Index: head/sys/net/if_llatbl.c =================================================================== --- head/sys/net/if_llatbl.c (revision 288300) +++ head/sys/net/if_llatbl.c (revision 288301) @@ -1,831 +1,830 @@ /* * Copyright (c) 2004 Luigi Rizzo, Alessandro Cerri. All rights reserved. * Copyright (c) 2004-2008 Qing Li. All rights reserved. * Copyright (c) 2008 Kip Macy. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include "opt_ddb.h" #include "opt_inet.h" #include "opt_inet6.h" #include #include #include #include #include #include #include #include #include #include #include #ifdef DDB #include #endif #include #include #include #include #include #include #include #include #include #include #include MALLOC_DEFINE(M_LLTABLE, "lltable", "link level address tables"); static VNET_DEFINE(SLIST_HEAD(, lltable), lltables); #define V_lltables VNET(lltables) static void vnet_lltable_init(void); struct rwlock lltable_rwlock; RW_SYSINIT(lltable_rwlock, &lltable_rwlock, "lltable_rwlock"); static void lltable_unlink(struct lltable *llt); static void llentries_unlink(struct lltable *llt, struct llentries *head); static void htable_unlink_entry(struct llentry *lle); static void htable_link_entry(struct lltable *llt, struct llentry *lle); static int htable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg); /* * Dump lle state for a specific address family. */ static int lltable_dump_af(struct lltable *llt, struct sysctl_req *wr) { int error; LLTABLE_LOCK_ASSERT(); if (llt->llt_ifp->if_flags & IFF_LOOPBACK) return (0); error = 0; IF_AFDATA_RLOCK(llt->llt_ifp); error = lltable_foreach_lle(llt, (llt_foreach_cb_t *)llt->llt_dump_entry, wr); IF_AFDATA_RUNLOCK(llt->llt_ifp); return (error); } /* * Dump arp state for a specific address family. */ int lltable_sysctl_dumparp(int af, struct sysctl_req *wr) { struct lltable *llt; int error = 0; LLTABLE_RLOCK(); SLIST_FOREACH(llt, &V_lltables, llt_link) { if (llt->llt_af == af) { error = lltable_dump_af(llt, wr); if (error != 0) goto done; } } done: LLTABLE_RUNLOCK(); return (error); } /* * Common function helpers for chained hash table. */ /* * Runs specified callback for each entry in @llt. * Caller does the locking. * */ static int htable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg) { struct llentry *lle, *next; int i, error; error = 0; for (i = 0; i < llt->llt_hsize; i++) { LIST_FOREACH_SAFE(lle, &llt->lle_head[i], lle_next, next) { error = f(llt, lle, farg); if (error != 0) break; } } return (error); } static void htable_link_entry(struct lltable *llt, struct llentry *lle) { struct llentries *lleh; uint32_t hashidx; if ((lle->la_flags & LLE_LINKED) != 0) return; IF_AFDATA_WLOCK_ASSERT(llt->llt_ifp); hashidx = llt->llt_hash(lle, llt->llt_hsize); lleh = &llt->lle_head[hashidx]; lle->lle_tbl = llt; lle->lle_head = lleh; lle->la_flags |= LLE_LINKED; LIST_INSERT_HEAD(lleh, lle, lle_next); } static void htable_unlink_entry(struct llentry *lle) { if ((lle->la_flags & LLE_LINKED) != 0) { IF_AFDATA_WLOCK_ASSERT(lle->lle_tbl->llt_ifp); LIST_REMOVE(lle, lle_next); lle->la_flags &= ~(LLE_VALID | LLE_LINKED); #if 0 lle->lle_tbl = NULL; lle->lle_head = NULL; #endif } } struct prefix_match_data { const struct sockaddr *addr; const struct sockaddr *mask; struct llentries dchain; u_int flags; }; static int htable_prefix_free_cb(struct lltable *llt, struct llentry *lle, void *farg) { struct prefix_match_data *pmd; pmd = (struct prefix_match_data *)farg; if (llt->llt_match_prefix(pmd->addr, pmd->mask, pmd->flags, lle)) { LLE_WLOCK(lle); LIST_INSERT_HEAD(&pmd->dchain, lle, lle_chain); } return (0); } static void htable_prefix_free(struct lltable *llt, const struct sockaddr *addr, const struct sockaddr *mask, u_int flags) { struct llentry *lle, *next; struct prefix_match_data pmd; bzero(&pmd, sizeof(pmd)); pmd.addr = addr; pmd.mask = mask; pmd.flags = flags; LIST_INIT(&pmd.dchain); IF_AFDATA_WLOCK(llt->llt_ifp); /* Push matching lles to chain */ lltable_foreach_lle(llt, htable_prefix_free_cb, &pmd); llentries_unlink(llt, &pmd.dchain); IF_AFDATA_WUNLOCK(llt->llt_ifp); LIST_FOREACH_SAFE(lle, &pmd.dchain, lle_chain, next) lltable_free_entry(llt, lle); } static void htable_free_tbl(struct lltable *llt) { free(llt->lle_head, M_LLTABLE); free(llt, M_LLTABLE); } static void llentries_unlink(struct lltable *llt, struct llentries *head) { struct llentry *lle, *next; LIST_FOREACH_SAFE(lle, head, lle_chain, next) llt->llt_unlink_entry(lle); } /* * Helper function used to drop all mbufs in hold queue. * * Returns the number of held packets, if any, that were dropped. */ size_t lltable_drop_entry_queue(struct llentry *lle) { size_t pkts_dropped; struct mbuf *next; LLE_WLOCK_ASSERT(lle); pkts_dropped = 0; while ((lle->la_numheld > 0) && (lle->la_hold != NULL)) { next = lle->la_hold->m_nextpkt; m_freem(lle->la_hold); lle->la_hold = next; lle->la_numheld--; pkts_dropped++; } KASSERT(lle->la_numheld == 0, ("%s: la_numheld %d > 0, pkts_droped %zd", __func__, lle->la_numheld, pkts_dropped)); return (pkts_dropped); } /* * * Performes generic cleanup routines and frees lle. * * Called for non-linked entries, with callouts and * other AF-specific cleanups performed. * * @lle must be passed WLOCK'ed * * Returns the number of held packets, if any, that were dropped. */ size_t llentry_free(struct llentry *lle) { size_t pkts_dropped; LLE_WLOCK_ASSERT(lle); KASSERT((lle->la_flags & LLE_LINKED) == 0, ("freeing linked lle")); pkts_dropped = lltable_drop_entry_queue(lle); LLE_FREE_LOCKED(lle); return (pkts_dropped); } /* * (al)locate an llentry for address dst (equivalent to rtalloc for new-arp). * * If found the llentry * is returned referenced and unlocked. */ struct llentry * llentry_alloc(struct ifnet *ifp, struct lltable *lt, struct sockaddr_storage *dst) { struct llentry *la, *la_tmp; IF_AFDATA_RLOCK(ifp); la = lla_lookup(lt, LLE_EXCLUSIVE, (struct sockaddr *)dst); IF_AFDATA_RUNLOCK(ifp); if (la != NULL) { LLE_ADDREF(la); LLE_WUNLOCK(la); return (la); } if ((ifp->if_flags & (IFF_NOARP | IFF_STATICARP)) == 0) { la = lltable_alloc_entry(lt, 0, (struct sockaddr *)dst); if (la == NULL) return (NULL); IF_AFDATA_WLOCK(ifp); LLE_WLOCK(la); /* Prefer any existing LLE over newly-created one */ la_tmp = lla_lookup(lt, LLE_EXCLUSIVE, (struct sockaddr *)dst); if (la_tmp == NULL) lltable_link_entry(lt, la); IF_AFDATA_WUNLOCK(ifp); if (la_tmp != NULL) { lltable_free_entry(lt, la); la = la_tmp; } LLE_ADDREF(la); LLE_WUNLOCK(la); } return (la); } /* * Free all entries from given table and free itself. */ static int lltable_free_cb(struct lltable *llt, struct llentry *lle, void *farg) { struct llentries *dchain; dchain = (struct llentries *)farg; LLE_WLOCK(lle); LIST_INSERT_HEAD(dchain, lle, lle_chain); return (0); } /* * Free all entries from given table and free itself. */ void lltable_free(struct lltable *llt) { struct llentry *lle, *next; struct llentries dchain; KASSERT(llt != NULL, ("%s: llt is NULL", __func__)); lltable_unlink(llt); LIST_INIT(&dchain); IF_AFDATA_WLOCK(llt->llt_ifp); /* Push all lles to @dchain */ lltable_foreach_lle(llt, lltable_free_cb, &dchain); llentries_unlink(llt, &dchain); IF_AFDATA_WUNLOCK(llt->llt_ifp); LIST_FOREACH_SAFE(lle, &dchain, lle_chain, next) { if (callout_stop(&lle->lle_timer)) LLE_REMREF(lle); llentry_free(lle); } llt->llt_free_tbl(llt); } #if 0 void lltable_drain(int af) { struct lltable *llt; struct llentry *lle; register int i; LLTABLE_RLOCK(); SLIST_FOREACH(llt, &V_lltables, llt_link) { if (llt->llt_af != af) continue; for (i=0; i < llt->llt_hsize; i++) { LIST_FOREACH(lle, &llt->lle_head[i], lle_next) { LLE_WLOCK(lle); if (lle->la_hold) { m_freem(lle->la_hold); lle->la_hold = NULL; } LLE_WUNLOCK(lle); } } } LLTABLE_RUNLOCK(); } #endif /* * Deletes an address from given lltable. * Used for userland interaction to remove * individual entries. Skips entries added by OS. */ int lltable_delete_addr(struct lltable *llt, u_int flags, const struct sockaddr *l3addr) { struct llentry *lle; struct ifnet *ifp; ifp = llt->llt_ifp; IF_AFDATA_WLOCK(ifp); lle = lla_lookup(llt, LLE_EXCLUSIVE, l3addr); if (lle == NULL) { IF_AFDATA_WUNLOCK(ifp); return (ENOENT); } if ((lle->la_flags & LLE_IFADDR) != 0 && (flags & LLE_IFADDR) == 0) { IF_AFDATA_WUNLOCK(ifp); LLE_WUNLOCK(lle); return (EPERM); } lltable_unlink_entry(llt, lle); IF_AFDATA_WUNLOCK(ifp); llt->llt_delete_entry(llt, lle); return (0); } void lltable_prefix_free(int af, struct sockaddr *addr, struct sockaddr *mask, u_int flags) { struct lltable *llt; LLTABLE_RLOCK(); SLIST_FOREACH(llt, &V_lltables, llt_link) { if (llt->llt_af != af) continue; llt->llt_prefix_free(llt, addr, mask, flags); } LLTABLE_RUNLOCK(); } struct lltable * lltable_allocate_htbl(uint32_t hsize) { struct lltable *llt; int i; llt = malloc(sizeof(struct lltable), M_LLTABLE, M_WAITOK | M_ZERO); llt->llt_hsize = hsize; llt->lle_head = malloc(sizeof(struct llentries) * hsize, M_LLTABLE, M_WAITOK | M_ZERO); for (i = 0; i < llt->llt_hsize; i++) LIST_INIT(&llt->lle_head[i]); /* Set some default callbacks */ llt->llt_link_entry = htable_link_entry; llt->llt_unlink_entry = htable_unlink_entry; llt->llt_prefix_free = htable_prefix_free; llt->llt_foreach_entry = htable_foreach_lle; llt->llt_free_tbl = htable_free_tbl; return (llt); } /* * Links lltable to global llt list. */ void lltable_link(struct lltable *llt) { LLTABLE_WLOCK(); SLIST_INSERT_HEAD(&V_lltables, llt, llt_link); LLTABLE_WUNLOCK(); } static void lltable_unlink(struct lltable *llt) { LLTABLE_WLOCK(); SLIST_REMOVE(&V_lltables, llt, lltable, llt_link); LLTABLE_WUNLOCK(); } /* * External methods used by lltable consumers */ int lltable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg) { return (llt->llt_foreach_entry(llt, f, farg)); } struct llentry * lltable_alloc_entry(struct lltable *llt, u_int flags, const struct sockaddr *l3addr) { return (llt->llt_alloc_entry(llt, flags, l3addr)); } void lltable_free_entry(struct lltable *llt, struct llentry *lle) { llt->llt_free_entry(llt, lle); } void lltable_link_entry(struct lltable *llt, struct llentry *lle) { llt->llt_link_entry(llt, lle); } void lltable_unlink_entry(struct lltable *llt, struct llentry *lle) { llt->llt_unlink_entry(lle); } void lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa) { struct lltable *llt; llt = lle->lle_tbl; llt->llt_fill_sa_entry(lle, sa); } struct ifnet * lltable_get_ifp(const struct lltable *llt) { return (llt->llt_ifp); } int lltable_get_af(const struct lltable *llt) { return (llt->llt_af); } /* * Called in route_output when rtm_flags contains RTF_LLDATA. */ int lla_rt_output(struct rt_msghdr *rtm, struct rt_addrinfo *info) { struct sockaddr_dl *dl = (struct sockaddr_dl *)info->rti_info[RTAX_GATEWAY]; struct sockaddr *dst = (struct sockaddr *)info->rti_info[RTAX_DST]; struct ifnet *ifp; struct lltable *llt; struct llentry *lle, *lle_tmp; u_int laflags = 0; int error; KASSERT(dl != NULL && dl->sdl_family == AF_LINK, ("%s: invalid dl\n", __func__)); ifp = ifnet_byindex(dl->sdl_index); if (ifp == NULL) { log(LOG_INFO, "%s: invalid ifp (sdl_index %d)\n", __func__, dl->sdl_index); return EINVAL; } /* XXX linked list may be too expensive */ LLTABLE_RLOCK(); SLIST_FOREACH(llt, &V_lltables, llt_link) { if (llt->llt_af == dst->sa_family && llt->llt_ifp == ifp) break; } LLTABLE_RUNLOCK(); KASSERT(llt != NULL, ("Yep, ugly hacks are bad\n")); error = 0; switch (rtm->rtm_type) { case RTM_ADD: /* Add static LLE */ laflags = 0; if (rtm->rtm_rmx.rmx_expire == 0) laflags = LLE_STATIC; lle = lltable_alloc_entry(llt, laflags, dst); if (lle == NULL) return (ENOMEM); bcopy(LLADDR(dl), &lle->ll_addr, ifp->if_addrlen); if ((rtm->rtm_flags & RTF_ANNOUNCE)) lle->la_flags |= LLE_PUB; lle->la_flags |= LLE_VALID; lle->la_expire = rtm->rtm_rmx.rmx_expire; laflags = lle->la_flags; /* Try to link new entry */ lle_tmp = NULL; IF_AFDATA_WLOCK(ifp); LLE_WLOCK(lle); lle_tmp = lla_lookup(llt, LLE_EXCLUSIVE, dst); if (lle_tmp != NULL) { /* Check if we are trying to replace immutable entry */ if ((lle_tmp->la_flags & LLE_IFADDR) != 0) { IF_AFDATA_WUNLOCK(ifp); LLE_WUNLOCK(lle_tmp); lltable_free_entry(llt, lle); return (EPERM); } /* Unlink existing entry from table */ lltable_unlink_entry(llt, lle_tmp); } lltable_link_entry(llt, lle); IF_AFDATA_WUNLOCK(ifp); if (lle_tmp != NULL) { EVENTHANDLER_INVOKE(lle_event, lle_tmp,LLENTRY_EXPIRED); lltable_free_entry(llt, lle_tmp); } /* * By invoking LLE handler here we might get * two events on static LLE entry insertion * in routing socket. However, since we might have * other subscribers we need to generate this event. */ EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_RESOLVED); LLE_WUNLOCK(lle); #ifdef INET /* gratuitous ARP */ if ((laflags & LLE_PUB) && dst->sa_family == AF_INET) arprequest(ifp, &((struct sockaddr_in *)dst)->sin_addr, &((struct sockaddr_in *)dst)->sin_addr, (u_char *)LLADDR(dl)); #endif break; case RTM_DELETE: return (lltable_delete_addr(llt, 0, dst)); default: error = EINVAL; } return (error); } static void vnet_lltable_init() { SLIST_INIT(&V_lltables); } VNET_SYSINIT(vnet_lltable_init, SI_SUB_PSEUDO, SI_ORDER_FIRST, vnet_lltable_init, NULL); #ifdef DDB struct llentry_sa { struct llentry base; struct sockaddr l3_addr; }; static void llatbl_lle_show(struct llentry_sa *la) { struct llentry *lle; uint8_t octet[6]; lle = &la->base; db_printf("lle=%p\n", lle); db_printf(" lle_next=%p\n", lle->lle_next.le_next); db_printf(" lle_lock=%p\n", &lle->lle_lock); db_printf(" lle_tbl=%p\n", lle->lle_tbl); db_printf(" lle_head=%p\n", lle->lle_head); db_printf(" la_hold=%p\n", lle->la_hold); db_printf(" la_numheld=%d\n", lle->la_numheld); db_printf(" la_expire=%ju\n", (uintmax_t)lle->la_expire); db_printf(" la_flags=0x%04x\n", lle->la_flags); db_printf(" la_asked=%u\n", lle->la_asked); db_printf(" la_preempt=%u\n", lle->la_preempt); - db_printf(" ln_byhint=%u\n", lle->ln_byhint); db_printf(" ln_state=%d\n", lle->ln_state); db_printf(" ln_router=%u\n", lle->ln_router); db_printf(" ln_ntick=%ju\n", (uintmax_t)lle->ln_ntick); db_printf(" lle_refcnt=%d\n", lle->lle_refcnt); bcopy(&lle->ll_addr.mac16, octet, sizeof(octet)); db_printf(" ll_addr=%02x:%02x:%02x:%02x:%02x:%02x\n", octet[0], octet[1], octet[2], octet[3], octet[4], octet[5]); db_printf(" lle_timer=%p\n", &lle->lle_timer); switch (la->l3_addr.sa_family) { #ifdef INET case AF_INET: { struct sockaddr_in *sin; char l3s[INET_ADDRSTRLEN]; sin = (struct sockaddr_in *)&la->l3_addr; inet_ntoa_r(sin->sin_addr, l3s); db_printf(" l3_addr=%s\n", l3s); break; } #endif #ifdef INET6 case AF_INET6: { struct sockaddr_in6 *sin6; char l3s[INET6_ADDRSTRLEN]; sin6 = (struct sockaddr_in6 *)&la->l3_addr; ip6_sprintf(l3s, &sin6->sin6_addr); db_printf(" l3_addr=%s\n", l3s); break; } #endif default: db_printf(" l3_addr=N/A (af=%d)\n", la->l3_addr.sa_family); break; } } DB_SHOW_COMMAND(llentry, db_show_llentry) { if (!have_addr) { db_printf("usage: show llentry \n"); return; } llatbl_lle_show((struct llentry_sa *)addr); } static void llatbl_llt_show(struct lltable *llt) { int i; struct llentry *lle; db_printf("llt=%p llt_af=%d llt_ifp=%p\n", llt, llt->llt_af, llt->llt_ifp); for (i = 0; i < llt->llt_hsize; i++) { LIST_FOREACH(lle, &llt->lle_head[i], lle_next) { llatbl_lle_show((struct llentry_sa *)lle); if (db_pager_quit) return; } } } DB_SHOW_COMMAND(lltable, db_show_lltable) { if (!have_addr) { db_printf("usage: show lltable \n"); return; } llatbl_llt_show((struct lltable *)addr); } DB_SHOW_ALL_COMMAND(lltables, db_show_all_lltables) { VNET_ITERATOR_DECL(vnet_iter); struct lltable *llt; VNET_FOREACH(vnet_iter) { CURVNET_SET_QUIET(vnet_iter); #ifdef VIMAGE db_printf("vnet=%p\n", curvnet); #endif SLIST_FOREACH(llt, &V_lltables, llt_link) { db_printf("llt=%p llt_af=%d llt_ifp=%p(%s)\n", llt, llt->llt_af, llt->llt_ifp, (llt->llt_ifp != NULL) ? llt->llt_ifp->if_xname : "?"); if (have_addr && addr != 0) /* verbose */ llatbl_llt_show(llt); if (db_pager_quit) { CURVNET_RESTORE(); return; } } CURVNET_RESTORE(); } } #endif Index: head/sys/net/if_llatbl.h =================================================================== --- head/sys/net/if_llatbl.h (revision 288300) +++ head/sys/net/if_llatbl.h (revision 288301) @@ -1,246 +1,245 @@ /* * Copyright (c) 2004 Luigi Rizzo, Alessandro Cerri. All rights reserved. * Copyright (c) 2004-2008 Qing Li. All rights reserved. * Copyright (c) 2008 Kip Macy. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #ifndef _NET_IF_LLATBL_H_ #define _NET_IF_LLATBL_H_ #include #include struct ifnet; struct sysctl_req; struct rt_msghdr; struct rt_addrinfo; struct llentry; LIST_HEAD(llentries, llentry); extern struct rwlock lltable_rwlock; #define LLTABLE_RLOCK() rw_rlock(&lltable_rwlock) #define LLTABLE_RUNLOCK() rw_runlock(&lltable_rwlock) #define LLTABLE_WLOCK() rw_wlock(&lltable_rwlock) #define LLTABLE_WUNLOCK() rw_wunlock(&lltable_rwlock) #define LLTABLE_LOCK_ASSERT() rw_assert(&lltable_rwlock, RA_LOCKED) /* * Code referencing llentry must at least hold * a shared lock */ struct llentry { LIST_ENTRY(llentry) lle_next; union { struct in_addr addr4; struct in6_addr addr6; } r_l3addr; union { uint64_t mac_aligned; uint16_t mac16[3]; uint8_t mac8[20]; /* IB needs 20 bytes. */ } ll_addr; uint32_t spare0; uint64_t spare1; struct lltable *lle_tbl; struct llentries *lle_head; void (*lle_free)(struct llentry *); struct mbuf *la_hold; int la_numheld; /* # of packets currently held */ time_t la_expire; uint16_t la_flags; uint16_t la_asked; uint16_t la_preempt; - uint16_t ln_byhint; int16_t ln_state; /* IPv6 has ND6_LLINFO_NOSTATE == -2 */ uint16_t ln_router; time_t ln_ntick; int lle_refcnt; LIST_ENTRY(llentry) lle_chain; /* chain of deleted items */ struct callout lle_timer; struct rwlock lle_lock; }; #define LLE_WLOCK(lle) rw_wlock(&(lle)->lle_lock) #define LLE_RLOCK(lle) rw_rlock(&(lle)->lle_lock) #define LLE_WUNLOCK(lle) rw_wunlock(&(lle)->lle_lock) #define LLE_RUNLOCK(lle) rw_runlock(&(lle)->lle_lock) #define LLE_DOWNGRADE(lle) rw_downgrade(&(lle)->lle_lock) #define LLE_TRY_UPGRADE(lle) rw_try_upgrade(&(lle)->lle_lock) #define LLE_LOCK_INIT(lle) rw_init_flags(&(lle)->lle_lock, "lle", RW_DUPOK) #define LLE_LOCK_DESTROY(lle) rw_destroy(&(lle)->lle_lock) #define LLE_WLOCK_ASSERT(lle) rw_assert(&(lle)->lle_lock, RA_WLOCKED) #define LLE_IS_VALID(lle) (((lle) != NULL) && ((lle) != (void *)-1)) #define LLE_ADDREF(lle) do { \ LLE_WLOCK_ASSERT(lle); \ KASSERT((lle)->lle_refcnt >= 0, \ ("negative refcnt %d on lle %p", \ (lle)->lle_refcnt, (lle))); \ (lle)->lle_refcnt++; \ } while (0) #define LLE_REMREF(lle) do { \ LLE_WLOCK_ASSERT(lle); \ KASSERT((lle)->lle_refcnt > 0, \ ("bogus refcnt %d on lle %p", \ (lle)->lle_refcnt, (lle))); \ (lle)->lle_refcnt--; \ } while (0) #define LLE_FREE_LOCKED(lle) do { \ if ((lle)->lle_refcnt == 1) \ (lle)->lle_free(lle); \ else { \ LLE_REMREF(lle); \ LLE_WUNLOCK(lle); \ } \ /* guard against invalid refs */ \ (lle) = NULL; \ } while (0) #define LLE_FREE(lle) do { \ LLE_WLOCK(lle); \ LLE_FREE_LOCKED(lle); \ } while (0) typedef struct llentry *(llt_lookup_t)(struct lltable *, u_int flags, const struct sockaddr *l3addr); typedef struct llentry *(llt_alloc_t)(struct lltable *, u_int flags, const struct sockaddr *l3addr); typedef void (llt_delete_t)(struct lltable *, struct llentry *); typedef void (llt_prefix_free_t)(struct lltable *, const struct sockaddr *addr, const struct sockaddr *mask, u_int flags); typedef int (llt_dump_entry_t)(struct lltable *, struct llentry *, struct sysctl_req *); typedef uint32_t (llt_hash_t)(const struct llentry *, uint32_t); typedef int (llt_match_prefix_t)(const struct sockaddr *, const struct sockaddr *, u_int, struct llentry *); typedef void (llt_free_entry_t)(struct lltable *, struct llentry *); typedef void (llt_fill_sa_entry_t)(const struct llentry *, struct sockaddr *); typedef void (llt_free_tbl_t)(struct lltable *); typedef void (llt_link_entry_t)(struct lltable *, struct llentry *); typedef void (llt_unlink_entry_t)(struct llentry *); typedef int (llt_foreach_cb_t)(struct lltable *, struct llentry *, void *); typedef int (llt_foreach_entry_t)(struct lltable *, llt_foreach_cb_t *, void *); struct lltable { SLIST_ENTRY(lltable) llt_link; int llt_af; int llt_hsize; struct llentries *lle_head; struct ifnet *llt_ifp; llt_lookup_t *llt_lookup; llt_alloc_t *llt_alloc_entry; llt_delete_t *llt_delete_entry; llt_prefix_free_t *llt_prefix_free; llt_dump_entry_t *llt_dump_entry; llt_hash_t *llt_hash; llt_match_prefix_t *llt_match_prefix; llt_free_entry_t *llt_free_entry; llt_foreach_entry_t *llt_foreach_entry; llt_link_entry_t *llt_link_entry; llt_unlink_entry_t *llt_unlink_entry; llt_fill_sa_entry_t *llt_fill_sa_entry; llt_free_tbl_t *llt_free_tbl; }; MALLOC_DECLARE(M_LLTABLE); /* * LLentry flags */ #define LLE_DELETED 0x0001 /* entry must be deleted */ #define LLE_STATIC 0x0002 /* entry is static */ #define LLE_IFADDR 0x0004 /* entry is interface addr */ #define LLE_VALID 0x0008 /* ll_addr is valid */ #define LLE_REDIRECT 0x0010 /* installed by redirect; has host rtentry */ #define LLE_PUB 0x0020 /* publish entry ??? */ #define LLE_LINKED 0x0040 /* linked to lookup structure */ /* LLE request flags */ #define LLE_EXCLUSIVE 0x2000 /* return lle xlocked */ #define LLATBL_HASH(key, mask) \ (((((((key >> 8) ^ key) >> 8) ^ key) >> 8) ^ key) & mask) struct lltable *lltable_allocate_htbl(uint32_t hsize); void lltable_free(struct lltable *); void lltable_link(struct lltable *llt); void lltable_prefix_free(int, struct sockaddr *, struct sockaddr *, u_int); #if 0 void lltable_drain(int); #endif int lltable_sysctl_dumparp(int, struct sysctl_req *); size_t llentry_free(struct llentry *); struct llentry *llentry_alloc(struct ifnet *, struct lltable *, struct sockaddr_storage *); /* helper functions */ size_t lltable_drop_entry_queue(struct llentry *); struct llentry *lltable_alloc_entry(struct lltable *llt, u_int flags, const struct sockaddr *l4addr); void lltable_free_entry(struct lltable *llt, struct llentry *lle); int lltable_delete_addr(struct lltable *llt, u_int flags, const struct sockaddr *l3addr); void lltable_link_entry(struct lltable *llt, struct llentry *lle); void lltable_unlink_entry(struct lltable *llt, struct llentry *lle); void lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa); struct ifnet *lltable_get_ifp(const struct lltable *llt); int lltable_get_af(const struct lltable *llt); int lltable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg); /* * Generic link layer address lookup function. */ static __inline struct llentry * lla_lookup(struct lltable *llt, u_int flags, const struct sockaddr *l3addr) { return (llt->llt_lookup(llt, flags, l3addr)); } int lla_rt_output(struct rt_msghdr *, struct rt_addrinfo *); #include enum { LLENTRY_RESOLVED, LLENTRY_TIMEDOUT, LLENTRY_DELETED, LLENTRY_EXPIRED, }; typedef void (*lle_event_fn)(void *, struct llentry *, int); EVENTHANDLER_DECLARE(lle_event, lle_event_fn); #endif /* _NET_IF_LLATBL_H_ */ Index: head/sys/netinet/tcp_input.c =================================================================== --- head/sys/netinet/tcp_input.c (revision 288300) +++ head/sys/netinet/tcp_input.c (revision 288301) @@ -1,3730 +1,3715 @@ /*- * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1994, 1995 * The Regents of the University of California. All rights reserved. * Copyright (c) 2007-2008,2010 * Swinburne University of Technology, Melbourne, Australia. * Copyright (c) 2009-2010 Lawrence Stewart * Copyright (c) 2010 The FreeBSD Foundation * Copyright (c) 2010-2011 Juniper Networks, Inc. * All rights reserved. * * Portions of this software were developed at the Centre for Advanced Internet * Architectures, Swinburne University of Technology, by Lawrence Stewart, * James Healy and David Hayes, made possible in part by a grant from the Cisco * University Research Program Fund at Community Foundation Silicon Valley. * * Portions of this software were developed at the Centre for Advanced * Internet Architectures, Swinburne University of Technology, Melbourne, * Australia by David Hayes under sponsorship from the FreeBSD Foundation. * * Portions of this software were developed by Robert N. M. Watson under * contract to Juniper Networks, Inc. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)tcp_input.c 8.12 (Berkeley) 5/24/95 */ #include __FBSDID("$FreeBSD$"); #include "opt_ipfw.h" /* for ipfw_fwd */ #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" #include "opt_tcpdebug.h" #include #include #include #include #include #include /* for proc0 declaration */ #include #include #include #include #include #include #include #include #include /* before tcp_seq.h, for tcp_random18() */ #include #include #include #include #include #define TCPSTATES /* for logging */ #include #include #include #include #include #include #include #include /* required for icmp_var.h */ #include /* for ICMP_BANDLIM */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef TCPDEBUG #include #endif /* TCPDEBUG */ #ifdef TCP_OFFLOAD #include #endif #ifdef IPSEC #include #include #endif /*IPSEC*/ #include #include const int tcprexmtthresh = 3; int tcp_log_in_vain = 0; SYSCTL_INT(_net_inet_tcp, OID_AUTO, log_in_vain, CTLFLAG_RW, &tcp_log_in_vain, 0, "Log all incoming TCP segments to closed ports"); VNET_DEFINE(int, blackhole) = 0; #define V_blackhole VNET(blackhole) SYSCTL_INT(_net_inet_tcp, OID_AUTO, blackhole, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(blackhole), 0, "Do not send RST on segments to closed ports"); VNET_DEFINE(int, tcp_delack_enabled) = 1; SYSCTL_INT(_net_inet_tcp, OID_AUTO, delayed_ack, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_delack_enabled), 0, "Delay ACK to try and piggyback it onto a data packet"); VNET_DEFINE(int, drop_synfin) = 0; #define V_drop_synfin VNET(drop_synfin) SYSCTL_INT(_net_inet_tcp, OID_AUTO, drop_synfin, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(drop_synfin), 0, "Drop TCP packets with SYN+FIN set"); VNET_DEFINE(int, tcp_do_rfc3042) = 1; #define V_tcp_do_rfc3042 VNET(tcp_do_rfc3042) SYSCTL_INT(_net_inet_tcp, OID_AUTO, rfc3042, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_do_rfc3042), 0, "Enable RFC 3042 (Limited Transmit)"); VNET_DEFINE(int, tcp_do_rfc3390) = 1; SYSCTL_INT(_net_inet_tcp, OID_AUTO, rfc3390, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_do_rfc3390), 0, "Enable RFC 3390 (Increasing TCP's Initial Congestion Window)"); SYSCTL_NODE(_net_inet_tcp, OID_AUTO, experimental, CTLFLAG_RW, 0, "Experimental TCP extensions"); VNET_DEFINE(int, tcp_do_initcwnd10) = 1; SYSCTL_INT(_net_inet_tcp_experimental, OID_AUTO, initcwnd10, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_do_initcwnd10), 0, "Enable RFC 6928 (Increasing initial CWND to 10)"); VNET_DEFINE(int, tcp_do_rfc3465) = 1; SYSCTL_INT(_net_inet_tcp, OID_AUTO, rfc3465, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_do_rfc3465), 0, "Enable RFC 3465 (Appropriate Byte Counting)"); VNET_DEFINE(int, tcp_abc_l_var) = 2; SYSCTL_INT(_net_inet_tcp, OID_AUTO, abc_l_var, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_abc_l_var), 2, "Cap the max cwnd increment during slow-start to this number of segments"); static SYSCTL_NODE(_net_inet_tcp, OID_AUTO, ecn, CTLFLAG_RW, 0, "TCP ECN"); VNET_DEFINE(int, tcp_do_ecn) = 0; SYSCTL_INT(_net_inet_tcp_ecn, OID_AUTO, enable, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_do_ecn), 0, "TCP ECN support"); VNET_DEFINE(int, tcp_ecn_maxretries) = 1; SYSCTL_INT(_net_inet_tcp_ecn, OID_AUTO, maxretries, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_ecn_maxretries), 0, "Max retries before giving up on ECN"); VNET_DEFINE(int, tcp_insecure_syn) = 0; #define V_tcp_insecure_syn VNET(tcp_insecure_syn) SYSCTL_INT(_net_inet_tcp, OID_AUTO, insecure_syn, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_insecure_syn), 0, "Follow RFC793 instead of RFC5961 criteria for accepting SYN packets"); VNET_DEFINE(int, tcp_insecure_rst) = 0; #define V_tcp_insecure_rst VNET(tcp_insecure_rst) SYSCTL_INT(_net_inet_tcp, OID_AUTO, insecure_rst, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_insecure_rst), 0, "Follow RFC793 instead of RFC5961 criteria for accepting RST packets"); VNET_DEFINE(int, tcp_recvspace) = 1024*64; #define V_tcp_recvspace VNET(tcp_recvspace) SYSCTL_INT(_net_inet_tcp, TCPCTL_RECVSPACE, recvspace, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_recvspace), 0, "Initial receive socket buffer size"); VNET_DEFINE(int, tcp_do_autorcvbuf) = 1; #define V_tcp_do_autorcvbuf VNET(tcp_do_autorcvbuf) SYSCTL_INT(_net_inet_tcp, OID_AUTO, recvbuf_auto, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_do_autorcvbuf), 0, "Enable automatic receive buffer sizing"); VNET_DEFINE(int, tcp_autorcvbuf_inc) = 16*1024; #define V_tcp_autorcvbuf_inc VNET(tcp_autorcvbuf_inc) SYSCTL_INT(_net_inet_tcp, OID_AUTO, recvbuf_inc, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_autorcvbuf_inc), 0, "Incrementor step size of automatic receive buffer"); VNET_DEFINE(int, tcp_autorcvbuf_max) = 2*1024*1024; #define V_tcp_autorcvbuf_max VNET(tcp_autorcvbuf_max) SYSCTL_INT(_net_inet_tcp, OID_AUTO, recvbuf_max, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_autorcvbuf_max), 0, "Max size of automatic receive buffer"); VNET_DEFINE(struct inpcbhead, tcb); #define tcb6 tcb /* for KAME src sync over BSD*'s */ VNET_DEFINE(struct inpcbinfo, tcbinfo); static void tcp_dooptions(struct tcpopt *, u_char *, int, int); static void tcp_do_segment(struct mbuf *, struct tcphdr *, struct socket *, struct tcpcb *, int, int, uint8_t, int); static void tcp_dropwithreset(struct mbuf *, struct tcphdr *, struct tcpcb *, int, int); static void tcp_pulloutofband(struct socket *, struct tcphdr *, struct mbuf *, int); static void tcp_xmit_timer(struct tcpcb *, int); static void tcp_newreno_partial_ack(struct tcpcb *, struct tcphdr *); static void inline cc_ack_received(struct tcpcb *tp, struct tcphdr *th, uint16_t type); static void inline cc_conn_init(struct tcpcb *tp); static void inline cc_post_recovery(struct tcpcb *tp, struct tcphdr *th); static void inline hhook_run_tcp_est_in(struct tcpcb *tp, struct tcphdr *th, struct tcpopt *to); /* * TCP statistics are stored in an "array" of counter(9)s. */ VNET_PCPUSTAT_DEFINE(struct tcpstat, tcpstat); VNET_PCPUSTAT_SYSINIT(tcpstat); SYSCTL_VNET_PCPUSTAT(_net_inet_tcp, TCPCTL_STATS, stats, struct tcpstat, tcpstat, "TCP statistics (struct tcpstat, netinet/tcp_var.h)"); #ifdef VIMAGE VNET_PCPUSTAT_SYSUNINIT(tcpstat); #endif /* VIMAGE */ /* * Kernel module interface for updating tcpstat. The argument is an index * into tcpstat treated as an array. */ void kmod_tcpstat_inc(int statnum) { counter_u64_add(VNET(tcpstat)[statnum], 1); } /* * Wrapper for the TCP established input helper hook. */ static void inline hhook_run_tcp_est_in(struct tcpcb *tp, struct tcphdr *th, struct tcpopt *to) { struct tcp_hhook_data hhook_data; if (V_tcp_hhh[HHOOK_TCP_EST_IN]->hhh_nhooks > 0) { hhook_data.tp = tp; hhook_data.th = th; hhook_data.to = to; hhook_run_hooks(V_tcp_hhh[HHOOK_TCP_EST_IN], &hhook_data, tp->osd); } } /* * CC wrapper hook functions */ static void inline cc_ack_received(struct tcpcb *tp, struct tcphdr *th, uint16_t type) { INP_WLOCK_ASSERT(tp->t_inpcb); tp->ccv->bytes_this_ack = BYTES_THIS_ACK(tp, th); if (tp->snd_cwnd <= tp->snd_wnd) tp->ccv->flags |= CCF_CWND_LIMITED; else tp->ccv->flags &= ~CCF_CWND_LIMITED; if (type == CC_ACK) { if (tp->snd_cwnd > tp->snd_ssthresh) { tp->t_bytes_acked += min(tp->ccv->bytes_this_ack, V_tcp_abc_l_var * tp->t_maxseg); if (tp->t_bytes_acked >= tp->snd_cwnd) { tp->t_bytes_acked -= tp->snd_cwnd; tp->ccv->flags |= CCF_ABC_SENTAWND; } } else { tp->ccv->flags &= ~CCF_ABC_SENTAWND; tp->t_bytes_acked = 0; } } if (CC_ALGO(tp)->ack_received != NULL) { /* XXXLAS: Find a way to live without this */ tp->ccv->curack = th->th_ack; CC_ALGO(tp)->ack_received(tp->ccv, type); } } static void inline cc_conn_init(struct tcpcb *tp) { struct hc_metrics_lite metrics; struct inpcb *inp = tp->t_inpcb; int rtt; INP_WLOCK_ASSERT(tp->t_inpcb); tcp_hc_get(&inp->inp_inc, &metrics); if (tp->t_srtt == 0 && (rtt = metrics.rmx_rtt)) { tp->t_srtt = rtt; tp->t_rttbest = tp->t_srtt + TCP_RTT_SCALE; TCPSTAT_INC(tcps_usedrtt); if (metrics.rmx_rttvar) { tp->t_rttvar = metrics.rmx_rttvar; TCPSTAT_INC(tcps_usedrttvar); } else { /* default variation is +- 1 rtt */ tp->t_rttvar = tp->t_srtt * TCP_RTTVAR_SCALE / TCP_RTT_SCALE; } TCPT_RANGESET(tp->t_rxtcur, ((tp->t_srtt >> 2) + tp->t_rttvar) >> 1, tp->t_rttmin, TCPTV_REXMTMAX); } if (metrics.rmx_ssthresh) { /* * There's some sort of gateway or interface * buffer limit on the path. Use this to set * the slow start threshhold, but set the * threshold to no less than 2*mss. */ tp->snd_ssthresh = max(2 * tp->t_maxseg, metrics.rmx_ssthresh); TCPSTAT_INC(tcps_usedssthresh); } /* * Set the initial slow-start flight size. * * RFC5681 Section 3.1 specifies the default conservative values. * RFC3390 specifies slightly more aggressive values. * RFC6928 increases it to ten segments. * * If a SYN or SYN/ACK was lost and retransmitted, we have to * reduce the initial CWND to one segment as congestion is likely * requiring us to be cautious. */ if (tp->snd_cwnd == 1) tp->snd_cwnd = tp->t_maxseg; /* SYN(-ACK) lost */ else if (V_tcp_do_initcwnd10) tp->snd_cwnd = min(10 * tp->t_maxseg, max(2 * tp->t_maxseg, 14600)); else if (V_tcp_do_rfc3390) tp->snd_cwnd = min(4 * tp->t_maxseg, max(2 * tp->t_maxseg, 4380)); else { /* Per RFC5681 Section 3.1 */ if (tp->t_maxseg > 2190) tp->snd_cwnd = 2 * tp->t_maxseg; else if (tp->t_maxseg > 1095) tp->snd_cwnd = 3 * tp->t_maxseg; else tp->snd_cwnd = 4 * tp->t_maxseg; } if (CC_ALGO(tp)->conn_init != NULL) CC_ALGO(tp)->conn_init(tp->ccv); } void inline cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type) { INP_WLOCK_ASSERT(tp->t_inpcb); switch(type) { case CC_NDUPACK: if (!IN_FASTRECOVERY(tp->t_flags)) { tp->snd_recover = tp->snd_max; if (tp->t_flags & TF_ECN_PERMIT) tp->t_flags |= TF_ECN_SND_CWR; } break; case CC_ECN: if (!IN_CONGRECOVERY(tp->t_flags)) { TCPSTAT_INC(tcps_ecn_rcwnd); tp->snd_recover = tp->snd_max; if (tp->t_flags & TF_ECN_PERMIT) tp->t_flags |= TF_ECN_SND_CWR; } break; case CC_RTO: tp->t_dupacks = 0; tp->t_bytes_acked = 0; EXIT_RECOVERY(tp->t_flags); tp->snd_ssthresh = max(2, min(tp->snd_wnd, tp->snd_cwnd) / 2 / tp->t_maxseg) * tp->t_maxseg; tp->snd_cwnd = tp->t_maxseg; break; case CC_RTO_ERR: TCPSTAT_INC(tcps_sndrexmitbad); /* RTO was unnecessary, so reset everything. */ tp->snd_cwnd = tp->snd_cwnd_prev; tp->snd_ssthresh = tp->snd_ssthresh_prev; tp->snd_recover = tp->snd_recover_prev; if (tp->t_flags & TF_WASFRECOVERY) ENTER_FASTRECOVERY(tp->t_flags); if (tp->t_flags & TF_WASCRECOVERY) ENTER_CONGRECOVERY(tp->t_flags); tp->snd_nxt = tp->snd_max; tp->t_flags &= ~TF_PREVVALID; tp->t_badrxtwin = 0; break; } if (CC_ALGO(tp)->cong_signal != NULL) { if (th != NULL) tp->ccv->curack = th->th_ack; CC_ALGO(tp)->cong_signal(tp->ccv, type); } } static void inline cc_post_recovery(struct tcpcb *tp, struct tcphdr *th) { INP_WLOCK_ASSERT(tp->t_inpcb); /* XXXLAS: KASSERT that we're in recovery? */ if (CC_ALGO(tp)->post_recovery != NULL) { tp->ccv->curack = th->th_ack; CC_ALGO(tp)->post_recovery(tp->ccv); } /* XXXLAS: EXIT_RECOVERY ? */ tp->t_bytes_acked = 0; } #ifdef TCP_SIGNATURE static inline int tcp_signature_verify_input(struct mbuf *m, int off0, int tlen, int optlen, struct tcpopt *to, struct tcphdr *th, u_int tcpbflag) { int ret; tcp_fields_to_net(th); ret = tcp_signature_verify(m, off0, tlen, optlen, to, th, tcpbflag); tcp_fields_to_host(th); return (ret); } #endif -/* Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. */ -#ifdef INET6 -#define ND6_HINT(tp) \ -do { \ - if ((tp) && (tp)->t_inpcb && \ - ((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0) \ - nd6_nud_hint(NULL, NULL, 0); \ -} while (0) -#else -#define ND6_HINT(tp) -#endif - /* * Indicate whether this ack should be delayed. We can delay the ack if * following conditions are met: * - There is no delayed ack timer in progress. * - Our last ack wasn't a 0-sized window. We never want to delay * the ack that opens up a 0-sized window. * - LRO wasn't used for this segment. We make sure by checking that the * segment size is not larger than the MSS. * - Delayed acks are enabled or this is a half-synchronized T/TCP * connection. */ #define DELAY_ACK(tp, tlen) \ ((!tcp_timer_active(tp, TT_DELACK) && \ (tp->t_flags & TF_RXWIN0SENT) == 0) && \ (tlen <= tp->t_maxopd) && \ (V_tcp_delack_enabled || (tp->t_flags & TF_NEEDSYN))) static void inline cc_ecnpkt_handler(struct tcpcb *tp, struct tcphdr *th, uint8_t iptos) { INP_WLOCK_ASSERT(tp->t_inpcb); if (CC_ALGO(tp)->ecnpkt_handler != NULL) { switch (iptos & IPTOS_ECN_MASK) { case IPTOS_ECN_CE: tp->ccv->flags |= CCF_IPHDR_CE; break; case IPTOS_ECN_ECT0: tp->ccv->flags &= ~CCF_IPHDR_CE; break; case IPTOS_ECN_ECT1: tp->ccv->flags &= ~CCF_IPHDR_CE; break; } if (th->th_flags & TH_CWR) tp->ccv->flags |= CCF_TCPHDR_CWR; else tp->ccv->flags &= ~CCF_TCPHDR_CWR; if (tp->t_flags & TF_DELACK) tp->ccv->flags |= CCF_DELACK; else tp->ccv->flags &= ~CCF_DELACK; CC_ALGO(tp)->ecnpkt_handler(tp->ccv); if (tp->ccv->flags & CCF_ACKNOW) tcp_timer_activate(tp, TT_DELACK, tcp_delacktime); } } /* * TCP input handling is split into multiple parts: * tcp6_input is a thin wrapper around tcp_input for the extended * ip6_protox[] call format in ip6_input * tcp_input handles primary segment validation, inpcb lookup and * SYN processing on listen sockets * tcp_do_segment processes the ACK and text of the segment for * establishing, established and closing connections */ #ifdef INET6 int tcp6_input(struct mbuf **mp, int *offp, int proto) { struct mbuf *m = *mp; struct in6_ifaddr *ia6; struct ip6_hdr *ip6; IP6_EXTHDR_CHECK(m, *offp, sizeof(struct tcphdr), IPPROTO_DONE); /* * draft-itojun-ipv6-tcp-to-anycast * better place to put this in? */ ip6 = mtod(m, struct ip6_hdr *); ia6 = in6ifa_ifwithaddr(&ip6->ip6_dst, 0 /* XXX */); if (ia6 && (ia6->ia6_flags & IN6_IFF_ANYCAST)) { struct ip6_hdr *ip6; ifa_free(&ia6->ia_ifa); ip6 = mtod(m, struct ip6_hdr *); icmp6_error(m, ICMP6_DST_UNREACH, ICMP6_DST_UNREACH_ADDR, (caddr_t)&ip6->ip6_dst - (caddr_t)ip6); return (IPPROTO_DONE); } if (ia6) ifa_free(&ia6->ia_ifa); return (tcp_input(mp, offp, proto)); } #endif /* INET6 */ int tcp_input(struct mbuf **mp, int *offp, int proto) { struct mbuf *m = *mp; struct tcphdr *th = NULL; struct ip *ip = NULL; struct inpcb *inp = NULL; struct tcpcb *tp = NULL; struct socket *so = NULL; u_char *optp = NULL; int off0; int optlen = 0; #ifdef INET int len; #endif int tlen = 0, off; int drop_hdrlen; int thflags; int rstreason = 0; /* For badport_bandlim accounting purposes */ #ifdef TCP_SIGNATURE uint8_t sig_checked = 0; #endif uint8_t iptos = 0; struct m_tag *fwd_tag = NULL; #ifdef INET6 struct ip6_hdr *ip6 = NULL; int isipv6; #else const void *ip6 = NULL; #endif /* INET6 */ struct tcpopt to; /* options in this segment */ char *s = NULL; /* address and port logging */ int ti_locked; #define TI_UNLOCKED 1 #define TI_RLOCKED 2 #ifdef TCPDEBUG /* * The size of tcp_saveipgen must be the size of the max ip header, * now IPv6. */ u_char tcp_saveipgen[IP6_HDR_LEN]; struct tcphdr tcp_savetcp; short ostate = 0; #endif #ifdef INET6 isipv6 = (mtod(m, struct ip *)->ip_v == 6) ? 1 : 0; #endif off0 = *offp; m = *mp; *mp = NULL; to.to_flags = 0; TCPSTAT_INC(tcps_rcvtotal); #ifdef INET6 if (isipv6) { /* IP6_EXTHDR_CHECK() is already done at tcp6_input(). */ if (m->m_len < (sizeof(*ip6) + sizeof(*th))) { m = m_pullup(m, sizeof(*ip6) + sizeof(*th)); if (m == NULL) { TCPSTAT_INC(tcps_rcvshort); return (IPPROTO_DONE); } } ip6 = mtod(m, struct ip6_hdr *); th = (struct tcphdr *)((caddr_t)ip6 + off0); tlen = sizeof(*ip6) + ntohs(ip6->ip6_plen) - off0; if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID_IPV6) { if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR) th->th_sum = m->m_pkthdr.csum_data; else th->th_sum = in6_cksum_pseudo(ip6, tlen, IPPROTO_TCP, m->m_pkthdr.csum_data); th->th_sum ^= 0xffff; } else th->th_sum = in6_cksum(m, IPPROTO_TCP, off0, tlen); if (th->th_sum) { TCPSTAT_INC(tcps_rcvbadsum); goto drop; } /* * Be proactive about unspecified IPv6 address in source. * As we use all-zero to indicate unbounded/unconnected pcb, * unspecified IPv6 address can be used to confuse us. * * Note that packets with unspecified IPv6 destination is * already dropped in ip6_input. */ if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) { /* XXX stat */ goto drop; } } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { /* * Get IP and TCP header together in first mbuf. * Note: IP leaves IP header in first mbuf. */ if (off0 > sizeof (struct ip)) { ip_stripoptions(m); off0 = sizeof(struct ip); } if (m->m_len < sizeof (struct tcpiphdr)) { if ((m = m_pullup(m, sizeof (struct tcpiphdr))) == NULL) { TCPSTAT_INC(tcps_rcvshort); return (IPPROTO_DONE); } } ip = mtod(m, struct ip *); th = (struct tcphdr *)((caddr_t)ip + off0); tlen = ntohs(ip->ip_len) - off0; if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) { if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR) th->th_sum = m->m_pkthdr.csum_data; else th->th_sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htonl(m->m_pkthdr.csum_data + tlen + IPPROTO_TCP)); th->th_sum ^= 0xffff; } else { struct ipovly *ipov = (struct ipovly *)ip; /* * Checksum extended TCP header and data. */ len = off0 + tlen; bzero(ipov->ih_x1, sizeof(ipov->ih_x1)); ipov->ih_len = htons(tlen); th->th_sum = in_cksum(m, len); /* Reset length for SDT probes. */ ip->ip_len = htons(tlen + off0); } if (th->th_sum) { TCPSTAT_INC(tcps_rcvbadsum); goto drop; } /* Re-initialization for later version check */ ip->ip_v = IPVERSION; } #endif /* INET */ #ifdef INET6 if (isipv6) iptos = (ntohl(ip6->ip6_flow) >> 20) & 0xff; #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET iptos = ip->ip_tos; #endif /* * Check that TCP offset makes sense, * pull out TCP options and adjust length. XXX */ off = th->th_off << 2; if (off < sizeof (struct tcphdr) || off > tlen) { TCPSTAT_INC(tcps_rcvbadoff); goto drop; } tlen -= off; /* tlen is used instead of ti->ti_len */ if (off > sizeof (struct tcphdr)) { #ifdef INET6 if (isipv6) { IP6_EXTHDR_CHECK(m, off0, off, IPPROTO_DONE); ip6 = mtod(m, struct ip6_hdr *); th = (struct tcphdr *)((caddr_t)ip6 + off0); } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { if (m->m_len < sizeof(struct ip) + off) { if ((m = m_pullup(m, sizeof (struct ip) + off)) == NULL) { TCPSTAT_INC(tcps_rcvshort); return (IPPROTO_DONE); } ip = mtod(m, struct ip *); th = (struct tcphdr *)((caddr_t)ip + off0); } } #endif optlen = off - sizeof (struct tcphdr); optp = (u_char *)(th + 1); } thflags = th->th_flags; /* * Convert TCP protocol specific fields to host format. */ tcp_fields_to_host(th); /* * Delay dropping TCP, IP headers, IPv6 ext headers, and TCP options. */ drop_hdrlen = off0 + off; /* * Locate pcb for segment; if we're likely to add or remove a * connection then first acquire pcbinfo lock. There are three cases * where we might discover later we need a write lock despite the * flags: ACKs moving a connection out of the syncache, ACKs for a * connection in TIMEWAIT and SYNs not targeting a listening socket. */ if ((thflags & (TH_FIN | TH_RST)) != 0) { INP_INFO_RLOCK(&V_tcbinfo); ti_locked = TI_RLOCKED; } else ti_locked = TI_UNLOCKED; /* * Grab info from PACKET_TAG_IPFORWARD tag prepended to the chain. */ if ( #ifdef INET6 (isipv6 && (m->m_flags & M_IP6_NEXTHOP)) #ifdef INET || (!isipv6 && (m->m_flags & M_IP_NEXTHOP)) #endif #endif #if defined(INET) && !defined(INET6) (m->m_flags & M_IP_NEXTHOP) #endif ) fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL); findpcb: #ifdef INVARIANTS if (ti_locked == TI_RLOCKED) { INP_INFO_RLOCK_ASSERT(&V_tcbinfo); } else { INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); } #endif #ifdef INET6 if (isipv6 && fwd_tag != NULL) { struct sockaddr_in6 *next_hop6; next_hop6 = (struct sockaddr_in6 *)(fwd_tag + 1); /* * Transparently forwarded. Pretend to be the destination. * Already got one like this? */ inp = in6_pcblookup_mbuf(&V_tcbinfo, &ip6->ip6_src, th->th_sport, &ip6->ip6_dst, th->th_dport, INPLOOKUP_WLOCKPCB, m->m_pkthdr.rcvif, m); if (!inp) { /* * It's new. Try to find the ambushing socket. * Because we've rewritten the destination address, * any hardware-generated hash is ignored. */ inp = in6_pcblookup(&V_tcbinfo, &ip6->ip6_src, th->th_sport, &next_hop6->sin6_addr, next_hop6->sin6_port ? ntohs(next_hop6->sin6_port) : th->th_dport, INPLOOKUP_WILDCARD | INPLOOKUP_WLOCKPCB, m->m_pkthdr.rcvif); } } else if (isipv6) { inp = in6_pcblookup_mbuf(&V_tcbinfo, &ip6->ip6_src, th->th_sport, &ip6->ip6_dst, th->th_dport, INPLOOKUP_WILDCARD | INPLOOKUP_WLOCKPCB, m->m_pkthdr.rcvif, m); } #endif /* INET6 */ #if defined(INET6) && defined(INET) else #endif #ifdef INET if (fwd_tag != NULL) { struct sockaddr_in *next_hop; next_hop = (struct sockaddr_in *)(fwd_tag+1); /* * Transparently forwarded. Pretend to be the destination. * already got one like this? */ inp = in_pcblookup_mbuf(&V_tcbinfo, ip->ip_src, th->th_sport, ip->ip_dst, th->th_dport, INPLOOKUP_WLOCKPCB, m->m_pkthdr.rcvif, m); if (!inp) { /* * It's new. Try to find the ambushing socket. * Because we've rewritten the destination address, * any hardware-generated hash is ignored. */ inp = in_pcblookup(&V_tcbinfo, ip->ip_src, th->th_sport, next_hop->sin_addr, next_hop->sin_port ? ntohs(next_hop->sin_port) : th->th_dport, INPLOOKUP_WILDCARD | INPLOOKUP_WLOCKPCB, m->m_pkthdr.rcvif); } } else inp = in_pcblookup_mbuf(&V_tcbinfo, ip->ip_src, th->th_sport, ip->ip_dst, th->th_dport, INPLOOKUP_WILDCARD | INPLOOKUP_WLOCKPCB, m->m_pkthdr.rcvif, m); #endif /* INET */ /* * If the INPCB does not exist then all data in the incoming * segment is discarded and an appropriate RST is sent back. * XXX MRT Send RST using which routing table? */ if (inp == NULL) { /* * Log communication attempts to ports that are not * in use. */ if ((tcp_log_in_vain == 1 && (thflags & TH_SYN)) || tcp_log_in_vain == 2) { if ((s = tcp_log_vain(NULL, th, (void *)ip, ip6))) log(LOG_INFO, "%s; %s: Connection attempt " "to closed port\n", s, __func__); } /* * When blackholing do not respond with a RST but * completely ignore the segment and drop it. */ if ((V_blackhole == 1 && (thflags & TH_SYN)) || V_blackhole == 2) goto dropunlock; rstreason = BANDLIM_RST_CLOSEDPORT; goto dropwithreset; } INP_WLOCK_ASSERT(inp); if ((inp->inp_flowtype == M_HASHTYPE_NONE) && (M_HASHTYPE_GET(m) != M_HASHTYPE_NONE) && ((inp->inp_socket == NULL) || (inp->inp_socket->so_options & SO_ACCEPTCONN) == 0)) { inp->inp_flowid = m->m_pkthdr.flowid; inp->inp_flowtype = M_HASHTYPE_GET(m); } #ifdef IPSEC #ifdef INET6 if (isipv6 && ipsec6_in_reject(m, inp)) { goto dropunlock; } else #endif /* INET6 */ if (ipsec4_in_reject(m, inp) != 0) { goto dropunlock; } #endif /* IPSEC */ /* * Check the minimum TTL for socket. */ if (inp->inp_ip_minttl != 0) { #ifdef INET6 if (isipv6 && inp->inp_ip_minttl > ip6->ip6_hlim) goto dropunlock; else #endif if (inp->inp_ip_minttl > ip->ip_ttl) goto dropunlock; } /* * A previous connection in TIMEWAIT state is supposed to catch stray * or duplicate segments arriving late. If this segment was a * legitimate new connection attempt, the old INPCB gets removed and * we can try again to find a listening socket. * * At this point, due to earlier optimism, we may hold only an inpcb * lock, and not the inpcbinfo write lock. If so, we need to try to * acquire it, or if that fails, acquire a reference on the inpcb, * drop all locks, acquire a global write lock, and then re-acquire * the inpcb lock. We may at that point discover that another thread * has tried to free the inpcb, in which case we need to loop back * and try to find a new inpcb to deliver to. * * XXXRW: It may be time to rethink timewait locking. */ relocked: if (inp->inp_flags & INP_TIMEWAIT) { if (ti_locked == TI_UNLOCKED) { if (INP_INFO_TRY_RLOCK(&V_tcbinfo) == 0) { in_pcbref(inp); INP_WUNLOCK(inp); INP_INFO_RLOCK(&V_tcbinfo); ti_locked = TI_RLOCKED; INP_WLOCK(inp); if (in_pcbrele_wlocked(inp)) { inp = NULL; goto findpcb; } } else ti_locked = TI_RLOCKED; } INP_INFO_RLOCK_ASSERT(&V_tcbinfo); if (thflags & TH_SYN) tcp_dooptions(&to, optp, optlen, TO_SYN); /* * NB: tcp_twcheck unlocks the INP and frees the mbuf. */ if (tcp_twcheck(inp, &to, th, m, tlen)) goto findpcb; INP_INFO_RUNLOCK(&V_tcbinfo); return (IPPROTO_DONE); } /* * The TCPCB may no longer exist if the connection is winding * down or it is in the CLOSED state. Either way we drop the * segment and send an appropriate response. */ tp = intotcpcb(inp); if (tp == NULL || tp->t_state == TCPS_CLOSED) { rstreason = BANDLIM_RST_CLOSEDPORT; goto dropwithreset; } #ifdef TCP_OFFLOAD if (tp->t_flags & TF_TOE) { tcp_offload_input(tp, m); m = NULL; /* consumed by the TOE driver */ goto dropunlock; } #endif /* * We've identified a valid inpcb, but it could be that we need an * inpcbinfo write lock but don't hold it. In this case, attempt to * acquire using the same strategy as the TIMEWAIT case above. If we * relock, we have to jump back to 'relocked' as the connection might * now be in TIMEWAIT. */ #ifdef INVARIANTS if ((thflags & (TH_FIN | TH_RST)) != 0) INP_INFO_RLOCK_ASSERT(&V_tcbinfo); #endif if (!((tp->t_state == TCPS_ESTABLISHED && (thflags & TH_SYN) == 0) || (tp->t_state == TCPS_LISTEN && (thflags & TH_SYN)))) { if (ti_locked == TI_UNLOCKED) { if (INP_INFO_TRY_RLOCK(&V_tcbinfo) == 0) { in_pcbref(inp); INP_WUNLOCK(inp); INP_INFO_RLOCK(&V_tcbinfo); ti_locked = TI_RLOCKED; INP_WLOCK(inp); if (in_pcbrele_wlocked(inp)) { inp = NULL; goto findpcb; } goto relocked; } else ti_locked = TI_RLOCKED; } INP_INFO_RLOCK_ASSERT(&V_tcbinfo); } #ifdef MAC INP_WLOCK_ASSERT(inp); if (mac_inpcb_check_deliver(inp, m)) goto dropunlock; #endif so = inp->inp_socket; KASSERT(so != NULL, ("%s: so == NULL", __func__)); #ifdef TCPDEBUG if (so->so_options & SO_DEBUG) { ostate = tp->t_state; #ifdef INET6 if (isipv6) { bcopy((char *)ip6, (char *)tcp_saveipgen, sizeof(*ip6)); } else #endif bcopy((char *)ip, (char *)tcp_saveipgen, sizeof(*ip)); tcp_savetcp = *th; } #endif /* TCPDEBUG */ /* * When the socket is accepting connections (the INPCB is in LISTEN * state) we look into the SYN cache if this is a new connection * attempt or the completion of a previous one. */ if (so->so_options & SO_ACCEPTCONN) { struct in_conninfo inc; KASSERT(tp->t_state == TCPS_LISTEN, ("%s: so accepting but " "tp not listening", __func__)); bzero(&inc, sizeof(inc)); #ifdef INET6 if (isipv6) { inc.inc_flags |= INC_ISIPV6; inc.inc6_faddr = ip6->ip6_src; inc.inc6_laddr = ip6->ip6_dst; } else #endif { inc.inc_faddr = ip->ip_src; inc.inc_laddr = ip->ip_dst; } inc.inc_fport = th->th_sport; inc.inc_lport = th->th_dport; inc.inc_fibnum = so->so_fibnum; /* * Check for an existing connection attempt in syncache if * the flag is only ACK. A successful lookup creates a new * socket appended to the listen queue in SYN_RECEIVED state. */ if ((thflags & (TH_RST|TH_ACK|TH_SYN)) == TH_ACK) { INP_INFO_RLOCK_ASSERT(&V_tcbinfo); /* * Parse the TCP options here because * syncookies need access to the reflected * timestamp. */ tcp_dooptions(&to, optp, optlen, 0); /* * NB: syncache_expand() doesn't unlock * inp and tcpinfo locks. */ if (!syncache_expand(&inc, &to, th, &so, m)) { /* * No syncache entry or ACK was not * for our SYN/ACK. Send a RST. * NB: syncache did its own logging * of the failure cause. */ rstreason = BANDLIM_RST_OPENPORT; goto dropwithreset; } if (so == NULL) { /* * We completed the 3-way handshake * but could not allocate a socket * either due to memory shortage, * listen queue length limits or * global socket limits. Send RST * or wait and have the remote end * retransmit the ACK for another * try. */ if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "Socket allocation failed due to " "limits or memory shortage, %s\n", s, __func__, V_tcp_sc_rst_sock_fail ? "sending RST" : "try again"); if (V_tcp_sc_rst_sock_fail) { rstreason = BANDLIM_UNLIMITED; goto dropwithreset; } else goto dropunlock; } /* * Socket is created in state SYN_RECEIVED. * Unlock the listen socket, lock the newly * created socket and update the tp variable. */ INP_WUNLOCK(inp); /* listen socket */ inp = sotoinpcb(so); /* * New connection inpcb is already locked by * syncache_expand(). */ INP_WLOCK_ASSERT(inp); tp = intotcpcb(inp); KASSERT(tp->t_state == TCPS_SYN_RECEIVED, ("%s: ", __func__)); #ifdef TCP_SIGNATURE if (sig_checked == 0) { tcp_dooptions(&to, optp, optlen, (thflags & TH_SYN) ? TO_SYN : 0); if (!tcp_signature_verify_input(m, off0, tlen, optlen, &to, th, tp->t_flags)) { /* * In SYN_SENT state if it receives an * RST, it is allowed for further * processing. */ if ((thflags & TH_RST) == 0 || (tp->t_state == TCPS_SYN_SENT) == 0) goto dropunlock; } sig_checked = 1; } #endif /* * Process the segment and the data it * contains. tcp_do_segment() consumes * the mbuf chain and unlocks the inpcb. */ tcp_do_segment(m, th, so, tp, drop_hdrlen, tlen, iptos, ti_locked); INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); return (IPPROTO_DONE); } /* * Segment flag validation for new connection attempts: * * Our (SYN|ACK) response was rejected. * Check with syncache and remove entry to prevent * retransmits. * * NB: syncache_chkrst does its own logging of failure * causes. */ if (thflags & TH_RST) { syncache_chkrst(&inc, th); goto dropunlock; } /* * We can't do anything without SYN. */ if ((thflags & TH_SYN) == 0) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "SYN is missing, segment ignored\n", s, __func__); TCPSTAT_INC(tcps_badsyn); goto dropunlock; } /* * (SYN|ACK) is bogus on a listen socket. */ if (thflags & TH_ACK) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "SYN|ACK invalid, segment rejected\n", s, __func__); syncache_badack(&inc); /* XXX: Not needed! */ TCPSTAT_INC(tcps_badsyn); rstreason = BANDLIM_RST_OPENPORT; goto dropwithreset; } /* * If the drop_synfin option is enabled, drop all * segments with both the SYN and FIN bits set. * This prevents e.g. nmap from identifying the * TCP/IP stack. * XXX: Poor reasoning. nmap has other methods * and is constantly refining its stack detection * strategies. * XXX: This is a violation of the TCP specification * and was used by RFC1644. */ if ((thflags & TH_FIN) && V_drop_synfin) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "SYN|FIN segment ignored (based on " "sysctl setting)\n", s, __func__); TCPSTAT_INC(tcps_badsyn); goto dropunlock; } /* * Segment's flags are (SYN) or (SYN|FIN). * * TH_PUSH, TH_URG, TH_ECE, TH_CWR are ignored * as they do not affect the state of the TCP FSM. * The data pointed to by TH_URG and th_urp is ignored. */ KASSERT((thflags & (TH_RST|TH_ACK)) == 0, ("%s: Listen socket: TH_RST or TH_ACK set", __func__)); KASSERT(thflags & (TH_SYN), ("%s: Listen socket: TH_SYN not set", __func__)); #ifdef INET6 /* * If deprecated address is forbidden, * we do not accept SYN to deprecated interface * address to prevent any new inbound connection from * getting established. * When we do not accept SYN, we send a TCP RST, * with deprecated source address (instead of dropping * it). We compromise it as it is much better for peer * to send a RST, and RST will be the final packet * for the exchange. * * If we do not forbid deprecated addresses, we accept * the SYN packet. RFC2462 does not suggest dropping * SYN in this case. * If we decipher RFC2462 5.5.4, it says like this: * 1. use of deprecated addr with existing * communication is okay - "SHOULD continue to be * used" * 2. use of it with new communication: * (2a) "SHOULD NOT be used if alternate address * with sufficient scope is available" * (2b) nothing mentioned otherwise. * Here we fall into (2b) case as we have no choice in * our source address selection - we must obey the peer. * * The wording in RFC2462 is confusing, and there are * multiple description text for deprecated address * handling - worse, they are not exactly the same. * I believe 5.5.4 is the best one, so we follow 5.5.4. */ if (isipv6 && !V_ip6_use_deprecated) { struct in6_ifaddr *ia6; ia6 = in6ifa_ifwithaddr(&ip6->ip6_dst, 0 /* XXX */); if (ia6 != NULL && (ia6->ia6_flags & IN6_IFF_DEPRECATED)) { ifa_free(&ia6->ia_ifa); if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "Connection attempt to deprecated " "IPv6 address rejected\n", s, __func__); rstreason = BANDLIM_RST_OPENPORT; goto dropwithreset; } if (ia6) ifa_free(&ia6->ia_ifa); } #endif /* INET6 */ /* * Basic sanity checks on incoming SYN requests: * Don't respond if the destination is a link layer * broadcast according to RFC1122 4.2.3.10, p. 104. * If it is from this socket it must be forged. * Don't respond if the source or destination is a * global or subnet broad- or multicast address. * Note that it is quite possible to receive unicast * link-layer packets with a broadcast IP address. Use * in_broadcast() to find them. */ if (m->m_flags & (M_BCAST|M_MCAST)) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "Connection attempt from broad- or multicast " "link layer address ignored\n", s, __func__); goto dropunlock; } #ifdef INET6 if (isipv6) { if (th->th_dport == th->th_sport && IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &ip6->ip6_src)) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "Connection attempt to/from self " "ignored\n", s, __func__); goto dropunlock; } if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || IN6_IS_ADDR_MULTICAST(&ip6->ip6_src)) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "Connection attempt from/to multicast " "address ignored\n", s, __func__); goto dropunlock; } } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { if (th->th_dport == th->th_sport && ip->ip_dst.s_addr == ip->ip_src.s_addr) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "Connection attempt from/to self " "ignored\n", s, __func__); goto dropunlock; } if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) || IN_MULTICAST(ntohl(ip->ip_src.s_addr)) || ip->ip_src.s_addr == htonl(INADDR_BROADCAST) || in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "Connection attempt from/to broad- " "or multicast address ignored\n", s, __func__); goto dropunlock; } } #endif /* * SYN appears to be valid. Create compressed TCP state * for syncache. */ #ifdef TCPDEBUG if (so->so_options & SO_DEBUG) tcp_trace(TA_INPUT, ostate, tp, (void *)tcp_saveipgen, &tcp_savetcp, 0); #endif TCP_PROBE3(debug__input, tp, th, mtod(m, const char *)); tcp_dooptions(&to, optp, optlen, TO_SYN); syncache_add(&inc, &to, th, inp, &so, m, NULL, NULL); /* * Entry added to syncache and mbuf consumed. * Only the listen socket is unlocked by syncache_add(). */ if (ti_locked == TI_RLOCKED) { INP_INFO_RUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; } INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); return (IPPROTO_DONE); } else if (tp->t_state == TCPS_LISTEN) { /* * When a listen socket is torn down the SO_ACCEPTCONN * flag is removed first while connections are drained * from the accept queue in a unlock/lock cycle of the * ACCEPT_LOCK, opening a race condition allowing a SYN * attempt go through unhandled. */ goto dropunlock; } #ifdef TCP_SIGNATURE if (sig_checked == 0) { tcp_dooptions(&to, optp, optlen, (thflags & TH_SYN) ? TO_SYN : 0); if (!tcp_signature_verify_input(m, off0, tlen, optlen, &to, th, tp->t_flags)) { /* * In SYN_SENT state if it receives an RST, it is * allowed for further processing. */ if ((thflags & TH_RST) == 0 || (tp->t_state == TCPS_SYN_SENT) == 0) goto dropunlock; } sig_checked = 1; } #endif TCP_PROBE5(receive, NULL, tp, mtod(m, const char *), tp, th); /* * Segment belongs to a connection in SYN_SENT, ESTABLISHED or later * state. tcp_do_segment() always consumes the mbuf chain, unlocks * the inpcb, and unlocks pcbinfo. */ tcp_do_segment(m, th, so, tp, drop_hdrlen, tlen, iptos, ti_locked); INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); return (IPPROTO_DONE); dropwithreset: TCP_PROBE5(receive, NULL, tp, mtod(m, const char *), tp, th); if (ti_locked == TI_RLOCKED) { INP_INFO_RUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; } #ifdef INVARIANTS else { KASSERT(ti_locked == TI_UNLOCKED, ("%s: dropwithreset " "ti_locked: %d", __func__, ti_locked)); INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); } #endif if (inp != NULL) { tcp_dropwithreset(m, th, tp, tlen, rstreason); INP_WUNLOCK(inp); } else tcp_dropwithreset(m, th, NULL, tlen, rstreason); m = NULL; /* mbuf chain got consumed. */ goto drop; dropunlock: if (m != NULL) TCP_PROBE5(receive, NULL, tp, mtod(m, const char *), tp, th); if (ti_locked == TI_RLOCKED) { INP_INFO_RUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; } #ifdef INVARIANTS else { KASSERT(ti_locked == TI_UNLOCKED, ("%s: dropunlock " "ti_locked: %d", __func__, ti_locked)); INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); } #endif if (inp != NULL) INP_WUNLOCK(inp); drop: INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); if (s != NULL) free(s, M_TCPLOG); if (m != NULL) m_freem(m); return (IPPROTO_DONE); } static void tcp_do_segment(struct mbuf *m, struct tcphdr *th, struct socket *so, struct tcpcb *tp, int drop_hdrlen, int tlen, uint8_t iptos, int ti_locked) { int thflags, acked, ourfinisacked, needoutput = 0; int rstreason, todrop, win; u_long tiwin; char *s; struct in_conninfo *inc; struct mbuf *mfree; struct tcpopt to; #ifdef TCPDEBUG /* * The size of tcp_saveipgen must be the size of the max ip header, * now IPv6. */ u_char tcp_saveipgen[IP6_HDR_LEN]; struct tcphdr tcp_savetcp; short ostate = 0; #endif thflags = th->th_flags; inc = &tp->t_inpcb->inp_inc; tp->sackhint.last_sack_ack = 0; /* * If this is either a state-changing packet or current state isn't * established, we require a write lock on tcbinfo. Otherwise, we * allow the tcbinfo to be in either alocked or unlocked, as the * caller may have unnecessarily acquired a write lock due to a race. */ if ((thflags & (TH_SYN | TH_FIN | TH_RST)) != 0 || tp->t_state != TCPS_ESTABLISHED) { KASSERT(ti_locked == TI_RLOCKED, ("%s ti_locked %d for " "SYN/FIN/RST/!EST", __func__, ti_locked)); INP_INFO_RLOCK_ASSERT(&V_tcbinfo); } else { #ifdef INVARIANTS if (ti_locked == TI_RLOCKED) INP_INFO_RLOCK_ASSERT(&V_tcbinfo); else { KASSERT(ti_locked == TI_UNLOCKED, ("%s: EST " "ti_locked: %d", __func__, ti_locked)); INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); } #endif } INP_WLOCK_ASSERT(tp->t_inpcb); KASSERT(tp->t_state > TCPS_LISTEN, ("%s: TCPS_LISTEN", __func__)); KASSERT(tp->t_state != TCPS_TIME_WAIT, ("%s: TCPS_TIME_WAIT", __func__)); /* * Segment received on connection. * Reset idle time and keep-alive timer. * XXX: This should be done after segment * validation to ignore broken/spoofed segs. */ tp->t_rcvtime = ticks; if (TCPS_HAVEESTABLISHED(tp->t_state)) tcp_timer_activate(tp, TT_KEEP, TP_KEEPIDLE(tp)); /* * Scale up the window into a 32-bit value. * For the SYN_SENT state the scale is zero. */ tiwin = th->th_win << tp->snd_scale; /* * TCP ECN processing. */ if (tp->t_flags & TF_ECN_PERMIT) { if (thflags & TH_CWR) tp->t_flags &= ~TF_ECN_SND_ECE; switch (iptos & IPTOS_ECN_MASK) { case IPTOS_ECN_CE: tp->t_flags |= TF_ECN_SND_ECE; TCPSTAT_INC(tcps_ecn_ce); break; case IPTOS_ECN_ECT0: TCPSTAT_INC(tcps_ecn_ect0); break; case IPTOS_ECN_ECT1: TCPSTAT_INC(tcps_ecn_ect1); break; } /* Process a packet differently from RFC3168. */ cc_ecnpkt_handler(tp, th, iptos); /* Congestion experienced. */ if (thflags & TH_ECE) { cc_cong_signal(tp, th, CC_ECN); } } /* * Parse options on any incoming segment. */ tcp_dooptions(&to, (u_char *)(th + 1), (th->th_off << 2) - sizeof(struct tcphdr), (thflags & TH_SYN) ? TO_SYN : 0); /* * If echoed timestamp is later than the current time, * fall back to non RFC1323 RTT calculation. Normalize * timestamp if syncookies were used when this connection * was established. */ if ((to.to_flags & TOF_TS) && (to.to_tsecr != 0)) { to.to_tsecr -= tp->ts_offset; if (TSTMP_GT(to.to_tsecr, tcp_ts_getticks())) to.to_tsecr = 0; } /* * If timestamps were negotiated during SYN/ACK they should * appear on every segment during this session and vice versa. */ if ((tp->t_flags & TF_RCVD_TSTMP) && !(to.to_flags & TOF_TS)) { if ((s = tcp_log_addrs(inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: Timestamp missing, " "no action\n", s, __func__); free(s, M_TCPLOG); } } if (!(tp->t_flags & TF_RCVD_TSTMP) && (to.to_flags & TOF_TS)) { if ((s = tcp_log_addrs(inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: Timestamp not expected, " "no action\n", s, __func__); free(s, M_TCPLOG); } } /* * Process options only when we get SYN/ACK back. The SYN case * for incoming connections is handled in tcp_syncache. * According to RFC1323 the window field in a SYN (i.e., a * or ) segment itself is never scaled. * XXX this is traditional behavior, may need to be cleaned up. */ if (tp->t_state == TCPS_SYN_SENT && (thflags & TH_SYN)) { if ((to.to_flags & TOF_SCALE) && (tp->t_flags & TF_REQ_SCALE)) { tp->t_flags |= TF_RCVD_SCALE; tp->snd_scale = to.to_wscale; } /* * Initial send window. It will be updated with * the next incoming segment to the scaled value. */ tp->snd_wnd = th->th_win; if (to.to_flags & TOF_TS) { tp->t_flags |= TF_RCVD_TSTMP; tp->ts_recent = to.to_tsval; tp->ts_recent_age = tcp_ts_getticks(); } if (to.to_flags & TOF_MSS) tcp_mss(tp, to.to_mss); if ((tp->t_flags & TF_SACK_PERMIT) && (to.to_flags & TOF_SACKPERM) == 0) tp->t_flags &= ~TF_SACK_PERMIT; } /* * Header prediction: check for the two common cases * of a uni-directional data xfer. If the packet has * no control flags, is in-sequence, the window didn't * change and we're not retransmitting, it's a * candidate. If the length is zero and the ack moved * forward, we're the sender side of the xfer. Just * free the data acked & wake any higher level process * that was blocked waiting for space. If the length * is non-zero and the ack didn't move, we're the * receiver side. If we're getting packets in-order * (the reassembly queue is empty), add the data to * the socket buffer and note that we need a delayed ack. * Make sure that the hidden state-flags are also off. * Since we check for TCPS_ESTABLISHED first, it can only * be TH_NEEDSYN. */ if (tp->t_state == TCPS_ESTABLISHED && th->th_seq == tp->rcv_nxt && (thflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ACK)) == TH_ACK && tp->snd_nxt == tp->snd_max && tiwin && tiwin == tp->snd_wnd && ((tp->t_flags & (TF_NEEDSYN|TF_NEEDFIN)) == 0) && LIST_EMPTY(&tp->t_segq) && ((to.to_flags & TOF_TS) == 0 || TSTMP_GEQ(to.to_tsval, tp->ts_recent)) ) { /* * If last ACK falls within this segment's sequence numbers, * record the timestamp. * NOTE that the test is modified according to the latest * proposal of the tcplw@cray.com list (Braden 1993/04/26). */ if ((to.to_flags & TOF_TS) != 0 && SEQ_LEQ(th->th_seq, tp->last_ack_sent)) { tp->ts_recent_age = tcp_ts_getticks(); tp->ts_recent = to.to_tsval; } if (tlen == 0) { if (SEQ_GT(th->th_ack, tp->snd_una) && SEQ_LEQ(th->th_ack, tp->snd_max) && !IN_RECOVERY(tp->t_flags) && (to.to_flags & TOF_SACK) == 0 && TAILQ_EMPTY(&tp->snd_holes)) { /* * This is a pure ack for outstanding data. */ if (ti_locked == TI_RLOCKED) INP_INFO_RUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; TCPSTAT_INC(tcps_predack); /* * "bad retransmit" recovery. */ if (tp->t_rxtshift == 1 && tp->t_flags & TF_PREVVALID && (int)(ticks - tp->t_badrxtwin) < 0) { cc_cong_signal(tp, th, CC_RTO_ERR); } /* * Recalculate the transmit timer / rtt. * * Some boxes send broken timestamp replies * during the SYN+ACK phase, ignore * timestamps of 0 or we could calculate a * huge RTT and blow up the retransmit timer. */ if ((to.to_flags & TOF_TS) != 0 && to.to_tsecr) { u_int t; t = tcp_ts_getticks() - to.to_tsecr; if (!tp->t_rttlow || tp->t_rttlow > t) tp->t_rttlow = t; tcp_xmit_timer(tp, TCP_TS_TO_TICKS(t) + 1); } else if (tp->t_rtttime && SEQ_GT(th->th_ack, tp->t_rtseq)) { if (!tp->t_rttlow || tp->t_rttlow > ticks - tp->t_rtttime) tp->t_rttlow = ticks - tp->t_rtttime; tcp_xmit_timer(tp, ticks - tp->t_rtttime); } acked = BYTES_THIS_ACK(tp, th); /* Run HHOOK_TCP_ESTABLISHED_IN helper hooks. */ hhook_run_tcp_est_in(tp, th, &to); TCPSTAT_INC(tcps_rcvackpack); TCPSTAT_ADD(tcps_rcvackbyte, acked); sbdrop(&so->so_snd, acked); if (SEQ_GT(tp->snd_una, tp->snd_recover) && SEQ_LEQ(th->th_ack, tp->snd_recover)) tp->snd_recover = th->th_ack - 1; /* * Let the congestion control algorithm update * congestion control related information. This * typically means increasing the congestion * window. */ cc_ack_received(tp, th, CC_ACK); tp->snd_una = th->th_ack; /* * Pull snd_wl2 up to prevent seq wrap relative * to th_ack. */ tp->snd_wl2 = th->th_ack; tp->t_dupacks = 0; m_freem(m); - ND6_HINT(tp); /* Some progress has been made. */ /* * If all outstanding data are acked, stop * retransmit timer, otherwise restart timer * using current (possibly backed-off) value. * If process is waiting for space, * wakeup/selwakeup/signal. If data * are ready to send, let tcp_output * decide between more output or persist. */ #ifdef TCPDEBUG if (so->so_options & SO_DEBUG) tcp_trace(TA_INPUT, ostate, tp, (void *)tcp_saveipgen, &tcp_savetcp, 0); #endif TCP_PROBE3(debug__input, tp, th, mtod(m, const char *)); if (tp->snd_una == tp->snd_max) tcp_timer_activate(tp, TT_REXMT, 0); else if (!tcp_timer_active(tp, TT_PERSIST)) tcp_timer_activate(tp, TT_REXMT, tp->t_rxtcur); sowwakeup(so); if (sbavail(&so->so_snd)) (void) tcp_output(tp); goto check_delack; } } else if (th->th_ack == tp->snd_una && tlen <= sbspace(&so->so_rcv)) { int newsize = 0; /* automatic sockbuf scaling */ /* * This is a pure, in-sequence data packet with * nothing on the reassembly queue and we have enough * buffer space to take it. */ if (ti_locked == TI_RLOCKED) INP_INFO_RUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; /* Clean receiver SACK report if present */ if ((tp->t_flags & TF_SACK_PERMIT) && tp->rcv_numsacks) tcp_clean_sackreport(tp); TCPSTAT_INC(tcps_preddat); tp->rcv_nxt += tlen; /* * Pull snd_wl1 up to prevent seq wrap relative to * th_seq. */ tp->snd_wl1 = th->th_seq; /* * Pull rcv_up up to prevent seq wrap relative to * rcv_nxt. */ tp->rcv_up = tp->rcv_nxt; TCPSTAT_INC(tcps_rcvpack); TCPSTAT_ADD(tcps_rcvbyte, tlen); - ND6_HINT(tp); /* Some progress has been made */ #ifdef TCPDEBUG if (so->so_options & SO_DEBUG) tcp_trace(TA_INPUT, ostate, tp, (void *)tcp_saveipgen, &tcp_savetcp, 0); #endif TCP_PROBE3(debug__input, tp, th, mtod(m, const char *)); /* * Automatic sizing of receive socket buffer. Often the send * buffer size is not optimally adjusted to the actual network * conditions at hand (delay bandwidth product). Setting the * buffer size too small limits throughput on links with high * bandwidth and high delay (eg. trans-continental/oceanic links). * * On the receive side the socket buffer memory is only rarely * used to any significant extent. This allows us to be much * more aggressive in scaling the receive socket buffer. For * the case that the buffer space is actually used to a large * extent and we run out of kernel memory we can simply drop * the new segments; TCP on the sender will just retransmit it * later. Setting the buffer size too big may only consume too * much kernel memory if the application doesn't read() from * the socket or packet loss or reordering makes use of the * reassembly queue. * * The criteria to step up the receive buffer one notch are: * 1. Application has not set receive buffer size with * SO_RCVBUF. Setting SO_RCVBUF clears SB_AUTOSIZE. * 2. the number of bytes received during the time it takes * one timestamp to be reflected back to us (the RTT); * 3. received bytes per RTT is within seven eighth of the * current socket buffer size; * 4. receive buffer size has not hit maximal automatic size; * * This algorithm does one step per RTT at most and only if * we receive a bulk stream w/o packet losses or reorderings. * Shrinking the buffer during idle times is not necessary as * it doesn't consume any memory when idle. * * TODO: Only step up if the application is actually serving * the buffer to better manage the socket buffer resources. */ if (V_tcp_do_autorcvbuf && (to.to_flags & TOF_TS) && to.to_tsecr && (so->so_rcv.sb_flags & SB_AUTOSIZE)) { if (TSTMP_GT(to.to_tsecr, tp->rfbuf_ts) && to.to_tsecr - tp->rfbuf_ts < hz) { if (tp->rfbuf_cnt > (so->so_rcv.sb_hiwat / 8 * 7) && so->so_rcv.sb_hiwat < V_tcp_autorcvbuf_max) { newsize = min(so->so_rcv.sb_hiwat + V_tcp_autorcvbuf_inc, V_tcp_autorcvbuf_max); } /* Start over with next RTT. */ tp->rfbuf_ts = 0; tp->rfbuf_cnt = 0; } else tp->rfbuf_cnt += tlen; /* add up */ } /* Add data to socket buffer. */ SOCKBUF_LOCK(&so->so_rcv); if (so->so_rcv.sb_state & SBS_CANTRCVMORE) { m_freem(m); } else { /* * Set new socket buffer size. * Give up when limit is reached. */ if (newsize) if (!sbreserve_locked(&so->so_rcv, newsize, so, NULL)) so->so_rcv.sb_flags &= ~SB_AUTOSIZE; m_adj(m, drop_hdrlen); /* delayed header drop */ sbappendstream_locked(&so->so_rcv, m, 0); } /* NB: sorwakeup_locked() does an implicit unlock. */ sorwakeup_locked(so); if (DELAY_ACK(tp, tlen)) { tp->t_flags |= TF_DELACK; } else { tp->t_flags |= TF_ACKNOW; tcp_output(tp); } goto check_delack; } } /* * Calculate amount of space in receive window, * and then do TCP input processing. * Receive window is amount of space in rcv queue, * but not less than advertised window. */ win = sbspace(&so->so_rcv); if (win < 0) win = 0; tp->rcv_wnd = imax(win, (int)(tp->rcv_adv - tp->rcv_nxt)); /* Reset receive buffer auto scaling when not in bulk receive mode. */ tp->rfbuf_ts = 0; tp->rfbuf_cnt = 0; switch (tp->t_state) { /* * If the state is SYN_RECEIVED: * if seg contains an ACK, but not for our SYN/ACK, send a RST. */ case TCPS_SYN_RECEIVED: if ((thflags & TH_ACK) && (SEQ_LEQ(th->th_ack, tp->snd_una) || SEQ_GT(th->th_ack, tp->snd_max))) { rstreason = BANDLIM_RST_OPENPORT; goto dropwithreset; } break; /* * If the state is SYN_SENT: * if seg contains an ACK, but not for our SYN, drop the input. * if seg contains a RST, then drop the connection. * if seg does not contain SYN, then drop it. * Otherwise this is an acceptable SYN segment * initialize tp->rcv_nxt and tp->irs * if seg contains ack then advance tp->snd_una * if seg contains an ECE and ECN support is enabled, the stream * is ECN capable. * if SYN has been acked change to ESTABLISHED else SYN_RCVD state * arrange for segment to be acked (eventually) * continue processing rest of data/controls, beginning with URG */ case TCPS_SYN_SENT: if ((thflags & TH_ACK) && (SEQ_LEQ(th->th_ack, tp->iss) || SEQ_GT(th->th_ack, tp->snd_max))) { rstreason = BANDLIM_UNLIMITED; goto dropwithreset; } if ((thflags & (TH_ACK|TH_RST)) == (TH_ACK|TH_RST)) { TCP_PROBE5(connect__refused, NULL, tp, mtod(m, const char *), tp, th); tp = tcp_drop(tp, ECONNREFUSED); } if (thflags & TH_RST) goto drop; if (!(thflags & TH_SYN)) goto drop; tp->irs = th->th_seq; tcp_rcvseqinit(tp); if (thflags & TH_ACK) { TCPSTAT_INC(tcps_connects); soisconnected(so); #ifdef MAC mac_socketpeer_set_from_mbuf(m, so); #endif /* Do window scaling on this connection? */ if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) == (TF_RCVD_SCALE|TF_REQ_SCALE)) { tp->rcv_scale = tp->request_r_scale; } tp->rcv_adv += imin(tp->rcv_wnd, TCP_MAXWIN << tp->rcv_scale); tp->snd_una++; /* SYN is acked */ /* * If there's data, delay ACK; if there's also a FIN * ACKNOW will be turned on later. */ if (DELAY_ACK(tp, tlen) && tlen != 0) tcp_timer_activate(tp, TT_DELACK, tcp_delacktime); else tp->t_flags |= TF_ACKNOW; if ((thflags & TH_ECE) && V_tcp_do_ecn) { tp->t_flags |= TF_ECN_PERMIT; TCPSTAT_INC(tcps_ecn_shs); } /* * Received in SYN_SENT[*] state. * Transitions: * SYN_SENT --> ESTABLISHED * SYN_SENT* --> FIN_WAIT_1 */ tp->t_starttime = ticks; if (tp->t_flags & TF_NEEDFIN) { tcp_state_change(tp, TCPS_FIN_WAIT_1); tp->t_flags &= ~TF_NEEDFIN; thflags &= ~TH_SYN; } else { tcp_state_change(tp, TCPS_ESTABLISHED); TCP_PROBE5(connect__established, NULL, tp, mtod(m, const char *), tp, th); cc_conn_init(tp); tcp_timer_activate(tp, TT_KEEP, TP_KEEPIDLE(tp)); } } else { /* * Received initial SYN in SYN-SENT[*] state => * simultaneous open. * If it succeeds, connection is * half-synchronized. * Otherwise, do 3-way handshake: * SYN-SENT -> SYN-RECEIVED * SYN-SENT* -> SYN-RECEIVED* */ tp->t_flags |= (TF_ACKNOW | TF_NEEDSYN); tcp_timer_activate(tp, TT_REXMT, 0); tcp_state_change(tp, TCPS_SYN_RECEIVED); } KASSERT(ti_locked == TI_RLOCKED, ("%s: trimthenstep6: " "ti_locked %d", __func__, ti_locked)); INP_INFO_RLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(tp->t_inpcb); /* * Advance th->th_seq to correspond to first data byte. * If data, trim to stay within window, * dropping FIN if necessary. */ th->th_seq++; if (tlen > tp->rcv_wnd) { todrop = tlen - tp->rcv_wnd; m_adj(m, -todrop); tlen = tp->rcv_wnd; thflags &= ~TH_FIN; TCPSTAT_INC(tcps_rcvpackafterwin); TCPSTAT_ADD(tcps_rcvbyteafterwin, todrop); } tp->snd_wl1 = th->th_seq - 1; tp->rcv_up = th->th_seq; /* * Client side of transaction: already sent SYN and data. * If the remote host used T/TCP to validate the SYN, * our data will be ACK'd; if so, enter normal data segment * processing in the middle of step 5, ack processing. * Otherwise, goto step 6. */ if (thflags & TH_ACK) goto process_ACK; goto step6; /* * If the state is LAST_ACK or CLOSING or TIME_WAIT: * do normal processing. * * NB: Leftover from RFC1644 T/TCP. Cases to be reused later. */ case TCPS_LAST_ACK: case TCPS_CLOSING: break; /* continue normal processing */ } /* * States other than LISTEN or SYN_SENT. * First check the RST flag and sequence number since reset segments * are exempt from the timestamp and connection count tests. This * fixes a bug introduced by the Stevens, vol. 2, p. 960 bugfix * below which allowed reset segments in half the sequence space * to fall though and be processed (which gives forged reset * segments with a random sequence number a 50 percent chance of * killing a connection). * Then check timestamp, if present. * Then check the connection count, if present. * Then check that at least some bytes of segment are within * receive window. If segment begins before rcv_nxt, * drop leading data (and SYN); if nothing left, just ack. */ if (thflags & TH_RST) { /* * RFC5961 Section 3.2 * * - RST drops connection only if SEG.SEQ == RCV.NXT. * - If RST is in window, we send challenge ACK. * * Note: to take into account delayed ACKs, we should * test against last_ack_sent instead of rcv_nxt. * Note 2: we handle special case of closed window, not * covered by the RFC. */ if ((SEQ_GEQ(th->th_seq, tp->last_ack_sent) && SEQ_LT(th->th_seq, tp->last_ack_sent + tp->rcv_wnd)) || (tp->rcv_wnd == 0 && tp->last_ack_sent == th->th_seq)) { INP_INFO_RLOCK_ASSERT(&V_tcbinfo); KASSERT(ti_locked == TI_RLOCKED, ("%s: TH_RST ti_locked %d, th %p tp %p", __func__, ti_locked, th, tp)); KASSERT(tp->t_state != TCPS_SYN_SENT, ("%s: TH_RST for TCPS_SYN_SENT th %p tp %p", __func__, th, tp)); if (V_tcp_insecure_rst || tp->last_ack_sent == th->th_seq) { TCPSTAT_INC(tcps_drops); /* Drop the connection. */ switch (tp->t_state) { case TCPS_SYN_RECEIVED: so->so_error = ECONNREFUSED; goto close; case TCPS_ESTABLISHED: case TCPS_FIN_WAIT_1: case TCPS_FIN_WAIT_2: case TCPS_CLOSE_WAIT: so->so_error = ECONNRESET; close: tcp_state_change(tp, TCPS_CLOSED); /* FALLTHROUGH */ default: tp = tcp_close(tp); } } else { TCPSTAT_INC(tcps_badrst); /* Send challenge ACK. */ tcp_respond(tp, mtod(m, void *), th, m, tp->rcv_nxt, tp->snd_nxt, TH_ACK); tp->last_ack_sent = tp->rcv_nxt; m = NULL; } } goto drop; } /* * RFC5961 Section 4.2 * Send challenge ACK for any SYN in synchronized state. */ if ((thflags & TH_SYN) && tp->t_state != TCPS_SYN_SENT) { KASSERT(ti_locked == TI_RLOCKED, ("tcp_do_segment: TH_SYN ti_locked %d", ti_locked)); INP_INFO_RLOCK_ASSERT(&V_tcbinfo); TCPSTAT_INC(tcps_badsyn); if (V_tcp_insecure_syn && SEQ_GEQ(th->th_seq, tp->last_ack_sent) && SEQ_LT(th->th_seq, tp->last_ack_sent + tp->rcv_wnd)) { tp = tcp_drop(tp, ECONNRESET); rstreason = BANDLIM_UNLIMITED; } else { /* Send challenge ACK. */ tcp_respond(tp, mtod(m, void *), th, m, tp->rcv_nxt, tp->snd_nxt, TH_ACK); tp->last_ack_sent = tp->rcv_nxt; m = NULL; } goto drop; } /* * RFC 1323 PAWS: If we have a timestamp reply on this segment * and it's less than ts_recent, drop it. */ if ((to.to_flags & TOF_TS) != 0 && tp->ts_recent && TSTMP_LT(to.to_tsval, tp->ts_recent)) { /* Check to see if ts_recent is over 24 days old. */ if (tcp_ts_getticks() - tp->ts_recent_age > TCP_PAWS_IDLE) { /* * Invalidate ts_recent. If this segment updates * ts_recent, the age will be reset later and ts_recent * will get a valid value. If it does not, setting * ts_recent to zero will at least satisfy the * requirement that zero be placed in the timestamp * echo reply when ts_recent isn't valid. The * age isn't reset until we get a valid ts_recent * because we don't want out-of-order segments to be * dropped when ts_recent is old. */ tp->ts_recent = 0; } else { TCPSTAT_INC(tcps_rcvduppack); TCPSTAT_ADD(tcps_rcvdupbyte, tlen); TCPSTAT_INC(tcps_pawsdrop); if (tlen) goto dropafterack; goto drop; } } /* * In the SYN-RECEIVED state, validate that the packet belongs to * this connection before trimming the data to fit the receive * window. Check the sequence number versus IRS since we know * the sequence numbers haven't wrapped. This is a partial fix * for the "LAND" DoS attack. */ if (tp->t_state == TCPS_SYN_RECEIVED && SEQ_LT(th->th_seq, tp->irs)) { rstreason = BANDLIM_RST_OPENPORT; goto dropwithreset; } todrop = tp->rcv_nxt - th->th_seq; if (todrop > 0) { if (thflags & TH_SYN) { thflags &= ~TH_SYN; th->th_seq++; if (th->th_urp > 1) th->th_urp--; else thflags &= ~TH_URG; todrop--; } /* * Following if statement from Stevens, vol. 2, p. 960. */ if (todrop > tlen || (todrop == tlen && (thflags & TH_FIN) == 0)) { /* * Any valid FIN must be to the left of the window. * At this point the FIN must be a duplicate or out * of sequence; drop it. */ thflags &= ~TH_FIN; /* * Send an ACK to resynchronize and drop any data. * But keep on processing for RST or ACK. */ tp->t_flags |= TF_ACKNOW; todrop = tlen; TCPSTAT_INC(tcps_rcvduppack); TCPSTAT_ADD(tcps_rcvdupbyte, todrop); } else { TCPSTAT_INC(tcps_rcvpartduppack); TCPSTAT_ADD(tcps_rcvpartdupbyte, todrop); } drop_hdrlen += todrop; /* drop from the top afterwards */ th->th_seq += todrop; tlen -= todrop; if (th->th_urp > todrop) th->th_urp -= todrop; else { thflags &= ~TH_URG; th->th_urp = 0; } } /* * If new data are received on a connection after the * user processes are gone, then RST the other end. */ if ((so->so_state & SS_NOFDREF) && tp->t_state > TCPS_CLOSE_WAIT && tlen) { KASSERT(ti_locked == TI_RLOCKED, ("%s: SS_NOFDEREF && " "CLOSE_WAIT && tlen ti_locked %d", __func__, ti_locked)); INP_INFO_RLOCK_ASSERT(&V_tcbinfo); if ((s = tcp_log_addrs(inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: %s: Received %d bytes of data " "after socket was closed, " "sending RST and removing tcpcb\n", s, __func__, tcpstates[tp->t_state], tlen); free(s, M_TCPLOG); } tp = tcp_close(tp); TCPSTAT_INC(tcps_rcvafterclose); rstreason = BANDLIM_UNLIMITED; goto dropwithreset; } /* * If segment ends after window, drop trailing data * (and PUSH and FIN); if nothing left, just ACK. */ todrop = (th->th_seq + tlen) - (tp->rcv_nxt + tp->rcv_wnd); if (todrop > 0) { TCPSTAT_INC(tcps_rcvpackafterwin); if (todrop >= tlen) { TCPSTAT_ADD(tcps_rcvbyteafterwin, tlen); /* * If window is closed can only take segments at * window edge, and have to drop data and PUSH from * incoming segments. Continue processing, but * remember to ack. Otherwise, drop segment * and ack. */ if (tp->rcv_wnd == 0 && th->th_seq == tp->rcv_nxt) { tp->t_flags |= TF_ACKNOW; TCPSTAT_INC(tcps_rcvwinprobe); } else goto dropafterack; } else TCPSTAT_ADD(tcps_rcvbyteafterwin, todrop); m_adj(m, -todrop); tlen -= todrop; thflags &= ~(TH_PUSH|TH_FIN); } /* * If last ACK falls within this segment's sequence numbers, * record its timestamp. * NOTE: * 1) That the test incorporates suggestions from the latest * proposal of the tcplw@cray.com list (Braden 1993/04/26). * 2) That updating only on newer timestamps interferes with * our earlier PAWS tests, so this check should be solely * predicated on the sequence space of this segment. * 3) That we modify the segment boundary check to be * Last.ACK.Sent <= SEG.SEQ + SEG.Len * instead of RFC1323's * Last.ACK.Sent < SEG.SEQ + SEG.Len, * This modified check allows us to overcome RFC1323's * limitations as described in Stevens TCP/IP Illustrated * Vol. 2 p.869. In such cases, we can still calculate the * RTT correctly when RCV.NXT == Last.ACK.Sent. */ if ((to.to_flags & TOF_TS) != 0 && SEQ_LEQ(th->th_seq, tp->last_ack_sent) && SEQ_LEQ(tp->last_ack_sent, th->th_seq + tlen + ((thflags & (TH_SYN|TH_FIN)) != 0))) { tp->ts_recent_age = tcp_ts_getticks(); tp->ts_recent = to.to_tsval; } /* * If the ACK bit is off: if in SYN-RECEIVED state or SENDSYN * flag is on (half-synchronized state), then queue data for * later processing; else drop segment and return. */ if ((thflags & TH_ACK) == 0) { if (tp->t_state == TCPS_SYN_RECEIVED || (tp->t_flags & TF_NEEDSYN)) goto step6; else if (tp->t_flags & TF_ACKNOW) goto dropafterack; else goto drop; } /* * Ack processing. */ switch (tp->t_state) { /* * In SYN_RECEIVED state, the ack ACKs our SYN, so enter * ESTABLISHED state and continue processing. * The ACK was checked above. */ case TCPS_SYN_RECEIVED: TCPSTAT_INC(tcps_connects); soisconnected(so); /* Do window scaling? */ if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) == (TF_RCVD_SCALE|TF_REQ_SCALE)) { tp->rcv_scale = tp->request_r_scale; tp->snd_wnd = tiwin; } /* * Make transitions: * SYN-RECEIVED -> ESTABLISHED * SYN-RECEIVED* -> FIN-WAIT-1 */ tp->t_starttime = ticks; if (tp->t_flags & TF_NEEDFIN) { tcp_state_change(tp, TCPS_FIN_WAIT_1); tp->t_flags &= ~TF_NEEDFIN; } else { tcp_state_change(tp, TCPS_ESTABLISHED); TCP_PROBE5(accept__established, NULL, tp, mtod(m, const char *), tp, th); cc_conn_init(tp); tcp_timer_activate(tp, TT_KEEP, TP_KEEPIDLE(tp)); } /* * If segment contains data or ACK, will call tcp_reass() * later; if not, do so now to pass queued data to user. */ if (tlen == 0 && (thflags & TH_FIN) == 0) (void) tcp_reass(tp, (struct tcphdr *)0, 0, (struct mbuf *)0); tp->snd_wl1 = th->th_seq - 1; /* FALLTHROUGH */ /* * In ESTABLISHED state: drop duplicate ACKs; ACK out of range * ACKs. If the ack is in the range * tp->snd_una < th->th_ack <= tp->snd_max * then advance tp->snd_una to th->th_ack and drop * data from the retransmission queue. If this ACK reflects * more up to date window information we update our window information. */ case TCPS_ESTABLISHED: case TCPS_FIN_WAIT_1: case TCPS_FIN_WAIT_2: case TCPS_CLOSE_WAIT: case TCPS_CLOSING: case TCPS_LAST_ACK: if (SEQ_GT(th->th_ack, tp->snd_max)) { TCPSTAT_INC(tcps_rcvacktoomuch); goto dropafterack; } if ((tp->t_flags & TF_SACK_PERMIT) && ((to.to_flags & TOF_SACK) || !TAILQ_EMPTY(&tp->snd_holes))) tcp_sack_doack(tp, &to, th->th_ack); /* Run HHOOK_TCP_ESTABLISHED_IN helper hooks. */ hhook_run_tcp_est_in(tp, th, &to); if (SEQ_LEQ(th->th_ack, tp->snd_una)) { if (tlen == 0 && tiwin == tp->snd_wnd) { /* * If this is the first time we've seen a * FIN from the remote, this is not a * duplicate and it needs to be processed * normally. This happens during a * simultaneous close. */ if ((thflags & TH_FIN) && (TCPS_HAVERCVDFIN(tp->t_state) == 0)) { tp->t_dupacks = 0; break; } TCPSTAT_INC(tcps_rcvdupack); /* * If we have outstanding data (other than * a window probe), this is a completely * duplicate ack (ie, window info didn't * change and FIN isn't set), * the ack is the biggest we've * seen and we've seen exactly our rexmt * threshhold of them, assume a packet * has been dropped and retransmit it. * Kludge snd_nxt & the congestion * window so we send only this one * packet. * * We know we're losing at the current * window size so do congestion avoidance * (set ssthresh to half the current window * and pull our congestion window back to * the new ssthresh). * * Dup acks mean that packets have left the * network (they're now cached at the receiver) * so bump cwnd by the amount in the receiver * to keep a constant cwnd packets in the * network. * * When using TCP ECN, notify the peer that * we reduced the cwnd. */ if (!tcp_timer_active(tp, TT_REXMT) || th->th_ack != tp->snd_una) tp->t_dupacks = 0; else if (++tp->t_dupacks > tcprexmtthresh || IN_FASTRECOVERY(tp->t_flags)) { cc_ack_received(tp, th, CC_DUPACK); if ((tp->t_flags & TF_SACK_PERMIT) && IN_FASTRECOVERY(tp->t_flags)) { int awnd; /* * Compute the amount of data in flight first. * We can inject new data into the pipe iff * we have less than 1/2 the original window's * worth of data in flight. */ awnd = (tp->snd_nxt - tp->snd_fack) + tp->sackhint.sack_bytes_rexmit; if (awnd < tp->snd_ssthresh) { tp->snd_cwnd += tp->t_maxseg; if (tp->snd_cwnd > tp->snd_ssthresh) tp->snd_cwnd = tp->snd_ssthresh; } } else tp->snd_cwnd += tp->t_maxseg; (void) tcp_output(tp); goto drop; } else if (tp->t_dupacks == tcprexmtthresh) { tcp_seq onxt = tp->snd_nxt; /* * If we're doing sack, check to * see if we're already in sack * recovery. If we're not doing sack, * check to see if we're in newreno * recovery. */ if (tp->t_flags & TF_SACK_PERMIT) { if (IN_FASTRECOVERY(tp->t_flags)) { tp->t_dupacks = 0; break; } } else { if (SEQ_LEQ(th->th_ack, tp->snd_recover)) { tp->t_dupacks = 0; break; } } /* Congestion signal before ack. */ cc_cong_signal(tp, th, CC_NDUPACK); cc_ack_received(tp, th, CC_DUPACK); tcp_timer_activate(tp, TT_REXMT, 0); tp->t_rtttime = 0; if (tp->t_flags & TF_SACK_PERMIT) { TCPSTAT_INC( tcps_sack_recovery_episode); tp->sack_newdata = tp->snd_nxt; tp->snd_cwnd = tp->t_maxseg; (void) tcp_output(tp); goto drop; } tp->snd_nxt = th->th_ack; tp->snd_cwnd = tp->t_maxseg; (void) tcp_output(tp); KASSERT(tp->snd_limited <= 2, ("%s: tp->snd_limited too big", __func__)); tp->snd_cwnd = tp->snd_ssthresh + tp->t_maxseg * (tp->t_dupacks - tp->snd_limited); if (SEQ_GT(onxt, tp->snd_nxt)) tp->snd_nxt = onxt; goto drop; } else if (V_tcp_do_rfc3042) { cc_ack_received(tp, th, CC_DUPACK); u_long oldcwnd = tp->snd_cwnd; tcp_seq oldsndmax = tp->snd_max; u_int sent; int avail; KASSERT(tp->t_dupacks == 1 || tp->t_dupacks == 2, ("%s: dupacks not 1 or 2", __func__)); if (tp->t_dupacks == 1) tp->snd_limited = 0; tp->snd_cwnd = (tp->snd_nxt - tp->snd_una) + (tp->t_dupacks - tp->snd_limited) * tp->t_maxseg; /* * Only call tcp_output when there * is new data available to be sent. * Otherwise we would send pure ACKs. */ SOCKBUF_LOCK(&so->so_snd); avail = sbavail(&so->so_snd) - (tp->snd_nxt - tp->snd_una); SOCKBUF_UNLOCK(&so->so_snd); if (avail > 0) (void) tcp_output(tp); sent = tp->snd_max - oldsndmax; if (sent > tp->t_maxseg) { KASSERT((tp->t_dupacks == 2 && tp->snd_limited == 0) || (sent == tp->t_maxseg + 1 && tp->t_flags & TF_SENTFIN), ("%s: sent too much", __func__)); tp->snd_limited = 2; } else if (sent > 0) ++tp->snd_limited; tp->snd_cwnd = oldcwnd; goto drop; } } else tp->t_dupacks = 0; break; } KASSERT(SEQ_GT(th->th_ack, tp->snd_una), ("%s: th_ack <= snd_una", __func__)); /* * If the congestion window was inflated to account * for the other side's cached packets, retract it. */ if (IN_FASTRECOVERY(tp->t_flags)) { if (SEQ_LT(th->th_ack, tp->snd_recover)) { if (tp->t_flags & TF_SACK_PERMIT) tcp_sack_partialack(tp, th); else tcp_newreno_partial_ack(tp, th); } else cc_post_recovery(tp, th); } tp->t_dupacks = 0; /* * If we reach this point, ACK is not a duplicate, * i.e., it ACKs something we sent. */ if (tp->t_flags & TF_NEEDSYN) { /* * T/TCP: Connection was half-synchronized, and our * SYN has been ACK'd (so connection is now fully * synchronized). Go to non-starred state, * increment snd_una for ACK of SYN, and check if * we can do window scaling. */ tp->t_flags &= ~TF_NEEDSYN; tp->snd_una++; /* Do window scaling? */ if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) == (TF_RCVD_SCALE|TF_REQ_SCALE)) { tp->rcv_scale = tp->request_r_scale; /* Send window already scaled. */ } } process_ACK: INP_WLOCK_ASSERT(tp->t_inpcb); acked = BYTES_THIS_ACK(tp, th); TCPSTAT_INC(tcps_rcvackpack); TCPSTAT_ADD(tcps_rcvackbyte, acked); /* * If we just performed our first retransmit, and the ACK * arrives within our recovery window, then it was a mistake * to do the retransmit in the first place. Recover our * original cwnd and ssthresh, and proceed to transmit where * we left off. */ if (tp->t_rxtshift == 1 && tp->t_flags & TF_PREVVALID && (int)(ticks - tp->t_badrxtwin) < 0) cc_cong_signal(tp, th, CC_RTO_ERR); /* * If we have a timestamp reply, update smoothed * round trip time. If no timestamp is present but * transmit timer is running and timed sequence * number was acked, update smoothed round trip time. * Since we now have an rtt measurement, cancel the * timer backoff (cf., Phil Karn's retransmit alg.). * Recompute the initial retransmit timer. * * Some boxes send broken timestamp replies * during the SYN+ACK phase, ignore * timestamps of 0 or we could calculate a * huge RTT and blow up the retransmit timer. */ if ((to.to_flags & TOF_TS) != 0 && to.to_tsecr) { u_int t; t = tcp_ts_getticks() - to.to_tsecr; if (!tp->t_rttlow || tp->t_rttlow > t) tp->t_rttlow = t; tcp_xmit_timer(tp, TCP_TS_TO_TICKS(t) + 1); } else if (tp->t_rtttime && SEQ_GT(th->th_ack, tp->t_rtseq)) { if (!tp->t_rttlow || tp->t_rttlow > ticks - tp->t_rtttime) tp->t_rttlow = ticks - tp->t_rtttime; tcp_xmit_timer(tp, ticks - tp->t_rtttime); } /* * If all outstanding data is acked, stop retransmit * timer and remember to restart (more output or persist). * If there is more data to be acked, restart retransmit * timer, using current (possibly backed-off) value. */ if (th->th_ack == tp->snd_max) { tcp_timer_activate(tp, TT_REXMT, 0); needoutput = 1; } else if (!tcp_timer_active(tp, TT_PERSIST)) tcp_timer_activate(tp, TT_REXMT, tp->t_rxtcur); /* * If no data (only SYN) was ACK'd, * skip rest of ACK processing. */ if (acked == 0) goto step6; /* * Let the congestion control algorithm update congestion * control related information. This typically means increasing * the congestion window. */ cc_ack_received(tp, th, CC_ACK); SOCKBUF_LOCK(&so->so_snd); if (acked > sbavail(&so->so_snd)) { tp->snd_wnd -= sbavail(&so->so_snd); mfree = sbcut_locked(&so->so_snd, (int)sbavail(&so->so_snd)); ourfinisacked = 1; } else { mfree = sbcut_locked(&so->so_snd, acked); tp->snd_wnd -= acked; ourfinisacked = 0; } /* NB: sowwakeup_locked() does an implicit unlock. */ sowwakeup_locked(so); m_freem(mfree); /* Detect una wraparound. */ if (!IN_RECOVERY(tp->t_flags) && SEQ_GT(tp->snd_una, tp->snd_recover) && SEQ_LEQ(th->th_ack, tp->snd_recover)) tp->snd_recover = th->th_ack - 1; /* XXXLAS: Can this be moved up into cc_post_recovery? */ if (IN_RECOVERY(tp->t_flags) && SEQ_GEQ(th->th_ack, tp->snd_recover)) { EXIT_RECOVERY(tp->t_flags); } tp->snd_una = th->th_ack; if (tp->t_flags & TF_SACK_PERMIT) { if (SEQ_GT(tp->snd_una, tp->snd_recover)) tp->snd_recover = tp->snd_una; } if (SEQ_LT(tp->snd_nxt, tp->snd_una)) tp->snd_nxt = tp->snd_una; switch (tp->t_state) { /* * In FIN_WAIT_1 STATE in addition to the processing * for the ESTABLISHED state if our FIN is now acknowledged * then enter FIN_WAIT_2. */ case TCPS_FIN_WAIT_1: if (ourfinisacked) { /* * If we can't receive any more * data, then closing user can proceed. * Starting the timer is contrary to the * specification, but if we don't get a FIN * we'll hang forever. * * XXXjl: * we should release the tp also, and use a * compressed state. */ if (so->so_rcv.sb_state & SBS_CANTRCVMORE) { soisdisconnected(so); tcp_timer_activate(tp, TT_2MSL, (tcp_fast_finwait2_recycle ? tcp_finwait2_timeout : TP_MAXIDLE(tp))); } tcp_state_change(tp, TCPS_FIN_WAIT_2); } break; /* * In CLOSING STATE in addition to the processing for * the ESTABLISHED state if the ACK acknowledges our FIN * then enter the TIME-WAIT state, otherwise ignore * the segment. */ case TCPS_CLOSING: if (ourfinisacked) { INP_INFO_RLOCK_ASSERT(&V_tcbinfo); tcp_twstart(tp); INP_INFO_RUNLOCK(&V_tcbinfo); m_freem(m); return; } break; /* * In LAST_ACK, we may still be waiting for data to drain * and/or to be acked, as well as for the ack of our FIN. * If our FIN is now acknowledged, delete the TCB, * enter the closed state and return. */ case TCPS_LAST_ACK: if (ourfinisacked) { INP_INFO_RLOCK_ASSERT(&V_tcbinfo); tp = tcp_close(tp); goto drop; } break; } } step6: INP_WLOCK_ASSERT(tp->t_inpcb); /* * Update window information. * Don't look at window if no ACK: TAC's send garbage on first SYN. */ if ((thflags & TH_ACK) && (SEQ_LT(tp->snd_wl1, th->th_seq) || (tp->snd_wl1 == th->th_seq && (SEQ_LT(tp->snd_wl2, th->th_ack) || (tp->snd_wl2 == th->th_ack && tiwin > tp->snd_wnd))))) { /* keep track of pure window updates */ if (tlen == 0 && tp->snd_wl2 == th->th_ack && tiwin > tp->snd_wnd) TCPSTAT_INC(tcps_rcvwinupd); tp->snd_wnd = tiwin; tp->snd_wl1 = th->th_seq; tp->snd_wl2 = th->th_ack; if (tp->snd_wnd > tp->max_sndwnd) tp->max_sndwnd = tp->snd_wnd; needoutput = 1; } /* * Process segments with URG. */ if ((thflags & TH_URG) && th->th_urp && TCPS_HAVERCVDFIN(tp->t_state) == 0) { /* * This is a kludge, but if we receive and accept * random urgent pointers, we'll crash in * soreceive. It's hard to imagine someone * actually wanting to send this much urgent data. */ SOCKBUF_LOCK(&so->so_rcv); if (th->th_urp + sbavail(&so->so_rcv) > sb_max) { th->th_urp = 0; /* XXX */ thflags &= ~TH_URG; /* XXX */ SOCKBUF_UNLOCK(&so->so_rcv); /* XXX */ goto dodata; /* XXX */ } /* * If this segment advances the known urgent pointer, * then mark the data stream. This should not happen * in CLOSE_WAIT, CLOSING, LAST_ACK or TIME_WAIT STATES since * a FIN has been received from the remote side. * In these states we ignore the URG. * * According to RFC961 (Assigned Protocols), * the urgent pointer points to the last octet * of urgent data. We continue, however, * to consider it to indicate the first octet * of data past the urgent section as the original * spec states (in one of two places). */ if (SEQ_GT(th->th_seq+th->th_urp, tp->rcv_up)) { tp->rcv_up = th->th_seq + th->th_urp; so->so_oobmark = sbavail(&so->so_rcv) + (tp->rcv_up - tp->rcv_nxt) - 1; if (so->so_oobmark == 0) so->so_rcv.sb_state |= SBS_RCVATMARK; sohasoutofband(so); tp->t_oobflags &= ~(TCPOOB_HAVEDATA | TCPOOB_HADDATA); } SOCKBUF_UNLOCK(&so->so_rcv); /* * Remove out of band data so doesn't get presented to user. * This can happen independent of advancing the URG pointer, * but if two URG's are pending at once, some out-of-band * data may creep in... ick. */ if (th->th_urp <= (u_long)tlen && !(so->so_options & SO_OOBINLINE)) { /* hdr drop is delayed */ tcp_pulloutofband(so, th, m, drop_hdrlen); } } else { /* * If no out of band data is expected, * pull receive urgent pointer along * with the receive window. */ if (SEQ_GT(tp->rcv_nxt, tp->rcv_up)) tp->rcv_up = tp->rcv_nxt; } dodata: /* XXX */ INP_WLOCK_ASSERT(tp->t_inpcb); /* * Process the segment text, merging it into the TCP sequencing queue, * and arranging for acknowledgment of receipt if necessary. * This process logically involves adjusting tp->rcv_wnd as data * is presented to the user (this happens in tcp_usrreq.c, * case PRU_RCVD). If a FIN has already been received on this * connection then we just ignore the text. */ if ((tlen || (thflags & TH_FIN)) && TCPS_HAVERCVDFIN(tp->t_state) == 0) { tcp_seq save_start = th->th_seq; m_adj(m, drop_hdrlen); /* delayed header drop */ /* * Insert segment which includes th into TCP reassembly queue * with control block tp. Set thflags to whether reassembly now * includes a segment with FIN. This handles the common case * inline (segment is the next to be received on an established * connection, and the queue is empty), avoiding linkage into * and removal from the queue and repetition of various * conversions. * Set DELACK for segments received in order, but ack * immediately when segments are out of order (so * fast retransmit can work). */ if (th->th_seq == tp->rcv_nxt && LIST_EMPTY(&tp->t_segq) && TCPS_HAVEESTABLISHED(tp->t_state)) { if (DELAY_ACK(tp, tlen)) tp->t_flags |= TF_DELACK; else tp->t_flags |= TF_ACKNOW; tp->rcv_nxt += tlen; thflags = th->th_flags & TH_FIN; TCPSTAT_INC(tcps_rcvpack); TCPSTAT_ADD(tcps_rcvbyte, tlen); - ND6_HINT(tp); SOCKBUF_LOCK(&so->so_rcv); if (so->so_rcv.sb_state & SBS_CANTRCVMORE) m_freem(m); else sbappendstream_locked(&so->so_rcv, m, 0); /* NB: sorwakeup_locked() does an implicit unlock. */ sorwakeup_locked(so); } else { /* * XXX: Due to the header drop above "th" is * theoretically invalid by now. Fortunately * m_adj() doesn't actually frees any mbufs * when trimming from the head. */ thflags = tcp_reass(tp, th, &tlen, m); tp->t_flags |= TF_ACKNOW; } if (tlen > 0 && (tp->t_flags & TF_SACK_PERMIT)) tcp_update_sack_list(tp, save_start, save_start + tlen); #if 0 /* * Note the amount of data that peer has sent into * our window, in order to estimate the sender's * buffer size. * XXX: Unused. */ if (SEQ_GT(tp->rcv_adv, tp->rcv_nxt)) len = so->so_rcv.sb_hiwat - (tp->rcv_adv - tp->rcv_nxt); else len = so->so_rcv.sb_hiwat; #endif } else { m_freem(m); thflags &= ~TH_FIN; } /* * If FIN is received ACK the FIN and let the user know * that the connection is closing. */ if (thflags & TH_FIN) { if (TCPS_HAVERCVDFIN(tp->t_state) == 0) { socantrcvmore(so); /* * If connection is half-synchronized * (ie NEEDSYN flag on) then delay ACK, * so it may be piggybacked when SYN is sent. * Otherwise, since we received a FIN then no * more input can be expected, send ACK now. */ if (tp->t_flags & TF_NEEDSYN) tp->t_flags |= TF_DELACK; else tp->t_flags |= TF_ACKNOW; tp->rcv_nxt++; } switch (tp->t_state) { /* * In SYN_RECEIVED and ESTABLISHED STATES * enter the CLOSE_WAIT state. */ case TCPS_SYN_RECEIVED: tp->t_starttime = ticks; /* FALLTHROUGH */ case TCPS_ESTABLISHED: tcp_state_change(tp, TCPS_CLOSE_WAIT); break; /* * If still in FIN_WAIT_1 STATE FIN has not been acked so * enter the CLOSING state. */ case TCPS_FIN_WAIT_1: tcp_state_change(tp, TCPS_CLOSING); break; /* * In FIN_WAIT_2 state enter the TIME_WAIT state, * starting the time-wait timer, turning off the other * standard timers. */ case TCPS_FIN_WAIT_2: INP_INFO_RLOCK_ASSERT(&V_tcbinfo); KASSERT(ti_locked == TI_RLOCKED, ("%s: dodata " "TCP_FIN_WAIT_2 ti_locked: %d", __func__, ti_locked)); tcp_twstart(tp); INP_INFO_RUNLOCK(&V_tcbinfo); return; } } if (ti_locked == TI_RLOCKED) INP_INFO_RUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; #ifdef TCPDEBUG if (so->so_options & SO_DEBUG) tcp_trace(TA_INPUT, ostate, tp, (void *)tcp_saveipgen, &tcp_savetcp, 0); #endif TCP_PROBE3(debug__input, tp, th, mtod(m, const char *)); /* * Return any desired output. */ if (needoutput || (tp->t_flags & TF_ACKNOW)) (void) tcp_output(tp); check_delack: KASSERT(ti_locked == TI_UNLOCKED, ("%s: check_delack ti_locked %d", __func__, ti_locked)); INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(tp->t_inpcb); if (tp->t_flags & TF_DELACK) { tp->t_flags &= ~TF_DELACK; tcp_timer_activate(tp, TT_DELACK, tcp_delacktime); } INP_WUNLOCK(tp->t_inpcb); return; dropafterack: /* * Generate an ACK dropping incoming segment if it occupies * sequence space, where the ACK reflects our state. * * We can now skip the test for the RST flag since all * paths to this code happen after packets containing * RST have been dropped. * * In the SYN-RECEIVED state, don't send an ACK unless the * segment we received passes the SYN-RECEIVED ACK test. * If it fails send a RST. This breaks the loop in the * "LAND" DoS attack, and also prevents an ACK storm * between two listening ports that have been sent forged * SYN segments, each with the source address of the other. */ if (tp->t_state == TCPS_SYN_RECEIVED && (thflags & TH_ACK) && (SEQ_GT(tp->snd_una, th->th_ack) || SEQ_GT(th->th_ack, tp->snd_max)) ) { rstreason = BANDLIM_RST_OPENPORT; goto dropwithreset; } #ifdef TCPDEBUG if (so->so_options & SO_DEBUG) tcp_trace(TA_DROP, ostate, tp, (void *)tcp_saveipgen, &tcp_savetcp, 0); #endif TCP_PROBE3(debug__input, tp, th, mtod(m, const char *)); if (ti_locked == TI_RLOCKED) INP_INFO_RUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; tp->t_flags |= TF_ACKNOW; (void) tcp_output(tp); INP_WUNLOCK(tp->t_inpcb); m_freem(m); return; dropwithreset: if (ti_locked == TI_RLOCKED) INP_INFO_RUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; if (tp != NULL) { tcp_dropwithreset(m, th, tp, tlen, rstreason); INP_WUNLOCK(tp->t_inpcb); } else tcp_dropwithreset(m, th, NULL, tlen, rstreason); return; drop: if (ti_locked == TI_RLOCKED) { INP_INFO_RUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; } #ifdef INVARIANTS else INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); #endif /* * Drop space held by incoming segment and return. */ #ifdef TCPDEBUG if (tp == NULL || (tp->t_inpcb->inp_socket->so_options & SO_DEBUG)) tcp_trace(TA_DROP, ostate, tp, (void *)tcp_saveipgen, &tcp_savetcp, 0); #endif TCP_PROBE3(debug__input, tp, th, mtod(m, const char *)); if (tp != NULL) INP_WUNLOCK(tp->t_inpcb); m_freem(m); } /* * Issue RST and make ACK acceptable to originator of segment. * The mbuf must still include the original packet header. * tp may be NULL. */ static void tcp_dropwithreset(struct mbuf *m, struct tcphdr *th, struct tcpcb *tp, int tlen, int rstreason) { #ifdef INET struct ip *ip; #endif #ifdef INET6 struct ip6_hdr *ip6; #endif if (tp != NULL) { INP_WLOCK_ASSERT(tp->t_inpcb); } /* Don't bother if destination was broadcast/multicast. */ if ((th->th_flags & TH_RST) || m->m_flags & (M_BCAST|M_MCAST)) goto drop; #ifdef INET6 if (mtod(m, struct ip *)->ip_v == 6) { ip6 = mtod(m, struct ip6_hdr *); if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || IN6_IS_ADDR_MULTICAST(&ip6->ip6_src)) goto drop; /* IPv6 anycast check is done at tcp6_input() */ } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { ip = mtod(m, struct ip *); if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) || IN_MULTICAST(ntohl(ip->ip_src.s_addr)) || ip->ip_src.s_addr == htonl(INADDR_BROADCAST) || in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) goto drop; } #endif /* Perform bandwidth limiting. */ if (badport_bandlim(rstreason) < 0) goto drop; /* tcp_respond consumes the mbuf chain. */ if (th->th_flags & TH_ACK) { tcp_respond(tp, mtod(m, void *), th, m, (tcp_seq)0, th->th_ack, TH_RST); } else { if (th->th_flags & TH_SYN) tlen++; tcp_respond(tp, mtod(m, void *), th, m, th->th_seq+tlen, (tcp_seq)0, TH_RST|TH_ACK); } return; drop: m_freem(m); } /* * Parse TCP options and place in tcpopt. */ static void tcp_dooptions(struct tcpopt *to, u_char *cp, int cnt, int flags) { int opt, optlen; to->to_flags = 0; for (; cnt > 0; cnt -= optlen, cp += optlen) { opt = cp[0]; if (opt == TCPOPT_EOL) break; if (opt == TCPOPT_NOP) optlen = 1; else { if (cnt < 2) break; optlen = cp[1]; if (optlen < 2 || optlen > cnt) break; } switch (opt) { case TCPOPT_MAXSEG: if (optlen != TCPOLEN_MAXSEG) continue; if (!(flags & TO_SYN)) continue; to->to_flags |= TOF_MSS; bcopy((char *)cp + 2, (char *)&to->to_mss, sizeof(to->to_mss)); to->to_mss = ntohs(to->to_mss); break; case TCPOPT_WINDOW: if (optlen != TCPOLEN_WINDOW) continue; if (!(flags & TO_SYN)) continue; to->to_flags |= TOF_SCALE; to->to_wscale = min(cp[2], TCP_MAX_WINSHIFT); break; case TCPOPT_TIMESTAMP: if (optlen != TCPOLEN_TIMESTAMP) continue; to->to_flags |= TOF_TS; bcopy((char *)cp + 2, (char *)&to->to_tsval, sizeof(to->to_tsval)); to->to_tsval = ntohl(to->to_tsval); bcopy((char *)cp + 6, (char *)&to->to_tsecr, sizeof(to->to_tsecr)); to->to_tsecr = ntohl(to->to_tsecr); break; #ifdef TCP_SIGNATURE /* * XXX In order to reply to a host which has set the * TCP_SIGNATURE option in its initial SYN, we have to * record the fact that the option was observed here * for the syncache code to perform the correct response. */ case TCPOPT_SIGNATURE: if (optlen != TCPOLEN_SIGNATURE) continue; to->to_flags |= TOF_SIGNATURE; to->to_signature = cp + 2; break; #endif case TCPOPT_SACK_PERMITTED: if (optlen != TCPOLEN_SACK_PERMITTED) continue; if (!(flags & TO_SYN)) continue; if (!V_tcp_do_sack) continue; to->to_flags |= TOF_SACKPERM; break; case TCPOPT_SACK: if (optlen <= 2 || (optlen - 2) % TCPOLEN_SACK != 0) continue; if (flags & TO_SYN) continue; to->to_flags |= TOF_SACK; to->to_nsacks = (optlen - 2) / TCPOLEN_SACK; to->to_sacks = cp + 2; TCPSTAT_INC(tcps_sack_rcv_blocks); break; default: continue; } } } /* * Pull out of band byte out of a segment so * it doesn't appear in the user's data queue. * It is still reflected in the segment length for * sequencing purposes. */ static void tcp_pulloutofband(struct socket *so, struct tcphdr *th, struct mbuf *m, int off) { int cnt = off + th->th_urp - 1; while (cnt >= 0) { if (m->m_len > cnt) { char *cp = mtod(m, caddr_t) + cnt; struct tcpcb *tp = sototcpcb(so); INP_WLOCK_ASSERT(tp->t_inpcb); tp->t_iobc = *cp; tp->t_oobflags |= TCPOOB_HAVEDATA; bcopy(cp+1, cp, (unsigned)(m->m_len - cnt - 1)); m->m_len--; if (m->m_flags & M_PKTHDR) m->m_pkthdr.len--; return; } cnt -= m->m_len; m = m->m_next; if (m == NULL) break; } panic("tcp_pulloutofband"); } /* * Collect new round-trip time estimate * and update averages and current timeout. */ static void tcp_xmit_timer(struct tcpcb *tp, int rtt) { int delta; INP_WLOCK_ASSERT(tp->t_inpcb); TCPSTAT_INC(tcps_rttupdated); tp->t_rttupdated++; if (tp->t_srtt != 0) { /* * srtt is stored as fixed point with 5 bits after the * binary point (i.e., scaled by 8). The following magic * is equivalent to the smoothing algorithm in rfc793 with * an alpha of .875 (srtt = rtt/8 + srtt*7/8 in fixed * point). Adjust rtt to origin 0. */ delta = ((rtt - 1) << TCP_DELTA_SHIFT) - (tp->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)); if ((tp->t_srtt += delta) <= 0) tp->t_srtt = 1; /* * We accumulate a smoothed rtt variance (actually, a * smoothed mean difference), then set the retransmit * timer to smoothed rtt + 4 times the smoothed variance. * rttvar is stored as fixed point with 4 bits after the * binary point (scaled by 16). The following is * equivalent to rfc793 smoothing with an alpha of .75 * (rttvar = rttvar*3/4 + |delta| / 4). This replaces * rfc793's wired-in beta. */ if (delta < 0) delta = -delta; delta -= tp->t_rttvar >> (TCP_RTTVAR_SHIFT - TCP_DELTA_SHIFT); if ((tp->t_rttvar += delta) <= 0) tp->t_rttvar = 1; if (tp->t_rttbest > tp->t_srtt + tp->t_rttvar) tp->t_rttbest = tp->t_srtt + tp->t_rttvar; } else { /* * No rtt measurement yet - use the unsmoothed rtt. * Set the variance to half the rtt (so our first * retransmit happens at 3*rtt). */ tp->t_srtt = rtt << TCP_RTT_SHIFT; tp->t_rttvar = rtt << (TCP_RTTVAR_SHIFT - 1); tp->t_rttbest = tp->t_srtt + tp->t_rttvar; } tp->t_rtttime = 0; tp->t_rxtshift = 0; /* * the retransmit should happen at rtt + 4 * rttvar. * Because of the way we do the smoothing, srtt and rttvar * will each average +1/2 tick of bias. When we compute * the retransmit timer, we want 1/2 tick of rounding and * 1 extra tick because of +-1/2 tick uncertainty in the * firing of the timer. The bias will give us exactly the * 1.5 tick we need. But, because the bias is * statistical, we have to test that we don't drop below * the minimum feasible timer (which is 2 ticks). */ TCPT_RANGESET(tp->t_rxtcur, TCP_REXMTVAL(tp), max(tp->t_rttmin, rtt + 2), TCPTV_REXMTMAX); /* * We received an ack for a packet that wasn't retransmitted; * it is probably safe to discard any error indications we've * received recently. This isn't quite right, but close enough * for now (a route might have failed after we sent a segment, * and the return path might not be symmetrical). */ tp->t_softerror = 0; } /* * Determine a reasonable value for maxseg size. * If the route is known, check route for mtu. * If none, use an mss that can be handled on the outgoing interface * without forcing IP to fragment. If no route is found, route has no mtu, * or the destination isn't local, use a default, hopefully conservative * size (usually 512 or the default IP max size, but no more than the mtu * of the interface), as we can't discover anything about intervening * gateways or networks. We also initialize the congestion/slow start * window to be a single segment if the destination isn't local. * While looking at the routing entry, we also initialize other path-dependent * parameters from pre-set or cached values in the routing entry. * * Also take into account the space needed for options that we * send regularly. Make maxseg shorter by that amount to assure * that we can send maxseg amount of data even when the options * are present. Store the upper limit of the length of options plus * data in maxopd. * * NOTE that this routine is only called when we process an incoming * segment, or an ICMP need fragmentation datagram. Outgoing SYN/ACK MSS * settings are handled in tcp_mssopt(). */ void tcp_mss_update(struct tcpcb *tp, int offer, int mtuoffer, struct hc_metrics_lite *metricptr, struct tcp_ifcap *cap) { int mss = 0; u_long maxmtu = 0; struct inpcb *inp = tp->t_inpcb; struct hc_metrics_lite metrics; int origoffer; #ifdef INET6 int isipv6 = ((inp->inp_vflag & INP_IPV6) != 0) ? 1 : 0; size_t min_protoh = isipv6 ? sizeof (struct ip6_hdr) + sizeof (struct tcphdr) : sizeof (struct tcpiphdr); #else const size_t min_protoh = sizeof(struct tcpiphdr); #endif INP_WLOCK_ASSERT(tp->t_inpcb); if (mtuoffer != -1) { KASSERT(offer == -1, ("%s: conflict", __func__)); offer = mtuoffer - min_protoh; } origoffer = offer; /* Initialize. */ #ifdef INET6 if (isipv6) { maxmtu = tcp_maxmtu6(&inp->inp_inc, cap); tp->t_maxopd = tp->t_maxseg = V_tcp_v6mssdflt; } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { maxmtu = tcp_maxmtu(&inp->inp_inc, cap); tp->t_maxopd = tp->t_maxseg = V_tcp_mssdflt; } #endif /* * No route to sender, stay with default mss and return. */ if (maxmtu == 0) { /* * In case we return early we need to initialize metrics * to a defined state as tcp_hc_get() would do for us * if there was no cache hit. */ if (metricptr != NULL) bzero(metricptr, sizeof(struct hc_metrics_lite)); return; } /* What have we got? */ switch (offer) { case 0: /* * Offer == 0 means that there was no MSS on the SYN * segment, in this case we use tcp_mssdflt as * already assigned to t_maxopd above. */ offer = tp->t_maxopd; break; case -1: /* * Offer == -1 means that we didn't receive SYN yet. */ /* FALLTHROUGH */ default: /* * Prevent DoS attack with too small MSS. Round up * to at least minmss. */ offer = max(offer, V_tcp_minmss); } /* * rmx information is now retrieved from tcp_hostcache. */ tcp_hc_get(&inp->inp_inc, &metrics); if (metricptr != NULL) bcopy(&metrics, metricptr, sizeof(struct hc_metrics_lite)); /* * If there's a discovered mtu in tcp hostcache, use it. * Else, use the link mtu. */ if (metrics.rmx_mtu) mss = min(metrics.rmx_mtu, maxmtu) - min_protoh; else { #ifdef INET6 if (isipv6) { mss = maxmtu - min_protoh; if (!V_path_mtu_discovery && !in6_localaddr(&inp->in6p_faddr)) mss = min(mss, V_tcp_v6mssdflt); } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { mss = maxmtu - min_protoh; if (!V_path_mtu_discovery && !in_localaddr(inp->inp_faddr)) mss = min(mss, V_tcp_mssdflt); } #endif /* * XXX - The above conditional (mss = maxmtu - min_protoh) * probably violates the TCP spec. * The problem is that, since we don't know the * other end's MSS, we are supposed to use a conservative * default. But, if we do that, then MTU discovery will * never actually take place, because the conservative * default is much less than the MTUs typically seen * on the Internet today. For the moment, we'll sweep * this under the carpet. * * The conservative default might not actually be a problem * if the only case this occurs is when sending an initial * SYN with options and data to a host we've never talked * to before. Then, they will reply with an MSS value which * will get recorded and the new parameters should get * recomputed. For Further Study. */ } mss = min(mss, offer); /* * Sanity check: make sure that maxopd will be large * enough to allow some data on segments even if the * all the option space is used (40bytes). Otherwise * funny things may happen in tcp_output. */ mss = max(mss, 64); /* * maxopd stores the maximum length of data AND options * in a segment; maxseg is the amount of data in a normal * segment. We need to store this value (maxopd) apart * from maxseg, because now every segment carries options * and thus we normally have somewhat less data in segments. */ tp->t_maxopd = mss; /* * origoffer==-1 indicates that no segments were received yet. * In this case we just guess. */ if ((tp->t_flags & (TF_REQ_TSTMP|TF_NOOPT)) == TF_REQ_TSTMP && (origoffer == -1 || (tp->t_flags & TF_RCVD_TSTMP) == TF_RCVD_TSTMP)) mss -= TCPOLEN_TSTAMP_APPA; tp->t_maxseg = mss; } void tcp_mss(struct tcpcb *tp, int offer) { int mss; u_long bufsize; struct inpcb *inp; struct socket *so; struct hc_metrics_lite metrics; struct tcp_ifcap cap; KASSERT(tp != NULL, ("%s: tp == NULL", __func__)); bzero(&cap, sizeof(cap)); tcp_mss_update(tp, offer, -1, &metrics, &cap); mss = tp->t_maxseg; inp = tp->t_inpcb; /* * If there's a pipesize, change the socket buffer to that size, * don't change if sb_hiwat is different than default (then it * has been changed on purpose with setsockopt). * Make the socket buffers an integral number of mss units; * if the mss is larger than the socket buffer, decrease the mss. */ so = inp->inp_socket; SOCKBUF_LOCK(&so->so_snd); if ((so->so_snd.sb_hiwat == V_tcp_sendspace) && metrics.rmx_sendpipe) bufsize = metrics.rmx_sendpipe; else bufsize = so->so_snd.sb_hiwat; if (bufsize < mss) mss = bufsize; else { bufsize = roundup(bufsize, mss); if (bufsize > sb_max) bufsize = sb_max; if (bufsize > so->so_snd.sb_hiwat) (void)sbreserve_locked(&so->so_snd, bufsize, so, NULL); } SOCKBUF_UNLOCK(&so->so_snd); tp->t_maxseg = mss; SOCKBUF_LOCK(&so->so_rcv); if ((so->so_rcv.sb_hiwat == V_tcp_recvspace) && metrics.rmx_recvpipe) bufsize = metrics.rmx_recvpipe; else bufsize = so->so_rcv.sb_hiwat; if (bufsize > mss) { bufsize = roundup(bufsize, mss); if (bufsize > sb_max) bufsize = sb_max; if (bufsize > so->so_rcv.sb_hiwat) (void)sbreserve_locked(&so->so_rcv, bufsize, so, NULL); } SOCKBUF_UNLOCK(&so->so_rcv); /* Check the interface for TSO capabilities. */ if (cap.ifcap & CSUM_TSO) { tp->t_flags |= TF_TSO; tp->t_tsomax = cap.tsomax; tp->t_tsomaxsegcount = cap.tsomaxsegcount; tp->t_tsomaxsegsize = cap.tsomaxsegsize; } } /* * Determine the MSS option to send on an outgoing SYN. */ int tcp_mssopt(struct in_conninfo *inc) { int mss = 0; u_long maxmtu = 0; u_long thcmtu = 0; size_t min_protoh; KASSERT(inc != NULL, ("tcp_mssopt with NULL in_conninfo pointer")); #ifdef INET6 if (inc->inc_flags & INC_ISIPV6) { mss = V_tcp_v6mssdflt; maxmtu = tcp_maxmtu6(inc, NULL); min_protoh = sizeof(struct ip6_hdr) + sizeof(struct tcphdr); } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { mss = V_tcp_mssdflt; maxmtu = tcp_maxmtu(inc, NULL); min_protoh = sizeof(struct tcpiphdr); } #endif #if defined(INET6) || defined(INET) thcmtu = tcp_hc_getmtu(inc); /* IPv4 and IPv6 */ #endif if (maxmtu && thcmtu) mss = min(maxmtu, thcmtu) - min_protoh; else if (maxmtu || thcmtu) mss = max(maxmtu, thcmtu) - min_protoh; return (mss); } /* * On a partial ack arrives, force the retransmission of the * next unacknowledged segment. Do not clear tp->t_dupacks. * By setting snd_nxt to ti_ack, this forces retransmission timer to * be started again. */ static void tcp_newreno_partial_ack(struct tcpcb *tp, struct tcphdr *th) { tcp_seq onxt = tp->snd_nxt; u_long ocwnd = tp->snd_cwnd; INP_WLOCK_ASSERT(tp->t_inpcb); tcp_timer_activate(tp, TT_REXMT, 0); tp->t_rtttime = 0; tp->snd_nxt = th->th_ack; /* * Set snd_cwnd to one segment beyond acknowledged offset. * (tp->snd_una has not yet been updated when this function is called.) */ tp->snd_cwnd = tp->t_maxseg + BYTES_THIS_ACK(tp, th); tp->t_flags |= TF_ACKNOW; (void) tcp_output(tp); tp->snd_cwnd = ocwnd; if (SEQ_GT(onxt, tp->snd_nxt)) tp->snd_nxt = onxt; /* * Partial window deflation. Relies on fact that tp->snd_una * not updated yet. */ if (tp->snd_cwnd > BYTES_THIS_ACK(tp, th)) tp->snd_cwnd -= BYTES_THIS_ACK(tp, th); else tp->snd_cwnd = 0; tp->snd_cwnd += tp->t_maxseg; } Index: head/sys/netinet/tcp_reass.c =================================================================== --- head/sys/netinet/tcp_reass.c (revision 288300) +++ head/sys/netinet/tcp_reass.c (revision 288301) @@ -1,333 +1,332 @@ /*- * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1994, 1995 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)tcp_input.c 8.12 (Berkeley) 5/24/95 */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_tcpdebug.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef TCPDEBUG #include #endif /* TCPDEBUG */ static SYSCTL_NODE(_net_inet_tcp, OID_AUTO, reass, CTLFLAG_RW, 0, "TCP Segment Reassembly Queue"); static int tcp_reass_maxseg = 0; SYSCTL_INT(_net_inet_tcp_reass, OID_AUTO, maxsegments, CTLFLAG_RDTUN, &tcp_reass_maxseg, 0, "Global maximum number of TCP Segments in Reassembly Queue"); static uma_zone_t tcp_reass_zone; SYSCTL_UMA_CUR(_net_inet_tcp_reass, OID_AUTO, cursegments, 0, &tcp_reass_zone, "Global number of TCP Segments currently in Reassembly Queue"); /* Initialize TCP reassembly queue */ static void tcp_reass_zone_change(void *tag) { /* Set the zone limit and read back the effective value. */ tcp_reass_maxseg = nmbclusters / 16; tcp_reass_maxseg = uma_zone_set_max(tcp_reass_zone, tcp_reass_maxseg); } void tcp_reass_global_init(void) { tcp_reass_maxseg = nmbclusters / 16; TUNABLE_INT_FETCH("net.inet.tcp.reass.maxsegments", &tcp_reass_maxseg); tcp_reass_zone = uma_zcreate("tcpreass", sizeof (struct tseg_qent), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); /* Set the zone limit and read back the effective value. */ tcp_reass_maxseg = uma_zone_set_max(tcp_reass_zone, tcp_reass_maxseg); EVENTHANDLER_REGISTER(nmbclusters_change, tcp_reass_zone_change, NULL, EVENTHANDLER_PRI_ANY); } void tcp_reass_flush(struct tcpcb *tp) { struct tseg_qent *qe; INP_WLOCK_ASSERT(tp->t_inpcb); while ((qe = LIST_FIRST(&tp->t_segq)) != NULL) { LIST_REMOVE(qe, tqe_q); m_freem(qe->tqe_m); uma_zfree(tcp_reass_zone, qe); tp->t_segqlen--; } KASSERT((tp->t_segqlen == 0), ("TCP reass queue %p segment count is %d instead of 0 after flush.", tp, tp->t_segqlen)); } int tcp_reass(struct tcpcb *tp, struct tcphdr *th, int *tlenp, struct mbuf *m) { struct tseg_qent *q; struct tseg_qent *p = NULL; struct tseg_qent *nq; struct tseg_qent *te = NULL; struct socket *so = tp->t_inpcb->inp_socket; char *s = NULL; int flags; struct tseg_qent tqs; INP_WLOCK_ASSERT(tp->t_inpcb); /* * XXX: tcp_reass() is rather inefficient with its data structures * and should be rewritten (see NetBSD for optimizations). */ /* * Call with th==NULL after become established to * force pre-ESTABLISHED data up to user socket. */ if (th == NULL) goto present; /* * Limit the number of segments that can be queued to reduce the * potential for mbuf exhaustion. For best performance, we want to be * able to queue a full window's worth of segments. The size of the * socket receive buffer determines our advertised window and grows * automatically when socket buffer autotuning is enabled. Use it as the * basis for our queue limit. * Always let the missing segment through which caused this queue. * NB: Access to the socket buffer is left intentionally unlocked as we * can tolerate stale information here. * * XXXLAS: Using sbspace(so->so_rcv) instead of so->so_rcv.sb_hiwat * should work but causes packets to be dropped when they shouldn't. * Investigate why and re-evaluate the below limit after the behaviour * is understood. */ if ((th->th_seq != tp->rcv_nxt || !TCPS_HAVEESTABLISHED(tp->t_state)) && tp->t_segqlen >= (so->so_rcv.sb_hiwat / tp->t_maxseg) + 1) { TCPSTAT_INC(tcps_rcvreassfull); *tlenp = 0; if ((s = tcp_log_addrs(&tp->t_inpcb->inp_inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: queue limit reached, " "segment dropped\n", s, __func__); free(s, M_TCPLOG); } m_freem(m); return (0); } /* * Allocate a new queue entry. If we can't, or hit the zone limit * just drop the pkt. * * Use a temporary structure on the stack for the missing segment * when the zone is exhausted. Otherwise we may get stuck. */ te = uma_zalloc(tcp_reass_zone, M_NOWAIT); if (te == NULL) { if (th->th_seq != tp->rcv_nxt || !TCPS_HAVEESTABLISHED(tp->t_state)) { TCPSTAT_INC(tcps_rcvmemdrop); m_freem(m); *tlenp = 0; if ((s = tcp_log_addrs(&tp->t_inpcb->inp_inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: global zone limit " "reached, segment dropped\n", s, __func__); free(s, M_TCPLOG); } return (0); } else { bzero(&tqs, sizeof(struct tseg_qent)); te = &tqs; if ((s = tcp_log_addrs(&tp->t_inpcb->inp_inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: global zone limit reached, using " "stack for missing segment\n", s, __func__); free(s, M_TCPLOG); } } } tp->t_segqlen++; /* * Find a segment which begins after this one does. */ LIST_FOREACH(q, &tp->t_segq, tqe_q) { if (SEQ_GT(q->tqe_th->th_seq, th->th_seq)) break; p = q; } /* * If there is a preceding segment, it may provide some of * our data already. If so, drop the data from the incoming * segment. If it provides all of our data, drop us. */ if (p != NULL) { int i; /* conversion to int (in i) handles seq wraparound */ i = p->tqe_th->th_seq + p->tqe_len - th->th_seq; if (i > 0) { if (i >= *tlenp) { TCPSTAT_INC(tcps_rcvduppack); TCPSTAT_ADD(tcps_rcvdupbyte, *tlenp); m_freem(m); if (te != &tqs) uma_zfree(tcp_reass_zone, te); tp->t_segqlen--; /* * Try to present any queued data * at the left window edge to the user. * This is needed after the 3-WHS * completes. */ goto present; /* ??? */ } m_adj(m, i); *tlenp -= i; th->th_seq += i; } } tp->t_rcvoopack++; TCPSTAT_INC(tcps_rcvoopack); TCPSTAT_ADD(tcps_rcvoobyte, *tlenp); /* * While we overlap succeeding segments trim them or, * if they are completely covered, dequeue them. */ while (q) { int i = (th->th_seq + *tlenp) - q->tqe_th->th_seq; if (i <= 0) break; if (i < q->tqe_len) { q->tqe_th->th_seq += i; q->tqe_len -= i; m_adj(q->tqe_m, i); break; } nq = LIST_NEXT(q, tqe_q); LIST_REMOVE(q, tqe_q); m_freem(q->tqe_m); uma_zfree(tcp_reass_zone, q); tp->t_segqlen--; q = nq; } /* Insert the new segment queue entry into place. */ te->tqe_m = m; te->tqe_th = th; te->tqe_len = *tlenp; if (p == NULL) { LIST_INSERT_HEAD(&tp->t_segq, te, tqe_q); } else { KASSERT(te != &tqs, ("%s: temporary stack based entry not " "first element in queue", __func__)); LIST_INSERT_AFTER(p, te, tqe_q); } present: /* * Present data to user, advancing rcv_nxt through * completed sequence space. */ if (!TCPS_HAVEESTABLISHED(tp->t_state)) return (0); q = LIST_FIRST(&tp->t_segq); if (!q || q->tqe_th->th_seq != tp->rcv_nxt) return (0); SOCKBUF_LOCK(&so->so_rcv); do { tp->rcv_nxt += q->tqe_len; flags = q->tqe_th->th_flags & TH_FIN; nq = LIST_NEXT(q, tqe_q); LIST_REMOVE(q, tqe_q); if (so->so_rcv.sb_state & SBS_CANTRCVMORE) m_freem(q->tqe_m); else sbappendstream_locked(&so->so_rcv, q->tqe_m, 0); if (q != &tqs) uma_zfree(tcp_reass_zone, q); tp->t_segqlen--; q = nq; } while (q && q->tqe_th->th_seq == tp->rcv_nxt); - ND6_HINT(tp); sorwakeup_locked(so); return (flags); } Index: head/sys/netinet/tcp_var.h =================================================================== --- head/sys/netinet/tcp_var.h (revision 288300) +++ head/sys/netinet/tcp_var.h (revision 288301) @@ -1,773 +1,761 @@ /*- * Copyright (c) 1982, 1986, 1993, 1994, 1995 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95 * $FreeBSD$ */ #ifndef _NETINET_TCP_VAR_H_ #define _NETINET_TCP_VAR_H_ #include #ifdef _KERNEL #include /* * Kernel variables for tcp. */ VNET_DECLARE(int, tcp_do_rfc1323); #define V_tcp_do_rfc1323 VNET(tcp_do_rfc1323) #endif /* _KERNEL */ /* TCP segment queue entry */ struct tseg_qent { LIST_ENTRY(tseg_qent) tqe_q; int tqe_len; /* TCP segment data length */ struct tcphdr *tqe_th; /* a pointer to tcp header */ struct mbuf *tqe_m; /* mbuf contains packet */ }; LIST_HEAD(tsegqe_head, tseg_qent); struct sackblk { tcp_seq start; /* start seq no. of sack block */ tcp_seq end; /* end seq no. */ }; struct sackhole { tcp_seq start; /* start seq no. of hole */ tcp_seq end; /* end seq no. */ tcp_seq rxmit; /* next seq. no in hole to be retransmitted */ TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */ }; struct sackhint { struct sackhole *nexthole; int sack_bytes_rexmit; tcp_seq last_sack_ack; /* Most recent/largest sacked ack */ int ispare; /* explicit pad for 64bit alignment */ uint64_t _pad[2]; /* 1 sacked_bytes, 1 TBD */ }; struct tcptemp { u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ struct tcphdr tt_t; }; #define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ -/* Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. */ -#ifdef INET6 -#define ND6_HINT(tp) \ -do { \ - if ((tp) && (tp)->t_inpcb && \ - ((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0) \ - nd6_nud_hint(NULL, NULL, 0); \ -} while (0) -#else -#define ND6_HINT(tp) -#endif - /* * Tcp control block, one per tcp; fields: * Organized for 16 byte cacheline efficiency. */ struct tcpcb { struct tsegqe_head t_segq; /* segment reassembly queue */ void *t_pspare[2]; /* new reassembly queue */ int t_segqlen; /* segment reassembly queue length */ int t_dupacks; /* consecutive dup acks recd */ struct tcp_timer *t_timers; /* All the TCP timers in one struct */ struct inpcb *t_inpcb; /* back pointer to internet pcb */ int t_state; /* state of this connection */ u_int t_flags; struct vnet *t_vnet; /* back pointer to parent vnet */ tcp_seq snd_una; /* sent but unacknowledged */ tcp_seq snd_max; /* highest sequence number sent; * used to recognize retransmits */ tcp_seq snd_nxt; /* send next */ tcp_seq snd_up; /* send urgent pointer */ tcp_seq snd_wl1; /* window update seg seq number */ tcp_seq snd_wl2; /* window update seg ack number */ tcp_seq iss; /* initial send sequence number */ tcp_seq irs; /* initial receive sequence number */ tcp_seq rcv_nxt; /* receive next */ tcp_seq rcv_adv; /* advertised window */ u_long rcv_wnd; /* receive window */ tcp_seq rcv_up; /* receive urgent pointer */ u_long snd_wnd; /* send window */ u_long snd_cwnd; /* congestion-controlled window */ u_long snd_spare1; /* unused */ u_long snd_ssthresh; /* snd_cwnd size threshold for * for slow start exponential to * linear switch */ u_long snd_spare2; /* unused */ tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ u_int t_maxopd; /* mss plus options */ u_int t_rcvtime; /* inactivity time */ u_int t_starttime; /* time connection was established */ u_int t_rtttime; /* RTT measurement start time */ tcp_seq t_rtseq; /* sequence number being timed */ u_int t_bw_spare1; /* unused */ tcp_seq t_bw_spare2; /* unused */ int t_rxtcur; /* current retransmit value (ticks) */ u_int t_maxseg; /* maximum segment size */ int t_srtt; /* smoothed round-trip time */ int t_rttvar; /* variance in round-trip time */ int t_rxtshift; /* log(2) of rexmt exp. backoff */ u_int t_rttmin; /* minimum rtt allowed */ u_int t_rttbest; /* best rtt we've seen */ u_long t_rttupdated; /* number of times rtt sampled */ u_long max_sndwnd; /* largest window peer has offered */ int t_softerror; /* possible error not yet reported */ /* out-of-band data */ char t_oobflags; /* have some */ char t_iobc; /* input character */ /* RFC 1323 variables */ u_char snd_scale; /* window scaling for send window */ u_char rcv_scale; /* window scaling for recv window */ u_char request_r_scale; /* pending window scaling */ u_int32_t ts_recent; /* timestamp echo data */ u_int ts_recent_age; /* when last updated */ u_int32_t ts_offset; /* our timestamp offset */ tcp_seq last_ack_sent; /* experimental */ u_long snd_cwnd_prev; /* cwnd prior to retransmit */ u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */ tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ int t_sndzerowin; /* zero-window updates sent */ u_int t_badrxtwin; /* window for retransmit recovery */ u_char snd_limited; /* segments limited transmitted */ /* SACK related state */ int snd_numholes; /* number of holes seen by sender */ TAILQ_HEAD(sackhole_head, sackhole) snd_holes; /* SACK scoreboard (sorted) */ tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/ int rcv_numsacks; /* # distinct sack blks present */ struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */ tcp_seq sack_newdata; /* New data xmitted in this recovery episode starts at this seq number */ struct sackhint sackhint; /* SACK scoreboard hint */ int t_rttlow; /* smallest observerved RTT */ u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */ int rfbuf_cnt; /* recv buffer autoscaling byte count */ struct toedev *tod; /* toedev handling this connection */ int t_sndrexmitpack; /* retransmit packets sent */ int t_rcvoopack; /* out-of-order packets received */ void *t_toe; /* TOE pcb pointer */ int t_bytes_acked; /* # bytes acked during current RTT */ struct cc_algo *cc_algo; /* congestion control algorithm */ struct cc_var *ccv; /* congestion control specific vars */ struct osd *osd; /* storage for Khelp module data */ u_int t_keepinit; /* time to establish connection */ u_int t_keepidle; /* time before keepalive probes begin */ u_int t_keepintvl; /* interval between keepalives */ u_int t_keepcnt; /* number of keepalives before close */ u_int t_tsomax; /* TSO total burst length limit in bytes */ u_int t_tsomaxsegcount; /* TSO maximum segment count */ u_int t_tsomaxsegsize; /* TSO maximum segment size in bytes */ u_int t_pmtud_saved_maxopd; /* pre-blackhole MSS */ u_int t_flags2; /* More tcpcb flags storage */ uint32_t t_ispare[8]; /* 5 UTO, 3 TBD */ void *t_pspare2[4]; /* 1 TCP_SIGNATURE, 3 TBD */ uint64_t _pad[6]; /* 6 TBD (1-2 CC/RTT?) */ }; /* * Flags and utility macros for the t_flags field. */ #define TF_ACKNOW 0x000001 /* ack peer immediately */ #define TF_DELACK 0x000002 /* ack, but try to delay it */ #define TF_NODELAY 0x000004 /* don't delay packets to coalesce */ #define TF_NOOPT 0x000008 /* don't use tcp options */ #define TF_SENTFIN 0x000010 /* have sent FIN */ #define TF_REQ_SCALE 0x000020 /* have/will request window scaling */ #define TF_RCVD_SCALE 0x000040 /* other side has requested scaling */ #define TF_REQ_TSTMP 0x000080 /* have/will request timestamps */ #define TF_RCVD_TSTMP 0x000100 /* a timestamp was received in SYN */ #define TF_SACK_PERMIT 0x000200 /* other side said I could SACK */ #define TF_NEEDSYN 0x000400 /* send SYN (implicit state) */ #define TF_NEEDFIN 0x000800 /* send FIN (implicit state) */ #define TF_NOPUSH 0x001000 /* don't push */ #define TF_PREVVALID 0x002000 /* saved values for bad rxmit valid */ #define TF_MORETOCOME 0x010000 /* More data to be appended to sock */ #define TF_LQ_OVERFLOW 0x020000 /* listen queue overflow */ #define TF_LASTIDLE 0x040000 /* connection was previously idle */ #define TF_RXWIN0SENT 0x080000 /* sent a receiver win 0 in response */ #define TF_FASTRECOVERY 0x100000 /* in NewReno Fast Recovery */ #define TF_WASFRECOVERY 0x200000 /* was in NewReno Fast Recovery */ #define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */ #define TF_FORCEDATA 0x800000 /* force out a byte */ #define TF_TSO 0x1000000 /* TSO enabled on this connection */ #define TF_TOE 0x2000000 /* this connection is offloaded */ #define TF_ECN_PERMIT 0x4000000 /* connection ECN-ready */ #define TF_ECN_SND_CWR 0x8000000 /* ECN CWR in queue */ #define TF_ECN_SND_ECE 0x10000000 /* ECN ECE in queue */ #define TF_CONGRECOVERY 0x20000000 /* congestion recovery mode */ #define TF_WASCRECOVERY 0x40000000 /* was in congestion recovery */ #define IN_FASTRECOVERY(t_flags) (t_flags & TF_FASTRECOVERY) #define ENTER_FASTRECOVERY(t_flags) t_flags |= TF_FASTRECOVERY #define EXIT_FASTRECOVERY(t_flags) t_flags &= ~TF_FASTRECOVERY #define IN_CONGRECOVERY(t_flags) (t_flags & TF_CONGRECOVERY) #define ENTER_CONGRECOVERY(t_flags) t_flags |= TF_CONGRECOVERY #define EXIT_CONGRECOVERY(t_flags) t_flags &= ~TF_CONGRECOVERY #define IN_RECOVERY(t_flags) (t_flags & (TF_CONGRECOVERY | TF_FASTRECOVERY)) #define ENTER_RECOVERY(t_flags) t_flags |= (TF_CONGRECOVERY | TF_FASTRECOVERY) #define EXIT_RECOVERY(t_flags) t_flags &= ~(TF_CONGRECOVERY | TF_FASTRECOVERY) #define BYTES_THIS_ACK(tp, th) (th->th_ack - tp->snd_una) /* * Flags for the t_oobflags field. */ #define TCPOOB_HAVEDATA 0x01 #define TCPOOB_HADDATA 0x02 #ifdef TCP_SIGNATURE /* * Defines which are needed by the xform_tcp module and tcp_[in|out]put * for SADB verification and lookup. */ #define TCP_SIGLEN 16 /* length of computed digest in bytes */ #define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */ #define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */ /* * Only a single SA per host may be specified at this time. An SPI is * needed in order for the KEY_ALLOCSA() lookup to work. */ #define TCP_SIG_SPI 0x1000 #endif /* TCP_SIGNATURE */ /* * Flags for PLPMTU handling, t_flags2 */ #define TF2_PLPMTU_BLACKHOLE 0x00000001 /* Possible PLPMTUD Black Hole. */ #define TF2_PLPMTU_PMTUD 0x00000002 /* Allowed to attempt PLPMTUD. */ #define TF2_PLPMTU_MAXSEGSNT 0x00000004 /* Last seg sent was full seg. */ /* * Structure to hold TCP options that are only used during segment * processing (in tcp_input), but not held in the tcpcb. * It's basically used to reduce the number of parameters * to tcp_dooptions and tcp_addoptions. * The binary order of the to_flags is relevant for packing of the * options in tcp_addoptions. */ struct tcpopt { u_int64_t to_flags; /* which options are present */ #define TOF_MSS 0x0001 /* maximum segment size */ #define TOF_SCALE 0x0002 /* window scaling */ #define TOF_SACKPERM 0x0004 /* SACK permitted */ #define TOF_TS 0x0010 /* timestamp */ #define TOF_SIGNATURE 0x0040 /* TCP-MD5 signature option (RFC2385) */ #define TOF_SACK 0x0080 /* Peer sent SACK option */ #define TOF_MAXOPT 0x0100 u_int32_t to_tsval; /* new timestamp */ u_int32_t to_tsecr; /* reflected timestamp */ u_char *to_sacks; /* pointer to the first SACK blocks */ u_char *to_signature; /* pointer to the TCP-MD5 signature */ u_int16_t to_mss; /* maximum segment size */ u_int8_t to_wscale; /* window scaling */ u_int8_t to_nsacks; /* number of SACK blocks */ u_int32_t to_spare; /* UTO */ }; /* * Flags for tcp_dooptions. */ #define TO_SYN 0x01 /* parse SYN-only options */ struct hc_metrics_lite { /* must stay in sync with hc_metrics */ u_long rmx_mtu; /* MTU for this path */ u_long rmx_ssthresh; /* outbound gateway buffer limit */ u_long rmx_rtt; /* estimated round trip time */ u_long rmx_rttvar; /* estimated rtt variance */ u_long rmx_bandwidth; /* estimated bandwidth */ u_long rmx_cwnd; /* congestion window */ u_long rmx_sendpipe; /* outbound delay-bandwidth product */ u_long rmx_recvpipe; /* inbound delay-bandwidth product */ }; /* * Used by tcp_maxmtu() to communicate interface specific features * and limits at the time of connection setup. */ struct tcp_ifcap { int ifcap; u_int tsomax; u_int tsomaxsegcount; u_int tsomaxsegsize; }; #ifndef _NETINET_IN_PCB_H_ struct in_conninfo; #endif /* _NETINET_IN_PCB_H_ */ struct tcptw { struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */ tcp_seq snd_nxt; tcp_seq rcv_nxt; tcp_seq iss; tcp_seq irs; u_short last_win; /* cached window value */ u_short tw_so_options; /* copy of so_options */ struct ucred *tw_cred; /* user credentials */ u_int32_t t_recent; u_int32_t ts_offset; /* our timestamp offset */ u_int t_starttime; int tw_time; TAILQ_ENTRY(tcptw) tw_2msl; void *tw_pspare; /* TCP_SIGNATURE */ u_int *tw_spare; /* TCP_SIGNATURE */ }; #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) #define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb) #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) /* * The smoothed round-trip time and estimated variance * are stored as fixed point numbers scaled by the values below. * For convenience, these scales are also used in smoothing the average * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). * With these scales, srtt has 3 bits to the right of the binary point, * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the * binary point, and is smoothed with an ALPHA of 0.75. */ #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ /* * The initial retransmission should happen at rtt + 4 * rttvar. * Because of the way we do the smoothing, srtt and rttvar * will each average +1/2 tick of bias. When we compute * the retransmit timer, we want 1/2 tick of rounding and * 1 extra tick because of +-1/2 tick uncertainty in the * firing of the timer. The bias will give us exactly the * 1.5 tick we need. But, because the bias is * statistical, we have to test that we don't drop below * the minimum feasible timer (which is 2 ticks). * This version of the macro adapted from a paper by Lawrence * Brakmo and Larry Peterson which outlines a problem caused * by insufficient precision in the original implementation, * which results in inappropriately large RTO values for very * fast networks. */ #define TCP_REXMTVAL(tp) \ max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) /* * TCP statistics. * Many of these should be kept per connection, * but that's inconvenient at the moment. */ struct tcpstat { uint64_t tcps_connattempt; /* connections initiated */ uint64_t tcps_accepts; /* connections accepted */ uint64_t tcps_connects; /* connections established */ uint64_t tcps_drops; /* connections dropped */ uint64_t tcps_conndrops; /* embryonic connections dropped */ uint64_t tcps_minmssdrops; /* average minmss too low drops */ uint64_t tcps_closed; /* conn. closed (includes drops) */ uint64_t tcps_segstimed; /* segs where we tried to get rtt */ uint64_t tcps_rttupdated; /* times we succeeded */ uint64_t tcps_delack; /* delayed acks sent */ uint64_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ uint64_t tcps_rexmttimeo; /* retransmit timeouts */ uint64_t tcps_persisttimeo; /* persist timeouts */ uint64_t tcps_keeptimeo; /* keepalive timeouts */ uint64_t tcps_keepprobe; /* keepalive probes sent */ uint64_t tcps_keepdrops; /* connections dropped in keepalive */ uint64_t tcps_sndtotal; /* total packets sent */ uint64_t tcps_sndpack; /* data packets sent */ uint64_t tcps_sndbyte; /* data bytes sent */ uint64_t tcps_sndrexmitpack; /* data packets retransmitted */ uint64_t tcps_sndrexmitbyte; /* data bytes retransmitted */ uint64_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */ uint64_t tcps_sndacks; /* ack-only packets sent */ uint64_t tcps_sndprobe; /* window probes sent */ uint64_t tcps_sndurg; /* packets sent with URG only */ uint64_t tcps_sndwinup; /* window update-only packets sent */ uint64_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ uint64_t tcps_rcvtotal; /* total packets received */ uint64_t tcps_rcvpack; /* packets received in sequence */ uint64_t tcps_rcvbyte; /* bytes received in sequence */ uint64_t tcps_rcvbadsum; /* packets received with ccksum errs */ uint64_t tcps_rcvbadoff; /* packets received with bad offset */ uint64_t tcps_rcvreassfull; /* packets dropped for no reass space */ uint64_t tcps_rcvshort; /* packets received too short */ uint64_t tcps_rcvduppack; /* duplicate-only packets received */ uint64_t tcps_rcvdupbyte; /* duplicate-only bytes received */ uint64_t tcps_rcvpartduppack; /* packets with some duplicate data */ uint64_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ uint64_t tcps_rcvoopack; /* out-of-order packets received */ uint64_t tcps_rcvoobyte; /* out-of-order bytes received */ uint64_t tcps_rcvpackafterwin; /* packets with data after window */ uint64_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ uint64_t tcps_rcvafterclose; /* packets rcvd after "close" */ uint64_t tcps_rcvwinprobe; /* rcvd window probe packets */ uint64_t tcps_rcvdupack; /* rcvd duplicate acks */ uint64_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ uint64_t tcps_rcvackpack; /* rcvd ack packets */ uint64_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ uint64_t tcps_rcvwinupd; /* rcvd window update packets */ uint64_t tcps_pawsdrop; /* segments dropped due to PAWS */ uint64_t tcps_predack; /* times hdr predict ok for acks */ uint64_t tcps_preddat; /* times hdr predict ok for data pkts */ uint64_t tcps_pcbcachemiss; uint64_t tcps_cachedrtt; /* times cached RTT in route updated */ uint64_t tcps_cachedrttvar; /* times cached rttvar updated */ uint64_t tcps_cachedssthresh; /* times cached ssthresh updated */ uint64_t tcps_usedrtt; /* times RTT initialized from route */ uint64_t tcps_usedrttvar; /* times RTTVAR initialized from rt */ uint64_t tcps_usedssthresh; /* times ssthresh initialized from rt*/ uint64_t tcps_persistdrop; /* timeout in persist state */ uint64_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */ uint64_t tcps_mturesent; /* resends due to MTU discovery */ uint64_t tcps_listendrop; /* listen queue overflows */ uint64_t tcps_badrst; /* ignored RSTs in the window */ uint64_t tcps_sc_added; /* entry added to syncache */ uint64_t tcps_sc_retransmitted; /* syncache entry was retransmitted */ uint64_t tcps_sc_dupsyn; /* duplicate SYN packet */ uint64_t tcps_sc_dropped; /* could not reply to packet */ uint64_t tcps_sc_completed; /* successful extraction of entry */ uint64_t tcps_sc_bucketoverflow;/* syncache per-bucket limit hit */ uint64_t tcps_sc_cacheoverflow; /* syncache cache limit hit */ uint64_t tcps_sc_reset; /* RST removed entry from syncache */ uint64_t tcps_sc_stale; /* timed out or listen socket gone */ uint64_t tcps_sc_aborted; /* syncache entry aborted */ uint64_t tcps_sc_badack; /* removed due to bad ACK */ uint64_t tcps_sc_unreach; /* ICMP unreachable received */ uint64_t tcps_sc_zonefail; /* zalloc() failed */ uint64_t tcps_sc_sendcookie; /* SYN cookie sent */ uint64_t tcps_sc_recvcookie; /* SYN cookie received */ uint64_t tcps_hc_added; /* entry added to hostcache */ uint64_t tcps_hc_bucketoverflow;/* hostcache per bucket limit hit */ uint64_t tcps_finwait2_drops; /* Drop FIN_WAIT_2 connection after time limit */ /* SACK related stats */ uint64_t tcps_sack_recovery_episode; /* SACK recovery episodes */ uint64_t tcps_sack_rexmits; /* SACK rexmit segments */ uint64_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ uint64_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */ uint64_t tcps_sack_send_blocks; /* SACK blocks (options) sent */ uint64_t tcps_sack_sboverflow; /* times scoreboard overflowed */ /* ECN related stats */ uint64_t tcps_ecn_ce; /* ECN Congestion Experienced */ uint64_t tcps_ecn_ect0; /* ECN Capable Transport */ uint64_t tcps_ecn_ect1; /* ECN Capable Transport */ uint64_t tcps_ecn_shs; /* ECN successful handshakes */ uint64_t tcps_ecn_rcwnd; /* # times ECN reduced the cwnd */ /* TCP_SIGNATURE related stats */ uint64_t tcps_sig_rcvgoodsig; /* Total matching signature received */ uint64_t tcps_sig_rcvbadsig; /* Total bad signature received */ uint64_t tcps_sig_err_buildsig; /* Mismatching signature received */ uint64_t tcps_sig_err_sigopt; /* No signature expected by socket */ uint64_t tcps_sig_err_nosigopt; /* No signature provided by segment */ uint64_t _pad[12]; /* 6 UTO, 6 TBD */ }; #define tcps_rcvmemdrop tcps_rcvreassfull /* compat */ #ifdef _KERNEL #include VNET_PCPUSTAT_DECLARE(struct tcpstat, tcpstat); /* tcp statistics */ /* * In-kernel consumers can use these accessor macros directly to update * stats. */ #define TCPSTAT_ADD(name, val) \ VNET_PCPUSTAT_ADD(struct tcpstat, tcpstat, name, (val)) #define TCPSTAT_INC(name) TCPSTAT_ADD(name, 1) /* * Kernel module consumers must use this accessor macro. */ void kmod_tcpstat_inc(int statnum); #define KMOD_TCPSTAT_INC(name) \ kmod_tcpstat_inc(offsetof(struct tcpstat, name) / sizeof(uint64_t)) /* * TCP specific helper hook point identifiers. */ #define HHOOK_TCP_EST_IN 0 #define HHOOK_TCP_EST_OUT 1 #define HHOOK_TCP_LAST HHOOK_TCP_EST_OUT struct tcp_hhook_data { struct tcpcb *tp; struct tcphdr *th; struct tcpopt *to; long len; int tso; tcp_seq curack; }; #endif /* * TCB structure exported to user-land via sysctl(3). * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been * included. Not all of our clients do. */ #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_) struct xtcp_timer { int tt_rexmt; /* retransmit timer */ int tt_persist; /* retransmit persistence */ int tt_keep; /* keepalive */ int tt_2msl; /* 2*msl TIME_WAIT timer */ int tt_delack; /* delayed ACK timer */ int t_rcvtime; /* Time since last packet received */ }; struct xtcpcb { size_t xt_len; struct inpcb xt_inp; struct tcpcb xt_tp; struct xsocket xt_socket; struct xtcp_timer xt_timer; u_quad_t xt_alignment_hack; }; #endif /* * Identifiers for TCP sysctl nodes */ #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ #define TCPCTL_MSSDFLT 3 /* MSS default */ #define TCPCTL_STATS 4 /* statistics (read-only) */ #define TCPCTL_RTTDFLT 5 /* default RTT estimate */ #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ #define TCPCTL_SENDSPACE 8 /* send buffer space */ #define TCPCTL_RECVSPACE 9 /* receive buffer space */ #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ #define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */ #define TCPCTL_DROP 15 /* drop tcp connection */ #ifdef _KERNEL #ifdef SYSCTL_DECL SYSCTL_DECL(_net_inet_tcp); SYSCTL_DECL(_net_inet_tcp_sack); MALLOC_DECLARE(M_TCPLOG); #endif VNET_DECLARE(struct inpcbhead, tcb); /* queue of active tcpcb's */ VNET_DECLARE(struct inpcbinfo, tcbinfo); extern int tcp_log_in_vain; VNET_DECLARE(int, tcp_mssdflt); /* XXX */ VNET_DECLARE(int, tcp_minmss); VNET_DECLARE(int, tcp_delack_enabled); VNET_DECLARE(int, tcp_do_rfc3390); VNET_DECLARE(int, tcp_do_initcwnd10); VNET_DECLARE(int, tcp_sendspace); VNET_DECLARE(int, tcp_recvspace); VNET_DECLARE(int, path_mtu_discovery); VNET_DECLARE(int, tcp_do_rfc3465); VNET_DECLARE(int, tcp_abc_l_var); #define V_tcb VNET(tcb) #define V_tcbinfo VNET(tcbinfo) #define V_tcp_mssdflt VNET(tcp_mssdflt) #define V_tcp_minmss VNET(tcp_minmss) #define V_tcp_delack_enabled VNET(tcp_delack_enabled) #define V_tcp_do_rfc3390 VNET(tcp_do_rfc3390) #define V_tcp_do_initcwnd10 VNET(tcp_do_initcwnd10) #define V_tcp_sendspace VNET(tcp_sendspace) #define V_tcp_recvspace VNET(tcp_recvspace) #define V_path_mtu_discovery VNET(path_mtu_discovery) #define V_tcp_do_rfc3465 VNET(tcp_do_rfc3465) #define V_tcp_abc_l_var VNET(tcp_abc_l_var) VNET_DECLARE(int, tcp_do_sack); /* SACK enabled/disabled */ VNET_DECLARE(int, tcp_sc_rst_sock_fail); /* RST on sock alloc failure */ #define V_tcp_do_sack VNET(tcp_do_sack) #define V_tcp_sc_rst_sock_fail VNET(tcp_sc_rst_sock_fail) VNET_DECLARE(int, tcp_do_ecn); /* TCP ECN enabled/disabled */ VNET_DECLARE(int, tcp_ecn_maxretries); #define V_tcp_do_ecn VNET(tcp_do_ecn) #define V_tcp_ecn_maxretries VNET(tcp_ecn_maxretries) VNET_DECLARE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST + 1]); #define V_tcp_hhh VNET(tcp_hhh) int tcp_addoptions(struct tcpopt *, u_char *); int tcp_ccalgounload(struct cc_algo *unload_algo); struct tcpcb * tcp_close(struct tcpcb *); void tcp_discardcb(struct tcpcb *); void tcp_twstart(struct tcpcb *); void tcp_twclose(struct tcptw *, int); void tcp_ctlinput(int, struct sockaddr *, void *); int tcp_ctloutput(struct socket *, struct sockopt *); struct tcpcb * tcp_drop(struct tcpcb *, int); void tcp_drain(void); void tcp_init(void); #ifdef VIMAGE void tcp_destroy(void); #endif void tcp_fini(void *); char *tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *, const void *); char *tcp_log_vain(struct in_conninfo *, struct tcphdr *, void *, const void *); int tcp_reass(struct tcpcb *, struct tcphdr *, int *, struct mbuf *); void tcp_reass_global_init(void); void tcp_reass_flush(struct tcpcb *); int tcp_input(struct mbuf **, int *, int); u_long tcp_maxmtu(struct in_conninfo *, struct tcp_ifcap *); u_long tcp_maxmtu6(struct in_conninfo *, struct tcp_ifcap *); void tcp_mss_update(struct tcpcb *, int, int, struct hc_metrics_lite *, struct tcp_ifcap *); void tcp_mss(struct tcpcb *, int); int tcp_mssopt(struct in_conninfo *); struct inpcb * tcp_drop_syn_sent(struct inpcb *, int); struct tcpcb * tcp_newtcpcb(struct inpcb *); int tcp_output(struct tcpcb *); void tcp_state_change(struct tcpcb *, int); void tcp_respond(struct tcpcb *, void *, struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int); void tcp_tw_init(void); #ifdef VIMAGE void tcp_tw_destroy(void); #endif void tcp_tw_zone_change(void); int tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *, struct mbuf *, int); void tcp_setpersist(struct tcpcb *); #ifdef TCP_SIGNATURE struct secasvar; struct secasvar *tcp_get_sav(struct mbuf *, u_int); int tcp_signature_do_compute(struct mbuf *, int, int, u_char *, struct secasvar *); int tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int); int tcp_signature_verify(struct mbuf *, int, int, int, struct tcpopt *, struct tcphdr *, u_int); int tcp_signature_check(struct mbuf *m, int off0, int tlen, int optlen, struct tcpopt *to, struct tcphdr *th, u_int tcpbflag); #endif void tcp_slowtimo(void); struct tcptemp * tcpip_maketemplate(struct inpcb *); void tcpip_fillheaders(struct inpcb *, void *, void *); void tcp_timer_activate(struct tcpcb *, uint32_t, u_int); int tcp_timer_active(struct tcpcb *, uint32_t); void tcp_timer_stop(struct tcpcb *, uint32_t); void tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int); /* * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo) */ void tcp_hc_init(void); #ifdef VIMAGE void tcp_hc_destroy(void); #endif void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *); u_long tcp_hc_getmtu(struct in_conninfo *); void tcp_hc_updatemtu(struct in_conninfo *, u_long); void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *); extern struct pr_usrreqs tcp_usrreqs; tcp_seq tcp_new_isn(struct tcpcb *); void tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq); void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend); void tcp_clean_sackreport(struct tcpcb *tp); void tcp_sack_adjust(struct tcpcb *tp); struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); void tcp_sack_partialack(struct tcpcb *, struct tcphdr *); void tcp_free_sackholes(struct tcpcb *tp); int tcp_newreno(struct tcpcb *, struct tcphdr *); u_long tcp_seq_subtract(u_long, u_long ); void cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type); static inline void tcp_fields_to_host(struct tcphdr *th) { th->th_seq = ntohl(th->th_seq); th->th_ack = ntohl(th->th_ack); th->th_win = ntohs(th->th_win); th->th_urp = ntohs(th->th_urp); } #ifdef TCP_SIGNATURE static inline void tcp_fields_to_net(struct tcphdr *th) { th->th_seq = htonl(th->th_seq); th->th_ack = htonl(th->th_ack); th->th_win = htons(th->th_win); th->th_urp = htons(th->th_urp); } #endif #endif /* _KERNEL */ #endif /* _NETINET_TCP_VAR_H_ */ Index: head/sys/netinet6/nd6.c =================================================================== --- head/sys/netinet6/nd6.c (revision 288300) +++ head/sys/netinet6/nd6.c (revision 288301) @@ -1,2448 +1,2407 @@ /*- * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $KAME: nd6.c,v 1.144 2001/05/24 07:44:00 itojun Exp $ */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define ND6_SLOWTIMER_INTERVAL (60 * 60) /* 1 hour */ #define ND6_RECALC_REACHTM_INTERVAL (60 * 120) /* 2 hours */ #define SIN6(s) ((const struct sockaddr_in6 *)(s)) /* timer values */ VNET_DEFINE(int, nd6_prune) = 1; /* walk list every 1 seconds */ VNET_DEFINE(int, nd6_delay) = 5; /* delay first probe time 5 second */ VNET_DEFINE(int, nd6_umaxtries) = 3; /* maximum unicast query */ VNET_DEFINE(int, nd6_mmaxtries) = 3; /* maximum multicast query */ VNET_DEFINE(int, nd6_useloopback) = 1; /* use loopback interface for * local traffic */ VNET_DEFINE(int, nd6_gctimer) = (60 * 60 * 24); /* 1 day: garbage * collection timer */ /* preventing too many loops in ND option parsing */ static VNET_DEFINE(int, nd6_maxndopt) = 10; /* max # of ND options allowed */ VNET_DEFINE(int, nd6_maxnudhint) = 0; /* max # of subsequent upper * layer hints */ static VNET_DEFINE(int, nd6_maxqueuelen) = 1; /* max pkts cached in unresolved * ND entries */ #define V_nd6_maxndopt VNET(nd6_maxndopt) #define V_nd6_maxqueuelen VNET(nd6_maxqueuelen) #ifdef ND6_DEBUG VNET_DEFINE(int, nd6_debug) = 1; #else VNET_DEFINE(int, nd6_debug) = 0; #endif static eventhandler_tag lle_event_eh; /* for debugging? */ #if 0 static int nd6_inuse, nd6_allocated; #endif VNET_DEFINE(struct nd_drhead, nd_defrouter); VNET_DEFINE(struct nd_prhead, nd_prefix); VNET_DEFINE(int, nd6_recalc_reachtm_interval) = ND6_RECALC_REACHTM_INTERVAL; #define V_nd6_recalc_reachtm_interval VNET(nd6_recalc_reachtm_interval) int (*send_sendso_input_hook)(struct mbuf *, struct ifnet *, int, int); static int nd6_is_new_addr_neighbor(const struct sockaddr_in6 *, struct ifnet *); static void nd6_setmtu0(struct ifnet *, struct nd_ifinfo *); static void nd6_slowtimo(void *); static int regen_tmpaddr(struct in6_ifaddr *); static void nd6_free(struct llentry *, int); static void nd6_free_redirect(const struct llentry *); static void nd6_llinfo_timer(void *); static void clear_llinfo_pqueue(struct llentry *); static void nd6_rtrequest(int, struct rtentry *, struct rt_addrinfo *); static int nd6_resolve_slow(struct ifnet *, struct mbuf *, const struct sockaddr_in6 *, u_char *, uint32_t *); static int nd6_need_cache(struct ifnet *); static VNET_DEFINE(struct callout, nd6_slowtimo_ch); #define V_nd6_slowtimo_ch VNET(nd6_slowtimo_ch) VNET_DEFINE(struct callout, nd6_timer_ch); static void nd6_lle_event(void *arg __unused, struct llentry *lle, int evt) { struct rt_addrinfo rtinfo; struct sockaddr_in6 dst; struct sockaddr_dl gw; struct ifnet *ifp; int type; LLE_WLOCK_ASSERT(lle); if (lltable_get_af(lle->lle_tbl) != AF_INET6) return; switch (evt) { case LLENTRY_RESOLVED: type = RTM_ADD; KASSERT(lle->la_flags & LLE_VALID, ("%s: %p resolved but not valid?", __func__, lle)); break; case LLENTRY_EXPIRED: type = RTM_DELETE; break; default: return; } ifp = lltable_get_ifp(lle->lle_tbl); bzero(&dst, sizeof(dst)); bzero(&gw, sizeof(gw)); bzero(&rtinfo, sizeof(rtinfo)); lltable_fill_sa_entry(lle, (struct sockaddr *)&dst); dst.sin6_scope_id = in6_getscopezone(ifp, in6_addrscope(&dst.sin6_addr)); gw.sdl_len = sizeof(struct sockaddr_dl); gw.sdl_family = AF_LINK; gw.sdl_alen = ifp->if_addrlen; gw.sdl_index = ifp->if_index; gw.sdl_type = ifp->if_type; if (evt == LLENTRY_RESOLVED) bcopy(&lle->ll_addr, gw.sdl_data, ifp->if_addrlen); rtinfo.rti_info[RTAX_DST] = (struct sockaddr *)&dst; rtinfo.rti_info[RTAX_GATEWAY] = (struct sockaddr *)&gw; rtinfo.rti_addrs = RTA_DST | RTA_GATEWAY; rt_missmsg_fib(type, &rtinfo, RTF_HOST | RTF_LLDATA | ( type == RTM_ADD ? RTF_UP: 0), 0, RT_DEFAULT_FIB); } void nd6_init(void) { LIST_INIT(&V_nd_prefix); /* initialization of the default router list */ TAILQ_INIT(&V_nd_defrouter); /* start timer */ callout_init(&V_nd6_slowtimo_ch, 0); callout_reset(&V_nd6_slowtimo_ch, ND6_SLOWTIMER_INTERVAL * hz, nd6_slowtimo, curvnet); nd6_dad_init(); if (IS_DEFAULT_VNET(curvnet)) lle_event_eh = EVENTHANDLER_REGISTER(lle_event, nd6_lle_event, NULL, EVENTHANDLER_PRI_ANY); } #ifdef VIMAGE void nd6_destroy() { callout_drain(&V_nd6_slowtimo_ch); callout_drain(&V_nd6_timer_ch); if (IS_DEFAULT_VNET(curvnet)) EVENTHANDLER_DEREGISTER(lle_event, lle_event_eh); } #endif struct nd_ifinfo * nd6_ifattach(struct ifnet *ifp) { struct nd_ifinfo *nd; nd = (struct nd_ifinfo *)malloc(sizeof(*nd), M_IP6NDP, M_WAITOK|M_ZERO); nd->initialized = 1; nd->chlim = IPV6_DEFHLIM; nd->basereachable = REACHABLE_TIME; nd->reachable = ND_COMPUTE_RTIME(nd->basereachable); nd->retrans = RETRANS_TIMER; nd->flags = ND6_IFF_PERFORMNUD; /* A loopback interface always has ND6_IFF_AUTO_LINKLOCAL. * XXXHRS: Clear ND6_IFF_AUTO_LINKLOCAL on an IFT_BRIDGE interface by * default regardless of the V_ip6_auto_linklocal configuration to * give a reasonable default behavior. */ if ((V_ip6_auto_linklocal && ifp->if_type != IFT_BRIDGE) || (ifp->if_flags & IFF_LOOPBACK)) nd->flags |= ND6_IFF_AUTO_LINKLOCAL; /* * A loopback interface does not need to accept RTADV. * XXXHRS: Clear ND6_IFF_ACCEPT_RTADV on an IFT_BRIDGE interface by * default regardless of the V_ip6_accept_rtadv configuration to * prevent the interface from accepting RA messages arrived * on one of the member interfaces with ND6_IFF_ACCEPT_RTADV. */ if (V_ip6_accept_rtadv && !(ifp->if_flags & IFF_LOOPBACK) && (ifp->if_type != IFT_BRIDGE)) nd->flags |= ND6_IFF_ACCEPT_RTADV; if (V_ip6_no_radr && !(ifp->if_flags & IFF_LOOPBACK)) nd->flags |= ND6_IFF_NO_RADR; /* XXX: we cannot call nd6_setmtu since ifp is not fully initialized */ nd6_setmtu0(ifp, nd); return nd; } void nd6_ifdetach(struct nd_ifinfo *nd) { free(nd, M_IP6NDP); } /* * Reset ND level link MTU. This function is called when the physical MTU * changes, which means we might have to adjust the ND level MTU. */ void nd6_setmtu(struct ifnet *ifp) { nd6_setmtu0(ifp, ND_IFINFO(ifp)); } /* XXX todo: do not maintain copy of ifp->if_mtu in ndi->maxmtu */ void nd6_setmtu0(struct ifnet *ifp, struct nd_ifinfo *ndi) { u_int32_t omaxmtu; omaxmtu = ndi->maxmtu; switch (ifp->if_type) { case IFT_ARCNET: ndi->maxmtu = MIN(ARC_PHDS_MAXMTU, ifp->if_mtu); /* RFC2497 */ break; case IFT_FDDI: ndi->maxmtu = MIN(FDDIIPMTU, ifp->if_mtu); /* RFC2467 */ break; case IFT_ISO88025: ndi->maxmtu = MIN(ISO88025_MAX_MTU, ifp->if_mtu); break; default: ndi->maxmtu = ifp->if_mtu; break; } /* * Decreasing the interface MTU under IPV6 minimum MTU may cause * undesirable situation. We thus notify the operator of the change * explicitly. The check for omaxmtu is necessary to restrict the * log to the case of changing the MTU, not initializing it. */ if (omaxmtu >= IPV6_MMTU && ndi->maxmtu < IPV6_MMTU) { log(LOG_NOTICE, "nd6_setmtu0: " "new link MTU on %s (%lu) is too small for IPv6\n", if_name(ifp), (unsigned long)ndi->maxmtu); } if (ndi->maxmtu > V_in6_maxmtu) in6_setmaxmtu(); /* check all interfaces just in case */ } void nd6_option_init(void *opt, int icmp6len, union nd_opts *ndopts) { bzero(ndopts, sizeof(*ndopts)); ndopts->nd_opts_search = (struct nd_opt_hdr *)opt; ndopts->nd_opts_last = (struct nd_opt_hdr *)(((u_char *)opt) + icmp6len); if (icmp6len == 0) { ndopts->nd_opts_done = 1; ndopts->nd_opts_search = NULL; } } /* * Take one ND option. */ struct nd_opt_hdr * nd6_option(union nd_opts *ndopts) { struct nd_opt_hdr *nd_opt; int olen; KASSERT(ndopts != NULL, ("%s: ndopts == NULL", __func__)); KASSERT(ndopts->nd_opts_last != NULL, ("%s: uninitialized ndopts", __func__)); if (ndopts->nd_opts_search == NULL) return NULL; if (ndopts->nd_opts_done) return NULL; nd_opt = ndopts->nd_opts_search; /* make sure nd_opt_len is inside the buffer */ if ((caddr_t)&nd_opt->nd_opt_len >= (caddr_t)ndopts->nd_opts_last) { bzero(ndopts, sizeof(*ndopts)); return NULL; } olen = nd_opt->nd_opt_len << 3; if (olen == 0) { /* * Message validation requires that all included * options have a length that is greater than zero. */ bzero(ndopts, sizeof(*ndopts)); return NULL; } ndopts->nd_opts_search = (struct nd_opt_hdr *)((caddr_t)nd_opt + olen); if (ndopts->nd_opts_search > ndopts->nd_opts_last) { /* option overruns the end of buffer, invalid */ bzero(ndopts, sizeof(*ndopts)); return NULL; } else if (ndopts->nd_opts_search == ndopts->nd_opts_last) { /* reached the end of options chain */ ndopts->nd_opts_done = 1; ndopts->nd_opts_search = NULL; } return nd_opt; } /* * Parse multiple ND options. * This function is much easier to use, for ND routines that do not need * multiple options of the same type. */ int nd6_options(union nd_opts *ndopts) { struct nd_opt_hdr *nd_opt; int i = 0; KASSERT(ndopts != NULL, ("%s: ndopts == NULL", __func__)); KASSERT(ndopts->nd_opts_last != NULL, ("%s: uninitialized ndopts", __func__)); if (ndopts->nd_opts_search == NULL) return 0; while (1) { nd_opt = nd6_option(ndopts); if (nd_opt == NULL && ndopts->nd_opts_last == NULL) { /* * Message validation requires that all included * options have a length that is greater than zero. */ ICMP6STAT_INC(icp6s_nd_badopt); bzero(ndopts, sizeof(*ndopts)); return -1; } if (nd_opt == NULL) goto skip1; switch (nd_opt->nd_opt_type) { case ND_OPT_SOURCE_LINKADDR: case ND_OPT_TARGET_LINKADDR: case ND_OPT_MTU: case ND_OPT_REDIRECTED_HEADER: case ND_OPT_NONCE: if (ndopts->nd_opt_array[nd_opt->nd_opt_type]) { nd6log((LOG_INFO, "duplicated ND6 option found (type=%d)\n", nd_opt->nd_opt_type)); /* XXX bark? */ } else { ndopts->nd_opt_array[nd_opt->nd_opt_type] = nd_opt; } break; case ND_OPT_PREFIX_INFORMATION: if (ndopts->nd_opt_array[nd_opt->nd_opt_type] == 0) { ndopts->nd_opt_array[nd_opt->nd_opt_type] = nd_opt; } ndopts->nd_opts_pi_end = (struct nd_opt_prefix_info *)nd_opt; break; /* What about ND_OPT_ROUTE_INFO? RFC 4191 */ case ND_OPT_RDNSS: /* RFC 6106 */ case ND_OPT_DNSSL: /* RFC 6106 */ /* * Silently ignore options we know and do not care about * in the kernel. */ break; default: /* * Unknown options must be silently ignored, * to accomodate future extension to the protocol. */ nd6log((LOG_DEBUG, "nd6_options: unsupported option %d - " "option ignored\n", nd_opt->nd_opt_type)); } skip1: i++; if (i > V_nd6_maxndopt) { ICMP6STAT_INC(icp6s_nd_toomanyopt); nd6log((LOG_INFO, "too many loop in nd opt\n")); break; } if (ndopts->nd_opts_done) break; } return 0; } /* * ND6 timer routine to handle ND6 entries */ void nd6_llinfo_settimer_locked(struct llentry *ln, long tick) { int canceled; LLE_WLOCK_ASSERT(ln); if (tick < 0) { ln->la_expire = 0; ln->ln_ntick = 0; canceled = callout_stop(&ln->lle_timer); } else { ln->la_expire = time_uptime + tick / hz; LLE_ADDREF(ln); if (tick > INT_MAX) { ln->ln_ntick = tick - INT_MAX; canceled = callout_reset(&ln->lle_timer, INT_MAX, nd6_llinfo_timer, ln); } else { ln->ln_ntick = 0; canceled = callout_reset(&ln->lle_timer, tick, nd6_llinfo_timer, ln); } } if (canceled) LLE_REMREF(ln); } /* * Gets source address of the first packet in hold queue * and stores it in @src. * Returns pointer to @src (if hold queue is not empty) or NULL. * */ static struct in6_addr * nd6_llinfo_get_holdsrc(struct llentry *ln, struct in6_addr *src) { struct ip6_hdr hdr; struct mbuf *m; if (ln->la_hold == NULL) return (NULL); /* * assume every packet in la_hold has the same IP header */ m = ln->la_hold; if (sizeof(hdr) < m->m_len) return (NULL); m_copydata(m, 0, sizeof(hdr), (caddr_t)&hdr); *src = hdr.ip6_src; return (src); } /* * Switch @lle state to new state optionally arming timers. */ void nd6_llinfo_setstate(struct llentry *lle, int newstate) { struct ifnet *ifp; long delay; delay = 0; switch (newstate) { case ND6_LLINFO_INCOMPLETE: ifp = lle->lle_tbl->llt_ifp; delay = (long)ND_IFINFO(ifp)->retrans * hz / 1000; break; case ND6_LLINFO_REACHABLE: if (!ND6_LLINFO_PERMANENT(lle)) { ifp = lle->lle_tbl->llt_ifp; delay = (long)ND_IFINFO(ifp)->reachable * hz; } break; case ND6_LLINFO_STALE: delay = (long)V_nd6_gctimer * hz; break; case ND6_LLINFO_DELAY: lle->la_asked = 0; delay = (long)V_nd6_delay * hz; break; } if (delay > 0) nd6_llinfo_settimer_locked(lle, delay); lle->ln_state = newstate; } void nd6_llinfo_settimer(struct llentry *ln, long tick) { LLE_WLOCK(ln); nd6_llinfo_settimer_locked(ln, tick); LLE_WUNLOCK(ln); } static void nd6_llinfo_timer(void *arg) { struct llentry *ln; struct in6_addr *dst, *pdst, *psrc, src; struct ifnet *ifp; struct nd_ifinfo *ndi = NULL; int send_ns; KASSERT(arg != NULL, ("%s: arg NULL", __func__)); ln = (struct llentry *)arg; LLE_WLOCK(ln); if (callout_pending(&ln->lle_timer)) { /* * Here we are a bit odd here in the treatment of * active/pending. If the pending bit is set, it got * rescheduled before I ran. The active * bit we ignore, since if it was stopped * in ll_tablefree() and was currently running * it would have return 0 so the code would * not have deleted it since the callout could * not be stopped so we want to go through * with the delete here now. If the callout * was restarted, the pending bit will be back on and * we just want to bail since the callout_reset would * return 1 and our reference would have been removed * by nd6_llinfo_settimer_locked above since canceled * would have been 1. */ LLE_WUNLOCK(ln); return; } ifp = ln->lle_tbl->llt_ifp; CURVNET_SET(ifp->if_vnet); ndi = ND_IFINFO(ifp); send_ns = 0; dst = &ln->r_l3addr.addr6; pdst = dst; if (ln->ln_ntick > 0) { if (ln->ln_ntick > INT_MAX) { ln->ln_ntick -= INT_MAX; nd6_llinfo_settimer_locked(ln, INT_MAX); } else { ln->ln_ntick = 0; nd6_llinfo_settimer_locked(ln, ln->ln_ntick); } goto done; } if (ln->la_flags & LLE_STATIC) { goto done; } if (ln->la_flags & LLE_DELETED) { nd6_free(ln, 0); ln = NULL; goto done; } switch (ln->ln_state) { case ND6_LLINFO_INCOMPLETE: if (ln->la_asked < V_nd6_mmaxtries) { ln->la_asked++; send_ns = 1; /* Send NS to multicast address */ pdst = NULL; } else { struct mbuf *m = ln->la_hold; if (m) { struct mbuf *m0; /* * assuming every packet in la_hold has the * same IP header. Send error after unlock. */ m0 = m->m_nextpkt; m->m_nextpkt = NULL; ln->la_hold = m0; clear_llinfo_pqueue(ln); } EVENTHANDLER_INVOKE(lle_event, ln, LLENTRY_TIMEDOUT); nd6_free(ln, 0); ln = NULL; if (m != NULL) icmp6_error2(m, ICMP6_DST_UNREACH, ICMP6_DST_UNREACH_ADDR, 0, ifp); } break; case ND6_LLINFO_REACHABLE: if (!ND6_LLINFO_PERMANENT(ln)) nd6_llinfo_setstate(ln, ND6_LLINFO_STALE); break; case ND6_LLINFO_STALE: /* Garbage Collection(RFC 2461 5.3) */ if (!ND6_LLINFO_PERMANENT(ln)) { EVENTHANDLER_INVOKE(lle_event, ln, LLENTRY_EXPIRED); nd6_free(ln, 1); ln = NULL; } break; case ND6_LLINFO_DELAY: if (ndi && (ndi->flags & ND6_IFF_PERFORMNUD) != 0) { /* We need NUD */ ln->la_asked = 1; nd6_llinfo_setstate(ln, ND6_LLINFO_PROBE); send_ns = 1; } else nd6_llinfo_setstate(ln, ND6_LLINFO_STALE); /* XXX */ break; case ND6_LLINFO_PROBE: if (ln->la_asked < V_nd6_umaxtries) { ln->la_asked++; send_ns = 1; } else { EVENTHANDLER_INVOKE(lle_event, ln, LLENTRY_EXPIRED); nd6_free(ln, 0); ln = NULL; } break; default: panic("%s: paths in a dark night can be confusing: %d", __func__, ln->ln_state); } done: if (send_ns != 0) { nd6_llinfo_settimer_locked(ln, (long)ndi->retrans * hz / 1000); psrc = nd6_llinfo_get_holdsrc(ln, &src); LLE_FREE_LOCKED(ln); ln = NULL; nd6_ns_output(ifp, psrc, pdst, dst, NULL); } if (ln != NULL) LLE_FREE_LOCKED(ln); CURVNET_RESTORE(); } /* * ND6 timer routine to expire default route list and prefix list */ void nd6_timer(void *arg) { CURVNET_SET((struct vnet *) arg); struct nd_defrouter *dr, *ndr; struct nd_prefix *pr, *npr; struct in6_ifaddr *ia6, *nia6; callout_reset(&V_nd6_timer_ch, V_nd6_prune * hz, nd6_timer, curvnet); /* expire default router list */ TAILQ_FOREACH_SAFE(dr, &V_nd_defrouter, dr_entry, ndr) { if (dr->expire && dr->expire < time_uptime) defrtrlist_del(dr); } /* * expire interface addresses. * in the past the loop was inside prefix expiry processing. * However, from a stricter speci-confrmance standpoint, we should * rather separate address lifetimes and prefix lifetimes. * * XXXRW: in6_ifaddrhead locking. */ addrloop: TAILQ_FOREACH_SAFE(ia6, &V_in6_ifaddrhead, ia_link, nia6) { /* check address lifetime */ if (IFA6_IS_INVALID(ia6)) { int regen = 0; /* * If the expiring address is temporary, try * regenerating a new one. This would be useful when * we suspended a laptop PC, then turned it on after a * period that could invalidate all temporary * addresses. Although we may have to restart the * loop (see below), it must be after purging the * address. Otherwise, we'd see an infinite loop of * regeneration. */ if (V_ip6_use_tempaddr && (ia6->ia6_flags & IN6_IFF_TEMPORARY) != 0) { if (regen_tmpaddr(ia6) == 0) regen = 1; } in6_purgeaddr(&ia6->ia_ifa); if (regen) goto addrloop; /* XXX: see below */ } else if (IFA6_IS_DEPRECATED(ia6)) { int oldflags = ia6->ia6_flags; ia6->ia6_flags |= IN6_IFF_DEPRECATED; /* * If a temporary address has just become deprecated, * regenerate a new one if possible. */ if (V_ip6_use_tempaddr && (ia6->ia6_flags & IN6_IFF_TEMPORARY) != 0 && (oldflags & IN6_IFF_DEPRECATED) == 0) { if (regen_tmpaddr(ia6) == 0) { /* * A new temporary address is * generated. * XXX: this means the address chain * has changed while we are still in * the loop. Although the change * would not cause disaster (because * it's not a deletion, but an * addition,) we'd rather restart the * loop just for safety. Or does this * significantly reduce performance?? */ goto addrloop; } } } else { /* * A new RA might have made a deprecated address * preferred. */ ia6->ia6_flags &= ~IN6_IFF_DEPRECATED; } } /* expire prefix list */ LIST_FOREACH_SAFE(pr, &V_nd_prefix, ndpr_entry, npr) { /* * check prefix lifetime. * since pltime is just for autoconf, pltime processing for * prefix is not necessary. */ if (pr->ndpr_vltime != ND6_INFINITE_LIFETIME && time_uptime - pr->ndpr_lastupdate > pr->ndpr_vltime) { /* * address expiration and prefix expiration are * separate. NEVER perform in6_purgeaddr here. */ prelist_remove(pr); } } CURVNET_RESTORE(); } /* * ia6 - deprecated/invalidated temporary address */ static int regen_tmpaddr(struct in6_ifaddr *ia6) { struct ifaddr *ifa; struct ifnet *ifp; struct in6_ifaddr *public_ifa6 = NULL; ifp = ia6->ia_ifa.ifa_ifp; IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { struct in6_ifaddr *it6; if (ifa->ifa_addr->sa_family != AF_INET6) continue; it6 = (struct in6_ifaddr *)ifa; /* ignore no autoconf addresses. */ if ((it6->ia6_flags & IN6_IFF_AUTOCONF) == 0) continue; /* ignore autoconf addresses with different prefixes. */ if (it6->ia6_ndpr == NULL || it6->ia6_ndpr != ia6->ia6_ndpr) continue; /* * Now we are looking at an autoconf address with the same * prefix as ours. If the address is temporary and is still * preferred, do not create another one. It would be rare, but * could happen, for example, when we resume a laptop PC after * a long period. */ if ((it6->ia6_flags & IN6_IFF_TEMPORARY) != 0 && !IFA6_IS_DEPRECATED(it6)) { public_ifa6 = NULL; break; } /* * This is a public autoconf address that has the same prefix * as ours. If it is preferred, keep it. We can't break the * loop here, because there may be a still-preferred temporary * address with the prefix. */ if (!IFA6_IS_DEPRECATED(it6)) public_ifa6 = it6; } if (public_ifa6 != NULL) ifa_ref(&public_ifa6->ia_ifa); IF_ADDR_RUNLOCK(ifp); if (public_ifa6 != NULL) { int e; if ((e = in6_tmpifadd(public_ifa6, 0, 0)) != 0) { ifa_free(&public_ifa6->ia_ifa); log(LOG_NOTICE, "regen_tmpaddr: failed to create a new" " tmp addr,errno=%d\n", e); return (-1); } ifa_free(&public_ifa6->ia_ifa); return (0); } return (-1); } /* * Nuke neighbor cache/prefix/default router management table, right before * ifp goes away. */ void nd6_purge(struct ifnet *ifp) { struct nd_defrouter *dr, *ndr; struct nd_prefix *pr, *npr; /* * Nuke default router list entries toward ifp. * We defer removal of default router list entries that is installed * in the routing table, in order to keep additional side effects as * small as possible. */ TAILQ_FOREACH_SAFE(dr, &V_nd_defrouter, dr_entry, ndr) { if (dr->installed) continue; if (dr->ifp == ifp) defrtrlist_del(dr); } TAILQ_FOREACH_SAFE(dr, &V_nd_defrouter, dr_entry, ndr) { if (!dr->installed) continue; if (dr->ifp == ifp) defrtrlist_del(dr); } /* Nuke prefix list entries toward ifp */ LIST_FOREACH_SAFE(pr, &V_nd_prefix, ndpr_entry, npr) { if (pr->ndpr_ifp == ifp) { /* * Because if_detach() does *not* release prefixes * while purging addresses the reference count will * still be above zero. We therefore reset it to * make sure that the prefix really gets purged. */ pr->ndpr_refcnt = 0; /* * Previously, pr->ndpr_addr is removed as well, * but I strongly believe we don't have to do it. * nd6_purge() is only called from in6_ifdetach(), * which removes all the associated interface addresses * by itself. * (jinmei@kame.net 20010129) */ prelist_remove(pr); } } /* cancel default outgoing interface setting */ if (V_nd6_defifindex == ifp->if_index) nd6_setdefaultiface(0); if (ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV) { /* Refresh default router list. */ defrouter_select(); } /* XXXXX * We do not nuke the neighbor cache entries here any more * because the neighbor cache is kept in if_afdata[AF_INET6]. * nd6_purge() is invoked by in6_ifdetach() which is called * from if_detach() where everything gets purged. So let * in6_domifdetach() do the actual L2 table purging work. */ } /* * the caller acquires and releases the lock on the lltbls * Returns the llentry locked */ struct llentry * nd6_lookup(const struct in6_addr *addr6, int flags, struct ifnet *ifp) { struct sockaddr_in6 sin6; struct llentry *ln; bzero(&sin6, sizeof(sin6)); sin6.sin6_len = sizeof(struct sockaddr_in6); sin6.sin6_family = AF_INET6; sin6.sin6_addr = *addr6; IF_AFDATA_LOCK_ASSERT(ifp); ln = lla_lookup(LLTABLE6(ifp), flags, (struct sockaddr *)&sin6); return (ln); } struct llentry * nd6_alloc(const struct in6_addr *addr6, int flags, struct ifnet *ifp) { struct sockaddr_in6 sin6; struct llentry *ln; bzero(&sin6, sizeof(sin6)); sin6.sin6_len = sizeof(struct sockaddr_in6); sin6.sin6_family = AF_INET6; sin6.sin6_addr = *addr6; ln = lltable_alloc_entry(LLTABLE6(ifp), 0, (struct sockaddr *)&sin6); if (ln != NULL) ln->ln_state = ND6_LLINFO_NOSTATE; return (ln); } /* * Test whether a given IPv6 address is a neighbor or not, ignoring * the actual neighbor cache. The neighbor cache is ignored in order * to not reenter the routing code from within itself. */ static int nd6_is_new_addr_neighbor(const struct sockaddr_in6 *addr, struct ifnet *ifp) { struct nd_prefix *pr; struct ifaddr *dstaddr; /* * A link-local address is always a neighbor. * XXX: a link does not necessarily specify a single interface. */ if (IN6_IS_ADDR_LINKLOCAL(&addr->sin6_addr)) { struct sockaddr_in6 sin6_copy; u_int32_t zone; /* * We need sin6_copy since sa6_recoverscope() may modify the * content (XXX). */ sin6_copy = *addr; if (sa6_recoverscope(&sin6_copy)) return (0); /* XXX: should be impossible */ if (in6_setscope(&sin6_copy.sin6_addr, ifp, &zone)) return (0); if (sin6_copy.sin6_scope_id == zone) return (1); else return (0); } /* * If the address matches one of our addresses, * it should be a neighbor. * If the address matches one of our on-link prefixes, it should be a * neighbor. */ LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) { if (pr->ndpr_ifp != ifp) continue; if (!(pr->ndpr_stateflags & NDPRF_ONLINK)) { struct rtentry *rt; /* Always use the default FIB here. */ rt = in6_rtalloc1((struct sockaddr *)&pr->ndpr_prefix, 0, 0, RT_DEFAULT_FIB); if (rt == NULL) continue; /* * This is the case where multiple interfaces * have the same prefix, but only one is installed * into the routing table and that prefix entry * is not the one being examined here. In the case * where RADIX_MPATH is enabled, multiple route * entries (of the same rt_key value) will be * installed because the interface addresses all * differ. */ if (!IN6_ARE_ADDR_EQUAL(&pr->ndpr_prefix.sin6_addr, &((struct sockaddr_in6 *)rt_key(rt))->sin6_addr)) { RTFREE_LOCKED(rt); continue; } RTFREE_LOCKED(rt); } if (IN6_ARE_MASKED_ADDR_EQUAL(&pr->ndpr_prefix.sin6_addr, &addr->sin6_addr, &pr->ndpr_mask)) return (1); } /* * If the address is assigned on the node of the other side of * a p2p interface, the address should be a neighbor. */ dstaddr = ifa_ifwithdstaddr((const struct sockaddr *)addr, RT_ALL_FIBS); if (dstaddr != NULL) { if (dstaddr->ifa_ifp == ifp) { ifa_free(dstaddr); return (1); } ifa_free(dstaddr); } /* * If the default router list is empty, all addresses are regarded * as on-link, and thus, as a neighbor. */ if (ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV && TAILQ_EMPTY(&V_nd_defrouter) && V_nd6_defifindex == ifp->if_index) { return (1); } return (0); } /* * Detect if a given IPv6 address identifies a neighbor on a given link. * XXX: should take care of the destination of a p2p link? */ int nd6_is_addr_neighbor(const struct sockaddr_in6 *addr, struct ifnet *ifp) { struct llentry *lle; int rc = 0; IF_AFDATA_UNLOCK_ASSERT(ifp); if (nd6_is_new_addr_neighbor(addr, ifp)) return (1); /* * Even if the address matches none of our addresses, it might be * in the neighbor cache. */ IF_AFDATA_RLOCK(ifp); if ((lle = nd6_lookup(&addr->sin6_addr, 0, ifp)) != NULL) { LLE_RUNLOCK(lle); rc = 1; } IF_AFDATA_RUNLOCK(ifp); return (rc); } /* * Free an nd6 llinfo entry. * Since the function would cause significant changes in the kernel, DO NOT * make it global, unless you have a strong reason for the change, and are sure * that the change is safe. */ static void nd6_free(struct llentry *ln, int gc) { struct nd_defrouter *dr; struct ifnet *ifp; LLE_WLOCK_ASSERT(ln); /* * we used to have pfctlinput(PRC_HOSTDEAD) here. * even though it is not harmful, it was not really necessary. */ /* cancel timer */ nd6_llinfo_settimer_locked(ln, -1); ifp = ln->lle_tbl->llt_ifp; if (ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV) { dr = defrouter_lookup(&ln->r_l3addr.addr6, ifp); if (dr != NULL && dr->expire && ln->ln_state == ND6_LLINFO_STALE && gc) { /* * If the reason for the deletion is just garbage * collection, and the neighbor is an active default * router, do not delete it. Instead, reset the GC * timer using the router's lifetime. * Simply deleting the entry would affect default * router selection, which is not necessarily a good * thing, especially when we're using router preference * values. * XXX: the check for ln_state would be redundant, * but we intentionally keep it just in case. */ if (dr->expire > time_uptime) nd6_llinfo_settimer_locked(ln, (dr->expire - time_uptime) * hz); else nd6_llinfo_settimer_locked(ln, (long)V_nd6_gctimer * hz); LLE_REMREF(ln); LLE_WUNLOCK(ln); return; } if (dr) { /* * Unreachablity of a router might affect the default * router selection and on-link detection of advertised * prefixes. */ /* * Temporarily fake the state to choose a new default * router and to perform on-link determination of * prefixes correctly. * Below the state will be set correctly, * or the entry itself will be deleted. */ ln->ln_state = ND6_LLINFO_INCOMPLETE; } if (ln->ln_router || dr) { /* * We need to unlock to avoid a LOR with rt6_flush() with the * rnh and for the calls to pfxlist_onlink_check() and * defrouter_select() in the block further down for calls * into nd6_lookup(). We still hold a ref. */ LLE_WUNLOCK(ln); /* * rt6_flush must be called whether or not the neighbor * is in the Default Router List. * See a corresponding comment in nd6_na_input(). */ rt6_flush(&ln->r_l3addr.addr6, ifp); } if (dr) { /* * Since defrouter_select() does not affect the * on-link determination and MIP6 needs the check * before the default router selection, we perform * the check now. */ pfxlist_onlink_check(); /* * Refresh default router list. */ defrouter_select(); } /* * If this entry was added by an on-link redirect, remove the * corresponding host route. */ if (ln->la_flags & LLE_REDIRECT) nd6_free_redirect(ln); if (ln->ln_router || dr) LLE_WLOCK(ln); } /* * Save to unlock. We still hold an extra reference and will not * free(9) in llentry_free() if someone else holds one as well. */ LLE_WUNLOCK(ln); IF_AFDATA_LOCK(ifp); LLE_WLOCK(ln); /* Guard against race with other llentry_free(). */ if (ln->la_flags & LLE_LINKED) { /* Remove callout reference */ LLE_REMREF(ln); lltable_unlink_entry(ln->lle_tbl, ln); } IF_AFDATA_UNLOCK(ifp); llentry_free(ln); } /* * Remove the rtentry for the given llentry, * both of which were installed by a redirect. */ static void nd6_free_redirect(const struct llentry *ln) { int fibnum; struct rtentry *rt; struct radix_node_head *rnh; struct sockaddr_in6 sin6; lltable_fill_sa_entry(ln, (struct sockaddr *)&sin6); for (fibnum = 0; fibnum < rt_numfibs; fibnum++) { rnh = rt_tables_get_rnh(fibnum, AF_INET6); if (rnh == NULL) continue; RADIX_NODE_HEAD_LOCK(rnh); rt = in6_rtalloc1((struct sockaddr *)&sin6, 0, RTF_RNH_LOCKED, fibnum); if (rt) { if (rt->rt_flags == (RTF_UP | RTF_HOST | RTF_DYNAMIC)) rt_expunge(rnh, rt); RTFREE_LOCKED(rt); } RADIX_NODE_HEAD_UNLOCK(rnh); } } /* - * Upper-layer reachability hint for Neighbor Unreachability Detection. - * - * XXX cost-effective methods? - */ -void -nd6_nud_hint(struct rtentry *rt, struct in6_addr *dst6, int force) -{ - struct llentry *ln; - struct ifnet *ifp; - - if ((dst6 == NULL) || (rt == NULL)) - return; - - ifp = rt->rt_ifp; - IF_AFDATA_RLOCK(ifp); - ln = nd6_lookup(dst6, LLE_EXCLUSIVE, NULL); - IF_AFDATA_RUNLOCK(ifp); - if (ln == NULL) - return; - - if (ln->ln_state < ND6_LLINFO_REACHABLE) - goto done; - - /* - * if we get upper-layer reachability confirmation many times, - * it is possible we have false information. - */ - if (!force) { - ln->ln_byhint++; - if (ln->ln_byhint > V_nd6_maxnudhint) { - goto done; - } - } - - nd6_llinfo_setstate(ln, ND6_LLINFO_REACHABLE); -done: - LLE_WUNLOCK(ln); -} - - -/* * Rejuvenate this function for routing operations related * processing. */ void nd6_rtrequest(int req, struct rtentry *rt, struct rt_addrinfo *info) { struct sockaddr_in6 *gateway; struct nd_defrouter *dr; struct ifnet *ifp; gateway = (struct sockaddr_in6 *)rt->rt_gateway; ifp = rt->rt_ifp; switch (req) { case RTM_ADD: break; case RTM_DELETE: if (!ifp) return; /* * Only indirect routes are interesting. */ if ((rt->rt_flags & RTF_GATEWAY) == 0) return; /* * check for default route */ if (IN6_ARE_ADDR_EQUAL(&in6addr_any, &SIN6(rt_key(rt))->sin6_addr)) { dr = defrouter_lookup(&gateway->sin6_addr, ifp); if (dr != NULL) dr->installed = 0; } break; } } int nd6_ioctl(u_long cmd, caddr_t data, struct ifnet *ifp) { struct in6_ndireq *ndi = (struct in6_ndireq *)data; struct in6_nbrinfo *nbi = (struct in6_nbrinfo *)data; struct in6_ndifreq *ndif = (struct in6_ndifreq *)data; int error = 0; if (ifp->if_afdata[AF_INET6] == NULL) return (EPFNOSUPPORT); switch (cmd) { case OSIOCGIFINFO_IN6: #define ND ndi->ndi /* XXX: old ndp(8) assumes a positive value for linkmtu. */ bzero(&ND, sizeof(ND)); ND.linkmtu = IN6_LINKMTU(ifp); ND.maxmtu = ND_IFINFO(ifp)->maxmtu; ND.basereachable = ND_IFINFO(ifp)->basereachable; ND.reachable = ND_IFINFO(ifp)->reachable; ND.retrans = ND_IFINFO(ifp)->retrans; ND.flags = ND_IFINFO(ifp)->flags; ND.recalctm = ND_IFINFO(ifp)->recalctm; ND.chlim = ND_IFINFO(ifp)->chlim; break; case SIOCGIFINFO_IN6: ND = *ND_IFINFO(ifp); break; case SIOCSIFINFO_IN6: /* * used to change host variables from userland. * intented for a use on router to reflect RA configurations. */ /* 0 means 'unspecified' */ if (ND.linkmtu != 0) { if (ND.linkmtu < IPV6_MMTU || ND.linkmtu > IN6_LINKMTU(ifp)) { error = EINVAL; break; } ND_IFINFO(ifp)->linkmtu = ND.linkmtu; } if (ND.basereachable != 0) { int obasereachable = ND_IFINFO(ifp)->basereachable; ND_IFINFO(ifp)->basereachable = ND.basereachable; if (ND.basereachable != obasereachable) ND_IFINFO(ifp)->reachable = ND_COMPUTE_RTIME(ND.basereachable); } if (ND.retrans != 0) ND_IFINFO(ifp)->retrans = ND.retrans; if (ND.chlim != 0) ND_IFINFO(ifp)->chlim = ND.chlim; /* FALLTHROUGH */ case SIOCSIFINFO_FLAGS: { struct ifaddr *ifa; struct in6_ifaddr *ia; if ((ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) && !(ND.flags & ND6_IFF_IFDISABLED)) { /* ifdisabled 1->0 transision */ /* * If the interface is marked as ND6_IFF_IFDISABLED and * has an link-local address with IN6_IFF_DUPLICATED, * do not clear ND6_IFF_IFDISABLED. * See RFC 4862, Section 5.4.5. */ IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; ia = (struct in6_ifaddr *)ifa; if ((ia->ia6_flags & IN6_IFF_DUPLICATED) && IN6_IS_ADDR_LINKLOCAL(IA6_IN6(ia))) break; } IF_ADDR_RUNLOCK(ifp); if (ifa != NULL) { /* LLA is duplicated. */ ND.flags |= ND6_IFF_IFDISABLED; log(LOG_ERR, "Cannot enable an interface" " with a link-local address marked" " duplicate.\n"); } else { ND_IFINFO(ifp)->flags &= ~ND6_IFF_IFDISABLED; if (ifp->if_flags & IFF_UP) in6_if_up(ifp); } } else if (!(ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) && (ND.flags & ND6_IFF_IFDISABLED)) { /* ifdisabled 0->1 transision */ /* Mark all IPv6 address as tentative. */ ND_IFINFO(ifp)->flags |= ND6_IFF_IFDISABLED; if ((ND_IFINFO(ifp)->flags & ND6_IFF_NO_DAD) == 0) { IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; ia = (struct in6_ifaddr *)ifa; ia->ia6_flags |= IN6_IFF_TENTATIVE; } IF_ADDR_RUNLOCK(ifp); } } if (ND.flags & ND6_IFF_AUTO_LINKLOCAL) { if (!(ND_IFINFO(ifp)->flags & ND6_IFF_AUTO_LINKLOCAL)) { /* auto_linklocal 0->1 transision */ /* If no link-local address on ifp, configure */ ND_IFINFO(ifp)->flags |= ND6_IFF_AUTO_LINKLOCAL; in6_ifattach(ifp, NULL); } else if (!(ND.flags & ND6_IFF_IFDISABLED) && ifp->if_flags & IFF_UP) { /* * When the IF already has * ND6_IFF_AUTO_LINKLOCAL, no link-local * address is assigned, and IFF_UP, try to * assign one. */ IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; ia = (struct in6_ifaddr *)ifa; if (IN6_IS_ADDR_LINKLOCAL(IA6_IN6(ia))) break; } IF_ADDR_RUNLOCK(ifp); if (ifa != NULL) /* No LLA is configured. */ in6_ifattach(ifp, NULL); } } } ND_IFINFO(ifp)->flags = ND.flags; break; #undef ND case SIOCSNDFLUSH_IN6: /* XXX: the ioctl name is confusing... */ /* sync kernel routing table with the default router list */ defrouter_reset(); defrouter_select(); break; case SIOCSPFXFLUSH_IN6: { /* flush all the prefix advertised by routers */ struct nd_prefix *pr, *next; LIST_FOREACH_SAFE(pr, &V_nd_prefix, ndpr_entry, next) { struct in6_ifaddr *ia, *ia_next; if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr)) continue; /* XXX */ /* do we really have to remove addresses as well? */ /* XXXRW: in6_ifaddrhead locking. */ TAILQ_FOREACH_SAFE(ia, &V_in6_ifaddrhead, ia_link, ia_next) { if ((ia->ia6_flags & IN6_IFF_AUTOCONF) == 0) continue; if (ia->ia6_ndpr == pr) in6_purgeaddr(&ia->ia_ifa); } prelist_remove(pr); } break; } case SIOCSRTRFLUSH_IN6: { /* flush all the default routers */ struct nd_defrouter *dr, *next; defrouter_reset(); TAILQ_FOREACH_SAFE(dr, &V_nd_defrouter, dr_entry, next) { defrtrlist_del(dr); } defrouter_select(); break; } case SIOCGNBRINFO_IN6: { struct llentry *ln; struct in6_addr nb_addr = nbi->addr; /* make local for safety */ if ((error = in6_setscope(&nb_addr, ifp, NULL)) != 0) return (error); IF_AFDATA_RLOCK(ifp); ln = nd6_lookup(&nb_addr, 0, ifp); IF_AFDATA_RUNLOCK(ifp); if (ln == NULL) { error = EINVAL; break; } nbi->state = ln->ln_state; nbi->asked = ln->la_asked; nbi->isrouter = ln->ln_router; if (ln->la_expire == 0) nbi->expire = 0; else nbi->expire = ln->la_expire + (time_second - time_uptime); LLE_RUNLOCK(ln); break; } case SIOCGDEFIFACE_IN6: /* XXX: should be implemented as a sysctl? */ ndif->ifindex = V_nd6_defifindex; break; case SIOCSDEFIFACE_IN6: /* XXX: should be implemented as a sysctl? */ return (nd6_setdefaultiface(ndif->ifindex)); } return (error); } /* * Calculates new isRouter value based on provided parameters and * returns it. */ static int nd6_is_router(int type, int code, int is_new, int old_addr, int new_addr, int ln_router) { /* * ICMP6 type dependent behavior. * * NS: clear IsRouter if new entry * RS: clear IsRouter * RA: set IsRouter if there's lladdr * redir: clear IsRouter if new entry * * RA case, (1): * The spec says that we must set IsRouter in the following cases: * - If lladdr exist, set IsRouter. This means (1-5). * - If it is old entry (!newentry), set IsRouter. This means (7). * So, based on the spec, in (1-5) and (7) cases we must set IsRouter. * A quetion arises for (1) case. (1) case has no lladdr in the * neighbor cache, this is similar to (6). * This case is rare but we figured that we MUST NOT set IsRouter. * * is_new old_addr new_addr NS RS RA redir * D R * 0 n n (1) c ? s * 0 y n (2) c s s * 0 n y (3) c s s * 0 y y (4) c s s * 0 y y (5) c s s * 1 -- n (6) c c c s * 1 -- y (7) c c s c s * * (c=clear s=set) */ switch (type & 0xff) { case ND_NEIGHBOR_SOLICIT: /* * New entry must have is_router flag cleared. */ if (is_new) /* (6-7) */ ln_router = 0; break; case ND_REDIRECT: /* * If the icmp is a redirect to a better router, always set the * is_router flag. Otherwise, if the entry is newly created, * clear the flag. [RFC 2461, sec 8.3] */ if (code == ND_REDIRECT_ROUTER) ln_router = 1; else { if (is_new) /* (6-7) */ ln_router = 0; } break; case ND_ROUTER_SOLICIT: /* * is_router flag must always be cleared. */ ln_router = 0; break; case ND_ROUTER_ADVERT: /* * Mark an entry with lladdr as a router. */ if ((!is_new && (old_addr || new_addr)) || /* (2-5) */ (is_new && new_addr)) { /* (7) */ ln_router = 1; } break; } return (ln_router); } /* * Create neighbor cache entry and cache link-layer address, * on reception of inbound ND6 packets. (RS/RA/NS/redirect) * * type - ICMP6 type * code - type dependent information * */ void nd6_cache_lladdr(struct ifnet *ifp, struct in6_addr *from, char *lladdr, int lladdrlen, int type, int code) { struct llentry *ln = NULL, *ln_tmp; int is_newentry; int do_update; int olladdr; int llchange; int flags; uint16_t router = 0; struct sockaddr_in6 sin6; struct mbuf *chain = NULL; IF_AFDATA_UNLOCK_ASSERT(ifp); KASSERT(ifp != NULL, ("%s: ifp == NULL", __func__)); KASSERT(from != NULL, ("%s: from == NULL", __func__)); /* nothing must be updated for unspecified address */ if (IN6_IS_ADDR_UNSPECIFIED(from)) return; /* * Validation about ifp->if_addrlen and lladdrlen must be done in * the caller. * * XXX If the link does not have link-layer adderss, what should * we do? (ifp->if_addrlen == 0) * Spec says nothing in sections for RA, RS and NA. There's small * description on it in NS section (RFC 2461 7.2.3). */ flags = lladdr ? LLE_EXCLUSIVE : 0; IF_AFDATA_RLOCK(ifp); ln = nd6_lookup(from, flags, ifp); IF_AFDATA_RUNLOCK(ifp); is_newentry = 0; if (ln == NULL) { flags |= LLE_EXCLUSIVE; ln = nd6_alloc(from, 0, ifp); if (ln == NULL) return; /* * Since we already know all the data for the new entry, * fill it before insertion. */ if (lladdr != NULL) { bcopy(lladdr, &ln->ll_addr, ifp->if_addrlen); ln->la_flags |= LLE_VALID; } IF_AFDATA_WLOCK(ifp); LLE_WLOCK(ln); /* Prefer any existing lle over newly-created one */ ln_tmp = nd6_lookup(from, LLE_EXCLUSIVE, ifp); if (ln_tmp == NULL) lltable_link_entry(LLTABLE6(ifp), ln); IF_AFDATA_WUNLOCK(ifp); if (ln_tmp == NULL) { /* No existing lle, mark as new entry */ is_newentry = 1; nd6_llinfo_setstate(ln, ND6_LLINFO_STALE); } else { lltable_free_entry(LLTABLE6(ifp), ln); ln = ln_tmp; ln_tmp = NULL; } } /* do nothing if static ndp is set */ if ((ln->la_flags & LLE_STATIC)) { if (flags & LLE_EXCLUSIVE) LLE_WUNLOCK(ln); else LLE_RUNLOCK(ln); return; } olladdr = (ln->la_flags & LLE_VALID) ? 1 : 0; if (olladdr && lladdr) { llchange = bcmp(lladdr, &ln->ll_addr, ifp->if_addrlen); } else if (!olladdr && lladdr) llchange = 1; else llchange = 0; /* * newentry olladdr lladdr llchange (*=record) * 0 n n -- (1) * 0 y n -- (2) * 0 n y y (3) * STALE * 0 y y n (4) * * 0 y y y (5) * STALE * 1 -- n -- (6) NOSTATE(= PASSIVE) * 1 -- y -- (7) * STALE */ do_update = 0; if (!is_newentry && llchange != 0) do_update = 1; /* (3,5) */ if (lladdr) { /* (3-5) and (7) */ /* * Record source link-layer address * XXX is it dependent to ifp->if_type? */ bcopy(lladdr, &ln->ll_addr, ifp->if_addrlen); ln->la_flags |= LLE_VALID; nd6_llinfo_setstate(ln, ND6_LLINFO_STALE); EVENTHANDLER_INVOKE(lle_event, ln, LLENTRY_RESOLVED); if (do_update) { if (ln->la_hold != NULL) nd6_grab_holdchain(ln, &chain, &sin6); } } /* Calculates new router status */ router = nd6_is_router(type, code, is_newentry, olladdr, lladdr != NULL ? 1 : 0, ln->ln_router); ln->ln_router = router; /* Mark non-router redirects with special flag */ if ((type & 0xFF) == ND_REDIRECT && code != ND_REDIRECT_ROUTER) ln->la_flags |= LLE_REDIRECT; if (flags & LLE_EXCLUSIVE) LLE_WUNLOCK(ln); else LLE_RUNLOCK(ln); if (chain != NULL) nd6_flush_holdchain(ifp, ifp, chain, &sin6); /* * When the link-layer address of a router changes, select the * best router again. In particular, when the neighbor entry is newly * created, it might affect the selection policy. * Question: can we restrict the first condition to the "is_newentry" * case? * XXX: when we hear an RA from a new router with the link-layer * address option, defrouter_select() is called twice, since * defrtrlist_update called the function as well. However, I believe * we can compromise the overhead, since it only happens the first * time. * XXX: although defrouter_select() should not have a bad effect * for those are not autoconfigured hosts, we explicitly avoid such * cases for safety. */ if ((do_update || is_newentry) && router && ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV) { /* * guaranteed recursion */ defrouter_select(); } } static void nd6_slowtimo(void *arg) { CURVNET_SET((struct vnet *) arg); struct nd_ifinfo *nd6if; struct ifnet *ifp; callout_reset(&V_nd6_slowtimo_ch, ND6_SLOWTIMER_INTERVAL * hz, nd6_slowtimo, curvnet); IFNET_RLOCK_NOSLEEP(); TAILQ_FOREACH(ifp, &V_ifnet, if_link) { if (ifp->if_afdata[AF_INET6] == NULL) continue; nd6if = ND_IFINFO(ifp); if (nd6if->basereachable && /* already initialized */ (nd6if->recalctm -= ND6_SLOWTIMER_INTERVAL) <= 0) { /* * Since reachable time rarely changes by router * advertisements, we SHOULD insure that a new random * value gets recomputed at least once every few hours. * (RFC 2461, 6.3.4) */ nd6if->recalctm = V_nd6_recalc_reachtm_interval; nd6if->reachable = ND_COMPUTE_RTIME(nd6if->basereachable); } } IFNET_RUNLOCK_NOSLEEP(); CURVNET_RESTORE(); } void nd6_grab_holdchain(struct llentry *ln, struct mbuf **chain, struct sockaddr_in6 *sin6) { LLE_WLOCK_ASSERT(ln); *chain = ln->la_hold; ln->la_hold = NULL; lltable_fill_sa_entry(ln, (struct sockaddr *)sin6); if (ln->ln_state == ND6_LLINFO_STALE) { /* * The first time we send a packet to a * neighbor whose entry is STALE, we have * to change the state to DELAY and a sets * a timer to expire in DELAY_FIRST_PROBE_TIME * seconds to ensure do neighbor unreachability * detection on expiration. * (RFC 2461 7.3.3) */ nd6_llinfo_setstate(ln, ND6_LLINFO_DELAY); } } int nd6_output_ifp(struct ifnet *ifp, struct ifnet *origifp, struct mbuf *m, struct sockaddr_in6 *dst) { int error; int ip6len; struct ip6_hdr *ip6; struct m_tag *mtag; #ifdef MAC mac_netinet6_nd6_send(ifp, m); #endif /* * If called from nd6_ns_output() (NS), nd6_na_output() (NA), * icmp6_redirect_output() (REDIRECT) or from rip6_output() (RS, RA * as handled by rtsol and rtadvd), mbufs will be tagged for SeND * to be diverted to user space. When re-injected into the kernel, * send_output() will directly dispatch them to the outgoing interface. */ if (send_sendso_input_hook != NULL) { mtag = m_tag_find(m, PACKET_TAG_ND_OUTGOING, NULL); if (mtag != NULL) { ip6 = mtod(m, struct ip6_hdr *); ip6len = sizeof(struct ip6_hdr) + ntohs(ip6->ip6_plen); /* Use the SEND socket */ error = send_sendso_input_hook(m, ifp, SND_OUT, ip6len); /* -1 == no app on SEND socket */ if (error == 0 || error != -1) return (error); } } m_clrprotoflags(m); /* Avoid confusing lower layers. */ IP_PROBE(send, NULL, NULL, mtod(m, struct ip6_hdr *), ifp, NULL, mtod(m, struct ip6_hdr *)); if ((ifp->if_flags & IFF_LOOPBACK) == 0) origifp = ifp; error = (*ifp->if_output)(origifp, m, (struct sockaddr *)dst, NULL); return (error); } /* * Do L2 address resolution for @sa_dst address. Stores found * address in @desten buffer. Copy of lle ln_flags can be also * saved in @pflags if @pflags is non-NULL. * * If destination LLE does not exists or lle state modification * is required, call "slow" version. * * Return values: * - 0 on success (address copied to buffer). * - EWOULDBLOCK (no local error, but address is still unresolved) * - other errors (alloc failure, etc) */ int nd6_resolve(struct ifnet *ifp, int is_gw, struct mbuf *m, const struct sockaddr *sa_dst, u_char *desten, uint32_t *pflags) { struct llentry *ln = NULL; const struct sockaddr_in6 *dst6; if (pflags != NULL) *pflags = 0; dst6 = (const struct sockaddr_in6 *)sa_dst; /* discard the packet if IPv6 operation is disabled on the interface */ if ((ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED)) { m_freem(m); return (ENETDOWN); /* better error? */ } if (m != NULL && m->m_flags & M_MCAST) { switch (ifp->if_type) { case IFT_ETHER: case IFT_FDDI: case IFT_L2VLAN: case IFT_IEEE80211: case IFT_BRIDGE: case IFT_ISO88025: ETHER_MAP_IPV6_MULTICAST(&dst6->sin6_addr, desten); return (0); default: m_freem(m); return (EAFNOSUPPORT); } } IF_AFDATA_RLOCK(ifp); ln = nd6_lookup(&dst6->sin6_addr, 0, ifp); IF_AFDATA_RUNLOCK(ifp); /* * Perform fast path for the following cases: * 1) lle state is REACHABLE * 2) lle state is DELAY (NS message sent) * * Every other case involves lle modification, so we handle * them separately. */ if (ln == NULL || (ln->ln_state != ND6_LLINFO_REACHABLE && ln->ln_state != ND6_LLINFO_DELAY)) { /* Fall back to slow processing path */ if (ln != NULL) LLE_RUNLOCK(ln); return (nd6_resolve_slow(ifp, m, dst6, desten, pflags)); } bcopy(&ln->ll_addr, desten, ifp->if_addrlen); if (pflags != NULL) *pflags = ln->la_flags; LLE_RUNLOCK(ln); return (0); } /* * Do L2 address resolution for @sa_dst address. Stores found * address in @desten buffer. Copy of lle ln_flags can be also * saved in @pflags if @pflags is non-NULL. * * Heavy version. * Function assume that destination LLE does not exist, * is invalid or stale, so LLE_EXCLUSIVE lock needs to be acquired. */ static int nd6_resolve_slow(struct ifnet *ifp, struct mbuf *m, const struct sockaddr_in6 *dst, u_char *desten, uint32_t *pflags) { struct llentry *lle = NULL, *lle_tmp; struct in6_addr *psrc, src; int send_ns; /* * Address resolution or Neighbor Unreachability Detection * for the next hop. * At this point, the destination of the packet must be a unicast * or an anycast address(i.e. not a multicast). */ if (lle == NULL) { IF_AFDATA_RLOCK(ifp); lle = nd6_lookup(&dst->sin6_addr, LLE_EXCLUSIVE, ifp); IF_AFDATA_RUNLOCK(ifp); if ((lle == NULL) && nd6_is_addr_neighbor(dst, ifp)) { /* * Since nd6_is_addr_neighbor() internally calls nd6_lookup(), * the condition below is not very efficient. But we believe * it is tolerable, because this should be a rare case. */ lle = nd6_alloc(&dst->sin6_addr, 0, ifp); if (lle == NULL) { char ip6buf[INET6_ADDRSTRLEN]; log(LOG_DEBUG, "nd6_output: can't allocate llinfo for %s " "(ln=%p)\n", ip6_sprintf(ip6buf, &dst->sin6_addr), lle); m_freem(m); return (ENOBUFS); } IF_AFDATA_WLOCK(ifp); LLE_WLOCK(lle); /* Prefer any existing entry over newly-created one */ lle_tmp = nd6_lookup(&dst->sin6_addr, LLE_EXCLUSIVE, ifp); if (lle_tmp == NULL) lltable_link_entry(LLTABLE6(ifp), lle); IF_AFDATA_WUNLOCK(ifp); if (lle_tmp != NULL) { lltable_free_entry(LLTABLE6(ifp), lle); lle = lle_tmp; lle_tmp = NULL; } } } if (lle == NULL) { if (!(ND_IFINFO(ifp)->flags & ND6_IFF_PERFORMNUD)) { m_freem(m); return (ENOBUFS); } if (m != NULL) m_freem(m); return (ENOBUFS); } LLE_WLOCK_ASSERT(lle); /* * The first time we send a packet to a neighbor whose entry is * STALE, we have to change the state to DELAY and a sets a timer to * expire in DELAY_FIRST_PROBE_TIME seconds to ensure do * neighbor unreachability detection on expiration. * (RFC 2461 7.3.3) */ if (lle->ln_state == ND6_LLINFO_STALE) nd6_llinfo_setstate(lle, ND6_LLINFO_DELAY); /* * If the neighbor cache entry has a state other than INCOMPLETE * (i.e. its link-layer address is already resolved), just * send the packet. */ if (lle->ln_state > ND6_LLINFO_INCOMPLETE) { bcopy(&lle->ll_addr, desten, ifp->if_addrlen); if (pflags != NULL) *pflags = lle->la_flags; LLE_WUNLOCK(lle); return (0); } /* * There is a neighbor cache entry, but no ethernet address * response yet. Append this latest packet to the end of the * packet queue in the mbuf, unless the number of the packet * does not exceed nd6_maxqueuelen. When it exceeds nd6_maxqueuelen, * the oldest packet in the queue will be removed. */ if (lle->la_hold != NULL) { struct mbuf *m_hold; int i; i = 0; for (m_hold = lle->la_hold; m_hold; m_hold = m_hold->m_nextpkt){ i++; if (m_hold->m_nextpkt == NULL) { m_hold->m_nextpkt = m; break; } } while (i >= V_nd6_maxqueuelen) { m_hold = lle->la_hold; lle->la_hold = lle->la_hold->m_nextpkt; m_freem(m_hold); i--; } } else { lle->la_hold = m; } /* * If there has been no NS for the neighbor after entering the * INCOMPLETE state, send the first solicitation. * Note that for newly-created lle la_asked will be 0, * so we will transition from ND6_LLINFO_NOSTATE to * ND6_LLINFO_INCOMPLETE state here. */ psrc = NULL; send_ns = 0; if (lle->la_asked == 0) { lle->la_asked++; send_ns = 1; psrc = nd6_llinfo_get_holdsrc(lle, &src); nd6_llinfo_setstate(lle, ND6_LLINFO_INCOMPLETE); } LLE_WUNLOCK(lle); if (send_ns != 0) nd6_ns_output(ifp, psrc, NULL, &dst->sin6_addr, NULL); return (EWOULDBLOCK); } int nd6_flush_holdchain(struct ifnet *ifp, struct ifnet *origifp, struct mbuf *chain, struct sockaddr_in6 *dst) { struct mbuf *m, *m_head; struct ifnet *outifp; int error = 0; m_head = chain; if ((ifp->if_flags & IFF_LOOPBACK) != 0) outifp = origifp; else outifp = ifp; while (m_head) { m = m_head; m_head = m_head->m_nextpkt; error = nd6_output_ifp(ifp, origifp, m, dst); } /* * XXX * note that intermediate errors are blindly ignored */ return (error); } static int nd6_need_cache(struct ifnet *ifp) { /* * XXX: we currently do not make neighbor cache on any interface * other than ARCnet, Ethernet, FDDI and GIF. * * RFC2893 says: * - unidirectional tunnels needs no ND */ switch (ifp->if_type) { case IFT_ARCNET: case IFT_ETHER: case IFT_FDDI: case IFT_IEEE1394: case IFT_L2VLAN: case IFT_IEEE80211: case IFT_INFINIBAND: case IFT_BRIDGE: case IFT_PROPVIRTUAL: return (1); default: return (0); } } /* * Add pernament ND6 link-layer record for given * interface address. * * Very similar to IPv4 arp_ifinit(), but: * 1) IPv6 DAD is performed in different place * 2) It is called by IPv6 protocol stack in contrast to * arp_ifinit() which is typically called in SIOCSIFADDR * driver ioctl handler. * */ int nd6_add_ifa_lle(struct in6_ifaddr *ia) { struct ifnet *ifp; struct llentry *ln, *ln_tmp; struct sockaddr *dst; ifp = ia->ia_ifa.ifa_ifp; if (nd6_need_cache(ifp) == 0) return (0); ia->ia_ifa.ifa_rtrequest = nd6_rtrequest; dst = (struct sockaddr *)&ia->ia_addr; ln = lltable_alloc_entry(LLTABLE6(ifp), LLE_IFADDR, dst); if (ln == NULL) return (ENOBUFS); IF_AFDATA_WLOCK(ifp); LLE_WLOCK(ln); /* Unlink any entry if exists */ ln_tmp = lla_lookup(LLTABLE6(ifp), LLE_EXCLUSIVE, dst); if (ln_tmp != NULL) lltable_unlink_entry(LLTABLE6(ifp), ln_tmp); lltable_link_entry(LLTABLE6(ifp), ln); IF_AFDATA_WUNLOCK(ifp); if (ln_tmp != NULL) EVENTHANDLER_INVOKE(lle_event, ln_tmp, LLENTRY_EXPIRED); EVENTHANDLER_INVOKE(lle_event, ln, LLENTRY_RESOLVED); LLE_WUNLOCK(ln); if (ln_tmp != NULL) llentry_free(ln_tmp); return (0); } /* * Removes either all lle entries for given @ia, or lle * corresponding to @ia address. */ void nd6_rem_ifa_lle(struct in6_ifaddr *ia, int all) { struct sockaddr_in6 mask, addr; struct sockaddr *saddr, *smask; struct ifnet *ifp; ifp = ia->ia_ifa.ifa_ifp; memcpy(&addr, &ia->ia_addr, sizeof(ia->ia_addr)); memcpy(&mask, &ia->ia_prefixmask, sizeof(ia->ia_prefixmask)); saddr = (struct sockaddr *)&addr; smask = (struct sockaddr *)&mask; if (all != 0) lltable_prefix_free(AF_INET6, saddr, smask, LLE_STATIC); else lltable_delete_addr(LLTABLE6(ifp), LLE_IFADDR, saddr); } static void clear_llinfo_pqueue(struct llentry *ln) { struct mbuf *m_hold, *m_hold_next; for (m_hold = ln->la_hold; m_hold; m_hold = m_hold_next) { m_hold_next = m_hold->m_nextpkt; m_freem(m_hold); } ln->la_hold = NULL; return; } static int nd6_sysctl_drlist(SYSCTL_HANDLER_ARGS); static int nd6_sysctl_prlist(SYSCTL_HANDLER_ARGS); #ifdef SYSCTL_DECL SYSCTL_DECL(_net_inet6_icmp6); #endif SYSCTL_NODE(_net_inet6_icmp6, ICMPV6CTL_ND6_DRLIST, nd6_drlist, CTLFLAG_RD, nd6_sysctl_drlist, ""); SYSCTL_NODE(_net_inet6_icmp6, ICMPV6CTL_ND6_PRLIST, nd6_prlist, CTLFLAG_RD, nd6_sysctl_prlist, ""); SYSCTL_INT(_net_inet6_icmp6, ICMPV6CTL_ND6_MAXQLEN, nd6_maxqueuelen, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(nd6_maxqueuelen), 1, ""); SYSCTL_INT(_net_inet6_icmp6, OID_AUTO, nd6_gctimer, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(nd6_gctimer), (60 * 60 * 24), ""); static int nd6_sysctl_drlist(SYSCTL_HANDLER_ARGS) { struct in6_defrouter d; struct nd_defrouter *dr; int error; if (req->newptr) return (EPERM); bzero(&d, sizeof(d)); d.rtaddr.sin6_family = AF_INET6; d.rtaddr.sin6_len = sizeof(d.rtaddr); /* * XXX locking */ TAILQ_FOREACH(dr, &V_nd_defrouter, dr_entry) { d.rtaddr.sin6_addr = dr->rtaddr; error = sa6_recoverscope(&d.rtaddr); if (error != 0) return (error); d.flags = dr->flags; d.rtlifetime = dr->rtlifetime; d.expire = dr->expire + (time_second - time_uptime); d.if_index = dr->ifp->if_index; error = SYSCTL_OUT(req, &d, sizeof(d)); if (error != 0) return (error); } return (0); } static int nd6_sysctl_prlist(SYSCTL_HANDLER_ARGS) { struct in6_prefix p; struct sockaddr_in6 s6; struct nd_prefix *pr; struct nd_pfxrouter *pfr; time_t maxexpire; int error; char ip6buf[INET6_ADDRSTRLEN]; if (req->newptr) return (EPERM); bzero(&p, sizeof(p)); p.origin = PR_ORIG_RA; bzero(&s6, sizeof(s6)); s6.sin6_family = AF_INET6; s6.sin6_len = sizeof(s6); /* * XXX locking */ LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) { p.prefix = pr->ndpr_prefix; if (sa6_recoverscope(&p.prefix)) { log(LOG_ERR, "scope error in prefix list (%s)\n", ip6_sprintf(ip6buf, &p.prefix.sin6_addr)); /* XXX: press on... */ } p.raflags = pr->ndpr_raf; p.prefixlen = pr->ndpr_plen; p.vltime = pr->ndpr_vltime; p.pltime = pr->ndpr_pltime; p.if_index = pr->ndpr_ifp->if_index; if (pr->ndpr_vltime == ND6_INFINITE_LIFETIME) p.expire = 0; else { /* XXX: we assume time_t is signed. */ maxexpire = (-1) & ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1)); if (pr->ndpr_vltime < maxexpire - pr->ndpr_lastupdate) p.expire = pr->ndpr_lastupdate + pr->ndpr_vltime + (time_second - time_uptime); else p.expire = maxexpire; } p.refcnt = pr->ndpr_refcnt; p.flags = pr->ndpr_stateflags; p.advrtrs = 0; LIST_FOREACH(pfr, &pr->ndpr_advrtrs, pfr_entry) p.advrtrs++; error = SYSCTL_OUT(req, &p, sizeof(p)); if (error != 0) return (error); LIST_FOREACH(pfr, &pr->ndpr_advrtrs, pfr_entry) { s6.sin6_addr = pfr->router->rtaddr; if (sa6_recoverscope(&s6)) log(LOG_ERR, "scope error in prefix list (%s)\n", ip6_sprintf(ip6buf, &pfr->router->rtaddr)); error = SYSCTL_OUT(req, &s6, sizeof(s6)); if (error != 0) return (error); } } return (0); } Index: head/sys/netinet6/nd6.h =================================================================== --- head/sys/netinet6/nd6.h (revision 288300) +++ head/sys/netinet6/nd6.h (revision 288301) @@ -1,461 +1,460 @@ /*- * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $KAME: nd6.h,v 1.76 2001/12/18 02:10:31 itojun Exp $ * $FreeBSD$ */ #ifndef _NETINET6_ND6_H_ #define _NETINET6_ND6_H_ /* see net/route.h, or net/if_inarp.h */ #ifndef RTF_ANNOUNCE #define RTF_ANNOUNCE RTF_PROTO2 #endif #include #include struct llentry; #define ND6_LLINFO_NOSTATE -2 /* * We don't need the WAITDELETE state any more, but we keep the definition * in a comment line instead of removing it. This is necessary to avoid * unintentionally reusing the value for another purpose, which might * affect backward compatibility with old applications. * (20000711 jinmei@kame.net) */ /* #define ND6_LLINFO_WAITDELETE -1 */ #define ND6_LLINFO_INCOMPLETE 0 #define ND6_LLINFO_REACHABLE 1 #define ND6_LLINFO_STALE 2 #define ND6_LLINFO_DELAY 3 #define ND6_LLINFO_PROBE 4 #define ND6_IS_LLINFO_PROBREACH(n) ((n)->ln_state > ND6_LLINFO_INCOMPLETE) #define ND6_LLINFO_PERMANENT(n) (((n)->la_expire == 0) && ((n)->ln_state > ND6_LLINFO_INCOMPLETE)) struct nd_ifinfo { u_int32_t linkmtu; /* LinkMTU */ u_int32_t maxmtu; /* Upper bound of LinkMTU */ u_int32_t basereachable; /* BaseReachableTime */ u_int32_t reachable; /* Reachable Time */ u_int32_t retrans; /* Retrans Timer */ u_int32_t flags; /* Flags */ int recalctm; /* BaseReacable re-calculation timer */ u_int8_t chlim; /* CurHopLimit */ u_int8_t initialized; /* Flag to see the entry is initialized */ /* the following 3 members are for privacy extension for addrconf */ u_int8_t randomseed0[8]; /* upper 64 bits of MD5 digest */ u_int8_t randomseed1[8]; /* lower 64 bits (usually the EUI64 IFID) */ u_int8_t randomid[8]; /* current random ID */ }; #define ND6_IFF_PERFORMNUD 0x1 #define ND6_IFF_ACCEPT_RTADV 0x2 #define ND6_IFF_PREFER_SOURCE 0x4 /* Not used in FreeBSD. */ #define ND6_IFF_IFDISABLED 0x8 /* IPv6 operation is disabled due to * DAD failure. (XXX: not ND-specific) */ #define ND6_IFF_DONT_SET_IFROUTE 0x10 #define ND6_IFF_AUTO_LINKLOCAL 0x20 #define ND6_IFF_NO_RADR 0x40 #define ND6_IFF_NO_PREFER_IFACE 0x80 /* XXX: not related to ND. */ #define ND6_IFF_NO_DAD 0x100 #ifdef _KERNEL #define ND_IFINFO(ifp) \ (((struct in6_ifextra *)(ifp)->if_afdata[AF_INET6])->nd_ifinfo) #define IN6_LINKMTU(ifp) \ ((ND_IFINFO(ifp)->linkmtu && ND_IFINFO(ifp)->linkmtu < (ifp)->if_mtu) \ ? ND_IFINFO(ifp)->linkmtu \ : ((ND_IFINFO(ifp)->maxmtu && ND_IFINFO(ifp)->maxmtu < (ifp)->if_mtu) \ ? ND_IFINFO(ifp)->maxmtu : (ifp)->if_mtu)) #endif struct in6_nbrinfo { char ifname[IFNAMSIZ]; /* if name, e.g. "en0" */ struct in6_addr addr; /* IPv6 address of the neighbor */ long asked; /* number of queries already sent for this addr */ int isrouter; /* if it acts as a router */ int state; /* reachability state */ int expire; /* lifetime for NDP state transition */ }; #define DRLSTSIZ 10 #define PRLSTSIZ 10 struct in6_drlist { char ifname[IFNAMSIZ]; struct { struct in6_addr rtaddr; u_char flags; u_short rtlifetime; u_long expire; u_short if_index; } defrouter[DRLSTSIZ]; }; struct in6_defrouter { struct sockaddr_in6 rtaddr; u_char flags; u_short rtlifetime; u_long expire; u_short if_index; }; #ifdef _KERNEL struct in6_oprlist { char ifname[IFNAMSIZ]; struct { struct in6_addr prefix; struct prf_ra raflags; u_char prefixlen; u_char origin; u_long vltime; u_long pltime; u_long expire; u_short if_index; u_short advrtrs; /* number of advertisement routers */ struct in6_addr advrtr[DRLSTSIZ]; /* XXX: explicit limit */ } prefix[PRLSTSIZ]; }; #endif struct in6_prlist { char ifname[IFNAMSIZ]; struct { struct in6_addr prefix; struct prf_ra raflags; u_char prefixlen; u_char origin; u_int32_t vltime; u_int32_t pltime; time_t expire; u_short if_index; u_short advrtrs; /* number of advertisement routers */ struct in6_addr advrtr[DRLSTSIZ]; /* XXX: explicit limit */ } prefix[PRLSTSIZ]; }; struct in6_prefix { struct sockaddr_in6 prefix; struct prf_ra raflags; u_char prefixlen; u_char origin; u_int32_t vltime; u_int32_t pltime; time_t expire; u_int32_t flags; int refcnt; u_short if_index; u_short advrtrs; /* number of advertisement routers */ /* struct sockaddr_in6 advrtr[] */ }; #ifdef _KERNEL struct in6_ondireq { char ifname[IFNAMSIZ]; struct { u_int32_t linkmtu; /* LinkMTU */ u_int32_t maxmtu; /* Upper bound of LinkMTU */ u_int32_t basereachable; /* BaseReachableTime */ u_int32_t reachable; /* Reachable Time */ u_int32_t retrans; /* Retrans Timer */ u_int32_t flags; /* Flags */ int recalctm; /* BaseReacable re-calculation timer */ u_int8_t chlim; /* CurHopLimit */ u_int8_t receivedra; } ndi; }; #endif struct in6_ndireq { char ifname[IFNAMSIZ]; struct nd_ifinfo ndi; }; struct in6_ndifreq { char ifname[IFNAMSIZ]; u_long ifindex; }; /* Prefix status */ #define NDPRF_ONLINK 0x1 #define NDPRF_DETACHED 0x2 /* protocol constants */ #define MAX_RTR_SOLICITATION_DELAY 1 /* 1sec */ #define RTR_SOLICITATION_INTERVAL 4 /* 4sec */ #define MAX_RTR_SOLICITATIONS 3 #define ND6_INFINITE_LIFETIME 0xffffffff #ifdef _KERNEL /* node constants */ #define MAX_REACHABLE_TIME 3600000 /* msec */ #define REACHABLE_TIME 30000 /* msec */ #define RETRANS_TIMER 1000 /* msec */ #define MIN_RANDOM_FACTOR 512 /* 1024 * 0.5 */ #define MAX_RANDOM_FACTOR 1536 /* 1024 * 1.5 */ #define DEF_TEMP_VALID_LIFETIME 604800 /* 1 week */ #define DEF_TEMP_PREFERRED_LIFETIME 86400 /* 1 day */ #define TEMPADDR_REGEN_ADVANCE 5 /* sec */ #define MAX_TEMP_DESYNC_FACTOR 600 /* 10 min */ #define ND_COMPUTE_RTIME(x) \ (((MIN_RANDOM_FACTOR * (x >> 10)) + (arc4random() & \ ((MAX_RANDOM_FACTOR - MIN_RANDOM_FACTOR) * (x >> 10)))) /1000) TAILQ_HEAD(nd_drhead, nd_defrouter); struct nd_defrouter { TAILQ_ENTRY(nd_defrouter) dr_entry; struct in6_addr rtaddr; u_char flags; /* flags on RA message */ u_short rtlifetime; u_long expire; struct ifnet *ifp; int installed; /* is installed into kernel routing table */ }; struct nd_prefixctl { struct ifnet *ndpr_ifp; /* prefix */ struct sockaddr_in6 ndpr_prefix; u_char ndpr_plen; u_int32_t ndpr_vltime; /* advertised valid lifetime */ u_int32_t ndpr_pltime; /* advertised preferred lifetime */ struct prf_ra ndpr_flags; }; struct nd_prefix { struct ifnet *ndpr_ifp; LIST_ENTRY(nd_prefix) ndpr_entry; struct sockaddr_in6 ndpr_prefix; /* prefix */ struct in6_addr ndpr_mask; /* netmask derived from the prefix */ u_int32_t ndpr_vltime; /* advertised valid lifetime */ u_int32_t ndpr_pltime; /* advertised preferred lifetime */ time_t ndpr_expire; /* expiration time of the prefix */ time_t ndpr_preferred; /* preferred time of the prefix */ time_t ndpr_lastupdate; /* reception time of last advertisement */ struct prf_ra ndpr_flags; u_int32_t ndpr_stateflags; /* actual state flags */ /* list of routers that advertise the prefix: */ LIST_HEAD(pr_rtrhead, nd_pfxrouter) ndpr_advrtrs; u_char ndpr_plen; int ndpr_refcnt; /* reference couter from addresses */ }; #define ndpr_raf ndpr_flags #define ndpr_raf_onlink ndpr_flags.onlink #define ndpr_raf_auto ndpr_flags.autonomous #define ndpr_raf_router ndpr_flags.router /* * Message format for use in obtaining information about prefixes * from inet6 sysctl function */ struct inet6_ndpr_msghdr { u_short inpm_msglen; /* to skip over non-understood messages */ u_char inpm_version; /* future binary compatibility */ u_char inpm_type; /* message type */ struct in6_addr inpm_prefix; u_long prm_vltim; u_long prm_pltime; u_long prm_expire; u_long prm_preferred; struct in6_prflags prm_flags; u_short prm_index; /* index for associated ifp */ u_char prm_plen; /* length of prefix in bits */ }; #define prm_raf_onlink prm_flags.prf_ra.onlink #define prm_raf_auto prm_flags.prf_ra.autonomous #define prm_statef_onlink prm_flags.prf_state.onlink #define prm_rrf_decrvalid prm_flags.prf_rr.decrvalid #define prm_rrf_decrprefd prm_flags.prf_rr.decrprefd struct nd_pfxrouter { LIST_ENTRY(nd_pfxrouter) pfr_entry; struct nd_defrouter *router; }; LIST_HEAD(nd_prhead, nd_prefix); /* nd6.c */ VNET_DECLARE(int, nd6_prune); VNET_DECLARE(int, nd6_delay); VNET_DECLARE(int, nd6_umaxtries); VNET_DECLARE(int, nd6_mmaxtries); VNET_DECLARE(int, nd6_useloopback); VNET_DECLARE(int, nd6_maxnudhint); VNET_DECLARE(int, nd6_gctimer); VNET_DECLARE(struct nd_drhead, nd_defrouter); VNET_DECLARE(struct nd_prhead, nd_prefix); VNET_DECLARE(int, nd6_debug); VNET_DECLARE(int, nd6_onlink_ns_rfc4861); #define V_nd6_prune VNET(nd6_prune) #define V_nd6_delay VNET(nd6_delay) #define V_nd6_umaxtries VNET(nd6_umaxtries) #define V_nd6_mmaxtries VNET(nd6_mmaxtries) #define V_nd6_useloopback VNET(nd6_useloopback) #define V_nd6_maxnudhint VNET(nd6_maxnudhint) #define V_nd6_gctimer VNET(nd6_gctimer) #define V_nd_defrouter VNET(nd_defrouter) #define V_nd_prefix VNET(nd_prefix) #define V_nd6_debug VNET(nd6_debug) #define V_nd6_onlink_ns_rfc4861 VNET(nd6_onlink_ns_rfc4861) #define nd6log(x) do { if (V_nd6_debug) log x; } while (/*CONSTCOND*/ 0) VNET_DECLARE(struct callout, nd6_timer_ch); #define V_nd6_timer_ch VNET(nd6_timer_ch) /* nd6_rtr.c */ VNET_DECLARE(int, nd6_defifindex); VNET_DECLARE(int, ip6_desync_factor); /* seconds */ VNET_DECLARE(u_int32_t, ip6_temp_preferred_lifetime); /* seconds */ VNET_DECLARE(u_int32_t, ip6_temp_valid_lifetime); /* seconds */ VNET_DECLARE(int, ip6_temp_regen_advance); /* seconds */ #define V_nd6_defifindex VNET(nd6_defifindex) #define V_ip6_desync_factor VNET(ip6_desync_factor) #define V_ip6_temp_preferred_lifetime VNET(ip6_temp_preferred_lifetime) #define V_ip6_temp_valid_lifetime VNET(ip6_temp_valid_lifetime) #define V_ip6_temp_regen_advance VNET(ip6_temp_regen_advance) union nd_opts { struct nd_opt_hdr *nd_opt_array[16]; /* max = ND_OPT_NONCE */ struct { struct nd_opt_hdr *zero; struct nd_opt_hdr *src_lladdr; struct nd_opt_hdr *tgt_lladdr; struct nd_opt_prefix_info *pi_beg; /* multiple opts, start */ struct nd_opt_rd_hdr *rh; struct nd_opt_mtu *mtu; struct nd_opt_hdr *__res6; struct nd_opt_hdr *__res7; struct nd_opt_hdr *__res8; struct nd_opt_hdr *__res9; struct nd_opt_hdr *__res10; struct nd_opt_hdr *__res11; struct nd_opt_hdr *__res12; struct nd_opt_hdr *__res13; struct nd_opt_nonce *nonce; struct nd_opt_hdr *__res15; struct nd_opt_hdr *search; /* multiple opts */ struct nd_opt_hdr *last; /* multiple opts */ int done; struct nd_opt_prefix_info *pi_end;/* multiple opts, end */ } nd_opt_each; }; #define nd_opts_src_lladdr nd_opt_each.src_lladdr #define nd_opts_tgt_lladdr nd_opt_each.tgt_lladdr #define nd_opts_pi nd_opt_each.pi_beg #define nd_opts_pi_end nd_opt_each.pi_end #define nd_opts_rh nd_opt_each.rh #define nd_opts_mtu nd_opt_each.mtu #define nd_opts_nonce nd_opt_each.nonce #define nd_opts_search nd_opt_each.search #define nd_opts_last nd_opt_each.last #define nd_opts_done nd_opt_each.done /* XXX: need nd6_var.h?? */ /* nd6.c */ void nd6_init(void); #ifdef VIMAGE void nd6_destroy(void); #endif struct nd_ifinfo *nd6_ifattach(struct ifnet *); void nd6_ifdetach(struct nd_ifinfo *); int nd6_is_addr_neighbor(const struct sockaddr_in6 *, struct ifnet *); void nd6_option_init(void *, int, union nd_opts *); struct nd_opt_hdr *nd6_option(union nd_opts *); int nd6_options(union nd_opts *); struct llentry *nd6_lookup(const struct in6_addr *, int, struct ifnet *); struct llentry *nd6_alloc(const struct in6_addr *, int, struct ifnet *); void nd6_setmtu(struct ifnet *); void nd6_llinfo_setstate(struct llentry *lle, int newstate); void nd6_llinfo_settimer(struct llentry *, long); void nd6_llinfo_settimer_locked(struct llentry *, long); void nd6_timer(void *); void nd6_purge(struct ifnet *); -void nd6_nud_hint(struct rtentry *, struct in6_addr *, int); int nd6_resolve(struct ifnet *, int, struct mbuf *, const struct sockaddr *, u_char *, uint32_t *); int nd6_ioctl(u_long, caddr_t, struct ifnet *); void nd6_cache_lladdr(struct ifnet *, struct in6_addr *, char *, int, int, int); void nd6_grab_holdchain(struct llentry *, struct mbuf **, struct sockaddr_in6 *); int nd6_flush_holdchain(struct ifnet *, struct ifnet *, struct mbuf *, struct sockaddr_in6 *); int nd6_add_ifa_lle(struct in6_ifaddr *); void nd6_rem_ifa_lle(struct in6_ifaddr *, int); int nd6_output_ifp(struct ifnet *, struct ifnet *, struct mbuf *, struct sockaddr_in6 *); /* nd6_nbr.c */ void nd6_na_input(struct mbuf *, int, int); void nd6_na_output(struct ifnet *, const struct in6_addr *, const struct in6_addr *, u_long, int, struct sockaddr *); void nd6_ns_input(struct mbuf *, int, int); void nd6_ns_output(struct ifnet *, const struct in6_addr *, const struct in6_addr *, const struct in6_addr *, uint8_t *); caddr_t nd6_ifptomac(struct ifnet *); void nd6_dad_init(void); void nd6_dad_start(struct ifaddr *, int); void nd6_dad_stop(struct ifaddr *); /* nd6_rtr.c */ void nd6_rs_input(struct mbuf *, int, int); void nd6_ra_input(struct mbuf *, int, int); void prelist_del(struct nd_prefix *); void defrouter_reset(void); void defrouter_select(void); void defrtrlist_del(struct nd_defrouter *); void prelist_remove(struct nd_prefix *); int nd6_prelist_add(struct nd_prefixctl *, struct nd_defrouter *, struct nd_prefix **); void pfxlist_onlink_check(void); struct nd_defrouter *defrouter_lookup(struct in6_addr *, struct ifnet *); struct nd_prefix *nd6_prefix_lookup(struct nd_prefixctl *); void rt6_flush(struct in6_addr *, struct ifnet *); int nd6_setdefaultiface(int); int in6_tmpifadd(const struct in6_ifaddr *, int, int); #endif /* _KERNEL */ #endif /* _NETINET6_ND6_H_ */ Index: head/sys/netinet6/nd6_nbr.c =================================================================== --- head/sys/netinet6/nd6_nbr.c (revision 288300) +++ head/sys/netinet6/nd6_nbr.c (revision 288301) @@ -1,1591 +1,1589 @@ /*- * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $KAME: nd6_nbr.c,v 1.86 2002/01/21 02:33:04 jinmei Exp $ */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" #include "opt_mpath.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef RADIX_MPATH #include #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #define SDL(s) ((struct sockaddr_dl *)s) struct dadq; static struct dadq *nd6_dad_find(struct ifaddr *, struct nd_opt_nonce *); static void nd6_dad_add(struct dadq *dp); static void nd6_dad_del(struct dadq *dp); static void nd6_dad_rele(struct dadq *); static void nd6_dad_starttimer(struct dadq *, int); static void nd6_dad_stoptimer(struct dadq *); static void nd6_dad_timer(struct dadq *); static void nd6_dad_duplicated(struct ifaddr *, struct dadq *); static void nd6_dad_ns_output(struct dadq *, struct ifaddr *); static void nd6_dad_ns_input(struct ifaddr *, struct nd_opt_nonce *); static void nd6_dad_na_input(struct ifaddr *); static void nd6_na_output_fib(struct ifnet *, const struct in6_addr *, const struct in6_addr *, u_long, int, struct sockaddr *, u_int); static void nd6_ns_output_fib(struct ifnet *, const struct in6_addr *, const struct in6_addr *, const struct in6_addr *, uint8_t *, u_int); static VNET_DEFINE(int, dad_enhanced) = 1; #define V_dad_enhanced VNET(dad_enhanced) SYSCTL_DECL(_net_inet6_ip6); SYSCTL_INT(_net_inet6_ip6, OID_AUTO, dad_enhanced, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(dad_enhanced), 0, "Enable Enhanced DAD, which adds a random nonce to NS messages for DAD."); static VNET_DEFINE(int, dad_maxtry) = 15; /* max # of *tries* to transmit DAD packet */ #define V_dad_maxtry VNET(dad_maxtry) /* * Input a Neighbor Solicitation Message. * * Based on RFC 2461 * Based on RFC 2462 (duplicate address detection) */ void nd6_ns_input(struct mbuf *m, int off, int icmp6len) { struct ifnet *ifp = m->m_pkthdr.rcvif; struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); struct nd_neighbor_solicit *nd_ns; struct in6_addr saddr6 = ip6->ip6_src; struct in6_addr daddr6 = ip6->ip6_dst; struct in6_addr taddr6; struct in6_addr myaddr6; char *lladdr = NULL; struct ifaddr *ifa = NULL; int lladdrlen = 0; int anycast = 0, proxy = 0, tentative = 0; int tlladdr; int rflag; union nd_opts ndopts; struct sockaddr_dl proxydl; char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; rflag = (V_ip6_forwarding) ? ND_NA_FLAG_ROUTER : 0; if (ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV && V_ip6_norbit_raif) rflag = 0; #ifndef PULLDOWN_TEST IP6_EXTHDR_CHECK(m, off, icmp6len,); nd_ns = (struct nd_neighbor_solicit *)((caddr_t)ip6 + off); #else IP6_EXTHDR_GET(nd_ns, struct nd_neighbor_solicit *, m, off, icmp6len); if (nd_ns == NULL) { ICMP6STAT_INC(icp6s_tooshort); return; } #endif ip6 = mtod(m, struct ip6_hdr *); /* adjust pointer for safety */ taddr6 = nd_ns->nd_ns_target; if (in6_setscope(&taddr6, ifp, NULL) != 0) goto bad; if (ip6->ip6_hlim != 255) { nd6log((LOG_ERR, "nd6_ns_input: invalid hlim (%d) from %s to %s on %s\n", ip6->ip6_hlim, ip6_sprintf(ip6bufs, &ip6->ip6_src), ip6_sprintf(ip6bufd, &ip6->ip6_dst), if_name(ifp))); goto bad; } if (IN6_IS_ADDR_UNSPECIFIED(&saddr6)) { /* dst has to be a solicited node multicast address. */ if (daddr6.s6_addr16[0] == IPV6_ADDR_INT16_MLL && /* don't check ifindex portion */ daddr6.s6_addr32[1] == 0 && daddr6.s6_addr32[2] == IPV6_ADDR_INT32_ONE && daddr6.s6_addr8[12] == 0xff) { ; /* good */ } else { nd6log((LOG_INFO, "nd6_ns_input: bad DAD packet " "(wrong ip6 dst)\n")); goto bad; } } else if (!V_nd6_onlink_ns_rfc4861) { struct sockaddr_in6 src_sa6; /* * According to recent IETF discussions, it is not a good idea * to accept a NS from an address which would not be deemed * to be a neighbor otherwise. This point is expected to be * clarified in future revisions of the specification. */ bzero(&src_sa6, sizeof(src_sa6)); src_sa6.sin6_family = AF_INET6; src_sa6.sin6_len = sizeof(src_sa6); src_sa6.sin6_addr = saddr6; if (nd6_is_addr_neighbor(&src_sa6, ifp) == 0) { nd6log((LOG_INFO, "nd6_ns_input: " "NS packet from non-neighbor\n")); goto bad; } } if (IN6_IS_ADDR_MULTICAST(&taddr6)) { nd6log((LOG_INFO, "nd6_ns_input: bad NS target (multicast)\n")); goto bad; } icmp6len -= sizeof(*nd_ns); nd6_option_init(nd_ns + 1, icmp6len, &ndopts); if (nd6_options(&ndopts) < 0) { nd6log((LOG_INFO, "nd6_ns_input: invalid ND option, ignored\n")); /* nd6_options have incremented stats */ goto freeit; } if (ndopts.nd_opts_src_lladdr) { lladdr = (char *)(ndopts.nd_opts_src_lladdr + 1); lladdrlen = ndopts.nd_opts_src_lladdr->nd_opt_len << 3; } if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src) && lladdr) { nd6log((LOG_INFO, "nd6_ns_input: bad DAD packet " "(link-layer address option)\n")); goto bad; } /* * Attaching target link-layer address to the NA? * (RFC 2461 7.2.4) * * NS IP dst is unicast/anycast MUST NOT add * NS IP dst is solicited-node multicast MUST add * * In implementation, we add target link-layer address by default. * We do not add one in MUST NOT cases. */ if (!IN6_IS_ADDR_MULTICAST(&daddr6)) tlladdr = 0; else tlladdr = 1; /* * Target address (taddr6) must be either: * (1) Valid unicast/anycast address for my receiving interface, * (2) Unicast address for which I'm offering proxy service, or * (3) "tentative" address on which DAD is being performed. */ /* (1) and (3) check. */ if (ifp->if_carp) ifa = (*carp_iamatch6_p)(ifp, &taddr6); else ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp, &taddr6); /* (2) check. */ if (ifa == NULL) { struct route_in6 ro; int need_proxy; bzero(&ro, sizeof(ro)); ro.ro_dst.sin6_len = sizeof(struct sockaddr_in6); ro.ro_dst.sin6_family = AF_INET6; ro.ro_dst.sin6_addr = taddr6; /* Always use the default FIB. */ #ifdef RADIX_MPATH rtalloc_mpath_fib((struct route *)&ro, ntohl(taddr6.s6_addr32[3]), RT_DEFAULT_FIB); #else in6_rtalloc(&ro, RT_DEFAULT_FIB); #endif need_proxy = (ro.ro_rt && (ro.ro_rt->rt_flags & RTF_ANNOUNCE) != 0 && ro.ro_rt->rt_gateway->sa_family == AF_LINK); if (ro.ro_rt != NULL) { if (need_proxy) proxydl = *SDL(ro.ro_rt->rt_gateway); RTFREE(ro.ro_rt); } if (need_proxy) { /* * proxy NDP for single entry */ ifa = (struct ifaddr *)in6ifa_ifpforlinklocal(ifp, IN6_IFF_NOTREADY|IN6_IFF_ANYCAST); if (ifa) proxy = 1; } } if (ifa == NULL) { /* * We've got an NS packet, and we don't have that adddress * assigned for us. We MUST silently ignore it. * See RFC2461 7.2.3. */ goto freeit; } myaddr6 = *IFA_IN6(ifa); anycast = ((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST; tentative = ((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_TENTATIVE; if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DUPLICATED) goto freeit; if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) { nd6log((LOG_INFO, "nd6_ns_input: lladdrlen mismatch for %s " "(if %d, NS packet %d)\n", ip6_sprintf(ip6bufs, &taddr6), ifp->if_addrlen, lladdrlen - 2)); goto bad; } if (IN6_ARE_ADDR_EQUAL(&myaddr6, &saddr6)) { nd6log((LOG_INFO, "nd6_ns_input: duplicate IP6 address %s\n", ip6_sprintf(ip6bufs, &saddr6))); goto freeit; } /* * We have neighbor solicitation packet, with target address equals to * one of my tentative address. * * src addr how to process? * --- --- * multicast of course, invalid (rejected in ip6_input) * unicast somebody is doing address resolution -> ignore * unspec dup address detection * * The processing is defined in RFC 2462. */ if (tentative) { /* * If source address is unspecified address, it is for * duplicate address detection. * * If not, the packet is for addess resolution; * silently ignore it. */ if (IN6_IS_ADDR_UNSPECIFIED(&saddr6)) nd6_dad_ns_input(ifa, ndopts.nd_opts_nonce); goto freeit; } /* * If the source address is unspecified address, entries must not * be created or updated. * It looks that sender is performing DAD. Output NA toward * all-node multicast address, to tell the sender that I'm using * the address. * S bit ("solicited") must be zero. */ if (IN6_IS_ADDR_UNSPECIFIED(&saddr6)) { struct in6_addr in6_all; in6_all = in6addr_linklocal_allnodes; if (in6_setscope(&in6_all, ifp, NULL) != 0) goto bad; nd6_na_output_fib(ifp, &in6_all, &taddr6, ((anycast || proxy || !tlladdr) ? 0 : ND_NA_FLAG_OVERRIDE) | rflag, tlladdr, proxy ? (struct sockaddr *)&proxydl : NULL, M_GETFIB(m)); goto freeit; } nd6_cache_lladdr(ifp, &saddr6, lladdr, lladdrlen, ND_NEIGHBOR_SOLICIT, 0); nd6_na_output_fib(ifp, &saddr6, &taddr6, ((anycast || proxy || !tlladdr) ? 0 : ND_NA_FLAG_OVERRIDE) | rflag | ND_NA_FLAG_SOLICITED, tlladdr, proxy ? (struct sockaddr *)&proxydl : NULL, M_GETFIB(m)); freeit: if (ifa != NULL) ifa_free(ifa); m_freem(m); return; bad: nd6log((LOG_ERR, "nd6_ns_input: src=%s\n", ip6_sprintf(ip6bufs, &saddr6))); nd6log((LOG_ERR, "nd6_ns_input: dst=%s\n", ip6_sprintf(ip6bufs, &daddr6))); nd6log((LOG_ERR, "nd6_ns_input: tgt=%s\n", ip6_sprintf(ip6bufs, &taddr6))); ICMP6STAT_INC(icp6s_badns); if (ifa != NULL) ifa_free(ifa); m_freem(m); } /* * Output a Neighbor Solicitation Message. Caller specifies: * - ICMP6 header source IP6 address * - ND6 header target IP6 address * - ND6 header source datalink address * * Based on RFC 2461 * Based on RFC 2462 (duplicate address detection) * * ln - for source address determination * nonce - If non-NULL, NS is used for duplicate address detection and * the value (length is ND_OPT_NONCE_LEN) is used as a random nonce. */ static void nd6_ns_output_fib(struct ifnet *ifp, const struct in6_addr *saddr6, const struct in6_addr *daddr6, const struct in6_addr *taddr6, uint8_t *nonce, u_int fibnum) { struct mbuf *m; struct m_tag *mtag; struct ip6_hdr *ip6; struct nd_neighbor_solicit *nd_ns; struct ip6_moptions im6o; int icmp6len; int maxlen; caddr_t mac; struct route_in6 ro; if (IN6_IS_ADDR_MULTICAST(taddr6)) return; /* estimate the size of message */ maxlen = sizeof(*ip6) + sizeof(*nd_ns); maxlen += (sizeof(struct nd_opt_hdr) + ifp->if_addrlen + 7) & ~7; KASSERT(max_linkhdr + maxlen <= MCLBYTES, ( "%s: max_linkhdr + maxlen > MCLBYTES (%d + %d > %d)", __func__, max_linkhdr, maxlen, MCLBYTES)); if (max_linkhdr + maxlen > MHLEN) m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR); else m = m_gethdr(M_NOWAIT, MT_DATA); if (m == NULL) return; M_SETFIB(m, fibnum); bzero(&ro, sizeof(ro)); if (daddr6 == NULL || IN6_IS_ADDR_MULTICAST(daddr6)) { m->m_flags |= M_MCAST; im6o.im6o_multicast_ifp = ifp; im6o.im6o_multicast_hlim = 255; im6o.im6o_multicast_loop = 0; } icmp6len = sizeof(*nd_ns); m->m_pkthdr.len = m->m_len = sizeof(*ip6) + icmp6len; m->m_data += max_linkhdr; /* or M_ALIGN() equivalent? */ /* fill neighbor solicitation packet */ ip6 = mtod(m, struct ip6_hdr *); ip6->ip6_flow = 0; ip6->ip6_vfc &= ~IPV6_VERSION_MASK; ip6->ip6_vfc |= IPV6_VERSION; /* ip6->ip6_plen will be set later */ ip6->ip6_nxt = IPPROTO_ICMPV6; ip6->ip6_hlim = 255; if (daddr6) ip6->ip6_dst = *daddr6; else { ip6->ip6_dst.s6_addr16[0] = IPV6_ADDR_INT16_MLL; ip6->ip6_dst.s6_addr16[1] = 0; ip6->ip6_dst.s6_addr32[1] = 0; ip6->ip6_dst.s6_addr32[2] = IPV6_ADDR_INT32_ONE; ip6->ip6_dst.s6_addr32[3] = taddr6->s6_addr32[3]; ip6->ip6_dst.s6_addr8[12] = 0xff; if (in6_setscope(&ip6->ip6_dst, ifp, NULL) != 0) goto bad; } if (nonce == NULL) { struct ifaddr *ifa = NULL; /* * RFC2461 7.2.2: * "If the source address of the packet prompting the * solicitation is the same as one of the addresses assigned * to the outgoing interface, that address SHOULD be placed * in the IP Source Address of the outgoing solicitation. * Otherwise, any one of the addresses assigned to the * interface should be used." * * We use the source address for the prompting packet * (saddr6), if saddr6 belongs to the outgoing interface. * Otherwise, we perform the source address selection as usual. */ if (saddr6 != NULL) ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp, saddr6); if (ifa != NULL) { /* ip6_src set already. */ ip6->ip6_src = *saddr6; ifa_free(ifa); } else { int error; struct sockaddr_in6 dst_sa; struct in6_addr src_in; struct ifnet *oifp; bzero(&dst_sa, sizeof(dst_sa)); dst_sa.sin6_family = AF_INET6; dst_sa.sin6_len = sizeof(dst_sa); dst_sa.sin6_addr = ip6->ip6_dst; oifp = ifp; error = in6_selectsrc(&dst_sa, NULL, NULL, &ro, NULL, &oifp, &src_in); if (error) { char ip6buf[INET6_ADDRSTRLEN]; nd6log((LOG_DEBUG, "%s: source can't be " "determined: dst=%s, error=%d\n", __func__, ip6_sprintf(ip6buf, &dst_sa.sin6_addr), error)); goto bad; } ip6->ip6_src = src_in; } } else { /* * Source address for DAD packet must always be IPv6 * unspecified address. (0::0) * We actually don't have to 0-clear the address (we did it * above), but we do so here explicitly to make the intention * clearer. */ bzero(&ip6->ip6_src, sizeof(ip6->ip6_src)); } nd_ns = (struct nd_neighbor_solicit *)(ip6 + 1); nd_ns->nd_ns_type = ND_NEIGHBOR_SOLICIT; nd_ns->nd_ns_code = 0; nd_ns->nd_ns_reserved = 0; nd_ns->nd_ns_target = *taddr6; in6_clearscope(&nd_ns->nd_ns_target); /* XXX */ /* * Add source link-layer address option. * * spec implementation * --- --- * DAD packet MUST NOT do not add the option * there's no link layer address: * impossible do not add the option * there's link layer address: * Multicast NS MUST add one add the option * Unicast NS SHOULD add one add the option */ if (nonce == NULL && (mac = nd6_ifptomac(ifp))) { int optlen = sizeof(struct nd_opt_hdr) + ifp->if_addrlen; struct nd_opt_hdr *nd_opt = (struct nd_opt_hdr *)(nd_ns + 1); /* 8 byte alignments... */ optlen = (optlen + 7) & ~7; m->m_pkthdr.len += optlen; m->m_len += optlen; icmp6len += optlen; bzero((caddr_t)nd_opt, optlen); nd_opt->nd_opt_type = ND_OPT_SOURCE_LINKADDR; nd_opt->nd_opt_len = optlen >> 3; bcopy(mac, (caddr_t)(nd_opt + 1), ifp->if_addrlen); } /* * Add a Nonce option (RFC 3971) to detect looped back NS messages. * This behavior is documented as Enhanced Duplicate Address * Detection in RFC 7527. * net.inet6.ip6.dad_enhanced=0 disables this. */ if (V_dad_enhanced != 0 && nonce != NULL) { int optlen = sizeof(struct nd_opt_hdr) + ND_OPT_NONCE_LEN; struct nd_opt_hdr *nd_opt = (struct nd_opt_hdr *)(nd_ns + 1); /* 8-byte alignment is required. */ optlen = (optlen + 7) & ~7; m->m_pkthdr.len += optlen; m->m_len += optlen; icmp6len += optlen; bzero((caddr_t)nd_opt, optlen); nd_opt->nd_opt_type = ND_OPT_NONCE; nd_opt->nd_opt_len = optlen >> 3; bcopy(nonce, (caddr_t)(nd_opt + 1), ND_OPT_NONCE_LEN); } ip6->ip6_plen = htons((u_short)icmp6len); nd_ns->nd_ns_cksum = 0; nd_ns->nd_ns_cksum = in6_cksum(m, IPPROTO_ICMPV6, sizeof(*ip6), icmp6len); if (send_sendso_input_hook != NULL) { mtag = m_tag_get(PACKET_TAG_ND_OUTGOING, sizeof(unsigned short), M_NOWAIT); if (mtag == NULL) goto bad; *(unsigned short *)(mtag + 1) = nd_ns->nd_ns_type; m_tag_prepend(m, mtag); } ip6_output(m, NULL, &ro, (nonce != NULL) ? IPV6_UNSPECSRC : 0, &im6o, NULL, NULL); icmp6_ifstat_inc(ifp, ifs6_out_msg); icmp6_ifstat_inc(ifp, ifs6_out_neighborsolicit); ICMP6STAT_INC(icp6s_outhist[ND_NEIGHBOR_SOLICIT]); /* We don't cache this route. */ RO_RTFREE(&ro); return; bad: if (ro.ro_rt) { RTFREE(ro.ro_rt); } m_freem(m); return; } #ifndef BURN_BRIDGES void nd6_ns_output(struct ifnet *ifp, const struct in6_addr *saddr6, const struct in6_addr *daddr6, const struct in6_addr *taddr6,uint8_t *nonce) { nd6_ns_output_fib(ifp, saddr6, daddr6, taddr6, nonce, RT_DEFAULT_FIB); } #endif /* * Neighbor advertisement input handling. * * Based on RFC 2461 * Based on RFC 2462 (duplicate address detection) * * the following items are not implemented yet: * - proxy advertisement delay rule (RFC2461 7.2.8, last paragraph, SHOULD) * - anycast advertisement delay rule (RFC2461 7.2.7, SHOULD) */ void nd6_na_input(struct mbuf *m, int off, int icmp6len) { struct ifnet *ifp = m->m_pkthdr.rcvif; struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); struct nd_neighbor_advert *nd_na; struct in6_addr daddr6 = ip6->ip6_dst; struct in6_addr taddr6; int flags; int is_router; int is_solicited; int is_override; char *lladdr = NULL; int lladdrlen = 0; int checklink = 0; struct ifaddr *ifa; struct llentry *ln = NULL; union nd_opts ndopts; struct mbuf *chain = NULL; struct sockaddr_in6 sin6; char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; if (ip6->ip6_hlim != 255) { nd6log((LOG_ERR, "nd6_na_input: invalid hlim (%d) from %s to %s on %s\n", ip6->ip6_hlim, ip6_sprintf(ip6bufs, &ip6->ip6_src), ip6_sprintf(ip6bufd, &ip6->ip6_dst), if_name(ifp))); goto bad; } #ifndef PULLDOWN_TEST IP6_EXTHDR_CHECK(m, off, icmp6len,); nd_na = (struct nd_neighbor_advert *)((caddr_t)ip6 + off); #else IP6_EXTHDR_GET(nd_na, struct nd_neighbor_advert *, m, off, icmp6len); if (nd_na == NULL) { ICMP6STAT_INC(icp6s_tooshort); return; } #endif flags = nd_na->nd_na_flags_reserved; is_router = ((flags & ND_NA_FLAG_ROUTER) != 0); is_solicited = ((flags & ND_NA_FLAG_SOLICITED) != 0); is_override = ((flags & ND_NA_FLAG_OVERRIDE) != 0); memset(&sin6, 0, sizeof(sin6)); taddr6 = nd_na->nd_na_target; if (in6_setscope(&taddr6, ifp, NULL)) goto bad; /* XXX: impossible */ if (IN6_IS_ADDR_MULTICAST(&taddr6)) { nd6log((LOG_ERR, "nd6_na_input: invalid target address %s\n", ip6_sprintf(ip6bufs, &taddr6))); goto bad; } if (IN6_IS_ADDR_MULTICAST(&daddr6)) if (is_solicited) { nd6log((LOG_ERR, "nd6_na_input: a solicited adv is multicasted\n")); goto bad; } icmp6len -= sizeof(*nd_na); nd6_option_init(nd_na + 1, icmp6len, &ndopts); if (nd6_options(&ndopts) < 0) { nd6log((LOG_INFO, "nd6_na_input: invalid ND option, ignored\n")); /* nd6_options have incremented stats */ goto freeit; } if (ndopts.nd_opts_tgt_lladdr) { lladdr = (char *)(ndopts.nd_opts_tgt_lladdr + 1); lladdrlen = ndopts.nd_opts_tgt_lladdr->nd_opt_len << 3; } /* * This effectively disables the DAD check on a non-master CARP * address. */ if (ifp->if_carp) ifa = (*carp_iamatch6_p)(ifp, &taddr6); else ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp, &taddr6); /* * Target address matches one of my interface address. * * If my address is tentative, this means that there's somebody * already using the same address as mine. This indicates DAD failure. * This is defined in RFC 2462. * * Otherwise, process as defined in RFC 2461. */ if (ifa && (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_TENTATIVE)) { nd6_dad_na_input(ifa); ifa_free(ifa); goto freeit; } /* Just for safety, maybe unnecessary. */ if (ifa) { ifa_free(ifa); log(LOG_ERR, "nd6_na_input: duplicate IP6 address %s\n", ip6_sprintf(ip6bufs, &taddr6)); goto freeit; } if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) { nd6log((LOG_INFO, "nd6_na_input: lladdrlen mismatch for %s " "(if %d, NA packet %d)\n", ip6_sprintf(ip6bufs, &taddr6), ifp->if_addrlen, lladdrlen - 2)); goto bad; } /* * If no neighbor cache entry is found, NA SHOULD silently be * discarded. */ IF_AFDATA_RLOCK(ifp); ln = nd6_lookup(&taddr6, LLE_EXCLUSIVE, ifp); IF_AFDATA_RUNLOCK(ifp); if (ln == NULL) { goto freeit; } if (ln->ln_state == ND6_LLINFO_INCOMPLETE) { /* * If the link-layer has address, and no lladdr option came, * discard the packet. */ if (ifp->if_addrlen && lladdr == NULL) { goto freeit; } /* * Record link-layer address, and update the state. */ bcopy(lladdr, &ln->ll_addr, ifp->if_addrlen); ln->la_flags |= LLE_VALID; EVENTHANDLER_INVOKE(lle_event, ln, LLENTRY_RESOLVED); - if (is_solicited) { + if (is_solicited) nd6_llinfo_setstate(ln, ND6_LLINFO_REACHABLE); - ln->ln_byhint = 0; - } else + else nd6_llinfo_setstate(ln, ND6_LLINFO_STALE); if ((ln->ln_router = is_router) != 0) { /* * This means a router's state has changed from * non-reachable to probably reachable, and might * affect the status of associated prefixes.. */ checklink = 1; } } else { int llchange; /* * Check if the link-layer address has changed or not. */ if (lladdr == NULL) llchange = 0; else { if (ln->la_flags & LLE_VALID) { if (bcmp(lladdr, &ln->ll_addr, ifp->if_addrlen)) llchange = 1; else llchange = 0; } else llchange = 1; } /* * This is VERY complex. Look at it with care. * * override solicit lladdr llchange action * (L: record lladdr) * * 0 0 n -- (2c) * 0 0 y n (2b) L * 0 0 y y (1) REACHABLE->STALE * 0 1 n -- (2c) *->REACHABLE * 0 1 y n (2b) L *->REACHABLE * 0 1 y y (1) REACHABLE->STALE * 1 0 n -- (2a) * 1 0 y n (2a) L * 1 0 y y (2a) L *->STALE * 1 1 n -- (2a) *->REACHABLE * 1 1 y n (2a) L *->REACHABLE * 1 1 y y (2a) L *->REACHABLE */ if (!is_override && (lladdr != NULL && llchange)) { /* (1) */ /* * If state is REACHABLE, make it STALE. * no other updates should be done. */ if (ln->ln_state == ND6_LLINFO_REACHABLE) nd6_llinfo_setstate(ln, ND6_LLINFO_STALE); goto freeit; } else if (is_override /* (2a) */ || (!is_override && (lladdr != NULL && !llchange)) /* (2b) */ || lladdr == NULL) { /* (2c) */ /* * Update link-local address, if any. */ if (lladdr != NULL) { bcopy(lladdr, &ln->ll_addr, ifp->if_addrlen); ln->la_flags |= LLE_VALID; EVENTHANDLER_INVOKE(lle_event, ln, LLENTRY_RESOLVED); } /* * If solicited, make the state REACHABLE. * If not solicited and the link-layer address was * changed, make it STALE. */ - if (is_solicited) { - ln->ln_byhint = 0; + if (is_solicited) nd6_llinfo_setstate(ln, ND6_LLINFO_REACHABLE); - } else { + else { if (lladdr != NULL && llchange) nd6_llinfo_setstate(ln, ND6_LLINFO_STALE); } } if (ln->ln_router && !is_router) { /* * The peer dropped the router flag. * Remove the sender from the Default Router List and * update the Destination Cache entries. */ struct nd_defrouter *dr; struct in6_addr *in6; in6 = &ln->r_l3addr.addr6; /* * Lock to protect the default router list. * XXX: this might be unnecessary, since this function * is only called under the network software interrupt * context. However, we keep it just for safety. */ dr = defrouter_lookup(in6, ln->lle_tbl->llt_ifp); if (dr) defrtrlist_del(dr); else if (ND_IFINFO(ln->lle_tbl->llt_ifp)->flags & ND6_IFF_ACCEPT_RTADV) { /* * Even if the neighbor is not in the default * router list, the neighbor may be used * as a next hop for some destinations * (e.g. redirect case). So we must * call rt6_flush explicitly. */ rt6_flush(&ip6->ip6_src, ifp); } } ln->ln_router = is_router; } /* XXX - QL * Does this matter? * rt->rt_flags &= ~RTF_REJECT; */ ln->la_asked = 0; if (ln->la_hold != NULL) nd6_grab_holdchain(ln, &chain, &sin6); freeit: if (ln != NULL) LLE_WUNLOCK(ln); if (chain != NULL) nd6_flush_holdchain(ifp, ifp, chain, &sin6); if (checklink) pfxlist_onlink_check(); m_freem(m); return; bad: if (ln != NULL) LLE_WUNLOCK(ln); ICMP6STAT_INC(icp6s_badna); m_freem(m); } /* * Neighbor advertisement output handling. * * Based on RFC 2461 * * the following items are not implemented yet: * - proxy advertisement delay rule (RFC2461 7.2.8, last paragraph, SHOULD) * - anycast advertisement delay rule (RFC2461 7.2.7, SHOULD) * * tlladdr - 1 if include target link-layer address * sdl0 - sockaddr_dl (= proxy NA) or NULL */ static void nd6_na_output_fib(struct ifnet *ifp, const struct in6_addr *daddr6_0, const struct in6_addr *taddr6, u_long flags, int tlladdr, struct sockaddr *sdl0, u_int fibnum) { struct mbuf *m; struct m_tag *mtag; struct ifnet *oifp; struct ip6_hdr *ip6; struct nd_neighbor_advert *nd_na; struct ip6_moptions im6o; struct in6_addr src, daddr6; struct sockaddr_in6 dst_sa; int icmp6len, maxlen, error; caddr_t mac = NULL; struct route_in6 ro; bzero(&ro, sizeof(ro)); daddr6 = *daddr6_0; /* make a local copy for modification */ /* estimate the size of message */ maxlen = sizeof(*ip6) + sizeof(*nd_na); maxlen += (sizeof(struct nd_opt_hdr) + ifp->if_addrlen + 7) & ~7; KASSERT(max_linkhdr + maxlen <= MCLBYTES, ( "%s: max_linkhdr + maxlen > MCLBYTES (%d + %d > %d)", __func__, max_linkhdr, maxlen, MCLBYTES)); if (max_linkhdr + maxlen > MHLEN) m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR); else m = m_gethdr(M_NOWAIT, MT_DATA); if (m == NULL) return; M_SETFIB(m, fibnum); if (IN6_IS_ADDR_MULTICAST(&daddr6)) { m->m_flags |= M_MCAST; im6o.im6o_multicast_ifp = ifp; im6o.im6o_multicast_hlim = 255; im6o.im6o_multicast_loop = 0; } icmp6len = sizeof(*nd_na); m->m_pkthdr.len = m->m_len = sizeof(struct ip6_hdr) + icmp6len; m->m_data += max_linkhdr; /* or M_ALIGN() equivalent? */ /* fill neighbor advertisement packet */ ip6 = mtod(m, struct ip6_hdr *); ip6->ip6_flow = 0; ip6->ip6_vfc &= ~IPV6_VERSION_MASK; ip6->ip6_vfc |= IPV6_VERSION; ip6->ip6_nxt = IPPROTO_ICMPV6; ip6->ip6_hlim = 255; if (IN6_IS_ADDR_UNSPECIFIED(&daddr6)) { /* reply to DAD */ daddr6.s6_addr16[0] = IPV6_ADDR_INT16_MLL; daddr6.s6_addr16[1] = 0; daddr6.s6_addr32[1] = 0; daddr6.s6_addr32[2] = 0; daddr6.s6_addr32[3] = IPV6_ADDR_INT32_ONE; if (in6_setscope(&daddr6, ifp, NULL)) goto bad; flags &= ~ND_NA_FLAG_SOLICITED; } ip6->ip6_dst = daddr6; bzero(&dst_sa, sizeof(struct sockaddr_in6)); dst_sa.sin6_family = AF_INET6; dst_sa.sin6_len = sizeof(struct sockaddr_in6); dst_sa.sin6_addr = daddr6; /* * Select a source whose scope is the same as that of the dest. */ bcopy(&dst_sa, &ro.ro_dst, sizeof(dst_sa)); oifp = ifp; error = in6_selectsrc(&dst_sa, NULL, NULL, &ro, NULL, &oifp, &src); if (error) { char ip6buf[INET6_ADDRSTRLEN]; nd6log((LOG_DEBUG, "nd6_na_output: source can't be " "determined: dst=%s, error=%d\n", ip6_sprintf(ip6buf, &dst_sa.sin6_addr), error)); goto bad; } ip6->ip6_src = src; nd_na = (struct nd_neighbor_advert *)(ip6 + 1); nd_na->nd_na_type = ND_NEIGHBOR_ADVERT; nd_na->nd_na_code = 0; nd_na->nd_na_target = *taddr6; in6_clearscope(&nd_na->nd_na_target); /* XXX */ /* * "tlladdr" indicates NS's condition for adding tlladdr or not. * see nd6_ns_input() for details. * Basically, if NS packet is sent to unicast/anycast addr, * target lladdr option SHOULD NOT be included. */ if (tlladdr) { /* * sdl0 != NULL indicates proxy NA. If we do proxy, use * lladdr in sdl0. If we are not proxying (sending NA for * my address) use lladdr configured for the interface. */ if (sdl0 == NULL) { if (ifp->if_carp) mac = (*carp_macmatch6_p)(ifp, m, taddr6); if (mac == NULL) mac = nd6_ifptomac(ifp); } else if (sdl0->sa_family == AF_LINK) { struct sockaddr_dl *sdl; sdl = (struct sockaddr_dl *)sdl0; if (sdl->sdl_alen == ifp->if_addrlen) mac = LLADDR(sdl); } } if (tlladdr && mac) { int optlen = sizeof(struct nd_opt_hdr) + ifp->if_addrlen; struct nd_opt_hdr *nd_opt = (struct nd_opt_hdr *)(nd_na + 1); /* roundup to 8 bytes alignment! */ optlen = (optlen + 7) & ~7; m->m_pkthdr.len += optlen; m->m_len += optlen; icmp6len += optlen; bzero((caddr_t)nd_opt, optlen); nd_opt->nd_opt_type = ND_OPT_TARGET_LINKADDR; nd_opt->nd_opt_len = optlen >> 3; bcopy(mac, (caddr_t)(nd_opt + 1), ifp->if_addrlen); } else flags &= ~ND_NA_FLAG_OVERRIDE; ip6->ip6_plen = htons((u_short)icmp6len); nd_na->nd_na_flags_reserved = flags; nd_na->nd_na_cksum = 0; nd_na->nd_na_cksum = in6_cksum(m, IPPROTO_ICMPV6, sizeof(struct ip6_hdr), icmp6len); if (send_sendso_input_hook != NULL) { mtag = m_tag_get(PACKET_TAG_ND_OUTGOING, sizeof(unsigned short), M_NOWAIT); if (mtag == NULL) goto bad; *(unsigned short *)(mtag + 1) = nd_na->nd_na_type; m_tag_prepend(m, mtag); } ip6_output(m, NULL, &ro, 0, &im6o, NULL, NULL); icmp6_ifstat_inc(ifp, ifs6_out_msg); icmp6_ifstat_inc(ifp, ifs6_out_neighboradvert); ICMP6STAT_INC(icp6s_outhist[ND_NEIGHBOR_ADVERT]); /* We don't cache this route. */ RO_RTFREE(&ro); return; bad: if (ro.ro_rt) { RTFREE(ro.ro_rt); } m_freem(m); return; } #ifndef BURN_BRIDGES void nd6_na_output(struct ifnet *ifp, const struct in6_addr *daddr6_0, const struct in6_addr *taddr6, u_long flags, int tlladdr, struct sockaddr *sdl0) { nd6_na_output_fib(ifp, daddr6_0, taddr6, flags, tlladdr, sdl0, RT_DEFAULT_FIB); } #endif caddr_t nd6_ifptomac(struct ifnet *ifp) { switch (ifp->if_type) { case IFT_ARCNET: case IFT_ETHER: case IFT_FDDI: case IFT_IEEE1394: case IFT_L2VLAN: case IFT_IEEE80211: case IFT_INFINIBAND: case IFT_BRIDGE: case IFT_ISO88025: return IF_LLADDR(ifp); default: return NULL; } } struct dadq { TAILQ_ENTRY(dadq) dad_list; struct ifaddr *dad_ifa; int dad_count; /* max NS to send */ int dad_ns_tcount; /* # of trials to send NS */ int dad_ns_ocount; /* NS sent so far */ int dad_ns_icount; int dad_na_icount; int dad_ns_lcount; /* looped back NS */ int dad_loopbackprobe; /* probing state for loopback detection */ struct callout dad_timer_ch; struct vnet *dad_vnet; u_int dad_refcnt; #define ND_OPT_NONCE_LEN32 \ ((ND_OPT_NONCE_LEN + sizeof(uint32_t) - 1)/sizeof(uint32_t)) uint32_t dad_nonce[ND_OPT_NONCE_LEN32]; }; static VNET_DEFINE(TAILQ_HEAD(, dadq), dadq); static VNET_DEFINE(struct rwlock, dad_rwlock); #define V_dadq VNET(dadq) #define V_dad_rwlock VNET(dad_rwlock) #define DADQ_RLOCK() rw_rlock(&V_dad_rwlock) #define DADQ_RUNLOCK() rw_runlock(&V_dad_rwlock) #define DADQ_WLOCK() rw_wlock(&V_dad_rwlock) #define DADQ_WUNLOCK() rw_wunlock(&V_dad_rwlock) static void nd6_dad_add(struct dadq *dp) { DADQ_WLOCK(); TAILQ_INSERT_TAIL(&V_dadq, dp, dad_list); DADQ_WUNLOCK(); } static void nd6_dad_del(struct dadq *dp) { DADQ_WLOCK(); TAILQ_REMOVE(&V_dadq, dp, dad_list); DADQ_WUNLOCK(); nd6_dad_rele(dp); } static struct dadq * nd6_dad_find(struct ifaddr *ifa, struct nd_opt_nonce *n) { struct dadq *dp; DADQ_RLOCK(); TAILQ_FOREACH(dp, &V_dadq, dad_list) { if (dp->dad_ifa != ifa) continue; /* * Skip if the nonce matches the received one. * +2 in the length is required because of type and * length fields are included in a header. */ if (n != NULL && n->nd_opt_nonce_len == (ND_OPT_NONCE_LEN + 2) / 8 && memcmp(&n->nd_opt_nonce[0], &dp->dad_nonce[0], ND_OPT_NONCE_LEN) == 0) { dp->dad_ns_lcount++; continue; } refcount_acquire(&dp->dad_refcnt); break; } DADQ_RUNLOCK(); return (dp); } static void nd6_dad_starttimer(struct dadq *dp, int ticks) { callout_reset(&dp->dad_timer_ch, ticks, (void (*)(void *))nd6_dad_timer, (void *)dp); } static void nd6_dad_stoptimer(struct dadq *dp) { callout_drain(&dp->dad_timer_ch); } static void nd6_dad_rele(struct dadq *dp) { if (refcount_release(&dp->dad_refcnt)) { ifa_free(dp->dad_ifa); free(dp, M_IP6NDP); } } void nd6_dad_init(void) { rw_init(&V_dad_rwlock, "nd6 DAD queue"); TAILQ_INIT(&V_dadq); } /* * Start Duplicate Address Detection (DAD) for specified interface address. */ void nd6_dad_start(struct ifaddr *ifa, int delay) { struct in6_ifaddr *ia = (struct in6_ifaddr *)ifa; struct dadq *dp; char ip6buf[INET6_ADDRSTRLEN]; /* * If we don't need DAD, don't do it. * There are several cases: * - DAD is disabled (ip6_dad_count == 0) * - the interface address is anycast */ if (!(ia->ia6_flags & IN6_IFF_TENTATIVE)) { log(LOG_DEBUG, "nd6_dad_start: called with non-tentative address " "%s(%s)\n", ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr), ifa->ifa_ifp ? if_name(ifa->ifa_ifp) : "???"); return; } if (ia->ia6_flags & IN6_IFF_ANYCAST) { ia->ia6_flags &= ~IN6_IFF_TENTATIVE; return; } if (!V_ip6_dad_count) { ia->ia6_flags &= ~IN6_IFF_TENTATIVE; return; } if (ifa->ifa_ifp == NULL) panic("nd6_dad_start: ifa->ifa_ifp == NULL"); if (ND_IFINFO(ifa->ifa_ifp)->flags & ND6_IFF_NO_DAD) { ia->ia6_flags &= ~IN6_IFF_TENTATIVE; return; } if (!(ifa->ifa_ifp->if_flags & IFF_UP) || !(ifa->ifa_ifp->if_drv_flags & IFF_DRV_RUNNING) || (ND_IFINFO(ifa->ifa_ifp)->flags & ND6_IFF_IFDISABLED)) { ia->ia6_flags |= IN6_IFF_TENTATIVE; return; } if ((dp = nd6_dad_find(ifa, NULL)) != NULL) { /* DAD already in progress */ nd6_dad_rele(dp); return; } dp = malloc(sizeof(*dp), M_IP6NDP, M_NOWAIT | M_ZERO); if (dp == NULL) { log(LOG_ERR, "nd6_dad_start: memory allocation failed for " "%s(%s)\n", ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr), ifa->ifa_ifp ? if_name(ifa->ifa_ifp) : "???"); return; } callout_init(&dp->dad_timer_ch, 0); #ifdef VIMAGE dp->dad_vnet = curvnet; #endif nd6log((LOG_DEBUG, "%s: starting DAD for %s\n", if_name(ifa->ifa_ifp), ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr))); /* * Send NS packet for DAD, ip6_dad_count times. * Note that we must delay the first transmission, if this is the * first packet to be sent from the interface after interface * (re)initialization. */ dp->dad_ifa = ifa; ifa_ref(dp->dad_ifa); dp->dad_count = V_ip6_dad_count; dp->dad_ns_icount = dp->dad_na_icount = 0; dp->dad_ns_ocount = dp->dad_ns_tcount = 0; dp->dad_ns_lcount = dp->dad_loopbackprobe = 0; refcount_init(&dp->dad_refcnt, 1); nd6_dad_add(dp); if (delay == 0) { nd6_dad_ns_output(dp, ifa); nd6_dad_starttimer(dp, (long)ND_IFINFO(ifa->ifa_ifp)->retrans * hz / 1000); } else { nd6_dad_starttimer(dp, delay); } } /* * terminate DAD unconditionally. used for address removals. */ void nd6_dad_stop(struct ifaddr *ifa) { struct dadq *dp; dp = nd6_dad_find(ifa, NULL); if (!dp) { /* DAD wasn't started yet */ return; } nd6_dad_stoptimer(dp); /* * The DAD queue entry may have been removed by nd6_dad_timer() while * we were waiting for it to stop, so re-do the lookup. */ nd6_dad_rele(dp); if (nd6_dad_find(ifa, NULL) == NULL) return; nd6_dad_del(dp); nd6_dad_rele(dp); } static void nd6_dad_timer(struct dadq *dp) { CURVNET_SET(dp->dad_vnet); struct ifaddr *ifa = dp->dad_ifa; struct ifnet *ifp = dp->dad_ifa->ifa_ifp; struct in6_ifaddr *ia = (struct in6_ifaddr *)ifa; char ip6buf[INET6_ADDRSTRLEN]; /* Sanity check */ if (ia == NULL) { log(LOG_ERR, "nd6_dad_timer: called with null parameter\n"); goto err; } if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) { /* Do not need DAD for ifdisabled interface. */ log(LOG_ERR, "nd6_dad_timer: cancel DAD on %s because of " "ND6_IFF_IFDISABLED.\n", ifp->if_xname); goto err; } if (ia->ia6_flags & IN6_IFF_DUPLICATED) { log(LOG_ERR, "nd6_dad_timer: called with duplicated address " "%s(%s)\n", ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr), ifa->ifa_ifp ? if_name(ifa->ifa_ifp) : "???"); goto err; } if ((ia->ia6_flags & IN6_IFF_TENTATIVE) == 0) { log(LOG_ERR, "nd6_dad_timer: called with non-tentative address " "%s(%s)\n", ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr), ifa->ifa_ifp ? if_name(ifa->ifa_ifp) : "???"); goto err; } /* Stop DAD if the interface is down even after dad_maxtry attempts. */ if ((dp->dad_ns_tcount > V_dad_maxtry) && (((ifp->if_flags & IFF_UP) == 0) || ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0))) { nd6log((LOG_INFO, "%s: could not run DAD, driver problem?\n", if_name(ifa->ifa_ifp))); goto err; } /* Need more checks? */ if (dp->dad_ns_ocount < dp->dad_count) { /* * We have more NS to go. Send NS packet for DAD. */ nd6_dad_ns_output(dp, ifa); nd6_dad_starttimer(dp, (long)ND_IFINFO(ifa->ifa_ifp)->retrans * hz / 1000); goto done; } else { /* * We have transmitted sufficient number of DAD packets. * See what we've got. */ if (dp->dad_ns_icount > 0 || dp->dad_na_icount > 0) /* We've seen NS or NA, means DAD has failed. */ nd6_dad_duplicated(ifa, dp); else if (V_dad_enhanced != 0 && dp->dad_ns_lcount > 0 && dp->dad_ns_lcount > dp->dad_loopbackprobe) { /* * Sec. 4.1 in RFC 7527 requires transmission of * additional probes until the loopback condition * becomes clear when a looped back probe is detected. */ log(LOG_ERR, "%s: a looped back NS message is " "detected during DAD for %s. " "Another DAD probes are being sent.\n", if_name(ifa->ifa_ifp), ip6_sprintf(ip6buf, IFA_IN6(ifa))); dp->dad_loopbackprobe = dp->dad_ns_lcount; /* * Send an NS immediately and increase dad_count by * V_nd6_mmaxtries - 1. */ nd6_dad_ns_output(dp, ifa); dp->dad_count = dp->dad_ns_ocount + V_nd6_mmaxtries - 1; nd6_dad_starttimer(dp, (long)ND_IFINFO(ifa->ifa_ifp)->retrans * hz / 1000); goto done; } else { /* * We are done with DAD. No NA came, no NS came. * No duplicate address found. Check IFDISABLED flag * again in case that it is changed between the * beginning of this function and here. */ if ((ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) == 0) ia->ia6_flags &= ~IN6_IFF_TENTATIVE; nd6log((LOG_DEBUG, "%s: DAD complete for %s - no duplicates found\n", if_name(ifa->ifa_ifp), ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr))); if (dp->dad_ns_lcount > 0) log(LOG_ERR, "%s: DAD completed while " "a looped back NS message is detected " "during DAD for %s.\n", if_name(ifa->ifa_ifp), ip6_sprintf(ip6buf, IFA_IN6(ifa))); } } err: nd6_dad_del(dp); done: CURVNET_RESTORE(); } static void nd6_dad_duplicated(struct ifaddr *ifa, struct dadq *dp) { struct in6_ifaddr *ia = (struct in6_ifaddr *)ifa; struct ifnet *ifp; char ip6buf[INET6_ADDRSTRLEN]; log(LOG_ERR, "%s: DAD detected duplicate IPv6 address %s: " "NS in/out/loopback=%d/%d/%d, NA in=%d\n", if_name(ifa->ifa_ifp), ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr), dp->dad_ns_icount, dp->dad_ns_ocount, dp->dad_ns_lcount, dp->dad_na_icount); ia->ia6_flags &= ~IN6_IFF_TENTATIVE; ia->ia6_flags |= IN6_IFF_DUPLICATED; ifp = ifa->ifa_ifp; log(LOG_ERR, "%s: DAD complete for %s - duplicate found\n", if_name(ifp), ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr)); log(LOG_ERR, "%s: manual intervention required\n", if_name(ifp)); /* * If the address is a link-local address formed from an interface * identifier based on the hardware address which is supposed to be * uniquely assigned (e.g., EUI-64 for an Ethernet interface), IP * operation on the interface SHOULD be disabled. * [RFC 4862, Section 5.4.5] */ if (IN6_IS_ADDR_LINKLOCAL(&ia->ia_addr.sin6_addr)) { struct in6_addr in6; /* * To avoid over-reaction, we only apply this logic when we are * very sure that hardware addresses are supposed to be unique. */ switch (ifp->if_type) { case IFT_ETHER: case IFT_FDDI: case IFT_ATM: case IFT_IEEE1394: case IFT_IEEE80211: case IFT_INFINIBAND: in6 = ia->ia_addr.sin6_addr; if (in6_get_hw_ifid(ifp, &in6) == 0 && IN6_ARE_ADDR_EQUAL(&ia->ia_addr.sin6_addr, &in6)) { ND_IFINFO(ifp)->flags |= ND6_IFF_IFDISABLED; log(LOG_ERR, "%s: possible hardware address " "duplication detected, disable IPv6\n", if_name(ifp)); } break; } } } static void nd6_dad_ns_output(struct dadq *dp, struct ifaddr *ifa) { struct in6_ifaddr *ia = (struct in6_ifaddr *)ifa; struct ifnet *ifp = ifa->ifa_ifp; int i; dp->dad_ns_tcount++; if ((ifp->if_flags & IFF_UP) == 0) { return; } if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { return; } dp->dad_ns_ocount++; if (V_dad_enhanced != 0) { for (i = 0; i < ND_OPT_NONCE_LEN32; i++) dp->dad_nonce[i] = arc4random(); /* * XXXHRS: Note that in the case that * DupAddrDetectTransmits > 1, multiple NS messages with * different nonces can be looped back in an unexpected * order. The current implementation recognizes only * the latest nonce on the sender side. Practically it * should work well in almost all cases. */ } nd6_ns_output(ifp, NULL, NULL, &ia->ia_addr.sin6_addr, (uint8_t *)&dp->dad_nonce[0]); } static void nd6_dad_ns_input(struct ifaddr *ifa, struct nd_opt_nonce *ndopt_nonce) { struct in6_ifaddr *ia; struct ifnet *ifp; const struct in6_addr *taddr6; struct dadq *dp; if (ifa == NULL) panic("ifa == NULL in nd6_dad_ns_input"); ia = (struct in6_ifaddr *)ifa; ifp = ifa->ifa_ifp; taddr6 = &ia->ia_addr.sin6_addr; /* Ignore Nonce option when Enhanced DAD is disabled. */ if (V_dad_enhanced == 0) ndopt_nonce = NULL; dp = nd6_dad_find(ifa, ndopt_nonce); if (dp == NULL) return; dp->dad_ns_icount++; nd6_dad_rele(dp); } static void nd6_dad_na_input(struct ifaddr *ifa) { struct dadq *dp; if (ifa == NULL) panic("ifa == NULL in nd6_dad_na_input"); dp = nd6_dad_find(ifa, NULL); if (dp != NULL) { dp->dad_na_icount++; nd6_dad_rele(dp); } }