/* $OpenBSD: ospfe.c,v 1.83 2011/07/04 04:34:14 claudio Exp $ */ /* * Copyright (c) 2005 Claudio Jeker * Copyright (c) 2004 Esben Norby * Copyright (c) 2003, 2004 Henning Brauer * * Permission to use, copy, modify, and distribute this software for any * purpose with or without fee is hereby granted, provided that the above * copyright notice and this permission notice appear in all copies. * * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "ospf.h" #include "ospfd.h" #include "ospfe.h" #include "rde.h" #include "control.h" #include "log.h" void ospfe_sig_handler(int, short, void *); void ospfe_shutdown(void); void orig_rtr_lsa_all(struct area *); struct iface *find_vlink(struct abr_rtr *); struct ospfd_conf *oeconf = NULL, *nconf; struct imsgev *iev_main; struct imsgev *iev_rde; int oe_nofib; /* ARGSUSED */ void ospfe_sig_handler(int sig, short event, void *bula) { switch (sig) { case SIGINT: case SIGTERM: ospfe_shutdown(); /* NOTREACHED */ default: fatalx("unexpected signal"); } } /* ospf engine */ pid_t ospfe(struct ospfd_conf *xconf, int pipe_parent2ospfe[2], int pipe_ospfe2rde[2], int pipe_parent2rde[2]) { struct area *area; struct iface *iface; struct redistribute *r; struct passwd *pw; struct event ev_sigint, ev_sigterm; pid_t pid; switch (pid = fork()) { case -1: fatal("cannot fork"); case 0: break; default: return (pid); } /* create ospfd control socket outside chroot */ if (control_init(xconf->csock) == -1) fatalx("control socket setup failed"); /* create the raw ip socket */ if ((xconf->ospf_socket = socket(AF_INET, SOCK_RAW, IPPROTO_OSPF)) == -1) fatal("error creating raw socket"); /* set some defaults */ if (if_set_mcast_loop(xconf->ospf_socket) == -1) fatal("if_set_mcast_loop"); if (if_set_ip_hdrincl(xconf->ospf_socket) == -1) fatal("if_set_ip_hdrincl"); if (if_set_recvif(xconf->ospf_socket, 1) == -1) fatal("if_set_recvif"); if_set_sockbuf(xconf->ospf_socket); oeconf = xconf; if (oeconf->flags & OSPFD_FLAG_NO_FIB_UPDATE) oe_nofib = 1; if ((pw = getpwnam(OSPFD_USER)) == NULL) fatal("getpwnam"); if (chroot(pw->pw_dir) == -1) fatal("chroot"); if (chdir("/") == -1) fatal("chdir(\"/\")"); setproctitle("ospf engine"); ospfd_process = PROC_OSPF_ENGINE; if (setgroups(1, &pw->pw_gid) || setresgid(pw->pw_gid, pw->pw_gid, pw->pw_gid) || setresuid(pw->pw_uid, pw->pw_uid, pw->pw_uid)) fatal("can't drop privileges"); event_init(); nbr_init(NBR_HASHSIZE); lsa_cache_init(LSA_HASHSIZE); /* setup signal handler */ signal_set(&ev_sigint, SIGINT, ospfe_sig_handler, NULL); signal_set(&ev_sigterm, SIGTERM, ospfe_sig_handler, NULL); signal_add(&ev_sigint, NULL); signal_add(&ev_sigterm, NULL); signal(SIGPIPE, SIG_IGN); signal(SIGHUP, SIG_IGN); /* setup pipes */ close(pipe_parent2ospfe[0]); close(pipe_ospfe2rde[1]); close(pipe_parent2rde[0]); close(pipe_parent2rde[1]); if ((iev_rde = malloc(sizeof(struct imsgev))) == NULL || (iev_main = malloc(sizeof(struct imsgev))) == NULL) fatal(NULL); imsg_init(&iev_rde->ibuf, pipe_ospfe2rde[0]); iev_rde->handler = ospfe_dispatch_rde; imsg_init(&iev_main->ibuf, pipe_parent2ospfe[1]); iev_main->handler = ospfe_dispatch_main; /* setup event handler */ iev_rde->events = EV_READ; event_set(&iev_rde->ev, iev_rde->ibuf.fd, iev_rde->events, iev_rde->handler, iev_rde); event_add(&iev_rde->ev, NULL); iev_main->events = EV_READ; event_set(&iev_main->ev, iev_main->ibuf.fd, iev_main->events, iev_main->handler, iev_main); event_add(&iev_main->ev, NULL); event_set(&oeconf->ev, oeconf->ospf_socket, EV_READ|EV_PERSIST, recv_packet, oeconf); event_add(&oeconf->ev, NULL); /* remove unneeded config stuff */ while ((r = SIMPLEQ_FIRST(&oeconf->redist_list)) != NULL) { SIMPLEQ_REMOVE_HEAD(&oeconf->redist_list, entry); free(r); } LIST_FOREACH(area, &oeconf->area_list, entry) { while ((r = SIMPLEQ_FIRST(&area->redist_list)) != NULL) { SIMPLEQ_REMOVE_HEAD(&area->redist_list, entry); free(r); } } /* listen on ospfd control socket */ TAILQ_INIT(&ctl_conns); control_listen(); if ((pkt_ptr = calloc(1, READ_BUF_SIZE)) == NULL) fatal("ospfe"); /* start interfaces */ LIST_FOREACH(area, &oeconf->area_list, entry) { ospfe_demote_area(area, 0); LIST_FOREACH(iface, &area->iface_list, entry) { if_init(xconf, iface); if (if_fsm(iface, IF_EVT_UP)) { log_debug("error starting interface %s", iface->name); } } } event_dispatch(); ospfe_shutdown(); /* NOTREACHED */ return (0); } void ospfe_shutdown(void) { struct area *area; struct iface *iface; /* stop all interfaces and remove all areas */ while ((area = LIST_FIRST(&oeconf->area_list)) != NULL) { LIST_FOREACH(iface, &area->iface_list, entry) { if (if_fsm(iface, IF_EVT_DOWN)) { log_debug("error stopping interface %s", iface->name); } } LIST_REMOVE(area, entry); area_del(area); } nbr_del(nbr_find_peerid(NBR_IDSELF)); kr_shutdown(); close(oeconf->ospf_socket); /* clean up */ msgbuf_write(&iev_rde->ibuf.w); msgbuf_clear(&iev_rde->ibuf.w); free(iev_rde); msgbuf_write(&iev_main->ibuf.w); msgbuf_clear(&iev_main->ibuf.w); free(iev_main); free(oeconf); free(pkt_ptr); log_info("ospf engine exiting"); _exit(0); } /* imesg */ int ospfe_imsg_compose_parent(int type, pid_t pid, void *data, u_int16_t datalen) { return (imsg_compose_event(iev_main, type, 0, pid, -1, data, datalen)); } int ospfe_imsg_compose_rde(int type, u_int32_t peerid, pid_t pid, void *data, u_int16_t datalen) { return (imsg_compose_event(iev_rde, type, peerid, pid, -1, data, datalen)); } /* ARGSUSED */ void ospfe_dispatch_main(int fd, short event, void *bula) { static struct area *narea; static struct iface *niface; struct ifaddrdel *ifc; struct imsg imsg; struct imsgev *iev = bula; struct imsgbuf *ibuf = &iev->ibuf; struct area *area = NULL; struct iface *iface = NULL; struct kif *kif; struct auth_md md; int n, link_ok, stub_changed, shut = 0; if (event & EV_READ) { if ((n = imsg_read(ibuf)) == -1) fatal("imsg_read error"); if (n == 0) /* connection closed */ shut = 1; } if (event & EV_WRITE) { if (msgbuf_write(&ibuf->w) == -1) fatal("msgbuf_write"); } for (;;) { if ((n = imsg_get(ibuf, &imsg)) == -1) fatal("ospfe_dispatch_main: imsg_read error"); if (n == 0) break; switch (imsg.hdr.type) { case IMSG_IFINFO: if (imsg.hdr.len != IMSG_HEADER_SIZE + sizeof(struct kif)) fatalx("IFINFO imsg with wrong len"); kif = imsg.data; link_ok = (kif->flags & IFF_UP) && LINK_STATE_IS_UP(kif->link_state); LIST_FOREACH(area, &oeconf->area_list, entry) { LIST_FOREACH(iface, &area->iface_list, entry) { if (kif->ifindex == iface->ifindex && iface->type != IF_TYPE_VIRTUALLINK) { iface->flags = kif->flags; iface->linkstate = kif->link_state; if (link_ok) { if_fsm(iface, IF_EVT_UP); log_warnx("interface %s" " up", iface->name); } else { if_fsm(iface, IF_EVT_DOWN); log_warnx("interface %s" " down", iface->name); } } } } break; case IMSG_IFADDRDEL: if (imsg.hdr.len != IMSG_HEADER_SIZE + sizeof(struct ifaddrdel)) fatalx("IFADDRDEL imsg with wrong len"); ifc = imsg.data; LIST_FOREACH(area, &oeconf->area_list, entry) { LIST_FOREACH(iface, &area->iface_list, entry) { if (ifc->ifindex == iface->ifindex && ifc->addr.s_addr == iface->addr.s_addr) { if_fsm(iface, IF_EVT_DOWN); log_warnx("interface %s:%s " "gone", iface->name, inet_ntoa(iface->addr)); break; } } } break; case IMSG_RECONF_CONF: if ((nconf = malloc(sizeof(struct ospfd_conf))) == NULL) fatal(NULL); memcpy(nconf, imsg.data, sizeof(struct ospfd_conf)); LIST_INIT(&nconf->area_list); LIST_INIT(&nconf->cand_list); break; case IMSG_RECONF_AREA: if ((narea = area_new()) == NULL) fatal(NULL); memcpy(narea, imsg.data, sizeof(struct area)); LIST_INIT(&narea->iface_list); LIST_INIT(&narea->nbr_list); RB_INIT(&narea->lsa_tree); SIMPLEQ_INIT(&narea->redist_list); LIST_INSERT_HEAD(&nconf->area_list, narea, entry); break; case IMSG_RECONF_IFACE: if ((niface = malloc(sizeof(struct iface))) == NULL) fatal(NULL); memcpy(niface, imsg.data, sizeof(struct iface)); LIST_INIT(&niface->nbr_list); TAILQ_INIT(&niface->ls_ack_list); TAILQ_INIT(&niface->auth_md_list); RB_INIT(&niface->lsa_tree); niface->area = narea; LIST_INSERT_HEAD(&narea->iface_list, niface, entry); break; case IMSG_RECONF_AUTHMD: memcpy(&md, imsg.data, sizeof(struct auth_md)); md_list_add(&niface->auth_md_list, md.keyid, md.key); break; case IMSG_RECONF_END: if ((oeconf->flags & OSPFD_FLAG_STUB_ROUTER) != (nconf->flags & OSPFD_FLAG_STUB_ROUTER)) stub_changed = 1; else stub_changed = 0; merge_config(oeconf, nconf); nconf = NULL; if (stub_changed) orig_rtr_lsa_all(NULL); break; case IMSG_CTL_KROUTE: case IMSG_CTL_KROUTE_ADDR: case IMSG_CTL_IFINFO: case IMSG_CTL_END: control_imsg_relay(&imsg); break; default: log_debug("ospfe_dispatch_main: error handling imsg %d", imsg.hdr.type); break; } imsg_free(&imsg); } if (!shut) imsg_event_add(iev); else { /* this pipe is dead, so remove the event handler */ event_del(&iev->ev); event_loopexit(NULL); } } /* ARGSUSED */ void ospfe_dispatch_rde(int fd, short event, void *bula) { struct lsa_hdr lsa_hdr; struct imsgev *iev = bula; struct imsgbuf *ibuf = &iev->ibuf; struct nbr *nbr; struct lsa_hdr *lhp; struct lsa_ref *ref; struct area *area; struct iface *iface; struct lsa_entry *le; struct imsg imsg; struct abr_rtr ar; int n, noack = 0, shut = 0; u_int16_t l, age; if (event & EV_READ) { if ((n = imsg_read(ibuf)) == -1) fatal("imsg_read error"); if (n == 0) /* connection closed */ shut = 1; } if (event & EV_WRITE) { if (msgbuf_write(&ibuf->w) == -1) fatal("msgbuf_write"); } for (;;) { if ((n = imsg_get(ibuf, &imsg)) == -1) fatal("ospfe_dispatch_rde: imsg_read error"); if (n == 0) break; switch (imsg.hdr.type) { case IMSG_DD: nbr = nbr_find_peerid(imsg.hdr.peerid); if (nbr == NULL) break; /* * Ignore imsg when in the wrong state because a * NBR_EVT_SEQ_NUM_MIS may have been issued in between. * Luckily regetting the DB snapshot acts as a barrier * for both state and process synchronisation. */ if ((nbr->state & NBR_STA_FLOOD) == 0) break; /* put these on my ls_req_list for retrieval */ lhp = lsa_hdr_new(); memcpy(lhp, imsg.data, sizeof(*lhp)); ls_req_list_add(nbr, lhp); break; case IMSG_DD_END: nbr = nbr_find_peerid(imsg.hdr.peerid); if (nbr == NULL) break; /* see above */ if ((nbr->state & NBR_STA_FLOOD) == 0) break; nbr->dd_pending--; if (nbr->dd_pending == 0 && nbr->state & NBR_STA_LOAD) { if (ls_req_list_empty(nbr)) nbr_fsm(nbr, NBR_EVT_LOAD_DONE); else start_ls_req_tx_timer(nbr); } break; case IMSG_DD_BADLSA: nbr = nbr_find_peerid(imsg.hdr.peerid); if (nbr == NULL) break; if (nbr->iface->self == nbr) fatalx("ospfe_dispatch_rde: " "dummy neighbor got BADREQ"); nbr_fsm(nbr, NBR_EVT_SEQ_NUM_MIS); break; case IMSG_DB_SNAPSHOT: nbr = nbr_find_peerid(imsg.hdr.peerid); if (nbr == NULL) break; /* add LSA header to the neighbor db_sum_list */ lhp = lsa_hdr_new(); memcpy(lhp, imsg.data, sizeof(*lhp)); db_sum_list_add(nbr, lhp); break; case IMSG_DB_END: nbr = nbr_find_peerid(imsg.hdr.peerid); if (nbr == NULL) break; /* snapshot done, start tx of dd packets */ nbr_fsm(nbr, NBR_EVT_SNAP_DONE); break; case IMSG_LS_FLOOD: nbr = nbr_find_peerid(imsg.hdr.peerid); if (nbr == NULL) break; l = imsg.hdr.len - IMSG_HEADER_SIZE; if (l < sizeof(lsa_hdr)) fatalx("ospfe_dispatch_rde: " "bad imsg size"); memcpy(&lsa_hdr, imsg.data, sizeof(lsa_hdr)); ref = lsa_cache_add(imsg.data, l); if (lsa_hdr.type == LSA_TYPE_EXTERNAL) { /* * flood on all areas but stub areas and * virtual links */ LIST_FOREACH(area, &oeconf->area_list, entry) { if (area->stub) continue; LIST_FOREACH(iface, &area->iface_list, entry) { noack += lsa_flood(iface, nbr, &lsa_hdr, imsg.data); } } } else if (lsa_hdr.type == LSA_TYPE_LINK_OPAQ) { /* * Flood on interface only */ noack += lsa_flood(nbr->iface, nbr, &lsa_hdr, imsg.data); } else { /* * Flood on all area interfaces. For * area 0.0.0.0 include the virtual links. */ area = nbr->iface->area; LIST_FOREACH(iface, &area->iface_list, entry) { noack += lsa_flood(iface, nbr, &lsa_hdr, imsg.data); } /* XXX virtual links */ } /* remove from ls_req_list */ le = ls_req_list_get(nbr, &lsa_hdr); if (!(nbr->state & NBR_STA_FULL) && le != NULL) { ls_req_list_free(nbr, le); /* * XXX no need to ack requested lsa * the problem is that the RFC is very * unclear about this. */ noack = 1; } if (!noack && nbr->iface != NULL && nbr->iface->self != nbr) { if (!(nbr->iface->state & IF_STA_BACKUP) || nbr->iface->dr == nbr) { /* delayed ack */ lhp = lsa_hdr_new(); memcpy(lhp, &lsa_hdr, sizeof(*lhp)); ls_ack_list_add(nbr->iface, lhp); } } lsa_cache_put(ref, nbr); break; case IMSG_LS_UPD: /* * IMSG_LS_UPD is used in three cases: * 1. as response to ls requests * 2. as response to ls updates where the DB * is newer then the sent LSA * 3. in EXSTART when the LSA has age MaxAge */ l = imsg.hdr.len - IMSG_HEADER_SIZE; if (l < sizeof(lsa_hdr)) fatalx("ospfe_dispatch_rde: " "bad imsg size"); nbr = nbr_find_peerid(imsg.hdr.peerid); if (nbr == NULL) break; if (nbr->iface->self == nbr) break; memcpy(&age, imsg.data, sizeof(age)); ref = lsa_cache_add(imsg.data, l); if (ntohs(age) >= MAX_AGE) /* add to retransmit list */ ls_retrans_list_add(nbr, imsg.data, 0, 0); else ls_retrans_list_add(nbr, imsg.data, 0, 1); lsa_cache_put(ref, nbr); break; case IMSG_LS_ACK: /* * IMSG_LS_ACK is used in two cases: * 1. LSA was a duplicate * 2. LS age is MaxAge and there is no current * instance in the DB plus no neighbor in state * Exchange or Loading */ nbr = nbr_find_peerid(imsg.hdr.peerid); if (nbr == NULL) break; if (nbr->iface->self == nbr) break; if (imsg.hdr.len - IMSG_HEADER_SIZE != sizeof(lsa_hdr)) fatalx("ospfe_dispatch_rde: bad imsg size"); memcpy(&lsa_hdr, imsg.data, sizeof(lsa_hdr)); /* for case one check for implied acks */ if (nbr->iface->state & IF_STA_DROTHER) if (ls_retrans_list_del(nbr->iface->self, &lsa_hdr) == 0) break; if (ls_retrans_list_del(nbr, &lsa_hdr) == 0) break; /* send a direct acknowledgement */ send_direct_ack(nbr->iface, nbr->addr, imsg.data, imsg.hdr.len - IMSG_HEADER_SIZE); break; case IMSG_LS_BADREQ: nbr = nbr_find_peerid(imsg.hdr.peerid); if (nbr == NULL) break; if (nbr->iface->self == nbr) fatalx("ospfe_dispatch_rde: " "dummy neighbor got BADREQ"); nbr_fsm(nbr, NBR_EVT_BAD_LS_REQ); break; case IMSG_ABR_UP: memcpy(&ar, imsg.data, sizeof(ar)); if ((iface = find_vlink(&ar)) != NULL && iface->state == IF_STA_DOWN) if (if_fsm(iface, IF_EVT_UP)) { log_debug("error starting interface %s", iface->name); } break; case IMSG_ABR_DOWN: memcpy(&ar, imsg.data, sizeof(ar)); if ((iface = find_vlink(&ar)) != NULL && iface->state == IF_STA_POINTTOPOINT) if (if_fsm(iface, IF_EVT_DOWN)) { log_debug("error stopping interface %s", iface->name); } break; case IMSG_CTL_AREA: case IMSG_CTL_IFACE: case IMSG_CTL_END: case IMSG_CTL_SHOW_DATABASE: case IMSG_CTL_SHOW_DB_EXT: case IMSG_CTL_SHOW_DB_NET: case IMSG_CTL_SHOW_DB_RTR: case IMSG_CTL_SHOW_DB_SELF: case IMSG_CTL_SHOW_DB_SUM: case IMSG_CTL_SHOW_DB_ASBR: case IMSG_CTL_SHOW_DB_OPAQ: case IMSG_CTL_SHOW_RIB: case IMSG_CTL_SHOW_SUM: case IMSG_CTL_SHOW_SUM_AREA: control_imsg_relay(&imsg); break; default: log_debug("ospfe_dispatch_rde: error handling imsg %d", imsg.hdr.type); break; } imsg_free(&imsg); } if (!shut) imsg_event_add(iev); else { /* this pipe is dead, so remove the event handler */ event_del(&iev->ev); event_loopexit(NULL); } } struct iface * find_vlink(struct abr_rtr *ar) { struct area *area; struct iface *iface = NULL; LIST_FOREACH(area, &oeconf->area_list, entry) LIST_FOREACH(iface, &area->iface_list, entry) if (iface->abr_id.s_addr == ar->abr_id.s_addr && iface->type == IF_TYPE_VIRTUALLINK && iface->area->id.s_addr == ar->area.s_addr) { iface->dst.s_addr = ar->dst_ip.s_addr; iface->addr.s_addr = ar->addr.s_addr; iface->metric = ar->metric; return (iface); } return (iface); } void orig_rtr_lsa_all(struct area *area) { struct area *a; /* * update all router LSA in all areas except area itself, * as this update is already running. */ LIST_FOREACH(a, &oeconf->area_list, entry) if (a != area) orig_rtr_lsa(a); } void orig_rtr_lsa(struct area *area) { struct lsa_hdr lsa_hdr; struct lsa_rtr lsa_rtr; struct lsa_rtr_link rtr_link; struct iface *iface; struct ibuf *buf; struct nbr *nbr, *self = NULL; u_int16_t num_links = 0; u_int16_t chksum; u_int8_t border, virtual = 0; log_debug("orig_rtr_lsa: area %s", inet_ntoa(area->id)); if ((buf = ibuf_dynamic(sizeof(lsa_hdr), IP_MAXPACKET - sizeof(struct ip) - sizeof(struct ospf_hdr) - sizeof(u_int32_t) - MD5_DIGEST_LENGTH)) == NULL) fatal("orig_rtr_lsa"); /* reserve space for LSA header and LSA Router header */ if (ibuf_reserve(buf, sizeof(lsa_hdr)) == NULL) fatal("orig_rtr_lsa: ibuf_reserve failed"); if (ibuf_reserve(buf, sizeof(lsa_rtr)) == NULL) fatal("orig_rtr_lsa: ibuf_reserve failed"); /* links */ LIST_FOREACH(iface, &area->iface_list, entry) { if (self == NULL && iface->self != NULL) self = iface->self; bzero(&rtr_link, sizeof(rtr_link)); if (iface->state & IF_STA_LOOPBACK) { rtr_link.id = iface->addr.s_addr; rtr_link.data = 0xffffffff; rtr_link.type = LINK_TYPE_STUB_NET; rtr_link.metric = htons(iface->metric); num_links++; if (ibuf_add(buf, &rtr_link, sizeof(rtr_link))) fatalx("orig_rtr_lsa: ibuf_add failed"); continue; } switch (iface->type) { case IF_TYPE_POINTOPOINT: LIST_FOREACH(nbr, &iface->nbr_list, entry) if (nbr != iface->self && nbr->state & NBR_STA_FULL) break; if (nbr) { log_debug("orig_rtr_lsa: point-to-point, " "interface %s", iface->name); rtr_link.id = nbr->id.s_addr; rtr_link.data = iface->addr.s_addr; rtr_link.type = LINK_TYPE_POINTTOPOINT; /* RFC 3137: stub router support */ if (oeconf->flags & OSPFD_FLAG_STUB_ROUTER || oe_nofib) rtr_link.metric = MAX_METRIC; else rtr_link.metric = htons(iface->metric); num_links++; if (ibuf_add(buf, &rtr_link, sizeof(rtr_link))) fatalx("orig_rtr_lsa: ibuf_add failed"); } if (iface->state & IF_STA_POINTTOPOINT) { log_debug("orig_rtr_lsa: stub net, " "interface %s", iface->name); bzero(&rtr_link, sizeof(rtr_link)); if (nbr) { rtr_link.id = nbr->addr.s_addr; rtr_link.data = 0xffffffff; } else { rtr_link.id = iface->addr.s_addr; rtr_link.data = iface->mask.s_addr; } rtr_link.type = LINK_TYPE_STUB_NET; rtr_link.metric = htons(iface->metric); num_links++; if (ibuf_add(buf, &rtr_link, sizeof(rtr_link))) fatalx("orig_rtr_lsa: ibuf_add failed"); } continue; case IF_TYPE_BROADCAST: case IF_TYPE_NBMA: if ((iface->state & IF_STA_MULTI)) { if (iface->dr == iface->self) { LIST_FOREACH(nbr, &iface->nbr_list, entry) if (nbr != iface->self && nbr->state & NBR_STA_FULL) break; } else nbr = iface->dr; if (nbr && nbr->state & NBR_STA_FULL) { log_debug("orig_rtr_lsa: transit net, " "interface %s", iface->name); rtr_link.id = iface->dr->addr.s_addr; rtr_link.data = iface->addr.s_addr; rtr_link.type = LINK_TYPE_TRANSIT_NET; break; } } /* * do not add a stub net LSA for interfaces that are: * - down * - have a linkstate which is down */ if (!(iface->flags & IFF_UP) || !LINK_STATE_IS_UP(iface->linkstate)) continue; log_debug("orig_rtr_lsa: stub net, " "interface %s", iface->name); rtr_link.id = iface->addr.s_addr & iface->mask.s_addr; rtr_link.data = iface->mask.s_addr; rtr_link.type = LINK_TYPE_STUB_NET; rtr_link.num_tos = 0; /* * backup carp interfaces are anounced with high metric * for faster failover. */ if (iface->media_type == IFT_CARP && iface->linkstate == LINK_STATE_DOWN) rtr_link.metric = MAX_METRIC; else rtr_link.metric = htons(iface->metric); num_links++; if (ibuf_add(buf, &rtr_link, sizeof(rtr_link))) fatalx("orig_rtr_lsa: ibuf_add failed"); continue; case IF_TYPE_VIRTUALLINK: LIST_FOREACH(nbr, &iface->nbr_list, entry) { if (nbr != iface->self && nbr->state & NBR_STA_FULL) break; } if (nbr) { rtr_link.id = nbr->id.s_addr; rtr_link.data = iface->addr.s_addr; rtr_link.type = LINK_TYPE_VIRTUAL; /* RFC 3137: stub router support */ if (oeconf->flags & OSPFD_FLAG_STUB_ROUTER || oe_nofib) rtr_link.metric = MAX_METRIC; else rtr_link.metric = htons(iface->metric); num_links++; virtual = 1; if (ibuf_add(buf, &rtr_link, sizeof(rtr_link))) fatalx("orig_rtr_lsa: ibuf_add failed"); log_debug("orig_rtr_lsa: virtual link, " "interface %s", iface->name); } continue; case IF_TYPE_POINTOMULTIPOINT: log_debug("orig_rtr_lsa: stub net, " "interface %s", iface->name); rtr_link.id = iface->addr.s_addr; rtr_link.data = 0xffffffff; rtr_link.type = LINK_TYPE_STUB_NET; rtr_link.metric = htons(iface->metric); num_links++; if (ibuf_add(buf, &rtr_link, sizeof(rtr_link))) fatalx("orig_rtr_lsa: ibuf_add failed"); LIST_FOREACH(nbr, &iface->nbr_list, entry) { if (nbr != iface->self && nbr->state & NBR_STA_FULL) { bzero(&rtr_link, sizeof(rtr_link)); log_debug("orig_rtr_lsa: " "point-to-multipoint, interface %s", iface->name); rtr_link.id = nbr->addr.s_addr; rtr_link.data = iface->addr.s_addr; rtr_link.type = LINK_TYPE_POINTTOPOINT; /* RFC 3137: stub router support */ if (oe_nofib || oeconf->flags & OSPFD_FLAG_STUB_ROUTER) rtr_link.metric = MAX_METRIC; else rtr_link.metric = htons(iface->metric); num_links++; if (ibuf_add(buf, &rtr_link, sizeof(rtr_link))) fatalx("orig_rtr_lsa: " "ibuf_add failed"); } } continue; default: fatalx("orig_rtr_lsa: unknown interface type"); } rtr_link.num_tos = 0; /* RFC 3137: stub router support */ if ((oeconf->flags & OSPFD_FLAG_STUB_ROUTER || oe_nofib) && rtr_link.type != LINK_TYPE_STUB_NET) rtr_link.metric = MAX_METRIC; else rtr_link.metric = htons(iface->metric); num_links++; if (ibuf_add(buf, &rtr_link, sizeof(rtr_link))) fatalx("orig_rtr_lsa: ibuf_add failed"); } /* LSA router header */ lsa_rtr.flags = 0; /* * Set the E bit as soon as an as-ext lsa may be redistributed, only * setting it in case we redistribute something is not worth the fuss. * Do not set the E bit in case of a stub area. */ if (oeconf->redistribute && !area->stub) lsa_rtr.flags |= OSPF_RTR_E; border = (area_border_router(oeconf) != 0); if (border != oeconf->border) { oeconf->border = border; orig_rtr_lsa_all(area); } if (oeconf->border) lsa_rtr.flags |= OSPF_RTR_B; /* TODO set V flag if a active virtual link ends here and the * area is the transit area for this link. */ if (virtual) lsa_rtr.flags |= OSPF_RTR_V; lsa_rtr.dummy = 0; lsa_rtr.nlinks = htons(num_links); memcpy(ibuf_seek(buf, sizeof(lsa_hdr), sizeof(lsa_rtr)), &lsa_rtr, sizeof(lsa_rtr)); /* LSA header */ lsa_hdr.age = htons(DEFAULT_AGE); lsa_hdr.opts = area_ospf_options(area); lsa_hdr.type = LSA_TYPE_ROUTER; lsa_hdr.ls_id = oeconf->rtr_id.s_addr; lsa_hdr.adv_rtr = oeconf->rtr_id.s_addr; lsa_hdr.seq_num = htonl(INIT_SEQ_NUM); lsa_hdr.len = htons(ibuf_size(buf)); lsa_hdr.ls_chksum = 0; /* updated later */ memcpy(ibuf_seek(buf, 0, sizeof(lsa_hdr)), &lsa_hdr, sizeof(lsa_hdr)); chksum = htons(iso_cksum(buf->buf, ibuf_size(buf), LS_CKSUM_OFFSET)); memcpy(ibuf_seek(buf, LS_CKSUM_OFFSET, sizeof(chksum)), &chksum, sizeof(chksum)); if (self) imsg_compose_event(iev_rde, IMSG_LS_UPD, self->peerid, 0, -1, buf->buf, ibuf_size(buf)); else log_warnx("orig_rtr_lsa: empty area %s", inet_ntoa(area->id)); ibuf_free(buf); } void orig_net_lsa(struct iface *iface) { struct lsa_hdr lsa_hdr; struct nbr *nbr; struct ibuf *buf; int num_rtr = 0; u_int16_t chksum; if ((buf = ibuf_dynamic(sizeof(lsa_hdr), IP_MAXPACKET - sizeof(struct ip) - sizeof(struct ospf_hdr) - sizeof(u_int32_t) - MD5_DIGEST_LENGTH)) == NULL) fatal("orig_net_lsa"); /* reserve space for LSA header and LSA Router header */ if (ibuf_reserve(buf, sizeof(lsa_hdr)) == NULL) fatal("orig_net_lsa: ibuf_reserve failed"); /* LSA net mask and then all fully adjacent routers */ if (ibuf_add(buf, &iface->mask, sizeof(iface->mask))) fatal("orig_net_lsa: ibuf_add failed"); /* fully adjacent neighbors + self */ LIST_FOREACH(nbr, &iface->nbr_list, entry) if (nbr->state & NBR_STA_FULL) { if (ibuf_add(buf, &nbr->id, sizeof(nbr->id))) fatal("orig_net_lsa: ibuf_add failed"); num_rtr++; } if (num_rtr == 1) { /* non transit net therefore no need to generate a net lsa */ ibuf_free(buf); return; } /* LSA header */ if (iface->state & IF_STA_DR) lsa_hdr.age = htons(DEFAULT_AGE); else lsa_hdr.age = htons(MAX_AGE); lsa_hdr.opts = area_ospf_options(iface->area); lsa_hdr.type = LSA_TYPE_NETWORK; lsa_hdr.ls_id = iface->addr.s_addr; lsa_hdr.adv_rtr = oeconf->rtr_id.s_addr; lsa_hdr.seq_num = htonl(INIT_SEQ_NUM); lsa_hdr.len = htons(ibuf_size(buf)); lsa_hdr.ls_chksum = 0; /* updated later */ memcpy(ibuf_seek(buf, 0, sizeof(lsa_hdr)), &lsa_hdr, sizeof(lsa_hdr)); chksum = htons(iso_cksum(buf->buf, ibuf_size(buf), LS_CKSUM_OFFSET)); memcpy(ibuf_seek(buf, LS_CKSUM_OFFSET, sizeof(chksum)), &chksum, sizeof(chksum)); imsg_compose_event(iev_rde, IMSG_LS_UPD, iface->self->peerid, 0, -1, buf->buf, ibuf_size(buf)); ibuf_free(buf); } u_int32_t ospfe_router_id(void) { return (oeconf->rtr_id.s_addr); } void ospfe_fib_update(int type) { int old = oe_nofib; if (type == IMSG_CTL_FIB_COUPLE) oe_nofib = 0; if (type == IMSG_CTL_FIB_DECOUPLE) oe_nofib = 1; if (old != oe_nofib) orig_rtr_lsa_all(NULL); } void ospfe_iface_ctl(struct ctl_conn *c, unsigned int idx) { struct area *area; struct iface *iface; struct ctl_iface *ictl; LIST_FOREACH(area, &oeconf->area_list, entry) LIST_FOREACH(iface, &area->iface_list, entry) if (idx == 0 || idx == iface->ifindex) { ictl = if_to_ctl(iface); imsg_compose_event(&c->iev, IMSG_CTL_SHOW_INTERFACE, 0, 0, -1, ictl, sizeof(struct ctl_iface)); } } void ospfe_nbr_ctl(struct ctl_conn *c) { struct area *area; struct iface *iface; struct nbr *nbr; struct ctl_nbr *nctl; LIST_FOREACH(area, &oeconf->area_list, entry) LIST_FOREACH(iface, &area->iface_list, entry) LIST_FOREACH(nbr, &iface->nbr_list, entry) { if (iface->self != nbr) { nctl = nbr_to_ctl(nbr); imsg_compose_event(&c->iev, IMSG_CTL_SHOW_NBR, 0, 0, -1, nctl, sizeof(struct ctl_nbr)); } } imsg_compose_event(&c->iev, IMSG_CTL_END, 0, 0, -1, NULL, 0); } void ospfe_demote_area(struct area *area, int active) { struct demote_msg dmsg; if (ospfd_process != PROC_OSPF_ENGINE || area->demote_group[0] == '\0') return; bzero(&dmsg, sizeof(dmsg)); strlcpy(dmsg.demote_group, area->demote_group, sizeof(dmsg.demote_group)); dmsg.level = area->demote_level; if (active) dmsg.level = -dmsg.level; ospfe_imsg_compose_parent(IMSG_DEMOTE, 0, &dmsg, sizeof(dmsg)); } void ospfe_demote_iface(struct iface *iface, int active) { struct demote_msg dmsg; if (ospfd_process != PROC_OSPF_ENGINE || iface->demote_group[0] == '\0') return; bzero(&dmsg, sizeof(dmsg)); strlcpy(dmsg.demote_group, iface->demote_group, sizeof(dmsg.demote_group)); if (active) dmsg.level = -1; else dmsg.level = 1; log_warnx("ospfe_demote_iface: group %s level %d", dmsg.demote_group, dmsg.level); ospfe_imsg_compose_parent(IMSG_DEMOTE, 0, &dmsg, sizeof(dmsg)); }