 ip6_sprintf(l3s, &sin6->sin6_addr);
816 db_printf(" l3_addr=%s\n", l3s);
821 db_printf(" l3_addr=N/A (af=%d)\n", la->l3_addr.sa_family);
826 DB_SHOW_COMMAND(llentry, db_show_llentry)
830 db_printf("usage: show llentry <struct llentry *>\n");
834 llatbl_lle_show((struct llentry_sa *)addr);
838 llatbl_llt_show(struct lltable *llt)
843 db_printf("llt=%p llt_af=%d llt_ifp=%p\n",
844 llt, llt->llt_af, llt->llt_ifp);
846 for (i = 0; i < llt->llt_hsize; i++) {
847 CK_LIST_FOREACH(lle, &llt->lle_head[i], lle_next) {
849 llatbl_lle_show((struct llentry_sa *)lle);
856 DB_SHOW_COMMAND(lltable, db_show_lltable)
860 db_printf("usage: show lltable <struct lltable *>\n");
864 llatbl_llt_show((struct lltable *)addr);
867 DB_SHOW_ALL_COMMAND(lltables, db_show_all_lltables)
869 VNET_ITERATOR_DECL(vnet_iter);
872 VNET_FOREACH(vnet_iter) {
873 CURVNET_SET_QUIET(vnet_iter);
875 db_printf("vnet=%p\n", curvnet);
877 SLIST_FOREACH(llt, &V_lltables, llt_link) {
878 db_printf("llt=%p llt_af=%d llt_ifp=%p(%s)\n",
879 llt, llt->llt_af, llt->llt_ifp,
880 (llt->llt_ifp != NULL) ?
881 llt->llt_ifp->if_xname : "?");
882 if (have_addr && addr != 0) /* verbose */
883 llatbl_llt_show(llt);